EViews.com

Hi,

I'm working with a model in which my dependent variable is I(1), two of my independent variables are I(1), but my third independent variable is I(0), in this case, if I find that the model as a cointegrating vector (using Phillips-Ouliaris), is it still correct to have a model which include a I(0) variable in a cointegration model (ECM or DOLS).

Thanks.

You can include it in the short run dynamics but no in the long run. So you exclude it from the equation in levels (which will estimate the error corretion term) and include it in the equation in differences. If it's a VAR, eviews allows you to impose restrictions in the VECM matrices.

Try the ardl cointegration method (Pesaran 2001).
http://davegiles.blogspot.com/2013/06/a ... tests.html

Actually, we have our own three part blog series on ARDL estimation, and your are strongly encouraged to read it.

http://blog.eviews.com/2017/04/autoregr ... l.html?m=1

http://blog.eviews.com/2017/05/autoregr ... 8.html?m=1

http://blog.eviews.com/2017/05/autoregr ... l.html?m=1