A Real-Time GARCH-MIDAS model

This paper proposes the Real-Time GARCH-MIDAS model to model and forecast volatility. An empirical application to the Shanghai Stock Exchange Composite Index (SSEC) and Shenzhen Stock Exchange Component Index (SZSEC) of China shows that the Real-Time GARCH-MIDAS model outperforms competing models in terms of both empirical return fitting and out-of-sample volatility forecasting. Moreover, the superior forecasting performance of the Real-Time GARCH-MIDAS model is robust to alternative rolling windows, alternative benchmark models, alternative MIDAS lags and alternative volatility proxy. Further discussion illustrates the flexibility of the Real-Time GARCH-MIDAS model.