Panel Cointegration Test

Questions related to panel (pooled cross-section time series) data.
TomDoan
Posts: 7774
Joined: Wed Nov 01, 2006 4:36 pm

Re: Panel Cointegration Test

Unread post by TomDoan »

NONE is actually not an option (we include it to make this similar to other procedures, but there's no theory for FM without at least the constant). If there's no trend in the data, the choice should be obvious. Otherwise, read Pedroni(2007) and see how he decides.
sanjeev
Posts: 191
Joined: Mon Jun 18, 2012 6:51 am

Re: Panel Cointegration Test

Unread post by sanjeev »

Dear Tom,
When I take more than eight endogenous variables in my model while testing for panel cointegration, I get the following message:
###@PANCOINT cannot handle more than 8 endogenous variables.
My question to you is whether this limitation is imposed by the software or the technique itself?

Please reply soon!Its urgent.

Thanks and Regards.
TomDoan
Posts: 7774
Joined: Wed Nov 01, 2006 4:36 pm

Re: Panel Cointegration Test

Unread post by TomDoan »

It's limited by common sense. The method is based upon a single (potentially) cointegrating relationship. If you have a large number of I(1) variables, if they are cointegrated there are almost certainly cointegrated with rank bigger than 1.
Post Reply