Estimation Error: Log data density [Laplace approximation] i

Dynare help (legacy posts)
1

Dear All,

I am trying to estimate a SOE model for Egypt, using quarterly data from 3q2001 to 4q2014,

The model is running fine on calibrated parameters (at least i think so as it does not give me any errors), however, the estimation gives me the following error

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.

Any ideas why or how should I tackle this problem?

the data attached was logged and filtered as recommended in jpfeifer’s observation equations document.

Thanks for your help.

EDIT NOTE: I edited the data file
filtereddata.xls (32 KB)
compact.mod (11.6 KB)

2

Running identification says:

[quote]
WARNING !!!
The rank of J (moments) is deficient!

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

E_B is collinear w.r.t. all other params!
rho is collinear w.r.t. all other params!
r_y is collinear w.r.t. all other params!
r_dy is collinear w.r.t. all other params![/quote]

Thus, given your data you cannot identify these parameters in the current model

3

Dear Dr. Pfeifer,

thank you for your continuous support on the forum and input,

I took out the unidentifiable parameters you suggested from the estimated_params block, yet still have the same error,
or did you mean to approach it differently?

thank you again

4

Dear Dr. Pfeifer,

after removing the suggested parameters, i added

identification(ar=10); command

it provided

[code]WARNING !!!
The rank of J (moments) is deficient!

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

So I removed it at well, and I no longer get an error,
attached is the final “hopefully” version after your comments.
compact_jp.mod (11.8 KB)