Bargaining with Learning

Abstract

We analyze a continuous-time bargaining game of two-sided incomplete information without time discounting. Consistent with existing results, no trade occurs in the unique equilibrium of this game. Next we assume that players have imperfect information about their types; that is, they do not exactly know their private valuations. We suggest a learning model according to which players learn about their types during the bargaining process. Under some conditions, there exist equilibria of this game where trade occurs with positive probability. Moreover, there exists an equilibrium that is ex-post efficient. This shows that imperfect information and a very simple boundedly rational model of learning can circumvent the Myerson-Satterthwaite theorem. These results continue to hold when the game is extended to allow alternating offers from the buyer and the seller.

Keywords: Bargaining, incomplete information, imperfect information, bounded rationality, Myerson-Satterthwaite theorem

JEL Classification: C78, D83