Page 1 of 1

posterior standard deviation

PostPosted: Tue Aug 04, 2015 2:37 pm
by earsmall
One of output resulting from DSGE-estimation is 'oo_' and as far as I know, this variable contains various estimated results.
My question is that whether 'oo_.posterior_std' includes standard deviations of MCMC-simulated draws.
One thing strange is that there is also 'oo_posterior_variance' and I was sure that I could get values in 'oo_.posterior_std'
after imposing square root on them. However, resulting values are little bit different.
Is there any one could explain these outputs?

Thanks.

Re: posterior standard deviation

PostPosted: Wed Aug 05, 2015 6:24 am
by jpfeifer
This is an unfortunate inconsistency. The
oo_.posterior_std

field is set after mode-finding, i.e. it is the parameter standard deviation at the posterior mode.
The
oo_.posterior_variance

field is set after the MCMC algorithm. That explains why they are not the same. We will fix this in future versions.