2

Survival models

2.1 Summary
In this chapter we represent the future lifetime of an individual as a random
variable, and show how probabilities of death or survival can be calculated
under this framework. We then define an important quantity known as the force
of mortality, introduce some actuarial notation, and discuss some properties of
the distribution of future lifetime. We introduce the curtate future lifetime ran-
dom variable. This is a function of the future lifetime random variable which
represents the number of complete years of future life. We explain why this
function is useful and derive its probability function.

2.2 The future lifetime random variable

In Chapter 1 we saw that many insurance policies provide a benefit on the
death of the policy holder. When an insurance company issues such a policy, the
policyholder's date of death is unknown, so the insurer does not know exactly
when the death benefit will be payable. In order to estimate the time at which
a death benefit is payable, the insurer needs a model of human mortality, from
which probabilities of death at particular ages can be calculated, and this is the
topic of this chapter.
We start with some notation. Let (x) denote a life aged x, where x '.'.".: 0. The
death of (x) can occur at any age greater than x, and we model the future
lifetime of (x) by a continuous random variable which we denote by Tx. This
means that x + Tx represents the age-at-death random variable for (x).
Let Fx be the distribution function of Tx, so that

Fx(t) = Pr[Tx :S t].

Then Fx (t) represents the probability that (x) does not survive beyond age
x + t, and we refer to Fx as the lifetime distribution from age x. In many life

17
18 Survival models

insurance problems we are interested in the probability of survival rather than

death, and so we define Sx as

Sx(t) = 1- Fx(t) = Pr[Tx > t].

Thus, Sx (t) represents the probability that (x) survives for at least t years, and
Sx is known as the survival function.
Given our interpretation of the collection of random variables {Tx} x::: o as
the future lifetimes of individuals, we need a connection between any pair of
them. To see this, consider To and Tx for an individual who is now aged x. The
random variable To represented the future lifetime at birth for this individual,
so that, at birth, the individual's age at death would have been represented by
To. This individual could have died before reaching age x - the probability of
this was Pr[ To < x] - but has survived. Now that the individual has survived
to age x, so that we know that To > x, her future lifetime is represented by Tx
and her age at death is now x + Tx. If she dies within t years from now, then
Tx:::; t and To:::; x + t. Loosely speaking, we require the events [Tx:::; t] and
[To:::; x + t] to be equivalent, given that the individual survives to age x. We
achieve this by making the following assumption for all x :::: 0 and for all t > 0

j Pr[Tx :::; t] = Pr[To:::; x +ti To > x].1 (2.1)

This is an important relationship.

Now, recall from probability theory that for two events A and B
Pr[A and B]
Pr[AIB] = Pr[B]
,
so, interpreting [To :::; x + t] as event A, and [To > x] as event B, we can rear-
range the right-hand side of (2.1) to give
Pr[x < To :::; x + t]
Pr[Tx :::; t] = Pr[To > x] ,

that is,
Fo(x + t) - Fo(x)
(2.2)
Fx(t) = .
So(x)
Also, using Sx(t) = 1- Fx(t),

S (t) = So(x + t) (2.3)

x So(x) '

which can be written as

ISo(x + t) = So(x) Sx(t). I (2.4)

 2. 2 The future lifetime random variable 19

This is a very important result. It shows that we can interpret the probability
of survival from birth to age x + t as the product of

(1) the probability of survival to age x from birth, and

(2) the probability, having survived to age x, of further surviving to age x + t.
Note that Sx (t) can be thought of as the probability that (0) survives to at least
age x + t given that (0) survives to age x, so this result can be derived from the
standard probability relationship

Pr[A and B] = Pr[AIB]Pr[B]

where the events here are A= [To > x + t] and B =[To> x], so that
Pr[AjB] = Pr[To > x + tJTo > x],

which we know from (2.1) is equal to Pr[Tx > t].

Similarly, any survival probability for (x), for, say, t + u years can be split
into the probability of surviving the first t years, and then, given survival to age
x + t, subsequently surviving another u years. That is,

So(x + t + u)
Sx(t + u) = So(x)
So(x + t) So(x + t + u)
::::} Sx(t + u) = --------
So(x) So(x+t)
::::} Sx(t + u) = Sx(t)Sx+1(u). (2.5)

We have already seen that if we know survival probabilities from birth, then,
using formula (2.4), we also know survival probabilities for our individual from
any future age x. Formula (2.5) takes this a stage further. It shows that if we
know survival probabilities from any age x ( ::=: 0), then we also know survival
probabilities from any future age x + t ( ::=: x).
Any survival function for a lifetime distribution must satisfy the following
conditions to be valid.

