Introduction to Life Contingencies

Chris Groendyke, FSA

Copyright © 2012, Society of Actuaries

Contents
1. Introduction ........................................................................................................................................ 1

2. Life Tables and Expected Future Lifetimes ........................................................................... 4

2.1. Introduction................................................................................................................................ 4

2.2. Life Tables ................................................................................................................................... 4

2.3. Expected Future Lifetimes ................................................................................................ 10

2.4. Multiple Decrement Models ............................................................................................. 11

2.5. Exercises ................................................................................................................................... 13

3. Expected Present Values of Life‐Contingent Benefits ................................................... 15

3.1. Review of Selected Interest Theory Concepts .......................................................... 15

3.2. Present Values of Life‐Contingent Payments ............................................................ 17

3.3. Expected Present Values of Life Annuities ................................................................. 19

3.4. Expected Present Values of Life Insurance Benefits .............................................. 23

3.5. More General Life‐Contingent Benefits and EPV Relationships ....................... 24

3.6. Exercises ................................................................................................................................... 26

4. Premiums and Reserves ............................................................................................................. 28

4.1. Benefit Premiums and the Equivalence Principle................................................... 28

4.2. Expenses and Gross Premiums ....................................................................................... 30

4.3. Reserves .................................................................................................................................... 34

4.4. Exercises ................................................................................................................................... 40

5. Adjustments for Non‐Annual Cases....................................................................................... 43

5.1. The Uniform Distribution of Deaths Assumption ................................................... 43

5.2. Approximating the Continuous Case ............................................................................ 47

5.3. Exercises ................................................................................................................................... 49

References .................................................................................................................................................. 51
1. Introduction
In 2011, the Society of Actuaries adopted a new pathway to its Chartered Enterprise
Risk Analyst (CERA) credential. One of the changes was the addition of the
Fundamentals of Actuarial Practice (FAP) course to the set of requirements.
However, the Models for Life Contingencies (MLC) exam continues to not be
required for candidates seeking the CERA credential. Nonetheless, the content of the
FAP requires an understanding of basic life contingencies. This note has been
written for the benefit of FAP candidates who have not studied the material covered
on the MLC exam syllabus. This note does not cover the full range of life
contingencies topics, but is sufficient to understand the readings and complete the
exercises in the FAP course. While this note is not required reading for the FAP
course, it will prove useful for candidates who have not previously studied the
subject or who may need a refresher.

Upon completion of this study note, you will:

 Understand life tables as a type of discrete survival model;

 Be able to use life tables to perform survival calculations;

 Be familiar with the concept of a future lifetime and be able to use a life
table to calculate a discrete expected future lifetime;

 Understand the meaning of the present value of a life‐contingent benefit;

 Be able to calculate the expected present value of various life‐contingent

benefits, including basic life insurance and annuity benefits;

 Know how these expected present values relate to each other;

 Be able to calculate benefit premiums for various life‐contingent benefits,

including basic life insurance and annuity benefits;

 Understand how expenses are expressed and how they can be

incorporated into the process of premium determination;

 Understand benefit premium reserves and gross premium reserves, and

how they can be calculated using prospective and recursion relationships;

 Be able to modify the above concepts to apply to non‐annual life‐

contingent cash flows.

Because this study note is intentionally limited in scope, there are many topics
(indeed, too many to enumerate here) that appear in most of the standard actuarial
mathematics and life contingencies textbooks, but which are not covered here.
Currently available books that provide full coverage of the subject include

1|Page
[BGHJN97], [CHL11], and [DHW09]. When reference is made to consulting “standard
life contingencies texts” any of these (and some others) will suffice.

The development presented in this study note largely follows a deterministic

approach, with a focus on expected values. While this suffices for the goals of this
study note, more detailed analyses of life contingencies will necessarily require
examination of the full distributions of the underlying random variables considered
here. Readers interested in an in‐depth, stochastic‐based treatment of actuarial
mathematics are referred to standard life contingencies texts.

The field of actuarial mathematics has developed a thorough system of notation

(often referred to as “International Actuarial Notation”) to denote many of the
quantities commonly used in the analyses of life‐contingent benefits. This study note
does not attempt to inventory the full range of actuarial symbols or to attempt to
explain the logic underlying this system of notation. Actuarial notation will be
introduced only as necessary. For treatments fully exploiting this notation, see
[BGHJN97] and [CHL11].

In order to fully understand the concepts discussed in this study note, you are
assumed to have some prerequisite knowledge in two main areas:

 Actuarial interest theory: It is assumed you are familiar with the concept
of the time value of money. You should also understand the mechanics of
compound interest. A brief review of selected interest theory concepts
(and associated actuarial notation) is presented at the beginning of
Chapter 3; for an in‐depth treatment of actuarial interest theory, the
reader is referred to [DV07], [Bro08], or [Kel08].

 Basic probability: You are expected to understand some elementary

principles of probability theory and be able to apply these principles in
order to calculate probabilities (conditional and unconditional) and
expectations. Any standard probability textbook, such as [HT09],
provides the requisite background.

To demonstrate the ideas discussed in this study note, a sample life table, the 1989‐
1991 Life Table for the Female Population of the United States [NCHS], will be used.
This is done to have a concrete example of a life table. While the mortality pattern
given by this table is broadly representative of that of many human populations, it is
merely an illustrative table. Unless otherwise specified, it is assumed throughout the
study note that mortality is governed by this table. An Excel file with this table and
solutions to all the exercises are available at the same location where you obtained
this document.

In a similar vein, where an interest assumption is required, a flat annual interest

rate of 6% will be used. This assumption is for convenience and may not necessarily
be appropriate for any given application. The methods and formulas presented here,
though, can be applied in different and more general interest rate settings.
2|Page
The author would like to acknowledge the helpful comments and suggestions made
during the writing of this study note by Nick Yuan and Stuart Klugman.

3|Page
2. Life Tables and Expected Future Lifetimes
In this chapter, calculations and concepts related to the measurement of human
mortality are presented. The tool used to accomplish this is the life table. After a
brief introduction in Section 2.1, Section 2.2 introduces life tables and discusses
some of the basic calculations that they are used to perform, Section 2.3 explores the
concept of an expected future lifetime and shows how a life table can be used to
calculate this quantity, while Section 2.4 extends these notions to the case where
individuals are subject to multiple types of hazards or decrements.

2.1. Introduction
One of the fundamental principles of actuarial science is the recognition that the
valuation of life‐contingent benefits requires consideration of both survivorship as
well as the time value of money. This notion, which combines ideas from probability
with those from finance, has great importance in the fields of insurance, annuities,
pensions, risk management, and many other areas.

The first step is to explore the topic of mortality models, especially as expressed by
life tables. Then, in the next chapter, various concepts related to the measurement of
interest are introduced. Finally, these ideas are combined, allowing for the valuation
of various types of life‐contingent benefits.

2.2. Life Tables

Central to the study of the subject of life contingencies is the specification of a
suitable model for the lifetimes of the individuals1 in the population of interest. All
of the quantities subsequently described will be calculated relative to this specified
model; thus it forms the basis for all calculations involving life contingencies or life
contingent payments, such as life insurance and annuity contracts.

One of the commonly used methods of describing a human lifetime model is through
the use of a life table. This type of specification has long been used in the study of
life contingencies,2 and is the framework used in this study note.

A life table describes the pattern of mortality3 by positing an initial number of living
people of the same age. The expected number of people still living in this cohort of

1 While these individuals are usually people, the concepts, notation and formulas
presented here are not specific to human populations and can be used more
generally to describe the lifetimes of entities (or even non‐life forms such as
batteries or light bulbs) in other types of populations as well.
2 For more information regarding the history of actuarial science, see Chapter 2 of

[Klu12], and references therein.

3 Again note that the use of the terms “life table” and “mortality” is for convenience.

The entity being model need not be living and the cause of failure need not be death.
4|Page
individuals is given as a function of their attained age. The individuals are taken to
be homogeneous (that is, they each have the same probability distribution with
regard to the length of their future lifetime), and no additional information is known
beyond the number expected to be alive at each subsequent integral age.

More formally, the members of the cohort are all taken to initially be exact (integral)
age x0 . In many life tables, x0 is 0, so that the cohort consists of newborns. The
number of individuals in this cohort, which is called the radix of the life table, is
denoted  x0 . Then the number of people out of this initial group expected to be
living at attained age x is given by  x . The life table provides a tabular summary of
 x for each integral x , up to some advanced attained age. This upper bound is
typically chosen so that the proportion of the initial cohort expected to still be alive
at this attained age is very small. In fact,  x will often be set to 0 for some sufficiently
large attained age x ; the model then implies that individuals cannot live beyond this
age. In general, when a model for human lifetimes has such a limiting age, this age is
denoted by  .

