US Individual Income Tax Public Use Sample Documentation
This directory includes documentation for each of the (nearly) annual Tax Model Files. The Tax Model Files are public use samples of US Federal Individual Income Tax returns, with individual identifiers removed and the remaining data sufficiently rounded and blurred to prevent the identification of individual taxpayers. These have been issued since 1960, and are of special interest to researchers studying the income tax and the distribution of income. The files are poor in demographic information (no age, race or sex variables) but are rich in income information, especially for higher income taxpayers whose incomes are not likely to be accurately covered in survey data.
General Description Booklets
Each year a small booklet with a record layout and copies of the tax forms was issued by the SOI. These are available here as image only PDF files through 1994, then as PDF normal (searchable) format. These include information about disclosure avoidance techniques used in the file, tax forms showing the exact fields extracted and some summary statistics.
Basic documentation for each file is available in text form. This is a subset of the General Description Booklets.
For some recent years SOI has supplied us with SAS programs that generate the PUF files. These are easily edited to read the same files.
Obtaining the data files
Research Associates and Faculty Fellows of the National Bureau of Economic Research should contact Daniel Feenberg to use the data on the NBER Cambridge Unix system.
Others may obtain the files from the SOI division of the IRS (202-376-0081) or (for the older files) from the National Archives (301-713-6640). You will have to sign a confidentiality agreement for files obtained from the SOI.
There were no public use files in 1961, 1963 and 1965.
Last updated 27 Jan 2016