Can risk-neutral skewness and kurtosis subsume the information content of historical jumps?

Abstract

We examine the relation between jump variations and risk-neutral moments in volatility forecasting. We propose a method that involves no extrapolation in computing the risk-neutral moments of Bakshi et al. (2003) and document that risk-neutral skewness and kurtosis subsume the information content of historical jumps. While historical jumps have significant explanatory power for future volatility and such power is actually not weakened by the inclusion of risk-neutral volatility in models, their predictability does disappear when risk-neutral skewness and kurtosis are included.