Wake me up before you GO-GARCH

"The `holy grail' in multivariate GARCH modelling is without any doubt a parameterization of the covariance matrix that is feasible in terms of estimation at a minimum loss of generality" (van der Weide, 2002). Recent models that aspire such favourable position in this trade-off are the DCC model by Engle (2002) and the GO-GARCH model by van der Weide (2002). These models have gained generality on the earliest models designed to be feasible, CCC and O-GARCH, without losing too much of their practical attractiveness. Generality may be measured by the ability to model the key stylized facts of multivariate data:(i) Persistence in volatility and covariation; (ii) Time-varying correlation; and (iii) Spill-over effects in volatility. The DCC model incorporates the first two items, but trades the third for particular ease of estimation. On the other hand, GO-GARCH which is nested in the general BEKK model meets all three key aspects of empirical data, while it may seem to give in a little on DCC in terms of practicability. This paper proposes an alternative method of estimating GO-GARCH that will substantially increase feasibility while preserving generality. In effect, the approach does not become more complicated than estimating a Vector Autoregressive Model along the way. As the procedure may easily be implemented in any popular software package, such as EViews, it should meet the convenience of DCC