Used in MLC manual

EXAM 3, FALL 2003

Please note: On a one-time basis, the CAS is releasing annotated solutions to Fall 2003 Examination 3 as a study aid to
candidates. It is anticipated that for future sittings, only the correct multiple-choice answers will be released.

1) Given:

i) 3 p40 = 0.990
ii) 6 p40 = 0.980

iii) 9 p40 = 0.965

iv) 12 p40 = 0.945

v) 15 p40 = 0.920

vi) 18 p40 = 0.890

For two independent lives aged 40, calculate the probability that the first death occurs after 6 years, but before
12 years.

A. Less than 0.050

B. At least 0.050, but less than 0.055
C. At least 0.055, but less than 0.060
D. At least 0.060, but less than 0.065
E. At least 0.065

P(First death after 6 years, but before 12 years)

= P(Neither dies during first 6 and not both alive at 12)
= P(Neither dies during first 6 and (not both alive at 12 given both alive after 6))
= P(Neither dies during first 6) * P(not both alive at 12 given both alive after 6) these are independent
2 2 −2
= 6 p40 (1 − 12 p40 6 p40 )
=0.067375

ANS: E

CONTINUED ON NEXT PAGE

1
 EXAM 3, FALL 2003

Use the following information for questions 2) and 3).

2) For a special fully discrete life insurance on (45), you are given:

i) i = 6%
ii) Mortality follows the Illustrative Life Table.
iii) The death benefit is 1,000 until age 65, and 500 thereafter.
iv) Benefit premiums of 12.51 are payable at the beginning of each year for 20 years.

Calculate the actuarial present value of the benefit payment.

A. Less than 100

B. At least 100, but less than 150
C. At least 150, but less than 200
D. At least 200, but less than 250
E. At least 250

The benefit payments of 1000 until age 65, and 500 thereafter are equivalent to 1000 A45 - 500 20| A45 .

1000 A45 - 500 20| A45 = 1000 A45 - 500 20 E45 A65
= 1000(0.20120) – 500 (0.25634)(0.43980)
= 144.83

Ans: B

(Since at issue the actuarial present value of the benefit premium equals the actuarial present value of the
benefit payment, this could also be done by evaluation the actuarial present value of the given premiums.)

CONTINUED ON NEXT PAGE

2
 EXAM 3, FALL 2003

The following is repeated for convenience.

3) For a special fully discrete life insurance on (45), you are given:

i) i = 6%
ii) Mortality follows the Illustrative Life Table.
iii) The death benefit is 1,000 until age 65, and 500 thereafter.
iv) Benefit premiums of 12.51 are payable at the beginning of each year for 20 years.

Calculate 19V , the benefit reserve at time t=19, the instant before the premium payment is made.

A. Less than 200

B. At least 200, but less than 210
C. At least 210, but less than 220
D. At least 220, but less than 230
E. At least 230

The desired reserve plus the final premium payment plus one year’s investment income needs to be able to
pay the year-20 death claims and to purchase a single premium policy of 500 on a 65-year old. That is:

( 19V + P)(1 + i) = 1000 q + p 64 500 A65

64

( 19V + 12.51)(1.06) = 1000 q + p 64 500 A65

64

( 19V + 12.51)(1.06) = 19.52 + 215.61 = 235.13

So: 19V = 209.31

Ans: B

CONTINUED ON NEXT PAGE

3
 EXAM 3, FALL 2003

4) Given:

2
µx = , for 0 ≤ x < 100
(100 − x)

Calculate q .
10| 65

1
A.
25
1
B.
35
1
C.
45
1
D.
55
1
E.
65

q is the probability of survival for 10 years and then death in the eleventh year for a life aged 65. That
10| 65
is survival to age 75, but not to age 76, given survival to age 65.

S ′( x)
µ x
=−
S (x )
equivalently, ∫µ x
= − log( S ( x )) + C

(100− x)
2

So, S ( x) = 2
The constant is determined by the fact that S(0) = 1.
100
−2
25 − 24
2 2
S (75) − S (76) 100 (25 − 24)(25 + 24) 1
The probability that we want is = −2 2 = 2 2 =
S (65) 100 35 57 25

Ans: A

CONTINUED ON NEXT PAGE

4
 EXAM 3, FALL 2003

5) Given:

i) Mortality follows De Moivre’s Law.

ii) eo20 = 30

Calculate q 20 .

1
A.
60
1
B.
70
1
C.
80
1
D.
90
1
E.
100

De Moivre’s law means that deaths are uniformly distributed from age 0 to age ω . We must determine ω .

A twenty year old is expected to live 30 more years, so the average time of death between 20 and ω is 20+30.
Since deaths are uniformly distributed, we must have ω =80.

q 20 is the probability that a twenty year old dies in the next year. As we saw above, times of death for 20 year
1
olds are uniformly distributed over the next 60 years, so q20 = .
60

Ans: A

CONTINUED ON NEXT PAGE

5
 EXAM 3, FALL 2003

6) Let Z1 be the present value random variable for an n-year term insurance of 1 on (x), and let Z2 be the
present value random variable for an n-year endowment insurance of 1 on (x). Claims are payable at the
moment of death.

Given:

i) v n = 0.250
ii) n p x = 0.400

iii) E[ Z2 ] = 0.400
iv) Var[ Z 2] = 0.055

Calculate Var[ Z1] .

