Cointegration

[alex9]

Here is the problem
I have to test for cointegration between a variable Y and another one X. Y has a deterministic trend while X is a I(1). (I used the ADF test). I know that if you want to test for cointegartion between two I(1) variables you should run a model with them and test if the residuals have a unit root. So I'm wondering wheter the procedure is the same for a variable with a deterministic trend. Should I compute something like
y c @trend x
and then create a residual series and use ADF test?
Thanks in advance

[EViews Glenn]

Yes, you can do that, but the inference on the unit root test won't be correct since the critical values differ. The test generally goes under the name Engle-Granger or Phillips-Ouliaris depending on whether you do an ADF or PP unit root test. EViews 7 computes this kind of single equation cointegration test for you.

[alex9]

thanks for your answer.
Yes, I have eviews 7 and I used eviews to test cointegration.. I mean Estimate equation> Method: Cointegrating regressor. How should I interpret the results? I imagine I should look at p-value but I'm not completely sure

[EViews Glenn]

You should look at the manual. I think the index has an entry for Engle-Granger cointegration test. It'll describe how to perform the test from a group of series, or from an estimated cointegrating regression equation.

[startz]

If you meant in your original post that Y is I(0) with a trend (I wasn't sure if that's what you were saying), then the whole deal is off. An I(0) and I(1) variable can't be cointegrated.