Sequential Choice and Non-Bayesian Observational Learning

Models of observational learning in settings of sequential choice have two key features. The first is that players make decisions by using Bayes’ rule to update their beliefs about payoffs from a common prior. The second is that each agent’s decision rule is common knowledge, so that subsequent players can draw inferences about unobserved private signals from observable actions. In this paper, I relax the first assumption while maintaining the second. In particular, I look at observational learning by players who choose among actions using nonparametric methods for estimating payoffs. When players are identical and learn according to the maximum score method, an informational cascade must result. If players of different types with respect to payoffs use kernel or nearest neighbor learning rules, there are cases in which neither a herd nor a cascade need arise. If a cascade does occur, it must be one in which all players, regardless of type, choose the same action. In some situations, these alternative learning rules actually perform better than Bayesian updating. * Department of Economics, University of Texas at Austin, Austin, TX 78712; [email protected].