st: RE: problems combining xtivreg and xtoverid or using xtivreg2 with random effects estimation

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

Daniel, 

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Schneng
> Sent: 29 September 2011 16:09
> To: statalist@hsphsun2.harvard.edu
> Subject: st: problems combining xtivreg and xtoverid or using 
> xtivreg2 with random effects estimation
> 
> --> I posted this earlier, but it didn't appear at the 
> statalist, so I 
> --> tried
> to post it again. I hope that the first post doesn't pop up 
> somewhere and confuses you.
> 
> 
> Dear Statalist,
> 
> for my Master's Thesis I'm currently dealing with a Panel 
> Data Analysis and Instrumental Variables, in fact with a 
> xtivreg. The problem that occurred is, that there is no 
> proper and adequate way to make a random effects estimation. 
> Xtivreg does one though, but the post estimation for relevant 
> and weak instruments doesn't work well. I tried xtoverid, but 
> this fails due to the omitted variable problem (have a look here:
> http://www.stata.com/statalist/archive/2011-09/msg00885.html).

The "omitted variable problem" is as described in the post you link to:
"-xtoverid- was written before Stata introduced the omitted variable
operator "o".  Variables are being dropped from your estimation and
being reported as such with an "o." in front of them, and -xtoverid-
doesn't understand this."

>  Even if I drop all omitted variables xtoverid doesn't work.

But this is peculiar.  If you estimate using -xtivreg- but exclude all
the variables that Stata is dropping, -xtoverid- should now work.  

Can you show us the output of -xtivreg- with the variables omitted?

There's one other idea that comes to mind, namely version control.  "o."
was introduced in Stata 11, if I recall.  It's possible that if you
precede your -xtivreg- estimation with the line

version 10

-xtivreg- will not report using "o." for omitted variables, and the
subsequent call to -xtoverid- may then work.  The -version- statement
will need to be in a do file along with the subsequent -xivreg- call.

HTH,
Mark
 
> Xtivreg2 on the other hand can't deal with random effects. 
> In a few posts and e-mails on Statalist I read, that there is 
> a xtivreg2 version developed, which can handle the random 
> effects and the statistics around that. 
> Maybe this would solve my problem?
> (http://statalist.1588530.n2.nabble.com/XTIVREG-first-stage-F-
> statistic-td5660893.html#a5661971) 
> 
> Thank you very very much.
> 
> Kind regards, Daniel Schalling.
> 
> 
> --
> View this message in context: 
> http://statalist.1588530.n2.nabble.com/problems-combining-xtiv
> reg-and-xtoverid-or-using-xtivreg2-with-random-effects-estimat
> ion-tp6844236p6844236.html
> Sent from the Statalist mailing list archive at Nabble.com.
> 
> *
> *   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.

Heriot-Watt University is the Sunday Times
Scottish University of the Year 2011-2012



*
*   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/