An estimate of the revenue effect of a change in the tax law has long been called a "revenue score" in Washington, and the act of making such estimates "scoring".
It is possible to do some simple scoring directly from the publications of the IRS, which tabulate many tax items by AGI. However since each AGI class includes taxpayers in differing marginal tax rates, and because the tax law is quite non-linear, the confidence placed in such back-of-the-envelope estimates is likely to be low. For example, a change in the personal exemption will affect taxpayers in the lowest income bracket hardly at all, since they typically have no taxable income after exemptions and the standard deduction. But some of those taxpayers are minimum tax payers who will see a significant change in tax. Similar considerations affect nearly every calculation done from the "Statistics of Income" tables.
This web service will allow students and others to make estimates of the revenue effect of changes in many of the parameters of the US Federal Income Tax for 2013. Once you hit the "submit" button below, our server will calculate the individual tax liabilities for each taxpayer in a large sample, with and without the change in the law you specify. A summary page tabulated by AGI will return 11 variables including taxable income, tax, and credits.
Visitors can specify changes to bracket rates and breakpoints, various ceilings, floors and phaseouts but can't change the basic structure of the tax system. For each change in the law, estimates are returned assuming no change in taxpayer behavior, and also assuming a (user specifiable) elasticity of income with respect to the after-tax share of income.
The database is a one in ten subset of the public use file provided by the Statistics of Income Division of the IRS. This is an anonymized and censored sample of tax returns for the 2007 tax year, with sufficient detail to calculate tax liabilities to a high degree of accuracy, without disclosing any identifiable returns. We age this to 2013 levels according to the growth in the adult population and the CBO forecast of income growth. We do not do anything to estimate the change in the distribution of income. This is a probability weighted sample, with high sampling rates for high incomes, so accuracy is good even for very high income strata. There are 2,862 returns in the subsample. More sophisticated micro-data samples projected into the future and with imputations for variables not shown on the tax return are the basis for official revenue estimates done by the Treasury, Joint Tax Committee, and by non-government organizations such as the NBER, Brookings-UI TPC, and various accounting firms.
Some cautions are in order. If you reduce the standard deduction there will be no increase in the number of itemizers, since the database, reflecting as it does actual tax returns, includes no itemized deductions for current standard deductors. Similarly for reductions in the medical deduction floor, etc. We will try to include mechanical links to warn users of such issues, but no doubt will miss some. Think about your request, and how it would play out on an existing tax return.
The appropriate elasticity is a highly controversial choice. A survey of the literature is available in an excellent CBO working paper. The details of the behavioral calculation are in Feldstein and Feenberg (1995) which is suggested reading if you intend to use the behavioral simulation.
The tax calculator used here is the TAXSIM program described in Feenberg and Coutts (198x) which should be read before any results from this page are propagated. Also, please give me a call before quoting ANY results from this page. If you prefer to write, please include a phone number and a good time to call, as I will want to talk to you about your application. There are many pitfalls in scoring, and I may be able to warn you away from a missuse of this resource (which is intended more as a teaching tool than a legislative resource).
Daniel Feenberg
617-863-0343
feenberg@nber.org