Using a dummy variable in an observation equation

Hello:

I am wondering if there is a way to use a deterministic (time-dependent) dummy variable in an observation equation for estimation. One of the observable variables has some unusual changes during a specific time period and I think it is very difficult to explain it endogenously. So, I am thinking of removing those changes using a dummy variable. But it seems that I cannot simply add the dummy variable as an additional observable because it requires an equation that describes the dummy variable. Thanks in advance!

WY

It’s not possible in Dynare, I think.

An apparent alternative is: why not just deduct that deterministic dummy variable from your data…?

1 Like

I agree with @wupeifan. Dynare is not good in handling such deterministics. You may want to remove this component before estimation.

Many thanks for your reply. I wanted to separate a part of the change that could be explained by my model’s structural shocks and another that could not be plausibly explained by them in a joint framework.

A feature that allows a parameter or an exogenous shock to apply for only a part of the sample would be very useful. For example, this “anticipated monetary policy shocks” in some DSGE models apply only during the ZLB period. But I guess it is easier said than done… Thanks!

If it was just a parameter or an exogenous shock, then you could transform your data accordingly, no?

For example, this “anticipated monetary policy shocks” in some DSGE models apply only during the ZLB period

ZLB is another thing, and it’s beyond the classical “perturb around the steady state of a state space model” trick. In dynare, lmccp, DSGE_mod/Gali_2015_chapter_5_commitment_ZLB.mod at master · JohannesPfeifer/DSGE_mod · GitHub.