Condition 1 Sx(O) = 1; that is, the probability that a life currently aged x
survives 0 years is 1.

Condition 2 limt--+oo Sx (t) = O; that is, all lives eventually die.

Condition 3 The survival function must be a non-increasing function oft; it

cannot be more likely that (x) survives, say 10.5 years than 10 years, because
in order to survive 10.5 years, (x) must first survive 10 years.

These conditions are both necessary and sufficient, so that any function Sx
which satisfies these three conditions as a function of t ( ::=: 0), for a fixed
20 Survival models

x ( :=::: 0), defines a lifetime distribution from age x, and, using formula (2.5),
for all ages greater than x.
For all the distributions used in this book, we make three additional assump-
tions:
Assumption 2.1 Sx(t) is differentiable for all t > 0. Note that together with
fr
Condition 3 above, this means that Sx (t) ::::; 0 for all t > 0.

Assumption 2.2 limr--+oo t Sx (t) = 0.

Assumption 2.3 limt---+oo t 2 Sx (t) = 0.
These last two assumptions ensure that the mean and variance of the distri-
bution of Tx exist. These are not particularly restrictive constraints - we do not
need to worry about distributions with infinite mean or variance in the context
of individuals' future lifetimes. These three extra assumptions are valid for all
distributions that are feasible for human lifetime modelling.

Example 2.1 Let

1 - (1 - t/120) 116 for 0 ::::; t ::::; 120
Fo(t) =
{1 fort > 120

Calculate the probability that

(a) a newborn life survives beyond age 30,
(b) a life aged 30 dies before age 50, and
(c) a life aged 40 survives beyond age 65.
Solution 2.1 (a) The required probability is
16
S0(30) = 1 - F 0(30) = (1 - 30/120) 1 = 0.9532.

(b) From formula (2.2), the required probability is

F ( 0) = F 0(50) - Fo(30) = 0.04 10.
30 2 1 - Fo(30)
(c) From formula (2.3), the required probability is
65
S40(25) = So( ) = 0.9395.
So(40) 0

We remark that in the above example, So(120) = 0, which means that under
this model, survival beyond age 120 is not possible. In this case we refer to 120
as the limiting age of the model. In general, if there is a limiting age, we use
the Greek letter w to denote it. In models where there is no limiting age, it is
often practical to introduce a limiting age in calculations, as we will see later
in this chapter.
 2.3 The force of mortality 21

2.3 The force of mortality

The force of mortality is an important and fundamental concept in modelling
future lifetime. We denote the force of mortality at age x by /J.,x and define it as
1
/J.,x = lim -Pr[To :::; x + dx I To > x]. (2.6)
dx--+O+ dx

From equation (2.1) we see that an equivalent way of defining /J.,x is

. 1
/J.,x = hm -Pr[Tx :::; dx],
dx--+O+ dx

which can be written in terms of the survival function Sx as

. 1 (2.7)
/J.,x = hm - (1 - Sx(dx)).
dx--+O+dx
Note that the force of mortality depends, numerically, on the unit of time; if we
are measuring time in years, then /J.,x is measured per year.
The force of mortality is best understood by noting that for very small dx,
formula (2.6) gives the approximation

/J.,x dx ~ Pr[To :::; x + dx I To > x]. (2.8)

Thus, for very small dx, we can interpret /J.,x dx as the probability that a life
who has attained age x dies before attaining age x + dx. For example, sup-
pose we have a life aged exactly 50, and that the force of mortality at age
50 is 0.0044 per year. A small value of dx might be a single day, or 0.00274
years. Then the approximate probability that the life dies on his 50th birthday
is 0.0044 X 0.00274 = 1.2 X 10- 5 •
We can relate the force of mortality to the survival function from birth,
So. As
S (dx) = So(x + dx)
x So(x) '

formula (2.7) gives

1 . So(x) - So(x + dx)
/J.,x = - - lim
So(x) dx--+O+ dx
1
=- (-!:__So(x)).
So(x) dx
Thus,

(2.9)
22 Survival models

From standard results in probability theory, we know that the probability

density function for the random variable r., which we denote f,, is related
to the distribution function F,, and the survival function Sx by
d d
fx (t) = dt Fx (t) = - dt Sx (t).
So, it follows from equation (2.9) that

fo(x)
f.Lx = So(x) ·

We can also relate the force of mortality function at any age x + t, t > 0,
to the lifetime distribution of Tx. Assume x is fixed and t is variable. Then
d(x + t) =dt and so
1 d
f.Lx+t = - So(x + t) d(x + t) So(x + t)
1 d
=- -So(x+t)
So(x + t) dt
1 d
= - So(x + t) dt (So(x)Sx(t))
So(x) d
= - So(x + t) dt Sx(t)
-1 d
= ---Sx(t).
Sx (t) dt
Hence
fx(t)
f.Lx+t = Sx (t)' (2.10)

This relationship gives a way of finding f.Lx+t given Sx(t). We can also use
equation (2.9) to develop a formula for Sx (t) in terms of the force of mortality
function. We use the fact that for a function h whose derivative exists,
d 1 d
-logh(x) = - - h ( x )
dx h(x) dx '
so from equation (2.9) we have
d
f.Lx = - - log So(x),
dx
and integrating this identity over (0, y) yields

lay f.Lxdx = - (log So(y) - log So(O)).

