Re: st: quantile regression and endogenous variables

[jean ries <ries@ires.ucl.ac.be>]

Justina,

As far as your first question is concerned, there are some papers that 
discuss endogeneity in quantile regressions. A fairly recent one is:

Sokbae Lee, 2004, Endogeneity in quantile regression models: a control 
function approach, CEMMAP WP 08/04.

downloadable here:

http://cemmap.ifs.org.uk/docs/cwp0804.pdf

As it seems, the method suggested in this paper can be implemented using 
the Stata commands for quantile regression.  Moreover, Lee also discusses 
some related methods. Thus, it could be a good starting point for you.

Hope this helps?

Jean


At 14:10 14/09/2004, you wrote:


> Dear All,
>
> using individual data (cross section) I want to run a quantile regression 
> (qreg, sqreg.) on the 0.1, 0.25, etc. percentile of the conditional 
> distribution. Unfortunately, I suspect that one or two determinants (X) 
> are endogenous.
> Does anybody of the specialists here know a way to handle this problem ( 
> theoretically and particularly using sqreg / Stata commands)?
>
> Furthermore, if I estimated OLS I would use the cluster-option as the 
> variables are measured at different levels ( individual, school, region).
> Is there an analogon to clustering for quantile regressions or, using this 
> qreg-method, did I not have to 'take into account' different levels ?
>
> Any comments are very welcome.
>
> Sincerely,
>
> Justina Fischer
*
*   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/