Analysis Plan

Appendix E

Draft Analysis Plan
for the FY2006 HUDQC Study
Quality Control for Rental Assistance
Subsidy Determinations

Prepared for:
Office of Policy Development and Research
Department of Housing and Urban Development
Washington, DC 20410

Contract #: GS-23F-0777H
Task Order #: C-CHI-00829, CHI-T0001

Prepared by:
ORC/Macro
11785 Beltsville Drive
Calverton, MD 20705-3119

July 24, 2006

Table of Contents
Introduction................................................................................................................................... 1
The Dependent Variable............................................................................................................... 2
Preparation of Analytic Files ....................................................................................................... 4
Analysis Plan by Study Objective................................................................................................ 5
List of Exhibits
Exhibit 1 Percent of Households with Proper Payments (FY2005 and FY2006) .........................5
Exhibit 2 Rent Error: Percent of Households in Error, Average Gross Dollars in Error, and
Error Rate (FY2005 and FY2006) .................................................................................6
Exhibit 3a Underpayment Households: Percent of Households and Average Monthly Dollar
Amount of Error (FY2005 and FY2006) .......................................................................6
Exhibit 3b Overpayment Households: Percent of Households and Average Monthly Dollar
Amount of Error (FY2005 and FY2006) .......................................................................7
Exhibit 4 Percent of Households with Calculation and Consistency Errors (FY2006).................8
Exhibit 5 Timeliness of Certification Status (FY2005 and FY2006) ............................................8
Exhibit 6 Average Monthly Underpayment and Overpayment Dollar Amount Averaged
across All Households (FY2005 and FY2006)..............................................................9
Exhibit 7 Procedural Error: Percent of Households, Average Dollars in Error, All
Households with 50058/50059 Recalculated Rent (FY2006) .....................................10
Exhibit 8 50058/50059 Procedural Error: Percent of Households, Average Dollars in Error
(FY2006)......................................................................................................................11
Exhibit 9 Gross and Net Dollar Rent Error (Monthly) for All Households (FY2005 and
FY2006) .......................................................................................................................12
Exhibit 10 50058/59 Rent Calculation Error Compared to QC Rent Error (FY2005 and
FY2006) .......................................................................................................................13
Exhibit 11 Percent of Households in Error and Dollar Error by Error Basis (FY2005 and
FY2006) .......................................................................................................................13
Exhibit 12 The Impact of Program Type on Gross and Net Dollar Error (FY2006).....................14
Exhibit 13 Verification of 50058/50059 Rent Components (FY2005 and FY2006) ....................15
Exhibit 14 Verification of 50058/50059 Rent Components (FY2006) .........................................16
Exhibit 15 QC Error Households with Missing Verification in the Tenant File (FY2005 and
FY2006) .......................................................................................................................16
Exhibit 16 Rent Components Responsible for the Largest Dollar Error Households with
Rent Error (listed by amount of dollar error) (FY2005 and FY2006) .........................17
Exhibit 17 Income and Expense Component Error by Payment Type for All Households
(FY2006)......................................................................................................................18
Exhibit 18 Percent of Households and Standard Error by Rent Component and Payment
Type (FY2006).............................................................................................................18
Exhibit 19 Elderly/Disabled Allowances and Dependent Allowances (FY2006).........................19
Exhibit 20 PHA Section 8 Unit Size Standards.............................................................................19
Exhibit 21 Percent of Households in Units with Correct Number of Bedrooms (According to
Study Guidelines) (FY2005 and FY2006)...................................................................20

Exhibit 22 Percent of All Households by Number of Bedrooms and Number of Household
Members (in thousands) (FY2006)..............................................................................20
Exhibit 23 Percent of Newly Certified Households Meeting Certification Criteria (FY2006).....21
Exhibit 24 Percent of Newly Certified Households Meeting Certification Criteria (FY2006).....22
Exhibit 25 Rent Reasonableness Determination Methods (FY2006)............................................22
Exhibit 26 Rent Reasonableness Documents in Files for New Admissions (FY2005 and
FY2006) .......................................................................................................................23
Exhibit 27 Timing of Most Recent Rent Reasonableness Determination—New Admissions
(FY2005 and FY2006).................................................................................................23
Exhibit 28 Rent Reasonableness Documents for Annual Recertifications (FY2005 and
FY2006) .......................................................................................................................24
Exhibit 29 Timing of Most Recent Rent Reasonableness Determination—Annual
Recertifications (FY2005 and FY2006) ......................................................................24
Exhibit 30 Percent of Households by Fair Market Rent Category After Comparing Payment
Standard to Fair Market Rent (FMR; FY2006) ...........................................................25
Exhibit 31 Percent of Households Meeting Payment Standard Requirements (FY2006).............25
Exhibit 32 Households Meeting Utility Allowance Requirements (FY2006) ..............................26
Exhibit 33 Negative Subsidy Households (Under-subsidies) Percent of Households and
Average Monthly Dollar Amount of Error (FY2005 and FY2006) ............................27
Exhibit 34 Positive Subsidy Households (Over-subsidies) Percent of Households and
Average Monthly Dollar Amount of Error (FY2005 and FY2006) ............................27
Exhibit 35 Average Monthly Dollar Amounts of Error for Negative (Under-) and Positive
(Over-) Subsidies Averaged Across All Households (FY2005 and FY2006).............28
Exhibit 36 Percent of Projects Using Computer Software for Administrative Tasks in the
Past 12 Months (FY2006)............................................................................................28
Exhibit 37 Percent of Projects Using Computer Software Uses in the Past 12 Months, by
Project Size (FY2006) .................................................................................................29
Exhibit 38 Average Dollars in Error by and TRACS/PIC Data (FY2006) ...................................30
Exhibit 39 Average Dollars in Error by Payment Type and TRACS/PIC Data (FY2006) ...........31
Exhibit 40 Percent of Matched and Non-Matched Dollar Amounts for Key Variables
Matching Variables from the 50058/50059 Form and TRACS/PIC Data Files
(FY2006)......................................................................................................................31
Exhibit 41 Percent of Gross Dollar Rent Errors for Cases Where Key Variables Did Not
Match (FY2006) ..........................................................................................................32
Exhibit 42 Percent of Procedural Errors for Cases Where Key Variables Did Not Match
(FY2006)......................................................................................................................32
Final Report Outline................................................................................................................... 33
Appendix A: Definitions of Key Terms
Appendix B: Source Tables Responding to Each Objective
Appendix C: National Estimates Source Tables

INTRODUCTION
The purpose of this document is to describe how analyses will be conducted for the FY2006 HUDQC Study: Quality Control For Rental Assistance Subsidy Determinations. The Department of
Housing and Urban Development (HUD) provides housing assistance through several rental
assistance programs. Subsidies are based on HUD regulations defining financial need, eligibility
requirements, and subsidy amounts. Generally, eligibility for a HUD-assisted housing unit requires
a total income equal to or below the very-low-income standard (50% of the median family income of
the area). The tenant payment is set at the higher of two amounts: 10 percent of total income, or 30
percent of adjusted income, based on certain types of deductions.
This study examines the following rent subsidy programs:
•

PIH-administered Public Housing (i.e., Public Housing)

•

PIH-administered Section 8 projects
- Moderate Rehabilitation
- Vouchers

•

Office of Housing-administered projects (i.e., owner-administered)
- Section 8 New Construction/Substantial Rehabilitation
- Section 8 Loan Management
- Section 8 Property Disposition
- Section 202 Project Rental Assistance Contracts (PRAC)
- Section 202/162 Project Assistance Contracts (PAC)
- Section 811 PRAC

The HUD QC Study focuses on the nature and extent of errors in rental assistance subsidies in the
assisted housing programs listed above. The overall purpose of the study is to determine the type,
severity, and cost of errors associated with income certification and rent calculations. This study
will produce national estimates of error in each program. A nationally representative sample of
2,400 tenants in approximately 600 projects nationwide will be selected for review and verification
of information used to determine rental assistance subsidies in their most recent (re) certification.1
In order to conduct this review and verification, we will execute the following steps:
1.

Review Tenant File. ORC Macro staff will use computer-assisted data collection
technology to review and extract information contained in each sampled tenant’s file. The
focus of the review is HUD’s forms 50058 and 50059 which are used by housing managers
to record information required for determining rental assistance eligibility and subsidy
amount; and the specific pieces of information contained in the file that are used by

1

The timing of the verification information is a key aspect of the study. This study seeks to verify information as of the
most recent (re)certification, or in the absence of a (re)certification, to verify information when the (re)certification was
due. If the (re)certification is more than one year overdue, verification will be obtained for the month the recertification
would have been effective if it had been completed on time. The fact that the study is being conducted after the
(re)certification has occurred, requires more attention to obtaining accurate reports and verifications than would be
needed if the study was done at the time of (re)certification. In order for the study to represent the population of tenants,
it is necessary to select all tenants with equal probability, even if it means that their most recent (re)certifications were
performed up to a year before.

Analysis Plan

1

July 24, 2006

management to verify the figures used in the 50058 or 50059. The 50058/9 forms also
contain the rent calculated by management.
2.

Determine Procedural Errors. Using the information in the tenant file, ORC Macro will
re-calculate the rent on the basis of verification documentation and information contained in
the file. Discrepancies between the rent recorded on the 50058/59 and this recalculation will
indicate procedural errors.

3.

Interview Tenants. Each tenant will participate in a detailed item by item interview,
capturing each element in the rent calculation. This interview will probe on all financial
resources and household circumstances, including those that may not be contained in the
tenant file. Tenants will be asked to sign releases permitting ORC Macro to obtain
verification from relevant third parties for items lacking verification documentation in the
tenant file.

4.

Conduct Enhanced Verification. Based on new or more accurate information provided by
the tenant, ORC Macro will record and then obtain verification from third parties of the
amounts reported. In addition, verification of benefits and earned income will be obtained
directly from official Federal-level sources by matching household member identifying
information (name, Social Security number, date of birth) with Social Security
Administration files and the National Directory of New Hires.

