Essays on the cross-sectional predictability of stock returns

Mihai B Ion, Purdue University

Abstract

I use several recent methodological advances in the field of empirical asset pricing to analyze the expected (as opposed to realized) returns associated with twelve well cited asset pricing anomalies. First, I jointly estimate the expected equity premium and the time-varying market betas of anomaly portfolios in a conditional CAPM framework and I find that nine of the twelve anomaly spread portfolios exhibit significantly higher conditional market betas in periods when the equity premium is at its highest. Second, using a regime switching model with time varying transition probabilities to estimate the joint distribution of returns on the long and short portfolio for each anomaly, I find that the spread portfolios of seven out of the twelve anomalies have significantly higher expected returns in states of the world characterized by high volatility in stock returns. Finally, using a method of estimating expected rates of capital gain using dividend growth rates, I decompose expected returns into expected dividend growth rates and expected dividend yields and find that expected returns on anomaly spread portfolios are almost entirely due to expected dividend growth rates. I interpret my findings as being supportive of the hypothesis that a significant portion of the cross-sectional variation in average returns associated with each anomaly is attributable to variation in risk.

Degree

Ph.D.

Advisors

Gulen, Purdue University.

Subject Area

Finance

COinS