Warning: Matrix is singular to working precision.
In trust_region>dogleg (line 198)
In trust_region (line 115)
In dynare_solve (line 255)
In evaluate_steady_state (line 221)
In resol (line 78)
In compute_decision_rules (line 35)
In dynare_resolve (line 69)
In dsge_likelihood (line 270)
In mode_check (line 161)
In dynare_estimation_1 (line 331)
In dynare_estimation (line 105)
In mnk.driver (line 604)
In dynare (line 293)
I use the logarithm of data and take the one-side hp-filter. And my model is exp(log(x)) form, the observation equation form is like this x_obs=x-steady_state(x), is this correct?
here is my .mod file
mnk.mod (5.9 KB)
You did not provide the data file to replicate the issue.
thank you, professor, for you reply, this is my data file.
data.mat (4.0 KB)
I was able to run the model:
mediumdsge_mode.mat (945 Bytes)
mediumdsge.mod (5.9 KB)
The problem is computing the steady state for all parameters. Increasing steady(maxit=1000) helps in that respect, but it would be better to have an analytical steady state.
Thanks a lot professor.
I still have some question about estimated a medium-scale DSGE model.
1.If I add wage or price index into the model,
P_{t} \left ( i \right ) =\pi_{t-1}^{ \kappa_{p} } P_{t-1} \left ( i \right)
it seems will get more feedback of this error message.
2. If I want to estimation a model with non-zero inflation, the result will be worse than former case.
In all of these cases, the steady state is harder to compute numerically. So it’s expected behavior.
thank you professor, I get it.