Re: st: Quantile regression with sample selection

[Luciana Mèndez <lucimerrico@gmail.com>]

Hy Gustavo,

There´s no module in Stata as far as I know to do that
,
what I am doing to replicate Buchinsky´s method was  first estimate a
single index ( semiparametric maximum likelihood estimator of Klein
and Spady (1993)) using sml command in stata,

then you have to calculate Mill´s ratio, and then incorporate this
ratio on your quantile regression,

**Mills ratio

generate mills = normalden(index)/normal(index) if female

replace  mills = -normalden(index)/1-normal(index) if male

gen mills2=mills^2

***include the power series into the QR**

qreg depvar indepvar  mills mills2 if female, q(0.1)



Buchinksy suggest including power series of order two, but you can try
with power series of order 3 and see whether the correction terms are
significant.
I hope this could help you,

Luciana

2012/3/28 Gustavo Palmeira <gustavopalmeira@gmail.com>:
>
>  I like to do quantile regression analysis to understand the earning
> gap evolution between formal and informal sector in Brazil. I’ve read
> a paper by Buchinsky "Quantile regression with sample selection:
> estimating women returns to education in the U.S." where he developed
> a sample selection procedure (a non parametric method similar to one
> developed by Heckman (1980 and Newey (1991)) and I'm having problems
> to repeat this technique in Stata. Can anyone help me with this
> technique? Is there a stata module for this procedure or can you help
> me constructing this selection equation in stata format step by step?
>
> *
> *   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/

*
*   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/