VAR, VECM, COINTEGRATING REGRESSIONS

[lnp3]

dear Group,

do VAR, VECM and co-integrating regressions (FMOLS, DOLS, COLS) always require FIRST DIFFERENCE STATIONARY variables ( they must also be cointegrated, with the exception VAR) ?What if all my variables are stationary (no unit root) AT LEVEL? Can I still use these estimation techniques?

I truly appreciate your assistance.

:(

[eobumneke]

If your variables are integrated at the order I(0) at their level forms, there is no point conducting co-integration test

[onursrmeli]

Hello,

I would like to make a model that estimate central bank interest reaction function. To do this, I took Taylor Rule as an based model. In this regard my variables are

Goutput -> I(0),
Ginf -> I(0)
on -> (0)
dollar currency -> I(1).

dollar currency is the only variable which is not stationary in level. The others are stationary in level.

In a such case, I know that I could not intestigate existence of co integration. But I want to estimate a model with using VAR model. Can I proceed in this way? Does an econometrics error happen?


Yours sincerely,