Actuarial Mathematics Ii: About

Actuarial Mathematics II
Course Notes
2007 David Kirk
ACTUARIAL MATHEMATICS II
About
This course has been created for the Lebanese University's Actuarial
Mathematics II course as part of their Masters in Actuarial Science
postgraduate programme. It is a concentrated course touching on many
of the basic actuarial techniques applied in life- and non-life insurance
mathematics.
Table of Contents
Chapter 1 Fundamentals of actuarial mathematics................................................... ......2
Chapter 2 Introduction to reserves and liabilities....................................................... ...21
Chapter 3 Actuarial reserving techniques for life insurance................................ .........38
Chapter 4 Unit and non-unit reserves.................................................... .......................54
Chapter 5 Actuarial reserving techniques for non-life insurance.................................61
Chapter 6 Assets and Liability considerations................................. .............................79
Chapter 7 Pension fund reserving................................................................................ ..96
Course Notes
2007 David Kirk
Chapter 1 Fundamentals of actuarial mathematics

About:
This chapter summarises the fundamentals of actuarial mathematics and
introduces standard actuarial notation that will be used throughout the
course. This material is not directly examinable, but knowledge of this
chapter is vital to understanding the course and being able to answer
exam questions on other sections.
Chapter Contents
1 Basic financial mathematics......................................................................................... ..3
1.1 Compound interest................................................................................... ...............3
1.2 Actuarial notation for financial mathematics................................ .......................6
2 Introduction to concepts of life contingencies....................................................... .....10
2.1 Survival probabilities............................................................................... ..............10
2.2 Death probabilities.............................................................................................. ..10
2.3 Deterministic modelling.................................................................... ...................12
2.4 Omega age................................................................................... ..........................13
3 Introduction to actuarial mathematics......................................... ..............................14
3.1 Life-contingent annuity factors................................................................. ............14
3.2 Premium payable for an annuity.................................................................. .........15
3.3 Assurance factors................................................................................... ................17
3.4 Guaranteed Endowments................................................................ .....................19
3.5 Premium calculations for assurance benefits......................................................19
3.6 Fractional year factors................................................................................. ..........19
3.7 Joint-life factors....................................................................... .............................20
Chapter 1 : Fundamentals of actuarial mathematics

Section 1 : Basic financial mathematics
1 BASIC FINANCIAL MATHEMATICS

1.1 Compound interest
1.1.1 Simplifying assumptions regarding interest rates
For most of the work in this course we will assume interest rates are constant. This
means two things:
1. The term structure of interest rates is flat. The yield curve is flat.
2. Interest rates do not change over time. Volatility of interest rates is zero.
Note that neither of these assumptions is realistic.
1.1.2 Definitions of interest rate notation

Let i be an interest rate.
i will apply for a certain period (a year, half a year, a quarter or a month typically)
i will be payable (and this compounded) at some interval (a year, a month etc.)
Table 1.1: Types of Interest rates

Type of interest rate
Applies for Compounds
Conversion to
Annual Effective
NACA = Annual Effective

Nominal Annual Compounded
Annually
year
yearly
NACS = i 2
Semi-annually
Year
Twice a year
i
1 1
2
NACQ = i 4
Quarterly
Year
Quarterly
i
1 1
4
2 2
4 4
David Kirk
Page 3 / 107

Type of interest rate
Applies for Compounds
NACM = i 12
Monthly
Year
Monthly
NMCM
Nominal Monthly Compounded
Monthly
Month
Monthly
NQCM
Nominal Quarterly Compounded
Monthly
Quarter
Monthly
Year
Continuously
NACC =
Continuously
Conversion to
Annual Effective
12 12
i
1
12
[ 1i 121]
12
i
1 1
4
e 1
This last row is called the force of interest and is described in the following
sections.
1.1.3 Continuously compounded rates of interest

(delta) is the symbol commonly used for the continuously compounded rate of
interest. It is also known as the force of interest. It is not generally used to quote an
interest rate in the market. It is used when performing actuarial calculations when
the cashflows are paid continuously over time.
Note that you may also have seen (or will see) the force of interest when it comes to
financial economics and the Black-Scholes option pricing formula.
1.1.4 Derivation of
t=lim n
nt

i n
1
n
t=[ ln 1i ]
t=tln 1i
i=e 1
David Kirk
Page 4 / 107

t is the time-dependent force of interest applicable at duration t.

Our usual simplification is that:
t= t
1.1.5 Application of force of interest in actuarial mathematics

Example
Premiums are paid monthly in advance. To calculate the interest applicable to
premiums we could use a NMCM interest rate such that the present value of a
premium in 1 month's time is:
PV Premium1= Premium11i 1
Or, we could use an annual effective (NACA) rate to perform the same calculation:
PV Premium1= Premium11i
1
12
These make sense for premiums since premiums are paid as specified intervals
according to the policy terms. Still, we could write the same calculation using a
continuously compounded (NACC in this case) force of interest like this:
PV Premium1= Premium1e
12
However, when are claims paid? Typically, we assume they are paid at the end of the
period we are considering. Most of our examples will be annual examples (for
simplicity) but we will also work with monthly examples. But is it correct to assume
that policyholders will wait until the end of the month for their death benefit?
A more accurate approach would be to assume that claims are paid continuously
over time. The PV of benefits paid during a year could be written as follows:
1
PV Benefits =SA e e
t
x t
x t dt
t=0
Where x t is the instantaneous annualised probability of death at age x+t. We

won't deal much with continuous-time assurance functions. This is given as an
example to better understand why we may need a force of interest or continuously
compounded rate of interest.
David Kirk
Page 5 / 107

1.1.6 Interest, discounts and discount factors

If i is the annual effective rate of interest, then the following also exist:
Discount factor:
v t =1i t
The discount factor is used to take present values of cashflows in a more concise
way.
Discount:
d =1v
The discount is the rate of interest when the interest is paid in advance. For
example, if interest rates are 10%, borrowing USD100 for one year will require
repayment of the USD100 and USD10 of interest for a total payment of
USD110=USD1001i
However, some loans are structured so that the interest payable is deducted off the
amount borrowed. If the loan is the discounted value of USD100, then the borrower
receives
USD1001 d =USD90.91
and the total interest paid is
USD100 USD1001d =USD100[ 1 1d ]=USD100d

As an exercise, prove that d =
d =1v =1
i
1i
1
1i1
i
=
=
1i 1i 1i
1.2 Actuarial notation for financial mathematics

This section describes some of the basic actuarial notation used for non-contingent
financial maths. That is, financial maths that doesn't not depend on contingent
event such as death.
1.2.1 Annuity certain payable in arrears

Consider a regular stream of constant payments of amount 1, payable annual in
arrears for n years. With a constant annual effective interest rate of i, the formula for
this can be written as:
David Kirk
Page 6 / 107

n
PV = v t
t=1
This is a geometric sum with geometric increases of
1
or 1i1 for n
1i
periods. Thus:
1 1in
PV =
i
If you need to, show the derivation of the above as an exercise.
The actuarial notation for this formula is:
11i n
an =
i
Check that you can calculate the following:
a 10 = 6.14 for i = 10%
a 10 = 7.72 for i = 5%
a 1 = 0.95 for i = 5%
1.2.2 Annuity certain payable in advance

Consider a regular stream of constant payments of amount 1, payable annual in
arrears for N years. With a constant annual effective interest rate of i, the formula for
this can be written as:
n
PV = v t1
t=1
1 1in
PV =
d
As an exercise, prove that the previous two statements are equivalent
The actuarial notation for this formula is:
David Kirk
Page 7 / 107

11i n
a n =
d
Check that you can calculate the following:
a 10 = 6.76 for i = 10%
a 10 = 8.11 for i = 5%
a 1 = 1.00 for i = 5%
Now show that
a n =1ia n =1a n1
Explain in words what this relationship means.
1.2.3 Accumulation factors

For each symbol a there is an equivalent symbol s. a is the present value function for a
constant series of cashflows. s is the future value function for a constant series of cashflows.
Payments in arrears:
n
FV = Cvt1
t=1
1i n 1
sn =
i
Payments in advance:
n
FV = Cvt
t=1
1i n1
s n =
d
Prove the following:
David Kirk
Page 8 / 107

v ns n =a n
v ns n = a n
David Kirk
Page 9 / 107

Section 2 : Introduction to concepts of life contingencies
2 INTRODUCTION TO CONCEPTS OF LIFE CONTINGENCIES

2.1 Survival probabilities
The probability of survival for a life aged x for t years is shown as
px .
If the t subscript is omitted, it is assumed to be 1. Thus,

p x =1 p x
Mortality in each year is independent. Thus, the following equation holds:

p x =1 p x 1 p x1
and further
p x =1 p x 1 p x1 1 p xn1
In terms of standard life table notation,
p x=
l xt
where l x is the number of lives
lx
in the population aged x .
2.2 Death probabilities

The probability of death for a life aged x within t years is shown as
qx .
q x =1 p x
q =1t p x
t x
Prove that
2
q x q x q x1
David Kirk
Page 10 / 107

Proof:
q x =12 p x
2q x =1 p x p x1
2q x =11q x 1q x1
2q x =1[ 1q x q x1q xq x1 ]
2q x =q x q x1q xq x1
2
Intuitively, the second last line can be interpreted as The probability of dying
within two years is the sum of the probability of dying in the first year and the
probability of dying in the second year, less the probability of dying in both
years. Take a moment to ensure that you understand why this (mathematically) true
If q x =0.01 and q x1=0.02 what is
qx ?
q x =11 p x 1 p x1 =10.990.98=0.02980
2.2.1 Force of mortality

In the same way that for interest we have a force of interest that reflects the
continuous compounding of interest over time, we have a force of mortality which
can be interpreted as the annualised instantaneous probability of death at a
particular point.
x t
mdm
p x =e
m= x
Where t is the force of mortality at time t .

x1
If t= x t [ x , x1 then
Thus
m dm
1 p x =e
m= x
simplifies to become
p x =e
q =1e and so x =ln 11q x which is also x =ln 1 p x

x
1 x
So we have assumed a constant force of mortality within an age bracket in order to

calculate the force of mortality operating over that age bracket. This is a common
method of calculating monthly mortality rates from annual life tables.
1
12
x /12
q x=1e
Prove the above equation.
David Kirk
Page 11 / 107

If q 45 = 0.01, calculate
Then, calculate
1
12
q 45
12
q 45 assuming a constant force of mortality for lives aged 45. Show
you result to at least 5 decimal places.
q 45
?
12
Is this the same as
If they are different, explain why this is.
We have used the assumption of a constant force of mortality in deriving these

results. Another possible assumption is the uniform distribution of deaths or
UDD.
Under UDD:
1
12
q x=
qx
12
or more generally:
t
q x =t q x for any integer x and 0 < t < 1
2.3 Deterministic modelling

In this chapter, we consider deterministic models. A deterministic model is one
where we consider only a single path of a random variable or time series. We do not
consider the potential variation around this sample path.
If we are going to use a single path, the obvious choice is the Expected Outcome.
For all our calculations, we will be calculating expected future cashflows, or equating
expected present values.
The expected value of a cashflow at time t for a poisson distribution is:
E [ CF t ]= pmf CF xx=
x=0
x=0
e
x
x!
Fortunately, our modelling is a lot simpler. We are considering a single life which is
either alive or dead in any particular period. Thus, the random variable follows a
Bernoulli distribution. If the cashflow in question is a premium, then a premium on 1
is payable if the life is alive, and 0 is paid if the life is dead.
David Kirk
Page 12 / 107

1
E [ CF t ]= pmf CF xx=1t p x 01t p x =t p x

t
x=0
The Expected Present Value of that same premium payable as described above is:
EPV [ CF t ]=v
pmf CF xx=v t [ 1t p x01 t px ] =v nt p x

x=0
2.4 Omega age

Life tables have a maximum age called the Omega Age. The probability of death at
this age is 1. Thus, nobody can live to be older than age .
We will use this concept in later sections.
David Kirk
Page 13 / 107

Section 3 : Introduction to actuarial mathematics
3 INTRODUCTION TO ACTUARIAL MATHEMATICS

3.1 Life-contingent annuity factors
3.1.1 Immediate Annuity
An immediate annuity is an annuity that begins paying 1 immediately (not deferred
until some point in the future) but is paid annual in arrears for the lifetime of the
life. Thus, it is not really paid immediately but rather after 1 year. The first cashflow
takes place at the end of the period.
An immediate annuity factor is the Expected Present Value of a stream of payments
of 1 as long as the life aged x is still alive.
1
EPV=v 1 p x v 2 p x v 3 p x
In practice, because our life table has a cut-off maximum age, we might express this
as:
EPV=v 11 p x v 22 p x v 33 p x v x x p x
The standard actuarial notation for this is:
a x =v 11 p x v 22 p x v 33 p x v x x p x
x
a x = v ii p x
i=1
3.1.2 Immediate Annuity in advance

As the name suggests, this is the expected present value of an annual payment of 1 for
every year that the life currently aged x is alive, with payments occurring at the start
of the year.
0
a x =v 0 p x v 1 p x v 2 p x v
x p x
a x = v ii p x
i=0
Show that a x =a x 1
David Kirk
Page 14 / 107

3.1.3 Temporary annuities

A temporary annuity is a contingent annuity that is payable annually until the sooner
of death, or a specified term.
In arrears:
a x: n =v 11 p x v 22 p x v 33 p x v nn p x
a x : n = v ii p x
i=1
In advance:
a x : n =v 00 p x v 11 p x v 22 p x v n1n 1 p x
n1
a x : n = v ii p x
i=0
Show that a x : n =1a x: n1
Show that a x: n =a x v nn p xa xn
3.2 Premium payable for an annuity

A policyholder usually purchases an immediate annuity by paying a single
premium. The insurer invests this money to earn interest (i) and pays regular
payments every year to the policyholder as long as he or she is alive. The
policyholder is insured against the risk of living too long. The insurer has taken
on the risk that the policyholder lives longer than expected.
The premium charged by the insurer for the annuity depends on the interest rates
available in the market, and the level of mortality expected.
Explain whether (and why) the insurer will have to charge a higher or lower
premium if:
interest rates rise
David Kirk
Page 15 / 107

mortality rates decrease
If interest rates rise, the discount rate will increase. If the discount rate increases,
the present value of future cashflows will decrease. Thus, a lower premium can be
charged. Another way of looking at this is to consider that with higher interest rates,
the premium received will earn higher investment return (interest) and will thus
grow faster to be able to pay higher benefits. For the same benefit, a lower premium
can be charged.
If mortality rates decrease, policyholders are expected to survive for longer. Annuity
payments are made as long as the policyholder survives. Thus, lower mortality
implies more payments. The present value of these higher payments will be higher
and the premium charged will have to increase.
To calculate the premium for any type of policy in the traditional actuarial way, we
equate the expected present value of income to the expected present value of outgo.
In this case we are only considering premiums and benefits.
Premiums are paid by the policyholder to the insurer. Benefits are paid by the
insurer to the policyholder. Unless told otherwise, the premiums will be constant
and denoted as P
The first step should always be:
EPV Premiums=EPV Benefits
Now, we fill in the LHS and RHS separately with more detailed formulae.
LHS =EPV Premiums=P
Since there is only one premium paid, and it is paid at the start of the contract, the
EPV is simply P.
Now if the annual benefit payment (in arrears) is B,
RHS =EPV Benefits =Ba x
Make sure you understand the above equation. Why is
value of benefits?
Ba x the expected present
Next step is to complete the original equation:
David Kirk
Page 16 / 107

