Peoblems about bayesian estimation

Estimation ML/Bayesian estimation
1

Dear all,
My bayesian estimation went wrong. When I used Dynare to run my mod file, there were some bad results:
Error in computing likelihood for initial parameter values
Initial value of the log posterior (or likelihood): -Inf

==========================================================
Change in the posterior covariance matrix = 9.9995.
Change in the posterior mean = 0.040833.
Current mode = 44.3099
Mode improvement = Inf
New value of jscale = 9.9854e-06

==========================================================
Change in the posterior covariance matrix = 0.00062842.
Change in the posterior mean = 0.040833.
Current mode = Inf
Mode improvement = Inf
New value of jscale = 0

==========================================================
Change in the posterior covariance matrix = 3.1419e-08.
Change in the posterior mean = 0.
Current mode = Inf
Mode improvement = NaN
New value of jscale = 0

Optimal value of the scale parameter = 0

Final value of minus the log posterior (or likelihood):Inf

MODE CHECK

Fval obtained by the optimization routine: Inf

And the mode check plots:

untitled
untitled700×525 9.04 KB

How can I solve this problem?
newsbayes.m (1.5 KB)
newsbayes.mod (3.9 KB)
obs.xlsx (15.5 KB)

2

You are missing the diffuse_filter option needed due to the unit root in your model. Also, you should provide a proper steady state file instead of just initial values. Estimation is having a hard time with steady state finding.

3

Thanks for your reply. I used the diffuse_filter option and recompiled the code, but a new problem arose. When I run the code, I get the following error:
initial_estimation_checks:: The forecast error variance in the multivariate Kalman filter became singular.
initial_estimation_checks:: This is often a sign of stochastic singularity, but can also sometimes happen by chance
initial_estimation_checks:: for a particular combination of parameters and data realizations.
initial_estimation_checks:: If you think the latter is the case, you should try with different initial values for the estimated parameters.

what should I do next?
newsbayes1.mod (5.8 KB)
obs.xlsx (23.3 KB)
news1.m (1.6 KB)

4

See e.g.

5

It worked! Thank you very much.