Hi,
I am a fairly advanced user of dynare, and I have built several models, calibrated them, simulated them and estimated them. As I have done several times, I estimated a model, usually SOE-NK, and then run simulations with the estimation. After a succesfull estimation, to get the irfs with the estimated parameters, I change the code in the estimation part to not calculate the mode, not run the mh replication and use the mh results and run and stochastic simulation, as in:
estimation(datafile='data_XMAS_24Q4_energy_new_ends19q3_newnew.xlsx',
xls_sheet=DataDemean, xls_range = A1:Ap80,
mode_compute=0,
% additional_optimizer_steps=[4],
% mh_replic=300000,
% mode_check,
mode_file=nk_open_elec_LCP_mh_mode,
% load_mh_file,
mh_nblocks=2,
mh_drop=0.5,
plot_priors=0,
posterior_nograph,
mh_jscale=0.35);
stoch_simul(irf=40,order=1, hp_filter=1600,nograph);
(The nk_open_elec_LCP_mh_mode.mat file is copied next to the .mod file). This procedure has worked perfectly. However, in the current model it starts again a mh replication procedure, taking close to two hours instead of the 5-10 minutes it usually takes. I googled this behavior and it states that this is due to an estimation procedure that did not finished properly, but it does not seem to be the problem by looking at the log file. ( I am attaching the log file). What could the reason for this behavior?
nk_open_elec_LCP.log (52.0 KB)
estimation_tests4a.inc (5.3 KB)