| Abstrakt: |
This article focuses on performance of GARCH models in forecasting stock market volatility. Many conventional time series and econometric models work only if the variance is constant. Until lately, the financial and economic researchers have started modeling time variation in second or higher-order moments. R.F. Engle, a researcher, has characterized the changing variances using the Autoregressive Conditional Heteroscedasticity (ARCH) model and its extensions and modifications. Since then, hundreds of researchers have applied these models to financial time series data. The alternative and more flexible lag structure is the generalized ARCH (GARCH) introduced by T. Bollerslev,a researcher. It is proven that a small lag such as GARCH is sufficient to model the variance changing overlong sample periods. The data used in this paper is daily observed stock market indices in the Kuala Lumpur Stock Exchanges including Composite Index, Tins Index, Plantations Index, Properties Index and Finance Index. This data was collected from 1 January, 1989, to 31 December, 1990. This study considers the daily closing prices as the daily observations. The family of GARCH models are estimated using the maximum likelihood method. It was observed that the weaknesses of imposing the parameter estimates of the GARCH model to certain constraints such as stationary or non-negativity. In contrast, the GARCH model has no restriction on the parameters. |
