Problem Set 19 (b) Show that g(X) = (−2)X is an unbiased estimator

Mathematical Methods for Economists1 for θ. (In fact, it is the best unbiased estimator, in
the sense of minimizing the average squared error.)
1. Let X and Y have a continuous distribution with joint (c) Explain intuitively why g(X) is a sillly choice for es-
p.d.f: timating θ, despite (b), and show how to improve it
( by finding an estimator h(X) for θ that is always at
x + y for 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1, least as good as g(X) and sometimes strictly better
f (x, y) =
0 otherwise than g(X). That is,

Compute the covariance Cov(X, Y ). |h(X) − θ| ≤ |g(X) − θ|,

2. Let random variable X has a continuous distribution with with the inequality sometimes strict.

in
p.d.f: ( 11. Consider a random variable X ∼ Expo(1), and define Y
2x for 0 < x < 1, to be the integer part of X + 1, that is
f (x) =
0 otherwise

e.
Y = i + 1, if and only if i ≤ X < i + 1,
Determine the value of d that minimizes

nc
for i = 0, 1, 2, . . .
(a) E([X − d]2 ) (a) Find the distribution of Y . What well-known distri-
(b) E(|X − d|) bution does Y have?

ra
(b) Find the conditional distribution of X − 4 given Y ≥
3. Suppose that a point X is chosen in accordance with the
5.
uniform distribution on the interval [0, 1]. Also suppose

nt
that after the value X = x has been observed (0 < x < 1), 12. Amar wants to sell his car. He decides to sell it to the first
a point Y is chosen in accordance with a uniform distri- person to offer at least Rs. 6 Lakhs for it. Assume that
se
bution on the interval [x, 1]. Determine E(Y ). the offers are independent exponential random variables
with mean Rs. 4 Lakh.
4. If a stick is broken at random into three pieces, what is
ic
the probability that the pieces can be put together in a (a) Find the expected number of offers Amar will have.
triangle? (b) Find the expected amount of money that Amar gets
om

for the car.

5. Let X1 and X2 be independent N (0, 1) random variables.
Find the probability density function of (X1 − X2 )2 /2. 13. Let X1 , . . . , Xn be independent, with Xj ∼ Expo(λj ). Let
M = min{X1 , . . . , Xn }. Show that M ∼ Expo(λ1 + · · · +
6. Let Y = eX , where X ∼ N (µ, σ 2 ). Then Y is said to have
on

λn ), and interpret this intutively.

a Log-Normal distribution; this distribution is of great
importance in Economics, Finance, and elsewhere. Find 14. A system, which is composed of two components, will func-
the CDF and PDF of Y (the CDF should be in terms of tion as long as at least one of the two components func-
ec

Φ). tions. When both components are operating, the lifetime

distribution of each is exponential with mean 1. However,
7. For X ∼ Pois(λ), find E(X!) (the average factorial of X), the distribution of the remaining lifetime of the good com-
if it is finite. ponent, after one fails, is exponential mean 21 . (The idea
is that after one component fails the other component car-
8. Let Z ∼ N (0, 1). Find E(|Z|). ries twice the load and hence has only half the expected
life time.) Find the lifetime distribution of the system.
9. Let Z ∼ N (0, 1) and S be the “random sign” indepen-
dent of Z, i.e., S is 1 with probability 1/2 and −1 with 15. If X has MGF M (t), what is the MGF of −X? What is
probability 1/2. Show that SZ ∼ N (0, 1). the MGF of a + bX, where a and b are constants?

10. Let X be a Pois(λ) random variable, where λ is fixed but 16. Let U1 , U2 , . . . , U40 be i.i.d Uniform(0,1) and X = U1 +
unknown. Let θ = e−3λ , and suppose that we are in- U2 + · · · + U40 . Find the MGF of X.
terested in estimating θ based on the data. Since X is
17. Let X ∼ Pois(λ), and let M (t) be the MGF of X. The cu-
what we observe, our estimator is a function of X, call
mulant generating function is defined to be g(t) = ln M (t).
it g(X). The bias of the estimator g(X) is defined to be
Expanding g(t) as a Taylor series
E(g(X)) − θ, i.e., how far off the estimate is on average;
the estimator is unbiased if its bias is 0. ∞
X cj j
g(t) = t
(a) For estimating λ, the r.v. X itself is an unbiased j=1
j!
estimator. Compute the bias of the estimator T =
(the sum starts at j = 1 because g(0) = 0), the coefficient
e−3X . Is it unbiased for estimating θ?
cj is called the jth cumulant of X. Find the jth cumulant
1 Topics: Probability distributions. Contact: econschool@gmail.com of X, for all j ≥ 1.

