Since I am coding some linear New Keynesian models in python and solving them under rational expectations (using the method of undetermined coefficients/linear time iteration), I was wondering whether anyone codes non-linear rational expectations models (e.g. with the ZLB)? I heard these are much more complex to solve and it is recommended to just solve them using dynare + OBC / occbin.

The question is, is it generally sensible to solve non-linear models in python/matlab by coding the solution algorithm yourself? Or is this too complex and you would recommend using dynare + OBC / occbin?

  • $\begingroup$ Could you please elaborate on what exactly you mean by "sensible"? :) I mean what are your goals, why would you prefer not to use dynare + OBC if that is an option? $\endgroup$
    – Giskard
    Jul 31, 2021 at 14:26
  • 1
    $\begingroup$ The reason is that using dynare+OBC you are not solving the model yourself, but using a "black-box" solution method where the software does it all for you. It would be preferable to show that I can do it myself, if feasible with reasonable amount of work. If it is too complex, I will have to resort to dynare + OBC anyway. I want to implement the ZLB in a RE model and show what implications this has, i.e. derive some IRFs for different types of shocks e.g. $\endgroup$
    – Suttungr
    Jul 31, 2021 at 14:29


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.