# Cointegration tests

I'm trying to figure out how to perform cointegration tests in R between 2 time series. I'm using po.test from the package tseries and ca.po and ca.jo from the package urca, and I have a couple of questions:

• If I use PO which is non symmetric and one time series is stationary and the other isn't, I get opposite results. Is this normal? Also related I guess, if I have two time series and don't know yet whether they are stationary, can I use cointegration tests right away or do I need to pretest for unit roots?
• When I use po.test and ca.po I get very different results, why is this?
-
This may help. stats.stackexchange.com/questions/14615/… –  vinux Apr 20 '13 at 3:13