A. 0.025
B. 0.100
C. 0.115
D. 0.190
E. 0.215

Let M be an n-year pure endowment of 1 on (x). Then:

E[ Z2 ] = 0.400 = E[ Z1 + M ] = E[ Z1] + (0.25)(0.40) So: E[ Z1 ] = 0.300

2 2 2
Var[ Z 2] = 0.055 = E[ Z 2 ] − E[ Z 2 ] So: E[ Z 2 ] = 0.055 + 0.160 = 0.215
2 2 2
E[ Z 2 ] = E[( Z1 + M ) ] = E[ Z1 ] + E[2 Z1M ] + E[ M ]
2

Now, it is impossible to collect on both the term insurance and the pure endowment, so the middle term is 0.
2 2
E[ M ] = (0.400)(0.250) = 0.025 So: E[ Z1 ] = 0.215 − 0.025 = 0.190
2

2 2
E[Z1 ] = 0.300 = 0.090
2 2
Var[ Z1] = E[ Z1 ] − E[ Z 1 ] = 0.190 − 0.090 = 0.100

Ans: B

CONTINUED ON NEXT PAGE

6
 EXAM 3, FALL 2003

7) Given:

i) i = 5%
ii) The force of mortality is constant.
iii) eo = 16.0
x

Calculate 20| Ax .

A. Less than 0.050

B. At least 0.050, but less than 0.075
C. At least 0.075, but less than 0.100
D. At least 0.100, but less than 0.125
E. At least 0.125

δ = ln( 1 + 0.05) = 0.0488

o
µ = 1 / ex = 0.0625
µ 0.0625
20 Ax = e − 20( µ +δ ) = e −20( 0.0625+ 0. 0488)
µ +δ 0.0625 + 0.0488
= 0.0606

Ans: B

CONTINUED ON NEXT PAGE

7
 EXAM 3, FALL 2003

8) Given:

i) i = 6%
ii) 10 E40 = 0.540

iii) 1000 A40 = 168

iv) 1000 A50 = 264

Calculate 1000 10 P( A40 ) , the benefit premium for a 10-payment fully discrete life insurance of 1,000 on (40).

A. 21.53
B. 21.88
C. 22.19
D. 22.51
E. 22.83

10 P( A40 )a&&40:10 = A40

a&&40 = a&&40:10 + 10 E40a&&50

1
i=6% so d = 1 − = 0.0566
1.06

da&&40 + A40 = 1 So: a&&40 = 14.70

da&&50 + A50 = 1 So: a&&50 = 13.00

So: 14.70 = &&

a40:10 + (0.540)(13.00) Hence: a&&40:10 = 7.68

168
And:100010 P( A40 ) = = 21.875
7.68

ANS: B

CONTINUED ON NEXT PAGE

8
 EXAM 3, FALL 2003

9) I a and I a&& represent the standard increasing annuities. A person aged 20 buys a special five-year
temporary life annuity-due, with payments of 1, 3, 5, 7, and 9.

Given:
i) a&&20:4 = 3.41
ii) a 20:4 = 3.04
iii) ( I a&&) 20:4 = 8.05
iv) ( I a ) 20:4 = 7.17

Calculate the net single premium.

A. Less than 18.0

B. At least 18.0, but less than 18.5
C. At least 18.5, but less than 19.0
D. At least 19.0, but less than 19.5
E. At least 19.5

The (conditional) payments of 1, 3, 5, 7, and 9 are the same as payments of 1, 1+2(1), 1+2(2), 1+2(3), 1+2(4).

The net single premium is 1 + a 20:4 + 2 ( I a ) 20:4 = 1 + 3.04 + 2(7.17) = 18.38

Ans: B

CONTINUED ON NEXT PAGE

9
 EXAM 3, FALL 2003

10) For a special fully continuous last-survivor life insurance of 1 on (x) and (y), you are given:

i) δ = 0.05
ii) T(x) and T(y) are independent.
iii) µ (x+t) = µ (y+t) = 0.07, for t>0.
iv) Premiums are payable until the first death.

Calculate the level annual benefit premium.

A. Less than 0.050

B. At least 0.050, but less than 0.075
C. At least 0.075, but less than 0.100
D. At least 0.100, but less than 0.125
E. At least 0.125

The force of mortality for (x) and (y) is constant. The lives are independent so the force of mortality for the
joint life status is 0.07+0.07 = 0.14.

In the case of constant mortality of µ and constant force of interest ofδ , the actuarial present value of a
µ
continuous insurance is .
µ +δ

0.14
Applying this to the joint life status we see that a first-to-die insurance has value = 0.737
0.14 + 0.05
1 − 0.737
The corresponding annuity, the first-to-die annuity, has actuarial present value = 5.263
0.05
This is the annuity corresponding to the premium payments.

To compute the actuarial present value of the last-to-die insurance, we observe that this insurance is identical
to insurance on each of the two lives minus the first-to-die insurance.

0.07
Insurance on each of the two lives has value = 0.583
0.07 + 0.05

So, the net single premium for the last-to-die insurance is 2(0.583) – 0.737 = 0.430

0.430
And the level annual benefit premium is = 0.082
5.263

Ans: C

CONTINUED ON NEXT PAGE

10
 EXAM 3, FALL 2003

11) Given:

q (1)
x q (2)
x q (3)
x q (xτ )
x < 40 0.10 0.04 0.02 0.16
x ≥ 40 0.20 0.04 0.02 0.26

Calculate q38(1) .
5|

A. Less than 0.06

B. At least 0.06, but less than 0.07
C. At least 0.07, but less than 0.08
D. At least 0.08, but less than 0.09
E. At least 0.09

We need to compute the probability of surviving all perils for five years and the succumbing to peril 1.

