Page 1 of 1

Forecast, shock decomposition after Bayesian estimation

PostPosted: Mon Dec 12, 2016 1:10 pm
by JoseDNino
I’ve already done Bayesian estimation with my model and have the results using the following command

Code: Select all
estimation(datafile=observables_gali2, mode_check, mh_conf_sig=0.95,mh_replic=1000000,selected_variables_only, mode_compute=6) y_obs;


However, I now need to use this bayesian estimation to do unconditional forecasts, historical variance decomposition of shocks and Bayesian IRFs. How do you recommend to do this without having to run the time consuming estimation command again?

Thanks for your valuable help.

Jose D. Niño

Re: Forecast, shock decomposition after Bayesian estimation

PostPosted: Mon Dec 12, 2016 1:23 pm
by jpfeifer
First of all, do a backup of the folder. Then, load the mode_file, use the load_mh_file-option, and set mh_replic=0 so as to not add additional draws. Also enable the options you need like forecast

Code: Select all
estimation(mh_replic=0,mode_compute=0,mode_file=whatever_the_name_of_the_mode_file_is,load_mh_file,forecast=8,bayesian_irfs...)

The shock_decomposition is a separate command that can be put after estimation.

Re: Forecast, shock decomposition after Bayesian estimation

PostPosted: Mon Dec 12, 2016 7:30 pm
by JoseDNino
Thanks Dr. Johannes for your quick and helpful reply.