ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

CHAPTER 1
SURVIVAL MODELS I

Let T be a non-negative random variable representing the waiting time until the occurrence
of an event of interest (such as death, disability, sickness, retirement, unemployment,
recovery after diagnosis of a disease, survival following treatment or surgery, time-until-
default of credit payment in a bond, time-until-bankruptcy of a company etc). For
simplicity we will adopt the terminology of survival analysis, referring to the event of
interest as ‘death’ and to the waiting time as ‘survival’ time, but the techniques to be studied
have much wider applicability.

1.1) The Survival Function

Assume T is a continuous random variable with probability density function (p.d.f.) f (t )

and cumulative distribution function (c.d.f.) F (t ) = P(T  t ) . Recall that f (t ) = F (t ) and
t
F (t ) =  f ( )d .
0
The survival function S (t ) is defined as


S (t ) = P(T  t ) = 1 − P(T  t ) = 1 − F (t ) =  f ( )d
t

Since f (t ) = F (t ) and S (t ) = 1 − F (t ) , therefore f (t ) = −S (t ) .

The survival function S (t ) is the probability of surviving at least to time t, or the

probability of being alive at duration t, or the probability that the event of interest has not
occurred by duration t, or the probability that the time of occurrence of the event of interest
is later than some specified time t.

The survival function S (t ) is defined on the domain t  [0, ) and S (t )  [0,1]

Properties of Survival Function

The survival function S (t ) has the following properties

• S (0) = 1
• lim S (t ) = 0
t →

• S (t ) is non-increasing, ie S (t1 )  S (t 2 ) for t1  t 2

School of Mathematical Sciences (SMS) Page 1 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 1a

−2
Determine whether the function S (t ) = (1 + t ) qualifies to be a survival function.

Solution

Note that

• S (0) = 1
−2
• lim S (t ) = lim (1 + t ) =0
t → t →

• S (t ) = −2(1 + t ) −3  0 , hence S (t ) is non-increasing. Alternatively, S (t ) is non-

−2 −2
increasing because (1 + t1 ) = S (t1 )  S (t 2 ) = (1 + t 2 ) for t1  t 2 .

−2
Therefore S (t ) = (1 + t ) qualifies to be a survival function.

Example 1b

− t
Determine whether the function S (t ) = e (   0 ) qualifies to be a survival function.

Solution

School of Mathematical Sciences (SMS) Page 2 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

1.2) The Hazard Function (Force of Mortality)

The hazard function h(t ) is defined as

P(t  T  t + t | T  t )
h(t ) = lim
t →0 t

The hazard function h(t ) is the instantaneous rate of occurrence of an event. The hazard
function h(t ) is NOT a probability. The hazard function is sometimes referred to as hazard
rate function, or hazard rate, or failure rate. In the actuarial context, the hazard function is
normally called the force of mortality.

The numerator of the above expression is the conditional probability that the event will
occur in the interval (t , t +  t ] given that it has not occurred before, and the denominator is
the width of the interval. Dividing one by the other we obtain a rate of occurrence of the
event per unit of time. Taking the limit as the width of the interval goes down to zero, we
obtain an instantaneous rate of occurrence.

The conditional probability in the numerator may be written as the ratio of the joint
probability that T is in the interval (t , t +  t ] and T  t to the probability of the condition
T  t . The former may be written as f (t )t for small t , while the latter is S (t ) . Dividing
by t , we obtain another definition for the hazard function as shown below

P(t  T  t + t | T  t )
h(t ) = lim
t → 0 t
P (t  T  t + t  T  t )
= lim
t → 0 P (T  t )t
P (t  T  t + t )
= lim
t → 0 P (T  t )t
f (t )t
= lim
t → 0 P (T  t )t
f (t )t
= lim
t → 0 S (t )t
f (t )
=
S (t )

School of Mathematical Sciences (SMS) Page 3 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Since f (t ) = − S (t ) , we have

f (t )
h(t ) =
S (t )
S (t )
=−
S (t )
d
= − ln S (t )
dt

Hence,

h( )d = −d (ln S ( ))

t t
 h( )d = − d (ln S ( ))
0 0
t
 h( )d = − ln( S (t ))
0
− 0t h ( ) d
S (t ) = e

t
The cumulative hazard function, H (t ) is defined as H (t ) =  h( )d , so that
0
− H (t )
S (t ) = e . Like the hazard function, the cumulative hazard function is NOT a
probability. The cumulative hazard function H (t ) can be thought of as the sum of the risks
one face going from duration 0 to t.

From the above results, we see that the survival function S (t ) and hazard functions h(t )
provide alternative but equivalent characterizations of the distribution of T. Given the
survival function, we can always differentiate to obtain the probability density function
f (t )
f (t ) and then calculate the hazard function using h(t ) = . Given the hazard function
S (t )
t
h(t ) we can always integrate to obtain the cumulative hazard function H (t ) =  h( )d
0
− H (t )
and then exponentiate to obtain the survival function using S (t ) = e .

Properties of Hazard Function

The hazard function h(t ) has the following properties

• h(t )  0

•  0
h(t ) dt = 

School of Mathematical Sciences (SMS) Page 4 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 3

Determine whether the following functions qualify to be a hazard function. Find the
corresponding survival function S (t ) .

1 1 − rt
(i) h(t ) = t 2 (ii) h(t ) = (iii) h(t ) = e , r  0
4 (t + 1) 2
Solution

School of Mathematical Sciences (SMS) Page 5 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 4

Consider the hazard function defined by h(t ) =  where  is a constant.

Find the cumulative hazard function H (t ) , the survival function S (t ) , the probability
density function f (t ) , and the cumulative distribution function F (t ) .

Solution

t t
H (t ) =  h( )d =  d = t .
0 0

S (t ) = e − H (t ) = e − t

f (t )
Since h(t ) = , therefore
S (t )

f (t ) = h(t ) S (t ) = e − t (Exponential probability density function with parameter  )

t t
F (t ) =  f ( )d =  e −  d = 1 − e − t
0 0

Example 5

Consider the hazard function defined by h(t ) = kt where n  0, k  0 .

n

Find the cumulative hazard function H (t ) , the survival function S (t ) , the probability
density function f (t ) , and the cumulative distribution function F (t ) .

Solution

t
H (t ) =  h( )d =
0

S (t ) = e − H (t ) =

f (t )
Since h(t ) = , therefore
S (t )

f (t ) = h(t ) S (t ) =

F (t ) =

School of Mathematical Sciences (SMS) Page 6 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 6

Provide a summary of results between f (t ) , F (t ) , S (t ) , h(t ) , and H (t )

Solution

f (t ) = F (t ) = − S (t )

t
F (t ) =  f ( )d
0

S (t ) = P(T  t ) = 1 − F (t ) =  f ( )d = e − H (t ) = e 0
 − h ( ) d

f (t ) f (t ) S (t ) d
h(t ) = = =− = − ln S (t )
S (t ) 1 − F (t ) S (t ) dt

t
H (t ) =  h( )d = − ln S (t )
0

School of Mathematical Sciences (SMS) Page 7 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

1.3) Age-at-failure random variable

The age-at-failure (or death) random variable, denoted as X, is assumed to be continuous

and non-negative. X is interpreted as the lifetime of a newborn in the actuarial context.

d d
f (t ) = f 0 ( x) = F0 ( x) = − S 0 ( x)
dx dx
x
F (t ) = F0 ( x) =  f 0 ( )d
0

S (t ) = S 0 ( x) = P( X  x) = 1 − F0 ( x) =  f 0 ( )d = e 0 
 −  d

f 0 ( x) f 0 ( x) − S 0 ( x) d
h(t ) =  x = = = = − ln S 0 ( x)
S 0 ( x) 1 − F0 ( x) S 0 ( x) dx

x
H (t ) = U x =   d = − ln S 0 ( x)
0

Properties of S 0 ( x )
• S 0 ( 0) = 1
• lim S 0 ( x) = 0
x →

• S 0 ( x ) is non-increasing

Properties of  x
• x  0

• 0
 x dx = 

School of Mathematical Sciences (SMS) Page 8 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 7a

The age-at-failure random variable X has a cumulative distribution function defined by

1
F0 ( x) = 1 − 0.1(100 − x) 2 for 0  x  100

Find the

(a) probability density function f 0 ( x )

(b) survival function S 0 ( x )
(c) hazard function  x

Solution

School of Mathematical Sciences (SMS) Page 9 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 7b

The age-at-failure random variable X has a cumulative distribution function defined by

x
F0 ( x) = for 0  x  


Find the

(a) probability density function f 0 ( x )

(b) survival function S 0 ( x )
(c) hazard function  x

Solution

School of Mathematical Sciences (SMS) Page 10 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

1.4) Time-to-failure random variable

A life aged x is denoted as (x) . For a person currently aged x, the time-to-failure (or death)
random variable, or future lifetime random variable, is denoted as Tx . Note that Tx = X − x
. For a newborn, x = 0 , so that we have T0 = X .

Notation
S x (t ) : probability that (x) will survive for another t years = t p x
Fx (t ) : probability that (x) will fail (die) within t years = t qx

d d d S0 ( x + t ) f0 ( x + t )
f (t ) = f x (t ) = Fx (t ) = − S x (t ) = − =
dt dt dt S 0 ( x) S 0 ( x)

P ( X  x + t  X  x ) S0 ( x) − S0 ( x + t )
F (t ) = Fx (t ) = P (Tx  t ) = P ( X  x + t | X  x) = =
P( X  x) S0 ( x)

P( X  x + t  X  x) P ( X  x + t ) S0 ( x + t )
S (t ) = S x (t ) = P (Tx  t ) = P ( X  x + t | X  x) = = =
P( X  x) P( X  x) S0 ( x)

f x (t ) f ( x + t ) S 0 ( x) f (x + t) d
h(t ) =  x (t ) = = 0 . = 0 =  x +t = − ln( S x (t ))
S x (t ) S 0 ( x) S 0 ( x + t ) S 0 ( x + t ) dt

t
H (t ) = U x (t ) =   x + d = − ln S x (t )
0

Properties of S x (t )
• S x ( 0) = 1
• lim S x (t ) = 0
t →

• S x (t ) is non-increasing

Properties of  x (t )
•  x (t )  0

• 0
 x (t ) dt = 

School of Mathematical Sciences (SMS) Page 11 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Define

t|u qx : probability that (x) will survive for another t years and die within the next u years

: probability that (x) will die between the ages x + t and x + t + u

We write 1 p x = p x , 1 q x = q x and t|1 q x = t| q x

Example 8
t |u qx = P (t  Tx  t + u )
= Fx (t + u ) − Fx (t )
= t +u qx − t qx
= t px − t +u px
= S x (t ) − S x (t + u )

Example 9

P( X  x + u + t  X  x + u ) P( X  x + u + t )
S x +u (t ) = P( X  x + u + t | X  x + u ) = =
P( X  x + u ) P( X  x + u )
S0 ( x + u + t ) S x (u + t )
= =
S0 ( x + u ) S x (u )

Example 10

Let x = 0 and u = x in Example 9, we get

P( X  x + t  X  x ) P( X  x + t ) S0 ( x + t )
S x (t ) = P ( X  x + t | X  x ) = = =
P( X  x ) P( X  x ) S0 ( x )

School of Mathematical Sciences (SMS) Page 12 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 11a

Consider the survival function given by

1
S 0 ( x) = 101 (100 − x) 2 for 0  x  100

(a) Derive the following

(i) S x (t )
(ii) S x+u (t )
(iii) t qx
(iv) t|u q x

(b) Find the probability that

(i) A person age 50 will survive another 30 years

(ii) A newborn survives to age 75 but does not survive to age 84
(iii) A person age 20 survives to age 75 but not to age 84
(iv) A person age 60 will die sometime between 2 and 5 years from now

Solution

School of Mathematical Sciences (SMS) Page 13 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 11b

Consider the survival function given by

x
S0 ( x ) = 1 − for 0  x  100
100

(a) Derive the following

(i) S x (t )
(ii) S x+u (t )
(iii) t qx
(iv) t|u q x

(b) Find the probability that

(i) A person age 40 will survive another 20 years

(ii) A newborn survives to age 70 but does not survive to age 85
(iii) A person age 20 survives to age 70 but not to age 85
(iv) A person age 50 will die sometime between 3 and 7 years from now

Solution

School of Mathematical Sciences (SMS) Page 14 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 11c

