Hi all,
What is the state-of-art method for incorporating xregressors in variance equations in GARCH models? So far, I have encountered BEKK models, but they impose severe sign restrictions on the impact of regressors, and EGARCH models of various flavors (i.e. constant and time-varying correlations, etc.) that have been discussed here on the forum.
Is there any other model type out there that you are aware of that is able of accommodating xregressors, with no restriction on xregressor coefficient sign, while guaranteeing positive-definiteness of the conditional covariance matrix?
Thanks,
Marin