EPV Premiums=EPV Benefits
P=Ba x
Therefore the premium that the insurer should charge is simply
Ba x .
What is the premium payable for a temporary annuity paid annually in advance for
up to 25 years? Define all symbols used. Will this be smaller or greater than the
premium for an immediate annuity paid in advance?
3.3 Assurance factors

3.3.1 Assurance factor
An assurance factor is the expected present value of a sum assured of 1 for a life aged
x paid at the end of the year.
An assurance factor is the Expected Present Value of a stream of payments of 1 in the
year of death of the policyholder.
1
EPV=v 0 p xq x v 1 p xq x1v 2 p xq x2
In practice, because our life table has a cut-off maximum age, we might express this
as:
1
EPV=v 0 p xq x v 1 p xq x1v 2 p xq x2v
x1
x p xq
The standard actuarial notation for this is:

1
A x =v 0 p xq x v 1 p xq x1v 2 p xq x2v
x1
x p xq
Ax = v i1i p xq xi
i=0
3.3.2 Term Assurance factor

n1
= v i p xq xi = v ii1 p xq xi1
x :n
i1
i= 0
i=1
Explain the above equation in words. What does the 1 over the x signify?
Confirm that you understand why both definitions (from i = o and from i = 1) are correct.
David Kirk
Page 17 / 107

3.3.3 Pure Endowment factor

A pure endowment is a benefit that pays only only survival to time n. The EPV of the benefit
is given by the following equation:
=v nn p x
x :n
Who would want to buy such a policy?

Why might it be unpopular?
A policyholder with a specified financial need in future might purchase the policy
as a savings vehicle, provided the financial need exists only for that policyholder
and not for any dependants. An example might be
Saving for retirement (with no family). If the policyholder dies, no savings are
needed.
Saving for university education. The education will not be needed if the
policyholder is dead.
It is often unpopular because no benefit is paid on death. On death occurring

shortly before maturity, a large number of premiums will have been paid but no
benefit is received. Problems arise through poor selling processes where the policy
may not be fully explained. Further, the family of the deceased policyholder may be
unhappy that there is no benefit to be paid to them. As a result, these policies are
usually not sold by insurers any more.
3.3.4 Endowment Assurance factor

An Endowment Assurance (often called just an endowment - for this course use the full
name to prevent confusion) is a policy that pays a benefit at the end of year of death of the
life assured, or at the end of year of maturity given survival to that period.
The EPV of benefits is given as:
Ax : n = A
x: n
x :n
since it combines the benefit of the Term Assurance with the
benefit of a pure endowment. One can also write:

n
A x : n = v ii1 p xq xi1v nn p x
i=1
Show that this can further be written as:
David Kirk
Page 18 / 107

n1
A x : n = v ii1 p xq xi1v nn1 p x
i=1
An n-year endowment assurance pays a benefit at the end of year of death for n-1 years. On
survival to n-1 years, a benefit is paid on death or survival at the end of year n.
3.4 Guaranteed Endowments

A guaranteed endowment is not really an insurance policy. It pays a benefit at maturity on
death or survival to maturity. The guaranteed endowment factor is simply v n
Show that
is equal to
n1
v n px i p xq xiv n
i=0
A guaranteed endowment pays a benefit on death before time n at the end of year n, and
also on survival to time n.
3.5 Premium calculations for assurance benefits

Given an expression for the regular annual premium (paid in advance) for an endowment
assurance policy.
Remember, start with EPV Premiums=EPV Benefits .
Give similar expressions for:
regular annual premium (paid in advance) for a term assurance policy
regular annual premium (paid in advance) for a pure endowment
regular annual premium (paid in advance) for an endowment assurance policy
regular annual premium (paid in advance) for a term assurance policy
regular annual premium (paid in advance) for a pure endowment
How does the single premium for a pure endowment policy compare with the amount
you would need to invest in a bank account earning the same amount of interest, over the
same period? Why is this?
3.6 Fractional year factors

All the above factors can be considered with payments more frequent than once per
year. This will not be covered in this course to make room for other concepts.
David Kirk
Page 19 / 107

3.7 Joint-life factors

Many of the above factors also exist in joint-life versions. These depend on the
survival or death of one or both of two lives (typically a married couple). These are
not frequently used in practice and have also been excluded from this course to allow
other concepts to be covered in more detail.
David Kirk
Page 20 / 107
Course Notes
2007 David Kirk
Chapter 2 Introduction to reserves and liabilities

About:
This chapter introduces the concepts of actuarial reserving and explores
the differences between reserving principles for solvency, and reserving /
liability principles for financial reporting. The Actuarial Control Cycle will
be explored as a framework for solving actuarial problems (including
reserve calculations).
Chapter Contents
1 The Actuarial Control Cycle.................................................................. .......................23
1.1 Overview......................................................................................................... ........23
1.2 Identifying the problem (or defining the problem).................................. ........24
1.3 Developing the solution.............................................................................. ..........24
1.4 Monitoring the experience................................................................................. ...25
1.5 Environment.......................................................................................... ................25
1.6 Professionalism.................................................................................. ...................26
2 Risk management in context of Actuarial Control Cycle...........................................27
2.1 Steps in Risk Management.......................................................................... ..........27
2.2 Risk Identification...................................................................... ..........................27
2.3 Risk Measurement........................................................................ ........................27
2.4 Risk Management.......................................................................................... .......28
2.5 Risk Monitoring........................................................................................ ............28
3 Cashflows arising from insurance products................................. ..............................29
3.1 Definition of cashflow.......................................................................................... .29
3.2 Operating cashflows...................................................................................... .......29
3.3 Cashflows of a capital nature.......................................................... .....................29
3.4 Mismatch in timing of cashflows in insurance products...................................30
4 Actuarial reserves and liabilities.................................................. ...............................32
4.1 Actuarial reserves for solvency.............................................................................. 32
Chapter 2 : Introduction to reserves and liabilities

Section Chapter 2 : Introduction to reserves and liabilities
4.2 Actuarial liabilities for financial reporting............................................. .............32
4.3 Difference in purpose between actuarial reserves and actuarial liabilities........33
4.4 Actuarial Balance Sheet......................................................... ..............................34
4.5 Typical actuarial reserves................................................................................... ...35
4.6 Special reserves.................................................................................................... .36
David Kirk
Page 22 / 107

Section 1 : The Actuarial Control Cycle
1 THE ACTUARIAL CONTROL CYCLE

1.1 Overview
The Actuarial Control Cycle (ACC) is a problem solving tool used in actuarial work.
There are control cycles in many professions and industries that follow a similar
form. The key idea is that it is a cycle that controls how to ensure a systematic
approach to ongoing problem solving.
It is a useful tool to solve actuarial problems in practice. It is also an extremely useful
tool for developing answers for actuarial exams. It encourages thinking around the
problem and following the problem through to its logical, long-term result.
The Environment
Monitoring
the
experience
Identifying
the problem
Developing
the Solution
Professionalism
Illustration 2.1: The Actuarial Control Cycle
David Kirk
Page 23 / 107

1.2 Identifying the problem (or defining the problem)

The first step is to identify the problem. Sometimes this will be apparent from the
real-life (or exam) problem, but sometimes significant work is required to
understand what the problem is. This is an active step (rather than a passive step)
that requires the actuary to consider the drivers of the problems, risks inherent in the
problem, different stakeholders who may have an interest in the problem.
Tip think about why the problem isn't trivial to solve. What makes it difficult, why
hasn't this problem already been solved? What solutions have already been tried?
Who is happy with the current solution, and who is unhappy?
The problem may be:
Our Term Assurance policies are not profitable
What will future mortality rates be for our policyholders?
Our business is exposed to several risk
This last problem will probably need to be devolved further by identifying each of the
risks to identify specific problems. This identification process is part of the Identify
the Problem step.
This step is also called defining the problem since to accurately identify the
problem, it is necessary to properly define the key aspects of the problem.
1.3 Developing the solution

Once the problem has been properly identified and defined we need to work to solve
it. This step is called Developing the Solution. Given the information specified in
the previous step, several solutions will be proposed, with flawed solutions being
discarded. The problem characteristics identified in the previous step should be kept
in mind as an effective solution will have to address these elements.
Solutions may involve:
We will increase term assurance premium rates
We will estimate future mortality based on past mortality experience, allowing for
apparent trends. We will make use of reinsurers' data to supplement our data
where it is not credible.
We will make use of reinsurance and pooling to reduce risks to our business.
David Kirk
Page 24 / 107

1.4 Monitoring the experience

This step is what make the ACC a control cycle. Once we have identified the
problem and developed a solution, it is critically important to monitor the
experience. We monitor the results to ensure that any assumptions we made in
developing the solution are reasonable. We look for unintended consequences and
gather new information for the next step. The next step is to Identify the Problem
or redefine the problem based on updated information from our experience, and
given the solutions we have already implemented.
We must monitor sales of term assurances. If our increased premium rates lead to
significant lower volumes of business, our unit expenses may increase resulting in
lower profitability.
Future mortality may differ from past mortality. Our past experience may not have
been a reliable estimate of future mortality. Mortality improvements (through
improved medical care) or worse mortality (HIV/AIDS or pandemics such as bird
flu) result in current and future mortality being very different from past mortality.
Our reinsurance programme may not be sufficient and our business may still be
exposed to too much risk. Alternatively, our reinsurance programme may be too
extensive, with the result that our profits are too low.
1.5 Environment
This part of the ACC is not a step in the process. Rather, it is a factor that must be
considered at each stage of the cycle. The specific elements considered will vary
depending on the problem at hand, but this list gives an idea of the typical categories
to consider:
Economic environment (interest rates, inflation, economic growth, currency

stability, GDP per capita, income inequality)
Regularly environment (licensing requirements, asset admissibility rules,

policyholder protection, reporting requirements, capital requirements, taxation
rules)
Business environment (number, size and strength of competitors, distribution

channels, employee considerations)
Political environment (political stability, potential for political interference in

business, impact on the economic environment)
Social environment (policyholder activism, treating customers fairly, socially

responsible investing, limits on underwriting, limits on exclusions)
David Kirk
Page 25 / 107

1.6 Professionalism
All aspects of actuarial work must be performed with utmost professionalism and
ethics. Much of our work relates to individuals long-term savings and other critical
events in their life such as the loss of a bread-winner for a family. Trust in placed in
insurers companies to look after individuals in their time of need and actuaries are
usually the custodians of this trust.
The following list represents just a few of the typical considerations under
professionalism:
Actuaries should consider the impact of their actions on all stakeholders.
It may often be important to convey this understanding to non-actuaries in a clear

manner.
Actuaries should not perform work for which they are not qualified or in which
they do not have sufficient experience.
Actuaries should avoid conflicts of interest, and if it is necessary to work under

conflicts of interest, then this should be communicated to all relevant parties.
Actuaries must maintain the confidentiality of their clients.
You will receive a copy of the UK's Professional Conduct Standards which governs the
professional conduct of members of the Institute and Faculty of Actuaries in the UK.
It also applies to Fellows of the Actuarial Society of South Africa. The content is not
examinable for this course, but you will be expected to consider aspects of
professionalism where relevant in exam questions.
David Kirk
Page 26 / 107

Section 2 : Risk management in context of Actuarial Control Cycle
2 RISK MANAGEMENT IN CONTEXT OF ACTUARIAL CONTROL CYCLE

2.1 Steps in Risk Management
Risk Identification
Risk Measurement
Risk Management
Risk Monitoring
Each of these steps is outlined in the sections below.
2.2 Risk Identification

Risk Identification is commonly performed through maintaining a risk register. The
risk register is a structured record of all the important risks faced by the organisation.
It can be difficult to list all relevant risks in a useful way, as many risks will interact
with one another and these links must be made clear in the risk register.
2.3 Risk Measurement

The risks should then be measured. There are three primary methods used to
measure risks.
2.3.1 Detailed modelled based on actual data

If reliable, credible data is available from the company's own experience, or through
industry data, then the risks can be modelled in detail through fitting probability
distributions to the magnitude and frequency of losses, allowing for correlations and
dependencies between risks.
This is typically the case for insurance risks, persistency risks and market risks. Credit
risks is increasingly being modelled accurately as banks collected relevant default
data for Basel II compliance.
2.3.2 Approximate modelling based on subjective estimates

Some risks are difficult to model exactly. Credit risk is still difficult to model
accurately for some classes of business. Operational risk is another that is typically
measured in an approximate manner. This can be based on subjective estimation of
distribution functions, or through an arbitrary factor.
David Kirk
Page 27 / 107

Section 2 : Risk management in context of Actuarial Control Cycle
2.3.3 Qualitative assessment

Some risks (often operational or catastrophe risks) cannot be measured at all with
current information and tools. For these, a subjective assessment of the probability
(or frequency or likelihood) and severity (or impact) is made.
For example:
an earthquake might be very infrequent, but have a severe impact
Fraud by an employee might be fairly frequent, but low in impact.
2.4 Risk Management

This section is called Risk Management which is also the name of the overall
process! However, it is in this section that the risks already identified and measured
are acted upon. Some of the possible actions include:
Avoid the risk
Reduce the risk
Mitigate the risk
Transfer the risk
Pool the risk
Share the risk
Hedge the risk
Insure the risk
Accept the risk
2.5 Risk Monitoring

As with the Actuarial Control Cycle, it is critically important that risks are monitored
on a regular basis. Risks may change in nature over time, and unless the risk registers
are regularly updated and the measurements re-performed, the risk management
actions may no longer be appropriate.
Additionally, if the risks are regularly monitored, it serves to focus management's
attention on risk management as a critical component of their job, rather than as
something peripheral to their main function.
David Kirk
Page 28 / 107

Section 3 : Cashflows arising from insurance products
3 CASHFLOWS ARISING FROM INSURANCE PRODUCTS

