Markov Regime-Switching Asymmetric GARCH model and Evaluation of TXO

• Main Authors: Lu Yen Chen, 陳律延
• Other Authors: Y. W. Shyu
• Format: Others
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/00878642324285446502

碩士 === 長庚大學 === 企業管理研究所 === 97 === In this paper we compare the classic Generalized Autoregressive Conditional Heteroscedasticity models with the ones linked Markov Regime switching model together in terms of their goodness of fit of TAIEX and forecastability of price of TXO by B-S model. Not only normal but also fat-tailed leptokurtic conditional distributions for the innovations are assumed, and the degrees of freedom can switch between the different regimes to draw time-varying kurtosis. The goodness of fit of the competing models are evaluated with the value of maximum likelihood function, AIC, and SBC. However, the forecasting performances of them are measured by the statistical loss functions. To obtain an official outcome on statistic, we apply nonparametric Wilcoxon signed rank test to ensure some model is significantly better than the others. The empirical result demonstrates that MRS-GARCH family do overall outperform GARCH family in fitting TAIEX, forecasting volatility, and reducing the inaccuracy of the evaluation of TXO in B-S option pricing model.