Ramsey_policy command - some important clarifications

Dear all,

A couple of clarifications to ask on ramsey_policy/ramsey_model commands in Dynare 4.5.

  1. In the online manual, it is stated that only 1st order approximation of the Ramsey set of FOCs is available with ramsey_policy. However, when setting ‘order=2’ as option, Dynare seems to run. Just wondering if it produces a 2nd order approximation this way.

  2. If that is not the case, would it work using the sequence ramsey_model and stoch_simul (order=2)? With the first command producing the non linear Ramsey FOCs and the second performing the 2nd order approximation over the the so produced set of Ramsey conditions?

  3. I noticed that the indication of the planner instrument(s) is only optional. I was therefore wondering what the ramsey_policy does when no instrument is indicated? Is the ramsey_policy In such case just choosing the decentralised economy equilibrium, therefore returning the same results as stoch_simul?

  4. I notice lead/lag variables are not accepted in the Planner objective function. This may be a limitation when Household utility contains habits and we want Planner objective to be same as household. To work this around I just defined a new variable, say Fake_t=C_t-1, and then inserted Fake_t in the objective in place of C_t-1. That seems to work. Do you anticipate an issue with this solution?

Many thanks,

  1. Yes, it will produce a second order approximation, but the evaluation of the objective function is still missing. See https://github.com/DynareTeam/dynare/issues/564
  2. This would be equivalent to what is done at 1.
  3. No, you do not need to specify an instrument to compute an optimal allocation. You can see this e.g. in Gali’s textbook where optimal policy is directly computed without going through the nominal interest rate. Having an instrument only simplifies perturbation solutions by making it easier to find the steady state to the Ramsey problem via the provision of a conditional steady state file. Conditional on that steady state, finding the steady state of the Ramsey problem is just solving a linear equation system as opposed to the nonlinear system you would need to solve otherwise.
  4. No, this is the suggested way of handling this problem. See Questions about Ramsey optimal policy