Using mode_compute=6 but still having "error using chol"

Dynare help (legacy posts)
1

Dear all,
I use Dynare 4.3.3 to estimate my model but even if I use MCMC and the cov matrix implied by lik_init=2, I still have:

[code]Initial value of the log posterior (or likelihood): -3583984986302561

==========================================================
Change in the covariance matrix = 2.25.
Mode improvement = 3583648927135472
New value of jscale = 1.2389e-05

Error using chol
Matrix must be positive definite.

Error in gmhmaxlik (line 196)
dd = transpose(chol(CovJump));

Error in dynare_estimation_1 (line 352)
[xparam1,PostVar,Scale,PostMean] =
…

Error in dynare_estimation (line 70)
dynare_estimation_1(var_list,dname);
[/code]

I tried also mode_compute=9 but get a message about the posterior kernel optimization and also “Error using dynare_estimation_1 (line 921) Mode values are outside prior bounds. Reduce prior_trunc.”
Can anyone help me please?
Please find my code and data (Smets and Wouters dataset).
thanks
J.A.
QMM.mod (6.96 KB)

2

Hi,
I corrected a typo and attached the paper I am trying to replicate. I have still the same error message.
Please help me, it’s becoming urgent!!

Thanks
JA
j.1538-4616.2008.00180.x.pdf (590 KB)
QMM.mod (7.03 KB)

3

Please provide the data-file.

4

Hello sir,
thanks for your attention. Here is the data file.

5

Sorry for the mistake, here it is.
usmodel_data.xls (196 KB)

6

It is hard to tell. There is no obvious issue with the model. My guess would be that it is really hard to find the correct mode. My only suggestion is to follow the iterative mode-finding suggested in:
[Usual Bayesian estimation problem)

7

Thanks!
I will have a look at the Gill and King paper.
Yet, I have changed my code (see attached) today and got a new error message (cf below): “Index exceeds matrix dimensions”.

[code]MH: I’m computing the posterior mean and covariance… Warning: Matrix is close to singular or badly
scaled. Results may be inaccurate. RCOND =
6.916154e-20.

In compute_mh_covariance_matrix at 74
In marginal_density at 50
In dynare_estimation_1 at 951
In dynare_estimation at 70
In QMM at 631
In dynare at 120
Warning: Matrix is close to singular or badly
scaled. Results may be inaccurate. RCOND =
6.916154e-20.
In marginal_density at 56
In dynare_estimation_1 at 951
In dynare_estimation at 70
In QMM at 631
In dynare at 120
Done!

MH: I’m computing the posterior log marginale density (modified harmonic mean)…
Warning: Matrix is close to singular or badly
scaled. Results may be inaccurate. RCOND =
6.916154e-20.

In marginal_density at 67
In dynare_estimation_1 at 951
In dynare_estimation at 70
In QMM at 631
In dynare at 120
MH: The support of the weighting density function is not large enough…
MH: I increase the variance of this distribution.
Warning: Matrix is close to singular or badly
scaled. Results may be inaccurate. RCOND =
9.147316e-20.
In marginal_density at 102
In dynare_estimation_1 at 951
In dynare_estimation at 70
In QMM at 631
In dynare at 120
MH: Let me try again.
Warning: Matrix is close to singular or badly
scaled. Results may be inaccurate. RCOND =
6.422115e-20.
In marginal_density at 108
In dynare_estimation_1 at 951
In dynare_estimation at 70
In QMM at 631
In dynare at 120
/several times/
MH: There’s probably a problem with the modified harmonic mean estimator.

ESTIMATION RESULTS

Log data density is -Inf.

parameters
prior mean post. mean conf. interval prior pstdev

rhoz 0.500 0.6023 0.3992 0.7544 beta 0.2000
rhob 0.500 0.6158 0.4703 0.7553 beta 0.2000
rhoi 0.500 0.7820 0.7441 0.8618 beta 0.2000
rhop 0.500 0.6992 0.4895 0.8076 beta 0.2000
rhoeta 0.500 0.3112 0.1804 0.4026 beta 0.2000
rhog 0.500 0.9547 0.9211 0.9932 beta 0.2000
rhor 0.500 0.3072 0.2090 0.3777 beta 0.2000
psinu 0.500 0.3159 0.0477 0.5991 beta 0.1000
etak 4.000 2.1750 0.8119 3.3814 norm 1.5000
h 0.500 0.0557 0.0513 0.0618 beta 0.1000
eta 0.500 0.7044 0.6716 0.7457 beta 0.1000
bb 0.500 0.4326 0.4095 0.4612 beta 0.1000
lambda 0.750 0.3458 0.2623 0.3629 beta 0.1000
lambdap 0.660 0.5186 0.4118 0.6091 beta 0.1000
gamma 0.500 0.5695 0.5573 0.5897 unif 0.2887
epsilonp 1.150 1.3426 1.3151 1.3792 norm 0.0500
rpi 1.700 3.0661 2.6012 3.5459 norm 0.3000
ry 0.125 0.0135 0.0045 0.0217 gamma 0.1000
rhos 0.750 0.7742 0.7105 0.8575 beta 0.1000
gammaz 1.250 1.2283 1.1881 1.2646 unif 0.1443

standard deviation of shocks
prior mean post. mean conf. interval prior pstdev

sig_z 0.150 1.0646 0.7574 1.4483 invg 0.1500
sig_b 0.150 1.5624 1.2243 1.7433 invg 0.1500
sig_i 0.150 0.7181 0.2870 1.1315 invg 0.1500
sig_p 0.150 2.9297 2.6467 3.2452 invg 0.1500
sig_eta 0.150 7.0011 6.3637 7.5131 invg 0.1500
sig_g 0.150 1.7323 1.6660 1.8568 invg 0.1500
sig_r 0.150 1.7527 1.5219 1.9990 invg 0.1500
Index exceeds matrix dimensions.

Error in DsgeSmoother (line 80)
constant = SteadyState(bayestopt_.mfys);

Error in dynare_estimation_1 (line 984)
[atT,innov,measurement_error,updated_variables,ys,trend_coeff,aK,T,R,P,PK,decomp]
=
DsgeSmoother(xparam1,dataset_.info.ntobs,dataset_.data,dataset_.missing.aindex,dataset_.missing.state);

Error in dynare_estimation (line 70)
dynare_estimation_1(var_list,dname);

Error in QMM (line 631)
dynare_estimation(var_list_);

Error in dynare (line 120)
evalin(‘base’,fname) ;
[/code]

I looked in the posts but I don’t figure out from where it comes.
Any help would be really really appreciated!
Thanks again,
JA
QMM.mod (6.83 KB)