The class of Multivariate BiLinear GARCH (MBL-GARCH) models is
proposed and its statistical properties are investigated. The model can be regarded as a
generalization to a multivariate setting of the univariate BL-GARCH model proposed
by Storti and Vitale (Stat Methods Appl 12:19–40, 2003a; Comput Stat 18:387–400,
2003b). It is shownhowMBL-GARCHmodels allowto account for asymmetric effects
in both conditional variances and correlations. An EM algorithm for the maximum
likelihood estimation of the model parameters is derived. Furthermore, in order to test
for the appropriateness of the conditional variance and covariance specifications, a
set of robust conditional moments test statistics are defined. Finally, the effectiveness
of MBL-GARCH models in a risk management setting is assessed by means of an
application to the estimation of the optimal hedge ratio in futures hedging