Re: st: Panel cointegration

[Manhal Mohammad Ali <manhal.ali@gmail.com>]

Dear Mite,

Traditionally panel GMM techniques like D-GMM and S-GMM there were
used and on micro data sets where N is large for example labour, or
patients, or companies as in Arellano and Bond's (1991) paper and T is
small for 8 years or 5 years etc. However GMM techniques are also
increasingly applied in economic growth for example Bond, Hoeffler and
Temple 2001 where N, non of countries is small and T i.e. time is
relatively large for eg 30 years. If you are motivated, I would highly
recommend to read the papers on Roodman (2006) and Roodman (2009). In
a nutshell, these two papers provide a pedagogic introduction to panel
GMM techniques and warns us about the problems that arises when T is
large and also mentions ways to deal with it. But judging from your
case where N=5, and T=20, I think panel time series due to Pesaran
will be more appropriate. Because as in your case some panel data sets
have relatively small N and large T. So its best to use methods that
are hybrid of time series and panel data methods. The pooled mean
group estimator of Pesaran,Shin and Smith (JASA) does this. In stata,
it is the xtpmg syntax as you mentioned. I think if you read a couple
of papers on these then it will give a clear picture and a robust
idea.

I hope this helps you.

Manhal Ali
University of Bristol.


On Wed, Aug 24, 2011 at 3:32 AM, dar mark <darmar09@gmail.com> wrote:
>
> Dear Statalisters,
>
> I want to do a panel cointegration analysis for N=5 T=20 panel. I have
> searched for estimators and found Pesaran's xtpmg. Can anyone tell me
> can I apply it and if not suggest another estimator?
>
> Thanks,
> Mite
> *
> *   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/