That is: (1-0.16)(1-0.16)(1-0.26)(1-0.26)(1-0.26)(0.20) = 0.0572

Ans: A

CONTINUED ON NEXT PAGE

11
 EXAM 3, FALL 2003

12) A driver is selected at random. If the driver is a “good” driver, he is from a Poisson population with a
mean of 1 claim per year. If the driver is a “bad” driver, he is from a Poisson population with a mean of 5
claims per year. There is equal probability that the driver is either a “good” driver or a “bad” driver. If the
driver had 3 claims last year, calculate the probability that the driver is a “good” driver.

A. Less than 0.325

B. At least 0.325, but less than 0.375
C. At least 0.375, but less than 0.425
D. At least 0.425, but less than 0.475
E. At least 0.475

This can easily be done using Bayes’ Theorem. Here is an alternative solution using life table techniques.
Suppose we had, say, 20,000 drivers. Since “good” and “bad” are equally likely we may assume that we have
3

The 10,000 “good” drivers each have probability = 1 e

−1
10,000 of each. = 0.0613 of having exactly 3
3!
accidents --- we expect 613 of them to have 3 accidents.
3

The 10,000 “bad” drivers have probability = 5 e

−5
=0.1404 of having exactly 3 accidents --- we expect 1404
3!
of them to have 3 accidents.

So, we expect 613+1404 = 2017 drivers to have 3 accidents and 613 of them are “good” drivers, so the
613
probability of a driver with 3 accidents being a good driver is = 0.304 .
2017

Ans: A

CONTINUED ON NEXT PAGE

12
 EXAM 3, FALL 2003

13) The Allerton Insurance Company insures 3 indistinguishable populations. The claims frequency of each
insured follows a Poisson process.

Given:

Population Expected Probability Claim

(class) time between of being in cost
claims class
I 12 months 1/3 1,000
II 15 months 1/3 1,000
III 18 months 1/3 1,000

Calculate the expected loss in year 2 for an insured that had no claims in year 1.

A. Less than 810

B. At least 810, but less than 910
C. At least 910, but less than 1,010
D. At least 1,010, but less than 1,110
E. At least 1,110

Populations I, II, and III are Poisson with annual expected frequency 12/12, 12/15, and 12/18 respectively.

P(0) = P(0| I ) P( I ) + P(0| II )P (II ) + P(0| III ) P ( III ) where 0 is the event 0 claims and I, II, and III are the
events the insured is in the respective population.

12 12 12
− 1 − 1 − 1
P(0) = e 12 + e 15 + e 18 = 0.4435
3 3 3

1 12
P (I ) 3 −
P( I |0) = P(0| I ) = e 12 = 0.2765
P (0) 0.4435
1 12
P( II ) 3 −
P( II |0) = P(0| II ) = 15 = 0.3377
P(0) 0.4435 e
1 12
P( III ) 3 −
P( III |0) = P(0| III ) = 18 = 0.3858 Note that these sum to 1 as they should.
P(0) 0.4435 e
12 12 12
So, the expected number of claims is (0.2765) + (0.3377) + (0.3858) = 0.8038
12 15 18
So, expected loss next year is 804.

Ans: A

CONTINUED ON NEXT PAGE

13
 EXAM 3, FALL 2003

14) The Independent Insurance Company insures 25 risks, each with a 4% probability of loss. The
probabilities of loss are independent.

On average, how often would 4 or more risks have losses in the same year?

A. Once in 13 years
B. Once in 17 years
C. Once in 39 years
D. Once in 60 years
E. Once in 72 years

 25  25 −25
P(0 losses) =   0.96 0.04 = 0.3604
25

 0
 25  25 −24
P(1 loss) =   0.96 0.04 = 0.3754
24

1
 25  25 −23
P(2 losses ) =   0.96 0.04 = 0.1877
23

2
 25  25− 22
P(3 losses ) =   0.96 0.04 = 0.0600
22

3
So, probability of 4 or more = 1-0.3604-0.3754-0.1877-0.600 = 0.0165

We would expect such an event to occur once every 1/0.0165 years = 60.61 years

Ans: D

CONTINUED ON NEXT PAGE

14
 EXAM 3, FALL 2003

15) Two actuaries are simulating the number of automobile claims for a book of business. For the
population that they are studying:

i) The claim frequency for each individual driver has a Poisson distribution.
ii) The means of the Poisson distributions are distributed as a random variable, Λ .
iii) Λ has a gamma distribution.

In the first actuary’s simulation, a driver is selected and one year’s experience is generated. This process of
selecting a driver and simulating one year is repeated N times.

In the second actuary’s simulation, a driver is selected and N years of experience are generated for that driver.

Which of the following is/are true?

I. The ratio of the number of claims the first actuary simulates to the number of claims the second
actuary simulates should tend towards 1 as N tends to infinity.
II. The ratio of the number of claims the first actuary simulates to the number of claims the second
actuary simulates will equal 1, provided that the same uniform random numbers are used.
III. When the variances of the two sequences of claim counts are compared the first actuary’s sequence
will have a smaller variance because more random numbers are used in computing it.

