Page 1 of 1

Posterior standard deviations and autocorrelations

PostPosted: Sat Jul 23, 2016 10:25 am
by mbov
Hi there,

(1)Would you know how to obtain model standard deviations and model autocorrelations of endogenous variables after bayesian estimation.
Ie,Does "oo_.PosteriorTheoreticalMoments.dsge.covariance.Variance.y.y" deliver the variance of y?

(2) Also, would "oo_.PosteriorTheoreticalMoments.dsge.correlation.mean.y" deliver the model implied autocorrelation of the variable y?

Thank you for your help

Re: Posterior standard deviations and autocorrelations

PostPosted: Sun Jul 24, 2016 1:41 pm
by jpfeifer
The fields of
Code: Select all
oo_.PosteriorTheoreticalMoments.dsge

denote the moments you are looking at. The next field provides the statistic for that moment considered.

1.
Code: Select all
oo_.PosteriorTheoreticalMoments.dsge.covariance.Mean.y.y
should be the mean covariance between y and y, i.e. the variance. What you wanted to look at is the variance of the covariance over the MCMC draws.
2. The autocorrelation of y, starting from lag 1, should be in
Code: Select all
oo_.PosteriorTheoreticalMoments.dsge.correlation.Mean.y.y
If you want contemporaneous correlations, you need to use the unstable version with the
Code: Select all
contemporaneous_correlation

option and then look at e.g.
Code: Select all
oo_.PosteriorTheoreticalMoments.dsge.contemporeaneous_correlation.Mean.y.c
to get the correlation between y and c

Re: Posterior standard deviations and autocorrelations

PostPosted: Mon Jul 25, 2016 11:26 am
by mbov
Thanks