Thanks for the clarification Brian.
I have an additional question why is the cluster option available for
ivtobit but not for tobit? I know that if there is heteroskedasticity tobit
will produce estimators that are inconsistent and cluster only takes care of
the problem in the standard errors. But what if I have say a panel of two
years and I am doing pooled tobit and I want to use cluster to account for
the fact that the observations are repeated twice? I think that will be an
instance where cluster in tobit is valid. So far I have been using intreg
which allows for cluster but it will be easier if tobit had that option.
Anyway, I was wondering why the option is not available for tobit but it is
for ivtobit.
Thanks,
Alejandro
----- Original Message -----
From: "Brian P. Poi" <[email protected]>
To: <[email protected]>
Sent: Sunday, December 18, 2005 6:06 PM
Subject: Re: st: Marginal Effects after ivtobit
> On Sat, 17 Dec 2005, Alejandro Lopez-Feldman wrote:
>
> >
> > I want to calculate marginal effects afterivtobit using mfx but I have a
> > question about the standard errors.
> >
> > The Stata manual ([R] page 552) states that the option "force" has to be
> > used otherwise mfx refuses to calculate standard errors. I have done
this
> > and obtained standard errors. My question is are those standard errors
> > correct or should I worry that they don't account for the fact that
tobit
> > is using a predicted value?
> >
> > Thanks,
> > Alejandro
> >
> >
>
> Alejandro,
>
> Using the "force" option with -mfx- after -ivprobit- or -ivtobit- is fine.
>
> The warning message you get when you do not use that option arises from
> the fact that the endogenous variable appears on the righthand side of the
> structural equation and on the lefthand side of the equation that relates
> it to the instruments. In this case, once we've consistently estimated
> the parameters of the structural equation and their covariance matrix, we
> can go ahead and get marginal effects and their standard errors in the
> usual way.
>
> -- Brian
> -- [email protected]
> *
> * 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/
