Download:
pdf |
pdfIRS Research Bulletin
Recent Research on Tax
Administration and Compliance
Selected Papers Given at the
2010 IRS Research Conference
The Liaison Capitol Hill
Washington, DC
June 29–30, 2010
Compiled and edited by Martha Eller Gangi and Alan Plumley*
Research, Analysis, and Statistics, Internal Revenue Service
*Prepared under the direction of Barry W. Johnson, Chief, Special Studies Branch
IRS Research Bulletin
iii
Foreword
This edition of the IRS Research Bulletin (Publication 1500) features selected
papers from the latest IRS Research Conference, held at the Liaison Capitol
Hill in Washington, DC, on June 29-30, 2010. As in prior years, conference
presenters and attendees included researchers from all areas of the IRS, officials from other government agencies, and academic and private sector experts on tax policy, tax administration, and tax compliance.
The conference began with a keynote address by Mark Ernst, Deputy IRS
Commissioner for Operations Support. Mr. Ernst stated that the IRS has
made great strides toward modernizing its systems and offering taxpayers
excellent service, with research continuing to play an important role in that
progress. He said that one example of our success is that the IRS is increasingly asked to administer nontax government initiatives, and it does so quickly
and well. After his prepared remarks, Mr. Ernst answered a few questions
from the audience.
Rosemary Marcuss, the Director of Research, Analysis, and Statistics, then
led a panel discussion on the impact of globalization on tax administration.
Panelists from Her Majesty’s Revenue and Customs (United Kingdom), the
Mexican Tax Administration Service, and the IRS discussed several growing
trends and efforts being made to address the challenges. The remainder of the
conference included sessions on the tax compliance of large business entities,
influencing individual taxpayer behavior, drivers of noncompliance, tax code
complexity and compliance burden, and enforcement strategies. The conference also included an after-hours poster session that highlighted additional
IRS research.
We trust that this volume will enable IRS executives, managers, employees, stakeholders, and tax administrators elsewhere to stay abreast of the latest trends and research findings affecting Federal tax administration. We also
hope that the research featured here will stimulate improved tax administration and additional helpful research.
iv
IRS Research Bulletin
Acknowledgments
The IRS Research Conference was the result of substantial effort and preparation over a number of months by many people. The conference program
was assembled by a committee representing research organizations throughout the IRS. Members of the program committee included Janice Hedemann
(National Headquarters Office of Research); Melissa Kovalick (Research,
Analysis, and Statistics); Alan Plumley, John Guyton, Kara Leibel, Sandy Lin,
Larry May, Rahul Tikekar, and Leann Weyl (National Headquarters Office of
Research); Elizabeth Kruse (Office of Program Evaluation and Risk Analysis);
Barry Johnson and Tamara Rib (Statistics of Income); Anne Parker and
Katherine Fox (Small Business and Self-Employed); Scott Leary (Tax Exempt
and Government Entities); Patti Davis (Wage and Investment); Tom Beers
(Taxpayer Advocate); Davy Sparkman (Criminal Investigation); and Lois
Petzing (Large and Midsize Business). Melissa Kovalick, Elizabeth Kruse,
Marcella Garland, Craig Swinford, Barbara Vaira, Michelle Chu, Gregory
Baldwin, Leann Weyl, Linda Addison, Daniel Wagner, and Ruth Schwartz
oversaw numerous details to ensure that the conference ran smoothly.
This volume was prepared by Lisa Smith, Paul Bastuscheck, Clay Moulton, and
Camille Swick (layout and graphics) and Martha Eller Gangi (editor), all of the
Statistics of Income Division. The authors of the papers are responsible for
their content and views expressed in these papers do not necessarily represent
the views of the Department of the Treasury or the Internal Revenue Service.
We appreciate the contributions of everyone who helped make the IRS
Research Conference a success.
Janice M. Hedemann
Director, National Headquarters Office of Research
Chair, 2010 IRS Research Conference
IRS Research Bulletin
v
2010 IRS Research Conference
Contents
Foreword.................................................................................................. iii
1. Compliance of Large Business Entities
Partnerships with Reportable Entity Partners, Charles E. Boynton
and Barbara A. Livingston.................................................................. 3
Temporary and Permanent Book-Tax Differences: Complements
or Substitutes?, Jennifer Blouin, Jason DeBacker,
and Stephanie Sikes ......................................................................... 21
2. Influencing Individual Taxpayer Behavior
Solving Information Asymmetry for Offshore Accounts,
Susan Morse...................................................................................... 49
Facilitated Self-Assistance Enhances Taxpayers’ Taxpayer
Assistance Center (TAC) Experiences, Kirsten Davis, Melissa Hayes,
and Erica Jenkins ............................................................................. 87
3. Drivers of Noncompliance
A Balance Due Before Remittance: The Effect on Reporting
Compliance, Paul Corcoro and Peter Adelsheim............................ 107
Predicting Intentional and Inadvertent Noncompliance, Kathleen
M. Carley, Brian Hirshman, Ju-Sung Lee, Michael Martin, Dawn
Roberston, and Jesse St. Charles...................................................... 159
vi
IRS Research Bulletin
4. Tax Code Complexity and Compliance Burden
Individual Taxpayer Compliance Burden: The Role of Assisted
Methods in Taxpayer Response to Increasing Complexity, George
Contos, John Guyton, Patrick Langetieg, and Melissa Vigil............ 191
Enhancing Compliance Through Improved Readability: Evidence
From New Zealand’s Rewrite “Experiment,” Adrian Sawyer........ 221
Tax Compliance Costs: The Effect of Authority Behavior and
Taxpayer Services, Sebastian Eichfelder, Chantal Kegels, and
Michael Schorn............................................................................... 255
5. Enforcement Strategies
Collecting Collected Taxes, Keith Fogg ......................................... 291
Measuring and Tackling the Illicit Market for Excise Goods,
Anthony Rourke.............................................................................. 495
Inspectors or Google Earth? Optimal Fiscal Policies Under
Uncertain Detection of Evaders, Martin Besfamille and
Pablo Olmos.................................................................................... 513
6. Poster Session
An Analysis of Preparer Testing on Compliance, Chris Hess, Karen
Yeager, Michael Bourque, Amy Sriuthai, and Christine Glass........ 533
7. Appendix
Conference Program...................................................................... 553
List of Attendees............................................................................. 559
1
D
Compliance of Large Business Entities
Boynton Livingston
Blouin Sikes
Partnerships with Reportable
Entity Partners1
Charles E. Boynton and Barbara A. Livingston, Internal Revenue Service
P
artnerships offer incredible flexibility as building blocks in complex organizations. Partnerships can be arranged in tiers and used as substitutes for corporate subsidiaries. Each partnership within an organization requires its
own tax return. Multiple partnerships, and thus multiple tax returns, within an
organization create the possibility that the details of a given transaction may be
distributed across several tax returns. Multiple tax returns within a single economic organization potentially decrease transparency to tax authorities as to the
true nature of the economic transactions.
Partnerships are an important and growing component of the U.S. tax system.
For example, in 2005, just over 2.7 million partnerships filed tax returns, steadily
increasing in 2006 to 2.9 million, then in 2007 to almost 3.1 million. Empirical
research on structures employing partnerships is limited.2
We focus on partnerships effectively controlled by other partnerships or by
corporations as an interesting sub-sample of the partnership population. We ask
whether partnerships effectively controlled by a corporation are different from
partnerships effectively controlled by another partnership. We look at the relation
between the asset size of the controlling entity and the asset size of the effectively
controlled partnership. Furthermore, we look at the relation between the industry
of the controlling entity and the industry of the controlled partnerships.
To do so, this report uses Tax Year 2007 partnership data made available by
the Statistics of Income (SOI) Division of the IRS. We believe this report is the
first publicly available descriptive study using tax data of partnerships effectively
controlled by other entities.
This report is organized as follows. We first provide technical background on
the U.S. tax reporting requirements for partnerships during Tax Year 2007. Next,
we outline the steps in identifying reportable entity partners in the SOI data.
Third, we give a descriptive overview of the partnership population, followed by
an analysis of reportable entity partners. The last section of the report provides
concluding observations.
Schedule M-3 and Partnerships with REPs
The IRS introduced the Form 1120 Schedule M-3 in 2004 to reconcile corporation
financial statement income with corporation taxable income for corporations with
4
Boynton and Livingston
assets of $10 million or more at the end of the tax year.3 In 2006, the IRS introduced Form 1065 Schedule M-3 to reconcile partnership financial statement income with partnership taxable income. The Form 1065 Schedule M-3 is required
of all partnerships with assets of $10 million or more at the end of the tax year.
The Form 1065 Schedule M-3 is also required of any smaller partnership if the
partnership had any of the following: adjusted total assets of $10 million or more
for the tax year, total receipts of $35 million or more for the tax year, or a reportable entity partner on any day of the tax year.4
As of 2006, the instructions for the Form 1065 and Form 1120 Schedules M-3
define a reportable entity partner. A reportable entity partner (REP) with respect
to a partnership is a corporation or partnership that owns, directly or, under the
Schedule M-3 instructions, indirectly, 50 percent or more of the partnership’s profit, loss, or capital on any day of the tax year, and itself was required to file Schedule
M-3 on its most recently filed U.S. tax return filed prior to that day.
A corporation or partnership that becomes a REP with respect to a partnership
must inform the partnership within 30 days of its name, employer identification
number (EIN), and maximum (direct or indirect) ownership interest.
A partnership with a REP must file Schedule M-3 even if it is not otherwise
required to do so and must report the REP name, EIN, and maximum ownership
interest on the partnership’s own Schedule M-3. If the partnership has two or
more REPs for the year, it reports the two with the maximum ownership interest.
The indirect ownership provisions for REPs follow an effective control model
testing for 50 percent or more ownership at each link.5 In general, an entity owning 50 percent or more of another entity is deemed to own all the corporate and
partnership interests of the owned entity.6
In particular, the parent corporation of a tax consolidated corporate group is
deemed to own all the corporate and partnership interests owned by any subsidiary. For example, if two subsidiaries each own 50 percent of a partnership, the
parent corporation is deemed to own 100 percent of the partnership.
Steps in Identifying REP Data
This section outlines the steps we took to identify the reportable entity partners
within the SOI data. The 2007 SOI partnership file is a weighted sample research
file statistically designed to describe the population of all partnerships filing a
Form 1065 in Tax Year 2007 (Processing Year 2008). We assume that the tax characteristics of a REP reported in the 2007 partnership file for a record weighted to
represent more than one partnership in the population represent the tax characteristics of a REP for each of the partnerships in the population represent by that
weighted record.
Partnerships with Reportable Entity Partners
yy We extract all REP EINs reported in the 2007 SOI partnership file.
yy We treat an EIN of 000000000 or 999999999, as well as names listed
without EINs, as reported but not identifiable.
yy We search for the REP EIN in the IRS corporation and partnership
files for forms subject to M-3 (and therefore subject to REP
reporting).
yy The order of the files searched within a year was 1065, 1065-B,
1120-S, 1120, 1120-PC, 1120-L, 1120-C, and 1120-F.
yy Searching stopped when an EIN was identified.
yy The Form 851 file of tax consolidated corporate subsidiaries and
parents was searched if the general search did not identify the
REP tax return to determine if the reported REP EIN is that of a
corporate subsidiary.
yy If the Form 851 file identified the reported EIN for a REP as that of
a subsidiary, the EIN and name of the parent of the subsidiary were
substituted as the actual REP for this research.
yy We extract the REP income-tax-return form type, tax period, total
assets, and industry NAICS code for the identified REP after any
Form 851 file substitutions.
yy We replace the REP’s NAICS code as extracted from the IRS file
with the SOI NAICS code from the 2007 SOI corporation and
partnership files if different. The SOI NAICS code is edited for
consistency across years. The IRS NAICS code is as-filed and is not
subject to any consistency check.
yy If two REPs are reported and identified, we choose the REP with
the largest total assets as the REP for this research.
yy REPs reported but not identified (6,226 cases) include:
▶▶ REPs reported with no EIN or an EIN of 000000000 or
999999999 (955 or about 15 percent of the not identified
cases), or
▶▶ REPs reported with a plausible EIN (5,271 or about 85 percent
of the not identified cases) not identified for this report either
because
»» The EIN is reported with error (we estimate in perhaps
5 percent of the not identified cases) or
»» The EIN based on name inspection is the EIN of an
entity not required to file Schedule M-3, and therefore,
5
6
Boynton and Livingston
not subject to the REP provisions of Schedule M-3
(we estimate in about 80 percent of the not identified
cases). These entities include regulated investment
companies, real estate investment trusts, personal
trusts, and governmental units. We do not extract or
analyze data for these entities for this report because
voluntary reporting is inherently incomplete.
The 2007 Partnership Population: Tables 1–3
In Tables 1 through 3, we report partnership assets, tax income, and book-tax difference for the entire population of 2007 partnerships (3,096,334 partnerships)
and subdivide by whether or not Schedule M-3, is or is not required and is or is not
present. We subdivide partnerships with Schedule M-3 both required and present
(190,711 partnerships), by whether a REP is reported or not, and if reported, by
whether we are able to identify the REP by EIN in the IRS return files for return
forms subject to Schedule M-3. We are particularly interested in the 63,847 partnerships with a REP reported and identified.
In Table 1, the 190,711 partnerships with Schedule M-3 required and present (6.2
percent of 3,096,334) report $18.1 trillion in assets (88.6 percent of $20.4 trillion in
assets reported by all partnerships). The 63,847 partnerships with a REP identified
(2.1 percent of all partnerships) report $4.6 trillion in total partnership assets (22.7
percent of $20.4 trillion for all partnerships). Stated differently, partnerships with
a REP represent one-third of the partnerships with Schedule M-3 required and
present, and report a quarter of the Schedule M-3 partnership population’s assets.
Table 1 indicates that 16,536 partnerships not required to file Schedule M-3 in
fact do so voluntarily rather than file the older Schedule M-1. These voluntary filers are smaller, less complex partnerships with total assets and adjusted total assets
of less than $10 million, total receipts of less than $35 million, and no REP. Such a
large number of voluntary filers among smaller partnerships suggests:
yy Schedule M-3 is not a burden for smaller partnerships;
yy Smaller partnerships or their tax practitioners have access to
accounting computer software packages featuring Schedule
M-3; and
yy An amendment to IRS section 6011(e) to expand mandated
electronic filing by partnerships with $10 million or more in assets
and by certain smaller partnerships filing Schedule M-3 would
probably not impose a burden on the smaller partnerships because,
in general, those smaller partnerships and their tax practitioners
have access to accounting software packages to facilitate
electronic filing.7
7
Partnerships with Reportable Entity Partners
TABLE 1. All Partnerships: Total Returns and Assets by Schedule M-3 Status, 2007
(dollar amounts in millions)
M-3 Status
Returns
Assets $
Returns %
Assets %
Not Required Not Present
2,882,188
1,942,739
93.1%
9.5%
Not Required but Present
16,536
29,828
0.5%
0.1%
Required but Not Present
6,900
354,084
0.2%
1.7%
2,905,623
2,326,651
93.8%
11.4%
120,637
12,231,250
3.9%
60.0%
63,847
4,626,270
2.1%
22.7%
6,226
1,201,964
0.2%
5.9%
190,711
18,059,483
6.2%
88.6%
3,096,334
20,386,134
100.0%
100.0%
Subtotal
Not Required or Not Present
No REP
REP Identified
REP Not Identified
Subtotal
Required and Present
Total All Returns
Note: M-3 is treated as “Not Present” if both book income and tax income are zero.
TABLE 2. All Partnerships: Total Returns and Tax Income by Schedule M-3 Status, 2007
(dollar amounts in millions)
M-3 Status
Negative Tax Income
Returns
Sum
Returns %
Sum %
Not Required Not Present
1,257,279
(98,444)
92.8%
28.5%
Not Required but Present
9,262
(4,955)
0.7%
1.4%
Required but Not Present
3,164
(2,535)
0.2%
0.7%
Subtotal
Not Required or Not Present
1,269,705
(105,934)
93.7%
30.7%
No REP
45,245
(142,073)
3.3%
41.1%
REP Identified
37,104
(90,435)
2.7%
26.2%
2,517
(7,073)
0.2%
2.0%
84,865
(239,581)
6.3%
69.3%
1,354,570
(345,515)
100.0%
100.0%
REP Not Identified
Subtotal
Required and Present
Total All Returns
M-3 Status
Positive Tax Income
Returns
Sum
Returns %
Sum %
Not Required Not Present
1,624,909
326,976
93.3%
17.6%
Not Required but Present
7,274
8,064
0.4%
0.4%
Required but Not Present
3,736
22,166
0.2%
1.2%
Subtotal
Not Required or Not Present
1,635,919
357,206
93.9%
19.2%
No REP
75,392
1,115,849
4.3%
60.1%
REP Identified
26,744
311,756
1.5%
16.8%
3,709
72,308
0.2%
3.9%
105,845
1,499,913
6.1%
80.8%
1,741,764
1,857,119
100.0%
100.0%
REP Not Identified
Subtotal
Required and Present
Total All Returns
Note: M-3 is treated as “Not Present” if both book income and tax income are zero.
Zero tax income returns reported in negative tax income column.
Add return row totals for negative and positive amounts to obtain return row totals in Table 1 (subject to rounding).
8
Boynton and Livingston
TABLE 3. All Partnerships: Total Returns and Book Tax Difference by Schedule M-3
Status, 2007
(dollar amounts in millions)
M-3 Status
Negative Book Tax Difference
Returns
Sum
Not Required Not Present
2,882,188
*
95.5%
Not Required but Present
9,083
(1,195)
0.3%
Required but Not Present
6,900
*
0.2%
Subtotal
Not Required or Not Present
Returns %
Sum %
*
0.3%
*
2,898,170
(1,195)
96.1%
0.3%
No REP
73,135
(301,112)
2.4%
73.9%
REP Identified
41,733
(89,924)
1.4%
22.1%
3,517
(15,241)
0.1%
3.7%
118,385
(406,277)
3.9%
99.7%
3,016,555
(407,472)
100.0%
100.0%
REP Not Identified
Subtotal
Required and Present
Total All Returns
M-3 Status
Positive Book Tax Difference
Returns
Not Required Not Present
0
Not Required but Present
7,453
Required but Not Present
0
Subtotal
Not Required or Not Present
Sum
*
2,717
*
Returns %
0.0%
9.3%
0.0%
Sum %
*
0.8%
*
7,453
2,717
9.3%
0.8%
No REP
47,502
266,643
59.5%
74.9%
REP Identified
22,114
72,790
27.7%
20.5%
2,709
13,654
3.4%
3.8%
72,325
353,087
90.7%
99.2%
79,779
355,805
100.0%
100.0%
REP Not Identified
Subtotal
Required and Present
Total All Returns
Note: M-3 is treated as “Not Present” if both book income and tax income are zero.
Zero Book Tax Difference (BTD) returns reported in negative BTD column.
Add return row totals for negative and positive amounts to obtain return row totals in Table 1 (subject to rounding).
Asterisk (*) in “Sum” column indicates M-3 data not present.
Table 1 also indicates that the 6,900 partnerships required to file Schedule M-3
either did not have a Schedule M-3 present or anomalously filed the form but reported zero for both total book income and for total tax income. We treat any such
anomalous Schedule M-3 as in fact not present. Tax Year 2007 was the second year
for the partnership Form 1065 Schedule M-3. Based on experience with the corporate Form 1120 Schedule M-3, introduced in 2004, we expect better partnership
compliance with the partnership Schedule M-3 in its third and later years (Tax
Years 2008 and later).
Partnerships with Reportable Entity Partners
9
Tables 2 and 3 separately tabulate returns by negative and positive tax income
amounts and book-tax differences amounts. Returns with a zero amount are tabulated with the returns with negative amounts. In Tables 2 and 3, the sum of the row
totals for the number of returns with negative amounts and the number with positive amounts equal the returns totals for the row in Table 1 (subject to rounding).
Table 2 indicates that partnerships with Schedule M-3 required and present
report 69.3 percent of the negative tax income and 80.8 percent of the positive
tax income of all partnerships. Positive tax income of approximately $1.5 trillion
is about six times negative tax income of approximately -$240 billion for these
partnerships.
The 63,847 partnerships with a REP identified report 26.2 percent of the negative tax income and 16.8 percent of the positive tax income of all partnerships.
Positive tax income of approximately $312 billion is about three and one-half times
negative tax income of approximately -$90 billion. Proportionately, partnerships
with a REP have more negative tax income than the Schedule M-3 partnership
population in general.
Table 3 reports Schedule M-3 book-tax difference (BTD) for partnerships. For
Schedule M-3, BTD is tax income minus book income. Negative BTD means
book income exceeds tax income. Only Schedule M-3 BTD is reported in Table
3. BTD reported on Schedule M-1 by partnerships not filing Schedule M-3 is not
included in Table 3. Total negative BTD is -$407 billion and total positive BTD
is $356 billion. These BTD amounts are the same order of magnitude as the BTD
amounts for the corporate population filing the 2005 Form 1120 Schedule M-3
(total negative BTD of -$436 billion and total positive BTD of $421 billion).8 The
partnership BTD amounts are substantial. The 63,847 with a REP identified report
22.1 percent of the negative BTD and 20.5 percent of the positive BTD.
Analysis of the REP Population: Tables 4–9
In discussing Tables 4 through 9, we often refer to partnerships with a REP simply as controlled partnerships. We identify how the distribution of the number
of partnerships and the distribution of partnerships’ assets are affected, as we use
characteristics of the controlling REP to define the columns and the characteristics
of the controlled partnership to define the rows. In all cases, the total number of
partnerships is 63,847, and the total asset amount is $4.6 trillion. In other words,
both the number of partnerships and the amount of assets are fixed in Tables 4
through 9, but we change the criteria along which we partition the data.
The first row of Table 4 reports that 73.0 percent of all controlled partnerships
report less than $10 million in assets and collectively report only 2.3 percent of
the assets reported by controlled partnerships. Reading down the third column
we see that REPs reporting $5 billion or more in assets effectively control 19.5
10
Boynton and Livingston
TABLE 4. Partnerships with REP: Partnership Asset Size by REP Asset Size
Panel A: 2007 Total Returns (% All) and Assets (% All)
REP Asset Size
Partnership
Assets Size
Under $10M
$10M < $250M
$250M and up
Total
Under $10M
$10M < $5B
Returns
Assets
Returns
14.5%
0.4%
3.2%
1.6%
0.2%
17.8%
$5B and up
Total
Assets
Returns
Assets
Returns
Assets
44.7%
1.6%
13.8%
0.3%
73.0%
2.3%
16.6%
10.8%
4.5%
3.7%
24.3%
16.0%
3.3%
1.4%
21.3%
1.2%
57.1%
2.8%
81.7%
5.2%
62.7%
33.7%
19.5%
61.1%
100.0%
100.0%
Panel B: 2007 Total Returns and Assets
(dollar amounts in millions)
REP Asset Size
Partnership
Assets Size
Under $10M
$10M < $5B
Returns
Assets
Returns
Under $10M
9,251
17,957
28,524
75,744
8,817
$10M < $250M
2,018
73,063
10,581
498,585
2,895
119
150,539
902
985,989
11,387
241,559
$250M and up
Total
Assets
$5B and up
Returns
Assets
Total
Returns
Assets
12,382
46,591
106,083
169,875
15,494
741,523
742 2,642,136
1,762 3,778,664
40,006 1,560,318 12,453 2,824,393
63,847 4,626,270
percent of controlled partnerships that collectively report 61.1 percent of the assets
of controlled partnerships. REPs reporting $5 billion or more in assets effectively
control many partnerships with less than $10 million in assets (13.8 percent), but
the smaller number of controlled partnerships with $250 million or more in assets
(1.2 percent) and a REP with $5 billion or more in assets collectively report 57.1
percent of the assets reported by controlled partnerships.
Interestingly, as reported in the first column third row of Table 4, a small number of REPs (119) reporting less than $10 million in assets effectively control partnerships with $250 million or more in assets. Many of these REPs report zero or
negative assets. The IRS is well aware that many corporations and partnerships,
both large and purportedly small, fail to present a proper balance sheet as part of
the tax return.9 Since 2006, the IRS has been working to correct balance sheet
reporting through changes to the tax return instructions.
In Table 5, we group together as “Form 1120” all corporate return types requiring Schedule M-3 other than Form 1120-S for S corporations. Specifically “Form
1120” includes Forms 1120, 1120-C, 1120-F, 1120-L, and 1120-PC. In Table 5, the
bottom line of the second column indicates that REPs filing a corporate Form 1120
effectively control 34.0 percent of controlled partnerships and these partnerships
collectively report 53.7 percent of the assets reported by controlled partnerships.
The first column of Table 5 indicates that 58.3 percent of controlled partnerships
are effectively controlled by partnerships filing Form 1065. In other words, a
11
Partnerships with Reportable Entity Partners
TABLE 5. Partnerships with REP: Partnership Asset Size by REP Return Type
Panel A: 2007 Total Returns (% All) and Assets (% All)
REP Return Type
Partnership
Asset Size
Form 1065
Form 1120
Form 1120S
Total
Returns
Assets
Returns
Assets
Returns
Assets
Returns
Under $10M
41.7%
1.5%
24.9%
0.6%
6.4%
0.2%
73.0%
2.3%
$10M < $250M
15.3%
9.5%
7.7%
5.8%
1.3%
0.8%
24.3%
16.0%
1.3%
32.8%
1.4%
47.3%
0.1%
1.6%
2.8%
81.7%
58.3%
43.7%
34.0%
53.7%
7.8%
2.6%
100.0%
100.0%
$250M and up
Total
Assets
Panel B: 2007 Total Returns and Assets
(dollar amounts in millions)
REP Return Type
Partnership
Asset Size
Form 1065
Returns
Under $10M
$10M < $250M
$250M and up
Total
Form 1120
Assets
Form 1120S
Returns
Assets
Total
Assets
Returns
Returns
Assets
26,607
67,490
15,907
27,908
4,076
9,760
437,277
4,893
267,709
841
10,684
46,591
106,083
36,537
15,494
831 1,517,927
879 2,187,891
52
72,846
1,762 3,778,664
741,523
37,198 2,022,694
21,680 2,483,508
4,969
120,067
63,847 4,626,270
majority of controlled partnerships are controlled by partnerships, but a majority
of controlled partnership assets are controlled by 1120 corporations.
In Table 6, the first column indicates that REPs in the Finance/Holding industry effectively control 29.8 percent of controlled partnerships, and they collectively report 55.8 percent of the assets reported by controlled partnerships. In
Table 6, the third row of the first column indicates that partnerships with $250
million or more in assets and a REP in Finance/Holding are only 1.2 percent of
controlled partnerships but report 49.3 percent of all assets reported by controlled
partnerships.
Table 6 indicates that REPs in Real-Estate/Rental effectively control 44.9 percent of controlled partnerships, but these partnerships collectively only report 11.7
percent of assets reported by controlled partnerships. Stated differently, at the REP
level, Real Estate REPs dominate in terms of numbers of partnerships controlled,
but Finance REPs dominate in terms of dollars of partnership assets controlled.
The assets effectively controlled by REPs in Manufacturing (10.6 percent),
Information (10.4 percent), and other industries (11.5 percent) are comparable to
those effectively controlled by Real-Estate/Rental REPs (11.7 percent). Note that
small controlled partnerships (less than $10 million in assets) with a REP in the
information industry anomalously report collective negative total assets of -$9,304
million. Since 2006, Forms 1065 and 1120 instructions have stated that negative
total assets may not be reported.
12
Boynton and Livingston
TABLE 6. Partnerships with REP: Partnership Asset Size by REP Industry
Panel A. 2007 Total Returns (% All) and Assets (% All)
REP Industry
Partnership
Asset Size
Finance/Holding
Real Estate/Rental
Manufacturing
Returns
Assets
Returns
Assets
Returns
21.3%
0.8%
33.8%
1.2%
2.2%
0.1%
$10M < $250M
7.3%
5.8%
10.6%
5.7%
1.3%
1.1%
$250M and up
1.2%
49.3%
0.4%
4.8%
0.4%
9.4%
29.8%
55.8%
44.9%
11.7%
3.9%
10.6%
Under $10M
Total
Assets
REP Industry
Partnership
Asset Size
Information
Other
Total
Returns
Assets
Returns
Assets
Returns
Assets
Under $10M
1.1%
-0.2%
14.6%
0.4%
73.0%
2.3%
$10M < $250M
0.7%
0.4%
4.3%
3.0%
24.3%
16.0%
$250M and up
0.2%
10.2%
0.5%
8.1%
2.8%
81.7%
Total
1.9%
10.4%
19.5%
11.5%
100.0%
100.0%
Panel B. 2007 Total Returns and Assets
(dollar amounts in millions)
REP Industry
Partnership
Asset Size
Finance/Holding
Returns
Under $10M
$10M < $250M
$250M and up
Total
Assets
Real Estate/Rental
Returns
Assets
Manufacturing
Returns
Assets
13,576
35,747
21,598
55,939
1,423
3,713
4,692
266,628
6,768
264,590
828
50,336
776
2,279,007
286
220,118
240
436,112
19,044
2,581,382
28,652
540,648
2,491
490,161
REP Industry
Partnership
Asset Size
Information
Returns
Assets
Other
Returns
Total
Assets
Returns
Assets
Under $10M
676
(9,304)
9,318
19,987
46,591
$10M < $250M
444
20,114
2,763
139,854
15,494
741,523
$250M and up
122
470,127
338
373,301
1,762
3,778,664
1,242
480,936
12,418
533,143
63,847
4,626,270
Total
106,083
In Table 7, the first row indicates that partnerships in the Finance/Holding industry that have a REP are 15.1 percent of all controlled partnerships with a REP
but collectively report 57.1 percent of assets reported by controlled partnerships.
In particular, the partnerships in the Finance/Holding industry that have a REP
13
Partnerships with Reportable Entity Partners
TABLE 7. Partnerships with REP: Partnership Industry by REP Asset Size
Panel A: 2007 Total Returns (% All) and Assets (% All)
REP Asset Size
Partnership
Industry
Under $10M
$10M < $5B
$5B and up
Total
Returns
Assets
Returns
Assets
Returns
Assets
Returns
Assets
2.1%
3.2%
8.1%
16.0%
5.0%
37.9%
15.1%
57.1%
12.0%
1.2%
36.7%
9.5%
10.3%
4.2%
58.9%
14.9%
Manufacturing
0.1%
0.2%
1.3%
2.1%
0.3%
3.9%
1.8%
6.1%
Information
0.8%
0.1%
0.4%
0.4%
1.2%
9.3%
2.4%
9.8%
Construction
0.6%
0.1%
5.1%
0.8%
0.3%
0.7%
6.0%
1.5%
Retail/Wholesale
0.2%
0.1%
1.5%
0.8%
0.2%
0.9%
2.0%
1.8%
All Other
2.0%
0.4%
9.6%
4.1%
2.3%
4.2%
13.9%
8.7%
17.8%
5.2%
62.7%
33.7%
19.5%
61.1%
100.0%
100.0%
Finance/Holding
Real Estate/Rental
Total
Panel B: 2007 Total returns and Assets
(dollar amounts in millions)
REP Asset Size
Partnership
Industry
Under $10M
$10M < $5B
Assets
Finance/Holding
1,329
146,969
5,149
741,385
3,162 1,753,389
Real Estate/Rental
7,631
56,161
23,428
439,398
6,548
192,090
37,607
687,649
91
7,413
820
96,884
210
179,435
1,121
283,733
Information
508
4,107
264
18,879
752
431,325
1,524
454,311
Construction
370
4,470
3,245
36,083
205
30,210
3,820
70,763
Retail/Wholesale
155
2,994
958
38,588
138
42,838
1,251
84,420
1,304
19,444
6,142
189,100
1,439
195,106
8,885
403,650
11,388
241,559
All Other
Total
Assets
40,006 1,560,318
Returns
Assets
Total
Returns
Manufacturing
Returns
$5B and up
12,453 2,824,393
Returns
Assets
9,639 2,641,743
63,847 4,626,270
with $5 billion or more in assets (5.0 percent) collectively report 37.9 percent of assets reported by controlled partnerships. In Table 7, the second row indicates that
partnership in the Real-Estate/Rental industry that have a REP are 58.9 percent
of all partnerships with a REP but collectively report only 14.9 percent of assets
reported by partnerships with a REP. At the controlled-partnership level, just as at
the controlling-REP level, Real Estate dominates in terms of numbers, but Finance
dominates in terms of dollars.
Partnerships in the Real-Estate industry with a REP generally have a REP with
$10 million to $5 billion in assets. Such partnerships (36.7 percent) report 9.5
percent of assets reported by partnerships with a REP. Partnerships in the RealEstate industry account for about two-thirds of the controlled partnerships with a
REP reporting under $10 million in assets (12.0 percent compared to 17.8 percent).
14
Boynton and Livingston
TABLE 8. Partnerships with REP: Partnership Industry by REP Return Type
Panel A: 2007 Total Returns (% All) and Assets (% All)
REP Return Type
Partnership
Industry
Form 1065
Form 1120
Form 1120S
Total
Returns Assets Returns Assets Returns Assets Returns Assets
Finance/Holding
9.2%
28.7%
5.6%
27.1%
0.3%
1.3%
15.1%
57.1%
39.7%
10.2%
15.5%
4.3%
3.7%
0.3%
58.9%
14.9%
Manufacturing
0.4%
1.0%
0.9%
4.9%
0.4%
0.2%
1.8%
6.1%
Information
0.8%
0.2%
1.5%
9.5%
0.1%
0.1%
2.4%
9.8%
Construction
3.3%
0.4%
1.6%
0.9%
1.0%
0.2%
6.0%
1.5%
Retail/Wholesale
0.4%
0.6%
0.9%
1.0%
0.6%
0.2%
2.0%
1.8%
All Other
4.4%
2.6%
8.0%
5.9%
1.5%
0.2%
13.9%
8.7%
58.3%
43.7%
34.0%
53.7%
7.8%
2.6%
100.0%
100.0%
Real Estate/Rental
Total
Panel B: 2007 Total Returns and Assets
(dollar amounts in millions)
REP Return Type
Partnership
Industry
Form 1065
Form 1120
Form 1120S
Total
Returns Assets Returns Assets Returns Assets Returns Assets
Finance/Holding
5,856 1,326,763
Real Estate/Rental 25,371
212
61,584
473,097
3,572 1,253,396
9,878
199,871
2,358
14,681
37,607
9,639 2,641,743
687,649
Manufacturing
254
45,578
590
228,480
277
9,675
1,121
283,733
Information
497
10,796
949
438,518
78
4,997
1,524
454,311
2,137
20,232
1,019
41,794
664
8,736
3,820
70,763
286
28,211
571
46,610
393
9,599
1,251
84,420
2,799
118,017
5,100
274,839
986
10,795
8,885
403,650
37,198 2,022,694 21,680 2,483,508
4,969
120,067
Construction
Retail/Wholesale
All Other
Total
63,847 4,626,270
Controlled partnerships in the Information industry generally have a REP with
$5 billion or more in assets. Such partnerships (1.2 percent) report 9.3 percent of
assets reported by controlled partnerships.
In Table 8, the first row shows that partnerships in the Finance/Holding industry that have a REP that is a Form 1065 partnership (9.2 percent) report 28.7
percent of assets reported by controlled partnerships, while those with a Form 1120
corporation as a REP (5.6 percent) report 27.1 percent.
In Table 9, we see that REPs, in general, stay close to their own industry
in terms of the industries of the partnerships they control. The exception is
Finance/Holding, which seems to be comfortable in controlling partnerships in
15
Partnerships with Reportable Entity Partners
TABLE 9. Partnerships with REP: Partnership Industry by REP Industry
Panel A: 2007 Total Returns (% All) and Assets (% All)
REP Industry
Partnership
Industry
Finance/Holding
Real Estate/Rental
Returns
Assets
Finance/Holding
12.7%
50.9%
0.7%
0.4%
0.4%
2.9%
Real Estate/Rental
13.4%
2.3%
40.0%
10.6%
1.1%
0.7%
Manufacturing
0.2%
0.8%
*
*
1.1%
4.6%
Information
0.8%
0.2%
*
*
Construction
0.7%
0.1%
2.0%
0.3%
Retail/Wholesale
0.4%
0.2%
0.1%
0.0%
0.4%
All Other
Total
Returns
Assets
Manufacturing
Returns
Assets
*
*
*
*
0.3%
1.6%
1.4%
2.1%
0.4%
0.8%
1.7%
29.8%
55.8%
44.9%
11.7%
3.9%
10.6%
REP Industry
Partnership
Industry
Information
Returns
Returns
Assets
0.0%
0.5%
1.4%
2.3%
15.1%
57.1%
Real Estate/Rental
0.1%
0.3%
4.3%
0.9%
58.9%
14.9%
0.3%
0.7%
1.8%
6.1%
0.0%
0.2%
2.4%
9.8%
3.3%
1.2%
6.0%
1.5%
1.8%
Information
Construction
*
1.6%
*
*
9.2%
*
Returns
Total
Finance/Holding
Manufacturing
Assets
Other
Assets
Retail/Wholesale
0.0%
0.0%
1.0%
1.3%
2.0%
All Other
0.2%
0.3%
9.2%
5.0%
13.9%
8.7%
Total
1.9%
10.4%
19.5%
11.5%
100.0%
100.0%
Note: Asterisk (*) indicates data suppressed to preserve taxpayer confidentiality. One or more of the suppressed cells
involves a return count of 1 or 2. Table 9 Panel B appears on the next page.
all industries with a preference for controlling Finance and Real Estate partnerships. Partnerships in the Finance/Holding industry with a REP in the Finance/
Holding industry (12.7 percent) report 50.9 percent of assets reported by controlled partnerships. Partnerships in the Real-Estate/Rental industry with a REP
in the Real-Estate/Rental industry (40.0 percent) report 10.6 percent of assets
reported by controlled partnerships.
Paraphrasing our comment on Table 6, in Table 9, Real Estate REPs controlling
Real Estate partnerships dominate in terms of numbers of partnerships controlled,
but Finance REPs controlling Finance partnerships dominate in terms of dollars of
partnership assets controlled.
16
Boynton and Livingston
TABLE 9. Partnerships with REP: Partnership Industry by REP Industry— continued
Panel B: 2007 Total Returns and Assets
(dollar amounts in millions)
REP Industry
Partnership
Industry
Finance/Holding
Real Estate/Rental
Manufacturing
Returns
Assets
Returns
Assets
Finance/Holding
8,091
2,355,080
419
17,748
224
Real Estate/Rental
8,533
107,545
25,531
490,433
709
31,210
Manufacturing
152
37,479
*
*
717
214,722
Information
493
7,015
*
*
*
*
Construction
449
2,969
1,265
12,183
*
*
Retail/Wholesale
275
7,830
43
343
286
16,117
All Other
Total
Returns
Assets
135,990
1,050
63,464
1,343
18,252
521
78,129
19,044
2,581,382
28,652
540,648
2,491
490,161
REP Industry
Partnership
Industry
Information
Returns
Other
Assets
Returns
Total
Assets
Returns
Assets
Finance/Holding
26
25,074
880
107,851
9,639
2,641,743
Real Estate/Rental
89
14,860
2,745
43,601
37,607
687,649
*
*
200
30,426
1,121
283,733
425,921
18
7,890
1,524
454,311
2,087
53,525
3,820
70,763
Manufacturing
Information
991
Construction
*
*
Retail/Wholesale
7
834
640
59,296
1,251
84,420
121
13,251
5,850
230,555
8,885
403,650
1,242
480,936
12,418
533,143
63,847
4,626,270
All Other
Total
Note: Asterisk (*) indicates data suppressed to preserve taxpayer confidentiality. One or more of the suppressed cells
involves a return count of 1 or 2. Table 9 Panel A appears on the prior page.
Closing Observations
At the level of the REP, the general story about effective control of controlled partnership assets is the importance of:
yy Large REPs ($5 billion or more in assets),
yy REPs that are 1120 corporations, and
yy REPs that are in Finance/Holding.
At the controlled partnership level, the story about controlled partnership assets is the importance of:
yy Large controlled partnerships ($250 million or more in assets), and
yy Controlled partnerships that are in Finance/Holding.
Partnerships with Reportable Entity Partners
17
We look forward to doing further research into effectively controlled partnerships. When we move forward to Tax Year 2008 data, we will have a broad set
of ownership data added to the Form 1065 and Form 1120 tax returns including
information on entity owners that are U.S. or foreign partnerships, corporations,
or trusts. That set of controlling entity owners is far broader than the Schedule
M-3 REP population.
Acknowledgements
Charles E. Boynton is a Program Manager and Senior Program Analyst, Office
of Research and Workload Identification (RWI), Large and Mid-Size Business
(LMSB) Division, IRS. From 2000 to 2006, he was a Stanley S. Surrey Senior
Research Fellow at the Treasury’s Office of Tax Analysis (OTA). Before joining
Treasury, he taught tax accounting at the University of Wisconsin and tax and
financial accounting at the University of North Texas.
Barbara A. Livingston is an IRS Senior Operations Research Analyst with
the Office of Research and Workload Identification (RWI), Large and MidSize Business (LMSB) Division, IRS. She has held positions with the Statistics
of Income (SOI) Division (1987–1997), and with the Office of Pre-Filing and
Technical Guidance (PFTG), Large and Mid-Size Business (LMSB) Division
(1998–2008), IRS.
The authors thank the management of the Statistics of Income (SOI) Division,
IRS, and of the Office of Research and Workload Identification (RWI), Large
and Mid-Size Business (LMSB) Division, IRS, for permission to use the 2007
SOI partnership data file and to conduct this study. We thank the organizers of
the 2010 IRS Research Conference for including the study as part of the panel
“Compliance of Large Entities.”
We particularly wish to thank for comments, assistance, and support:
Kevin Cecco, Joseph Koshansky, Kenneth Szeflinski, Tim Wheeler, and Nina
Shumofsky at SOI; Elizabeth Kruse, Alan Plumley, and Melissa Kovalick at the
2010 IRS Research Conference; Drew Lyon at PricewaterhouseCoopers, our
discussant at the Conference; Don McPartland, John Davidson, Jim Clarkson,
Darla Riggan, Fredericka Bunting, John Miller, Donald Lee, Lois Petzing, Ellen
Legel, Bill Wilson, Mark Silva, Cary Russ, and Robert Adams at LMSB; and Petro
Lisowsky at the University of Illinois-Urbana-Champaign.
DISCLAIMER: The opinions expressed are those of the authors and do not
necessarily reflect positions of the IRS or U.S. Department of the Treasury.
18
Boynton and Livingston
References
Banoff, Shelton (2009a), “FAQ-Filled Guidance on Computing A Partner’s
Interest in Profits, Losses, and Capital—Part 1,” Journal of Taxation.
Banoff, Shelton (2009b), “FAQ-Filled Guidance on Computing A Partner’s
Interest in Profits, Losses, and Capital—Part 2,” Journal of Taxation.
Blouin, Jennifer; Jason DeBacker; and Stephanie Sikes (2010), “Temporary and
Permanent Book-Tax Differences: Complements or Substitutes?” University of
Pennsylvania working paper.
Boynton, Charles; Portia DeFilippes; and Ellen Legel (2006), “A First Look at
2004 Schedule M-3 Reporting by Large Corporations,” Tax Notes, September
11, p. 943. Available at http://www.irs.gov/businesses/corporations/
article/0,,id=163246,00.html.
Boynton, Charles; Portia DeFilippes; and Ellen Legel (2008), “A First Look
at 2005 Schedule M-3 Corporate Reporting,” Tax Notes, November
3, p. 563. Available at http://www.irs.gov/businesses/corporations/
article/0,,id=163246,00.html.
Boynton, Charles; Portia DeFilippes; Petro Lisowsky; and Lillian Mills (2004),
“Consolidation Anomalies in Form 1120 Corporation Tax Return Data,” Tax
Notes, July 26, p. 405.
Boynton, Charles and Lillian Mills (2004), “The Evolving Schedule M-3: A New
Era of Corporate Show and Tell?” National Tax Journal 57, no. 3:757–772.
Boynton, Charles and William Wilson (2006), “A Review of Schedule M-3, The
Internal Revenue Service’s New Book-Tax Reconciliation Tool,” Petroleum
Accounting and Financial Management Journal 25, no. 1, 1–6. Available at
http://www.irs.gov/businesses/corporations/article/0,,id=163246,00.html
Dunbar, Amy; John Phillips; and George Plesko (2009), “The Effect of FIN 48 on
Firms’ Tax Reporting Behavior,” University of Connecticut working paper.
Everett, John; Cherie Hennig; and William Raabe (2007), “The Schedule M-3
Compliance Maze: Unanswered Questions,” Tax Notes, March 19, p. 1103.
Hennig, Cherie; John Everett; and William Raabe (2009), “Reportable Entity
Partner Attribution Rules Lack Authority,” Tax Notes, April 6, p. 86.
Levy, David and Matthew J. Hofheimer (2010), “Bankrupt Partnerships and
Disregarded Entities,” Tax Notes, June 7, p1103.
Lisowsky, Petro (2009), “Inferring U.S. Tax Liability from Financial Statement
Information,” Journal of the American Taxation Association, Volume 31,
Number 1, pp. 29–63.
Partnerships with Reportable Entity Partners
19
Lisowsky, Petro; Leslie Robinson; and Andrew Schmidt (2010), “Do Publicly
Disclosed Tax Reserves Tell Us About Privately Disclosed Tax Shelters?”
University of Illinois at Urbana-Champaign working paper.
Mills, Lillian; and George Plesko (2003), “Bridging the Gap: A Proposal for More
Informative Reconciling of Book and Tax Income,” National Tax Journal 56,
no. 4: 865–893.
Petska, Tom; Michael Parisi; Kelley Luttrell; Lucy Davitian; and Matt Scoffice
(2005), “An Analysis of Business Organizational Structure and Activity from
Tax Data,” Special Studies in Federal Tax Statistics: 9–44. Available at http://
www.irs.gov/pub/irs-soi/05petska.pdf.
U.S. Department of the Treasury (2010), General Explanation of the
Administration’s Fiscal Year 2011 Revenue Proposals, Washington, D.C.:
Government Printing Office, February: 103. Available at http://www.treas.gov/
offices/tax-policy/library/greenbk10.pdf.
Weiner, Joann M. (2007), “Closing the Other Tax Gap: The Book-Tax Income
Gap,” Tax Notes, May 29, p. 849.
Wheeler, Tim and Nina Shumofsky (2009), “Partnership Returns, 2007,” Statistics
of Income Bulletin, Volume 29, Number 2: 70–159. Available at http://www.irs.
gov/pub/irs-soi/09fallbulpartret.pdf.
Endnotes
1
This paper was first published by Tax Notes 128 No. 9 (August 30, 2010):949–
958. It is reprinted with permission. It was prepared for the 2010 IRS Research
Conference. The opinions expressed are those of the authors and do not
necessarily reflect positions of the IRS or U.S. Department of the Treasury.
2 For an excellent review of the 2007 partnership population, see Wheeler and
Shumofsky (2009). For a study of the growth in partnership business receipts
from 1980 to 2002, see Petska, Parisi, Luttrell, Davitian, and Scoffice (2005).
For a recent study of bankruptcy risks for individual and corporate partners
selecting to do business in the increasingly popular limited liability company
(LLC) form, see Levy and Hofheimer (2010).
3 For discussions relating to the development of Schedule M-3, see Mills and
Plesko (2003), Boynton and Mills (2004), Boynton and Wilson (2006), and
Boynton, DeFilippes, and Legel (2006 and 2008). For a summary of research
on book-tax differences and Schedules M-1 and M-3 through 2007, see Weiner
(2007). For a discussion of the relationship between financial accounting
current Federal income tax expense on SEC Form 10K (and now on Schedule
M-3) and Form 1120 tax liability, see Lisowsky (2009). Research using
20
Boynton and Livingston
Schedule M-3 data has developed further as data have become available. For
example, Lisowsky, Robinson, and Schmidt (2010) discuss the relation between
uncertain tax positions, tax shelters, and reportable transaction amounts
reported on Schedule M-3; Dunbar, Phillips, and Plesko (2009) examine public
versus private firms’ book-tax reporting and tax planning before and after rules
were passed for more public disclosures of tax reserves; and Blouin, DeBacker,
and Sikes (2010) examine the relation between temporary and permanent
book-tax differences on Schedule M-3 for public versus private firms.
4 For technical details on Schedule M-3 filing requirements, including the
definition of adjusted total assets and Reportable Entity Partner, see the
current instructions for Forms 1065 and 1120 Schedules M-3. Go to w
ww.irs.
gov, click on Forms and Publications, click on Form and instruction number
(PDF), insert 1065 or 1120, and click on Schedule M-3 instructions.
5 For comments strongly opposing the Schedule M-3 Reportable Entity Partner
indirect ownership attribution rules, see Hennig, Everett, and Raabe (2009).
For an earlier discussion, see Everett, Hennig and Raabe (2007), Questions 32
through 36.
6 IRC section 267(c) provides an alternative attribution model for corporate
stock ownership based on proportional allocation. IRC section 707(b) makes
section 267(c) applicable to attribution of interest in partnership profits, loss,
and capital. In 2008, partnership tax return Form 1065 Schedule B ownership
questions 3 and 4 and corporation tax return Form 1120 Schedule K ownership
questions 4 and 5 were added and use section 267(c) attribution but with a
limit on family attribution among individuals. See the current instructions
for Form 1065 Schedule B and Form 1120 Schedule K. For Frequently Asked
Questions (FAQs) on the 2008 ownership questions, go to www.irs.gov, click
on businesses, click on partnerships, click on 2008 Changes to Form 1065—
Frequently Asked Questions. For comments on the 2008 ownership questions
and the web-based FAQs, see Banoff (2009a, 2009b). Banoff (2009b) in text
before and following his footnote 64 comments on Hennig, Everett, and Raabe
(2009) cited in our prior note.
7 For a proposal to require electronic tax return filing of all corporations and
partnerships filing Schedule M-3, see U.S. Department of the Treasury (2010)
page 103.
8 See Boynton, DeFilippes, and Legel (2008).
9 For a discussion of balance sheet and other consolidation anomalies of large
corporations, see Boynton, DeFilippes, Lisowsky, and Mills (2004).
Temporary and Permanent
Book-Tax Differences:
Complements or Substitutes?
Jennifer Blouin, University of Pennsylvania; Jason DeBacker, U.S. Department
of Treasury; Stephanie Sikes, University of Pennsylvania
T
axable income reported to the tax authorities almost always differs from
book income reported to the capital markets. Such differences arise
from tax legislation that mandates departures from Generally Accepted
Accounting Principles (GAAP) for various economic, political, and administrative
reasons. There are two types of book-tax differences: permanent and temporary.
Permanent book-tax differences arise when an item (1) affects taxable income, but
never affects book income, or (2) affects book income, but never affects taxable income. Examples include municipal bond income, fines, and meals and entertainment. Temporary book-tax differences arise when book and tax treatment for a
transaction differ in a particular year but have the same cumulative effect over the
life of the firm (ignoring issues related to the time value of money). Depreciation,
bad debt expenses, and loss contingencies are some of the more common temporary book-tax differences.
In recent years, researchers have exerted considerable effort to determining
the cause of book-tax differences (e.g., are they caused by aggressive reporting
of book income, aggressive reporting of taxable income, or both).1 Using data
from firms’ Schedule M-3 for years 2005–2007, we examine whether firms treat
permanent and temporary book-tax differences as complements or as substitutes,
and whether capital markets incentives affect this trade-off. Our study has two
advantages over prior studies of book-tax differences that rely on publicly available information. First, rather than estimating a firm’s temporary and permanent
book-tax differences, we incorporate the actual amounts of a firm’s temporary and
permanent book-tax differences as reported on its Schedule M-3. Second, there
are no publicly available data for private firms. Although all companies have incentives to reduce taxable income, public firms face much greater capital markets
pressure to report high book income. Therefore, the non-tax costs of tax planning
could lead public companies to have different preferences for permanent and timing book-tax differences. Incorporating private firms into our analysis allows us
to attribute any differences in the use of permanent and temporary book-tax differences between public and private firms to the capital markets pressures faced
by public firms.
20
Blouin, DeBacker, and Sikes
We cannot predict ex ante whether public firms will view temporary and permanent differences as substitutes or as complements. Some (e.g., Weisbach 2002;
Plesko 2004; Neubig 2006) conjecture that public companies facing pressure to
report high GAAP earnings prefer tax planning that creates permanent booktax differences because permanent book-tax differences decrease taxable income
without reducing book income. Larger firms also have more resources available
for tax planning (e.g., Rego 2003). For both of these reasons, we expect that public
firms might view the two as substitutes and only rely on temporary differences
when they have exhausted available permanent book-tax differences.
Permanent differences could create financial statement risk. Only permanent
book-tax differences have an impact on a firm’s effective tax rate and thus on reported net income. However, if tax authorities later overturn a permanent booktax difference, then income tax expense will increase, and GAAP net income will
decrease. Because of this financial statement risk, public firms facing heightened
capital market pressure (i.e., to maintain a series of increasing earnings (Barth et
al. 1999)) might prefer temporary differences over permanent differences.
Our paper contributes to prior literature that compares and contrasts aggressive financial reporting and aggressive tax reporting between public and private
firms (Cloyd et al. 1996; among others). Based on the results of Cloyd et al. (1996),
we predict that private firms will have smaller book-tax differences (scaled by assets) than public firms because private firms are more likely to choose conforming
transactions that reduce taxable income and book income. Cloyd et al. (1996)
find that executives of private firms choose conforming transactions that decrease
both taxable income and reported book income in order to save on taxes and to
increase the probability of successfully defending the transaction if challenged by
the Internal Revenue Service (IRS) (see Mills 1998). To the extent that private
firms have book-tax differences, we have little expectation that they would prefer
permanent to temporary differences, because private firms are likely less sensitive
than public firms to the financial statement benefits associated with permanent
book-tax differences. Since private firms rely on debt for their external financing needs, private firms may be more focused on cash flows (generated by both
temporary and permanent differences) than net income. However, if private firm
managers are compensated based on net income, then private firms may prefer
permanent to temporary differences.
We also incorporate a set of firms that are a public-private hybrid: quasi-public
firms. We define quasi-public firms as those firms with public debt and private
equity. Using a similar sample, Badertscher et al. (2010) finds that private firms
owned by private equity firms (“PE-backed firms”) pay 14.2 percent less income tax
per dollar of pre-tax income than private firms without private equity firm ownership and that PE-backed firms have lower marginal tax rates and participate in
more non-conforming transactions than do non-PE-backed private firms. Based
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
21
on these findings, we anticipate that quasi-public firms have levels of permanent
book-tax differences similar to those of public firms.
The results of univariate analysis are largely consistent with our expectations.
We find that public firms have significantly more total book-tax differences scaled
by total assets, temporary differences scaled by total assets, and permanent differences scaled by total assets than do private firms. These results suggest that
public firms are either a) more aggressive tax planners or b) undergo relatively less
conforming tax planning. Surprisingly, the quasi-public sample appears to have
income increasing book-tax differences. However, we are hesitant to draw inferences from this analysis, since there are so few quasi-public firms in our sample.
Next, we examine whether firms treat permanent and temporary book-tax
differences as substitutes or as complements and if the treatment varies between
public and private firms.2 We do not have any ex ante predictions for this analysis.
Overall, we find that firms treat temporary and permanent book-tax differences
as substitutes. When a firm decreases (increases) its temporary book-tax differences from one year to the next, it increases (decreases) its permanent book-tax
differences. However, when we investigate whether the substitution effect varies
by ownership type, we find little evidence that public firms have a different rate of
substitution than private firms.
The remainder of the paper proceeds as follows. Section 2 discusses the motivation for the Schedule M-3 and prior literature related to tax reporting practices
of public, private, and quasi-public firms. Section 3 outlines our empirical design.
Section 4 summarizes the results. Section 5 concludes.
Background and Hypothesis Development
During the 1990s, the growing disparity between income reported for financial
statements and income reported to tax authorities caught the attention of legislators and regulators. The growing divide between financial income and taxable income may have reinforced instances of outrageous corporate misbehavior to help
motivate regulatory legislation including Sarbanes-Oxley and tax-shelter restrictions. A Treasury report in 1999, as well as testimony in 2000 by then Treasury
Assistant Secretary for Tax Policy Jonathan Talisman, highlighted the growing divide between pre-tax book income and taxable income and expressed concerns
that tax-sheltering activity could be responsible for the divide (see Treasury (1999)
and Talisman (2000)). At the time, the Schedule M-1 on a corporation’s tax return
reconciled the firm’s pre-tax book income with its taxable income. Many felt that
the information provided on the Schedule M-1 was inadequate to address whether
the growing divide was due to tax planning or financial reporting aggressiveness.
As a result, the Treasury created a new schedule for large corporate filers. For tax
22
Blouin, DeBacker, and Sikes
years 2004 forward, Schedule M-3 replaces Schedule M-1 for corporate tax returns
reporting total assets of $10 million or more.3
Recent research in accounting has focused on corporate tax avoidance and
earnings management as potential sources for the book-tax gap (Plesko 2002;
Manzon and Plesko 2002; Desai 2003; McGill and Outslay 2004; Seidman 2010).
Some link the growing divide between tax and financial reporting to aggressive
tax planning and/or tax shelter use (Mills 1998; Desai 2003), while others link
differences between tax and financial reporting to more aggressive earnings management (Phillips et al. 2003; Phillips et al. 2004; Hanlon 2005). Of course, some
firms are aggressive with their financial reporting as well as their tax reporting
(Frank et al. 2009). For instance, Wilson (2009) finds that firms that are actively
engaged in tax shelters have larger ex post book-tax differences and are more aggressive in their financial reporting. Although a consensus has not been reached
as to whether the growth in the book-tax gap is primarily driven by tax planning
or earnings management, there is strong evidence to suggest that the divide between pre-tax book income and taxable income is due to both tax sheltering and
aggressive financial reporting.
The goal of the Schedule M-3 is greater transparency, and, as such, it provides
much more granular information related to book-tax differences than does the
Schedule M-1. Moreover, unlike the Schedule M-1, the Schedule M-3 separates
total book-tax differences into temporary book-tax differences and permanent
book-tax differences. Permanent book-tax differences arise when an item (1) affects taxable income, but never affects book income, or (2) affects book income,
but never affects taxable income. Temporary book-tax differences arise when
book and tax treatment for a transaction differ in a particular year but have the
same cumulative effect over the life of the firm (ignoring issues related to the time
value of money). In this paper, we examine whether firms treat permanent and
temporary book-tax differences as substitutes or as complements and whether this
varies according to a firm’s ownership structure. In particular, we examine the
variation between publicly traded firms, quasi-public firms (i.e., firms with publicly held debt), and private firms.
We have no ex ante prediction about whether firms will view timing and permanent book-tax differences as complements or as substitutes. If firms tax plan
to the fullest extent possible, then they might have high levels of permanent and
temporary book-tax differences. However, if firms are constrained from undertaking all tax planning opportunities because a) they view extreme tax planning as
too risky or b) they have limited firm resources to invest in tax planning, then they
might have to choose between permanent and temporary planning opportunities.
All else equal, we expect firms to prefer permanent book-tax difference because
of the potential financial statement benefits. However, temporary differences provide cash flow benefits with little financial statement risk.
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
23
Because of the lack of publicly available data on private firms, we know little
about the tax reporting behaviors of private firms with the exception of the findings in a handful of papers. In one of the earliest studies of book-tax differences,
Mills (1998) finds that IRS audit adjustments increase as book-tax differences increase. Her results support the notion that firms face a tradeoff between book and
tax incentives for earnings management. Moreover, using data from firms’ tax
returns, she predicts that public firms will be less aggressive in transactions where
book and tax treatment conform because it is more costly for public firms than
for private firms to report lower book income for tax savings purposes. Thus, she
predicts that as long as IRS detection is equivalent for public and private firms,
then public firms should have smaller IRS adjustments. However, she only finds
evidence that this is so in one of the eight years of her 1982-1989 sample period.
She also finds that audit adjustments as a percentage of book-tax differences are
smaller for public firms than for private firms, and this difference is significant for
three of the years of her sample period.
Cloyd et al. (1996) conducts a survey of public and private firms. The authors
study whether public firms are less likely than private firms to partake in conforming tax planning transactions (i.e., transactions that reduce both book income and
taxable income) as a result of public firms facing greater non-tax costs associated
with income-decreasing tax planning. Of the 1,920 surveys that they mailed, they
received a 32-percent response rate for public and private firms (423 public firms
responded, and 172 private firms responded). They find that conformity is more
likely when it increases the probability of successfully defending the aggressive tax
position upon audit and that managers of public firms are less likely to conform
than are private firm managers. Moreover, they find that public firm managers
perceive conformity to have higher non-tax costs. Examples of non-tax costs associated with downward conforming transactions are debt covenant violations,
lower management compensation, and lost promotions when compensation and
performance are tied to reported income. In addition, managers could perceive
that lower reported income could lead to lower market value if investors are fixated on earnings (Dietrich 1984; Hand 1990; Chen and Schoderbek 2000).4
Several recent papers examine the relation between the incentives of managers
of public firms to increase reported book income and a firm’s tax planning behavior. In the 1990s, some firms began to evaluate their tax departments as profit centers, or “contributors to the bottom line.” Using confidential survey taken in 1999
of Chief Financial Officers of Fortune 500 companies, Robinson et al. (2010) find
that firms that evaluate their tax departments as profit centers have significantly
lower effective tax rates than do firms that evaluate their tax departments as cost
centers. Similarly, using a proprietary data set with detailed executive compensation, Blouin et al. (2010) find a significant negative relation between the incentives
of tax directors and firms’ effective tax rates. Dyreng et al. (2010) find that individual executives play a significant role in determining the level of tax avoidance that
24
Blouin, DeBacker, and Sikes
their firms undertake and can have a significant impact on their firms’ effective tax
rates. Consistent with findings in these papers, in particular the incentives given
to tax departments examined in Robinson et al. (2010) and Blouin et al. (2010), we
expect for public firms to have large amounts of permanent book-tax differences
since only permanent book-tax differences can reduce a firm’s effective tax rate,
thereby increasing its net income.
We include quasi-public firms in our analysis because a recent paper
(Badertscher et al. 2010) finds that private firms that are owned by private equity
firms (“PE-backed firms”) are more tax aggressive than are private firms that are
not owned by private equity firms. Badertscher et al. (2010) find that PE-backed
firms pay 14.2 percent less income tax per dollar of pre-tax income than non-PE
backed firms. They also find that PE-backed firms have lower marginal tax rates
and participate in more non-conforming transactions than do other private firms.
The fact that PE-backed firms participate in more non-conforming transactions
suggests that PE-backed firms have larger book-tax differences than do non-PE
backed private firms. We will examine whether this is indeed true. Badertscher
et al. (2010) conclude that private equity firms view tax avoidance as a source of
economic value, the benefits of which exceed any potential reputational costs associated with corporate tax avoidance.
Based on the findings in the literature above, we expect public firms to have
larger book-tax differences than private firms. Public firms and private firms both
have an incentive to report lower taxable income; however, there are more non-tax
costs associated with downward conforming transactions for public firms than
for private firms. Based on the results in Robinson et al. (2010) and Blouin et al.
(2010), we expect that public firms will want to engage in permanent book-tax
differences because only permanent book-tax differences can reduce a firm’s effective tax rate. However, if we find that public firms actually have more temporary
differences than permanent differences, we will attribute such a finding to the fact
that there is some financial statement risk associated with permanent book-tax
differences (i.e., if the IRS later overturns permanent book-tax differences, a firm
has to increase income tax expense, thereby increasing its effective tax rate and
decreasing reported net income). Based on the finding in Badertscher et al. (2010)
that PE-backed private firms engage in more nonconforming transactions than do
non-PE-backed private firms, we expect our sample of quasi-public firms to have
larger amounts of book-tax differences than our sample of private firms.
Sample and Research Design
Sample
Our sample consists of 2,799 public firms, 21,445 private firms, and 23 quasi-
public firms from 2005–2007. For a firm to be included in our sample, it must
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
25
be a C corporation, report total assets of at least $10 million, and have positive
book income. We also exclude firms whose foreign net income is more than 25
percent of their worldwide net income.5 Although multinational tax planning is an
interesting topic, it is difficult to disentangle whether the deferral of foreign income
from taxation until repatriation represents a permanent or temporary difference for
the purposes of our study. Note that Louie (2005) describes that deferral of foreign
income is reported as a timing book-tax difference on the Schedule M-3. However,
under APB 23, the deferral of foreign income does not create a deferred tax liability
for reporting purposes, which would decrease firms’ GAAP effective tax rates.
Unlike Badertscher et al. (2010), we do not know which private firms in our
sample have a private equity firm as either a majority or minority owner. Thus, we
identify our public and quasi-public samples by merging the IRS and Compustat
data using the employer identification number (EIN). If a firm is found in both datasets, we then look to see whether the firm has reported a stock price (Compustat
variable prcc_f) in any of the past 5 years. If price (no price) information is available, then we designate the firm as public (quasi-public). Like the PE-backed firms
in Badertscher et al’s (2010) sample, the quasi-public firms in our sample have
publicly-traded debt. Our sample of private firms consists of firms for which we
have a tax return but that are not included in Compustat.6
Table 1 outlines the sample derivation. As expected, there are far more private
firms than public and quasi-public firms in our sample. However, the representation across industries is similar. The elimination of firms with less than $10 millions of assets disproportionately affects the private sample. Also, the public firms
included in our analysis constitute approximately 15 percent of the assets of all
Compustat firms in our sample period.
Research Design
For each of the three groups (public, quasi-public, and private), we estimate their
total book-tax differences, permanent book-tax differences, and temporary booktax differences using information from the Schedule M-3. We then divide their
book-tax differences into quintiles with quintile 5 representing the most aggressive
book-tax differences (i.e., greatest difference between pre-tax book income and
taxable income, where pre-tax book income exceeds taxable income). Quintile 1
represents the least aggressive differences. In univariate analyses, we test whether
there are significant differences in the mean scaled total, permanent, and temporary book-tax differences across the public, quasi-public, and private firms. We
also test whether the means for quintiles 1 and 5 are significantly different across
the three groups.
For each of the three ownership groups, we also examine the percentage of
total, permanent, and temporary book-tax differences that are associated with
“reportable transactions” under Treasury Regulation Section 1.6011–4(b). There
26
Blouin, DeBacker, and Sikes
are six categories of reportable transactions over our sample period. They are
listed transactions, confidential transactions, transactions with contractual protection, loss transactions, transactions with a significant book-tax difference, and
transactions involving a brief asset holding period.7 Treasury Regulation Section
1.6011– 4 (b) (2) defines a listed transaction as “a transaction that is the same as or
substantially similar to one of the types of transactions that the Internal Revenue
Service (IRS) has determined to be a tax avoidance transaction and identified by
notice, regulation, or other form of published guidance as a listed transaction.”
Ideally, we would like to identify those reportable transactions that are listed transactions; however, we do not have access to such detailed information. Thus, the
percentage of book-tax differences related to reportable transactions that we report are at best a noisy indication of the degree of tax avoidance related to the
book-tax differences.
To investigate whether public firms have more or less total, temporary and
permanent book-tax differences than do private firms, we undertake a multivariate analysis where we control for other variables related to book-tax differences.
Unfortunately, we do not have enough observations for quasi-public firms to include them in this analysis. We begin by estimating the following Ordinary Least
Squares (OLS) regression:
BTDit = β0 + β1PUBLICi + β2ROAit + β3PPEit + β4LEVit + β5INTANit +
β6GROWit + IndustryDummiesi + YearDummiest + ε
(1)
The dependent variable, BTD, is either total book-tax difference, SCBTD (lines
26 (b) (c) and 27 (b) (c) on Part II of Schedule M-3 plus lines 1 (b) (c)–7 (b) (c) on
Part III of Schedule M-3, divided by total assets at the end of the year (box D at
top of Form 1120)); permanent book-tax differences, PERM_SCBTD (lines 26 (c)
and 27 (c) on Part II of Schedule M-3 plus lines 1 (c)–7 (c) on Part III of Schedule
M-3, divided by total assets at the end of the year (box D at top of Form 1120)); or
temporary book-tax differences, TEMP_SCBTD (lines 26 (b) and 27 (b) on Part II
of Schedule M-3 plus lines 1 (b)–7 (b) on Part III of Schedule M-3, divided by total
assets at the end of the year (box D at top of Form 1120)). To study differences
between public and private firms’ book-tax differences, we include PUBLIC, which
equals 1 if the firm is a public firm and 0 if it is a private firm.
Equation (1) also includes several control variables. ROA, included as a control
for profitability, is defined as net book income (line 11 on Schedule M-3) divided
by total assets at the end of the year (box D at top of Form 1120). Presumably, less
profitable firms will have relatively less incentive to tax plan, yielding a positive
association with book-tax differences.
PPE is net property, plant, and equipment (sum of lines 10 (a) (c) and 10 (b) (c)
on Schedule L of Form 1120) divided by total assets at the end of the year (box D
at top of Form 1120). We control for property, plant, and equipment because prior
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
27
research finds that differences in book and tax depreciation is one of the determinants of book-tax differences (Seidman 2010). Depending on where a firm’s
long-term assets are in their life cycle, PPE could be negatively or positively related
to total and temporary book-tax differences. LEV, the sum of lines 17 (d), 20 (d),
and 21 (d) on Schedule L of Form 1120 divided by total assets at the end of the year
(box D at top of Form 1120), is included to control for the fact that firms with
more debt-related tax shields may have less of a need to be aggressive with other
tax planning. Consistent with Chen et al. (2010) and Robinson et al. (2010), we
include the variable INTAN, sum of lines 13 (a) (c) and 13 (b) (c) on Schedule L of
Form 1120, at the end of the year divided by total assets at the end of the year (box
D at top of Form 1120), to control for differing book and tax treatments of intangible assets. Opportunities to shift income could also be represented by INTAN
(Grubert and Slemrod 1998). GROW, one year percentage growth in sales (line 1c
on Form 1120), is our proxy for growth. Bankman (1994) finds that high-growth
firms generally place less emphasis on tax planning. We also include Industry
Dummies using Barth et al. (2001) industry classifications and Year Dummies.
Next, we examine whether firms treat temporary and permanent book-tax differences as substitutes or as complements. As discussed in Section 2 above, we
do not have an ex ante prediction regarding whether firms trade off one form
of tax planning for another. In addition, it is not clear whether firms have some
order in their preferences for different types of book-tax differences. Said another way, firms may first choose their permanent differences and then choose
their timing differences (or vice versa). Alternatively, they might jointly determine
the two types of tax planning.8 To test whether firms view the different types of
book-tax differences as complements or as substitutes, we estimate the following
regressions:
TEMP _ SCBTD = β0 + β1PERM _ SCBTD + β2ROA + β3PPE + β4LEV +
β5INTAN + β6GROW + IndustryDummies + YearDummies + ε
(2a)
PERM _ SCBTD = γ0 + γ1TEMP _ SCBTD + γ2ROA + γ3PPE + γ4LEV +
γ5INTAN + γ6GROW + IndustryDummies + YearDummies + ε
(2b)
If firms choose to maximize both timing and permanent differences, then we
expect a positive β1 and γ1 . If firms tax plan to some firm-specific optimal level,
then we expect firms to trade off one type of tax planning for the other leading to
a negative β1 and γ1 .
We also estimate a changes specification. If firms are actively trading off planning opportunities, then changes in the book-tax differences should also be correlated. Therefore, we estimate the following regressions:
28
Blouin, DeBacker, and Sikes
∆TEMP _ SCBTD = α0 + α1∆PERM _SCBTD + α2∆ROA + α3∆PPE + α4∆LEV +
α5∆INTAN + α6GROW + IndustryDummies + YearDummies + ε
(3a)
∆PERM _ SCBTD = ϕ0 + ϕ1∆TEMP _ SCBTD + ϕ2∆ROA + ϕ3∆PPE + ϕ4LEV +
ϕ5∆INTAN + ϕ6GROW + IndustryDummies + YearDummies + ε
(3b)
Finally, we investigate whether firms’ treatment of book-tax differences varies by ownership structure. Because public firms presumably face more earnings
pressure from the capital markets, they may have a stronger preference for permanent differences. However, if public firms are concerned with financial statement risk, then they may prefer temporary differences. Yet, if the capital markets are focused on the firms’ cash outflows for taxes, then these firms may be
extreme tax planners leading to public firms treating timing and permanent booktax differences as complements. To investigate whether ownership structure affects the association between the types of book-tax differences, we estimate the
following regressions:
TEMP _ SCBTD = β0 + β1PERM _ SCBTD + β2PUBLIC*PERM _ SCBTD +
β3ROA + β4PPE + β5LEV + β6INTAN + β7GROW + IndustryDummies +
YearDummies + ε
(2aʹ)
PERM _ SCBTD = γ0 + γ1TEMP _ SCBTD + γ2PUBLIC*TEMP _ SCBTD +
γ3ROA + γ4PPE + γ5LEV + γ6INTAN + γ7GROW + IndustryDummies +
YearDummies + ε
(2bʹ)
∆TEMP _ SCBTD = α0 + α1∆PERM _SCBTD + α2PUBLIC*∆PERM _ SCBTD +
α3∆ROA + α4∆PPE + α5∆LEV + α6∆INTAN + α7GROW + IndustryDummies +
YearDummies + ε
(3aʹ)
∆PERM _ SCBTD = ϕ0 + ϕ1∆TEMP _ SCBTD + ϕ2PUBLIC*∆TEMP _ SCBTD +
ϕ2∆ROA + ϕ3∆PPE + ϕ4LEV + ϕ5∆INTAN + ϕ6GROW + IndustryDummies +
YearDummies + ε
(3bʹ)
If substitution between temporary and permanent book-tax differences varies
between public and private firms, then we expect β2 , γ2, α2, and φ2 to be significant.
Note that we winsorize all continuous variables by year at the 1st and 99th percentiles throughout all of our analyses. In addition, all significance levels are reported using robust standard errors.
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
29
Results
Univariate
Table 2 presents descriptive statistics for our sample of firms by ownership type.
The mean Total Assets of public and quasi-public firms is larger ($1.42 billion and
$2.24 billion, respectively) than that of private firms. However, untabulated statistics reveal that private firms comprise more of the total economy ($12.3 trillion in total assets as compared to $8.9 trillion and $0.09 trillion for public and
quasi-public firms, respectively). ROA, net book income over assets, is greatest
for public firms (mean of 0.063). However, mean TI ROA, taxable income (line 30
on Form 1120) over total assets at the end of the year, is actually greater for private
firms than for public firms. But median TI ROA is identical between public and
private firms.
Quasi-public have the greatest growth opportunities (mean GROW equal to
0.23, mean INTAN equal to 0.10). Public firms appear to have greater growth opportunities relative to private firms as evidenced by their mean and median values
for GROW and INTAN. Private firms have more debt (mean value of LEV equal
to 0.264) than public firms (mean value of LEV equal to 0.241), consistent with
private firms’ need to replace equity capital with debt capital. Quasi-public firms’
mean LEV of 0.498 double that of the private and the public firms, which suggests
that many of these firms have undergone a leveraged buyout (LBO) or have private
equity ownership. Also, the high level of tangible assets of the quasi-public (mean
PPE equal to 0.327 firms) is consistent with the LBO model as assets provide security for the debt. Notice that all three categories of firms are reasonably mature
with an average number of years since incorporation of approximately a quarter of
a century. Finally, public and private firms have a similar number of firms reporting small levels of income, defined as taxable income (line 30 on Form 1120) less
than one percent of total assets (box D at top of Form 1120).
In Table 3, we present descriptive statistics of the book-tax differences, by ownership category. Consistent with public firms undertaking less conforming tax
planning, we find that total (SCBTD), temporary (TEMP_SCBTD) and permanent
(PERM_SCBTD) scaled book-tax differences are greater for public firms than for
private firms.9
Notice that the difference between public and private firms is not being driven
by differences in the extreme book-tax difference quintile (Quintile 5), as mean
SCBTD, TEMP_SCBTD and PERM_SCBTD for Quintile 5 are very similar across
the public and private ownership groups. Also, note that the Quintile 1 booktax differences, i.e., the book-tax differences that increase taxable income relative
to book income, are much smaller in magnitude than the Quintile 5 book-tax
differences. Consistent with public firms being relatively more concerned with
the financial statement implications of tax planning, public firms have larger
30
Blouin, DeBacker, and Sikes
PERM_SCBTD than do private firms. Finally, public firms have more reportable
transactions than do private firms. However, because any book-tax difference
greater than $10 million must be included as a reportable transaction, we hesitate
to draw any inference about whether this item captures aggressive tax planning.
Surprisingly, we find that mean SCBTD, TEMP_SCBTD, and PERM_SCBTD
for the quasi-public group is actually income increasing. Note that the (untabulated) medians are also negative suggesting that the mean scaled book-tax differences are not driven by extreme observations in the tails. As there are often passthrough entities involved in an LBO and/or private equity backed transaction, it
may be that some of the negative book-tax differences stem from including the
firms’ proportional interest in those entities in their taxable income.
In Table 4, we report SCBTD, TEMP_SCBTD, and PERM_SCBTD by ownership type and by industry. Consistent with tax incentives varying across industries, we find significant variation in the book-tax differences. Notice that Mining
and Extractive Industries, where depreciation and percentage depletion are sizeable, have the largest book-tax differences.
Multivariate
In Table 5, we report the result of estimating Equation (1) across each of scaled
measures of book-tax differences. The positive coefficient on PUBLIC in each of
the three models suggests that public firms have relatively more book-tax differences than private companies. Since the univariate statistics suggest that public
and private firms ultimately have similar taxable incomes, our results are consistent with public firms undertaking less conforming tax planning. This result is
consistent with those in Cloyd et al. (1996).
Many of the control variables load consistent with expectations. ROA is significant and positive, which suggests that profitable firms engage in more tax planning. PPE loads positively (negatively) in the PERM_SCBTD (TEMP_SCBTD)
model. The negative relation between PPE and TEMP_SCBTD could be consistent with long-term assets nearing the end of their depreciable lives such that
book depreciation exceeds tax depreciation, and with firms not replacing the assets with new assets. Moreover, the positive relation between PPE and PERM_
SCBTD could be consistent with firms substituting the former tax shields from
depreciation with permanent book-tax differences. The positive association between LEV and PERM_SCBTD suggests that firms manage their effective tax rate
to prevent debt covenant violation. The negative association between LEV and
TEMP_SCBTD could be explained by the fact that debt is a natural tax shield.
Hence, firms with high levels of debt need relatively less incremental tax planning (Mackie-Mason 1990). The positive association between PERM_SCBTD and
INTAN could be due to the fact that intangible assets often give rise to creditable
research and development expenses. Finally, the negative association between
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
31
GROW and PERM_SCBTD is consistent with Bankman’s (1994) argument that
high growth firms generally place less emphasis on tax planning.
In Table 6, we investigate whether firms treat permanent and temporary booktax differences as substitutes or as complements. In Panel A, we report the results
of the estimation of Equations (2a), (2b), (3a), and (3b). Consistent with firms
treating permanent and temporary book-tax differences as substitutes, in columns
(1) and (2) we find a significant negative association between PERM_SCBTD and
TEMP_SCBTD. Notice that the negative association is independent of the dependent variable. These results suggest that firms have some “optimal” tax planning
level that they reach using a mixture of temporary and permanent differences.
The results from columns (3) and (4) suggest that as one category of book-tax
differences increases, the other decreases. Hence, these findings imply that firms
recognize that there are some costs to tax planning either due to an increase in
the likelihood of an audit (Mills 1998) or an increase in the likelihood of being assessed a penalty if they are overly aggressive tax planners.
In Panel B of Table 6, we report the results of estimating Equations (2aʹ), (2bʹ),
(3aʹ), and (3bʹ). By incorporating the interaction term between PUBLIC and the
relevant book-tax difference measure, we are able to determine whether capital
market pressure alters the rate of substitution between temporary and permanent
book-tax differences documented in Table 6, Panel A. The results in columns (1)
and (2) suggest that public and private firms do not have a significantly different
rate of substitution between their types of book-tax differences (i.e., the coefficients on PUBLIC*PERM_SCBTD and PUBLIC*TEMP_SCBTD are not statistically significant). When we move to the changes analysis, we find no evidence that
changes in temporary difference have a significant difference impact on changes in
permanent differences for public firms (see column (3)).
In column (4), we find that changes in permanent differences have less influence on changes in temporary differences (PUBLIC*ΔPERM_SCBTD 0.176,
p-value < 0.01). One explanation for this result could be that capital markets incentives lead to some weakening of the substitution between the types of book-tax
differences.
Conclusion
We use confidential data from the Schedule M-3 that all C corporations with assets
over $10 million must attach to their corporate tax return to examine the use of
total, permanent, and temporary book-tax differences among a sample of public,
private, and quasi-public firms over the years 2005-2007. The Treasury created the
Schedule M-3 in response to concern among legislators and regulators regarding
the growing divide between net income that firms report to the public and taxable
income that they report to tax authorities. Beginning in 2004, the Schedule M-3
32
Blouin, DeBacker, and Sikes
replaced the Schedule M-1 for all firms with total assets of $10 million or more.
The Schedule M-3 contains much more detail about firms’ book-tax differences
than does the Schedule M-1. This paper is one of the first academic studies to utilize the data on the Schedule M-3. In addition to allowing us to accurately measure
firms’ book-tax differences and to separate them between permanent and temporary differences, the Schedule M-3 data allows us to examine the financial and tax
reporting behavior of private firms.
We expect that both public and private firms engage in tax planning. However,
because private firms face less pressure from capital markets to report higher earnings, we expect that private firms engage in more conforming transactions (i.e.,
transactions that reduce book income and taxable income) and, as a result, have
smaller book-tax differences than public firms. Consistent with our expectation,
we find that public firms have greater total book-tax differences, greater permanent book-tax differences, and greater temporary book-tax differences (all scaled
by total assets) than do private firms.
A recent study (Badertscher et al. 2010) finds that private firms that have a private equity firm as either a majority or minority owner are more aggressive in
their tax reporting than are private firms that are not owned by a private equity
firm. We extend Badertscher et al. (2010) by collecting a sample of firms that
have private equity and public debt. We label these firms “quasi-public” firms.
Although we do not know if the quasi-public firms in our sample are owned by a
private equity firm, they exhibit traits that are common to firms that are owned by
private equity firms (e.g., high growth opportunities and high levels of debt and
tangible assets). Based on the findings in Badertscher et al. (2010), we expect that
the quasi-public firms in our sample have greater book-tax differences than the
private firms in our sample. Inconsistent with this expectation, we find that the
quasi-public firms in our sample actually have income-decreasing book-tax differences. We do not place much weight on this result, however, due to the very low
number of quasi-public firms in our sample.
Next, we use multivariate analysis to examine whether public and private firms
treat temporary and permanent book-tax differences as substitutes or as complements.
We find that firms treat permanent and temporary book-tax differences as substitutes.
However, the substitution effect does not appear to vary by ownership type.
Acknowledgements
The authors thank Charles Boynton, Petro Lisowsky, and Drew Lyon for helpful
comments.
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
33
References
Badertscher, B., S. Katz, and S. Rego (2010), “The Impact of Private Equity
Ownership on Portfolio Firms’ Corporate Tax Planning,” Working paper
(Notre Dame University, Columbia Business School, and the University of
Iowa).
Bankman, J. (1994), “The Structure of Silicon Valley start-ups,” UCLA Law
Review 41 (September): 1737–1768.
Barth, M.E., J. Elliott, and M. Finn (1999), “Market Rewards Associated with
Patterns of Increasing Earnings,” Journal of Accounting Research 37(2): 387–413.
Barth, M.E., W.H. Beaver, and W.R. Landsman (2001), The Relevance of the
Value Relevance Literature for Financial Accounting Standard Setting: Another
View. Journal of Accounting and Economics 31 (September 2001): 77–104.
Beatty, A., and D.G. Harris, (1998), “The Effects of Taxes, Agency Costs and
Information Asymmetry on Earnings Management: A Comparison of Public
and Private Firms”. Review of Accounting Studies 3: 299–326.
Blouin, J., C. Armstrong, and D. Larcker (2010), “The Incentives for Tax
Planning”. Working paper, University of Pennsylvania and Stanford University.
Boynton C.E., P. DeFilippes, and E.J. Legel (2008), “A First Look at 2005 Schedule
M-3 Corporate Reporting”. Tax Notes (November 3, 2008): 563–596.
Chen, K., and M. Schoderbek (2000), “The 1993 Tax Rate Increase and Deferred
Tax Adjustments: A Test of Functional Fixation”. Journal of Accounting
Research 38(1): 23–44.
Chen, S., X. Chen, Q. Cheng, and T. Shevlin 2010, “Are Family Firms More or
Less Tax Aggressive Than Non-Family Firms?” Journal of Financial Economics
95 (1): 41–61.
Cloyd, B., J. Pratt, and T. Stock (1996), “The Use of Financial Accounting Choice
to Support Aggressive Tax Positions: Public and Private Firms”. Journal of
Accounting Research 34 (1): 23–43.
Desai, M. (2003), “The Divergence Between Book and Tax Income”. In Tax Policy
and the Economy, Volume 17, edited by James M. Poterba. Cambridge: NBER
and MIT Press: 169–206.
Dietrich, J.R. (1984), “Effects of Early Bond Refunding: An Empirical
Investigation of Security Returns”, Journal of Accounting and Economics 6 (1):
67–96.
Dyreng, S., M. Hanlon, and E. Maydew (2010), “The Effects of Executives on
Corporate Tax Avoidance”, The Accounting Review, forthcoming.
34
Blouin, DeBacker, and Sikes
Frank M.M., L.J. Lynch, and S.O. Rego (2009), “Tax Reporting Aggressiveness
and its Relation to Aggressive Financial Reporting Aggressiveness”, The
Accounting Review 84 (2): 467–496.
Graham, J., J. Raedy, and S. Shackelford (2009), “Research in Accounting for
Income Taxes”, Working Paper, Duke University and the University of North
Carolina-Chapel Hill.
Grubert, H. and J. Slemrod (1998), “The Effect of Taxes on Investment and
Income Shifting to Puerto Rico”, Review of Economics and Statistics 80 (3):
365–373.
Hand, J.R.M. (1990), “A Test of the Extended Functional Fixation Hypothesis,”
The Accounting Review 65 (4): 740–763.
Hanlon, M. (2005), “The Persistence and Pricing of Earnings, Accruals, and Cash
Flows When Firms Have Large Book-Tax Differences,” The Accounting Review
80: 137–166.
Louie, Henry J. (2005), “A First Look at the Book-Tax Differences in the ForeignSource Income of U.S. Multinational Companies,” National Tax Association
Proceedings from 98th Annual Conference on Taxation: 138–144.
MacKie-Mason, Jeffrey K. (1990), “Do Taxes Affect Corporate Financing
Decisions?” Journal of Finance 45: 1471–1493.
Manzon, G. and G. Plesko. (2002), “The Relation Between Financial and Tax
Reporting Measures of Income,” Tax Law Review 55: 175–214.
McGill, G. and E. Outslay, (2004), “Lost in Translation: Detecting Tax Shelter
Activity in the Financial Statements”, National Tax Journal 57 (3): 739–756.
Mikhail, M.B. (1999), “Coordination of Earnings, Regulatory Capital and Taxes
in Private and Public Companies,” Working paper, Arizona State University.
Mills, L. (1998), “Book-Tax Differences and Internal Revenue Service
Adjustments,” Journal of Accounting Research 36 (2): 343–356.
Mills, L., and G. Plesko (2003), “Bridging the Gap: A Proposal for More
Informative Reconciling of Book and Tax Income,” National Tax Journal 56 (4):
865–893.
Neubig, T. (2006), “Where’s the Surplus? Why Most Corporations Prefer a Lower
Tax Rate,” Tax Notes 111: 483–486.
Penno, M., and D.T. Simon (1986), “Accounting Choices: Public Versus Private
Firms,” Journal of Business Finance and Accounting 13 (4): 561–569.
Phillips, J., M. Pincus, and S. Rego (2003), “Earnings Management: New
Evidence Based on Deferred Tax Expense,” The Accounting Review 78 (2):
491–521.
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
35
Phillips, J., M. Pincus, S. Rego, and H. Wan (2004), “Decomposing Changes in
Deferred Tax Assets and Liabilities to Isolate Earnings Management Activities,”
The Journal of the American Taxation Association 26 (Supplement): 43–66.
Plesko, G. (2002), “Reconciling Corporation Book and Tax Net Income, Tax
Years 1996–1998,” SOI Bulletin, U.S. Government Printing Office, Washington
DC, (Spring): 111–132.
Plesko, G. (2004), “Corporate Tax Avoidance and the Properties of Corporate
Earnings,” National Tax Journal 58: 729–737.
Rego, S. (2003), “Tax Avoidance Activities of U.S. Multinational Corporations,”
Contemporary Accounting Research 20, 805–833.
Robinson, J.R., S.A. Sikes, and C.D. Weaver (2010), “Performance Measurement
of Corporate Tax Departments,” The Accounting Review 85 (3): 1035–1064.
Seidman, J. (2010), “Interpreting the Book-Tax Gap as Earnings Management or
Tax Sheltering,” Working paper, University of Texas at Austin.
Talisman, J. (2000), “Corporate Tax Shelters and the Corporate Tax Base,” pp.
4–6, in “Penalty and interest provisions, corporate tax shelters,” Testimony of
Jonathan Talisman, Assistant Secretary (Tax Policy), in U.S. Department of the
Treasury, Before the U.S. Senate, Committee on Finance, Washington, March
8, 2000.
U.S. Department of Treasury (1999), “Evidence of growth in corporate tax
shelters,” The Problem of Corporate Tax Shelters: Discussion, Analysis, and
Legislative Proposals. Washington, D.C.: Government Printing Office, July
1999: 31–33.
Weisbach, D. (2002), “Ten truths about tax shelters,” Tax Law Review 55: 215–253.
Wilson, R. (2009), “An examination of corporate tax shelter participants,” The
Accounting Review 84 (3): 969–999.
Endnotes
1
See Graham et al. (2009) for a review of the literature on book-tax differences
and accounting for income taxes and Seidman (2010) for an interpretation
of whether book-tax differences are due to earnings management or tax
sheltering.
2 We do not have enough quasi-public firm observations to include them in this
analysis.
3 The composition of the Schedule M-3 heavily reflects recommendations in
Mills and Plesko (2003). We refer readers to Boynton et al. (2008) for an
excellent summary of the details of Schedule M-3.
36
4
5
6
7
8
9
Blouin, DeBacker, and Sikes
Other papers that analyze public and private firms include Beatty and Harris
(1998), Mikhail (1999), and Penno and Simon (1986). Beatty and Harris (1998)
analyzes 297 public and 553 private bank year observations during 1991 and
1992. They find that public banks partake in earnings management significantly
more than do private banks. They conclude that information asymmetry
in public banks motivates the earnings management. Mikhail (1999) finds
that tax management is more prevalent among private insurance companies
than among public insurance companies. Mikhail (1999) provides evidence
suggesting that the differences in incentive compensation contracts, designed
to control agency costs, are partially responsible for the difference. Penno and
Simon (1986) finds that publicly-traded firms are more likely to use incomeincreasing accounting methods (i.e., inventory choice and depreciation) than
are privately-held firms.
If the absolute value of ((Schedule M-3 line 5a + Schedule M-3 line b5)/
Schedule M-3 line 4) > 0.25, then we remove the firm-year from our analysis.
Note that we will likely incorrectly designate some public firms as private
because the EIN that firms report on their tax return is different that the EIN
they include in their Form 10-K. However, this misclassification should bias
against us finding results.
Note that transactions with significant book-tax differences were eliminated
from the list of reportable transactions as of January 6, 2006.
In an untabulated test, we replace PERM_SCBTD in equations (2a) and (2a’)
with its one-period lagged value and replace TEMP_SCBTD in equations
(2b) and (2b’) with its one-period lagged value. We do so to ensure that
the potential endogenous relation between a firm’s level of permanent and
temporary book-tax differences is not responsible for our results. Using
lagged values does not change the results reported in columns (1) and (2) of
Tables 6 and 7, which confirms that the reported results are not attributable to
endogeneity.
On the Schedule M-3, a positive pre-tax difference means that taxable income
is greater than book income. As explained in the previous section, in the
1990s, regulators and academics became concerned that firms’ reported pretax book income began to exceed reported taxable income. The concern
relates to the fact that the growing divide could be due to firms over-stating
their pre-tax book income to shareholders, debt holders, and other interested
parties, and/or firms engaging in tax shelters that reduce their taxable income.
In order to be consistent with this idea and with the direction of the difference
used in prior studies, we flip the sign on the pre-tax difference reported on the
Schedule M-3. Thus, in our paper, a positive pre-tax difference suggests that
pre-tax book income exceeds taxable income.
37
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
TABLE 1. Sample Derivation
Panel A: Firm and Firm-year Reconciliation
In IRS data
Firms
Firm-years
135,406
304,806
Less: non C Corporations
65,772
146,851
Less: total assets less than $10 million
34,463
77,446
977
2,805
Less: no Schedule M-3 attached
Less: large foreign operations
2,140
7,271
Less: negative book income
7,681
20,300
24,373
50,133
Total in Sample
Panel B: Sample reconciliation by Ownership Type
Public
Quasi-Public
Private
Firms
2,799
23
21,445
Firm Years
6,255
40
43,838
Industry
Public
Quasi-Public
Private
1. Mining
1.63%
0.00%
0.86%
2. Food
1.95%
0.00%
2.10%
3. Textiles, printing, publishing
3.23%
2.50%
2.66%
4. Chemicals
1.14%
0.00%
1.43%
5. Pharmaceuticals
2.41%
0.00%
0.40%
6. Extractive Industries
3.93%
7.50%
1.34%
14.00%
5.00%
11.28%
8. Computers
8.36%
5.00%
2.40%
9. Transport
4.94%
2.50%
4.66%
10. Utilities
4.43%
10.00%
0.94%
11. Retail
11.45%
42.50%
18.76%
12. Finance and Insurance
Panel C: Firm-Year Industry Representation
7. Durable Manufacturing
31.35%
15.00%
15.29%
13. Real Estate, trusts, bank holding co
1.81%
0.00%
23.22%
14. Services
9.24%
10.00%
8.57%
15. Other
0.13%
0.00%
6.10%
Sum
0.183
0.241
0.102
0.210
0.369
PPE
LEV
INTAN
GROW
# Small Inc. Firms
17.000
0.105
0.016
0.159
0.070
0.038
0.018
0.038
526.78
Median
Sum
18
89,412
23.925
0.450
0.229
0.100
0.498
0.327
0.048
0.035
0.034
2,235.29
Mean
Quasi-Public
Median
14.500
0.075
0.018
0.488
0.233
0.023
0.012
0.023
1,055.94
Sum
26.612
16,769
12,248,235
20.000
0.383
0.167
0.043
0.264
0.166
0.072
0.062
0.054
279.40
Mean
Private
0.075
0.000
0.148
0.050
0.030
0.018
0.026
66.88
Median
NOTES: Total Assets equals total assets at the end of the year (box D at top of Form 1120). ROA equals net book income (line 11 on Schedule M-3) divided by total assets at the end of
the year (box D at top of Form 1120). TI ROA equals taxable income (line 30 on Form 1120) divided by total assets at the end of the year (box D at top of Form 1120). NI ROA equals
taxable income before net operating loss deduction and special deductions (line 28 on Form 1120) divided by total assets (box D at top of Form 1120). PPE equals net property, plant and
equipment (sum of lines 10(a)(c) and 10(b)(c) on Schedule L) divided by total assets at the end of the year (box D at top of Form 1120). LEV equals the sum of lines 17(d), 20(d), and
21(d) on Schedule L of Form 1120, divided by total assets at the end of the year (box D at top of Form 1120). INTAN equals the sum of lines 13(a)(c) and 13(b)(c) on Schedule L at the
end of the year divided by total assets at the end of the year (box D at top of Form 1120). GROW equals one year percentage growth in sales (line 1c on Form 1120). Small Income Firms
are defined as firms whose taxable income (line 30 on Form 1120) are less than one percent of total assets at the end of the year (box D at top of Form 1120). # Yrs. Incorporated denotes
the number of years that a firm has been incorporated. All continuous variables are winsorized at the 1st and 99th percentiles.
22.569
0.070
NI ROA
Number Years Incorporated
0.059
TI ROA
2,310
Mean
1,422.52
0.063
8,887,838
Public
ROA
Total Assets (in $ Millions)
Variable
TABLE 2. Descriptive Statistics
38
Blouin, DeBacker, and Sikes
0.185
0.024
Quintile 5 SCBTD
% Reportable Transactions
0.156
0.018
Quintile 5 TEMP_SCBTD
% Reportable Transactions
-0.001
0.068
0.011
Quintile 1 PERM_SCBTD
Quintile 5 PERM_SCBTD
% Reportable Transactions
0.001
0.064
-0.002
0.001
473.885
0.003
0.156
-0.009
0.002
469.558
0.004
0.185
-0.009
0.004
1,054.234
Private
0.000
0.028
-0.004
-0.003
1,763.064
0.050
0.117
-0.019
-0.007
965.002
0.050
0.117
-0.026
-0.010
4,361.277
Quasi-Public
7.08
16.61
3.23
20.14
27.94
9.03
0.00
-6.52
5.65
7.24
10.36
0.00
5.19
14.61
19.98
Public-Private
8.15
152.25
10.88
34.12
20.57
-18.40
72.70
12.59
22.84
5.62
-13.30
105.00
31.27
36.34
11.41
Public-Qpub
7.86
12.45
0.70
1.28
-0.59
-1.356
6.69
1.76
1.58
-0.10
-1.33
10.37
2.58
2.15
-0.63
Private-Qpub
T-Statistic for Differences Between
NOTES: BTD equals sum of permanent and temporary book-tax differences (lines 26(b)(c) and 27(b)(c) on Part II of Schedule M-3 plus lines 1(b)(c)-7(b)(c) on Part III of Schedule M-3).
SCBTD equals the sum of permanent and temporary book-tax differences (lines 26(b)(c) and 27(b)(c) on Part II of Schedule M-3 plus lines 1(b)(c)-7(b)(c) on Part III of Schedule M-3),
divided by total assets at the end of the year (box D at top of Form 1120). TEMP_BTD equals temporary book-tax differences (lines 26(b) and 27(b) on Part II of Schedule M-3 plus lines
1(b)-7(b) on Part III of Schedule M-3). TEMP_SCBTD equals temporary book-tax differences (lines 26(b) and 27(b) on Part II of Schedule M-3 plus lines 1(b)-7(b) on Part III of Schedule
M-3), divided by total assets at the end of the year (box D at top of Form 1120)). PERM_BTD equals permanent book-tax differences (lines 26(c) and 27(c) on Part II of Schedule M-3
plus lines 1(c)-7(c) on Part III of Schedule M-3). PERM_SCBTD equals permanent book-tax differences (lines 26(c) and 27(c) on Part II of Schedule M-3 plus lines 1(c)-7(c) on Part III
of Schedule M-3), divided by total assets at the end of the year (box D at top of Form 1120)). Firms are divided into five quintiles based on the amounts of their SCBTD, TEMP_SCBTD,
and PERM_SCBTD, with Quintile 1 (5) including the firms with the smallest (largest) amounts of these variables. % Reportable Transactions are the percent of the total, temporary, or
permanent book-tax differences that are associated with “reportable transactions” under Treasury Regulation Section 1.6011-4(b). All continuous variables are winsorized at the 1st and
99th percentiles.
0.006
PERM_SCBTD
PERM_BTD
5,362.793
-0.012
Quintile 1 TEMP_SCBTD
Permanent book-tax Differences
0.005
TEMP_SCBTD
TEMP_BTD
2,678.956
-0.006
Quintile 1 SCBTD
Temporary book-tax Differences
0.014
8,762.823
Public
SCBTD
BTD
Total book-tax Differences
Variable (Means)
TABLE 3. Descriptive Statistics on Book-Tax Difference
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
39
0.006
0.010
0.016
0.016
0.066
0.017
0.022
0.024
0.016
0.008
0.002
0.025
0.018
0.000
3. Textiles, printing, publishing
4. Chemicals
5. Pharmaceuticals
6. Extractive Industries
7. Durable Manufacturing
8. Computers
9. Transport
10. Utilities
11. Retail
12. Finance and Insurance
13. Real Estate, trusts, bank holding co
14. Services
15. Other
-0.012
0.008
0.008
0.000
-0.001
0.012
0.019
0.007
0.003
0.057
-0.002
0.006
0.003
0.000
-0.003
TEMP_
SCBTD
0.003
0.006
0.013
0.002
0.007
0.003
0.003
0.011
0.012
0.005
0.011
0.007
0.006
0.008
0.014
PERM_
SCBTD
NOTES: Industry groups based on Barth et al. (2001). See Table 3 for variable definitions.
0.015
2. Food
SCBTD
1. Mining
Industry (Means)
Public
0.004
0.003
0.006
0.003
-0.001
0.008
0.013
0.002
0.003
0.042
0.002
0.006
0.002
0.010
0.034
SCBTD
0.003
0.004
0.003
0.002
0.000
0.003
0.008
0.000
0.000
0.031
-0.010
0.002
0.001
0.006
0.016
TEMP_
SCBTD
Private
0.001
-0.002
0.002
0.001
-0.001
0.003
0.003
0.000
0.002
0.006
0.007
0.002
0.001
0.003
0.013
PERM_
SCBTD
TABLE 4. Variation in Book-Tax Differences across Industries
0.000
-0.039
0.000
-0.002
-0.010
0.005
0.007
-0.098
-0.040
0.068
0.000
0.000
0.001
0.000
0.000
SCBTD
0.000
-0.036
0.000
-0.002
-0.007
-0.008
0.011
-0.043
-0.044
0.068
0.000
0.000
0.003
0.000
0.000
TEMP_
SCBTD
Quasi-Public
0.000
-0.003
0.000
0.000
-0.003
0.013
-0.004
-0.054
0.004
0.000
0.000
0.000
-0.002
0.000
0.000
PERM_
SCBTD
40
Blouin, DeBacker, and Sikes
41
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
TABLE 5. Total, Temporary, and Permanent Book-Tax Differences as a Function of
Public vs. Private Organizational Form
Dependent Variable
PUBLIC
ROA
PPE
LEV
SCBTD
PERM_SCBTD
TEMP_SCBTD
(1)
(2)
(3)
0.008***
0.002***
0.005***
(0.001)
(0.001)
(0.000)
0.126***
0.064***
0.030***
(0.007)
(0.006)
(0.002)
0.002
0.004***
-0.001***
(0.001)
(0.001)
(0.000)
0.009***
0.010***
-0.002***
(0.000)
(0.001)
(0.001)
0.011***
0.012***
-0.001
(0.002)
(0.002)
(0.001)
-0.001***
-0.001***
0.000
(0.000)
(0.000)
(0.000)
Industry Fixed Effects Included
YES
YES
YES
Year Fixed Effects Included
YES
YES
YES
0.081
0.047
0.060
41,900
41,900
41,900
INTAN
GROW
R-squared
N
*p<0.10, **p<0.05, ***p<0.01.
NOTES: See Tables 2 and 3 for variable definitions. Heteroskedasticity-robust standard errors are reported in parentheses below the coefficient point estimates.
42
Blouin, DeBacker, and Sikes
TABLE 6. Treatment of Temporary and Permanent Book-Tax Differences as
Complements or Substitutes as a Function of Public or Private Status
Panel A: Basic Complement or Substitution
Dependent Variable
TEMP_SCBTD
PERM_SCBTD
TEMP_SCBTD
(1)
(2)
∆PERM_SCBTD ∆TEMP_SCBTD
(3)
-0.063***
(0.004)
-0.326***
PERM_SCBTD
(0.021)
-0.071***
∆TEMP_SCBTD
(0.006)
-0.423***
∆PERM_SCBTD
ROA or ∆ROA
PPE or ∆PPE
LEV or ∆LEV
INTAN or ∆INTAN
(4)
(0.031)
0.035***
0.075***
0.046***
0.146***
(0.002)
(0.006)
(0.005)
(0.012)
-0.001
0.004***
0.004
0.037***
(0.000)
(0.001)
(0.003)
(0.009)
-0.002***
0.009***
0.001
0.011**
(0.005)
(0.000)
(0.001)
(0.002)
0.002***
0.013***
0.004
0.008
(0.001)
(0.002)
(0.005)
(0.008)
0.000
-0.001***
0.046***
0.146***
(0.000)
(0.000)
(0.005)
(0.012)
Industry Fixed Effects
Included
YES
YES
YES
YES
Year Fixed Effects
Included
YES
YES
YES
YES
GROW
R-squared
N
0.067
0.066
0.042
0.057
41,900
41,900
23,421
23,421
*p<0.10, **p<0.05, ***p<0.01.
NOTES: See Tables 2 and 3 for variable definitions. Heteroskedasticity-robust standard errors are reported in parentheses below the coefficient point estimates.
43
Temporary and Permanent Book-Tax Differences: Complements or Substitutes?
TABLE 6. Treatment of Temporary and Permanent Book-Tax Differences as
Complements or Substitutes as a Function of Public or Private Status (Continued)
Panel B: Complements or Substitution by Ownership
Dependent Variable
TEMP_SCBTD
PUBLIC*TEMP_SCBTD
PERM_SCBTD
TEMP_SCBTD
(1)
(2)
∆PERM_SCBTD ∆TEMP_SCBTD
(3)
(4)
-0.061***
(0.005)
-0.018
(0.012)
-0.340***
PERM_SCBTD
(0.025)
0.058
PUBLIC*PERM_SCBTD
(0.043)
-0.069***
∆TEMP_SCBTD
-0.006
-0.009
PUBLIC*∆TEMP_SCBTD
-0.015
-0.467***
∆PERM_SCBTD
(0.037)
0.176***
PUBLIC*∆PERM_SCBTD
(0.063)
ROA or ∆ROA
PPE or ∆PPE
LEV or ∆LEV
INTAN or ∆INTAN
GROW
Industry Fixed Effects
Included
Year Fixed Effects Included
R-squared
N
0.035***
0.074***
0.046***
0.146***
(0.002)
(0.006)
-0.005
(0.012)
-0.001
0.004***
0.004
0.037***
0.000
(0.001)
-0.003
(0.009)
-0.002***
0.009***
0.001
0.011**
(0.005)
0.000
(0.001)
-0.002
0.002***
0.013***
0.005
0.008
(0.001)
(0.002)
-0.005
(0.008)
0.000
-0.001***
0.000
-0.001***
0.000
0.000
0.000
-0.001
YES
YES
YES
YES
YES
YES
YES
YES
0.068
0.067
0.042
0.058
41,900
41,900
23,421
23,421
*p<0.10, **p<0.05, ***p<0.01.
NOTES: See Tables 2 and 3 for variable definitions. Heteroskedasticity-robust standard errors are reported in parentheses below the coefficient point estimates.
2
D
Influencing Individual Taxpayer Behavior
Morse
An Analysis of the FBAR
High-Penalty Regime
Susan C. Morse, University of California Hastings College of Law
H
ow can the U.S. government find wealthy Americans with assets in offshore financial accounts and make sure they pay their taxes? One tool
is the Report of Foreign Bank and Financial Accounts, or FBAR, rules
that the U.S. government has begun to systematically enforce. The FBAR regime
requires individual taxpayers to submit statements disclosing their holdings in offshore accounts or face enormous penalties based on the account asset value. Will
the regime work? Maybe.
The first part of this paper describes how a high-penalty regime can successfully
push taxpayers to comply and self-identify as compliers through the mechanisms
of deterrence, separation, and/or signaling. These mechanisms can succeed if (1)
taxpayers perceive that penalties for noncompliers and rewards offered to compliers are credible; (2) taxpayers lack close-substitute, penalty-free choices; and (3)
taxpayers perceive that the government will detect noncompliers trying to masquerade as compliers. The second part contends that it is possible for the U.S. to
achieve all of these elements in the FBAR case, although it would require changes
to the current administration of the rules and sustained litigation and publicity.
Framework for Analyzing High-Penalty Regimes
Penalties and Rewards
High penalties can increase compliance in several ways. One mechanism is deterrence. The hypothetical fully rational taxpayer decides whether to evade tax by
comparing the amount of saved tax to the penalties for cheating weighted by the
chance that they will be detected.1 Risk aversion modifies this analysis, adding a
compliance bias to the fully rational model.2
Another mechanism is separation. If some taxpayers are willing to comply,
they may be more inclined to self-identify as compliant if they know that failing
to do so subjects them to the possibility of high penalties. Moreover, complier
self-identification can permit the application of a more understanding enforcement approach for compliers and reduce the risk that compliant behavior will be
crowded out by threats of severe penalties.3
High penalties can also serve as signals that may change compliance norms.
Compliance behavior is not only the product of inherent individual preference.
4
Morse
The identification of an illegal behavior by the government—including through
the announcement and implementation of high penalties applicable to the behavior—can act as a norm signal that may cause some significant number of taxpayers
to adopt compliance behavior.4 Peer-to-peer influences, in turn, may induce still
others to comply.5 Some empirical and experimental research supports the conclusion that compliance norms can evolve over time depending on the compliance
decisions of other taxpayers with whom the taxpayer identifies or communicates.6
And, once an individual starts to act in a more compliant fashion, the change can
become an entrenched part of the way the individual views his or her personality
and values.7
Yet high penalties have the potential to crowd out compliant behavior as well
as serving the compliance-enhancing functions of deterrence, separation, and signaling.8 They may commoditize and thereby undermine previous social norms of
compliance.9 Or they may be interpreted by a compliant taxpayer as a defecting
move in the previously reciprocal tit-for-tat compliance relationship the taxpayer
had built with the government.10
One solution to the problem of crowding out is to apply and articulate different penalties proportional to the severity of different offenses, and to also
publicize rewards, such as better taxpayer service, offered to compliers.11 This
is part of the reason why rewards for compliance are also important. Rewards
may also comprise an important part of a high-penalty strategy because taxpayers
who self-identify as compliers may be more likely to remain compliers if rewards
engage the government with the taxpayer in a mutually reinforcing tit-for-tat
reciprocity cycle.12
Sometimes rewards can be specific to a certain set of rules. But it is also true
that a framework that rewards compliers is already built into existing tax administration practice. The two-pronged service and enforcement mission of the
Internal Revenue Service (IRS), increasingly emphasized over the last 10 years
or so, reflects the government’s view that “good” and “bad” taxpayers should be
treated differently.13 In the IRS audit and appeals processes, there is every reason
to believe that taxpayers with records of compliance receive better treatment than
taxpayers with records of noncompliance. In some cases, the idea of better service
for more compliant taxpayers has been formalized into specific initiatives, such as
the Compliance Assurance Program, or CAP, which is available to certain large
corporate taxpayers.14
Penalty and Reward Credibility
The ability of penalties and rewards to achieve deterrence, separation, and / or signaling goals without falling into a crowding-out trap depends on more than the
penalties and rewards as stated in the statute books. A gap often exists between a
de jure penalty or reward and a de facto penalty or reward policy. Several causes
An Analysis of the FBAR High-Penalty Regime
5
can produce this gap between an on-the-books penalty or reward and its enforcement in practice. These may include litigation risk management;15 internal agency
politics, such as a desire to stick to prior practice or avoid adversarial relationships
with regulatees;16 national politics, including the goal of avoiding backlash legislation that could curb the agency’s power or resources in response to an excessively
tough public image;17 and international politics, including a reluctance to upset
foreign governments with U.S. policies that appear harsh and unilateral.18
A conceptually distinct—and more important 19—gap also often exists between
a de jure penalty or reward and taxpayers’ perception of a de facto penalty or reward policy. Taxpayers’ internal perception of the likelihood of penalty imposition
or reward enjoyment drives their compliance decisions and hence this perception
is the real key to the success of a regulatory default strategy. Elements that influence this perception include how the agency actually imposes penalties; whether
it says it will impose penalties; and how information about penalty imposition and
rhetoric is made public and, separately, publicized.
Close Substitutes
Like any other kind of rule, the operation of a high-penalty regime will also be
affected by the ability of taxpayers to avoid the whole scheme by making choices
that are sufficiently close substitutes for the penalized behavior.20 David Weisbach
has conceptualized the idea of minimizing close substitutes for a taxed activity as
the goal of minimizing the “marginal efficiency cost function,” which is lower if
fewer behavioral distortions result from the imposition of a tax.21 David Schizer
has categorized the factors that may determine whether a particular “friction”
prevents taxpayers from planning around a particular rule. Schizer notes that
strong and not-malleable frictions, which may come in the form of business choice
preferences, technology limitations, and legal and accounting costs, can hinder or
prevent the development of close substitutes.22 The absence of close substitutes or,
similarly, the existence of strong and inflexible frictions, is thus key to the success
of a high penalty strategy.
The problem with close substitutes is fairly clear for the deterrence and separation goals. To the extent that taxpayers can avoid a penalty, it will neither deter
their noncompliance behavior nor incent them to self-identify as compliers. The
problem with close substitutes for a signaling goal is somewhat different.
It is possible that the simple enactment of a new rule will serve to signal the
advent of a new norm, even if the rule is not enforced. This expressive theory
suggests that the key to compliance is the persuasion of a material portion of the
population to voluntarily obey the law. Once a voluntarily obedient group reaches
a tipping point, others will follow suit. Enforcement can play a role by raising the
salience of the new law and helping to achieve a tipping-point amount of compli-
6
Morse
ance by persuading rational actors who are susceptible to deterrence strategies to
comply, but the mere existence of the rule can also act as a signal.23
So if enforcement is not essential to the success of a signaling penalty, what
is the problem with close substitutes? The problem is that a close substitute can
function as a competing signal that undermines the signaling power of the enacted law, so long as the close substitute is sufficiently well known. Regulatees
may gather around the workaround rather than around the law as enacted, just as
motorists informally agree that driving several miles above the speed limit is close
enough. The process of gathering around a close substitute is expressive and peerenforced just as it could be for an enacted law.
Detection and Information Strategies
As Alex Raskolnikov has persuasively argued, a key task in tax administration is
to identify noncompliers who masquerade as compliers.24 This point is highly
relevant to a high-penalty regime, whether the high penalty is intended to serve
only the separation purpose that Raskolnikov identified in the context of menubased regulatory penalty default structures in tax administration or whether the
high penalty also functions as a deterrent and/or signal. The deterrence function
will also be frustrated if it is possible for noncompliers to hide behind the mask of
compliance as compliers. Signaling will be weaker, as well, if masked noncompliance is a known workaround, for the workaround can serve as a competing signal.
In tax administration, a high-penalty regime will typically include an information filing requirement that can serve to identify the compliers as those taxpayers
not subject to the high penalty. This information filing requirement presents a
key opportunity to increase the government’s ability to detect noncompliers who
masquerade as compliers. In the case of the FBAR rules considered here, it takes
the form of a specifically crafted information return.
In focusing here in this detection and information strategies point on the identification of noncompliers masquerading as compliers, I do not mean to dismiss
the importance of discovering and penalizing noncompliers. But the place for that
goal within this conceptual framework is in the consideration of whether taxpayers perceive penalties and rewards as credible possibilities. Assuming that they do,
and that they self-identify as compliers, the necessity of detection and information
strategies to determine whether they are telling the truth is a separate and important component of an effective high-penalty strategy.
The issue of detection is also separate from the question of whether the very
fact of self-identifying to the government as compliant, including through actions
as simple as signing one’s name to a regulatory filing, improves the chance that a
regulatee will comply. It probably does—commitment consistency is a powerful
heuristic.25 However, this detection point means to leave that to the side, and fo-
An Analysis of the FBAR High-Penalty Regime
7
cus on ways to improve the government’s ability to detect noncompliers amid the
compliance group.
One way that information filing can improve detection is through its interaction with audit strategies. In simplest form, regulatees who identify themselves as
compliers may be subject to more frequent or more thorough audit. This may be
a sufficient strategy for a small population of regulatees, if it is possible to craft the
audit approach in a way that does not interfere with the goal of rewarding compliant taxpayers with better service.
Larger populations of regulatees require an audit selection strategy that sorts
out compliance filers who are more likely to be in fact noncompliant. Part of
this can be based solely on the compliance information provided by regulatees, as
they can be sorted based on statistical information about the likelihood of compliance by regulatees who meet certain descriptive characteristics. This works only if
those characteristics are available in information provided to the regulating agency, and it works best if they are provided in a form that allows for the performance
of automatic information search functions.
A different audit-selection strategy may be available if there are alternative
sources of information about regulatees. Third-party reporting is the most prevalent in tax administration, but other “non-tax documentation” sources26—booktax balance sheet differences provide one example—might also be used. Strategies
here go beyond sorting based on a statistical model built from taxpayer-provided
data. Instead, the regulator may analyze different sources of data to check whether they match and/or to feed a richer statistical model of the likelihood of compliance. Because of the importance of interactions between alternative sources
of data and the taxpayer-provided information that signals compliance, careful
design of the compliance report to maximize its usefulness in combination with
other data sources will increase the chance of success for a high-penalty regime.
Especially under an assumption of limited administrative resources, efforts to
improve detection of noncompliers who masquerade as compliers may appear to
be at odds with the goal of increasing taxpayers’ perception of the credibility of
penalties and rewards. One is directed at vetting taxpayers inside the system, and
the other is targeted at finding taxpayers who are outside the system. Yet it is not
clear that the two enforcement efforts are diametrically opposed. In each case,
the truly important metric is taxpayers’ perception—in one case of the likelihood
of penalty imposition for noncompliers, and in the other case of the likelihood
of being found out if a noncomplier tries to masquerade as a complier. Publicity
of successful government enforcement efforts could, depending on how they are
absorbed by the taxpayer population, enhance both perceptions at the same time.
Or they could enhance only one and be neutral as to the other, or they could enhance one at the expense of the other. A clever publicity strategy would seek the
first option.
8
Morse
Applying the Framework to the Problem of Offshore
Account Information Asymmetry
A 2002 Treasury report estimated that there were about 1 million offshore accounts
held by U.S. persons and that less than 20 percent of foreign bank account reports,
or FBARs, were duly filed as required annually.27 A separate recent estimate suggests that these accounts might contain in the neighborhood of $1.5 trillion. The
tax collection shortfall resulting from the failure to pay tax on the income from
funds placed in unreported offshore accounts might amount to as much as $50
billion annually.28
The IRS has said that account holders come from “all walks” of (relatively
wealthy) life.29 One official has been reported as saying that of 50,000 accounts
targeted by a subpoena discussed below—which requested all accounts with U.S.
connections at a certain bank, without any filtering mechanism as to size or otherwise—a few thousand were enormous accounts of tens or hundreds of millions of
dollars, and the vast majority smaller, less than ten million dollars.
Offshore account holders include heirs, immigrants, and expatriates with some
personal connection to the location of their offshore account.30 Account holders
who lack any non-U.S. connection may have various reasons for forming the account, including misguided acceptance of an unscrupulous planner’s advice,31 or
nontax asset protection, as well as determined and conscious tax evasion. And
determined tax evaders may have legal or illegal sources for their deposited funds,
tax-paid or not.
Offshore account noncompliance presents a problem of information asymmetry, rather than an issue of legal uncertainty. It is perfectly clear that U.S. citizens
and residents must pay U.S. tax on their worldwide income, including income that
accrues to an offshore account.32 The challenge is to make offshore account holders disclose the relevant information. The FBAR rules attempt to do just that, in a
framework that threatens high penalties for nondisclosure. In the second part of
this paper, I apply the analytical framework developed above to evaluate whether
the FBAR regime can succeed, asking whether (1) taxpayers perceive that penalties
and rewards are credible, (2) close substitutes are absent, and (3) taxpayers perceive that the government can detect noncompliers masquerading as compliers.33
FBAR Reporting Could Succeed as a High
Penalty Regime
The FBAR
U.S. owners of offshore accounts must annually file Reports of Foreign Banks
and Financial Accounts, or FBARs, with respect to their non-U.S. holdings. This
An Analysis of the FBAR High-Penalty Regime
9
requirement links to the individual income tax return through Line 7 of Form
1040, Schedule B, which requires a taxpayer to specify whether he or she “had
an interest in or a signature or other authority over a financial account in a foreign country.” 34 The FBAR requirement is separate from recently enacted I.R.C.
§ 6038D—a “shadow FBAR” provision—which imposes similar self-reporting
requirements.35
A regulation36 authorized by a provision of the Bank Secrecy Act37 requires the
filing of FBARs. A central purpose of the Bank Secrecy Act is to collect information on financial transactions in order to track down money laundering related to
drug and other crimes.38 A neighboring statutory section requires banks to file
currency transaction reports, or CTRs, with respect to nonexempt bank transactions in excess of $10,000.39 The statute also contains other bank reporting40 and
self-reporting41 requirements.42
Despite the characterization of the Bank Secrecy Act as an anti-money-laundering statute, there are at least three partially overlapping concerns with offshore
accounts. First, the depositor may have illegally obtained the funds that go into an
account. Second, the depositor, whether or not he or she has obtained the funds
illegally, may not have properly paid taxes with respect to them. Third, the depositor may fail to pay taxes on the income from the accounts.43 The second and third
issues are tax enforcement concerns.
The FBAR regulations are broad. They require “every person subject to the
jurisdiction of the United States … having a financial interest in, or signature or
other authority over, a bank, securities or other financial account” to file a report.44 Under a de minimis rule, a report is required if the aggregate value of the
financial accounts exceeds $10,000 at any time during the calendar year. The form
instructions give more specifics, but retain the broad character of the regulatory
requirements, both with respect to the definition of persons required to report45
and with respect to the definition of accounts required to be reported.46 Filings
are required of entities such as corporations, partnerships, and trusts47 and with
respect to holdings in or through corporations, partnerships, trusts, or other entities.48 Taxpayers must report information that should be readily available to them:
the existence and size of an offshore account.49 This paper considers the core requirement to report bank accounts financially owned by individual U.S. taxpayers
directly or through a corporation or other entity over which the U.S. owner has
signatory authority.50
There are several civil and criminal statutory penalties specified for FBAR violations.51 This paper focuses on the civil willful violation penalty, which equals the
greater of $100,000 or 50 percent of the balance in the account “at the time of the
violation.” 52 This is a huge potential penalty, and significantly more than before
the statute was amended in 2004.53
The Financial Crimes Enforcement Network, or FinCEN, division of the
Treasury had enforcement responsibility for FBAR compliance until 2003,
10
Morse
when enforcement authority was transferred to the IRS under a Memorandum
of Understanding that did not explicitly anticipate the issuance of regulations.54
Perhaps in part for this reason, and certainly in part because other elements necessary for effective enforcement—such as a way to access information from foreign
banks—were not in place, FBAR enforcement activity did not immediately ramp
up.55 Greater attention began to be paid to the FBAR requirement, including the
submission of Treasury reports on the widespread noncompliance with the requirement under a 2001 statute.56 But a voluntary disclosure offer made in 2003,
which followed efforts to investigate offshore credit card issuers and encompassed
FBAR filing requirements, did not result in a big enforcement success.57
For taxable years beginning after March 18, 2010, new § 6038D of the code imposes a similar annual reporting requirement for “specified foreign assets” if the
total value of such assets is in excess of $50,000.58 This requirement is in addition to the banking-law-based FBAR reporting requirement.59 This paper focuses
mainly on the banking-law-based FBAR requirement rather than the § 6038D requirement, because FBAR reporting more clearly fits the high-penalty model that
I am concerned with in this paper, at least as long as a willfulness-based penalty is
perceived as a credible possibility.60
Applying the High-Penalty Analytical Framework
to the FBAR
PENALTY AND REWARD CREDIBILITY
The first part of this paper argued that penalty and reward credibility is one factor
necessary to support the success of a high-penalty regime as a deterrence, separation, and/or signaling mechanism. In the case of the FBAR, the government has
done a good job so far of establishing the credibility of penalties and rewards in
the minds of taxpayers. Government efforts to articulate and publicize applicable
penalties crystallized in litigation relating to accounts at Swiss bank UBS and in
the administration of the 2009 FBAR voluntary disclosure program.
UBS publicity leverages availability bias. Starting in 2007, a U.S. native and UBS
banker named Bradley Birkenfeld channeled evidence to the government of serious misconduct at the Swiss bank. He informed on the elaborate James-Bondworthy secrecy practices in the cross-border private banking division at UBS,
for example, “say[ing] he once transported diamonds, bought with client money
abroad, into the United States in a tube of toothpaste.” 61 Birkenfeld pled guilty
in June 2008 to conspiring to help wealthy American Igor Olenicoff evade taxes62
and, in August 2009, received a 40-month prison sentence.63
There ensued a criminal fraud case against UBS. The key to the case was the
deliberately designed UBS process for working around the “qualified intermediary,” or QI, agreement that UBS had entered into with the U.S. government.64 The
An Analysis of the FBAR High-Penalty Regime
11
main thrust of the QI agreement was to permit UBS to forward non-U.S. client
information to U.S. withholding agents in summary form and still obtain statutory
withholding exemptions or lower treaty-based withholding rates on the payments
of U.S.-source investment income to non-U.S. persons.65 But the QI agreement
also included a less-than-airtight provision that required UBS to disclose U.S. account holders to the U.S. government,66 and it was this provision that UBS helped
clients to deliberately plan around.67 The criminal case ended with a $780 million
fine and a deferred prosecution agreement in February 2009.68
The IRS then submitted a request for enforcement of a broad subpoena to disclose the names of more than 50,000 U.S. clients of UBS. In August 2009, after
the intervention of the Swiss government as amicus in the case and top-level negotiations, the civil case settled under an agreement requiring UBS to disclose
more than 4,000 names through the information exchange provisions of the U.S.Switzerland treaty.69 After considerable debate, the Swiss parliament approved the
agreement in the June 2010.70 As of August 2010, the IRS had received information
about 2,000 clients.71
The Justice Department used the UBS case to support the criminal prosecution
of a number of offshore account holders, and it obtained a number of plea bargains, which then supported well-executed availability-bias-based publicity.72 The
UBS case also helped the cause of the 2009 voluntary disclosure program targeted
at delinquent FBAR filers. The volume of publicity of the 2009 disclosure program
in contrast to the 2003 program is striking. One rough measure derives from the
indispensable Tax Notes database, a touchstone for tax practitioners. Before 2008,
only nine Tax Notes articles mentioned “FBARs.” Between September 2008 and
October 11, 2009, 58 articles did so—partly because the earlier settlement did not
focus as intensively on the FBAR as the central disclosure tool, but also because
practitioners had less to say about their clients’ compliance experience in 2003.
Other data is instructive as well. In a similar 2003 program targeting offshore
credit and debit card accounts, a total of about 1300 applications were filed.73 In
the 2009 program, almost 15,0000 applications were received.74 This is far fewer
than the estimated hundreds of thousands of unreported offshore accounts, and
also fewer than the 50,000 or so UBS accounts initially targeted by the U.S. subpoena, but several times more than the 4,000 or so accounts expected to be disclosed in the UBS settlement.
It remains to be seen whether there will be an enormous difference in the resulting number of criminal prosecutions. Contemporaneous with the 2003 program, reportedly a total of 10 individuals were prosecuted.75 As of April 2010,
about 15 taxpayers had been charged and most of those had pled guilty; 76 the IRS
had reported several months earlier that it was investigating “dozens” of taxpayers
in the aftermath of the voluntary disclosure program.77
A large wave of prosecutions would increase the persuasiveness of the FBAR
high-penalty regime, but the fact of a large number of cases is not dispositive, in
12
Morse
part because taxpayers’ estimation of the likelihood of being caught is a perception. A central purpose of audit and compliance publicity is to increase taxpayers’
or tax preparers’ perception of the risk of detection.78 These efforts should leverage the well-established cognitive availability bias, which prompts us to estimate
the “likelihood of an event on the basis of how quickly instances or other associations come to mind.” 79 Studies support the existence of an “indirect” audit effect
related to taxpayers’ decisions to comply because they hear news of others getting
caught.80 Estimates of the ratio between the dollars brought in because of other
taxpayers’ compliance compared to the additional collections resulting from the
audit itself are in the range of 11 or 15:1.81
Associations come more quickly to mind if the stories are familiar.82 Publicizing
famous and/or egregious taxpayers may produce some indirect audit effect, but it
should not be expected to maximize the possible effect, because many taxpayers
whom the government seeks to influence are neither famous nor egregious. To
take advantage of the powerful tool of availability bias, a publicity strategy should
effectively communicate to taxpayers that people like them get caught by the IRS
or settle with the IRS because of a fear of being caught.83
The 2008–2010 plea bargain publicity does a nice job of leveraging availability
bias. Historically, the IRS has managed to attract publicity mainly for the most
famous or egregious offenders (such as Leona Helmsley or tax protestors like Ed
Brown, who barricaded himself in his New Hampshire home against a Federal
agent siege). But in the UBS case, the media has run stories on plea bargains
entered into by offshore account holders whose stories are somewhat egregious,
but not the worst or largest stories out there. This average-rich-person storyline
maximizes the availability bias power of the plea bargain publicity.
Some of the taxpayers in the news for tax evasion through offshore accounts
are Forbes-400 rich.84 But featured taxpayers also include Steven Michael
Rubinstein, a Florida accountant with a UBS account allegedly worth “at least $6
million;” 85 and Robert Moran, Florida resident whose company builds and rents
yachts and the alleged owner of an account containing “at least $3.7 million,” 86
and Jeffrey Chernick, a New York resident who runs a toy company and concealed
“more than $8 million.” 87 They include Juergen Homann of Saddle River, New
Jersey, who runs a chemical company and allegedly concealed “about $6.1 million
in assets,” 88 John McCarthy, a Malibu businessman whose account allegedly held
“more than $1 million,” 89 and Roberto Cittadini, a retired Boeing sales manager
who pled guilty to “hiding nearly $2 million.” 90
These are not small numbers, but they are also not among the largest accounts
out there. Of the 52,000 UBS clients on the original summons list, one description put the number of “ultrawealthy” taxpayers with accounts worth “tens to
hundreds of millions of dollars” at several thousand and suggested that the government would focus its attention there.91 Yet that is not where all the action
has been. Smaller UBS clients were reportedly included on the list selected for
An Analysis of the FBAR High-Penalty Regime
13
disclosure.92 For a typical offshore account holder, the news about indictments
and plea bargains of the merely very wealthy, rather than the Forbes 400, has more
salience and taps more effectively into availability bias. The IRS should publicize different kinds of taxpayers that have gotten caught to the extent it legally
can.93 The government’s apparent focus on marshaling simple and easily decided
(or plea bargained) charges makes sense, as does its emphasis on continuing its
prosecution, plea bargain and publicity program94 and in covering banks other
than UBS,95 particularly in light of reports that Swiss bank clients may be moving
their accounts to other banks, for example in Singapore and Hong Kong.96 The
government appears aware of the need to broaden the net beyond UBS and has
instituted criminal proceedings against another large bank, HSBC, and at least
two of its clients.97
The IRS is fortunate in this case that various media outlets are following this
story closely, because Section 6103 of the Code, which prohibits the IRS from
disclosing confidential taxpayer “return information,” 98 limits the government’s
direct publicity efforts.99 The enumerated exceptions in the statute do not even
include explicit permission for the IRS to publicize return information that has
already been disclosed publicly, whether through a posted lien, civil or criminal
litigation, taxpayer discussion of the case in a public forum, or otherwise,100 although in light of the case law101 the IRS has gotten comfortable with the strategy
of posting basic press releases, or links to such releases, on its website.102
Voluntary disclosure penalty transcends legal uncertainty. The IRS approach
to its FBAR voluntary disclosure program also supported taxpayer perception of
credible penalties—subject to the close substitutes issue discussed below. In general, a valid voluntary disclosure is a full disclosure of unpaid tax, made before the
IRS has begun an investigation of the taxpayer and including a good faith undertaking by the taxpayer to pay all tax, interest and penalties determined by the IRS
to be due. The IRS will take such a disclosure into account in determining whether
to recommend criminal prosecution to the Justice Department.103
The government chose a high monetary penalty benchmark for this program.
In particular, it took the civil willfulness penalty equal to 50 percent of the account
balance for each annual failure to file,104 as its starting point. In addition to requiring taxpayers to file returns going back 6 years and pay all back taxes, interest, and
either accuracy or delinquency penalties,105 participants in the FBAR voluntary
disclosure program faced a maximum penalty of 20 percent of the account balance
for the year (of the six years covered) with the highest balance.106
The IRS stated unequivocally, and repeatedly, that in its view all taxpayers
who have failed to pay tax on income related to the offshore accounts—no matter
whether they are among the super-rich—are intentionally concealing income and
assets from the government rather than negligently remaining unaware of filing
and taxpaying obligations. The government declined to recognize a distinction
between business accounts and savings and investment accounts for purposes of
14
Morse
applying the 20-percent penalty in the voluntary disclosure program.107 Some
reports suggest that an anticipated reduction to a 5-percent penalty apparently
meant to apply to inherited accounts would be rarely, if ever, granted.108
The IRS wanted to “draw a clear line between those individual taxpayers with
offshore accounts who voluntarily come forward to get right with the government
and those who continue to fail to meet their tax obligations.” 109 But it certainly
did not suggest that those who came forward would be let off scot-free. The government indicated that taxpayers who voluntarily disclosed would not be recommended for criminal prosecution,110 but the 20-percent-of-account-value monetary fine, derived from the benchmark of the statutory willfulness penalty, is a
substantial amount.111
Remarkably, the government managed to establish 20 percent of the account
value as a credible penalty—in other words, it successfully publicized that penalty
level in its program, and voluntarily disclosing taxpayers appear to have accepted
it as a benchmark—despite legal uncertainty about how a court would apply the
willfulness standard in the offshore account situation. Under the Supreme Court’s
Cheek case, “willful” violation of a legal duty to file a tax form generally requires
that the defendant know of the legal duty.112 It is conceivable, given the historic
lack of publicity and enforcement about the FBAR filing requirements, that a defendant might be able to show a lack of willfulness.113 There is one circuit case
decided under the Cheek standard that rejected an “ostrich” defense theory in an
FBAR filing case, but it involved egregious facts.114
FBAR compliance rewards. As argued above in the description of the analytic
framework, perceived rewards for compliant taxpayers reduce the risks of crowding out compliant behavior and complement perceived penalties for noncompliant taxpayers, thus supporting the deterrence, separation and signaling goals of
a high-penalty regime. Existing features of U.S. tax administration, such as its
articulated service/enforcement goal and the tendency to treat historically compliant taxpayers more gently in the audit and Appeals process, serve as rewards for all
compliant taxpayers.115 Specific elements of the FBAR regime aimed at shaping or
explaining the compliance option could do an even better job in this specific case.
In particular, the government should keep FBARs confidential from third parties
and publicize more effectively the benefits of the compliance option.
The FBAR does not receive the confidentiality protection extended to most tax
filings. This is because it is not denominated a tax return for purposes of Section
6103, as it is not required by Title 26 of the U.S. Code, but rather by Title 31.116 It
is not clear whether an FBAR attached to a tax return would count as return information.117 The FBAR instructions direct that taxpayers not file FBARs with their
tax returns, but the voluntary disclosure guidance is less clear.118
In any case, carving FBARs out of Section 6103 is apparently intentional.
Although the taxpayer confidentiality provisions include some exceptions for
sharing information with other Federal agencies, the concept of the FBAR was
An Analysis of the FBAR High-Penalty Regime
15
to provide a more generally available database.119 A banking statute requires the
Secretary of the Treasury to develop “standards and guidelines” relating to who
has access to information administered by the FinCEN division of Treasury.120 In
2009, Senator Max Baucus considered proposing the classification of FBAR filings
as confidential tax return information.121
But even though increased FBAR confidentiality might conceivably act as a
compliance-inducing reward, it is not clear how the government could go about
ensuring this for taxpayers. Making FBARs tax return information by statute
has the disadvantage of undermining the access of other agencies to FBAR information, in contravention of the original law’s intention. But without a statutory
amendment, the government presumably cannot promise that it will keep FBARs
secret against, say, Freedom of Information Act requests or court orders emerging
from civil litigation to the same extent tax return information is kept secret under
I.R.C. § 6103.122 The FBAR form warns of possible information sharing with other
“state, local and foreign” government entities but is silent on the question of sharing with other third parties.123 The advantage of extending rewards to compliant
taxpayers suggests that Treasury should strongly resist any non-government third
party information requests, and, of course, publicize any wins.
A better-publicized explanation of what happens to taxpayers who choose the
compliance option would also increase the power of compliance rewards to shape
compliance behavior. The main available pieces of information are the penalty
limit of 20 percent of account value indicated in the 2009 voluntary disclosure
program and the 25 percent penalty benchmark put forth in the follow-on 2011
program. Compliant taxpayers presumably enjoy other benefits, such as the peace
of mind that comes from getting right with the government and (hopefully) cordial and competent handling of the FBAR filings and related matters. However,
clarifications of the rewards for compliance face two central challenges: taxpayer
confidentiality and menu complexity.
Taxpayer confidentiality concerns limit the government’s ability to tell salient
stories about taxpayers who choose to comply. Even the broadest view regarding the ability of the IRS to disclose information also available in public records
would not permit the IRS to publicize taxpayers whose cases are not litigated or
otherwise publicized, such as through liens. Public discussion by the taxpayer, for
example, does not waive the confidentiality protection.124 The statute does permit
disclosure to persons designated in writing by the taxpayer.125 Accordingly, an
explicit waiver of taxpayer confidentiality and permission to publicize might work
to permit the IRS to disclose specific taxpayer information. But getting the waiver
and connecting it to a publicity strategy would be a time-consuming and often
futile case-by-case exercise.126
Publicity of different categories of taxpayers who, for example, settled with the
IRS would likely be permitted under the flush language of Section 6103 (b) (2),
which excludes from the definition of protected return information “data in a form
16
Morse
which cannot be associated with, or otherwise identify, directly or indirectly, a
particular taxpayer.” 127 Typically this rule—the Haskell amendment—is used to
permit “statistical studies or other compilations of data,” as Senator Haskell explained when proposing it on the floor.128 The Supreme Court has held that it cannot support the disclosure of return information from which identifying details
have merely been redacted.129
However, several courts have concluded that information assembled in a more
granular way than the macro-level IRS statistics on income tables might fit within
the Haskell amendment’s description of data that falls outside the return information definition. For example, the Ninth Circuit held that Section 6103 did not
block a FOIA request from logging companies for a report the IRS had prepared
on tax preparation in their business.130 In another case, the Court of Federal
Claims, in response to a discovery request from an oil company seeking information about production methods of other companies claiming a Section 43 credit,
concluded that “[a] list of the various production methods could be complied. If
only this list, and no other information, were delivered to plaintiff, then Section
6103 would not be violated.” 131
The IRS can describe compiled data in a more engaging way than in tabular
statistical form without violating Section 6103. In particular, it should be able
to describe general types of offshore account taxpayers with the goal of more effectively communicating the possibility of audit and prosecution and the benefits
of disclosure and settlement. It need not stick to dry categorical descriptions.132
More creative and salient tactics are needed. The government should consider
fictional portrayals, taxpayer testimonials, or more abstract, but salient, messages
about the different results produced by the compliance and penalty regimes.133
THE CLOSE SUBSTITUTE OF QUIET DISCLOSURE
As the first part of this paper discussed, the problem of close substitutes can also
bar a high-penalty regime from achieving its deterrence, separation, and/or signaling goals. This is an issue for the FBAR filing requirement. The possibility of
a “quiet disclosure” option may exist as a close substitute alternative to voluntary
disclosure.
“Quiet disclosure” is the practice of simply filing amended tax returns for the
years in question.134 It is not endorsed by any government guidance, in contrast
to official “voluntary disclosure,” which is described in the Internal Revenue
Manual.135 Voluntary disclosure includes a list of conditions—and features an undertaking by the IRS to consider the fact of disclosure when deciding whether to
forward a case to the Justice Department for criminal prosecution, such as for tax
evasion. In practice, it is generally thought that voluntary disclosure bars criminal
prosecution.136
Even though quiet disclosure is not officially endorsed,137 it is a fairly well established practice,138 and taxpayers’ expectation that quiet disclosure offers at least
An Analysis of the FBAR High-Penalty Regime
17
some protection against criminal prosecution is also well entrenched.139 This
presents a problem for the integrity of the high-penalty FBAR rules, because the
quiet disclosure option probably will not subject the taxpayer to the significant
willful-failure-to-file-derived penalties that the IRS has applied to voluntarily disclosing taxpayers. The quiet disclosure option weakens the ability of the highpenalty FBAR regime to serve its deterrence, separation and signaling functions.
The deterrence power of the FBAR, grounded in taxpayers’ comparison of the
risks and rewards of filing and not filing, depends on taxpayers’ belief that failure
to file the FBAR will lead to the government imposing penalties and withholding rewards. A no-penalty quiet disclosure option would suggest that there is
little cost to failing to file the form initially, and that the taxpayer may wait to see
whether the government seems to have the ability to discover his or her offshore
accounts by other means. If the government does, then quiet disclosure is an easy
solution.140
The separation goal of a high-penalty system is similarly undermined by the
quiet disclosure option. Compliant taxpayers might choose up-front compliance,
by filing the FBAR, or delayed compliance, through quiet disclosure. The quiet
disclosure option does not clearly identify compliant taxpayers in the way that
filing an FBAR does, and therefore makes it more difficult for the government to
target taxpayer service or tailored detection strategies to the compliance group.
The signaling potential of the high-penalty FBAR system is also muffled by the
availability of quiet disclosure, because quiet disclosure constitutes a competing
signal around which taxpayers may gather instead.
To permit FBAR reporting to function as a high-penalty regime that promotes
deterrence, separation and signaling, this quiet disclosure close substitute should
be removed. The government has taken the first step toward doing so, by providing that it will not respect quiet disclosure—in contrast to voluntary disclosure—as a reason to refrain from criminal prosecution in the offshore account
context.141 But taxpayers’ perception is what counts. So the plan for eliminating
a quiet disclosure option should include appropriate, availability-bias-motivated
publicity, such as publicity of taxpayers subject to civil and/or criminal penalties
despite efforts at quiet disclosure.
DETECTION AND INFORMATION STRATEGIES
A key possible weakness in a high-penalty regime is the possibility that taxpayers
who wish to game the system may pretend to be compliers.142 Excellent audit of
FBAR filers is therefore essential, as is publicity of successful audit. The availability of data and the nature of the FBAR filing group as a small population with
established publicity avenues can shape the audit strategy in the case of the FBAR.
In the short term, until third-party data can be used to cross-check the accuracy of FBAR filing, audit filters must derive from statistical models containing the
18
Morse
information on FBAR filings themselves, together with other predictive variables
such as reported Form 1040 income level and demographic characteristics. The
shadow FBAR tax return filing mandated for taxable years starting after March
18, 2010 by § 6038D is therefore important to the audit project. This is because
taxpayer confidentiality limitations restrict the IRS’s ability to use tax return information to enforce FBAR requirements. The shadow FBAR filing required under
§ 6038D is intended to solve this problem and permit the IRS to develop a program to automatically match § 6038D data with other tax return information.143
Fortunately, the taxpayers targeted by the FBAR filing requirement are not an
enormous group—perhaps one million or so. The actual audit rate for wealthier
taxpayers—6.42 percent for Fiscal Year 2009 for taxpayers with income in excess
of $1 million—exceeds substantially the 1.03 percent rate for individual taxpayers
on average.144 And the IRS has formed a special group to coordinate offshore account examinations for high-net-worth individuals.145 The small size and high
net worth characteristics of the target population also facilitate effective publicity.
In fact, the government has a proven publicity strategy: the distribution of press
releases that national and international newspapers then report on. It is likely
that this publicity and newspaper coverage reaches some significant portion of the
taxpayers required to file FBARs.
The possible future availability of third-party data, perhaps from non-U.S.
banks or governments, should shape the way in which the government collects
FBAR and shadow FBAR data now. In particular, data fields should be simplified
in anticipation of establishing a standardized global format for third-party reports
in the future. The essential contents of an FBAR or shadow FBAR form filed on
behalf of an individual usually can be reduced to four information fields: taxpayer
identity, which should often reduce to a TIN; the identity of the financial institution at which the account is held; the maximum value of the account for the year;
and the account number.146 Even if electronic filing—which would require statutory authorization147—is not yet feasible, assigning numeric codes for these fields
would facilitate data entry and sorting based on paper source documents. For
example, foreign financial institutions should have identification numbers to be
used on FBAR and other filings.148 Without these simplification and automation
measures, the government may face a situation where it has gobs of paper FBAR
information about taxpayers and does not know what to do with it.149 And it may
also find it more difficult than necessary to crosscheck FBAR filings against information provided through a global reporting system, if and when such a system
ultimately develops.
Conclusion
The FBAR rules have the necessary ingredients to support the high-penalty compliance mechanisms of deterrence, separation, and signaling. But to maximize
An Analysis of the FBAR High-Penalty Regime
19
their effectiveness, the government should adjust several aspects of its administration of the rules. Tax administrators should continue to work to increase taxpayers’ perception of the credibility of the penalties and rewards specified under the
FBAR system, by expanding the reach of their criminal and civil investigations to
other banks and by publicizing both cases where taxpayers failed to file FBARs and
got caught and also the advantages of compliance. They should defend third-party
confidentiality to FBAR filers. They should also eliminate the close substitute option of quiet disclosure as a remedy for the failure to file an FBAR.
Finally, the government should pursue the goal of increasing taxpayers’ perception of the likelihood that noncompliers who masquerade as compliers will be
detected. This last goal should involve good audit coverage of FBAR and shadow
FBAR filers, publicity of successful audits to the extent consistent with taxpayer
confidentiality limitations, and the development of a limited number of standardized, numerically coded data fields for FBAR and shadow FBAR reports which
may ultimately be cross-checked against global information reports about U.S. account holders.
Acknowledgements
The author began this project while serving as Research Assistant Professor at
Santa Clara University School of Law. Many thanks to my discussant, Leandra
Lederman, and to participants at the June 2010 IRS Research Conference. Thanks
also for helpful comments to Joe Bankman, Pat Cain, David Gamage, Mark Gergen
and Dennis Ventry; and to participants at presentations at Santa Clara University
School of Law and the University of California Hastings College of the Law.
Endnotes
1
See Michael G. Allingham & Agnar Sandmo, Income Tax Evasion: A
Theoretical Analysis, 1 J. Pub. Econ. 323 (1972).
2 See id.
3 See Alex Raskolnikov, Revealing Choices: Using Taxpayer Choice to Target
Tax Enforcement, 109 Colum. L. Rev. 689, 704–05 (2009) (describing the
separation function of penalties and noting the targeting benefit of avoiding
crowding out); see also Dan M. Kahan, The Logic of Reciprocity: Trust,
Collective Action, and Law, 102 Mich. L. Rev. 71, 83–84 (2003) (suggesting
that emphasizing the possibility of audit will not encourage reciprocal
compliance behavior); Marjorie Kornhauser, Normative and Cognitive
Aspects of Tax Compliance: Literature Review and Recommendations for
the IRS Regarding Individual Taxpayers, in 2 Nat’l Taxpayer Advocate,
20
4
5
6
7
8
9
10
Morse
2007 Annual Report to Congress 151 (2008) (noting that “commoditization
of a behavior” can “crow[d] out positive normative influences on that
behavior”).
See Robert Cooter, Expressive Law and Economics, 27 J. Leg. Stud. 585,
607 (1998) (“Law provides an instrument for changing social norms by
expressing commitments.”); Eric A. Posner, Law and Social Norms: The
Case of Tax Compliance, 86 Va. L. Rev. 1781, 1798–99 (2000) (observing that
the government is a player in the norm creation game and can send signals
that affect taxpayer behavior); Cass Sunstein, Legal Interference with Private
Preferences, 53 U. Chi. L. Rev. 1129, 1137–38 (1986) (explaining that law helps
to determine preferences and exploring related democratic theory problems).
See Dan M. Kahan, Social Influence, Social Meaning and Deterrence, 83 Va.
L. Rev. 349, 365 (1997) (arguing that individuals decide whether or not to
commit crimes in large part based on their perception of others’ criminal
behavior).
See James Alm, Gary H. McClelland and William D. Schulze, Changing
the Social Norm of Tax Compliance by Voting, 52 KYKLOS 141, 153, 161
(reporting increased compliance if experimental subjects were permitted
to communicate about their compliance decisions); Michael Wenzel,
Motivation or Rationalisation? Causal Relations Between Ethics, Norms
and Tax Compliance, 46 J. Econ. Psychol. 491, 504–05 (2005) (reporting
longitudinal study results showing that group norms affect personal ethics
when a taxpayer identifies with the group).
See Robert Cialdini, Influence: Science and Practice 61–90 (4th ed.
2001) (detailing results of commitment consistency studies).
See, e.g., Eric Fleisig-Greene, Law’s War With Conscience: The Psychological
Limits of Enforcement, 2007 B.Y.U. L. Rev. 1203, 1222, 1233-1235 (2007)
(arguing that law can have an adverse impact on previously existing positive
norms and citing one empirical study suggesting that taxpayers who received
letters notifying them of a likely audit reported less income than other
taxpayers).
