Testing for autocorrelation in systems of equations

This paper deals with the problem of testing for the presence of autocorrelation in a system of general linear models (Seemingly Unrelated Regressions, SUR) when the model is formulated as a vector autoregression (VAR) with exogenous variables. The solution presented in this paper is a generalization of the h-statistic for the single equation single parameter case given in Durbin (1970a). All derivations are based on first principles and no use is made of Durbin's original arguments.