question on the two news shock dynare examples

[zhanshuo]

Dear Prof. Pfeifer,
I have been confused about two examples mods on news shock.
one is https://github.com/DynareTeam/dynare/bl ... recast.mod.
which uses both varexo_det and eps_a_antic(-7) to describe an anticipated shock in time 8.
another is https://github.com/JohannesPfeifer/DSGE ... _model.mod
where it uses the lagged value of a stochastic shock to describe an anticipated shock in time 8.
Here my question is, what is the difference between using a deterministic way "varexo_det" to describe news shock ,and a stochastic way using the lagged value of a stochastic shock?
Thanks.

[jpfeifer]

The second one is the generally correct one. The first file is Dynare's unit test, checking that

Code: Select all

varexo_det

works correctly. It makes use of the fact that due to certainty equivalence at first order a stochastic news shock and a perfectly anticipated deterministic shock are the same. But that is not true at higher order. The stochastic version is also more flexible.

[zhanshuo]

thanks a lot!