“…As a matter of fact, Cipollini et al (2013) show that there is a definite improvement in the forecasting performance of the (multivariate) modeling of realized volatility over realized variance. Finally, the HARQ model has been recently extended to the multivariate case by Bollerslev et al (2018). The challenge would be to develop a similar exercise, dealing with multivariate MEM (Cipollini et al, 2013;Taylor and Xu, 2017) with and without the presence of Markov Switching dynamics in the time-varying covariance matrix (see, for example, Otranto, 2010).…”