Re: Son [or DAUGHTER] of unit roots

[ecwfm@xxxxxxxxxxxxxxxxxxx (bill mitchell)]

John from Hong kong asks:
>
>Um... *low power* ADFs? I'm afraid I don't get you. Again, I'm betraying my
>ignorance, but TS is trend stationarity?
>
The problem with ADFs which take the DS hypothesis as the null is that they
are of low power. the functional point of this is that they find it very
difficult to distinguish TS processes with high autoregressivity from a pure
unit root DS processes.

And some more:

>If r2 is not as
>relevant as these other tests, is there some level at which r2 can *reject*
>a model (provided, again, that tests of residuals generate satisfactory
>results)?
>
Well and r2 does not come into rejection unless of course it is zero.
It tells how much variation in one variable is explained (using this
cautiously - the reason will appear in the next post i send!), by the group
of regressors.

And again:

>I think I've asked the question wrong. I mean, when you do the short-run
>equation using differenced variables plus the lagged ECM, do all of the
>variables appearing in the long-run equation have to appear, in differenced
>form, in the short-run equation as well?
>
>
What you are implying is that a variable that influences the "steady-state"
(caution here -!) only influences the dynamic path via the ECM term. That is
the short run dynamics of the two (or more variables) are orthogonal. What
sort of economic model would you have in mind where that could occur?

But the real answer is that if data-based restrictions are acceptable then
you can delete the higher order terms from the dynamic model, even if the
level still appears (implicit in the ECM).

hope this helps

kind regards
con, sorry bill
**************************************************************************
 William F. Mitchell            Telephone: +61-49-215027      .-_|\
 Department of Economics                   +61-49-705133     /     \
 The University of Newcastle    Fax:       +61-49-216919     \.--._/*<--
 Callaghan   NSW  2308                                            v
 Australia                      Email : ecwfm@xxxxxxxxxxxxxxxxxxx

 WWW Home Page: http://econ-www.newcastle.edu.au/~bill/billyhp.html
**************************************************************************