Dynamic Behavioral Responses and Resulting Biases

Abstract

I develop theoretical models of behavioral responses to information and network conditions that are determined endogenously in order to analyze how certain biases may appear in equilibrium. Chapter 1 constructs a model of endogenous network creation and the spread of information to dynamically determine how certain ideologies may become endemic within a population. One half of the model handles population dynamics through proportional mixing and the other half handles individuals' choices around participation in the network based on the utility derived from interacting with types of other agents. I find that, paradoxically, more unpleasant ideas like conspiracy theories have a greater likelihood of becoming endemic within a population. Chapter 2 considers overconfidence among agents in a tournament setting, as well as optimal tournament design for a principal maximizing profit. I posit a model of component skills that influence overall performance, which is the ultimate basis for judgment between individuals. By overestimating the importance of the skills in which they are talented, participants overexert effort relative to a fully rational benchmark. This allows a principal to manipulate tournament parameters to increase effort exertion, which increases profits, without promising additional rewards. Chapter 3 analyzes a model of sequential decisions under incomplete recall to produce certain observed biases as optimal responses for a decision-maker in this the setting. By limiting the decision-maker's ability to evaluate past choices to only the realized outcomes, they gradually become biased toward repeating those choices, even if a full history of signals would suggest otherwise. This can even lead to complete disregard of new information in favor of the status quo.