RE: st: RE: xtabond2--number of instruments.

[DE SOUZA Eric <eric.de_souza@coleurope.eu>]

The abdata file is unbalanced and, therefore, difficult to follow the instrument count.
If you execute the following you will see how the instruments are reduced as a year is dropped from the sample. (I'm sure it can be succinctly written as a program). The example is not meant to make any sense and is purely used for the instrument count

webuse grunfeld.dta
sum
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1935
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1936
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1937
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1938
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1939
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1940
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1941
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1942
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1943
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1944
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1945
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1946
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1947
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1948
xtabond2 invest l.invest, gmm(l.invest) noleveleq
drop if year == 1949
xtabond2 invest l.invest, gmm(l.invest) noleveleq
clear

Eric


Eric de Souza
College of Europe
Brugge (Bruges), Belgium
http://www.coleurope.eu


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Steven Archambault
Sent: 29 June 2011 19:31
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: RE: xtabond2--number of instruments.

Hi Eric,

I appreciate the links, those definitely help. But, I still feel like I cannot predict what the number of instruments will be before I run the regression. For example, using the sample data for xtabond2 (http://www.stata-press.com/data/r7/abdata.dta), I run the following simple regressions and get the number of instruments.

xtabond2 n l.n k, gmm(l.n) noleveleq   (28 instruments for 751 obs and
141 groups)

I would have expected a minimum 4 instruments per group, since there are 5-7 annual observations per group. Where is my miscalculation?

Thanks,
Steve

On Tue, Jun 28, 2011 at 7:19 AM, DE SOUZA Eric <eric.de_souza@coleurope.eu> wrote:
>
> To your question concerning whether it is counting instruments for each year, the answer is yes. I forgot to answer this in my previous reply.
> See in -xtabond2- the difference between instruments introduced with 
> the option -iv- and instruments introduced with the option -gmm-
>
> Eric
>
>
> Eric de Souza
> College of Europe
> Brugge (Bruges), Belgium
> http://www.coleurope.eu
>
>
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Steven 
> Archambault
> Sent: 28 June 2011 00:15
> To: statalist@hsphsun2.harvard.edu
> Subject: st: xtabond2--number of instruments.
>
> Hi all,
>
> I am trying to figure out xtabond2. I don't see why this command gives 
> me 42 instruments
>
> xtabond2 lnyp l(1/2).lnyp yr1988-yr2009, gmm(l(1).lnyp, laglimits(1 
> 2)) robust noleveleq
>
> This equation gives me 56 instruments
>
>  xtabond2 lnyp l(1/2).lnyp yr1988-yr2009, gmm(l(1).lnyp, laglimits(1 
> 3)) robust noleveleq
>
> Is it counting instruments for each year? I don't quite get it.
>
> By the way, there are 1752 observations, 122 panels.
>
> I'd appreciate the info. Also, I am reading "How to do xtabond2". Its very informative, but I am still unclear.
>
> Thanks!
>
> -Steve
>
>
> *
> *   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/
>
> *
> *   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/

*
*   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/

*
*   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/