R help - Jun 2011 - serial autocorrelation tests for VAR model under heteroskedasticity

Hi all,

I have the problem that in an estimated vector autoregressive model I find
arch effects using the arch-lm test. As far as I know, the tests for
autocorrelation in residuals are then biased. Am I right? So my question: Is
there any possibility to conduct the serial tests using heteroskedasticity
robust covariance matrices? The function serial.test in the vars package
does not seem to offer such possibility.

Thank you in advance.

Regards

	[[alternative HTML version deleted]]