3.1 Definition of cashflow
Cashflow is the transfer of cash from one entity to another. It only counts as a
cashflow when it has moved or flowed. Thus, cash at a bank is not a cashflow.
Premiums flow from the policyholder to the insurance company and are thus a
cashflow.
Profit is not a cashflow since no money flows anywhere. It is an accounting result.
3.2 Operating cashflows

Operating cashflows are those cashflows that occur during the normal operating
activities of the company
3.2.1 Cash in-flows
Premiums
Investment income (interest and dividends, proceeds from sale of financial

instruments)
Reinsurance claims
Reinsurance commission
3.2.2 Cash out-flows
Claims
Expenses
Commission
Reinsurance premiums
Profit commission
Tax
3.3 Cashflows of a capital nature

Some cashflows arise due to the company increasing or decreasing the amount of
capital it has. These are not operational cashflows but are capital cashflows.
David Kirk
Page 29 / 107

3.3.1 Capital Cash in-flows
Issue Shares
Issue Preferred Shares (or preference shares)
Issue bonds / debentures (borrow money from investors)
3.3.2 Capital Cash out-flows
Pay dividends
Pay interest on bonds / debentures
Repay capital on bonds / debentures
Buyback shares
Make a capital distribution (pay out some share capital. Normal dividends are paid
from retained earnings)
3.4 Mismatch in timing of cashflows in insurance products

For most insurance products, cash in-flows occur mostly at the start of the policy, and
cash out-flows mostly at the end of the policy.
3.4.1 Immediate annuity example
Single premium is received up front at t = 0
Claims and expenses are paid every month or year from t = 1 for up to 30 or 40 years
3.4.2 10 year Endowment Assurance example
Premiums are paid regularly from t = 0
Small death claims are paid regularly from t = 1 to 10. These cashflows are much
smaller than the premiums
Expenses are paid regularly from t = 0 to 10. These cashflows are much smaller than
the premiums
At t = 10, a large maturity payment is made to the survivors. This is much larger
than the premium received in the last year.
3.4.3 Term Assurance example

Term assurance has closer matching of in- and out-flows. The major mismatch is
because the premium is level, but the risk increases over time due to increasing
mortality rates.
David Kirk
Page 30 / 107

Level premiums are paid each year
Expenses are paid each year
Claims are small at the start of the policy, but increase over time
It is likely that
Premiums > Expenses + Benefits for early durations
Premiums < Expenses + Benefits for later durations
David Kirk
Page 31 / 107

Section 4 : Actuarial reserves and liabilities
4 ACTUARIAL RESERVES AND LIABILITIES

This section covers the need for actuarial reserves and liabilities, explains the
difference between them and the types of reserves and liabilities calculated.
4.1 Actuarial reserves for solvency

As we saw in the previous section, cashflows for insurance company are
mismatched. That is, we receive cash before having to pay it out. Sometimes this
difference can be many years. The case of the immediate annuity is an easy example.
We receive cash upfront and then have to make payments for many years (sometimes
20 or 30 or even 40 years). It is important that we ensure we have enough money to be
able to make these payments when they fall due.
An insurance company is solvent if it can meet is liabilities as they fall due.
Thus, actuaries calculated the amount of money that must be held in reserve to
ensure that liabilities can be met as they fall due. It is common for this calculation to
be prudent since we have a responsibility to our policyholders to look after them.
In general, the more prudence in the calculation of the solvency reserves the better.
4.2 Actuarial liabilities for financial reporting

Financial reporting is the disclosure of:
the company's financial position (balance sheet); and
financial performance (income statement)
to allow current and potential future shareholders to:
assess the value of their shareholding;
estimate how the company will perform in future; and
make informed investment decisions about whether to buy, hold or sell the
company
and to assist management to:
make good operational and strategic decisions in the management of the company.
David Kirk
Page 32 / 107

4.2.1 External reporting

Shareholders and investors need to know the current financial position and past
financial performance of the company. They will also be interested in the solvency of
the company (if the company is insolvent and cannot pay policyholders, there will be
no money left for shareholders and other investors). However, they also need to know
accurately what the balance sheet and income statement of the insurer looks like.
Since most insurance products have large cashflows upfront followed by negative
cashflows at later durations, without calculating the liabilities due to policyholders
and placing this on the balance sheet, insurance companies will show large profits in
the early years of a policy, followed by large losses at the end of a policy. This clearly
doesn't reflect the realistic scenario for the insurer.
4.2.2 Internal reporting

Internal management also need to know the accurate financial position of the
company so that they can manage it effectively and make appropriate strategic and
operational decisions. They need to know the assets and liabilities on the balance
sheet, as well as how much profit the company makes as a whole, and how much
profit is made by each business unit or product line.
4.3 Difference in purpose between actuarial reserves and actuarial liabilities

Actuarial reserves for solvency are intended to protect policyholders against the risk
that the insurer becomes insolvent. Prudence is a good thing. More prudence is
generally better.
Actuarial liabilities for financial reporting are to provide shareholders and other
investors with accurate information about the financial position of the company. A
small element of prudence is good to ensure that profits are recognised prematurely.
However, too much prudence paints an unrealistic picture of the financial position of
the insurance company and can be misleading.
This is an important difference in the purpose of solvency calculations and financial
reporting.
A secondary difference is that financial reporting generally requires accurate
information for individual business units and product lines to enable management to
make appropriate decisions on this level. On the solvency side, as long as the entire
company is solvent, we are less concerned (although not unconcerned) with
individual product lines.
David Kirk
Page 33 / 107

4.4 Actuarial Balance Sheet

The largest elements of an insurers' balance sheet are the actuarial reserves or
liabilities. The illustration below gives an example of a life insurance company with
total assets = 100 and total liabilities = 75.
Actuarial Balance Sheet
80
60
40
20
Assets (Market Value or

Amortised Cost)
100
Shareholders Net
Asset Value
Free assets
Risk Capital
Margins
Liabilities
Best estimate
liabilities
0
Assets
Liabilities
Liabilities
Illustration 2.2: Actuarial balance sheet

The two sides of the balance sheet are the assets and liabilities. The illustration above
shows the assets on the left hand side, and two different breakdowns of the liabilities
on the right hand side.
4.4.1 Insurance company assets

Insurance companies' assets can be valued at market value or some form of
historical cost or amortised cost.
4.4.2 Liabilities
The middle column shows the most basic breakdown. 75 of liabilities and 25 of
shareholders' net asset value, which is also known as shareholders' equity or
shareholders' capital.
David Kirk
Page 34 / 107

The column on the right provides more detail. Total liabilities of 75 consist of best
estimate liabilities of 60 and margins for prudence of 15. The level of prudence will
differ between companies, between countries, and may be different for solvency
reserving and financial reporting liabilities. Prudence is usually introduced through
conservative assumptions.
4.4.3 Risk capital and free assets

Risk capital is capital held as an additional buffer against fluctuations in experience
and other risks to the insurer. Regulators may stipulate the minimum level of risk
capital that must be held over and above the reserves. Risk capital belongs to
shareholders, but may be needed to pay benefits to policies if experience is much
worse than expected.
4.5 Typical actuarial reserves

4.5.1 Prospective reserves
Net Premium Valuation
Gross Premium Valuation or Realistic Discounted Cashflow Valuation
4.5.2 Retrospective reserves
Unit fund reserves
Unearned Premium reserves
4.5.3 Approximate reserves

Some reserves or liabilities are difficult to calculate, or there may not be sufficient
information to calculate them accurately. In these cases, approximate methods may
be used. This is often the case when the overall size of these reserves / liabilities is
small.
Common approximations may include:
multiple of premium
multiple of sum assured
retrospective accumulation of premiums, expenses and benefits
4.5.4 Non-life reserves

Non-life insurance generally does not used discounted cashflow reserves. The
reserves commonly used for non-life insurance are:
David Kirk
Page 35 / 107

Unearned Premium Reserves (UPR) or Unearned Premium Provision (UPP)
Additional Unexpired Risk Reserves (AURR) or Additional Unexpired Risk

Provision (AURP) .
This is sometimes called the Unexpired Risk Reserve or (URR) but this is not
technically correct
Incurred But Not Reported Reserve
Outstanding claims reserve.
4.6 Special reserves

In addition to the above reserves, special reserves are sometimes held to meet
specific needs.
4.6.1 Bad data reserves

If there are concerns about the data on which the reserves are based (number of
policies, premium, age etc.) then a bad data reserve may be held to reflect the
potential additional outgo in future arising from corrections to the data.
A bad data reserve should not be held forever. The data problems should be fixed and
the reserves released. When a reserve is no longer held it is said to be released.
4.6.2 Mismatch reserves

A mismatch reserve may be held when assets and liabilities are not matched.
4.6.3 Investment Option and Guarantee Guarantee Reserves

Investment options and guarantees can present significant risks to an insurer. Many
insurance company failures have been the result of problems with investment
guarantees and options.
The reserves for this risks are usually calculated using stochastic models that
explicitly model the possible variation in investment returns and interest rates.
4.6.4 Non-Investment Option & Guarantee Reserves

Options and guarantees that do not involve investment returns are also significant
sources of risk. Some examples include:
Guaranteed Insurability Benefit
Optional premium increases
Implicit option to lapse a policy
David Kirk
Page 36 / 107

4.6.5 Regulatory special reserves

The insurance regulator may require additional special reserves to be held. These
could be a percentage of premiums, or a fixed amount for each class of business.
David Kirk
Page 37 / 107
Course Notes
2007 David Kirk
Chapter 3 Actuarial reserving techniques for life

insurance
About
This chapter covers the core techniques of actuarial reserving. We will
cover the net premium valuation and gross premium valuation.
Chapter Contents
1 Practical Reserving Principles........................................................ .............................40
1.1 Prospective valuation........................................................................................... ..40
1.2 Assumptions required (the basis)........................................ .............................40
2 Gross Premium Valuation...................................................................................... ......42
2.1 Definition of Gross Premium Valuation................................ ..............................42
2.2 Impact of basis changes on gross premium valuation........................................42
2.3 Examples of gross premium valuations......................................... ......................42
2.4 Example of impact of basis changes on the gross premium valuation...............43
3 Net Premium Valuation................................................................................. ..............44
3.1 Definition of Net Premium Valuation........................................................... .......44
3.2 Implicit allowance for expenses in net premium valuation................................44
3.3 Impact of basis changes on net premium valuation....................................... .....45
3.4 Examples of net premium valuations........................................... .......................45
3.5 Single premium policies under the net premium valuation...............................46
3.6 Alternative interpretations of net premium valuation.......................................46
3.7 Example of impact of basis changes on the net premium valuation..................47
4 Retrospective versus prospective calculation of reserves.......................................... .48
4.1 Assumptions required for equality.................................................................... ...48
4.2 Typical form of relationship................................................................................ .48
4.3 Intuitive interpretation of relationship.................................................. .............48
4.4 Derivation of equality for a Whole of Life Assurance.................................... .....49
Chapter 3 : Actuarial reserving techniques for life insurance

Section Chapter 3 : Actuarial reserving techniques for life insurance
5 Impact of reserves on profit and loss................................................. .........................50
5.1 New business strain.......................................................................................... .....50
5.2 Zillmerisation............................................................................ ............................51
5.3 Basis changes................................................................................................. ........52
5.4 Pattern of reserves over time.................................................................. ..............53
David Kirk
Page 39 / 107

Section 1 : Practical Reserving Principles
1 PRACTICAL RESERVING PRINCIPLES

1.1 Prospective valuation
In a prospective valuation, we take the reserve to be the present value of future
cashflows.
If we invest the reserves in assets that earn the discount rate, then the reserve will be
sufficient to meet future cashflows.
1.2 Assumptions required (the basis)

In valuing the expected present value of future cashflows, we need to make certain
assumptions regarding what these expected future cashflows will be. These
assumptions are collectively called the basis.
1.2.1 Mortality ( q x )
Expected mortality rates for each age are required. Often, this will be different for
males and females. These rates will be based on:
past mortality experience of the insurance company
industry data if this is available
data from reinsurers
mortality experience from similar countries
known or expected trends, such as annuitant mortality improvements and

deteriorating mortality from HIV/AIDS
1.2.2 Discount rate ( i )

The discount rate (or valuation rate) is used to calculate the present value of
cashflows for the reserves. It can be set in two ways:
1. The expected future investment return (or yield) on the assets backing the
reserves
2. The market-consistent yield expected on similar instruments in the market.
If the assets backing the reserves closely match the liability cashflows, then approach
1 and 2 will produce the same result.
David Kirk
Page 40 / 107

Section 1 : Practical Reserving Principles
Note, the discount rate we are considered is the expected future investment return
or yield. It does not necessarily reflect recent past investment returns and does not
directly affect the actual interest received on the investments.
1.2.3 Expenses
Future expenses (commission, salaries, rent, depreciation, electricity etc.) are
sometimes modelled explicitly (typically with a Gross Premium Valuation, but not for
a Net Premium Valuation). An assumption regarding the expected level of future
expenses per policy will be required.
1.2.4 Other decrements

A decrement is a change from one state to another. For example, death is a
decrement from the state alive to the state dead. Other decrements that are of
interest in insurance are:
Lapse (the policyholder stops paying premiums so the contract is cancelled)
Surrender (the policyholder stops paying premiums and receives a surrender

value and the contract is cancelled)
Paid-up (the policyholder stops paying premiums, by the policy continues under
modified terms, usually with a lower Sum Assured)
None, some, or all of these decrements may be modelled in a Gross Premium

Valuation.
1.2.5 Other assumptions

Depending on the product and valuation methodology, a range of other assumptions
may be required.
David Kirk
Page 41 / 107

Section 2 : Gross Premium Valuation
2 GROSS PREMIUM VALUATION

2.1 Definition of Gross Premium Valuation
A Gross Premium Valuation is a reserving technique that discounts the expected
value of all future cashflows, including the actual premium paid by the policyholder
(the Gross Premium) and any expenses expected to be paid in future.
We will use the symbol GP for the Gross Premium. In these examples, we model
expenses as E payable annual in advance as a fixed amount per policy.
2.2 Impact of basis changes on gross premium valuation

If the basis is changed (e.g. Mortality, discount rate or expenses) then the calculation
of the gross premium reserve must take these new assumptions into account. Thus,
the basis change is directly incorporated into the reserve.
2.3 Examples of gross premium valuations

The reserve for a policy in force (still active) at time t is calculated as
for each regular premium product below.
V as shown
2.3.1 Whole of Life Assurance

t
V =SAA x t GPa x t Ea xt
2.3.2 Term Assurance

t
V =SAA
xt : nt
GPa xt : nt Ea xt : nt
2.3.3 Endowment Assurance

t
V =SAA xt : nt GPa xt : nt Ea xt : nt
2.3.4 Annuity
For this annuity, we assume payments annually in arrears = Benefit, and expenses
also incurred annually in arrears.
t
V = Benefit Ea xt
David Kirk
Page 42 / 107

