Biases in Perceptions, Beliefs and Behavior

Abstract

This paper presents a model where individuals have imperfect information about their preferences (or the environment) and there is an opportunity cost of learning. It shows that the endogenous decision to collect information before taking an action creates a systematic bias in the aggregate behavior of a population of rational, profit-maximizing agents. More precisely, individuals will favor actions with large payoff-variance, i.e., those which may potentially generate the highest benefits even if they may also generate the biggest losses. The paper thus concludes that systematically biased choices do not necessarily imply that agents have irrational, systematically biased beliefs. It also provides testable implications about the propensity of individuals to incur different types of errors. Some applications such as biases in judicial decision-making and career choices are discussed.