Pricing Derivatives: Implementing Heston and Nandi's (2000) Model on the Swedish Stock Index

Abstract: This thesis is based on Heston and Nandi’s (2000) paper. The aim is to check how their closed-form discrete-time GARCH option pricing model performs on Swedish data, and if there are any significant changes to its performance when estimating it via maximum likelihood using the Normal- and the Student-t distribution. The model is compared with the Black-Scholes- and the ad hoc Black-Scholes model of Dumas, Fleming, and Whaley (1998). The results show that when the model is estimated under the Student-t distribution, its out-of-sample valuation performance increases in general when its parameters are updated. However, it is also shown that the model suffers from significant mispricing errors as it is greatly outperformed by both the BS- and the ad hoc BS model. This error is caused by poor estimates of two of its most important parameters, namely the volatility of volatility- and the skewness parameter. It is also shown that the model generally performs better when it is estimated using the Normal- than the Student-t distribution.