ESTIMATION OF RETURN VOLATILITY OF THE SET50 INDEX FUTURES USING GARCH-X MODEL

Abstract

This study focuses on GARCH-X techniques applied to predict the direction of SET50 index futures in order to estimate volatility of the returns. This will be beneficial for reducing the risk of changes in price index. In this study, the daily closing price of the index, the SET50 index futures, and daily net trading of foreign, institutional and local investors from January 2, 2008 to June 28, 2012 were analyzed. The total number of the data was 1,172 for all groups. This study comprises two parts. The first part is the study of relationships of the data movement. The return value of the SET50 Index Futures and net purchase of foreign investors, institutional investor and local investor were evaluated to create a conditional variance model. After constructing an appropriate model, it was used as a tool to compare the predicted data with the actual value. Then the trustworthiness of predicted data was set. From the study of relationships of the direction of the SET50 index futures rate of return, after the consistency stationary test of unit root at level had been done, the result showed that there were some trend and changeable features appearing in the graph. Hence, the data were non-stationary. However, when the movement of the net purchase by foreign investors, institutional investor and local investor were considered, the data were stable at level. Thus these data could be used to predict ARMA to make the most desirable model. For model selection done by choosing from the lowest AIC and SIC value, it was found that the most reasonable models were AR(1), MA(1) and MA(2). When the models were, next, evaluated by GARCH-X techniques, the SET50 index futures return rate significantly depended not only on the variance in the past, but it also depended on the net foreign institutional and local investors statistics.