Ramsey model and steady state problem

Hello everyone,

I am trying to compare the effect on welfare of different monetary policy rules, for this, I will use conditional welfare (starting in the deterministic steady state). As a reference regime, I have the welfare derived from the Ramsey policy with one instrument (the nominal interest rate), but the Ramsey policy has a different steady state than the other rules. Although dynare returns the conditional welfare of the Ramsey model, I’m not sure if I can compare the conditional welfare of Ramsey with the conditional welfare of the other rules for calculating consumption equivalence because the steady states are different. Is this possible?

Thanks in advance for your comments

You need to ask yourself what the comparison is that you are trying to do. If the question is what welfare would be if you switched to optimal Ramsey policy today and were still stuck at the old steady state, then obviously you need to compute conditional welfare at the old steady state.

Thanks for your reply

  1. To see if I understood correctly, you say that if I want to compare welfare starting at the deterministic steady state, I need to obtain the conditional welfare of the Ramsey model (conditional on the deterministic steady state not the Ramsey steady state)? If so, is there an example of how to calculate the welfare of the Ramsey model conditional on the deterministic (old) steady state? I guess the conditional welfare calculated directly by dynare is conditional in the Ramsey steady state.

  2. Another question related to the Ramsey model. My model is in non-linear form and following the dynare manual I calculated the steady state solution CONDITIONAL on the value of the instruments (the nominal interest rate, R), and used initval so that dynare calculates the steady state value of R, however, I notice that the value of the steady state is very sensitive to the initial values assumed for R, how can I make sure that I choose the correct initial value for my isntrument?

Thank you again for your time.

  1. You can set initial conditions for conditional welfare in the Ramey-context using histval (see the manual). Otherwise, the steady state of the new regime will be taken.
  2. That suggests there are multiple Ramsey steady states. Ideally, you are looking for the globally best one. That often requires testing different starting values.

Thank you professor.

  1. Just to clarify, if I want conditional (deterministic steady state) welfare for Ramsey model I can use histval, however, in a post from previous years you commented that “the policy computed at order=2 is indeed the optimal one, but that the planner objective values reported are not the full second-order approximation” and that “You can get the correct Ramsey policy at second order. But what is not yet implemented is the evaluation of the planner_objective”

In the latest versions of dynare can we rely on the second order “planner_objective” calculation? I looked in the release notes but couldn’t find anything about it.

  1. To finish, what factor or value do I have to look at to know which is the steady state that guarantees the global maximum?
  1. That statement does not apply to Dynare 5.2 or later.
  2. You would have to look at the unconditional welfare.

Hello professor, it’s me again, I followed all your advice, however you told me that from dynare 5.2 onwards the conditional welfare of the Ramsey model is already reliable, however I did the test with dynare 5.1 and the conditional welfare is identical to what i get in dynare 5.2, is there any reason for that?


Dynare 5.1 had a bug at order=1. For order=2 results were correct starting in Dynare 5.0.