[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: bootstrapping and time series
I'm sure Stas (and Jeff Pitblado) are right here.
However, do note that there is literature on special
bootstrapping methods for time series (see e.g. Politis _Statistical
Science_ 2003). The point is that Stata's
-bootstrap- implements none of these methods.
How far such methods extend to panel data I
do not know.
> DON"T TWEAK WITH THE BOOTSTRAP!
> I would argue that statistically the bootstrap of the time series as
> if it were i.i.d. is at least misleading. The bootstrap resamples
> should resemble the data generating mechanism.
> Suppose you had I(1) process originally, and you bootstrap it naively.
> Then what you have is not a time series anymore, but just a bunch of
> heteroskedastic observations, and an AR(1) process fit to this data
> would give you zero correlation.
> If you bootstrap your dynamic panel data naively and then run
> -ivreg-... well, you are on your own with your referees. If you are
> lucky, and they don't know much about the bootstrap, and your paper
> gets through... then somebody will have a chance to publish a paper
> just to rebut your results. Nobody needs this sort of a professional
> Stas Kolenikov
> * 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/
* For searches and help try: