Steady State Computation - Larger Models

Hi all,

This question is related to the steady state computation of larger DSGE models. At the moment I solve these model with pen and paper, which can be a very inefficient and long process especially when only one specific feature has been introduced into the model. For example, at the moment Dynare produces the well-known error message that model and steady state do not match, but I am struggling to find my mistake(s). One missing parameter could already cause this. I was thinking to a run a solver on the model equation but then we still don’t know whether the steady state is correct.

Hence my question to the community, do you use pen and paper or do you use a solver approach?

Many thanks.



Because of exactly the type of situation you outline, I use pencil and paper. In the situation you describe, you don’t know whether the mistake is in your model or the steady state computation. If it is in the model and you run a numerical solver, it will then find the steady state for a wrong model. Having the steady state analytically allows you to debug based on the discrepancy.

Thanks Johannes, that’s what I was thinking.