Computing theoretical moments using identification toolbox

Identification/Sensitivity
1

Dear Colleagues:

After running the identification procedure on the linearized code of my DSGE model, I have de following results:

==== Identification analysis ====

Testing prior mean
Evaluating simulated moment uncertainty … please wait
Doing 8289 replicas of length 300 periods.
Simulated moment uncertainty … done!

All parameters are identified in the model (rank of H).

All parameters are identified by J moments (rank of J)

==== Identification analysis completed ====

59.6% of the prior support gives unique saddle-path solution.
40.4% of the prior support gives explosive dynamics.

Smirnov statistics in driving acceptable behaviour
phi_y d-stat = 0.927 p-value = 0.000

Smirnov statistics in driving instability
phi_y d-stat = 0.553 p-value = 0.000

Starting bivariate analysis:

Correlation analysis for prior_stable
[Omega_H,rho_zRP]: corrcoef = -0.233
[Omega_NT,phi_inf]: corrcoef = -0.186
[w_T,rho_muM]: corrcoef = -0.118

Correlation analysis for prior_unacceptable
[Omega_H,rho_zRP]: corrcoef = 0.334
[Omega_NT,phi_inf]: corrcoef = 0.269
[Omega_NT,phi_y]: corrcoef = -0.192
[w_T,rho_muM]: corrcoef = 0.169
[phi_inf,phi_y]: corrcoef = 0.193

Correlation analysis for prior_unstable
[Omega_H,rho_zRP]: corrcoef = 0.334
[Omega_NT,phi_inf]: corrcoef = 0.269
[Omega_NT,phi_y]: corrcoef = -0.192
[w_T,rho_muM]: corrcoef = 0.169
[phi_inf,phi_y]: corrcoef = 0.193
Computing theoretical moments …
Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 1.953907e-16.

In lyapunov_symm at 145
In th_autocovariances at 114
In th_moments at 38
In mc_moments at 36
In map_ident_ at 58
In dynare_sensitivity at 239
In mmt at 2290
In dynare at 180

My concern lies in the computation of theoretical moments because eventhough my model’s parameters are identified, but it is unable to compute the theoretical moments. How can I solve this problem?

Thank you,

Jesus

2

What do you mean with

?

Is there a unit root in your model so that those moments do not exist?

3

No, the model does not have any unit root. In fact, I solved the model by using stoch_simul with order 1 (since the model is linear) and all the theoretical moments are computed and well defined. Troubles arise at the moment of computing theoretical moments when using identification commands. Thank you

4

Where does the identification command compute theoretical moments?

5

Dear Mr Pfeifer, the command that I use to compute the identification is:

identification(parameter_set=prior_mean,advanced=0,ar=1);
dynare_sensitivity(identification=1,morris=2,ar=1);

but after computing the correlation analyses, it says

Computing theoretical moments …
Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 1.953907e-16.

6

Please provide the full files. Usually this is just a warning that can often be ignored.