Fall 2003 Society of Actuaries

Course 3 Solutions

Question #1
Key: E

2 q30:34 = 2 p30:34 − 3 p30:34

2 p30 = ( 0.9 )( 0.8 ) = 0.72

2 p34 = ( 0.5 )( 0.4 ) = 0.20
2 p30:34 = ( 0.72 )( 0.20 ) = 0.144
2 p30:34 = 0.72 + 0.20 − 0.144 = 0.776
3 p30 = ( 0.72 )( 0.7 ) = 0.504
3 p34 = ( 0.20 )( 0.3) = 0.06
3 p30:34 = ( 0.504 )( 0.06 ) = 0.03024
3 p30:34 = 0.504 + 0.06 − 0.03024
= 0.53376

2 q30:34 = 0.776 − 0.53376

= 0.24224

Alternatively,

2 q30:34 = 2 q30 + 2 q34 − 2 q30:34

b
= 2 p30q32 + 2 p34q36 − 2 p30:34 1 − p32:36 g
= (0.9)(0.8)(0.3) + (0.5)(0.4)(0.7) – (0.9)(0.8)(0.5)(0.4) [1-(0.7)(0.3)]
= 0.216 + 0.140 – 0.144(0.79)
= 0.24224

Alternatively,

2 q30:34 = 3 q30 × 3 q34 − 2 q30 × 2 q34

= (1 − 3 p30 )(1 − 3 p34 ) − (1 − 2 p30 )(1 − 2 p34 )
= (1 − 0.504 )(1 − 0.06 ) − (1 − 0.72 )(1 − 0.20 )
= 0.24224

(see first solution for 2 p30 , 2 p34 , 3 p30 , 3 p34 )

Question #2
Key: E

1000 Ax = 1000 ⎡ Ax1:10 + 10 Ax ⎤

⎣ ⎦
= 1000 ⎡ ∫ e −0.04t e −0.06t (0.06)dt + e−0.4e−0.6 ∫ e−0.05t e−0.07t (0.07)dt ⎤
10 ∞
⎢⎣ 0 0 ⎥⎦

= 1000 ⎡ 0.06∫ e −0.1t dt + e −1 (0.07) ∫ e−0.12t dt ⎤

10 ∞

⎣⎢ 0 0 ⎦⎥
⎡ −0.10 t 10 −0.12 t ∞ ⎤
= 1000 ⎢ 0.06 ⎡ − e0.10 ⎤ + e −1 (0.07) ⎡ − e0.12 ⎤ ⎥
⎣ ⎣ ⎦0 ⎣ ⎦0 ⎦

= 1000 ⎡ 0.06 ⎡1 − e −1 ⎤ + 0.07 −1 ⎡ −1.2 ⎤ ⎤

⎦ 0.12 e ⎣1 − e ⎦ ⎦
⎣ 0.10 ⎣
= 1000 ( 0.37927 + 0.21460 ) = 593.87

Because this is a timed exam, many candidates will know common results for constant force
and constant interest without integration.

µ
For example Ax1:10 = (1 − 10 Ex )
µ +δ
−10 ( µ +δ )
10 E x =e
µ
Ax =
µ +δ

With those relationships, the solution becomes

1000 Ax = 1000 ⎡⎣ Ax1:10 + 10 Ex Ax +10 ⎤⎦
⎡⎛
= 1000 ⎢⎜
0.06 ⎞
⎟ 1− e
⎣⎝ 0.06 + 0.04 ⎠
(
−( 0.06+ 0.04 )10
)
− 0.06+ 0.04 )10 ⎛
+e ( ⎜
0.07 ⎞ ⎤
⎟⎥
⎝ 0.07 + 0.05 ⎠ ⎦

( )
= 1000 ⎡( 0.60 ) 1 − e −1 + 0.5833 e −1 ⎤⎥
⎣ ⎦
= 593.86
Question #3
Key: A

⎧c ( 400 − x ) x < 400

B=⎨
⎩ 0 x ≥ 400
100 = E ( B ) = ci400 − cE ( X ∧ 400 )
⎛ 300 ⎞
= ci400 − ci300 ⎜1 − ⎟
⎝ 300 + 400 ⎠
⎛ 4⎞
= c ⎜ 400 − 300i ⎟
⎝ 7⎠
100
c= = 0.44
228.6

Question #4
Key: C

Let N = # of computers in department

Let X = cost of a maintenance call
Let S = aggregate cost

Var ( X ) = ⎡⎣Standard Deviation ( X ) ⎤⎦ = 2002 = 40,000

2

( )
E X 2 = Var ( X ) + ⎡⎣ E ( X ) ⎤⎦
2

= 40,000 + 802 = 46, 400

E ( S ) = N × λ × E ( X ) = N × 3 × 80 = 240 N

( )
Var ( S ) = N × λ × E X 2 = N × 3 × 46, 400 = 139, 200 N

We want 0.1 ≥ Pr ( S > 1.2 E ( S ) )

⎛ S − E (S ) 0.2 E ( S ) ⎞ 0.2 × 240 N
≥ Pr ⎜ > ⎟⇒ ≥ 1.282 = Φ ( 0.9 )
⎝ 139, 200 N 139, 200 N ⎠ 373.1 N
2
⎛ 1.282 × 373.1 ⎞
N ≥⎜ ⎟ = 99.3
⎝ 48 ⎠
Question #5
Key: B

If you happen to remember this distribution from the Simulation text (example 4d in third
edition), you could use:

⎛ log (1 − u ) ⎞ log 0.95

n = Int ⎜ ⎟ + 1 = Int +1 = 0 +1 = 1
⎝ log q ⎠ log 0.1

For mere mortals, you get the simulated value of N from the definition of the inverse
transformation method:

f(1) = F(1) = 0.9

0.05 ≤ 0.9 so n = 1

log(
1 1−v1 ) 1
x1 = =− log 0.7 = 35.67
λ 0.01

The amount of total claims during the year = 35.67

Question #6
Key: D

µ x(τ ) = 0.02 ⇒ time of death = −50ln ( 0.35 ) = 52.491

µ(
adb )
= 0.25 ⇒ Prob ( non adb ) = 0.75
µ (τ )
0.775 > Prob ( non adb ) so death is accidental and benefit is 2 (see below for more detail)

L = 2e −δ T − 0.025 aT
⎛ 1 − e −δ t ⎞
=2e-δ T − 0.025 ⎜ ⎟
⎝ δ ⎠
−( 0.05 )( 52.491)
⎛ 1 − e −( 0.05)( 52.91) ⎞
= 2e − 0.025 ⎜ ⎟
⎜ 0.05 ⎟
⎝ ⎠
= 0.1449 − ( 0.025 )(18.55 )
= −0.319

Another way of seeing that Prob(non adb) = 0.75.

