Page 1 of 1
Cross country correlations
Posted:
Wed Jul 20, 2016 5:50 am
by mbov
Hi there,
Is there a code after bayesian estimation of a DSGE model which can obtain the cross-country correlations if I have used data between Canada and the US?
Ie, if their is US output and Canada Output, obtain correlations for Corr(US_t,CAN_(t-k)) where t indexes time (t) and k>0.
Thank you
Re: Cross country correlations
Posted:
Wed Jul 20, 2016 6:49 am
by jpfeifer
Are you talking about cross-correlations in the data? Or for model variables? The latter are simple cross-correlation between model variables that are e.g. saved if you enable
- Code: Select all
moments_varendo
Re: Cross country correlations
Posted:
Wed Jul 20, 2016 8:44 am
by mbov
Thank you for your response. Cross correlations for both the data and DSGE model variables.
Re: Cross country correlations
Posted:
Wed Jul 20, 2016 9:06 am
by jpfeifer
For the data you have to compute it yourself.
Re: Cross country correlations
Posted:
Thu Jul 21, 2016 12:59 pm
by mbov
Thankyou. For the model cross correlations, how would you obtain the (10,90) or (5,95) posterior bands from:
oo_.PosteriorTheoreticalMoments.dsge.correlation.deciles.dp_obs.dpS_obs
It gives me a 5x1 cell
Thanks
Re: Cross country correlations
Posted:
Thu Jul 21, 2016 1:06 pm
by jpfeifer
The HPDI should give you 90% posterior bands.