See, e.g., Marjorie Kornhauser, supra note 3, at 164 (arguing that a
commodified exchange view of taxation can undermine voluntary
compliance and prompt taxpayers to believe that the government is not
fair); Marjorie Kornhauser, Tax Compliance and the Education of John
(and) Jane Q. Taxpayer, Tax Notes 737, 739 & n.21 (Nov. 10, 2008) (noting
the importance of procedural fairness and reciprocal trust for compliant
taxpayers).
See Kahan, supra note 3, at 83-84 (noting the inconsistency of audit threats
with a tit-for-tat relationship).
An Analysis of the FBAR High-Penalty Regime
11
12
13
14
15
16
17
21
See Susan Cleary Morse, Using Salience and Influence to Narrow the Tax
Gap, 40 Loy. U. Chi. L. J. 483, 510–512 (2009) (arguing for penalties and
rewards commensurate with taxpayer behavior).
This idea of government-taxpayer reciprocity is distinguishable from
the idea of taxpayer-taxpayer reciprocity connected to the common
provision of public goods. See Sagit Leviner, An Overview: A New Era
of Tax Enforcement—From “Big Stick” to Responsive Regulation, 2 Reg.
& Governance 360, 365 (2008); Dennis J. Ventry, Jr., Cooperative Tax
Regulation, 41 Conn. L. Rev. 431, 436 (2009).
The IRS website, for example, plainly states its service and enforcement
mission. See IRS, The IRS Mission, available at http://www.irs.gov/irs/
article/0,,id=98141,00.html.
See IRS Announcement 2005-87, 2005-50 I.R.B. 1144 (anticipating
government-taxpayer cooperation in the CAP early issue resolution
program); Cliff Jernigan, Corporate Tax Audit Survival: A View of
the IRS Through Corporate Insider Eyes 76-77 (2005) (explaining that
the IRS invited taxpayers with a “history of honest dealings” to participate in
CAP).
See, e.g., Ian Ayres & John Braithwaite, Responsive Regulation
44-47 (1992) (arguing that an agency that threatens serious punishments
may be “vulnerable to a litigious firm determined to shatter its myth of
invincibility”).
For example, although the IRS has broad statutory powers to summon
documents, see I.R.C. § 7602; and the Supreme Court has vindicated its
authority to use these powers to summon tax accrual workpapers prepared
by accountants, see United States v. Arthur Young, 465 U.S. 805, 816 (1984),
the IRS has historically followed a “policy of restraint” under which it will
only seek workpapers “to obtain collateral sources of data, not to fish for
new issues.” Thomas J. Monks, Your Papers, Please: Requests for FIN 48
Workpapers, 125 Tax Notes 901, at nn. 72–75 (Oct. 28, 2009). This restraint
may stem from habit as well as from a desire to dodge litigation risks, limit
exposure to restrictive statutory changes, and/or avoid souring relationships
with taxpayers. For a summary of recent developments in the area of tax
accrual workpapers and a view that they can never constitute protected work
product, see Dennis J. Ventry Jr., A Primer on Tax Work Product for Federal
Courts, 123 Tax Notes 875 (May 18, 2009).
See Bryan T. Camp, Tax Administration as Inquisitorial Process and the Partial
Paradigm Shift in the IRS Restructuring and Reform Act of 1998, 56 Fla. L.
Rev. 1, 79 (2004) (describing the public perception of an overzealous IRS that
led to the passage of a statute curtailing the agency’s power).
22
Morse
18
Cf. Joseph S. Nye, Jr., Soft Power 63–65 (2004) (connecting “unilateral”
approach to American foreign policy and “decline of America’s attractiveness
abroad”).
See, e.g., Ayres & Braithwaite, supra note 15, at 44–47 (identifying that
regulatees’ perception of an agency’s “invincibility” as a key factor).
See, e.g., Leandra Lederman, Reducing Information Gaps to Reduce the
Tax Gap: When Is Information Reporting Warranted?, 78 Ford. L. Rev.
1733, 1740–41 (2010) (arguing that the absence of alternative arrangements
increases the likelihood of success of an information reporting provision).
See David Weisbach, Line Drawing, Doctrine and Efficiency in the Tax Law,
84 Corn. L. Rev. 1627, 1665–1668 (1999) (defining marginal efficiency cost of
funds as the ratio between the revenue from a tax change with no behavioral
distortion and the actual (presumably lower but still positive) revenue
including the impact of behavioral effects).
See David Schizer, Frictions as a Constraint on Tax Planning, 101 Colum. L.
Rev. 1312, 1323–34 (2001) (noting the strength and malleability properties of
frictions and listing different possible sources of frictions including business,
technology, and legal constraints).
See Cooter, supra note 4, at 595 (explaining an expressive theory of
enactment and enforcement of law).
See Raskolnikov, supra note 3, at 724–728 (exploring several ways to increase
the likelihood of detection in the compliance group).
See, e.g., Kahan, supra note 5, at 358–59 (noting that the desire to avoid
cognitive dissonance motivates individuals to conform their behavior).
Joshua D. Blank, Overcoming Overdisclosure: Toward Tax Shelter Detection,
56 U.C.L.A. L. Rev. 1629, 1685–86 (2009).
See Sec’y of Treas., A Report to Congress in Accordance with
§ 361(b) of the Uniting and Strengthening America by Providing
Appropriate Tools Required to Intercept and Obstruct Terrorism
Act of 2001 6 (noting that the IRS estimated the number of foreign bank
accounts at 1 million and the number of annual FBAR filings at about
180,000). The IRS reported that 322,414 FBARs were filed in 2007. See IR
2008–79. Without more information about the number of offshore accounts,
which may have increased, the frequency with which one FBAR filer listed
more than one account or more than one FBAR filer reported one account
and so forth, better estimates are difficult to produce.
See Martin Sullivan, U.S. Citizens Hide Hundreds of Billions in Cayman
Accounts, 103 Tax Notes 956 (May 24, 2004) (citing $70 billion estimate);
Hearing on Issues Involving Banking Secrecy Practices and Wealthy American
19
20
21
22
23
24
25
26
27
28
An Analysis of the FBAR High-Penalty Regime
23
Taxpayers, 111th Cong., 1st Sess. (Mar. 31, 2009) (statement of Reuven S. AviYonah) [hereinafter Avi-Yonah Statement] (assembling various estimates
of cash and securities deposits offshore and translating those figures into
an estimate of approximately $50 billion in unpaid U.S. Federal income
tax annually). The Avi-Yonah calculation includes the assumption that the
accounts earn a 10% return. See id.
The overall 2001 tax gap estimates were based on National Research Program
(NRP) audit studies. See Eric Toder, What is the Tax Gap?, 117 Tax Notes
367, 370–74 (Oct. 22, 2007) (describing NRP estimation methodology). The
cited estimates of offshore account noncompliance derive from independent
estimates of the size of offshore accounts held by U.S. individual taxpayers,
and according to one report “it is doubtful that the $345 billion estimate
includes the entire international tax gap.” Treas. Inspector Gen’l for Tax
Admin., A Combination of Legislative Actions and Increased IRS
Capability and Capacity Are Required to Reduce the Multi-Billion
Dollar U.S. International Tax Gap (Jan. 27, 2009), available at http://
www.treas.gov/tigta/iereports/2009reports/2009IER001fr.html. Despite the
different methodology and incomplete overlap, comparing the $350 billion
measure of overall noncompliance to the $50 billion estimate of offshore
account noncompliance should give an idea of the general size of the tax gap
problem and relative importance of its components, as intended.
29 See IR 2003-95 (July 30, 2003) (“People from all walks of life applied for the
[2003 voluntary disclosure] program, including lawyers, dentists, business
executives, estate heirs, and numerous other occupations.”).
30 See Jeremiah Coder, IRS Extends Offshore Voluntary Disclosure Deadline, 124
Tax Notes 1297 (Sept. 28, 2009) (reporting practitioner report of inquiries
from “immigrants or their children who created accounts before coming
to the United States”). Cf. Fred Feingold, Further Guidance Needed on Who
Must Report Foreign Accounts, 123 Tax Notes 1023 (May 25, 2009) (arguing
that many FBAR nonfilers fail to file due to ignorance of the requirement).
31 At a Senate committee hearing in 2002, for example, lawmakers heard
testimony from an orthopedic surgeon and Federal inmate. He had gotten
into financial trouble, refused the offers of several tax protestor promoters,
and then entered into a offshore “business trust” arrangement supported
by “legal opinions and letters from several attorneys.” He thought things
were legal, he claimed, until he discovered that the trust routed funds from
Utah to the Isle of Man and then to Austria and provided false receipts for
the funds. He stated that he was attempting to extricate himself from the
situation when he was found out. Transcript of Hearing on Schemes, Scams
and Cons, Part II: The IRS Strikes Back, 107th Cong., 2d Sess. (Apr. 11, 2002)
16–22 (statement of Dr. Daniel Bullock).
24
Morse
32
A U.S. citizen or resident alien may exclude certain income earned abroad
from the performance of services, but this foreign earned income exclusion
does not exempt investment income from U.S. tax. See I.R.C. § 911. One
of the requirements for the foreign earned income exclusion is that the
individual demonstrate his or her substantial foreign presence abroad
by meeting either the bona fide residence test or the 330-day test. See
I.R.C. § 911(d). Such a taxpayer may also be able to reduce his or her U.S. tax
by the amount of foreign taxes paid, but must record the foreign tax credit
claim on the tax forms submitted to the U.S. government. See I.R.C. § 901
(providing for foreign tax credit election); IRS, Form 1116 (enabling foreign
tax credit election); IRS, Form 1116 Instructions at 1 (explaining that in
certain circumstances a taxpayer with “passive category income” only may
claim a foreign tax credit on line 47 of From 1040 without filing Form 1116).
In most offshore account situations, in any case, there is no foreign tax to
credit.
Another law, the Foreign Account Tax Compliance Act, or FATCA, also
attempts to solve the problem of offshore account information asymmetry.
FATCA was enacted as part of a larger jobs-oriented tax package in March
2010. See P.L. 111–147 § 501 (codified at I.R.C. §§ 1471–1474). It requires nonU.S. financial institutions to tell the U.S. government about their U.S. account
holders and includes very high penalties for noncompliance. In future work,
I hope to analyze FATCA using the framework developed here.
IRS Form 1040, Schedule B, Line 7.
See I.R.C. § 6038D. The provision is effective for tax years beginning after the
date of enactment, March 18, 2010. See P.L. 111–147 § 511(c). See also infra text
accompanying notes 58–60 (contrasting § 6038D and FBAR requirements).
See 31 C.F.R. § 103.24.
See 31 U.S.C. § 5314. Section 5314 was enacted in 1982. See Pub. L. 97–258,
Sept. 13, 1982, 96 Stat. 997. See generally Bank Secrecy Act of 1970 (“BSA”),
Pub. L. No. 91–508, 84 Stat. 114 (1970) (codified as amended at 12 U.S.C.
§§ 1829b & 1951–59 and 31 U.S.C. 5311-5332; Steven Mark Levy, Federal
Money Laundering Regulation: Banking, Corporate and Securities
Compliance § 3.02 [B] (2003 & Supp. 2009) (explaining that Title I of the
BSA, codified in the 12 U.S.C. sections, requires banks to maintain certain
records and that Title II, codified in the 31 U.S.C. sections, requires certain
reporting of “secret foreign bank transactions”).
See Levy, supra note 37, at § 3.02 (“The grande dame of money laundering
regulation is the statute commonly known as the Bank Secrecy Act of 1970.”).
See 31 U.S.C. § 5313 (a); 31 C.F.R. § 103.22.
33
34
35
36
37
38
39
An Analysis of the FBAR High-Penalty Regime
40
41
42
43
44
45
46
47
48
25
See, e.g., 31 U.S.C. § 5318 (g) (regarding “Suspicious Activity Reports,” or
SARs).
See 31 U.S.C. § 5316 (a) (requiring persons who physically transport more than
$10,000 cross border to make a “Currency and Monetary Instrument Report”,
or CMIR).
See generally Levy, supra note 37, at § 6.02 (providing an overview of
reporting requirements).
The legislative history of the 1970 enactment of the Bank Secrecy Act
includes a concern for these tax evasion issues. Swiss bank accounts are not
a recent phenomenon. “One of the most damaging effects of an American’s
use of secret foreign financial facilities is its undermining of the fairness of
our tax laws. Secret foreign financial facilities, particularly in Switzerland,
are available only to the wealthy. To open a secret Swiss account normally
requires a substantial deposit, but such an account offers a convenient means
of evading U.S. taxes. … [I]t is grossly unfair to leave the secret foreign bank
account open as a convenient avenue of tax evasion.” H.R.Rep. No. 91–975,
reprinted in 1970 U.S.C.C.A.N. 4394, 4397–98 (1970).
31 C.F.R. § 103.24.
The definition of a person subject to the jurisdiction of the United States
is narrowed in the form instructions to “a citizen or resident of the United
States, or a person in or doing business in the United States.” Instructions
to TD F 90-22.1 [hereinafter FBAR Instructions] at 6. See Boris I. Bittker
& Lawrence Lokken, Federal Taxation of Income, Estate & Gifts
¶ 65.5.8, available at Lawrence Lokken, The Big, Bad FBAR: Reporting Foreign
Bank Accounts to the U.S. IRS 2-4 (2009), available at http://ssrn.com/
abstract=1429744 (describing persons required to report and interpretation
of “doing business” language). Proposed regulations and other guidance also
narrow FBAR filing requirements somewhat. See infra notes 48 and 50.
See FBAR Instructions, at 6 (defining financial account).
An entity account may be required to be reported because of a U.S. person’s
financial interest in or signatory authority over such account. See FBAR
Instructions at 6–7 (describing financial interest and signatory authority
rules); Bittker & Lokken, supra note 45, at ¶¶ 65.5.8.4 and 65.5.8.5 (same).
Various other requirements to report ownership in and transactions with
foreign entities also exist. See IRS Forms 5471, 5472, 3520-A, 8865.
Recent government comments and guidance have expanded practitioners’
previous understanding of the breadth of the financial account definition.
See FBAR Instructions at 6 (providing that financial accounts “generally
also encompass any accounts in which the assets are held in commingled
funds”); see also Letter from New York State Bar Association to Neal S.
26
49
50
51
52
53
Morse
Wolin, Deputy Secretary, Department of the Treasury, et al. (July 17, 2009),
available at LEXIS, TNT library, 2009 TNT 137–13, [hereinafter NYSBA July
17, 2009 Letter] at text accompanying notes 13–22 (charging that a “flurry” of
informal guidance and media comments indicating that holders of accounts
in commingled funds such as non-U.S. hedge funds or private equity
funds presented a “tension” with earlier guidance and took practitioners
by surprise). Proposed regulations, however, “reserve the treatment of
investment companies other than mutual funds or similar pooled funds”
despite some concerns about the possible use of entities such as hedge funds
for tax evasion. See Financial Crimes Enforcement Network: Amendment to
the Bank Secrecy Act Regulations—Reports of Foreign Financial Accounts,
75 Fed. Reg. 8844, 8846 (proposed Feb. 26, 2010) (to be codified at 31
C.F.R. Part 103). See also Notice 2009–62, 2009–35 I.R.B. 260 (requesting
comments on various FBAR filing requirements).
The FBAR form requires the reporting of the maximum amount in the
account during the year reported. See Treasury Department Form TD F
90.22–1.
Proposed regulations and other guidance would not disturb the FBAR filing
requirement in this paradigm case. See Financial Crimes Enforcement
Network: Amendment to the Bank Secrecy Act Regulations—Reports of
Foreign Financial Accounts, 75 Fed. Reg. 8844 (proposed Feb. 26, 2010) (to
be codified at 31 C.F.R. Part 103); Notice 2010–23 (providing administrative
relief for FBAR filing requirements such as for certain signatories without
financial interest in the account); Announcement 2010–16 (suspending FBAR
filing requirement for certain non-U.S. persons).
There is also a voluminous list of possible penalties for tax evasion and other
offenses that may be linked to failure to file an FBAR. See IRS, Voluntary
Disclosure Questions and Answers, available at www.irs.gov/newsroom/
article/0,,id=210027,00.html, Q&A 14 and 15 (listing possible civil and
criminal penalties).
See 31 U.S.C. § 5321(a)(5) (specifying willful civil penalty). See also 31
U.S.C. § 5321 (a) (5) (B) (specifying $10,000 civil penalty with reasonable
cause exception); 31 U.S.C. § 5322 (a) and (b) (specifying criminal penalties
including imprisonment). See generally Bittker & Lokken, supra note 45,
at ¶ 65.5.8.7 (summarizing penalties).
The legislative history indicates that the increased penalty responded
to Treasury’s reporting of widespread disregard for the FBAR filing
requirement. See Joint Committee on Taxation, General Explanation
of Tax Legislation Enacted in the 108th Congress 377–38 (2005)
(proving explanation for § 821 of the Act, codified at 31 U.S.C. § 5321 of the
Code.
An Analysis of the FBAR High-Penalty Regime
54
55
56
57
58
59
60
27
See IRS News Release 2003–48; 31 C.F.R. § 103.56 (g). See also NYSBA July
17, 2009 Letter, supra note 48, at text accompanying notes 6–7 (describing
delegation of authority).
One 2008 Tax Court case, based on tax years 1993–2000, is notable for the
imposition of FBAR penalties in the presence of particularly egregious
facts and for the Tax Court’s lack of jurisdiction to review the propriety of
the imposition of FBAR penalties by the IRS. See Williams v. Comm’r, 131
T.C. 54, 58–59 (2008) (finding no jurisdiction in the absence of a notice
of deficiency, lien or levy); T.C. Memo 2009–81 (concluding on summary
judgment that Williams’ criminal tax fraud guilty plea collaterally estopped
him from contesting civil fraud charges).
The reports were filed under section 361 (b) of the USA Patriot Act of 2001,
which “requires the Secretary of the Treasury to (1) study ways to improve
compliance with the reporting requirements set forth in Section 5314, and (2)
submit an annual report on the study to Congress.” Levy, supra note 37, at
§ 10.02.
The offshore credit card initiative of 2000–2003 sought information about
credit card holders from MasterCard, Visa and other payment processors.
See, e.g., John Hembera, IRS Targets AmEX, MasterCard in Offshore Fishing
Expedition, Tax Notes (Oct. 26, 2000). In general that initiative did not face
a bank secrecy obstacle, since it targeted U.S. payment processors. See, e.g.,
Dorsey v. United States, 2004–1 U.S.T.C. ¶ 50,164 (D. Md. 2004) (refusing
to quash summons under § 7602; bank secrecy issue not raised). However,
it culminated in only 10 or so prosecuted cases, plus settled cases that did
not get publicized; it is reportedly considered not a great success. See Rev.
Proc. 2003–11, 2003–1 C.B. 311 (announcing offshore initiative directed in
part at credit cards); Lee Sheppard, Now What? Dealing With UBS Account
Disclosures, 124 Tax Notes 847, 851–52 (Aug. 31, 2009) (recalling results of
credit card initiative); Heather Bennett, IRS Offshore Compliance Initiative
Collects $170M, 102 Tax Notes 713 (Feb. 9, 2004) (reporting that the
initiative collected 1300 applications and $170 million).
I.R.C. § 6038D (a).
See Joint Committee on Tax’n, Technical Explanation of the
Revenue Provisions Contained in Senate Amendment 3310,
the “Hiring Incentives to Restore Employment Act,” Under
Consideration by the Senate 60 (Feb. 23, 2010) (noting that § 6038D does
not modify or replace the FBAR requirements).
The basic § 6038D penalty is $10,000, increasing to a maximum of $50,000
after notification by the Secretary. See I.R.C. § 6038D (d) (providing $50,000
maximum for “any failure,” presumably meaning a limit for each annual
28
61
62
63
64
65
Morse
failure to file). Another provision increases substantial underpayment
penalty for “any transaction involving a foreign financial asset” from 20
percent to 40 percent. See I.R.C. § 6662 (b) (6). But these penalties do not
approach the size of FBAR penalties such as the 50 percent of account value
willful civil penalty and the possibility of imprisonment.
Evan Thomas & Mark Hosenball, Cracking the Vault, Newsweek (Mar. 23,
2009).
See Lynnley Browning, U.S. Said to Expand Tax Inquiry, N.Y. Times, (Dec. 1,
2008).
See Joanna Chung, Former UBS Banker Given Prison Term, Fin. Times (Aug.
22, 2009).
Over 5000 foreign banks, such as UBS, Credit Suisse and Deutsche Bank,
have signed qualified intermediary agreements with the U.S. See Letter from
New York State Bar Association to Sen. Max Baucus et al. (Sept. 10, 2009)
available at LEXIS, TNT library, 2009 TNT 175–67, [hereinafter NYSBA
Sept. 10, 2009 Letter]. The alternative is nonqualified intermediary, or NQI
treatment, which requires the submission of beneficial owner information for
each specific account to avoid U.S. withholding on U.S. source payments of
investment income such as interest and dividends. See generally Treas. Regs.
§§ 1.1441–1 et. seq. (containing QI and NQI documentation and withholding
rules. Importantly, there is no presumption of U.S. status for purposes of
backup withholding with respect to gross security sale proceeds. See Treas.
Regs. 1.6049–5 (d) (3) (ii) (providing that withholding on gross proceeds is
not required for payment to a non-U.S. intermediary unless the payer has
actual knowledge that a nonexempt U.S. person is the beneficial owner of the
payment).
See Rev. Proc. 2000–12, 2000–1 C.B. 387 (outlining model QI agreement).
Prior to the adoption of these nonresident withholding rules, the U.S. had
little assurance that the rules for reducing rates on U.S.-source investment
income payments to non-U.S. investors were properly enforced. See
Susan C. Morse & Stephen E. Shay, Qualified Intermediary Status: A
New U.S. Withholding Role for Foreign Financial Institutions Under Final
U.S. Withholding Regulations, 27 Tax Mgm’t Int’l J. 331, 332–33 (1998)
(noting that the regulations require foreign financial institutions to provide
information about “foreign status, eligibility for treaty benefits, and
qualification for other statutory withholding tax exemptions such as those
applicable to effectively connected income and foreign government or
international organization status” and “plac[e] the burden of investigating
beneficial ownership on QIs rather than on U.S. custodians”). See also
Reuven S. Avi-Yonah, International Tax as International Law 27, 28,
An Analysis of the FBAR High-Penalty Regime
66
67
68
69
70
71
72
29
68–78 (2008) (outlining exceptions to the default 30 percent U.S. withholding
tax on U.S.-source investment income).
A compromise struck in the model QI agreement, in deference to bank
secrecy rules, does not flatly require QIs to disclose the identity of their U.S.
clients. Instead, it describes the option of reconciling the existence of a U.S.
account holder with bank secrecy laws by excluding U.S. securities or other
assets that generate U.S.-source reportable payments from the U.S. client’s
account. See Rev. Proc. 2000–12, 2000 C.B. 387 § 6.02 (“If QI is prohibited
by law, including by contract, from disclosing to a withholding agent … the
account holder’s name … then QI must (i) request … the authority to make
such a disclosure; (ii) request … the authority to sell any assets that generate
… reportable payments or (iii) request that the account holder disclose
himself.”).
UBS apparently recommended to U.S. clients that they hold accounts
through a nominee blocker corporation in a tax haven or that they take
advantage of the fact that the qualified intermediary reporting rules only
applied to assets that generated U.S. source income by moving U.S. account
holders out of assets that produced U.S. source income, perhaps trading in
U.S. treasuries for British gilts. . See, e.g. UBS, Qualified Intermediary
System: U.S. Withholding Tax on Dividends and Interest Income
From U.S. Securities 1 (Oct. 2004) (“A QI has to ensure that US Persons
… either declare themselves to the US tax authorities … or are no longer
permitted to invest in US securities.”).
See Lynnley Browning, U.S. Reports Agreement With UBS in Tax Case, N.Y.
Times at B3 (Aug. 13, 2009) (noting earlier resolution of criminal case against
UBS).
See id. (reporting civil case settlement).
See Lynnley Browning, Swiss Approve Deal for UBS to Reveal U.S. Clients
Suspected of Tax Evasion, N.Y. Times (June 17, 2010). This followed a decision
by the Swiss Federal Administrative Court that the failure to file a W-9 with
UBS for transmission to the U.S. tax authorities did not constitute “tax fraud
and the like” and therefore did not meet a requirement under a 1996 treaty
for an exception to bank secrecy protection. See Daniel Pruzin, Switzerland
for Now to Hand Over Data on Only 250 Secret Accounts with UBS, BNA Tax
Management Weekly Report 144–45 (Feb. 1, 2010).
See Lynnley Browning, IRS to Drop Suit Against UBS Over Tax Havens, N.Y.
Times (Aug. 26, 2010).
The 2009 program followed another initiative, in 2003, which was generally
characterized as having produced “limited success” in large part due to a lack
of enforcement action and publicity. See supra note 57 (describing results of
30
73
74
75
76
77
78
79
80
Morse
2003 voluntary disclosure program launched in part as a response to offshore
credit card initiative); Joint Committee on Taxation, Tax Compliance
and Enforcement Issues With Respect to Offshore Accounts and
Entities 48–49 (Mar. 30, 2009) (hereinafter JCT 2009 Offshore Account
Report) (reporting the view that the lack of parallel enforcement actions
and publicity limited the success of the 2003 program). In 2003 at least one
onerous penalty provision, the 50 percent provision for willful failure to file
an FBAR, was not yet law.
See JCT 2009 Offshore Account Report at 49 (citing IRS report of 1299
applications in 2003).
See Marie Sapirie, New Era of Enforcement Follows UBS Saga, 59 Tax Notes
Int’l 501 (Aug. 16, 2010) (noting 14,700 disclosures under the program).
See Sheppard, supra note 57, at 851.
See, e.g., Lynnley Browning, UBS Client Pleads Guilty to Tax Fraud, New York
Times (Apr. 12, 2010) (reporting on the ninth taxpayer to be caught up in
UBS-related cases); Press Release, U.S. Department of Justice, Tax Division,
Seven UBS Clients Charged With Hiding Over $100 Million in Secret Swiss
Bank Accounts to Defraud the IRS (Apr. 15, 2010), available at http://www.
justice.gov/tax/txdv10_USB_Clients.htm (reporting seven prosecutions, out
of which two defendants had plea bargained). The tendency of such press
releases to cluster in April has not gone unnoticed. See Joshua D. Blank &
Daniel Z. Levin, When is Tax Enforcement Publicized?, 30 Va. Tax Rev. 1
(2010).
See U.S. Department of Justice, Press Release, Justice Department & IRS
Announce Results of UBS Settlement & Unprecedented Response in
Voluntary Tax Disclosure Program (Nov. 17, 2009), available at http://www.
justice.gov/tax/txdv091241.htm.
See, e.g., James Andreoni et al., Tax Compliance, 36 J. Econ. Lit. 818, 846
(1998) (summarizing tax compliance studies associating a high subjective
probability of detection with significantly higher compliance rates).
Jon Hanson & David Yosifon, The Situational Character: A Critical Realist
Perspective on the Human Animal, 93 Geo. L. J. 1, 39–41 (2004) (quoting
Susan T. Fiske & Shelley E. Taylor, Social Cognition 384 (2d ed.
1991)). See generally Edward J. McCaffery & Joel Slemrod, Toward an Agenda
for Behavioral Public Finance, in Behavioral Public Finance (Edward J.
McCaffery & Joel Slemrod eds., 2006).
See, e.g., James Alm, Betty R. Jackson & Michael McKee, Getting the Word
Out: Enforcement Information Dissemination and Compliance Behavior 93 J.
Pub. Econ. 392, 401 (2009) (reporting results of laboratory study showing
that subject-to-subject messaging about audit outcomes significantly affects
An Analysis of the FBAR High-Penalty Regime
81
82
83
84
85
31
compliance decisions and also showing different responses to different
combinations of government information); Jeffrey Dubin, Criminal
Investigation Enforcement Activities and Taxpayer Noncompliance, 35 Pub.
Fin. Rev. 500, 516, 518 (2007) (concluding from a longitudinal study of state
segmented data that audits and criminal investigation activities significantly
influence compliance behavior). See also James Alm & Mohammad Yunus,
Spatiality and Persistence in U.S. Individual Income Tax Compliance, 57 Nat’l
Tax J. 101, 121 (2009) (finding correlation between geographic residence and
evasion behavior).
See Alan H. Plumley, I.R. Publication 1916: The Determinants of
Individual Income Tax Compliance: Estimating the Impacts of Tax
Policy, Enforcement, and IRS Responsiveness 35 (1996) (estimating the
indirect audit effect at 11.6 times the direct audit effect); Dubin, supra note
86, at 519 (reporting result of 15.1:1 under simulation doubling audit rates).
See Susan T. Fiske & Shelley E. Taylor, Social Cognition 270–71
(1984) (noting “retrieval biases,” “strength of association biases” and ease
of imagining events); Amos Tversky & Daniel Kahneman, Availability: A
Heuristic for Judging Frequency and Probability, 163, 163 in Judgment Under
Uncertainty: Heuristics and Biases (Daniel Kahneman, Paul Slovic
& Amos Tversky, eds., 1982). (“Life-long experience has taught us that
instances of large classes are recalled better and faster than instances of less
frequent classes, that likely occurrences are easier to imagine than unlikely
ones, and that associative connections are strengthened when two events
frequently co-occur.”) See also Ronald Chen & Jon Hanson, Categorically
Based: The Influence of Knowledge Structures on Law and Legal Theory, 77
S. Cal. L. Rev. 1106, 1179 (2004) (“[C]ues that are prominent or catch our
attention are more likely to activate associated categories and schemas.”).
See Morse, supra note 11, at 510 (“[An audit] publicity campaign featuring
more typical taxpayers would have more salience.”).
See Joanna Chung & Haig Simoniam, Former UBS Employee Charged
With Helping Billionaire Evade Tax, Fin. Times, May 14, 2008 (noting the
December 2008 guilty plea of real estate magnate Igor Olenicoff, who
agreed to pay $52 million in back taxes related to “income earned on about
$200 million of assets kept offshore”); see also Lynnley Browning, Suicide
Victim May Have Hidden Millions Abroad, N.Y. Times, Sept. 15, 2009, at B1
(reporting that the government had begun to build a criminal tax evasion
case involving as much as $100 million in back taxes against Finn Caspersen
before his death).
See, e.g., Lynnley Browning, Florida Man, a UBS Client, Pleads Guilty to Tax
Fraud, N.Y. Times, June 26, 2009 (reporting Rubinstein guilty plea); Lynnley
32
Morse
Browning, First Client from U.S. Is Arrested in UBS Case, Apr. 3, 2009
(reporting Rubinstein arrest).
86 See Lynnley Browning, UBS Client Pleads Guilty in Tax Case, N.Y. Times,
Apr. 15, 2009 (reporting Moran guilty plea).
87
88
89
90
91
92
93
See Lynnley Browning, Inquiry Widens as UBS Client Pleads Guilty, N.Y.
Times, July 29, 2009 (reporting Chernick guilty plea).
See Samantha Henry, UBS Client Pleads Guilty to Hiding Assets, Associated
Press, Sept. 26, 2009 (reporting Homann guilty plea).
See David Voreacos & Carlyn Kolker, UBS Client to Admit Failure to Report
Swiss Account to IRS, Bloomberg, Aug. 15, 2009 (reporting anticipated
McCarthy guilty plea).
See Kim Dixon, Ex-Boeing Manager Pleads Guilty in UBS Tax Case, Reuters,
Oct. 5, 2009.
Lynnley Browning, Settlement Anticipated in UBS Case, N.Y. Times, June 22,
2009 (reporting the description of a government official).
See Laura Saunders, IRS Extends Deadline to Declare Foreign Accounts,
Wall St. J., Sept. 22, 2009, (reporting “no discernible pattern as to which
customers were selected” for required disclosure under UBS settlement and
repeating on practitioner’s comment that “ ‘[s]everal of our clients with ‘plain
vanilla’ accounts well under $1 million have gotten these letters.’ ”). The
U.S. John Doe summons request did not discriminate based on the size of
the account. See Memorandum in Support of Ex Parte Petition for Leave
to Serve John Doe Summons at 5, In re Tax Liabilities of John Does (S.D.
Fla. No. 08–21864) (June 30, 2008) (describing John Doe class as any U.S.
taxpayer with “signature or other authority … with respect to any financial
accounts,” except for taxpayers who had supplied UBS with Forms W-9 and
been subject to Form 1099 reporting). However, since the description of
account selection criteria under the summons settlement is not yet available,
it is difficult to tell whether targeting a range of accounts was an intentional
strategy. See Sheppard, supra note 57, at 850 (speculating that the U.S.
targeted large accounts and accounts with particularly creative planning).
Publicizing taxpayers who have been caught is likely more important
that publicizing the audit rate or the compliance rate, both of which draw
mixed results in terms of their ability to promote additional compliance.
Taxpayers may interpret the audit rate as communicating that audit activity
exists or communicating that an audit is too unlikely to worry about. Cf.
Alm, Jackson & McKee, supra note 80, at 401 (noting conflicting results for
“official” publication of audit information in laboratory study). The IRS does
publish audit rates, though it keeps the factors that affect its audit selection
mechanism secret.
An Analysis of the FBAR High-Penalty Regime
94
95
96
97
98
33
The typically cited problem with publicizing the compliance rate, as opposed
to quietly disclosing it, is that taxpayers can interpret the figure as meaning “a
clever minority cheats” instead of “most people pay their taxes.” In one reallife experiment, Minnesota taxpayers received a letter from the Minnesota
Department of Revenue stating that nearly all taxpayers—93 percent—were
compliant. Increased compliance, measured by reference to actual tax
returns filed, was not statistically significant for those who received the letter.
The possibility that the audience will self-identify with or aspire to be part of
the “clever minority” makes this a risky strategy. See Marsha Blumenthal et
al., Do Normative Appeals Affect Tax Compliance? Evidence from a Controlled
Experiment in Minnesota, 54 Nat’l Tax J. 125, 135 (2001) (stating that a
statement of high compliance “may be interpreted to mean that the revenue
department is unable to detect cheating”).
See BNA, Tax Mg’t Weekly Report at 100 (Jan. 25, 2010) (noting 150
ongoing offshore account criminal investigations and that “hundreds of
taxpayers are still coming in under IRS’s basic procedures for voluntary
disclosure”). Plea bargain publicity has continued to emerge, and continues
to feature the average wealthy. See, e.g., Lynnley Browning, UBS Client Pleads
Guilty to Tax Fraud, N.Y. Times (Apr. 12, 2010) (reporting guilty plea of
Harry Abrahamsen of Oradell, New Jersey, whose UBS account was allegedly
financed by claiming $1.3 million in inflated expenses—which would have
produced a tax benefit of perhaps approximately $500,000).
See Sheppard, supra note 57, at 850 (suggesting that the IRS should pursue
and publicize 50 UBS cases and 20 from other banks).
See Lynnley Browning, Seeking Bank Secrecy in Asia, N.Y. Times (Sept. 22,
2010) (reporting hundreds of billions of dollars in account value reductions
in Europe and gains in Hong Kong and Singapore).
See Lynnley Browning, U.S. Widens Tax Inquiry Into HSBC, N.Y. Times (July
9, 2010) (reporting criminal investigation of London-based HSBC and two of
its clients).
I.R.C. § 6103. The statute defines “return information” very broadly and it
includes “any information developed or obtained by the IRS during the
course of an audit or investigation of the taxpayer, as well as the mere fact
that the taxpayer’s return has been or is being audited or investigated.”
Stephen W. Mazza, Taxpayer Privacy and Tax Compliance, 51 Kan. L. Rev.
1065, 1091 (2003). A series of exceptions permits disclosure of return
information in certain specific circumstances, which include several thirdparty disclosure permissions necessary to effective administration. For
example, the IRS may disclose information in connection with judicial
proceedings, see, e.g., I.R.C. § 6103 (h) (4), and under certain circumstances to
34
Morse
obtain relevant information, see I.R.C. § 6103 (k) (6), or put an interested party
on notice, see, e.g., 6103 (e).
99 See, e.g., I.R.C. § 6103 (b) (1) (permitting disclosures to the general public
when it publicizes “data in a form which cannot be associated with, or
otherwise identify, directly or indirectly, a particular taxpayer”).
100 See, e.g., Mazza, supra note 98, at 1121 (“The IRS’s current efforts to
communicate strong and meaningful deterrence messages are hampered
by the lack of an exception in section 6103 permitting disclosure of return
information to criminal tax proceedings.”).
101 The circuit courts have divided into three camps. The Ninth and Sixth
Circuits have adopted a “public records” exception that permits the IRS
to publicize taxpayer information that has been disclosed in litigation,
including in an indictment or other filing that precedes a final determination.
See Rowley v. United States, 76 F.3d 796, 801 (6th Cir. 1996) (holding valid
IRS disclosure of taxpayers’ names and tax deficiency in an advertisement for
the sale of property under tax lien despite the later release of the lien due to
improper notice); Schrambling v. United States, 937 F.2d 1485, 1488–89 (9th
Cir. 1991) (concluding that the filing of a tax lien destroyed confidentiality);
Lampert v. United States, 854 F.2d 335, 338 (9th Cir. 1988) (focusing on
press releases relating to charges and final resolutions and declining to use a
“strict, technical reading of the statute” because such a reading would “defeat
the purposes of the statute”), cert. denied, 490 U.S. 1034 (1989). The Fourth
Circuit adopts the technical statutory reading rejected by the Ninth Circuit
and holds that no disclosure of return information is permitted regardless
of the public disclosure of such information elsewhere. See Mallas v. United
States, 993 F.2d 1111, 1120–21 (4th Cir. 1993) (finding a violation of Section
6103 under a strict statutory reading and on facts including the disclosure
of more facts than appeared in the court opinion, which was subsequently
unanimously reversed by an en banc Fourth Circuit decision). The Fifth,
Seventh and Tenth Circuits have adopted forms of an “immediate source”
exception, which permits disclosure if the IRS in fact drew the relevant
information from court or other public proceedings and not from inside
agency information. See Thomas v. United States, 890 F.2d 18, 21 (7th Cir.
1989) (noting that Section 6103 “is not a prohibition of any kind against the
disclosure of opinions of the Tax Court”); see also Rice v. United States, 166
F.3d 1088 (10th Cir. 1999) (finding no Section 6103 violation where IRS press
official had obtained press release information from public findings and trial
and sentencing proceedings); Johnson v. Sawyer, 120 F.3d 1307, 1325–26 (5th
Cir. 1997) (finding a violation of Section 6103 where information disclosed
by IRS employee “came either from Johnson’s return file or from information
‘in [the IRS employee’s] head’”). See generally Mazza, supra note 98, at
An Analysis of the FBAR High-Penalty Regime
35
1105–14, 1121–22 (analyzing case law and related cases in other contexts
considering when public disclosure diminishes privacy rights and describing
and evaluating Joint Committee and Treasury recommendations “which
essentially adopt the Ninth Circuit’s public records exception”).
102
See IRS, Offshore Tax Avoidance and IRS Compliance Efforts, http://www.irs.
gov/newsroom/article/0,,id=110092,00.html (last visited Oct. 12, 2009).
103 See Internal Revenue Manual § 9.5.11.9 (1)–(4). An earlier variant of the
voluntary disclosure program included a more definite undertakings to not
recommend criminal prosecution. See Michael I. Saltzman & Leslie
Book, IRS Practice and Procedure ¶ 12.07 [3] (2009) (describing policy
between 1945 and 1952).
104 See 31 U.S.C. § 5321 (a) (5) (providing for a penalty of the greater of $100,000
or 50 percent of the balance in the account at the time of the violation).
Prior to 2004, the maximum penalty for a willful violation was the lesser
of $100,000 or the account balance at the time of violation. See Bittker &
Lokken, supra note 45.
105 See id.; IRS, Voluntary Disclosures: Questions and Answers, supra note 51,
at Q & A 22 (giving penalty example). Delinquency penalties for failure
to file and failure to pay are typically calculated as a percentage of the tax
due per month of failure to file or pay, up to a maximum of 25 percent
each. See I.R.C. § 6651. The accuracy penalty equals 20 percent of certain
underpayments including an underpayment attributable to negligence,
disregard of rules or regulations, a substantial underpayment or other
misconduct. See I.R.C. § 6662. See generally Leandra Lederman &
Stephen W. Mazza, Tax Controversies: Practice and Procedure
§ 10.02 (2d ed. 2002).
106 Cf. Fred Feingold, Further Guidance Needed on Who Must Report Foreign
Accounts, 123 Tax Notes 1023, May 25, 2009 (arguing that the FBAR
proposal goes too far, as ignorance of reporting requirements, not willful
intent to evade tax, may cause failure to comply with FBAR filing).
107 See IRS, Voluntary Disclosure: Questions and Answers, supra note 51, at
Q & A 32.
108 In guidance, the IRS stated that a 5 percent penalty might apply to accounts
that the taxpayer “did not open or cause … to be opened, [where] there
has been no activity … during the period the account … was controlled by
the taxpayer, and … all applicable U.S. taxes have been paid on the funds
[deposited] in the accounts.” Memorandum from Linda E. Stiff, Deputy
IRS Commissioner for Services and Enforcement, to Commissioner,
Large and Mid-Size Business Division and Commissioner, Small Business/
Self-Employed Division (March 23, 2009), at 2. An inherited account, for
36
Morse
example, might fit these criteria. However, practitioners report that as a
practical matter taxpayers cannot persuade the government to apply only a
5 percent penalty. See, e.g., Remarks of Frank Agostino, Kathryn Keneally &
Bryan Skarlatos, The Prosecution and Defense of Offshore Bank Accounts,
ABA Tax Section Teleconference and Live Audio Webcast (Mar. 3, 2010).
109 Statement from IRS Commissioner Doug Shulman on Offshore Income, Mar.
26, 2009, available at www.irs.gov/newsroom/article/0,,id=206014,00.html.
110 See id.; see also IRS Extends Deadline for Disclosing Hidden Offshore
Accounts, IR 2009–84, Sept. 21, 2009.
111 As this article went to press, the government announced a second FBARtargeted voluntary disclosure program that used a 25-percent-of-accountvalue fine for most accounts and added a 12.5 percent penalty for smaller
accounts whose value did not exceed $75,000 in any covered year. See Second
Special Voluntary Disclosure Initiative Opens; Those Hiding Assets Offshore
Face Aug. 31 Deadline, IR 2001-14 (Feb. 8, 2011).
112 See Cheek v. United States, 498 U.S. 192, 201–02 (1991) (considering case
involving alleged willful failure to file a Federal income tax return).
113 See, e.g., Fred Feingold, Further Guidance Needed on Who Must Report
Foreign Accounts, 123 Tax Notes 1023 (May 25, 2009) (arguing that many
FBAR nonfilers fail to file due to ignorance of the requirement).
114 See United States v. Sturman, 951 F.2d 1466, 1476–77 (6th Cir. 1991) (holding
that actions taken to conceal assets from the government, including the
use of different corporations to transfer funds, together with admitted
“knowledge of and failure to answer a question concerning signature
authority at foreign banks on Schedule B of his income tax return”
“provid[ed] a sufficient basis to establish willfulness on the part of the
defendant”).
115 See supra text accompanying notes 13–14 (outlining reward elements built
into tax administration).
116 See I.R.C. § 6103 (b) (1) (defining “return”); 31 U.S.C. § 5314 (authorizing
statute for FBAR regulation); 31 C.F.R. § 103.24 (a) (requiring FBAR filing).
117 See I.R.C. § 6103 (b) (1) (defining “return” as “including supporting schedules,
attachments, or lists which are supplemental to, or part of, the return so
filed”).
118 See IRS, Voluntary Disclosure: Questions and Answers, supra note 51, Q & A
6, 26 (indicating that all missing “returns” may be filed with the voluntary
disclosure letter and not specifying that FBARs should be sent separately).
119 See Bittker & Lokken, supra note 45 (“[A]lthough this reporting regime
is administered by the IRS, it is not the only Federal agency having access
An Analysis of the FBAR High-Penalty Regime
37
to the reported information, and government use of the information is not
restricted to tax enforcement.”); Levy, supra note 37.
120 See 31 U.S.C. § 310 (c) (2); see also Lee Sheppard, FBAR Filing for Hedge
Funds, 125 Tax Notes 496, 500 (Aug. 17, 2009) (calling practitioner’s § 6103
concerns a “red herring”).
121 Cf. Baucus Seeks to Deter, Detect, Discourage Offshore Tax Evasion, Tax
Notes (Mar. 12, 2009), available at 2009 TNT 46-19 (reporting on legislation
discussion draft that would have required FBARs to be filed along with
tax returns and to require tax preparers to ask due diligence questions
specifically relating to FBAR compliance).
122 The banking law, at 31 U.S.C. § 310(c), references the Privacy Act, 5 U.S.C.
§ 552a, which includes an exception for any court order, see 5 U.S.C. § 552a(b)
(11) and has been classified by Congress as a statute that does not provide
general protection against FOIA disclosure for the information that it covers,
see 5 U.S.C. § 552a(b)(2). The law relating to the disclosure of tax return
information, for example to third parties under FOIA and in civil litigation,
has developed differently. See, e.g., Lederman & Mazza, supra note 105,
at § 3.04 (noting issues related to the “tension between FOIA and Section
6103”).
123 See Department of the Treasury, Form TD F 90-22.1.
124 In one case, a district court awarded a taxpayer $75,000 in compensatory
damages and $250,000 in punitive damages plus costs as a result of IRS
employees’ discussion of a taxpayer’s case on a radio show with the taxpayer
and submission of a letter to the editor concerning the case in response to
the taxpayer’s prior letter to the editor. See Ward v. United States, 973 F.
Supp. 996, 1000–02 (D. Colo. 1997) (imposing damages pursuant to I.R.C.
§ 7431 (c) ).
125 See I.R.C. § 6103 (c).
126 I am aware of at least one such settlement. See IRS News Release IR-2004-151
(Dec. 16, 2004), available at www.irs.gov/newsroom/article/0,,id=132350,00.
html) (reporting tax shelter settlement involving corporate taxpayer
Hercules, which waived privacy rights in connection with the press release).
127 I.R.C. § 6103 (b) (2).
128 See 122 Cong. Rec. 24012 (1976).
129 Church of Scientology of California v. IRS, 484 U.S. 9 (1987). See also Long
v. IRS, 891 F.2d 222, 223–24 (9th Cir. 1989) (holding on remand that audit
“check sheets” were not in a form that constitutes a reformulated data base of
the sort that is eligible for disclosure under the Haskell amendment).
38
130
Morse
See Willamette Industries, Inc. v. United States, 689 F.2d 865 (9th Cir. 1982);
see also Gary, Plant, Mooty, Mooty & Bennett, P.A.v IRS, 1990 U.S. Dist.
LEXIS 18799 (following Willamette and holding that Section 6103 did not
block a FOIA claim for the IRS to produce the “Brown Report,” relating to
industry-level data about computer company audits).
131 Shell Petroleum, Inc. v. United States, 46 Fed. Cl. 719, 725 (2000).
132 Listing types of taxpayers in a press release was apparently not effective
when used in connection with the 2003 voluntary disclosure program. See
IR 2003–95 (July 30, 2003) (“People from all walks of life applied for the
[2003 voluntary disclosure] program, including lawyers, dentists, business
executives, estate heirs and numerous other occupations.”).
133 Pennsylvania has recently adopted an interesting, salient, Orwellian
approach to publicizing a tax amnesty program. See Pennsylvania Dep’t
of Revenue, http://www.pataxpayup.com/portal/server.pt/community/
resources_advertising/18999 (last visited May 21, 2010) (including links
to communications including TV commercial titled “We Know Who You
Are.”).
134 See Treas. Regs. § 1.451–1 (a) (“If a taxpayer ascertains that an item should
have been included in gross income in a prior taxable year, he should,
if within the period of limitation, file an amended return and pay any
additional tax due.”); Lederman & Mazza, supra note 105, at § 2.02 [D]
(describing generally amended return practice).
135 See supra note 103 and accompanying text (describing voluntary disclosure
guidance).
136 See, e.g., Letter from Stuart E. Abrams et al. to The Honorable Douglas
H. Shulman, Commissioner of Internal Revenue & John DiCicco, Esq.,
Acting Assistant Attorney General, Department of Justice, Tax Division 2
(Mar. 30, 2010) (asserting that to maintain consistency with taxpayer and
practitioner expectations, the government should ensure that taxpayers who
attempt voluntary disclosure in “good faith” are not prosecuted, even if their
disclosures are technically late).
137 See Saltzman & Book, supra note 103, at ¶¶ 12.07[3][d] & [e] (distinguishing
quiet disclosure from voluntary disclosure and noting disadvantages such
as the waiver of Fifth Amendment protection and the possibility of an
additional violation if the amended returns are incorrect).
138 Remarks of Frank Agostino at The Prosecution and Defense of Offshore
Bank Accounts, ABA Tax Section Teleconference & Live Audio Webcast
(Mar. 3, 2010) (describing long-standing “quiet disclosure” approach based
on private practice experience in Hackensack, New Jersey).
An Analysis of the FBAR High-Penalty Regime
139
39
Cf. Lawrence R. Jones, Jr., Dealing With the IRS Collection Division
§ 1412, at 235–26 (1995) (stating that a taxpayer has a very limited chance of
criminal prosecution if failure to file is corrected by filing tax returns and
recommending the resolution of “all questionable items on the delinquent
tax return … in favor of the IRS” to minimize the risk of fraud charges).
140 Of course, the taxpayer’s willingness to choose the quiet disclosure option
instead of the voluntary disclosure option with its more explicit commitment
to avoid a criminal prosecution recommendation depends in part on the
taxpayer’s risk aversion.
141 See IRS, Voluntary Disclosure Questions and Answers, supra note 51, at
Q & A 10 (“Those taxpayers making ‘quiet’ disclosures should be aware of
the risk of being examined and potentially criminally prosecuted for all
applicable years.”).
142 See Raskolknikov, supra note 3, at 724 (noting that very high compliance
regime penalties will induce gamers, particularly aggressive gamers, to try to
hide behind the compliance regime).
143 See Treas. Inspector Gen’l for Tax Admin., New Legislation Could
Affect Filers of the Report of Foreign Bank and Financial
Accounts, but Potential Issues Are Being Addressed (Sept. 29, 2010).
144 See Internal Revenue Service, Fiscal Year 2009 Enforcement Results
2, 3, available at http://www.irs.gov/newsroom/article/0,,id=217442,00.html.
145 See David D. Stewart, New IRS Group to Examine Wealthier Individuals Using
Offshore Arrangements for Evasion, Tax Notes (Sept. 2, 2009) available at
LEXIS, TNT library, 2009 TNT–168–1.
146 See, e.g., I.R.C. § 6038D(c).
147 I.R.C. § 6011 (e) generally specifies the Secretary’s ability to require electronic
filing. Robert Foley of State Street Bank has suggested that taxpayers at least
be able to elect electronic FBAR filing, citing in part the ability of the IRS to
more effectively use electronically submitted data. See email from Robert J.
Foley to Notice Comments (Aug. 27, 2009), available at LEXIS: TNT library,
2009 TNT 173–19 or Doc. 2009–20081).
148 The applicable FBAR regulation delegates to the Secretary of the Treasury
the authority to prescribe the information that must be listed on the form. 31
C.F.R. § 103.24 (a).
149 Cf. Blank, supra note 26, at 1632 (describing the problem of overdisclosure
under tax shelter disclosure rules).
Facilitated Self-Assistance
Enhances Taxpayers’ Taxpayer
Assistance Center (TAC)
Experiences
Kirsten Davis, Melissa Hayes, and Erica Jenkins, Internal Revenue Service
I
n the summer of 2005, Congress mandated the Internal Revenue Service (IRS)
to research, taxpayer needs and IRS service delivery at present and in the future. The goal of the mandate was to ensure that service-related decisions
are informed by research and guided by stakeholder engagement. To fulfill the
mandate, the IRS formed a team to create Taxpayer Assistance Blueprint, or TAB,
products. The TAB Phase 1 Report, released in April 2006, presented preliminary
IRS research relative to taxpayer needs, preferences, and behaviors.1 The Phase 2
Report, released in April 2007, built upon the baselines and improvement themes
identified in Phase 1 and produced a 5-year strategic plan for taxpayer service.2
The 5-year plan was to be the future of IRS service delivery as envisioned collaboratively by the IRS, the IRS Oversight Board, and the National Taxpayer Advocate.
TAB Phase 2 research from the 2006 IRS Oversight Board Taxpayer Customer
Service and Channel Preference Survey indicated that about two in five (43 percent) Taxpayer Assistance Center (TAC, IRS office) users said they would consider
contacting the IRS on the web instead of in person to get needed information
or assistance.3 Another key finding from TAB Phase 2 was that there is greater
taxpayer value in getting forms and publications online rather than through any
other channel.4 These findings suggested the IRS pursue opportunities to enhance
services provided in the TACs, and the Facilitated Self-Assistance Project (FSRP)
began.
The FSRP was a collaborative effort among Wage & Investment Research and
Analysis (WIRA), Field Assistance (FA), and Modernization and Information
Technology Services (MITS) to test the alternative TAC business model. With the
exception of forms and publications racks, the TAC business model included only
face-to-face service in which taxpayers must wait for assistance, irrespective of
service task. Face-to-face service in TACs is the IRS’s second most expensive delivery channel.5
The goal of the FSRP was to test a service option in which taxpayers, with in-person
assistance readily available, conducted certain tax-related tasks using IRS.gov. After
40
Davis, Hayes, and Jenkins
using the new service option, study participants were asked to share their opinions
about the service experience.
The FSRP was originally administered at 15 TACs from March 10, 2008, to April
15, 2008. Due to data collection problems, a second phase of data collection was
necessary.6 The second phase of the FSRP was conducted in 50 TACs from January
2 to April 30, 2009. A formal research report was released in the autumn of 2009,
and our paper is based on that report.7
Research Problem
The objective of the FSRP was to determine if self-assistance computer workstations, with IRS assistors available, are an effective and efficient means of providing
customer service in TACs. The primary research question of this study was:
Does changing the TAC business model to more differentiated service
delivery increase taxpayer and/or government value?
The market segment for the study included taxpayers seeking tax-related services available on IRS.gov in TACs that were selected to offer and test facilitated
self-assistance (FSA).
This paper begins with a profile of FSA users, including demographic and
service-task characteristics and is followed by an examination of taxpayers’ opinions on the value of FSA. Taxpayers’ verbatim comments are presented to support
quantitative findings and enhance the reader’s understanding of user’s opinions.
The paper concludes with data limitations, challenges, and recommendations regarding future implementation of FSA. The terms “users” and “participants” are
used interchangeably. Findings presented in this paper are from FSRP Phase 2
unless otherwise noted.
Research Methodology
FSRP Phase 2 was conducted at 50 TAC sites from January 2, 2009, to April 30,
2009. FA selected participating TACs by judgmental sample—a non-probability
sampling method that uses basic criteria specified as relevant to addressing the
research objective. FSA was to be offered to all taxpayers who sought service for
project eligible tasks and who were judged eligible to participate by an IRS employee. Since TACs and taxpayers included in the project were not randomly selected; results from this study may not be statistically representative of all TAC
visitors.
The FSRP included several data collection instruments: participant eligibility,
Intake Survey, and Exit Survey. IRS employees stationed at TAC reception desks
evaluated eligibility and performed study recruitment, as well as administered the
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
41
Intake Survey to willing participants. Exit Surveys were completed online by project participants after completion of their FSA service task.
Eligibility for FSRP participation was determined by three factors: 1) English
language proficiency, 2) physical ability to use computer workstations without
adaptive technology, and 3) service tasks that were within the scope of the project.
FSRP service tasks were selected based on electronic services that were available
on IRS.gov during the project period. In-scope service tasks for the project were:
1. Free File or Free File Fillable Forms
2. Where’s My Refund?
3. Frequently Asked Tax Questions and Answers
4. Forms and Publications
5. Earned Income Tax Credit (EITC) Assistant
6. IRS Withholding Calculator
7. Online Payment Agreement
8. E-File Locator Service
9. Economic Stimulus Payment/Rebate
10. Employer Identification Number (EIN) Application
11. Electronic Federal Tax Payment System (EFTPS)
If the taxpayer was judged eligible and was prepared to receive service, the taxpayer was invited to participate in the study and obtain service using a computer
with an IRS employee available to help them. Project protocol required that an
IRS employee, called a facilitator, be available to assist FSA users, as needed, in
performing their tasks. The facilitators were directed to assist project participants
but not to enter any data for them.
Research Findings
TAC Visitors’ Willingness to Try FSA
Participant eligibility data indicated that, of taxpayers who were assessed as capable, had FSRP eligible tasks, and were prepared to conduct their business, 49
percent chose to try the new service option.
Figure 1 displays the number of FSA users who participated in the study’s surveys, including those who responded to more than one survey. Of 7,858 FSA users,
6,490 responded to one or both of the project’s surveys (Intake Survey and/or Exit
Survey).
42
Davis, Hayes, and Jenkins
FIGURE 1. FSRP Survey Participation
1,468 Participants
Took Intake Survey
Only
1,042 Participants
Took Both Intake
& Exit Survey
3, 980 Participants
Took Exit Survey
Only
Source: FSRP Intake and Exit Survey data, January 2, 2009, to April 30, 2009.
Data for Figure 1 based on:
Of 7,858 FSA users, 6,490 responded to one or both of the project’s surveys (Intake Survey and/or
Exit Survey). Intake Survey data were obtained from 2,510 FSA participants, and 5,022 FSA users
participated in the Exit Survey; 1,042 FSA users participated in both the Intake and Exit Surveys.
Profile of FSA Users
The Intake Survey captured demographics and service task for taxpayers who participated in the FSRP. Demographic information collected included respondents’
age, total household income, and gender. Intake Survey data indicated the following characteristics:
yy Most FSA users were 54 or under (78 percent), and nearly half were
25 to 44 (48 percent).
yy Approximately half (51 percent) of respondents had a total
household income of $35,000 or less.
yy Males and females made up 54 and 44 percent of the sample,
respectively.8
Figure 2 displays the most common service tasks among FSA users were seeking tax forms or publications (47 percent) and Free File (34 percent).
43
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
FIGURE 2. Distribution of Main Service Tasks
Other
9%
Forms and
Publications
47%
EIN
Application
10%
Free File
34%
Source: FSRP Intake Survey data, January 2, 2009, to April 30, 2009.
Data for Figure 2 based on:
Primary Service Task
Forms or Publications
Free File
EIN Application
Where’s My Refund?
Frequently Asked Tax Questions and Answers
Online Payment Agreement (OPA)
Earned Income Tax Credit (EITC) Assistant
E-File Locator Service
Economic Stimulus Payment/Rebate
Electronic Federal Tax Payment System (EFTPS)
IRS Withholding Calculator
Count
Percent Respondents
1,176
844
255
100
62
22
18
10
9
9
5
47%
34%
10%
4%
2%
1%
1%
0%
0%
0%
0%
Service task was also examined in conjunction with FSA user demographic
characteristics. Younger respondents appeared most likely to use FSA for Free File:
60 percent of users 24 or under used FSA for Free File (Figure 3). In contrast, most
seniors 65 or over (88 percent), used FSA to obtain forms or publications.
44
Davis, Hayes, and Jenkins
FIGURE 3. Distribution of FSA Users’ Service Task by Age Group
Age
Groups
Other
Tasks
24 or under
14%
25 – 34
15%
35 – 44
7%
45 – 54
EIN
Application
Forms or
Publications
Free
File
9%
16%
60%
12%
35%
39%
14%
46%
32%
9%
9%
61%
21%
55 – 64
7%
7%
75%
11%
65 or over
5%
3%
88%
4%
100%
80%
39%
32%
21%
Free File
11%
4%
60%
60%
35%
40%
46%
61%
75%
88%
0%
9%
14%
EIN Application
Other tasks
16%
20%
Forms or
Publications
12%
15%
24 or under 25-34
3%
14%
9%
7%
9%
7%
7%
35-44
45-54
55-64
5%
65 or over
Age Groups
Sources: FSRP Intake and Exit Survey data, January 2, 2009, to April 30, 2009 (n = 994).
Similarly, the use of FSA for forms and publications increased with income
level, while use of FSA for Free File decreased with income level.
FSA Participants’ Intentions to Use IRS.gov in the
Future
After completing tasks using FSA, nearly three quarters (73 percent) of respondents answered the Exit Survey question, “In the future, would you use the IRS
website (www.IRS.gov) again for any of the following services?” As shown in
Figure 4, over half of the respondents (58 percent) indicated that they would prepare returns online, followed by getting forms or publications (50 percent). Use of
the Online Payment Agreement was the FSA service task least selected (7 percent).
45
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
FIGURE 4. Intentions to Perform FSA Tasks Using IRS.gov in the Future
EFTPS
10%
EIN Application
8%
FSRP Task
Stimulus Payment
12%
E-File Locator Service
9%
Online Payment Agreement
7%
Witholding Calculator
9%
EITC
11%
Forms or Publications
50%
FAQs
31%
Where's My Refund?
25%
58%
Free File
0%
20%
40%
60%
80%
100%
Source: FSRP Exit Survey data, January 2, 2009, to April 30, 2009 (n = 3,648).
Note: FSA users could make multiple responses; total will not sum to 100 percent
Additional Exit Survey analyses were conducted to examine whether FSA users
would use IRS.gov again in the future for the same task. Among FSA users who
used Free File and said they would use IRS.gov in the future, 95 percent said they
would use the IRS Web site for Free File again. For those seeking forms or publications who said they would use IRS.gov in the future, 83 percent said they would
use the IRS Web site to obtain forms or publications again.
Wait Time for Service
In order to assess taxpayer value of FSA, wait time for the new service option
versus traditional TAC service (i.e., face-to-face service) was examined. For FSA
users, wait time was measured from the time the IRS employee designated the taxpayer as willing to participate to the time the taxpayer logged into an FSA computer workstation. The average workstation wait time for the two most common FSRP
tasks—getting forms or publications or using Free File—was 3.7 minutes and 7.1
minutes, respectively. Wait time for traditional TAC service is measured from the
time a customer receives a Q-Matic ticket to the time she or he is called to see an
assistor. These data are captured in Business Objects in categories, (i.e., 0 minutes,
1 to 30 minutes, 31 to 45 minutes, etc.); therefore, FSRP data were grouped similarly for the two most common and comparable FSRP tasks.
Figure 5 indicates that wait times between FSA and traditional face-to-face
TAC service were similar for obtaining forms or publications. For both FSA and
traditional service, nearly all taxpayers (99 and 95 percent, respectively) waited 30
minutes or less to obtain forms or publications. However, wait time for Free File
was shorter for taxpayers using FSA than it was for those using traditional TAC
46
Davis, Hayes, and Jenkins
service. Nearly all taxpayers (95 percent) who used an FSRP workstation for Free
File waited 30 minutes or less, while 79 percent of those who used face-to-face
service had similar wait times for paper or electronic return preparation.
FIGURE 5. Wait Time for FSA vs. Traditional Service
Wait Time Comparison: FSA vs Traditional
0–30 min
31+ min
FSA, Forms or Publications (n = 2,085)
99%
1%
Traditional, Forms or Publications (n = 312,301)
95%
5%
FSA, Free File (n = 1,910)
95%
5%
Traditional, Return Preparation (n = 103,438)
79%
21%
100%
1%
5%
5%
21%
80%
60%
31+ min
0–30 min
40%
99%
95%
95%
79%
FSA,
Forms or
Publications
(n = 2,085)
Traditional,
Forms or
Publications
(n = 312,301)
FSA,
Free File
(n = 1,910)
Traditional,
Return
Preparation
(n = 103,438)
20%
0%
Service Channel and Task
Sources: FSRP Screening and Exit Survey data, January 2, 2009, to April 30, 2009; Business Objects,
January 2, 2009, to April 30, 2009.
Taxpayers who used FSA had shorter wait times than taxpayers who used traditional service in TACs. Findings indicate that 75 percent of FSA users who sought
assistance for Free File had a wait time of less than 5 minutes.
FSA users’ comments regarding wait time:
I came into the office today to have an IRS staf (sic) person prepare my
tax return, but the self e-file service was offered and I used it. I found
it more convenient than waiting to see a representative and relatively
easy for anyone who is computer literate.
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
47
It was my pleasure to be able to receive the type of services that was
render (sic) to me without a long period of waiting
A wonderful experience ~ the assistance I received from the IRS
Employee made the process even easier…I plan to use the system
again for my 2009 filing. GREAT Job IRS ~ keep up the good work!
No more Waiting!
According to the IRS Customer Satisfaction Survey Field Assistance National
Report (January 2009 to April 2009), promptness of service remains the top improvement priority for FA customers. Similarly, wait time was a key concern from
FSA users’ perspectives. FSA was acknowledged as a positive improvement in this
area. Reduced wait time decreases taxpayer burden and improves taxpayers’ overall experience using TACs.
Service Time
Unlike wait time, service time for the two primary FSRP tasks was longer than
traditional face-to-face service. Service time was defined as the time FSA users
were logged into a workstation. Service time for traditional TAC service is defined
as the time between when the customer is called to see an assistor and the time the
customer’s ticket is closed by that assistor.9
On average, it took FSA users about 2 minutes longer to obtain forms or publications, compared to taxpayers using traditional TAC service for the same task
(8.7 minutes vs. 6.5 minutes). Regarding Free File, FSA users who completed an
electronic return took an average of 53.3 minutes to do so. Taxpayers who sought
traditional service for paper or electronic return preparation experienced a lower
average service time (45.8 minutes) than FSA users. This difference was possibly
a function of experience between IRS employees and most taxpayers in preparing
tax returns. The difference could also have been due to the learning curve associated with using FSA. Since using computers to complete their service tasks might
have been unfamiliar to some taxpayers, it is possible that it took them longer to
complete their tasks.
Despite it taking longer, nearly all FSA respondents (95 percent) who completed the task of filing a return using an FSRP workstation reported they were
“Satisfied” to “Very Satisfied” with the service they received. Exit Survey data and
taxpayers’ open-ended comments appear to indicate that shorter wait times but
slightly longer service time was a valuable trade-off.
48
Davis, Hayes, and Jenkins
Perceived Issue Resolution
To determine how perceived issue resolution compared between taxpayers using
FSA and taxpayers who used traditional TAC service, FSRP data were compared
with national customer satisfaction data from FA’s transactional survey for the
same period. Data regarding perceived issue resolution using traditional TAC service were obtained from the FA National Report for January 2009 through April
2009.10 Figure 6 shows that there was no difference in perceived resolution rates
between FSA and traditional TAC service for forms or publications. However, perceived issue resolution for Free File using traditional TAC service was higher than
using FSA by 10 percentage points. As mentioned earlier, this was possibly related
to differing experience levels regarding tax preparation between taxpayers and IRS
employees.
FIGURE 6. Issue Resolution for FSA Workstations vs. Traditional TAC Service
FSA
Traditional TAC Service
Forms or Publications
95%
95%
Free File
86%
96%
100%
95%
96%
95%
86%
80%
60%
FSA
40%
Traditional
TAC Service
20%
0%
Forms or
Publications
Free
File
Primary Service Task
Sources: FSRP Exit Survey data, January 2, 2009, to April 30, 2009 (n = 4,603); Field Assistance
National Report, January 2009 through April 2009 (n = 112,188).
Overall, most FSA users (89 percent) reported that they were able to get answers
to their questions or complete their transactions. Issue resolution was examined
by age, total household income, and service task to determine if resolution rates
among taxpayer segments varied. Younger FSA users (44 and under) achieved
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
49
lower rates of issue resolution than their older counterparts (84 percent vs. 91 percent). FSA users with a total household income of $45,000 or less reported lower
issue resolution rates than their counterparts (83 percent vs. 91 percent). FSA users whose task was forms or publications had the highest issue resolution rate (95
percent), and taxpayers who had other less common tasks had the lowest issue
resolution rate (76 percent). A high resolution rate among FSA users who sought
forms or publications was expected, since this is a less complex service task.
These findings were not surprising as FSA users whose total household income
was more than $45,000 and/or who were 45 or older were more likely to visit a
TAC for a form or publication, while younger and/or lower income FSA users
were more likely to use FSA for Free File. The lower issue resolution rate among
Free File users may possibly be attributed to the users’ experience level using computers, IRS.gov, or tax preparation software.
FSA users’ comment regarding issue resolution (and service time):
I was really impressed by the fast service that I got today! All of my
questions and concerns were answered. I know that if I have any more
questions then I can use the website or come to my local IRS office to
receive further assistance.
Ease of Use
Taxpayer value was assessed by asking FSA users how easy it was to use the new
service option. Respondents who stated that their issue was resolved through FSA
were asked how easy it was to use FSA. Almost all taxpayers (94 percent) indicated
that FSA was “Just About Right” to “Very Easy to Use.” Further, the majority (65
percent) of Exit Survey respondents who reported issue resolution felt that FSA
was “Very Easy to Use.”
Due to a programming error in the Exit Survey, respondents who did not receive issue resolution were not asked to indicate how easy FSA was to use or their
satisfaction with the new service option; therefore data are limited to FSA respondents who perceived that they achieved issue resolution. However, as stated above,
a majority (89 percent) of FSA respondents perceived that their issue was resolved.
FSA users’ comments regarding ease of use:
This was very easy and stress free
The information was easily accessable (sic) and user friendly. I
appreciate this service and will recommend it to my family and
friends.
50
Davis, Hayes, and Jenkins
The experience for e-filing was easy to navigate. Would recommend
it to family and friends.
Having the assistance of [IRS facilitator name], made the experience
of filing on line much easier. As long as there is someone present for
assistance, I think that people will get used to filing on line and it will
become easier.
Satisfaction with FSA
Taxpayer value was examined relative to the level of satisfaction with FSA. In general, nearly all (96 percent) respondents who reported issue resolution indicated
that they were satisfied with FSA.
Findings indicate that there was a direct relationship between satisfaction and
ease of use. Almost all (98 percent) individuals who indicated that using FSA was
“Very Easy” to “Just About Right” reported that they were satisfied with services.
Of those taxpayers who indicated that FSA was either “Somewhat Difficult” or
“Very Difficult” to use, 62 percent reported that they were satisfied with services.
This finding suggests that, although some individuals had a more difficult time using FSA, the majority were still satisfied with the service they received.
FSA users’ comments regarding satisfaction with service:
Service was excellent! They provided me with any and every answer
I needed to know. I am very satisfied with my first attempt to file
my own taxes. It was a wonderful experience and I will do it again.
Thank you.
THANK GOD FOR E-Z TAX RETURN.COM I WAS VERY MUCH
SATISFIED, AND I LEFT SMILING. ALSO MY ASSISTANT [IRS
FACILITATOR NAME], WAS A GREAT HELP. THANKS AGAIN.
KEPT (sic) UP THE GOOD WORK.
This was a very productive trip to the IRS. I never knew that I could
excess (sic) the computer to receive forms and publications that were
not available at the office upon my visit there. Your people at the office
were very helpful to me. I enjoyed my visit.
Taxpayer Expectations
Expectations about service may have played a role in how satisfied taxpayers were
with FSA. Data from an open-ended question on the Exit Survey suggested that
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
51
some taxpayers’ dissatisfaction may not be with FSA but with not being able to get
the service that they expected.
FSA users’ comments regarding expectations:
A supply of commonly used forms, such as those used to file automatic
extensions for both Trusts and partnerships should be kept in the
IRS office and immediately available. It should not be necessary to
download such common forms.
I came in for HELP and they send you to a computor (sic) where I
would like to talk to a person.
I came for 2 forms for 2007 tax year and had to wait in line, and then
had to use computer. Took 20 minutes…To (sic) long!
As shown by the above FSA users’ comments, some individuals entered the
TAC with specific service expectations. In particular, some individuals expected
immediate access to certain forms or publications that were not readily available
through traditional TAC services.
Areas for Improvement
Survey respondents who perceived that they did not receive issue resolution (11
percent) were asked why their main issue was not resolved. The most common
response selected was “Other, please specify” (28 percent). The second most common response selected was “Could not find the information I needed” (22 percent).
Despite a high proportion of respondents reporting issue resolution, 164 respondents provided written comments related to not getting their issue resolved.
When examining open-ended comments, several themes and areas for improvement emerged including:
yy FSRP computer workstation issues,
yy Lack of assistance in finding what they needed,
yy Inappropriate study recruitment and
yy Free File website issues.
FSA users’ comments regarding suggestions for improvements:
You need more then (sic) one person to help taxpayers on the computer.
After filing an e-mail or notification should be available. This would at
least confirm that the information (data) was received and reviewed.
52
Davis, Hayes, and Jenkins
It would have been more helpful to have someone at this site trained
to answer my questions which I did not think were that uncommon:
my question had to do with Sch E rental income...
Need more terminals to assist tax payers
When the Federal government website says “free e-file” I believe the
Federal government should make sure the chosen contractors honor
the advertisement.
Data Limitations and Study Challenges
In addition to the project’s sampling methods, as discussed in the Research
Methodology section, other issues potentially affecting the research results were
uncovered. These limitations were related to IRS network performance, survey
design, and inconsistency of project operations at TAC sites. Most of these difficulties related to the challenges of performing operational research in live production
and service environments.
Intake Survey data transmission issues were experienced by all project sites during the entire data collection period—tax season 2009. There were no detectable
patterns for missing data at particular project sites, within FA Areas, or in general.
However, there was one project-wide spike in lost data during early February, but
MITS was unable to identify the cause of the increased loss. Although efforts were
made to remedy the problem, changes to the system were not implemented prior
to the end of data collection. Therefore, information regarding demographics was
not available for analyses for about 23 percent of study participants.
Survey design influenced how data were collected, and, ultimately, the findings
derived from data analysis. One such design issue was a skip pattern that was inaccurately programmed into the Exit Survey. This issue directly affected how Exit
Survey questions regarding ease of use, taxpayer satisfaction with service received,
and plans to seek further assistance can be interpreted. Respondents only received
these questions if they indicated that they received resolution for the tax issue that
brought them to the TAC. As a result, it was not possible to determine opinions
about satisfaction, ease of use, or plans to seek further assistance from individuals
who did not achieve issue resolution. While analysis of responses to open-ended
questions provided some information to address these areas, it would have been
more informative to know how participants who did not achieve issue resolution
felt about FSA.
Another survey design issue was that Exit Survey question 7 was not exhaustive in its response offerings. This question stated, “In the future, would you use
the IRS website (IRS.gov) again for any of the following services? (Please select all
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
53
that apply).” Unfortunately, the option, “No future intended use,” was not offered.
Therefore, it was not possible to distinguish people who did not intend to use IRS.gov
in the future from those who did not answer the question.
In spite of standardized training sessions and materials for IRS managers and
frontline employees, and regular, frequent conference calls with each of these
groups, project operations varied among FSRP sites. Differences were observed
regarding the proportion of taxpayers identified as eligible as well as study participation rates. For example, the participation rate in TACs ranged from 5 percent
to 99 percent. This evidence, along with data from “shopping” visits to project
sites, suggested that participant recruitment was not always performed according
to project protocols. Employee buy-in, manager support, and adequate staffing appeared to be predictors of following prescribed project operations.
Despite data limitations, sufficient amounts of data were collected in order to
increase the IRS’s understanding of potential taxpayer and government value of
FSA. Approximately 8,000 taxpayers used FSA workstations and over 5,000 responded to questions regarding their experience.
Conclusions
Taxpayers who were willing to use FSA generally reported that issue resolution
was obtained, that the system was easy to use, and that they were satisfied with the
service received.
yy The time expenditure trade-off between wait and service time
appears to be valuable to taxpayers.
yy FSA is a practical option for taxpayers who visit TACs for services
available on IRS.gov.
yy Findings suggest that facilitators are a necessary component of
FSA, and that knowledgeable facilitators helped make taxpayers’
experience positive and beneficial.
yy FSA provides value to taxpayers by increasing their awareness of
IRS online services.
yy Areas for improvement of FSA include computer systems and staff
support.
Recommendations
Based on the findings of this report, several recommendations were presented in order
to assist FA in making business decisions regarding the future delivery of FSA.
54
Davis, Hayes, and Jenkins
yy WIRA and MITS continue to collaborate with Field Assistance to
develop selection criteria for the potential future implementation
of FSA.
yy If it is not feasible for all TACs to administer FSA, focus
implementation in TACs with high volumes of taxpayers seeking
tax return preparation assistance.
yy Since a large number of taxpayers enter TACs to obtain forms,
consider implementation of separate dedicated “express lane”
computers for tasks that do not require the input of personal
identifiers (i.e., names, social security numbers, etc.), such as
obtaining forms or publications.
yy Self-assistance remain facilitated with adequate staffing made
available, to ensure that taxpayers are receiving needed assistance.
yy FA continue to “sell” FSA to employees before selling it to taxpayers.
Successful implementation of FSA is dependent upon TAC staff
and managerial engagement. IRS employees must understand the
importance of their role in making FSA a success, by increasing
awareness of IRS.gov and in helping taxpayers develop confidence
in performing tasks using FSA. This may ultimately free TAC
employees to answer more complicated tax questions as well as
increase taxpayers’ awareness of additional service channels to
address tax needs.
yy Facilitators remain available and knowledgeable about IRS.gov
to ensure taxpayers receive needed assistance. In open-ended
comments, many taxpayers discussed how facilitators played an
important role in helping them navigate the Web site, particularly
for Free File, as well as how the experience was more positive with
assistance.
yy It is recommended that the TACs have adequate staffing in order
for FSA to make taxpayers’ experience positive and beneficial. It
may be more beneficial to taxpayers to have at least one person
dedicated as a facilitator.
yy Before implementing FSA in additional TACs, FA and MITS reevaluate the current system design. It is important for the IRS
to determine how the system as a whole can be reengineered to
better support the future dynamic needs of FA. Expansion should
include additional thorough testing of network capabilities, and all
identified computer system limitations should be understood and
addressed. Although the majority of customers indicated that they
were satisfied with services, many of the individuals who gave a
Facilitated Self-Assistance Enhances Taxpayers’ TAC Experiences
55
reason for not getting their issue resolved cited computer or system
problems. While the goal is to encourage taxpayers to use IRS.gov,
if they have bad experiences using FSA in the TACs, they may be
less inclined to use IRS.gov, including FSA, in the future.
Endnotes
1
Internal Revenue Service, The 2006 Taxpayer Assistance Blueprint Phase 1.
Internal Revenue Service, The 2007 Taxpayer Assistance Blueprint Phase 2.
3 Ibid., page 109.
4 Ibid., page 113.
5 Ibid., Figure 2–24, page 52.
6 For details regarding the first phase of FSRP, see W&I Planning, Research,
and Analysis, “Facilitated Self-Assistance Research Project Research Report,”
September 8, 2008.
7 W&I Research and Analysis, “Facilitated Self-Assistance Research Project
Phase 2, Research Report for Field Assistance” September 30, 2009.
8 An additional 2 percent of the sample were identified as “couples,” thus lacking
a specific gender assignment.
9 Service time for face-to-face service is collected in Business Objects in
taxpayer units and hours such that average service times could be calculated
and compared with FSRP data.
10 The FA transactional survey asks questions regarding the nature of the
taxpayer’s visit, demographic characteristics, perceived issue resolution, and
satisfaction with TAC service.
2
3
D
Drivers of Noncompliance
Corcoro Adelsheim
Carley Hirshman Lee Martin
Robertson St. Charles
A Balance Due Before
Remittance: The Effect on
Reporting Compliance
Paul Corcoro and Peter Adelsheim, Internal Revenue Service
T
he relationship between a taxpayer’s prepayment position and the understatement of taxes has been observed since at least 1969, but has not been
the direct focus of much tax compliance literature. There is evidence that
balance-due taxpayers have been found to understate their taxes more often than
refund-due taxpayers.1 The goal is to examine the hypothesis that prepayment
position causes a portion of reporting noncompliance.
This research provides the Internal Revenue Service (IRS) with insight to the
nature and behavior of the taxpayer population. There have been policy changes that temporarily change taxpayer’s withholding intending to stimulate the
economy. Congress enacted such a stimulus in 1992 with the intention of injecting $2 billion per month in spending that year.2 However, since the tax liability
was not adjusted there was a greater possibility of taxpayers being under withheld.
Thus in 1993, there was an increased chance in having to pay money back to the
government. More recently, it is anticipated that the Making Work Pay Credit in
section one of the American Recovery and Reinvestment Act of 2009 (ARRA)
may cause an increase in balance-due taxpayers since not all taxpayers qualify for
the additional tax credits to offset the change in their withholding.3 These policies of temporary withholding changes could cause an unanticipated prepayment
position and in turn increase underreporting noncompliance. With this research,
the IRS could address potential compliance problems associated with any future
withholding changes, as well as increase the transparency of the taxes collected.
Whether a taxpayer has a balance due or refund due is determined by the timing
of the payments within the year. Prepayment position does not change the amount
of tax liability, yet the different behavioral shifts in reporting compliance violate
standard economic assumptions. In a way, this research also adds to a debate on
standard economic theory versus behavioral economics, reference-dependent
theories. Growing literature within behavioral economics has noted that individuals violate standard economic assumptions and exhibit behavioral shifts dependent
on frames of reference. Other literature in support of standard economic theory
found that these behavioral shifts dissipate with increased information and experience. This research can act as an empirical data-driven test for the behavioral
theory’s viability to complement the experimental evidence already found.
4
Corcoro and Adelsheim
Purpose and Structure of Report
The report is organized as follows:
• Background and Objectives—Presents a review of previous tax
compliance work outside the IRS and overviews goals of this paper.
• Research Methods—Provides a description of the methodology
used to analyze the interactions of a taxpayer’s prepayment position
and reporting compliance.
• Research Findings—Presents the results of the models.
• Conclusions and Recommendations—Summarizes the report and
proposes future work based on the information gained from the
models and hypothetical policy changes.
• Appendices—Provides a detailed description of the methodology
and data analysis developed for this study.
Background and Objectives
Tax Compliance
One of the earliest theoretical tax compliance models was established by Allingham
and Sandmo (AS) (1972). A taxpayer makes compliance decisions based on expectations of an audit, their risk aversion, and the total amount of their assets.
Their seminal paper focused on varying tax rates and reporting compliance. Their
theory implies that tax compliance can be increased by increasing penalties or
increasing the probability of an audit by increased examinations. AS’s framework has been used as a baseline model to analyze a number of influences to tax
compliance.
A recent paper utilizing the AS framework is Kleven et al. (2009). Their research is particularly noteworthy, for its ingenious new dataset that is similar to
the Taxpayer Compliance Measurement Program (TCMP) and National Research
Program (NRP) datasets. With cooperation from the Danish Inland Revenue
(SKAT), Kleven et al. set aside a stratified sample of roughly 40,000 Danish individual tax filers who were followed for the 2007 and 2008 tax years. The first year,
half of the sample was audited while the other half was deliberately not audited.4
The following year the whole sample was broken into three mailing groups. Two
of the groups received letters informing that they were randomly selected to face a
‘threat of audit.’ This left the final third as a deliberately not audited control group.
The exogenous audit probabilities allowed them to examine the causal effects of
prior audits and threats of audits.
A Balance Due Before Remittance
5
Like previous IRS research (Ho (2003), SB/SE Research Seattle/San Jose (2007))
focusing on varying levels of income visibility (self-reported to third-party
reported income) Kleven et al. found that the Danish population as a whole was
largely compliant, but there was significant tax evasion in self-reported income. In
addition, both prior audits and threat of audits increased the self-reported income
compliance.
The tax compliance survey by Andreoni, Errard, and Feinstein (1998) does an
excellent job summarizing the variety of theoretical and empirical research on tax
compliance including how it has evolved from the AS model. As noted in their
survey, examples of empirical tests using the AS model have focused on the influences of tax preparers, interactions with labor supply decisions, and cases when
the audit probability is endogenously determined. The difference in a taxpayer’s
balance due or refund prepayment position is the timing of the payments within
the year; in both cases, the tax liability remains the same. This survey like many
others is silent on the influence of a taxpayer’s prepayment position on compliance.
Prepayment Position
Under the Allingham and Sandmo framework, a taxpayer’s prepayment position
would not matter since their total tax liability remains unchanged; due to this,
much of the tax compliance literature is largely silent on prepayment position being a contributing factor to noncompliance. Taxpayers with a balance due have received an interest-free loan from the government, whereas refund taxpayers have
given the government an interest-free loan.
In times of a sluggish economy, changes in Federal Withholding Tables have
been used to invigorate the economy. In theory, the reduced withholding in selected tax brackets would stimulate the economy by boosting the respective households’ take-home pay increasing the demand for goods and services. Congress
enacted such a stimulus in 1992 with the intention of increasing consumer spending by $2 billion per month.5 However, tax liabilities were not adjusted; thus an
estimated 8.9 million taxpayers would hold an unexpected balance due when
filing taxes in 1993. More recently, it was estimated that the Making Work Pay
Credit portion of ARRA may cause an increase in balance due taxpayers. Not all
taxpayers are qualified for the additional tax credits to offset the change in their
withholding.6 According to a 2009 TIGTA report, “the Making Work Pay Credit
is to be advanced to taxpayers through their wages by a decrease in Federal income
tax withholding. This creates the vulnerability that some taxpayers may have their
taxes underwithheld at the end of Tax Years 2009 and 2010.” The report found
that more than 15.4 million taxpayers could unexpectedly owe taxes for the 2009
tax year since they were advanced more of the credit than they were entitled to
6
Corcoro and Adelsheim
receive. TIGTA analyzed the changes to the withholding tables to identify groups
that could potentially be advanced more Making Work Pay Credit than they are
entitled to receive. They determined the tables did not account for taxpayer situations such as:
• Dependents who receive wages;
• Single taxpayers with more than one job;
• Joint filers in households where both spouses work or where one or
both spouses have more than one job;
• Individuals who file a return with an Individual Taxpayer
Identification Number (ITIN);
• Taxpayers who receive pension payments; and
• Taxpayers who are employed and receive Social Security or similar
benefits.
In response, efforts have been made to publicize this issue and instruct affected
taxpayers to adjust their withholding accordingly. How well the message was received is unknown.
Absent of outside influences, theoretical discussions attributed transaction
costs and self-imposed, forced savings as reasons for a taxpayer’s prepayment position.7 From the U.S. Treasury—Fact Sheet on the History of the U.S. Tax System—
“[Prepayments] greatly eased the collection of the tax for both the taxpayer and
the Bureau of Internal Revenue. However, it also greatly reduced the taxpayer’s
awareness of the amount of tax being collected, i.e. it reduced the transparency of
the tax.” The transaction cost argument is the following—given that withholding
reduces the transparency of the tax liability, individuals may optimize and find
that the costs of properly adjusting their withheld income exceed the benefits.
The forced-savings argument was born from some observations that a number of
taxpayers voluntarily overprepay. The taxpayer optimizes by realizing that they
cannot properly save income on their own, thus purposely over pay their withholding tax.
Christian et al. (1993) examined the relationship between prepayment position
and tax preparers. Their work found that “paid-prepared returns have lower tax
liabilities and that the reduction in tax liability is larger than the reduction in total
prepayments.” Noncompliance was out of the scope of the paper, but they suggested that further research be done in regards to compliance and prepayment.
There has been anecdotal and some experimental evidence that found balance due
taxpayers are more likely to be noncompliant. Empirical examinations by Chang
and Shultz (1990), and Adelsheim (1997) using the 1988 Taxpayer Compliance
Measurement Program (TCMP) data found this positive relationship between a
taxpayer’s prepayment position and their underreported tax liability.
A Balance Due Before Remittance
7
Prospect Theory
Kahneman and Tversky (1979) created prospect theory to explain situations not
addressed by expected-utility theory, in particular shifts from risk seeking to risk
aversion and vice versa.8 Prospect theory, a branch of behavioral economics, differentiates itself from expected-utility theory in three distinct ways. First, individuals
make decisions based on a self-assigned intrinsic value relative to a neutral reference point rather than the individual’s final monetary assets. Second, individuals
are risk-seekers in loss domains and risk-averse in gain domains and the value
function is steeper for losses than for gains. The curvature implies that a loss has
a greater impact in an individual when compared to an equivalent gain. Finally,
individuals’ underweigh probable events when compared to certain events.9
This theory fits well as a framework for the taxpayers’ apparent change in behavior due to their prepayment positions. Come filing season, a taxpayer would
have an expectation of their tax liability, assumed to be zero additional liability.10
This expectation would be the individual’s neutral reference point. Depending on
the taxpayer’s prepayment position, the individual may act risk seeking to lessen
a balance due payment (a perceived loss) or risk averse to preserve a tax refund
(a perceived gain). The taxpayer’s behavior was presumably influenced by the
taxpayer’s perception of his probability of an audit and that probability of audit
(and his perception of the probability) is generally influenced by what he reports
on his return.11
Such behavior has been observed in laboratory experiments. In these experiments all subjects hypothetically received the same level of net income by using
“framing effects,” net incomes were framed from either a gain (a refund due) or a
loss (an additional tax balance due). An individual in the gain framework would
be told that, in addition to their $800 of assets, they would expect a refund of $200.
An individual in the loss framework would be told that their current assets were
$1100, and they would expect to pay $100. Either way, all individuals would have
a net income at $1000 if a compliant return was filed. Depending on the frame of
reference, the experiments found that those in a balance-due position were more
likely to understate their tax liability.12
How the Current Study Differs from the Previous IRS
Research
The main hypothesis in this research is that prepayment position causes a portion
of underreporting noncompliance, whether this relationship exists or is simply a
statistical illusion. Given a statistical relation, the direction of causation would
need to be examined and the magnitude of the effect estimated. This question
has not been examined in the context of the 2001 National Research Program
(NRP). Unlike the TCMP data, the 2001 NRP can be linked to prior tax year
8
Corcoro and Adelsheim
data. Examining the prepayment relationship with the TCMP and NRP datasets
may lead to some general insights, for example, whether the incidence and size of
the balance due effect is increasing, or whether the characteristics of balance due
taxpayers are changing.
The previous IRS work mentioned above used Analysis of Variance (ANOVA)
and did not analyze the prepayment effect in a fully multivariate framework. The
ANOVA method controlled for single categorical causes of noncompliance one at
a time. The previous work indicated a relationship but did not quantify it. Here
a regression framework is used to simultaneously control for both categorical and
continuous causes of underreporting noncompliance in order to isolate and quantify the hypothesized marginal impact of a taxpayer’s prepayment position. The
regression analysis in this study is not meant to predict reporting noncompliance;
all other predictor variables were included specifically to isolate the impact of a
taxpayer’s prepayment position not to predict reporting compliance.
To further test the issue of causation, refinements to the regression model were
needed. With the use of nonexperimental data like the TCMP and NRP datasets,
a taxpayer’s prepayment position and underreporting compliance are both determined by the taxpayer’s behavior; in other words, they are endogenous. A prepayment position could cause a taxpayer to underreport, or both instances could be
jointly caused by another external factor. If this external factor is not accounted
for, then the estimated marginal impact of prepayment is potentially biased and
inconsistent.13 Two additional methods are used to account for this causation
issue. The first is to focus on a subsample of the population where the taxpayer’s
prepayment position is assumed to be less endogenous. The other is to use instrumental variables (IV).14 One equation estimates the taxpayer’s prepayment position, while a second estimates understatement of tax liability.
In addition, this research could further contribute to the debate regarding the
viability of prospect theory.15 The research can act as an empirical data-driven
test to complement or refute the experimental evidence already found in support
of prospect theory. If the results show that prepayment position causes noncompliance, this knowledge may help quantify consequences of adjustments to the
Federal Withholding schedules. Finally, this research can also distinguish other
characteristics of noncompliant returns by looking at the significance of the demographic predictors in the model.
Research Methods
Data
Research used archival taxpayer data collected for the 1988 TCMP and the 2001
NRP. Both datasets are stratified random samples and all regression analysis was
weighted using NRP population weights to account for the stratification. The
A Balance Due Before Remittance
9
older TCMP dataset was used to verify whether the paper’s results were in line
with the previous IRS research using TCMP. Like the Danish dataset generated in
Klevin et al., all taxpayers in the TCMP and NRP were audited at random so the
probability of audit is not determined by any characteristics of the return.
Two methods of regression analysis were performed on the data and will be
discussed in detail in the methodology section. One method used the audit year’s
return information, the other required prior-year taxpayer information. In order
to keep the dataset consistent between the two estimation methods, roughly 3,000
cases in the 2001 NRP dataset were excluded.16
All the analysis used the audited ‘as corrected’ values for the predictor variables
rather than the ‘per return’ or ‘reported’ values. This was done to minimize the
correlation between all predictor variables and the immeasurable taxpayer noncompliance. The line item values reported ‘per return’ are likely highly correlated
with a taxpayer’s compliance inclination. Because of this high correlation, analysis
using reported values would lead to the opposite conclusions. A noncompliant
taxpayer would likely want to report a smaller than usual balance due or larger
refund. It is counterintuitive for one to be noncompliant by paying a large sum
to the government. The only variable using the taxpayer ‘reported’ values is underreporting, which by definition is the difference from the reported and the as
corrected liability. All the results derived in the following sections were from the
audited/corrected values and cannot be directly translated to any operational data.
Operational data would only contain the reported values.
A LARGE NUMBER OF THOSE WHO HAVE A BALANCE DUE ALSO
UNDERREPORTED TAXES
Figure 1 and Tables 1–4 show the descriptive statistics of the stratified TCMP and
NRP datasets. Figure 1 shows a pair of unweighted histograms of the difference between reported and actual tax liability for both datasets with positive values corresponding to underreporting.17 If the errors were the result of mistakes, one would
assume that errors were randomly distributed, with more or less equal numbers of
over- and underreporting cases. However, both graphs exhibit a longer right-hand
side tail in the distribution. This implies some amount of underreported taxes that
cannot be attributed to random error.
The issue being addressed here is whether some of this underreporting noncompliance can be attributed to a taxpayer’s prepayment position. Table 1 outlines a descriptive relationship between the taxpayers’ prepayment position and
their compliance in reported tax liability after both datasets were adjusted for the
sampling stratification. The total percentages at the bottom of the table represent the estimated percent of the population that is in either prepayment position.
According to the estimates about one in four taxpayers carried a balance due. The
TCMP and NRP datasets roughly relay the same information: a higher proportion
10
Corcoro and Adelsheim
FIGURE 1. Distribution of Change in Tax Liability (Underreport > 0 , Over report < 0)
SOURCE: 2001 NRP raw data reflecting only what the examiners detected.
SOURCE: 1988 TCMP raw data reflecting only what the examiners detected.
11
A Balance Due Before Remittance
of taxpayers who had a balance due also underreported their taxes (60.02 percent
in TCMP, 61.99 percent in NRP) when compared to the proportion of taxpayers
without a balance due after prepayments and underreported taxes (32.51 percent
in TCMP, 32.28 percent in NRP).18
TABLE 1. Descriptive Statistics—By Prepayment Position† & Reporting Accuracy
Reported tax
liability
TCMP
TCMP
NRP
NRP
no balance
due
balance
due
no balance
due
balance
due
5,500,863
2,012,321
6,120,977
1,728,002
7.15%
7.34%
7.50%
6.06%
46,420,000
8,949,615
49,140,000
9,105,742
Overreported
Correctly reported
Underreported
Total
60.34%
32.64%
60.22%
31.95%
25,010,000
16,450,000
26,350,000
17,670,000
32.51%
60.02%
32.28%
61.99%
76,930,000
27,420,000
8,161,000,000
28,500,000
73.73%
26.27%
74.12%
25.88%
Frequency
Col Pct
Frequency
Col Pct
Frequency
Col Pct
Frequency
Percent
*Samples are weighted to compensate from stratification.
**Totals are the weighted estimates of the population as a whole.
†As
established by the audits.
SOURCE: Raw 1988 TCMP and 2001 NRP data.
In both datasets, of the taxpayers who carried a balance due prepayment position, less than 8 percent (7.34 percent TCMP, 6.06 percent NRP) were found to
have overreported their tax liability. Slightly less than one-third (32.64 percent of
TCMP and 31.95 percent of NRP) of the balance due group did not need an adjustment to their tax liability. Amongst the taxpayers who had a balance due, 60 percent also underreported their taxes. Both random audit data sets imply a correlation between reporting noncompliance and a balance-due prepayment position.
The previous IRS work by Adelsheim (1997) noted that in the taxpayer profiles
of those who were slightly underwithheld and had a small balance due had a very
similar profile to taxpayers who were owed a refund. Whether a taxpayer holds a
slight balance due or a refund might be due to some randomness in slight reporting errors and miscalculations.
Taxpayers with large levels of underwithholding (resulting in large balance due
payments) appeared to act more risky with larger underreporting of their tax liability. A reproduced version of the profile is Table 2. Returns with large balance
due prepayment positions were associated with a large amount of underreported
tax and had higher proportions of additional schedules. The average understated
tax was $104 for the non-balance due group and $134 for the group with small
balance due. Both were a fraction of the average underreported tax of the large
balance-due group, which was $976.
12
Corcoro and Adelsheim
TABLE 2. TCMP Descriptive Statistics—Profiles by Prepay Position†
Exact withheld/
refund due
Balance
due < $120
Balance
due > $120
Mean / %
Mean / %
Mean / %
$104.00
$134.00
$976.00
Prepayment position
($934.00)
$54.00
$3,142.00
% w/ Interest income
57.30%
73.60%
79.34%
% w/ Dividend income
16.70%
22.68%
34.28%
% w/ Sch C
7.17%
8.90%
26.60%
% w/ Sch D
8.12%
7.96%
22.99%
% w/ Rental
9.21%
8.83%
24.04%
% w/ Sch F
1.36%
1.30%
4.57%
N
26,124
1,637
22,649
Variable
Understatement of tax
*Negative values denote overreporting / refund due.
†As
established by the audits.
SOURCE: Weighted 1988 TCMP / Adelsheim 1997.
Table 3 is a similar profile with the weighted NRP data set. Again the data shows
that the change in tax liability discovered via the random NRP audit is roughly the
same magnitude for taxpayers who had a refund due and those who had a small
balance due (an additional $189.94 for refund due taxpayers and $293.43 for small
balance-due taxpayers). Taxpayers who had a large balance due had a tax change
roughly 10 times as large as the refund-due and small balance-due taxpayers.
Also, taxpayers with a large balance due often had a higher proportion of attached
schedules.
TABLE 3. NRP Descriptive Statistics—Profiles by Prepayment Position†
Variable
Understatement of tax
Exact
withheld
Refund
due
Balance
due—small*
Balance
due—large
Mean / %
Mean / %
Mean / %
Mean / %
($23.30)
$189.94
$293.43
$2,468.17
Prepayment position
$0.00
($2,284.63)
$179.52
$6,135.30
Primary age
57.84
41.89
46.50
50.92
% w/ Sch Aw
18.48%
36.53%
29.78%
52.26%
% w/ Sch C
10.09%
10.79%
15.91%
37.07%
% w/ Sch D
33.05%
19.43%
23.73%
33.93%
% w/ Sch E
22.40%
10.90%
11.27%
27.15%
% w/ Sch F
2.81%
1.25%
1.90%
3.94%
% w/ Interest income
75.18%
58.38%
67.04%
75.68%
% w/ Dividend income
46.26%
26.15%
31.66%
40.28%
589
21,011
2,027
18,252
N
*10th percentile of balance due.
**Negative values denote overreporting / refund due.
†As
established by the audits.
SOURCE: Weighted 2001 NRP raw data reflecting only what the examiners detected.
13
A Balance Due Before Remittance
The higher frequency of a Schedule C and other attachments along with balance due and underreporting falls in line with previous research.19 This link has
been attributed to a number of reasons. The method of withholding is different for
self-reported income. Reasons have been the added complexity of the return or
the presence of less visible income. Taxpayers who are self-employed small business owners (ones with Schedules C or F) may also be inherently less risk-averse.
Considerations were made in the model specifications to disentangle the effects
of pure prepayment positions and prepayment position as a proxy for these other
covariates.
Net Misreporting Percentage (NMP) is another way to view reporting compliance. NMP is defined as the sum across all observations of the net amount
misreported divided by the sum of the absolute values of the amounts that should
have been reported. Since misreporting induced by prepayment position might
show up on any line, including refundable credits, the NMP was calculated on tax
after refundable credits. Table 4 displays the NMP by various prepayment positions and income levels in the 2001 NRP dataset to show the major reasons why
the amount of misreporting varies, and to what extent the rate varies. The table
divides prepayment position by the 75th, interquantile range, and 25th percentiles of both balance due and refund due while excluding cases where the taxpayer
had neither a balance due or refund due (2 percent of the sample). Total Positive
Income (TPI) is also broken into three bins, TPI levels 1, 2, and 3, respectively corresponding to the 25th, interquantile range, and 75th percentiles.20
TABLE 4. Net Misreporting Percentage by Income and Prepayment Position
Corrected Prepayment Position
Reported Prepayment Position
TPI Level
(1)
(2)
(3)
(1)
(2)
(3)
Large refund
10.95%
3.20%
0.80%
55.12%
12.38%
3.83%
Medium
refund
8.19%
4.03%
2.06%
14.97%
7.87%
6.13%
Small refund
11.47%
4.90%
1.41%
12.02%
7.16%
22.47%
Small
balance due
16.32%
5.35%
1.56%
12.19%
6.25%
10.39%
Medium
balance due
26.55%
10.39%
2.81%
10.86%
7.85%
7.10%
Large
balance due
34.15%
28.95%
10.17%
2.71%
9.27%
3.96%
Large, Medium, and Small prepayment are designated by 75th, interquartile range, and 25th of BD and RD even prepay
comprised of 2 percent of sample and was included in small refund group. estimates weighted to compensate for
stratification
SOURCE: 2001 NRP raw data reflecting only what the examiners detected.
This table breaks NMP by prepayment position defined on a per-return basis in addition to the as-corrected definition and shows evidence that there is a
14
Corcoro and Adelsheim
difference in compliance broken down by corrected and reported prepayment position. Because of this, it must be emphasized that any of the results derived from
this research cannot be used with operational or reported data. If a taxpayer was
to be noncompliant, they would be very unlikely to report a large balance due,
but may be found to have a large balance due after the audit. The noncompliant
taxpayer would more likely report a smaller balance due or a larger refund rather
than a smaller refund. In terms of reported prepayment position, a taxpayer who
reports a large balance due is likely to be compliant.
Observing NMP by audited corrected prepayment position, there is a near
monotonic increase in NMP going from a large refund due to large balance due
at all income levels. A consistently higher portion of noncompliance was found
in the balance-due domain than the refund due domain. A higher percentage of
net misreporting was found in persons who are found to owe a large balance due.
The distribution of NMP by reported prepayment position is different than the
corrected values. While noisier, the table does show what was expected: a higher
proportion of taxpayers claimed large refunds due and a smaller proportion of
taxpayers claimed large balances due. Specifically when comparing the large refund group to the small refund group in TPI level 2. Of those in the large refund
group, there was 12.4 percent NMP compared to 7.2 percent in the small refund
group and a large balance due prepayment position does not necessarily translate
to larger proportions of NMP. The between-group differences are less consistent
due to the fact that each group is muddled with a greater mix of compliant and
noncompliant returns. A higher proportion of taxpayers have shifted into claiming better prepayment positions.
Theoretical Methodology
Taxpayers face a decision between complying by paying their full liability and noncomplying and facing an increased chance of an audit. The prospect theory model
assumes that an individual taxpayer’s compliance decision is partially determined
by their actual prepayment position. The taxpayer is assumed to sequentially do
the following:
• Enter the filing process with some expectation of either a refund or
balance owed based on prior experience;
• Draft a return;
• Realize the true prepayment position of a balance due, refund due
or neither; and
• Make a compliance decision (finalize draft or change income/
deductions/credits).
A Balance Due Before Remittance
15
In more detail, the model assumes that a taxpayer does not directly calculate
his or her full tax liability to make a compliance decision. Instead the taxpayer
assumes that after credits and deductions and their withholding prepayments that
they have properly paid their tax liability. This is their reference point when filing
their taxes, an expected zero additional liability.21 Once the tax return is drafted,
the individual realizes their prepayment position. Either they fall in a zero-
prepayment position, a balance-due position or a refund-due position. Based on
the prepayment position, the taxpayer makes a compliance decision. The taxpayer
then considers oneself to be in the loss domain if he or she has a balance due and is
more likely to act risk-seeking to minimize loss. If the taxpayer is in a refund due
position, he or she is likely to act risk averse to preserve their perceived gain. The
result is the characteristic S-shaped value curve associated with prospect theory.22
Empirical Methodology
WEIGHTED LEAST SQUARES
Least-Squares Regression analysis was used with the two datasets to explore the
hypothesis that taxpayers with a balance due before remittance understate their
tax liability more than an equivalent refund-due taxpayer.23 The NRP data used in
the analysis excluded roughly 3,000 cases. The excluded cases’ primary TIN did
not match a primary TIN used on a return for the prior 2 years. This was done
to keep the dataset consistent with the dataset used in the Instrumental Variable
refinement described later.
The weights adjusted to compensate for the oversampling stratification performed in TCMP and NRP. Necessary for the Weighted Least Squares (WLS)
estimation to be valid, the right-hand-side regressors must be exogenously determined. An exogenous prepayment position could result from a policy that
changed withholding for a given year. The regression was intended to simultaneously control for other factors that influence taxpayer underreporting and quantify
how much a person’s prepayment position affects underreporting. To get a baseline set of results, a WLS dummy variable regression was performed. Research
estimated with two model specifications; one using WLS on the NRP dataset, the
other to estimate after segmenting the NRP dataset, into three income groups.24
The model for an individual’s reporting decision can be stated as follows:
(1)
Where in equation (1) the parameters βj are to be estimated from the regression
and εu is the random error term.25
The individual’s understatement of taxes (u) will be measured as the difference
between the tax liability prior to refundable credits as determined by the examiner
and tax liability stated by the taxpayer.26
16
Corcoro and Adelsheim
VARIABLES OF INTEREST
Prepayment position is measured by the difference between the tax liability and
the total tax payments, both as determined by the tax examiner. This is broken
up between balance due and refund due.27 The outputs of interest are the para
meters β1 and β2 which are the marginal effects of the two prepayment positions
on reporting compliance holding all other predictors constant. If prospect theory
holds, it is expected that the parameter on the balance-due prepayment would
be positive (β1 > 0); a taxpayer would be more willing to run the risk of an audit
and underreport because of the perceived loss. Likewise the parameter associated
with refund due would be negative (β2 < 0); the taxpayer would act risk averse to
preserve his or her gain. Alternatively, if taxpayer behavior is guided by expected
utility as outlined in the standard AS model, prepayment position would have no
influence on the taxpayer’s reporting compliance; the parameters β1 and β2 would
not be significantly different from zero. An additional indicator variable captured
cases of no prepayment position where withholding exactly covers the tax liability. Its associated parameter estimate is expected to not be significantly different
from zero.
OTHER COVARIATES
In order to ensure that the parameter estimates β1 and β2 report the change in
underreporting caused only by prepayment position, the regression equation includes other profiling variables. The vector xc contains return-specific characteristics noted in the previous literature that may also explain this withholding
phenomenon, details are provided below. The intent is to isolate the effect of a
taxpayer’s prepayment position. This vector contains sets of dummy variables for
the following characteristics—different occupation codes, varying types of income
sources, and return complexity.28 The vector xCD contains select dummy variables
from xc which are interacted with the prepayment variables. The interactions were
done in attempt to further isolate the marginal effect of prepayment position on
compliance.
Liquidity constraints can explain why a taxpayer carried both a balance due
and underreported tax liability. The constraint could be caused by a myopic savings plan throughout the year or from a negative financial shock (car repair, medical costs) realized by the individual. If the individual had a balance due and liquid
assets to pay, they would pay the balance rather than face an audit. However if
the individual was liquidity constrained, without the available assets to pay off
the balance due, they might resort to underreporting. The indicator for interest
income serves as a proxy for liquidity if the taxpayer has sufficient interest income
to generate a Form 1099-Int then he or she likely has enough liquidity to pay off
their balance due.
A Balance Due Before Remittance
17
List (2004) found that the behavior outlined by prospect theory declines with
further experience, so the taxpayer’s age may play a role with their underreporting
compliance and prepayment position. Income is thought to play a role in compliance; however the pure TPI amount can be misleading since the purchasing power
can vary by states. An indicator on the existence of a state income tax and rescaling the taxpayer’s TPI by the 2001 median state income adjusts for regional effects
and influences of the current condition of the economy.
Indicators for Schedules C and F are intended to account for some of the characteristics such as return complexity, opportunity (or visibility) of certain fungible
income, and expenses items as detailed in the data section. In addition, the analysis included the calculated Discriminant Function System (DIF) score for each
taxpayer. In standard audits, some returns are selected for examination based on a
DIF score. The DIF score rates the potential for change in a return, based on past
IRS experience with similar returns. The highest-scoring returns are screened by
IRS personnel identifying the returns most likely to need review. Here, the DIF
score is used as a catch-all variable for noncompliance and accounts for a taxpayer’s expectation for audit. Because of game theoretic inner monologues, the
taxpayer’s expectation of their audit probability would influence what is reported
on their return. Hopefully the DIF captures these variations not identified with
the indicator variables.
Inherent with the least-squares regression analysis, research can observe
whether these other factors significantly contribute to underreporting. Research
can test the significance of the estimated vectors β3 to identify other drivers to
underreporting. The statistical significance of the covariates could also be used to
check the validity of the models. For example, if the models find that underreporting does not increase with higher DIF scores then there are likely issues with the
specification of the model.
EMPIRICAL REFINEMENTS
An issue in the model is causation. As shown in Figure 2, the amount that taxpayers underreport and their prepayment position are both determined by the taxpayer. Since both are determined by the taxpayer, it is possible that the two variables are jointly caused by a taxpayer’s unaccounted noncompliant behavior, so the
resulting estimates from the previous model could be biased and inconsistent.29
The DIF score can be used as a proxy for a taxpayer’s noncompliant behavior, but
it may not be sufficient.30 To attempt to account for the endogeneity of prepayment position, two approaches were used. Each used the NRP dataset joined with
return information from the previous 2 years.31
18
Corcoro and Adelsheim
FIGURE 2. Flow Chart of Causation/Endogeneity Issues
Taxpayer Compliance
Inclination
Withholding
Prepayment
Position
Reporting
Compliance
INSTRUMENTAL VARIABLES
One approach used to correct for endogeneity was Instrumental Variables (IV)
estimation.32 The system consisted of a pair of equations. Since an individual’s
underreporting depends both on prepayment position and his or her inherent
compliance inclination, the estimation must be performed in stages. The first
stage (equation 2) estimated a taxpayer’s prepayment position in terms of all the
external variables. The first stage regression fitted estimates of prepayment position were assumed to have corrected for the correlation between prepayment and
noncompliant behavior. The second stage (equation 3) consisted of a modified
version of the previously defined underreporting equation. The fitted estimates
used from the equation for prepayment position that are no longer correlated with
the error term.
(2) First stage:
(3) Second stage: X =
33 = δ R + δ 33W − + δ 33W − + ] ′į + ε 33
β R + β %Ö ' + β 5Ö ' + [&′ ȕ + LQWHUDFWLR Q WHUPV + ε X
For an individual taxpayer, the vector z contains profiling variables of prepayment position. In order to make the model tractable, identifying assumptions
mentioned below were made on the structural form of the system of equations.
Some of the variables in the vector z may also reside in xc but the vectors cannot be
identical or linear combinations of each other. The variables that reside in vector
z directly instrument for prepayment position are assumed to be correlated with
prepayment position, but are not correlated with the decision to underreport in
2001 (reside in vector xc).
A Balance Due Before Remittance
19
Variables that influence prepayment and the reporting decision are assumed to
be the current (NRP) year line items. These line items reside in both vectors xc and
z. It is assumed that the decision on reporting compliance is made annually; the
decision largely depends on the individual’s current tax situation. The taxpayer’s
decision on the amount to withhold, and thus his or her prepayment position is
thought to be more backward looking, adapting from the prior years. The taxpayer’s prior year’s prepayment positions from 1999 and 2000 and the change in
tax liability from 1999 to 2000 and from 2000 to 2001 (PPt-1 and PPt-2) were used to
instrument for the 2001 NRP tax year’s prepayment positions. While it is feasible
that what an individual’s prior years’ filing has influence on the current year, it is
assumed that prior years’ results only influence the 2001 reporting compliance via
the 2001 prepayment position.
By running the IV estimation on the system of equations, it can be verified
whether the results of estimating the previous WLS equation (1) may be an illusion
of statistical feedback. The Hausman test was used to determine the severity of the
endogeneity problem with the prepayment position variables.
SUBSET ANALYSIS
The second approach utilized the prior-year information to create a subset which
had taxpayers with relatively stable withholding throughout the prior 2 years, but
then realized a large absolute change in tax liability during the NRP year. Here it
is assumed that the steady withholding but large change in tax liability (as found
by the audit exam) resulted in an unexpected large prepayment position shock. If
this large prepayment position is unanticipated then it is not determined by the
taxpayer, thus reducing the endogeneity and the issue of bias. The rationale behind
the reduction of endogeneity is consumption smoothing. If a taxpayer was aware
of a large change in liability, one way to reduce the large financial shock would be
to change the amount withheld; distributing the additional tax paid (or money
received) throughout the year, rather than in a lump sum during tax season.
What accounted for ‘stable’ withholding was a withholding amount that stayed
within an upper bound of an absolute change of 15 percent of the previous year. A
‘large’ absolute change in tax liability had a lower bound of a 20-percent change.
Other bounds for ‘stable’ withholding and ‘large’ tax change were tested by incrementing each by ±5 percent and ±10 percent. The chosen subset data consisted
of 7,365 cases. This specification struck a balance between a robust sample and
sample size.33 Table 5 compares the profiles of this subset to the remaining population within the NRP dataset. Based on the attached schedules, the two populations appear similar, with the subset population having slightly higher proportions
of taxpayers with Schedule A attachments and interest income. For the subset
population, there is a near one-to-one relationship with the average change in tax
20
Corcoro and Adelsheim
liability and prepayment position. If there was a causal relationship, it would appear that the balance due amount nearly accounts for the change in tax.
TABLE 5. NRP Descriptive Statistics—Comparing Subset Population
Stable withheld &
large tax change
Remaining sample
Mean / %
Mean / %
Primary Age
49.600
42.744
Change in tax
547.00
674.01
Prepayment position
608.89
1337.48
% w/ Sch A
0.459
0.367
% w/ Sch C
0.133
0.169
% w/ Sch D
0.208
0.233
% w/ Sch E
0.124
0.147
% w/ Sch F
0.019
0.018
% w/ Interest income
0.713
0.602
% w/ Dividend income
0.308
0.293
N
7363
34499
Variable
SOURCE: Weighted 2001 NRP raw data reflecting only what the examiners detected.
Limitations and Deviations
Deviations from the original research plan revolve around one model assumption that was changed: the causal relationship between prepayment position and
reporting compliance. Initially, the plan was to test whether prepayment position
caused reporting compliance or reporting compliance caused prepayment position (via withholding amount). After the plan was submitted, it was thought that
prepayment position is not likely caused by the reporting compliance decision;
rather both were caused by the taxpayer’s willingness to comply with the tax authority. Because of the change in the assumptions, the estimation method changed
from a multivariate system of equations to Least Squares/Instrumental Variable
estimation.
Also, the emphasis on reporting the marginal effects of the other covariates has
been downplayed. All the regression results can be found in the appendix to see
the influences of the other variables. The results largely fall in line with previous
research, such as a higher DIF score corresponding to a higher degree of reporting
noncompliance. Some parameter estimates appeared counter to prior research,
such as greater reporting compliance from taxpayers with an attached Schedule
C in the TCMP regressions. However, accounting for interaction terms with the
attachment aligns the regression results to prior research. Fully dedicating a section outlining all the intricacies from the multiple models were thought to be too
A Balance Due Before Remittance
21
tangential and take focus away from the main objective in testing the significance
of prepayment position on reporting compliance.
One limitation of the research involves correcting for the endogeneity of a
taxpayer’s prepayment position. With the endogeneity, the WLS analysis would
report an upper-bound estimate of the effect of a balance-due payment and a
lower-bound estimate for a refund. The two model refinements may not fully
compensate for the bias generated from not being able to quantify a taxpayer’s
willingness to comply. The subset analysis of the 2001 NRP may not have completely isolated taxpayers who had an unexpected prepayment position. The fact
that the taxpayer had ‘stable’ withholding and a ‘large’ tax liability change from the
prior year does not necessarily mean that the prepayment position was a shock.
If the prepayment position was not a shock, then the subset did not correct for
the endogeneity. The IV estimation may have been mis-specified. The line items
chosen as instruments may be weakly correlated with prepayment position so the
estimates were a poor fit to the model or the instruments chosen are still correlated
with the reporting compliance error term.
Another potential limitation involves the reference point in terms of this prepayment behavior phenomenon. Schepanski and Shearer (1995) argued that the
neutral reference is not zero additional liability (no balance due or refund due)
but rather the expected prepayment position. With their expected asset condition
as the true reference point, they argue that a taxpayer who expects a large balance
due, but only realizes a small balance due would consider that within the gain domain, thus act risk averse. Likewise, a taxpayer who expects a large refund but gets
a marginal refund would view that as a loss, and may act risk seeking to capture a
larger refund. Given that a taxpayer’s expectations are not coded in the random
audit data, it is operationally difficult to test this behavior.
Research Findings
This research is to examine which prevailing economic theory coincides with the
reality of reporting compliance. Prospect theory posits that an individual will be
risk seeking in the face of a perceived loss; the same individual will act cautious
to maintain a perceived gain. Thus prospect theory would expect a balance-due
individual to seek relatively more risk, a refund individual, relatively less risk. In
addition, the theory states that a loss has a greater impact on an individual’s behavior when compared to an equivalent gain. The subsequent regression analyses
were formed in light of this theory. Standard utility theory would suggest that
prepayment position has no significant influence on tax reporting compliance;
if so, the estimates should not be significantly different from zero. The following
regressions were not intended to be used as a way to forecast or predict reporting
noncompliance.
22
Corcoro and Adelsheim
Impact of Prepayment Position on Reporting
Compliance—WLS
Using WLS estimation under a number of specifications, it was consistently found
that taxpayers’ behavior changed when presented with a balance-due or refunddue prepayment position and loss aversion.34 The parameter estimates from WLS
are reported in Table 6 and is segmented by modeling with and without interaction terms. The interactions were done in attempt to further isolate the marginal
effect of prepayment position on compliance and are further examined in Table 7.
The interpretation of the parameter estimates are marginal effects; ‘for an incremental dollar balance due/refund due, the tax change from audit is $X,’ with all
other variables held constant. The balance-due and refund-due prepayment position variables were coded in dollar terms. In the regression without interaction
terms, the marginal effects of the prepayment positions are the actual parameter
estimates associated with the prepayment variables.35
Referencing the ‘Exactly Withheld indicator variable, it was consistently found
that taxpayers who had perfectly prepaid their liability (received neither a balance
due or refund) were less likely to underreport their tax liability. Under the different model specifications, all of their associated parameter estimates were negative
and or not significantly different from zero.
TABLE 6. Prepayment Parameter Estimates from WLS
Dependent Variable: Misreported Tax
(underreport > 0, over report < 0)
Without Interaction Terms
Parameter
Bal due
Refund
No prepay pos
Full sample
TPI Level 1
0.162**
TPI Level 2
0.287**
(0.002)
0.465**
(0.006)
0.032**
(0.002)
(0.004)
TPI Level 3
0.151
(0.004)
−0.066**
−0.006
0.030
(0.004)
(0.005)
(0.004)
−304.656
−70.534**
−110.639
−2588.335
(208.849)
(28.699)
(191.140)
(7553.931)
With Interaction Terms
Parameter
Bal due
Refund
No prepay pos
Full sample
TPI Level 1
0.404**
0.378**
(0.011)
(0.022)
TPI Level 2
0.501**
(0.013)
TPI Level 3
0.383
(0.024)
−0.052*
−0.074**
−0.074**
−0.067
(0.030)
(0.020)
(0.016)
(0.157)
−340.829*
−64.225**
−300.256
−942.581
(200.949)
(28.138)
(185.499)
(7332.515)
Appendix F has Table F1, which outlines the parameter estimations different model specifications.
Standard errors in parenthesis, ** p < 0.05; * p < 0.10.
A Balance Due Before Remittance
23
Under the two model specifications with and without interaction terms, at a statistically significant level, holding a balance-due prepayment position was consistently shown to result in an increase in underreporting noncompliance. This
result holds using the full sample and at each income level. A different behavior is shown when taxpayers have a refund due. All specifications show that a
refund due is associated with smaller absolute changes in reporting compliance.
The reaction to the loss condition is much greater than the reaction to the gain
condition. For example, the regression performed on the population with income
between the 25th and 75th percentile (TPI level 2). The marginal effect of an additional dollar balance due is an additional $0.47 in underreported taxes discovered.
This model predicts that a taxpayer found in a balance due would not underreport
to completely remove their additional amount owed. This can be attributed to
the taxpayer rationalizing using game theoretic elements. It would be better to
underreport by a fraction and be required to pay back a smaller amount rather
than run the greater risk of an audit by underreporting until the amount owed is
zeroed. The marginal effect of an overpayment is a tax change of -0.006; the taxpayer would overreport his or her liability by less than a cent. In the taxpayer in a
refund position is not likely to have reporting noncompliance.
These parameter estimates further support prospect theory, in that an individual is much more risk seeking. The individual is willing to underreport their tax
liability to reduce a perceived loss when compared to the risk aversion to maintain
a perceived gain. If expected-utility theory held, the estimates for balance-due
and refund due would be identical or not statistically significant. The WLS regression using the 1988 TCMP dataset resulted in similar parameter estimates. At all
income groups a balance due prepayment position was found to increase noncompliance by around $0.30 for every dollar due. These results are in Appendix G.
While promising, the results from the analysis without interaction terms reported first section of Table 6 may be confounded with other causes of underreporting noncompliance. Table 7 extends the results of the second section of Table
6 and shows the marginal effects of the prepayment positions on tax reporting
compliance when accounting for interaction terms.36
The information in Table 7 also agrees with prospect theory after the interaction
terms are considered. Any interaction term that was found to be not significantly
different from zero was excluded in Table 7. The parameter estimates on the interaction terms report that among balance-due taxpayers, those with interest income
are less likely to underreport their tax liability. It appears that liquidity indeed
plays a role.
The presence of increased complexity and fungible income not subject to thirdparty reporting was captured by the presence of a Schedule C or Schedule F attached to the Form 1040. The inclusion of a Schedule C was found to increase
underreporting when interacted with the balance due variable and it was found
largely insignificant when interacted with the refund-due prepayment position.
24
Corcoro and Adelsheim
Unexpectedly, in the most general regressions, inclusion of a Schedule F decreased
underreporting. In those instances, the marginal effect was relatively small, and
in most of the income-group regressions, the interaction with Schedule F attachments were found not significantly different from zero.
TABLE 7. Net Misreporting Percentage by Income and Prepayment Position
Parameter Estimates—Prepayment and Interaction Terms
Balance Due
Full Sample
TPI Level 1
TPI Level 2
0.404
0.378
0.501
0.383
−0.225
—
−0.136
−0.211
x Sched C
0.218
0.230
0.193
0.188
x Sched F
−0.020
—
−0.050
—
x Age > 65
−0.110
−0.140
−0.175
−0.111
Full Sample
TPI Level 1
−0.052
−0.074
−0.074
x Interest
0.113
0.021
0.058
—
x Sched C
0.058
—
—
0.063
x Sched F
−0.012
—
—
—
x Age > 65
−0.095
0.047
—
−0.102
BD/RD
x Interest
TPI Level 3
Refund
BD/RD
TPI Level 2
TPI Level 3
−0.067
*All estimates from iteractions that were not statistically significant were excluded and can be found in Appendix F.
Cumulative Effect of Prepayment With Interaction Terms
Balance Due
Full Sample
TPI Level 1
TPI Level 2
TPI Level 3
−0.110
0.468
0.333
0.250
No interest
0.492
0.468
0.469
0.461
No sch C
0.050
0.238
0.141
0.062
No sch F
0.287
0.468
0.383
0.250
Age < 65
0.377
0.608
0.508
0.361
TPI Level 2
TPI Level 3
All interactions
Refund
Full Sample
TPI Level 1
All interactions
−0.095
−0.006
−0.016
−0.107
No interest
−0.101
−0.027
−0.074
−0.107
No sch C
−0.045
−0.006
−0.016
−0.169
No sch F
0.025
−0.006
−0.016
−0.107
Age < 65
0.108
−0.053
−0.016
−0.005
SOURCE: Raw 2001 NRP data.
List (2004) found that the behavior predicted by prospect theory dissipates
with experience in a given market. Thus we might anticipate the prepayment effect to dissipate due to taxpayer age. The interaction with the senior citizen dummy (along with an age by years) variable was included. The marginal effect of age
did not have a significant effect on underreporting compliance, but the parameter
A Balance Due Before Remittance
25
estimates were found to be mostly negative. Interacted terms with the age indicator variable predicted a decrease of around $0.10 in underreporting noncompliance. ‘Experience’ with tax filing did not fully explain the balance due noncompliance and the over 65 years indicator might be capturing other effects associated
with those types of taxpayers.
The cumulative effect of a prepayment position inclusive of all the interaction
terms is also reported in Table 7. With all the interaction terms included (along
with different combinations of interactions) the taxpayers still appear to be more
compliant if they have overpaid and more noncompliant if they have a balance
due. Once again, the results follow the predictions of prospect theory.
PREDICTED NONCOMPLIANCE ASSOCIATED WITH TIGTA REPORT
The 2009 TIGTA report has a series of illustrative examples on how the Making
Work Pay Credit and a taxpayer’s situation might result in the individual being
in the balance-due prepayment position. Some of the scenarios are duplicated
below. Following the examples, Table 8 estimates the underreported tax liability
predicted from the OLS regression.37
Example 1: A taxpayer is claimed by his or her parents and works for
the entire year during TY 2009. By the end of the year, this taxpayer
will have had $400 less withheld from his or her wages. Since he
or she is claimed as a dependent, this taxpayer is not eligible for
the Making Work Pay Credit and will therefore have to pay back
the $400 that he or she was advanced in the form of decreased
withholding during the year. If this taxpayer usually receives a
$200 refund, he or she will owe $200 when his or her TY 2009 tax
return is filed.
Example 2: An unmarried taxpayer has two jobs for all of Calendar
Year 2009. By the end of the year, this taxpayer will have received
$800 through reduced withholding. As a single filer, the taxpayer
is eligible for only $400 of the Making Work Pay Credit and will,
therefore, have to pay back the extra $400 that he or she was
advanced in the form of decreased withholding during the year. If
this taxpayer usually receives a $200 refund, he or she will owe $200
when his or her TY 2009 tax return is filed.
Example 3: A single taxpayer receives pension payments, receives
Social Security benefits, and is employed for the whole year during
Calendar Year 2009. This individual will receive $400 through his or
her pension, $400 through his or her wages, and $250 from the Social
26
Corcoro and Adelsheim
Security Administration. By the end of the year, this household will
have received an extra $1,050. As a single filer who is employed, this
individual is eligible for $400 and will, therefore, have to pay back
the extra $650 that he or she received during the year. This scenario
is exacerbated if taxpayers have more than one job.
TABLE 8. Predicted Effect of Balance Due on Underreported Income*
Example 1
Example 2
Example 3
Balance due
$400
$200
$650
Marginal effect
0.387
0.366
0.191
$154.80
$73.20
$124.15
Predicted underreporting
*Assumes different marginal effects based on demographics.
SOURCE: WLS regressions using 2001 NRP.
If taxpayers act as prospect theory predicts, then people presented with an unexpected balance due would underreport their tax liability.
Refinements
The previous analysis implicitly assumed that a taxpayer’s prepayment position
was not determined by the taxpayer. It assumed that the balance due or refund
due was caused by something external from the taxpayer. If the taxpayer’s prepayment position and the amount of underreported tax are determined by taxpayer’s
annual compliance behavior, then the estimates predicting how much prepayment
causes underreported taxes are biased and inconsistent.38 Due to this omitted
variable bias, the estimates predicting the effect of a balance-due prepayment position will be biased upwards, overstating the actual effect. Likewise, the effect of a
refund would be bias downwards, understating this prepayment position’s effect
on underreporting noncompliance.39 To address the issue of mutual causation
(endogeneity) of a taxpayer’s prepayment position and underreporting compliance, two refinements to the original model were performed. The first used the
instrumental variable (IV) estimation approach; the second restricted the dataset
to a subset of taxpayers who may have a more exogenous prepayment position.
IMPACT OF PREPAYMENT POSITION ON REPORTING COMPLIANCE—
IV ESTIMATION
Withholding and prepayment positions and changes in tax liability from previous years were used as instruments for the 2001 NRP year prepayment positions.
Hausman test statistics verified the need for IV estimation for the lower two TPI
levels. The results from the IV regressions are in Appendix H. Table 9 focuses
on the prepayment position variables. The sign of the IV parameters shows that
27
A Balance Due Before Remittance
the balance-due prepayment position increases underreporting and refund due
decreases underreporting. However, the influences for all prepayment parameter
estimates are not significantly different from zero.
TABLE 9. IV Prepayment Parameter Estimates
TPI Lev 1
Bal due
Refund
BD x interest
RD x interest
TPI Lev 2
TPI Lev 3
0.524
2.631
1.253
(1.96)
(3199.20)
(1.23)
−0.439
−3.335
−6.044
(1.27)
(3.88)
(17.51)
−0.468
−3.132
−1.268
(1.19)
(0.68)
(1.45)
0.444
3.465
6.143
(1.26)
(1048.80)
(17.36)
Standard errors in parentheses.
SOURCE: 2001 NRP.
Since the parameter estimates are not significantly different from zero, the IV
estimates imply that the timing of the payments does not matter in a taxpayer’s
underreporting compliance decisions. These results align with traditional economic theory.
As reported in Appendix H, many of the other predictor variables are found
to have little influence on underreporting. The DIF score is found to have a positive correlation with underreporting noncompliance for the TPI level 1 and TPI
level 3 groups, but not for TPI level 2 taxpayers that make up the interquartile
range (nearly half of the sample). With the high-income (TPI level 3 group), the
Hausman Test statistic could not reject the hypothesis that the WLS estimation
was efficient. However, the strength of the Hausman test is reliant on the strength
of the IV regression. This concern is discussed in more detail in the conclusions
and Appendix E.
IMPACT OF PREPAYMENT POSITION ON REPORTING COMPLIANCE—
SUBSAMPLE
Restricting the sample to taxpayers with consistent withholding implies that the
taxpayer expected business as usual from the previous years. The hope was that
this subgroup captured those who were not making their prepayment position
as a joint compliance decision with their reporting of tax liability, thus an exogenous taxpayer prepayment position. These results are reported in Table 10 and
in Appendix I. The results with the subset of data are similar to the prior WLS
analysis—a balance due prepayment position significantly contributes to greater
underreporting of taxes but a refund due has a much smaller absolute influence
on tax compliance.
28
Corcoro and Adelsheim
TABLE 10. NRP WLS Regression—Stable Withholding
Dependent Variable: Misreported Tax Liability
(underreport > 0, over report < 0)
Parameter
State income tax
Bal due
Refund
Full Sample
estimate
tpi_lev=1
estimate
tpi_lev=2
estimate
tpi_lev=3
estimate
−20.206
52.801
−59.054
−361.690
(65.278)
(48.147)
(44.861)
(553.374)
0.730**
(0.028)
1.465**
(0.088)
0.594**
(0.035)
0.709
(0.067)
−0.047
−0.048
−0.055
0.311
(0.031)
(0.049)
(0.026)
(0.426)
Standard errors in parentheses.
** p < 0.05; * p < 0.10.
SOURCE: 2001 NRP, stable withhold ≤ ±15% change in withholding.
Conclusions and Recommendations
Conclusions
TAXPAYERS’ REPORTING COMPLIANCE IS CORRELATED WITH
PREPAYMENT POSITION
The first estimation method by weighted least-squares assumed that the prepayment position was exogenously determined. This assumption could be valid in
cases where policy changes temporarily adjust the withholding tables, but ultimately the amount a taxpayer withholds and prepays is determined by the taxpayer. The subset model filtered the NRP data to taxpayers who most likely had
an exogenously determined prepayment position and the IV model used fitted estimates of the prepayment variables. Table 11 summarizes the analysis performed
to test which economic model aligns with taxpayer behavior. Different signs of the
balance-due and refund-due parameter estimates indicate that behavior is reference dependent. Coupled with oppositely signed estimates, a greater magnitude
of the balance-due parameter implies loss aversion, a perceived loss has a greater
impact in an individual when compared to an equivalent gain. Parameter estimates with opposite signs and different magnitudes support Prospect theory if, at
minimum, the balance due parameter is statistically significant.
Most of the results suggest that taxpayers react to a balance-due prepayment
position with an increase to underreporting noncompliance. To a lesser extent,
there appears to be a decrease in noncompliance when presented with a refund.
This is consistent with the behavior outlined by Prospect Theory: individuals behave differently depending on their perceived loss or gain in reference to a neutral
point; a perceived loss generates a stronger reaction than an equivalent gain.
29
A Balance Due Before Remittance
Among all the different regression analyses and robustness checks performed,
the results from the high income group (TPI level 3) were found to be the strongest. Balance due significantly contributed to underreporting and refunds reduced underreporting.
TABLE 11. Summary of Report Findings
WLS
Subset
IV
Different Signs
x
x
x
Loss aversion
x
x
Statistically significant BD
x
x
Statistically significant RD
x
SOURCE: NRP and TCMP regressions.
INSTRUMENTAL VARIABLES RESULT IN QUESTION
Whether or not the significant parameter estimates found in the WLS and subset
analysis was a statistical illusion has not been resolved. Due to endogeneity, these
results report upper-bound estimates of the effect of a balance-due payment and a
lower-bound estimates for a refund. The lack of statistically significant predictors
with the Instrumental Variable estimation is a concern. The presence of additional
Schedules and many of the other covariates that have historically been found to increase reporting noncompliance were found not to be statistically significant. The
DIF score used for audit selection was not found to be a good predictor of taxpayers within TPI level 2, the bulk of the taxpaying population. This can be attributed
to either a model mis-specification with poor instruments or improper weighting
by using the wrong tool in the statistical package. If the model is improperly specified, then an alternate model with different instruments must be considered. By
not properly accounting for the sample’s stratification, the statistical package is not
properly weighting the data; the parameter estimates are consistent but then the
standard errors and test statistics are incorrect and the computer may understate
the statistical significance of the estimates. Because the IV estimates are in question, Research cannot fully conclude that the results fully prove or disprove the
change in reporting behavior due to prepayment position, but there is an indication that there may be a causal relationship.
Recommendations
The results show, at minimum, that this relationship between a taxpayer’s prepayment position and underreporting tax compliance should be further examined.
Understanding the behavior of taxpayers increases the efficiency of future research
and targeted policy changes involving taxpayers’ withheld tax prepayments.
30
Corcoro and Adelsheim
INCREASED EDUCATION ON WITHHOLDING AND RELATED POLICIES
NOT CLASSIFYING RETURNS
Most of the results show that there are differences in reporting compliance dependent on the taxpayer’s prepayment position. The results should support preventative measures against unexpected prepayment positions. The biggest reasons
for underwithholding are life changes, and unexpected income that’s not fully
withheld, all coupled with lack of information. Better education on withholding and policy changes and wider access to the withholding calculator would
alert the taxpayer about the proper amount of income to prepay. Reducing the
number of balance-due taxpayers potentially reduces the amount of reporting
noncompliance.
These results—that prepayment position plays a role in underreporting compliance—should not be misconstrued into using prepayment position to classify returns. At the time of classification, only the reported values are known, and there
is a distinction between the reported and the as-corrected by audit prepayment
positions in NRP and TCMP.
ADDITIONAL RESEARCH NEEDED
Most of the results show that there are differences in reporting compliance dependent on the taxpayer’s prepayment position. This should not be a complete
recommendation in support of prospect theory. Even with random audit data
from the NRP and TCMP, the endogeneity of taxpayer’s withholding, thus prepayment position, is still an issue. The Instrumental Variable method was used in
an attempt to correct for this endogeneity issue. Other model specifications and
methods of analysis could test the robustness of the results. Some of these refined
models to test the interaction could be made with minimal additional data acquisition and manipulation.
ADDITIONAL RESEARCH—EXAMINE REFERENCE POINTS
This model assumed that the taxpayer expects that his or her remaining liability
after prepayments is at or near zero additional dollars. Schepanski and Shearer
(1995) noted that a taxpayer’s reference point is likely not zero liability but is likely close to the taxpayer’s expected prepayment position. If this is the reference
point, then analysis with as-corrected prepayment positions becomes less clear.
Taxpayers who expect a large refund (because of last year’s return) but actually
realize a small refund could consider themselves in the loss domain. Likewise,
those who expect to owe a large balance due, but realize a small balance due could
interpret this realization as a gain. Extending the data to include more prior years
could help this line of research. An expectation of prepayment position can be
A Balance Due Before Remittance
31
generated by averaging the taxpayer’s income-adjusted prepayment position from
previous years. The NRP data can easily be linked to add operational data from
the prior years not included in the NRP. Differencing the NRP-year prepayment
position from these estimated averages expresses the data in terms of this new
reference point.
ADDITIONAL RESEARCH—CHANGE IN REPORTED TAX LIABILITY
AFTER REFUNDABLE CREDITS
There were subtle differences when reporting compliance was determined by the
total tax via the actual line item and when it was determined by the total tax less
any refundable credits like the EIC. Total changes in reporting compliance should
also take into consideration these additional refundable credits. There can be
cases where after considering the additional credits a taxpayer’s tax liability can
change. Minor analysis in this report used this definition of underreporting. The
results were similar to the WLS results but there were no additional predictors
used to control for the EIC or additional refundable credits. Calculating tax liability after refundable credits is not a reported line item, but can be easily calculated
with the NRP data.
ADDITIONAL RESEARCH—ALTERNATE DISTRIBUTIONS OF
UNDERREPORTING
Underreported tax is likely skewed. The models used in this research assume normality in the dependent variable (and error terms). Least squares estimation can
be distorted due to outliers and thick distribution tails. The other model specifications accounting for the different distributions would lead to other parameter
estimates that might be more insightful. If a hurdle involving negative values can
be overcome, a log-linear model specification would have a log-normal distribution of underreporting. The resulting parameter estimates would reflect a taxpayer’s cross elasticity of prepayment and underreporting. Alternatively, it would
measure a proportional or percentage change in underreporting compliance in
response to changes in prepayment.
Acknowledgements
The views expressed are those of the author not the official positions of the Internal
Revenue Service. This report was prepared by the SB/SE Seattle/San Jose Research
Office under the direction of Peter Adelsheim, Senior Research Analyst SB/SE
Research Seattle/San Jose. The project was conducted by Paul Corcoro, Operations
Research Analyst. The following should be acknowledged: Alan Plumley, Drew
32
Corcoro and Adelsheim
Johns, RAS; John DeWald, M.C. Fusco, Tani Hunt, Sandra Miller, Anne Parker,
Sarah Shipley, Alex Turk, James Zanetti, SB/SE Research; Elaina Rose, Lan Shi,
Hendrik Wolff, University of Washington, for their support, helpful comments
and suggestions.
Appendices
Appendix A—References
Adelsheim, Peter D., “Prepayment Position and Income Tax Noncompliance,”
Pacific—Northwest DORA Research Report, January 1997.
Angrist J.D., Imbens G.W., Rubin, D.B., “Identification of Causal Effects Using
Instrumental Variables” Journal of the American Statistical Association, 1996.
Behavioural Finance, “Prospect Theory” http://prospect-theory.behaviouralfinance.net
Chang, Otto H., Joseph J. Shultz, Jr., “The Income Tax Withholding
Phenomenon: Evidence From TCMP Data,” JATA, Vol. 12, No. 1, Fall 1990.
Christian, Charles, et. al., “The Relation Between the Use of Tax Preparers and
Taxpayers’ Prepayment Position,” JATA, Vol. 16., No. 1, 1994.
Clotfelter, Charles, “Tax Evasion and Tax Rates: An Analysis of Individual
Returns,” Review of Economics and Statistics, Vol. 63, No. 3, 1983.
Fudenberg, Drew, “Advancing Beyond Advances in Behavioral Economics,”
Journal of Economic Literature, Vol. XLIV, September 2006.
Greene, William H., “Econometric Analysis 5th Ed.,” Prentice Hall 2003.
Gronau, Reuben, “Sex-related Wage Differentials and Women’s Interrupted Labor
Careers—the Chicken or the Egg,” Journal of Labor Economics, Vol. 6, No. 3,
1988.
Hershey Jr., Robert, “The Federal Budget: Taxes; Adjustment Brings In Cash and
Risk.” The New York Times, Jan. 30, 1992.
Heukelom, Floris, “Kahneman and Tversky and the Origin of Behavioral
Economics,” Tinbergen Institute Discussion Paper, September 2006.
Ho, Chih-Chin, ”Predicting Reporting Accuracy of Individual Tax Returns,”
Washington D.C., Internal Revenue Service, 2003.
Kahneman, Daniel, and Amos Tversky, “ Prospect Theory: An Analysis of
Decision Under Risk,”Econometrica, Vol. 47, No. 2, March 1979.
Kleven, H.J., M.B. Knudsen, C.T. Kreiner, S. Pedersen, E. Saez, “Unwilling or
Unable to Cheat? Evidence from a Randomized Tax Audit Experiment in
Denmark,” NBER Working Paper, No. 15769, February 2010.
A Balance Due Before Remittance
33
List, John A., “Notes and Comments—Neoclassical Theory Versus Prospect
Theory: Evidence from the Marketplace.” Econometrica, Vol. 72, March 2004.
Pope, Devin G. and Maurice E. Scweitzer, “Is Tiger Woods Loss Averse?—
Persistent Bias in the Face of Experience, Competition and High Stakes,”
Working Paper, The Wharton School, June 2009.
SB/SE Research Seattle/San Jose, “Tax Compliance among Varying Levels of
Income Visibility,” Project SEA0033, June 2007.
Schepanski, A., D. Kelsey, “Testing for Framing Effects in Taxpayer Compliance
Decisions,” JATA Fall, Vol. 12, No. 1, 1990.
Schepanski, A. and T. Shearer, 1995. A Prospect Theory Account of the Income
Tax Withholding Phenomenon. Organizational Behavior and Human Decision
Processes.
Shapiro, Matthew D. and Joel Slemrod, “Consumer Response to the Timing of
Income: Evidence from a Change in Tax Withholding,” American Economic
Review, March 1995.
Slemrod, Joel, “An Empirical Test for Tax Evasion,” Review of Economics and
Statistics, Vol. 67, No. 2, 1985.
Treasury Inspector General For Tax Administration, “Millions of Taxpayers
May Be Negatively Affected by the Reduced Withholding Associated With the
Making Work Pay Credit,” Nov. 2009, Reference Number: 2010–41–002.
Tversky, Amos, Daniel Kahneman, “Rational Choice and the Framing of
Decisions,” Journal of Business, Vol. 59, No. 4, 1986.
White, Richard, et al., “The Impact of Income Tax Withholding on Taxpayer
Compliance: Further Empirical Evidence,” JATA 1993.
34
Corcoro and Adelsheim
Appendix B —Unweighted Histograms of Change in
Reported Liability
NRP TPI Level 1—Percentiles: [0, 25th)
NRP TPI Level 2—Percentiles: [25th, 75th]
A Balance Due Before Remittance
NRP TPI Level 3—Percentiles: (75th, 99th] [1]
[1] Removed large outlier.
TCMP TPI Level 1—Percentiles: [0, 25th)
35
36
Corcoro and Adelsheim
TCMP TPI Level 2—Percentiles: [25th, 75th]
TCMP TPI Level 3—Percentiles: (75th, 99th]
37
A Balance Due Before Remittance
Appendix C—Discussion on Risk and Utility
A brief discussion on risk and utility is needed to highlight the theoretical differences between work based on the standard utility model (Allingham and Sando),
and work based on Prospect theory. An extensive discussion on utility can be
found in any intermediate economics textbook. Economics gives a framework
to model an individual’s preferences—a utility function. From its background in
philosophy, utility quantifies the level of happiness or satisfaction from different
levels of a particular good; for the following examples, money income is the good.
Risk aversion is equivalent to having a concave utility function. This individual
gets greater satisfaction with an increase in income (u' > 0) but at a diminishing
rate (u" < 0); an individual is happier with more money, but at a decreasing rate.
A person is said to be risk averse if the person prefers a certain prospect over
any risky prospect of equivalent expected value. The graph in Figure 2 exhibits
risk aversion. The solid curve represents an individual’s utility function, or his
or her level of happiness given a level of income. The dashed line represents a
fair gamble of a 50-percent chance of an income of $200 (point a) and 50-percent
chance of $1,000 (point c). The expected value of this fair gamble is $600 (point
d). Contrasting the points b and d illustrates that this risk averse person would
not take the gamble.
FIGURE C1. Utility Curve with Risk Aversion
11.25
Utility
10.25
c
b
9.25
d
8.25
a
7.25
6.25
0
200
400
600
800
1000
1200
Income
utility
risky gamble
1400
1600
38
Corcoro and Adelsheim
Point b illustrates what is called the certainty equivalent to the expected payoff
from the risky gamble (point d). By the curvature of the utility function, the graph
shows that this individual derives greater satisfaction from $600 for certain, versus
the gamble between $200 and $1000, thus is averse to risk.
Expected utility theory itself does not presuppose risk aversion. If an individual
has a linear utility function they are said to be risk neutral. This is the case where
the individual is indifferent between the risky gamble and the certainty equivalent.
The other case is of risk seeking; here the individual has a convex utility function
(u' > 0, u" > 0) with a preference for the risky gamble over the certainty equivalent.
Kahneman and Tversky (1979) created prospect theory to explain the behavioral shifts from risk averse to risk seeking observed in their experiments that are
not addressed in expected utility theory. Prospect theory differentiates itself from
expected utility theory by having a value function dependent to a neutral reference
point rather than a utility function dependent on the individual’s final assets. The
curvature of the value function is shaped such that individuals are risk seekers in
loss domains and risk averse in gain domains and the value function is steeper for
losses than for gains. In addition, the shape also implies that a loss has a greater
impact in an individual when compared to an equivalent gain.
FIGURE C2. Prospect Theory Value Function
10
5
0
Value
-300
-200
-100
0
-5
-10
-15
-20
Loss / Gain in Income
100
200
300
39
A Balance Due Before Remittance
Appendix D —Omitted Variable Bias
Whether a taxpayer underreports on his or her taxes and their prepayment position is ultimately determined by the taxpayer; because of this, the two variables
can be considered endogenously determined by taxpayer and both influenced by
the taxpayer’s noncompliant behavior. The varying compliance behavior is not observable in the data and thus cannot be controlled for (random effect). This creates
a potential bias in the parameter estimates of the prepayment position variables.
Suppose the true real-world model for underreported taxes is as follows:1
U i = βX i + γN i + ε i'
Where U is the amount of underreported taxes
X are variables of interest (i.e. a taxpayer’s prepayment position)
N represents the taxpayer’s level of tax noncompliance behavior,
ε i' is the random error term of the model which can include mistakes due to com-
plexities of the tax code,
and
β , γ are parameters of interest.
With the data, Research may be able to estimate the following (suppressing the
subscripts for clarity):
U = βX + ε
This is due to the fact that a taxpayer’s level of noncompliance is hard to estimate
and would reside in this model’s error term
ε = γN + ε '
)
β = (X ′ X
(
= ∑X2
1 This
) (X ′U )
−1
) (∑ (X (βX
−1
standard OLS estimate of the parameter of interest
+ γN + ε '
)))
substituting the true model
into the equation
simplified model suppresses all the other covariates, the intercept terms and the distinction of the two prepayment positions. The usual intercept term can be suppressed by taking the difference from the mean for all the variables
which allows the constant intercept terms to cancel out. This is done for clarity on how the bias can arise. These
econometric results still hold in the estimated model. See Greene pp. 76 for more details.
40
Corcoro and Adelsheim
(
= ¦;
) (β ¦ ;
−
(β ∑ X
2
+ γ ¦ ;1 + ¦ ;ε
)
+ γ ∑ X N + ∑ Xε '
=
2
X
∑
n
)
n
istributing and
d
combining terms
=u
sing algebra to set up variances
and covariances.
Taking the probability limit of the parameter estimate gives the following:
plim β =
βσ ; + γσ ;1 + σ ;ε
σ ;
We assumed that after accounting for N the remaining regressors are uncorrelated with the true random error error ( σ ' = 0). After simplification we get that
Xε
the probability limit of a parameter estimate for the effect of prepayment position
on underreporting is the following.
plim β =
β+
γσ ;1
σ ;
Given that a taxpayer’s noncompliance behavior is unaccounted for, this estimate
will not tend to the true value as the sample size increases, but will be bias and
inconsistent.
