Estimating DSGE model with extensive margin labour

Estimation ML/Bayesian estimation
1

Hi all,

I am estimating the parameters and shocks for this model; I have checked the identification for the set of estimated parameters and they are all identified. However, when I run the estimation, it only works if I remove the priors of 3 parameters: alp, sigma_c, and s. In contrast, if I include these parameters, it would returns errors as below

Error using chol
Matrix must be positive definite.

Error in posterior_sampler_initialization (line 84)
d = chol(vv);

Error in posterior_sampler (line 60)
posterior_sampler_initialization(TargetFun, xparam1, vv,
mh_bounds,dataset_,dataset_info,options_,M_,estim_params_,bayestopt_,oo_);

Error in dynare_estimation_1 (line 474)
posterior_sampler(objective_function,posterior_sampler_options.proposal_distribution,xparam1,posterior_sampler_options,bounds,dataset_,dataset_info,options_,M_,estim_params_,bayestopt_,oo_);

Error in dynare_estimation (line 105)
dynare_estimation_1(var_list,dname);

Error in paper2.driver (line 573)
oo_recursive_=dynare_estimation(var_list_);

Error in dynare (line 293)
evalin(‘base’,[fname ‘.driver’]) ;

I have tried estimating them with different mode_compute (4, 5, 6, 9) and several mode_file but they do not help.
Please have a look and let me know what I should do to improve this estimation.
Thank you!

Best,
Anh Pham

paper2.mod (4.2 KB)
paper2_mode.mat (778 Bytes)
paper2_steadystate.m (2.0 KB)
uk_data.mat (2.9 KB)
paper2_mh_mode.mat (984 Bytes)

2

For some reason, when estimating the three parameters you mention, your system gets stuck between indeterminacy and not being able to solve for the steady state. Maybe you need more different priors that push you further away from indeterminacy. Your model seems to require strong inflation feedback, which in turn partially comes from the extreme output feedback.

4

Dear Professor Pfeifer,

Thank you for your answer. Does it mean that I need to include more estimated parameters in my model? In this case, it seems that some parameters are correlated or partially correlated to the others, what should I do in this case to improve the estimation?

5

You should try to understand the economics behind your model. Why do you think a priori there is such strong output feedback?

6

Thank you, I got it now.