Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: different lag length system GMM


From   "Martin Baumgarten" <Pandaros@gmx.de>
To   statalist@hsphsun2.harvard.edu
Subject   st: different lag length system GMM
Date   Fri, 20 Nov 2009 22:35:59 +0100

Dear statalist-members,

maybe someone can help me out here. I have the following equation using the xtdpd command

y(i,t)=a0+a1*y(i,t-1)+a2*x1(i,t)+a3*x2(i,t)+x3*time+e(i,t)

where l.y is the lagged value of the lhs variable, x1 and x2 are different regressors, and a1,a2(..) are coefficients. Assume now I specify the model as follows by using 

xtdpd  y  l.y  x1  x2  dgmmiv(l.y x1,lagrange(3 3))dgmmiv(x2,lagrange(2 2)) lgmmiv(l.y x1,lag(2))lgmmiv( x2,lag(3))iv(time) artests(2) hascons vce(robust)

Note how I specified different lag length for different variables. I do that, to have a much better p-value for the Sargan-test and for the serial correlation test. However, I do not know if this econometrically or conventionally allowed. Is it? I also cannot remember to have read in the literature about 
It would be great if somebody could give me answer with a short explanation.

Best wishes

Martin 

-- 
GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index