Assumptions need to be made in order to make inferences about the param2
eters to be estimated. The variance ( σ X ) is always a positive number. Research
assumes that the amount a taxpayer underreports is positively correlated with a
taxpayer’s noncompliance decision thus γ > 0 .2 The sign for the covariance between prepayment position and non-compliance (σ ;1 ) needs to be determined.
Recall that a balance due prepayment position is coded in positive values and refund due in negative values. Using prospect theory as a guide the correlation
between prepayment position and taxpayer’s compliance decision would depend
on whether the taxpayer is balance-due or refund due. If the taxpayer has a balance due then the correlation would be positive (σ ;1 > 0).3 The taxpayer views
2 Signs
could change depending on whether Research focuses on compliance versus non-compliance.
signs would be different if reported prepayment position was used rather than actual prepayment position. A
noncompliant taxpayer would likely report a larger refund rather than not.
3 The
A Balance Due Before Remittance
41
him- or herself in the loss condition thus more willing to be noncompliant and
risk an audit. The opposite is true with a refund ( σ ;1 < 0), a taxpayer views himor herself in a gain condition and less willing to risk losing the refund due to an
audit.
With these correlations, if noncompliance is not properly accounted for, then
the model would report an upward bias effect of a balance due position on underreported taxes and a downward bias of a refund position on underreporting.
To account for this, an instrumental variable (IV, two-staged least squares, 2SLS)
approach was used to create a fitted measure of prepayment position. Appendix
E provides a detailed description of the IV approach used to estimate for this
research.
Appendix E—Instrumental Variable Methodology and
Results
(2) First stage: PP = δ0 + Z1 δ 1 + PPt-1 δ 2 + PPt-2 δ 3 + εpp
ˆ + β RDˆ +interaction terms+ ε
(3) Second stage: U = β0 + X1 β1 + β2 BD
3
u
Critical to the OLS model is the assumption that the predictor variables are uncorrelated with the model’s error term (unaccounted or immeasurable variables).
However, there is a problem. A taxpayer’s specific compliance disposition may
help to explain both their prepayment position and how much they are likely to
underreport his or her tax liability. These varying dispositions are not observable
in the data and thus cannot be controlled for (random effect).
It is plausible that prepayment positions are endogenous since the amount to
withhold throughout the year is determined by the taxpayer. This implies that the
variables BD and RD are correlated with random error (εu). The assumptions of
them being independent in the WLS specification are invalid. This creates issues
of bias and inconsistency in the parameter estimates.
Given that prepayment position is endogenous, Research used an Instrumental
Variable (IV) approach to model the change in reported and actual tax liability.
IV, first involves modeling the taxpayer’s prepayment position as a function of
exogenous or predetermined variables
The Hausman test was used to determine the severity of the endogeneity problem with the prepayment position. In general, this specification test involved the
estimated covariance matrices from the WLS and IV estimations under the hypothesis that both are consistent.
42
Corcoro and Adelsheim
If Research could not reject the null hypothesis associated with the Hausman
test, then the endogeneity problem associated with prepayment position was not
severe and Research could proceed with WLS. If Research rejected the null hypothesis, then there was an endogeneity problem and the IV estimation would be
used.4
Table E1 reports the Hausman Test statistics. In cases where the TPI is below
the 75th percentile, the null hypothesis that the WLS estimates are efficient is rejected. These results support the belief that there is an endogeneity issue with
prepayment position. In the high income group (TPI level 3), the WLS estimates
appear to be efficient.
TABLE E1. Hausman’s Specification Test Results
Comparing WLS to 2SLS
Ho:WLS and IV consistent, WLS efficient
Ha: IV consistent and efficient
TPI Level 1
DF
Statistic
Pr > ChiSq
TPI Level 2
TPI Level 3
31
31
28
417.7
289.8
16.04
<.0001
<.0001
0.9652
SOURCE: WLS and IV estimations using NRP data.
4 Caveat: the Hausman test assumes that the 2SLS approach is consistent. If the instruments are weak or are correlated
with the error term, then 2SLS is inconsistent as well. These issues of proper instruments were brought up in Klevin
et al. (2009) and Andreoni, Erard, and Feinstein (1998).
-
-
-
-
-
-
-
Full Sample
0.158945
41417
Estimate
-147.709
(74.080)
0.162
(0.002)
0.032
(0.002)
-304.656
(208.849)
-
**
**
**
Intercept terms are supressed, standard errors in parentheses.
** p < 0.05; * p < 0.10
R-Square
N
Parameter
State Income Tax
Income Tax
Bal Due
Bal Due
Refund
Refund
No Prepay Pos
No Prepay Pos
BD x Interest
BD x Interest
RD x Interest
RD x Interest
BD x Sched C
BD x Sched C
RD x Sched C
RD x Sched C
BD x Sched F
BD x Sched F
RD x Sched F
RD x Sched F
BD x Age > 65
BD x Age > 65
BD x Age > 65
Dependent Variable: Understatement of Tax Liability
No Interaction Terms
With Interaction Terms
TPI Level 1
TPI Level 2
TPI Level 3
Full Sample
TPI Level 1
TPI Level 2
0.298356
0.437014
0.170145
0.223177
0.339663
0.471652
10503
20902
10478
41417
10503
20902
Estimate
Estimate
Estimate
Estimate
Estimate
Estimate
**
**
**
**
40.129
17.959
-1971.662
-159.196
35.116
12.665
(13.903)
(27.705)
(553.436)
(71.223)
(13.502)
(26.858)
**
**
**
**
**
0.287
0.465
0.151
0.404
0.378
0.501
(0.006)
(0.004)
(0.004)
(0.011)
(0.022)
(0.013)
**
**
*
**
-0.066
-0.006
0.030
-0.052
-0.074
-0.074
(0.004)
(0.005)
(0.004)
(0.030)
(0.020)
(0.016)
**
*
**
-70.534
-110.639
-2588.335
-340.829
-64.225
-300.256
(28.699)
(191.140)
(7553.931)
(200.949)
(28.138)
(185.499)
**
-0.225
-0.007
-0.136
(0.010)
(0.013)
(0.010)
**
**
0.113
0.021
0.058
(0.030)
(0.009)
(0.014)
**
**
0.218
0.230
0.193
(0.005)
(0.014)
(0.009)
**
0.058
-0.004
0.004
(0.006)
(0.014)
(0.011)
**
-0.020
0.029
-0.050
(0.005)
(0.053)
(0.017)
*
-0.012
-0.076
-0.036
(0.007)
(0.054)
(0.032)
**
**
-0.110
-0.140
-0.175
(0.005)
(0.015)
(0.012)
**
**
-0.095
0.047
0.007
(0.006)
(0.014)
(0.010)
Table F1. WLS Regression 2001 NRP
**
**
**
**
**
**
**
TPI Level 3
0.219011
10478
Estimate
-2045.001
(537.365)
0.383
(0.024)
-0.067
(0.157)
-942.581
(7332.515)
-0.211
(0.022)
0.133
(0.157)
0.188
(0.011)
0.063
(0.012)
-0.015
(0.010)
-0.005
(0.015)
-0.111
(0.010)
-0.102
(0.013)
**
**
**
**
**
**
**
A Balance Due Before Remittance
43
Appendix F—NRP Weighted Least Squares (WLS)
Parameter Estimations
R_65
OVR_65
meter
Full Sample
0.158945
41417
Estimate
288.447
(67.359)
-86.632
(244.072)
55.917
(92.888)
585.931
(87.620)
-178.825
(1253.029)
548.453
(179.739)
233.438
(1166.139)
-33.463
(67.310)
452.930
(103.869)
-171.858
(135.861)
-151.191
(5.825)
1.042
(2.664)
1.543
(0.270)
**
**
**
**
**
**
* p < 0.05; ** p < 0.10
Intercept terms are supressed, standard errors in parentheses.
Parameter
Indicator - Sched A
Indicator - Sched A
Indicator - Sched C
Indicator - Sched C
Indicator - Sched D
Indicator - Sched D
Indicator - Sched E
Indicator - Sched E
Indicator - Sched F
Indicator - Sched F
Num Sched C
Num Sched C
Num Sched F
Num Sched F
Indicator - Interest Income
Indicator - Interest Income
Indicator - Other Income
Indicator - Other Income
Indicator for Age > 65
Indicator for Age > 65
Relative TPI
Relative TPI
Primary Age
Primary Age
DIF Score
DIF Score
R-Square
N
Dependent Variable: Understatement of Tax Liability
No Interaction Terms
With Interaction Terms
TPI Level 1
TPI Level 2
TPI Level 3
Full Sample
TPI Level 1
TPI Level 2
0.298356
0.437014
0.170145
0.223177
0.339663
0.471652
10503
20902
10478
41417
10503
20902
Estimate
Estimate
Estimate
Estimate
Estimate
Estimate
36.381 *
158.501 **
-1716.267 **
255.289 **
26.290
159.809
(18.622)
(22.704)
(795.244)
(64.894)
(18.178)
(22.057)
**
**
*
**
260.036
271.411
-1630.748
-389.183
134.800
111.663
(60.952)
(82.213)
(1260.053)
(234.883)
(60.827)
(81.803)
-48.004 **
-69.599 **
-310.889
45.204
-42.397 *
-72.651
(22.761)
(31.102)
(505.121)
(89.349)
(22.157)
(30.150)
**
**
25.465
27.039
1318.747
505.100
18.664
84.222
(23.167)
(29.516)
(439.650)
(84.350)
(22.509)
(28.692)
34.521
-283.781
-1395.734
-368.167
78.498
-138.739
(392.825)
(414.545)
(5018.637)
(1204.450)
(384.568)
(404.114)
**
*
*
**
-95.266
113.045
1388.230
166.512
-91.284
46.778
(47.927)
(59.967)
(831.385)
(172.948)
(46.522)
(58.155)
37.808
216.296
510.643
536.790
37.001
274.249
(368.933)
(385.394)
(4601.631)
(1121.167)
(358.450)
(373.712)
**
**
**
**
-25.102
-156.825
-2923.862
-68.169
-44.724
-130.378
(12.502)
(25.182)
(958.691)
(74.160)
(15.220)
(33.138)
**
**
*
**
**
72.979
181.030
1009.333
422.362
86.994
227.908
(24.175)
(35.713)
(531.266)
(99.853)
(23.559)
(34.654)
29.956
21.284
845.300
114.245
35.819
218.804
(24.699)
(57.947)
(1189.555)
(131.025)
(25.978)
(59.505)
521.296 **
-55.642 **
-158.162 **
-133.432 **
505.110 **
-15.893
(32.258)
(21.246)
(11.411)
(6.164)
(31.407)
(20.701)
0.140
-0.254
-38.465
-1.705
0.001
-0.697
(0.481)
(1.049)
(23.568)
(2.562)
(0.468)
(1.020)
0.572 **
0.876 **
16.370 **
1.443 **
0.548 **
0.904
(0.065)
(0.092)
(2.829)
(0.260)
(0.063)
(0.090)
Table F1. WLS Regression 2001 NRP Continued
**
**
**
**
**
**
**
TPI Level 3
0.219011
10478
Estimate
-639.102
(773.422)
-3447.102
(1229.098)
-287.425
(490.482)
1199.182
(427.443)
-1633.645
(4872.764)
550.009
(808.280)
1187.281
(4469.211)
-744.484
(1038.964)
968.908
(515.924)
3032.248
(1169.715)
-144.795
(12.241)
-41.352
(22.886)
13.222
(2.778)
44
Corcoro and Adelsheim
45
A Balance Due Before Remittance
Appendix G—TCMP WLS Parameter Estimates
Table G1. TCMP WLS Regressions
(underreport > 0, over report < 0)
R-Sq
N
parameter
State Income Tax
Income Tax
Bal Due
Bal Due
Refund
Refund
Indicator - Interest Income
Indicator - Interest Income
BD x Interest
BD x Interest
RD x Interest
RD x Interest
BD x Sched C
BD x Sched C
RD x Sched C
RD x Sched C
BD x Sched F
BD x Sched F
RD x Sched F
RD x Sched F
Indicator - Sched C
Indicator - Sched C
Indicator - Sched D
Indicator - Sched D
Indicator - Sched E
Indicator - Sched E
Indicator - Sched F
Indicator - Sched F
Indicator - Dividend Income
Indicator - Dividend Income
Indicator - Alimony Income
Indicator - Alimony Income
Indicator - Capital Gain Income
Indicator - Capital Gain Income
Indicator - Other Income
Indicator - Other Income
Indicator - IRA Income
Indicator - IRA Income
Indicator - Pension Income
Indicator - Pension Income
Indicator - Unemp Income
Indicator - Unemp Income
Indicator - Soc Security Income
Indicator - Soc Security Income
Indicator - Other Gain Income
Indicator - Other Gain Income
Relative TPI
Relative TPI
DIF Score
DIF Score
full sample
0.119114
54088
tpi_lev=1
0.38378
13522
estimates
estimates
-24.923987
(30.649)
0.3639
(0.014)
-0.107112
(0.033)
56.67273
(27.735)
-0.31988
(0.014)
0.090321
(0.033)
0.062502
(0.002)
0.085694
(0.006)
-0.076865
(0.002)
-0.059951
(0.014)
50.168572
(37.209)
-140.61411
(353.814)
133.343963
(31.290)
291.711844
(73.732)
-53.601178
(25.599)
-57.772068
(154.921)
271.088234
(38.667)
225.784224
(34.905)
64.621606
(54.029)
-5.838312
(27.999)
-0.742442
(36.988)
-45.468038
(53.274)
258.774994
(94.411)
-14.681065
(2.335)
0.969843
(0.064)
-11.91629
(6.619)
0.342891
(0.012)
-0.158673
(0.008)
-24.738222
(5.962)
-0.249134
(0.012)
0.045317
(0.010)
0.2029
(0.012)
0.067042
(0.014)
0.125294
(0.028)
0.018134
(0.038)
-42.152099
(13.012)
-42.943226
(143.816)
23.515383
(9.418)
-29.034204
(27.971)
-15.750057
(6.899)
23.287686
(32.864)
-13.143108
(12.202)
85.678201
(9.616)
-6.93238
(14.684)
-7.087528
(6.882)
41.184614
(8.328)
247.60049
(61.245)
62.274545
(32.919)
115.897938
(12.989)
0.530928
(0.016)
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
**
SOURCE: 1988 TCMP.
Intercept terms supressed, standard errors in parentheses.
** p < 0.05; * p < 0.10
tpi_lev=2
0.420734
27044
*
**
**
**
**
**
**
**
**
**
**
**
**
**
**
*
**
**
tpi_lev=3
0.152767
13522
estimates
estimates
4.9213981
(15.047)
0.2889049
(0.010)
-0.0579062
(0.016)
4.919593
(17.761)
-0.0924134
(0.009)
0.0079406
(0.016)
0.2518219
(0.007)
0.0484029
(0.010)
-0.0474706
(0.010)
-0.0369991
(0.019)
-92.491659
(17.875)
212.013634
(154.720)
-6.1561525
(13.378)
104.036927
(35.371)
-58.264569
(10.953)
-1.3602876
(76.212)
3.4793031
(16.855)
102.701713
(15.353)
45.8155355
(22.685)
-14.863958
(12.748)
28.3615722
(16.486)
-166.21938
(19.945)
33.486513
(41.508)
129.573124
(13.848)
0.7623807
(0.038)
-247.51298
(224.298)
0.34388
(0.038)
-0.10939
(0.276)
555.66327
(625.834)
-0.30779
(0.038)
0.09183
(0.276)
0.05922
(0.005)
0.08299
(0.012)
-0.06997
(0.005)
-0.07596
(0.028)
46.10944
(196.172)
-926.36842
(1139.952)
64.52353
(155.286)
1057.65052
(388.198)
-511.13955
(145.945)
-436.35422
(1173.665)
283.25353
(156.356)
321.86276
(163.827)
130.44194
(305.235)
-29.5074
(167.532)
312.10128
(449.298)
176.37579
(227.242)
250.93513
(331.820)
-16.23099
(4.390)
2.98313
(0.581)
**
**
**
**
**
**
*
**
**
**
**
**
*
**
**
**
**
**
**
**
**
**
**
**
*
**
**
**
46
Corcoro and Adelsheim
Table G1. TCMP WLS Regressions
(underreport > 0, over report < 0)
full sample
0.119114
54088
tpi_lev=1
0.38378
13522
parameter
estimates
estimates
estimates
estimates
Indicator - Alimony Income
-57.772068
23.287686
-1.3602876
-436.35422
R-Sq
N
Indicator - Alimony Income
Indicator - Capital Gain Income
Indicator - Capital Gain Income
Indicator - Other Income
Indicator - Other Income
Indicator - IRA Income
Indicator - IRA Income
(154.921)
271.088234
tpi_lev=3
0.152767
13522
(32.864)
(76.212)
(1173.665)
**
-13.143108
3.4793031
283.25353
**
85.678201
(38.667)
225.784224
tpi_lev=2
0.420734
27044
(12.202)
(16.855)
**
102.701713
(34.905)
(9.616)
(15.353)
64.621606
-6.93238
45.8155355
(156.356)
**
321.86276
**
130.44194
(54.029)
(14.684)
(22.685)
-5.838312
-7.087528
-14.863958
-29.5074
Indicator - Pension Income
(27.999)
(6.882)
(12.748)
(167.532)
Indicator - Unemp Income
-0.742442
41.184614
Indicator - Unemp Income
(36.988)
(8.328)
-45.468038
247.60049
Indicator - Soc Security Income
Indicator - Soc Security Income
Indicator - Other Gain Income
Indicator - Other Gain Income
Relative TPI
Relative TPI
(53.274)
258.774994
DIF Score
(0.064)
** p < 0.05; * p < 0.10
115.897938
**
0.530928
312.10128
**
176.37579
(449.298)
-166.21938
(19.945)
(227.242)
*
33.486513
250.93513
(41.508)
**
129.573124
**
0.7623807
(12.989)
(0.016)
*
**
(32.919)
(2.335)
0.969843
Standard errors in parentheses.
**
(94.411)
-14.681065
DIF Score
SOURCE: 1988 TCMP
62.274545
(305.235)
(16.486)
(61.245)
**
28.3615722
(331.820)
**
-16.23099
**
2.98313
(13.848)
(0.038)
**
(163.827)
Indicator - Pension Income
**
*
(4.390)
(0.581)
**
**
47
A Balance Due Before Remittance
Appendix H—Instrumental Variables (IV 2SLS)
Parameter Estimations
Table H1. IV Estimation—First Stage Results
Dependent Variable: Balance Due/Refund Due Amount
TPI Level 1
R-Sq
N
parameter
intercept
intercept
1999 Prepay Pos
1999 Prepay Pos
2000 Prepay Pos
2000 Prepay Pos
Change in Tax '99 - '00
Change in Tax '99 - '00
Change in Tax '00 - '01
Change in Tax '00 - '01
State Income Tax
Income Tax
Indicator - 20% tax change
Indicator - 20% tax change
Indicator - stable withholding
Indicator - stable withholding
No Prepay Pos
No Prepay Pos
Indicator - Sched A
Indicator - Sched A
Indicator - Sched C
Indicator - Sched C
Indicator - Sched D
Indicator - Sched D
Indicator - Sched E
Indicator - Sched E
Indicator - Sched F
Indicator - Sched F
Num Sched C
Num Sched C
Num Sched F
Num Sched F
SOURCE: 2001 NRP.
Standard errors in parentheses.
** p < 0.05; * p < 0.10
TPI Level 2
0.3245
10503
0.3352
20902
Estimate
-699.009
(84.04)
0.180
(0.01)
0.271
(0.01)
0.037
(0.00)
0.248
(0.01)
13.518
(37.86)
-319.232
(40.05)
-242.785
(47.75)
357.295
(76.96)
-38.508
(50.08)
900.995
(153.90)
-112.076
(61.28)
8.629
(62.42)
675.260
(996.10)
-152.314
(134.10)
-393 284
-393.284
(963.10)
TPI Level 3
0.439
10478
Estimate
**
**
**
**
**
**
**
**
**
*
-1,630.390
(45.47)
0.111
(0.01)
0.353
(0.00)
0.051
(0.00)
0.330
(59.09)
43.274
(125.80)
65.344
(59.04)
-127.891
(404.70)
656.202
(48.21)
-482.385
(156.10)
1,129.881
(66.08)
-354.143
(62.63)
290.440
(841.40)
1,252.561
(128.10)
544.584
(810.60)
-694 780
-694.780
(53.92)
Estimate
**
**
**
**
**
**
**
**
**
**
**
5,333.142
(5519.60)
0.246
(0.01)
0.359
(0.01)
0.041
(0.01)
0.218
(0.00)
-1,353.740
(1757.40)
3,707.037
(1364.40)
-952.550
(2166.40)
-3,915.020
(23900.70)
-10,026.700
(2516.70)
778.027
(3451.90)
146.857
(1610.40)
2,512.339
(1400.60)
3,369.736
(15560.60)
2,095.834
(2664.40)
482 078
482.078
(14568.90)
**
**
**
**
**
**
*
48
Corcoro and Adelsheim
Table H1. IV Estimation—First Stage Results Continued
Dependent Variable: Balance Due/Refund Due Amount
TPI Level 1
R-Sq
N
parameter
Indicator - Interest Income
Indicator - Interest Income
Indicator for Age > 65
Indicator for Age > 65
Indicator - Dividend Income
Indicator - Dividend Income
Indicator - Alimony Income
Indicator - Alimony Income
Indicator - Capital Gain Income
Indicator - Capital Gain Income
Indicator - Other Gain Income
Indicator - Other Gain Income
Indicator - IRA Income
Indicator - IRA Income
Indicator - Pension Income
Indicator - Pension Income
Indicator - Unemp Income
Indicator - Unemp Income
Indicator - Soc Security Income
Indicator - Soc Security Income
Indicator - Other Income
Indicator - Other Income
Relative TPI
Relative TPI
Primary Age
Primary Age
Indicator - Dependent Status
Indicator - Dependent Status
DIF Score
DIF Score
SOURCE: 2001 NRP.
Standard errors in parentheses.
** p < 0.05; * p < 0.10
TPI Level 2
0.3245
10503
0.3352
20902
Estimate
42.943
(33.91)
-229.391
(66.57)
-9.729
(50.01)
-68.153
(389.90)
1.770
(64.86)
406.913
(244.80)
292.223
(223.90)
-18.657
(85.01)
136.243
(50.49)
343.574
(64.31)
375.227
(64.97)
-727.599
(86.95)
9.743
(1.30)
349.981
(56.13)
2.123
(0.17)
TPI Level 3
0.439
10478
Estimate
**
*
**
**
**
**
**
**
**
-157.626
(123.20)
-524.467
(55.06)
-263.839
(385.80)
-500.321
(72.03)
333.576
(234.90)
1,148.922
(169.80)
265.707
(76.26)
4.338
(79.61)
160.285
(109.60)
-20.512
(76.39)
660.904
(44.59)
251.877
(2.24)
26.956
(585.20)
-730.385
(0.20)
-0.690
0.00
Estimate
**
**
**
**
**
**
**
**
**
**
-4,116.810
(3079.30)
385.830
(3801.70)
-1,394.530
(1632.40)
8,326.480
(16795.30)
4,207.996
(1482.70)
3,907.094
(3734.30)
-2,902.470
(3161.20)
80.258
(1669.90)
362.867
(3159.30)
-3,853.180
(3556.60)
3,350.963
(1697.20)
-1,220.390
(41.21)
54.328
(75.18)
57,556.060
(15759.20)
40.334
(9.11)
**
**
**
49
A Balance Due Before Remittance
Table H2. IV Estimation—Second Stage Results
Dependent Variable: Change in Tax Liability
(Underreport > 0, Overreport < 0)
TPI Level 1
N
parameter
TPI Level 2
10503
20902
Estimate
Estimate
TPI Level 3
10478
Estimate
Intercept
-297.847
-1,813.790
-17,243.900
Intercept
(355.70)
(392.00)
(39004.30)
-16.582
-2,443.130
(2.06)
(1214.00)
Income Tax
31.574
Income Tax
(18.50)
*
Bal Due
0.524
2.631
1.253
Bal Due
(1.96)
(3199.20)
(1.23)
Refund
-0.439
-3.335
-6.044
Refund
(1.27)
(3.88)
(17.51)
Indicator - 20% tax change
-6.993
337.238
1,417.561
Indicator - 20% tax change
(529.30)
(80.36)
(966.50)
Indicator - stable withholding
15.017
-46.803
-444.788
Indicator - stable withholding
(87.44)
(2.20)
(1506.00)
BD x 20% tax change
-0.160
0.322
-0.098
BD x 20% tax change
(1.13)
(4.01)
(0.07)
BD x Interest
-0.468
-3.132
-1.268
BD x Interest
(1.19)
(0.68)
(1.45)
RD x Interest
0.444
3.465
6.143
RD x Interest
(1.26)
(1048.80)
(17.36)
No Prepay Pos
-15.981
-380.719
-6,162.040
No Prepay Pos
(202.50)
(226.90)
(9897.10)
38.787
-159.050
-2,472.750
Indicator - Sched A
(65.86)
(800.30)
(1791.20)
Indicator - Sched C
387.854
-21.377
280.428
Indicator - Sched C
(405.70)
(132.60)
(2588.30)
Indicator - Sched D
-67.048
-105.411
-673.212
Indicator - Sched D
(39.01)
(293.90)
(1141.50)
Indicator - Sched E
20.543
312.232
1,661.845
Indicator - Sched A
*
Indicator - Sched E
(36.89)
(1028.00)
(929.50)
Indicator - Sched F
250.481
-87.916
-2,401.500
Indicator - Sched F
(482.10)
(305.80)
(6267.30)
Num Sched C
-94.874
154.405
2,323.124
Num Sched C
(95.21)
(913.40)
(1304.50)
SOURCE: 2001 NRP.
Standard errors in parentheses.
** p < 0.05; * p < 0.10
**
*
*
50
Corcoro and Adelsheim
Appendix I—Stable Withholding Subsample Parameter
Estimations
Table I1. NRP WLS Regression —Stable Withholding
Dependent Variable: Change in Tax Liability
(underreport > 0, over report < 0)
R-sq
N
parameter
State Income Tax
Income Tax
Bal Due
Bal Due
Refund
Refund
Indicator - Interest Income
Indicator - Interest Income
Indicator - 20% tax change
Indicator - 20% tax change
BD x Interest
BD x Interest
RD x Interest
RD x Interest
BD x 20% tax change
BD x 20% tax change
RD x 20% tax change
RD x 20% tax change
BD x Sched C
BD x Sched C
RD x Sched C
RD x Sched C
BD x Sched F
BD x Sched F
RD x Sched F
RD x Sched F
BD x Age > 65
BD x Age > 65
RD x Age > 65
RD x Age > 65
No Prepay Pos
No Prepay Pos
Indicator - Sched A
Indicator - Sched A
Indicator - Sched C
Indicator - Sched C
Full Sample
0.461733
7365
tpi_lev=1
0.425449
1463
tpi_lev=2
0.453458
4482
tpi_lev=3
0.569697
1420
estimate
estimate
-20.206
(65.278)
0.730
(0.028)
-0.047
(0.031)
-147.294
(77.457)
141.659
(54.050)
-0.274
(0.026)
0.095
(0.030)
-0.067
(0.014)
-0.041
(0.011)
-0.103
(0.011)
0.000
(0.012)
0.105
(0.017)
-0.073
(0.028)
-0.269
(0.011)
0.019
(0.012)
-1,099.411
(795.751)
133.075
(55.240)
107.772
(239.815)
52.801
(48.147)
1.465
(0.088)
-0.048
(0.049)
-124.790
(64.437)
330.490
(59.547)
-0.230
(0.081)
0.039
(0.032)
-0.840
(0.091)
-0.082
(0.048)
-0.375
(0.100)
-0.025
(0.057)
-0.020
(0.283)
0.040
(0.166)
-0.606
(0.115)
0.128
(0.058)
-956.037
(371.181)
-50.854
(65.183)
95.285
(318.908)
estimate
-59.054
(44.861)
0.594
(0.035)
-0.055
(0.026)
-114.718
(60.416)
253.130
(50.270)
-0.092
(0.030)
0.043
(0.026)
-0.162
(0.022)
-0.022
(0.019)
0.283
(0.023)
0.056
(0.022)
-0.162
(0.039)
-0.112
(0.054)
-0.145
(0.033)
0.020
(0.024)
estimate
-361.690
(553.374)
0.709
(0.067)
0.311
(0.426)
1,547.723
(1477.473)
-1,623.948
(480.887)
-0.315
(0.066)
-0.271
(0.426)
0.083
(0.030)
-0.022
(0.023)
-0.206
(0.023)
-0.011
(0.024)
0.143
(0.036)
-0.069
(0.057)
-0.282
(0.024)
0.018
(0.025)
-1,139.609
(5517.760)
11.266
(761.227)
769.969
(1347.828)
**
*
**
**
**
**
**
**
**
**
**
**
**
*
**
**
**
**
**
**
**
**
**
**
-
SOURCE: 2001 NRP, stable withhold <= -+15% change in withholding .
Standard errors in parentheses.
* p < 0.05; ** p < 0.10
152.536
(35.911)
-252.862
(151.808)
**
*
**
**
**
**
**
**
**
**
**
**
**
*
**
**
**
**
**
**
**
**
**
**
51
A Balance Due Before Remittance
Table I1 . NRP WLS Regression —Stable Withholding Continued
Dependent Variable: Change in Tax Liability
Full Sample
(underreport > 0, over report < 0)
tpi_lev=2
tpi_lev=1
parameter
estimate
estimate
estimate
Indicator - Sched D
Indicator - Sched D
Indicator - Sched E
Indicator - Sched E
Indicator - Sched F
Indicator - Sched F
Num Sched C
Num Sched C
Num Sched F
Num Sched F
Indicator - Dividend Income
Indicator - Dividend Income
Indicator - Alimony Income
Indicator - Alimony Income
Indicator - Capital Gain Income
Indicator - Capital Gain Income
Indicator - Other Gain Income
Indicator - Other Gain Income
Indicator - IRA Income
Indicator - IRA Income
Indicator - Pension Income
Indicator - Pension Income
Indicator - Unemp Income
Indicator - Unemp Income
Indicator - Soc Security Income
Indicator - Soc Security Income
Indicator - Other Income
Indicator - Other Income
Indicator for Age > 65
Indicator for Age > 65
Relative TPI
Relative TPI
Primary Age
Primary Age
DIF Score
DIF Score
-65.187
(81.091)
150.031
(78.159)
987.671
(2535.644)
285.393
(186.400)
-824.093
(2511.941)
-83.104
(63.052)
-305.963
(414.093)
-29.466
(86.002)
475.131
(327.234)
-167.384
(217.653)
23.123
(89.156)
74.999
(102.698)
19.387
(97.559)
266.934
(92.381)
-34.548
(114.313)
-92.098
(17.943)
-1.330
(2.473)
1.007
(0.237)
-194.100
(104.419)
3.536
(89.993)
536.642
(7486.036)
18.770
(273.886)
-694.127
(7475.372)
-122.334
(60.705)
75.888
(382.090)
103.667
(102.842)
-239.960
(789.210)
0.831
(308.507)
-180.926
(94.468)
58.503
(78.842)
-112.863
(72.085)
488.191
(95.827)
4.941
(94.518)
462.484
(114.648)
-0.722
(1.774)
0.297
(0.226)
-23.617
(48.558)
198.630
(49.043)
1,243.002
(1425.722)
197.329
(112.276)
-573.397
(1406.304)
-52.801
(38.958)
-219.612
(246.893)
-5.744
(52.929)
-59.369
(202.203)
-231.168
(131.301)
-6.676
(59.264)
70.590
(67.897)
-107.423
(79.991)
110.477
(58.279)
-17.489
(99.077)
-20.607
(33.124)
-2.004
(1.787)
1.035
(0.155)
*
**
**
**
*
**
**
**
SOURCE: 2001 NRP, stable withhold <= -+15% change in withholding.
Standard errors in parentheses.
* p < 0.05; ** p < 0.10
tpi_lev=3
estimate
*
**
**
**
-398.500
(490.399)
-293.317
(427.974)
2,316.243
(13690.284)
844.532
(956.954)
-2,904.946
(13649.349)
-347.367
(474.882)
-30,984.763
(12364.958)
-158.073
(467.102)
1,237.954
(1327.058)
-584.702
(1070.143)
-16.224
(510.441)
130.299
(1002.272)
1,440.529
(1069.679)
307.603
(502.890)
-940.256
(1225.884)
-19.478
(42.471)
-7.171
(23.438)
1.254
(3.586)
*
**
**
**
52
Corcoro and Adelsheim
Endnotes
1
Clotfelter (1969), Adelsheim (1997), Christian et. al. (1993).
Hershey (1992); New York Times.
3 Treasury Inspector General For Tax Administration (TIGTA) Millions of
Taxpayers May Be Negatively Affected by the Reduced Withholding Associated
With the Making Work Pay Credit (2009) Reference Number: 2010–41–002.
4 SKAT maintained the option of retroactive audits on taxpayers who would
have been regularly flagged for an audit. The threat of audit letter read that
either 100 percent of the taxpayers in their group would be audited, or 50
percent of taxpayers in the group would be audited. This experiment used up
nearly 1/5th of SKAT’s resources devoted to tax audits for the given years.
5 Hershey (1992); New York Times.
6 Treasury Inspector General For Tax Administration (TIGTA) (2009).
7 Christian et al. (1993).
8 Prospect theory has been analyzed and tested in barter markets (List 2004) and
in the PGA Golf tour (Pope and Schweitzer 2009).
9 See Appendix C for a discussion in risk and utility.
10 Schepanski & Shearer (1995) focused on the reference point in terms of this
withholding phenomenon. They argue that the neutral reference is not the
actual prepayment position (as they use ‘current asset position’) but rather
the expected prepayment position (expected asset condition). With their
expected asset condition as the true reference point, they argue that a taxpayer
who expects a large balance due, but only realizes a small balance due would
consider that within the gain domain, thus act risk averse. Likewise, a taxpayer
who expects a large refund but gets a marginal refund would view that as a
loss, and may act risk seeking to capture a larger refund.
11 This research focuses on the first two properties of Prospect Theory. The actual
audit probability is unknown to the taxpayer so their assumed probability
of audit (weight) and the actual audit probably would be identical for their
compliance decision.
12 Schepanski and Kelsey (1990), White et al (1993).
13 See Appendix D for the discussion on bias and inconsistency.
14 There has been discussion that none of the available instruments are likely
to satisfy the assumptions for IV-estimation. Andreoni, Erard and Feinstein
(1998), Kleven et al. (2009).
2
A Balance Due Before Remittance
53
15 List
(2004) found that the effects of prospect theory dissipate with more
experience in the memorabilia trade market. Pope and Schweitzer (2009)
found that prospect theory holds in PGA golf player’s actions even with the
most experienced golfers.
16 The
hope is that a subset of taxpayers was not systematically excluded. Reasons
for not having a match to prior years’ returns: newly filing in 2001, change in
filing status, and spouses alternate as primary taxpayer in different years.
17 The graphs show each interquartile range. The remaining histograms are in the
appendix. In order to see the distribution of over/underreporting, each graph
suppresses the cases where there is no change in the tax liability (one-third
of each sample). In both datasets, even with the cases with zero tax change
removed, a tall spike at the mode (of $14 in underreported taxes) remains in
each dataset.
18 Due to the weights used, Table 1 does not report what was found in the sample
but rather extrapolations to the whole taxpayer population.
19 Adelsheim (1997), Ho (2003), SB/SE Seattle/San Jose IRS (2007), Kleven et al.
(2009).
20 Interquantile range lies between the 75th and 25th percentiles.
21 This current iteration ignores the reference point concerns brought up by
Schepanski and Shearer, but further research could be done adapting the
previous year’s prepayment position as the expected asset position.
22 See Appendix C for a review on varying risk behavior.
23 Alternate model specifications including tobit and multinomial probit. OLS
was used for ease of interpretation.
24 The income groups (TPI Levels) were segmented by the 25th percentile,
inter-quartile range, and 75th percentile, with the assumption that there are
behavioral differences between income groups.
25 Bolded terms in (1) are vectors.
26 Refundable credits like the Earned Income Tax Credit and the additional
child tax credit are additional line items that can have issues with reporting
compliance.
27 The variables were coded when BD > 0 then RD = 0 and RD > 0 then BD = 0.
Given this specification, to convert back to a continuous Prepayment position
variable coded in the NRP dataset, PP = BD – RD.
28 Occupation codes were taken from the reported occupation on the return,
an additional field was created if the occupation code was absent. Return
complexity was proxied by a series of indicators for the existence of attached
schedule forms. The other variables can be found in Appendix F.
54
29 Details
Corcoro and Adelsheim
can be found in Appendix D.
standard example of this issue is the economic returns of schooling. It
can be modeled that the income an individual earns can be caused by his or
her performance in school. However, both earnings and school grades are
jointly caused by an immeasurable ‘ability.’ The taxpayer’s DIF score would
be analogous to a student’s SAT score. An SAT score likely does not entirely
capture a student’s ability, the DIF score likely does not entirely capture
noncompliance.
31 A previous iteration of the research proposed using a simultaneous equation
model to examine causation between underreported tax liability and
prepayment position. After further consideration it was considered unlikely
that underreporting done at tax filing caused a taxpayer’s prepayment position.
More likely, it is an unaccounted taxpayer compliance inclination that causes
both.
32 For a discussion of instrumental variables and causation, see Appendix E or
Greene (2008) pp 74–85 and Angrist J.D., G.W. Imbens, and D.B. Rubin (1996)
for a more detailed discussion.
33 Restricting the upper bound for what could be deemed as ‘stable’ withholding,
or increasing the lower bound for a ‘large’ tax change would severely limit the
number of cases in the sample and reduce its robustness. Reducing the lower
bound for ‘large’ tax change could contaminate the sample by including some
cases where the tax change was not a shock.
34 The income groups (TPI Levels) were segmented by the 25th percentile, interquartile range, and 75th percentile.
35 The specified model without interactions: U = β + X β + β BD+ β RD +ε
0
1 1
2
3
u
The resulting marginal effect for refund due is ∂U ∂RD = β3
36 The specified model with interactions:
U = β0 + X1 β1 + β2 BD+ β3RD + X1D ·BD β4 + X1D ·RD β5 +εu.
The resulting marginal effect for balance due is w8 w%' = β2 +X1D β4.
37 The marginal effects in Table 6 assume different demographic characteristics
for the different taxpayers in each example. Example 1—taxpayer does not have
any interest income, Schedules C, F, and is under 65 years of age. Example 2—
taxpayer has interest income, no Schedules C, F, and is under 65 years of age.
Example 3—taxpayer has interest income, no Schedules C, F, and is over 65
years of age.
38 See Appendix D.
39 This is using correlations assumed using prospect theory. Numerical
simulations would help verify whether these assumptions are valid.
30 A
Predicting Intentional and
Inadvertent Noncompliance
Kathleen M. Carley, Dawn C. Robertson, Michael K. Martin, Ju-Sung Lee,
Jesse L. St. Charles, and Brian R. Hirshman; Carnegie Mellon University
T
ax noncompliance is socially harmful, as it can reduce revenues, distort
labor markets, and undermine state stability by feeding perceptions of
cheating and fraud. Reducing noncompliance can be facilitated if one understands the basis for that noncompliance. Kinsey (1984) defined noncompliance
with tax laws as the “failure, intentional or unintentional, of taxpayers to meet
their tax obligation.” Estimates of errors place the number of returns containing
either an intentional or inadvertent error, or both, above 50 percent. Minimizing
the number and size of such errors requires attending to both types of error. This
point was made in 2007 by Michael Brostek in his testimony on tax compliance
before the Committee on the Budget, U.S. Senate. For example, he noted that the
Government Accountability Office had found that simplification had the potential
to reduce the tax gap because it would reduce inadvertent errors by eliminating
confusion, decrease misuse by making it harder to hide noncompliance, and increase willingness to comply due to increased understanding. In the case of simplification, the same action can reduce both intentional and inadvertent errors.
However, when simplification is not possible, different strategies may be necessary to reduce the tax gap due to inadvertent and intentional errors. Educational
outreach, for example, is more likely to impact inadvertent errors; whereas, enforcement, withholding, and information requirements may have a greater impact
on reducing intentional errors. In order to provide a more nuanced approach to
reducing the tax gap that is tuned to the needs of the taxpayers, understanding
both intentional and inadvertent error is critical.
The majority of research on taxpayer noncompliance has been concerned with
intentional errors on tax returns (i.e., evasion). The term intentional tax error is
often used synonymously with “noncompliance” and “tax evasion.” Intentional tax
errors comprise any form of willful misrepresentation while completing a tax return, for the purposes of minimizing the tax owed or maximizing a tax refund.
Typically, these acts include under-reporting income, over-reporting deductions,
and erroneously claiming credits with the intent of noncompliance. In contrast,
inadvertent tax errors include mistakes, math errors, forgetting, and unintentional
mis-interpretation or misunderstanding.
Our research, conducted for the Internal Revenue Service (IRS), explores both
intentional and inadvertent error. We ask, is it possible, given the information on
a return, to tell whether an error is intentional or inadvertent? Thus, this work
54
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
addresses the lack of knowledge concerning unintentional errors on tax returns
and may provide potential guidance to examiners, while helping the Service better
meet taxpayer needs by identifying factors that lead to inadvertent error.
The goal is to determine when it is possible to predict intentional and inadvertent errors given only the information available on a tax return. Once the contributing factors to the commission of errors are identified, the IRS can address these
factors with the intent of reducing future errors. Also, profiles resulting from these
models may be used in a similar fashion. This would enable more customized support to taxpayers. In addition, models gleaned from this study could be used in
simulation models of taxpayer behavior enabling the IRS to explore the potential
impact of various services and interventions.
Background on Noncompliance Modeling
Theories of noncompliance generally break down into those that emphasize economic deterrence and those that emphasize fiscal psychology (Milliron and Toy,
1988). Researchers in the economic deterrence paradigm tend to employ expected
utility theory and view the taxpayer as a rational actor seeking to maximize personal gain by minimizing taxes paid. However, the evidence is mixed and taxpayers often fail to behave in an objectively rational manner. Researchers in the
fiscal psychology paradigm tend to employ prospect theory (e.g., Kahneman and
Tversky, 1979) and consider factors such as the cost of compliance and social context (Smith and Kinsey, 1987). Supporting evidence includes the generally high
rates of compliance and the fact that compliance increases with the expectation
of a refund and as knowledge of the tax law increases. Additionally, from a purely empirical perspective, there exist key correlates of noncompliance, of general
intentional noncompliance, and of inadvertent error. For example, income level,
youth and unfamiliarity with the tax laws, and gender are all highly correlated
with noncompliance. Despite this body of information, no single, clear picture of
the correlates of noncompliance exists.
This lack of a single, clear picture suggests that a multimodeling perspective
is needed. We developed the first principles models using the open-source literature, which includes the results of psychology experiments and social empirical
(survey-based) research. These models were developed in order to identify factors
outside of those derivable directly from the tax returns that might account for errors. Further, it was felt that such models might provide greater insight into why
errors occurred. Since the rationale for intentional and inadvertent errors in such
first principles models is based on different social and psychological factors that
may or may not be visible to examiners, these models should help distinguish the
two types of error. The statistical machine learning models were developed in order to identify factors that were directly derivable from tax returns. Such models
Predicting Intentional and Inadvertent Non-compliance
55
were expected to be potentially predictive, but more related to tax law in its current form and with less ability to predict the impact of changes. Since the statistical
distribution of intentional and inadvertent errors was likely to be different, the
statistical models should help distinguish the two types of error.
Modeling Errors
In this study, we take a dual-teaming approach. We have two teams, working independently from different sources, to develop models of error. Team A works
from open-source literature and has developed a model of intentional error and
another of inadvertent error from theory using only the data and information in
the published literature, much of which does not consider taxpayer applications.
These are referred to as the first principles models. Team B works from the Exam
Office Automation Database (EOAD) and the Individual Return Transaction
Files (IRTF) database provided by the IRS and, utilizing statistical and machinelearning approaches, estimates a set of empirical models which are then combined into a unified empirical model. The first principles and the empirical models are then compared and contrasted by Team C, who uses a subset of the empirical data and applies the models from Teams A and B to that data, creating a
combined model.
Compliance was modeled first for the tax return as a whole, and then for specific line items. Two line items have been modeled to date. The first line item examined was the earned income tax credit (EITC), as it is one of the most adjusted
line items. The second is wages, salaries, and tips. Other potential line items to be
modeled in the future include those found to be critical in the first principles intentional error model: capital gains, self-employment, farm income, student loans
and Social Security income.
Data Used by Teams B & C1
The IRS EOAD data includes 2.66 million records containing 2,379,523 exams
with corresponding line items and valid incomes, filing statuses, and timeliness
codes from the period 2002–2007, most of which were in 2006–2007. Of these,
only the data from 2006 and 2007 was used, as it matched with the IRTF. In addition, in 2006, examiners switched to identifying “penalty” or “no penalty” before
assigning reason codes. As this produced noticeable differences in the way reason
codes were used, we used only the 2006–2007 for consistency. It is important to
note that these are operational exams, and the returns included are those that were
thought to be noncompliant. As such, this is a biased sample. However, it was the
only available data with any non-researcher-proposed indication of error. Having
56
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
such an indication is a requirement for the specific statistical learning models employed in this exploratory study.
Of these 2,379,523 returns, all of which are in 2006–2007, 65,547 were marked
as having intentional errors, 1.22 million tax returns were marked as having unintentional or inadvertent errors, and the remaining were not marked with either
type of error by the examiners. This is a second source of bias, the examiners cannot know the intent for sure, have incentives not to mark a tax return as having an
intentional error, and the taxpayers have incentives to provide support for inadvertent error. Consequently, even among this nonrepresentative sample, there may
be fewer tax returns marked as containing an intentional error than is actually the
case. These records include 1.12 million campus (correspondence) examinations,
216,774 field exams, and the remainder are office, no-show, no-response, or undeliverable mail. Although not itself a source of bias, the type of exam is indirect
information about the likelihood of error and is information that would not be
available with a tax return not in this operational set.
The EOAD data set contains two tables, E and C. The C table contains tax return
data without specific line item information. Example fields are exam date, adjusted
gross income, and preparer. The E table contains information about the line items
examined during the audit. Every line item examined is included in this table, and
some fields included are monetary adjustment by line item, reason for the adjustment and line item identification. The C table was cleaned and duplicate keys and
records were removed. All records without valid filing statuses or adjusted gross
income fields were dropped, resulting in 2.48 million records left. The C and E sets
were combined in such a way that the tax return information was preserved from
C along with summary information from the line item set.
Intent for the tax returns was determined from the intent of corresponding line
items. If a tax return had at least one line item issue that was considered intentional, the whole tax return was marked as intentional. If a return had at least one unintentional line item, then it was considered to have inadvertent errors. This procedure resulted in some tax returns being marked as containing both intentional
and inadvertent errors. Note that an alternative would have been to consider all
the returns for which the error led to an underpayment of taxes to include intentional errors. In Figure 1, the distribution of level of error by level of adjusted gross
income is shown. As can be seen, most of the errors result in under-reporting of
income (right-hand side). However, both under- and over-reporting occur at all
income levels. Based on our research of the general factors leading to intentional
and inadvertent errors, as well as discussions with examiners, we found that it
should not be assumed that all cases of under-reporting are intentional, nor that
all over-reporting is inadvertent. In both cases, there are a number of factors that
can lead to inadvertent errors in particular, the complexity of the return.
Predicting Intentional and Inadvertent Non-compliance
57
In Figure 1, it will be seen that there are returns with an error of zero dollars. A
return that is marked with an error of size zero is one that, after the exam, either
it was determined that no adjustments need to be made or the adjustments were
such that those in the positive direction cancelled those in the negative leading to
zero total adjustment.
When analyses of individual line items were done, expected burden was used to
determine complexity. Information from an IRS-provided burden study was used
in conjunction with an estimation of the number of lines of instruction a taxpayer
would need in order to read to fill out that line item. This results in an estimation
of low to high complexity per line item using a 5-point scale. For the return as a
whole, its complexity was set based on the complexity of the line items used. To
minimize error, this was turned into a 3-point scale as follows:
• Low complexity—Form 1040, 1040A, or 1040EZ without schedules
• Intermediate complexity—Form 1040A with schedules and 1040
with Schedules A, B, D, Additional Child Tax Credit, Educational
Credits, Child Care Credit, Credit for the Elderly, or EIC
• High complexity—Form 1040 with Schedules C, E, or F, or other
schedules and all other specific Forms 1040, e.g. 1040PR, etc.
We only have the line items examined to determine which schedules were used.
As such, it is likely that we are underestimating complexity.
The IRTF data came in several tables as it is a much larger database. It includes
information about all tax returns from 2006 and 2007. In the IRS IRTF data, there
are 139 million records that exist in both years. The records were matched via
keys for EITC eligibility and age (which were calculated from the return year).
The IRTF data has fewer variables per tax return, and the data is less in-depth
than the EOAD set. However, it does contain returns not examined. We used only
those records in the IRTF that could be matched to records in the EOAD. There
were a few key pieces of data gleaned from this set for use with the EOAD data
when modeling intent. Those included date of birth, additional preparer information, and additional line items.
For the purposes of this study, for each variable, the data was placed into
predetermined categories or “bins.” These same bins are used for both the first
principles and the statistical models. The purpose of binning is four-fold: first, it
reduces error by decreasing the granularity of the data; second, it enables comparability with existing studies in the literature; third, it enables the results to
be used directly by field operatives; and fourth, it allows the results to be used
directly in the construct simulation model (Hirshman, Martin, and Carley, 2008;
Carley and Maxwell, 2006; Carley, 1990) and the SmartCard (Carley et al., 2010;
Altman et al., 2009).
ures
58
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
FIGURE 1. Distribution of Under/Over-Reporting (Loss) by Income Level
Figure The
1. Distribution
of for
Under/Over-Reporting
(Loss)
by Income
Level
income field used
our analysis was the adjusted
gross income
reported
on the return. See Table 1. Additional variables created at this step were itemization, preparer use, exemptions claimed, and an initial capital gains variable. If the
return indicated itemized rather than standard deductions, the itemized flag was
set to 1. Preparer use was gleaned from the preparer variables and categorized as
self-prepared, paid preparer used and IRS-prepared. The IRS/center-prepared tax
returns included any prepared with IRS assistance, whether by an IRS employee
or the TCE/VITA programs. We note that future work might want to separate
these two types of returns. The number of exemptions claimed on each return was
used as the exemption variable up to five. If there were more than five exemptions
claimed, the variable value was set to 6. If the capital loss field was negative, then
an initial capital gains flag is set to 1. Later, using line item data, a more robust flag
may be set.
Another variable that required binning was the monetary adjustment of each
overall tax-return: rar_ovedef_amt. See Table 2. When this field is negative, it indicates that the exam resulted in a lower tax liability than the original return indicated, i.e., the taxpayer is owed a refund. If it is positive, then the taxpayer owes
additional money to the IRS.
59
Predicting Intentional and Inadvertent Non-compliance
TABLE 1. Income Bins
Initial Bins
Super Bins
AGI < $0
Negative
AGI = 0
Low
$0 < AGI < $15,000
Low
$15,000 < AGI < $30,000
Middle
$30,000 < AGI < $50,000
Middle
$50,000 < AGI < $80,000
Middle
$80,000 < AGI < $120,000
Middle
AGI > $120,000
High
TABLE 2. Monetary Adjustment Bins
Due/Owed Bins
Owe < $0
Owe = $0
$0 < Owe < $2,000
$2,000 < Owe < $3,000
$3,000 < Owe < $4,000
$4,000 < Owe < $5,000
$5,000 < Owe < $6,000
Owe > $6,000
Bins were set so that there was an approximately uniform distribution.
After the initial adjustments and additions to the tax return set, the line item
set, E, adjustments due to line item E were made. Of the line items included, 11.3
million corresponded with tax returns from C and were used. The first thing done
was a determination of intent by reason code and by penalties. Very few line items,
82,000, were assessed penalties. Each line item had a reason code assigned by the
examiner. These reason codes were split into intent groups after correspondence
with the IRS. Possible values were intentional, unintentional or inadvertent, neutral, possible intentional, and “discard.” It is important to note that only a subset
of the reason codes was used to distinguish between intentional and inadvertent.
If a line item had a penalty associated with it, it was also considered intentional.
Later study revealed that this may not always be accurate. Finally, 57 percent of
tax returns are marked as having inadvertent errors, and 4 percent are marked as
having intentional errors.
60
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
Both the first principles and the empirical models used the same bins if they
used the same variables. There are, however, some differences in variables available
to the two modeling teams. For example, first principles models considered information about gender, which is not readily available from the tax returns. However,
the empirical models are based on information on the level of the monetary return
that is not readily available without access to the tax returns. By combining
the models, a more comprehensive view of the correlates of noncompliance
is possible.
Additional information from the IRTF data set was fused with the EOAD data.
We were only provided with a subset of the IRTF database, and, as such, the corresponding records for some of the tax returns in the EOAD were not available.
Hence, the set of tax returns used from the EOAD was pared down to just those
1.9 million records for which IRTF data was also available. The IRTF set contains
information about the superset of taxpayers, including the date of birth and additional line items used: EITC, student loan interest, capital gains, and Social Security benefits. The taxpayers’ ages and filing statuses were added to the tax return
data set. Ages were binned accordingly: under 30, between 30 and 60, and over 60
years of age.
In the EOAD data, the rates of inadvertent and intentional errors, as marked by
the examiners, increase with income (AGI) when looking at the percentages from
the actual tax return errors. See Table 3. The exception is the negative-income
category, which has an even higher rate of error than the high-income group. Note
that the error rate is significantly lower across the board for intentional error as
compared with inadvertent error. In part, this is due to a reluctance of examiners
to mark a return as containing an intentional error, as explained later.
TABLE 3. Empirical Distribution of Inadvertent Error by Adjusted Gross
Income Level
Intent/Income
Negative
Low
Middle
High
Total
23498
270356
630648
117910
1042412
Not Inadvertent
5133
308807
424153
46392
784485
Intentional
4671
10290
46396
14110
75467
Not Intentional
23960
568873
1008405
150192
1751430
Total
Inadvertent
28631
579163
1054801
164302
1826897
Inadvertent (percent)
82%
47%
60%
72%
57%
Intentional (percent)
16%
2%
4%
9%
4%
In Figure 2, the percentage of errors of each type by income level is shown. As
can be seen, the distributions are different for intentional and inadvertent errors.
In general, more tax returns are labeled inadvertent than intentional. Further, for
Predicting Intentional and Inadvertent Non-compliance
61
both intentional and inadvertent, as the level of income increases, the tendency
to label the exam with an error increases. However, for all errors marked, there
is a greater tendency to label tax returns as containing intentional errors if the
report’s income is high or negative; that is, of the returns marked with errors in the
negative-income level 16.3 percent of the marked errors are labeled as intentional,
and, of the high-income level, 11.1 percent are labeled as intentional. However, at
the low-income level, of all the returns with errors only 4 percent are labeled as
intentional. If there were no monetary differences, we would expect the fraction
of errors labeled as intentional to be similar, regardless of income level. This may
reflect a bias on the part of the examiners due to the fact that the tax loss is higher
in the negative- and high-income areas, or it may reflect a greater lack of financial literacy at low-income levels. This difference in the distribution, and the lack
of clarity on its cause, is one of the factors suggesting the need for a more comprehensive model of errors, rather than simply assuming that underpayment are
intentional errors.
FIGURE 2. Percentage of Labeled Errors by Adjusted Gross Income Level
Figure 2. Percentage of Labeled Errors by Adjusted Gross Income Level
Model Details
The first principles and machine-learning models employ different variables due
to the way in which they are constructed. These differences are summarized in
Table 4. These first principles models did not make use of the EOAD/IRTF data.
The intentional error model contains variables that are available on the tax return
62
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
and, so, can be applied to the combined EOAD/IRTF data. The inadvertent error
model, at this point, contains less of that information and, so, cannot be applied to
the EOAD/IRTF data as easily. As part of the next phase of study, we will impute
the relation between the variables in the EOAD/IRTF data and the variables used
in the first principles, inadvertent error model. In this latter case, future work will
seek to find a mapping between the variables in the inadvertent error, first principles model and those items available on tax returns.
TABLE 4. Variables Used by the Different Models
Variable
1st Principle
Intentional
1st Principle
Inadvertent
Machinelearning
EITC
no
no
yes
Age
yes
yes
yes
Burden/Complexity
no
yes
yes
Late
no
yes
yes
Filing Status
no
no
yes
Itemization
no
no
yes
Exemptions
no
no
yes
Preparer
no
no
yes
Error Amount
no
no
yes
Income
yes
yes
yes
Gender
yes
yes
no
Belief in obey law
yes
no
no
Education
yes
yes
no
Team A
The first principles models, as they are derived from the general literature and
not the EOAD/IRTF data, provide a principled way of characterizing errors that
can be applied to any return. The model of intentional errors from first principles predicts the probability that individuals will commit some error as determined by their sociodemographic traits, namely gender, age, education, and
income, as well as their attitudes toward obedience to the law (Lee and Carley,
2009). This model incorporates scientific findings from several published papers on tax evasion and represents their weighted average, taking into account
their similarities to the recent U.S. population. In Figure 3, the intentional error
model, for the standardized regression models (or path coefficients) for predicting an intentional error, as derived from the open source literature, is shown. As
can be seen, tendency to believe that laws should be obeyed, age, and, indirectly,
education are primary drivers.
Predicting Intentional and Inadvertent Non-compliance
63
The inadvertent error model from first principles takes into account issues of
literacy, the relative complexity of the tax law, stress due to time of filing, and basic
sociodemographic correlates of error to predict taxpayer mistakes. The basic inadvertent error model is shown in Figure 4. In this case, general sociodemographic
traits have a diagnostic role only to the extent they correlate with financial literacy
and the expectation to receive a refund. In general, the dominant factor in producing an inadvertent error is task complexity; in other words, the burden in filling
out the relevant line items.
FIGURE 3. First Principles Intentional Error Model
Age
Figure 3. First Principles Intentional Error Model
Taken together, the two first principles models suggest about a 45- to 50-percent error rate, of which about 30 percent are inadvertent and 20 to 30 percent are
intentional. And, although we have not yet been able to estimate it, these models
suggest that there are likely to be returns with both intentional and inadvertent
error, particularly when the complexity of the return is high.
Team B
The empirical model of errors is a composite model employing three machinelearning and statistical techniques: the Proc Logistic regression model developed
in SAS, a Bayesian Network Prediction model, and a j48 decision tree classifier
with multiboosting. The models for error were formulated with 10 explanatory
64
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
variables and a binary response variable. For one set, the response variable is intentional error, and the other set has inadvertent error. The 10 explanatory variables are: income, error amount as determined by the exam, complexity (burden),
late code, preparer used, exemptions, filing status, age, EIC, and itemization.
Proc Log is a linear regression procedure used to model dichotomous outcomes
of interest, such as the error variables. A linear function is produced to model the
relationship between the explanatory and dependent variables. The error variables
were coded as “0” for no error and “1” for an error in order to be used with Proc Log.
Proc Log can produce several “goodness of fit” indicators, but Proc Log was used primarily to produce classification tables for the IRS data. Once the classification tables
were produced from the labeled set, they were used to predict outcomes in both the
labeled and unlabeled sets for the intentional and inadvertent models.
The other software used for prediction was the Belief Net (BN) Power
Constructor. This Bayesian network predictive software uses a conditional independence-based algorithm to construct a directed acyclic graph. Given the binned
variables, this software can produce a graph that will calculate error probabilities
for each tax return. Like the Proc Log classifiers, the resulting models are applied
to the labeled and unlabeled sets for comparison. The predictive software (SAS
and BNP) uses a tolerance of 0.5 to determine whether the model predicts that a
particular tax return has an error. Changing this tolerance lowers or increases the
threshold for prediction. We used a tolerance of 0.5 for inadvertent errors and 0.1
for intentional errors. This difference is a direct result of the fact that there are so
few known cases of intentional error.
The models are learned using data gleaned from the EOAD and IRTF data sets
provided by the IRS. The EOAD data is split into two sets: “labeled” and “unlabelled.” The labeled set is further divided into two overlapping sets: “intentional”
and “inadvertent.” This was done at the full tax return level and by line item. The
unlabelled set had neither intentional nor inadvertent errors. The data was again
split by four income groups: negative, low, medium, and high. Each of these income groups has a substantially different profile in terms of taxpaying behavior
and, so, errors. These splits were applied overall and by line item. Several line items
or issues associated with each tax return were derived from the line item set. These
include tips, self-employment income, farm income, alimony, as well as another
indicator for capital gains. In lieu of learning separate models for exam types, such
as field or campus, we simply controlled for the exam type.
Comparing, Contrasting and Testing the Models—
Team C
The first principles intentional error models and the empirical models for intentional and inadvertent errors are applied to the labeled sets to determine how
Predicting Intentional and Inadvertent Non-compliance
65
well the models work. This is done for the overall tax return and by selected line
items. After the models are assessed using the labeled data, they are then applied
to the unlabeled sets to determine how many of these forms can be characterized.
Finally, to create a composite model, the predictions of the various independent
models are combined. Both intersection and union are explored.
Model results are strongest when controlling for income as cause, type, and
level of error are different. There is substantial overlap among models suggesting a
class of cases for which there is strong ability to discriminate between intentional
and inadvertent errors. However, each of the models has a different strength with
respect to the cases with less clear signals. Hence, a composite model, formed by
combining the diverse models, provides a more comprehensive assessment.
Results
Both first principles and machine-learning models were built separately for inadvertent errors and for intentional errors. These models suggest that it is possible
to discriminate apparently intentional from inadvertent errors for most returns.
Of the 1,042,412 tax returns marked as inadvertent by the examiners, 81 percent
are predicted to be inadvertent using machine-learning models. Of the 784,485
tax returns marked as intentional by the examiners, approximately 50 percent are
predicted to be intentional using the machine-learning models. Of the records
marked as both intentional and inadvertent by the examiners, approximately 84
percent are predicted to be both inadvertent and intentional using the respective
machine-learning models. Using the first principles models, a higher percentage
of the tax returns are marked as containing intentional errors.
Of the tax returns marked as inadvertent, 2 percent are predicted to be intentional by the empirical models. There are two possibilities:
1. The flags that are set by the examiner are wrong
2. The flags are correct and the intentional error models are “over”
predicting
If the flags are wrong, then this 2 percent means that these models identify an
additional 2 percent of the cases as containing intentional errors. If the flags are
correct, this 2 percent error means that we would expect these intentional error
models to incorrectly suggest that returns might contain intentional errors 2 percent of the time for returns already selected as thought to contain an error. This
would be the cap on the inaccuracy of these models.
We expect that refined models that look at line items and explore the correlations among those may further increase the predictive value of the results. We
also expect that combining the final models from Teams A and B will result in a
66
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
better general model that can be used by the Service in a variety of ways, including compliance-related education for both IRS enforcement staff and the taxpayer.
We now turn to a more detailed analysis of the modeling results for inadvertent and then intentional errors. In this more detailed analysis, we consider both
labeled and unlabeled exams.
Inadvertent Errors
We developed from the open literature a general or first principles model of inadvertent errors. This is shown in Figure 4. As can be seen, two factors that drive
inadvertent errors are complexity of the problem and financial literacy. That is,
higher return complexity combined with lesser financial literacy translates to an
increased likelihood of inadvertent error.
FIGURE 4. First Principles Inadvertent Error Model
Figure 4. First Principles Inadvertent Error Model
Predicting Inadvertent Errors (0.5 Tolerance)
LABELED SET—ONLY THOSE CLASSED AS INADVERTENT
Accuracy results from applying the learned models to the known or labeled set of
tax returns are shown in Table 5. Note that the predictive models return a percent
likelihood of error for each tax return. The tolerance for these outcomes is set at
the default of 0.5. At 0.5, the sum of the percentage of correct positives and correct
67
Predicting Intentional and Inadvertent Non-compliance
negatives is usually maximized, for both types of error. The tolerance is not a confidence interval. It is simply a cut-off point for whether the exam is predicted to
have an error or not. Moving away from 0.5 increases the likelihood of false positives or false negatives. In an operational context, a different tolerance might be
used for intentional errors if, e.g., the policy was to examine all possible cases of
intentional error even if there is a high chance that the error, if there was one was
not intentional. Similarly, for inadvertent errors, a policy that “education never
hurts” might use a tolerance that produces a high level of false positives.
TABLE 5. Inadvertent Error Predictions by Models Independently and Collectively
Given Labeled Tax Returns
Negative
Income
Confirmed Potential Predicted
Error
Error
Error
Low
Inaccuracy
Confirmed Potential Predicted
Error
Error
Error
Inaccuracy
BNP
80.96%
16.51%
97.47%
1.04%
30.03%
16.51%
46.54%
16.97%
PL
80.98%
16.51%
97.49%
1.02%
29.80%
17.16%
46.96%
17.20%
BNP ⋂ PL
80.51%
16.96%
97.47%
1.49%
28.24%
18.52%
46.76%
18.76%
BNP ⋃ PL
81.43%
16.06%
97.49%
0.57%
31.59%
15.14%
46.73%
15.41%
Average of
Models
80.97%
16.51%
97.48%
1.03%
29.92%
16.83%
46.75%
17.09%
Confirmed
Maximum
82.00%
47.00%
Middle
Income
Confirmed Potential Predicted
Error
Error
Error
High
Inaccuracy
Confirmed Potential Predicted
Error
Error
Error
Inaccuracy
BNP
49.93%
23.51%
73.44%
10.07%
71.76%
28.24%
100.00%
0.24%
PL
49.42%
24.39%
73.81%
10.58%
70.29%
26.64%
96.93%
1.71%
BNP ⋂ PL
47.02%
26.46%
73.48%
12.98%
70.29%
28.24%
98.53%
1.71%
BNP ⋃ PL
52.33%
21.44%
73.77%
7.67%
71.76%
26.64%
98.40%
0.24%
Average of
Models
49.68%
23.95%
73.63%
10.33%
71.03%
27.44%
98.47%
0.97%
Confirmed
Maximum
60.00%
72.00%
In table 5, 8, 11, and 14, the percentage errors for labeled tax returns are shown. To
generate the values shown, the following factors were considered. Note that there are
two ways for a model to match the conclusions of the examiner. A model can label
the tax return as having the same type of error (inadvertent or intentional) as marked
by the examiner. We refer to these as confirmed errors. Or, a model can label the tax
return as not having an error of that type and the examiner also marks the tax return
as not having an error of that type. These are confirmed non-errors and will not be reported. Similarly, there are two ways in which the models can mismatch the examiners.
68
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
A model can label the tax return as having an error of that type, and the examiner did
not mark it as such. We refer to these as potential errors, as they are tax returns that
the models would also characterize as having an error of that type. Or, a model can
label the tax return as not having an error of that type, but the examiner did mark it as
having an error of that type. We refer to these as mistakes.2 It should be noted that the
difference between the percentage of returns marked as having that type of error by
the examiner (see Table 3) and the percentage of the returns that are confirmed errors
are the mistakes. The basic idea behind this demarcation is that, although examiners
may under-report errors, if they do mark an exam as containing a particular type of
error, they are unlikely to be wrong. The percentages under mistakes can be thought
of as the minimum level of inaccuracy expected when these models are used. Another
feature of many of these tables is that we present results for both an intersected and a
union approach on confirmed errors. In tables with these combinations, ∩ is used for
intersection on confirmed and ∪ to denote union on confirmed. This refers to the way
in which the models were combined for the confirmed errors, as well as confirmed
nonerrors. In the case of the union, the potential errors are those cases for which none
of the models suggested it was not in error.
More exams are marked as inadvertent than as intentional by the examiners, 57
percent and 4 percent respectively. If a model were to exactly match the examiners
findings, the maximum number of labeled tax returns the model would label 4 percent as intentional. A model that exactly matches the examiners would for inadvertent
errors have a higher percentage of the returns characterized as confirmed errors and
for intentional errors have a higher percentage characterized as confirmed non-errors.
The sum of confirmed errors and potential errors is the number of labeled exams a
model suggests has that type of error. This is the predicted error. The maximum possible predicted error that can be confirmed is also shown in these tables. The predicted
error will be higher than the confirmed maximum when the model predicts exams to
contain an error of that type and the examiner did not. If the model does not mark a
labeled exam as having an error when the examiner does label that exam as having an
error, we refer to that as an inaccuracy in the model; alternatively, we could view this
as cases where, if the model is correct, the examiner has erroneously labeled the exam.
Looking at Table 5, we can see that, for inadvertent error, the minimum level of inaccuracy is highest (i.e., the difference between the confirmed error and the confirmed
maximum) when the return is from someone in the low- and middle-income area. In
contrast, for negative- and high-income cases, the models tend to mark as inadvertent
the same cases marked by the examiners. Specifically, the models estimate that more
than 90 percent of these returns contain inadvertent errors. However, the models suggest that only 40 percent to 50 percent of the low-income returns and 65 percent to 75
percent of the middle-income returns contain inadvertent errors.
The accuracy is highest for negative and high incomes. Also, there is a great deal
of overlap between the two models. However, the percentage of false positives is
quite high. Increasing the tolerance or threshold for a positive result will minimize
the false positives but at a cost to overall accuracy. If examiners have a tendency to
69
Predicting Intentional and Inadvertent Non-compliance
mark exams as inadvertent, even if they are not, then these models can be interpreted as suggesting, on average, that for low- and middle-income cases, 17.09 percent and 10.33 percent of the cases, respectively, may have been erroneously marked
as inadvertent.
UNLABELLED DATA
Table 6 contains the models’ predictions for inadvertent error in the unlabeled
set of tax returns. Note that the percentages predicted are higher than the actual
percentages based off the labeled set. This may be because something about the tax
return or the taxpayer alerted the examiner that the case was inadvertent, so they
just did not mark it. Or, this may be because there were other signals during the
exam for the labeled cases that suggested they were intentional.
TABLE 6. Percentage of Unlabeled Exams Predicted to be Inadvertent by Models
Negative
Low
Middle
High
BNP
Model/Income
83.4%
69.0%
81.4%
100.0%
PL
80.3%
66.9%
84.0%
94.9%
BNP ∩ PL
76.3%
62.8%
76.6%
94.9%
BNP ∪ PL
87.4%
73.1%
88.8%
100.0%
PROFILES OF TAX RETURNS WITH ERRORS FOR WHICH THE ERRORS
ARE LIKELY TO BE INADVERTENT
Because so many examined returns have inadvertent errors, picking definitive
profiles is challenging. Many returns have both intentional and inadvertent errors. Nevertheless, trends definitely emerge. Illustrative profiles by income level are
shown in Table 7. For all income groups, higher burden is associated with inadvertent error. We note that the first principles model for inadvertent error also suggests
that complexity (and therefore burden) is a major contributor to inadvertent error.
In this table, NA means not applicable.
Burden is consistently higher for erroneous tax returns. Although it is not always
higher for every single tax return, when looking at the percentages of erroneous tax
returns versus ones without inadvertent error, a clear pattern is shown. For example,
90 percent of nonerroneous tax returns in the negative group are in the lowest burden
group. Eighty percent of those in the error group were in the highest burden group.
Also, the percentage of those married filing jointly increases in each erroneous group.
This may be a result of more opportunity for error as more lines of tax returns must be
completed compared with those filing singly or as a head of household. Also, younger
taxpayers (in the under 30 bin) have lower percentages of erroneous tax returns. Again,
this may be due to younger people having less complicated tax situations in general.
70
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
TABLE 7. Profiles Consistent with Inadvertent Errors
Age
Low
Middle
High
Negative
Mixed
Use Paid
Itemized
Preparer
Less
Likely
More
Likely
Income
Late
Burden
EIC
Mixed
Mixed
High
More
Likely
Mixed
High
More
Likely
Joint–
More
likely
High
NA
Mixed
Mixed
Joint–
More
likely
Older
Less
Likely
More
Likely
Higher
More
extensions
Slightly
Older
Slightly
more
likely
Mixed
NA
Mixed
Slightly
Older
Slightly
more
likely
Mixed
NA
Mixed
High
FS
Predicting Intentional Errors (0.1 Tolerance)
LABELED SET—ONLY THOSE CLASSIFIED AS INTENTIONAL
The accuracy results from applying the learned models to the known or labeled set of
tax returns are shown in Table 8. While the accuracy percentage is quite high (typically inaccuracy is less than 5 percent), there are many false negatives. Essentially,
the models underpredict intentional errors at the 0.5 level, resulting in a high number of correct negatives. When the tolerance is set to 0.1, a wider net is cast, and more
tax returns will be classified as intentional. This lowers the number of cases for which
a model claims there is no intentional error, and the examiner marks the exam as
containing an intentional error (false negatives). And, it increases the number of
cases for which a model claims that the error is intentional, and the examiner does
not (false positives). While there is a great deal of overlap in the Bayes Net and Proc
Log models, the first principles model has different, yet still similar, results. Overall,
by combining the models, a stronger result is produced.
We set the tolerance lower for intentional than for inadvertent errors for two reasons. First, there were simply far fewer tax returns marked as intentional. Second,
by setting it lower, the overall mismatch with the examiners is lower. However, even
though the overall mismatch is lower, the number of returns for which a model suggests there is an intentional error and the examiner does not will be higher. Thus, we
erred on the side of forecasting potential errors.
In table 8, we see that the first principles model and the union of models with the
first principles models tend to predict more intentional errors and tend to have lower
minimum levels of inaccuracy. As with the inadvertent errors, the models are more
accurate for negative and high income than for low income.
71
Predicting Intentional and Inadvertent Non-compliance
TABLE 8. Intentional Error Predictions by Models Independently and Collectively
Given Labeled Tax Returns
Negative
Income
Confirmed Potential Predicted
Error
Error
Error
Low
Inaccuracy
Confirmed Potential Predicted
Error
Error
Error
Inaccuracy
BNP
14.17%
56.02%
70.19%
1.83%
1.03%
4.36%
5.39%
0.97%
PL
14.28%
55.43%
69.71%
1.72%
1.03%
4.66%
5.69%
0.97%
FP
13.15%
59.41%
72.56%
2.85%
1.43%
42.20%
43.63%
0.57%
FP ⋂ PL
11.86%
71.35%
83.21%
4.14%
0.94%
42.59%
43.53%
1.06%
FP ⋃ PL
15.58%
43.49%
59.07%
0.42%
1.53%
4.27%
5.80%
0.47%
FP ⋂ BNP
11.78%
71.98%
83.76%
4.22%
0.93%
42.52%
43.45%
1.07%
FP ⋃ BNP
15.54%
43.45%
58.99%
0.46%
1.53%
4.04%
5.57%
0.47%
BNP ⋂ PL
13.48%
62.01%
75.49%
2.52%
0.93%
5.33%
6.26%
1.07%
BNP ⋃ PL
14.97%
49.45%
64.42%
1.03%
1.14%
3.69%
4.83%
0.86%
⋂ all
11.29%
73.61%
84.90%
4.71%
0.85%
42.68%
43.53%
1.15%
Union All
15.77%
39.13%
54.90%
0.23%
1.55%
3.46%
5.01%
0.45%
Average of
Models
13.81%
56.85%
70.65%
2.19%
1.17%
18.16%
19.34%
0.83%
Confirmed
Maximum
16.00%
2.00%
Middle
Income
Confirmed Potential Predicted
Error
Error
Error
High
Inaccuracy
Confirmed Potential Predicted
Error
Error
Error
Inaccuracy
BNP
2.72%
13.14%
15.86%
1.28%
5.71%
24.92%
30.63%
3.29%
PL
2.68%
13.06%
15.74%
1.32%
6.09%
27.04%
33.13%
2.91%
FP
2.61%
37.80%
40.41%
1.39%
4.79%
33.43%
38.22%
4.21%
FP ⋂ PL
2.02%
41.14%
43.16%
1.98%
3.73%
46.45%
50.18%
5.27%
FP ⋃ PL
3.27%
9.72%
12.99%
0.73%
7.15%
14.01%
21.16%
1.85%
FP ⋂ BNP
2.03%
41.40%
43.43%
1.97%
3.65%
44.52%
48.17%
5.35%
FP ⋃ BNP
3.30%
9.54%
12.84%
0.70%
6.85%
13.82%
20.67%
2.15%
BNP ⋂ PL
2.57%
14.28%
16.85%
1.43%
5.42%
29.93%
35.35%
3.58%
BNP ⋃ PL
2.83%
11.93%
14.76%
1.17%
6.38%
22.03%
28.41%
2.62%
⋂ all
1.95%
41.74%
43.69%
2.05%
3.51%
48.03%
51.54%
5.49%
Union All
3.34%
8.92%
12.26%
0.66%
7.29%
12.51%
19.80%
1.71%
Average of
Models
2.67%
22.06%
24.73%
1.33%
5.51%
28.79%
34.30%
3.49%
Confirmed
Maximum
4.00%
9.00%
72
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
UNLABELED DATA
Since the predictive models for intent determine so few errors, lowering the tolerance to 0.1 results in percentages of erroneous tax returns more in keeping with the
actual exam error percentages. These results are shown in Table 9.
TABLE 9. Percentage of Unlabeled Exams Predicted to be Intentional by Models
Model/Income
Negative
Low
Middle
High
7.0%
BNP
39.3%
2.0%
6.3%
PL
30.5%
2.3%
5.7%
9.0%
FP
35.7%
30.2%
23.4%
25.1%
FP ⋂ P PL
14.7%
0.9%
2.5%
2.2%
FP ⋃ PL
51.5%
31.5%
26.7%
32.0%
FP ⋂ BNP
13.9%
1.0%
2.4%
2.1%
FP ⋃ BNP
61.1%
31.2%
27.3%
30.1%
BNP ⋂ PL
23.4%
1.2%
4.9%
5.5%
BNP ⋃ PL
46.4%
3.1%
7.2%
10.6%
Intersect all
11.0%
0.7%
2.0%
1.6%
Union All
64.5%
32.1%
27.7%
33.1%
PROFILES OF TAX RETURNS WITH ERRORS FOR WHICH THE ERRORS
ARE LIKELY TO BE INTENTIONAL
By income level, the profiles of tax returns with intentional and unintentional errors are somewhat different. For all four income groups, markers for intentional
error include self preparation, age greater than 30 years, high complexity, and no
EITC. For all income groups, except low income, itemized deductions were also
well represented. One consistent difference is the representation of head of household filers. They are consistently more represented in the “no error” group. Fewer
of them and more married taxpayers appear in the group that make intentional
errors. It should be noted that people may claim head of household status, even
if they are not eligible to do so. We did not control for this. If we could determine
that this claim was wrong, then that might move some of these cases to the intentional error category, if, in fact, this error was not inadvertent. However, the
complexity of determining eligibility for head of household status in and of itself
is likely to increase both intentional and inadvertent errors. Taking into account
errors on other factors, such as head of household status, is a point for future research. These profiles for intentional errors are shown in Table 10.
73
Predicting Intentional and Inadvertent Non-compliance
TABLE 10. Profiles Consistent with Intentional Errors
Itemized
Late
Exemptions
Error Amount
Filing Status
Low
No
Extension
and No File
<2
Very High
and Low
Single and
Married-J
Middle
Yes
Extension
Mixed
Very High
and Low
Single and
Married-J
High
Yes
Mixed
Mixed
Very High
and Low
Married-J
Negative
Yes
Extension
Mixed
High
Married-J
Line Items—EITC
The first line item modeled was the earned income tax credit (EITC). This was
because it is one of the most examined line items, being concentrated in low- and
middle-income groups. It is also one of the most complex of the line items. As
such, according to the theoretical first principles models, the likelihood of both
intentional and inadvertent errors is likely to be higher than for other line items.
Over 940,000 EITC line items were examined in the set. The average adjustment
was −$2,285 and the total was −$2.15 billion. Almost all returns were labeled as
containing inadvertent errors (more than 99 percent), while there were very few
returns marked as containing intentional errors (less than 1 percent) for all income
groups except the high-income group. Due to the nature of the EITC line items,
there are no tax returns that employ this line item that are in the high-income
bracket. The models behaved accordingly. We note that the distribution of errors
for the EITC line item is not symmetric about zero; i.e., in most cases the errors
result in tax-loss (under-reporting). The distribution is slightly more symmetric
for taxpayers with low income who take the EITC than for other income levels.
As with the entire set of tax returns, we do not make the assumption that errors
resulting in under-reporting are intentional.
EITC MODELS—LABELED SET—ONLY THOSE CLASSIFIED AS
INADVERTENT
The BNP and Proc Log models for error on the EITC line item, unlike the corresponding models for error somewhere on the overall tax return, do not use taking
the EITC credit as a control. The EITC error results are shown in Table 11. The BNP
line item model for EITC results in a much higher percentage of false positives
than the full tax return BNP model. The Proc Log model outperforms the full tax
74
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
return model significantly and, remarkably, does not overlap very much with the
BNP model. This line item may be a good candidate for ensemble learning because
of this lack of overlap. By combining the models in an ensemble, the strengths of
both individual models can be exploited. It is likely that the Proc Log model is
overestimating the likelihood of inadvertent errors. As such, in this case, it would
not be reasonable to use the union of the two models as the composite model of
inadvertent errors. Another important point is that the minimum level of inaccuracy is much lower for the Proc Log model than the BNP.
TABLE 11. Inadvertent Error Predictions by Models Independently and Collectively
Given Labeled Tax Returns for Inadvertent Error on the EITC Line Item
Negative
Income
Confirmed
Potential
Error
Error
BNP
41.98%
0.00%
PL
100.00%
0.00%
BNP ⋂ PL
41.98%
BNP ⋃ PL
100.00%
Low
Confirmed
Potential
Error
Error
58.02%
33.53%
0.75%
0.00%
86.57%
4.88%
0.00%
58.02%
29.24%
0.00%
0.00%
90.87%
Mistakes
Middle
Confirmed
Potential
Error
Error
65.72%
36.80%
0.12%
63.08%
8.54%
88.07%
4.23%
7.71%
4.93%
65.84%
32.65%
4.25%
63.10%
0.71%
8.42%
92.22%
0.10%
7.69%
Mistakes
Mistakes
EITC MODELS—UNLABELED INADVERTENT EITC LINE ITEMS
COMPARED WITH INTENT ON OVERALL TAX RETURN
As previously noted, the unlabeled set had no error designation, so, for the sake of
comparison, the results of the EITC line item models were compared with the intent
ascribed to the overall tax return. This compares, for a specific return the type of error
on a line item with the type of error on the tax return as a whole. For the unlabeled set,
the Bayes Net model outperformed the Proc Logistic model. Proc Log tended to mark
the vast majority of the line items as inadvertent, which resulted in large percentages
of false positives. For the line items, the model can be applied to the unlabeled data
directly, and/or in comparison with the predicted intentionality of the tax return as a
whole. In Table 12, the latter is shown. In this case, we assume that the predicted type of
error for the tax return as a whole is correct. Then if a model labels the EITC line item
as inadvertent, and the parent model labeled the overall tax return to be inadvertent,
we would say that is a confirmed error. If a model labels the EITC line item as inadvertent, when the overall tax return was not labeled as inadvertent, then that model
is suggesting there is a potential error on that line item. If a model does not label the
EITC line item as inadvertent, but the overall return was labeled as inadvertent, then
that model is either mistaken, or the source of error is on a different line item. From a
conservative point of view, then, the minimum inaccuracy would be that all of these
last cases are actually model mistakes and the percentage shown can be thought of as
the minimum possible mistakes.
75
Predicting Intentional and Inadvertent Non-compliance
TABLE 12. Match of the Models Independently and Collectively for Inadvertent
Errors on the EITC Line Item for Unlabeled Tax Returns Assuming that
the Prediction for the Overall Tax Return holds
Negative
Income
Matches
Overall
Exam
Low
Potential
Error
Mistake
Matches
Overall
Exam
Potential
Error
Middle
Mistake
Matches
Overall
Exam
Potential
Error
Mistake
BNP
50.98%
4.25%
44.77%
77.55%
0.42%
22.02%
77.19%
3.56%
19.25%
PL
50.98%
48.53%
0.49%
24.74%
73.30%
1.95%
23.76%
74.74%
1.50%
BNP ⋂ PL
6.70%
48.53%
44.77%
4.66%
73.31%
22.03%
5.99%
74.74%
19.27%
INADVERTENT EITC PROFILE
The EITC profiles are somewhat different from those for the tax returns when
viewed in their entirety. See Table 13. All but two negative income tax returns were
classified as having inadvertent errors, so there was no basis for a comparison.
Also, high-income tax returns were excluded, as so few claimed EITC. Again, the
people who tended to make errors were a bit older and the complexity somewhat
higher, but it was not as pronounced as in the whole tax returns. There were fewer
distinguishing characteristics between those who made errors and those who did
not. This is at least partially due to the high percentage (over 99 percent) of those
making errors.
TABLE 13. Profiles Consistent with Inadvertent Errors on EITC Line Item
Age
Preparer
Complexity
Exemptions
Error Amount
FS
Low
Mixed
Slightly
more self
Higher
More < 2
Higher
More Singles
Mid
Slightly Older
Mixed
Mixed
More < 2
Higher
More Singles
and HOH
EITC MODELS—LABELED SET—ONLY THOSE CLASSIFIED AS
INTENTIONAL
Very few of the examiners marked the EITC line item as containing an intentional
error. Table 14 clearly demonstrates the effects of “rare events” on our models. The
rare event in this case is the designation of an intentional error on the EITC line
item. The Bayes Net Model marked every single tax return as not having an error
and was not included. The other two models marked some, though very few, tax
returns as having intentional errors on the EITC line item. As with the set of tax
returns as a whole, for the EITC line item, the models are quite likely to label a
return as not having an intentional error when the examiner also marks it as such.
76
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
Overall, these models are more able to identify inadvertent errors on the EITC line
item than on the return as a whole and are unable to identify intentional errors
marked by examiners. This shortcoming is likely to be overcome by simply building the models on more data.
TABLE 14. Intentional Error Predictions by Models Independently and Collectively
Given Labeled Tax Returns for Intentional Error on the EITC Line Item
Negative
Income
Confirmed
Potential
Error
Error
PL
0.41%
0.00%
FP
0.00%
0.00%
FP ⋂ PL
0.00%
FP ⋃ PL
0.41%
Low
Confirmed
Potential
Error
Error
0.00%
0.01%
0.00%
0.41%
0.00%
0.00%
0.00%
0.41%
0.00%
0.00%
0.00%
0.01%
Mistakes
Middle
Confirmed
Potential
Error
Error
0.58%
0.00%
0.00%
0.77%
0.58%
0.00%
0.00%
0.77%
0.00%
0.58%
0.00%
0.00%
0.77%
0.00%
0.58%
0.00%
0.00%
0.77%
Mistakes
Mistakes
EITC MODELS—UNLABELED INTENTIONAL EITC LINE ITEMS
COMPARED WITH A LABEL OF INTENTIONALITY ON THE TAX RETURN
AS A WHOLE
Compared with the predictions for the return as a whole, the EITC line item model produces similar results. See Table 15. Again, the Bayes Net Model was excluded
as it marked no errors. This data suggests that, even when the overall exam is likely
to include an intentional error, intentional error is likely not to be on the EITC line
item. One must be cautious in over interpreting the EITC result, however, as it is
based on so little data.
TABLE 15. Match of the Models Independently and Collectively for Intentional
Errors on the EITC Line Item for Unlabeled Tax returns Assuming that the
Prediction for the Overall Tax return holds
Negative
Income
Low
Middle
Confirmed
Confirmed
Potential
Error
Error
Mistakes
Confirmed
Potential
Error
Error
Mistakes
Error
Potential
Overall
Error
Mistakes
Exam
PL
0.16%
0.33%
3.92%
0.00%
0.05%
0.13%
0.00%
0.01%
0.11%
FP
0.00%
0.00%
4.08%
0.00%
0.00%
0.13%
0.00%
0.00%
0.11%
FP ⋂ PL
0.00%
0.33%
4.08%
0.00%
0.05%
0.13%
0.00%
0.01%
0.11%
FP ⋃ PL
0.16%
0.00%
3.92%
0.00%
0.00%
0.13%
0.00%
0.00%
0.11%
77
Predicting Intentional and Inadvertent Non-compliance
INTENTIONAL PROFILES
Very few intentional tax returns were identified by the models, which is not surprising, as intentional errors are truly rare events as labeled in the data by examiners. Like whole tax returns, for the EITC line item, the preparer tends to be self and
the complexity high. Also, once again, head of household is not as represented in
the error group. See Table 16.
TABLE 16. Profiles Consistent with Intentional Errors on EITC Line Item
Age
Low
Preparer Itemized Income
<60
Middle
<60
Negative
>30
<60
Self
Self
Mixed
No
Yes
Mixed
Lower
Late
Less
on time
Complexity Exemptions
High
More < 2
Error
Amount
FS
High
Very
Few
HOH
Mixed
Mixed
High
More < 2
Mixed
Very
Few
HOH
Mixed
On
Time
High
Mixed
Mixed
Married-J
Discussion
It is important to recognize that these models are not true models of error so much
as models of error as determined by IRS examiners. This is due to the data used.
The EOAD data contain only operational exams. Consequently, the tax returns
are not representative of the population. They were selected for examination because of some perceived noncompliance. In deciding which tax returns to further
examine, a set of selection criteria are used resulting in a set of tax returns that
are suspected to contain errors. Thus, the first source of bias is selection on the
dependent variable-error. Future work should take the proposed models and test
against a random sample of all tax returns.
The second limitation is the criterion for defining error. The criterion we used
for asserting that the tax return contained an intentional error was that the examiner marked it as such. If an examiner did not mark a tax return as intentional,
then we would not have marked it as such. In general, examiners cannot know for
sure whether an error is intentional or inadvertent. Making that judgment requires
knowledge of the taxpayer’s true motives at the time of preparing the return or
possibly an admission of intent. However, such information is generally not available. In addition, examiners have very significant incentives not to characterize
an error as intentional, since that generally carries with it a higher standard of
proof. While taxpayers have every incentive to claim that they forgot, lost, or did
not know something, for one taxpayer, that may be true and inadvertent, but for
78
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
another similarly situated taxpayer, it may be a simple attempt to cover up intentional noncompliance. To mitigate this bias, we used a jittering approach in which
we tested the models by relabeling a few of the tax returns as intentional or not and
rebuilding the models. This did not appreciably change the results.
Discussions with IRS staff also led to the conclusion that expectations about
the source of error and/or level of error impacted the type of exam; e.g., field or
campus. This source of bias is related to the differing proportions of intentional
marking given the different types of exams. To mitigate this source of bias, all tax
returns were considered collectively with controls for types of exams considered.
However, future work might create separate models for field and campus exams,
for the empirical models, and see how different these are as compared to the unified model that controls for the type of exam. In particular, differences vis-a-vis
the profiles should be examined. In doing this, it will be important to keep in mind
that the first principles models consider motivation and, as such, represent factors prior to the exam; hence, there should not be separate first principles models
vis-a-vis the exam type. Additionally, it should be recognized that differences in
profiles resulting from separate models of the field and campus exams will still be
subject to the criticism of sampling on the dependent variable.
This research focused on accuracy; i.e., “was there an error ?” Other key avenues of exploration are the degree of error (e.g., did it lead to a 1 percent or 30
percent difference in payments?) and the ambiguity of the error. We found that
many tax returns contained both intentional and inadvertent errors and that, in
some cases, one could not discern whether a particular error was intentional or
inadvertent. This suggests a “gradient” in the “intentionality” of the error. Thus,
one can think of some errors as being, e.g., 60 percent inadvertent. Future research
might consider whether there are systematic factors that lead to classes of errors
that are diagnosable as having a certain percentage of inadvertency. This could
lead to a new support for the taxpayer; e.g., if there are a set of line items for which
those errors that occur tend to appear as between (30 and 70 percent ) inadvertent
and for which the dominant factors are degree of burden or complexity and experience of the taxpayer then, this might suggest that such items provide sufficient
ambiguity that they both confuse the taxpayer and that they present opportunities
for those inclined to noncompliance.
In this research, the data did not support a systematic examination of the differences in compliance relative to math and nonmath errors or use or nonuse of
software in preparing taxes. Future work should explore alternative data to see if,
in fact, there is greater symmetry in errors with respect to math errors than other
errors and whether, in fact, the errors in favor or not in favor of the taxpayer with
respect to math errors have reduced proportionally with increasing use of computation support such as e-preparation and e-filing of tax returns. In addition, our
simulation work shows that the use of software and paid preparers are also critical
codeterminants of error. In part, e-filing and the use of software in preparing taxes
reduced math errors in particular.
Predicting Intentional and Inadvertent Non-compliance
79
Conclusion
This research suggests that it is possible to identify factors associated with intentional and inadvertent noncompliance on tax returns. From a theoretical perspective, the core difference in causes of errors from the first principles models is that
a belief in obeying laws will decrease intentional errors and is irrelevant for inadvertent errors; whereas, complexity or burden is a strong predictor of inadvertent
errors and is not a direct predictor of intentional errors. The machine-learning
models suggest that, for inadvertent errors, age, use of paid preparers (no for negative and low income, yes for middle and high income), taking the EIC, and the
overall burden/complexity of the exam are diagnostics for assessing when an error
is inadvertent; whereas, filing late, taking multiple exemptions, and larger errors
are diagnostic of the error being intentional.
The most challenging part of this effort has been dealing with rare events. In
general, many statistical-leaning models work better when there are vast quantities of data and when the data contains a uniform set of results. While a 50/50
split on the results (inadvertent/intentional) is not required, a more even split than
99.9/.01 is helpful. Despite the rarity of the event (the intentional error), trends are
definitely emerging for both inadvertent and intentional errors; however, more
work is needed on the models to increase the accuracy and robustness of the results. This challenge is difficult for the tax returns as a whole; but, it is even worse
for the individual line items. One possible way of mitigating this would be to get
more data. Another extension would be to see if imputing labels for line items that
are not labeled, when the exam as a whole is labeled, would alter the results.
Our investigations suggest that the key to improved accuracy is to employ an
ensemble of techniques that blend results from multiple diverse models. As noted
previously, the various models have different strengths and weaknesses and, as
such, tend to pick up on different aspects of the factors that lead to errors. By
blending the models, a more robust comprehensive picture emerges. We note that
even blending the Bayes Net, the Proc Log, and the first principles models improve
the predictive model for intentional errors. We expect the same will be true for
the inadvertent errors. The gains, however, will be larger for intentional than for
inadvertent errors, as a higher percentage of the tax returns marked as containing
inadvertent errors as opposed to intentional errors by the examiners were classified as inadvertent by the machine-learning models. In addition, the gains will be
larger at the individual line-item level. If sufficient gain is made at the line-item
level, it might be possible to then re-estimate the type of error for the exam as a
whole using a composite of line-item characteristics and overall exam characteristics. The lower accuracy of the machine-learning models for intentional errors
means greater room for improvement as additional machine learning techniques
are employed. Although not reported here, we are currently investigating models
that may reach as high as 80-percent accuracy.
80
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
The examination of individual line items is another way of improving the overall accuracy of the results vis-a-vis classifying at least those returns containing
errors as an examiner would have. With individual line items, there is still the
problem that intentional errors are a rare event; however, restrictions on which
taxpayers can utilize which line items does alter the proportions and makes the
distribution slightly less rare. Further, by building models of errors for key line
items, an overall improved ensemble model is made possible. Future work should
expand on this by focusing on an exploration of additional line items and building a composite model using line-item and overall predictions. Self-employment
promises to be a fruitful line item to consider.
Future research should explore more components of the tax return, such
as Schedule C. For Schedule C, we note that it should not be treated as a uniformly complex item. For example, returns with a Schedule C of greater than
$10,000 are more complex than are Schedule C returns under $10,000. Using a
more fine-grained assessment of complexity of the various parts of the tax form
seems warranted, particularly as complexity has turned out to be a dominant
underlying root cause.
Other ensemble techniques should also be used. For example, the intersection/
union results for intentional models show that, by adding in the first principles
model, accuracy can be improved. The next step here is to employ the specific coefficients from the first principles models for intentional and inadvertent models
in the statistical models.
Having a wider range of data would also help improve the model, as it would
provide more cases and examples of returns without errors. This would support
the use of unsupervised learning techniques and enable us to make better use of
the first principles models. Using such techniques is critical if we are to move further beyond the constraints imposed by training models on the basis of exam results. The core issue will be determining the extent to which these techniques can
provide useful models of error, intentional and inadvertent, that are independent
of known biases. The lack of data also influences the number of variables available. For this study, we did expand the data by fusing and cleaning portions of two
different datasets—EOAD and IRTF. The result of this data fusion and cleaning is
that we were left with a small set of variables for which there was data on all tax
returns. However, many of these variables showed no relation to errors in any of
the models. The end result was a set of variables for which there was clean data,
sufficient variance, and some relation with errors. We find that, adding other of
the available variables did not tend to significantly change the results. If more data
were available, that might lead to additional variables in the models. Based on the
data currently available, the main additional variables we could add have to do
with metrics on where errors occur in the tax return assessments.
Finally, we note that, while it is useful to know whether an error is intentional or
not, the underlying core issue is simply, “what is the root cause of the error?” This
Predicting Intentional and Inadvertent Non-compliance
81
work has suggested a number of root causes: individuals simply not believing that
they should obey laws, complexity of the tax form (and, therefore, burden placed
on the taxpayer with differential impacts based on whether this is a numerical calculation burden or general cognitive/verbal complexity), assumptions about risk,
timepressure, and inexperience in paying taxes. Further work should be done to
refine this list and identify those classes of taxpayers and portions of the tax forms
where one or more of these root causes is dominant. Such further work should
consider using, as feasible, the codes used by the National Research Program and
the reason codes. We note that the ultimate goal is to identify interventions that
could be focused on types of taxpayers who are predicted to have a high likelihood
of either inadvertent or intentional noncompliance. Distinguishing inadvertent
from intentional is a first, albeit insufficient, step to address root cause. This research lays the groundwork for identifying those interventions.
Acknowledgments
This work is part of the Taxpayer Simulation and Dynamics Networks projects
at the Center for Computational Analysis of Social and Organizational Systems
in the Institute for Software Research and the School of Computer Science at
Carnegie Mellon University. Support was provided by the IRS Office of Program
Evaluation and Risk Analysis. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the
official policies, either expressed or implied, of the Internal Revenue Service or the
U.S. Government. The authors would like to thank Chris Hess, Amy Sriuthai, and
Alan Plumley for their insightful and helpful comments.
References
Altman, N.; M.K. Martin; D. Robertson; and K. M. Carley (2009), SmartCard
Prototype, Carnegie Mellon University, School of Computer Science, Institute
for Software Research technical Report: CMU–ISR–09–122, Pittsburgh, PA.
Brostek, M. (2007), “Tax Compliance: Multiple Approaches are Needed to
Reduce the Tax Gap.” Testimony before the Committee on the Budget, U.S.
Senate. GAO–07–391T.
Carley, K.M. and D.T. Maxwell (2006), “Understanding Taxpayer Behavior and
Assessing Potential IRS Interventions using Multiagent Dynamic-Network
Simulations,” Recent Research on Tax Administration and Compliance:
Proceedings of the 2006 IRS Research Conference, Dalton and Kilss (eds.),
Washington, D.C., pp. 96–103.
82
Carley, Robertson, Martin, Lee, St. Charles, and Hirshman
Carley, K.M.; N. Altman; M. Martin; J. Meikle, and T. Suiter (2010), “Compliance,
Assistance, and the SmartCard.” Poster presented at the June 2010 IRS Research
Conference, Washington D.C.
Carley, K.M. (1990), “Group Stability: A Socio-Cognitive Approach,” in Lawler
E.; B. Markovsky; C. Ridgeway; and H. Walker (eds.), Advances in Group
Processes: Theory and Research. Vol. VII. Greenwhich, CN: JAI Press, pp. 1–44.
Hirshman, B.R.; M.K. Martin; and K.M. Carley (2008), “Modeling Information
Access in Construct,” Carnegie Mellon University, School of Computer
Science, Institute for Software Research, Technical Report, CMU–ISR–08–115.
Kahneman, D., and A. Tversky (1979), “Prospect Theory: An Analysis of Decision
Under Risk,” Econometrica, 47, pp. 263–291.
Kinsey, K. (1984), “Survey Data on Tax Compliance: A Compendium and
Review,” Working Paper #8716, (Chicago: American Bar Foundation, 1984).
Lee, Ju-Sung and Kathleen M. Carley (2009), “Predicting Intentional Tax Error
(Noncompliance) Using Open Source Literature and Data,” Carnegie Mellon
University, School of Computer Science, Institute for Software Research,
Technical Report CMU–ISR–09–125.
Milliron, V. and D. Toy (1988), “Tax Compliance: An Investigation of Key
Features,” The Journal of the American Taxation Association. 9, pp. 84–104.
Smith, K., and K. Kinsey (1987), “Understanding Taxpayer Behaviour: A
Conceptual Framework with Implications for Research.” Law and Society
Review, 21, pp. 639–663.
Endnotes
1
The EOAD and IRTF data were held in a secure facility on a standalone machine following the CASOS technical control policy guidelines.
Only members of the Center for Computational Analysis of Social and
Organizational Systems team at Carnegie Mellon University who were cleared
by the IRS to handle sensitive data were allowed access.
2 From an experimental perspective, these categories of match and mismatch
are the same as the traditional false+ and false− distinction that is used when
ground truth is known. Since there is reason to suspect that the examiner
markings contain errors, we use the term match and mismatch instead of
correct and false. In summary, potential errors are, from a ground-truth
perspective, false positives; whereas, mistakes are, from a ground-truth
perspective, false negatives.
4
D
Tax Code Complexity and Compliance Burden
Contos Guyton Langetieg Vigil
Sawyer
Eichfelder Schorn
Individual Taxpayer
Compliance Burden: The
Role of Assisted Methods
in Taxpayer Response to
Increasing Complexity
George Contos, John Guyton, Patrick Langetieg, and Melissa Vigil,
Internal Revenue Service
D
esirable features of any tax system are fairness, simplicity, and efficiency.
These characteristics encourage voluntary participation—raising revenue for government operations, while minimizing the cost of collection.
People certainly argue about the fairness of the U.S. income tax system, but there
is a definite consensus that it is highly complicated. The U.S. income tax is progressive; higher marginal tax rates are applied at higher income levels, but these
are offset by a myriad of tax deductions and credits (not to mention all the rules
and exceptions to rules that apply to each). There has been a steady trend toward
increased complexity since the Tax Reform Act of 1986. In recent years, the trend
has been compounded by increased administration of economic stimulus and
social benefits programs (e.g., health care reform) directly through the tax system.
Another desirable feature of a tax system is efficiency—that is, a tax system
that does not distort economic decisions. For example, high marginal tax rates
may discourage labor market participation leading to lower economic output.
Alternatively, a generous tax credit that targets home ownership may result in
overinvestment in the purchase of homes, and underinvestment in alternatives
such as retirement or savings. Compliance burden, the subject of this paper, is a
form of inefficiency in the tax system.
A highly complicated tax system, such as that of the U.S., presents opportunities for evasion, impedes compliance, and requires significant resources to administer. Allingham and Sandmo (1972) and Slemrod (1985) analyzed the role of
risk and the taxpayer’s decision to avoid taxes in the context of determining the
optimal level of tax law enforcement, or compliance. Slemrod notes that governments, however, do not expend the level of resources necessary to enforce the
tax laws optimally. Andreoni et al. (1998) reinforce this by noting that the IRS
budget has been declining in recent decades. They point to further evidence in
declining examination rates, which stood at about 4 percent in the 1960s, but
have declined to about 1 percent in recent years. Furthermore, the frequency with
4
Contos, Guyton, Langetieg, and Vigil
which penalties are applied to taxpayers who underreport income or otherwise
evade taxes has also declined. According to the authors, these conditions encourage taxpayers to increasingly conclude that the benefits of increased evasion are
greater than the potential costs associated with being detected and the likelihood
of paying penalties.
A study by Karlinsky et al. (2004) further reinforces this conclusion. In their
study, participants were surveyed regarding their attitudes toward the severity of
various crimes, revealing that tax evasion is generally not considered a serious
crime and, furthermore, is seen in a somewhat positive light by some taxpayer
segments—perhaps a reflection of frustration on the part of the taxpaying public
concerning tax system complexity. Nonetheless, this is against the backdrop of
a Federal tax gap estimated at $345 billion for Tax Year 2001 alone (Mazur and
Plumley, 2007).
Because the U.S. tax system is so complicated, compliance burden—the time
and money taxpayers expend to comply with Federal tax law—can be a barrier to
compliance. Slemrod & Sorum (1984) showed that the resource costs of filing the
individual income tax was 5 percent to 7 percent of Federal and state revenue collected among employers who participate in the withholding system—or, equivalently, 1.4 percent of aggregate adjusted gross income. Erard and Ho (2003) showed
that compliance burden, defined as the average time in hours to complete a tax
return, was positively related to noncompliance, either through frustration of the
taxpayer or ignorance of tax provisions resulting from complexity.1 Another study,
Slemrod (1985), concluded that relatively more complicated tax returns, such as
those filed by the self-employed, and those that report capital gains income, experience higher compliance burden expenditures. In addition, taxpayers with a higher
value of time (measured as the after-tax wage rate) and taxpayers who itemize
tended to use the services of paid professionals. The author further concluded that
“complexity encourages taxpayers to interpret the tax law to their advantage” and
“unpredictability and the existence of complicated ways to avoid taxes may erode
confidence in the fairness of the tax system and thereby affect voluntary compliance.” More recently, Auerbach et al. (2010) discuss compliance costs as a factor
expected to impede compliance with the new health insurance mandates.
The IRS began reporting compliance burden on tax forms beginning in 1988,
using the A.D. Little (ADL) study. This model focused on compliance burden primarily associated with the time spent on gathering tax materials, recordkeeping,
form preparation, and form submission, and ignored other components of compliance burden such as tax planning and out-of-pocket costs. This latter omission
became increasingly relevant as more taxpayers availed themselves of paid assistance from preparers and software. Furthermore, the estimates became increasingly out of date as time passed.
In 1998, the IRS contracted with IBM to develop an improved methodology to
measure and model the compliance burden imposed by the tax system. The result
Individual Taxpayer Compliance Burden
5
was a model able to measure compliance burden (the time and out-of-pocket costs
associated with filing taxes), inform tax policy (measure the compliance burden
resulting from a change in tax policy), and guide administrative initiatives.
The current model is an extension and significant reformulation of the original
IRS/IBM model. It also measures components of burden associated with changes
in tax policy and tax administration, but it benefits from new survey data and
updated econometric analysis. The current formulation also benefits from a common modeling framework with the business burden model. Another notable
change is that the current model measures total monetized burden, comprised of
both time and out-of-pocket costs.
The most recent model follows the general framework discussed in Guyton et
al. (2003) and, similarly, by the President’s Advisory Panel on Tax Reform (2005).
As such, it measures total burden as the direct compliance costs and inefficiencies
that would disappear if the Federal tax system did not exist. Specifically, total
burden is divided into two components: tax liability and excess burden. Tax liability is defined as the tax, penalty, and interest paid by taxpayers. Excess burden
represents the remaining costs: compliance burden, efficiency costs, psychological
costs, and administrative costs. Compliance burden, the time and money taxpayers spend to comply with the Federal tax system, includes all activities associated with filing a tax return: tax planning, recordkeeping, gathering tax materials, etc. As mentioned above, efficiency costs are costs associated with distortions
in the allocation of capital resulting from the incentive effects of the tax system.
Psychological costs reflect the frustration or anxiety induced by the tax system.
Finally, administrative costs are direct budgetary costs of administering the tax
system. The primary cost of administering the Federal tax system is the budget
of the IRS.
Components of total burden often interact with one another. For example,
taxpayers may spend more time and money on additional tax planning in order
to reduce their income tax liability; taxpayers may forego benefits associated with
voluntary credits in order to avoid the costs associated with filing the credits; or
some taxpayers may choose to hire a paid professional to prepare their return in
response to a change in tax policy that resulted in increased complexity.
This paper presents the current econometric specifications of the individual
taxpayer burden model developed using recently collected Tax Year 2007 compliance burden data. The resulting model is then used to estimate compliance
burden using data from an earlier survey which covered Tax Years 1999 and 2000.
Finally, the model and data from each survey are used to develop compliance burden estimates for the intervening years. This was necessary since the modeling
effort will support estimation of compliance burden in future years, as well as support longitudinal compliance analysis by estimating burden for prior years. In
doing so, we attempt to understand the effects of changes in tax system complexity,
technology, and the use of assisted tax-preparation methods, since these factors
changed dramatically during this period.
6
Contos, Guyton, Langetieg, and Vigil
Individual Taxpayers
From Tax Years 2000 through 2007, individual taxpayers increasingly adopted assisted preparation methods (paid preparers, tax preparation software, and
IRS-sponsored tax preparation services—VITA and TCE) to complete their
tax returns.2
Table 1 shows that, in Tax Year 2000, 72 percent of all taxpayers used an assisted method to complete their 1040-series (1040, 1040A, 1040EZ) tax returns.
That year, paid preparers completed 55 percent of all individual returns, while selfpreparers using software accounted for 17 percent of all individual returns. The
percentage of taxpayers using assisted methods increased steadily until Tax Year
2006, reaching 87 percent, before declining in Tax Year 2007 to 85 percent (that
year a large volume of simple returns were filed only to collect Economic Stimulus
Payments), before rising again to 92 percent in Tax Year 2008. During the same
time period, the percentage of taxpayers that prepared their own tax returns declined from 28 percent in Tax Year 2000 to only 8 percent in Tax Year 2008.
TABLE 1. TY 2000–2008 1040-Series Returns Prepared With Assistance in
Percentages
Tax Year
Third Party
Prep Paid
Preparer
Self Prepared with Software
Assistance (On-Line Filed/Self
V-Coded/Telefiled)
Taxpayers Using a
Paid Preparer or
Software A
ssistance
or TCE & VITA
2000
55%
17%
72%
2001
57%
19%
76%
2002
57%
20%
79%
2003
61%
22%
84%
2004
59%
25%
85%
2005
59%
25%
86%
2006
60%
26%
87%
2007
56%
26%
85%
2008
60%
29%
92%
SOURCE: R:S–97 Report, IRS Master File System, Compliance Data Warehouse, ETA, SOI.
TABULATIONS: IRS:RAS:RFDA, April 2010.
NOTES:
TY 2004 software filings include telefiled returns.
TY 2006 data includes 1040/A/EZ returns filed only to claim Telephone Excise Tax Refund.
TY 2007 data includes an estimated 15 million taxpayers that filed solely to claim an Economic Stimulus Payment.
A major reason for the trend toward increased usage of tax preparation assistance is increased tax system complexity.3 One proxy for tax system complexity
is the number of times the tax code can be subdivided. Preliminary IRS research
indicates, as illustrated in Figure 1, that the number of subdivisions and crossreferences, proxies for the complexity of the Internal Revenue Code, increased
by about 30 percent between Tax Years 2000 and 2007. In addition, a growing
7
Individual Taxpayer Compliance Burden
number of taxpayers—about an 8-percent increase over the same timeframe—
have found themselves subject to form lines that require involved recordkeeping
and complicated calculations, such as the alternative minimum tax; increasingly
complicated capital gains rules; and an increasing number of elective benefits.
Although a major driver of individual taxpayers’ increasing migration to
assisted-preparation methods is tax system complexity, we find evidence that
some taxpayer segments that had not encountered increased tax system complexity are, nonetheless, migrating to assisted methods. This suggests that there are
reasons beyond tax system complexity taxpayers consider when choosing a preparation method.
Figure 1: U.S. Tax Code Complexity: Individual Taxpayers by Tax Year
FIGURE 1. U.S. Tax Code Complexity: Individual Taxpayers by Tax Year
20,000
18,000
16,000
14,000
12,000
10,000
10 000
8,000
6,000
4,000
2,000
-
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Number of Subdivisions
Number of Cross-References
Total
One explanation is that changes in the taxpayer population over time may have
resulted in the filing of more complicated tax returns. In Table 2, we compare
select taxpayer data to illustrate changes in the demographic and economic characteristics of the taxpayer population between Tax Years 2000 and 2007. The number of returns and all of the remaining filing categories increased during this time
period. For example, all returns filed rose from 129.4 million in Tax Year 2000 to
143.0 million in Tax Year 2007, an increase of 10.5 percent. However, when we
consider the growth in the share of the categories, a more appropriate measure
of changing taxpayer composition, we see that taxpayers filing joint returns decreased by 2.7 percent, while taxpayers filing head of household increased by 5.2
percent, suggesting that more tax returns are being filed by nondependent taxpayers. In addition, the share of returns reporting salaries and wages, returns traditionally associated with relatively lower burden, declined by 0.7 percent. The opposite is true for a number of categories associated with higher burden. Categories
such as business income or loss, pension and social security income, statutory
adjustments, alternative minimum tax, and the earned income credit all increased
8
Contos, Guyton, Langetieg, and Vigil
significantly, ranging from percentage share growth of 15.1 percent to 185.1 percent.
Overall, we see evidence that returns filed for Tax Year 2007 were more complicated than those filed for Tax Year 2000, due to an increase in return complexity
beyond increases in tax law complexity.
TABLE 2. Change in the Demographic and Economic Characteristics of the
Individual Taxpayer Population between Tax Years 2000 and 2007
Number of Returns
Shares
Tax Year
2000
Tax Year
2007
Growth:
% change
Tax Year
2000
Tax Year
2007
All returns [1]
129,373,500
142,978,806
10.5%
Joint returns
50,268,249
54,065,030
7.6%
38.9%
37.8%
Growth:
% change
-2.7%
Head household
18,208,359
21,169,039
16.3%
14.1%
14.8%
5.2%
Salaries & Wages
110,168,714
120,844,802
9.7%
85.2%
84.5%
-0.7%
Business or profession
net income [2]
13,312,586
16,932,476
27.2%
10.3%
11.8%
15.1%
Business or profession
net loss [2,3]
4,287,423
5,696,992
32.9%
3.3%
4.0%
20.2%
Pensions & Annuities
[4]
21,765,211
27,678,148
27.2%
16.8%
19.4%
15.1%
Social Security Benefits
[5]
10,608,572
15,011,961
41.5%
8.2%
10.5%
28.0%
Total statutory adjustments [6]
23,197,425
36,050,434
55.4%
17.9%
25.2%
40.6%
Alternative Minimum
Tax [7]
1,304,198
4,108,964
215.1%
1.0%
2.9%
185.1%
Earned income credit
[8]
19,277,225
24,583,940
27.5%
14.9%
17.2%
15.4%
SOURCE: SOI Bulletin Historical Table 1. Individual Income Tax Returns: Selected Income and Tax Items for Tax Years
1999–2008.
[1] Year-to-year comparability of the “all returns” total is affected by changes in dollar income filing thresholds, while
year-to-year comparability of the number of returns by type of tax form used is affected by changes in the specific filing
require.
[2] Rent net income (or loss) excludes sole proprietorship (including farm) rental income or loss; these are included in
business or profession net income or loss. Rental losses are before “passive loss” limitation and, therefore, exceed the
amount included.
[3] Losses are after “passive loss” limitation.
[4] Excludes Individual Retirement Arrangement (IRA) distributions.
[5] Up to 85 percent of Social Security benefits were taxable.
[6] Includes adjustments not shown separately in this table. Total excludes certain business-related expenses, deduction for two-earner married couples, and certain alimony payments.
[7] Under “alternative minimum tax” (AMT), generally high-income taxpayers to whom the tax applied were required to
pay the larger of the regular income tax or the AMT.
[8] In Table 1, the amounts “used to offset income tax before credits” and “used to offset other taxes” (that are incomerelated) are reflected in the statistics for “total tax credits”; however, “excess earned income credit (refundable)” is
reflected.
Individual Taxpayer Compliance Burden
9
Another possible explanation is improvements in labor productivity over time,
since labor productivity is conceptually related to the cost and ease of preparation.4
We believe that productivity improvements in technology have mitigated some of
the increased burden associated with both tax and nontax system complexity and
is also contributing to the general trend toward assisted tax preparation. Bureau
of Labor Statistics data indicate that labor productivity in the tax preparation services sector, as well as overall labor productivity, increased by about 15 percent to
20 percent between Tax Years 2000 and 2007.5 In their survey of large businesses,
Slemrod and Blumenthal (1993) found that these taxpayers increasingly managed
complexity through computerization—purchasing computer hardware or filing
software. We observe results consistent with the interpretation of productivity
gains when we compare compliance burden data from the Tax Year 2007 survey of
taxpayers to the surveys conducted for Tax Years 1999 and 2000. While additional
study of the role of labor productivity and its impact on choice of preparation
methods is warranted, the data from these studies are consistent with the interpretation that taxpayer compliance costs are at least staying constant in real dollar
terms over this period and may be decreasing despite increases in tax system and
tax return complexity. A similar qualitative interpretation can be found in a recent
temporally comparative report on the compliance costs for small businesses conducted by New Zealand Inland Revenue (2010).
Individual Taxpayer Burden Modeling Approach
The primary objective of the individual taxpayer burden model is to measure and
explain individual compliance burden. We developed a model reflecting the recent public finance literature and used current statistical techniques. When developing this model, we identified two criteria that the model should possess. First,
it was important that the model be easily adaptable to changes in the economy and
the tax system. Second, the model should be sufficiently generalized so that compliance burden of other taxpayer populations, such as businesses and tax-exempt
entities, could be modeled. The resulting model meets both criteria.
Economic Model
Following the methodology in Contos et al. (2009a) and Contos et al. (2009b),
which modeled the compliance burden of small businesses, we employ a log-linear specification in which the natural logarithm of burden is linearly related to a
set of explanatory variables. The dependent variable, log(Burden), is monetized
time and money and is based on data obtained from surveys of individual taxpayers. To control for the type and volume of activities performed by each taxpayer,
tax items from the primary forms and schedules were organized into one of four
10
Contos, Guyton, Langetieg, and Vigil
complexity categories: low, medium-low, medium, and high. We use the logarithm of modified positive income as a proxy for the taxpayer’s income in the current period. Modified positive income generally represents the income amounts
on the 1040-series returns with some modifications.6
We assume that taxpayers select the combination of time and money necessary
to fully comply with the requirements of the Federal tax system while minimizing compliance costs. This assumption may not hold true for all taxpayers all
the time, but we believe that taxpayers tend to adopt a compliance process that
reduces complexity. For example, low-complexity taxpayers are more likely than
high-complexity taxpayers to undertake all of the prefiling and filing activities
related to their tax returns without assistance. As modified positive income increases, and, with it, complexity, taxpayers face a higher opportunity cost on these
activities. Taxpayers with relatively high complexity may invest in tax preparation
software or hire a professional tax preparer to advise them on issues related to tax
planning, as well as the preparation and filing of their tax returns. The increased
out-of-pocket costs reduce the time they spend on tax-related activities so we see a
substitution of money for time. For many of these higher income and more complicated returns, our model shows lower compliance costs associated with assisted
methods, despite higher fixed costs and additional consumption of often higher
quality services.
The Data Set
As mentioned previously, compliance burden data was collected by surveying
taxpayers who filed an individual tax return for Tax Years 1999, 2000, and 2007.
The most recent survey, which covered Tax Year 2007 taxpayers, resulted in 6,968
responses and a response rate of 48 percent. The population was defined as individual taxpayers who filed any of the tax forms in the 1040 series: Forms 1040,
1040A, or 1040EZ. The sample was a stratified random sample, which, when
weighted, represents the individual taxpayer population. Nonresponse bias analysis was conducted and the sample weights were adjusted accordingly, as discussed
in Brick et al. (2009).
The earlier survey effort, hereafter referred to as the 2000 survey, was conducted
in two phases. The Wage and Investment (W&I) and Self-Employed (SE) surveys
focused on taxpayers who filed a return for Tax Years 1999 and 2000, respectively.
In total, 6,366 responses were completed from the W&I population for a response
rate of just under 61 percent. Likewise, 9,081 responses were collected from SE
taxpayers, for a response rate of 56 percent. Again, the samples were stratified
random samples representative of the individual taxpayer population.
The surveys collected information on both the time and money individuals
spent on prefiling and filing activities, as well as some demographic data. Each
survey was then linked to the matching administrative record to create the estimation dataset. The administrative record includes select items from the primary
Individual Taxpayer Compliance Burden
11
tax forms and various secondary forms and schedules. Both the survey and administrative records were extensively reviewed and cleaned for memory recall and
administrative and processing errors. The data received further cleaning through
the application of robust regression methods discussed in the Simulation Issues
section of this paper. Data missing as a result of incomplete responses or robust
regression were imputed using multiple imputation techniques as discussed in
Contos et al. (2009b).
Econometric Model
To model the conditional distribution of taxpayer compliance burden, we employ
a log-linear regression specification in which the natural logarithm of burden is
linearly related to a set of explanatory variables. This type of model is supported by
the survey data, as well as the survey findings of large and mid-size business taxpayers conducted by Slemrod and Blumenthal (1993) and Slemrod and Venkatesh
(2002). The model was developed and finalized using the Tax Year 2007 survey
data, since they are the most recent survey data available. The 2000 survey data was
then run through the same model. For comparability reasons, all money amounts
were converted to 2007 constant dollars. In addition, the model was updated to
reflect tax law relevant for each year by adjusting the complexity categories.
Next, we considered pooling the data from the two surveys and generating
econometric specifications from this data. As discussed earlier, this is an important exercise since the model will be used to estimate burden in future years, as
well as support longitudinal compliance analysis by estimating burden in prior
years. The Chow test was used to determine if pooled coefficients for both data
years were superior in explanatory power to a model that estimates a distinct set
of coefficients for each year. Based on the results of the Chow test, we concluded
that separate coefficients should be estimated for each year..7
Since the pooled model was not an option, we proceeded to account for changes in technology and use of assisted tax preparation methods over the 2000 to
2007 time period by making two assumptions. We assumed that the changes occurred at a constant rate and that productivity changes in self- and paid-prepared
returns were closely related to overall labor productivity. First, we blended the
results from the two models. The blended estimates for Tax Years 2001 to 2006
were produced using a weighting scheme that favored estimates from the closest
survey year..8 Next, we used Bureau of Labor Statistics data using overall labor
productivity to adjust the estimates for changes in productivity between Tax Years
2000 and 2007.9, 10
Because one of the objectives of this modeling effort was to estimate burden
in future years using IRS administrative data, the choice of independent variables
was limited to IRS data.11 Using administrative data from subsequent tax years
allows us to produce burden estimates for forecast years. The dependent variable,
12
Contos, Guyton, Langetieg, and Vigil
log(Burden), is based on survey responses. It is defined as the natural logarithm of
total prefiling and filing compliance costs—that is, the monetized time and money
taxpayers spend to comply with Federal tax laws. Monetizing compliance costs
enables us to account for the substitution of time and money and allows aggregation of burden across activities to create a single measure of compliance burden. The key choice was whether to monetize the value of time and add it to the
out-of-pocket costs or rather to “chronotize” the out-of-pocket costs and add it to
time. Monetizing time was adopted for both technical and program management
reasons.12
As mentioned previously, we expect compliance burden to increase with modified positive income (mpi), but at a decreasing rate, so we expect the coefficient to
be positive and less than 1.
The most unique aspect of modeling compliance burden is the need to control
for the type and volume of activities performed by taxpayers to meet their Federal
tax obligations. We developed a proxy for the type of activities performed. Each
tax item from the primary forms and schedules was organized into one of four
complexity categories: low, medium-low, medium, and high. These complexity
categories are based on the notion that burden increases as a function of both the
type of tax-related activities completed by the taxpayer, as well as the volume completed. For example, if an individual completes an additional tax item one year,
holding everything else constant, compliance burden should increase since the
taxpayer will have adjusted his recordkeeping, familiarized himself with the relevant taxpayer instructions, or perhaps paid higher preparation fees. A complete
listing of the variables and complexity assignments is included in Appendix A.
To develop the complexity categories, we initially placed the various tax items
into categories based on recordkeeping intensity and tax planning activities.13 To
test the assignment criteria, the model was then run with the natural logarithm of
the money amount for each item as a separate right-hand-side variable. The magnitude of the estimated coefficients was compared with the rest of the items in that
complexity category. Items that had coefficients significantly different from their
peers were moved to a more suitable category.
As a proxy for the volume of activities, we used the money amounts reported
by each taxpayer for that item. This is based on the notion that the larger the
amount reported on a tax item, the more transactions typically associated with
the activities related to that line. The variable Low is then obtained by summing
the natural logarithms of all values on lines categorized as having low complexity.
The remaining complexity variables, Medium-Low, Medium, and High are defined
analogously. By utilizing the properties of logarithms in the complexity categories,
the equation acquires a desirable property, that is, each tax item included in the
categories acts as a separate regressor, but the coefficients of all items of the same
category are restricted to be the same.
We included dummy variables to measure the effect of preparation method
on compliance burden where self-preparation is the reference category. The
Individual Taxpayer Compliance Burden
13
remaining preparation categories represent paid (Paid) and software preparation
(Soft). Taxpayers who use assisted preparation methods typically incur additional
costs through the purchase of assistance from tax professionals such as certified
public accountants, enrolled agents, or from tax preparation software such as
TurboTax, TaxCut, etc. As discussed earlier, the trade-off is a reduction in the
amount of time it would have taken taxpayers to research and complete each line
of the tax forms themselves. In addition to preparation of their tax returns, taxpayers may also receive tax-planning advice and can be reasonably assured that
they receive tax benefits (elective credits) they may not have received had they
prepared their own tax return. In addition to an accurately prepared tax return,
taxpayers may also benefit from representation in the event they are contacted
by the IRS about their tax return. We expect the coefficients for the preparation
dummies to be positive since there are fixed costs associated with using assisted
methods.
To control for efficiency gains associated with hiring a paid professional, we
include in the specification an interaction term between Paid and the natural logarithm of mpi, Paid-mpi. As discussed above, a taxpayer who has hired a paid
professional may experience lower marginal compliance costs for additional items
than he would have experienced if he either self-prepared or software prepared
his own tax return. Although complexity is already captured in the complexity
categories, this interaction term captures any additional effect of complexity on
burden unaccounted for by the complexity categories. Since this interaction term
represents the reductions in burden at the margin associated with hiring a tax preparer, we anticipate the coefficient to be negative and smaller in magnitude than
the mpi coefficient.
To control for additional efficiency gains associated with hiring a paid professional or using software, we include a variable that measures the percentage of
forms a self-prepared taxpayer did not have to consider when completing their
return. This variable, Consider, is equal to 1 minus the ratio of the number of lines
on the forms the taxpayer filed over the total number of lines on all individual tax
forms. As such, it represents the percentage of all form lines associated with forms
a taxpayer did not fill out when completing his or her tax return.14 We anticipate
the coefficient to be negative since simpler returns will have a value for Consider
closer to 1. For example, a self-prepared Form 1040EZ taxpayer would have a value
for Consider close to 1 since most of the forms and schedules are irrelevant to this
taxpayer. In contrast, a self-prepared Form 1040 taxpayer who itemizes, has a family, and has a business with corresponding credits would have a value for Consider
closer to zero.
To control for the contribution of specific individual taxpayer characteristics, three additional variables were included in the model. The first variable,
HH/Widow, is a dummy variable and is equal to 1 if the taxpayer’s filing status
is head of household or qualifying widow(er). The second variable, Married, is
14
Contos, Guyton, Langetieg, and Vigil
a dummy variable and is equal to 1 if the taxpayer’s filing status is married filing
jointly. We expect both of these dummy variables to be positively associated with
compliance burden since both of these taxpayer groups tend to have more complicated family and lifestyles than single taxpayers.
Finally, the third variable, log(Exemptions), is the natural logarithm of the number of exemptions claimed by the taxpayer (Form 1040, line 6).15 The number of
exemptions is associated with completion of additional lines on the tax form, such
as child tax credits, earned income credit, etc. It controls for increased compliance
burden typically associated with taxpayers who claim exemptions for household
members who may not be fully captured by the complexity categories.
Total monetized compliance burden was estimated using the following
equation:
16
Log(Burden) = b0 + b1 Log(mpi) i + b2 Lowi + b3 Medium_Lowi + b4 Mediumi +
b5 Highi + b6 Paidi + b7 Softi + b8 Paid_mpii + b9 Consideri + b10 HH/Widowi +
b11 Marriedi + b12 Log(Exemptionsi ) + εi (1)
where the letter i indexes the taxpayer.
Simulation Issues
The individual taxpayer population is highly diverse and covers returns in a large
range of modified positive income; Table 3 shows the ratio of average burden
to modified positive income by percentile range. It is clear that the survey data
are skewed with a heavy tail. Taxpayers in the lowest decile have average mpi of
$6,237, and their burden represents 2.2 percent of mpi; those in the highest decile
have average mpi of $1.7 million and average burden of 0.6 percent of mpi, indicating that lower income taxpayers experience a larger share of burden, measured as
a percentage of their income.
TABLE 3. 2007 Average Individual Income Tax Compliance Burden as a Percentage
of Modified Positive Income
Decile
Average mpi
Average Burden as a
Percentage of mpi
0 to 10
10 to 20
20 to 30
30 to 40
40 to 50
50 to 60
60 to 70
70 to 80
80 to 90
90 to 100
$6,237
$13,209
$20,270
$27,874
$36,295
$47,164
$61,849
$83,279
$124,541
$1,716,546
2.2%
1.6%
1.2%
1.1%
1.0%
0.9%
0.9%
0.8%
0.8%
0.6%
Individual Taxpayer Compliance Burden
15
Our log-linear regression specification addresses the inherent skewness in
the compliance burden data as recommended by Manning and Mullahy (2001).
Although there are a variety of alternative functional forms to address skewness, a
Box-Cox test for the optimal transformation of the dependent variable confirmed
logarithmic transformation as the best option.17 Although both the survey and
administrative data were cleaned and standardized early in the process, there was
still concern that outliers could affect the robustness of the model. The detection
of potential outliers was of particular interest, since the survey required respondents to recall the intensity of the various activities performed and to isolate and
report only the activities they incurred as a result of the Federal tax system.
Given the complexity of the multivariate outlier detection process, robust regression was used to identify and adjust the weights of observations with reported
values furthest from the initial regression line. Robust regression is an iterative
process that reduces the importance of observations with large residuals by lowering their weights (based on a weight function) and then re-estimates the regression with the new weights, repeating the process until it converges.
Since total monetized compliance burden was transformed into logs, an important issue was how to accurately transform the estimates back to levels. In a
standard regression model, the error (ε) is ignored when predicting the value of
the dependent variable. However, when one retransforms the dependent variable
in a log-linear regression specification, the level of the dependent variable depends
on the value of the anti-log of the error term (exp{ε}). In general, the contribution
of this non-linear function of the error term cannot be ignored when predicting
the level of the dependent variable. In a model where the error is heteroskedastic,
this process becomes more complicated. In addition, since the model’s objective is
to support tax policy-making through “what-if ” analysis, the model needs to perform satisfactorily in estimating compliance burden for subgroups of the business
population and across the overall population distribution. All these issues led us
to use a number of statistical techniques that improved the representativeness of
the model across the entire population. The technical aspects of these techniques
are discussed in detail in Contos et al. (2009b).
Estimated Coefficients
Robust OLS regression results for both survey collections are presented in Table 4.
Estimated coefficients for log(mpi) are positive and less than 1 as expected, 0.491
and 0.439, respectively, and are significant at the 1-percent level, implying that as
income increases, burden increases but at a decreasing rate. The coefficients are of
similar magnitude for both years, but the 2007 coefficient is lower—perhaps partially confirming our hypothesis that technological improvements have reduced
burden.
16
Contos, Guyton, Langetieg, and Vigil
All the coefficients for the complexity categories are positive and statistically
significant at the 1-percent level. The coefficients for Low are 0.009 and 0.005;
for Medium-Low they are 0.009 and 0.008; for Medium, they are 0.012 and 0.013;
and for High, they are 0.023 and 0.014. Positive coefficients imply that increases
in complexity and the volume of an activity increase total burden. The magnitudes of these coefficients confirm the construction of the complexity categories.
A dollar increase in a medium complexity item, holding all else constant, will increase burden more than a dollar increase in a low complexity item. Again, notice
that although the coefficients are of similar magnitude, the 2007 coefficients are
lower for three of the four complexity categories. As discussed earlier, this may
be confirming our hypothesis of reduced burden due to technological improvement. However, differences in the underlying data and limitations in our ability
to control for differences between the surveys may also be contributing to these
differences.
TABLE 4: Regression Results*
Variable
Survey 2000 Coefficients
Estimate
T-Stat
Survey 2007 Coefficients
Estimate
T-Stat
Intercept
0.247
1.53
1.163
4.91
Log (mpi)
0.491
35.62
0.439
26.01
Low Complexity
0.009
11.59
0.005
5.69
Medium-Low Complexity
0.009
14.15
0.008
8.97
Medium Complexity
0.012
18.86
0.013
13.63
High Complexity
0.023
22.46
0.014
10.39
Paid Professional Prepared
Return
1.843
9.87
1.299
4.74
Self Prepared Return Using
Software
-0.558
-8.39
-1.025
-7.23
Log (mpi) and Paid Professional
Prepared Return
-0.224
-14.84
-0.178
-9.15
Consider
-1.556
-17.91
-1.697
-9.26
Head of Household or Widow
-0.047
-1.48
-0.013
-0.29
Married
-0.270
-8.68
-0.306
-6.57
Log (Exemptions)
0.142
5.41
0.186
5.03
Adj. R-Squared
0.603
0.574
*T-statistics in bold are statistically significant at the one percent level.
The coefficients for returns prepared by paid professional are 1.843 and 1.299,
and both are statistically significant at the 1-percent level. This implies that, controlling for the size and complexity of the return, self-prepared returns have lower
fixed costs than paid-prepared returns.
Individual Taxpayer Compliance Burden
17
The coefficients for returns prepared by software are −0.558 and −1.025 and
are statistically significant at the 1-percent level. Although the negative sign for
the software coefficient is counterintuitive, it should be interpreted in the context
of the Consider variable. For example, the 2007 coefficient for Consider is negative, −1.697, as expected, and statistically significant at the 1-percent level. Because
a self-preparer filing Form 1040EZ does not consider, on average, 95 percent of
the lines on all individual tax forms, this taxpayer’s log(Burden) is reduced by
−1.697 * 0.95, or −1.61215. Similarly, a W&I taxpayer that self-prepares and files
Form 1040 does not consider, on average, 58 percent of the lines on all the tax
forms, reducing his log(Burden) by −1.697 * 0.58, or −0.98426.18 This implies that a
taxpayer filing a 2007 self-prepared Form 1040EZ return has lower fixed costs, and
total monetized burden, than an otherwise similar taxpayer using software. While
a taxpayer filing a 2007 self-prepared Form 1040 return has higher fixed costs,
and total monetized burden, than an otherwise similar taxpayer using software.
Overall, lower 2007 coefficients for paid preparer and software returns indicate
that the cost of software and professional preparation services has been declining
in real terms.
The coefficients for Paid-mpi are negative, −0.224 and −0.178, and are significant at the 1-percent level. As expected, as income increases, the burden associated with returns prepared by paid professionals increases at a lower rate than
returns that are self-prepared. This implies that, although paid prepared returns
have higher fixed costs than self- and software-prepared returns, as mpi increases
above a certain level, the lower marginal cost leads to lower total monetized burden for paid prepared returns. If we combine log(mpi) and Paid-mpi to estimate
the growth rate of burden associated with paid-prepared returns, we see that the
effective coefficients for 2000 and 2007 remain virtually the same, 0.267 and 0.261.
As discussed earlier, additional costs associated with software- and paid-
prepared returns may reflect some combination of self-selection or demand for
a different quantity or quality of services above and beyond the impact of lower
marginal costs for paid-prepared returns, along with more effective handling by
software of complex returns as suggested in the model.
The coefficients for HH/Widow are negative, but small and not statistically significant. The negative sign is counterintuitive implying that after controlling for
size, complexity, preparation method, etc., single taxpayers have higher burden
than head of household or qualifying widowed taxpayers. This coefficient can
be interacted with log(Exemptions), which has a positive coefficient of 0.186, and
is statistically significant at the 1-percent level. If we combine the coefficients of
the two variables, holding all else constant, we see that a taxpayer filing head of
household with one child has a final coefficient of 0.116 (−0.013 + 0.186 * log (2)
exemptions) whereas a single taxpayer has, by construct, zero effect.
18
Contos, Guyton, Langetieg, and Vigil
The coefficient for Married is negative and statistically significant at the 1-percent level. Again, the sign is counterintuitive, but if considered in conjunction
with log(Exemptions), the model shows that married taxpayers have higher burden
than single taxpayers. Similarly, the model shows that married taxpayers without children have lower burden than those that file head of household with one
child. Another possible explanation is that married taxpayers tend to have more
tax preparation and filing experience. A detailed discussion on how to estimate
burden using 2000 and 2007 coefficients is included in Appendix B.
Findings
In this section, we present our findings and suggest some interpretations of the
results. Table 5 shows the estimated distribution of monetized burden for the
8-year period covered by this study. From Tax Years 2000 to 2007, total monetized
burden is estimated to have increased for the bottom 80 percent of the burden
distribution and decreased significantly for the top 20 percent. This suggests that
taxpayers that filed more complicated returns, such as the self-employed, have
benefited disproportionately from productivity gains, better integration of recordkeeping and tax software, because of their disproportionate usage of assisted
methods. For the lower four-fifths of the distribution, estimated burden actually
increased. This is consistent with changes in the demographics of the taxpayer
population, resulting in the filing of more complicated tax returns, as well as increased usage in assisted preparation methods and the fixed costs associated with
these preparation methods.
TABLE 5. Distribution of Monetized Burden, by Tax Year
Decile
2000
2001
2002
2003
2004
2005
2006
2007
95%
2128
2013
1952
1934
1883
1910
1917
1842
90%
1235
1194
1171
1157
1145
1154
1170
1134
75%
531
536
535
533
537
546
559
548
Median
217
228
232
237
242
248
258
256
25%
86
94
97
102
107
111
116
118
10%
34
39
41
45
48
51
53
55
5%
19
22
23
26
28
31
31
33
Table 6 presents estimates of taxpayer burden using blended coefficients for Tax
Years 2000 through 2007 in constant 2007 dollars.
Although it is widely accepted that tax system complexity has increased during the 8-year period we studied, we estimate that the average time burden per
taxpayer declined from 23 hours in Tax Year 2000 to 19 hours in Tax Year 2007, a
17.4-percent decline. We similarly estimate that the average constant dollar money
19
Individual Taxpayer Compliance Burden
burden increased from $220 per taxpayer in Tax Year 2000 to $258 in Tax Year
2007, an increase of 17.3 percent. Taken together, we see evidence of the trade-off
between time and money. Average monetized burden is estimated as having decreased 7.5 percent in constant dollars from $652 in Tax Year 2000 to $603 in Tax
Year 2007. Finally, total aggregate monetized burden for all taxpayers is estimated
to have grown in constant dollars from $84.3 billion in Tax Year 2000 to $86 billion
in Tax Year 2007—an increase of 2 percent—despite a much larger increase in the
size of the filing population. We see evidence that, after adjusting for productivity, average and total monetized burden remained relatively constant in real dollar
terms over this period and may have even decreased, despite increased complexity.
TABLE 6. Individual Taxpayer Burden by Tax Year: Using Blended 2000 and 2007
Coefficients
Variable
2000
2001
2002
2003
2004
2005
2006
23
22
21
20
20
20
20
19
Average Money**
$220
$215
$215
$233
$232
$248
$257
$ 258
Average Monetized
Burden**
$652
$614
$601
$603
$588
$610
$615
$ 603
Total Monetized
Burden***
$84.3
$79.8
$78.0
$78.7
$77.4
$81.7
$85.1
$86.0
Average Time
2007*
* The 2007 total monetized burden estimate does not include 10.6 million stimulus only taxpayers that accounts for
approximately $1.08 billion of additional burden.
** In constant 2007 dollars.
*** In billions of constant 2007 dollars.
Table 7 presents per capita individual taxpayer burden by tax year. An advantage of estimating per capita burden is that the measure is less sensitive to temporary changes in the filing population due to one-time events such as the telephone
excise tax refund and the economic stimulus payment. Again, we see evidence of
the trade-off between time and money. Per capita time burden declined from 10.5
hours in 2000 to 9.2 hours in Tax Year 2007. Over the full time frame, time burden
declined by just over 12 percent. In contrast, average money burden increased by
nearly 22 percent from $101 in Tax Year 2000 to $123 in Tax Year 2007. Average
monetized burden declined over the 8-year period from $299 in Tax Year 2000 to
$289 in Tax Year 2007.
Finally, Table 8 presents estimates of taxpayer burden using blended coefficients for Tax Years 2000 through 2007 in nominal dollars. As expected, average
monetized burden increased for all but three years: 2001, 2002, and 2007, whereas
total monetized burden decreased in 2001 and 2002 and then increased for all
subsequent years.
20
Contos, Guyton, Langetieg, and Vigil
TABLE 7. Per Capita Individual Taxpayer Burden by Tax Year: Using Blended 2000
and 2007 Coefficients
2000
2001
2002
2003
2004
2005
2006
Average Time
Variable
10.5
9.8
9.5
9.2
8.8
9.0
9.0
2007
9.2
Average Money*
$101
$98
$97
$105
$104
$112
$119
$123
Average Monetized
Burden*
$299
$280
$271
$271
$264
$276
$285
$289
* In constant 2007 dollars
TABLE 8. Individual Taxpayer Burden in Nominal Dollars by Tax Year: Using Blended
2000 and 2007 Coefficients
2000
2001
2002
2003
2004
2005
2006
2007*
Average Monetized
Burden
Variable
$541
$525
$522
$535
$536
$574
$614
$603
Total Monetized
Burden**
$70.0
$68.1
$67.7
$69.8
$70.5
$76.9
$85.0
$86.0
* The 2007 total monetized burden estimate does not include 10.6 million stimulus only taxpayers that accounts for
approximately $1.08 billion of additional burden.
** In billions of dollars.
Conclusions
In this paper, we reported estimated differences in compliance burden for Tax
Years 2000 and 2007. In doing so, we considered the effects on individual taxpayer
compliance burden resulting from changes in tax return complexity, technological
and related productivity changes, and increased use of assisted tax preparation and
filing methods—since these factors changed dramatically during this time period.
With these insights we presented one plausible interpolation of compliance burden for the intermediate years.
Our results suggest that average real monetized compliance burden may have
declined by as much as 7.5 percent in constant dollars from $652 in Tax Year 2000
to $603 in Tax Year 2007. More conservatively, this result and the related analysis suggests that compliance burden did not materially increase over this period
despite increasing complexity of the tax law, economic activity, and demographic
characteristics. This interpretation is at least partially corroborated by the New
Zealand Inland Revenue finding that real dollar compliance costs for New Zealand
small businesses decreased by 1.3 percent between 2004 and 2009, despite increases in the complexity of New Zealand tax law.19 The conceptual framework we
outline in this paper is expected to assist the IRS in its compliance burden forecasts
for policymakers and the public. Other explanations for the differences in reported compliance burden between 2000 and 2007 include sampling, measurement,
and modeling error. It will take additional compliance burden surveys over time
to more definitively disentangle some of these competing explanations.
Individual Taxpayer Compliance Burden
21
Although we estimate average monetized burden as having decreased 7.5 percent in constant dollars from Tax Year 2000 to Tax Year 2007, we do not find this
decrease uniformly over the population. In particular, the overall average decrease
is primarily attributable to a significant decrease in burden for the top 20 percent
of the burden distribution. This suggests that increasing complexity is nonetheless imposing significant costs on the public and is likely one of the factors driving
increasing use of assisted methods. We expect to continue to examine the drivers
of compliance costs and the implications for tax administration in future studies
as we collect more data on the subject.
Acknowledgements
The authors wish to acknowledge comments and assistance from Lauren Craigie,
Ardeshir Eftekharzadeh, Sebastian Eichfelder, Allen Lerman, Sandy Lin, Rachel
Maguire, Karen Masken, Susan Nelson, Jim Nunns, Mark Payne, Alan Plumley,
Stephen Rhody, and Eric Toder. Any errors are solely the responsibility of
the authors.
References
Abramowitz, Milton and Irene A. Stegun, eds. (1972), Handbook of Mathematical
Functions with Formulas, Graphs, and Mathematical Tables, New York: Dover,
ch. 7.
Allingham, Michael and Agnar Sandmo (1972), “Income Tax Evasion: A
Theoretical Analysis.” Journal of Public Economics 1 323–338.
Allison, Paul D. (2007), Missing Data Workshop, New York.
Andreoni, James; Brian Erard; and Jonathan Feinstein (1998), “Tax Compliance.”
Journal of Economic Literature. Vol. XXXVI 818–869.
Auerbach, David; Janet Holtzblatt; Paul Jacobs; Alexandra Minicozzi; Chapin
White, Congressional Budget Office; and Pamela Moomau, Joint Committee
on Taxation (2010), “Would Health Insurance Mandates Increase Coverage?
Synthesizing Perspectives from Health, Tax, and Behavioral Economics.” The
National Tax Association 40th Annual Spring Symposium.
Brick, Michael; George Contos; Karen Masken; and Roy Nord (2009), “Response
Mode and Bias Analysis in the IRS’s Individual Taxpayer Burden Survey,” 2009
Joint Statistical Meeting (forthcoming).
Carr, Karmen W. (2010),“Evaluation of Taxpayer Decision Points,” a Report for
the Internal Revenue Service Wage & Investment Division. Concentrance
Consulting Group.
22
Contos, Guyton, Langetieg, and Vigil
Contos, George; John Guyton; Patrick Langetieg; and Susan Nelson (2009),
“Taxpayer Compliance Costs For Small Businesses: Evidence From
Corporations, Partnerships, and Sole Proprietorships.” Proceedings of the
2009 National Tax Association Meetings (forthcoming).
Contos, George; Ardeshir Eftekharzadeh; Brian Erard; John Guyton; and Scott
Stilmar (2009), “Econometric Simulation of the Income Tax Compliance
Process for Small Business.” Proceedings of the 2009 Winter Simulation
Conference.
Erard, Brian and Chih-Chin Ho (2003), “Explaining the U.S. Income Tax
Compliance Continuum.” eJournal of Tax Research, Volume 1, Number 2.
Gupta, A. and John O’Hare (2000), Practical Microsimulation Models. In
“Contributions to Economic Analysis: Microsimulation Modeling in
Government,” North-Holland.
Guyton, John, L.; John F. O’Hare; Michael P. Stavrianos; and Eric J. Toder (2003),
“Estimating the Compliance Cost of the U.S. Individual Income Tax.” National
Tax Journal, 673–688.
IRS Small Business Self-Employed Division (2008), “Differences in Return
Characteristics When Tax Preparers Switch from Self-Preparing to a Paid
Preparer.”
Johnston, Jack. and John Dinardo (1984), “Econometric Methods.” 3rd Edition,
New York: McGraw-Hill, 303.
Karlinsky, Stewart; Hughlene Burton; and Cindy Blanthorne (2004), “Perceptions
of Tax Evasion as a Crime.” eJournal of Tax Research, Volume 2, Number 2.
Lerman, Allen (2007), “The Federal Individual Income Tax Form: How Does Its
Appearance Matter?” 100th Proceedings of the National Tax Association.
Manning, Willard G., and John Mullahy (2001), “Estimating log Models: to
Transform or not to Transform?” Journal of Health Economics, Vol. 20,
461–494.
Mazur, Mark J., and Alan H. Plumley (2007), “Understanding the Tax Gap.”
National Tax Journal, 569–576.
New Zealand Inland Revenue Office of Evaluation Services (2010), “Changes in
SME Tax Compliance Costs 2004 to 2009: Evaluation Report 2.
The President’s Advisory Panel on Federal Tax Reform (2005), “Simple, Fair, and
Pro-Growth: Proposals for Fixing America’s Tax System.”
Rosen, Harvey S. (2002), Public Finance, 6th ed. New York: McGraw-Hill Irwin.
Slemrod, Joel and Marsha Blumenthal (1993), “The Income Tax Compliance Cost
of Big Business.” Tax Foundation.
23
Individual Taxpayer Compliance Burden
Slemrod, Joel (1985), “The Return to Tax Simplification: An Econometric
Analysis.” NBER Working Paper No. 175.
Slemrod, Joel and Nikki Sorum (1984), “The Compliance Cost of the U.S.
Individual Income Tax System.” NBER Working Paper No. 144.
Slemrod, Joel, and Varsha Venkatesh (2002), “The Income Tax Compliance Cost
of Large and Mid-Size Businesses,” A Report for the Internal Revenue Service
Large and Mid-Sized Business Division.
Slemrod, Joel and Shlomo Yitzhaki (1985), “The Optimal Size of a Tax Collection
Agency.” NBER Working Paper No. 175.
