Improving GARCH Volatility Forecasts

Many researchers use GARCH models to generate volatility forecasts. We show, however, that such forecasts are too variable. To correct for this, we extend the GARCH model by distinguishing two regimes with different volatility levels. GARCH effects are allowed within each regime, so that our model generalizes existing regime-switching models that allow for ARCH terms only. The empirical application on U.S. dollar exchange rates shows that our model indeed yields better volatility forecasts than single-regime GARCH and that the allowance for GARCH terms besides ARCH terms can be crucial for the forecast quality.