CARLETON UNIVERSITY

SCHOOL OF
MATHEMATICS AND STATISTICS

HONOURS PROJECT

TITLE: Profit Testing of Equity-Linked

Insurance Policies

AUTHOR: Jacob Lewis

SUPERVISOR: Dr M Ould Haye

DATE: April 13, 2020

Contents
1 Introduction 2

2 Notation 3

3 Life Insurance Policies 4

3.1 Traditional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Modern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

4 Profit Testing 8
4.1 Profit Testing Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.1 Expenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.2 Mortality model . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.3 Probability of Surrender . . . . . . . . . . . . . . . . . . . . . . 11
4.1.4 Interest rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.1.5 Expected costs from benefits paid . . . . . . . . . . . . . . . . . 12
4.1.6 Reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.2 Deterministic Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2.1 Emerging Surplus . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2.2 Setting Reserves . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2.3 Equity-Linked Life Insurance Profit Testing . . . . . . . . . . . 14
4.3 Stochastic Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.3.1 Determining a Sample of Interest Rates . . . . . . . . . . . . . . 15
4.3.2 Changes to the Deterministic Approach . . . . . . . . . . . . . . 16
4.3.3 Reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

5 Application 18
5.1 Term Life Insurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.2 Deterministic Testing of an Equity-Linked Insurance Policy . . . . . . . 22
5.3 Stochastic Testing of an Equity-Linked Insurance Policy . . . . . . . . 25

6 Conclusion 30

7 References 32

A Life Table Excerpt - Population of Canada, Both Sexes, Age 55 to 75 33

B Code used to Simulate the Stochastic Profit Test 1000 times 34

C Inputs for EQUITYLINKED Function 37

D CTE Function 38

E t px function 39

1
1 Introduction
When insurance companies offer life insurance to potential clients, they are offering
an opportunity for the policyholder to reduce their own risk by paying the insurer to
take it for them. The insurer must then evaluate this risk and price their products
accordingly. Traditionally, actuarial methods would involve the use of the expected
present value approach to assess a contract and determine premiums and reserves.
This approach comes with many problems as it is not flexible with variable interest
rates or profit measures, and other more complex insurance contracts. Every change in
the variables requires a manual recalculation of all other variables. Thus, we introduce
cash flow analysis, also called profit testing when used for life insurance, as a method
to determine profit, set reserves, and set premiums, while allowing for variability in
interest rates and payment methods. This technique forecasts the income and expenses
of an insurance company which relate to the product of interest and finding the net
present value of these cash flows. Furthermore, profit testing can still be used to value
a product even when interest rates are uncertain, simulating interest rates will generate
a distribution of net present values, with which one can determine profitability of an
insurance product.

2
2 Notation

Mortality Model

t px Probability of a person age x surviving to age x + t

t qx Probability of a person age x dying before age x + t

Kx Random variable for the curtate future lifetime of a person at age x

Tx Random variable for the future lifetime of a person at age x

Insurance Policy

ν Discount rate

Πt Profit signature at time t

tV Reserves held at time t for a policy

CVt Cash for surrender payable at time t if the policyholder surrenders the contract
in the period [t − 1, t)

E tV Expected cost of setting the reserves at time t − 1 for a policy

Et Expenses incurred at time t

EDBt Expected cost of death benefits in the period [t − 1, t), paid at time t

ESBt Expected cost of surrender benefits in the period [t − 1, t), paid at time t

Ft Amount invested in the policyholder’s fund

It Interest earned on insurer’s assets from time t − 1 to t

N P V Net present value

Pt Premium paid at time t − 1

P rt Emerging surplus for the period [t − 1, t) at time t

St Benefit to be paid at time t if the policyholder dies in the period [t − 1, t)

3
3 Life Insurance Policies
This section will provide a basic understanding of how life insurance policies work
and how the insurer might assess such policies in order to make them profitable. Life
insurance policies are an agreement between the policyholder and the insurer. The
policyholder pays the insurer an amount to the insurer so that in the future, should the
policyholder die, the insurer will pay a benefit, denoted by S or St to the beneficiary.
The payments are called premiums, denoted by P or Pt , and can either be paid as a lump
sum, or as an annuity to the insurer. Typically the premiums are paid as an annuity
and the benefit is a lump sum paid at the end of the year of death, however these can
be reversed where the premium is a lump sum paid at the start of the contract and the
benefits paid either as a lump sum or an annuity. The insurer uses the premiums first
to pay its expenses, then uses the remaining amount to either invest or hold in reserve
in case of a claim. To set the premiums and reserves, the insurer makes assumptions
about the mortality of the policyholder, the expenses, and the interest rates.

3.1 Traditional
The most common types of policies are the whole life and term insurance policies. In the
whole life case, the policyholder will pay premiums regularly until the year of death,
when the insurer pays the benefit amount to the beneficiary. A term life insurance
policy is similar except upon maturity, the policy ends and the premiums and benefits
will not be paid. When the insurer evaluates these policies, they must predict the
expected present value, also called the actuarial present value, of both the premiums
and the benefits. We denote Z as a random variable for the present value of a benefit
payable at the end of the year of death, defined by

Z = e−δKx +1

where Kx is a random variable for the curtate future lifetime of a person age x and δ is
the continuously compounding interest rate. Then the AP V of a whole life insurance
policy with a benefit of $1 is denoted by Ax and is defined

Ax = E[Z] = E[e−δ(Kx +1) ]. (3.1)

The AP V of the premiums is also evaluated, the present value random variable of an
annuity due is denoted Y and is defined

1 − e−δ(Kx +1)
Y = ä Kx +1 = . (3.2)
d
Then the AP V of level premiums of $1 paid annually is denoted äx and is defined

1 − e−δ(Kx +1) 1 − E[e−δ(Kx +1) ]

äx = E[Y ] = E[ ]= (3.3)
d d

4
where d is the discount rate. Then by equation (3.1), we can see that

1 − Ax
äx = (3.4)
d
Similarly, the AP V of an n-year term life insurance policy with a benefit of $1 is denoted
1
by Ax:n and is defined
1
Ax:n = E[Z] (3.5)
where Z is now defined as
(
e−δTx if Kx + 1 ≤ n,
Z=
0 if Kx + 1 > n.

The term life AP V can also be expressed as

Ax:n = Ax − Ax+n .

The P V of the annuity of premiums paid until death or time n is denoted Y , and
defined (
ä Kx +1 if Kx + 1 ≤ n,
Y = (3.6)
ä n if Kx + 1 > n.
Then the AP V of the term life annuity due with payments of $1 is denoted äx:n and is
defined
äx:n = E[Y ] (3.7)
where Y is defined as in equation (3.6).

The insurer is then able to set premiums using the net present value of the
loss. We define the future loss at time t as the amount the insurer will pay in an
insurance policy, and denote it as Lt . The random loss at time 0 is defined as

L0 = SZ − P Y (3.8)

where S is the benefit amount, Z is the present value of a benefit of $1, P is the
premium amount, and Y is the present value of an annuity of $1. Then taking the
expected value and setting E[L0 ] = 0 we can solve for the premium amount, where this
amount gives a net present value of 0 for the policy.

E[L0 ] = SE[Z] − P E[Y ].

