Can I make a brief comment on this debate.  A residual based test for cointegration is base on the residuals from a cointegrating regression.  Depending on whether the variables in the cointegrating regression have non zero drifts on should include a constants or a trend in the cointegration.  If this is done properly there is no need to include a constant or a trend in the cointegrating regression.  The ADF statistic may, however, depend to some extent on the inclusion of a constant or a trend in the original cointegrating regression.  Can I recommend the account in Hayashi , F (2000), Econometrics, Princeton University Press.
John C Frain
Economics Department 
Trinity College Dublin
Dublin 2  

On 26/07/06, john w <> wrote:
Yes, maybe there could be a possibility to include seasonal dummies too?
Now this is possibile but ADF test is done for them too with, off course,
wrong results.

>From: Allin Cottrell <>
>Reply-To: Gretl list < >
>To: Gretl list <>
>Subject: Re: [Gretl-users] Cointegration Engle-Granger ?
>Date: Tue, 25 Jul 2006 16:19:38 -0400 (EDT)
>On Tue, 25 Jul 2006, john w wrote:
>>I think that skip function is not working...
>>When this option is checked then initial variable DF testing should be
>>excluded. Right?
>There is now a new Windows snapshot in place, with a few more Engle-Granger
>Gretl-users mailing list

Windows Live™ Messenger has arrived. Click here to download it for free!

Gretl-users mailing list

John C Frain