− µ x( ) ( r )
t τ
ADB deaths in time t = ∫ 0.005 e dr
0
(τ ) r
= 0.25∫ 0.02 e − µ x ( ) dr
t
0
(τ ) r
= 0.25∫ µ x( ) ( r ) e − µ x ( ) dr
t τ
0

= 0.25 × all deaths in time t

More detail on why benefit is 2:

Benefit (b) f(b) F(b)

1 0.75 0.75
2 0.25 1

F (1) < 0.775 < F ( 2 ) so benefit = 2

Question #7
Key: B

µ x(τ ) = µ x(1) + µ x( 2 ) + µ x( 3) = 0.0001045

px( ) = e −0.0001045t
τ
t

∞
APV Benefits = ∫ e −δ t 1,000,000 t px ( ) µ x( ) dt
τ 1
0
∞ −δ t
500,000 t px ( ) µ x( ) dt
τ
+∫ e
2
0
∞
+ ∫ e −δτ 200,000 t px ( ) µ x( ) dt
τ 3
0
1,000,000 ∞ −0.0601045t 500,000 ∞ −0.0601045t 250,000 ∞ −0.0601045t
= ∫
2,000,000 0
e dt + ∫
250,000 0
e dt +
10,000 ∫0
e dt

= 27.5 (16.6377 ) = 457.54

Question #8
Key: B

∞
APV Benefits = 1000 A40:20
1
+ ∑ k E401000vq40+k
k = 20
∞
APV Premiums = π a40:20 + ∑ k E401000vq40+k
k = 20
Benefit premiums ⇒ Equivalence principle ⇒
∞ ∞
1
1000 A40:20 + ∑ k E401000vq40 + k = π a40:20 + ∑ k E401000vq40 + k
k = 20 20

π = 1000 A40:
1
20
/ a40:20
161.32 − ( 0.27414 )( 369.13)
=
14.8166 − ( 0.27414 )(11.1454 )
= 5.11

While this solution above recognized that π = 1000P40:20

1
and was structured to take advantage
of that, it wasn’t necessary, nor would it save much time. Instead, you could do:

APV Benefits = 1000 A40 = 161.32

∞
APV Premiums =π a40:20 + 20 E40 ∑ k E60 1000vq60+ k
k =0
= π a40:20 + 20 E40 1000 A60
= π ⎡⎣14.8166 − ( 0.27414 )(11.1454 ) ⎤⎦ + ( 0.27414 )( 369.13)
= 11.7612π + 101.19
11.7612π + 101.19 = 161.32
161.32 − 101.19
π= = 5.11
11.7612
Question #9
Key: C

ln (1.06 )
A70 = δ A70 = ( 0.53) = 0.5147
i 0.06
1 − A70 1 − 0.5147
a70 = = = 8.5736
d 0.06 /1.06
⎛ 0.97 ⎞
a69 = 1 + vp69a70 = 1 + ⎜ ⎟ ( 8.5736 ) = 8.8457
⎝ 1.06 ⎠
( )
= α ( 2 ) a69 − β ( 2 ) = (1.00021)( 8.8457 ) − 0.25739
2
a69
= 8.5902

Note that the approximation ax( ) ≅ ax −

m ( m − 1) works well (is closest to the exact answer, only
2m
1
off by less than 0.01). Since m = 2, this estimate becomes 8.8457 − = 8.5957
4
Question #10
Key: C

The following steps would do in this multiple-choice context:

1. From the answer choices, this is a recursion for an insurance or pure endowment.
2. Only C and E would satisfy u(70) = 1.0.
1+ i
3. It is not E. The recursion for a pure endowment is simpler: u ( k ) = u ( k − 1)
pk −1
4. Thus, it must be C.

More rigorously, transform the recursion to its backward equivalent, u ( k − 1) in terms of u ( k ) :

⎛ q ⎞ ⎛ 1+ i ⎞
u ( k ) = − ⎜ k −1 ⎟ + ⎜ ⎟ u ( k − 1)
⎝ pk −1 ⎠ ⎝ pk −1 ⎠
pk −1u ( k ) = − qk −1 + (1 + i ) u ( k − 1)
u ( k − 1) = vqk −1 + vpk −1 u ( k )

This is the form of (a), (b) and (c) on page 119 of Bowers with x = k − 1 . Thus, the recursion
could be:

Ax = vqx + vpx Ax +1
or A1x: y − x = vq x + vp x Ax1+1: y − x −1
or Ax: y − x = vqx + vpx Ax +1: y − x −1

Condition (iii) forces it to be answer choice C

u ( k − 1) = Ax fails at x = 69 since it is not true that

(
A69 = vq69 + vp69 (1) )
u ( k − 1) = A1x: y − x fails at x = 69 since it is not true that
1
A69:1 (
= vq69 + vp69 (1) )
u ( k − 1) = Ax: y − x is OK at x = 69 since

(
A69:1 = vq69 + vp69 (1) )
Note: While writing recursion in backward form gave us something exactly like page 119 of
Bowers, in its original forward form it is comparable to problem 8.7 on page 251. Reasoning
from that formula, with π h = 0 and bh +1 = 1 , should also lead to the correct answer.
Question #11
Key: A

You arrive first if both (A) the first train to arrive is a local and (B) no express arrives in the 12
minutes after the local arrives.
P ( A) = 0.75
Expresses arrive at Poisson rate of ( 0.25 )( 20 ) = 5 per hour, hence 1 per 12 minutes.

e −110
f ( 0) = = 0.368
0!
A and B are independent, so
P ( A and B ) = ( 0.75 )( 0.368 ) = 0.276
Question #12
Key: E

(This is covered in Probability Models, section 4.6 in the eighth edition).

Let states be 1, 2, 3 for acutely ill, in remission, cured/dead.

The transition matrix is:

⎛ 0.6 0.3 0.1⎞

⎜ ⎟
⎜ 0.2 0.5 0.3 ⎟
⎜ 0 1 ⎟⎠
⎝ 0

States 1 and 2 are transient. The matrix for transitions from transient states
to transient states is.

