My model has two steady state, after the permanent shock, the model turn to the steady state characteristics by endval. Because there is some occasional binding condition, I have to use occbin toolkit. However, I find occbin can only perturbation around the original steady? How should I let Dynare make perturbation around endval steady state?

This sounds like a case where you need to use a `lmmcp`

perfect foresight solver, not OccBin.

Thanks a lot professor!!! But the simulated plot seems to be weird, it jump to the new steady state in the last few periods, instead of gradually approach it. Do you know why it the case? I have upload my mod file and simulated plot. Thanks very much for your reply!

transition_risk_NK_e_lmmca.mod (5.1 KB)

You may also need this m file to run the mod file.

transition_risk_NK_e_steadystate_helper.m (2.5 KB)

Can you get the model to work without the OBC?

No, it doesn’t work. Even if I use the model without OBC, the model is still unstable in the last few periods like the above figure shows.

Then you should try to investigate why that is the case. Have you tried smaller shocks/perturbations of the model?

Thank you a lot professor, I’m still wondering.

The smaller shocks also don’t work. The last few periods to new steady state are still not smoothed.

As my model solves the transition path between two steady states, the perfect forward looking method, which exactly solve the equation system, is suitable here. But perturbation only can be applied around steady state, I think it only useful for IRF of a temporary shock, and can’t solve transition path like my model. Am I right?

Thanks a lot!!!

That is the part that bothers me. Something is weird in the model. Did you give the model enough periods to smoothly transition to the new steady state.

Note that you can have permanent shocks in a stochastic perturbation simulation as well. Unit roots are allowed. The issue is that you cannot capture nonlinearities in this case.