Section 2 : Gross Premium Valuation
2.4 Example of impact of basis changes on the gross premium valuation

Table 3.1.Basis change impact on gross premium valuation for Term Assurance
Basis change
Impact of basic change

on GP
Impact of basis change

on Reserve
Increase mortality
No change
Increase
Increase discount rate
No change
Decrease
Increase expenses
No change
Increase
David Kirk
Page 43 / 107

Section 3 : Net Premium Valuation
3 NET PREMIUM VALUATION

3.1 Definition of Net Premium Valuation
A Net Premium Valuation is a reserving technique that discounts the expected
value of future benefit cashflows and future Net Premium cashflows. The Net
Premium is not the same as the Gross Premium actually paid by the insured. This is
an important difference with the Net Premium Valuation.
We will use the symbol NP for the Net Premium. The Net Premium is calculated
as the premium necessary, at the outset of the policy (i.e. Where t= 0) to meet future
benefits.
No expenses are modelled as is the nature of the Net Premium Valuation.
3.2 Implicit allowance for expenses in net premium valuation

The net premium valuation does not explicitly allow for expenses. However, the net
premium used is generally lower than the gross premium actually paid by the
policyholder. The net premium is calculated without considering expenses, so it is
lower by approximately the amount required to cover expenses.
When we use the net premium in the net premium valuation, because it is lower than
the gross premium, the reserve is higher by approximately the same amount as the
allowance for expenses in the gross premium.
Consider the example where NP = GP E for a Whole of Life Assurance
Gross Premium Valuation
Net Premium Valuation
V =SAA xt NPa xt
which is also
V =SAA xt GPa xt GP NP a x t
which is also
And so the net premium valuation and gross premium valuation are equal if NP =
GP E. This is not always the case, particularly where the basis has changed since
the policy was issued.
David Kirk
Page 44 / 107

3.3 Impact of basis changes on net premium valuation

If the basis is changed (e.g. Mortality, discount rate or expenses) then the calculation
of the net premium reserve must take these new assumptions into account. However,
there are two impacts for every basis change:
1. The Net Premium is calculated on the new basis, as at the start of the contract
using the standard formula. Thus, the Net Premium changes when we change the
valuation basis (or the reserving basis or the set of assumptions used to calculate
the reserve.
2. The Net Premium Valuation reserve is then calculated on the new basis, but also
allowing for the recalculated Net Premium.
The impact of a particular basis change on the Net Premium will be opposite of the
direct impact of the basis change on the reserve. This means that the overall impact
on the net premium valuation will be less than that for a gross premium valuation.
We will consider some examples of this later in this section.
3.4 Examples of net premium valuations

The reserve for a policy in force (still active) at time t is calculated as
for each regular premium product below.
V as shown
3.4.1 Whole of Life Assurance

t
V =SAA xt NPa xt
SAA x
where NP =
a x
3.4.2 Term Assurance

t
V =SAA
xt : nt
NPa xt : nt
SAA
where NP=
a x : n
x :n
3.4.3 Endowment Assurance

t
V =SAA xt : nt NPa xt : nt
where NP=
SAA x : n
a x : n
David Kirk
Page 45 / 107

3.4.4 Annuity
For this annuity, we assume payments annually in arrears. However, because there is
no premium to be valued, we cannot allow for expenses as the difference between
gross and net premiums. Thus it is typical to allow explicitly for expenses even for the
net premium valuation. Once method is given below.
t
V = Benefit Ea xt
3.5 Single premium policies under the net premium valuation

The net premium valuation allows implicitly for expenses through the difference in
the net premium and the gross premium. However, for single premium policies, the
premium is not a component of the reserve. Unless we make an adjustment to the net
premium valuation when dealing with single premium policies, our reserves will not
be sufficient.
3.5.1 Single Premium Endowment Assurance

t
V =SAA xt : nt expense reserve
The expense reserve could be calculated in different ways. A common (and sensible)
approach would be as follows:
expense reserve = Ea xt : nt
In this case, the net premium valuation and gross premium valuation are the same.
3.6 Alternative interpretations of net premium valuation

Some practitioners also use a net premium valuation to be a valuation using:
net premium = gross premium expenses

The difference from this and a traditional net premium valuation is that the net
premium is calculated explicitly as gross premium less expenses, and that this
premium is not recalculated if the basis changes.
This is not strictly a net premium valuation in the formal sense. There is also the
requirement to test the net premium to ensure there is a sufficient gap between the
net and gross premium to cover future expenses
David Kirk
Page 46 / 107

3.7 Example of impact of basis changes on the net premium valuation

Table 3.2.Basis change impact on net premium valuation for Term Assurance
Basis change
Impact of
basic change
on NP
Impact of
Impact of NP
basis change
change on
(excluding NP
Reserve
change) on
Reserve
Combined
Impact of
basis change
and NP
change on
Reserve
Increase
mortality
Increase
Increase
Decrease
Small Increase
Increase
discount rate
Decrease
Decrease
Increase
Small Decrease
No change
No change
No change
No change
Increase
expenses
David Kirk
Page 47 / 107

Section 4 : Retrospective versus prospective calculation of reserves
4 RETROSPECTIVE VERSUS PROSPECTIVE CALCULATION OF RESERVES

Both the Gross Premium Valuation and the Net Premium Valuation are prospective
reserving techniques. Prospective means that we look forward to future cashflows
and take the present values of these cashflows to calculate the current reserve.
A retrospective calculation of reserves considers past cashflows to estimate the
current reserve. Under certain assumptions, we can show that the prospective and
retrospective calculation will give the same result.
4.1 Assumptions required for equality
Actual cashflows must be equal to expected cashflows under the basis used for the
prospective valuation
Premiums must have been calculated on the same basis as the valuation
4.2 Typical form of relationship

The equation below gives the typical form of the relationship. This version is for a
whole of life assurance.
t 1V =
[ tV P 1i SAq x t ]
p xt
The equation below gives the same equation for an annuity with annual benefits paid
in arrears.
t1
V =t
V 1i Benefitp xt
p xt
4.3 Intuitive interpretation of relationship

The reserves at time t, plus the premium received at the start of the year and
investment return on the reserve and premium just received, is sufficient to pay
claims during that year and set up the reserve required at the end of the year for all
surviving policyholders.
David Kirk
Page 48 / 107

Section 4 : Retrospective versus prospective calculation of reserves
4.4 Derivation of equality for a Whole of Life Assurance

Required to Prove:
[ tV P 1i SAq xt ]
V
=
t 1
p xt
Proof:
a xt 1
t 1V = SAA xt 1 P
V = SA[ A xt q xt /1 i ] 1 i / p xt P a xt 1 1 i / p x t
t 1
t 1V =
t 1V =
t 1V =
[ SA{ Ax
q x t / 1 i } 1i P a xt 1 1 i
p x t
1i SAA xt Pa x t P SAq xt
p xt
[ tV P 1i SAq x t ]
p xt
David Kirk
Page 49 / 107

Section 5 : Impact of reserves on profit and loss
5 IMPACT OF RESERVES ON PROFIT AND LOSS

5.1 New business strain
Reserves are often calculated on more prudent or conservative assumptions. This is
to ensure that
profits are not recognised prematurely
the insurance company has sufficient funds to meet its obligations to policyholders
as they fall due
If the reserving basis is more prudent that the basis used to calculate the premium,
then the reserve at time 0 will be greater than zero. Thus leads to new business
strain. New business strain is the loss incurred at the start of a policy due to a
prudent reserving basis (and also initial expenses, although this is not covered here).
This loss is temporary, and profits will emerge over time as the prudent assumptions
turn out to be more conservative than actual experience.
Consider the example below:
A policy pays a benefit of 110 in one year's time, regardless of death or survival. The
valuation discount rate is assumed to be 5% as this is a prudent assessment of the
expected future investment returns available in the market.
However, the premium is calculated using a discount rate of 10% because this is a
best estimate of the expected future investment returns available in the market. IT
should be easy to see that the single premium payable at the start of the policy is:
GP =100
The reserve at time 0 is
V =110 1.0511001 =4.76 0
Thus, before the first premium is received, we must create a reserve of 4.76. This
will cause a loss of 4.76 at t = 0. This is the new business strain.
After the first premium is received, the reserve will simply be the present value of
the benefit of 110 at 5%. This is 110 1.051 =104.76 . This reserve will earn
interest at 10% (not 5%, since 10% is our best estimate of actual investment returns).
At t = 1 the assets have grown from 104.76 to 115.24 ( 115.24 =104.761.10 ). We
must pay the benefit of 110 at t= 1, which leaves us with 5.24 profit.
David Kirk
Page 50 / 107

Thus, the new business strain was a temporary loss. It has been reversed by the
profit of 5.24 at t = 1.
Importantly, the present value of the profit at t = 1 5.241.101 =4.76
which is the same as the new business strain or loss at t = 0.
The present value of profits arising from the contract does not depend on the level
of prudence within the reserve. It depends on the actual cashflows (premiums,
claims and expenses) incurred. The reserves affect only the timing of the profits and
losses.
5.2 Zillmerisation
5.2.1 Impact of initial expenses on the Gross Premium
Zillmerisation is a technique used to allow for initial expenses incurred at the outset
of the policy. This is only necessary for the net premium valuation. For the gross
premium valuation, the premium would be calculated as follows:
GPa x : n = SAA x : n I Ea x : n
GP =
SAA x: n I
E
a x : n
Where E is the ongoing expenses incurred at the start of every year, and I is the initial
expenses incurred once at the outset of the policy.
David Kirk
Page 51 / 107

5.2.2 Zillmerised Net Premium Valuation

The Zillmerised Net Premium Valuation is given as:
t
a
V =SAA xt : nt NPa xt : nt I xt : nt
a x : n
where NP=
SAAx : n
a x : n
I =initial expenses
At t = 0, the third time in the above reserves equation = -I. As t approaches n, the
third term in the reserve equation above approaches zero. The reserve is decreased
by the amount of initial expenses initially to reduce new business strain. The
adjustment reduces over time as t approaches n and the policy approaches maturity.
This can be intuitively understood as increasing the net premium to reflect the
component of the Gross Premium that will be used to recover initial expenses is
recognised in the Net Premium Valuation. Without this Zillmer Adjustment, the Net
Premium will be too much lower than the gross premium and will result in very high
new business strain.
5.3 Basis changes

5.3.1 Gross Premium valuation
The full impact of the basis changes is reflected in the liability valuation. This change
flows directly through the income statement.
5.3.2 Net Premium valuation

The impact of the basis changes is reduced because the net premium itself also
changes when the basis is changed. The basis change has a smaller impact on the
income statement than the gross premium valuation.
5.3.3 Impact on financial reporting

If basis changes result in a large change in reserves or liabilities, then basis changes
will have a large impact on the profit of the company for a particular year.
David Kirk
Page 52 / 107

For example, if the total liabilities of a life insurer are USD10m and the average
annual profit is USD400,000, then a 5% increase in liabilities resulting from a basis
change (maybe an increase in mortality) will produce a USD500,000 increase in
liabilities, which is a corresponding decrease in profits of USD500,000. This will
erase all the profits for the year and make the insurance company make a loss.
1. Do you understand how a basis change can lead to a change in liabilities?

2. Do you understand how a change in liabilities will lead to a profit or loss in the
period in which the basis change is made?
3. Do you understand why one needs to be careful not to change the basis
unnecessarily?
4. Do you understand how basis changes might be used to manipulate profits?
1. Changing assumptions changes the expected present value of income and outgo.
Since liabilities (and reserves) are calculated as the expected present value of
outgo less the expected present value of income, the liability figure will change.
2. The accounting equation is A = E + L where A is assets, E is equity and L is
liabilities. A change in L without a corresponding change in A must lead to a
change in E. This change comes about through a profit or a loss.
3. Unnecessary basis changes will lead to inappropriate profits and losses,
distorting the financial performance of the company. This is particularly relevant
given the subjectivity involved in setting assumptions and issues of credibility of
data and random fluctuations.
5.4 Pattern of reserves over time

We have already covered the graphs of
V over time in lectures.
David Kirk
Page 53 / 107
Course Notes
2007 David Kirk
Chapter 4 Unit and non-unit reserves

About
This chapter will not be covered in the course and as a result the notes
provide only the outline of the methodology..
Chapter Contents
1 Unit-linked products........................................................................... .........................55
1.1 Charging structure............................................................................. ....................55
1.2 Benefit flexibility........................................................................................... ........56
1.3 Investment flexibility............................................................................ ................56
1.4 Investment Guarantees...................................................................... ...................57
2 Unit Fund or Unit Reserves................................................................ .........................58
2.1 Unit pricing........................................................................................................... .58
2.2 Unit reserve................................................................................................. ..........58
3 Non-unit reserves................................................................................................ .........59
3.1 Introduction................................................................................ ..........................59
3.2 Market practice around the use of non-unit reserves.........................................59
3.3 Names for non-unit reserves............................................................. ...................59
3.4 Calculation principles....................................................................... ...................60
Chapter 4 : Unit and non-unit reserves

Section 1 : Unit-linked products
1 UNIT-LINKED PRODUCTS
Unit-linked products (also known as Variable Universal Life products) are
sophisticated life insurance products that combine risk cover and savings benefits in
a flexible structure. The flexibility is two-fold:
1. The mix of risk benefits and savings contributions can be specified by the
policyholder at inception of the policy, and often changed within limits during the
term of the policy.
2. The investment vehicle(s) into which funds are invested can be selected by the
policyholder at the outset, and often changed within limits during the term of the
policy.
1.1 Charging structure

Premiums are paid by the policyholder. Of this premium, a certain proportion is
allocated to the unit fund. The unit fund operates much like a bank account and fees
/ charges are deducted from the unit fund. Investment return is credited to the unit
fund. The unit fund is also sometimes known as the unit account.
Death benefits are paid first out of the unit fund, and topped up by the insurer if the
Sum Assured is greater than the value of the unit funds. Thus, the risk premium
charged each month is calculated as Max SA Unit Fund , 0 q x t .
An annual asset management fee is also charged on a monthly basis. Policy fees,
administration charges, advice levies, bid-offer spreads and a range of other charges
may be levied.
One of the advantages of a unit-linked product is that the charging structure is
transparent in that the policyholder can see exactly what charges are levied.
However, in practice, policyholders often find that the charging structure is
extremely complicated and difficult to understand. High charges will inevitably lead
to poor value for money for policyholders.
The unallocated premium plus fees and charges plus risk premiums are income for
the insurer. Expenses and death benefits in excess of the unit fund are outgo for the
insurer.
David Kirk
Page 55 / 107

