How can I use filtered_vars options?


#1

Dear professor,
I am trying to do one step ahead forecast, my model was calibrate, dynare report an error about ‘VAROBS statement is missing!’. I am not sure if it is because I am a calibration model, so I can’t use the estimation command?

BEST REGARDS.shuju.xlsx (12.2 KB)
mixrule_baseline.mod (20.9 KB)
parameterfile.mat (2.4 KB)


About Forecast of endogenous variable
#2

You should run the calib_smoother on your mod-file. See the example at https://git.dynare.org/Dynare/dynare/blob/master/tests/fs2000/fs2000_calib.mod


#3

Thank you very much Professor jpfeifer. Your reply has helped me a lot.
I don’t quite understand the meaning of 3 and 4 in filter_step_ahead = [3:4].
I added ‘calib_smoother’ to the original code and I have this error reminder.
Error using print_info (line 42)
Blanchard Kahn conditions are not satisfied: no stable equilibrium

Error DsgeSmoother (line 108)
Print_info(info,options_.noprint, options_);

Error evaluate_smoother (line 105)
DsgeSmoother(parameters, dataset_.nobs, transpose(dataset_.data), dataset_info.missing.aindex, dataset_info.missing.state, M_, oo_, options_, bayestopt_, estim_params_);

Error mixrule_baseline (line 630)
[oo_,M_,options_,bayestopt_]=evaluate_smoother(‘calibration’,var_list_,M_,oo_,options_,bayestopt_,estim_params_);

Error dynare (line 235)
Evalin(‘base’,fname) ;
shuju.xlsx (13.3 KB)
mixrule_baseline.mod (21.1 KB)parameterfile.mat (2.4 KB)


#4
  1. I am getting an error that the steady state cannot be computed. Is there maybe a steady state file missing.
  2. The [3:4] specifies how many steps ahead the filtered variables are computed.
  3. Regarding the Blanchard-Kahn conditions: if your model runs with simulations, check whether you have a unit root. If yes, you need to specify the diffuse_filter option.

#5

mixrule_baseline_steadystate.m (5.0 KB)
Sorry, here is the steadystate file. If I don’t add the last one, the mod file will run successfully. So I am confused.

‘varobs R_nom m1;

calib_smoother(datafile=shuju, filtered_vars, filter_step_ahead = [3:4]) YD ND g CD ID pi;

run calibrated diffuse filter
bayestopt_=[];
calib_smoother(diffuse_filter,datafile=shuju, filtered_vars, filter_step_ahead = [3:4]) YD ND g CD ID pi;’


#6

You never reach the line with the diffuse_filter, because the mod-file crashes in the calib_smoother-command without that option. You need to take it out.


#7

You mean take the ‘diffuse_filter’ out in the ‘calib_smoother(diffuse_filter,datafile=shuju, filtered_vars, filter_step_ahead = [3:4]) YD ND g CD ID pi;’ centence? Or say to add % in front of this sentence ‘run calibrated diffuse filter’,There is a % in front of this sentence in the code, I accidentally deleted it when I replied.:laughing:
But I tried to remove the diffuse_filter, which means that this sentence is changed to calib_smoother(datafile=shuju, filtered_vars, filter_step_ahead = [3:4]) YD ND g CD ID pi; The program still prompts
Error using print_info (line 42)
Blanchard Kahn conditions are not satisfied: no stable equilibrium

Error DsgeSmoother (line 108)
Print_info(info,options_.noprint, options_);

Error evaluate_smoother (line 105)
DsgeSmoother(parameters, dataset_.nobs, transpose(dataset_.data), dataset_info.missing.aindex, dataset_info.missing.state, M_, oo_, options_, bayestopt_, estim_params_);

Error mixrule_baseline (line 630)
[oo_,M_,options_,bayestopt_]=evaluate_smoother(‘calibration’,var_list_,M_,oo_,options_,bayestopt_,estim_params_);

Error dynare (line 235)
Evalin(‘base’,fname) ;


#8

Only have

calib_smoother(diffuse_filter,datafile=shuju, filtered_vars, filter_step_ahead = [3:4]) YD ND g CD ID pi;’

#9

Thank you very much Professor jpfeifer!
Is the output in oo_.Smoother.Trend? But in this set, all endogenous variables have a value of zero. And the folder mixrule_baseline\Output is empty.


#10

That depends on what you are looking for. Please see the manual. Filtered variables will be in oo_.FilteredVariables.


#11

Thank you Professor Jpfeifer, you have given me a lot of help!:blush:


#12

I am very sorry, Professor Jpfeifer, bothering you again, I hope to get the result of Figure 16 in An introduction to Graphs in Dynare, but I use the data I uploaded (the result of logarithm difference) to get a large value of the output data. (The results calculated by dynare and the data published by the state uses HP filter to take the cycle item are slightly different ).
I also Try to replace the data with only logarithmic or unprocessed data, and the results are relatively large. Do you know why?
image


#13

It seems your observation equations are wrong. The data means and the means of the associated variables in the model differ significantly.


#14

Thank you very much for your help Professor Jpfeifer! You have given me a lot of help!:relaxed: