Scribd
Upload
19 views3 pages

MTH 515a: Inference-II Assignment No. 5: Asymptotically Efficient Estimators

1. This document contains 19 problems related to asymptotically efficient estimators. The problems cover topics like consistent estimators, asymptotic distributions, UMVUEs, MLEs, and comparing estimators based on their limiting relative efficiency and asymptotic relative efficiency. 2. Many of the problems involve showing that certain sample statistics are consistent estimators or finding their asymptotic distributions. Some problems calculate variances and biases of estimators. 3. Several problems compare different estimators for a parameter in terms of their limiting relative efficiency and asymptotic relative efficiency to determine which is more efficient.

Uploaded by

Mulyana Kahfi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
19 views3 pages

MTH 515a: Inference-II Assignment No. 5: Asymptotically Efficient Estimators

1. This document contains 19 problems related to asymptotically efficient estimators. The problems cover topics like consistent estimators, asymptotic distributions, UMVUEs, MLEs, and comparing estimators based on their limiting relative efficiency and asymptotic relative efficiency. 2. Many of the problems involve showing that certain sample statistics are consistent estimators or finding their asymptotic distributions. Some problems calculate variances and biases of estimators. 3. Several problems compare different estimators for a parameter in terms of their limiting relative efficiency and asymptotic relative efficiency to determine which is more efficient.

Uploaded by

Mulyana Kahfi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

MTH 515a: Inference-II

Assignment No. 5: Asymptotically Efficient Estimators


1. Let X1 , X2 , . . . be a sequence of i.i.d. random variables with finite mean µ and finite
1
Pn √ d
variance σ 2 > 0. Let S 2 = n−1 2
i=1 (Xi − X) . Show that n(S 2 − σ 2 ) → Z ∼
N(0, (γ − 1)σ 4 ), as n → ∞, where γ is the kurtosis of the parent distribution. Hence
show that S 2 is a consistent estimator of σ 2 .
√
 
X −µ
2. Consider the set-up of Problem 1. Find the asymptotic distributions of n
S 2 − σ2
√
and n(X + S − µ − σ).
 
Xi
3. Let , i = 1, 2, . . . be a sequence of i.i.d. random vectors with E(X1 ) = µ1 ∈
Yi
R, E(Y1 ) = µ2 ∈ R, Var(X1 ) = σ12 > 0, Var(Y1 ) = σ22 > P 0, Corr(X1 , Y1 ) = ρ ∈
(−1, 1), Var(X1 ) < ∞ and Var(Y1 ) < ∞. Let S1 = n ni=1 (Xi − X)2 , S2 2 =
4 4 2 1

1
Pn 2 1
Pn S1,2
n i=1 (Yi − Y ) , S1,2 = n i=1 (Xi − X)(Yi − Y ) and r = S1 S2 . Show that, as
n → ∞,  2   2 
√ S1 σ1
d
n   2 
S2 −  σ22  → N3 (0, Σ∗ ),
S1,2 ρσ1 σ2
 
∗ ∗ Xi
for some p.d. matrix Σ . What are elements of Σ ?. When , i = 1, 2, . . .
Yi
follow bivariate normal distribution, show that, as n → ∞,
√ d
n(r − ρ) → N (0, (1 − ρ2 )2 )
√
 
1 1+r 1 1+ρ d
and n ln − ln → N (0, 1)
2 1−r 2 1−ρ

4. Let X1 , X2 , . . . be a sequence of i.i.d. U(0, θ) random variables, where θ ∈ Θ =


d
(0, ∞). Let X(n) = max{X1 , . . . , Xn }, n = 1, 2, . . .. Show that n(θ − X(n) ) → Z ∼
n
Exp(θ), as n → ∞. Hence show that δn (X) = n+1 X(n) is a consistent estimator of
θ.
5. Let X1 , X2 , . . . be a sequence of i.i.d. random variables with common p.m.f./p.d.f.

gθ (x) = eθT (x)−ψ(θ) , x ∈ χ ⊆ R, θ ∈ Θ ⊆ R,


√ d
that n( n1 ni=1 T (Xi ) − ψ 0 (θ)) → N (0, ψ 00 (θ)), as
P
where Θ is an open set. Show
n → ∞. Hence show that n1 ni=1 T (Xi ) is a consistent estimator of ψ 0 (θ).
P

6. Let X1 , . . . , Xn be i.i.d. N(θ, 1) random variables. Find a consistent estimator of



0, if θ 6= 0
g(θ) = .
1, if θ = 0
7. Let the random variables X1 , . . . , Xn have the common mean µ ∈ R and variance
σ 2 ∈ (0, ∞), and that Cov(Xi , Xj ) = ρj−i , j 6= i. For estimating µ, show that:

(a) the sample mean X may not be consistent if ρj−i = ρ 6= 0, ∀ i 6= j;


(b) the sample mean X is consistent if |ρj−i | ≤ M rj−i , with |r| < 1.

