Forecasting volatility for financial markets: An application of MIDAS Regression

This study presents a holistic approach to volatility forecasting. It covers a wide range of definitions, concepts and alternative estimation methods related to volatility forecasting. Not only evaluating the literature in a broader perspective in volatility forecasting, it also presents more detailed results on stock market indices. The study adopted a comparatively new technique named MIDAS (Mixed Data Sampling) regression as an alternative to commonly used GARCH methods. With that purpose, I studied 8 developed and 7 developing country's stock markets for the 2008 financial crisis period. I evaluated the one month out-of-sample volatility forecast performance of MIDAS, GARCH and EGARCH regression models. Our results suggest that MIDAS produce superior forecast performance compared to GARCH and EGARCH models. MIDAS model can be a sophisticated tool for researchers and analysts to forecast future volatility among other research topics, with its ability to process mixed-frequency data.