Lecture Notes (Chapter 2) ASC2014 Life Contingencies I
Lecture Notes (Chapter 2) ASC2014 Life Contingencies I
CHAPTER 2
SURVIVAL MODELS II
2.1) Moments of T
Recall from Section 1 that T is 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).
E (T ) = t f (t ) dt = S (t ) dt
0 0
In other words, the mean is simply the integral of the survival function S (t ) .
(See Example 1a for the proving.)
Example 1a
Use integration by parts to show that 0
t f (t ) dt = S (t ) dt provided lim t S (t ) = 0 .
0 t →
Solution
E (T 2 ) = t 2 f (t ) dt
0
Var (T ) = E (T 2 ) − [ E (T )] 2
For the age-to-failure random variable X, the mean or expected value of X is called the
complete expectation of life at birth
o
e0 = E ( X ) = x f 0 ( x )dx =
0 0
S 0 ( x ) dx
E ( X 2 ) = x 2 f 0 ( x ) dx
0
Var ( X ) = E ( X 2 ) − [ E ( X )] 2
For the time-to-failure random variable T x , the mean or expected value of T x is called the
complete future lifetime at age x (also called the complete expectation of life at age x)
o
ex = E (Tx ) = t f x (t )dt =
0 0
t t p x x +t dt = 0
S x (t ) dt =
0
t p x dt
o
e 20 = 71.94 means that averagely a person aged 20 is going to live for another 71.94 years.
E (Tx ) = t 2 f x (t )dt = 2 t t p x dt
2
0 0
Example 1b
Use integration by parts to show that 0
t 2 f x (t )dt = 2 t t p x dt provided lim t 2 S x (t ) = 0 .
0 t →
Solution
The n-year temporary complete future lifetime at age x, the average future lifetime of
(x) within the next n years, is defined as
o n n
e x: n = t t p x x +t dt + n n p x = t p x dt
0 0
o
e 20:10 = 9.68 means that averagely a person aged 20 is going to live for another 9.68 years
within the next 10 years.
Given that
x =
(i) t p x = e − t
o
1
(ii) ex =
o
2
(b) Find an expression for e x:n , E (Tx ) and Var (T x ) .
Solution
(a) (i) t p x = e − 0 x + d = e − 0 d = e − t
t t
o
1 1
(ii) e x = t p x dt = e − t dt = − [e − t ]0 =
0 0
o n n
1 1
(b) e x:n = p x dt = e − t dt = − [e − t ]0n = [1 − e − n ]
t
0 0
E (Tx ) = 2 t t p x dt
2
0
= 2 te − t dt
0
1 − t 1
= 2 t ( − e ) ]0 − − e − t .1 dt
0
1 − t
= 2 e dt
0
2 1 2
= − e − t = 2
0
2 1 1
= − ( )2 = 2
2
Given that
1
x = 0 x
−x
t
(i) px = 1 −
−x
t
o
−x
(ii) ex =
2
o
2
(b) Find an expression for e x:n , E (Tx ) and Var (T x ) .
Solution
Given that
x = 0 x , 0
−x
t
(i) p x = (1 − )
−x
t
o
−x
(ii) ex =
+1
o
2
(b) Find an expression for e x:n , E (Tx ) and Var (T x ) .
Solution
2( − x ) 2 ( − x) 2
Answer: E (Tx 2 ) = , Var(Tx ) =
( + 1)( + 2) +1 ( + 2)( + 1)
0.01 t 5
x +t =
0.02 t 5
o o
Calculate e x and e x:n .
First, we find t p x
For t 5 , t p x = e − 0 x + d = e − 0 0.01d = e −0.01t
t t
For t 5 , t p x = e − 0 x + d
t
− [ 0 x + d + 5 x + d ]
5 t
=e
−[ 5
0.01d + t
0.02 d ]
= e 0 5
o 5
e x = t p x dt = t p x dt + t p x dt
0 0 5
5
= e − 0.01t
dt + e 0.05−0.02t dt
0 5
1 1
=− [ e − 0.01t ]50 + e 0.05 . − [ e − 0.02 t ]5
0.01 0.02
−0.05
= 100 [1 − e ] + e .50 [ e −0.10 ]
0.05
o n n
o n
For t 5 , e x: n = t p x dt
0
5 n
= e −0.01t dt + e 0.05−0.02t dt
0 5
0.04 0 x 40
x =
0.05 x 40
o o
Calculate e 25 and e 25:25 .
Solution
The curtate future lifetime of (x) , denoted K x , refers to the number of future years
completed by (x) prior to death. K x is the greatest integer of T x .
