Page 1 of 1

Difference between EViews and R.

Posted: Wed May 07, 2014 8:37 am
by yanivd
Hello,

By any chance, does anyone know if there is a fundamental difference between EViews' unrestricted VAR procedure and the VAR function (vars package) in R?
I am running the exact same model over the exact same data and getting different results (coefficients).

Thanks!

Re: Difference between EViews and R.

Posted: Wed May 07, 2014 8:39 am
by EViews Gareth
A simple unrestricted VAR is just a set of OLS equations. There should be no difference, assuming you have specified both identically.

Perhaps you could post your results and/or data.

Re: Difference between EViews and R.

Posted: Wed May 07, 2014 9:09 am
by yanivd
Yes, that's why I am confused.

I can't post the data, it is proprietary, but the call in R is simply:

VAR(data, p = 5, type = "const",season=24)

And in EViews I simply estimate using the "Estimate" command and add "c" and a seasonal dummy to the exog. variables.

Maybe it is in how the VAR function handles seasonality?

Re: Difference between EViews and R.

Posted: Wed May 07, 2014 9:26 am
by EViews Gareth
Yep, my guess would be that the seasonal dummy in R is different from the seasonal dummy you're using. I don't know enough about the R code to speculate further though.

Try it without the seasonal term and make sure the results are identical then.