Working Papers
Sequential Learning under Informational Ambiguity, R&R at Econometrica [SSRN]
Abstract: This paper studies a sequential learning problem where individuals are ambiguous about other people's signal structures. It finds that ambiguity has an important impact on social learning and provides new insights on the mechanism behind herding behavior. This paper claims that whether an information cascade occurs is a result of individuals' ambiguity level instead of specific statistical features of the actual signal processes as suggested by previous literature. When there is sufficient ambiguity, for all possible data-generating processes, an information cascade occurs almost surely. Moreover, a slight degree of ambiguity suffices to produce a cascade when signals are bounded and destroys full learning when signals are unbounded. As an extension, this paper also investigates the case where there is an outside option. It finds that an information cascade occurs on this outside option when there is sufficient ambiguity and individuals are ambiguity-averse.
Biased Learning under Model Uncertainty [SSRN]
Abstract: This paper proposes a model of how biased individuals update beliefs in the presence of model uncertainty. Individuals are ambiguous about the actual signal-generating process and interpret signals according to the model that can best support their biases. This paper provides a complete characterization of the limit beliefs under this rule. The presence of model ambiguity has the following effects. First, it destroys correct learning even if infinitely many informative signals can be observed. When the ambiguity is sufficiently high, individuals can self-confirm their biases, leading to belief extremism and polarization. Second, an ambiguous individual can exhibit greater confidence than a Bayesian individual with any feasible model perception. This phenomenon comes from a novel complementary effect of different models in the belief set. As an extension, this paper also discusses the case where the bias can change with beliefs.
Abstract: This paper studies a sequential learning problem where individuals are ambiguous about other people's signal structures. It finds that ambiguity has an important impact on social learning and provides new insights on the mechanism behind herding behavior. This paper claims that whether an information cascade occurs is a result of individuals' ambiguity level instead of specific statistical features of the actual signal processes as suggested by previous literature. When there is sufficient ambiguity, for all possible data-generating processes, an information cascade occurs almost surely. Moreover, a slight degree of ambiguity suffices to produce a cascade when signals are bounded and destroys full learning when signals are unbounded. As an extension, this paper also investigates the case where there is an outside option. It finds that an information cascade occurs on this outside option when there is sufficient ambiguity and individuals are ambiguity-averse.
Biased Learning under Model Uncertainty [SSRN]
Abstract: This paper proposes a model of how biased individuals update beliefs in the presence of model uncertainty. Individuals are ambiguous about the actual signal-generating process and interpret signals according to the model that can best support their biases. This paper provides a complete characterization of the limit beliefs under this rule. The presence of model ambiguity has the following effects. First, it destroys correct learning even if infinitely many informative signals can be observed. When the ambiguity is sufficiently high, individuals can self-confirm their biases, leading to belief extremism and polarization. Second, an ambiguous individual can exhibit greater confidence than a Bayesian individual with any feasible model perception. This phenomenon comes from a novel complementary effect of different models in the belief set. As an extension, this paper also discusses the case where the bias can change with beliefs.