[R] about kpss.test()

Achim Zeileis Achim.Zeileis at wu-wien.ac.at
Thu Mar 10 13:14:01 CET 2005


On Wed, 9 Mar 2005 14:32:09 -0500 (EST) Weiguang Shi wrote:

>  --- Achim Zeileis <Achim.Zeileis at wu-wien.ac.at>
> wrote: 
> > > First of all, could you tell me what the "KPSS
> > Level" 
> > > in the output of the test means?
> Thanks. But I meant to also ask about the number after
> that, e.g., 
>     KPSS Level = 0.0027

If you wouldn't have deleted my reply to that question, you could still
see the answer! It was:

  <quote>
  It means that you tested for level stationarity and gives you the test
  statistic.
  </quote>

Note the second half of the sentence.

> > Please check the references of kpss.test or any book
> > on economectric
> > time series analysis for which alternatives the KPSS
> > test was designed.
> Will soon do.
> 
> > Hint: it's not designed for detecting cyclical
> > variations.
> OK. I guess there must be some pre-requisite on the 
> data before kpss-ing it.

In particular it means: There are infinitely many conceivable ways of
deviation from stationarity and the KPSS test is only good at detecting
certain kinds of deviation (which do not include cyclical variations).
Z




More information about the R-help mailing list