as n -→ ∞, fn - g -→p 0, then D21 -→ 0
(33)
and since i∞=1 ln f(θ, xi)g(xi) = ln f(θ, x)g(x)dx, then D22 -→ 0.
We can conclude that KIjn (θ) -→p KIj (θ), hence it is a contrast relative to the contrast function KIj (θ)
according to the Definitions 3 and 4 in Dhrymes (1998).
Further since KIjn (θ) can be rewritten as Hn (θ) - Hn(fn), where
nn
Hn(fcn) = - lnfcn(xi) fcn(xi) and Hn(θ) = - ln f (θ, xi)fcn(xi)
then
(34)
(35)
Hn(θ1) - Hn(θ2) = [Hn(θ1) -Hn(fcn)] - [Hn (θ2) -Hn(fcn)].
It follows that
Hn(θ1) - Hn(θ2) →p KIj(θ1) - KIj(θ2). (36)
By the continuity of Kullback-Leibler Information and by A3, assumption (iii) of Theorem 1 in Dhrymes
MΛΛ∩', ∙ ∙ j ∙ Γ∙ 1 ГГ11 Jl ∙ J PjI Л /Г/ɔ j∙ J ^Λ^ C 11 ∙ T J 1 1 Jl ∙ Jl
(1998) is justified. Then the consistency of the MC estimator θMj follows immediately by this same theorem.
9.2 Proof Theorem 2:
By the mean value theorem around the parameter θ*
0 = VKI(fn, fb) ` VKI(fn, fb) ∣θ. +V2KI(fn, fθ) ∣θ (bn - θn) (37)
(bn - θ∙) ' -(V2KI(fn, fθ) ∣θ)-1 ∙ VKI(fn, fθ) ∣θ. (38)
,b fl∙A (∖'d2 log f (θ,xi) bl Λ (∖'d log f (θ*,xi) bl Λ
(bn - θ) ' -(∑ ∂θi∂θj fn(xiζl ∙(Σ-----∂θ-----fn(xiζl
Г b a∙A 11 X d2 logf (θ,xi) bl Λ 1 1 X d logf (θ*,xi) bl Λ ∕9∩λ
√n(θn - θ ) ' - (n^ ∂θi∂θj fn(xi)J ∙ (√n∑ -----∂θ-----fn(x,)J . (39)
Let us define s(θ,x) = d logfθθ,xi) = dff(⅛)dθ
√n(bn - θ* ) ' -
/1 X θ fn( ∖ 1f√s X s(θ.,χi)fn(χ.A
n ∂θ n
-(An(θ))-1Wn(θ*).
(40)
Rewriting Wn (θ. ) as a second order U-statistic of the form
n n -1 n-1 n
Un = S⅛-Γ)Σ∑hk(xj-x)s(θ∙,χi)= n) ∑∑hK(xj-xi)[s(θ∙,χi)-s(θ∙,χj)] (41)
n n -
i=1 j=1 i=1 j=i+1
j6=i
23