Bayesian IRF Matching Estimation - Posteriors equal to Priors

Estimation SMM/GMM/IRF-Matching
1

Dear all,

I am trying to estimate some of the parameters of my model using impulse response function matching. However, even when I isolate the estimation to only one of the parameters, the priors and posteriors are almost identical, which seems very odd beacuse it would mean that all parameters are not well identified. Am I doing anything wrong? In a previous version of the code I received the error message below:

The steady state file internally changed the values of the following estimated parameters:
b
This will override the parameter values and may lead to wrong results.

But after changing the steady state file, I don’t get any error message.

Attached are two versions of the file trying to estimate only ETA or PHI (the other priors are commmented out).

Thanks for the help.

IRF_PVAR_EM.mat (10.9 KB)
MCBB_BIRF_EM_ETA.mod (10.1 KB)
MCBB_BIRF_EM_ETA_steadystate.m (5.4 KB)
MCBB_BIRF_EM_PHI.mod (10.1 KB)
MCBB_BIRF_EM_PHI_steadystate.m (5.4 KB)

2

From what I can see, your objective function is very insensitive to changes in PHI. If I center PHI at 2 instead of 10, the objective function hardly changes. That should explain the issue.

3

Thanks, but this happens also with all the other parameters, I tried each one separetely and get the same issue, it doesn~t make sense for me, have you seen something like this before? Thanks

4

You should try to check manually how the IRFs change when you change the parameters. That may give you an indication.

5

Thanks, it looks like if a re-scale the IRFs I am able to estimate the model. Thanks for your help!