EViews.com

Hi guys,

I hope someone can clear up something that i was not sure about. I have these 5 series that I found unit roots for in levels and no unit-root in first differences. Now to run the Johansen cointegration test, should I run it on the original series or run on the first differenced series that I generated (these are I(0) i assume)?

If I run the Johanse test on the original I(1) series, I see no cointergration. But if i run it on the first differenced series, then I get very high ranks of cointegration, but does that second test even make sense since differenced series are already i(0)? If someone can just freshen me up on how cointegration works and why does the second way not make sense?

If there is a unit root present in your series at level, this means that the series are not integrated of order I(0). When first differenced, there is no unit root present, it means that your series are now stationary and of order I(1). Since, both series are of same order, there is a chance of cointegration. So, you perform Johansen Test on the differenced series.