st: RE: Panel with multiple problems

["Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>]

Phil,

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> statalist@hsphsun2.harvard.edu] On Behalf Of Philip Bromiley
> Sent: 25 March 2014 23:41
> To: statalist@hsphsun2.harvard.edu
> Subject: st: Panel with multiple problems
> 
> I have a panel data set with many firms and 2 to 20 annual observations per
> firm.  I want fixed effects.
> 
> The model includes an endogenous regressor.  I am using additional lags of the
> endogenous regressor as instruments.  I expect substantial heteroscedasticity
> by firm and serial correlation within firms.
> 
> My experience is that estimators that correct for serial correlation (e.g.,
> xtregar) give very different betas than those that do not (e.g., xtreg), even if
> both estimators are consistent.
> 
> I know xtivreg2 with cluster and xtivreg2 with the bw() robust options are both
> robust to arbitrary heteroskedasticity and arbitrary autocorrelation.  The
> documentation says the cluster option is robust to "arbitrary intragroup
> autocorrelation" but for the bw()robust option, it just says
> "arbitrary  autocorrelation".

Just a quick comment on this.  The standard cluster-robust VCE relies on large-N asymptotics, i.e., you need the number of clusters N to go off to infinity.  The bw(.) option uses kernels and bandwidths that typically rely on large-T asymptotics, i.e., you need the number of periods T to go off to infinity.

You don't say how many firms you have, but 2-20 annual observations isn't very many, so relying on large-T asymptotics would be iffy.  Given the choice between cluster-robust and kernel-robust, you'd probably want to opt for the former.

HTH,
Mark

>  However, the gmm estimator (option gmm2s)
> also appears robust to heteroskedasticity and some autocorrelation.
> 
> I am considering using xtivreg with aweights along with the options fe gmm2s
> cluster(gvkey).
> 
> Suggestions would be welcome.  Thank you.
> 
> Phil
> 
> 
> Philip Bromiley
> Dean's Professor in Strategic Management
> Merage School of Business
> University of California, Irvine
> Irvine, CA 92697-3125
> Phone: (949) 824-6657
> Fax: (949) 725-2898
> Email: Bromiley@uci.edu
> 
> "The subject who is truly loyal to the Chief Magistrate will neither advise
> nor submit to arbitrary measures."   Junius
> 
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/


----- 
Sunday Times Scottish University of the Year 2011-2013
Top in the UK for student experience
Fourth university in the UK and top in Scotland (National Student Survey 2012)


We invite research leaders and ambitious early career researchers to 
join us in leading and driving research in key inter-disciplinary themes. 
Please see www.hw.ac.uk/researchleaders for further information and how
to apply.

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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/