Three Essays on Behavioral Economics: Narrow Bracketing and Ambiguity Aversion

Abstract

This collection of essays investigates the theory and applications of narrow bracketing and ambiguity aversion. In Chapter 1, I study two heuristics, narrow bracketing and correlation neglect, that decision makers adopt to simplify the evaluation of risk from multiple sources and axiomatize them as behavioral deviations from the expected utility benchmark by relaxing the independence axiom. When different sources represent different streams of income, my model can explain experimental evidence on violations of first order stochastic dominance and risk aversion over small gambles. When different sources represent consumption in different periods, I show that an Epstein-Zin type utility function can emerge because of narrow bracketing and it can resolve a critique of the original Epstein-Zin model. In Chapter 2 (co-authored with Rui Tang), we study the implementation problem of a mechanism designer with ambiguity averse agents. The mechanism designer, desiring to implement a choice correspondence, can create ambiguity for agents by committing to multiple allocation rules and transfer schemes without revealing which one to use. By extending the cyclical monotonicity condition from choice functions to choice correspondences, we show that the condition can fully characterize implementable choice correspondences. We then study the implementation problem in supermodular environments and apply our results to public procurement. In Chapter 3 (co-authored with Pëllumb Reshidi and João Thereze), we study asymptotic learning when the decision-maker is ambiguous about the precision of her information sources. She aims to estimate a state and evaluates outcomes according to the worst-case scenario. Under prior-by-prior updating, ambiguity regarding information sources induces ambiguity about the state. We show this induced ambiguity does not vanish even as the number of information sources grows indefinitely, and characterize the limit set of posteriors. The decision-maker's asymptotic estimate of the state is generically incorrect. We then consider several applications. For example, we provide a foundation for disagreement among agents with access to the same large dataset.