Appendices
Appendix A
Form
Line
F1040
7
Wages
Short Description
Low
Complexity
F1040
8.1
Taxable Interest Income
Low
F1040
8.2
Tax Exempt Interest
Medium/Low
F1040
9.1
Taxable Dividends Amount
Medium/Low
F1040
9.2
Qualified Taxable Dividends Amount
Medium
F1040
10
State Income Tax Refund
Low
F1040
11
Alimony Received
Medium/Low
F1040
14
Supplemental Gains/Losses
Medium
F1040
15.1
Gross IRA Distributions
Medium/Low
F1040
15.2
Taxable IRA Distribution
Medium
F1040
16.1
Gross Pension Annuity Amount
Medium/Low
F1040
16.2
Taxable Pension/Annuity Amount
Medium
F1040
19
Taxable Unemployment Compensation
Low
F1040
20.1
Gross SS Amount
Medium/Low
F1040
20.2
Taxable Social Security
Medium
F1040
21
Other Income
Medium
F1040
23
Educator Expenses
Medium/Low
F1040
24
Reservist and Other Business Expense Amount
Medium
F1040
25
Health Savings Account Deduction Amount
Medium
F1040
26
Moving Expense Amount
Medium
F1040
28
SE Retirement Plans Deduction
Medium/High
F1040
29
SE Health Insurance Deduction
Medium/High
F1040
30
Penalty on Early Savings Withdrawal
Medium/Low
F1040
31.1
Alimony Paid
Medium/Low
F1040
32
IRA Payment
Medium/Low
F1040
33
Student Loan Interest Deduction
Medium/Low
F1040
34
Tuition and Fees Deduction Amount
Medium/Low
24
Contos, Guyton, Langetieg, and Vigil
Form
Line
F1040
35
Domestic Production Activity Deduction
Short Description
Medium
Complexity
F1040
36
Other Adjustments
Medium
F1040
36
MSA Deduction
Medium/Low
F1040
36.2
Jury Duty Pay Deduction
Medium/Low
F1040
45
AMT
Medium
F1040
47
Child/Dependent Credit Amount
Medium/Low
F1040
48
Credit for the Elderly of Disabled
Medium
F1040
50
Education Credit
Medium/Low
F1040
51
Foreign Tax Credit
Medium
F1040
52
Child Tax Credit Amount
Medium/Low
F1040
53
Retirement Savings Contribution Credit
Medium/Low
F1040
54
Adoption Credit Amount
Medium/Low
F1040
54
Mortgage Certificate Credit
Medium/Low
F1040
55
Residential Energy Credit
Medium
F1040
55
Other Statutory Credits
Medium
F1040
55
DC First Time Homebuyer Credit
Medium/Low
F1040
58
Combined SE Tax
Medium
F1040
61
Advance EIC Amount
Medium/Low
F1040
62
Schedule H Combined Household Employment Tax
Medium
F1040
63
Accumulation Distribution Tax (ADT)
Medium
F1040
64
Withholding Tax Amount
Low
F1040
65
Estimated Tax Credit
Medium/Low
F1040
66.1
EIC
Low
F1040
66.2
Nontaxable Combat Pay Amount
Low
F1040
67
Excess SS Tax and RRTA Tax Withheld
Low
F1040
68
Additional Child Tax Credit Amount
Medium/Low
F1040
69
Request for Extension Payment
Medium/Low
F1040
70
Health Coverage Credit
Medium
F1040
72
Total Payments
Low
F1040
72
Phone Excise Tax Credit
Medium
F1040
73.1
Balance Due / Refund Amount
Low
F1040
75
Credit Elect Amount
Medium/Low
F1040
77
Estimated Tax Penalty
Medium
F6251
4
Home Mortgage Interest Amount
Medium
F6251
9
Investment Interest Expenses
Medium/High
F6251
11
Net Operating Loss
High
F6251
14
Incentive Stock Options Amount
High
F6251
15
Benefit of Estate and Trust Amount
Medium/High
F6251
17
Adjusted Gain/Loss
Medium
F6251
18
Depreciation on Assets
High
F6251
19
Passive Activities Amount
High
F6251
27
Other Adjustment Amount
High
F6251
28
Alternative Tax Net Operation Loss
High
25
Individual Taxpayer Compliance Burden
Form
Line
F6251
33
Foreign Tax Credit
Short Description
Medium
Complexity
Sch A
4
Total Medical Deduction
Medium
Sch A
5
State and Local Income Tax
Medium/Low
Sch A
6
Real Estate Tax
Medium/Low
Sch A
9
Taxes Deduction
Low
Sch A
10
Financial Home Mortgage Interest Amount
Medium/Low
Sch A
11
Personal Seller Home Mortgage Interest Paid
Medium/Low
Sch A
12
Deductible Points
Medium/Low
Sch A
14
Investment Interest Paid
Medium/High
Sch A
15
Total Interest Deduction
Low
Sch A
17
Other Contributions
Medium/Low
Sch A
18
Carryover Contributions
Medium/High
Sch A
19
Total Contributions Deduction
Low
Sch A
20
Total Casualty Theft Loss
Medium
Sch A
24
Gross Limited Miscellaneous Deduction
Medium
Sch A
28
Other Non Limited Miscellaneous Deductions
Medium
Sch B
3
Excludable Saving Bond Interest
Medium
Sch C
2
Returns and Allowances
Medium/High
Sch C
3
Net Gross Receipts
Medium/Low
Sch C
4
Cost of Goods Sold
Medium/High
Sch C
9
Car and Truck Expense
Medium/High
Sch C
13
Depreciation
High
Sch C
16
Mortgage Interest
Medium
Sch C
17
Legal and Professional Services
Medium
Sch C
21
Repairs and Maintenance
Medium/High
Sch C
24
Travel Expense
Medium
Sch C
26
Wages Expense
Medium
Sch C
27
Other Expenses
Medium
Sch C
30
Business Use of Home Expense
Medium/High
Sch D
7
Net Short Term Gain/Loss
Medium
Sch D
7
Net Short Term Gain/Loss (Post May)
Medium
Sch D
13
Capital Gains Distribution
Medium/Low
Sch D
15
Net Long Term Gain/Loss (Post May)
Medium
Sch D
15
Net Long Term Gain/Loss
Medium
Sch D
18
Sch D 28 percent Gain
High
Sch D
18
Sch D 28 percent Gain (Post 2003)
High
Sch D
19
Unrecaptured Section 1250 Gain
High
Sch E
3
Total Rents Received
Medium
Sch E
4
Total Royalties Received
Medium/High
Sch E
12
Mortgage Interest Amount
Medium
Sch E
19
Rental / Royalty Deduction
Medium
Sch E
19
Rental / Royalty Deduction
Medium
Sch E
20
Rental Depreciation
Medium/High
26
Contos, Guyton, Langetieg, and Vigil
Form
Line
Sch E
24
Rents and Royalties Income
Short Description
Medium/High
Complexity
Sch E
25
Rents and Royalties Loss
Medium/High
Sch E
30
Partnership/S-Corp Income
Medium/High
Sch E
35
Estate and Trust Income
Medium
Sch E
36
Estate and Trust Loss
Medium/High
Sch F
2
Cost of Purchased Item Cash
Medium
Sch F
5
Gross Co-op Distribution Cash
Medium
Sch F
6
Gross Agriculture Program Payments Cash
Medium
Sch F
7
CCC Loans Forfeited Cash Amount
Medium
Sch F
8
Gross Crop Insurance
Medium
Sch F
9
Custom Hire Cash
Medium
Sch F
15
Machine Hire Expense
Medium
Sch F
21
Gas, Fuel, Oil Deduction
Medium/High
Sch F
23
Mortgage Interest Amount
Medium/Low
Sch F
27
Repairs and Maintenance Expense
Medium/High
Sch F
31
Supplies Purchased Expense
Medium
Appendix B: Technical Appendix
Since total monetized compliance burden was transformed into logs for purposes
of regression analysis, the estimates had to be retransformed back to levels. As
discussed extensively in Contos et al. (2009b), this is not a trivial exercise. In a
standard regression model, the error term (ε) has a mean of zero and is thus ignored when predicting the dependent variable. However, when one retransforms
the dependent variable in a log-linear regression specification, the dependent variable depends on the value of the anti-log of the error term, (exp(ε)). In general, the
contribution of this non-linear function of the error term cannot be ignored when
predicting the level of the dependent variable. To illustrate, consider the log-linear
specification:
ln(Yi ) = β’Xi + εi , (1)
where i indexes observations, Xi is a column vector of explanatory variables, β is a
column vector of coefficients, and εi (conditional on Xi ), is a normally distributed
error term with zero mean. In this specification, the natural log function has been
used to transform the dependent variable, Yi . As in a standard regression, the
mean of our transformed dependent variable is equal to β’Xi . However, when we
retransform this specification to obtain the level of Yi , we obtain:
Yi = exp{β’Xi } exp{εi |Xi } (2)
Therefore, the conditional expectation of Yi given Xi may be computed as:
Individual Taxpayer Compliance Burden
27
E(Yi |Xi ) = exp{β’Xi } E(exp{εi |Xi }) (3)
Although E(εi |Xi ) is zero, the value of E(exp{εi |Xi }) is a nonlinear function of
the error variance.
An additional complication was that the errors were heteroskedastic, indicating
that the expectation of the anti-log of the error was no longer constant. To account
for heteroskedasticity, we assumed that the conditional distribution of Yi given the
explanatory variables Xi was normal. As such, equation (3) simplified to:
E(exp{Yi |Xi }) = exp{β’Xi }exp{1/2σ2i } (4)
The first term is estimated by replacing β with its regression estimate. However,
the second term requires estimation of the variance of the error term (σ2i ) for each
observation in our sample. To address this problem, we defined and estimated a
parametric specification for the variance of the error term. The detailed steps are
as follows:
A. Regress ln(Yi ) on Xi and obtain estimated residuals, εi .
B. Define vi and set it equal to ε2i . Regress v on x and compute the
predicted value (ˆvi ) for each observation.
C. Perform a weighted regression of ln(Yi ) on Xi using 1/ˆvi as a weight
variable. A new weight variable will be created by multiplying the
sample weights by 1/ˆvi.
D. Use the result from step C to compute the predicted linear value
of Y as:
yˆi = exp(β’Xi + ˆvi/2) (5)
where β’Xi uses estimated coefficients from step C and ˆvi is the estimated squared
error from step B.
A simplified method of estimating total monetized burden for Tax Year 2007
using micro level data is to add the estimated coefficients of the weighted regression of ln(Yi) (analogous to step C, above) and one half of the estimated coefficients of the regression of v on x (analogous to step B, above). Coefficient estimates
using data obtained from a survey of Tax Year 2007 taxpayers are presented in
Table 8. As an example, log(mpi) can be calculated as 1.163 + 1.427*0.5 = 1.8765.
Estimates for Tax Year 2000 would be produced similarly, using coefficients presented in Table 9. Estimates can then be blended and adjusted for productivity, as
discussed in the Econometric Model section, to produce estimates for Tax Years
2001 through 2006.
To use the model for tax policy-making through “what-if ” type analysis another issue must be considered. As discussed in Contos et al. (2009b), an undesirable feature of assigning an estimate of the expected taxpayer burden to each
taxpayer in the sample is that it causes the predicted burden values to be much less
28
Contos, Guyton, Langetieg, and Vigil
dispersed than the actual reported values. Given that taxpayer burden is highly
skewed, this approach also causes the median of the predicted burden amounts
to be significantly higher than the median of the reported burden distribution. To
better match the reported burden distribution, we developed a stochastic microsimulation methodology that simulates burden according to the distributional assumptions inherent in our model. The mean of the normal distribution we draw
from is set equal to zero, and the variance is set equal to the estimated variance of
the error term from our regression analysis; since we allow for heteroskedasticity,
the estimated variance varies across observations in our sample. We elected to
repeat this process 30 times for each observation, thereby yielding 30 simulated
values of the dependent variable for each observation in our sample.
TABLE 8. Compliance Burden Coefficients Estimated using Tax Year 2007 Taxpayer
Survey Data
Variable
STEP C
Estimated Coefficients of
Weighted Regression of ln(Yi)
Estimate
T-Stat
STEP B
Estimated Coefficients
of Regression of v on x
Estimate
T-Stat
Intercept
1.163
4.91
1.427
5.31
Log (mpi)
0.439
26.01
-0.076
-3.56
Low Complexity
0.005
5.69
-0.001
-0.75
Medium-Low Complexity
0.008
8.97
0.000
-0.44
Medium Complexity
0.013
13.63
-0.001
-0.50
High Complexity
0.014
10.39
0.003
2.19
Paid Professional Prepared
Return
1.299
4.74
-0.979
-3.29
Self Prepared Return Using
Software
-1.025
-7.23
0.085
0.56
Log (mpi) and Paid Professional
Prepared Return
-0.178
-9.15
0.092
4.19
Consider
-1.697
-9.26
0.194
0.99
Head of Household or Widow
-0.013
-0.29
-0.049
-1.02
Married
-0.306
-6.57
0.015
0.30
Log (Exemptions)
0.186
5.03
0.000
-0.01
Adj. R-Squared
0.574
0.012
29
Individual Taxpayer Compliance Burden
TABLE 9. Compliance Burden Coefficients Estimated using Tax Year 2000 Taxpayer
Survey Data
Variable
STEP C
Estimated Coefficients of
Weighted Regression of ln(Yi)
Estimate
T-Stat
STEP B
Estimated Coefficients of
Regression of v on x
Estimate
T-Stat
Intercept
0.247
1.53
1.283
6.43
Log (mpi)
0.491
35.62
-0.065
-3.69
Low Complexity
0.009
11.59
0.001
1.17
Medium-Low Complexity
0.009
14.15
-0.002
-2.43
Medium Complexity
0.012
18.86
0.001
1.53
High Complexity
0.023
22.46
0.003
2.43
Paid Professional Prepared
Return
1.843
9.87
0.017
0.08
Self Prepared Return Using
Software
-0.558
-8.39
0.202
2.58
Log (mpi) and Paid Professional
Prepared Return
-0.224
-14.84
0.004
0.20
Consider
-1.556
-17.91
0.387
3.77
Head of Household or Widow
-0.047
-1.48
-0.077
-1.96
Married
-0.270
-8.68
0.042
1.14
Log (Exemptions)
0.142
5.41
0.000
0.01
Adj. R-Squared
0.603
0.017
Table 10 compares the distribution of burden as reported in the Tax Year 2007
survey and how it changes following various adjustments. The first two columns
compare the distribution of reported burden with the distribution of the prediction after adjusting for heteroskedasticity (transformation adjustment). Medians
in both columns are significantly lower than the means, reflecting the fact that the
median of a highly right-skewed distribution falls well below the mean of the distribution. In addition, the median of the predictions (332) is higher than the median reported burden (262), whereas the estimated mean (583) is much closer to the
reported mean (554). The final column of Table 10 shows the distribution of the
predicted burden after all adjustments. The predicted median (268) is now much
closer to the reported median (262), while the mean is still slightly over-predicted
(583 vs. 554) by about five percent. As the results indicate, our stochastic microsimulation approach does a much better job of representing the overall distribution of reported burden than the non-stochastic micro-simulation methodology.
While, on average, we over-predict the Tax Year 2007 mean by approximately five percent and the Tax Year 2000 mean by approximately 10 percent, this is
still a substantial improvement compared with the results given an assumption of
homoskedasticity. An assumption of homoskedasticity would have led us to overpredict level burden by about 30 percent. To avoid over-predicting Tax Year 2007
30
Contos, Guyton, Langetieg, and Vigil
level burden by five percent would require us to better fit a model of the variance.
As we see above, even a fairly weak model for the variance substantially improves
our overall prediction of burden. To account for the residual effects of heteroskedasticity that we cannot model, we apply a further correction factor of 5.14 percent
(10.27 percent for Tax Year 2000) on average predicted monetized burden to fit the
reported level average and hence total monetized burden.
TABLE 10. Distribution of Tax Year 2007 Reported Burden and Adjustments
Reported Burden
Predicted with
Transformation
Adjustment
Final Predicted
95%
1,918
1,654
1,871
90%
1,226
1,083
1,182
75% Q3
560
596
572
50% Median
262
332
268
25% Q1
117
206
125
10%
55
125
60
5%
32
85
37
554
583
583
Quantile
Mean
Endnotes
1
2
3
4
5
6
Specifically, a 1-hour increase in time burden was associated with an additional
$119 of noncompliance.
Volunteer Income Tax Assistance and Tax Counseling for the Elderly are
IRS services that offer free assistance with tax return preparation and tax
counseling using specially trained volunteers. These programs assist seniors
and individuals with low to moderate incomes, those with disabilities, and
those for whom English is a second language.
For a qualitative study of this issue, see Carr (2010). The study reports that
monetary cost and ease of preparation are the most often cited factors in
preparation method choice, with complexity primarily being a factor for usage
of a paid preparer.
Ibid.
See ftp://ftp.bls.gov/pub/special.requests/opt/dipts/and http://data.bls.gov/
PDQ/servlet/SurveyOutputServlet?data_tool=latest_numbers&series_id=
PRS85006092.
Modified positive income is defined as the sum of wages and salaries, taxable
and tax-exempt interest, ordinary and qualified dividends, state income tax
refunds, alimony received, net profit reported on Schedule C, capital and other
gains, taxable and non-taxable retirement income (IRA distributions, pensions
Individual Taxpayer Compliance Burden
31
and annuities, social security benefits), gross profits from active participation
in a Partnership or S-Corporation reported on Schedule E, gross farm profits
reported on Schedule F, unemployment compensation, and other income
reported on the tax return.
7
The Chow test statistic is equal to:
where Sc is the sum of squared residuals from the combined data. S1 is the sum
of squared residuals from the first group and S2 is the sum of squared residuals
from the second group. N1 and N2 are the number of observations in each
group, and k is the total number of parameters. The resulting test statistic had
a numerator of 8,310,268 and a denominator 8,064, which resulted in a test
statistic equal to 1,031. Using an F-distribution with 10 parameters and 708,130
(N1 + N2−2K) degrees of freedom gives a level of confidence over 99 percent
that the null hypothesis should be rejected. Based on the results of the Chow
test, we concluded that the alternative hypothesis should be accepted, and two
sets of coefficients were estimated.
8 Blended estimates were calculated using a simple weighting scheme that uses
estimates of the two survey years. To calculate estimates for Tax Year 2001: the
2001 estimates produced using 2000 coefficients were weighted by 6/7ths, the
2001 estimates produced using 2007 coefficients were weighted by 1/7ths; for
Tax Year 2002 estimates, the 2002 estimates produced using 2000 coefficients
were weighted by 5/7ths, the 2002 estimates produced using 2007 coefficients
by 2/7ths, etc.
9 See http://data.bls.gov/PDQ/servlet/SurveyOutputServlet?data_tool=latest_
numbers&series_id=PRS85006092.
10 Estimates were adjusted for year specific productivity: A proxy for technology
was introduced into estimates produced using 2000 coefficients and removed
from the estimates produced using 2007 coefficients. To calculate estimates
for Tax Year 2003, 2003 estimates produced using 2000 coefficients were
multiplied by the change in total productivity between 2000 and 2003, 1.116.
2003 estimates produced using 2007 coefficients were multiplied by 0.930, etc.
11 Another objective was parsimony in trying to avoid over-fitting the data.
We expect to explore extensions to the model in the future as we gather
additional data.
12 To monetize the value of time a wage rate was calculated for each taxpayer by
dividing modified positive income by either 2,080 or 4,160, depending on the
taxpayer’s filing status. For example, the modified positive income of taxpayers
who filed married filing jointly was divided by 4,160 to arrive at a wage rate.
32
Contos, Guyton, Langetieg, and Vigil
Modified positive income of taxpayers with any other filing status was divided
by 2,080. The resulting wage rate was restricted to be at least as large as the
minimum wage rate in order to avoid zero or unreasonably small values.
Separate maximum limits were set for three groups of prefiling and filing
activities. For example, the maximum hourly cost for recordkeeping time was
set equal to the fees charged by professional bookkeepers.
13 More specifically, the low category includes items that are reported on
information returns or require very little recordkeeping. The medium-low
category includes items that are reported at an aggregate level but require some
recordkeeping. The medium category includes items that require additional
recordkeeping and are reported to the IRS separately. Many of the items
included in the medium category require attaching worksheets that document
how the totals were determined. Finally, the high category includes items that
may require a separate recordkeeping system or a process with potentially
separate rules for each item. Tracking records across years is an additional
component for most in this category.
14 Construction and use of this variable was motivated by the discussion in
Lerman (2007) of the design and estimated impact of the Schedule O for
Form 1040.
15 For purposes of calculating this variable, exemptions are not allowed to be less
than 1.
16 Forecasting the logarithm of costs indicates that a change in any of the
explanatory variables is associated with a certain percentage change in
compliance costs, regardless of the initial level of these costs, Slemrod and
Blumenthal (1993).
17 It is worth noting that, following the model selection process described by
Manning and Mullahy (2001), we tested whether a Generalized Linear Model
(GLM) model would perform better than OLS. First, the kurtosis of the logscale residual was calculated from one of the consistent GLM estimators.
Since the kurtosis was less than 3, the Park test was then used to select the
appropriate GLM model. The estimated l was equal to 1.58. If l is equal to
1 (raw-scale variance is proportional to the raw-scale prediction) the Park
test suggests considering a Poisson-like model. If l is equal to 2 (raw-scale
variance is quadratic in the raw-scale prediction) then consider the gamma
model or the homoskedastic log OLS model. All three specifications were tried
and the results were qualitatively similar so the simpler and more efficient OLS
method was selected.
18 Wage & Investment (W&I) taxpayers are those individual taxpayers not filing
Schedule C, Schedule E, Schedule F, or Form 2106, typically covering business,
farm, partnership, rental and royalty income and expenses.
19 See Table 1.2.
Enhancing Compliance
Through Improved Readability:
Evidence from New Zealand’s
Rewrite “Experiment”
Adrian Sawyer,
University of Canterbury
T
ax law complexity is an international phenomenon that is often criticized
but infrequently tackled. United States (US) federal income tax law is no
exception. In 1993, New Zealand (NZ) embarked upon an ambitious project to respond to calls for reducing complexity to stimulate further compliance by
taxpayers through rewriting its income tax legislation. The project was essentially
a reorganization of existing material followed by a progressive rewriting of the
statutory language, with minor policy changes implemented throughout the process. No attempt was made to address the complexity of the underlying concepts,
yet concurrent with the rewrite project, legislative amendments and new policy
initiatives (including administrative simplification measures and social policy developments) had to be incorporated. The rewrite project, originally intended to
take 5 years, took 15 years and considerable expense to achieve (Sawyer, 2007). To
an extent this project was part of a larger experimental exercise that both Australia
and the United Kingdom (UK) embarked upon (see James et al., 1998). Australia
appeared to have “given up” part way through, although the Assistant Treasurer
released the rewrite of 149 pages of income tax provisions for consultation in
November 2009. The UK is nearing completion of its rewrite project. Thus the
NZ experiment is the first to be completed and in itself comprised a number of
unique features, including the establishment of the Rewrite Advisory Panel (the
Panel) (Sawyer, 2008).
To put this experiment in its context, the rewrite project was intended to assist
the NZ Government’s simplification aspirations through reducing sentence length
and improved readability of the Income Tax Act (ITA). Initial analysis of the success of the progressive outputs of the projects has been undertaken through employing readability measures such as the Flesch Reading Ease Index (and FleschKincaid Grade Level (FKGL)), and to a lesser extent, the Cloze Procedure (see for
example, Tan and Tower, 1992; Richardson and Sawyer, 1998; Castle, 2006a; Castle,
2006b; Harrison, 2006; Pau et al., 2007; and Sawyer, 2007).
This study provides further empirically tested insights into the success or otherwise of the simplification exercise, primarily through application of the Cloze
32
Sawyer
Procedure to important statutory provisions. Specifically, this study uses the Cloze
Procedure focusing on the extent to which subjects can correctly fill in the gaps.
This study tests undergraduate business students, at the commencement of their
first tax paper (a course with over 250 students) on their understanding of several
sections from the ITA, both as it stood in 1993, prior to the rewrite project commencing, and as it is now written in 2007. This was also tested on advanced level
undergraduate tax majors during their final year tax papers (a course with over
100 students) to ascertain the extent to which reader knowledge, interest in the
subject matter, and other issues not able to be captured in readability measures can
be gauged to have met with success through the rewritten legislation. While the
intended subjects are to be students rather than business taxpayers and tax practitioners, the literature has established (see for example, Richardson and Sawyer,
1998) that the major users of tax legislation are tax practitioners and implicitly, at
least, students studying taxation.
This study enables triangulation of data utilizing several readability measures
(Flesch, FKGL, and Cloze Procedure) applied to four versions of the ITA to assess
the readability and understandability of the ITA against intended NZ Government
policy outcomes for the rewrite project. The NZ evidence is anticipated to provide
an excellent example of the degree of success in an experiment to reduce complexity (and consequently increase simplicity) through the expression of statutory
provisions, and enhance tax compliance.
The remainder of this paper is organized as follows. Section 2 discuses the
relevant literature on readability generally, and in the context of tax legislation,
focusing on developments in NZ supported by comparative efforts undertaken
in Australia and the UK. Then, in section 3, the reasoning behind adopting NZ
as an experimental case study is explained. Section 4 provides an overview of
prior research using Flesch and other readability measures in NZ, and presents the
results of the Cloze Procedure employed in this study. Finally, section 5 sets out
the conclusions from combining the results of this study with those of prior NZ
readability research, considers a number of policy implications, acknowledges the
limitations of readability research, and suggests areas for future research.
Prior Research on Readability Measures and Tax
Legislation
Complexity and Noncompliance
A number of previous studies have recognized complexity of tax laws as a potential factor in tax noncompliance (see for example, Jackson and Milliron, 1986).
Long and Swingen (1988, p. 132) provide a comprehensive definition of complexity that includes the ambiguity of tax laws; the need for numerous calculations;
Enhancing Compliance Through Improved Readability
33
the frequency of change in the tax laws; the excessive detail in the tax laws, such
as rules and exceptions to the rules; the obligation to keep the records; and taxpayer forms and instructions. In keeping with previous NZ rewrite project studies
(Tan and Tower, 1992; Richardson and Sawyer, 1998; Pau et al., 2007; and Saw and
Sawyer, 2010), this study measures the readability of tax laws and other tax-related
materials as a proxy for the complexity of the NZ income tax legislation.
Previous studies have found mixed results on the effect of complexity on noncompliance. For instance, Richardson and Sawyer (2001, p. 185) note that the mixed
results imply that complexity can have different effects on compliance: complexity
can open up opportunities for both intentional and nonintentional noncompliance; the uncertainty of the complex tax laws may encourage compliance among
risk-averse taxpayers; and complexity can reduce the willingness of taxpayers to
comply with tax laws (implying intentional noncompliance). Although previous
studies have provided mixed evidence, the general conclusion from more recent
evidence is that tax complexity can have a negative effect on compliance, whether
intentional or nonintentional (see for example, McKerchar, 2003, p. 108).
Before moving on, it is important to understand what is meant by compliance.
While there are a number of definitions, for the purposes of this paper compliance
means (Roth et al., 1989, p. 21):
“Compliance with reporting requirements means that the taxpayer
files all required tax returns at the proper time and that the returns
accurately report tax liability in accordance with the Internal
Revenue Code, regulations, and court decisions applicable at the
time the return is filed.”
New Zealand’s Tax Rewrite Project and its International
Context
There are a number of ways to encourage voluntary compliance, one of which is to
simplify the tax law. In relation to NZ’s rewriting its income tax legislation project,
NZ’s Inland Revenue Department (IRD) (2001, pp 4, 7) stated:
“Rewriting the Income Tax Act has always been seen as integral
to increasing voluntary compliance with tax laws. This is because
legislation that is clear, uses plain language and is structurally
consistent should make it easier for taxpayers to identify and
comply with their income tax obligations. …
The rewrite cannot, however, eliminate all the complexity and
inconsistency of tax law because the subject matter is inherently
complex. The challenge is to ensure the complexity results from
34
Sawyer
the concepts rather than from the way the information is presented.
Even then, the least complex way of expressing the concepts should
be found.”
The NZ Income Tax legislation has grown significantly in both complexity and
in size over time. When the initial income tax statute, the Land and Income Tax
Assessment Act 1891, was first enacted, it was only 24 pages long. This is very small
relative to the first major installment of the rewrite project, the Income Tax Act
1994 (ITA 1994), which extended to approximately 1,300 pages when the Taxation
(Core Provisions) Act 1996 was enacted in July 1996. Since then, the income tax
legislation has grown to approximately 2,000 pages with the Income Tax Act 2004
(ITA 2004) (Pau et al, 2007) and 2,850 pages with the Income Tax Act 2007 (ITA
2007) (Saw and Sawyer, 2010).
Much has been written concerning the NZ rewrite project, and it is not the
intention of this paper to reproduce that discussion other than to summarize key
themes and briefly overview the process. The Working Party on the Reorganization
of the Income Tax Act 1976-1993 (the Working Party, 1993) was established and
suggested the income tax legislation (Income Tax Act 1976 (ITA 1976) and Inland
Revenue Department Act 1974 (IRDA 1974)) be rewritten in stages over a period of
4 to 5 years. The New Zealand Government issued a discussion document in 1994
(Birch and Creech, 1994), setting out the approach to the rewrite of the income tax
legislation. The objective of the rewrite process was to make the legislation easier
to understand without changing the effect of the tax laws. The drafting styles used
in the rewrite process include the use of plain language, section headings, subheadings, diagrams, and flowcharts.
Stage One of the rewrite of the income tax legislation was completed in 1994
with the enactment of the ITA 1994, the Tax Administration Act 1994 (TAA 1994),
and the Taxation Review Authorities Act 1994 (TRAA 1994). The ITA 1994 was
reorganized using a new alphanumeric numbering system. Stage Two of the rewrite process involved the rewriting of Part A (Interpretation) and Part B (Core
Provisions) of the ITA 1994. Richardson and Sawyer (1998) found evidence that
the reorganization and the rewrite up to this point were successful in terms of
significantly reducing the average sentence length from 324 words to 53 words
and some improvement in readability. However, the readability of tax legislation,
as indicated by the Flesch Reading Ease Score, had not improved dramatically.
Stage Three of the rewrite resulted in the enactment of the ITA 2004, with further changes made to Parts A and B, as well as the rewritten sections of Parts C
(Income), D (Deductions), E (Timing and Quantifying Rules) and Y (definitions).
Pau et al (2007) reviewed this stage of the process and found further improvement
in readability. The final stage (Stage Four) of the rewrite process involved drafting
Enhancing Compliance Through Improved Readability
35
and enacting the ITA 2007, which incorporates rewriting of Parts F to the end of
the ITA 2004, in addition to all the intermediate amending Acts. Saw and Sawyer
(2010) review this final stage and find further improvements in readability. Thus
the rewrite process of ITA was complete with the enactment of the ITA 2007 with
effect from April 1, 2008.
During the early stages of the rewrite process, the Panel was established in 1995,
consisting of one representative each from the New Zealand Institute of Chartered
Accountants (NZICA), the New Zealand Law Society (NZLS), the NZ Inland
Revenue Department (IRD), and the NZ Treasury. The Panel was chaired for most
of the rewrite project by former President of the NZ Court of Appeal, the Rt. Hon
Sir Ivor Richardson, and its initial brief was to consider and advise on issues arising during the rewriting of the ITA 1994. Sawyer (2008) provides a comprehensive
review of the operation of the Panel. His study demonstrates that the rewrite project would not have been such a successful exercise (in relation to dealing with the
detail and associated drafting issues) without the involvement of the Panel and the
dedication of its members.
That said, the rewrite project is in a sense incomplete with regard to other key
statutes. No effort has been made to apply a similar rewriting process to the TAA
1994 and TRAA 1994, nor to the Goods and Services Tax Act 1985 (GSTA 1985). A
call to rewrite GSTA 1985 from the highest judicial level in NZ has to date “fallen
on deaf ears.” The Hon. Justice Blanchard, a member of the Supreme Court, stated
in 2006 (Blanchard, 2007, p 92) that “… it is to be hoped that once the redrafting
exercise on the Income Tax Act is completed the team will move on to the [GSTA
1985], which is not, and never has been, a user-friendly statute.” The NZ GST is an
important tax, contributing over 20 percent to the NZ Government’s annual tax
revenues, and is extensively utilized for ascertaining tax liability.
In relation to developments in two other countries that embarked upon rewrite
projects (Australia and UK), the outcomes to date differ. James et al. (1998) provide an early review of the three countries’ projects, highlighting the differences
in approach to achieve a common goal of simpler and more understandable tax
legislation. In relation to the UK, Saw and Sawyer (2010) observe that the project
is nearing an end in terms of the installment process for the rewrite of the income
tax legislation. Once finished, the project will be evaluated further before any
further rewriting is considered in relation to other revenue statutes. Likewise, Saw
and Sawyer (2010) comment on the unfortunate situation in Australia, where two
income tax statutes are in operation, the original Income Tax Assessment Act 1936
(ITAA 1936) and the partially rewritten Income Tax Assessment Act 1997 (ITAA
1997). Until late in 2009, it appeared that the suspended rewrite project would
forever leave two statutes, but the process has been revitalized with further draft
legislation under review.
36
Sawyer
Readability measures
Redish and Selzer (1985) identify numerous mathematical formulas that can be
used to measure readability. Readability formulas were originally developed in the
1920s to enable textbook publishers to assign children’s books to the suitable grade
level. Rudolph Flesch developed the Flesch Index in 1948 to measure the readability of adult reading materials such as popular magazine articles (see Flesch, 1979).
While technical writing differs significantly from popular magazine articles, the
Flesch Reading Ease Index has been commonly applied to technical writing, such
as tax instruction booklets and legislation.
Much of the prior research on income tax readability in NZ, as well as Australia
and the UK, has employed the Flesch Reading Ease Index, which is based on word
length and sentence length, and is calculated using the following formula:
Flesch Reading Ease Score = 206.835 – 0.846wl – 1.015sl
where:
wl = number of syllables per 100 words
sl = average sentence length in words.
In the late 1940s, Edgar Dale and Jeanne Chall developed the Dale-Chall
Formula for adults and children above 4th grade as a way to improve upon the
Flesch Reading Ease Formula. This readability formula is rarely used in the context of technical materials and has not been utilized in taxation research.
Smith and Richardson (1999), in addition to using the Flesch Reading Ease
Score, computed the F-KGL Index. The F-KGL, also developed by Rudolph
Flesch, rates a given text based on a US-grade school level. The F-KGL Index is
computed as follows:
F-KGL Index = 0.39 (words/sentence) + 11.8 (syllables/word) – 15.59.
The Gunning FOG Index, developed in the 1940s by Robert Gunning, is a figure
in years of education required to read and understand text and is computed as
follows:
1.
The total number of words is divided by the total number of
sentences to equal the average number of words per sentence;
2. The number of words with more than three syllables is divided
by the total number of words to equal the percentage of difficult
words; and
3. The figures derived in #1 and #2 are totaled, and then multiplied
by 0.4.
Enhancing Compliance Through Improved Readability
37
Other readability formulae include the Coleman-Liau Readability Score, designed by Meri Coleman and T. L. Liau in the 1970s to gauge the understandability
of a text. The formula for the Coleman-Liau Grade Level score is:
CLGL = (5.89 x (AWL / ASL)) - (30 x ANS / ASL) - 15.8
where:
AWL = average word length or number of characters per
word (number of characters divided by the number of words);
ASL = average sentence length in words or average number of words in
sentence (number of words divided by the number of sentences); and
ANS = average number of sentences.
The Bormuth Readability Score is based on a count of characters rather than
syllables per word and words per sentence to determine a score corresponding to
the estimated grade level. It was designed to evaluate more academic documents,
such as school texts. The formula for the Bormuth readability score formula is:
BGL = 0.886593 - (AWL x 0.03640) + (AFW x 0.161911) - (ASL x
0.21401) - (ASL x 0.000577) - (ASL x 0.000005)
where:
AWL = average word length or number of characters per word
(number of characters divided by the number of words);
AFW = average familiar words per word (the number of words in
the original Dale-Chall list of 3,000 simple words divided by the
number of words); and
ASL = average sentence length in words or average number of
words in sentence (number of words divided by the number of
sentences).
The FOG Index, Coleman-Liau Readability Score, and Bormuth Readability
Score have been used on technical materials, but rarely in the area of taxation, and
are thus not considered further in this research. Bormuth used extensive selections of text, ranging in difficulty from first grade to college, covering a wide range
of subject matter. He applied the Cloze Procedure to this text. He also developed
multiple regression equations to predict word length, minimal punctuation unit
length, and sentence length work led to development of the Cloze Procedure in
the late 1960s.
According to Redish and Selzer (1985), readability formulae are an inadequate
measure of the difficulty of technical reading material on the basis that readability
formulae do not take into consideration the content, organization, and layout of
38
Sawyer
the reading material, all of which are major contributing factors to the readability
of the reading material. Other factors that influence readability, including the
frequency of changes in tax laws, the background knowledge of the reader, the interest of the reader, and the use of diagrams and flowcharts, are not accounted for
by such readability measures. Mathematical formulae do not necessarily consider
conceptual difficulty, semantics, reader characteristics, and presentation of the
material such as font size, layout of text, graphics, and tables. Consequently, these
formulae cannot result in an absolute measure of clarity (Smith and Richardson,
1999).
Stephens (2000) provides an excellent overview of what is meant by readability
and how it may be measured. In relation to the Cloze Procedure (which was developed in 1953) she comments (p. 5):
“It was popular because its scoring was objective; it was easy to use
and analyze; it used the text itself for analysis; and it yields high
correlations to other formulas.
The Cloze technique does not predict whether the materials are
comprehensible; it is an actual try-out of the material. It tells you
whether a particular audience group can comprehend the writing
well enough to complete the cloze test. …
Cloze testing has been called a ‘rubber yardstick’ because Cloze
scores reflect both the difficulty of the text and the reader’s abilities
or resources. …
In particular, critics suggest that Cloze is inappropriate for
measuring text or reader’s abilities in languages other than their
native language. The results of close testing reflect the reader’s basic
intuition about the structure and vocabulary of the target language
-- and that does not exist for the language student.”
This last comment is particularly important as a number of students in the tax
classes in which the Cloze Procedure was applied have English as their second
language. Stephens continues in discussing the strengths and weaknesses of readability tests (p. 7):
“THINGS THEY CAN DO
1.
Their primary advantage is they can serve as an early warning
system to let the writer know that the writing is too dense.
They can give a quick, on-the-spot assessment. They have been
Enhancing Compliance Through Improved Readability
39
described as ‘screening devices’ to eliminate dense drafts and
give rise to revisions or substitutions.
2. In some organizational settings, readability tests are considered
useful to show measurable improvement in written documents.
They provide a quantifiable measure of improvement or
simplification.
THINGS THEY CAN’T TELL YOU AND WHY
1.
How complex the ideas are;
2. Whether or not the content is in a logical order;
3. Whether the vocabulary is appropriate for the audience;
4. Whether there is a gender, class or cultural bias;
5. Whether the design is attractive and helps or hinders the
reader;
6. Whether the material appears in a form and type style that is
easy or hard to read.”
Stephens (2000) also suggests combining readability formulae with questionnaires to seek out features of documents that could be altered to make reading
easier, using experts and testing drafts with individuals that correspond with the
intended audience. Readability formula can be used to provide feedback. That is,
after a portion of text is written, an appropriate formula may be applied, the text
revised and then retested to see if it has improved. In part, the progressive rewriting of the ITA in NZ followed this approach through preparation of numerous
drafts of rewritten legislation, revisions released for feedback following submissions, and use of experts in reviewing internal drafting within the IRD.
Green (2001, p. 95), who applies the Cloze Procedure to economics materials,
suggests that the extent to which the Cloze Procedure assesses global coherence
depends upon the deletion strategy employed. Normally the Cloze Procedure is a
means of testing the readability of a passage in which every nth word (usually every
five words) is systematically removed from a passage, and the participants are then
required to insert the missing word.
Guillemette (1989) applies the Cloze Procedure to an IEEE Professional
Standard, observing that while the readability scores from formulas can be
40
Sawyer
converted into grade-level equivalents, the usefulness of these results largely depends on the validity of user assumptions concerning the reading level and background of the target audience. Guillemette (1989, p. 41) concludes:
“The Cloze Procedure is a direct measure of readability
which correlates with other measures such as judgments and
comprehension tests. It is able to distinguish differences in
readability not determinable by readability formulas. This study
demonstrated that the Cloze is sensitive enough to detect differences
in understanding between alternate audiences and in readability
among passages in a text. This type of information can provide
valuable feedback to authors by pointing out readability problems
for revision efforts.”
Smith and Taffler (1992, p. 93) apply the Cloze Procedure to different narratives
used in company annual reports. They conclude:
“The absolute level of Cloze score differed significantly according
to the level of accounting sophistication of the user. This suggests
that understandability is related both to complexity of context and
to education and experience, and constitutes a different measure
to readability indices calculated independently of either context or
user.”
Stevens et al. (1993) compare readability formula with the Cloze Procedure
and comment on why readability formulas are inappropriate measures of adult
reading comprehension and why the Cloze Procedure should be the method of
choice in assessing adult readers. That said, prior research has established that
readability indices can be used to assist in predicting the readability of business
and legal documents. Saw and Sawyer (2010) observe that the same mathematical formulae have been used by the United States (US) Internal Revenue Service
(IRS) to measure the readability of taxation forms and instruction booklets (for
example, Tan and Tower, 1992), which provides further support and justification in
terms of their usefulness for measuring readability (and consequently understandability). Importantly, the results should be considered in conjunction with other
measures of readability through triangulation of results in order to corroborate
and strengthen the research findings.
Limited use of other forms of readability measures, such as the Cloze Procedure,
have been applied in NZ, although greater use has been made of such measures in
Australia. As a consequence, the results from prior Cloze Procedure testing and
Flesch (and FKGL) readability measures, can be compared with other readability
measures to gain a richer picture of the understandability of tax-related materials.
Enhancing Compliance Through Improved Readability
41
Research Method: New Zealand as A Case Study
Experiment
Case Study Approach
At this point, one might ask why would the developments in NZ, a small country with approximately 4.5 million people (1.5 percent size of the United States),
be of interest to other countries, including the United States, and in particular,
the IRS. In the writer’s view, the reasons are numerous (his personal bias accepted). Importantly, the subject matter under consideration (income tax legislation) is universally complex in terms of its concepts and expression in developed
countries, and an attempt to reduce complexity and enhance understandability
has been completed (the first of three experiments in developed common law
countries—the others being Australia and the UK). Furthermore, analysis of the
process adopted in NZ through employing various readability measures suggests
that the efforts of the IRD (including the drafting team), the Panel, and contributions by way of submissions from tax practitioners have led to a commendable
result—more readable (and potentially more understandable) income tax legislation. Furthermore, even if the impact of the rewrite project only produces a
small reduction in legislative complexity and consequently minor improvement in
compliance, this still makes the exercise worthwhile (assuming the benefits exceed
the costs).
Case study research is often maligned and considered to be a nonscientific approach to undertaking research. Notwithstanding this view, case study research
is used extensively in academic enquiry in traditional social science disciplines
as well as practice-oriented fields, with the design and analysis considerations of
prime importance, more so often than the description of events or the scenario
under review (see Yin, 2003).
Specifically, in this paper, I adopt the explanatory case study approach for a
single case set in its context, in relation to a significant event, namely the development and completion of NZ’s tax rewrite project with respect to the ITA. Thus
the paper outlines the major steps in the rewrite project, and since it was conducted in defined stages rather than in one major legislative enactment, each stage
is able to be evaluated through use of appropriate readability techniques. While
the project is complete, enabling researchers to reflect upon its impact, overall it is
too early to assess whether the benefits will outweigh the sizeable compliance and
associated costs incurred in drafting, preparing submissions, re-education of tax
professionals, revising teaching material for students of taxation, and the legacy
effect experienced through litigation based upon prior versions of the relevant legislation. Furthermore, increased compliance costs will be incurred as tax advisers
42
Sawyer
review all transactions and become familiar with the new section locations and
groupings.
In terms of being an experiment, we can, at most, draw preliminary conclusions at this point, in terms of assessing the potential benefits through analyzing
the impact that the new drafting style has had on the readability of the resulting
legislation. In terms of the costs, Sawyer (2007) provides an early assessment, although if there are to be significant benefits that will outweigh the costs, most will
not arise until the longer term, perhaps in the next 5 to 10 years.
Research Method: Application of the Cloze Procedure
In this study, the tool for assessing readability (and potentially understandability)
of NZ’s Income Tax legislation is the Cloze Procedure, applied to selected sections
from four versions of the Income Tax Act: the ITA 1976; the ITA 1994; the ITA
2004 and the ITA 2007. The selection of four key sections was taken from the ITA
2007, which is the latest version of the ITA and one that the subjects, undergraduate taxation students, should generally be familiar with. The four sections covered
key areas of tax residence for natural persons, sources of NZ income, the general
deeming provision for income and exempt income, and the general permission
for allowing deductions. The equivalent section in earlier versions of the ITA for
these four sections was then selected from each of the preceding Acts (the ITA
2004, the ITA 1994 and the ITA 1976). The sections selected for each of the four
versions of the ITA are set out in Table 1.
Table 1: Sections from the ITA used in the Cloze Procedure instrument
Version of ITA / Section
ITA 2007
ITA 2004
ITA 1994
ITA 1976
Income & exempt income
BD 1
BD 1
BD 1
242
General permission – deduction
DA 1
DA 1
BD 2(1)
104
Residence of natural persons
YD 1
OE 1
OE 1
241
NZ-sourced income
YD 4
OE 4(1)
OE 4
243
Four versions of the Cloze Procedure instrument were created, one for each
of the four versions of the ITA. The instrument commenced with an explanatory
cover sheet, indicating amongst other things the purpose of the study and that the
study had Human Ethics approval from the University of Canterbury. The instrument concluded with a request for brief demographic information. A copy of the
Enhancing Compliance Through Improved Readability
43
instrument (without the blanks) for the ITA 2007 is attached as an Appendix to
this paper.
The four versions of the instrument were randomly allocated to students in
the second-year undergraduate class “Introduction to Taxation” at the commencement of their first week of classes (class size over 250) and to students in the thirdyear undergraduate class “Advanced Issues in Taxation” at the commencement of
their first week of classes (class size over 100). Students would typically take the
introductory class in their second or third year of study, with the advanced class
taken a year later. Students were given a limited period of 15-20 minutes (due to
time constraints for teaching) to complete as much of the instrument as they could
and were encouraged to ensure that they completed the demographic information.
Upon receipt of the completed instruments, the data was inputted by a research
assistant, from which basic statistical results would be generated.
Overview of Prior New Zealand Research
Findings and Results of This Study
Prior New Zealand Research using Flesch and FKGL
Formulae
Prior NZ studies (see, for example, Saw and Sawyer, 2010) have employed both the
Flesch Reading Ease Index and FKGL Index on the Income Tax legislation resulting from the rewrite project, along with testing these measures on Tax Information
Bulletins (TIBs) prepared by the IRD, and binding rulings issued by the IRD. In
providing an overall analysis of the rewrite project through use of the Flesch and
FKGL indices, Saw and Sawyer (2010) include a series of tables with the Flesch and
FKGL results over the complete period of the rewrite project. They suggest that
the results highlight the success of the NZ rewrite project in terms of its goal of
tax simplification as measured through readability measures. Through enacting
the ITA 2007 average sentence length has been reduced from over 135 words per
sentence generated in Tan and Tower’s (1992) study (with some sentences exceeding 300 words) to approximately 25 words per sentence.
The results of Saw and Sawyer (2010) also depict a remarkable improvement
in the average Flesch Reading Ease Score, which indicates that the income tax
legislation should now be more readable. The scores indicate that a university
undergraduate should be able to read and understand most of the sections in the
ITA 2007. Although there are still a number of sections that exceed the suggested benchmark of 30 words per sentence set by the NZ Government (Birch and
Creech, 1994), this can be attributed to the nature of the complexity of the underlying concepts, a situation that is inevitable with income tax legislation in almost any
jurisdiction. Table 2 reproduces Table 5 from Saw and Sawyer (2010) and provides
44
Sawyer
an overview of the Flesch Reading Ease Score results for the various stages of the
NZ rewrite project, and a comparison with Australia.
Saw and Sawyer (2010), like Pau et al. (2007), also observe that in comparing
the readability of the ITA 2007 with other tax related materials, the average Flesch
Reading Ease Score is higher for the ITA 2007 and binding rulings, suggesting
they are easier to read and understand compared to the IRD’s TIBs. This finding
suggests that the drafters of TIBs should re-examine the drafting style adopted
Table 2: Summary of Flesch Reading Ease Scores (Income Tax Legislation)
New Zealand
Flesch
Reading Ease
Score
2007
2004
Australia
1994
1976
1997
Education level
General
Reading Ease
Scale
#
%
#
%
#
%
#
%
#
%
Below 30
35
18
16
20
21
80.7
40
100
11
12
University
Graduate
Very Difficult
30-50
92
48
50
61
2
7.7
0
0
47
49
University
Undergraduate
Difficult
50-60
44
23
6
7.3
1
3.8
0
0
22
23
Years 11-13
Fairly
Difficult
60-70
15
8
7
8.5
1
3.8
0
0
11
12
Years 9-10
Standard
70-80
4
2
3
3.7
0
0
0
0
4
4
Year 8
Fairly Easy
80-90
2
1
0
0
0
0
0
0
0
0
Year 7
Easy
90-100
0
0
0
0
0
0
0
0
0
0
Year 6
Very Easy
Total
192
100
82
100
25
100
40
100
95
100
-
-
for preparing the content of TIBs to bring them more in line with the ITA 2007.
Interestingly, binding rulings, which are also drafted by the IRD, were found to
perform much better than TIBs that are prepared for a general audience. TIBs are
provided by the IRD “as a service to people with an interest in New Zealand taxation,” and contain information about changes to tax-related legislation, proposed
legislation, judgments, rulings, and other specialist tax topics. Binding rulings are
a service provided by the IRD on a fee-basis for private, product, and status rulings.
Public rulings are issued free of charge. All binding rulings reflect the IRD’s interpretations of the tax laws and are formally binding on the Commissioner. Table 3
reproduces Table 8 from Saw and Sawyer (2010) and reveals the relative successes
in terms of relative readability of the ITA 2007, TIBs, and binding rulings.
45
Enhancing Compliance Through Improved Readability
Based on the 2006 NZ Census, approximately 14 percent of the NZ population had a university bachelor’s degree or higher, indicating a relatively small
proportion of the population should be able to read and understand income tax
legislation. While this sector of the population will include most tax practitioners
Table 3: Summary of Flesch Reading Ease Scores (New Zealand Income Tax Legislation)
Flesch
Reading Ease
Score
ITA 2007
TIBs
Binding Rulings
Education level
General
Reading Ease
Scale
#
%
#
%
#
%
Below 30
35
18
4
25
8
44
University
Graduate
Very Difficult
30-50
92
48
10
63
8
44
University
Undergraduate
Difficult
50-60
44
23
2
13
1
6
Years 11-13
Fairly
Difficult
60-70
15
8
0
0
1
6
Years 9-10
Standard
70-80
4
2
0
0
0
0
Year 8
Fairly Easy
80-90
2
1
0
0
0
0
Year 7
Easy
90-100
0
0
0
0
0
0
Year 6
Very Easy
192
100
16
100
18
100
-
-
Total
(and tax students in due course), it is unlikely to include a significant number of
business taxpayers that operate small and medium enterprises (SMEs).
Prior New Zealand research using the Cloze Procedure
and other measures
In March 2006, the Panel, in conjunction with IRD tax policy officials, commissioned a staged, post-publication review of the ITA 2004 by Richard Castle. The
first stage was to identify the various methodologies available for evaluating the
readability of the legislation (although readability formulae and a full empirical
survey with questionnaires were excluded from consideration – Castle, 2006a).
In his follow-up report, Castle (2006b) includes comments from an expert linguist’s report (prepared by Harrison, 2006), in which she commences with a
46
Sawyer
comprehensive review of the prior readability literature, including a discussion
of the Cloze Procedure. Harrison (2006) concludes that the prior studies have
reinforced the validity of the Cloze Procedure as a useful means of judging comprehension of a particular text for a particular reader. Harrison (2006) observes
that a major disadvantage of the Cloze Procedure is that it does not provide any
explanation as to the difficulties of particular extracts and versions. Nevertheless,
she observes that a substantial number of researchers agree that it measures understandability rather than readability.
In terms of the findings from Harrison’s (2006) research, Sawyer (2008) provides a comprehensive analysis. Extracts from the ITA 2004 that were analyzed
using the Cloze Procedure produced an average score of 68.1 percent and were
marginally more understandable for intended readers (in this case tax professionals and revenue officials) than were extracts from the ITA 1976 (average score
62.5 percent). Both of these scores exceed the level of 44 percent suggested by
Bormuth (1967) for his instructional range. Overall, the results suggest that the
rewritten legislation (ITA 2004) may be easier for its primary users to read than
the ITA 1976.
Harrison (2006) suggests that there could be many reasons for the varying levels
of difficulty between Cloze versions. She also suggests that rather than speculate
on the causes, it is more useful to triangulate the data by applying a second readability evaluation technique to these sections in a future study; a recommendation
with which the current author concurs. Nevertheless, she cautions that, unless
the combined score for both versions of the legislation is very low relative to other
sections, such further analysis does not seem justified. While Harrison’s (2006)
findings are important and support earlier research using the Flesch Reading Ease
Index and the F-KGL Index, it is important to note that the comparison is ITA
1976 to ITA 2004—the reorganized ITA 1994 was not compared using the Cloze
Procedure. Prior research suggests that the reorganization itself may have a significant impact on readability (se Richardson and Sawyer, 1998).
Woellner et al. (2007) utilized Cloze Procedure testing on undergraduate students (considered to be “tax novices”) and tax practitioners and tax officers (considered to be “tax experts”). The authors found that the tax experts scored over
70 percent on both the ITAA 1936 and ITAA 1997, well over the benchmark of 44
percent. The tax novices failed to achieve the benchmark on both the ITAA 1936
and ITAA 1997, but found the ITAA 1997 marginally easier (at 35 percent) compared to 24 percent for the ITAA 1936.
Enhancing Compliance Through Improved Readability
47
Results of Using the Cloze Procedure on
Undergraduate Tax Students in New Zealand
As noted in the previous section, the key contribution of this study is to assess the
“success” of the NZ Government’s initiative of simplifying the income tax legislation through use of the Cloze Procedure on four versions of the ITA: the ITA 1976;
the ITA 1994; the ITA 2004, and the ITA 2007. A total of 221 useable instruments
were received, comprising 155 from the introductory tax class (60 percent) and 66
from the advanced class (62 percent). Table 4 sets out the demographic information of the 221 useable responses.
The four versions of the instrument were relatively evenly spread, ranging between 48 to 63 useable instruments for each of the four versions of the ITA. The
four versions were also well spread between the two undergraduate tax classes,
ranging between 34 and 46 for the introductory class for each version, and 14 to
18 for the advanced tax class. Given that there were fewer than 20 useable instruments for the four versions spread over the advanced tax class, it is considered
inappropriate to test for significance of differences between the various versions
of the instrument and the particular tax class. Nevertheless, descriptive statistics
(including mean and standard deviations) were determined for the two classes and
four variations of the instrument.
Several interesting observations from the background demographic data are
worth noting. First, in terms of language of the subjects, over 30 percent did not
have English as their first language, making the task of interpretation and understanding of the text more difficult, and in part reducing the effectiveness of the
Cloze Procedure. English as a second language for student subjects ranged from
as low as 27 percent for those completing the ITA 1976 version of the instrument to
as high as 36 percent for those completing the ITA 1994 version of the instrument.
In terms of age, most subjects were in the range 20-29 years which is unsurprising
given that students would typically take these courses in their second or third year
at university. The gender mix is very close to the university ratios as a whole (55:45
female to male).
Unsurprisingly, given that most students come to university from school in NZ,
and that those with wage and salary earnings generally do not need to file tax
returns (since income is taxed comprehensively at source and there are no deductions for employees), tax experience and related work was negligible for nearly 80
percent of the subjects. Nevertheless, nearly 20 percent had some minimal level
of experience with tax issues outside of the classroom. Most respondents found
the exercise to be difficult to extremely difficult, which is also unsurprising given
the complexity of taxation law, the time frame in which they had to complete the
exercise, and their limited tax experience. Of the four versions, the ITA 1976 was
found to be the least difficult, a finding counterintuitive to the expectations from
48
Sawyer
Table 4: Demographic background information for Cloze Procedure testing
Item
Frequency
Percentage
Introductory course
155
70.14
Advanced course
66
29.86
English
152
68.78
Other: Chinese
40
18.10
Japanese
5
2.26
European
5
2.26
Other
19
8.60
15-19 years
50
22.62
20-29 years
158
71.49
30-39 years
9
4.07
40-49 years
2
0.90
50+ years
2
0.90
Female
120
54.30
Male
101
45.70
None
172
78.90
1-4 years
41
18.81
5-9 years
2
0.90
10-19 years
1
0.50
20+ years
2
0.90
Course
Language
Age
Gender
Experience with taxation
filing or related work (218)
Difficulty of exercise (210)
Extremely easy
1
0.50
Very easy
2
1.00
Easy
8
3.81
Neutral
35
16.67
Difficult
52
24.76
Very difficult
58
27.62
Extremely difficult
54
25.71
(Unless otherwise indicated,
these results are based on 221
useable responses)
the prior readability research on the ITA in NZ. This finding is probably attributable to the sections selected which for the ITA 1976 were much less wordy that in
the more recent versions of the ITA.
49
Enhancing Compliance Through Improved Readability
Turning to the basic descriptive statistical data, Table 5 sets out the means and
standard deviations based on the four versions of the instrument and for the two
undergraduate tax classes. For the purposes of this exercise, Bormuth’s (1967)
44-percent instructional range was used as a benchmark since both groups of students can be considered to be tax novices rather than tax experts. Only legible
exact responses are treated as correct.
The data in Table 5 indicate that only in one instance did a particular version
of the ITA exceed 44 percent, namely the ITA 2004 for the advanced taxation
class, where over 55 percent of the responses exceeded the 44 percent mark. The
data also indicate that those in the advanced tax class overall performed better
than those in the introductory class. This suggests that the students benefited
from their prior instruction in taxation where they would have been familiar with
the concepts behind the four sections included in the instrument, although most
would have experienced the ITA 2007. Interest in the subject matter is also likely
to have been a factor as the advanced taxation course is not compulsory for accounting majors until 2011. In terms of the range of correct responses, the highest
Table 5: Basic Statistical Data for Cloze Procedure
Item
Class: Introductory
Class: Advanced
Overall
Means
%
%
%
ITA 1976
30.10
38.42
32.25
ITA 1994
29.59
35.95
31.75
ITA 2004
33.56
48.78
40.59
ITA 2007
30.17
35.84
32.09
Average
30.86
39.75
34.17
ITA 1976
12.88
13.10
12.89
ITA 1994
15.19
8.92
13.66
ITA 2004
17.40
15.11
17.34
ITA 2007
13.27
17.05
14.59
ITA 1976
16.67
17.65
16.95
ITA 1994
26.67
20.00
24.44
ITA 2004
25.00
55.56
36.96
ITA 2007
17.86
38.46
24.39
21.55
32.92
25.69
Standard Deviations
Number of responses
exceeding 44 percent
Average
50
Sawyer
number was 72 percent for the ITA 2004 (by a student in each of the introductory
and advanced tax classes). In total four student subjects exceed 70 percent, the
level considered to designate the subject as an expert. The lowest correct level of
responses was 5 percent (a student from the introductory tax class). A total of 50
out of 221 subjects exceed the 44 percent threshold (22.6 percent). In breaking
this down between the two classes, 21 of the 66 students from the advanced class
(31.8 percent) exceeded 44 percent, while 29 out of 155 (18.7 percent) exceed this
level from the introductory class. Table 6 provides an overview of the frequency
of correct responses across versions of the ITA and by tax class (introductory and
advanced).
Table 6: Frequency of Correct Responses
Range (number) /
Version of Act
0-20%
20-40%
40-60%
60-80%
80-100%
ITA 1976
11
20
11
0
0
ITA 1994
8
13
8
1
0
ITA 2004
7
11
8
2
0
ITA 2007
6
17
4
1
0
ITA 1976
2
7
7
1
0
ITA 1994
0
11
3
0
0
ITA 2004
0
5
8
5
0
ITA 2007
2
5
5
1
0
Introductory Class
(155)
Advanced Class
(66)
Overall, it appears that the student subjects found the ITA 2004 version of the
four sections easiest to understand, as measured by the level of correct responses
in filling in the gaps. The ITA 2007, the final version of the rewritten legislation,
came in a close third behind the ITA 1976 (the pre-rewritten version)! As noted
earlier, the number of observations for the advanced tax class makes significance
testing comparison between classes unsuitable.
Comparing the results to those of Woellner et al. (2007), the results overall are
strikingly comparable. In Woellner et al. (2007, p 723), students scored, on average, 35 percent for their partially rewritten statute (ITAA 1997), compared to 24
Enhancing Compliance Through Improved Readability
51
percent for the ITAA 1936. All four versions of the ITA in NZ scored higher than
the ITAA 1936, with the ITA 2004 exceeding the ITAA 1997.
Conclusions, Policy Implications, Limitations and
Future Research
The NZ Government’s key objective in rewriting the ITA 1976 and IRDA 1974 was
to make the legislation clearer and easier to read without changing the content
of the current legislation (except in limited identified circumstances). Since the
beginning of the rewrite project, NZ’s income tax legislation has been subjected
to reorganization, re-enactment, and extensive rewriting. This large-scale rewrite
project has required all income tax legislation users to review their transactions
and familiarize themselves with the new section locations and groupings (Sawyer,
2007). Thus, in achieving the long-term goals of tax simplification, legislative
complexity in the short term may have been compromised.
That said, it is clear that the NZ Government is committed to reducing the
complexity in New Zealand’s tax laws with respect to the drafting style employed.
The results of research to date provide some preliminary evidence that the NZ
Government’s effort has been successful in terms of improving the readability
of the tax laws. Research on the rewrite project provides evidence that different
drafting styles can affect the readability of income tax legislation. For example, the
use of shorter sentences and active voice, and the use of alphanumeric numbering,
can improve the readability of legislation. The NZ Government should take this
into consideration when drafting other tax legislation, such as the TAA 1994 and
GSTA 1985.
The NZ rewrite project experiment itself comes with a number of limitations,
including the use of student subjects as proxies for users of tax legislation (recognized to be at the level of novices only), and the small sample size (particularly in
the advanced tax class) making statistical significance testing unreliable. Within
the context of NZ as a case study, it is important to recognize a number of unique
features. The relative simplicity of the political processes in NZ, coupled with
the relatively less complex starting point with regard to the income tax legislation that the NZ legislative drafters were faced with compared to that facing other
countries, and the colonial history that forms a backdrop to much of NZ’s income
tax policy, facilitated the rewrite project. This should unsurprisingly enhance the
Cloze Procedure results. Nevertheless, the IRS and others involved in US tax policy would benefit from the opportunity to assess the costs and benefits of rewriting
legislative prose into a more simplified format, and, to this end, the New Zealand
experiment, buttressed by the emerging empirical research, provides much “food
for thought.”
52
Sawyer
The use of readability tests provides evidence to support the NZ Government’s
intention behind the rewrite project, and it has been largely achieved in terms of
more readable (and potentially understandable) legislation. However, while the
Flesch-and FKGL-level research suggests incremental gains with each version of
the ITA, the early Cloze Procedure analysis using tax students is less encouraging
in that the ITA 2007 does not appear to be the easiest to read of the four versions of
the ITA (the ITA 2004 occupies this place). More importantly, less than 25 percent
of the subjects (acknowledged to be novices) exceeded the 44 percent instructional guideline. Various explanations for this situation can be extracted from the
data, including the large number of subjects with English as their second language
(31.2 percent), the tight time frame to complete the exercise, and the underlying
complexity of tax concepts, collectively make expression in a readily understandable statutory format a very difficult task for drafters.
In terms of ongoing research in this area, it is intended to extend this work
further in the future to incorporate application of the Cloze Procedure with tax
practitioners and revenue officials as a group of experts. Future research could
also incorporate scenarios requiring application of statutory provisions under various versions of the ITA, using undergraduate tax students and tax practitioners
as subjects. Such research will allow for comparison against student subjects, as
well as to studies undertaken in other countries, such as Woellner et al. (2007) on
Australian students and tax practitioners and revenue officials.
Having data from two readability measures, one purely based on the results
from assessing the text and the other using subjects’ experiences, enables triangulation of results which may buttress the conclusions that can be drawn. To this
end, the results provide limited support for enhanced understandability through
the rewrite project’s efforts to simplify the statutory language. A positive feature of
the iterative rewrite project approach is that researchers (and hopefully drafters, as
well) have been able to assess their work, both through testing it with drafts made
available for public submission, and use of readability methods (there is evidence
that the IRD undertook limited in-house testing in this regard—see Sawyer, 2007).
This would then facilitate the process of drafters refining their text as a result of
submissions (and potentially though redrafting to improve readability and understandability) to provide a better quality final product.
As Pau et al. (2007) observe, the NZ Government considers the rewrite process
to be successful if the rewritten income tax legislation is accepted by all main users as clearer and easier to apply (Birch and Creech, 1994). This analysis is yet to
be completed in full and, thus, future research could identify all the main users of
income tax legislation, such as accountants, lawyers, revenue authority officials,
and the judiciary, and examine (potentially through use of a questionnaire survey)
their perceptions of the usability and readability of the legislation. Thus, future
research could provide more conclusive evidence as to whether the rewrite process
has in fact reduced the complexity of tax laws and enhanced compliance. Future
Enhancing Compliance Through Improved Readability
53
research should also be undertaken into quantifying the compliance cost impact
of the rewrite project, once the long-term benefits have been achieved.
Notwithstanding the above observations, the question needs to be asked: “Does
anyone really care if no one other than tax experts (for example, tax practitioners, tax lawyers, the Judiciary, tax academics, and tax officials) can read and understand the Income Tax Act?” Provided that taxpayers can determine their tax
obligations through other means, such as from tax agents and tax authority publications, do they really need to be able to read and understand tax legislation? I
would argue that it is a fundamental right for all citizens to be able to ascertain
their basic legal obligations (including tax obligations) readily without incurring
substantial cost and in an informed and unbiased manner. To this end, taxpayers
with an “average” level of education should be able to read and understand (tax)
legislation individually, should they choose to do so.
While this study has focused on NZ as an experimental case study, it is not the
only common law country that has experimented with rewriting of its income
tax legislation. Australia and the United Kingdom being major players, with research by Castle (2006b) suggesting that expert, yet nontax professional managers prefer the Australian partially rewritten legislation and the author himself the
almost complete UK rewritten legislation! Further research should be conducted
on the rewrite project in the UK, which is nearing completion, and now that the
Australian project has recommended, future research should build on that of
Smith and Richardson (1999) and Woellner et al. (2007).
The collective results of readability research on the NZ tax rewrite project provide evidence of improvements in readability (and to a lesser degree suggest improvements in understandability) through the process of simplifying the text of
the ITA. Such an outcome should enable taxpayers and their advisors to more
readily determine their tax obligations, thereby facilitating an environment that is
conducive to improvements in the level of tax compliance.
This paper presents the latest in a growing literature of research on the completed rewrite project in NZ. It offers further insights into this important case study
of a small country, without the complexity of the US, that undertook a massive
project to completely overhaul and redraft its income tax legislation. While the
findings provide their own insights, they should be read in the context of research
into all aspects of the rewrite project, including the initial proposals and strategy
adopted by the IRD, the involvement of the Panel, an assessment of the costs and
benefits (including when and how these may be measured), and assessments of
whether these simplification efforts have produced more understandable legislation. With the ITA 2007 still relatively new, ongoing re-education and reviews of
transactions remain prominent, and disputes continue to be based on earlier versions of the ITA, meaning that compliance costs continue to rise and the benefits
remain, in part, at least, elusive. Assessment of the ultimate impact on compliance
levels must be left to another day.
54
Sawyer
Acknowledgements
I would like to thank the IRS Research Conference 2010 organizing committee
for both inviting me to submit this paper and for providing financial support to
enable me to attend and present the paper, along with the College of Business
and Economics at the University of Canterbury for their financial support and for
providing the necessary ethical clearance to enable the instrument to be used on
students.
References
Birch, Hon. B. and Hon. W. Creech (December 1994), Rewriting the Income
Tax Act 1994—Objectives, Process, Guidelines: A Discussion Document,
(Wellington).
Blanchard, Rt. Hon P. (2007), “Some Basic Concepts of New Zealand GST,” in R.
Krever and D. White (eds), GST in Retrospect and Prospect (Wellington), 91 at
92.
Bormuth, J. (1967), “Comparable cloze and multiple-choice comprehension test
scores,” 10 Journal of Reading 291-299.
Castle, R. (2006a), Income Tax Act 2004: Parts A to E—A Model for a Review
(Wellington).
Castle, R. (2006b), The Income Tax Act 2004: Parts A to E—A postimplementation review: A Report to the Inland Revenue Department
(Wellington).
Flesch, R. (1979), How To Write Plain English, (New York).
Greene, B.B. (2001), “Testing reading comprehension of theoretical disclose with
cloze,” 24(1) Journal of Research in Reading, 82–98.
Guillemette, R.A. (1989), “The Cloze Procedure: Assessing the Understandability
of an IEEE Standard,” 32(1) IEEE Transactions on Professional Communication
41–47.
Harrison, J. (2006), The Readability of the Income Tax Act 2004: A Report to
Inland Revenue for the Rewrite Advisory Panel (Auckland).
Inland Revenue Department (2001), Rewriting the Income Tax Act: Exposure
Draft: General Commentary, (Policy Advice Division, Wellington).
Jackson, B. and V. Milliron (1986), “Tax Compliance Research: Findings,
Problems and Prospects,” 5 Journal of Accounting Literature, 125–165.
James, S.R., A.J. Sawyer, and I.G. Wallschutzky (1998), “The Complexities of Tax
Simplification: Progress in Australia, New Zealand and the United Kingdom,”
14(1) Australian Tax Forum, 29–68.
Enhancing Compliance Through Improved Readability
55
Long, S. and J. Swingen (1988), “The role of legal complexity in shaping taxpayer
compliance,” in Van Koppen, P., D. Hessing, and G. Van den Heuvel, Lawyers
on Psychology and Psychologists on Law, (Amsterdam, the Netherlands).
McKerchar, M. Complexity (2003), Fairness and Compliance: A study of
personal income taxpayers in Australia, (Australian Tax Research Foundation,
Sydney).
Pau, C., A.J. Sawyer, and A.J. Maples (2007), “Complexity of the New Zealand’s
Tax Laws: An Empirical Study,” 22(2) Australian Tax Forum, 59–92.
Redish, J. and J. Selzer (1985), “The Place of Readability Formulas in Technical
Communication,” Technical Communication, 1.
Richardson, M. (1998), and A.J. Sawyer, “Complexity in the Expression of New
Zealand Tax Laws: An Empirical Analysis,” 14(3) Australian Tax Forum,
325–360.
Richardson, M. and A.J. Sawyer (2001), “A Taxonomy of the Tax Compliance
Literature: Further Findings, Problems and Prospects,” 16(2) Australian Tax
Forum, 137–320.
Roth, J.A., J.T. Scholz and A.D. Witte (eds.) (1989), Taxpayer compliance: an
agenda for research (Philadelphia).
Saw, K. and A.J. Sawyer (2010), “Complexity of New Zealand’s Income Tax
Legislation: The Final Installment,” 25 Australian Tax Forum, 213-245.
Sawyer, A.J. (2007), “New Zealand’s Tax Rewrite Program—In Pursuit of the
(Elusive) Goal of Simplicity,” 4 British Tax Review, 405–427.
Sawyer, A.J. (2008), “RAP(ping) in Taxation: A Review of New Zealand’s Rewrite
Advisory Panel and its Potential for Adaptation to Other Jurisdictions,” 37(3)
Australian Tax Review, 148–163.
Smith, D. and G. Richardson (1999), “The Readability of Australia’s Taxation
Laws and Supplementary Materials: An Empirical Investigation,” 20(3) Fiscal
Studies, 321–349.
Smith, M. and R. Taffler (1992), “Readability and Understandability: Different
measures of the textual Complexity of Accounting Narrative,” 5(4) Accounting,
Auditing and Accountability Journal 84–98.
Stephens, C. (2000), All about Readability, available at: http://plainlanguage.com/
newreadability.html (accessed 22 April 2010).
Stevens, K.C. (1993), K.T. Stevens and W.P. Stevens, “A Response to ‘Measuring
Readability: A Comparison of Accounting Textbooks’,” 11 Journal of Accounting
Education 287–292.
Tan, L-M. and G. Tower (1992), “The Readability of Tax Laws: An Empirical
Study in New Zealand,” 9 Australian Tax Forum, 355–372.
56
Sawyer
Woellner, R., C. Coleman, M. McKerchar, M. Walpole, J. and Zetler (2007), “Can
Simplified Legal Drafting Reduce the Psychological Costs of Tax Compliance?
An Australian Perspective,” British Tax Review, No 6, 717–734.
Yin, R.K. (2003), Case study research: Design and methods (California).
Working Party on the Reorganization of the Income Tax Act 1976–1993 (1993),
Second Report of the Working Party on the Reorganization of the Income Tax Act
(Wellington).
Enhancing Compliance Through Improved Readability
57
Appendix
College of Business and Economics
Adrian Sawyer
Department of Accounting and Information Systems
Tel: +64 3 364 2617, Fax: + 64 3 364 2727
Email: [email protected]
25 February 2010
Rewrite of the Income Tax Act—Assessing the Readability of Legislation
First please let me introduce myself. I am Professor Adrian Sawyer from the
Department of Accounting and Information Systems. I have been involved for a
number of years in conducting research evaluating the effectiveness of the Rewrite
of the Income Tax Act Project in New Zealand. The rewrite project commenced
in 1993 with the reorganization of the existing key revenue Acts and involved a
progressive rewriting of the Income Tax Act resulting in the Income Tax Act 2007.
To date the research has focused on applicable readability formula such as the
Flesch Readability Index. In this further extension of the research I wish to test
readability using the method known as the Cloze Procedure. Briefly the Cloze
Procedure is a technique in which words are deleted from a passage according to
a word-count formula or various other criteria. The passage is presented to the
intended subjects (in this study, you as tax students) who insert words to complete
and construct meaning from the text.
Instructions:
What I would like you to do is to read the legislative sections and to fill in the
blanks with the word that you believe has been removed in each instance. The
intention of this study is to determine the degree to which students, as readers of
tax legislation, collectively can correctly determine the missing words within the
time frame provided. Here every fifth word has been deleted from four sections
of the Income Tax Act. I would also ask you to complete several questions that
follow which ask you for some background information. You have 15 minutes to
complete this task to the extent that you can in this time.
This research has been reviewed and approved by the University of Canterbury’s
Human Ethics Committee and is therefore subject to strict guidelines. All responses will be treated in the strictest confidence and will only be used for this
research project and possibly used collectively for comparison purposes with
58
Sawyer
future research involving New Zealand tax professionals. No one other than me,
as the researcher, will have access to these responses. There are no markings on
the documents and as such, it will not be possible to identify you in any papers
derived from this study.
Completion of this documentation is completely voluntary. You do not have to
answer any individual item or question if you do not wish to. You may withdraw
your participation at any time until you have handed in the documentation. By
completing the documentation and handing it in, however, it will be understood
that you have consented to participate in the project, and that you consent to publication of the results of the project with the understanding that your anonymity
will be preserved.
Yours sincerely
Dr. Adrian Sawyer,
Professor of Taxation, Tel 3642617 (direct),
Email: [email protected]
Enhancing Compliance Through Improved Readability
59
Sections from the Income Tax Act 2007
BD 1: Income, exempt income, excluded income, non-residents’ foreignsourced income, and assessable income
Amounts of income
(1) An amount is income of a person if it is their income under a provision in
Part C (Income).
Exempt income
(2) An amount of income of a person is exempt income if it is their exempt
income under a provision in subpart CW (Exempt income) or CZ (Terminating
provisions).
Excluded income
(3) An amount of income of a person is excluded income if—
(a) it is their excluded income under a provision in subpart CX (Excluded
income) or CZ; and
(b) it is not their non-residents' foreign-sourced income.
Non-residents' foreign-sourced income
(4) An amount of income of a person is non-residents' foreign-sourced income
if—
(a) the amount is a foreign-sourced amount; and
(b) the person is a non-resident when it is derived; and
(c) the amount is not income of a trustee to which section HC 25(2) (Foreignsourced amounts: non-resident trustees) applies.
Assessable income
(5) An amount of income of a person is assessable income in the calculation of
their annual gross income if it is not income of any of the following kinds:
(a) their exempt income:
(b) their excluded income:
(c) their non-residents' foreign-sourced income.
Defined in this Act:
amount, annual gross income, assessable income, excluded income, exempt
income, foreign-sourced amount, income, non-resident, non-residents' foreignsourced income
60
Sawyer
DA 1: General permission
Nexus with income
(1) A person is allowed a deduction for an amount of expenditure or loss,
including an amount of depreciation loss, to the extent to which the expenditure
or loss is—
(a) incurred by them in deriving—
(i) their assessable income; or
(ii) their excluded income; or
(iii) a combination of their assessable income and excluded income; or
(b) incurred by them in the course of carrying on a business for the purpose
of deriving—
(i) their assessable income; or
(ii) their excluded income; or
(iii) a combination of their assessable income and excluded income.
General permission
(2) Subsection (1) is called the general permission.
Avoidance arrangements
(3) Section GB 33 (Arrangements involving depreciation loss) may apply to
override the general permission in relation to an amount of depreciation loss.
Defined in this Act:
amount, assessable income, business, deduction, depreciation loss, excluded
income, general permission, loss
YD 1: Residence of natural persons
What this section does
(1) This section contains the rules for determining when a person who is not a
company is a New Zealand resident for the purposes of this Act.
Permanent place of abode in New Zealand
(2) Despite anything else in this section, a person is a New Zealand resident if
they have a permanent place of abode in New Zealand, even if they also have a
permanent place of abode elsewhere.
183 days in New Zealand
(3) A person is a New Zealand resident if they are personally present in New
Zealand for more than 183 days in total in a 12-month period.
Person treated as resident from first of 183 days
Enhancing Compliance Through Improved Readability
61
(4) If subsection (3) applies, the person is treated as resident from the first of the
183 days until the person is treated under subsection (5) as ceasing to be a New
Zealand resident.
Ending residence: 325 days outside New Zealand
(5) A person treated as a New Zealand resident only under subsection (3) stops
being a New Zealand resident if they are personally absent from New Zealand for
more than 325 days in total in a 12-month period.
Person treated as non-resident from first of 325 days
(6) The person is treated as not resident from the first of the 325 days until they
are treated again as resident under this section.
Government servants
(7) Despite subsection (5), a person who is personally absent from New Zealand
in the service, in any capacity, of the New Zealand Government is treated as a
New Zealand resident during the absence.
Presence for part-days
(8) For the purposes of this section, a person personally present in New Zealand
for part of a day is treated as—
(a) present in New Zealand for the whole day; and
(b) not absent from New Zealand for any part of the day.
YD 4: Classes of income treated as having New Zealand source [extract]
What this section does
(1) This section lists the types of income that are treated as having a source in
New Zealand for the purposes of this Act.
Business in New Zealand
(2) Income derived from a business has a source in New Zealand if—
(a) the business is wholly carried on in New Zealand:
(b) the business is partly carried on in New Zealand, to the extent to which the
income is apportioned to a New Zealand source under section YD 5.
Contracts made or performed in New Zealand
(3) Income derived by a person from a contract has a source in New Zealand if
the contract is—
(a) made in New Zealand, except to the extent to which the person wholly
or partly performs the contract outside New Zealand, and the income is
apportioned to a source outside New Zealand under section YD 5:
62
Sawyer
(b) made outside New Zealand but the person wholly or partly performs
the contract here, to the extent to which the income is apportioned to a New
Zealand source under section YD 5.
Personal services in New Zealand
(4) An amount that is income under section CE 1 (Amounts derived in
connection with employment) has a source in New Zealand if the amount is
earned in New Zealand, even if the employer is not a New Zealand resident.
Accident compensation payments
(5) An accident compensation payment as defined in section CF 1(2) (Benefits,
pensions, compensation, and government grants) has a source in New Zealand.
Enhancing Compliance Through Improved Readability
63
Background Information
Question 1:
Which age group are you currently in?
15-19
20-29
30-39
40-49
50+ years
Question 2:
What gender are you?
Female
Male
Question 3:
What is your first language for reading and writing?
English
Other (please state)___________________________
Question 4:
Which tax course are you completing this material for?
ACIS 254: Introduction to Taxation
ACIS 358: Advanced Issues in Taxation
Question 5:
How many years (part-time and/or full-time) tax-related work experience and/or
tax filing experience do you have?
None
1 - 4
5 - 9
10 - 19
20+ years
Question 6:
On the scale below please rate how easy you found it to complete the gaps in the
sections of legislation by circling the corresponding number:
Extremely Very Easy Neutral Difficult Very
Extremely
easy
easy
difficult
difficult
1
2
3
4
5
6
7
Tax Compliance Costs: the
Effect of Authority Behavior
and Taxpayer Services
Sebastian Eichfelder, University of Wuppertal, Schumpeter School of Business
and Economics; Chantal Kegels, Federal Planning Bureau, Brussels; and Michael
Schorn, Institute for Economic and Political Research, Cologne
I
n addition to the tax law itself, the optimal enforcement of tax regulations is
an important economic issue (Slemrod and Yitzhaki 2002, McCubbin 2004).
Within the literature on tax compliance decisions, the question has been
raised if a shift from a control-oriented to a customer-oriented approach of tax
administration could reduce tax evasion (Hansford and Hasseldine 2002, Murphy
2004, Freedman et al. 2009). According to Kirchler (2007), instead of a “cops and
robbers” mentality, educating and supporting customers could be more promising
to convince taxpayers to comply.
As has been stated by Gale and Holtzblatt (2002), administrative issues may
not only affect the compliance decisions of private taxpayers, but also their burden of red tape. According to the literature, the compliance costs of private businesses substantially exceed the administrative costs of the tax authorities (for a review, see Evans 2003, Vaillancourt and Clemens 2008). Vaillancourt and Clemens
(2008) estimate the compliance costs of Canadian households and businesses at a
range from 1.2 percent to 1.8 percent of the GDP, while the administrative costs of
the Canadian authorities lie in a range from 0.2 percent to 0.5 percent.
Taking into account economies of scale within the tax compliance process
(Sandford et al. 1989, Allers 1994, Evans 2003), the support especially of small and
medium-sized businesses by government agencies could result in a decrease of the
cost burden for the overall tax system. From this perspective, a more customeroriented approach of tax administration could result in an improvement of productivity for the economy as a whole (Barton 2001).
From our point of view, there is a lack of empirical evidence regarding the
hypothesized relationship between customer orientation and compliance costs.
From an efficiency perspective, it is especially of interest to quantify potential cost
reductions. Furthermore, the identification of the key cost drivers is an important
question of research. Up to our knowledge, we estimate for the first time the effect
of taxpayer services on the tax compliance costs of private businesses.
Using ratings on administrative quality as measure for customer orientation,
we find empirical support for a relationship of authority behavior and compliance burdens. Our results suggest that service orientation may yield a significant
36
Eichfelder, Kegels, and Schorn
cost reduction for private businesses. In the German case, a positive evaluation
of the authorities’ service orientation is correlated with a cost reduction of about
30 percent. In Belgium, businesses with a negative rating regarding the information obtained from the revenue service bear on average an about 26 percent higher
cost burden.
The paper is organized as follows. In the next sections, we illustrate the two
data bases and our estimation strategy. We discuss potential caveats of cost measurement and especially the potential problem of endogeneity of our proxies for
authority behavior. Thereafter, we analyze the regression results for Germany and
Belgium. The last section concludes the paper, while the appendices contain additional regression results and cross checks for the Belgian data set.
Database
German Data
The German data has been raised by the Institute for Small and Medium Businesses
Research in Bonn and the Institute for Economic and Political Research in Cologne
in 2003 on behalf of the German Ministry of Economics and Labor. It contains information on the compliance costs resulting from business taxes, social insurance
contributions, statistics, and labor market and environmental regulations. Further
information is given by Kayser et al. (2004).
Corresponding to investigations in other countries (OECD 2001, European
Communities 2004, DeLuca et al. 2007), the overall cost burden is calculated by
the sum of internal personnel costs, expenses for external advice, and other monetary expenses. The cost burdens are subjective estimates given by the survey participants. The same holds for the labor costs per hour. The tax-related costs TC
and the social insurance-related costs SC are described by a proportion of the
overall burden of red tape. The sample has been selected to represent the German
business population.
To our knowledge, the data is the best survey available concerning the tax compliance costs of German businesses. Nevertheless, some measurement issues have
to be taken into account. A basic problem associated with measuring tax compliance costs is the reliability of the taxpayers’ statements. As Tate (1988, p. 352)
argues, the respondents may overstate their compliance cost burden to impose
pressure on political authorities. On the other hand, the literature gives also some
empirical evidence for a possible cost perception deficit. From this perspective,
respondents may underestimate tax compliance costs by failing to remember parts
of their cost burden.1 Regarding our data set, we find a strong correlation between
the compliance cost estimates and the “perceived” compliance burden.2 We may
37
Tax Compliance Costs: Authority Behavior and Taxpayer Services
therefore assume that there is no systematic overestimation or underestimation of
the burden of red tape.
Because of a relatively low response rate of 7.3 percent, the empirical results could
be affected by a non-response bias.3 According to the literature, there are theoretical
and empirical arguments for a positive and a negative non-response bias. Therefore,
the net effect of a self-selection on average cost estimates is unclear and could result
in “random noise.” A selection bias would not necessarily distort the regression results if it is not correlated with the investigated variables. Taking into account the
small differences between the descriptive results of our database and international
estimates (OECD 2001, European Communities 2004, Klun and Blažić 2005), there
is no reason to suspect a major distortion due to non-response bias.
Table 1 contains the mean and median values (in parentheses) for the overall
compliance costs (including statistics as well as labor market and environmental
regulations), the relative burdens per turnover and employee, and the proportion
of compliance costs caused by business taxes and social insurance contributions.
According to the size criteria of the European Union, we define enterprises with
less than 50 (between 50 and 249) associates as small (medium).4, 5 The case number is also considered [in parentheses].
TABLE 1. German Data: Compliance Cost Burdens
Size class
Small
Medium
Big
103,323 (53,000) [196] 649,716 (140,000) [97]
Cases
Compliance costs per business (€)
37,726 (25,000) [434]
Compliance costs per associate (€)
3,296 (2,000) [434]
1,090 (499) [196]
894 (169) [97]
727
727
Compliance costs per turnover (%)
3.27 (1.83) [417]
1.06 (0.48) [184]
0.59 (0.11) [91]
692
Tax-related costs (%)
47.96 (45.00) [580]
37.39 (35.00) [225]
32.81 (30.00) [116]
921
Social insurance-related costs (%)
29.37 (30.00) [580]
30.20 (30.00) [225]
27.78 (26.00) [116]
921
Due to economies of scale, the relative cost burdens (per turnover or per
employee number) are significantly higher for small businesses. Resulting from
strong effects of business size on the compliance cost burden, we find also a remarkable difference between mean and median values that is driven by businesses
at the edge of a specific size class. Table 1 documents further that the majority
of the overall cost burden results from taxes and social insurance contributions.
Regarding small businesses, about 75 percent of the overall cost burden are caused
by corresponding administrative obligations.
In addition to the compliance cost burden, the survey contains general information on the participants (size, location of head office, legal form, sector), information on specific forms of employment (casual workers, fluctuation of employees, etc.), the accounting method used for tax purposes, the use of electronic
38
Eichfelder, Kegels, and Schorn
submission methods, Likert scale values regarding the “perceived” burden of complying with the legislation and value judgements on administrative quality. The
ratings on tax administration are given on a 5-point Likert scale regarding the
qualification, the service orientation and the processing time of the German tax
and social insurance administration.
The distribution of ratings is documented by Table 2. A rating of 1 denotes very
positive and 5 very negative. The proportion of a specific rating to the overall number of ratings is given in parentheses. Furthermore, we document also the total
number of cases and the mean rating for each administrative issue.
TABLE 2. German Data: Ratings for Tax and Social Insurance Administration
Variable
1
2
3
4
5
Cases
Mean
QUALIFICATION TAX
42 (3.81)
506 (45.87)
397 (35.99)
98 (8.88)
60 (5.44)
1,103
2.66
SERVICE TAX
17 (1.57)
230 (21.24)
324 (29.92)
268 (24.75)
244 (22.53)
1,083
3.45
PROCESSING TAX
20 (1.84)
203 (18.66)
312 (28.68)
294 (27.02)
259 (23.81)
1,088
3.52
QUALIFICATION SIA
39 (3.80)
482 (46.98)
355 (34.60)
95 (9.26)
55 (5.36)
1,026
2.65
SERVICE SIA
29 (2.83)
340 (33.17)
349 (34.05)
187 (18.24)
120 (11.71)
1,025
3.03
PROCESSING SIA
26 (2.57)
327 (32.38)
371 (36.73)
187 (18.51)
99 (9.80)
1,010
3.01
Evidently, the qualification of the tax and social insurance administration is
positively evaluated. Regarding this aspect we find in both cases about 50 percent
of positive (1 or 2), but only 15 percent of negative (4 or 5) ratings. Taking into
account the processing time and the service orientation, we can state divergent
results. While the ratings for SERVICE TAX and PROCESSING TAX are relatively
negative, we do not find a similar result with reference to the social insurance administration (SERVICE SIA, PROCESSING SIA).
Belgian Data
The Belgian data has been raised by the Federal Planning Bureau in Brussels by
order of the Belgian Council of Ministers. It contains information on compliance
costs resulting exclusively from business taxes.6 The data consists of four crosssections regarding the years 2000, 2002, 2004 and 2006. Due to the fact that our
data source is not a panel, most of the records are one-shot observations. Each
survey contains two separated samples for enterprises (generally in the legal form
of a corporation) and independent businesses (sole proprietorships). For further
information, see De Vil and Kegels (2002), Joos and Kegels (2004), Janssen et al.
(2006) and Kegels (2008).
The cost measurement is similar to the German data. However, except from
tax adviser costs, monetary expenses are not considered in each survey year and
are therefore excluded from further analysis.7 The response rates are higher than
39
Tax Compliance Costs: Authority Behavior and Taxpayer Services
in the German survey, but still relatively low. Regarding the enterprises surveys
(independents surveys) the response rates lie in a range from 15 percent to 23
percent (from 8 percent to 17 percent). Taking into account that the compliance cost burdens are similar to international estimates (OECD 2001, European
Communities 2004, Klun and Blažić 2005), there is no sign for a significant bias
of the cost burden.
Table 3 documents the mean and the median (in parentheses) compliance cost
burdens of sampled Belgian businesses in euro. The table contains separate values
for independent businesses, as well as for small, medium, and big enterprises. As
elucidated above, we use the size criteria of the European Union. In contrast to the
German case, size is measured by the number of employees and not by the number
of associates.8
TABLE 3. Belgian Data: Absolute Cost Burdens
Survey
Independent
Small
enterprise
Medium
enterprise
Big
enterprise
Cases
2000
4,550 (2,975) [117]
40,110 (10,055) [87]
66,738 (17,105) [12]
201,506 (87,382) [32]
248
2002
11,044 (2,856) [174]
171,232 (14,310) [106]
85,681 (39,000) [23]
145,108 (62,250) [40]
343
2004
8,054 (3,240) [142]
74,490 (12,060) [77]
36,004 (25,020) [32]
304,529 (62,400) [61]
312
2006
2,400 (1,250) [113]
30,801 (10,000) [72]
39,024 (14,563) [20]
74,009 (30,750) [47]
252
546
342
87
180
1,155
Cases
As should be expected, the cost burden increases in business size. Furthermore,
the lion's share of our data base consists of small businesses. The differences between median and mean values, as well as between different survey years, are
remarkable. This results especially from the strong effect of business size on the
compliance cost burden and from the variance of average business size between
the different survey years. Due to economies of scale, business size does not only
affect the absolute, but also the relative compliance cost burden. This is exemplified by Table 4 (relative costs in percent of turnover).
TABLE 4. Belgian Data: Relative Cost Burdens
Independent
Small
enterprise
Medium
enterprise
Big
enterprise
Cases
2000
14.29 (5.83) [109]
10.19 (0.96) [80]
0.65 (0.17) [12]
9.98 (0.04) [32]
233
2002
221.03 (4.77) [158]
4.15 (0.80) [99]
0.33 (0.13) [23]
0.11 (0.04) [37]
316
2004
301.12 (4.64) [130]
17.90 (0.80) [74]
4.54 (0.22) [31]
2.99 (0.09) [60]
295
2006
11.03 (3.04) [100]
77.05 (0.66) [67]
0.13 (0.10) [20]
0.10 (0.03) [47]
234
497
320
85
176
1,078
Survey
Cases
40
Eichfelder, Kegels, and Schorn
Corresponding to the literature, the cost ratios are significantly higher for independents and small businesses compared to the bigger size classes. Nevertheless,
we also find a high variance of cost ratios between different survey years. There
are two main reasons for this outcome. On the one hand, size classes in our data
are based on the employee number, but not on the turnover. A very low turnover
results in the proportion of compliance costs to turnover converging to infinity.
For example, the independents surveys in 2002 and 2004 contain cases with a
turnover of less than 20 euro and very high cost ratios. On the other hand, especially high differences between mean and median values may also be a sign for
outliers that could bias regression results. Therefore, an analysis for outliers seems
to be necessary.
In addition to the cost burden, the data includes “demographic” information
on business size, industry, and region, as well as ratings on the Belgian tax policy
and the Belgian tax administration. For specific years, there is also information
regarding business age, the number of establishments in Belgium, the legal form,
the use of different information technology tools for tax purposes, and proposals
to simplify the overall tax system.9
The questionnaire includes seven statements on administrative quality (original statements are in French language). The answers were given on a 5-point
Likert scale:
1. It is easy to assess, which tax agency should be contacted
(AGENCY).
2. It is easy to contact the right tax agency (CONTACT).
3. The tax administration gives precise answers (ANSWER).
4. Administrative decisions are clearly motivated (MOTIVATION).
5. The administration gives an answer within the expected delay
(DELAY).
6. The answer is the same regardless of the contacted service personnel
(CONTRADICTION).
7. The obtained information corresponds to your needs
(INFORMATION).
Similar to the German case, we find a considerable variance in the distribution
of ratings. For example, only about 30 percent of the requested businesses did give a
positive statement regarding possible contradictions in the answers of administrative staff members to businesses’ requests (CONTRADICTION). By contrast, the
majority was convinced to receive the required information (INFORMATION).
41
Tax Compliance Costs: Authority Behavior and Taxpayer Services
TABLE 5. Belgian Data: Ratings for Tax Administration
Variable
1
2
3
4
5
AGENCY
108 (6.99)
CONTACT
89 (5.79)
ANSWER
Cases Mean
581 (37.58)
95 (6.14)
487 (31.50)
275 (17.79)
1,546
2.79
535 (34.81)
109 (7.09)
517 (33.64)
287 (18.67)
1,537
2.86
72 (4.67)
629 (40.79)
144 (9.34)
487 (31.58)
210 (13.62)
1,542
2.82
MOTIVATION
61 (3.96)
514 (33.40)
192 (12.48)
563 (36.58)
209 (13.58)
1,539
2.97
DELAY
96 (6.25)
709 (46.13)
173 (11.26)
356 (23.16)
203 (13.21)
1,537
2.77
CONTRADICTION
49 (3.20)
416 (27.17)
351 (22.93)
491 (32.07)
224 (14.63)
1,531
3.27
INFORMATION
59 (3.84)
725 (47.20)
194 (12.63)
426 (27.73)
132 (8.59)
1,536
2.79
Unlike the German case, only a relatively low proportion of respondents did
neither give a positive nor a negative rating. This should result from the fact that 3
does not label a neutral rating, but no opinion.
Estimation Strategy
Reliability of Rating Behavior
In both countries, the data contains information on tax compliance costs and ratings on administrative quality from a taxpayers’ perspective. Interpreting these
ratings as proxies for authority behavior, we would expect that a customer-oriented
administration results in a positive rating and, therefore, in a decrease of compliance costs. By contrast, a negative rating should be a proxy for administrative
problems resulting in a significantly higher cost burden for the taxpayer.
A problem in estimating the effect of authority behavior on compliance costs
lies in a potential endogeneity of the rating variable. A correlation could not only
result from the authority behavior itself, but also from a dissatisfaction of the taxpayer with the compliance burden or the overall tax system. According to this
argument, taxpayers with high compliance costs could “punish” the tax administration by negative ratings. On the other hand, it has to be expected that negative
and positive ratings are significantly affected by the experiences of taxpayers with
the administrative authorities. Therefore, if the ratings are reliable they should be
a good proxy for authority behavior from a taxpayers’ perspective.
Our data contain not only ratings on administrative quality but also on similar
factors. This information can be used as a control parameter for the reliability
of the administrative ratings. If the evaluation behavior of private businesses results mainly from a single factor like the compliance cost burden (endogeneity of
the value judgment), we would expect a high degree of collinearity of all rating
42
Eichfelder, Kegels, and Schorn
variables. In this case, the taxpayer would punish or reward the authorities with
a generally positive or negative statement that results especially from the compliance costs and not vice versa.
In the German data, we find the following correlations of rating behavior. We
also include the correlations with the “perceived” compliance burden given on a
5-point Likert scale. If evaluation behavior is mainly driven by compliance costs,
we would expect a strong correlation between rating variables and this “perceived”
burden that may be interpreted as a proxy for psychological compliance costs.
QUALIFICATION
TAX
SERVICE
TAX
PROCESSING
TAX
QUALIFICATION
SIA
SERVICE
SIA
PROCESSING
SIA
PERCEIVED
CC
TABLE 6. German Data: Rating Correlations
1.000
0.523
0.369
0.432
0.258
0.223
0.155
SERVICE TAX
—
1.000
0.550
0.282
0.441
0.340
0.240
PROCESSING TAX
—
—
1.000
0.194
0.280
0.351
0.187
QUALIFICATION SIA
—
—
—
1.000
0.625
0.557
0.118
SERVICE SIA
—
—
—
—
1.000
0.659
0.212
PROCESSING SIA
—
—
—
—
—
1.000
0.185
PERCEIVED CC
—
—
—
—
—
—
1.000
QUALIFICATION TAX
According to Table 6, there is a wide variation of correlation coefficients.
Coefficients for similar issues (ratings on tax administration) are considerably stronger than correlations between ratings on tax administration and ratings on social insurance administration. Furthermore, we find only a relatively
weak connection between rating behavior and the “perceived” cost burden.
Using all rating variables as exogenous factors of a linear regression analyzing
PERCEIVED CC, we obtain an R2 of only 7.93 percent. Therefore, no more than
8 percent of the “perceived” burdens’ variance may be described by rating behavior. From this perspective, there is no convincing evidence for an endogeneity of
the taxpayers’ statements.
The Belgian data set contains not only ratings on tax administration, but also
seven statements on tax legislation (original statements are in French language).
The answers were given on a 5-point Likert scale:
1. The information on a tax regulation is obtained in advance before
it is adapted (ADVANCE).
2. Tax regulations are easy to understand (UNDERSTANDABILITY).
3. Their objectives are clear (OBJECTIVE).
43
Tax Compliance Costs: Authority Behavior and Taxpayer Services
4. They are sufficiently adapted to all situations (ADAPTION).
5. The information on a tax regulation is obtained in a reasonable
time to comply with the tax law (TIME).
6. The different tax regulations are coherent to each other
(COHERENCY).
7. They include sufficient and adequate information content
(ENTROPY).
This distribution does not seem to support the hypothesis of a very strong correlation between administrative and legislative ratings. On average, the requested
businesses have a lower rating for tax legislation compared to the tax administration. Furthermore, there are also remarkable differences regarding the distribution
of the legislative ratings (for example TIME and UNDERSTANDABILITY).
The following table documents the correlations of all rating variables within the
Belgian data set.
TABLE 7. Belgian Data: Ratings for Tax Legislation
Variable
1
2
3
4
5
133 (8.74)
447 (29.37)
191 (12.55)
422 (27.73)
329 (21.62)
1,522
3.00
UNDERSTANDABILITY
39 (2.54)
328 (21.33)
87 (5.66)
639 (41.55)
445 (28.93)
1,538
3.14
OBJECTIVE
73 (4.79)
428 (28.07)
130 (8.52)
597 (37.97)
315 (20.66)
1,525
3.00
ADAPTION
35 (2.30)
299 (19.61)
184 (12.07)
612 (40.13)
395 (25.90)
1,525
3.26
112 (7.28)
604 (39.25)
115 (7.47)
432 (28.07)
276 (17.93)
1,539
2.79
COHERENCY
42 (2.74)
307 (20.03)
244 (15.92)
587 (38.29)
353 (23.03)
1,533
3.29
ENTROPY
45 (2.93)
417 (27.18)
153 (9.97)
589 (38.40)
330 (21.51)
1,534
3.08
ADVANCE
TIME
Cases Mean
Evidently, the coefficients are considerably higher for correlations within administrative or legislative ratings. For example, the coefficients of ADVANCE for
correlations with other legislative issues lie in a range from 0.349 to 0.434, while
the range for correlations with administrative ratings is about 10 percentage points
lower (from 0.257 to 0.329). Furthermore, we also find a considerable variance
of the correlation coefficients. While there is a very strong correlation between
AGENCY and CONTACT, the coefficient for the connection between AGENCY
and ADVANCE is relatively low.
The observed diversity and interdependency of rating behavior does not support the hypothesis that ratings are mainly driven by a single factor. Hence, the
empirical support for a strong effect of tax compliance costs on the overall evaluation behavior is relatively weak. Furthermore, we find also that ratings for similar issues (like AGENCY and CONTACT) are strongly correlated to each other
compared to ratings for separate aspects (like ADVANCE and AGENCY). As an
exogenous event should affect ratings for related aspects in a similar direction,
44
Eichfelder, Kegels, and Schorn
this can be interpreted as an empirical support for a rating behavior that is mainly
driven by exogenous factors like the experiences of the taxpayer.
ADVANCE
UNDERSTANDABILITY
OBJECTIVE
ADAPTION
TIME
COHERENCY
ENTROPY
AGENCY
CONTACT
ANSWER
MOTIVATION
DELAY
CONTRADICTION
INFORMATION
TABLE 8. Belgian Data: Rating Correlations
1.000
0.434
0.356
0.372
0.396
0.349
0.394
0.295
0.252
0.291
0.329
0.257
0.301
0.308
UNDERSTANDABILITY
—
1.000
0.610
0.540
0.467
0.472
0.566
0.417
0.353
0.376
0.339
0.304
0.231
0.336
OBJECTIVE
—
—
1.000
0.555
0.425
0.504
0.503
0.333
0.305
0.345
0.347
0.299
0.258
0.358
ADAPTION
—
—
—
1.000
0.452
0.604
0.556
0.310
0.282
0.359
0.375
0.326
0.316
0.393
TIME
—
—
—
—
1.000
0.424
0.550
0.317
0.287
0.374
0.351
0.328
0.200
0.345
COHERENCY
—
—
—
—
—
1.000
0.592
0.288
0.280
0.365
0.405
0.334
0.344
0.386
ENTROPY
—
—
—
—
—
—
1.000
0.358
0.328
0.366
0.413
0.330
0.302
0.391
AGENCY
—
—
—
—
—
—
—
1.000
0.692
0.532
0.426
0.446
0.322
0.489
CONTACT
—
—
—
—
—
—
—
—
1.000
0.566
0.447
0.476
0.301
0.490
ANSWER
—
—
—
—
—
—
—
—
—
1.000
0.573
0.546
0.398
0.624
MOTIVATION
—
—
—
—
—
—
—
—
—
—
1.000
0.546
0.414
0.554
DELAY
—
—
—
—
—
—
—
—
—
—
—
1.000
0.451
0.596
CONTRADICTION
—
—
—
—
—
—
—
—
—
—
—
—
1.000
0.470
INFORMATION
—
—
—
—
—
—
—
—
—
—
—
—
—
1.000
ADVANCE
Model Specification
Corresponding to the literature (Hudson and Godwin 2000, Slemrod and
Venkatesh 2002), we use a logarithmic OLS model for our econometric analysis.10
Hence, regression coefficients may be interpreted as elasticities of the compliance
cost burden related to an exogenous factor. The logarithmic transformation accounts for economies of scale in the tax compliance process and ensures linearity
of the OLS regression:
CCOST = α0 + α1 ∙ SIZE + α2 ∙ ADMINISTRATION + α3 ∙ X + ε
(1)
CCOST denotes the logarithm of compliance costs, SIZE the appropriate measure for business size, ADMINISTRATION the vector for rating behavior and X
the vector of further control parameters. The coefficients (or the corresponding
vectors) are described by α0 to α3, while the error term is labelled by ε.
As the statements for administrative and (in Belgium) legislative issues are
based on Likert scales, it seems appropriate to account for rating behavior by
Tax Compliance Costs: Authority Behavior and Taxpayer Services
45
dummy variables. We use separate dummies for positive and negative statements.
Therefore, we compare in each specification of our regression businesses with a
positive (negative) rating to all other businesses in the data set. This is due to the
fact that businesses with a neutral rating do not seem to be a sufficient control
group in our setting.11
In detail we account for the following variables analyzing the German data set:12
CCOST—Logarithm of compliance costs: these are defined as sum
of personnel costs, external costs and other monetary expenses.
The sum is exclusively calculated if there are no missing values.
Tax compliance costs are defined as the overall compliance costs
multiplied with the proportion of tax-related costs. Regarding social
insurance-related costs, we add 1 percent to the corresponding
proportion to prevent undefined logarithmic values.13
SIZE—Regarding business taxes, we use the logarithm of turnover
as size measure. In terms of the social insurance-related compliance
costs, the logarithm of employees is more appropriate. In this case,
we add one employee to prevent undefined logarithmic values.
As an additional size measure, we include a dummy variable for
employment-related activities (EMPLOYMENT) in case of at least
two associates.14
ADMINISTRATION—Set of dummy variables regarding the
following administrative issues: QUALIFICATION, SERVICE and
PROCESSING. We account for positive ratings (1, 2) and negative
ratings (4, 5) by a separate dummy variable for the tax and social
insurance administration (SIA).
INDUSTRY—Set of dummy variables: we control for industrial
businesses (INDUSTRIAL), traders (TRADE), construction
businesses (CONSTRUCTION) and services for enterprises
(ESERVICE). The remaining businesses are in the services sector as
well (other services). Furthermore, we include dummies for liberal
professions (PROFESSION) and crafts enterprises (CRAFTS).
LEGAL FORM—Dummy variables for legal form including
individual enterprises (INDEPENDENT), partnerships
(PARTNERSHIP) and the combination of a limited partnership
and a limited liability company (German: GMBH & CO. KG). The
remaining businesses are corporations.
46
Eichfelder, Kegels, and Schorn
OUTSOURCING—We utilize the logarithm of the proportion of
external costs to total compliance costs increased by 1 percent as
measure for the use of external advice.
AGE—Natural logarithm of business age raised by 1: the variable
accounts for possible start-up costs of young businesses, which are
documented by Hansford et al. (2003).
E-FILING—We take into account electronic accounting methods
for the tax and social insurance administration (E-FILING).
Furthermore, we also consider potential problems resulting from
e-filing (E-FILING PROBLEM).
CASH ACCOUNTING—This dummy variable accounts for
businesses using a cash accounting method. It is exclusively included
for tax-related costs, but not for social insurance-related costs.
Within the German data set, we do not account for regional dummies. This is
due to the fact that the German fiscal administration is organized by the states and
not by a federal agency. Therefore, the use of regional dummies (German states)
could intercept the effect of authority behavior. The following variables are exclusively considered within the social insurance-related models:
EMPLOYEES—Set of variables regarding specific forms of
employment: we consider part time workers (PART TIME), casual
workers (CASUAL), handicapped workers (HANDICAPPED)
and trainees (TRAINEE). We use the logarithm of the
proportion of these specific forms of employment to the total
number of employees raised by 1 percent. Furthermore, we also
consider specific obligations for foreign employees by a dummy
variable (FOREIGN).
FLUCTUATION—This dummy variable takes a value of 1 if there has
been a significant increase or decrease in the number of employees
in the last three years of a business.
Within the Belgian data set, we exclusively account for control parameters that
are available for all survey years. Hence, we include the following variables:
CCOST—Logarithm of compliance costs: the compliance costs
are defined as sum of personnel costs and external costs. The sum
is calculated if there are no missing values. Amounts in Belgian
francs are converted into euro. Inflation effects are controlled by
the year dummies.
Tax Compliance Costs: Authority Behavior and Taxpayer Services
SIZE—Businesses in the independents surveys do not have
employees. For that reason, we deploy the logarithm of turnover
as size measure.
ADMINISTRATION—Set of dummy variables regarding
the following administrative issues: AGENCY, CONTACT,
ANSWER, MOTIVATION, DELAY, CONTRADICTION and
INFORMATION. We account for positive ratings (1, 2) and negative
ratings (4, 5) by a separate dummy variable.
LEGISLATION—Set of dummy variables regarding the following
legislative issues: ADVANCE, UNDERSTANDABILITY,
OBJECTIVE, ADAPTION, TIME, COHERENCY and ENTROPY.
We account for positive ratings (1, 2) and negative ratings (4, 5) by
a separate dummy variable.
INDEPENDENT—The dummy variable controls for the requested
group of the survey. It takes a value of 1 (0) for an independents
(enterprises) survey.
YEAR—Set of dummy variables: we consider dummies for 2002,
2004 and 2006 to control for time series effects.
INDUSTRY—Set of dummy variables: within the enterprises
survey, we control only for industrial businesses (EINDUSTRIAL).
Regarding independents, we control in addition to IINDUSTRIAL
for the primary sector (IPRIMARY) and construction
(ICONSTRUCTION). The other independents are active in the
services sector.
REGION—Within Belgium, the tax administration is at least for
the major business taxes (Business Income Tax, VAT) organized by
the Federal Public Service Finance. Therefore, it seems appropriate
to control for businesses in WALLONIA and BRUSSELS. The
remaining businesses are located in Flanders.
ADVICE—Variables measuring the use of external advice:
similar to Germany, we utilize the logarithm of the proportion
of external costs to total compliance costs increased by 1 percent
(OUTSOURCING). Due to a considerable number of observations
without any external advice in the Belgian data, we additionally
take into account a dummy variable for businesses without external
advice (INHOUSE).
47
48
Eichfelder, Kegels, and Schorn
Taking into account that our rating variables describe aspects that are closely
related to each other, a complete model may also give a mistakable picture due to
an “overspecification” of the model. That holds especially for Belgium, where we
analyze the influence of 14 different aspects of tax administration and legislation.
Therefore, we initially analyze the isolated effects of each rating variable on the
compliance cost burden. In these basic models, we account for all other control
variables, but do not consider any further rating variables. In the following, we develop an extended model, including all variables on administrative and legislative
quality for which we account.
The differences between mean and median values in Table 1, Table 3, and Table
4 document that our data contains a number of potential outliers, which could
bias our regression results. Taking into account that misperceptions of compliance costs by the requested businesses are possible, it seems appropriate to exclude
cases with unusually high or low cost values. We estimate a simplified regression
including only a limited set of variables and exclude all cases where the residuals
exceed two standard deviations of the estimated standard residual: 15
CCOST = α0 + α1 ∙ SIZE + α2 ∙ Y + ε
(2)
In the German data set, we identified 39 outliers for tax-related costs and 22
outliers for social insurance-related costs. In the case of Belgium, 60 cases were
excluded as outliers.
As documented by Hudson and Godwin (2000) and Eichfelder and Schorn (2009),
heteroscedasticity may be a problem regarding the estimation of tax compliance costs.
For that reason, we use robust standard errors for our econometric analysis.
Results
German Data
Table 9 documents the basic model for the German data set. Hence, each correlation coefficient results from a different regression controlling exclusively for one
rating variable.
49
Tax Compliance Costs: Authority Behavior and Taxpayer Services
TABLE 9. German Data: Basic Model
Model
Positive Rating
Variable
Coefficient
QUALIFICATION TAX
−0.038
0.085
0.3631
SERVICE TAX
−0.247**
0.099
0.3717
PROCESSING TAX
0.021
0.104
QUALIFICATION SIA
0.016
SERVICE SIA
PROCESSING SIA
RSTD
DEV
Negative Rating
RSTD
DEV
R2
Cases
0.309***
0.119
0.3706
544
0.089
0.085
0.3656
535
0.3620
0.056
0.083
0.3625
542
0.103
0.3727
−0.052
0.140
0.3729
485
−0.171
0.106
0.3784
0.182
0.114
0.3786
484
0.053
0.106
0.3769
0.146
0.114
0.3787
483
R2
Coefficient
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related or social insurance-related
compliance costs. Each correlation coefficient represents one OLS regression including the control parameters
described by the section “model specification.” Case numbers are identical for models with positive and negative
dummies for rating behavior.
We find significant effects for the qualification and service orientation of the
tax authorities. By contrast, there is no significant effect for the ratings on the
social insurance administration (SIA) behavior. However, including all rating variables into one regression, we obtain the following results for selected variables (see
Appendix A for the complete regression results):
TABLE 10. German Data: Extended Model
Model
Positive Rating
Negative Rating
0.3734
0.3686
R2
Cases
526
Variable
QUALIFICATION TAX
526
Coefficient
RSTD DEV
Coefficient
RSTD DEV
0.061
0.095
0.242*
0.128
−0.360***
0.129
0.026
0.094
PROCESSING TAX
0.190
0.123
0.025
0.091
SIZE
0.339***
0.031
0.344***
0.032
Constant
4.360***
0.451
4.118***
SERVICE TAX
R2
0.3776
Cases
QUALIFICATION SIA
0.464
0.3765
472
472
0.135
0.140
−0.263
0.176
−0.353**
0.157
0.227
0.152
PROCESSING SIA
0.190
0.146
0.093
0.141
SIZE
0.375***
0.051
0.366***
0.051
Constant
3.431***
0.561
3.327***
0.583
SERVICE SIA
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related or social insurance-related
compliance costs. We estimate separate models for positive and negative statements on administrative quality.
50
Eichfelder, Kegels, and Schorn
We find in this specification especially an effect of a positive rating for service
orientation. Contrasting Table 9, this holds not only for taxes, but also for the
social insurance administration. Furthermore, there exists only a relatively weak
effect of the qualification of the tax administration in this model. Similar to Table
9, we do not find any significant effect of the processing time.
According to our estimate, the compliance cost burden of businesses with a
positive rating for service orientation is about 30 percent lower compared to the
other businesses in our data.16 From this perspective, there is evidence for a significant and substantial reduction of compliance costs resulting from a serviceoriented administration approach.
The effect of authority behavior on the compliance costs of private businesses
may vary with business size. Therefore, we estimate the extended model for small
businesses (less than 50 associates including the entrepreneur) and medium and
big businesses (50 and more associates including the entrepreneur). Regarding
small businesses, we obtain the following selected results:
TABLE 11. German Data: Extended Model for Small Businesses
Model
Positive Rating
Negative Rating
0.3106
0.2997
319
319
R2
Cases
Variable
Coefficient
RSTD DEV
Coefficient
RSTD DEV
0.307**
0.124
QUALIFICATION TAX
−0.010
0.110
SERVICE TAX
−0.436***
0.148
−0.074
0.107
PROCESSING TAX
0.340**
0.133
0.074
0.104
SIZE
0.327***
0.052
0.328***
0.053
Constant
4.654***
0.701
4.486***
R2
0.3469
Cases
QUALIFICATION SIA
0.705
0.3466
262
262
0.120
0.177
−0.105
0.212
−0.158
0.175
−0.089
0.190
PROCESSING SIA
0.082
0.173
0.172
0.165
SIZE
0.442***
0.101
0.432***
0.100
Constant
4.302***
0.696
4.302***
0.703
SERVICE SIA
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related or social insurance-related
compliance costs. We estimate separate models for positive and negative statements on administrative quality.
Unlike in the overall data set, we do not find a significant correlation of ratings regarding the social insurance administration. However, the effect of the tax
authority behavior seems to be stronger than in the overall data. A negative rating
for QUALIFICATION TAX yields an increase in the compliance cost burden of
about 34 percent, while a positive rating for SERVICE TAX is connected with a
cost reduction of about 35 percent.
51
Tax Compliance Costs: Authority Behavior and Taxpayer Services
Contrasting our hypotheses, we also find a positive and significant correlation
for a positive rating of processing time of the tax administration (PROCESSING
TAX). This should be interpreted cautiously. Taking into account the results for the
overall data set (Table 9 compared to Table 10), this outcome is especially driven
by the interaction of PROCESSING TAX and SERVICE TAX. In an isolated analysis excluding any further rating variables, the coefficient of PROCESSING TAX
would be 0.090 and not significant. In this setting, we would also obtain a lower,
but still significant, coefficient for a positive rating of SERVICE TAX (−0.299).
Therefore, also the very high value of the correlation coefficient in Table 11 for this
variable should be interpreted cautiously. Due to the interdependency of the different rating variables, a corresponding estimation problem does not seem to be
a big surprise.
Regarding the medium and big businesses in the German data set, we obtain
the following selected regression results:
TABLE 12. German Data: Extended Model for Medium and Big Businesses
Model
Positive Rating
Negative Rating
0.3298
0.3248
207
207
R2
Cases
Variable
Coefficient
RSTD DEV
Coefficient
RSTD DEV
0.181
0.183
0.024
0.357
−0.292
0.244
0.151
0.179
PROCESSING TAX
0.034
0.239
−0.043
0.173
SIZE
0.414***
0.059
0.411***
Constant
2.316**
0.941
2.326**
QUALIFICATION TAX
SERVICE TAX
R2
0.3450
Cases
QUALIFICATION SIA
0.059
0.963
0.3532
210
210
0.183
0.222
−0.569*
0.303
PROCESSING SIA
0.423
0.277
SIZE
0.312***
0.101
0.317***
0.101
Constant
5.030***
0.756
4.919***
0.785
SERVICE SIA
−0.363
0.588**
−0.018
0.313
0.242
0.239
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related or social insurance-related
compliance costs. We estimate separate models for positive and negative statements on administrative quality.
Contrasting Table 11, we find significant effects for the service orientation of the
social insurance administration, but no substantial impact of tax authority behavior. Businesses with a negative rating for SERVICE SIA bear on average an about
80 percent cost higher burden than comparable observations.
Hence, while the compliance costs of small businesses seem to be driven by the
qualification and service orientation of tax authorities, medium and big businesses should be especially affected by the behavior of the German social insurance
52
Eichfelder, Kegels, and Schorn
administration. From this perspective, the effect of administrative issues on compliance costs depends on business size.
Belgian Data
Similar to the German case, we initially analyze the isolated effects of each rating
variable on the compliance cost burden. We obtain under these conditions the
following results:
TABLE 13. Belgian Data: Basic Model
Model
Variable
Positive Rating
Coefficient
RSTD
DEV
Negative Rating
R2
Coefficient
RSTD
DEV
R2
Cases
1,003
AGENCY
−0.049
0.062
0.6550
0.070
0.062
0.6552
CONTACT
−0.024
0.063
0.6579
0.073
0.063
0.6583
997
ANSWER
−0.129**
0.063
0.6554
0.175***
0.063
0.6566
1,000
MOTIVATION
−0.094
0.064
0.6540
0.145**
0.063
0.6551
1,001
DELAY
−0.175***
0.062
0.6584
0.225***
0.064
0.6599
1,000
CONTRADICTION
−0.013
0.068
0.6532
0.085
0.064
0.6538
993
INFORMATION
−0.220***
0.063
0.6578
0.315***
0.065
0.6618
997
ADVANCE
−0.067
0.064
0.6544
0.150**
0.062
0.6560
993
UNDERSTANDABILITY
−0.241***
0.069
0.6576
0.270***
0.067
0.6590
1,001
OBJECTIVE
−0.076
0.066
0.6550
0.134**
0.064
0.6561
993
ADAPTION
−0.180**
0.075
0.6539
0.232***
0.066
0.6560
994
TIME
−0.190***
0.062
0.6579
0.218***
0.061
0.6590
1,001
COHERENCY
−0.231***
0.072
0.6572
0.299***
0.064
0.6610
998
ENTROPY
−0.215***
0.064
0.6566
0.260***
0.062
0.6587
999
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related or social insurance-related
compliance costs. Each correlation coefficient represents one OLS regression including the control parameters
described by the section “model specification.” Case numbers are identical for models with positive and negative
dummies for rating behavior.
Similar to the German case, we find positive regression coefficients for negative ratings and an opposite result for positive ratings. Hence, we can determine
a significant and negative correlation between rating behavior and the burden of
red tape for administrative and legislative issues. This demonstrates clearly the hypothesized influence of tax administration and tax legislation on the compliance
cost burden. As we find highly significant coefficients for most administrative aspects,
the evidence is clearly stronger than in the German case. This could partially result
from the higher number of observations in the Belgian data.
Furthermore, the effect of administrative issues depends on the considered
aspect. While problems of finding and contacting the correct agency (AGENCY,
53
Tax Compliance Costs: Authority Behavior and Taxpayer Services
CONTACT) do not seem to have a considerable effect, we can state a significantly
lower cost burden for businesses who obtained precise answers (ANSWER) and
the requested information (INFORMATION) in a reasonable time (DELAY).
We find also significant effects for the motivation of administrative decisions
(MOTIVATION), while potential contradictions between the statements of administrative staff members (CONTRADICTION) do not seem to be a major problem.
The same holds for the included aspects of tax legislation. We find an
especially strong correlation for the UNDERSTANDABILITY and the
COHERENCY of the tax law. By contrast, a clear OBJECTIVE of tax regulations does not seem to be an important cost driver. The evidence for the announcement of tax regulations is mixed. While there is relatively weak effect
of ADVANCE (compared to the other rating variables), we find a strong correlation of TIME. Furthermore, we also find a highly significant effect for the
information content of tax regulations (ENTROPY).
Including all rating variables in one regression, we obtain the following
outcome:
TABLE 14. Belgian Data: Extended Model
Model
Positive Rating
Negative Rating
0.6630
0.6674
937
937
R2
Cases
Variable
AGENCY
CONTACT
ANSWER
MOTIVATION
DELAY
CONTRADICTION
INFORMATION
ADVANCE
UNDERSTANDABILITY
OBJECTIVE
ADAPTION
TIME
COHERENCY
ENTROPY
SIZE
Constant
Coefficient
RSTD DEV
Coefficient
RSTD DEV
0.030
0.103
0.016
0.018
−0.121
0.066
−0.160*
0.050
−0.171**
0.066
−0.022
−0.149**
−0.097
−0.080
0.281***
5.397***
0.077
0.078
0.077
0.077
0.074
0.074
0.085
0.070
0.081
0.079
0.088
0.074
0.090
0.076
0.019
0.293
−0.062
−0.086
−0.032
−0.023
0.144*
−0.060
0.229***
0.013
0.143*
−0.043
0.001
0.137*
0.154*
0.067
0.279***
4.920***
0.076
0.078
0.079
0.078
0.074
0.073
0.086
0.069
0.081
0.077
0.081
0.074
0.083
0.075
0.019
0.295
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. We estimate
separate models for positive and negative statements on administrative and legislative quality.
In the extended model, we find significant effects only for a part of the considered rating variables. This result should be interpreted with caution. In the
extended model, we include 14 different statements on administrative and legislative quality into one model. As these variables measure similar issues, it should
54
Eichfelder, Kegels, and Schorn
not be expected to obtain significant results for each aspect. By contrast, only the
strongest and most important issues should “survive” into such a kind of setting.
However, that allows us to draw conclusions on the most important cost drivers
in the case of Belgium.
We can state that businesses, which were not convinced to receive the requested information, bear an about 26 percent higher compliance burden.
Furthermore, we find a slightly significant increase of 15 percent in the cost burden for businesses who received the requested information with an unexpected
delay. From this perspective, it seems to be essential that private businesses are
supported with the necessary information by the administrative authorities in a
reasonable time.
The model exemplifies further that the impact of administrative issues may
be separated from the effect of the tax law itself. Taking into account legislative
issues, we find significant effects of the UNDERSTANDABILITY and the timely
announcement (TIME) of tax regulations. Furthermore, there is also an effect
resulting from the COHERENCY of tax legislation. The relatively low significance of these variables results from the fact that they measure similar aspects.
Hence, the model seems to be in some way “overspecified.”
Similar to the German data, we made separate regressions for different size
classes. Due to the structure of the data set, we differentiate between independents (sole proprietorships) and enterprises (generally in form of a corporation).
In contrast to Germany, we observed considerable differences between the basic
model and the extended model resulting from the high number of rating variables. Therefore, it seems appropriate to estimate basic models and extended
models for independents surveys and enterprises surveys.
In case of the independents we obtain the following outcome (Table 15 and
Table 16) for the basic model and the extended model. The results are similar
to the overall data set. However, the evidence is stronger. Regarding the basic
model, we obtain significant results for all administrative rating variables (for
CONTRADICTION only on a 10-percent level). In the extended model, the estimated coefficients for DELAY and INFORMATION are higher than in the overall data. According to our estimate, problems regarding INFORMATION result
in an increase of the cost burden by about 39 percent. An unexpected DELAY
adds further 26 percent.
Again, we find a considerable difference between the basic and the extended
model results. Due to the high number of rating variables in the extended model, we observe only significant effect for the main aspects. Thus, we can conclude
that a timely and accurate information of small businesses is the most important subject regarding tax administration, while the coherency of tax regulations
seem to be the main issue for the ratings on tax legislation. Businesses with a
negative rating regarding this aspect bear on average an about 28 percent higher
cost burden.
55
Tax Compliance Costs: Authority Behavior and Taxpayer Services
TABLE 15. Belgian Data: Basic Model for Independents
Model
Variable
Positive Rating
Coefficient
RSTD
DEV
Negative Rating
R2
Coefficient
RSTD
DEV
R2
Cases
AGENCY
−0.218***
0.083
0.4259
0.255***
0.083
0.4291
472
CONTACT
−0.186**
0.086
0.4252
0.265***
0.085
0.4316
472
ANSWER
−0.139
0.086
0.4190
0.239***
0.086
0.4257
469
MOTIVATION
−0.230***
0.087
0.4245
0.318***
0.085
0.4335
472
DELAY
−0.263***
0.084
0.4314
0.369***
0.086
0.4419
471
CONTRADICTION
−0.011
0.089
0.4166
0.145*
0.085
0.4203
469
INFORMATION
−0.327***
0.083
0.4369
0.451***
0.085
0.4521
469
ADVANCE
−0.043
0.090
0.4176
0.165*
0.085
0.4221
472
UNDERSTANDABILITY
−0.338***
0.098
0.4316
0.348***
0.092
0.4344
473
OBJECTIVE
−0.198**
0.090
0.4216
0.261***
0.085
0.4272
470
ADAPTION
−0.173
0.110
0.4169
0.227**
0.093
0.4215
469
TIME
−0.220***
0.083
0.4281
0.244***
0.082
0.4302
474
COHERENCY
−0.366***
0.094
0.4365
0.406***
0.086
0.4464
472
ENTROPY
−0.151
0.092
0.4213
0.243***
0.086
0.4281
472
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. Each correlation coefficient represents one OLS regression including the control parameters described by the section “model
specification.” Case numbers are identical for models with positive and negative dummies for rating behavior.
TABLE 16. Belgian Data: Extended Model for Independents
Model
Positive Rating
Negative Rating
0.4582
0.4697
449
449
R2
Cases
Variable
AGENCY
CONTACT
ANSWER
MOTIVATION
DELAY
CONTRADICTION
INFORMATION
ADVANCE
UNDERSTANDABILITY
OBJECTIVE
ADAPTION
TIME
COHERENCY
ENTROPY
SIZE
Constant
Coefficient
RSTD DEV
Coefficient
RSTD DEV
−0.127
0.033
0.157
−0.099
−0.153
0.107
−0.256**
0.120
−0.199*
0.001
0.122
−0.167
−0.262**
−0.168
0.240***
5.509***
0.097
0.105
0.118
0.123
0.103
0.101
0.118
0.098
0.121
0.107
0.129
0.104
0.126
0.109
0.029
0.370
0.052
−0.023
−0.150
0.071
0.231**
−0.066
0.329***
0.003
0.155
0.028
−0.105
0.130
0.244**
−0.140
0.235***
4.934***
0.097
0.105
0.121
0.126
0.102
0.099
0.120
0.091
0.120
0.104
0.114
0.099
0.118
0.109
0.029
0.362
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. We estimate
separate models for positive and negative statements on administrative and legislative quality.
56
Eichfelder, Kegels, and Schorn
In terms of the enterprises surveys, we obtain the following regression results
for the basic model and the extended model:
TABLE 17. Belgian Data: Basic Model for Independents
Model
Variable
Positive Rating
Coefficient
RSTD
DEV
Negative Rating
R2
Coefficient
RSTD
DEV
R2
Cases
AGENCY
0.077
0.090
0.4047
−0.079
0.090
0.4048
531
CONTACT
0.083
0.092
0.4100
−0.069
0.092
0.4098
525
ANSWER
−0.166*
0.091
0.4078
0.159*
0.092
0.4074
531
MOTIVATION
−0.005
0.094
0.4036
0.012
0.091
0.4036
529
DELAY
−0.120
0.091
0.4073
0.121
0.091
0.4072
529
CONTRADICTION
−0.033
0.101
0.4017
0.027
0.092
0.4016
524
INFORMATION
−0.172*
0.095
0.4067
0.224**
0.095
0.4092
528
ADVANCE
−0.121
0.091
0.4075
0.151*
0.090
0.4086
521
UNDERSTANDABILITY
−0.164*
0.097
0.4076
0.195**
0.095
0.4091
528
OBJECTIVE
0.018
0.095
0.4050
0.028
0.093
0.4051
523
ADAPTION
−0.211**
0.103
0.4086
0.246***
0.094
0.4111
525
TIME
−0.204**
0.090
0.4124
0.229**
0.090
0.4140
527
COHERENCY
−0.099
0.110
0.4059
0.189**
0.096
0.4091
526
ENTROPY
−0.288***
0.089
0.4134
0.282***
0.089
0.4136
527
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. Each correlation coefficient represents one OLS regression including the control parameters described by the section “model
specification.” Case numbers are identical for models with positive and negative dummies for rating behavior.
TABLE 18. Belgian Data: Extended Model for Independents
Model
Positive Rating
Negative Rating
0.4388
0.4387
488
488
R2
Cases
Variable
AGENCY
CONTACT
ANSWER
MOTIVATION
DELAY
CONTRADICTION
INFORMATION
ADVANCE
UNDERSTANDABILITY
OBJECTIVE
ADAPTION
TIME
COHERENCY
ENTROPY
SIZE
Constant
Coefficient
RSTD DEV
Coefficient
RSTD DEV
0.142
0.188
−0.108
0.058
−0.057
0.014
−0.126
−0.005
−0.107
0.170
−0.167
−0.163
0.053
−0.267***
0.302***
5.041***
0.112
0.110
0.104
0.102
0.104
0.108
0.126
0.102
0.109
0.115
0.117
0.108
0.123
0.102
0.025
0.356
−0.159
−0.178
0.079
−0.076
0.052
−0.051
0.193
0.010
0.115
−0.142
0.115
0.183*
0.041
0.208**
0.299***
4.715***
0.112
0.112
0.109
0.103
0.104
0.106
0.127
0.104
0.111
0.111
0.111
0.111
0.117
0.104
0.025
0.373
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. We estimate
separate models for positive and negative statements on administrative and legislative quality.
Tax Compliance Costs: Authority Behavior and Taxpayer Services
57
Exclusively in the basic model, we can observe significant effects for administrative issues (INFORMATION and ANSWER) that are in most cases on a 10-percent level. Hence, we find only weak evidence for a correlation between ratings on
administrative quality and tax compliance costs. By contrast, there is significant
evidence for an impact of legislative issues that can be observed in the basic model
as well as in the extended model.
Nevertheless, there are also considerable differences regarding legislative issues. In the overall data set, we observed especially strong effects for TIME,
COHERENCY, and UNDERSTANDABILITY, while for enterprises ENTROPY,
TIME, and ADAPTION seem to be the most important influence factors. In addition to the results for administrative issues, this may be interpreted as evidence
for considerable differences in the tax administration process of small businesses
compared to the bigger size classes.
Conclusion
In our paper, we analyzed empirically the effects of authority behavior and taxpayer services on the compliance costs of private businesses in Germany and
Belgium. Using ratings of survey participants as measure for administrative quality, we found evidence for a considerable reduction of compliance burdens by a
customer-oriented administration approach.
Due to the fact that the data contain a considerable number of rating aspects,
we estimated a basic model concentrating on one rating aspect and an extended
model including all rating parameters. While the basic models documents the effects in general (with the rating variable interpreted as proxy for administration or
legislation), the extended model gives evidence, which aspects are the main cost
drivers. Based on the extended model, we could also demonstrate that the effect of
tax administration may be separated from the effect of tax legislation.
A positive rating for the service orientation of the German tax and the German
social insurance administration results on average in a cost reduction of about
30 percent. In terms of small businesses, we also found a significant effect for the
perceived qualification of the tax authorities. Problems to obtain the requested
information in Belgium yield on average an increase in the cost burden of about
26 percent. An unexpected delay adds further 15 percent.
According to our results, service orientation and especially an accurate handling of the taxpayers’ requests is an important issue in reducing the overall
burden of red tape. By contrast, we find only in case of the Belgian independents
(sole proprietorships) significant results for problems of getting access to the
authorities (AGENCY and CONTACT in the basic model). We also do not find
evidence that contradictions between statements of administrative staff members are an important problem from a compliance cost perspective. Within the
58
Eichfelder, Kegels, and Schorn
extended models, we do not find a significant correlation for clear answers to
taxpayers’ requests (ANSWER) and the motivation of administrative decisions
(MOTIVATION). However, these aspects could be implicitly included within
the INFORMATION variable.
Considering the processing time of administrative authorities, we find contradictory evidence. On the one hand there is no convincing effect of PROCESSING
in the German data set. On the other hand DELAY significantly affects the compliance costs of Belgian businesses and especially independents. The ambiguity of
the results may be explained by the different wording of the survey questionnaires.
The Belgian questionnaire asks explicitly for an unexpected delay if taxpayers try
to obtain information from the authorities. By contrast, PROCESSING in the
German questionnaire describes an abstract operating time. From this perspective, a long processing time seems only to be a problem if it prolongs latency time
and increases the taxpayers’ uncertainty.
We find a stronger effect of tax authority behavior on the compliance costs of
smaller size classes in both countries. This should be due to the lower information capacity of small businesses compared to the bigger size classes. While big
businesses do not depend on the advice of the authorities, small businesses have
only limited resources to spend on tax compliance and information requirements.
Therefore, it is especially important that they are supported by the revenue service.
Taking into account the economies of scale within the compliance process, improving taxpayer services for this group could be a promising measure to enhance
the productivity of the overall tax system.
In terms of the bigger size classes, we find only weak evidence for an impact
of tax authority behavior on the burden of red tape. However, there are significant effects regarding the service orientation of the German social insurance
administration. This could result from specific aspects of withholding taxes on
wage income. The complexity and automatization of payroll accounting should
increase in the number of employees. Potential problems result for example
from tax incentives for retirement plans, fringe benefits (childcare, free lunch,
free transportation, company car, etc.) or working-time accounts that are typically an issue for the bigger size classes.
Therefore, big businesses depend to a higher degree on the service orientation of the social insurance authorities including the compatibility of the corresponding processes. From this perspective, service orientation could be especially important for employment taxes and payroll taxation. However, the
evidence is not very strong. Therefore, further research seems to be necessary
regarding this aspect.
The Belgian data contain also information on legislative variables that could be
of interest from an administrative perspective. In addition to the understandability and coherency of the tax law, we found a significant effect for TIME. Therefore,
the announcement and timing of new tax regulations should have a considerable
Tax Compliance Costs: Authority Behavior and Taxpayer Services
59
impact on the burden of red tape. Regarding the bigger size classes also the information content of fiscal regulations (denoted as ENTROPY) seems to affect the
burden of red tape. Hence, the administrative burden of the bigger size classes
does not seem to result from a lack of the understandability of the tax law (like
in the case of independents) but from the limited information content of specific
regulations. This emphasizes the higher information capacity of big businesses.
Our results imply that a service-oriented approach of tax administration significantly reduces the burden of private businesses to comply with the tax law.
Within the administration process, it seems essential to provide timely and accurate information for the taxpayer. That includes especially that taxpayers’ requests are answered carefully and without an unexpected or unreasonable delay.
Furthermore, taxpayers should be informed about new regulations in sufficient
time to arrange their affairs properly. Tax regulations should be easy to understand
and coherent to other regulations in the legal system. From this perspective, time
pressure in the process of fiscal legislation should result in an unnecessarily high
burden of red tape.
Administrative issues seem to be most important for self-employed people
and small businesses. Therefore, enhancing taxpayer services for this target
group could be a promising step to optimize the productivity of the tax system.
Taking into account the literature on linkages between compliance costs and tax
evasion (Hasseldine 2001, Erard and Ho 2003), a reduction of the burden of red
tape by administrative actions could also increase the overall compliance with
the tax system.
Acknowledgements
We are very thankful for the support of our research by the Federal Planning Bureau
in Brussels and the German Ministry of Economics. Furthermore, we would like
to thank the organizers of the IRS Research Conference for a perfect conference in
Washington D.C. Especially we are thankful to Martha Gangi, Marcella Garland,
Mellissa Kovalick, Kara Leibel, Lisa Smith, and our discussant James E. Nunns for
helpful advice and support.
References
Allers, M. (1994), Administrative and Compliance Costs of Taxation and Public
Transfers in the Netherlands, Wolters-Noordhoff: Groningen.
Barton, T. (2001), Supporting the Tax System by Improving Taxpayer Services, In
Evans, C.; Walpole, M. (Eds.), Tax administration in the 21st century, Prospect
Media: St. Leonards NSW, 191–198.
60
Eichfelder, Kegels, and Schorn
Chittenden, F.; Kauser, S.; Poutzouris, P. (2005), PAYE–NIC Compliance Costs:
Empirical Evidence from the UK SME Economy, International Small Business
Journal, Vol. 23, 635–656.
Delgado Lobo, M.L.; Salinaz-Jimenez, J.; Sanz Sanz, J.F. (2001), Hidden Tax
Burden of the Personal Income Tax: Evidence from the recent tax reform in
Spain, Australian Tax Forum, Vol. 16, 463–482.
DeLuca, D.; Stilmar, S.; Guyton, J.; Lee, W.–L.; O’Hare, J. (2007), Aggregate
Estimates of Small Business Taxpayer Compliance Burden, SOI Tax
Stats—2007 IRS Research Conference, 147–184.
De Vil, G.; Kegels, C. (2002), Les Charges Administratives en Belgique pour
l’année 2000, Planning Paper No. 92, Federal Planning Bureau: Brussels.
Eichfelder, S.; Schorn, M. (2009), Tax Compliance Costs: A Business
Administration Perspective, Diskussionsbeiträge des Fachbereichs
Wirtschaftswissenschaft der Freien Universität Berlin-Betriebswirtschaftliche
Reihe, No. 2009/3, Free University: Berlin.
Erard, B.; Ho, C.–C. (2003), Explaining the U.S. Income Tax Compliance
Continuum, eJournal of Tax Research, Vol. 1, 93–105.
European Communities (Eds.) (2004), European Tax Survey, Working Paper
No. 3/2004, Office for Official Publications of the European Communities:
Luxembourg.
Evans, C. (2003), Studying the Studies: An Overview of Recent Research into
Taxation Operating Costs, eJournal of Tax Research, Vol. 1, 64–92.
Freedman, J.; Loomer, G.; Vella, J. (2009), Analyzing the Enhanced Relationship
Between Corporate Taxpayers and Revenue Authorities: A U.K. Case Study,
SOI Tax Stats—2009 IRS Research Conference, 103–148.
Gale, W.; Holtzblatt, J. (2002), The Role of Administrative Issues in Tax Reform:
Simplicity, Compliance and Administration, In Zodrow, G.; Mieszkowski, P.
(Eds.), United States Tax Reform in the 21st Century, Cambridge University
Press: Cambridge, 179–214.
Hasseldine, J. (2001), Linkages Between Compliance Costs and Taxpayer
Compliance Research, In Evans, C.; Pope, J.; Hasseldine, J. (Eds.), Tax
Compliance Costs: A Festschrift for Cedric Sandford, Prospect Media: St.
Leonards NSW, 3–14.
Hansford, A.; Hasseldine, J., (2002), Best Practice in Tax Administration, Public
Money and Management, Vol. 22, 5–6.
Hansford, A.; Hasseldine, J.; Howorth, C. (2003), Factors Affecting the Costs of
UK VAT Compliance for Small and Medium Sized Enterprises, Environment
and Planning C: Government and Policy, Vol. 21, 479–492.
Tax Compliance Costs: Authority Behavior and Taxpayer Services
61
Hudson, J.; Godwin, M.R. (2000), The Compliance Costs of Collecting Direct
Taxes in the UK: An Analysis of PAYE, Journal of Public Economics, Vol. 77,
29–44.
Janssen, L.; Kegels, C.; Verschueren, F. (2006), Les Charges Administratives en
Belgique pour l'année 2004, Planning Paper No. 100, Federal Planning Bureau:
Brussels.
Joos, A.; Kegels, C. (2004), Les Charges Administratives en Belgique pour l'année
2002, Planning Paper No. 94, Federal Planning Bureau: Brussels.
Kayser, G.; Clemens, R.; Wolter, H.–J.; Schorn, M. (2004), Bürokratiekosten
kleiner und mittlerer Unternehmen—Gutachten im Auftrag des
Bundesministeriums für Wirtschaft und Arbeit, Deutscher Universitäts-Verlag:
Wiesbaden.
Kegels, C. (2008), Les Charges Administratives en Belgique pour l'année 2006,
Planning Paper No. 103, Federal Planning Bureau: Brussels.
Kirchler, E. (2007), The Economic Psychology of Tax Behaviour, Cambridge
University Press: Cambridge.
Klein-Blenkers, F. (1980), Die Belastung der Industrieunternehmen durch
administrative Leistungen für den Staat—unter besonderer Berücksichtigung
kleiner und mittlerer Unternehmen, Otto Schwarz: Göttingen.
Klun, M.; Blažić, H. (2005), Tax Compliance Costs for Companies in Slovenia
and Croatia, Finanz-Archiv, Vol. 61, 418–437.
Murphy, K. (2004), The Role of Trust in Nurturing Compliance: A study of
Accused Tax Avoiders, Law and Human Behavior, Vol. 28, 187–209.
McCubbin, J.G. (2004), Optimal Tax Enforcement: A Review of the Literature and
Practical Implications, OTA Working Paper No. 90, Office of Tax Analysis—
Department of the Treasury, Washington D.C.
OECD (Eds.) (2001), Businesses' Views on Red Tape: Administrative and
Regulatory Burdens on Small and Medium Enterprises, OECD: Paris.
Oster, C.V.; Lynn, A.D. (1955), Compliance Costs of the Ohio Axle Mile Tax: A
Case Study, National Tax Journal, Vol. 8, 209–214.
Rametse, N.; Pope, J. (2002), Start-Up Tax Compliance Costs of the GST:
Empirical Evidence from Western-Australian Small Businesses, Australian Tax
Forum, Vol. 17, 407–422.
Sandford, C.T.; Godwin, M.R.; Hardwick, P.J. (1989), Administrative and
Compliance Costs of Taxation, Fiscal Publications: Bath.
Slemrod, J.; Venkatesh, V. (2002), The Income Tax Compliance Cost of Large and
Mid-Size Businesses: A Report to the IRS LMSB Division, Working Paper No.
914, Ross School of Business/University of Michigan: Ann Arbor.
62
Eichfelder, Kegels, and Schorn
Slemrod, J.; Yitzhaki, S. (2002), Tax Avoidance, Evasion and Administration, In
Auerbach, A.J.; Feldstein, M. (Eds.), Handbook of Public Economics, Vol. 3,
Elsevier: Amsterdam et al., 1423–1470.
Tate, A.A. (1988), Value Added Tax—International Practice and Problems,
International Monetary Fund: Washington D.C.
Vaillancourt, F.; Clemens, J. (2008), Compliance and Administrative Costs of
Taxation in Canada, In Clemens, J. (Ed.), The Impact and Cost of Taxation
in Canada: The Case for Flat Tax Reform, The Fraser Institute: Calgary et al.,
55–102.
Wicks, J.H. (1965): Taxpayer compliance costs from the Montana business tax,
Montana Business Quarterly, Vol. 3, No. 3, 36–42.
Appendices
Appendix A: Complete Regression Results
Regarding the German data set, we obtain the following complete results for the
extended model of tax administration:
TABLE A1. German Data: Complete Results for Tax Administration
Model
Positive Rating
Negative Rating
0.3734
526
0.3686
526
R2
Cases
Variable
Coefficient
QUALIFICATION TAX
SERVICE TAX
PROCESSING TAX
SIZE
EMPLOYMENT
INDUSTRIAL
CONSTRUCTION
TRADE
ESERVICE
PROFESSION
CRAFTS
INDEPENDENT
PARTNERSHIP
GMBH & CO. KG
OUTSOURCING
AGE
E-FILING
E-FILING PROBLEM
CASH ACCOUNTING
Constant
0.061
−0.360***
0.190
0.339***
−0.407**
−0.001
−0.230*
0.036
0.282**
−0.050
0.159
0.115
0.028
0.044
−0.270***
0.009
−0.034
0.093
−0.351
4.360***
RSTD
DEV
0.095
0.129
0.123
0.031
0.182
0.145
0.136
0.142
0.133
0.116
0.116
0.146
0.178
0.159
−0.056
0.045
0.117
0.214
0.240
0.451
VIF Value Coefficient
1.24
1.58
1.44
1.80
1.32
1.70
2.20
1.51
1.42
1.49
1.83
1.40
1.08
1.18
1.03
1.58
1.24
1.24
1.53
—
0.242*
0.026
0.025
0.344***
−0.341*
0.014
−0.204
0.047
0.265**
−0.035
0.137
0.123
0.043
0.032
−0.273***
0.011
−0.024
0.118
−0.351
4.118***
RSTD
DEV
VIF Value
0.128
0.094
0.091
0.032
0.175
0.143
0.139
0.146
0.133
0.118
0.117
0.144
0.175
0.159
0.056
0.046
0.117
0.212
0.235
0.464
1.14
1.24
1.17
1.81
1.32
1.70
2.21
1.52
1.44
1.47
1.83
1.40
1.09
1.18
1.03
1.60
1.23
1.25
1.53
—
Regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a 1-percent,
5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. We estimate separate
models for positive and negative statements on administrative quality.
63
Tax Compliance Costs: Authority Behavior and Taxpayer Services
In terms of the German social insurance administration, we obtain the following complete results for the extended model (overall data set):
TABLE A2. German Data: Complete Results for Social Insurance Administration
Model
Positive Rating
Negative Rating
0.3776
0.3765
472
472
R2
Cases
Variable
QUALIFICATION SIA
Coefficient
RSTD
DEV
VIF Value Coefficient
RSTD
DEV
VIF Value
0.135
0.140
1.62
−0.263
0.176
1.43
−0.353**
0.157
1.94
0.227
0.152
1.63
PROCESSING SIA
0.190
0.146
1.78
0.093
0.141
1.51
SIZE
0.375***
0.051
1.82
0.366***
0.051
1.83
EMPLOYMENT
2.737***
0.236
1.09
2.815***
0.243
1.09
SERVICE SIA
INDUSTRIAL
−0.132
0.176
1.73
−0.111
0.177
1.74
CONSTRUCTION
−0.281
0.176
2.37
−0.253
0.176
2.38
TRADE
0.097
0.177
1.55
0.107
0.177
1.55
ESERVICE
0.098
0.165
1.41
0.110
0.168
1.41
PROFESSION
−0.217
0.143
1.43
−0.209
0.143
1.43
CRAFTS
−0.016
0.134
1.77
−0.021
0.134
1.78
INDEPENDENT
−0.128
0.172
1.17
−0.110
0.175
1.18
PARTNERSHIP
0.133
0.211
1.11
0.158
0.199
1.10
GMBH & CO. KG
0.087
0.184
1.21
0.063
0.181
1.21
OUTSOURCING
−0.460***
−0.079
1.07
−0.474***
0.073
1.06
0.082
0.063
1.58
0.102
0.064
1.57
−0.057
0.121
1.28
−0.050
0.120
1.27
0.055
0.210
1.22
0.009
0.214
1.21
PART TIME
−0.051
0.041
1.27
−0.042
0.042
1.27
CASUAL
−0.028
0.039
1.13
−0.020
0.040
1.13
HANDICAPPED
−0.126*
0.065
1.17
−0.117*
0.066
1.17
TRAINEE
0.026
0.043
1.21
0.027
0.044
1.21
FOREIGN
0.137
0.107
1.13
0.133
0.108
1.13
FLUCTUATION
0.233**
0.113
1.04
0.248**
0.115
1.04
Constant
3.431***
0.561
—
3.327***
0.583
—
AGE
E-FILING
E-FILING PROBLEM
Regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a 1-percent,
5-percent, and 10-percent level; dependent variable: logarithm of social insurance-related compliance costs. We estimate separate models for positive and negative statements on administrative quality.
