Forecasting volatility with the VIX index: is it worthwhile?

This paper explores the information content and the forecasting power of the VIX index, computed by CBOE. As a benchmark, the forecasting performance of VIX is compared to the Garch (1;1) model and historical volatility. The total period of 20 years taken into consideration (January 1995-December 2014) is split into two sub-periods, precisely before and after March 2006. This is when the trading of option contracts having as underlying VIX index began. By comparing the two sub-periods, we can judge if the information content of VIX increased after becoming a negotiable asset. The results of the analysis are not clear-cut. The VIX index shows strong information content, but is an upward biased forecast of realized performance. When comparing VIX to Garch and historical volatility, the former is dominant, when the outlier period of the subprime crisis is excluded from the sample. The information content of VIX seems unaffected by the event of becoming the underlying of option contracts.