stt455 - Actuarial Models PDF

STT 455-6: Actuarial Models
Albert Cohen
Actuarial Sciences Program
Department of Mathematics
Department of Statistics and Probability
A336 Wells Hall
Michigan State University
East Lansing MI
48823
albert@math.msu.edu
acohen@stt.msu.edu
Albert Cohen (MSU) STT 455-6: Actuarial Models MSU 2011-12 1 / 283
Copyright Acknowledgement
Many examples and theorem proofs in these slides, and on in class exam
preparation slides, are taken from our textbook Actuarial Mathematics for
Life Contingent Risks by Dickson,Hardy, and Waters.
Please note that Cambridge owns the copyright for that material.
No portion of the Cambridge textbook material may be reproduced
in any part or by any means without the permission of the
publisher. We are very thankful to the publisher for allowing posting
of these notes on our class website.
Also, we will from time-to-time look at problems from released
previous Exams MLC by the SOA. All such questions belong in
copyright to the Society of Actuaries, and we make no claim on
them. It is of course an honor to be able to present analysis of such
examples here.
Survival Models
An insurance policy is a contract where the policyholder pays a
premium to the insurer in return for a benet or payment later.
The contract species what event the payment is contingent on. This
event may be random in nature
Assume that interest rates are deterministic, for now
Consider the case where an insurance company provides a benet
upon death of the policyholder. This time is unknown, and so the
issuer requires, at least, a model of of human mortality
Survival Models
Dene (x) as a human at age x. Also, dene that persons future lifetime
as the continuous random variable T
x
. This means that x + T
x
represents
that persons age at death.
Dene the lifetime distribution
F
x
(t) = P[T
x
t] (1)
the probabiliity that (x) does not survive beyond age x + t years, and its
complement, the survival function S
x
(t) = 1 F
x
(t).
Conditional Equivalence
We have an important conditional relationship
P[T
x
t] = P[T
0
x + t [ T
0
> x]
=
P[x < T
0
x + t]
P[T
0
> x]
(2)
and so
F
x
(t) =
F
0
(x + t) F
0
(x)
1 F
0
(x)
S
x
(t) =
S
0
(x + t)
S
0
(x)
(3)
In general we can extend this to
S
x
(t + u) = S
x
(t)S
x+t
(u) (4)
P[T
x
t] = P[T
0
x + t [ T
0
> x]
=
P[x < T
0
x + t]
P[T
0
> x]
(2)
and so
F
x
(t) =
F
0
(x + t) F
0
(x)
1 F
0
(x)
S
x
(t) =
S
0
(x + t)
S
0
(x)
(3)
S
x
(t + u) = S
x
(t)S
x+t
(u) (4)
P[T
x
t] = P[T
0
x + t [ T
0
> x]
=
P[x < T
0
x + t]
P[T
0
> x]
(2)
and so
F
x
(t) =
F
0
(x + t) F
0
(x)
1 F
0
(x)
S
x
(t) =
S
0
(x + t)
S
0
(x)
(3)
S
x
(t + u) = S
x
(t)S
x+t
(u) (4)
Conditions and Assumptions
Conditions on S
x
(t)
S
x
(0) = 1
lim
t
S
x
(t) = 0 for all x 0
S
x
(t
1
) S
x
(t
2
) for all t
1
t
2
and x 0
Assumptions on S
x
(t)
d
dt
S
x
(t) exists t R
+
lim
t
t S
x
(t) = 0 for all x 0
lim
t
t
2
S
x
(t) = 0 for all x 0
The last two conditions ensure that E[T
x
] and E[T
2
x
] exist, respectively.
Example 2.1
Assume that F
0
(t) = 1
_
1
t
120
_1
6
for 0 t 120. Calculate the
probability that
(0) survives beyond age 30
(30) dies before age 50
P[(0) survives beyond age 30] = S
0
(30) = 1 F
0
(30)
=
_
1
30
120
_1
6
= 0.9532
P[(30) dies before age 50] = F
30
(20)
=
F
0
(50) F
0
(30)
1 F
0
(30)
= 0.0410
40
(25) =
S
0
(65)
S
0
(40)
= 0.9395
(5)
Example 2.1
Assume that F
0
(t) = 1
_
1
t
120
_1
6
for 0 t 120. Calculate the
probability that
(30) dies before age 50
0
(30) = 1 F
0
(30)
=
_
1
30
120
_1
6
= 0.9532
P[(30) dies before age 50] = F
30
(20)
=
F
0
(50) F
0
(30)
1 F
0
(30)
= 0.0410
40
(25) =
S
0
(65)
S
0
(40)
= 0.9395
(5)
The Force of Mortality
Recall from basic probability that the density of F
x
(t) is dened as
f
x
(t) :=
d
dt
F
x
(t).
It follows that
f
0
(x) :=
d
dx
F
0
(x) = lim
dx0
+
F
0
(x + dx) F
0
(x)
dx
= lim
dx0
+
P[x < T
0
x + dx]
dx
(6)
Recall from basic probability that the density of F
x
(t) is dened as
f
x
(t) :=
d
dt
F
x
(t).
It follows that
f
0
(x) :=
d
dx
F
0
(x) = lim
dx0
+
F
0
(x + dx) F
0
(x)
dx
= lim
dx0
+
P[x < T
0
x + dx]
dx
(6)
However, we can nd the conditional density, also known as the Force of
Mortality via
x
= lim
dx0
+
P[x < T
0
x + dx [ T
0
> x]
dx
= lim
dx0
+
P[T
x
dx]
dx
= lim
dx0
+
1 S
x
(dx)
dx
= lim
dx0
+
1 S
x
(dx)
dx
= lim
dx0
+
1
S
0
(x+dx)
S
0
(x)
dx
=
1
S
0
(x)
lim
dx0
+
S
0
(x) S
0
(x + dx)
dx
=
1
S
0
(x)
d
dx
S
0
(x) =
f
0
(x)
S
0
(x)
(7)
In general, we can show
x+t
=
1
S
x
(t)
d
dt
S
x
(t) =
f
x
(t)
S
x
(t)
(8)
and integration of this relation leads to
S
x
(t) =
S
0
(x + t)
S
0
(x)
=
e
x+t
0

s
ds
e
x
0

s
ds
= e
x+t
x

s
ds
= e
t
0

x+s
ds
(9)
Example 2.2
Assume that F
0
(t) = 1
_
1
t
120
_1
6
for 0 t 120. Calculate
x
d
dx
S
0
(x) =
1
6

_
1
x
120
_
5
6
1
120
_

x
=
1
_
1
x
120
_1
6
_
1
6

_
1
x
120
_
5
6
1
120
__
=
1
720 6x
(10)
Example 2.2
Assume that F
0
(t) = 1
_
1
t
120
_1
6
for 0 t 120. Calculate
x
d
dx
S
0
(x) =
1
6

_
1
x
120
_
5
6
1
120
_

x
=
1
_
1
x
120
_1
6
_
1
6

_
1
x
120
_
5
6
1
120
__
=
1
720 6x
(10)
Gompertz Law / Makehamss Law
One model of human mortality, postulated by Gompertz, is
x
= Bc
x
,
where (B, c) (0, 1) (1, ). This is based on the assumption that
mortality is age dependent, and that the growth rate for mortality is
proportional to its own value. Makeham proposed that there should also
be an age independent component, and so Makehams Law is
x
= A + Bc
x
(11)
Of course, when A = 0, this reduces back to Gompertz Law.
By denition,
S
x
(t) = e
x+t
x

s
ds
= e
x+t
x
(A+Bc
s
)ds
= e
At
B
ln c
c
x
(c
t
1)
(12)
Keep in mind that this is a multivariable function of (x, t) R
2
+
Some online resources:
CDC National Vital Statistics report, Dec. 2002 .
Of course, when A = 0, this reduces back to Gompertz Law.
By denition,
S
x
(t) = e
x+t
x

s
ds
= e
x+t
x
(A+Bc
s
)ds
= e
At
B
ln c
c
x
(c
t
1)
(12)
Keep in mind that this is a multivariable function of (x, t) R
2
+
Some online resources:
CDC National Vital Statistics report, Dec. 2002 .
Comparison with US Govt data
Actuarial Notation
Actuaries make the notational conventions
t
p
x
= P[T
x
> t] = S
x
(t)
t
q
x
= P[T
x
t] = F
x
(t)
u|t
q
x
= P[u < T
x
u + t] = S
x
(u) S
x
(u + t)
(13)
u|t
q
x
, also known as the deferred mortality probability, is the probability
that (x) survives u years, and then dies in the subsequent t years.
Another convention is that p
x
:=
1
p
x
and q
x
:=
1
q
x
.
Actuarial Notation-Relationships
Consequently,
t
p
x
+
t
q
x
= 1
u|t
q
x
=
u
p
x

u+t
p
x
t+u
p
x
=
t
p
x

u
p
x+t
x
=
1
x
p
0
d
dx
(
x
p
0
)
(14)
Similarly,
x+t
=
1
t
p
x
d
dt
t
p
x

d
dt
t
p
x
=
x+t

t
p
x
x+t
=
f
x
(t)
S
x
(t)
f
x
(t) =
x+t

t
p
x
t
p
x
= e
t
0

x+s
ds
(15)
Also, since F
x
(t) =
_
t
0
f
x
(s)ds, we have as a linear approximation
t
q
x
=
_
t
0
s
p
x

x+s
ds
q
x
=
_
1
0
s
p
x

x+s
ds
=
_
1
0
e
s
0

x+v
dv

x+s
ds
_
1
0
x+s
ds

x+
1
2
(16)
Also, since F
x
(t) =
_
t
0
f
x
(s)ds, we have as a linear approximation
t
q
x
=
_
t
0
s
p
x

x+s
ds
q
x
=
_
1
0
s
p
x

x+s
ds
=
_
1
0
e
s
0

x+v
dv

x+s
ds
_
1
0
x+s
ds

x+
1
2
(16)
Mean and Standard Deviation of T
x
Actuaries make the notational denition e
x
:= E[T
x
], also known as the
complete expectation of life. Recall f
x
(t) =
t
p
x

x+t
=
d
dt
t
p
x
, and
e
x
=
_

0
t f
x
(t)dt
=
_

0
t
t
p
x

x+t
dt
=
_

0
t
d
dt
t
p
x
dt
=
_

0
t
p
x
dt
E[T
2
x
] =
_

0
t
2
f
x
(t)dt =
_

0
2t
t
p
x
dt
V [T
x
] := E[T
2
x
] (e
x
)
2
(17)
x
x
:= E[T
x
x
(t) =
t
p
x

x+t
=
d
dt
t
p
x
, and
e
x
=
_

0
t f
x
(t)dt
=
_

0
t
t
p
x

x+t
dt
=
_

0
t
d
dt
t
p
x
dt
=
_

0
t
p
x
dt
E[T
2
x
] =
_

0
t
2
f
x
(t)dt =
_

0
2t
t
p
x
dt
V [T
x
] := E[T
2
x
] (e
x
)
2
(17)
x
x
:= E[T
x
x
(t) =
t
p
x

x+t
=
d
dt
t
p
x
, and
e
x
=
_

0
t f
x
(t)dt
=
_

0
t
t
p
x

x+t
dt
=
_

0
t
d
dt
t
p
x
dt
=
_

0
t
p
x
dt
E[T
2
x
] =
_

0
t
2
f
x
(t)dt =
_

0
2t
t
p
x
dt
V [T
x
] := E[T
2
x
] (e
x
)
2
(17)
x
x
:= E[T
x
x
(t) =
t
p
x

x+t
=
d
dt
t
p
x
, and
e
x
=
_

0
t f
x
(t)dt
=
_

0
t
t
p
x

x+t
dt
=
_

0
t
d
dt
t
p
x
dt
=
_

0
t
p
x
dt
E[T
2
x
] =
_

0
t
2
f
x
(t)dt =
_

0
2t
t
p
x
dt
V [T
x
] := E[T
2
x
] (e
x
)
2
(17)
Example 2.6
Assume that F
0
(x) = 1
_
1
x
120
_1
6
for 0 x 120. Calculate e
x
, V [T
x
]
for a.)x = 30 and b.)x = 80.
Since S
0
(x) =
_
1
x
120
_1
6
, it follows that in keeping with the model where
survival is constrained to be les than 120,
t
p
x
=
S
0
(x + t)
S
0
(x)
=
_
_
_
_
1
t
120x
_1
6
: x + t 120
0 : x + t > 120
Example 2.6
Assume that F
0
(x) = 1
_
1
x
120
_1
6
for 0 x 120. Calculate e
x
, V [T
x
]
for a.)x = 30 and b.)x = 80.
Since S
0
(x) =
_
1
x
120
_1
6
, it follows that in keeping with the model where
survival is constrained to be les than 120,
t
p
x
=
S
0
(x + t)
S
0
(x)
=
_
_
_
_
1
t
120x
_1
6
: x + t 120
0 : x + t > 120
Example 2.6
So,
e
x
=
_
120x
0
_
1
t
120 x
_1
6
dt =
6
7
(120 x)
E[T
2
x
] =
_
120x
0
2t
_
1
t
120 x
_1
6
dt
=
_
6
7

6
13
_
2(120 x)
2
(18)
and
(e
30
, e
80
) = (77.143, 34.286)
(V[T
30
], V[T
80
]) =
_
(21.396)
2
, (9.509)
2
_ (19)
Exam MLC Spring 2007: Q8
Kevin and Kira excel at the newest video game at the local arcade,
Reversion. The arcade has only one station for it. Kevin is playing. Kira is
next in line. You are given:
(i) Kevin will play until his parents call him to come home.
(ii) Kira will leave when her parents call her. She will start playing as
soon as Kevin leaves if he is called rst.
(iii) Each child is subject to a constant force of being called: 0.7 per
hour for Kevin; 0.6 per hour for Kira.
(iv) Calls are independent.
(v) If Kira gets to play, she will score points at a rate of 100,000 per
hour.
Calculate the expected number of points Kira will score before she leaves.
(A) 77,000
(B) 80,000
(C) 84,000
(D) 87,000
(E) 90,000
Dene
t
p
x
= P[Kevin still there]
t
p
y
= P[Kira still there]
(20)
and so
E[Kiras playing time] =
_

0
(1
t
p
x
)
t
p
y
dt
=
_

0
_
1 e
0.7t
_
e
0.6t
dt
=
_

0
_
e
0.6t
e
1.3t
_
dt
=
1
0.6

1
1.3
= 0.89744 hrs.
(21)
It follows that
E[Kiras winnings] = 100000
$
hr
E[Kiras playing time]
= $89744.
(22)
Hence, we choose (E).
Numerical Considerations for T
x
In general, computations for the mean and SD for T
x
will require
numerical integration. For example,
Table: 2.1: Gompertz Model Statistics: (B, c) = (0.0003, 1.07)
x e
x
SD[T
x
] x +e
x
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 34.252 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
Curtate Future Lifetime
Dene
K
x
:= T
x
| (23)
and so
P[K
x
= k] = P[k T
x
< k + 1]
=
k|
q
x
=
k
p
x

k+1
p
x
=
k
p
x

k
p
x
p
x+k
=
k
p
x
q
x+k
(24)
E[K
x
] := e
x
=
k=0
k P[K
x
= k]
=
k=0
k (
k
p
x

k+1
p
x
)
=
k=1
k
p
x
by telescoping series..
E
_
K
2
x
k=0
k
2
P[K
x
= k]
= 2
k=1
k
k
p
x

k=1
k
p
x
= 2
k=1
k
k
p
x
e
x
(25)
E[K
x
] := e
x
=
k=0
k P[K
x
= k]
=
k=0
k (
k
p
x

k+1
p
x
)
=
k=1
k
p
x
by telescoping series..
E
_
K
2
x
k=0
k
2
P[K
x
= k]
= 2
k=1
k
k
p
x

k=1
k
p
x
= 2
k=1
k
k
p
x
e
x
(25)
Relationship between e
x
and e
x
Recall
e
x
=
_

0
t
p
x
dt =
j =0
_
j +1
j
t
p
x
dt (26)
By trapezoid rule for numerical integration, we obtain
_
j +1
j
t
p
x
dt
1
2
(
j
p
x
+
j +1
p
x
), and so
e
x

j =0
1
2
(
j
p
x
+
j +1
p
x
)
=
1
2
+
j =1
j
p
x
=
1
2
+ e
x
(27)
As with all numerical schemes, this approximation can be rened when
necessary.
Comparison of e
x
and e
x
Approximation matches well for small values of x
Table: 2.2: Gompertz Model Statistics: (B, c) = (0.0003, 1.07)
x e
x
e
x
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 13.55
80 8.354 8.848
90 4.944 5.433
100 2.673 3.152
Notes
An extension of Gompertz - Makeham Laws is the GM(r , s) formula
x
= h
1
r
(x) + e
h
2
s
(x)
, where h
1
r
(x), h
2
s
(x) are polynomials of degree r
and s, respectively.
Hazard rate in survival analysis and failure rate in reliability theory is
the same as what actuaries call force of mortality.
Homework Questions
HW: 2.1, 2.2, 2.5, 2.6, 2.10, 2.13, 2.14, 2.15
Life Tables
Dene for a model with maximum age and initial age x
0
the radix l
x
0
,
where
l
x
0
+t
= l
x
0

t
p
x
0
(28)
Life Tables
It follows that
l
x+t
= l
x
0

x+tx
0
p
x
0
= l
x
0

xx
0
p
x
0

t
p
x
= l
x

t
p
x
t
p
x
=
l
x+t
l
x
(29)
We assume a binomial model where L
t
is the number of survivors to age
x + t.
Life Tables
It follows that
l
x+t
= l
x
0

x+tx
0
p
x
0
= l
x
0

xx
0
p
x
0

t
p
x
= l
x

t
p
x
t
p
x
=
l
x+t
l
x
(29)
We assume a binomial model where L
t
is the number of survivors to age
x + t.
Life Tables
So, if there are l
x
independent individuals aged x with probability
t
p
x
of
survival to age x + t, then we interpret l
x+t
as the expected number of
survivors to age x + t out of l
x
independent individuals aged x.
Symbolically,
E[L
t
[ L
0
= l
x
] = l
x+t
= l
x

t
p
x
(30)
Also, dene the expected number of deaths from year x to year x + 1 as
d
x
:= l
x
l
x+1
= l
x

_
1
l
x+1
l
x
_
= l
x
(1 p
x
) = l
x
q
x
(31)
Life Tables
So, if there are l
x
independent individuals aged x with probability
t
p
x
of
survival to age x + t, then we interpret l
x+t
as the expected number of
survivors to age x + t out of l
x
independent individuals aged x.
Symbolically,
E[L
t
[ L
0
= l
x
] = l
x+t
= l
x

t
p
x
(30)
Also, dene the expected number of deaths from year x to year x + 1 as
d
x
:= l
x
l
x+1
= l
x

