Which fear index matters for predicting US stock market volatilities: Text-counts or option based measurement?

Abstract

Although VIX has long been recognized as an index to measure fear sentiment in US stock markets, a set of similar measurements called Equity Market Volatility (EMV) trackers are newly created based on the text-counts of newspaper articles including several keywords related to US economy or stock market volatility. In this paper, we use GARCH-MIDAS method to quantify the in-sample explanatory and out-of-sample predictive powers of these two kinds of fear indices in US stock markets. Our empirical results show that VIX has larger in-sample impacts on US stock market volatility than EMV trackers. However, the out-of-sample volatility predictive performances of EMV trackers are generally superior to VIX across different US stock indices and prediction time horizons. In addition, policy-related EMV tracker acts better than VIX and other EMV trackers in predicting volatilities of US stock markets.