A simultaneous equation system presumes that endogenous explanatory Va
riables are correlated with equation-specific structural disturbances.
Our paper proposes estimation methods and tests for switching orthogo
nality of a subset of the endogenous variables over a sample subset wh
en the switching point is presumed known. Dividing the full sample per
iod into two sub-periods, the null hypothesis of switching orthogonali
ty/endogeneity is that the subset of endogenous variables is correlate
d with the disturbance in one of the two periods only, whereas the alt
ernative hypothesis of endogeneity is that the subset of variables is
correlated with the disturbance in both periods.