EViews.com

Hi,

I am estimating causality between 4 variables in panel data and part of the process includes testing for cointegration. When I test for Johansen-Fisher cointegration through Quick>Group Statistics>Johansen Cointegration Test, I get different results from when I set-up a VAR with my 4 variables and then go to View>Cointegration test.

Is there a reason for which this is happening and if so, which output should I follow?

I am using EviewS 9. The two different outputs are attached.

Thank you for your help.

The VAR/VEC estimator and corresponding test is doing nothing panel related, whereas the group cointegration test is doing the panel version of the tests.

Therefore, I should follow the Group Statistics cointegration test.

Thank you for your reply!!!