Beyond Bayes: information, beliefs, and economic outcomes
Publication Date
July 27, 2026
Creator
University of Nottingham
Abstract
This dissertation studies the economic consequences of how individuals process information, form beliefs, and make decisions, particularly when these processes depart from the Bayesian benchmark. While standard economic models assume that decision-makers seek and objectively incorporate relevant information, a growing body of evidence shows that individuals often avoid information, interpret it selectively, and maintain biased beliefs. Understanding the consequences of these departures is important because beliefs shape decisions in domains ranging from personal finance and health to organizational behavior and labor markets.
Chapter 1 investigates whether and how information avoidance —the deliberate refusal of freely available and instrumentally valuable information— is interdependent across individuals. Building on a model inspired by Bénabou (2013), it shows that when remaining ignorant is socially harmful, information acquisition decisions may exhibit strategic complementarity, strategic substitutability, or be independent of others' choices. A laboratory experiment finds substantial individual heterogeneity consistent with the theoretical predictions. Around 40% of subjects condition their choices on others' behavior, split roughly evenly between complements —who avoid information when many others do so, making ignorance contagious— and substitutes —who avoid information when many others acquire it, limiting the spread of ignorance. On average, complementarity dominates, making ignorance contagious. Evidence suggests the contagion is mediated by a deterioration of anticipatory utility induced by others' ignorance. The coexistence of heterogeneous individual responses to other information decisions implies that group composition strongly shapes the equilibrium level of information acquisition and can determine whether groups converge to widespread awareness or collective ignorance, as demonstrated by simulation results.
Chapter 2 studies the role of inaccurate beliefs and confirmation-biased learning in the evaluation of workers' performance, wage discrimination, and segregation in labor markets. The chapter develops a model in which employers evaluate workers by observing a sequence of noisy signals of performance and update beliefs under confirmation bias —so that they interpret signals in a way that aligns with their prior beliefs about the worker's performance. As a result, employers do not correct beliefs fully, even after observing an infinite number of unbiased signals, leading to persistent wage discrimination. Negative stereotypes generate discrimination against minority workers upon entry to the labor market, but are not enough to have discrimination in the long run, and reversals in discrimination are possible. The chapter also discusses whether interventions aimed at reducing discrimination would succeed if confirmation bias is an important source of discrimination. Finally, it considers segregation in an extension where employers are heterogeneous in their prior beliefs.
Together, the essays show that departures from Bayesian information processing can have important economic consequences. Information avoidance and confirmation-biased learning affect not only individual decisions but also collective outcomes, helping explain collective blindness to information, the formation and maintenance of inaccurate beliefs, and the persistence of inequality in organizations and markets. The dissertation therefore highlights the importance of incorporating psychological motives and social interactions into models of information acquisition and belief formation.
Item Type
ethesis
Thesis Type
PhD
Supervisors
University of Nottingham
University of Nottingham
Subjects (LC)
Associated Schools / Departments
UoN Repository URI
Except where otherwise noted, this item's license is described as
File(s)![Thumbnail Image]()
Name
Thesis.pdf
Size
7.63 MB
Format
Adobe PDF
Checksum (MD5)
67f0414278096f837ce574b621edb232