Modelling Volatility Cycles: the (MF)^2 GARCH Model

We propose a multiplicative factor multi frequency component GARCH model which exploits the empirical fact that daily standardized forecast errors of one-component models behave counter-cyclical when averaged at lower frequency. For new model, we derive unconditional variance returns, news impact function and multi-step-ahead volatility forecasts. When applied to S&P 500, significantly outperforms nested GJR-GARCH log-HAR in terms out-of-sample forecasting.