A Note on the Bivariate Orthogonal GARCH Model

This short paper discusses some econometric flaws in the way the conditional covariance matrix is estimated in a bivariate orthogonal GARCH model. The practical implications of these problems are illustrated in two empirical cases. Although this kind of model is usually employed with many more than two time series, some downturns of the bivariate version might apply also to the more general case.