Essays on the analysis of cross-sectional stock returns

Abstract

This dissertation consists of three studies on cross-sectional pricing in equity market. The first chapter discusses co-jumps. I use high-frequency stock returns to estimate downside jump beta and upside jump beta, and find that downside jump beta is greater than upside jump beta for most stocks included in Dow Jones 30 Index. I propose a measure for co-jump asymmetry (CJA) and find that stocks with high CJA (more left-skewed co-jumps) significantly outperform their counterpart. The prediction effects of CJA are stronger for moreliquid and less volatile stocks, implying that the relation between CJA and cross-sectional stock returns is driven by risk-return trade-off rather than the behavioral bias of risk-seeking investors. In the second chapter, I adopt the semimartingale framework to extract the forward-looking jump risk factors from option prices, and find the asymmetry in option-implied jump size can predict cross-sectional stock returns. The empirical evidence shows that the predictive power of jump asymmetry is driven by multiple channels: information asymmetry, jackpot hypothesis and short risk premium. I investigate into the relation between salience and cross-sectional stock returns in the third chapter. Investors’ beliefs towards the future pay-offs of a security are always distorted by its most different pay-off from other securities. I construct the salience measure using the distribution of historical stock returns to quantify the distortion in beliefs, and test the hypothesis that stocks with salient upsides underperform their counterpart. The empirical evidence is in favor of the implications of Bordalo et al. (2013), and the crosssectional return predictability is particularly strong among small-cap, volatile and illiquid stocks where less sophisticated investors are more likely to invest in. Moreover, the predictive power of salience is related to fears and recessions. No evidence can show upside salience is positively correlated with option-implied skewness, which implies that option traders are not salience-followers.