-reg3/sureg- is the contemporaneous cross-equation correlation only relevant for the same observations?

[Christopher Parker]

Dear Statalists,

Im analyzing productivity across four industries (i) and 18 countries (c) over a 22 year time period (t). I want to regress separate industry equations and I am thinking about applying a SUR setup by using reg3/sureg. When using industry equations, are only contemporaneous correlations per country used to improve efficiency, or are also cross equation and cross country correlations of relevance?

What I mean: are only correlations corr(Uic,Ujc) incorporated or also corr(Uia,Ujc) (a and c denoting separate countries)?

Many thanks ahead

Chris

[Christopher Parker]

And if the correlations across equations and countries are not used by the estimation procedure, is there some way to incoporate this?