ACTL3151 UNSW Tutorial Exercise

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 124

Contents

Module 1: Life Insurance Benefits 2


Fixed term and whole life insurance benefits: the continuous case . . . . . . . . . . . . . . . . . . . 2
Endowment insurance and deferred insurance benefits: the continuous case . . . . . . . . . . . . . 2
Insurance benefits payable at the end of the year of death (the annual case) . . . . . . . . . . . . . 3
Life insurance benefits: 1/m cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Life insurance relationships between the continuous and discrete cases . . . . . . . . . . . . . . . . 4
Solutions to Module 1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Module 2: Annuities 14
Whole life annuity-due: the annual case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Term annuity-due and two other types of annuity due . . . . . . . . . . . . . . . . . . . . . . . . . 14
Immediate annuities: The annual case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Annuities payable 1/m-thly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Annuities payable continuously . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Increasing annuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Approximations for annuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Solutions to Module 2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Module 3: Net Premium Valuation 25


Net premium and the equivalence principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Fully discrete annual premiums - WLI and term insurance . . . . . . . . . . . . . . . . . . . . . . . 25
Fully discrete annual premiums of some other types of insurance contracts . . . . . . . . . . . . . . 26
Fully continuous premium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Premiums paid m times a year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Solutions to Module 3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Module 4: Net Premium Reserves 33


Net premium reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Retrospective reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Recursive calculation of reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Death strain at risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Thiele’s differential equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Participating life insurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Solutions to Module 4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Module 5: Gross Premiums and Reserves 50


Gross premium and reserve calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Zillmerized reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Surrender values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
The portfolio percentile premium principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Solutions to Module 5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

1
Module 6: Profit Testing 64
Cash-flow projection of conventional products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Cash-flow projection for unit-linked products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Profit measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Zeroisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Universal life insurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Solutions to Module 6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Module 7: Multiple Decrement Tables 78


Multiple decrement tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
The associated single decrement tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Construction of MDT from the ASDTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Statistical treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Solutions to Module 7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

Module 8: Multiple State Models 87


Markov chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
EPV of cashflows under multiple state models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Multiple state model and multiple decrement table . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Solutions to Module 8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Module 9: Insurance and Annuities for Multiple Lives 102


Joint distribution of future lifetimes, the joint life status and benefits . . . . . . . . . . . . . . . . . 102
The last-survivor status and benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Joint life functions under some special laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Simple contingent functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Reversionary annuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Solutions to Module 9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

Module 10: Pension Mathematics 112


Expected value of lump sum and pension benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Salary scales, death benefits and age retirement benefits . . . . . . . . . . . . . . . . . . . . . . . . 112
Salary related ill-health retirement benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Solutions to Module 10 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

2
University of New South Wales

School of Risk and Actuarial Studies

ACTL3151/ACTL5105
Tutorial Problems
Module 1: Life Insurance Benefits

Fixed term and whole life insurance benefits: the continuous case
1. [Solution] Calculate Ā45 , the expected present value of a whole of life insurance on (45) of $1 payable
at the moment of death. Assume the force of interest is constant and equal to δ = 0.03 and the force
of mortality is equal to µ(t) = 0.005. Also calculate the variance of this benefit.

2. [Solution] For a whole life insurance of 1000 on (x) with benefits payable at the moment of death, you
are given: 
0.04, 0 < t ≤ 10
δt =
0.05, t > 10
and 
0.06, 0 < t ≤ 10
µx+t = .
0.07, t > 10
Calculate the expected present value of this insurance.

1
3. [Solution] Calculate the exact value of Ā70:1 , assuming the force of mortality is constant between
consecutive integer ages. You are given the following basis:

ˆ Mortality: ELT15 (Males)


ˆ Interest: 7.5% per annum

4. Dickson et al. (2009a) Exercise 4.9 (Dickson et al., 2013a, Exercise 4.11)

5. [Solution] Each of 100 independent lives purchases a single premium 5-year deferred whole life insurance
of 10 payable at the moment of death. Using the Normal approximation, calculate F such that the
probability the insurer has sufficient funds to pay all claims is 0.95. You are given:

ˆ µ = 0.004;
ˆ δ = 0.006;
ˆ F is the aggregate amount the insurer receives from the 100 lives; and
ˆ the 95th percentile of the standard Normal distribution is 1.645.

Endowment insurance and deferred insurance benefits: the contin-


uous case
6. [Solution] Suppose we have an insurance policy with sum assured $10,000 payable at the moment of
death, provided the insured (currently aged 35) survives the first fifteen years. Calculate the expected
present value of this benefit.
Basis:

ˆ δ(t) = 0.02

2

0.003, 0 ≤ t < 15

ˆ µ(x + t) = 0.005, t ≥ 15

7. [Solution] Calculate the expected value and the variance of a 15 year term Endowment Insurance policy
issued at age 50. A death benefit of 100, 000 is payable immediately upon death, and 50, 000 is payable
on survival to the end of the policy term. Assume µx+t = 0.04 is constant and the interest rate is
i = 6% per annum.
8. Dickson et al. (2009a) Exercise 4.12 (Dickson et al., 2013a, Exercise 4.14)

Insurance benefits payable at the end of the year of death (the


annual case)
9. [Solution] Calculate the following using the AM92 table with 4% interest:

(a) 15 E20
(b) A120:15
(c) 2 A120:15
(d) (IA)[20]:15
1
10. [Solution] Given that Ax = 0.2, Ax+25 = 0.3, Ax:25 = 0.6 and i = 0.06, calculate Ax:25 and 25| Ax .

11. [Solution] Suppose we have an insurance policy on (35) providing sum assured $500 if the insured dies
in the first year, $480 if they die in the second year, with the benefit decreasing by $20 each year. All
benefits are payable at the end of the year of death. Mortality follows the AM92 Ultimate table with
6% interest. Calculate the expected present value of these benefits.
12. [Solution] For a whole life insurance of 1 on (41) with death benefit payable at the end of the year of
death, let Z be the present value random variable for this insurance. You are given: (i) i = 0.05; (ii)
p40 = 0.9972; (iii) A41 − A40 = 0.00822; and (iv) 2 A41 −2 A40 = 0.00433. Calculate Var(Z).

13. Dickson et al. (2009a) Exercise 4.3 (Dickson et al., 2013a, Exercise 4.4)
14. Dickson et al. (2009a) Exercise 4.4 (Dickson et al., 2013a, Exercise 4.5)

Life insurance benefits: 1/m cases


15. Dickson et al. (2009a) Exercise 4.2 (Dickson et al., 2013a, Exercise 4.3)
16. Dickson et al. (2009a) Exercise 4.6 (Dickson et al., 2013a, Exercise 4.7)
17. Dickson et al. (2009a) Exercise 4.17 (Dickson et al., 2013a, Exercise 4.27)
Remark: (1) Sometimes the exercise question does not specify the interest rate and the life table. Note
that it has been stated in the beginning of the textbook exercises, quoted, “unless otherwise stated you
should assume the mortality follows the Standard Select Survival Model as specified in Section 6.3, that
interest is at 5% per year effective, and the equivalence principle is used for the calculation of premi-
ums.” You can find the values for the corresponding actuarial symbols in Section 6.3. Alternatively,
you can also find the Standard Ultimate Survival Model here: click this link.
(2) In the solution to part (b), we need to calculate the second moment of the present value variable
which involves the 1/m life insurance and term life insurance with double the force of mortality. We
can still apply the UDD assumption to these values. For example, we still have

2 (4) i0 2
A50 = A50 ,
i0m

3
where i0 and i0m are the corresponding annual interest rate and 1/m nominal interest rate calculated
with double the force of mortality, i.e.,
m
i0

2
1+ m = 1 + i0 = e2δ = (1 + i) .
m

The 1/m term life insurance with double the force of mortality value can be calculated in the same way.
18. [Solution] Calculate (I¯Ā)x (the expected present value of a whole life assurance issued to a life aged 20
exact payable immediately on death where the benefit paid on death at time t is t) using the following
basis: Basis: Mortality µ(x) = 0.03 for x < 40 inclusive and 0.04 for x ≥ 40. Force of interest 5% p.a.

Life insurance relationships between the continuous and discrete


cases
19. [Solution] Assume that deaths are uniformly distributed between integer ages.
(i) By considering a term insurance policy as a series of one year deferred term insurance policies, show
that:
1 i 1
Āx:n = Ax:n .
δ
(ii) Calculate the expected present value and the variance of the present value of a term insurance of 1
payable immediately on death for a life aged 40 exact, if death occurs within 30 years. You are given
the following basis:

ˆ Interest: 4% per annum


ˆ Mortality: AM92 Select
ˆ No expenses.

4
Solutions to Module 1 Exercises
 R   R 
t t
1. [Question] For a constant force of mortality, we have t px = exp − 0 µ(s)ds = exp − 0 0.005ds =
e−0.005t .
Z ∞
Ā45 = v t t px µx+t dt
Z0 ∞
= e−0.03t e−0.005t × 0.005dt
0
 −0.035 ∞
e
= 0.005
0.035 0
1
=
Z7 ∞
E[Z 2 ] = v 2t t px µx+t dt
0
Z ∞ 2
= e−0.03t e−0.005t × 0.005dt
0
 −0.065 ∞
e
= 0.005
0.065 0
1
=
13
 2
1 1
Var(Z) = −
13 7
= 0.0565149

2. [Question] Define the present value random variable Z such that


(
1000e−0.04T 0 < T ≤ 10
Z=
1000e−0.04×10−0.05×(T −10) = e0.1−0.05T T > 10

where T is the future lifetime of (x). We can also calculate the survival probability as
(
− 0t µx+s ds
R e−0.06t 0 < T ≤ 10
t px = e = −0.06∗10−(t−10)∗0.07 0.1−0.07t
.
e =e T > 10

We may write the actuarial present value of this as


Z 10 Z ∞
E (Z) = 1000 e−0.04t t px µx+t dt + 1000 e0.1−0.05t t px µx+t dt
t=0 10
 Z 10
= 1000 e−0.04t × e−0.06t 0.06 dt
0
Z ∞ 
+ e0.1−0.05t × e0.1−0.07t × 0.07dt
10
 Z ∞ 
0.06 
× 1 − e−1 + 0.07 × e−0.12t+0.2 dt

= 1000
0.1 10
 
0.06  0.07
× 1 − e−1 + × e0.2 e−1.2 − 0
  
= 1000
0.1 0.12
= 593.87.

3. [Question] We know that in the general case, we compute


Z 1
1
Ā70:1 = e−δ.075 t t p70 µ70+t dt
0

5
Assuming µ is constant for 0 < t < 1, we have for 0 ≤ s ≤ 1
Rs
s p70 = e− 0
µds

and in particular when s = 1,


R1
p70 = e− 0
µds
= e−µ =⇒ µ = − log (p70 )
l71
The probability p70 can then be calculated using the life-tables; p70 = l70 . Therefore, we find that
µ = − log(p70 ) = − log(1 − 0.03930) = 0.040093
and
t p70 = exp (−µt) = exp (−0.040093t)
and
δ.075 = log(1.075) = 0.07232.
Therefore, we have
Z 1
1
Ā70:1 = e−0.07232t e−0.040093t (0.040093)dt
0
= 0.0379.

4. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual


5. [Question] Define the aggregate claim random variable as
100
X
S = 10S1 + 10S2 + ...... + 10S100 = 10 Si
i=1

where Si is the present value of the claim payout of $1 for the ith policyholder, and is defined such
that (where Ti is the exact future lifetime for the ith policy holder)

0, 0 ≤ Ti < 5
Si = .
v Ti , Ti ≥ 5
We wish to find the capital reserve with a 95% probability that this will be sufficient to pay out claims.
This will be the 95% quantile value of the of S, and will be denoted F . Hence, the equation to solve is
! !
S − E(S) F − E(S) F − E(S)
0.95 = Pr (S ≤ F ) = Pr p ≤ p ≈ Pr Z ≤ p
Var(S) Var(S) Var(S)
where Z ∼ Normal (0, 1) ,which is an approximation coming from the central limit theorem. Thus, the
capital reserve simplifies to
F − E(S) p
1.645 = p ⇐⇒ F = 1.645 Var(S) + E(S),
Var(S)
and hence, naturally we must obtain the expectation and variance of the aggregate claims random
variable. Consider first the expectation
100
!
X
E(S) = E 10 Si = 1000E (Si ),
i=1

as Si are iid. Now, we have


Z ∞ Z ∞
E(Si ) = v t fTi (t)dt = e−δt t px µx+t dt
Zt=5

t=5
Z ∞
−0.006t −0.004t
= e ×e 0.004dt = 0.004 e−0.01t dt
t=5 t=5
0.004 −0.05 
= e = 0.380492.
0.01

6
Hence E(S) = 1000 × 0.380492 = 380.492.
100
!
X
Var(S) = Var 10 Si = 10 000Var(Si )
i=1

2
where we use Var(Si ) =E(Si2 ) − (E(Si )) . The second moment of Si is
Z ∞ Z ∞
t 2
2
e−2δt t px µx+t dt

E(Si ) = v fTi (t)dt =
Zt=5

t=5
Z ∞
−0.012t −0.004t
= e ×e 0.004dt = 0.004 e−0.016t dt
t=5 t=5
0.004 −0.08 
= e = 0.230 78.
0.016
2
So, V ar(Si ) = 0.230 78 − (0.380492) = 0.08600 6 and therefore, Var(S) = 10000Var(Si ) = 860.06. The
capital required to ensure that we will be able to pay claims on average 95% of the time is

F = 1.645 × 860.06 + 380.492 = 428.73.

6. [Question] The expected present value is given by 1000015| Ā35 .

15| Ā35 = 15 E35 Ā50


Z ∞
= v 15 15 p35 v t t p50 µ50+t dt
0
Z ∞
−0.02×15 −0.003×15
=e e e−0.02t e−0.005t × 0.005dt
0
 −0.025t ∞
−0.345 e
=e × 0.005
−0.025 0
0.005e−0.345
=
0.025
= 0.1416

Thus the expected present value is $1,416.44.


1 1
7. [Question] Expected value of this endowment insurance policy is 100, 000Ā50:15 + 50, 000A50:15 with
the force of interest δ = log(1.06).
Z 15 Z 15
1
Ā50:15 = v t µx+tt px dt = e−δt e−µt µdt
0 0
Z 15
µ  
=µ e−(δ+µ dt = 1 − e−(δ+µ)15
0 δ+µ
0.04  
= 1 − e−(log(1.06)+0.04))∗15
log(1.06) + 0.04
=0.31383

1
A50:15 =v 15 1 5p50 = e−δ∗15 e−µ∗15
 15
1
= e−0.04∗15 = 0.22900
1.06

The expected value is

100, 000 ∗ (0.31383) + 50, 000 ∗ (0.22900) = 42, 833.27

7
We have
Z 15 Z 15
2 1
Ā50:15 = 2t
v µx+tt px dt = e−2δt e−µt µdt
0 0
Z 15
µ  
=µ e−(2δ+µ) dt = 1 − e−(2δ+µ)15
0 2δ + µ
0.04  
= 1 − e−(2 log(1.06)+0.04)∗15 = 0.23111
2 log(1.06) + 0.04

and
2 1
A50:15 =v 2∗15 1 5p50 = e−2δ∗15 e−µ∗15
 30
1
= e−0.04∗15 = 0.09555
1.06

hence the variance is

100, 0002 (0.2311) + 50, 0002 (0.09555) − (42, 833.27)2 = (26, 745.4743)2

8. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual


9. [Question] Using the AM92 table at 4% interest, we have:

(a)

15 E20 = 1.04−15 15 p20


9894.4299
= 1.04−15 ×
9982.2006
= 0.55038

Alternatively, we may use the commutation functions:


D35
15 E20 =
D20
2507.4
=
4555.75
= 0.55038

(b)

A120:15 = A20 − 15 E20 A35


= 0.11226 − 0.55038 × 0.19219
= 0.0064822

(c)
2
2
A120:15 = 2 A20 − v 15 15 p20
2
A35
2 9894.4299
= 0.01982 − 1.04−15 × × 0.04874
9982.2006
= 0.0049247

8
(d) We will firstly calculate the endowment assurance function here, 15 E[20] as a useful intermediary
calculation.

D35
15 E[20] =
D[20]
2507.4
=
4554.85
= 0.55049
(IA)[20]:15 = 1515 E[20] + (IA)1[20]:15 (∗)
= 1515 E[20] + (IA)[20] − 15 E[20] (15A35 + (IA)35 ) (∗∗)
= 15 × 0.55049 + 5.74637 − 0.55049 (15 × 0.19219 + 7.46909)
= 8.30509

In line (∗), we decompose the increasing endowment into a pure endowment of $15 and an in-
creasing 15-year term assurance. In line (∗∗) we further decompose the increasing 15-year term
assurance into a whole of life assurance to [20] and deduct all cash flows occurring after the 15-year
term. The aforementioned cash flows may be visualised in the chart below.

The chart depicts the cash flows of (IA)[20] . After removing the maroon cash flows (representing
15A35 ) and the orange cash flows (representing (IA)35 ), we are left with only the blue cash flows
(representing (IA)1[20]:15 ).

9
10. [Question]

Ax = A1x:25 + 25 Ex Ax+25
 
0.2 = Ax:25 − 25 Ex + 25 Ex Ax+25
= Ax:25 − 25 Ex (1 − Ax+25 )
= 0.6 − 25 Ex (1 − 0.3)
0.6 − 0.2
25 Ex =
1 − 0.3
4
=
7
25| Ax = 25 Ex Ax+25
4
= × 0.3
7
6
=
35

11. [Question] We have a benefit of $500 decreasing at a rate of $20 per year. This is a decreasing assurance,
but we may calculate it using increasing assurances. We can write it as a level term assurance of $520,
less an increasing term assurance (that increases at $20 per year).

20(DA)135:25 = 520A135:25 − 20(IA)135:25


 
= 520 A35:25 − 25 E35 − 20 ((IA)35 − 25 E35 (25A60 + (IA)60 ))
= 520 (0.24208 − 0.2187) − 20 (3.33702 − 0.2187 (25 × 0.32692 + 5.46572))
= 5.07304

Note that we calculated 25 E35 = v 25 25 p35 = 1.06−25 × 9287.2164


9894.4289 = 0.2187 as an intermediary calculation.

12. [Question] V ar(Z) = 2 A41 − (A41 )2 .


Note that
A40 = vq40 + vp40 A41 ,
so that this implies that
0.9972 0.0028
A41 − A40 = − .
1.05 1.05
Combining this with the given A41 − A40 = 0.00822 leads us to
(0.00822 + (0.0028/1.05))
A41 = = 0.21649621.
(1 − (0.9972/1.05))
On a similar note, we have
2
A40 = v 2 q40 + v 2 p40 2 A41 ,
so that we have
0.9972 2 0.0028
A41 − 2 A40 = − .
1.052 1.052
Combining this with the given 2 A41 − 2 A40 = 0.00433 leads us to

2 0.00433 + (0.0028/1.052 )
A41 = = 0.07192616.
(1 − (0.9972/1.052 ))
Finally, we have the variance:

Var(Z) = 0.07192616 − (0.21649621)2 = 0.02505555.

10
13. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual
14. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual
15. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual
16. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual
17. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual
For part (b), the answers in the textbook are incorrect.The following is the correct answer.
2 (4) By UDD i@2δ
A(4)
x = Ax@2δ = (4)
× Ax@2δ (1.1)
i@2δ
i@2δ = e2δ − 1 = (1.05)2 − 1 = 0.1025 (1.2)
(4) 2δ 1
i@2δ = 4(e 4 − 1) = 4((1.05) − 1) = 0.098780
2 (1.3)
By (1), (2) and (3) we obtain

2
A(4) 2
x = 1.037659 × Ax@2δ = 1.037659 × Ax (1.4)

2 (4)1 (4)1
A 50:15
= A 50:15 @2δ
(4) 15 (4)
= A50@2δ − v@2δ × 15 p50 × A65@2δ
 15
2 (4) 1 l65 2 (4)
= A50 − 2
× × A65
1.05 l50
 
By (4) 2 −30 94579.73 2
= 1.037659 A50 − 1.05 × × A65
98576.37
= 1.037659(0.05108 − 0.221997 × 0.15420)
= 0.017483 (1.5)

2 (4) (4)
15| A50 = 15| A50@2δ
15 (4)
= v@2δ × 15 p50 × A65@2δ
94579.73 2 (4)
= 1.05−30 × × A65
98576.37
= 0.035521 (1.6)

   
(4)1 (4)
E[Z 2 ] = 20002 2
A 50:15
+ 10002 2
15| A50

By (5) and (6)


= 20002 (0.017483) + 10002 (0.035521)
= 105453

18. [Question]
Z ∞
(I¯Ā)x = tv t t px µx+t dt
0
Z 20 Z ∞
= te−0.05t e−0.03t 0.03dt + e−20∗0.03 te−0.05t e−0.04(t−20) 0.04dt
0 20
Z 20 Z ∞
= 0.03 te−0.08t dt + e−20∗0.03 e0.04∗20 te−0.05t e−0.04t 0.04dt
0 20
Z 20 Z ∞
−0.08t 0.2 −0.09t
= 0.03 te dt + 0.04e te dt
0 20

11
(
20 Z 20 )
te−0.08t 1 −0.08t
= 0.03 + e dt
0.08 0 0.08 0
 −0.09t ∞ Z ∞ 
te 1
+0.04e0.2 + e−0.09t dt
0.09 20 0.09 20
( )
−1.6
20e 1  −1.6 
= 0.03
0.08
+0+ 2 −e +1
(0.08)
( )
0.2 20e−1.8 1  −1.8

+0.04e −0 + + 2 −0 + e
0.09 (0.09)
= 0.03 × 74.23 + 0.04 × 1.2214 × 57.140
= 5.019

19. [Question] (i) First, we write the n-year term insurance as


Z n
1
Āx:n = v t t px µx+t dt
0
Z 1 Z 2 Z n
t t
= v t px µx+t dt + v t px µx+t dt + · · · + v t t px µx+t dt
0 1 n−1
Z 1 Z 1
= v t t px µx+t dt + vpx v t t px+1 µx+1+t dt + · · ·
0 0
Z 1
+v n−1 n−1 px v t t px+n−1 µx+n−1+t dt
0
n−1
X Z 1
k
= v k px v t t px+k µx+k+t dt
k=0 0
n−1
X n−1
X
k 1 1
= v k px Āx+k:1 = k| Āx:1
k=0 k=0

Thus, the n-year term insurance can be considered as the sum of a series of one year deferred term
insurance policies.
Assuming a uniform distribution of deaths between integer ages, we would have
fTx+k (t) =t px+k µx+k+t = qx+k for integer x, k, and 0 < t < 1,
so that
1 1
1 − e−δ 1−v
Z Z
1 t d iv
Āx+k:1 = v fTx+k (t)dt = qx+k v t dt = qx+k = qx+k = qx+k = qx+k · for integer k.
0 0 δ δ δ δ
Thus,
n−1 n−1 n−1
1
X X iv i X k+1 i 1
Āx:n = v k k px Āx+k:1
1
= v k k px qx+k · = v k px qx+k = Ax:n
δ δ δ
k=0 k=0 k=0
(ii) For a term insurance of 1 payable immediately on death to (40) using AM92 Select rates, if death
occurs within 30 years, the actuarial present value is calculated as
1 i 1 0.04
Ā[40]:30 = A[40]:30 = A 1
δ ln(1.04) [40]:30
0.04
= 0.078970 = 0.080539,
ln(1.04)
where
1
A[40]:30 = A[40] − v 30 30 p[40] A70
8054.0544
= 0.23041 − v 30 0.60097 = 0.078970.
9854.3036

12
Now to compute the variance, we note that
2
Var Z̄ = 2 Āx:n
1 1

− Āx:n

where Z̄ is the continuous present value random variable for the term insurance.
Notice
Z n
2 1
Āx:n = (v t )2 t px µx+t dt
0
n−1
X Z 1
2k
= v k px v 2t t px+k µx+k+t dt
k=0 0
n−1
X Z 1
= v 2k k px v 2t qx+k dt under UDD assumption
k=0 0

1
1 − e−2δ 1 − v2 v 2 (i2 + 2i)
Z
v 2t dt = = =
0 2δ 2δ 2δ
Hence,
n−1 n−1
2 1
X
2k v 2 (i2 + 2i) (i2 + 2i) X 2(k+1) (i2 + 2i) 2 1 
Āx:n = v k px qx+k = v k px qx+k = Ax:n|
2δ 2δ 2δ
k=0 k=0

Note that this formula can also be obtained by directly applying the result in (i) with twice the force
of interest. Consequently,
 2 
i2 + 2i
  2
2
 i
A1x:n| A1x:n| .

Var Z̄ = −
2δ δ

Note
2
A1[40]:30 = 2
A[40] − v 30∗2 30 p[40] 2 A70
8054.0544
= 0.06775 − v 30∗2 × × 0.38975 = 0.037469,
9854.3036
where here, discount is at 2δ so that v = 1/1.0816. Therefore,
2
0.042 + 2(0.04)

0.04
0.0789702 = 0.032491.

Var Z̄ = 0.037469 −
2(0.039221) ln(1.04)

13
Module 2: Annuities

Whole life annuity-due: the annual case


1. [Solution] John (currently aged 52 exact) purchases a whole of life annuity-due of $10,000 p.a. for
$183,290. Suppose Thomas (currently aged 50 exact) is considering purchasing a whole of life annuity-
immediate paying $5,000 annually in arrears. The insurer believes his probability of survival until age
51 (i.e. 1 p50 ) is 0.994 and the probability of survival until age 52 (i.e. 2 p50 ) is 0.986. Using the principle
of equivalence, how much should Thomas pay for his annuity-immediate? Assume John and Thomas
follow the same mortality basis, and that the interest rate is 3%.
2. [Solution] The values of p0x in a new mortality table are related to the px values of a standard table by

p0x = (1 + r) · px

for some fixed constant r and for all age x. By considering a (discrete) whole life annuity-due, show
that the annuity values on the new mortality table at the rate of interest i are equal to those on the
standard table but at rate of interest i0 . Find expression for i0 . Using your intuition, what would you
expect the relationship between the two rates of interest: higher or lower?
3. [Solution] A substandard mortality table is derived by taking the force of mortality µ0x to be

µ0x = (1 + k) · µx

for some fixed constant k where µx is given by a standard Gompertz table. It is found that the values
of the substandard life annuities ä0x can be obtained by rating up the age in a table of the standard
life annuities, that is, we have
ä0x = äx+r .
Find an expression for the constant addition r. Interpret this result.

Term annuity-due and two other types of annuity due


4. [Solution] A life insurance company issues a 20-year temporary annuity of 1 payable annually in advance
to a life aged 50 exact. Calculate the expected present value of this annuity using the mortality table
PMA92C20 Ultimate with 4% interest. Also calculate the variance of this annuity.
5. [Solution] Derive a simplified expression for the actuarial present value for an n-year term insurance
payable at the end of year of death of (x), under which the death benefit in case of death in year k + 1
is s̈k+1| , 0 ≤ k < n. Interpret your result.

6. [Solution] Dickson et al. (2013a) Exercise 5.7


Given 10| äx = 4, äx = 10, 10 Ex = 0.375 and v = 0.94, calculate A1x:10| .

7. [Solution] A person age 40 wins 100,000 in the actuarial lottery. Rather than receiving the money at
once, the winner is offered the actuarially equivalent option of receiving an annual payment of K (at
the beginning of each year) guaranteed for 10 years and continuing thereafter for life.
You are given that interest rate i = 4% and the following values:

14
ˆ A40 = 0.23056;
ˆ A50 = 0.32907; and
ˆ A140:10| = 0.01151.

Calculate K.

Immediate annuities: The annual case


8. Dickson et al. (2009a) Exercise 5.3 (Dickson et al., 2013a, Exercise 5.3)
9. Dickson et al. (2009a) Exercise 5.4 (Dickson et al., 2013a, Exercise 5.4)

Annuities payable 1/m-thly


10. [Solution] You are given lx = 100, lx+1 = 72, lx+2 = 39, lx+3 = 0 and the annual interest rate i = 0.06.
(2)
Assuming that deaths are uniformly distributed between integer ages, find äx:2| .

11. [Solution] You are given:

ˆ Mortality: AM92.
ˆ Deaths are uniformly distributed within each year of age.
ˆ Interest rate: 6%.
(2)
Calculate ä30:1 .

Annuities payable continuously


12. [Solution] Consider a continuous whole life annuity of 1 on (x). We have:

ˆ T (x) is the future lifetime random variable for (x);


ˆ The force of interest and the force of mortality are constant and equal;
ˆ āx = 10

(a) Calculate the force of interest and the force of mortality.


(b) Calculate the standard deviation of ā T (x) .

13. [Solution] For a continuous whole life annuity of 1 on (x), you are given that T (x), the future lifetime
has
 a constant force of mortality of 0.06. Assume the force of interest is also constant at 4%. Calculate
P āT (x)| > āx . Interpret this probability.