Consider the survival function given by

x 2 − 190 x + 9000
S0 ( x) = for 0  x  90
9000

(a) Derive the following

(i) S x (t )
(ii) S x+u (t )
(iii) t qx
(iv) t|u q x

(b) Find the probability that

(i) A person age 45 will survive another 25 years

(ii) A newborn survives to age 60 but does not survive to age 75
(iii) A person age 25 survives to age 60 but not to age 75
(iv) A person age 40 will die sometime between 3 and 5 years from now

Solution (Answer: 0.2424, 0.0917, 0.1692, 0.0680)

School of Mathematical Sciences (SMS) Page 15 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 12a (Gompertz’s Law)

Consider the hazard function (force of mortality) given by

 x = Bc x for B  0, c  1

(a) Derive the following

(i) t px
(ii) t qx
(iii) t|u q x

(b) Find the probability that (assuming B = 0.001 and c = 1.05 )

(i) A person age 36 will survive to age 75

(ii) A person age 25 surviving 20 years and then dying in the next 10 years
(iii) A person age 35 will die sometime between ages 40 and 50

Solution (Answer: 0.5080, 0.09750, 0.08408)

School of Mathematical Sciences (SMS) Page 16 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 12b (Makeham’s Law)

Consider the hazard function (force of mortality) given by

 x = A + Bc x for A  0, B  0, c  1

(a) Derive the following

(i) t px
(ii) t qx
(iii) t|u q x

(b) Find the probability that (assuming A = 0.05 , B = 0.001 and c = 1.03 )

(i) A person age 36 will survive to age 75

(ii) A person age 35 surviving 10 years and then dying in the next 20 years
(iii) A person age 25 will die sometime between ages 30 and 50

Solution (Answer: 0.1150, 0.3922, 0.5049)

School of Mathematical Sciences (SMS) Page 17 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 12c (Weibull’s Distribution)

Consider the hazard function (force of mortality) given by

 x = kx n for n  0, k  0

(a) Derive the following

(i) t px
(ii) t qx
(iii) t|u q x

(b) Find the probability that (assuming n = 0.5 and k = 0.001)

(i) A person age 20 will survive to age 65

(ii) A person age 20 surviving 30 years and then dying in the next 10 years
(iii) A person age 30 will die sometime between ages 45 and 65

Solution (Answer: 0.7485, 0.05992, 0.1256)

School of Mathematical Sciences (SMS) Page 18 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 12d (Constant Force)

Consider the hazard function (force of mortality) given by

x = 

(a) Derive the following

(i) t px
(ii) t qx
(iii) t|u q x

(b) Find the probability that (assuming  = 0.05 )

(i) A person age 25 will survive to age 45

(ii) A person age 25 surviving 20 years and then dying in the next 10 years
(iii) A person age 35 will die sometime between ages 40 and 50

Solution (Answer: 0.3679, 0.1447, 0.3064)

School of Mathematical Sciences (SMS) Page 19 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 12e (Uniform Distribution)

Consider the hazard function (force of mortality) given by

1
x = for 0  x  
−x

(a) Derive the following

(i) t px
(ii) t qx
(iii) t|u q x

(b) Find the probability that (assuming  = 100 )

(i) A person age 35 will survive to age 75

(ii) A person age 35 surviving 20 years and then dying in the next 10 years
(iii) A person age 35 will die sometime between ages 40 and 50

Solution (Answer: 0.3846, 0.1538, 0.1538)

School of Mathematical Sciences (SMS) Page 20 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 12f (Beta Distribution)

Consider the hazard function (force of mortality) given by


x = for 0  x   ,   0
−x

(a) Derive the following

(i) t px
(ii) t qx
(iii) t|u q x

(b) Find the probability that (assuming  = 100 and  = 0.5 )

(i) A person age 36 will survive to age 75

(ii) A person age 25 surviving 30 years and then dying in the next 10 years
(iii) A person age 40 will die sometime between ages 50 and 70

Solution (Answer: 0.6250, 0.0915, 0.2058)

School of Mathematical Sciences (SMS) Page 21 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