8. Let X1 , . . . , Xn be i.i.d. random variables with mean µ and finite variance σ 2 > 0.
Let 
X, w.p. 1 − n
δn (X) = ,
An , w.p. n
where n and An (n = 1, 2, . . .) are constants satisfying n → 0 and An n → ∞,
as n → ∞. Show that δn is consistent for estimating µ but Eθ ((δn − µ)2 ) 9 0, as
n → ∞.

9. Let X1 , . . . , Xn be i.i.d. U (0, θ), θ > 0, and let T = max{X1 , . . . , Xn }. Let h be


a four times differentiable function on (0, ∞) with bounded fourth derivative on
(0, ∞). Show that
θ 0 1 1
Eθ (h(T )) = h(θ) − h (θ) + 2 [θh0 (θ) + θ2 h00 (θ)] + O( 3 )
n n n
2
θ 0 1
and Varθ (h(T )) = 2
[h (θ)]2 + O( 3 ).
n n

10. Let X1 , X2 , . . . be i.i.d. Bin(1, θ), where θ ∈ Θ = (0, 1). For estimating g(θ) =
Varθ (X1 ), let δ1 be the UMVUE. Find the limiting bias, limiting variance, asymp-
totic bias and asymptotic variance of δ1 .

11. Let X1 , X2 , . . . be i.i.d. Bin(1, θ), where θ ∈ Θ = (0, 1). For estimating g(θ) = θ, let
(1) (2)
δn be the UMVUE and let δn be the minimax estimator under the SEL function.
(1) (2)
Find the LRE and ARE of δn relative to δn . Are these estimators asymptotically
efficient?

12. Let X1 , . . . , Xn be i.i.d. gaussian random variables with finite mean θ and variance
1. Let u0 be a given real P constant. For estimating g(θ) = Pθ (X1 ≤ u0 ), let δ1 be the
UMVUE and let δ2 = n ni=1 I(Xi ≤ u0 ). Find the LRE lδ1 ,δ2 and the ARE eδ1 ,δ2 .
1

Are these estimators asymptotically efficient? Which estimator will you prefer?

13. Let X1 , . . . , Xn be i.i.d. normal random variables having unknown mean µ and finite
known variance σ 2 > 0. Let hr (µ) = µr , r = 2, 3, 4. Determine up to O( n1 ),

(a) the variance of the UMVUE of hr (µ), r = 2, 3, 4;


(b) the bias of the MLE of hr (µ), r = 2, 3, 4.

Repeat (a) and (b) when µ is known and σ is unknown and hr (σ) = σ r , r = 2, 4.

14. Let X1 , . . . , Xn be i.i.d. Poisson random variables having mean θ > 0.


(a) Find the variance of the UMVUE of Pθ (X1 = 0) up to O( n1 ). Is the UMVUE
CAN?
(b) For estimating h(θ) = e−θ , find the LRE and ARE of δ2,n = [#Xi = 0]/n
relative to δ1,n = e−X .
15. Let X ∼ Bin(n, θ) and let h(θ) = θ(1 − θ). Find the UMVUE of h(θ). Is it CAN?
16. Let X1 , . . . , Xn be i.i.d. N(0, σ 2 ), where σ > 0.
(a) Show that δ1,n = nk ni=1 |Xi | is a consistent estimator of σ iff k = π2 ;
P p

(b) Determine the LRE and ARE of δ1,n , with k = π2 , relative to MLE δ2,n =
p
q P
1 n 2
n i=1 Xi .

17. Let X1 , . . . , Xn be i.i.d. with Eθ (X1 ) = µ, Varθ (X1 ) = 1 and Eθ ((X1 − µ)4 ) = µ4 ∈
2
R. Consider the unbiased estimators δ1,n = n1 ni=1 Xi2 − 1 and δ2,n = X − n1 of the
P
estimand h(θ) = µ2 .
(a) Determine the ARE e2,1 ;
(b) Show that e2,1 ≥ 1 if the distribution of X1 is symmetric;
(c) Find a distribution for which e2,1 < 1.
(d) Is δ2,n CAN?
18. Let X1 , . . . , Xn be i.i.d. Pareto random variables with p.d.f.
 θ
xθ+1
, if x > 1
fθ (x) = ,
0, otherwise

where θ ∈ (0, ∞) = Ω is unknown. Show that the MME estimator based on


ln Xi , i = 1, . . . , n, has asymptotic variance equal to CRLB. Compare the variance
of UMVUE of θ with CRLB.
19. Let X1 , . . . , Xn be i.i.d. double exponential random variables with p.d.f.
1 − |x|
fθ (x) = e θ , −∞ < x < ∞,
2θ
where θ ∈ (0, ∞) = Ω is unknown. Suppose that the estimand is h(θ) = θ.
(a) Show that the MME and the UMVUE are the same and their variances attain
CRLB;
(b) Show that X is asymptotically normal but not consistent;
(1)
(c) Find a MME (say, δn ) based on Xi2 , i = 1, . . . , n and show that it is CAN.
(2)
Compare its asymptotic variance with that of δn = n1 ni=1 |Xi |. Find the
P
(1) (2) (1) (2)
LRE and ARE of δn relative to δn . Are δn and δn asymptotically efficient?

You might also like