Difference on variance with listing smoothed variables after estimation and without


#1

Dear all:
Will there be any difference on the variance if I list several smoothed variables after Bayesian estimation and if not?
For example:
code 1:
estimation(datafile=hpplus,order=1,mode_compute=6);
stoch_simul(irf=20,nograph) y inv k n q lev pi i r c;
code 2:
estimation(datafile=hpplus,order=1,mode_compute=6) y inv k n q lev pi i r c;
stoch_simul(irf=20,nograph) y inv k n q lev pi i r c;
will the two codes above produce different results on oo.var_ and irf results?

Many thanks in advance.


#2

Sorry, I have checked that there will be no difference.

I have one more question, I run stoch_simul after estimation without the code

xparam1= get_posterior_parameters(‘mean’);
M_ = set_all_parameters(xparam1,estim_params_,M_);

I read in the manual that after estimation with iterations(the default value of mh_replic is larger than 0, so I think I don’t need to declare again), dynare will set the paramters M.params_ to posterior means automatically, so I think I can put stoch_simul after estimation without these codes. Am I right?
Any response will be appreciated.
Thank you indeed.


#3

Yes, Dynare will automatically set the parameters to the posterior mean after you run an MCMC