Positive self image and asymmetries in information processing: Existence and implications for economic analysis

This dissertation investigates the existence and relevance of a class of information-processing biases from an economic point of view. There is widespread evidence from psychology that individuals perceive their own abilities more positively than others perceive these abilities. The dissertation shows how rational self-interested decision makers may exhibit positive self image and investigates the implications of positive self image and related properties for economic behavior.; The first essay is joint work with Joel Sobel. It proposes a mechanism that organizes the evidence on individuals' positive self image. The mechanism assumes that individuals: (1) have an endowment of skills and increase them by making investments; (2) have heterogeneous production functions that determine ability as a function of skills; and (3) make egocentric ability comparisons, that is, use their own production functions to compare their final skills to those of others. Under these conditions, the population demonstrates positive self image in the sense that more than x percent of the population ranks themselves as being better than 100 - x percent of the population. The essay presents conditions that imply a number of stylized observations from psychology and makes several predictions that have not been tested.; The second essay shows that positive self image alters the standard equilibrium predictions of economic theory regarding the design of incentives in organizations. The essay shows that workers' mistaken beliefs of their peers' abilities make interdependent incentive schemes more attractive to firms than individualistic ones. The essay also shows that if positive self image and effort are complements, then a firm's welfare is higher in a rank-order tournament where workers have positive self images than in a rank-order tournament where workers are accurate.; The third essay analyzes the implications of asymmetries in information processing in individuals' decisions and welfare. The essay presents a Bayesian decision-theoretical model of information processing and defines what it means for one individual to be more sensitive to positive information than another. The essay links this definition to a partial ordering over decision problems. Individuals who are sensitive to positive information prefer decision problems with more variable incremental returns in favorable states of nature.