1
18. A random point (X, Y, Z) is chosen uniformly in the ball 27. Given X has a density
B = {(x, y, z) : x2 + y 2 + z 2 ≤ 1}. (
0.5, if −1 < x < 1
(a) Find the joint PDF of X, Y, Z. fX (x) =
0 otherwise
(b) Find the joint PDF of X, Y .
(c) Find the expression for the marginal PDF of X, as Define
an integral. (
X + 1, if 0 < X < 1
19. Let X and Y be i.i.d. Unif(0,1). Find the expected value Y =
−2X if −1 < X ≤ 0
and the standard deviation of the distance between X and
Y , that is, E(|X − Y |), σ(|X − Y |).
Obtain the density of Y , fY (y).
20. Let U1 , U2 , U3 be i.i.d. Unif(0,1), and let L =
28. Suppose X and Y are independent with E(X) = 1,

in
min{U1 , U2 , U3 }, M = max{U1 , U2 , U3 },
V(X) = 1, E(Y ) = 2, and V(Y ) = 1. Define Z = X + Y
(a) Find the marginal CDF and marginal PDF of M , and and W = XY . Calculate Cov(Z, W ).

e.
the joint CDF and joint PDF of L, M .
Hint: for the latter, start by considering Pr(L ≥ 29. Suppose that U and V are independently and identically
distributed with density

nc
l, M ≤ m)
(b) Find the conditional PDF of M given L. (
e−t , if t > 0
f (t) =

ra
21. A chicken lays a number N ∼ Pois(λ) of eggs. Each egg, 0 otherwise
independently, hatches a chick with probability p. Let X
be the number which hatch, so X|N ∼ Bin(N, p). Find Find the conditional density of X given Y if X = U and

nt
the correlation between N (the number of eggs) and X Y =U +V.
(the number of eggs which hatch). Simplify; your final
answer should work out to a simple function of p (the λ 30. Let X = 1 with probability p and 0 with probability 1 − p.
se
should cancel out) Let the conditional density of Y given X = 1 be uniform
over 0 < y < 1 and given X = 0 be uniform over 0 < y <
22. Let X be Hypergeometric with parameters b, w, n.
ic
2. Calculate Cov(X, Y ), V(Y ), V(X).
 
X
(a) Find E by thinking, without any complicated 31. Prove or disprove: If E(Y |X) = X, E(X|Y ) = Y , and
om

2 both E(X 2 ) and E(Y 2 ) are finite, then Pr(X = Y ) = 1.

calculations. (Possible HINT: Pr(X = Y ) = 1 if V(X − Y ) = 0.)
(b) Use (a) to get the variance of X:
32. An urn contains four balls; two of the balls are numbered
N −n
on

V(X) = npq with a 1, and the other two are numbered with a 2. Two
N −1 balls are drawn from the urn without replacement. Let
where N = w + b, p = w/N , q = 1 − p. X denote the smaller of the numbers on the drawn balls
ec

and Y the larger. Find the joint distribution of X and Y .

23. Let B ∼ Beta(a, b). Find the distribution of 1 − B. Also Find Cov(X, Y ).
explain why the result makes sense in terms of Beta being
the conjugate prior for the Binomial. 33. Suppose X1 is a standard normal random variable. Define
24. Harsh waits X ∼ Gamma(a, λ) minutes for the bus to (
−X1 if |X1 | < 1
work, and then waits Y ∼ Gamma(b, λ) for the bus go- X2 =
ing home, with X and Y independent. Is the ratio X/Y X1 otherwise
independent of the total wait time X + Y ?
(a) Show that X2 is also a standard normal random vari-
25. Let X ∼ Beta(a, b) and Y ∼ Beta(a + b, c) be independent able.
random variables. The joint PDF of (X, Y ) is fX,Y (x, y) =
(b) Obtain the cumulative distribution function of X1 +
Γ(a + b) a−1 Γ(a + b + c) a+b−1 X2 in terms of the cumulative distribution function
x (1 − x)b−1 y (1 − y)c−1 of a standard normal random variable.
Γ(a)Γ(b) Γ(a + b)Γ(c)
for 0 < x < 1, 0 < y < 1. Consider the transormation 34. The trinomial distribution of two random variables X and
U = XY and V = X. Find the joint PDF of U and V . Y is given by
Also determine the marginal PDF of U .
n!
26. Let the joint density of (X, Y ) be given by fX,Y (x, y) = px q y (1 − p − q)n−x−y
x!y!(n − x − y)!
(
(x + y), if 0 < x < 1, 0 < y < 1 for x, y = 0, 1, . . . , n and x + y ≤ n, where p ≥ 0, q ≥ 0,
fX,Y (x, y) =
0 otherwise and p + q ≤ 1.