 2.3 The force of mortality 23

As log So (0) = log Pr[To > O] = log 1 = 0, we obtain

So(Y) =exp {-lay /,lxdx} ,

from which it follows that

Sx (t) = So(x + t) =exp { - 1x+t µ,,.dr } = exp { - lot /,lx+sds } .

So(x) x o
(2.11)
This means that if we know /,lx for all x :=:: 0, then we can calculate all the sur-
vival probabilities Sx (t), for any x and t. In other words, the force of mortality
function fully describes the lifetime distribution, just as the function So does.
In fact, it is often more convenient to describe the lifetime distribution using
the force of mortality function than the smvival function.

Example 2.2 As in Example 2.1, let

Fo(x) = 1 - (1 - x/120) 116

for 0 ~x ~ 120. Derive an expression for /,lx·

Solution 2.2 As So(x) = (1 - x/120) 116 , it follows that

:xSo(x) = tO -x/120)- 1 (-iio),

5 6

and so
-1 d 1 1 1
/,lx = ---So(x) = 720 (1-x/120)- = - - -
So(x) dx 720 - 6x
As an alternative, we could use the relationship

/,lx =-~log So(x) = -~ (~ log(l - x/120)) = _ _ _l _ _

dx dx 6 720(1-x/120)
1
=---
720- 6x
D

Example 2.3 Let /,lx =Bex, x > 0, where B and c are constants such that
0 < B < 1andc>1. This model is called Gompertz' law of mortality. Derive
an expression for Sx (t).

Solution 2.3 From equation (2.11 ),

Sx (t) =exp {- lx+t Ber dr} .

 24 Survival models

Writing er as exp{r log c},

1~+;,. dr = B lxe-+:.p{r log c }dr

B lx+t
= - - exp{r log c}
loge x

B
= __ (cx+t _ex),
loge
giving

Sx(t) =exp { -B cx(ct -

loge
1)}.
D

The force of mortality under Gompertz' law increases exponentially with age.
At first sight this seems reasonable, but as we will see in the next chapter, the
force of mortality for most populations is not an increasing function of age
over the entire age range. Nevertheless, the Gompertz model does provide a
fairly good fit to mortality data over some age ranges, particularly from middle
age to early old age.

Example 2.4 Calculate the survival function and probability density function
for Tx using Gompertz' law of mortality, with B = 0.0003 and c = 1.07, for
x = 20, x = 50 and x = 80. Plot the results and comment on the features of the
graphs.

Solution 2.4 For x = 20, the force of mortality is µ 20+1 = Bc20+t and the
survival function is

S20(t) =exp { --B

- c 20 (c t - 1) } .
loge
The probability density function is found from (2.10):

ho(t)
f-l20+1 = - - =} ho(t) = f-l20+1 S20(t) = Be 20+1 exp { --B
- c 20 (c t - 1) } .
~~) ~c
Figure 2.1 shows the survival functions for ages 20, 50 and 80, and Figure 2.2
shows the corresponding probability density functions. These figures illustrate
some general points about lifetime distributions.
First, we see an effective limiting age, even though, in principle, there is no
age to which the survival probability is exactly zero. Looking at Figure 2.1,
we see that although Sx(t) > 0 for all combinations of x and t, survival beyond
age 120 is very unlikely.
-
2.3 The force of mortality 25

0.9

0.8

0.7
~
.g 0.6
'8
k
\'.>< 0.5
';rj
.::: 0.4
~
r/J
0.3

0.2

0.1

0
0 10 20 30 40 50 60 70 80 90 100
Time, t

Figure 2.1 Sx (t) for x = 20 (bold), 50 (solid) and 80 (dotted).

O.D7

0.06

0.05

0.04
~

~
O.D3

0.02

0.01

0
0 10 20 30 40 50 60 70 80 90 100
Time, t

Figure 2.2 fx (t) for x = 20 (bold), 50 (solid) and 80 (dotted).

