Identification of VIRFs

[Jules89]

Dear Tom,

I have a question related to the identification of the shocks within an VIRF analysis as in Hafner and Herwatz (2006).
As far as I understand it correctly, the shocks have to be independent and non-normal to be unique.
If these two conditions have to be met, in addition to a non-normality test, shouldn't independence of the residuals resulting from the BEKK model be tested?

Could you give me a suggestion of a procedure that tests the independence between the residulas resulting from the BEKK model?

In the literature I just see normality tests and univariate tests, but no tests that test for joint independence of the elements of the residual vector.

Thank you in advance

Kind regards

Jules