Testing the Information Matrix Equality with Robust Estimators



wherefrom, using (4),

ξ ( Y ; θ )= 1

σ2


u2 - 1

u3 - 3u
u
4 - 5u2 +2


2 / IF( Y ; ^; Fθ )

⅛ ( θ

σ ∖ IF( Y ; ^; Fθ )


Note that ξ ( Y ; θ ) does not depend on IF( Y ; /; Fθ ). Take ^ to be equivariant,
i.e.
σ ( aY1 + b, , aYn+b ) = ασ ( Y1, ∙ ∙ ∙ , Yn ), so IF( Y ; ^; Fθ ) = σ IF( u ; σ; Φ)
and

1 /    u2 - 1 - 2IF(u; ^; Φ)

ξ ( Y ; θ ) = — I           u3 3 u

(6)


σ \—4 5u2 + 2 2IF(u; ^; Φ)

A straightforward calculation shows that V = σ-4B , where B is a 3 × 3
matrix with elements
Bij given by

B u = 2 + 4ASV( ^),

B22 =6,

B33 = 46 + 4ASV( ^) 4 E φ ( u 4IF),

B13 = 10 + 4ASV( ^) 2 E φ ( u 4IF) = B 31,

B12 = B21 = B23 = B32 =0,

with IF = IF(u; ^; Φ) and ASV(σ) = Eφ(IF2), the asymptotic variance of
σ when σ = 1. Note that V does not depend on the estimator β that is

chosen. For a given estimator θ = (β, σ)1, let

ʌ .

u i = ( Yi — β)/σ,


n

-1 j
μj = n X"i,

i=1


. ) ʌ

and write M as σ 2 N with

/ μ2 1

I          .ʌ          n .ʌ

I μ3 3 μ1

ξ μ4 — 5μ2 + 2

Taking V+ equal to V+ with σ replaced by ^ yields the test statistic

T = nM '1V+MM = nN^ B+N^.

(7)




More intriguing information

1. Tastes, castes, and culture: The influence of society on preferences
2. Økonomisk teorihistorie - Overflødig information eller brugbar ballast?
3. If our brains were simple, we would be too simple to understand them.
4. Fiscal federalism and Fiscal Autonomy: Lessons for the UK from other Industrialised Countries
5. Altruism with Social Roots: An Emerging Literature
6. Ability grouping in the secondary school: attitudes of teachers of practically based subjects
7. The name is absent
8. Credit Market Competition and Capital Regulation
9. The name is absent
10. Strengthening civil society from the outside? Donor driven consultation and participation processes in Poverty Reduction Strategies (PRSP): the Bolivian case