Historical counterfactual analysis

Stochastic simulations
1

Hi everyone!

I am running a historical counterfactual analysis of Gerali et al. (2010).
I followed a previous example in saving the shocks and variables in a mat file.

load (‘original_modified_results’)

ex_=[oo_.SmoothedShocks.Median.e_A_e oo_.SmoothedShocks.Median.e_eps_K_b oo_.SmoothedShocks.Median.e_j oo_.SmoothedShocks.Median.e_l oo_.SmoothedShocks.Median.e_me oo_.SmoothedShocks.Median.e_mi oo_.SmoothedShocks.Median.e_mk_be oo_.SmoothedShocks.Median.e_mk_d oo_.SmoothedShocks.Median.e_r_ib oo_.SmoothedShocks.Median.e_qk oo_.SmoothedShocks.Median.e_y oo_.SmoothedShocks.Median.e_z ] ;
save results ex_;

vrbls_=[oo_.SmoothedVariables.Median.C oo_.SmoothedVariables.Median.Y oo_.SmoothedVariables.Median.D oo_.SmoothedVariables.Median.B oo_.SmoothedVariables.Median.w_p oo_.SmoothedVariables.Median.J_B oo_.SmoothedVariables.Median.q_h oo_.SmoothedVariables.Median.K oo_.SmoothedVariables.Median.PIW oo_.SmoothedVariables.Median.r_k oo_.SmoothedVariables.Median.ee_z oo_.SmoothedVariables.Median.A_e oo_.SmoothedVariables.Median.ee_j oo_.SmoothedVariables.Median.mk_d oo_.SmoothedVariables.Median.mk_be oo_.SmoothedVariables.Median.mk_bh oo_.SmoothedVariables.Median.ee_qk oo_.SmoothedVariables.Median.m_i oo_.SmoothedVariables.Median.eps_y oo_.SmoothedVariables.Median.eps_l oo_.SmoothedVariables.Median.eps_K_b ] ;
save results1 vrbls_;

  • and uploaded them in the counterfactual.mod

load results ex_;
load results1 vrbls_;
load(‘my_data’);
T = size(ex_,1);

% Simulate shocks after initial period
shocks = ex_ (2:end,:);

%state variables at initial period
y0=oo_.dr.ys;

% update decision rules
[oo_.dr,info,M_,options_] = resol(0,M_,options_,oo_);

dr = oo_.dr;
iorder = 1;
y_ = simult_(y0,dr,ex_,iorder);

However, I’m running into errors, and also simply following the attached link is also not working.
https://forum.dynare.org/t/simulate-a-model-using-smoothed-shocks/3008/6

I feel like I’m confusing some things here. Does anybody have a clue what I’m doing wrong?
Thanks in advance!
count1.mod (18.5 KB)

2

You did not provide the full files required to run. Please have a look at

3

Hi professor Pfeifer,

firstly, thank you very much for your prompt reply. I followed the example file that you linked.

I’ve tried loading my data, smoothedshocks and smoothedvariables in a mat file, following the previous example: load results ex_
However, the following error message is still popping up:

Reference to non-existent field ‘SmoothedShocks’.
Error in count1.driver (line 1037)
ex_=[ex_ oo_.SmoothedShocks.Mean.(M_.exo_names{shock_iter})];
Error in dynare (line 293)
evalin(‘base’,[fname ‘.driver’]) ;

I can’t figure out why dynare does not recognize smoothedshocks, as I have the results from the estimation in the exact same folder.

Would be grateful for any further clues!
count1.mod (18.7 KB)

4

Again, you are loading stuff

load results ex_ ;
load results1 vrbls_;
load('deviation_data_net');

you did not provide.

5

I hope it’s complete now

results.mat (4.3 KB)
results1.mat (7.3 KB)
deviation_data_net.mat (5.5 KB)

6

Where do you actually run the Kalman smoother to compute the smoothed objects?

7

Dear prof Pfeifer,

sorry for the late reply, we have been trying some ways to run the counterfactual. we estimated a model before, which is named “estimation”. we extracted the smoothed variables and shocks from this estimation by using

load estimation_results

%%%%%%%%%%%%%%%

vParam = oo_.posterior_mode.parameters;

save(‘post_param.mat’,’-struct’,‘vParam’)

% save posterior std dev

vStd = oo_.posterior_mode.shocks_std;

save(‘post_std.mat’,’-struct’,‘vStd’).

However, uploading a zip file of estimation_results is not possible due to the large size. I hope the attached files are sufficient.
count1-2.mod (18.7 KB)
estimation.mod (20.7 KB)
post_param.mat (1.5 KB)
post_std.mat (821 Bytes)

8

Those are estimation results you mention. They are not the smoothed objects like oo_.SmoothedShocks