Johansen Test for Cointegration

[aacasiobuk]

Hi,

I'm currently running test for cointegration between the bank rate, libor and the standard variable rate between 2000 and 2007 and 2000 and 2010. However, for the longer sample period, no cointegration exists between libor and the standard variable rate. The papers that I am following seem to find the same conclusion but continue their analysis anyway and I am trying to work out how, as I was under the impression that if cointegration cannot be established then one cannot proceed with an ECM. I've attached links to the two papers.

http://carecon.org.uk/DPs/1016.pdf

http://www.socialsciences.manchester.ac ... bcr141.pdf

I'd be really grateful for a swift reply as this has been troubling me for some time.

I was also wondering whether or not I should set the variables I am assuming to be weakly exogenous as exogenous in the test - as this gives very different results.

Thanks