The identification assumption here is that birth order affects education but does not directly affect wages (except through its effect on education). You can test this assumption by checking the correlation between birth order and the residuals from the second stage regression. how to do this