Estimating and Forecasting Volatility of the Malaysian Stock Market Using a Combination of Kalman Filter and GARCH Models

Abstract

Abstract: The Kuala Lumpur Composite Index plays an important role as an indicator to the growth of investment in share equity and economic development in Malaysia. It has been an area of interest to investigate how changes in volatility affect share prices. The Kalman filter is an algorithm for sequentially updating a linear projection for the system. Since its development by Kalman and Bucy in the 1960s, the applications of the Kalman filter technique in financial time series have been very few and far in between. This study aims to investigate the nature and behaviour of the volatility of the Kuala Lumpur Composite Index return using a combination of Kalman filter and ARCH-type models. Three ARCH-type models (GARCH, GARCH-in-Mean and Exponential-GARCH) were considered. Monthly data from May 1986 to February 2005 were used. In general, the results strongly indicate the unsuitability of the constant variance Kalman filter model, suggesting the importance for modeling time varying volatility in the return series. When ARCH structure was taken into account in the conditional variance of the Kalman filter framework, the model was found to provide better results over the pure Kalman filter model. Further analysis also revealed that a combination of Kalman filter and ARCH-type models better fitted the KLCI series than the simple ARCH-type models. In conclusion, the Kalman filter model with time-varying conditional variance can better capture the behaviour of return series and successfully model the changing of variances.