14. Dickson et al. (2009a) Exercise 5.2 (Dickson et al., 2013a, Exercise 5.2)

Increasing annuities
15. [Solution] Consider a special increasing life annuity on (45). The annuity is continuously paid and
increasing, with the benefit at time t equal to $ 300t + 700 per annum. Benefits are payable for a
maximum of 20 years. Calculate the expected present value of this annuity under the following basis:

ˆ Force of interest: δ(t) = 0.03.


ˆ Force of mortality: µ45+t = 0.03t + 0.04.

16. [Solution] Express each of the following by a single (actuarial) symbol:

15
(a) the present value at age 35 of an annuity of 1 per annum, first payment at age 42;
(b) the present value at age 35 of a 15-year temporary annuity-due of 10 per month;
(c) the present value at age 50 of an annuity of 1 payable every 6 months, first payment at the end
of 3 months;
(d) the present value at age 65 of an annuity-immediate payable monthly with first payment 1, second
payment 2, increasing by 1 each month; and
(e) the present value at age 20 of a 15-year deferred 10-year temporary continuous annuity of 1 per
annum.

17. Dickson et al. (2009a) Exercise 5.5 (Dickson et al., 2013a, Exercise 5.8)
18. Dickson et al. (2009a) Exercise 5.7 (Dickson et al., 2013a, Exercise 5.10)
19. [Solution] A 10-year deferred index-linked whole life annuity due is sold to a 50-year old male. The first
annuity payment is $10,000 at age 60, and the annuity is assumed to increase at 3% p.a. Calculate the
expected present value of this product. Assume PMA92C20 mortality and an interest rate i = 7.12%.

Approximations for annuities


20. [Solution] A life insurance company issues an annuity to a life aged 60 exact. The purchase price is
$200,000. The annuity is payable monthly in advance and is guaranteed to be paid for a period of 10
years and for the whole of life thereafter. Calculate the annual annuity payment. Basis: Mortality
AM92 Ultimate, Interest 6% per annum.
21. [Solution] Dickson et al. (2013a) Exercise 5.5
Using AM92 (4%), calculate:

(a) ä40:20| ,
(4)
(b) ä40:20| using Woodhouse’s formula with two terms,
(c) ā25:10| assuming UDD,
(12)
(d) a50:20| using Woodhouse’s formula with two terms,
(12)
(e) 20 |ä45 assuming UDD

16
Solutions to Module 2 Exercises
1. [Question] The present value of Thomas’ life annuity-immediate is 5000a5 0. We are given that i = 0.03,
1 p50 = 0.994 and 2 p50 = 0.986. We will first write a present-value expression for John’s life annuity-due,
and then use the recursive relationship äx = 1 + vpx äx+1 .

10000ä52 = 183290
ä52 = 18.329
ä50 = 1 + vp50 + v 2 2 p50 ä52
0.994 0.986
=1+ + × 18.329
1.03 1.032
= 19
a50 = ä50 − 1
= 18
5000a50 = 5000 × 18
= 90000

Thus Thomas can buy the life annuity-immediate for $90,000.

2. [Question] First, note that the probability of survival for t years under the new mortality table can be
expressed as
0 t t
t px = p0x · p0x+1 · · · p0x+t−1 = (1 + r) × px · px+1 · · · px+t−1 = (1 + r) ×t px .

Thus, the APV of a (discrete) whole life annuity-due can be evaluated as



X
0t t
äx0 = 1 + t px v
t=1

X
=1+ (1 + r)t t px v t
t=1
∞  t
X 1+r
=1+ t px
t=1
1+i
∞  t
X 1+r
= t px
t=0
1+i
= ä@j
x

which is the APV of a (discrete) whole life annuity-due evaluated using the old mortality table but at
the rate of interest j where j satisfies
1 1+r
= .
1+j 1+i
Solving for this new rate of interest, we get

i−r
j=
1+r

provided i > r. [This requirement of i > r is to ensure we are not discounting at a negative interest
rate.] If r > 0, we would expect survival probabilities in the new table to be higher which will make
annuity payments to be more likely and therefore will increase the annuity value. For the annuity value
to be higher, then the discount rate must be smaller. Vice versa holds.

17
3. [Question] Note that, under Gompertz law, we have that the probability of survival for t years can be
written as
 Z t   Z t 
x s
t px = exp − µx+s ds = exp −bc c ds
0 0
x t
x
c − 1 / log c = e−bc (c −1)/ log c .
t
 
= exp −bc

Now, consider the substandard case. The probability of survival for t years in the substandard mortality
table can be written as
 Z t   Z t 
0 0
t px = exp − µx+s ds = exp − (1 + k) µx+s ds
0 0
−bcx (1+k)(ct −1)/ log c −bcx+r (ct −1)/ log c
= e =e =t px+r

where r satisfies
cx+r = cx (1 + k)
or equivalently
r = log(1 + k)/ log c.
Thus, it becomes straightforward to see that

X ∞
X
ä0x = vtt p0x = v t t px+r = äx+r .
t=0 t=0

Note that the APV of life annuities decreases with age, i.e. the older one gets, the cheaper the cost
of life annuities is (generally). This is because survival becomes less likely as we age. Thus, in this
case where we have substandard mortality rates, survival is therefore less likely for substandard than
for the standard, in which case, we would therefore expect a cheaper cost of life annuities. To make
the cost less expensive, one could therefore rate up the age, as in this situation where Gompertz law is
assumed. [Indeed, you can extend this rating up of age principle in the case of Makeham law. Try it!]

4. [Question] Firstly, we will calculate 20 E50 as an intermediary calculation, and then calculate the ex-
pected value and variance of the benefits.

20 E50 = v 20 20 p50
9238.134
= 1.04−20 ×
9941.923
= 0.42408
h i
E a∗ ∗ min(K50 + 1, 20) = a∗ ∗ 50 : 20
= a∗ ∗ 50 − 20 E50 a∗ ∗ 70
= 18.843 − 0.42408 × 11.562
= 13.9398
1 − v min(K50 +1,20)
   
Var a∗ ∗ min(K50 + 1, 20) = Var
d
1  
= 2 Var v min(K50 +1,20)
d
 2
2
A50:20 − A50:20
=
d2

0.04
Now, we will calculate the value of 2 A50:20 and A50:20 . Note that d = iv = 1.04 ≈ 0.038462. Also note

18
that we will use Ax = 1 − da∗ ∗ x, because the values of Ax are not given in PMA92C20.
2
2 2
A50:20 = 2 A50 − v 20 20 p50
2
A70 + v 20 20 p50

9238.134 9238.134
= 0.08802 − 1.04−40 × × 0.33469 + 1.04−40 ×
9941.923 9941.923
= 0.21678693
A50:20 = A50 − 20 E50 A70 + 20 E50
= (1 − da∗ ∗ 50) − 20 E50 (1 − da∗ ∗ 70) + 20 E50
= (1 − 0.038462 × 18.843) − 0.42408 (1 − 0.038462 × 11.562) + 0.42408
= 0.4638479
2
  0.21678693 − (0.46384798)
∴ Var a∗ ∗ min(K50 + 1, 20) = 2
(0.038462)
= 1.103

5. [Question] The APV for the term insurance can be derived as


n−1
X n−1
X
APV = s̈k+1| v k+1 k| qx = äk+1| k| qx
k=0 k=0
= äx:n| − än| n px

The benefit is indeed an accumulated value of a series of payments paid prior to death. This is then
(almost) equivalent to a temporary life annuity, except that if you survive to reach the end of the term,
there would have been no benefits paid. Hence, the reduction in the second term in the second line of
the equation above.

6. [Question] Dickson et al. (2013b)


We use

A1x:10| = Ax:10| −10 Ex

and

Ax:10| = 1 − däx:10|

from the given information, we calculate

äx:10| = äx −10 |äx = 6,

giving

Ax:10| = 1 − (1 − 0.94)6 = 0.64

and hence

A1x:10| = 0.64 − 0.375 = 0.265

7. [Question] By actuarial equivalence, we are required to solve for K from the equation

100000 = Kä10| + K 10| ä40 = Kä10| + K · v 10 10 p40 · ä50

so that
100000
K= .
ä10| + v 10 10 p40 · ä50

19
We know that
1 − v 10
ä10| = = 8.435332
d
and that
1 − A50 1 − 0.32907
ä50 = = = 17.44418.
d 0.04/1.04
Now, from the relationship
1
A40:10| = A40 − v 10 10 p40 A50
we get
1
A40 − A40:10| 0.23056 − 0.01151
v 10 10 p40 = = = 0.665664.
A50 0.32907
Hence, finally we get
100000 100000
K= = = 4, 988.21.
8.435332 + 0.665664 (17.44418) 20.04729

8. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual .


9. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual .
10. [Question] We have

(2) 1 0 1 3
ä = (v 0 px + v 2 12 px + vpx + v 2 32 px )
x:2| 2
1 1 (100 + 72)/2 72 3 (72 + 39)/2
= (1 + 1.06− 2 + 1.06−1 + 1.06− 2 )
2 100 100 100
= 1.51155

11. [Question] We have

(2) 1 1 1 1 1 1
ä30:1 = + v2 1 p30 = + v 2 (1 − 12 q30 )
2 2 2 2 2

Under UDD assumption,


1
1 q30 = q30 .
2 2

Therefore,

(2) 1 1 1 1
ä30:1 = + ×√ × (1 − × (0.000590)) = 0.9855
2 2 1.06 2

12. [Question] Under a constant force of mortality and force of interest, we may write:
Z ∞
āx = v t t px dt
Z0 ∞
10 = e−δt e−µt dt
0
Z ∞
10 = e−(δ+µ)t dt
0
1
10 =
δ+µ
δ + µ = 0.1
δ = µ = 0.05

20
To calculate the variance of the continuous annuity payment, we first require Āx and 2 Āx .
Z ∞
Āx = v t t px µx+t dt
0
Z ∞
=µ v t t px dt
0
µ
=
δ+µ
0.05
=
0.05 + 0.05
= 0.5
2 µ
Likewise, Āx =
2δ + µ
1
=
3
  2 Ā − Ā 2
x x
Var ā T (x) = 2
δ
1 2
− 0.5
= 3 2
(0.05)
100
=
3

q
100
Therefore, the standard deviation of the payments is 3 ≈ 5.7735

13. [Question] We are interested in the probability that the present value random variable of the annuity
exceeds the expectation. Now we know that

1 − e−δT (x)
āT (x)| = .
δ
Rt
Now under the constant force assumption, t px = e− 0 µx+s ds = e−0.06t . Also, we know that the APV
of the life annuity is
Z ∞ Z ∞
āx = t
v t px dt = e−0.06t .e−0.04t dt
t=0 t=0
Z ∞
= e−0.1t dt = −10 [0 − 1] = 10
t=0

So, the probability of interest is

1 − e−δT (x)
     
− log (1 − 10δ) − log 0.6
P > 10 = P T (x) > = P T (x) >
δ δ 0.04

Given the constant force of mortality, we can easily obtain the distribution of the exact future lifetime.
We know that t px = P (T (x) > t), so that
    
− log 0.6 − log 0.6
P T (x) > = exp −0.06 = 0.46476.
0.04 0.04

Thus, the required probability is equal to 0.4648. If the net single premium is āx for the life annuity,
then it has the interpretation of the probability that the premium is insufficient to pay the benefits.

14. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual .

21
15. [Question] We can first find the probability of survival as:
 Z t 
p
t 45 = exp − µ45+s ds
0
 Z t 
= exp − 0.03s + 0.04ds
0
= exp −0.015t2 − 0.04t

Z 20
EP V = (300t + 700)v t t p45 dt
0
Z 20
2
= (300t + 700)e−0.03t e−0.015t −0.04t
dt
0
Z 20
2
= 10000 (0.03t + 0.07)e−0.015t −0.07t
dt
0

du
Let u = 0.015t2 + 0.07t, so that dt = 0.03t + 0.07. The interval of the integral are transformed to 0
and 7.4. Then,
Z 7.4
EP V = 10000 e−u du
0
= 9993.89

16. [Question]
(a) 7| ä35

(12)
(b) 120ä35:15|

(2) (2)
(c) 2 1/4| ä50 = 2v 1/4 1/4 p50 · ä50+(1/4)

(12)
(d) 12 I (12) a 65

(e) 15| ā20:10|

17. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual .

18. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual

19. [Question] By the equivalence principle, the net single premium is simply the expected present value
1 1.03 1
of benefits. Let g = 0.03, v = 1+7.12% . Note that this means (1 + g)v = 1.0712 = 1.04 = v4% . Now,

EP V = 10000 v 10 10 p50 + (1 + g)v 11 11 p50 + (1 + g)2 v 12 12 p50 + . . .




= 10000v 10 10 p50 1 + (1 + g)v 1 p60 + (1 + g)2 v 2 2 p60 + . . .




= 10000v 10 10 p50 1 + v4%1 p60 + v4% 2



2 p60 + . . .

= 10000v 10 10 p50 ä60@4%


1 9826.131
= 10000 × × × 15.632
1.071210 9941.923
= 77664.23

22
20. [Question] This is a guaranteed annuity-due. We may write the present value as the sum of the term
annuity-certain plus the deferred life annuity.

(12) (12) (12)


ä = ä 10 + 10| ä60
50:10
1 − v 10 (12)
= + 10 E60 ä70
d(12)
1 − v 10 (12)
= (12) + v 10 10 p60 ä70
d  
1 − 0.558395 8054.0544 11
= + 0.558395 × 9.140 −
0.0581277 9287.2164 24
= 11.8013
(m) (12)
Note that we used the approximation äx ≈ äx − m−12m , giving ä70 ≈ ä70 − 12−1
2×12 . d(12) may be
obtained from page 58 of the Formulae and Tables book.
Finally, the annual annuity payment may be calculated using:
(12)
C ä = 200000
50:10
200000
C= = 16, 947.34
11.8013

21. [Question] Dickson et al. (2013b)


(a) We have
l60
ä40:20| = ä40 − v 20 ä60
l40
9287.2164
= 20.005 − 0.45639 × × 14.134
9856.2863
= 13.9268.
(b) Using Woolhouse’s formula with two terms
   
(4) 3 l60 3 9287.2164
ä40:20| = ä40:20| − 1 − v 20 = 13.9268 − 1 − 0.45639 ×
8 l40 8 9856.2863
= 13.7131.
(c) Under the UDD assumption
 
10 l35
a25:10| = α(∞)ä25:10| − β(∞) 1 − v
l25
We have
id 0.04 × 0.038462 i−δ
α(∞) = = = 1.000126, β(∞) = = 0.5064075
δ2 0.0392212 δ2
where
1
d=1− = 0.038462
1+i
δ = ln 1.04 = 0.039221
This gives
   
9894.4299 9894.4299
a25:10| = 1.000126 22.520 − 0.67556 × × 21.003 − 0.5064075 × 1 − 0.67556 ×
9953.6144 9953.6144
= 8.250308

23
(d) Using Woolhouse’s formula with two terms
 
(12) (12) 1 l70
a50:20| = ä50:20| − 1 − v 20
12 l50
   
11 l70 1 l70
= ä50:20| − 1 − v 20 − 1 − v 20
24 l50 12 l50
 
l70 13 l 70
= ä50 − v 20 ä70 − 1 − v 20
l50 24 l50
 
8054.0544 13 8054.0544
= 17.444 − 0.45639 × × 10.375 − 1 − 0.45639 ×
9712.0728 24 9712.0728
= 13.18065.

(e) Under the UDD assumption,

id 0.04 × 0.038462 i − i(12)


α(12) = = = 1.000128 and β(12) = = 0.464804
i(12) d(12) 0.039285 × 0.039157 i(12) d(12)
where

j = ((1 + i)(1/12) − 1) = 0.00327374


i(12) = 12 × j = 0.039285
 
1
d(12) = 1 − × 12 = 0.039157
1+j

This gives

(12) l65 (12)


20 |ä45 = v 20 ä
l45 65
l65
= v 20 (α(12)ä65 − β(12))
l45
8821.2612
= 0.45639 × × (1.000128 × 12.276 − 0.464804)
9801.3123
= 4.85215

24
Module 3: Net Premium Valuation

Net premium and the equivalence principle


1. Dickson et al. (2009a) Exercise 6.2 (Dickson et al., 2013a, Exercise 6.4).
Remark: Sometimes the exercise question does not specify the interest rate and the life table. Note that
it has been stated in the beginning of the textbook exercises, quoted, “unless otherwise stated you should
assume the mortality follows the Standard Select Survival Model as specified in Section 6.3, that interest
is at 5% per year effective, and the equivalence principle is used for the calculation of premiums.” You
can find the values for the corresponding actuarial symbols in Section 6.3. Alternatively, you can also
find the Standard Ultimate Survival Model here: click this link.

2. [Solution] Consider the following special 3-year endowment on (69):

ˆ Net level premium is payable at the beginning of each year.


ˆ A death benefit is payable at the end of the year of death, provided that death occurs within the
first 3 years.
ˆ The amount of the death benefit is the sum of the net premiums paid without interest.
ˆ A lump sum of 10,000 is payable at the end of year 3, provided that the life is still alive then.
ˆ Interest rate: 4% per annum.
ˆ p69 = 0.90, p70 = 0.88, p71 = 0.85.

Calculate the annual premium.

3. [Solution] Consider a 3-year term assurance issued to a select life aged [40]. The select period is 1 year,
with q[x] = 0.8qx . The sum assured is $10,000. Net level premiums are payable annually in advance
and all benefits are payable at the end of year of death. Suppose that mortality follows Gompertz’
Law with B = 0.02, c = 1.03. Calculate the annual premium under an interest rate of 5%.

Fully discrete annual premiums - WLI and term insurance


4. [Solution] Calculate the following using the AM92 table at 4% interest:

(a) P55
1
(b) P55:10
 
(4)
(c) P A55:10
 
(4)
(d) P A30:20

5. [Solution] A new mortality table is constructed by increasing the px values of a standard table by a
(small) constant percentage so that

p∗x = (1 + r) px for all x.

25
Prove that the net annual premium for a fully discrete whole life insurance policy calculated under this
new mortality basis can be expressed as

1
Px∗ = −d
äx @i∗

where d = iv and äx @i∗ is calculated from the standard table at the rate of interest i∗ . Find an
expression for i∗ .

6. [Solution] For a fully discrete whole life insurance of 1, 000 issued to (60), the annual benefit premium
was calculated using the following assumptions: i = 6%, q60 = 0.01376, 1000A60 = 369.33, and
1000A61 = 383.00. A particular insured is expected to experience a first-year death probability 10 times
the death probability used to calculate the annual benefit premium. The expected death probabilities
for all other years are the ones originally used. Calculate the expected loss at issue for this insured
based on the original benefit premium. Why do you think there is a loss?

7. [Solution] Consider a life insurance that provides a death benefit of $2,000 payable immediately upon
death and a survival benefit of $1,000 payable on every fifth anniversary of the inception of the policy.
Assume a constant force of mortality, 0.05, and a constant force of interest, 0.04. Calculate the level
premium payable annually in advance for life.

Fully discrete annual premiums of some other types of insurance


contracts
8. [Solution] An n-year endowment policy issued to (x) provides for the payment of ät| if death occurs
during the first policy year, ä2t| if death occurs during the second policy year, etc., and änt| if death
occurs during the n-th policy year or in the event that (x) survives the n-year period. Show that the
net annual premium P payable for r years can be expressed as

Ax:n| − A∗x:n|
P =
1 − Ax:r|
t+1
where A∗x:n| is calculated at the rate of interest i∗ = (1 + i) − 1, where i is the rate of interest
assumed in the premium calculations.

9. [Solution] Suppose that mortality follows the AM92 (Select) mortality table and interest rates are 4%
for the first 15 years and 6% thereafter. A contract issued to life (50) provides for a life annuity
beginning at age 65. The annual annuity payment is 1,000 for the first 10 years and 2,000 thereafter
for life. If death occurs before annuity payments begin, there is a death benefit of 10,000 paid at the
end of the year of death. Level annual premiums are payable for 15 years.
Calculate the amount of the level annual premium.

Fully continuous premium


10. [Solution] Consider a fully continuous increasing whole life insurance contract issued to (30). Assume
that the force of interest δ and the force of mortality µx = µ for every age x are constants. Express the
fully continuous level annual benefit premium, P̄ (I¯Ā)x , in terms of δ and µ, and calculate it assuming
the force of interest δ = 0.06 and the force of mortality µx = µ = 0.04 for every age x.

Premiums paid m times a year


11. [Solution] Express each of the following premiums in terms of actuarial notations for annuities and
assurances using the principle of equivalence:

26
(a) the net annual premium payable continuously for 20 years to provide a term insurance to age 60
on a life now aged 25. Assume the benefit is payable immediately upon death.
(b) the net annual premium payable quarterly for 15 years to provide a deferred annuity of 10 per
month to (45) with first payment at age 60

12. [Solution] A life aged exactly 45 buys a 20-year endowment assurance policy with a sum assured of
$100, 000 payable on maturity or at the of year of earlier death. Level premiums are payable monthly
in advance. Calculate the monthly premium assuming AM92 Ultimate mortality and 4% pa interest.
Ignore expenses.

27
Solutions to Module 3 Exercises

1. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

2. [Question] EPV(premiums)=EPV(benefits)

P (1 + vp69 + v 2 2 p69 ) = P vq69 + 2P v 2 p69 q70 + 3P v 3 2 p69 q71 + 10000v 3 3 p69

2 p69 = p69 p70 = 0.90 × 0.88 = 0.792


3 p69 = p69 p70 p71 = 0.90 × 0.88 × 0.85 = 0.6732
q69 = 1 − p69 = 1 − 0.9 = 0.1
q70 = 1 − p70 = 1 − 0.88 = 0.12
q71 = 1 − p71 = 1 − 0.85 = 0.15

1 1 2
P (1 + × 0.9 + ( ) × 0.792)
1.04 1.04 
1 1 2 1 3 1 3
=P × 0.1 + 2( ) × 0.9 × 0.12 + 3 ( ) × 0.792 × 0.15 + 10000( ) × 0.6732
1.04 1.04 1.04 1.04

P = 3015.

3. [Question] We will write an expression for the EPV of benefits and EPV of premiums and equate them
according to the equivalence principle.

APV(Benefits) = 10000 vq[40] + v 2 p[40] q41 + v 3 p[40] p41 q42




APV(Premiums) = P 1 + vp[40] + v 2 p[40] p41




Now, we will calculate the probabilities of survival and death.According to Gompertz law, µx = Bcx ,
x t
which may be written as t px = g c (c −1) where g = exp − lnBc (F & T, page 32).

 
0.02
g = exp −
ln 1.03
= 0.50833
40
p40 = 0.508331.03 (1.03−1)

= 0.93593
q40 = 0.06407
q[40] = 0.8 × 0.06407
= 0.051256
p[40] = 0.94874
41
p41 = 0.508331.03 (1.03−1)

= 0.93407

28
42
p42 = 0.508331.03 (1.03−1)

= 0.93216
 
0.051256 0.94874 (1 − 0.93407) 0.94874 × 0.93407 × (1 − 0.93216)
APV(Benefits) = 10000 + +
1.05 1.052 1.053
= 1574.83
 
0.94874 0.94874 × 0.93407
APV(Premiums) = P 1 + +
1.05 1.052
= 2.70736P

Finally, we equate the APV of benefits and APV of premiums and solve for the net level premium P .

2.70736P = 1574.83
P = 581.68

4. [Question] Using the AM92 table at 4% interest, we have:


A55 0.38950
(a) P55 = ä55
= 15.873 = 0.0245385
D
A155:10 A55:10 −10 E55 A55:10 − D65 689.23
0.68388− 1105.41
1
(b) P55:10 = ä
= ä
= ä
55
= 8.219 = 0.0073456
55:10 55:10 55:10
(4)
A55:10 i
A155:10 +10 E55 0.04
 
(4) i(4) 0.039414 ×0.06037+0.6235
(c) P A55:10 = ä
≈ ä
= 8.219 = 0.08331566
55:10 55:10
(4) i (1) 
D

D
A A + E i
(A30 −20 E30 A50 )+20 E30 i
A30 − D50 A50 + D50
30:20 20 30
 
(4) 30:20 i(4) i(4)
i(4)
(d) P A30:20 = ≈ = = 30
D
30
=

30:10
ä30 −20 E30 ä50 ä30 −20 E30 ä50 ä30 − D50 ä50
30
0.04
0.039414 (0.16023− 1366.61
3060.13 ×0.32907)+ 3060.13
1366.61
0.460055
21.834− 1366.61
= 14.04376 = 0.03275868
3060.13 ×17.444

5. [Question] Note
1
Px∗ = − d.
ä∗x
We have

X ∞
X ∞
X
ä∗x = v n n p∗x = v n (1 + r)n n px = (v ∗ )n n px ,
n=0 n=0 n=0

where v = (1 + r)v.
Hence, that the life annuity-due for the new mortality basis can be expressed as the old mortality basis
but computed at the rate of interest
1 i−r
i∗ = ∗
−1=
v 1+r
Hence, the result immediately follows because we know

A∗x = 1 − däx

where d = iv, still computed under the old basis.

6. [Question] A loss is expected because with a higher mortality rate than expected, there will be greater
chance of paying out a death benefit in the first year. Now with the original benefit premium, we have

A60 dA60 (0.06/1.06)0.36933


P60 = = = = 0.0331480357.
ä60 1 − A60 1 − 0.36933

29
The new single benefit premium would have been

AN
60
EW N EW
= vq60 + vpN60
EW
A61
1
= (0.1376 + 0.8624 × 0.383) = 0.4414143396.
1.06

Therefore, the expected value of the loss is

1 − AN EW
 
60
AN EW
P60 äN EW
AN EW

1000 60 − 60 = 1000 60 − P60
d
 
1 − 0.4414143396
= 1000 0.4414143396 − 0.0331480357
(0.06/1.06)
= 114.298.

7. [Question] The value of the death benefit is:


Z ∞
Death benefit = 2000 v t t px µx+t dt
Z0 ∞
= 2000 e−0.04t e−0.05t × 0.05dt
0
0.05
= 2000 ×
0.09
= 1111.111

The value of the survival benefit is:

Survival benefit = 1000 v 5 5 px + v 10 10 px + v 15 15 px + . . .




= 1000 e−0.04×5 e−0.05×5 + e−0.04×10 e−0.05×10 + e−0.04×15 e−0.05×15 + . . .




= 1000 e−0.45 + e−0.9 + e−1.35 + . . .




e−0.45
= 1000 ×
1 − e−0.45
= 1759.596

Let P be the annual premium. Then the value of premiums is given by:

EPV(Premiums) = P vpx + v 2 2 px + v 3 3 px + . . .


= P 1 + e−0.09 + e−0.09×2 + e−0.09×3 + . . .




1
=P ×
1 − e−0.09
= 11.619P

By the principle of equivalence, the value of premiums must equal the total EPV of death benefits and
survival benefits. Thus,

11.619P = 1111.111 + 1759.596


P = 247.07

30
8. [Question] The APV of future benefits can be derived as
n−1
X
v k+1 ä(k+1)t| k| qx + änt| v n n px
k=0
n−1
1 − v (k+1)t 1 − v nt
X    
k+1
= v k| qx + v n n px
d d
k=0
"n−1 n−1
#
1 X k+1 X
(t+1)(k+1)

n (t+1)n

= v k| qx − v k| qx + v n px − v n px
d
k=0 k=0
  
1
= A1x:n| − A∗1 + A 1 − A∗ 1
d x:n| x:n| x:n|
1 
= Ax:n| − A∗x:n|
d
where A∗1 , A∗ 1 and A∗x:n| are all computed at the rate of interest
x:n|
x:n|

t+1
i∗ = (1 + i) − 1.

Therefore, our premium equation becomes


1 
P äx:r| = Ax:n| − A∗x:n|
d
and the result should immediately follow because däx:n| = 1 − Ax:n| .

9. [Question] Denote by π the amount of the level annual premium. Thus, the APV of future premiums
will be
APV(Future Premiums) = π · ä[50]:15|@4% = 11.259π.
The APV of future benefits can be expressed as

APV(Future Benefits) = 10000A1[50]:15|@4%


15 10

+v@4% 15 p[50] 1000ä65@6% + 1000v@6% 10 p65 ä75@6% .

Note that we have

A1[50]:15|@4% = A[50]:15|@4% − A 1
50:15|@4%
15
= A[50]:15|@4% − v@4% 15 p[50]
8821.2612
= 0.56695 − (1.04)−15
9706.0977
= 0.062309,

and that we have


15 10

v@4% 15 p[50]
1000ä65@6% + 1000v@6% 10 p65 ä75@6%
8821.2612
= (1.04)−15
 9706.0977 
−10 6879.1673
× 1000(10.569) + 1000(1.06) 7.679
8821.2612
= 7021.067.

