Wednesday, January 22, 2014

When is a class not a class?

Depending on how you define your terms, yesterday either did or didn't witness the semester's first PM meeting of the NYU Tax Policy Colloquium, in its nineteenth year of operations.

The introductory AM class, which is private (i.e., just for the enrolled students) went off fine despite the approach of threatening weather.  But the public PM session, scheduled to meet at 4 pm, ran into a bit of a hitch when it was decided, at about 3 pm, to cancel all NYU Law School classes for the day that were scheduled to begin at 4 pm or later.

This was an entirely reasonable, and indeed probably necessary, decision given the steadily falling snow, severe temperatures, and threat of travel disruptions for people not living in the immediate area.  But when you have your speaker at hand from out-of-town, have prepared extensively for the discussion, and have separately met earlier in the day both with the students and with the author to lay the groundwork for a good session, it's less than entirely welcome.  We ended up holding a voluntary session (i.e., not an official class at which student attendance was expected) and had a good discussion notwithstanding.

Yesterday's author was Saul Levmore, discussing two short papers (available here and here) concerning internalities and regulation.  Consider mandated saving (such as under Social Security) or cigarette taxes and smoking bans.  The papers contrast the "old view," in which the underlying regulatory aims sound in externalities or paternalism, with a "new view" in which they instead aim to address people's self-control problems.

At the risk of headlining mere semantics, I don't view these alternatives as "old view" versus "new view."  Usually, when we contrast such things, the two views actually contradict each other.  The "old view" of corporate dividends holds that the shareholder-level tax discourages paying them.  The "new view" shows that there is no such discouragement under specified circumstances.  The "old view" of transition relief holds that, when there is a legal change (such as repealing the income tax exemption for municipal bonds) relief such as grandfathering should be granted, in order to protect reliance interests.  The "new view" instead emphasizes incentive effects from anticipation, and thus in many circumstances favors not granting transition relief.

But in the internalities / regulatory setting we instead get related and somewhat overlapping rationales for approximately the same policies.  For example, one can favor forced saving via Social Security both so the elderly won't need public support by reason of indigence and to help them optimize, if we fear they might otherwise save too little from the standpoint of their own long-term self-interest, although it is true that the two alternative rationales may have different implications for program design.  (E.g., the fiscal externality would suggest requiring just enough saving to avoid public support, and indeed would suggest nothing if public support were repealed.)

I also would quibble with the papers' definition of an internality, which they extended to cover collective action problems in which each individual is acting rationally and time-consistently.  An example that the papers discuss is helmet regulation.  In many places that don't require motorcycle helmets, nobody wears them.  Arguably, the riders have some desire to wear the helmets for safety but don't want to out themselves as nerds by doing so.  But once a helmet is legally mandated, you can get the safety without causing people to view you as a nerd.

This example (probably in truth, but certainly under the stipulated facts) is a benign example of regulation working well, but I wouldn't say that it addresses an internality problem.  After all, it's entirely rational not to want look like a nerd.  (As Hume famously said, reason is merely the slave of the passions, and most of us have "passions" that extend to caring about how people perceive us.)  There is no time-inconsistency or self-control failure in this story.  Rather, it's an externality issue - if I don't wear a helmet, I affect the social meaning when someone else does - that may be difficult to solve privately due to collective action problems.

More generally, I don't have the sense that internalities transform the regulatory analysis as much as the papers suggest.  Within neoclassical economics, allowing for internalities is a radical step because it challenges the basic rationality assumption, as well as greatly muddying the effort to discern preferences and utility from behavior.  But once you allow yourself to take this step, the analysis goes forward in largely familiar ways.  (Not entirely so, because the potentially missing "transaction" is between present and future selves, rather than different people.)  The papers have some interesting things to say about how interest group politics and information problems in discerning the proper policy can be big issues in a regulatory setting where internalities are important.  But this is also the case if one just has externalities to deal with.

Although I don't here comment on the sessions themselves, in order to preserve their being off-the-record, I will note that Levmore is always good value as a speaker and guest, which is one reason we scheduled him for Week 1, and also helps to explain why I did not want to lose the session.

No comments: