Heterogeneity and equilibrium

Abstract

The research reported in this dissertation explores the observable effects that individual heterogeneity implies in strategic environments. The first chapter provides a focused experimental test of mixed strategy play in strictly competitive games. The experiment directly tests whether serial correlation results from subjects' inability to generate sequences of actions that appear to be time independent, or instead from the play of non-equilibrium strategies. This is achieved by allowing the subjects to generate actions via a simple randomizing device. It is found that serial correlation is not reduced and that subjects adopt a wide variety of non-equilibrium mixed strategies. This wide variety of mixtures potentially explains the seeming paradox of minimax winning proportions with a high variance of win rates across pairs of players. In the second chapter a theoretical model is developed for simultaneous move games in which the observable outcomes are allocations of monetary payoffs or commodity bundles, not expected utility levels. It is assumed that the players' mappings from the uncertain money amounts or commodity bundle allocations to expected utility levels are heterogeneous and are private information. The third chapter applies this framework to investigate the incentives to form agricultural marketing pools.