ACTUARIAL MODELS

Introduction to Survival Models

Dr. F. T. Oduro

Kwame Nkrumah University of Science and Technology

[email protected]

January 3, 2016

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 1 / 59

1 Continuous Survival Models

2 The Cumulative distribution function of X

3 Other Reserve Formulae

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 2 / 59
Continuous Survival Models

The distribution of X, the random lifetime of a newborn life may be

specified using any of the following:
the cumulative distribution function of X
the probability density function of X
the survival function
the life table function
the force of mortality

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 2 / 59

The Cumulative distribution function of X

The cdf, FX (x), represents the probability that a newborn life will die at or
before age x.

FX (x) = Pr (X ≤ x)
FX (x) is continuous and non-decreasing with FX (0) = 0 and FX (ω) = 1
where ω is the first age at which death is certain to have occurred for a
newborn life.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 3 / 59

Probability density function of X

The probability density function is given by:

fX (x) = FX0 (x)

and Z ω
fX (x)dx = 1
0

The pdf fX (x) is non-negative and continuous on the interval [0, ω)

The probability that a newborn life dies between ages a and b is:
Z b
Pr (a ≤ X ≤ b) = fx (u)du = F (b) − F (a)
a

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 4 / 59

Properties of pdf and cdf

The probability that a newborn life dies in the interval [x, x + 4x] can be
estimated as
Pr (x ≤ X ≤ x + 4x) = fX (x) 4 x
We also note that
Z x
FX (x) = Pr (X ≤ x) = fx (u)du
0

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 5 / 59

The Survival Function

The survival function is given by

sX (x) = Pr (X > x)

The survival function gives the probability that a newborn life dies
after age x. This is the same as saying that the newborn survives to
age x, or is alive at age x.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 6 / 59

Properties of the Survival Function

From the distributions of the lifetime variable X, we can deduce the

following properties of the survival function.

sX (x) = 1 − FX (x)
Z b
Pr (a ≤ X ≤ b) = fX (x)dx = sX (a) − sX (b)
a

fX (x) = −sX0 (x)

sX (x) is continuous and non-increasing with sX (0) = 1 and sX (ω) = 0

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 7 / 59

Random Variable L(x)

Let the number of survivors at age x be a random variable L(x). The

life function lx represents the expected number of survivors. Thus we
have
lx = E [L(x)]
Assume L(x) follows a binomial distribution. Each life is viewed as an
independent Bernoulli trial with number of trials n = l0 . ”success” is
survival to age x, with p =x P0

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 8 / 59

The Life Table Function

The life table function lx is defined as the expected number of

survivors at age x. Hence

lx = E [L(x)] = np = l0 sX (x)

Here, the random variable L(x) follows a binomial distribution with

n = l0 trials. We define ”success” as survival to age x, with
probability p = Pr (X > x) = sX (x).

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 9 / 59

Properties of the life table function

1 lx is the expected number of survivors at age x from a group of l0

newborn lives
2 lx = l0 sX (x) is continuous and non-increasing with lω = 0
lx
3 sX (x) =
l0

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 10 / 59

The life table - a discrete survival model

lx is the number of lives expected to survive to age x from a group of

l0 newborn lives.
A life table displays in a table format the values lx at ages x for x
equal to 0, 1, 2, ..., ω, where ω is the first whole number age at which
there are no remaining lives in the group.
dx represents the number of lives among l0 newborn lives that die in
the range [x, x + 1). It is given as:

dx = lx − lx+1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 11 / 59

Hypothetical Life Table

x 0 1 2 3 4 5 6 7 8 9
Lx 1000 991 985 982 979 976 972 968 964 959
x 9 6 3 3 3 3 4 4 4 6

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 12 / 59

Probabilities computed from the life table n Px

The probability that a life currently age x will survive the n years
denoted n Px and is given by
lx+n
n Px =
lx
NOTE: It is standard to omit the n subscript when n = 1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 13 / 59

Probabilities computed from the life table (n qx )

The probability that a life currently age x will die within n years is
denotedn qx and is given by
lx+n lx − lx+n
n qx = 1 −n px = 1 − =
lx lx
NOTE: It is standard to omit the n subscript when n = 1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 14 / 59

Probabilities computed from the life table (m|n qx )

The probability that a life currently age x will survive for m years and
then die within n years after is denoted m|n qx and is given by

lx+m − lx+m+n
m|n qx =
lx
NOTE: It is standard to omit the n subscript when n = 1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 15 / 59

Examples

The probability that a life currently age x will survive 1 year is:
lx+1
Px =
lx
For example, the probability that a life age 5 survives for 2 years to
age 7 is
l7 968
2 P5 = =
l5 976

