Monday, August 4, 2008

NBER Summer Institute

Two weeks ago I went to the NBER Summer Institute in Boston. I took in three of the five days, which past experience dictates (correctly) is when burn-out sets in. The days I took in included one day of the labor studies program, one day of the education program (joint with labor) and one day of the crime program (partly joint with labor). All of the session are held in parallel (along with the aging program, the children program and a couple of others) so that shopping is possible, but seemed to be discouraged this year by having the starting times for sessions in different programs staggered a bit.

I've resolved to be as postive as possible when I discuss conferences on here, so I will not talk about the couple of papers that seemed only so-so or the presenter I found irritating due to a lack of interest in getting the applied econometrics right. Instead, I'll remark on the papers that I liked and have stuck with me over the ensuing couple of weeks.

My favorite is this paper by Alan Krueger and and Andreas Mueller entitled "The Lot of the Unemployed: A Time Use Perspective". It uses time use data from a number of countries to look at what the unemployed do and, in a small sample suggestive sort of way tries to link the time use patterns to the institutions surrounding unemployment in different countries. The fun cocktail party fact is that unemployed Americans engage in job search related activities about 30 minutes a day while, for example, unemployed Belgians spend about six minutes a day and unemployed Swedes only about five minutes a day. The question I found myself pondering in regard to the figure for the US was: is 30 minutes a lot or a little? I am still pondering this as it is not so obvious. Clearly it should vary a lot by education and skill level. For workers with only limited general skills, there are many possible jobs and, at least in urban areas, one could presumably spend all day every day trying to track them down. In highly skilled labor markets, say for aeronautical engineers, there may be only a few really relevant employers, so that one applies to them and then waits. More research along these lines would be useful.

My friend and colleague John DiNardo presented - the crime group had the discussants present the papers when not meeting jointly with labor - this paper by Ethan Cohen-Cole, Steve Durlauf, Jeffrey Fagan and Dan Nagin entitled "Model Uncertainty and the Deterrent Effect of Capital Punishment". The paper attempts a quantitative summary of the literature on the deterrent effect of capital punishment. It does so not with meta-analysis, which is the classical statistics way of doing such things, but with a variant of what is called Bayesian model averaging. The discussion, as discussions of papers using Bayesian methods often do, ran up against the boundary of philosophy, which made it extra fun, as did the presence of several refugees from the labor group including David Card and Steve Pischke.

Finally, I also quite enjoyed the paper entitled "Estimating Teacher Impacts on Student Achievement: An Experimental Evaluation" by Tom Kane and Doug Staiger but they are not circulating their paper yet so I will leave it at that. Suffice it to say that if the results hold up they will get a zillion cites.