As I mentioned in my previous post besides other issues, I cannot get the oo_.FilteredVariablesKStepAhead structure after running estimation with the filter_step_ahead option:
[code]estimation(order=1,datafile=myOriginalData, mh_replic=300000, mh_nblocks=1, mh_drop=.3, mh_jscale=0.2,
mode_compute=6, filter_step_ahead = [1 4 8 12], forecast=20,smoother, filtered_vars, diffuse_filter,
% do this only for the final model b/c takes a lot of time
conditional_variance_decomposition=[1 4 8 20],
lmcU lrkbarU lwbarU lRfU lpxU lpixU lkU lRnustarU
lHU lpiU lmcxU lpimcU lmcmcU lpimxU lmcmxU lpimiU
lmcmiU lpicU lpmcU lpiiU lpmiU lkbarU liU lpsizplusU
lcU lpkprimeU luU lpinvestU lsU aU lqU lxU
lpcU lRU lyU lRxU lpmxU lRkU
lylessdU nxdivGDPU lSlevU
This occurs both for ver. 4.3.3. and the snapshot ver. 2013-09-22.
There is no error message during the dynare session.
I presume this is due to the conflicting options (unknown to me) or the model’s size.
All files attached.
myOriginalData.txt (11.8 KB)
baseline70log.txt (432 KB)
baseline70_steadystate.m (34.1 KB)
baseline70.mod (57.7 KB)
baseline_steadystate_find_AL_delta_varphi.m (3.29 KB)
Could you also please provide the mode-file? That way I can save on the mode_compute=6.
This might be an error in the manual. If you use Bayesian estimation with the smoother command, the field that is set is just oo_.FilteredVariables. oo_.FilteredVariablesKStepAhead is for ML estimation
This doesn’t seem to be right, since I’m asking for multiple-period forecasts filter_step_ahead = [1 4 8 12] but the oo_.FilteredVariables structure gives just vectors which I presume is one-step ahead forecasts which correspond to the option filtered_vars which I’m also using. It seems that the filtered_vars option overwrites the filter_step_ahead option, but this is not a welcome behavior.
Sorry, but this is a bug, which will be hopefully fixed in one of the next unstable versions. In case you are using the current unstable version, I am attaching the two files you need to replace to fix the issue. Please report any strange behavior you might encounter.
pm3.m (8.57 KB)
prior_posterior_statistics.m (10.7 KB)
Thanks, Johannes. Seems to be working.