Estimation with structural breaks (perfect foresight)

Estimation SMM/GMM/IRF-Matching
1

Dear all,

I need to estimate a model using SMM and assuming perfect foresight. I want to allow for structural breaks in 3 parameters. Namely, to allow these 3 parameters to change over time (once every 25 years). From what I understand, this is not feasible in Dynare, am I wrong? However, all the posts I read concerned stochastic simulations. Do you have any advice on how to do this?

Thanks in advance.

2

Can you elaborate on what the matching targets are? From what you describe, a structural break may be handled via the

perfect_foresight_with_expectation_errors_setup;
perfect_foresight_with_expectation_errors_solver;

commands. The changes in parameters would be unexpected shifts in the exogenous variables.

3

Thanks for your answer. Let me rephrase my doubt. How can I estimate my time-varying parameters with SMM if they’re defined as exogenous variables? Isn’t it incompatible with estimated_params block? That was my doubt. Even if I treat my parameters as parameters and model their change as an exogenous variable (p + e_t) still I don’t obtain what I want, do I?

It is a model of the academic system. I can observe human capital supplied by scholars, n. of active scholars and GDP. The 3 time-varying parameters represent resp. the productivity of the acad. sector, a utility shock to join academia relative to doing another job and TFP. Ideally, I want to obtain an estimate of, say, acad. productivity in each simulated period.

I hope I didn’t get you wrong.

4

Estimation with

doesn’t work with perfect foresight in any case. If you do your own matching, the distinction between parameters and exogenous shocks should be unimportant. Both are taken as given by the agents in the model.

6

Dynare does not support conditioning on t. It is not a defined object.