Hi All
I have been estimating a DSGE model using only a subset of the shocks, setting some shock st. deviations to zero. I noted that dynare produces impulse responses for even the inactive shocks! It must be noted that in the ‘inactive-shock triggered’ IRFs the deviations of the endogenous variables from steady-state are very very small, to the order of 10^-10. Have any of you experienced the same? And if so, what could be the explanation?
Thanks
Reuben
To prevent dealing with non-positive definite covariance matrices, Dynare adds a small number to the diagonal of the covariance matrix:
SS(M_.exo_names_orig_ord,M_.exo_names_orig_ord)=M_.Sigma_e+1e-14*eye(M_.exo_nbr);
In most cases, no IRFs are displayed as IRFs are only printed for cases where the response exceeds
options_.impulse_responses.plot_threshold=1e-10;
In your case, the IRFs seem to barely exceed this threshold, resulting in displayed graphs. But this is nothing to worry about.
Thanks, Johannes! I must thank you for your work on this website. It is fantastic.