Calculate Cov(X, Y ), E(X|X < Y ). (a) Find the marginal distribution of Y .

2
 (b) Find the conditional distribution of X given Y , and 40. Let X1 , X2 , . . . , Xn be independent and identically dis-
obtain its expected value. tributed positive random variables. For k ≤ n, find
(c) Find ρ(X, Y ). 
Xk

 Xi 
35. There are 10 empty boxes numbered 1, 2, . . . , 10 placed
 
 i=1 
E n
sequentially on a circle (i.e. for each 1 < i < 10,

X 
the box numbered i has box numbered i − 1 and i + 1
 Xi 
as neighbours. For box numbered 1 the neighbouring i=1

boxes are numbered 10 and 2 and for the box num-

bered 10 the neighbouring boxes are numbered 9 and 1). 41. A lot of n items contains k defectives, and m are selected
randomly and inspected. How should the value of m be
4 3
chosen so that the probability that at least one defective

in
item turns up is 0.90? Apply your answer to (a) n = 1000,
5 2 k = 10, and (b) n = 10, 000, k = 100.

e.
42. A prisoner is trapped in a cell containing 3 doors. The
first door leads to a tunnel that returns him to his cell
6 1

nc
after 2 days’ travel. The second leads to a tunnel that
returns him to his cell after 4 days’ travel. The third door
leads to freedom after 1 day of travel. If it is assumed that

ra
7 10 the prisoner always select doors 1, 2, and 3 with respective
probabilities 0.5, 0.3 and 0.2, what is the expected number
of days until the prisoner reaches freedom?

nt
8 9

We perform 100 independent trials. At each trial, one box 43. Let Y be a non-negative random variable and k be any
positive constant. Show that Pr(Y ≥ k) ≤ E(Y )/k, where
se
is selected with probability 1/10 and one ball is placed in
each of the two neighbouring boxes of the selected one. Pr denotes probability and E(·) denotes expectation.
Define Xk to be the number of balls in the kth box at the 44. Let X1 and X2 have the joint pdf f (x1 , x2 ) = 6x2 , 0 <
ic
end of 100 trials. x2 < x1 < 1, zero elsewhere. Find the marginal pdf of X1
om

and X2 . Find the conditional pdf of X2 , given X1 = x1 .

(a) Find E(Xk ) for 1 ≤ k ≤ 10. Find the conditional mean of X2 , given X1 = x1 . Verify
(b) Find Cov(Xk , X5 ) for 1 ≤ k ≤ 10. that E(X2 ) = E(E(X2 |X1 )) and V(E(X2 |X1 )) ≤ V(X2 ).
(Note: V(X) = E((X − E(X))2 ) = E(X 2 ) − (E(X))2 )
on

36. Let X1 , . . . , Xn be random variables such that E(Xi2 ) ≤ 1

for all i = 1, . . . , n. Let Si = X1 + · · · + Xi for i = 1, . . . , n.
Show that
ec

E( max Si2 ) ≤ n2
1≤i≤n

37. Let X1 , X2 , . . . , X2n be random variables such that

V (Xi ) = 4, i = 1, 2, . . . , 2n and Cov(Xi , Xj ) = 3,
1 ≤ i 6= j ≤ 2n. Find the variance V(X1 − X2 + X3 − X4 +
· · · + X2n−1 − X2n ).

38. Let Pn be the set of all subsets of the set [n] =

{1, 2, . . . , n}. If two elements of P8 are chosen at ran-
dom, find the expected number of elements (of [8]) that
they have in common.

39. Consider the following model for aggregate claim amounts

S:
S = X1 + X2 + · · · + XN

where the Xi are independent, identically distributed ran-

dom variables representing individual claim amounts and
N is a random variable, independent of the Xi , and rep-
resenting the number of claims. Let X have mean µX and
2
let N have mean µN and variance σN . Find E(SN ).