Flight to Safety: Does It Matter for Volatility Forecasting?
63 Pages Posted: 9 Mar 2018
Date Written: September 30, 2016
We propose the use of a risk measure built on flight-to-safety (FTS) episodes into a volatility forecasting model. We assign to each day in the sample a probability of being a FTS day after observing (ab)normal movements in the US equity, US bond and gold markets. By allowing each FTS day to be an indicator of higher future volatility, we document statistically significant improvements in the accuracy of the 1-day ahead forecasts for the realized variance of the S&P 500 equity index. Superior model performance is found over some of the most common volatility forecasting models proposed in the literature that do not specifically account for these, generally short-lived, market stress episodes.
Keywords: Flight to safety, safe haven, gold, realized variance, volatility forecasting.
JEL Classification: C22, C53, C58, G17
Suggested Citation: Suggested Citation