Hi, I’m trying to incorporate a penalty function in an otherwise standard model with borrowing constraints (similar to Mendoza 2010): a_{t+1} <= k_t (the total amount borrowed cannot be higher than a fraction of the beginning of period capital). The penalty function that I am using is:

I’m solving the model as deviations from steady state (exp), so my Dynare code for the penalty looks like: \frac{b}{c}e^{c(exp(a)-\alpha exp(k(-1)} and I was wondering whether this is correct since the results using this penalty function are not different from the ones obtained from assuming that the constraint is always binding (of course c is high enough). Thanks.