Thanks for your reply. I’m going to plot the graph in Matlab then.

Another question if I may: why are my smooth estimates (oo_.SmoothedVariables) slightly different from the smoothed variables I can find in the shock decomposition (i.e. aggregating all shocks + initial condition) ?

That should be Jensen’s Inequality. The mean of the smoothed objects over the different parameter vectors from estimation is different from the smoothed object at the mean parameter vector.

for var_iter=1:M_.orig_endo_nbr
var_name=M_.endo_names(var_iter,:);
xlswrite('Shock_decomp_results',[cellstr(M_.exo_names);'Initial';'Variable Value'],var_name,'A2')
xlswrite('Shock_decomp_results',1:options_.nobs,var_name,'B1')
xlswrite('Shock_decomp_results',squeeze(oo_.shock_decomposition(var_iter,:,:)),var_name,'B2')
end