5.

Calculate QC Rent. A rent calculation will be performed on the basis of verified
information, including that contained in the original tenant file and that obtained through the
interview process and third party verification.

6.

Determine Error. Errors are defined as the difference between the rent calculation on the
50058/59 and rent determined by the QC rent calculation.

Using the data collected in the above steps and the error determinations, the data analysis will
proceed to address the study’s objectives.
THE DEPENDENT VARIABLE
The dependent variable in this study has several dimensions and definitions. At a very basic level,
an error pertains to the condition in which a tenant is receiving an incorrect amount of subsidy,
based on verified information.
Rent Used in Error Determinations. Error is determined by the difference between the rent
actually paid by the tenant and the rent that should have been paid, based on verified information
obtained by the HUD-QC study:
•
•

Actual Rent—the monthly rent indicated on the 50058/9 forms or, if this item was missing,
this information is obtained from other sources in the tenant file. This is the monthly rent for
the year to follow the most recent (re)certification.
Quality Control (QC) Rent—the monthly rent calculated by ORC Macro using the
information reported by the household and verified, if possible, as well as the verified
information contained in the tenant file.

Analysis Plan

2

July 24, 2006

Calculation of Quality Control Rent. HUD specifies the formulas for determining tenant rents for
each of its programs. These formulas generally consider adjusted annual income, which is the total
of household members’ earned and unearned income, less specific allowances. There are several
different calculation formulas, depending on the program and the specifics of each tenant’s situation.
These formulas are defined in the HUD-QC Study Standards document delivered under separate
cover.
Error Definitions. Study objectives require that several different types of errors be estimated on the
basis of data collected in this study. The two primary distinctions are total errors and error rates.
Total Errors
•

•
•

Dollar Rent Error—the dollar amount of Actual Rent minus QC Rent for an individual
tenant. A negative number indicates an underpayment, meaning the household paid less than
it should and HUD’s subsidy was higher than it should have been. A positive number
indicates a household overpayment, meaning HUD’s contribution was less than it should
have been.
Total Gross Rent Error—the weighted sum of the absolute values of positive and negative
individual tenant Rent Dollar Errors.
Total Net Rent Error—the arithmetic value of the weighted sum of individual tenant Rent
Dollar Errors.

Error Rates
•
•

Dollar Error Rate—the quotient of Total Gross Rent Error divided by the weighted sum of
individual tenant QC rents.
Case Error Rate—the quotient of the weighted sum of Dollar Rent Errors in excess of $5
per month divided by the total weighted number of tenants.

Errors in rental assistance subsidies relate to both eligibility and amount of subsidy:
•
•

Eligibility Error—a tenant may not be eligible, which places the entire subsidy in error.2
Subsidy Error—the amount of subsidy may be too high or too low.

Error sources are classified into two broad types:
•
•

Rent Component Error—any of the components used to determine rent (e.g., earned
income, household size, medical expenses) could be in error. These are often attributed to
tenant misreporting, but they can also be due to misunderstanding.
Procedural Error—local housing administrative staff may make mistakes (e.g., calculation
errors, transcription errors, improper application of income or allowances) or they may fail to
follow HUD requirements (e.g., fail to recertify on time). Some procedural errors (e.g., not
requesting a Social Security number) do not produce rent errors.

2

Eligibility is determined at the time of initial certification; therefore, eligibility errors will be assessed only for
certifications, not recertifications.

Analysis Plan

3

July 24, 2006

Errors may be made in either the determination of initial eligibility or in the determination of the
correct tenant payment. Two types of payment errors may occur:3
•
•

Overpayment—tenant payment is above the correct amount, and HUD’s subsidy is too low.
Underpayment—tenant payment is below the correct amount, and HUD’s subsidy is too
high.

Appendix A contains the definitions of all key terms used in this analysis plan.
PREPARATION OF ANALYTIC FILES
The main analytic files will be based on the results of tenant file reviews, tenant interviews, and
third party verification. While we will be using the third party verified information to determine
errors, the analytic files must also contain the information collected from the tenant files and tenant
interviews to address the study objectives pertaining to error sources and causes. The tenant file
information is needed to identify the incidence of procedural errors; the tenant interview data is
needed to determine the incidence of tenant misreporting; both files and the verification file are
needed to determine the extent that various types of resources contribute to error.
Our core master analytic file will consist of a tenant record containing:
Tenant Record Review Data—all information collected from the 50058/9, the items that are
verified and the type of verification observed; and the tenant rent.
Tenant Interview Data—all information collected during the tenant interview pertaining to items
needed to calculate rent and determine eligibility.
QC Verification Data—all information used to calculate the QC rent, consisting of verified
information obtained from third parties, information provided by the tenant, and verified information
obtained from the tenant file.
We will construct a series of analytic files to address the research questions, using the data in the
master analytic file. Error values (as defined by the methods described above) will be calculated and
appended to the main analytic file, and identify discrepancies and dollar differences between the
three sources of tenant data listed above. Additional variables will be constructed, including error
type (e.g., transcription, calculation). Weights equal to the inverse of the sampling fractions will be
appended so that national estimates can be produced. Variance estimates will be produced using a
replication or Taylor series procedure.
We will use two additional data sources. One of the study objectives is to determine whether tenant
data entered into TRACS/PIC has associated QC errors. Another objective is to determine whether
errors can be predicted from tenant and project characteristics. To obtain information on housing
project characteristics, we conduct a survey of local housing managers from which we obtain
information on characteristics of the housing project and management practices. We will create
separate analytic files to conduct the analyses associated with these two data sets. Relevant tenant
3

It is possible that the result may produce no error in rent payment or subsidy amount. Some errors may “cancel” others
out, or the individual items may not be of sufficient magnitude to have an effect on rents or subsidies.

Analysis Plan

4

July 24, 2006

information will be appended to the project survey file. The study sample will be matched with
TRACS/PIC, and the 50058/9 data from TRACS/PIC will be appended to the tenant data for
analysis.
ANALYSIS PLAN BY STUDY OBJECTIVE
This section of the Analysis Plan discusses the study objectives and describes the analysis that will
address each objective. Appendix B contains a summary of the objectives and the source tables that
address each objective. Appendix C contains shells for the source tables. Source tables will be used
to produce the analytic exhibits displayed in the body of the report. We describe specific analytic
exhibits and provide shells for these in the discussion below.
Objective 1:

Identify the various types of rent errors and rent error rates and their variance
estimates.

This objective requires us to identify types of errors and produce national estimates of the proportion
of tenant cases with errors, along with associated variance estimates. These errors include the
percent of households paying correct and incorrect rent, average dollar rent error, and dollar error
rate. Analyses will cross-tabulate national estimates to produce a series of tables as described
below. To assure comparability with prior studies, the tabular displays will follow the previously
used formats and will include FY2005 study results alongside the FY2006 study results. Variance
estimates are displayed in tables discussed under Objective 3.
Exhibit 1 illustrates how we will display the percent of households with proper payments. It
provides the national estimate of the proportion of tenants whose QC rent is exactly equal to the
Actual Rent, and the proportion within $5 of an exact match. This exhibit also provides a
comparison between FY2005 and FY2006 results, and a comparison of results for by program type.
Exhibit 1
Percent of Households with Proper Payments (FY2005 and FY2006)
Percent Matched Within $5

Percent Matched Exactly

Program Type
FY2005

FY2006

FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Tables 1a and 1b

Exhibit 2 provides further information on the tenant error rate, displaying the average dollars in error
and gross dollar error rate for the total population in PHA-administered and owner-administered
projects. It compares the FY2005 results with the FY2006 results. Gross dollars in error for
FY2004 will be calculated using actual FY2004 dollars.

Analysis Plan

5

July 24, 2006

Exhibit 2
Rent Error: Percent of Households in Error, Average Gross Dollars in Error, and Error Rate
(FY2005 and FY2006)

Program Type

Percent of Households in
Error
FY2005

FY2006

Average Gross Dollars in
Error
FY2005

FY2006

Gross Dollar Error Rate
FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Tables 1a and 2

Exhibits 3a and 3b display the dollar amount of error associated with tenant over- and underpayments. Exhibit 3a displays the percent of households paying less than the proper amount and the
average dollar underpayment error. Exhibit 3b displays the same information for households paying
more than the proper amount.
Exhibit 3a
Underpayment Households: Percent of Households and Average Monthly Dollar Amount of Error
(FY2005 and FY2006)

Program Type

Percent of Households with
Underpayment
FY2005

FY2006

Average Dollar Error for Households
with Underpayment
FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Tables 1b and 3

Analysis Plan

6

July 24, 2006

Exhibit 3b
Overpayment Households: Percent of Households and Average Monthly Dollar Amount of Error
(FY2005 and FY2006)
Percent of Households with
Overpayment

Program Type

FY2005

FY2006

Average Dollar Error for Households
with Overpayment
FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Tables 1b and 3

Objective 2:

Identify the dollar costs of various types of error.

Previous QC studies identified several types of error that can be detected using information in the
tenant file. These errors are identified using data obtained from the 50058/59 directly as it appears
on the 50058/59 form, and other information from files used to determine which information should
be recorded on the 50058/59. Procedural errors are detectable through the analysis of the tenant file
data, and may or may not result in rent errors. This analysis will not use QC rent error as a standard
because the QC rent will be based on information obtained during the tenant interview as well as
verification obtained from third parties.
Calculation errors are detected by recalculating section subtotals and final rent based on the exact
information in the 50058/59 forms. The rent will be calculated using the detailed information on the
50058/59 and compared to the tenant rent on the 50058/59. If the two rents differ, this indicates a
calculation error.
Consistency errors are identified by assessing the logical conformity between elements within the
50058 or 50059 forms. For example, assistance status information must correspond (e.g., elderly
status) must be consistent with other information (e.g., age of the head of household or spouse).
Transcription errors are detected by comparing 50058/59 data with information obtained from the
tenant file. As the data are entered, the automated data entry system compares the subtotals in each
section of the 50058/59 forms. If the 50058/59 data do not match the tenant file data, the system
alerts data collectors to a possible transcription error. The data collectors will then that confirm the
data were entered correctly.
The improper application of allowances and incorrect calculation of income are a subset of
transcription errors. Failure to apply allowances correctly and identify income correctly will be
identified by comparing tenant file information to 50058/59 data. Allowance errors will be detected
by calculating the allowances based on the tenant file and comparing this QC allowance to the
Actual Allowance on the 50058/59. Similarly, income will be calculated based on the types and
amounts of income reported in the tenant file.

