Joint calibration of VIX and VXX options: does volatility clustering matter?

This paper studies the effects of volatility clustering on the joint calibration of VIX and VXX options. We find that model which incorporates volatility clustering outperforms other models without this feature in joint calibration of VIX and VXX options both in-sample and out-of-sample; the superiority of the model with volatility clustering is statistically significant. Moreover, the information contained in the VXX options is not fully spanned by the VIX options, as a result, one can achieve better joint pricing performance by employing both VIX and VXX derivatives data when calibrating the model, compared to the case when only VIX data are used in calibration.