_
1
l
x+1
l
x
_
= l
x
(1 p
x
) = l
x
q
x
(31)
Example 3.1
Table: 3.1: Extract from a life table
x l
x
d
x
30 10000.00 34.78
31 9965.22 38.10
32 9927.12 41.76
33 9885.35 45.81
34 9839.55 50.26
35 9789.29 55.17
36 9734.12 60.56
37 9673.56 66.49
38 9607.07 72.99
39 9534.08 80.11
Example 3.1
Calculate:
a.) l
40
b.)
10
p
30
c.) q
35
d.)
5
q
30
e.) P[(30) dies between age 35 and 36]
Example 3.1
Calculate:
a.) l
40
= l
39
d
39
= 9453.97
b.)
10
p
30
=
l
40
l
30
= 0.94540
c.) q
35
=
d
35
l
35
= 0.00564
d.)
5
q
30
=
l
30
l
35
l
30
= 0.02107
e.) P[(30) dies between age 35 and 36] =
l
35
l
36
l
30
= 0.00552
Example 3.1
Calculate:
a.) l
40
= l
39
d
39
= 9453.97
b.)
10
p
30
=
l
40
l
30
= 0.94540
c.) q
35
=
d
35
l
35
= 0.00564
d.)
5
q
30
=
l
30
l
35
l
30
= 0.02107
l
35
l
36
l
30
= 0.00552
Example 3.1
Calculate:
a.) l
40
= l
39
d
39
= 9453.97
b.)
10
p
30
=
l
40
l
30
= 0.94540
c.) q
35
=
d
35
l
35
= 0.00564
d.)
5
q
30
=
l
30
l
35
l
30
= 0.02107
l
35
l
36
l
30
= 0.00552
Example 3.1
Calculate:
a.) l
40
= l
39
d
39
= 9453.97
b.)
10
p
30
=
l
40
l
30
= 0.94540
c.) q
35
=
d
35
l
35
= 0.00564
d.)
5
q
30
=
l
30
l
35
l
30
= 0.02107
l
35
l
36
l
30
= 0.00552
Example 3.1
Calculate:
a.) l
40
= l
39
d
39
= 9453.97
b.)
10
p
30
=
l
40
l
30
= 0.94540
c.) q
35
=
d
35
l
35
= 0.00564
d.)
5
q
30
=
l
30
l
35
l
30
= 0.02107
l
35
l
36
l
30
= 0.00552
Fractional Age Assumptions
So far, the life table approach has mirrored the survival distribution
method we encountered in the previous lecture. However, in detailing the
life table, no information is presented on the cohort in between whole
years. To account for this, we must make some fractional age
assumptions. The following are equivalent:
UDD1 For all (x, s) N [0, 1), we assume that
s
q
x
= s q
x
UDD2 For all x N, we assume
R
x
:= T
x
K
x
U(0, 1)
R
x
is independent of K
x
.
Fractional Age Assumptions
So far, the life table approach has mirrored the survival distribution
method we encountered in the previous lecture. However, in detailing the
life table, no information is presented on the cohort in between whole
years. To account for this, we must make some fractional age
assumptions. The following are equivalent:
UDD1 For all (x, s) N [0, 1), we assume that
s
q
x
= s q
x
UDD2 For all x N, we assume
R
x
:= T
x
K
x
U(0, 1)
R
x
is independent of K
x
.
Proof of Equivalence
Proof:
UDD1 UDD2: Assume for all (x, s) N [0, 1), we assume that
s
q
x
= s q
x
. Then
P[R
x
s] =
k=0
P[R
x
s, K
x
= k]
=
k=0
P[k T
x
k + s]
=
k=0
k
p
x

s
q
x+k
=
k=0
k
p
x
s q
x+k
= s
k=0
k
p
x
q
x+k
= s
k=0
P[K
x
= k] = s
(32)
and so R
x
U(0, 1).
Proof of Equivalence
To show independence of R
x
and K
x
,
P[R
x
s, K
x
= k] = P[k T
x
k + s]
=
k
p
x

s
q
x+k
= s
k
p
x
q
x+k
= P[R
x
s] P[K
x
= k]
(33)
UDD2 UDD1: Assuming UDD2 is true, then for (x, s) N [0, 1)
we have
s
q
x
= P[T
x
s]
= P[K
x
= 0, R
x
s]
= P[R
x
s] P[K
x
= 0]
= s q
x
(34)
QED
Corollary
Recall that
s
q
x
=
l
x
l
x+s
l
x
. It follows now that
s
q
x
= sq
x
= s
d
x
l
x
=
l
x
l
x+s
l
x
l
x+s
= l
x
s d
x
which is a linear decreasing function of s [0, 1)
q
x
=
d
ds
[
s
q
x
] = f
x
(s) =
s
p
x

x+s
(35)
But, since q
x
is constant in s, we have f
x
(s) is constant for s [0, 1).
Read over Examples 3.2 3.5
Corollary
Recall that
s
q
x
=
l
x
l
x+s
l
x
s
q
x
= sq
x
= s
d
x
l
x
=
l
x
l
x+s
l
x
l
x+s
= l
x
s d
x
q
x
=
d
ds
[
s
q
x
] = f
x
(s) =
s
p
x

x+s
(35)
But, since q
x
x
Corollary
Recall that
s
q
x
=
l
x
l
x+s
l
x
s
q
x
= sq
x
= s
d
x
l
x
=
l
x
l
x+s
l
x
l
x+s
= l
x
s d
x
q
x
=
d
ds
[
s
q
x
] = f
x
(s) =
s
p
x

x+s
(35)
But, since q
x
x
Constant Force of Mortality for Fractional Age
For all (x, s) N[0, 1), we assume that
x+s
does not depend on s, and
we denote
x+s
:=
x
. It follows that
p
x
= e
1
0

x+s
ds
= e
x
s
p
x
= e
s
0

x
du
= e
x
s
= (p
x
)
s
s
p
x+t
= e
s
0

x
du
= (p
x
)
s
when t + s < 1
q
x
= 1 e
x
=
k=1
(1)
k+1
(
x
)
k
k!

x
s
q
x
= 1 e
x
t

x
t,
(36)
where the last two lines assume
x
1.
Read Examples 3.6, 3.7 and Sections 3.4, 3.5, 3.6.
Homework Questions
HW: 3.1, 3.2, 3.4, 3.7, 3.8, 3.9, 3.10
Contingent Events
We have spent the previous two lectures on modeling human mortality.
The need for such models in insurance pricing arises when designing
contracts that are event-contingent. Such events include reaching
retirement before the end of the underlying life (x) .
However, one can also write contracts that are dependent on a life (x)
being admitted to college (planning for school), and also on (x)
s
external
portfolios maintaining a minimal value over a time-interval (insuring
external investments.)
Contingent Events: General Case
Consider a probability space (, T, P) and an event A T.
If we are working with a force of interest
s
() and the time of event W
as
W
, then we have under the stated probability measure P the Expected
Present Value of a payo K() contingent upon W
EPV = E[K()e
W
0

s
()ds
] (37)
Actuarial Encounters of the Third Kind !!
Some Initial Simplifying Assumptions
K() = 1 for all (a.s.)
s
() = for all (a.s.)
W := event that (x) dies
W
:= T
x
P is obtained via historical observation and is thus a physical
measure. Specically, we use
t
p
x
obtained from life tables or via
models of human mortality
We do not assume now that a unique risk-neutral pricing measure

P
exists.
Standard Ultimate Survival Model with assumes Makehams law
with (A, B, c) = (0.00022, 2.7 10
6
, 1.124)
Recall...
The equivalent interest rate i := e
1 per year
The discount factor v :=
1
1+i
= e
per year
The nominal interest rate i
(p)
= p
_
(1 + i )
1
p
1
_
compounded p
times per year
The eective rate of discount d := 1 v = i v = 1 e
per year
The nominal rate of discount d
(p)
:= p
_
1 v
1
p
_
compounded p
times per year
Whole Life Insurance: Continuous Case
Consider now the random variable
Z = v
T
x
= e
T
x
(38)
which represents the present value of a dollar upon death of (x). We are
interested in statistical measures of this quantity:
E[Z] =

A
x
:= E[e
T
x
] =
_

0
e
t
t
p
x
x+t
dt
E[Z
2
] =
2
A
x
:= E[e
2T
x
]
=
_

0
e
2t
t
p
x
x+t
dt
Var (Z) =
2
A
x

_
A
x
_
2
P[Z z] = P
_
T
x

ln (z)
_
(39)
Whole Life Insurance: Yearly Case
Assuming payments are made at the end of the death year, our random
variable is now Z = v
K
x
+1
= e
K
x
and so
E[Z] = A
x
:= E[v
K
x
+1
] =
k=0
v
k+1
P[K
x
= k]
=
k=0
v
k+1
k|
q
x
E[Z
2
] =
k=0
v
2k+2
k|
q
x
Var (Z) =
2
A
x
(A
x
)
2
P[Z z] = P
_
K
x

ln (z)
_
(40)
Whole Life Insurance:
1
m
thly
Case
Instead of only paying at the end of the last whole year lived, an insurance
contract might specify payment upon the end of the last period lived. In
this case, if we split a year into m periods, and dene
K
(m)
x
=
1
m
mT
x
| (41)
For example, if K
x
= 19.78, then
K
(m)
x
=
_
_
19 m = 1
19
1
2
= 19.5 m = 2
19
3
4
= 19.75 m = 4
19
9
12
= 19.75 m = 12
1
m
thly
Case
It follows that we need r
_
0,
1
m
,
2
m
, ...,
m1
m
, 1,
m+1
m
, ...
_
P
_
K
(m)
x
= r
_
= P
_
r T
x
< r +
1
m
_
=
r |
1
m
q
x
(42)
to compute statistics for our random variable
Z = v
K
(m)
x
+
1
m
(43)
1
m
thly
Case
E[Z] = A
(m)
x
:= E[v
K
(m)
x
+
1
m
] =
k=0
v
k+1
m
k
m
|
1
m
q
x
E[Z
2
] =
2
A
(m)
x
=
k=0
v
2k+2
m
k
m
|
1
m
q
x
Var (Z) =
2
A
(m)
x

_
A
(m)
x
_
2
P[Z z] = P
_
K
(m)
x

ln (z)

1
m
_
(44)
Recursion Method
One of the computational tools we share directly with quantitative nance
is the method of backwards-pricing. In option pricing, we assume the
contract has a nite term. Here, we assume a nite lifetime maximum of
< . It follows that
A
1
= E
_
v
K
1
+1
_
= E
_
v
1
= v (45)
At age 2, we have P[K
2
= 0] = q
2
and so
A
2
= E
_
v
K
2
+1
_
= q
2
v + p
2
E
_
v
(1+K
1
)+1
_
= q
2
v + p
2
v E
_
v
K
1
+1
_
= q
2
v + p
2
v
2
(46)
Recursion Method
One of the computational tools we share directly with quantitative nance
is the method of backwards-pricing. In option pricing, we assume the
contract has a nite term. Here, we assume a nite lifetime maximum of
< . It follows that
A
1
= E
_
v
K
1
+1
_
= E
_
v
1
= v (45)
At age 2, we have P[K
2
= 0] = q
2
and so
A
2
= E
_
v
K
2
+1
_
= q
2
v + p
2
E
_
v
(1+K
1
)+1
_
= q
2
v + p
2
v E
_
v
K
1
+1
_
= q
2
v + p
2
v
2
(46)
Recursion Method
In general, we have the recursion equation for a life (x) that satises
A
x
= vq
x
+ vp
x
A
x+1
A
1
= v
(47)
in the whole life case, and
A
(m)
x
= v
1
m
1
m
q
x
+ v
1
m
1
m
p
x
A
(m)
x+
1
m
A
(m)
1
m
= v
1
m
(48)
in the
1
m
thly
case.
Recursion Method
Recall that for Makehams law we have, respectively
1
m
p
x
= e
A
m
Bc
x
ln (c)
c
1
m
1
1
p
x
= e
A
Bc
x
ln (c)
(c1)
(49)
and for the power law of survival, S
0
(x) =
_
1
x
_
a
for some a, > 0
1
m
p
x
=
_
x
1
m
x
_
a
1
p
x
=
_
x 1
x
_
a
(50)
Recursion Method
For any positive integer m, it follows that for Makehams law:
A
(m)
x
= v
1
m
_
1 e
A
m
Bc
x
ln (c)
c
1
m
1
_
+ v
1
m
e
A
m
Bc
x
ln (c)
c
1
m
1
A
(m)
x+
1
m
A
(m)
1
m
= v
1
m
(51)
Recursion Method
For any positive integer m, it follows that for Power law:
A
(m)
x
= v
1
m
_
1
_
x
1
m
x
_
a
_
+ v
1
m
_
x
1
m
x
_
a
A
(m)
x+
1
m
A
(m)
1
m
= v
1
m
(52)
HW Project: For Power law with a =
3
5
and = 101
Generate a spreadsheet like Table 4.1 in the text, including values for
2
A
(m)
x
Repeat Example 4.3 with the Power law model
Term Insurance: Continuous Case
Consider now the case where payment is made in the continuous case, and
death benet is payable to the policyholder only if T
x
n. Then, we are
interested in the random variable
Z = e
T
x
1
{T
x
n}
(53)
and so
A
1
x:n
= E[Z] =
_
n
0
e
t
t
p
x
x+t
dt
2
A
1
x:n
= E
_
Z
2
=
_
n
0
e
2t
t
p
x
x+t
dt
(54)
Term Insurance:
1
m
thly
Case
Consider again the case where the death benet is payable at the end of
the
1
m
thly
period in the death year to the policyholder only if
K
(m)
x
+
1
m
n. Then, we are interested in the random variable
Z = e
(K
(m)
x
+
1
m
)
1
K
(m)
x
+
1
m
n
(55)
and so
A
(m)1
x:n
= E[Z] =
mn1
k=0
v
k+1
m
k
m
|
1
m
q
x
2
A
(m)1
x:n
= E
_
Z
2
=
mn1
k=0
v
2k+2
m
k
m
|
1
m
q
x
(56)
Pure Endowment
Pure endowment benets depend on the survival policyholder (x) until at
least age x + n. In such a contract, a xed benet of 1 is paid at time n.
This is expressed via
Z = e
n
1
{T
x
n}
A
1
x:n
= E[Z] = v
n
n
p
x
= e
n
n
p
x
(57)
Endowment Insurance
Endowment insurance is a combination of term insurance and pure
endowment. In such a policy, the amount is paid upon death if it occurs
with a xed term n. However, if (x) survives beyond n years, the sum
insured is payable at the end of the n
th
year. The corresponding present
value random variable is
Z = e
min{T
x
,n}
E[Z] =

A
x:n
=
_
n
0
e
t
t
p
x
x+t
dt + e
n
n
p
x
=

A
1
x:n
+ A
1
x:n
(58)
This can be generalized once again to the
1
m
thly
case and for E[Z
2
]
Deferred Insurance Benets
Suppose policyholder on a life (x) receives benet 1 if u T
x
< u + n.
Then
Z = e
T
x
1
{uT
x
<u+n}
E[Z] =
u|
A
1
x:n
=
_
u+n
u
e
t
t
p
x
x+t
dt
=
_
n
0
e
(s+u)
s+u
p
x
x+s+u
ds
= e
u
_
n
0
e
s
u
p
x

s
p
x+u
x+s+u
ds
= e
u
u
p
x
A
1
x+u:n
=

A
1
x:u+n

A
1
x:u
(59)
Relationships
By denition, we have
A
x
= A
1
x:n
+
n|
A
x
= A
1
x:n
+ v
n
n
p
x
A
x+n
(60)
What about relationship between

A
x
and A
x
?
Relationships
By denition, we have
A
x
= A
1
x:n
+
n|
A
x
= A
1
x:n
+ v
n
n
p
x
A
x+n
(60)
What about relationship between

A
x
and A
x
?
Employing UDD:

A
x
vs A
x
If expected values are computed via information derived from life tables,
then certainly

A
x
must be approximated using techniques from previous
lecture.
Recall that by the denition of
s
p
x
and the UDD, we have
s
p
x
x+s
= f
x
(s)
=
d
ds
P[T
x
s]
=
d
ds
(
s
q
x
) =
d
ds
(s q
x
)
= q
x
(61)
Employing UDD:

A
x
vs A
x
If expected values are computed via information derived from life tables,
then certainly

A
x
must be approximated using techniques from previous
lecture.
Recall that by the denition of
s
p
x
and the UDD, we have
s
p
x
x+s
= f
x
(s)
=
d
ds
P[T
x
s]
=
d
ds
(
s
q
x
) =
d
ds
(s q
x
)
= q
x
(61)
Employing UDD:

A
x
vs A
x
It follows that using the UDD approximation leads to
A
x
=
_

0
e
t
t
p
x
x+t
dt =
k=0
_
k+1
k
e
t
t
p
x
x+t
dt
=
k=0
k
p
x
v
k+1
_
1
0
e
e
s
s
p
x+k
x+k+s
ds
k=0
k
p
x
v
k+1
q
x+k

_
1
0
e
e
s
ds
=
k=0
v
k+1
P[K
x
= k]
_
1
0
e
e
s
ds = A
x

_
1
0
e
e
s
ds
= A
x

i
A
(m)
x

i
i
(m)
A
x
=
i
m
_
(1 + i )
1
m
1
_
A
x
(62)
Employing UDD:

A
x
vs A
x
It follows that using the UDD approximation leads to
A
x
=
_

0
e
t
t
p
x
x+t
dt =
k=0
_
k+1
k
e
t
t
p
x
x+t
dt
=
k=0
k
p
x
v
k+1
_
1
0
e
e
s
s
p
x+k
x+k+s
ds
k=0
k
p
x
v
k+1
q
x+k

_
1
0
e
e
s
ds
=
k=0
v
k+1
P[K
x
= k]
_
1
0
e
e
s
ds = A
x

_
1
0
e
e
s
ds
= A
x

i
A
(m)
x

i
i
(m)
A
x
=
i
m
_
(1 + i )
1
m
1
_
A
x
(62)
Claims Accelaration Approach : A
(m)
x
vs A
x
Consider now a policy that pays the holder at the end of the
1
m
thly
period
of death. In this case, the benet is paid at one of the times r where
r
_
K
x
+
1
m
, K
x
+
2
m
, ..., K
x
+
m
m
_
(63)
and so under the UDD,
E[T
payment
[ K
x
= k] =
m
j =1
1
m

_
k +
j
m
_
= k +
m + 1
2m
(64)
Claims Accelaration Approach : A
(m)
x
vs A
x
Once again, it follows using the UDD aproximation
A
(m)
x
= E[v
K
(m)
x
+
1
m
] =
k=0
v
k+1
m
k
m
|
1
m
q
x
k=0
v
E[T
payment
|K
x
=k]
P[K
x
= k]
=
k=0
v
k+
m+1
2m
k|
q
x
= (1 + i )
m1
2m
k=0
v
k+1
k|
q
x
= (1 + i )
m1
2m
A
x
(1 + i )
1
2
A
x
(65)
as m . Note that using the UDD approximation, both