Setting E[L0 ] equal to zero and rearranging for P , we find

E[Z]
P =S .
E[Y ]

5
The future loss random variable is also used to calculate the reserves. We define Lt as

Lt = PV(Future Benefits at time t) - PV(Future Premiums at time t)

The expected future loss at time t gives the amount the insurer must hold in order to
be able to pay for the expected costs. This number is called the reserve or the policy
value. We denote the policy value at time t for a policyholder at age x as t V and it is
defined as

tV = APV(Future Benefits at time t) - APV(Future Premiums at time t).

In the case of a whole life insurance policy, in actuarial notation this is written

tV = SAx+t − P äx+t

This is the method that an insurer would use to set premiums and reserves and with
some modification this allows for expenses, investments, and profit. The problem with
this process is that it does not allow for flexible premiums or interest rates. A change
would mean a manual recalculation of all parts of the policy.

3.2 Modern
Insurance companies offer alternative forms of insurance which provide the policyholder
the opportunity to make an investment with the insurance company. Participating
insurance is similar to term and whole life insurance except the insurer shares some
of the investment income made on the premiums with the policyholders. Another
common life insurance policy is Universal Life insurance, which is similar to Partic-
ipating insurance, but with the added flexibility of premiums. The policyholder can
change the premium amount that they wish to pay, as long as it is still enough to
cover the insurance benefit set. The insurer pools the premiums with its own in-
vestments and pays a portion of the returns back to the policyholder in regular intervals.

The most investment focused product that a life insurance company offers is the
equity-linked insurance policy. Unlike the previously mentioned policies, equity-linked
insurance is primarily an investment, with the added benefit of life insurance. The
insurer offers many different investment funds to the policyholders that have different
expected returns and riskiness, and the policyholder chooses which fund their premiums
will be paid into. The policyholder also has the option to choose the amount and
number of premiums paid into this fund, with a minimum amount set by the insurer to
make sure the expenses are covered. The insurer also offers the option of guaranteed
minimum benefits upon maturity or upon death of the policyholder.

A key difference between equity-linked insurance and other life insurance types
is that these premiums must be held separately from the insurance company’s assets.

6
For the insurer this means there is less need to hold reserves for these contracts, as the
policyholder fund acts as a reserve itself. The risk of the investment is taken on by
the policyholder, however, risk for the insurer is added when the guaranteed minimum
benefits are added to the policy.

7
4 Profit Testing
4.1 Profit Testing Model
To begin profit testing, a basis of assumptions and details must be made by the insurer.
First of these is the time interval. The time interval is set by an insurer to determine at
what time the cash flows occur. The insurer can choose any interval they like, such as
monthly or annually, as long as they use the appropriate discount rate for calculation.
It is most convenient for the insurer to use the same interval that the premiums are
paid in, unless the premium payments do not follow any defined structure, in which
case monthly or annual cash flows would be the most convenient.

The insurer must also determine what is called the profit test basis. This is
formed by assumptions made about the profit test model such as expenses, the
mortality model, and interest rates.

4.1.1 Expenses
Usually when a contract is written by an insurer there are initial expenses, also called
acquisition costs, and recurring expenses that continue for the duration of the contract.
The initial expenses must always be allocated to time 0, as any expenses must be
allocated as early as possible. The later expenses are costs arising from the maintenance
of the policy and collection of premiums are much smaller than the acquisition expenses
and are allocated to the beginning of the period they are paid. After defining the
expenses, the insurer can set the amount to charge from the premiums paid.

4.1.2 Mortality model

An insurer must also make assumptions about the mortality model for policyholders
as well as the probability of the policyholder withdrawing from the contract. Often
life tables can be used to estimate mortality of the policyholder as shown in Appendix
A, as they are very easy to derive probabilities of survival for each year. An alterna-
tive would be to use a distribution based on the force of mortality µx demonstrated here.

Let Fx be the distribution function of Tx , then

Fx (t) = P r[Tx ≤ t]. (4.1)
Fx represents the probability of a person aged x dies before age x + t. We can also
define the survival probability, Sx .
Sx (t) = P r[Tx > t] = 1 − Fx (t). (4.2)
We denote the probability density function of the random variable Tx , as fx and its
relationship to Fx and Sx is shown by
d d
fx (t) = Fx (t) = − Sx (t). (4.3)
dt dt

8
An important property for the random variable Tx is that the probability a person aged
x dies within t more years is equal to the probability a person aged 0 dies within x + t
years given that they survive x years.

P r[Tx ≤ t] = P r[T0 ≤ x + t|T0 ≤ x]. (4.4)

Now by definiton

P r[x < T0 ≤ x + t]
P r[Tx ≤ t] = P r[T0 ≤ x + t|T0 ≤ x] = ,
P r[T0 > x]

which is equivalent to
F0 (x + t) − F0 (x)
Fx (t) = ,
S0 (x)
and using equation (4.2)
S0 (x + t)
Sx (t) = . (4.5)
S0 (x)
The force of mortality for a life at age x is denoted µx and is defined as
1
µx = lim+ P r[T0 ≤ x + dx|T0 > x].
dx→0 dx
or, by equation (4.4),
1
µx = lim+ P r[Tx ≤ dx].
dx→0 dx

Then we relate µx to Sx by equations (4.1) and (4.2)

1
µx = lim+ (1 − Sx (dx)). (4.6)
dx→0 dx
Recall that by equation (4.5), we can rewrite this as
 
1 S0 (x + dx)
µx = lim+ 1−
dx→0 dx S0 (x)
1 S0 (x) − S0 (x + dx)
= lim+
dx→0 dx S0 (x)
1 S0 (x) − S0 (x + dx)
= lim
S0 (x) dx→0+ dx
1 d
= (− S0 (x))
S0 (x) dx

and finally,
1 d
µx = − S0 (x). (4.7)
S0 (x) dx

9
which can also be written
f0 (x)
µx = .
S0 (x)
By the property stated in equation 4.5, we can define µx+t similarly. We assume x
constant and t variable, then d(x + t) = dt.

1 d
µx+t = − S0 (x + t)
S0 (x + t) d(x + t)
1 d
=− S0 (x)Sx (t)
S0 (x)Sx (t) dt
1 d
=− Sx (t)
Sx (t) dt

resulting in
fx (t)
µx+t = . (4.8)
Sx (t)
Now we want to write Sx (t) in terms of µx+t . We rewrite equation (4.7) as

d
µx+t = − log Sx (t)
dx
Then integrating both sides over 0 to y gives
Z y
µx+t = log Sx (0) − log Sx (y)
0

and since Sx (0) = 1,  Z y 

Sx (y) = exp − µx+t dt (4.9)
0

From this equation, we are able to define a force of mortality and determine a mortality
model. In 1825, Benjamin Gompertz proposed that the force of mortality increases
exponentially with age, that is
µx = Bcx .
where B and c are constants such that 0 < β < 1 and c > 1. Later William Makeham
added to this theory, saying that there should also be a fixed term in the force of
mortality, independent of age.
µx = α + Bcx .
So from these examples we see ways in which an insurer could model mortality using
the force of mortality.

10
4.1.3 Probability of Surrender
Depending on the type of policy, the insurer may also have to determine the probability
of the policyholder withdrawing from the contract in any given period. This can be done
based on historical results and forecasting. In such cases the insurer must use stochastic
modelling for multiple states of the policyholder. In a 3-state model we denote state 0
as the survival with policy in force state, state 1 as survive and withdraw from policy,
and state 2 as death. In this case the insurer is only concerned with transitions from
state 0, so we use a probability vector for these probabilities. The single step probability
vector for a person age x using this model is defined as
 00 01 02 
px px px . . (4.10)

Under the assumption of independent and identically distributed lives, we can use this
probability vector to predict how many policies are still in force in any given year.
Then these probabilities can be used to define many terms in the profit test.

