What’s the best top marginal income tax rate?

-opportunity-for-all_152949441244Amid all the hoopla over French economist Thomas Piketty‘s lengthy new book, which calls for an 80 percent top marginal tax rate in the United States, this observer read a much shorter book the American Enterprise Institute published recently. It’s titled Opportunity For All. Through a collection of columns, research reports, journal articles, and congressional testimony transcripts, AEI tackles the issue of income inequality.

There’s an over-reliance in spots on jargon and mathematical formulas, and several articles repeat the same themes — using the same language. Still, AEI has put together a nice resource for those looking for effective responses to President Obama’s claims about the seriousness of an inequality problem in the United States.

Piketty came to mind because the last chapter of Opportunity For All features a response to the suggestion a couple of years ago from economists Peter Diamond and Emmanuel Saez that “the socially optimal top marginal income tax rate is 73 percent, with a range from 54 to 80 percent.” AEI resident scholars Aparna Mathur, Sita Slavov, and Michael Strain put that assertion to the test.

While Diamond and Saez base their recommendation on mathematical calculations, the AEI team points out early on why that approach can lead to problems.

It is extremely difficult to take the results in an academic journal article and apply them to real-world policy questions because the method used in much of economics research is to start with assumptions and to derive conclusions from them. Much of economics research is a series of if-then statements. But as economists readily admit, the ifs are often wildly incorrect. That is not a problem for academic research, in which all results come with a long list of caveats and in which readers are professional economists who spend years in PhD programs learning the caveats. But it does present a problem for economists who try to base specific real-world policy prescriptions on highly stylized models.

Then Mathur, Slavov, and Strain set about explaining why Diamond and Saez’s “ifs” are likely to be incorrect. One major “if” that falls short of the mark is the assumption that a much higher top marginal tax rate would have no impact on the behavior of high-income earners and potential future high-income earners.

Diamond and Saez’s short-run estimate completely ignores the effects an increase in the top marginal income tax rate has on choice of schooling, occupation, entrepreneurship, and business development. They argue that they ‘‘unfortunately have little compelling empirical evidence to assess whether taxes affect earnings through those long-run channels.’’

We agree with Diamond and Saez that the economics literature does not have good estimates of the long-run effects of high top marginal tax rates on human capital accumulation, career, entrepreneurship, and business development choices. But most economists would agree that those effects exist and may be important. We do not have estimates of their magnitude only because we lack appropriate data. Indeed, the most important questions in economics are often the hardest to answer because of data limitations. This is certainly one of those cases.

Change one assumption in Diamond and Saez’s formula, and the resulting “optimal” marginal tax rate is much different.

Assume Diamond and Saez have the right formula. Let’s use a value for the behavioral elasticity for high earners of 1.3, as estimated [in two other cited research reports]. Using their model, the optimal top tax rate is 34 percent.

Really? 34 percent? A rate almost 6 percentage points lower than the top marginal tax rate now in place?

Because of the limitations of applying mathematical models to the discussion of optimal tax rates, the AEI team reaches an important conclusion.

Perhaps we have stumbled on a better algorithm than Diamond and Saez’s formula? We think so. In a real-world policy setting, the optimal top rate should be determined by taking the status quo seriously and deciding how to deviate from it in a way that will be welcomed by society and implementable by the president and Congress. Advocates for particular policies should not allow the pursuit of the perfect to become the enemy of the good. …

… We don’t know what to do about the fact that economists haven’t credibly and precisely estimated many effects that are of first-order importance in policy, but we’re pretty sure that the solution isn’t to pretend the effects don’t matter.

No comments yet. You should be kind and add one!

Our apologies, you must be registered and logged in to post a comment.