Three Essays On Option Pricing PDF Download

This dissertation contains three essays on observable covariates in option pricing. In the first essay, I propose firm-specific public news arrival from Factiva database as an observable covariate in equity options market and study how the public news arrival is priced. I first establish the empirical relationship between the firm-specific public news arrival and jumps in individual equity returns. Subsequently, I build a continuous-time stochastic volatility jump diffusion model where news arrivals driving the jump dynamics. When estimated on equity options data for 20 individual firms, the premia placed on jump frequency and size turn out to be consistent with the theories highlighting both positive and negative effects of public news arrival. The second essay, based on a joint work with Peter Christoffersen, Bruno Feunou and Chayawat Ornthanalai, studies how the stock market illiquidity affects the market crash risk. Our empirical approach is to estimate a continuous-time model with stochastic volatility and dynamic crash probability where stock market illiquidity is used as an observable covariate driving the crash probability. While the crash probability is time-varying, its dynamic depends only weakly on return variance once we include market illiquidity as an economic variable in the model. This finding suggests that the relationship between variance and jump risk found in the literature is largely due to their common exposure to market illiquidity. Our study highlights the importance of equity market frictions in index return dynamics and explains why prior studies find that crash risk increases with market uncertainty level. The third essay, based on a joint work with Peter Christoffersen and Bruno Feunou, proposes the realized jump variation measure constructed from the intraday S returns data as an observable covariate that helps pricing of index options. The volatility and jump intensity dynamics in the model are directly driven by model-free empirical measures of diffusive volatility and jump variation. Because the empirical measures are observed in discrete intervals, our option valuation model is cast in discrete time, allowing for straightforward filtering and estimation of the model. When estimated on S index options and returns the new model performs well compared with standard benchmarks.