Analysis Plan

7

July 24, 2006

A series of exhibits will display errors detected in tenant file data. Exhibit 4 presents the percent of
households with calculation and consistency errors in different sections of the 50058 and 50059
forms. More detailed data will be presented in Source Tables 4 (calculation errors) and 5
(consistency errors). Note that the 50058 form is formatted differently and in some sections
provides more line items of information than the 50059 form. Consequently, the number and types
of calculation and consistency errors on the forms will be different, and the findings from the two
forms will not be comparable.
Exhibit 4
Percent of Households with Calculation and Consistency Errors (FY2006)
Percent of Households
50058/50059 Item

Calculation Errors
50058

General Information

n/a

50059
n/a

Consistency Errors
Total

50058

50059

Total

n/a

Household Composition
Net Family Assets and Income
Allowances and Adjusted Income
Family Rent and Subsidy Information
Total
Source Tables 4 and 5

Overdue Recertifications also produce errors in rents because rents are calculated using old
information. We will calculate the error amount due to overdue recertification, based on the
difference between Actual and QC Rent. Exhibit 5 will display the percent of cases with overdue
recertifications, timely recertifications, and new certifications. This exhibit will provide this
information by program type.
Exhibit 5
Timeliness of Certification Status (FY2005 and FY2006)
New Certifications

Timely Recertifications

Overdue Recertifications

Rent Component
FY2005

FY2006

FY2005

FY2006

FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Table 7

Analysis Plan

8

July 24, 2006

Our analysis will also replicate the graphics produced in the FY2005 study, illustrated by Figure 1,
which shows the proportion of cases that are new certifications, timely recertifications, and overdue
recertifications.
Figure 1: Case Type (FY2006)
Overdue
Recertifications,
7%

New Certifications,
12%

Timely
Recertifications,
81%

Exhibit 6 illustrates the error associated with overdue recertifications compared to errors from
certification and timely recertifications. In cases with overdue recertifications, the information at the
time the recertification was due will be used to determine rent, as it was in the previous studies.
Exhibit 6
Average Monthly Underpayment and Overpayment Dollar Amount
Averaged across All Households
(FY2005 and FY2006)

Action Type

Underpayment
Average Dollar Amount
FY2005

FY2006

Overpayment Average
Dollar Amount
FY2005

FY2006

New Certification
Timely Recertification
Overdue Recertification
Total
Source Table 6

Analysis Plan

9

July 24, 2006

As in FY2005, we will conduct additional analyses to summarize the information that addresses this
objective. Exhibit 7 provides the proportion of cases with procedural error, the estimated average
cost of each type of error, and the standard error of the estimated average (i.e., the variance estimate
of the average). That cost will be the difference between the actual rent and the recalculated rent,
using the tenant file information and correcting transcription and calculation errors.
Exhibit 7
Procedural Error: Percent of Households, Average Dollars in Error, All Households with 50058/50059
Recalculated Rent (FY2006)
Gross Rent Error
Error Type

Percent of
Households in
Error

Average
Dollars in
Error

Net Rent Error
Standard
Error of
Mean

Percent of
Households in
Error

Average
Dollars in
Error

Standard
Error of
Mean

Transcription Error
Calculation Error – Allowances
Calculation Error – Income
Calculation Error – Other
Overdue Recertifications
Any Procedural Error
All Households with Procedural
Errors
Source Table 8

Exhibit 8 provides a summary of the errors identified from the 50058/50059 forms. These data are
produced using cross-tabulations and show the error rates and costs for households with each type of
procedural error, for households without procedural errors, and for the total weighted sample. The
exhibit presents the percent of households in error, the average dollar error, and the standard errors
for both households with recalculated 50058/50059 error (error determined using only the
50058/50059 form), and households with QC Rent error.

Analysis Plan

10

July 24, 2006

Exhibit 8
50058/50059 Procedural Error: Percent of Households, Average Dollars in Error (FY2006)
Households with Recalculated 50058/9 Error
Error Type Based on 50058/59 Recalculation

Percent of
Households in
Error

Standard
Error of
Percent

Average
Dollar
Error

Standard
Error of
Mean

Households with QC Rent Error
Percent of
Households in
Error

Standard
Error of
Percent

Average
Dollar
Error

Standard
Error of
Mean

Households with Consistency Error
Households without Consistency Error
Households with Allowance Calculation Error
Households without Allowance Calculation Error
Households with Income Calculation Error
Households without Income Calculation Error
Households with Other Calculation Error
Households without Other Calculation Error
Overdue Recertifications
On-time Recertifications
Certifications
Unduplicated Count, Any Type of 50058/50059 Error
Unduplicated Count, No 50058/50059 Error
Total
Source Tables 4 - 8

Analysis Plan

11

July 24, 2006

Incorrectly applied flat rent schedule will be identified by obtaining flat rent schedules from
PHAs and comparing them to the actual rent amount recorded on the 50058. This examination only
applies to public housing program tenants. HUD policy requires that “for families who choose flat
rents, the PHA must conduct a reexamination of family composition at least annually, and must
conduct a reexamination of family income at least once every three years.” [24 CFR 960.257 (a)(2)].
Therefore, multiple flat rent schedules to cover the three year period prior to the data collection
effort must be obtained and documented as to when they became effective. The correct flat rent for
a particular case will vary depending on when the last annual recertification was conducted. The
examination of flat rents and schedules will be accompanied by a discussion of the issues identified
during the analysis.
Objective 3:

Estimate national-level net costs for total errors and major error types.

This analysis will replicate the cross-tabulations developed in the previous studies that address error
dollars. Results from FY2005 and FY2006 will be presented for comparison. The gross rent error is
obtained by adding together the absolute values of the dollar amount of overpayments to the dollar
amount of underpayments. The net cost for total errors is an arithmetic calculation of the sum of
positive and negative nationally weighted error costs. This sum represents the net amount of tenant
payments in error and will be displayed by program type. Exhibit 9 provides this information with
its associated standard error.
Exhibit 9
Gross and Net Dollar Rent Error (Monthly) for All Households
(FY2005 and FY2006)
Gross Rent Error
Program Type

Average Dollars in Error
FY2005

FY2006

Net Rent Error
Standard
Error
FY2006

Average Dollars in Error
FY2005

FY2006

Standard
Error
FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Owner-administered
Total
Source Table 9

Objective 4:

Determine the relationship between errors detectable by using the HUD-50058 and
HUD-50059 forms and total errors found in the study.

Objective 2 estimates procedural error that can be attributed to mistakes made by the housing
manager. Except for overdue recertifications, it does not estimate QC error detected through the
verification process. The purpose of Objective 4 is to determine the relationship between those
procedural errors detected from the 50058/59 forms and the total error found after all information
was verified in the study. Exhibit 10 illustrates this analysis.

Analysis Plan

12

July 24, 2006

Exhibit 10
50058/59 Rent Calculation Error Compared to QC Rent Error
(FY2005 and FY2006)
Percent of Households with
Correctly Calculated Rent

Rent Calculation Method

FY2005

FY2006

Percent of Households with
Incorrectly Calculated Rent
FY2005

FY2006

Using Information on the 50058/50059 Form
According to the QC Rent Calculation
Both 50058/50059 calculation and QC Rent calculation
Source Table 1

Since HUD collects 50058/9 forms centrally on the TRACS/PIC System, it may beneficial for the
agency to re-calculate information on the 50058/9 forms to help identify cases likely to be in error.
This decision could be made on the basis of the results of the descriptive analysis, or HUD may
choose to use more sophisticated techniques. Additional discussion of the use of PIC and TRACS
data to predict error is found under Objective 14.
Exhibit 11 presents the percent of households in error and the total annual program dollar errors,
comparing error obtained from tenant file information alone to error obtained from tenant file
information plus household interview information.
Exhibit 11
Percent of Households in Error and Dollar Error by Error Basis
(FY2005 and FY2006)
Percent of Households in Error

Total Annual Dollar Errors

Error Basis
FY2005

FY2006

FY2005

FY2006

Error based on tenant file and interview
information
Error based on tenant file information only
Source Table 3

Objective 5:

Determine whether error rates and error costs have statistically significant
differences from program to program.

Regression analysis will be used to determine whether differences in error rates by programs are
statistically significant. Regression models will estimate the impact of program type on gross and
net dollar errors as dependent variables. The gross error rate is the sum dollar amount of gross error
divided by the sum dollar amount of QC Rent, and the net error rate, which is the sum dollar amount
of net error divided again by the sum dollar amount of QC Rent. The models will include two
dummy variables for program type: public housing project versus PHA-administered Section 8 and
owner-administered projects, and PHA-administered Section 8 projects versus public housing and
owner-administered projects. Generally, if a variable has m categories, the model must contain m-1
dummy variables to represent the categorical variable. The omitted category, in this case the owneradministered projects, is referred to as the reference group because the model’s comparisons are
Analysis Plan

13

July 24, 2006

made against that program type. The estimated coefficients can be interpreted as the dollar amount
of error for that program type, on average. For the omitted category (i.e., owner-administered
projects), the model intercept represents the average dollar amount of error. Using the estimated
coefficients, we can then calculate the average gross and net error rates for each program type. Ttests will indicate whether program differences in error are statistically significant. Exhibit 12
illustrates how these results might be displayed.
Exhibit 12
The Impact of Program Type on Gross and Net Dollar Error (FY2006)
Program Type

Average Gross Error

Average Net Error

Public Housing
PHA-administered Section 8
Owner-administered

Objective 6:

Determine the apparent cause of significant rent errors, either on a sample or a
comprehensive basis, to provide HUD with information on whether the error was
caused primarily by the tenant or by program sponsor staff.