Thus, the life table provides a method of describing the mortality patterns of
individuals between the attained ages of x0 and  . Because the life table gives
information corresponding only to integral ages, it is a discrete model. However, the
information in the life table can be augmented to produce a continuous model; this
is discussed further in Chapter 5.

Another quantity that sometimes appears in life tables is d x , which is the number of
individuals (out of the original  x0 ) that are living at age x, but are expected to die
prior to reaching age x+1. Then d x is calculated as

d x   x   x 1 ; (2.1)

this quantity is only included for convenience and does not provide any additional
information.

Table 2.1 is an example of a basic life table and is the table used for calculations
throughout this study note, unless otherwise specified. This table, which is the U.S.
1989‐1991 Female Lives Table [NCHS], has a radix of  x0   0  100, 000 and a
limiting age of   111 . This table is available in an Excel file that is available at the
same location from which this document was downloaded.

5|Page
 Table 2.1 – U.S. 1989‐1991 Female Lives Mortality Table

x lx x lx x lx
0 100,000 40 97,033 80 56,986
1 99,172 41 96,895 81 54,077
2 99,105 42 96,747 82 51,016
3 99,063 43 96,586 83 47,820
4 99,031 44 96,412 84 44,512
5 99,006 45 96,222 85 41,115
6 98,983 46 96,012 86 37,643
7 98,962 47 95,779 87 34,113
8 98,943 48 95,522 88 30,573
9 98,927 49 95,240 89 27,075
10 98,911 50 94,932 90 23,666
11 98,896 51 94,594 91 20,372
12 98,881 52 94,222 92 17,231
13 98,863 53 93,813 93 14,310
14 98,842 54 93,367 94 11,672
15 98,814 55 92,881 95 9,346
16 98,780 56 92,355 96 7,339
17 98,740 57 91,784 97 5,641
18 98,695 58 91,163 98 4,243
19 98,647 59 90,483 99 3,124
20 98,597 60 89,742 100 2,251
21 98,546 61 88,939 101 1,584
22 98,492 62 88,075 102 1,086
23 98,438 63 87,146 103 725
24 98,382 64 86,147 104 469
25 98,325 65 85,075 105 293
26 98,267 66 83,929 106 177
27 98,208 67 82,707 107 103
28 98,147 68 81,405 108 57
29 98,082 69 80,014 109 30
30 98,013 70 78,522 110 15
31 97,939 71 76,919 111 0
32 97,860 72 75,197
33 97,777 73 73,350
34 97,689 74 71,378
35 97,596 75 69,287
36 97,497 76 67,082
37 97,393 77 64,764
38 97,281 78 62,321
39 97,161 79 59,733

6|Page
A graphical display of Table 2.1is given in Figure 2.1. The survival pattern shown in
this figure is relatively typical of human populations; see [DHW09, Chapter 3] for
further discussion of these patterns and features.

Equipped with a life table, many of the probabilities and other quantities required
for the study of life contingencies can be calculated. The probability that a person
currently age x will still be alive at age x  n , that is, the probability that the person
will live at least n more years, is
 xn
n px  . (2.2)
x

The complementary probability, i.e., the probability that a person currently age x
will die prior to reaching age x  n , is

 xn  x   xn
n qx  1  n px  1   . (2.3)
x x

A useful quantity is the probability that an individual alive at age x will attain age
x+m, but will die prior to age x+m+n; that is, the probability that this person will live
at least m more years, but fewer than m+n more years. The symbol m |n qx denotes this
probability, sometimes referred to as a deferred mortality probability and its
value is:
 x m   x m n
| qx 
m n . (2.4)
x

Example 2.1: Calculate the following probabilities and write a sentence interpreting
each.

a) 20 p45
b) 10 q65

c) 20 |10 q45

7|Page
 100000
80000
60000
lx

40000
20000
0

0 20 40 60 80 100

Age (x)

Figure 2.1 ‐ Graphical display of Table 1.

Solution:
 65 85, 075
a) 20 p45    0.88415 . This is the probability that a person
 45 96, 222
currently age 45 lives at least 20 more years, i.e., attains age 65.

b) 10 q65  1  75  0.18558 . This is the probability that a person currently
 65
age 65 dies within the following 10 years, i.e., prior to attaining
age 75.

8|Page
  65   75
c) | q45 
20 10  0.16408 . This is the probability that a person
 45
currently age 45 dies between the ages of 65 and 75.

Through the following examples, some simple relationships between the

probabilities defined above are obtained.

Example 2.2: Show that m |n qx  m px  n qx  m .

 x  m   x m n  x m   x  mn  x m  xm   xmn  xm

Solution: m | n qx       m px  n qx  m .
x x  xm  xm x

Thus, the deferred mortality probability m | n qx can interpreted as the probability that
a person currently age x lives m years, and then the resulting person of age x  m dies
in the following n years.

Example 2.3: Show that m n px  m px  n px  m .

 x m n  x mn  x m  x  m  x  mn
Solution: m n px       m px  n px  m .
x  x  xm  x  xm

The intuition behind this relation is similar to that of the previous example:
calculate the probability a person currently age x attains age x+m+n by first
considering the probability that this person attains age x+m and then finding the
probability that the resulting (x+m)‐year‐old person attains age x+m+n by living an
additional n years. For a simple demonstration, letting x  25 , m  30 , and n  20 ,
calculate 30 p25  0.94463 and 20 p55  0.74598 so that
30 p25  20 p55  (0.94463)  (0.74598) = 0.70467 . Calculating 50 p25  0.70467 directly

confirms that the relation holds here.

Example 2.4: Show that m | n qx  m px  m  n px  m n qx  m qx .

The solution is left as an exercise.

Commonly, when calculating the probabilities described above, periods of one year
will be of greatest interest, i.e., n  1 . Standard actuarial shorthand notation omits
the subscript n from Equations (2.2), (2.3), and (2.4) in the cases when the value of
n is 1, allowing these probabilities to be written px , qx , and m| qx respectively. Thus:

 x 1  x   x 1  x  m   x  m 1
px  , qx  , and m | qx  .
x x x

The type of life table described to this point – one in which mortality is solely a
function of attained age – is sometimes known as an ultimate table (or aggregate

9|Page
table). Select tables, on the other hand, incorporate a durational component, so that
mortality is a function of both attained age as well as time elapsed since a selection
event.

Often, this selection event corresponds to the process of being underwritten for an
insurance policy; the knowledge that a person has recently been issued a policy
provides more information regarding their health status. In some situations, though,
the effect of selection may wear off over time. For these cases, it may be reasonable
to assume that mortality is related to both the age at selection and the time since
selection. This often holds during some fixed period after selection (often referred
to as the select period), after which mortality probabilities revert to being a
function solely of attained age. Life tables possessing this property are referred to as
select and ultimate tables.

While slightly complicating the form of the life table (and associated notation),
select and ultimate tables do not fundamentally change any of the concepts
discussed above. All of the formulas and calculations shown here for ultimate tables
can be translated in a straightforward manner to select and ultimate tables. Thus,
for simplicity, only ultimate tables are used here; further information regarding
select and ultimate tables can be found in standard life contingencies texts.

2.3. Expected Future Lifetimes

A key quantity of interest is the expected future lifetime of an individual, that is, the
number of additional years in the future that a person is expected to live. Before
showing how to use life table to perform this type of calculation, some context
regarding the notion of an “expected” future lifetime is needed.

Consider, for a person currently age x (where, again, x is a non‐negative integer), the
number of complete years (or whole years) that this person will live in the future.
For example, if someone currently age 30 were to live an additional 41.5 years, then
the complete years lived in the future is 41. If, on the other hand, this person were to
die after 0.8 more years, that person has lived an additional 0 complete years.

The quantity described above is necessarily a non‐negative integer whose value is

unknown at present, which is to say that it is a discrete random variable. In actuarial
language, this random variable is referred to as the curtate future lifetime4 of the
individual and denoted by K x .

From the life table, the full probability mass function (pmf) for K x can be
constructed so that its distribution can be studied. This random variable takes on
the non‐negative integer values 0, 1, …, each with probabilities that can be
calculated using the life table:

4The term “curtate duration at failure” is also sometimes used; further, the notation
K ( x) may also be used to refer to this random variable.
10 | P a g e
 Pr( K x  k )  k | qx for k  0,1, 2,  (2.5)

The main quantity of interest here is the expected value (or mean) of this random
variable. It is called the curtate expectation of life and denoted by ex .

