Testing Causality Between Two Vectors in Multivariate GARCH Models

Spillover and contagion eects have gained significant interest in the recent years of financial crisis. Attention has not only been directed to relations between returns of financial variables, but to spillovers in risk as well. I use the family of Constant Conditional Correlation GARCH models to model the risk associated with financial time series and to make inferences about Granger causal relations between second conditional moments. The restrictions for second-order Granger noncausality between two vectors of variables are derived. To assess the credibility of the noncausality hypotheses, I employ Bayes factors. Bayesian testing procedures have not yet been applied to the problem of testing Granger noncausality. Contrary to classical tests, Bayes factors make such testing possible, regardless of the form of the restrictions on the parameters of the model. Moreover, they relax the assumptions about the existence of higher-order moments of the processes required in classical tests.