Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

st: RE: ivreg2 warning but ALL statistics are reported


From   "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: ivreg2 warning but ALL statistics are reported
Date   Tue, 30 Mar 2010 15:40:46 +0100

Umid,

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Umid Aliev
> Sent: Tuesday, March 30, 2010 3:13 PM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: ivreg2 warning but ALL statistics are reported
> 
> Reposting the question as the old text seemed a bit unclear.
> 
> Dear Stata users,
> 
> In my work on estimating wage equation using cross-sectional 
> data I am using
> ivreg2 command for 2-step GMM. Since the data is 
> cross-regional I am tryig to
> use cluster-robust standard errors. I have 26 clusters 
> (regions) and 70
> regressors (including industry and regional dummies). To 
> avoid the problem of
> having weighting matrix to be rank deficient I partialled out 
> industrial and
> regional dummies, which left me with only 3 regressors (1 endogenous
> instrumented by 3 excluded) for 26 clusters (regions). 
> Nevertheless I got the
> warning message -
> 
> > warning - estimated covariance matrix of moment conditions 
> not of full rank.
> > overidentification statistic not reported, and standard 
> errors and model
> > tests should be interpreted with caution.
> >
> > possible causes -  number of clusters insufficient to 
> calculate robust
> > covariance matrix singleton dummy variable (dummy with one 
> 1 and N-1 0s or
> > vice versa) partial option may address problem.
> 
> Overidentification test Hansen J statistic is reported 
> contrary to what is
> stated in the warning message. The only statistic missing is 
> Stock-Wright LM S statistic.

This is a bit odd.  Partialling-out is the right strategy, and your
results should be valid.  I don't understand, though, why ivreg2 is
still complaining about a var-cov matrix of orthogonality conditions
that isn't full rank, since, as you point out, it was able to calculate
the J stat.

We released a major update of ivreg2 not long ago.  Are you using the
most recent version?  I've got version 3.0.01 on my machine here.

We also released a version of ivreg2 that runs under Stata 9 and later,
ivreg29.  The internal code of ivreg2 and ivreg29 are rather different
(ivreg2 is heaviily Mata-ized, ivreg29 is not).  Do you get the same
output with ivreg29?

Cheers,
Mark

> I am slightly confused on how to treat these 
> results. Are they
> valid? or does the appearance of warning suggest that further 
> work needs to be
> done?
> 
> Many thanks,
> Umid
> 
> Umid Aliev
> Leeds University Business School
> Maurice Keyworth Building
> Leeds
> 
> *
> *   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/
> 


-- 
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.


*
*   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   |   Site index