Forecasting volatility and Value-at-Risk of the Karachi Stock Exchange 100 index: Comparing distribution-type and asymmetry-type models

Abstract

We investigate the daily volatility and Value-at-Risk (VaR) forecasts for the Karachi Stock Exchange 100 Index (KSE-100) series from 1998 to 2008. The forecasting performance of the distribution-type volatility models (GARCH-N, -t, -SGT, and -HT) are compared with that of asymmetry-type models (GJR-GARCH and EGARCH) in order to ascertain the crucial determinants for improving forecast accuracy of daily volatility and VaR. Empirical results indicate that the GARCH-HT and GARCH-SGT models generate far more accurate daily volatility forecasts as compared to their competitors. For VaR calculation, the GARCH-t and GARCH-SGT are the appropriate models to predict the daily VaRs of KSE-100 stock index under high confidence level.