RE: st: RE: difficulty in explaining GMM sargan overid

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

Binta,

Clive is exactly right.  I said you need to reduce the number of IVs (a lot), but as he correctly points out, you can also address the problem by increasing N (a lot).

The reason for using 2SLS in my explanation is simply that it's easier to get the intution for what is going on with 2SLS.  2SLS and the A-B GMM estimator are very close relatives; in fact, if you add the assumption of conditional homoskedasticity to the latter, you will get a 2SLS version of the estimator.

Cheers,
Mark

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of B B
> Sent: 25 June 2010 09:18
> To: statalist@hsphsun2.harvard.edu
> Subject: Re: st: RE: difficulty in explaining GMM sargan overid
> 
> ha!
> 
> right, I did for once think he meant to reduce the 
> instruments but then I thought it might just be the method. 
> Thank you very much for clarifying this.l It is very helpful.
> 
> I think I cant increase the Ns because I only restricted the 
> sample to a certain group of emerging and frontier markets 
> and plus I have a limited time to do this, however, if after 
> every thing you've suggested, (reducing the IV marginally) it 
> does need me to increase the N then maybe I have no option.
> 
> Thank you, Thank you, Thank you
> 
> Binta
> 
> --- On Fri, 25/6/10, Clive Nicholas <clivelists@googlemail.com> wrote:
> 
> > From: Clive Nicholas <clivelists@googlemail.com>
> > Subject: Re: st: RE: difficulty in explaining GMM sargan overid
> > To: statalist@hsphsun2.harvard.edu
> > Date: Friday, 25 June, 2010, 5:05
> > Binta Sarat replied to Mark
> > Schaffer:
> > 
> > [...]
> > 
> > > Just wanted to clarify, are you suggesting using 2sls?
> > And if yes, from your response it seems like when I use 
> 2sls, I might 
> > end up with similar output to OLS, is this correct? Please help 
> > clarify this. I'm sorry it sounds really lame,but Im new to most of 
> > these things and Im trying to not get confused with them.
> > 
> > I suspect your tiredness is preventing you from following what Mark 
> > wrote closely enough! He's simply saying that you're using far too 
> > many instrumental variables (350) for the observations 
> you've got to 
> > play with (209).
> > 
> > Remember that, in many ways, IVs are not much different 
> from any other
> > variables: adding each one to your model will consume 1 degree of 
> > freedom. Since you don't have nearly enough data to handle this, 
> > you'll need to either:
> > 
> > (1) substantially increase your Ns, if that's possible; or
> > (2) substantially reduce your instruments, exactly as Mark advises.
> > 
> > As for whether "to 2SLS or not to 2SLS?", we don't know 
> your data to 
> > really advise you on IV model selection, and in any case, 
> this really 
> > is your call.
> > 
> > --
> > Clive Nicholas
> > 
> > [Please DO NOT mail me personally here, but at 
> > <clivenicholas@hotmail.com>.
> > Please respond to contributions I make in a list thread 
> here. Thanks!]
> > 
> > "My colleagues in the social sciences talk a great deal about 
> > methodology. I prefer to call it style." -- Freeman J.
> > Dyson.
> > 
> > *
> > *   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/
> 


-- 
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/