64
Eichfelder, Kegels, and Schorn
Regarding the Belgian case, we receive the following complete results for the
extended model (overall data set):
TABLE A3. Belgian Data: Complete Results for the Extended Model
Model
R2
Cases
Positive Rating
Negative Rating
0.6630
0.6674
937
937
Coefficient
RSTD
DEV
VIF Value
Coefficient
RSTD
DEV
VIF Value
AGENCY
0.030
0.077
1.50
−0.062
0.076
1.49
CONTACT
0.103
0.078
1.55
−0.086
0.078
1.63
ANSWER
0.016
0.077
1.57
−0.032
0.079
1.66
MOTIVATION
0.018
0.077
1.37
−0.023
0.078
1.46
−0.121
0.074
1.33
0.074
1.32
0.066
0.074
1.20
0.073
1.31
−0.160*
0.085
1.71
0.229***
0.086
1.71
0.050
0.070
1.22
0.013
0.069
1.25
−0.171**
0.081
1.50
0.143*
0.081
1.56
OBJECTIVE
0.066
0.079
1.37
−0.043
0.077
1.40
ADAPTION
−0.022
0.088
1.36
0.001
0.081
1.47
TIME
−0.149**
0.074
1.30
0.137*
0.074
1.33
COHERENCY
−0.097
0.090
1.42
0.154*
0.083
1.55
ENTROPY
−0.080
0.076
1.50
0.067
0.075
1.57
0.281***
0.019
2.63
0.279***
0.019
2.63
−0.352***
0.120
3.42
−0.340***
0.120
3.42
2002
0.105
0.084
1.65
0.130
0.084
1.67
2004
0.123
0.092
1.64
0.153*
0.093
1.65
2006
−0.339***
0.108
2.05
−0.321***
0.108
2.08
EINDUSTRIAL
0.151
0.100
1.51
0.161
0.099
1.51
IINDUSTRIAL
0.318
0.241
1.09
0.316
0.238
1.09
IBUIDING
0.016
0.102
1.41
0.036
0.101
1.42
IPRIMARY
−0.851***
0.104
1.55
−0.834***
0.103
1.56
OUTSOURCING
−0.282***
0.052
3.40
−0.270***
0.052
3.42
INHOUSE
−1.756***
0.226
3.45
−1.710***
0.228
3.47
BRUSSELS
0.041
0.101
1.58
0.060
0.098
1.57
WALLONIA
−0.136*
0.080
1.14
−0.149*
0.080
1.14
0.293
—
0.295
—
Variable
DELAY
CONTRADICTION
INFORMATION
ADVANCE
UNDERSTANDABILITY
SIZE
INDEPENDENT
Constant
5.397***
0.144*
−0.060
4.920***
Regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a 1-percent,
5-percent, and 10-percent level; dependent variable: logarithm of social insurance-related compliance costs. We estimate separate models for positive and negative statements on administrative quality.
65
Tax Compliance Costs: Authority Behavior and Taxpayer Services
Appendix B: Cross Checks
In contrast to the German data, the Belgian data set does not only include oneshot observations, but also second-shot and third-shot observations. If the effect
of administrative and legislative issues on compliance costs is especially driven
by cases with more than one observation, this could bias our regression results.
Therefore, we estimate an alternative regression ignoring second and third observations of the same case. Excluding 329 observations, we obtain in the basic model
the following results:
TABLE A4. Belgian Data (Adjusted): Basic Model
Model
Variable
Positive Rating
Coefficient
RSTD
DEV
Negative Rating
R2
Coefficient
RSTD
DEV
R2
Cases
AGENCY
−0.024
0.067
0.6736
0.030
0.067
0.6736
769
CONTACT
−0.010
0.070
0.6743
0.040
0.069
0.6745
769
ANSWER
−0.158**
0.068
0.6752
0.194***
0.069
0.6763
766
MOTIVATION
−0.108
0.069
0.6726
0.154**
0.069
0.6737
766
DELAY
−0.198***
0.067
0.6781
0.249***
0.069
0.6798
767
CONTRADICTION
−0.039
0.072
0.6724
0.108
0.069
0.6734
761
INFORMATION
−0.186***
0.067
0.6755
0.280***
0.070
0.6792
763
ADVANCE
−0.055
0.070
0.6738
0.131*
0.067
0.6752
764
UNDERSTANDABILITY
−0.243***
0.076
0.6776
0.257***
0.074
0.6784
769
OBJECTIVE
−0.076
0.072
0.6763
0.129*
0.069
0.6774
764
ADAPTION
−0.072
0.084
0.6715
0.139*
0.073
0.6727
763
TIME
−0.177***
0.067
0.6774
0.205***
0.067
0.6785
770
COHERENCY
−0.236***
0.079
0.6769
0.295***
0.071
0.6805
767
ENTROPY
−0.195***
0.070
0.6767
0.249***
0.068
0.6790
768
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. Each correlation coefficient represents one OLS regression including the control parameters described by the section “model
specification.” Case numbers are identical for models with positive and negative dummies for rating behavior.
66
Eichfelder, Kegels, and Schorn
We find no evidence that our regression results are mainly driven by second
and third observations. As in the original model, there are negative coefficients for
positive ratings and positive coefficients for negative ratings that are significant in
most cases. Including all rating variables, we obtain:
TABLE A5. Belgian Data (Adjusted): Extended Model
Model
Positive Rating
Negative Rating
0.6814
0.6861
731
731
R2
Cases
Coefficient
RSTD DEV
AGENCY
Variable
0.050
0.082
−0.089
0.081
CONTACT
0.134
0.085
−0.145
0.084
ANSWER
−0.056
0.086
0.034
0.088
0.006
0.082
−0.021
0.084
−0.143*
0.082
0.036
0.080
−0.105
0.093
0.181*
0.094
0.041
0.076
0.032
0.073
−0.198**
0.088
0.154*
0.088
OBJECTIVE
0.060
0.083
−0.046
0.081
ADAPTION
0.098
0.094
−0.094
0.086
TIME
−0.140*
0.081
0.117
0.080
COHERENCY
−0.140
0.098
0.195**
0.090
ENTROPY
−0.074
0.082
0.080
0.081
0.020
0.282***
0.020
MOTIVATION
DELAY
CONTRADICTION
INFORMATION
ADVANCE
UNDERSTANDABILITY
SIZE
SURVEY
0.286***
Coefficient
0.177**
−0.016
RSTD DEV
0.074
0.079
−0.230*
0.123
−0.230*
0.123
2002
0.042
0.086
0.077
0.087
2004
0.121
0.117
0.091*
0.117
2006
−0.285***
0.110
−0.335***
0.105
5.204***
0.307
4.846***
0.310
Constant
Selected regression coefficients, robust standard errors, R2 and case numbers; ***, **, * indicate significance on a
1-percent, 5-percent, and 10-percent level; dependent variable: logarithm of tax-related compliance costs. We estimate
separate models for positive and negative statements on administrative and legislative quality.
While the effects of INFORMATION and TIME are relatively weak compared
to the original setting, we find a stronger influence of COHERENCY and DELAY.
Taking into account the results of Table A4 and Table A5, we do not find evidence
that the results of our original setting are mainly driven by second and third observations of the same case.
Tax Compliance Costs: Authority Behavior and Taxpayer Services
67
Endnotes
1
2
3
4
5
6
A possible underestimation of the cost burden has already been mentioned by
Oster and Lynn (1955). Klein-Blenkers (1980, p. 140) asked German enterprises
for the sum of overall compliance costs as well as for the sum of itemized
cost elements. According to his findings, the sum of overall compliance costs
was considerably lower (by about 50 percent). Hence, within the overall cost
burden estimation some cost elements must have been “forgotten”. This can
be interpreted as evidence for a cost perception deficit. Similar results are
reported by Rametse and Pope (2002) and Chittenden et al. (2005). These
authors try to estimate the psychological costs of tax compliance by computing
the difference of the sum of overall compliance costs and the sum of itemized
cost elements. In contrast to the qualitative findings of Delgado Lobo et al.
(2001) and other authors, this difference is generally identified as negative. In
the questionnaire of the German survey, businesses were asked for personnel
costs, external costs and other monetary costs without an allocation to specific
activities. This could result in a possible underestimation of the corresponding
cost burden.
This estimate is represented by a 5-point Likert scale, whereby businesses could
declare their “perceived” burden resulting from compliance activities. In a
logarithmic OLS model, we obtain a correlation coefficient of 0.483 (robust
standard error of 0.101) for businesses with a high “perceived” cost burden
(Likert scale rating of 4 or 5). Regarding social insurance-related costs, we
receive a coefficient of 0.456 (robust standard error of 0.128). Hence, these
businesses bear on average about 60 percent higher compliance costs than
other businesses in the data set.
Pressure on the political authorities may be an incentive for private taxpayers
with high compliance costs to participate in a survey. However, these taxpayers
may also be particularly reluctant to take part in a survey, because they do not
want to waste their time. Empirical investigations provide evidence for both
arguments (Wicks 1965; Allers 1994; Rametse and Pope 2002).
See the recommendation of the European Communities K (2003) 1422 from
the 6th of May 2003.
Instead of the number of employees, the German questionnaire exclusively
quotes the number of associates including the entrepreneur. Hence, we use this
information as size measure.
Employment taxes and social insurance contributions are not included. They
are part of additional statistical material of the Federal Planning Bureau about
the compliance costs of labor legislation.
68
Eichfelder, Kegels, and Schorn
7
Within the years 2000 and 2002, the data contains information about
depreciations resulting from monetary expenses for tax compliance. However,
this information has not been collected in the years 2004 and 2006.
8 This is caused by differences in the survey questionnaires.
9
Business age is available for 2000 and 2002. The same holds for the legal
form and the number of establishments in case of an enterprises survey. The
questionnaires of 2004 and 2006 contain statements on proposals to simplify
the tax law. The use of different information technology tools is questioned in
2004 and—in another form—in 2006.
10 Regarding the Belgian data source, also an unbalanced panel regression should
be possible. However, due to the survey design, most cases are one-shot
observations. For that reason, we would lose the greater part of the overall
information if we would choose a panel estimator.
11 That holds especially for Belgium. Within the questionnaire, a rating of 3 does
not account for a neutral rating, but for no opinion. Therefore, it is not evident
that businesses with a positive rating have a lower cost value than businesses
with a “rating” of 3.
12 Due to the endogeneity of “real” compliance costs and the “perceived” cost
burden based on Likert scale values, we do not consider a dummy variable for
this proxy of psychological costs.
13 In contrast to tax-related costs, also a compliance burden of zero may
be realistic regarding social insurance-related costs if a business has no
employees.
14 The first associate is assumed to be the entrepreneur.
15 Corresponding to our main analysis, size is measured as logarithm of turnover
or as logarithm of the employee number raised by one. In the German
case, we consider no alternative control parameters for the tax-related costs
and exclusively EMPLOYMENT in case of social insurance-related costs.
Considering the Belgian data set, we include INDEPENDENT and YEAR.
16 The additional cost burdens are slightly higher than the coefficients in Table 10.
This is due to the fact that these coefficients document exclusively the marginal
but not the absolute effects of our logarithmic regression model. Regarding
dummy-variables, there is no meaningful interpretation of a marginal effect.
5
D
Enforcement Strategies
Fogg
Rourke
Besfamille Olmos
Collecting Collected Taxes
T. Keith Fogg, Villanova University School of Law
L
ong ago Congress made the decision to collect taxes through business entities rather than to have every tax dollar collected directly by the government.1 The decision created an effective and efficient means to collect most
individual income taxes, employment taxes and excise taxes. This method of collecting taxes places businesses in the front line of tax collection as trustees for the
government with respect to the taxes they collect.
The system works well as long as the entities collecting the taxes remit those
taxes to the United States after collecting them. While the vast majority of entities remit the collected taxes in a timely fashion, unfortunately, not all entities do
so.2, 3, 4 The failure to pay over collected taxes occurs in only a small percentage of
cases; however, even this small percentage adds up to a significant dollar amount
when spread across the entire economy.5 As of 2008 the failure to pay over collected taxes represented a $58 billion piece of the overall tax gap.6
Predicting the general types entities that will fail to pay over the collected taxes
does not require sophisticated modeling. The entities that will struggle to pay over
the collected taxes will almost always be smaller businesses run by entrepreneurs.7
The failure will typically, but not always occur during the startup years. The failure
to pay the collected taxes will frequently, but not always, signal that the business
is about to fail.8
This paper examines techniques that could improve the current system for
collecting from entities these collected taxes with the goal of creating a system
that better encourages the entities to pay over the collected taxes at the outset and
more effectively addresses the situations in which the failure has occurred. The
paper makes several suggestions on improving compliance in this area. First, the
returns reporting collected taxes should be made public. As will be discussed
further, these returns do not carry with them the same need for privacy driving
the disclosure provisions which current keep private most tax returns and making
them public will have significant compliance benefits. This section of the paper
addresses the benefits of transparency. Second, structures must exist to funnel the
entrepreneurial businesses into compliance. The current system lacks sufficient
structure and contains some structural provisions that actually encourage noncompliance.9 This section of the paper addresses the benefits of better structure.
Third, the alternative mechanism for recovering collected taxes, the responsible
officer provisions, needs changing to allow interest (and penalties) to be charged
against the responsible officer from the due date of the entity return rather than
from the date of assessment of the personal liability.10 This section of the paper addresses the benefits of changing the law to make responsible individuals
more accountable.
4
Fogg
Disclosing Public Trusts—Transparency
Proposal
A brief history of the twists and turns of the provisions in the laws of the United
States governing disclosure of tax information is set out below. That history focuses on individual income tax returns, corporate income tax returns, returns of
exempt organizations, pension plans and political contributions. Very little, if
any, effort in this debate has been expended focusing on the returns reporting the
money held by business entities in trust for the United States. This Article will
return, below, to a more extended discussion of why the returns reporting the
money held in trust for the United States most closely resemble the pension plan
returns reporting the monies held in trust for employees. Like the returns of pension plans which Congress makes publicly available, the returns reporting money
held in trust for the United States should likewise enjoy public availability.
Currently, returns reporting money held in trust contain information about
both money held in trust and liabilities that do not stem from a trust relationship.11 Those returns should be split into two parts with one part reporting the
collected taxes and the other part reporting the taxes directly due from the entity.
The “new” return reporting only the collected taxes should become publicly available while the “new” return for the entity liability would remain subject to the current disclosure provisions. The return reporting the collected taxes should report
not only the obligation for the taxes but also the amount of payments made toward
that obligation during the return period and with the return so that anyone viewing the return could ascertain if the trust obligation had been fulfilled or remained
partially or fully unmet.
By creating returns reporting just the money held in trust and by making
those returns public, everyone can determine if a business entity meets its basic
obligation regarding the duty of handling the public’s money with which it was
entrusted. Making this information public would allow everyone to make decisions concerning businesses entrusted with public funds just as everyone makes
decisions concerning public officials entrusted with public funds. The monies reported on these returns do not belong to the taxpayers filing the returns, do not
reveal business secrets and bear none of the reasons for protection that ordinary
tax information carries. The effect of making this information public should allow everyone to make informed decisions on which businesses to support or not
support, which businesses have a strong likelihood of failure and which competing
businesses have gained an improper competitive advantage. Once this information becomes public, those entities failing to pay over the collected taxes should
find a non-receptive public just as public officials would find a non-receptive public if they failed to properly handle public monies. The pressure caused by this
Collecting Collected Taxes
5
situation should inspire entities collecting taxes to properly report and pay these
taxes thereby improving compliance in this segment of the tax gap.
Brief History of Disclosing Tax Return Information
Almost since the adoption of an income tax system, Congress has debated the appropriateness of publishing the returns of individuals and entities reporting that
income.12 In addition to income taxes, Congress has imposed several other types
of taxes in its quest to gather enough money to satisfy its spending appetite.13 Most
of the taxes reported to the IRS fall under the disclosure provisions of Section 6103,
which prohibits the IRS from disclosing the information on those returns except
in specifically prescribed situations.14
While the United States experimented initially with public disclosure of tax
returns and return information, it evolved fairly early in the income tax era into
a restrictive posture with respect to the general availability of information from
tax returns.15 This more restrictive posture treated returns as public documents
but subject to disclosure rules established by the President.16 Under this system,
public disclosure of returns generally did not occur. Broad disclosure of returns
and return information, however, took place within the federal government. In
the disclosure provisions prior to 1977, Congress deferred to the Executive Branch
to create rules governing this area. Within this context, a significant shift occurred
in 1977 in reaction to President Richard Nixon’s use of tax information.17
The Nixon White House used tax return information to attack the President’s
“enemies,” and consequently Congress began to more carefully monitor the use of
tax information.18 Its review of the situation resulted in a significant revamping
of Section 6103 in 1976 to the statutory structure that exists today.19 Through the
Tax Reform Act of 1976, Congress set out to eliminate the ability of the Executive
Branch to obtain and use tax information and it successfully terminated that practice by removing the President’s control of disclosure exceptions.20 Instead of
granting broad discretion to the Executive Branch, Congress took the disclosure
power upon itself and created a series of narrow exceptions to govern disclosure
of tax information. These limited exceptions produced a scheme in which nondisclosure of tax information now serves as the guiding premise.21
The debate over privacy of returns has not uniformly marched towards keeping
private all tax information but rather has meandered as different types of tax information came under scrutiny. Disclosure of individual income tax information
came up for debate with the Revenue Act of 1864, which provided that tax lists
would be public.22 This debate continued in 1870 when the Commissioner ended
publication in newspapers, but the information remained open to inspection.23
Congress stepped into the debate in 1894 with the reenactment of the income tax
6
Fogg
by prohibiting publishing of tax information and imposing criminal sanctions
for violations.24
The issues surrounding the disclosure of returns by business entities did not
surface until much later.25 In 1909 the Payne-Aldrich Tariff passed, which imposed an excise tax on corporations.26 This law contained conflicting provisions
on the public nature of corporate returns with one paragraph explicitly making
them public records and the next punishing the divulgence of information.27 The
confusion caused by the conflicting provisions of the 1909 legislation resulted in
an amendment to the provision in 1910 which stated that “any and all such returns
shall be open to inspection only upon the order of the President under rules and
regulations to be prescribed by the Secretary of the Treasury and approved by the
President.”28 This language essentially created a compromise between those who
thought that corporate returns should be fully open to the public and those who
did not.29 The amendment also left the corporate returns as “public records” but
only open to public inspection with the President’s authorization.30
After the passage of the Sixteenth Amendment permitting income taxes,
Congress passed tax legislation in 1913 to exercise its newly created taxing authority.31 In this legislation Congress essentially adopted the compromise on disclosure
adopted in the 1910 provision.32 The debate surrounding the confidentiality of tax
return information continued for two more decades with each side citing the policy reasons for and against publicity.33 In 1924 Congress ordered the Commissioner
to prepare and make publicly available the names, addresses and amounts of tax
of individuals and corporations filing returns.34 In 1934 Congress enacted further
disclosure and then repealed it less than a year later.35 Concerns over kidnapping, resulting from the publicity of individual income, overrode concern for the
public’s need for this information, causing repeal of the 1934 disclosure provisions
in 1935.36 From 1935 until 1976, little changed in tax disclosure provisions, with
Presidential order controlling disclosure of return information.37 During this
period, Presidential decree inhibited the publicity of tax return information, but
availability of this information increased among government agencies.38
In 1976 Congress enacted sweeping changes to Section 6103, severely restricting the use of tax return information.39 Essentially, Congress assumed the role
through legislation of determining which information to disclose, removing this
authority from the Executive Branch.40 Since the 1976 revisions to Section 6103,
merely cosmetic changes have occurred. In section 3802 of the Revenue Reform
Act of 1998, Congress provided for a major study of the disclosure laws.41 That
section ordered the Joint Committee on Taxation and the Treasury Department
to submit reports to Congress on the state of the disclosure laws and any needed
changes. These reports provide a significant overview of the disclosure laws from
both historical and policy perspectives and also outline legislative proposals.42
Nothing in these reports or in any legislative history specifically addresses the rec-
Collecting Collected Taxes
7
ommendation of this Article that returns of collected taxes present different issues
than income tax returns and other returns reporting taxes of taxable entities.
In addition to Section 6103 which provides the primary directives on disclosure
issues, two other statutes exist in the Internal Revenue Code which provide significant guidance concerning disclosure issues—sections 6104 and 6110. Section 6104
got its legislative start in 1950 when Congress first gave legislative attention to the
different disclosure considerations present regarding returns of tax exempt organizations.43 Essentially, 6104 takes the opposite approach to 6103 and provides for
disclosure of the tax information of tax exempt organizations.44 This disclosure
occurs because of the tax benefits received by the tax exempt organizations and a
perceived need for public awareness of the affairs of organizations that receive a
public subsidy.45
Section 6110 resulted from litigation under the Freedom of Information Act
(FOIA) seeking disclosure of private letter rulings.46, 47 In 1976 as Congress revised Section 6103, it added 6110 to create a more open system for parties trying
to understand the IRS positions on specific transactions. Prior to 6110 certain law
firms that regularly made private letter ruling requests had significant information
on IRS ruling positions that was unavailable to the general public.48 Section 6110
opened up the IRS decision making process. The IRS removes taxpayer identifying information and certain other data in the published rulings before the data is
made public.49 The inclusion of Chief Counsel Advice in 1998 significantly expanded Section 6110.50 Litigation by Tax Analyst has increasingly expanded the
interpretation of 6110’s disclosure provisions, as it constantly pushes for disclosure
of more information.51
The history of the disclosure provisions demonstrates a fairly broad consensus
that privacy interests trump publicity of most tax returns and return information
except in narrowly drawn circumstances. Broad exceptions to that consensus exist with respect to the returns of tax exempt organizations, political organizations,
and pension plans. This Article argues that the private collection of federal taxes
should trigger application of the broad exception to the general rule of privacy. To
understand why, it is necessary to understand how the private collection of federal
taxes operates.
Third Party Collection
Using business entities to collect taxes for the government results in efficient and
often seamless tax collection as demonstrated by the significant percentage of federal taxes collected in this manner.52 Incorporating the collection of taxes into
the purchase price of goods and services, a process which occurs with sales and
excise taxes, requires little additional time or effort to collect the tax than to make
payment for the underlying item. Similarly, using employers to collect income
8
Fogg
and social security taxes directly out of employees’ wages produces efficiencies
and reduces compliance concerns because the taxpayer never sees the money but
merely receives a net paycheck. By collecting taxes through such transactions,
the government uses efficient structural tax principles which increase compliance
while simultaneously lowering the both collection costs and the bitterness associated with making tax payments.53
One of the most common ways in which businesses collect excise taxes on the
government’s behalf involves telephone companies. For example, telephone companies collect most of the communications excise tax as they collect telephone bill
payments from their customers, simply adding the excise tax to the amount of the
bill.54, 55 The bill clearly details the amount of the excise tax, separating the amount
from the bill’s total. Upon receipt of payment, the telephone company sets aside
the portion of the payment that represents the tax. The telephone company then
reports the excise tax to the IRS on a Form 720, which is filed by the telephone
company on a quarterly basis. Payments of the communications excise tax occur
along with the filing of the Form 720.56 At present the Form 720 reports both
excise taxes collected by the entity from others as well as excise taxes for which an
entity has its own liability. Proposed revisions to Form 720 to create a new form
specifically for reporting collected taxes are discussed in more detail below.
Another common tax collected by business entities concerns employment
taxes—income and social security taxes of employees withheld or collected by
their employer for the benefit of the United States. With respect to employment
taxes, the employer calculates the amount of taxes it should withhold from each
employee’s paycheck. Each time the employer pays its employees, it pays them the
net amount of wages after withholding income and social security taxes and any
other deductions.57 The employer should set aside the money it holds back from
the employees for payment of their income and social security taxes.58 Unlike
excise taxes where the entity actually collects the taxes from a third party, the employer “collects” these taxes from itself. The theory is that an employer with a
gross payroll of $10,000 will have $10,000 with which to pay the wages. It will pay
$7,000 to its employees and place the other $3,000 into a trust account and in such
a manner collect the taxes.59 Since some employers may have only have $7,000
at the time of the payment of payroll, the concept of collected taxes sometimes
breaks down when cash poor employers lack the resources necessary to satisfy the
tax obligations of its employees. The law, however, does not distinguish between
taxes collected from third parties as part of an excise tax and taxes collected from
employees to satisfy the employees’ income and social security taxes.60
The public nature of the trust comes not only from the description of the monies held in Section 7501 but also in the manner in which the money is treated once
collected. When a taxpayer pays a telephone bill that includes the communications
excise tax, the taxpayer’s liability ends there because that taxpayer receives credit
for the payment of the excise tax regardless of whether the telephone company
Collecting Collected Taxes
9
actually pays over the tax.61 Similarly, the employee whose wages are withheld
does not need to worry about whether the employer pays the withheld income
and social security taxes over to the IRS because that employee receives credit for
the payment regardless of whether the employer pays over the withheld taxes.62
In essence the entity collecting the taxes becomes an agent of the United States. It
does not hold the collected money for the benefit of the individuals whose taxes
are collected but rather for the benefit of the United States Treasury. Because the
funds are held for the public benefit, the public nature of the trust exists not only
by virtue of the statutory language which labels it a trust but also because of the
operation of the trust and the monies it holds.63, 64
The legislative history of the disclosure provisions does not contain a discussion
concerning why public trusts such as those held by business entities with collected
tax dollars are subject to the same disclosure laws, or rather, nondisclosure laws as
income tax returns. Congress did, however, provide for disclosure of certain types
of returns and identified the reason for it treatment of those returns. The benefits
that tax exempt organizations receive often serve as a basis for the policy argument
behind disclosing their tax return information.65
While business entities holding these trusts of collected taxes do not receive
the same subsidies received by tax exempt organizations, some similarities exist
between the benefits these entities receive and the benefits received by tax exempt
organizations and pension plans.66 First, the businesses do control funds for days
or weeks, depending on their size, as the money passes from the taxpayer to the
IRS.67 For businesses with a high number of employees or large amounts of excise
tax, the cash flow benefit could be substantial, even if short lived. Temporary
control of this money helps to offset the cost of administering the tax, even though
many businesses may not view it as much of a subsidy. Second, businesses are
granted the right to operate subject to certain obligations that exist regardless of
whether the business is tax exempt. The grant of authority to operate a business is
the grant of a potentially valuable benefit which should not entirely be overlooked.
Collecting taxes is a price the business must pay for the privilege of operating.
Third, the money held in trust for the public in the collected tax situation is not
unlike the money held in trust by a pension for its beneficiaries. It also bears similarities to other public trusts which keep their records open to the public.68
Other reasons exist for disclosing returns of collected taxes, particularly employment tax returns reporting withheld income and social security taxes. The first
of these ancillary reasons stems from the peculiar circumstances of employment
tax returns. Many of these returns are prepared by “payroll tax providers.” These
providers prepare the returns, sign the returns, pull the money from taxpayer’s
checking accounts and file the returns and the required remittances. Taxpayers
essentially turn over everything about payroll taxes to these firms that provide
this service.69
10
Fogg
If collected tax returns were publicly posted, the accessibility of information
on a public website would allow taxpayers who rely on payroll providers to pay
their taxes to ensure that their taxes were paid. Of course, these taxpayers could
go to the IRS now and make a request for their transcripts, but the availability of a
website with an easy search feature might help to reduce the problem that payroll
providers with a bent to steal cause—a small collateral benefit to this proposal.
A second ancillary reason for disclosing collected tax returns involves the
Federal government and its relationships with federal contractors. The Federal
government has a goal of not contracting with those who do not pay their federal
taxes.70 On January 20, 2010, President Obama signed a memorandum directing
government officials to recommend how to ensure that no new federal contracts
were awarded contractors delinquent in paying their federal taxes.71 One obvious way to accomplish this goal would be to publish the delinquent collected tax
data in a form easily retrieved by federal contracting officers since collected taxes
comprise over 90 percent of the unpaid federal tax debts of contractors seeking
federal contracts.72
This Article does not seek to change the practice of having third parties collect
taxes for the IRS or the method by which third parties collect these taxes. Rather,
it seeks to shed light on that process by changing the disclosure law regarding
these taxes. Amending current disclosure law will not only significantly enhance
the chances of closing the multi-billion dollar tax gap that exists because of the
failure to pay over these collected taxes, but this change will also correctly align the
disclosure laws with their policy considerations.
Reconciling Competing Goals of Increased Collection
and Privacy of Tax Information
Disclosing tax return information brings together competing policies of openness
and transparency against privacy rights, fiercely held individualism and concerns
for unnecessary government intrusion. Disclosure also brings up competing
claims concerning the benefits of openness. Proponents of opening up more information to the public cite the positive effects they perceive such openness will
have on compliance.73 Opponents on the other hand cite it as a concern, suggesting that it will detract from compliance as taxpayers become fearful that accurately
reporting their taxes will negatively affect another aspect of life.74
In order to determine when transparency should trump privacy and vice versa,
it is necessary to examine the benefits and concerns raised on each side of the
policy coin. Privacy concerns heighten when disclosure of tax information 1)
concerns individuals rather than entities; 2) may disclose trade secrets or other
information that might damage the taxpayer’s business; 3) discourages rather than
promotes accurate reporting of information; 4) results in associated costs which
Collecting Collected Taxes
11
outweigh the benefits of the information so disclosed; 5) fosters misunderstanding; and 6) politicizes the process.75
Concerns for the need to disclose information heighten when disclosure of tax
information 1) involves an entity whose information that receives significant tax
subsidies such as tax exempt organizations; 2) involves an entity that is reporting
information about funds held in trust for others such as pension plans; 3) involves
an entity that, while receiving tax subsidies, exerts influence without adequate accountability of those exerting the influence such as the concerns driving Section
527 (j); and 4) when the tax return contains valuable information to other government entities under circumstances where further release of the information can
be controlled.76, 77
The Joint Committee on Taxation Report (“JCT Report”), which Congress directed the Joint Committee on Taxation to prepare on disclosure law, describes
presumptions either for or against disclosure.78 The general recommendation of
the staff of the Joint Committee with respect to returns and return information
was that information “should not be provided unless the requesting agency can
establish a compelling need for the disclosure that clearly outweighs the privacy
interests of the taxpayer” (emphasis added).79 In contrast to this general rule with
respect to tax returns stands the policy recommendation concerning tax exempt
organizations, stating that “disclosure of information regarding tax-exempt organization … should be disclosed unless there are compelling reasons for nondisclosure that clearly outweigh the public interest in disclosure.” 80 Tax exempt organizations, pension plans and political organizations thus receive a presumption for,
instead of against, disclosure.
As a starting point, this Article adopts the two general principles set out by the
Joint Committee staff that a presumption of nondisclosure of return information
governs most return information and that a presumption of disclosure governs
the information of tax exempt organizations.81 These principles fit the consensus
on disclosure matters that has essentially controlled disclosure rules during the
modern era of tax administration and certainly reflects the consensus in effect
since 1934 and the repeal of the “pink slips.” 82 Exploring the reasons behind these
general principles provides an opportunity to determine where the returns of collected taxes should fall, and allows a testing of these principles against a specific
type of tax information that has received very little, if any, attention in the policy
debates surrounding disclosure.
The Joint Committee Report identifies the principal reason for the general rule
of nondisclosure: privacy. The right to privacy is a bedrock principle in the United
States.83 It has driven the policy debate concerning disclosure from its inception.
A second reason for the rule of nondisclosure is the view that confidentiality promotes accuracy on the returns submitted because taxpayers do not need to worry
about collateral effects of reporting accurate information if they know that the
returns stay within the IRS.84 The principal countervailing interest to privacy in
12
Fogg
this debate is the benefit that disclosure provides by shedding light on corrupt
practices. This was a principle that weighed heavily for Progressives in the early
part of the 20th century and drove the disclosure provisions enacted in 1909, 1924
and 1933, discussed above. While privacy eventually defeated the Progressive position and the presumption of nondisclosure won with respect to most tax returns,
the victory has not meant complete confidentiality. As the JCT Report states, the
showing of a compelling interest can overcome the general principle.85
The table below clearly outlines the disadvantages and benefits of disclosing tax
return information.
TABLE 1
Disadvantages of Disclosure
Benefits of Disclosure
1. Tax returns contain private information which
the government compels taxpayers to report,
and when disclosed, the individual may lose
significant privacy protections.1
1. The disclosure of information may be necessary in order to protect taxpayer rights.2
2. Tax return information that concerns individuals implicates greater privacy concerns.3
2. The informational value of the data from the
return may outweigh the privacy concerns
and safeguards exist to protect privacy to the
greatest extent possible.4
3. Tax return information that contains trade secrets of a business implicates greater privacy
concerns.
3. The disclosure of information assists in closing
the tax gap.6
4. When the disclosure of return information
would discourage accurate reporting of
information, the benefits of disclosure must
overcome the concerns of inaccuracy.7
4. When an entity is publicly traded, certain information on the return could influence investor
behavior.8
5. When the costs of disclosure outweigh the
benefits, the decision to disclose becomes
impractical.9
6. Disclosure has the potential to foster misunderstanding of the information in a manner
that disadvantages the tax system or the
taxpayer whose information was disclosed.10
1 See
JCT Report (Vol. I), at 5.
2 Without
disclosure of the existence of the federal tax lien, the Government cannot perfect its lien interest with respect to
certain competing creditors. I.R.C. § 6323. Alternatively, if the lien of the Government can defeat competing creditors
without their ability to know of the lien, lending would dry up as creditors feared for the security of their loans.
3 See
United States v. Morton Salt, 338 U.S. 632, 652 (1950) (holding “corporations can claim no equality with individuals in the enjoyment of a right to privacy.”); “The Committee decided that the information that the American citizen is
compelled by our tax laws to disclose to the Internal Revenue Service was entitled to essentially the same degree of
privacy as those private papers maintained in his home.” S. Rep. No. 94–938, 94th Cong., 2d Sess. (1976) This discussion focused on the ability to obtain tax information in non-tax criminal matters and highlights the kind of sensitivity
surrounding tax information of individuals.
4 Statistical
disclosures and state matching programs fall into this benefit category.
Collecting Collected Taxes
13
TABLE 1— continued
5 In
a Letter from Michael P. Boyle, International President of Tax Executive’s Institute to Senators Grassley and Baucus
dated June 12, 2006, Mr. Boyle expressed concerns about expanded disclosure of corporate tax returns listing several
reasons. One of his concerns specifically addressed the issue of proprietary information: “Public disclosure of tax
returns of publicly traded corporations would also reveal confidential and proprietary data not currently contained in
consolidated financial statements, including revenue and expense information by legal entity, jurisdiction, and functional
category (e.g. sales, dividends, cost of sale). Although much if not all of the information in a tax return would be
confusing to the majority of investors, disclosure would clearly aid a company’s competitors enormously in understanding the taxpayer’s business practices. Where a company’s competitors are not subject to U.S. taxing jurisdiction (and
hence, not subject to the same disclosure rules), the comparative disadvantage would be even more pronounced.”
See also Robert P. Strauss, State Disclosure of Tax Return Information: Taxpayer Privacy v. The Public’s Right to
Know, 5 State Tax Notes Magazine 24, 29 (July 5, 1993).(stating that disclosure “in this instance could disadvantage the
individual company as competitors learn the private details of the company’s activities. For small public companies,
and for companies with foreign competitors this problem is most pronounced, because for small companies there will
be a close relationship between their state and federal return and what they provide to the Massachusetts secretary of
state for public review. They would now have their private financial affairs subject to competitive scrutiny. Foreign competitors of a domestic firm would not have to disclose the financial circumstances of their offshore parent companies,
while now gaining access to information about the financial circumstance of the domestic firm.”)
6 States
7 JCT
have used this in adopting their shaming provisions.
Report (Vol. I), at 5.
8 Joe
Thorndike, Tax History: Promoting Honesty by Releasing Corporate Tax Returns, Tax Notes (July 15, 2002), at
324; Majorie E. Kornhauser, More Historical Perspective on Publication of Corporate Returns, Tax Notes, July 29, 2002.
These articles describe the perceived benefits of disclosing corporate tax returns as a means of informing investors.
In
arguing that publication of corporate tax shelter participation may have the opposite effect desired by proponents of
such publication, Joshua Blank points out that investors have been positively motivated to invest in corporations seen
as aggressively seeking to lower their taxes. What’s Wrong With Shaming Corporate Tax Abuse, 62 Tax L. Rev. 539,
563 (2009) (citing Michelle Hanlon & Joel Slemrod, What Does Tax Aggressiveness Signal? Evidence from Stock Price
Reactions to news About Tax Shelter Involvement, 93 J. Pub. Econ. 126, 128 (2009)).
9 See
Lederman, supra note 60, at n.183 (citing Theodore P. Seto, The Assumption of Selfishness in the Internal Revenue Code: Reframing the Unintended Tax Advantages of Gay Marriage 6 (Loyola Law Sch. L.A., Legal Studies Paper
No. 2005–33, 2005), available at http:// ssrn.com/abstract=850645). Of course a cost benefit analysis is essential in
every policy decision. The benefits listed below are simply a part of this analysis.
10 See
Charles W. Shewbridge, “Taxpayer Confidentiality Must Remain Paramount,” TEI Says, 1999 TNT 206–60 (In
comments on taxpayer confidentiality submitted to the Joint Committee on Taxation and the Treasury Department,
the Tax Executives Institute (TEI) stresses the necessity of the confidentiality of taxpayer information “to the integrity
of the tax system.”) This is a big concern of TEI. TEI has also expressed concern that public disclosure of corporate
tax returns would implicate the need to protect taxpayers from their return information being misused for political
purposes. TEI Opposes Public Disclosure of Corporate Tax Returns, Tax Executive (May 2006–June 2006). Of
course, this is a big concern in general about the disclosure of return information and is essentially reflected in the
first reason.
The two statements from the JCT Report setting out the policies governing
disclosure create several factors against which to test a request for a disclosure
exception. The application of these tests permits a reasonable determination of
whether a new proposed change to disclosure laws follows established policies.
These policies are embedded in the subparagraphs of Section 6103 that contain the
exceptions to the general rule of nondisclosure.86
14
Fogg
IRC 6103
Section 6103 outlines the disclosure principles of tax information, beginning with
the general rule of confidentiality. The code section then creates exceptions to this
rule through a series of four basic steps, detailed below.
First, does the disclosure contain “return” or “return information?” 87 If the
information sought is not return or return information, then more general federal
laws concerning disclosure of information take over.88 If the information sought
is return or return information, however, then the general rule of nondisclosure
kicks into effect with no disclosure absent an exception.
Second, does disclosure of the information raise privacy concerns? If the disclosure occurs to the taxpayer or to the taxpayer’s proxy, privacy concerns are not
implicated. In this situation, the reason for disclosure need not be compelling and
may be simply that a taxpayer wants to view his own tax return.
When disclosure does not occur to the taxpayer or the taxpayer’s proxy, the
next inquiry is whether the tax information concerns individuals. An individual’s
tax information has the greatest presumption of nondisclosure and requires the
greatest showing of a compelling interest. A business entity’s tax information also
requires the demonstration of a compelling interest, but not quite as high as is
needed for individuals.89
Other factors enter into this step of the privacy analysis as well: 1) The nature
of the tax information sought affects privacy concerns. Disclosure of a taxpayer’s
entire return will implicate greater privacy concerns than a discreet portion of
the return. 2) The type of disclosure also impacts privacy concerns. If the tax
information clearly identifies the taxpayer and is published in a public place, then
privacy concerns are elevated. Passing tax information to a limited group with
restrictions on further publication creates less of a privacy concern. 3) The potential for publication of the tax information to reveal trade secrets will implicate a
greater level of privacy concerns. 4) The potential for disclosure of the information to discourage accurate reporting on the return will create a stronger presumption of nondisclosure. 5) The potential for disclosure of the information to foster
misunderstanding will also implicate greater privacy concerns.
Each of these factors affecting privacy can be seen as moving the needle on a
dial, with one side of the dial representing complete nondisclosure and the other
representing full disclosure. The needle sits on the nondisclosure side of the dial
for disclosure of most tax information. When more of these factors are present
and greater privacy interests are involved, the needle moves even further onto the
nondisclosure side, and the more compelling the reasons must be to move the
needle over to the disclosure side of the dial.
Third, do the benefits of the disclosure outweigh the privacy concerns? This
step requires an analysis of the disclosure’s purpose and the gains derived from
disclosing information. Many benefits can result from disclosing tax information,
Collecting Collected Taxes
15
which serve as the basis for the numerous exceptions that currently exist to the
rule of nondisclosure. Disclosing tax information can help close the tax gap, catch
criminals, protect the rights of others and serve many other useful purposes. Each
exception represents an example of successful arguments for the benefits that disclosing tax information can bring.90
Fourth, if the disclosure is to an “agency,” are adequate safeguards in place to
limit disclosure of the information beyond that agency? Clear limitations on the
use of the information must accompany any disclosure outside the IRS that is not
to the taxpayer or the taxpayer’s proxy. In addition to the general admonition
against further disclosure contained in Section 6103 (a), almost every subsection
of 6103 involving disclosure to an agency contains explicit safeguards regarding
further disclosure as well as citations to agreements regarding disclosure, which
will also contain restrictions on further disclosure.91
These four steps encompass the inquiry necessary to implement the disclosure
policy stated in the JCT Report on Section 6103.
Section 6103 currently contains 13 exceptions to the principle of nondisclosure
representing instances in which Congress found a compelling reason to override the principle. Congress has also created exceptions for tax exempt returns
through 6104, opinions through 6110 and information concerning political organizations through 527. Examining the situations in which Congress has applied the
four-step test and determined to create exceptions provides the basis for a system
to test further exceptions to the rule of nondisclosure.
Testing the Policy
Section 6103 (a) provides that “[r]eturns and return information shall be confidential, and except as authorized by the title,” no official or anyone else with access to
this information “shall disclose any return or return information… .” This very
broad statement prohibiting disclosure follows the rule that absent a compelling
showing of a need for disclosure, the information remains inside the IRS. Due to
its breadth, this rule does not distinguish between individuals and entities.
The second test first concerns disclosure to the taxpayer or the taxpayer’s proxy.
This portion of the test drives two of the exceptions set out in Section 6103.
6103 (C) DISCLOSURE TO TAXPAYER OR TAXPAYER’S DESIGNEE
Although almost unnecessary, Congress created this exception with a limitation
that the Secretary can restrict the disclosure of return information if such disclosure would “seriously impair Federal tax administration.” 92 Permitting disclosure
upon the request of a taxpayer avoids policy concerns because the taxpayer waives
his right to privacy. No policy reasons for nondisclosure stand as a barrier to
this exception and, therefore, there is no need to analyze the benefits side of the
16
Fogg
equation.93 The limitation within (c) stems principally from the Government’s
interest in protecting the identity of informants.94 If a taxpayer or a taxpayer’s designee could access all information in a taxpayer’s file, then the taxpayer could learn
the identity of any IRS informants who may have instigated investigation of the
taxpayer’s return.
6103 (E) DISCLOSURE TO PERSONS HAVING A MATERIAL INTEREST
This exception covers a variety of persons who have a material interest of a return
filed by a taxpayer, viz, the taxpayer himself, the taxpayer’s spouse and children,
administrators of estates, trustees of trusts, trustees or guardians of incompetent
individuals, executors and administrators, receivers and bankruptcy trustees, attorneys in fact, former spouses and responsible officers.95–105 Due to the lack of
a need to protect privacy, the policy basis for the exception follows a similar path
as that in Section 6103 (c), which involves the taxpayer’s own information. Most
of the persons with a material interest in the tax return essentially step into the
taxpayer’s shoes, have a direct connection with the return, or have an interest in
knowing the information in order to make reasoned decisions.106 Since few, if any,
privacy concerns exist, little effort is needed to move the needle from the nondisclosure side to the disclosure side of the dial.
The remaining exceptions to the rule of nondisclosure set out in Section
6103 (a) and in the policy pronounced by the JCT Report all raise privacy concerns. Therefore, they require applying a combination of factors: the party seeking disclosure must demonstrate a compelling interest, benefits must exceed the
costs, and rules must exist to limit further disclosure. These tests are met in each
of the exceptions to the general rule of nondisclosure set out in the subsections of
Section 6103. Because these disclosures implicate privacy interests, the reason for
disclosure must be sufficiently compelling to move the needle over to the disclosure side. As will be seen with each exception discussed below, applying the fourstep test outlined above provides a clear demonstration of the underlying policy
reasons for disclosure:
6103 (D) DISCLOSURE TO STATE TAX OFFICIALS AND LAW
107
ENFORCEMENT AGENCIES
This exception fully discloses both returns and return information, the broadest
possible array of information, to a limited party—state and local taxing agencies.
Disclosure to this limited party fully implicates all privacy concerns and has drawn
many lawsuits over concerns of lost privacy.108 The privacy issues here affect both
individuals and entities, implicating heightened scrutiny of this exception. The
cost of this disclosure does not outweigh the benefits because the taxpayer incurs
Collecting Collected Taxes
17
no direct dollar cost. The information transfer takes place directly, usually electronically, between the IRS and the receiving state or local entity.109 The states
perceive a significant benefit in receiving this information.110 This disclosure will
not cause misunderstanding because the recipients of the information are tax collectors with specific knowledge and interest in the information.
Disclosure of tax return information to the state taxing authorities raises the
traditional privacy concerns; however, none of the other factors suggest that this
information should remain within the IRS and not be shared with states. The
states perceive a significant benefit from the receipt of this information as demonstrated by their many letters to the Joint Committee.111 For ease of tax administration, most states have chosen to base their income taxes on the federal model.112
One consequence of this conformity is that states rely heavily on federal tax information to confirm the limited data they require from taxpayers.113 Currently,
the state returns ask for less information from taxpayers since the states know that
they can obtain additional information from the federal government.114 This system creates efficiencies because it keeps taxpayers from duplicating information in
two parallel systems. Because the states could ask for the same information that
appears on the federal return, their willingness to obtain this information through
the disclosure exchange does not really subject taxpayers to a greater intrusion.
In addition to the overall benefits this disclosure provides to the tax system,
other reasons exist in support of disclosure. The states must carefully safeguard
the tax information they receive from the IRS as a part of this bargain.115 This
safeguarding represents an integral part of this policy decision to allow disclosure,
because this exception is so broad that state failure to safeguard the information
could compromise the integrity of the entire taxpayer information database. The
exception limits the use of the information, stating that the disclosure is “for the
purpose of, and only to the extent necessary in, the administration of such law,
including any procedures with respect to locating any person who may be entitled
to a refund.” 116 Additionally, the use is limited by the agreement entered into between the IRS and the state or local agency.117
Looking at how this provision would affect the needle on the disclosure dial,
the needle would start on the nondisclosure side but no specific privacy of accuracy concerns would push it further to that side of the dial. The importance of the
material to the states coupled with the elimination of duplication by sharing this
information pulls the needle over to the disclosure side of the dial.
6103 (J) DISCLOSURE OF INFORMATION FOR STATISTICAL
118
PURPOSES
This exception fully discloses both return and return information to some federal
agencies, and discloses only return information to other agencies.119 The exception permits disclosure to allow certain agencies to use the tax information to
18
Fogg
create statistics, specifically limiting the disclosure for this purpose alone.120, 121
Even though the disclosure implicates privacy concerns by releasing information
about individuals and entities to the agencies, the overall effect of the disclosure
here moves the needle to the disclosure side of the dial. The implication of the
privacy concerns initially moves the needle further towards nondisclosure; however, the limited use of the information by the agencies, the protection on further
disclosure and the importance of the data pull the needle to the disclosure side.
Similar to the reasoning for release of data to the states, the release of this data may
also have the effect of reducing burden on taxpayers by keeping them from receiving duplicate data requests from different government agencies.
The cost of this disclosure does not outweigh the benefits because there is no
direct dollar cost to the taxpayer. The information transfer takes place directly,
usually electronically, between the IRS and the receiving agency. Disclosure of this
rich database of information benefits all taxpayers by aiding the economy in running more smoothly and reducing intrusions on privacy by the census data collectors. In addition, the statistical information that these agencies produce must
protect the privacy of individual taxpayers.122 The importance of the data to the
specific programs satisfies the compelling need test, even where, as here, many of
the agencies receive data about individuals as well as entities.
6103 (K) DISCLOSURE FOR TAX ADMINISTRATION PURPOSES
This subsection contains a number of discrete circumstances in which disclosure
occurs, only one of which will be discussed here.123 This provision permits disclosure to the public of specific taxpayer information, including information about
individual taxpayers.124 The information disclosed with filing a notice of federal
tax lien (NFTL) is very specific, and therefore economically harmful to the named
taxpayer.125 Because of the sensitive and private nature of the tax data and the
public nature of the disclosure, the filing of the NFTL would move the needle far
to the nondisclosure side of the dial. Only the compelling need to protect the lien
interest of the Government allows the needle to swing to the disclosure side.
The compelling need to disclose taxpayer information by filing an NFTL comes
under the umbrella of tax administration. An NFTL is filed only when a federal tax lien exists, and the lien exists only when taxes remain unpaid. To collect the unpaid taxes, Congress created the federal tax lien to protect the United
States’ interest in the taxpayer’s assets. The administrative problem with the lien
is that without publication, only the IRS and the taxpayer know of its existence.
Creditors remain unaware of the existence of the lien until its publication. In the
1966 Federal Tax Lien Act, Congress acknowledged that most creditors would defeat the federal tax lien unless a notice of the lien was properly filed.126 It devised
a system of filing as a mechanism for fairly treating creditors competing with the
federal tax lien.127 Filing the lien, however, discloses the taxpayer’s identity and
Collecting Collected Taxes
19
address, the existence of an outstanding tax liability, the amount and type of that
liability and the year(s) related to the liability.128
The costs associated with filing the NFTL do not outweigh the benefits because
the IRS secures its interest in the taxpayer’s assets by filing the lien.129 Even though
this disclosure enables the availability of damaging information in an unlimited
fashion, it meets the compelling need to disclose test. The only alternative to disclosure that would protect the IRS’ secured status is a law that would make competing creditors vulnerable to losing their secured claims, without the opportunity
to know of the competing tax lien.130 Here, the benefit to the IRS and to competing creditors outweighs the privacy interests of the taxpayer. This exception to the
rule of nondisclosure only occurs because of the compelling need to disclose the
lien to protect the interests of the Government and competing creditors.
PROPOSED SHAMING LAWS
Even though Congress has not passed laws similar to the shaming provisions
enacted by some states, applying this test to shaming laws provides insight
into Congress’ failure to follow the lead of the states. Shaming laws would
greatly implicate privacy concerns. The shaming laws of most states do so in
the broadest way by listing the names of individuals as well as entities. The
proposals of the past decade seeking to shame corporations engaged in tax
shelters still invoke privacy concerns, although not at the same level. Broad
shaming laws, such as those many states have adopted, create a level of privacy
concern similar to the level created by filing the NFTL—essentially the highest
level of concern short of publishing an individual’s return. Given the privacy
interests presented by the proposal, proponents need to show a very compelling need for such a proposal to pass. As noted by the JCT Report, a more
in-depth study on the benefits of shaming is needed to make a compelling
case for such a law.131 In 2000 when the JCT Report was written, insufficient
empirical data existed to support a compelling case for the benefit of disclosing information in this manner. The same concerns still exist today based on
some of the articles discussing corporate shaming.132 Nothing like the compelling case presented by the filing of the NFTL exists with respect to shaming. Until it does, shaming should continue to stand on the sidelines of federal
disclosure law.133
Assuming that returns containing collected tax information contain only information about collected taxes and the entity, the disclosure of these returns can be
tested similarly to the exceptions under Section 6103. Making these returns public
would not implicate privacy concerns of individuals because all of the information
concerns a business entity. So, this disclosure is not deserving of the strongest possible protections. Still, the proposal in this Article is to fully disclose the return,
making all of the entity information about the collected taxes available to anyone
seeking information about the entity.
20
Fogg
Because the collected tax information is information about others paying their
taxes through the entity, the information does not directly provide private tax information about the entity. If viewed strictly in that light, it is possible to argue
that privacy concerns are not implicated. Nor does the disclosure involve privacy
information about the individuals whose taxes have been collected because the
reporting of collected tax data would occur only in an aggregate form. The inquiry
does not stop here, however.
The tax information on a collected tax return does reveal entity information
about the number and potentially the compensation levels of the employees. More
specifically, the excise tax information reveals information about sales by the entity. This indirect revelation of information deserves some protection or at least
a basis for disclosure. The revelation of this information may cause the entity to
make an incorrect tax filing for the purpose of hiding trade secrets. It is also possible that an entity, knowing that the information would become public, would fail
to file a return in order not to reveal the extent to which it was not paying taxes.
Even though the privacy interests of the entity may be weak, the entity has
privacy interests in the conclusions that could be drawn from the tax data and the
seriousness of those privacy interests push the needle onto the nondisclosure side
a reasonable distance. It may not be possible to overcome these concerns from
a Section 6103 perspective. The reasons for disclosing the collected tax returns
derive from both the disclosure perspective and a collection perspective. From a
disclosure perspective, the nature of the information serves as the basis for disclosing the collected tax returns. The information concerns money held in trust, and
the public has a right to know what is happening to its money. This argument is
unlike the reasons for other exceptions to Section 6103 and is the reason that this
Article proposes that the change instead be made to permit this information to
become public pursuant to Section 6104. This argument, if persuasive, could move
the needle on the dial from the nondisclosure position to disclosure.
This Article will next examine the broadest exception to the rule of nondisclosure, Section 6104. This provision provides further background for this proposal concerning collected taxes and their placement within the Internal Revenue
Code. Unlike the exceptions to Section 6103 discussion in this section, Section
6104 takes the view that certain returns have a different starting point from a disclosure perspective.
IRC 6104
Section 6104 begins with the opposite presumption of the 6103 provision, with its
governing principle that tax information should be disclosed unless a reason exists
for nondisclosure.134 The tax exempt organizations, pension plans and political organizations governed by Section 6104 relinquish their privacy rights, in large part,
Collecting Collected Taxes
21
because of the tax benefits they receive.135 The public has a legitimate interest in the
information on the tax returns and applications of these organizations. This interest
outweighs the privacy concerns and other policy concerns driving the nondisclosure policy behind Section 6103.
The history of Section 6104 starts later than that of Section 6103, in part because
the history of tax exempt organizations, pension plans and political organizations
trails the income taxes that these organizations receive exemptions from paying.136
Tax exempt status was formally recognized in 1939.137 Reporting requirements for
these organizations followed in 1943.138 Concerns about abuses in the charitable
sector resulted in passage of additional reporting requirements for these organizations in 1950 and additional disclosure provisions.139, 140 In 1958, applications for
tax exempt status became available after an amendment to Section 6104.141 Pension
plans were added to Section 6104 in 1974 as part of the passage of ERISA.142 As
discussed further below, political organizations were added in 2000.143
The JCT report cited four reasons for increased disclosure of information concerning tax exempt organizations: 144
“(1) increasing public oversight of tax-exempt organizations; (2)
increasing compliance with Federal tax and other applicable laws;
(3) promoting the fair application and administration of the Federal
tax laws; and (4) advancing the policies underlying the federal tax
rules regarding such organizations.” 145
To the extent that the basis for presumption of disclosure of tax information
of the entities described in 6104 rests on the benefits they receive, as the Joint
Committee staff cited with respect to tax exempt organizations, it is difficult to draw
a parallel to the returns reporting collected taxes. While entities that collect taxes
on the Government’s behalf receive some small benefits for holding the taxes, the
argument that those benefits outweigh the burdens has little merit.146 Therefore,
the reason for categorizing returns reporting collected taxes under 6104 comes from
policies creating 6104 that extend beyond simply the grant of benefits to tax exempt
organizations. For that reason, other types of taxpayers and returns that 6104 involves are discussed here as well.
One type of tax exempt organization with a special return that receives partial
disclosure pursuant to Section 6104 is the trust for black lung patients.147 Black
Lung Benefits Trusts (BLBT) collect money for beneficiaries held in a public trust
for them administered by the Treasury Department.148 The money paid into BLBTs
comes from coal mine operators seeking to “self-insure for liabilities under federal
and state black lung benefits laws.” 149 These trusts file a return on Form 990–BL,
portions of which are public pursuant to Section 6104. The money paid by coal
mine operators into BLBTs is not a collected tax.150 BLBTs serve a different purpose
than most exempt organizations. They do, however, have a certain quasi-government aspect demonstrated by their ability to pour money into a trust administered
22
Fogg
by the Treasury Department, the Black Lung Disability Trust Fund.151 Congress
created BLBTs for the benefit of coal mine operators who had a requirement to pay
black lung benefits.152
Unlike most tax exempt organizations which receive public benefits, BLBTs instead serve a benefit to coal mine operators. The JCT Report did not address BLBTs
and the policy issues behind their creation as tax exempt organizations. In this case
the policy argument for disclosing a BLBT’s return information cannot easily derive
from the grant of government benefits as with most tax exempt organizations and
particularly the tax exempt organizations that existed in 1950. The trust created
here more resembles a public trust than a tax exempt organization. In this regard
it represents an instance of disclosure not unlike the disclosure proposed in this
Article for collected taxes.
BLBTs are singled out for discussion here because they have a different policy
foundation than most tax exempt organizations. The policy basis for BLBTs as organizations whose returns face a presumption for disclosure more closely mirrors
the basis for making public collected tax returns, since both circumstances involve
trusts in which the public has an interest. Moving from tax exempt organizations,
even those such as BLBTs, to pension plans makes this parallel more apparent. The
reasons for disclosing pension plan information do not mirror those for tax exempt
organizations, although some overlap exists.153 Pension plans hold money paid by
employers into a trust for their employees. The public trust created by pension
plans more closely resembles the public trust created by collected taxes than the
circumstances of most tax exempt organizations.154 The disclosure of the tax return
information of pension plans increases public oversight just as with a tax exempt
organization. Publication allows plan beneficiaries to observe the finances of their
pension plan. Even though pension plans serve a defined population of employees
and former employees of a business, the health of the plan implicates significant
public interest.
A failed pension plan invokes the intervention of the Pension Benefit Guarantee
Corporation (PBGC), a quasi-government agency that pays pension benefits
when a pension plan fails.155 Because of government is standing behind the pension plan, the interest of the general public in the information about pension plans
is heightened. Publication of pension plan information also, arguably, increases
compliance with Federal tax laws because plan administrators know that they are
being watched.
In addition to tax exempt organizations and pension plans, political organizations156 described in Section 527 also have their returns disclosed under Section
6104.157 Political organizations only came under the disclosure provisions of Section
6104 in 2000 as a result of Congressional desire to make public both contributors to
political organizations and the expenditures of political organizations.158, 159 When
the Supreme Court struck down and limited as unconstitutional some of the reporting requirements of the Federal Election Campaign Act of 1971 (FECA), Congress
Collecting Collected Taxes
23
relied on Section 6104 as a mechanism for shining light on those who stood behind the curtain of political organizations.160 This use of Section 6104 served more
to benefit campaign finance law that to promote tax disclosure.161 Using Section
6104 and Section 527 (j) to publicly name donors to political organizations stands in
contrast to the shielding of donors to Section 501 organizations by 6104.162 While
the information disclosure with respect to political organizations that occurs under
Section 6104 differs significantly from the disclosure of information about collected
taxes proposed in this Article, the use of Section 6104 for the purpose of disclosing
donations and expenditure information of political organizations demonstrates that
Section 6104 does not exist solely to shine a light on charities. Here, Congress used
it for primarily a non-tax purpose.
Another possible reason cited by the JCT Report as a basis for publication of
the tax information of tax exempt organizations is the fact that these organizations
often fill a void that a Government organization would otherwise fill. The governmental nature of the operation of these tax exempt organizations provides a reason for opening up their records just as the records of the Government accessible
to all.163
The JCT Report contained a quote from Senator Carl Curtis made in 1969 during the legislative debates that led to significant overhaul and restructuring of the
tax exempt sections of the Internal Revenue Code. The language used by Senator
Curtis provides a powerful argument for placing the returns of collected taxes into
the same category as tax exempt returns:
“[T]ax exemption is a high privilege. I believe the operation of a tax
exempt foundation is public trust; and starting from the premise, I
believe that all the business, all the transactions, all the receipts, all
the investments, all the grants and all contributions made by the
foundation to individuals or to institutions, are of public concern.”
(Emphasis added)164
This quote helps to tie the returns of tax exempt organizations and the policy
driving their disclosure with the returns reporting collected taxes. Senator Curtis’
use of the term “public trust” very accurately describes the effect of Section 7501.165
That statute provides, in part, that “[w]henever any person is required to collect or
withhold any internal revenue tax from any other person and to pay over such tax
to the United States, the amount of tax so collected or withheld shall be held to be
a special fund in trust for the United States.”166 The statutory language describes a
public trust held by the business entity. The monies so held are certainly of public
concern. As described above, the persons paying the taxes receive credit whether or
not the entity holding the funds in trust pays over the taxes to the IRS.167 Therefore,
the public has a direct concern with the public trust created when business entities
hold collected taxes, since the persons whose taxes are collected received credit for
24
Fogg
those payments whether or not the IRS ever receives the money.168 The nature of
the public trust created when business entities hold these taxes and the quasi governmental nature of this activity can perhaps more easily be seen if viewed through
the lens of the policy debate in recent years surrounding private debt collectors.
During the past decade Congress has enacted Section 6306 which established
“Qualified Tax Collection Contracts,” the statutory language for private debt collectors.169 Even though the authority to enter into private collection contracts still
exists in the Code, the IRS has recently decided not to renew any contracts and does
not plan to renew. One of the biggest concerns with private debt collectors was
that detractors of the program viewed collection of taxes as an inherently government function.170 Even though the program did not allow private debt collectors
to handle any money, the actions of these companies in assisting the IRS to collect
taxes was viewed as too closely tied to government action to permit their actions to
continue.171, 172 It is interesting how the post-assessment use of private collectors
could be such a hot topic because of the inherently governmental nature of the activity while most pre-assessment taxes are collected by “private collectors” without
even a whisper of complaint and without public disclosure of what they collected
and whether they paid over the taxes.
While the carefully vetted private debt collectors were not permitted to handle
any dollars, business entities handle over a trillion collected tax dollars every year
with no vetting prior to assumption of that responsibility.173 The point here is not
that the collected tax system requires dismantling in the same manner that the
private debt collection program has been dismantled, but rather that the collected
tax system is one of an inherently governmental function—the collection of taxes.
Further, the collected tax system allows private parties to hold tax dollars which
even the private tax collectors could not do. The governmental nature of the action
coupled with the holding of large amounts of federal tax dollars makes the returns
reporting collected taxes like the returns currently listed in Section 6104.
Placement of Collected Taxes within Disclosure Regime
While most of the businesses submitting returns to report collected taxes do not
receive subsidies in the same manner as tax exempt organizations, they operate as
businesses with the understanding that they have an obligation to collect federal
taxes as a part of the grant of the right to do business. In this sense their role as
tax collectors, while not subsidized, is a role in which they carry out a government
function. In addition to carrying out a government function, these businesses also
receive the benefits of holding this money as well as the burden of reporting on it.
The JCT Report cited two reasons for public disclosure that would apply equally
to reporting collected taxes as to the entities list in Section 6104: 1) disclosure enables the public to provide oversight and 2) disclosure allows the public to determine
Collecting Collected Taxes
25
which organizations to support.174 If tax returns reporting collected taxes became
public through Section 6104, the public would have the opportunity to view those
returns and report anomalies. The public would also have the opportunity to decide
whether to support businesses that did not properly treat the collected taxes they
held. Businesses, and government agencies, seeking to contract with the taxpayer
would have an easy means of checking on this important measure of tax compliance.175 Compliance or lack of compliance could form an important part of the
decision to contract with the taxpayer.
A few states have opted to disclose certain collected tax information such as sales
tax, excise tax, use tax and gasoline tax data.176 The policies of these states essentially reach the same result as the result proposed here that disclosure of collected tax
data is beneficial. A close look at these state laws and the policies behind those laws
is warranted.
Wisconsin, home of the Progressives who lead the early 20th Century charge
to disclose tax returns, has permitted disclosure of some aspects of its income tax
returns since 1923.177 In 1953 access to the entire return was paired back to access
to the net taxes paid.178 Public access to the amount of income tax paid extends
to individuals as well as corporations; however, the information is available only
upon a specific request to the Wisconsin Department of Revenue satisfying certain
conditions.179 While the Wisconsin disclosure provisions do not cover returns of
collected taxes, other states do.
Vermont allows disclosure of a number of taxes.180 Specifically, the plain language of the statute allows for anyone to obtain information about an entity holding
money in trust concerning the compliance of that entity. The publicity of this tax
data closely correlates with the collected tax data for which disclosure is proposed
here. Vermont permits oral or written requests. The Tax Department responds by
advising the requester whether the taxpayer is in “good standing,” which is the code
phrase for fully paid upon the collected taxes, or is “not in good standing,” which
is the phrase for a delinquent taxpayer. Vermont does not allow the public to view
the returns.
Massachusetts passed a law in 1992 making public a host of tax information regarding publicly traded corporations, banks, and insurance companies.181
Businesses are currently required to disclose the following:
1. Name
2. Address of principal office
3. Massachusetts taxable income
4. Total Massachusetts excise tax due
5. Non-income excise tax due
6. Gross receipts or sales
7. Either gross profit or credit carries over to future years
8. Income subject to apportionment182
26
Fogg
The Massachusetts provisions require reporting of both income taxes and the
sales and excise taxes more like the collected tax which are the focus of this
Article. One problem with the Massachusetts statute is its focus on publicly
traded companies. As will be discussed more fully below, companies of this size
are very unlikely to have problems with reporting and paying collected taxes.
The purpose for disclosing the liabilities in Massachusetts appears driven by a
somewhat populist desire to insure that large companies pay their “fair share.” To
the extent a goal exists for reporting collected taxes aside from the goal of aligning collected taxes with the proper disclosure provisions, limiting the reporting
of collected taxes to public corporations would serve no collection purpose. The
reporting of this information has now been in place for almost two decades with
little data gathered showing any negative impact from this reporting.183
The returns reporting collected taxes differ from almost all other tax returns
because they do not contain information about a tax liability incurred by the
taxpayer.184 Rather they contain information about taxes collected and held in
trust for the United States. These returns do not calculate a tax rate nor do they
contain “secret” information about a business that would enable competitors to
obtain an advantage. These returns simply report the amount of money held in
trust by the tax collecting entity. This type of return information should not
raise privacy concerns that drive the underlying secrecy of federal tax information.185 Rather, this type of information should exist in the public domain in
order that everyone has a transparent view of the money collected on our behalf
by the entities serving as agents of the federal government. The disclosure policy
reasons behind the decision to make public the returns reporting tax exempt
and pension return information should apply to the returns reporting collected
tax information.
Because the money is held in trust, there is no basis for distinguishing between the various entities reporting this information.186 The information should
be readily available in an unfiltered manner and posted on the internet so that
it is easily accessible. Reporting all of the information in an unfiltered manner
would make the task administratively easier for the IRS and allow those using
the data to access it all without limitations on size of business or other limiting
criteria. The reasons for disclosing the returns apply to all returns containing
collected taxes.
Disclosing all returns fits with the collection aspect of the policy consideration as well as the disclosure piece. By disclosing all returns, businesses filing
these returns know from the outset that the information on these returns differs
from the information on other tax returns of the business. Knowing that it is
different helps them understand why this debt obligation differs from other debt
obligations of the business which should make it more likely that businesses
would pay this debt, or go out of business, rather than paying the debts of trade
creditors in an attempt to stay afloat.
Collecting Collected Taxes
27
Changes to Current Return Forms
Currently, returns reporting money held in trust contain information about both
money held in trust and liabilities that do not stem from a trust relationship.187
Those returns should be split into two parts: one part reporting the collected taxes
(the “collected taxes return”) and the other reporting the taxes directly due from
the entity (the “entity liability return”). The collected taxes return should become
publicly available while the entity liability return would remain subject to the current disclosure provisions.188 The collected taxes return should report not only
the obligation for the taxes but also the amount of payments made toward that
obligation during the return period and with the return itself. This would allow
anyone viewing the return to ascertain if the trust obligation had been fulfilled or
remained partially or fully unmet.189
Two return forms require revision in order to accomplish this result. First,
the employment tax return, Form 941, requires changing. Form 941, due on a
quarterly basis, currently reports three primary tax liabilities of the entity having
employment tax obligation. These tax liabilities consist of the amount of income
taxes withheld from employees, the amount of social security tax withheld from
employees, and the entity’s own liability for social security taxes.190 Instead of one
form that reports both collected taxes and the entity’s own obligation, two forms
should exist. One form would report the collected taxes, described here as Form
941T (the T stands for “trust”) and the other would report the entity’s obligation,
described here as Form 941E (the E stands for “entity”).
Form 941T should contain relatively little information in order to limit the
disclosure of information and avoid confusion for anyone reading it. It should
report the total amount of income taxes collected from its employees, the total
amount of social security taxes collected from its employees and the total amount
of taxes paid to the IRS during the quarter. Some additional information could
be placed on the return similar to the information currently reported on Form
990 with respect to tax exempt organizations.191 This information is general information about the entity such as the type of organization, year of formation and
state of domicile. Certain information required on the Form 990–BL might also
provide some benefit such as “The books are in the care of:”, “Phone number:”, and
“Located at:.”192
Form 941E should track the information on the current Form 941, but will
exclude the information on the collected taxes reported on the companion
Form 941T.
The second return requiring revision is the Form 720 used to report excise
taxes. Like the Form 941, this form currently reports excise taxes directly owed
by the employer as well as excise taxes collected from others. Two forms, the
Form 720T and Form 720E, should replace the current Form 720. The Form 720T
should report only the excise taxes collected from others, identify the type of tax
28
Fogg
collected and report the total amount paid to the IRS during the reporting period
for the form. Some additional information could be placed on the return similar
to the information reported on Form 941T discussed above.
The Form 720 E should retain the information on the current Form 720, but
will exclude the information on the collected taxes reported on the companion
Form 720T.
How Mechanics of Disclosure Should Take Place
The disclosure of the collected tax information should take place through posting
every filed Form 941T and Form 720T on the internet. The posting should adopt
a format that is easily searchable. Section 6104 (a) (3) currently contemplates posting on the internet certain returns disclosed under Section 6104 and 527. That
same mechanism for dissemination of information should apply with respect to
the returns reporting collected taxes. The posting of returns should occur as soon
as possible after receipt. Neither the business entity nor the IRS should be required to produce copies of the returns posted on the internet.
The IRS should post any failure to receive a return on the internet. Individuals
interested in collected tax returns of an entity should not be forced to guess whether a return was filed and not posted.
Proposed change to Section 6104 (a) (1) (E)—Returns Reporting Collected
Taxes—If a business is required to collect taxes for the United states and holds the
collected taxes in trust pursuant to Section 7501 (a), the returns of the business
reporting the collection and payment of the collected taxes shall be open to public
inspection and posted on the internet.
Proposed Change to Section 6104 (a) (3) (C) Information Available on the
Internet—The Secretary shall make publicly available on the internet the tax returns described in 6103 (a) (1) (E).
Policy Reasons for Creating an Exception to the Rule
of Nondisclosure
While disclosure policy drives the recommendation in this Article that collected
tax returns should be disclosed under Section 6104 rather than kept private under Section 6103, the decision to disclose these returns could impact collection
policy as well. This Article proceeds with the belief that the disclosure of collected
tax returns would benefit compliance. In this unsubstantiated belief, the Article
adopts the unsubstantiated position of the JCT Report that disclosing tax exempt organization information increases compliance whereas disclosing returns
and return information with respect to taxable persons generally compromises
voluntary compliance.193
Collecting Collected Taxes
29
Assuming that disclosing collected tax returns will have the beneficial compliance effect that such disclosures controlled by Section 6104 currently have, the
next issue concerns the costs associated with publishing this information. Under
this proposal the taxpayer would bear little direct costs. The cost of preparing
the returns would increase, if at all, only marginally. The IRS would bear the
cost of publication. The real costs of this proposal would potentially consist of
a decrease in compliance, as a result of publishing the returns. This disclosure
proposal must then consider whether a taxpayer’s likelihood of filing returns and
reporting accurate information will decrease because of fears that information on
these returns would disclose proprietary information or otherwise harm the business. Publication is unlikely to impact the accuracy of the withholding tax returns
because of the direct link between these returns and the social security/withholding benefits of the employees including the employees responsible for filing the
returns. This accuracy is checked each year for employment tax returns under
the CAWRS program.194 While it is possible that some taxpayers would react
to publication by failing to file returns, this failure also has a detrimental effect
on those responsible for filing the returns since it indefinitely extends the statute
of limitations on assessment of their liability as responsible officers for the trust
fund recovery penalty.195 Although no definitive answer exists on possible detriments to publication of collected tax returns, no specific negative consequences
immediately appears.
Creating collected taxes returns which report only the money held in trust and
then making those returns public would enable everyone to determine if a business entity meets its basic obligation to properly handle the public’s money with
which it was entrusted. Publishing this information would also allow the public to
make decisions concerning businesses entrusted with public funds just as everyone makes decisions concerning public officials entrusted with public funds. The
monies reported on these returns do not belong to the taxpayers filing the returns
and implicate few of the reasons for protection that ordinary tax information carries. Publishing this information facilitates informed decision-making regarding
which businesses to support, which businesses have a strong likelihood of failure
and which competing businesses have gained an improper competitive advantage.
Once this information becomes public, those entities failing to pay over the collected taxes should find a non-receptive public just as public officials would find
a non-receptive public if they improperly handled public monies. The pressure
caused by this situation should encourage entities to properly report and pay collected taxes, thereby improving compliance in this segment of the tax gap.
The proposal in this Article to disclose all returns reporting collected taxes
under the regime of Section 6104 turns on an interpretation of disclosure policy
that places collected taxes into public view because of the trust nature of these
returns.196 It is possible to approach this problem based on the collection policy
perspective rather than disclosure policy, by considering possibilities of increasing
30
Fogg
transparency without moving collected tax returns under Section 6104. One such
possibility would be to use tools essentially available already under Section 6103,
which would require minor changes in that statute to the manner of publication
of information about taxpayers who owe collected taxes. This Article does not
recommend the collection policy approach but addresses it below as a potential
path to increased compliance with a smaller change in the approach to disclosure
policy with respect to collected taxes.
Shaming
As discussed below, the ability to disclose information concerning unpaid collected taxes already exists in almost all instances.197 Once the IRS files an NFTL,
the taxpayer’s liability for collected taxes (or at least for liabilities on returns on
which collected taxes are reported) becomes a matter of public record. This public
record will be quickly found by credit reporting agencies and others tracking the
filing of the federal tax lien.198 Disclosure of this information is currently permitted under Section 6103 (k) (2). This information goes to the county clerk’s office
where the taxpayer resides or where the taxpayer has property.199 If the taxpayer
is a corporation or partnership, the NFTL is filed as designated by the state where
the entity’s principal place executive office is located.200
Given that the information of an unpaid collected tax can become public
through the filing of an NFTL as soon as ten days after the assessment of the tax,
the next collection policy question is whether a more public pronunciation of the
liability should occur in order to more effectively convince taxpayers with unpaid
collected taxes (or potentially any unpaid taxes) to quickly satisfy the obligation.
Starting in the late 1990s and continuing as an increasing trend, states have turned
to further publicity dubbed shaming. 201, 202
In a tight market, one business may be able to hold a business advantage over
its competitors if it avoids paying to the IRS the taxes collected from or on behalf
of others. Publicizing the names of entities that fail to pay these taxes could potentially serve to level the playing field in such business areas A business advantage
obtained in this manner should instead become a business liability if competitors
have knowledge of the situation and can use it in the marketplace. Much of the literature in this area characterizes this type of disclosure as “shaming.” 203 Shaming
seeks to alter taxpayer behavior through the use of social pressure.204 In recent
years over half of the states have adopted a limited disclosure exception allowing publication of the names of certain delinquent taxpayers.205 States enact such
statutes with the hope that the individual or entity, seeking to avoid the negative
publicity associated with this publication, will ultimately comply.206 The Article
does not recommend that the United States adopt a shaming policy as a basis for
the publication of taxpayers delinquent in paying their collected taxes; however,
Collecting Collected Taxes
31
the relatively recent policy debate surrounding the state shaming provisions provides a basis for examining one relevant policy reason for creating an exception to
disclosure that would cover those taxpayers who were delinquent in paying over
collected taxes.
If the United States were to adopt shaming as a basis for addressing unpaid
collected taxes, it has several models to choose from as it reviews the statutes adopted by the states. The most common shaming provisions choose a numerical limit, such as the 100 taxpayers with delinquent collected taxes who owe the
most outstanding liabilities, and publish the names of those taxpayers on a website
or other prominent location. Another common method involves publishing the
names of all delinquent taxpayers whose outstanding liabilities exceed a selected
dollar amount. The dollar level for publication of an entity with debt should reflect an amount high enough to avoid information overload from all of the published names but low enough to provide meaningful information to competitors
and consumers.
Using ABC, Inc. to illustrate the proposal, the IRS would consider posting the
name of ABC, Inc. on its website at a special location designed to publicize delinquent taxpayers. The IRS would only publish ABC’s name if ABC owed a sufficient
amount, for example $25,000, of unpaid collected taxes. Once ABC crossed the
dollar threshold, the IRS would enter ABC’s name onto the list of tax delinquents.
The list would be available to anyone with internet access.
Currently, the IRS may not disclose tax information about any taxpayer without specific authorization under Section 6103. No exception exists for listing the
names of entities that do not pay taxes, whether the taxes are income, excise, employment, or some other type. In many instances entities with unpaid collected
taxes find themselves saddled with a filed federal tax lien; however, even when the
lien is filed, their competitors and companies with whom they do business might
not know about the existence of the federal tax lien.207
At present one exception to this general rule of nondisclosure in the Internal
Revenue Code could fairly be characterized as a shaming provision, rather than
simply a disclosure exception based on one of the traditional reasons.208 In 1996
Congress enacted Section 6039G.209 This section addresses a problem perceived
by Congress when an individual renounces U.S. citizenship for the purpose of
avoiding the payment of U.S. taxes.210 The shaming remedy created by Congress
to address this situation appears to be both too broad and too obscure.211 The
remedy reaches too broadly because shaming, or publication of the names of individuals renouncing U.S. citizenship, occurs for all who renounce rather than just
those who renounce for tax motivated reasons. The breadth of this reach diminishes the effectiveness of the publication of the names, because inclusion on this
list does not tie directly to improper tax behavior. The remedy is also too obscure
because the names of the shamed individuals are published in the Federal Register
on a quarterly basis. The Federal Register seems a rather remote and inaccessible
32
Fogg
place to publish names if its purpose is to have the individuals ostracized by their
community of peers.212
Regardless of its effectiveness, Section 6039G demonstrates a Congressional
willingness to resort to shaming as an enforcement technique.213 More recently
Congress has flirted with the idea of using shaming to identify corporate taxpayers who seek to reduce or eliminate their tax liability by employing “abusive” tax shelters.214 While numerous states have adopted shaming as a means
of increasing revenue, no state has yet adopted shaming based on corporate tax
shelter activity.215
The concept of shaming has received much attention among writers seeking
ways to promote tax compliance.216 Earlier writing concerning shaming addressed
its effectiveness in the criminal context.217 Toni Massaro provided a critical analysis of shaming in this context and identified five conditions that a shaming statute
should meet to be an effective remedy: 1) offenders should be members of an identifiable group; 2) sanctions must compromise social standing within the group;
3) group awareness of the sanctions and withdrawal from offenders; 4) offenders
must fear withdrawal by the group; and 5) offenders must have means to regain
social standing.218 Massaro concludes that these tests are rarely met in modern
America so she does not favor shaming as an effective remedy for criminals. Her
article demonstrates that shaming fell from grace as an appropriate sanction because it lost its effectiveness as a punishment tool as American society evolved
over the past 200 years.219 Because the factors for effective shaming in a criminal
case do not currently exist in America, she concludes that a reprise of shaming as
a tool for effective criminal punishment and rehabilitation would be a mistake.
The concerns expressed by Massaro have validity for analyzing whether shaming would work in certain tax contexts, but they also fail to address certain issues
presented by civil tax issues.220 Kirsch identified some of the shortcomings of
shaming in the tax context, at least as applied to the expatriate situation currently
adopted in the Code.221 Assuming that Massaro’s often cited tests provide the
most appropriate structure for determining the effectiveness of shaming, how do
these tests apply to the context of the failure to pay over taxes held in trust by an
entity? Is it worthwhile to consider the publication of the names of entities that fail
to pay their trust fund taxes, or would such publication fail to motivate the named
entities to pay the taxes while broadcasting to the world that the government has
been unsuccessful in fixing the problem in this area of noncompliance? 222
Many of the concerns raised about the effectiveness of shaming in the criminal
context do not apply to the naming of liable entities in the trust fund context.
Arguably, the publishing of names in the trust fund context serves not so much to
shame the offending party as to inform competitors and potential customers. If
the principal function publishing names is to inform rather than to shame, then
the tests for effectiveness would be quite different than those set out in Massaro’s
article. The focus moves from the impact of publication on the offender’s feelings
Collecting Collected Taxes
33
to the impact of publication on the behavior of its customers and, in their reaction, on the offender. Other than the few anecdotal consequences cited herein,
the effect of the knowledge of an entity’s failure to pay over its trust fund taxes is
not known.
In addition to the concerns about shaming in the criminal context, Kirsch
raised concerns about shaming in the civil context because of the way in which it
was handled in 6039G.223 His concerns raise slightly different issues than the ones
identified by Massaro and likewise need to be addressed in deciding whether to
pursue publication as an effective remedy for failing to pay over trust fund taxes.
Perhaps the largest single distinction between the expatriation statute and the
proposal to publish names of entities not paying trust fund taxes is the failure of
the definite link between having a tax motivated purpose for expatriation and the
publication of the individual’s name in the Federal Register implying that such a
link may exist.224 The link between non-payment of trust fund taxes and publication would clearly exist. The employment or excise tax that gives rise to the trust
fund liability is not a tax situation in which uncertainty exists. This is a situation
with a straightforward tax and an unpaid liability that is almost always a certainty.
The issue for trust fund taxes turns on non-payment and not the sometimes ambiguous language of the Internal Revenue Code in which the existence of a liability
itself can be in play.225
Knowledge that an entity has failed to pay its employment taxes could modify
the behavior of competitors of that entity or its customers.226 Competitors would
seek to find ways to exploit that information and would feel disadvantaged that
prior competition occurred on a non-level playing field. In addition, customers
might make decisions about entering into long term contacts with an entity that
could not keep current on its employment taxes since this failure would suggest
a lack of financial stability.227 The information could assist both competitors and
customers in making decisions.228
Many states have embraced shaming as a basis for altering taxpayer behavior
in a manner resulting in greater success in tax law enforcement.229 The movement
toward shaming in tax laws has increased significantly in the past decade.230 State
shaming laws generally follow a pattern of disclosing the 100 or 200 largest delinquent accounts or disclosing accounts exceeding a certain dollar amount.231 They
generally do not distinguish between types of taxes. No state, however, focuses its
shaming laws on collected taxes.
Balanced against providing a list that discloses outstanding tax obligations
is the general policy that tax information has privacy protections other types of
information about an entity do not. The question becomes whether protecting
an entity’s privacy with respect to its tax information should extend to money it
holds in trust for the United States. The money held in trust for the United States
does not reveal any business secrets about an entity. Because this type of shaming would occur with respect to an unpaid liability, an exception for disclosure of
34
Fogg
the information already exists in Section 6103 (k) (2). In this way, Congress has
already demonstrated a willingness to reveal this information in a format designed
to alert competing creditors of the existence of the liability making the issue of
shaming or other disclosure listing of this information one of formatting rather
than disclosing.232
Shaming seeks to modify behavior by targeting specific taxpayers with the
highest unpaid taxes or some other identifying negative tax trait. While some
states have expressed what they characterize as success through their shaming
laws, shaming has limitations in a modern society as discussed by Massaro. The
theory underpinning shaming applies equally to all types of unpaid taxes and,
in fact, is applied by states adopting shaming laws to a broad spectrum of delinquent taxes.233 Because no proof exists that shaming laws succeed, because they
represent a departure from the disclosure laws for a somewhat penal reason and
because they represent a broad based exception to the disclosure laws rather than
one targeted to collected taxes, this Article does not propose shaming laws as the
remedy for increasing collected tax compliance.234
In addition to broader policy implications for rejecting shaming as a remedy
for collecting collected taxes, a more specific reason exists for the circumstances
of these taxes. Shaming would not serve as an adequate deterrent to individuals
and entities considering the improper use of collected taxes. Tax shaming occurs
well after the use of this money in a circumstance in which the money is frequently
faced with a more immediate and real form of shaming, business failure.
While some persons may fail to pay collected taxes motivated purely by the
personal gain of “embezzling” collected taxes, the majority of persons using collected taxes do so because of liquidity issues with the business.235 When collected
taxes become the operating capital of businesses with liquidity issues, the people
making the decision to do so already face very real shaming issues. These people
face the shame of losing their business and perhaps losing their home and other
personal assets.236 The shame of having their name published on a list by the IRS
at some distance point in the future may come far down the list of matters causing
them deep personal pain. The shaming remedy when applied to collected taxes
seeks to shame the individual or entity responsible into paying the taxes at a point
when the business has often failed and the individual is broke. No amount of
shame can bring money into the Government when the party shamed has no ability to pay the taxes. Publication of the information of non-payment must come
at an earlier stage when business decisions concerning the use of the trust fund
money still have meaning.237
While shaming might deter a large corporation from investing in a tax shelter that will marginally improve its profits, the issues facing most entrepreneurs
who tap collected taxes for working capital differ significantly and suggest that
the shame from publication of non-payment of taxes may pale in comparison to
the shame they seek to avoid by using the collected tax dollars.238 For this specific
Collecting Collected Taxes
35
reason, as well as for the more general reasons discussed here, shaming is not
recommended as a better policy alternative to broad disclosure of collected tax
returns.
Disclosing Some Returns Containing Collected Tax
Information
As discussed above the failure to pay over collected taxes occurs in small businesses, usually during their start up phase when working capital needs achieve
acute status. Since large businesses almost never have issues with failure to
pay over collected taxes, should these businesses suffer the requirement of
disclosure of their collected tax return information when such information
will rarely disclose anything other than the timely filing and payment of the
required taxes. Given the realities of when the failure to pay collected taxes
occurs, would a disclosure provision targeted at the businesses most likely to
have difficulty be preferable to the broad disclosure of these tax returns.
Through a targeted use of disclosure the possibility exists that the benefits
of making information available could exist without burdening all entities that
collect taxes with disclosure. Disclosure could occur for those entities in the
target group which failed to timely file or pay their collected taxes.239 This
approach would resemble shaming in the sense that it would not publish all
entities, only the names of the “bad” entities. It would also resemble general disclosure from the perspective that it would provide information about
all entities because it would provide information about all entities within the
target group.
The exceptions to the rule of disclosure for tax exempt organizations, political organizations and pension plans do not provide for disclosure of only a
part of the group of impacted entities. In each of those exceptions, all of the
returns of exempt organizations or pension plans are displayed openly. No
effort exists in the provisions opening those returns to the public to distinguish between good and bad taxpayers or large and small taxpayers.240 Such
a distinction would not make sense in the disclosure of the returns of exempt
organizations, political organizations or pension plans since the goal of disclosure stems from a broad desire for knowledge about all of the organizations.
One distinction, however, between pension plans and collected taxes is that
the information on the pension plan return provides a picture into a complex
investment situation. The payment or non-payment of collected taxes, however, is a black and white situation—either they were paid or they were not
paid; the same simplicity of compliance does not exist in the pension plan situation. The amount necessary to properly fund a pension plan, while calculated
by actuaries, does not represent the same type of clear cut picture presented
36
Fogg
by collected taxes. For this reason publication of all pension returns provides
information beyond the payment or non-payment situation presented with
collected taxes. Therefore, it makes sense to publicize all pension plan returns
because of the information such publication provides where a similar publication of the returns of collected taxes does not serve the same function.
If not all collected tax returns were published, the next issue concerns how to
make the division between publishing and not publishing. This decision could
rest on whether the return has unpaid taxes. The policy decision made along
such grounds would parallel, in many ways, the policies present with respect to
shaming. As mentioned above, at least one of those policy decisions has already
been made in the area of federal tax liens. A decision to publish all collected tax
returns on which the taxpayer has an outstanding balance in actuality provides
little more information to the public, if any, than would already exist with the
NFTL.241 Such a decision involves small policy issues of the formatting of information but not broader policy issues of whether to allow such information into
the public realm.242
Another way to limit publication of collected tax returns would be to publish
all collected tax returns of entities of a certain size or age. Size measurement
could occur in a number of ways; however, the ideal method for such a limitation would turn on finding the break point at which entities, based on size or
some similar criteria, no longer fail to pay over the collected taxes. Disclosure
of all returns reporting collected taxes would occur below that break point. This
method, like the reporting of all entities, might create administrative simplicity
while avoiding publishing information about collected taxes that in almost all
instances would simply report that they were paid.
Many numerical cut off points exist in the Internal Revenue Code which base
reporting, and other, decisions on size or similar criteria. Creating another such
break point would not create precedent but would add a small layer of complexity
in administration that simply reporting all returns would not create. While placing a limit on reporting holds some allure because it avoids dumping information into the public with very limited benefit, the simplicity of a policy decision
that requires publication of all returns of collected taxes holds the greater allure.
For that reason, the limited publication of returns reporting collected taxes is
not recommended.
Conclusion of Transparency Section
Returns reporting collected taxes differ from other tax returns both in the type of
information they report and the underlying nature of that information. Disclosing
these returns is consistent with current disclosure policy when these returns are
viewed as similar to the returns disclosed under Section 6104. Disclosing these
Collecting Collected Taxes
37
returns is consistent with good collection policy because their disclosure informs
the taxpayer of the important and different nature of collected taxes as well as
informing the public of compliance regarding collected taxes. For these reasons,
the returns of collected taxes should move from the restrictive circumstances of
Section 6103 to the openness of Section 6104.
Creating Structure for Payment of Collected
Taxes—Structure
Proper Compliance Incentives
Before addressing each proposal and the compliance incentives it provides, it is
important to look at the research that exists on how to best influence compliant
tax behavior to see how the current system fits with the thinking on this subject,
and to see how these proposals might fit into that thinking. Academics wrote
much on tax compliance in the past few decades while, as discussed in the next
section, little has changed in the compliance regime of collected taxes. Most of
the writing does not specifically address the issue of failure to pay collected taxes
where the unpaid liability is not at issue and the obligation for the tax is straightforward. Nonetheless, the literature provides a basis for approaching methods to encourage compliant behavior in the area of collected taxes as well as in
more widely discussed areas of noncompliance such as the reporting of taxes by
self-employed individuals.
Recent scholarship identified structural systems as an important mechanism
for reducing behavior that the IRS seeks to prohibit and as a better alternative
to statutory commands.243 Professor Edward K. Cheng describes statutory commands, terming these “fiat,” as a direct but not always effective method of regulating behavior. According to Professor Cheng, the alternative to fiat is an indirect
method of regulation, terming this “structure,” to “establish mechanisms or procedures that push citizens toward compliance by making the undesirable behavior
less profitable or more troublesome.” 244
While direct regulation provides many benefits and is always an appealing option for legislatures, it relies principally on deterrence. Reliance on deterrence
does not necessarily achieve the desired result when large numbers of people violate the law.245 At some point, the passage of unenforceable laws simply fails to
provide the desired or necessary results, leaving structural laws as a potentially
more effective alternative. Structural laws seek to regulate behavior by removing
or minimizing the opportunities to violate the law rather than by punishing the
violations.246 Two types of structural laws exist: (1) laws that create “a process
that facilitates enforcement” and (2) laws that make it more difficult to achieve the
undesirable activity.247, 248
38
Fogg
As discussed in the following section, Congress passed numerous statutory
commands concerning the responsibilities of entities to collect certain taxes on
behalf of the IRS and pay the taxes over to the IRS. These statutory commands
created a system that works extremely well for collecting taxes from individuals
who are employees.249 The system also works well to achieve payment from large
corporations of the collected taxes where the leaders are managers and executives
rather than entrepreneurs. The system works much less well with small entities in
which the entity and the individual essentially share an identity. These small entities fail to pay over the collected taxes at a rate that has created a $58 billion unpaid
amount on the books of the IRS.250 While most people think of the tax gap as the
result of underreporting of taxes, the failure to pay acknowledged tax obligations
each year represents ten percent of the gap and amounted to $33 billion of the tax
gap in 2001.251
Congress created an effective system for tax administration using structural
laws to withhold taxes, and only used statutory commands to move the collected
taxes from the accounts of the entities doing the collecting into the hands of the
IRS. States are using some structural laws to assist in obtaining their collected
taxes. It is time for Congress to consider some of the techniques the states have
used and try to close the payment gap with techniques that extend beyond the
current ability to collect presented by statutory commands and enforced collection
action. In selecting the appropriate structural laws to address the payment of collected taxes, Congress and the Tax Administrator should determine the incentives
that influence those paying over collected taxes.252
One issue that emerges repeatedly when analyzing the payment of collected
taxes concerns the competing claims on the entities that owe these taxes. The
competing claims force the responsible persons to choose whether to honor their
personal commitments to trade creditors instead of impersonal creditors such as
governmental entities. A second issue concerns the entrepreneurial spirit of the
individuals who run the entities with collected tax problems and their belief that
the business will turn the corner “at any moment,” so they can make good on
unpaid obligations such as the collected taxes. Both of these issues present “legitimate” societal reasons for the decisions that responsible officers make that cause
them to fail to pay over the collected taxes. Neither issue presents a legal basis for
the decisions of the responsible officers.253
Nonetheless, the very real competing interests placed on the responsible officers
of entities with outstanding collected tax obligations put incentives on these individuals that conflict with their best interest from a legal perspective and conflict
with the societal goals of payment of taxes. The question is how to prioritize the
payment of collected taxes so that it causes responsible officers to properly balance
this requirement with their desires to pay their friends and their hopes to continue
a flagging business because success lies just around the corner. Additionally, an
appropriate structure must exist for those responsible officers who simply want to
Collecting Collected Taxes
39
use the collected taxes for their own benefit knowing that their use is an evasion of
the payment of tax obligations.
The proposals set out below seek to address these concerns by providing a mix
of information to allow responsible officers to make better informed decisions.
Hopefully, better informed decisions will lead to higher compliance with the tax
law.254 The proposals also seek to address the situation by providing the IRS with
better information about who the IRS needs to pursue because earlier pursuit of
the proper individuals will result in the collection of a significantly higher amount
of the unpaid collected taxes.255 The proposals also seek to provide some benefits
for compliant action and greater punishment for noncompliant action. The mixture of behaviors the proposals seek to influence mirrors the mixture of reasons
for current noncompliance.
The variety of bases for the proposals fits with the research suggesting that “one
size” enforcement activity does not fit all.256 Individuals and entities that fail to
pay over the collected taxes will have unpaid taxes for different reasons.257 The
path to non-payment may follow the same course as the path taken by those who
seek to evade the payment of their taxes or it may follow a course driven by different considerations.258 Creation of a variety of responses that include information, incentives, and enforcement provides a much greater opportunity for success
in reaching all of the individuals and entities within the target audience. Norms
based appeals and enforced compliance action can both increase overall tax compliance, although each may impact a different group.259
Past and Current Compliance Efforts
The failure of entities to pay over collected taxes the entities held in trust has
existed as a problem since the time these entities received the responsibility to
collect taxes for the government. Many administrative and legislative responses
have attempted to address the issue with varying degrees of success. As discussed in more detail below, attempts to address the problem through criminal
tax provisions have proved especially ineffective.
Current IRC section 6672 has its origins in a criminal tax provision.260, 261
When the 1954 IRC created section 6672 as the civil enforcement mechanism
allowing the IRS to pierce the corporate veil and pursue individuals responsible
for the failure to pay the collected taxes, IRC section 7202 came into existence as
its criminal counterpart tracing its roots back to the same origins.262, 263 Almost
no one has been prosecuted under section 7202 despite the fact it dates back
to 1954 and that $58 billion of unpaid collected taxes exist on the books of the
IRS today.264 Section 7202 presents prosecutors with significant problems as
they must prove that an individual willfully violated the act.265 The difficulty
is grounded in the discomfort felt by fact finders, whether the bench or a jury,
40
Fogg
convicting someone trying to keep their business afloat.266 As section 7202 is
almost never used as a basis for prosecution even though large amounts of collected taxes go unpaid, it represents a failure as an enforcement mechanism.267
Not long after the passage of the 1954 Code, Congress recognized the need
for more enforcement in the collected tax area and sought to address that need
through the creation of another criminal tax provision, IRC section 7215, in
1958.268, 269 Section 7215 specifically refers to and is triggered by the IRS taking action pursuant to section 7512.270 This statutory scheme has now existed for over fifty years. During that time twelve reported decisions existed of
individuals prosecuted pursuant to section 7215.271 As with section 7202 this
criminal provision had almost no impact on individuals and entities failing to
pay over collected taxes.272 Subsequent to the passage of section 7215, Congress
has not enacted any criminal laws addressing the issue of the failure to pay
collected taxes.273
Congress addressed a shortcoming of section 6672 in 1966 with the passage of
IRC section 3505.274 The reason for creating section 3505 was to close a loophole
in section 6672.275 Although the number of cases brought under section 3505 is
relatively small, the statute appears to have the intended effect of stopping entities from making net payroll lending.276 One reason this provision may have
succeeded where the criminal provisions did not is that the target audience of
this statute, banks and other lenders, represents an audience that pays careful
attention to statutory obligations and receives competent advice on how to meet
such obligations.
Some small changes to section 6672 were enacted by Congress in 1996 and
1998 as a part of taxpayer bill of rights provisions to expand rights and codify
procedures. However, those provisions did nothing to address shortcomings in
the ability of the IRS to recover collected taxes.277
In July 2008, the GAO issued a report concerning outstanding payroll taxes.278
In this report, GAO identified five problems caused by the failure of businesses to
remit payroll taxes: (1) public perception; (2) compliant taxpayers must shoulder
greater burden; (3) unfair competitive advantage to noncomplying taxpayers; (4)
prolonging the life of failed businesses through the noncompliance subsidy; and
(5) unmet government financial needs in era of deficits.279 Next, GAO made six
proposals to “fix” or improve the situation of non-payment of collected taxes: (1)
develop a better process for monitoring collection actions taken by IRS collection officers; (2) review case assignment prioritization; (3) file notice of federal
tax liens faster; (4) develop procedures for better monitoring how fast IRS collection officers cause assessment of section 6672 liability; (5) develop performance
goals and measures; and (6) work with states to develop better measures and
implementation of goals.280
Essentially all of GAO’s recommendations to “fix” the problem of compliance
in the collected tax area have fallen on the IRS and not on the noncompliant
Collecting Collected Taxes
41
taxpayers. GAO sees the solution to the compliance problem as one which can
be fixed by having the IRS work smarter, better, and faster rather than creating a
structure in which the noncompliant entities are led to a path of compliance by
structures that move them down that path.281 The proposals below seek to build
a better path rather than simply to find solutions to the problem in the efficiency
of the IRS or by additional criminal or civil penalty statutes that seek to treat the
problem by fiat.
Proposal
There are five structural proposals which will be discussed in turn:
1. Make the responsible persons of entities collecting taxes identify
themselves to the IRS and provide these individuals with
information about the legal consequences of collecting taxes
2. Provide incentives for new small business to timely pay collected
taxes
3. Require bonds of individuals or entities with a history of failure to
pay collect taxes
4. Eliminate the withholding and social security credit for responsible
individuals who fail to pay the entity’s collected employment taxes
5. Create an incentive for the responsible officers to pay the internal
revenue code section 6672 liability where more than one responsible
officer exists
The adoption of these proposals would put in place a system in which businesses collecting taxes and the individuals who run them would find themselves
both informed and motivated to pay over the collected taxes to a much greater
degree than exists today. With the implementation of these incentives, this corner
of the tax gap should shrink.
Self Identification of Responsible Officers
As mentioned above, to become a trustee for the United States, an individual or
entity merely needs to start a business in which (a) employees exist causing withholding of employment taxes, or (b) collection of excise taxes occurs. This Article
proposes that the IRS should: (1) require identification of the individuals within
an entity who have responsibility for the financial decisions of that entity and who
control the funds of the entity in such a manner that they determine whether the
taxes held in trust for the IRS get paid on time; (2) require the identified individuals to personally sign a document acknowledging responsibility as well as the
42
Fogg
duties and consequences that flow from that status; (3) provide the identified individual with detailed information concerning their duties; (4) require that the
list of responsible individuals be updated as positions change; (5) impose actual
penalties on persons found responsible who were not on the list of responsible
individuals for an entity that fails to pay collected taxes; and (6) set up a system
for contacting the responsible individuals within thirty days following the end
of a quarter in which collected taxes were not timely paid by the entity. Several
states have decided to gather information about who is responsible for payment
of collected taxes when a new business incorporates.282 The IRS has a similar opportunity to gather this data and to use that information to create a structure that
better fosters the payment of collected taxes. The IRS should do so.
When an individual or entity starts a business, it must request an employer
identification number (EIN) from the IRS and use the EIN on the return reporting
the employment or excise taxes. Currently, the IRS does not ask entities seeking
an EIN to provide any information concerning the individuals who have responsibility for ensuring that any trust funds held by the entity get paid.283 By failing
to request information about responsible officers at this juncture in the life of an
entity, the IRS misses the opportunity to obtain valuable information and simultaneously educate the individual on their duties as a trustee.
Understanding the process of obtaining collected taxes is necessary to understand why the IRS should want to know who is responsible at an entity for making
sure that the trust fund taxes get paid. To illustrate the process, a typical employment tax liability will set the scene:
ABC, Inc. (ABC) is a small manufacturing company that makes
widgets. It has a payroll of twenty-five employees that it pays
weekly. At the end of each week, ABC’s bookkeeper writes
twenty-five payroll checks. In the first quarter of 2008 each of its
employees makes $500 per week. The payroll checks made out to
the employees each week are for $400 because $100 is taken from
their weekly check to pay withholding taxes, social security, and
other employee expenses. For purposes of this illustration, $50
each week is withheld for employment taxes and $25 each week
is withheld for social security. After paying the payroll each week,
ABC should hold $1,875 in trust for the United States. If there are
thirteen weeks in the quarter, ABC should remit to the United States
$24,375 at the end of the quarter to cover its obligation as a trustee
for the withheld income and social security taxes. ABC will also
have a corporate liability for half of the social security liability of its
employees. However, that liability is not held in trust for the United
States but rather represents a tax liability of the entity.
Collecting Collected Taxes
43
If ABC does not remit $24,375 to the IRS at the end of the quarter,
it is liable for this unpaid collected tax liability. With respect to
this liability, the individuals at ABC and any other individuals or
entities having responsibility for paying over the collected taxes that
willfully caused the taxes not to be paid are also individually liable
for this amount. This liability arises under IRC section 6672 that is
variously known as the “one-hundred percent penalty” or the “trust
fund recovery penalty.” 284 The provisions of section 6672 allow the
IRS to pierce the corporate veil and hold liable anyone who meets
the tests of responsibility and willfulness. Assuming that ABC files
its employment tax return for the quarter, the IRS will know at that
point that ABC has an outstanding liability; however, the IRS does
not currently know who the responsible people of ABC are. The
IRS takes about two to three years to find out who is responsible
at ABC and whether they acted willfully in failing to pay the
collected taxes.285
To obtain the collected taxes that the entity did not pay over to the IRS, the
IRS will seek to collect those taxes from the entity itself, using its administrative
enforcement powers, such as lien and levy. In many cases in which the entity
does not remit the employment taxes, the entity faces severe cash flow problems
and the collection efforts of the IRS fail because no assets exist from which tax
collection can occur. The failure to pay the employment taxes serves as a marker
that the business failed.286 If this occurs, the only recourse for seeking the unpaid
collected taxes that the entity held in trust lies in the pursuit of the individuals who had responsibility for running the entity. At this point, the IRS would
benefit from knowing the names and other identifying information of these individuals. It does not currently know this information, and therefore must embark on a search to ascertain the identity of the individuals. This search, and
the time it takes to begin this search, is a big part of the determination of who is
liable. This proposal seeks to eliminate the search for who is responsible and simultaneously educate those individuals on their special responsibilities regarding
collected taxes.287
As discussed above, many states have already identified this issue and have administrative requirements that entities seeking to incorporate a business in their
jurisdiction must identify the individuals responsible for the monies held in trust
by the newly incorporating entity.288 Creating this requirement at the time an
entity seeks an EIN would not require a legislative change but merely an administrative one. The IRS would side step much of its currently lengthy process of
determining who to assess for the section 6672 liability if it knew the individuals responsible for paying over the monies held in trust by the entity. Armed
with knowledge of the individuals responsible, the IRS could simply send them
44
Fogg
correspondence giving them a reasonable, but relatively short, period of time to
explain why they were not willful in the failure to pay over the collected taxes.
For individuals failing to respond to such an inquiry or responding acknowledging liability, assessment could take place immediately after the correspondence
closing date. This might collapse the current two to three year waiting period
for assessment to a period lasting two to three months after the return due date.
Obviously, some individuals identified as responsible for payment of the collected taxes would respond to such an inquiry with an explanation detailing why
they were not willful in the failure to pay the collected taxes or an explanation
of why they were no longer responsible at the time the collected taxes were not
paid. Even an early response denying liability would allow the IRS to focus its
collection efforts at a point much closer to the end of the quarter. This would be
possible because of the known identity(ies) of the individual(s) who had claimed
responsibility for the payment of the collected taxes.
For individuals who argued they were not willful in the failure to pay the collected taxes, their responses would provide an early focus on the specific issues
in the case. A narrow range of defenses to the liability based on willfulness exists.289 The discussion concerning liability would focus on that narrow range
within which the admittedly responsible individual would need to fall to avoid liability under section 6672. For individuals who denied responsibility at the time
of the failure to pay over the collected taxes, the discussion would quickly shift to
the third parties such individual alleges to have taken over the responsibility and
the facts in support of such a takeover of responsibility. This could all take place
before a field collection officer had to track down individuals associated with the
corporation and would allow the collection officer to focus efforts on collecting
information before beginning the field work.290
The IRS would want to establish a presumption concerning responsibility for
those individuals listed by the entity as responsible. The presumptions would
not necessarily control the outcome but would put an emphasis on individuals
paying attention to the information provided to the IRS concerning responsibility.291 The presumption should provide that any individual listed with the IRS as
responsible for the payment of the collected taxes meets the test of responsibility
absent clear and convincing evidence that such individual did not hold a responsible position during the quarter(s) at issue.
The second aspect of this proposal concerns the acknowledgment of the individuals designated as responsible and the information provided to those individuals. The proposal contemplates that anyone listed as responsible for payment of an entity’s collected taxes would personally sign a form, under penalty
of perjury, acknowledging his or her status as a responsible person.292 Having
individuals sign such an acknowledgement serves several purposes that promote
effective tax administration.293 First, it forces the individuals signing such a form
to acknowledge their position as a trustee of the IRS. Second, it provides them
Collecting Collected Taxes
45
with knowledge that the liability of the entity for collected taxes has a different
character than other liabilities that the entity owes, because this liability extends
to individuals responsible for non-payment who have the protection of the corporate shield from almost all other debts of the entity.294 Third, it informs the
individual of the seriousness of this obligation which, in turn, should cause that
individual to prioritize which creditors to pay when cash flow problems occur.295
This proposal has two components. One is acknowledgment by the individual
that he or she has responsibility for the payment of the collected tax. The second
component has equal value and it concerns the information provided to these
self-identifying individuals. The acknowledgment will take place at or near the
time of incorporation when the entity has not yet had cash flow problems. This
should be a teachable moment for the individuals running the entity. This teachable moment should not pass without an effort to teach the individuals the duties
of a trustee and the consequences of the failure to carry out those duties.296 To accomplish the teaching of these principles, the IRS should create a detailed handout and provide it to the individuals as a part of a package they receive with the
acknowledgement form.297
The handout should explain the reasons why the IRS uses business entities to
collect taxes for it, the way the collected tax system works, and the proper method
for paying over collected employment and excise taxes. The booklet should further explain the civil and criminal penalties applicable to individuals who breach
those duties. Finally, it should explain the bankruptcy consequences of the liability for collected taxes and the fact that this liability, unlike all other tax liabilities, can never result in a discharge of this debt through bankruptcy filed by the
entity or a responsible officer.298 This type of information will allow individuals
embarking on this trust relationship to enter it with their eyes open rather than
discovering years later that their decisions to pay trade creditors rather than the
IRS created disastrous financial results. By being better informed, the individuals
responsible for the collection of the monies held in trust should take compliance
with the payment of the trust taxes more seriously.299
No matter how often the individuals associated with an entity receive reminders concerning the need to accurately identify responsible officers of the entity,
it is certain that either the wrong individuals will be identified initially or no one
will update the information as the entity business changes over time. For this
reason a third component to this proposal exists. This component concerns the
consequences for failing to update the information to the IRS as new individuals
become responsible for the payment of the collected taxes.
The directive requiring entities to identify responsible officers at the time of
the EIN application needs to contain a further directive requiring identification
of individuals as changes occur. While the IRS can administratively order that
the list of responsible officers initially contain all of the responsible individuals
and that any additional individuals inform the IRS as they become responsible,
46
Fogg
these administrative directions will have few teeth without some consequence for
failure to provide this information. To ensure that the initial group contains all of
the responsible persons and that necessary changes to the group are provided to
the IRS as the business changes, legislation may become necessary to enact a penalty structure for failure to register changes with the IRS. This penalty structure
should be a penalty separate from the section 6672 penalty. The penalty would
need to be significant enough to deter individuals from trying to hide from the
section 6672 liability by refusing or neglecting to identify themselves to the IRS at
incorporation or as newly responsible individuals joined the entity.
Many small companies contract out their employment tax compliance work to
payroll provider companies.300 These companies offer a range of services concerning payroll and payroll tax compliance. A typical contract between a payroll tax
provider and a small business entity might have the payroll tax provider preparing
payroll, paying payroll, preparing the quarterly Form 941 form, and paying the
Form 941 taxes. The payroll provider typically has an agreement allowing it to
withdraw the necessary funds from the entity’s bank account. Problems have arisen with payroll provider companies.301 Entities contracting with payroll provider
companies and the responsible officers of these entities may think that the hiring
of these type companies resolves all employment tax issues, but it does not. The
hiring of such a company does not absolve the entity and its officers of their duties
regarding the trust fund taxes. Michigan specifically addresses this issue with a
form it requires.302 The IRS should adopt something similar to the Michigan form
because so many small companies use payroll providers and may not realize that
the use of these entities does not absolve them from liability.
Individuals seeking to hide from the IRS by refusing or neglecting to self-identify as a responsible officer should bear a cost for that behavior. At present, no
penalty exists beyond that for failure to pay for individuals who cause the collected taxes to go unpaid. The section 6672 liability, while labeled a penalty, simply
serves as a collection device for retrieving the unpaid collected taxes. Arguably,
all individuals responsible for the failure to pay over the collected taxes should
have true penalties assessed against them in addition to the liability for the unpaid
collected tax. This proposal does not go that far and seeks only to impose a “true”
penalty on those individuals who fail to identify themselves as responsible for the
payment of the collected taxes, and thereby cause the IRS to do additional work
and take additional time in determining who should have the section 6672 liability
assessed against them.303 The clearly willful failure to self-identify should serve as
a factor in determining whether to prosecute someone for failure to pay the trust
fund taxes since, coupled with the failure to pay the liability, the failure to selfidentify could demonstrate an intent to evade the payment of the liability both at
the corporate and individual level.
In addition to a penalty for failing to self-identify as a responsible officer, the
statute of limitations for assessment should be suspended with respect to anyone
Collecting Collected Taxes
47
who has not self-identified as a responsible officer because the IRS will be misled
by the names on file and need time to recover the correct information. A statutory
change should treat the statute of limitations on assessment of the section 6672
liability as not beginning until a person identified himself or herself to the IRS
as a responsible person of an entity. If the person never provides the necessary
self-identification, then the statute of limitations would never run, similar to the
situation with unfiled returns.
This proposal does impose an additional burden on small businesses because
it causes more information to be provided at the time of the application for the
EIN, and because of the ongoing obligation to update the names of the responsible officers. It imposes additional burdens on the IRS to keep track of the
responsible officer information provided through this process. The information
sought, however, is consistent with the type of information generally available
about trustees and should allow the IRS to move much more quickly to assess
section 6672 liabilities, which should in turn, promote compliance. Also, the
targeted information provided to responsible officers at the time of incorporation should also enhance compliance.
Most of the changes needed to implement this proposal fall within the administrative powers of the IRS. It controls the EIN process and can change it with its
own authority. The aspects of the proposal imposing a penalty on persons who
do not update the responsible person information and tolling the statute of limitations on assessment for those persons require a legislative change.
Incentives for Small Businesses to Timely Pay
Collected Taxes
Businesses do not have a choice whether to become trustees for the IRS.304 If a
business has employees or if it engages in an industry in which excise taxes exist, a business must become a trustee for the IRS to the extent that it engages in
conduct that has these aspects.305 Becoming a trustee imposes burdens on businesses, particularly small businesses, which are not currently compensated by the
IRS.306 Many small businesses with payroll obligations hire companies called payroll providers to assist them through the thicket of rules and regulations necessary
to comply with the employment tax rules—particularly the withholding rules.307
About half of the states have addressed the cost of administering the sales tax
receipts by carving out from the payment of sales taxes a small fraction of the cost
which the business entity may keep as an acknowledgement of the cost of administering the tax system for their state government.308 A similar system could be
adopted for the IRS to compensate small businesses for the cost of handling the
withholding or the excises taxes they must collect.309 The system would be built
using incentives that reward compliance, creating a carrot for entities that timely
48
Fogg
complied with their handling of collected taxes.310 As the early years of a small
business are challenging both financially and educationally, it is recommended
that the incentive be limited in time to the first three years of the new entity’s
existence. These are the years the incentive would do the most good to aid the
business over a difficult financial time and to train the business leaders in good
management of the money held in trust.
At present, penalties serve as the only motivator to timely pay collected taxes
over to the IRS. While penalties may have a deterrent effect, they do not provide
a motivating influence on small businesses struggling to meet the cash demands
placed upon them.311 The duty to collect and pay excise and employment taxes is
an expense that falls most heavily on small businesses. Providing some relief from
this expense while simultaneously providing a financial incentive to timely file and
pay should improve compliance.312
Using ABC, Inc. as the model again, an illustration of the way the incentive
would work may be seen. Assume that ABC was a small business and it timely
filed its employment tax returns for all quarters of 2008. ABC would qualify for a
small discount on its employment tax liability for each of those quarters. If, in a
future quarter, it failed to timely file or timely pay its employment tax for a quarter,
it would lose the ability to receive a discount for that quarter.
Providing a small incentive to pay collected taxes not only follows policy considerations for effective tax administration but also relieves the cost burden imposed on small business by the current structure.313 Compliance with the employment and excise tax rules concerning collecting taxes for the IRS simply imposes
a duty on small companies to aid in governmental function without compensation.314 This proposal would recognize that cost to the entities while withholding
such payments if the business entity did not meet its collected tax obligations.315
This proposal limits itself to the costs incurred by small businesses and does
not provide any monetary incentive for large and mid-size businesses. The reason
for that limitation has two bases. First, the vast majority of unpaid collected taxes
occur in small businesses and not large or mid-size businesses. Larger businesses have legal counsel who advise the owners on the potential individual liability
which exists in failing to pay over the collected taxes. Individuals running these
businesses may have a large equity stake in the businesses but usually they do not.
In these businesses the entity does not serve as an extension of the individual;
rather the individuals responsible for running the entity have a separate identity
from the entity. These individuals know the severe risk to their personal finances
that exists if the collected taxes are not timely paid over to the IRS. Rewarding
these individuals with a small monetary incentive to pay the taxes does nothing to
provide them with a true incentive to pay. Those incentives already exist. Second,
the cost to the small business of compliance with the regime of holding taxes in
trust is much higher on a percentage basis than the cost to a large business.316
Collecting Collected Taxes
49
On the contrary, with small businesses, where the entity often serves merely as
an extension of the individual, incentives of this type could provide a meaningful
alternative to the individual and the entity. These small businesses typically have
thin capitalization and a constant need for ready cash. Providing these businesses
with a tangible monetary benefit for timely paying their taxes serves not only to
provide an incentive to them to timely pay the taxes, but also to remind them of
the responsibility to do so.
This proposal distinguishes between businesses based on size and age to target
the entities most likely have compliance problems with payment of collected taxes
and most likely to experience financial burdens in complying. Through the carrot
of this financial assistance these fragile entities may learn good habits at the outset with the hope that the good habits will remain once the incentive is removed.
The proposal would increase the administrative burden on the IRS by causing it
to create and run the incentive program, but the expectation is that the overall
administrative burden may be decreased by higher compliance.
Implementation of this proposal would require a legislative change to permit
incentive payments as well as administrative rules to implement the system.
Require Bonds of Individuals or Entities with History of
Failure to Pay Collected Taxes
The trustor in creating a trust can make whatever provision with respect to the duties, powers, and requirements of the trustee as he or she may deem wise.317 This
concept crossed over into the area of trustees for collected taxes, but only at the
state level and essentially only for sales taxes.318 This proposal suggests that the
concept of bonding trustees should migrate to the collected federal tax regime to
provide protection for the IRS and an incentive for prompt payment for the entities and their responsible officers acting as trustees. As state laws on this subject
essentially only impose the bonding requirement on sales taxes and because the
primary federal tax situation in which bonding would arise is employment taxes, a
brief discussion of the differences in those situations is necessary.
This proposal permits the IRS to require a bond from an entity if the entity
failed to pay collected taxes on a timely basis or if one of the principals of the entity
was previously assessed a responsible officer penalty. Together with the power to
require a bond, the IRS needs an enforcement mechanism when an entity continues to operate without posting the required bond. That enforcement mechanism
should include quick access to the federal district court or Tax Court to enjoin the
entity from continued operation without the posting of the bond. Without the
ability to enjoin the entity, the bond requirement will have little effect.319
Using ABC, Inc. again, an example of how the bonding provisions would work
can be shown. When ABC begins its business, the IRS would not require a bond
50
Fogg
unless one of the persons responsible for payment of ABC’s collected taxes was
someone who was previously assessed a responsible officer penalty. As long as
ABC continued to timely file and pay its collected tax obligations, the IRS would
not request a bond from ABC. However, if at some point ABC fell behind in filing or paying its collected tax returns, then the IRS could require that ABC post a
bond to ensure payment of collected taxes in the future. The amount of the bond
would relate to the quarterly liability incurred by ABC.
The state laws cited above almost uniformly impose the bonding requirement
only in sales tax situations.320 State sales taxes operate in a manner essentially
identical to excise taxes on the federal level. In these situations a customer comes
to the entity seeking to purchase goods or services. A sales or excise tax exists on
the goods or services purchased. The entity selling the goods or services collects
the sales or excise tax at the time of the purchase of the goods or services, and then
holds the tax so collected for the governmental unit to which it relates.
In these situations the entity actually received money from a third party that it
holds in trust. Cold hard cash, or its electronic equivalent, exists in the bank account of the entity collecting such payments. In contrast, employment taxes do not
involve the receipt of a payment from any outside source. In the employment tax
context, the entity makes payroll and pays its employees wages in the contracted
amount at the contracted time. At the time of the payment of the wages, the entity
takes on an obligation to pay over to the appropriate governmental unit an amount
equal to the withheld income and social security taxes credited to each employee
at the time of the payroll payment. The entity holds the withheld income and social security taxes in trust for the governmental unit; however, there may or may
not actually be any “cold hard cash” associated with the trust the entity holds for
the governmental unit. In some situations, entities will set up separate trust bank
accounts and deposit into those accounts the amount of money necessary to fund
the employment tax trust. In many situations, the entity simply has an entry on its
books that it owes the governmental unit a sum certain for the withheld income
and social security taxes, but no actual dollars are set aside and the entity may have
no cash reserves with which to pay the collected taxes.
The sales tax situation involves the entity actually holding a third party’s money
paid to the entity to hold in trust, whereas the payroll tax situation involves a
sometimes fictional trust in which the entity never actually deposited dollars and
certainly does not hold money belonging to third parties for the payment of the
trust liability.321 Therefore, a question exists as to whether a distinction can be
drawn between the two types of collected tax situations for purposes of imposing a
bond or requiring other actions. States seem to have drawn a distinction between
collected taxes that a business entity receives in hand, e.g., a sales tax, and a collected tax a business entity should establish, e.g., employment taxes. The distinctions states have drawn contain no discussion of why states do not require a bond
Collecting Collected Taxes
51
for unpaid employment taxes. States have simply created laws only applicable to
sales taxes even in states in which both sales and employment taxes exist.322 This
proposal does not recognize the distinction between sales and employment taxes
for purposes of setting up a bond requirement. Both situations involve taxes collected for a governmental unit. The fact that in one situation the entity holds
money coming from third parties for the governmental unit and in the other it
holds money the entity itself must set aside does not seem an adequate basis for
distinguishing between the two situations for purposes of determining whether a
bond requirement advances the collection of the unpaid collected tax.
Section 6672 has a rarely used provision that an individual tagged with the
liability under that section can use a bond to forestall collection on the liability while a lawsuit for refund takes place.323, 324 The bond described in section
6672 (c) occurs totally at the discretion of the individual allegedly responsible and
occurs after non-payment of the collected tax has resulted in a section 6672 assessment. Almost no one used this provision before 1998 because the IRS had a
policy of generally not collecting on the section 6672 liability while the refund action played out. However, with the passage of IRC section 6331 (i) as a part of the
Revenue Reform Act of 1998, collection action during a refund proceeding for a
divisible tax essentially became impossible absent jeopardy or consent by the taxpayer.325, 326 Nothing in section 6672 or the employment or excise tax provisions
allows the IRS to require a bond of certain individuals or entities, even if those
individuals or entities have demonstrated in the past that they do not timely file
the appropriate returns and pay the collected taxes.327
The bonding requirement provides incentives for entities to remain current on
their payments to the IRS for collected taxes since no bond is needed for entities
that remain current. A new entity would not need to post a bond to ensure payment of its collected taxes unless the IRS had concerns about the entity’s ability
to timely pay the collected taxes. Such a concern, in a new entity, would arise
if one or more of the individuals identified as responsible for that entity had a
previous section 6672 assessment. Assuming none of the responsible officers had
a previous section 6672 assessment and the entity had no history of noncompliance with payment of collected taxes, the IRS would generally not seek to impose a bond unless some demonstrated concern about payment of the collected
taxes existed.328
If an entity fell behind in filing the returns for collected taxes or paying the collected taxes, then the IRS could impose a bond on the entity to ensure payment
of the taxes. The decision to require the posting of a bond would belong solely to
the IRS. The amount of the bond should relate to the amount of the collected tax
exposure the IRS faces. Current section 6672 (c) (3) sets a good limit on the bond.
It requires a person assessed a section 6672 liability to post a bond equal to one
and one-half times the amount of the assessed section 6672 liability. A similar
52
Fogg
limit of one and one-half of the unpaid collected taxes due from the entity for past
quarters and projected due from the entity in the quarter in which the bond imposition occurs would serve as an adequate safety net for the IRS. The trigger for
the imposition of the bond could be a letter from the IRS to the entity notifying
the entity of the unpaid collected taxes, calculating the necessary amount of the
bond, and giving the taxpayer fifteen days to obtain the bond and provide proof
to the IRS of its existence. The failure to obtain and present the bond within fifteen days after notification could trigger the right of the IRS to obtain an injunction requiring cessation of all business activities to avoid further increases in the
amount of the unpaid collected taxes.329
Another aspect of the bonding requirement concerns termination of the bond
upon timely performance of the taxpaying obligations over a period of time. At
least one state that requires the posting of a bond in certain circumstances allows
the bonding requirement to lapse after the entity has met its collected tax obligations for a specified period.330 The federal statute should adopt a similar lapsing
provision to provide an incentive to entities to meet their obligations and thereby
reduce their operating costs.
The bonding statute should have a triggering mechanism tailored to the overriding concern in the collected tax arena—pyramiding of liabilities. Frequently,
the IRS has no basis for closing a business that engages in pyramiding and it can
do nothing to collect from the entity as the entity continues operations but does
not pay over its collected tax obligation. This means that the government subsidizes the continuation of the entity with little or no expectation of ever receiving
payment on the collected taxes.
Entities that do not pay their employment taxes for more than one quarter
or that have an outstanding collected tax liability in excess of $10,000 (or some
other number sufficiently high to raise concerns about pyramiding) should be
subjected to the bonding requirement. If the bonding requirement applies the
IRS can contact the entity in person or by certified mail and request the posting
of a bond. The taxpayer would have fifteen days (or some other relatively short
period of time) to obtain the bond and notify the IRS of its procurement.
The bond would remain in place until the taxpayer cured the outstanding liability and kept current on its collected tax liability for two consecutive quarters.
If the liability was satisfied and the taxpayer kept current, then the IRS would
release the taxpayer from the bond obligation. If the taxpayer incurred further
collected tax liabilities while the bond was in place, then the IRS could call the
bond and liquidate it to satisfy all outstanding federal tax liabilities of the entity.
To continue operations the taxpayer would need to post a new bond, within fifteen days after liquidation of the first bond, equal to twice the first bond.
In situations in which the taxpayer was notified by the IRS of the need to post
a bond or in which a bond was liquidated, the failure of the taxpayer to purchase
Collecting Collected Taxes
53
a bond as required would trigger for the IRS the right to seek an injunction. The
injunction would prohibit the business from incurring further collected tax liability without first paying the outstanding balance on the current collected tax
liability and posting a bond to insure proper payment in the future.331
For the injunction process to be effective, it should involve a summary proceeding similar to a summons enforcement action.332 The proof required to
obtain the injunction should require merely a transcript of account demonstrating a collected tax liability in more than one quarter or an outstanding
collected tax liability in excess of $10,000 together with proof of the mailing
or hand delivery of the request for a bond and the refusal or failure to procure
the bond. Once the IRS proved the existence of the liability, the request for
the bond and the failure on the taxpayer’s part to procure the bond, the federal
district court should enjoin the taxpayer from further operation until the payment of the liability or the posting of an appropriate bond. In this manner the
IRS would have the tools necessary to stop further pyramiding of the collected
tax liability.
The bonding requirement conforms not only with current trust law but also
with the literature on incentives. As mentioned above, many trustees, particularly
trustees having control over large amounts of money, must post a bond even if
they operate within a bank that has handled trust matters for hundreds of years.
Requiring a bond for tax collection trustees who have demonstrated a problem
with timely meeting their obligations seems only logical as a method for controlling a group of trustees that the IRS has no choice in selecting. The existence of
bonding requirements for entities also provides an incentive for entities collecting
taxes to meet their obligations to avoid the bonding requirement, or to rid themselves of the bonding requirement if already imposed.333 Coupling the bonding
requirement with the ability of the IRS to obtain an injunction provides teeth to
the bonding requirement. The goal of this provision is to stop pyramiding at the
earliest possible point and to avoid having to seek collection from an entity that
probably has nothing from which to collect.
This requirement will impose burdens on both the taxpayer and the IRS.
Entities hit with the bonding requirement will face a significant burden because of the cost of posting a bond. This burden may lead to some business
closures at an earlier point than would otherwise occur. The implementation
of the bonding requirement will also impose additional administrative duties
on the IRS although the increased compliance brought about by the bonds
may, on the whole, decrease the administrative burden the IRS faces with respect to collected taxes.
The bonding proposal requires legislative action both to create the bond and to
create an enforcement mechanism through the injunction. Significant administrative rules as well as structure will also be necessary to implement this proposal.
54
Fogg
Eliminate the Withholding and Social Security Credit
for Responsible Individuals Who Fail to Pay the Entity’s
Collected Employment Taxes
The current system rewards individuals who make decisions causing the entities
they control to fail to pay over collected employment taxes. While the reward received by these individuals may not itself create an incentive to fail to pay the employment taxes, the manner in which the failure manifests itself on the individual’s
return certainly does not deter future failures. A change is needed in the system
for crediting individuals for withheld income and social security taxes to eliminate
the benefits available to those who cause these taxes to go unpaid.
This proposal seeks to eliminate the credit for income and social security tax
payments received by employees of an entity if those employees were responsible
for the failure of the entity to pay over the collected taxes. Currently, IRC section
31 grants a credit to all employees for the amount of income and social security
taxes withheld from their paychecks.334 The credit should not extend to individuals who caused the collected taxes not to be paid. This proposal would eliminate
the credit for responsible officers if the return on which the collected taxes are
reported is not timely filed and all taxes shown thereon are not timely paid.
Perhaps the best way to illustrate the current system is through an example. The
circumstances at ABC, Inc., introduced above, can also work for this illustration.
Assume that George serves as the President and Chief Executive
Officer of ABC during all quarters of 2008, and that George holds
one-hundred percent of the stock of ABC. No one other than
George has responsibility for the decision of whether to pay the
taxes collected for the IRS over to it. When the IRC section 6672
assessment occurs for ABC’s unpaid collected taxes, the assessment
will only be made against George.335 In addition to his ownership of
the company, George also serves as one of the twenty-five employees
of ABC making $500 per week for his work.
During each week in 2008, ABC holds out $50 of income tax and
$25 of social security tax from George. The total amount of income
tax withheld during 2008 is $2,600, and the total amount of social
security tax withheld is $1,300. ABC issues George a Form W–2
at the conclusion of 2008 showing his wage income of $26,000
together with his withheld income and social security taxes. George
files his return for 2008 showing the income reported on the W–2
and claiming credit for the withheld income tax. A report goes to
the Social Security Administration (SSA) showing the total amount
of compensation George received in 2008 together with the social
security contribution made by him through his employer. George
Collecting Collected Taxes
55
receives credit from SSA for the four quarters he worked in 2008
and he also receives credit on his 2008 income tax return for the
withheld income taxes.336, 337 George individually receives credit
from SSA and from the IRS even if ABC never ends up paying the
collected taxes to the IRS and the IRS never collects these taxes
from George pursuant to section 6672.
In this situation, George receives a significant benefit from the government,
the government receives nothing, and George created the situation in which the
government receives nothing. Imagine further that most individuals who hold
the position of President and Chief Executive Officer of an entity do not receive
the same compensation as the other workers. Usually, the individual who caused
the failure to pay the collected taxes receives the highest compensation of all of the
employees. While it makes sense to credit all of the non-responsible employees
whose taxes the company withheld, it does not make sense to provide this benefit
to the very individual (or individuals in the case of more than one responsible officer) who caused the problem in the first place. For the “innocent” employees,
they had no choice concerning the withholding of the tax from their paychecks
and no choice concerning the failure of the company to properly remit the collected amount to the IRS. The company served as the agent of the IRS and not of
the individual “innocent” employees.
In contrast, the individual responsible for failing to pay over the collected taxes
had total control. That individual, acting through the entity, served in a position
of trustee. That individual breached the trust. That individual made the decisions
causing the failure to remit the collected taxes to the IRS. That individual should
not receive a reward for these failures by having the unpaid collected taxes credited to his or her account. Instead, that individual should bear the burden of the
loss as a consequence of the actions. How can that individual properly bear the
burden created by their actions? Eliminate IRC section 31 (a) with respect to the
responsible person for withholding taxes and eliminate the parallel provision with
respect to the withheld social security taxes.338, 339 Make the individual personally
responsible for payment of these taxes prior to delivery of any benefits under social security for those quarters or any benefits for withheld income taxes.
These benefits should be eliminated for each quarter in which the entity does
not timely file the appropriate return and remit the full amount of the collected
taxes. Any shortfall in remission of the collected taxes should be treated as the
failure to pay the taxes withheld from the responsible officer(s). As a consequence,
the late filing of the employment tax return by the entity (absent reasonable cause)
or the late payment of the taxes reflected on the return (absent reasonable cause)
will result in the loss of the credit for withheld income taxes on the return of the
individual responsible officer(s), and will result in the loss of any social security
benefits for the individual responsible officer(s) for the quarter for which the return was late or unpaid.340
56
Fogg
Receiving credit for the withheld taxes not remitted to the IRS provides a
significant benefit to those individuals whose actions cause the problem.
Eliminating this benefit places consequences on the actions of the individual
responsible officer that have immediate and tangible effects. The responsible
officer will immediately need to find the funds to pay the personal income tax liability or the IRS will have the right to begin collection of that liability as soon as
the assessment occurs on the income tax return. While the first proposal, above,
seeks to collapse the long waiting period before collection begins on responsible
officers, this proposal will have that effect with respect to a portion of the liability, assuming that the individual timely files an income tax return.
This provision also creates a real penalty for those individuals who fail in
their responsibilities to see that collected employment taxes are remitted to the
IRS. The current situation provides no penalty for the failure to ensure the payment of the collected taxes and actually provides a perverse incentive since the
individual committing the bad act receives the full tax benefit, and the IRS must
bear the full brunt of the loss where the tax goes uncollected. Reversing that
equation eliminates the inequity of having responsible officers benefit from their
actions. With this approach, the responsible officer may never pay the income
tax liability and the section 6672 liability may never get paid, but, at least the
responsible officer remains liable for both. Further, the elimination of the social
security benefits may impact the responsible officer in a tangible way. It reduces
the amount of eligible quarters the individual can accumulate. It potentially
reduces from the equation “phantom” high earning quarters in which the individual gets credit for high earnings yet the taxes themselves never make it to the
social security fund.
These tangible and immediate impacts on an individual may have a benefit in
driving the individual’s behavior with respect to the collected tax of the entity.
The current system, which insulates the individual from their own bad acts at
the entity, further fosters a culture of treating the collected taxes as just another
liability instead of a liability held in trust. This proposal may cause potentially
responsible persons to ensure the payment of the collected taxes because of the
direct impact payment has on their personal income tax liability.
If the collected taxes never get paid, this proposal creates logical symmetry
between the unpaid collected taxes on the employment tax return and the individual’s income and social security taxes. The current system can hardly be justified by anything other than administrative convenience. Allowing the persons
who breached their duties as trustees to receive the tangible benefit of a credit for
the very taxes they caused to go unpaid is almost impossible to justify.
Even in situations in which the collected taxes eventually get paid, this proposal
simply serves as a penalty that seeks to modify behavior and impose a true penalty
on inappropriate conduct. The potential loss of the credit on their personal return should create a very tangible incentive to insure timely payment of collected
Collecting Collected Taxes
57
taxes. The administrative burden on the IRS of administering this penalty should
not be great.
This proposal will require a legislative change to section 31 to prevent the
crediting of the withheld tax to the responsible officer’s individual income tax
liability. It will also require a change to the social security laws.
Create an Incentive for the Responsible Officers to
Pay the Internal Revenue Code Section 6672 Liability
Where More Than One Responsible Officer Exists
A truly perverse system exists today when multiple individuals owe the liability
under IRC section 6672.341 The system provides exactly the opposite incentive
that one would want for a tax collection system. The system can best be described
through an illustration. Again, ABC, Inc. will supply the background for the
illustration.
Assume that three individuals bear responsibility for the payment
of the collected taxes at ABC. During the first quarter of 2008,
ABC fails to pay its employment taxes. During the entire quarter,
George, Mary, and Bob each had the requisite responsibility and
willfulness to make them liable for the section 6672 assessment.
A section 6672 liability of $24,375 is made against each of them
on April 1, 2009, and notice and demand made to each of them
pursuant to IRC section 6303.342 Then collection begins.
Because of policy statement P–5–14, the IRS treats the section 6672 liability
as an alternative means of collecting the unpaid collected taxes and not as a penalty.343, 344 Therefore, it seeks to collect the liability only once.345 In seeking to
collect the liability once, however, it will seek to collect it from each of the three
individuals until it obtains full satisfaction of the liability.346
Assume that George sends the IRS $24,375 which it receives on
April 6, 2009; that Mary sends the IRS $24,375 which it receives on
April 7, 2009; and that Bob sends the IRS $24,375 which it receives
on April 8, 2009. Setting aside for purposes of this example the
issue of interest and penalties and assuming that $24,375 provides
full remittance, the IRS now has three times the amount it seeks
to collect. Since it seeks to collect the tax only once, it will return
the excess to the responsible officers, and it has a protocol for
returning the excess proceeds.347 The IRS protocol calls for it
to send Mary and Bob back the full amount that they paid to
the IRS because it keeps the first money that it collects. George
has the right, pursuant to section 6672 (d) to sue Mary and Bob
58
Fogg
for the right of contribution but no right to refund from the IRS
(assuming that he was correctly assessed).
The application of Internal Revenue Manual (IRM) 5.17.7.1.9 creates precisely
the wrong type of incentive for people in this situation. This provision essentially
states to multiple responsible officers that you are a chump if you choose to pay the
liability and you should do everything in your power to avoid paying the liability
before your fellow responsible officers do so. Presumably, IRM 5.17.7.1.9 seeks to
foster the rule of only collecting once as set forth in Policy Statement P–5–14; however, it promotes in responsible officers a strong desire to do everything in their
power to avoid paying the section 6672 liability in hopes that the IRS will collect
first from one or more of the other responsible officers. If the reason for adopting
IRM 5.17.7.1.9 was to foster the rule of only collecting once, that rule can be honored while still promoting the prompt payment of the collected taxes.
Instead of driving people away from payment, the IRS should adopt a policy
that promotes prompt payment of the collected taxes. To do this it could adopt a
policy somewhat parallel to the whistleblower provisions that promote individuals
to come forward to the IRS with information by providing rewards. Here the goal
would be to promote the responsible officers to come forward with early payment
in return for a potential reward.348 The potential reward in the multiple responsible officer contexts is the refund of the money paid by the earliest responsible
officer to step forward and pay the liability. The IRS should seek to reverse the
scenario described in the example above in which George, the first to pay, lost all
rights to the money while Mary and Bob, who paid later, had their money returned
to them in full.
This proposal suggests that the IRS treat the first person to pay as someone who is
eligible for a reward of having all or a portion of the money returned to them should
the IRS succeed in collecting from one or more of the other responsible officers.349
This proposal contemplates that the IRS would continue to collect the responsible officer liability from everyone against whom it was assessed. Collection efforts in this
regard could be fairly passive such as filing the notice of federal tax lien and making
offsets. Even these relatively passive efforts have the result of bringing a fair amount
of money into the IRS coffers. The goal of continued collection allows the money
recovered from the responsible officers who did not pay first to be refunded to the
responsible officer who paid first. Through this mechanism the responsible officer
paying first has the potential to fully recover and the IRS stands indemnified from
the time of the payment by the first responsible officer. A further incentive could
be created if the late payment penalties recovered from the later paying responsible
officers were given to the first paying responsible officer as a bonus for paying first.
All of this could be accomplished without significant collection effort by the IRS
other than filing the notice of federal tax lien and making available offsets. The
IRS has no responsibility to take enforced collection action against the later paying
Collecting Collected Taxes
59
responsible officers, although the IRS could take such action if it so chose. The IRS
could also file claims in bankruptcy if one or more of the responsible officers filed
bankruptcy petitions. The responsible person paying first has no guarantee that the
other responsible officers will pay and that through such payments some or all of the
money paid over will be returned as a refund, but at least the first paying responsible
officer has a chance for that to happen and an incentive to pay quickly to reap that
potential benefit. This proposal maintains the ability of the first responsible office to
seek contributions from the other responsible officers as an alternative to seeking a
refund from the IRS as it collects their payments. This proposal does not change the
policy of only collecting the unpaid collected tax once. It simply modifies the manner in which the one payment is achieved by providing an incentive for one of the responsible officers to pay quickly and serve a role that is essentially that of guarantor.
The early payment proposal should effectively guarantee the IRS payment of the
collected taxes in those situations in which someone steps forward. By providing
an incentive to pay and removing a significant impediment to payment, this system
should greatly increase collection effectiveness. This proposal should not present a
significant burden to either the responsible officers or the IRS. This proposal merely
requires a change in the administrative rule concerning the posting of payments.
Conclusion of Structural Change Section
Having created a largely successful structural system of using business entities to
collect federal taxes directly from the parties subject to the tax, Congress and the
IRS have failed to devise a structure that will insure the business entities that collect these taxes actually pay them over to the IRS. The proposals herein seek to
begin a discussion designed to create an appropriate structural system that ensures
the IRS ultimately receives payment. The variety of the proposals addresses the variety of reasons for non-payment. Giving thought to creating the proper structure
to ensure payment should help in reducing this corner of the tax gap.
Charge Interest and Penalties to Responsible
Officers Based on the Liability of the Entity for
Unpaid Collected Taxes—Accountability
As will be discussed in more detail below, almost 100 years ago Congress reacted to
the circumstance of entities that collected customs duties for the United States and
failed to pay those collected taxes over to the United States by enacting a criminal
penalty for such behavior. The criminal penalties soon expanded to include civil
penalties as well. Those civil penalties permit the IRS to assert a liability against all
parties responsible for failing to pay over the collected taxes. This “civil penalty” is
60
Fogg
not a true penalty but rather a collection device designed to permit recovery of the
unpaid collected taxes. As a result of the placement of this responsible person liability in the Internal Revenue Code during the 1954 codification process, interest
does not run on this liability due from the responsible persons until the separate
assessment occurs with respect to each responsible person. The uncoupling of
interest and penalties from the liability of the entity liable for the payment of the
collected taxes creates incentives harmful to the effectiveness of this provision.
Proposal
Charge interest and penalties to responsible persons of entities that fail to pay
over collected taxes equal to the interest and penalties for which the entity itself
is liable. By charging interest and penalties to responsible persons equal to the
amounts charged to the business entity, the purpose of the statute to recover the
unpaid collected taxes receives fulfillment. Additionally, incentives for responsible persons to delay assessment are removed as well as incentives to pay state
taxing authorizes who currently charge interest and penalties back to the business
entity due date for collected taxes.
Operation of 6672
Sections 3102 (a) and 3402 (a) obligate every employer to withhold (or collect)
from its employees’ wages income and social security taxes.350 The statutes require
the employers to pay over to the Treasury these collected taxes and section 7501 (a)
provides that these collected taxes constitute a special trust fund for the benefit of
the United States.351 The term “person” used in 7501 (a) means the employer and
“person” for this purpose is defined by 7701 (a) (1) to mean an individual, a trust,
estate, partnership, association, company or corporation.
Section 7501 (b) provides that persons violating the trust established in 7501 (a)
bring into play the penalties imposed in 6672 and 7202. Section 7202 provides
criminal sanctions for failure to pay over collected taxes in certain circumstances.352 IRC 6672 lays out what the code describes as a civil penalty but which acts
as a collection device.353
Section 6672 is sometimes called the Trust Fund Recovery Penalty (TFRP) and
in other instances the responsible officer penalty or 100 percent penalty. The TFRP
label derives from the fact that the taxes collected by the company constitute a
trust for the United States and that 6672 seeks to provide an alternate means of
recovering the trust fund when the company does not pay over the monies held
in that trust. This article will use the term “collected tax” rather than trust fund to
describe the taxes collected by the company and not paid over to the United States.
While the article will focus its discussion on income taxes withheld by employers,
Collecting Collected Taxes
61
the types of taxes in which the government uses third parties to collect spans a
broad range making collected tax a more appropriate term than simply withheld
taxes or trust fund taxes.354 Also, the term trust fund tax implies that a trust exists when, in fact, it often does not because the trust res does not exist or has not
been identified.
The responsible officer penalty label comes from the person to whom the
penalty applies. This article will use the term responsible officer to describe the
persons who meet the tests in section 6672 for piercing the corporate veil and
imposing derivative, personal liability. The term responsible officer penalty will
not be used to describe 6672 except as its use comes from specific case language.
Similarly, the term 100 percent penalty derives from the imposition of a liability
upon responsible officers equal to 100 percent of the unpaid collected tax. Except
for occasions when use of that term comes from specific case language, this article
will not refer to 6672 with the term 100 percent penalty.
Collection of taxes through withholding operates as an efficient and effective
means of collecting taxes; however, when the business collecting the taxes has cash
flow problems, the collected taxes which the business should hold in trust for the
government become a potential source of salvation that proves too tempting for
some to resist.355 Typically, the business owner faced with an inability to meet
ongoing expenses and a bank unwilling to extend further credit seeks a “short”
term solution by not paying over to the government the funds it holds or should
be holding in trust.356 Unlike other creditors who know the taxpayer’s business
and who are generally quick to react to nonpayment, the government responds
slowly to nonpayment. This slowness may encourage the business to continue the
practice of nonpayment of the trust fund taxes in the mistaken belief that either
the business will soon turn around or the government does not care about the
nonpayment. When the government finally arrives to recover the taxes due to it,
the unpaid tax bill for collected taxes has reached levels the business cannot repay.
The business ceases to exist leaving a large unpaid bill to the government for the
taxes it “held” in trust.
It may help in the overall understanding of what happens in these cases to look
at the situation briefly from the government’s perspective. The IRS does not know
how much income tax each company withholds during a specific quarter until the
company files its quarterly employment tax return. With the possible exception
of some very large corporations or corporations with past delinquencies, no one
at the IRS monitors the daily, weekly, monthly or quarterly practices of a particular company with regard to the payment of the income taxes collected from its
employees. If a company files a tax return and on that return it lists a liability for
which it does not remit payment, the IRS will assess the liability reported on the
return and initiate the collection process. If a company fails to file a return, the
IRS will usually notice that failure within a few months and initiate the collection process. Even the initiation of the collection process does not mean that an
62
Fogg
actual person will make contact with the company for weeks or months after the
collection process begins because correspondence will usually occur first followed
afterward by the assignment of a human.357 This explanation of the typical process
merely shows how a company that has collected taxes for payment over to the IRS
could fail to pay the collected taxes over to the IRS for a reasonable period of time
before the IRS will enter the scene and demand its money. It is easy to contrast the
IRS action with trade creditors and commercial creditors who usually notice nonpayment much sooner. Consequently, a company experiencing cash flow problems may naturally tend to keep current with trade and commercial creditors and
delay on payment of the collected taxes.
At this point IRC 6672 comes into play. Section 6672 allows the IRS to impose
a liability, labeled a penalty, equal to the unpaid collected taxes on those persons
who were (a) responsible for the payment of the collected taxes to the government
and (b) willful in their failure to pay the trust fund taxes over to the government.
The IRS may assess more than one person if more than one person meets the statutory tests.358 The IRS policy takes the view that the unpaid trust fund taxes should
be collected only once.359 Even though it is possible for several people to be assessed the 6672 liability, the IRS will usually first attempt to collect from the entity
that incurred the liability.360 In circumstances in which the IRS cannot collect
from that entity, it will turn its enforcement mechanisms toward the responsible
officers usually collecting from the responsible officer who presents that easiest
case for collection.361 If the IRS collects full payment from the entity or from one
of the responsible officers, then it will stop and not seek collection further.362 If
the IRS collects from more than one responsible officer and collects more than
the total liability for collected taxes, then it will refund the excess to the person(s)
from whom it collected after it received full payment.363
It is not uncommon for the inquiry into the liability under IRC 6672 to take
several months after the taxes were due and the inquiry itself lasts several months
longer. So by the time the IRS makes an assessment against a responsible officer,
one or two years have passed since the return for the collected taxes was due and
since interest (and penalties) began accruing on the underlying tax obligation of
the entity.364
This paper presumes collected taxes went unpaid by the entity and that one
or more persons were responsible for that underpayment. As mentioned above
the tests for who is liable for the IRC 6672 penalty involve both responsibility
and willfulness. Much has been written on these tests and on other aspects of
this liability; however, for purposes of the discussion in this paper, liability exists
leaving the question of payment and more precisely the payment of interest on
the obligation.365
Collecting Collected Taxes
63
An Example
The example below illustrates the manner in which collection of the 6672 liability
is collected in the current system.
ABC, Inc. employs 50 people. It has a quarterly payroll of $300,000.
