NEW EVIDENCE ON THE INFORMATION CONTENT OF IMPLIED VOLATILITY OF S&P 500: MODEL-FREE VERSUS MODEL-BASED

This paper provides new evidence to compare the information content of model-free implied volatility (MFIV) and model-based volatility for forecasting future volatility of the S&P 500. We choose Black and Scholes (BS) implied volatility as our model-based volatility and VIX as our measure of MFIV. By using non-overlapping monthly samples from January 2004 to June 2019, we find that both BS implied volatility and MFIV are informationally efficient and subsume information contained in the historical realized volatility for forecasting future volatility. This is the first study show that BS implied volatility and MFIV contain the same information and there is no winner for forecasting future volatility. This implied that a forecast model could include both BS implied volatility and MFIV