Re: st: Is there a fixed effect quantile regression in STATA?

[Bo MacInnis <bo@macinnis.org>]

Thank you, Stas. You questioned: "Is this really how a quantile regression 
to be interpreted? " I checked out some papers on quantile regression on 
www.jstor.org. The following paper by Buchinsky, who did the famous wage 
inequality in the early 90's,:

Moshe Buchinsky, "Recent Advances in Quantile Regression Models: A 
Practical Guideline for Empirical Research", The Journal of Human 
Resources, 1998; 33(1):88-126

states that:

on page 89: "Useful feaures of the quantile regression and censored 
quantile regression models can be summarized as follows: ......, (e) 
potentially different solutions at distinct quantiles may be interpretated 
as differences in the response of the dependent variable to the changes in 
the regreessors at various points in the conditional distribution of the 
dependent variable; ....."

on page 98: "1. Interpretation of Quantile Regression Estimation, ...... 
How can the quantile's coefficients be interpreted? Consider the partial 
derivative of the conditional (theta-th) quantile of y with repspect to one 
of the regressors, say, x. ... This derivative is to be interpreted as the 
marginal change in the (theta-th) quantiledue to the marginal change in x. 
... One should be cautious in intreprating this result. It doesn't imply 
that a person who happens to be in the theta-th quantile of one conditional 
distribution will also find himself/herself at the same quantile had 
his/her x changed."

Seperately, thank you for suggesting the -xtdata- and I will try it out.

Once again thank you,
Bo
a graduate student at University of California at Berkeley

At 11:33 PM 7/30/2004, you wrote:

> Date: Fri, 30 Jul 2004 14:35:50 -0400 (EDT)
> From: Stas Kolenikov <skolenik@email.unc.edu>
> Subject: Re: st: Is there a fixed effect quantile regression in STATA?
>
> > I have a cross-sectional data on children and their familes. I want to
> > "control for" the family unobservable (the family fixed effect). I have
> > used
> >
> > xtreg y x, fe (family_id)
> >
> > Now I am interested in the effect of x on a particular percentile
> > (quantile) of y. I would like to "differ out" the family unobservable in
> > _qreg_ or _sqreg_.
>
> Is this really how a quantile regression to be interpreted? I was rather
> thinking of it giving a quantile of a conditional distribution. If x is
> age, and y is height, then -qreg- will give you something like the 25% or
> 75% percentiles of the children height (I would specify something like
> B-splines in place of the linear term for this regression to be serious).
> If y is an educational attainment, and x is income, then -qreg- will give
> you the 25% or 75% or whatever percentile of the grade distribution
> conditional on income... but that is not the effect of income on grades in
> the poorest 25% of population.
>
> > Your suggestion about creating dummies may not work very well in my
> > case, because (1) I have a large number of families, and each family has
> > a small (and unbalanced) number of children, say from 1 to 6; and (2) I
> > do not need to estimate the coefficients of the fixed effect, all I want
> > is to control for.  I am looking for something (a STATA command) to do
> > what _areg_ does for the mean response estimates.
>
> Have you checked -xtdata- yet? It can do fixed-effect demeaning for you,
> and then you can run -qreg- (even though it won't give you cluster
> corrected standard errors... which is theoretically possible given that a
> quantile is an M-estimate, and thus can be corrected for clustering with a
> regular sandwich formula).
>
>  ---                                    Stas Kolenikov
>  --       Ph.D. student in Statistics at UNC-Chapel Hill
>  - http://www.komkon.org/~tacik/ -- Stas.Kolenikov@unc.edu

*
*   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/