Second, we note that the survival functions are ordered according to age,
with the probability of survival for any given value of t being highest for age
20 and lowest for age 80. For survival functions that give a more realistic rep-
resentation of human mortality, this ordering can be violated, but it usually
26 Survival models

holds at ages of interest to insurers. An example of the violation of this order-

ing is that So(l) may be smaller than Sx(l) for x ~ 1, as a result of perinatal
mortality.
Looking at Figure 2.2, we see that the densities for ages 20 and 50 have
similar shapes, but the density for age 80 has a quite different shape. For ages
20 and 50, the densities have their respective maximums at (approximately)
t = 60 and t = 30, indicating that death is most likely to occur around age 80.
The decreasing form of the density for age 80 also indicates that death is more
likely to occur at age 80 than at any other age for a life now aged 80. A fur-
ther point to note about these density functions is that although each density
function is defined on (0, oo), the spread of values of fx(t) is much greater for
x = 20 than for x = 50, which, as we will see in Table 2.1, results in a greater
variance of future lifetime for x = 20 than for x = 50. D

2.4 Actuarial notation

The notation used in the previous sections, Sx(t), F,,(t) and fx(t), is standard
in statistics. Actuarial science has developed its own notation, International
Actuarial Notation, that encapsulates the probabilities and functions of great-
est interest and usefulness to actuaries. The force of mortality notation, /Lx,
comes from International Actuarial Notation. We summarize the relevant actu-
arial notation in this section, and rewrite the important results developed so
far in this chapter in terms of actuarial functions. The actuarial notation for
survival and mortality probabilities is

It Px = Pr[Tx > t] = Sx(t), I (2.12)

(2.13)

/ ultqx = Pr[u < Tx :Su+ t] = Sx(U) - Sx(u + t). j (2.14)

That is

t Px is the probability that (x) survives to at least age x + t,

tqx is the probability that (x) dies before age x + t,
u ltqx is the probability that (x) survives u years, and then dies in the sub-
sequent t years, that is, between ages x + u and x + u + t.

We may drop the subscript t if its value is 1, so that Px represents the proba-
bility that (x) survives to at least age x + 1. Similarly, qx is the probability that
(x) dies before age x + 1. In actuarial terminology qx is called the mortality
rate at age x. We call ultqx a deferred mortality probability, because it is
 2.4 Actuarial notation 27

the probability that death occurs in the interval oft years, following a deferred
period of u years.
The relationships below follow immediately from the definitions above and
the previous results in this chapter:

tPx +tqx = 1,
ultqx = uPx - u+t Px•
t+uPx = t Px uPx+t from (2.5), (2.15)
1 d
/Lx = - - -d xPO from (2.9). (2.16)
xPO X

Similarly,

1 d d
/Lx+t = -t Px
- -dt tPx::::} -d tPx = -tPx /Lx+t• (2.17)
t
fx(t) (2.18)
/Lx+t = Sx (t) ::::} fx (t) = t Px /Lx+t from (2.10),

tPx =exp {-lat fLx+sds} from (2.11). (2.19)

As Fx is a distribution function and f x is its density function, it follows that

Fx(t) =lat fx(s)ds,

which can be written in actuarial notation as

(2.20)

This is an important formula, which can be interpreted as follows. Consider

time s, where 0 ~ s < t. The probability that (x) is alive at time s is s Px,
and the probability that (x) dies between ages x + s and x + s + ds, having
survived to age x + s, is (loosely) /Lx+sds, provided that ds is very small. Thus
s Px fLx+sds can be interpreted as the probability that (x) dies between ages
x + s and x + s + ds. Now, we can sum over all the possible death intervals s
to s + ds - which requires integrating because these are infinitesimal intervals
- to obtain the probability of death before age x + t.
We can illustrate this event sequence using the time-line diagram shown in
Figure 2.3.
This type of interpretation is important as it can be applied to more compli-
cated situations, and we will employ the time-line again in later chapters.
 28 Survival models
Time 0 s s+ds t

Age
I
x I
x+s
I
x+s+ds I
x+t
Event (x) survives s years (x)
\_
~
Probability sPx µx +sds

Figure 2.3 Time-line diagram for 1qx.

In the special case when t = 1, formula (2.20) becomes

1
qx = fo sPx µx+sds.

When qx is small, it follows that Px is close to 1, and hence s Px is close to 1

for 0 =:: s < 1. Thus

~ fo ~ µx+l/2,
1
qx µx+sds

where the second relationship follows by the mid-point rule for numerical inte-
gration.

Example 2.5 As in Examples 2.1 and 2.2, let

Fo(x) = 1 - (1 - x/120) 116

for 0 :": x :": 120. Calculate both qx and µx+l/2 for x = 20 and for x = 110,
and comment on these values.

