Re: [Gretl-users] Johansen question
On Sun, 1 May 2011, Talha Yalta wrote:
I want to add a table which will show Johansen tests can give
different results in small samples when different parameters are used.
In the table, I just use the trace test results (since this is just to
prove a point) and choose lags 1, 2, and 3. I also use the default
case as well as the --ct option. This results in 6 combinations.
Here is how I proceed:
For a bivariate case, if the trace test rejects c=0 and does not
reject c=1, I report c=1.
OK.
If it is the other way around, then I report c=0 as the test
result.
On the trace test, that result would seem anomalous, and
indicative of a small-sample problem. If there's "enough evidence"
to reject c=1 (with an alternative of c=2), then concluding that
c=0 is problematic.
As you know, sometimes the results can be contradictory so that c=0
and c=1 are rejected (or not rejected) simultaneously.
That doesn't seem contradictory: if both c=0 and c=1 are rejected,
that favors the hypothesis that both series are stationary. If
neither c=0 nor c=1 is rejected that suggests they're both
non-stationary, and not cointegrated.
Allin Cottrell