4.1.4 Interest rates

The insurer must also determine multiple interest rates to use in the profit test. First
is the discount rate used to calculate the net present value of all of the cash flows in
each period. This rate can be determined using the weighted average cost of capital,
rW ACC . To calculate rW ACC you must use the cost of equity, rE , the cost of debt, rD ,
and the cost of preferred shares, rP f d , as well as the weights associated with these rates,
wE , wD , wP f d . The cost of equity can be found using the capital asset pricing model,
which says that

rE = risk free rate + βfirm equity × market risk premium.

The cost of debt is found simply by taking the interest rate associated to the outstanding
debt for the firm. Then the effective cost of debt is found using the cost of debt and
the corporate tax rate for the firm, Tc .

effective cost of debt = rD × (1 − Tc ).

The cost of preferred equity is calculated using the dividends paid per share of preferred
stock, Div1 , the price per share of preferred stock, P0 , and the growth rate of the
dividends, g.
Div1
rP f d = + g.
P0
The weights are calculated as the percentage of the firm value financed by equity,
preferred stock, and debt. Now we can calculate rW ACC then we use this rate to find
the discount rate ν.

rW ACC = wD × rD (1 − Tc ) + wE × rE + wP f d × rP f d .

11
The discount rate is defined as
1
ν= .
1 + rW ACC
The insurer must also determine what interest is earned on its assets. We denote the
amount of interest earned in the period [t − 1, t) at time t as It . If we say the interest
rate earned on insurer assets over period [t − 1, t) is it , then the interest earned is

It = Amount invested × it . (4.11)

Finally, an insurer may also choose to invest the reserves held in short term, low risk
assets. The rates on these assets must also be accounted for in the profit test.

4.1.5 Expected costs from benefits paid

Once the profit test assumptions are made, the insurer must then find the expected costs
in each period in order to determine the emerging surplus. These costs are from the
multiple states in which a policyholder could be in each period. Using the probability
vector from equation (4.10), we define the following: The expected death benefit in
year t, EDBt
EDBt = p02 x+t−1 × St , (4.12)
the expected cost of surrender (or withdraw) benefits, ESBt

ESBt = p01
x+t−1 × CVt , (4.13)

and the expected cost of survivor benefits (also called endowment) EEBt

EEBt = p00
x+t−1 × Endowment. (4.14)

Usually an endowment occurs at the end of the contract, in the case of equity-linked
life insurance this could be a percentage of the policyholder’s fund, Ft . This method
can be extended to any number of states where the transition probabilities between
other states may become necessary, and a probability matrix must be formed to predict
future proportions of policyholders in each state.

4.1.6 Reserves
The expected cost of reserves is also calculated using the survival model, which is
considered a cost at the end of the period, then the actual reserve is carried forward
as income at the start of the next period. To find the reserves we want to make the
expected value of the policy equal to 0 in each period. The net present value of a policy
in the period [t − 1, t) is given by

N P Vt = E[Benef itst ] + E[Expensest ] − E[P remiumst ] − t V. (4.15)

12
where Benef itst , Expensest , and P remiumst each represent the future amounts, start-
ing from time t to the end of the term. We set N P Vt = 0 for every t to find the reserve
in each period. Note that reserves can never be negative, in such cases we set the
reserve equal to zero. Then the expected reserve for the period t is defined as

E t V = p00
x+t−1 t V. (4.16)

Note that in the final policy period of whole life and term insurance policies, the reserve
and expected reserve should be zero, since there is no benefit payment to be made in
the future.

4.2 Deterministic Approach

In this section we look at the deterministic approach to cash flow analysis to show the
basic ideas behind the method. We assume that interest rates are constant over the
duration of the policy term.

4.2.1 Emerging Surplus

Using the cash flows defined in the previous section, the emerging surplus for the insurer
in each year is defined as

P rt = t−1 V + Pt − Et + It − EDBt − ESBt − EEBt − E t V. (4.17)

We define the profit signature in period t, Πt , as the expected surplus emerging at the
end of period t for a contract in force at the issue date, time t = 0 and is calculated as
follows,
Πt = p00
x+t−1 P rt . (4.18)
Then the net present value of a policy in force up to time t can be found.
t
X
NP V = Πk ν k . (4.19)
k=0

With the net present value computed, it can be compared to the desired profit measure
to determine if the product if profitable. Often this profit measure is the net present
value itself, but alternatives include the internal rate of return, discounted payback
period, and the profit margin.

4.2.2 Setting Reserves

The profit test can be used to set the reserves through a method called zeroization. We
do this by first performing a profit test without reserves, then starting with the final
period of the contract, set the reserve in the previous period such that the emerging

13
surplus of the period is equal to 0. If we assume an interest rate of j on the amount
held in reserves, then the amount needed at the end of the period to set the emerging
surplus equal to zero is  
P rt
n−1 V = max − ,0 . (4.20)
1+j
For every period before the final period, the expected reserve must also be accounted
for. We find the reserve of the periods before the final one using the following formula.
 
P rt
t V = max − + E t V, 0 (4.21)
1+j
where E t V is defined as in Equation (4.16). This pattern is repeated for every period
of the contract working backwards from the final period to the first period. Once the
zeroized reserves have been set, the net present value can be recalculated. This is done
because the emerging surplus will come at an earlier point in time, and is therefore
more valuable due to discounting.

4.2.3 Equity-Linked Life Insurance Profit Testing

Equity-linked life insurance policies add multiple factors to the calculation of emerging
surplus. Since the insurer often takes a percentage of the policyholder’s fund to cover
expenses, called the management charge, or M Ct (this is also often called the manage-
ment expense ratio, MER), the first step must be to determine what the policyholder
fund will be in each period. To begin, the premiums paid are split into what is allocated
to the policyholder’s fund, APt , and what is allocated to the insurer’s fund (or unal-
located to the policyholder’s fund), U APt . The policyholder fund starts with the first
allocated premium at the beginning of the period, accrues interest, then the manage-
ment charge is subtracted at the end of the period, which results in the policyholder’s
fund amount.
F1 = AP1 × (1 + i) − M C1 .
Then the future fund values will add the previous value to the allocated premium before
interest and management charges are accounted for.

Ft = (Ft−1 + APt ) × (1 + i) − M Ct . (4.22)

Once the policyholder fund is determined, the insurer can then calculate the emerging
surplus from it’s own cash flows. This is done similarly to a traditional life insurance
policy except instead of collecting the premiums, the insurer collects the unallocated
premiums and management charges. Equation (4.17) becomes

P rt = t−1 V + U APt + M Ct − Et + It − EDBt − ESBt − EEBt − E t V.

Since the management charge is usually a percentage of the policyholder fund, we can
denote this as M ER,
M Ct = M ER(Ft−1 + APt )(1 + i),

14
and then we change the equation (4.22)
Ft = (Ft−1 + APt ) × (1 + i) × (1 − M ER).
Often are no reserves needed for an equity-linked life insurance policy, due to the
policyholder fund being the benefit being paid. Also there may be no difference between
surrender benefits and endowment benefits, if the insurer chooses to do so. In such cases
the calculation of emerging surplus simplifies to
P rt = U APt + M Ct − Et + It − EDBt − ESBt .

