| Chapter One analyzes a setting in which the presence of decision makers who are irrationally overconfident reduces trading costs and provides a net benefit to society. Overconfidence is modeled by extending a familiar financial market model to allow some agents' beliefs about 2nd moments to be incorrect. We provide conditions on beliefs under which equilibrium exists. In addition, we show several monotonic relationships relating equilibrium prices and quantities to beliefs. We propose a method for objectively comparing individual well-being under different beliefs and find that, by this measure, overconfidence can be Pareto improving.; In Chapter Two, we develop a probability model for handling data sets with both missing and misclassified survey responses. We provide a technique for estimating the probability that an individual decides to tell the truth, lie, or refuse to respond to a survey question, conditioning on other observed characteristics. Using 1991–1996 General Social Survey data which include questions on sexual orientation, our estimates indicate that factors such as marital status and income play a role in conditioning the probabilities of misclassification and nonresponse. We explore the relationship between individual income and the probability that a homosexual individual reports “heterosexual” on the survey. High-earning male homosexuals tend to respond untruthfully compared with other male homosexuals, while low-earning female homosexuals are more likely to refuse to answer a survey question about sexual orientation.; Finally, in Chapter Three, we analyze the connections between tracking error decision rules, risk taking, and accumulated wealth. There is abundant evidence that typical decision makers care about controlling the difference between their own payoffs and a benchmark such as one's own past payoffs, historical average payoffs, or the payoffs of competitors. Referring to that difference as “tracking error,” we compare decision makers who care about tracking error (in addition to risk and return) against decision makers with standard mean-variance preferences. We analyze average levels of risk taking and wealth accumulations across these two types after calibrating the types' risk aversion to be identical when tracking error is zero. On average, tracking error types take more risk and accumulate more wealth. In an equilibrium, the increased risk taking of tracking error types hurts mean-variance types, but the net effect of the tracking error types is to improve aggregate accumulated wealth over that which would prevail in a world populated exclusively by mean-variance types. |
