Testing for Granger causality in variance in the presence of causality in mean

In this paper we investigate the effects of neglected causality in mean, on the size properties of some recently proposed tests for causality in variance. The results from Monte Carlo simulations suggest that the tests for causality in variance suffer from severe size distortions when strong causality-in-mean effects are left unaccounted for. Therefore, any conditional mean effects should be filtered out by a parametric model that explicitly allows for the presence of causality in mean before any inferences on causality in variance are drawn.