Implementation of Heston-Nandi GARCH model on OMXS30

Sammanfattning: This paper evaluates the performance of Heston and Nandi’s closed form option pricing model (2000) on the OMXS30 (Swedish stock index), pre and post the financial crisis. The main purpose is to investigate if the more realistic assumptions of Heston and Nandi yield more accurate price estimates, than the computationally more simplistic Black-Scholes model. Both periods are evaluated in-sample and out-of-sample and the parameters of the model are generated by Maximum Likelihood Estimation. The out-of-sample analysis reveals some mixed results, but put options are in general more accurately estimated than call options, especially out-of-the money. Some periods experience large pricing errors, due to poor parameter estimates. One natural extension would thus be to perform the study by estimating the parameters by the Nonlinear Least Squares method, indeed implemented by Heston and Nandi.