SEMIPARAMETRIC EFFICIENCY BOUNDS FOR THE BINARY CHOICE AND SAMPLE SELECTION MODELS UNDER CONDITIONAL SYMMETRY

We consider a binary choice model and a sample selection model in whic h the conditional distribution of the error term, given the explanator y variables, has a symmetric form. We show that the semiparametric eff iciency bounds are infinity for both models under the symmetry restric tion. Hence, it is not possible to obtain a root n consistent and asym ptotic normal estimator for the two models under the restriction.