tau vs. z statistic for engle-granger and phillips-ouliaris

[maragloria]

Hi there,

I am running engle-granger and phillips-ouliaris cointegration tests on 2 variables for about 25 countries (not panel).

I have weekly data and I'm doing rolling regression of 2 years to perform the tests to see for which periods the variables are cointegrated or not.

I am saving the results of both the tau and z statistic for each test. I would like to know if there's a preference for one of these ADF test statistics when their conclusions diverge.

I could not find the reference (Stock 1986, Hayashi 2000) in the reference list of the User guide to further guide me.

Thanks in advance,

Mara