Clement Montane Reyes and Zivot Andrews Unit Root Tests

[John Baker]

Hello All,

I have downloaded the zandrews and clemao1 / clemao2 unit root tests which have been developed by Stata users.

dfuller and pperron and all of the xtunitroot tests (except one) have a null hypothesis of unit root.

dfuller for example tests the null hypothesis that beta = 0 where beta = (rho - 1) and rho is the autoregressive parameter.

The zandrews help page doesn't indicate the null hypothesis. And the clemao1/clemao2 help pages say "The test considers the null hypothesis that (rho - 1) is different from zero" which would seem to suggest a null hypothesis of stationarity.

Can anyone confirm the null hypotheses for theses two tests?

Many thanks in advance.

John.

[Anat Tchetchik]

John,
I'm having the same question now, did you manage to resolve the issue?
Anat

[Anat Tchetchik]

OK, according to Table 1 in the attached paper. the null indeed support hypothesis of stationary variable. So rejecting the null indicates a unit root var (as oppose to the case -dfuller-)
Anat

[Zied OUELHAZI]

Thank you very much for your help. I thought from the beginning that the null was the presence of unit root

[Zied OUELHAZI]

Guys look at this paper

[Anat Tchetchik]

Thank you Zied,
I think the answer provided by me in my post from June 2014 was wrong. according to paper attached by Zied, "The language of choice for time-series analysis?" p.57 (by C.F. Baum) the null hypothesis of a unit root.
Sorry for misleading,
Anat

[JOHN MGANGA]

Hallow members!
I would like to seek for assistance on how to test for stationarity by using zivot- andrews test in stata and how do we interpret the output. I am not familiar to this test.