The deviation in the IRF appears to be much smaller than it should be

Stochastic simulations Semistructural Modeling
1

Dear community,

Here is an endogenous variable, df, in my DSGE model. Its equation is as follows:
1
The two variables highlighted in red are endogenous variables, while the remaining ones are parameters. The shock response of the two variables is as follows:
2
3
Based on the deviations of the two variables in the initial period, I have calculated that the deviation of variable df in the first period is 0.509543.

4
41335×146 31.6 KB

However, the impulse response curve of variable df indicates a deviation of 0.0018 in the first period.
5
I am uncertain as to why there is such a discrepancy and what may be causing it. Could you please give me some help? Your help would be greatly appreciated.

2

Can you try whether this is an approximation issue? The gap should become smaller for smaller shocks and if you increase the approximation order.

3

Dear Professor,

     Thanks for your advice.
     Here are the equations for the varable kb and the shock:

11
In the case described above, the shock is set as:
12

     Now, as I increase the approximation order from 1 to 2, the following hints were obtained:

13
What does this mean? How to fix it?
And the gap become smaller indeed when I set the shock smaller.
I look forward to hearing from you.

4

Dear Professor，

     I changed the code from "stoch_simul(order=1,irf=100)"   to "stoch_simul(order=2,irf=100,pruning)", Then the following impulse response curve was obtained:

14
However, the gap is still much smaller than it should be.
I’m really comfused.

5

You may need pruning and a higher number of replic at higher order.