The thesis contains three essays. These essays investigate the market pricing of the derivative warrants and stock options by using a semi-parametric method, which couples a nonparametric technique with a parametric option pricing model. The method proposed in our study can significantly improve the accuracy of option pricing. Moreover, our analysis brings a better understanding of the pricing behavior in the derivative warrants and stock options markets....[ Read more ]

The thesis contains three essays. These essays investigate the market pricing of the derivative warrants and stock options by using a semi-parametric method, which couples a nonparametric technique with a parametric option pricing model. The method proposed in our study can significantly improve the accuracy of option pricing. Moreover, our analysis brings a better understanding of the pricing behavior in the derivative warrants and stock options markets.

The first essay is entitled "Semi-parametric Pricing of Derivative Warrants". In this essay we develop a pricing method based on the Black-Scholes option pricing model with a nonlinear correction to account for differences in contract specifications. The nonlinear correction is based on the local linear kernel regression technique with time to maturity of the warrant, moneyness of the warrant and volatility of the underlying stock as the regressors. The derivative warrants written on the HSBC common stock traded in the Stock Exchange of Hong Kong are used as the data sample. Our semi-parametric approach is found to substantially improve the model's ability to describe the market pricing structure of derivative warrants. The performance improvement due to the nonlinear correction is significant in the in-sample analysis and remains stable when moving to out-sample applications. In addition, we find consistency in the pricing behavior across warrants issued at different times and by different financial institutions. Specifically, we find no evidence that the identity of warrant issuer can distort the pricing of derivative warrants.

The second essay is entitled "Pricing of Stock Options Using Derivative Warrants on the Same Underlying Asset". In this essay we compare the pricing behaviors between derivative warrants and stock options. Both stock options and derivative warrants written on the same underlying asset are traded concurrently in Hong Kong. These two categories of contracts differ in their trading mechanisms and contract specifications even though both of them are fundamentally American style options. There is a common belief, based upon cursory evidence on average implied volatilities, that derivative warrants are more expensive, perhaps due to the short-sale restriction placed on these warrants. Given our knowledge of smile/smirk phenomenon in option prices, such an observation clearly fails to take into account the differences in contract specifications. We employ the semi-parametric pricing technique developed in Essay One to price the stock options relative to a pool of derivative warrants as a way of correcting the differences in contract specifications. Our findings suggest a biased prediction in the direction consistent with overpricing of derivative warrants. We formally examine this issue by adding a parameter to the historical volatility in the semi-parametric pricing function. Our results indicate that this constant value is negative and statistically significant, which in turn implies that stock options are relatively cheaper after controlling for the differences in contract specifications. In addition, we find the divergent pricing behaviors between derivative warrants and stock options tend to disappear toward the end of the sample period.

The last essay is entitled "Empirical Performance of Derivative Warrants Pricing - Parametric vs. Semi-parametric Methods". In this essay, we compare the performance of a more flexible parametric model - GARCH option pricing model vis-a-vis the semi-parametric method in pricing derivative warrants. Since all the HSBC derivative warrant contracts in our sample are of American type, we use the numerical method of Duan and Simonato (2000) to price the derivative warrants in the GARCH framework. Our results indicate that the GARCH option pricing model performs as well as the semi-parametric method constructed around the Black-Scholes model. Moreover, we reconsider the semi-parametric method developed in Essay One by replacing the Black-Scholes model with the GARCH option pricing model. Our results show that the non-parametric correction based on a local linear kernel regression can not improve the pricing biases under the GARCH model.