Understanding the sources and causes of significant rent errors is important for determining
corrective actions. First, it is necessary to have an understanding of which items in the rent
calculation formula contribute most to error. Second, it is important to understand whether this error
is caused by the tenant or by the project’s administrative staff. However, it is sometimes difficult to
disentangle the source and cause of errors. Transcription and calculation errors, improper
application of allowances, and failure to recertify on time are procedural errors. These are clear
responsibilities of the project’s management and administrative staff. The cause of discrepancies
between the information used to calculate rent by the project and that obtained through the QC
verification process is not always clear. Tenants may have failed to report an income item because
they intentionally withheld the information to pay less rent; they may not have been asked to report
an item during the interview; or they may have misunderstood the requirement. For that reason, we
prefer not to ascribe to the tenant all errors attributed to discrepancies between information in the
project files and the QC verification process. It may often be the case that the error is due to the
tenant, but this study will not be able to make that determination. Therefore, we consider
discrepancies between information used to determine rent and verified information as sources of
error, rather than ascribing cause to tenants or project staff.
For the purposes of analysis and corrective action, it is useful to learn which elements in the rent
computation formula contribute to QC errors. Even if we don’t know why items such as income or
medical expenses were inaccurate, HUD will know that these items should be given more careful
attention by local project staff when they obtain information from tenants and/or verify information
from third parties.
We propose two levels of analysis to address this issue. First, we will provide descriptive
information on the sources of discrepancies between housing file information and verified
information, and describe the incidence of procedural errors and their impacts. Exhibit 7, already
presented in our discussion of objective 4, describes the proportion of cases with procedural errors
(i.e., calculation, transcription, improper application of allowances, improper calculation of income,
and overdue certification), and their corresponding QC rent error. It shows the relationship between
Analysis Plan

14

July 24, 2006

these procedural errors and QC errors. Second, we will produce exhibits that illustrate another type
of procedural error—failure to verify information or inappropriate application of verification
information, as shown in Exhibits 14, 15, and 16 below. Source Table 11 will provide the data for
these exhibits.
Exhibit 13 presents the number of households by rent component where verification was not
obtained, where it was obtained but the verification amount did not match the amount used on the
50058/59, and where verification was obtained and it did match the amount used on the 50058/59.
Exhibit 13
Verification of 50058/50059 Rent Components
(FY2005 and FY2006)

Rent Component

No ProjectVerification
FY2005

Item Verified by Project

FY2006

FY2005

FY2006

Verification Matched
50058/59
FY2005

FY2006

Earned Income
Pensions
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly Allowance
Child Care Allowance
Handicapped Allowance
Medical Allowance
Source Table 11

Exhibit 14 provides case file verification information by program type. It displays the percent of
households where the rent component was verified, as well as the percent of the verification for each
rent component found in the tenant file that matched the data on the 50058/50059 form within $100.

Analysis Plan

15

July 24, 2006

Exhibit 14
Verification of 50058/50059 Rent Components (FY2006)
PHA-administered Section 8
Rent Component

Verified

Matched*

Owner-administered
Verified

Public Housing

Matched*

Verified

Matched*

Earned Income
Pensions
Public Assistance
Other Income
Asset Income
Child Care Expense
Disability Expense
Medical Expense

* Matched within $100
Source Table 13

Exhibit 15 takes the analysis a step further. It provides data on whether failure to verify sources of
income and expenses was a contributor to QC error. It displays the percent of households with QC
error for which verification was missing in the tenant file. Each error is presented by rent
component.
Exhibit 15
QC Error Households with Missing Verification in the Tenant File
(FY2005 and FY2006)
50058
Rent Component

Households with
QC Error
FY2005 FY2006

50059

Households with QC Errors Households with QC
and Missing Verification
Error
FY2005

FY2006

FY2005

FY2006

Households with QC Errors
and Missing Verification
FY2005

FY2006

Earned Income
Pensions
Public Assistance
Other Income
Asset Income
Child Care Expense
Handicapped Expense
Medical Expense
No Component Error
Source Table 11

Analysis Plan

16

July 24, 2006

Rent components—the elements used to calculate rent—are another source of error, so we will
conduct analyses of rent component error. Exhibit 16 shows the relationship between errors in each
rent component and the average dollar amount for cases in error.
Exhibit 17 compares the percent of total households with and without component error by
component type and payment type. It also compares this information for households in different
housing program types.
Exhibit 18 presents the standard errors for the total number of households with and without
component error by component type and payment type.
Exhibit 16
Rent Components Responsible for the Largest Dollar Error
Households with Rent Error (listed by amount of dollar error)
(FY2005 and FY2006)

Rent Component

% of Households
in Error
FY2005

FY2006

Average Dollar Amount
FY2005

FY2006

Earned Income
Other Income
Pensions
Asset Income
Public Assistance
Child Care Expenses
Medical Expenses
Dependent Allowance
Disability Expenses
Elderly/Disabled Allowance
No Rent Component Error
Source Table 12

Analysis Plan

17

July 24, 2006

Exhibit 17
Income and Expense Component Error by Payment Type for All Households (FY2006)
Underpayment

Proper Payment

Overpayment

Income/Expense Component
PHA

Owner

Total

PHA

Owner

Total

PHA

Owner

Total

Earned Income
Pension Income
Public Assistance Income
Other Income
Asset Income
Dependent Allowance
Elderly Household Allowance
Child Care Allowance
Disability Assistance Expense
Medical Expense
No Rent Component Error
Source Table 13

Exhibit 18
Percent of Households and Standard Error by Rent Component and Payment Type (FY2006)
Underpayment
Component

Percent of
Total
Households

Proper Payment

Standard
Error

Percent of
Total
Households

Standard
Error

Overpayment
Percent of
Total
Households

Standard
Error

Earned Income
Pension Income
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Expenses
Disability Expenses
Medical Expenses
No Rent Component Error
Source Table 14

Analysis Plan

18

July 24, 2006

Exhibit 19 will explore whether elderly/disabled and dependent allowances4 are applied correctly.
Exhibit 19
Elderly/Disabled Allowances and Dependent Allowances (FY2006)
Elderly Allowance
Non-Elderly/
Non-Disabled
Households

Elderly/
Disabled
Households

Dependent Allowance

All
Households

Households
With
Dependents

Households
Without
Dependents

All
Households

No Allowance
Incorrect Allowance
Correct Allowance
Source Table 15

Additional multivariate analyses will explore associations between procedural errors and income
components and QC Dollar Rent Error.
Objective 7:

Determine the extent to which households are overhoused relative to HUD’s
occupancy standards.

This objective addresses whether households reside in units with the correct number of bedrooms.
Generally acceptable standards5 specifying the appropriate size unit for PHA Section 8 households
are shown in Exhibit 20 below.
Exhibit 20
PHA Section 8 Unit Size Standards
Number of Persons in Household
Number of Bedrooms
Minimum

Maximum

0

1

1

1

1

2

2

2

4

3

3

6

4

5

8

5

7

10

4

Households with an elderly or disabled head or spouse are entitled to one $400 allowance (i.e., deduction from gross
annual income) in calculating rent. Households are entitled to a $480 allowance for each dependent (defined as children
under 18, full-time students, and disabled members other than the head or spouse).
5
Local projects have discretion in determining unit size, and may determine unit size differently than shown.
Analysis Plan

19

July 24, 2006

There are exceptions to these guidelines. If a tenant is elderly, disabled, pregnant, or meets other
criteria, they may be allowed a larger bedroom unit. There are also circumstances when households
are allowed smaller bedroom units. The determination of appropriate bedroom size is locally based.
For this study it will be based on the Standards, delivered under separate cover, which specify rules
for bedroom size.
Overhousing refers to tenants occupying units that exceed the bedroom size allowed by HUD
regulation for their actual household size. This study will replicate the analysis completed in
previous studies, identifying by bedroom size and program, the proportion of households in
compliance with and in violation of occupancy standards. This analysis will be conducted with
national estimates of proportions in tabular displays showing the results for FY2005 and FY2006.
Exhibit 21 presents the percent of households in units with the correct number of households by
program type with information for both the FY2005 and FY2006 study. Exhibit 22 presents the
overall findings. The shaded cells generally indicate incorrect unit assignments.
Exhibit 21
Percent of Households in Units with Correct Number of Bedrooms
(According to Study Guidelines)
(FY2005 and FY2006)
PHA-administered
Number of
Bedrooms

Public Housing
FY2005

Owner Administered

Section 8

FY2006

FY2005

FY2006

FY2005

Total

FY2006

FY2005

FY2006

0
1
2
3
4
5
All Units
Source Table 16

Exhibit 22
Percent of All Households by
Number of Bedrooms and Number of Household Members (in thousands) (FY2006)
Number of
Bedrooms

Number of Household Members
1

2

3

4

5

6

7

8

9

10

0
1
2
3
4
5
Source Table 16

Analysis Plan

20

July 24, 2006

Objective 8:

Provide information on the extent to which errors are concentrated in projects and
programs.

We will determine the degree to which errors are concentrated in certain projects, as opposed to
randomly distributed across the sample. On the one hand, if most errors are caused by the project
staff, we would expect to find errors clustered in certain projects. On the other hand, if errors are
mostly caused by the tenant, we would expect to find errors randomly distributed throughout the
sample.
Using information obtained from the Project Staff Questionnaire, we will conduct multivariate
analyses to explore the association between project characteristics (e.g., program type, SMSA/non
SMSA location, percent of elderly tenants, management practices) and error rates. This analysis will
identify how each of these variables contributes to differences in error. The results will provide
HUD with information to guide the management of error rates, and will elaborate relationships
between management practices and project/tenant characteristics associated with error rates.
Objective 9:

Identify the percentage of newly certified tenants who were incorrectly determined
eligible for program admission.