Therefore, the annual premium, according to the equivalence principle, is equal to


623.09 + 7021.067
π= = 678.937.
11.259

31
10. [Question] Through integration by parts, we can have
Z ∞ α
β
tα−1 e−βt dt = 1.
0 Γ (α)

A shortcut to solve the above integral: the function inside is the p.d.f of a Gamma distribution,
therefore, the integral is the total probability of 1.
By constant force assumption where µ and δ are both constants, we have that
Z ∞ Z ∞
T
 t −µt Γ (2) (µ + δ)2 2−1 −(µ+δ)t µ
E Tv = tv · µe dt = µ 2
t ·e dt = .
0 (µ + δ) 0 Γ (2) (µ + δ)2
| {z }
1

Since the contract is fully continuous, the premium can be shown as


T
 µ
E T v (µ+δ)2 µ
π = P̄ (I¯Ā)x =   = 1 =
E āT | (µ+δ)
µ + δ

and
µ 0.04
= = 0.4.
µ+δ 0.04 + 0.06
Note that P̄ means that premiums are paid continuously and (I¯Ā)x indicates that the death benefit is
continuously increasing at a rate of $1 p.a. and is paid immediately on death.
11. [Question]
A125:35
1
(a) 20 P̄ (A25:35 ) = ä25:20
    (12)
(12) (12) ä
(b) 12 × 10 × P (4) 15| ä45 or 12 × 10 ×15 P (4) 15| ä45 = 120 15|(4)45
ä45:15

1 (12)
12. [Question] Let P denote the monthly premium. In this case, P = 12 P45:20| . (You can denote the
symbol P as the monthly or annual premium for your convenience, but must remember to adjust the
formula accordingly.)
The expected present value of the premium is (remind that we define P as the monthly premium, so
the annual premium is 12P instead of P ):
  
(12) 11 D65
12P ä45:20 = 12P ä45:20| − 1−
24 D45
  
11 689.23
= 12P 13.780 − 1−
24 1677.97
= 162.1191P

The expected present value of the benefit is

100, 000A45:20| = 100, 000 × 0.46998 = 46, 998

so the monthly premium is


46998
≈ 289.90
P =
162.1191
The Woolhouse’s approximation is listed in the Orange book (page 36)

32
Module 4: Net Premium Reserves

Net premium reserves


1. Dickson et al. (2009a) Exercise 7.1 (Dickson et al., 2013a, Exercise 7.1).

2. Dickson et al. (2009a) Exercise 7.8 (a) (b) (Dickson et al., 2013a, Exercise 7.9).

3. [Solution] Suppose that 90,000 lives aged 30 each purchase an ordinary whole life insurance of 1. The
net annual premium is 0.02. If interest rate is at 5% and if the mortality table used provides for 700
deaths at age 30, compute 1 V30 .

4. [Solution] An individual aged 25 purchases an ordinary whole life insurance policy. Each year he invests
elsewhere the difference between the net premium for his policy and the corresponding net premium for
a twenty-year endowment policy. Approximately what interest rate must he earn on this investment
in order that it may accumulate at the end of twenty years to the difference between the face amount
and the reserve for his policy. You are given the following:

ä25 = 23 ä20| at 3.5% = 14.7

ä45 = 16 ä20| at 4% = 14.1

ä25:20| = 15 d = 0.0338 for i = 3.5%

d = 0.0385 for i = 4%

Explain briefly why the rate he must earn on his investment differs from that used in the calculation
of the annual premium.

5. [Solution] A large machine in the ABC Paper Mill Company is 25 years old when ABC purchases a
5-year term insurance paying a benefit in the event the machine breaks down. Calculate the benefit
reserve for this insurance at the end of the third year. You are given the following:

ˆ Annual benefit premiums of 6,643 are payable at the beginning of the year.
ˆ A benefit of 500,000 is payable at the moment of breakdown.
ˆ Once a benefit is paid, the insurance contract is terminated.
ˆ Machine breakdowns follow De Moivre’s Law with `x+t = 100 − (x + t).
ˆ i = 6%.

6. [Solution] Prove the following are true:


 
1+i 1+i
(a) k Vx = Ax+k 1+ Px − Px .
i i
Qk
(b) k Vx = 1 − j=1 (1 − 1 Vx+j−1 ).

33
7. [Solution] You are given:
k äk| k−1| qx

1 1.000 0.33
2 1.930 0.24
3 2.795 0.16
4 3.600 0.11
Calculate 2 Vx:4| .

8. [Solution] A special endowment assurance issued to a man aged 40 exact has a term of 20 years. A
sum assured of $10, 000 is payable immediately on death, and a sum assured of $15, 000 is payable on
survival to the end of the term.
The policy is secured by annual premiums of P during the first five years, 2P during the second five
years and 3P during the remainder of the term. Premiums are payable annually in advance for 20
years or until death, if earlier.

(a) Find the initial net annual premium for this policy.
(b) Determine the prospective net premium policy value (reserves) at the end of the twelfth policy
year, immediately before the payment of the thirteenth premium.
Basis: AM 92 Ultimate Mortality Table. 4% per annum interest.

9. [Solution] Consider a life (x) with qx = 0.02, qx+1 = 0.03, and qx+2 = 0.04. Let v = 0.97.

(a) Find the level benefit premium and the benefit reserves for a special 3-year term insurance with
payments of b1 = 10, b2 = 20, and b3 = 15 .
(b) Do the same for a 3-year endowment insurance with the death benefits as above and with a
payment of 20 at the beginning of the 4-th year if the insured attains age x + 3.

10. [Solution] If 0 L = T (x)v T (x) − πāT (x)| and the forces of mortality and interest are constant, express
(a) π and (b) t V̄ in terms of µ and δ.

Retrospective reserves
11. [Solution] Consider a life (x) with qx = 0.02, qx+1 = 0.03, and qx+2 = 0.04. Let v = 0.97.

(a) Suppose that the level benefit premium determined by the equivalence principle is 0.445621.
Calculate the retrospective reserves for a special 3-year term insurance with payments of b1 = 10,
b2 = 20, and b3 = 15 .
(b) Suppose the level benefit premium determined by the equivalence principle for a 3-year endowment
insurance with the death benefits as above and with a payment of 20 at the beginning of the 4-th
year if the insured attains age x + 3 is 6.300649. Calculate the retrospective reserves.

Recursive calculation of reserves


12. Dickson et al. (2009a) Exercise 7.3 (Dickson et al., 2013a, Exercise 7.4).
Remark: The Standard Ultimate and Select Survival Model is introduced in Appendix D of the textbook.
The life table for interest rate i=5% is also included in the Appendix. However, the life tables for i=6%
and other interest rates are not included. You will need to create your own spreadsheets by using the
Standard Select Survival Model assumption and the desired interest rate.
For part (b), a common misunderstanding is that, since there is no death benefit or premium from time
2 to time 2.25, the reserve 2.25 V should simply be the reserve at time 2 multiplied by the time value,

34
0.25
i.e., 2.25 V = 2 V · (1 + i) . However, the solution reveals that this understanding is wrong:

2V = 20000 q52 · v + p52 · v · 3V


0.75
2.25 V = 20000 0.75 q52.25 · v + 0.75 p52.25 · v 0.75 · 3V
0.25
2.25 V 6= 2V · (1 + i) .

To clarify this confusion, it is essential to understand the definition of 2.25 V , which is the reserve for
each policy that is still effective at time 2.25. Indeed, for those who died from 2 to time 2.25, the
payments will be paid at time 3, which seems to be the prospective cashflows for the reserve at time
2.25. But remember that the definition of 2.25 V is conditioned on the survival of (50) until time 2.25,
so those past deaths should not be included in 2.25 V .

13. [Solution] An insured under an ordinary whole life policy for $1, 000, issued at age 20, is subject during
the 11-th policy year to an extra mortality not covered by the terms of the policy. This extra risk
may be expressed as an addition of 0.01 to the normal probability of death during that year. If the
11-th year reserve is $81.54 and the probability of death at age 30 is 0.008427, calculate the theoretical
reduction in the amount of insurance the company should require during the 11-th year if the policy
is to be amended to include the extra risk.

Death strain at risk


14. Dickson et al. (2009a) Exercise 7.12 (Dickson et al., 2013a, Exercise 7.13).

15. [Solution] Write down in the form of symbols, and also explain in words, the expressions “death strain
at risk”, “expected death strain” and “actual death strain”.

16. [Solution] A life insurance company issues 20-year term insurances with a sum assured of $100, 000
where the death benefit is payable at the end of the year of death and premiums are payable annually
in advance.
On 1 January 2002, the company sold 5, 000 term insurance policies to male lives aged 45 exact. During
the first two years, there were fifteen actual deaths from these policies written.

(a) Calculate the death strain at risk during 2004.


(b) During 2004, there were eight actual deaths from these policies written. Calculate the total
mortality profit or loss to the company in the year 2004.
Basis:
Interest 4% per annum
Mortality AM92 Ultimate Table

Thiele’s differential equation


17. Dickson et al. (2009a) Exercise 7.13 (Dickson et al., 2013a, Exercise 7.14).

18. Dickson et al. (2009a) Exercise 7.14 (Dickson et al., 2013a, Exercise 7.15).
Remark: Note that in the solution to (d), they have used an argument
Z 20
Ā1[60]+t:20−t| = e−δ(s−t) s−t p[60]+t µ[60]+s ds,
t

which can be easily understood in the following way:

ˆ To receive the continuous payment at age [60] + s, the policyholder, currently aged [60] + t, must
first survive to age [60] + s, represented by the survival probability s−t p[60]+t .
ˆ Then, the policyholder needs to die immediately at age [60]+s, represented by the force of mortality
µ[60]+s .

35
ˆ The death payment of $1 at age [60] + s needs to be discounted back to the initial age [60] + t,
represented by the discounting factor e−δ(s−t) .
ˆ Finally, given the definition of the term insurance Ā1[60]+t:20−t| , the maximum age for the death
payment is [60] + t + 20 − t = 80. Therefore, the payment date [60] + s must be between [60] + t
and 80, i.e., the range of s is from t to 20.
ˆ Combining all above items, we have the above integral expression for Ā1[60]+t:20−t| .

Some may see the range of this kind of integrals being from 0 to 20 − t in some similar questions. This
can be achieved by simply changing the variable u = s − t, and then we have
Z 20−t
Ā1[60]+t:20−t| = e−δu u p[60]+t µ[60]+u+t du.
0

The different expressions of the integral are equivalent.

19. [Solution] Consider a 15-year term assurance on (35). The survival benefit payable at maturity is
$30,000. The death benefit (payable immediately upon death before maturity) is $10,000 in the first
5 years, 30% of the reserve in the following 5 years, and 100% of the reserve in the last 5 years. The
force of interest is δ(t) = 0.002 + 0.01t.
Write down Thiele’s differential equation and state any boundary conditions.

20. [Solution] [Institute of Actuaries question, April 2006]

(a) Show that



s px+t =s px+t (µx+t − µx+t+s )
∂t
(b) Prove Thiele’s differential equation for a whole life insurance issued to a life aged x to be as
follows:

t V̄x = −µx+t (1 −t V̄x ) + δ t V̄x + P̄x .
∂t
21. [Solution] Consider a 10-year without-profit endowment assurance on (60). $10,000 is payable at
maturity, or 25% of the reserve held at the time of death is payable immediately on death, if the
insured dies before maturity. A net single premium is payable at policy outset.

(a) Write down Thiele’s differential equation for this endowment. State any boundary conditions.
(b) Calculate the net single premium of this endowment assurance, using the AM92 table at 6%
interest.

Participating life insurance


22. [Solution] A life insurance company issues a 20-year with profits endowment insurance policy to a
female life aged 45 exact. The sum assured of $150, 000 plus declared reversionary bonuses are payable
upon survival to the end of the term or immediately on death, if earlier.
Calculate the monthly premium payable in advance throughout the term of the policy if the company
assumes that future reversionary bonuses will be declared at a rate of 1.92308% of the sum assured,
compounded and vesting at the end of each policy year.
Basis:
Interest 6% per annum
Mortality AM92 Select Table

23. [Solution] A life insurance company issues a with profits whole life insurance policy to a life aged 20
exact, on 1 July 2002. Under the policy, the basic sum assured of $100, 000 and attaching bonuses are
payable immediately on death. The company declares simple reversionary bonuses at the end of each
year. Level premiums are payable annually in advance under the policy.

36
(a) Give an expression for the net future loss random variable under the policy at the outset (i.e. at
issue). Define symbols where necessary.
(b) Calculate the annual premium using the principle of equivalence.

Basis:
Interest 6% per annum
Mortality AM92 Select Table
Bonus loading 3% simple per annum.
Assume bonus entitlement earned immediately on payment of premium.

(c) On 30 June 2005 the policy is still in force. A total of $10, 000 has been declared as a simple
bonus to date on the policy.
The company calculates provisions for the policy using a prospective basis, with the following
assumptions:

Interest 4% per annum


Mortality AM92 Ultimate
Bonus loading 4% simple per annum

Calculate the provision (reserve) for the policy as at 30 June 2005.

37
Solutions to Module 4 Exercises
1. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

2. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

3. [Question] We know that the required reserve

1 V30 = A31 − P30 · ä31

where A31 is the APV of future benefits and P30 · ä31 is the APV of future premiums. From the given,
we have
A30 dA30
P30 = = = 0.02
ä30 1 − A30
so that A30 = 0.295775. We also know from the recursive relation A30 = vq30 + vp30 A31 so that

(1 + i) A30 − q30 1.05 (0.295775) − (700/90000)


A31 = = = 0.305159
p30 89300/90000

and that
1 − A31 1 − 0.305159
ä31 = = = 14.59166.
d 0.05/1.05
We finally have

1 V30 = A31 − P30 · ä31


= 0.305159 − 0.02 (14.59166)
= 0.013326.

4. [Question] The net premium for the ordinary whole life insurance policy is
1 1
P25 = −d= −d
ä25 23
and for the 20-year endowment, we have
1 1
P25:20| = −d= −d
ä25:20| 15

so that the amount to invest is the difference


1 1
P25:20| − P25 = − = 0.023188.
15 23
Now, in 20 years, the difference between the face amount and the reserve is
ä45 16
1− 20 V25 = = = 0.695652.
ä25 23
Thus, we find the approximate i that solves

0.023188 · s̈20| = 0.695652

or equivalently
s̈20| = 30.
From the given, we can deduce that the interest rate is between 3.5% and 4% since
20
s̈20| at 3.5% = ä20| at 3.5% × (1 + 3.5%) = 29.2499

38
and
20
s̈20| at 4% = ä20| at 4% × (1 + 4%) = 30.89.

By linear interpolation, we assume that the quantity s̈20| is a function of the interest rate, and is
approximately a linear function between 3.5% and 4%. Then the slope of this linear function should
be
s̈20| at 4% − s̈20| at 3.5% 30.89 − 29.2499
=
4% − 3.5% 4% − 3.5%
and this function is
30.89 − 29.2499
s̈20| at i% = 29.2499 + (i% − 3.5%) .
4% − 3.5%
Remember that the previous derivations reveal that s̈20| = 30, so the underlying interest rate should
be
30 − 29.2499
i% = 3.5% + 0.5% × = 3.73%.
30.89 − 29.2499
The rate he must earn will be different from that rate used to compute premiums because there is no
life contingent probabilities in the calculations.

5. [Question] The benefit reserves at the end of the 3rd year can be expressed as

3V = 500000 · Ā128:2| − 6643 · ä28:2| .

lx+t 100−(x+t)
Note t px = lx = 100−x
100−(x+t)
dlog(t px ) dlog( 100−x ) 1
and µx+t = − dt =− dt = 100−(x+t)
Thus,
71/72
ä28:2| = 1 + vp28 = 1 + = 1.930294
1.06
and
Z 2
Ā128:2| = v t t p28 µ28+t dt
0
2
100 − (28 + t)
Z
1
= vt dt
0 100 − 28 100 − (28 + t)
Z 2
1 1 h
−2
i
= v t dt = −1 (1.06) − 1 = 0.026220282.
72 0 72 log (1.06)

Therefore, we have

3V = 500000 (0.026220282) − 6643 (1.930294) = 287.20.

6. [Question] To prove:

(a) Starting with the RHS,


   
1+i 1+i Px Px
Ax+k 1 + Px − Px = Ax+k 1 + −
i i d d
Px
= Ax+k − (1 − Ax+k )
d 
1 − Ax+k
= Ax+k − Px
d
= Ax+k − Px · äx+k =k Vx .

39
äx+k
(b) We know that k Vx = 1 − and beginning with the RHS, we have
äx
 
k k äx+j
1 − Πj=1 (1 − 1 Vx+j−1 ) = 1 − Πj=1
äx+j−1
 
äx+1 äx+2 äx+k
= 1− · ···
äx äx+1 äx+k−1
äx+k
= 1− =k Vx .
äx

7. [Question] The required reserve can be written as

2 Vx:4| = Ax+2:2| − Px:4| · äx+2:2| .

Consider
( !
äK+1| , K<3
äx:4| = E Y =
ä4| , K≥3
= 1(0.33) + 1.93(0.24) + 2.795(0.16) + 3.6(1 − 0.33 − 0.24 − 0.16)
= 2.2124.
Remind that as long as (x) survives to the end of the third year, he or she has received four payments,
therefore the probability for receiving ä4| is actually (1 − 0.33 − 0.24 − 0.16).
We know from ä2| = 1 + v = 1.93 so that d = 1 − v = 1 − 0.93 = 0.07 . Therefore,

Ax:4| = 1 − däx:4| = 1 − 0.07(2.2124) = 0.845132.

Now, similarly, we have


( !
äK+1| , K(x + 2) < 2
äx+2:2| = E Y =
ä2| , K(x + 2) ≥ 2
= ä1| · qx+2 + ä2| (1 − qx+2 )
= qx+2 + 1.93(1 − qx+2 ) = 1.93 − 0.93 · qx+2

To calculate qx+2 , we have the relationship that

2| qx = P ((x) survives 2 years and dies within the third year)


= 2 px × qx+2 ,
where

2 px = P ((x) survives 2 years)


= 1 − P ((x) dies in the first year) − P ((x) dies in the second year)
= 1− 0| qx − 1| qx .

Therefore, we have
2| qx 0.16
qx+2 = = = 0.372093,
1− 0| qx − 1| qx 1 − 0.33 − 0.24
äx+2:2| = 1.93 − 0.93 · 0.372093 = 1.583953,
and
Ax+2:2| = 1 − däx+2:2| = 1 − 0.07(1.583953) = 0.889123.
Finally,  
0.845132
2 Vx:4| = 0.889123 − · 1.583953 = 0.284056.
2.2124

40
8. [Question] We consider a special endowment assurance

(a) The APV of future benefits, from issue, can be expressed as

APV(Benefits) = 10, 000Ā140:20| + 15, 000A 1


40:20|

where we can evaluate from the AM 92 Ultimate Mortality Table the following:
`60 −20 9287.216
A 1 = v 20 20 p40 = v 20 = (1.04) = 0.430 04.
40:20| `40 9856.286

 
i 1 0.04
Ā140:20| ≈ A40:20| = A40:20| − A 1
δ ln 1.04 40:20|
0.04
= × (0.46433 − 0.430 04) = 0.034971.
ln 1.04
Therefore, we have 10000(0.034971) + 15000(0.430 04) = 6800. 3

APV(Benefits) = 10000(0.034971) + 15000(0.430 04) = 6800. 3.

Now, for the APV of future premiums, we have

APV(Premiums) = P ä40:20| + P 5| ä40:15| +P 10| ä40:10|


5
= P ä40:20| + P · v 5 p40 ä45:15| + P · v 10 10 p40 ä50:10|

= P · 28.76772

Therefore, we have
6800.30
P = = 236.391.
28.76772
(b) The prospective reserves after 12 years is given by

12 V = APV (F B12 ) − APV (F P12 )


= 10000Ā152:8| + 15000A 1 − 3P ä52:8|
52:8|

where again we can evaluate from the Table the following:


`60 −8 9287.2164
A 1 = v 8 8 p52 = v 8 = (1.04) = 0.702 46.
52:8| `52 9660.5021

 
i 1 0.04
Ā152:8| ≈ A = A52:8| − A 1
δ 52:8| ln 1.04 52:8|
0.04
= × (0.73424 − 0.702 46) = 0.03241.
ln 1.04
ä52:8| = 6.910.
Thus, the reserves (or policy value) at the end of 12 years is

12 V = 10000(0.03241) + 15000(0.70246) − 3(236.39)(6.910) = 5960.64.

9. [Question]

(a) Suppose P denotes the level benefit premium. The APV of future premiums then is given by

APV(premiums) = P 1 + vpx + v 2 2 px = 2.84502P.




41
The APV of future benefits is given by

APV(benefits) = 10vqx + 20v 2 1| qx + 15v 3 2| qx


2
= 10(0.97)(0.02) + 20(0.97) (0.98)(0.03)
+15(0.97)3 (0.98)(0.97)(0.04)
= 1.267801.

By the equivalence principle, we have

P = 1.267801/2.84502 = 0.445621.

The beginning terminal reserves is clearly 0 V = 0. In year 1, we would have

1V = APV(FB) − APV(FP)
= 20vqx+1 + 15v 2 − P · (1 + vpx+1 )
1| qx+1

20(0.97)(0.03) + 15(0.97)2 (0.97)(0.04) − (0.445621) (1 + (0.97)(0.97))



=
= 0.264697.

In year 2, we would have

2V = 15vqx+2 − P = 15(0.97)(0.04) − 0.445621 = 0.136379.

Finally, because of the term insurance at the end of year 3, the terminal reserve at that point
must be equal to 3 V = 0.
(b) Now we suppose P denotes the level benefit premium for this endowment policy. The APV of
future premiums then is given by

APV(premiums) = P 1 + vpx + v 2 2 px = 2.84502P,




(similar to the previous one). However, the APV of future benefits is given by

APV(benefits) = 10vqx + 20v 2 1| qx + 15v 3 2| qx + 20v 3 3 px


2
= 10(0.97)(0.02) + 20(0.97) (0.98)(0.03)
+15(0.97)3 (0.98)(0.97)(0.04) + 20(0.97)3 (0.98)(0.97)(0.96)
= 17.92547.

By the equivalence principle, we have

P = 17.92547/2.84502 = 6.300649.

The beginning terminal reserves is again clearly 0 V = 0. In year 1, we would have

1V = APV(FB) − APV(FP)
20vqx+1 + 15v 2 + 20v 2 2 px+1 − P · (1 + vpx+1 )

= 1| qx+1

20(0.97)(0.03) + 15(0.97)2 (0.97)(0.04) + 20(0.97)2 (0.97)(0.96)



=
−(6.300469) (1 + (0.97)(0.97))
= 6.423995.

In year 2, we would have

2V = 15vqx+2 + 20vpx+2 − P
= 15(0.97)(0.04) + 20(0.97)(0.96) − 6.300649
= 12.90535.

Finally, because of the endowment at the end of year 3, the terminal reserve at that point must
be equal to 3 V = 20.

42
10. [Question] Note

β α α−1 −βt
Z
t e dt = 1.
0 Γ (α)
(a) By constant force assumption, we have that
Z ∞ Z ∞
T
 t −µt Γ (2) (µ + δ)2 2−1 −(µ+δ)t µ
E Tv = tv · µe dt = µ 2
t ·e dt = .
0 (µ + δ) Γ (2) (µ + δ)2
|0 {z }
1

Thus, the premium  µ


E T vT (µ+δ)2 µ
π=   = 1 = .
E āT | (µ+δ)
µ+δ

(b) For the reserves, we know that the prospective loss, conditioned on (x) survives t years and aged
x + t at time t, is given by

tL = [t + T (x + t)] · v T (x+t) − πāT (x+t)|


= t · v T (x+t) + [T (x + t)] · v T (x+t) − πāT (x+t)|
= t · v T (x+t) + T (x + t) · v T (x+t) − πāT (x+t)| , T (x) > t

where T (x+t) is the future lifetime of (x+t). Therefore taking the expectation of this prospective
loss, we have
  h i
E (t L|T (x) > t) = E t · v T (x+t) + E T (x + t) · v T (x+t) − πāT (x+t)|
  µ
= t · E v T (x+t) + 0 = t ×
µ+δ
which gives the reserve. Note that we have used the fact that by constant force, the density of
T h(x) gives the same density of Ti(x + t) and that is why
E T (x + t) · v T (x+t) − πāT (x+t)| = 0.

11. [Question]

(a) We know
P = 0.445621
The beginning terminal reserves is clearly 0 V = 0. In year 1, we would have

1V = Accumulated Value (Past Premiums) − Accumulated Value (Past Benefits)


P 10vqx
= −
vpx vpx
0.445621 10(0.02)
= −
(0.97)(0.98) 0.98
= 0.264697.
In year 2, we would have
P (1 + vpx ) 10vqx + 201| qx v 2
2V = 2
− 2
2 px v 2 px v
0.445621(1 + (0.97)(0.98)) 10(0.02)(0.97) + 20(0.97)2 (0.98)(0.03)
= −
(0.97)2 (0.98)(0.97) (0.97)2 (0.98)(0.97)
= 0.136379.
Finally, because of the term insurance at the end of year 3, the terminal reserve at that point
must be equal to 3 V = 0.

43
(b) We are given
P = 6.300649.
The beginning terminal reserves is again clearly 0 V = 0. In year 1, we would have

1V = Accumulated Value(Past Premiums) − Accumulated Value (Past Benefits)


P 10vqx
= −
vpx vpx
6.300649 10(0.02)
= −
(0.97)(0.98) 0.98
= 6.423995.

In year 2, we would have

P (1 + vpx ) 10vqx + 201| qx v 2


2V = 2
− 2
2 px v 2 px v
6.300649(1 + (0.97)(0.98)) 10(0.02)(0.97) + 20(0.97)2 (0.98)(0.03)
= −
(0.97)2 (0.98)(0.97) (0.97)2 (0.98)(0.97)
= 12.90535.

Finally, because of the endowment at the end of year 3, the terminal reserve at that point must
be equal to 3 V = 20.

12. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

13. [Question] Denote the required reduction in death benefit for the 11-th policy year by R and the new
rate of mortality for that year by

q30 = q30 + 0.01
where the original rate of mortality q30 = 0.008427. Now applying the recursive formula

(10 V + P ) (1 + i) = bh+1 qx+h +h+1 V px+h


= bh+1 qx+h +h+1 V (1 − qx+h )
= h+1 V + (bh+1 − h+1 V ) qx+h

= 11 V + q30 (1000 − R −11 V )
= 11 V + (q30 + 0.01) (1000 − R −11 V )
= 11 V + q30 (1000 −11 V )
−R · q30 + 0.01 (1000 − R −11 V ) .

Thus, we must have


R · q30 = 0.01 (1000 − R −11 V ) .
Intuitively this means that, on the left-hand side, the extra cost of mortality for that year should be
funded by the extra expected death strain for the year. Now solving for the reduction in benefit, we
get
0.01 (1000 −11 V ) 0.01 (1000 − 81.54)
R= = = 498.43.
q30 + 0.01 0.008427 + 0.01

14. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

15. [Question] During year t + 1, for t = 0, 1, 2, ..., the death strain per policy is the excess of the sum
insured for the policy over the end of the year reserves or provision. That is,

DS for year t + 1 = DB − t+1 V

44
where DB is the sum insured or benefit payable on death in the year.
The “expected death strain” for year t + 1 is then the amount the insurance company expects to pay
on top of the end of the year reserves (which has been set aside to fund benefits) for the policy. That
is,
EDS for year t + 1 = q · (DB − t+1 V ) .
where q is the probability of paying out the benefit for the policy during the year.
The “actual death strain” for year t+1 is the observed value at t+1 of the death strain random variable.
It is then zero if the life survived to reach t + 1; else, it is equal to the death strain, (DB − t+1 V ).

16. [Question] First we compute the reserves at the end of year 2004:

3V = 100000 · A 148:17| − P · ä48:17|

where the net premium is computed as


 
100000 · A 145:20| 100000 · A 45:20| − 20 E45
P = =
ä45:20| ä45:20|
100000 · (0.46998 − (689.23/1677.97))
= = 429.81.
13.780
Therefore the reserves gives

3V = 100000 · (0.52617 − (689.23/1484.43)) − 429.81 (12.32)


= 891.12.

(a) The death strain at risk for the year, per policy, is therefore

DSAR = 100000 − 3V = 100000 − 891.12 = 99, 108.88.

(b) The total expected death strain for the year is

Total EDS = 4985 × q47 × 99108.88


= 4985 × 0.001802 × 99108.88
= 890, 292.10

and the total actual death strain is

Total ADS = 8 × 99108.88 = 792, 871.04

so that the mortality profit/loss for the year is the difference between the two:

890, 292.10 − 795, 871.04 = 97, 421.06.

17. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

18. See Dickson et al. (2009b) Solutions Manual (Dickson et al., 2013b).

19. [Question] We split the differential equation across three cases, since the benefit payable changes:
 
δt × t V̄ + P − 10000 − t V̄ µx+t


for 0 < t < 5
d  δt × t V̄ + P − 0.3 t V̄ − t V̄ µx+t for 5 < t < 10
t V̄ = 
dt δt × t V̄ + P − t V̄ − t V̄ µx+t
 for 10 < t < 15

45
The boundary conditions are:

lim+ t V̄ =0
t→0
 
lim− t V̄ = lim+ t V̄
t→5 t→5
 
lim t V̄ = lim t V̄
t→10− t→10+

lim t V̄ = 30000
t→15−

20. (a) [Question] We have


1 ∂ ∂
× s px+t = log (s px+t )
s px+t ∂t ∂t

= (log `x+t+s − log `x+t ) = −µx+t+s + µx+t
∂t
Multiplying through by s px+t gives the required result.
(b) It has been shown (in class) that we have
āx+t
t V̄x =1− ,
āx
and applying result of previous part, we have
Z ∞
∂ ∂
āx+t = e−δs s px+t ds
∂t ∂t 0
Z ∞

= e−δs s px+t ds
0 ∂t
Z ∞
= e−δs s px+t (µx+t − µx+t+s ) ds
0
= µx+t āx+t − Āx+t

This implies therefore that



∂ − µx+t āx+t − Āx+t
t V̄x =
∂t āx
 (1 − δāx+t )
= −µx+t 1 − t V̄x +

 x 
 āx+t 1
= −µx+t 1 − t V̄x + δ 1 − −δ+
āx āx
 
 1 − δāx
= −µx+t 1 − t V̄x + δ t V̄x +
āx

= −µx+t 1 − t V̄x + δ t V̄x + P̄x ,

which gives the desired result.


21. [Question] Let N SP be the net single premium for the policy, payable at outset. This also means that
no other premiums are received continuously during the policy term.

(a) Thus, we may write Thiele’s differential equation:


d  
t V̄ = δ t V̄ + 0 − 0.25t V̄ −t V̄ µ60+t
dt
= δ t V̄ + 0.75 t V̄ µ60+t

46
As t → 0+ , the reserve will approach the net single premium received at outset. Likewise, as
t → 10− the reserve will approach the survival benefit. Therefore, the boundary conditions are:

lim t V̄ = N SP
t→0+

lim t V̄ = 10000
t→10−

(b) We may solve Thiele’s differential equation as follows:

d 
t V̄ = δ t V̄ + 0.75 t V̄ µ60+t
dt
d

dt t V̄
= δ + 0.75µ60+t
t V̄
Z 10 d  Z 10
dt t V̄
dt = δ + 0.75µ60+t dt
0 t V̄ 0
Z 10
 
ln 10 V̄ − ln 0 V̄ = 10δ + 075 µ60+t dt
0
ln 10000 − ln N SP = 10δ + 0.75 (− ln (10 p60 ))
ln N SP = ln 10000 − 10δ + 0.75 ln 10 p60
  
8054.0544
N SP = exp ln 10000 − 10 ln 1.06 + 0.75 ln
9287.2164
N SP = 5018.086
 R 
t Rt
Note that in the fourth line we used t px = exp − 0 µx+s ds ⇔ 0 µx+s ds = − ln (t px ).

22. [Question] To find the premium, we equate the APVFP to APVFB at issue. We then have

(12)
APVFP = P · ä[45]:20|

and
h   i
20
APVFB = 150000 E (1 + b)K([45]) v T ([45]) · I (T ([45]) ≤ 20) + (1 + b) · v 20 I (T ([45]) > 20)

where b = 1.92308% annual compound rate of bonus. Note that


   
E (1 + b)K([45]) v T ([45]) · I (T ([45]) ≤ 20) = E (1 + b)K([45]) v K([45])+S · I (T ([45]) ≤ 20)
 
= E v S ((1 + b)v)K([45]) · I (T ([45]) ≤ 20)

where S = T ([45]) − K([45]), the extra portion of the year lived in the year of death. Assuming UDD,
then S is uniform on (0,1) and is independent of T (45) so that
   
E v S ((1 + b)v)K([45]) · I (T ([45]) ≤ 20) = E v S · E ((1 + b)v)K([45]) · I (T (45) ≤ 20) .


and
Z 1 Z 1
d
E vS = v s ds = e−δs ds =

.
0 0 δ
Moreover, define
1
= w = (1 + b)v,
1+j

47
then
   
E ((1 + b)v)K([45]) · I (T (45) ≤ 20) = E wK([45]) · I (T (45) ≤ 20)
1  K([45])+1 
= E w · I (T (45) ≤ 20)
w
1
= Aj 1 .
(1 + b)v [45]:20|

Thereofe, we have
  d 1
E v S ((1 + b)v)K([45]) · I (T ([45]) ≤ 20) = Aj 1
δ (1 + b)v [45]:20|
i 1
= Aj 1
δ 1 + b [45]:20|
0.06 1
= Aj 1
ln(1.06) 1.0912308 [45]:20|
= 1.01028 · Aj 1
[45]:20|

where the 20-year term insurance Aj 1 is evaluated at the rate of interest j = (1.06/1.0192308)−1 =
[45]:20|
4%,
D65 689.23
Aj=4%
1 = Aj=4% −20 E j=4%
[45] = 0.46982 − = 0.46982 − = 0.058933.
[45]:20| [45]:20| D[45] 1677.42

Therefore, we have
h i
20
APVFB = 150000 1.01028(0.058933) + (1.0912308) v 20 20 p[45]
= 150000 1.01028(0.058933) + (v@j = 4%)20 20 p[45]
 
 
D65 @j = 4%
= 150000 1.01028(0.058933) +
D[45] @j = 4%
 
689.23
= 150000 1.01028(0.058933) +
1677.42
= 70, 563.87.

The annual premium then is


70, 563.87 70, 563.87 70, 563.87
P = (12)
≈  =  
D65
ä[45]:20| ä[45]:20| − 11
24 1 − D[45] 11.888 − 11
24 1 − (1.06)−20 ``[45]
65

70, 563.87 70, 563.87


= 11 −20 8821.2612 = 11.55833 = 6, 105.023

11.888 − 24 1 − (1.06) 9798.0837

so that the monthly premium equals P/12 = 508.75.


23. [Question] Denote by K([20]) and T ([20]) the curtate future lifetime and the future lifetime of a person
select at age 20.

(a) The net future loss random variable from issue is given by

0L = 100, 000 (1 + bK([20])) v T ([20]) − P · äK([20])+1|

where P is the annual premium, b is the annual rate of simple bonus.


(b) By the equivalence principle, the premium is given by
h  i
P · ä[20] = 100, 000 · Ā[20] + 3, 000 · I Ā [20]
− Ā[20]

48
so that

i 97, 000 · A[20] + 3, 000 · (IA)[20]


P =
δ ä[20]
0.06 97, 000 · 0.04472 + 3, 000 · 2.00874
=
ln(1.06) 16.877
= 632.3376.

(c) It is explicitly stated that the total simple bonus declared to date is $10,000 . Hence, there
is $100, 000 + $10, 000 = $110, 000 sum assured on year 3. The example provided in the video
lecture makes no provision for the future simple bonus rate, so we usually do not include any
future bonuses into the reserve calculations for conservative reserves. However, here the question
also specifically makes a provision for a simple bonus rate of 4 % to be applied in the future.
Hence,we need to include te future bonus for this question. The policy reserve as at 30 June 2005
is given by

= 110, 000Ā23 + 4, 000 · I Ā 23 − Ā23 − P 0 · ä23


  
3V
i i
= 106, 000 A23 + 4, 000 · (IA)23 − P 0 · ä23
δ δ
where, to be on the conservative side, P 0 is set to be the net premium assuming no bonuses in
the future:
Ā20 i A20
P0 = 100, 000 ≈ 100, 000
ä20 δ ä20
0.04 0.11226
= 100, 000 = 496.0380
ln(1.04) 23.081

so that
0.04 0.04
3V = 106, 000 (0.12469) + 4, 000 · (6.09644) − 496.0380(22.758)
ln(1.04) ln(1.04)
= 27, 061.21.

Remark: Many students are confused why this question considers future bonuses while the lecture
slides demonstrate that future bonuses should not be considered. Note that, in this problem, it was
explicitly made to provision for a 4% annual rate of simple bonus. So we include these future
bonuses into our consideration. However, normally, on the conservative side, if no such explicit
assumption is made, you would assume zero bonuses from then on.

49
Module 5: Gross Premiums and
Reserves

Gross premium and reserve calculations


1. [Solution] For a fully discrete whole life policy of amount 1, the level expense-loaded premium is based
on the following schedule of expenses:

ˆ An initial expense of e0
ˆ Each policy year, including the first, an expense of e1 + e2 Px
ˆ The cost of claims settlement, paid along with the claim, of e3 per unit of insurance.

If G = aPx + c, determine a and c.


2. [Solution] A life aged 35 purchases a 30-year endowment assurance policy where level premiums are
payable monthly in advance throughout the duration of the contract. The sum assured of $250, 000 is
payable at maturity or at the end of the year of death, if earlier.Calculate the monthly premium using
thefollowing basis:

Basis:
Mortality: AM92 select
Interest: 4% per annum
Initial expenses: $250 plus 50% of the gross annual premium
Renewal expenses: 3% of the second and subsequent monthly premiums
Claims expenses: $300 on death; $150 on maturity

3. [Solution] Consider a 10-year endowment insurance policy with sum insured = 1000, issue age x = 40,
De Moivre’s mortality with ω = 100, and i = 4%.

(a) The acquisition expense is 50. No other expenses are recognized. Calculate the expense-loaded
premium and the expense-loaded premium reserves for each policy year.
(b) Determine the maximum value of acquisition expense if negative expense-loaded reserves are to
be avoided.

4. [Solution] A life office issues to a proposer aged 35 a whole life policy participating in profits for a sum
assured of $10, 000. The sum assured and bonuses are payable immediately on death.

(a) Calculate the monthly premium payable for a maximum of 30 years, if the office assumes that
future bonuses will be at a rate of 1.92308% of the sum assured, compounded annually and vesting
at the start of each policy year starting from the 2nd policy year.
Basis:
ˆ Mortality: AM92 Select
ˆ Interest: 6% per annum
ˆ Expenses: none

50
(b) The office also prepared a new type of policy participating in profits that provides both family
income benefit and whole life death benefit. Under this scheme, on death within 20 years, monthly
installments of $250 each month would be payable, the first installment being due on the date
of death. The installments would continue for the remainder of the 20 year period. The sum
assured of $10, 000 is payable immediately on death. Future bonuses will be at a rate of 1.92308
of the sum assured, compounded annually and vesting at the start of each policy year starting
from the 2nd policy year. Bonuses are also paid immediately on death. Calculate the monthly
premium payable for a maximum of 20 years of this policy.

Basis:
Mortality: AM 92 Select.
Interest: 6% per annum.
Initial expenses: $50 plus 2% of the basic sum assured.
Renewal expenses: $10 on each annual anniversary of the date of issue of the policy throughout
the life of the policyholder, 2 21 % of each premium after the first monthly premium.

Zillmerized reserves
5. [Solution] On 1 January 2000, a life insurance company issued a 20-year annual premium without profits
endowment assurance policy to a life then aged exactly 40, which is still in force. The sum assured of
$100, 000 is payable at the end of the year of death within the term of the policy, or on survival. The
company values the policy using a modified net premium method, with a Zillmer adjustment.

(a) Calculate the reserve for the policy on 31 December 2009.


Basis:
Mortality: AM 92 Select.
Interest: 4%.per annum.
Zillmer adjustment: 2% of the sum assured.
(b) Without carrying out further calculations, explain how the value of the policy would differ if
the company used a Zillmer adjustment of 1% of the sum assured, with the same mortality and
interest assumptions.

6. [Solution] On 1 January 2000, a life insurance company issued an endowment assurance policy to a life
aged exactly 50 for a term of 10 years.
Under the policy, a sum assured of $100,000 is payable on survival to age 60 exact or at the end of the
year of death if earlier. Level premiums are payable annually in advance for 10 years or until earlier
death.
On 1 January 2003, the policy is still in force and the life insurance company calculates on a prospective
basis both the gross premium reserve and the net premium reserve for the policy at this date, using
the assumptions shown below. The same assumptions were used to calculate the gross premium at
inception as follows:
Mortality: AM92 ultimate
Interest: 4% per annum
Initial expenses: $300 incurred at the outset
Renewal expenses: 5% of each premium

(a) Calculate the gross premium reserve as at 1 January 2003.


(b) Calculate the net premium reserve, with Zillmer adjustment, as at 1 January 2003. Identify clearly
the Zillmer adjustment.
(c) Explain why the net premium reserve with Zillmer adjustment calculated in part (b) might be
used in preference to the net premium reserve with no Zillmer adjustment, calculated as at 1
January 2003, using the same assumptions.

51
Surrender values
7. [Solution] A life aged 40 purchases a 25 year endowment assurance contract. Level quarterly premiums
are payable throughout the duration of the contract. The sum assured $100, 000 is payable at maturity
or at the end of the year of death.

(a) Calculate the quarterly premium using the following basis:


Basis:
Mortality: AM92 select
Interest: 4% p.a.
Initial expenses: $250 plus 60% of the gross annual premium
Renewal expenses: 3% of the second and subsequent quarterly premiums
Claims expenses: $500 on death; $100 on maturity
(b) At age 60, immediately before the premium then due, the life wishes to surrender the policy. The
insurer calculates a surrender value equal to the gross retrospective policy value, assuming the
same basis as in (a) above, but with 3% p.a. interest.
i. Calculate the surrender value.
ii. Explain with reasons whether the surrender value would have been larger, the same or smaller
than in (b) i above if the office had used the prospective gross premium policy value, on the
same basis.

8. [Solution] A life insurance policy generally includes a clause in the contract which states that “if the
age of the insured is misstated at issue, but later discovered, the benefit under the contract will be
adjusted, as of issue, to whatever the premium actually paid would purchase at the correct age”.
A life office issues a 10-year endowment insurance policy to an applicant whose stated age is 30 exact.
The sum insured of $500,000 is payable on survival to the end of the term or immediately upon death,
if earlier.
The office prices the policy using the following basis:

ˆ Mortality: AM92 Select


ˆ Interest: 4% per annum
ˆ Initial expenses: 50% of the first premium plus 1/2% of the sum insured
ˆ Renewal expenses: 1.5% of the second and subsequent monthly premiums

(a) Calculate the monthly premium payable in advance throughout the term of the policy, assuming
the correct age of the insured is 30 exact.
(b) Assume that the insured survived to reach 10 years, in which case, he is now eligible to receive the
endowment benefit. However, while reviewing the policy, the life office discovers that the insured
at the time of the application was actually 35 years old. Calculate the amount of the adjusted
endowment benefit to be paid to the insured.

The portfolio percentile premium principle


9. Dickson et al. (2009a) Exercise 6.15 (Dickson et al. (2013a) Exercise 6.20)
10. [Solution] A life insurance company issues life annuities to males aged exactly 60. The annuities are
single premium annuities with annual payments of $20,000 in arrears (at the end of the year). Assume
PMA92C20 with 4% interest.

(a) Show that the single premium, P, is $292,640.


(b) Give an expression for the random variable given by the present value of the profit of the life
annuity (premium less annuity payments). Determine the probability that the present value of
the profit on a policy will be positive.

52
(c) Show that the standard deviation of the present value of the profit on a policy is $74,422.
(d) Assume the life insurance company sells 1,000 of these policies to independent male lives aged 60.
(i) Determine the profit loading so that the probability that the present value of the profit to the
insurance company will be positive is 95%.
(ii) If the life insurance company charged a profit loading of 10%, determine the probability of a
positive profit.

53
Solutions to Module 5 Exercises
1. [Question] The only work required here is writing out the APV of future expenses which is given by

APVFE = e0 + (e1 + e2 · Px ) · äx + e3 · Ax .

By the principle of equivalence, we have

APVFP = APVFB + APVFE


G · äx = Ax + e0 + (e1 + e2 · Px ) · äx + e3 · Ax
= e0 + (e1 + e2 · Px ) · äx + (1 + e3 ) · Ax

which implies the expense-loaded premium is

e0 + (e1 + e2 · Px ) · äx + (1 + e3 ) · Ax
G =
äx
= e0 · (Px + d) + (e1 + e2 · Px ) + (1 + e3 ) · Px
= (1 + e0 + e2 + e3 ) · Px + (e1 + de0 )

so that clearly, a = (1 + e0 + e2 + e3 ) and c = (e1 + de0 ).

2. [Question] Denote by G the monthly premium for the contract. Then,


(12)
APV(Premiums) = 12Gä
[35]:30|
  
11 D65
= 12G ä[35]:30| − 1−
24 D[35]
  
11 689.23
= 12G 17.631 − 1−
24 2507.02
= 207.5841G.

and

APV(Benefits + Expenses)
(12)
= 250300A[35]:30| + 250 + 0.5 · 12G + 0.03 · 12Gä
[35]:30|
−0.03G − 150 30 E[35]
= 250300 · 0.32187 + 250 + 6G + 0.03 · 12G · 17.298675
689.23
−0.03G − 150 ·
2507.02
= 80772.82 + 12.19752G.

Equating the two APVs and solving for the monthly premium, we have
80772.82
G= = 413.4001.
195.3866

3. [Question] It is not stated in the problem whether this is fully-discrete, fully continuous, or otherwise.
So, we make the assumption that this is a fully-discrete 10-year endowment policy. Note that for issue
age x = 40, we have under De Moivre’s law, that

`40+k 60 − k
k p40 = = .
`40 60

(a) Let G be the expense-loaded (or gross) annual premium. Thus, we have, by equating APVFP
with APVFB+APVFE,
G · ä40:10 = 1000A40:10 + 50

54
where
9 9
X 1 X k
ä40:10 = v k ·k p40 = v · (60 − k)
60
k=0 k=0
1
= ä10 − (Ia)9 = 7.848055
60
and
A40:10 = 1 − dä40:10 = 0.6981517
so that
1000A40:10 + 50 1000 (0.6981517) + 50
G= = = 95.329576.
ä40:10 7.848055
For year k = 1, 2, ..., 9, the gross premium reserve is given by

kV = 1000A40+k:10−k − G · ä40+k:10−k
= 1000A40+k:10−k − 95.329576 · ä40+k:10−k

= 1000 1 − dä40+k:10−k − 95.329576 · ä40+k:10−k
= 1000 − (1000(0.04/1.04) + 95.329576) ä40+k:10−k
= 1000 − 133.791114ä40+k:10−k
9−k
!
X
s
= 1000 − 133.791114 v ·s p40+k
s=0
9−k  !
X
s s
= 1000 − 133.791114 v · 1−
s=0
60 − k
 
1
= 1000 − 133.791114 ä10−k − (Ia)9−k .
60 − k
The values are summarized below:
k 1 2 3 4 5 6 7 8 9
kV 30.99 116.40 206.52 301.67 402.20 508.50 620.96 740.04 866.21

(b) If we let a be the acquisition expense, in general, the gross annual premium will be
1000A40:10 + a 1000 (0.6981517) + a a
Ga = = = 88.9585702 +
ä40:10 7.848055 ä40:10
and the gross premium reserve will be

kV = 1000A40+k:10−k − Ga · ä40+k:10−k
ä40+k:10−k
 
= 1000A40+k:10−k − 88.9585702 · ä40+k:10−k − a
ä40:10
ä40+k:10−k
 

= 1000 1 − dä40+k:10−k − 88.9585702 · ä40+k:10−k − a

 40:10
ä40+k:10−k

= 1000 − (1000(0.04/1.04) + 88.9585702) ä40+k:10−k − a
ä40:10
ä40+k:10−k
 
= 1000 − 127.420109ä40+k:10−k − a .
ä40:10
Reserves will be negative if it ever hits negative at the end of the first year. This is because reserve
is increasing obviously (because of the nature of the endowment insurance - not all reserves do
increase over time) with time and if it ever hits negative, it must be in the first year. Therefore,
we must have  
ä41:9
1000 − 127.420109ä41:9 − a >0
ä40:10

55
or equivalently,

1000 − 127.420109ä41:9
 
a < ä40:10
ä41:9
 
1000 − 127.420109(6.69627266)
= 7.848055
6.69627266
= 83.5830964.

4. [Question] Denote by P the annual premium payable monthly.

(a) Unless otherwise stated, present values are at 6% interest. Assume UDD holds. The APVFP at
issue is given by
    
(12) 11  11 30 `65
P · ä[35]:30 = P ä[35]:30 − 1 −30 E[35] = P ä[35]:30 − 1−v
24 24 `[35]
  
11 8821.2612
= P 14.352 − 1 − 0.17411 · = 13.965P
24 9892.9151

The APVFB at issue, on the other hand, is given by


h i
K
APVFB0 = 10000 × E v T (1 + b)

where bonuses are, by assumption, added and vested at the start of each policy year. Note that
bonus does not start until after the first policy year. Now, set the interest rate

i−b 6% − 1.92308%
j= = = 4%
1+b 1.0192308
Now writing
h i h i h i
K K K
E v T (1 + b) E (v (1 + b)) v S = E (v (1 + b)) · E v S
 
=
1 h
K+1
i
· E vS
 
= · E (v (1 + b))
v(1 + b)
1 d 1 i j
= · Aj[35] · = A
v(1 + b) δ (1 + b) δ [35]
1 0.06
= Aj = 1.01028Aj[35]
(1.0192308) ln(1.06) [35]

Note that in the second line we have equated E[v S ] to dδ . This holds true because we can evaluate
R1 R1
the expectation as E[v S ] = 0 v s ds = 0 e−δs ds
= δ 
1
(1 − e−δ ) = 1δ (1 − v) = dδ . Also, we have
h i K+1
K+1
transformed the quantity E (v (1 + b)) =E 1+j 1
=Aj[35] .
Now we have the APV of future benefits as

APV of benefits = 10000 (1.01028) Aj[35]

where, from the AM92 Select table, we have

Aj[35] = 0.19207.

Hence, finally, equating the two APV’s, we derive the annual premium:

13.965P = 10000 (1.01028) Aj[35] = 1940.445

Therefore, we have P = 138.952 so that the monthly premium is P/12 = 11.58.

56
(b) Let P ∗ be the (annual) premium for this new policy. To find out the new premium, we can apply
the equivalence principle which requires us to find the present value of all the death benefits and
family income.
First, we calculate the family income benefit. Since the first installments is payable at the date of
death, it is hard to quantify the present value. We first consider a simplified situation where the
installments are payable at the end of the month following death. Considering that the monthly
installment is 250, the annual installment is 250 ∗ 12 = 3000, and the family income benefit is
given by  
(12) (12)
APV (FIB) = 3000 a20 − a[35]:20
Note that result can also be derived mathematically from basic principles.

239
(12)
X
APV (FIB) = 3000 v j/12 · a × (j/12)|(1/12) q[35]
(240−j)/12
j=0
239  j/12
− v 20

X v
= 3000 × (j/12)|(1/12) q[35]
j=0
i(12)
239  j/12
− v 20 + 1 − 1

X v
= 3000 × (j/12)|(1/12) q[35]
j=0
i(12)
 
239
(12) (12)
X
= 3000 a20 × 20 q[35] − a × (j/12)|(1/12) q[35]

j/12
j=0
  
239
(12) (12) (12)
X
= 3000 a20 −  a × (j/12)|(1/12) q[35] + a20 ·20 p[35] 
j/12
j=0
 
(12) (12)
= 3000 a20 − a[35]:20

Now we have calculated the family income benefit assuming that the first installment is paid at
the end of the month following death. However, the first installment for this policy is actually due
on the date of death, hence using UDD assumption we assume that deaths occur in the middle of
the month following monthly anniversary. In this case, to adjust our above calculations, we need
to adjust the cash flows from the end of each month to the middle of the month. Therefore, we
account for the lost interest earned in half a month using
 
1/24 (12) (12)
APV (FIB) = (1 + i) · 3000 a20 − a[35]:20

where we have
1/24
(1 + i) = 1.002431,
(12) i
a20 = a = 1.027211(11.4699) = 11.78201,
i(12) 20
and
(12) (12) (12) `55
a[35]:20 = a[35] − a55 · v 20
`[35]
   
(12) 1 (12) 1 `55
= ä[35] − − ä55 − · v 20
12 12 `[35]
     
11 1 11 1 `55
≈ ä[35] − − − ä55 − − · v 20
24 12 24 12 `[35]
   
13 13 9557.8179
= 15.993 − − 13.057 − · 0.31180
24 24 9892.9151
= 11.6812322.

57
From part (a), we have
1 i j
APV (Death benefits + bonuses) = 10000 A = 1940.445,
(1 + b) δ [35]
where j = 4%.

We also have
(12) (12) 1 1 `55 1 1 9557.8179
ä[35]:20 = a[35]:20 + − v 20 = 11.6812322 + − 0.31180 = 11.41312
12 12 `[35] 12 12 9892.9151
Using principle of equivalence we have
(12)

(12) (12)
 1 i j
P ∗ ä[35]:20 = 3000 · (1 + i)1/24 a20 − a[35]:20 + 10000 A + 240 + 10ä[35]
(1 + b) δ [35]
(12) 0.025P ∗
+ 0.025P ∗ ä[35]:20 −
12
which gives
 
(12) 0.025 
(12) (12)
 1 i j
0.975ä[35]:20 − P ∗ = 3000 · (1 + i)1/24 a20 − a[35]:20 + 10000 A + 240 + 10ä[35] ,
12 (1 + b) δ [35]
 
0.025
0.975(11.13699) − P ∗ = 3000 · 1.002431 (11.78201 − 11.6812322) + 1940.445 + 240 + 10(15.993),
12
10.85648P ∗ = 3624.850
Hence
3624.850
P∗ = = 333.8881
10.85648
or a monthly premium of P ∗ /12 = 27.8240.

5. [Question] Recall that the Zillmerized reserve has the form of the “net premium reserve” reduced by
the amortization of the Zillmer adjustment (usually equivalent to the first year expense).

(a) We first calculate the net premium reserve. In this case, it has been calculated using the ratio of
annuities formula which can be found on page 3 of Module 4.2.1, which is
!
ä50:10
100000 · 1 − .
ä[40]:20

Then we apply the Zillmer adjustment. Recall that the Zillmer adjustment is
äx+t:n−t|
I
äx:n|

where I is the Zillmer expense (2% of $100, 000).

The policy reserve as at 31 Dec 1999 is given by


!
ä50:10 ä50:10
10 V = 100000 · 1− − 0.02 × 100000 ·
ä[40]:20 ä[40]:20

where
ä[40]:20 = 13.929 and ä50:10 = 8.313.
Therefore, we have
 
8.313 8.313
10 V = 100000 · 1 − − 2000 ·
13.929 13.929
= 39125.13

58
Remark: Many students understand Zilmer adjustments as I∗ a ratio of annuities. But the
question says the Zilmer adjustment is 2% of the sum assured. Sometimes, Zilmer adjustment
refers to the I instead of I∗ a ratio of annuities. A simple way to decide whether it refers to I
or I∗ a ratio of annuities, is to see whether it varies with the time of the reserve. The term I∗
a ratio of annuities changes as the time changes (the numerator annuity depends on the time;
it is the annuity starting from the time for the reserve until the end of the policy), which means
that the adjustment reserves for different times need to be adjusted differently, instead of just
taking a constant amount for any year of time. In this case, the Zilmer adjustment described in
the question is 2% of the sum assured, which is a constant regardless of time, and therefore is I
instead of I∗ a ratio of annuities.
(b) Using a Zillmer adjustment has the effect of reducing the policy reserve because this is a reduction
to the net premium reserve, assuming of course the same mortality and interest basis. Changing
the Zillmer adjustment from 2% of the sum insured to 1% of the sum insured therefore has the
effect of reducing the amount of the Zillmer adjustment, and hence, increasing the policy reserve
as at 31 Dec 1999.
6. [Question] On 1 January 2000, a life insurance company issued an endowment assurance policy to a
life aged exactly 50 for a term of 10 years.
(a) Denote by G the gross premium so that we have
0.95Gä50:10 = 300 + 100000A50:10 .
From the table, we have ä50:10 = 8.314 and A50:10 = 0.68024 so that the gross premium is
G = 8, 650.47.
The (3rd year) gross premium reserve is therefore
GP reserve = 100000A53:7 − 0.95Gä53:7 .
Noting that ä53:7 = 6.166 and A53:7 = 0.76286, we have
GP reserve = 25, 614.14.
(b) The net premium reserve with the Zillmer adjustment can be computed as
!
ä53:7 ä
100000 1 − − 300 · 53:7 = 25, 835.94 − 222.49
ä50:10 ä50:10
= 25, 613.45.
Thus, 222.49 refers to the Zillmer adjustment. Indeed this is the difference between the gross
premium reserves and the net premium reserves.
(c) The net premium reserve with Zillmer adjustment equals the gross premium reserve calculated in
part (a) (subject to rounding errors). If the insurance company actuary is satisfied that there are
sufficient margins in the gross premium reserve then the net premium reserve with the Zillmer
adjustment would be adequate. In addition, the use of the net premium reserve with Zillmer
adjustment compared with the use of the reserve without adjustment would reduce the company’s
funding requirements.
7. [Question] We denote the quarterly premiums by P .
(a) To find the premium, we equate the APVFP to APVFB at issue. We have
(4)
APVFP = 4P · ä[40]:25
  
3 D65
≈ 4P ä[40]:25 − 1−
8 D[40]
  
3 689.23
= 4P 15.887 − 1−
8 2052.54
= 62.55169055P,

59
for the APV of benefits and claims expenses, we have

D65
APVFB = 100, 500A[40]:25 − 400
D[40]
689.23
= 100, 500 · 0.38896 − 400 ·
2052.54
= 38, 956.16,

and for the APV of other expenses, we have

APVFE = 0.03 × 62.55169055P − 0.03P + 0.6 × 4P + 250


= 4.246551P + 250.

