Multivariate GARCH Models: Software Choice and Estimation Issues

Abstract: A large number of important practical tasks can be accomplished using a multivariate GARCH model. This paper examines the relatively small number of software packages that are currently available for estimating such models, in spite of their widespread use. The review focuses upon estimation issues and differences in available options for controlling the optimisation, and the review then considers an application to the estimation of optimal hedge ratios. Large differences in estimated parameters and standard errors are observed, but these are found to generate only modest differences in optimal hedge ratios and virtually indiscernible differences in model performance measures.