The guaranteed minimum maturity benefit (or death benefit) can increase risk for the
insurer. In a profit test this is included after the net present value of the cash flows
is calculated. In the case the guaranteed minimum maturity benefit (or death benefit)
exceeds the policyholder’s fund at the time the benefit must be paid, then this difference
must be subtracted from the emerging surplus of the final period of the contract. If we
suppose the policyholder survives until the end of the policy term, period n, then
( t
)
X
EEBt = t px max Ft , GM M B × Pk . (4.23)
k=1

Similarly for the guaranteed minimum death benefit, if the policyholder dies in period
t, ( )
Xt
EDBt = qx+t max Ft , GM DB × Pk (4.24)
k=1

4.3 Stochastic Approach

Due to the variable nature of interest rates, it is not enough to simply take the expected
interest rate and use it to perform a profit test on equity-linked life insurance type
policies. Instead an insurer must account for this variability by making an assumption
about the distribution of the interest rates on the policyholder’s assets, and use it to
generate a large sample. This sample can then be used to perform a large number of
profit tests, and will give a sample of net present values, from which sample statistics
can be drawn.

4.3.1 Determining a Sample of Interest Rates

If we assume the interest rates are independent, identically distributed, and follow a
lognormal distribution with mean µ and variance σ 2 , LN (µ, σ 2 ), with n periods in
the term of the policy, we generate a sample with size n from the standard normal
distribution, N(0,1). Then by definition of the lognormal distribution, the sample of
interest rates, ik , is found by
ik = eµ+σzk , (4.25)
where zk is the k th element from the sample of standard normal random variables.

15
4.3.2 Changes to the Deterministic Approach
Once the interest rates have been determined the insurer can use them to perform a
profit test. This is done very similarly to the deterministic approach, changing the
interest rates in each period. When we examine the equity-linked life insurance policy,
the effect of the variable interest rates on the policyholder’s fund can be very large. This
is especially important when there are guaranteed minimum maturity or death benefits
offered in the policy. The formula for calculating the policyholder’s fund becomes

Ft = (Ft−1 + APt ) × (1 + it ) × (1 − M ER) (4.26)

and the rest of the calculation continues in the same way as demonstrated in section
(4.2). This process is then repeated as many times as the insurer wishes to get a
sufficiently large sample of independent net present values. Statistics can be drawn
from this distribution to determine profit measures and assess if a product is worth
accepting for an insurer. For example, an insurer may choose to accept a policy if, after
1000 simulations, the 5th percentile is positive. An insurer may also be interested in a
confidence interval for the expected value of the N P V . For a sufficiently large sample
size, N , we can apply the Central Limit Theorem, and say a (1−α%) confidence interval
for E[N P V ] is given by
 
s s
N P V − zα/2 √ , N P V − zα/2 √ ,
N N
where s is the sample standard deviation of N P V .

4.3.3 Reserves
Insurers must use alternative methods to determine reserves when the risk associated
with the policy is nondiversifiable, as is the case with equity-linked life insurance.
The two most common methods are called quantile reserving and conditional tail
expectation reserving (CTE), also commonly known as Value-at-Risk (VaR) and Tail
Value-at-Risk (or TVaR) respectively.

The Value-at-Risk for a specified parameter α, where 0 ≤ α ≤ 1, is the amount

such that the probability that a loss exceeds the reserve is equal to 1 − α. That is, for
a random loss amount L with continuous distribution function FL , we set the quantile
reserve Qα such that,
P r[L ≤ Qα ] = α. (4.27)
Alternatively,
Qα = FL−1 (α). (4.28)

The random loss, L, is defined by the future expected profits. Using the sets of interest
rates simulated, each set will determine one element of the sample of loss random

16
variables, Li . Let j denote the rate for interest earned on reserves, and let t p00
x be the
probability the policy is still in force at t and P rt,i the profit emerging in period t of
simulation i. Then n 00
t−1 px P rt,i
X
Li = − t
(4.29)
t=1
(1 + j)
This will give an empirical distribution for L from which the reserve 0 V can be drawn.
This method of reserving does not require any changes after setting the initial reserve
provided that the assumptions made (see sections 3.2.1 to 3.2.4) are accurate. Then
quantile reserving will give an amount such that the insurer will be able to cover the
expected benefits with probability 0.95. This does not mean that the insurer should
not adjust the reserve, in fact the insurer can adjust the reserve based on the result
from the interested earned on the policyholder fund. If the fund does well, this means
the insurer is less likely to have to pay the guaranteed minimum, as the fund will
likely exceed the minimum. The insurer could then decrease the reserves held in order
to profit more from other assets. Similarly if the fund does poorly, the insurer must
increase the reserves in order to be able to pay the expected costs.

The conditional tail expectation reserving method, uses the conditional expecta-
tion of the random loss L given that the loss is greater than the quantile reserve, Qα .
That is, for a given level α, the CTE reserve is given by

CT Eα = E[L|L > Qα ] (4.30)

CTE reserving looks at what happens if the loss is to actually exceed the quantile
reserve, and set the reserves based on the tail distribution of the loss. This leads to
a higher amount set for the reserves, and decreases risk for the insurer, at the cost of
some potential profit earned on other assets.

17
5 Application
In this section we will see how an insurer would profit test for a term life insurance
policy using a deterministic method, then compare the deterministic method to
the stochastic method for an equity-linked insurance policy. First we must make
our assumptions about the model. The mortality model used is the life table for
all Canadians provided by Statistics Canada over the years 2016-2018. All other
assumptions about the model are derived from the 2019 Annual Report of Canada Life
Insurance. These assumptions are a rough estimate based on the total expenses, total
investment income, total investments, and total premiums provided in the Annual
Report. The information about the Core Conservative Growth segregated fund, such
as historical returns, guaranteed minimums offered, and management expense ratio,
can be found in the Canada Life segregated funds information folder.

The expenses depend on the policy, for term insurance we assume the initial ex-
pense is $98.85 per policy, with annual expenses of $107.10, and for the segregated
fund policy, expenses depend on the fund chosen and premiums paid. We assume there
is a 5% commision owed to the advisor as an initial expense. The annual expense
estimated for segregated fund insurance is $1.70. We assume the insurer’s assets earn
interest at 7.7% annually, and reserves are held as cash with no interest earned. The
discount rate used to calculate any net present values is 10.3%. The policyholder fund
in the segregated fund policy is the Core Conservative Growth fund, with guaranteed
minimum maturity/death benefits of 75%/100%, offers an expected return of 3.21%
and standard deviation of 2.16%. The management expense ratio, or management
charge, of this fund is 2.48%.

Summary of Model Assumptions and Data

Table 1: 10-year Term Life Insurance

Name Amount
Initial Expense $98.85
Annual Expenses $107.10
Interest Rate 7.7%
Discount Rate 10.3%
Age of Policyholder 65
Amount Insured $100000
Premium Amount $1500

18
 Table 2: 10-year Equity Linked Insurance

Name Amount
Initial Expense 0.05P1
Annual Expenses $1.70
Initial Premium $500
Annuity Premium $100
Age of Policyholder 65
Expected Interest Rate of Fund 3.21%
Standard Deviation of Fund Interest Rate 2.161%
Guaranteed Minimum Maturity Benefit 75%
Guaranteed Minimum Death Benefit 100%
Management Expense Ratio 2.48%
Confidence Level 0.05
CTE Level 0.75

In the following examples year 5 will be shown as an example calculation, as well as

any years where the calculation is different from the other years. The policies will
be evaluated with the following profit measures: Net Present Value, Internal Rate of
Return (IRR), Discounted Payback Period (DP P ), and Profit Margin.

