Individual Income Tax Return Panel Data at SOI: What SOI Has and How You Can Use It Mike Weber Statistics of Income March 13, 2009 Michael.e.weber@irs.gov
Sources of Data available to SOI Individual Income Tax Returns Individual Information Returns (W-2s, 1099s, etc.) Date of birth and death, gender March 13, 2009 Weber/SOI 2
How SOI Samples Individual Income Tax Returns Computerized sampling of the IRS Individual Returns Transaction File (IRTF). Primary SSNs are transformed into random numbers. More than 100 income-based stratifications in the 2006 cross section. Cross section is highly stratified. March 13, 2009 Weber/SOI 3
How SOI Sample Individual Income Tax Returns - Continued Transformed SSN ranges are selected for each stratification based on expected populations. The transformed SSN ranges generally do not change from year to year. SOI also samples returns with specific SSN endings in the primary SSN position. Subsample of the Continuous Work History Sample (CWHS). March 13, 2009 Weber/SOI 4
SOI Individual Income Tax Return Panels Overlap Panels CWHS Primary SSN Panels CWHS Primary/Secondary SSN Panels Stratified Panels Prospective Stratified Panels Retrospective March 13, 2009 Weber/SOI 5
Overlap Panels Since SOI generally uses the same transformed SSN ranges and CWHS endings each year, there is a large overlap of taxpayers who are sampled every year (provided their income remains the same or increases.) Can start a panel in any year. Lose taxpayers whose income drops. March 13, 2009 Weber/SOI 6
CWHS Panel (Primary SSN) 1979 to present CWHS endings used varies from 1-10. 1979-1990 released as a Public-Use File through the University of Michigan. Random sample. Can start a panel in any year. Women who marry, and women who are widowed, are lost from the sample. Weak at high-income levels March 13, 2009 Weber/SOI 7
CWHS Panel (Primary & Secondary SSN) 1999 to present. 10 CWHS endings. Random Sample Can start a panel in any year. No gender bias Weak at high-income levels March 13, 2009 Weber/SOI 8
Prospective Panels 1987-1996 Family Panel 1999 Forward Edited Panel 2007 Forward Edited Panel All are subsamples of their base year cross sectional sample Highly stratified. Panel drift. March 13, 2009 Weber/SOI 9
Retrospective Panels SOI has never actually produced a retrospective panel. But the ability to do so exists through the IRS Compliance Data Warehouse (CDW). The CDW contains the Form 1040 population files back to 1997. No panel drift. March 13, 2009 Weber/SOI 10
The Big Problem with Individual Income Tax Return Panels What is the unit of analysis? The tax return or the individual? Jointly filed returns of married couples reflect their combined income. How do you separate their incomes? March 13, 2009 Weber/SOI 11
Information Returns When linked to tax returns, information returns can provide data on wage and income splits on joint returns When linked to a sample, they can identify nonfilers. We expect to have information documents available for all panels from 1999 forward. March 13, 2009 Weber/SOI 12
Panel Data Dissemination & Outside Researcher Access Tabulations Panels do not lend themselves to standardized tabulations. Public-Use Files The last one was in 1990. Reimbursable Requests Very expensive. Work with friends on the inside (OTA/JCT/CBO) with access to the data. March 13, 2009 Weber/SOI 13
New Approaches to Outside Researcher Access to Panel Data Tax Statistics Research Projects Annual program. SOI solicits research projects and then contracts with the winning researchers to perform the work. Access provided to the real data but the work must be performed at IRS facilities. Limited number for panel study. March 13, 2009 Weber/SOI 14
New Approaches to Outside Researcher Access to Panel Data SOI Panel Data Users Group - A group of external economists who are assisting SOI in the use of panel data sets. - Principle purpose is to serve as a link between external researchers and SOI staff willing to co-author papers with external researchers. March 13, 2009 Weber/SOI 15
Panel Data Users Group Members Dan Feenberg, NBER Peter Brady, Investment Company Institute (ICI) John Diamond, Baker Institute, Rice University Austin Nichols, Urban Institute March 13, 2009 Weber/SOI 16
The Process of Co-authoring a paper with SOI Staff External researchers will develop research proposals in consultation with the Panel Data Users Group. The Group will evaluate and rank the proposals. As SOI workloads permit, SOI Staff will be paired with the outside researcher. The effort will culminate in a paper for presentation at an economics or tax conference. March 13, 2009 Weber/SOI 17
Co-authoring Ground Rules The work will be done using SAS. The outside researcher is responsible for writing the SAS code. The SOI co-author will provide the necessary documentation and file expertise. They will also run the SAS programs and check for disclosure issues. March 13, 2009 Weber/SOI 18
Co-authoring Ground Rules The final SAS code will be available on the internet for others to critique and build upon. Over time, this library of SAS code will grow and allow more sophisticated research projects. March 13, 2009 Weber/SOI 19
Proposal Evaluation Criteria The Panel Data Users Group will evaluate proposals based on a number of factors. Quality of proposal. Quality of prior work or references General relevance to the use of panel data or tax policy. Knowledge of SOI data. Amount of file preparation. March 13, 2009 Weber/SOI 20
Mike Weber Statistics of Income (SOI) (202) 874-0877 Michael.e.weber@irs.gov March 13, 2009 Weber/SOI 21