A
x
A
x
and
A
(m)
x
A
x
are independent of x.
Variable Insurance Benets
Upon death, we have considered policies that pay the holder a xed
amount. What varied was the method and time of payment. If, however,
the actual payo amount depended on the time T
x
of death for (x), then
we term such a contract a Variable Insurance Contract.
Specically, if the payo amount dependent on T
x
is h(T
x
), then
Z = h(T
x
)e
T
x
E[Z] =
_

0
h(t)e
t
t
p
x
x+t
dt
_
I

A
_
x
:=
_

0
te
t
t
p
x
x+t
dt
_
I

A
_
1
x:n
:=
_
n
0
te
t
t
p
x
x+t
dt
(66)
Example 4.8
Consider an nyear term insurance issued to (x) under which the death
benet is paid at the end of the year of death. The death benet if death
occurs between ages x + k and x + k + 1 is valued at (1 + j )
k
. Hence,
using the denition i
:=
1+i
1+j
1,
Z = v
K
x
+1
(1 + j )
K
x
E[Z] =
n1
k=0
v
k+1
(1 + j )
k
k|
q
x
=
1
1 + j

n1
k=0
v
k+1
(1 + j )
k+1
k|
q
x
=
1
1 + j

n1
k=0
k|
q
x
_
1+i
1+j
_
k+1
=
1
1 + j
A
1
x:n
(67)
Homework Questions
HW: 4.1, 4.2, 4.3, 4.7, 4.9, 4.11, 4.12, 4.14, 4.15, 4.16, 4.17, 4.18
Life Annuities
A Life Annuity refers to a series of payments to or from an individual as
long as that person is still alive. For a xed rate i and term n , we recall
the deterministic pricing theory:
a
n i
= 1 + v + ... + v
n1
=
1 v
n
d
a
n i
= v + ... + v
n
= a
n i
1 + v
n
=
1 v
n
i
a
n i
=
_
n
0
v
t
dt =
1 v
n
a
(m)
n i
=
1
m

_
1 + v
1
m
+ ... + v
n
1
m
_
=
1 v
n
d
(m)
a
(m)
n i
=
1
m

_
v
1
m
+ ... + v
n
1
m
+ v
n
_
=
1 v
n
i
(m)
(68)
Whole Life Annuity Due
Consider the case where 1 is paid out at the beginning of every period
until death. Our present random variable is now
Y := a
K
x
+1
=
1 v
K
x
+1
d
(69)
and so
a
x
= E[Y] = E
_
1 v
K
x
+1
d
_
=
1 A
x
d
(70)
V[Y] = V
_
1 v
K
x
+1
d
_
=
1
d
2
V[1] +
1
d
2
V[v
K
x
+1
]
= 0 +
2
A
x
A
2
x
d
2
(71)
Whole Life Annuity Due
The present value random variable can also be represented as
Y =
k=0
v
k
1
{T
x
>k}
(72)
As P[T
x
> k] =
t
p
x
, we have the alternate expression for a
x
a
x
= E[Y] = E
_

k=0
v
k
1
{T
x
>k}
_
=
k=0
E[v
k
1
{T
x
>k}
]
=
k=0
v
k
k
p
x
=
k=0
k|
q
x
a
k+1
(73)
Term Annuity Due
Dene the present value random variable
Y =
_
a
K
x
+1
: K
x
0, 1, 2, ..., n 1
a
n
: K
x
n, n + 1, n + 2, ...
Another expression is
Y = a
min{K
x
+1,n}
=
1 v
min{K
x
+1,n}
d
(74)
and so
a
x:n
= E[Y] =
1 E
_
v
min{K
x
+1,n}
d
=
1 A
x:n
d
=
n1
t=0
v
t
t
p
x
=
n1
k=0
k|
q
x
a
k+1
+
n
p
x
a
n
(75)
Whole Life and Term Immediate Annuity
Dene Y
k=1
v
k
1
{T
x
>k}
. Then we have an annuity immediate that
begins payment one unit of time from now. It follows that
a
x
= a
x
1
V[Y
] = V[Y]
(76)
Also, if we dene Y = a
min{K
x
,n}
, then
a
x:n
=
n
t=1
v
t
t
p
x
= a
x:n
1 + v
n
n
p
x
(77)
Whole Life Continuous Annuity
Dene
Y = a
T
x
=
1 v
T
x
=
_

0
e
t
1
{T
x
>t}
dt
a
x
= E[Y] =
1
A
x
=
_

0
e
t
t
p
x
dt
(78)
Note that if = 0, then a
x
= e
x
Term Continuous Annuity
Dene Y = a
min{T
x
,n}
.
Then
Y =
1 v
min{T
x
,n}
a
x:n
= E[Y] =
1
A
x:n
=
_
n
0
e
t
t
p
x
dt
(79)
Deferred Annuity
Consider now the case of an annuity for (x) that will pay 1 at the end of
each year, beginning at age x + u and will continue until death age
x + T
x
. We dene
u|
a
x
to be the Expected Present Value of this policy. It
should be apparent that
u|
a
x
= a
x
a
x:u
=
t=u
v
t
t
p
x
= v
t
t
p
x

t=0
v
t
t
p
x
= v
t
t
p
x
a
x+u
(80)
holds in the discrete case, and similarly in the continuous case,
u|
a
x
= a
x
a
x:u
(81)
Term Deferred Annuity
For the cases of an annuity for (x) that will pay 1 at the end of each year,
beginning at age x + u and will continue until death age x + T
x
up to a
term of length n, or annuity-due payable
1
m
thly
. Then
u|
a
x:n
= v
u
u
p
x
a
x+u:n
u|
a
(m)
x
= v
u
u
p
x
a
(m)
x+u
(82)
respectively.
These combine with the previous slide to reveal the useful formulae:
a
x:n
= a
x
v
n
n
p
x
a
x+n
a
(m)
x:n
= a
(m)
x
v
n
n
p
x
a
(m)
x+n
(83)
Guaranteed Annuities
There are instances where an age (x) wishes to buy a policy where
payments are guaranteed to continue upon death to a beneciary. In this
case, dene the present random variable as Y = a
n
+ Y
1
, where
Y
1
=
_
0 : K
x
0, 1, 2, ..., n 1
a
K
x
+1
a
n
: K
x
n, n + 1, n + 2, ...
and so
E[Y
1
] = E
__
a
K
x
+1
a
n
_
1
{K
x
n}
_
=
n|
a
x
= v
n
n
p
x
a
x+n
E[Y] := a
x:n
= a
n
+ v
n
n
p
x
a
x+n
and E[Y
(m)
] := a
(m)
x:n
= a
(m)
n
+ v
n
n
p
x
a
(m)
x+n
(84)
Guaranteed Annuities
There are instances where an age (x) wishes to buy a policy where
payments are guaranteed to continue upon death to a beneciary. In this
case, dene the present random variable as Y = a
n
+ Y
1
, where
Y
1
=
_
0 : K
x
0, 1, 2, ..., n 1
a
K
x
+1
a
n
: K
x
n, n + 1, n + 2, ...
and so
E[Y
1
] = E
__
a
K
x
+1
a
n
_
1
{K
x
n}
_
=
n|
a
x
= v
n
n
p
x
a
x+n
E[Y] := a
x:n
= a
n
+ v
n
n
p
x
a
x+n
and E[Y
(m)
] := a
(m)
x:n
= a
(m)
n
+ v
n
n
p
x
a
(m)
x+n
(84)
Example 5.4
A pension plan member is entitled to a benet of 1000 per month, in
advance, for life from age 65, with no guarantee. She can opt to take a
lower benet with a 10year guarantee. The revised benet is calculated
to have equal EPV at age 65 to the original benet. Calculate the revised
benet using the Standard Ultimate Survival Model, with interest at 5%
per year.
Example 5.4
Let B denote the revised monthly benet. Then the two options are
12000 per year, paid per month with Present Value Y
1
12B per year, paid per month with Present Value Y
2
Hence E[Y
1
Y
2
] = 0 implies
12000a
(12)
65
= 12Ba
(12)
65:10
= 12B
_
a
(12)
10
+ v
10
10
p
65
a
(12)
75
_
B = 1000
a
(12)
65
a
(12)
10
+ v
10
10
p
65
a
(12)
75
= 1000
13.0870
13.3791
= 978.17
V[Y
1
Y
2
] = 0?
(85)
Example 5.4
1
2
Hence E[Y
1
Y
2
] = 0 implies
12000a
(12)
65
= 12Ba
(12)
65:10
= 12B
_
a
(12)
10
+ v
10
10
p
65
a
(12)
75
_
B = 1000
a
(12)
65
a
(12)
10
+ v
10
10
p
65
a
(12)
75
= 1000
13.0870
13.3791
= 978.17
V[Y
1
Y
2
] = 0?
(85)
Example 5.4
1
2
Hence E[Y
1
Y
2
] = 0 implies
12000a
(12)
65
= 12Ba
(12)
65:10
= 12B
_
a
(12)
10
+ v
10
10
p
65
a
(12)
75
_
B = 1000
a
(12)
65
a
(12)
10
+ v
10
10
p
65
a
(12)
75
= 1000
13.0870
13.3791
= 978.17
V[Y
1
Y
2
] = 0?
(85)
Linearly Increasing Annuities
Dene an annuity where the payments increase linearly at times
t = 0, 1, 2, .. provided that (x) is alive at time t
(I a)
x
=
t=0
(t + 1) v
t
t
p
x
(I a)
x:n
=
n1
t=0
(t + 1) v
t
t
p
x
(86)
Linearly Increasing Annuities
If then the annuity is payable continuously, with payments increasing by 1
at each year end and the rate of payment in the t
th
year constant and
equal to t for t 1, 2, ..m, .., n,
then h(t) = m1
{mt<m+1}
, and the EPV is
(I a)
x:n
=
n1
m=0
(m + 1)
m|
a
x:1
(87)
If h(t) = t, then
(
I a)
x:n
=
_
n
0
te
t
t
p
x
dt (88)
Evaluating Annuities Using Recursion
By recursion, we observe
a
x
= 1 + vp
x
+ v
2
2
p
x
+ v
3
3
p
x
+ ....
= 1 + vp
x
_
1 + vp
x+1
+ v
2
2
p
x+1
+ v
3
3
p
x+1
+ ....
_
= 1 + vp
x
a
x+1
a
(m)
x
=
1
m
+ v
1
m
1
m
p
x
a
(m)
x+
1
m
(89)
Consider the case where there is a maximum age in the model, and so
a
1
= 1
a
(m)
1
m
=
1
m
(90)
a
x
= 1 + vp
x
+ v
2
2
p
x
+ v
3
3
p
x
+ ....
= 1 + vp
x
_
1 + vp
x+1
+ v
2
2
p
x+1
+ v
3
3
p
x+1
+ ....
_
= 1 + vp
x
a
x+1
a
(m)
x
=
1
m
+ v
1
m
1
m
p
x
a
(m)
x+
1
m
(89)
a
1
= 1
a
(m)
1
m
=
1
m
(90)
a
x
= 1 + vp
x
+ v
2
2
p
x
+ v
3
3
p
x
+ ....
= 1 + vp
x
_
1 + vp
x+1
+ v
2
2
p
x+1
+ v
3
3
p
x+1
+ ....
_
= 1 + vp
x
a
x+1
a
(m)
x
=
1
m
+ v
1
m
1
m
p
x
a
(m)
x+
1
m
(89)
a
1
= 1
a
(m)
1
m
=
1
m
(90)
Evaluating Annuities Using UDD
Recall that under the UDD assumption,
A
(m)
x
=
i
i
(m)
A
x
A
x
=
i
A
x
(91)
and by denition,
a
x
=
1 A
x
d
a
(m)
x
=
1 A
(m)
x
d
(m)
a
x
=
1
A
x
(92)
It follows that
a
(m)
x
=
1 A
(m)
x
d
(m)
=
1
i
i
(m)
A
x
d
(m)
=
i
(m)
iA
x
i
(m)
d
(m)
=
i
(m)
i (1 da
x
)
i
(m)
d
(m)
=
id
i
(m)
d
(m)
a
x

i i
(m)
i
(m)
d
(m)
:= (m)a
x
(m)
a
x
=
id
2
a
x

i
2
(93)
For term annuities, we have
a
(m)
x:n
= a
(m)
x
v
n
n
p
x
a
(m)
x+n
= (m)a
x
(m) v
n
n
p
x
((m)a
x+n
(m))
= (m)
_
a
x
v
n
n
p
x
a
(m)
x+n
_
(m) (1 v
n
n
p
x
)
= (m) a
x:n
(m) (1 v
n
n
p
x
)
a
x:n

m 1
2m
(1 v
n
n
p
x
)
(94)
Woolhouses Formula
Consider a function g : R
+
R such that lim
t
g(t) = 0, then
_

0
g(t)dt = h
k=0
g(kh)
h
2
g(0) +
h
2
12
g
(0)
h
4
720
g
(0) + ... (95)

Woolhouses Formula
Dene
g(t) = v
t
t
p
x
g
(t) =
t
p
x
e
t
v
t
t
p
x
x+t
g
(0) =
x
(96)
and so for h = 1,
a
x

k=0
g(k)
1
2
+
1
12
g
(0)
=
k=0
v
k
k
p
x

1
2

1
12
( +
x
)
= a
x

1
2

1
12
( +
x
)
(97)
Woolhouses Formula
Correspondingly, for h =
1
m
,
a
x

1
m
k=0
g
_
k
m
_
1
2m
+
1
12m
2
g
(0)
=
k=0
v
k
m
k
m
p
x

1
2m

1
12m
2
( +
x
)
= a
(m)
x

1
2m

1
12m
2
( +
x
)
(98)
Woolhouses Formula
Equating the previous two approximations for a
x
, we obtain
a
(m)
x
a
x

m 1
2m

m
2
1
12m
2
( +
x
) (99)
Woolhouses Formula
For term annuities, we obtain the approximation
a
(m)
x:n
a
x:n
m 1
2m
(1v
n
n
p
x
)
m
2
1
12m
2
(+
x
v
n
n
p
x
(+
x+n
)) (100)
Woolhouses Formula
Letting m , we get
a
x
a
x

1
2

1
12
( +
x
)
a
x:n
a
x:n

1
2
(1 v
n
n
p
x
)
1
12
( +
x
v
n
n
p
x
( +
x+n
))
(101)
For a
x
with = 0, the approximation above reduces further to
e
x
(e
x
+ 1)
1
2

1
12
x
(102)
NB: For life tables, we can compute these quantities using the
approximation
x

1
2
[ln (p
x
) + ln (p
x+1
)]
Select and Ultimate Survival Models
Notation:
Aggregate Survival Models: Models for a large population, where
t
p
x
depends only on the current age x.
Select (and Ultimate) Survival Models: Models for a select group
of individuals that depend on the current age x and
Future survival probabilities for an individual in the group depend on
the individuals current age and on the age at which the individual
joined the group
d > 0 such that if an individual joined the group more than d years
ago, future survival probabilities depend only on current age. So, after
d years, the person is considered to be back in the aggregate
population.
Ultimately, a select survival model includes another event upon which
probabilities are conditional on.
Select and Ultimate Survival Models
Notation:
d is the select period
The mortality applicable to lives after the select period is over is
known as the ultimate mortality.
A select group should have a dierent mortality rate, as they have been
oered (selected for) life insurance. A question, of course, is the eect on
mortality by maintaining proper health insurance.
Example 3.8
Consider men who need to undergo surgery because they are suering
from a particular disease. The surgery is complicated and
P[survive one year after surgery] = 0.5
(d, l
60
, l
61
, l
70
) = (1, 89777, 89015, 77946)
(103)
Calculate P[A], P[B], P[C], where
A = (60),about to have surgery, will be alive at age 70
B = (60),had surgery at age 59, will be alive at age 70
C = (60),had surgery at age 58, will be alive at age 70
Example 3.8
P[A] = P[(60),about to have surgery, alive at age 61]
l
70
l
61
= 0.5
77946
89015
= 0.4378
P[B] =
l
70
l
60
= 0.8682
P[C] =
l
70
l
60
= 0.8682
(104)
Select Survival Models
S
[x]+s
(t) = P[(x + s) selected at (x), survives to(x + s + t)]
t
q
[x]+s
= P[(x + s) selected at (x)dies before(x + s + t)]
[x]+s
= force of mortality at (x + s) for select at (x)
= lim
h0
+
_
1 S
[x]+s
(h)
h
_
t
p
[x]+s
= 1
t
q
[x]+s
= S
[x]+s
(t)
= e
t
0

[x]+s+u
du
(105)
For t < d, we refer to to the above as part of the select model. For
t d, they are part of the ultimate model. Please read through section
on Select Life Tables.
Select Life Tables
Sometimes, we wish to compute values from life tables. Consider again a
model where x x
0
, where x
0
is the initial age, and 0 t d. Then
l
x+d
=
dt
p
[x]+t
l
[x]+t
(106)
Example 3.9
Theorem
Consider y x + d > x + s > x + t x x
0
. Then
yxt
p
[x]+t
=
l
y
l
[x]+t
st
p
[x]+t
=
l
[x]+s
l
[x]+t
(107)
Example 3.9
Proof.
yxt
p
[x]+t
=
yxd
p
[x]+d

dt
p
[x]+t
=
yxd
p
x+d

dt
p
[x]+t
=
l
y
l
x+d
l
x+d
l
[x]+t
=
l
y
l
[x]+t
st
p
[x]+t
=
dt
p
[x]+t
ds
p
[x]+s
=
l
x+d
l
[x]+t
l
[x]+s
l
x+d
=
l
[x]+s
l
[x]+t
(108)
Example 3.11
A select survival model has a select period of three years. Its ultimate
mortality is equivalent to the US Life Tables, 2002 Females of which an
extract is shown below. Information given is that for all x 65,
_
p
[x]
, p
[x1]+1
, p
[x2]+2
_
= (0.999, 0.998, 0.997). (109)
Table: 3.5: Extract from US LIfe Tables, 2002 Females
x l
x
70 80556
71 79026
72 77410
73 75666
74 73802
75 71800
Example 3.11
Calculate the probability that a woman currently aged 70 will survive to
age 75 given that
1
she was select at age 67:
2
she was select at age 68
3
4
Example 3.11
5
p
[703]+3
=
5
p
70
=
l
75
l
70
= 0.8913
5
p
[702]+2
=
l
[68]+2+5
l
[68]+2
=
l
75
l
[68]+2
=
l
75
l
[68]+3
1
p
[68]+2
=
l
75
l
71
1
p
[68]+2
=
4
p
71
1
p
[68]+2
=
71800
79026
0.997 = 0.9058
(110)
Example 3.11
5
p
[701]+1
=
l
[69]+1+5
l
[69]+1
=
l
75
l
[69]+1
=
l
75
l
[69]+3
(
1
p
[69]+1
)(
1
p
[69]+2
)
=
l
75
l
72
(
1
p
[69]+1
) (
1
p
[69]+2
)
=
71800
77410
0.997 0.998 = 0.9229
5
p
[70]
=
l
75
l
73
(
1
p
[70]
) (
1
p
[70]+1
) (
1
p
[70]+2
)
=
71800
75666
0.997 0.998 0.999 = 0.9432
(111)
Example 3.12
Given a table of values for q
[x]
, q
[x1]+1
, q
x
and the knowledge that the
model incorporates a 2year selct period, compute
4
p
[70]
3
q
[60]+1
4
p
[70]
= p
[70]
p
[70]+1
p
[70]+2
p
[70]+3
= p
[70]
p
[70]+1
p
72
p
73
=
_
1 q
[70]
_
_
1 q
[70]+1
_
(1 q
72
) (1 q
73
)
3
q
[60]+1
= q
[60]+1
+ p
[60]+1
q
62
+ p
[60]+1
p
62
q
63
= q
[60]+1
+
_
1 q
[60]+1
_
q
62
+
_
1 q
[60]+1
_
(1 q
62
) q
63
(112)
Example 3.12
[x]
, q
[x1]+1
, q
x
4
p
[70]
3
q
[60]+1
4
p
[70]
= p
[70]
p
[70]+1
p
[70]+2
p
[70]+3
= p
[70]
p
[70]+1
p
72
p
73
=
_
1 q
[70]
_
_
1 q
[70]+1
_
(1 q
72
) (1 q
73
)
3
q
[60]+1
= q
[60]+1
+ p
[60]+1
q
62
+ p
[60]+1
p
62
q
63
= q
[60]+1
+
_
1 q
[60]+1
_
q
62
+
_
1 q
[60]+1
_
(1 q
62
) q
63
(112)
Example 3.12
[x]
, q
[x1]+1
, q
x
4
p
[70]
3
q
[60]+1
4
p
[70]
= p
[70]
p
[70]+1
p
[70]+2
p
[70]+3
= p
[70]
p
[70]+1
p
72
p
73
=
_
1 q
[70]
_
_
1 q
[70]+1
_
(1 q
72
) (1 q
73
)
3
q
[60]+1
= q
[60]+1
+ p
[60]+1
q
62
+ p
[60]+1
p
62
q
63
= q
[60]+1
+
_
1 q
[60]+1
_
q
62
+
_
1 q
[60]+1
_
(1 q
62
) q
63
(112)
Example 3.13
A select survival model has a two-year select period and is specied as
follows. The ultimate part of the model follows Makehams law, where
(A, B, c) = (0.00022, 2.7 10
6
, 1.124):
x
= 0.00022 + (2.7 10
6
) (1.124)
x
(113)
The select part of the model is such that for 0 s 2,
[x]+s
= 0.9
2s
x+s
(114)
and so for 0 t 2,
t
p
[x]
= e
t
0

