| In this dissertation I explore the role of beliefs in affecting behavior in three contexts. The first is a laboratory experiment that identifies patterns of optimistic belief formation through manipulation of financial stakes. The second complements the first, by examining how individuals update beliefs in response to new information. The third and final context is a lab in the field experiment that documents the role of communication in altering beliefs, leading to behavioral changes in experimental games in Rwanda.;The first chapter examines belief bias, through an experimental analysis that is able to shed light on patterns of optimistic belief formation and on the theories that underlie these beliefs. I conduct an experiment examining beliefs about binary events with financial stakes. By varying financial prizes in outcomes, as well as incentive payments for accuracy, the experiment is able to distinguish between two leading theories of optimistic belief formation that differ in their assumptions about how such beliefs are constrained. Existing theories either model beliefs as being constrained through the future costs of holding incorrect beliefs, or through mental costs of distorting reality. The experimental results support the hypothesis that people hold optimistically biased beliefs, and the comparative statics indicate that these beliefs are not constrained by increasing the costs of making inaccurate judgments.;The second chapter studies information processing across different domains. Bayesian updating remains the benchmark for dynamic modeling within economics, however recent evidence suggests individuals may process information differently when it relates to desirable characteristics such as intelligence. Using the empirical framework of Mobius et al. (2013) which nests the Bayesian benchmark model, I am able to investigate whether updating is asymmetric, with positive signals receiving more weight. I find updating appears to be well approximated by the Bayesian model, and that this is stable across domains. In particular I do not find that positive signals receive more weight, contrary to recent evidence.;The third chapter explores the role of communication between individuals in Rwanda in altering beliefs and subsequent behavior in experimental games. In a sequence of public goods games conducted in Rwandan villages I study a natural experiment that generated variation in the order of implementation. Due to geographic variation in village locations, some villages were more likely to receive information more quickly than others. I document how communication changed contribution rates, using this idiosyncratic variation across villages as exogenous variation. I provide evidence that communication likely took the form of advice, and led to an overall 10% increase in contribution rates in the full sample, with heterogeneous effects as high as 45% for certain individuals. These considerations exist in a context where, theoretically, communication should have no ex-ante effects. |
