Identification and sensitivity

Identification/Sensitivity
1

Hello!
I got stock in performing the identification test before estimation of my model. When I try running the code with identification and sensitivity analysis command, I get the following:

 0.2% of the prior support gives unique saddle-path solution.
99.8% of the prior support gives explosive dynamics.

Smirnov statistics in driving unique solution
Parameter            d-stat         p-value
phipi                 0.968           0.000




Smirnov statistics in driving instability
Parameter            d-stat         p-value
phipi                 0.968           0.000

==== 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

I cant’ fix the problem . Any hint is welcome.
Thanks

2

Typically, your model not satisfying the Blanchard-Kahn conditions for most parameter values indicates a timing problem, i.e. a mistake in the model setup. The less likely second possibility is that your prior region was chosen poorly and only allows for a narrow range of parameters that satisfy the BK conditions.
Note that if you model features a unit root, you can get a behavior like this during identification analysis if you forget to specify the diffuse_filter-option.

3

I think my model is fine. Calibrated version and estimated versions work (phi_pi=1.212, phi_y=0.945). But when I do sensitivity analysis, I get:

19.4% of the prior support gives unique saddle-path solution.
80.6% of the prior support gives explosive dynamics.

Estimated parameters block in estimated version is

estimated_params;
phi_pi,1.212, 0, 2,UNIFORM_PDF,1.212,0.2;
phi_y,0.945, 0.5, 1.5,UNIFORM_PDF,1.212,0.2;

Why dynare_sensitivity; can’t find prior support that gives a unique solution? Is it because the algorithm searches randomly? Thanks!!

.

4

Yes, this does random sampling over the prior.