Valerie,
Have you considered saving your residuals and entering them into a VAR. You could run the
var at the appropriate lag and then type: varlmar, mlag(#)
where mlag is the maximum lag, This is a laGrange Mutliplier test for the joint autocorrelation at
the lags up to mlag.
- Regards,
Bob Yaffee
Robert A. Yaffee, Ph.D.
Research Professor
Silver School of Social Work
New York University
----- Original Message -----
From: Valerie Orozco <[email protected]>
Date: Wednesday, July 2, 2008 9:16 am
Subject: RE : st:How to test autocorrelation in the disturbance in a system of equations?
To: "[email protected]" <[email protected]>
> Thank you Nick and thank you Robert,
> I know that I can treat each equation separately (saving the residual
> and testing all equations independently) but since the system I am
> estimating is a simultaneous one with some correlations between
> equations, such test (equation by equation independently) is
> incorrect. For example, one residi of one equation coul be correlated
> to a lagged residj with j different from i.
> I was wondering if a joint test (of all the system) for
> autocorrelation in the full system exists...
>
> thank you
>
> val�rie
>
> ________________________________________
> De : Valerie Orozco
> Date d'envoi : mardi 1 juillet 2008 17:06
> � : [email protected]
> Objet : How to test autocorrelation in the disturbance in a system of
> equations?
>
> Dear all,
>
> I posted my question some days ago but have no answer. I try to
> explain it in a better way.
> I'm estimating a system of equations (by 3SLS with "reg3") (20 equations)
> I'm wondering if a joint test for autocorrelation in the disturbance
> exists in such simultaneous model.
>
> I know the durbin Watson test, breush godfrey test, and ljung box test
> to test the correlation in the disturbance of one equation (after "regress").
> Thus, in my system of equations, I am able to test each equation
> separately (programming the durbin Watson or ljung box formula for
> each of all the equation). But I would like to know if there exists a
> way to test the autocorrelation globally (i.e a joint test)
> (even if I have to program it)
>
> If you have any idea...
>
> Thank you very much.
>
> val�rie
>
> -------------------------------
> Val�rie OROZCO
> Toulouse School of Economics (INRA-GREMAQ)
> 21, all�e de Brienne
> F-31000 Toulouse, France
> -------------------------------
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