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 16 / 59

Example

The probability that a life currently age 1 will die within 3 years is:
l1 − l4 991 − 979 12
3 q1 = = =
l1 991 991
Note that the probability that a life currently age x will die within 1
year is:
lx − lx+1 dx
qx = =
lx lx

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 17 / 59

Example

The probability that a life age 4 survives for 3 years and then dies
within the following 2 years is:
l7 − l9 968 − 959 9
3|2 q4 = = =
l4 979 979
The probability that a life currently age x will survive for m years and
then die within 1 year is:
lx+m − lx+m+1 dx+m
m| qx = =
lx lx

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 18 / 59

Example

Using the life table, the mean and variance of L(5) are:

E [L(5)] = np = l05 P0 = (1000)(0.976) = 976 = l5

Var [L(5)] = npq = l05 P05 q0 = (1000)(0.976)(0.024) = 23.424

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 19 / 59

Example

Suppose that the lifetime X of a newborn is exponentially distributed with

mean 75 years.
(i) Identify the survival function sX (x)
(ii) Calculate the probability that a newborn is still alive at the age 100
(iii) Calculate the probability that newborn ages 60 and 75

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 20 / 59

Solution

x
FX (x) = 1 − e − 75
implies
100
sX (100) = e − 75
thus
100
sX (100) = e − 75 = 0.26360
75
sX (60) − sX (75) = e − 75 = 0.08145

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 21 / 59

Example

Suppose that the lifetime X of a newborn is uniformly distributed on

[0,100].
(i) Identify the survival function sX (x)
(ii) Identify the table function lx if l0 = 100

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 22 / 59

Solution

The survival function is given by:

Z 100 Z 100
sX (x) = Pr (X > x) = fX (u)du = 0.01du
x x

= 0.01(100 − x) = 1 − 0.01x
The life table function is:

lx = l0 sX (x) = 100(1 − 0.01x) = 100 − x

for 0 ≤ x ≤ 100

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 23 / 59

Example

Let lx = 10, 000(100 − x)2 for 0 ≤ x ≤ 100 . Find the cdf and the pdf for
the associated lifetime variable X.
lx 10, 000(100 − x)2 (100 − x)2
sX (x) = = =
l0 10, 000(100)2 1002

FX (x) = 1 − sX (x)
(100 − x)2
=1−
1002

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 24 / 59

Example contd

The pdf is given by

fX (x) = FX0 (x)
(100 − x)2
 
d
= 1−
dx 1002
2(100 − x)
=
1002
100 − x
=
5000

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 25 / 59

Force of mortality

The force of the mortality is denoted µ(x) and is an instantaneous

measure of mortality at age x. It can be defined as

fX (x) s 0 (x) l0
µ(x) = =− X = −(ln(sX (x)))0 = − x
sX (x) sX (x) lx
The equalities can be easily verified.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 26 / 59

example

Using the data in the previous example, we have:

sX0 (x) −2(100 − x)/1002 2

µ(x) = − =− =
sX (x) (100 − x)2 /1002 100 − x
x
A previous example with FX (x) = 1 − e − 75 gives
 x 0 1
µ(x) = −(ln(sX (x)))0 = − − =
75 75

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 27 / 59

example

Using the data in the previous example, we have:

sX0 (x) −2(100 − x)/1002 2

µ(x) = − =− 2 2
=
sX (x) (100 − x) /100 100 − x
x
A previous example with FX (x) = 1 − e − 75 gives
 x 0 1
µ(x) = −(ln(sX (x)))0 = − − =
75 75

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 28 / 59

Survival function from the force of mortality

Survival function can be written in terms of the force of mortality:

Z x
µ(y )dy = − ln(sX (x))
0
 Z x 
⇒ sX (x) = exp − µ(y )dy
0

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 29 / 59

Example

Let the force of mortality be given by the formula:

0.9
µ(x) =for 0 ≤ x < 90
90 − x
Calculate the survival function.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 30 / 59

Solution

 Z x 
sX (x) = exp − µ(y )dy
0
 Z x 
0.9
= exp − dy
0 90 − y

= exp (0.9 ln(90 − y )|x0 )

  
90 − x
= exp 0.9 ln
90

90 − x 0.9
 
=
90

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 31 / 59

The meaning of the force of mortality

We rewrite the defining formula in the form:

fX (x) 4 x = Pr (x < X ≤ x + 4x)

= Pr (X ≤ x + 4x|X > x)Pr (X > x)

= Pr (X ≤ x + 4x|X > x)sX (x)