Thus, we have the quarterly premium equal to


39206.16
P = = 672.43.
58.3399

(b) The cash surrender value is computed retrospectively.


i. Thus, we have cash surrender value at the end of 20 years,

(4) 1 1 1
CSV20 = 4P × 0.97 × ä[40]:20 × + 0.03P × − 0.6 × 4P ×
E
20 [40] E
20 [40] E
20 [40]
1 1
−250 × − 100, 500A1[40]:20 ×
E
20 [40] E
20 [40]

where everything is computed at 3%. We can re-write this as


1 h
(4)
i
CSV20 = · 2609.03ä[40]:20 − 1843.66 − 100, 500A1[40]:20 .
20 E[40]

Unfortunately, the Formulae book does not provide the needed insurance/annuity values at
3% from the AM92 select tables.
ii. Retrospectively: the lower interest rate values accumulated premium less benefits as if they
had earned only 3% instead of the required 4%.

Prospectively: the lower interest rate indicates a lower rate to be earned in the future. This
implies that more reserves is required now in order to meet future liabilities.
So the prospective reserve at 3% > the prospective reserve at 4% = the retrospective reserve
at 4%> the retrospective reserve at 3%

8. [Question]

(a) Denote by G, the gross annual premium payable monthly. The APV of future benefits and
expenses is given by

G (12)
AP V F B = 500000 Ā[30]:10 + 0.485 + 0.005 × 500000 + 0.015Gä[30]:10
12
The APV of future premiums is given by
(12)
AP V F P = Gä[30]:10

Equating these two APVs and solving for G gives

500000 Ā[30]:10 + 2500


G= (12) G
0.985ä[30]:10 − 0.485 12

60
Now, we calculate the continuous assurance and annuity values using AM92:
(1)
Ā[30]:10 = Ā[30]:10 + 10 E[30]
i 
= A[30] − 10 E[30] A40 + 10 E[30]
δ  
0.04 2052.96 2052.96
= 0.16011 − × 0.23056 +
ln 1.04 3059.68 3059.68
= 0.67649
 
(12) 12 − 1 D40
ä[30]:10 = ä[30]:10 − 1−
2 × 12 D[30]
 
11 2052.96
= ä[30] − 10 E[30] ä40 − 1−
24 3059.68
 
2052.96 11 2052.96
= 21.837 − 20.005 × − 1−
3059.68 24 3059.68
= 8.263398

D40 2052.96
Note that we used 10 E[30] = D[30] = 3059.68 . Finally, we substitute the values to obtain:

500000 × 0.67649 + 2500


G= = 42072.34
0.985 × 8.263398 − 0.485/12
Then, the monthly premium is 42072.34/12 = 3506.03.
(b) Let the revised benefit be B ∗ . According to the clause in the contract, we simply apply the
equivalence principle at issue using age 35 using the gross premiums calculated above. Thus, the
revised APV of future benefits and expenses is:
G (12)
B ∗ Ā[35]:10 + 0.485 + 0.005B ∗ + 0.015Gä[35]:10
12
The revised APV of future premiums is:
(12)
Gä[35]:10

We then equate these and solve for B ∗ to obtain:


(12) G
0.985Gä[35]:10 − 0.485 12

B =
Ā[35]:10 + 0.005

We must now recalculate the continuous assurance and annuity values using age 35:
i 
Ā[30]:10 = A[35] − 10 E[35] A45 + 10 E[35]
δ  
0.04 1677.97 1677.97
= 0.19207 − × 0.27605 +
ln 1.04 2507.02 2507.02
= 0.676761
 
(12) 12 − 1 D45
ä[35]:10 = ä[35]:10 − 1−
2 × 12 D[35]
 
11 1677.97
= ä[35] − 10 E[35] ä45 − 1−
24 2507.02
 
1677.97 11 1677.97
= 21.006 − 18.823 × − 1−
2507.02 24 2507.02
= 8.256038

61
Finally, we substitute into the equation to obtain B ∗ .
42072.34

∗ 0.985 × 42072.34 × 8.256038 − 0.485 12
B = = 499354
0.676761 + 0.005
9. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.

10. [Question]

(a) Single premium is

P = 20, 000a60
= 20, 000 (ä60 − 1)
= 20, 000 (15.632 − 1)
= 292, 640

(b) The random variable for the present value of the profit is

P − 20, 000aK for death at age 60 + K, K = 0, 1, 2, ...

Probability profit will be positive is


 
Pr 292, 640 − 20, 000aK ≥ 0
 
= Pr aK ≤ 14.632
" #
−K
1 − (1.04)
= Pr ≤ 14.632
0.04
= Pr [K ln (1.04) ≤ − ln (0.41472)]
= Pr [K ≤ 22.441]
= P r(K ≤ 22)
= P r(T < 23).

Required prob is
l83 5772.378
1− = 1−
l60 9826.131
= 0.41255

(c) Standard deviation of present value of profit is square root of variance

20, 0002 V ar aK
 

= 20, 0002 V ar äK − 1


 

1 − vK
 
= 20, 0002 V ar
d
2
20, 000
V ar v K
 
= 2
d
20, 0002
V ar v K+1
 
=
d2 v 2
20, 0002 h 2
i
= 0.17950 − (1 − dä60 )
d2 v 2
2
20, 000 h
2
i
= 0.17950 − (1 − 0.0384615.15.632)
0.03846152 × 0.96153852
= 5, 990, 582, 642

Standard deviation is 77,399.

62
(d) Denote by P ∗ the premium calculated such that the probability of a positive profit is 95%. The PV
of profit X is approximately normally distributed with mean 1000 (P ∗ − 292, 640) and variance
1000 × 77, 3992 .
−1000 (P ∗ − 292, 640)
 
Pr [X > 0] = Pr Z > √
77, 399 1000
and we require to solve for P ∗ such that

Pr [Z > −1.6449] = 0.95

We then have
1000 (P ∗ − 292, 640)
√ = 1.6449
77, 399 1000
so that
77, 399
P ∗ = 292, 640 + √ × 1.6449 = 296, 511
1000
This means that the loading required for a 95% probability of a positive profit is 296, 511 −
292, 640 = 3, 871.
Now, a 10% loading would be a premium of 1.1 (292, 640) = 321, 904 which would give

1000 (321, 904 − 292, 640)


Z= √ = 11.956
77, 399 1000
so almost certain that profit will be positive.

63
Module 6: Profit Testing

Cash-flow projection of conventional products


1. Dickson et al. (2009a) Exercise 11.3 (a) (Exercise 12.6 (a) - Dickson et al., 2013a)
2. [Solution] A special 5-year policy is issued to a life aged 60. The death claim (payable at the end of
the year of death) is $100,000 and on maturity there is a survival benefit of $10,000. Level annual
premiums are payable in advance. Reserves are held by the office equal to the net premium policy
value using an interest rate of 4% and the AM92 ultimate life table for mortality.

(a) Calculate the level annual premium using the principle of equivalence.
(b) Show that the reserve at the end of the first year is $2010.99 (assuming the policy is still in force)
and calculate the reserves at the end of each subsequent year assuming the policyholder survives.
(c) The office profit tests the contract assuming initial expenses of 60% of the premium plus $50,
renewal expenses of $50, an interest rate of 7% and the AM92 select life table for mortality. It
is incorporating the reserves found in (b), using the premium found in (a). Calculate the profit
signature.

Cash-flow projection for unit-linked products


3. [Solution] A life office issues a 4-year unit linked endowment policy to a life aged 50 exact under which
level premiums of $2, 000 per annum are payable in advance. In the first year, 25% of the premium is
allocated to units and 102.5% in the second and subsequent years. There is a bid/offer spread in the
prices of the units of 5% of the offer price. A management charge of 0.5% per annum of the bid value
of the units is deducted at the end of each year.
If the policyholder dies during the term of the policy, the death benefit of $5, 000 or the bid value of
the units after the deduction of the management charge, whichever is higher, is payable at the end of
the year of death. On surrender or on survival to the end of the term, the bid value of the units is
payable at the end of the year of exit.
The company uses the following assumptions in its profit test of this contract:

Rate of interest on unit investments: 8% per annum


Rate of interest on non-unit fund cashflows: 6% per annum
Independent rates of mortality: AM92 Select
Independent rates of withdrawal: 10% of policies in force,
uniformly distributed over the year
Initial expenses: $150 plus 20% of first premium
Renewal expenses: $50 per annum on the second and
subsequent premium dates
Initial commission: 25% of first year premium
Renewal commission: 2.5% of the second and
subsequent premiums
Risk discount rate: 12% per annum

64
Project end-of-the-year unit funds and non-unit fund cash flows.

4. [Solution] A three-year unit-linked policy is issued on the following terms:

ˆ premiums of $1, 000 are payable annually in advance;


ˆ the allocation proportion is 50% in the first year and 100% in the following two years;
ˆ the bid/offer spread is 5%;
ˆ the management charge, which is deducted at the end of the year before payment of the death
benefit, is 0.5% of the bid price of the unit fund;
ˆ the death benefit, which is payable at the end of the year of death, is the bid price of the units or
$1, 500, whichever is greater; and
ˆ the maturity proceeds are the value of the bid price of the units.

The company issuing the policy estimates the profit signature on the following (conservative) assump-
tions:

ˆ the units will grow at 3% p.a.;


ˆ the interest rate for reserves is 3% p.a.;
ˆ there are no withdrawals and the probability of death in any given year is 0.008; and
ˆ the expenses are $400 at the start of the first year and $100 at the start of each of the following
two years.

Project end-of-the-year unit funds and non-unit fund cash flows.

Profit measures
5. Dickson et al. (2009a) Exercise 11.3 (b), (c), (d) (Exercise 12.6 (b), (c), (d) - Dickson et al., 2013a)

6. [Solution] A life insurance issues an endowment insurance with a term of five years to a life aged exactly
55. The sum insured is $100, 000, payable at the end of the five years, or at the end of the year of
death, if earlier. Premiums are payable annually in advance throughout the term of the policy.
The insurer assumes that initial expenses will be $300 and renewal expenses, which are incurred at the
beginning of the second and subsequent years of the policy, will be $30 plus 2.5% of the premium. The
funds invested for the policy are expected to earn 7.5% p.a., and mortality is expected to follow the
AM92 Select life table. The company holds net premium reserves, calculated using AM92 Ultimate
mortality and interest of 4% p.a.
The insurer sets premiums so that the net present value of the profit on the contract is 15% of the
annual premium, using a risk discount rate of 12% p.a.

(a) Calculate the annual premium.


(b) Without carrying out any further calculations, state with brief reasons what the effect on the
premium would be in each of the following cases:
(i) The reserves are calculated using a lower rate of interest;
(ii) The insurer uses a risk discount rate of 15%; and
(iii) Mortality is assumed to be AM92 Ultimate.

7. [Solution] A life insurance company issues a 4-year unit-linked policy to a life aged 50 exact under
which level premiums of $2,000 per annum are payable in advance. The following non-unit cash flows,
NUCFk (k = 1,2,3,4), are obtained at the end of each year k per policy in force at the start of the year k:

65
Year k 1 2 3 4
N U CFk 497.659 -43.199 -25.777 -13.205

The company uses the following assumptions in its profit test of this contract:

Independent rates of mortality: AM92 Select


Independent rates of withdrawal: 10% of all policies still in force at the end
of the first and subsequent years
Risk discount rate: 12% per annum
Calculate the profit margin on the assumption that the company does not zeroize future negative
cashflows.

8. [Solution] A life office issues a 5-year unit-linked policy to a life aged 60 exact with annual payment of
$1, 500. Assume a 100% allocation percentage in each year.
Policyholders purchase units from the office at the offer price and the life office buys back units at the
beginning of each year’s bid price in the case of earlier death or at maturity. Benefits are paid at the
end of the year of death or at maturity. Initial expenses are $1, 200 and renewal expenses are $100
excluding the first per annum.

Basis: AM92 Ultimate. Interest 4%. Annual management charge: $200. Bid/Offer spread 10%.

(a) Calculate the profit made at the end of each of the five years provided the unit trust is in force
at the beginning of the year, assuming that the unit fund grows at 10% per annum before the
deduction of management charge.
(b) Calculate the net present value of the profit for a 5-year unit trust using a risk discount rate of
15% per annum.
(c) Calculate the profit margin using a risk discount rate of 15% per annum.

Zeroisation
9. [Solution] A life insurance company issues 4-year unit-linked contracts to a male aged 50 exact. The
following non-unit fund cash flows, NUCFk, (k = 1, 2, 3, 4) are obtained at the end of each year k per
contract in force at the start of the year k:

Year k 1 2 3 4
N U CFk 375.4 -152.0 -136.2 -118.0

The rate of interest earned on non-unit reserves is 5.5% per annum and mortality follows the AM92
Select table. Calculate the reserves required at times k = 1, 2 and 3 in order to zeroise future negative
cash flows.

10. [Solution] A three-year unit-linked policy is issued to a male under which premiums of $1, 000 are
payable annually in advance; the interest rate is 3% p.a.; there are no withdrawals and the probability
of death in any given year is 0.008. The following non-unit cash flows, NUCFk (k = 1,2,3), are obtained
at the end of each year k per policy in force at the start of the year k:

Year k 1 2 3
N U CFk 123.091 -44.320 -39.024

(a) Calculate the profit signature of the policy, assuming that no reserves are held at the end of each
policy year.

66
(b) Find the zeroized profit signature.
(c) Giving reasons for your answer, state the effect on the zeroized profit signature of an increase in
the assumed rate of growth of unit prices.

Universal life insurance


11. [Solution] Dickson et al. (2013a) Excercise 13.10
A life insurance company issues a four-year universal life policy to (65). The main features of the
contract are as follows:

ˆ Premiums: $3000 per year, payable yearly in advance.


ˆ Expense charges: 4% of premium is deducted at the start of the first year; $100 plus 0.4% of the
account value (before premium) is deducted at the start of each subsequent policy year.
ˆ CoI: $25 is deducted from the account value at the start of each year.
ˆ Death benefit: greater of $12 000 or 1.5 times the account value at year end.
ˆ Maturity benefit: 100% of the account value.

The company uses the following assumptions in carrying out a profit test of this contract.

ˆ Interest rate: 4.5% per year in year 1, 5.5% per year in year 2, and 6.5% per year in years 3 and
4.
ˆ Credited interest: Earned rate minus 1%, with a 4% minimum.
ˆ Survival model: Standard ultimate survival model.
ˆ Withdrawals: None.
ˆ Initial expenses: $200 pre-contract expenses.
ˆ Renewal expenses: payable annually at each premium date, initial cost (with first premium) $50,
increasing with inflation of 2% per year.
ˆ Risk discount rate: 8% per year.

There are no reserves held other than the account value.

(a) Calculate the profit signature and NPV of a newly issued contract.
(b) Calculate the profit signature and NPV for the policy given that the policyholder dies in the first
year of the contract.
(c) Calculate the profit signature and NPV for the policy given that the policyholder survives to the
contract end.
(d) Calculate the profit signature and NPV for the policy given that the policyholder surrenders at
the end of the second year, assuming
(i) that the cash value is 100% of the year end account value
(ii) that the cash value is 90% of the year end account value
(e) Calculate the surrender penalty at time 2, as a proportion of reserves at time 2, which gives
the same profit margin for surrendering policyholders as for policyholders who remain in force
throughout.
(f) Comment on your results

67
Solutions to Module 6 Exercises
1. Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.
2. [Question] The premium is calculated as follows:

P ä60:5 = 100000A160:5 + 100005 E60


 
= 100000 A60:5 − 5 E60 + 100005 E60
 
D65 D65
= 100000 A60:5 − + 10000
D60 D60
 
689.23 689.23
4.55P = 100000 0.82499 − + 10000 ×
882.85 882.85
P = 2689.48

The reserve may be calculated recursively using:

(h V + P ) (1 + i) − 100000qx+h
h+1 V =
px+h

Substituting these values yields:

0V =0
(0 + 2689.48) (1.04) − 100000 × 0.008022
1V = = 2010.99
1 − 0.008022
(2010.99 + 2689.48) (1.04) − 100000 × 0.009009
2V = = 4023.83
1 − 0.009009
(4023.83 + 2689.48) (1.04) − 100000 × 0.010112
3V = = 6031.64
1 − 0.010112
(6031.64 + 2689.48) (1.04) − 100000 × 0.011344
4V = = 8026.61
1 − 0.011344
5V = 10000

Finally, we tabulate the projection of cash flows as follows:

Year 1 2 3 4 5
[1] Initial Reserves 0 2010.99 4023.83 6031.64 8026.61
[2] Premiums 2689.48 2689.48 2689.48 2689.48 2689.48
[3] Expenses 1663.69 50 50 50 50
[4] Subtotal [1]+[2]-[3] 1025.79 4650.47 6663.31 8671.12 10666.09
[5] Interest 1.07[4] 71.81 325.53 466.43 606.98 746.63
[6] Probability of death 0.005774 0.00868 0.010112 0.011344 0.012716
[7] Expected death claims 100000[6] 577.4 868 1011.2 1134.4 1271.6
[8] Probability of survival 1 - [6] 0.994226 0.99132 0.989888 0.988656 0.987284
[9] Expected end of year reserves 1999.38 3988.9 5970.65 7935.56 9872.84
[10] Profit [4]+[5]-[7]-[9] -1479.18 119.1 147.89 208.14 268.28
[11] Probability of survival t years 1 0.994226 0.985596 0.97563 0.964562
[12] Profit signature [10]*[11] -1479.18 118.41 145.76 203.07 258.77

Note that [6] is simply qx under the AM92 Select table (so it is q[60] in the first year, q[60]+1 in the
next, q62 after that, etc.). [9] is calculated as [8]×t+1 V . [11] is simply t px .

68
3. [Question] First, we need to convert the independent decrements to dependent decrements. From the
AM92 Select table and a 10% yearly withdrawal rate, we have

x qxd qxw
50 0.001971 0.10
51 0.002732 0.10
52 0.003152 0.10
53 0.003539 0.10

where decrement d refers to death and w withdrawal or surrender. Next, assuming uniform distribution
of decrement, we know that the corresponding dependent decrement gives
 
1
(aq)dx = qxd · 1 − qxw
2

and
 
1 d
(aq)w
x = qx
w
· 1 − q .
2 x

And so we have
(ap)x = 1 − (aq)dx − (aq)w
x

which gives the probability of survival (i.e. not death nor withdrawal) during the year. We thus have

x (aq)dx (aq)w
x (ap)x t−1 (ap)x
50 0.001872 0.099901 0.898226 1.000000
51 0.002595 0.099863 0.897541 0.898226
52 0.002994 0.099842 0.897163 0.806195
53 0.003362 0.099823 0.896815 0.723288

Step 1: Project the unit funds:

Year 1 Year 2 Year 3 Year 4


value of units at
start of the year 0.000 510.435 2,641.297 4,931.121

premium received 2,000 2,000 2,000 2,000

premium allocated 500 2,050 2,050 2,050

B/O spread 25 102.5 102.5 102.5

interest 38.000 196.635 367.104 550.290

mngt charge 2.565 13.273 24.780 37.145

value of units at
end of the year 510.435 2,641.297 4,931.121 7,391.766

Note that management charges are imposed after crediting of interest.

69
Step 2: Project the non-unit cashflows:

Year 1 Year 2 Year 3 Year 4

premium unallocated (+) 1,500 -50 -50 -50

B/O spread (+) 25 102.5 102.5 102.5

expenses (-) 1,050 100 100 100

interest (+) 28.50 -2.85 -2.85 -2.85

mngt charge (+) 2.565 13.273 24.780 37.145

extra mortality cost (-) 8.406 6.122 0.206 0.000

end of the year


cash flow 497.659 -43.199 -25.777 -13.205

Note that management charges is positive because this is an (addition) cash flow to the insurer. The
extra mortality costs were computed as

max (0, 5000 − ut ) · (aq)dx+t−1

which provides for the shortfall of the units to provide for the minimum death benefit guaranteed.

4. [Question] Step 1: Project the unit funds:

Year 1 Year 2 Year 3


value of units at
start of the year 0.000 486.804 1,472.508

premium received 1,000 1,000 1,000

premium allocated 500 1,000 1,000

B/O spread 25 50 50

interest 14.250 43.104 72.675

mngt charge 2.446 7.400 12.476

value of units at
end of the year 486.804 1,472.508 2,482.708

Note that management charges are imposed after crediting of interest.

70
Step 2: Project the non-unit cashflows:
Year 1 Year 2 Year 3

premium unallocated (+) 500 0 0

B/O spread (+) 25 50 50

expenses (-) 400 100 100

interest (+) 3.750 -1.500 -1.500

mngt charge (+) 2.446 7.400 12.476

extra mortality cost (-) 8.106 0.220 0

end of the year


cash flow 123.091 -44.320 -39.024
Note that management charges is positive because this is an (addition) cash flow to the insurer. The
extra mortality costs were computed as
max (0, 1500 − ut ) · (aq)dx+t−1
which provides for the shortfall of the units to provide for the minimum death benefit guaranteed.

5. Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual


Correction to the solution of the textbook question 11.3 part (d):
The profit margin is:
288.64
= 7.8%
3698.36
6. [Question] One can use “Solver” in Excel, for example, to find the desired annual premium, by varying
the annual premium and seeking a 15% profit margin.

(a) In the end, the result is an annual premium of $19, 847.84. The details of the resulting emerging
cash flows with this annual premium is summarized below:
Year 1 Year 2 Year 3 Year 4 Year 5

Reserves at beginning of the year 0 18,320.61 37,404.58 57,360.96 78,189.75

Premium 19,847.84 19,847.84 19,847.84 19,847.84 19,847.84

Expenses 300.00 526.20 526.20 526.20 526.20

Interest on (Prem-Expns) 1,466.09 1,449.12 1,449.12 1,449.12 1,449.12


Interest on reserves 0.00 1,374.05 2,805.34 4,302.07 5,864.23
Total interest income 1,466.09 2,823.17 4,254.47 5,751.19 7,313.35

Death claims (expected) 335.80 490.30 565.00 635.20 714.00

Surrenders - - - - -

Expected reserves at E.O.Y. 18,259.09 37,221.19 57,036.87 77,693.09 99,286

E.O.Y. profits 2,419.03 2,753.93 3,378.82 4,105.51 4,824.74

71
where reserves are calculated using AM92 ultimate and the expected reserves at the end of the
n-th year is calculated using
n V · p[55]+n−1

Instead of calculating reserves at the beginning of the year and at the end of the year separately,
we can also calculate increase in reserves in the n−th year using

nV · p[55]+n−1 −n−1 V

Therefore, in computing the resulting profit margin, we have:

Year 1 Year 2 Year 3 Year 4 Year 5

Probability in force 1.000000 0.996642 0.991755 0.986152 0.979888

Discount factor 0.8928571 0.7971939 0.7117802 0.6355181 0.5674269

Expected PV profits 2,159.85 2,188.05 2,385.15 2,572.99 2,682.63

Expected PV premiums 19,847.84 17,661.77 15,692.12 13,931.66 12,359.97

One can then easily verify that the resulting profit margin will be around the neighborhood of
15%.
(b) If reserves were calculated using a lower interest rate, this increases the necessary value of reserves
to hold in each year because the reserve fund will be expected to be earning less. There will then
be larger increases in reserves expected in each year, thereby emerging of lower profits. To achieve
the 15% profit margin, one will have to increase the annual premium. If risk discount rate of 15%
were used instead of 12%, this decreases the present value of the profits, thereby requiring higher
annual premium to achieve the same profit goal. Lastly, if the AM92 ultimate table were used
instead of AM92 Select, then mortality cost will be higher than expected which reduces profit in
turn. Thereby higher premium is required.

7. [Question] In the calculation of probability in force in this question, we do not use UDD assumptions
because surrenders are only allowed at the end of each year. The MDT can be constructed as follows
d
(aq)x = qxd
w
(aq)x = 1 − qxd · 10%


d w
(ap)x = 1 − (aq)x − (aq)x = 0.9 − 0.9 qxd

so

(ap)[50] = 0.9 − 0.9 · 0.001971 = 0.8982261


(ap)[50]+1 = 0.9 − 0.9 · 0.002732 = 0.8975412
(ap)[50]+2 = 0.9 − 0.9 · 0.003152 = 0.8971632

Therefore, the probability in force for years 1 to 4 are calculated as

0 (ap)[50] = 1
(ap)[50] = 0.8982261
2 (ap)[50] = 0.8982261 · 0.8975412 = 0.8061949
3 (ap)[50] = 0.8061949 · 0.8971632 = 0.72328840

72
We can then calculate the expected p.v. of profits, premiums and then the profit margin (without
zeroizing future negative cash flows).

Year 1 Year 2 Year 3 Year 4


probability
in force 1.000000 0.898226 0.806195 0.72328840

discount factor 0.892857143 0.797193878 0.711780248 0.635518078

expected p.v. of profits 444.338 -30.933 -14.792 -6.062


sum = 392.55

expected p.v. of premiums 2,000.000 1,603.975 1,285.387 1029.6448


(annual premium*discount factor*probability in force)
expected p.v. premiums 5,917.730

profit margin 6.63%

8. [Question]

(a) Step 1: Project the unit funds:

Year 1 Year 2 Year 3 Year 4 Year 5


value of units at
start of the year 0 1285 2698.5 4273.35 6045.093

premium received 1500 1500 1500 1500 1500

premium allocated 1500 1500 1500 1500 1500

B/O spread 150 150 150 150 150

interest 135 263.5 424.85 621.7434 815.4982

mngt charge 200 200 200 200 200

value of units at
end of the year 1285 2698.5 4273.35 6045.093 8010.591

Note that management charges are imposed after crediting of interest.

Step 2: Project the non-unit cashflows:

The death benefit and maturity benefit are same as the balance of the unit fund. So there will be

73
no extra mortality and withdraw cost in the non-unit fund for any year.

Year 1 Year 2 Year 3 Year 4 Year 5

premium unallocated (+) 0 0 0 0 0

B/O spread (+) 150 150 150 150 150

expenses (-) 1200 100 100 100 100

interest (+) -42 2 2 2 2

mngt charge (+) 200 200 200 200 200

extra mortality cost (-) 0 0 0 0 0

end of the year


cash flow -892 252 252 252 252

Note that management charges is positive because this is an (addition) cash flow to the insurer.
So the profit vector turns out to be

PRO = (−892, 252, 252, 252, 252) .

(b) The NPV of the profits is equal to −163.007.


(c) This gives a profit margin of −2.86%.

9. [Question]

118.0
3V = = 111.85
1.055
2 V × 1.055 − p52 ×3 V = 136.2 →2 V = 234.78

1V × 1.055 − p[50]+1 ×2 V = 152.0 →1 V = 366.01

where
Year k q[50]+t−1 p[50]+t−1
1 0.001971 0.998029
2 0.002732 0.997268
3 0.003152 0.996848
4 0.003539 0.996461

10. [Question]

(a) Calculation of the profit signature without reserves held.

Year 1 Year 2 Year 3


probability
in force 1.000000 0.992000 0.984064

profit vector 123.091 -44.320 -39.024

profit signature 123.091 -43.966 -38.402

74
t AVt−1 P EC CoI ic AVt DBt
1 0.00 3 000 120.00 25 4.0% 2 960.20 12 000.00
2 2 969.20 3 000 111.88 25 4.5% 6 094.78 12 000.00
3 6 094.78 3 000 124.38 25 5.5% 9 437.40 14 156.09
4 9 437.40 3 000 137.75 25 5.5% 12 949.75 19 424.63

(b) We have V3 = 0.
39.024
2V = = 37.887
1.03
1 V × 1.03 − px+2 ×2 V = 44.320 →1 V = 79.518

and V0 = 0.

The end of year cash flows at the end of the first year after zeroization is then

123.091 −1 V × px = 44.209

The resulting profit signature is then


(44.209, 0, 0) .
(c) It is expected that the zeroized profit signature will increase with increased rate of growth in the
unit funds primarily for two reasons: (1) means that when the unit fund growth rate increases,
the value of the unit fund increase and thus lead to increased amount of management charge. This
will be added to the non-unit account and thus the non-unit account value will increase. When
the unit fund growth rate increases, the unit fund value after management charge also increases
for management charge is proportional to the unit fund value.
(2) Recall if there is a guaranteed minimum death benefit, part of it (the part that equals the
amount of the value fund) will be covered by the unit fund and the shortfall (the minimum
guaranteed death benefit minus the unit fund value), if any, will be covered by the non-unit fund.
If the unit fund value increase, then it is less likely for this to happen: the minimum guaranteed
death benefit - the unit fund value > 0.

11. [Question] Dickson et al. (2013b) Solutions Manual - Exercise 13.10.

