SCHOOL OF FINANCE AND APPLIED ECONOMICS

Bachelor of Business Science – Actuarial Science, Finance & Financial Economics

END OF SEMESTER EXAMINATION

BSA 2206 INTRODUCTION TO STOCHASTIC MODELLING

DATE: Time: 2 Hours

Instructions
1. This examination consists of FIVE questions.
2. Answer Question ONE (COMPULSORY) and any other TWO questions
Section A- Question One [30 Marks]

a. A No-Claims Discount system operated by a motor insurer has the following four levels:

Level 1: 0% discount Level 2: 25% discount

Level 3: 40% discount Level 4: 60% discount

The rules for moving between these levels are as follows:

Following a year with no claims, move to the next higher level, or remain at level 4.
Following a year with one claim, move to the next lower level, or remain at level 1.
Following a year with two or more claims, move back two levels, or move to level 1
(from level 2) or remain at level 1.

For a given policyholder the probability of no claims in a given year is 0.85 and the
probability of making one claim is 0.12.

i. Write down the transition matrix of this chain. [2]

1
ii. Calculate the probability that a policyholder who is currently at level 2 will be at level
2 after year two. [3]

2
b.

i. State what is meat by the Memoryless property of an exponential distribution [1]

P X > a + b X > a = P(X > b)

ii. Prove the Memoryless property for an exponential distribution [3]

If + ∽ exp (0) then:

1 + > 2 + 3 + > 2 = 1(+ > 3)
1[+ > 2 + 3, + > 2] 1[+ > 2 + 3]
1 + >2+3 + >2 = =
1[+ > 2] 1[+ > 2]
89(:;<)
7
= 89:
= 7 89< = 1[+ > 3]
7

Example: 1 + > 30 + 10 + > 30 = 1(+ > 10)

c. Let @A be a standard Brownian Motion, explain the following characteristics of standard

Brownian motion:

i. Stationary increments [1]

A stochastic process X is said to be Stationary if the distribution XBC ;D and XBE ;D are
identical. i.e. their statistical properties remain unchanged (variance, mean e.t.c.) as time
elapses. The statistical properties only depend on the time lag

ii. Rescalling Property [2]

If @A is a standard Brownian motion and c is a positive constant, then the rescaled

G
process @ ∗ A = @HA is also a standard Brownian Motion.
H

1 1
I @HA = I @HA = 0
J J

3
 1 1 1
K2L @HA = K2L @HA = ×JN = N
J J J

iii. Time-inversion property [2]

If @A is a standard Brownian motion and c is a positive constant, then the rescaled

process @ ∗ A = N@C is also a standard Brownian Motion.
O

1
I N@G = I @G = 0
A J A

1
K2L N@G = N P K2L @G = N P × = N
A A N

iv. Show that Cov WB , WU = min (s, t) [2]

[\] @A , @^ = [\] @^ + @A − @^ , @^ = [\] @^ , @^ + [\] @A − @^ , @^

[\] @^ , @^ = ` = min (`, N)

v. Show that the following is a Martingale [4]

E
aA = 7 9bO 8A9
For ` < N
E
a^ = 7 89bd ;^9
E bd 8bd ;A9E
I[aA e^ = I 7 89bO ;A9 e^ = I 7 89bd 89 e^
Applying measurability
E E
= I 7 89bd 89(bd 8bd );A9 e^ = 7 89bd ;A9 I 7 89(bd 8bd ) e^
Applying independence
E
7 89bd ;A9 I[7 89(bd 8bd ) ]

4
 Applying the MGF of a normal distribution
E E ;^9E E
7 89bd ;A9 7 8A9 = 7 89bd ;^9 = a^
Hence it is a Martingale

d. An insurer knows from past experience that the number of claims received per month
has a Poisson distribution with mean 15, and that claim amounts have an exponential
distribution with mean 500. The insurer uses a security loading of 30%. Calculate the
insurer’s adjustment coefficient and give an upper bound for the insurer’s probability of
ruin, if the insurer sets aside an initial surplus of 1,000. [4]

f ag h − 1 − hJ = 0
0
hJ = f ag h − 1 = f( − 1)
0−h
1
f 0 1 500
h=0− =0− = − = 0.0000461538
J 1+i 500 1.3
The upper bound from Lundberg inequality:
o p ≤ 7 8rs = 7 8t.ttttuvGwxy×Gttt = 0.630

e.

i. For each of the following give an example of an application which may be useful to
the Strathmore University Cafeteria.