⎛ 0.6 0.3 ⎞
PT = ⎜ ⎟
⎝ 0.2 0.5 ⎠

⎛ 0.4 −0.3 ⎞ ⎛1 0⎞
I − PT = ⎜ ⎟ where I = ⎜ ⎟ is the identity matrix
⎝ −0.2 0.5 ⎠ ⎝0 1⎠

⎛ 0.5 0.14 ⎞
( )
−1 0.3
S = I − PT = ⎜ 0.2 0.14 ⎟
⎝ 0.14 0.14 ⎠
0.4

In that matrix, the entry sij is the expected time in state j, given that you started in state i. In this
problem, we started in state 1, so the relevant expected times are s11 and s12 .

Total cost = ( 10× expected time in state 1) + (1 × expected time in state 2)

= 10 s11 + s12
⎛ 0.5 ⎞ ⎛ 0.3 ⎞
= (10 ) ⎜ ⎟+⎜ ⎟
⎝ 0.14 ⎠ ⎝ 0.14 ⎠
= 37.84
Question #13
Key: C

Since all claims are for 1000,

claims rate = 1000λ = 800 .
With relative security loading 0.25, premium rate = (800)(1.25) = 1000.

Per period x f ( x) F(x) 1-F(x)

0 0.4493 0.4493 0.5507
1 0.3595 0.8088 0.1912
2 0.1438 0.9526 0.0474

Ruin if 2 or more claims by t = 1 or 3 or more claims by t = 2

2 or more claims by t = 1 is ruin since we must pay 2000 or more when we started with 1000 and
have collected less than 1000 in premium.
3 or more by t = 2 is similar: we can’t pay 3000 or more with the initial 1000 and less than 2000
in premium.
Evaluate the probability of ruin by summing the probabilities of three separate cases:
1. Two or more claims before time 1.
2. No claims by time 1, and three or more between 1 and 2.
3. Exactly one claim by time 1, and two more between 1 and 2.

P(ruin) = 1-F(1) = 0.1912

+ f(0)(1-F(2)) (0.4493)(0.0474) = 0.0213
+ f(1)(1-F(1)) (0.3595)(0.1912) = 0.0687
0.2812
Question #14
Key: E

d = 0.05 → v = 0.095

At issue

( ) ( )
49
A40 = ∑ v k +1 k q40 = 0.02 v1 + ... + v50 = 0.02v 1 − v50 / d = 0.35076
k =0

and a40 = (1 − A40 ) / d = (1 − 0.35076 ) / 0.05 = 12.9848

1000 A40 350.76
so P40 = = = 27.013
a40 12.9848

( )
K ( 40 ) ≥ 10 = 1000 A50 = 549.18 − ( 27.013)( 9.0164 ) = 305.62
Revised Revised
E 10 L − P40a50

where

( ) ( )
24
= ∑ v k +1 k q50
Revised Revised
A50 = 0.04 v1 + ... + v 25 = 0.04v 1 − v 25 / d = 0.54918
k =0

and
Revised
a50 = (1 − A
Revised
50 ) / d = (1 − 0.54918) / 0.05 = 9.0164
Question #15
Key: E

Let NS denote non-smokers and S denote smokers.

The shortest solution is based on the conditional variance formula

(
Var ( X ) = E Var ( X Y ) + Var E ( X Y ) ) ( )
Let Y = 1 if smoker; Y = 0 if non-smoker
1 − AxS
(
E aT Y = 1 = ax =S
) δ
1 − 0.444
=
= 5.56
0.1
1 − 0.286
(
Similarly E aT Y = 0 = )0.1
= 7.14

E ( E ( aT Y ) ) = E ( E ( aT 0 ) ) × Prob ( Y=0 ) + E ( E ( aT 1) ) × Prob ( Y=1)

= ( 7.14 )( 0.70 ) + ( 5.56 )( 0.30 )
= 6.67

( (
E ⎡⎢ E aT Y )) ( ) ( )
2⎤
⎥⎦ = 7.14 ( 0.70 ) + 5.56 ( 0.30 )
2 2
⎣
= 44.96
( (
Var E aT Y ) ) = 44.96 − 6.672 = 0.47
E ( Var ( aT Y ) ) = ( 8.503)( 0.70 ) + ( 8.818 )( 0.30 )
= 8.60
( )
Var aT = 8.60 + 0.47 = 9.07

Alternatively, here is a solution based on

( )
Var(Y ) = E Y 2 − ⎡⎣ E (Y ) ⎤⎦ , a formula for the variance of any random variable. This can be
2

( )
transformed into E Y 2 = Var (Y ) + ⎡⎣ E (Y ) ⎤⎦ which we will use in its conditional form
2

((
E aT )
2
) (
NS = Var aT NS + ⎡ E aT NS ⎤
⎣ ⎦ ) ( )
2

Var ⎡⎣ aT ⎤⎦ = E ⎡ aT ( ) ( )
2⎤ 2
− E ⎡⎣ aT ⎤⎦
⎣⎢ ⎦⎥
E ⎡⎣ aT ⎤⎦ = E ⎡⎣ aT S⎤⎦ × Prob [S] + E ⎡⎣ aT NS⎤⎦ × Prob [ NS]
 = 0.30axS + 0.70axNS

=
(
0.30 1 − AxS ) + 0.70 (1 − A ) NS
x

0.1 0.1
0.30 (1 − 0.444 ) + 0.70 (1 − 0.286 )
= = ( 0.30 )( 5.56 ) + ( 0.70 )( 7.14 )
0.1
= 1.67 + 5.00 = 6.67

( )
E ⎡ aT
2⎤
= E ⎡⎣ aT 2 S⎤⎦ × Prob [S] + E ⎡⎣ aT 2 NS⎤⎦ × Prob [ NS]
⎢⎣ ⎥⎦

( ( ) (
= 0.30 Var aT S + E ⎡⎣ aT S⎤⎦ ⎞⎟ )
2

( (
+0.70 Var aT NS + E aT NS ) ( )
2
)
= 0.30 ⎡8.818 + ( 5.56 ) ⎤ + 0.70 ⎡8.503 + ( 7.14 ) ⎤
2 2
⎣ ⎦ ⎣ ⎦
11.919 + 41.638 = 53.557

