A routine for BVAR model selection

[CharlieEVIEWS]

Dear All,

I'm looking to write some kind of subroutine which automatically evaluates BVAR models (primarily the lag length for simplicity, possibly also across the hypers, but not prior type) and gives a suggestion about which model to choose. However, a cursory search of the literature, the forum, and the summary statistics provided by default in EViews 8 has left me a little confused on where to begin. While I appreciate that this kind of thing may be beyond the scope of the forum, any suggestions you might have would be more than extremely useful.

Very best wishes as always,

Charlie