STAT3058/6058 Tutorial 1 Solutions

1. An individual faces the following possible losses:

Loss Size ($) Probability

1,000 0.001
100 0.1
0 0.899

If the utility function of a potential purchaser of insurance is:

u(x) = x0.6

where x > 0 is the individual’s wealth.

(a) Show that this person is risk averse.
(b) Calculate the maximum premium this individual would pay for insurance given the above loss
distribution and initial wealth of $2,000.

Solution:
(a) Based on the given utility function, for x > 0 we get:

u(x) = x0.6 > 0

u′ (x) = 0.6x−0.4 > 0

u′′ (x) = −0.24x−1.4 < 0
So we have decreasing marginal utility, which indicates the individual is risk averse.
(b) With insurance that cost $ G, the outcome is known and equals 2, 000 − G with utility (2, 000 −
G)0.6 .
Without insurance, we have a loss distribution with three possible outcomes and resulting
expected utility:

0.001($2, 000 − $1, 000)0.6 + 0.1($2, 000 − $100)0.6 + 0.899($2, 000)0.6

= 0.06309573 + 9.2736811 + 85.97608907
= 95.3128666

So set ($2, 000 − G)0.6 = 95.3128666 and we obtain G = $11.22.

1
2. An individual is facing a random loss, X, that is uniformly distributed on (0, 200). The individual can
buy partial insurance cover against this under which the individual would pay Y = min(X, 100), so that
the individual would pay the loss in full if the loss was less than 100, and would pay 100 otherwise. The
individual makes decisions using the utility function u(x) = x2/5 . Is the individual prepared to pay 80
for this partial insurance cover if the individual’s wealth is 300?

Solution:
The individual is choosing between not insuring (resulting in wealth of 300 − X) and insuring, in
which case the resulting wealth is a random variable as the individual is buying partial insurance
cover. So the individual is prepared to pay 80 for this partial insurance cover if

E[u(300 − X)] ≤ E[u(300 − 80 − Y )].

Noting that the density function of X is 1/200, we have

Z 200
1
E[u(300 − X)] = (300 − x)2/5 dx
200 0
200
−5
= (300 − x)7/5
200 × 7 0
= 8.2375,

and Z 100 Z 200 

1
E[u(300 − 80 − Y )] = (220 − x)2/5 dx + 1202/5 dx
200 0 100
!
100
1 −5
= (220 − x)7/5 + 100 × 1202/5
200 7 0

= 7.2805 < 8.2375.

Hence, the individual is not prepared to pay 80 for this partial insurance cover.

3. Prove Jensen’s inequality: if v(x) is convex, then E[v(X)] ≥ v(E[x]). Consider especially the case
v(x) = x2 .
Hint: Use the following characterisation of convexity: a function v(x) is convex if and only if, for every
x0 a line l0 (x) = a0 x + b0 exists, such that l0 (x0 ) = v(x0 ) and moreover l0 (x) ≤ v(x) for all x (usually
l0 (·) is a tangent line of v(·)). Take x0 = E[X].

Solution:
The hint suggests to use the following characterisation: a function v(x) is convex if and only if, for
every x0 a line l0 (x) = a0 x + b0 exists, such that l0 (x0 ) = v(x0 ) and moreover l0 (x) ≤ v(x) for all x
(usually l0 (·) is a tangent line of v(·)).
Take x0 = E[X], then from v(x) ≥ l0 (x) ∀x, we have

E[v(X)] ≥ E[l0 (x)] = E[a0 X + b0 ] = a0 E[X] + b0 = l0 (E[X]) = v(E[X]).

2
 The tangent line in x0 satisfies l0 (x0 ) = v(x0 ) and l0′ (x0 ) = v ′ (x0 ). If v(x) = x2 , then l0 (x) =
x20 + 2x0 (x − x0 ) must hold. We have v(x) − l0 (x) = (x − x0 )2 ≥ 0, so E[X 2 ] = E[v(X)] ≥ E[l0 (x)] =
v(E[X]) = µ2 .
Here we also have:
E[X 2 ] = V [X] + (E[X])2 ≥ (E[X])2 .

4. (a) Given the utility function u(·), how can we approximate the maximum premium P + for a risk X?
(b) Show that the approximation in (a) is exact if X ∼ N (µ, σ 2 ) and u(·) is exponential.
(c) It is known that the premium P that a policyholder with exponential utility function is willing
to pay for a N (1000, 1002 ) distributed risk satisfies P ≤ 1250. What can be said about their risk
aversion α? If the unit of risk is in dollars ($), then what is the unit of α?
(d) An insurer undertakes a risk X and after collecting the premium, he owns a capital w = 100. What
is the maximum premium the insurer is willing to pay to a reinsurer to take over the complete risk,
if his utility function u(w) = log(W ) and P (X = 0) = P (X = 36) = 0.5? Find not only the exact
value, but also the approximation as in (a).

