Shutting down shocks during specified periods in estimation

Estimation
1

Hi everyone,

I’m trying to replicate in Dynare one feature of the New York Fed DSGE, in particular, the method of handling the ZLB and forward guidance based on Laseen and Svensson (2011, https://www.ijcb.org/journal/ijcb11q3a1.htm).
In brief, the method consists of augmenting the policy rule with anticipated shocks. Also, model-implied expectations of the policy rate are matched with market expectations of the Fed funds rate (used as observables). These anticipated shocks are “simply” set to 0 in every period when the ZLB is not binding, without resorting to regime-switching, as described in this section: https://frbny-dsge.github.io/DSGE.jl/latest/advanced_usage/#Handling-of-the-Zero-Lower-Bound-(ZLB)-1.

Does Dynare allow fixing the standard deviation of a given shock to zero during specific periods of an estimation sample?

Thanks,
Giorgio

2

So you are trying to estimate a model with the ZLB?

3

Exactly. The sample would cover (also) the ZLB period in the US.
I’m able to introduce anticipated shocks in the policy rule and to match model-implied with observed Fed funds rate expectations. However, the standard deviation of those shocks is obviously estimated to be positive throughout the whole sample, whereas I would like it to be 0 in the non-ZLB periods (as is achieved by the NY Fed DSGE code in Julia).
I was wondering if this is possible in Dynare too, without using more complex tools such as Occbin.

Thanks,
Giorgio

4

You should be able to do this in the unstable version employing the heteroskedastic_shocks-block.

See https://git.dynare.org/Dynare/dynare/-/commit/1645f38269e445e6a686a6889420012106794985

5

Thank you @jpfeifer. I was not aware of this new feature, and it worked!