Solution 2.5 We have

So(x + 1) ( 1 ) l/ 6
1
Px = So(x) = - 120 - x '

giving q20 = 0.00167 and quo = 0.01741, and from the solution to Exam-
ple 2.2, µ 20 1=0.00168 and µ 110 1 = 0.01754. We see that µx+l/2 is a good
approximatibn to qx when the mohality rate is small, but is not such a good
approximation, at least in absolute terms, when the mortality rate is not
close to 0. 0

2.5 Mean and standard deviation of Tx

Next, we consider the expected future lifetime of (x), E[Tx], denoted in actu-
arial notation by ~ x. We call this the complete expectation of life. In order to
 2.5 Mean and standard deviation ofTx 29

0
evaluate ex, we note from formulae (2.17) and (2.18) that
d (2.21)
fx(t) = tPx /J.,x+t = - dt tPx·
From the definition of an expected value, we have

~x = LXJ t fx(t)dt
= loot t Px /J.,x+1dt.

We can now use (2.21) to evaluate this integral using integration by parts as

~x = - loot (:t tPx) dt

-(t1Px\~ -
00

= l tPxdt).

In Section 2.2 we stated the assumption that lim1--+oo t 1 Px = 0, which gives

(2.22)

Similarly, for E[T}J, we have

E[T;] =loo t 2 1Px /J.,x+tdt

00
2
=-l t (:ttPx)dt

= -(t 2
tPx\: - loo tPx 2t dt)

(2.23)
= 2 loo t1Px dt.
So we have integral expressions for E[Tx] and E[T}J. For some lifetime distri-
butions we are able to integrate directly. In other cases we have to use numer-
ical integration techniques to evaluate the integrals in (2.22) and (2.23). The
variance of Tx can then be calculated as

Example 2.6 As in Example 2.1, let

1 6
Fo(x) = 1 - (1 - x/120) 1

for 0 :S x :S 120. Calculate ~x and V[Tx] for (a) x = 30 and (b) x = 80.
 30 Survival models

Solution 2.6 As So(x) = (1 - x/120) 116 , we have

16
So(x+t) ( t ) 1
tPx = So(x) = l - 120-x

Now recall that this formula is valid for 0 :S t :S (120 - x), because, under this
model, survival beyond age 120 is impossible. Technically, we have

tPx --I 0
(1 - 12 6-x )
116
for x
for x
+t
+t
::::: 120,
> 120.
So the upper limit of integration in equation (2.22) is 120 - x, and

0
1120-x ( t ) 1/6
ex = 1- dt.
0 120 - x
We make the substitution y = 1 - t/(120 - x), so that t = (120 - x)(l - y),
giving

1
~x = (120 - x) fo y 116dy

= ~(120 - x).
0 0
Then e3o = 77 .143 and eso = 34.286.
Under this model the expectation of life at any age x is 6/7 of the time to
age 120.
For the variance we require E[T}]. Using equation (2.23) we have
{120-x
E [r;] = 2 lo ftPxdt

= 2 {120-x t
lo
(1- t
120 - x
)l/6 dt.

Again, we substitute y = 1 - t/(120 - x) giving

E [ r}] = 2(120 - x)2 la1 (yl/6 - y7/6) dy

= 2(120 - x ) (
2
~- 6
13) .

Then

V[Tx] = E[T}] - (~x

2
r = (120 - x) 2 ( 2(6/7 - 6/13) - (6/7) 2)

= (120 - x) (0.056515) = ((120 - x) (0.23773)) 2 .

So V[T30] = 21.396 2 and V[T80 ] = 9.509 2 .

 2.5 Mean and standard deviation ofTx 31

Table 2.1 Values of ~x' SD[Tx] and expected age at death

for the Gompertz model with B = 0.0003 and c = 1.07.
0 0
x ex SD[Tx] x +ex
0 71.938 18.074 71.938
10 62.223 17.579 72.223
20 52.703 16.857 72.703
30 43.492 15.841 73.492
40 34.752 14.477 74.752
50 26.691 12.746 76.691
60 19.550 10.693 79.550
70 13.555 8.449 83.555
80 8.848 6.224 88.848
90 5.433 4.246 95.433
100 3.152 2.682 103.152

Since we know under this model that all lives will die before age 120, it
makes sense that the uncertainty in the future lifetime should be greater for
younger lives than for older lives. D

