6 Sn(x ) ' |
/ ʌ | |
n↑[n Sn& z (x ) |
⇒ |
c(ɪ)ʃwty )dy к c (1) W(x ) , |
√n |
(A.4)
It follows from Lemma 9.6.3 in Bierens (1994, p.200) that
JL . c .
FX-(t/n)z, F F(1)S∙(1) - f(x)'∙.(x)dx,
t= 1 j
(A.5)
where fk is the derivative of Fk, and similarly for ∆zt. Using (A.4), (A.5), and the straightforward
qualities
Fk(1)Jw(x)dx - ʃfk(x)ʃW(y)dydx = ʃFk(x) W(x)dx,
0
Var(jFk(x)W(x)dx∣ ʃ ʃʃFk(x)Fk(y)min(x,y)dxdy ∙ Iq,
(A.6)
(A.7)
Var Fk(1) W(1) - ʃf (x) W(x)dj
(A.8)
= [f (1)2 - 2Fk(1) ʃxfk(x)dx + ʃʃfk(x)fk(y)min(x,y)dxdy∣ ■ 1q = (ʃFk(x)2dxl
Cov
6. ` l Ô
(ʃFk(x) W(x)dx∖, (Fk(1) W(1)-ʃfk(x) W(x)dx}^
ʃʃFk(x )fk(y)min(x y ) dxdy^
(A.9)
ʃFk((x ) ʃFk(y ) dy dx
= ʌ(ʃFk(x)dxI ∙ 1q ,
it follows that (11) holds. The independence of the random vectors Xk and Yk over k follows
from:
38