A. I only
B. I and II only
C. I and III only
D. II and III only
E. None of I, II, or III is true

The first actuary is generating the negative binomial random variable for the population. The second actuary
is generating the Poisson random variable for the driver that he picked.

There is no reason that the first two should be true, the second actuary’s result depends on which driver he
picked. The variances they observe could be higher or lower again depending on which driver the second
actuary picked; the number of random numbers used is irrelevant.

None of these statements is true.

Ans: E

CONTINUED ON NEXT PAGE

15
 EXAM 3, FALL 2003

16) According to the Klugman study note, which of the following is/are true, based on the existence of
moments test?

I. The Loglogistic Distribution has a heavier tail than the Gamma Distribution.
II. The Paralogistic Distribution has a heavier tail than the Lognormal Distribution.
III. The Inverse Exponential has a heavier tail than the Exponential Distribution.

A. I only
B. I and II only
C. I and III only
D. II and III only
E. I, II, and III

According to the existence of moments test, distribution A has a heavier tail than distribution B if A has fewer
moments than B. The moments for the various distributions are in the supplied tables, in all three cases the
distribution on the left has only finitely many moments and the distribution on the left has all moments, so all
three statements are true.

Ans: E

CONTINUED ON NEXT PAGE

16
 EXAM 3, FALL 2003

17) Losses have an Inverse Exponential distribution. The mode is 10,000.

Calculate the median.

A. Less than 10,000

B. At least 10,000, but less than 15,000
C. At least 15,000, but less than 20,000
D. At least 20,000, but less than 25,000
E. At least 25,000

From the supplied tables for an Inverse Exponential, mode = 10,000 è θ = 20,000.

The median of a continuous distribution is that value x so that F(x) = ½.

−x
F ( x) = e θ
for the Inverse Exponential (again from the table).
1 −20000
We want to find x so that: = e x
. So, x = −20000/ln( 1 2 ) = 28,854
2

Ans: E

CONTINUED ON NEXT PAGE

17
 EXAM 3, FALL 2003

18) A new actuarial student analyzed the claim frequencies of a group of drivers and concluded that they
were distributed according to a negative binomial distribution and that the two parameters, r and β , were
equal.

An experienced actuary reviewed the analysis and pointed out the following:

“Yes, it is a negative binomial distribution. The r parameter is fine, but the value of the β
1
parameter is wrong. Your parameters indicate that of the drivers should be claim-free, but
9
4
in fact, of them are claim-free.”
9

Based on this information, calculate the variance of the corrected negative binomial distribution.

A. 0.50
B. 1.00
C. 1.50
D. 2.00
E. 2.50

(1+ β )
−r
The probability of 0 claims from a negative binomial with parameters r and β is . We are told that
1
originally these two parameters were equal and that the result was . By inspection, r = β = 2 works.
9

So, the corrected r value is 2, as we were told it didn’t change.

−2
The corrected β value needs to satisfy (1+ β ) = . So, β = .
4 1
9 2

1 3
The variance of a negative binomial with parameters r and β is r β (1+ β ) = 2 ( ) = 1.5
2 2

Ans: C

CONTINUED ON NEXT PAGE

18
 EXAM 3, FALL 2003

19) For a loss distribution where x ≥ 2 , you are given:

z2
i) The hazard rate function: h ( x ) = , for x ≥ 2
2x
ii) A value of the distribution function: F(5) = 0.84

Calculate z.

A. 2
B. 3
C. 4
D. 5
E. 6

S (5) = 1 − F (5) = 0.16

∞

S ( x) = e ∫− ∞
− h ( x )dx

5 2

S (5) = 0.16 = e ∫2
− z dx
2x

2
−( z )(ln5 − ln 2 )
=e 2

2
ln 0.16 = −( z 2 )(ln 5 − ln 2)
z2 = 4
z=2

Ans: A

CONTINUED ON NEXT PAGE

19
 EXAM 3, FALL 2003

20) Let X be the size-of-loss random variable with cumulative distribution function F(x) as shown below:

F(x)

Which expression(s) below equal(s) the expected loss in the shaded region?

∞
I. ∫
K
xdF ( x)

II. E ( x ) − ∫ xdF ( x ) − K[1 − F ( K )]

K

0
∞
III. ∫
K
[1 − F ( x)] dx

A. I only
B. II only
C. III only
D. I and III only
E. II and III only

∞
I is false, since the correct representation would be ∫
K
xdF ( x) − KG (K )
II and III are true.

Ans: E

CONTINUED ON NEXT PAGE

20
 EXAM 3, FALL 2003

10 6
21) The cumulative loss distribution for a risk is F ( x ) = 1− .
( x + 103 ) 2

Calculate the percent of expected losses within the layer 1,000 to 10,000.

A. 10%
B. 12%
C. 17%
D. 34%
E. 41%

Total losses:
∞ ∞ 10 6
∫0
G ( x ) dx = ∫
0 ( x + 1000) 2
dx

∞ 1
= 10 6 ∫1000 du = 1000
u2

Layer losses:
10000 10000 10 6
∫1000
G( x) dx = ∫
1000 ( x + 1000) 2
dx

1
= 10 6 ∫2000
11000
dx = 409
u2

Percentage = 409/1000 = 41%

Ans: E

CONTINUED ON NEXT PAGE

21
 EXAM 3, FALL 2003

22) The severity distribution function of claims data for automobile property damage coverage for Le
Behemoth Insurance Company is given by an exponential distribution, F(x).

