Hi Mike,
thanks for your email.
My understanding is that your procedure gives the correct twostep estimator
as detailed by Lee, Lung-Fei (1978) "Unionism and Wage Rates: A Simultaneous
Equations Model with Qualitative and Limited Dependent Variables",
International Economic Review, Vol. 19(2), pp. 415-433 and reproduced in
textbooks such as Maddala, G. S. (1983) Limited-Dependent and Qualitative
Variables in Econometrics.
To use the same notation as in the -movestay- article, your procedure
basically substitutes equation (9) and (10) into (11). Call the resulting
equation (11')
And then estimates by -heckman, towstep- (9)-and-(11') and (10)-and-(11')
-movestay- implements full information maximum likelihood, therefore results
are bound to be different. There is the usual trade off. If you believe the
assumption on the joint distribution of the error then -movestay- is both
consistent and efficient, while the twostep estimator is only consistent. If
you believe the hypothesis for the marginal distribution but not the joint
then the twostep procedure remains consistent while the full information
maximum likelihood is not.
The -movestay- article talks about the necessity to adjust standard errors
if the twostep technique is to be used. I have not looked into that.
Best,
Renzo
-----Original Message-----
*From: Michael Johnson [mailto:[email protected]]
Sent: Monday, September 20, 2004 6:55 PM
To: [email protected]
Cc: Renzo Comolli
Subject: Re: st: -movestay- issue with variable abbreviation
Hi, Renzo:
I wonder if we can bypass the movestay command by using two heckman:
use http://www.econ.yale.edu/~rc255/data/for_question_movestay.dta
heckman income age, sel (group= age race_2 race_3) twostep
gen non_group=1-group
heckman income age, sel (non_group = age race_2 race_3) twostep
Do you think this works?
Thanks.
Mike
*
* 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/