ABC’s management is led by Bob Smith, President; Mary Jones, Vice
President and John Doe, Treasurer. For the first quarter of 2008,
ABC experienced a sharp dip in orders due to a recession in the
U.S. economy. The dip in orders led to cash flow problems at ABC.
Bob, Mary and John met to discuss the cash flow problems. They
decided that ABC could keep afloat without incurring significant
additional bank debt if it delayed paying the payroll taxes to the
IRS. So, instead of paying the $75, 000 in payroll taxes to the IRS,
ABC mailed in its quarterly Form 941 reporting this amount of
liability with no remittance. The $75,000 in payroll taxes consist
of three parts: the employer liability, withheld social security taxes
and withheld income taxes. For purposes of this example, the
withheld income taxes make up $50,000 of this amount in each of
the quarters and the amount of the withheld social security taxes,
also a collected tax subject to 6672, is ignored.
Although ABC’s management expected an upturn in the second
quarter that would allow them to catch up with the missed payment,
things only got worse. Consequently, they again decided not to
send into the IRS the payroll tax payment for the second quarter
which again would have totaled $75,000. The IRS continued not to
bother the company. Management knew things would get better
and that in the 3rd quarter, they would catch up. Unfortunately,
orders continued to decline as the year progressed. ABC was again
unable in the 3rd quarter to pay its payroll taxes of $75,000. Other
bills were also being delayed or being left unpaid. Creditors were
calling at an ever increasing pace. Finally, in late September, ABC
heard from the IRS asking where the payroll taxes were for the
first three quarters. When ABC did not immediately pay the back
payroll taxes, the IRS filed a notice of federal tax lien on September
25, 2008. The filing of that notice triggered the termination of the
company’s line of credit with the bank. Without that line of credit
and with no ability to replace it, ABC could no longer meet payroll
or pay for new goods. ABC closed its doors on September 30, 2008.
64
Fogg
At that time, it owed $225,000 in payroll taxes of which $150,000
stemmed from income taxes that it withheld from its employees.
In March 2009, having concluded that ABC could not pay the back
payroll taxes, the IRS investigated ABC to determine why it did
not pay its payroll taxes. The IRS determined that the failure to
pay was due to decisions made by John, Mary and Bob. The IRS
asked John, Mary and Bob to consent to the assessment of the 6672
penalty against them. Each of them told the IRS that they were not
responsible for failing to pay over the withheld income and social
security taxes withheld from the wages of the employees of ABC
and that the problem was a direct result of decisions made by the
other two parties. Each officer appealed the IRS determination of
responsibility to the Appeals Office. In October 2009 a conference
was held in the IRS appeals office with respect to each of their cases.
The information exchange with the appeals officer and the time it
took him to reach a conclusion meant that the decision to hold Bob,
Mary and John liable under IRC 6672 came in February, 2010. The
assessment against each of them for $150,000 was made on April 30,
2010, two years after the due date of the return for the first quarter
for which the withheld payroll taxes were not paid.
Interest Analysis
Interest begins running on each of Bob, Mary and John’s 6672 liabilities on April
30, 2010, the day of the assessment of the 6672 liability. The liability of ABC for
these employment taxes arose on the payment of the employee wages. A failure to
deposit penalty could be imposed against ABC beginning on the due date of the
payment of the employment tax. That due date depends on the amount of the payroll. The due date of the return for each quarter marks the day on which interest
begins to run against ABC on the outstanding employment taxes. The quarterly
return is due on the last day of the month following the end of the quarter. For the
quarter ending March 31, 2008, the liability for interest began running on April
30, 2008.
For purposes of this illustration, the amount of interest reflects only interest
on the withheld income tax portion of the employment tax liability. Interest on
$50,000 from April 30, 2008 to April 30, 2010, at the applicable rate (using 5 percent simple interest for all quarters) would be $5,000.366 Interest on the liability for
the second quarter would run from July 31, 2008 and at the applicable rate would
be $4,380. Interest on the liability for the third quarter would run from October
31, 2008 at the applicable rate would be $3,760. The total amount of interest due
from ABC on the employment tax liabilities for these three quarters as of April 30,
65
Collecting Collected Taxes
2010, would be $13,140. As of April 30, 2010, Bob, Mary and John owe $0 in interest for the employment tax liabilities assessed against them with respect to the first
three quarters of 2008 because the 6672 liability is treated as an assessable penalty
for which interest does not begin until the liability is assessed.
The tax and interest liabilities in table form are as follows:
TABLE 2
Unpaid
Employment
Taxes
Withheld Income
Tax Portion of
Employment Taxes
Interest on
Withheld Tax, Due
Date of Return to
Assessment of
6672 Liability
1st Q 2008
$75,000
$50,000
$5,000
2nd Q 2008
$75,000
$50,000
$4,380
3rd Q 2008
$75,000
$50,000
$3,760
$225,0001
$150,0002
$13,1403
Period
Total
1 Total
unpaid employment taxes due from ABC.
amount of unpaid employment taxes due from ABC that represents collected taxes. This is the amount for
which responsible persons may be held liable pursuant to I.R.C. § 6672.
3 Total interest owed on the unpaid collected taxes between date the liability arose—the due dates of the employment tax returns—and the date of the responsible officer assessments—April 30, 2010. This amount represents the
amount owed by the entity for failing to pay over the collected taxes that is not charged under current federal law
to the responsible officers because the liability of responsible officers for interest does not start until the date of the
6672 assessment while the liability of the entity for interest runs from the due date of the return.
This chart does not display other liabilities that would be due from the corporation that are not charged to the
individual responsible officers under existing federal law but are charged to the individual responsible officers under
the laws of most states. These liabilities are the failure to deposit penalty and the failure to pay penalty. The failure
to deposit penalty arises under I.R.C. § 6656 (a). The failure to pay penalty arrives under I.R.C. § 6651(a) (2). In the
aggregate these penalties would almost always exceed the amount of unpaid interest.
2 Total
As is seen in this example, the savings to the responsible officers resulting from
the delay equals $13,140. The current system not only causes the Treasury to forego revenue for the time value of money on the unpaid taxes, it also makes the
administration of the laws much more difficult because the IRS must use its investigative powers to identify the individuals responsible for the failure to pay the
employment taxes while those individuals do little to work with the IRS to resolve
the matter.367 Instead, they will do everything in their power to avoid resolving the
matter since resolution not only means they have been tagged with the liability but
also that the liability no longer exists in an interest-free setting.
Purpose of 6672
The legislative history of 6672 traces back to a penal statute. The penal nature
displays itself in both civil and criminal manifestations. This history supports a
reasonable inference that 6672’s location in the 1954 Code in the assessable penalty
66
Fogg
provisions followed, or at least did not contradict, the purpose of the statute as
developed in the decades prior to codification. In contrast to the legislative history, the purpose of 6672 as expressed in Congressional policy, in IRS policy and
in court decisions is simply that 6672 serves as a backup mechanism for insuring payment of collected taxes. Furthermore, the Congressional, IRS and court
expressions on the policy of 6672 make clear that this statute does not create a
separate liability.
Legislative History
Penal provisions imposing criminal liability for failure to pay over collected taxes
were created in the Corporate Excise Tax Act of 1909.368 Section 6672’s history
flows though the Revenue Act of 1916 shortly after the establishment of the modern tax system following the passage of the 16th Amendment.369 Like many tax
provisions it traces this part of its history directly to a war—in this case World War
I. Congress passed a criminal penalty which applied to violations relating to the
failure to pay of certain excise taxes.370 At that time withholding of income taxes
did not exist and excise taxes provided a substantial portion of the total federal
tax revenues. Because this segment of 6672’s history manifests itself as a criminal
provision, interest did not come into play.
The statute provided:
“That whoever fails to make any return required … or who makes
any false or fraudulent return, and whoever evades, or attempts to
evade any tax … or fails to collect or truly to account for and pay
over any such tax, shall be subject to a penalty of not more than
$1,000, or to imprisonment for not more than one year or both,
at the discretion of the court, and in addition thereto a penalty of
double the tax evaded, or not collected, or accounted for and paid
over, to be assessed and collected in the same manner as taxes are
assessed and collected in any case in which the punishment is not
otherwise specifically provided.” 371
In the Revenue Act of 1918, Congress enacted section 1308 creating three tiers of
civil and criminal penalties applicable to non-compliance with excise taxes.372 The
first tier, the civil tier, provided for a monetary penalty of up to $1,000. The statute
does not specifically tie the penalty to an amount of unpaid tax and in that respect
looks more like a “regular” penalty.373 The second tier began the criminal sanctions by creating a misdemeanor liability.374 The third tier most closely resembles
the current 6672 except that this third tier imposed a criminal liability. It hit the
offending party with a “penalty of the amount of the tax evaded, or not paid, collected or accounted for and paid over… .” 375
Collecting Collected Taxes
67
The Revenue Act of 1924 made further changes.376 The changes continued to
move the language toward the current language and “except for the minor phrase
reversal from ‘any person who willfully fails’ of the 1924 Revenue Act to ‘any person required to collect, … who willfully fails to collect’ of present section 6672, the
Revenue Act of 1924 represents the last substantive amendment to the language of
what became section 6672.” 377
In 1935, the passage of the Social Security Act expanded the scope of the penalty
for failure to pay collected taxes making it applicable to unpaid Social Security taxes collected at the source in addition to excise taxes.378 This penalty was codified
in 1939 and “remained intimately and exclusively related to the criminal sanctions
until its ‘civil’ pigeonholing in the 1954 Code.” 379
In the same year it passed the Social Security Act, 1935, Congress brought into
the Internal Revenue Code the predecessor of current section 7501.380 Section
7501 provides that “the amount of tax so collected or withheld shall be held to be
a special fund in trust for the United States.” The goal behind the statute centered
on a desire to make administrative assessment and collection provisions available
and to provide further protection for the collected funds.381
Just as World War I caused Congress to create the criminal penalty predecessor
of 6672, World War II inspired another change which significantly impacted the
penalty for failure to pay over collected taxes. The Current Tax Payment Act of
1943 created the regime of tax collection from employees that still exists today.382
Congress once again grafted the withholding tax provisions into the penalty regime initially set up for excise taxes expanding this collected tax penalty provision
to reach essentially the same provisions it currently covers.383
The next act in the progression of the collected tax penalty to its modern provision occurred in the codification effort in 1954.384 In this effort the penalty for
collected taxes moved into Subtitle F, subpart B—Assessable Penalties of the newly
revised Internal Revenue Code.385 The legislative history of 6672 contains basically no explanation concerning the placement of the civil liability creating personal liability for failure to pay collected taxes in the assessable penalty section
of the newly revised Code.386 The positioning of 6672 in the assessable penalty
subpart of the Code together with the absence of any specific language in 6672
concerning interest has led to the current state of affairs in which interest does not
accrue until the liability is assessed. None of the changes to 6672 since 1954 have
addressed the issue of interest. Its placement within subpart B of Subtitle F has
remained unchanged.
Assessable penalties generally exist separate from taxes imposed under the
Internal Revenue Code. Because these penalties do not relate to a specific tax,
they do not relate back to a specific return due date or taxable event.387 Assessable
penalties generally stand alone as their own separate liability with the exception of
6672. Consequently, a separate interest provision imposes interest from the time
these penalties arise—at the time of their assessment.388
68
Fogg
Congressional Policy
The expression of Congressional policy concerning 6672 discussed here will
focus on the bankruptcy provisions concerning the liability for collected taxes.
Congressional policy expressed through the bankruptcy code demonstrates that
almost no liability shares the importance of collected taxes.
Creditors in bankruptcy cases basically belong to one of two groups: secured
or unsecured. Generally, secured creditors who do not sleep on their rights have
little concern about bankruptcy because they look to their security rather than the
debtor’s solvency for repayment. Unsecured creditors, however, have much to fear
from bankruptcy since so many debtors have little or no unencumbered assets
with which to repay their unsecured debts. Congressional policy addresses the
plight of the unsecured creditors by making some of them more equal than others.
The provisions that differentiate unsecured creditors come in two forms: priority
status and exceptions to discharge.
Bankruptcy code section 507 (a) sets forth a list of unsecured creditors that
Congress has designated as entitled to payment before other unsecured creditors. Placement of an unsecured creditor on this list significantly improves its
chances of receiving payment through the bankruptcy proceeding. The higher on
the list created by 507 (a) an unsecured creditor achieves, the more likely it will
receive payment.389
In a similar manner bankruptcy code section 523 (a) creates a list of unsecured
creditors whose debts Congress has determined receive an exception to discharge.
Creditors on this list may continue to seek collection from individual debtors even
after the individual debtor has obtained a discharge from the bankruptcy court.390
Every unsecured creditor wants recognition on this list because the “next best
thing” to receiving payment through the bankruptcy estate is having the continued ability to pursue collection after discharge. Some unsecured creditors have
sufficient fortune or influence to have their debt recognized as both a priority debt
and one excepted from discharge.391
With the exception of the liability imposed by 6672, assessable tax penalties do
not make the priority list in bankruptcy code section 507 (a).392 An unsecured
claim for an assessable penalty receives general unsecured classification rather
than receiving any priority. Assessable penalties and tax penalties in general receive even worse treatment than a general unsecured classification for cases administered under chapter 7.393 In chapter 7 cases, these penalties only receive
payment after all general unsecured claims have been paid.394 This subgeneral
unsecured classification even applies to penalty claims for which a notice of federal tax lien was filed and would, except for their origin in the penalty provisions,
receive secured status.395
Although assessable penalties do not receive priority claim classification, they
do receive an exception to discharge pursuant to B.C. 523 (a) (7).396 To qualify for
Collecting Collected Taxes
69
an exception to discharge, an assessable penalty must relate to an act or a return
due date that occurred within three years before the filing of the bankruptcy petition. The exception to discharge that applies to assessable penalties arises under
a different subparagraph than the exception to discharge applicable to 6672. The
exception to discharge applicable to 6672 is much preferred because of the lack of
a time limitation.
As further discussed below, the treatment of the liability imposed by 6672 is not
only different and more favorable than the treatment of other assessable penalties,
it is more favorable than the treatment of almost any other unsecured liability.
This special treatment appears to result from Congressional recognition of the
importance of the payment of collected taxes.397
A debate concerning the treatment of 6672 in bankruptcy occurred in the
late 1950s and early 1960s when proposals were pending in Congress to reform
the discharge provisions to reduce or eliminate the broad exception to discharge
then available to taxes.398 In 1961, Assistant Secretary of the Treasury Stanley
S. Surrey wrote to Senator Eastland, the Chairman of the Senate Committee on
the Judiciary:
“Delinquency in this area has increased in recent years, and the
Department considers it most undesirable to permit persons who
are charged with the responsibility of paying over to the Federal
Government monies collected from third persons to be relieved
of their obligations in bankruptcy when they have converted such
monies for their own use.”399
In 1966 Congress did scale back the discharge exception previously granted for
taxes but added subsection (e) to Section 17a (1) of the Bankruptcy Act.400 Prior to
the 1966 amendments, all taxes basically benefited from the exception to discharge
in bankruptcy. This broad exception provoked significant complaints from the
bankruptcy bar and certain commercial interests. With the passage of the amendments in 1966, the exception to discharge for taxes took on a form similar to that
carried into the current Bankruptcy Code, that is, the exception primarily applies
to taxes incurred within three years of the filing of the bankruptcy petition.401
A new section 17a (1) (e) provided: “That a discharge in bankruptcy shall not
release a bankrupt from any taxes… , which the bankrupt has collected or withheld from others as required by the laws of the United States … but has not paid
over… .” As the House Committee explained in reporting out the measure, the
purpose of the amendment was “to exempt from the provisions of this bill taxes
which the bankrupt has collected or withheld from others under Federal or State
law.” 402 In the House Committee’s view, “[t]he objection of Treasury to the discharge of so-called trust fund taxes has been met by the amendment to this bill.” 403
Likewise, the Senate Reports confirm that the purpose of Section 17a (1) (e) was to
render trust fund taxes nondischargeable in bankruptcy.404
70
Fogg
In the 1970s Congress spent several years creating a new bankruptcy code to
replace the Bankruptcy Act of 1898.405 In creating the new bankruptcy code,
Congress reviewed, inter alia, the types of unsecured debts that should receive
priority status and that should receive an exception to discharge.406 Ultimately,
the type of debt it singled out for an exception to discharge in 1966, collected and
withheld taxes including the 6672 liability, received special recognition in the new
bankruptcy code as a priority tax claim and as a claim excepted from discharge.407
Not only did 6672 receive priority status under the bankruptcy code enacted in
1978 when no other assessable penalty achieved such status, the liability imposed
under 6672, and for any unpaid collected tax, also received better treatment under
the bankruptcy code than any other tax of any type.408
A taxpayer entering bankruptcy with unpaid income, employment or excise
taxes, other than taxes of those types collected from others, essentially has a time
limit cap on the life of that liability before it loses its status as an unsecured priority claim.409 The time limit essentially makes income, employment or excise taxes
older than three years at the time of the bankruptcy petition, general unsecured
claims rather than claims entitled to unsecured priority status. Contrast that with
the treatment of the 6672 liability and the liability for unpaid collected taxes. A
taxpayer entering bankruptcy with unpaid 6672 liabilities has a liability that will
receive unsecured priority status no matter how old the 6672 liability is at the time
of the bankruptcy petition.
Granting the 6672 liability unsecured priority status no matter its age provides
significant recognition of the importance of this liability from Congress’ viewpoint. Priority status gives the government a much greater chance to receive payment on this liability from the bankruptcy estate that it would have as a general
unsecured claim. The unlimited time period for priority status also means that the
6672 liability will always receive the exception to discharge under B.C. 523 (a) (1)
while other taxes lose their exception to discharge with age. The combination essentially makes it impossible to get rid of the 6672 liability through bankruptcy.
This total protection from bankruptcy evinces a significant policy statement by
Congress concerning the importance of this liability. No other tax and almost no
other liability receive this type of protection.
In 2005 Congress addressed the protection for the 6672 liability again in order
to close a loophole that had arisen through case law. The change in 2005 once
again demonstrated Congress’ view of the importance of this type of liability. The
change occurred in the discharge provisions of chapter 13.410 Persons liable under
6672 are not always known to the IRS at the time they file a bankruptcy petition
because the liability is a derivative liability. Generally, the IRS does not know who
has liability under 6672 until it investigates a company after it has failed to pay over
the collected taxes. Because the identity of the debtor as a responsible officer is not
known by the IRS prior to the bar date, it fails to file a timely proof of claim. The
failure to file a timely proof of claim does not affect the exception to discharge in
Collecting Collected Taxes
71
chapter 7 and 11 cases of individuals because the exception ties itself to the status
of the IRS claim and not whether the IRS timely filed such a claim.
The IRS argued for a similar result in chapter 13 but lost that argument in
Tomlan v. United States.411 The IRS failed to timely file its claim in Tomlan. The
debtor’s plan proposed to pay in full all timely filed priority claims. The District
Court found that the plan discharged the debtor’s liability under 6672 because
of the finality of the plan and the wording of B.C. 1328 (a).412 The IRS essentially
acquiesced in the decision in the publication of its litigation position on the issue;
however, it sought to change 1328 (a) when Congress appointed a Commission
in 1994 to look into changes to the bankruptcy code.413, 414 When Congress ultimately passed the laws resulting primarily from the proposals of the Commission
in the 2005, those bankruptcy amendments included a provision addressing this
concern of the IRS.415 The result of this process is a change to 1328 (a) that prevents discharge of the liability imposed under 6672 in a chapter 13, whether or not
the IRS files a timely claim.
Congressional policy toward 6672 as expressed in bankruptcy code provisions
from 1966 to 2005 could not more strongly suggest how important Congress views
the requirement to pay the collected taxes and how different 6672 is from any
other assessable penalty. Its difference comes from its status as an alternate means
for the government to collect those taxes which have been collected for it and
which should be held in trust and paid over to the government.
IRS Policy
The principal IRS position concerning 6672 comes in policy statement P–5–14.416
This policy statement currently provides that “[t]he withheld employment and income taxes or collected excise taxes will be collected only once, whether from the
business, or from one or more of its responsible persons.” 417 This policy statement
goes back to 1956.418 The Supreme Court has cited to the policy statement and to
a Comptroller General Opinion based on this policy statement in describing the
purpose of 6672.419, 420
The policy of the IRS regarding 6672 has remained constant for over fifty
years.421 The IRS imposes the 6672 liability against all of the persons responsible
for the failure to pay a collected tax. Consequently, it may have assessments on
its books for the original liability due from a corporation plus one or more assessments of the amount of the unpaid collected taxes against persons responsible for
the corporation’s failure to pay those collected taxes over to the IRS. Despite having numerous assessments and despite the apparent ability under 6672 to collect
the full amount from each party, the IRS has consistently said that it will not use
6672 as a mechanism for collecting the full amount of the unpaid collected taxes
from each party assessed. Rather, it takes the view that 6672 is simply a device for
72
Fogg
the collection of the unpaid taxes collected by the corporation. It is not a separate
liability. As such, the IRS seeks to collect only once from either the corporation or
any of the responsible persons.422
Using the example of Bob, Mary, and John as responsible officers of ABC for
three quarters of 2008 for a total of $150,000 in unpaid collected taxes, an illustration is possible. Once the IRS makes the responsible officer assessments against
Bob, Mary and John it will have four assessments on its books for the recovery of
this same $150,000. Because of the policy statement, the IRS links these four accounts in order to insure that it only collects $150,000 in tax (plus any applicable
penalties and interest).423 This policy leads to the IRS position on repayment of
proceeds received in excess of one full payment of the tax. If Bob, Mary and John
are each assessed a $150,000 6672 liability on March 10, 2010, each owes $150,000
at that moment. Suppose Bob pays the $150,000 that day at 10 a.m., Mary pays
the $150,000 at 11 a.m. and John pays the $150,000 at noon. After a thorough investigation to ascertain when the payments were received, the IRS would refund
to Mary and John their entire payments leaving Bob as the person who paid it all.
If Bob wishes to have Mary and John contribute to 6672 liability, he must bring
a separate suit against each of them for that purpose, obtain a judgment and successfully collect on the judgment.424
The IRS policy toward 6672 does not match its policy with respect to any of
the other assessable penalties. For all other assessable penalties, the IRS seeks to
collect the total amount of the taxes assessed. Unlike 6672, the other assessable
penalties are separate and distinct from any taxes to which they may relate. The
other assessable penalties perform a penal function rather than a function to recover unpaid taxes.
Court Decisions
Two court decisions provide significant insight into the view courts take toward 6672. These decisions adopt the IRS policy that 6672 exists as a collection device and, in one, reinforce the Congressional policy view concerning the
importance of 6672.
United States v. Sotelo 425
Arising in bankruptcy, Sotelo presented the Supreme Court with the opportunity
to consider the nature and purpose of 6672. Mr. Sotelo filed his bankruptcy petition in 1973 when the Bankruptcy Act (rather than the current Bankruptcy Code)
was in effect. He initially contested the determination that he owed the government pursuant to 6672; however, he did not appeal the determination that he was
liable. Instead, he shifted his argument to one based on discharge arguing that the
Collecting Collected Taxes
73
Bankruptcy Act 17a (1) (e) discharged penalties imposed under 6672. Although he
lost at the bankruptcy court and district court level, he prevailed on this argument
before the Seventh Circuit.426
First, Mr. Sotelo argued that “the liability described in 6672 itself as a ‘penalty’
and as such had been discharged in bankruptcy.” 427 Second, he argued that section 17a (1) (e) of the Bankruptcy Act did not except from discharge the penalty
imposed under 6672 but rather excepted from discharge only the liability for collected taxes due from the corporation that incurred the debt.428
The Supreme Court examined both the history of the 1966 amendments to the
Bankruptcy Act as well as the purpose of 6672 as it related to the 1966 amendments.
Through that examination it determined that the Mr. Sotelo’s 6672 liability was
excepted from discharge by section 17a (1) (e) of the Bankruptcy Act.429 It further
determined that the penalty label placed on 6672 by the Internal Revenue Code
did not matter when the Court analyzed the language and purpose of Bankruptcy
Act 17a (1) (3).430 Through this analysis, the Court determined that 6672 acted as
a device for collecting the types of taxes described in 17a (1) (e).431 As such, simply
seeking to label 6672 as a penalty did not advance the taxpayer’s argument because
the label did not control the true purpose of 6672 as it related to the discharge
provisions in Bankruptcy Act 17a (1) (e).432
The Court did not explicitly say that 6672 is not a penalty. Instead, it focused
on how 6672 operated with respect to the language of the discharge provision. In
doing so, the Court did quote from a letter prepared in 1976 by the Comptroller
General concerning IRS practices with regard to 6672: “IRS uses the 100–percent
penalty only when all other means of securing the delinquent taxes have been
exhausted. It is generally used against responsible officials of corporations that
have gone out of business… . [I]t is IRS policy that the amount of the tax will be
collected only once. After the tax liability is satisfied, no collection action is taken
on the remaining 100-percent penalties.” 433
The dissent in this 5–4 decision disagreed strongly that the “taxes” excepted
from discharge in Bankruptcy Act 17a (1) (e) equated to the “penalty” imposed
by 6672.434 The dissent pointed to the legislative history of 6672 in support of
the penal underpinnings of the statute.435 The harsh result that the majority
opinion created for the individual business owner by holding the 6672 liability
as an exception to discharge was cited as support for the wrong policy direction
taken by the majority. In stark terms the dissent described the same bankruptcy result, made much clearer in the legislative history of the Bankruptcy Code
that is described above in Section 3.B.436 While the dissent expresses its significant concerns that neither the language of the Bankruptcy Act nor the policies behind it could support the majority’s decision, Senator DeConcini made
it clear just a few months later in his explanation of the Bankruptcy Code that
the position adopted by the majority in Sotelo was the position adopted for the
new legislation.437
74
Fogg
Lauckner v. United States 438
The government discovered that Mr. Lauckner met the tests as a responsible officer of AAA Trucking Corporation. The discovery, however, came after the previously presumed date on which the statute of limitations expired for making a 6672.
Prior to Lauckner, the government used as the statute of limitation for the 6672 liability the date on which the statute expired with respect to additional assessments
against the corporation that failed to pay the collected taxes. In support of its assessment after the date on which one could be made against the corporation, the
government argued that 6672 was an assessable penalty and, as such, did not have
a statute of limitations on assessment. Mr. Lauckner argued that the assessment
was time barred citing the previous position of the IRS concerning the statute of
limitations for the 6672 liability.
The only Circuit Court addressing the issue of the statute of limitations on
assessment of the liability imposed by 6672 determined that the Government
does not have an unlimited amount of time to assess this liability, as with other
assessable penalties, but has a limitations period established by the underlying
liability with respect to the corporation that collected the unpaid tax.439 This
determination, having nothing specifically to do with interest on the 6672 liability, aligns perfectly with the position that interest on the 6672 liability should
not look to the interest provisions applicable to assessable penalties but rather to
should run from the due date of the return giving rise to the liability.
No specific code section sets out a statute of limitations for the assessment of
the 6672 liability.440 For many years the IRS took the position that the statute of
limitations on assessment of the 6672 liability mirrored the statute of limitations
for the underlying tax and ran for three years from April 15th of the year following the end of the quarter in question.441 This position followed the general rule
found in 6501. Section 6501, however, applies to liabilities based on tax returns.
Probably because of a series of victories regarding the statute of limitations
in cases under 6700 and 6701, in 1994 the IRS suddenly seemed to come to the
realization that 6672 was placed into the Code as an assessable penalty that was
not based on a tax return.442 It then concluded that 6672 was a statute without
a controlling provision with respect to the statute of limitations and made an
assessment that would have been time barred under its previous interpretation
of the statute of limitations as it applied to 6672.443
To support its “new” position, the IRS made numerous arguments, all of which
were rejected. First, the IRS argued that the 6672 liability constitutes a “separate
and distinct” liability from the IRC 3403 liability imposed on the employer.444
The opening paragraph of the Lauckner opinion sets the tone for the court’s view
of the Government position; however, its rejection of the Government’s position stems from exactly the reasoning that supports imposing interest on the
responsible officer equal to the interest on the entity. In describing the nature of
Collecting Collected Taxes
75
the liability under 6672, the court stated “[a]s of the moment payment was due
and not made, both the employer and any responsible officer became liable.” 445
The court found that “it seems clear that courts have based the lower standard of conduct necessary to trigger § 6672 liability [for willfulness] on their
understanding, unchallenged until now, that § 6672 functions only as a collection device, not as a truly ‘separate and distinct’ penalty.” 446 The court went on
to hold that the 6672 assessment is separate only for purposes of collection.
The Government argued that “because the responsible person assessed under
§ 6672 files no return with respect to the assessment, the assessment is not made
with respect to any return, and the § 6501 (a) limitations period on assessments
is never triggered.” 447 On this issue the court found, however, that the 6672 liability was in fact based on the employment tax return triggering the running of
the statute of limitations under 6501. It examined several cases that had noted
‘no return’ is filed concerning 6672 liabilities and determined that “[t]hese cases
do not stand for the proposition that § 6672 penalties are not assessed with respect to any return.” 448 Therefore, it concluded that to the extent that there was
something about 6672 that was “separate and distinct” from the employer liability it was “only for purposes of collection.” 449 Important for purposes of
this article, the court held that “the assessment itself is based on the underlying
liability of the employer.” 450
If the assessment is based on the underlying liability of the employer and
is not a separate and distinct liability, then separating the two for purposes of
assessing interest makes little sense. The logical point for imposing interest
against the responsible officer is the same point when it arises with respect to
the corporation since the liabilities are separate and distinct only for purposes
of collection.451 The separateness for purposes of collection describes the separateness of the actual assessment. The IRS creates an assessment for each individual or entity liable for the 6672 liability and has a separate assessment for the
person liable for the underlying tax which includes not only the collected taxes
but also the employer portion of the liability as well.452
In an Action on Decision dated July, 15, 1996, the IRS acquiesced in the result in Lauckner.453 While the sudden change in the IRS position on the statute
of limitations seemed to also influence the court’s decision in Lauckner,454 the
basis for its conclusion supports the policy behind the position that the 6672
liability should result in interest running with the employment tax liability to
which it is so closely aligned. While the decision in Lauckner did not leave
the IRS in any worse position than it was in before it realized that 6672 was an
assessable penalty just like 6701, the picture now clearly presents 6672 as an assessable penalty with the worst of both worlds. It does not have the unlimited
statute of limitations enjoyed by other assessable penalties since it is viewed as
being tied to a return; however it does have the interest provisions of 6601 (e),
discussed infra, with other assessable penalties, denying the running of interest
76
Fogg
until the assessment occurs. This is an odd combination of handicaps to place
on a liability protecting the funds held in trust for the United States and a liability so important that, unlike any other assessable penalty, Congress gives
it not only priority status in bankruptcy but priority status for the life of the
collection statute.
The Problem of Interest
As alluded to above, the placement of 6672 in the assessable penalty provisions
positions it for treatment with respect to interest that contradicts the purpose of
6672 and that creates a lack of parallel structure with similarly situated taxpayers.
To understand how this works requires analyzing the interest provisions.
The Mechanics of the Interest Provisions in IRC 6601
Section 6601 sets out the rules, not the rates, for interest on liabilities imposed in
Title 26. The rule for taxes found in 6601 (a) provides that interest runs “from the
last date prescribed for payment of the tax to the date on which payment is received.” 455 The last date prescribed for payment of taxes generally coincides with
the due date of the tax return for that tax without taking into account extensions of
the date for filing.456 For example, the due date for individual income taxes falls on
the 15th day of the fourth month following the close of the tax year.457 That date,
April 15th, starts the running of interest for individual taxpayers for the calendar
year that ended immediately prior to that April. If an individual remits payment
for a income tax prior to April 15th either by withholding, estimated payments or
payment with the return, then no interest accrues with respect to that year’s tax
liability (unless a subsequent assessment occurs). If an individual does not remit
sufficient funds by April 15th to cover the tax liability for the preceding year, then
interest begins to run on April 16th and runs until full remittance reaches the IRS
or the IRS abates the tax.
The general rule for interest on income taxes described above also applies to
employment taxes. The employment tax return due date comes at the end of the
month immediately following the end of the quarter, e.g., April 30th for the first
quarter. If the employment taxes due for the first quarter remain unpaid as of the
April 30th immediately following the end of the quarter then interest begins to run
and continues running until paid.
The rule for interest on assessable penalties follows a different path. The
Treasury Regulation interpreting IRC 6601 (e) (2) provides that “interest will not
be imposed on any assessable penalty, … if the amount is paid within 21 calendar
days … from the date of the notice and demand. If interest is imposed, it will be
imposed only for the period from the date of the notice and demand to the date on
Collecting Collected Taxes
77
which payment is received.” 458 Notice and demand occurs simultaneously with or
immediately following the assessment of a liability.459 Since 6672 falls into the assessable penalty section of the Code, this provision and not the provision for taxes
applies to the running of interest on assessments made pursuant to this statute.
That difference appears to result solely from the placement of 6672 in the Code
and no explanation for its placement with respect to interest exists in the legislative history of 6672 or 6601.
Problems Created by the Interest Provisions Applicable
to 6672
Because of 6672’s placement as an assessable penalty and the consequent application of IRC 6601 (e) rather than IRC 6601 (a), three problems exist with respect to
the application of 6672. First, the delay in the running of interest against responsible officers treats similarly situated taxpayers in a disparate manner. This creates a fairness issue. Second, the treatment of interest for those liable pursuant to
the derivative liability of 6672 works differently than the interest charged to those
derivatively liable under similar statutes. This highlights a lack of a consistent approach with respect to parties held liable when the initial taxpayer did not fulfill its
obligation. Third, the delay in the running of interest creates problems for the IRS.
It loses the time value of money for the period between the return due date and the
date of assessment of the responsible officer. This interest free period also harms
the IRS because it creates an incentive for responsible officers to delay and burden
the system of administration to gain the benefit of the time value of money. The
postponement of interest also puts the federal government at odds with its state
counterparts providing an incentive for responsible officers to satisfy their state
obligations for unpaid collected taxes first.
The Disparate Treatment of Similarly Situated
Taxpayers
That the form of an entity or a transaction can control the outcome in a tax matter
needs no citations. Nonetheless, in certain matters varying results based simply
on form can create cries for fairness.460 Section 6672 creates a lack of fairness with
respect to the imposition of interest between those individuals who incorporate
their business and those who do not. Changing the statute to charge persons liable
under 6672 with interest back to the due date of the entity’s return would eliminate
that inequality.
The form of the entity definitely matters to the person who decides not to pay
over collected taxes. If the person has employed corporate form, then the collection against that individual will take place through 6672 with no interest accruing
78
Fogg
and no penalties until the point of assessment. If, however, the person does business as a sole proprietorship or a single member LLC treated as a disregarded
entity, then that person is liable for interest on the unpaid employment taxes from
the due date of the return.
Looking at the original example used in this article can provide some insight
into this problem:
Assume that Bob, Mary and John ran ABC as a partnership in which
they were general partners rather than a corporation. If ABC fails
to pay $50,000 in withheld income taxes for the first quarter of
2008 over to the government by April 30, 2008, the due date of the
employment tax return for the first quarter, interest will run from
April 30, 2008. Each of them as general partners is liable for the
full amount of the unpaid withheld income taxes ($50,000) plus
interest from April 30, 2008. If Bob ran ABC as a sole proprietorship
instead he would also be liable for the full amount of the tax plus
interest from April 30, 2008. If, however, ABC were incorporated,
interest would not run against Bob, Mary and John as responsible
officers until the 6672 was made against them. Assuming 5
percent simple interest and an assessment on April 30, 2010, two
years after the return due date, the savings in interest would be
approximately $5,000.
Senator Ervin in his floor statements concerning the amendment of the
Bankruptcy Act to include section 17a (1) (e) and the majority of the Supreme Court
in Sotelo both expressed concerns about the equality of treatment persons liable
for collected taxes who had incorporated versus those who ran their businesses as
a sole proprietorship. The two parties addressed the subject from the perspective
of the discharge in bankruptcy at issue in Sotelo; however, the reasoning could
apply to the difference created with respect to the running of interest. The Court
in Sotelo quoted Senator Ervin’s statements made during the consideration of the
amendment to the Bankruptcy Act of section 17a (1) (e) “The inequity between a
corporate officer and an individual entrepreneur, both of whom have a similar liability to the Government, frequently would turn on nothing more than whether
the individual was ‘sophisticated’ enough ‘to, in effect, incorporate himself.’ ” 461
Justice Marshall, writing for the majority in Sotelo, expressed similar concerns
of fairness as a basis for the majority’s decision.
“The dissenting opinion recognizes Congress’ unquestioned
concern about eliminating corporations’ ‘unfair’ advantage over
individual entrepreneurs. Elsewhere our Brother Rehnquist appears
to concede that Congress meant ‘to ameliorate the lot’ of only ‘some
bankrupts’ when it passed the 1966 amendment to the Bankruptcy
Collecting Collected Taxes
79
Act. There is every indication that the 1966 amendment was not
intended ‘to ameliorate the lot’ of corporations and their principal
officers, at least with regard to taxes collected from employees. And
the dissenting opinion has not even attempted to explain how a
Congress concerned about ‘[discrimination] against the private
individual or the unincorporated small businessman,’ could have
thought it just to relieve corporate officers of § 6672 liability in
bankruptcy, as the dissent’s approach would do, while leaving
other owners of ‘small family [businesses],’ those who happen
to operate through non corporate entities—subject to the same
kind of liability.” 462
While slightly different in nature, the same type of disparity that concerned
Senator Ervin and Justice Marshall still exists in the application of 6672 because of
the manner in which interest and penalties are treated. Those who fail to incorporate and fail to pay collected taxes pay the higher price even though the policy
for collecting the tax seems identical in both instances. This disparity prevents a
parallel result between similarly situated taxpayers. This lack of parallelism does
not promote effective tax administration and fails to achieve a sense of fairness
desired in tax legislation.463 The disparity also fosters the wrong incentive to promote prompt payment and compliance.464
The Misalignment with Similar Statutes
Section 6672 provides a mechanism for holding third parties liable for the payment of a tax for which a corporate entity has primary liability. The derivative
nature of the liability imposed by 6672 creates an exception to the normal rule
for liability but not a unique situation. Other statutes also create derivative tax
liabilities for third parties not primarily liable for the tax. The executor provision
of 31 U.S.C. 3713, the transferee liability provisions of 6901, and the lender liability
provisions of 3505 each provide a parallel situation to 6672. These statutes offer
another view regarding the accrual of interest against third parties. Of these three,
3505 deserves the most attention since it sprang from perceived inadequacies in
6672.
The so-called insolvency statute found in 31 U.S.C. 3713 applies to situations
broader than just tax. Essentially, it holds someone like an executor personally
liable for the payment of the taxes of an estate when the executor distributes assets
to beneficiaries or pays out lower priority creditors without satisfying the taxes of
the estate. The person liable under the insolvency statute must pay interest (and
penalties) on their personal liability stemming from misapplication of estate assets to the extent that the value of the assets distributed exceeds the amount of the
outstanding liability.465
80
Fogg
A transferee under 6901 also must pay interest depending on the value of the
property transferred. The extent of the liability is the transferor’s unpaid taxes
(including interest and penalties) for the transfer year and prior years to the extent
of the value of the assets plus interest.466 Whether a transferee is liable to the full
extent of the transferor generally depends on the value of the asset(s) transferred
together with the amount of the liability at issue. Where the transferred assets
exceed in value the amount of the liability, the transferee will generally be liable
for interest and penalties on the transferor’s taxes.467 If the transferred assets are
less than the amount of the liability, the transferee’s liability is generally capped
at the value of the assets received with the possibility that under state law interest
might accrue on the value of the assets received.468 If a notice of transferee liability
is sent, the transferee is liable for interest on the amount assessed pursuant to that
notice. The interest runs from the date of the notice of transferee liability.469
Someone tagged with liability under the insolvency or the transferee statutes
must pay interest back to the due date of the return of the person primarily liable.
This general rule is tempered in some situations by the amount of assets the third
party received vis-a-vis the amount of the total tax liability. If the value of assets in
the estate or the value of assets transferred are below the amount of the primary liability, the liability of the third party is capped at the value of the assets. 6672 does
not have a direct parallel to this provision limiting interest; however, the manner
in which the taxpayer whose tax is collected receives full credit for that payment
provides a basis for viewing the 6672 situation as one in which the corporation
and the responsible officer received assets equal in value to the unpaid liability.470
The policies leading to imposition of interest against these third parties support
the imposition of interest back to the due date of the underlying return for those
persons responsible under 6672.
The most significant of the three derivative liability provisions with respect to the
treatment of interest is 3505 because it developed out of a loophole in 6672 and deals
with a subset of the same liability that 6672 does.
In the 1950s and early 1960s, lending practices in the construction industry exposed a gap in the coverage of 6672 with respect to income and social security taxes
withheld from employees of troubled businesses in that employment sector. In closing that gap, Congress created a new statute that specifically provides that the third
party liable under the new statute has liability for the interest from the due date of
the return of the party primarily liable for the unpaid tax. While the legislative history of the new statute does not provide any insight as to why the interest provision
appears in this new statute (and not in 6672), the adoption of the interest provision
in 3505 supports the imposition of interest for all collected tax situations.
Section 3505 was enacted as part of the Federal Tax Lien Act of 1966 at the request
of the Government to plug what it perceived to be a loophole in the collection of
withheld employment taxes.471 The specific loophole 3505 sought to address concerned third parties who paid, either directly or indirectly, the wages for an employer
Collecting Collected Taxes
81
in such a manner that the withheld employment taxes did not get paid over to the
Government.472 Section 3505 does not address all types of collected taxes. It imposes liability on lenders, sureties, and others who lend net payroll in a manner that
causes a failure to pay over to the Government withheld federal income tax, FICA
tax and railroad retirement tax. The statute has two components: 3505 (a) imposes
liability for the full amount of the unpaid tax on third parties who pay net wages
directly to the employees, and 3505 (b) imposes a limited personal liability on third
parties who provide the funds used to make net payroll payments of no more than
25 percent of the unpaid employment taxes.473 As mentioned above, the collection
problem the IRS primarily sought to address through this legislation involved the
construction industry.474
Prior to the enactment of 3505, the IRS lost several cases in which it attempted to
assert the 6672 penalty against the type of third parties described in 3505.475 Courts
determined that such individuals were willful but not responsible. At the same time,
the alleged responsible persons would win their cases under 6672 because they were
responsible but not willful. These responsible persons would testify, usually quite
correctly, that the lender would not permit them to pay the employment taxes even
though they had tried to do so. Section 3505 filled the gap caused by the circumstances of the lender who essentially controlled the finances of a cash-poor entity but
whose role did not neatly fit the statutory scheme of 6672.
Two common situations occurred that posed problems for the IRS in attaching the 6672 penalty where employment taxes were not paid. The first, and perhaps most common, scenario involved companies with cash flow problems. These
companies would negotiate a line of credit with a bank. As business worsened
the bank’s involvement increased. At some point in the relationship, a loan officer at the bank essentially took over the duty of approving every check written by
the company. The loan officer then made decisions to pay employees their wages
but also refused to allow the company to write the employment tax check to the
IRS. The company would eventually fail and the IRS would come looking for the
employment taxes.
The second variation on this theme usually occurred in the construction industry.476 A general contractor would hire a subcontractor to complete a specific task
on a larger job. The subcontractor would encounter financial difficulty. The general contractor needed the subcontractor to complete the task for which it had been
hired or the entire project would fall behind with all of the attendant consequences.
The employees of the subcontractor who were not getting paid would refuse to work
without pay. So, the general contractor or its surety would step into the breach and
pay the net wages of the employees of the subcontractor. At some point the subcontractor would fail before the employment taxes were paid.
As mentioned above, 3505 has two components which attempt to address problems presented by both direct payment of net payroll by a third party and indirect
payment of net payroll. Section 3505 (a) holds a person liable in an amount equal
82
Fogg
to the entire amount of the payroll taxes required to be withheld and paid over in
those situations in which the third party directly pays the wages of the employees
of the company that fails to pay its employment taxes. The 3505 (a) liability arises
upon the payment of the wages whether or not the third party knows taxes should be
paid or withheld.477
Section 3505 (a) prevents third parties from taking over a company’s payroll and paying net wages. The third party becomes liable not only for taxes on
the wages from the date the wages are paid but the third party is liable also for
interest back to the due date of the return. Imposing liability on a third party in
this situation was viewed as “fair” because the third party knows the finances of
the employer.478
The liability under 3505 (a) is not imposed by way of assessment as with 6672.
In order to hold a third party liable under 3505 the Government must bring a
suit against the third party. The statute of limitations for the suit is the statute of
limitations on collection of the underlying liability.479 The Government bears the
burden of proof in the litigation to show that the third party directly made net
payroll payments to the employer.
The existence of 3505 (a) has undoubtedly caused lenders to change their practices to avoid this pitfall.480 Very few 3505 (a) cases exist.481 A lender directly paying net payroll has little room to hide. This provision serves an important role in
prevention but receives little enforcement attention because of the straightforward
and predictable outcomes it creates.
Section 3505 (b) does not impose the broad liability set out in subparagraph
(a). Nor does 3505 (b) involve the relatively easy to identify direct payment of
net wages. Instead, 3505 (b) concerns the actions of those who provide funds to
the employer knowing that the funds will be used to meet payroll and that the
employment taxes will not be paid.
As with 3505 (a) the liability under subsection (b) does not occur through an
administrative assessment but rather the Government must bring a suit to establish the liability. The Government has the burden of proof in the suit. The liability
under 3505 (b) has a limitation of 25 percent of the amounts paid to the employer
for the purpose of making net wage payments. The statute provides for interest back to the due date of the return; however, case law has limited the amount
of interest recoverable by treating it as a part of the 25 percent cap and not an
addition to that amount.482
A lender has knowledge for purposes of 3505 (b) from (1) the time the lender receives notice of this fact or (2) the time the lender would have known if exercising
due diligence.483, 484 Because the liability under 3505 (b) requires the Government
to show this knowledge, the Government can experience difficulty establishing
this liability. The Government does not need to prove, however, that a formal loan
agreement existed. Honoring overdrafts over a period of time can also trigger
this liability.485
Collecting Collected Taxes
83
An exception to 3505 (b) liability for lending for net payroll occurs for working
capital loans. Perhaps the exception need not have been explicitly stated in the
statute because of the knowledge provision of the statute, nonetheless it exists as
a stated exception. Lenders must take care when making working capital loans if
they learn that the loans finance net payroll.486 Likewise, lenders pursuing remedies upon default of a loan do not enter 3505 (b) territory unless they become too
entwined in the business of the distressed entity.487
Because the liability under 3505 is not considered a tax liability, the interest
component referred to in the statutes does not represent interest on the employment taxes themselves.488 The liability of the lender is for a sum equal to the unpaid trust fund portion of the employment taxes rather than the taxes due from
the employer.489 Depending on the type of 3505 liability, the third party may have
no interest liability because of the interpretation of the 25 percent cap in 3505 (b).
Nonetheless, the statute does contemplate in general that the third party engaged
in the actions described by 3505 is liable for interest on the delinquent employment
taxes. Two examples demonstrate how the interest component of 3505 works:
Example 1: ABC Inc. experiences financial difficulty during 2007
and XYZ Bank becomes increasingly involved in its finances. It
looks like ABC might get a contract that will pull it out of the
tailspin but it is totally out of gas pending the award of that
contact. It must keep the business going, however, to remain
competitive. During the first quarter of 2008 XYZ Bank directly
pays the payroll of ABC. Neither the bank nor the company pays
the withholding taxes of $25 over to the IRS. ABC dissolves
without making any payments on the employment tax liability
for the first quarter of 2008. The IRS cannot pursue Bob, Mary
and John because XYZ bank had assumed financial control of
ABC making Bob, Mary and John either not responsible or not
willful or both. The IRS pursues XYZ pursuant to 3505 (a) and
obtains a judgment for $25 for the full amount of the unpaid
withholding taxes plus interest from April 30, 2008, the due date
of the employment tax return.
Example 2: ABC Inc. experiences financial difficulty in 2007
and XYZ Bank becomes increasingly involved as in Example
1. During the first quarter of 2008 XYZ lends to ABC $80 so
that ABC can pay net payroll. XYZ knows the finances of ABC
and knows that ABC does not have sufficient funds to pay over
the withholding taxes. ABC dissolves without making any
payment on the employment tax liability for the first quarter
of 2008. The withholding tax obligation of ABC for the first
quarter of 2008 is $25. Again, the IRS would probably fail if
84
Fogg
it pursued a 6672 liability against Bob, Mary or John. The IRS
pursues XYZ pursuant to 3505 (b) and obtains a judgment for
$20 equal to 25 percent of the net payroll lending. The IRS
cannot obtain interest on this amount because it is limited to a 25
percent recovery.490
One case that highlights the differences between 6672 and 3505 and explores the reach of the term “responsible person” in 6672 is Pacific National
Insurance, Co., v. United States.491 Pacific National, a surety, loaned money to
Central States Construction and Equipment Company (Central) from May to
September, 1955. This time period predated the enactment of 3505 even though
the 9th Circuit’s opinion followed the passage. Because of the period in issue,
the IRS had to argue for the assessment against Pacific National under 6672.
Naturally, Pacific National argued that 3505 applied to its circumstances, albeit
not literally since it did not exist in 1955, and 6672 did not reach the situation
presented by this case.
The court examined the legislative history and found little aid in determining
the scope of persons who were included. It determined that Pacific National met
the language of responsible person set forth in 6672 and sustained the decision
of the District court holding Pacific National liable for taxes under 6672. The
case points out that 3505 and 6672 overlap. In fact, the Government recommends looking to hold parties liable under both provisions when possible.492
The result in Pacific National, displaying the overlap between 3505 and 6672
on collected employment taxes, points out the possibility that the Government
could obtain interest from a responsible party back to the due date of the return
by pursuing one statute, 3505, while a parallel result remains unavailable if it
pursues the responsible officer under 6672.493
Did Congress intend to provide interest on collected taxes back to the due
date of the return only for that narrow class of collected taxes represented by
“net lenders” of wages? Does this class of responsible officers have some special
responsibility not borne by all others who convert funds held in trust to some
other purpose? Since 3505 updates 6672 and closes a narrow loophole on one
aspect, could Congress have included interest in 3505 without realizing that interest back to the due date of the return did not apply in other responsible officer
situations? Answers to these questions do not exist in the legislative history of
3505. One possible answer, the answer that Congress simply did not think about
the lack of symmetry on the interest issue between the two statutes imposing
liability on responsible persons, provides support for seeking symmetry now to
close the gap between the two statutes. While 3505 came into existence to close
one loophole in 6672, Congress inadvertently exposed a fundamental flaw in
6672. The fixing of that flaw requires imposing interest on the responsible officers back to the due date of the return.
Collecting Collected Taxes
85
The Problems Created for the IRS
The first problem that the failure to charge interest under 6672 creates for the IRS is the
loss of the time value of money. As the example with ABC shows, the amount of interest that runs between the time the employment tax return is due and an assessment
occurs with respect to one or more responsible officers can be significant. When that
example multiples across a system, the lost revenue begins to mount. Of course, the
IRS will not collect all of the interest that runs on its 6672 assessments but it certainly
would collect some of the money were it allowed to charge interest back to the due date
of the return.
The second problem for the IRS concerns resources. Many of the individuals identified by the IRS as responsible officers subject to the 6672 liability know that they
satisfy the statutory tests for liability under 6672. These individuals, who know they
are liable, can agree to that liability at the first moment the IRS revenue officer appears seeking to investigate the liability; however, they have no incentive to do so. The
minute they agree to the liability, an assessment will occur and interest will start to
run. Consequently, the system provides an incentive for even the individuals who
know they owe to exhaust their administrative remedies.494 This places a burden
on IRS resources that might significantly diminish if liable individuals lost their
incentive to delay.
The third problem created for the IRS results from the approach that most states
have taken with respect to individuals liable for collected taxes not paid by the corporation that had primary responsibility. The significant majority of states have adopted
an approach similar to the one suggested for the federal government in this article.
These states hit individual responsible officers with the same liability, including interest and penalties, that are due from the corporation. They do not build in a period of
delay for interest and penalties. Consequently, knowledgeable individuals faced with
responsible officer liability to both the state and the federal government, which is often
the case, will pay their money first to the states to stop the running of further penalties
and interest while they continue to exhaust their administrative remedies with the IRS.
This situation obviously puts the federal government at a competitive disadvantage in
seeking to collect from these individuals.
All states with the exception of Wyoming have responsible officer statutes that
work with many similarities to section 6672. Most state statutes draw directly from
6672 and, even if indirectly, certainly draw from the same policy framework that
drove the creation of 6672. Despite their similar origins, the overwhelming majority of states have specifically written into their responsible officer statutes or have
judicially interpreted their statutes in such a manner that their responsible officers
are charged interest from the due date of the underlying return (and penalties as
well). The manner in which the states have chosen to treat interest with respect to
individuals responsible for paying over collected taxes supports the recommendation of this article.
86
Fogg
While not controlling, viewing the manner in which states treat their delinquent
trust fund obligations vis-a-vis responsible officers provides some insight from which
the federal government can draw. As discussed previously, IRC 6672 covers a variety
of taxes.495 Employment taxes offer the best known example of trust fund taxes in
the federal system under IRC 6672 but the excise tax on telephone service which is
collected by the phone company on behalf of the federal government actually touches
more people.496 Because the number of telephone companies is relatively small and
they typically do not have financial difficulties at the same rate as “regular” businesses,
this particular trust fund tax has received little attention. The excise tax on telephone
service, like the one on airplane tickets and motor fuel, behaves much like a sales or
use tax common in state taxing schemes.497–499 So, in looking at states for comparative
purposes, both state withholding and sales tax provisions must be analyzed.
Because some states have no income tax and some states have no sales tax, there
exists at times only one type of tax for comparison within a specific state.500, 501
Surprisingly, some states with both sales and income taxes treat failures with respect to the payment of each of these taxes differently when imposing the liabilities on individuals responsible for the trust fund taxes.502 Those differences
merit further exploration because within those states exist two models for trust
fund treatment.
The vast majority of states with trust fund tax regimes choose to impose upon
the individual trustees (responsible officers) the precise liability imposed upon
the entity that failed to meet its trust fund obligation.503, 504 Stated another way,
these states have adopted, with respect to interest, the same result advocated herein. These states also uniformly impose penalties on unpaid trust fund taxes against
the responsible officers going back to the due date of the entity’s return.505 The
combination of imposing the penalty and interest due from the entity against the
responsible persons creates a significant additional liability against these individuals compared to the liability imposed using the current federal regime under
IRC 6672.506 This additional liability for interest and penalty charged to the responsible officer could, if collected, reduce the tax gap; however, the stronger reason for imposing interest, as argued herein, is the removal of the incentive to delay
the assessment.507, 508
Some states explicitly provide for interest in the flush language of their statutes creating responsibility.509 Other states have reached the same result by judicial decision.510
States reaching the result by judicial decision with the courts referencing the fact that
the responsible officer liability is an alternative means of collecting the trust fund tax is
yet another model.511 Based upon that reason for the liability of the responsible officer,
the courts conclude the officer is liable for everything for which the entity is liable.512
A minority of states treat interest in the same manner as the federal government.513
These states have adopted statutes imposing the responsible officer liability that essentially mirror the language of IRC 6672.514 In interpreting their statutes, these states follow the federal interpretation and do not impose liability until the assessment against
the responsible officer occurs.515
Collecting Collected Taxes
87
A still smaller minority of states chooses to impose a larger liability against the individuals liable for the trust fund liability than simply the amount of the unpaid trust
fund taxes.516 These states impose the liability by means of a penalty. In Colorado
the trust fund penalty is 150 percent of the collected tax.517 In Florida the trust fund
penalty is 200 percent of the collected tax.518 Penalties at these levels cause the responsible officer to have a liability that reflects something close to the liability imposed by
those states hitting the taxpayer with the interest and penalty imposed on the entity.
By adopting a scheme that imposes a penalty for late payment of collected taxes rather
than one which simply seeks to collect the unpaid tax, plus interest, these states are at
a disadvantage in bankruptcy proceedings.519
The state provisions for treating interest on the liability for collected taxes imposed
upon responsible officers provide an interesting window from which to view the federal statutes. Most states impose liability upon individuals because they build on the
federal model. Yet, almost all states go past the federal model to cause their statutes
to work in a manner that, with respect to interest, is philosophically consistent with
the underlying reason for the statute.520 The most interesting states are the states that
“split the baby.” 521 These states charge responsible officers with interest back to the due
date of the entity’s liability for sales taxes but charge interest only from the assessment
date of the responsible officer’s liability for unpaid withholding taxes. This division in
approach, adopted by a small but diverse set of states, suggests a statutory scheme built
upon placement rather than consistent philosophy similar to the federal model. These
states all have different language in different sections of their codes for dealing with
persons responsible for collecting sales taxes versus persons responsible for withholding employment taxes. In both circumstances the money is to be collected and held
by the employer for the state yet the code sections, adopted at different times for the
different specific purposes, fail to take into account the essentially identical trust fund
situation created in each situation.
The fact that the overwhelming majority of states choose to impose interest on responsible officers from the due date of the return of the entity suggests that the states
see the direct link between the liability of the responsible officers and the liability of the
entity. The state statutes reaching this result contain diverse language. The position
adopted by a majority of the states has happened without the apparent benefit of any
model other than 6672 itself. The laws adopted by the majority of the states represent a
significant expression of how the derivative liability imposed upon responsible officers
should be structured with respect to interest and validates the legal reasons expressed
herein for modifying the federal statute.
Conclusion of Accountability
The current system using business entities to collect taxes and turn them over to
the IRS works efficiently and effectively for the vast majority of business entities.
88
Fogg
The business entities that fail to pay over collected taxes are almost always relatively young, small businesses run by entrepreneurs who “need” the collected
taxes for operating capital and who may not appreciate the special nature of the
collected tax funds which they hold. By better informing the public of the practices of those handling collected taxes, better informing the business entities of
their responsibilities, creating more structure and incentives for the payment of
these collected taxes, this corner of the tax gap can be reduced.
Overall Conclusion
Collecting collected taxes requires a thoughtful plan that does not currently exist.
The need for transparency, structure and accountability in the related provisions
provides an opportunity for both the IRS and Congress to step in to create a system that will significantly reduce the current tax gap in this area.
Appendices
Appendix A—States with Shaming Laws and Their
Websites (as of August 2010)
State
Statute
Alabama
Ala. Code § 40-5-23 (LexisNexis 2010)
The tax collector must publish twice during the month of July a list of delinquent taxpayers. The publication shall be made in a daily newspaper
printed and published in the county in which the taxpayer lives. If no such
paper is published, a weekly paper will suffice. If there is neither a daily nor
a weekly newspaper of any sort published in said county, the tax collector
shall publish the list in the courthouse and in other conspicuous places in
said county. The tax collector must keep said posting available for the public
during the entire month of July.
Alaska
Alaska does not have a shaming statute.
Arizona
Arizona does not have a shaming statute.
Arkansas
Ark. Code Ann. § 26-36-203 (2010)
No later than December 1 of each year, the county tax collector shall prepare a list of delinquent personal property taxes and deliver a copy of the
list to a legal newspaper in the county. The newspaper shall publish the list
within seven days. The list must be in at least seven-point font. The list shall
show the name of the taxpayer, the taxpayer’s school district, and the total
amount of taxes delinquent.
Collecting Collected Taxes
California
89
Cal. Rev. & Tax. Code § 19195 (Deering 2010)
The Franchise Tax Board shall make available as a matter of public record each calendar year a list of the 250 largest delinquencies in excess of
$100,000 as of December 31 of the preceding year.
Colorado
Colo. Rev. Stat. § 24-35-117 (2010)
The executive director of the department of revenue shall annually disclose
a list of all taxpayers delinquent in the payment of tax liabilities collected
by the department. The list shall include only those taxpayers with total
delinquent final liabilities for all taxes collected by the department in an
amount greater than $20,000 for a period of six months from the time that
a distraint warrant issues or may issue. The list shall contain the name,
address, types of taxes, month and year in which each tax liability was assessed, the amount of each tax outstanding of each delinquent taxpayer,
and, in the case of a corporate taxpayer, the name of the current president of
the corporation.
Connecticut
Conn. Gen. Stat. § 12-7a (2010)
The Commissioner of Revenue Services shall prepare and maintain a list
related to each type of tax levied by the state, containing the name and address of any person or corporation liable for payment of any such tax and
the amount thereof which tax is unpaid and a period in excess of ninety days
has elapsed following the date on which such tax was due. Such lists shall
be available to the public for inspection by any person.
Delaware
Del. Code Ann. tit. 30, § 359(b) (2010)
The Secretary of Finance shall prepare, maintain, and publish on the Division of Revenue Internet Website, two lists of taxpayers owing unpaid tax
and additions to tax finally determined to be due under Title 30 for personal
income tax and business taxes administered by the Department of Finance.
Each list shall consist of the 100 taxpayers owing to Delaware the greatest
amount of unpaid tax and shall contain the name and address of each such
taxpayer, the total type and amount of tax and additions to tax due and the
date the amount was finally determined to be due. In the case of entities
other than natural persons, the list may also name any persons who were at
least 25% owners or beneficial owners or who were responsible officers of
such entity at or after the time the liability was created.
District of Columbia
District of Columbia Office of Tax and Revenue, Delinquent Taxpayers,
http://otr.cfo.dc.gov/otr/cwp/view,a,1330,q,593715,otrNav_gid,1679,|33288|
.asp (last visited August 9, 2010).
The District of Columbia publishes a list of its delinquent taxpayers as part of
an overall program to encourage voluntary compliance with the District’s tax
laws. The list contains the taxpayer’s name, address, and amount owed. In
the case of a business, the responsible officer and his/her address is listed.
Florida
Georgia
Florida does not have a shaming statute.
Ga. Code Ann. § 48-3-29 (2010)
The commissioner may publish in the media or on the internet for public
access any or all information with respect to executions issued for the collection of any tax, fee, license, penalty, interest, or collection costs due the
state which are recorded on the public records of any county.
90
Fogg
Hawaii
Haw. Rev. Stat. Ann. § 231-32 (LexisNexis 2010)
Hawaii Department of Taxation, List of Delinquent Taxpayers With Large
Balances, http://www6.hawaii.gov/tax/a2_b2_2delinq.htm (last visited August 9, 2010).
The department of taxation shall prepare and maintain, open to public inspection, a complete record of the amounts of taxes assessed in each district that have become delinquent with the name of the delinquent taxpayer
in each case. This list may be published on the Internet after taxpayers have
had a final opportunity to settle their debt.
Idaho
Illinois
Idaho does not have a shaming statute.
20 Ill. Comp. Stat. Ann. 2505/2505-425 (LexisNexis 2010)
State of Illinois Department of Revenue Public List of Delinquent Taxpayers,
http://www.revenue.state.il.us/AboutIdor/DelinquentList.html (last visited
August 9, 2010).
The Director may annually disclose a list of all taxpayers that are delinquent
in the payment of tax liabilities collected by the Department. The list shall include only those taxpayers with total final liabilities for all taxes collected by
the Department in an amount greater than $1,000 for a period of six months
from the time that the taxes were assessed. The list shall contain the name,
address, types of taxes, month and year in which each tax liability was assessed, the amount of each tax outstanding of each delinquent taxpayer,
and, in the case of a corporate taxpayer, the name of the current president.
Illinois is in the process of creating a website for publication of this list.
Indiana
Ind. Code Ann. § 6-8.1-3-16 (LexisNexis 2010)
The Department shall compile each month a list of the taxpayers subject to
tax warrants that were issued at least twenty-four months before the date of
the list and are for amounts that exceed $1,000. The list must identify each
taxpayer liable for a warrant by name, address, and amount of tax. The department shall publish the list on accessIndiana and make the list available
for public inspection and copying. The department may not publish a list that
identifies a particular taxpayer unless at least two weeks before the publication of the list the department sends notice to the taxpayer.
Iowa
Kansas
Kentucky
Iowa does not have a shaming statute.
Kansas does not have a shaming statute.
Ky. Rev. Stat. Ann. § 131.650 (LexisNexis 2010)
The department may publish a list or lists of taxpayers that owe delinquent
taxes of fees administered by the Department of Revenue. A taxpayer may
be included on the list if the taxes owed remain unpaid at least forty-five
days after the dates they became due and payable and a tax lien or judgment has been filed of public record against the taxpayer. If the listed taxpayers are business entities, the Department of Revenue may also list the
names of responsible persons assessed. Notice must be given to the affected taxpayers before any list is published.
Collecting Collected Taxes
Louisiana
91
La. Rev. Stat. Ann. § 47:1508 (2010)
The secretary may disclose the name and address of the taxpayer, the type
of delinquent taxes due, and the total amount of tax, penalty, and interest
due. If the taxpayer is a business entity, the secretary may additionally
name any owner who owns at least a fifty percent ownership interest in the
entity. The disclosure may be made in a newspaper, magazine, or in electronic media, such as television or the Internet. The secretary must provide
written notice by registered mail to the taxpayer.
Maine
Maryland
Maine does not have a shaming statute.
Comptroller of Maryland Caught In the Web, http://compnet.comp.state.
md.us/Compliance_Division/Collections/General_Collections_Information/
Caught_in_the_Web.shtml (last visited August 9, 2010)
Maryland publishes the names of businesses, individuals and corporate officers having large unresolved liabilities (including individuals who have large
unresolved personal income tax liabilities). All of the information is public,
because liens and judgments have been recorded in the judgment dockets
of one or more circuit courts of Maryland.
Massachusetts
Mass. Ann. Laws ch. 62C, § 21(b)(11) (LexisNexis 2010)
Massachusetts Department of Revenue, Public Disclosure, https://wfb.dor.
state.ma.us/dorcommon/PublicDisclosure/disclosure.aspx (last visited August 9, 2010).
Massachusetts allows disclosure by the commissioner of a list of all taxpayers that are delinquent in the payment of their tax liabilities in an amount
greater than $25,000 for a period of six months from the time the taxes were
assessed. The list shall contain the names, address, types of taxes, month
and year assessed, and amounts outstanding of said delinquent taxpayer.
Massachusetts publishes this list online.
Michigan
Michigan does not have a shaming statute.
Minnesota
Minnesota no longer has a shaming statute.
Mississippi
Mississippi does not have a shaming statute.
Missouri
Email from Kathy Mantle, Collections and Tax Assistance, State of Missouri,
to Fleming Ware, Research Assistant, Villanova University School of Law
(July 8, 2009, 14:11 EST) (on file with author).
Missouri publishes a list of businesses that have had their sales licenses
revoked for failure to remit sales tax, but does not publish a list of the state’s
largest delinquent taxpayers.
Montana
Email from Russ Hyatt, Accounts Receivable and Collections Bureau, Business and Income Tax Division, State of Montana, to Fleming Ware, Research Assistant, Villanova University School of Law (July 8, 2009, 14:09
EST) (on file with author); Mont. Code Ann. §§ 3-5-508-09 (2010).
The Montana Department of Revenue publishes a list of the state’s delinquent taxpayers. The list includes only taxpayer’s names for tax debts that
Montana has filed a warrant for distraint against them for the tax debt they
owe. Authority is derived from cited statute.
Nebraska
Nebraska does not have a shaming statute.
92
Fogg
Nevada
Nev. Rev. Stat. Ann. § 361.300 (LexisNexis 2010)
On or before January 1 of each year, the county assessor shall transmit
to the county clerk, post at the front door of the courthouse and publish in
a newspaper published in the county a notice that the tax roll is complete
and open for public inspection. Additionally, the list may be posted in public
areas of public libraries, in public areas of courthouses, and on a website.
New Hampshire
New Jersey
New Hampshire does not have a shaming statute.
Email from New Jersey Taxation, to Fleming Ware, Research Assistant,
Villanova University School of Law (July 8, 2009, 13:00 EST) (on file with
author); New Jersey Division of Taxation’s Largest Judgmented Taxpayer
Listing, http://www.state.nj.us/treasury/taxation/jdgdiscl.shtml (last visited
August 9, 2010).
New Jersey publishes a list of delinquent taxpayers; however, the website is
currently under construction.
New Mexico
New York
North Carolina
New Mexico no longer has a shaming website.
New York does not have a shaming statute.
North Carolina Tax Debtors, http://www.dor.state.nc.us/collect/debtor_info.
html (last visited August 9, 2010).
North Carolina publishes a list of delinquent taxpayers names, the type of
tax owed, and the amount of the tax.
North Dakota
Ohio
North Dakota does not have a shaming statute.
Ohio Rev. Code Ann. § 5719.04 (LexisNexis 2010)
Ohio prepares a tax list containing the name of the person charged and the
amount of such taxes and the penalty. The auditor shall cause a copy of
the delinquent personal and classified property tax list to be published twice
within sixty days in a newspaper published in the English language in the
county and of general circulation thereof.
Oklahoma
Email from Tim Rudek, Oklahoma Tax Division - Account Maintenance Division, to Fleming Ware, Research Assistant, Villanova University School of
Law (July 13, 2009, 08:25) (on file with author). Oklahoma Tax Commission,
http://www.tax.ok.gov/top100.html (last visited August 9, 2010).
Oklahoma publishes a hard list of delinquent taxpayers owing taxes for
which a warrant has been issued.
Oregon
Oregon does not have a shaming statute.
Pennsylvania
Pennsylvania no longer has a shaming website.
Rhode Island
R.I. Gen. Laws § 44-1-34 (2010); Rhode Island Division of Taxation, Top 100
Tax Delinquents, http://www.tax.ri.gov/misc/top100.php (last visited August
9, 2010).
The tax administrator is authorized by statute to prepare a list of names of
the 100 delinquent taxpayers who owe the largest amount of state tax and
whose taxes have been unpaid for a period in excess of ninety days following the date their tax was due.
Collecting Collected Taxes
South Carolina
93
South Carolina’s Debtor’s Corner, http://www.sctax.org/delinquent/
delinquent.shtml (last visited August 9, 2010).
The South Carolina Department of Revenue publishes information pertaining to some of the largest uncollected liabilities owed to the citizens of South
Carolina. All of the information provided on the list is public information as a
result of the Department of Revenue’s having filed a tax lien with the Clerk
of Court/Register of Deeds in the county of residence. Debt information may
also be obtained directly for the Department of Revenue. The list includes
the name of the taxpayer, the taxpayer’s address, and the amount owed.
South Dakota
Tennessee
South Dakota does not have a shaming statute.
Tennessee does not have a shaming statute.
Texas
Texas does not have a shaming statute.
Utah
Utah does not have a shaming statute.
Vermont
Virginia
Vermont does not have a shaming statute.
Email from VA Tax Customer Service, to Fleming Ware, Research Assistant,
Villanova University School of Law (July 9, 2009, 09:19 EST) (on file with
author); Virginia Delinquent Taxpayer List, http://www.tax.virginia.gov/site.
cfm?alias=delinquentdebtors (last visited August 9, 2010).
Virginia publishes the names of businesses having unresolved tax liabilities. The list includes the name of the business, address, and amount of
tax owed. The information contained in the list is public information as a
Memorandum of Lien has been filed on the debts listed in the Circuit Court.
Washington
Wash. Rev. Code Ann. § 82.32.330(3)(c) (LexisNexis 2010)
Washington may publish the names of taxpayers against whom a warrant
has been either issued or filed and remains outstanding for a period of at
least ten working days.
West Virginia
Wisconsin
West Virginia does not have a shaming statute.
Wis. Stat. § 73.03(62) (2010).
It shall be the duty of the department of revenue, and it shall have the power
and authority to prepare and maintain a list of all persons who owe delinquent taxes to the department, in excess of $5,000, which are unpaid for
more than ninety days after all appeal rights have expired. The department
shall post the names of persons from this list on the internet at a site that
is created and maintained by the department for this purpose. The department shall distribute the posted information to Internet search engines so
the information is searchable. The Internet site shall list the name, address,
type of tax due, and amount of tax due, and the Internet site shall contain
a special a special page for the 100 largest delinquent taxpayer accounts.
Wyoming
Wyoming does not have a shaming statute.
94
Fogg
Appendix B —States Requiring Identification of
Responsible Parties
ALABAMA
Alabama Depart. of Revenue CombinedRegistration/Application, http://www.excelpay.com/files/file/tax%20forms/ALCom101%20–%20CombinedRegistration
Application.pdf; Alabama Department of Revenue Starting a New Business
Educational Brochure, www.revenue.alabama.gov/taxpayerassist/newbus.pdf.
Alabama requires identification of current owners, partners, corporate officers,
and employers. The Alabama Starting A New Business brochure discusses trust
fund taxes; it clearly states that if owners, partners, or corporate officers occupy
positions within a business in which they have authority and control over the payment of creditors, and they choose to pay other creditors over the government, the
owner, partner, or corporate officer can be held personally liable for the tax. The
brochure further states that bankruptcy cannot waive this debt.
ALASKA
Alaska Admin. Code tit. 8, § 85.020 (2009); Alaska Employer Registration Form
(2009), www.payroll.com/support/PDFs/State/AK_erf.pdf; Alaska Employer
Resource Manual (2009), www.jobs.state.ak.us/handbook/aerm.pdf; Alaska First
Time Filers Form (2009), labor.state.ak.us/estax/forms/2009FirstTimeFilerForm.pdf.
Alaska has no income or sales tax, but employment security tax, administered
by the Department of Labor, requires withholding and remittance by employers.
The Department of Labor requires disclosure of corporate officers and there is a
duty to notify the Department of changes; however, there is no duty to specifically
identify responsible officers.
ARIZONA
Ariz. Rev. Stat. § 43–414–435 (2009); Arizona Joint Tax Application, www.
revenue.state.az.us/ADOR_Forms/70…/74–4002_fillable.pdf; Arizona With
holding Tax Basics, http://www.azdor.gov/Business/WithholdingTax.aspx (last
visited February 13, 2010).
Arizona requires identification of owners, partners, and corporate officers, but
does not require designation of a responsible party for taxation purposes. Arizona
statute simply states that the employer is liable.
ARKANSAS
26–51–916 Ark. Code R. (Weil 2009); State of Arkansas Withholding Registration,
https://www.ark.org/dfa/withholding/index.php (last visited August 1, 2009).
Arkansas requires the signature of the owner or responsible party when registering to withhold wages.
Collecting Collected Taxes
95
CALIFORNIA
California Seller’s Permit Application (2009), available at www.boe.ca.gov/
pdf/boe400spa.pdf; California Publication 73—Your California Seller’s Permit
(2009), available at www.boe.ca.gov/pdf/pub73.pdf; Registration for Commercial
Employers (2009), available at www.edd.ca.gov/pdf_pub_ctr/de1.pdf.
California requires identification of corporate officers for withheld income taxes. For sales taxes, California requires identification of corporate officers, cautioning that failure to update will cause the listed corporate officers to be personally
liable for unpaid taxes.
COLORADO
Colo. Rev. Stat. § 39–22–604 (2009); Colorado Business Registration, https://
secure.cdle.state.co.us/CR100/ (last visited August 1, 2009); Colorado Income
Withholding for Employers, http://www.colorado.gov/cs/Satellite?blobcol=
urldata&blobheader=application%2Fpdf&blobkey=id&blobtable=MungoBlobs&
blobwhere=1191399221944&ssbinary=true (last visited August 1, 2009).
Colorado requires disclosure of owners, partners, and corporate officers, but
does not require designation of a responsible party.
CONNECTICUT
Conn. Gen. Stat. §§ 12–705–06 (2009); Connecticut Form REG–1 & Instructions
(2009), available at http://www.ct.gov/drs/lib/drs/forms/2006forms/applications/
reg-1instructions.pdf; Business Tax Registration Application (2009), available at
www.ct.gov/drs/cwp/view.asp?a=1433&q=265880.
Connecticut requires disclosure of owners, partners, and corporate officers, but
does not require disclosure of a responsible party.
DELAWARE
30–11–VII Del. Code Regs. § 1151–1156 (Weil 2009); Delaware Combined
Registration Application (2009), available at revenue.delaware.gov/services/
current_bt/cra.pdf; Email from Margaret Boyle, Delaware Division of Revenue, to
Fleming Ware, Research Assistant, Villanova University School of Law (June 23,
2009, 07:02 EST) (on file with author).
Delaware requires disclosure of owners, partners, and corporate officers, but
does not require identification of a responsible party.
96
Fogg
DISTRICT OF COLUMBIA
D.C. Code § 47–4491 (2009); Combined Business Tax Registration Application
(2009), available at https://www.taxpayerservicecenter.com/fr500/.
The District of Columbia requires disclosure of owners, partners, and principal
officers. Statutes state that an officer or director of a corporation, general partner
of a partnership, or similar principal of business shall be liable for a penalty equal
to the tax not paid over to the state.
FLORIDA
Florida Tax Guide (2009), available at www.stateofflorida.com/flortaxguid.html.
Florida does not collect personal income tax.
GEORGIA
Ga. Comp. R. & Regs. 560–7–8–.33 (2009); Georgia State Tax Registration and
Application Instructions (2009), available at https://etax.dor.ga.gov/ctr/formsreg.aspx.
Georgia requires either disclosure of an important person in the business, including that person’s SSN or ITIN, or disclosure of all owners, partners, or corporate officers.
HAWAII
Employer’s Withholding Guide, http://hawaii.gov/ag/csea/main/info_for_
employers/hawaii_empl_guide_iw (last visited August 1, 2009); Hawaii Form
BB–1 and Instructions (2009), available at hawaii.gov/labor/forms/bb1_04.pdf.
Hawaii requires identification of the corporate officers but not responsible
officers.
IDAHO
Idaho Business Registration Form IBR–1 (2009), available at http://tax.idaho.gov/
forms/EFO00147_06–09–2004.pdf.
Idaho requires identification of corporate officers but not responsible officers.
ILLINOIS
Illinois Business Registration Application (2009), available at www.revenue.state.
il.us/taxforms/Reg/REG1.PDF. Illinois’s Business Registration Form requires responsible party information.
Collecting Collected Taxes
97
INDIANA
Ind. Code § 4–1–8–1 (2009); Indiana Business Tax Application (2009), available at
http://www.in.gov/dor/files/bt-1.pdf. Indiana requires identification of corporate
officers but not responsible officers.
IOWA
Iowa Code §§ 701–38.1 (8), 46.3 (422), 46.3 (1) (2009); The Iowa Business Tax
Registration Form requires identification of responsible parties.
KANSAS
Kansas Business Tax Application (2009), available at http://www.ksrevenue.org/
pdf/forms/cr16.pdf; Kansas Withholding Tax Guide (2009), available at www.
ksrevenue.org/pdf/forms/kw100.pdf.
Kansas requires identification of all owners, partners, corporate officers, and
directors. The Kansas Withholding Tax Guide states that officers and directors of
corporations, sole proprietors, and partners are personally liable for any unpaid
withheld tax, along with any other person determined to be a responsible party.
KENTUCKY
Ky. Rev. Stat. Ann. § 141.340 (2009); Kentucky Tax Registration Application
(2009), available at http://revenue.ky.gov/NR/rdonlyres/4A9BEB16–844E–4F8B–
B095–8825257E54B5/0/10A100409.pdf.
Kentucky requires identification of owners and responsible parties. A Kentucky
statute states that parties such as the president, vice president, secretary, treasurer,
or any other person holding a equivalent corporate office shall be held personally
liable, joint and severally, for any tax required to be withheld.
LOUISIANA
La. Admin. Code tit. 61, § 1511 (2009); Louisiana Application for Revenue Account
Number and Instructions (2009), available at revenue.louisiana.gov/sections/
business/intro.aspx; Louisiana Withholding Brochure.
Louisiana requires identification of corporate officers, partners, or owners.
Administrative law states that a withholding agent is personally liable for amounts
required to be withheld.
98
Fogg
MAINE
Revenue Services and Department of Labor Application for Tax Registration
(2009), available at www.maineboats.com/files/MEApplicationTaxRegistration.
pdf; Email from Maine Revenue Service, Compliance Division, to Fleming Ware,
Research Assistant, Villanova University School of Law (June 18, 2009, 10:51 EST)
(on file with author).
Maine requires disclosure of the person responsible for the finances of the corporation and all directors, partners, and officers. Any or all listed individuals may
be held individually liable for any taxes withheld.
MARYLAND
Maryland Combined Business Application (2009), available at http://forms.
marylandtaxes.com/current_forms/cra.pdf.
Maryland requires identification of the person(s) responsible for remitting
taxes.
MASSACHUSETTS
830 Mass. Code Regs. 62B.2.1 (2009); Massachusetts Department of Revenue:
A Guide to Employer Tax Obligations, http://www.mass.gov/?pageID=dorsub
topic&L=4&L0=Home&L1=Businesses&L2=Current+Tax+Year+Information
&L3=Guide+to+Employer+Tax+Obligations&sid=Ador (last visited August 1,
2009); Massachusetts Department of Revenue: A Guide to Withholding of Taxes
on Wages, http://www.mass.gov/?pageID=dorterminal&L=6&L0=Home&L1=
Individuals+and+Families&L2=Personal+Income+Tax&L3=Forms+%26+
Publications&L4=Publications&L5=Publications+Index&sid=Ador&b=terminal
content&f=dor_publ_withholding&csid=Ador (last visited August 1, 2009); Email
from Massachusetts Department of Revenue, to Fleming Ware, Research Assistant,
Villanova University School of Law, July 1, 2009, 14:50 EST) (on file with author).
Massachusetts requires new businesses to designate someone as the Master
Business Administrator. This individual receives authority over all electronic
business. Massachusetts Department of Revenue requires all withholding forms
be completed online.
MICHIGAN
Michigan Business Taxes Registration Booklet (2009), available at www.michigan.
gov/uia/0,1607,7–118–1360–78056—,00.html; Michigan Tax Form 3683 (2009),
available at www.michigan.gov/documents/3683f_2906_7.pdf.
Collecting Collected Taxes
99
Michigan requires identification of corporate officers. Listed corporate officers
can be personally liable for unpaid taxes. If the corporation hires a payroll provider, the corporation must file Form 3683, which has to be signed by the corporate
officer responsible for paying taxes acknowledging the personal liability for unpaid taxes. There is a duty to update this form if the responsible person changes.
MINNESOTA
Minn. Stat. §§ 270C.56, 290.92 (2007); Minnesota Application for Business
Registration Instruction Booklet (2009), available at http://www.taxes.state.
mn.us/instructions/abr_in.pdf. Minnesota requires identification of corporate
officers.
Minnesota further states that the supplied information will be used for assessment of unpaid taxes.
MISSISSIPPI
Miss. Code Ann. § 27–7–307 (2008). Mississippi requires a list of the officers,
directors, managing partners, or members who have any responsibility for fiscal
management.
MISSOURI
Mo. Rev. Stat. § 143.241.1–2 (2008); Missouri Employer Tax Guide (2005), available at http://dor.mo.gov/tax/business/withhold/forms/2005/4282.pdf; Missouri
Tax Registration Application (2009), available at http://dor.mo.gov/tax/business/
register/forms/2643f.pdf.
Missouri requires identification of all officers, partners, and sole proprietors,
but not of responsible parties.
MONTANA
Mont. Code Ann. §§ 15–30–203, 207.
The Registration Application requires listing of the President or Partner,
Secretary or Partner, and Treasurer or Partner but not specifically the partner
whom is responsible for remitting withholding taxes. The Montana Annual Wage
Withholding Tax Reconciliation requires a name and address when filing the form.
100
Fogg
NEBRASKA
Neb. Admin. Code § 316–33 (2009); Nebraska Circular EN (2008), available
at http://www.revenue.ne.gov/circ-en/2008/circ_en_2008.htm; Nebraska Tax
Application (2009), available at Nebraska Tax Application.
Nebraska requires identification of all corporate officers but does not require
disclosure of the responsible party.
NEVADA
Nev. Rev. Stat. §§ 360.780, 360.784 (2009); Nevada Business Registration
Form and Instructions (2005), available at http://tax.state.nv.us/documents/
APP–01.00%20Nevada%20Business%20Registration%2002–17–05.doc.
Nevada requires identification of corporate officers but not responsible officers.
NEW HAMPSHIRE
New Hampshire Department of Revenue Administration, http://www.nh.gov/
revenue/ (last visited August 1, 2009).
New Hampshire does not collect personal income tax.
NEW JERSEY
Cooperstein v. Director, Division of Taxation, 13 NJ Tax 68 (1993); New Jersey
Business Registration Application & Instructions (2009), available at www.state.
nj.us/treasury/taxation/pdf/other_forms/git-er/njwt.pdf; New Jersey Gross In
come Tax Instruction Booklet (2008), available at http://www.docstoc.com/
docs/766762/2007–New-Jersey-Gross-Income-Tax; State of New Jersey, Division
of Taxation, Responsible Persons, www.state.nj.us/treasury/taxation/respons.
shtml (last visited August 1, 2009).
New Jersey requires taxpayers to disclose the name and SSN of all owners,
partners, or responsible corporate officers. The New Jersey Department of the
Treasury states that a responsible party may be any officer or employee of a corporation who is under the duty to collect and remit trust fund taxes to the State on
behalf of the corporation. The Tax Court of New Jersey looks to factors laid out in
Cooperstein to determine who is a responsible party.
NEW MEXICO
N.M. Code R. § 3.3.2.8–10 (Weil 2009); Application for Business Tax Identification
Number (2009, available at www.tax.state.nm.us/forms/year98/acd31015.pdf; New
Collecting Collected Taxes
101
Mexico Withholding Tax Pamphlet (2009), available at www.tax.state.nm.us/
forms/year99/crsforms.htm.
New Mexico requires disclosure of owners, partners, corporate officers, and
shareholders, but does not require disclosure of a responsible party.
NEW YORK
N.Y. Tax Law §§ 50, 100 (Gould 2009); New York State Tax Guide for New
Businesses (2009), available at www.tax.state.ny.us/pdf/publications/multi/
pub20_1007.pdf; Business Tax Account Update (October 2004), available at www.
tax.state.ny.us/pdf/2004/fillin/misc/dtf95_1004_fill_in.pdf.
The Department of Taxation and Finance requires identification of corporate
officers. Businesses are required to update corporate officer information. The
update form references responsible persons.
NORTH CAROLINA
17 N.C. Admin. Code 6C.0204 (2009); North Carolina Income Tax Withholding
Tables and Instructions for Employers (2009), available at www.dornc.com/
downloads/nc30.pdf; North Carolina Business Registration Application for Income
Tax Withholding (2009), available at www.dornc.com/downloads/fillin/NCBR_
webfill.pdf.
The North Carolina Department of Revenue requires taxpayers register for
withholding. Taxpayers must disclose responsible persons on the withholding
registration form.
NORTH DAKOTA
N.D. Cent. Code § 57–38–60.1–2; N.D. Admin. Code 81–03–03.1–02
(2009); Application to Register for Income Tax Withholding and Sales and
Use Tax Permit (2009), available at http://www.nd.gov/tax/salesanduse/forms/
withholdsalesapplication-enabled.pdf; Income Tax Withholding Guidelines, www.
nd.gov/tax/indwithhold/pubs/guide/index.html.
The State Tax Department requires identification of corporate officers for registration. Guidelines warn of potential personal liability of corporate officers if the
entity fails to remit taxes; however, corporate officers can completely elect out of
personal responsibility if the entity posts a bond.
102
Fogg
OHIO
Ohio Rev. Code Ann. § 5747.13 (2009); Ohio Admin. Code 5703:7–15 (2009);
Ohio Employer Withholding Tax General Guidelines (2009), available at http://
www.tax.ohio.gov/documents/forms/employer_withholding/2006/WTH_
GeneralGuidelines_2006_BW_081006.pdf.
Ohio Withholding Tax Registration requires disclosure of the name and title of
individuals responsible for remitting Ohio withholding tax.
OKLAHOMA
Okla. Stat. Ann. tit. 68, § 253; Okla. Admin. Code 710:90 (2009); Oklahoma
Business Registration Packet (2009), available at www.tax.ok.gov/forms/busregpk.
pdf.
All managers and members of any LLC shall be liable for the failed remittance
of any withholding tax, unless, during the period for which the assessment was
made, a manager(s) or member(s) was specified as a responsible party for withholding tax purposes.
OREGON
Oregon Combined Employers Registration (2008), available at www.oregon.gov/
DOR/BUS/docs/211–055.pdf; Oregon Combined Payroll Tax Report (2008), available at http://www.doc.state.or.us/DOR/BUS/docs/2008Forms/211–155–2–08fill.
pdf.
The Oregon Employment Department requires all employers to disclose all
corporate officers and for what each officer is responsible (filing tax returns, paying taxes, hiring/firing, determining which creditors to pay first).
PENNSYLVANIA
61 Pa. Code § 13.3b (2009); Pennsylvania Enterprise Registration Form and
Instructions (2009), available at www.revenue.state.pa.us/revenue/lib/revenue/
pa-100.pdf.
Pennsylvania requires identification of responsible parties on the Enterprise
Registration Form and identification of the type of tax each person is responsible
for remitting. The Enterprise Registration Form also states that responsible parties can be personally assessed if the entity does not pay.
Collecting Collected Taxes
103
RHODE ISLAND
State of Rhode Island Business Registration, https://www.ri.gov/taxation/BAR/
(last visited August 1, 2009).
Rhode Island requires identification of corporate officers but not identification
of responsible officers.
SOUTH CAROLINA
South Carolina Tax Form 111 and Instructions, available at http://www.sctax.org/
Forms+and+Instructions/Current+Years+Forms+and+Instructions/default.htm.
South Carolina requires identification of corporate officers on its registration
form but not responsible officers.
SOUTH DAKOTA
S.D. Codified Laws § 10–45–24 (2009); S.D. Admin. R. 64:06:01:07:01 (2009);
South Dakota License Requirements For Sales, Use, and Exise Tax (2009), available at http://www.state.sd.us/drr2/businesstax/publications/taxfacts/basic.pdf.
South Dakota requires businesses to register for a sales tax permit with the
Department of Revenue. The online application requires identification of corporate officers, but not responsible parties. A related publication warns against
personal liability of corporate officers.
TENNESSEE
Tenn. Code. Ann. § 67–6–601–02 (2009); Tennessee Application for Business
Registration (2009), available at http://state.tn.us/revenue/forms/general/
f13005_1.pdf. Tennessee requires identification of corporate officers but not responsible officers.
TEXAS
Tex. Tax Code Ann. § 151.202 (Vernon 2009); Texas Sales and Use Tax Permit
Application (2009), available at www.window.state.tx.us/taxinfo/taxforms/ap-201.pdf.
The Comptroller of Public Accounts requires identification and signature of
corporate officers for sales tax registration.
104
Fogg
UTAH
Utah Code Ann. § 59–10–405.5 (2009); Utah State Business and Tax Registration,
TC–69 (2009), available at tax.utah.gov/forms/current/tc-69.pdf; Utah Business
Guide, www.utah.gov/services/business.html (last visited August 1, 2009).
Utah requires identification of corporate officers on its registration form, but
not responsible parties. The form states that all officers will be reviewed by the
Department of Revenue for past unpaid tax debts and may be required to post a
bond. The form also requires acknowledgment of potential personal liability of
officers and the requirement to update corporate officer status.
VERMONT
Application for Business Tax Account (2009), available at http://tax.vermont.gov/
pdf.word.excel/forms/business/s-1&instr.pdf; Guide to Vermont Business Taxes
(2009), available at tax.vermont.gov/pdf.word.excel/business/guidetobustaxes.pdf.
The Department of Taxes requires businesses open a Vermont Business
Account. Businesses must identify all business principals with fiscal responsibility
on this account.
VIRGINIA
Virginia Business Registration Application (2009), available at www.tax.virginia.gov/
web_pdfs/busforms/fr199.pdf; Virginia Business Registration Application Instructions
(2009), available at www.payroll.com/support/PDFs/State/VA_r-1Inst.pdf.
The Department of Taxation requires business to register with the State. The
registration form includes a separate section for listing responsible persons.
WASHINGTON
Wash. Rev. Code §§ 50.12.070, 82.32.030, 82.32.210 (2009); Wash. Admin. Code
§ 458–20–101 (2009).
Washington requires identification of all owners and spouses, but does not require designation of responsible officers.
WEST VIRGINIA
W.Va. Code R. §§ 11–12–3, 11–12–4 (2009); West Virginia State Tax Department
Publication TSD–100 (2009), available at www.state.wv.us/taxrev/taxdoc/tsd100.
pdf; West Virginia Working to Serve You Better! Publication (2009), available at
www.state.wv.us/taxrev/uploads/busapp.pdf.
Collecting Collected Taxes
105
The Tax Commissioner requires identification of corporate officers to obtain
the required business registration certificate. Responsible parties do not have to
be identified.
WISCONSIN
Wis. Stat. §§ 71.67 (6), 73.03 (50) (a) (2009); Application for Business Tax
Registration (2009), available at http://www.dor.state.wi.us/forms/sales/index.
html. Wisconsin requires identification of corporate officers but not responsible
officers.
WYOMING
Wyoming Sales/Use Tax License Application (2009), available at http://revenue.
state.wy.us/PortalVBVS/uploads/ETS%20Form%20001.pdf; Wyoming MultiLevel Marketer’s Sales/Use Tax Application (2009), available at http://revenue.
state.wy.us/PortalVBVS/uploads/ETS%20Form%20001ML.pdf. Wyoming has no
trust fund regime, but requires identification of corporate officers of out-of-state
businesses applying for a sales tax permit to cover more than one independent
sales contractor.
Appendix C—State Bonding Requirements
ALABAMA
Alabama does not have a bonding requirement.
ALASKA
Alaska does not have a bonding requirement.
ARIZONA
Ariz. Rev. Stat. Ann. §§ 42–1102, 42–5006–07 (2009); Ariz. Admin. Code
§§ R15–5–2207, R15–5–601 (2009). Arizona may require a bond to secure the
payment of any tax if the taxpayer has not previously paid taxes. A bond may
be required if a licensee, filing monthly, has been delinquent four times within
two years, if an applicant had its prior license revoked, if an applicant had delinquencies under a previous license, or if the Department concludes that an applicant would be unable to remit taxes. A contractor licensed by the Registrar of
Contractors and a dealer of manufactured housing that has not had a principal
106
Fogg
place of business in Arizona for over one year is generally required to furnish a
bond. Any person who has not had a principal place of business in Arizona for
over one year who enters a prime construction contract in excess of $50,000 must
furnish a bond equal to transaction privilege tax on the contract price and obtain
a certificate from the Director.
ARKANSAS
Arkansas does not have a bonding requirement.
CALIFORNIA
Cal. Rev. & Tax. Code § 6701 (Deering 2009). California allows the Board, whenever it deems it necessary to ensure compliance, to require any person to place
with it any security that the Board determines. Security held by the Board shall be
released after a three-year period in which the person has timely filed all returns
and paid all taxes to California.
COLORADO
Colorado does not have a bonding requirement.
CONNECTICUT
Conn. Gen. Stat. § 12–430 (1) (2009). The Commissioner of Revenue can require
any person subject to sales or use tax to deposit security for payment. Security
can be in any form and amount the Commissioner determines, up to six times the
taxpayers estimated tax liability for the applicable reporting period.
DELAWARE
Delaware does not have a bonding requirement.
DISTRICT OF COLUMBIA
D.C. Code § 47–2208 (2009). The Office of Tax and Revenue may require a security bond from vendors engaged in business in the District to ensure payment
of tax. A bond can also be required from vendors not engaged in business in the
District, but authorized to pay the tax and collect reimbursement from it.
Collecting Collected Taxes
107
FLORIDA
Fla. Stat. § 221.16 (2) (2009). The Department of Revenue can require a bond
to ensure payment of the tax prior to issuing a certificate of registration. In determining if a bond is needed, and the amount thereof, the Department must
consider: (1) applicant’s prior compliance history; (2) type of business; (3) type of
inventory; (4) business location; (5) applicant’s financial status; and (6) anticipated
volume of business. The Department may also require the posting of a bond to
guarantee the payment of the taxes before issuing an importer’s permit to a person
importing taxable personal property in its own trucks in connection with that
person’s business.
GEORGIA
Ga. Code Ann. §§ 48–8–57 (a), 48–2–51, 48–2–59 (c), 48–8–63 (e) (2009); Ga.
Comp. R. & Regs. 560–12–1–.16, 560–12–2–.26 (8) (2009); Georgia provides for the
posting of a bond in the following situations: (1) chronically delinquent dealers; (2)
direct payment permit holders; (3) jeopardy assessments; (4) appeals to superior
court; and (5) subcontractors.
HAWAII
Haw. Rev. Stat. § 231–24 (d) (2009). Collection of a tax pursuant to a jeopardy
bond can be stayed by filing with the Department a bond in whatever amount
and with whatever surety the Department deems appropriate, together with any
further security the department deems necessary to ensure the collection of taxes.
IDAHO
Idaho Code Ann. § 63–3625 (2009). Idaho may require security if the State Tax
Commissioner believes a retailer is not complying with applicable tax statutes.
The Commissioner may require security from habitual delinquents.
ILLINOIS
35 Ill. Comp. Stat. 120/12 (2009). Any person filing an action under the
Administrative Review Law to review a final assessment or revised final assessment issued by the Department must post security within twenty days after filing
the complaint.
108
Fogg
INDIANA
Ind. Code §§ 6–2.5–6–12 (a), 6–2.5–7–8 (2009). The Department of Revenue may
require a retail merchant to post security if it determines the bond is necessary to
insure payment of sales and use tax. The department may also require a motor fuel
distributor to file a bond.
IOWA
Iowa Code § 421.27, 423.35 (2009); Iowa Admin. Code r. 701–11–10 (2009). The
Director may require any person subject to tax to file a bond in order to secure the
collection of the tax due. A bond is required under the following situations: (1)
when the Director determines that certain segments of the business community
are experiencing above average financial failures which might jeopardize the collection of the tax; (2) when an applicant for a new sales tax permit, after a complete
investigation of the applicant’s financial status, would be unable to timely remit the
tax; (3) when a new applicant’s record under a permit for a prior business shows
delinquencies; (4) when the department experienced collection problems while a
new applicant was engaged in a prior business; (5) when a new applicant is substantially similar to a person who would have been required to post a bond under
the certain set guidelines. Existing permit holders may be required to file a bond
if they have: (1) one or more delinquencies in remitting sales tax or filing timely
returns in the last 24 months; (2) two or more delinquencies remitting sales tax or
filing timely returns during the last 24 months if filing quarterly; (3) four or more
delinquencies in remitting sales tax or filing timely deposits or returns during the
last 24 months if filing returns on a monthly basis; and (5) eight or more delinquencies during the last 24 months if filing on a semimonthly basis.
KANSAS
Kan. Stat. Ann. § 79–3616 (2009). Any person subject to tax may be required
to post a bond if the Director of Taxation determines that the collection of taxes
needs to be secure. Corporations applying for a registration certification must
post a bond.
KENTUCKY
Ky. Rev. Stat. Ann. § 139.660 (1) (2009). Security may be required of any taxpayer
subject to state sales and use taxes.
Collecting Collected Taxes
109
LOUISIANA
La. Rev. Stat. Ann. §§ 47.9, 47.306 (D) (1) (2009); La. Admin. Code tit. 61I,
§ 4373 (C) (1). Contractors and subcontractors who are nonresidents must file a
surety bond for all contracts.
MAINE
Me. Rev. Stat. Ann. tit. 36, §§ 5231 (2), 145A (2009). A bond may be required
when an extension of time to pay the tax is granted. A bond is required to stay
collection following a jeopardy assessment.
MARYLAND
Code Ann. Tax-Gen. § 13–825 (g) (West 2009). To protect tax revenue, the comptroller can set an amount to secure payment of tax that is due or may become due
and required acceptable security to be posted.
MASSACHUSETTS
Mass. Gen. Laws ch. 64H, § 30A (2009); Mass. Gen. Laws ch. 64I, § 31A (2009);
830 Mass. Code Regs. 62C.66.1, 64H.3.1 (5) (2009); Massachusetts DOR Directive
06–6, 11/20/2006. Nonresident contractors must deposit with the Department of
Revenue an amount equal to five percent of the total amount of the contract or
furnish a guarantee bond to ensure payment of the tax unless the project is less
than $20,000 or the owner of the real estate being constructed is a tax-exempt
organization or a government agency exempt from tax. Transient vendors are
required to post and maintain a bond to ensure payment of the tax if: (1) the transient vendor solicits taxable sales without first registering as a vendor; or (2) a
notice of assessment is issued to the transient vendor and the vendor fails to pay
the assessed amount. The Commissioner may require a purchaser with direct payment authority to post a bond in an amount acceptable to the Commissioner as a
condition of receiving a Form ST–14, Direct Payment Certificate. Taxpayers must
post security if the Commissioner assesses the tax for failing to file a return, for
filing a false return, or if the Commissioner determines that: 1) the collection of
the tax will be jeopardized by delay; 2) the return filing or payment history of the
taxpayer raises doubt as to the collection of the tax if delayed; or 3) an abatement
application or petition is frivolous and has been filed primarily to avoid prompt
payment of the tax.
110
Fogg
MICHIGAN
Mich. Comp. Laws § 205.53 (2009). If a person engages in a business for which a
privilege tax is imposed, the person shall apply for and obtain a license from the
Department for the current tax year. If the Department considers it necessary in
order to secure collection of the tax, if a taxpayer has at any time failed, refused, or
neglected to pay any tax or interest or penalty upon a tax or has attempted to evade
the payment of any tax, or if the applicant is a corporation and the Department
has reason to believe that the management or control of the corporation is under
persons who have failed to pay any tax or interest or penalty upon a tax under this
act, the Department shall require a surety bond payable to the state.
MINNESOTA
Minn. Stat. § 297A.92 (2009). The commissioner may require a retailer to deposit
security with the commissioner. In lieu of the security, the commissioner may
require a retailer to file a bond.
MISSISSIPPI
Miss. Code Ann. §§ 27–65–21, 23, 33, 61 (2009); IV–1–03 Miss. Code R. § 35
(2009). The commissioner may require taxpayers to file a bond in an amount
double the aggregate tax liability of the taxpayer for any previous three months period within the last calendar year or estimated three-month tax liability. The bond
is to be conditioned for the prompt payment of taxes that are due for each return.
The Commissioner may require a taxpayer to post a bond for other circumstances
where the Commissioner feels it is necessary. Any taxpayer operating a business
from their home or from a temporary location must post a cash or surety bond
sufficient to cover the estimated tax liability for a six-month period. Persons in the
business of selling mobile homes must post a cash or surety bond in an amount
not less than two times estimated liability for three months or prepay the tax before a sales tax permit may be issued. Any person who fails to comply with all the
provisions of the sales tax law forfeits his right to do business in the state until the
person complies with all the provisions, posts an adequate bond, and pays all taxes
due. A surety bond is required of persons who: (1) fail to obtain a sales tax permit
before going into business; (2) continues to operate a business after revocation of
the sales tax permit; (3) fails to keep adequate records and invoices as required
by the state sales tax laws; (4) fails or refuses to permit inspection of records; or
(5) fails to pay any taxes due under the state sales tax laws. Resident contractors
and nonresident contractors subject to the contractor’s tax must either prepay the
estimated contractor’s tax or file a bond with sufficient sureties approved by the
Collecting Collected Taxes
111
Commissioner conditioned on the payment of all sales, use, income, withholding,
and motor fuels tax.
MISSOURI
Mo. Rev. Stat. §§ 144.087 (1), 144.625 (2009); Mo. Code Regs. Ann. tit. 12, § 10–
104.020 (1) (2009). Taxpayers must pay tax owed plus interest and penalties before
a license may be issued. A taxpayer who has defaulted on the filling of payment
of sales or use tax may obtain or reinstate their retail sales license by filling a bond
with the Department of Revenue. The Department may also require vendors to
file bonds where necessary to secure payment of use tax.
MONTANA
Mont. Code Ann. § 15–68–512 (2009). The Department of Revenue may require
a retailer to deposit security in a form and amount that the Department determines is appropriate. In lieu of security, the Department may require a seller to
file a bond to guarantee solvency and responsibility.
NEBRASKA
Neb. Rev. Stat. § 77–2710 (2009). Transient and itinerant vendors are required to
post a bond. Nonresident contractors may be required to post a bond.
NEVADA
Nev. Rev. Stat. §§ 372.510, 372.825 (2009). The Department, whenever it deems it
necessary to insure compliance with this chapter, may require any person subject
to tax to place with it such security as the Department deems necessary. A person
who obtains a permit to collect sales tax after July 1, 1985, shall deposit with the
Department security in an amount equal to twice the estimated yearly tax due
quarterly, or three times the estimated average tax due monthly if filing monthly
returns, but not less than $100. Any person holding a permit in good standing on
July 1, 1985, who becomes delinquent, files a late return, or whose check tendered
as payment is dishonored shall deposit additional security with the Department.
The Department shall require an organization which is habitually delinquent to
deposit with the Department security in an amount equal to three times the average actual tax due quarterly if the organization files its returns quarterly or five
times the average actual tax due monthly if the organization files its returns for
monthly periods.
112
Fogg
NEW HAMPSHIRE
New Hampshire does not have a bonding requirement.
NEW JERSEY
N.J. Rev. Stat. §§ 54:32B–18, 54:49–2 (2009). If the Director deems it necessary
to protect the revenues to be obtained under the state sales and use tax law, he or
she may require any person required to collect the tax to file a bond to secure the
payment of any tax due or which may become due.
NEW MEXICO
N.M. Stat. § 7–1–54 (2009). Whenever it is necessary to ensure payment of any
tax, the Department is authorized to require or allow any person to furnish an
acceptable surety bond in an appropriate amount. If any person neglects to comply with a notice to furnish security, the Department may demand compliance.
If a risk exists that the tax due will not be paid, the Secretary may require any
person liable to furnish security. The Secretary may require taxpayers who protest an assessment or the payment of tax to furnish security for amounts greater
than $200,000.
NEW YORK
N.Y. Tax Law §§ 1137–38, 1252 (Gould 2009); N.Y. Comp. Codes R. & Regs. tit.
20, § IV(A) (539.5) (2009). The Department of Taxation and Finance can require
persons who are required to collect tax to file a bond to secure payment of any tax,
interest, or penalties. The collection of any jeopardy assessment may be stayed
by filing a bond with the Department. Where the Department of Taxation and
Finance deems it necessary to protect the revenues to be collected, it has the power
to require a bond under the procedures set forth in the state sales and compensating use tax law. Whenever the Department deems it necessary to protect the revenues due, it may require any holder of a certificate of authority who is required to
collect tax to file a bond in such amount as may be determined by the department
to secure payment of the tax, penalties, or interest due or which may become due.
NORTH CAROLINA
N.C. Gen. Stat. § 105–164.40 (2009). The Secretary can require a taxpayer to file
an indemnity bond sufficient to protect the interest of the state when a jeopardy
assessment is made.
Collecting Collected Taxes
113
NORTH DAKOTA
N.D. Cent. Code § 57–39.2–12 (3); N.D. Admin. Code 81–04.1–04–05 (2009).
The Commissioner can require a taxpayer to file a bond when it is necessary to
secure the collection of tax. Security must be posted by operators of carnivals,
circuses, show troupes, and similar organizations.
OHIO
Ohio Rev. Code Ann. §§ 5739.05, 5739.15, 5739.30, 5741.06 (LexisNexis 2009);
Ohio Admin. Code 5703:9–08 (2009). The Tax Commissioner may require a
bond under the following circumstances: (1) if a vendor or seller is authorized to
prepay the tax; (2) failure to file two consecutive monthly returns, or three or more
returns in a twelve month period; or (3) if a person issued a jeopardy assessment
wishes to stay execution by filling a petition for reassessment.
OKLAHOMA
Okla. Stat. tit. 68, § 1368 (2009); Okla. Admin. Code § 710: 65–9–2 (a) (1) (2009).
The Tax Commissioner may require every person who holds a sales tax permit and
who is a delinquent taxpayer to furnish to the Commission a cash bond or other
security as the Commission deems necessary to secure payment of taxes. Group
three vendors must make a sufficient cash deposit or sufficient bond with the Tax
Commission as to secure payment of semiannual tax liability before doing business in Oklahoma.
OREGON
Oregon does not have a bonding requirement.
PENNSYLVANIA
Pa. Cons. Stat. § 7277 (2009). The Department of Revenue can require a bond to
be filed by a nonresident individual or foreign business entity that is subject to the
Pennsylvania sales-use tax and is not authorized to do business in Pennsylvania
or does not have an established place of business in the state. If such out-of-state
individual is a building contractor or a supplier delivering materials for work in
Pennsylvania, a bond is required. The Department can require a bond to be filed
by a person who petitions it for reassessment if the assessment exceeds $500. A
transient vendor must post a bond in the amount of $500.
114
Fogg
RHODE ISLAND
R.I. Gen. Laws § 44–19–23 (2009). The Tax Administrator may require any person subject to the taxes imposed by Rhode Island to file with the Administrator a
bond to secure the payment of taxes due or which may become due.
SOUTH CAROLINA
S.C. Code Ann. § 12–54–200 (A). The Department of Revenue may, at its discretion, require taxpayers to post a cash or surety bond upon the taxpayers’ failure
to file a timely return or pay any tax for as many as two tax filing periods during a
twelve-month period.
SOUTH DAKOTA
South Dakota does not have a bonding requirement.
TENNESSEE
Tenn. Code Ann. § 67–6–522 (2009). Dealers who become delinquent for more
than ninety days in the payment of any sales or use tax due to Tennessee must
post with the Commissioner cash or an indemnity bond with good and solvent
surety to secure proper payment of sales and use taxes for which the dealers may
become liable.
TEXAS
Tex. Tax Code Ann. § 151.251 (Vernon 2009); 34 Tex. Admn. Code § 3.327 (a)
(2009). Applicants for sales tax permits and those required to register as retainers
must furnish a bond or security.
UTAH
Utah does not have a bonding requirement.
VERMONT
Vt. Stat. Ann. tit. 32, § 8916 (2009). When deemed necessary to protect motor
vehicle purchase and use tax revenues, the Commissioner of Motor Vehicles may
require a rental company to file a bond issued by a surety company in an amount
Collecting Collected Taxes
115
fixed by the Commissioner that does not exceed the total potential tax liability of
the company to secure payment of any tax, penalties, or interest due.
VIRGINIA
Va. Code Ann. § 58.1–630 (2009). The Commissioner may require any person
subject to the tax to file a bond.
WASHINGTON
Washington does not have a bonding requirement.
WEST VIRGINIA
W. Va. Code R. §§ 11–15–8b, 11–12–20–21, 11–15A–12 (2009); West Virginia
Taxpayer Services Division Publications TSD–330, 11/01/1999, TSD–317, 4/00/1993.
Nonresident contractors are required to post a cash bond, a corporate surety
bond, or an umbrella surety bond prior to performing any contracting activity in
West Virginia. Transient vendors are required to post a $500 bond with the Tax
Commissioner. The Tax Commissioner is authorized to require any taxpayer or
retailer required to collect tax to file a bond as security for payment of the tax.
WISCONSIN
Wis. Stat. § 77.61 (2009). The Department may require any person who is or will
be liable to it for tax to place with it a security that the Department determines.
In determining the amount of the security, the taxpayer’s payment of other taxes
administered and any other relevant facts may be considered. A certified service
provider who has contracted with a seller and filed an application to collect and
remit sales and use taxes on behalf of the seller must submit a surety bond to the
Department to guarantee payment of the taxes.
WYOMING
Wyo. Stat. Ann. § 39–15–306 (b) (2009). To secure payment of sales taxes by
nonresident prime or general contractors, each nonresident contractor must file
a surety bond.
116
Fogg
Appendix D —States Imposing Interest on Responsible
Officers From Due Date of Return
ALABAMA
Ala Code §§ 40–29–72, 40–29–73; Arthur W. McPhillips, Jr. v. State of Alabama
Department of Revenue, No. P–04–377 (Ala. Dept. of Rev. April 5, 2006) (RIA
Checkpoint, Ala. Case Law) (Sales taxes); O. Hugh Campbell v. State of Alabama
Dep’t of Revenue, No. P–04–359, 2005 Ala. Tax LEXIS 81 (Ala. Dept. of Rev. Nov.
10, 2005) (Withholding tax); State of Alabama v. King, No. CV 91–B–2121–S, 1995
WL 423171 (N.D. Ala. 1995) (comparing Alabama law with IRC 6672 side by side
in B.C. 505 case)
ALASKA
Alaska Stat. 23.20.240 (2006); Breck v. State, 862 P.2d 854, 856 n.1 (Alaska 1993)
ARIZONA
Ariz. Rev. Stat. Ann. § 42–5028 (2006) (Sales tax); Ariz. Rev. Stat. Ann. § 43–435
(2006) (Withholding tax); but see Arizona Dep’t of Revenue v. Action Marine, 161
P.3d 1248 (Ariz. Ct. App. 2007); In re Inselman, 334 B.R. 267 (Bankr. D. Ariz. 2005)
(There does, however, appear to be some controversy about how the Arizona statute operates Compare Action Marine and Inselman with state cases decided by
administrative law judges after Inselman.)
CALIFORNIA
Cal. Rev. & Tax. Code § 6829 (West 1998) (Sales tax); Cal. Unemp. Ins. Code § 1735
(West 1986) (Withholding tax); State Bd. of Equalization v. Wirick, 112 Cal. Rptr.
2d 919, 924 (Cal. Ct. App. 2001).
CONNECTICUT
Conn. Gen. Stat. Ann. § 12–414a (West 2000) (Sales); Conn. Gen. Stat. Ann.
§ 12–736 (West 2000) (Withholding); (Sales tax statute is clear that interest applies
back to corporate due date on responsible officer. Withholding tax statute mirrors
6672 and is unclear.)
Collecting Collected Taxes
117
GEORGIA
Ga. Code Ann. § 48–2–52 (Supp. 2007) (Withholding); Ga. Code Ann. § 34–8–
167 (2004); E-mail from Warren R. Calvert, Senior Assistant Attorney General,
Georgia Department of Law, to T. Keith Fogg, Visiting Associate Professor of Law,
Villanova University School of Law (Mar. 18, 2008, 18:13:54 EST)
HAWAII
Haw. Rev. Stat. § 235–64 (b) (Supp. 2007).
IDAHO
Idaho Code Ann. § 63–3078 (2007) (Withholding); Idaho Code Ann. § 63–3627
(2007) (Sales); Idaho State Tax Commission Ruling, No. 19641, 2007 WL 2297072
(Idaho Tax. Comm. April 3, 2007); Idaho State Tax Comission Ruling, No. 17949,
17950, 2004 WL 5215791 (Idaho Tax. Comm. Oct. 7, 2004)
ILLINOIS
35 Ill. Comp. Stat. Ann. 735/3–7 (West 2006) (All trust fund taxes)
INDIANA
Ind. Code § 6–3–4–8 (g) (West 2006) (Withholding); Ind. Code § 6–2.5–2–1 (b),
6–2.5–9–3 (West 2006) (Sales); Russell v. Indiana, No. 49T10–0103–SC–31, 2001
Ind. Tax Lexis 68 (Dec. 6, 2001)
IOWA
Iowa Code Ann. § 422.16.4, 10 (West Supp. 2008); Iowa Code Ann. § 422.4.19 (West
2006) (Withholding); Iowa Code Ann. § 421.26 (West Supp. 2008) (Sales)
KANSAS
Kan. Stat. Ann. § 79–3643 (Supp. 2007) (Sales); Kan. Stat. Ann. § 79–2971 (Supp.
2007) (Excise); Kan. Stat. Ann. § 79–32,107 (e) (1997) (Withholding); Kan. Stat.
Ann. § 79–32,100 (b), (c) (Supp. 2007)
118
Fogg
KENTUCKY
Ky. Rev. Stat. Ann. § 141.340 (2) (LexisNexis Supp. 2007) (Withholding); Ky.
Rev. Stat. Ann. § 139.185 (LexisNexis 2007) (Sales); Koppel v. Revenue Cabinet,
Commonwealth of Kentucky, 777 S.W.2d 938 (Ky. Ct. App. 1989); Nienaber
v. Revenue Cabinet, Commonwealth of Kentucky, No. K92–R–71, 1996 Ky. Tax
LEXIS 379 (Bd. Tax App. March 13, 1996)
LOUISIANA
La. Rev. Stat. Ann. § 47:1561.1 (2006) (Withholding and Sales)
MAINE
Me. Rev. Stat. Ann. Tit. 36, § 177.1 (Supp. 2007) (Withholding and Sales); Prescott
v. State Assessor, 721 A.2d 169 (Me. 1998)
MARYLAND
Md. Code Ann., [Tax-General] § 10–906 (d) (LexisNexis 2004) (Withholding);
Md. Code Ann., [Tax-General] § 11–601 (d) (LexisNexis 2004) (Sales); Nissenbaum
v. Comptroller of the Treasury, No. 3374, 1991 WL 322992 (Md. Tax 1991); Fox v.
Comptroller of Treasury, 728 A.2d 776 (Md. Ct. Spec. App. 1999), cert. denied, 735
A.2d 1106 (Md. 1999) (Sales)
MASSACHUSETTS
Mass. Gen. Laws Ann. ch. 62B, § 5 (West Supp. 2008) (Withholding); Mass. Gen.
Laws Ann. ch. 62C, § 31A (West Supp. 2008); Mass. Gen. Laws Ann. ch. 64H,
§ 16 (West Supp. 2008) (Sales); Berenson v. Comm’r, 604 N.E.2d 704 (Mass.
1992) (Sales); Hazard v. Comm’r of Revenue, No. C261103, 2006 WL 724543 (Mass.
App. Tax. Bd. 2006) (Withholding)
MICHIGAN
Mich. Comp. Laws Ann. § 205.27a(5) (West 2007); Mich. Comp. Laws Ann.
§ 205.65 (West 2007) (Sales and Withholding); Jamian v. Department of Treasury,
No. 256522, 2005 Mich. App. LEXIS 2601 (Mich. Ct. App. Oct. 10, 2005); Dickow
v. Department of Treasury, No. 0329530, 2007 Mich. Tax LEXIS 38 (Mich. Tax
Tribunal Nov. 27, 2007)
Collecting Collected Taxes
119
MINNESOTA
Minn. Stat. Ann. § 270C.56 (West 2007); Minn. Stat. Ann. § 290.92 subdiv. 1 (4)
(West 2007) (Withholding); Minn. Stat. Ann. § 297A.61 subdiv. 2 (West 2007)
(Sales)
MISSISSIPPI
Miss. Code Ann. § 27–7–307 (West 2006) (Withholding)
MISSOURI
(The failure to file the sales or withholding tax return is a prerequisite to trigger
the imposition of tax against the responsible officer; however, if triggered, interest
is charged back to the due date of the entity’s return); Mo. Ann. Stat. § 143.241.2
(West 2006) (Withholding); Mo. Ann. Stat. § 144.157.1 (West 2006) (Sales); Jones
v. Director of Revenue, 981 S.W.2d 571 (Mo.1998); see also Garland v. Director of
Revenue, 961 S.W. 2d 824 (Mo. 1998)
MONTANA
Mont. Code Ann. § 15–68–811 (2007) (Sales); Mont. Code Ann. § 15–30–203
(2007).
NEBRASKA
Neb. Rev. Stat. § 77–1783.01 (2003); Neb. Dep’t. of Revenue, 4–787–1989 Rev.
10–2007, Statutory Responsibilities for Collecting, Reporting, and Remitting
Nebraska Taxes (2007)
NEVADA
Nev. Rev. Stat. Ann. § 360.297 (LexisNexis 2007) (Sales)
NEW HAMPSHIRE
N.H. Rev. Stat. Ann. § 78–A:7 (LexisNexis Supp. 2006); N.H. Rev. Stat. Ann.
§ 78–A:20 (LexisNexis 2001)
120
Fogg
NEW JERSEY
N.J. Stat. Ann. § 54A:9–6 (f), (g), (l) (West Supp. 2008) (Withholding); N.J. Stat.
Ann. § 54:32B–14 (West Supp. 2008); N.J. Stat. Ann. § 54–32B–2 (w) (West Supp.
2008) (Sales); Skaperdas v. Director, Division of Taxation, 14 N.J. Tax 103 (1994),
aff ’d, 685 A.2d 18 (N.J. Super. Ct. App. Div. 1996) (Sales) (It is possible that New
Jersey follows New York and is split on income taxes where interest does not
accrue.)
NEW MEXICO
N.M. Stat. Ann. § 7–3–5 (LexisNexis 2004) (Withholding); In re Baker, No. 2001–
30, 2001 WL 35723190 (N.M. Tax. Rev. Dept. Nov. 1, 2001)
NEW YORK
N.Y. [Tax] Law § 1133 (McKinney 2008) (Sales); Lorenz v. Dep’t of Taxation, 623
N.Y.S.2d 455 (N.Y. App. Div. 1995) (Sales); N.Y. [Tax] Law § 685 (g) (McKinney
Supp. 2008); Yellin v. N.Y. Tax Commission, 81 A.D.2d 196 (N.Y. App. Div. 1981)
(As mentioned above, New York imposes interest back to the due date of the entity
return for unpaid sales taxes but not for unpaid withholding taxes.)
NORTH CAROLINA
N.C. Gen. Stat. Ann. § 105–253 (West 2007) (Withholding and Sales); In re Jonas,
318 S.E.2d 869 (N.C. App. 1984); In re Proposed Assessments, No. 2000–70, 2000
N.C. Tax LEXIS 23 (N.C. Dept. Rev. Oct. 10, 2000); In re Proposed Assessments,
No. 2004–45, 2004 N.C. Tax LEXIS 28 (N.C. Dept. Rev. June 25, 2004); N.C. Dep’t
of Revenue, Individual Income Tax Gift Tax Estate Tax Rules and Bulletins Taxable
Years 2007 and 2008 (2008)
NORTH DAKOTA
N.D. Cent. Code § 57–38–60.1 (2000) (Withholding); N.D. Cent. Code 57–39.2–
18.1 (2000) (Sales)
OHIO
Ohio Rev. Code Ann. § 5747.07 (LexisNexis 2005) (Withholding); Ohio Rev.
Code Ann. § 5739.33 (LexisNexis 2005) (Sales); Soltesiz v. Tracy, Tax Comm’r, 663
N.E.2d 1273 (Ohio 1996)
Collecting Collected Taxes
121
OKLAHOMA
Okla. Stat. Ann. tit. 68, § 2385.3 (E) (West Supp. 2008) (Withholding); Okla. Stat.
Ann. tit. 68, § 2385.6 (A) (West Supp. 2008); Okla. Stat. Ann. tit. 68, § 1361 (A)
(West 2008) (Sales); Oklahoma Tax Commission Decision, 2005–08–16–15, 2005
Okla. Tax LEXIS 15 (Okla. Tax Comm. Aug. 16, 2005)
OREGON
Or. Rev. Stat. § 316.207 (3) (2007) (Withholding); see also Or. Rev. Stat.
§ 316.162 (3) (b) (2007); (No sales tax in Oregon)
PENNSYLVANIA
72 Pa. Stat. Ann. § 7320 (West 2000) (Withholding); 72 Pa. Stat. Ann. § 7201 (e)
(West Supp. 2008) (Sales); In re Hartman, 375 B.R. 740 (W.D. Pa. 2007)
RHODE ISLAND
R.I. Gen. Laws § 44–30–76, 85 (2005) (Withholding); In re Payroll Tax, No. 98–14,
1998 R.I. Tax LEXIS 12 (R.I. Div. of Tax. Aug. 20, 1998); In re Payroll Tax, No.
93–22, 1993 R.I. Tax LEXIS 24 (R.I. Div. of Tax. July 14, 1993); In re Withholding
Tax, No. 94–18, 1994 R.I. Tax LEXIS 17 (R.I. Div. of Tax. June 2, 1994); R.I. Gen.
Laws § 44–19–35 (2005)
SOUTH CAROLINA
S.C. Code Ann. § 12–8–2010 (2000) (Withholding); S.C. Code Ann. § 12–54–195
(Supp. 2007) (Sales)
SOUTH DAKOTA
S.D. Codified Laws § 10–45–55 (2004); S.D. Codified Laws § 10–59–1, 6 (2004,
Supp. 2008)
TENNESSEE
Tenn. Code Ann. § 67–1–1443 (a) (2006) (Sales) (No state income tax)
122
Fogg
TEXAS
Tex. [Tax] Code Ann. § 111.016 (Vernon 2008) (Sales); Dixon v. State, 808 S.W.2d
721 (Tex. App. 1991); (No state income tax)
VERMONT
Vt. Stat. Ann. tit. 32, § 5844 (2007) (Withholding); Vt. Stat. Ann. tit. 32, § 9703
(Supp. 2007) (Sales); Rock v. Dep’t of Taxes, 742 A.2d 1211 (Vt. 1999) (Rock says
withholding and sales tax provisions are treated the same even though they
have slight variation. Withholding statute silent on interest while sales tax
statute explicit)
VIRGINIA
Va. Code Ann. § 58.1–1813 (2004) (Withholding and Sales); In re Individual
Income Tax, No. 05–132, 2005 Va. Tax LEXIS 159 (Va. Dept. of Tax. Aug. 10, 2005)
WASHINGTON
Wash. Rev. Code Ann. § 82.32.145 (West 2000) (Sales); In re Petition for Correction
of Trust Fund Accountability, No. 05–0066, 24 Wash. Tax. Dec. 454 (Wash. Dept.
of Revenue Appeals Div. Mar. 30, 2005); (Washington has no income tax)
WASHINGTON, D.C.
D.C. Code Ann. 47–4491 (Withholding and Sales); Michael v. District of Columbia,
No. 5490–93 (D.C. Super. Ct. Tax Div. Dec. 30, 1997)
WEST VIRGINIA
W. Va. Code Ann. § 11–15–17 (LexisNexis 2005) (Sales); W.Va. Code Ann.
§ 11–10–19 (LexisNexis 2003) (Withholding); Frymier-Halloran v. Paige, 458 S.E.2d
780 (W.Va. 1995); In re Bowen, 116 B.R. 477 (S.D. W. Va. 1990); In re Audia, Nos.
93–384 CS, 93–385 WS, 1994 W. Va. Tax LEXIS 81 (W. Va. Dept. Tax and Revenue
May 27, July 7, 1994) (West Virginia imposes interest back to the due date of the
return for sales taxes but not for income taxes.)
Collecting Collected Taxes
123
WISCONSIN
Wis. Stat. Ann. § 71–83 (1) (b) (West 2004) (Withholding); Wis. Stat. Ann.
§ 77.60 (9) (West 2004) (Sales); Omegbu v. Wisconsin Dept. of Rev., No. 97–W–
342, 1999 Wisc. Tax LEXIS 46 (Tax App. Comm. Oct. 14, 1999)
Appendix E—International Law Treatment of
Responsible Officers
The concept of using businesses to collect taxes for the government exists in other
countries. A brief survey of English speaking countries suggests that even more
support exists for the concept of charging interest back to the due date of the corporate return relating to the collected tax. While these countries all have systems
that differ from the responsible officer system used in the United States, their systems also have much in common with the United States. The concept of holding
individual corporate officers responsible for the failure to pay over taxes collected
by a corporation for the government is one which the countries share even if their
systems of affecting the liability differ.
CANADA
Canada has a provision similar to that of the United States for withholding income
taxes. Persons paying salary, wages or other remuneration must withhold taxes
and remit them to the Receiver General at the time prescribed by regulation.522
The amounts withheld pursuant to this provision are held “in trust for Her Majesty
and for payment to Her Majesty in the manner and at the time provided under this
Act.” 523 Corporate directors of the entity that fails to withhold or remit such taxes
are jointly and severally liable to pay that amount plus any interest and penalties
relating to it.524 The Soper case discusses the objective and subjective tests applied with respect to any director to determine if the director meets the exception
for liability.525
In addition to the liability for the unpaid withholding taxes, the director may
be liable for a 10 percent or 20 percent penalty for a knowing failure to remit the
withheld taxes or gross negligence in the failure to remit.526 The Canadian statute
holding directors liable developed in the 1980s when Canada felt too many companies were failing to pay collected employment taxes.527 The current Canadian
law makes clear that the directors of a company that does not pay over withheld
employment taxes are personally liable for not only the taxes but the penalties and
interest as well that are due from the company.528
124
Fogg
ENGLAND
England requires an employer to withhold its mandated social security contribution from wages. Corporate officers incur liability if the “failure [to pay] appears
to the [Inland Revenue] to be attributable to fraud or neglect on the part of one
or more individuals who, at the time of the fraud or neglect, were officers of the
body corporate.” 529 Unless only one corporate officer exists, England apportions
the liability among the officers based on relative responsibility.530 The amount of
liability asserted against the officers(s) includes interest and penalty amounts due
from the corporation.531 Interest then runs on the amount of the liability Inland
Revenue specifies as due from the individual.532
AUSTRALIA
Corporate directors face personal liability for failure of the corporation to withhold income taxes.533 The directors also face personal liability for failure of the
corporation to timely pay the withheld taxes over to the Government.534 The liability for failure to collect and pay over can equal the full amount of the taxes not
paid. Interest on the unpaid liability is subject to the discretion of the tribunal
imposing the liability.535
Collecting Collected Taxes
125
Endnotes
1
“The tax imposed by section 3101 shall be collected by the employer of the
taxpayer, by deducting the amount of the tax from the wages as and when
paid.” IRC § 3101 (a).
“Except as otherwise provided in this section, every employer making
payment of wages shall deduct and withhold upon such wages a tax.”
IRC § 3402 (a).
2 The term “collected taxes” will be used in this Article to describe taxes that an
individual or entity must collect on behalf of the United States and hold for
payment over to the United States at some future point. The most common
collected taxes are employment taxes that consist of two parts: withheld
income taxes and withheld social security taxes.
3 See U.S. Gov’t Accountability Office, GAO–08–617, Tax Compliance:
Businesses Owe Billions in Federal Payroll Taxes, 19 (2008).
4 See id.
5 See id.
6 See id. at 5.
7 See id.
8 These recommendations are drawn from Transparency in Private Collection
of Federal Taxes, __Florida Tax Review__ (2011). They are reprinted here
with permission from Florida Tax Review.
9 These recommendations are drawn from In Whom We Trust, 43 Creighton
L. Rev. 357 (2010). They are reprinted here with permission from Creighton
Law Review.
10 These recommendations are drawn from Leaving Money on the Table and
Providing an Incentive not to Pay—A Failed Collection Device, 5 Hastings
Business Law Journal 1 (2009). They are reprinted here with permission
from the Hastings Business Law Journal.
11 Take, for example, the Form 941 which reports three different types of
information: withheld income taxes (trust information); withheld social
security taxes (trust information) and the employer’s portion of the social
security taxes (not trust information). Similarly, Form 720 sets out a
reporting mechanism for a variety of excise taxes some of which result from a
trust relationship where the entity filing the Form 720 has collected the excise
taxes from third parties and some of which result from excise taxes directly
imposed on the entity.
12 See Joint Committee on Taxation, Study of Present-Law Taxpayer
Confidentiality and Disclosure Provision as Required by Section 3802 of the
126
Fogg
Internal Revenue Service Restructuring and Reform Act of 1998, Volume I:
Study of General Disclosure Provisions (JCS–1–00), 246–79, January 28, 2000
for a comprehensive discussion of the history of the disclosure laws. In 1998
Congress mandated in Section 3802 of the Revenue Reform Act that the
Joint Committee on Taxation and the Treasury Department prepare reports
on disclosure law. The Joint Committee on Taxation report (JCT Report)
was issued in three volumes: Volume I covers more general issues, Volume
II discusses issues involving exempt organizations and Volume III contains
letters from states and tax authorities on the costs and benefits of disclosure.
The report prepared by the Joint Committee on Taxation contains a thorough
history of disclosure laws in the United States in Volume I at pages 246–79.
While some disclosure history will be briefly summarized below, a more
detailed summary of the history exists in the JCT Report. See Staff of the
Joint Committee on Taxation, Study of Present-Law Taxpayer Confidentiality
and Disclosure Provision as Required by Section 3802 of the Internal Revenue
Service Restructuring and Reform Act of 1998, Volume I: Study of General
Disclosure Provisions (JCS–1–00) January 28, 2000. Although Volume I of
the Treasury Department report (Treasury Report) notes that a Volume II
will be published focusing on IRC § 6104, no Volume II was published by
the Treasury Department. Documented in email dated July 7, 2010 from
Channprett Singh, IRS Office of the Chief Counsel (on file with the author).
Department of the Treasury, Office of Tax Policy, Report to the Congress
on Scope and Use of Taxpayer Confidentiality and Disclosure Provisions
(October 2000) (Treasury Report). Therefore, all references to the Treasury
Report herein are to this Volume I. This section does not seek to provide
comprehensive information concerning this history but only to assist the
reader in understanding the policy debates that have occurred concerning
disclosure of tax information.
13 See, e.g., IRC § 1 (imposing income tax on individuals); IRC § 11 (imposing
income tax on corporations); IRC § 1201 (outlining capital gains tax); IRC
§ 2001 (a) (imposing tax on transfers of estates); IRC § 2501 (imposing tax on
gifts); IRC § 4001 (imposing tax on luxury vehicles); IRC § 4051 (imposing
tax on heavy trucks and trailers); IRC § 4064 (imposing tax on gas guzzlers);
IRC § 4191 (imposing tax on medical devices); IRC § 4261 (a) (imposing tax
on air travel); IRC § 4261 (b) (imposing tax on transportation); IRC § 4375
(imposing fee on health insurance); IRC § 4401 (imposing tax on wagers);
IRC § 4471 (imposing tax on covered voyages); IRC § 4611 (imposing tax on
petroleum); IRC § 5701 (imposing tax on cigarettes).
14 The general rule of nondisclosure of tax information is set out in IRC 6103 (a)
which provides in part that “[r]eturns and return information shall be
confidential,” and except as provided in 6103 the information cannot
be disclosed.”
Collecting Collected Taxes
127
15
See Robert P. Strauss, State Disclosure of Tax Return Information: Taxpayer
Privacy vs. The Public’s Right to Know, 5 State Tax Notes Magazine 24, 25
(July 5, 1993).
16 Id. at 25–26.
17
Id. at 26.
Id.; David Lenter, Joel Slemrod, Douglas Shackelford, Public Disclosure
of Corporate Tax Return Information: Accounting, Economics, and Legal
Perspectives, National Tax Journal Volume LVI, No. 4, December 2003,
at 813.
19 The Privacy Protection Study Commission, created as part of the Privacy Act
of 1974, recommended that Congress make major changes to the disclosure of
federal tax information. The Watergate scandal and ensuing disclosure policy
recommendations caused Congress to evaluate access to taxpayer records.
JCT Report (Vol. I), at 256.
20 Pub. L. No. 94–455 sec. 1202 (a) (1), 90 Stat. 1667 (1976); IRC § 6103 (a).
21 Compare IRC 6103 as it existed in 1976 with the new revised version effective
in 1977.
22 Lenter, Slemrod & Shackelford, supra note 18, at 807.
23 Act of July 14, 1870, 16 Stat. 256, 259 (“[N]o collector, deputy collector,
assessor or assistant assessor, shall permit to be published in any manner
such income return, or any part thereof, except such general statistics not
specifying the names of individuals or firms, as he may make public, under
such rules and regulations as the Commissioner of Internal Revenue
shall prescribe”).
24 28 Stat. 509, c. 349. (“By section 34, sections 3167, 3172, 3173, and 3176 of
the Revised Statutes of the United States as amended were amended so as to
provide that it should be unlawful for the collector and other officers to make
known, or to publish, amount or source of income, under penalty”).
25 See JCT Report (Vol. I), at 248–49. The Joint Committee on Taxation Report
does not discuss any debate concerning disclosure of these types of returns,
suggesting that returns filed by business entities did not become an issue
until later.
26 Act approved Aug. 5, 1909, ch. 6, § 38, 36 Stat. 11, 112.
27 Section 38 of the legislation read as follows:
Sixth—When the assessment shall be made, as provided in this section,
the returns, together with any corrections thereof which may have
been made by the Commissioner, shall be filed in the office of the
commissioner of Internal revenue and shall constitute public records
and be open to inspection as such.
18
128
Fogg
Seventh—It shall be unlawful for any collector, deputy collector, agent,
clerk, or other officer or employee of the United States to divulge or
make known in any manner whatever not provided by law to any
person any information obtained by him in the discharge of his official
duty, or to divulge or make known in any manner not provided by
law any document received, evidence taken, or report made under
this section except upon the special direction of the President; and
any offense against the foregoing provision shall be a misdemeanor
and be punished by a fine not exceeding on thousand dollars, or by
imprisonment not exceeding one year, or both, at the discretion of
the court.
28 Appropriations Act of 1910, Act of June 17, 1910, ch. 197, 36 Stat. 468, 494.
29 Disclosure and Privacy Law Reference Guide, 1-3–1-4, available at: http://
www.irs.gov/pub/irs-pdf/p4639.pdf
30 Id.
31 The Sixteenth Amendment states that “[t]he Congress shall have the power
to lay and collect taxes on incomes from whatever source derived, without
apportionment among the several States, and without regard to any census
or enumeration.”
32 Section G.(d)1. of the Tariff Act of 1913 provided: When the assessment
shall be made, as provided in this section, the returns, together with any
corrections thereof which may have been made by the Commissioner, shall
be filed in the office of the Commissioner of Internal Revenue and shall
constitute public records and be open to inspection as such: Provided, That
any and all such returns shall be open to inspection only upon the order of
the President, under rules and regulations to be prescribed by the Secretary
of the Treasury and approved by the President. 38 Stat. 177.
33 Compare the statement of former President Benjamin Harrison before the
Union League Club of Chicago in 1898:
Each citizen has a personal interest, a pecuniary interest in the tax
return of his neighbor. We are members of a great partnership, and it
is the right of each to know what every other member is contributing
to the partnership and what he is taking from it.
With the statement of Secretary of the Treasury—Mellon:
[w]hile the government does not know every source of income of a
taxpayer and must rely upon the good faith of those reporting income,
still in the great majority of cases this reliance is entirely justifiable,
principally because the taxpayer knows that in making a truthful
disclosure of the sources of his income, information stops with the
government. It is like confiding in one’s lawyer.
Collecting Collected Taxes
34
129
Lenter, Slemrod & Shackelford, supra note 18 (citing Revenue Act of June 2,
1924, Ch. 234, § 257 (b), 43 Stat. 293 (1924)).
35 Section 55 (b) of Revenue Act of 1934, 48 Stat. 680, provided:
“Every person required to file an income return shall file with his return,
upon a form prescribed by the commissioner; a correct statement of
the following items shown upon return: (1) name and address, (2) total
gross income, (3) total deductions, (4) net income, (5) total credits
against net income for purposes of normal tax, and tax payable… .
Such statements or copies thereof shall as soon as practicable be made
available to public examination and inspection in such manner as the
Commissioner, with the approval of the Secretary, may determine, in
the office of the collector with which they are filed, for period of not
less than three years from the date they are required to be filed.”
36 Act of April 19, 1935, Ch. 74; 49 Stat. 158 (repealing the pink slips).
37 JCT Report (Vol. I), at 254 & n.1056 (“In 1939, the disclosure provisions were
codified at section 55 of the Internal Revenue Code. In 1954, the disclosure
provisions moved to their present location in section 6103. No material
change was made from existing law”).
38 See JCT Report (Vol. I), at 255–56.
39 Tax Reform Act of 1976 (P.L. 94–455).
40 See generally Staff of Joint Committee on Taxation, 94th Cong., 2d Sess.,
General Explanation of the Tax Reform Act of 1976, 313–316 (Comm. Print
1976), reprinted in 1976–3 C.B. (Vol. 2) 325–328.
41 “The Joint Committee on Taxation and the Secretary of the Treasury shall each
conduct a separate study on the scope and use of provisions regarding taxpayer
confidentiality and shall report the findings of such study … to Congress.”
42 The JCT Report and the Treasury Report are so thorough that they must be
read by anyone with an interest in this area. As discussed below, this Article
takes off from the point of many of the disclosure policies stated in the Joint
Committee Report.
43 Pub. L. No. 81–814, sec. 341 (1950). The provision sec. 153 (c) of the 1939
Code, was the earliest version of Section 6104. Section 153 (c) was later
codified as Section 6104 (a) of the 1954 Code, without amendment. JCT
Report (Vol. II), at 124 (citing Pub. L. No. 83–591 (1954)).
44 Unlike IRC 6103 which starts with a blanket statement prohibiting disclosure
without an exception, IRC 6104 outlines what will be open to the public,
addressing first tax exempt organizations in IRC 6104 (a) (1) and then pension
plans in (a) (2).
45 See JCT Report (Vol. II), at 5–6, 121.
130
46
47
48
49
50
51
52
53
54
Fogg
Public Law 89–554, 80 Stat. 383. For a discussion on disclosure, see infra note 88.
See JCT Report (Vol. I), at 82 and footnote 293 thereof, citing Tax Analysts
and Advocates v. IRS, 505 F.2d 350 (D.C. Cir. 1974); Fruehauf Corp. v. IRS,
75–2 U.S.T.C. 16,189 (6th Cir. 1975).
See JCT Report (Vol. I), at 82.
Treasury Report, at 27.
Internal Revenue Service Restructuring and Reform Act of 1998, Pub. L.
No. 105–206, § 3509, 112 Stat. 685, 772–74 (codified at IRC § 6110 (i) (2000));
H.R. Rep. No. 105–599, at 298–99 (1998) (Conf. Rep.), reprinted in 1998
U.S.C.C.A.N. See Mitchell Rogovin and Donald Korb, Four R’s Revisited:
Regulations, Rulings, Reliance, and Retroactivity in the 21st century, 46 Dug.
L. Rev. 357 (2007–2008) (“The Act also added Chief Counsel Advice to the
definition of a “written determination” in IRC § 6110 (b). By including Chief
Counsel Advice within the disclosure framework of IRC § 6110, Congress
intended to remove the advice from disclosure under the Freedom of
Information Act, 5 U.S.C. § 552 (2000 & Supp. V 2005). See IRC § 6110 (m)
(2000) (providing that “written determinations” are not subject to mandatory
disclosure); H.R. Rep. No. 105–599, at 302 (1998) (Conf. Rep.), reprinted in
1998 U.S.C.C.A.N.).
Tax Analysts v. IRS, 350 F.3d 100 (D.C. Cir. 2003).
See Written Testimony of IRS Deputy Commissioner Linda Stiff Before the
Senate Committee on Homeland Security and Governmental Affairs Permanent
Subcommittee on Investigations of the collection of federal Employment Taxes,
2 (July 29, 2008) (“Today employment taxes represent the largest portion
of total tax dollars collected by the IRS. In FY 2007 for example, of the $2.7
trillion in taxes collected by the IRS, $1.7 trillion was payroll taxes. This
means that approximately two out of every three dollars collected by the
IRS are from required withholding on employment tax returns. Of this $1.7
trillion collected in withholding and FICA taxes, approximately $778 billion
was collected for Social Security and Medicare and approximately $992
billion was collected for individual withholding taxes”).
See Edward K. Cheng, Structural Laws and the Puzzle of Regulating Behavior,
100 NW. U.L. Rev. 655 (2006); Leandra Lederman, Statutory Speed Bumps:
The Roles Third Parties Play in Tax Compliance, 60 Stan. L. Rev. 695 (2007);
Richard Thaler & Cass Sunstein, Nudge: Improving Decisions About Health,
Wealth and Happiness (Yale University Press 2008); Erich Kirchler, The
Economic Psychology of Tax Behavior (Cambridge 2007).
See IRC § 4251 (imposing tax on communication, including local telephone
service, toll telephone service, and teletypewriter exchange service).
Collecting Collected Taxes
55
56
57
58
59
60
61
131
IRC § 4291 (“Every person receiving any payment for facilities or services on
which a tax is imposed upon the payor thereof under this chapter shall collect
the amount of the tax from the person making such payment.”) Airlines use
a similar system as they collect the airline excise tax from their customers.
The amount of the excise tax is added to the cost of the ticket and collected at
the time of purchase.
Instructions for Form 720 (Rev. January 2010) (“Use Form 720 and
attachments to report liability by IRS No. and pay the excise taxes listed on
the form”).
Publication 15, (Circular E), Employer’s Tax Guide, Internal Revenue Service,
available at http://www.irs.gov/pub/irs-pdf/p15.pdf.
Id. There may be instances when the entity need not set this money aside in a
separate account but can continue to hold it in the entity’s general account.
See discussion of Begier v. IRS, 496 U.S. 53 (1990) infra at note 165.
IRC § 6672, 7501 (addressing collected taxes, but neither provision specifying
which type of collected taxes.) States do make distinctions between sales
taxes and withholding taxes which can ultimately only be attributed to this
difference. See Appendix C (listing states which have adopted bonding
laws for unpaid sales taxes with no complimentary bonding laws for unpaid
withholding taxes). For a discussion of how certain states charge interest
to responsible officers from due date of sales tax return but not due date of
withholding tax return, see infra part three, Charge Interest and Penalties to
Responsible Officers Based on the Liability of the Entity for Unpaid Collected
Taxes—Accountability.
IRC § 31 (a) (1) provides that “the amount withheld as tax under chapter
24 shall be allowed to the recipient of the income as a credit against the
tax imposed by this subtitle.” No parallel credit provision exists for excise
taxes such as the communication or airline excise taxes; however, the same
rules of principal and agent govern the transaction. When a taxpayer pays
their phone bill, including the communication excise tax, that taxpayer
expects credit for such payment and would not welcome an appearance by
the IRS seeking to collect the tax from the taxpayer for a second time. This
circumstance is acknowledged by Charles Schriebman in his text IRS Tax
Collection Procedures where he stated that “[i]f a collecting agency (other
than a partnership or sole proprietorship) has failed to pay over excise taxes
it had collected from patrons or members, the IRS will explore the possibility
of asserting the trust fund recovery penalty against the collecting agency’s
responsible persons.” P. 9132 (CCH 3d. 2002). His statement acknowledges
the liability of the collecting entity for the excise tax.
132
Fogg
62
IRC § 31 (a) (1) (“The amount withheld as tax under chapter 24 shall be
allowed to the recipient of the income as a credit against the tax imposed by
this subtitle”).
63 See IRC § 7501.
64
65
66
67
68
69
70
71
72
See discussion of Begier v. IRS, 496 U.S. 53 (1990), infra note 165.
JCT Report (Vol. II), at 6.
Tax exempt organizations also must comply with the employment tax
provisions. Many tax exempt organizations have large employee bases and
collect vast amounts of taxes from their employees.
The payment of excise taxes generally occurs semimonthly, with several
narrow exceptions. IRS Publication 510, Ch. 11 (available at: http://www.irs.
gov/publications/p510/ch11.html).
The IRC currently provides for eleven public trust funds: IRC § 9501 Black
Lung Disability Trust Fund, IRC § 9502 Airport and Airway Trust Fund,
IRC § 9503 Highway Trust Fund, IRC § 9504 Sport Fish Restoration and
Boating Trust Fund, IRC § 9505 Harbor Maintenance Trust Fund, IRC § 9506
Inland Waterways Trust Fund, IRC § 9507 Hazardous Substance Superfund,
IRC § 9508 Leaking Underground Storage Tank Trust Fund, IRC § 9509
Oil Spill Liability Trust Fund, IRC § 9510 Vaccine Injury Compensation
Trust Fund, and IRC § 9511 Patient-Centered Outcomes Research Trust
Fund. The Treasury website contains monthly financial reports for trust
funds administered by the Treasury Department. Available at: http://www.
treasurydirect.gov/govt/reports/tfmp/tfmp.htm.
Because of the “trusting” nature of taxpayers who rely on payroll providers,
a number of these providers have perpetrated Ponzi style schemes in which
they take the money from the taxpayers’ accounts and use some of it for
personal gain rather than using the money to pay the taxes. By the time the
schemes collapse, potentially thousands of taxpayers who actually had money
drawn out of their accounts to pay over the collected taxes find themselves
with a tax bill for these taxes. See, e.g., In re FirstPay, Inc., Nos. 09–1076,
09–1107, 2010 WL 3199858 (4th Cir. Aug. 13, 2010). See infra note 301.
See GAO 2005 report (U.S. Gov’t Accountability Office, GAO–05–637,
Financial Management: Thousands of Civilian Agency Contractors Abuse the
Federal Tax System with Little Consequence, 2 (2005)).
See Doc 2010–1397; see also Michael Joe, Obama Seeks to Block Tax Debtors
from Receiving Federal Contracts, Tax Notes Today, January 21, 2010.
GAO 2005 report (U.S. Gov’t Accountability Office, GAO–05–637, Financial
Management: Thousands of Civilian Agency Contractors Abuse the Federal Tax
System with Little Consequence, 2 (2005)).
Collecting Collected Taxes
73
74
75
76
77
78
79
133
See Christopher S. Rizek, Taxpayer Privacy and Disclosure Issues Will
Continue to Touch Us All, in The Future of American Taxation: Essays
Commemorating the 30th Anniversary of Tax Notes, 81, 89 (“The short answer
is, unfortunately, that no one really knows as a factual matter what the
link is between the confidentiality of taxpayer information and voluntary
compliance. The claim that confidentiality fosters compliance is rather,
something like an article of faith, for or against which only anecdotal and not
particularly conclusive evidence can be offered”).