Incorrect initial eligibility determinations create long-term problems for assisted-housing programs.
It is key to prudent housing management practices to correctly determine initial eligibility criteria.
Eligibility for housing assistance is based on five certification criteria: family composition,
citizenship, verification of Social Security numbers, signed consent forms, and low and very low
income limits. This study will examine eligibility criteria and verify the accuracy of collected
information. We will examine citizenship, Social Security number, consent form and low income
criteria, and present results as shown in Exhibit 23, and by program type, as in Exhibit 24.
Exhibit 23
Percent of Newly Certified Households Meeting Certification Criteria (FY2006)
Percent of Households
Certification Criteria
Met Criterion

Did Not Meet
Criterion

Unable to Determine

Citizenship
Social Security Number
Consent Form
Low and Very Low Income
Meets All Eligibility Criteria
Source Table 17

Analysis Plan

21

July 24, 2006

Exhibit 24
Percent of Newly Certified Households Meeting Certification Criteria (FY2006)
Percent of Households Meeting the Criteria
Certification Criteria

PHA-administered
Section 8

Public Housing

Owner-administered Section
8

Citizenship
Social Security Number
Consent Form
Low and Very Low Income
Meets All Eligibility Criteria
Source Table 18

Objective 10: Determine the extent to which Section 8 voucher rent comparability
determinations are found in the tenant file, and indicate the method used to
support the determination. Determine whether voucher payment standards are
within 90 – 110 percent of fair market rents, and determine whether the correct
utility allowances are being applied.
Objective 10 examines several issues related to the Section 8 Voucher program that have important
but indirect influences on rent errors.
Rent Reasonableness Analysis
To comply with the rent reasonableness requirement, housing authorities must determine that
Section 8 voucher rents are reasonable in comparison to rents for similar housing in the private,
unassisted market. We will determine, based on information obtained from PHAs, their usual
method for assessing rent reasonableness. Exhibit 25 illustrates these results.
Exhibit 25
Rent Reasonableness Determination Methods (FY2006)
PHAs Using Method
Method for Assessing Rent Reasonableness
Number

Percent

Unit-to-Unit Comparison
Unit-to-Market Comparison
Point System
Professional Judgment
Other or Rent Control
No Information Provided
Total

Analysis Plan

22

July 24, 2006

Using information collected from tenant files, we will estimate the proportion of new admission
Section 8 voucher recipients with rent reasonableness documentation. We will also determine the
timing of their most recent determination, and compare this to the results from FY2005. Exhibits 26
and 27 illustrate these results.
Exhibit 26
Rent Reasonableness Documents in Files for New Admissions
(FY2005 and FY2006)
FY2005
Status

Units in
1000s

Percent

FY2006
Units in
1000s

Percent

Determination documented
A signed statement certifying that the rent is reasonable
Comparable units documented by the property owner in
section 12a of HUD 52517
Comparable units documented on other documents
Any other reference to rent reasonableness
Missing reference
No determination documented
Total

Exhibit 27
Timing of Most Recent Rent Reasonableness Determination—New Admissions
(FY2005 and FY2006)
FY2005
Determination-Certification Chronology

Units in
1000s

Percent

FY2006
Units in
1000s

Percent

More than 4 months before lease date
Up to 4 months before lease date
After lease date—up to 2 months
After lease date—greater than 2 months
Date missing
Total

Annual recertifications require rent reasonableness documents only when owners increased rental
rates. We will examine case files to determine when the current rent first became effective, and
whether rent reasonableness documentation is present in the files. This analysis is displayed in
Exhibit 28. We will also compare timing of determinations from FY2004 and FY2005, as Exhibit
29 illustrates.

Analysis Plan

23

July 24, 2006

Exhibit 28
Rent Reasonableness Documents for Annual Recertifications
(FY2005 and FY2006)
FY2005
Status

Units in
1000s

Percent

FY2006
Units in
1000s

Percent

Determination documented
A signed statement certifying that the rent is reasonable
Comparable units documented by the property owner in
section 12a of HUD 52517
Comparable units documented on other documents
Any other reference to rent reasonableness
Missing reference
No determination documented
Total

Exhibit 29
Timing of Most Recent Rent Reasonableness Determination—Annual Recertifications
(FY2005 and FY2006)
FY2005

FY2006

Determination-Certification Chronology
Units in
1000s

Percent

Units in 1000s

Percent

More than 4 months before lease date
Up to 4 months before lease date
After lease date—up to 2 months
After lease date—greater than 2 months
Date missing
Total

Payment Standards Analysis
HUD will supply the published Fair Market Rents (FMR) to ORC Macro. This information will be
compared to payment standard data from the 50058 form, which will be captured during the data
collection process. Household rents outside of the 90-110 percent band of the FMR will be
appropriately flagged and this information will be sent to HUD. The comparison of FMRs and
payment standard data will result in a table that summarizes the number and percent of households
below, in, and above the 90-110 percent band. Exhibit 30 displays this.

Analysis Plan

24

July 24, 2006

Exhibit 30
Percent of Households by Fair Market Rent Category
After Comparing Payment Standard to Fair Market Rent (FMR; FY2006)
Percent of Households
Fair Market Rent Category
Under 90% FMR

90–110 % FMR

Over 110% FMR

Less than $500
$500–$599
$600–$799
$800–$999
$1,000–$1,199
$1,FY200–or Higher
All Voucher Households

For households that fall outside the 90-110 band, we will determine whether they received an
exemption. Exhibit 31 illustrates this analysis.
Exhibit 31
Percent of Households Meeting Payment Standard Requirements (FY2006)
Percent of Households

Under 90%
FMR

90–110 %
FMR

Over
110%
FMR

Total Percent
Outside the
90–110%
Band

Payment Standard Compared with Fair Market Rent
Households Granted an Exemption
Households (without exemptions) with Elderly or Disabled Members
Households Not Meeting Requirements

Utility Schedules
Voucher utility allowances will be evaluated in by comparing the utility allowance amount recorded
in the tenant file utility worksheet to the utility allowance recorded on the 50058/9 form, and to the
amount calculated using the PHA utility allowance schedule. Macro will obtain utility schedules in
use by the PHAs and the utility allowance worksheet from the tenant file. We will compare the total
utility allowance amount, the number of bedrooms, and the address. Exhibit 32 illustrates this
analysis.

Analysis Plan

25

July 24, 2006

Exhibit 32
Households Meeting Utility Allowance Requirements (FY2006)
Outcome

Number

Percent

QC Utility Allowance Compared to 50058
Utility Allowance
Matched
Did not match
Possible reasons for discrepancy:
Number of bedrooms did not match
Address did not match
Type of unit did not match
Dates did not match
QC Utility Allowance Compared to Project
Utility Allowance Schedule
Matched
Did not match
Possible reasons for mismatch:
Number of bedrooms did not match
Type of unit did not match
Dates did not match

Objective 11: Estimate total positive and negative errors in terms of HUD subsidies.
The actual cost of errors to HUD is expressed in terms of subsidy payments. HUD subsidies for
assisted housing programs equal the allowed expense level or payment standard minus the tenant
rent. In the previous study, proper payments were defined as those in which the Actual Rent equals
the QC Rent (i.e., there is no dollar error in the tenant payment). Errors can be either overpayments
(Actual Rent greater than QC Rent) or tenant underpayments (Actual Rent less than QC Rent).
Overpayment error rates are computed by dividing the total amount of overpayment by the total
Actual Rent; underpayment error rates are calculated by dividing the total amount of underpayments
by the total Actual Rent. Tenant overpayments are negative subsidy errors; tenant underpayments
are positive subsidy errors. Tables as shown in Exhibits 33, 34 and 35 below will illustrate the
results of these comparisons.

Analysis Plan

26

July 24, 2006

Exhibit 33
Negative Subsidy Households (Under-subsidies)
Percent of Households and Average Monthly Dollar Amount of Error
(FY2005 and FY2006)
Average Dollar Amount of Error
Negative Subsidy
Households (with errors >
$5)

Percent of Households
in Error

Program Type

FY2005

FY2006

FY2005

FY2006

All Households
FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Tables 1b and 3

Exhibit 34
Positive Subsidy Households (Over-subsidies)
Percent of Households and Average Monthly Dollar Amount of Error
(FY2005 and FY2006)
Average Dollar Amount of Error

Administration Type

Positive Subsidy
Households
(with errors > $5)

Percent of Households
in Error
FY2005

FY2006

FY2005

FY2006

All Households
FY2005

FY2006

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Tables 1b and 3

Analysis Plan

27

July 24, 2006

Exhibit 35
Average Monthly Dollar Amounts of Error for Negative (Under-) and Positive (Over-) Subsidies
Averaged Across All Households
(FY2005 and FY2006)

Household Type

Negative Subsidy Average Dollar
Amount of Error
FY2005

FY2006

Positive Subsidy Average Dollar
Amount of Error
FY2005

FY2006

Certifications
Non-overdue Recertifications
Overdue Recertifications
Total
Source Tables 1b and 3

Objective 12: Determine the extent to which error rates in projects that use an automated rent
calculation system differ from errors in those that do not.
Rent calculation using an automated system could eliminate calculation errors. Automated systems
may also facilitate accurate collection and storage of tenant information. In the FY2005 study, we
found that between 84 and 96 percent of projects used computers for various administrative
processes. For the FY2006 study, we will determine (via the PSQ) the tasks for which PHAs and
projects use computers (e.g., tracking certifications; conducting accounting tasks; submitting tenant
information to HUD). Exhibit 36 displays this analysis.
Exhibit 36
Percent of Projects Using Computer Software for Administrative Tasks in the Past 12 Months (FY2006)
Percent Using Computer Software
Administrative Tasks