⇒ sX (x)µ(x) 4 x = Pr (X ≤ x + 4x|X ≥ x)sX (x)

⇒ µ(x) 4 x = Pr (X ≤ x + 4x|X > x)
So, µ(x) 4 x is approximately equal to the conditional probability that a
newborn that has survived to age x subsequently dies during the next 4x
years.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 32 / 59

Example

Let the force of mortality be given by the formula:

0.9
µ(x) =
90 − x
for 0 ≤ x < 90
Calculate the approximate probability that a life aged 40 dies within the
next week.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 33 / 59

Solution

7
Setting 4x = 365 , the required probability is:

0.9 7
µ(40) 4 x = . = 0.00035
90 − x 365

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 34 / 59

Summary of the properties of the force of mortality

fX (x) s 0 (x) l0
1 µ(x) = =− X = −(ln(sX (x)))0 = x
sX (x) s (x) lx
Rx X

2 sX (x) = exp − 0 µ(y )dy
3 µ(x) 4 x = Pr (X ≤ x + 4x|X > x)
4 µ(x) is non-negative and piece-wise continuous
Rω
0 µ(y )dy = ∞ so that sX (ω) = 0
5

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 35 / 59

Standard Probabilities

The probability that a life currently aged x will survive t years

lx+t sX (x + t) Pr (X > x + t)
t Px = = = = Pr (X > x + t|X > x)
lx sX (x) Pr (X > x)

The probability that a life currently aged x will die within the next t years

lx − lx+t Pr (X > x) − Pr (X > x + t)

t qx = = = Pr (X ≤ x + t|X > x)
lx Pr (X > x)

The probability that a life currently aged x will survive s years but die
within the following t years
lx+s − lx+s+1
s|t qx = = Pr (x + s < X ≤ x + t|X > x)
lx

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 36 / 59

Suppose that the force of mortality for a survival model is given by the
formula
0.9
µ(x) =
90 − x
for 0 ≤ x < 90
Compute the following probabilities
(a) 2.5 P20
(b) 2.5 q20
(c) 2.5| q20

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 37 / 59

Key relations concerning standard probabilities

These are
t Px +t qx = 1
s+t Px = s Px t Px+s
s|t qx = s Px t qx+s =s Px (1 −t Px+s ) =s Px −s+t Px =s+t qx −s qx
n Px = Px Px+1 ...Px+n−1
n qx =0| qx +1| qx + ... +n−1| qx

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 38 / 59

Continuous future lifetime after age x

The conditional distribution of the time lived after age x, given survival to
age x, is
T (x) = X − x|X > x
The continuous random variable T (x) is a survival model defined on the
interval [0, ω − x]
Note that the corresponding survival function is

lx+t sX (x + t)
sT (x) (t) = P(T (x) > t) =t Px = =
lx sX (x)

since lx = l0 sX (x)

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 39 / 59

Example

1 Compute the table survival function for newborn lives.

2 Compute the survival function for lives currently aged 20.
where lx = 10, 000(100 − x)2

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 40 / 59

Solution

The survival function for newborn lives sX (x) is

2
10, 000(100 − x)2

lx 100 − x
sX (x) =x P0 = = =
l0 10, 000(100 − 0)2 100

0 ≤ x ≤ 100
The survival function for lives currently aged 20 sT (20)(t) is
2
10, 000(100 − (20 + t))2

l20+t 80 − t
sT (20)(t) =t P20 = = =
l20 10, 000(100 − 20)2 80

0 ≤ x ≤ 80

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 41 / 59

cdf and pdf for the distributions X and T(x)

Now
FT (t) = Pr (T (x) ≤ t) = Pr (X − x ≤ t|X > x)
Pr (x < X ≤ x + t) FX (x + t) − FX (x)
= =
Pr (X > x) 1 − FX (x)
Thus  
0 d FX (x + t) − FX (x)
fT (t) = (FT (t)) =
dt 1 − FX (x)
fX (x + t) fX (x + t) s 0 (x + t) l0
= = = X = − x+t
1 − FX (x) sX (x) sX (x) lx

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 42 / 59

fT (x) (t) and force of mortality

We have
fX (x) = sX (x)µ(x) =x P0 µ(x)
fX (x + t)
⇒ fT (t) =
sX (x)
x+t P0 µ(x + t) x P0t Px µ(x + t)
= =
x P0 x P0
=t Px µ(x + t)

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 43 / 59

Example

Compute the cdf and pdf for the future lifetime of a life aged 20 for
lx = 10, 000(100 − x)2 , 0 ≤ x ≤ 100.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 44 / 59

