Observational Learning with Ordered States
Participate
Department of Economics & Decision Sciences
Speaker : Navin Kartik
Professor at Columbia University
Videoconference
Abstract :
When does society eventually learn the truth, or underlying state, via sequential observational learning? This paper develops the interplay of preferences satisfying single-crossing differences (SCD) and a new informational condition, directionally unbounded beliefs (DUB). SCD preferences and DUB information are a jointly minimal pair of sufficient conditions for learning. When there are more than two states, DUB is weaker than unbounded beliefs, which characterizes learning for all preferences (Smith and Sorensen, 2000). Unlike unbounded beliefs, DUB is compatible with the monotone likelihood ratio property, and satisfied, for example, by normal information.