1.2 Benefit flexibility

1.2.1 Risk cover
A wide range of benefits can be offered. Each benefit is charged with a separate risk
premium. As long as the total premium paid by the policyholder is greater than the
risk premiums (required for the risk benefits) and expenses, the remaining premium
is added to the unit fund and will grow the unit fund over time.
1.2.2 Maturity benefits

At maturity, unless there is a guaranteed minimum maturity value (see investment
guarantees below) the unit fund will be paid out as a maturity value.
1.2.3 Surrender benefits

Surrender values are usually paid as the unit fund less some deduction for
unrecouped initial expenses incurred
some amount of the future profit foregone by the insurer
Poor surrender values have been a major criticism of unit-linked policies around the
world. This is exacerbated by complicated charging structures and poor
communication of surrender terms.
1.3 Investment flexibility

The investment return credited to the policyholder is the actual investment return
earned on the underlying funds as selected by the policyholder. Usually several
different funds are offered such as:
Balanced portfolio (equities, bonds, property and cash)
Aggressive portfolio (mostly equity)
Defensive portfolio (greater proportion of cash and bonds)
Cash-only portfolio (investing in money market instruments up to 90 days)
International portfolios (offering exposure to international currencies,

multinational companies and the domestic companies of other countries)
Sector fund (based on a particular sector of a market, e.g. financials or industrials

or commodities)
Absolute Return Funds / Hedge Funds
David Kirk
Page 56 / 107

The funds can either be own-branded (and managed) funds or funds offered by other
financial service providers. Some companies make use of white labelled where they
repackage another company's fund under their own brand.
1.4 Investment Guarantees

Investment Guarantees can be offered, ensuring that at least a minimum return is
offered. These will generally be charged for explicitly and require a great deal of care
to ensure that the fee charged is appropriate given the risks involved. The risks will
vary depending on the nature of the underlying fund selected by the policyholder so
the fee charged must also depend on the underlying fund or combination of funds
selected.
David Kirk
Page 57 / 107

Section 2 : Unit Fund or Unit Reserves
2 UNIT FUND OR UNIT RESERVES

2.1 Unit pricing
The unit account or unit fund is calculated as a price per unit or unit price and a
number of units. The unit fund for a particular policyholder is:
UF t =UP tN t
where UF t =Unit Fund at time t
UP t = Unit Price at tme t
N = Number of units at time t
The Unit Price will be the same for all policyholders, but the number of units will
differ so the unit fund will differ. Allocated premiums and charges are converted to
numbers of units before being added or subtracted from the unit fund. Investment
return increases the unit price of each unit.
This is the basic approach. The actual detail of unit pricing is complex and beyond
the scope of this course.
2.2 Unit reserve

The unit fund or unit reserve is taken as the market value of the units.
UF t 1=UF t APt chargest RP t 1 investment return t

where AP = Allocated Premium
RP =Risk Premiums
Allocated Premiums are the proportion of the total Gross Premium paid by
policyholders that is allocated to the unit fund. The unallocated premium is income
for the company as is effectively another form of charge. The Risk Premiums charged
are the actual risk premiums deducted from the unit fund to provide the risk
benefits.
This formula is a very simplistic version based on a very simple charging structure. In
practice, the calculation can be more complex.
David Kirk
Page 58 / 107

Section 3 : Non-unit reserves
3 NON-UNIT RESERVES
3.1 Introduction
Non-unit reserves are reserves held in respect of unit-linked business separate from
the unit fund or unit reserve. It is calculated as the expected present value of nonunit cashflows such as
unallocated premiums
fees and charges
expenses
non-unit risks benefits paid.
Since some of the charges and benefits paid on unit-linked policies depend on the
size of the unit fund, we need to project the unit fund into the future in order to
calculate the expected future non-unit cashflows.
The total reserve for a unit-linked product is the sum of the unit reserve and the nonunit reserve.
Non-unit reserves can be positive or negative, but are usually negative. A negative
non-unit reserve decreases the total reserve for a policy.
3.2 Market practice around the use of non-unit reserves

Since non-unit reserves are usually negative, it is not uncommon for a company to set
the non-unit reserve to zero. There are two primary reasons for this:
1. Prudence. By not holding a negative reserve the overall reserves are higher and
thus more prudent
2. Simplicity. The calculation of non-unit reserves is fairly complex.
3.3 Names for non-unit reserves

In many countries, non-unit reserve are named according the local currency. So, in
the United Kingdom, non-unit reserves are called Sterling Reserves and in South
Africa they are called Rand Reserves. The term non-unit reserve is generic and can
be used in all markets.
David Kirk
Page 59 / 107

Section 3 : Non-unit reserves
3.4 Calculation principles

1. Using a set of assumptions, project the unit fund forward allowing for allocated
premiums, charges, risk premiums and investment returns and taking decrements
into account.
2. Calculate the expected future level of non-unit cashflows for all policyholders.
3. Calculate the present value of these expected future cashflows
Why are non-unit reserves usually negative?
What does a positive non-unit reserve imply?
Non-unit reserves are usually negative for two reasons:
profitability products will have higher future income than outgo. If this difference
is greater than any prudential margins included in the valuation basis, the nonunit reserves will be positive. An advanced concept, which is not examinable and
will not be explained in detail is that of distortions caused by projecting and
discounting the unit funds in a real world (rather than market consistent)
manner.
Initial expenses are usually large. The unit-linked product will have been
designed to recoup these costs through future charges. These future charges are
capitalised to offset the cost of the high initial expenses.
A positive non-unit reserve is usually an indicator of an unprofitable or very low

profit product. This is unusual and the reasons behind positive non-unit reserves
should be investigated.
David Kirk
Page 60 / 107
Course Notes
2007 David Kirk
Chapter 5 Actuarial reserving techniques for nonlife insurance

About
This chapter covers the core techniques of actuarial reserving for non-life
insurance. We will cover the material in less detail than for life insurance
Chapter Contents
1 Introduction to reserving for non-life companies.............................................. .........63
1.1 Differences between life insurance non-life insurance........................................63
1.2 Need for reserves / liabilities for non-life insurance...........................................64
1.3 Explicit versus implicit prudence................................................. ........................64
2 Unearned Premium Reserves / Provisions..................................... ............................65
2.1 Introduction.................................................................................. ........................65
2.2 Distribution of risk over policy term.......................................... .........................65
2.3 UPR under uniform distribution of risk................................................. .............66
2.4 UPR under non-uniform distribution of risk............................................ ..........67
2.5 UPR for monthly premiums.......................................................................... .......68
3 Additional Unexpired Risk Reserves / Provisions......................................... .............69
3.1 Need for an AURR.............................................................................. ...................69
3.2 Calculation of the Unexpired Risk Reserve................................................ .........69
3.3 Calculation of the AURR.................................................................... ..................69
3.4 Alternative naming conventions........................................... ..............................69
4 Outstanding Claims Reserves / Provisions............................................................. ....70
4.1 Introduction........................................................................................ ..................70
4.2 Case estimates........................................................................ ..............................70
4.3 Statistical methods.................................................................................. .............70
4.4 Combined methods................................................................... ...........................71
5 Incurred But Not Reported Reserves / Provisions........................................... ...........72
Chapter 5 : Actuarial reserving techniques for non-life insurance

Section Chapter 5 : Actuarial reserving techniques for non-life insurance
5.1 Introduction............................................................................ ..............................72
5.2 Percentage of premium approach..................................................... ...................72
5.3 Loss ratio approach.................................................................................. .............72
5.4 Basic chain ladder methods.................................................... .............................73
5.5 Average cost per claim method......................................................................... ....74
5.6 Bornhuetter-Ferguson.................................................................................. ........78
David Kirk
Page 62 / 107

Section 1 : Introduction to reserving for non-life companies
1 INTRODUCTION TO RESERVING FOR NON-LIFE COMPANIES

1.1 Differences between life insurance non-life insurance
Life insurance and non-life insurance are different in many respects. The table below
outlines some of these differences.
Non-life Insurance
Life Insurance
Policies are short-term (often one year)
Policies are long-term, often ten years

or more
Claim frequencies are high
Claim frequencies are very low
More than one claim can occur on a

policy
Usually only one claim per policy
Claim amounts are variable
Claim amounts are usually fixed
Policyholders can easily change insurer
Policyholders can change insurer, but

given the long-term nature of the
policies it doesn't happen as often
The insured person is usually the

policy beneficiary
The insured person is usually not the

beneficiary
Claims are usually for indemnity

against loss
Claims meet a general financial need
Claim amounts are affected by

inflation
Claim amounts are generally fixed
Policies do not provide a savings

benefit
Policies frequently combine risk cover

and a savings element
Table 5.1.Comparison of life and non-life insurance

Group Life Assurance (GLA) policies have many of the same characteristics as
non-life insurance. Thus, many of the techniques used for non-life reserving
and also used for group life business.
David Kirk
Page 63 / 107

Section 1 : Introduction to reserving for non-life companies
1.2 Need for reserves / liabilities for non-life insurance

1.2.1 Solvency
Non-life insurance companies must hold reserves to ensure that they can meet
policyholder liabilities (claim payments and associated expenses) as they fall due.
1.2.2 Financial reporting

As with life insurance companies, accurate assessment of the company's financial
position and financial performance requires an assessment of the outstanding
liabilities.
1.2.3 Monitoring experience and re-pricing

Accurate pricing is critically important for non-life insurance. Premium rates are
regularly re-assessed based on claims experience. To measure recent claims
experience accurately, the insurer needs an assessment of claims that have been
incurred but have not yet been paid. If only paid claims are considered, the total
claims incurred will be underestimated and premiums rates may be set too low.
1.3 Explicit versus implicit prudence

Prudence is usually introduced to non-life reserves in an implicit manner through:
not discounting future cashflows
subjective increases in case estimates of claims
arbitrary increases in the size of reserves held (e.g. 10% higher)
David Kirk
Page 64 / 107

Section 2 : Unearned Premium Reserves / Provisions
2 UNEARNED PREMIUM RESERVES / PROVISIONS

2.1 Introduction
Premiums are typically received in advance of paying claims. These premiums when
received should not be immediately counted as profit as future claims would then
lead to losses.
At any point in time, a certain proportion of premiums received will relate to risk
cover to be provided in future. We consider these premiums as unearned because
the insurance company has not yet earned the premiums. Once the risk cover is
provided to the policyholders in future, and claims have been incurred, then the
premiums will be considered to be earned.
In the simplest sense, if premiums are paid annually in advance on 1 January each
year, half the premiums will have been earned by 30 June. The other half of the
premium will be unearned.
2.2 Distribution of risk over policy term

For most non-life insurance policies, risks are uniform over the lifetime of the policy.
This means that the probability of a claim, and the expected size of claims, is
constant over the term of the policy.
Some exceptions include:
Crop insurance.
Crop (wheat, grapes, fruit) insurance is highly dependent on weather conditions
which are seasonal.
Motor vehicle accident insurance

In some countries, hail, snow and other adverse weather conditions lead to higher
claims in certain seasons (typically winter).
Some examples where claims are uniform throughout the year:
Motor vehicle theft insurance

The risk of theft of a motor vehicle will usually be uniform over the term of a policy.
Public liability insurance
David Kirk
Page 65 / 107

Will buildings and household contents insurance (perils of fire, flood, water
damage, subsidence etc.) depend on the time of year? As an exercise, give some
examples of claim types that will depend on time of year, and some that are
uniform.
Flood and to some extent subsidence claims are usually season-dependent. Some
fires may be created from heating appliances used in the house during winter. Theft
and most other fires should be uniform throughout the year.
If houses are frequently left unoccupied for extended periods (summer houses, or
houses during extended holiday periods when families may vacate the house for
several weeks at a time) may be correlated with an increase in theft claims.
2.3 UPR under uniform distribution of risk

When risk is distributed uniformly over the term of the policy, the unearned
premium is taken as the proportion of the premium payment interval that lies in the
future for the most recently paid premium.
2.3.1 365ths method

The 365ths method consider the proportion of premium unearned measured in days
(a year divided into 365 days).
UPR =
days until next premium

Premium
days from last premium to next premium
2.3.2 24ths method

The 24ths method consider the proportion of premium unearned measured in halfmonths (a year divided into 24 half-months).
This method is an approximation to the 365ths method. Premiums are usually
assumed to be received halfway through a month rather that on the exact date of
premium receipt.
David Kirk
Page 66 / 107

2.4 UPR under non-uniform distribution of risk

If risk is not uniformly distributed over the term of the policy, the UPR needs to be
adjusted to reflect not only the proportion of time remaining for the most recently
paid premium, but the proportion of risk.
Example:
A policy has annual premiums (USD1,000) paid in advance on the 1st of January each
year. Risk increases linearly over the lifetime of the policy. What is the UPR at 30
June of each year?
Risk t = ct Where c is the gradient and t is the time in years

Total risk for the policy year is:
1
Total Risk = risk t dt

0
1
= ct dt
0
1
[ ]
ct 2
=
2
c
=
2
Risk remaining @ t = 0.5 = risk t dt

0.5
1
[ ]
ct 2
=
2 0.5
c c 3
= = c
2 8 8
Percentage risk remaining @ t = 0.5=
c3 /8 6
= =75 %
c/2
8
So the UPR required is USD750
David Kirk
Page 67 / 107

This can be expressed graphically as the area under the function of risk as a function
of time as shown below:
Earned Premiums under non-uniform distribution of risk
100%
Risk
75%
50%
25%
Risk Unearned
= 75%
Risk Earned =
25%
0%
0%
25%
50%
75%
100%
Risk Unearned
Risk Earned
Illustration 5.1: Graphical representation of unearned risk under non-uniform distribution

of risk
As an exercise, calculate the proportion of premiums unearned at 30 September if:
Risk t=e ct
Risk t =t 2t
ct
e 0.75c1
therefore UPR= c
premium
e 1
Risk t =e
Risk t=t 2t therefore UPR= 20.75330.752 premium
2.5 UPR for monthly premiums

For monthly premium policies, the UPR at the end of the month will always be zero.
Check that you understand why this is true.
David Kirk
Page 68 / 107

Section 3 : Additional Unexpired Risk Reserves / Provisions
3 ADDITIONAL UNEXPIRED RISK RESERVES / PROVISIONS

3.1 Need for an AURR
If the future expected claims are higher than the UPR calculated as described in the
previous section, the insurance company will need to hold additional reserves.
This usually occurs for one of two reasons:
1. Premium rates are too low and will not be sufficient to meet normal levels of
future claims
2. Significant additional risks have arisen since the premiums were set. For example,
court rulings may create a new precedent that will expose the insurer to higher
claims in future. Alternatively, an unexpected increase in vehicle replacement
parts may mean that future motor accident claims will be more costly to pay.
3.2 Calculation of the Unexpired Risk Reserve