F ( x ) =1 − exp( 5000
−x )

To improve the profitability of this portfolio of policies, Le Behemoth institutes the following policy
modifications:

i) It imposes a per-claim deductible of 500.

ii) It imposes a per-claim limit of 25,000.

Previously, there was no deductible and no limit.

Calculate the average savings per (old) claim if the new deductible and policy limit had been in place.

A. 490
B. 500
C. 510
D. 520
E. 530

Unlimited coverage:

∞
E ( X ) = ∫ G( x ) dx
0
∞
where G(x) = 1 – F(x)
∫
−x
= e 5, 000
dx = 5,000
0

Limited coverage:

L
E ( X ; d , L) = ∫ G( x) dx
d
25, 000
∫
−x
= e 5 , 000
dx = −5,000(e − 5 − e −0.1 ) = 4,490
500

Savings = 5,000 – 4,490 = 510

Ans: C

CONTINUED ON NEXT PAGE

22
 EXAM 3, FALL 2003

23) F(x) is the cumulative distribution function for the size-of-loss variable, X.

P, Q, R, S, T, and U represent the areas of the respective regions.

What is the expected value of the insurance payment on a policy with a deductible of “DED” and a limit of
“LIM”? (For clarity, that is a policy that pays its first dollar of loss for a loss of DED + 1 and its last dollar of
loss for a loss of LIM.)

P
LIM

Q R
DED

S T U

F(X)

A. Q
B. Q+R
C. Q+T
D. Q+R+T+U
E. S+T+U

Lee integrates horizontally: G(x) = 1 – F(x)

∫
LIM
G( x) dx = Q + R
DED

Ans: B

CONTINUED ON NEXT PAGE

23
 EXAM 3, FALL 2003

24) Zoom Buy Tire Store, a nationwide chain of retail tire stores, sells 2,000,000 tires per year of various
sizes and models. Zoom Buy offers the following road hazard warranty:

“If a tire sold by us is irreparably damaged in the first year after purchase, we’ll replace it free, regardless of
the cause.”

The average annual cost of honoring this warranty is $10,000,000, with a standard deviation of $40,000.
Individual claim counts follow a binomial distribution, and the average cost to replace a tire is $100.

All tires are equally likely to fail in the first year, and tire failures are independent.

Calculate the standard deviation of the replacement cost per tire.

A. Less than $60

B. At least $60, but less than $65
C. At least $65, but less than $70
D. At least $70, but less than $75
E. At least $75

E(X) = 100; E(S) = 10,000,000; Var(S) = 40,0002

m = tires sold = 2 million
E(N) = E(S)/E(X) = 100,000
Frequency of loss = E(N)/m = 0.05 = q
Var(N) = mq(1-q) = (2,000,000)(0.05)(1-0.05) = 95,000

Var(S) = E(N)Var(X) + Var(N)E(X)2

40,0002 = (100,000)Var(X) + (95,000)(1002 )

Var(X) = 6,500
Standard Deviation = 80.62

Ans: E

CONTINUED ON NEXT PAGE

24
 EXAM 3, FALL 2003

25) Daily claim counts are modeled by the negative binomial distribution with mean 8 and variance 15.
Severities have mean 100 and variance 40,000. Severities are independent of each other and of the number of
claims.

Let σ be the standard deviation of a day’s aggregate losses.

On a certain day, 13 claims occurred, but you have no knowledge of their severities.

Let σ ′ be the standard deviation of that day’s aggregate losses, given that 13 claims occurred.

σ
Calculate −1 .
σ′

A. Less than –7.5%

B. At least –7.5%, but less than 0
C. 0
D. More than 0, but less than 7.5%
E. At least 7.5%

Beginning of day:

Var(S) = E(N)Var(X) + Var(N)E(X)2

= (8)(40,000) + (15)(1002 )
= 470,000
Std Dev (S) = 685.56

End of day:

Var(S) = N * Var(X)
= (13)(40,000)
= 520,000
Std Dev (S) = 721.11

Increase = 685.56 / 721.11 – 1 = – 4.9%

Ans: B

CONTINUED ON NEXT PAGE

25
 EXAM 3, FALL 2003

26) A fair coin is flipped by a gambler with 10 chips. If the outcome is “heads,” the gambler wins 1 chip; if
the outcome is “tails,” the gambler loses 1 chip.

The gambler will stop playing when he either has lost all of his chips or he reaches 30 chips.

Of the first ten flips, 4 are “heads” and 6 are “tails.”

Calculate the probability that the gambler will lose all of his chips, given the results of the first ten flips.

A. Less than 0.75

B. At least 0.75, but less than 0.80
C. At least 0.80, but less than 0.85
D. At least 0.85, but less than 0.90
E. At least 0.90

After 10 flips, the gambler is down 2 and has 8 chips left.

He will quit if he loses his remaining 8 chips or reaches 30 chips (22 more than 8).

