NYU Tax Policy Colloquium on David Gamage's Optimal Tax Theory Meets Tax Avoidance

I guess I've been doing these colloquium sessions for long enough now that, at some level of generality, nearly everything I encounter has happened before.  The genre for today's session, which happens at least every other year or so (maybe more), was being initially really irked by a paper, leading to skirmishing at lunch (today, including eye-rolling by all 3 of us), followed by a turn for the better after hammering it out.  For this to work, one has to try to be open-minded, and to voice objections clearly (up to the limits of civility, the franker the better), but also to convey wanting to increase mutual understanding, not win a battle.  Also, as a matter of social dynamics, I think one really needs somehow the group of 3 - two conveners (in this case, myself and Mihir Desai) along with the author.  One-on-one seems to develop worse dynamics and less engagement.

Anyway, Gamage's paper irked me as I read it in advance because it seemed to combine making very aggressive claims about errors and omissions in the prior literature (including renowned articles by Nobel economists) with not being entirely clear or persuasive about what was new.   The issue goes to how taking account of certain tax-responsive behaviors, such as reducing taxable income through adjustments to behavior other than reducing work or saving, might, if they have a certain cost structure, lead us to think very differently about optimal instruments.   The early draft we read suggests we might like "double taxation" of various kinds, but the true thrust turned out to be considerably different.

I don't quite have the energy, as I write this late at night, to go through all the permutations of where the discussion led us.  But in sum, if avoidance (i.e., tax-responsive real behavior particular to the institutional details of exactly how we are collecting the tax) turns out to have a rising marginal cost structure, then in some cases we might want to have as many separate collection points as possible, all else equal.

As a collectively developed analogy put it, to combat fare-jumping in the NYC subway, we might want to collect half of the fare when you enter and the other half when you leave, if in practice this means people will no longer find it worthwhile to jump over the turnstile and risk spraining their knees.  Leaving for later (i.e., for Gamage as he works on the draft) the question of how generalizable and important to real world situations this insight is.  I honestly don't know how this will come out, but I certainly wish him the best, and am reasonably hopeful that he will either develop the analysis in a more clearly useful fashion or turn to something else.