I use dynare 4.2 for my code, but still get NAN for the theoretical moments of most variables such as output and consumption, for both stoch_simul(order=1) and stoch_simul(order=2). From the past posts on the forum, it seems that this problem has been fixed by 4.2. What would be the mistakes that I possibly make?

Moreover, IRFs can be generated under some simple specification, but will fail for a more complicated specification of policy rule.

If you have a unit root in the model, these moments are not finite and thus set to NaN. In contrast, stationary variables like inflation or hours should have finite moments.

Indeed, I have a NKPC model with Calvo pricing and used the nonlinear modelling in dynare. Since the central bank controls the monetary aggregate growth rate, money is essentially a unit root. After I write the model in terms of the real money balance, I finally get some results, but they are unreasonably large.

Since the examples I found on the dynare website are mostly linear, are there good examples to demonstrate the dynare nonlinear code on a NKPC model with Calvo pricing?