A Regime-Switching GARCH-MIDAS Approach to Modelling Stock Market Volatility

This paper investigates macroeconomic effects on the volatility of the S&P500 stock market index in the GARCH-MIDAS setup of Engle et al. (2013). This setup is extended to a two-regime Markov-switching model. The results indicate a significant effect of macroeconomic variables on volatility. There are four variables leading volatility: real consumption, consumer sentiment, housing starts and the term spread. Implementation of the Markov-switching extensions shows that a second regime primarily corresponds to “outliers” in volatility. Descriptively it is on par with the single-regime GARCH-MIDAS in the 2000-2010 period, but in terms of forecasting the regime-switching model is inferior, with decreasing accuracy as the forecast horizon increases. This suggests that regime information is sufficiently contained in the macroeconomic variables, such that explicitly accounting for regimes becomes unnecessary when considering a diversified stock index for a developed country.