Gregory-Hansen Cointegration test error

[akn1]

Hi,
I would really appreciate it if any expert or any one who has tried this could help me out. So I found the G-H code by trubador and applied it to my data. I was able to run the test for the level shift model and the level shift with trend model. However, when I attempted to run it for the regime model, I obtained the error 'Near singular matrix error. Regressors may be perfectly colliner in "DO_GHC.LS Y C (@TREND>399-2) G".'

I am not really sure how to rid myself of this problem. Could anyone kindly be of assistance ASAP? I have attached my workfile and program just in case.

Thanks,
AKN

[trubador]

The variable "lmaladl$" seems to be the source of the problem. It remains unchanged until 7/21/2005 from the beginning of the sample and leads to perfect multicollinearity at some point during the analysis. If that is the actual characteristic of that indicator, then you can either use a different sample or drop the variable from the analysis.

[akn1]

Thank you very much Trubador!! It seems that variable was really the issue...i just omitted it from the test. Thanks again. ^^ :D

AKN