The Unexpired Risk Reserve or URR is the reserve for unexpired risks. Unexpired
risks are risks arising from current policies. This is an estimate of future claim
payments for which the insurance company has a legal obligation to pay. This will
reflect the remaining term on existing policies as the insurer is not obligated to
renew policies and may increase the premium rates on policy anniversary in most
cases.
Typically, as is the case with most non-life insurance reserves, future claims are not
discounted in this calculation as an implicit form of prudence. Although it might be
considered more accurate to discount claims and use an explicit amount of prudence,
this is not market practice.
3.3 Calculation of the AURR

The calculation of the AURR is straightforward:
AURR =Max URR UPR , 0

If the URR is less than the UPR, no AURR is held.
3.4 Alternative naming conventions

Some insurance companies use the term URR to mean AURR.
David Kirk
Page 69 / 107

Section 4 : Outstanding Claims Reserves / Provisions
4 OUTSTANDING CLAIMS RESERVES / PROVISIONS

4.1 Introduction
The process to pay a claim begins once a claim has been reported to the insurance
company. However, there may be a delay of a few days to several years from the date
the claim is reported to the date the claim is finally settled.
Motor accident claims will be paid within a short period of time. The delays here are
caused by the need to examine the vehicle for damage, assess culpability and liability,
repair the damage and for the final claim amount to be determined.
For product liability claims, it may take several years for the courts to decide what
liability is owed by the insured.
Once the claim has been reported to the insurance company, the insurance company
must create an outstanding claims reserve (OCR) based on the expected claim
amount to be paid. There are two primary methods of assessing outstanding claims
reserves.
4.2 Case estimates

Specially trained (and experienced) claims assessors or loss adjusters consider the
details of each claim on an individual basis and estimate the size of the claim. This
relies on the expertise of the claims assessor.
This approach is useful when:
There are a small number of claims so the workload of assessing individual claims
is manageable.
The claims are large or unusual in nature so human judgement is required.
The claims can be accurately assessed based on the details of the claim. For
example, for a specific model vehicle, the cost of replacing a specific part or
repairing a specific body panel should be known with high certainty.
4.3 Statistical methods

Statistical methods are used when there is a very large number of claims.
Considerable time would be required to estimate claims individually, and the impact
of errors in any individual claim estimate should be small given the large number of
claims.
David Kirk
Page 70 / 107

Section 4 : Outstanding Claims Reserves / Provisions
The method might allow an average claim amount per claim type, policy type and
peril type. More complicated approaches can be adopted if required.
Only very large insurers with very large, homogeneous books use statistical methods.
Example:
Average claim amount per claim estimated over recent years has been USD250. We
have 34 reported claims that have not yet been paid. What is the total OCR?
OCR =34250 =USD 8,500
4.4 Combined methods

In some cases, both methods are used together. For example, claims assessors may
make case estimates for each individual claim. The insurance company may then use
statistical methods to adjust these estimates for inflation or assessor bias.
David Kirk
Page 71 / 107

Section 5 : Incurred But Not Reported Reserves / Provisions
5 INCURRED BUT NOT REPORTED RESERVES / PROVISIONS

5.1 Introduction
Claims are not always reported to the insurance company immediately. Delays may
be:
Several days for a motor vehicle accident
Several weeks for a house burglary while the occupants were away on holiday
Several years for product liability where the product defect is not known for several
years.
At any given valuation date (the date at which we calculate reserves) there will be an
unknown number of claims that have been incurred (i.e. have happened) but have
not been reported. Each claims will have an unknown size. This poses a seemingly
difficult question, How do you reserve for claims that you don't know exist?
This is the subject of this section. These claims are called Incurred But Not
Reported claims or simply IBNR claims. The reserves for these claims are known
strictly as IBNR Reserves but often simply IBNRs.
5.2 Percentage of premium approach

The most basic approach is to calculate IBNRs as a fixed percentage of premium.
Given that claims will generally increase with increased premium volumes (more
policies, larger risks), this is a reasonable simple approach. The problem is that this
approach alone does not help us to assess what percentage of premiums to use.
5.3 Loss ratio approach

The simplest approach that does help to assess the absolute level of the IBNR reserve
is called the loss ratio approach. A Loss Ratio (LR) is the ratio of total claims
incurred to total premiums earned over a certain period for a certain line of business
of other grouping.
LR =
Claims Incurred
Premiums Earned
The LR approach assumes that claims (losses) can be expressed as a stable percentage
of premiums. This is typically the way policies are priced, and for large portfolios of
business the LR may be relatively stable. The LR will usually be estimated in
conjunction with the pricing and underwriting department(s).
David Kirk
Page 72 / 107

Year
Earned
Assumed Expected
Premiums Loss Ratio Claims
Reported
Claims
Unreported
Claims
200x
EP
LR
EC=
EP * LR
2004
50,430
60%
30,258
30,100
158
2005
67,800
60%
40,680
38,400
2,280
2006
66,300
70%
46,410
41,310
5,100
2007
75,070
55%
41,289
23,650
17,639
Total IBNR
RC
IBNR =
EC - RC
25,177
Table 5.2: Loss Ratio approach to calculating IBNR reserve
The table above shows the typical way in which the LR approach is applied.
5.4 Basic chain ladder methods

The Basic Chain Ladder (BCL) method is a more sophisticated method of estimating
IBNR reserves.
5.4.1 Assumptions required
The past development pattern of claims will continue in future. (Future reporting
delays will be similar to past reporting delays)
Future claims inflation will be a weighted average of past claims inflation
The past claims experience on which the reporting delays are based includes at
least one year that is fully run-off. (At least one past year of claims must be 100%
reported)
David Kirk
Page 73 / 107

5.5 Average cost per claim method

Another method often adopted is called the average cost per claim. It is similar to
the BCL method in that it assumes past development patterns (reporting delays) will
continue in future. However, it assumes the past development pattern measured in
terms of number of claims will be constant in future and similar to the past. (The
BCL assumes past development patterns measured in amounts of claims will be
constant in future and similar to the past).
The reserve is calculated in two steps:
1. Using the past development pattern, calculate how many unreported claims have
been incurred
2. Separately, calculate an average cost per claim based on past claims information.
This figure may be adjusted if necessary to reflect known changes in claim sizes
due to inflation, court precedents, changes in policy conditions etc.
More basic methods are sometimes used for step 1 such as subjective assessment of
the number of unreported claims, or the average number of unreported claims over
similar periods in the past.
David Kirk
Page 74 / 107

5.5.1 Overview of approach

The approach can be separated into distinct steps. We will consider the example set
at 31 December 2006. Note that in this example we only use three years of claims
information. In practice more detail could be (and generally would be) used.
1. Tabulate reported claims year in which claim was incurred, and the number of
years delay to the claim being reported.
Cclaim year, reporting delay are shown in their correct positions in the table below
Table 5.3.Tabulation of claims by year of incidence and reporting delay
Reporting Delay
2004
C2004,0
C2004,1
C2004,0
2005
C2005,0
C2005,1
Not available
2006
C2006,0
Not available
Not available
Claim Year
The not available cells are not yet available since (at the end of year 2006) we do not
know about claims that will be reported in 2007 and 2008.
2. Then we cumulate the claims in a similar table where each column includes the
total claims reported up to that point as shown in the table below.
Table 5.4.Tabulation of cumulative claims by year of incidence and reporting delay
Reporting Delay
2004
C2004,0
C2004,0+C2004,1
C2004,0+C2004,1 + C2004,2
2005
C2005,0
C2005,0+C2005,1
Not available
2006
C2006,0
Not available
Not available
Claim Year
David Kirk
Page 75 / 107

3. Once we have the cumulative claims reported for each year and each reporting
delay, we can calculate link ratios that link the claims reported by reporting delay
x to the claims reported by reporting year x+1.
link 0,1=
C 2004,1 C 2005,1
C 2004,0 C 2005,0
link 1,2=
C 2004,2
C 2004,1
There are fewer data points available for link 1,2 than for link 0,1 because of
the triangular nature of the information available.
4. We can use the link ratios to estimate the empty cells of the cumulative claims
table:
Table 5.5.Tabulation of cumulative claims by year of incidence and reporting delay
Reporting Delay
2004
C2004,0
C2004,0+C2004,1
C2004,0+C2004,1+C2004,2
2005
C2005,0
C2005,0+C2005,1
(C2005,0+C2005,1) x link1,2
2006
C2006,0
C2006,0 x link0,1
C2006,0 x link0,1 x link1,2
Claim Year
5. The right-most column now includes estimates of total claims incurred (provided
the first row is fully run-off). If we subtract total reported claims from the total
estimated claims incurred, we get an estimate of total IBNR claims.
David Kirk
Page 76 / 107

6. A slight modification of the approach is to calculate cumulative development
factors (cdf)based on the product link ratios. The calculations are algebraically
equivalent, but the cumulative development factors are easier to work with and
have another useful interpretation.
cdf 0=link 0,1link 1,2

cdf 1=link 1,2
7.
1
= percentage of claims reported by time t.
cdf t
5.5.2 More detailed worked example

Make sure you receive a copy of the more detailed worked example in an electronic
spreadsheet.
5.5.3 Problems with BCL method

The BCL method works well if the assumptions underlying it are appropriate.
1. When might the assumption of stable development patterns not hold?
2. When might the assumption relating to inflation not hold?
1. The development pattern will change if any of the factors affecting any of the
causes of delays change. For example, different brokers may report claims with
different delays, so changes in brokers or the amount of business arising from
certain brokers could change the pattern. Administration systems and processes
may change resulting in shorter or longer delays (usually longer in the short
term, and hopefully shorter in the long-term as the benefits of the system change
are recognised). Outsourcing of administration functions can have a dramatic
change in the reporting delays.
2. If the economy has moved from a high inflation phase to a low inflation phase or
vice versa. This has happened to many developing economies as tighter monetary
policy and more careful fiscal policy have led to greater stability and lower
inflation.
David Kirk
Page 77 / 107

5.6 Bornhuetter-Ferguson.
More advanced methods exist for estimate IBNR reserves. None of these will be
covered in this course. They are generally covered in a specialised non-life course.
The most commonly used advanced method is known as the BornhuetterFerguson method or BF method. It uses Bayesian techniques to combine
estimates using the loss-ratio approach and the basic chain ladder method based on
the credibility of the available data.
David Kirk
Page 78 / 107
Course Notes
2007 David Kirk
Chapter 6 Assets and Liability considerations

About
This chapter covers the some basic principles for considering assets and
liabilities together. The major aim of this section is to introduce the
concepts of matching and immunisation.
Chapter Contents
1 Yield Curves................................................................................................ ..................80
1.1 Factors affecting the shape of the yield curve......................................................80
1.2 Zero Coupon Yield Curve......................................................... .............................81
2 Matching of assets and liabilities............................................................ ....................82
2.1 Introduction to matching............................................................. ........................82
2.2 Perfect cashflow matching and the Law of Once Price......................................82
2.3 Difficulties of perfect cashflow matching.......................................................... ..83
3 Measures of the term of financial instruments....................................... ...................84
3.1 Discounted Mean Term......................................................................................... 84
3.2 Volatility............................................................................................... .................85
3.3 Comparison of DMT and volatility........................................................ ..............86
4 Linear approximations to change in prices....................................................... ..........87
4.1 General form of first-order linear approximation...............................................87
4.2 Example for a zero coupon bond............................................... ..........................87
4.3 Example for 2 year coupon bond............................................... ..........................88
4.4 Second-order approximation........................................................................... ....88
5 Immunisation................................................................................. .............................90
5.1 Purpose of immunisation.................................................................. ...................90
5.2 Assumptions......................................................................................... ................90
5.3 Requirement of immunisation...................................................... .......................91
5.4 Example of immunisation............................................................................ ........92
6 Mismatch example........................................................................ ..............................94
Chapter 6 : Assets and Liability considerations

Section 1 : Yield Curves
1 YIELD CURVES
A yield curve is a function (or graph, if presented graphically) of the Yield To
Maturity (YTM) for fixed interest securities with term to maturity T.
An example is given below:
Yield Curve
9.
5
10
.5
11
.5
12
.5
13
.5
14
.5
8.
5
7.
5
6.
5
5.
5
4.
5
3.
5
2.
5
1.
5
6.0%
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
0.
5
YTM
9.0%
8.0%
7.0%
Term
Illustration 6.1: Example of a Yield Curve
1.1 Factors affecting the shape of the yield curve

This have been covered in lectures and are only listed below:
Expectations of future short-term interest rates
Preference for liquid (short-term) instruments
Market Segmentation
A fourth factor, called Market Preference, is sometimes included. It is a weaker form

of the Market Segmentation theory, where different investors in the market have a
preference for instruments of a certain term. This contrasts with Market
Segmentation where the investors only purchase instruments within a particular
range of terms.
David Kirk
Page 80 / 107

Section 1 : Yield Curves
1.2 Zero Coupon Yield Curve

A normal yield curve shows the YTM for a bond with a given maturity date. However,
the YTM is the single discount rate used to discount all the cashflows (coupons and
principal) for a given bond.
A zero coupon yield curve (sometimes known simply as a zero curve) shows the
yield to maturity on zero coupon bonds for the given maturity. This YTM is applies
only to cashflows of a specific term. Thus, the rates from the zero coupon yield curve
can be used to discount specific cashflows of the appropriate term.
Illustration 6.2: Graph comparing a zero coupon yield curve against a yield curve.
Yield Curve and Zero Coupon Yield Curve

10.50%
9.50%
Yield
8.50%
7.50%
6.50%
5.50%
4.50%
3.50%
1
term
Yield Curve
Zero Coupon YC
Note that the slope (gradient) of the Zero Coupon Yield Curve is greater in
magnitude than the Yield Curve. The Yield Curve can be thought of as a form of
weighted average of the Zero rates, where the weights are related to the size of the
cashflows (coupons and principal).
Coupon bonds can also be priced using the zero coupon yield curve by discounting
each cashflow (coupon and principal again) at the appropriate rate from the zero
coupon yield curve.
David Kirk
Page 81 / 107

Section 2 : Matching of assets and liabilities
2 MATCHING OF ASSETS AND LIABILITIES

2.1 Introduction to matching
Matching assets and liabilities means to choose assets with the same characteristics
as the liabilities to reduce risk. Assets and liabilities are both financial instruments
and may vary in value independently. If the assets and liabilities have similar
characteristics, we would expect them to vary in a similar fashion to changes in
external conditions.
The three primary characteristics considered are:
1. Term
2. Currency
3. Nature (fixed or real)
However, there may be other factors such as sensitivity to particular market prices
that could be considered.
The concept of matching is similar to that of hedging.
2.2 Perfect cashflow matching and the Law of Once Price

If we have two instruments, A and B, which produce the exact same cashflows as each
other under all circumstances, they must have the same price. This is known as the
Law of One Price.
This should be fairly obvious, but it can be explained further by considering what
might happen if A and B weren't priced the same.
Assume P A P B
Then, we purchase instrument B and sell instrument A. Since P A P B , this yields
profit =P A P B . Since all future cashflows from B will perfectly match
cashflows required to pay to the owner of A, we cannot make a loss in future.
Thus, we have made risk-free profit. The profit we made, , is made with no
risk. This breaks the no arbitrage assumption.
David Kirk
Page 82 / 107

Section 2 : Matching of assets and liabilities
Further, we could continue to buy B and sell A to make profit over and over again.
However, buying B would increase the price of B, and selling A would decrease the
price of A. Eventually, the price of A and B would be equal. Since there is no risk to
this activity, anybody who found the price difference would immediately buy as
much of B as possible, and sell as much of A as possible. Thus, the pressure to
equalise prices A and B would act very quickly and with great force to make the
prices equal.
2.3 Difficulties of perfect cashflow matching

In practice, it is often difficult to find two securities with the exact same cashflows
under all circumstances. In this case, we make use of more approximate matching
techniques. We attempt to match the currency, nature and average term of cashflows.
One should also try to match the convexity of the instruments as closely as possible.
Convexity is covered in later sections in this Chapter.
1. Why is it important to match annuity cashflows?