Var ⎡⎣ aT ⎤⎦ = 53.557 − ( 6.67 ) = 9.1

2

1 − vT
Alternatively, here is a solution based on aT =
δ
⎛1 v ⎞ T
( )
Var aT = Var ⎜ − ⎟
⎝δ δ ⎠
⎛ − vT ⎞
= Var ⎜ ⎟ since Var ( X + constant ) = Var ( X )
⎝ δ ⎠

=
Var vT ( ) since Var ( constant × X ) = constant 2
× Var ( X )
δ2
Ax − ( Ax )
2 2

= which is Bowers formula 5.2.9

δ2

( )
This could be transformed into 2Ax = δ 2 Var aT + Ax2 , which we will use to get
2
Ax NS and 2AxS .
2
Ax = E ⎡⎣ v 2T ⎤⎦

= E ⎡⎣ v 2T NS⎤⎦ × Prob ( NS) + E ⎡⎣ v 2T S⎤⎦ × Prob ( S )

( ) ( )
= ⎡⎢δ 2 Var aT NS + Ax NS ⎤⎥ × Prob ( NS)
2

⎣ ⎦

( ) ( )
+ ⎡⎢δ 2Var aT S + AxS ⎤⎥ × Prob ( S)
2

⎣ ⎦
= ⎡⎣( 0.01)( 8.503) + 0.2862 ⎤⎦ × 0.70

+ ⎡⎣( 0.01)( 8.818 ) + 0.4442 ⎤⎦ × 0.30

= ( 0.16683)( 0.70 ) + ( 0.28532 )( 0.30 )
= 0.20238

Ax = E ⎡⎣ vT ⎤⎦

= E ⎡⎣ vT NS⎤⎦ × Prob ( NS) + E ⎡⎣ vT S⎤⎦ × Prob ( S)

= ( 0.286 )( 0.70 ) + ( 0.444 )( 0.30 )
= 0.3334
Ax − ( Ax )
2 2

( )
Var aT =
δ2
0.20238 − 0.33342
= = 9.12
0.01
Question #16
Key: E

2 p0 = 0.7
3 p0 = 0.4

Since hyperbolic,
1 1 1
= ( 0.25 ) + ( 0.75 )
s ( 2.25 ) s ( 3) s ( 2)
= ( 0.25 ) / ( 0.4 ) + ( 0.75 ) / ( 0.7 )
= 1.69643
s ( 2.25 ) = 2.25 p0 = 0.58947

Alternatively, since hyperbolic

0.4
p2 = = 0.57143
0.7
p2
0.25 p2 =
1 − (1 − 0.25 ) q2
0.57143
=
1 − ( 0.75 )(1 − 0.57143)
= 0.84211
2.25 p0 = ( 0.7 )( 0.84211) = 0.58948

No matter which way we got 2.25 p0 (ignoring rounding in last digit)

2.25 q0 = 1 − 0.58947 = 0.41053
Probe not transmitting means all three failed
Prob = ( 0.41053) = 0.0692
3
Question #17
Key: A

To be a density function, the integral of f must be 1 (i.e., everyone dies eventually). The solution
is written for the general case, with upper limit ∞ . Given the distribution of f 2 ( t ) , we could
have used upper limit 100 here.

Preliminary calculations from the Illustrative Life Table:

l50
= 0.8951
l0
l40
= 0.9313
l0

∞ ∞
1 = ∫ fT ( t ) dt = ∫ k f1 ( t ) dt + ∫ 1.2 f 2 ( t )dt
50
0 0 50
∞
f1 ( t ) dt + 1.2∫ f 2 ( t )dt
50
= k∫
0 50

= k F1 ( 50 ) + 1.2 ( F2 ( ∞ ) − F2 ( 50 ) )
= k (1 − 50 p0 ) + 1.2 (1 − 0.5 )
= k (1 − 0.8951) + 0.6
1 − 0.6
k= = 3.813
1 − 0.8951

For x ≤ 50, FT ( x ) = ∫ 3.813 f1 ( t ) dt = 3.813F1 ( x )

x
0

⎛ l ⎞
FT ( 40 ) = 3.813 ⎜ 1 − 40 ⎟ = 3.813 (1 − 0.9313) = 0.262
⎝ l0 ⎠
⎛ l ⎞
FT ( 50 ) = 3.813 ⎜ 1 − 50 ⎟ = 3.813 (1 − 0.8951) = 0.400
⎝ l0 ⎠
1 − FT ( 50 ) 1 − 0.400
p40 = = = 0.813
1 − FT ( 40 ) 1 − 0.262
10
Question #18
Key: D

Let NS denote non-smokers, S denote smokers.

Prob (T < t ) = Prob (T < t NS) × Prob ( NS ) + P rob (T < t S) × P rob ( S )

( ) (
= 1 − e −0.1t × 0.7 + 1 − e −0.2t × 0.3 )
= 1 − 0.7e −0.1t − 0.3 e−0.2t

S ( t ) = 0.3e−0.2 t + 0.7e−0.1t
Want tˆ such that 0.75 = 1 − S ( tˆ ) or 0.25 = S ( tˆ )

0.25 = 0.3e−2t + 0.7e −0.1t = 0.3 e−0.1t

ˆ ˆ
( )
ˆ 2
+ 0.7e−0.1t
ˆ

Substitute: let x = e −0.1t

ˆ

0.3 x 2 + 0.7 x − 0.25 = 0

−0.7 ± 0.49 + ( 0.3)( 0.25 ) 4

This is quadratic, so x =
2 ( 0.3)

x = 0.3147

e −0.1t = 0.3147
ˆ
so tˆ = 11.56
Question #19
Key: D

The modified severity, X*, represents the conditional payment amount given that a payment
occurs. Given that a payment is required (X > d), the payment must be uniformly distributed
between 0 and c ⋅ (b − d ) .

