VAR Estimation...

[eViews User]

Hi everyone. I have the following issue in eViews:

I am trying to run a VAR for the 3 month treasury bill rate and the 1 year UK libor rate but I have come unstuck when deciding whether to include a trend in my VAR model. A graph of the two series indicates that both are trending downwards across the sample, and unit root tests (ADF and KPSS) confirm that both of the series are trend stationary at the 5% level. Given this, should I include a trend as an exogenous variable when I estimate a VAR process?

Also, do I need to run any unit root tests prior to my VAR estimation? I know that in the univariate case it is usually done but I am unsure as to whether this is true in the multivariate case.

Thanks for your time.

[dealsfe]

If you use VAR, your variables must be I(0). if your variables have trend, you should not use VAR.