Engle-Granger cointegration test

[hsmellin]

Hi,

I am wondering if this is valid:

1. Take the residuals from an OLS regression and make a series, say resid01
2. Then inputting the code:

"coint(method = eg, trend =0) resid01"

I have been told i can do this with the PO test and wondered if the same held for this method of test.

Thanks,
M

[EViews Glenn]

You can do a unit root test on RESID01, but the critical values will be wrong. The best thing to do is to create a group with the series of interest, and then do the Engle-Granger or Phillips-Ouliaris test directly.

[hsmellin]

Thanks Glenn.