|
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: RE: stata drops variable at random
Schaffer, Mark E wrote:
Marck,
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of
Marck Bulter
Sent: 09 December 2007 17:24
To: [email protected]
Subject: st: stata drops variable at random
Dear all,
Currently I am working on a panel data regression. Main
problem is that Stata drops variables at random immediately
after a performed regression.
For example if the following regression is performed:
xtivreg pstrmon price mat age coup pstrmonprev pstrprev intr
ivol compl (precmon = price mat age coup pstrmonprev pstrprev
intr ivol compl
precmonprev)
Stata drops often one of the above variables, or the ID-var
or the Time-Var
There's a basic misunderstanding here. In
xtivreg y x1 x2 x3 (y1 y2 y3 = z1 z2 z3)
the sets of xs, ys and zs shouldn't overlap: the xs are the included
exogenous regressors, the ys are the included endogenous regressors, and
the zs are the excluded instruments.
In your example, you have included all your xs in the list of zs.
Naturally, Stata finds collinearities and drops variables. The right
way to run the regression in your example is
xtivreg pstrmon price mat age coup pstrmonprev pstrprev intr
ivol compl (precmon = precmonprev)
Cheers,
Mark
Any help would be very much appreciated, not much hair left,
Kind regards,
Marck Bulter
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
Dear Mark,
Thank you for your reply. I have corrected the specification and the
regression seem to be more reliable/stable. ( I think I mixed it up with
Eviews, where you have to declear all instruments). Sometimes variables
still disappear, but less often than it used to be. I think there is
still some higher degree of (multi)collinearity between variables.
Thanks again,
regards,
Marck
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/