[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Re: number of obs and GMM
> Dear Mark,
> I am using the ivreg2 command with a time-series cross section, about 200
> - 300 observations ( N = 26). We know that GMM performs best in large
> Do you know the minimal number of obs for ensuring the nice large sample
> properties of GMM?
The short answer is I don't know, partly because I don't know all the
papers on this in enough detail, and partly because the literature I have
seen is usually oriented towards specific problems. For example, I think
there are some papers on this in a special issue of the Journal of
Business and Economic Statistics (14:3, 1996), but my recollection is that
they investigate particular case such as time series applications.
You are using panel data, so you are probably interested in Monte Carlo
studies of the small-sample properties of panel data GMM estimators. I
think Windmeijer has written on this, also Blundell and Bond. Windmeijer
has a 2000 Institute for Fiscal Studies Working Paper in which he proposes
a finite sample correction for 2-step GMM that may be useful for you. I'm
not a panel data expert, though, so I'm posting this to Statalist because
there are panel data experts out there who may be able to offer
BTW, if you want to use simple fixed effects or first differences, you can
use -xtivreg2- instead. You can get more sophisticated panel estimators
with David Roodman's -xtabond2-.
Hope this helps.
> (I was not able to fnd according literature in the web)
> Thanks a lot,
> Dr. Justina A.V. Fischer, M.A.
> Research Associate
> SIAW - University of St. Gallen
> Bodanstrasse 8
> CH-9000 St. Gallen
> Tel.:++41-71-224 2345
> Fax: ++41-71-224 2298
Prof. Mark Schaffer
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3294
This e-mail message is subject to http://www.hw.ac.uk/disclaim.htm
* For searches and help try: