Richard Green is a professor in the Sol Price School of Public Policy and the Marshall School of Business at the University of Southern California. This blog will feature commentary on the current state of housing, commercial real estate, mortgage finance, and urban development around the world. It may also at times have ruminations about graduate business education.
Saturday, May 21, 2011
A Plot of Effective Marginal Tax Rates and Per Capita GDP by State
Jared Bernstein recently posted a scatter plot of Federal Marginal Tax Rates and GDP growth, and found no correlation between the two. The graph above depicts the top potential marginal effective tax rate by state as calculated by the Tax Foundation (I will explain their calculation below) against GDP per capita by state. The correlation is actually positive, at about .2. If one removes the "DC effect," the correlation drops to about .19.
The top rate number I use from the calculation is the number produced under the GOP tax plan from late 2010, since they essentially got everything they wanted from the president in their tax deal. State taxes also move fairly slowly, so there is some persistence in the data across time.
I would not use this plot to argue that taxes on the richest cause higher living standards; but it sure is hard to argue that they cause living standards to fall.
I see two problems here: endogeneity and bad data. Tax rates and income may be endogenous because low income states cut tax rates to attract high income people. States with amenities that attract and hold high income people raise taxes.
ReplyDeleteThe data are suspect because published rates are not always true rates. In Iowa, for example, the published top rate is 8.98%, but federal taxes are deductible, so the true rate is just over 6%. Also, Iowa recently had a program to subsidize movie makers. Hollywood companies who shot movies here received Iowa tax credits, which they could sell. I recently bought a lot of them for 83 cents per dollar of credit, lowering my rate to about 5%. Taxes are far more complicated than most people realize, and I think a lot of tax research doesn't take this into account.