1.5) Life Table

A life table is a tabular representation of the mortality evolution of a cohort group of lives.
The life table usually begins with l0 number of lives ( l0 is called the radix of the life table).
The limiting age of the table is usually denoted as  , which means the table ends at age
 . No one survives beyond this age so that S 0 ( − 1)  0 but S 0 ( ) = 0 .

l x : number of lives who are age x (number of survival to age x)

l x = l0 S 0 ( x ) = l0 x p0

Hence, we see that l x can be interpreted as the number of survivors to age x out of a cohort
group of l0 lives.

d x : number of deaths between ages x and x +1

n d x : number of deaths between ages x and x + n

From the life table we will be able to calculate the values for n p x and n q x for integer
values of x and n . Recall that

S x (t )= t p x : probability that (x) will survive for another t years

Fx (t )= t q x : probability that (x) will die within t years
t|u q x : probability that (x) will die between the ages x + t and x + t + u

n d x = l x − l x+n
lx+n
S 0 ( x + n) lx + n
n px = = =
l0
lx
S0 ( x) l0 lx
lx − lx+n n d x
n q x = 1− n p x = =
lx lx
m d x+n l −l
n| m q x = n p x − n + m p x = = x+n x+n+m
lx lx
n| m q x = n p x m q x + n

School of Mathematical Sciences (SMS) Page 22 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 13

Verify the following

 −1
(a) d
x =0
x = l0

 −1
(b) x =0
x| q0 = 1

 − x −1
(c) d
k =0
x+k = lx

n −1
(d) d
k =0
x+k =n d x

Solution

(a)
 −1

d
x =0
x = d 0 + d1 + d 2 + ... + d −1

= (l0 − l1 ) + (l1 − l2 ) + (l2 − l3 ) + ... + (l −1 − l )

= l0 − l
= l0

School of Mathematical Sciences (SMS) Page 23 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 14

Complete the following life table with radix l0 = 10,000

x px lx dx
0 0.8
1 0.7
2 0.4
3 0.0

Solution

l x +1
First we obtain the values for l x using the formula p x = .
lx
l1
For example, from p0 = , we obtain l1 = l0 p0 = (10,000)(0.8) = 8000
l0
The values of d x are obtained from d x = l x − l x +1

Example 15

Complete the following life table

x lx px dx qx
40 24,983
41 24,541
42 24,175
43 23,880
44 23,656
45 23,495 - - -

School of Mathematical Sciences (SMS) Page 24 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 16

Based on the life table below, calculate the probability that a person age 60 will die
sometime between 2 and 5 years from now. (Answer: 0.011917)

x qx
60 0.001
61 0.002
62 0.003
63 0.004
64 0.005

Example 17

Based on the life table below, calculate the probability that a person age 71 survive to age
74. (Answer: 94 )

x lx dx
70 100 10
71 90 15
72 75 15
73 60 20
74 40 18

Example 18

Based on the life table below, calculate the probability that a newborn will survive to age
2. (Answer: 0.9288)

x lx dx px
0 50
1 0.98
2 890

School of Mathematical Sciences (SMS) Page 25 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 19a

A life table is developed using a radix of l0 = 100,000 . The number of survival to age x is
given by

 90 − x 
l x = 100,000   for 0  x  90
 90 + x 

Find the

(a) probability of surviving from birth to age 60

(b) probability a life age 10 will survive for another 35 years

(c) probability a life age 10 will survive to age 35

(d) probability a life age 10 will die between age 30 and age 45

(e) probability a life age 10 will die between age 25 and age 45

Solution

School of Mathematical Sciences (SMS) Page 26 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

Example 19b

A life table is developed using a radix of l0 = 100,000 . The number of survival to age x is
given by
1
l x = 10,000(100 − x) 2 for 0  x  100

Find the

(a) probability of surviving from birth to age 50

(b) probability a life age 20 will survive for another 25 years

(c) probability a life age 20 will survive to age 25

(d) probability a life age 20 will die between age 40 and age 55

(e) probability a life age 20 will die between age 35 and age 55

Solution (Answer: 0.7071, 0.8292, 0.9682, 0.1160, 0.1514)

School of Mathematical Sciences (SMS) Page 27 April 2023 Semester

ASC2014 Life Contingencies I BSc (Hons) in Actuarial Studies

School of Mathematical Sciences (SMS) Page 28 April 2023 Semester