P ( K x = k ) = P ( k Tx k + 1)
= P (Tx k ) − P (Tx k + 1)
= S x ( k ) − S x ( k + 1)
= k p x − k +1 p x
= k +1 qx −k qx
= k| q x
= probabilit y that ( x ) will die between ages x + k and x + k + 1
k k
P( K x k ) = h| q x = ( h +1 q x − h q x ) = k +1 q x (See Tutorial 2)
h=0 h=0
The mean or expected value of K x is called the curtate future lifetime of life at age x
ex = E ( K x ) = k P( K x = k ) = k k |q x = k p x (See Tutorial 2)
k =0 k =0 k =1
E ( K x ) = k 2 k | q x = ( 2 k − 1) k p x
2
k =0 k =1
Var ( K x ) = E ( K x ) − [ E ( K x )] 2
2
The n-year temporary curtate future lifetime at age x, the average future lifetime of (x)
within the next n years, is defined as
n −1 n
ex:n = k k |q x + n n p x = k p x
k =0 k =1
Given that
x =
(i) k| q x = e − k (1 − e − )
1
(ii) ex =
e −1
Solution
Then, k | q x = k p x − k +1 p x = e − k − e − ( k +1) = e − k (1 − e − )
n
ex = k p x ex:n = k px
k =1 k =1
n
= e − k = e− k
k =1 k =1
− −2 −3
=e +e +e + ... = e − + e −2 + e −3 + ... + e − n
e− e − (1 − e − n )
= =
1 − e− 1 − e−
=
1 1 − e − n
=
e −1 e −1
a (1 − r n )
Sn =
1− r
a
S =
1− r
Given that
1
x = 0 x
−x
1
(i) qx =
−x
k|
− x −1
(ii) ex =
2
Solution
n
Sn = (a + l )
2
A 100 percentile of survival time is the time t such that there is a 100 % probability
that survival time is less than t. In other words, it is t such that t q x = or t p x = 1 − . The
median or the 50th percentile of future lifetime at age x is the value of t such that
t p x = t q x = 0 .50 . In other words, the median time-to-failure m is the solution to
Determine the median future lifetime of a life age x, under each the following forces of
mortality assumptions
Solution
Determine the median future lifetime of a life age x, for each the following survival
functions:
x 1
(a) S 0 ( x ) = (1 − )2
(b) S 0 ( x ) = 1 − kx
1
(c) S 0 ( x ) =
1+ x
Solution
Example 7 (Percentiles)
1
Future lifetime of a life age x is subjected to the force of mortality x + t = . Determine
t + 50
the 75th percentile of future lifetime.
Solution
Future lifetime of a life age 45 is subjected to the force of mortality x = 0.0003 (1.065) x .
Determine the 60th percentile of future lifetime.
Solution
Recall that for both complete and curtate future lifetime, we have formulas expressing them
as the sums or integrals of probabilities of survival. We can actually obtain nice recursive
o o
formula to ease our calculation of e x , e x:n , e x and e x:n . Some of the useful recursive
formulas are given below
e x = e x:n + n p x e x + n
e x: n
= e x:m + m px e x + m:n −m
e x = p x (1 + e x +1 )
e x: n
= px (1 + e x +1:n −1 )
Example 9 (Recursions)
• e x = 24
Calculate e x + 3 .
Solution
Example 10 (Recursions)
Solution
Solution
When adopting a life table which may contain only integer ages, some assumptions are
needed about the distribution between integers
l x + t = (1 − t ) l x + t l x +1 or
l x+t = l x − t d x
t qx = t qx
qx
x +t =
1 − t qx
t p x x +t = q x
t qx
qx+s = 0 s +t 1
1 − s qx
t
o
e x = ex + 0.5
o
e x:n| = e x:n| + 0.5n q x
o
e x:n| = n n p x + n2 n q x
o
e x:1| = p x + 0.5q x
Example 12
(a) t q x = t q x
(b) t p x = 1 − t q x
qx
(c) x + t =
1 − t qx
(d) t p x x + t = q x
t qx
(e) qx+s = 0 s +t 1 (Hint: q x = s p x t q x + s = s p x − s +t p x )
1 − s qx
t s|t
Solution
l x + t = (l x )1−t ( l x +1 ) t or
ln l x + t = (1 − t ) ln l x + t ln l x +1
q x = 1 − (1 − q x ) t
t
x + t = = − ln p x
t p x x + t = e − t
t q x + s = 1 − e − t 0 s +t 1
Example 13
Prove that under Constant Force of Mortality assumption for fractional ages,
(a) t qx = 1 − (1 − qx )t