A feature of the model used in Example 2.6 is that we can obtain formulae for
quantities of interest such as ~x. but for many models this is not possible. For
example, when we model mortality using Gompertz' law, there is no explicit
formula for ~x and we must use numerical integration to calculate moments of
Tx. In Appendix B we describe in detail how to do this.
Table 2.1 shows values of ~x and the standard deviation of Tx (denoted
SD[TxD for a range of values of x using Gompertz' law, f-Lx = BcX, where
B = 0.0003 and c = 1.07. For this survival model, l30PO = 1.9 x 10- 13 , so that
using 130 as the maximum attainable age in our numerical integration is accu-
rate enough for practical purposes.
We see that ~x is a decreasing function of x, as it was in Example 2.6. In
that example ~x was a linear function of x, but we see that this is not true in
Table 2.1.
We are sometimes interested in the future lifetime random variable subject
to a cap of n years, which is represented by the random variable min(Tx, n).
For example, suppose that (x) is entitled to a benefit payable continuously
for a maximum of n years, conditional on survival. Then min(Tx, n) would
represent the payment period for the benefit. We derive the mean and variance
of this random variable, using a similar approach to the derivation of the mean
----------------..................
32 Survival models

and variance of Tx. The expected value of min(Tx, n) is denoted ~x:ii1' and is
called the term expectation of life.

E[min(Tx, n)] = ~x:/il =Ion t tPx µx+t dt + £ 00

n tPx µx+t dt

=font (-:ttPx)dt+nnPx

= -(ttPxl~ -lantPxdt) +nnPx

=:} ~x:/il = lantPx dt.

The iil notation is used to denote a period of n years Gust as in annuity-certain

notation), and is used extensively in later chapters.

2.6 Curtate future lifetime

2.6.1 Kx and ex
In many insurance applications we are interested not only in the future lifetime
of an individual, but also in what is known as the individual's curtate future
lifetime. The curtate future lifetime random variable is defined as the integer
part of future lifetime, and is denoted by Kx for a life aged x. If we let L j
denote the floor function, we have

We can think of the curtate future lifetime as the number of whole years lived
in the future by an individual. As an illustration of the importance of curtate
future lifetime, consider the situation where a life aged x at time 0 is entitled
to payments of 1 at times 1, 2, 3, ... provided that (x) is alive at these times.
Then the number of payments made equals the number of complete years lived
after time 0 by (x). This is the curtate future lifetime.
We can find the probability function of Kx by noting that fork= 0, 1, 2, ... ,
K x = k if and only if (x) dies between the ages of x + k and x + k + 1. Thus
fork= 0, 1, 2, ...

Pr[Kx = k] = Pr[k :S Tx < k + 1]

= klqx
= kPx - k+IPx
= kPx - kPx Px+k
= kPx qx+k·
- 2.6 Curtate future lifetime 33

The expected value of Kx is denoted by ex, so that ex= E[Kx], and is referred
to as the curtate expectation of life (even though it represents the expected
curtate lifetime). So

E[Kx] =ex
00

= L)Pr[Kx = k]
k=O
00

=I> (kPx - k+lPx)

k=O
= (1Px - 2Px) + 2(2Px - 3Px) + 3(3Px - 4Px) + · ··
00

= L kPx· (2.24)
k=l

Note that the lower limit of summation is k = 1.

Similarly,

00

E[K;] = Lk 2 ( kPx - k+lPx)

k=O
= (1Px - 2Px) + 4(2Px - 3Px) + 9(3Px - 4Px) + l6(4Px - 5Px) + · ··
00 00

= 2 L k kPx - L kPx
k=l k=l
00

= 2 Lk kPx - ex.
k=l

As with the complete expectation of life, there are a few lifetime distributions
that allow E[Kx] and E[K~] to be calculated analytically. For more realistic
models, such as Gompertz', we can calculate the values easily using Excel or
other suitable software. Although in principle we have to evaluate an infinite
sum, at some age the survival probability will be sufficiently small that we can
treat it as an effective limiting age.
Analogous to the random variable min(Tx, n) we have the random variable
min(Kx, n ). For example, if a life aged x is entitled to payments of 1 at times
1, 2, 3, ... , n, where n is an integer, then min(Kx, n) represents the number of
payments made. An important difference between these two random variables
is that min(Tx, n) is a mixed random variable (with a density over (0, n) and
a mass of probability at n), whereas min(Kx, n) is a discrete random variable
since Kx is a discrete random variable. The expected value of min(Kx, n) is
 34 Survival models

denoted ex:lil• and when n is an integer is given by

n

ex:lil = L kPx ·
k=I

The proof of this result is set as Exercise 2.16.

