Looping over DSGE-VAR options: dsge_prior_weight

Estimation DSGE-VAR
1

Dear all,
I want to get Marginal Density from DSGE-VAR over differents values of dsge_prior_weight = [0.5:0.25:10],
Please could you give some advice to do this?

I found some examples about looping over parameters but I didn´t find something about looping over options like dsge_prior_weight.

Thanks

Cheers

Aldo

2

You would need something along the lines of

global oo_
n_grid_points=5;
param_grid=linspace(0.3,0.5,n_grid_points);
options_.bayesian_irf = 0;
options_.plot_priors=0;
options_.silent_optimizer=1;
data_density=NaN(n_grid_points,1);

for par_iter=1:n_grid_points
    set_param_value('dsge_prior_weight',param_grid(par_iter));
    oo_recursive_=dynare_estimation(var_list_); 
    data_density(par_iter)=oo_.MarginalDensity.LaplaceApproximation;
end
3

Thanks jpfeifer,
Now I am getting this message:

WARNING: When estimating a DSGE-Var, declaring dsge_prior_weight as a parameter is deprecated. The preferred method is to do this via the dsge_var option in the estimation statement.
ERROR: If dsge_prior_weight is declared as a parameter, it must either be initialized or placed in the estimated_params block.

How could I solve this problem?
I attach my .mod and data.

Thanks in advance

Aldo
ls.mod (6.12 KB)
newmfile_peru.m (5.82 KB)

4

The attached file should work with the unstable version
ls.mod (6.32 KB)

5

Thanks a lot dear jpfeifer.

Your work is very worthfull. Thanks.

Cheers

Aldo

6

Dear Sir,

Thanks you for the contribution. Can you please clarify how we can specify the value for the dsge_prior_weight, as such that it can range from 0 to infinity as mentioned in MARCO DEL NEGRO AND FRANK SCHORFHEIDE, 2006, paper, so that we can ge the liklehood function based on hyperparamter with 0-infinity values.

Thanks.

7

What exactly are you trying to achieve here?

8

Dear sir,
For different values of lambda ranging 0 to infinity, I want to get the corresponding marginal liklehood values, so that I can plot the Marginal liklehhood function against the values of hyperparamter which ranges from 0 to infinity after scaling as mentioned in the paper discussed above.

The code below don’t consider the range of hyperpamter from its minimum value to infinity.
n_grid_points=5;
param_grid=linspace(0.3,0.5,n_grid_points);
options_.bayesian_irf = 0;
options_.plot_priors=0;
options_.silent_optimizer=1;
data_density=NaN(n_grid_points,1);

Thanks

9

You need to adjust the grid to your liking, e.g.

n_grid_points=50;
param_grid=linspace(0,15,n_grid_points);
10

Thank you sir.