What Matchings Can Be Stable? The Testable Implications of Matching Theory

This paper studies the falsifiability of two–sided matching theory when agents' preferences are unknown. A collection of matchings is rationalizable if there are preferences for the agents involved so that the matchings are stable. We show that there are nonrationalizable collections of matchings; hence, the theory is falsifiable. We also characterize the rationalizable collections of matchings, which leads to a test of matching theory in the spirit of revealed–preference tests of individual optimizing behavior.