Hi Dynare team!

I am performing the Bayesian estimation on a model with 7 shocks and 7 observable. The model is log-linearized model and the observed data are filtered. The error message I received is the following:

POSTERIOR KERNEL OPTIMIZATION PROBLEM!

(minus) the hessian matrix at the “mode” is not positive definite!

=> posterior variance of the estimated parameters are not positive.

You should try to change the initial values of the parameters using

the estimated_params_init block, or use another optimization routine.

I don’t really understand why this problem occurs.

I await your suggestions.

Thanks

Are the observation equations correct? And what do the `mode_check`

-plots say?

the observation equation are as follows:

y_obs=y;

pie_obs=pie;

r_obs=r+e_r;

dq_obs=dq+e_q;

de_obs=de;

pie_w_obs=pie_w;

tot_obs=tot+e_s;

NB: e_s; e_q, e_r are exogenous shocks in the corresponding equations of the model

The mode check for option mode_compute=4 and 5 yields graphs showing that the estimated mode is not at the maximum of the posterior likelihood. In addition, the posterior density graphs do not appear. Only prior shapes appear and the program fails to complete the estimation

I also observed that the graphs generated by mode_check have red dots

I need to see the mod-file, the data-file and the mode-file

Hello Professor!

I finally got it run. However, there are red dots on the mode_check graphs for few parameters (4 out of 23 parameters). Does it mean something wrong on the model or I can conclude on the output? I would also like to use the command shock_decomposition which cannot run in the .mod file. I wonder whether I am using it the wrong way.

- Use
`options_.debug=1`

to see why the red dots appear. - What is the error message from
`shock_decomposition`

?

I got the shock decomposition graphs when I inserted the command after the estimation command. However, in the stoch_simul result, I do not get the variance adding up to 100 and I can’t explain why. Could you tell me the command to get them add up to 100?

Thank you