Behavioral Social Learning

We revisit the economic models of social learning by assuming that individuals update their beliefs in a non-Bayesian way. Individuals either overweigh or underweigh (in Bayesian terms) their private information relative to the public information revealed by the decisions of others and each individual's updating rule is private information. First, we consider a setting with perfectly rational individuals with a commonly known distribution of updating rules. We show that introducing heterogeneous updating rules in a simple social learning environment reconciles equilibrium predictions with laboratory evidence. Additionally, a model of social learning with bounded private beliefs and sufficiently rich updating rules corresponds to a model of social learning with unbounded private beliefs. A straightforward implication is that heterogeneity in updating rules is efficiency-enhancing in most social learning environments. Second, we investigate the implications of heterogeneous updating rules in social learning environments where individuals only understand the relation between the aggregate distribution of decisions and the state of the world. Unlike in rational social learning, heterogeneous updating rules do not lead to a substantial improvement of the societal welfare and there is always a non-negligible likelihood that individuals become extremely and wrongly conï¬dent about the state of the world.Social learning, Non-Bayesian updating, Herding, Informational cascades