Public Housing
Projects

PHA-Administered Owner-Administered
All Projects
Section 8 Projects
Projects

Interview tenants and record answers
Keep track of pending verifications
Input verified information
Calculate rent
Print the 50058/50059 form
Conduct accounting tasks
Track maintenance activities
Print letters to the tenants
Assign recertification dates/appointments
Print checks
Submit tenant information to HUD
Conduct rent reasonableness comparisons
Maintain demographics on the population
Keep other types of statistics
DO NOT USE COMPUTERS
TOTAL NUMBER OF PHA/PROJECTS

Analysis Plan

28

July 24, 2006

We will also examine use of computers by project size, as illustrated by Exhibit 37.
Exhibit 37
Percent of Projects Using Computer Software Uses in the Past 12 Months, by Project Size (FY2006)
Percent Using Computer Software
Administrative Tasks

Projects With <150
Units

Projects With 150 To Projects With >500
500 Units
Units

Interview tenants and record answers
Keep track of pending verifications
Input verified information
Calculate rent
Print the 50058/50059 form
Conduct accounting tasks
Track maintenance activities
Print letters to the tenants
Assign recertification dates/appointments
Print checks
Submit tenant information to HUD
Conduct rent reasonableness comparisons
Maintain demographics on the population
Keep other types of statistics
DO NOT USE COMPUTERS
TOTAL NUMBER OF PHA/PROJECTS

Objective 13: Determine whether other tenant or project characteristics on which data are
available are correlated with higher or low error rates.
Many Federal and state agencies use error-prone modeling techniques to identify cases with a high
probability of being in error. These techniques are often used in welfare, Medicaid, student aid, food
assistance, and tax compliance programs. A variety of tools have been used, including regression
analysis, sequential search techniques, discriminant analysis, correlation and regression trees
(CART), and other statistical methods, depending on the nature of the available data. Ideally, these
methods are used to develop equations that predict the likelihood a case is in error or an
administrative unit is making errors.
Error prone models provide a cost-effective means to target quality control monitoring efforts by
identifying specific types of tenants and projects likely to exhibit high error rates. We will use
multivariate regression techniques, path analysis, and classification and regression trees (CART) to
develop error-prone models. The dependent variable in these analyses will be rent errors. Project
characteristics (e.g. PHA/project size; staff training methods) and tenant characteristics (e.g., number
of sources of income; type of expenses) will be used as independent variable. Where possible, we
will incorporate data from TRACS/PIC into the models to provide HUD with more information for
identifying projects and tenants likely to exhibit high error rates. Although the explanatory findings
of error-prone models are important, we believe that such models will be most useful to HUD if its
analysts can combine the findings from program data (e.g., TRACS/PIC) to target projects and

Analysis Plan

29

July 24, 2006

tenants likely to exhibit high error rates. In this proposed study, our error-prone modeling efforts
will focus on producing practical tools that HUD analysts can use in ongoing quality control efforts.
Objective 14: Determine whether cases for which 50058/9 data had been submitted to HUD were
more or less likely to have errors than those for which data had not been
submitted.
A national database of tenant 50058/9s is maintained by HUD on the TRACS/PIC system.
However, not all tenants are on the system. There are concerns about projects that fail to routinely
transmit information to TRACS/PIC, and it is hypothesized that a reason for this failure is that
(re)certifications are not performed on a timely basis. The existence of TRACS/PIC in concert with
the QC study provides the opportunity to investigate the relationship between TRACS/PIC reporting
and rent accuracy.
In the FY2003 study, we found that 97 percent of study households were included in the
TRACS/PIC databases. The total cost of errors was higher for the households not matched to
TRACS/PIC. In the FY2004 study, TRACS/PIC matching was attempted but match results were not
available.
If HUD is able to provide the data necessary to conduct the analysis, we will replicate the FY2003
analysis in the present study. We will match the sampled tenants with the TRACS and PIC
databases and then produce error estimates for matched and nonmatched cases.
The analysis will compare QC error rates for sampled tenants who appear on TRACS/PIC with those
who do not. Any difference that is greater than sampling error would be considered significant. The
results will be presented, as shown in Exhibits 38 and 39, for program type and payment type. The
total population will be used to determine the average dollars in error.
Exhibit 38
Average Dollars in Error by and TRACS/PIC Data (FY2006)
TRACS/PIC PRESENT
Administration Type

Percent of All
Households in
Error

Average Dollars in
Error

TRACS/PIC ABSENT
Percent of All
Households in
Error

Average Dollars in
Error

Public Housing
PHA-administered Section 8
Total PHA-administered
Total Owner-administered
Total
Source Table 19

Analysis Plan

30

July 24, 2006

Exhibit 39
Average Dollars in Error by Payment Type and TRACS/PIC Data (FY2006)
TRACS/PIC PRESENT
Payment Type

Percent of
Households in
TRACS/PIC

TRACS/PIC ABSENT

Average Dollars in
Error

Percent of Households
Not in TRACS/PIC

Average Dollars in
Error

Overpayment
Underpayment
Proper Payment
Total
Source Table 20

Analyses will identify the number of households where the effective date of action on the
50058/50059 used in the study matches the effective date of action in the TRACS/PIC file. For
those households that match on effective date of action, we will determine whether certain key
variables match. Variables included in this analysis will be gross income, net income, tenant rent,
and total tenant payment (TTP). Exhibit 40 provides the percent of households where key variables
on the 50058/59 forms matched the TRACS/PIC data.
Exhibit 40
Percent of Matched and Non-Matched Dollar Amounts for Key Variables
Matching Variables from the 50058/50059 Form and TRACS/PIC Data Files (FY2006)
Gross Income

Net Income

Total Tenant Payment

Tenant Rent

Match Status
PIC

TRACS

PIC

TRACS

PIC

TRACS

PIC

No Match
Match
Subtotal
Missing
Total
Source Table 20

The households which included variables where the 50058/50059 data did not match the
TRACS/PIC data will be reviewed to determine if these households’ rent was calculated in error.
Exhibit 41 displays the cases with discrepancies in gross income, net income, total tenant payment,
and tenant rent, and the percents that also have rent errors.

Analysis Plan

31

July 24, 2006

Exhibit 41
Percent of Gross Dollar Rent Errors for Cases Where Key Variables Did Not Match (FY2006)
Gross Income

Net Income

Total Tenant Payment

Tenant Rent

Rent Error Status
PIC

TRACS

PIC

TRACS

PIC

TRACS

PIC

Rent Error
No Rent Error
Total

Analysis will also be conducted to determine whether non-matching households included
consistency, transcription or calculation errors within the 50058/50059. Exhibit 42 presents these
households by type of error.
Exhibit 42
Percent of Procedural Errors for Cases Where Key Variables Did Not Match (FY2006)
Calculation and
Consistency Error
Status

Gross Income
PIC

TRACS

Net Income
PIC

TRACS

Total Tenant Payment
PIC

TRACS

Tenant
Rent
PIC

Consistency Error
Allowance Calculation
Error
Income Calculation
Error
Other Calculation Error
Transcription Error

Analysis Plan

32

July 24, 2006

FINAL REPORT OUTLINE
The final report will communicate all study findings and recommendations to HUD, the assisted
housing community, the Congress, and other interested parties. As such, it must provide accurate
and clear findings in a fashion that is easy to read and understand. While many of the overall goals
of the project are straightforward, the processes for addressing them may be analytically complex.
The challenge in preparing the report is to present important findings without burdening the reader
with all of the complexity that went into conducting the analysis. Our approach to report preparation
is to use simple tabular and graphical displays that illustrate key findings.
The outline that will be followed to structure the report is presented below.
EXECUTIVE SUMMARY
I.

INTRODUCTION (Purpose, background, and organization of the report)

II. METHODOLOGY (Requirements and study standards, sample description, data collection
process, data sources, and analysis processes)
III. STUDY OBJECTIVES (Discussion of each of the study’s analytic objectives)
IV. FINDINGS (Narrative, tabular, and graphical presentations of the findings)
A. Rent Error
B. Sources of Error
C. Errors Detected Using Information Obtained from Project Files
D. Occupancy Standards Analysis
E. PIC/TRACS Analysis
F. Project Staff Questionnaire Analysis
G. The Impact of Project and Tenant Characteristics on Procedural Errors
H. Rent Reasonableness Analysis
I. Utility Allowance Analysis
J. Flat Rent Analysis
K. Payment Standards Analysis
V. RECOMMENDATIONS (Policy implications, and a discussion of how study
methodologies can be improved)
VI. APPENDICES
A.
B.
C.
D.
E.
F.
G.

Rent Calculations
Weighting Procedures
Source Tables
Consistency and Calculation Errors
Project Staff Questionnaire Analysis
Utility Allowance Analysis
Other Technical Analysis

Analysis Plan

33

July 24, 2006

Appendix A
Definitions of Key Terms

Definitions
Actual TTP—from the 50058/50059
Administration Type—PHA or Owner
Aggregate Error—the difference between the actual rental payment and the QC rental payment
Case Type—certification, recertification, and overdue recertification
Dollar Error Rate—the quotient of dividing the Total Gross Rent Error by the weighted sum of
the QC rents.
Case Error Rate—the quotient of dividing the sum of the weights of tenant cases with dollar
error rates in excess of $5 per month by the total sum of the weights of tenant cases.
Gross Rent Error—the sum of the absolute values of under- and overpayments
Largest Dollar Error—the annual dollar amount of error in the component with the largest
error
Overpayment—results when the tenant paid more than he/she should have paid; HUD's
contribution was less than it should have been
Payment Type—underpayment, proper payment, and overpayment
Program Type—Public Housing, Section 8 Vouchers, Section 8 Moderate Rehabilitation,
Section 8 Substantial Rehabilitation and New Construction, Section 8 Loan Management,
Section 8 Property Disposition, Section 202 PRAC/PAC, and Section 811 PRAC/PAC.
Quality Control Month (QCM)—the effective date of the most recent action in the file
Quality Control (QC) Total Tenant Payment (TTP)—calculated by using the household
interview and verification data
Rent Component—the five sources of income (earned, pensions, public assistance, other
income, and assets) and the five types of deductions (medical, child care, disability, dependent
allowance, and elderly/disabled family allowance)
Rent Dollar Error—the dollar amount of the Actual Rent minus the QC Rent for an individual
tenant. A negative number indicates an underpayment, meaning the household paid less than it
should and HUD’s subsidy was higher than it should have been. A positive number indicates a
household overpayment, meaning HUD’s contribution was less than it should have been.
Total Gross Rent Error—the weighted sum of the absolute values of positive and negative
Dollar Rent Errors.
Total Net Rent Error—the arithmetic value of the weighted sum of individual tenant Rent
Dollar Errors.
Underpayment—results when the tenant paid less than he/she should have paid; HUD's
contribution was higher than it should have been.

