xtabond2 hansen test

[Marialena Petrakou]

Dear All

i am using the following command to estimate a twostep diff GMM growth model:
xtabond2 growthrate gdpcapt_1 fdi l.fdi findepth rd caplab popden _Iyear_2007 _Iyear_2008 _Iyear_2009 _Iyear_2010 _Iyear_2011 _Iyear_2012 _Iyear_2013 , gmm(gdpcapt_1) gmm(l.fdi l.rd l.caplab )iv(findepth popden _Iyear_2007 _Iyear_2008 _Iyear_2009 _Iyear_2010 _Iyear_2011 _Iyear_2012 _Iyear_2013 ) noleveleq twostep robust

my endogenous vars are in bold and my results in the attached file

can you pls explain how do i interpret this: as far as i know if Hansen test is so low it means that i have instrument proliferation right? ive tried all the possible ways to mitigate this but it still appears 0.0 so maybe i use a wrong command??

Sargan test of overid. restrictions: chi2(114) = 454.70 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(114) = 204.10 Prob > chi2 = 0.000
(Robust, but weakened by many instruments.)

many thanks
marialena

[Sebastian Kripfganz]

Indeed, there seems to be a problem with instrument proliferation. I recommend to use the collapse suboption of the gmm() options, and possibly also to restrict the lags used as instruments with the laglimits() suboption.

[Marialena Petrakou]

thanks Sebastian
the collapse option didnt really improve it and also it messed up my coefficients

can you pls give me an example of a command specification using the laglimits option? it appears invalid in stata when i run it

thank you