A simple random walk The number of customers in the shop each time the door is
opened
A Poisson process The number of customers at any given time during the day

A compound Poisson process value of goods sold at any time during the day

A markov Chain The number of customers owning using cards (as opposed to
cash) at the end of each week

5
ii. State the conditions under a general Random walk becomes a simple symmetric
random walk [2]

In a special case where the steps +z ′` can only take values of either +1 or -1, the process
is known as simple random walk.
Simple symmetric Random walk
In more special case where the the values of the steps can be +1 or -1 with equal
probabilities (0.5). I.e.
+1 |}Nℎ 2 L\323}Ä}NÅ \Ç 0.5
+z =
−1 |}Nℎ 2 L\323}Ä}NÅ \Ç 0.5
such a process is called simple symmetric random walk.
It is equally likely to step upwards or downwards.

[1]

Question Two

A. A gambler wins or loses one pound in each round of betting. The probability of winning
one pound in each round is  while that of losing one pound is É and independently of
the past events. |ℎ7L7  ≠ É . She starts betting with a firm determination that she
will stop gambling when either she wins b pounds or loses a pounds. We can model the
experiment using a simple symmetric random walk as follows:

We let ÖÜ , Öá , … , Öâ be the independent and identically distributed random wins and

â
losses and define the gambler’s wealth as äâ = ã + åçÜ Öå . We would also define

é = èêë â > í| äâ = í, äâ = î

a. Show that each of the following is a Martingale:

ñ äâ
i. ïâ = [3]
ó

6
ii. òâ = äâ − â(ó − ñ) [4]

iii. Find the value of C for which the following is a following is a Martingale

ôâ = ö â õ ä â [3]

7
b. Find the the expected number of her betting rounds before she will stop playing

[10]

ôâ = ö â õ ä â

ôâ;Ü = öâ;Ü õäâ ;ÖâúÜ = öõòâ ôâ

8
 I ùû;G eû = I öõòâ ôâ eû = öôâ I[õòâ |eû ]

öôâ I[õòâ ]

öôâ óõ + ñõ8Ü = ôâ

Ü
ö=
óõ + ñõ8Ü

Question Three

a. Distinguish between a Poisson process and a Compound Poisson process. [4]

b. Give two examples of quantities that can be modeled using each, Poisson process and
compound Poisson process. [4]

Risk Theory: The Poisson process is used in modelling (counting) the number of claims
while the compound Poisson process is used to model the total claim amounts.

Queuing process: The number of customers arriving at a service counter can be

modelled by a Poisson process while the total time of service for serving all the
customers/ the total amount of amount of deposits paid into a bank by all the
depositors.

9
c. Let NB be a random variable representing the number of claims arising from a portfolio
of insurance policies. Let Öå denote the size of the } A† claim and suppose
that ÖÜ , Öá , … + Ö°¢ are independent identically distributed random variables, all having
the same distribution as X. The claim sizes are independent of the number of claims.
Let £A = ÖÜ + Öá + ⋯ + Ö°¢ denote the total claim size. Show that: [5]

a •O L = a ¶O (log ag (L)) = 7 9A ©™ (´)8G

≠O
a•O 8•d L = I 7 ´ •O 8•d
= I[7 ´ ¨ÆC g¨ ]

≠OØd ≠OØd
= I I 7´ ¨ÆC g¨ |∞A8^ = I I 7´ ¨ÆC g¨ |∞A = I I 7 ´gC ;´gE ;⋯;´g≠O

I I 7 ´gC I 7 ´gC I 7 ´gE …×I 7 ´g≠O

¶O
I ag (L)×ag (L)× …×ag (L) = I[ ag L ]
¶O ≠O
I ag L = I 7 ±≤ ©™ ´
= I 7 ¶O ±≤ ©™ (´) = a¶O (ln ag (L))

= 7 [9A ©™ (´)8G ]

d. Consider a Poisson process NB which is Poisson distributed with parameter fN . Let ≥G be

the arrival time of the first event, ≥û be the inter-arrival time between (¥ − 1)`N event
and ¥Nℎ event. Show that ≥G , ≥P , … , 2L7 ≥û have exponential distribution with parameter
f. [7]

10
 Question Four

a. Let us consider two players A and B. They toss a fair coin. If the coin is a head, A wins 1
dollar from B, otherwise A losses 1 dollar to B. Given that A and B start with 2 and 3
initially. Find:
i. The model for the wealth process of player A, µû [2]

The wealth process of player A, the wealth of player A after the ¥ − Nℎ ∂2∑7 can
be modelled as follows:
û
µû = 2 + +z
zçG

1 ∏ |}¥` 1
Where: +z =
−1 ∏ Ä\`7` 1

ii. Show that µû is a Martingale [2]

11
 û;G
µû;G = 2 + +z = µû + +û;G
zçG

I µû;G eû = I µû + +û;G eû
Applying measurability
= µû + I +û;G eû
Applying independence
= µû + I[+û;G ]
I +û;G = 1×0.5 + −1×0.5 = 0
= µû + 0 = µû
Hence the Wealth process of player A is a Martingale

iii. The probability that A wins all the money from B [3]

