Hi! I am trying to stochastically simulate a basic (3-equation) NK model with the ZLB using DynareOBC. The IRFs for the model are the same when simulating in Dynare and DynareOBC, hence I assume the ZLB must not be binding. When I increase the shockscale or the size of the shocks to make the ZLB binding, I get the error message that I should try to increase the timetoescapebounds, which I have done, but I still get the same error message. [Sidenote: I did manage to get the bbw2016.mod (found in the OBC example files) to bind at the ZLB using the shockscale.]
nk3z.mod (747 Bytes)
For DynareOBC questions, you should tag @cfp. He is the one to ask.
Sorry for the slow reply.
If DynareOBC gives that message, it means there’s no solution to the problem as you have set it up. It may always be the case that increasing TimeToEscapeBounds would then find a solution, but you are probably not interested in solutions that spend 200 periods at the bound in any case (so there’s not much point having TimeToEscapeBounds much larger than 200).
I expect that there will be an intermediate value of ShockScale for which model will have a solution.
The reasons why your model will not have a solution more generally are outlined in my OBC existence and uniqueness theory paper. You have a lot of endogenous persistence here, which makes existence less likely. I expect that responding to expected inflation and output gaps also makes existence less likely. Try a more standard (contemporary) monetary rule. Or try including a small response to the price level.
@cfp No worries and thank you for your detailed response!
I managed to make the ZLB bind and have the model recover again from the ZLB by (doing as you suggest) a) removing the endogenous persistence terms, b) having a contemporary Taylor rule and c) responding to output growth rather than output gap (as you discuss in your OBC existence and uniqueness theory paper) in the Taylor rule.
(I have attached the amended mod file for anyone interested.)
nk3z.mod (688 Bytes)