Solution:
(a) Recall that the Taylor series of a function f of a variable x around the point x∗ is
2 3
f (x) = f (x∗ ) + f ′ (x∗ ) (x − x∗ ) + f ′′ (x∗ ) (x − x∗ ) /2 + f ′′′ (x∗ ) (x − x∗ ) /3! + · · · .

Let µ and σ 2 denote the mean and variance of X. So the Taylor series expansion of u around
mean risk µ (with constant wealth w) is

u(w − X) = u(w − µ) + u′ (w − µ)(µ − X) + u′′ (w − µ)(µ − X)2 /2 + u′′′ (w − µ)(µ − X)3 /3! + · · · .

Using the first several terms in the series expansion of u(·) in w − µ, we obtain

u w − P + ≈ u(w − µ) + µ − P + u′ (w − µ);

1
u(w − X) ≈ u(w − µ) + (µ − X)u′ (w − µ) + (µ − X)2 u′′ (w − µ).
2
Taking expectations on both sides of the latter approximation yields
1
E[u(w − X)] ≈ u(w − µ) + (µ − µ)u′ (w − µ) + (µ2 − 2µ2 + σ 2 + µ2 )u′′ (w − µ)
2
1
⇒ E[u(w − X)] ≈ u(w − µ) + σ 2 u′′ (w − µ).
2
Via equating E[u(w − X)] with u (w − P + ), we get
1 2 ′′
σ u (w − µ) ≈ µ − P + u′ (w − µ).


2

Hence, the maximum premium P + for a risk X is approximately

1 u′′ (w − µ)
P + ≈ µ − σ2 ′ .
2 u (w − µ)

3
 Moreover, recall that the (absolute) risk aversion coefficient r(w) of the utility function u(·) at
wealth w is given by
u′′ (w)
r(w) = − ′ .
u (w)

Then the maximum premium P + to be paid for a risk X is approximately

1
P + ≈ µ + σ 2 × r(w − µ).
2
Note that r(w) does not change when u(·) is replaced by au(·) + b. From the approximation
equation above, we can see that the risk aversion coefficient indeed reflects the degree of risk
aversion: the more risk averse one is, the larger the premium one is willing to pay.
(b) Recall that exponential utility function is u(w) = −αe−αw (α > 0).
Solving the premium equation we obtain:

u(w − P + ) = E[u(w − X)]

+
h i
⇒ −αe−α(w−P ) = E −αe−α(w−X)
+
⇒ −αe−αw × eαP = −αe−αw × E eαX


 

+
⇒ eαP = E eαX
 

In addition, recall that the mgf of the normal distribution N (µ, σ 2 ) is

Z
1 1 2 2 1 2 2
MX (t) = E ext = ext √ e− 2 (x−µ) /σ dx = eµt+ 2 σ t .
 
2πσ 2

Then we get
+
eαP = MX (α)
1
⇒ P + = log (MX (α)) /α = µ + σ 2 α.
2
Moreover, for the exponential utility function we have

u′′ (w − µ) −α3 e−α(w−µ)

r(w − µ) = − ′
= − 2 −α(w−µ) = α.
u (w − µ) α e
Via applying the approximation in (a), we get
1 1
P + ≈ µ + σ 2 × r(w − µ) = µ + σ 2 α.
2 2
Therefore, the premium approximation equals the exact maximum premium under this scenario.
(c) Following the results from (b), we get
1
P + = µ + σ 2 α = 1000 + 5000α = 1250
2
Thus we have α = 0.05.
 
The unit of σ 2 = E (X − µ)2 is ($) 2 , of σ 2 α it is ($) 1
(involved in premium calculation),
hence α must have the unit ($) −1 .

4
(d) First, solving the premium equation we get:

u(w − P + ) = E(u(w − X))

⇒ log(100 − P + ) = 0.5 × log(100 − 0) + 0.5 × log(100 − 36)

⇒ log(100 − P + ) = log(80)
⇒ P + = 20
Hence, based on utility equilibrium, we get the exact maximum premium P + = 20.
2 2
Next, with µ = E[X] = 18, σ 2 = E[X 2 ]−(E[X]) = 364 and r(w−µ) = 1
w−µ , the approximation
in (a) gives
1
P + ≈ µ + σ 2 × r(w − µ) = 19.9756.
2