Bayesian IRF Matching


I was wondering whether Dynare supports Bayesian IRF Matching a la Christiano, Eichenbaum, Trabandt (2016 Econometrica)

Thanks so much!!

Not yet, but Dynare 6 will feature it. We are currently in the process of adding the routines, e.g. Options for Bayesian IRF Matching in method_of_moments command (!85) · Merge requests · Dynare / preprocessor · GitLab

Thanks so much! Is it an acceptable practice to give special weight to certain variables in (non-bayesian) IRF matching? Like I would like to target a specific spread that is important for my story even though it is not super precisely estimated?

In a similar fashion, is it permissible to put bounds on parameters?

I wrote my own IRF-matching code but I am not well aware of putting ad-hoc constraints.


As long as you are able to justify your weighting procedure, you should be fine. In our 2014 JME paper “Policy and the business cycle” we even augmented non-Bayesian moment matching with a “prior”.

1 Like