Essays on Empirical Asset Pricing

This thesis concerns the empirical relation between risk and return in equities. It studies why the expected return on stocks as a whole varies over time and why there are predictable cross-sectional di↵erences in the return on individual stocks. The thesis consists of three chapters which can be read independently. The first chapter addresses why the expected return on the market portfolio varies over time. The market portfolio is a claim to all future cash flows earned by the firms in the stock market. I study the expected return to these future cash flows individually. I find that the expected return to the distant-future cash flows increases by more in bad times than the expected return to near-future cash flows does. This new stylized fact is important for understanding why the expected return on the market portfolio as a whole varies over time. In addition, it has strong implications for which economic model that drives the return to stocks. Indeed, I find that none of the canonical asset pricing models can explain this new stylized fact while also explaining the previously documented facts about stock returns. The second chapter, called Conditional Risk, studies how the expected return on individual stocks is influenced by the fact that their riskiness varies over time. We introduce a new ”conditional-risk factor”, which is a simple method for determining how much of the expected return to individual stocks that can be explained by time variation in their market risk, i.e. market betas. Using this new factor, we find that around 20% of the cross-sectional variation in expected stock returns worldwide can be explained by such time variation in market betas. The third chapter studies why stocks with low market betas have high risk-adjusted returns. To shed light on this low-risk e↵ect, we decompose all stocks’ market betas into their volatility and their correlation with the market portfolio. We find that both stocks with lower volatility and stocks with lower correlation have higher risk-adjusted returns. The last fact, that stocks with low correlation have high risk-adjusted returns, is particularly important because it helps distinguish between competing theories of the low-risk e↵ect. Indeed, the high risk-adjusted returns to low-correlation stocks are consistent with leverage based theories of the low-risk e↵ect, but it is not immediately implied by competing behavioral theories we consider in the paper.