This repository contains all working files for a project to develop validated input files for use in Tax-Calculator.
The current version is 2.0.0, which was released on March 29, 2026, and includes the following significant improvements:
- generate national, state, and Congressional district, input files for 2022: #470 #471 #472 #473 #474
- improve the selection of CPS tax units to represent nonfilers: #438
- vastly improve the reweighting algorithm: #416
- impute values for three variables used in new OBBBA deductions: #397
In order to use the code in this repository, you need to license the
2015 PUF from IRS/SOI. Once you have done that, you will have two
CSV-formatted files from IRS/SOI: puf_2015.csv and
demographics_2015.csv.
To generate the TMD files from the PUF files, do this:
- Copy the two 2015 PUF files to the
tmd/storage/inputfolder - Install the SIPP files described in
tmd/storage/input/SIPP24/README.md - Install the CEX files described in
tmd/storage/input/CEX23/README.md - Run
make cleanin the repository's top-level folder - Run
make datain the repository's top-level folder
The make data command creates and tests the three national
tmd*csv* data files, which are located in the tmd/storage/output
folder. Read this
documentation
on how to use these three files with Tax-Calculator. Also, you can
look at the tests in this repository to see Python code that uses the
TMD files with Tax-Calculator.