The IRR is defined the solution to the following equation,

n
X
t
0= Πt νIRR
t=0

where
1
νIRR = .
1 + IRR
The solution for IRR can be computed in Excel using the command IRR(). It is
important to note that in some cases it is impossible to find a solution to this equation.

The DP P is the amount of time required for an investment to break even (ie.
N P V ≥ 0). To find this we simply look at the N P V column and set the DP P equal
to the time at which the N P V first becomes nonnegative.

The profit margin is ratio of the net present value and the actuarial present
value of the premiums.
N P Vn
Profit margin = P
n
Pt t p x ν t
t=0

19
5.1 Term Life Insurance
In the first example we are an insurer offering a 10-year term life insurance policy for a
person age 65 with a death benefit of $100000 payable at the end of the year of death.
Level premiums of $1500 are paid annually with the first payment at the beginning of
the first year. To begin we set up a table with columns: Time, Premiums, Expenses,
Interest, Probability of Death, Expected Death Benefit, Profit, Profit Signature, and
Net Present Value. The first 5 columns are all in the assumptions and can be entered
immediately as seen in Table (3).

Table 3: Assumptions of Term Life Policy

Time Premium Expenses Interest qx
0 $0 $98.85 $0 0
1 $1,500.00 $107.10 $115.49 0.00918
2 $1,500.00 $107.10 $115.49 0.01009
3 $1,500.00 $107.10 $115.49 0.0111
4 $1,500.00 $107.10 $115.49 0.01222
5 $1,500.00 $107.10 $115.49 0.01346
6 $1,500.00 $107.10 $115.49 0.01484
7 $1,500.00 $107.10 $115.49 0.01637
8 $1,500.00 $107.10 $115.49 0.01808
9 $1,500.00 $107.10 $115.49 0.01997
10 $1,500.00 $107.10 $115.49 0.02208

Table 4: Calculations
Time EDB Profit Profit Signature NP V
0 $0.00 -$98.85 -$98.85 -$98.85
1 $918.00 $590.39 $584.97 $642.00
2 $1,009.00 $499.39 $494.35 $1,068.20
3 $1,110.00 $398.39 $393.97 $1,383.57
4 $1,222.00 $286.39 $282.89 $1,593.83
5 $1,346.00 $162.39 $160.20 $1,704.39
6 $1,484.00 $24.39 $24.03 $1,719.79
7 $1,637.00 -$128.61 -$126.50 $1,644.52
8 $1,808.00 -$299.61 -$294.19 $1,482.00
9 $1,997.00 -$488.61 -$478.85 $1,236.37
10 $2,208.00 -$699.61 -$684.16 $910.51

The expected death benefit in each year is determined by Equation (4.12). In this case
we assume there is no probability of surrender, so we have p02
x+t−1 = qx .

EDB5 = Sqx+5 = 100000(0.01346) = 1346.

20
Once the expected death benefit is calculated, the emerging surplus is then calculated
using Equation (4.17).
P r5 = P5 − E5 + I5 − EDB5 .
= 1500 − 107.1 + 115.49 − 1346
= 162.39
Then the profit signature is calculated using Equation (4.18).
Π5 = px+5 P r5
= (1 − qx+5 )P r5
= (1 − 0.01346)(162.39)
= 160.2
Finally the N P V is calculated using Equation (4.19).
5
X
N P V5 = Πk ν k
k=0
 1  5
1 1
= −98.85 + (584.97) + ... + (160.2)
1.103 1.103
= 1704.39
We note in Table (4) that the net present value of the policy is always positive after
the first period, but in the later years of the policy there is negative emerging surplus,
which must be avoided for proper risk management. We use the profit test to calculate
the reserves using zeroization as shown in Section (4.2.2). Since there is no interest
earned on reserves, we set 9 V = −P r10 = $699.61. We then calculate backwards and
find the values shown in Table (5).

4V = max{−P r5 + E 5 V, 0}
= −162.39 + 1489.63
= 1327.24.
Note that holding reserves decreases the N P V , but the emerging surplus is nonnegative
for every year after the initial expenses which will decrease the risk for the insurer in
the future years.

Summary of Profit Measures

N P V = $542.40
IRR = 383%
DP P = 1
$542.40
Profit margin = = 4.99%
$1500a65:10

21
 Table 5: Reserve Calculations
Time P rt Before t−1 V Et V P rt After Πt NP V
0 -$98.85 $0.00 $0.00 -$98.85 -$98.85 -$98.85
1 $590.39 $0.00 $108.27 $482.12 $477.70 $542.40
2 $499.39 $109.27 $608.66 $0.00 $0.00 $542.40
3 $398.39 $614.87 $1,013.26 $0.00 $0.00 $542.40
4 $286.39 $1,024.63 $1,311.02 $0.00 $0.00 $542.40
5 $162.39 $1,327.24 $1,489.63 $0.00 $0.00 $542.40
6 $24.39 $1,509.95 $1,534.34 $0.00 $0.00 $542.40
7 -$128.61 $1,557.46 $1,428.85 $0.00 $0.00 $542.40
8 -$299.61 $1,452.63 $1,153.02 $0.00 $0.00 $542.40
9 -$488.61 $1,174.25 $685.64 $0.00 $0.00 $542.40
10 -$699.61 $699.61 $0.00 $0.00 $0.00 $542.40

5.2 Deterministic Testing of an Equity-Linked Insurance Pol-

icy
Now that a basic understanding of the profit test is demonstrated, we look at
equity-linked insurance to see how an insurer would analyze the cash flows of this
policy and set reserves if necessary to minimize risk. We use the information provided
from the Canada Life segregated fund information folder about the Core Conservative
Growth segregated fund to build the model of the policy. Suppose the policyholder
is 65 years old and they wish to pay the minimum premiums as an annuity. The
minimum amount set by Canada Life is an initial investment of $500.00, and any
future premiums must be at least $100.00. When the policyholder sets up an annuity
account, there is no initial fee charged, and the management expense ratio is 2.48%.
Note that the entire premium is allocated to the policyholder’s fund, thus there is no
unallocated premium amount added to the insurer’s fund (U APt = 0 for all t).

We begin the profit test by first predicting the policyholder fund, since the MER
depends on the amount in the fund, then we calculate the emerging surplus using
the MER, expenses and expected death benefit. The profit signature and net present
value are calculated in the same way as before. We set up a table with columns:
Time, Premium, Fund Before MC, MC, Fund After MC, Expenses, Probability of
Death, Expected Death Benefits, Reserves at Time t − 1, Expected Reserves Emerging
Surplus, Profit Signature, and Net Present Value. First we find the fund value before
subtracting the management charges, then we calculate the charges based on this
value, and the final fund value at the end of the year is what remains after subtracting
management charges.
F1 Before = P1 (1 + i)
= 500(1.0321)
= 516.05.

22
We then find the management charge in the first year,

M C1 = F1 Before × M ER
= 516.05(0.0248)
= 12.8

The final value is then

F1 = F1 Before − M C1
= 516.05 − 12.8
= 503.25

After finding the fund value after the first year, it must be accounted for in the future
fund values.
F5 Before = (F4 + P5 )(1 + i)
= (817.05 + 100)(1.0321)
= 946.49.

M C5 = F5 Before × M ER
= 946.49(0.0248)
= 23.47
F5 = F5 Before − M C5
= 946.49 − 23.47
= 923.02
Table (6) shows the result of the calculations of the policyholder’s fund.