The modified frequency, N*, represents the number of losses that result in a payment. The
b−d
deductible eliminates payments for losses below d, so only 1 − Fx ( d ) = of losses will
b
require payments. Therefore, the Poisson parameter for the modified frequency distribution is
b−d
λ⋅ . (Reimbursing c% after the deductible affects only the payment amount and not the
b
frequency of payments).
Question #20
Key: C

Let N = number of sales on that day

S = aggregate prospective loss at issue on those sales
K = curtate future lifetime

N ∼ Poisson ( 0.2*50 ) ⇒ E [ N ] = Var [ N ] = 10

0L = 10,000v K +1 − 500aK +1 ⇒ E [ 0 L ] = 10,000 A65 − 500a65

2
⎛ 500 ⎞ K +1 500 ⎛ 500 ⎞ ⎡ 2
⇒ Var [ 0 L ] = ⎜ 10,000 + ⎟ ⎣ A65 − ( A65 ) ⎦
2⎤
0 L = ⎜ 10,000 + ⎟v −
⎝ d ⎠ d ⎝ d ⎠

S = 0 L1 + 0 L2 + ... + 0 LN
E [ S ] = E [ N ] ⋅ E [ 0 L]

Var [ S ] = Var [ 0 L ] ⋅ E [ N ] + ( E [ 0 L ]) ⋅ Var [ N ]

2

⎛ 0 − E [S ] ⎞
Pr ( S < 0 ) = Pr ⎜ Z < ⎟
⎜
⎝ Var [ S ] ⎟⎠

Substituting d = 0.06/(1+0.06), 2 A65 = 0.23603, A65 = 0.43980 and a65 = 9.8969 yields
E [ 0 L ] = −550.45
Var [ 0 L ] = 15,112,000
E [ S ] = −5504.5
Var [ S ] = 154,150,000

Std Dev (S) = 12,416

⎛ S + 5504.5 5504.5 ⎞
Pr ( S < 0 ) = Pr ⎜ <
⎝ 12,416 12, 416 ⎟⎠
= Pr ( Z < 0.443)
= 0.67

With the answer choices, it was sufficient to recognize that:

0.6554 = φ ( 0.4 ) < φ ( 0.443) < φ ( 0.5 ) = 0.6915
By interpolation, φ ( 0.443) ≈ ( 0.43) φ ( 0.5 ) + ( 0.57 ) φ ( 0.4 )
= ( 0.43)( 0.6915 ) + ( 0.57 )( 0.6554 )
= 0.6709
Question #21
Key: A

A40 161.32
1000 P40 = = = 10.89
a40 14.8166
⎛ a ⎞ ⎛ 11.1454 ⎞
1000 20V40 = 1000 ⎜ 1 − 60 ⎟ = 1000 ⎜ 1 − ⎟ = 247.78
⎝ a40 ⎠ ⎝ 14.8166 ⎠
( 20V + 5000 P40 ) (1 + i) − 5000q60
21V =
P60

=
( 247.78 + (5)(10.89) ) × 1.06 − 5000 ( 0.01376 ) = 255
1 − 0.01376

[Note: For this insurance, 20V = 1000 20V40 because retrospectively, this is identical to whole
life]

Though it would have taken much longer, you can do this as a prospective reserve. The
prospective solution is included for educational purposes, not to suggest it would be suitable
under exam time constraints.

1000 P40 = 10.89 as above

1000 A40 + 4000 20 E40 A60:5
1
= 1000 P40 + 5000 P40 × 20 E40 a60:5 + π 20 E40 × 5 E60 a65
1
where A60:5 = A60 − 5 E60 A65 = 0.06674
a40:20 = a40 − 20 E40 a60 = 11.7612
a60:5 = a60 − 5 E60 a65 = 4.3407
1000 ( 0.16132 ) + ( 4000 )( 0.27414 )( 0.06674 ) =
= (10.89 )(11.7612 ) + ( 5 )(10.89 )( 0.27414 )( 4.3407 ) + π ( 0.27414 )( 0.68756 )( 9.8969 )

161.32 + 73.18 − 128.08 − 64.79

π=
1.86544
= 22.32

Having struggled to solve for π , you could calculate 20 V prospectively then (as above)
calculate 21V recursively.

20V = 4000 A60:5

1
+ 1000 A60 − 5000 P40 a60:5 − π 5 E60 a65
= ( 4000 )( 0.06674 ) + 369.13 − ( 5000 )( 0.01089 )( 4.3407 ) − ( 22.32 )( 0.68756 )( 9.8969 )
= 247.86 (minor rounding difference from 1000 20V40 )
Or we can continue to 21V prospectively

21V = 5000 A61:4

1
+ 1000 4 E61 A65 − 5000 P40 a61:4 − π 4 E61 a65
l65 4 ⎛ 7,533,964 ⎞
where 4 E61 = v =⎜ ⎟ ( 0.79209 ) = 0.73898
l61 ⎝ 8,075, 403 ⎠

1
A61:4 = A61 − 4 E61 A65 = 0.38279 − 0.73898 × 0.43980
= 0.05779
a61:4 = a61 − 4 E61 a65 = 10.9041 − 0.73898 × 9.8969
= 3.5905

21V = ( 5000 )( 0.05779 ) + (1000 )( 0.73898 )( 0.43980 )

− ( 5 )(10.89 )( 3.5905 ) − 22.32 ( 0.73898 )( 9.8969 )
= 255

Finally. A moral victory. Under exam conditions since prospective benefit reserves must equal
retrospective benefit reserves, calculate whichever is simpler.

Question #22
Key: C

Var ( Z ) = 2 A41 − ( A41 )

2

A41 − A40 = 0.00822 = A41 − (vq40 + vp40 A41 )

= A41 − (0.0028 /1.05 + ( 0.9972 /1.05 ) A41 )
⇒ A41 = 0.21650

2
(
A41 − 2 A40 = 0.00433 = 2 A41 − v 2 q40 + v 2 p40 2 A41 )
( )
2
= 2 A41 − (0.0028 /1.052 + 0.9972 /1.052 A41 )
2
A41 = 0.07193

Var ( Z ) = 0.07193 − 0.216502

= 0.02544
Question #23
Key: A

Let L, be the amount by which surplus first drops below 40, given that it does drop
below 40.

∞
PR [ 40 − L1 < 35] = PR [5 < L1 ] = ∫ f L1 ( y )dy
5
∞ 1
=∫ ⎡1 − P ( y ) ⎤⎦ dy
5 p1 ⎣

The last step above is equation 13.5.3 in Bowers (page 416)

For claim size distributed uniformly over (0, 10)

p1 = 5
⎧0 y<0
⎪
and P ( y ) = ⎨ y /10 0 < y < 10
⎪1 y > 10
⎩

2
−1 1
(1 − y /10 ) dy = × × 10 × ⎛⎜ 1 − ⎞⎟
10 1 y 10
PR = ∫ = 0.25
5 5 5 2 ⎝ 10 ⎠ 5
Question #24
Key: D

This solution looks imposing because there is no standard notation. Try to focus on the big
picture ideas rather than starting with the details of the formulas.