2. Why is it less important to match term assurance cashflows?
1. Annuities have a high DMT and large reserves. Changes in interest rates give rise
to a large percentage change in the reserves, and a large absolute change since the
reserves are large.
2. Term Assurances generally have fairly small DMT and small reserves. A change in
interest rates gives rise to a smaller percentage change in reserves than annuities,
and as the reserves themselves are usually very small compared with annuities,
the absolute financial impact is low.
David Kirk
Page 83 / 107

Section 3 : Measures of the term of financial instruments
3 MEASURES OF THE TERM OF FINANCIAL INSTRUMENTS

3.1 Discounted Mean Term
The Discount Mean Term (DMT) is the average term of the financial instrument,
where the term of each cashflow is weighted by the Present Value of that cashflow.
Below we give the derivation of DMT for an N-year bond with annual coupons C
payable in arrears.
P i=price of instrument
Cv t 100v n
P i=
t=1
tw t
DMT i= t=1
N
wt
t=1
w t =Cv t for t[1, N 1]

w N =Cv N 100v N
[
[
N
tCv n100v
DMT i=
t=1
Cv t 100v n
t=1
From the equations above, it should be clear than the DMT of an N-year zero
coupon bond is N.
A zero coupon bond is a bond that does not pay coupons. It only repays the
principal on maturity. Thus, the formula for the price of a zero coupon bond is
simply P =100v N
David Kirk
Page 84 / 107

3.2 Volatility
Volatility in this sense means the sensitivity of price to interest rates. It is related to,
but not the same as, volatility in the sense of random variability of market prices of
financial instruments.
3.2.1 General form

P i=price of instrument
Pi
P / i
/Pi=
=volatility
i
Pi
3.2.2 Example for a bond

Consider an n-year bond with coupons C payable annually in arrears.
P i=
Cvt
t=1
100v n
Pi
P i
/ Pi =
=
i
P i
tCvt1
t=1
Cv t
t=1
tCvt
t=1
1i
100v n
]
]
n100v n
Cvt
t=1
n100v n1
100v n
We obtain the derivative through use of the chain rule.

Taking the derivative requires some care because v=
1
1i
David Kirk
Page 85 / 107

3.3 Comparison of DMT and volatility

DMT and volatility are similar measures. However, they are defined slightly
differently and differ in numerical value for the same instrument.
DMTi P / i
=
1i
Pi
Thus, the DMT for an N-year zero coupon bond is N, but the volatility is
N
.
1i
David Kirk
Page 86 / 107

Section 4 : Linear approximations to change in prices
4 LINEAR APPROXIMATIONS TO CHANGE IN PRICES

4.1 General form of first-order linear approximation
For a small change in interest rate from i to i h , we can approximate
P A i h as:
P i
P A i
P A i h PA i h A =P A i h
P i
i
iP A i A
4.2 Example for a zero coupon bond
Take an N-year zero coupon bond with price P A i at interest rate i and par value =
100. The formula for the price is P A i =100 v
derivative with respect to i of
and has volatility
N
and
1i
N
100 v N .
1i
To continue the example, we will assume
N =10
10
We calculate the value of the bond at 5% to be: P A 5 % =100v =61.39

Now, if we need to calculate the value of the bond at 5.1%, we can use the linear
approximation as:
P A 5 %0.1 %=P A 5%0.1%

=61.39
0.001N
P 5 %
10.05 A
0.01
1
61.39=61.39
=61.390.58=60.81
1.05
105
So the approximation yields 60.81. If we now calculate the value directly using the
price formula, we get:
P A 5.1 %=1001.05110 =60.81

In this case, for this instrument, for a 0.1% change in interest rates, the result is the
same up to two decimals places.
Check for yourself that the results differ from the third decimal place.
David Kirk
Page 87 / 107

4.3 Example for 2 year coupon bond

Consider a 2-year bond, par value 100 that pays a 5% annual coupon in arrears.
1. Calculate the price at i = 5% and at i = 6% directly using the price formula.
2. Calculate the DMT at i = 6%
3. Calculate the volatility at i = 6%
4. Calculate the price at i = 6.2% using the first-order linear approximation
5. Calculate the price at i = 6.2% directly using the price formula.
4.4 Second-order approximation

The first-order linear approximation is only useful for very small changes in i. For
larger changes, we need to consider higher order derivatives to more closely match
the shape of the price formula curve (the graph of price against interest rates).
The formula for the second-order approximation extends the approach for the firstorder linear approximation:
2
P A i
2 P A i
P A i h PA ih
h 2
i
i /2
The graph below shows the blue curved line of the price of a 10-year ZCB for different
interest rates. The straight dotted orange line is the first-order linear approximation.
The approximation is reasonably close close the point of tangent, but gets
progressively worse as we move away from that point.
The curve dotted green line is the second-order approximation. As can be seen from
the graph, it is a much better fit at most interest rates.
David Kirk
Page 88 / 107

%
10
.0
0%
9.
0%
8.
0%
7.
0%
6.
0%
5.
0%
4.
0%
3.
0%
2.
1.
0.
0%
100.0
90.0
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0%
Price
10-year Zero Coupon Bond Price
YTM
Price
First Order Approximation
Second Order Approximation
Illustration 6.3: Price sensitivity of a 10-year Zero Coupon Bond
David Kirk
Page 89 / 107

Section 5 : Immunisation
5 IMMUNISATION
5.1 Purpose of immunisation
The purpose of immunisation is to ensure that no losses will be made on changes in
interest rates. It is a method of matching that can be used when exact cashflow
matching is not possible or too expensive.
Note that immunisation in practice is almost never possible. It requires the
assumption of a flat yield curve where changes in interest rates occur only through a
parallel shift in the yield curve. This does not happen in practice, since it implies an
arbitrage opportunity where risk-free profits can be made by investors.
The discussion of arbitrage opportunities is beyond the scope of this course.
5.2 Assumptions
Interest rates are equal for all terms. (The yield curve is flat.)
Changes in interest rates are equal at all durations.
Note that this second point is a necessary result of the first point. If the yield curve
is flat (the same interest rate at all durations) then any change in the yield curve at
one point must result in the same change at every point on the yield curve.
All changes in interest rates are small. Portfolios can be rebalanced without cost after
all small changes in interest rates.
Again, this assumption is not necessarily true in practice. Immunisation only
protects against small changes in interest rates. Large changes occur in practice.
Secondly, there are costs to rebalancing portfolios to ensure an immunised position.
David Kirk
Page 90 / 107

5.3 Requirement of immunisation

5.3.1 Some definitions
All definitions given for interest rate i
f A i =value of assets
f L i=value of liabilities
f A i
/
=f A i=first derivate of asset value
i
f L i
=f /L i =first derivate of liability value
i
2
f A i
=f //A i=second derivate of asset value
2
i
2 f L i
i
//
=f L i =second derivate of liability value
5.3.2 Requirements for immunisation

For immunisation, the following three criteria must hold:
f A i =f L i The value of assets and liabilities must be equal
f /A i =f /L i The volatility of assets and liabilities must be equal
f //A i f //L i The convexity of assets must be greater than convexity of

liabilities
Volatility here is defined as the sensitivity of the price of the instrument to changes in
interest rates. Similarly, convexity is defined as the sensitivity of volatility to changes
in interest rates.
David Kirk
Page 91 / 107

5.3.3 Practical interpretation
Assets and liability values must be the same (i.e. USD100 of Assets = USD100 of
Liabilities)
The volatility or DMT of assets and liabilities must be the same (we call this being
duration matched)
The spread of assets (by term) must be greater than that of liabilities
5.4 Example of immunisation

We have an obligation of USD100 in 5 years time. Current interest rates are 5% per
annum. We have two assets backing this liability.
Asset X pays 52.5 in year 6
Asset Y pays 47.62 in year 4.
Assets A = X + Y
Check that
f L 5%=78.35
f X 5 %=39.18
f Y 5 %=39.18
f A 5 %=f X Y 5 %=78.35
f L i=1001i5
f /L i=51001i 6
f /A i =652.51i7 447.621i5
f /L 5% =f /A 5%
Note that volatility, or sensitivity to changes in interest rates is negative. This
confirms the principle that bond prices move inversely with changes in interest
rates.
f //L i=301001i 7
f /A i = 4252.51i8 2047.621i 6
f //A i f //L i
David Kirk
Page 92 / 107

So it appears that all the requirements have been met. Let's consider what happens
the the assets and liabilities if we increase or decrease interest rates. If
immunisation is working, we should make no profit or a small profit (A>L) in
both cases.
For i = 4%
Check that
f L 4 %=82.19
f X 4 %=41.49
f Y 4 %=40.70
f A 4 %=f X Y 4 %=82.20f L 4 %=82.19
For i = 6%
Check that f L 6%=74.73
f X 6%=37.01
f Y 6 %=37.72
f A 6 %=f X Y 4 %=74.73f L 6 %=74.73
For both an interest rate increase and decrease, our assets are higher (or not lower)
than the liabilities. Thus, we are immune to small changes in interest rates.
David Kirk
Page 93 / 107

Section 6 : Mismatch example
6 MISMATCH EXAMPLE
Take an immediate annuity for a life aged 50. The DMT for this liability is 20 at an
interest rate of 4%. The reserve or liability is USD80,000. We invest the assets
backing the reserve into a Zero Coupon Bond with term of ten years.
1. What is the DMT of the ZCB?
2. What is the Volatility of the ZCB?
3. What is the Volatility of the annuity liability?
4. What happens if interest rates increase to 4.5%?
5. What happens if interest rates decrease to 3.5%?
6. What practical implications does this have?
7. What should you do to reduce this risk?
8. Why might it be difficult to perfectly match the liability cashflows arising from
annuity products?
1. 10
2.
3.
10
1.04
=9.62
20
=19.23
1.04
4. Assets decrease to 76,254. Liabilities decrease to 72,683. We make a profit of

3,571.
5. Assets increase to 83,950. Liabilities increase to 88,095. We make a loss of 4,145.
6. The company is exposed to interest rate risk since changes in interest rates have
the potential to cause fluctuations in financial performance. Further, we should
be particularly concerned about the potential for interest rate decreases since this
will lead to a loss.
David Kirk
Page 94 / 107

Section 6 : Mismatch example
7. We should try to match our liabilities more closely by selling the 10-year bond
and buying longer term fixed interest securities. The first step should be to have
the DMT of assets = to that of liabilities. Next steps would be to consider perfect
cashflow matching, or attempting immunisation. More complicated hedging
strategies such as entering into derivative contracts such as swaps could be
considered.
8. Annuity cashflows have a long average term. It might be difficult to find assets in
the market with sufficiently long term. This will prevent DMT matching,
cashflow matching and immunisation.
Even if available, the assets may be very expensive due to large demand and
limited supply of these long-term instruments. Even if the expected annuity
cashflows could be matched perfectly matched, the annuity cashflows depend
on mortality risk. If policyholders experience different mortality from assumed
(either due to incorrect or basis or simply random fluctuations) then the actual
cashflows will not be matched. This could lead to profits or losses, but would
also require a rebalancing of the portfolio to match the updated cashflows. This
rebalancing can be expensive and the updated assets may not be available.
Perfect cashflow matching also requires the premiums received from new
policies to be invested in exactly matching assets as soon as the funds are
received. In practice, new business cashflows are matched monthly in large,
fairly sophisticated companies. A few companies match new business on a weekly
basis. Again, the costs of this frequent rebalancing can be large.
David Kirk
Page 95 / 107
Course Notes
2007 David Kirk
Chapter 7 Pension fund reserving

About
This chapter covers the basic actuarial techniques used for the calculation
of retirement fund reserves.
Chapter Contents
1 The Nature of Pension Funds................................................................................ .......97
1.1 Defined Benefit (DB) structures...................................................... .....................97
1.2 Defined Contribution (DC) structure........................................................... .......98
1.3 Movement towards Defined Contribution............................... ...........................98
1.4 Spouse benefits.................................................................................... .................99
2 Pre-funding versus cash-based approaches................................. .............................100
2.1 Cash or pay as you go system......................................................... ...................100
2.2 Complete full funding from inception.................................................... ............101
2.3 Practical pre-funding systems......................................................................... ....102
3 Projected Unit Credit Method............................................................................ ........103
3.1 Reserving for benefits in payment..................................................................... ..103
3.2 Reserving for active employees....................................................... ....................103
3.3 Change in reserve.................................................................. ..............................105
Chapter 7 : Pension fund reserving

Section 1 : The Nature of Pension Funds
1 THE NATURE OF PENSION FUNDS

1.1 Defined Benefit (DB) structures
1.1.1 Typical defined benefit structure
A defined benefit scheme is one where the benefit is defined and guaranteed by
the employer. A typical benefit definition would be:
Pension NRA =Salary NRAService p %

where PensionNRA is the annual pension payable at Normal Retirement Age
SalaryNRA is the final salary paid at normal retirement age
Service is the number of years service to the employer
p% is the specified percentage, usually somewhere between 1.5% and 2.5%.
This could also be defined as:
Pension 0=Salary NRAServicep % where Pension0 is the pension payable 0

years after Normal Retirement Age.
The exact notation is not as important as the concept and the calculation. You must
consider how the terms are defined and use them appropriately or define your own.
No equation is useful unless all terms have been appropriately defined. This is a
general comment that applies across all sections (and should apply to your other
courses too).
Inflationary increases in the pension payable during retirement are usually at the
discretion of the pension fund trustees and will depend on available funds and
employee contributions. Some level of increase (either absolute or relative to
inflation) may be implied or even guaranteed.
1.1.2 Interpretation of typical defined benefit structure

