R help - Oct 2006 - Quantile Regression: Measuring Goodness of Fit

Hi,
 
how to measure the goodness of fit, when using the rq() function of quantreg? I
need something like an R^2 for quantile regression, a single number which tells
me if the fit of the whole quantile process (not only for a single quantile) is
o.k. or not.
Is it possible to compare the (conditional) quantile process with the
(unconditional) empirical distribution function? Perhaps with a Chi^2 or
Kolmogorv-Smirnov Test?
 
Thanks for feedback.
 
Martin

Hi Martin,

Efferz, Martin <efferz <at> finance.uni-mainz.de> writes:
> 
> Hi,
> 
> how to measure the goodness of fit, when using the rq() function of
quantreg?
I need something like an R^2 for
> quantile regression, a single number which tells me if the fit of the whole
quantile process (not only for a
> single quantile) is o.k. or not. 
> Is it possible to compare the (conditional) quantile process with the
(unconditional) empirical
> distribution function? Perhaps with a Chi^2 or Kolmogorv-Smirnov Test?
> 
> Thanks for feedback.
> 
> Martin
Please see:

http://www.econ.uiuc.edu/~roger/research/rq/rq.html

Anupam.