These job-market clearinghouses are built on a simple theory first formalized in 1962 by mathematicians David Gale and Lloyd Shapely. They designed an influential "deferred acceptance" algorithm showing how prospective market partners could pair off without any preferred matches being ignored. But in the medical job market, that theory breaks down when couples apply for jobs at the same time; the 1962 algorithm depends on everyone making decisions independently.
In medicine, the centralized job-placement system originated in the 1950s, and by the 1970s, it was increasingly common to find married couples entering the medical job market together. As a result, these clearinghouses began to change their mechanisms by the 1990s, allowing married couples to rank preferences for pairs of jobs.
However, "There had never been any real formal basis for those tweaks," says Pathak, an associate professor in MIT's Department of Economics. That meant it was not clear if there were a better system.
Moreover, some previous theoretical work in the field suggested that placing married couples in the same applicant pool as single doctors could make it impossible to find a stable matching. Mathematicians have shown that in some cases, even computing whether a stable matching exists is practically impossible.
"The theoretical results seemed to contradict the actual performance of the real-world markets," Pathak says. Together, Kojima, Pathak, and Roth worked out the assumptions under which it is possible to reconcile the theory with the practice, and used data from the job market for clinical psychologists to test their theory.
Good test results for d
|Contact: Sarah McDonnell|
Massachusetts Institute of Technology