Two essays on stock market anomalies
Metadata[+] Show full item record
[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] In this dissertation, I consider two topics in stock market anomalies, i.e., cross- sectional patterns in stock returns that cannot be explained by traditional asset pricing models. Building on a comprehensive sample of anomalies, I provide new insights on characterizing common features among anomalies and understanding the sources of abnormal returns. The first essay focuses on the performance of volatility managed version of market anomalies. Recent studies find that volatility managed strategies (i.e., long-short portfolios with investment positions scaled by lagged volatility) exhibit substantial improvements in alphas and Sharpe ratios. They document this effect for the momentum, market, and several other asset pricing factors. Using a sample of 90 anomalies, I show that the superior performance of volatility managed portfolios is primarily driven by downside volatility. Strategies that scale returns by lagged downside volatility produce significantly larger alphas not only relative to the original portfolios but also to those scaled by total volatility. A decomposition indicates that this incremental performance arises from both volatility timing and return timing. The enhanced abnormal returns of downside volatility managed portfolios pose a considerable challenge to traditional asset pricing models. The second essay conducts a comprehensive investigation of the performance of 90 asset pricing anomalies relative to the conditional CAPM. For most strategies, I find that conditional betas display meaningful time-series variation. All but four anomalies have conditional alphas smaller than their unconditional counterparts, and conditioning leads to significant reductions in abnormal returns for 60% of the strategies considered. I conclude that although the conditional CAPM does not fully resolve cross-sectional anomalies, properly specifying conditioning information is critical in evaluating their performance.
Access to files is limited to the University of Missouri--Columbia.