Solution

From the previous problem, the survival function sT (20) (t) for lives
currently aged 20, is
2
10, 000(100 − (20 + t))2

l20+t 80 − t
sT (20) (t) =t P20 = = =
l20 10, 000(100 − 20)2 80

0 ≤ x ≤ 80

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 45 / 59

Thus
80 − t 2
 
FT (20) (t) = 1 − sT (20) = 1 −
80
 0!
80 − t 2
  
0 (80 − t) (80 − t)
fT (20) (t) = (FT (20) (t)) = 1 − =2 2
=
80 80 3200

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 46 / 59

Alternative Solution for fT (20) (t)

We may also proceed using as follows

lx0 2
µ(x) = =−
lx 100 − x

80 − t 2 2
 
80 − t
fT (20) (t) =t P20 µ(20 + t) = =
80 80 − t 3200

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 47 / 59

Key relations between T(x) and X

These are
lx+t sX (x + t)
sT (t) = Pr (T (x) > t) =t Px = =
lx sX (x)

FX (x + t) − FX (x)
FT (t) =t qx = 1 − sX (t) =
sX (x)
fX (x + t)
fT (t) = =t Px µ(x + t)
sX (x)

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 48 / 59

Curtate future lifetime K(x) after age X
The curtate lifetime is a discrete random variable defined by
K(x)=[T(x)] ie the integer part of T(x).
Since it is a function of T(x), it is simple to calculate the probability
function of K(x) from what we know about T(x). The values of K(x)
are the numbers 0, 1, 2, ..., ω − 1
If the life (70) eventually dies at age 85.8, then the continuous future
lifetime is T(70)=15.8 and the curtate lifetime is
K (70) = [15.8] = 15.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 49 / 59

Probability function for K(x)

Note that if K (x) = k, then

k ≤ T (x) < k + 1

⇒ Pr (K (x) = k) = Pr (k ≤ T (x) < k + 1)

dx+k
=k| qx =
lx
for k = 0, 1, 2, ..., ω − x − 1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 50 / 59

Example

Given lx = 100 − x, we compute the probability function for K(75) as

follows:
d75+k l75+k − l75+k+1
Pr (K (75) = k) = =
l75 l75
(100 − 75 − k) − (100 − 75 − k − 1)
=
(100 − 75)
1
=
25
Thus K(75) has 25 possible values (0, 1, 2, ?, 24) that are equally
likely to occur.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 51 / 59

Example

Given lx = 1000e −0.015x , we compute the probability function for

K(75) as follows:
d75+k
Pr (K (75) = k) =
l75
l75+k − l75+k+1
=
l75
e −0.015(75+k) − e −0.015(75+k+1)
=
e −0.015(75)
= e −0.015k (1 − e −0.015 )
Note that this is a geometric distribution.

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 52 / 59

Survival function for the curtate future lifetime

For any random variable FX (x) = Pr (X ≤ x) Thus

FK (x) (k) = Pr (K (x) ≤ k)

= Pr (K (x) = 0) + Pr (K (x) = 1) + ... + Pr (K (x) = k)

dx dx+1 dx+2 dx+k
= + + + ... +
lx lx lx lx
lx − lx+k+1
=
lx
=k+1 qx

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 53 / 59

for k = 0, 1, ..., ω − x − 1 ⇒ sK (x) (k) = Pr (K (x) > k)

= 1 − FK (x) (k)

= 1 −k+1 qx
=k+1 Px
for k = 0, 1, ..., ω − x − 1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 54 / 59

Example

Given lx = 100 − x, we compute the survival function for K(75) as

follows:
sK (75) (k) =k+1 P75
l75+k+1
=
l75
100 − (75 + k + 1)
=
100 − 75
24 − k
=
25
for k = 0, 1, ..., 24

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 55 / 59

Example

Given lx = 1000e −0.015x , we compute the survival function for K(75)

as follows:
sK (75) (k) =k+1 P75
l75+k+1
=
l75
e −0.015(75+k+1)
=
e −0.015(75)
= e −0.015(k+1)
for k = 0, 1, 2, ...

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 56 / 59

Key relations concerning K(x)

These are
lx+k+1
sK (x) (k) =k+1 Px = lx for k = 0, 1, ..., ω − 1
dx+k
fK (x) (k) = Pr (K = k) =k| qx = lx for k = 0, 1, ..., ω − 1
FK (x) (k) =k+1 qx = 1 −k+1 Px for k = 0, 1, ..., ω − 1

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 57 / 59

 The End

Dr. F. T. Oduro (K.N.U.S.T) MPHIL(ACTUARIAL SCIENCE) January 3, 2016 58 / 59