OccBin simulation problem

Occbin
1

Dear professor, I use the Occbin method to simulate ZLB, then the interest rate figure seems not smooth,

2021-12-02 10.49.04
2021-12-02 10.49.042880×1800 251 KB

Here is my mod filebaseline.mod (9.8 KB)
zlb.mod (8.8 KB)
run_zlb.m (3.2 KB)

2

Hi,
your constraint is wrong.

Instead of

constraint = 'i<1-1/beta';
constraint_relax ='i>1-1/beta';

it is better to use the shadow rate within constraints. E.g.

constraint = 'inot<1-1/beta';
constraint_relax ='inot>1-1/beta';
3

Thank a lot to you!!!

4

ascari.mod (4.3 KB)
Dear Professor, can You please, tell me what is wrong with my code where I want to impose a ZLB on my interest rate using Occbin toolkit.

5

Hi, first of all I am not a Professor. I am just a PhD Student.

You should check whether the periodic solution is a bug or a feature of your model.
My hunch is that there could be a typo somewhere in the equations.
For instance, if I change the calibration of alpha to alpha = 0.25; the steady state is no longer computed properly.

6

Thanks

7

Hi Max, sorry for bothering You again :sweat_smile:
I take the code from Christiano closed economy model and tried to impose a ZLB, but the code doesn’t work in a proper way, I guess, because, for example, when I set a shock value to be 1 the model crushes but when I put a smaller number it works. Can You, please, take a look and help me?
Thank You in advance
closed_economy_zlb.mod (2.7 KB)

8

It might be the case that the shock size matters for a solution to be feasible for the OccBin Toolkit.
You should also be aware of this issue if you work with a single shock in the first period.

10

Thank You for Your reply, is there a way to make the model work for any shock size?

11

But why should a 100 percent shock ever make sense? It would take forever to return to steady state and the ZLB not to bind.

12

Thank You Professor, I got it. Also, I have another question. Is it possible to the a welfare analysis using Occbin toolkit, since it solves the model using first order approximation?

13

Generally, I would say no, because welfare does often depend on higher order properties. But sometimes linear solutions suffice. It should depend on the model at hand.

14

Thanks