Combining Stock Market Volatility Forecasts with Analysis of Stock Materials under Regime Switching

Abstract

Forecasting stock market volatility is an important and challenging task for both academic researchers and business practitioners. The recent trend to improve the prediction accuracy is to combine individual forecasts using a simple average or weighted average where the weight reflects the inverse of the prediction error. However, a problem in the existing forecast combination methods is that the weights remain fixed over time. This may prove inadequate, especially in economics data, where changes in policy regimes may induce structural change in the pattern of forecast errors of the different models, thereby altering the relative effectiveness of each model over time. In this paper, we present a new forecast combination approach where the forecasting results of the Generalized Autoregressive Conditional Heteroskedastic (GARCH), the Exponential GARCH (EGARCH), stochastic volatility(SV), and Moving average(MAV) models are combined based on time-varying weights that can be driven by regime switching in a latent state variable. The results of an empirical study indicate that the proposed method has a better accuracy than the GARCH, EGARCH, SV and MAV models, and also combining forecast methods with constant weights.