A diagnostic m-test for distributional specification of parametric conditional heteroscedasticity models for financial data

This paper proposes a convenient and generally applicable diagnostic m-test for checking the distributional specification of parametric conditional heteroscedasticity models for financial data such as the customary Student t GARCH model. The proposed test is based on the moments of the probability integral transform of the innovations of the assumed model. Monte-Carlo evidence indicates that our test performs well both in terms of size and power. An empirical example illustrates the practical usefulness of the test and some of its possible extensions are outlined.