Identification help

[Nice]

Hello jpfeifer ,

I have performed the identification analysis of my DSGE model and the results are as follows:

==== Identification analysis ====

Testing prior mean

All parameters are identified in the model (rank of H)
All parameters are identified by J moments (rank of J)
Monte Carlo Testing
Testing MC sample
All parameters are identified in the model (rank of H).
WARNING !!!
The rank of J (moments) is deficient for 2 out of 677 MC runs!

However what concerns me are the following messages which appeared prior to this which report:

SOLVE: maxit has been reached
SOLVE: maxit has been reached
33.1% of the prior support gives unique saddle-path solution.
30.7% of the prior support gives explosive dynamics.

For 36.279\% of the prior support dynare could not find a solution.

For 36.279\% Cannot find the steady state.

Smirnov statistics in driving acceptable behaviour
H d-stat = 0.327 p-value = 0.000
GAMMA_K d-stat = 0.490 p-value = 0.000
GAMMA_L d-stat = 0.121 p-value = 0.000
EPSI_K d-stat = 0.183 p-value = 0.000

Smirnov statistics in driving instability
H d-stat = 0.287 p-value = 0.000
GAMMA_G d-stat = 0.133 p-value = 0.000
GAMMA_L d-stat = 0.136 p-value = 0.000
GAMMA_Z d-stat = 0.240 p-value = 0.000

Smirnov statistics in driving no solution
GAMMA_G d-stat = 0.100 p-value = 0.000
GAMMA_K d-stat = 0.249 p-value = 0.000
GAMMA_Z d-stat = 0.198 p-value = 0.000
EPSI_K d-stat = 0.153 p-value = 0.000


Starting bivariate analysis:

Correlation analysis for prior_stable
[GAMMA,H]: corrcoef = -0.179
[H,GAMMA_G]: corrcoef = -0.195
[H,GAMMA_K]: corrcoef = -0.213
[H,GAMMA_Z]: corrcoef = -0.279
[RHO_A,EPSI_L]: corrcoef = 0.233
[RHO_TK,RHO_TL]: corrcoef = 0.119
[GAMMA_K,EPSI_K]: corrcoef = 0.286
[GAMMA_L,GAMMA_Z]: corrcoef = -0.133
[EPSI_L,PHI_KL]: corrcoef = 0.140
[EPSI_Z,PHI_LC]: corrcoef = 0.147

Correlation analysis for prior_unacceptable
[H,GAMMA_K]: corrcoef = -0.195
[H,GAMMA_Z]: corrcoef = 0.099
[H,EPSI_K]: corrcoef = 0.112
[RHO_A,EPSI_L]: corrcoef = -0.115

Correlation analysis for prior_unstable
[KAPPA,H]: corrcoef = -0.160
[H,GAMMA_G]: corrcoef = -0.156
[H,GAMMA_K]: corrcoef = -0.154
[H,GAMMA_L]: corrcoef = -0.239
[H,GAMMA_Z]: corrcoef = -0.241
[RHO_TK,RHO_TL]: corrcoef = -0.140
[GAMMA_K,EPSI_K]: corrcoef = 0.173
[EPSI_L,PHI_KL]: corrcoef = -0.182
[EPSI_Z,PHI_LC]: corrcoef = -0.186

Correlation analysis for prior_wrong
[GAMMA,H]: corrcoef = 0.127
[H,GAMMA_K]: corrcoef = -0.119
[H,EPSI_K]: corrcoef = 0.133
[RHO_A,EPSI_L]: corrcoef = -0.194
[GAMMA_G,GAMMA_Z]: corrcoef = -0.143
[GAMMA_K,GAMMA_Z]: corrcoef = 0.134
[GAMMA_Z,EPSI_K]: corrcoef = -0.177
Computing theoretical moments ...

... done !

==== Identification analysis ====

Testing prior mean

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


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


Monte Carlo Testing

Testing MC sample

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


WARNING !!!
The rank of J (moments) is deficient for 2 out of 677 MC runs!

Am I to take away that my parameters are identified or are there identification problems that I need to fix prior to my estimation?

Many Thanks,

[jpfeifer]

You don't have an identification problem, but an issue with steady state finding. Consider increasing maxit or using a steady state file.