(a) First, project the account values and the death benefit, assuming the policy remains in force for
four years. In general,
AVt = (AVt−1 + P − EC − CoI)(1 + iC t )

and
DBt = max(12000, 1.5Vt ).
The calculations are set out in the following table. Next, use AVt and DBt from this projection
to profit test the insurer’s cash flows. At time t = 0, we have the pre-premium expenses of $200,
giving P r0 = −200. For t = 1, 2, 3, 4, the emerging profit at the year end for a policy in force at
the start of the year is

P rt = (AVt−1 + P − E)(1 + it ) − q65+t−1 DBt − p65+t−1 AVt

The full profit vector calculation is tabulated below: The profit signature, Π, is the vector of
emerging profits per policy issued, where Πt =t−1 p65 P rt for t = 1, 2, 3, 4, and Π0 = P r0 . So

Π = (−200.00, 60.14, 109.19, 156.20, 182.16)0

and
4
X
t
N P V = Π0 + Πt v8% = 207.19
t=1

75
t AVt−1 P E I EDBt EAVt P rt
0 200.00 -200.00
1 0.00 3 000 50.00 132.75 70.98 2 951.64 60.14
2 2 969.20 3 000 51.00 325.50 79.42 6 054.44 109.84
3 6 094.78 3 000 52.02 587.78 104.88 9 367.47 158.18
4 9 437.40 3 000 53.06 804.98 161.16 12 842.31 185.84

(b) Given that the policyholder dies in the first year, the profit signature is

Π = (−200.00, −8917.15, 0, 0, 0)0

Note that the profit signature and profit vector are the same when there is no uncertainty about
the year of death.

We see that there is no impact on P r0 , as the value was not dependent on survival in any
case. The second term is from the year 1 cash flows, given that the death benefit is paid in the
first year:
Π1 = 1.045(3000 − 50) − 12000 = −8917.25
The NPV is
−200 − 8917.25/1.08 = −200 − 8256.71 = −8456.71

(c) The calculations are are similar to those in part (a), but assuming the policyholder survives the
term. For example,

Π1 = 1.045(3000 − 50) − AV1 = 1.045(3000 − 50) − 2960.20 = 122.55


Π2 = 1.055(AV1 + 3000 − 51) − AV2 = 1.055(2960.20 + 3000 − 51) − 6094.78 = 139.426
Π3 = 1.065(AV2 + 3000 − 52) − AV3 = 1.065(6094.78 + 3000 − 52) − 9437.40 = 193.16
Π4 = 1.065(AV3 + 3000 − 53) − AV4 = 1.065(9437.40 + 3000 − 53) − 12949.75 = 239.64

The profit signature, given survival to the end of the term, is

Π = (−200, 122.55, 139.426, 193.16, 239.64)0

and the NPV is $362.23.


(d) (i) The calculations are similar to those in part (c), but assuming the policyholder survives to
time 2, and then receives the full account value, AV2 , and there are no further cash flows.
Using part (c), we have
Π = (−200, 122.55, 139.426, 0, 0)0
and the NPV is $33.01.
(ii) Assuming the policyholder survives to time 2, and then receives 90% of the account value,
we have
Π2 = 1.055(AV1 + 3000 − 51) − 0.9AV2 = 748.904
giving
Π = (−200, 122.55, 748.904, 0, 0)0
and the NPV is $555.54.
(e) The profit margin, denoted pm for policyholders who remain in force throughout the contract is

NPV
pm = given survival to the contract end
P V P remiums
NPV 362.23
= = = 3.38%
3000ä4|8% 10731.29

76
The profit margin for a policyholder who surrenders at time 2 is
NPV
pm = given surrender at time 2
P V premiums
NPV NPV
= =
3000(1 + v8% ) 5777.78

Setting this profit margin equal to 3.38% gives

N P V = 0.0338 × 5777.78 = 195.02


P2 k
and so, since we also have N P V = k=0 Πk v8% ,
2
195.02 = −200 + 122.55v8% + Π2 v8% ,

giving Π2 = 328.40. Then if α represents the surrender penalty,

328.40 = 1.055(2969.20 + 3000 − 51) − (1 − α)AV2

giving α = 2.9%. so a cash value of 97.1% of the account value would generate 3.38% profit
margin required from policyholders surrendering during the second policy year.
(f) We see from part (a) that, ignoring surrenders, there will be a profit for the insurer, at the 8%
discount rate, of $207.19 per policy. (The profit margin is 1.95%.) However, part (b) shows that
early death is quite costly, as we would expect from a policy with a term insurance component.
On the other hand, survivals generate a small profit in NPV terms, of $362.23 each. The mix of
more severe losses from death benefit claims, which are assumed to be relatively rare, and smaller
gains from survivals, which are more common, gives the average results found in part (a).

If the policyholder surrenders at time 2, we see from part (d) that the cash value determina-
tion is significant. If there is no surrender penalty, so policyholders take the full account value on
early exit, the NPV is really small, at $32.82; the profit margin (which is a useful measure as it
adjusts for fewer premiums) is 0.6% in this case, compared with 1.96% overall.

If the cash value is reduced to 90% of the account value, the NPV is much stronger, at $555.35,
representing a profit margin if 9.61%. At this level of penalty, the insurer gains an NPV greater
than that earned from the full term survivors in part (c). It is not ideal to have surrenders gener-
ating significantly higher returns than those who stay - it creates perverse incentives for insurers
to stimulate higher withdrawal rates, and indicates a lack of equitable treatment for leavers com-
pared with stayers.

From part (e) we see that a cash value of 96.9% of the account value for the time 2 surrenders
would generate the same profit margin for leaving policyholders as for those who stay throughout
the term. This would appear to be more equitable. The NPV generated by the surrendering
policyholders is less than for those staying for the full term, but is slightly more than the overall
expected NPV, which allows for mortality but not for surrenders.

77
Module 7: Multiple Decrement Tables

Multiple decrement tables


1. [Solution] A life (x) is subject to two causes of death with:
1
(aµ)x+t = 0.1 for t ≥ 0

and
2 1
(aµ)x+t = for 0 ≤ t < 10.
10 − t
Calculate:

(a) the probability that the cause of death will be 1; and


(b) the probability that the cause of death will be 2.

2. [Solution] A two-year fully discrete term insurance policy issued to (50) pays $3, 000 if death is due to
accidental causes. It pays nothing if death is due to non-accidental causes. Interest rate i is given to be
20%. You are given the following double decrement table in which decrement 1 represents accidental
death and decrement 2 represents death from all other causes:
1 2
x (a`)x (a`)x
50 3,600 6,400
51 2,160 5,040
52 432 2,540

(a) Find the annual premium using the equivalence principle.


(b) Find the net premium reserve after one year.

3. [Solution] A whole life insurance policy provides that upon accidental death as a passenger on an
airplane, a benefit of $1,000,000 will be paid. If death occurs from other accidental causes, a death
benefit of $500,000 will be paid. If death occurs from a cause other than an accident, a death benefit
of $250,000 will be paid.
You are given:

ˆ death benefits are payable at the moment of death;


ˆ µ(1) = 1/2000000 where (1) indicates accidental death as a passenger on an airplane;
ˆ µ(2) = 1/250000 where (2) indicates death from other accidental causes;
ˆ µ(3) = 1/10000 where (3) indicates non-accidental death; and
ˆ force of interest is 6%.

Calculate the actuarial present value of the benefits provided by this insurance.
4. [Solution] A special whole life insurance of $100,000 payable at the moment of death of (x) includes a
double indemnity accidental benefit. This benefit pays during the first 10 years an additional benefit
of $100,000 at the moment of death for accidental cause of death.
You are given that:

78
(τ )
ˆ µx+t = 0.002 for all t ≥ 0;
(1) (1)
ˆ µx+t = 0.0001 for all t ≥ 0, where µx+t denotes the force of decrement due to accidental causes
of death; and
ˆ δ = 0.05.

Calculate the net single premium for this insurance.

The associated single decrement tables


5. [Solution] You are given the following double decrement table:
1 2
Age x (a`)x (ad)x (ad)x

24 901,020 299 92,762


25 807,959 314 86,632
26 721,013 324 80,385
27 640,304 329 74,117
28 565,858 329 67,909
29 497,620 324 61,839

Construct the two associated single-decrement tables. State any assumptions and approximations you
use whenever necessary.
6. [Solution] A person aged 20 exactly is subject to two modes of decrement, death and withdrawal from
the insurance portfolio. Decrements due to death take place uniformly over the year of age in the
double-decrement table and the person can only withdraw at the end of the first month after birthday.
d
Find the two equations that stating the relationship between the independent rates of decrement, q20
w d w
and q20 , and the dependent rates of decrement, (aq)20 and (aq)20 .

Construction of MDT from the ASDTs


7. [Solution] A man aged 40 purchases a one-year term insurance policy because of upcoming frequent
air travel. The policy will pay $1, 000, 000 upon his death, with an additional $1, 000, 000 if his death
is due to an accident. Claims are payable at the end of the year of death.
You are given that the independent rates of decrements are
1 2
q40 = 0.04 and q40 = 0.06

where decrement 1 represents non-accidental death and decrement 2 is accidental death.

(a) If the valuation rate of interest is 10%, find the actuarial present value of the benefit.
(b) If the actuarial present value of the benefit is $146, 246.45, find the valuation rate of interest.

8. [Solution] Suppose for a double decrement table with causes (1) and (2), you are given the following:

ˆ In the single decrement table associated with cause (1), decrements occurs at only two points
during the year. Half of the decrements occur at time 1/4 and the remaining half occur at time
3/4.
ˆ In the single decrement table associated with cause (2), decrements are uniformly distributed over
the year.
ˆ q30
1 2
= 0.12 and q30 = 0.10.

(a) Find the probability of a life aged 30.75 will die at 30.75 in the single decrement table associated
with cause (1).

79
(b) Find the probability of a life aged 30.25 will survive to at least age 30.75 in the single decrement
table associated with cause (2).
(c) Calculate (aq)130 .

9. [Solution] A multiple decrement table that allows for age retirements and deaths between the ages of
61 and 63 is as follows:
Age x (al)x (ad)rx (ad)dx
61 10,000 1,291 494
62 8,215 1,471 662
63 6,082

Following improvements in the mortality experience, it is decided to construct a new table with the
independent rates of mortality reduced by 30%. Construct the new multiple decrement table.

Statistical treatment
10. [Solution] A life aged (20) is subject to two causes of death. Let T represent the future lifetime of this
life and J represent the cause of death. Suppose

fT,J (t, 1) = 0.005e−0.005t , 0 ≤ t ≤ 96.242,


fT,J (t, 2) = 0.01e−0.01t , 0 ≤ t ≤ 96.242.

Calculate t (ap)20 , (aµ)j20+t , j


10 (aq)20 , 10 (aq)20 and the probability that the life dies due to cause j for
j = 1, 2.

80
Solutions to Module 7 Exercises
1. [Question] Note that the total force of mortality is
1 2
(aµ)x+t = (aµ)x+t + (aµ)x+t
1 1
= + , for 0 ≤ t < 10
10 10 − t
Therefore, the probability of survival for t years is given by
 Z t  
1 1
t (ap)x = exp − + ds
0 10 10 − s
 
t
= 1− e−0.1t , for 0 ≤ t < 10
10
Probability of survival after 10 years is 0.
(a) The probability that the cause of death will be 1 is given by
Z 10
1 1
10 (aq)x = t (ap)x · (aµ)x+t dt
0
Z 10  
t 1
= 1− e−0.1t · dt
0 10 10
= 0.36788.
(b) Thus, the probability that the cause of death will be 2 is given by
Z 10
2 2
10 (aq)x = t (ap)x · (aµ)x+t dt
0
Z 10  
t 1
= 1− e−0.1t · dt
0 10 10 − t
Z 10
1 −0.1t
= e dt
0 10
= 0.63212.
One can also evaluate the complement of probability in (a).
2. [Question] Constructing the double-decrement table we get:
1 2
x (a`)x (ad)x (ad)x
50 10,000 1,440 1,360
51 7,200 1,728 2,500
52 2,972 - -

(a) Let P be the net annual premium. The APV(Premiums) is given by


APV(Premiums) = P × (1 + v · (ap)50 )
 
1 7200
= P × 1+ ·
1.2 10000
= 1.6P
while the APV(Benefits) is given by
 
1 1
APV(Benefits) = 3000 × v · (aq)50 + v 2 · (ap)50 · (aq)51
 
1 1440 1 7200 1728
= 3000 × · + · ·
1.2 10000 1.44 10000 7200
= 720.
Therefore, by the equivalence principle, we have 720/1.6 = 450.

81
(b) The net premium reserve at the end of year 1 is given by
1
1V = 3000 × v · (aq)51 − P
1 1728
= 3000 × · − 450
1.2 7200
= 600 − 450 = 150.

3. [Question] Note that


1 1 1
(aµ) = (aµ)1 + (aµ)2 + (aµ)3 = µ(1) + µ(2) + µ(3) = + + = 0.0001405.
2000000 250000 10000
Rt
t (ap)x = e− 0
(aµ)x+s ds
= e−0.0001405t
The APV of the benefits from accidental death as an airplane passenger is
Z ∞
1 0.5
1000000 · e−0.06t e−0.0001405t · dt = .
0 2000000 0.0601405
The APV of the benefits from other accidental death is
Z ∞
1 2
500000 · e−0.06t e−0.0001405t · dt = .
0 250000 0.0601405
And finally, the APV of the benefits for non-accidental death is
Z ∞
1 25
250000 · e−0.06t e−0.0001405t · dt = .
0 10000 0.0601405
The sum of the three components gives the APV of all the benefits:
27.5
= 457.262577.
0.0601405

4. [Question] The APV is simply the sum of the APV of the regular death benefit and the accidental
death benefit.
Z ∞
(τ )
APV(Regular benefit) = 100000 v t t px µx+t (τ )dt
Z0 ∞ R t (τ )
= 100000 v t e− 0 µx+s ds µx+t (τ )dt
Z0 ∞
= 100000 e−0.05t e−0.002t 0.002dt
0
0.002
= 100000 ×
0.05 + 0.002
= 3846.15
Z 10
(τ )
APV(Accidental benefit) = 100000 v t t px µx+t (1)dt
0
Z 10 Rt (τ )
= 100000 v t e− 0
µx+s ds
µx+t (1)dt
0
Z 10
= 100000 e−0.05t e−0.002t 0.0001dt
0
1 − e−0.052
= 10 ×
0.05 + 0.002
= 77.98

Thus the total APV is the sum, 3924.13.

82
1 2
5. [Question] First, we compute the dependent rates (aq)x and (aq)x and convert them into independent
rates. Assuming other decrements are exposed for half of the year, we use the approximations:
 
1 1 2
qx1 = (aq)x · 1 + (aq)x
2

and  
2 1 1
qx2 = (aq)x · 1 + (aq)x .
2
We have
1 2
Age x (aq)x (aq)x qx1 qx2

24 0.0003318 0.1029522 0.0003489 0.1029693


25 0.0003886 0.1072233 0.0004095 0.1072441
26 0.0004494 0.1114890 0.0004744 0.1115140
27 0.0005138 0.1157528 0.0005436 0.1157826
28 0.0005814 0.1200107 0.0006163 0.1200456
29 0.0006511 0.1242695 0.0006916 0.1243100

The associated single decrement tables are therefore summarized below:


Age x `1x d1x `2x d2x

24 901,020 314 901,020 92,777


25 900,706 369 808,243 86,679
26 900,337 427 721,563 80,464
27 899,910 489 641,099 74,228
28 899,421 554 566,871 68,050
29 898,866 622 498,821 62,008

6. [Question]
1
w (ad)w d w
20 + 12 (ad)20 q20 1
q20 = = (aq)w20 + (aq)d20 q20
w
(al)20 12
11
d (ad)d20 + 12 (ad)w d
20 q20 11
q20 = = (aq)d20 + (aq)w d
20 q20
(al)20 12

First, we look at the withdraw probability in the single-decrement table (SDT). Because withdraws only
happens at the end of the first month, there have been deaths in the first month before withdrawing.
In the SDT, when the only decrement is withdrawing, then those deaths in the first month would
have survived to the end of the first month, and also subjected to the influence of withdrawing. To
summarize, compared to the multiple-decrement table (MDT), the additional withdraw in the SDT are
1
the number of deaths in the first month, which is 12 (ad)d20 due to the uniform distribution, multipled
by the independent withdraw probability. Therefore we have
1
w (ad)w d w
20 + 12 (ad)20 q20 1
q20 = = (aq)w
20 + (aq)d20 q20
w
.
(al)20 12

Next, we look at the death probability in the SDT. To quantify the independent deaths, we have to
bring back the number of withdraw at the end of the first month, and then applying the remaining
death probability to this population. It is easy to see that the number of withdraw is (ad)w
20 , but the
death probability for the remaining 11 months is NOT 11 q d
12 20 . Remember that we only have the
uniform assumption for the MDT model but NOT the SDT model. We need to calculate

83
the death probability for the remaining 11 months by ourselves. Note that in the MDT, the remaining
population at the end of the first month is
1 d w
(al)20 − (ad)20 − (ad)20 ,
12
and the number of death during the next 11 months is
11 d
(ad)20 .
12
Therefore, for someone who is alive at the end of the first month, the death probability in the next 11
months is
11 d 11 d
d 12 (ad)20 12 (aq)20
11 (aq) 1 = = ,
20+ 12 1 d w 1 d w
12
(al)20 − 12 (ad)20 − (ad)20 1 − 12 (aq)20 − (aq)20
and we further have
d d
11
12
q20+ 1 = 11
12
(aq)20+ 1
12 12

because death is the only decrement in the last 11 months. Finally, in the SDT for death, we can bring
back the number of withdraw (ad)w 20 and then apply the death probbaility 1211 (q) 1 for the next 11
20+ 12
months. The death probability in the SDT should be
d
(ad)20 + (ad)w
20 × 12
11 (q)
20+ 1
d 12
q20 =
(al)20
11 d
d (aq)20
= (aq)20 + (aq)w
20 × 1
12
d w
.
1− 12 (aq)20 − (aq)20

7. [Question] First, convert the ASDT rates into dependent rates using:
 
1 1 1 2
(aq)40 = q40 1 − q40 = 0.04 · 0.97 = 0.0388
2
and  
2 2 1 1
(aq)40 = q40 1 − q40 = 0.06 · 0.98 = 0.0588.
2
The APV of benefits can be written as
h i
1 2
APV(Benefits) = 1, 000, 000 × v · (aq)40 + 2 · (aq)40
= 156, 400v.
(a) If i = 10%, then this gives
1
APV(Benefits) = 156, 400 · = 142, 181.80.
1.1
(b) If APV(Benefits) is 146,246.45, this gives
156, 400
1+i=
146, 246.45
so that i = 6.94%.

8. [Question]

(a) In the associated single decrement table for decrement 1:


No one aged 30 dies before age 30.25,
1
Prob. of a live which reaches age 30.25 dying at 30.25: 0.5 × q30 = 0.5 × 0.12 = 0.06
Prob. of (30) surviving to age 30.75: 1 − 0.06 = 0.94
Prob. of a live (30) dying at 30.75: 0.5 × 0.12 = 0.06
Prob. of a life which reaches 30.75 will die at 30.75: 0.06/0.94 = 0.06383

84
(b) In the associated single decrement table for decrement 2:
2 2
t p30 = 1 − t q30 = 1 − 0.1t
Hence, 0.25 p30 = 1 − 0.025 = 0.975 0.75 p230 = 1 − 0.075 = 0.925
2
2
2 0.75 p30 0.925
0.5 p30.25 = 0.25 p2 = 0.975 = 0.94872
30

(c) In the double decrement table:


Prob. of (30) surviving to 30.25: 0.25 p230 = 0.975 because no one dies from cause (1) before 0.25
Prob. of (30) dying at age 30.25 from cause (1): 0.975 × 0.06 = 0.0585, since the prob. of a life
reaches 30.25 dying at 30.25 is 0.06.
Prob. of (30) surviving beyond age 30.25: 0.975 − 0.0585 = 0.9165
Prob. of (30) surviving to 30.75: 0.25 (ap)30 × 0.5 p230.25 = 0.9165 × 0.94872 = 0.8695, since no one
dies from cause (1) between age 30.25 and age 30.75
Prob. of (30) dying at age 30.75 from cause (1): 0.8695 × 0.06383 = 0.0555

(aq)130 =Prob. of (30) dying at age 30.25 from cause (1)+Prob. of (30) dying at age 30.75 from
cause (1)
=0.0585+0.0555=0.114

9. [Question] We will first apply the following formulas to obtain the independent rates of decrement from
the dependent rates:
(aq)dx (aq)rx
qxd ≈ and qx
r

1 − 12 (aq)rx 1 − 21 (aq)dx

Age x (aq)rx (aq)dx qxr qxd


60 0.1291 0.0494 0.132369527 0.052808809
61 0.17906269 0.080584297 0.186580416 0.08850859

Let qxd∗ be the adjusted independent rate of death. Since these rates have decreased by 30%, we
calculate them as:
qxd∗ = 0.7 × qxd

Age x qxr qxd∗ (aq)r∗


x (aq)d∗
x
60 0.13237 0.03697 0.12992 0.03452
60 0.18658 0.06196 0.18080 0.05618

Note that we have defined (aq)r∗ d∗


x as the dependent rate of retirement after adjustment, and (aq)x as
the dependent rate of death after decrement. Finally, we multiply these by the population to obtain
the new life table:

Age x (al)x (ad)rx (ad)dx


61 10000 1299 345
62 8356 1511 469
63 6376

10. [Question] Note

fT (t) = f (t, 1) + f (t, 2) = 0.005 exp(−0.005t) + 0.01 exp(−0.01t).

85
We can then calculate
Z t Z t
t (aq)20 = P (T < t) = fT (s)ds = (0.005 exp(−0.005s) + 0.01 exp(−0.01s))ds
0 0
= 2 − exp(−0.005t) − exp(−0.01t)
t (ap)20 = 1 −t (aq)20 = exp(−0.005t) + exp(−0.01t) − 1.

Hence
fT,J (t, 1) 0.005 exp(−0.005t)
(aµ)120+t = =
t (ap)20 exp(−0.005t) + exp(−0.01t) − 1
fT,J (t, 2) 0.01 exp(−0.01t)
(aµ)220+t = =
t (ap)20 exp(−0.005t) + exp(−0.01t) − 1

We can also calculate


Z 10 Z 10
1
10 (aq)20 = f(T,J) (t, 1)dt = 0.005 exp(−0.005t)dt = 1 − exp(−0.05) = 0.048771
0 0
Z 10 Z 10
2
10 (aq)20 = f(T,J) (t, 2)dt = 0.01 exp(−0.01t)dt = 1 − exp(−0.1) = 0.095163
0 0

and subsequently

10 (aq)20 =10 (aq)120 +10 (aq)220 = 2 − exp(−0.05) − exp(−0.1) = 0.143933

Finally, the probability that occurs due to cause j = P (J = j) = fJ (j):


Z 96.242 Z 96.242
fJ (1) = f (t, 1)dt = 0.005 exp(−0.005t)dt = 1 − exp(−0.005 × 96.242) = 0.3819649
0 0
Z 96.242 Z 96.242
fJ (2) = f (t, 2)dt = 0.01 exp(−0.01t)dt = 1 − exp(−0.01 × 96.242) = 0.6180326.
0 0

86
Module 8: Multiple State Models

Markov chains
1. Dickson et al. (2009a) Exercise 8.1 (a) (Dickson et al. (2013a) Exercise 8.6 (a))
Remark: There is a typo in the solution to Exercise 8.1 (a) (ii). We should have the differential
equation
∂ 01
tp = t p00 01
30 · µ30+t ,
∂t 30
from which we conclude that
Z 10
01 00
10 p30 = t p30 · µ01
30+t dt.
0

From Exercise 8.1 (a) (i), we can easily conclude the expression for t p00 30 . Bring in the expression of
00 01
t p30 , this integral is complicated but can be easily solved numerically, eventually giving 10 p30 = 0.000099.

2. Dickson et al. (2009a) Exercise 8.9 (a) (Dickson et al. (2013a) Exercise 8.12 (a))

3. [Solution] Dickson et al. (2013a) Exercise 8.9 Consider the permanent disability model in figure below
and suppose that µ02 12
x = µx for all x.

(i) Deduce that


 Z t   Z n 
00 11 01
p p
t x n−t x+t = exp − µx+s ds exp − µ02
x+s ds .
0 0

(ii) Hence show that


 Z n   Z n 
01
n px = exp − µ02
x+s ds 1 − exp − µ01
x+s ds .
0 0

87
4. [Solution] Prove the Kolmogorov equations:

∂ t pgh
x
X  gj jh gh hj

= t px µx+t − t px µx+t
∂t
j6=h

∂ t pgg X
x
= −t pgg
x µgj
x+t
∂t
j6=g

EPV of cashflows under multiple state models


5. [Solution] A life insurance company uses the following three-state model to calculate the annual pre-
mium for a 3-year critical illness policy issued to healthy policyholders aged 55 at entry.

In return for premiums payable annually in advance for 3 years, the insurer will pay benefits of:

$60, 000 if the policyholder dies from the healthy state;


$40, 000 if the policyholder is diagnosed as having become critically ill; and
$35, 000 if the policyholder dies from the critically ill state.

All benefits are payable at the end of the relevant policy year.

Now, let St denote the state of the policyholder at age 55 + t, so that S0 = H, and for t = 1, 2, 3,
St = H, C or D. The transition probabilities are given as follows:

P (St+1 = C |St = H ) = pHC


55+t = 0.05
P (St+1 = D |St = H ) = pHD
55+t = 0.03
P (St+1 = D |St = C ) = pCD
55+t = 0.24

(a) Calculate the probabilities of being in each state at times t = 1, 2, 3.


(b) Calculate the gross annual premium on this contract based on the following assumptions:

88
Interest rate: 5% per annum
Initial expenses: $200 incurred on payment
of the first premium
Initial commission: 40% of the first premium
Renewal expenses: $40 at times t = 1, 2
regardless of whether healthy or not

6. [Solution] An insurer sells combined death and sickness policies to healthy lives aged 40. The policies,
which are for a term of 2 years, pay a lump sum of $20, 000 at death, with an additional $10, 000 if
the deceased was sick in the prior period. These is also a benefit of $3, 000 per annum payable to sick
policyholders. Annual premiums are payable at the beginning of each year by healthy policyholders.

Changes in state occur at the end of the year.

The mortality and sickness of the policyholders are described by the following multiple state model:

The annual transition probabilities are as follows:

H S D
H 0.8 0.1 0.1
S 0.1 0.7 0.2
D 0.0 0.0 1.0

Assume that the interest rate is 5% pa. Ignoring expenses, calculate the annual premium payable for
this insurance.

7. [Solution] A life insurance company uses the following three-state model, to calculate the premiums
for a 3-year sickness policy issued to healthy policyholders age 50 exact at inception.

89
In return for a single premium of P payable at the outset, the company will pay a benefit of $10, 000
at the end of each of the three years if the policyholder is sick at that time.

Let St represent the state of the policyholder at age 50 + t, so that S0 = H and for t = 1, 2, and 3, St
could either be H, S or D. The life insurance company uses transition probabilities defined as follows:
pij
50+t = P (St+1 = j |St = i ) .

For t = 0, 1 and 2, the transition probabilities are:


pHD
50+t = 0.05 pSD
50+t = 0.15

pSH
50+t = 0.80 pHS
50+t = 0.10

The life insurance company calculates P as the expected present value of the benefit payments, assum-
ing interest rate of 6% per year and expenses of 5% of P . Calculate the value of P .
8. [Solution] A life insurance company uses the following three state model, to estimate the profit in
respect of a 2-year combined death benefit and sickness policy issued to a healthy policyholder aged
exactly 55 at inception:

90
In return for a single premium of $6, 000 payable at the outset while the policyholder is healthy, the
company will pay for the following benefits:

$16, 000 if the policyholder dies within 2 years, payable at the end of the year of death;

$8, 000 at the end of each of the 2 years if the policyholder is sick at those times.

Let St denote the state of the policyholder at age 55 + t, so that S0 = H and for t = 1 and 2, St could
be H, S or D. The company uses the transition probabilities defined by:

pij
55+t = P (St+1 = j |St = i ) .

For t = 0 and 1, the transition probabilities are given by:

pHD
55+t = 0.08, pSD
55+t = 0.15, pSH HS
55+t = 0.75, and p55+t = 0.12.

The transitions in the multiple state model are the only sources of randomness.

(a) One possible outcome for this policy is that the policyholder is healthy at all times. List all the
possible outcomes together with their associated cash flows.
(b) Calculate the probability that each outcome occurs.
(c) Now assuming a rate of interest of 8% per annum, calculate the Net Present Value at time 0 of
the profit arising from each outcome.
(d) Calculate the mean and standard deviation of the Net Present Value of the profits at time 0 for
the policy.

9. [Solution] A life insurance company uses the following multiple state model for pricing annual premium
long-term care contracts, which are sold to lives that are healthy at onset.

