A volatility model must be able to forecast volatility even in extreme situations. Thus, the main objective of this paper, and due to the most recent increase in international stock markets' volatility, is to check which one of the most popular autoregressive conditional heteroskedasticity models (GARCH, GJR, EGARCH or APARCH) is more able to predict the extreme volatility in 2008 considering the daily returns of eight major international stock market indexes: CAC 40 (France), DAX 30 (Germany), FTSE 100 (UK), NIKKEI 225 (Japan), HANG SENG (Hong Kong), NASDAQ 100, DJIA and S&P 500 (United States). Goodness-of-fit measures demonstrate that EGARCH and APARCH models are able to correctly fit the conditional heteroskedasticity dynamics of the return's series under study. In terms of volatility forecast comparisons, using the Harvey-Newbold test for multiple forecasts encompassing and the ranking of forecasts based on the coefficient of determination (R-2) resulting from the Mincer-Zarnowitz regression, we conclude that EGARCH dominates competing standard asymmetric models.

--

Forecasting volatility,EGARCH,APARCH,GJR

• Mathematics - Natural Sciences
• Economics and Business - Social Sciences