Supplement to “Sequential Learning under Informational Ambiguity”

Abstract

This online appendix provides the following materials: Section S1 provides a necessary and sufficient condition for complete learning with power-tail DGPs. Section S2 provides conditions that are close to necessary and sufficient for information cascades. Sections S3 and S4 explore extensions to settings with multiple actions and multiple states, respectively. Section S5 examines an extension in which individuals use mixed strategies to hedge against ambiguity.  Section S6 explores social learning under consistent ambiguity, illustrating how cascades arise from ambiguity attitudes rather than model misspecification. Section S7 investigates the case in which individuals face ambiguity about their own DGPs.  Section S8 analyzes learning under heterogeneous ambiguity. Section S9 discusses heterogeneous preferences.  Section S10 discusses an alternative updating rule: the alpha-maximum likelihood rule. Section S11 considers an extension in which individuals face ambiguity about the network

structure. Appendix A contains omitted proofs.