I am trying to estimate a small DSGe model and I just encountered a problem that some more of uas have faced in the past given some previous posts.
Dynare reports that the number of “explosive” eigenvalues is equal to the number of forward-looking variables in the model and that the rank condition is verified but then immediately after that an error message is returned saying that the Blanchard-Kahn conditions are not satisfied and there is no stable equilibrium!

How can this be? And how can one overcome this problem?

Thanks for your reply to my question. What Paolo has written is of course very much sensible, indeed the initial values must be such that a unique and stable equilibrium exists.
Now,I have exactly 2 explosive eigenvalues for exactly 2 forward-looking variables based on the initial values that I have given and that should be enough for Dynare to conclude that the BK condition is satisfied. Yet,it does not say so.
Am I missing a point here?