3 Equation NK model - 3 periods shock

Hi all,
I have the code for a 3 equations New Keynesian model. It is working fine.
Now, what I am trying to do is to simulate the IRF function for a 3 period shock in the monetary policy. I would like to have a 1% shock in the first 3 quarters. How can I do this? I tried to make a deterministic shock in eps_v variable, but it did not work. Any suggestions?

Thanks in advance!NK28.mod (1.5 KB)
NK_new_tombini.mat (1.2 KB)

I’ve been checking Dynare User Guide and I found the following explanation, for stochastic NK model:

In these models, shocks hit today (with a surprise), but thereafter their
expected value is zero. Expected future shocks, or permanent changes
in the exogenous variables cannot be handled due to the use of Taylor
approximations around a steady state.

Is there any workaround so I can have a 1% shock in interest rate for 3 periods?
Thanks again.


I think you might want to check simult_ command in the manual, if I recall correctly…

You need to use the simult_-function. See https://github.com/JohannesPfeifer/DSGE_mod/blob/master/RBC_news_shock_model/RBC_news_shock_model.mod

Hi. Thanks!
I will take a look!

Hi guys.
I made it work using simult_ function, but the result is not as I expected. I want to make a 1 p.p. shock in interest rate, that will last for four quarters, and then, starting from the fifth quarter, the interest rate path will follow the taylor rule. Is that right what I am doing in the attached code?
I was expecting an impulse response function following the shape of the following document - page 82 (chart 2). This is the inflation response to a monetary policy shock.

Any clue of what/ if I am doing something wrong here?

Thanks again

teste_choque.mod (1.4 KB)

  1. The file you uploaded does not contain a simult_-command.
  2. Please describe exactly the experiment you have in mind. It seems you want a model where the Taylor rule is not operative for some periods (as opposed to having a shock in a Taylor rule for some periods). That is only feasible in a perfect foresight context in Dynare as the model is not time-invariant.