Adding deterministic shock

Hello,
I have added a deterministic exogenous shock and mixed with stochastic shocks. The deterministic shock is iR_det which duly appears in varexo_det block. I do not see any impulse response coming out of it. Not sure what is wrong.

I am attaching teh mod file. Any help will be appreciated.boj_project_July24a.mod (6.8 KB)

Hi,

That’s normal, Dynare does not display an impulse response function for this expected shock (see page 13 in the reference manual). If you want to see it’s marginal effect you just have to do a deterministic simulation with the simul command (in another mod file).

What is bothering me is that I do not see any differences in the displayed IRFs when I change the specification of this expected shock (or remove it). That is not what I expected. I need to check on a simpler model…

Best,
Stéphane.

Dear Stephane,
Many thanks for the prompt response. Exactly so. What is then the utility of bringing a deterministic shock in a stochastic model. I did exactly what you said in a different mod file and I get results with simul. The issue that I am investigating is what changes when I mix stochastic and deterministic shocks. The stochastic property should remain unaffected by addition of deterministic shocks which are known as news shocks.

Perhaps I don’t understand something here.

PB

I must confess that I never used these deterministic shocks it a stochastic context (I do not see the value added compared to the simulation of perfect foresight models). Looking around, I have the impression that these shocks will only affect the output of the forecast command (I do not understand why we do not integrate the consequence of these expected shocks in the IRFs or stochastic simulations). Hence, apparently, the way to go for you is to use this forecast command. There is an example in the reference manual (and mod files are available in the tests subfolder of Dynare, on Github). But my preferred solution would be to use a perfect solver (which is possible I think to solve the problem you seem to have). As far as I understand, but I may miss something, the only advantage I see is that the forecasts produced with these expected shocks includes confidence bands related to the (other) stochastic shocks.

Best,
Stéphane.

@pb2415 What exactly are you trying to achieve? At first order, you can implement news shocks either as a stochastic or a perfect foresight shock. Why are you going for varexo_det?
You might find https://github.com/DynareTeam/dynare/blob/master/tests/forecast/Hansen_exo_det_forecast.mod useful.

Many thanks for thinking about my problem.

I am trying to understand how a perfectly anticipated reduction in policy interest rate from positive to negative impacts the economy in an otherwise stochastic environment. Your lead about forecast command is useful. I expect that the impulse response function of relevant macro variables with respect to stochastic shocks will undergo shift when such interest rate regime change happens in a preditable manner. The only complication is that such regime change will lead the economy traverse from one steady state to another. In other words, it will cause the stationary distribution of endigenous variables to shift in a preditable manner, i am not sure how I can implement this in dynare.
PB

What do you mean with

I expect that the impulse response function of relevant macro variables with respect to stochastic shocks will undergo shift when such interest rate regime change happens in a preditable manner?

In a linearized model the point in the state space where you analyze the IRFs does not matter. Thus, if you consider a low interest rate regime or a high interest rate regime will not affect the IRF to a stochastic shock. If you want to consider the path of the joint stochastic and perfect foresight shock, it will simply be the sum of the two components. What you can easily do in this case is to add a untit root news shock instead of a varexo_det and then simulate the model with the permanent news shock and the surprise shock hitting simultaneously. You would have to use the simult_ command for this. See https://github.com/JohannesPfeifer/DSGE_mod/blob/master/RBC_news_shock_model/RBC_news_shock_model.mod

Thank you. Will try to grasp this.