Big picture ideas:

1. We can express the present values of the perpetuity recursively.
2. Because the interest rates follow a Markov process, the present value (at time t ) of the
future payments at time t depends only on the state you are in at time t , not how you got
there.
3. Because the interest rates follow a Markov process, the present value of the future
payments at times t1 and t2 are equal if you are in the same state at times t1 and t2 .

Method 1: Attack without considering the special characteristics of this transition matrix.

Let sk = state you are in at time k ( thus sk = 0, 1 or 2 )

Let Yk = present value, at time k , of the future payments.

Yk is a random variable because its value depends on the pattern of discount factors, which are
random. The expected value of Yk is not constant; it depends on what state we are in at time k.

Recursively we can write

Yk = v × (1 + Yk +1 ) , where it would be better to have notation that indicates the v’s are not
constant, but are realizations of a random variable, where the random variable itself has different
distributions depending on what state we’re in. However, that would make the notation so
complex as to mask the simplicity of the relationship.

Every time we are in state 0 we have

(
E ⎡⎣Yk sk = 0 ⎤⎦ = 0.95 × 1 + E ⎡⎣Yk +1 sk = 0 ⎤⎦ )
( {( )) (
= 0.95 × 1 + E ⎡⎣Yk +1 sk +1 = 0 ⎤⎦ × Pr ob s k+1 = 0 sk = 0] )
(
+ E ⎡⎣Yk +1 sk +1 = 1⎤⎦ ) × Pr ob ( sk +1 = 1 s = 0])
k

= ( E ⎡⎣Yk +1 sk +1 = 2 ⎤⎦ ) × Pr ob ( s
k +1 = 2 s = 0])}
k

(
= 0.95 × 1 + E ⎡⎣Yk +1 sk +1 = 1⎤⎦ )
That last step follows because from the transition matrix if we are in state 0, we always move to
state 1 one period later.
Similarly, every time we are in state 2 we have
(
E ⎡⎣Yk sk = 2 ⎤⎦ = 0.93 × 1 + E ⎡⎣Yk +1 sk = 2 ⎤⎦ )
(
= 0.93 × 1 + E ⎡⎣Yk +1 sk +1 = 1⎤⎦ )
That last step follows because from the transition matrix if we are in state 2, we always move to
state 1 one period later.

Finally, every time we are in state 1 we have

(
E ⎡⎣Yk sk = 1⎤⎦ = 0.94 × 1 + E ⎡⎣Yk +1 sk = 1⎤⎦ )
( { })
= 0.94 × 1 + E ⎡⎣Yk +1 sk +1 = 0 ⎤⎦ × Pr ⎡⎣ sk +1 = 0 sk = 1⎤⎦ + E ⎡⎣Yk +1 sk +1 = 2⎤⎦ × Pr ⎡⎣ sk +1 = 2 sk = 1⎤⎦

( { })
= 0.94 × 1 + E ⎡⎣Yk +1 sk +1 = 0 ⎤⎦ × 0.9 + E ⎡⎣Yk +1 sk +1 = 2 ⎤⎦ × 0.1 . Those last two steps follow
from the fact that from state 1 we always go to either state 0 (with probability 0.9) or state 2
(with probability 0.1).

Now let’s write those last three paragraphs using this shorter notation: xn = E ⎡⎣Yk sk = n ⎤⎦ . We
can do this because (big picture idea #3), the conditional expected value is only a function of the
state we are in, not when we are in it or how we got there.

x0 = 0.95 (1 + x1 )
x1 = 0.94 (1 + 0.9 x0 + 0.1x2 )
x2 = 0.93 (1 + x1 )

That’s three equations in three unknowns. Solve (by substituting the first and third into the
second) to get x1 = 16.82 .

That’s the answer to the question, the expected present value of the future payments given in
state 1.

The solution above is almost exactly what we would have to do with any 3 × 3 transition matrix.
As we worked through, we put only the non-zero entries into our formulas. But if for example
the top row of the transition matrix had been ( 0.4 0.5 0.1) , then the first of our three
equations would have become x0 = 0.95 (1 + 0.4 x0 + 0.5 x1 + 0.1x2 ) , similar in structure to our
actual equation for x1 . We would still have ended up with three linear equations in three
unknowns, just more tedious ones to solve.

Method 2: Recognize the patterns of changes for this particular transition matrix.

This particular transition matrix has a recurring pattern that leads to a much quicker solution.
We are starting in state 1 and are guaranteed to be back in state 1 two steps later, with the same
prospective value then as we have now.
Thus,
E [Y ] = E ⎡⎣Y first move is to 0 ⎤⎦ × Pr [ first move is to 0] + E ⎡⎣Y first move is to 2 ⎤⎦ × Pr [first move is to 2 ]

⎣ ( ⎦ ) ⎣ ⎣(
= 0.94 × ⎡ 1 + 0.95 × (1 + E [Y ] ⎤ × 0.9 + ⎡ 0.94 × ⎡ 1 + 0.93 × (1 + E [Y ]) × 0.1⎤
⎦

(Note that the equation above is exactly what you get when you substitute x0 and x2 into the
formula for x1 in Method 1.)

= 1.6497 + 0.8037 E [Y ] + 0.1814 + 0.0874 E [Y ]

1.6497 + 0.1814
E [Y ] =
(1 − 0.8037 − 0.0874 )
= 16.82
Question #25
Key: A

Let state s = number of stocks with market price > strike price.
s = 0,1, 2

Transition probability matrix:

0 1 2
9 6 1
0
16 16 16
1 1 1
1
4 2 4
1 6 9
2
16 16 16

π 0 = limiting probability of 0 stocks with strike price < market price.

π 1 = limiting probability of 1 stock with strike price < market price.
π 2 = limiting probability of 2 stocks with strike price < market price.

9 1 1
π0 = π 0 + π1 + π 2
16 4 16
3 1 3
π1 = π 0 + π1 + π 2
8 2 8
π 0 + π1 + π 2 = 1

Solving algebraically the three equations in three unknowns gives us π 2 = 2 .

7
Question #26
Key: E

The number of problems solved in 10 minutes is Poisson with mean 2.

