Interpretation of cointegration equation graph of a VECM?

[Rene Macon]

I have estimated the cointegrating rank of a VECM with the "vecrank" command. The result indicates the existence of one cointegration relationship. I have then estimated the corresponding VEC to identify the parameters. The relevant parameters are significant and have the right economic signs. All kind of checks (veclmar, vecnorm, ARCH & GARCH tests of residuals, vecstable) provide reasonable results.

HOWEVER: Printing the "cointegration equation" with "predict coequation1 if e(sample), ce equ(#1)" yields the following graph:

Graph.gph

To my mind, this does not actually look like a cointegration relationship. How is the possible given the above results? Has anybody an explanation?

Thanks a lot for any help!
Rene