Bayesian estimation problems

Estimation ML/Bayesian estimation
1

Dear dynare helpers,

Previously I did a successful Bayesian estimation of a closed economy. Recently I have extended it to an open economy. I have done a successful deterministic simultation with this model too (at least it seems to be). But when I start Bayesian estimation, dynare keeps returning an error message.

I have tried getting rid of the new measurement equations and observables and running the estimation again, but I still receive the same error message.

I am getting a bit lost now.

Do you have any idea how I could find out where the mistake(s) is?

Please find my code and data attached.

Posterior_Generation.mod (37.9 KB)
data.mat (57.5 KB)
Posterior_Generation_mode.mat (16.7 KB)

Best

Jing

2

When I run mode_compute=5 I get

The following parameters are at the prior bound: SE_epinf_mi, chabb

Apart from that, various parameters are right at the boundary to the determinacy region.

3

Thanks. Really? But when I run mode_compute=5, I get an orange warning at the beginning, saying that “Matrix is singular, close to singular or badly scaled.” Is that a matter of concern?

And for the prior bound problem, do I just need to change the prior bound or need to change parameters/intial values too?

4
  1. That warning is not a big concern.
  2. That depends. If you consider the prior bounds to be reasonable, you need to find out why the model assigns a large likelihood to the parameters being right at the bound. Otherwise, modify your prior.
5

Thanks very much! I will take a closer look.