Flexible Multivariate GARCH Modeling With an Application to International Stock Markets
We develop an estimation method for the Diagonal Multivariate GARCH model. For a vector of size N unidimensional GARCH processes for the diagonal elements of the conditional covariance matrix, and N(N-1)/2 bivariate GARCH processes for the off-diagonal elements of the conditional covariance matrix. The coefficient matrices are then transformed in such a way that ensures the positive semi-definiteness of the conditional covariance matrix. Under a technical assumption, the estimator has the same asymptotic properties as the univariate and bivariate maximum likelihood GARCH estimators. The method is computationally feasible for large problems, of size N=100 or larger. We do not need to impose any particular simplifying structure on the coefficient matrices. The conditional covariance is ensured to be stationary, and is, in general, well conditioned. We provide an empirical application in the context of international stock markets and offer Monte Carlo evidence of the good small sample properties of the model.