Table 6: Policyholder Fund

Time Premium Ft Before M C MC Ft
0 $0.00 $0.00 $0.00 $0.00
1 $500.00 $516.05 $12.80 $503.25
2 $100.00 $622.62 $15.44 $607.18
3 $100.00 $729.88 $18.10 $711.77
4 $100.00 $837.83 $20.78 $817.05
5 $100.00 $946.49 $23.47 $923.02
6 $100.00 $1,055.86 $26.19 $1,029.67
7 $100.00 $1,165.94 $28.92 $1,137.02
8 $100.00 $1,276.73 $31.66 $1,245.07
9 $100.00 $1,388.24 $34.43 $1,353.81
10 $100.00 $1,500.48 $37.21 $1,463.27

23
The calculations of the fund value at the end of the year can also be done in one step
as follows.
F1 = P1 (1 + i)(1 − M ER)
= 500(1.0321)(1 − 0.0248)
= 503.25.

The fund values after time t = 1 are calculated similarly,

F5 = (F4 + P5 )(1 + i)(1 − M ER)

= (817.05 + 100)(1.0321)(1 − 0.0248)
= 923.02

The test continues by calculating the emerging surplus as in Equation (??). In this case
there is no negative emerging surplus in any year, so reserves do not need to be held.
The interest earned on insurer assets is also zero since there is no amount invested in
relation to this policy.

P r5 = M C5 − E5 − EDB5
= 23.47 − 1.7 + 12.42
= 9.35

Notice the only instance where the emerging surplus is negative is at time t = 0, since

Table 7: Insurer Cash Flows From Equity-Linked Policy

Time M Ct Et qx+t EDBt P rt
0 $0.00 $25.00 0 $0.00 -$25.00
1 $12.80 $1.70 0.00918 $4.62 $6.48
2 $15.44 $1.70 0.01009 $6.13 $7.61
3 $18.10 $1.70 0.0111 $7.90 $8.50
4 $20.78 $1.70 0.01222 $9.98 $9.09
5 $23.47 $1.70 0.01346 $12.42 $9.35
6 $26.19 $1.70 0.01484 $15.28 $9.20
7 $28.92 $1.70 0.01637 $18.61 $8.60
8 $31.66 $1.70 0.01808 $22.51 $7.45
9 $34.43 $1.70 0.01997 $27.04 $5.69
10 $37.21 $1.70 0.02208 $32.31 $3.20

this expense is paid immediately there is no need to hold reserves for this policy, the
policyholder fund itself acts as the reserves.

Next we calculate the profit signature and net present value as before in order

24
to value the policy.

Π5 = px+5 P r5
= (1 − qx + 5)P r5
= (1 − 0.01346)(9.35)
= 9.22

Table 8: Net Present Value Of Equity-Linked Policy

Time P rt px+t Πt NP V
0 -$25.00 1 -$25.00 -$25.00
1 $6.48 0.99082 $6.42 -$19.18
2 $7.61 0.98991 $7.54 -$12.99
3 $8.50 0.9889 $8.41 -$6.72
4 $9.09 0.98778 $8.98 -$0.65
5 $9.35 0.98654 $9.22 $5.00
6 $9.20 0.98516 $9.07 $10.03
7 $8.60 0.98363 $8.46 $14.29
8 $7.45 0.98192 $7.32 $17.63
9 $5.69 0.98003 $5.58 $19.94
10 $3.20 0.97792 $3.13 $21.11

Summary of Profit Measures

N P V = $21.11
IRR = 28%
DP P = 5
$21.11
Profit margin = = 1.99%
$500 + $100ν a66:9

5.3 Stochastic Testing of an Equity-Linked Insurance Policy

The deterministic test is good to get an idea of what to expect from the product, but
it is not realistic as interest rates tend to vary. With the stochastic test we follow
the same procedure as before while changing the expected interest rates earned on
the policyholder fund in each period. In this case we assume interest rates follow a
lognormal distribution with mean 0.0321 and standard deviation 0.0216. First we do
one simulation to calculate the N P V of the policy including an interest rate column in
the table for calculations. The change in Equation (4.22) is with interest rate notation,
where we use it instead of i to reflect the annual change in interest rates. We calculate

25
the policyholder fund as follows.
F1 = P1 (1 + i1 )(1 − M ER)
= 500(1.03597)(1 − 0.0248)
= 505.14.

F5 = (F4 + P5 )(1 + i5 )(1 − M ER)

= (819.8 + 100)(1.03815)(1 − 0.0248)
= 931.21.

Table 9: Stochastic Test of Policyholder Fund

Time Pt i1,t Ft Before M C Ft
0 $0.00 0 $0.00 $0.00 $0.00
1 $500.00 0.03597 $517.98 $12.85 $505.14
2 $100.00 0.05327 $637.37 $15.81 $621.57
3 $100.00 0.03221 $744.81 $18.47 $726.34
4 $100.00 0.01732 $840.65 $20.85 $819.80
5 $100.00 0.03815 $954.89 $23.68 $931.21
6 $100.00 0.01635 $1,048.06 $25.99 $1,022.07
7 $100.00 0.05934 $1,188.66 $29.48 $1,159.18
8 $100.00 0.06573 $1,341.95 $33.28 $1,308.67
9 $100.00 0.02903 $1,449.55 $35.95 $1,413.60
10 $100.00 0.01898 $1,542.32 $38.25 $1,504.08

Once the expected policyholder fund has been calculated, we then look at the cash flows
for the insurer. We take the management charges which have already been calculated
in the first step, then subtract expenses and the expected death benefit payable.
( 5
)
X
EDB5 = qx+5 max F5 , GM DB × Pt
t=1
= 0.01346 max{931.21, 500 + 4(100)}
= 0.1345(931.21)
= 12.53
It is important to remember the guaranteed minimum death benefit as this is the source
of risk for the insurer. In this case the minimum death benefit is 100% of the premiums
paid, GM DB = 1. Note that if we set GM DB = 1.5, the expected death benefit will
increase
EDB5 = 0.01346 max{931.21, 1.5(500 + 4(100))}
= 0.01345 max{931.21, 1350}
= 18.16.

26
We then calculate the emerging surplus in each year as before, except we include the
guaranteed minimum maturity benefit in the final year.

P r5 = M C5 − E5 − EDB5
= 23.68 − 1.7 − 12.53
= 9.45
1
X
P r10 = M C10 − E5 − EDB5 − 10 p65 max{F10 − GM M B 0Pt , 0}
t=1
10
! 1
Y X
= M C10 − E5 − EDB5 − (1 − q65+t ) max{F10 − GM M B 0Pt , 0}
t=1 t=1
= 38.25 − 1.7 − 33.21 − 0.8619(max{500 + 9(100) − 1504.08, 0})
= 3.34
We can build Table (10) with these calculations.

Table 10: Stochastic Test of Insurer Cash Flows

Time M Ct Et qx+t EDBt P rt
0 $0.00 $25.00 0 $0.00 -$25.00
1 $12.85 $1.70 0.00918 $4.64 $6.51
2 $15.81 $1.70 0.01009 $6.27 $7.83
3 $18.47 $1.70 0.0111 $8.06 $8.71
4 $20.85 $1.70 0.01222 $10.02 $9.13
5 $23.68 $1.70 0.01346 $12.53 $9.45
6 $25.99 $1.70 0.01484 $15.17 $9.12
7 $29.48 $1.70 0.01637 $18.98 $8.80
8 $33.28 $1.70 0.01808 $23.66 $7.92
9 $35.95 $1.70 0.01997 $28.23 $6.02
10 $38.25 $1.70 0.02208 $33.21 $3.34

Then we finally calculate the profit signature and net present value as before.

