cointegration

[cychang]

hi everyone

I am confused. If I have three variables, two of them are I(1), and the other one is I(0), is it appropriate to use the bulit-in package(Johansen error correction model) to analysis? It sounds strange to use Johansen error correction model to analysis this kind of mixed group, but someone told me that because the linear combination of those variables are still possible to be a I(0) variable, so its still ok to use error correction model. If anyone knows the answer, please let me know. I will very appreciated for that.

sincerely

Andy

[Arun.stat]

In your system you have variables with maximum order of integration 1. Therefore it is an I(1) system. And as you have 1 stationary variable there is trivially one cointegration relationship i.e. trivially rank of the cointegration matrix is minimum 1. However this rank can be max. of 2 depending on the structure of other 2 integrated variables.

HTH,

Arun