Comparing figures 1u and 1m for the same model - can give an insight into the performance of the covariance forecast by the model. If the performance in the multivariate context drops drastically compared to the univariate context, this may indicate that the model may be salvaged by refining the methodology for covariance forecasting in conjunction with the model. (This is of course generally true when comparing any u figure with a corresponding m figure for the same model.)