Π5 = p65+5 P r5
= 0.9865(9.45)
= 9.32
5
X
NP V = Πt ν t
t=0
1 1 5
= −25 + 6.45 + ... + 9.32( )
1.103 1.103
= 5.44

27
 Table 11: NPV of Equity-Linked Policy
Time P rt px+t Πt NP V
0 -$25.00 1 -$25.00 -$25.00
1 $6.51 0.99082 $6.45 -$19.15
2 $7.83 0.98991 $7.76 -$12.78
3 $8.71 0.9889 $8.61 -$6.36
4 $9.13 0.98778 $9.02 -$0.27
5 $9.45 0.98654 $9.32 $5.44
6 $9.12 0.98516 $8.99 $10.43
7 $8.80 0.98363 $8.66 $14.79
8 $7.92 0.98192 $7.78 $18.34
9 $6.02 0.98003 $5.90 $20.78
10 $3.34 0.97792 $3.27 $22.01

We can build the Table (11) using these equations.

Summary of Profit Measures

N P V = $22.01
IRR = 29%
DP P = 5
$22.01
Profit margin = = 2.07%
$500 + $100ν a66:9
Now that it is determined how to do one instance of the profit test, we want to
simulate this same process many times. In this example we will do 1000 simulations to
determine a distribution for N P V and the loss L. From these distributions we want
to find the maximum, minimum, average, standard deviation, and a 95% confidence
interval for the expected net present value, as well as use the loss to determine the
reserve using the CTE method. We set the level of the CTE reserve at 75%. The
reserve value will then be equal to the average loss in the losses that exceed the 75th
percentile. To see the code used to perform this simulation see Appendix B. We get
the following results from the simulation:

Statistics
Average = $12.38
Standard Deviation = $1.64
Maximum = $19.87
Minimum = $6.77
95% CI = [$12.28, $12.48]
CTE75 reserve = $0.00

28
From these statistics the insurer would be able to determine whether the product is
profitable enough.

An important result of simulation is the ease of changing factors in the policy.

Specifically, the insurer may want to change the guaranteed minimum benefits offered.
We see that with the 75/100 maturity/death minimums, there is no need to hold
reserves for this policy. Lets see how the policy looks with multiple different guaranteed
minimum benefits.

100/100

Average = $11.18
Standard Deviation = $4.76
Maximum = $18.18
Minimum = −$20.30
95% CI = [$10.89, $11.48]
CTE75 reserve = $0.00

In this case we offer 100% minimum maturity and death benefits to the policyholder.
We see the average drops slightly and a large increase in the standard deviation. Also
there are now negative values appearing for the net present value when the fund does
poorly. The reserves held are still zero as the risk of a large loss is still low for the insurer.

100/125

Average = −$7.13
Standard Deviation = $6.57
Maximum = $7.52
Minimum = −$41.78
95% CI = [−$7.54, −$6.73]
CTE75 reserve = $0.00

Here is the same policy with 100% guaranteed minimum maturity benefit and 125%
death benefit. This is clearly not worth offering as the insurer would be losing money
in most cases. The insurer might considering adjusting the premium so that a only
a portion of the premium is allocated to the policyholder fund, with the rest being
allocated to insurer assets. What is interesting here is that even though the policy is
not profitable for the insurer, there is no need for reserves as the loss amounts remain
negative. This is due to the initial expense being paid at the start of the contract,
after which all emerging surplus amounts are positive, while not being sufficient to
make up for the initial expense.

29
75/75

Average = $12.79
Standard Deviation = $1.30
Maximum = $19.53
Minimum = $10.23
95% CI = [$12.71, $12.87]
CTE75 reserve = $0.00

We look at the case of 75% guaranteed minimum benefits offered in both categories.
We notice that the numbers are very similar to the 75/100 policy, but the insurer does
a little bit better as there would be fewer cases where the fund falls below 75% of total
premiums paid.

110/100

Average = −$13.84
Standard Deviation = $18.57
Maximum = $20.02
Minimum = −$62.16
95% CI = [−$14.99, −$12.68]
CTE75 reserve = $63.81

This policy offers a 110% guaranteed minimum maturity benefit, with 100% death
benefit. This is clearly the most volatile of the policies we have looked at and it also
would not be worthwhile for the insurer to offer, as most cases are negative N P V .
Similar to the 100/125 a solution may be to allocate some of the premiums to the
insurer fund to make up for the loss.

6 Conclusion
In this paper we have seen how an insurer would evaluate its products using cash
flow analysis. Specifically the traditional 10-year term life insurance and 10-year
equity-linked insurance policies were examined. Cash flow analysis using software
offers added flexibility to the manual calculation method as it is very easy to change
inputs provided the program is set up properly. An insurer is able to easily implement
variable premiums, expenses, interest rates to test the product. We also looked at how
an insurer can use the profit test to set premiums and reserves. Using the conditional
tail expectation of a loss, the insurer can set the desired level of risk, and maximizes
profit at this risk level. Finally we examined some types of profit measures an insurer

30
can use based on the profit test such as the net present value and profit margin in the
case of the deterministic test, and statistics of the distribution of the net present value
in the stochastic test.

When the equity-linked insurance policy was tested, we find that adjustments in
the guaranteed minimum benefits can change the profitability for the insurer, yet even
with reasonable changes to these amounts, the CTE75 reserve is still zero, as the risk is
not so high with this policy to need reserves. When the guaranteed minimums offered
are very large amounts, the insurer would then have to hold reserves.

31
7 References
D.C.M. Dickson, M.R. Hardy, H.R. Waters, Actuarial Mathematics for Life Contin-
gent Risks, Cambridge University Press, 2013.

The Canada Life Assurance Company. 2019 Annual Report,

available at: https://www.canadalife.com/about-us/financial-information.html

The Canada Life Assurance Company. Canada Life Segregated Funds Information
Folder, available at: https://www.canadalife.com/investing-saving/segregated-funds/
audited-financial-statements.html

Statistics Canada. Life Tables, Canada, Provinces and Territories, 2016/2018

available at: https://www150.statcan.gc.ca/n1/pub/84-537-x/84-537-x2019002-eng.htm

32
A Life Table Excerpt - Population of Canada, Both
Sexes, Age 55 to 75

Age lx dx qx px ex
55 94814 352 0.00371 0.99629 30.3
56 94462 382 0.00405 0.99595 29.4
57 94080 416 0.00442 0.99558 28.6
58 93664 452 0.00483 0.99517 27.7
59 93212 492 0.00528 0.99472 26.8
60 92720 536 0.00578 0.99422 25.9
61 92184 584 0.00633 0.99367 25
62 91600 636 0.00694 0.99306 24.2
63 90964 692 0.00761 0.99239 23.3
64 90272 754 0.00836 0.99164 22.5
65 89517 822 0.00918 0.99082 21.7
66 88696 895 0.01009 0.98991 20.8
67 87801 975 0.0111 0.9889 19.2
68 86826 1061 0.01222 0.98778 18.4
69 85765 1155 0.01346 0.98654 17.7
70 84611 1256 0.01484 0.98516 16.9
71 83355 1365 0.01637 0.98363 16.1
72 81991 1482 0.01808 0.98192 15.4
73 80508 1608 0.01997 0.98003 14.7
74 78900 1742 0.02208 0.97792 13.3
75 77158 1885 0.02443 0.97557 12.6

lx - Number of people alive at age x per 100000 alive at age 0

dx - Number of deaths at age x per 100000 alive at age 0

qx - Probability of death during the year at age x

px - Probability of surviving the year at age x

ex - Expected future lifetime at age x

33
B Code used to Simulate the Stochastic Profit Test
1000 times
This function is for use in Microsoft Visual Basic for Applications, also known as
VBA in Excel. See appendix C for the inputs of this function. The function returns
the statistics shown in section (5.3).

