How much progressivity is optimal? (copied)

Over at the Hamilton Project, Jason Furman, Lawrence H. Summers, and Jason Bordoff say we should make the tax code more progressive:

At a minimum, the tax cuts of 2001 and 2003 should be repealed or allowed to expire for married couples making above $200,000 a year.
But they say we should not go too far:

To be sure, there are limits to how far policy should go in using progressive taxation to offset inequality.
The paper does not, however, say what those limits are. How do we know that the current top rate of 35 percent is not already past the limit?

Unfortunately, the authors do not spell out what economic model they have in mind in reaching their conclusions. Here is my guess, based in part on their footnote 8, about the model in their heads. They believe:

Tax policy should be set to maximize a social welfare function.
The social welfare function has strong aversion to inequality.
The tax-setting social planner must take into account that taxes are distortionary, but elasticities are small enough that the distortions are not much of a problem until tax rates reach much higher levels than we have today.
Note that points 1 and 2 are about political philosophy rather than economics. Economists can have opinions about these matters, but our opinions are not much better or worse by virtue of our training than those of anyone else. One can even reject the whole idea of a social welfare function. Utilitarian philosophers argue for interpersonal comparisons of utility, but nothing inherent in economics permits such comparisons. If we reject the idea of a social welfare function, then we are left with Pareto efficiency as the only valid criterion for judging tax policy, and economics has little to say about optimal progressivity--a point emphasized in a recent paper by MIT wunderkind Ivan Werning.

Point 3 is more economic, but it raises an issue about which economists hotly debate. See here and here. And even if that debate is resolved, point 3 on its own tells you nothing about optimal policy. You need the social welfare function to draw normative conclusions.

As a political document aimed at advising Democratic candidates and elected officials, the Hamilton paper makes sense. But as an intellectual document, it would been more compelling if the authors had spelled out the model of optimal tax policy they had in mind. It would have been nice to see explicitly their assumed social welfare function and elasticity estimates and then some calculations of optimal progressivity based on these. As it stands, the paper leaps to conclusions without stating premises, so it is hard for the reader to reach a reasoned judgment about whether to agree or disagree.

By the way, I bet if the Hamilton authors took their framework seriously, they would conclude that taxing height is an inevitable implication of their normative axioms.

Update: While my former student Jason Furman argues for greater progressivity, my former student Alan Viard takes the opposing view.