Pr(down 8 before up 22) = 22/(8+22) = 0.733

Ans: A

CONTINUED ON NEXT PAGE

26
 EXAM 3, FALL 2003

27) Not-That-Bad-Burgers employs exactly three types of full-time workers with the following annual
salaries:

Title Annual Salary

Burger Flipper (B) 6,000
Cashier (C) 8,400
Manager (M) 12,000

Each month the employees are promoted or demoted according to the following transition probability matrix:

B C M
1 1 0
B 2 2
C 14 1
2
1
4
M 0 1 2
3 3

Calculate the average long-term annual salary of an employee at Not-That-Bad-Burgers.

A. 8,133
B. 8,533
C. 9,067
D. 9,200
E. 9,467

∑ r ( j)π j
j

(1)π B = 1 π B + 1 π C
2 4
( 2)π C = 1 2 π B + 1 2 π C + 13 π M

(3)π M = 1 4 π C + 2 3 π M
( 4)1 = π B + π C + π M
(1) → 1 2 π B = 1 4 π C → π B = 1 2 π C

(3) → 1 π M = 1 π C → π M = 3 π C
3 4 4
( 4) → 1 = 1 π C + π C + 3 π C
2 4
π B = 9 ,π C = 9 ,π M = 3 9
2 4
(2/9)(6,000)+(4/9)(8,400)+(3/9)(12,000)=9,067

Ans: C

CONTINUED ON NEXT PAGE

27
 EXAM 3, FALL 2003

28) A Markov chain with five states has the following transition probability matrix:

0.50 0.25 0.20 0.05 0.00

0.25 0.20 0.05 0.00 0.50
0.20 0.05 0.00 0.50 0.25
0.50 0.00 0.00 0.50 0.00
0.50 0.00 0.00 0.00 0.50

How many classes does this Markov chain have?

A. 1
B. 2
C. 3
D. 4
E. 5

State 0 communicates with state 1.

State 1 communicates with state 2.
State 0 communicates with state 3.
State 4 is accessible from state 1. State 1 is accessible from state 4 (4-0-1).
Therefore, state 4 and state 1 communicate.

Therefore, all states communicate, so there is one class.

Ans: A

CONTINUED ON NEXT PAGE

28
 EXAM 3, FALL 2003

29) A ferry transports fishermen to and from a fishing dock every hour on the hour. The probability of
catching a fish in the next hour is a function of the number of fish caught in the preceding hour as follows:

Number of fish Probability of 0 Probability of 1 Probability of

caught in the fish caught in fish caught in 2+ fish caught
preceding hour the next hour the next hour in the next
hour
0 0.7 0.2 0.1
1 0.3 0.5 0.2
2+ (2 or more) 0.1 0.5 0.4

If a fisherman has not caught a fish in the hour before the next ferry arrives, he leaves; otherwise he stays and
continues to fish. A fisherman arrives at 11:00 AM and catches exactly one fish before noon.

Calculate the expected total number of hours that the fisherman spends on the dock.

A. 2 hours
B. 3 hours
C. 4 hours
D. 5 hours
E. 6 hours

Since the fisherman only stays if he catches a fish, the probabilities associated with 0 fish caught are not
relevant. Therefore,

 0 .5 0 .2 
P T =  
 0 .5 0 .4 
 0 .5 − 0 .2 
I − P =  
T

 − 0 .5 0 .6 
 0. 6 0. 2 0. 2 0. 2 
( I − P T ) −1 =  
 0. 2 0. 2 
0. 5 0. 5

E(Time in 1 | Started in 1) = 3
E(Time in 2 | Started in 1) = 1
E(Time in 0) = 1

Total expected time on the dock = 3 + 1 + 1 = 5

Ans: D

CONTINUED ON NEXT PAGE

29
 EXAM 3, FALL 2003

30) Speedy Delivery Company makes deliveries 6 days a week. Accidents involving Speedy vehicles occur
according to a Poisson process with a rate of 3 per day and are independent. In each accident, damage to the
contents of Speedy’s vehicles is distributed as follows:

Amount of damage Probability

$ 0 ¼
$2,000 ½
$8,000 ¼

Using the normal approximation, calculate the probability that Speedy’s weekly aggregate damages will not
exceed $63,000.

A. 0.24
B. 0.31
C. 0.54
D. 0.69
E. 0.76

Let Yi represent damages for accident i, and X(6) represent weekly damages.

E[Yi ] = 0(1 / 4) + 2,000(1 / 2) + 8,000(1 / 4) = 3,000

E[Yi 2 ] = 0(1 / 4) + 2,000 2 (1 / 2) + 8,000 2 (1 / 4) = 18million
E[ X ( 6)] = ( 6)(3)(3,000) = 54,000
Var[ X ( 6)] = (6)( 3) E[Yi 2 ] = 324 million
X (6) − 54,000 63,000 − 54,000
Pr( X ( 6) ≤ 63,000) ≅ Pr( ≤ )
324mil 324mil
= Φ (0.5) = 0.6915

Ans: D

CONTINUED ON NEXT PAGE

30
 EXAM 3, FALL 2003

31) Vehicles arrive at the Bun-and-Run drive-thru at a Poisson rate of 20 per hour. On average, 30% of
these vehicles are trucks.

Calculate the probability that at least 3 trucks arrive between noon and 1:00 PM.

A. Less than 0.80

B. At least 0.80, but less than 0.85
C. At least 0.85, but less than 0.90
D. At least 0.90, but less than 0.95
E. At least 0.95

λtp = ( 20)(1)(0.30) = 6
e − λtp (λtp) n
P( n) =
n!
−6
P( 0) = e = 0.0025
e − 6 (6) 1
P(1) = = 0.0149
1!
e − 6 (6) 2
P( 2) = = 0.0446
2!

