Dear all,
I tried to use the below command to do recursive forecast when I replicated RMSE of SW2007.
estimation(optim=('MaxIter',200),datafile='usmodel_data.mat',mode_compute=0,mode_file=usmodel_hist_dsge_f19_7_71_100_mode,nobs=[71:170],presample=4,lik_init=2,prefilter=0,mh_replic=2500,mh_nblocks=2,mh_jscale=0.20,mh_drop=0.2,forecast=14,nograph, nodiagnostic)
I wanted to get the forecast from period 171 to period184, so I set forecast=14. I just thought [71:170] was the sample data which were used to estimate parameters, but after running the command, I got a 100*14 matrix in the oo_.MeanForecast.Mean. I checked the manual, but I was not sure if I got the right meaning.
the first row means that dynare uses only one perid data(period 71) to forecast [171:184] data. then the 2nd row means that dynare uses 2 period data(from 71 to 72) to forecast [171:184], then 3rd row uses 71:73] to forecast [171:184] and the last one uses [71:170]to forecast [171:184]. Am I right?
Then when I caculate the RMSE, I just need use the last one? I do not know how to use these data to get RMSE as the same as SW2007.
Thank you for your help.