See Treasury Report (Vol. I), which makes the concern for collateral nonreporting a basis for its recommendations concerning correct policy in this
area placing much more emphasis on this factor than the JCT Report.
See JCT Report (Vol. I), at 127–33 and Treasury Report, at 33–37 for a policy
discussion. On the last two points, see also 2006 TNT 115–18 for a discussion
by Mark Boyle, writing as President of Tax Executives, Inc. to Senators Grassley
and Baucus. Mr. Boyle strongly opposed disclosure of corporate tax returns
citing many reasons for his opposition. Interestingly, one reason was that “public
disclosure of corporate tax returns would effectively represent the outsourcing of
a core governmental function—the examination of tax returns—to the public or
the media.” That comment provides an interesting bookend to the discussion of
disclosure surrounding private tax collectors, infra.
On these three points, see the policy discussions in the JCT Report (Vol. II),
at 5–9.
For example, the Department of Justice may access returns and return
information for use in tax administration proceedings. IRC § 6103 (h) (3).
The Department of the Treasury may access returns and return information
when a need to know is demonstrated. IRC § 6103 (h) (1). The Department
of Commerce, the Federal Trade Commission, and the Department of
Agriculture all may access returns and return information upon written
request to the IRS. See JCT Report (Vol. I) at 43–44; see also Rizek, supra
note 73, at 86–87 (discussing ever-present pull from outside sources at
incredible database maintained by IRS which each claimant for exception
sees as important resource).
Compare JCT Report (Vol. I), at 5 (discussing IRC 6103), with JCT Report
(Vol. II), at 6 (discussing IRC 6104).
JCT Report (Vol. I), at 6. Emphasis added to the word “agency” because
the JCT’s use of that word makes an important statement as a part of this
policy. That term basically speaks of disclosures to government entities and
not to individuals. Yet, two of the exceptions contained in IRC 6103 concern
disclosures to individuals and not to government entities. IRC § 6103 (c), (e).
The essential exclusion of these two exceptions in the JCT Report’s
134
Fogg
conclusion concerning disclosure policy reflects, as discussed below, that
disclosures to individuals almost always occur only in the absence of
privacy interests.
80 JCT Report (Vol. II), at 6.
81
82
83
84
85
86
87
88
Perhaps the better view of the second principle is that it is simply a broad
exception to the first and not really a second principle unto its own. Many
exceptions to the general rule of nondisclosure exist in IRC 6103, and the
disclosure of tax exempt returns in IRC 6104 simply represents one of those
exceptions, albeit a broad one.
Act of April 19, 1935, Ch. 74; 49 Stat. 158.
Although privacy has a strong foundation in this country, individuals
arguably have severely diminished privacy expectations due to the advances
of the Internet Age. A simple search in a search engine of an individual’s
name may produce results detailing that individual’s political party
affiliation, locations where the individual owns property and how much
each is worth, phone numbers and even relatives of the individual, and a
link the individual’s Facebook profile. With all of this information “floating”
around and easily accessible by the public, privacy considerations for tax
return information, which may reveal less than what an internet search may
uncover, are potentially worth less than they used to be.
JCT Report, (Vol. I), at 5; Treasury Report, at 34; see also Rizek, supra note 73,
at 89.
JCT Report (Vol. I), at 238.
The rules listed here do not include the disclosure exceptions carved out in
IRC 6104 which will be discussed separately below.
1) The entity receives substantial subsidies from the government such as tax
exempt organizations.
2) The entity exists to hold funds in trust for the public such as pension
plans.
3) The entity exerts political influence without adequate accountability such
as the political organizations described in IRC 527.
These terms are defined in IRC § 6103 (b) (1) & (2) and are discussed infra at
note 92.
Around the same time Congress amended the disclosure provisions in
the Internal Revenue Code in 1976 to usher in the modern era, it was also
looking at similar issues from a broader perspective. In 1966, Congress
passed the Freedom of Information Act (FOIA) (5 U.S.C. 552) and in 1974 it
passed the Privacy Act (5 U.S.C. 552 (a)). FOIA established a right to access
certain information held by the federal government. The purpose was to
Collecting Collected Taxes
89
90
91
92
93
135
allow citizens of the United States to be better informed so they could fight
corruption and hold those governing accountable. NLRB v. Robbins Tire
& Rubber Co., 437 U.S. 214, 242 (1978). The Privacy Act created rules to
govern the use of personnel information concerning individuals working for
the federal government. All of these changes occurred as the government
recognized the massive databases that it maintained and the good or evil that
could result from the dissemination of information in those databases.
United States v. Dickey, 268 U.S. 378, 387 (1925).
Reading the letters from the state taxing authorities to the Joint Committee
provides an easy source of the benefits which stem from disclosing tax
information to state taxing authorities. Study of Present-Law Taxpayer
Confidentiality and Disclosure Provisions as Required by Section 3802 of
the Internal Revenue Service Restructuring and Reform Act of 1998, Volume
III: Public Comments and General Accounting Office Reports (JCS–1–00),
January 28, 2000.
“The IRS maintains standing agreements with the States and the District
of Columbia for disclosure of returns and return information. The basic
agreement, Agreement on Coordination of Tax Administration, provides for
the mutual exchange of returns and return information between a specific
State tax agency and the IRS.” JCT Report (Vol. I), at 28.
The limiting language requires an explanation of terms. The terms
“return” and “return information” are defined terms in the statute.
IRC § 6103 (b) (1) – (2). A “return” is “any tax or information return,
declaration of estimated tax, or claim for refund required by, or provided
for or permitted under” title 26 of the United States Code. The term “return
information” is much longer, comprising four subparts. Essentially, return
information encompasses all of the data associated with a taxpayers file for a
particular return.
While the policy issue here presents little challenge, the administration of this
provision does provide some challenges for the IRS. It must determine the
form of adequate consent and the execution of such consent. See Huckaby v.
IRS, 794 F.2d 1041 (5th Cir. 1986) (unlawful disclosure based on oral consent);
Olsen v. Egger, 594 F. Supp. 644 (S.D.N.Y. 1984) (consent in divorce decree
not binding on IRS); Tierney v. Schweiker, 718 F.2d 449 (D.C. Cir. 1983)
(open ended consent for “all years” not valid; consents were coerced based
on fear of losing social security benefits); Hefti v. Loeb et al., 1992 U.S. Dist.
LEXIS 12644 (C.C. Ill. 1992) (disclosure to one spouse permitted with respect
to joint return); Ward v. United States, 973 F. Supp. 996 (D. Colo. 1997)
(disclosure in public form during radio broadcast unauthorized because
consent did not designate persons to whom disclosure could be made).
136
94
Fogg
IRC § 6103 (d) (1) (“[S]uch return information shall not be disclosed to the
extent that the Secretary determines that such disclosure would identify
a confidential informant or seriously impair any civil or criminal tax
investigation.”) Another concern is disruption if taxpayer invites persons to
participate in a meeting whose goal in the meeting might be to impair tax
administration. See United States v. Finch, 434 F. Supp. 1085 (D. Colo. 1977);
Treas. Reg. 301.6103 (c) –1 (c) and Delegation Order No. 156.
95 It includes return information “if the Secretary determines that such
disclosure would not seriously impair Federal tax administration.”
IRC § 6103 (e) (7).
96 With respect to joint returns filed by the spouses but not other returns.
IRC § 6103 (e) (1) (A) (ii).
97 With respect to those portions returns filed by the child’s parents which
contain information necessary for the child to comply with IRC § 1 (g). IRC
§ 6103 (e) (1) (A) (iii).
98 IRC § 6103 (d) (1) (E). Heirs can also obtain tax information from an estate tax
return to the extent that the heirs demonstrate a material interest in the estate
to the IRS.
99 IRC § 6103 (d) (1) (F). Beneficiaries of trusts can also obtain tax information
from a trust tax return to the extent that the beneficiaries demonstrate a
material interest in the trust to the IRS.
100 IRC § 6103 (e) (2).
101 IRC § 6103 (e) (3). Heirs can also obtain tax information concerning deceased
individuals to the extent that the heirs demonstrate a material interest in the
information contained in those income tax returns.
102 IRC § 6103 (e) (4) & (5). These individuals can receive the returns filed by the
estate being administered or prior returns of the individual or entity whose
estate they administer if they can demonstrate a material interest in the
information contained in the prior returns.
103 IRC § 6103 (e) (6).
104 IRC § 6103 (e) (8). This exception allows a former spouse to receive
information concerning collection action with regard to a tax liability for
which the former spouse is jointly liable with the taxpayer.
105 IRC § 6103 (e) (9). This exception allows a person responsible for taxes
pursuant to IRC § 6672 to learn if others have also been held liable for the
same penalty and, if so, the collection actions taken with respect to the other
responsible officers.
106 This last basis applies to responsible officers. The 6672 liability does not
strictly relate to a tax return. No return is filed that reports such a liability.
Collecting Collected Taxes
137
Rather, the liability is derivative resulting from a failure of certain persons to
meet their statutory obligations to collect and pay over certain taxes. This
provision, like IRC § 6103 (e) (8) that addresses collection information on
former spouses, was added to IRC § 6103 in 1996. By adding this provision,
Congress acknowledged that joint liability creates a need to know that
overrides individual privacy concerns. The policy reasons behind this
provision are distinct from most other material disclosures in that the need
for information actually outweighs the individual’s policy concerns, rather
than the requesting party eliminating privacy concerns by stepping into the
taxpayer’s shoes. While the information disclosure is based on a material
interest, the nature of the material interest here differs from that of most of
the persons on this list. (Disclosure to a child for compliance with IRC § 1 (g)
and disclosure to a spouse concerning collection on a joint return also fall
within this basis for an exception.)
107 See 2000 Disclosure Lit. Reference Book at 8-2–8-5 for cases and details on
form of disclosure; see also Jct Report (Vol. I) at 163.
108 Long v. United States, 972 F.2d 1174 (10th Cir. 1992); Smith v. United States, 964
F.2d 630 (7th Cir. 1992), reh’g en banc, denied, 1992 U.S. App. LEXIS 19344
(7th Cir. 1992), cert. denied 506 U.S. 1067 (1993); Bator v. Internal Revenue
Service, 89–1 U.S.T.C. 9138 (D. Nev. 1988), aff ’d without published opinion sub
nom, Bator v. United States, 899 F.2d 1224 (9th Cir. 1990), cert. denied, 498
U.S. 893 (1990); Rueckert v. Internal Revenue Service, 775 F.2d 208 (7th Cir.
1985); Taylor v. United States 106 F.3d 833 (8th Cir. 1997), aff ’g 915 F. Supp.
10115 (N.D. Iowa 1996); White v. Commissioner, 537 F. Supp. 679 (D. Colo.
1982); Loomis v. Internal Revenue Service, 81–1 U.S.T.C. 9341 (D. Colo. 1981);
Davis v. United States, 80–2 U.S.T.C. 9794 (D. Mass. 1980), aff ’d 81–1 U.S.T.C.
9458 (1st Cir. 1981).
109 See IRS, Electronic Data Exchange Pilot Project (EDS), http://www.irs.gov/
privacy/article/0,,id=132017,00.html.
110 California uses federal tax information to locate tax debtors, especially those
who are out of state and cannot be located through the post office or other
skip tracing methods, to identify the amount and sources of tax debtors’
assets, [and] to verify the accuracy of taxpayer-supplied information…” JCT
Report (Vol. III), at 66. Colorado stated that the federal tax information is
the “cornerstone of our income tax compliance program” and that it is used
for statistical analysis for informed economic decision-making. Id. at 69.
Hawaii also states that it uses federal tax information on individuals and
businesses for “statistical and compliance purposes.” Id. at 72.
111 See preceding footnote. In addition to the letters from the individual states,
the Federation of Tax Administrators submitted a detailed letter addressing
the need for states to “use tax return information and the adequacy of
138
Fogg
present-law protections governing taxpayer privacy.” See JCT Report (Vol.
III), ¶ 98.
112 Letter from Federation of Tax Administrations, JCT Report (Vol. III), ¶ 100.
113 Id., ¶ 101.
114
See id.
IRC § 6103 (d) (6). Those safeguards are detailed in IRS Publication 1075,
Tax Information Security Guidelines for State, Local and Federal Agencies.
States safeguard confidential taxpayer data in accordance with IRS guidelines.
Many states implement training and education programs to instruct
employees on proper procedures to protect this data. See, e.g., JCT Report
(Vol. III), at Paragraph 83, 339, 347, 363, 387, 422, 505, 519, 533.
116 IRC § 6103 (d) (1).
117 “A prerequisite to disclosure is a written request by the head of the agency,
body or commission. The IRS maintains standing agreements with the
States and the District of Columbia for disclosure of returns and return
information. The basic agreement, Agreement on Coordination of Tax
Administration, provides for the mutual exchange of returns and return
information between a specific State tax agency and the IRS.” JCT Report
(Vol. I), at 28.
118 For a general discussion of these provisions see General Accounting Office,
Taxpayer Confidentiality: Federal, State, and Local Agencies Receiving
Taxpayer Information, GAO–GDD–99–164 (August 1999) ; see also Jct
Report (Vol. I), at 43–45.
119 Compare the disclosure to the Department of Commerce for the Bureau
of the Census which allows both return and return information with the
disclosure to Commerce’s Bureau of Economic Analysis which only releases
return information. IRC § 6103 (j) (1) (A) (Census Bureau) with 6103 (j) (1) (B)
(Economic Analysis Bureau).
120 These agencies consist of the Commerce Department, the Congressional
Budget Office, the Federal Trade Commission, the Treasury Department and
the Agriculture Department. The exceptions granted here do not reach all
federal agencies but only agencies that demonstrated a specific need related
to the statutorily mandated tasks governed by that agency’s directives.
121 To the Commerce Department the statute says “for the purpose of, but
only to the extent necessary in, the structuring of censuses and national
economic accounts and conducting related statistical activities authorized
by law.” IRC § 6103 (j) (1). To the Treasury Department the statute says
“only to the extent necessary in, preparing economic or financial forecast,
projections, analyses, and statistical studies and conducting related activities.”
IRC § 6103 (j) (3). Each grant to an agency has similar limiting language.
115
Collecting Collected Taxes
122
139
IRC § 6103 (j) (4) provides that “[n]o person who receives a return or return
information under this subsection shall disclose such return or return
information to any person other than the taxpayer to whom it relates except
in a form which cannot be associated with or otherwise identify, directly or
indirectly, a particular taxpayer.”
123 IRC § 6103 (k) (2).
124 IRM EXH 1.2.49–2.
125 National Taxpayer Advocate, Report to Congress Fiscal Year 2011 Objectives,
14 (“[T]he filing of the NFTL in the public record might actually prevent
the taxpayer from borrowing money to fully pay the outstanding tax
liability”) available at http://www.irs.gov/pub/irs-utl/nta2011objectivesfinal.
pdf. National Taxpayer Advocate, Report to Congress Fiscal Year 2010
Objectives, 54 (“Filing an NFTL on outstanding liabilities may create serious
consequences for a taxpayer, including making it more difficult to obtain
credit”) available at http://www.irs.gov/pub/irs-utl/fy2010_objectivesreport.pdf.
126 See William T. Plumb, Jr., Federal Tax Liens 53–54 (The American Law
Institute 1981) (1962).
127 See id.
128 IRC § 6323 (f); Reg §301.6323 (f)–1 (d) (“A Form 668 must identify the
taxpayer, the tax liability giving rise to the lien, and the date the assessment
arose regardless of the method used to file the notice of Federal tax lien”).
129 File a Notice of Federal Tax Lien, http://www.irs.gov/businesses/small/
article/0,,id=108339,00.html (last visited July 21, 2010).
130 A vulnerability of this type currently exists with respect to liens for unpaid
real estate taxes. In most, if not all, jurisdictions, these liens can jump ahead
of mortgages and other liens created prior in time. To protect themselves,
mortgage lenders require borrowers to escrow their real estate taxes. In this
manner, mortgage lenders protect themselves from nasty surprises. If federal
tax liens could, without notice, similarly defeat lenders, lenders would either
be required to fashion some type of protective mechanism as with mortgage
liens, be exposed to defeat or forego lending. The problem with fashioning
a protective mechanism is that unlike real estate taxes, which are relatively
easily ascertained and predicted, federal taxes could be assessed for very
unpredictable amounts.
131 Jct Report (Vol. I), at 238–242. The specific proposal before the JCT
staff concerned publication of the names of persons who did not file tax
returns. The JCT staff ’s concerns extended beyond whether such a proposal
would reap collection benefits and into the area of the reliability of the data
concerning who had not filed a return. The combination of both concerns
effected the view of the staff on the failure of such a shaming provision to
140
Fogg
demonstrate a compelling interest for disclosure. The concerns about the
reliability of the data listing persons with unfiled returns raises issues on the
benefits side of the test since disclosure of incorrect data could destroy any
benefits received even if collection from some persons increased as a result of
the disclosure.
132 See Joshua Blank, What’s Wrong With Shaming Corporate Tax Abuse, 62 Tax
L. Rev. 539, 563 (2009).
133 The way shaming laws can meet the tests necessary to qualify as an exception
to the rule of nondisclosure is to ride on the back of the exception allowing
the publication of the notice of lien. Several of the states that permit
shaming have explicitly stated this as their basis for publishing the shaming
lists. See, e.g., Maryland, South Carolina and Virginia, listed in Appendix
A. Essentially, these states have determined that the taxpayer has little or no
privacy interest in the information because the information already exists
in the public domain. Since it exists in the public domain and no privacy
interest are implicated, the benefits derived by publishing the information
need not be as great in order to move the needle over to the disclosure side.
These states view the shaming provision as merely a formatting issue more
than a disclosure issue.
The reasoning used some by states, an absence of privacy interests in
the disclosed information, in adopting shaming laws would not work for
corporate shaming provisions. With corporate shaming, the taxpayer’s
privacy interests have not been removed by the public filing of an NFTL.
While the corporate interests in privacy may not equal those of individuals,
these interests remain substantial. The benefits side of the equation would
need to pull full weight in order to move the needle on the dial over to the
disclosure side.
134 See Jct Report (Vol. II), at 6.
135 “The present law rules requiring disclosure of returns and return information
relating to tax-exempt organizations reflect a determination that, because
such organizations are supported by the public, both through the tax benefits
associated with tax-exempt status and, in some cases, direct contributions,
such organizations have a different expectation of privacy than taxable
persons and the public has a strong interest in information regarding such
organizations.” JCT Report (Vol. II), at 63.
136 For a general discussion of the history of IRC § 6104, see JCT Report (Vol. II,
Appendix A), at 120–129.
137 See Section 101 of the Internal Revenue Code of 1939.
138 Revenue Act of 1943, Pub. L. No. 78–235, ch. 63, sec. 117, 58 Stat. 21, 36–37.
Collecting Collected Taxes
139
141
Revenue Act of 1950, Pub. L. No. 81–814 (1950). S. Rep. No. 81–2375, 125
(1950).
140 Form 990, already in existence at that time, was opened to public inspection.
Obtaining the form required a written request to the IRS. Public Law 81–814
became Section 153 (c) of the 1939 Code which then became Section 6104 of
the 1954 Code. Pub. L. No. 83–591 (1954).
141 Pub. L. No. 85–866, sec. 75. “The committee believes that making these
applications available to the public will provide substantial additional aid
to the Internal Revenue Service in determining whether organizations
are actually operating in the manner in which they have stated in their
applications for exemption.” H.R. Rep. No. 85–262, at 41–42 (1957).
142 Pub. L. No. 93–406, sec. 1022 (g) (1) effective for applications filed after
September 2, 1974.
143 Pub. L. No. 106–230, 1 (b) (a) (A) (i) – (vi) amended IRC § 6104 to include
political organizations, effective July 1, 2000.
144 Volume II of the JCT Report which specifically deals with IRC § 6104 did not
address issues concerning pension plans or political organizations but only
tax exempt organizations. The reasons for pension plans may not mirror
those of exempt organizations because of the trust nature of the pension
plans. The JCT Report also did not discuss the Black Lung Trust information
made public under IRC § 6104, which is discussed elsewhere in this report.
145 JCT Report (Vol. II), at 6.
146 The benefits are discussed briefly supra at footnotes 67– 68 and
accompanying text.
147 See generally Bruce Hopkins, The Law of Tax-Exempt Organizations 406, 408
(8th Ed. 2003) (discussing the tax exempt organizations created under IRC
501 (c) (21) called “Black Lung Benefits Trusts”); see also John Lopatto III,
The Federal Black Lung Program: A 1983 Primer, 85 W. Va. L. Rev. 677 (1983)
(discussing the general law of black lung benefits with a discussion in Section
XI of the creation of the public trust under IRC § 9501).
148 Hopkins, supra note 147, at 406.
149 IRC § 4121 imposes the excise tax on extraction of coal. See also 30 U.S.C.
§ 934 for creation of the trust into which the excise is paid.
150 Money paid into BLBTs is not considered a collected tax because Black Lung
Benefits Trusts are not funded by taxes. Rather, the mine operators pay this
money as an alternative to commercial insurance coverage or state workman’s
compensation for pneumoconiosis. The payments by the mine operator to this
trust are deductible under IRC § 192. See Hopkins, supra note 165, at 406–08.
142
151
Fogg
IRC § 9501. The Black Lung Disability Trust Fund was established on the
books of the Treasury in fiscal year 1978 according to the Black Lung Benefits
Revenue Act of 1977 (Public Law 95–227). The Black Lung Benefits Revenue
Act of 1981 (Public Law 97–119) reestablished the fund in the IRC § 9501. The
Black Lung Disability Trust Fund is one of 10 public trusts created in the
Internal Revenue Code. See IRC § 9501 to 9510. It is the only one of these 10
to accept a portion of its contributions from a tax exempt organization. IRC
§ 9501 (a) (2) (C) provides that a portion of the receipts of this trust fund can
come from Black Lung Disability Trust Funds described in IRC § 501 (c) (21).
152 “Congress established this form of self-insurance program, with similar tax
consequences (from the point of view of the operator) as would result if the
operator had purchased non-cancellable accident and health insurance.”
Hopkins, supra note 147, at 406–07, citing S. Rep. No. 95–336, 95th Cong., 2d
Sess. 11–12 (1978).
153 Pension plans are subject to public inspection so that participants may
comment on employer plan submissions and to ensure compliance with certain
antidiscrimination rules. See David S. Preminger, E. Judson Jennings, and
John Alexander, What Do You Get with the Gold Watch—An Analysis of the
Employee Retirement Income Security Act of 1974, 17 Ariz. L. Rev. 426 (1975).
154 Some overlap exists between disclosing pension plan information under IRC
§ 6104 and the disclosure exception under IRC § 6103 (c) (1) (F) to beneficiaries
of trusts. See Duncan v. Northern Alaska Carpenters Retirement Fund, 1991
WL 165052 (W.D. Wash. 1991).
155 http://www.pbgc.gov/about/operation.html; http://www.pbgc.gov/docs/
egovrept2008.pdf
156 For a general description of political organizations, see Hopkins, supra note
147, at 411–17.
157 IRC § 6104 (a) (1) (A).
158 Pub. L. No. 106–230, 1 (b) (a) (A) (i) – (vi) amended IRC § 6104 to include
political organizations, effective July 1, 2000.
159 The information required to be made public is set out in IRC § 527 (j). For
a general discussion of the history of IRC § 527 and the history leading to
its inclusion in the list of organizations subject to the disclosure rules of
IRC § 6104, see Donald Tobin, Anonymous Speech and Section 527 of the
Internal Revenue Code, 37 Ga. L. Rev. 611 (2003). The political organization
provision of 527 came into existence in 1974 as a recognition that
organizations engaged strictly in political activity did not fit under IRC § 501,
but were also not traditional taxable entities. By 2000 these organizations
had morphed into something very different than Congress initially
envisioned.
Collecting Collected Taxes
160
143
See Buckley v. Valeo, 424 U.S. 1 (1976).
Senator Lieberman, a sponsor of the changes to IRC §§ 527 and 6104 to
permit disclosure of information about the political organizations stated:
“None of us should doubt that the proliferation of these groups with
their potential to serve as secret slush funds for candidates and parties,
their ability to run difficult to trace attack ads, and their promise
of anonymity to those seeking to spend huge amounts of money
to influence our elections poses a real and significant threat to the
integrity and fairness of our elections…”
146 Cong. Rec. S5995 (daily ed. June 28, 2000) (statement of Sen. Lieberman).
162 Some have criticized this distinction, arguing that some 501 (c) organizations
can engage in limited political activity and the inability to see the donors
of those organizations leaves the public in the same place it was before
IRC § 6104 required publication of the donors of political organizations.
See Recent Legislation: Campaign Finance Reform—Issue Advocacy
Organizations—Congress Mandates Contributions and Expenditure
Requirements for Section 527 Organizations, 114 Harv. L. Rev. 2209 (2001)
and Note, The Political Activity of Think Tanks: The Case for Mandatory
Contributor Disclosure, 115 Harv. L. Rev. 1502 (2002).
163 JCT Report, (Vol. II), at 63
164 Statement of Senator Carl T. Curtis, Cong. Rec. S15646 (Daily Ed. December
4, 1969)
165 The language of 7501 describes what can fairly be described as a public trust
in function but it does not lay out the terms of that trust. The Supreme
Court tried to do that in Begier v. IRS, 496 U.S. 53 (1990). The Court sought
to describe the res of the trust created under IRC § 7501 where the monies
paid to the IRS for the collected taxes came from the general account of
the entity that collected the tax rather than from a specifically designated
trust account. This inquiry was important because the payment to the IRS
came less than 90 days before bankruptcy. If the payment represented trust
funds held for the United States then the payment would not get pulled back
into the bankruptcy estate under the preference rules. On the other hand,
if the payment came from the taxpayer’s money rather than a trust, then a
preference payment would exist. The Court held the receipt of the collected
taxes created the trust res at the moment of payment. The fact that the funds
were held in the corporation’s general account did not destroy the trust res
and payment of the money to the IRS for purposes of satisfying the collected
tax obligation identified the trust res. Therefore, it held that the payment to
the IRS was not a preferential payment.
161
144
Fogg
As mentioned above, the Internal Revenue Code does specifically
establish eleven trust funds in IRC § 9501 through 9511 that definitely fit the
description of public trusts. One of these trusts is the Black Lung disability
Trust, described above. Three of these trusts are funded in whole or in
part with collected taxes—the Airport and Airway Trust Fund of IRC 9502;
the Highway Trust Fund of IRC 9503; and the Sports Fish Restoration
and Boating Trust Fund of IRC 9504. These public trusts are managed by
Treasury’s Office of Public Debt Accounting—Trust Funds Management
Branch which maintains a website where it discloses the management
of these funds. The fact that some collected taxes end up in public trust
managed by the Treasury Department, some taxes with Social Security and
some taxes go into the general fund of the Treasury Department does not
change the fact that entities collecting this tax hold it in trust as described in
IRC § 7501 and in Begier.
166 IRC § 7501 (a).
167 See IRC § 31 (a) (1), supra note 61.
168 Id.
169 For a general overview of private debt collection see Gary Guenther, CRS
Updates Report on Private Debt Collection Program, 2007 TNT 236–21.
170 See Hearing on the Internal Revenue Service’s Use of Private Debt Collection
Companies to Collect Federal Income Taxes: Hearing Before the H. Comm.
on Ways & Means, 110th Cong. 43 (2007) (statement of Nina E. Olson, the
National Taxpayer Advocate). See also Internal Revenue Service Budget for
FY 2008: Hearing Before the H. Appropriations Subcomm. on Financial Serv.
& General Government, 110th Cong. (2007) (statement submitted by Colleen
Kelley, President of the National Treasury Employees’ Union); Use of Private
Collection Agencies to Improve IRS Debt Collection: Hearing Before the H.
Subcomm. on Oversight of the Comm. on Ways & Means, 108th Cong. 21 (2003)
(statement of Earl Pomeroy, Member, H. Comm. on Ways and Means).
171 Guenther, supra note 169 (stating that “all revenue collected through the
efforts of PCAs has to go into a revolving fund. PCAs are not allowed to
receive or process any of this money; only the IRS can do so. The IRS may
use up to 25 percent of the money in the fund to compensate PCAs for their
services—though IRC section 6306 offers no guidance on the factors the IRS
should consider in compensating a PCA for its services. In addition, the
IRS may transfer up to 25 percent of the revenue in the revolving fund to its
budget for tax law enforcement”).
172 David Lawder, U.S. IRS to End Contracts with Private Tax Debt Collectors,
Reuters (Mar. 2009), available at http://www.reuters.com/article/
idUSN0536345520090306, quoting Sen. Richard Durbin, “Until private
Collecting Collected Taxes
145
debt collectors can prove they can do the job … more efficiently at a lower
cost than the IRS, there is no reason we should continue this program.”
Senator Durbin agreed with the IRS decision not to renew contracts with the
private tax debt collectors, arguing that tax collection is a “core government
function.”
173 See Statement of Deputy Commissioner Linda Stiff, supra note 60.
174 JCT Report (Vol. II) at 64.
175 See infra note 226; U.S. Gov.’t Accountability Office, GAO–07–742T, Tax
Compliance Thousands of Federal Contractors Abuse the Federal Tax System,
at 3–4; U.S. Gov’t Accountability Office, GAO–05–637, Financial
Management: Thousands of Civilian Agency Contractors Abuse the Federal Tax
System with Little Consequence, 2 (2005).
176 Ark. Code Ann. § 26–18–303 (b) (18) (West 2010) (“For the purpose of the
timely and accurate collection of local sales and use tax and state income
tax withholding for employees, disclosure of the name and address of a
taxpayer that has failed three (3) times within any consecutive twentyfour-month period to either report or remit state or local gross receipts or
compensating use tax or state income tax withholding for employees and
has been served with a business closure order under § 26–18–1001 et seq.”);
Fla. Stat. § 213.053 (8) (d) (2010) (“the department may provide … [n]ames,
addresses, and sales tax registration information, and information relating
to a hotel or restaurant having an outstanding tax warrant, notice of lien, or
judgment lien certificate, to the Division of Hotels and Restaurants of the
Department of Business and Professional Regulation in the conduct of its
official duties.”); Ind. Code § 6–8.1–7–1 (n) (2010); Mass. Gen. Laws Ch.
62C §21 (b) (3); Nev. Rev. Stat. § 366.160 (1) (“All records of mileage operated,
origin and destination points within Nevada, equipment operated in this
state, gallons or cubic feet consumed, and tax paid must at all reasonable
times be open to the public.” ) ; Utah Code Ann. § 59–1–403 (3) (e) (“[A] t
the request of any person, the commission shall provide that person sales
and purchase volume data reported to the commission on a report, return,
or other information filed with the commission under … Motor Fuel or …
Aviation Fuel ” ).
177 1923 Wis. Laws 39.
178 1953 Wis. Laws 303.
179 Wis. Stat. Ann. § 71.78 (2) (West 2010).
180 32 V.S.A. § 3102 (providing that “the commissioner shall disclose a return
or return information … to any person who inquires, provided that the
information is limited to whether a person is registered to collect Vermont
income withholding, sales and use, or meals and rooms tax; whether a person
146
Fogg
is in good standing with respect to the payment of these taxes; whether a
person is authorized to buy or sell property free of tax; or whether a person
holds a valid license…”). The practical explanation of Vermont’s application
of this provision is based on a conversation between the author and Molly
Bachman, Vermont’s General Counsel for the Tax Department. Telephone
conversation with Molly Bachman, Vermont’s General Counsel for the Tax
Department (Aug. 18, 2010).
181 Mass. Gen. L. ch. 62C, § 83 (c) (1992); currently Mass. Gen. Laws Ann. Ch.
62C, § 83 (c) (West 2010).
182 Id.
183 Not only is there little data that evidences any negative impact, but there is
little data concerning the beneficial effects of the disclosure. See Richard
Pomp, The Disclosure of State Corporate Income Tax Data: Turning the Clock
Back to the Future, 22 Cap. U. L. Rev. 373 (discussing benefits of disclosure at
state level).
184 Current employment tax returns do reflect the liability of the employer
portion of the social security tax. As discussed below, this Article
recommends removing that section from returns reporting collected taxes.
185 In some ways the debate on disclosure of tax information has become less
important since 1976 when the last great debate occurred. Tax information
no longer exists as the single greatest source of information about an
individual or an entity. Tax information has been replaced by a host of other
information sources including but not limited to the Bank Secrecy Act, the
SEC rules and other broad rules seeking transparency in corporate affairs.
Interestingly, the IRS even uses third party data gathering sources such as
ChoicePoint which is built upon publicly available data as it tries to gather
information about taxpayers. The IRS’ use of this information provider
serves as a poignant statement of where much information lies today—it lies
in a wide array of public venues available to those who know how to mine
such data. Additionally, other rich sources of information about individuals
and entities exist in the public domain, provided by the federal government
through such sources as PACER, which provides public information on
individuals filing bankruptcy or other court proceedings. Again, far more
data about an individual can readily be accessed electronically through
PACER than is found on the individual’s income tax return.
186 While disclosure policy provides no basis for distinguishing among different
taxpayers whose information will be disclosed, collection policy with respect
to collected taxes suggests that the most likely taxpayers to fail to pay over
collected taxes are small and newly formed businesses. With this in mind,
an alternate proposal, discussed below, addresses the disclosure of collected
Collecting Collected Taxes
147
tax returns for certain smaller entities or entities that have experienced
difficulties fulfilling their collected tax obligations.
187 Take, for example, the Form 941 which reports three different types of
information: withheld income taxes (trust information); withheld social
security taxes (trust information) and the employer’s portion of the social
security taxes (not trust information). Similarly, Form 720 sets out a
reporting mechanism for a variety of excise taxes, some of which result from
a trust relationship where the entity filing the Form 720 has collected the
excise taxes from third parties and some of which result from excise taxes
directly imposed on the entity.
188 This article focuses on policy reasons for changing the disclosure laws to
provide for disclosure of the returns reporting collected taxes. For a discussion
of these policy reasons, see supra notes 174–186 and accompanying text.
189 See following discussion for what should be included on the collected
tax return.
190 IRS Form 941 and instructions.
191 Form 5500 used for returns of pensions requires extensive information;
however, the information sought on the Form 5500 seems irrelevant to the
information that would make the Form 941T and Form 720T useful.
192 As discussed above, the Form 990–BL concerns the type of tax most closely
related to collected taxes of all of the returns made public pursuant to IRC
6104 at present.
193 See JCT Report (Vol. II), at 65 and accompanying footnotes. See also Rizek,
supra note 73, at 88–90 for a pragmatic view that may represent the only
realistic point of view on this subject in the absence of credible supporting
data for either point.
194 IRM 1.15.19, 1.15.35 (available at http://www.irs.gov/irm/part1/irm_01–015–019.
html; http://www.irs.gov/irm/part1/irm_01–015–035.html.) The Combined
Annual Wage Reporting (CAWR) falls under the division of Small Business/
Self-Employed (SB/SE). CAWR ensures that employers accurately report
annual wage data on IR Forms in the 940 series to IRS and Form W–3 to
Social Security Administration (SSA). When there is a discrepancy between
the two forms, a case is created and worked within the SB/SE campuses. The
CAWR system consists of five Tier 1 sub-projects maintained by National
Office Modernization and Information Technology Services (MITS) and one
Tier 2 system maintained by Ogden Development Center MITS. CAWR
runs on both the Tier 1 IBM platform and on the Teir 2 Sun platform. The
Tier 1 processing is known as Combined Annual Wage Reporting Mainframe
(CAWR MAINFRAME). The Tier 2 processing is known as the Combined
Annual Wage Reporting Automation Program (CAP).
148
Fogg
The CAP system houses the CAWR for cases for a three year period,
it allows notice/letter generation and user updates, monitors cases for
responses/no responses etc. and creates reports.
195 See Lipsig v. United States, 187 F. Supp. 826. (E.D.N.Y. 1960); Michael I.
Saltzman, IRS Practice and Procedure, ¶17.09[4] (Warren Gorham
Lamont, 2nd ed. 2002).
196 Publishing all returns of collected taxes, as recommended in this Article,
does go further in disseminating information than allowed under
IRC § 6103 (k) (2). In some ways such disclosure mirrors the disclosure
exception in IRC § 6103 (e) (1) (F) which permits disclosure of information
to trust beneficiaries. Here the disclosure of information benefits the
beneficiaries of the trust on collected taxes created under IRC 7501. The
beneficiaries are the people of the United States. The exception to disclosure
concerning the NFTL is discussed in notes 141–48 and accompanying text.
197 Collected taxes fall outside the deficiency tax procedure of IRC 6213. When
collected taxes go unpaid, the IRS can, if it has not already done so based on
a return with insufficient remittance, assess the taxes due on the collected tax
return and almost immediately begin collection. Because these taxes can go
immediately or almost immediately into the collection stream, the federal tax
lien exists once the liability goes unpaid. The existence of the federal tax lien
occurs when a federal tax assessment has taken place, followed by notice and
demand pursuant to IRC 6303, followed by 10 days (the usual period the IRS
gives taxpayers to pay as a policy matter) in which the taxes remain unpaid.
If this sequence occurs, a federal tax lien exists as described in IRC sections
6321 and 6322. If a federal tax lien exists, then the IRS can make the liability
public when it wants by filing an NFTL pursuant to IRC § 6323 (f). The
publication of the liability to the world through the NFTL represents one of
the many exceptions promulgated in IRC 6103. See IRC § 6103 (k) (2). Since
the collected tax liabilities in almost all instances fit this disclosure exception,
publishing these liabilities presents few hurdles from a disclosure perspective
if a liability exists.
198 Based on correspondence to clients of the Villanova Federal Tax Clinic for
whom federal tax liens are filed, a number of business organizations track
federal tax lien filings in order to offer taxpayers assistance in working out
their debts with the IRS.
199 IRC § 6323 (f) (1) (A).
200 IRC § 6323 (f) (1) (A); IRM 5.12.2.8 (“The principal executive office is deemed
to be the residence of the corporation or partnership. It is the place where
the major management decisions are made. Do not confuse the principal
executive office with the principal place of business”).
Collecting Collected Taxes
201
149
As discussed in note 197, supra, assessment triggers issuance of the notice
and demand letter under IRC 6303 giving the taxpayer 10 days to pay. If
the taxpayer does not pay within the 10 days, the assessment lien arises
automatically. Once the assessment lien exists, it is up to the IRS to decide
when to make that lien public with the filing of an NFTL.
202 Section 3802 of the Revenue Reform Act of 1998 directed the Joint
Committee on Taxation and the Treasury Department to comment on the
feasibility of shaming among many other disclosure issues. The JCT Report
addresses shaming, recommending against federal shaming program for
non-filers and expresses concern that publishing non-filer information might
incorrectly identify individuals with no filing requirement. See JCT Report
(Vol. I), at 238–40. As of 2000 only five jurisdictions had adopted shaming
provisions. Contrast that number with the twenty-six states and the District
of Columbia that now use shaming, listed in Appendix A.
203 See Blank, supra note 132, at 539, 547–48.
204 Id.
205 See Appendix A.
206 JCT Report (Vol. I), at 238.
207 The failure to pay collected taxes creates a competitive advantage for the
company that fails to pay over related companies that do pay these taxes.
This competitive advantage creates a strong reason for publishing this
information. If competitors learn of the failure to pay, they may be able to
publicize that fact and potentially remove the advantage. Some discussion
of the competitive advantage has surfaced although little has been written on
the scope of this advantage.
208 The disclosure under IRC § 6039G is the disclosure of the taxpayer’s name.
Although the filing of the IRC § 6039G information return acts as the
triggering mechanism for the disclosure, the disclosure itself simply consists
of the listing of the taxpayer’s name with no identifying tax information.
In this respect, the IRC § 6039G disclosure differs from other disclosure
exceptions described in IRC 6103, 6104 or 6110.
209 IRC § 6039G provides that any individual to whom section 877 (b) or 877A
applies for any taxable year shall provide a statement for such taxable year
which includes the following information: (1) the taxpayer’s TIN, (2) the
mailing address of such individual’s principal foreign residence, (3) the
foreign country in which such individual is residing, (4) the foreign country
of which such individual is a citizen, (5) information detailing the income,
assets, and liabilities of such individual, (6) the number of days during any
portion of which that the individual was physically present in the United
States during the taxable year, and (7) such other information as the Secretary
150
Fogg
may prescribe. The statute also provides that an individual who is required
to file a statement under subsection (a) for any taxable year, and fails to
file such a statement, fails to include all required information, or includes
incorrect information, must pay a penalty of $10,000 unless it is shown that
such failure is due to reasonable cause and not to willful neglect. Finally,
the statute provides that any Federal agency or court which collects the
statement under subsection (a) shall provide to the Secretary a copy of any
such statement, and the name (and any other identifying information) of
any individual refusing to comply with the provisions of subsection (a). The
Secretary of State shall provide to the Secretary a copy of each certificate as
to the loss of American nationality under section 358 of the Immigration and
Nationality Act which is approved by the Secretary of State, and the Federal
agency primarily responsible for administering the immigration laws shall
provide to the Secretary the name of each lawful permanent resident of the
United States whose status has been revoked or abandoned. No later than 30
days after the close of each calendar quarter, the Secretary shall publish in the
Federal Register the name of each individual losing United States citizenship
with respect to whom the Secretary receives information under the preceding
sentence during such quarter.
210 See Michael S. Kirsch, Alternative Sanctions and the Federal Tax Law:
Symbols, Shaming, and Social Norm Management As a Substitute For Effective
Tax Policy, 89 Iowa L. Rev. 863 (2004) for a detailed discussion of this law
and of the policies behind the law as well as the shortcomings of the law.
211 The remedy also appears ineffective, as more and more American citizens
renounce their citizenships in order to avoid this taxation. See Ellen
Kelleher, Americans Forfeit Citizenship to Avoid Tax, Financial Times, July
17, 2010, available at http://www.ft.com/cms/s/0/bab42a32–9126–11df-b297–
00144feab49a.html.
212 Kirsch, supra at note 210, at 888 (discussing the effectiveness of shaming
sanctions).
213 The IRS and the Department of Justice use a form of shaming in some of
their information releases and website postings. The IRS publishes a “dirty
dozen” list of transactions it finds abhorrent and contrary to the law. The list
serves both to “shame” the promoters and investors in the promotion as well
as to inform prospective investors of the toxic tax nature of the transaction.
Internal Revenue Service, Beware of IRS’ “Dirty Dozen” Tax Scams, http://
www.irs.gov/newsroom/article/0,,id=220238,00.html. Similarly, but in
less of a shaming mode, the IRS publishes “listed transactions” in an effort
to let people know that certain transactions have gain the attention of the
IRS in such a way that settlement of the cases is no longer an option. The
listing of a transaction serves to shame those engaged in that transaction
Collecting Collected Taxes
151
although not by name as well as to inform. Internal Revenue Service,
Recognized Abusive and Listed Transactions, http://www.irs.gov/businesses/
corporations/article/0,,id=120633,00.html. The Service sometimes back-ends
the shaming provisions on these transactions by requiring a disclosure waiver
in settlements it reaches with taxpayers engaged in such transactions so it
can publicize the concession by the offending taxpayer. See Blank, supra note
132, at 82–85. The Department of Justice regularly publicizes the convictions
that it obtains and the civil injunctions that it obtains in promoter and return
preparer cases. See, e.g. Press Release, U.S. Department of Justice, Cincinnati
Area Return Preparer Pleads Guilty to Tax Crimes (June 8, 2010) (available at
http://www.justice.gov/tax/txdv10671.htm); Press Release, U.S. Department
of Justice, Federal Jury Convicts Local Tax Preparer—Faces up to 33 Years
in Federal Prison (Mar. 12, 2010) (available at http://www.justice.gov/usao/
txn/PressRel10/watson_tax_convict_pr.html). The publication of the name
of the person convicted or enjoined serves not only to shame the individual
so named but to deter others who might engage in similar behavior. Here,
the shaming comes after enforcement so the shaming does not motivate the
convicted or enjoined individual to change their behavior. The enforcement
activity, hopefully, accomplishes that purpose.
214 Jumpstart Our Business Strength (JOBS) Act, S. 1637, 108th Cong. 402, 150
Cong. Rec. S. 5622, 5643 (May 18, 2004). For a detailed discussion of this
provision, see Blank, supra note 132, at 553 & n.74. Blank argues that shaming
corporations that use tax shelters would not promote tax compliance for a
variety of reasons. In many ways the proposal to shame corporations in this
context carries many of the symbolic but ineffective concerns expressed by
Kirsch about expatriate shaming. Kirsch, supra note 210, at 921. Congress
feels a need to express displeasure about certain behavior but chooses to
make its expression in a manner that does not affect future behavior in the
manner in which it seeks.
215 See Appendix A for a listing of states with shaming laws and, where used,
their websites. All of the state shaming provisions focus on collection of
unpaid taxes rather than corporate shaming.
216 See Dennis J. Ventry, Jr., Cooperative Tax Regulations, 41 Conn. L. Rev.
431 (2009); Maryann Richardson and Adrian Sawyer, A Taxonomy of the
Tax Compliance Literature: Further Finings, Problems and Perspectives, 16
Australia Tax F. 137, 168 (2001); Leandra Lederman, The Interplay Between
Norms and Enforcement in Tax Compliance, 64 Ohio St. L. J. 1453, 1493
(2003); Susan Cleary Morse, Using Salience and Influence to Narrow the Tax
Gap, 40 Loyola University Chicago Law Journal 483 (2009); Dan Kahan,
The Logic of Reciprocity, 102 Mich. L. R. 71 (2003); Marjorie Kornhauser, Tax
Compliance and the Education of John (and Jane) Q. Taxpayer, 121 Tax Notes
152
Fogg
737 (Nov. 10, 2008); Joshua Rosenberg, Narrowing the Tax Gap: Behavioral
Options, 117 Tax Notes 517 (Oct. 29, 2007); Jay Soled and Dennis Ventry, Jr., A
Little Shame Might Just Deter Tax Cheaters, USA Today April 10, 2008.
217 Toni M. Massaro, Shame, Culture, and American Criminal Law, 89 Mich.
L. Rev. 1880 (1991); Dan M. Kahan & Eric A. Posner, Shaming White Collar
Criminals: A Proposal for Reform of the Federal Sentencing Guidelines, 42
J.L. & Econ. 365 (1999); John B. Owens, Have We No Shame?; Thoughts on
Shaming, “White Collar” Criminals, and the Federal Sentencing Guidelines,
49 Am. U. L. Rev. 1047 ( 2000); James Q. Whitman, What is Wrong with
Inflicting Shame Sanctions?, 107 Yale L.J. 1055 (1998). For more recent
discussion continuing this debate see Dan Kahan, What’s Really Wrong with
Shaming Sanctions, 84 Tex. L. Rev. 2015 (2006) and the articles cited therein.
218 Massaro, supra note 217, at 1883.
219 One concern with shaming provisions is that shaming not publicize a general
failure of society to comply with the tax law. Shaming should not cause
less compliance by alerting the compliant to the fact they may constitute a
disadvantaged minority of individuals complying with present laws. This
circumstance graphically displayed itself in bankruptcy courts around
the county in the 1980s and 1990s as the IRS and Department of Justice
sought to hold up plan confirmation of individuals who had not filed their
tax returns. It did so by objecting to every chapter 13 plan in which the
debtor had outstanding tax returns. The Bankruptcy Judge in Richmond,
Virginia before whom the author practiced initially took the time to publicly
berate each chapter 13 debtor coming before him who failed to file their tax
returns explaining to the individual how the failure to file the tax returns
was a federal crime for which the individual could be sent to jail, etc. After
seeing these motions in case after case, the judge eventually gave up on the
failure to file return lecture almost undoubtedly after realizing the extent
of the problem and the lack of effect his lectures were having. The problem
eventually led to changes in the bankruptcy law in 2005 theoretically
preventing debtors from moving forward in chapter 13 cases without the
submission of the prior four years returns. 11 U.S.C.A. § 1308.
220 The IRS engages in some publication that could be classified as shaming as
it publicizes the “dirty dozen” most offensive tax shelters which plays the
dual role of shaming the transaction and warning people away from the
transaction. The IRS listed transactions could be viewed as a similar type of
shaming as is the IRS publication of certain settlements with corporations
engaged in tax shelters. See Blank, supra note 132, at 554. The Department of
Justice regularly publicizes the names of individuals it successfully prosecutes
or whom it successfully enjoins from promoting tax shelters or improperly
preparing tax returns. Supra note 213.
Collecting Collected Taxes
153
221
Kirsch, supra note 210, at 908–12.
Massaro, supra note 217, at 1930–32.
223 Kirsch, supra note 210, at 889–90.
222
224
The manner in which states publish the names of the individuals and entities
provides a good insight into effective use of publication of non-payment.
Some states, such as Wisconsin, create an easy to use link right on the front
page of their website. This model makes it quite easy to locate entities that
fail to pay. Other states bury the listing of names well into the website
making it very difficult, if not impossible to locate the names. For the same
reason that publication only in the Federal Register does not make much
sense in this context, neither does publication on a website that is relatively
inaccessible.
225 The uncertainty of the liability created one of the concerns expressed by
Joshua Blank in his article. See Blank, note 132, at 544. With corporate
tax shelters, the Government may believe that the claims abuse the tax
code but until case law settles the issue, the alleged abuse lacks certainty.
Uncertainty is also one of the problems with the publication of the names
of the expatriates since the list sweeps up all expatriates and even if it were
targeting only those who left for tax motivated reasons, it would be difficult
to determine those situations in which the tax motive was the sole or primary
reason for renouncing citizenship. None of that uncertainty exists with
unpaid collected taxes. The liability is almost always a certainty usually
stemming from self-assessment but even when it results from adjustments by
the IRS the dollar amount of the assessment is rarely at issue.
226 The failure of federal contractors to pay their collected taxes was the subject
of a GAO report “Thousands of Federal Contractors Abuse the Federal
Tax System” GAO–07–742T, April 19, 2007. This report not only found
that entities contracting with the United States owed billions of dollars
in unpaid employment taxes but determined that the United States had
not previously requested information that would allow it to factor such
behavior into its decision making process. As a result of this GAO report,
the Federal Government proposed to revise the information that contractors
must disclose as they seek to contract with the Federal Government. This
caused proposed changes to the Federal Acquisition Regulation (FAR)—
Representations and Certifications—Tax Delinquency, 72 Fed. Reg. 15093
(proposed Mar. 30, 2007). The GAO report represents a clear example of
how knowledge of the failure to pay collected taxes impacts a potential
customer. With that type of customer reaction, one would expect that in
the area of federal contractors the incidence of failure to pay collected taxes
should significantly decrease.
154
Fogg
This GAO report was one of several on a similar theme. A follow up
report was issued later in 2007. U.S. Gov’t Accountability Office, GAO–
07–563, Thousands of Organizations Exempt from Federal Income Tax Owe
Nearly $1 Billion in Payroll and Other Taxes (June 2007). This report shows
how the failure to pay collected taxes could impact charitable organizations
and the entities making donations to those organizations. This is yet
another example of how knowledge of the failure to pay the collected
taxes could impact behavior. See, e.g., Farah Stockman, Shell companies in
Cayman Islands allow KBR to avoid Medicare, Social Security Deductions,
The Boston Globe, March 6, 2008, available at http://www.boston.com/
news/world/articles/2008/03/06/top_iraq_contractor_skirts_us_taxes_
offshore/ (“Payroll taxes can be a significant cost, he said, speaking on the
condition of anonymity. If you are bidding against [rival construction
firms] Fluor and Bechtel, it might give you a competitive advantage.”)
The issue in this article is not so much Brown & Roots’ failure to pay
employment taxes as its setting up a foreign entity to employ individuals
in a manner in which it would have no employment tax obligation
whatsoever; see also U.S. Gov.’t Accountability Office, GAO–07–742T, Tax
Compliance Thousands of Federal Contractors Abuse the Federal Tax System,
at 3–4 (“[F]or wage-based businesses that provide goods and services,
federal contractors with unpaid federal taxes have an unfair advantage in
price competition when competing against other businesses for federal
contracts. Companies that do not pay their payroll tax, which is typically
over 15 percent of the employees’ wages, would have a significantly lower
costs advantage and therefore have a substantive competitive advantage
over similarly situation businesses that pay their taxes. For example, we
identified instances in which companies that had unpaid payroll taxes were
competitively awarded contracts over companies that had paid their
federal taxes”).
227 The author knows of one situation in which knowledge that the entity had
outstanding collected tax obligations had a direct impact on a potential
customer’s decision and drove the customer away. The potential customer
was the IRS. The IRS sought to contract with a hotel in which it would hold
a continuing professional education conference for its employees in one state.
The contracting officer chose a hotel that had a longstanding problem with
the payment of its collected taxes. When the revenue officers knowledgeable
about the outstanding taxes learned of the potential contract with the
hotel, they became quite vocal about how improper contracting with that
hotel would be. Their voices were heard and another location was selected.
Perhaps this example is extreme because of the close nature between the
potential customer and the unpaid collected taxes; however, it is not hard
to imagine other circumstances in which a potential customer would make
Collecting Collected Taxes
155
a decision not to contact with an entity that did not pay its collected taxes.
Indeed, the hope in publicizing this information is to assist in creating a
culture in which not paying these taxes makes the entity somewhat of a
pariah and causes entities in general to want to pay these taxes in order to
avoid the stigma that would come from failure to pay.
228 While slightly different in its factual underpinnings, the actions of Kellogg,
Brown & Root (KBR) with respect to its workers in Iraq provides some
insight into how information can impact customer and competitor decisions.
Based on the information provided in an article in The Boston Globe
on March 6, 2008, by Farah Stockman, KBR apparently avoided paying
employment taxes altogether with respect to approximately 20,000 employees
it had in Iraq by treating the individuals as employees of a Cayman Island
subsidiary. KBR’s customer, the Defense Department, knew “since at least
2004 that KBR was avoiding taxes by declaring its American workers as
employees of Cayman Islands shell companies, and officials said the more
allowed KBR to perform the work more cheaply, saving Defense dollars.”
The reaction of KBR’s customer is somewhat surprising because of the
overall losses to the United States and its citizens from the employment tax
maneuver executed by KBR but at least it shows a reaction from a customer
aware of the situation. A former executive at Halliburton, the parent of KBR,
said “Payroll taxes can be a significant cost, … speaking on the condition of
anonymity. ‘If you are bidding against [rival construction firms] Fluor and
Bechtel, it might give you a competitive advantage.’ ” The article did not
contain statements from the competitors but one can imagine what they
might say. Farah Stockman, Shell companies in Cayman Islands allow KBR
to avoid Medicare, Social Security Deductions, The Boston Globe, March 6,
2008, available at http://www.boston.com/news/world/articles/2008/03/06/
top_iraq_contractor_skirts_us_taxes_offshore/.
229 See Appendix A for a list of states that have shaming provisions.
230 Compare the currently list of states engaged in shaming from Appendix A
with the five states that had adopted this practice in 1999 at the time the Joint
Committee on Taxation report to Congress was prepared in 2000. See pp—
of that report; see also U.S. Gov’t Accountability Office, GOA–GDD 99–164,
Few State and Local Government Publicly Disclose Delinquent Taxpayers. Like
the Joint Committee Report, this GAO report was ordered by Congress as
a result of the Revenue Reform Act of 1998, Section 3802. While the states
felt the disclosure of delinquent taxpayers was aiding in the collection of
outstanding taxes, no studies quantified the impact of the disclosure.
231 Several states have provisions that disclose the greatest delinquent accounts:
California, Delaware, Rhode Island. See Appendix A. Several other states
156
Fogg
have provisions that disclose accounts exceeding a certain dollar amount:
Colorado, Illinois, Indiana, Massachusetts, Wisconsin. See Appendix A.
232 The IRS can file an NFTL against any taxpayer with an assessed liability
which is unpaid. Upon assessment of a tax, the IRS computer searches a
taxpayer’s account for credits with which to satisfy the assessed liability.
If insufficient credits exist on the account, the IRS sends the taxpayer a
notice and demand letter pursuant to IRC 6303 demanding payment of the
outstanding liability within 10 days. If payment is not forthcoming within
the 10 day period, IRC 6321 and 6322 cause the creation of a lien against all
of the taxpayer’s property and rights to property. This lien, known only to
the taxpayer and the IRS, is sometimes called the secret lien or assessment
lien. In order for this secret lien to defeat certain creditors described in IRC
6323 (a), the IRS must file a public notice of the lien pursuant to IRC 6323 (f).
That notice is available to the world. The filing of an NFTL has serious
consequences for credit and financial well-being.
233 R.I. Gen. Laws § 44–1–34 (2010). Rhode Island’s Division of Taxation
website lists the top 100 delinquent taxpayers which includes all types of
state tax delinquencies, including personal, sales, withholding, corporate and
inheritance taxes.
234 Although articulated almost solely on the unproven aspect of the success of
shaming, the Joint Committee on Taxation reached the same conclusion in
its 2000 report. See JCT Report (Vol. I), at 238–40. At the time of that report
only five states had shaming laws. Obviously, the allure of shaming to states
has grown since that time. Because of the difficulty of separating the positive
effect that shaming has on compliance from other causes, the empirical case
for shaming still lacks a strong underpinning. The concerns voiced by the
Joint Committee and others as cited above, still raise a cautionary flag to this
approach. It also has some disconnects with the policy reasons underlying
disclosure unless you view the shaming provisions solely as an extension of
the lien filing as discussed further below.
235 Shaming serves as an unlikely deterrent to those setting out to cheat. For
those persons strong enforcement measures must deter.
236 The stress of these types of situations also leads to the loss of relationships.
Financial difficulties of the type encountered by those running failing
business frequently lead to the dissolution of marriages which further serves
to drag individuals in this circumstance down a financial and emotional hole.
In this situation shaming will not cause the person to pay over the money. It
simply puts more fuel on the fire of a life situation going up in flames.
237 The publication of returns of collected taxes comes at this early stage and
would seem a much more effective mechanism for effecting behavior of
Collecting Collected Taxes
157
those making decisions about this money than the much later publications of
shaming lists.
238 Not everyone would agree with this point. See Blank, supra note 132, at 540.
Here, it serves merely as an illustration in contrast.
239
Ark. Code Ann. § 26–18–303 (b) (18) (“For the purpose of the timely
and accurate collection of local sales and use tax and state income tax
withholding for employees, disclosure of the name and address of a taxpayer
that has failed three (3) times within any consecutive twenty-four-month
period to either report or remit state or local gross receipts or compensating
use tax or state income tax withholding for employees and has been served
with a business closure order under § 26–18–1001 et seq.”) See Arkansas
Department of Finance & Administration Revenue Division, Sales Tax
Business Closures Update, State Revenue Tax Quarterly, Volume XI, No. 1
(2005), at 3–4 (describing this provision with respect to sales taxes).
240 There are some distinctions concerning the publication of pension plan
information which leaves out some of the information of the smaller plans in
an apparent recognition that the smaller plans do not raise the same overall
concerns as the large ones. IRM 11.3.10.3. (“Documents relating to plans
with 25 or fewer participants are available only to plan participants, the plan
sponsor, or their authorized representatives.”)
241 As discussed above in notes 128–31 and accompanying text, the IRS can
decide to make public the outstanding liability on any collected tax by simply
filing an NFTL. IRC §§ 6323 (f), 6103 (k) (2). Filing an NFTL notifies the
“world” that a taxpayer has an unpaid federal tax liability. Because credit
reporting agencies almost always search for filings of the NFTL, these filings
generally have significant negative consequences to the taxpayers against
whom the liens are filed. See 2010 NTA Annual Report, at 54 (discussing
effect of filing an NFTL and urging for more measured approach to filing
of NFTL). Despite that fact that the “world” knows about the lien when
the NFTL occurs, many people do not know because of where lien filing
occurs. Section 6323 (f) requires filing of the notice in the place where
the taxpayer resides in order to perfect the lien as to personalty and in the
location of any real property with respect to such property. Unless one
frequents courthouses or their online databases, where available, knowledge
of the filing of the NFTL would require some searching. Public knowledge
would come easier if a national tax lien registry were adopted. T. Keith Fogg,
National Tax Lien Registry, 120 Tax Notes 783 (August 25, 2008). Still, even
a national registry would lump all types of taxes together not highlighting
collected taxes. Some states take the position that the existence of a
published lien allows them to highlight liabilities in their shaming websites.
E-mail from VA Tax Customer Service, to Fleming Ware, Research Assistant,
158
Fogg
Villanova University School of Law (July 9, 2009, 09:19 EST) (on file with
author). The IRS could not take that approach because of the uncertain
state of the law regarding the public records exception. The Circuits have
split on the issue of whether allowable public disclosure of information in
one setting allows publication of that same information by the IRS in other
settings. See JCT Report (Vol. I), at 70–81 (citing Lampert v. United States,
854 F.2d 335, 338 (9th Cir. 1988), cert. denied, 490 U.S. 1034 (1989) (holding
that “if a taxpayer’s return is lawfully disclosed in a judicial proceeding … [t]
he information is no longer confidential and may be disclosed again without
regard to section 6103”); Rowley v. United States, 76 F.3d 796 (6th Cir. 1996)
(holding that once return information becomes public through filing and
recording of judicial lien, it is no longer confidential); Mallas v. United States,
993 F.2d 1111 (4th Cir. 1993) (holding that the United States is liable when is
discloses return information that was previously made part ofpublic records)).
242 The debate over the public disclosure exception seems like a “small” policy
issue of the format and procedure for disclosure rather than the larger policy
decision of whether to disclose.
243 See Cheng, supra note 53 (“[t]he use of structure to encourage tax compliance
has been an unqualified success”); see also Lederman, supra note 53 (arguing
that substantive federal income tax law can, and often does, foster compliance
by harnessing the structural incentives of third parties).
244 Id. at 657; see also Thaler & Sunstein, supra note 53 (arguing a similar
approach but from the perspective of behavioral psychologists).
245 See Cheng, supra note 53, at 659 (discussing fiat and its problems); see
also Erich Kirchler, The Economic Psychology of Tax Behavior
(Cambridge 2007) (discussing why such structures may be needed given
taxpayer perceptions and attitudes). Non-compliant taxpayers who owed
VAT were younger, more egotistical, and less knowledgeable about VAT.
Many thought money received from others in payment of VAT was coming
from their own business and did not appreciate they were holding the money
in trust. People put money in different boxes mentally and if they do so it
effects how they are willing to spend it. Most people surveyed believed that
VAT money was theirs. Id. at 164–65.
246 See Cheng, supra note 53, at 662 (arguing that structural laws offer a more
effective alternative for influencing everyday behavior than statutory
prohibitions).
247 Id. at 664.
248 Id.
249 Perhaps the greatest tax example of a structural law is the withholding
provisions that underlie the collected tax issue that is the subject of this
Collecting Collected Taxes
159
paper. The withholding tax laws require employees to have a portion of
their salary taken each pay period resulting, usually, in full payment of
their income taxes over the course of the year. Payment, and compliance,
is achieved because of the structure of the withholding provisions. The
withholding tax provisions make it difficult for employees to fail in their
tax paying obligations and provide an incentive for them to file their taxes.
See generally Piroska Soos, Self-Employed Evasion and Tax Withholding: A
Comparative Study and Analysis of the Issues, 24 U.C. Davis L. Rev. 107 (Fall
1990) (discussing the non-compliance of withholding agents).
250 See U.S. Gov’t Accountability Office, GAO–08–617, Tax Compliance:
Businesses Owe Billions in Federal Payroll Taxes 15 (2008) (finding a
significant number of businesses are not paying payroll taxes).
251 James M. Bickley, Tax Gap and Tax Enforcement, Report for Congress No.
RL33882 (2008), available at http://opencrs.com/document/RL33882/.
252 See Robert B. Cialdini, Social Motivations to Comply: Norms,
Values and Principles in 2 Taxpayer Compliance: Social Science
Perspectives 200 (Jeffrey A. Roth & John T. Scholz, eds., University of
Pennsylvania Press 1989) (1989) (establishing six sets of principles that most
powerfully and regularly influence compliance decisions: 1) be consistent
with prior commitments; 2) return an earlier fit, favor or service; 3) follow
the lead of similar others; 4) conform to the directives of legitimate authority;
5) seize opportunities that are scarce or dwindling in availability; and 6)
accede to the requests of those we like).
253 See U.S. Gov’t Accounting Office, GAO–99–256, Payroll Taxes:
Billions in Delinquent Taxes and Penalties Due but Unlikely to Be
Collected, 13 (1999) (containing anecdotal comments from IRS revenue
officers on why the collected taxes do not get paid).
254 See Margaret McKerchar, Understanding Small Business Taxpayers: Their
Sources of Information and Levels of Knowledge of Taxation, 12 Australian
Tax Forum 25 (1995) (discussing the level of knowledge of small business
taxpayers); see also Thaler, supra note 53, at 188 (discussing the importance of
providing information as a basis for achieving desired outcomes).
255 Numerous studies have shown that the longer the delay in collecting
delinquent liabilities the less likely collection becomes. See, e.g., 2004
National Taxpayer Advocate’s Annual Report to Congress 233, available at
http://www.irs.gov/advocate/article/0,,id=133967,00.html (showing success
in collection based on the number of months since the assessment was
made). But see U.S. Gov’t Accountability Office, GAO–08–617, Tax
Compliance: Businesses Owe Billions in Federal Payroll Taxes, 32–33
(2008) (finding that determining the persons responsible for the payment of
160
Fogg
tax and assessing the liability against them typically took between two and
three years); see also U.S. Gov’t Accountability Office, GAO–05–637,
Financial Management: Thousands of Civilian Agency Contractors
Abuse the Federal Tax System with Little Consequence, 2 (2005)
(discussing civilian contractor tax abuse).
256 See Leslie Book, The Poor and Tax Compliance: One Size Does Not Fit All, 51
U. Kan. L. Rev. 1145 (2003) (examining the scope of low income taxpayers’
compliance problem and why the IRS vigorously directs its compliance effort
toward low-income taxpayers).
257 See Dan Kahan, Trust, Collective Action and Law, 81 B. U. L. Rev. 333
(2001) (noting that the level of tax compliance in a community depends on
“individuals beliefs about the willingness of others to comply with tax laws:
those individuals who believe that most other individuals are complying form
the belief that paying taxes is an important moral duty, whereas those who
believe cheating is widespread conclude that evading is not a particularly
serious wrong”).
258 The group of individuals or entities that fails to pay its collected taxes is
drawn from the same group that exhibits the highest non-compliance with
reporting of taxes—the small businesses within the IRS SBSE classification.
See Lederman, supra note 216, at 1505 (2003). As Lederman discusses, the
non-compliance among this group could result from a desire to remain
competitive with others in the group who are also not fully paying their taxes
or it could result because this group consists of entrepreneurs, a self-selected
group of risk takers who would almost always be among the least compliant
taxpayers. The same factors causing this group to be non-compliant in
reporting their taxes could also drive their failure to pay over collected
taxes. At least one study suggests that non-compliance in one aspect of
paying taxes has a carryover effect to other aspects of tax compliance.
The one consideration that differs in the payment mode compared to the
reporting one is the existence of third parties. While the general notion
that third party competitors or peers may not be properly reporting their
taxes could drive the behavior of a member of this group, the interplay
with third parties exists more closely in the payment situation than the
reporting one. When deciding whether to pay collected taxes over to the
government, the individual or entity often faces a dilemma of whom to pay
when insufficient funds exists. The resolution of that dilemma often results
in payment of creditors with whom the individual or entity has a close or
reciprocal relationship that does not exist between the individual or entity
and the government. See Book, supra note 256 (examining the scope of low
income taxpayers’ compliance problem and why the IRS vigorously directs its
compliance effort toward low-income taxpayers).
Collecting Collected Taxes
259
161
See Lederman, supra note 216, at 1500 (describing three types of taxpayers: 1)
those committed to compliance; 2) those susceptible to influence (the largest
group); and 3) those committed to non-compliance). The government wants
to enact provisions with the greatest impact on the second group but it also
wants compliance actions that will keep the third group in check.
260 26 U.S.C. § 6672 (2010).
261 See Gerald P. Moran, Willfulness: The Inner Sanctum or Unnecessary Element
of Section 6672, 11 U. Tol. L. Rev. 709, 723–751 (1980) (discussing the
legislative history of section 6672). For a discussion of the legislative history
of section 6672, see infra notes 368–88 and accompanying text.
262 26 U.S.C. § 7202 (2010); This section provides “[a]ny person required under
this title to collect, account for, and pay over any tax imposed by this title
who willfully fails to collect or truthfully account for and pay over such tax
shall, in addition to other penalties provided by law, be guilty of a felony and,
upon conviction thereof, shall be fined not more than $10,000, or imprisoned
not more than 5 years, or both, together with the costs of prosecution.” IRC
§ 7202 (2010). This section has existed in its current form since 1954 with
no changes.
263 See Moran, supra note 261, at 750 (discussing the legislative history of section
6672). For further discussion of the legislative history and purpose of section
6672, see infra notes 368–88 and accompanying text.
264 The Department of Justice, Tax Division, Criminal Tax Manual (CTM)
reports that section 7202 is rarely used. See Department of Justice: Tax
Division, Criminal Tax Manual 9.03 (2009), available at http://www.usdoj.
gov/tax/readingroom/2001ctm/titlepg.htm; see also Ian Comisky et al., Tax
Fraud and Evasion ¶ 2.08 (RIA 2009) (“IRC 7202, a rarely used statute… .”);
U.S. Gov’t Accountability Office, GAO–08–617, Tax Compliance:
Businesses Owe Billions in Federal Payroll Taxes, 15 (2008). Twenty
reported cases exist: (1) United States v. Adam, 296 F.3d 327 (5th Cir. Tex.
2002); (2) United States v. Anglin, 999 F. Supp. 1378 (D. Haw. 1998); (3)
United States v. Bailey, 789 F. Supp. 788 (D. Tex. 1992), aff ’d, 996 F.2d 305 (5th
Cir. Tex. 1993); (4) United States v. Blanchard, 2007–2 U.S. Tax Cas. (CCH)
P50596 ( E.D. Mich. 2007); (5) United States v. Brennick, 134 F.3d 10 (1st Cir.
1998); (6) United States v. Cordell, 237 Fed. Appx. 998 (5th Cir. 2007); (7)
United States v. Creamer, 370 F. Supp. 2d 715 (D. Ill. 2005), vacated in part,
370 F. Supp. 2d 715 (N.D. Ill. 2005); (8) United States v. Easterday, 564 F.3d
1004 (9th Cir. 2009); (9) United States v. Ellis, 2007 U.S. Dist. LEXIS 58453
(D. Ind. 2007); (10) United States v. Evangelista, 122 F.3d 112 (2d Cir. 1997),
aff ’d, Evangelista v. Ashcroft, 359 F.3d 145 (2d Cir. N.Y. 2004); (11) United
States v. Gilbert, 266 F.3d 1180 (9th Cir. 2001); (12) United States v. Goins,
593 F.2d 88 (8th Cir. 1979); (13) United States v. Gollapudi, 947 F. Supp. 768
162
Fogg
(D.N.J. 1996), aff ’d, 130 F.3d 66 (3d Cir. N.J. (1997); (14) United States v.
Hovind, 305 Fed. Appx. 615 (11th Cir. 2008); (15) United States v. Lombardo,
281 Fed. Appx. 78 (3d Cir. 2008); (16) United States v. Mussacchia, 900 F.2d
493 (2d Cir. N.Y. 1990); (17) United States v. Pflum, 2004 U.S. Dist. LEXIS
26217 (D. Kan. Dec. 7, 2004); (18) United States v. Poll, 521 F.2d 329 (9th Cir.
1975), overruled by, United States v, Easterday, 539 F.3d 1176 (9th Cir. Cal.
2008); (19) United States v. Porth, 426 F.2d 519 (10th Cir. 1970); (20) United
States v. Ramirez, 92 A.F.T.R.2d (RIA) 7378 (W.D. Tex. 2003). The trend for
prosecutions of employment cases is up. See IRS Examples of Employment
Tax Investigations FY2008, www.irs.gov/compliance/enforcement/
article/0,,id=174631,00.html (last visited August 9, 2009) (reporting recent
employment tax investigations).
265 See Jeffrey A. Dubin, Criminal Investigation Enforcement Activities and
Taxpayer Noncompliance, 35 Pub. Fin. R. 500 (2007) (discussing the
effectiveness of criminal tax enforcement in raising overall compliance). Of
course with so few prosecutions under IRC section 7202 the effect of criminal
enforcement in this area is extremely limited.
266 Department of Justice: Tax Division, Criminal Tax Manual 9.03 (2009),
available at http://www.usdoj.gov/tax/readingroom/2001ctm/titlepg.htm;
United States v. Poll, 521 F.2d 329 (9th Cir. 1975), overruled by United States v.
Easterday, 539 F.3d 1176 (9th Cir. Cal. 2008). Poll is no longer good law. The
court in Poll gave a jury instruction putting wrong limitations on willfulness
based on the money in the corporate bank account. It is not cited for its
legal correctness but rather to demonstrate an attitude that has pervaded
regarding prosecutions for failure to pay over collected taxes—particularly
employment taxes. That attitude may be shifting and there has been an
uptick in the number of prosecutions for failure to pay collected taxes but the
uptick is simply a circumstance of going from almost none to several each
year in recent years. This number of prosecutions is still unlikely to generate
significant compliance. For a discussion of the correct legal standard in a
case that makes clear Poll has been overturned see United States v. Easterday,
564 F.3d 1004, 1007 (9th Cir. 2009) (“The statute under which Easterday
was found guilty is 26 U.S.C. § 7202, a fairly rarely invoked provision that
criminalizes a willful failure to pay over employees federal income tax
withholding on wages”). See U.S. Gov’t Accounting Office, GAO–99–
256, Payroll Taxes: Billions in Delinquent Taxes and Penalties Due
but Unlikely to Be Collected, 16 (1999) (containing observations of IRS
field collection personnel on the likelihood of criminal tax prosecutions for
the failure to pay collected taxes).
267 Perhaps it could be argued that so few prosecutions occur under this statute
because the statute has achieved its intended result of modifying behavior
so that prosecution is unnecessary. That does not appear to be the case here
Collecting Collected Taxes
163
given the discussion in the recent GAO report, the discussion in the Criminal
Tax Manual, the amount of the unpaid collected tax liability and the passage
of section 7215 in a later attempt to address the problem. More prosecution
of employment tax issues, though not necessarily under section 7202 or
section 7215, has occurred in recent years and the IRS is doing a good job of
publicizing these prosecutions on its web site for those who go there to look.
See IRS Examples of Employment Tax Investigations FY2008, www.irs.gov/
compliance/enforcement/article/0,,id=174631,00.html (last visited August 9,
2009) (providing examples of recent employment tax investigations).
268 26 U.S.C. § 7215 (2010).
269 This section provides:
“(a) Penalty—Any person who fails to comply with any provision of section
7512 (b) shall, in addition to any other penalties provided by law, be
guilty of a misdemeanor, and, upon conviction thereof, shall be fined
not more than $5,000, or imprisoned not more than one year, or both,
together with the costs of prosecution.
(b) Exceptions—This section shall not apply—
(1) to any person, if such person shows that there was reasonable doubt
as to (A) whether the law required collection of tax, or (B) who was
required by law to collect tax, and
(2) to any person, if such person shows that the failure to comply with
the provisions of section 7512 (b) was due to circumstances beyond
his control.
For purposes of paragraph (2), a lack of funds existing immediately after the
payment of wages (whether or not created by the payment of such wages)
shall not be considered to be circumstances beyond the control of a person.”
IRC § 7215 (2009).
270 This section provides:
“(a) General Rule—Whenever any person who is required to collect, account
for, and pay over any tax imposed by subtitle C, or chapter 33—at the
time and in the manner prescribed by law or regulations (A) fails to
collect, truthfully account for, or pay over such tax, or (B) fails to make
deposits, payments, or returns of such tax, and is notified, by notice
delivered in hand to such person, or any such failure, than all the
requirements of subsection (b) shall be complied with. In the case of a
corporation, partnership, or trust, notice delivered in hand to an officer,
partner, or trustee, shall for purposes of this section, be deemed to be
notice delivered in hand to such corporation, partnership, or trust and
to all officers, partners, trustees, and employees thereof.
164
Fogg
(b) Requirements—Any person who is required to collect, account for, and
pay over any tax imposed by subtitle C, or chapter 33, if notice has been
delivered to such person in accordance with subsection (a), shall collect
the taxes imposed by subtitle C, or chapter 33 which become collectible
after delivery of such notice, shall (not later than the end of the second
banking day after any amount of such taxes is collected) deposit such
amount in a separate account in a bank (as defined in section 581), and
shall keep the amount of such taxes in such account until payment over
to the United States. Any such account shall be designated as a special
fund in trust for the United States, payable to the United States by such
person as trustee.
(c) Relief From Further Compliance with Subsection (b)—Whenever the
Secretary is satisfied with respect to any notification made under
subsection (a), that all requirements of law and regulations with respect
to the taxes imposed by subtitle C, or chapter 33, as the case may be will
be henceforth be complied with, he may cancel such notification. Such
cancellation shall take effect at such time as is specified in the notice of
such cancellation.” IRC § 7512 (2009).
271 The twelve reported decisions are: United States v. Christopher, 1 Fed. Appx.
533 (7th Cir. Ill. 2001); United States v. Dreske, 536 F.2d 188 (7th Cir. 1976);
United States v. Erne, 576 F.2d 212 (9th Cir. 1978); United States v. Gay, 576
F.2d 1134 (5th Cir. Fla. 1978); United States v. Gordon, 495 F.2d 308 (7th Cir. Ill.
1974); United States v. Hemphill, 544 F.2d 341 (8th Cir. Mo. 1976); United States
v. Meriwether, 329 F. Supp. 1156 (D. Ala. 1971), aff ’d, 469 F.2d 1406 (5th Cir.
Ala. 1972); United States v. Paulton, 540 F.2d 886 (8th Cir. 1976); United States
v. Plotkin, 239 F. Supp. 129 (E.D. Wis. 1965); United States v. Polk, 550 F.2d 566
(9th Cir. 1977); United States v. Randolph, 588 F.2d 931 (5th Cir. Ga. 1979).
272 Edward Cheng aptly described the reaction of Congress to the compliance
problems with collected taxes, “When faced with undesirable behavior,
legislatures almost invariably turn to the criminal law to regulate.” Cheng,
supra note 53, at 656. The system of regulation created by IRC section 7202
and section 7215 sought to control the undesirable behavior of non-payment
of collected taxes but it failed because the problem was not one which
criminal prohibitions could fix.
273 It should be noted that prosecution for evasion of collected taxes is also
possible under IRC section 7201 and prosecution for failure to file tax returns
related to collected taxes is also possible under IRC section 7203. These
two provisions describe criminal sanctions that cast a broader net than just
seeking to prosecute for failures with respect to collected taxes. Occasionally,
a prosecution under one of these provisions will occur related to a
collected tax.
Collecting Collected Taxes
274
165
26 U.S.C. § 3505 (2010)
(“(a) Direct Payment by Third Parties—For purposes of sections 3102, 3202,
3402, and 3403, if a lender, surety, or other person, who is not an
employer under such sections with respect to an employee or group
of employees, pays wages directly to such an employee or group of
employee, employed by one or more employers, or to an agent on
behalf of such employee or employees, such lender, surety, or other
person shall be liable in his own person and estate to the United States
in a sum equal to the taxes (together with interest) required to be
deducted and withheld from such wages by such employer.
(b) Personal Liability Where Funds Are Supplied—If a lender, surety or
other person supplies funds to or for the account of an employer
for the specific purpose of paying wages of the employees of such
employer, with actual notice or knowledge (within the meaning of
section 6323 (i) (1)) that such employer does not intend to or will not
be able to make timely payment or deposit of the wages, such lender,
surety, or other persons shall be liable in his own person and estate to
the United States in a sum equal to the taxes (together with interest)
which are not paid over to the United States by such employer with
respect to such wages. However, the liability of such lender, surety, or
other person shall be limited to an amount equal to 25 percent of the
amount so supplied to or for the account of such employer for
such purpose.
(c) Effect of Payment—Any amounts paid to the United States pursuant to
this section shall be credited against the liability of the employer”).
275 H.R. Rep. No. 89–1884 (1966), reprinted in 1966–2 C.B. 815, 828–30; S. Rep.
No. 89–1708 (1966), reprinted in 1966 U.S.C.C.A.N. 3722, 2724, 2742–45. For
further discussion of Section 3505, see infra notes 471–93 and
accompanying text.
276 Net payroll lending refers to the practice of lending to permit an employer
to pay the salaries of its employees as reduced by the withheld taxes while
simultaneously refusing to lend the employer the funds to allow it to pay the
government the tax amounts.
277 The Taxpayer Bill of Rights 2 in 1996 made the most changes to section 6672.
Consistent with the title of the legislation, these changes did not seek to
impose additional duties or obligations on responsible persons but rather to
improve procedures for handling these cases in a manner that gave taxpayers
more rights in the determination of their liability. New subparagraph (b)
provided the requirement that a notice be issued prior to assessment and
gave taxpayers a right to have a hearing in the Appeals Division prior to an
166
Fogg
assessment. New subparagraph (d) gave responsible officers the right to seek
contribution from fellow responsible officers. New subparagraph (e) made
it more difficult to hold volunteer members of the board of directors of a
tax-exempt organization liable as responsible officers. The IRS Restructuring
and Reform Act of 1998 (some parts of which are referred to as Taxpayer Bill
of Rights 3) made minor changes to subparagraph (b) clarifying the method
of delivery of the required notice. None of these changes to section 6672
had any impact, or were intended to have any impact, on the effectiveness of
section 6672 as a tool to reduce the amount of unpaid collected taxes.
278 See U.S. Gov’t Accountability Office, GAO–08–617, Tax Compliance:
Businesses Owe Billions in Federal Payroll Taxes (2008) (discussing
outstanding payroll taxes).
279 Id. at 50.
280 Id. at 52.
281 See American Bar Association, Report and Recommendations on Taxpayer
Compliance, 41 Tax Law. 329 (1988) (identifying four types of noncompliance: 1) underreporting of income; 2) overstating deductions: 3) failure
to file returns: and 4) failure to pay established liabilities). The fourth type of
non-compliance is the subject of concern here. With respect to this type of
non-compliance, the ABA states “there is a persistent and difficult problem
of unsuccessful businesses failing to pay over to the IRS the taxes already
withheld from their employees’ wages. The Commission recommends that
more effective procedures be developed to allow the Service to intervene
quickly in such situations, before the tax becomes unmanageable.” Id.
282 As might be expected, states vary widely in how and whether they gather
this information. Included, as Appendix B, is a listing of the information
required by all states regarding the identification of responsible officers
at the time of incorporation. Many states do not collect information
concerning who is responsible for paying over collected taxes. Those states
that do collect the information do so in a variety of ways that offer several
models for how this might best occur. Fourteen states require identification
of persons responsible for payment of collected taxes at the time the entity
seeks its business registration: Illinois, Iowa, Kansas, Kentucky, Maine,
Maryland, Mississippi, New Jersey, North Carolina, Ohio, Oklahoma,
Oregon, Pennsylvania, and Virginia. Only Kansas appears to require the
signatures of all responsible officers which means in other states seeking
this information it is possible for someone to be listed as responsible
without their knowledge. See Appendix B (detailing state requirements
regarding identification of persons responsible for payment of collected
taxes at the time the entity seeks its business registration).
Collecting Collected Taxes
283
167
An Employer Identification Number (EIN) is a nine-digit number that
the IRS issues to certain business entities. Employers, sole proprietors,
corporations, partnerships, non-profit associations, trusts, estates of
decedents, government agencies, and certain individuals can apply for
EINs. Applicants may apply by telephone, fax, mail, or online. If applying
by fax or mail, applicants must complete Form SS–4 and submit it to the
IRS. EINs are issued within four days for faxed applications and three
weeks for mailed applications. The IRS prefers applicants apply online.
Using the online application process, the IRS asks the applicant a series
of questions (very similar to questions asked on Form SS–4), attempts to
validate the information, and, if the information is validated, issues the
applicant a permanent EIN. If the information is not validated, an EIN will
not be issued; however, the applicant will have an opportunity to correct the
information during the same session. The telephone application process is
very similar to the online application process; applicants receive an EIN at
the end of the conversation.
The Application for Employer Identification Number, Form SS–4,
requires the applicant to disclose its legal name, trade name, and physical
and mailing addresses. The applicant must designate an individual to
contact regarding tax matters and identify a principal officer, general
partner, grantor, owner, or trustor and that individual’s social security
number, taxpayer identification number (ITIN), or EIN. The applicant
must disclose its type of organizational structure (sole proprietorship,
partnership, corporation, etc.) and the reason the entity is applying for an
EIN. The application asks approximately how many employees the entity
plans to employ, if the entity expects its tax liability to be greater than
$1,000, and if the entity has received an EIN previously. The applicant must
also check one of twelve boxes describing the principal activity of the entity.
Finally, the applicant must sign the application.
Form SS–4 has a Third Party Designee option whereby the applicant
of the EIN authorizes a named individual to receive the entity’s EIN and
answer questions regarding Form SS–4. If the applicant uses a third party
designee, the application requires identification of the third party designee
and disclosure of the third party’s telephone number, fax number, and
address. A third party designee may also be used to obtain an EIN by
telephone, but the applying entity will need to fax the Third Party Designee
section to the telephone application center. See IRS Employer Identification
Numbers, http://www.irs.gov/businesses/small/article/0,,id=98350,00.html
(last visited October 11, 2009) (explaining EINs generally); IRS Application
for Employer Identification Number (2009), available at http://www.roa.
org/site/DocServer/irs_ss4.pdf?docID=10324 (setting forth procedure to
apply for an EIN).
168
284
Fogg
26 U.S.C. § 6672 (2010).
U.S. Gov’t Accountability Office, GAO–08–617, Tax Compliance:
Businesses Owe Billions in Federal Payroll Taxes, 32–33 (2008).
286 See Report and Recommendations, supra note 281 (stating “[t]here is a
tendency for the media to depict the Service in such cases as somehow
victimizing the business and causing it to fail. But, in fact, the business had
already failed; theft of employee withholdings only postpones the day of
reckoning, at a substantial cost to the public treasury”).
287 See U.S. Gov’t Accountability Office, GAO–07–742T, Tax Compliance:
Thousands of Federal Contractors Abuse the Federal Tax System
(2007) (addressing noncompliance by entities who have contracts with the
federal government). To address the issue of identifying these entities prior
to entering into contracts, a new federal government rule has been adopted
requiring companies seeking federal contracts to certify whether they have
any outstanding federal tax debts over $3,000. See Dave Rifkin, A Primer on
the ‘Tax Gap’ and Methodologies for Reducing It, 375 Quinnipiac L. Rev. 375,
416 (2009) (discussing shaming).
288 See supra note 282.
289 Willfulness under section 6672 is not defined in the Internal Revenue Code.
The standard in section 6672 differs from willfulness in a criminal case.
The criminal standard is “a voluntary, intentional violation of a known legal
duty.” Cheek v. United States, 498 U.S. 192, 192, 111 S. Ct. 604, 605 (1991).
The standard for willful conduct in section 6672 requires only a “voluntary,
conscious, and intentional—as opposed to accidental-decision not to remit
funds properly withheld to the government.” Kalb v. United States, 505
F.2d 506, 511 (2d Cir. 1974). Other courts have defined willfulness under
section 6672 as the decision to prefer other creditors over the government.
Muck v. United States, 3 F.3d 1378, 1381 (10th Cir. 1993). Courts will
generally find willfulness present if either the responsible individual knows
that other creditors are being paid when the government is not or the
responsible person should have known that was happening given all of
the circumstances. Usually, defenses to willfulness involve attempts to pin
knowledge of the facts on others coupled with a reliance that the others
were doing what they were supposed to do. The success of this type defense
usually turns on the nature of the job held by the responsible person and
the reasonableness of any reliance given the nature of the job and past
actions by others on whom they relied.
Not all states have a willfulness component to their determination of
liability. Some states also hold officers strictly liable. For these reasons
285
Collecting Collected Taxes
169
use of state law data on both responsibility and willfulness is not a perfect
match for federal law.
290 IRS Revenue Officers begin considering the potential applicability of the
trust fund recovery penalty (“TFRP”) during the first field visit to the
entity taxpayer. IRS Internal Revenue Manual (August 1, 2009), § 5.1.10.3.2,
available at http://www.irs.gov/irm/part5/irm_05–001–010.html#d0e227.
After the first meeting with the entity taxpayer, revenue officers decide
whether to pursue the TFRP, but the presumption is to pursue the TFRP. Id.
at § 5.7.4.1. In investigating the TFRP, revenue officers follow four general
procedures, establishing responsibility, establishing willfulness, conducting
collectability determinations, and conducting Form 4180 interviews. Id.
at § 5.7.3.1.2. The revenue officer identifies potential responsible persons
by reviewing the corporate by-laws to determine duties of officers and
determining who has the authority to hire and fire employees, determine
which creditors to pay, control payroll and disbursements, sign employment
tax returns, and make federal tax deposits. Id. at § 5.7.3.3.1. During the first
field visit to the entity taxpayer, revenue officers must conduct interviews
with the greatest number of potentially responsible persons possible. Id. at
§ 5.7.4.2. The revenue officer’s personal interview achieves a dual purpose,
to give information and gather information. Id. The revenue officer
discusses the TFRP and potential personal liability, while also gathering bank
statements and cancelled checks to support asserting the TFRP. Id. The
revenue officer attempts to fill out Form 4180 to record as much information
about the person’s duties and authority in the entity to determine if the
statutory tests for a responsible person have been met to support asserting
the TFRP. Id. The revenue officer also must review corporate records such
as articles of incorporation, minute books, payroll records, and employment
tax returns to determine the duties of the officers and the responsibilities
of the individuals to file and pay collected taxes. Id. Revenue officers also
review bank records such as canceled checks, signature cards and loan
applications to determine who had authority to sign checks, deposit funds,
and obligate the entity by borrowing money. Id. The revenue officer reviews
all information and documentation collected during the investigation and
prepares a report on Form 4183. Id. at § 4.2.1. The revenue officer makes
a determination as to whether each person meets the statutory tests for
responsible persons and conducts a collectability inquiry for each person. Id.
The report details the revenue officer’s recommendations for assertion or non
assertion of the TFRP as to each potentially responsible person investigated
by the revenue officer. Id.
291 The IRS can seek to create an administrative presumption that listing
your name as responsible with the IRS is treated as meeting the test for
170
Fogg
responsibility. If section 6672 were amended to specifically provide for this
presumption, it would be safer for the IRS to engage in this reliance for a part
of its proof of section 6672 liability.
292 Presently, Kansas imposes this requirement in Part 11 of the Kansas Business
Tax Application. Part 11 requires that the entity “List all owners, partners,
corporate officers and directors. Provide the personal information and
signatures of all persons who have control or authority over how business
funds or assets are spent.” Kansas Business Tax Application (2009),
available at http://www.ksrevenue.org/pdf/forms/cr16.pdf. By requiring
each responsible officer to personally sign the form, Kansas makes each of
these individuals acknowledge that they are indeed responsible rather than
relying on one person to list others who may, or may not, agree that they are
responsible once problems occur. Requiring these signatures is an integral
part of the process of proving these individuals have responsibility. The form
Kansas uses could be adopted by the IRS.
293 Having individuals sign not only provides proof of responsibility when
an issue arises but it provides the opportunity to give those self-identified
individuals information about the duties of a responsible officer. Providing
information to the responsible officers could impact their decisionmaking when the time comes for making payments to the IRS versus
other creditors. Several states provide information in their registration
booklets about the personal liability that may apply when collected
taxes are not paid. For example, Virginia cites to its state statute holding
individuals personally liable on the second page of its Business Registration
Application where it requires a listing of all responsible persons, and again
on the seventh page where it describes the consequences of failure to pay
the collected taxes. See Virginia Business Registration Application (2009),
available at www.tax.virginia.gov/web_pdfs/busforms/fr199.pdf (setting
forth requirements to register a business in Virginia). The literature
examining effective compliance mechanisms suggest the providing of
information as one of the components of effective administration. Report
and Recommendations, supra note 281 at 368–383.
294 See Corrie Lynn Lyle, The Wrath of IRC 6672: The Renewed Call for Change—
Is Anyone Listening? If You are a Corporate Official, You Had Better Be, 74
S. Cal. L. Rev. 1133, 1135 (2001) (“Most corporate officers … do not realize
that this [the failure to pay over the collected taxes] is a Pandora’s Box that
can result in personal liability for thousands or even millions of dollars in
unpaid payroll taxes”). The author of the article does not cite to any authority
for the statement that most corporate officers do not realize the difference
between failure to pay these collected taxes and failure to pay other corporate
liabilities, but the sentiment expressed in the article is widely held by those
Collecting Collected Taxes
171
who have encountered individuals trying to explain why the section 6672
liability should not attach to them. Almost universally, they profess a lack of
knowledge of the consequence of their action in preferring other creditors
over the payment of the collected taxes to the United States.
295
See U.S. Gov’t Accounting Office, GAO–99–256, Payroll Taxes:
Billions in Delinquent Taxes and Penalties Due but Unlikely to Be
Collected, 15 (1999)
(“The GAO report contains a discussion of IRS efforts at taxpayer
education in this area. Some collection officials observed that the
persons most in need of this training did not attend. At least, the IRS
offered the training; however, if the education effort is to succeed it
must reach everyone who is potentially responsible”).
In testimony before the Senate Committee on Homeland Security and
Government Affairs on July 29, 2008, Linda Stiff recited a litany of actions
taken by the IRS to inform small businesses of their employment tax
responsibilities:
“ One of the means to accomplish this is through the use of the
Federal Tax Deposit Alert process which helps to identify, at
an early stage, taxpayers classified as bi-weekly depositors who
have not made federal tax deposits during the current quarter,
or have made deposits in substantially lower amounts from
prior quarters. Virtually all of the IRS functional and operating
divisions participate in employment tax outreach and education… .
Applicants for a new Employer Identification Numbers receive
considerable information based on what they indicate on their
Form SS–4. Those who state they have or will have employees
automatically get a copy of Circular E, Employer’s Tax Guide;
information for making Federal Tax Deposits; and enrollment
information for the Electronic Federal Tax Payment System …
The IRS also works with community partners to present Small
Business Tax Workshops throughout the United States. These
workshops instruct new and prospective business owners in federal
tax responsibilities, including employment taxes… . Quarterly,
the IRS sends out approximately seven million SSA IRS Reporter
newsletters with Form 941, Employer’s Quarterly Federal Tax
Return. The newsletter contains information on subjects such as
Social Security law, the Electronic Federal Tax Payment System,
or changes in Social Security or IRS electronic filing systems. In
addition, in 2007, the IRS developed a new brochure, Publication
4591—Small Business Federal Tax Responsibilities, which includes
information about employment taxes.”
172
Fogg
Written Testimony of IRS Deputy Commissioner Linda Stiff Before
the Senate Committee on Homeland Security and Governmental Affairs
Permanent Subcommittee on Investigations on the Collection of Federal
Employment Taxes, 3–10 (July 29, 2008). Publication 4591 itself provides little
information. Circular E addresses the trust fund penalty on page 24 of a 69
page publication and even then does not talk about the fact the collected
taxes are held in trust or that the trust fund penalty is non-dischargeable in
bankruptcy. Much information exists on the web and undoubtedly good
information is transmitted in training sessions for those who attend; however,
as noted in the GAO report, the persons who most need to know are the least
likely to attend such sessions. This is why requiring each responsible officer
to sign and acknowledge the duty at the outset is critical.
296 Oklahoma has a powerful statement above the signature line:
“I, the undersigned applicant or authorized representative, declare
under the penalties of perjury that I have examined this application
and attachments and to the best of my knowledge the facts set forth
are true and correct, and that the requirements hereunder will be
carried out in accordance with the laws of the State of Oklahoma and
the rules and regulations of the Oklahoma Tax commission. I further
acknowledge and agree that sales, withholding and motor fuel taxes
are trust funds for the State of Oklahoma and that any use of these
trust funds other than timely remittance to the State of Oklahoma is
embezzlement and can result in criminal prosecution.”
Having the responsible person further acknowledge that the failure of the
business entity to pay the funds held in trust becomes a personal liability that
cannot be discharged in bankruptcy might further enhance the statement by
impressing the serious personal consequences of the action. See Oklahoma
Business Registration Packet (2009), available at http://www.tax.ok.gov/
forms/busregpk.pdf (setting forth requirements to register a business in
Oklahoma).
297 Many states do this in their business registration package. See,
e.g., Alabama Department of Revenue Combined Registration/
Application (2009), available at www.excel-pay.com/.../ALCom101%20
–%20CombinedRegistrationApplication.pdf (Alabama) (requiring
acknowledgement by an individual that he or she has responsibility
for the payment of trust fund taxes); Kansas Business Tax Application
(2009), available at http://www.ksrevenue.org/pdf/forms/cr16.pdf
(Kansas) (requiring acknowledgement by an individual that he or she has
responsibility for the payment of trust fund taxes); Minnesota Application
for Business Registration Instruction Booklet (2009), available at http://
www.taxes.state.mn.us/instructions/abr_in.pdf (Minnesota) (requiring
Collecting Collected Taxes
173
acknowledgement by an individual that he or she has responsibility for
the payment of trust fund taxes); Business Registration Application &
Instructions (2009), available at www.state.nj.us/treasury/taxation/pdf/
other_forms/git-er/njwt.pdf (New Jersey) (requiring acknowledgement
by an individual that he or she has responsibility for the payment of trust
fund taxes); North Carolina Business Registration Application for Income
Tax Withholding (2009), available at www.dornc.com/downloads/fillin/
NCBR_webfill.pdf (North Carolina) (requiring acknowledgement by an
individual that he or she has responsibility for the payment of trust fund
taxes); Application to Register for Income Tax Withholding and Sales and
Use Tax Permit (2009), available at http://www.nd.gov/tax/salesanduse/
forms/withholdsalesapplication-enabled.pdf (North Dakota) (requiring
acknowledgement by an individual that he or she has responsibility for the
payment of trust fund taxes); Pennsylvania Enterprise Registration Form
and Instructions (2009), available at www.revenue.state.pa.us/revenue/
lib/revenue/pa-100.pdf (Pennsylvania) (requiring acknowledgement by an
individual that he or she has responsibility for the payment of trust fund
taxes); Utah State Business and Tax Registration, TC–69 (2009), available at
tax.utah.gov/forms/current/tc-69.pdf (Utah) (requiring acknowledgement
by an individual that he or she has responsibility for the payment of trust
fund taxes); Virginia Business Registration Application (2009), available at
www.tax.virginia.gov/web_pdfs/busforms/fr199.pdf (Virginia) (requiring
acknowledgement by an individual that he or she has responsibility for the
payment of trust fund taxes).
298 11 U.S.C. §§ 507 (a) (8) (C), 523 (a) (1) (A) (2009). The liability for collected
taxes receives priority status without regard to its age. Because it has priority
status, the exceptions to discharge always apply to individual debtors.
Because it has priority status, an entity in a Chapter 11 reorganization
must provide for full payment in order to obtain plan confirmation. 11
U.S.C. § 1129 (a) (9) (C) (2009).
299 For a link between information and compliance see supra note 254 and
accompanying text.
300 One state, Michigan, not on the list for requiring information on responsible
officers, has identified the situation with payroll providers as another
problem area for collected taxes, which is causing noncompliance. Michigan
specifically informs entities that the hiring of a payroll provider does not
remove responsible individuals from their obligation to pay over the collected
taxes. See Michigan Tax Form 3683 (2009), available at www.michigan.gov/
documents/3683f_2906_7.pdf (informing businesses that the hiring of a
payroll provider does not remove responsible officers from their obligation to
pay trust fund taxes). “Payroll Service Provider” is a name given to members
174
Fogg
of an industry that manages the administrative aspects of a company’s
payroll. Usually, the company provides the payroll service provider
(PSP) with employee names and compensation and the PSP does payroll
processing, check generation and distribution, payroll tax filing, and W–2
generation. In more sophisticated arrangements, the PSP might also handle
401 (k) management, employee handbook development, and direct deposit
registration. Almost all PSPs handle the payroll tax filing. The IRS even has a
list of approved PSPs. See IRS Payroll Service Providers, http://www.irs.gov/
efile/lists/0,,id=101120,00.html (last visited October 12, 2009) ( listing payroll
service providers). The companies on that list all have web sites that detail
their services. Most companies working with PSPs have little trouble meeting
their collected tax obligations to the IRS as long as they have cash to make
the required payments. A few PSPs have acted irresponsibly with the money
from their clients. When this happens, the company suffers the loss rather
the IRS because the PSP is the agent of the company.
301 Several payroll provider companies found their clients’ money too tempting
and took the money intended to pay the taxes of their clients without paying
it over to the IRS. Usually, these schemes ran something like a Ponzi scheme
in order for the payroll provider to keep out of trouble as long as possible.
See, e.g., Atlas Hotels v. United States, 140 F.3d 1245 (9th Cir. Cal. 1998)
(affirming summary judgment to defendant United States in plaintiffs, payroll
tax service providers, action seeking a refund of late payment penalties and
interest paid by the company to whom plaintiffs provided payroll tax service);
Morin v. Fronties Business Technologies, 288 B.R. 663 (W.D.N.Y. 2003)
(affirming orders of bankruptcy court); Pediatric Affiliates v. United States,
230 Fed. Appx. 167 (3d Cir. N.J. 2007) (affirming the district court’s order
dismissing the taxpayer’s complaint); Wolff v. United States, 372 B.R. 244
(D. Md. 2007) (affirming order of the bankruptcy court with respect to its
dismissal of all counts, except avoidance of preferential transfers made within
ninety days preceding the petition date and turnover of avoided preferential
payments which was remanded to the bankruptcy court).
302 On line 22 of Michigan’s Form 518 “Registration for Michigan Taxes”
Michigan requests those filling out the form to check a box “if you use
a payroll service that produces your payroll checks and sends income
tax withholding payments to the State.” Michigan also requests that
taxpayers using a payroll provider fill out and attach to their registration
Michigan Form 3683 “Payroll Service Provider Combined Power of
Attorney Authorization and Corporate Officer Liability (COL) Certificate
for Businesses.” See Michigan Tax Form 3683 (2009), available at www.
michigan.gov/documents/3683f_2906_7.pdf ( detailing Michigan
requirements for businesses using a payroll provider to withhold and pay
Collecting Collected Taxes
175
over employment taxes); see also IRS Outsourcing Payroll Duties, http://
www.irs.gov/businesses/small/article/0,,id=176943,00.html (last visited
October 11, 2009) (providing IRS statements on the use of payroll providers).
303 The penalty for failure to list oneself as a responsible officer could either be
imposed irrespective of substantive compliance similar to the FBAR penalties
or it could require substantive compliance as a triggering mechanism.
Because the penalty should be substantial enough to inflict pain for noncompliance, requiring a substantive compliance triggering mechanism makes
it easier to support a higher penalty. Alternatively, the penalty could exist
for simple failure to notify with a higher penalty amount if it is coupled
with non-compliance. Compliance should be made easy by having a line on
the Form 941, or similar collected tax document, asking if there have been
any changes in the ranks of responsible officers since the last filing (or the
initial EIN request) and providing an attachment to the form for the new
responsible officers to sign.
304 See IRC § 3102 (2009) (social security taxes); IRC § 3402 (2009) (income taxes).
305 See Treas. Reg. § 31.3102–1 (a) (“The employer shall collect from each of
his employees, the employee tax with respect to wages for employment
performed for the employer by the employee… . The employer is required
to collect the tax, notwithstanding the wages are paid in something other
than money, and to pay over the tax in money.”); Treas. Reg. 31.3402 (a)-1 (b)
(“The employer is required to collect the tax by deducting and withholding
the amount thereof from the employee’s wages… .”); Treas. Reg. 49.4251–2 (c)
(regarding telephone excise taxes “[t]he taxes imposed by section 4251 are
payable by the person paying for the services rendered, and must be paid
to the person rendering the services who is required to collect the tax and
return and pay over the tax”).
306 See Mark Crain, The Impact of Regulatory Costs on Small Firms 28,
50 (SBA Office of Advocacy 2005) (discussing the disproportionality of the
burden imposed by federal regulations on small business); J. Scott Moody,
The Cost of Complying with the Federal Income Tax, Special Report
No. 114 10 (Tax Foundation 2002) (estimating how much it costs individuals
and businesses to read the rules, fill out forms, and do all the necessary
things to comply with the nation’s tax laws); Francis Chittenden, et al., Tax
Regulation and Small-business in the USA, UK, Australia and New Zealand, 21
Int’l Small Bus. J. 93, 98–99 (2002) (determining that small businesses face
much higher costs per employee compared with large firms).
307 For a general explanation of this practice see Corrie Lyle, The Wrath of
IRC 6672: The Renewed Call for Change—Is Anyone Listening? If You Are a
Corporate Official, You Had Better Be, 74 S. Cal. L. Rev. 1133, 1156–57 (Spring
2001). For a discussion of how the payroll provider business works see
176
Fogg
generally Sandra Boncek, Yahoo Small Business How Payroll Systems Price
Their Services, http://smallbusiness.yahoo.com/r-article-a-2111–m-2–sc-52–
how_payroll_systems_price_their_services-i (last visited August 3, 2009);
Sandra Boncek, Yahoo Small Business, What Do Payroll Providers Typically
Offer?, http://smallbusiness.yahoo.com/r-article-a-2140–m-2–sc-52–what_
do_payroll_companies_typically_offer-i (last visited August 3, 2009); ADP,
http://www.adp.com/ (last visited July 31, 2009).
308 See John F. Due & John L. Mikesell, Sales Taxation: State and Local
Structure and Administration 327 (John Hopkins University Press 2d
ed. 1983); Phillip Mattera with Leigh McIlvaine, Skimming the Sales Tax:
How Wal-Mart and other Big Retailers (Legally) Keep a Cut of the Taxes We
Pay on Everyday Purchases, 50 State Tax Notes 713 (2008). As discussed
in Skimming the Sales Tax the manner in which states affect this carve-out
varies wildly among the states. Usually, the carve-out takes the form of a
percentage of sales taxes collected. The percentage varies wildly from state to
state. Some states cap the amount of recovery a retailer can obtain under this
statute while others have no cap. As discussed further below, this proposal
suggests a modest percentage of the collected taxes be used, as the base from
which to compute the incentive payment while a reasonable cap is placed on
the total amount of incentive payment available to avoid creating a significant
windfall to any one recipient.
The concept of paying taxpayers for collecting taxes is somewhat
controversial. Certainly, real economic costs exist with the program of
collecting these taxes. Some cooperation from the taxpayer may be built if
the program is properly administered. Such an incentive could also assist
with respect to the filing of tax returns.
309 This proposal only applies to small businesses because the incentive to pay
appears much more needed for small businesses run by entrepreneurs than
large business run by executives and managers. The benefit could be cut
off at the point used to divide taxpayers categorized as Small Business Self
Employed under the IRS scale or some other categorization pegged more
closely to the compliance issues presented by nonpayment of collected
taxes. If you allow all businesses to participate in this type of incentive,
the types of problems that exist in those states with unlimited sales tax
discounts will result. See Mattera, supra note 308. In states with unlimited
refunds the large retailers, who would comply in any event and for whom
the cost of compliance is insignificant relative to their overall operation,
soak up the lion’s share of the of discounts paid out under the program. See
Pennsylvania Budget and Policy Center, A Tax Windfall Whose Time Has
Passed, Understanding Pennsylvania’s Sales Tax vendor Discount Program
(2008) (“The sales tax discount program cost Pennsylvania nearly $74 million
Collecting Collected Taxes
177
in 2007–2008. Nearly $12 million of that went to 10 corporations with sales
exceeding $1 billion in the state. On average, each of those vendors kept $1.16
million in sales tax they collected. Small businesses, by contrast, received
only a few dollars from the program. A little more than half of the 301,000
licensed vendors in the state, with sales less than $100,000, got average
discounts of $9, while another 21 percent got nothing”).
310 Because the goal is to get the taxpayers off on the right foot, one way to hold
down the cost of this incentive other than limiting it to small businesses is to
limit it in time to the first two or three years of a taxpayer’s business. These
early years are critical times for small businesses because over half fail within
the first four years. See Rafael Efrat, The Tax Burden and the Propensity of
Small Business Entrepreneurs to File for Bankruptcy, 4 Hastings Bus. L.J.
175, 204–06 (2008) (discussing the danger of failure for the self-employed
and small businesses). Having learned good behavior in the early years, the
surviving taxpayers would then be well trained on these tax obligations and
posed to continue with good compliance.
311 See Dennis J. Ventry, Jr., Cooperative Tax Regulations, 41 Conn. L. Rev.
431 (2009) (making a case for discounted rates for timely and accurate
compliance); see also Joshua D. Rosenberg, The Psychology of Taxes: Why
They Drive Us Crazy, and How We Can Make Them Sane, 16 Va. Tax Rev.
155, 168 (1996) (“People are significantly more likely to actually adopt desired
behaviors in response to a system that: (1) uses rewards for correct behaviors
in addition to, and where possible, in place of, punishments for wrong
behaviors: and (2) ensures that both positive reinforces and punishments are
administered swiftly and consistently”).
312 For a discussion regarding reimbursement to taxpayers for their costs in
a slightly different context see Joseph Bankman, Tax Enforcement: Tax
Shelters, The Cash Economy, and Compliance Costs, 31 Ohio N. U. L. Rev. 1, 9
(2005); Joseph Bankman, Who Should Bear Tax Compliance Costs? (Stanford
Law School, John M, Olin Program in Law & Econ., Working Paper No.
279, 2004), available at http://ssrn.com/abstract=519783; Sarah Lawsky,
Fairly Random: On Compensating Audited Taxpayers, 41 Conn. L. Rev. 161
(2008). Professor Bankman proposes reimbursing taxpayers for the costs
they incur in going through an audit. Professor Lawsky opposes this idea;
however, at least a part of her concern is the randomness of those receiving
reimbursement. The proposal here imparts all businesses, within the selected
economic range, that serve as trustees for payroll taxes.
313 As with bonds, discussed infra, providing some compensation for the
service of acting as a trustee fits into the traditional model of serving as a
trustee. The Government definitely benefits from the work done by the
business entity serving as a trustee and collecting the taxes. Just as certain
178
Fogg
trustees should be bonded to insure compliance with the trustee duties
so should certain trustees be compensated for their efforts on behalf of
the Government. Because this proposal excludes large entities and noncompliant entities from the incentive program, it is not a perfect match
with outside trustee, but the proposal does reflect a more traditional model
for interaction between the trustee and the beneficiary. Interim Report,
Small Seller and Vendor Compensation Task Force of the Streamlined Sales
Tax Governing Board (December 15, 2008); see also Sales Tax Fairness and
Simplification Act, H.R. 3396, 110th Cong. § 6 (a) (14) (2007) (detailing vendor
compensation); Sales Tax Fairness and Simplification Act, S. 34, 110th Cong.
§ 7 (a) (14) (2007).
314 While merely a tangential thought, the recent debate concerning private
debt collection of federal taxes raises an interesting contrast with the use
of businesses to collect the bulk of federal taxes. Many commentators
assailed the use of private debt collectors for what was an “inherently
governmental function.” Yet, they failed to address the fact that most federal
taxes are privately collected at the front end. The objections to private debt
collection caused it to end. See IRS Private Debt Collection, http://www.
irs.gov/businesses/small/article/0,,id=155136,00.html (last visited August
1, 2009) (discussing private debt collection); National Taxpayer Advocate’s
Report to Congress, 27 (2009), available at http://www.irs.gov/pub/irs-utl/
fy09objectivesreport.pdf (discussing private debt collection).
315 Only trustees that timely file the returns for the collected taxes and timely
pay the taxes reflected on those returns would be eligible for the incentive
discount. This limitation on the discount reflects the practices of most,
but not all, of the states with the incentive discount program. See Good
Jobs First, Skimming the Sales Tax: How Wal-Mart and Other Big Retailers
(Legally) Keep a Cut of the Taxes We Pay on Everyday Purchases 20–22
(November 2008) (finding [t]wenty-six states have this feature).
316 A recent national survey commissioned by the Joint Cost of
Collection Study, a public/private sector group, and conducted by
PricewaterhouseCoopers LLP, has shown that in fiscal year 2003 the total
cost to sellers to collect state and local sales taxes was $6.8 billion. This
amount was calculated after subtractions for state vendor discounts and
retailer float on the sales tax revenues.
The study showed that for fiscal year 2003, for retailers selling between
$150,000 and $ 1 million the average cost was 13.47 percent of the sales taxes
collected or approximately $2,386; for mid-size retailer, between $1 million
and $10 million in sales, the average cost was 5.2 percent or approximately
$5,279; and for the larger retailers, over $10 million in sales, the average
cost of collection was 2.17 percent or approximately $18,233. See National
Collecting Collected Taxes
179
Economic Consulting, Retail Sales Tax Compliance Costs: A National
Estimate 12 (PriceWaterhouseCoopers 2006). It is important to remember
that these amounts, including the total cost for all retailers of $6.8 billion,
are not reimbursed to the retailer by the state or local government, these
costs come out of the retailer’s own pocket. See Sales Tax Fairness and
Simplification Act: Hearing on H.R. 3396 Before the Subcomm. on Admin
and Commercial Law Comm. on the Judiciary, 110th Cong. (97) (2007)
(statement of Steven Rauschenberger, Past President, National Conference
of State Legislatures).
317 See Austin Wakeman Scott & Mark Ascher, 2 The Law of Trusts 39 (Aspen
6th ed. 2006) (1886).
318 Appendix C contains a list of the laws of the states concerning bonding
of retailers who incur state sales tax obligations. Four out of five states
have some form of bonding requirement. Some states have a list of
criteria. For example, Arizona, Florida, Georgia, and Iowa have bonding
requirements. Other states simply leave the decision to require a bond to
the tax administrator. For example, Maryland requires a bond “to protect
tax revenue”, New Jersey requires a bond whenever the “director deems
necessary”, and New Mexico requires a bond whenever it is necessary to
ensure payment of any tax. States that provide some guidance to their
tax administrators concerning when a bond is necessary, create a better
atmosphere for administration since both the tax administrators and the
public know when to expect a bond. Some states describe in their statute
when a bond should be released due to prolonged compliance. For example,
California requires that security held by the Board shall be released after a
three-year period in which the person has timely filed all returns and paid all
taxes to California.
319 It would have little effect because the entity could simply refuse to post
the bond, continue operations and continue to add on additional unpaid
employment tax liabilities. The ability to enjoin entities from continued
operation arguably exists as an option for the IRS already. See U.S. Gov’t
Accounting Office, GAO–99–256, Payroll Taxes: Billions in
Delinquent Taxes and Penalties Due but Unlikely to Be Collected,
16 (1999) (“A few field personnel noted that IRS could seek injunctions
through the U.S. Attorney’s Office to prevent taxpayers from accumulating
multiple payroll tax delinquencies and that District Counsel prefer not to
seek such injunctions due to the time and expense required to prosecute
such cases.”); See also Written Testimony of IRS Deputy Commissioner Linda
Stiff Before the Senate Committee on Homeland Security and Governmental
Affairs Permanent Subcommittee on Investigations on the Collection of Federal
Employment Taxes, 2 (July 29, 2008) (suggesting that an injunctive remedy
180
Fogg
exists while not citing to any instances when one was actually obtained:
“The IRS is developing and testing streamline procedures that will assist
Field Collection when developing and requesting a suit for injunctive relief.
Specifically, through an understanding with the Department of Justice, the
IRS will be putting forward injunction suits that are based on more flexible
standards for showing that the government is being irreparably harmed by
the non-payment of employment taxes, that further administrative activity
would be futile, and that no adequate remedy at law exists”). Attempts to
enjoin entities have essentially gone nowhere to date because of the difficulty
of bringing a traditional injunction suit, the restraint on business formation
such a restraint would place, and the resource issue bringing such suits
imposes on the IRS and the Department of Justice.
320 See Appendix C (listing state bonding requirements).
321 See Begier v. United States, 496 U.S. 53 (1990) (discussing this fiction); see
also Harris J. Diamond, Tracing Cash Proceeds in Insolvency Proceedings
Under Revised Article 9, 9 Am. Bankr. Inst. L. Rev. 385, 410 (2001); Christian
Onsager, et al., Trust Fund Taxes in Bankruptcy Beiger v. IRS Five Years Later,
15–1 Am. Bankr. Ist. L. Rev. 15 (Feb, 1996); Richard Orrell-Jones, Blinded
By the Debtor’s Headlights: Deer Park’s Liquidation of United States v. Energy
Resources, 12 Emory Bankr. Dev. J. 451 (1996).
322 See Appendix B (identifying states that impose interest on responsible
officers from the due date of return). Most states do have employment
taxes; however, no states appear to require a bond for the payment of
employment taxes. Compare Appendix B (identifying states that impose
interest on responsible officers from the due date of return) with Appendix
C (identifying states with bonding requirements). While most states do
not distinguish how they charge interest to responsible persons based
on whether the underlying tax is a sales tax or an employment tax, four
states (Idaho, New York, West Virginia, and South Carolina) make that
distinction. See IRS Service Center Advisory, IRS SCA 200026024, 2000
WL 33116108 (June 30, 2000). It is unclear whether states have made these
distinctions knowingly based on specific policies or have simply come to
these results through different legislative paths when the employment taxes
and sales taxes did not move in tandem.
323 26 U.S.C. § 6672 (2010).
324 See IRC § 6672 (c) (1) (C), (c) (3) (2009) (setting forth procedure if a
taxpayer fails to collect and pay over tax to the IRS or attempts to evade or
defeat a tax).
325 26 U.S.C. § 6331 (i) (2010).
Collecting Collected Taxes
326
181
IRC § 6331 (i) (2009) (“(1) In General—No levy may be made under
subsection (a) on the property or rights to property of any person with
respect to any unpaid divisible tax during the pendency of any proceeding
brought by such person to a proper Federal trial court for the recovery of any
portion of such divisible tax which was paid by such person…”).
327 The effect of requiring a bond may terminate the business because the cash
strapped entity will not have the resources to post the bond. This result
could have long-term beneficial effects for tax administration by removing
at an early stage those businesses that simply are not viable, but obviously,
it also has the effect of placing more pressure on already vulnerable small
businesses. Because many small businesses with outstanding collected taxes
due to the IRS will have difficulty finding a surety willing to post a bond or
because the cost of the bond will be prohibitively high, this provision is likely
to result in few bonds. Taxpayers will be forced to pay the tax, go out of
business, or be enjoined.
328 If a responsible person or officer of the entity had an outstanding liability at
the time of incorporation, then a bond would be required unless the prior tax
issues were resolved. See Utah State Business and Tax Registration, TC–69
(2009), available at http://tax.utah.gov/forms/current/tc-69.pdf (setting forth
requirements to register a business in Utah). Even if the prior issues were
resolved, the existence of prior liabilities by one or more of the responsible
persons or officers of a newly formed entity, may provide a basis for requiring
a bond since their past behavior may suggest the potential for problems
with payment of the collected taxes. This is an area where the suggestion for
identification of responsible persons provides the government with a basis
to research the background of the individuals running the entity that seeks
to become a trustee of the government and allows the government to inform
itself whether a bond or some other action is needed in order to protect itself.
329 An entity that continually increases the amount of unpaid collected taxes
as it keeps its doors open is engaged in the “pyramiding” of taxes. The
pyramiding of collected taxes has long been a major problem for the IRS.
Allowing the IRS a quick entrance into court to stop the pyramiding of
further taxes would solve a problem for the IRS in dealing with pyramiding
situations. See Written Testimony of IRS Deputy Commissioner Linda Stiff
Before the Senate Committee on Homeland Security and Governmental
Affairs Permanent Subcommittee on Investigations on the Collection of Federal
Employment Taxes, 2 (July 29, 2008) (“Taxpayers often attempt to ‘pyramid’
their liabilities as a means of deferring payment or delaying enforcement
action. One form of pyramiding occurs when the same business fails to remit
payroll taxes for multiple quarters. The second form of pyramiding occurs
when the owner of a delinquent business closes down once enforcement
182
Fogg
action begins. The owner will then simply incorporate as another entity
often selling the same products as before. If he fails to remit for that business
and enforcement action begin, he will start a third business. Meanwhile, his
payroll tax liabilities continue to pyramid higher”).
330
See Cal. Rev. & Tax. Code § 6701 (Deering 2009) (allowing bonding
requirement to lapse after entity has met its collected tax obligations for a
three year period).
331 The IRS has the ability to enjoin certain noncompliance with tax laws
currently using one of three provisions in the Internal Revenue Code: 1) IRC
section 7402 provides a broad injunction to enforce the tax laws. It is little
used because it is broad and relies on traditional equitable considerations
rather than specific statutory conditions. Occasionally, the IRS has
considered using this provision to seek to stop the pyramiding of collected
taxes but essentially has not done so because the statute is too unwieldy for
that purpose; 2) IRC section 7407 provides for an injunction against return
preparers—this injunction has been routinely used during the past decade;
3) IRC section 7408 provides for an injunction against specified activities
related to tax shelters and certain reportable transactions—this injunction
provision has also been routinely used during the past decade. See Rifkin,
supra note 287, at 402 (providing an overview of the injunctive remedies).
These provisions do not sufficiently address the pyramiding of taxes to fit
the purpose of this proposal. Only section 7402 could arguably apply and is
too slow and unwieldy. A narrowly tailored provision like section 7407 or
section 7408 is needed as well as an expedited path to obtaining
an injunction.
332 Summons enforcement occurs pursuant to IRC sections 7604 and 7402 (b).
The proceeding to enforce the summons is described as a summary
proceeding because of the relatively small amount of information the
government must put into evidence before the burden shifts to the taxpayer
and because of the expedited nature of the procedure. These are traits
needed for an effective injunction procedure with respect to taxpayers who
fail to timely post a bond because speed is necessary to prevent pyramiding
of taxes. See Internal Revenue Manual, available at http://www.irs.gov/irm/
part34/irm_34–006–003.html#d0e10 (describing summons procedures);
Department of Justice Summons Enforcement Manuel, available at http://
www.usdoj.gov/tax/readingroom/summonsmn/SumEnfMan_May2006.pdf
(describing summons enforcement).
333 Posting a bond could easily prove financially overwhelming for a small
business and could cause the business to fail if imposed. Small businesses
have a significant incentive not to have this requirement imposed because of
the greatly increased risk of failure.
Collecting Collected Taxes
334
183
26 U.S.C. § 31 (2010).
26 U.S.C. § 6672 (2010).
336 Morales v. United States, 805 F. Supp. 1062, 1067 n.5 (D. Puerto Rico 1992)
(“Where the employer withholds social security taxes but fails to pay over
the funds, the employee is not liable to the government for the amount of
the withheld taxes and is entitled to credit notwithstanding the employer’s
default…”).
337 IRC § 31 (a) (2009); Treas. Reg. § 1.31–1. See also IRS Publication 505 (Rev.
March 2009), available at www.irs.gov/pub/irs-pdf/p505.pdf (generally
explaining the posting of this credit).
338 26 U.S.C. § 31 (a) (2010).
339 See Morales, 805 F. Supp at 1067 n.5.
340 A similar loss of social security benefits exists already for self-employed
individuals. A self-employed individual who fails to correctly report selfemployment income before three years, three months and 15 days after the
close of the tax years loses all social security benefits on the self-employment
income even if it is later determined that such income exists and even if the
individual must pay the self-employment tax on said income. See 42 U.S.C.
§ 405 (2009); 20 C.F.R. 404.802 (2009); Maone v. United States, 212 Ct. Cl. 591,
1977 WL 25823 (Ct. Cl. 1977)).
To create rough parallel treatment between responsible officers, employees,
and self-employed individuals, responsible officers should lose the social
security taxes that are not remitted with the return when due.
341 26 U.S.C. § 6672 (2010).
342 26 U.S.C. § 6303 (2010).
343 Formerly, P–5–60.
344 P–5–14 states: “The withheld employment and income taxes or collected
excise taxes will be collected only once, whether from the business, or from
one or more of its responsible persons.”
345 See IRS Internal Revenue Manual § 1.2.14.1.3 (August 1, 2009), available at
http://www.irs.gov/irm/part1/irm_01–002–014.html (“The full unpaid trust
fund amount will be paid only once in a particular case…”); id. § 5.7.3.1 (8),
available at http://www.irs.gov/irm/part5/irm_05–007–003.html#d0e10
( “If, after the assertion of the TFRP, the corporation pays the delinquent
tax, the FFRP assessment will be abated.”); id. at § 5.17.7.1.9 (2), available
at http://www.irs.gov/irm/part5/irm_05–017–007.html; see generally
POSTN–124416–08, 2008 TNT 184–8.
335
184
346
Fogg
See IRS Internal Revenue Manual § 5.17.7.1.9 (August 1, 2009), available at
http://www.irs.gov/irm/part5/irm_05–017–007.html#d0e10, 5.7.3; IRS Service
Center Advisory 200026024, 2000 WL 331116108 (April 20, 2000).
347 See IRS Internal Revenue Manual § 5.17.7.1.9 (August 1, 2009), available at
http://www.irs.gov/irm/part5/irm_05–017–007.html.
348 Because the goal is to promote the responsible officers to come forward to
pay the liability, this proposal does not seek to reward those responsible
officers from whom the IRS collects the tax through some form of collection
action, including offset. The policy here would parallel the current law
concerning designation of payments. See Ida v. Commissioner, 108 F. Supp.2d
1181, 1183 (D. Kan. 2000) (“A taxpayer who “voluntarily” makes payments
to the IRS has a right to designate the tax liability to which the payment will
apply); O’Dell v. United States, 326 F.2d 451, 456 (10th Cir. 1964). When a
taxpayer makes a voluntary payment without directing the application of
the funds, the IRS may decide how to apply it. Liddon v. United States, 448
F.2d 509, 513 (5th Cir. 1971). However, where the taxpayer makes a payment
“involuntarily,” the IRS will decide how to apply the payment. Muntwyler v.
United States, 703 F.2d 1030, 1032 (7th Cir. 1983).
In Amos v. Commissioner, 47 T.C. 65, 69 (1966), the Tax Court defined
“involuntary payment.” If a taxpayer voluntarily pays the liability, the
taxpayer can designate how that payment is applied to any outstanding
liabilities. If the taxpayer does not voluntarily pay, the IRS can apply the
payment in a manner that preserves its best interest. Similarly, involuntary
payments of the collected tax by responsible officers would not trigger the
rewards this proposal seeks to provide to responsible officers who step
forward and satisfy the liability.
349 In addition to the issue of voluntary vs. involuntary payment of the liability
as a triggering mechanism for reaping the benefits of this statute, an issue
exists concerning the treatment of a volunteer who steps forward and pays
a portion but not all of the liability. The proposal suggests that responsible
officers who pay a portion of the liability do not garner the full benefit of the
repayment reward. These payments should post on a quarter-by-quarter
basis. To the extent that a responsible officer makes a payment that fully pays
one or more quarters but not the entire liability, then that individual should
receive the reward for the quarters that are fully paid. To the extent that
quarters are partially paid, the individual would receive no reward.
350 “The tax imposed by section 3101 shall be collected by the employer of the
taxpayer, by deducting the amount of the tax from the wages as and when
paid.” IRC § 3101 (a).
Collecting Collected Taxes
185
“Except as otherwise provided in this section, every employer making
payment of wages shall deduct and withhold upon such wages a tax.”
IRC § 3402 (a).
351 “Whenever any person is required to collect or withhold any internal revenue
tax from any other person and to pay over such tax to the United States, the
amount of tax so collected or withheld shall be held to be a special fund in
trust for the United States.” IRC § 7501 (a).
352 This paper does not address criminal sanctions available to the IRS for failure
to pay these taxes. These criminal provisions receive so little use that a
citizen of the United States stands a much greater likelihood of being struck
by lightning than being prosecuted under one of these provisions. About
400 people each year are struck by lightning in the United States. http://
www.llightningsafety.noaa.gov/medical.htm. Almost no one is prosecuted
for the crime described by this section. See Wilson v. United States, 250
F.2d 312, 314 (9th Cir. 1957) (Court stated that there “does not appear to be
a single [prior] reported decision involving a felony prosecution for failure
to pay withholding taxes.”); United States v. Poll, 521 F.2d 329, 334 n.3 (9th
Cir. 1975), cert. denied, 429 U.S. 977 (1976) (After citing two other cases of
felony prosecution for withholding tax violations, the court stated that “[t]o
our knowledge these are the sum of the reported prosecutions under 7202 as
applied to withholding taxes.”).
353 6672 serves as a collection device because of the policy adopted by the
IRS regarding this liability. That policy, set out originally in P–5–60, is
discussed below. However, taxpayers must be careful to properly account
for withholding to ensure the withholding credit they receive matches the
amount of tax owed. See It’s Time to Adjust Withholding, but Can You Do
the Calculations?, http://mauledagain.blogspot.com/2008_04_01_archive.
html (Apr. 4, 2008, 8:39 EST).
354 See supra note 350.
355 Numerous articles have been written on the trust fund recovery penalty
of 6672. It is also the single most litigated federal tax issue litigated in the
refund context. For a general overview of the statute see John W. Schmehl &
Richard L. Fox, Responsible Person and Lender Liability for Trust Fund
Taxes—§§ 6672 and 3505, 639–2d Tax Mgt. (BNA) A–45 (2000); see also P.
Prestin Weidner, Note, The Misappropriation of Trust Fund Taxes Under the
Guise of Reasonable Cause, 57 Vand. L. Rev. 287 (2004); Moran, supra note
261, at 721–22 n.36.
356 As mentioned previously this article does not seek to address whether a
particular individual fits the responsible officer definition. This article
presumes that a responsible officer exists and moves forward from that point.
186
Fogg
Although the discussion in this section provides background information
about a “typical” situation, numerous reasons for not paying the trust fund
taxes exists. Nothing in this article seeks to portray the individuals held
responsible as miscreants or evil doers. Some individuals who do not pay the
trust fund taxes do so with bad motives knowing that their actions seek to
deprive the government of the trust fund taxes their business has collected.
Many times, however, the person who ends up responsible has a good faith
belief that the taxes will eventually be paid and just misjudges the economics
of the situation. Numerous articles exist discussing whether someone
meets the statutory criteria for assessment. See Doreen McCall, Who is a
“Responsible Person”—The Overreaching Power of the Internal Revenue Service
to Collect Employer Withholding Taxes, 18 Ohio N.U. L. Rev. 905 (1992); Mary
A. Bedikian, The Pernicious Reach of 26 U.S.C. Section 6672, 13 Va. Tax Rev.
225 (1993). This article starts at a different point and concerns only how the
liability should attach once the determination of liability has occurred.
357 A discussion of the collection process is provided in the U.S. Gov’t
Accountability Office, GAO–08–728, Tax Debt Collection: IRS Has
a Complex Process to Attempt to Collect Billions of Dollars in
Unpaid Tax Debts, 8–13 (2008).
358 Brown v. United States, 591 F.2d 1136 (5th Cir. 1979); Turner v. United States,
423 F.2d 448, 449 (9th Cir. 1970); Bowlen v. United States, 956 F.2d 723, 728
(7th Cir. 1992).
359 P–5–14 states: “The withheld employment and income taxes or collected
excise taxes will be collected only once, whether from the business, or from
one or more of its responsible persons.” IRS Internal Revenue Manual
1.2.14.1.3 (June 9, 2003). The IRS renumbered P–5–60 in 2003 to P–5–14.
360 IRS Internal Revenue Manual 5.7.4.4 (Apr. 13, 2006), 5.7.3.1 (Oct. 30, 2007).
361 Taxpayer Bill of Rights II Legislative History, Pub. L. No. 104–168, § 903, 1996
U.S.C.C.A.N. 1143, 1163.
362 IRS Internal Revenue Manual 5.7.7.7, 5.7.7.7.2 (Apr. 13, 2006).
363 IRS Internal Revenue Manual 5.17.7.1.9 (Nov. 02, 2007), 5.7.7.3 (Apr. 13,
2006). IRS Service Center Advisory 200026024, 2000 WL 33116108, April
20, 2000. This policy also promotes the same tactics of running, hiding and
delaying, attributed here to the failure to charge interest, when more than one
responsible officer exists. A detailed discussion of the effect of this policy on
compliance exceeds the scope of this article but an example demonstrates
why this policy promotes delay. Assume ABC Corporation withholds
$100,000 in income taxes which it fails to pay over to the Government. ABC
goes out of business without paying this debt. Bob, Mary and John are the
responsible officers of ABC and on July 1 each are assessed a $100,000 liability
Collecting Collected Taxes
187
based on 6672. Bob fully pays the liability on July 5. Mary fully pays the
liability on July 6. John fully pays the liability on July 7. The IRS will keep
Bob’s money and refund to Mary and John all of the money that they paid.
Since this occurred after the passage of 6672 (d) in 1996, Bob has the right
to sue Mary and John for contribution. He will probably have to bear the
cost of that litigation as well as the risk associated with collecting upon any
judgment he obtains. This policy does not entice responsible officers to step
forward with payment but rather to stand back waiting and hoping that one
of the others will pay willingly or by enforced collection. For a discussion
of the intersection of incentives and tax compliance see Leslie Book,
Freakonomics and the Tax Gap: An Applied Perspective, 56 Am. U. L. Rev.
1163.(2007); Lederman, supra note 256.
364 The delay in IRS entry onto the scene is spelled out in detail in U.S. Gov’t
Accountability Office, GAO–08–617, Tax Compliance: Businesses
Owe Billions in Federal Payroll Taxes, 7 (2008). The consequences
of the delay are spelled out in 2006 Nat’l Taxpayer Advocate Ann. Rep.
vol. 1, at 69.
365 See David M. Richardson, Jerome Borison & Steve Johnson, Civil Tax
Procedure Ch. 14 (LexisNexis 2d ed. 2008) (2005); Michael I. Saltzman,
IRS Practice and Procedure, 17.07–09 (Warren Gorham Lamont 1991)
(1981); Doreen McCall, Who is a “Responsible Person”—The Overreaching
Power of the Internal Revenue Service to Collect Employer Withholding Taxes,
18 Ohio N.U. L. Rev. 905 (1992); Mary A. Bedikian, The Pernicious Reach of
26 U.S.C. Section 6672, 13 Va. Tax Rev. 225 (1993).
366 The current interest rate does not use simple interest but interest that
compounds daily so, if the assumed interest rate is correct, the total interest
in a real case would, of course, be higher. IRC § 6622 (a ).
367 As seen below in the section discussing state laws on this issue, the
administration of this issue by the IRS will also prove more difficult in most
states if the taxpayer or the taxpayer’s representative is well informed since
the first payments will go to the state to stop the running of interest and
penalties there rather than to the IRS.
368 “That if any cosigner, seller … or other person … shall be guilty of any willful
act or omission by means whereof the United States shall or may be deprived
of the lawful duties, or any portion thereof, accruing upon the merchandise
… such person or persons shall, upon conviction, be fined for each offense …
or be imprisoned … or both.”
Corporate Excise Tax Act of 1909, Pub. L. No. 5, 36 Stat. 11, 97 (1909).
188
369
Fogg
Act of Sept. 8, 1916, Pub. L. No. 271 ch. 463, § 16, 39 Stat. 756, 773–75. For a
detailed discussion of the legislative history of 6672 see Moran, supra note
261, at 723–53.
370 Act of Sept 8, 1916, Pub. L. No. 271 §§ 200,300,400,500, 39 Stat. 756, 777, 780,
783, 793.
371 Act of Oct. 3, 1917, ch. 63, § 1004, 40 Stat. 300, 325–26.
372 Revenue Act of 1918, Pub. L. No. 254 ch. 18, § 1308 (c), 40 Stat. 1057, 1143.
373 Id. § 1308 (a):
“That any person required under Titles V, VI, VII, VII, IX, X, or Xii, to
pay, or to collect, account for and pay over any tax, or required by law
or regulations made under authority thereof to make a return or supply
any information for the purposes of the computation, assessment or
collection of any such tax, who fails to pay, collect, or truly account for
and pay over any such tax, make any such return or supply any such
information at the time or times required by law or regulation shall in
addition to other penalties provided by law be subject to a penalty of
not more than $1,000.”
374 Id. § 1308 (b).
375 Id. § 1308 (c). The statute also contains a definitional provision similar to
current section 6672:
“The term person as used in this section includes an officer or employee
of a corporation or a member or employee of a partnership, who as
such officer, employee, or member is under a duty to perform the act in
respect of which the violation occurs.”
Id. § 1308 (d).
376 Revenue Act of 1924, Pub. L. No. 176 ch. 234, 43 Stat. 253.
377 Moran, supra note 261, at 740–41. Revenue Act of 1928, Pub. L. No. 562 ch.
852, § 146, 45 Stat. 791, 835; See also Revenue Act of 1924, Pub. L. No. 176, ch.
234, § 1017 (d), 43 Stat. 253, 344. Section 1017 (d) provides:
“Any person who willfully fails to pay, collect or truthfully account for
and pay over, any tax imposed by Titles IV, V, VI, VII, and VIII, or
willfully attempts in any manner to evade or defeat any such tax or
the payment thereof, shall, in addition to other penalties provided by
law, be liable to a penalty of the amount of the tax evaded, or not paid,
collected or accounted for and paid over, to be assessed and collected in
the same manner as taxes are assessed and collected.”
378 Act of Aug. 14, 1935, Pub. L. No. 271, ch.531, § 807 (c), 49 Stat. 620, 638 stating:
Collecting Collected Taxes
189
“All provisions of law, including penalties, applicable with respect to any
tax imposed by section 600 [excise tax provisions] … of the Revenue
Act of 1926, … shall, insofar as applicable and not inconsistent with the
provisions of this title, be applicable with respect to the taxes imposed
by this title.”
379 Moran, supra note 261, at 747.
380 Revenue Act of 1934, Pub. L. No. 216 ch. 277, § 607, 48 Stat. 680, 768.
381 S. Rep. No. 73–558, at 53 (1934). The Supreme Court interpreted the scope
of this trust and detailed the history behind the creation of the provision in
Begier v. Internal Revenue Service, 496 U.S. 53 (1990).
382 Current Tax Payment Act of 1943, Pub. L. No. 68, ch. 120, 57 Stat. 126.
383 Moran, supra note 261, at 748. As noted in Professor Moran’s article at
footnote 261, no judicial interpretation of the penalty for collected taxes had
yet occurred.
384 See Regan & Co., Inc. v. United States, 290 F.Supp. 470, 479–480 (E.D.N.Y.
1968) for a discussion of the broad scope of 6672 as it tied together more
narrowly crafted statutes imposing similar liabilities in piecemeal fashion.
385 Int. Rev. Code of 1954, Pub. L. No. 591, 68A Stat. 3, 828. At the same
time section 7202 was enacted in Subtitle F, subpart A—Crimes creating a
criminal liability for similar conduct but with penalty provisions that did not
incorporate the 100 percent liability for the unpaid tax. Id. at 851. Moran,
supra note 261, at 750.
386 Moran, supra note 261, at 750; H.R. Rep. No. 83–1337, (1954), as reprinted in,
1954 U.S.C.C.A.N. 4017, 4025 “This revision includes a rearrangement of the
provisions to place them in more logical sequence, the deletion of obsolete
material, and an attempt to express the internal revenue laws in a more
understandable manner.”
387 IRS Chief Counsel Advice 200112003, 2001 WL 283666 (Nov. 28, 2000); Sage
v. United States, 908 F.2d 18 (5th Cir. 1990).
388 IRC § 6601 (e) (2) provides that interest is only imposed on an assessable
penalty if the person assessed such penalty fails to pay the liability after
receiving notice and demand.
389 Collier on Bankruptcy Ch. 507 (Alan N. Resnick & Henry J. Sommer
eds., 15th ed. 2008); Michael Herbert, Understanding Bankruptcy, Ch.
10.04 (Bender 2000).
390 Collier on Bankruptcy Ch. 523 (Alan N. Resnick & Henry J. Sommer
eds., 15th ed. 2008); David Epstein, Bankruptcy and Related Law in a
Nutshell, Ch. XVII, Sec. B (West 2005) (2002).
190
391
Fogg
Letter from David Lindsay, Assistant to the Secretary of the Treasury, to
Emanuel Celler, Chairman of the House Committee on the Judiciary (June
24, 1959) in H.R. Rep. No. 86–735, at 6 (1959); Letter from Stanley Surrey,
Assistant Secretary of the Treasury, to James Eastland, Chairman of the
Senate Committee on the Judiciary (July 14, 1961) in S. Rep. No. 89–114, at
7 (1965); Letter from Stanley Surrey, Assistant Secretary of the Treasury, to
James Eastland, Chairman of the Senate Committee on the Judiciary (Aug.
24, 1961) in S. Rep. No. 89–114, at 10 (1965).
392 B.C. 507 (a) (8) (C) provides priority status for collected taxes, “a tax required
to be collected or withheld and for which the debtor is liable in whatever
capacity.” 11 U.S.C.A. § 507 (a) (8) (C) (West 2004). This discussion focuses
on federal tax liability; however essentially the same results would occur with
respect to state tax liabilities. Subsequent references to “B.C.” refer to the
Bankruptcy Code as enacted in Title 11 of the United States Code.
393 11 U.S.C.A. §§ 726 (a) (4), 724 (a) (West 2004).
394 11 U.S.C.A. § 726 (a) (4) (West 2004).
395 11 U.S.C.A. §724 (a) (West 2004).
396 B.C. 523 (a) (1) (A) provides an exception to discharge for collected
taxes because collected taxes receive priority treatment pursuant to
B.C. 507. Because assessable penalties, other than 6672, do not receive
priority treatment pursuant to B.C. 507, they do not meet the test of B.C.
523 (a) (1) (A). They also do not meet the tests of B.C. 523 (a) (1) (B) or (C).
Assessable penalties do meet the test of B.C. 523 (a) (7). Only one reported
decision specifically addresses the application of the exception to discharge
to assessable penalties other than 6672. This decision was subsequently
withdrawn. Nielsen v. United States, No. 3–88–3164–H, 1991 WL 101552
(N.D. Tex. 1991), withdrawn, Nielson v. United States, No. 3–88–3164–H, 1991
WL 107412 (N.D. Tex. 1991); IRS Litigation Guideline Mem. GL–36, Effect of
Bankruptcy Case upon IRC 6700, 6701, and 7408 (Apr. 24, 1998).
397 S. Rep. No. 95–989, at 14 (1978).
398 Letter from David Lindsay, Assistant to the Secretary of the Treasury, to
Emanuel Celler, Chairman of the House Committee on the Judiciary (June
24, 1959) in H.R. Rep. No. 86–735, at 6 (1959); Letter from Stanley Surrey,
Assistant Secretary of the Treasury, to James Eastland, Chairman of the
Senate Committee on the Judiciary (July 14, 1961) in S. Rep. No. 89–114, at
7 (1965); Letter from Stanley Surrey, Assistant Secretary of the Treasury, to
James Eastland, Chairman of the Senate Committee on the Judiciary (Aug.
24, 1961) in S. Rep. No. 89–114, at 10 (1965); H.R. 2236, 86th Cong. (1959); S.
976, 89th Cong. (1965).
399 S. Rep. No. 89–114, at 10 (1965). See also H.R. Rep. No. 88–372, at 6 (1963).
Collecting Collected Taxes
191
400
S. Rep. No. 89–114, at 16–18 (1965).
H.R. 3438, 89th Cong., 80 Stat. 270 (1966).
402 H.R. Rep. No. 88–372, at 1 (1963).
401
403
Id. at 5.
S. Rep. No. 88–1134, at 1, 6 (1964); S. Rep. No. 89–114, at 6 (1965); Brief of
Petitioner-Appellant at 17–18, United States v. Sotelo, 436 U.S. 268 (1978) (No.
76–1800).
405 S. Rep. No. 95–989, at 1–4 (1978).
406 Congress created a commission to review the bankruptcy laws and make
recommendations. S.J. Res. 88, 91st Cong., 84 Stat. 468 (1970). That
commissions initial recommendation concerning taxes proposed a very
limited exception to discharge for taxes including collected taxes. “The
principal revisions are, first, the reduction from three years to one year of
the time period for the non-dischargeability and priority of tax debts, and,
second, the shift from reference to ‘due and owing’ and ‘assessed’ to special
rules tailored to major categories of the debts.” H.R. Doc. No. 93–137 Part II,
at 138 (1973).
407 S. Rep. No. 95–989, at 14 (1978); H.R. Rep. No. 95–595, at 191 (1977).
408 Basically, all other prepetition unsecured taxes received a time limited grant
of priority status and a time limited exception to discharge. See 11 U.S.C.A.
§§ 507 (a) (8), 523 (a) (1) (West 2004, Supp. 2008).
409 See 507 (a) (8) (A) for income taxes; (D) for employment taxes; and (E) for
excise taxes. 11 U.S.C.A. §507 (a) (8) (A), (D), (E) (West 2004, Supp. 2008).
With respect to each type of tax the time limit is generally three years from
the due date of the return or the event giving rise to the tax liability.
410 11 U.S.C.A. §1328 (a) (West Supp. 2008); H.R. Rep. No. 109–31, at 101 (2005).
411 102 B.R. 790 (E.D. Wash. 1989), aff ’d per curiam, 907 F.2d 114 (9th Cir. 1990).
412 Tomlan, 102 B.R. at 796.
413 IRS Litigation Guideline Mem. GL–37, Dischargeability of Untimely Filed
Liabilities in Chapter 13 Bankruptcies (Apr. 7, 1992) (obsoleted Jan. 13, 1998).
414 Jack Williams, A Comment on the Tax Provisions of the National Bankruptcy
Review Commission Report: The Good, The Bad and the Ugly, 5 Am. Bank.
Inst. L. Rev. 445, 455 (1997).
415 Bankruptcy Abuse Prevention and Consumer Protection Act of 2005, § 707,
119 Stat. 23, 126; 11 U.S.C.A. §1328 (a) (West Supp. 2008).
416 Policy Statement P–5–14 was renumbered and slightly rewritten in 2003.
Prior to that it was Policy Statement P–5–60.
417 IRM 1.2.14.1.3 (June 9, 2003).
404
192
Fogg
418
IRS Policy Statement P–5–60, MT 1218–56 (Approved Nov. 5, 1956); see
McCarty v. United States, 437 F.2d 961 (Ct. Cl. 1971) (discussing related
Internal Memorandum No. 56–46).
419 Slodov v. United States, 436 U.S. 238, 245–46 (1978).
420
United States v. Sotelo, 436 U.S. 268, 280 n.12 (1978).
IRS Service Center Advisory, 2000 WL 33116108 (June 30, 2000).
422 See Bryan T. Camp, Avoiding the Ex Post Facto Slippery Slope of Deer Park,
3 Am. Bank. Inst. L. Rev. 329, 330–32 (1995) for a general discussion of the
nature of the 6672 liability and how the IRS seeks to collect it.
423 IRS Service Center Advisory, 2000 WL 33116108 (June 30, 2000).
424 IRS Service Center Advisory, 2000 WL 33116108 (June 30, 2000).
425 United States v. Sotelo, 436 U.S. 268 (1978).
426 In re Sotelo, 551 F.2d 1090 (7th Cir. 1977).
427 Sotelo, 436 U.S. at 271; Brief of Respondent-Appellee at 5, United States v.
Sotelo, 436 U.S. 268 (1978) (No. 76–1800).
428 Brief for the Respondent, supra note 427, at 8.
429 “The fact that respondent was found liable under 6672 necessarily means
that he was ‘required to collect, truthfully account for, and pay over’ the
withholding taxes, and that he willfully failed to meet one or more of these
obligations.” Sotelo, 436 U.S. at 274; “It is therefore clear that the 6672 liability
was not imposed for a failure on the part of respondent to collect taxes but
was rather imposed for his failure to pay over taxes that he was required
both to collect and to pay over. Under these circumstances, the most natural
reading of the statutory language leads to the conclusion that respondent
‘collected or withheld’ the taxes within the meaning of Bankruptcy Act
17a (1) 9e).” Sotelo, 436 U.S. at 275.
430 “The funds here involved were unquestionable ‘taxes’ at the time they were
‘collected or withheld from others’… . That the funds due are referred to as
a ‘penalty’ when the government later seeks to recover them does not alter
their essential character as taxes for purposes of the Bankruptcy Act… .”
Sotelo, 436 U.S. at 275.
431 Id. at 279.
432 Id. at 280.
433 Id. at 280, quoting United States Comptroller General Opinion, B–137762
(May 3, 1977), in 9 Standard Federal Tax Reporter, ¶ 6614, (CCH) 71,438
(1977).
434 Sotelo, 436 U.S. at 287.
435 Id. at 288.
421
Collecting Collected Taxes
436
193
“[T]he lifelong liability which the Court imposes today falls on the shoulders
of one who was the chief executive officer of a small family business… .”
