Essays on Disaster Risk and Equity Return Predictability

This dissertation consists of two essays on disaster risk and equity return predictability. The first essay proposes new measures of firm-level and market level disaster risk from deviation of put-call symmetry, which is free from being contaminated by the asymmetry between option traders and equity investors. Compared with other known measures of disaster risk, the market-level disaster risk measure robustly predicts aggregate market returns, with out-of-sample (R^2=6.86%) for the next twelve months. The cross-sectional analysis shows that firm-level disaster risk also explains variations in expected stock returns. Stocks with high firm-level disaster risk earn an annual four-factor subsequent alpha 8.0% higher than stocks with low firm-level disaster risk. I explore potential mechanisms giving rise to these asset pricing facts.

The second essay finds that the investor’s learning of higher moments can account for the time-variation, size, and volatility of equity premium. I estimate the investor’s belief on skewness and kurtosis of consumption and dividend growth, and assume investor’s Bayesian learning about a skew student’s t-distribution with unknown fixed parameters. The predictive regressions show that more negative skewness and higher kurtosis predict higher subsequent market excess returns, which implies the investor’s learning generates the time variation of equity premium although the true distribution is static. The calibrated asset pricing model shows that the investor’s learning also explains the size and volatility of the equity premium observed in the data when the investor has a preference for early resolution of uncertainty.