BVAR-DSGE estimation problems

Dynare help (legacy posts)
1

I am attempting to use the BVAR_DSGE routines on the Lubik-Schorfheide EU model published in Schorfheide’s site under “nber-ma2005-public.zip\projects\Active\nber\public\dynare\euro_m1\euro_m1-mod”.
The model runs perfectly under normal estimation. Unfortunately, when attempting BVAR-DSGE estimation, despite many attempts changing various estimation options, it is impossible to get beyond the standard Bayesian estimation of the parameters, viz:

parameters
prior mean post. mean conf. interval prior pstdev

theta 0.750 0.4194 0.2422 0.6175 beta 0.1500
tau 2.000 2.2726 1.4040 3.0923 gamm 0.5000
h 0.300 0.3803 0.1931 0.5674 beta 0.1000
psi1 1.500 1.5818 1.2156 1.9178 gamm 0.2500
psi2 0.500 0.2966 0.0946 0.4766 gamm 0.2500
rhoA 0.800 0.8412 0.7408 0.9327 beta 0.1000
rhoR 0.500 0.7096 0.6126 0.8194 beta 0.2000
rhoG 0.800 0.8678 0.7701 0.9779 beta 0.1000
rhoZ 0.660 0.4057 0.2020 0.6085 beta 0.1500
rr_steady 0.500 0.7928 0.0000 1.7801 gamm 0.5000
gam_steady 0.400 0.4674 0.2372 0.6767 norm 0.2000
pi_steady 7.000 2.8271 1.8305 3.7817 gamm 2.0000
dsge_prior_weight 1.000 0.8999 0.4404 1.3644 unif 0.5774

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

EPS_A 0.501 0.5875 0.2945 0.8873 invg 0.2621
EPS_G 1.253 0.4827 0.3985 0.5751 invg 0.6551
EPS_R 0.251 0.1372 0.0996 0.1704 invg 0.1310
EPS_Z 0.627 0.3354 0.2264 0.4386 invg 0.3276
Loading 80 observations from us_euro_data.m

??? Error using ==> mtimes
Inner matrix dimensions must agree.

Error in ==> PosteriorIRF at 229
SIGMAtrOMEGA = SIGMAu_chol*OMEGAstar’;

Error in ==> dynare_estimation_1 at 1074
PosteriorIRF(‘posterior’);

Error in ==> dynare_estimation at 62
dynare_estimation_1(var_list,varargin{:});

Error in ==> euro_m1b at 165
dynare_estimation(var_list_);

Error in ==> dynare at 132
evalin(‘base’,fname) ;

I am unable to comprehend why there is reported the error “Error using ==> mtimes Inner matrix dimensions must agree”

Would Stéphane be able to help with this?

Regards
US_EURO.TXT (7.33 KB)
us_euro_data.m (381 Bytes)
euro_m1b.mod (3.69 KB)

2

Dear John, I was not able to run your file… The problem here is that you have more shocks than observed variables. The identification scheme proposed by Del Negro and Schorfheide requires that the number of structural innovations equals the number of observed variables in the VAR model.

Best, Stéphane.

3

Dear Stéphane,

Many thanks for your kind reply and apologies for this late response - I have been away.
Apologies also for having forgotten to include the mode_file with my submission; this is why you could not run my programme.
But in any case, you are quite right that the problem lay with the number of shocks permitted under the DelN_S identification scheme. Thank you for pointing this out.
Best regards

4

Hello again Stéphane. I have sorted out the earlier problem, thanks to your note about the identification problem, but now have another problem with a different model.
As you can see from the error messages below, this case is not supposed to happen and I am asked to contact you. Hence this. Any thoughts?

Error in computing likelihood for initial parameter values
??? Error using ==> print_info at 60
This case shouldn’t happen. Contact the authors of Dynare

Error in ==> initial_estimation_checks at 101
print_info(info, options_.noprint)

Error in ==> dynare_estimation_1 at 334
initial_estimation_checks(xparam1,gend,data,data_index,number_of_observations,no_more_missing_observations);

Error in ==> dynare_estimation at 62
dynare_estimation_1(var_list,varargin{:});

Error in ==> ls_d_vn at 188
dynare_estimation(var_list_);

Error in ==> dynare at 132
evalin(‘base’,fname) ;

Best regards
ls20074vn_jc4.xls (25 KB)
ls_d_vn.mod (2.94 KB)