Bayesian DSGE Estimation - Warning: Matrix is singular

[Lautaro2453]

Hello, I hope you're fine.

I'm estimating a DSGE Model using Bayesian techniques. The paper in which I base is "Bayesian Estimation of an Open Economy DSGE Model with
Incomplete Pass-Through" Adolfson (2007).

I have the next warning error:

Estimation::marginal density: I'm computing the posterior mean and covariance... Warning: Matrix is singular to working precision."

Estimation::marginal density: I'm computing the posterior log marginal density (modified harmonic mean)... Warning: Matrix is singular to working precision."

"Estimation::marginal density: The support of the weighting density function is not large enough..."
"Estimation::marginal density: I increase the variance of this distribution."

"Estimation::marginal density: Let me try again."

"Estimation::marginal density: There's probably a problem with the modified harmonic mean estimator."

This is the first time that I work with Dynare. I would like to know if someone face this issue.

How can I solve this issue? Where is the problem? In the initial value of the parameters? In the prior distribution? In the observable variables? In the amount of parameters which we decide to estimate?

I try to solve the issue reducing the amount of parameters to estimate, calibrating some parameters and estimating others, changing the observable variables but I continue with the issue.

How generally you solve this issue?

I attach the dynare .mod code and the xls with the data file.

I really appreciate your time and support.

Kind regards,

Lautaro

[jpfeifer]

The first and most pressing issue is identification. Run the identification command to see

Code: Select all

WARNING !!!
The rank of H (model) is deficient!

    xi_w is not identified in the model!
    [dJ/d(xi_w)=0 for all tau elements in the model solution!]



WARNING !!!
The rank of J (moments) is deficient!

    epsilon_epsilon is not identified by J moments!
    [dJ/d(epsilon_epsilon)=0 for all J moments!]
    xi_e is not identified by J moments!
    [dJ/d(xi_e)=0 for all J moments!]
    xi_w is not identified by J moments!
    [dJ/d(xi_w)=0 for all J moments!]



You need to fix this first.

[Lautaro2453]

Hi Johannes,

Thanks for your answer. I've a short doubt regarding your answer.

In what file do you see this message error? Do you execute an additional command to see this message error or is available in an external file?

When I estimated the model, I couldn't see this message. I only see the warning messages.

If I could see this type message it would be very helpful to solve the problems with myself.

Thank you very much. Kind regards.

Lautaro

[Lautaro2453]

I found how get this warning message

I will include in my code.

identification; dynare_sensitivity(identification=1, morris=2);

Thanks for clarify my doubt.

Regards