Since I[|µû;G − µû |] < 1 we can apply the Doob’s optional stopping theorem
Define √ = inf {N > 0|µ≈ = 2 + 3 \L µ≈ = 0}
I µ≈ = I µt = 2
I µ≈ = I(µ≈ |∏ |}¥` 2ÄÄ ÇL\∑ «)1(µ≈ = 2 + 3) + I(µ≈ |∏ }` Lp}¥7»)1(µ≈
= 0)
2 = 2 + 3 ×1 µ≈ = 2 + 3 + (0)×1(µ≈ = 0)
2 + 3 ×1 µ≈ = 2 + 3 = 2
Probability that A wins
2
1 µ≈ = 2 + 3 =
2+3

iv. Probability that A loses all the money to B [1]

But 1 µ≈ = 2 + 3 + 1 µ≈ = −2 = 1

2 3
1 µ≈ = 0 = 1 − =
2+3 2+3

b. A gambler has 5 pounds and has the opportunity of playing a game in which the
probability is 0.3 that he wins an amount equal to his stake, and probability of 0.7 that
he loses his stake. If his capital is increased to 8 pounds, he will stop playing. If his capital
becomes 0, he has to stop of course. He is allowed to decide how much to stake at each

12
game, and he decides each time to bet, when this is possible, just sufficient in order to
increase his to 8, or if he does not have enough for this, to bet all that he has. Let
ùt , ùG , … , ùû be his capital in pounds after 0,1,2 … , ¥ game such that ùz is a Markov
chain.

i. Define a Markov chain [2]

ii. Draw the transition diagrams, identifying the state space, S. [3]

iii. Find the transition matrix, 1. [2]

13
iv. Compute the probability of ultimately increasing his capital to 8 pounds. [5]

14
 Question Five

a. Adjustment Coefficient is defined as the unique positive value of h that makes

¶O
aA = 7 8r O , where µA is a surplus process defined as µA = p + JN − zçG +z .

i. Show that the above definition can be simplified as the unique positive value of R
that satisfies: [5]
f(ag h − 1 − hJ = 0

For ` < N

I 7 8rÀÃ e^ = 7 8rÀÕ

—Ã
I 7 8rÀà e^ = I 7 8r(s;ŒB8 –ÆC œ– ) e^

O g 8 ≠O ≠
8r “;HU;Œ B8U 8 ¨ÆC ≠ÕúC g¨
I 7 ¨
e^
≠
O g ;ŒB8ŒU8 O ≠
8r “;HU8 ¨ÆC ≠ÕúC g¨
=I 7 ¨
e^

— —
r —à œ – r —à œ –
7 8r Õ 8rHA;H^ I 7 ÕúC e^ = 7 8r Õ 8rHA;H^ I[7 ÕúC ]

7 8r Õ 8rHA;H^ I 7´ •O 8•d
= 7 8r Õ 8rHA;rH^ a•O 8•d h

From chapter two we know the MGF of the increments of a compound Poisson
process

7 8r d 8rHA;rH^ 7 [9 A8^ ©™ (r)8G ]

For this to be a Martingale, 7 8rHA;H^;9 A8^ ©™ (r)8G = 1

−hJN + hJ` + f N − ` ag h − 1 = 0

−hJ N − ` + f N − ` ag h − 1 = 0

−hJ + f ag h − 1 = 0

f ag h − 1 − hJ = 0

15
ii. Claims on a portfolio of insurance policies arrive as a Poisson process with annual
rate λ. Individual claims are uniformly distributed between 0 and 50, and the
insurance company uses a premium loading of 12%. Show that the insurance
company’s adjustment coefficient is 0.0066 to four decimal places. [3]

iii. Individual claim amounts, follow a gamma distribution with parameters 0 = 20

and ‘ = 0.5 The insurance company calculates premiums using a premium
loading factor of 15% and has an initial surplus of 300. Show that for this
portfolio the value of R is 0.00648 correct to three significant figures. [5]

16
b. Define:
√ = inf (N > 0|µA < 0)
9
Assume that J > . Show that the Probability of ultimate ruin is given by:
’
[7]

f 8 ’8
9
“
1 √<∞ = 7 ’
0J

f ag h − 1 − hJ = 0
0
hJ = f ag h − 1 = f( − 1)
0−h
f
h=0−
J
7 8rs
o p =
I[7 8rÀ◊ |√ < ∞]
−U≈ has an exponential distribution which has a memoryless property (as we discussed in
chapter one) hence:
0
I 7 8rÀ◊ √ < ∞ = I 7 8rÀ◊ =
0−h

17
 9
7 8rs 7 8rs 7 8(’8H )s f 8(’89)s
o p = = 0 = = 7 H
I[7 8rÀ◊ ] 0J
0−h 0
f
0 − (0 − )
J
G
But J = f 1 + i
’

1 8
’Ÿ
s
o p = 7 G;Ÿ
1+i

18