Modelling and Forecasting the Volatility of the Central European Stock Market

Abstract

We investigate the nature of the Central European stock market volatility before, during and after major emerging market crises. We analyze the Central European Stock Index over the period April 30, 1996 - May 31, 2002. The data is divided into three sample periods - pre-crisis period, crisis period and post-crisis period. We find significant autocorrelation in return series. The autoregressive process is attributed to both nonsynchrounous trading and asymmetric response to good and bad news. We reported significant leverage effect in conditional variance and high volatility persistence in all considered period. Both asymmetry in conditional volatility and volatility persistence tend to increase in crises periods. We employ two symmetric and six asymmetric GARCH models for in-sample and out-of-sample forecasting. In addition, we apply Engle and Ng (1993) diagnostic tests for news impac. Results lead us to the conclusion that following a financial crisis, the negative return shocks have higher volatility than positive return shocks. We find that asymmetric GARCH model with non-normal distributed residuals capture most of Central European stock market volatility characteristics: (1) asymmetric news impact, (2) volatility persistence and (3) fat-tailed distribution of stock market returns. The asAR(1)-VGARCH (1,1)-t is most appropriate model in case of in-sample forecast while the asAR(1)-NAGARCH (1,1)-t model can be regareded as most appropriate.