I am trying to use IRF matching to estimate the parameters of a model that sometimes runs into problems with the Blanchard-Kahn and/or the rank condition. I am estimating also the parameters of the monetary policy rule, and sometimes the combination of these is such that the model crashes. The way I deal with it is to use a penalty function that adds a large number X to the object I am trying to minimize whenever this happens.
My problem is the following: I use a small loop that adds X whenever info(1)>0. However, it sometimes happens that I get a message that the rank condition is not satisfied, yet I obtain info(1)=0. This means that I end up with a set of estimated parameters for which my model does not run.
How can it be that I get info(1)=0 when the rank condition is not satisfied, and is there a way for me to rule out or penalize these cases? I am using Dynare 4.5.4. Thanks in advance!