Essays on Information Acquisition and Learning

Abstract

This dissertation focuses on the economic consequences of frictions to information acquisition and processing. The first two chapters study how costly information acquisition affects economic outcomes in different markets. The last chapter shows how a particular departure from Bayesian information processing distorts statistical learning. Chapter 1 examines adverse selection markets where consumers can, at a cost, learn how much they value a product. Information acquisition improves choice quality but comes at the cost of worsening selection, imposing a negative externality to the market. Consumer's ability to acquire information may decrease welfare, and may lead to new forms of market breakdown. I propose information policies to correct inefficiencies, and offer a framework for policy analysis. In Chapter 2, I investigate a screening problem under endogenous information acquisition. A monopolist offers quality-differentiated products, and consumers can costly learn about their taste for quality. We identify the optimal menu of products and show that all types of buyers typically receive lower-quality products than optimal. I identify a novel distortion due to the threat from the buyer to obtain information that is undesirable to the seller. Profits are U-shaped in information costs, and policies that reduce information costs may lower consumer's welfare by eroding their strategic power. In Chapter 3 (co-authored with P\"ellumb Reshidi and Mu Zhang), we study learning under ambiguity about the precision of information sources. Ambiguity about precisions induces ambiguity about the state the decision-maker aims to estimate. We show that this induced ambiguity does not vanish even as the number of sources grows indefinitely, and characterize the asymptotic set of beliefs about the state that the decision-maker entertains. We prove that a small amount of ambiguity can greatly exacerbate the effect of model misspecification on learning.