BVAR grid search

[paolo.zanghieri]

Good morning,
I am trying to estimate a medium size BVAR to forecast a few US variables, using the Sims Zha Normal-Wishart priors. Ad far as I understand the best way to get the hyperpatameters “right” is to run a grid search, finding the combination that maximises the posterior marginal log-likelihood. Writing the loop program is (should be) straightforward, problem is that I need the command to convert the marginal LL, into a scalar, i.e. the equivalent of var.@logl. Any clue?
Thanks a lot
Paolo

[dakila]

Freeze VAR model then get the marginal likelihood.
For example:
freeze(tab01) var01
!mll=@val(tab01(r,c))

[paolo.zanghieri]

It works. Thanks a lot.