Dynare crash: varexo_det in a recursive model

[Stefan_Boeters]

Apparently Dynare crashes (with a Matlab error message that is cryptical, to me at least) when varexo_det is used in a purely recursive model (only backward-looking variables). See the attached example. As it stands, the example produces the crash, if you use the out-commented equation (with a forward-looking variable) instead, it works fine.

Admittedly, given that varexo_det is used to model anticipated, deterministic shocks, the combination with a backward-looking model does not really make sense. Still, it might appear occasionally (as in my case) if forward-linkages are switched off temporarily for testing purposes. It would be nice if someone found the time to introduce a telling Dynare error-message for this case.

[jpfeifer]

Thanks for pointing this out. In the future, you will get an instructive error message of the form

STOCHASTIC_SOLVER: Dynare does not solve purely backward models with var_exo_det.
STOCHASTIC_SOLVER: To circumvent this restriction, you can add a foward-looking dummy equation of the form:
STOCHASTIC_SOLVER: junk=0.9*junk(+1);