Weighted Average Marginal Tax Rates for Barro-Redlick

1. Based on each year's income file
2. Based on 1985 actual data in/deflated to the tabulation year.
3. Using data from one year PRIOR to the law year.

   With wagelike incomes increased by 1% before calculation of marginal rates.

4. Based on each year's income file
5. Based on 1985 actual data in/deflated to the tabulation year.

These tables were prepared for the NBER Working Paper w15369 Macroeconomic Effects from Government Purchases and Taxes by Robert J. Barro and Charles J. Redlick.

These tables shows the average marginal rate of income plus FICA tax, weighted by income by source, estimated from micro-data files provided by the Statistics of Income Division of the IRS. using the TAXSIM tax calculator. In columns 2 through 6, only 4 wage-like income sources are included in the average, in columns 7 through 10 all income sources other than capital gains listed on the 1040 are included in the finite difference.

In columns 11 and 12 the root-mean-square average total marginal rate (federal plus state plus FICA) is shown for wagelike and then all non-gain income. The idea here is that deadweight loss is proportional to the square of the tax rate, so the RMS average may be more relevant than the arithmetic average.

For example, the first table shows that in 2006, a uniform percentage increase in all 1040 income items would be taxed by the federal government at a rate of 20.81 percent, averaged over all US taxpayers in that year.

Marginal rates calculated on a finite difference of a 1% increase in incomes but no change to deductions or adjustments. Negative income items are not changed, but are included in the tax calculation. The marginal rate is just the ratio of the tax difference to the gross income difference expressed as a percent. Incomes are shown in billions of current dollars.

The first table shows the marginal rates using the public use file issued each year by the ISR SOI. The second table uses only the 1985 public use file, and inflates or deflates it to the tabulation year. This second table is for use when an instrument for the tax law is required that is insensitive to economic growth. The third table uses the prior year data (uninflated) and applies the law to that. Hence state results start in 1980 rather than 1979.

The last two tables are the same as the first two, but 4 income items are increased by 1% before any other calculation is done, which shows how an increase in income will affect the marginal rates.

We assume that the employer share of FICA is borne by the employee, therefore the denominator income is grossed up by the employer share of FICA and the FICA rate above includes both the employer and employee share. The denominator for the FICA rate is the change in income from the 4 (or all non-gain) sources, not just the change in wages, hence the rate is lower than might be expected from the statutory rate.

An additional table showing the share of wage-like income going to itemizers.

There are no public use files for 1961,1963 or 1965. Post 2004 data is an extrapolation of 2004 actual data. State tax calculators are available in TAXSIM for 1977 on, but no state identifier is available in the data before 1979. Therefore state identifiers for taxpayers with AGI above $200,000 are imputed from aggregate counts. The income split between husbands and wives is also imputed.

Before using this data for publication or other serious purpose, please speak with me about llmitations.

Daniel Feenberg
feenberg at nber dotte org
617-588-0343 (voice)
Source: http://www.nber.org/taxsim/barro-redlick
Date: June 2009