Introduction

Future

lifetime random variable of an individual

Force of mortality
Actuarial notations
Curtate future lifetime random variable

FINA3220A
Actuarial Models I
Chapter 1
Survival Models

FINA3220A (2015-2016)

The Future Lifetime Random Variable

Chapter 1 Survival Models I

Page 2

The Future Lifetime Random Variable

The

future lifetime or time-until-death random variable, Tx or T(x),

is the basic building block in life insurance and annuity
Used as a model of human mortality in order to estimate the
time at which a death benefit is payable
Another important distribution is the distribution of the age-atdeath random variable T0 or X

life aged x is denoted by (x) where x 0

The death of (x) can occur at any age greater than x
Let Tx be the continuous future lifetime random variable of (x)
The age-at-death for (x) is x + Tx, which is also a random
variable
The distribution function of Tx is given by
A

It

is the probability that (x) does not survive beyond age x + t

It is called the lifetime distribution from age x
The survival function for (x) is given by
It

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 3

is the probability that (x) survives for at least t years

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 4

The Future Lifetime Random Variable

The Future Lifetime Random Variable

Consider

the relationship between T0 and Tx

At birth, the individuals age at death would have been
represented by T0
Given that the individual has survived to age x
The future lifetime is represented by Tx and the age at death is
now x + Tx
If the individual dies within t years from now, then we have

In

probability term, we have

Using

basic probability result,

Assumption:

event that Tx t is equivalent to the event that T0 x + t

given that T0 > x
That is,
The

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 5

FINA3220A (2015-2016)

The Future Lifetime Random Variable

With

It

Chapter 1 Survival Models I

Page 6

Exercise 1.1

the same assumption, we have

Show

that the survival probability for (x), for t + u years is the

product of
The probability of surviving the first t years
The probability of surviving for u years, given survival to age
x+t

can be written as

The

survival probability from birth to age x + t is the product of

The probability of survival to age x from birth
The probability, having survived to age x of further surviving
to age x + t

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 7

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 8

The Future Lifetime Random Variable

The Future Lifetime Random Variable

Survival

function of a future lifetime random variable must satisfy

the following (necessary and sufficient) conditions
Sx(0) = 1
limt Sx(t) = 0
The survival function must be a non-increasing function of t
Three additional assumptions for future lifetime random variable
for human life
Sx(t) is differentiable for all t > 0
limt t Sx(t) = 0
limt t2 Sx(t) = 0
The last two assumptions imply finite mean and variance

FINA3220A (2015-2016)

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 9

limiting age may be imposed in a lifetime distribution

That is, survival beyond the limiting age is not possible
We use the Greek letter to denote limiting age

Exercise 1.2
You

Chapter 1 Survival Models I

Page 10

Force of Mortality

are given the following lifetime distribution

The

force of mortality is an important and fundamental concept in

modelling future lifetime
Recall that P(x < T0 z | T0 > x) is the conditional probability that
(0) will die between ages x and z, given survival to x
With z x held constant at c, then considered as a function of x
P(x < T0 x + c | T0 > x) describes the distribution of the
probability of death in the near future (between 0 and c) for a
life attained age x
Analogy for instantaneous death is obtained by using the density
of probability of death at attained age x, that is

Calculate
The

probability that a newborn life dies before age 60

The probability that a life aged 30 survives to at least age 70
The probability that a life aged 20 dies between ages 90 and
100

Here, fX(x)

is the p.d.f. of the continuous age-to-death random

variable
FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 11

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 12

Force of Mortality
We

Force of Mortality

define the force of mortality, x or (x), as

The

force of mortality can be used to specify the distribution of T0

Rearranging the equation for x, we have

conditional probability density interpretation:

For each age x, it gives the value of the conditional p.d.f. of
T0 at exact age x, given survival to that age
However, x itself is not a probability but x x is
Note that x 0 as both f0(x) 0 and 1 F0(x) 0
x is also called the failure rate or hazard rate or hazard rate
function

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 13

Integrating

Taking

More

this expression from 0 to x, we have

exponentials, we have

relationships:

FINA3220A (2015-2016)

Force of Mortality

Chapter 1 Survival Models I

Page 14

Exercise 1.3

Similarly,

we can use the lifetime distribution of Tx to find the

force of mortality function

For

the lifetime distribution in Exercise 1.2

Calculate the force of mortality at age 50
Calculate the median future lifetime at age 50

And

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 15

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 16

Exercise 1.4

Some Analytical Laws of Mortality

x = 2(100 x)1 for 0 x < 100 and let x = 40

Find fx(t) and Fx(t)
Suppose

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Three

principal justifications for postulating an analytic form for

mortality or survival functions:
Philosophical:
Some authors suggested that human survival is governed by
an simple law such as those in physics
Practical:
Easier to communicate a function with a few parameters
Some of the analytic forms have elegant properties that are
convenient in evaluating probability statements that involve
more than one life
Estimation:
Easier to estimate a few parameters for a simple analytic
survival function

Page 17

FINA3220A (2015-2016)

