Estimation error :Blanchard & Kahn conditions are not satisfied: no stable equilibrium

Prior distribution for parameter rhoApound has unbounded density!
Prior distribution for parameter rhoR has unbounded density!
Prior distribution for parameter rhoRpound has unbounded density!
Prior distribution for parameter rhoY has unbounded density!
Prior distribution for parameter rhoYpound has unbounded density!
PARAMETER INITIALIZATION: Some standard deviations of shocks of the calibrated model are 0 and
PARAMETER INITIALIZATION: violate the inverse gamma prior. They will instead be initialized with the prior mean.
Error in computing likelihood for initial parameter values

ESTIMATION_CHECKS: There was an error in computing the likelihood for initial parameter values.
ESTIMATION_CHECKS: If this is not a problem with the setting of options (check the error message below),
ESTIMATION_CHECKS: you should try using the calibrated version of the model as starting values. To do
ESTIMATION_CHECKS: this, add an empty estimated_params_init-block with use_calibration option immediately before the estimation
ESTIMATION_CHECKS: command (and after the estimated_params-block so that it does not get

print_info (line 32)
Blanchard & Kahn conditions are not satisfied: no stable equilibrium.

This is my prior dstribution set.

rhopsi, beta_pdf, 0.5, 0.2;
rhopsipound, beta_pdf, 0.5, 0.2;
rhoA, beta_pdf, 0.4, 0.2;
rhoApound, beta_pdf, 0.2, 0.2;
rhoCM , beta_pdf, 0.7, 0.2;
rhoR , beta_pdf, 0.9, 0.2;
rhoRpound , beta_pdf, 0.9, 0.2;
rhopi,normal_pdf,1.5, 0.2;
rhopipound,normal_pdf,1.5, 0.2;
rhoY, beta_pdf,0.1, 0.2;
rhoYpound, beta_pdf,0.1, 0.2;
rhoe, beta_pdf,0.5, 0.2;
stderr epsi, inv_gamma_pdf, 0.2236, 0.2;
stderr epsipound , inv_gamma_pdf, 0.2236, 0.2;
stderr eA, inv_gamma_pdf, 0.1,0.2;
stderr eApound, inv_gamma_pdf, 0.1, 0.2;
stderr eR, inv_gamma_pdf, 0.1,0.2;
stderr eRpound, inv_gamma_pdf, 0.1,0.2;
stderr eCM, inv_gamma_pdf, 0.1,0.2;
stderr eobs_c_h, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_gdp_h, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_m_h, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_pi_h, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_c_f, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_gdp_f, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_m_f, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_pi_f, inv_gamma_pdf, 0.01, 0.02;
stderr eobs_cm , inv_gamma_pdf, 0.01, 0.02;

I used diffusion filter and changed the prior values of some parameters, which made my mod run (of course, it still showed that the setting of inverse gamma distribution was violated). However, my results are very strange. The graph of mcmc shows that the mcmc of some parameters still cannot converge under the condition of 500000 samples. At the same time, the posterior distribution of output is mostly very close to the prior distribution.Here are some of my results.
1.fig (169.3 KB)
untitled.fig (168.4 KB)
untitled2.fig (60.2 KB)
untitled5.fig (38.6 KB)
untitled5.fig (38.6 KB)

Without the files it is impossible to tell what is going on.

I change a algorithm 9,which is recommended in other post,to find the mode.But it raise an error and the translation:

Log data density [Laplace approximation] is NaN.

Incorrect use of chol
The matrix must be a positive-definite matrix with a diagonal of real numbers.

error posterior_sampler_initialization (line 84)
d = chol(vv);

error posterior_sampler (line 60)
posterior_sampler_initialization(TargetFun, xparam1, vv,

error dynare_estimation_1 (line 471)

error dynare_estimation (line 118)

error cpippiestimation.driver (line 959)

error dynare (line 281)
evalin(‘base’,[fname ‘.driver’]);

error cpippiestimation (line 135)
dynare cpippiestimation

Again, without the codes it is impossible to tell what is going on.

cpippiestimation1.m (3.5 KB)
cpippiestimation1.mod (10.2 KB)
Sorry professor, I’ll give you the code right now.
obsdata2.mat (7.1 KB)


But the parameters I estimated were the autoregressive parameters of the AR(1) process, the Taylor rule parameters, and the standard deviation of exogenous shocks. Are there also dependencies on such parameters?

My mistake. If you only estimating non-dependent parameters that should not be an issue. But your mode_check-plots show corner solutions. You may need to put prior_trunc=0.