[fontsize=\small]
Public Function EQUITYLINKED(i As Double, initial_premium _
+ As Double, annuity_premium As Double, age As Integer, _
+ expected_PH_fund_interest_rate As Double, _
+ PH_fund_interest_standard_deviation As Double, _
+ gmmb As Double, gmdb As Double, _
+ mer As Double, initial_expenses As Double, _
+ annual_expenses As Double, ci_level, _
+ cte_level As Double) As Variant

’Define variables as arrays since we may want to remember

’values in previous years
Dim NPV(1 To 1000) As Variant
Dim loss(1 To 1000) As Variant
Dim interest(1 To 10) As Double
Dim fund(1 To 10) As Double
Dim charge(1 To 10) As Double
Dim edb(1 To 10) As Double
Dim expenses(1 To 10) As Double
Dim profit(1 To 10) As Double
Dim sig(1 To 10) As Double
Dim kNPV(1 To 10) As Double
Dim px(1 To 10) As Double
Dim nu(1 To 10) As Double
Dim kloss(1 To 10) As Double
Dim important_statistics(1 To 7) As Variant
Dim large_loss As Variant

’Counting variables
Dim t As Integer
Dim k As Integer

’Variables for the Normal mean and standard deviation

34
Dim mu As Double
Dim sd As Double

’Calculate Normal mean and standard deviation

mu = Log((expected_PH_fund_interest_rate) ^ (2) / _
+ (Sqr((expected_PH_fund_interest_rate) ^ (2) + _
+ (PH_fund_interest_standard_deviation) ^ (2))))
sd = Sqr(Log(1 + (PH_fund_interest_standard_deviation) ^ (2) / _
+ (expected_PH_fund_interest_rate) ^ (2)))

’Loop to calculate discount rates and set survival probabilities

’based on the Excel Sheet "Life Table"
’This sheet is provided from Statistics Canada
For t = 1 To 10
nu(t) = (1 / (1 + i)) ^ t
px(t) = Worksheets("Life Table").Cells(age + 5 + t, 6).Value
Next t

’Initialize profit test variables

For k = 1 To 1000
interest(1) = Exp(mu + WorksheetFunction.Norm_Inv_
+(Rnd, 0, 1) * sd)
charge(1) = initial_premium * (1 + interest(1)) * mer
fund(1) = initial_premium * (1 + interest(1)) - charge(1)
edb(1) = WorksheetFunction.Max((1 - px(1)) * fund(1), _
+ (1 - px(1)) * initial_premium * gmdb)
expenses(1) = annual_expenses * annuity_premium
profit(1) = charge(1) - expenses(1) - edb(1)
sig(1) = profit(1) * px(1)
kloss(1) = -profit(1)
kNPV(1) = sig(1) * nu(1) - initial_expenses

’loop through profit test

For t = 2 To 10
interest(t) = Exp(mu + WorksheetFunction._
+Norm_Inv(Rnd, 0, 1) + * sd)
charge(t) = (fund(t - 1) + annuity_premium) _
+ * (1 + interest(t)) * mer
fund(t) = (fund(t - 1) + annuity_premium) * _
+ (1 + interest(t)) - charge(t)

35
edb(t) = WorksheetFunction.Max((1 - px(t)) * _
+ fund(t), _
+ (1 - px(t)) * (initial_premium + annuity_premium * _
+ (t - 1)) * gmdb)
expenses(t) = annual_expenses * annuity_premium

’want to check the gmmb if it is time 10

If t = 10 Then
profit(t) = charge(t) - expenses(t) - edb(t) - TPX(age, 10) * _
+ WorksheetFunction.Max(0, (initial_premium + 9 * annuity_premium) _
+ *gmmb - fund(10))

Else
profit(t) = charge(t) - expenses(t) - edb(t)
End If

sig(t) = profit(t) * px(t)

kloss(t) = kloss(t - 1) - profit(t) * TPX(age, t - 1)
kNPV(t) = kNPV(t - 1) + sig(t) * nu(t)
Next t

’set the loss and npv arrays as the final values

loss(k) = kloss(10)
NPV(k) = kNPV(10)
Next k

’define the output values

important_statistics(1) = WorksheetFunction.Average(NPV)
important_statistics(2) = WorksheetFunction.StDev_S(NPV)
important_statistics(3) = WorksheetFunction.Max(NPV)
important_statistics(4) = WorksheetFunction.Min(NPV)
important_statistics(5) = important_statistics(1) - _
+ WorksheetFunction.Confidence_Norm(0.05, _
+ important_statistics(2), 1000)
important_statistics(6) = important_statistics(1) + _
+WorksheetFunction.Confidence_Norm(0.05, _
+ important_statistics(2), 1000)
important_statistics(7) = CTE(cte_level, loss)

EQUITYLINKED = important_statistics
End Function

36
C Inputs for EQUITYLINKED Function

Name Notation
Discount Rate i
Initial Premium P1
Annuity Premium Pt
Age of Insured N/A
PH Fund E[Interest] µ
PH Fund SD[Interest] σ
GMMB gmmb
GMDB gmdb
Management Charge mct
Initial Expense E0
Annual Expenses Et
CI alpha αCI
CTE alpha αCT E

37
D CTE Function
This function is used in the EQUITYLINKED function to calculate the CTE reserve.

Public Function CTE(level As Double, losses As Variant)

Dim losses_above_quantile() As Variant

Dim quantile_loss As Double
Dim k As Integer
Dim count As Integer
Dim avg As Double

’Find the minimum loss of the new distribution

quantile_loss = WorksheetFunction.Percentile(losses, level)

’Build an array of losses that exceed the set percentile

For k = 1 To UBound(losses)
If losses(k) >= quantile_loss Then

count = count + 1
ReDim Preserve losses_above_quantile(1 To count)
losses_above_quantile(count) = losses(k)

End If
Next k

’Take the average of the large losses

avg = WorksheetFunction.Average(losses_above_quantile)

’Set the reserve to 0 if the loss is negative

If avg > 0 Then
CTE = avg
Else
CTE = 0
End If
End Function

38
E t px function
The TPX function is used in the EQUITYLINKED function to determine t px when
needed.

Public Function TPX(age As Integer, time As Integer)

’Define variables and set their length equal to time input

Dim kpx() As Double
Dim px() As Double
Dim t As Integer
ReDim kpx(time)
ReDim px(time)

’take the survival probability from the life table

kpx(1) = Worksheets("Life Table").Cells(age + 5, 6).Value

If time >= 2 Then

For t = 2 To time
px(t) = Worksheets("Life Table").Cells(age + 5 + t - 1, 6).Value
kpx(t) = px(t) * kpx(t - 1)
Next t
End If

’Return the product of the single year survival probabilities

TPX = kpx(time)

End Function

39