SAS tax calculator - Example 1

This example show how to calculate the income tax liability for a sample of taxpayers in 2004 and compare the aggregate value from taxcalc.sas (c10300) to the SOI edited value (E10300).

Notes

The calculator is supplied as two SAS macros: %INIT and %COMP in a single %INCLUDE file. There are some other macros in that file that will be explained in later examples.

%INIT contains the ARRAY and RETAIN statements that characterize the tax law. Parameters that change over time are in arrays indexed by year.

%COMP contains the executable code to calculate the taxpayer liability and numerous intermediate values from the data supplied in the SOI individual tax files. _puf is set to 1 to inform that calculator that certain values are in "P" variables rather than "E" variables.

Note that the two macros together do not form a complete data step. They require introductory data and set statements, and allow for trailing statements also. This allows input or output data to be modified without adding the I/O and disk space burdens of multiple data steps. It will also allow changes in tax parameters and multiple calculations of tax without additional I/O, a feature to used in later examples.

The t.insoleNN files can be either the public use file distributed by SOI, or the internal file at SOI. The %INCLUDE file includes some code that detects the SOI SAS license, and redefines the libname if the code is running at SOI. For this reason you can modify the libname statement to point to your local PUF library, and no editing will be required when the job runs at SOI. Naturally this means that your libname statement must come before the inclusion of taxcalc.sas.

The calculator converts any missing input data to zeroes.

By default the tax is calculated according to the law for the filing period year given by the FLPDYR variable in the data. Here we replace the value of FLPDYR with the SOIYR (the file year) so that late returns are calculated according to the file year. Note that the SOIYR variable is a recent addition to the SOI datasets, and may not be available in exported files, in which case you will have to create it yourself.

E10300 is the element name given by SOI to "tax after credits". Wherever TAXCALC has calculated a value the name used for the claculated value is given the same numeric id, but the "E" is replaced with a "c". A list of input variables is here and calculated values here.

This exact program will run at SOI, and only the LIBNAME statement will change for other locations. The LIBNAME statemant above is required to run at NBER, but is overridden by a conditional LIBNAME statement in the include file when run at SOI. The explicit LIBNAME statement can be modified to work at your location, without affecting the ability to run without modification at SOI. OTA uses a slightly different SET statement.

The current version of TAXCALC is 97, and is distributed as taxcalc97.sas. The version number is printed in the log file at run time. At SOI a link from taxcalc.sas to the latest version should be maintained, but you can use the versioned name if it is important to maintain consistency with earlier work.