What predicts success in economics PhD programs?

When I was a graduate student, my mentor Jim Poterba told me more than once that my research should “always be about the economy and never about economists.” I took those words to heart and consistently resisted the temptation to do self-referential research about the economics profession.

I finally violated Poterba’s rule at this year’s American Economic Association meetings. Doubly ironic, one of the co-authors on the paper was Jim Poterba himself! We both got involved because we thought the project had some special merit for the profession.

Alan Krueger spearheaded an ambitious project to bring together parallel data on students from Princeton, Chicago, MIT, Stanford, and Harvard graduate programs in economics with the hopes that we could better understand the graduate admissions and training process. Susan Athey (who was at Stanford at the time) and Larry Katz were the other co-authors. You can read the paper here, and read about the paper at insidehighered.com.

I can’t really take any credit or blame for the paper — Alan Krueger did all the work and had all the ideas. It is a stretch to even consider me a co-author. (It would be more appropriate to make Julie Less, who worked tirelessly to assemble the U of C data, the U of C co-author.)

Our results thus far, unfortunately, are not particularly earth-shattering. Admissions committees are good at predicting who will do well in first-year classes, but not good at predicting who will find good jobs after graduate school. The first-year course grades are strong predictors of job placement. Once you take into account first-year grades, it turns out that GRE scores, gender, and doing a foreign undergraduate degree do not predict who will get a good job.

There were three findings I think are most interesting. First, the GRE analytical score turns out to be more important than the GRE math score in predicting first-year grades, despite the tremendous emphasis that is put on the GRE math score in the admissions process. (In contrast, ETS discontinued the analytical part of the test!). Second, women do very badly in first-year courses given their background characteristics, but they do fine on the job market. Third, after controlling for other factors, students who went to top undergraduate institutions do not do much better in classes, but they do a lot better on the job market.