SUPPORTING STATEMENT PART B - Final

SUPPORTING STATEMENT

U.S. Department of Commerce

U.S. Census Bureau

Quarterly Summary of State and Local Tax Revenues

(Forms F-71, F-72, F-73)

OMB Control Number 0607-0112

Part B – Collections of Information Employing Statistical Methods

1. Universe and Respondent Selection

Form F-72 is sent to all 50 state governments and the District of Columbia. Since universe coverage is achieved, no sampling or estimation is employed. The complete canvass methodology is the only way to achieve national totals for state government tax collections.

Form F-71 is a sample sent to 6902 unique local governments that collect property taxes.

Form F-73 is a sample sent to 3910 unique local governments that collect non-property taxes.

Unit

Form Sample Universe Response Rate

F-71 Local government tax collectors

Old Sample 5,379 35,000 64%

New Sample 5,351 35,000 64%

Overlap of Samples 3,828 N/A N/A

F-72 State governments 51 51 100%

F-73 Major local governments

Old Panel 111 N/A 81%

New Sample 3879 49,613 NA

Overlap of Sample/Panel 80 NA NA

Response rates are based upon receipts during the initial quarter mail-out period and are subject to increase over subsequent quarter collections as late data are revised.

The survey collects data using three forms:

The Quarterly Survey of Property Tax Collections (Form F-71) is sent to 6,902 local government tax collecting agencies. While some counties are served by a single county level tax collection agency, others have county, city, township, and even school district collectors. Each agency is asked to report the total property tax collections during the past quarter. Currently we are running two samples (old and new) concurrently that increases the overall collection size to 6,902 local government tax collecting agencies (with 3,828 local governments overlapping on both samples). This new sample will ensure that this portion of the survey meets OMB’s statistical standards. We will continue to collect the concurrent samples for two additional quarters, at which time a bridge study will be conducted and the old sample will be discontinued.

The Quarterly Survey of State Tax Collections (Form F-72) is sent to a state level revenue, finance, or budget agency in each state to report tax collection data during the past quarter.

The Quarterly Survey – Selected Local Taxes (Form F-73) is sent to 3,910 local government tax collection agencies. Each respondent is asked to report the total non-property tax collection during the past quarter. In addition to the old panel, we are implementing a sample of 3,879 local tax-collecting agencies for up to six quarters, at which time a bridge study will be conducted and the old sample will be discontinued.

2. Procedures for Collecting Information

The survey collects data using three forms:

The Quarterly Survey of Property Tax Collections (Form F-71) is sent to 6,902 local government tax collecting agencies. While some counties are served by a single county level tax collection agency, others have county, city, township, and even school district collectors. Each agency is asked to report the total property tax collections during the past quarter. In addition to the old sample of 5,379, we are implementing a new sample of 5,351 which will increase the overall collection size to 6,902 local government tax-collecting agencies in 609 county areas (with 3,828 local governments overlapping on both samples). This sample revision will ensure that this portion of the survey meets OMB’s statistical standards. We will collect the concurrent samples for six quarters, at which time a bridge study will be published and the old sample will be discontinued. Units in the universe of county areas were stratified by population and a cost indicator prior to allocating using Neyman allocation. At this point a simple weighted estimate will be used until we research the usefulness of a benchmarking procedure. The sample was designed to yield a coefficient of variation of less than three percent on a national estimate of property tax. Coefficients of variation will be calculated and disseminated with the new sample.

The Quarterly Survey of State Tax Collections (Form F-72) is sent to a state level revenue, finance, or budget agency in each state to report tax collection data during the past quarter. This is a census of state tax collections so there is no sampling error associated with this part of the survey.

The Quarterly Survey – Selected Local Taxes (Form F-73) is sent to a sample of 3879 local tax-collecting agencies known to have substantial collections of local general sales, local individual income taxes and/or other non-property taxes. The sample universe is stratified in three steps. The first step is to determine which variable contributed the most to total non-property tax in each state by type. The second step is to determine the number of sub-strata each cell would be divided into based on the overall size of the cell. The third step is to stratify the cell using the cumulative square root of the frequency method. A simple random sample of five units is selected from each stratum. If a stratum has fewer than five units, then every unit is selected in that stratum. This results in each unit in the stratum receiving the same weight, and simplifies variance calculations. The sample is designed to yield a coefficient of variation of less than three percent on a national estimate of individual income tax, and on general sales and gross receipts.

Additionally, the 111 units from the old non-probability panel will be collected for at least six quarters.

3. Methods to Maximize Response

Telephone and email follow-up are conducted throughout the collection period to maximize response. New contacts are obtained for chronic non-respondent units as well as compiling data where available from Internet sources.

Respondents can and do submit late data for prior quarters that were imputed as well as revisions to prior quarter data (for up to seven prior quarters). They are also encouraged to submit reasonably estimated data for the current collection when actual data is not available by the time the current collection period is closed. These data are then revised when actual data is available.

4. Test of Procedures or Methods

The redesigned F-73 form underwent extensive cognitive testing, with representatives from town, village, and county governments participating. Testing was conducted in the San Francisco and DC metro areas, as well as in parts of Texas. See the attached summary report for detailed findings.

5. Contacts for Statistical Aspects and Data Collection

Statistical Aspects

Carma R. Hogue

(301) 763-4882

Assistant Division Chief – Research and Methodology Area

Governments Division

Data Collection

Phillip Vidal

(301) 763-1749

Section Chief

State Finance and Tax Statistics Branch

Governments Division

Attachments

F-71 survey form and cover letter

F-72 survey form and cover letter

F-73 survey form and cover letter

Cognitive testing report