ordering of variables in pedroni panel cointegration

[machi]

hi guys!

i am doing pedroni panel cointegration tests in my research... my question is that if i change the order of variables to choose and open as a group to test for cointegration (i am just doing bivariate analysis) the results differ. what does this imply?

thank you very much in advance! :D

[EViews Glenn]

As with the Engle-Granger methodology on which the Pedroni tests are derived, the test results can depend on which of the variables is chosen to be the dependent variable. You should look at the Engle-Granger literature for discussion of this issue.

[machi]

thank you very much! ^^

[kzmh78]

Hi !

I am doing pedroni panel cointegration tests using same methods (open as a group to test for cointegration).
Then the tests results display that 'series: gdp exp'.
Is gdp a used dependent variables? (Is gdp the 'y' at equation(38.6) in Users Guide2 ?)