Analysis Plan

A-1

July 24, 2006

Appendix B
Source Tables Responding to Each Objective

Tables Responding to Objective(s)
OBJECTIVE
Objective 1: Identify the various types of errors and error rates and
related estimated variances.

TABLE

2.
3.

Percent of Households by Payment Type and Program Type
1a. Proper payment based on exact match of actual and QC rent
1b. Proper payment based on a match of actual and QC rent within $5
Dollar Rent Error by Program Type
Dollar Error Amount by Payment Type and Program Type

Objective 2: Identify the dollar costs of the various types of errors.

4.
5.
6.
7.
8.

Calculation Errors on Form 50058/59
Consistency Errors on Form 50058/59
Dollar Error Amount by Payment Type and Case Type
Case Type by Program Type
Administrative Error: percent of Households, Average Dollars in Error

Objective 3: Estimate national-level net costs for total errors and
major error types.

9.

Gross and Net Rent Error by Program Type

Objective 4

1.
3.

Percent of Households by Payment Type and Program Type
Dollar Error Amount by Payment Type and Program Type

Objective 5: Determine whether error rates and error costs have
statistically significant differences from program to
program.

9.
10.

Gross and Net Rent Error by Program Type
Total and Largest Dollar Error by Program Type for Households with
Rent Error

Objective 6: Determine the apparent cause of significant rent errors.

11.
•
•
•

Verification of Form 50058/59 Rent Component
Largest Component Error for Households with Rent Error
QC Rent Components by Payment Type and Administrative Type
Percent of Cases and Standard Error by Rent Component and Payment
Type
Allowances

Determine the relationship between errors detectable
using the HUD 50058 and HUD 50059 forms and total
errors found in the study.

1.

•

Multivariate regression analysis with error sources and error causes as
independent variables, and QC error as the dependent variable.
Objective 7: Determine the extent to which households are
overhoused relative to HUD’s occupancy standards.

16.

Occupancy Standards

Objective 8: Provide information on the extent to which errors are
concentrated in projects and programs.

2.

Dollar Rent Error by Program Type

These data are from the Project Staff Questionnaire
Objective 9: Estimate the percentage of newly certified tenants who
were incorrectly determined eligible for program
admission.
Analysis Plan

17.
18.

B-1

Percent of Newly Certified Households Meeting Certification Criteria
Percent of Newly Certified Households Meeting Certification Criteria by
Program Type
July 24, 2006

OBJECTIVE

TABLE

Objective 10: Determine the extent to which Section 8 voucher rent
comparability determinations are found in the tenant
file, and indicate the method used to support the
determination.
Objective 11: Estimate total positive and negative errors in terms of
HUD subsidies.

1.
3.

Percent of Households by Payment Type and Program Type
Dollar Error Amount by Program Type and Program Type

Objective 12: Determine the extent to which error rates in projects that
use an automated rent calculation system differ from
error rates in those that do not.

2.
3.

Dollar Rent Error by Program Type
Dollar Error Amount by Payment Type and Program Type

Objective 13: Determine whether other tenant or project characteristics
on which data are available are correlated with high or
low error rates.

Multivariate error prone analysis using tenant and project characteristics as
independent variables and QC error as the dependent variable.

Objective 14: Determine whether cases for which 50058/59 data had
been submitted to HUD were more or less likely to have
errors than those for which data had not been submitted.

19.
20.

Terms:

QC Errors by Match with TRACS/MTCS and Program
Payment Type by Program and Match with TRACS/MTCS

Rent Component: The five sources of income (earned, pensions, public assistance, other, and asset), three types of expense deductions (medical,
child care, and handicapped expenses), and two allowances (dependent and elderly allowances)
Rent Error: The difference between the Actual Rent (Total Tenant Payment) and the QC Rent; net rent error is the algebraic sum of over- and
underpayments; gross rent error is the sum of the absolute values of under- and overpayments.
Program Type: Public Housing, Section 8 Vouchers, Section 8 Moderate Rehabilitation, Section 8 Substantial Rehabilitation and New
Construction, Section 8 Loan Management, Section 8 Property Disposition, Section 202 PRAC/PAC, Section 811 PRAC/PAC
Administration Type: PHA or Owner
Payment Type: Underpayment, proper payment, and overpayment
Case Type: Certification, annual recertification, and overdue recertification

Analysis Plan

B-2

July 24, 2006

Appendix C
National Estimates Source Tables

National Estimate Source Tables
Table 1a.

Percent of Households by Payment Type by Program Type
Proper Payment based on exact match of Actual and QC Rent
UNDERPAYMENT
# of
Cases*

Row % of
Cases

Col %of
Cases

PROPER PAYMENT
# of
Cases*

Row % of
Cases

Col %of
Cases

OVERPAYMENT
# of
Cases*

Row % of
Cases

Col %of
Cases

# of
Cases*

TOTAL
Row % of
Cases

Col %of
Cases

PHA-Administered
Public Housing
PHA-Administered Sec. 8
Group Total
Owner-Administered
Group Total
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-1

July 24, 2006

National Estimate Source Tables
Table 1b.

Percent of Households by Payment Type and Program Type
Proper Payment based on a match of Actual and QC Rent Within $5
UNDERPAYMENT
# of
Cases*

Row %
of Cases

Col % of
Cases

PROPER PAYMENT
# of
Cases*

Row %
of Cases

Col % of
Cases

OVERPAYMENT
# of
Cases*

Row %
of Cases

Col % of
Cases

TOTAL
# of
Cases*

Row %
of Cases

Col % of
Cases

PHA-Administered
Public Housing
PHA-Adminisrtered Sec. 8
Group Total
Owner-Administered
Group Total
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-2

July 24, 2006

National Estimate Source Tables
Table 2. Dollar Rent Error by Program Type
ACTUAL RENT
(MONTHLY)

# of
Cases* (1)

Col %
of Cases

Sum
Dollar
Amount*
(2)

QC RENT
(MONTHLY)
Ave.
Dollar
Amount
(2)/(1)

# of
Cases*

Col % of
Cases

Sum
Dollar
Amount
* (3)

GROSS RENT ERROR
(MONTHLY)
Ave.
Dollar
Amount
(3)/(1)

Sum
Dollar
Amount*
(4)

Ave.
Dollar
Amount*
(4)

Error Rate

PHA-Administered
Public Housing
PHA-Adminisrtered Sec. 8
Group Total
Owner-Administered
Group Total
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-3

July 24, 2006

National Estimate Source Tables
Table 3. Dollar Error Amounts by Payment Type and Program Type
UNDERPAYMENT (MONTHLY)

# of Cases*

Col % of
Cases

Sum
Dollar
Amount*
(1)

Ave.
Dollar
Amount
(1) / (3)

OVERPAYMENT (MONTHLY)

# of Cases *

Col % of
Cases

Sum
Dollar
Amount*
(2)

Ave.
Dollar
Amount
(2) / (3)

QC RENT (MONTHLY)
# of
Cases*
(3)

Col % of
Cases

Sum
Dollar
Amount*
(4)

Ave.
Dollar
Amount
(4) / (3)

PHA-Administered
Public Housing
PHA-Adminisrtered Sec. 8

Group Total
Owner-Administered
Group Total
Table Total

Analysis Plan

C-4

July 24, 2006

National Estimate Source Tables
Table 4. Calculation Errors on Form 50058/59
FORM
50058

50059
Col %of
Cases

# of Cases*

Total Number of Cases

# of Cases*

Col %of
Cases

# of Cases*

Col % of
Cases

Age
Number of Family Members
Number of Foster Children & Live-in
Number of Dependents
Total Assets
Imputed Asset Income
Earned Income Sum
Pension, Etc., Income Sum
Public-Assistance Income Sum
Asset Income Sum
Other Income Sum
Total Non-asset Income
Income From Asset
Total Annual Income
Elderly/Disabled Allowance
Dependent Allowance
3% of Annual Income
Medical Allowance
Disability Allowance
Child Care Allowance
Total Allowance
Adjusted Annual Income
Gross Rent
Total Tenant Payment
Tenant Rent
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-5

July 24, 2006

National Estimate Source Tables
Table 5. Consistency Errors on Form 50058/59
50058
ITEM

# of
Errors*

# of Cases*

50059
# of
Errors*

# of Cases*

Total
# of
Errors*

# of Cases*

General Information
Household Composition
Net Family Assets and Income
Allowances & Adjusted Income
Family Rent and Subsidy Information
Flat Rent Schedule Information (PH only)
Note: * denotes values in the thousands

Analysis Plan

C-6

July 24, 2006

National Estimate Source Tables
Table 6. Dollar Error Amount by Payment Type and Case Type
UNDERPAYMENT (MONTHLY)
# of
Cases*

Sum
Col % of Dollar
Cases
Amount*

Ave.
Dollar
Amount

OVERPAYMENT (MONTHLY)
# of
Cases*

Sum
Col % of Dollar
Cases
Amount*

Ave.
Dollar
Amount

QC RENT (MONTHLY)
# of
Cases*

Sum
Col % of Dollar
Cases
Amount*

Ave.
Dollar
Amount

Certification
Group Total
Recertification
Non-Overdue
Overdue
Group Total
Table Total

Note: * denotes values in the thousands

Analysis Plan

C-7

July 24, 2006

National Estimate Source Tables
Table 7. Case Type by Program Type
CERTIFICATIONS
# of
Cases*