[x]+s
ds
= exp
_
0.9
2t
_
1 0.9
t
ln (0.9)
+
c
t
0.9
t
ln
_
0.9
c
_
__
(115)
Example 3.13
It follows that given an initial cohort at age x
0
, that is given l
x
0
, we can
compute the entries of a select life table via
l
x
= p
x1
l
x1
l
[x]+1
=
l
x+2
p
[x]+1
l
[x]
=
l
x+2
2
p
[x]
(116)
Homework Questions
HW: 3.1, 3.2, 3.4, 3.7, 3.8, 3.9, 3.10, 5.1, 5.3, 5.5, 5.6, 5.11, 5.14
What is a Premium?
When entering into a contract, the nancial obligations of all parties must
be specied. In an insurance contract, the insurance company agrees to
pay the policyholder benets in return for premium payments. The
premiums secure the benets as well as pay the company for expenses
attached to the administation of the policy
Premium Types
A Net Premium does not explicitly allow for companys expenses, while a
Oce or Gross Premium does. There may be a Single Premium or or a
series of payments that could even match with the policyholders salary
freequency.
It is important to note that premiums are paid as soon as the contract is
signed, otherwise the policyholder would attain coverage before paying for
it with the rst premium. This could be seen as an arbitrage opportunity -
non-zero probability of gain with no money up front.
Premium Types
Premiums cease upon death of the policyholder. The premium paying
term is the maximum length of time that premiums are required.
Certainly, premium term can be xed so that upon retirement, say, no
more payments are required.
Also, the benets can be secured in the future (deferred) by a single
premium payment up front. For example, pay now to secure annuity
payments upon retirement until death.
Assumptions
Recall the life model used in Example 3.13 : The select survival model has
a two-year select period and is specied as follows. The ultimate part of
the model follows Makehams law, where
(A, B, c) = (0.00022, 2.7 10
6
, 1.124):
x
= 0.00022 + (2.7 10
6
) (1.124)
x
(117)
The select part of the model is such that for 0 s 2,
[x]+s
= 0.9
2s
x+s
(118)
and so for 0 t 2,
t
p
[x]
= e
t
0

[x]+s
ds
= exp
_
0.9
2t
_
1 0.9
t
ln (0.9)
+
c
t
0.9
t
ln
_
0.9
c
_
__
(119)
Assumptions
Furthermore, we can extend the recursion principle when using a select life
model to obtain
a
x
= 1 + vp
x
a
x+1
a
[x]+1
= 1 + vp
[x]+1
a
x+2
a
[x]
= 1 + vp
[x]
a
[x]
(120)
Basic Model
In general, an insurance company can expect to have a total benet paid
out, along with expense loading and other related costs. We represent this
total benet as Z. Similarly, to fund Z, the company can expect the
policyholder to make a single payment, or stream of payments, that has
present value P Y. Here, P represents the level premium P and Y
represents the present value associated to a unit payment or payment
stream.
Future Loss Random Variable
For life contingent contracts, there is an outow and inow of money
during the term of the agreement. The premium income is certain, but
since the benets are life contingent, the term and total income may not
be certain up front. To account for this, we dene the Net Future Loss
L
n
0
(which includes expenses) and the Gross Future Loss L
g
0
(which does
not includes expenses) as
L
n
0
= PV [benet outgo] PV [net premium income]
L
g
0
= PV [benet outgo] + PV [expenses]
PV [gross premium income]
(121)
Example 6.2
An insurer issues a whole life insurance to [60], with sum insured S
payable immediately upon death. Premiums are payable annually in
advance, ceasing at 80 or on earlier death. The net annual premium is P.
What is the net future loss random variable L
n
0
for this contract in terms of
lifetime random variables for [60]?
L
n
0
= Sv
T
[60]
Pa
min
K
[60]
+1,20
(122)
Example 6.2
An insurer issues a whole life insurance to [60], with sum insured S
payable immediately upon death. Premiums are payable annually in
advance, ceasing at 80 or on earlier death. The net annual premium is P.
What is the net future loss random variable L
n
0
for this contract in terms of
lifetime random variables for [60]?
L
n
0
= Sv
T
[60]
Pa
min
K
[60]
+1,20
(122)
Equivalence Principle
Absent a risk-neutral type pricing measure, insurers price these
event-contingent contracts by setting the average value of the loss to be
zero. Symbolically, this is simply (for net premiums) nd P such that
E[L
n
0
] = 0 (123)
Note that this value P does not necessarily set Var [L
n
0
] = 0
Returning to our general set-up, we see that the equivalence pricing
principle can be summarized as
P =
E[Z]
E[Y]
(124)
As an introductory example, consider > 0 and a contract where (under
no selection)
Z = v
T
x
Y = a
T
x
t
p
x
= e
t
(125)
Hence, we have a unit whole-life insurance payable immediately upon death
of (x), where mortality is modeled to be exponential with parameter .
As an introductory example, consider > 0 and a contract where (under
no selection)
Z = v
T
x
Y = a
T
x
t
p
x
= e
t
(125)
Hence, we have a unit whole-life insurance payable immediately upon death
of (x), where mortality is modeled to be exponential with parameter .
We obtain
P
x
=
E
_
v
T
x
E
_
a
T
x
_
=
A
x
a
x
=
A
x
1
A
x
=
_
0
e
t
e
t
dt
1
_
0
e
t
e
t
dt
=
+
1

+
=
(126)
HW: repeat the above calculation if S
0
(x) =
x
for a nite lifetime

model with maximal age .
We obtain
P
x
=
E
_
v
T
x
E
_
a
T
x
_
=
A
x
a
x
=
A
x
1
A
x
=
_
0
e
t
e
t
dt
1
_
0
e
t
e
t
dt
=
+
1

+
=
(126)
HW: repeat the above calculation if S
0
(x) =
x
for a nite lifetime

model with maximal age .
If we repeat the previous example, but now for the case of of a unit
whole-life insurance contract with level annual premium payment and
benet paid at the end of the death year, then
Z = v
K
x
+1
Y = a
K
x
+1
t
p
x
= e
t
(127)
It follows that
P
x
= d
A
x
1 A
x
= d
k=0
e
(k+1)
(
k
p
x

k+1
p
x
)
1
k=0
e
(k+1)
(
k
p
x

k+1
p
x
)
= d
(1 e
k=0
e
(k+1)
e
k
1 (1 e
k=0
e
t
e
k
= d
(1 e
) e
1
1e
(+)
1 (1 e
) e
1
1e
(+)
= (1 e
) e
(128)
Example 6.5
Consider an endowment insurance with sum insured 100000 issued to a
select life aged [45] with term 20 years under which the death benet is
payable at the end of of the year of death. Using the Standard Select
Survival Model with interest at 5% per year, calculate the total amount of
net premium payable in a year if premiums are payable annually.
Example 6.5
By the EPP, the fact that d = 1
1
1.05
, and tables 6.1 and 3.7, we have
100000 A
[45]:20
= P a
[45]:20
P = 100000
A
[45]:20
a
[45]:20
=
100000
_
1 da
[45]:20
_
a
[45]:20
= 100000
_
1 d
_
a
[45]
l
65
l
[45]
v
20
a
[65]
__
a
[45]
l
65
l
[45]
v
20
a
[65]
= 100000
0.383766
12.94092
= 2965.52
(129)
New Business Strain
Starting up an insurance company requires start-up capital like most other
companies. Agents are charged with drumming up new business in the
form of nding and issuing new life insurance contracts. This helps to
diversify risk in the case of a large loss on one contract (more on this
later.)
However, new contracts can incur larger losses up front in the rst few
years even without a benet payout. This is due to initial commision
payments to agents as well as contract preparation costs. Periodic
maintenance costs can also factor into the premium calculation.
An Example..
Consider oering an nyear endowment policy to an age (x) in the
aggregate population where the benet B is paid at the end of the year of
death or on maturity. There are periodic renewal expenses of r per policy.
Then the premium P is calculated via the EPP as
Pa
x:n
= B A
x:n
+ r a
x:n
P = B
A
x:n
a
x:n
+ r
(130)
and we see that periodic expenses are simply passed on to the consumer!
An Example..
death or on maturity. There are periodic renewal expenses of r per policy.
Then the premium P is calculated via the EPP as
Pa
x:n
= B A
x:n
+ r a
x:n
P = B
A
x:n
a
x:n
+ r
(130)
and we see that periodic expenses are simply passed on to the consumer!
Another Example..
death or on maturity. There are periodic renewal expenses of r per policy
and an inital preparation expense of z per contract.
Then the premium P is calculated via
P = B
A
x:n
a
x:n
+ r +
z
a
x:n
(131)
and so the initial preparation expense is amortized over the lifetime of the
contract.
Another Example..
death or on maturity. There are periodic renewal expenses of r per policy
and an inital preparation expense of z per contract.
Then the premium P is calculated via
P = B
A
x:n
a
x:n
+ r +
z
a
x:n
(131)
and so the initial preparation expense is amortized over the lifetime of the
contract.
Example 6.6
An insurer issues a 25year annual premium endowment insurance with
sum insured 100000 to a select life aged [30]. The insurer incurs initial
expenses of 2000 plus 50% of the rst premium and renewable expenses of
2.5% of each subsequent premium. The death benet is payable
immediately upon death. What is the annual premium P?
Example 6.6
We can see that
L
g
0
= 100000v
min
T
[30]
,25
+ 2000 + 0.475P
+ 0.025Pa
min
K
[30]
+1,25
Pa
min
K
[30]
+1,25
P =
100000 E
_
v
min
T
[30]
,25
_
+ 2000
0.975 E
_
a
min
K
[30]
+1,25
_
0.475
(132)
Some more worked examples
Consider the following net loss random variables:
L
n
0
= v
T
x
Pa
min{T
x
,t}
L
n
0
= v
min{T
x
,n}
Pa
min{T
x
,t}
L
n
0
= v
K
x
+1
Pa
min{K
x
+1,t}
(133)
What are the fair premiums under the EPP?
P =
A
x
a
x:t
P =
A
x:n
a
x:t
P =
A
x
a
x:t
(134)
L
n
0
= v
T
x
Pa
min{T
x
,t}
L
n
0
= v
min{T
x
,n}
Pa
min{T
x
,t}
L
n
0
= v
K
x
+1
Pa
min{K
x
+1,t}
(133)
P =
A
x
a
x:t
P =
A
x:n
a
x:t
P =
A
x
a
x:t
(134)
L
n
0
= v
T
x
Pa
min{T
x
,t}
L
n
0
= v
min{T
x
,n}
Pa
min{T
x
,t}
L
n
0
= v
K
x
+1
Pa
min{K
x
+1,t}
(133)
P =
A
x
a
x:t
P =
A
x:n
a
x:t
P =
A
x
a
x:t
(134)
L
n
0
= v
T
x
Pa
min{T
x
,t}
L
n
0
= v
min{T
x
,n}
Pa
min{T
x
,t}
L
n
0
= v
K
x
+1
Pa
min{K
x
+1,t}
(133)
P =
A
x
a
x:t
P =
A
x:n
a
x:t
P =
A
x
a
x:t
(134)
L
n
0
= v
T
x
Pa
min{T
x
,t}
L
n
0
= v
min{T
x
,n}
Pa
min{T
x
,t}
L
n
0
= v
K
x
+1
Pa
min{K
x
+1,t}
(133)
P =
A
x
a
x:t
P =
A
x:n
a
x:t
P =
A
x
a
x:t
(134)
L
n
0
= v
T
x
Pa
min{T
x
,t}
L
n
0
= v
min{T
x
,n}
Pa
min{T
x
,t}
L
n
0
= v
K
x
+1
Pa
min{K
x
+1,t}
(133)
P =
A
x
a
x:t
P =
A
x:n
a
x:t
P =
A
x
a
x:t
(134)
Refunded Deferred Annual Whole-Life Annuity Due
Consider the case where a nyear deferred annual whole-life annuity due
of 1 on a life (x) where if the death occurs during the deferral period, the
single benet premium is refunded without interest at the end of the
year of death. What is this single benet premium P?
The net loss random variable is
L
n
0
= Pv
K
x
+1
1
{K
x
+1n}
+ v
n
1
{K
x
+1>n}
a
K
x
+1n
P (135)
and by the EPP we have
0 = PA
1
x:n
+
n|
a
x
P (136)
This implies that
P =
n|
a
x
1 A
1
x:n
=
A
x:n
A
1
x:n
1 A
1
x:n
a
x+n
(137)
The net loss random variable is
L
n
0
= Pv
K
x
+1
1
{K
x
+1n}
+ v
n
1
{K
x
+1>n}
a
K
x
+1n
P (135)
and by the EPP we have
0 = PA
1
x:n
+
n|
a
x
P (136)
This implies that
P =
n|
a
x
1 A
1
x:n
=
A
x:n
A
1
x:n
1 A
1
x:n
a
x+n
(137)
Prot
Consider a 1year term insurance contract issued to a select life [x], with
sum insured S = 1000, interest rate i = 0.05, and mortality
q
[x]
= P[T
[x]
1] = 0.01
It follows that L
0
, the future loss random variable calculated at the time of
issuance, is
L
0
= 1000v
1
1
T
[x]
1
P
P = E[1000v
1
1
T
[x]
1
] = 1000v P[T
[x]
1]
=
(1000)(0.01)
1.05
= 9.52
(138)
Prot
Consider a 1year term insurance contract issued to a select life [x], with
sum insured S = 1000, interest rate i = 0.05, and mortality
q
[x]
= P[T
[x]
1] = 0.01
It follows that L
0
, the future loss random variable calculated at the time of
issuance, is
L
0
= 1000v
1
1
T
[x]
1
P
P = E[1000v
1
1
T
[x]
1
] = 1000v P[T
[x]
1]
=
(1000)(0.01)
1.05
= 9.52
(138)
Prot
Consider that the company has issued a lot of these contracts, say N 1,
to independent select lives [x]. Let D
[x]
be the random variable
representing the number of deaths in a year of this population, and assume
D
[x]
Bin(N, q
[x]
) (139)
Prot
In general, we have the event that the insurer turns a prot on this group
of policies is Prot =
_
D
[x]
N q
[x]
_
, and so as N ,
P[Prot] = P[D
[x]
N q
[x]
]
= P[D
[x]
E[D
[x]
]]
= P
_
_
D
[x]
E[D
[x]
]
_
Var
_
D
[x]
0
_
_
(0) =
1
2
by the CLT.
(140)
HW: Compute
E
_
Prot [ D
[x]
N q
[x]
N P
(141)
Prot
Consider now a whole-life contract issued to [x] with sum insured S and
annual premium P. Then
P[Prot] = P[L
0
< 0] = P[Sv
K
[x]
+1
Pa
K
[x]
+1
< 0]
= P[Sv
K
[x]
+1
< P
1 v
K
[x]
+1
d
]
= P
_
v
K
[x]
+1
<
P
P + d S
_
= P
_
K
[x]
+ 1 >
1
ln
_
P
P + d S
__
= P
_
K
[x]
+ 1 >
1
ln
_
P
P + d S
_
|
_
=
ln
(
P
P+dS
)
p
[x]
(142)
Conditional Sums
Dene L
0
(k) = PV[Loss [ K
[x]
= k]. For a contract with a term n, it
follows that
L
0
= E[PV[Loss]] = E
_
n1
k=0
PV[Loss [ K
[x]
= k] 1
K
[x]
=k
_
+E
_
PV[Loss [ K
[x]
n] 1
K
[x]
n
_
=
n1
k=0
_
PV[Loss [ K
[x]
= k] P[K
[x]
= k]
_
+ L
0
(n) P[K
[x]
n]
=
n1
k=0
L
0
(k)
k|
q
[x]
+ L
0
(n)
n
p
[x]
(143)
Q: Can we use this for the case n = ?
Example 6.9
A life insurer is about to issue a 25year endowment insurance with a
basic sum insured S = 250000 to a select life aged exactly [30]. Premiums
are payable annually throughout the term of the policy. Initial expenses are
1200 plus 40% of the rst premium and renewal expenses are 1% of the
second and subsequent premiums. The insurer allows for a compound
reversionary bonus of 2.5% of the basic sum insured, vesting on each
policy anniversary (including the last.) The death benet is payable at the
end of the year of death. Assume the Standard Select Survival Model with
interest rate 5% per year.
Example 6.9
In this case, we have the reversionary bonus exponentially grow the sum
insured:
L
0
= B
0
(K
[x]
) + E
0
P
0
(K
[x]
)
B
0
(k) = 250000 1.025
k
v
k+1
for k 0, 1, 2, ..., 24
B
0
(25) = 250000 1.025
25
v
25
E
0
= 1200 + 0.39P
P
0
(k) = 0.99Pa
min {k+1,25}
(144)
Example 6.9
For 1 + j =
1+i
1.025
, we have j = 0.02439 and so
E[L
0
] = E[B
0
(K
[x]
)] + E
0
E[P
0
(K
[x]
)]
=
24
k=0
B
0
(k)
k|
q
[30]
+ B
0
(25)
25
p
[30]
+ E
0
0.99Pa
[30]:25
=
24
k=0
250000 (1.025)
t
(1.05)
t+1
k|
q
[30]
+ B
0
(25)
25
p
[30]
+ E
0
0.99Pa
[30]:25
=
250000
1.025
A
1
[30]25 j
+ B
0
(25)
25
p
[30]
+ 1200 + 0.39P 14.5838P
(145)
Under the EPP, we have P = 97644.44.
HW: Replicate Table 6.3 in the text by using a spreadsheet program.
Compare this example with Example 6.10.
Example 6.9
For 1 + j =
1+i
1.025
, we have j = 0.02439 and so
E[L
0
] = E[B
0
(K
[x]
)] + E
0
E[P
0
(K
[x]
)]
=
24
k=0
B
0
(k)
k|
q
[30]
+ B
0
(25)
25
p
[30]
+ E
0
0.99Pa
[30]:25
=
24
k=0
250000 (1.025)
t
(1.05)
t+1
k|
q
[30]
+ B
0
(25)
25
p
[30]
+ E
0
0.99Pa
[30]:25
=
250000
1.025
A
1
[30]25 j
+ B
0
(25)
25
p
[30]
+ 1200 + 0.39P 14.5838P
(145)
Under the EPP, we have P = 97644.44.
HW: Replicate Table 6.3 in the text by using a spreadsheet program.
Compare this example with Example 6.10.
Portfolio Percentile Premium Principle
Assume once again that a company is about to issue insurance to N
independent lives [x], each with loss L
0,i
for i 1, 2, .., N . In this case
L
0
=
N
i =1
L
0,i
E[L
0
] = E
_
N
i =1
L
0,i
_
=
N
i =1
E[L
0,i
] = N E[L
0,1
]
Var [L
0
] = N Var [L
0,1
]
(146)
Portfolio Percentile Premium Principle
If we require P such that P[L
0
< 0] = , we can use CLT once again to
show
= P[L
0
< 0]
= P
_
L
0
E[L
0
]
_
Var [L
0
]
<
E[L
0
]
_
Var [L
0
]
_
E[L
0
]
_
Var [L
0
]
_
as N
(147)
For an individual present value of loss, stated wlog as L
0,1
, we recover the
EPP as N
E[L
0,1
]
1
()
_
Var [L
0,1
]
N
0 (148)
Example 6.11
An insurer issues whole life insurance policies to select lives aged [30]. The
sum insured S = 100000 is paid at the end of the month of death and
level monthly premiums are payable throughout the term of the policy.
Initial expenses, incurred at the issue of the policy, are 15% of the total of
the rst years premiums. Renewal expenses are 4% of every premium,
including those in the rst year. Assume the SSSM with interest at 5% per
year.
Calculate the monthly premium P using the EPP and
Calculate the monthly premium P using the PPPP such that
= 0.95 and N = 10000.
Example 6.11
For the EPP calculation, we have
E[PV(Premiums)] = 12Pa
(12)
[30]
= 227.065P
E[PV(Benets)] = 100000A
(12)
[30]
= 7866.18
E[PV(Expenses)] = (0.15)(12P) + (0.04)(12Pa
(12)
[30]
)
= 10.8826P
E[PV(Premiums)] = E[PV(Benets)] +E[PV(Expenses)]
P = 36.39
(149)
Example 6.11
For the EPP calculation, we have
E[PV(Premiums)] = 12Pa
(12)
[30]
= 227.065P
E[PV(Benets)] = 100000A
(12)
[30]
= 7866.18
E[PV(Expenses)] = (0.15)(12P) + (0.04)(12Pa
(12)
[30]
)
= 10.8826P
E[PV(Premiums)] = E[PV(Benets)] +E[PV(Expenses)]
P = 36.39
(149)
Example 6.11
For all i 1, 2, .., N, we have the i.i.d. PV(Loss) random variables
L
0,i
= 100000v
K
(12)
[30]
+
1
12
+ (0.15)(12P)
(0.96)
_
12Pa
(12)
K
(12)
[30]
+
1
12
_
E[L
0,i
] = 100000A
(12)
[30]
+ (0.15)(12P) (0.96)(12Pa
(12)
[30]
)
= 7866.18 216.18P
(150)
Example 6.11
To nd the variance, we rewrite
L
0,i
=
_
100000 +
(0.96)(12P)
d
(12)
_
v
K
(12)
[30]
+ (0.15)(12P)
(0.96)(12P)
d
(12)
Var [L
0,i
] =
_
100000 +
(0.96)(12P)
d
(12)
_
2
_
2
A
(12)
[30]

