Modelling and Forecasting the Volatility of Stock Price Index : ASEAN-5 CountriesTools Hoh, Chai Ping (2014) Modelling and Forecasting the Volatility of Stock Price Index : ASEAN-5 Countries. [Dissertation (University of Nottingham only)] (Unpublished)
AbstractThis study aims to model and forecast the stock index volatility in ASEAN-5 countries. In addition, this study investigates the asymmetric characteristics of the returns series, evaluate if the use of non-normal distribution improves the model estimation and compare the forecasting performance of the GARCH-type models with conventional models for the five countries. The volatility is estimated by employing three GARCH-type models, GARCH(1,1), EGARCH and GJR-GARCH on the daily returns data. To increase the robustness of the models, we apply three different distributions; normal, Student-t and Generalized Error Distribution. The serial correlation and heteroscedasticity hypothesis in the data series is verified by using the Ljung-Box Q-statistic and Lagrange Multiplier tests. The in-sample estimation was evaluated by three information criterion measures while the out-of-sample forecasting performance was evaluated based on three symmetric statistical loss functions and an asymmetric loss function. Consistent with previous studies on emerging markets, the empirical results for this study indicate strong evidence of asymmetry in the ASEAN-5 stock markets. The study also demonstrated that the application of non-normal distributions to the models enhanced the goodness-of-fit for the model estimation. Based on the in-sample estimation, the best fitted model for all five countries is the asymmetric models with non-normal distributions. The empirical results for the out-of-sample forecasting performance however did not demonstrate evidence of superior performance for the GARCH models in comparison to conventional models. This study found that for three of the ASEAN countries the forecasting accuracy was dominated by the conventional models. Evidently, from this study, there is no dominant model for modelling and forecasting of volatility. Hence, for robust investment decisions, portfolio management and risk management implementations, it will be prudent to evaluate the purpose of forecasting volatility and applying the best estimation model for volatility forecasting to fit this purpose.
Actions (Archive Staff Only)
|
Tools
Tools
![[img]](/style/images/fileicons/application_pdf.png)
