SBR-9422563 Raymond Deneckere One of the most enduring puzzles in economics is why real world bargaining is characterized by so many apparent inefficiencies such as strikes, work slowdowns, and failures to reach agreements on contractural terms when it is known that both parties can benefit from an agreement. In such situations at least one of the parties must be uncertain about either its gains from trade and/or its opponent's gains from trade. We know this must be so because of Rubinstein's pioneering demonstration that when the gains to both parties from trading is common knowledge, then the unique bargaining equilibrium has the two parties agreeing immediately. Unfortunately, game theory has been unable to extend Rubinstein's uniqueness result to the case where the parties have incomplete information without making unrealistic assumptions about the ability of the bargainers to communicate with each other or imposing ad hoc restrictions on the equilibrium. This project will examine the conditions under which one very plausible restriction on the bargaining equilibrium-- that it satisfy Kohlberg and Merten's forward induction requirement-- is sufficiently strong so as to eliminate all bargaining equilibria but one. Such a result would uncover a fundamental link between rational bargaining delay and private information and move game theory forward in explaining the bargaining outcomes we actually observe. Preliminary work by the principal investigators concerning the case there the buyers are better informed than the sellers indicate that a key variable is the rate at which the players discount the future. When the discount rate is low, the bargaining equilibrium is semi-pooling and trade occurs in exactly three rounds. When the discount rate is high, all buyer types pool by making frivolous offers until the seller's beliefs cross a threshold, at which point the buyers make a serious offer that is then always accepted by the seller.
