Model-Free Implied Volatility and Its Information Content

We implement an estimator of the model-free implied volatility derived by Britten-Jones and Neuberger (2000) and investigate its information content in the S&P 500 index options. In contrast to the commonly used Black-Scholes implied volatility, the model-free implied volatility is not based on any specific option pricing model and thus provides a direct test of the informational efficiency of the option market. Our results suggest that the model-free implied volatility is an efficient forecast for future realized volatility and subsumes all information contained in the Black-Scholes implied volatility and past realized volatility. We also find that the model-free implied volatility is, under the log specification, an unbiased forecast for future realized volatility after a constant adjustment. These results are shown to be robust to alternative estimation methods and volatility series over different horizons. There has been considerable debate on the use of the Black-Scholes (B-S, hereafter) implied volatility as an unbiased forecast for future volatility of the underlying asset. Since option prices reflect market participants’ expectations of future movements of the underlying asset, the volatility implied from option prices is widely believed to be informationally superior to historical volatility of the underlying asset. If option markets are informationally efficient and the option pricing model (used to imply the volatility) is correct, implied volatility should subsume the information contained in other variables in explaining future volatility. In other words, implied volatility should be an efficient forecast of future volatility over the horizon spanning the remaining life of the option. This testable hypothesis has been the subject of many empirical studies. Early studies find that implied volatility is a biased and an inefficient forecast of future volatility and it contains little or no incremental information beyond that of historical volatility. For example, Canina and Figlewski (1993) study the daily closing prices of call options on the S&P 100 index from March 15, 1983 through March 28, 1987 and conclude that implied volatility is a poor forecast of subsequent realized volatility on the underlying index. Based on an encompassing regression analysis, they find that implied volatility has virtually no correlation with future return volatility and does not appear to incorporate information contained in historical return volatility. In contrast, several other studies including Day and Lewis (1992), Lamoureux and Lastrapes (1993), Jorion (1995), and Fleming (1998) report evidence supporting the hypothesis that implied volatility has predictive power for future volatility. Like Canina and Figlewski (1993), Day and Lewis (1992) also analyze S&P 100 options but over a longer period from 1983 to 1989. They find that the implied volatility has significant information content for weekly volatility. However, the information content of implied volatility is not necessarily higher than that of standard time series models such as GARCH/EGARCH. Lamoureux and Lastrapes (1993) reach a similar conclusion by