Essays in Empirical Asset Pricing

This dissertation consists of three essays in empirical asset pricing. In the first essay, I propose a machine learning based approach to monitor the relative forecasting performance between two forecasts and select the conditionally better forecast. When I apply this ap- proach to the combination forecast and the historical average benchmark forecast, the re- sulting new return predictor leads to statistically and economically significant out-of-sample gains consistently over time. Such improvements come from expecting a conditionally poor performance of the combination forecast and switching to the historical average benchmark. This approach also works well for other return forecasts using individual economic predictors and can be applied to combine individual forecasts more efficiently. Interestingly, the weight on the combination forecast produced by the machine learning based monitoring approach is high during NBER recessions and periods with high macro uncertainty, which captures the well-known fact that return predictability is concentrated in bad times. In the second essay, we compute implied dividend yields using equity options and show that they are negatively related to the subsequent stock returns. This finding is in contrast with the theory and evidence at the market level where dividend yield is positively related to the future market return. The panel data analysis reveals that the normal relation between the dividend yield and individual stock returns recovers in longer horizon. I further find that the mixed evidence regarding option implied skewness and stock returns could also be reconciled with varying forecast horizon. The opposite to theory relation between option implied measures and stock returns is stronger when analyst forecast dispersion is at a higher level. In the third essay, we investigate the relationship between systematic risk and credit default swap (CDS) returns and discovers that cross-sectional dispersion in future CDS returns can be rationalized by differences in firm’s sensitivities to the market return. Further analysis shows that investors in the CDS market demand higher compensation to provide default protection to firms with higher sensitivities to downside market movements. The reward for bearing downside risk is not simply the compensation for systematic risk nor is it explained by other firm characteristics. The relation between downside risk and CDS returns is stronger for longer maturity CDS contracts.