Sequential Learning under Informational Ambiguity
41 Pages Posted: 14 Nov 2019 Last revised: 30 Oct 2023
Date Written: September 1, 2019
Abstract
This paper studies a sequential social learning problem in which individuals face ambiguity regarding other people’s signal structures. It finds that ambiguity has a significant influence on social learning and offers new insights into the mechanism driving herding behavior. In contrast to previous findings that identified various learning outcomes based on fine details of the learning environments, such as the statistical properties of the signal structures, this paper establishes information cascades as the only robust outcome under ambiguity. Specifically, it demonstrates that in the presence of sufficient ambiguity, an information cascade will occur almost surely, regardless of the statistical properties of the signal structures or other specific details of the learning environments. Furthermore, this paper highlights that some standard results that feature the absence of a cascade can become fragile in the face of ambiguity. In some cases, even a slight degree of ambiguity can trigger a cascade when signals are bounded and can lead to incorrect learning when signals are unbounded.
Keywords: Social learning, information cascades, ambiguity, herding
JEL Classification: D81, D83, C72
Suggested Citation: Suggested Citation