This table is for scholars who need an instrument for the marginal tax rate for a regression study with US data.
The actual income tax rate on a taxpayer is endogenous, i.e. it depends upon his income. So any regression explaining income (or labor supply or charitable giving...) in terms of after-tax prices should be wary of using the actual tax rate directly as an explanatory variable. But variation in state tax laws across states and years is exogenous to individual labor supply and realization decisions. So an instumental variable that depends only on the state law (or, to a lesser extent, on the year), and not on the individual has the potential to be a valid instrument.
Sometimes the maximum income tax rate by state and year is a nice independent variable in a cross-state tax-price regression. Although not representative of typical marginal rates, it is representative of tax rates on property income and it is independent of individual decisions. Of course still higher effective rates may be levied at lower income levels due to clawbacks (phaseouts) but these are not considered here.
Available here is a list of tax rates by year and state for the maximum state tax. It is calculated from a run of the TAXSIM model. Note that the top federal rate varies across states because of the deductibility of state income taxes on the federal tax.
This calculation was originally requested by Josh Lerner of the Harvard Business School, and reports the maximum tax rate for an additional $1000 of income on an initial $1,500,000 of wage income (split evenly between husband and wife). The taxpayer is assumed to be married and filing jointly and with $10,000 of long term capital gains. A mortgage interest deduction of $150,000 and the calculated state income tax are present as personal deductions. To avoid a simultaneous determination, the federal and state calculation is iterated to a fixed point. (Note, in real life each taxing authority allows a deduction only for taxes paid to the other, not tax liability, which avoids any simultaneity). This deductibility would be the main difference between these rates and the maximum bracket rate which might be published in a summary of state tax laws.
The rates shown include the interaction of the regular bracket rates with the phaseout of exemptions and itemized deductions, and many other features of the tax law, but not the minimum tax which by 2003 was affecting our synthetic taxpayer if and only if he lived in Oregon or Washington, DC. Suggestions for treating the AMT are welcome. It will always reduce the marginal rate, but generally not apply to the highest income taxpayers, so it doesn't seem exactly right to include it here. These are rates for very high income taxpayers rather than the maximum rates (which may be very high at moderate income levels in phaseout ranges). These phaseouts explain why there are so few round numbers in the table.
In January 2000, the program was rerun to get rates through 1999. The old run used by Lerner is available.. The January run introduced some errors in Wisconsin, and was replaced in September 2001 with a new run, correcting all known errors and extending the series to 2001. An update to 2003 is now posted. In April 2005 wage income doubled to 500,000 and was split between husband and wife, so that states with separate taxation of spouses didn't show a low rate for capital gains. Also, a mortgage interest deduction of $50,000 was introduced, so that all taxpayers would be federal itemizers in all years. In future years this will cause some taxpayers to owe minimum tax, but I don't think that has happened yet. Thanks to Matt Gardner of CTJ for help tracking down the need for these modifications and corrections to Arkansas and Colorado. Comments and corrections are welcome and should be sent to email@example.com, or you may call me at 617-588-0343 to discuss your application.
In January of 2006, David Harris requested the mortgage subsidy rate by state and year for high income taxpayers, and three columns have been added to the tables with that information. The calculation was extended to the 2005 tax year at that time. In April 2005 was updated for minor corrections. In April 2010 the program was rerun through 2008, with minor corrections. In February of 2011 the program was updated to 2009 tax law. In September of 2011 Andrew Lai pointed out that New Jersey had a top bracket above the then $500,000 in income used for the high income taxpayer, so the income was increased to $1,500,000 for the calculations when the table was updated to 2010 tax law.
In April of 2013 the base deduction for mortgage interest was set to zero, and $150,000 of property tax deductions substituted. This avoided including any AMT preferences among the deductions. Also, the calculations were extended to 2011.
In September of 2015 the table was recalculated, this time through 2014.
In August of 2016 the table was recalculated, this time through 2015. In November 2018 the table was recalculated through 2017.
In January 2019 Medicare NIIT was added to the federal rate. I also added a line for "Federal Only" each year.
Note that the entire table is recalculated each time a change is made, to improve comparability across years.
Here are the column labels for the ASCII version:
If you suspect any case has been calculated incorrectly, please extract that case (or a exemplar, if there are many cases) and email it to me with a statement of how you think the tax calculation ought to have been made. I will get back to you within a couple of days with an explanation or a fix.
This paper and this URL (http://www.nber.org/~taxsim)should be cited if any results from TAXSIM are circulated or published.
1050 Mass Ave.
Cambridge MA 02138
NBER home page | TAXSIM home page