_
A
(12)
[30]
_
2
_
= (100000 + 236.59P)
2
(0.0053515)
(151)
Example 6.11
Collecting our results, we now have
0.95 = = P[L
0
< 0]

_
E[L
0
]
_
Var [L
0
]
_
=
_
N
E[L
0,1
]
_
Var [L
0,1
]
_
=
_
10000
216.18P 7866.18
(100000 + 236.59P)
0.0053515
_
(152)
Example 6.11
It follows that
1.645 =
1
(0.95)
10000
216.18P 7866.18
(100000 + 236.59P)
0.0053515
P = 36.99
(153)
For general N, we have
216.18P 7866.18
(100000 + 236.59P)
0.0053515
=
1.645
N
(154)
and as N , we have P 36.39, recovering the EPP premium as
expected.
Independent Exponential RVs
Imagine that a fully continuous whole life insurance is oered to N
individuals aged [x] with T
(i )
[x]
exp() for all i 1, .., N. For each
insured, the i.i.d. loss random variables are
L
0,i
= Sv
T
(i )
[x]
Pa
T
(i )
[x]
=
S + P
e
T
(i )
[x]
(155)
and so for L
0
=
N
i =1
L
0,i
, the PPPP seeks to determine P such that
P
_
N
i =1
_
S + P
e
T
(i )
[x]
_
< 0
_
= (156)
In this case, we can rewrite this as
P
_
1
N
N
i =1
e
T
(i )
[x]
<
P
P + S
_
= (157)
In this case, for each i if we dene Y
i
:= e
T
(i )
[x]
, then we know that
P[Y
i
> y] = P
_
e
T
(i )
[x]
> y
_
= 1 y

.
HW Using convolution techniques, nd the above probability
P
_
1
N
N
i =1
Y
i
<
P
P + S
_
= . (158)
Are there any ergodic theory results that we can use?
In this case, we can rewrite this as
P
_
1
N
N
i =1
e
T
(i )
[x]
<
P
P + S
_
= (157)
In this case, for each i if we dene Y
i
:= e
T
(i )
[x]
, then we know that
P[Y
i
> y] = P
_
e
T
(i )
[x]
> y
_
= 1 y

.
HW Using convolution techniques, nd the above probability
P
_
1
N
N
i =1
Y
i
<
P
P + S
_
= . (158)
Are there any ergodic theory results that we can use?
Capped Maximal Loss
Another principle is to nd P such that for any (0, 1), the probability
that any contract suers a loss of < S

N
< S is set to for a set of
i.i.d. exponentially distributed times T
(i )
[x]
:
P
_
max
i =1..N
_
S + P
e
T
(i )
[x]
_
<
_
= (159)
We rewrite this as
= P
_
min
i =1..N
_
T
(i )
[x]
_
>
1
ln
_
P +
P +
__
=
_
e
ln
(
P+
P+
)]
_
N
=
_
P +
P + S
_N
P =
S

N

1

N
as N
(160)
For this risk measure, is there a number of policy holders N that is too
high?
Capped Maximal Loss

N
(i )
[x]
:
P
_
max
i =1..N
_
S + P
e
T
(i )
[x]
_
<
_
= (159)
We rewrite this as
= P
_
min
i =1..N
_
T
(i )
[x]
_
>
1
ln
_
P +
P +
__
=
_
e
ln
(
P+
P+
)]
_
N
=
_
P +
P + S
_N
P =
S

N

1

N
as N
(160)
high?
Capped Maximal Loss

N
(i )
[x]
:
P
_
max
i =1..N
_
S + P
e
T
(i )
[x]
_
<
_
= (159)
We rewrite this as
= P
_
min
i =1..N
_
T
(i )
[x]
_
>
1
ln
_
P +
P +
__
=
_
e
ln
(
P+
P+
)]
_
N
=
_
P +
P + S
_N
P =
S

N

1

N
as N
(160)
high?
Comments on PPPP
Notice that the PPPP only guarantees that the probability of a loss is
1 .
It says nothing about the size of what that loss could be if it
arises.
This is a big problem if the loss is extremely large and bankrupts the
insurer. It may seem very unlikely, but recent economic events have
shown otherwise.
Further improvements to this model can be seen in the ERM for
Strategic Management (Status Report) by Gary Venter, posted on
the SOA.org website
Also, there is a close link, perhaps to be explored in a project, with
VAR in the nancial world. Click here for an informative article in the
NY Times
TM
for an article on VAR and the recent nancial crisis.
Homework Questions
HW: 6.1, 6.2, 6.5, 6.7, 6.8, 6.12, 6.14, 6.15
Policy Value Basis
When entering into a contract, the nancial obligations of all parties
should be specied at the time the agreement is signed. This includes
disclosure of health status, age, and premium payments expected to fund
benets and expenses associated with the contract.
The Policy Value
t
V is the expected value of the future loss random
variable L
t
at time t:
t
V = E[L
t
] = E[ Loss [ T
[x]
> t] (161)
Policy Value Basis
Denition
The gross premium policy value for a policy in force at duration
t 0 years after it was purchased is the expected value at that time
of the gross future loss random variable on a specied basis. The
premiums used in the calculation are the actual premiums payable
under the contract.
The net premium policy value for a policy in force at duration
t 0 years after it was purchased is the expected value at that time
of the net future loss random variable on a specied basis (which
makes no allowance for expenses.) The premiums used in the
calculation are the net premiums calculated on the policy value basis
using the equivalence principle, not the actual premiums payable
It is important to note that usual practice dictates that when calculating
t
V, premiums and premium-related expenses due at t are regarded as
future payments and any insurance benets and related expenses as past
payments.
Recursion
Dene
P
t
as the premium payable at time t
e
t
as the premium-related expense payable at time t
S
t+1
as the sum insured payable at time t + 1
E
t+1
as the expense of paying the sum insured at time t + 1
t+1
V as the gross premium policy value for a policy in force at time
t + 1
L
t
as the gross future loss random variable at time t
i
t
as the interest rate from time t to time t + 1.
Recursion
Then, using recursion, we obtain
t
V = e
t
P
t
+ q
[x]+t

S
t+1
+ E
t+1
1 + i
t
+ p
[x]+t

t+1
V
1 + i
t
. (162)
Notice that if there is a xed term to the contract, such as an endowment
or term insurance, then we have the boundary condition
n
V = 0 (163)
Also, if the premium is calculated using the EPP and the policy basis is
the same as the premium basis, then
0
V = E[L
0
] = 0 (164)
BC : Endowment
For an endowment insurance contract with sum insured S, however, we
have the pair of boundary conditions
n
V = lim
0
+
n
V = S
n
V = 0
(165)
In calculating
n1
V, we actually use
n
V instead of
n
V. See next example!
Example 7.7
Consider a zero-expense, 20 year endowment policy purchased by a life
aged 50. Level premiums of 23500 per year are payable annually
throughout the term of the policy. A sum insured of 700000 is payable at
the end of the term if the life survives to age 70. On death before age 70,
a sum insured is payable at the end of the year of death equal to the policy
value at the start of the year in which the policyholder dies. Assuming the
SSSM with interest at 3.5% per year, calculate
15
V, the policy value in
force at the start of the 16
th
year.
Example 7.7
It follows that
S
t+1
=
t
V
e
t
= 0 = E
t
S = 700000
P
t
= 23500
t
V = 23500 + q
[50]+t
t
V
1.035
+ p
[50]+t
t+1
V
1.035
(166)
Example 7.7
Combining with our boundary value, we obtain the dierence equation
t
V =
p
[50]+t

t+1
V 24322.50
p
[50]+t
+ 0.035
20
V = 700000.
(167)
Our initial iteration actually uses
20
V to obtain
19
V =
p
69
(
20
V) 24322.50
p
69
+ 0.035
= 652401 (168)
Use tables or spreadsheet to calculate SSSM values and obtain
15
V = 478063.
Example 7.7
t
V =
p
[50]+t

t+1
V 24322.50
p
[50]+t
+ 0.035
20
V = 700000.
(167)
20
V to obtain
19
V =
p
69
(
20
V) 24322.50
p
69
+ 0.035
= 652401 (168)
15
V = 478063.
Example 7.7
t
V =
p
[50]+t

t+1
V 24322.50
p
[50]+t
+ 0.035
20
V = 700000.
(167)
20
V to obtain
19
V =
p
69
(
20
V) 24322.50
p
69
+ 0.035
= 652401 (168)
15
V = 478063.
Example 7.1
Consider a 20year endowment policy purchased by a life aged 50. Level
premiums are payable annually throughout the term of the policy and the
sum insured, S = 500000, is payable at the end of the year of death or at
the end of the term, whichever is sooner. The basis used by the insurance
company for all calculations is under the SSSM with 5% per year interest
and no allowance for expenses. Calculate P under the EPPP and the
corresponding policy values
Example 7.1
Subsitituting the information contained in the problem formation, we
obtain
t
V = P + q
[50]+t

500000
1.05
+ p
[50]+t

t+1
V
1.05
=
t+1
V 500000
1.05
+
500000
1.05
P
0
V = 0 = E[L
0
] = Pa
[50]:20
500000A
[50]:20
20
V = 500000
(169)
Solving for P, we obtain P = 15114.33. Iteration of the resulting
dierence equation delivers the remaining policy values.
Example 7.4
A man aged 50 purchases a deferred annuity policy. The annuity will be
paid annually for life, with the rst payment on his 60
th
birthday. Each
annuity payment will be 10000. Level premiums of 11900 are payable
annually for at most 10 years. On death before age 60, all premiums paid
will be returned, without interest, at the end of the year of death. The
insurer uses the following basis for calculation of policy values:
SSSM with 5% interest per year
Expenses of 10% of the rst premium, 5% of subsequent premiums,
25 each time an annuity payment is paid, and 100 when a death claim
is paid.
Calculate
t
V for t 0, 1, ..., 9
Example 7.4
Our initial policy value is
0
V = P (IA)
1
50:10
+ 100A
1
[50]:10
+ 10025v
10
10
p
[50]
a
60
_
0.95a
[50]:10
0.05
_
P
= 485 > 0
(170)
This of course can now be used to forward iterate to nd
t
V
9
t=1
. Since
0
V = 485 > 0, the premiums charged correspond to a valuation basis that
is more conservative than the premium basis.
Example 7.4
For 1 t 9, we can see our recursion equation is
t
V = 0.95P + q
[50]+t

(t + 1) P + 100
1.05
+ p
[50]+t

t+1
V
1.05
(171)
For t 10, we have
t
V = 10025a
50+t
t
+V = 10025a
50+t
=
t
V 10025
(172)
Which do we use to nd
9
V,
10
V or
10
+V ?
Notes
The previous example shows that sometimes we need to calculate the
initial value, given the information contained in the problem
statement, to iterate forward, especially if there is no term n and
corresponding boundary condition
n
V. Also, no annuity payments
have occured yet and this reects in the expenses.
It is likely that DSAR := S
t+1
+ E
t+1
t+1
V ,= 0. The Death Strain
At Risk, or DSAR, is the extra amount needed to increase the policy
value to the death benet at time t + 1. This is a capital based risk
measure, as it is a direct measure of what the insurer may be at risk
of needing to close out a contract if a benet must be paid. If the
DSAR is large enough, management may want to purchase
reinsurance in case a large DSAR (even with low probability) occurs.
Example 7.3
A woman aged 60 purchases a 20 year endowment insurance with a sum
insured S = 100000 payable at the end of the year of death or on survival
to age 80, whichever occurs rst. An annual premium of 5200 is
payable for at most 10 years. The insurer uses the following basis for
calculation of policy values:
SSSM with 5% interest per year
Expenses of 10% of the rst premium, 5% of subsequent premiums,
and 200 on payment of the sum insured.
Example 7.3
Our initial recursion equation is
0
V = 0.9P + q
[60]
100200
1.05
+ p
[60]
1
V
1.05
. (173)
For 1 t 9, we have
t
V = 0.95P + q
[60]+t

100200
1.05
+ p
[60]+t

t+1
V
1.05
. (174)
For t = 10, we have
10
V = E[L
10
] = E
_
100200v
min{K
70
+1,10}
_
= 100200A
70:10
= 63703.
(175)
HW Compute
9
V and
12
V explicity.
Annual Prot - Example 7.8
An insurer issues a large number of policies identical to the policy in
Example 7.3 to women aged 60. Five years after they were issued, a total
of 100 of these policies were still in force. In the following year, one person
died d
5
= 1 and
expenses of 6% of each premium paid were incurred - i.e.
e
actual
5
= 0.06P
5
interest was earned at 6.5% on all assets - i.e. i
actual
5
= 0.065
expenses of E
actual
6
= 250 were incurred on the payment of the sum
insured for the policyholder who died.
Calculate a.) the prot or loss on the group of policies for this year and b.)
determine how much of this prot or loss is attributable to prot or loss
from mortality, from interest, and from expenses.
To solve, we compute the dierence between the growth of assets from
t = 5 to t = 6 and subtract the total asset value at t = 6:
Prot = N (
5
V + P
5
0.06P
5
) (1 + i
5
)
1
(d
5
(S + E
6
) + (N d
5
)
6
V)
= 100 (
5
V + (0.94)(5200)) (1.065)
1
(1 (S + E
6
) + 99
6
V)
= 106.5 (
5
V + 4888) (100250 + 99
6
V)
(176)
Furthermore,
5
V = E[L
5
] = 100200A
65:15
0.95 5200a
65:5
= 29068
6
V = E[L
6
] = 100200A
66:14
0.95 5200a
66:4
= 35324
Prot = 106.5 (29068 + 4888) (100250 + (99)(35324))
= 18919
(177)
HW: Read the solution for part b.) in the textbook.
Furthermore,
5
V = E[L
5
] = 100200A
65:15
0.95 5200a
65:5
= 29068
6
V = E[L
6
] = 100200A
66:14
0.95 5200a
66:4
= 35324
Prot = 106.5 (29068 + 4888) (100250 + (99)(35324))
= 18919
(177)
HW: Read the solution for part b.) in the textbook.
Dene AS
t
as the share of the insurers assets attributable to each policy
in force at time t. Consider now a policy identical to the policy studied in
Example 7.4 amd suppose that this policy has now been in force for ve
years. Suppose that over the past ve years, the insurers experience in
respect of similar policies has realized annual interest on investments as
(i
1
, i
2
, i
3
, i
4
, i
5
) = (0.048, 0.056, 0.052, 0.049, 0.047).
Furthermore,
Expenses at the start of the year in which a policy was issued were
15% of the premium
Expenses at the start of the year after the year in which a policy was
issued were 6% of the premium
The expense of paying a death claim was, on average, 120
The mortality rate q
[50]+t
0.0015 for t 0, 1, 2, 3, 4
Calculate AS
t
using the convention that AS
t
does not include the premium
and related expense due at time t. (This of course means that AS
0
= 0.)
We calculate AS
1
here and refer to Table 7.1 for the complete set of
calculations
At time 0, insurer receives premiums minus expenses of
0.85 11900N = 10115N.
At time 1, this accumulates to 10115N (1 + i
1
) = 10601N.
A total of 0.0015N policy holders die in the rst year and death
claims are 0.0015N (11900 + 120) = 18N.
Therefore, the value of the fund at the end of the rst year is
10601N 18N = 10582N.
It follows that
AS
1
=
Fund Value at time 1
Number of Policies in Force at time 1
=
10582N
0.9985N
= 10598
(178)
Now, read Section 7.4 on computing Valuation between premium dates.
Policy Values - Cts Cash Flows
Recall that
t
V = e
t
P
t
+ q
[x]+t

