Analysts at each of these sources rely on the same data sources (tax and Census records) in building their microsimulation tax models. And each, if asked the same question, would give about the same answer.
So why do the tables from these three sources seem to differ, sometimes radically, in their apparent results?
The primary reason for the differences reflects which provisions of the tax bills are shown in the tables. A secondary factor is whether provisions are shown fully in effect or only partially effective.
The congressional tax-writing committees, for example, have told their technical staff to prepare tables that leave out most of the tax benefits that primarily go to the very well off, in particular the proposed estate tax and capital gains tax cuts. That has allowed congressional leaders to assert that their proposed tax cuts primarily benefit the middle class.
The congressional tables also show some important provisions, such as new tax-free savings accounts, after they have been in effect for only a few years. Since the tax benefits from these provisions grow rapidly over time, this approach makes them look far smaller than when they are fully effective.
The following table outlines the key differences among the distribution tables:
Further Explanation of Difference Between CTJ and Congressional Distributional Tables
More on how the CTJ/ITEP Analysis of the Tax Plans Compares to Treasury's Analysis.