Take the regression model Y X0e E[e j X] 0 E e2 j X x 2(x) with X 2

Take the regression model Y Æ X0¯Åe E[e j X] Æ 0 E

£

e2 j X Æ x

¤

Æ ¾2(x)

with X 2 Rk . Assume that P[e Æ 0] Æ 0. Consider the infeasible estimator e¯ Æ

Ã

Xn iÆ1 e¡2 i Xi X0 i

!¡1 Ã

Xn iÆ1 e¡2 i XiYi

!

.

This is aWLS estimator using the weights e¡2 i .

(a) Find the asymptotic distribution of e¯.

(b) Contrast your result with the asymptotic distribution of infeasible GLS.