So the probability of at least 3 trucks = 1 – P(0) – P(1) – P(2)

= 1 – 0.0025 – 0.0149 – 0.0446 = 0.938

Ans: D

CONTINUED ON NEXT PAGE

31
 EXAM 3, FALL 2003

32) Ross, in Introduction to Probability Models, identifies four requirements that a counting process N(t)
must satisfy.

Which of the following is NOT one of them?

A. N(t) must be greater than or equal to zero.

B. N(t) must be an integer.
C. If s<t, then N(s) must be less than or equal to N(t).
D. The number of events that occur in disjoint time intervals must be independent.
E. For s<t, N(t)-N(s) must equal the number of events that have occurred in the interval (s,t].

D only applies if the counting process possesses independent increments.

Ans: D

CONTINUED ON NEXT PAGE

32
 EXAM 3, FALL 2003

33) Stock I and stock II open the trading day at the same price. Let X(t) denote the dollar amount by which
stock I’s price exceeds stock II’s price when 100t percent of the trading day has elapsed. {X (t ),0 ≤ t ≤ 1} is
modeled as a Brownian motion process with variance parameter σ 2 = 0.3695 .

After ¾ of the trading day has elapsed, the price of stock I is 40.25 and the price of stock II is 39.75.

Calculate the probability that X (1) ≥ 0 .

A. Less than 0.935

B. At least 0.935, but less than 0.945
C. At least 0.945, but less than 0.955
D. At least 0.955, but less than 0.965
E. At least 0.965

Pr{ X (1) > 0 | X ( 3 4) = 0.50}

= Pr{[ X (1) − X ( 34 )] > −0.50 | X ( 34 ) = 0.50}
= Pr{[ X (1) − X ( 34 )] > −0.50}
= Pr{ X ( 14) > −0.50}
X ( 14) − 0.50
= Pr{ > }
σ /4
2
σ 2 /4
X ( 14)
= Pr{ > −1 / σ }
σ 2 /4
= Φ (1 / σ ) = Φ(1.645) = 0.95

Ans: C

CONTINUED ON NEXT PAGE

33
 EXAM 3, FALL 2003

34) Given:

i) Initial surplus is 10.

ii) Annual losses are distributed as follows:
Annual Probability
Loss
0 0.60
10 0.25
20 0.10
30 0.05

iii) Premium, paid at the beginning of each year, equals expected losses for the year.
iv) If surplus increases in a year, a dividend is paid at the end of that year. The dividend is equal to half of
the increase for the year.
v) There are no other cash flows.

Calculate ψ (10,2) , the probability of ruin during the first two years.

A. Less than 0.20

B. At least 0.20, but less than 0.25
C. At least 0.25, but less than 0.30
D. At least 0.30, but less than 0.35
E. At least 0.35

Year 1:
U0 P L Rebate U1 Prob
10 6 0 3 13
10 6 10 0 6
10 6 20 0 -4 0.10
10 6 30 0 -14 0.05

Year 2:
U1 P L Rebate U2 Prob
13 6 0 3 16
13 6 10 0 9
13 6 20 0 -1 0.60*0.10=0.06
13 6 30 0 -11 0.60*0.05=0.03
OR
6 6 0 3 9
6 6 10 0 2
6 6 20 0 -8 0.25*0.10=0.025
6 6 30 0 -18 0.25*0.05=0.0125

0.10+0.05+0.06+0.03+0.025+0.0125=0.2775

Ans: C

CONTINUED ON NEXT PAGE

34
 EXAM 3, FALL 2003

35) In Loss Models --- From Data to Decisions, Klugman et al. discuss survival probabilities and ruin
probabilities in terms of discrete time vs. continuous time and finite time vs. infinite time.

Based on that discussion, which of the following are true?

I. φ% (u , τ ) ≥ φ (u )
II. φ (u , τ ) ≥ φ% (u ,τ )
III. φ (u ) ≥ ψ (u )
IV. ψ% (u ) increases as the frequency with which surplus is checked increases.

A. I and IV only.
B. I, II, and III only.
C. I, II, and IV only.
D. I, III, and IV only.
E. II, III, and IV only.

The discrete time survival probability with finite time horizon is greater than the continuous time survival
probability with the same finite horizon (short term ruins during the gaps aren’t seen) which in turn is greater
than the infinite time horizon survival probability (if ruin occurs, it occurs at some finite time), so I is true.

Short term ruins during the time gaps are caught in the continuous time, but might not be in discrete time, so
II is false (the inequality goes the other way).

The continuous time, infinite horizon survival probability and continuous time, infinite horizon ruin
probability add to one, but neither is necessarily larger than the other, so III is false.

The more often you look, the more likely you are to catch a short term ruin, so IV is true.

Ans: A

CONTINUED ON NEXT PAGE

35
 EXAM 3, FALL 2003

36) A random variable having density function f ( x ) = 30 x 2 (1 − x ) 2, 0 < x < 1 is to be generated using the
rejection method with g ( x) = 1, for 0 < x < 1 .

