Johansen Cointegration Test

[thuynguyen]

I want to examine the relationship between stock returns and five macroeconomic factors. They are all I(1). I run VAR test then Johansen cointegration test. Both the trace test and the maximum eigenvalue test indicate 6 cointegrating vectors. Could you please let me know that I can confirm the model is therefore stationary, or no stable long-run relationship? Could I continue with VEC model?
My supervisor suggests me to test the relationship between stock returns and individual macro factor first to see which one is statistical significant, then run the test for stock returns and all significant macro factors later. Should I do that way?
I am very new user of Eviews, so please help me to clear the problems. Many thanks!

[trubador]

This has nothing to do with EViews at all. You are having a difficulty in understanding the methodology that you have chosen. I am a fan of general-to-specific approach, so it is your call to follow your supervisor's advice. There have been a plenty of discussions in the forum with regards to VEC or modeling non-stationary series, so you can search for them to get some practical ideas. But, I suggest you to refer to a textbook for a deeper understanding of the methodology before going any further.

[thuynguyen]

Thanks for your suggestion. I'm gonna read text book to get deeper understanding for my method too. However, I've searched in the forum for VAR/VEC model for non-stationary series. There are three steps as I understood: check unit root test; select lag length; and run Johansen cointegration test to see whether there are conintegrating vectors. If there are more than one cointegrating vector, we should run VEC model afterwards. Please correct me if anything's wrong. Regards,