For every year of service (employment), the employer promises to pay p% of the
employees final salary as a pension once the employee retires at Normal Retirement
Age.
If p% is 1.5% and the employee works for the company for 40 years, the pension
payable is 60% of final salary.
David Kirk
Page 97 / 107

If p% is 2.5% and the employee works for 30 years, the pension payable will be 75%
of final salary.
The factor p% is part of each employer's remuneration package and is usually entirely
at the discretion of the company.
1.1.3 Some variations on typical defined benefit structure
SalaryNRA may be an average of the last x years of salary (often 3 years). This reduces
the potential for manipulation and unequal treatment between equal employees
Normal Retirement Age might be a band of ages rather than a single specific age,
with possibly adjustments to the final benefit depending on actual retirement age
p% may very outside the usual range specified. p% may differ for different levels of
employees.
Additional years of service may be purchased through once-off contributions.
1.2 Defined Contribution (DC) structure

A defined contribution pension scheme is one where the contributions are
defined clearly. The actual benefit paid depends entirely on the investment
performance of the underlying assets. This scheme operates as a savings vehicle with
no guarantees made in terms of replacement of actual final salary.
At retirement, the value of accumulated funds is used to purchase a retirement
income product (often from an insurance company). Depending on the exact legal,
tax and regulatory environment, some of the accumulated funds may be taken as
cash.
When might it make sense for a retiring employee to take a portion in cash rather
than in a lifetime annuity?
When would it not make sense?
1.3 Movement towards Defined Contribution

DB schemes represent significant risks to the employers. Many prominent companies
face severe difficulties in financing their DB liabilities. The employers are exposed to:
longevity risk
investment risk
As a result, many companies are moving towards DC schemes, where risk is

transferred to the employees.
David Kirk
Page 98 / 107

DC schemes involve little actuarial valuation work. The liabilities are retrospectively
accumulated fund values and do not require discounted cashflow valuation
techniques.
The rest of this chapter deals with the valuation of DB schemes only.
1.4 Spouse benefits

Many pension schemes provide a spouse's pension on death of the employee either
during employment, during retirement or both. The spouse's pension will usually be
smaller than the employees pension.
1. Why might an employer offer a spouse's pension?
2. Why might the spouse's pension be lower than the employee's own pension?
3. Would the spouse's pension be different before and after the year in which the
employee would have reached normal retirement age?
1. The employer might offer the benefit as a way to attract top employees to the
company. Also, if the company did not offer the benefit, and an employee with a
family died in employment (especially shortly before retirement) and not benefit
was paid to the family, the bad publicity might force the company to pay the
benefit in any event. Thus, many companies elect to offer the benefit in the first
place.
2. The employee's pension is intended to provide for his/her income in retirement,
and a proportion for his family. On the death of the employee, he part intended
for his/her own income is no longer required, and only the portion intended for
his/her family is required.
Also, by offering a lower spouse's pension, the cost of the benefit is reduced
which allows a higher salary or employee's pension to be paid at the same cost to
the company.
3. Before NRA, it is more likely that the employee will have a young family with
many financial obligations. After retirement, it is more likely that major
household debts will be paid off. This is just an example, the exact nature of the
benefits would depend on the rules of the pension fund as stipulated by the
employer or country-specific law.
David Kirk
Page 99 / 107

Section 2 : Pre-funding versus cash-based approaches
2 PRE-FUNDING VERSUS CASH-BASED APPROACHES

Pension funds can be funded (paid for) at different points in time. In this section, we
first address two extreme methods (pay as you go and complete full funding from
inception) before describing the more moderate approaches more often adopted in
practice.
2.1 Cash or pay as you go system

Under the cash system, pension benefit payments to retired employees are paid when
due, and no prior reserves are held. Each payment is an expense to the employer
when it is paid.
The risk with this approach to employees is that if the employer can no longer meet
these obligations, they have no security and will likely lose their pension.
Under what conditions might the pension be relatively safe?
Under what conditions might the pension be severely at risk?
A state (government-backed) pension might be relatively safe while the population

is growing rapidly, such that a large base of active workers are contributing towards
the income of pensioners. However, as soon as this population growth slows, the
population pyramid becomes a population column or even an inverted pyramid
and a small number of active workers must support the income of a large number of
pensioners. At this point, the sustainability of the system is at risk.
The pension fund of a company depends similarly on the ratio of active employees
to pensioners. As long as the company continues to grow in terms of number of
employees, the funding requirements are low. However, inevitably this changes and
the pension will become onerous to finance.
Pay-as-you-go systems are usually at risk in the long term.
A cash system is used by some companies in practice. It is more often used by the
governments of countries rather than individual countries.
Is a cash-based system more safe or less safe if the sponsoring employer is a
government?
What is the current problem with the United States Social Security System?
David Kirk
Page 100 / 107

If the government is backing the pension fund, they should be more able to pay the
pensions as they can raise additional funds through higher taxes. However, since it
is still the working population that are paying the taxes, the burden on the active
works can become high if the population growth rate slows down.
The US Social Security System has many problems. One of these is the high growth
in the number of people claiming from the system relative to the growth of those
contributing to the system.
2.2 Complete full funding from inception

This approach is not adopted in practice.
The employer calculates the full present value of pension payments due when an
employee joins the firm. This amount is set aside as a reserve. The full cost of the
entire pension is recognised at the date on which the employee joins the company.
Why is this likely to be unpopular with employers?
Are there any risks left to the employer?
Is this an accurate reflection of the cost of employee benefits?
A very high, upfront cost at the point of hiring an employee will decrease profits
when employees are hired. This acts as a disincentive to hire employees with full
benefits.
There are considerable risks left to the employer. The assets backing the pension fun
are exposed to market fluctuations. A decline in asset values will require additional
funds from the employer. Further, if employees live longer than expected and
require an income after NRA from the company for a longer period than expected.
These cashflow risks also give rise to assumption and valuation risks- at the point at
which future the company recognises changes in expected mortality or investment
returns, the reserves will change without a change in asset values.
No, the cost of employee benefits should be matched with the service provided by
the employees to the employer. Under this method, the entire expected cost is
recognised upfront.
David Kirk
Page 101 / 107

2.3 Practical pre-funding systems

In practice, many pension funds are gradually funded over the period of an
employee's service such that at retirement, the full pension has been reserved for.
Reserve 0 =0
Reserve NRA= Full PV of pension payments
NRA here means normal retirement age. The rate and pattern at which
Reserve 0 grows to Reserve NRA depends on the funding approach chosen.
In the next section we will discuss the projected unit credit method which is one
of the more common approaches.
David Kirk
Page 102 / 107

Section 3 : Projected Unit Credit Method
3 PROJECTED UNIT CREDIT METHOD

The Projected Unit Credit Method is one popular method used to reserve for prefunded pension funds. It is covered here as an example of an appropriate
methodology. Other methods are not covered in this course but involve similar
concepts and techniques.
3.1 Reserving for benefits in payment

3.1.1 Reserving methodology
The reserving methodology for benefits in payment are the same as those used in
valuing an immediate annuity. The expected present value of future cashflows is
calculated, allowing for:
probability of survival
inflationary increases in pension payments
discount rates based on available returns in the market and taking into
consideration the assets backing the pension liabilities.
3.1.2 Economic assumptions required
nominal discount rate and pension payment inflation rate
OR
real discount rate
3.1.3 Demographic assumptions required
base mortality (by gender, and sometimes class of worker)
mortality improvements
3.2 Reserving for active employees

3.2.1 Reserving methodology
The reserving methodology involves two key steps:
1. The first step is to calculate the expected present value of the pension payments
based on projected final salary.
2. The second step is to determine what proportion of this liability has been accrued
based on past service compared with expected total service.
David Kirk
Page 103 / 107

It is step 2 above that defines the Projected Unit Credit Method differently from
alternative pension reserving methodologies.
past service
Reserve = EPV future pension payments
expected total service
More rigorously, we might define this as:
j
Reserve x =v NRAxsalary x1eNRAxp% NRAEA a NRA
xEA
NRAEA
NRAx
Reserve x =salary x
1e
1i
p %a NRA x EA
where v=1i 1
i=nominal discount rate
e=salary inflation
salary x =current salary for employee aged x
1i
1i1g i g
j=
1=
=
1 g
1g
1 g
g=pension increases in retirement
x=current age
EA=age at start of service or employment
j
a NRA=immediate annuity in advance at discount rate j
3.2.2 Economic assumptions required
nominal discount rate and salary inflation rate
AND
nominal discount rate and pension payment increase rate
OR
real discount rate (usually different rates for the period before and after retirement
3.2.3 Demographic assumptions required
base mortality
mortality improvements
David Kirk
Page 104 / 107

3.3 Change in reserve

From t to t+1, the reserve required will change for both retired employees and active
employees.
3.3.1 Change in reserve for pensioners

The change in reserve here operates like an immediate annuity. The factors affecting
the change in reserve are:
unwind of the discount rate
benefit payments made
Survival (0 or 1 for each person) to next period (actual versus expected mortality)
Difference between assumed pension increases and actual pension increases

declared (actual versus expected pension increases)
Changes in mortality basis
Changes in the economic basis
For each of these factors, make sure you understand:

1. Why it impacts the reserve
2. In which direction it impacts the reserve
3. How large or frequent the changes are likely to be
4. How the company might manage the risks of these factors
The unwind of the discount rate is the increase in the reserve as a result of the
passage of time since future cashflows are discounted by less time. This is not a risk
since it will happen over time with certainty.
When benefit payments are made, they are in the past. Since the reserve takes into
account future cashflows, they are no longer part of future cashflows and thus
reduce the reserve. This also happens with certainty and so no risk management is
required.
David Kirk
Page 105 / 107

Each pensioner has a survival probability. However, each pension with either die or
survive. For every pensioner that survives, the reserve will increase (since mortality
was lighter than expected) and for every pensioner that dies the reserve will
decrease (since mortality was heavier than expected). If the total number of
pensioners that survive matches expectations, the overall reserve will not change as
a result since actual deaths are the same as expected. By having a large number of
pensioners, the random fluctuations will average out and the overall variability of
reserves due to actual mortality being different from expected will decrease.
Accurate reserving assumptions will help to ensure that overall mortality
expectations are accurate. Trends (such as annuitant mortality improvements)
should be taken into consideration.
If actual pension increases are higher than assumed, then the reserve will increase
since the expected future cashflows will now be higher than previously expected
since the base (current pension) is higher than previously assumed. This will
generally occur annually when pension increases are decided. The approach
followed to setting the pension increases should be taken into consideration when
determining the pension increase assumption. Another form of risk management is
automatic since a key input into increases declared will be the surplus available in
the pension fund (assets less liabilities). Thus, high pension increases will result in
an increase in pension liabilities (reserves) but this will often correspond with a
previous increase in asset values from good investment performance. Thus,
although liabilities (reserves) will increase, the surplus will not be inappropriately
decreased.
Any change in basis will affect the reserves through a change in expected future
cashflows. The basis will usually be considered annually. However, changes may be
made less frequently. The impact of economic basis changes can be managed
through matching assets and liabilities such that assets and liabilities move in line
with each other. Changes to the mortality basis are more difficult to hedge. Care
should be exercised when setting the basis in the first place, and spurious changes
(unnecessary small changes that might be reversed in future) should not be made.
3.3.2 Change in reserve for active employees

The change in reserve for active employees is an important part of the actuarial
calculation for DB schemes.
The change can be expressed as:
Reserve t = Reserve t 1 Reserve t
David Kirk
Page 106 / 107

The actual change depends on the approach adopted to move from
to
Reserve 0=0
ReserveNRA =PV of pension benefits .
This may be considered along with the Analysis of Profit in AMIII. This will not
form part of AMII.
David Kirk
Page 107 / 107

Actuarial Mathematics Ii: About

Uploaded by

Copyright:

Available Formats

Actuarial Mathematics Ii: About

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Actuarial Mathematics Ii: About

Uploaded by

Copyright:

Available Formats

Actuarial Mathematics II

2007 David Kirk

2007 David Kirk

Chapter 1 Fundamentals of actuarial mathematics

Chapter 1 : Fundamentals of actuarial mathematics

1 BASIC FINANCIAL MATHEMATICS

1.1.2 Definitions of interest rate notation

Table 1.1: Types of Interest rates

Applies for Compounds

NACA = Annual Effective

Chapter 1 : Fundamentals of actuarial mathematics

Applies for Compounds

1.1.3 Continuously compounded rates of interest

Chapter 1 : Fundamentals of actuarial mathematics

t is the time-dependent force of interest applicable at duration t.

1.1.5 Application of force of interest in actuarial mathematics

Where x t is the instantaneous annualised probability of death at age x+t. We

Chapter 1 : Fundamentals of actuarial mathematics

1.1.6 Interest, discounts and discount factors

USD100 USD1001d =USD100[ 1 1d ]=USD100d

1.2 Actuarial notation for financial mathematics

1.2.1 Annuity certain payable in arrears

Chapter 1 : Fundamentals of actuarial mathematics

This is a geometric sum with geometric increases of

Check that you can calculate the following:

a 10 = 6.14 for i = 10%

1.2.2 Annuity certain payable in advance

Chapter 1 : Fundamentals of actuarial mathematics

Check that you can calculate the following:

a 10 = 6.76 for i = 10%

Now show that

Explain in words what this relationship means.

1.2.3 Accumulation factors

Prove the following:

Chapter 1 : Fundamentals of actuarial mathematics

Chapter 1 : Fundamentals of actuarial mathematics

2 INTRODUCTION TO CONCEPTS OF LIFE CONTINGENCIES

If the t subscript is omitted, it is assumed to be 1. Thus,

Mortality in each year is independent. Thus, the following equation holds:

In terms of standard life table notation,

in the population aged x .

2.2 Death probabilities

Chapter 1 : Fundamentals of actuarial mathematics

2.2.1 Force of mortality

Where t is the force of mortality at time t .

q =1e and so x =ln 11q x which is also x =ln 1 p x

So we have assumed a constant force of mortality within an age bracket in order to

Prove the above equation.

Chapter 1 : Fundamentals of actuarial mathematics

q 45 assuming a constant force of mortality for lives aged 45. Show

you result to at least 5 decimal places.

Is this the same as

If they are different, explain why this is.

We have used the assumption of a constant force of mortality in deriving these

q x =t q x for any integer x and 0 < t < 1

2.3 Deterministic modelling

Chapter 1 : Fundamentals of actuarial mathematics

E [ CF t ]= pmf CF xx=1t p x 01t p x =t p x

pmf CF xx=v t [ 1t p x01 t px ] =v nt p x