Abstract:
Conventional Bayesian games of incomplete information are limited in their ability to represent severe in- completeness of information. Using an illustrative example of (seller offer) sequential bargaining with one- sided incomplete information, we analyze a dynamic game under ambiguity. The novelty of our model is the stark assumption that the seller has complete ignorance-represented by the set of all plausible prior distributions-over the buyer's type. We propose a new equilibrium concept-Perfect 0bjectivist Equi- librium (P0E)-in which multiple priors and full Bayesian updating characterize the belief system, and the uninformed player maximizes the infimum expected utility over non-weakly-dominated strategies. We provide a novel justification for refining P0E through Markov perfection, and obtain a unique refined equilib- rium. This results in a New "Coase Conjecture"-a competitive outcome arising from an apparent monopoly, which does not require the discount rate to approach zero, and is robust to reversion caused by reputation equilibria.
Key words: ambiguity, dynamic game with incomplete information, bargaining power, bargaining in- determinacy, Markov perfection, deflationary expectation, Coase Conjecture
JEL classification: D82, D83, C7, C78