Solving Ramsey problem

Dear Team
we are working with relatively large model and trying to get impulse responses. We have several related problems, maybe you could comment on them, please?

We have a version where we derive FOCs, and then just use stoch_simul command to solve the model. It probably works OK, but to be sure, we solve Ramsey directly (let’s call it Ramsey):
we specify objective, instruments, and then again use stoch_simul. We also provide steady state values for all variables except Lagrange Multipliers.

  • it seems to us that in the Ramsey case the steady state is found with tolerance 1e-5. This is not suitable for our purposes - how can we fix this?

  • in some parameterisations, the steady state cannot be found at all, how can we provide starting values for LMs? This probably resolves the above problem.

Also, can I confirm that when plotting IRFs, dynare starts with second period, de facto. so we just need to take predetermined state variables, and add zero point in period one? What are the initial conditions for predetermined LMs? We think these are zeros as well, but could you confirm, please?

Many many thanks for your help

  1. Can you provide me with an example to look into the tolerance issue?
  2. For IRFs, the default starting point at order=1 is the steady state. That also includes the Lagrange multipliers.