Source: Dissertation Abstracts International, Volume: 67-12, Section: A, page: 4643.

Adviser: Nicholas Barberis.

I investigate how certain investor behaviors affect individual stock return distributions. A common characteristic of these behaviors is that they do not arise directly from the traditional expected utility framework with known normally distributed probability distributions.

In the first chapter of my dissertation, I present evidence of performance chasing behavior in closed-end funds, and study the impact of performance chasing on the time series dynamics and cross-sectional variation of closed-end fund discounts. Prices and discounts seem to be "backward looking" but are not "forward looking". I also offer evidence that performance chasing is not induced by a simple form of parameter uncertainty, where the parameter measures managerial skill.

In the second chapter, I investigate how differences of opinion among traders, together with short-sales constraints, affect stock price reactions to public news announcements. In a parsimonious model two groups of traders differ from each other by a parameter which measures how responsive these groups are to public news announcements. As a result stock returns are positively skewed, and skewness is positively correlated with the degree of differences of opinion. I use measures of dispersion in analysts' earnings forecasts, as well as size, to proxy for differences of opinion in empirical tests of the model. These variables prove to be important cross-sectional predictors of skewness and the average price reaction to earnings news.

While the second chapter asks why individual stocks have positively skewed returns, the third chapter explores the asset pricing implications of skewness. I introduce a novel method to compute expected individual stock return skewness, and present strong evidence that this skewness measure predicts monthly stock returns in the cross-section. Instead of relying on a long history of past self returns, I use recent returns from a peer group to compute skewness for each stock. I group stocks by their industry membership, book-to-market ratio and size, and past return correlation with other stocks, and the skewness measure works in all cases.