re: st: gmm vs gmm2s

[Kit Baum <baum@bc.edu>]

<>
Gary said
I replicated the gmm example based on the Griliches data from chapter 8 of =
IMEUS using ivreg2.  The book uses gmm; when I ran the problem gmm was repl=
aced by gmm2s:

. use
 http://www.stata-press.com/data/imeus/griliches,
 clear
. ivreg2 lw s expr tenure rns smsa _I* (iq =3D med kww age mrt), gmm2s

The results (param estimates; test stats) differed a bit between gmm and gm=
m2s-I am wondering why?  Different algorithms?


No. As you can see below, if you invoke -ivreg2- from SSC with the gmm option, it tells you that the option is deprecated, and is equivalent to doing -gmm2s robust-.  In recent versions of -ivreg2-, using the option -gmm2s- does nothing: it gives back standard 2SLS results, as unless you state that the error VCE is non-iid, IV-GMM degenerates to 2SLS (see Hayashi's book or the 2007 Baum-Schaffer-Stillman paper, available from my website below). So when you say just -gmm2s- you are getting (non-robust) 2SLS results, which differ from those calculated by -gmm2s robust-, which are (a) robust and (b) calculated via IV-GMM, which for an overidentified equation, differ from robust 2SLS.

. ivreg2 lw s expr tenure rns smsa _I* (iq = med kww age mrt), gmm
-gmm- is no longer a supported option; use -gmm2s- with the appropriate option
      gmm             =  gmm2s robust
      gmm robust      =  gmm2s robust
      gmm bw()        =  gmm2s bw()
      gmm robust bw() =  gmm2s robust bw()
      gmm cluster()   =  gmm2s cluster()


Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/