how to estimate LB-Q tests in BEkK model?

[hardmann]

Dear tom:

thank you tom. I had done.
I found bdindtests and regcorrs.

[TomDoan]

First, that's not a BEKK model; you need the option MV=BEKK.

That's the correct way to compute and test the standardized residuals. How large a data set do you have, and how large are the correlations on the squares? With a large data set, it's quite possible for very small autocorrelations to be significant at the .05 level.