Problems with panel cointegration test

[Mito]

I have 2 variables which are LGDP and FDI. When I run panel cointegration test (Open as a Group, View/Cointegration test/ panel cointegration test ...) I use pedroni method. When I choose FDI first( which means my series will be FDI LGDP in the result table), the results are totally different with the results I got when I choose LGDP first (series: LGDP FDI). Anyone got any ideas about why? I think when I test for cointegration, no matter what variables are selected first, I will get the same results.
Many thanks.

[EViews Glenn]

You will not necessarily get the same results using different variables in your system as the dependent variable

http://forums.eviews.com/viewtopic.php? ... le+granger

[Mito]

in the topic you gave me, I still confuse about the conclusions. In my data, the results say that LGDP and FDI is cointegrated but FDI and LGDP is not. It doesnt make sense at all. Could you suggest me what I should say about these results?

[EViews Glenn]

That the results are in disagreement when you perform the test using different dependent variables.