Three Essays of Empirical Investigation in Asset Pricing

Abstract

This dissertation is a collection of three essays of empirical investigation in asset pricing identifying three distinct relations in the cross-section of U.S. stock returns. In Chapter 1, I show that the rate of capital gains of the market portfolio is vastly more volatile than the dividend yield. I propose a two-factor CAPM that includes a separate market dividend yield factor and find that this factor carries a significant negative premium in the post-1978 period that coincides with the persistent decline in the number of US dividend-paying firms. This finding is motivated with a theoretical model, which shows that the predictive information of the dividend yield can be high when investors have a behavioural bias against dividends. In Chapter 2, I examine the role of the market state in predicting asset pricing anomalies. I show that the sign, size, and significance of anomaly returns depend crucially on whether they follow a positive or negative market return. I show that the findings can be explained by market frictions for investors with short positions and behavioural biases for investors with long positions. This hypothesis is supported by empirical evidence on short interest, liquidity, and fund flows. In Chapter 3, I introduce an improved measure of earnings surprises to identify Post-Earnings-Announcement-Drift (PEAD). This new measure is based on standardized surprises defined as earnings minus the median of the cross section of analyst forecasts. In contrast to the standard earnings measures, the improved earnings measure is successful in identifying PEAD in the most recent five-year period. The improved measure also outperforms the standard measures throughout the sample in identifying long-short portfolio returns, thus providing an effective way to predict post-earnings-announcement-drift (PEAD). Overall, the findings support the weakening, but not the death, of PEAD in recent years.