2.6.2 The complete and curtate expected future lifetimes, ~x and ex

As the curtate future lifetime is the integer part of future lifetime, it is natural
to ask if there is a simple relationship between ~x and ex. We can obtain an
approximate relationship by writing

~x = f
lo
00

tPx dt = f
l=O jl
1
~i+ tPx dt.
If we approximate each integral using the trapezium rule for numerical inte-
gration (see Appendix B), we obtain

{}+! I
jl tPx dt;::::; z (Jpx + J+lPx),
and hence
00 00

~x;::::; L ~ (Jpx + J+!Px) = ~ + LiPx·

1=0 l=l
Thus, we have an approximation that is frequently applied in practice, namely

(2.25)
In Chapter 5 we will meet a refined version of this approximation. Table 2.2
shows values of ~x and ex for a range of values of x when the survival model
is Gompertz' law, with B = 0.0003 and c = 1.07. Values of ex were calculated
by applying formula (2.24) with an upper limit of summation of 130 - x, and
values of ~ x are as in Table 2.1. This table illustrates that formula (2.25) is a
very good approximation in this particular case for younger ages, but is less
accurate at very old ages. This observation is true for most realistic survival
models.

2.7 Notes and further reading

Although laws of mortality such as Gompertz' law are appealing due to their
simplicity, they rarely represent mortality over the whole span of human ages.
 2. 7 Notes and further reading 35

Table 2.2 Values of ex and ~x

for Gompertz' law with
B = 0.0003 and c = 1.07.
0
x ex ex
0 71.438 71.938
10 61.723 62.223
20 52.203 52.703
30 42.992 43.492
40 34.252 34.752
50 26.192 26.691
60 19.052 19.550
70 13.058 13.555
80 8.354 8.848
90 4.944 5.433
100 2.673 3.152

A simple extension of Gompertz' law is Makeham's law (Makeham, 1860),

which models the force of mortality as

(2.26)

This is very similar to Gompertz' law, but adds a fixed term that is not age
related, that allows better for accidental deaths. The extra term tends to improve
the fit of the model to mortality data at younger ages. See Exercise 2.11.
In recent times, the Gompertz-Makeham approach has been generalized fur-
ther to give the GM(r, s) (Gompertz-Makeham) formula,

fLx = h;(x) + exp{h;(x)},

where h; and h; are polynomials in x of degree r and s, respectively. A dis-

cussion of this formula can be found in Forfar et al. (1988). Both Gompertz'
law and Makeham's law are special cases of the GM formula.
In Section 2.3, we noted the importance of the force of mortality. A further
significant point is that when mortality data are analysed, the force of mortality
is a natural quantity to estimate, whereas the lifetime distribution is not. The
analysis of mortality data is a huge topic and is beyond the scope of this book.
An excellent summary article on this topic is Macdonald (1996). For more
general distributions, the quantity fo(x)/So(x), which actuaries call the force
of mortality at age x, is known as the hazard rate in survival analysis and
the failure rate in reliability theory.
36 Survival models

2.8 Exercises
Exercise 2.1 Let Fo(t) = 1 - (1 - t /105) 115 for 0 .::: t .::: 105. Calculate

(a) the probability that a newborn life dies before age 60,
(b) the probability that a life aged 30 survives to at least age 70,
(c) the probability that a life aged 20 dies between ages 90 and 100,
(d) the force of mortality at age 50,
(e) the median future lifetime at age 50,
(t) the complete expectation of life at age 50,
(g) the curtate expectation of life at age 50.

Exercise 2.2 The function

18 000 - llOx - x 2
G(x) = 18 000
has been proposed as the survival function So(x) for a mortality model.

(a) What is the implied limiting age w?

(b) Verify that the function G satisfies the criteria for a survival function.
(c) Calculate 20 po.
(d) Determine the survival function for a life aged 20.
(e) Calculate the probability that a life aged 20 will die between ages 30
and 40.
(t) Calculate the force of mortality at age 50.

Exercise 2.3 Calculate the probability that a life aged 0 will die between ages
19 and 36, given the survival function
1
So(x) = .J100 - x, 0.::: x .::: 100 (= w).
10
Exercise 2.4 Let

So(x) =exp {-(Ax+ ~Bx

2
2 + _E_Dx -
log D
_!!_)}
log D
where A, B, C and D are all positive.

(a) Show that the function So is a survival function.

(b) Derive a formula for Sx (t).
(c) Derive a formula for /,lx·
(d) Now suppose that

A= 0.00005, B = 0.0000005, C = 0.0003, D = 1.07.

(i) Calculate tP30 fort= 1, 5, 10, 20, 50, 90.

(ii) Calculate tq40fort=1, 10, 20.
 2.8 Exercises
37

(iii) Calculate tl1oq30fort=1, 10, 20.

(iv) Calculate ex for x =70, 71, 72, 73, 74, 75.
(v) Calculate e0x for x = 70, 71, 72, 73, 74, 75, using numerical integration.

Exercise 2.5 Let Fo(t) = 1 - e-M, where J... > 0.

