Dear professor

when I run a simpler model, the dynare has show this error "Blanchard Kahn conditions are not satisfied: no stable equilibrium "

what is the problem?

any help appreciated

As said repeatedly in this thread: without seeing the codes it is impossible to answer such general questions. The general answer is: usually a timing problem.

I sent the code on you email

tanks a lot for your reply

Dear professor

excuse me

the blanchard kahn solved but this error has emerged "initial_estimation_checks:: The forecast error variance in the multivariate Kalman filter became singular. "

Your timing for capital is wrong, In some equations you use the stock at the end of period, in other equations the stock at the beginning of period notation.

You have stochastic singularity. Most probably, there is an exact linear combination between your observables implied by the model. Try dropping e.g. `o`

from the observables

tanks a lot professor

Dear professor

after deleting observable variable “o”, the result does not change and again the error “initial_estimation_checks:: The forecast error variance in the multivariate Kalman filter became singular” has emerged

tanks a lot

Try dropping each of the observables at a time to see which one is causing the problem

Dear professor

tanks a lot for your time and I appreciated your contribution

when I run the DSGE, this error "Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 5.005722e-65. " has emerged,

this error caused that I do not see the result of posterior and acceptance rate, can I remove this Warning?

tanks a lot

Hi

by this warning “Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 5.005722e-65.”, the result of variance decomposition does not report.

what is the problem ?

As Johannes said, you probably have a stochastic singularity issue, which needs to be solved. You do not give a lot of details but I assume that the problem is related to the covariance matrix of the forecast errors. Did you try to estimate the model with less observed variables, as Johannes suggested? You can also do this the other way, starting with one observed variable and adding one by one new observed variables.

The stochastic singularity may have data and /or theoretical roots. You can check that you are not doing something stupid by simulating the model (with `stoch_simul`

, `order=1`

, and a large value for the `periods`

option) and then compute the covariance of the simulated observed variables. If this matrix is not full rank, then it means that your model says that at least one linear combination of the observed variables has zero variance. The number of such linear combinations is the number of zero eigenvalues in this covariance matrix. If you look at the associated eigenvector you will be able to identify the culprit(s) (the eigenvector associated to a zero eigenvalue defines the linear combination of variables with zero variance).

You can also check that there is no similar obvious problem with the data themselves, by doing the same exercise on the ``true’’ observed data.

You can also do the same eigenvalue/eigenvector analysis on the covariance matrix of the forecast errors, but you would have to hack the Dynare codes.

Best,

Stéphane.

Dear Stepan

tank you for your time

I do all the suggestion by Johannes, and solved all the problem

the only problem remained for me is the "Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 5.005722e-65. " .

also the variance decomposition does not report.

Best goli

But this message suggest that you did not fix the singularity issue… Or that you have another singularity elsewhere. Can you post the entire error message?

Best,

Stéphane.

As before, provide me with the files required to replicate the issue.

Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 5.005722e-65.

tank a lot

Dear professor

I sent the files on your email because the file must be secret

tanks a lot

This (warning) message is not complete. I would need all the message displayed in the Matlab command window (with the lines referring to where the problem occurs).

Best,

Stéphane.

I cannot even run your model due to the Blanchard-Kahn conditions not being satisfied. This seems to be due to Ricardian equivalence as bonds and lump-sum taxes are perfect substitutes. I get

model_diagnostics(M_,options_,oo_)

MODEL_DIAGNOSTICS: The Jacobian of the static model is singular

MODEL_DIAGNOSTICS: there is 1 colinear relationships between the variables and the equations

Colinear variables:

t

bo

Colinear equations

Columns 1 through 19

1 2 3 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Columns 20 through 21

21 23

You can only have one of the two in there in the current setup.

Dear professor

tanks a lot for your time

when I remove t or bo from my model, i see “Blanchard Kahn conditions are not satisfied: no stable equilibrium”,