Johansen Cointegration Test

[tba]

Hello,

I would appreciate if someone can answer some of my questions:

1) In Johansen cointegration test, in order to select the lag length I try to select option 6 (summarize all 5 options) first, but it gives the error 'near singular matrix'. Why does that happen? The error occurs only in the sixth option.

2) Is it okay to run VAR on stationary variables and find the lag length for Johansen that way?

3) If trace and max statistics give different results, is there any way to choose between them? (Both are very close to CV)

Thank you

[sawaviews]

Hi there,

I am having a similar problem. When i am evaluating the lag length criteria, if i enter 2 lags I am given the "near singular matrix" error but this does not appear if i test with just 1 lag. However, when i then go forward and try to perform the Johansen cointegration test Assumption6, or otherwise, it returns an error of "insufficient observations". I dont know how to rectify the problem and think the near singular matrix is causing problems throughout especially as my number of observations is 39 and I have performed the JCT with smaller samples.

Any help?!?!