(a) ShowthatSx(t)=e-AI.
(b) Show that f.1,x =A.
1
(c) Show that ex= (e" - 1)- .
(d) What conclusions do you draw about using this lifetime distribution to
model human mortality?
Exercise 2.6 Given Px =0.99, Px+l =0.985, 3Px+l =0.95 and
qx+3 = 0.02, calculate

(a) Px+3,
(b)2Px•
(c) 2Px+l,
(d) 3Px,
(e) 1i2qx.
Exercise 2.7 Given
1
Fo(x) = 1 - - - for x 2: 0,
l+x
find expressions for (a), (b), (c) below, simplifying as far as possible,

(a) So(x),
(b) fo(x),
(c) Sx(t),

and calculate:

(d) pzo, and

(e) 10 isq30.

Exercise 2.8 Given

So(x) =e-o.001x2 for x 2: 0,

find expressions for (a) and (b ), simplifying as far as possible,

(a) fo(x), and

(b) f.1,x·
Exercise 2.9 Show that
d
dx tPx =tPx (µ,x - f.1,x+t) ·
38 Survival models

Exercise 2.10 Suppose that Gompertz' law applies with µ,30 = 0.000130 and
f..lso = 0.000344. Calculate 10P40·

Exercise 2.11 A survival model follows Mak:eham's law, so that

for x :=::: 0.

(a) Show that under Mak:eham's law

(2.27)

wheres= e-A andg= exp{-B/logc}.

(b) Suppose you are given the values of 10Pso, 10P60 and 10P70· Show that

c = ( log(10P10) - log(10P60)) 0.1

log(10P60) - log(10Pso)

Exercise 2.12 (a) Construct a table of Px for Mak:eham's law with parame-
ters A= 0.0001, B = 0.00035 and c = 1.075, for integer x from age 0 to
age 130, using Excel or other appropriate computer software. You should
set the parameters so that they can be easily changed, and you should
keep the table, as many exercises and examples in future chapters will use
Mak:eham's law.
(b) Use the table to determine the age last birthday at which a life currently
aged 70 is most likely to die.
(c) Use the table to calculate e10.
(d) Using a numerical approach, calculate ~70·

Exercise 2.13 A life insurer assumes that the force of mortality of smokers at
all ages is twice the force of mortality of non-smokers.

(a) Show that, if* represents smokers' mortality, and the 'unstarred' function
represents non-smokers' mortality, then

(b) Calculate the difference between the life expectancy of smokers and non-
smokers aged 50, assuming that non-smokers mortality follows Gompertz'
law, with B = 0.0005 and c = 1.07.
(c) Calculate the variance of the future lifetime for a non-smoker aged 50 and
for a smoker aged 50 under Gompertz' law.

Hint: You will need to use numerical integration for parts (b) and (c).
 2.8 Exercises 39

Exercise 2.14 (a) Show that

~x < ~x+l + 1.
(b) Show that

(c) Explain (in words) why

(d) Is ~x always a non-increasing function of x?

Exercise 2.15 (a) Show that

~x = -
1
-1
So(x) x
00

So(t)dt,

where So(t) = 1 - Fo(t), and hence, or otherwise, prove that

d 0 0
dx ex = /J.,x ex - 1.

Hint: :x {ix g(t)dt} = g(x). What about :x lia g(t)dt}?

(b) Deduce that

0
x +ex

is an increasing function of x, and explain this result intuitively.

Exercise 2.16 Show that for integer n,

n

ex:iil = L kPx ·
k=l

Answers to selected exercises

2.1 (a) 0.1559

(b) 0.8586
(c) 0.1394
(d) 0.0036
(e) 53.28
(f) 45.83
(g) 45.18
40 Survival models
2.2 (a) 90
(c) 0.8556
(d) 1 - 3x/308 - x 2 /15 400
(e) 0.1169
(f) 0.021
2.3 0.1
2.4 (d) (i) 0.9976, 0.9862, 0.9672, 0.9064, 0.3812, 3.5 x 10-7
(ii) 0.0047, 0.0629, 0.1747
(iii) 0.0349, 0.0608, 0.1082
(iv) 13.046, 12.517, 12.001, 11.499, 11.009, 10.533
(v) 13.544, 13.014, 12.498, 11.995, 11.505, 11.029
2.6 (a) 0.98
(b) 0.97515
(c) 0.96939
(d) 0.95969
(e) 0.03031
2.7 (d) 0.95455
(e) 0.08218
2.10 0.9973
2.12 (b) 73
(c) 9.339
(d) 9.834
2.13 (b) 6.432
(c) 125.89 (non-smokers), 80.11 (smokers)