Bayesian estimation with unit root

[Cro]

Hello

I have a system of equations that contains a unit root (not a DSGE model). I have included intercepts to account for drift. The model check indicates that the steady state is not defined and estimation fails. When I block out the intercepts, the estimation proceeds. Alternatively, when I do not block out the intercepts but change the AR coefficient from 1 to 0.999, the estimation also proceeds. Would you know if this is a Dynare issue or if the system is not suitable for estimation in Dynare?

[jpfeifer]

Please have a look at

A nonstationary model for output gap with a const in drift

The problem is that Dynare tries to approximate around the steady state, which does not exist if there is a drift. But as the model is already linear, the approximation point does not really matter. For that reason, you cannot use an initval block, which gives just the starting values for the steady state finding. Rather, you need to provide a “steady state” via the steady_state_model block and make sure that the correctness of the “steady state” is not checked via the nocheck option. The diff…

[Cro]

Thanks. Seems to be working now.

[Cro]

Hello. Could I ask another question not related to unit root? Say I have a constraint where for example a1+a2/a3=1. (1) Is it correct in model specification to set #a1=1-a2/a3, and estimate a2 and a3? (2) if this is correct, how can I get mean, median and other statistics for a1?

[jpfeifer]

You can use a posterior_function, see examples/Gali_2015.mod · master · Dynare / dynare · GitLab