Hello everyone!I´m trying to conduct a cointegration analysis (Engle-Granger two step method) on some pair of stocks. If possible, I would like to knowyou do you guys cope with these issues:1) When I run the regression on the two price series, how can I decide if I need to consider a constant or not ?I mean, stockA = a + b*stockB or stockA = b*stockB ? 2) Now, with my spread (or residuals) series at hand, I need to perform a unit root test.I know that I have to use the critical values from McKinnon (those values are available for example in S+FinMetrics and in R´s urca package). But how can I decide between the "no constant" and "no trend" critical values variants?3) urca´s qunitroot method allows me to specify the number of observations or to use the asymptotic value instead. When it is appropriate to use one or another?Thank you!

Your questions might be pretty basic for some. I suggest you post such question on student forum next time.You may wanna look into the following book. It might be able to answer most of your questions (if not all). Unit roots, cointegration, and structural change / G.S. Maddala, In-Moo Kim.