If she solves exactly one, there is 1/3 probability that it is #3.
If she solves exactly two, there is a 2/3 probability that she solved #3.
If she solves #3 or more, she got #3.

f(0) = 0.1353
f(1) = 0.2707
f(2) = 0.2707

⎛1⎞ ⎛ 2⎞
P = ⎜ ⎟ ( 0.2707 ) + ⎜ ⎟ ( 0.2707 ) + (1 − 0.1353 − 0.2707 − 0.2707 ) = 0.594
⎝ 3⎠ ⎝ 3⎠
Question #27
Key: D

µ x(τ ) = µ x(1) ( t ) + µ x( 2 ) ( t )

= 0.2 µ x( ) ( t ) + µ x( ) ( t )
τ 2

⇒ µ x( ) ( t ) = 0.8µ x( ) ( t )
2 τ

1 k
− ∫ 0.2 k t 2
−0.2
qx( ) = 1 − px( ) = 1 − e 0
dt
= 1− e = 0.04
'1 '1 3

k
3 ⇒ ln (1 − 0.04 ) / ( −0.2 ) = 0.2041
k = 0.6123

( 2)
px( ) µ x( ) dt = 0.8∫ px( ) µ x( ) ( t ) dt
2 τ 2 τ τ
=∫
2
2 qx 0 t 0 t

= 0.8 2 qx( ) = 0.8 1 − 2 px( )

τ τ
( )
− ∫ µ x (t ) d t
2
(τ )
2 px = e
0

= e ∫0
− kt 2
dt

−8 k
= e 3
−( 8 ) ( 0.6123)
= e 3

= 0.19538

( 2)
2 qx = 0.8 (1 − 0.19538 ) = 0.644
Question #28
Key: A

k k ∧3 f(k) f ( k ) × ( k ∧ 3) f ( k ) × (k ∧ 3) 2
0 0 0.1 0 0
1 1 (0.9)(0.2) = 0.18 0.18 0.18
2 2 (0.72)(0.3) = 0.216 0.432 0.864
3+ 3 1-0.1-0.18-0.216 = 0.504 1.512 4.536
2.124 5.580

E ( K ∧ 3) = 2.124

(
E ( K ∧ 3)
2
) = 5.580
Var ( K ∧ 3) = 5.580 − 2.1242 = 1.07

Note that E [ K ∧ 3] is the temporary curtate life expectancy, ex:3 if the life is age x.
Problem 3.17 in Bowers, pages 86 and 87, gives an alternative formula for the variance,
basing the calculation on k px rather than k q x .

Question #29
Key: E

f ( x) = 0.01, 0 ≤ x ≤ 80
= 0.01 − 0.00025( x − 80) = 0.03 − 0.00025 x, 80 < x ≤ 120

( 0.03x − 0.00025x ) dx
80 120
E ( x) = ∫ 0.01x dx + ∫ 2
0 80

0.01x 2 80 0.03 x 2 120 0.00025 x3 120

= + −
2 0 2 80 3 80

= 32 + 120 − 101.33 = 50.66667

E ( X − 20) + = E ( X ) − ∫ x f ( x ) dx − 20(1 − ∫ f ( x ) dx)
20 20
0 0

= 50.6667 −
0.01x 2 20
2 0
(
− 20 1 − 0.01x
20
0
)
= 50.6667 − 2 − 20(0.8) = 32.6667

32.6667
Loss Elimination Ratio = 1 − = 0.3553
50.6667
Question #30
Key: D

Let q64 for Michel equal the standard q64 plus c. We need to solve for c.
Recursion formula for a standard insurance:

20V45 = ( 19V45 + P45 ) (1.03) − q64 (1 − 20V45 )

Recursion formula for Michel’s insurance

20V45 = ( 19V45 + P45 + 0.01) (1.03) − ( q64 + c ) (1 − 20V45 )

The values of 19 V45 and 20V45 are the same in the two equations because we are told
Michel’s benefit reserves are the same as for a standard insurance.

Subtract the second equation from the first to get:

0 = − (1.03) (0.01) + c(1 − 20V45 )

(1.03) ( 0.01)
c=
(1 − 20V45 )
0.0103
=
1 − 0.427
= 0.018
Question #31
Key: B

K is the curtate future lifetime for one insured.

L is the loss random variable for one insurance.
LAGG is the aggregate loss random variables for the individual insurances.
σ AGG is the standard deviation of LAGG .
M is the number of policies.

⎛ π⎞
L = v K +1 − π aK +1 = ⎜ 1 + ⎟ v K +1 − π
⎝ d⎠ d

E [ L ] = ( Ax − π ax ) = Ax − π
(1 − Ax )
d
⎛ 0.75095 ⎞
= 0.24905 − 0.025 ⎜ ⎟ = −0.082618
⎝ 0.056604 ⎠
⎛ π⎞
( )
2 2
Var [ L ] = ⎜ 1 + ⎟
⎝ d⎠
( 2
Ax − Ax2 ) ⎛
= ⎜1 +
0.025 ⎞
⎟ 0.09476 − ( 0.24905 ) = 0.068034
⎝ 0.056604 ⎠
2

E [ LAGG ] = M E [ L ] = −0.082618M
Var [ LAGG ] = M Var [ L ] = M (0.068034) ⇒ σ AGG = 0.260833 M
⎡L − E [ LAGG ] − E ( LAGG ) ⎤
Pr [ LAGG > 0] = ⎢ AGG > ⎥
⎣ σ AGG σ AGG ⎦
⎛ 0.082618M ⎞
≈ Pr ⎜ N (0,1) > ⎟
⎜ M ( 0.260833) ⎟⎠
⎝
0.082618 M
⇒ 1.645 =
0.260833
⇒ M = 26.97

⇒ minimum number needed = 27

Question #32
Key: D

1 − v K +1
Annuity benefit: Z1 = 12,000 for K = 0,1, 2,...
d
Death benefit: Z 2 = Bv K +1 for K = 0,1, 2,...
1 − v K +1
New benefit: Z = Z1 + Z 2 = 12,000 + Bv K +1
d
12,000 ⎛ 12,000 ⎞ K +1
= +⎜B− ⎟v
d ⎝ d ⎠

2
⎛
Var( Z ) = ⎜ B −
⎝
12,000 ⎞
d
⎟ Var v
⎠
(K +1
)
12,000
Var ( Z ) = 0 if B = = 150,000 .
0.08

In the first formula for Var ( Z ) , we used the formula, valid for any constants a and b and
random variable X,

Var ( a + bX ) = b 2Var ( X )
Question #33
Key: B

First restate the table to be CAC’s cost, after the 10% payment by the auto owner:

