Identification does not run

Dear all,

I get the following error message when running the identification command in Dynare (a similar problem appears when using dynare_sensitivity command):

==== Identification analysis ====

Testing prior mean


Parameter error:
The model does not solve for prior_mean with error code info = 3

info==3 %! Blanchard & Kahn conditions are not satisfied: no stable equilibrium.

Try sampling up to 50 parameter sets from the prior.


Identification stopped:
The model did not solve for any of 50 attempts of random samples from the prior

However, when simulating the model by my own setting parameters at the prior mean, it works fine and the BK conditions are well respected. Likewise, when I launch the model estimation (also starting at the prior mean, or at any other reasonable value), it also works and converges to a solution. Therefore I do not understand why I am having the above error message using the identification command. I also tried using different model specifications / different combination of estimated vs. calibrated parameters, and it still did not work.

A similar post (from 2014) is the following: Dynare forums • View topic - Identification failure , where it is mentioned that there is a bug leading to this issue. Does the same bug persist in Dynare 4.5.3, which I am using ?

Many thanks for the help!

Does your model have a unit root?

Yes it does. It is an optimal policy problem where some variables (such as Lagrange multipliers summarizing past commitments of the planner) have a random walk behaviour in the linearized solution.

Then you must set the diffuse_filter-option.

Thanks! Of course it is working fine now