For instance, for two endogenous variables y1 and y2,
if I have instruments z1 z2 z3 for y1 and z4 and z5
for y2, how do I write the ivreg command? Is the
following syntax ok:
ivreg y x1 x2 .. (y1 y2 = z1 z2 z3 z4 z5)? Does ivreg
recognize that z1 z2 z3 are instruments for y1?
Yes. But you should not speak of 'these instruments for those endogenous
variables'. ivreg, like any proper implementation of instrumental
variables, projects each of the endogenous on ALL of the instruments --
which include, in your case, x1, x2, z1, z2, z3, z4, z5, and 1. There is no
sense in which some of those instruments apply to some of the endogenous
variables.