The life table lends itself to the calculation of this quantity as the expected value
simplifies to:
 
ex  E ( K x )   k k | q x   k p x . (2.6)
k 0 k 1

These are standard formulas for the expected value of a discrete random variable,
where the bounds of the summation reflect the domain of K x , that is, the non‐
negative integers. Note that when there is an assumed limiting age in the survival
model, as is typically the case when using a life table, this summation will contain a
finite number of terms, as the probability of living beyond  is assumed to be zero.

Example 2.5: Calculate e25 .

Solution: From Equation (2.6),


e25   k p25
k 1

 1 p25  2 p25  3 p25    85 p25  86 p25  87 p25  

           
  26    27    28      110    111    112   
  25    25    25    25    25    25 
 0.99941  0.99881  0.99819    0.00015  0  0  
 54.5268.
It is important to remember that while the development and discussion in this
section has been chiefly focused on the expected future lifetime, this quantity is
derived as the expected value of a random variable. For many purposes, the future
lifetime random variable cannot simply be reduced to its mean; that is, the entirety
of the distribution must be considered. For a more thorough treatment of the
curtate future lifetime random variable, the interested reader is referred to any
standard textbook on life contingencies.

2.4. Multiple Decrement Models

The survival models discussed to this point have been restricted to the case in which
there is only a single cause by which an individual may leave the population, namely
death. In some cases, however, there may be multiple reasons why a person could
leave, and these various reasons for leaving may result in different financial
outcomes.

11 | P a g e
Consider a company offering a pension plan to its employees. These employees may
leave the service of the company either due to death, termination, or retirement;
these three causes are likely to result in different financial obligations for the
employer. In analyzing the pension plan, these causes would therefore need to be
considered separately.

As another example, it is common for life insurance policies to offer “surrender

benefits” or “cash values”; that is, to make a payment to the policyholder upon the
lapse or surrender of their policy. This payment is different – typically lower – than
the amount paid upon death of the insured. Thus, an analysis of such a policy needs
to consider both reasons for policy termination.

There are many situations in which the population is subject to multiple

decrements; these cases require a slightly more general approach to modeling the
statuses of the individuals in the population of interest.

Assume that a population is subject to m decrements, labeled 1, …, m . The

probability that an individual age x leaves the population within the next year as a
result of decrement j is denoted qx( j ) . The probability that the individual leaves the
population for any reason is denoted in the multiple decrement context by qx( ) ; it is
simply the sum of the individual decrement probabilities:
m
qx( )   qx( j ) . (2.7)
j 1

The probability that a person age x attains age x + 1 prior to succumbing to any of
the decrements (i.e., they survive all decrements) is given by px( ) and can be
calculated as

px( )  1  qx( ) . (2.8)

Table 2.2 gives an example of a fragment of a multiple decrement table for a

situation with two decrements. By using Equations (2.7) and (2.8), the two
rightmost columns can be calculated from the values of qx(1) and qx(2) .

Table 2.2 ‐ Excerpt from a sample multiple decrement table

x qx(1) qx(2) qx( ) px( )

75 0.04 0.12 0.16 0.84
76 0.05 0.12 0.17 0.83
77 0.06 0.12 0.18 0.82
78 0.07 0.08 0.15 0.85
79 0.08 0.08 0.16 0.84

12 | P a g e
The various probabilities of interest can then be calculated.

Example 2.6: Using Table 2, calculate the probability that a person currently age 76
will succumb to decrement 1 within the next two years.

Solution: The event in question can happen in two different ways, namely by (a)
leaving due to decrement 1 in the first year, or (b) surviving both decrements in the
first year and then succumbing to decrement 1 in the second year. The probability of
( )
these happening are q76(1)
 0.05 and p76  q77
(1)
 (0.83)  (0.06)  0.0498 , respectively.
Then the desired probability is 0.05 + 0.0498 = 0.0998.

Note that many of the other life table concepts and symbols explored earlier in this
chapter have analogs in the multiple decrement context. Readers interested in a full
treatment of multiple decrement life tables are referred to standard life
contingencies texts.

2.5. Exercises
(Note: Exercises marked with an asterisk are best completed using a spreadsheet;
Table 2.1 is available for download as an Excel spreadsheet at the same page where
this document was obtained. Assume an interest rate of i = 6% and that mortality is
governed by Table 2.1 unless otherwise specified.)

Exercise 2.1:

a) Show algebraically that m | n qx  m px  m  n px  m  n qx  m qx and give an

interpretation of this result.
b) Verify the above relationship numerically for the case where x  45 , m  20 ,
and n  10 .

Exercise 2.2*:

a) Augment Table 2.1 to include columns containing the quantities

d x , px , and qx for 0  x  110 .
b) As a special case of the expression demonstrated in Example 2.3, any n ‐year
survival probability can be written as the product of one‐year survival
probabilities:
n 1

n p x  p x  p x 1  p x  2    p x  n 1   p x  k .
k 0

Calculate 25 p50 using the expression above and the column of px values
created in a) and verify that this expression yields the same answer as that
obtained by using Equation (2.2).

Exercise 2.3: If the probability that a 30‐year old is still alive in n years is 0.68441,
determine n .

13 | P a g e
Exercise 2.4: For a person currently age 65, determine the most likely age at which
this person will die.

Exercise 2.5*: Calculate the values of e35 , e45 , and e55 and compare their values.

Exercise 2.6: Using Table 2.2, calculate the probability that a person currently age
75 will succumb to decrement 2 within the next three years.

Exercise 2.7: Assume that the survival distribution for a person age 75 is described
by Table 2.2. Given that this person is known to have left the population prior to age
78 (due to either decrement 1 or 2), calculate the probability that decrement 1 was
the cause of failure.

14 | P a g e
3. Expected Present Values of Life‐Contingent Benefits
In this chapter, the goal is to find the expected present value of payments that are
based on the life status of an individual. This can include life annuity and life
insurance benefits, among others. Before moving to life‐contingent benefits, some
essential elements relating to actuarial interest theory are reviewed in Section 3.1.
Then, in Section 3.2, present values (and expected present values) of life‐contingent
payments are considered. Sections 3.3 and 3.4 discuss expected present values of
life annuities and life insurances, respectively. Finally, Section 3.5 broadens the
discussion to consider more general types of life‐contingent benefits, as well as
relationships between some of the various quantities defined in the chapter.

3.1. Review of Selected Interest Theory Concepts

Prior to valuing life‐contingent benefits, a review of concepts (and corresponding
actuarial notation) associated with the measurement of interest is needed. These
concepts are in turn motivated by the notion of the “time value of money,” the idea
that a monetary sum has greater utility in the present than that of the same sum
received in the future.

Suppose you have some amount of money at present (time 0). Assume this amount
to be one unit for the sake of simplicity. The equivalent sum (in the sense that it has
the same utility, taking into account the time value of money) one year in the future
is 1  i , where i is the annual effective rate of interest. This amount, 1  i , is referred
to as the accumulated value. If the accumulation is continued further into the
future, say n years, the accumulated value at time n will be (1  i ) n ; see Figure 3.1
for a visual representation of this compound interest accumulation process.

If on the other hand, the goal is to find the value now of one unit payable at time 1,
this value at time 0 is referred to as a present value. It is found by reversing the
accumulation process. The present value is necessarily equal to 1/ (1  i ) , as this is
the quantity required at time 0 in order to accumulate to an amount of 1 at time 1.
In actuarial parlance, 1/ (1  i ) is referred to as a discount factor, and for the sake of
convenience, the shorthand notation v is used to represent this quantity; that is,

1
v . (3.1)
1 i

15 | P a g e
Figure 3.1 ‐ Accumulation of one unit under compound interest. At annual
effective interest rate i , the accumulated value is 1  i after one year and
(1  i ) n after n years.

Analogous to the accumulation process, the process of finding a present value can be
applied for multiple periods. The present value of one unit payable at time
n is v n or equivalently, (1  i )  n ; see Figure 3.2.

Figure 3.2 ‐ The present value (at time zero) of one unit payable at time n is v n .

The processes of finding find present values and accumulated values can be
extended to annuities, i.e., sequences of payments. When all of the payments in an
annuity are certain to be made, as will be the case with the annuities considered in
this section, the stream of payments is called an annuity‐certain. In particular, this
section considers annuities‐certain in which each payment amount is constant and
the payments are regularly spaced. For the sake of simplicity, assume n payments of
one unit each occurring on an annual basis.

For the case in which the first payment occurs at the end of the first year (i.e., at
time 1, so that the final payment occurs at time n ), the resulting annuity is referred

16 | P a g e
to as an annuity‐immediate; its present value is denoted by an and can be
calculated5 as
n
1  vn
an  v  v    v  
2 n
v k
. (3.2)
k 1 i

The corresponding accumulated value (value at time n ) is

n 1
(1  i ) n  1
sn  1  (1  i )  (1  i ) 2    (1  i ) n 1   (1  i ) k  . (3.3)
k 0 i

When the first payment happens at the start of the first year (so that the n payments
occur at times 0, 1, …, n  1 ), the result is called an annuity‐due and its present
value and accumulated value (at times 0 and n , respectively) are given by
n 1
1  vn
an  1  v  v 2    v n 1   v k  (3.4)
k 0 d

and
n
(1  i ) n  1
sn  (1  i )  (1  i ) 2    (1  i ) n   (1  i ) k 
 (3.5)
k 1 d

respectively, where d  1  v .

For a full treatment of actuarial interest theory, the interested reader is referred to
[Kel08], [Bro08], or [DV07].

3.2. Present Values of Life‐Contingent Payments

Now consider the present values of payments, which, unlike those considered in the
previous section, are not guaranteed to be made. Of particular interest are payments
whose amounts and/or timing are contingent on the life status of a particular
individual, that is, whether the individual is alive or dead. (Though, in some
circumstances, other statuses, such as disability or retirement are considered.)

In these cases, the concept of a present value necessarily becomes more

complicated. Because the life status of the individual is a random quantity, then the
present value of any payment(s) based on this life status must then also be random.

5The formulas given in this section for the present values and accumulated values of
annuities‐certain all make use of the following formula for the sum of a geometric
n 1
1  xn
series: for x  1, 1  x  x    x   x 
2 n 1 k
.
k 0 1 x
17 | P a g e
Thus, the present value of such payments is no longer simply a single value, but is
rather a probability distribution of values.

While acknowledging that, in general, the entire distribution of present values may
be of interest, the focus here is on the expected value of the distribution. This
expected value is given various names in the actuarial literature,6 but here the term
expected present value (EPV) is used.

The EPV of any single life‐contingent payment is

EPV  Amt  Prob  Disc , (3.6)

where Amt is the amount of the payment to be made, Prob is the probability that the
payment will be made, and Disc is the appropriate discount factor, which here is v n ,
where n is the time of the payment.

Example 3.1: Suppose that a person currently age 35 will receive a payment of
$100,000 if alive at age 55. Calculate the EPV of this payment. (Recall that for this
study note, Table 2.1 is the survival model and the interest rate is a flat i = 6% unless
otherwise specified.)

Solution: Using Equation (3.6):

EPV  Amt  Prob  Disc

 100000  20 p35  v 20
20
 1 
 100000  (0.95168)   
 1.06 
 29,674.

A payment that is only made to a person if they are still alive at some specified time
in the future, as was the case in the previous example, is known as a pure
endowment.

Equation (3.6) can be generalized to the case where there are multiple payments. To
find the EPV of a stream of contingent payments, use the formula

EPV   Amtk  Probk  Disck , (3.7)

k

where Amtk , Probk and Disck represent the amount, probability of payment, and
discount factor, respectively, corresponding to the kth payment. The summation is
taken over all possible payment times. Equation (3.7) is very versatile – indeed it
can be used to find the EPV of any arbitrary sequence of life‐contingent payments –

6The terms Actuarial Value, Net Single Premium, and Actuarial Present Value
are commonly used to describe this type of expected value.
18 | P a g e
and will be used repeatedly in this study note. Note that the equation follows
directly from the definition of expected value for discrete random variables.

Example 3.2: Now suppose that a 35‐year‐old will receive three payments, of
amounts $100,000, $200,000, and $400,000 upon reaching the ages of 55, 65, and
75, respectively. As in the previous example, each payment will only be made if the
person is still living at the time of the scheduled payment. Find the EPV of this
sequence of payments.

Solution: Using Equation (3.7):

EPV   Amtk  Probk  Disck

k

 100, 000  20 p35  v 20  200, 000  30 p35  v 30  400, 000  40 p35  v 40

 29,674  30,355 + 27,609
= 87,637.

This method can be extended to any number of payments of arbitrary amounts.

Under this framework, potential payments can be conditioned on various types of
life statuses.7 Examples of this are presented in the next two sections.

3.3. Expected Present Values of Life Annuities

Now consider the expected present values of life annuities, that is, annuities for
which some or all of the payments are life‐contingent. The contingent nature of
these annuity payments distinguish life annuities from the annuities‐certain that
were previously reviewed. The uncertainty surrounding the payments also means
that the present value of a life annuity is a random quantity rather than a single
number.

The first type of life annuity considered is the whole life annuity. This annuity
consists of a sequence of regularly spaced payments, continuing as long as the
annuitant is alive. Payments are assumed made on an annual basis and each
payment is one unit.

In a whole life annuity‐due, the first payment happens immediately, i.e., at time 0.
The EPV for this type of annuity, issued to a person currently age x, is denoted in
actuarial notation by ax . This EPV can be calculated from Equation (3.7), noting that
the possible payment times are 0, 1, 2, … , and each potential payment amount is 1:

7In this study note, only statuses related to a single life are considered. However,
statuses involving multiple lives are possible and are indeed common; see, for
example, [DHW09, Chapter 8] for discussion of benefits based on these types of
statuses.
19 | P a g e
  
ax   Amtk  Probk  Disck   k px  v k . (3.8)
k 0 k 0

Note that 0 px is trivially equal to 1 and that when the underlying survival model has
a limiting age, all of the terms in the summation corresponding to payments beyond
this age will be zero.

For a whole life annuity‐immediate, the first payment happens one year after
issue, that is, at time 1. The EPV for this type of annuity, issued to a person currently
age x, is denoted by ax . This EPV can be calculated in a similar manner to that of the
whole life annuity‐due, the only change being that the potential payments are made
at times 1, 2, … , so that
 
ax   Amtk  Probk  Disck   k px  v k . (3.9)
k 1 k 1

These EPV formulas (as well as the ones that will follow) readily lend themselves to
calculation by spreadsheet.

Example 3.3: Calculate a55 and a55 .

Solution: In order to calculate these EPVs, values of k p55 for k = 0, 1, 2, … are

needed. The expression given in Exercise 2.2, Part b) can be used to calculate these
values and then the EPVs are:

a55   k p55  v k
k 0

 0 p55  v 0  1 p55  v1    55 p55  v55  56 p55  v 56  

1 0.99434 0.00016 0
 0
 1
  55
 
(1.06) (1.06) (1.06) (1.06)56
 13.29133

a55   k p55  v k
k 1

 1 p55  v1    55 p55  v 55  56 p55  v 56  

0.99434 0.00016 0
 1
  55
 
(1.06) (1.06) (1.06)56
 12.29133

Note that a55  a55  1 , which is sensible, as the two annuities considered here differ
only by the payment made at time zero, whose EPV is 1. This relationship between

20 | P a g e
EPVs (and indeed between the underlying present value random variables) holds in
general; the proof of this relation is not difficult and is left as an exercise.

It is important to note that the EPVs described in this section represent expected
values of particular distributions. For instance, when calculating ax , the underlying
distribution under consideration consists of present values of annuities‐certain.
This is not the same as finding the mean lifetime of a person age x and then
calculating the present value of the corresponding annuity of this duration; this
latter present value is a e and will not in general be equal to ax . Also note that ex is
x

not necessarily an integer value. In this case, while the interpretation of the quantity
a e is perhaps a bit less intuitive, its numerical value is still given by Equation (3.4).
x

The following example numerically illustrates this point.

Example 3.4: Calculate a e .

55

Solution: First, from Equation (2.6), e55  26.55205 . Then

1  v 26.55205
a e  a 26.55205  0.06
 13.90628 . This is clearly not the same value as
55
1.06

a55  13.29133 .

The next type of annuity considered is a term life annuity.8 This type of annuity
differs from a whole life annuity in that the payments are only made for some
specified length of time. Each payment is again contingent on the annuitant being
alive at the time of the payment.

The term life annuity has both annuity‐due and annuity‐immediate versions; the
EPVs of these annuities (issued to a person age x , with a term of n years) are
denoted by ax:n and ax:n , respectively. These EPVs can be computed similarly to
those of the corresponding whole life annuities:
n 1 n 1
ax:n   Amtk  Probk  Disck   k px  v k (3.10)
k 0 k 0

n n
ax:n   Amtk  Probk  Disck   k px  v k (3.11)
k 1 k 1

As is reflected in the bounds of the summations in Equations (3.10) and (3.11), the
potential payment times for the n ‐year term annuity‐due are 0, 1,  , n  1 ,
whereas the possible payments in the n ‐year annuity‐immediate version are
1, 2,  , n .

8This type of annuity is also commonly referred to as a temporary life annuity.

21 | P a g e
Example 3.5: Consider a 20‐year term life annuity‐due, paying an amount of one
per year, issued to a 65‐year old.

a) Calculate the EPV of this annuity.

b) Now assuming this annuity pays 500 per year, calculate the EPV of the
annuity.

Solution:
19
a) EPV  a65:20   k p65  v k  10.32146
k 0

19 19
EPV   500  k p65  v k  500   p65  v k  500 a65:20  (500)  (10.32146)
b) k 0 k 0
k

= 5,160.73

The annuity EPV notation introduced in Equations (3.8) ‐ (3.11) only represent the
specific cases in which the annuity payments are all one unit. However, this example
shows that, in the event of a level annuity of a different amount, these EPVs scale
linearly with payment amount. That is, in order to calculate an EPV for any annuity
with level payments of amount C , first calculate the EPV of a corresponding annuity
with payment amounts of 1, and then simply multiply this EPV by C .

While the EPVs computed previously assume level payment amounts, Equation (3.7)
can be used to find the EPVs of annuities with more general payment patterns.

Example 3.6: Consider a whole life annuity‐due issued to a person age 55. The
amount of the first payment is 1,000, the second payment is 1,050, the third
payment is 1,100, and each subsequent annual payment increases in amount by 50.
Calculate the EPV of this annuity.

Solution: Again using Equation (3.7),

EPV   Amtk  Probk  Disck

k

 (1, 000)  0 p55  v 0  (1, 050)  1 p55  v1  (1,100)  2 p55  v 2  

 19,939.12.

While all of the life‐contingent benefits discussed thus far have been premised on a
person being alive at the time of a payment, as mentioned before, the general
framework allows EPVs to be calculated for instruments whose payments are based
on other life statuses as well, as explored in the next section.

22 | P a g e
 3.4. Expected Present Values of Life Insurance Benefits
Next, consider life insurance benefits, and in particular, how to find the EPV of such
benefits. A basic life insurance contract9 makes a payment to a named beneficiary
upon the death of the insured person, subject to the provisions and stipulations of
the policy. For now, assume that this benefit payment is made at the end of year of
death of the insured.

The simplest coverage is whole life insurance; this type of contract lasts for the
length of the life of the insured. It provides a level death benefit amount of one unit,
payable at the end of the year of death. Then the possible payment times are 1, 2, … ,
corresponding to death in the first year ( K x  0 ), second year ( K x  1 ), etc. From
Equation (2.5), the probability of payment at time 1 is qx , the probability of
payment at time 2 is 1| qx , and in general, the probability of payment at time k is
k 1 | q x . Then the EPV for this life insurance, which is denoted by Ax , is computed as

 
Ax   Amtk  Probk  Disck   k 1 | qx  v k . (3.12)
k 1 k 1

Another category of life insurance is known as term life insurance. This type of
insurance is distinguished by the fact that it only pays a benefit if the insured dies
within a fixed period from the issue of the contract. Again assume that the benefit is
paid at the end of the year of death, but only if death occurs within n years of the
policy’s issue date. Then, for such a term life insurance contract issued to a person
age x with a benefit amount of one unit, the EPV can be calculated in a manner
similar to that of a whole life insurance – the only difference is that the possible
payment times are 1, 2, … , n . This EPV is denoted by Ax1 :n and can be calculated as

n n
Ax1 :n   Amtk  Probk  Disck   k 1 | qx  v k . (3.13)
k 1 k 1

Example 3.7: Calculate A40 and A40:30

1
.

Solution: Using Equations (3.12) and (3.13) yields, respectively,


A40   k 1 | q40  v k
k 1

 q40 v  p40 q41 v 2  2 p40 q42 v 3  

 0.12401

9Only some very basic and generic types of life insurance benefits are considered
here, and no attempt is made to analyze the myriad varieties of life insurance
products; see, e.g., [Klu12] for a discussion of life insurance product designs.
23 | P a g e
and
30
A40:30
1  k 1 | q40  v k
k 1

 q40 v  p40 q41 v 2  2 p40 q42 v 3    29 p40 q69 v 30

 0.06203.

The term life EPV is smaller in magnitude than the whole life EPV. This is not
surprising, because the underlying whole life insurance pays a benefit regardless of
when death occurs, whereas the policy the term life policy only pays if the death of
the insured happens within 30 years of policy issue.

While the EPV notation described in this section only specifically applies to the
types of insurance policies described here (i.e., death benefit amounts of one unit,
payable at the end of the year of death), the same approach used in Equations (3.12)
and (3.13) can be used to find the EPV of life insurance policies with any arbitrary
pattern of benefit payments.

3.5. More General Life‐Contingent Benefits and EPV

Relationships
In this section, more general and complex forms of life‐contingent benefits are
introduced. Some useful relationships between the various EPV quantities detailed
in this chapter are also presented.

An endowment insurance is a type of policy providing both a life insurance benefit

upon death as well as an endowment benefit. The particular type of endowment
insurance considered here is a combination of an n‐year term insurance and an n‐
year pure endowment. That is, if the insured dies within n years of the policy issue,
the policy will pay a benefit (which, for convenience, is again assumed to be one
unit); if the insured is still alive at the end of the n‐year term, an endowment benefit
of one unit is paid at that time.

The EPV for this type of policy can be calculated as before; for such a policy issued to
a person age x , the EPV is denoted by Ax:n and is calculated by using Equation (3.7):

n
 n 
Ax:n  EPV   Amtk  Probk  Disck    | qx  v k   n px v n .
k 1 (3.14)
k 1  k 1 

The summation within the parentheses in Equation (3.14) corresponds to the term
insurance portion of the EPV, while the final term represents the EPV of an n‐year
pure endowment benefit.

Example 3.8: Calculate A70:20 .

Solution: Using Equation (3.14),

24 | P a g e
  20 
A70:20    k 1 | q70  v k   20 p70 v 20  0.36912 + 0.09398 = 0.46310 .
 k 1 

EPVs are additive in the sense that the EPV for an instrument containing multiple
benefits is simply the sum of the EPVs of the individual benefits. This can be
exploited to find the EPV of arbitrary life‐contingent instruments containing several
benefits, as demonstrated in the following example.

Example 3.9: Consider a policy sold to a person age 50 that pays endowment
benefits of 50,000 and 100,000 if the policyholder is still alive at the ages of 65 and
70, respectively. The policy also pays a death benefit of 200,000 at the end of the
year of death, should the policyholder die prior to reaching age 65. Calculate the EPV
of this policy.

Solution: The EPV of this policy is simply the sum of the EPVs for the endowment
and insurance benefits:

EPV  200, 000 A50:15

1
 50, 000 15 p50 v15  100, 000 20 p50 v 20
 12,310.59  18,696.98  25,790.60
= 56,798.16.

There are a few additional useful relationships between these annuity and life
insurance EPVs that are worth mentioning. The EPV of a whole life insurance is
related to that of a whole life annuity‐due (both issued to a person age x ) via the
following equation:
1  Ax
ax  . (3.15)
d
A similar relationship can be made between an n‐year endowment insurance and an
n‐year term life annuity‐due:

1  Ax:n
ax:n  . (3.16)
d
Because these relationships10 relate insurance and annuity EPVs, they are very
valuable in practice.

It is also useful to relate the EPV of a particular life‐contingent benefit for a person at
a particular age to that for a person at a subsequent age, e.g., ages x and x  1 ; this

10Note that relationships analogous to Equations (3.15) and (3.16) also hold
between the underlying random variables, not just their expectations; see, for
example, [BGHJN97] for a derivation and interpretation of these relationships.
25 | P a g e
type of relationship is known as a recursion relation. For example, for a whole life
annuity‐due,

ax  1  v  px  ax 1 . (3.17)

This equation essentially decomposes the EPV of the annuity into the EPV of the first
payment (which is 1) and the EPV of the payments in the subsequent years. This
type of recursion is often helpful when computing EPVs using a spreadsheet.
Equation (3.18) is an analogous relationship for whole life insurance EPVs – it also
splits the total EPV into the EPV corresponding to the first year benefit and that
corresponding to the benefits payable in later years:

Ax  v  qx  v  px  Ax 1 . (3.18)

Example 3.10: Calculate A75 , given the following information:

A77  0.43944, q76  0.03455, q75  0.03182, i  6% .

Solution: Use Equation (3.18) to successively find A76 and then A75 :

A76  v  q76  v  p76  A77

1 1
  (0.03455)   (1  0.03455)  (0.43944)
1.06 1.06
= 0.43284
so that

A75  v  q75  v  p75  A76

1 1
  (0.03182)   (1  0.03182)  (0.43284)
1.06 1.06
= 0.42537.

3.6. Exercises
(Assume an interest rate of i = 6% and that mortality is governed by Table 2.1 unless
otherwise specified.)

Exercise 3.1: Show algebraically that ax  ax  1 for any given value of x .

Exercise 3.2*: Show that ax:n  ax:n  1   v n  n px  . Give an interpretation of this

result and verify the relationship numerically for the case where x  45 and n  20 .

Exercise 3.3*: Consider a whole life annuity‐due issued to a person age 65 that pays
an amount of 1 at the beginning of the first year, 1.03 at the beginning of the second

26 | P a g e
year, (1.03) 2 at the beginning of the third, and whose payment amounts continue to
increase at a rate of 3% each year. Calculate the EPV of this annuity.

Exercise 3.4*: A person currently age 45 purchases a 20‐year term insurance. The
death benefit, which is paid at the end of the year of death, is $100,000 for the first
10 years and $200,000 for the last 10 years. Calculate the EPV of this insurance
policy.

Exercise 3.5*: A person currently age 30 purchases a whole life insurance with
death benefit payable at the end of the year of death. If the EPV of this insurance
policy is 5,300, calculate the amount of the death benefit.

Exercise 3.6*: Equation (3.18) can be used to find the value of Ax for any integer x in
the following manner: The life table value of 111  0 implies that
A110  v  (1.06) 1  0.9434 . Use this value to work backwards in a spreadsheet,
recursively calculating the values of Ax for x  110, 109,  , 0 .

Exercise 3.7*: Calculate the value of A40 under the standard assumptions (i.e., using
Table 2.1 for mortality and an interest rate of i = 6%). Then, change the interest rate
to i = 8% and calculate A40 under this new assumption. Finally, repeat the process
with i = 10% and compare the resulting three EPVs.

27 | P a g e
4. Premiums and Reserves
This chapter applies the concepts discussed in the previous chapter to some
common actuarial tasks. In particular, the determination of premiums and reserves
are discussed here with the tools previously developed for calculating EPVs used
extensively. Section 4.1 introduces the equivalence principle and describes how it
can be used to determine benefit premiums for life‐contingent policies. Section 4.2
generalizes the determination of premiums to consider expenses. Finally, Section
4.3 discusses reserves and develops formulas for determining benefit reserves as
well as gross premium reserves.

4.1. Benefit Premiums and the Equivalence Principle

In exchange for the benefits provided by insurance products, such as the whole life
or term life products discussed in the previous chapter, customers make payments,
known as premiums, to the insurance company. It will be assumed that the amount
and frequency of the premiums are specified in the policy contract, though for some
types of life insurance policies (Universal Life Insurance, for example) the premium
schedule may be more flexible.

Benefit premiums (sometimes known as net premiums) are first considered;

these are premiums meant to cover the benefits paid by the policy. Later, gross
premiums will be considered. These are the actual premiums paid by the
policyholder, and are meant to account for not only the benefits paid, but also other
items, such as, for example, expenses and profit.

The framework utilized in this study note for the determination of premiums is
called the equivalence principle, which stipulates that the EPV of the payments
(benefits and maybe expenses and profits) made by the insurer should be equal to
the EPV of the income (premiums) received by the insurer.

For the case in which the equivalence principle is applied on a net basis, benefits
will be taken to be strictly the actual benefits paid by the insurer, and premiums will
refer to the benefit premiums. Then the equivalence principle implies

EPV (benefit premiums) = EPV (benefits paid) , (4.1)

and this equation can be used to determine the benefit premium amount.

Example 4.1: Consider a whole life insurance policy issued to a person currently
age 40, with a death benefit of $100,000 payable at the end of the year of death.
Assume that a benefit premium of amount P is payable at the beginning of each year
that the contract is still in‐force. Calculate P.

Solution: In this example, the benefits paid consist of the whole life death benefit,
whereas the benefit premiums form a whole life annuity‐due of amount P.
28 | P a g e
EPV (benefit premiums) = EPV (benefits paid)
P  a40  100, 000 A40
100, 000 A40
P .
a40

Then, using Table 2.1, A40  0.12401 and a40  15.47581 so that P  801.32 .

The assumption in this example that premiums are payable at the beginning of each
year that the policy is in‐force is a typical one, and one that will be used for the
balance of this chapter. However, in practice, benefits are paid at (or shortly after)
the moment of death, not at the end of year. An appropriate adjustment will be
introduced later.

Example 4.2: Consider a 20‐year term life insurance policy issued to a person age
30, with a death benefit of $300,000 payable at the end of the year of death. Assume
that a premium of amount P is payable at the beginning of each year that the
contract is still in‐force. Calculate P.

Solution: The structure of both the benefits as well as the premiums here is
somewhat different than in the previous example. Specifically, the death benefit is a
20‐year term life insurance benefit, while the premium payment stream forms a 20‐
year term annuity‐due. Thus:
EPV (benefit premiums) = EPV (benefits paid)
P  a30:20  300, 000 A30:20
1

300, 000 A30:20

1

P .
a30:20

Then, A30:20
1
 0.01558 and a30:20  12.05600 so that P  387.73 .

While the above examples have involved premium payment patterns and benefits
that are both level, this need not be the case; the equivalence principle can be
applied more generally to arbitrary benefit and premium patterns.

Example 4.3: A person currently age 35 purchases a whole life insurance policy in
which the death benefit is $100,000 for the first 20 years of the policy, and then
doubles to $200,000 for the remainder of the policy. Premiums are paid annually
while the policy is still in‐force. The amount of the first premium (at time 0) is P, and
each subsequent premium increases by 3%. Calculate the amount of the first
premium paid, P.

Solution: The EPV of the benefit premiums can be calculated as

29 | P a g e
EPV (benefit premiums) = P  (1.03)v 1 p35 P  (1.03) 2 v 2 2 p35 P  
 25.12993 P,

and the EPV of the benefits paid can be calculated as

EPV (benefits paid) = 100, 000v q35  100, 000v 2 1| q35    100, 000v 20 19 | q35
 200, 000v 21 20| q35    17,044.73.

Then, equating these two expressions yields a value of P  678.26 .

The benefits in question need not be life insurance benefits; the equivalence
principle will hold for any type of life‐contingent benefit. Also, in the event that the
premium pattern consists of a single premium payment made at issue 0, then the
amount of this premium is simply the EPV of the benefits. The following example
illustrates these points.

Example 4.4: A man currently age 40 purchases a product that will make annuity
payments of $50,000 each year, starting at age 65 and continuing as long as he lives.
If he dies prior to reaching age 65, a death benefit of $100,000 will be paid at the
end of the year of his death. If he purchases this product with a single premium of
amount P at issue, calculate P.

Solution:
EPV (benefit premiums) = EPV (benefits paid)
P  100, 000 A40:25
1  50, 000v 25 25 p40 a65
P  118,974.

4.2. Expenses and Gross Premiums

Before discussing the determination of gross premiums, it is necessary to first
consider the various types of expenses that are typically incurred by insurance
companies. There are many possible bases on which expenses can be expressed; in
many cases, the nature of expense will influence the basis used.

Some of the more common ways to express expenses are given below.

 As a percent of premiums paid: Some expenses might naturally be expressed

as a percentage of all actual (gross) premiums paid. These might include, for
example, commissions paid to agents and premium taxes.

 As a flat amount per policy: Various types of administrative, maintenance, or

claims expenses may be considered to consist of fixed amounts, incurred at
different times, such as at policy issue, on a yearly basis, or upon policy
termination.

30 | P a g e
  As an amount per unit of death benefit amount: There may be some
expenses, such as those associated with underwriting a new policy, which
increase with the size of the policy being written. They may be expressed as
an amount per “unit” (say, $1,000) of the policy’s death benefit.

For the purposes of this study note, it will always be assumed that expense amounts
are known and given. Hence, there is no discussion here regarding how these
expenses have been determined, though the interested reader is referred to [Klu12]
for a discussion of life insurance company expenses.

In addition to benefits and expenses, gross premiums must also incorporate any
profits for the insurer. Insurers’ profits may be measured in many different ways;
see [Klu12] for a discussion. However, for the sake of simplicity, when profits are
considered in this study note, they will always be expressed as a percent of gross
premiums.

Then the equivalence principle, applied on a gross basis, requires gross premiums to
satisfy the following equation:

EPV (gross premiums)  EPV (benefits, expenses, and profit) . (4.2)

The use of the gross equivalence principle is demonstrated through the following
examples. In all examples and exercises that follow, any expenses and profits
beyond those specifically mentioned are ignored.

Example 4.5: A person age 50 purchases a whole life insurance policy with a death
benefit of $250,000. Gross premiums of amount G are payable annually. If the
company’s expenses are given below, calculate G.

Underwriting Expenses: $5 per $1,000 of policy death benefit, incurred at

policy issue

Other Issue Expenses: $200 per policy, incurred at policy issue

Maintenance Expenses: $50 per policy, incurred at the start of each year
the policy is in‐force including the first year

Solution: Using Equation (4.2),

EPV (gross premiums)  EPV (benefits and expenses paid)
5
G a50 = 250,000 A 50   (250, 000)  200  50a50
1, 000
5
14.14961G  250, 000 (0.19908)   (250, 000)  200  50 (14.14961)
1, 000
G  3,669.87

31 | P a g e
Note that the maintenance expenses in this example form a whole life annuity, since
they are payable annually for as long as the insured is alive. Also, it is common to
assume, as was done here, that underwriting expenses are incurred at the time of
policy issue, though in reality the expenses may be incurred prior to issue.

Example 4.6: (Extension of Example 4.5) A person age 50 purchases a whole life
insurance policy with a death benefit of B. Gross premiums of amount G are payable
annually. If the company’s expenses are as given in Example 4.5:

a) Write an expression for G as a function of B.

b) Calculate the gross premium per thousand dollars of insurance benefit for
death benefit amounts of $100,000, $250,000, and $400,000.

Solution:

a) The solution is similar to that of the previous example:

EPV (gross premiums)  EPV (benefits and expenses paid)
5
G a50 = B A50   ( B)  200  50a50
1, 000
5
14.14961G  B (0.19908)   ( B)  200  50 (14.14961)
1, 000
G  0.01442 B  64.13

Thus, G is a linear function of B.

b) The following table gives the gross premiums and gross premium amounts
per thousand of death benefit:

Death Benefit Gross Premium Gross Premium / $1000 Death Benefit

$100,000 $1,506.13 15.0613
$250,000 $3,669.13 14.6765
$400,000 $5,832.13 14.5803

It can be seen that, due to the constant term of 64.13 in the above equation,
the cost per thousand of the life insurance decreases as the death benefit
amount increases.

Example 4.7: A person age 30 is issued a 25‐year term life insurance policy with a
death benefit of $200,000. Gross premiums of amount G are payable annually. If the
company’s expenses are given below, calculate G.

Issue Expenses: $200 per policy, incurred at policy issue

Commissions: 40% of the first gross premium, and 6% of

32 | P a g e
 subsequent premiums

Premium Taxes: 2% of all gross premiums paid

Solution: Again, using Equation (4.2),

EPV (gross premiums)  EPV (benefits and expenses paid)
G a30:25 = 200, 000 A30:25
1  200  (0.34) G  (0.06) G a30:25  (0.02) G a30:25
13.39452 G  200, 000 (0.02102)  200  (0.34)G  (0.06)(13.39452) G
+ (0.02)(13.39452) G
G = 367.53.

There are a couple of noteworthy items in this example. First, since the expenses
related to commissions and premium taxes are expressed as a percent of gross
premium, this gross premium G appears on both sides of the equation. Also, the
commission expenses have been expressed in terms of an amount incurred in the
first year only, and an additional amount occurring in every year, including the first
year; this is simply done as a computational convenience.

Example 4.8: (Extension of Example 4.3) Re‐do Example 4.3, but now incorporate
the company’s expenses and profits as given below. The premium to be calculated
will be the gross premium, rather than the benefit premium.

Issue Expenses: $500 per policy, incurred at policy issue

Maintenance Expenses: $50 per policy, incurred at the start of every year
that the policy is in‐force

Termination Expenses: $100 per policy, incurred at the time of the

policy’s termination

Profits: 4% of gross premiums

Solution: As in Example 4.3, the EPV of the (gross) premiums can be calculated as

EPV (gross premiums) = G  (1.03)v 1 p35G  (1.03) 2 v 2 2 p35G  

 25.12993 G.

In Example 4.3, the EPV of these benefits was calculated to be 17,044.73. Thus,

33 | P a g e
EPV (benefits, expenses, and profits) = 17,044.73  500  50 a35  100 A35  0.04G a35
 17, 044.73  500  50(15.95377) + 100 (0.09696)
+ 0.04(15.95377)G
 18,352.11 + 0.63815G

Then equating these two expressions yields G = 749.32.

It is assumed in this example that the termination expenses are paid at the same
time as the payment of the death benefit, i.e., at the end of the year of death. Thus,
the EPV of these expenses can be written in terms of the EPV of a whole life
insurance policy.

4.3. Reserves
The previous sections in this chapter were concerned with the determination of
premiums. The calculations and formulas were based on EPVs at the time of policy
issue. These concepts will be slightly generalized here to consider policies that have
been in‐force for some period of time. Such an in‐force policy could have life‐
contingent benefits payable in the future; there may also be future premiums to be
collected.

For most life insurance policies – particularly those funded by a series of level
premium payments – the premiums paid in the early years of the policy exceed the
expected payments to be made by the insurer. As a result, the insurer will tend to
accumulate funds during these early policy years. This accumulation does not
represent a profit, however, since the insurer will expect to pay more than it
receives in the later policy years. Indeed, the insurer will need to retain some of
these accumulated funds in order to meet its future obligations; a measure of the
amount needed is referred to as the reserve11.

The process of calculating a reserve, which is often referred to as valuation, is the

subject of this section. The method for determining the reserve will depend on the
reason for the valuation, and may be specified by regulatory or accounting rules (see
[Klu12]). In this study note, we consider two types of reserves, namely benefit
reserves and gross premium reserves.

For a particular policy, the benefit reserve is calculated as the expected value of the
difference between the present value of the future benefits to be paid and the
present value of the future premiums to be collected. More formally, the benefit
reserve at time t for a given policy is defined as

t V  EPVt (future benefits to be paid)  EPVt (future premiums to be collected), (4.3)

11Other terms such as policy value are also sometimes used to describe the same
concept.
34 | P a g e
where the notation t V is used generically to denote any reserve value calculated at
time t. The notation EPVt indicates an EPV calculated at time t. This involves
discounting any payments with respect to mortality and interest back to time t. By
convention, this calculation includes any premiums that are due at time t, but does
not include any benefits payable at the end of the previous year. Also, any
calculation of any reserve at time t presupposes that the underlying policy is still in‐
force at this time.

Equation (4.3) is known as a prospective reserve formula since it looks forward

into the future to calculate the reserve value. Some authors also define a
retrospective reserve, which looks back from time t to the time of policy issue in
order to calculate the reserve value. Under particular circumstances, the
retrospective reserve can be shown to be equal to the prospective reserve. Because
the retrospective reserve is less‐commonly used in practice, it is not further
explored here; the interested reader is referred to [BGHJN97] or [CHL11].

The following examples demonstrate the calculation of benefit reserves.

Example 4.9: (Extension of Example 4.2) Consider a 20‐year term life insurance
policy issued to a person age 30, with a death benefit of $300,000 payable at the end
of the year of death. Assuming level premiums payable at the beginning of each year
that the contract is still in‐force, calculate the benefit reserve for this policy at time
5.

Solution: In Example 4.2, the annual benefit premium for this policy was calculated
to be P  387.73 . Then the benefit reserve at time 5 is calculated as

5 V  EPV5 (future benefits to be paid)  EPV5 (future premiums to be collected)

 300, 000 A35:15
1  P a35:15
 300, 000 (0.01615) - (387.73)(10.2108)
= 887.50.

At time 5, there are 15 years remaining in the policy, so that the future benefits
consist of a 15‐year term policy on person currently age 35. Likewise, the future
premiums form a 15‐year term life annuity‐due.

Example 4.10: Extending the previous example, determine the benefit reserve for
the policy described in Example 4.2 at time t, for t = 0, 1, 2, … 19, 20. Plot t V as a
function of t.

Solution: The benefit reserve values are shown in Figure 4.1.

Note that the benefit reserve is 0 at both times 0 and 20. The former is due to the
use of the equivalence principle to set the benefit premium, and the latter is because
at time 20 there are no future benefits or premiums for the policy.
35 | P a g e
 2000
1500
Benefit Reserve

1000
500
0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Year

Figure 4.1 – Benefit Reserve Values for Example 4.10.

Example 4.11: A man currently age 40 purchases a product that will make annuity
payments of $50,000 each year, starting at age 65 and continuing as long as he lives.
If he dies prior to reaching age 65, a death benefit of $100,000 will be paid at the
end of the year of his death. If he purchases this product with a single premium at
issue, calculate the benefit reserve at times 10, 20, and 30.

