about rank condition

[uvcindy]

Hi, I have a general question here.
I am trying to estimate some parameters by using SMM. However, I know that when the parameters fall in some range, which I don't know exactly, the model will fail the rank condition. Is there a way in dyanre I can use to index whether the rank condition is satisfied?

Thanks!

[MichelJuillard]

If you want to do SMM, you should use the Matlab functions in Dynare toolbox rather than Dynare keywords. In order to get the solution of a model, you should use resol.m
When something goes wrong, info is different from zero upon return. You can check the meaning of some of the info codes in print_info.m
I'm interested by SMM, but didn't have time to get into it. Tell me if you need additional help and I'm looking forward to see your results.

Best

Michel

[uvcindy]

Dear Michel
thanks a lot for your answer.
I tried to input [dr,info]=resol(ys_,checkflag) to use the toolbox resol.
But I still have questions:
(1) as I see some posts in the forum, resol. m is to compute the steady state and find the first-order optimal decision ruls. But in order to use SMM, I need to use either theoretical moments or simulated moments.
Can resol.m do that? or I need to use another toolbox.
(2) I understand ys_ is the steady state, but how can I specify checkflag?
How can I specify the filename in this command?

thanks again for your kindness

[MichelJuillard]

1) It depends a bit of which moments do you want to use. Look at DsgeLikelihood.m the first part of the function computes the decision rule and the covariance matrix. dynare_resolve builds a state space representation for the state variables and the observed variables

If you need other moments that means (mean is the steady state in the linear approximation) and variances, look at th_autocovariances.m

2) Use Dynare version 4 rather than version 3. Many of the global variables have been reorganized.

3) in resol.m checkflag = 1 only computes the eigenvalues. Use checkflag = 0

4) The file name is in M_.fname (version 4)

5) My advice is to write a regular *.mod file, then add the calls to your SMM functions at the bottom of it, instead of our estimation keyword.

Kind regards

Michel