15
Using c= , which of the following pairs (Y,U) would be rejected?
8

I. (0.35, 0.80)
II. (0.80, 0.40)
III. (0.15, 0.80)

A. II only
B. III only
C. I and II only
D. I and III only
E. I, II, and III

f (Y )
A pair (Y,U) is rejected if U > .
cg (Y )
g(Y) is identically 1.

It is easily verified that only III is rejected.

Ans: B

CONTINUED ON NEXT PAGE

36
 EXAM 3, FALL 2003

37) One of the most common methods for generating pseudorandom numbers starts with an initial value,
X 0 , called the seed, and recursively computes successive values, X n , by letting

X n = aX n−1 mod( m) .

Which of the following is/are criteria that should be satisfied when selecting a and m?

I. The number of variables that can be generated before repetition is large.

II. For any X 0 , generated numbers are independent Normal (0,1) variables.
III. The values can be computed efficiently on a computer.

A. I only
B. II only
C. I and III only
D. II and III only
E. I, II, and III

Once the pseudorandom numbers start to repeat, they really stop looking like random numbers, we want this
to occur as late as possible, so I is true.

The given procedure generates a sequence of residue classes mod(m), these are then associated with the
integers from 0 to m-1 (these are the X’s). The generated X’s are then divided by m to get number in the
interval [0,1]. Neither the X’s nor their rescaled counterparts are normally distributed, so II is false.

Typically, the pseudorandom numbers will be used by a computer program, so being able to generate them
efficiently on a computer is a good feature, thus III is true.

Ans: C

CONTINUED ON NEXT PAGE

37
 EXAM 3, FALL 2003

38) Using the Inverse Transform Method, a Binomial (10,0.20) random variable is generated, with 0.65 from
U(0,1) as the initial random number.

Determine the simulated result.

A. 0
B. 1
C. 2
D. 3
E. 4

Prob(0) = 0.810 = 0.1074 < 0.65 keep going

Prob(0 or 1) = 10(0.8) 9 (0.2)1 + 0.1074 = 0.2684 + 0.1074 < 0.65 keep going

10 ⋅ 9
Prob(0 or 1 or 2) = (0.8) 8 (0.2) 2 + 0.2684 + 0.1074 = 0.6778 > 0.65 Success.
2

So, generated value is 2.

Ans: C

CONTINUED ON NEXT PAGE

38
 EXAM 3, FALL 2003

39) When generating random variables, it is important to consider how much time it takes to complete the
process.

Consider a discrete random variable X with the following distribution:

k Prob(X=k)
1 0.15
2 0.10
3 0.25
4 0.20
5 0.30

Of the following algorithms, which is the most efficient way to simulate X?

A. If U<0.15 set X = 1 and stop. Expected number of tests = 1(.15) + 2(.10) + 3(.25) +4(.50)
If U<0.25 set X = 2 and stop.
If U<0.50 set X = 3 and stop.
If U<0.70 set X = 4 and stop.
Otherwise set X = 5 and stop.

B. If U<0.30 set X = 5 and stop. Expected number of tests = 1(.30) + 2(.20) + 3(.25) +4(.25)
If U<0.50 set X = 4 and stop.
If U<0.75 set X = 3 and stop.
If U<0.85 set X = 2 and stop.
Otherwise set X = 1 and stop.

C. If U<0.10 set X = 2 and stop. Expected number of tests = 1(.10) + 2(.15) + 3(.20) +4(.55)
If U<0.25 set X = 1 and stop.
If U<0.45 set X = 4 and stop.
If U<0.70 set X = 3 and stop.
Otherwise set X = 5 and stop.

D. If U<0.30 set X = 5 and stop. Expected number of tests = 1(.30) + 2(.25) + 3(.20) +4(.25)
If U<0.55 set X = 3 and stop.
If U<0.75 set X = 4 and stop.
If U<0.90 set X = 1 and stop.
Otherwise set X = 2 and stop.

E. If U<0.20 set X = 4 and stop. Expected number of tests = 1(.20) + 2(.15) + 3(.20) +4(.55)
If U<0.35 set X = 1 and stop.
If U<0.45 set X = 2 and stop.
If U<0.75 set X = 5 and stop.
Otherwise set X = 3 and stop.

Ans: D D has the smallest expected number of tests, so it is the most efficient of these algorithms.

CONTINUED ON NEXT PAGE

39
 EXAM 3, FALL 2003

40) W is a geometric random variable with parameter β = 7 3 .

Use the multiplicative congruential method:

X n+1 = aX n mod( m) where a = 6, m = 25, and X 0 = 7

and the inverse transform method.

Calculate W3 , the third randomly generated value of W.

A. 2
B. 4
C. 8
D. 12
E. 16

X1 = 6(7) = 42 = 17 mod(25)
X 2 = 6(17) = 102 = 2 mod(25)
X 3 = 6(2) = 12 = 12 mod(25)

12
The associated uniform pseudo-random variable is = 0.48
25

The geometric random variable with parameter 7/3 counts the number of Bernoulli trials (each with success
probability 0.3) needed to observe a success.

Using the inverse transform method, we have:

Probability of success on first try = 0.3 < 0.48 (didn’t happen on first trial) keep going

Probability of success on first or second try = 0.3 + (0.7)(0.3) = 0.51 > 0.48 Stop.

So, the first success occurred on the second trial.

Ans: A

(A grading adjustment was made for question 40 to account for an alternative approach for solving this problem.)

END OF EXAMINATION
40