Figure 8.1:

Under each contract, the life insurance company will pay the costs of long-term care while the poli-
cyholder satisfies the conditions for payment. These conditions are assessed every year on the policy
anniversary, just before payment of the premium then due. If the policyholder satisfies the conditions,
the annual amount of the benefit payable is paid immediately. A maximum of four benefit payments

91
may be made under the policy, after which time the policy expires. The policy also expires on earlier
death.

Premiums are payable annually in advance under the policy until expiry, and are waived if a benefit is
being paid at a policy anniversary.
For lives at claim level 1, benefits of 60% of the maximum level are paid, while lives at claim level 2
receive 100% of the maximum level. The current maximum level is $50, 000 per annum and is expected
to increase by 6% per annum compounded in the future.

Let pij
x denote the probability that a life aged x in state i will be in state j at age x + 1 and the insurer
uses the following probabilities for all values of x:
p00
x = 0.87 p01
x = 0.10 p02
x = 0.00

p11
x = 0.60 p12
x = 0.30 p22
x = 0.60

Assume that interest rate is 6% per annum and that expenses are 7.5% of each premium. Calculate
the annual premium under the contract.
10. [Solution] An insurance company issues a policy with a 30-year term to a life aged 35 exact. The policy
provides the following benefits:
(a) a benefit of $50, 000 payable immediately on the life becoming critically ill
(b) a death benefit of $80, 000 payable immediately on death if the life has been paid critical illness
benefit already, and $100, 000 otherwise
(c) a disability benefit of $30, 000 per year payable continuously during the first period of sickness and
20, 000 thereafter whilst the life is sick

There is a 3-month waiting period for both the sickness and disability benefits and 1 year waiting period
for the death benefit.The company uses the following multiple-state model for pricing and reserving.

Figure 8.2:

The transition forces are µij


x+t for i, j = 0, 1, 2, 3.

(i) Express the expected present value of the death benefit in integral form using the probabilities
ij
t px , i, j = 0, 1, 2, 3, t ≥ 0 and x ≥ 0, and the various forces of transition.
(ii) Express the expected present value of the disability benefit in integral form using the probabilities
ij
t px , i, j = 0, 1, 2, 3, t ≥ 0 and x ≥ 0, and the various forces of transition.

92
Multiple state model and multiple decrement table
11. [Solution] A person aged x is subject to two causes of decrements, retired (R) and dead (D) from the
portfolio with the following forces of decrements

(aµ)R
x+t = 0.004(x + t) for t ≥ 0
(aµ)D
x+t = 0.001(x + t) for t ≥ 0

(i) By considering a model with three states: Active (A), Retired (R) and Dead (D) in which R and D
are two absorbing states, derive the Kolmogorov differential equations for dependent probabilities
∂ R ∂ D
∂t t (aq)x and ∂t t (aq)x
(ii) Solve the above differential equations to obtain expressions for dependent probabilities t (aq)R
x and
D
t (aq)x

(iii) Find expressions for independent probabilities t qxR and t qxD

93
Solutions to Module 8 Exercises
1. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.
Note that there is a typo in the solution for part (ii). The formula for the required probability should
be
Z 10
01 00 01
10 p30 = t p30 µ30+t dt
0

2. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.

3. [Question] Dickson et al., 2013b Solutions Manual.

(i) Notice from the diagram that once you leave the state 0, it it impossible to come back to 0. In
order to be at state 0 after time t, one has to always stay at state 0 until time t, which means that
00 00 11 11
t px = t px . The same obervation also applies for state 1, that is, n−t px+t = n−t px+t . Therefore,
we have
 Z t   Z n−t 
00 11 01 02 12
p p
t x n−t x+t = exp − (µ x+s + µ x+s )ds exp − µ x+t+r dr
0 0
 Z t Z t Z n−t 
= exp − µ01
x+s ds − µ 02
x+s ds − µ 02
x+t+r dr
0 0 0
 Z t Z t Z n 
01 02 02
= exp − µx+s ds − µx+s ds − µx+s ds
0 0 t
 Z t Z n 
= exp − µ01
x+s ds − µ02
x+s ds
0 0
 Z t   Z n 
= exp − µ01
x+s ds exp − µ02
x+s ds ,
0 0

where the second equality comes from the condition that µ02 12
x = µx for all x.
(ii) From the diagram, we can see that
Z n
01 00 01 11
p
n x = t px µx+t n−t px+t dt
0
Z n  Z t   Z n−t 
01 02 01 12
= exp − (µx+s + µx+s )ds µx+t exp − µx+t+r dr dt
0 0 0
Z n  Z t Z n 
= µ01
x+t exp − (µ01 02
x+s + µx+s )ds − µ12
x+s ds dt
0 0 t
Z n  Z t Z n 
01 01 02 02
= µx+t exp − (µx+s + µx+s )ds − µx+s ds dt
0 0 t
Z n  Z t Z n 
= µ01
x+t exp − µ 01
x+s ds − µ 02
x+s ds dt
0 0 0
 Z n Z n  Z t 
= exp − µ02
x+s ds µ 01
x+t exp − µ01
x+s ds dt,
0 0 0

where we have again applied the condition that µ02 12


x = µx for all x. Defining a function
 Z t 
01
F (t) = exp − µx+s ds ,
0

we find out that  Z t 


d
F (t) = −µ01
x+t exp − µ01
x+s ds
dt 0

94
we get

 Z n Z n
01
n px = exp − (µ02
x+s ds) −dF (t)
0 0
 Z n 
= exp − (µ02
x+s ds) (F (0) − F (n))
 Z0 n   Z n 
02 01
= exp − (µx+s ds) 1 − exp − (µx+s ds) .
0 0

gh
4. [Question] We will first derive an expression for t+dt px .

gh
t+dt px = t pg1 1h g2 2h
x dt px+t + t px dt px+t + . . .
 
X gj jh
= t px dt px+t
 + t pgh hh
x dt px+t
j6=h
   
X   X hj
gj
= t px µjh
x+t dt
gh
+ o(dt)  + t px 1 − dt px+t

j6=h j6=h
   
X   X  hj 
gj
= t px µjh
x+t dt
gh
+ o(dt)  + t px 1 − µx+t dt + o(dt) 
j6=h j6=h

Note that we have written dt pjh jh


x+t = µx+t dt + o(dt). This means the probability of transitioning more
than once in a small unit of time dt is extremely small o(dt), and follows from the Markov assumption.
Finally, we will apply differentiation by first principles:

∂ t pgh pgh − t pgh x


x
= lim t+dt x
∂t dt→0 dt
    
gj jh gh hj
− t pgh
P P
j6=h t p x µ x+t dt + o(dt) + t px 1 − j6=h µx+t dt + o(dt) x
= lim
dt→0 dt
X gj jh gh
X hj
= t px µx+t − t px µx+t
j6=h j6=h
X gj jh gh hj

= t px µx+t − t px µx+t
j6=h

95
gg
∂ t px
The same process applies to derive the Kolmogorov equation for ∂t .

gg
t+dt px = t pgg gg
x dt px+t
 
X
= t pgg
x
1 − gj 
dt px+t
j6=g
 
X 
= t pgg
x
1 − µgj
x+t dt + o(dt)

j6=g
gg
∂ t pgg
x t+dt px − t pggx
= lim
∂t dt→0 dt
  
gg
1 − j6=g µgj − t pgg
P
t px x+t dt + o(dt) x
= lim
dt→0 dt
 
gg P gj
−t px j6=g µx+t dt + o(dt)
= lim
dt→0 dt
 
gg
X  gj  o(dt)
= lim −t px µx+t + 
dt→0 dt
j6=g
gg
X gj
= −t px µx+t
j6=g

5. [Question] Denote by P the vector of one-year transition probabilities:


 
0.92 0.05 0.03
P = 0.00 0.76 0.24
0.00 0.00 1.00

(a) Let π (t) denote the row vector of state probabilities at time t so that

π (t) = π (t − 1) × P

where π (0) = (1, 0, 0) since the policyholder is active/healthy at issue. Therefore, we have
 
0.92 0.05 0.03
π (1) = (1, 0, 0) × 0.00 0.76 0.24 = (0.92, 0.05, 0.03)
0.00 0.00 1.00

and similarly,
 
0.92 0.05 0.03
π (2) = (0.92, 0.05, 0.03) × 0.00 0.76 0.24 = (0.8464, 0.0840, 0.0696)
0.00 0.00 1.00
 
0.92 0.05 0.03
π (3) = (0.8464, 0.0840, 0.0696) × 0.00 0.76 0.24 = (0.7787, 0.1062, 0.1152) .
0.00 0.00 1.00

(b) Let P be the required gross premium. We summarize below all the possible states that might
happen by maturity and take the net present value of premiums, benefits and expenses under
each state:

96
possible states
0 1 2 3 probabilities net present value

H D 0.03000 (0.6P − 200) − 60000v


H H D 0.02760 (0.6P − 200 + P v) − 40v− 60000v 2 
H H H H 0.77869 0.6P − 200 + P v + P v 2  − 40v + 40v 2 
H H H D 0.02539 0.6P − 200 + P v + P v 2  − 40v + 40v 2 + 60000v 3 
H H H C 0.04232 0.6P − 200 + P v + P v 2 − 40v + 40v 2 + 40000v 
3
2 3
H H C D 0.01104 (0.6P − 200 + P v) − 40v + 40040v + 35000v
2
H H C C 0.03496 (0.6P − 200 + P v) − 40v + 40040v 
2
H C D 0.01200 (0.6P − 200) − 40040v + 35000v 
H C C D 0.00912 (0.6P − 200) − 40040v + 40v 2 + 35000v 3
H C C C 0.02888 (0.6P − 200) − 40040v + 40v 2
Then multiplying each net present value by the appropriate probabilities to get the actuarial
present value of premiums, benefits and expenses, and then equate this to zero, by the equivalence
principle. Solving for the gross annual premium, we get

P = 4, 740.40.

[Here we assumed that the renewal expense of $40 in years 1 and 2 is NOT payable upon death -
only if healthy or sick.]

6. [Question] Let P be the annual premium payable for this insurance. We summarize below all the
possible states that might happen by maturity and take the net present value of premiums and benefits
(no expenses) under each state:
possible states
0 1 2 probabilities net present value

H D 0.10 P − 20000v
H H D 0.08 (P + P v) − 20000v 2
H H H 0.64 P + Pv
H H S 0.08 (P + P v) − 3000v 2 
H S D 0.02 P − 3000v + 30000v 2
H S H 0.01 P − 3000v 
H S S 0.07 P − 3000v + 3000v 2
Then multiplying each net present value by the appropriate probabilities to get the actuarial present
value of premiums, benefits and expenses, and then equate this to zero, by the equivalence principle.
Solving for the annual premium, we get

P = 2, 607.50.

7. [Question] There is only sickness benefits in the policy and there is only one single premium. The
probability of being sick at time t = 1 is given by

pHS
50 = 0.10.

The probability of being sick at time t = 2 is given by

pHH HS HS SS
50 · p51 + p50 · p51 = 0.85 · 0.10 + 0.10 · 0.05 = 0.09.

Finally, the probability of being sick at time t = 3 is given by

pHH HH HS HH HS SS HS SS SS HS SH HS
50 · p51 · p52 + p50 · p51 · p52 + p50 · p51 · p52 + p50 · p51 · p52
= 0.85 · 0.85 · 0.1 + 0.85 · 0.1 · 0.05 + 0.1 · 0.05 · 0.05 + 0.1 · 0.8 · 0.1
= 0.08475.

97
Therefore,P can be solved from
P = 0.05P + 10, 000 0.1v + 0.09v 2 + 0.08475v 2


and this implies that


P = 2, 585.23.

8. [Question] There are a total of 7 possible outcomes by the end of the policy.

(a) These outcomes are listed below, together with their associated cash flows:
H→D: 6000, -16000
H→H→H: 6000, 0, 0
H→H→S: 6000, 0, -8000
H→H→D: 6000, 0, -16000
H→S→H: 6000, -8000, 0
H→S→S: 6000, -8000, -8000
H→S→D: 6000, -8000, -16000

(b) The probabilities of each of these outcomes are given below:


P (H → D) = 0.08
P (H → H → H) = 0.8 × 0.8 = 0.64
P (H → H → S) = 0.8 × 0.12 = 0.096
P (H → H → D) = 0.8 × 0.08 = 0.064
P (H → S → H) = 0.12 × 0.75 = 0.09
P (H → S → S) = 0.12 × 0.1 = 0.012
P (H → S → D) = 0.12 × 0.15 = 0.018
One can easily verify that the sum of all these probabilities equals to 1.
(c) The NPV of the profit at time 0 arising from each outcome:
H → D = 6000 − 16000v = −8814.81
H → H → H = 6000 + 0v + 0v 2 = 6000
H → H → S = 6000 + 0v − 8000v 2 = −858.71
H → H → D = 6000 + 0v − 16000v 2 = −7717.42
H → S → H = 6000 − 8000v + 0v 2 = −1407.41
H → S → S = 6000 − 8000v − 8000v 2 = −8266.12
H → S → D = 6000 − 8000v − 16000v 2 = −15124.83

(d) The expected value of the NPV of profits at time 0 is therefore given by
(−8814.81 × 0.08) + (6000 × 0.64) + (−858.71 × 0.096) + (−7717.42 × 0.064)
+ (−1407.41 × 0.09) + (−8266.12 × 0.012) + (−15124.83 × 0.018)
= 2060.36.
The variance of the net present value of profits at time 0 is
   
2 2
(−8814.81 − 2060.36) × 0.08 + (6000 − 2060.36) × 0.64 + · · ·
 
2
+ (−15124.83 − 2060.36) × 0.018
= 34, 009, 449.78.
So that the standard deviation is
p
SD = 34, 009, 449.78 = 5, 831.76.

98
9. [Question] With no recovery to the healthy state, premiums are payable only until the first claim, or
death (whichever occurs first). The Actuarial Present Value of premiums is therefore
∞ ∞
X X k 1
APV (Premiums) = P v k k p00
x =P v k (0.87) = P = 5.57894P.
1 − 0.87v
k=0 k=0

Now valuing the benefits from the point when the first claim arises (that is, conditional on the first
claim), we get the following probabilities:
the first claim payment will be at claim level 1, with probability 1;
the second claim payment will be at claim level 1, with probability 0.6 and at claim level 2 with prob-
ability 0.3;
the third claim payment will be at claim level 1, with probability 0.62 = 0.36, and at claim level 2,
with probability 0.6 ∗ 0.3 + 0.3 ∗ 0.6 = 0.36;
the fourth claim payment will be at claim level 1, with probability 0.63 = 0.216, and at claim level 2,
with probability 0.6 ∗ 0.3 ∗ 0.6 + 0.6 ∗ 0.6 ∗ 0.3 + 0.3 ∗ 0.6 ∗ 0.6 = 0.324.

If the first claim is in k years, the expected present value of any level 1 claim benefits will be

50000 ∗ 0.6 ∗ 1.06k ∗ v k .

But with v at 6%, this is 30,000 for all k. Similarly, the expected present value of any level 2 claim
benefits will be 50,000, so we can ignore interest in valuing claims.
The APV of all claims from the point of the first claim payment arising is therefore

30, 000 ∗ (1 + 0.6 + 0.36 + 0.216) + 50, 000 ∗ (0 + 0.3 + 0.36 + 0.324) = 114, 480.

Finally, the probability that the first claim occurs at the end of year 1 is 0.1, at the end of year 2
is 0.87 ∗ 0.1, at the end of year 3 is 0.872 ∗ 0.1, and so on, which in general at the end of year k is
0.87k−1 ∗ 0.1.
The probability of a claim is therefore
0.1
0.1 ∗ 1 + 0.87 + 0.872 + · · · =

= 0.76923.
1 − 0.87
The APV of all claims is therefore

APV (Claims) = 114, 480 ∗ 0.76923 = 88, 061.45.

With 7.5% of each premium for expenses, the equivalence principle leads us to the equation of value:

(1 − 0.075) ∗ 5.57894P = 88, 061.45

so that
88, 061.45
P = = 17, 064.43.
(1 − 0.075) ∗ 5.57894

10. [Question]

(i) EPV of death benefit is


Z 30 Z 30 Z 30
100, 000 v t t p00 µ02
35 35+t dt + 100, 000 v t t p01 12
35 µ35+t dt + 80, 000 v t t p03 32
35 µ35+t dt
1 1 1

(ii) EPV of disability benefits is


Z 30 Z 30 Z 30
t 01 00 01
20, 000 v t p35 dt + 10, 000 t p35 µ35+t v s s−t p11
35+t dsdt
0.25 0 t∨0.25

where Rt
(µ01 02 03
00
t p35 = e− 0 35+s +µ35+s +µ35+s )ds

99
and R s−t
(µ10 11 12
11
s−t p35+t = e− 0 35+t+y +µ35+t+y +µ35+t+y )ds

In the first term the payment of 20,000 is inclusive of all cases of sickness, even those that
experience the ”first period of sickness”. Hence in the second term we only include $10,000 extra
for those that got sick the first time (which makes $30,000 in total).
The person stays healthy (p00 ) until they becomes sick for the first time at age 35 + t. From this
point on they stays sick (p11 ) for a period of length s-t where s takes value from max(t, 0.25) to
30.

11. [Question]

(i) Given the forces of decrements in the question, we have the corresponding transition intensities

(aµ)R
x+t = 0.004(x + t) for t ≥ 0
(aµ)D
x+t = 0.001(x + t) for t ≥ 0

for transitions of the life aged x from A to R and A to D, respectively.

We can then set up the following Kolmogorov equations


∂ R D
t (ap)x = −((aµ)x+t + (aµ)x+t )t (ap)x
∂t
= − (0.004(x + t) + 0.001(x + t)) t (ap)x

which gives
 Z t 
t (ap)x = exp − (0.004(x + s) + 0.001(x + s)) ds
0
= exp −(0.005xt + 0.0025t2 )


= exp −0.0025((x + t)2 − x2 )




Hence
∂ R R
t (aq)x = (aµ)x+t × t (ap)x
∂t
= 0.004(x + t) × exp −0.0025((x + t)2 − x2 )


∂ D D
t (aq)x = (aµ)x+t × t (ap)x
∂t
= 0.001(x + t) × exp −0.0025((x + t)2 − x2 )


(ii) Solving Kolmogorov differential equations we have


Z t
R
0.004(x + s) × exp −0.0025((x + s)2 − x2 ) ds

t (aq)x =
0
Z t
= 0.004 exp 0.0025x2 (x + s) × exp −0.0025((x + s)2 ) ds
 
0

Use a change of variable y = (x + s)2 , dy = 2(x + s)ds we have:


Z (x+t)2
R
= 0.002 exp 0.0025x2

t (aq)x exp {−0.0025y} dy
x2
0.002
exp 0.0025x2 exp −0.0025x2 − exp −0.0025(x + t)2
  
=
0.0025
= 0.8 1 − exp −0.0025((x + t)2 − x2 )


100
Similarly we have
D
= 0.2 1 − exp −0.0025((x + t)2 − x2 )

t (aq)x

(iii) Using the link between multiple decrements and associated single decrement tables, we have

µR R
x+t = (aµ)x+t = 0.004(x + t) for t ≥ 0
µD
x+t = (aµ)D
x+t = 0.001(x + t) for t ≥ 0

Hence
Z t 
R
t px = exp −0.004(x + s)ds
0
= exp {−0.002t(2x + t)}

hence
R
t qx = 1 − exp {−0.002t(2x + t)}

Similarly we can find


D
t qx = 1 − exp {−0.0005t(2x + t)}

101
Module 9: Insurance and Annuities
for Multiple Lives

Joint distribution of future lifetimes, the joint life status and ben-
efits
1. Dickson et al. (2009a) Exercise 8.17 (c) (Dickson et al., 2013a Exercise 9.12 (c))
2. Dickson et al. (2009a) Exercise 8.18 (a) (Dickson et al., 2013a Exercise 9.13 (a))
3. [Solution] The following multiple state model represents the joint mortality of two lives, one male and
one female:

A life insurer sells a 2-year term insurance contract to a man age 50 and a woman age 50. The sum
insured of $100, 000 is payable at the end of the year in which the second life does, if both die within
2 years. Premiums are payable annually in advance.

The following transition probabilities are given (independent of the ages of the two lives, which we
denote x and y below):
11 12
1 px:y = 0.96 1 px:y = 0.02

13 14
1 px:y = 0.01 1 px:y = 0.01

24 34
1 px:y = 0.02 1 px:y = 0.04

102
(a) Calculate the net premium, at a rate of interest of 5% per annum.
(b) Calculate the standard deviation of the present value of the profit, at a rate of interest of 5% per
annum.

4. [Solution] A joint life annuity of 1 per annum is payable continuously to lives who are currently aged
x and y while both lives are alive. The present value of the annuity benefits is expressed as a random
variable, in terms of the joint future lifetime of x and y.
Derive and simplify as far as possible expressions for:

(a) the expected present value of this annuity; and


(b) the variance of the present value of this annuity.

5. [Solution] Let T (x) and T (y) be the complete future lifetimes of two lives now aged x and y, respectively.
Define the random variable g(T (x), T (y)) as
(
āT (x)| , if T (x) ≤ T (y)
g(T (x), T (y)) = .
āT (y)| , if T (x) > T (y)

(a) Describe the benefit which has present value equal to g(T (x), T (y)).
(b) Express E [g(T (x), T (y)] as an integral.
(c) Write down an expression for the variance of g(T (x), T (y)) using insurance functions.

The last-survivor status and benefits


6. Dickson et al. (2009a) Exercise 8.10 (a), (d) (Dickson et al., 2013a Exercise 9.2 (a), (d))
7. [Solution] Verbally describe the benefits payable to an annuity n| äxy . Show that n| äxy is not equivalent
to
v n n pxy · äx+n:y+n
nor to
v n n pxy · äx+n:y+n .
Instead, give a correct expression for n| äxy .

8. [Solution] On 1 January 2005, a life insurance company issued a joint life whole life insurance policy to
a couple, one male and one female. Both are then aged 50 exact on 1 January 2005. The policy pays
a sum assured of $200,000 immediately on the death of the second of the lives to die.
Premiums are payable annually in advance while at least one of the lives is alive.
Calculate the net annual premium payable under this policy.
Basis:
Mortality: PMA92C20 for the male and PFA92C20 for the female.
Interest: 4%.per annum.
9. [Solution] On 1 January 2001, a life insurance company issued joint life whole life insurance policies to
married couples. Each couple comprised one male and one female and both were aged 55 exact on 1
January 2001. Under each policy, a sum assured of $500,000 is payable at the moment of death of the
second of the lives to die.
Premiums for each policy are payable annually in advance while at least one of the lives is alive.
The insurance company prices the product using the following basis:

ˆ Mortality: PMA92C20 for male; PFA92C20 for female.


ˆ Interest: 4% per annum.
ˆ Expenses:

103
– Initial: $500
– Renewal (starting from the 2nd year): 2.5% of each premium payment

(a) Calculate the gross annual premium payable under each policy, using the equivalence principle.
State any assumptions you make.
(b) The company calculates reserves for these policies on a net premium basis, using PMA92C20 and
PFA92C20 mortality for the male and female lives respectively, and a 4% per annum interest.
Calculate the fifth-year reserve for each policy where:
1. only the male is alive; and
2. both the male and female are alive.

Joint life functions under some special laws


10. [Solution] Consider m independent lives with ages x1 , x2 , ..., xm and each follow the Makeham mortality
law
µx = A + Bcx .

(a) Prove that an annuity value on these m joint lives calculated at a rate of interest i is equal to the
value of a single-life annuity at age w but calculated at a rate of interest i0 .
(b) Find expression for i0 in terms of i.
(c) Show that the age w can be determined from the relation
µx1 + µx2 + · · · + µxm
µw+h =
m
where
h = − log m/ log c.

11. Dickson et al. (2009a) Exercise 8.16 (Dickson et al., 2013a Exercise 9.11)

Simple contingent functions


12. Dickson et al. (2009a) Exercise 8.11 (Dickson et al., 2013a Exercise 9.6)

13. [Solution] You are given two independent lives (x) and (y) where Z is the present value random variable
for an insurance on these two lives with
 T (y)
v , for T (y) > T (x)
Z= .
0, otherwise

The life (x) is subject to a constant force of mortality of 0.07 while (y) is subject to a constant force
of mortality of 0.09. The force of interest is a constant δ = 6%.

(a) Verbally describe this insurance provided.


(b) Calculate the actuarial present value for this insurance.
(c) Calculate the variance of the present value random variable.

Reversionary annuities
14. [Solution] A special life policy on two lives now aged x and y, respectively, provides for cash payments of
$10,000 and $20,000 immediately on the first and second deaths, respectively. In addition, an annuity
at the rate of $1,000 per annum will be paid continuously, commencing immediately on the first death
and ceasing immediately on the second death.

104
(a) Obtain an expression for the net single premium for this life policy.
(b) Express your answer in part (a) in terms of joint-life and single-life annuity functions.

15. Dickson et al. (2009a) Exercises 8.20 (Dickson et al., 2013a Exercise 9.16)
Remark: In the questions, the annuity is paying at an annual rate of 5000 payable continuously. This
is not a single payment but is a cash flow going out at rate 5000 and therefore we have a -5000 term
in the equation, just like a premium payable continuously at an annual rate of P. Therefore, there is a
-5000 included in the solution.

105
Solutions to Module 9 Exercises
1. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.
2. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.
3. [Question] Let P be the net premium per annum. We summarize below all the possible states that
might happen by maturity and take the net present value of premiums and benefits under each state:
possible states
0 1 2 probabilities net present value

1 1 1 0.9261 P + Pv
1 1 2 0.0192 P + Pv
1 1 3 0.0096 P + Pv
1 1 4 0.0096 P + P v − 100, 000v 2
1 2 2 0.0196 P + Pv
1 2 4 0.0004 P + P v − 100, 000v 2
1 3 3 0.0096 P + Pv
1 3 4 0.0004 P + P v − 100, 000v 2
1 4 0.0100 P − 100, 000v
[Here, we assumed that premiums are payable so long as at least one of the lives is alive.]

(a) Then multiplying each net present value by the appropriate probabilities to get the actuarial
present value of premiums, benefits and expenses, and then equate this to zero, by the equivalence
principle. Solving for the net annual premium, we get

P = 975.72.

(b) Recall the formula ofor calculating the variance of the profit variable, denoted by Z,
z
V ar (Z) = E[Z 2 ] − (E[Z]) .

According to the equivalence principle, the net premium has been set so that the expected profit
E[Z] is 0. In this case, the standard deviation is simply the square root of E[Z 2 ] , which can be
calculated by X 2
(net present value) × probabilities.
One can easily verify that this is equal to

13, 206.58.

4. [Question] The present value random variable for the annuity benefits can be expressed as

Y = āT (xy)|

where T (xy) is the joint-life status of (x) and (y).

(a) The expected present value for this annuity is given by


 
E (Y ) = E āT (xy)| = āxy .

(b) Its variance can be expressed as

1 − v T (xy)
   
Var (Y ) = Var āT (xy)| = Var
δ
1 
T (xy)
 1 h
2
2 i
= Var v = Ā xy − Ā xy .
δ2 δ2

106
5. [Question]

(a) This is the present value of a joint life annuity of $1 per annum payable continuously until the
first death of 2 lives (x) and (y).
(b) We write either Z ∞
E [g (T (x), T (y))] = t pxy µx+t:y+t āT | dt
0
or Z ∞
E [g (T (x), T (y))] = t pxy e−δt dt
0

(c) The variance can be expressed as

2
2
Āxy − Āxy
Var [g (T (x), T (y))] = ,
δ2
where 2 Āxy indicates that the function is to be evaluated at force of interest 2δ.

6. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.

7. [Question] Recall that the status (xy) fails upon the last death, so that (xy) survives so long as at
least one of them is alive. Therefore, n| äxy pays a $1 benefit, starting n years from now, so long as at
least one of (x) and (y) is alive. Its actuarial present value can be expressed as, using current payment
technique,

X ∞
X
n| äxy = v k k pxy = v k (k px +k py −k pxy )
k=n k=n

X ∗
= vk +n
(k∗ +n px +k∗ +n py −k∗ +n pxy )
k∗ =0
∞ ∞ ∞
X ∗ X ∗ X ∗
= v n n px vk k∗ px+n + v n n py vk k∗ py+n − v n n pxy vk k∗ px+n:y+n
k∗ =0 k∗ =0 k∗ =0
= v n n px · äx+n + v n n py · äy+n − v n n pxy · äx+n:y+n

and is clearly not equal to

v n n pxy · äx+n:y+n = v n n pxy · (äx+n + äy+n − äx+n:y+n )


= v n n pxy · äx+n + v n n pxy · äy+n − v n n pxy · äx+n:y+n
= v n n px · äx+n + v n n py · äy+n − v n n pxy · äx+n:y+n
− (v n n qy · äx+n + v n n qx · äy+n ) + (v n n qxy · äx+n + v n n qxy · äy+n )

nor

v n n pxy · äx+n:y+n = v n ( n px + n py − n pxy ) · (äx+n + äy+n − äx+n:y+n )


n n n
= v n px · äx+n + v n py · äy+n − v n pxy · äx+n:y+n
+v n n px · (äy+n − äx+n:y+n ) + v n n py · (äx+n − äx+n:y+n )
−v n n pxy · (äx+n + äy+n − 2äx+n:y+n )

even if lives are assumed to be independent.

