The three chapters of this dissertation are each separate research papers in behavioral economics, though the second and third respectively provide applications to macroeconomics and finance.

The first chapter studies what I call the weighted updating model, which is a generalization of Bayesian updating that allows for biased beliefs by weighting the functions that constitute Bayes' rule with real exponents. I provide an axiomatic development of the model. On the way, I show that weighting a distribution results in a systematic change in information entropy, suggesting that weighted updating can model biases in which individuals mistake the information content of the observations from which they make inferences. I augment the base model in two ways, allowing it to account for additional biases. The first expansion involves discrimination between data. The second allows the weights to vary over time. This chapter also presents a set of sufficient conditions for the uniqueness of parameter estimation through maximum likelihood, with log-concavity playing a key role. An application shows that self attribution bias can lead to optimism bias.

The second chapter builds upon some of the findings in the first chapter. Specifically, I examine the implications of self-attribution bias on consumption and savings decisions. When self-attributive learning replaces rational expectations in a model of intertemporal choice, two departures from the permanent-income hypotheses manifest. One is that consumers tend to under-save early in life. Another is a relatively high degree of covariance between changes in consumption and changes in income. No other factor on its own has been able to explain both of these empirical anomalies that the permanent-income hypothesis has faced.

Chapter three presents the results of an experiment designed to investigate whether order effects are due to the imperfections of memory and whether order effects influence valuations. After experimental participants observed a sequence of six random returns from an asset on a computer screen, I elicited the participants' valuations for a random return from the asset. The subjects were randomly assigned to an experimental group and a control group. The experimental group observed returns that remained on display once they were revealed, while the control group saw each return for only three seconds at a time. The analysis suggests that these groups utilized the observed returns differently in forming their valuations and that order effects do influence valuations. The results also provide evidence that the experimental group exhibited order effects, suggesting that memory is not the only cause of order effects.