Coarse matching with incomplete information

We study two-sided markets with heterogeneous, privately informed agents who gain from being matched with better partners from the other side. Our main results quantify the relative attractiveness of a coarse matching scheme consisting of two classes of agents on each side, in terms of matching surplus (output), an intermediarys revenue, and the agentswelfare (de\u85ned by the total surplus minus payments to the intermediary). Following Chao and Wilson (1987) and McAfee (2002), our philosophy is that, if the worst-case scenario under coarse matching is not too bad relative to what is achievable by more complex, \u85 ner schemes, a coarse matching scheme will turn out to be preferable once the various transaction costs associated with \u85ne schemes are taken into account. Similarly, coarse matching schemes can be signi\u85cantly better than completely random matching, requiring only a minimal amount of information