Tuesday, June 9, 2009

When matching markets go crazy

A long standing problem in matching markets is that they tend to become highly inefficient as one side rushes to get the best applicants. The classic example is the market for medical interns, where the candidates thought to be the best were receiving offers from hospitals barely after the start of medical school. The inefficiency here stems not only from the obvious incomplete information about the quality of the future interns, but also from the adverse incentives of already matched students, who will not study as hard. One could also see some of that inefficiency on the academic market in Economics, where schools rush to make offers as early as possible, and now sometimes even before students officially go on the market. In medicine, the solution was to force all hospitals and students to participate in a matching mechanism at the end of studies. Both sides submit preference lists, and an algorithm makes proper assignments, make can be shown to be efficient under some conditions, and they are certainly more efficient that under the previous regime.

But when should one put such a mechanism in place? In other words, when is a matching market inefficient when left alone? Muriel Niederle, Alvin Roth and Utku Ünver study this question and conclude that the following are needed: a surplus of applicants and a shortage of good applicants. Then, "firms" will try to make early offer to the best workers, and they are willing to accept such offers. This is reminiscent of early admission strategies in college, discussed before on this blog. This raises thus the question whether it would be more efficient to impose a matching mechanism there as well. Some countries have already done so, for example Germany for medical students.

No comments: