Re: st: clustered bootstrapped standard errors in quantile regressions

[Stas Kolenikov <skolenik@gmail.com>]

On Thu, Apr 7, 2011 at 7:52 AM, Jacqueline S. Zweig <smith2@usc.edu> wrote:
> Is there a program in Stata to calculate clustered bootstrapped
> standard errors in quantile regressions?  It seems that SQREG  does
> not support clustering and QREG does not support the bootstrap prefix.
>
> I could not find a response to this original post from a year ago with
> the same question:
> http://www.stata.com/statalist/archive/2010-03/msg01290.html

In a recent wishful discussion of dynamic time series models features
for a blockbuster Stata 12, people have been throwing references to
technical literature. Can you point to one that derives a bootstrap
procedure that would produce consistent standard errors for quantile
regression? Until such a paper appears in JASA/Econometrica/Journal of
Econometrics (CJS is also a possibility; I might have reservations
about TAS), clustered bootstrap in quantile regression, although
probably feasible via a combination of -bsweights-/-bs4rw-/-sqreg-,
essentially rests on hand-waving and computing tricks rather than
solid theory. If somebody stomps your paper because you did not use
the right procedure, that's your academic reputation to gamble with.

May be Austin Nichols or Steve Samuels think otherwise, and do know
such references, in which case you'd want to listen to them :).

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   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/