S
t+1
+ E
t+1
1 + i
t
+ p
[x]+t

t+1
V
1 + i
t
. (179)
Now, consider that t is real-valued and dene
P
t
as the annual rate of premium payable at time t
e
t
as the annual rate of premium-related expense payable at time t
S
t
as the sum insured payable at time t if the policy holder dies
exactly at t
E
t
as the expense of paying the sum insured at time t
[x]+t
as the force of mortality at age [x] + t
t
as the force of interest assumed at time t
t
V as the ipolicy value at time t.
Now, as the force of interest varies, we have
v(t) = e
t
0

u
du
v(t)
v(s)
= e
t
s

u
du
(180)
and so
t
V =
_

0
v(t + s)
v(t)

_
[S
t+s
+ E
t+s
]
s
p
[x]+t
[x]+t+s
_
ds
_

0
v(t + s)
v(t)

_
[P
t+s
e
t+s
]
s
p
[x]+t
_
ds
(181)
Q: What happens if there is a nite term to contract?
By the product rule and the identities
r t
p
[x]+t
=
r
p
[x]
t
p
[x]
d
dt
_
t
p
[x]
_
=
t
p
[x]
[x]+t
v
(t) =
t
v(t)
(182)
we obtain the ODE
d
dt
(
t
V) =
t

t
V + P
t
e
t
(S
t
+ E
t

t
V)
[x]+t
(183)
Boundary Conditions:
For S sum insured, we have
lim
tn
t
V = S for an endowment policy with term n years.
lim
tn
t
V = 0 for a term policy with term n years.
lim
t
t
V = S for a whole life policy with upper limit years.
(184)
Forward Euler:
t+h
V =
t
V + h
_
t

t
V + P
t
e
t
(S
t
+ E
t

t
V)
[x]+t
_
(185)
As an example, consider the case where for S sum insured, we have
S
t+s
= S
e
t+s
= 0 = E
t+s
t
=
[x]+t+s
=
P
t+s
= Pe
(t+s)
(186)
Then it follows that
t
V =
_

0
Se
s
e
s
ds
_

0
e
s
Pe
(t+s)
e
s
ds
=
S
+

Pe
t
+ +
(187)
Policy Alterations
In many cases, policyholders may wish to change the terms of their
contract if it is still in eect. For example:
They may wish to stop making premiums, or to change the terms of
their benet payout.
They may wish to cash out their position, or simply wish to shorten
the time remaining until payout.
One may argue that the insurer is under no obligation to make such
changes if they are not written expressly into the initial contract. For
example:
The policyholder (but not insurer) may know something about their
health status that would make it better for them to cash out now.
By having to liquidate assets that cover the policy, the insurer may
have to take a loss to be able to settle the alteration, and this could
aect other policyholders adversely.
Policy Alterations
In many cases, policyholders may wish to change the terms of their
contract if it is still in eect. For example:
They may wish to stop making premiums, or to change the terms of
their benet payout.
They may wish to cash out their position, or simply wish to shorten
the time remaining until payout.
One may argue that the insurer is under no obligation to make such
changes if they are not written expressly into the initial contract. For
example:
The policyholder (but not insurer) may know something about their
health status that would make it better for them to cash out now.
By having to liquidate assets that cover the policy, the insurer may
have to take a loss to be able to settle the alteration, and this could
aect other policyholders adversely.
Policy Alterations
Beacuse of these concerns, the lender may agree to alter the terms of the
contract, but only paying a Surrender (Cash) Value C
t
of a fraction of
t
V or AS
t
.
C
t
= E[PV
t
(future benets + expenses, altered contract)]
E[PV
t
(future premiums, altered contract)]
(188)
In allowing the policy to lapse, the policy holder is cashing out a policy
and using the proceeds to enter into a new contract. If the period between
lapsing and entering into a new contract is too short, then the insurer may
suer from not earning enough income to cover the new business strain of
writing the rst contract. Hence, some countries including the US have
non-forfeiture laws that allow for zero cash values for early surrenders.
Policy Alterations
Beacuse of these concerns, the lender may agree to alter the terms of the
contract, but only paying a Surrender (Cash) Value C
t
of a fraction of
t
V or AS
t
.
C
t
= E[PV
t
(future benets + expenses, altered contract)]
E[PV
t
(future premiums, altered contract)]
(188)
In allowing the policy to lapse, the policy holder is cashing out a policy
and using the proceeds to enter into a new contract. If the period between
lapsing and entering into a new contract is too short, then the insurer may
suer from not earning enough income to cover the new business strain of
writing the rst contract. Hence, some countries including the US have
non-forfeiture laws that allow for zero cash values for early surrenders.
Policy Alterations: Example 7.13
Consider the policy discussed in Examples 7.4 and 7.9. Given the
experience of the insurer detailed in Example 7.9, at the start of the 6
th
year but before paying the premium due, the policyholder requests that the
policy be altered in one of the following ways:
1
The policy is surrendered immediately
2
No more premiums are paid, and a reduced annuity is payable from
age 60. In this case, all premiums paid are refunded at the end of the
year of death if the policyholder dies before age 60.
3
Premiums continue to be paid, but the benet is altered from an
annuity to a lump sum (pure endowment) payable on reaching age 60.
Expenses and benets on death before age 60 follow the original
policy terms. There is an expense of 100 associated with paying the
sum insured at the new maturity date.
Calculate the surrender value, the reduced annuity, and sum insured
assuming the insurer uses
90% of the asset share less a charge of 200 or
95% of the policy value less a charge of 200
together with the assumptions in the policy value basis when calculating
revised benets and premimus. Use the associated values
5
V = 65470
AS
5
= 63509
(189)
1
C
assetshare
5
= 0.9 AS
5
200 = 56958
C
policyvalue
5
= 0.9
5
V 200 = 58723
(190)
2
C
5
= 5 11900A
1
55:5
+ 100A
1
55:5
+ (X + 25) v
5
5
p
55
a
60
X
assetshare
= 4859
X
policyvalue
= 5012
(191)
3
C
5
+ 0.95 11900a
55:5
= 11900
_
(IA)
1
55:5
+ 5A
1
55:5
_
+ 100A
1
55:5
+ v
5
5
p
55
(S + 100)
S
assetshare
= 138314
S
policyvalue
= 140594
(192)
1
C
assetshare
5
= 0.9 AS
5
200 = 56958
C
policyvalue
5
= 0.9
5
V 200 = 58723
(190)
2
C
5
= 5 11900A
1
55:5
+ 100A
1
55:5
+ (X + 25) v
5
5
p
55
a
60
X
assetshare
= 4859
X
policyvalue
= 5012
(191)
3
C
5
+ 0.95 11900a
55:5
= 11900
_
(IA)
1
55:5
+ 5A
1
55:5
_
+ 100A
1
55:5
+ v
5
5
p
55
(S + 100)
S
assetshare
= 138314
S
policyvalue
= 140594
(192)
1
C
assetshare
5
= 0.9 AS
5
200 = 56958
C
policyvalue
5
= 0.9
5
V 200 = 58723
(190)
2
C
5
= 5 11900A
1
55:5
+ 100A
1
55:5
+ (X + 25) v
5
5
p
55
a
60
X
assetshare
= 4859
X
policyvalue
= 5012
(191)
3
C
5
+ 0.95 11900a
55:5
= 11900
_
(IA)
1
55:5
+ 5A
1
55:5
_
+ 100A
1
55:5
+ v
5
5
p
55
(S + 100)
S
assetshare
= 138314
S
policyvalue
= 140594
(192)
Related Project Topics
Over- or underestimated interest rates are only one risk factor for
actuarial reserving. Another very real factor is known as longevity
risk, which is due to the possibility that a pensioner may live longer
than expected. Hedging against such a possibility is extremely
important, Please consult the paper by Tsai, Tzeng, and Wang on
Hedging Longevity Risk When Interest Rates Are Uncertain
Related Project Topics
For those of you interested in more sophisitcated, cutting edge coding
methods for reserving, code.google.com has a site dedicated to
ChainLadder (google code name chainladder) that contains an R
package providing methods which are typically used in insurance
claims reserving. Links to slides explaining the method are also on the
site
An Introduction to R: Examples for Actuaries by Nigel de Silva is a
very nice primer on using R.
Homework Questions
HW: 7.1, 7.2, 7.4, 7.5, 7.8, 7.12, 7.14, 7.15
Two State Model: Alive or Dead
Recall that for the survival time T
x
of an individual (x), we have
S
x
(t) = 1 F
x
(t) = 1 P[T
x
t] (193)
We now extend the model to include multiple states, but rst we dene
the random variable Y(t) 0, 1 as the state of the individual (x). If (x)
is alive at time x + t, then Y(t) = 0. Otherwise, Y(t) = 1.
Hence, we can dene
T
x
= max t [ Y(t) = 0 (194)
and the model ow 0 1.
Accidental Death Model
We can also dene
Y(t) =
_
_
_
0 if (x) is alive at time x + t
1 if (x) is dead at time x + t of accidental cause
2 if (x) is dead at time x + t of other cause
0
>
>
>
>
>
>
>
>
1 2
Figure: ADM Flow Chart
There is a sum insured upon leaving state 0, but that sum is dependent on
entering state 1 or 2.
Permanent Disability Model
However, we can go even further and dene
Y(t) =
_
_
_
0 if (x) is alive at time x + t
1 if (x) is disabled at time x + t
2 if (x) is dead at time x + t
0
//
>
>
>
>
>
>
>
>
1
2
Figure: PDM Flow Chart
There is a lump sum paid upon entering state 1, an annuity paid while in
state 1, and a lump sum paid upon entering state 2.
Disability Income Insurance Model
However, we can go even further and dene
Y(t) =
_
_
_
0 if (x) is alive and healthy at time x + t
1 if (x) is alive and sick at time x + t
2 if (x) is dead at time x + t
0
//
>
>
>
>
>
>
>
>
1
oo
2
Figure: DIIM Flow Chart
Premium is paid while in state 0, is a lump sum paid upon entering state
1, an annuity paid while in state 1, and a lump sum paid upon entering
state 2.
Joint Life - Last Survivor
Dene
Joint Life Annuity - annuity that pays until the rst death among a
group of lives
Last Survivor Annuity - annuity that pays until the last death
among a group of lives
A common feature is payment rate decreases upon each death
Reversionary Annuity - life annuity that starts payment upon death
of a specied life, as long as another member of group is alive
Joint Life Insurance - life annuity that starts payment upon rst
death of a member of group
Usually, group consists of two members, a husband and a wife
Joint Model
For example, consider a policy issued to a group (H, W) of age x, y).
Then,
Y(t) =
_
_
0 if H is alive at x + t and W is alive at y + t
1 if H is alive at x + t and W is dead at y + t
2 if H is dead at x + t and W is alive at y + t
3 if H is dead at x + t and W is dead at y + t
0
//
2
//
3
Figure: Joint Model Flow Chart
Notation
Assuming that the group (which can consist of 1,2, or more individuals)
can be found in an of the the n + 1 states 0, 1, 2, ..., n 1, n, we dene
the event
Y(t) = i (195)
to mean the group is in state i at time t.
It follows that Y(t)
t0
is a discrete valued stochastic process.
Assumptions
We make the following assumptions and denitions about transitions
between states and their associated probabilities:
P[Y(x + t) = j [ Y(x) = i ] :=
t
p
ij
x
(Markovity)
P[Y(x + s) = i for all s [0, t] [ Y(x) = i ] :=
t
p
ii
x
lim
h0
P[2 or more transitions in interval of length h]
h
= 0
lim
h0
+
h
p
ij
x
h
:=
ij
x
d
dt
_
t
p
ij
x
_
exists for all t 0
(196)
Assumptions
P[Y(x + t) = j [ Y(x) = i ] :=
t
p
ij
x
(Markovity)
t
p
ii
x
lim
h0
h
= 0
lim
h0
+
h
p
ij
x
h
:=
ij
x
d
dt
_
t
p
ij
x
_
exists for all t 0
(196)
Assumptions
P[Y(x + t) = j [ Y(x) = i ] :=
t
p
ij
x
(Markovity)
t
p
ii
x
lim
h0
h
= 0
lim
h0
+
h
p
ij
x
h
:=
ij
x
d
dt
_
t
p
ij
x
_
exists for all t 0
(196)
Assumptions
P[Y(x + t) = j [ Y(x) = i ] :=
t
p
ij
x
(Markovity)
t
p
ii
x
lim
h0
h
= 0
lim
h0
+
h
p
ij
x
h
:=
ij
x
d
dt
_
t
p
ij
x
_
exists for all t 0
(196)
Assumptions
P[Y(x + t) = j [ Y(x) = i ] :=
t
p
ij
x
(Markovity)
t
p
ii
x
lim
h0
h
= 0
lim
h0
+
h
p
ij
x
h
:=
ij
x
d
dt
_
t
p
ij
x
_
exists for all t 0
(196)
Note:
Note that
t
p
00
x
=
t
p
x
t
p
01
x
=
t
q
x
t
p
10
x
= 0
0
p
ij
x
= 1
{i =j }
01
x
=
x
h
p
ij
x
= h
ij
x
+ o(h)
t
p
ii
x

t
p
ii
x
(197)
Note:
Note that
t
p
00
x
=
t
p
x
t
p
01
x
=
t
q
x
t
p
10
x
= 0
0
p
ij
x
= 1
{i =j }
01
x
=
x
h
p
ij
x
= h
ij
x
+ o(h)
t
p
ii
x

t
p
ii
x
(197)
Example 8.2 3
We can in fact show that
h
p
ii
x
= 1 h
n
j =0,j =i
ij
x
+ o(h)
t
p
ii
x
= exp
_
_
_
t
0
n
j =0,j =i
ij
x+s
ds
_
_
(198)
See proof in book.
As an example, we can show that for the permanent disability model, we
have
u
p
01
x
=
_
u
0
t
p
00
x

01
x+t

ut
p
11
x+t
dt (199)
Example 8.2 3
We can in fact show that
h
p
ii
x
= 1 h
n
j =0,j =i
ij
x
+ o(h)
t
p
ii
x
= exp
_
_
_
t
0
n
j =0,j =i
ij
x+s
ds
_
_
(198)
See proof in book.
As an example, we can show that for the permanent disability model, we
have
u
p
01
x
=
_
u
0
t
p
00
x

01
x+t

ut
p
11
x+t
dt (199)
Kolmogorov Forward Equations
Using the vocabulary of probabilists, we dene the Kolmogorov forward
equations for the evolution of the densities of the birth death Markov
process Y as
d
dt
t
p
ij
x
=
n
k=0,k=j
t
p
ik
x

kj
x+t

t
p
ij
x
n
k=0,k=j
jk
x+t
(200)
In matrix notation, for a xed x, dene the matrices P(t), Q(t) such that
[P(t)]
i ,j
=
t
p
ij
x
Q(t) =
_
_
_
_
_
n
k=1

0k
x+t

01
x+t

0n
x+t
10
x+t

n
k=0,k=1

1k
x+t

1n
x+t
.
.
.
.
.
.
.
.
.
.
.
.
n0
x+t

n1
x+t

n1
k=0

nk
x+t
_
_
_
_
_
(201)
and the corresponding ODE system is
P
(t) = P(t)Q(t)
P(0) = I = Identity Matrix.
(202)
In matrix notation, for a xed x, dene the matrices P(t), Q(t) such that
[P(t)]
i ,j
=
t
p
ij
x
Q(t) =
_
_
_
_
_
n
k=1

0k
x+t

01
x+t

0n
x+t
10
x+t

n
k=0,k=1

1k
x+t

1n
x+t
.
.
.
.
.
.
.
.
.
.
.
.
n0
x+t

n1
x+t

n1
k=0

nk
x+t
_
_
_
_
_
(201)
and the corresponding ODE system is
P
(t) = P(t)Q(t)
P(0) = I = Identity Matrix.
(202)
Here, Q is referred to as the transition intensity matrix. We can work
with o diagonal entries as the diagonal entries are dependent on them.
Also, a whole row of the matrix is lled by zeroes if there is no transition
out of the state corresponding to the row.
Consider the case where Q is time-independent. Also, consider the
diagonalization of Q via
Q = UDU
1
D =
_
_
_
_
_
1
0 0
0
2
0
.
.
.
.
.
.
.
.
.
.
.
.
0 0
n
_
_
_
_
_
(203)
where
1
,
2
, ,
n
are the eigenvalues of Q and U is the matrix
composed of the corresponding eigenvectors.
Then P(t) = e
tQ
P(0), where e
tQ
=
k=0
t
k
k!
Q
k
.
If Q is diagonalizable, then
e
tQ
= Ue
tD
U
1
e
tD
=
_
_
_
_
_
e
t
1
0 0
0 e
t
2
0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 e
t
n
_
_
_
_
_
(204)
Question: What if Q is in fact time dependent?
Example 8.4
Suppose you are given the transition intensity matrix for the permanent
disability model as follows:
_
_
00
x

01
x

02
x
10
x

11
x

12
x
20
x

21
x

22
x
_
_
=
_
_
0.0508 0.0279 0.0229
0.0000 0.0229 0.0229
0.0000 0.0000 0.0000
_
_
(205)
Then
10
p
00
60
=
10
p
00
60
= e
10
0
(0.0279+0.0229)ds
= 0.60170
10
p
01
60
=
_
10
0
t
p
00
60

01
60+t

10t
p
11
60+t
dt
=
_
10
0
e
t
0
(0.0279+0.0229)ds
0.0279 e
10t
0
(0.0229)ds
dt
= 0.19363
(206)
Example 8.5
Suppose you are given the transition intensity matrix for the disability
income insurance model as follows:
_
_
00
x

