Mark Gritter (markgritter) wrote,
Mark Gritter

Corporate Tax Rate Analysis

The Strib ran a commentary today from Kevin Hassett: the case for lowering the corporate tax. He argues, based on empirical studies done by himself and Alex Brill, that the U.S. would raise greater revenue with a somewhat lower corporate tax rate. (The beauty of such an empirical study is that, should a link be found, it relieves you of the burden of explaining where the additional revenue actually comes from--- greater compliance, movements of profits, economic growth, etc. And his commentary does not.)

So I looked up the AEI working paper he published in 2007, since this seems to be the source of his claim. The authors offer a reasonable model (although the Laffer curve, if it can be said to exist at all, is *certainly* not quadratic) but I found several points that concerned me.

The paper suggests that the revenue-maximizing rate is somewhere between 29.5% and 37.1%. Or, using a 5-year lagged model (with provides the best fit), somewhere between 30.6% and 34.1%. So why suggest, in his commentary, a rate of 26.4%? (Could it be that we are not actually on a Laffer curve but just a "race to the bottom"? If so, why not just shortcut down to zero?)

The authors report R^2 values. The best fit is about 0.15; that is, about 15% of the variation in revenue is accounted for by corporate tax rate. I don't know what the standard is for economic results, but this does not seem to me a particularly good showing from a public-policy perspective. That is, are there other factors which could better explain or more dramatically affect the remaining 85% of revenue variability? On the other hand, I could believe that the government funds plenty of activities with less justification! But, it is clear that Hassett's assertion that lowering tax rates would provide "$748 billion over the next ten years" is vastly overstating what he actually demonstrated.

The measures of significance provided do show that the result is real--- i.e., statistically significant. However, I am less pleased that they do not carry through an analysis of the error on the prediction of revenue-maximizing rates.

From table 4:
beta_1 = Tax rate (5 year lag) coefficient = 0.175 w/ 0.018 standard error
beta_2 = Tax rate^2 coefficient = -0.286 w/ 0.030 standard error
alpha = Constant coefficient = 0.007 w/ 0.002 standard error

Their calculation:
revenue/GDP = 0.007 + 0.175*x + -0.286*x^2
y maximized at 0.175 - 0.572x = 0
x = 0.306

If instead, we plug in the standard errors on the coefficients, we get a recommended tax rate anywhere from 0.248 to 0.377. So, in fact, the corporate U.S. tax rate may already be at the revenue-maximizing point (though they present anecdotal evidence to suggest otherwise.)
Tags: economics, statistics, taxes
  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded