Hi All,

I am simulating a perfect foresight model, and one of the parameters(say alpha) of the model is a function of the endogenous variable. Here are my questions:

(1) Using a perfect foresight solver, will the value of the parameter (alpha) change as the steady-state changes?

(2) If yes, is it possible to declare the parameter(alpha) such that it does not change due to the shock to the endogenous variable?

In my file, I only have a value for the parameter (alpha). However, I noticed that the steady-state values before and after the shocks for some of the variables differ from what I got when I use fsolve to compute for the policy change.

In my fsolve file, I declare the alpha to be a function of the initial value of the endogenous variable, such that when computing for the new steady-state after the shock, I assume the alpha value from the previous steady state. Does this have any implication in the computation of steady-states in dynare?

I would need to see the codes. What do you mean with

? A parameter is a fixed number. Do you mean that the parameter value is set to achieve a target in a particular steady state? Then it would be a fixed number. Whether Dynare will update the parameter to satisfy calibration targets will depend on the implementation.

Yes, exactly. The parameter is an “implied target”

Please check your email, I sent you the files.

Thank you

Whenever you call `steady`

, the steady state file will update all parameters that you set in that file/`steady_state_model`

-block. In your case, that applies to `kap_c`

.

Is it possible to use the “if” command to tell dynare to change the values of Nss, such that the first value of Nss produces a steady-state that solves the model and fits the initial value; and the second value of Nss solves the model and fit the end value?

What exactly do you mean with that? The `initval`

- and `endval`

-blocks followed by `steady`

are for exactly that purpose.

The file I sent to you, the initval and endval from fsolve do not solve model in dynare.

Also, I thought if I have the initval and endval, I don’t need the steady state bloc. But when I remove the steady state bloc it produce error output of NaN in the residual. Am I doing something wrong?

I’ve seen examples where they excluded the steady state bloc when they have the initval and endval.

Dr @jpfeifer

I have figured out the problem. The initval and end values solve the model well.

Thank you for your assistance