01
x

02
x
10
x

11
x

12
x
20
x

21
x

22
x
_
_
=
_
_
01
x

02
x
a
1
+ b
1
e
c
1
x
a
2
+ b
2
e
c
2
x
0.1
_
01
x
_

10
x

12
x
a
2
+ b
1
e
c
2
x
0 0 0
_
_
(207)
for parameters
_
a
1
b
1
c
1
a
2
b
2
c
2
_
=
_
4 10
4
3.4674 10
6
0.138155
5 10
4
7.5858 10
6
0.087498
_
(208)
Example 8.5
Then Forward Euler applied the Kolmogorov equations leads to
t+h
p
00
60
=
t
p
00
60
h
t
p
00
60
01
60+t
+
02
60+t
_
+ h
t
p
01
60

10
60+t
+ o(h)
t+h
p
01
60
=
t
p
01
60
h
t
p
01
60
12
60+t
+
10
60+t
_
+ h
t
p
00
60

10
60+t
+ o(h)
(209)
Ignoring the o(h) terms, we can iterate forward using, for example, h =
1
12
.
Example 8.5
In matrix-vector notation, we have
_
t+h
p
00
60
t+h
p
01
60
_
= [I hA(t)]
_
t
p
00
60
t
p
01
60
_
I =
_
1 0
0 1
_
A(t) =
_
01
60+t
+
02
60+t

10
60+t
10
60+t

12
60+t
+
10
60+t
_
(210)
Keep in mind that A is determined by the given transition intensity matrix.
HW Compute this vector over the interval t [0, 10] using a time step of
h =
1
12
. Use any numerical solver you like, but please have the values
computed into a pair of columns.
Example 8.5
_
t+h
p
00
60
t+h
p
01
60
_
= [I hA(t)]
_
t
p
00
60
t
p
01
60
_
I =
_
1 0
0 1
_
A(t) =
_
01
60+t
+
02
60+t

10
60+t
10
60+t

12
60+t
+
10
60+t
_
(210)
h =
1
12
Example 8.5
_
t+h
p
00
60
t+h
p
01
60
_
= [I hA(t)]
_
t
p
00
60
t
p
01
60
_
I =
_
1 0
0 1
_
A(t) =
_
01
60+t
+
02
60+t

10
60+t
10
60+t

12
60+t
+
10
60+t
_
(210)
h =
1
12
Premiums
Consider an annuity issued to a life (x) that pays at rate 1 per year
continuously while the life is in state j . Then the EPV of this annuity at
force of interest per year is
a
ij
x
= E
__

0
e
t
1
{Y(t)=j |Y(0)=i }
dt
_
=
_

0
e
t
E
_
1
{Y(t)=j |Y(0)=i }
dt =
_

0
e
t
t
p
ij
x
dt
(211)
If the annuity is payable at the start of each year from the current time,
based on the conditional event Y(t) = j [ Y(0) = i , then the EPV is
a
ij
x
=
k=0
v
k
k
p
ij
x
(212)
Premiums
If a unit benet is payable to a life (x) on transition to state k, given that
it is currently in state i , then the EPV of this benet is
A
ik
x
=
_

0
j =k
e
t
t
p
ij
x
jk
x+t
dt (213)
Example 8.6
An insurer issues a 10year disability income insurance policy to a healthy
life aged 60. Use the model and parameters from Example 8.5. Assume an
eective rate of 5% per year and no expenses. Calculate the premiums for
the following designs
(a) Premiums are payable continuously while in the healthy state. A
benet of 20000 per year is payable continuously while in the disabled
state. A death benet of 50000 is payable immediately upon death.
(b) Premiums are payable monthly in advance conditional on the life
being in the healthy state at the premium date. The sickness benet
of 20000 per year is payable monthly in arrear, if the life is in the sick
state at the payment date. A death benet of 50000 is payable
immediatlely upon death.
Example 8.6
For case a.), we have via the EPP principle that
20000a
01
60:10
+ 50000
A
02
60:10
Pa
00
60:10
= 0 (214)
and so
P =
20000
_
10
0
e
t
t
p
01
60
dt
_
10
0
e
t
t
p
00
60
dt
+
50000
_
10
0
e
t
_
t
p
00
60
02
60+t
+ p
01
60
12
60+t
_
dt
_
10
0
e
t
t
p
00
60
dt
.
(215)
For a time step of h =
1
12
(monthly), we can use the forward-Euler results
from the previous example to calculate P = 3254.65
Example 8.6
For case b.), we have via the EPP principle that
20000a
(12)01
60:10
+ 50000
A
02
60:10
Pa
(12)00
60:10
= 0 (216)
and so
P =
20000
119
k=0
v
k
12
k
12
p
01
60
119
k=0
v
k
12
k
12
p
00
60
+
50000
_
10
0
e
t
_
t
p
00
60
02
60+t
+ p
01
60
12
60+t
_
dt
119
k=0
v
k
12
k
12
p
00
60
.
(217)
Again, we can use the previously computed values of
_
k
12
p
0j
60
_
(1,119)
(j ,k)=(0,0)
to
calculate P = 3257.20
Policy Values
Dene
ij
y
as the transition intensity between states i and j at age y
t
as the force of interest per year at time t
B
(i )
t
as the benet payment rate while the policyholder is in state i
S
(ij )
t
as the lump sum payment instantaneously at time t on transition
from state i to state j .
Assume the above are all members of C
0
[0, n]. Then i 0, 1, , n,
Thieles Dierential Equation is
d
dt
t
V
i
=
t

t
V
i
B
(i )
t

j =i
ij
x+t
_
S
(ij )
t
+
t
V
j
t
V
i
_
(218)
Multiple Decrement Models
Consider the special case for transition matrix
Q(t) =
_
_
_
_
_
n
k=1

0k
x+t

01
x+t

0n
x+t
0 0 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0
_
_
_
_
_
(219)
Here, there are multiple exits from state 0, but no further transitions.
0

A
A
A
A
A
A
A
A
''
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
1 2 n
Figure: Multiple Decrement Flow Chart
Multiple Decrement Models
t
p
00
x
=
t
p
00
x
= exp
_
_
t
0
n
i =1
0i
x+s
ds
_
t
p
0i
x
=
_
t
0
s
p
00
x

0i
x+s
ds
0
p
ij
x
= 1
{i =j }
(220)
Note:
Premium is now dierent when compared to a policy that only allows
transition 0 1.
This is because
00
x+t
=
n
k=1

0k
x+t
<
01
x+t
and so
t
p
00
x
changes
accordingly.
See Example 8.8, where for example an insurer may allow for lower
premiums via lapse support.
Joint Life - Last Survivor Benets
0
01
x+t:y+t
//
02
x+t:y+t
13
x+t
23
y+t
//
3
Figure: Joint Model Transition Rates
Dene the joint transition matrix via the ow chart above.
Dene
t
p
xy
=
t
p
00
xy
= P[(x) and (y) are both alive in t years]
t
q
xy
=
t
p
01
xy
+
t
p
02
xy
+
t
p
03
xy
= P[(x) and (y) are not both alive in t years]
t
p
xy
=
t
p
00
xy
+
t
p
01
xy
+
t
p
02
xy
= P[ at least one of (x) and (y) is alive in t years]
t
q
xy
= 1
t
p
xy
x+t:y+t
=
01
x+t:y+t
+
02
x+t:y+t
(221)
Also dene
t
q
1
xy
= P[(x) dies before (y) and within t years]
=
_
t
0
r
p
00
xy
02
x+r :y+r
dr
,=
t
p
02
xy
t
q
2
xy
= P[(x) dies after (y) and within t years]
=
_
t
0
r
p
01
xy
13
x+r
dr
(222)
Insurance Notation
a
xy
= a
00
xy
, the Joint Life Annuity with continuous payment of 1 per
year while both husband and wife are alive.
A
xy
the Joint Life Insurance with a unit payment immediately upon
rst death.
A
1
xy
, the Contingent Insurance, a unit payment immediately upon
death of the husband given that he dies before his wife.
Insurance Notation
A
xy
=

A
03
xy
, the Last Survivor Insurance with unit payment
immediately upon second death.
a
x|y
= a
02
xy
the Reversionary Annuity with a continuous payment at
unit rate per year while wife is alive given that husband has died..
a
xy
= a
00
xy
+ a
01
xy
+ a
02
xy
, the Last Survivor Annuity, a continuous
payment at rate 1 per year while at least one person is alive.
It can be shown that
a
xy
= a
x
+ a
y
a
xy
a
x|y
= a
y
a
xy
A
xy
=

A
x
+

A
y

A
xy
a
xy
=
1
A
xy
(223)
HW
Prove this using explicit integral formulations.
Read over Examples 8.10 and 8.11.
Transitions at Specic Ages
Example 8.12 : The employees of a large corporation can leave the
corporation in thre ways: they can withdraw from the corporation, they can
retire, or they can die while they are still employees. Consider the model
03
x

13
x

23
x
=
x
02
x
=
_
02
, if x < 60
0, if x 60
(224)
where retirement can take place only on an employees
60
th
, 61
st
, 62
nd
, 63
rd
, 64
th
, or 65
th
birthday. Assume that 40% of
employees reaching exact age 60, 61, 62, 63 or 64 will retire at that age
and that 100% of all employees who reach age 65 retire immediately.
The corporation oers the following benets to the employees:
For those employees who die while still employed, a lump sum of
200000 is payable immediately upon death.
For those employees who retire, a lump sum of 150000 is payable
immediately upon death after retirement.
Theorem
Assuming a constant force of interest of per year and the notation of

A
x
and
n
E
x
from single life computations based on a force of mortality
x
, it
follows that the EPV of the Death after retirement benet of an
employee currently aged 40 is
150000
20
E
40
e
20
02
_
0.4
_
4
k=0
0.6
k
k|
A
60
_
+ 0.6
5
5|
A
60
_
(225)
To begin our proof, we can compute
E
(40)
[PV(Benet) [ retire at age 60] = 150000e
20
A
60
(226)
20
p
00
40
= exp
_
_
20
0
_
02
+
03
40+t
_
dt
_
= exp
_
_
20
0
(
40+t
) dt
_
e
20
02
=
20
p
40
e
20
02
P
(40)
[retire at age 60] = 0.4
20
p
00
40
= 0.4
20
p
40
e
20
02
20
+p
00
40
= 0.6
20
p
00
40
21
p
00
40
=
20
+p
00
40
p
60
21
+p
00
40
= 0.6
21
p
00
40
= 0.6
2
21
p
40
e
20
02
(227)
Also,
P
(40)
[retire at age 61] = X
1
X
2
X
3
X
4
X
1
= P
(40)
[survive in employment to age 60
] =
20
p
00
40
X
2
= P
(40)
[will not retire at age 60] = 0.6
X
3
= P
(40)
[(60
+
) will survive to age 61
] =
1
p
60
X
4
= P
(40)
[will retire at age 61] = 0.4
E
(40)
21
A
61
(228)
We can repeat this until
P
(40)
[retire at age 65]
=
20
p
00
40
0.6
5
1
p
60
1
p
61
1
p
62
1
p
63
1
p
64
=
25
p
40
e
20
02
0.6
5
E
(40)
25
A
65
(229)
We now have enough information to complete the proof.
Proof.
For benet after retirement, we have
E
(40)
[PV(Benet)] =
5
k=0
B
k
(40)
P
k
(40)
B
k
(40)
= E
(40)
[PV(Benet) [ retire at age 60 + k]
= 150000e
(20+k)
A
60+k
for k 0, , 5
P
k
(40)
:= P
(40)
[retire at age 60 + k]
=
20+k
p
40
e
20
02
0.6
k
0.4 for k 0, , 4
P
5
(40)
=
25
p
40
e
20
02
0.6
5
(230)
Substitution and arithmetic lead to the form in the theorem statement.
Also notice that via the Tower property for conditional expectations, we
have a direct version of the proof:
E
(40)
[PV(Benet)] = E
(40)
_
E
(60)
[PV(Benet)]
=
20
E
40
e
20
02
k=0
B
k
(60)
P
k
(60)
(231)
where
B
k
(60)
= E
(60)
[PV(Benet) [ retire at age 60 + k]
= 150000e
k
A
60+k
for k 0, , 5
P
k
(60)
:= P
(60)
[retire at age 60 + k]
=
k
p
60
0.6
k
0.4 for k 0, , 4
P
5
(60)
=
5
p
60
0.6
5
(232)
Homework Questions
HW: 8.1, 8.2, 8.4, 8.5, 8.8, 8.10, 8.11, 8.15, 8.16, 8.21, 8.22
Plan Type
We consider two types of retirement plans.
A Dened Contribution plan species how much an employer will
contribute, as a percentage of salary, into a plan.
A Dened Benet plan species a level of benet, most likely
related to the employees salary near retirement. Here, contributions
may need to be updated based on the investment returns to ensure
that the benet is met.
Replacement Ratio
Also, dening the Replacement Ratio
R :=
pension income in the year after retirement
salary in the year before retirement
(233)
the benet under DB plans and target under DC plans may aim for
R (0.5, 0.7). (234)
This of course assumes the member survives the year following retirement.
Salary Scale Function
We can also use a deterministic model to dene the salary scale s
y
yx
0
beginning at some suitiable initial age x
0
where the ratio
s
y
s
x
=
salary received in year y to y + 1
salary received in year x to x + 1
(235)
and, assuming that salaries are increased continuously, the salary rate at
age x as s
x
1
2
.
Example 9.1
The nal average salary for the pension benet provided by a pension plan
is dened as the average salary in the three years before retirement.
Members salaries are increased each year, six months before the valuation
date
A member aged exactly 35 at the valuation date received 75000 in
salary in the year to the valuation date. Calculate his predicted nal
average salary assuming retirement at age 65.
A member aged exactly 55 at the valuation date was paid salary at a
rate of 100000 per year at that time. Calculate her predicted nal
average salary assuming retirement at age 65.
Assume a salary scale where s
y
= 1.04
y
.
Example 9.1
For rst case
s
avg
= 75000
1
3
s
62
+ s
63
+ s
64
s
34
=
75000
3

_
1.04
28
+ 1.04
29
+ 1.04
30
_
= 234019
(236)
For second case
s
avg
= 100000
1
3
s
62
+ s
63
+ s
64
s
54.5
=
100000
3

_
1.04
7.5
+ 1.04
8.5
+ 1.04
9.5
_
= 139639
(237)
Now read Example 9.2 for more practice and Example 9.3 for setting the
DC contribution.
Stochastic Pension Model
We can dene a multiple decrement model for a pension plan via states
Y(t) =
_
_
0 if (x) is a member at age x + t
1 if (x) has withdrawn by time x + t
2 if (x) has retired due to disability by age x + t
3 if (x) has retired due to age at x + t
4 if (x) has died in service by age x + t
0

>
>
>
>
>
>
>
>
''
N
N
N
N
N
N
N
N
N
N
N
N
N
N
1 2 3 4
Figure: Pension Plan Flow Chart. In a DC plan, benet on exit is the same.
However, in a DB plan dierent benets may be payable on dierent forms of exit.
Example 9.4
A pension plan member is entitled to a lump sum benet on death in
service of four times the salary paid in the year up to death. Assuming the
multiple decrement model with
01
x
=
w
x
02
x
=
i
x
= 0.001
03
x
=
r
x
04
x
=
d
x
= A + Bc
x
= 0.00022 + (2.7 10
6
) 1.124
x
(238)
Example 9.4
Assume
w
x
=
_
_
0.1, if x < 35
0.05, if 35 x < 45
0.02, if 45 x < 60
0, if x 60
r
x
=
_
0 if x < 60
0.1, if 60 < x < 65
and
P[(x) retires at (60) [ survives in employment to (60)] = 0.3
P[(x) retires at (65) [ survives in employment to (65)] = 1
(239)
Example 9.4
Calculate, for a member aged 35, the probability of retiring at age 65.
Notice the similarities to Example 8.12.
For t (0, 10), we have
t
p
00
35
= exp
_
_
t
0
_
w
35+s
+
i
35+s
+
r
35+s
+
d
35+s
_
ds
_
= exp
_
_
t
0
_
0.05 + 0.001 + 0 + A + Bc
35+s
_
ds
_
= exp
_
0.05122t +
2.7 10
6
ln (1.124)
1.124
35
_
1.124
t
1
_
_
(240)
It follows that
10
p
00
35
= 0.597342
Example 9.4
Calculate, for a member aged 35, the probability of retiring at age 65.
Notice the similarities to Example 8.12.
For t (0, 10), we have
t
p
00
35
= exp
_
_
t
0
_
w
35+s
+
i
35+s
+
r
35+s
+
d
35+s
_
ds
_
= exp
_
_
t
0
_
0.05 + 0.001 + 0 + A + Bc
35+s
_
ds
_
= exp
_
0.05122t +
2.7 10
6
ln (1.124)
1.124
35
_
1.124
t
1
_
_
(240)
It follows that
10
p
00
35
= 0.597342
Example 9.4
For t [10, 25), we compute
t
p
00
35
10
p
00
35
= exp
_
_
t10
0
_
w
45+s
+
i
45+s
+
r
45+s
+
d
45+s
_
ds
_
= exp
_
_
t
0
_
0.02 + 0.001 + 0 + A + Bc
45+s
_
ds
_
= exp
_
0.02122(t 10) +
2.7 10
6
ln (1.124)
1.124
35
_
1.124
t10
1
_
_
(241)
It follows that
25
p
00
35
= 0.597342 0.712105 = 0.425370
25
+p
00
35
= 0.7
25
p
00
35
= 0.297759
(242)
Example 9.4
For t (25, 30), we compute
t
p
00
35
25
+
p
00
35
= exp
_
_
t25
0
_
w
60+s
+
i
60+s
+
r
60+s
+
d
60+s
_
ds
_
= exp
_
0.10122(t 25) +
2.7 10
6
ln (1.124)
1.124
60
_
1.124
t25
1
_
_
(243)
It follows that the probability of retirement at exact age 65 is
30
p
00
35
=
_
25
+p
00
35
_
(0.590675)
= 0.175879
(244)
Now calculate: P
35
[withdrawal], P
35
[retirement], P
35
[disability retirement]
and P
35
[death in service].
Service Table
We can represent the multiple decrement model for pensions in tabular
form. Begin by dening a minimum integer entry age x
0
and
corresponding arbitrary radix (cohort) l
x
0
. With these, we can organize a
table with entries
w
x
0
+k
= l
x
0
k
p
00
x
0
p
01
x
0
+k
i
x
0
+k
= l
x
0
k
p
00
x
0
p
02
x
0
+k
r
x
0
+k
= l
x
0
k
p
00
x
0
p
03
x
0
+k
d
x
0
+k
= l
x
0
k
p
00
x
0
p
04
x
0
+k
l
x
0
+k
= l
x
0
k
p
00
x
0
(245)
and it follows that
l
x
= l
x1
w
x1
i
x1
r
x1
d
x1
(246)
Service Table
It follows that we can use the service table to answer questions like
P
35
[withdraws] =
24
k=0
w
35+k
l
35
P
35
[retires in ill health] =
29
k=0
i
35+k
l
35
P
35
[retires for age reasons] =
30
k=0
r
35+k
l
35
P
35
[dies in service] =
29
k=0
d
35+k
l
35
(247)
For long-horizon investments with uncertain returns (forecasts may only be
valid for a small horizon), using tabular methods with UDD approximation
is common in pension valuation. See Example 9.5 for a comparison
with exact methods.
Valuation:Contributions
Employees in a pension plan pay contributions of 6% of their previous
months salary at each month end until age 60. Calculate the EPV at
entry of contributions for a new entrant aged 35, with a starting salary
rate of 100000 using the model
01
x
= ,
02
x
= ,
03
x
= 0 and
04
x
=
for x (35, 60). Assume a constant force of interest and a salary scale
function s
y
= e
y
for y (35, 60).
Valuation
Per month, the contribution amount is a scaling of 0.06
100000
12
= 500. It
follows that
E[PV(Contributions)] = 500
300
k=1
k
12
p
00
35
e
k
12
e
k
12
= 500
300
k=1
e
(++)
k
12
e
k
12
e
k
12
= 500
e
x
1 e
x
1 e
300x
x =

