Dear Maria,
I have also had problems with achieving convergence using Heckman MLE. From
what I understand it is very common to have problems with convergence using
MLE (I think Wooldridge's book argue this), that is why a lot of people
still uses the twostep option. Also, do you have exlcusion restrictions?
Are they 'good'? That might be another reason for your problem.
A couple of references that might be useful are:
Vella, F. (1998). Estimating Models with Sample Selection Bias: A Survey.
The Journal of Human Resources. 33(1): 127-169.
Leung, S.F. and Yu. S. (1996). On the choice between sample selection and
two-part models. Journal of Econometrics. 72: 197-229.
I hope this helps.
> ----- Original Message -----
> From: "Santos, Maria Emma" <[email protected]>
> To: <[email protected]>
> Sent: Thursday, March 17, 2005 12:48 PM
> Subject: Re: st: Convergence not achieved in MLE Heckman
>
>
> > Dear Renzo,
> >
> > Thank you for the advice. The problem is that after excluding iterate (0)
> the
> > iteration log keeps showing "not concave" messages...
> >
> > Maria.
> >
> > Quoting Renzo Comolli <[email protected]>:
> >
> > > Dear Maria:
> > >
> > > Get rid of the -, iterate(0)- option. It is blocking the optimization
> > > process.
> > > -, iterate(#)- blocks the optimization process at # regardless of
> > > convergence. By blocking it at 0, you basically see the intial values.
> > >
> > > Renzo
> > >
> >
> > --------------------------------------------------------------------------
> --
> > > ----
> > > *From "Santos, Maria Emma" <[email protected]>
> > > To [email protected]
> > > Subject st: Convergence not achieved in MLE Heckman
> > > Date Thu, 17 Mar 2005 12:11:10 -0600
> > >
> >
> > --------------------------------------------------------------------------
> --
> > > ----
> > >
> > > Hello!
> > >
> > > I am using MLE Heckman procedure with standard errors corrected for
> > > heteroskedasticity and clustering, with the options iterate(0) and
> > > "difficult".
> > >
> > > I get estimations but it says: "Convergence not achieved". My question
> is if
> > > the
> > > fact that convergence has not been achieved makes the estimations
> > > unreliable.
> > > If that is the case, what should I do to try to improve estimations?
> > > I have tried dropping some variables but still convergence is not
> achieved.
> > >
> > > Also, I haven't been able to use the Heckman two steps procedure because
> I
> > > need
> > > to weight the variables.
> > >
> > > I would really appreciate if someone could give me some advice on this
> issue
> > > or
> > > at least some reference where I can get information.
> > >
> > > Sincerely yours,
> > > Maria.
> > > -----------------------------------------------------------------
> > > Santos, Maria Emma
> > > Vanderbilt University
> > > Email: [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/
> > >
> >
> >
> > -----------------------------------------------------------------
> > Santos, Maria Emma
> > Vanderbilt University
> > Email: [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/
>
>
-----------------------------------------
Alejandro Lopez-Feldman
Ph.D. Candidate
Agricultural and Resource Economics
University of California-Davis
*
* 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/