Row %
of Cases

Col %
of Cases

RECERTIFICATIONS/
NON-OVERDUE
# of
Cases*

Row %
of Cases

Col %
of Cases

RECERTIFICATIONS/OVERDUE
# of
Cases*

Row %
of Cases

Col % of
Cases

TOTAL
# of
Cases*

Row %
of Cases

Col % of
Cases

PHA-Administered
Public Housing
PHA-Administered Sec. 8
Group Total
Owner-Administered
Group Total
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-8

July 24, 2006

National Estimate Source Tables
Table 8. Administrative Error: Percent of Households, Average Dollars in Error
GROSS RENT ERROR (MONTHLY)
Error Type

# of
Cases*
(1)

Sum
Ave.
Dollar
Dollar
Col % of Amount * Amount
Cases
(2)
(2) / (1)

NET RENT ERROR (MONTHLY)
# of
Cases *
(3)

Col %
of
Cases

Sum
Ave.
Dollar
Dollar
Amount * Amount
(4)
(4) / (3)

QC RENT (MONTHLY)
# of
Cases *
(1)

Sum
Ave.
Col %
Dollar
Dollar
of
Amount* Amount
Cases
(5)
(4) / (3)

Transcription Errors
Calculation Errors - Allowances
Calculation Errors - Income
Calculation Errors - Other
Overdue Recertifications
Any Administrative Errors
Note: * denotes values in the thousands

Analysis Plan

C-9

July 24, 2006

National Estimate Source Tables
Table 9. Gross and Net Rent Error by Program Type
GROSS RENT ERROR (MONTHLY)
Sum
# of
Dollar
Cases * Col % Amount *
(1)
of Cases
(2)

Ave.
Dollar
Amount
(2) / (1)

NET RENT ERROR (MONTHLY)
Sum
Ave.
# of
Dollar
Dollar
Cases* Col % of Amount * Amount
(3)
Cases
(4)
(4) / (3)

QC RENT (MONTHLY)
# of
Cases*
(5)

Col %
of
Cases

Sum Dollar
Amount *
(6)

Ave.
Dollar
Amount
(6) / (5)

PHA-Administered
Public Housing
PHA-Adminisrtered Sec. 8

Group Total
Owner-Administered
Group Total
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-10

July 24, 2006

National Estimate Source Tables
Table 10. Total and Largest Dollar Error by Program Type for Households with Rent Errors
TOTAL DOLLAR IN ERROR
# of
Cases*

Col % of Cases

Sum Dollar
Amount*

LARGEST DOLLAR ERROR
Ave. Dollar
Amount

# of Cases*

Col % of Cases

Sum Dollar
Amount*

Ave. Dollar
Amount

PHA-Administered
Public Housing
PHA-Administered Sec. 8
Group Total
Owner-Administered
Group Total
Total
Note: * denotes values in the thousands

Analysis Plan

C-11

July 24, 2006

National Estimate Source Tables
Table 11.

Verification of Form 50058/59 Rent Component
Provided for Each Major Program Type
Third Party Verbal or in Writing, or Documentation
VERIFICATION

NO VERIFICATION

Dollar Amt. Not Matched

RENT COMPONENT
# of Cases*

Row %of
Cases

# of Cases*

Row %of
Cases

TOTAL

Dollar Amt. Matched
# of Cases*

Row %of
Cases

# of Cases*

Row %of
Cases

Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Elderly/Disabled
Allowance
Child Care Allowance
Disabled Allowance
Medical Allowance
Note: * denotes values in the thousands

Analysis Plan

C-12

July 24, 2006

National Estimate Source Tables
Table 12. Largest Component Error for Households with Rent Error (Annual Dollars)
RENT COMPONENT

# of Cases*

Col %of Cases

Sum Dollar Amount*

Ave. Dollar Amount

Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly Allowance
Child Care Allowance
Handicapped Allowance
Medical Allowance
No Error
Total
Note: * denotes values in the thousands

Analysis Plan

C-13

July 24, 2006

National Estimate Source Tables
Table 13. QC Rent Components by Payment Type and Administration Type
PHA-ADMINISTERED
RENT COMPONENT
# of Cases*

Col %of Cases

OWNER-ADMINISTERED
Row % of
Cases

# of Cases*

Col %of Cases

Row % of
Cases

TOTAL
# of Cases*

Col %of Cases

Underpayment
Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Allowance
Disability Allowance
Medical Allowance
No Error
Proper Payment
Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Allowance
Disability Allowance
Medical Allowance
No Error
Overpayment
Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Allowance
Disability Allowance
Medical Allowance
No Error
Total w/Rent Error Calc
Note: * denotes values in the thousands

Analysis Plan

C-14

July 24, 2006

Row % of
Cases

National Estimate Source Tables
Table 14. Percent of Cases and Standard Error by Rent Component and Payment Type
% PHA-ADMINISTERED
RENT COMPONENT
# of Cases*

% of Total
Cases

% OWNER-ADMINISTERED
SE (%)

# of Cases*

% of Total
Cases

SE (%)

TOTAL
# of Cases*

% of Total
Cases

Underpayment
Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Allowance
Disability Allowance
Medical Allowance
No Error
Proper Payment
Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Allowance
Disability Allowance
Medical Allowance
No Error
Overpayment
Earned Income
Pension, Etc.
Public Assistance
Other Income
Asset Income
Dependent Allowance
Elderly/Disabled Allowance
Child Care Allowance
Disability Allowance
Medical Allowance

No Error
Note: * denotes values in the thousands

Analysis Plan

C-15

July 24, 2006

SE (%)

National Estimate Source Tables
Table 15. Allowances
ELDERLY/DISABLED ALLOWANCE

NONELDERLY/DISABLED HH

# of Cases*

Col %of
Cases

Row % of
Cases

ELDERLY/DISABLED HH

# of Cases*

Col %of
Cases

Row % of
Cases

TABLE TOTAL

# of Cases*

Col %of Cases

Row % of
Cases

No Allowance
Incorrect Allowance
Correct Allowance
Table Total
Note: * denotes values in the thousands

DEPENDENT ALLOWANCE

HH W/OUT DEPENDENT

# of Cases*

Col %of Cases

Row % of
Cases

HH W/DEPENDENT

# of Cases*

Col %of
Cases

TABLE TOTAL
Row % of
Cases

# of Cases*

Col %of
Cases

Row % of
Cases

No Allowance
Incorrect Allowance
Correct Allowance
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-16

July 24, 2006

National Estimate Source Tables
Table 16. Occupancy Standards
Public Housing
Number of
Bedrooms

PHA-Administered Section 8

Table
Total

Owner-Administered

Underhoused

Correct

Overhoused

Group
Total

Underhoused

Correct

Overhoused

Group
Total

Underhoused

Correct

Overhoused

Group
Total

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

# of
Cases*

0
1
2
3
4
5
Table Total
Note: * denotes values in the thousands

Analysis Plan

C-17

July 24, 2006

National Estimate Source Tables
Table 16. Occupancy Standards (cont’d)
Percent of Cases
Public Housing
Number of
Bedrooms

Underhoused

Correct

PHA-Administered Section 8
Overhoused

Underhoused

Correct

Overhoused

Owner-Administered
Underhoused

Correct

Overhoused

0
1
2
3
4
5
Table Total

Analysis Plan

C-18

July 24, 2006

National Estimate Source Tables
Table 17. Percent of Newly Certified Households Meeting Certification Criteria
DID NOT MEET CRITERION

MET CRITERION
# of Cases
(in 1,000)

# of Cases
(in 1,000)

% of Cases

% of Cases

Citizenship
Social Security Number
Consent Form
Low and Very Low Income
Meets All Eligibility Criteria

Table 18. Percent of Newly Certified Households Meeting Certification Criteria by Program Type
MET CRITERION
# of Cases
(in 1,000)
PUBLIC HOUSING

% of Cases

DID NOT MEET CRITERION
# of Cases
(in 1,000)

% of Cases

Citizenship
Social Security Number
Consent Form
Low and Very Low Income
Meets All Eligibility Criteria

PHA-ADMINISTERED SECTION 8

Citizenship
Social Security Number
Consent Form
Low and Very Low Income
Meets All Eligibility Criteria

OWNER-ADMINISTERED

Citizenship
Social Security Number
Consent Form
Low and Very Low Income
Meets All Eligibility Criteria

Analysis Plan

C-19

July 24, 2006

National Estimate Source Tables
Table 19. QC Errors by Match with TRACS/PIC and Program
PERCENT OF CASES
50058/9 DATA ON TRACS/PIC

HOUSING AUTHORITY MANAGED

Public Housing

Section 8

Subtotal
PHA
Administered

TOTAL
OWNER
ADMINISTERED

All Projects

Matched With TRACS/PIC
% Cases in Error
Average Payment Error
Std. Error of Mean
Nonmatch with TRACS/PIC
% Cases in Error
Average Payment Error
Std. Error of Mean
Table Total
% Cases in Error
Average Payment Error
Std. Error of Mean

Analysis Plan

C-20

July 24, 2006

National Estimate Source Tables
Table 20. Payment Type by Program and Match with TRACS/PIC
UNDERPAYMENT
# of
Cases
(in 1,000)

Row
% of
Cases

Col %
of
Cases

PROPER PAYMENT
# of
Cases
(in 1,000)

Row
% of
Cases

Col %
of Cases

OVERPAYMENT
# of
Cases
(in 1,000)

Row %
of Cases

Col %
of Cases

TOTAL
# of
Cases
(in 1,000)

Row %
of Cases

Col % of
Cases

PHA-Administered
TRACS/PIC Present
TRACS/PIC Absent
Public Housing
TRACS/PIC Present
TRACS/PIC Absent
Section 8
TRACS/PIC Present
TRACS/PIC Absent
Owner-Administered
TRACS/PIC Present
TRACS/PIC Absent
Total
TRACS/PIC Present
TRACS/PIC Absent
Table Total

Analysis Plan

C-21

July 24, 2006