Three essays on empirical asset pricing

The dissertation consists of three essays on empirical asset pricing, each being one chapter of this dissertation.;In the first chapter, I study an options trading strategy known as dispersion strategy to investigate the apparent risk premium for bearing correlation risk in the options market. Previous studies have attributed the profits to dispersion trading to the correlation risk premium embedded in index options. The natural alternative hypothesis argues that the profitability results from option market inefficiency. Institutional changes in the options market in late 1999 and 2000 provide a natural experiment to distinguish between these hypotheses. This provides evidence supporting the market inefficiency hypothesis and against the risk-based hypothesis since a fundamental market risk premium should not change as the market structure changes.;The second chapter uses an anticipated volatility increasing event, the stock split, to investigate the informational content in implied volatility. We find that implied volatility jumps sharply at the announcement date and continues to rise gradually to the split ex-date, as predicted by standard option pricing theory. In general, we find that post-announcement implied volatility provides informative forecasts of future realized stock volatility over the announcement to post-split period, implying that the change of stock volatility at the ex-date is priced into option's prices after the announcement. In addition, a regime shift in the forecast performance happens around 2000, before which implied volatilities are more biased and less efficient.;The third chapter estimates the conditional variance of daily stock returns using an extended GARCH model with event-related dummy variables to capture the predictable components of volatility change, such as earnings announcements, macroeconomic announcements, day-of-the-week effects, etc. We examine the out-of-sample forecasting ability and find this model provides a better performance compared to the usual GARCH(1,1) volatility model. In addition, we find that the dependence on the random components increases after we include the predictable components. This implies that modeling volatilities using only past returns without other predictable variables could underestimate the persistence levels of volatilities and thus bias the volatility forecasts, especially those over long horizons.