Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: xtoverid error: internal reestimation of eqn differs from original


From   "Schaffer, Mark E" <[email protected]>
To   <[email protected]>
Subject   st: RE: xtoverid error: internal reestimation of eqn differs from original
Date   Fri, 24 Oct 2008 14:26:10 +0100

Statalist didn't make it through despite it being in plain text format.
If anyone has any suggestions about what went wrong with Claudio's email
and why it didn't make it through to the list, can they let him know?)

Claudio,

> -----Original Message-----
> From: [email protected] [mailto:[email protected]] 
> Sent: Friday, October 24, 2008 1:58 PM
> To: [email protected]; 
> [email protected]
> Cc: Schaffer, Mark E
> Subject: Fw: xtoverid error: internal reestimation of eqn 
> differs from original
> 
> 
> ----- Original Message ----- 
> From: <[email protected]>
> To: <[email protected]>
> Sent: Friday, October 24, 2008 2:10 PM
> Subject: xtoverid error: internal reestimation of eqn differs from
original
> 
> 
> > dear List, dear Mark
> > i have had a look to previous questions and answers about the
xtoverid 
> > error after Hausman-Taylor stata command.
> > The problem is quite clear, but i don't understand the strategy to
solve 
> > it. In particular, my specificatrion is close to Maria's one: among
my 
> > regressors, i have a set of interaction terms (dummies) between
countries 
> > and time. Some of these are dropped in the estimates due to
collinearity. 
> > When i run xtoverid,  endogenous variable(s) is collinear with 
> > instruments. Running the noisily option, it seems clear that
problems 
> > arise in the set of interaction terms. In fact, if i don't include
them in 
> > the regression, no problems arise with xtoverid .
> > I tried to change my engonous variables (variables included in
endog( ) 
> > list of Hausman-Taylor syntax) without any result; is the exclution
of 
> > interaction terms from my specification the only chance to perform
an 
> > overidentification test? or  there is some "trick" you can suggest
to me?

My guess is that the variables that are dropped by -xthtaylor- to remove
the collinearities are not the same variables that are dropped by
-xtoverid-.  -xtoverid- detects this and exits with an error.

What you need to do is to control the variables that are dropped by
explicitly dropping them yourself, i.e., supplying the minimal
non-collinear set of variables, so that neither -xthtaylor- nor
-xtoverid- have to drop anying.  You can do this by using _rmcoll to
detect collinearites, gradually dropping variables until _rmcoll doesn't
find anything, and then run the regression with the remaining variables.

Hope this helps.

Cheers,
Mark

> > Thanks
> > Claudio
> >
> > 
> **************************************************************
> **********
> > Claudio Vicarelli
> > Macroeconomics and International Economic Policy Dep.
> > Institute for Studies and Economic Analyses (ISAE)
> > Piazza Indipendenza n.4 00185, Rome, Italy
> > tel. +39 0644482370
> > 
> 
> 


-- 
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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index