(b) t p x = ( p x ) t = e − t
(c) x + t = − ln p x
(d) t p x x + t = e − t
(e) t q x + s = 1 − e − t 0 s +t 1 (Hint: s|t q x = s p x t q x + s = s p x − s +t p x )
Solution
Solution
1 q x + 3 = 1 q x + 3 + 1 p x + 3 1 q x +1
(a) 2 4 4 4 4 4 4
1
qx 1
4 qx
1
= + − 4 qx +1
4
1
1 − 34 qx 1 − 34 qx
=
(b) 1 q x + 3 = 1 q x + 3 + 1 p x + 3 1 q x +1
2 4 4 4 4 4 4
• q x = 0.10 , q x +1 = 0.15
• The force of mortality is constant between integral ages
o o
Calculate q
0.3|0.5 x + 0.4 , e x:0.4| , and e x + 0.4:0.6|
Solution
o 0.4
e x:0.4|
= t px dt
0
0.4
= 1 − t qx dt
0
0.4
= 1 − [1 − (1 − q x )t ] dt
0
0.4
= (1 − q x )t dt
0
=
• q x = 0.10 , q x +1 = 0.20
• Deaths are uniformly distributed between integral ages
o o o o o
Calculate e x + 0.5:1 and e x:1.5 (Hint: e x: n
= e x:m + m px e x + m:n −m )
Solution
o 1
e x + 0.5:1
= t px + 0.5 dt
0
0 .5 1
= t p x +0.5 dt + t p x+0.5 dt
0 0 .5
0.5 0.5
= 1 − t q x + 0.5 dt + 0.5 p x + 0.5 ( t p x +1 ) dt
0 0
=
=
= 89
95
o o o
e x:1.5
= e x:0.5 + 0.5 px e x + 0.5:1 (how and why?)
= 1.3775
An individual applying for insurance coverage usually has different mortality than the
general population. This is because the insurer will screen the application to determine the
level of mortality risk for that individual. Some tests (such as a medical examination) may
be required before insurance is offered. If the insurer make the restrictions that individual
of poorer health are not accepted, then it is reasonable to say that the population selected
for life insurance coverage will have lower mortality probability than the original
population of all individuals. Those offered the coverage is said to have met the selection
criteria of the insurer. A life table constructed based on which the above characteristics are
taken into consideration is called a select and ultimate table.
Denote
p[ x ] = probability that a person who is underwritten and selected for insurance at age x
will survive for another year
q[ x ] = probability that a person who is underwritten and selected for insurance at age x
will die within the next year
The first step in developing a select and ultimate table is to determine the select period r.
In this case the impact of selection diminishes after the select period r.
q[ x ] q x , q[ x ]+1 q x +1 , q[ x ]+ 2 q x + 2 , and q[ x ]+ 3 = qx + 3
[x ] [ x] + 1 [ x] + 2 x + 3
30 q[ 30] q[ 30 ]+1 q[ 30]+ 2 q 33
31 q[ 31] q[ 31]+1 q[ 31]+ 2 q 34
32 q[ 32] q[ 32 ]+1 q[ 32]+ 2 q 35
l[ x ]+ n l[ x ] − l[ x ]+ n l[ x ]+ n − l[ x ]+ n + m
n p[ x ] = , n q[ x ] = , and n|m q[ x ] =
l[ x ] l[ x ] l[ x ]
From this we can construct a select and ultimate table with entries in terms of l
[x ] [ x] + 1 [ x] + 2 x+3
30 l[ 30 ] l[ 30]+1 l[ 30]+ 2 l 33
31 l[ 31] l[ 31]+1 l[ 31]+ 2 l 34
32 l[ 32 ] l[ 32]+1 l[ 32]+ 2 l 35
Consider a select and ultimate table with a 2-year select period. Express the following in
terms of l function.
Solution
Consider a select and ultimate table with a 3-year select period. Express the following in
terms of l function.
Solution
l[ 40]+ 2
(a) 2 p[ 40] =
l[ 40]
l[ 41]+1 − l[ 41]+ 2
(b) 1| q[ 41] =
l[ 41]
l[ 40 ]+ 2 − l[ 40 ]+ 3 l[ 40]+ 2 − l 43
(c) 1| q[ 40]+1 = =
l[ 40 ]+1 l[ 40]+1
l[ 41]+ 2 − l 45
(d) 2 q[ 41]+ 2 =
l[ 41]+ 2
Consider a select and ultimate table with a 2-year select period shown below.
(a) 2 p[ 31] (b) 2 p[ 30]+1 (c) 1| q[ 31] (d) 1| q[30]+1 (e) 2 q[ 33] (f) 2 q[ 31]+ 2
Solution
l[ 30 ]+ 3 l33 988
(b) 2 p[ 30]+1 = = =
l[ 30 ]+1 l[ 30 ]+1 998
Calculate q
2|3 [ 41] .
Solution
• e[ 60] = 25
• q60 = 0.01 , q61 = 0.02
• q[ x ] = 0.6qx
Solution