“Matching markets: Theory and practice”

That's the title of this paper by Atila Abdulkadiroglu and Tayfun Sonmez, presented at the 2010 Econometric Society World Congress, in Shanghai.

Yeon-Koo Che has just posted his discussion of that paper. Among other things, he suggests that one reason we don’t see more centralized clearinghouses is that they don’t necessarily create Pareto improvements, and a hybrid approach is worth thinking about. (For example, he speculates, maybe a centralized clearinghouse for college admissions could be started if it explicitly allowed colleges to continue to engage in early admissions, and only tried to organize the "regular admissions" part of the market.)