Essays in empirical finance

Abstract

This thesis consists of three essays on various topics in empirical financial studies. In Chapter 1, I study the profitability of momentum trading from evidence in mutual fund performance. I find that mutual funds that exhibit a strong momentum trading pattern earn significant risk-adjusted returns relative to Fama-French 3-Factor model, and tend to outperform other funds that do not momentum trade as much. The superior performance of these funds persists across different investment objectives as well as after controlling for fund size or fund flow. The robustness of my results suggests that momentum profits are real and momentum trading has the potential to improve a funds return. However, I also find relatively weak evidence of persistence in mutual funds trading styles. In particular, most funds do not seem to maintain their aggressiveness in momentum trading from one year to another. The findings indicate that momentum trading patterns observed in these mutual funds are more likely to be caused by random chances than the managers intention to capture momentum profits. Results in this paper also favor the under-reaction hypothesis as explanation for momentum in stock returns. Chapter 2 is a joint work with Charles Chang and Albert Wang. We explore how financial firms trade on in-house, US equity recommendations. We match the quarterly trades of financial firms with their own recommendations and document their trading patterns before, in the same quarter as, and after issuing recommendations. We find that net trade is more positive around upgrades than downgrades for all periods, and these relations are particularly significant in the quarter of and quarter immediately after the recommendation change. These empirical relations suggest that

(cont.) by and large, financial firms actually do "put their money where their mouths are". Previous studies have found that the execution costs and volatilities of the securities traded in the auction market are generally lower than those of the comparable securities traded in the dealer market. However, due to the difficulty of identifying perfectly matched pairs of stocks, the conclusions drawn from those studies are always subject to the criticism of inadequate control for individual stock characteristics. In Chapter 3, I repeat previous studies of execution costs and volatilities using a sample of stocks that is chosen specifically to address this criticism. The sample is made up of stocks that are listed on both the NASDAQ and AMEX in 2003. Consistent with existing literature, I find that the volatilities are generally higher on the NASDAQ than on the AMEX. On the other hand, the transaction costs are higher on the AMEX, which is at odds with previous empirical studies. The difference in execution costs and volatilities can be partially explained by their different sensitivities to various stock characteristics in the two different markets.