8. [Question] Denote by P the required net annual premium. Clearly, we have by the equivalence principle:

APVFP = APVFB
P · ä50:50 = 200, 000 · Ā50:50

107
and therefore

Ā50:50 i 1 − dä50:50
P = 200, 000 · = 200, 000 ·
ä50:50 δ ä50:50
 
i 1
= 200, 000 · −d
δ ä50:50

where
f
ä50:50 = äm
50 + ä50 − ä50:50 = 18.843 + 19.539 − 17.688 = 20.694.

Therefore, we have
 
0.04 1 0.04
P = 200, 000 · − = 2, 011.52.
ln(1.04) 20.694 1.04

9. [Question] This is a last survivor benefit. To calculate continuous benefits, we will use the UDD
approximation.

(a) Let G be the gross annual premium.

AP V F P = G ∗ ä55:55
 
f
= G ∗ äm55 + ä 55 − ä55:55

= 19.558G

We may use the UDD approximation to find the APV of future benefits:

AP V F B = 500000 Ā55:55
i
= 500000 A55:55
δ
i
= 500000 (1 − dä55:55 )
δ  
0.04 0.04
= 500000 1− × 19.558
ln 1.04 1.04
= 126346.112
AP V F E = 500 + 0.025G (ä55:55 − 1)
= 500 + 0.025G × 18.558
= 500 + 0.46395G

We then use the equivalence principle to obtain:

126346.112 + 500
G= = 6643.23
19.558 − 0.46395

(b) To calculate net premium reserves, we need the net premium. This is calculated as:

Ā55:55
P = 500000
ä55:55
i 1 − dä55:55
= 500000
δ ä55:55
0.04
1 − 1.04 × 19.558
 
0.04
= 500000
ln 1.04 19.558
= 6460.073

108
1. If only the male is alive, then the net premium reserve in year 5 is:
m
5V = 500000Ām m
60 − P ä60
i
= 500000 Am − P äm
δ 60  60 
0.04 0.04
= 500000 1− × 15.632 − 6460.073 × 15.632
ln 1.04 1.04
= 102362.38

2. If both the male and female are alive, the net premium reserve in year 5 is:
mf
5V = 500000Ā60:60 − P ä60:60
i
= 500000 (1 − dä60:60 ) − P ä60:60
δ  
0.04 0.04
= 500000 1− × 18.194 − 6460.073 × 18.194
ln 1.04 1.04
= 35563.495

10. [Question] First, we notice that given the Makeham’s law, we can show (easily) that we can express
the probability
Z k ! Z k !
x k
A + Bcx+s ds = sk × g c (c −1)

k px = exp − µx+s ds = exp −
0 0

where the constants s = e−A and log g = −B/ log c. Moreover, given the independence of these lives,
the joint survival probability is simply the muliplication of all individual survival probability:

k px1 x2 ···xm = Πm
i=1 k pxi
 
cxi (ck −1)
= Πm
i=1 sk
× g
x1
+cx2 +···+cxm )(ck −1)
= smk × g (c .

(a) The annuity value on these m joint lives calculated at a rate of interest i is equal to
∞ ∞
x1
+cx2 +···+cxm )
(ck −1)
X X
äx1 x2 ···xm = v k k px1 x2 ···xm = v k × smk × g (c
k=0 k=0

x1
+cx2 +···+cxm )(ck −1)
X
= v k × s(m−1)k × sk × g (c
k=0

w k
v ∗k × sk × g c (c −1)
X
=
k=0

where
v ∗ = v × s(m−1)
and
cw = cx1 + cx2 + · · · + cxm .
Thus, we are able to write this joint annuity value as

äx1 x2 ···xm = ä∗w

as single-life annuity value but discounted at a different rate of interest.


(b) The interest rate used for discounting in the single life can be determined from
1 1
= × s(m−1)
1 + i0 1+i

109
so that
1+i
i0 = −1
s(m−1)
where s = e−A , as determined earlier.
(c) The single-life (w) is determined from

cw = cx1 + cx2 + · · · + cxm


mA + Bcw = (A + Bcx1 ) + (A + Bcx2 ) + · · · + (A + Bcxm )
= µx1 + µx2 + · · · + µxm

which is equivalent to (dividing both sides by m)


Bcw µx1 + µx2 + · · · + µxm
A+ =
m m
µx1 + µx2 + · · · + µxm
A + Bcw+h =
m
µx1 + µx2 + · · · + µxm
µw+h =
m
where clearly
1
ch =
m
or equivalently h = − log m/ log c.

11. Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.


12. Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.
13. [Question] Since the lives are independent, we note that the joint density of their future lifetimes T (x)
and T (y) can be expressed as

fT (x)T (y) (s, t) = fT (x) (s) fT (y) (t)


= s px · µx+s ×t py · µy+t
= 0.07e−0.07s × 0.09e−0.09t

(a) The insurance provides a benefit if life (y) outlives (x).


(b) The actuarial present value for this insurance can be computed directly from the joint density:
Z ∞Z t
E (Z) = e−0.06t 0.07e−0.07s × 0.09e−0.09t dsdt
Z0 ∞ 0
1 − e−0.07t × 0.09e−0.15t dt

=
0
Z ∞
9 9 9
= − 0.09 e−0.22t dt = − = 0.190909.
15 0 15 22

(c) The variance of the present value random variable can be computed, by first computing:
Z ∞Z t
E Z2 e−0.12t 0.07e−0.07s × 0.09e−0.09t dsdt

=
0 0
Z ∞
1 − e−0.07t × 0.09e−0.21t dt

=
0
Z ∞
9 9 9
= − 0.09 e−0.28t dt = − = 0.107143.
21 0 21 28
The variance is therefore
2
Var (Z) = 0.107143 − (0.190909) = 0.070697.

110
14. [Question] Reversionary annuities:

(a) The net single premium for the life policy is given by

NSP = 10, 000Āxy + 20, 000Āxy + 1, 000 āy|x + āx|y .

(b) In terms of joint life and single life functions, this NSP can be written as:

NSP = 10, 000Āxy + 20, 000 Āx + Āy − Āxy + 1, 000 (āx − āxy + āy − āxy )

= 20, 000 Āx + Āy − 10, 000Āxy + 1, 000 (āx + āy − 2āxy ) .

15. See Dickson et al. (2009b) (Dickson et al., 2013b) Solutions Manual.

111
Module 10: Pension Mathematics

Expected value of lump sum and pension benefits


1. Dickson et al. (2009a) Exercises 9.3 (Dickson et al. (2013a) Exercises 10.3).

Salary scales, death benefits and age retirement benefits


2. Dickson et al. (2009a) Exercises 9.1 (Dickson et al. (2013a) Exercises 10.1).

3. Dickson et al. (2009a) Exercises 9.2 (Dickson et al. (2013a) Exercises 10.2).

4. [Solution] Under the rules of a pension scheme, a member may retire due to age at any age from exact
age 60 to exact age 65.
On age retirement, the scheme provides a pension of 1/60th of Final Pensionable Salary for each year
of scheme service, subject to a maximum of 40/60ths of Final Pensionable Salary. Only complete years
of service are taken into account.
Final Pensionable Salary is defined as the average salary over the three-year period before the date of
retirement.
The pension scheme also provides a lump sum benefit of four times Pensionable Salary on death before
retirement. The benefit is payable immediately on death and Pensionable Salary is defined as the
annual rate of salary at the date of death.
You are given the following data in respect of a member:
Date of birth 1 January 1979
Date of joining the scheme 1 January 2000
Annual rate of salary at 1 January 2005 $50,000
Date of last salary increase 1 April 2004

(a) Derive commutation functions to value the past service and future service pension liability on
age retirement for this member as at 1 January 2005. State any assumptions that you make and
define all the symbols that you use.
(b) Derive commutation functions to value the liability in respect of the lump sum payable on death
before retirement for this member as at 1 January 2005. State any assumptions that you make
and define all the symbols that you use.

5. [Solution] A pension scheme provides the following benefits:

ˆ on retirement, a pension commencing immediately at the annual rate of 1/45 of Final Pensionable
Salary for each year of service, with fractions of a year to count proportionately. Final Pensionable
Salary is defined as the member’s average annual salary received during the last three years of
service;
ˆ on death in service, a lump sum of 4 times the annual rate of salary at the date of death.

112
Members may retire once they have attained age 60; members must retire on attaining age 65.
Data are available in respect of each member’s salary in the year to the valuation date.
Develop commutation functions you would use for valuing the death benefits and the past and future
service pension benefits, for a member aged 55 exactly, with exactly 5 years past service. State clearly
any assumptions that you make and define carefully all the symbols that you use.

Salary related ill-health retirement benefits


6. [Solution] A pension scheme provides an ill-health retirement pension of 1/60 of Final Pensionable
Salary for each year of company service, with fractions of a year to count proportionately, subject to
a maximum pension of 40/60 of Final Pensionable Salary.
Retirement due to ill-health may take place at any age before age 65. Final Pensionable Salary is
defined as the average annual salary over the three-year period preceding retirement.
Derive commutation functions to value the ill-health retirement pension for a member aged exactly 25,
who has completed exactly 5 years company service to date. Define carefully all the symbols that you
use.

7. [Solution] A pension scheme provides a pension on ill-health retirement of 1/80 th of Final Pension-
able Salary for each year of pensionable service subject to a minimum pension of 20/80 the of Final
Pensionable Salary. Final Pensionable Salary is defined as the average salary earned in the three years
before retirement. Normal retirement age is 65 exact.
Calculate the present value of the ill-health retirement benefit for a member currently aged 35 exact
with exactly 10 years past service and salary for the year before the calculation date of $20,000.
Basis: Pension Scheme tables from Formulae and Tables for Examinations
Interest: 4% per annum

Contributions
8. [Solution] Consider a pension scheme set up by a small employer for 1 employee who is at age 35. This
employee has 6 years of past service with the expected salary in next year being $20, 000. The scheme
will provide a pension of 1/60th of pensionable salary for each year of service (fractions of a year
counting proportionally) on retirement for any reason. Pensionable salary is the average annual salary
earned in the final 36 months of employment. The employer meets the full cost of the scheme. The
contribution rate is determined by equating the expected present value of the total scheme liabilities to
the expected present value of contributions. Contributions are calculated to be a constant percentage
of the total salaries of the members at any time.
Using the symbols defined in, and assumptions underlying, the Formulae and Tables for Actuarial
Examinations, calculate the contribution rate required for the scheme. Ignore the possibility of new
members joining the scheme.

9. [Solution] A company is about to establish a pension scheme that will provide an age retirement benefit
of n/60ths of final pensionable salary where n is total number of years of service. Final pensionable
salary is the average salary in the three years before retirement.
An employee who will become a member of the pension scheme is currently aged 55 exact has and will
be granted exactly 20 years of past service. The employee’s salary in the year before the valuation date
was $40,000.
Basis: Pension Scheme from the Formulae and Tables for Actuarial Examinations

(a) Calculate the present value of benefits for this member (including future service).
(b) Calculate the contribution required to fund this benefit as a percentage of future salaries.

113
10. [Solution] A new member aged 35 exact, expecting to earn $40,000 in the next 12 months, has just
joined a pension scheme. The scheme provides a pension on retirement for any reason of 1/60 th of final
pensionable salary for each year of service, with fractions counting proportionately. Final pensionable
salary is defined as the average salary over the three years prior to retirement.
Members contribute a percentage of salary, the rate depending on age. Those under age 50 contribute
4% and those age 50 exact and over contribute 5%. The employer contributes a constant multiple of
members contributions to meet exactly the expected cost of pension benefits.
Calculate the multiple needed to meet this new members benefits. Basis: Pension Scheme from the
Formulae and Tables for Examinations.

114
Solutions to Module 10 Exercises
1. Dickson et al. (2009b) (Dickson et al., 2009b) Solutions Manual.
2. Dickson et al. (2009b) (Dickson et al., 2009b) Solutions Manual.
3. Dickson et al. (2009b) (Dickson et al., 2009b) Solutions Manual.
4. [Question] First, we define a service table consisting of:

ˆ `x+t = number of members surviving to reach age x + t;


ˆ rx+t = number of members who retire age x + t; and
ˆ salary scale function sy so that sx+t
sx is the ratio of earnings in the year of age x + t to x + t + 1
to the earnings in the year of age x to x + 1.

Furthermore, define the average of the final 3-year salaries for which the Final Pensionable Salary is
based on:
1
zx = (sx−3 + sx−2 + sx−1 ) .
3
Denote by ārx the value of a life annuity of $1 per annum to a retiree aged exactly x.

(a) Retirement benefits related to past service:


Assume that retirements take place uniformly over the year of age between 60 and 65, and that
retirements for those who attain age 65 takes place at exact age 65.
Consider retirement between the ages 26 + t and 26 + t + 1. The actuarial present value of the
age retirement benefits related to past service, for 34 ≤ t ≤ 38, is

z26+t+1/2 v 26+t+1/2 r26+t r z ra


5 5 C 26+t
· 50000 26
ā26+t+1/2 = · 50000 · s
,
60 s25.25 v `26 60 D26

where
z
C ra
26+t = z26+t+1/2 v
26+t+1/2
r26+t ār26+t+1/2
and
s
D26 = s25.25 v 26 `26 .
For retirement at exact age 65, the APV of the benefits is expressed as

5 z65 v 65 r65 r 5 z ra
C 65
· 50000 26
ā65 = · 50000 · s
,
60 s25.25 v `26 60 D26

where
z
C ra 65 r
65 = z65 v r65 ā65 .

Thus, summing over all the possible ages at retirement, we have the value
P38 z ra
5 C +z C ra 5 z
M ra
· 50000 · t=34 s 26+t 65
= · 50000 · s 60 ,
60 D26 60 D26

where
39
X
z
M ra
60 =
z
C ra
26+t .
t=34

Retirement benefits related to future service:


Similarly, assume that retirements take place uniformly over the year of age between 60 and 65,
and that retirements for those who attain age 65 takes place at exact age 65.
If retirement takes place between ages 60 and 61, the number of future years of service to count
is 34. Moreover, the member joined the scheme at age 21 and is now age 26, so he has already

115
served 5 years. Therefore the number of future years that can count is capped at 40-5 = 35 from
age 61 onwards. Therefore, if retirement takes place at age 61 or after, the number of future years
of service to count is 35 (since there is a max of 40).
For retirement between ages 60 and 61, the APV of the retirement benefits is:

34 z60+1/2 v 60+1/2 r60 r 34 z ra


C 60
· 50000 ā60+1/2 = · 50000 · .
60 s25.25 v 26 `26 60 sD
26

For retirement in later years, the formula is similar to the one above, with 35 in place of 34.
Summing all these together gives:
50000 50000
s
[35 (z C ra z ra z ra z ra
60 + C 61 + · · · + C 65 ) − C 60 ] = (35z M ra z ra
60 − C 60 )
60 D26 60 ·s D26

where
5
X
z
M ra
60 =
z
C ra
60+t .
t=0

(b) In addition to the notations in the service table already defined, let us additionally define d26+t
to be the number of members dying between ages 26 + t and 26 + t + 1.
Usually, we assume that the death happens in the middle of the year, and the salary also increases
uniformly over the year. However, now it is clearly stated that the salary is increased at 1st April
and then remain constant over the year, which requires us to adopt different analyses. The $5000
actually corresponds to the rate of salary from 1st April 2004 to 31st March 2005, which is the
annual salary s25.25 . In the same way, for d26+t which uniformly happens at 1st June, the annual
rate of salary at death is equal to the salary rate at 1 April of age 26 + t, which is equal to the
annual salary s26.25+t .
Assume that deaths take place on average in the middle of the year of age. For deaths between
ages 26 + t and 26 + t + 1, the APV of the death benefit can be expressed as
s d
s26.25+t v 26+t+1/2 d26+t C 26+t
50000 · 4 · = 50000 · 4 · ,
s25.25 v 26 `26 sD
26

where
s
C d26+t = s26.25+t v 26+t+1/2 d26+t .
Adding the APV of these death benefits for all possible years of death gives us
38 s d s
X C 26+t M d26
50000 · 4 · sD
= 50000 · 4 · sD
,
t=0 26 26

where
38
X
s
M d26 = s
C d26+t .
t=0

5. [Question] Let S denote the salary earned in the year of the valuation date. What follows are some
useful definitions and symbols used:

ˆ Compound interest: v = 1/(1 + i) where i is the annual rate of interest.


ˆ Salary scale: the salary scale is sx where sx+t
sx is defined to be the ratio of the salary in the year
of age x + t and x + t + 1 to the salary in the year of age x and x + 1.
ˆ Service table: `x is defined as the number of lives attaining age x, rx is defined as the number
of lives retiring aged x last birthday for 60 ≤ x ≤ 64, and r65 is the number of retiring aged 65
exact, and dx is the number of lives dying between ages x and x + 1 .
ˆ Annuity values: define ārx to be the present value of an annuity of $1 per annum payable on
retirement at age x exact, using an appropriate mortality table for 60 ≤ x ≤ 65.

116
ˆ 3-yr average scale: define the 3-year average scale function zx = 1
3 (sx−3 + sx−2 + sx−1 ).

Assume that retirements which take place before age 65 are uniformly distributed over the year of age.
Consider first the age-retirement benefit arising from past service. Note that in this question, members
may retire only after they have turned 60, so we don’t consider any retirement before that and our
calculation of retirement benefit starts from 60. The value of the benefit is
4
!
5 X z60+t+1/2 v 60+t+1/2 r60+t r z65 v 65 r65 r
·S· ā + ā .
45 t=0
s55 v 55 `55 60+t+1/2 s55 v 55 `55 65

By defining
z60+t+1/2 · v 60+t+1/2 · r60+t · ār60+t+1/2 =z C ra
60+t ,

and
z65 · v 65 · r65 · ār65 =z C ra
65 ,

and
s55 · v 55 · `55 = s
D55 ,
we have the value of the benefit equal to
5 z ra z
5 X C 60+t 5 M ra
·S· s
= · S · s 60 ,
45 t=0
D 55 45 D 55

where
5
X
z
M ra
60 =
z
C ra
60+t
t=0

Next, we consider the age-retirement benefit arising from future service.


Remark: Here the situation is different from the lecture slides because no retirements can take place
before age 60. Therefore, we have to discuss the future benefits in two separate cases: (1) between 55
and 60; (2) between 60 and 65.
(1). The benefit arising from service between 55 and 60 is different from the usual calculation of future
benefits. In contrary, it is similar to the above calculation of past service benefit. This is because the
retirement can only happens after age 60. No matter what is the year of retirement after age 60, the
member can always receive 5 years of full service from 55 to 60. Therefore, the value of the benefit
arising from service between 55 and 60 is the same as the past service benefit:
z
5 M ra
· S · s 60
45 D55

(2). For year of service between 60 + t and 60 + t + 1, the calculation becomes the same as the lecture
slides because now the member can retire in any year. We have the value of the benefit expressed as
 1 z ra
1 C 60+t +z C ra z ra 
60+t+1 + · · · + C 65 S z M̄ ra
60+t
·S· 2 sD
=
45 55 45 s D55
where
z 1 z
M̄ ra z ra
60+t = M 60+t − C ra
60+t .
2
Summing over all possible values of t, that is, for t = 0, 1, 2, 3, 4, we have
4
S X z M̄ ra
60+t S z R̄ra
60
s
= ,
45 t=0 D 55 45 s D55

where
4
X
z
R̄ra
60 =
z
M̄ ra
60+t .
t=0

117
Finally, the value of the age-retirement benefits is therefore
z
10 M ra S z R̄ra
· S · s 60 + 60
.
45 D55 45 s D55

Remark: Here the coefficient of the first term is 10 instead of 5, which is different from the usual
formula in lecture slides. This is because the retirement can only happens after age 60. Due to this
constraint, the above discussions show that the benefit of the future years from 55 to 60 should be
calculated in the same way as the past service benefit, therefore having in total 10 years in the first
term.
For death benefits, assume a uniform distribution of deaths over the year of age. The value of the
death benefit can be expressed as
P9
s55+t v 55+t+1/2 d55+t s
M d55
4S · t=0 = 4S · ,
s55 v 55 `55 sD
55

where, clearly,
9
X
s
M d55 = s55+t v 55+t+1/2 d55+t .
t=0

6. [Question] First, we define a service table consisting of:

ˆ `x+t = number of members surviving to reach age x + t;


ˆ ix+t = number of members who retire due to ill-health age x + t; and
ˆ salary scale function sy so that sx+t
sx is the ratio of earnings in the year of age x + t to x + t + 1
to the earnings in the year of age x to x + 1.

Furthermore, define the average of the final 3-year salaries for which the Final Pensionable Salary is
based on:
1
zx = (sx−3 + sx−2 + sx−1 ) .
3
Denote by āix the value of a life annuity of $1 per annum to an ill-health retiree aged exactly x. Assume
that ill-health retirements take place uniformly over the year of age.
Suppose that (AS)25 denotes the salary earnings in the year of age 25 to 26. And now consider ill-
health retirement between the ages 25 + t and 25 + t + 1. The actuarial present value of the ill-health
retirement benefits related to future service, for t < 35, is

z25+t+1/2 v 25+t+1/2 i25+t i z ia


t + 1/2 t + 1/2 C
(AS)25 25
ā25+t+1/2 = (AS)25 · s 25+t ,
60 s25 v `25 60 D25

and, for 35 ≤ t < 40, we have


z ia
35 C
(AS)25 · s 25+t
60 D25
where
z
C ia
25+t = z25+t+1/2 v
25+t+1/2
i25+t āi25+t+1/2
and
s
D25 = s25 D25 = s25 v 25 `25 .
Remark: Note that here the calculation of future service benefits is different from the lecture slides.
Specifically, here t + 1/2 is the total number of years of future service at the valuation time 0. For
exmaple, if there are 2.5 years of future service, we have two ways to calculate the APV of benefits
arising from future service benefits: (1) one way (used in the lecture notes) is to split the future service
into: 1 year service in the time period (0,1), 1 year service in the time period (1,2), and 0.5 year of
service in the time period (2,3), and add the APV for all the three items together; (2) a second way is
to consider the 2.5 years of service as a whole (in this question).

118
Therefore, the APV of the benefits, in total, is expressed as
" 34 39
#
(AS)25 X z ia
X
z ia
(t + 1/2) · C 25+t + 35 · C 25+t .
60s D25 t=0 t=35

Re-writing it and simplifying we get,


" 39 39
#
(AS)25 X z ia
X
z ia
(t + 1/2) · C 25+t + (35 − t − 1/2) · C 25+t
60s D25 t=0 t=35
" 39 4
#
(AS)25 X z ia
X
z ia
= (t + 1/2) · C 25+t − (t + 1/2) · C 60+t
60s D25 t=0 t=0
" 39 4
#
(AS)25 X z ia X
z ia
= M̄ 25+t − M̄ 60+t
60s D25 t=0 t=0
(AS)25 z ia z ia 
= R̄25 − R̄60 ,
60s D25
where from the second line to the third line we applied
39 39
" t #
X
z ia
X X
z ia 1 z ia
(t + 1/2) · C 25+t = C 25+t − C 25+t
t=0 t=0
2
k=0
39 X
39 39
X
z 1 X z ia
= C ia
25+t − C 25+t
2 t=0
k=0 t=k
39 39−k 39
X X
z ia 1 X z ia
= C25+k+t − C 25+t
2 t=0
k=0 t=0
39 64−(25+k) 39
X X
z ia 1 X z ia
= C25+k+t − C 25+t
t=0
2 t=0
k=0
 
39 64−(25+k) z
X X
z ia 1 ia 
=  C25+k+t − C25+t
t=0
2
k=0
39
X
z z
= M̄ ia ia
25+t = R̄25 ,
t=0

and in the same way,


4 4
X
z
X z
(t + 1/2) · C ia
60+t = z
M̄ ia ia
60+t = R̄60 ,
t=0 t=0
with the commutations
64−x
z
X 1 z ia
M̄ ia
x =
z
C ia
x+t − Cx ,
t=0
2
and
64−x
X
z
R̄ia
x =
z
M̄ ia
x+t .
t=0

Similarly, the APV related to past service can be shown to be


39 z ia z
X 5 C 5 M ia
· (AS)25 · s 25+t = · (AS)25 · s 25
t=0
60 D25 60 D25

where
64−x
X
z
M ia
x =
z
C ia
x+t .
t=0

119
7. [Question] CT5 April 2005 Question 8
The benefits from past service can be written as:
29 1
10 X i35+t v 35+t+ 2 z35+t+ 12 1
Past Service = × 20000 35
a∗ ∗ 35 + t +
80 t=0
l35 v s34 2
z ia
10 M35
= × 20000
80 s34 D35
10 61843
= × 20000
80 6.389 × 4781
= 5061.496

Here, the salary 20000 is the earnings over the one-year period from age 35 to 36, which is represented
by the notation (AS)35 .
For the future benefits, remind that there is a minimum pension of 20/80 of Final Salary, which implies
that no matter how many future years of service there will be, in the formulae we will always include
at least 10 years for the future services. This idea is similar to the solution to Question 6. So the future
service benefits have been splitted into (i) the future service benefits arising from the first future years
(35, 45), and (ii) those arising from the future years (45, 65). Likewise, the benefits from future service
may be written as:

z ia z ia
10 M35 1 R45
Future Service = × 20000 + × 20000
80 s34 D35 80 s34 D35
10 61843 1 609826
= × 20000 + × 20000
80 6.389 × 4781 80 6.389 × 4781
= 10052.574

Recall that at age 35 this member has already had 10 years of pass service, and that there is a minimum
of 20 years service required for the pension to be paid. This means that, it will be guaranteed that 10
years of future service benefits will accrue from the next 10 years (the years from age 35 to 45). The
first term represents the EPV of benefits accrued from this 10-year service benefits. The second term
represents the EPV of benefits accrued from future service from age 45.
Therefore, the total benefits is $15,114.07.

8. [Question] Let n denote the number of years of past service.


EPV of past pensions:

(6/60)(Sal)(z M35
ia
+ z M35
ra s
)/ D35
= (6/60)(20, 000)(61, 843 + 128, 026)/s D35
20, 000
= (18986.9) s
D35

EPV future pensions:


ia ra
(1/60)(Sal)(z R35 + z R35 )/s D35
= (1/60)(20, 000)(1, 187, 407 + 3, 524, 390)/s D35
20, 000
= (78529.95) s
D35

Let C represent the contribution rate.


EPV of contributions:

C(Sal)(s N 35 )/s D35


= C(20, 000)(502, 836)//s D35

120
Hence,
20, 000 20, 000
(18986.9) sD
+ (78529.95) s = C(20, 000)(502, 836)//s D35
35 D35
C = 19.39%.

9. [Question] CT5 April 2010 Question 7

(a) We may write the age retirement benefit as:


ra
1 20 × z M55
ra
+ z R55
Age retirement benefit = × 40000
60 s54 D55
1 20 × 128026 + 963869
= × 40000
60 9.745 × 1389
= 173584

(b) Let the contribution percentage be K. We may write the contributions as:
s
N 55
Contributions = K × 40000 ×
s54 D55
88615
= K × 40000 ×
9.745 × 1389
= 261868K

Equating the two and solving for K yields:


173584
K= = 66.3%
261868

10. [Question] CT5 September 2008 Question 9


We may write the EPV of benefits as:

z ra ia
1 R + z R35
Benefits = × 40000 35s
60 D35
1 3524390 + 1187407
= × 40000
60 31816
= 98730

Note that the denominator is simply s D55 because we are given the salary over the next 12 months as
$40,000. (If we were instead given the salary over the year prior to the calculation date, the denominator
would be s34 D35 )
Now, the EPV of the employee’s own contribution is:

4%s N 35 + 1%s N 50
Employee’s contribution = 40000 × sD
35
0.04 × 502836 + 0.01 × 163638
= 40000 ×
31816
= 27345

The employer must contribute the difference; that is, 98730 − 27345 = 71385. Expressed as a multiple
of the employee’s contribution, this is 71385
27345 = 2.61 times the employee’s contribution.

121
References

Dickson, D.C.M., Hardy, M.R., Waters, H.R., 2009a. Actuarial Mathematics for Life Contingent Risks. 1st
ed., Cambridge University Press.

Dickson, D.C.M., Hardy, M.R., Waters, H.R., 2009b. Solutions Manual for Actuarial Mathematics for Life
Contingent Risks. 1st ed., Cambridge University Press.
Dickson, D.C.M., Hardy, M.R., Waters, H.R., 2013a. Actuarial Mathematics for Life Contingent Risks. 2nd
ed., Cambridge University Press.

Dickson, D.C.M., Hardy, M.R., Waters, H.R., 2013b. Solutions Manual for Actuarial Mathematics for Life
Contingent Risks. 2nd ed., Cambridge University Press.

122

You might also like