Some Analytical Laws of Mortality

for simple analytic survival functions has declined
recently
Some authors feel that the belief in universal laws of mortality
is nave
With the increasing speed and storage capacity of computers,
the advantages of some analytic forms in computations
involving more than one life are no longer of great importance
Several families of analytic survival functions:

Gompertz
Makeham

x
( x)1
Bcx
A + Bcx

Weibull

kxn

De Movire

FINA3220A (2015-2016)

Page 18

Some Analytical Laws of Mortality

Support

Originator

Chapter 1 Survival Models I

Note:
Gompertzs

law is a special case of Makehams law with A = 0

If c = 1 in Gompertzs and Makehams laws, the exponential
distribution results
In connection with Makehams law
The constant A has been interpreted as capturing the
accident hazard
The term Bcx as capturing the hazard of aging

Restrictions
0x<
B > 0, c > 1, x 0
B > 0, A B, c > 1, x 0
k > 0, n > 0, x 0
Chapter 1 Survival Models I

Page 19

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 20

Exercise 1.5
For

Actuarial Notation

the Gompertz law of mortality, show that

Find

standard set of actuarial symbols are used to promote research

and communication
The standard is called the International Actuarial Notation
(IAN)
These symbols differ from those used for probability notation
The probability that (x) survives to at least age x + t is denoted by

the probability density function for Tx

The

probability that (x) dies before age x + t is denoted by

If

t = 1, convention permits us to omit the prefix in the symbols, we

have

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 21

qx

is called the mortality probability or rate at age x

FINA3220A (2015-2016)

Actuarial Notation

Chapter 1 Survival Models I

Page 22

Actuarial Notation

The

probability that (x) survives t years, and then dies in the

subsequent u years, is denoted by

Restating

the previous results in actuarial notation

Some

authors call it the deferred mortality probability

As before, if u = 1, the prefix is deleted in
, and we have

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 23

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 24

Actuarial Notation
Consider

Exercise 1.6

the mortality probability at age x

Given px =
Calculate

An

approximating formula can be obtained as follows

When qx is small, px is close to 1, and hence spx is close to 1
Thus

0.99, px+1 = 0.985, 3px+1 = 0.95 and qx+3 = 0.02

px+3
2 px
2px+1
3 px

The

second relationship follows by the mid-point rule for

numerical integration

FINA3220A (2015-2016)

Chapter 1 Survival Models I

1|2qx

Page 25

Mean and Standard Deviation of Future Lifetime

Complete

expectation of life:
The expected value of Tx, denoted by

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 26

Mean and Standard Deviation of Future Lifetime

Second

moment of Tx:

, is

The

variance of T(x):

Here,

we use the integration by parts technique

The existence of E[T(x)] implies that
The

complete-expectation-of-life at various ages is often used to

compare levels of public health among different populations

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 27

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 28

Exercise 1.7

Exercise 1.8

For

the lifetime distribution in Exercise 1.2, calculate the complete

expectation of life and var(Tx) at age 50

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 29

Suppose

the mortality follows Gompertz law with B = 0.0003 and

c = 1.07
Calculate, using numerical integration, the complete expectation of
life and standard deviation of Tx at ages 0, 50 and 100

FINA3220A (2015-2016)

Mean and Standard Deviation of Future Lifetime

Term

expectation of life:
Sometimes, we are interested in the mixed random variable
min(Tx, n), i.e. partly continuous and partly discrete
For example, a certain benefit is payable continuously for a
maximum of n years, conditional on survival
The expected value of min(Tx, n) is denoted by
, called term
expectation of life

Chapter 1 Survival Models I

Page 30

Curtate Future Lifetime

The

curtate future lifetime of (x), is denoted by Kx

A discrete random variable associated with the future lifetime
The number of future years completed by (x) prior to death
Note that Kx is the greatest integer in Tx, i.e.
The probability function, p.f., is given by

Note

that under the assumption that Tx is a continuous-type

random variable

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 31

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 32

Curtate Future Lifetime

Curtate Future Lifetime

Curtate

expectation of life:
The expected value of Kx is denoted by ex

Variance

Term

curtate expectation of life

The expected value of the discrete random variable min(Kx, n)
The random variable is applicable in the case that (x) is
entitled to payments of 1 at times 1, 2, 3, , n, conditional on
survival
It can be known that

of Kx, by summation by parts:

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 33

FINA3220A (2015-2016)

Complete and Curtate Expectations of Life

Page 34

Exercise 1.9

The

relationship between
can be obtained based on
different assumptions or approximation
Here, we look at an approximation based on numerical integration
First, note that

We

Chapter 1 Survival Models I

Based

on the assumption in Exercise 1.8, verify the approximation

formula

then approximate the integrals by trapezium rule

Hence

we have

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 35

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 36

Exercise 1.10
F0(t) = 1 et, where > 0
Find Sx(t)
Find x
Show that ex = (e 1)1
What conclusions do you draw about using this lifetime
distribution to model human mortality?

Let

FINA3220A (2015-2016)

Chapter 1 Survival Models I

Page 37