12
(248)
Valuation:Benets
For a DB plan, the basic annual pension benet is equal to n S
Fin
,
where n is the total number of years of service, S
Fin
is the average salary
in a specied period before retirement (ie. three years preceding exit) and
is the accrual rate, usually between 0.01 and 0.02.
Valuation:Benets; Example 9.6
Estimate the EPV of the accrued age retirement pension benet for a
member aged 55 with 20 years of service, whose salary in the year prior to
the valuation date was 50000. Assume that mid-year age retirements
happen at exactly halfway into the year. Assume the nal average salary is
dened as the earnings in the three years before retirement. Describe this
value by using elements of a corresponding service table.
Note that for this problem,
n S
Fin
= 50000 0.015 20 = 15000
z
y
=
s
y3
+ s
y2
+ s
y1
3
(249)
as well as
E[Projected Final Salary [ Retirement at age y] = 50000
z
y
s
54
. (250)
E[PV(Benets)] = 15000
_
r
60
l
55
z
60
s
54
v
5
a
(12)
60
+
r
65
l
55
z
65
s
54
v
10
a
(12)
65
_
+ 15000
r
60
+
l
55
z
60.5
s
54
v
5.5
a
(12)
60.5
+ 15000
4
k=1
r
60+k
l
55
z
60.5+k
s
54
v
5.5+k
a
(12)
60.5+k
z
y
=
s
y3
+ s
y2
+ s
y1
3
(251)
One can program this using numerical software, using linear interpolation
for mid-year quantities. Read Examples 9.8, 9.9 for a discussion on
withdrawal pension.
Plan Funding
Assuming that all employer contributions to a fund are paid the start of
the year, that there are no employee contributions, and any benets
payable during the year are paid exactly half-way though the year, we
dene the normal contribution due at the start of the year t to t + 1 for
a member aged x at time t as C
t
. Using reserving principles studied
earlier, we have the equation
t
V + C
t
= E[PV(Benets for mid-year exits)] + v
1
p
00
x
t+1
V (252)
Example 9.9
Assume a pension plan with the following valuation methods:
Accrual rate: 1.5%
Final salary plan
Pension based on earnings in the year before age retirement
Normal retirement at age 65
The pension benet is a life annuity payable monthly in advance
There is no benet due on death in service
No exits other than by death before normal retirement age
Example 9.9
Calculate the value of the accrued pension benet and normal contribution
due at the start of the year using a projected unit funding (PUC), where
interest is set at 5% per year, salaries increase at 4% per year and assume
a constant mortality before and after retirement.
S
Fin
= 50000
s
64
s
49
= 50000(1.04)
15
= 90047
0
V = 0.015 20 S
Fin
15
p
50
v
15
a
(12)
65
= 12994.24 e
15
v
15
1
12
k=0
v
k
12
e
k
12
=
12994.24
12e
15
1.05
15
_
1
12
_
1
1.05e
_
(253)
Example 9.9
Calculate the value of the accrued pension benet and normal contribution
due at the start of the year using a projected unit funding (PUC), where
interest is set at 5% per year, salaries increase at 4% per year and assume
a constant mortality before and after retirement.
S
Fin
= 50000
s
64
s
49
= 50000(1.04)
15
= 90047
0
V = 0.015 20 S
Fin
15
p
50
v
15
a
(12)
65
= 12994.24 e
15
v
15
1
12
k=0
v
k
12
e
k
12
=
12994.24
12e
15
1.05
15
_
1
12
_
1
1.05e
_
(253)
Example 9.9
It follows that our equation for C is
1
V = 0.015 21 S
Fin
14
p
51
v
14
a
(12)
65
C = v p
51
1
V
0
V =
21
20
0
V
0
V =
0
V
20
(254)
Consider the traditional unit credit funding approach, and see how this
aects our previous calculation. Also, read over Example 9.10 which allows
for benets payable on exit during the year.
Homework Questions
HW: 9.1, 9.3, 9.5, 9.7, 9.9, 9.12, 9.13
Law of Total Variance
Recall that for two random variables X, Y in a probability space (, T, P)
we have the Tower Property
E
_
E[X [ Y]
_
= E[X]
E
_
E[Y [ X]
_
= E[Y]
(255)
and so it follows that
V[X] = E
_
X
2
_
E[X]
_
2
= E
_
E
_
X
2
[ Y
_
E
_
E[X [ Y]
__
2
= E
_
V [X [ Y] +E[X [ Y]
2
_
_
E
_
E[X [ Y]
__
2
= E
_
V [X [ Y]
_
+ V
_
E[X [ Y]
_
(256)
Deviation as a Risk Measure
Assuming a sequence of i.i.d. Random Variables
_
X
k
_
n
k=1
, one measure
of the risk associated to the average
X
n
:=
1
n
n
k=1
X
k
(257)
is the total variance
n
(X) := V
_
X
n
_
(258)
Correspondingly, we say that such a risk is Diversiable if
lim
n

n
(X) = 0, and not diversiable otherwise.
Note that if
_
X
k
_
n
k=1
are dependent but otherwise identically distrbuted
with correlation coecient , mean and variance
2
, then
n
(X) =
n
2
+ n(n 1)
2
n
2

2
,= 0 (259)
For a history of variance as a risk measure in Modern Portfolio Theory and
the corresponding use of diverscation, click here and references within.
Recall that for any random variable Y and i.i.d. sequence
_
X
k
_
n
k=1
with
identical copy X
V
_
1
n
n
k=1
X
k
_
= E
_
V
_
1
n
n
k=1
X
k
[ Y
__
+ V
_
E
_
1
n
n
k=1
X
k
[ Y
__
=
1
n
E
_
V
_
X [ Y
__
+ V
_
E
_
X [ Y
__
(260)
It follows that
n
(X) 0 as long as V
_
E
_
X [ Y
__
= 0.
_
X
k
_
n
k=1
with
identical copy X
V
_
1
n
n
k=1
X
k
_
= E
_
V
_
1
n
n
k=1
X
k
[ Y
__
+ V
_
E
_
1
n
n
k=1
X
k
[ Y
__
=
1
n
E
_
V
_
X [ Y
__
+ V
_
E
_
X [ Y
__
(260)
It follows that
n
(X) 0 as long as V
_
E
_
X [ Y
__
= 0.
_
X
k
_
n
k=1
with
identical copy X
V
_
1
n
n
k=1
X
k
_
= E
_
V
_
1
n
n
k=1
X
k
[ Y
__
+ V
_
E
_
1
n
n
k=1
X
k
[ Y
__
=
1
n
E
_
V
_
X [ Y
__
+ V
_
E
_
X [ Y
__
(260)
It follows that
n
(X) 0 as long as V
_
E
_
X [ Y
__
= 0.
Connection with CLT
Note that by the Central Limit Theorem,
lim
n
P
_

X
n
_
= lim
n
_
= 0 (261)
This says that for uncorrelated r.v.s, since the variance of the aggregate
mean is linear in n, we have the deviation of the aggregate mean from the
individual mean asymptotically disappears.
Note that if our sequence is correlated, then there is the adjust CLT that
states the above, except the limit is now
n
_
X
n
_
_
_
2
+
i =2
Cov[X
1
, X
i
]
_
Z N(0, 1) (262)
Example of Diversiable Risk
Consider the case where we have an i.i.d. sequence
_
X
k
_
n
k=1
X
k
0, 1
P[X
k
= 1] =
t
p
x
(1
s
p
x+t
).
(263)
It follows that
X
n
=
1
n
n
k=1
X
k
(264)
models the sample probability of deaths of a population of n alive at age x
where death occurs between age x +t and age x +t +s. This is of course
a binomial random variable with p =
t
p
x
(1
s
p
x+t
).
Example of Diversiable Risk
We can see that for a copy X of the sequence,
V
_
X
n
_
=
V
_
n
k=1
X
k
_
n
2
=
nV[X]
n
2
=
1
n

_
t
p
x
(1
s
p
x+t
)
_
_
1
t
p
x
(1
s
p
x+t
)
_
0
(265)
and so the risk is diversiable.
Example of Non-Diversiable Risk
Consider now the case where the X
k
model the loss associated with a
member of an insured population. If each member has loss function X
k
and the premiums are charged in keeping with the EPP, then we expect
that E
_
X
k
_
= 0 for all k 1, . . . , n .
If, however, the forecasted yield rate used is a random variable Y, then if
E[X
k
[ Y] ,= 0 we have non-diversiable risk as
V
_
X
n
_
V
_
E
_
X [ Y
__
,= 0. (266)
Example of Non-Diversiable Risk
Consider now the case where the X
k
model the loss associated with a
member of an insured population. If each member has loss function X
k
and the premiums are charged in keeping with the EPP, then we expect
that E
_
X
k
_
= 0 for all k 1, . . . , n .
If, however, the forecasted yield rate used is a random variable Y, then if
E[X
k
[ Y] ,= 0 we have non-diversiable risk as
V
_
X
n
_
V
_
E
_
X [ Y
__
,= 0. (266)
Homework Questions
HW: 10.1, 10.2, 10.3, 10.5, 10.6
Reserves
Recall the need for policy values when negative future cash ows were
expected. In this lecture, we cover the idea of reserves, which is the actual
amount of money held by the insurer to cover future liabilities associated
with contracts.
Reserves
The insurer may decide to set aside assets in reserve as equal to the net
premium policy values on a certain (reserve) basis.
For example, consider an nyear term insurance contract issued to a life x
with sum insured S. Since we use the net premium basis to compute xed
premiums, it follows that
P = S
A
1
x:n
a
x:n
R
t
=
t
V = SA
1
x+t:nt
Pa
x+t:nt
= SA
1
x:n

_
A
1
x+t:nt
A
1
x:n
a
x+t:nt
a
x:n
_
(267)
The cost of setting up, from t 1 to t, the reserve amount of
t
V is at
time t equal to
t
V p
x+t1
when valued at time t
i.e. the proportion of contracts that survive to the end of the year.
Reserves
P = S
A
1
x:n
a
x:n
R
t
=
t
V = SA
1
x+t:nt
Pa
x+t:nt
= SA
1
x:n

_
A
1
x+t:nt
A
1
x:n
a
x+t:nt
a
x:n
_
(267)
t
V is at
time t equal to
t
V p
x+t1
Reserves
P = S
A
1
x:n
a
x:n
R
t
=
t
V = SA
1
x+t:nt
Pa
x+t:nt
= SA
1
x:n

_
A
1
x+t:nt
A
1
x:n
a
x+t:nt
a
x:n
_
(267)
t
V is at
time t equal to
t
V p
x+t1
Reserves
P = S
A
1
x:n
a
x:n
R
t
=
t
V = SA
1
x+t:nt
Pa
x+t:nt
= SA
1
x:n

_
A
1
x+t:nt
A
1
x:n
a
x+t:nt
a
x:n
_
(267)
t
V is at
time t equal to
t
V p
x+t1
Notation
At time t, just before and just after, we have quantities that are assets
and costs.
At time (t 1)
+
, we have the cost E
t
associated from t 1 to t.
Between (t 1)
+
and t
, we have the payout S settled at time t
with expected value S q

x+t1
.
At time (t 1)
+
, we have the asset
t1
V which grows at the interest
rate i to value (1 + i )
t1
V at time t
At time (t)
, we have the cost

t
V p
x+t1
which grows at the
interest rate i to value (1 + i )
t1
V at time t
Prots
Correspondingly, we can set up an equation for the prot at time t,
denoted by Pr
t
:
Pr
t
=
_
t1
V + P E
t
_
(1 + i ) Sq
x+t1
t
Vp
x+t1
(268)
The Prot Vector
Pr :=
_
Pr
0
, . . . , Pr
n
_
(269)
is comprised of elements that represent the expected prot at the end of
the year given that the policy is in eect at the start of the year.
Prots
The Prot Signature is the the vector

comprised of elements
t
:=
t1
p
x
Pr
t
(270)
that represent the expected prot at the end of the year given that the
policy was in eect at age x.
Prot Measures
Recall that for any set of cash ows C
t
, the internal rate of return IRR (if
it uniquely exists) is the interest rate j such that
n
t=0
C
t
(1 + j )
t
= 0. (271)
In accordance with the IRR, the insurer may set a minimum hurdle or risk
discount rate r such that the contract is satisably protable if IRR > r .
EPV of Future Prot
If the IRR does not exist, the insurer may seek to measure the protability
via the Net Present Value computed using the risk discount rate:
NPV :=
n
t=0
t
(1 + r )
t
(272)
Prot Margin and DPP
Another measure is the ratio of NPV to E[PV(Premiums)]:
Prot Margin :=
NPV
E[PV(Premiums)]
(273)
as is the discounted payback period DPP:
DPP := min
_
m :
m
t=0
t
(1 + r )
t
0
_
(274)
which represents the time until the insurer starts to make a prot.
A question naturally arises of how to jointly measure interest risk and
prot. One may even compute the marginal changes in the prot measures
with respect to change in risk discount factor r .
Homework Questions
Read Example 11.1
HW: 11.1, 11.3, 11.6, 11.7
Equity Linked Insurance
Modern insurance contracts can include some form of guarantee. These
are known in America as Variable Annuities and Segregated Funds in
Canada. The accumulating premiums the policyholder pays is invested on
the policyholders behalf. These premiums form the policyholders fund,
from which regular management charges are deducted by the insurer and
paid into the insurers fund to cover expenses and insurance charges.
On survival to the end of the contract term the benet may be just the
policyholders fund and no more, or there may be a guaranteed minimum
maturity benet (GMMB). There may also be a guaranteed minimum
death benet (GMDB).
There are very real consequences to the dierences between nancial
pricing and actuarial reserving. A short but excellent analysis can be found
in the paper by Bangwon Ko and Elias S. W. Shiu on Financial Pricing
and Actuarial Reserving.
Also consider A Heavy Trac Approach to Modeling Large Life
Insurance Portfolios (Stochastic modeling of actuarial reserve, with Ito
integration of a time-changed Brownian Bridge.)
We follow the example set by Shiu and Ko now.
On survival to the end of the contract term the benet may be just the
policyholders fund and no more, or there may be a guaranteed minimum
maturity benet (GMMB). There may also be a guaranteed minimum
death benet (GMDB).
There are very real consequences to the dierences between nancial
pricing and actuarial reserving. A short but excellent analysis can be found
in the paper by Bangwon Ko and Elias S. W. Shiu on Financial Pricing
and Actuarial Reserving.
Also consider A Heavy Trac Approach to Modeling Large Life
Insurance Portfolios (Stochastic modeling of actuarial reserve, with Ito
integration of a time-changed Brownian Bridge.)
We follow the example set by Shiu and Ko now.
Stochastic Actuarial Reserving
Fix a probability space
_
, T, P
_
and a standard Brownian motion W
that lives on this space.
Consider now a term contact with term T and let denote the
management charges factor along with representing the policyholders
participation factor.
Furthermore, assume mean and standard deviation parameters (, )
respectively and the corresponding Geometric Brownian Mutual Fund Asset
S
t
= S
0
e
t+W
t
. (275)
Using this as the model of the asset returns upon which premiums are
invested, the policyholder wishes to purchase a contract that pays a
maturity benet credited at a rate of return which is the greater of
the customers risk discount rate r , where r < or
the participation rate of the stock index returns of S.
Symbolically, for a current premium P invested in the , the contract
payout value at maturity is
V(T) = (1 )P max
_
e
rT
, 1 +
_
S
T
S
0
1
_
_
. (276)
Assume that the policyholder is able to fully participate in the returns from
the fund (i.e. = 1.)
Then
V(T) = (1 )P
S
T
S
0
+ (1 )P max
_
e
rT
_
S
T
S
0
_
, 0
_
:= V
1
(T) + V
2
(T).
(277)
Here, V
1
(T) is the net premium, or payo, for investing in the index fund
and V
2
(T) is the guaranteed option payo if the index fund
under-performs relative to the risk discount rate r .
How does one reserve to meet the obligations of V
2
(T).
One can see that the probability of a payout, that V
2
(T) ,= 0 is for large T
P[V
2
(T) ,= 0] = P[rT > T + W
T
] =
_
r
T
_
0. (278)
Since it is a low probability event that we have to prepare for a payout
V
2
(T) and since we can directly replicate the payo V
1
(T) by initially
purchasing
(1)P
S
0
units of the index fund, an actuary may be tempted to
not reserve for the uncertain portion of the guarantee, V
2
(T), if the
contract has a relatively long term T.
Is this a wise decision?
2
P[V
2
(T) ,= 0] = P[rT > T + W
T
] =
_
r
T
_
0. (278)
V
2
1
(T) by initially
purchasing
(1)P
S
0
2
(T), if the
2
P[V
2
(T) ,= 0] = P[rT > T + W
T
] =
_
r
T
_
0. (278)
V
2
1
(T) by initially
purchasing
(1)P
S
0
2
(T), if the
Stochastic Reserving for non-diversiable risk
Given a random loss L, we dene the quantile reserve, also known as the
Value at Risk with parameter , as the amount which with probability
will not be exceeded by the loss.
Symbolically, if L has a continuous distribution function F
L
, then the
quantile reserve is Q
where
P[L Q
] = . (279)
Stochastic Reserving for non-diversiable risk
One feature that is missing in VaR is the description of what the loss could
be if it does exceed the quantile Q
. In this case, the Conditional Tail

Expectation (CTE
) is dened as
CTE
= E[L [ L Q
]. (280)
A risk manager should not rely on static measures of risk involved with a
portfolio of liabilities. Rather, the CTE or VaR reserve should be regularly
updated to incorporate market information as it arrives. This allows
reserves which are held in less-risky (and possibly more liquid) funds to be
invested higher return and higher risk assets if current market information
dictates that CTE reserves can be reduced.
Homework Questions
Read Example 12.1 and Table 12.8
HW: 12.1 12.5

stt455 - Actuarial Models PDF

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

stt455 - Actuarial Models PDF

Загружено:

Авторское право:

Доступные форматы

STT 455-6: Actuarial Models

(0) + ... (95)

for a nite lifetime

for a nite lifetime

, we have the payout S settled at time t

with expected value S q

, we have the cost

. In this case, the Conditional Tail

Вам также может понравиться