Towing Cost, x p(x)

72 50%
90 40%
144 10%

Then E ( X ) = 0.5*72 + 0.4*90 + 0.1*144 = 86.4

( )
E X 2 = 0.5*722 + 0.4*902 + 0.1*1442 = 7905.6
Var ( X ) = 7905.6 − 86.42 = 440.64
Because Poisson, E ( N ) = Var ( N ) = 1000
E ( S ) = E ( X ) E ( N ) = 86.4*1000 = 86, 400
Var( S ) = E ( N )Var( X ) + E ( X )2 Var( N ) = 1000* 440.64 + 86.42 *1000 = 7,905,600
⎛ S − E ( S ) 90,000 − 86, 400 ⎞
Pr( S > 90,000) + Pr ⎜ > ⎟⎟ = Pr( Z > 1.28) = 1 − Φ (1.28) = 0.10
⎜ Var( S )
⎝ 7,905,600 ⎠

Since the frequency is Poisson, you could also have used

( )
Var ( S ) = λ E X 2 = (1000 )( 7905.6 ) = 7,905,600
That way, you would not need to have calculated Var ( X ) .
Question #34
Key: C

LER =
E ( X ∧ d ) θ 1− e
=
(
− d /θ
)
= 1 − e− d /θ
E(X ) θ
Last year 0.70 = 1 − e− d /θ ⇒ − d = θ log 0.30
Next year: − d new = θ log(1 − LER new )
4
Hence θ log (1 − LER new ) = − d new = θ log 0.30
3
log (1 − LER new ) = −1.6053
(1 − LER new ) = e−1.6053 = 0.20
LER new = 0.80

Question #35
Key: E

E ( X ) = e(d ) S (d ) + E ( X ∧ d ) [Klugman Study Note, formula 3.10]

62 = e 40 × 40 p0 + E (T ∧ 40 )

62 = (e 40 )(0.6) + 40 − (0.005)(402 )

= 0.6 e 40 + 32

e40 =
( 62 − 32 ) = 50
0.6

The first equation, in the notation of Bowers, is e0 = e40 × 40 p0 + e0:40 . The corresponding
formula, with i > 0 , is a very commonly used one:

ax = ax:n + n Ex ax + n
Question #36
Key: C

n
ex:n = ∫ px dt
0 t

q90 = 0.18877

For U t px = 1 − t ⋅ qx so e 90:1 = 1 − 1 2 ( 0.18877 ) = 0.9056

q90 0.18877
px = ( px )
t
For C so e90:1 = = = 0.9023
− log p90 − log ( 0.81123)
t

px p90 0.81123
For H px = so e 90:1 = ( − log p90 ) = ( − log 0.81123) = 0.8990
px + tqx
t
q90 0.18877

Alternatively

qx
For U, µ ( x + z ) = is increasing for 0 < t < 1
1 − tqx
For C, µ ( x + t ) = − log px is constant for 0 < t < 1
qx
For H, µ ( x + t ) = is decreasing for 0 < t < 1
1 − (1 − t ) qx

With 1 px the same for all three, we must have t pUx > t pxC > t pxH for 0<t<1
1 1 1
∫0 dt > ∫ dt > ∫
U
and t p90 pC H
p90 dt
0 t 90 0 t

Alternatively (not rigorously)

0.5
U
p90 = 0.9056
0.5
C
p90 = 0.9007
0.5
H
p90 = 0.8958

1 1 1
∫0 dt > ∫ dt > ∫
U
That should strongly suggest that t p90 pC pH dt
0 t 90 0 t 90
You could compare additional values of p for greater comfort.
Question #37
Key: B

d = 0.05 ⇒ v = 0.95

Step 1 Determine px from Kevin’s work:

608 + 350vpx = 1000vqx + 1000v 2 px ( px +1 + qx +1 )
608 + 350 ( 0.95 ) px = 1000 ( 0.95 ) (1 − px ) + 1000 ( 0.9025 ) px (1)
608 + 332.5 px = 950 (1 − px ) + 902.5 px
px = 342 / 380 = 0.9

Step 2 Calculate 1000 Px:2 , as Kira did:

608 + 350 ( 0.95 )( 0.9 ) = 1000 Px:2 ⎡⎣1 + ( 0.95 )( 0.9 ) ⎤⎦

1000 Px:2 =
[ 299.25 + 608] = 489.08
1.855

The first line of Kira’s solution is that the actuarial present value of Kevin’s benefit premiums is
equal to the actuarial present value of Kira’s, since each must equal the actuarial present value of
benefits. The actuarial present value of benefits would also have been easy to calculate as
( )
(1000 )( 0.95)( 0.1) + (1000 ) 0.952 ( 0.9 ) = 907.25
Question #38
Key: E

Because no premiums are paid after year 10 for (x), 11Vx = Ax +11

Rearranging 8.3.10 from Bowers, we get =

( hV + π h ) (1 + i ) − bh+1qx+ h
h +1V
px + h

=
( 32,535 + 2,078) × (1.05) − 100,000 × 0.011 = 35,635.642
10V
0.989
( 35,635.642 + 0 ) × (1.05) − 100,000 × 0.012 = 36,657.31 = A
11V = x +11
0.988

Question #39
Key: B

⎛ x⎞
For De Moivre’s law where s ( x ) = ⎜ 1 − ⎟ :
⎝ ω⎠

° ω−x ⎛ t ⎞
ex = and t px = ⎜ 1 − ⎟
2 ⎝ ω −x⎠
° 105 − 45
e 45 = = 30
2
105 − 65
e°65 = = 20
2

° 40 40 60 − t 40 − t
e 45:65 = ∫ t p45:65dt = ∫ × dt
0 0 60 40

1 FG 60 + 40 2 1 3 IJ 40
=
60 × 40 H
60 × 40 × t −
2
t + t
3 K 0

= 1556
.

e 45:65 = e 45 + e 65 − e 45:65
= 30 + 20 − 1556 . = 34

In the integral for e45:65 , the upper limit is 40 since 65 (and thus the joint status also)
can survive a maximum of 40 years.
Question #40
Key: B

F(0) = 0.8
F(t) = 0.8 + 0.00025(t-1000), 1000 ≤ t ≤ 5000

0.75 ⇒ 0 found since F ( 0 ) ≥ 0.75

0.85 ⇒ 2000 found since F ( 2000 ) = 0.85
Average of those two outcomes is 1000.