Combination forecast based on financial stress categories for global equity market volatility: the evidence during the COVID-19 and the global financial crisis periods

Abstract

The 2008 global financial crisis and the COVID-19 pandemic both decrease economic growth and lead to high uncertainty in global stock markets, and financial stress information is closely linked to financial crises. To improve the predictability of the realized volatility of the global equity indices during crises, we examine the predictive role of the Global Financial Stress Index (GFSI) and its categories. We find that the combination predictions based on GFSI’s five incorporated categories and three region-based categories outperform the predictions based on the raw GFSI for most indices. Specifically, the DMSPE combination model with a low discount factor has accurate forecasts for 5- and 22-day-ahead realized volatility, and it also performs better than the equal-weighted and the trimmed mean combination methods. In this study, we present a comprehensive analysis of the predictive role of financial stress information in stock market volatility during crises, and the empirical evidence provides a positive case against the ‘forecast combination puzzle’. Our findings are very instructive for policymakers and investors to make their own short-term and long-term plans in crisis.