Sotelo, 436 U.S. at 290–91 (Emphasis added).
437 “Statement by the Hon. Dennis DeConcini, Chairman of the subcommittee
on improvements in judicial machinery of the Senate Committee on
the Judiciary, upon introducing the Senate Amendment to the House
Amendment to H.R. 8200 … Taxes which the debtor was required by law
to withhold or collect from others and for which he is liable in any capacity,
regardless of the age of the tax claims … In addition, this category includes
the liability of a responsible officer under the Internal Revenue Code
(sec. 6672) … and the priority will cover the debtor’s responsible officer
liability regardless of the age of the tax year to which the tax relates. The
U.S. Supreme Court has interpreted present law to require the same result
as will be reached under this rule. U.S. v. Sotelo, 436 U.S.268 (1978). This
category also includes the liability under section 3505 of the Internal Revenue
Code of a taxpayer who loans money for the payment of wages or other
compensation.” (Emphasis added). 1978 U.S.C.C.A.N. 6505, 6566.
438 Lauckner v. United States, No. 93–1594, 1994 WL 837464, at *7 (D.N.J. May 4,
1994), aff ’d per curiam, 68 F.3d 69 (3d Cir. 1995).
439 Id.
440 See IRS CCA 200112003 (March 23, 2001) for a general discussion of the
statute of limitations on penalties in subchapter 68B and a specific discussion
of whether a statute of limitation on assessment of the penalty imposed under
6707 exists.
441 United States v. Hodgekins, 805 F. Supp. 653 (N.D. Ind. 1992); Turk v. United
States, No. S92–307M, 1993 WL 497785, at *6 (N.D. Ind. Sept. 17, 1993);
Stallard v. United States, 12 F.3d 489 (5th Cir. 1994).
442 See Mullikin v. United States, 952 F.2d 920 (6th Cir. 1991); Lamb v. United
States, 977 F.2d 1296 (8th Cir. 1992); Capozzi v. United States, 980 F.2d 872
(2d Cir. 1992); Sage v. United States, 908 F.2d 18 (5th Cir. 1990).
443 Lauckner v. United States, No. 93–1594, 1994 WL 837464, at *7 (D.N.J. May 4,
1994), aff ’d per curiam, 68 F.3d 69 (3d Cir. 1995).
444 Numerous authorities exist for the proposition that the liability under 6672
is separate and distinct from the liability of the entity for the collected taxes.
None of the authorities set the issue up in quite the way that Bradley v. United
States, 936 F.2d 707 (2nd Cir. 1991) does. In Bradley the IRS assessed 6672
liabilities against two individuals, Charles Bradley and David Agnew for
failure of Maxim Industries, Inc. (Maxim) to pay withheld employment and
social security taxes. After the 6672 assessments were made against Bradley
and Agnew they paid a portion of the tax, filed a claim for refund and then
194
Fogg
filed suit. Also after the 6672 assessments were made, Maxim filed a chapter
11 bankruptcy petition. Because it appeared that Maxim might have sufficient
funds to fully pay the outstanding employment tax liability through the
bankruptcy case, the parties in the refund suit agreed to dismiss the refund
suit subject to reinstatement. Maxim did pay all of the employment taxes
through the bankruptcy case together with all of the interest for which it was
liable; however, because interest does not accrue in chapter 11 bankruptcy
cases on unsecured claims between the date of the petition and the
confirmation of the plan, the IRS sought to collect the interest for this period
of time from the two responsible officers. They resisted and the refund suit
was reinstated setting up the issue of whether the IRS could do so.
The Second Circuit laid out the issue: “Essentially, plaintiffs contend
that since Maxim has paid its tax liability and related interest, the Internal
Revenue Code provides no authority for charging plaintiffs with interest
for the period during which Maxim was in bankruptcy. This argument
mischaracterizes the legal basis for the assessments against plaintiffs. Strictly
speaking, liability under section 6672 (a) is not derived from, or dependent
upon, an employer’s outstanding tax obligation. Rather, the section imposes
a penalty upon persons who fail to perform a specified statutory task. We
have consistently held, therefore, that the liability for such a penalty is
separate and distinct from the employer’s liability for trust fund taxes.” Id.
at 710.
445 Lauckner, 1994 WL 837464 at *1. The similarities noted by the Court between
6672 and 3403 parallel the similarities in the treatment of these liabilities in
the bankruptcy code. These similarities for the basis for the Government’s
policy decision adopting Policy Statement 5–60 in which it states the liability
will only be collected once. The separateness discussed in Bradley, however,
does seem more separate than the discussion in Lauckner addresses.
446 Id. at *4.
447 Id. at *5.
448 Id.
449 Id.
450 Id.
451 See the argument in Brief of Petitioner-Appellant at 25, n.15, United States
v. Sotelo, 436 U.S. 268 (1978) (No. 76–1800). “Liability for taxes under 6672
is deemed ‘due and owing’ on the date the person responsible for seeing the
taxes are paid failed to do so—the date the corporate returns were due to be
filed.”
452 The IRS made additional arguments in Lauckner based on the relevant return
for purposes of IRC 6501 (a) and Congressional intent. These arguments
Collecting Collected Taxes
195
were also rejected with the reasoning that covers matters not related to this
paper.
453 IRS Action on Decision 1996–006, 1996–2 C.B. 1.
454 Lauckner, 1994 WL 837464 at *1. In the second paragraph of the opinion the
court clearly expresses its concern that the IRS position reflected a reversal
of its long held position concerning the applicable state of limitations. “It
argues for perpetual exposure despite its long-standing position to the
contrary, coupled with judicial acceptance and congressional acquiescence
for more than 30 years. Such a radical change must come from the legislature
and not the courts, particularly where it seeks to leave persons exposed to tax
liability in perpetuity.”
455 Treas. Reg. § 301.6601–1 (a) (as amended in 1997).
456 IRC § 6151 (a).
457 IRC § 6072 (a).
458 Treas. Reg. § 301.6601–1 (f) (3) (as amended in 1997).
459 IRC 6303; Treas. Reg. 301.6303–1 (as amended in 2001).
460 See Thomas E. Fritz, Flowthrough Entities and the Self-Employment Tax: Is it
Time for a Uniform Standard?, 17 Va. Tax Rev. 811 (1997–1998); Fred B. Brown,
Federal Income Taxation of US Branches of Foreign Corporations: Separate
Entity or Separate Rules?, 49 Tax L. Rev. 133 (1993).
461 United States v. Sotelo, 436 U.S. 268, 281 (1978) (quoting 112 Cong. Rec.
13809, 13817 (1966)).
462 Sotelo, 436 U.S. at 281 n.16. (internal citations omitted).
463 Much has been written on the role of horizontal equity and parallelism in
tax legislation. These concepts are important because taxpayer perceptions
are important. Enforced compliance measures by the IRS cannot account
for the level of compliance by the taxpaying public. Fairness in the system
is critical. Jeffrey H. Kahn, The Mirage of Equivalence and the Ethereal
Principles of Parallelism and Horizontal Equity, 57 Hastings L.J. 645
(2006). In this article, Professor Kahn devises a test to determine when
parallel treatment of a specific tax circumstance is desirable and when
countervailing considerations drive nonparallel treatment as the correct
result. He did not test this situation. Using his tests a strong argument exists
for parallel treatment with respect to interest between individuals who fail
to pay over monies held in trust for the Government. These individuals
whether operating as a sole proprietorship, partnership, or in corporate form
have already received parallel treatment—the very reason for piercing the
corporate veil. No reason exists for departing from that parallel treatment
in the imposition of interest. See also Dave Elkins, Horizontal Equity as
196
Fogg
a Principle of Tax Theory, 24 Yale L. & Pol’y Rev. 43 (2006); Richard
Winchester, The Gap in the Tax Gap: What Congress Should Do About It
(Thomas Jefferson School of Law Legal Studies Research Paper No. 1151363),
available at http://ssrn.com/abstract=1151363 (addressing parallelism in
employment tax issues).
464 Leandra Lederman, The Interplay Between Norms and Enforcement in Tax
Compliance, 64 Ohio St. L.J. 1453 (2003).
465 See United States v. Estate of Kime, 950 F.Supp. 950, 954 (D. Neb. 1996)
(finding the Insolvency Statute holds a representative of an estate liable for
the unpaid tax liability, interest, and penalties of the estate); United States
v. Coppola, 85 F.3d 1015, 1020–21 (2d Cir. 1996) (limiting the liability of the
representative to the amount of the payment made or the value of the assets
distributed before taxes are paid; importantly, the court found that the
executor was liable for the unpaid taxes plus interest under 31 U.S.C. 3713).
466 Papineau v. Commissioner, 28 T.C. 54 (1957); Yagoda v. Commissioner, 39 T.C.
170 (1962), aff ’d, 331 F.2d 485 (2d Cir. 1964), cert. denied, 379 U.S. 842 (1964).
The cap on the liability described here applies to transferee cases “in equity.”
Generally, no cap exists for transferee cases “at law.”
467 Lowy v. Commissioner, 35 T.C. 393 (1960). For a general discussion on the
issue of interest on transferee liability see Theodore D. Peyser, Transferee
Liability, 628–2d Tax Mgt. (BNA) A–25–27 (2003); Michael I. Saltzman,
IRS Practice and Procedure, 17.06 (Warren Gorham Lamont 1991) (1981).
468 Estate of Stein v. Commissioner, 37 T.C. 945 (1962), supp. op., 40 T.C. 275
(1963).
469 Patterson v. Sims, 281 F.2d 577 (5th Cir. 1960). The courts are split on
the liability for accrued interest if the underlying tax is an estate or gift
tax. Compare Baptiste v. Commissioner, 29 F.3d 433 (8th Cir. 1994), cert.
denied, 513 U.S. 1190 (1995) with Baptiste v. Commissioner, 29 F.3d 1533
(11th Cir. 1994).
470 See e.g., IRC § 31. Section 31 (a) provides that “the amount withheld …
shall be allowed to the recipient of the income as a credit against the tax
imposed by the subtitle.” This provision insures that a worker whose wages
have been reduced by the amount of the withheld taxes will receive credit
for payment of those taxes even if the company that withheld the taxes fails
to pay them over to the Government. This credit extends even to the taxes
withheld on the wages of individuals determined to be responsible for failure
to pay over the withheld taxes. As a consequence, the Government, through
this provision, grants full value for the withheld taxes whether it receives
that value or not. This granting of full value provides an equivalent to the
Collecting Collected Taxes
197
transferee who has received from the taxpayer material equal to or greater
than the value of the taxes owed by the transferor.
471 See H.R. Rep. No. 89–1884, (1966), reprinted in 1966–2 C.B. 815, 828–30;
S.Rep. No. 89–1708, (1966), reprinted in 1966 U.S.C.C.A.N. (80 Stat. 1125)
3722, 3724, 3742–45 (one of the statements provided in the legislative history
here describes 3505 as “intended to represent a reasonable accommodation
of the interests of the Government in collecting the taxes of delinquent
taxpayers with the rights of taxpayers and third parties.”)
One commentator states that “prior to 1966 no lender or other institutional
creditor had ever been held liable for the 6672 penalty.” Larry A. Makel &
James C. Chadwick, Lender Liability for a Borrower’s Unpaid Payroll Taxes, 43
Bus. Law. 507, 520 n.56 (1988).
472 The term for this practice, net payroll financing, does not appear in the
statute but found common usage during the discussion of the need for this
provision as described in United States v. Algernon Blair, Inc., 441 F.2d 1379,
1381 (5th Cir. 1971):
“Prior to the effective date of 3505 (b), problems arose with the
construction industry’s device known as ‘net payroll financing’.
Using this method, a sub-contractor-employer, who had financially
overextended himself would go to a lender, in this case the prime
contractor, for financial assistance. The prime contractor-lender,
desirous of having the sub-contractor complete the work, but also
wanting to minimize costs would provide only the net payroll funds.
In many of these situations, the United States would never receive the
withholding taxes due, even though the employees received credit on
the records of the Treasury Department as if the taxes had been paid.
While the sub-contractor-employer would still be liable for the taxes
under 3102 (b) and 3404 of the Code, recourse against the employer
was often fruitless, as he was financially unable to pay the taxes.”
473 IRC § 3505 provides:
“(a) Direct payment by third parties. For purposes of sections 3102, 3202,
3402 and 3403, if a lender, surety, or other person, who is not an
employer under such sections with respect to an employee or group
of employees, pays wages directly to such an employee or group of
employees, employed by one or more employers, to an agent on behalf
of such employee or employees, such lender, surety, or other person
shall be liable in his own person and estate to the United States in a sum
equal to the taxes (together with interest) required to be deducted and
withheld from such wages by such employer.
198
Fogg
(b) Personal liability where funds are supplied. If a lender, surety, or other
person supplies funds to or for the account of an employer for the
specific purpose of paying wages of the employees of such employer,
with actual notice or knowledge (within the meaning of section
6323 (i) (1) that such employer does not intend to or will not be able to
make timely payment or deposit of the amounts of tax required by this
subtitle to be deducted and withheld by such employer from such wages,
such lender, surety, or other person shall be liable in his own person
and estate to the United States in a sum equal to the taxes (together with
interest) which are not paid over to the United States by such employer
with respect to such wages. However, the liability of such lender surety,
or other person shall be limited to an amount equal to 25 percent of the
amount so supplied to or for the account of such employer for
such purpose.
(c) Effect of payment. Any amounts paid to the United States pursuant
to this section shall be credited against the liability of the employer.”
(emphasis added).
474 “As of January, 1965, the delinquent accounts in the construction industry
totaled $55,608,622.00, which was twenty-six percent of all delinquencies of
this type of taxes. As of the same date, $29,730,508.00 of the $55,608,622.00,
or fifty-three percent, had been overdue for more than one year. And
according to the same Treasury figures, it had written off in 1964, as
uncollectible from the construction industry, the sum of $16,290,098.00,
which was twenty-eight percent of all unpaid withholding taxes for all
industries written off during that year.” Edward Gallagher, The Good and
the Bad for Surety Companies Under the Federal Tax Lien Act of 1966, 34 Ins.
Couns. J. 214, 218 (1967).
475 United States v. Hill, 368 F.2d 617, 623 (5th Cir. 1966)
(“[T]he bank agreed to loan to the corporation funds to complete
jobs in progress. The only control which the bank exercised
during this period was in connection with the funds which it
loaned. Taxpayer Hill admitted at trial that he understood the
bank’s refusal to honor checks for taxes drawn on these funds
to be merely a statement that the bank would not loan the
corporation funds for the taxes. Certainly this refusal to make
a loan did not place the bank in control of the corporation’s
checking account or alter appellants’ control of the corporation.”)
Girard Corn Trust Exchange Bank v. United States, 259 F.Supp. 214 (E.D.
Pa. 1966); United States v. Park Cities Bank & Trust Co., 481 F.2d 738 (5th Cir.
1973); United States v. Algernon Blair, Inc., 441 F.2d 1379 (5th Cir. 1979); United
States v. Coconut Grove Bank, 545 F.2d 502,505 n.2 (5th Cir. 1977) (lists the
Collecting Collected Taxes
199
cases). Many of the cases cited in this footnote and listed in Coconut Grove
Bank include additional arguments by the United States in its attempts to
hold the third party lenders liable. Seeking to hold the lender liable under
a contract theory and seeking to hold it liable as the “true” employer were
the primary additional theories. Those arguments are not important to the
purpose of this article.
476 See case quoted, supra, note 472.
477 Abrams v. United States, 333 F. Supp. 1134, 1147 (S.D. W. Va. 1971); A later
court set up a two part test that must be satisfied in order for a payment to
be considered a direct payment of wages under 3505 (a). First, the payor
must have the ability to control the funds. If the employer controls the funds
then the situation would shift from 3505 (a) to some other provision such as
3505 (b) or 6672. Second, the payor must have the right and legal authority to
exercise that control. United States v. Fred A. Arnold, Inc., 573 F.2d 605, 608
(9th Cir. 1978). See also United States v. Kennedy Construction Co. of NSB, 572
F.2d 492 (5th Cir. 1978); Derr v. United States, 498 F. Supp. 337 (W.D.
Wis. 1980).
478 H.R. Rep. No. 89–1884, (1966), reprinted in 1966–2 C.B. 815, 829; S. Rep. No.
89–1708, (1966), reprinted in 1966 U.S.C.C.A.N. (80 Stat. 1125) 3722, 3743.
479 Treas. Reg. § 31.3505–1 (d) (as amended in 1995) provides: “In the event
the lender, surety, or other person does not satisfy the liability imposed by
Section 3505, the United States may collect the liability by appropriate civil
proceedings commenced within 10 years after assessment of the tax against
the employer.”
480 In fact Congress anticipated that lenders would take certain precautions to
avoid the liability under this statute. “ ‘[S]ureties can protect themselves
against any losses attributable to withholding taxes by including this risk of
liability in establishing their premiums, and lenders by including the amounts
in their loans and taking adequate security.’ ” Jersey Shore State Bank v.
United States, 479 U.S. 442, 449 (1987), citing, S. Rep. No. 89–1708, (1966),
reprinted in 1966 U.S.C.C.A.N. 3722, 3744; H.R. Rep. 89–1884, reprinted in
1966–2 C.B. 815, 830.
481 See Julius Thannhauser et al., Lender’s Liability for Unpaid Withholding
Taxes of Borrower—Employer—IRC Sections 3505 and 6672, 80 Com. L.
J. 137 (1975); See also Mark R. Hinkston, Dealing with the Disarray: The
Eighth Circuit Addresses Notice and Demand Applicability to Lenders’
Liability For Withholding Taxes Under IRC 3505 (b)—United States v.
Messina Builders and Contractors Co., 20 Creighton L. Rev. 1093, 1099
n.26 (1987) (This article has an excellent introductory section on the
legislative history of 3505).
200
482
Fogg
The IRS took the position initially that interest due from the third party
under 3505 (b) added onto the 25 percent. Treas. Reg § 31.3505–1 (b) (as
amended in 1995); Rev. Proc. 78–13, 1978–1 C.B. 591. It lost this issue in three
circuits. See United States v. Metro Construction Co., 602 F.2d 879 (9th Cir.
1979); United States v. Intercontinental Industries, Inc., 635 F.2d 1215 (6th Cir.
1980); United States v. Hannan Co., 639 F.2d 284 (5th Cir. 1981). After these
losses, the IRS abandoned the position that the person liable under 3505 (b)
must pay interest in addition to the 25 percent of net payroll. See IRS
Litigation Guideline Mem. GL–14 (May 4, 1994).
483 Once the fact that funds are being used for net payroll is brought to the
attention of the lender, the lender is deemed to meet the knowledge part of
this test. United States v. Park Cities Bank & Trust Co., 481 F.2d 738, 740 (5th
Cir. 1973); United States v. Estate of Swan, 441 F.2d 1082, 1087 (5th Cir. 1971).
484 The statute references 6323 (i) (1) which provides “An organization exercises
due diligence if it maintains reasonable routines for communicating
significant information to the person conducting the transaction and there
is a reasonable compliance with the routines. Due diligence does not require
an individual acting for the organization to communicate information unless
such communication is part of his regular duties or unless he has reason to
know of the transaction and that the transaction would be materially affected
by the information.” IRC § 6323 (i).
485 United States v. First Nat’l Bank of Circle, 652 F.2d 882 (9th Cir. 1981).
486 United States v. Intercontinental Industries, 635 F.2d 1215 (6th Cir. 1980);
Fidelity Bank, N.A. v. United States, 616 F.2d 1181 (10th Cir. 1980).
487 Werner v. United States, 374 F. Supp. 558 (D. Conn. 1974), aff ’d, 512 F.2d 1381
(2d Cir. 1975).
488 See Jersey Shore State Bank v. United States, 479 U.S. 442, 446 (1987) (“Section
3505 does not declare that a lender is ‘liable for the unpaid tax.’ Instead,
the section imposes liability the lender for all or part of ‘a sum equal to the
taxes’ ”).
489 Id. at 446–47. Because the 3505 liability was not a tax but rather a judgment
for a sum certain based on the tax, the Supreme Court held that the IRS
was not required to follow all of the notice provisions set out in the Internal
Revenue Code for collection of taxes. Specifically, IRC 6303 requiring notice
and demand prior to collection did not apply to this situation.
490 See supra note 482.
491 422 F.2d 26 (9th Cir. 1970).
492 “Thus, in considering the application of section 3505, the possibility of also
asserting the trust fund recovery penalty against the lender or an employee of
Collecting Collected Taxes
201
the lender should not be overlooked.” IRS Litigation Guideline Mem. GL–14
(May 4, 1994).
493 See also Commonwealth Nat’l Bank of Dallas v. United States, 665 F.2d 743
(5th Cir. 1982); Regan & Co. v. United States, 290 F. Supp. 470, 479 (E.D.N.Y.
1968) (“[C]ongress sought to plug the loopholes against the limitless
ingenuity of those whose métier it is to search for crevices between mortise
and tenon in the infinitely complex definition and imposition of obligations
in the Revenue Code”).
494 “Section 6672 does not refer to any liability of a responsible person for
interest on the delinquent taxes. A responsible person has no liability for
interest on the unpaid withholding taxes to the extent that it accrues between
the date that the employee’s tax should have been paid and the date the
IRS assesses the tax against the responsible person. Hence, a potentially
responsible person has reason to pursue all good faith defenses through the
administrative process.” John W. Schmehl & Richard L. Fox, Responsible
Person and Lender Liability for Trust Fund Taxes - §§ 6672 and 3505,
639–2d Tax Mgt. (BNA) A–45 (2000) (emphasis added); see also David M.
Richardson, Jerome Borison & Steve Johnson, Civil Tax Procedure
400 (LexisNexis 2d ed. 2008) (2005). “An important advantage to protesting
the penalty before it is assessed is that doing so stays the assessment of
the penalty; consequently, interest does not begin to accrue.” David M.
Richardson, Jerome Borison & Steve Johnson, Civil Tax Procedure
Teacher’s Manual 284 (LexisNexis 2007) (emphasis added).
. As seen below in the section discussing state laws on this issue, the
administration of this issue by the IRS will also prove more difficult in most
states if the taxpayer or the taxpayer’s representative is well informed since
the first payments will go to the state to stop the running of interest and
penalties there rather than to the IRS.
495 See supra note 353.
496 IRC § 4251. In 2006, total collections for the telephone excise tax equaled
4.6 million dollars. IRS SOI Bulletin Historical Table 20, Federal Excise
Taxes Reported to or Collected by the Internal Revenue Service, Alcohol and
Tobacco Tax and Trade Bureau, and Customs Service, by Type of Excise Tax.
497 IRC § 4261.
498 IRC § 4081.
499 These three federal excise taxes operate to charge the consumer of the item
(telephone service, plane tickets or motor fuel) with a federal tax. The tax is
collected by the provider of the service or item purchased. The tax is held
in trust by the provider for the federal government. Similarly, a state sales
tax imposes a liability on a purchaser of goods or other taxable items. The
202
Fogg
purchaser pays the tax at the time of the purchase of the goods. The vendor
receives the payment for the tax and holds that payment in trust for the
state which requires payment to it at certain intervals. One major difference
between the federal excise taxes and the state sales taxes is the breadth
of business impacted by these taxes. The federal excise taxes fall upon a
relatively small number of business entities in very specific businesses. The
state sales taxes fall upon almost every retailer and the state use taxes fall
upon many other types of business providing a service. The numbers of the
businesses being charged with preservation of trust funds by the states makes
its scope much more like the federal withholding taxes and gives rise to a
larger body of law concerning the failure to pay over sales taxes than exists
with the failure to pay over federal excise taxes.
500 E.g. Texas, Washington, Florida, Alaska, South Dakota, Nevada, and
Wyoming. Also, New Hampshire and Tennessee limit income taxation to
interest and dividends.
501 E.g., Oregon, Alaska, Delaware, and New Hampshire.
502 E.g., Idaho, New York and West Virginia; possibly also South Carolina
503 All states except Wyoming have some form of trust fund regime imposing
personal liability on persons who fail to pay over to the state the taxes
collected on its behalf.
504 See Appendix D.
505 Id.
506 This is best illustrated through an example. Suppose that John Smith is the
responsible officer of Acme, Inc. a Pennsylvania corporation which failed
to pay over the income taxes it withheld for the first quarter of 2005 in the
amount of $50,000. If we assume that the IRS takes 12 months after the due
date of the Form 941 on April 30, 2005, to initiate its trust fund recovery
investigation and further assume that John Smith avails himself of the full
range of administrative remedies prior to assessment while responding to
the IRS at a very deliberate speed, it may be two years (April 30, 2007) after
the due date of the return before the IRS assesses the trust fund recovery
penalty against him. Assuming John failed to pay over collected taxes to
Pennsylvania of the same amount for the same period, on April 30, 2007,
John Smith will owe the IRS $50,000 and on that same date he will owe
the Pennsylvania $61,000 consisting of $50,000 in trust fund taxes, $5,000
in interest and $6,000 in penalties. Moving forward from April 30, 2007,
John will owe interest and failure to pay penalties on both the federal
and state liabilities; however, the interest will be on the higher amount of
the Pennsylvania liability causing him to accrue even more interest (and
penalty) expenses compared with his federal tax liability. See U.S. Gov’t
Collecting Collected Taxes
203
Accountability Office, GAO–08–617, Tax Compliance: Businesses
Owe Billions in Federal Payroll Taxes, 32–33 (2008), stating that “from
the time the tax debt was assessed against the business, IRS took over 2 years,
on average, to assess [a 6672 liability] against the business owners/officers.”
507
This is money that the federal government chooses not to seek even though
the parallel state statute seeks it for the most states. The fact that most states
are seeking to pick up this money does not compel the result that the federal
government should do likewise; however, in a time of looking about for tax
gap provisions, the practice of the overwhelming majority of the states on
this issue should at least provide some food for thought for those writing the
federal statutes with an eye toward more revenue. Another consideration for
imposing the tax could be whether this class of individuals is one deserving
of a break on interest or whether it is perhaps a class most deserving of
making the government whole on the time value of the revenue lost through
their actions.
508 This also poses room for thought when comparing the state and federal
provisions. If most states impose interest from the due date of the entity
return and the federal government imposes interest only upon assessment of
the responsible officer liability, what rational taxpayer would pay the federal
government first? In addition to the general incentive provided by IRC 6672 to
delay the assessment, the juxtaposition of the state and federal statutes causes
the responsible officer aware of the manner in which the two statutes operate
to use his first funds to pay down the state liability and stop the running of
interest. By the time the federal government comes into the mix, the available
funds from the responsible officer, which are generally not great to begin with,
are further depleted, leaving the federal government to scramble harder to
collect its trust fund liabilities once they are finally assessed.
509 Alaska, California, Illinois, Indiana, Iowa, Kansas, Louisiana, Maine,
Minnesota, Mississippi, Montana, Nebraska, Nevada, New Hampshire, North
Dakota, Oklahoma, Oregon, Pennsylvania, South Dakota, Tennessee, Texas,
Washington, Washington D.C., Wisconsin. See Appendix D.
510 Alabama, Kentucky, Maryland, Massachusetts, Michigan, New Mexico, North
Carolina, Ohio, Rhode Island, Vermont, Virginia. See Appendix D.
511 Garland v. Director of Revenue, 961 S.W. 2d 824 (Mo. 1998).
512 Id.
513 Two states do not charge interest back to the due date of the corporate return
and follow the Federal model:
Delaware—Del. Code Ann. tit. 30, § 535 (e) (1997) (Withholding): (No
sales tax in Delaware); E-mail from Randy R. Weller, Manager Bankruptcy/
Decedents, Delaware Division of Revenue, to T. Keith Fogg, Visiting
204
Fogg
Associate Professor of Law, Villanova University School of Law (Mar. 10,
2008, 8:32:50 EST).
Utah—Utah Code Ann. § 59–1–302 (2007) (Withholding and Sales);
Telephone Interview with Gale Francis, Assistant Attorney General, Utah
Attorney General’s Office, in Salt Lake City, Utah, Utah (March 14, 2008).
514 Id.
515 Id.
516 Two states do not charge interest back to the due date of the corporate return;
however, they charge the responsible officer with a “penalty” equal to 150
percent or 200 percent of the unpaid trust fund taxes:
Colorado—Colo. Rev. Stat. Ann. § 39–21–116.5 (West 2007) (Sales and
Withholding) (150 percent).
Florida—Fla. Stat. Ann. § 213.29 (West 2005) (Sales); (Florida does not have
an income tax).
517 Colo. Rev. Stat. Ann. § 39–21–116.5 (West 2007).
518 Fla. Stat. Ann. § 213.29 (West 2005).
519 The scheme used by Florida and Colorado appears to transform the liability
from a collected tax which would have priority under B.C. 507 (a) (8) (C) to a
general unsecured claim for a penalty.
520 Although not a part of this survey, liability for paying over collected taxes
exists at the local level as well. One locality imposing liability for failure
to pay over withholding taxes is Columbus, Ohio. Columbus charges the
responsible officer with interest and penalty due from the entity that incurred
the tax.
“The officer or the employee having control or supervision of or charged
with the responsibility of filing the report and making payment is
personally liable for failure to file the report or pay the tax due as
required by this section. The dissolution of a corporation does not
discharge an officer’s or employee’s liability for a prior failure of the
corporation to file returns or pay tax due.”
City of Columbus v. Mid-Ohio Canopies, Inc., No. 95APG06–685, 1995 Ohio
App. LEXIS 4964 (Ohio Ct. App. 1995).
521 Idaho, New York, West Virginia, South Carolina. See Appendix D.
522 Income Tax Act, R.S.C., § 153 (1) (1985).
523 Income Tax Act, R.S.C. § 227 (4) (1985).
524 Income Tax Act, R.S.C. § 227.1 (1) (9185) provides:
“Where a corporation has failed to deduct or withhold an amount as
required… , has failed to remit such an amount or has failed to pay
Collecting Collected Taxes
205
an amount of tax for a taxation year as required… , the directors of
the corporation at the time the corporation was required to deduct,
withhold, remit or pay the amount are jointly and severally, or solidarily,
liable, together with the corporation, to pay that amount and any
interest or penalties relating to it.”
See Soper v. The Queen, [1997] 3 C.T.C. 242 (Can.) (Discussion of legislative
history of 227.1); and Veilleux v. The Queen, [2001] 3 C.T.C. 288 (Can.).
Income Tax Act 227.1 (2) provides some limitations on this liability as
does section 227.1 (3) which states “A director is not liable for a failure
under subsection (1) where the director exercised the degree of care,
diligence and skill to prevent the failure that a reasonably prudent person
would have exercised in comparable circumstances.” Income Tax Act,
R.S.C. 227.1 (2), (3) (1985).
See Barnett v. Minister of National Revenue, [1985] 2 C.T.C. 2336 (Can.)
(Sole shareholder liable and unable to successfully interpose defense of
delegation to comptroller); Fraser v. Minister of National Revenue, [1987]
1 C.T.C. 2311 (Can.) (Director who was vice-President liable and unable
to successfully argue that another officer could more easily pay); Beutler
v. Minister of National Revenue, [1988] 1 C.T.C. 2414 (Can.) (Director and
President liable and efforts to satisfy arrearages did not relieve him
of liability).
525 Soper v. The Queen, [1997] 3 C.T.C. 242 (Can.).
526 Income Tax Act, R.S.C. § 227 (9) (1985).
527 Soper 3 C.T.C. at 250 (Can.).
528 Income Tax Act, R.S.C. § 227.1 (1) (1985).
529 Social Security Administration Act, 1992, ch. 5, § 121C (1) (b).
530 Social Security Administration Act, 1992, ch. 5, § 121C (3) (b); England also
imposes personal liability for evasion of Value Added Tax (VAT) if dishonest
conduct played a role in its nonpayment. Sections 60 & 61. This personal
liability also has a relative responsibility component. Section 61 (2).
531 Social Security Administration Act, 1992, ch. 5, § 121C (2) (b) (i) & (9).
532 Social Security Administration Act, 1992, ch. 5, § 121C (2) (b) (ii).
533 Income Tax Assessment Act, 1936, § 222AOB (Austl.).
534 Income Tax Assessment Act, 1936, § 222AOC (Austl.).
535 District Court Act, 1973, § 83A (Austl.).
Measuring and Tackling the
Illicit Market for Excise Goods
Anthony Rourke, Her Majesty’s Revenue and Customs
G
lobally, there are large price differentials in most goods due to the large
variation in production costs and tax rates. Therefore, the profit margins
that are available to individuals who can successfully smuggle goods from
low-cost to high-price countries can be very attractive. Likewise, the revenue
losses for the tax administrations of the high-price countries can be significant,
in particular where tax accounts for a large portion of the retail price of the illicit
goods. This problem can be exacerbated when the goods are easily portable, transportation costs are low, and border controls are less stringent between the lowcost and high-price countries, as can occur in free-trade zones. The task facing
every tax administration is to reduce its revenue losses by identifying and tackling
those involved in the illicit trade. However, in order to determine the appropriate response to the illicit market, it is first necessary to determine the scale of the
problem.
Techniques for estimating the size of the illicit market can be separated into
two general methodologies, with different methodologies used both within and
between different tax administrations.1, 2, 3, 4, 5, 6
Bottom-up methodologies base their estimates on operational activities, such
as enquiries into the tax returns submitted to HM Revenue and Customs (HMRC)
or seizures of illicit goods made by HMRC. The advantage of this approach is
that the calculations are not dependent on any external third-party sources and so
HMRC has more control over the consistency of the data used to estimate the illicit markets. However, the disadvantage is that the calculations are dependent on
operational strategies and performance. Care needs to be made with bottom-up
calculations to ensure that the changes observed in the illicit market actually occur
and are not reflections of operational changes.
Top-down methodologies base their estimates on high-level statistics, such as
the total declarations made to HMRC or the results of national surveys of behaviours or attitudes. The advantage of this approach is that the calculations are independent of operational activity. However, the disadvantage is that it is dependent
on third-party data.
HMRC uses a top-down methodology to estimate the size of the illicit markets
for spirits, cigarettes, hand-rolling tobacco, and oils. As the models used for each
market are well documented elsewhere, the specific methodologies will not be detailed in this paper.7, 8 Instead, this paper focuses on fundamental considerations
of a top-down methodology.
206
Rourke
The general top-down methodology for estimating the size of any illicit market
is fundamentally a simple task. It is just a comparison of the total consumption of
a particular good with the legitimate consumption of that good:
Illicit Market = Total Consumption − Legitimate Consumption
where the total and legitimate consumptions are determined from a combination
of independent surveys and declarations made to HMRC.
It is the appropriate selection and handling of the survey and declaration data
that make the calculation complex.
Surveys
HMRC is fortunate to have a wide range of survey data available for analysis, covering not only behaviours such as consumption of or expenditure on particular
goods but also attitudes towards those goods and illicit markets for them. Some
of the research is funded by external bodies, whilst other questions are funded
directly by HMRC.
Although surveys of attitude can be of use in understanding and tackling the illicit market for any good, it is the consumption and expenditure surveys that are of
most use in estimating the size of the illicit market. However, there are strengths
and weaknesses to both types of survey. These need to be carefully considered
when analysing the survey data.
Ultimately, to estimate the size of the illicit market, the surveys are compared
with the clearances that are declared to HMRC when the goods are released for
sale and so when the duty on those goods is due. Both the volume and the duty
paid of the goods released for sale are declared to HMRC, where the duty paid is
based on the recommended retail price of those goods. As there can be discrepancies between the recommended retail price of a good and the price it is actually sold for, HMRC uses the volumes to determine the size of the legitimate UK
markets of excise goods. Therefore, ideally, the results of the surveys used would
reflect the volumes sold relatively synchronous with the clearances.
The relative strength of the consumption surveys is that these typically measure
the volume of goods consumed at any moment in time but that consumption is
not necessarily synchronous with the clearances. The relative strength of the expenditure surveys is that these are relatively synchronous with the clearances, as
retailers will not intentionally maintain high levels of stocks in their stores for a
long time, but these typically measure the amount spent on the goods and do not
record the volumes. Therefore, it can be seen that neither survey fits the ideal criteria for use in the tax-gap calculation, so a compromise needs to be made when
selecting the most appropriate survey to use.
Measuring and Tackling the Illicit Market for Excise Goods
207
Two factors that can be used to decide which type of survey to use are the
omestic shelf life of the goods and any seasonality in sales or consumption of
d
the goods.
Domestic Shelf Life
The domestic shelf life of a good in this instance refers not to the life span within
which a good will no longer be fit for consumption but the length of time that
a good is expected to last between when it is purchased and when it is totally
consumed.
For example, in 2008, the average smoker in the UK consumed between 10 and
16 cigarettes per day, with some variation depending on age and gender.9 As cigarettes in the UK are typically sold in packs of 10 or 20, the average domestic shelf
life for an individual pack of cigarettes is less than 2 days.
By contrast, in 2008, on average between 5 and 19 units of alcohol were consumed per week, with some variation depending on age and gender.1 As this consumption consists of a range of alcohols including beer, wine, and spirits, of which
spirits is a small component, and there are typically 30 units of alcohol in a bottle
of spirits, the average domestic shelf life of a bottle of spirits is likely to be over
2 months. In addition, as spirits cover a range of substitute goods (for example,
vodka, whisky, and gin) that an individual may consume in parallel (unlike the
relatively homogeneous tobacco brands), the domestic shelf life of a particular
brand of spirits is likely to be longer.
An extreme example of a long domestic shelf life is wine. In general, some wine
is consumed within a short time of it being purchased but there is a behaviour
within the wine market for “laying down” wine for a number of years before it is
consumed. Indeed, some of the wine that is laid down may never be consumed.
Figure 1 illustrates the implications of the domestic shelf life for the measurement of an illicit market.
If a good has a short domestic shelf life, as tobacco has, the purchase and total
consumption of that good are relatively synchronous with the clearances of that
good. Therefore, either survey can be used to determine a reasonable estimate of
the size of the total, legitimate, and illicit markets.
If a good has a medium domestic shelf life, as spirits have, the consumption
of the good is not synchronous with the clearances. Provided that there are no
large changes in the consumption trends for the good, and the total and legitimate
markets are compared over a longer period (for example, comparing the overall
markets for an entire year rather than on a monthly basis), the consumption surveys may still be useful for estimating the size of the illicit market.
208
Rourke
FIGURE 1. Excise Goods with Short, Medium, and Long Domestic Shelf Lives
Year 1
Short
(Tobacco)
Year 2
Year 3 +
Totally consumed
Medium
(Spirits)
Long
(Wine)
Sold
Cleared for sale
However, as the domestic shelf life increases, the consumption of the good cannot be regarded as synchronous, and it becomes inappropriate to compare with
the clearances. The size of the illicit market can only then be reliably estimated
using relatively synchronous expenditure surveys.
Seasonality
Seasonality can also be an important consideration in determining whether to use
consumption or expenditure surveys. Whilst there is not much seasonality in the
consumption of most excise goods, seasonality has been observed in the expenditure surveys for some goods. If there is also seasonality in the clearances, then a
comparison of the consumption survey with the clearances will result in an apparent seasonal illicit market that may not exist in reality and that would not appear
in a comparison of the expenditure survey with the clearances.
For example, Figure 2 (a) shows the purchases, consumption, and clearances of
a good with seasonal variation in the purchases and clearances but not consumption. Figure 2 (b) shows the effect that comparing the purchases or consumption
figures with the clearances in each month has on the estimate of the illicit market.
It can be clearly seen that the illicit market shares based on the purchases of goods
are more stable and consistent over the year.
Some seasonal behaviour has been observed in the purchases of alcohol, in
particular, spirits, with more being purchased towards the end of the calendar year
than the beginning.10 It is believed that this is driven by purchases either to give
209
Measuring and Tackling the Illicit Market for Excise Goods
as gifts at Christmas or for entertaining purposes over the holiday period, or most
likely by both factors. This seasonality is also reflected in the clearances of spirits.
Therefore, it is necessary to use expenditure surveys to estimate the size of the total market and the illicit market for spirits to account for both its longer domestic
shelf life and the seasonality of its sales.
FIGURES 2A and 2B. Seasonal Variations in the Clearances of Spirits in the UK
Purchases
Consumption
Clearances
80%
60%
Illiciit market share
40%
20%
0%
-20%
20%
-40%
-60%
-80%
Purchases
Consumption
210
Rourke
Market Research Data
In addition to surveys, empirical data on the volumes and prices of the retail sales
at any moment in time is also available via market research companies, which
monitor the demand for many goods. As much of the illicit market is unlikely to
be recorded passing through retail sites that co-operate with the research companies, for most goods, this data is more a measure of the legitimate market than the
total market and so is not used as the main measure of total consumption.
However, the market research data does provide an independent estimate of
the actual retail prices for those goods at any given moment in time. By combining the market research data with the expenditure surveys, a reasonable estimate
for the volume of sales can be made, which reduces the relative weakness of using
expenditure surveys rather than consumption surveys.
Trends
A final consideration when deciding which survey to use is the history of the surveys available. Typically, estimates of the size of an illicit market are not made in
isolation. Instead, these estimates are made over a number of years so that any
changes in the illicit market, including the effect of any activity to reduce the illicit
market, can be detected. Therefore, it is critical that the data used to estimate any
market exists over the period of interest and that either the data is consistent over
the whole period, or it is clear from any supporting documents what has caused
any changes to the definitions of particular data and how to make it consistent
over the whole period.
This is particularly important for excise goods such as alcohol and tobacco,
as various factors typically lead to surveys underreporting consumption or purchases of these goods. Therefore, it is necessary to uplift the survey results to take
account of this underreporting. The most practical method to do this is to identify
a time when the illicit market is negligible and to compare the total consumption
observed from the surveys with the legitimate consumption observed from surveys and clearances. Any differences in this base year can then be applied to the
survey results of all years. If the results of the survey are not consistent over time,
the validity of this uplift factor is undermined.
Clearances
In addition to being careful to select the most appropriate survey for the different excise goods, careful consideration of what the clearance data represents also
needs to be made before the survey and clearances are compared directly.
Measuring and Tackling the Illicit Market for Excise Goods
211
Whilst the clearance data are used in the estimates of the illicit market as a
measure of the legitimate sales of a good in the UK, excise duty is not actually paid
at the point of sale to the end consumer but at the point that it leaves a bonded
warehouse to be supplied either to a retailer or to an intermediate wholesaler. In
general, this distinction is academic, as it is in the retailer’s or wholesaler’s interest
to maintain only low levels of stock and to sell any goods on to the end consumer
as quickly as possible.
However, there are instances when there is a delay between the goods being
cleared for sale and being purchased by the end consumer. Two events that introduce such a delay are forestalling by the manufacturers and overanticipation
of demand by the retailers. These delays have implications for the illicit-market
models that need to be taken into account.
Forestalling
Forestalling is a form of tax avoidance, in which the tax is declared early on goods
or services in anticipation of an increase in the tax rate.
In the case of tobacco, forestalling operates by the tobacco manufacturers releasing more through clearances into the market than could reasonably be expected to be sold at that time. These clearances are then sold over the following
months, during which time the legitimate clearances are lower than average. As
there have been annual increases in tobacco duty rates since 1999, there has been
an annual incentive for the tobacco manufacturers to be involved in forestalling.
Forestalling has two impacts on HMRC. The primary impact is a reduction
in the amount of tax collected from sales of tobacco. The secondary impact is a
distortion of the legitimate market for tobacco and, so, a distortion of the estimates
of the illicit market.
To tackle the actual avoidance of tobacco duty, HMRC has introduced restrictions on the supply of tobacco over a period leading up to a change in the duty rate.
HMRC makes use of the Customs and Excise Management Act 1979, Section 128,
which allows HMRC to refuse to allow clearances of goods in quantities exceeding
those which appear to be reasonable.11, 12 HMRC, by default, uses these powers to
prevent all tobacco manufacturers from clearing any tobacco over the restricted
period. To enable legitimate supplies, HMRC then grants manufacturers, upon
application, an allocation of tobacco that can be cleared over that period.
For example, in December 2009, HMRC restricted the clearances of tobacco
between January 1, 2010, and midday on Budget Day, when any changes to duty
rates would be announced. HMRC then indicated that the manufacturers could
apply for an allocation for that period, where the maximum volume that they could
clear would be determined from their average daily clearances over the previous
year and the length of the period of restriction, with an uplift factor that allowed
for a growth in the legitimate demand for their product.
212
Rourke
These restrictions do have some impact on forestalling but do not remove all
of it from the tobacco manufacturers’ supply chain, and so some compensation
needs to be made in the estimates of the legitimate tobacco markets. As HMRC
reports its estimates of the tobacco markets on an annual basis, forestalling is not
necessarily a problem for the market estimates if it occurs halfway through a reporting year as the total forestalled and nonforestalled clearances would all occur
within the same reporting year. However, as Budget Day is traditionally around
the beginning of a financial year, and HMRC reports tobacco market estimates
by financial year, any forestalling at these times may over- and underestimate the
legitimate markets in the preceding and following financial years. To tackle the
effect of the residual forestalling on the market estimates, HMRC simply identifies
months closely preceding a Budget in which there were higher than average clearances and reallocates some of these clearances to the following 2 months.
Figure 3 shows the volumes of tobacco that have been cleared for sale on a
monthly basis over the last few years.13 The distinct peaks, followed by sharp
drops, related to Budget Day and forestalling can be clearly seen. It can also be
seen that the simple 3-month averages redistribute the volumes to a level close to
the average volume cleared throughout the rest of the year.
FIGURE 3. Forestalling Variation in the Monthly Clearances of Tobacco in the UK
Monthly clearances
Raw
Smoothed
Year 1
Year 9
Overanticipation of Demand
Overanticipation of demand for a good is a natural occurrence for individual retailers and wholesalers. As profit margins are often tight for retailers, they cannot
afford to maintain high levels of unpaid stock, but they will be equally reluctant
Measuring and Tackling the Illicit Market for Excise Goods
213
to lose sales by not having enough stock available. Therefore, retailers need to accurately estimate the demand that there will be for any good far enough in advance
so that they have the appropriate stock levels at the appropriate time.
If a retailer has been in operation long enough, and there are clear patterns
in the demand for their product, it is possible for the retailer to forecast the demand for that product in future months and years. However, with the exception
of monopolies, the demand for a particular retailer’s products cannot be treated in
isolation from the demand for competitors’ products. If consumers switch from
an existing retailer to a new retailer, the existing retailer will overanticipate their
demand and so result in temporary excess stocks. Similarly, if consumers switch
from a competitor to another retailer, the demand for the existing retailer will appear to grow faster than the overall demand for their combined goods and, at some
point, the existing retailer may overanticipate their demand.
If the legitimate and illicit markets are treated as a pair of competitors in a
duopoly, the legitimate market may overanticipate the demand in the same way
that any individual retailer may overanticipate their personal demand. This is illustrated in Figure 4.
FIGURE 4. Under- and Oversupply of the Total Market by Legitimate Suppliers
Over-supply
Total consumption
Illicit market
Legitimate sales
Year 1
►
214
Rourke
In the early years, the market is dominated by the illicit market. It can be seen
that, over time, both the total and legitimate demand for the good increases,with
the legitimate demand increasing faster than the total demand, which either causes
or is caused by a reduction in the illicit market. At some point, the illicit market
disappears, and the total demand is satisfied by the legitimate market. However,
as the legitimate market only has its own demand as an indicator of the size of the
total market, it may be unaware of this fact and so will assume that the market will
continue to grow along the trend it has observed. This will result in a temporary
oversupply of goods to the market.
It is the temporary oversupply of goods to the market that is the important effect of an overanticipation of demand for a good on the estimation of the size of
an illicit market.
To compensate for underreporting within the expenditure and consumption
surveys that form the basis of most of the models used by HMRC, these models
assume that, in a given base year, the illicit market is negligible and then scales the
survey results across all years so that the total consumption of the good is equal to
the legitimate consumption in that base year. If there had been an oversupply of
goods in that base year, errors would be introduced into this scaling factor and so
into the overall estimates of the illicit market.
In addition, as the price of most excise goods in the UK is higher than the price
of the same goods in most of the UK’s neighbours, it is believed that the illicit market will be strictly non-negative. The existence of an oversupply (or negative illicit
market) challenges this belief and may undermine the credibility of the model, if
not properly understood.
Note, the problem of over-anticipation will be greatest for those goods with
seasonal trends in demand. If the level of demand remains relatively stable over
the year, any overanticipation will be quickly detected and corrected for by the
manufacturers. However, if there is over-anticipation in a seasonal good, it may
take longer for this to be detected by the manufacturers. In addition, if the peak
in demand accounts for a large portion of the demand for a seasonal good, the
oversupply is likely to be greater.
Therefore, care needs to be taken to determine that the volumes released for
sale in the UK via clearances could be reasonably expected to supply the demand
at the time that they were released to avoid any distortions entering the estimates
of the illicit market. This can be most easily achieved by monitoring the clearances
both for unusual peaks and unusual troughs, typically with a trough closely following a peak. If the manufacturers have been forestalling their goods, this will
be a natural part of their supply chain. If the manufacturers have inadvertently
oversupplied the market, this will be a corrective factor until all the excess supply
has been sold.
Measuring and Tackling the Illicit Market for Excise Goods
215
International Trade
The problem facing HMRC regarding illicit markets for excise goods stems largely
from the international market for those goods, the fact that most are easily portable, and the relative high prices in the UK compared with much of the rest of the
world. This provides a strong financial incentive for smuggling of goods into the
UK. However, it also provides a strong financial incentive for legitimate international trade, which, if not accounted for in the HMRC models, would overestimate
the size of the illicit market in the UK.
In addition, whilst some of the illicit market is supplied with counterfeit goods,
there is also a problem of smugglers obtaining goods from legitimate international
sources and that some of these sources are knowingly supplied by international
manufacturers. Part of HMRC’s strategy is to tackle this facilitation of the illicit
market by legitimate suppliers.
Legitimate Personal Consumption
Each adult, when returning to the UK from abroad, is entitled to bring back a volume of goods for their own personal consumption. These purchases are legitimate
and so should be included in the estimates of the UK markets for excise goods to
avoid overestimating the size of the illicit market.
The consequences of not including legitimate cross-border shopping differs depending on the type of survey used to estimate the size of the total market for a
good in the UK.
If an expenditure survey is used, and the survey only covers expenditure within
the UK, the effect will be an underestimate of both the total market and the legitimate market, which in turn will result in an overestimate of the illicit market share
(due to the smaller total market) but a reasonable estimate of the illicit market.
Estimates of the size of legitimate cross-border shopping will need to be added to
both the total market and legitimate market estimates to avoid this overestimate.
If a consumption survey is used, the effect will be an underestimate of the legitimate market but a reasonable estimate of the total market, which will result in
an overestimate of both the illicit market and the illicit market share. Estimates
of the size of legitimate cross-border shopping will need to be added to just the
legitimate market estimates to avoid these overestimates.
Whichever type of survey is used, the effect of not including legitimate crossborder shopping in the models can be significant.
For example, in 2008, UK residents made 69 million international trips,of
which 50 million were to members of the European Union.14 Considering that
the total UK resident population in 2008 was 61 million,each UK resident made
just over 1 trip per year, on the return from which they would be entitled to bring
216
Rourke
tobacco into the UK.15 The duty-free limits when returning to the UK from outside the European Union is 200 cigarettes, which is what would be consumed by
the average smoker over 12 to 20 days. When returning from within the European
Union, there is no strict limit on what can be brought into the UK, but the assumption that it is solely for personal consumption will be challenged by UK officials if
more than 3,200 cigarettes are brought back, which is what would be consumed by
the average smoker over 200 to 320 days.
Therefore, if every smoker brought into the UK what they could bring back
without challenge and every smoker travelled outside the UK at least once a year,
the UK market would be dominated by legitimate cross-border shopping. If this
cross-border shopping was not included in the market estimates, the illicit market
would appear to account for over 50 percent of the total market.
Obviously, these are the upper estimates of the potential size of the legitimate
cross-border shopping market and it is not expected that all smokers will bring
back their limits or that all smokers will travel abroad at least once. To estimate
the actual size of this market, HMRC sponsors questions on the International
Passenger Survey that is carried out each year by the UK Office for National
Statistics.
The UK is fortunate that, with the exception of the Irish border, it has no land
boundaries with any other countries and so all legitimate international travel will
have to pass through a limited number of ports and airports. Therefore, the number of UK residents returning to the UK can be easily determined and, using the
results of the survey, the volume of excise goods that are brought into the UK for
personal consumption can estimated. These volumes can then be included in the
market estimates.
As can be seen in Figure 5, which shows a recent estimate of the size of the legitimate and illicit markets for cigarettes in the UK, legitimate cross-border shopping cannot be over-looked in the UK market estimates, accounting for almost 10
percent of the total UK market.
Facilitation
The less stringent controls on the movement of goods within the European Union
have made cross-border shopping easier not only for the legitimate travellers but
also for those who aim to smuggle goods into the UK. Whilst the illicit market
will include a range of non-UK duty paid products, brand loyalty within the UK
market may make smuggled versions of UK brands more attractive as there will
be a larger market for those goods. Therefore, the illicit market may seek to purchase large volumes of UK brands from international markets where the price will
be lower, resulting in a strong financial incentive for legitimate manufacturers to
knowingly supply international markets with goods that will end up in the illicit
UK market.
217
Measuring and Tackling the Illicit Market for Excise Goods
In response to evidence of facilitation by tobacco manufacturers, HMRC introduced legislation in the 2006 Finance Act that made the manufacturers liable to
take care that any product that they supplied to international markets would not
end up on the UK illicit market.16, 17 The legislation also made the manufacturers
liable for a penalty if they knowingly oversupplied the illicit market with tobacco.
If, as has happened in the past, the manufacturers oversupplied an international market with no domestic and little legitimate UK demand for UK branded
FIGURE 5. Legitimate and Illicit Markets for Tobacco in the UK
Total UK Cigarette Market 2000/01 - 2007/08
80
Ciga
arettes (Billion Sticks)
70
60
50
40
30
Illicit Market mid-point
Cross Border Shopping
20
UK Duty Paid
10
0
2000/01
2001/02
2002/03
2003/04
2004/05
2005/06
2006/07
2007/08
tobacco, the manufacturers could be easily challenged over the legitimacy of their
markets. However, there are some international destinations where large numbers
of UK residents regularly travel to and large numbers of UK citizens have emigrated to, and so where there are credible markets for UK brands. The challenge
to HMRC in these markets is to prove that the manufacturers are supplying more
than the legitimate demand in those markets.
The advantage of the methodology for estimating the size of the illicit market
for tobacco in the UK is that the same methodology can be applied to estimate
the size of the legitimate market in any international market. From the surveys
available to HMRC, the length of time UK residents spend in international destinations and, so, the average amount of tobacco they will consume outside the UK
can be estimated, as can the volume of tobacco that will be brought into the UK
when the UK residents return from their trips and the local demand for the UK
brands. These can then be combined to form an estimate of the legitimate market
218
Rourke
in that destination and compared with the volumes that have been supplied by the
tobacco manufacturers.
If there are large differences between these two figures, HMRC can discuss
these differences with the manufacturers. If no legitimate market can be identified for any excess supply, these calculations provide credible evidence of diversion
into the UK illicit market and of a need for the manufacturers to exercise stronger
controls of their supply chain in that international market.
Operational Effectiveness
Having combined the various surveys and clearances to estimate the size of the
total and illicit markets for a good, it may be possible to use these models to estimate the effectiveness of operational activity. Not only do these estimates provide
evidence of the benefits of HMRC activity, they can inform future decisions on the
resources required to reduce the illicit market or maintain it below a given level.
Whilst the size of the illicit market is dependent on a wide range of factors, if
operational activity has had an impact it should be observable. The simplest relationship that should be observable is:
Illicit volume
= Illicit proportion * Total consumption
- Seizure impact * Volume of seizures
+ Latent illicit volume
where the Latent illicit volume should be nonpositive for the model to be credible;
a positive value would indicate that an illicit market existed when there was no
actual market. In addition, the illicit rate should be between 0 and 1 but the illicit
rate could be greater than 1, which would indicate a serious problem in the market as any increase in the total market would lead to a reduction of the legitimate
market.
If the operational activity has had any effect, the Seizure impact will be greater
than 0.
Ideally, the Seizure impact would be greater than 1, indicating that the activity not only removes the volume of goods seized from the illicit market but also
prevents further goods from reaching the market. This could be achieved either
by deterring further illicit activity or reducing the illicit market’s financial capacity.
By contrast, if the Seizure impact is less than 1, it shows that the illicit market
is resilient enough to cope with operational activity and can replace its losses with
alternative supplies. If the Seizure impact is much less than 1, the operational activity is having very little effect on the illicit market.
219
Measuring and Tackling the Illicit Market for Excise Goods
Note, when considering the effect of the operational activity on the illicit market, it may be necessary to introduce a delay into the operational results. If the
illicit market has a robust supply chain, the effect of seizures may be delayed until
the existing illicit stocks have been depleted. For example, in Figure 6, a good
match between the observed and modelled illicit markets was found when a six
month delay in the effect of the seizures was considered.
FIGURE 6. Observed and Modelled Illicit Cigarette Volumes
Illicit v
volume (billion sticks)
15
12
Observed
9
Modelled
6
2001/02
2002/03
2003/04
2004/05
2005/06
Financial year
2006/07
2007/08
Conclusion
It has been shown in this paper that estimates of the total, legitimate, and illicit
markets for excise goods can be calculated using a combination of surveys and
clearance data. It has been explained that care has to be taken when choosing a
survey to take account of issues such as the domestic shelf life of the goods and any
seasonality in their consumption or purchases. It has also been shown that similar
care needs to be taken with the declared sales of goods that no distortions of the
legitimate market have been introduced, either deliberately through tax avoidance
schemes or inadvertently through overanticipation, and so oversupply, of the UK
market. Finally, it has been explained how the same techniques can be used to
estimate the international market for UK brands in an attempt to prevent the facilitation of the illicit market by legitimate manufacturers and how the efficiency
of operational activity can be assessed from the illicit market estimates.
220
Rourke
Endnotes
1
“Developments in VAT Compliance Management in Selected Countries,”
OECD Centre for Tax Policy and Administration, August 2009.
2 “Measuring the Non-observed Economy: A Handbook,” Paris, OECD, 2002.
3 “Reducing the Federal Tax Gap: A Report on Improving Voluntary
Compliance,” Internal Revenue Service, U.S. Department of the Treasury, 2007.
4 “Tax Compliance,” James Andreoni, Brian Erard and Jonathan Feinstein,
Journal of Economic Literature, Vol. 36, No. 2, (June 1998), pp. 818–860.
5 “The Shadow Economy in Germany, Great Britain and Scandinavia: A
Measurement Based on Questionnaire Surveys,” Soren Pedersen, Statistics
Denmark, 2003.
6 “Tax Gap Map for Sweden: How Was it Created and How Can it Be Used?”
Swedish National Tax Agency, 2008.
7 “Measuring Tax Gaps, 2009,” HM Revenue and Customs. http://www.hmrc.
gov.uk/stats/measuring-tax-gaps.htm
8 “Methodological Annex for Measuring Tax Gaps 2010,” HM Revenue and
Customs. http://www.hmrc.gov.uk/stats/measuring-tax-gaps.htm
9 “Smoking and drinking among adults, 2008,” General Lifestyle Survey 2008,
Office for National Statistics. http://www.statistics.gov.uk/downloads/theme_
compendia/GLF08/GLFSmoking&DrinkingAmongAdults2008.pdf
10 HM Revenue and Customs Spirits Bulletin. https://www.uktradeinfo.com/
index.cfm?task=bullspirits
11 Customs
and Excise Management Act 1979, UK. http://www.opsi.gov.uk/
RevisedStatutes/Acts/ukpga/1979/cukpga_19790002_en_1
12 128 Restriction of delivery of goods
(1) During any period not exceeding 3 months specified at any time by order
of the Commissioners for the purposes of this section, the Commissioners
may refuse to allow the removal for home use on payment of duty, or the
sending out for home use after the charging of duty, of goods of any class or
description chargeable with a duty of excise, notwithstanding payment of that
duty, in quantities exceeding those which appear to the Commissioners to be
reasonable in the circumstances.
(2) Where the Commissioners have during any such period exercised their
powers under this section with respect to goods of any class or description,
then, in the case of any such goods which are removed or sent out for home
use after the end of that period, the duties of excise and the rates thereof
chargeable on those goods shall, notwithstanding any other provision of the
Measuring and Tackling the Illicit Market for Excise Goods
221
customs and excise Acts relating to the determination of those duties and rates,
be those in force at the date of the removal or sending out of the goods.
13 HM Revenue and Customs Tobacco Bulletin. https://www.uktradeinfo.com/
index.cfm?task=bulltobacco
14
“Travel Trends 2008, Data and commentary from the International Passenger
Survey,” Office for National Statistics, UK. http://www.statistics.gov.uk/
statbase/Product.asp?vlnk=1391
15 “Population Estimates for UK,” Office for National Statistics, UK. http://www.
statistics.gov.uk/statbase/product.asp?vlnk=601
16 Finance Act 2006, UK. http://www.opsi.gov.uk/acts/acts2006/
ukpga_20060025_en_2#pt1-pb1
17 Duty not to facilitate smuggling
(1) A manufacturer of cigarettes or hand-rolling tobacco shall so far as is
reasonably practicable avoid—
(a) supplying cigarettes or hand-rolling tobacco to persons who are likely
to smuggle them into the United Kingdom,
(b) supplying cigarettes or hand-rolling tobacco where the nature or
circumstances of the supply makes it likely that they will be resupplied
to persons who are likely to smuggle them into the United Kingdom,
or
(c) otherwise facilitating the smuggling into the United Kingdom of
cigarettes or hand-rolling tobacco.
(2) In particular, a manufacturer—
(a) in supplying cigarettes or hand-rolling tobacco to persons carrying on
business in or in relation to a country other than the United Kingdom,
shall consider whether the size or nature of the supply suggests that the
products may be required for smuggling into the United Kingdom,
(b) shall maintain a written policy about steps to be taken for the purpose
of complying with the duty under subsection (1), and
(c) shall provide a copy of the policy to the Commissioners on request.
Investing in Detection
Technology as a Component
of an Optimal Tax-Enforcement
Policy
Martin Besfamille, Universidad Torcuato Di Tella
and Pablo Olmos, Yale University
F
or most governments, tax evasion is a problem because it threatens the
equity and the efficiency of their fiscal policies. For this reason, governments react and adopt actions to assure compliance with the tax law. For
example, audits are conducted to verify whether tax liabilities have been met and,
if this is not the case, evaders are penalized.
However, this “enforcement approach” is not sufficient to deal with tax evasion. As the public finance literature shows, the fight against tax evasion cannot be
isolated from the design of the fiscal policy; the extent of tax evasion depends not
only on parameters that characterize the enforcement policy carried out by the tax
administration (e.g., frequency of audits, level of fines) but also upon the structure
of the tax law (e.g., tax rates). Therefore, as suggested by Allingham and Sandmo
(1972) and then emphasized by Kolm (1973), the design of optimal fiscal policies
or ‘optimal tax systems’ (in Slemrod’s (1990) terminology) should also include all
instruments that help to enforce the tax law.
Since Sandmo (1981), many articles have analyzed, in different settings, optimal
tax-enforcement policies. All these contributions share a feature: audits are perfect. In other words, when the tax administration performs an audit, it observes
the taxpayer’s private information. But this assumption is far from being realistic,
as already recognized by Alm (1988), Scotchmer and Slemrod (1989), and Snow
and Warren Jr. (2005). Among others, Feinstein (1991) and Erard and Feinstein
(2010) estimate, using data of the Internal Revenue Service (IRS), that detection
rates vary between 30 percent and 50 percent. This failure to detect evaders clearly
modifies the analysis of optimal tax-enforcement policies, as shown by Boadway
and Sato (2000).
Once this is acknowledged, one has to address a second important issue: is
the detection probability exogenous or endogenous? From a theoretical point of
view, it should be clear that, if governments have the possibility to modify this
detection probability, for example by investing resources to improve the detection
20
Besfamille and Olmos
technology of the tax administration, they will probably undertake such kind of
investments. Moreover, adopting such investment decisions is not only a theoretical possibility but, as the following paragraph illustrates, it also has a strong
empirical support: governments do invest resources to improve the technology to
detect evaders.
In Argentina, the Dirección de Rentas de la Provincia de Buenos Aires (the tax
authority of the Buenos Aires province) subscribed to Google Earth, in February
2007, to download high-quality satellite images that could serve as evidence of
evasion to the property tax. This technology enabled the tax agents to discover, in
less than one semester, 68.844 undeclared properties, 1458 undeclared swimmingpools, and more than 13 million square meters with undeclared silos (La Nación,
August 27th 2007). The use of Google Earth is not limited to developing countries:
the Italian Guardia di Finanza has used it many times to discover inconsistencies
between actual and declared values of luxurious villas.
Investments made by governments to improve their tax administration’s detection capacity have been either mentioned informally (see Snavely 1988) or studied
empirically (see Hunter and Nelson 1996 and Cebula 2001). Here, we also incorporate them into the formal analysis of optimal fiscal policies. We address this
issue in a simple three-stage model, with two classes of active agents: individuals
and a government. Each individual can be poor or rich; the rich being the only
to earn a taxable income. The government follows a social welfare criterion that
incorporates aversion to inequality. In order to maximize its criterion, the government designs a fiscal policy, to be implemented by the tax administration. In
the first stage of the model, the government invests resources to improve the tax
administration’s detection capacity. In the second stage, the government designs
the tax law, which specifies the tax owed by the rich and the enforcement policy
to be conducted later by the tax administration. We assume that the government
has the ability to commit to this policy. Finally, in the last stage of the model,
the tax administration collects taxes and enforces the tax law, as follows. As incomes are private information, individuals are requested to report them. Then,
the tax administration audits reports according to the frequency pre-specified by
the government, and known to individuals. As in many other contributions to the
literature on tax evasion, we assume that audits are costly. But here audits discover
randomly whether a taxpayer has misreported, and the probability of catching
an evader (i.e., the detection probability) is an increasing function of the initial
investment and its productivity. If a misreport is detected, the tax administration
taxes the evader according to his true income and imposes him an additional fine.
With all revenues collected (taxes and fines, net of investment and audit costs), the
government finances the provision of a public good.
As a benchmark, we derive first the optimal fiscal policy under full information, when enforcement is not necessary. Then we move to asymmetric information, and we solve the model backwards. As the government can commit to the
21
Investing in Detection Technology
tax and to the enforcement policy to be conducted during the third stage, in the
second stage we characterize the optimal tax law adopting a mechanism design
approach. Depending upon the value of the detection probability, two regimes
emerge. In the first regime, when the detection probability is high, the tax administration only audits individuals that have reported to be poor. In order to
attenuate the stake for evasion of the rich, the optimal tax is downwardly distorted
with respect to the full-information optimal tax. We show that the optimal tax
increases with the detection probability. Regarding the optimal audit probability,
it can monotonically increase with detection or have an inverse U-shaped curve.
As is usual in this kind of models, the optimal fine has only a deterrent role and is
maximal. In the second regime, the detection probability is so low that the government prefers not to tax, and so no enforcement takes place. Anticipating these
decisions, in the first stage, the government chooses whether to invest in the tax
administration to improve the detection probability and, if so, by how much. This
choice has an impact not only on the expected social welfare (because tax revenues
are allocated to investment instead of being allocated to the public good) but can
also fix under which regime the government and the tax administration will be
afterwards. Although we prove that an optimal investment exists, we cannot completely characterize it in general, due to the non convexity of the problem. In fact,
we can show that, when the optimal investment is strictly positive, the levels of the
tax and the public good are higher than their respective levels when investments
are not an option for the government. Under this circumstance, we derive some
comparative statics results.
The remainder of this paper is organized as follows. Section 2 describes the
model and shows the optimal fiscal policy under full information. Section 3 analyzes the optimal fiscal policy under asymmetric information. Finally, Section 4
concludes. All proofs appear in the Appendix.
The Model
There is a continuum of individuals of measure one. Each individual l has an
income |l which is a random variable that takes values in the set if> |j> with
| A f= An individual with | income is henceforth called “rich”; otherwise, he
is called “poor.” Each individual’s income |l is his private information. All individual incomes are i.i.d., and the probability that |l ' | for any given individual l
is μ∈ (0,1), which is common knowledge.
Poor individuals only benefit from a public good, provided by the government,
in quantity j . Their ex-post welfare is given by
zs ' j1
22
Besfamille and Olmos
Rich individuals also derive utility from consumption of a private good q, the price
of which is normalized to one. Their ex-post welfare is
zu ' xEt n j
where the function x ( ) satisfies1
xEf ' f> xt A f> *4 xt ' 4> *4 xt ' f> xtt ? f> xttt A f dqg
t 0.
If a taxpayer is not audited, he pays the tax that corresponds to
his report.4 If he is audited, the tax administration discovers a
misreport with probability S : in other words, audits are imperfect.
In fact, the “detection probability” S is a continuous, strictly
increasing and concave function S EL> where > 0. The function
S ( ) satisfies
*4 S ' Sf dqg *4 S ' =
L 0:
A f G
the higher the parameter , the higher the capacity of any level of
investment to improve the initial detection probability Sf. This
is the reason to call the “investment productivity.” Its value
depends, among other things, on the training skills of the people
who are expected to operate the new technology.
Then, if a misreport is detected, the evader has to pay the tax
that he legally owes plus the additional fine ƒ.5 With all revenues
collected (taxes and fines, net of investment and audit costs), the
government finances the provision of the public good j , whose
cost is also normalized to one.
The goal of the paper is to characterize the optimal fiscal policy, which is the
e
ee
5-uple iL>
w> i>
eh > e
j j. Before doing that, and in order to have a benchmark, we
present the optimal fiscal policy under full information. In this case, the tax administration observes incomes and thus audits are useless. Anticipating this, the
government does not need to invest and simply solves the following problem,
where private consumption has been replaced, using the taxpayers’ budget constraint, by their disposable income
;
Pd{ xE| w n j
A
A
w>j
A
A
A
A
A
A
? vxemhfw wr
W
S
fw
A
A
A
A
A
w|
EOO
A
A
A
=
j ' w
EE
24
Besfamille and Olmos
We denote by (LL) the limited-liability constraint and by (B), the government’s
budget constraint. The following expressions
xt E| wW ' li wW A f
+4,
rwkhuzlvh
xt E| A
characterize the optimal full-information tax wW. On the one hand, when
xt E| @ , the government taxes the rich in order to equalize their social marginal utility of consumption with the social marginal utility of the last $ spent in
the public good. On the other hand, when xt E| A @ , taxation is too costly in
welfare terms and thus wW = 0. We denote by j W ' wW the optimal provision of the
public good.
Optimal Fiscal Policy Under Asymmetric
Information
In this section, we characterize the optimal fiscal policy under asymmetric information, when the tax administration tries to detect evaders by auditing income
reports. As usual, we solve the model backwards.
The Optimal Tax Law
When the government designs the tax law, it can commit to the audit probability |h . Thus, as shown in a similar setting by Mookherjee and Png (1990), the
Revelation Principle applies and the optimal tax law can be characterized adopting a mechanism design approach. According to Mookherjee and Png (1989), the
tax administration does not need to audit a taxpayer who has reported to be rich.
Thereinafter, π will denote the probability of auditing an announcement |h = 0. The
w> ie>
e> je solves the following problem, where again private conoptimal tax law Ee
sumptions have been replaced by taxpayers’ disposable income, but now at each
possible final state.
;
Pd{ xE| w n j
A
A
w>i> >j
A
A
A
A
A
A
vxemhfw wr
A
A
A
A
A
A
? fw
S
f
A
A
A
A wni |
A
EOO
A
A
A
A
A
A
x E| w E S x E| n S x E| w i ELF
A
A
A
=
j ' w E f L
EE
25
Investing in Detection Technology
Now we denote by (LL′ ) the after-audit limited-liability constraint6 and by (IC),
the incentive-compatibility constraint.7 With respect to the full-information setting, the government’s budget constraint (B′ ) now incorporates the aggregate audit
cost E f and the (sunk) initial investment L .8w As is usual in this kind of
models, the fine ƒ does not enter in the maximand of the problem S because it
only has a deterrent role.
At the optimum, (LL′ ) binds: increasing the fine ƒ up to its maximal legal level
| w relaxes (IC). Moreover, this constraint also binds. Thus the government
sets the audit strategy
x E| w
'
+5,
S
x E|
such that a potential evader is indifferent between truthfully reporting his income and misreporting.9 We compute the first-order condition of problem S .
Rearranging, we obtain the expression that characterizes an (interior) optimal tax
xt E| e
w '
E f
=
E f n S x E|
+6,
Under asymmetric information, the taxpayers’ social marginal utility of consumption again equals the social marginal utility of expenditure in the public good. But
now, due to the necessity of auditing reports to collect taxes, the expenditure in
the public good is less than the tax collection. Therefore, by concavity of the utility
w is downwardly distorted with respect to the opfunction x ( ), the optimal tax e
timal full-information tax wW. In the next proposition, we completely characterize
the optimal tax law. In particular, we explain in detail how the optimal tax and the
audit probability vary with the detection probability S .
E fxt E|
= Under asymmetric information, the
d xt E| oxE|
following two regimes emerge.
Proposition 1 Let S '
• Regime UQD : when S S > the government does not tax and the
tax administration does not audit.
• Regime UD: when S A S > the government taxes and the tax
w
administration audits reports |h = 0 randomly. The optimal tax e
is a continuous, strictly increasing and concave function of S that
satisfies
w ' f dqg
*4 e
S charges the tax administration to audit with probability
ernment sets the tax h
h
w> where f h
and to impose to evaders
the fine ih ' | h
w ? wW and 0 <
h < 1.10
Now consider a small decrease in the detection probability S . Ceteris paribus, π
increases, and so does the aggregate audit cost E f=. This causes a decrease
in the provision of the public good, with its consequent welfare loss. What should
be the optimal reaction of the government? To reduce the tax w and to increase the
fine ƒ, while keeping their sum constant, equal to | . Even if this change reduces
the tax collection and, a priori, decreases further the provision of the public good,
it has two other effects that attenuate the abovementioned welfare loss. First, the
decrease in w reduces the stake for evasion, i.e., the gain x E| x E| w , countering the initial increase of the optimal audit probability π.g Second,
l
k reducing the
tax makes private consumption of the rich to increase. A similar argument can be
w lincreases when the detection probability
used to explain why the distortion wW e
S decreases.
Under some parameter configurations of the model, for high values of S , the
abovementioned decrease in the stake for evasion x E| x E| w may be lower
than the decrease in S , and thus π increases. In other
g l words,
k from S = 1, the
government optimally reacts to a decrease in the detection probability by auditing
more frequently. But for lower values of S ,
e always increases with S . Therefore,
it may be the case that the profile of
e be inverse U-shaped.
This is not the end of the story. For a ‘sufficiently low’ value of the detection
probability S , namely S > the optimal tax e
w converges to 0. Then, when S S /
regime UQD emerges: aswlthe detection probability is very low, an incentive-compatible enforcement policy is prohibitively costly, and thus
e = 0. Under this regime, the unique incentive-compatible tax is 0, and the fine is irrelevant. So, no
redistribution takes place because no public good is provided.
Finally, the optimal provision of public good e
j is also a continuous and strictly
increasing function of S , that satisfies
*4 e
j ' f dqg *4 e
j'h
j=
S ' S =11 The expected welfare can now be written as a function of investment
L as
w follows
li Sf S HZ ' HZ D ' xE| ew n ew E ef L iru L f
+
iru L ' f
HZ QD ' xE|
li Sf ? S HZ '
D
HZ ' xE| ew n ew E ef L li L 4@ if> Lj
+7,
where the superscripts indicate the corresponding regime. The expression HZ QD
takes into account that, as the government raises no tax revenue, investment in the
tax administration cannot be afforded. As we can see, the value of Sf is important
to characterize the expected welfare. When Sf S > only regime UD emerges: no
matter the investment decision, S S . But this is not the case when Sf ? S :Gaccording to the level of L , wboth regimes UQD or UD can occur.
e we proceed as follows. First, we
In order to solve for the optimal investment L>
find LD , the investment that maximizes HZ D 1. Second, when it is pertinent to do
e
so, we compare HZ D ELD with HZ QD to take the overall maximum L1
Under regime UD, the optimal investment LD is the solution to the following
problem
;
A
Pd{ xE| e
w n e
w E
ef L
A
A
L
A
A
A
A
A
A
? vxemhfw wr
S2
e
w'e
wES >
e'
eES > S ' S EL>
A
A
A
A
A
4@ if> Lj L
A
A
A
A
=
w E
ef
L e
wES and
where e
eES are given by (3) and (2), respectively. From Proposition 1,
they areo uniquely
defined and continuous functions of the detection probability
g
S . The last two inequalities characterize the constraint set. The first inequality
28
Besfamille and Olmos
reflects that the lowest value of L supporting
regime UD is not unique because it
w
depends upon Sf, as it is clear from (4). The second inequality shows the resource
constraint of the government, at this initial stage.
A general characterization of the solution to S2 is difficult, for the following
reasons. First, when if> Lj ' L> the constraint set may be empty. Indeed, under
some parameter configurations of the model (e.g. high audit cost c), no investment
fulfills the resource constraint. Second, even if the constraint set is non-empty
and we can prove that the problem S2 has a maximum, it is often difficult to
find it with the usual techniques because the expected welfare HZ D EL
isDnot
always
concave in L and
w the second-order condition, evaluated at the critical points, cannot be verified analytically. Finally, even if one succeeds in identifying LD , the
comparison between HZ QD and HZ D ELD is not straightforward because it is a
comparison of levels. Still, we can prove the following results that apply when the
optimal investment satisfies Le A f .
Proposition 2. If the government invests a strictly positive amount of money to
w and
improve the tax administration’s capacity to detect evaders, the optimal tax e
j are higher than when such investments are not an option
the level of public good e
for the government.
If the optimal investment is strictly positive, the detection probability is higher
than Sf . Therefore, by Proposition 1, the optimal tax increases above the level
chosen by the government when it is (exogenously) not allowed to invest in the
tax administration. Intuitively, one could also have expected a similar result about
j . However, this intuition must not be based on the conjecture that investment,
via the increase in S , makes π to decrease, pushing downwards the aggregate audit cost and thus yielding j to increase, ceteris paribus. In fact, this conjecture
can be wrong: we already know from Proposition 1 that the optimal frequency of
audit may increase with S , pushing the aggregate audit cost upwards and making
the provision of public good to decrease. The reason for the higher provision of
the public good is the following: as e
w increases when investment is realized (via
the increase in S ), optimality implies that the provision of the public good must
increase, to compensate for the lower consumption of taxpayers.
Next, we present some comparative statics results under Regime UD.
Proposition 3. If the government invests a strictly positive amount of money to
improve the tax administration’s capacity to detect evaders and the optimal tax law
specifies auditing reports |h ' f randomly, the optimal investment Le increases with
taxable income | and with the degree of aversion to inequality E . With respect to the other parameters of the model, the change in Le is ambiguous.
Investing in Detection Technology
29
When the tax administration audits reports |h ' f randomly, an (interior) optimal investment Le A f is characterized by the following expression
e '
E
eS f=SL EL>
+8,
where
eS ' e
@S= The lhs of this expression is the marginal benefit of increasing investment, and the rhs, its marginal cost (hereinafter MCI), which is always
equal to 1. By an envelope argument, the marginal benefit is formed as the product of the effect of an increase in the detection probability on the aggregate audit
eS f, and the detection improvement due to a marginal increase
cost + E
e ,1 This product measures savings in the aggregate audit cost that
in L +SL EL>
obtain from a marginal increase in L . w Consequently, we denote by MSAAC the
lhs of (5).
From this expression, we can infer how a change in one parameter of the model
affects the value of Le. The variation of Le results from the combination of two
potential effects on the MSAAC. First, there is a direct effect that occurs when
e 1 Second, there is an
this parameter change affects only E > f>
eS or SL EL>
indirect effect, which appears provided the parameter
change modifies the value of
k
the optimal tax e
w , making the value of the derivative
eS to vary as well. So, if after
a parameter change the MSAAC is greater (lesser) than the MCI, the government
restores optimality by increasing (decreasing) investment. Having this in mind,
we explain in detail the comparative statics results presented in the proposition.
• When taxable income | increases, the two effects have opposite
signs. On the one hand, the direct effect is positive. For a given
tax w, by concavity of the utility function x ( ), the stake for evasion
xE| xE| w decreases when | increases. Therefore, in order to
ensure incentive compatibility, it is not necessary to audit so much,
and thus S increases. On the other hand, the indirect effect is
negative: an increase in | enables the government to tax more, but
also to audit more. This pushes downwards S . In spite of these
countervailing forces, the indirect effect dominates and thus
eS
decreases. So, as the MSAAC increases with | , the government
optimally invests more.
• When the degree of aversion to inequality E increases, only
the indirect effect appears. Higher aversion to inequality makes the
government to increase the tax. This pushes the stake for evasion
upwards: π increases, making
eS to decrease. As a consequence
of this, the MSAAC increases and thus Le also optimally increases.
• When the fraction of rich individuals in the population µ increases,
the two effects go in opposite directions. The direct effect is clearly
30
Besfamille and Olmos
negative. On the other hand, the indirect effect is positive: an
increase in μ pushes the government to tax more. As w increases,
so do π and the MSAAC. Hence, the total effect is ambiguous.
• When the audit cost f increases, the two effects go in opposite
directions. The direct effect is clearly positive. On the other
hand, the indirect effect is negative: an increase in f pushes the
government to tax less. As w decreases, so do π and the MSAAC.
Hence, the total effect is ambiguous.
• Finally, when the initial detection probability Sf or the investment
productivity increase, the two effects may appear. Their respec
tive value depend upon the retained functional specification of the
detection probability S EL> .
Conclusions
There is a large list of contributions that have analyzed optimal tax-enforcement
policies under the threat of tax evasion. Surprisingly, all assume that audits are
perfect. Not only audits are indeed imperfect but also, in practice, governments
invest many resources to improve the capacity of their tax administration to detect evaders. This paper incorporates these investment decisions in a very simple
model of an optimal fiscal policy. We have been able to characterize the optimal
tax-enforcement policy, adopting a mechanism-design approach. As many other
contributions to the costly-state verification literature, the optimal fine for evaders is maximal and the optimal audit probability is such that evasion is deterred.
However, in order to attenuate the stake for evasion, the government optimally
distorts taxes downward, distortion with respect to the fullinformation optimal
tax. Then we analyze the optimal investment. Although we prove its existence,
we cannot completely characterize the optimal investment in general. But we can
show that, when it is strictly positive, the levels of the tax and the public good
are higher than their respective levels without investment. Finally, under these
circumstances, we obtain some comparative statics results concerning the optimal
level of investment.
The model can be extended in several directions. First, it could be extended
dynamically, to analyze the path of the different elements of the optimal fiscal
policy. Second, our analysis can be generalized to incorporate more than two levels of income or more dimensions of heterogeneity (e.g., different degrees of risk
aversion). Then one could think to calibrate an optimal tax-enforcement model
with empirically founded parameters, and then to proceed by adding investments
to modernize the tax administration into the simulations. Finally, the model
generates some testable implications. All these are interesting venues for future
research.
Investing in Detection Technology
31
Acknowledgements
We are grateful to D. McPartland and 2010 IRS Research Conference participants
for helpful comments and discussions.
References
Alm, J. (1988) Uncertain Tax Policies, Individual Behavior, and Welfare,
American Economic Review 78, 237–245.
Allingham, M. and A. Sandmo (1972) Income Tax Evasion: A Theoretical
Analysis, Journal of Public Economics 1, 323–338.
Boadway, R. and M. Sato (2000) The Optimality of Punishing Only the Innocent:
The Case of Tax Evasion, International Tax and Public Finance 7, 641–664.
Cebula, R. (2001) Impact of income-detection technology and other factors on
aggregate income tax evasion: the case of the United States, Banca Nazionale
del Lavoro Quarterly Review 54, 401–415.
Eeckhoudt, L. and C. Gollier (1995) Risk: Evaluation, management and sharing,
Harvester Wheatsheaf: New York.
Erard, B. and J. Feinstein (2010) Econometric Models for Multi-Stage Audit
Processes: An Application to the IRS National Research Program, in
Developing Alternative Frameworks for Explaining Tax Compliance, Alm, J.,
Mártinez-Vazquez, J., and B. Torgler (eds). New York: Routledge Publishing.
Feinstein, J. (1991) An Econometric Analysis of Income Tax Evasion and its
Detection, RAND Journal of Economics 22, 14–35.
Hunter, W. and M. Nelson (1996) An IRS production function, National Tax
Journal 49 , 105–115.
Kolm, S. (1973) A Note on Optimum Tax Evasion, Journal of Public Economics 2,
265–270.
Laffont, J.-J. and D. Martimort (2005) The Design of Transnational Public Good
Mechanisms for Developing Countries, Journal of Public Economics 89,
159–196.
Ledyard, J. and T. Palfrey (1999) A Characterization of Interim Efficiency with
Public Goods, Econometrica 67, 435–448.
Marhuenda, F. and I. Ortuño-Ortín (1997) Tax Enforcement Problems,
Scandinavian Journal of Economics 99, 61–72.
Merton, R. (1971) Optimum Consumption and Portfolio Rules in a ContinuousTime Model, Journal of Economic Theory 3, 373–413.
32
Besfamille and Olmos
Mookherjee, D. and I. Png (1989) Optimal Auditing, Insurance and
Redistribution, The Quarterly Journal of Economics 104, 399–415.
Mookherjee, D. and I. Png (1990) Enforcement Costs and the Optimal
Progressivity of Income Taxes. Journal of Law, Economics, and Organizations
6, 411–431.
Sandmo, A. (1981) Income Tax Evasion, Labour Supply and the Equity-Efficiency
Trade-Off, Journal of Public Economics 16, 265–288.
Scotchmer, S. and Slemrod, J. (1989) Randomnes In Tax Enforcement, Journal of
Public Economics 38, 17–32.
Slemrod, J. (1990) Optimal Taxation and Optimal Tax Systems, Journal of
Economic Perspectives 4, 157–178.
Snavely, K. (1988) Innovations in State Tax Administrations, Public
Administration Review 48, 903–910.
Snow, A. and R. Warren (2005) Tax evasion under random audits with uncertain
detection, Economic Letters 88, 97–100.
Xie, D. (2000) Power Risk Aversion Utility Functions, Annals of Economics and
Finance 1, 265–282.
Endnotes
1 Throughout
this chapter, subscripts of functions denote partial derivatives.
exception is the quadratic utility function. But, as Eeckhoudt and Gollier
(1995) assert, this specification does not represent well the preferences of a risk
averse individual.
3 This social welfare function has been studied theoretically by Ledyard and
Palfrey (1999), and used, in a more applied context, by Laffont and Martimort
(2005). This criterion is especially interesting because, as α adopts values
between 0 and 1, it describes a family of standardly used social welfare
functions. In particular, when α = 0, the social welfare function is Rawlsian;
whereas, when α = 1, it is utilitarian.
4 For the sake of simplicity, we assume that tax collection in itself is costless,
both for taxpayers and for the tax administration. If this were not the case, the
model would be biased towards more investment in detection technology. We
thank D. McPartland for raising this issue.
5 In this setting, it is straightforward to verify that rewards for truthful reports
are worthless. Therefore, like Mookherjee and Png (1990) and Marhuenda and
Ortuño-Ortín (1997), we rule them out of the model.
6 As ƒ ≥ 0, imposing an after-audit limited-liability constraint also ensures t ≤ y.
2 One
33
Investing in Detection Technology
7 Since
the number of taxpayers is very large, none of them considers the impact
of non-complying with the tax law on the amount of public good. So the
public good does not appear in the incentive-compatibility constraint.
8 Without any loss of generality, we do not impose the net tax collection to be
(weakly) positive because this should hold at the optimum. If this were not
the case, it is straightforward to realize that not taxing the rich (and thus not
enforcing the tax law) would dominate.
9 Throughout the chapter we assume that the parameters of the model ensure
that π < 1. In view of, for example, the current IRS’s audit policy, where audits
represent near 1 percent of filed returns, this assumption is far from being
unrealistic.
10 We assume that the parameters of the model are such that these inequalities
are satisfied.
11 Given the properties of the function P( ), I is unique.
Appendix
Characterization of the Optimal Tax Law Ee
w> ie>
e
First-order condition
Formally, the government solves the following problem
S
;
Pd{ xE| w n j
A
A
w>i> >j
A
A
A
A
A
A
vxemhfw wr
A
A
A
A
A
A
? fw
f
A
A
A
A
A
wni |
EOO
A
A
A
A
A
A
x E| w E S x E| n S x E| w i ELF
A
A
A
=
j ' w E f L
EE
To solve this problem, we momentarily neglect the constraints on the audit probability, as mentioned in the text. At the optimum, (LL′ ) and (IC) bind. From these
binding constraints, we obtain the optimal audit probability
34
Besfamille and Olmos
'
S
x E| w
=
x E|
Then, we obtain j from (B′ ). So, replacing π and j in Z , the maximand of S
becomes
x E| w
xE| w n w E
f L=
+9,
S
x E|
The first-order condition that characterizes an interior optimal tax w is
xt E| e
w '
E f
E f n S x E|
+:,
By strict concavity of (6), (7) is also sufficient to find the unique optimal tax. Also,
w wW =
by strict concavity of the utility function x( ), the optimal tax verifies e
Comparative statics
w is a continuous function of the deBy the Maximum theorem, the optimal tax e
e In order to completely characterize the
tection probability S and so are
e and i=
e
w> i>
e > we apply the Implicit Function theorem to (7) and we
optimal tax law Ee
obtain
E xE| f
Ce
w
'
A f>
CS
V
#
$
xt | e
w Ce
Ce
w
'
e
W f>
CS
S
x E| CS
2 S xE|
Ce
E
Ce
j
w
'
n
fe
Af
CS
E f n S xE| CS
S
w dE f n S xE| o2 ? f=
where V ' xtt E| e
Domain of positive taxation
Replacing e
w by its lowest bound in (7) and rearranging, we obtain
S '
E fxt E|
=
d xt E| oxE|
Hence, when S S > e
w ' f> otherwise, e
w A f=
+;,
35
Investing in Detection Technology
Non-monotonicity of the optimal audit probability
@CS is ambiguous. Despite this fact, we can prove
As shown in (8), the sign of Ce
that, if there are parameter configurations of the model such that
e is non-monotonic, the profile of ES is inverse U-shaped.
Let’s compute
Ce
w
C 2e
w
'
CS 2
CS
and
#
$
w Ce
xttt | e
2xE|
w
?f
xtt E| CS
E f n S xE|
5
# $2 6
7 xt | ew C 2ew xtt | ew
C 2 e
Cew 8
'
CS 2 C e 'f
S
x E| CS 2
x E|
CS
CS
'
$
%
#
&
Cew Cew xt | ew xttt | ew
2xE|
xtt | ew
S x E| CS CS
xtt E| w
E f n S xE|
By assumption, the term in brakets is negative. So this second derivative is also
negative: when Ce
@CS = 0 the optimal audit probability
e attains a local maximum. By contradiction, this critical value of
e has to be a global maximum.
Characterization of an (interior) optimal investment LD
Here, we adopt a parametric configuration such that the constraint set is not
empty. Under this circumstance, this set is bounded by 0 and by wW (i.e. the tax
collection under full information). Moreover, this set is also closed because it is
eES are continuos in
wES and
defined by weak inequalities and the functions e
L . w Hence, the constraint set is compact. In addition, the maximand in S2 is also
continuos in L . wSo, by the Weierstrass theorem, the problem S2 has a maximum.
First-order condition
To find the (interior) optimal investment LD , the government solves the following
problem
;
A
Pd{ xE| e
w n e
w E
ef L
A
A L
?
S2
vxemhfw wr
A
A
A
=e e
w ' wES >
e'
eES > S ' S EL>
36
Besfamille and Olmos
By an envelope argument, the first-order condition for an interior solution of
problem S2 is given by
#
$ $
#
Ce
w
xt E| e
CHZ D
w
e
e
'
SL EL> xt | w n E
f
CL
CS
xE|
+<,
~}
=
0
e ' f=
E fe
S SL EL>
Comparative statics
Assuming that
]'
+#
$
,
xt E| ew Cew
e 2 n eS EL>
e SLL EL>
e E f ? f
2e SL EL>
e 2
xE| CS
S EL>
(i.e. that the solution characterized by (9) verifies the second-order condition for a
maximum), we apply the Implicit Function theorem to (9) and compute
r
E e
f q
C Le
e SSf EL>
e SLSf EL>
e B f
2SL EL>
'
e
CSf
] S EL>
r
E e
C Le
f q
e S EL>
e SL EL>
e B f
'
2SL EL>
e
C
] S EL>
$
#
E
C Le
w Ce
w
xt E| e
e
e
'
fn
eS EL> B f
S EL>
e 2 L
Cf
] S EL>
xE| Cf
$
#
e
e
C Le
E|
w
f
C
w
x
t
e B f
e
'
E
eS EL>
S EL>
e 2 L
C
] S EL>
xE| C
$
#
e
e
e
C Le
E f
xE|
w
x
E|
w
E|
C
w
x
t
t
e
'
E n
Af
S EL>
e 2 L
C|
] S EL>
xE| C|
xE|
e e
E f
CL
e xE| w C w ? f
'
SL EL>
e 2
C
] S EL>
xE| C
6
D
Poster Session
Hess Yeager Bourque Sriuthai
In addition to the papers presented at the main sessions of the conference, several researchers had been invited to describe their research projects at an informal
poster session during the evening of the first day of the conference. These are often
simple descriptive analyses, preliminary analyses, or otherwise limited research
projects. The purpose was to provide an opportunity for conference attendees to
talk with these researchers about their work and to exchange ideas, but did not
include any input from a discussant.
A Study of Preparer Testing:
Exploring the Relationship
Between Preparer Testing and
Tax Preparation Accuracy
Karen Yeager, Christopher Hess, Michael Bourque, Amy Sriuthai,
and Christine Glass, Internal Revenue Service
I
n 2009, the Internal Revenue Service (IRS) initiated a return preparer study to
inform decision-makers proposing regulation of the paid preparer industry.
Based on the study’s results, the IRS announced in January 2010 that all paid
preparers will be required to register with the Service and complete continuing
education requirements.1 In addition, paid preparers without a designation of enrolled agent (EA), certified public accountant (CPA), or attorney will be required
to pass a competency test. In July 2010, the IRS announced that beginning in 2011,
two competency exams will test preparers’ expertise in completing basic and complex Form 1040 series tax returns.2
Background
Prior to this change, the Internal Revenue Service did not require tax return preparers to pass a certification examination or competency test to assess their ability to accurately prepare tax returns for compensation. Other than the Special
Enrollment Examination (SEE) administered to candidates seeking enrolled agent
status with the IRS under Treasury Department Circular No. 230, the licensing
and examination of other tax practitioners, such as certified public accountants
and attorneys, has been the purview of the states.3 Two States already require
registration of paid preparers, Oregon since 1973 and California since 1997. New
York and Maryland have recently enacted legislation requiring paid preparers to
register with the state. However, only Oregon and Maryland require that licensed
preparers pass a competency exam.4 Although these requirements may assist the
government and general public in identifying professionals who have the expertise
to competently prepare tax returns, the range of regulatory frameworks governing
the practice of tax preparation is inconsistent across States.
There are few studies that examine the difference in tax preparation accuracy between certified and non-certified tax preparers. Recently, the Government Accountability Office (GAO) and the Treasury Inspector General for Tax
4
Yeager, Hess, Bourque, Sriuthai, and Glass
Administration (TIGTA) published reports that question the competency of preparers who have not passed a certifying exam.
This study explores the relationship between preparer testing and tax preparation accuracy by examining the adjustment rate for returns prepared by three
groups of paid preparers. The paper is organized as follows: the first section will
review and analyze past research detailing the role of paid preparers in tax compliance. The second section, in keeping with the main research objective, will
review and analyze past research relating to the relationship between testing and
competency. The third and fourth sections will discuss the research design and
study results. Overall, the results are consistent with the assumption that there
may be a relationship between testing and tax preparation accuracy; where differences are observed, the returns of tested preparers were more accurate than those
of untested preparers.
The Role of Paid Preparers in Tax Compliance
Prior to the analysis, a review of published research studies about paid preparers
was conducted to identify what is currently known about preparer accuracy and
competency. There is limited research that examines the performance of a tested
population of tax preparers relative to a non-tested population. However, there
are a few research studies that have examined errors made on returns prepared
by paid preparers. The findings suggest that errors made by paid preparers may
contribute significantly to the tax gap.
GAO (August 2008) audited a sample of 2001 tax returns collected by the National Research Program (NRP) to examine the effectiveness of paid preparer
regulation on compliance.5 Findings revealed that, when comparing the accuracy
of reported income, expenses, and deductions on each return, returns prepared
by paid preparers had higher error rates than self-prepared returns (GAO-08-781,
p.6-7). Empirical information collected by GAO auditors supports the suggestion
that paid preparers may be contributing to noncompliance. While the report did
not distinguish results between the types of paid preparers, they did, however,
segregate returns filed in both California and Oregon from those filed in the rest
of the country. At the time of this study, California and Oregon were the only two
States that required registration of paid preparers. GAO found that returns prepared by an Oregon paid preparer were more likely to be accurate compared with
the rest of the country, including California. At the time, Oregon was the only
State that required competency testing of paid preparers.
A TIGTA report (September 2008) independently reached similar conclusions.
Using a participant-observational research method, auditors posed as taxpayers
at various commercial and independently-owned tax return preparation offices to
assess paid preparer accuracy and competency. Some 28 returns were prepared
A Study of Preparer Testing
5
with 17 containing mistakes and omissions caused by human error and/or misinterpretation of the tax law. TIGTA auditors identified six preparers who acted
willfully or recklessly when determining the number of deductions for the taxpayer (TIGTA, 2008, p.2).
Holtzblatt and McCubbin (2004) examined the Earned Income Tax Credit
(EITC) overclaim rate for Tax Year 1999. Almost one-third of the EITC claimants used the services of attorneys, certified public accountants, enrolled agents,
or preparers affiliated with a nationally recognized tax preparation service. For
these preparers, 25.2 percent of the EITC was claimed in error. An additional
one-third of the EITC claimants reported using another type of paid preparer; the
EITC error rate for this group was 36.2 percent. Although there was an observable
difference between these two types of preparers in the amount of EITC claimed in
error, the researchers could not determine the extent to which that difference was
attributable to the skill of the preparer or characteristics of the clients.
Bloomquist, Albert, and Edgerton (2007) compared AUR discrepancy rates
among self-prepared and paid-prepared returns and found that the latter accounted for higher numbers and larger percentages of AUR cases. The study noted that,
within the paid preparer population, as firm size increased, the number of returns
with AUR discrepancies tended to decrease. In order to determine the influence
that paid preparers have on the outcome of a tax return, the study recommends
that the intentions and quality control procedures used by tax practitioners should
be examined further.
The common theme among these studies is that errors made by paid preparers may contribute to the tax gap and that additional research is needed to assess
the level of tax preparation accuracy and key characteristics of the paid-preparer
community. However, there is little demographic data available, making it difficult
to describe the type of preparer most likely to prepare inaccurate tax returns. In
addition, analysts cannot quantify the effect of mistakes made by paid preparers because there is no accurate count of active paid preparers.6 While there are
several IRS systems that collect limited program-specific information about tax
preparers, currently there is no single data source or common unique identifier
to assist researchers in differentiating paid preparers. In 2009, TIGTA examined
two of these systems, the Centralized Authorization File (CAF) and the Enrolled
Practitioner Program System (EPPS), to assess the quality and accuracy of the data
on paid preparers.7 The report sampled 139 preparers and found that a majority
had multiple identification numbers or inconsistencies in their street addresses or
locations (p. 2).
Both the GAO and TIGTA reports recommended implementing additional
data collection procedures to help monitor and track paid preparer accuracy and
competency. The 2009 TIGTA report urged the Service to prioritize the implementation of preparer registration to allow the IRS to identify paid preparers by
the 2011 tax filing season. Additionally, this report recommended that, in order to
6
Yeager, Hess, Bourque, Sriuthai, and Glass
prepare client returns for compensation, paid preparers should be required to be
compliant with their own Federal tax obligations.8
Certification and Performance in Business and
Industry
Since this study is one of the first to explore more comprehensive measures of tax
preparation accuracy to assess the competency of paid preparers, literature specific to that topic is very limited. However, empirical researchers in other disciplines,
such as nursing and information technology, have conducted studies exploring the
relationship of competency testing or certification to performance.
In a 1990 article, Blits and Gottfredson assessed the validity of general mental
ability tests as a predictor of job performance. Although these tests have been
criticized as racially biased, the authors referred to research studies that show
that, regardless of race, lower test scores are often accompanied by lower job performance. Specifically, their report cited a National Academy of Sciences investigation that found a 0.3 correlation between the General Aptitude Test Battery
(GATB) score and job performance. This means that with a “perfectly valid test,”
30 percent of the gains in workforce productivity could be attributed to tests like
the GATB (p. 21).
A 3-year study performed by International Data Corporation (2009) examined
the relationship between training and certification of information technology
professionals and its impact on network administration functions. 9 Researchers
found that teams with the most certified members had 10 percent more devices
in full compliance with security policies. Findings also revealed that, in organizations with greater concentrations of certified staff, applications and network
capabilities were about 10 percent more likely to be deployed on time and within
budget and unscheduled downtime was about 20 percent lower.
In a healthcare study, the differences in performance scores for certified and
non-certified nurses (n=83) were examined based on six dimensions of nursing
performance: leadership, critical care, teaching/collaboration, planning/evaluation, interpersonal relationship, and professional development. As rated by supervisors, certified nurses had consistently higher performance scores on each
of the six dimensions; however, only two, teaching/collaboration and planning/
evaluation, were found to be statistically significant (Redd and Alexander, 1997).
Despite suggestions from the pertinent empirical literature that testing or certification may contribute to improved accuracy or work performance, there is little
basis to definitively establish a causal relationship between certification testing or
training and performance. Such links may be difficult to establish due to the complex nature of human behavior. Instead, certification or testing may represent
“one factor in measuring competence [but] by no means the determining factor”
A Study of Preparer Testing
7
(Glassie and Jacobs, 2003, p. 18). The purpose of this study was to explore the
relationship between preparer testing and tax preparation accuracy by comparing
error rates across different segments of the preparer population to determine if
there were observable differences in return accuracy.
Research Design
This study utilizes a cross-sectional design to collect data for variables of interest
during a specified period of time to compare the accuracy of tested enrolled agents
and non-tested preparers (enrolled agents and others). Although it would be preferable to observe the change in accuracy before versus after testing within the
population of preparers required to pass a competency test, classical experimental
design of this nature was not possible since the IRS has not yet implemented mandatory preparer testing.10
Data Collection
Data from individual tax returns prepared by each preparer were extracted from
three databases that reside in the IRS Compliance Data Warehouse (CDW). Data
collection was limited to three tax years, 2005 to 2007. These tax years represent
the most current tax returns that have been subject to review.
• The Individual Returns Transaction File (IRTF) stores transactional
and entity information for each individual income tax return,
including identification of the signing preparer.
• The Examination Operational Automation Database (EOAD)
provides data that tracks examination results for returns by issue.11
EOAD is the primary means of sharing information from Federal
revenue agent reports (RARs) and consists of audit reports closed
by Small Business/Self Employed and Wage and Investment
business units in the field and campuses.12 The extract for this
study included issue category codes and adjustment amounts
believed to reflect tax-return accuracy.13 An error was defined as
either a negative or positive adjustment to taxes, credits, or taxable
income; examinations resulting in “no change” were excluded.
• The Automated Underreporter program (AUR) matches thirdparty information reporting against individual income tax returns
to verify that income and deductions are reported correctly. When
discrepancies are found between reported data and the tax return as
filed, the return is flagged with an indicator that identifies the type
of income or deduction and, if necessary, the tax is adjusted. Using
8
Yeager, Hess, Bourque, Sriuthai, and Glass
tax topics covered by the Special Enrollment Examination (SEE) as
a guide in choosing indicators of interest, tax records with selected
discrepancy indicators were extracted from the database.14 Only
records that had been selected by analysts using the AUR selection
tool, and subsequently worked by tax examiners and resulted in a
tax adjustment, were included.
To determine tax-return accuracy for each preparer, data was extracted for client returns that had been adjusted by AUR or EOAD. Preparer Social Security
numbers and Preparer Tax Identification Numbers (PTINs) were matched to the
paid-preparer identification field on client records present in the AUR and EOAD
databases.
For each preparer, a count was compiled of all individual tax returns prepared
for each tax year. Both adjusted and unadjusted returns were counted in IRTF
by matching preparer Social Security numbers and PTINs to the paid-preparer
identification field on client returns. The resulting tax-return volume was used
in the analysis to calculate adjustment rates for each preparer and was also used
as a proxy for preparer experience. Preparers who could not be matched to client
records were considered to have prepared no returns for the purposes of this study.
To avoid distortion of the adjustment rates that could skew the results, this analysis
was limited to include only those preparers who prepared at least 25 individual
returns for a tax year.
Target Population
There were three populations of interest for this study: two groups of enrolled
agents (EAs) and one group of other tax preparers (non-EAs).
Enrolled agents are granted authority to represent taxpayers before the Internal
Revenue Service to the same extent as attorneys and certified public accountants.15
The designation is awarded to practitioners who have demonstrated special tax
competency by passing a rigorous written examination or who were granted exemption from the Special Enrollment Examination based on a qualifying former
occupation with the IRS. All enrolled agents are subject to a 3-year renewal cycle
during which they are required to earn 72 hours of continuing professional education (CPE) in order to maintain their status. The enrolled agent population in this
study was limited to practitioners who were enrolled to practice between 1992 and
2004 and whose status was “active” (n=19,516). These practitioners were more
likely to be actively engaged in providing tax preparation services for the tax years
under study and would have experienced at least one renewal cycle. Identification
and demographic information was extracted from the Enrolled Agent database
which consists of registration and renewal data provided by the Office of Professional Responsibility (OPR).
9
A Study of Preparer Testing
This initial enrolled agent study population was reduced by 33 percent after
restricting the study to enrolled agents who prepared at least 25 returns in a tax
year. Of the resulting 12,996 practitioners, 12,507 achieved enrolled agent status by
passing the SEE (Tested EAs), and 489 received waivers from the testing requirement by qualifying as former IRS employees (Untested EAs).
The third group, “non-EAs” (n=19,450), was selected by a simple random sample from approximately 900,000 preparers who had been assigned a Preparer Tax
Identification Number (PTIN) but were not present in the Enrolled Agent database. From the data that was available, the professional status or level of education
of preparers in this group could not be identified; it may include attorneys, CPAs,
former IRS employees not granted EA status, as well as other preparers without
certification or professional designation. After compiling return counts for this
group of preparers, there were 11,157 that could not be identified as signing preparers, which reduced the group size by 57 percent to 8,293. The non-EA group
was further reduced to 5,737 after eliminating those who prepared fewer than 25
returns in a tax year.
For each of the three groups of tax preparers under study, Table 1 below shows
the return volume for each tax year from 2005 to 2007. The declining trend in the
number of returns prepared over the 3-year period is likely due to attrition as tax
preparers leave the profession.
TABLE 1: Total Number Of Returns Prepared By Preparer Type And Tax Year
Preparer Type
Untested EAs
Tested EAs
Non-EAs (Unweighted Sample)
TOTAL
2005
76,589
3,262,114
928,918
4,267,621
2006
74,043
3,222,342
878,648
4,175,033
2007
71,541
3,183,808
828,063
4,083,412
For each tax year from 2005 to 2007, Table 2 below shows the final number of
preparers in each group after extracting client-adjusted return data from AUR and
EOAD.
TABLE 2: Preparer Population By Type And Tax Year
Preparer Type
Untested EAs
Tested EAs
Non-EAs (Unweighted Sample)
TOTAL
2005
410
11,711
4,552
16,673
2006
388
11,330
4,226
15,944
2007
356
10,881
3,805
15,042
10
Yeager, Hess, Bourque, Sriuthai, and Glass
Limitations
The study utilized data collected during the tax administration process.
Compliance data records typically contain multiple observations per taxpayer for
a given tax year, which adds a layer of complexity. Efforts were made to omit any
irrelevant, incorrect, or duplicative data to prevent misleading results.
Tax preparation accuracy depends on a variety of factors, and this study did
not attempt to control for conditions that may impact paid-preparer performance
such as the complexity of a return, reliance on tax-preparation software, or taxpayer influence. Errors may be attributed to the taxpayer, the preparer, or both.
Little descriptive information is currently available that could have added depth to
the findings, such as education and professional designation other than enrolled
agent status (e.g., attorney or CPA) or data such as firm size or other characteristics which may influence tax return accuracy. Additionally, there was no control
for selection bias, either by the taxpayer or the preparer, which may be a subject
for future research.16
If a preparer could not be matched to client records as the return signer, the
preparer was excluded from analysis. Although IRS guidelines require tax preparers to sign returns prepared for compensation, tax code complexity often requires
additional assistance in completing a return. The inability to identify paid preparers who may have contributed to, but did not sign, the completed return required
that these contributing preparers be omitted from the study.
In some cases, there were a limited number of observations for various AUR
and EOAD adjustment categories and, although there may be an observed difference between the study groups, a small number of observations may reduce
any statistical or practical significance. The small population of untested enrolled
agents may also limit extrapolating results beyond this subset of the enrolled agent
population. The analysis relied on error rates from operational enforcement data
and did not take into account the nature and magnitude of the errors.17
Analysis of Error Rates
To compare the accuracy of the returns prepared by our three groups of preparers, three ratios were calculated utilizing the data extracted from IRTF, AUR, and
EOAD for each tax year:
• Adjusted Error Rate (AER)—the number of returns adjusted by
AUR divided by the total number of returns prepared;
• Exam Error Rate (EER)—the number of returns adjusted by EOAD
divided by the total number of returns prepared; and
11
A Study of Preparer Testing
• Combined Error Rate (CER)—the total number of returns
adjusted by AUR and EOAD divided by the total number of returns
prepared.
The calculation of ratio scores using operational data as a measure of preparer accuracy has an established precedence in preparer research and was used by
Bloomquist et al. (2007) to evaluate preparation accuracy, as well as by McKerchar,
Bloomquist, and Leviner (2008) in exploring the use of regulation to improve the
services offered by tax agents.18
FIGURE 1. Adjusted Error Rates (AER)
3.0%
2.79%
2.5%
2.58%
2.18%
2.0%
1.5%
1.0%
0.5%
0.0%
Untested EAs
Tested EAs
Non-EAs
Adjusted Error Rate: Figure 1 shows the adjusted error rate (AER) for the three
preparer populations for Tax Years 2005 to 2007. The tested EA population had
a lower error rate than both the untested and non-EA populations. For returns
prepared by tested enrolled agents, 2.18 percent had a discrepancy detected by the
Automated Underreporter program (AUR). Of the three groups, untested EAs
had the highest adjusted error rate, 2.79 percent. This rate is marginally higher
than the non-EA group, who experienced an adjusted error rate of 2.58 percent.
With a difference of more than 0.6 percent between tested and untested enrolled
agents, these results suggest that testing may influence tax preparation accuracy.
12
Yeager, Hess, Bourque, Sriuthai, and Glass
FIGURE 2: Exams Error Rates (EER)
0.6%
0.57%
0.5%
0.4%
0.3%
0.34%
0.34%
Untested EAs
Tested EAs
0.2%
0.1%
0.0%
Non-EAs
Exam Error Rate: Figure 2 shows the error rate for examined returns (EER) for
the 3-year period 2005–2007 by preparer type. Both tested and untested enrolled
agents at 0.34 percent have a lower exam error rate than non-EAs (0.57 percent).
Tested and untested enrolled agent populations had the same exam error rate suggesting that for adjustments made by examination, which is a subjective process,
testing may not be a factor. Rather, other factors may influence the observed difference between the enrolled agent and non-enrolled agent groups.
The 0.23-percent difference between the enrolled agent and non-enrolled agent
groups may suggest that practitioners who are granted the privilege to represent
clients before the IRS are better able to argue their clients’ positions. As noted by
Galanter (1974), enrolled agents and other practitioners governed by Treasury Department Circular No. 230 may have a better understanding of IRS practices and,
due to the frequent interactions with the Service, may have developed working
13
A Study of Preparer Testing
relationships with examination personnel. These two factors may improve their
chances in resolving issues to the benefit of their client (as cited in Ayres, Jackson,
and Hite, 1989).
FIGURE 3: Combined Error Rates (CER)
3.5%
3.0%
3.15%
3.13%
2.5%
2.52%
2.0%
1.5%
1.0%
0.5%
0.0%
Untested EAs
Tested EAs
Non-EAs
Combined Error Rates: Figure 3 illustrates the combined error rate (CER) for the
3-year period 2005–2007 by preparer type. Tested enrolled agents have a lower
combined error rate than non-EAs and untested EAs, 2.52 percent, 3.13 percent,
and 3.15 percent respectively. With a difference of more than 0.6 percent between
tested enrolled agents and both untested enrolled agents and non-EAs, the results
are consistent with the assumption that testing may be a contributing factor in
reducing errors and improving tax preparation accuracy.
14
Yeager, Hess, Bourque, Sriuthai, and Glass
FIGURE 4: Effect OF Continuing Professional Education On CER
4.0%
3.5%
3.0%
2.5%
3.49%
3.38%
3.27%
2.68%
3.29%
2.58%
2.73%
2.52%
2.0%
1.5%
1.0%
≤2 years
3 - 5 Years
Untested EAs
6 - 8 Years
≥ 9 Years
Tested EAs
Effect of Continuing Education: For both groups of enrolled agents, tested and
untested, the amount of required continuing education was thought to contribute to preparer accuracy. To approximate the amount of continuing professional
education (CPE) the practitioner had completed, the number of years since the
practitioner’s enrollment date (EA tenure) was calculated and used as a proxy.19
An assumption was made that longer tenure reflects more renewal cycles, which
should result in greater tax expertise gained from accumulated CPE hours.
Figure 4 above illustrates the combined error rates for both groups of enrolled
agents. Across all four levels of tenure, tested enrolled agents consistently had a
lower combined error rate (CER) than untested EAs. With both groups subject
to the same continuing education requirements, the results show little evidence
that professional development alone affects preparation accuracy. However, in
this study, there was no attempt to control for tax return complexity. It is likely
that as preparers become more experienced, the level of complexity in the returns
they prepare will also increase.
For tested enrolled agents, the number of years since enrollment was also used
to observe the test effect, or the potential impact that testing and test preparation has on tax return accuracy. Although there was almost a 0.59 percent difference in the combined adjustment rate between tested and untested EAs in the first
renewal cycle, this gap was smaller than the differences in the second and third
renewal cycles which were 0.80 percent and 0.97 percent, respectively. Any test
effect would be indicated by a larger gap in the first cycle.
15
A Study of Preparer Testing
FIGURE 5: Combined Error Rate By Volume Of Returns Prepared
4.0%
3.5%
3.0%
2.68%
2.94%
3.16%
3.60%
2.50%
2.41%
2.79%
2.52%
1.0%
2.46%
3.10%
1.5%
3.10%
2.0%
3.40%
2.5%
0.5%
0.0%
Untested EAs
Tested EAs
25-199
200-349
Non-EAs
350-749
750 ≤
Figure 5 shows the combined error rate (CER) for Tax Years 2005–2007 for
each preparer type segmented by the volume of tax returns prepared. Serving as
a proxy for experience, it is assumed that higher return volume indicates a preparer has more experience than a preparer with lower return volume. Overall, in
three (25-199, 200-349, and 350-749) of the four return-volume segments, tested
enrolled agents have the lowest combined error rate compared to both untested
enrolled agents and non-EAs.
For all of the preparer groups, there appears to be a decline in the combined
error rate as return volume increases. This decline suggests that, when using return volume as a proxy, more experienced return preparers are likely to be more
accurate.
16
Yeager, Hess, Bourque, Sriuthai, and Glass
Tax Issues by Frequency
Since differences in tax preparation accuracy are observable across the three preparer types, this study attempted to identify where the differences in tax-preparation accuracy are occurring by taking a closer look at the types of adjustments
experienced by each preparer segment.
TABLE 3: Issue Adjustment by Type as Reported by Examination
Issue Adjustment Type
Adj. to Tax before Credits
EIC
Prepayment Credit
EITC earned income per exam
Additional Child Care Credit
Excess Misc. Deductions Per Exam
Child Tax
T Credit
C dit
Self Employment Tax
EITC Investment income per exam
Contributions Per Exam
Tested EAs
18.56%
9.21%
6.30%
6.02%
6.01%
5.65%
5.40%
5 40%
3.72%
3.56%
2.93%
Rank Untested EAs
1
18.22%
2
8.31%
3
6.16%
4
4.08%
5
5.37%
6
4.87%
7
5.59%
5 59%
8
4.91%
9
4.08%
10
2.90%
Rank
1
2
4
9
6
8
5
7
9
11
Non-EAs
17.17%
14.64%
10.04%
3.86%
8.83%
3.71%
6.12%
6 12%
2.68%
2.36%
1.58%
Rank
1
2
3
6
4
7
5
8
9
11
Table 3 above shows the percent of paid-preparer returns adjusted by examination by type of issue. For the tested enrolled-agent group, the top 10 issues
were listed in rank order, with these results then compared to the other 2 groups.
There appears to be little variation in the top 10 issue types across the 3 preparer
groups with the exception of EITC earned income per exam, which ranks 4th for
tested EAs, 9th for untested EAs, and 6th for non-EAs. Despite tested EAs having
a greater overall accuracy rate, there is no consistent pattern in the ranking of adjustments among the three groups. However, this study did not attempt to control
for tax return complexity.
TABLE 4: Discrepancy by Type as Reported by AUR
Discrepancy Type
Withholding
Mortgage Interest
Wages
Interest Income
Other Income
Taxable SSA
Di id d
Dividends
Gambling Income
Unemployment
Pension
Grant/Debt Cancellation Income
IRA Payments
Tested EAs
14.31%
12.76%
12.36%
12.19%
10.67%
8.08%
7.82%
7 82%
5.81%
3.28%
2.91%
2.66%
2.40%
Rank Untested EAs
1
14.87%
2
14.56%
3
10.62%
4
13.25%
5
10.93%
6
8.15%
7
7.41%
7 41%
8
5.60%
9
3.07%
10
2.71%
11
2.37%
12
1.49%
Rank
1
2
3
4
5
6
7
8
9
10
11
12
Non-EAs
15.09%
11.16%
14.76%
11.71%
10.33%
7.10%
7.91%
7 91%
4.87%
3.88%
2.44%
2.44%
2.26%
Rank
1
4
2
3
5
7
6
8
9
12
11
14
A Study of Preparer Testing
17
Table 4 above illustrates the percent of returns adjusted by AUR according to
type of discrepancy for each preparer group. For the tested enrolled-agent group,
the top 12 adjustments were listed in rank order, with these results then compared
to the other 2 groups. The rank order of adjustments for tested and untested EAs
is identical. However, greater variation is observed when compared to the non-EA
group.
Conclusions
This study explored the relationship between competency testing and tax-preparation accuracy by examining the adjustment rate for the returns prepared by three
groups of paid preparers: tested enrolled agents, enrolled agents exempt from the
testing requirement, and a group of non-enrolled agent preparers. The findings
suggest there may be a relationship between preparer testing and tax-preparation
accuracy. Using this study’s measures, Federal individual income tax returns prepared by tested enrolled agents for Tax Years 2005–2007 had fewer errors than
those prepared by untested enrolled agents or non-EA preparers. However, based
on the data available it was not possible to confirm that testing alone caused the
difference in accuracy or whether other underlying factors influenced these results.
To explore these potential influences, the effects of testing and continuing professional education (CPE) were isolated for the two enrolled agent groups. There
was no observed relationship between accumulated professional development and
tax preparation accuracy nor did testing alone appear to affect return accuracy
rates for the two groups. Although returns prepared by tested enrolled agents
were consistently better over the 3-year period, it appears that these factors alone
cannot explain why these returns were more accurate.
This study also examined the influence of experience for all three preparer
groups by using return volume as a proxy. Overall, it was found that an inverse
relationship existed between return volume and return error rates. One explanation may be that as preparer experience increases, accuracy rates may improve.
Returns prepared by tested enrolled agents were just as or more accurate than
those prepared by untested EAs and non-EAs across all volume segments. The
non-EA preparer population had the highest combined error rates across all four
volume segments.
In addition to suggesting a possible relationship between competency testing
and tax preparation accuracy, this study also addresses the current deficiencies
in the literature. At present, few studies have investigated the relationship between testing and job performance, and little exists specific to tax preparation accuracy. This study makes an effort to bridge this knowledge gap and provide a
baseline for future research. Regulations enacted in 2010 will soon require all paid
18
Yeager, Hess, Bourque, Sriuthai, and Glass
tax preparers to register annually with the IRS and earn continuing professional
education credits. Preparers who are not enrolled agents, CPAs, or attorneys will
also be required to pass competency exams to demonstrate their expertise in order
to prepare tax returns for compensation. This registration and testing process will
provide rich detail to inform future research.
Acknowledgements
We wish to thank our colleagues in Research, Analysis, and Statistics and the
Office of Program Evaluation and Risk Analysis who provided their critiques,
comments, and suggestions in support of this research—in particular, Patricia
McGuire for her support and Kevin Cecco for his review. We also wish to thank
Karen Hawkins and members of the Office of Professional Responsibility for their
review and sponsorship of this research project.
References
Ayres, F.L., B.R. Jackson, and P.S. Hite (1989). The economic benefits of
regulation: evidence from professional tax preparers. The Accounting Review,
64(2), 300-312.
Blits, J. H. and L.S. Gottfredson (1990). Employment testing and job
performance. The Public Interest, 98, 18-25.
Bloomquist, K.M., M.F. Albert, and R.L. Edgerton (2007). Evaluating
Preparation Accuracy of Tax Practitioners: A Bootstrap Approach. The
IRS Research Bulletin: Proceedings of the 2007 IRS research Conference
(Publication Number 1500).
Glassie, J.C. and J.A. Jacobs (June 2003). Certification programs as a reflection of
competency. Association Management, 55(6), 17-18.
Government Accountability Office. Tax Preparers: Oregon’s regulatory regime
may lead to improved Federal tax return accuracy and provides a possible model
for national regulation (Report Number GAO-08-781).
Holtzblatt, J. and J. McCubbin (2004). Issues affecting low-income filers. In
Aaron, H.J. and J. Slemrod (Eds.), The Crisis in Tax Administration. 148-200.
Washington D.C.: The Brookings Institution.
International Data Corporation. (Nov. 2009). Impact of training on network
administration: Certification leads to operational productivity (Document
Number 220563). Retrieved from http://idcdocserv.com/220563
A Study of Preparer Testing
19
Mckerchar, M., K. Bloomquist, and S. Leviner (2008). Improving the quality of
services offered by tax agents: Can regulation assist? Australian Tax Forum,
23(4), 399-425.
Redd, M.L. and J.W. Alexander (1997). Does certification mean better
performance? Nursing Management, 28(2), 45-50.
Treasury Inspector General for Tax Administration. (2008, September). Most
tax returns prepared by a limited sample of unenrolled preparers contained
significant errors (Report Number 2008-40-171).
Treasury Inspector General for Tax Administration. (2009, July). Inadequate data
on paid preparers impedes effective oversight (Report Number 2009-40-098).
Endnotes
1
2
3
4
5
6
7
8
Department of the Treasury, Internal Revenue Service. Return Preparer
Review. Publication 4832 (Rev. 12-2009) Catalog Number 54419P http://www.
irs.gov/pub/irs-pdf/p4832.pdf.
Proposed New Requirements for Tax Return Preparers: Frequently Asked
Questions, http://www.irs.gov/taxpros/article/0,,id=218611,00.html#Testing
(August 2, 2010).
Treasury Department Circular No. 230 (Revised 4-2008). Regulations
Governing the Practice of Attorneys, Certified Public Accountants, Enrolled
Agents, Enrolled Actuaries, Enrolled Retirement Plan Agents, and Appraisers
before the Internal Revenue Service http://www.irs.gov/pub/irs-pdf/pcir230.
pdf.
Return Preparer Review http://www.irs.gov/pub/irs-pdf/p4832.pdf.
In 2000, IRS established the National Research Program (NRP) as part of its
efforts to develop and monitor strategic measures of taxpayer compliance.
The NRP seeks to increase public confidence in the fairness of our tax system
by helping the IRS identify where voluntary compliance problems occur so
that the IRS can efficiently utilize its resources to address those problems.
In 1999, IRS estimated there were up to 1.2 million paid preparers (GAO-08781, p.6).
Reference Number 2009-40-098.
The principals and responsible officials for firms that apply to participate in
the IRS e-file Program are currently subject to suitability checks that include
compliance with tax requirements. See Publication 1345, Handbook for
20
9
10
11
12
13
14
15
16
17
18
19
Yeager, Hess, Bourque, Sriuthai, and Glass
Authorized IRS e-file Providers of Individual Income Tax Return, http://core.
publish.no.irs.gov/pubs/pdf/64382a01.pdf.
International Data Corporation (IDC) is a provider of market
intelligence, advisory services, and events for the information technology,
telecommunications, and consumer technology markets. http://www.idc.
com/.
Those who are not enrolled agents, certified public accountants, or attorneys.
IRM 4.10.16.1
Examination Operational Database (11/2009), http://mysbse.web.irs.gov/CLD/
GLD/GL/Programs/Exam/4530.aspx (1109).
The extract included records through the 201004 processing period.
The extract for this study included records through the 200852 processing
period.
Treasury Department Circular No. 230 (Revised 4-2008), http://www.irs.gov/
pub/irs-pdf/pcir230.pdf.
Taxpayer compliance intentions may influence their preparer selection
decision.
Operational enforcement data is not representative of the entire population
and thus may reflect a lower bound of errors.
These studies also relied on operational data and are thus subject to the
similar limitations discussed in note 17.
During each 3-year enrollment cycle, enrolled agents must complete 72
contact hours of tax or tax-related education with a minimum of 16 hours
each year.
7
D
Appendix
Conference Program
List of Attendees
Conference Program
3
2010 IRS Research Conference Program
The Liaison Capitol Hill
June 29–30, 2010
DAY ONE : Wednesday, July 8
8:00–8:45
Registration
8:45–9:00
Welcome
9:00–9:30 Keynote Address
Mark Ernst, Deputy Commissioner for Operations Support,
Internal Revenue Service
9:30–10:30 Panel Discussion:
The Impact of Globalization on Tax Administration
Moderator:
Rosemary D. Marcuss, Director, Research, Analysis, and
Statistics, Internal Revenue Service
Panelists:
Michael Danilack, Large and Mid-Size Business, Internal
Revenue Service, Partho Shome, Knowledge, Analysis,
and Intelligence, Her Majesty’s Revenue and Customs, and
Patricia Arteaga, Mexican Tax Administration Service
10:30–10:50
BREAK
10:50–12:20
Compliance of Large Business Entities
Moderator:
Lois Petzing, Large and Mid-Size Business, Internal
Revenue Service
Papers:
• An Examination of FIN 48: Tax Shelters, Auditor
Independence, and Corporate Governance, Petro Lisowsky,
University of Illinois at Urbana-Champaign, Leslie A.
Robinson, Tuck School of Business at Dartmouth and
Andrew P. Schmidt, Columbia University
4
Conference Program
• Partnerships with Reportable Entity Partners, Charles E.
Boynton and Barbara A. Livingston, Large and Mid-Size
Business, Internal Revenue Service
• Temporary and Permanent Book-Tax Differences:
Complements or Substitutes?, Jennifer Blouin, The
Wharton School, University of Pennsylvania, Jason
DeBacker, Office of Tax Analysis, U.S. Department of
the Treasury, and Stephanie Sikes, The Wharton School,
University of Pennsylvania
iscussant:
D
Drew Lyon, PricewaterhouseCoopers
12:20–2:00
Lunch
2:00–2:15
Presentation of IRS Research Recognition Awards
2:15–3:45
Influencing Individual Taxpayer Behavior
Moderator:
Patti Davis-Smith, Wage and Investment, Internal Revenue
Service
Papers:
• Subsidizing Charitable Contributions with a Match vs.
Income Tax Rebate: What Happens to Donations and
Compliance?, Marsha Blumenthal, University of St.
Thomas, Laura Kalambokidis, University of Minnesota,
and Alex Turk, Small Business/Self-Employed Division,
Internal Revenue Service
• Solving Information Asymmetry for Offshore Accounts,
Susan Morse, Hastings College of the Law, University of
California
• Facilitated Self-Assistance Enhances Taxpayers’ Taxpayer
Assistance Center (TAC) Experiences, Kirsten Davis,
Melissa Hayes, and Erica Jenkins, Wage and Investment,
Internal Revenue Service
Discussant:
Leandra Lederman, Indiana University
Conference Program
5
3:45–4:00
BREAK
4:00–5:15
Drivers of Noncompliance
oderator:
M
Rahul Tikekar, Research, Analysis, and Statistics, Internal
Revenue Service
Papers:
• A Balance Due Before Remittance: The Effect on Reporting
Compliance, Paul Corcoro and Peter Adelsheim, Small
Business/Self-Employed Division, Internal Revenue
Service
• Predicting Intentional and Inadvertent Noncompliance,
Kathleen M. Carley, Brian Hirshman, Ju-Sung Lee,
Michael Martin, Dawn Roberston, and Jesse St. Charles,
Carnegie Mellon University
Discussant:
Peggy Hite, Indiana University
5:30–6:30
Poster Session and Social Hour
• Application of Text Mining to Uncover the Issues and
Concerns Surrounding Tax Preparers, Ririko Horvath,
Larry May, Rahul Tikekar, and Cheryl Wagner, Research,
Analysis, and Statistics, Internal Revenue Service
• Compliance, Assistance, and the SmartCard, Kathleen
M. Carley, Neal Altman, and Michael Martin, Carnegie
Mellon University and Joanne Meikle and Traci L. Suiter,
Research, Analysis, and Statistics, Internal Revenue
Service
• Understanding Tax Professionals’ Work Processes Using
the In Basket Technique, Courtney L. Rasey, Wage and
Investment, Internal Revenue Service
• An Analysis of Preparer Testing on Compliance, Chris
Hess, Karen Yeager, Michael Bourque, and Amy Sriuthai,
Research, Analysis, and Statistics, Internal Revenue
Service
6
Conference Program
DAY TWO : Thursday, July 9
8:30–10:00
Tax Code Complexity and Compliance Burden
Moderator:
Kara Leibel, Research, Analysis, and Statistics, Internal
Revenue Service
Papers:
• Individual Taxpayer Compliance Burden: The Role of
Assisted Methods in Taxpayer Response to Increasing
Complexity, George Contos, John Guyton, Patrick
Langetieg, and Melissa Vigil, Research, Analysis, and
Statistics, Internal Revenue Service
• Enhancing Compliance Through Improved Readability:
Evidence From New Zealand’s Rewrite “Experiment,”
Adrian Sawyer, University of Canterbury
• Tax Compliance Costs: The Effect of Authority Behavior
and Taxpayer Services, Sebastian Eichfelder, University
of Wuppertal, Chantal Kegels, Federal Planning Bureau,
and Michael Schorn, Institute for Economy and Policy
Research
Discussant:
James R. Nunns, The Urban Institute
10:00–10:20
BREAK
10:20–11:50
Enforcement Strategies
Moderator:
Katie Fox, Small Business/Self-Employed Division, Internal
Revenue Service
Papers:
• Collecting Collected Taxes, Keith Fogg, Villanova
University School of Law
• Measuring and Tackling the Illicit Market for Excise Goods,
Anthony Rourke, Knowledge, Analysis, and Intelligence,
Enforcement and Compliance, Her Majesty’s Revenue
and Customs
Conference Program
• Inspectors or Google Earth? Optimal Fiscal Policies Under
Uncertain Detection of Evaders, Martin Besfamille and
Pablo Olmos, Universidad Torcuato Di Tella
Discussant:
Don McPartland, Large and Mid-Size Business, Internal
Revenue Service
11:50–12:00
Closing Remarks
7
List of Attendees
IRS Research Conference
The Liaison Capitol Hill
June 29–30, 2010
Louis Acevedow
Andrea Behrmann
Jonathan Adams
Ken Beier
Robert Adams
James Bellefeuille
Peter Adelsheim
Philip Beram
Jane Agule
Martin Besfamille
Sarah Allen
Ted Black
Kay Anderson
Elizabeth Blair
Patricia Arteaga
Kim Bloomquist
Terry Ashley
Jennifer Blouin
Linda Baker
Marsha Blumenthal
Greg Baldwin
Lawrence Boateng
Aaron Barnes
Susan Boehmer
Paul Bastuscheck
Henry Bowman
Tom Beers
Charles Boynton
IRS, Small Business/Self-Employed
Eastport Analytics, Inc.
IRS, Research, Analysis, and Statistics
IRS, Small Business/Self-Employed
IRS, Communications & Liaison
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
Mexican Tax Administration
IRS, National Taxpayer Advocate
U.S. Government Accountability Office
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, National Taxpayer Advocate
IRS, Research, Analysis, and Statistics
Multistate Tax Commission
IRS, Tax Exempt & Government Entities
Philip C. Beram, CPA
Universidad Torcuato Di Tella
IRS, Research, Analysis, and Statistics
IRS, Wage & Investment
IRS, Research, Analysis, and Statistics
University of Pennsylvania
University of St. Thomas
IRS, Wage & Investment
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
8
List of Attendees
Renee Brandon
Renita Carter
David Brazell
Kevin Cecco
Spencer Brien
David Chandler
Michael Brostek
Winnie Chang
Robert Brown
Helen Choi
Tiffanie Bruch
Timothy Chriest
IRS, National Taxpayer Advocate
U.S. Department of the Treasury
IRS, Research, Analysis, and Statistics
U.S. Government Accountability Office
IRS, Research, Analysis, and Statistics
IRS, Wage & Investment
Melanie Bryson
IRS, Small Business/Self-Employed
Tania Buffone
Canada Revenue Agency
Fredericka Bunting
IRS, Large & Mid-Size Business
Thomas Burger
IRS, Research, Analysis, and Statistics
Jeff Butler
IRS, Research, Analysis, and Statistics
Taquesha Cain
IRS, Research, Analysis, and Statistics
Wanda Canada
IRS, Small Business/Self-Employed
Francis Cappelletti
IRS, Large & Mid-Size Business
Kathleen Carley
Carnegie Mellon University
Natalia Carro
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
U.S. Senate Committee on the Budget
IRS, Large & Mid-Size Business
Treasury Inspector General for Tax
Administration
Michelle Chu
IRS, Research, Analysis, and Statistics
James Clarkson
IRS, Large & Mid-Size Business
Lee Cogburn
IRS, Tax Exempt & Government Entities
Ed Cohen
Ernst and Young
Ciyata Coleman
IRS, Wage & Investment
Bill Conroy
IRS, Chief Counsel
George Contos
IRS, Research, Analysis, and Statistics
Paul Corcoro
IRS, Small Business/Self-Employed
Debbie Cortez
IRS, Small Business/Self-Employed
Laurent Corthay
The World Bank
9
List of Attendees
Debi Coyle
Don DeLuca
Lauren Craigie
Joel Deuth
Blaire Crowley
John DeWald
Michael Cullen
Nirmail Dhaliwal
Rafael Dacal
Julia Dhimitri
Kevin Daly
Bertha Dong
Charlie Daniel
Michael Donnelly
Michael Danilack
Steven Doskey
John Davidson
Zlatka Draganova
Jonathan Davis
Ken Drexler
Kirsten Davis
Nicole Duarte
Scott Davis
Alain Dubois
Patti Davis-Smith
Ann Dunbar
Jason DeBacker
Michael Dunn
Marcy Defiel
Ed Eck
Portia DeFilippes
Jonathan Edelson
IRS, Large & Mid-Size Business
ASR Analytics
IRS, Research, Analysis, and Statistics
Canada Revenue Agency
IRS, Small Business/Self-Employed
U.S. Government Accountability Office
U.S. Government Accountability Office
IRS, Large & Mid-Size Business
IRS, Large & Mid-Size Business
IRS, Office of the Commissioner
IRS, Wage & Investment
IRS, Large & Mid-Size Business
IRS, Wage & Investment
U.S. Department of the Treasury
IRS, Small Business/Self-Employed
U.S. Department of the Treasury
IBM
Deloitte Tax, LLP
IRS, Small Business/Self-Employed
Government of the District of Columbia
The World Bank
U.S. Government Accountability Office
IRS, Research, Analysis, and Statistics
MITRE Corporation
IRS, Large & Mid-Size Business
IRS, National Taxpayer Advocate
Tax Notes
IRS, Small Business/Self-Employed
BEA
IRS, Research, Analysis, and Statistics
IRS Oversight Board
IRS, Research, Analysis, and Statistics
10
List of Attendees
Adrienne Edisis
Marcella Garland
Ardeshir Eftekharzadeh
John Garnish
Sebastian Eichfelder
Gwen Garren
Brian Erard
Fikre Gedamu
Donald Evans
Russell Geiman
Ali Feizollahi
Mark Gillen
Mark Felbinger
Christine Glass
Dawn Fitzelle
Annika Glennon
Mary Fitzpatrick
Isaac Goodwin
Terence Fitzpatrick
Eric Gorman
George Washington University
IRS, Research, Analysis, and Statistics
University of Wuppertal
B. Erard & Associates
IRS, Tax Exempt & Government Entities
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Tax Exempt & Government Entities
Oregon Department of Revenue
Science Applications International
Corporation
Keith Fogg
Villanova University School of Law
Cecilia Fong
IRS, Chief Counsel
Katherine Fox
IRS, Small Business/Self-Employed
Javier Framinan
IRS, Wage & Investment
Jon Games
IRS, Small Business/Self-Employed
Martha Gangi
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Small Business/Self-Employed
IRS, Wage & Investment
IRS, Large & Mid-Size Business
IRS Oversight Board
Deloitte Consulting
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
U.S. Government Accountability Office
Paulo Gottgtroy
Inland Revenue Department, New Zealand
Michael Graham
IRS, Research, Analysis, and Statistics
Caryl Grant
IRS, Communications & Liaison
Leon Green
U.S. Government Accountability Office
Arnie Greenland
IBM
Stanley Griffin
IRS, Wage & Investment
11
List of Attendees
Eric Guthrie
Arlene Holen
George Guttman
R. David Holmgren
IRS, Research, Analysis, and Statistics
U.S. Government Accountability Office
John Guyton
IRS, Research, Analysis, and Statistics
Cynthia Hamill
IRS, Tax Exempt & Government Entities
William Hannon
IRS, Research, Analysis, and Statistics
Lance Harris
IRS, Research, Analysis, and Statistics
Carol Hatch
IRS, National Taxpayer Advocate
Melissa Hayes
IRS, Wage & Investment
Michael Hayes
Bureau of Economic Analysis
Janice Hedemann
IRS, Research, Analysis, and Statistics
Mary Henry
IRS, Tax Exempt & Government Entities
Andrea Hicks
IRS, Research, Analysis, and Statistics
Brian Hirshman
Carnegie Mellon University
Peggy Hite
Indiana University
Ronald Hodge
IRS, Research, Analysis, and Statistics
Bobby Hodges, Jr.
IRS, Research, Analysis, and Statistics
Technology Policy Institute
Treasury Inspector General for Tax
Administration
John Holt
Inland Revenue Department, New Zealand
Ririko Horvath
IRS, Research, Analysis, and Statistics
Diana Hubbard
IRS, Small Business/Self-Employed
Taukir Hussain
IRS, Research, Analysis, and Statistics
Mark Hutchens
IRS, National Taxpayer Advocate
Amy Ibbotson
IRS, National Taxpayer Advocate
Harvey Iglarsh
Georgetown University
Jael Jackson
IRS, Research, Analysis, and Statistics
Madeline Jackson
IRS, Research, Analysis, and Statistics
Larry Jacobs
IRS, Wage & Investment
Darien Jacobson
IRS, Research, Analysis, and Statistics
Jason Jarvis
IRS, Tax Exempt & Government Entities
Erica Jenkins
IRS, Wage & Investment
Drew Johns
IRS, Research, Analysis, and Statistics
12
List of Attendees
Barry Johnson
Lawrence Korb
Raul Junquera
Joseph Koshansky
Bonnie Kacher
Melissa Kovalick
Laura Kalambokidis
Elizabeth Kruse
Jason Kall
Leslie Kulick
John Kam
Charles Lacijan
Kenneth Kaufman
Melissa Laine
Chantal Kegels
Pat Langetieg
Lou Ann Kelleher
Eric Larsen
John Kennedy
Brian Latourell
Mark Kennedy
Scott Leary
Ashley Kent
Leandra Lederman
Dan Killingsworth
Jason Lee
Eurry Kim
Ju-Sung Lee
Irene Kim
Wu-Lang Lee
Stephen Klotz
Ellen Legel
IRS, Research, Analysis, and Statistics
The World Bank
IRS, Chief Counsel
University of Minnesota
IRS, Tax Exempt & Government Entities
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
Federal Planning Bureau
IRS, Small Business/Self-Employed
IRS, Small Business/Self-Employed
IRS, Office of Appeals
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Chief Counsel
IRS, Small Business/Self-Employed
U.S. Government Accountability Office
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Tax Exempt & Government Entities
IRS Oversight Board
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Wage & Investment
Canada Revenue Agency
IRS, Tax Exempt & Government Entities
Indiana University, Bloomington
IRS, Research, Analysis, and Statistics
Carnegie Mellon University
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
13
List of Attendees
Kara Leibel
Rachel Maguire
Allen Lerman
Michael Mahn
Pearson Liddell
Grant Mallie
Sandy Lin
Bridget Mallon
Richard Ling
Rosemary Marcuss
Petro Lisowsky
Nicholas Marino
Wei Liu
Michael Martin
Barbara Livingston
Carl Martinez
Christine Longo
Karen Masken
Jose Lopez
Jeff Matsuo
Melissa Ludlum
Thomas Matthews
David Ludlum
Larry May
Andrew Lyon
Mark Mazur
Elaine Maag
Michael McCall
David Macias
Gary McDonald
Alan Macnaughton
Jamie McGuire
IRS, Research, Analysis, and Statistics
U.S. Department of the Treasury
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
University of Illinois at Urbana, Champaign
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
PricewaterhouseCoopers
The Urban Institute
IRS, Large & Mid-Size Business
University of Waterloo
ASR Analytics
IRS, Tax Exempt & Government Entities
U.S. Government Accountability Office
IRS, National Taxpayer Advocate
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
Carnegie Mellon University
IRS, Large & Mid-Size Business
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
HM Revenue & Customs
IRS, Research, Analysis, and Statistics
U.S. Department of the Treasury
IRS, Small Business/Self-Employed
IRS, Small Business/Self-Employed
Joint Committee on Taxation
14
List of Attendees
Pat McGuire
Susan Morse
Megan McInerney Burnham
Carolyn Morton
Michael McKenney
Clay Moulton
IRS, Research, Analysis, and Statistics
Old Dominion University
Treasury Inspector General for Tax
Administration
Judy McNamara
IRS, Large & Mid-Size Business
Monica McNeil
IRS, Research, Analysis, and Statistics
Don McPartland
IRS, Large & Mid-Size Business
Jay McTigue
U.S. Government Accountability Office
Saima Mehmood
IRS, Wage & Investment
Joanne Meikle
IRS, Research, Analysis, and Statistics
Stacie Meredith
IRS, Small Business/Self-Employed
Otey Meriwether
CCH a Wolters Kluwer business
John Miller
IRS, Large & Mid-Size Business
Tri Mobley
IRS, Tax Exempt & Government Entities
Linda Morey
Bureau of Economic Analysis
Lisa Morley
Ernst and Young
Layne Morrison
IBM
University of California
IRS, Small Business/Self-Employed
IRS, Research, Analysis, and Statistics
Gerald Murphy
IRS, Wage & Investment
Maryamm Muzikir
IRS, Small Business/Self-Employed
Ed Nannenhorn
U.S. Government Accountability Office
Marsha Napper
Government of the District of Columbia
Frank Natsuhara
IRS, Small Business/Self-Employed
Susan Nelson
U.S. Department of the Treasury
Mike Nestor
IRS, National Taxpayer Advocate
Nonna Noto
Congressional Research Service
James Nunns
The Urban Institute
Thomas O’Connor
IRS, Research, Analysis, and Statistics
Karen O’Conor
U.S. Government Accountability Office
Jim O’Donnell
IRS, Research, Analysis, and Statistics
Nina Olson
IRS, National Taxpayer Advocate
15
List of Attendees
Roy Olson
Robin Rappaport
Daniel Opitz
Courtney Rasey
Jose Oyola
Brian Raub
Diana Palombo
Robert Reddy
David Paris
Mary Lane Renninger
Debbie Paul
Tammy Rib
Mark Payne
Mary-Helen Risler
Michael Petko
Dawn Roberston
Lois Petzing
Leslie Robinson
Kathleen Pippig
Reuben Robinson
Alan Plumley
Peter Rogers
Phillip Poirier
Gregory Rogozinski
Mark Pursley
Martha Rose
Mikael Pyka
Lisa Rosenmerkel
Brock Ramos
Anthony Rourke
Dave Ramsden
Scott Rutz
IRS, Tax Exempt & Government Entities
IRS, Research, Analysis, and Statistics
U.S. Government Accountability Office
IRS, Office of Appeals
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
National Education Association
IRS, Large & Mid-Size Business
IRS, Large & Mid-Size Business
IRS, Research, Analysis, and Statistics
Intuit, Inc.
IRS, Wage & Investment
Swedish National Audit Office
IRS, Research, Analysis, and Statistics
Inland Revenue Department, New Zealand
IRS, Research, Analysis, and Statistics
IRS, Wage & Investment
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
Carnegie Mellon University
Dartmouth College
IRS, Small Business/Self-Employed
IRS, Tax Exempt & Government Entities
IRS, Office of Appeals
The City University of New York
IRS, Research, Analysis, and Statistics
HM Revenue & Customs
IRS, National Taxpayer Advocate
16
List of Attendees
Deborah Sabatelli
Eric Smith
Cynthia Saunders
Lou Virginia Smith
Adrian Sawyer
Melissa Snell
Andrew Schmidt
Eric Spitzer
Michael Schorn
David Splinter
Ruth Schwartz
Jesse St. Charles
Michael Sebastiani
Kathy Stanek
Emily Shammas
Amy Stanton
Sarah Shipley
Rick Stephenson
Parthasarathi Shome
Anne Stevens
Thomas Short
Kimberly Stockton
Philip Shropshire
John Stout
IRS, Small Business/Self-Employed
U.S. Government Accountability Office
University of Canterbury
Columbia University
Institute for Economy and Policy Research
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Small Business/Self-Employed
HM Revenue & Customs
U.S. Government Accountability Office
Treasury Inspector General for Tax
Administration
Nina Shumofsky
IRS, Research, Analysis, and Statistics
Maxim Shvedov
Congressional Research Service
Stephanie Sikes
University of Pennsylvania
Ann Marie Smith
IRS, Small Business/Self-Employed
IRS, Communications & Liaison
U.S. Government Accountability Office
IRS, National Taxpayer Advocate
IRS, Research, Analysis, and Statistics
Congressional Budget Office
Carnegie Mellon University
National Association of Tax Professionals
IRS, National Taxpayer Advocate
IRS, Small Business/Self-Employed
U.S. Government Accountability Office
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
Andrew Strelka
IRS, Tax Exempt & Government Entities
Michael Strudler
IRS, Research, Analysis, and Statistics
Darren Stuart
IRS, Large & Mid-Size Business
Esmeralda Stuk
IRS, Research, Analysis, and Statistics
17
List of Attendees
Traci Suiter
Patti Turner
Clay Swanson
Wayne Turowski
Camille Swick
Jim Ungvarsky
Zhicheng Swift
Barbara Vaira
Craig Swinford
Stirling Van Winkle
Kelli Tait
Melissa Vigil
Gary Takahashi
Harish Wadhwa
Sarah Talboys
Cheryl Wagner
Bedell Terry
Jeff Wallbaum
Valerie Testa
Hilary Wallis
Katie Thamert
Joann Weiner
Rahul Tikekar
Leann Weyl
Erin Towery
Getaneh Yismaw
IRS, Research, Analysis, and Statistics
IRS, Small Business/Self-Employed
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Large & Mid-Size Business
IRS, Large & Mid-Size Business
Inland Revenue Department, New Zealand
Government of the District of Columbia
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
University of Texas at Austin
Alex Turk
IRS, Small Business/Self-Employed
IRS, Criminal Investigation
U.S. Government Accountability Office
U.S. Government Accountability Office
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
IRS, Tax Exempt & Government Entities
IRS, Chief Counsel
PoliticsDaily.com
IRS, Research, Analysis, and Statistics
IRS, Research, Analysis, and Statistics
| File Type | application/pdf |
| File Modified | 2011-07-27 |
| File Created | 2011-07-27 |