Essays in Empirical Asset Pricing
Analytics
33 views ◎15 downloads ⇓
Abstract
The dissertation is composed of three essays that address the cross-sectional relation between firm characteristics and expected stock returns. Chapter 1, "Regime-switching and the Cross-Section of Expected Stock Returns", incorporates regime switching techniques. Under a two-regime-switching model of stock market returns, the good (bad) regime is characterized by a high (low) market mean return and low (high) market volatility. A simple method is proposed to estimate good- and bad-regime means, volatility, and cross-correlations for a large number of individual stocks. We find that the cross-sectional relation between the bad-regime mean return and the expected stock return is significantly negative, and the relation between the average bad-regime cross-correlation and the expected stock return is significantly positive. The observed relations are consistent with hedging hypothesis that investors want to hedge against market downturns and volatile markets. Furthermore, stocks with high (low) one-step-ahead predicted returns estimated using only bad-regime variables earn substantially high (low) subsequent returns and abnormal returns. Chapter 2, "Short-Term Reversals and Trading Activity", takes the interaction between prior returns and prior trading activities into consideration. Using a sample that excludes micro-cap stocks, we find that short-term reversals in monthly stock returns are strongly linked to prior monthly trading activity. Stocks with low turnover display a pronounced reversal effect, whereas those with high turnover display a continuation effect (momentum). The results are similar if we restrict the sample to large-cap stocks. Our analysis suggests that turnover is linked to short-term autocorrelation patterns in returns because it proxies for the flow of news that spurs speculative trading, and that the likelihood of short-term reversals falls as the proportion of turnover that is driven by news increases. Chapter 3, "Portfolio Sorts via Nonparametric Regression: From B-Splines to Basis Portfolios", nests portfolio sorts within the B-spline regression framework.