Solution: Note that at time 10, there are no future premiums to be paid. Then the
benefit reserve is calculated as

10 V  EPV10 (future benefits to be paid)  EPV10 (future premiums to be collected)

 100, 000 A 1  15 p50 v15 (50, 000) a65  0
50:15

 6,155.29 + (0.89616)(0.41726)(50,000)(11.16929)
= 214,987.

Similarly, at times 20 and 30 the benefit reserve is 20V = 399,959 and 30V = 494,781.

36 | P a g e
As was the case for various EPVs, it is possible to derive recursion relations for
benefit reserve values. For a life insurance policy issued to a person age x, a general
relation between the benefit reserve value at duration t, t V , and the value at
duration t+1, t 1V , is given by:

(t V  P)(1  i )  qx t B  px t V,
t 1 (4.4)

where B is the benefit payable should the insured die during the year. The left side
of this equation is the benefit reserve at time t, combined with the benefit premium
at time t and accumulated with interest for a year. This is expected to be sufficient to
provide for the death benefit in the event that the insured dies during the year, or
alternatively, the next year’s benefit reserve, should the insured live. In Equation
(4.4), the death benefit, benefit premium and interest rate are all assumed to be
level. However, this formula easily generalizes to the case where any or all of these
values can vary with t.

Example 4.12: (Extension of Example 4.9) Using Equation (4.4), calculate the
benefit reserve at duration 6 for the policy described in Example 4.2.

Solution: It was determined in Example 4.9 that the benefit reserve at duration 5
was 887.50. Then, from Equation (4.4):

(5V  P)(1  i )  q35 B  p35 6V

(887.50  387.73)(1.06)  (0.00101)(300,000) + (0.99899) 6V
6 V  1,048.49.

Recursion reserve equations such as Equation (4.4) often simplify the valuation
process and are commonly used in practice.

Another type of reserve that is frequently calculated is the gross premium reserve.
The basic concept behind the calculation of gross premium reserves is similar to
that of benefit premium reserves in that the reserve is still the expected present
value of future loss. However, the gross premium reserve uses gross premiums
rather than benefit premiums and incorporates expenses (but not profits) into the
valuation process. Thus, the gross premium reserve at time t (again denoted by t V )
can be written as

t V  EPVt (future benefits and expenses)  EPVt (future gross premiums) . (4.5)

By convention, this calculation includes any expenses that are due at time t, but does
not include any expenses related to benefits payable at the end of the previous year.

At time 0, the gross premium reserve is usually a negative value with the magnitude
representing the EPV of future profits for the policy. The following examples
demonstrate the use of Equation (4.5).

37 | P a g e
Example 4.13: (Extension of Example 4.5) A person age 50 purchases a whole life
insurance policy with a death benefit of $250,000. Gross premiums are payable
annually. If the company’s expenses are as described in Example 4.5, calculate the
gross premium reserve for this policy at duration 10.

Solution: It was determined in Example 4.5 that the gross premium for this policy
was 3,669.87. Then, from Equation (4.5),

10V  EPV10 (future benefits and expenses)  EPV10 (future gross premiums)
 250, 000 A60  50a60  3669.87 a60
 31,482.77.

Note that the only expenses included in this calculation are the maintenance
expenses, as the other types of expenses for this policy are incurred prior to time 10.

Example 4.14: A person age 45 purchases a 20‐year endowment insurance policy

that will pay $500,000 either at the end of the year of death, or at the end of 20
years if he is still alive at that time. A gross premium of G is due at the beginning of
each year the policy is in‐force. The company’s expenses are given below.

Issue Expenses: $300 per policy, incurred at policy issue

Commissions: 40% of the first gross premium, and 6% of

subsequent premiums

Premium Taxes: 3% of all gross premiums paid

Maintenance Expenses: $50 per policy, incurred at the beginning of each

year the policy is in‐force

a) Calculate the gross premium G.

b) Calculate the gross premium reserve at duration 5, 5V .

Solution:

a) The gross premium can be calculated from Equation (4.2):

EPV (gross premiums)  EPV (benefits and expenses paid)

G a45:20 = 500,000 A 45:20  300  0.06G a45:20
 0.34G  0.03G a45:20  50 a45:20
G (11.79723)  500,000 (0.33223)  300  0.06G (11.79723)
 0.34G  0.03G (11.79723)  50 (11.79723)
G  16, 065.25

b) The gross premium reserve is calculated by using Equation (4.5):

38 | P a g e
 V  EPV5 (future benefits and expenses)  EPV5 (future gross premiums)
5

 500,000 A 50:15  0.06G a50:15  0.03G a50:15  50 a50:15  G a50:15

 500,000 A 50:15  0.09(16,065.25) a50:15  50 a50:15  (16,065.25) a50:15
 72,446.

Example 4.15: Extending the previous example, calculate the gross premium
reserve t V for the policy described in Example 4.14 at time t, for t = 0, 1, 2, … 19, 20.
Plot t V as a function of t.

Solution: The gross premium reserve values for this policy are shown in Figure 4.2.
500000
400000
Gross Premium Reserve

300000
200000
100000
0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Year

Figure 4.2 – Gross Premium Reserves for Example 4.15

Note that the value of the gross premium reserve for this policy approaches the
endowment amount of $500,000 as the policy nears expiration at time 20.

As was done for benefit reserves, it is possible to write recursion equations for gross
premium reserves. For a life insurance policy issued to a person age x, a recursion
relation between the gross premium reserves at times t and t+1 is given by:

(t V  G  eb )(1  i )  qx t ( B  ed )  px t V,
t 1 (4.6)

39 | P a g e
where G denotes the gross premium, eb represents the total amount of expenses
payable at the beginning of the year (which could include premium‐related
expenses, and maintenance or other expenses), and ed denotes any expenses
payable at the end of the year, on the death of the insured.

Example 4.16: (Extension of Example 4.13) Using Equation (4.6), calculate the
benefit reserve at duration 11 for the policy described in Example 4.13.

Solution: In Example 4.13, the gross premium reserve at time 10 for this policy was
calculated to be 31,482.77. The only expense payable during the year is the
maintenance expense at the beginning of the year. Then Equation (4.6) can be used
to calculate the gross premium reserve at time 11:

(10 V  P  eb )(1  i )  q60 ( B  ed )  p60 11V

(31, 482.77  3, 669.87  50)(1.06)  (0.00895)(250,000) + (0.99105) 11V
11V  35,287.12

Beyond the types of reserves discussed here, there are many other varieties of
reserves that are calculated for a myriad of reasons. Discussion of some of the other
common types of reserves, e.g., modified reserves, can be found in standard life
contingencies texts.

4.4. Exercises
(Assume an interest rate of i = 6% and that mortality is governed by Table 2.1 unless
otherwise specified.)

Exercise 4.1*: A 25‐year term life insurance policy issued to a person age 40 has a
death benefit of $250,000 payable at the end of the year of death.

a) Calculate the annual benefit premium for this policy.

b) Calculate the benefit reserve for this policy at time 10.

c) Use the answer to the previous part, along with Equation (4.4), to calculate
the benefit reserve for this policy at time 11.

d) Now assume that the death benefit is changed so that the policy will pay
$250,000 plus the sum total of all benefit premiums paid to the insurance
company prior to death. If this new death benefit is still paid at the end of the
year of death, calculate the new benefit premium for this policy.

Exercise 4.2*: A person currently age 60 purchases a 20‐year endowment

insurance. A death benefit of $300,000 is paid at the end of the year of death; or if
the person attains age 80, an endowment benefit of $500,000 is paid at this time.
Assuming annual premium payments, and the following expense structure:

40 | P a g e
Issue Expenses: $200 per policy, incurred at policy issue

Commissions: 20% of the first gross premium, and 5% of

subsequent premiums

Premium Taxes: 2% of gross premiums

Profits: 3% of gross premiums

a) Calculate the annual benefit premium.

b) Calculate the benefit reserve for this policy at time 5.

c) Calculate the annual gross premium.

d) Calculate the gross premium reserve for this policy at time 5.

Exercise 4.3*: A person age 30 wants to buy a 25‐year term insurance with level
annual premiums and death benefit B payable at the end of the year of death. She
can afford an annual gross premium of no more than $500 per year for this
insurance policy. Assuming that the insurance company has the following expense
structure:

Underwriting Expenses: $3 per $1,000 of policy death benefit, incurred at

policy issue

Other Issue Expenses: $200 per policy, incurred at policy issue

Maintenance Expenses: $20 per policy, incurred at the start of each year
the policy is in‐force

Calculate the maximum death benefit amount B that she can purchase.

Exercise 4.4*: You are given the following information about a whole life policy
purchased from a life insurance company by a 45‐year old:

Gross Premium: G paid at the beginning of each year the person is

alive

Death Benefit: $500,000 payable at the end of the year of death

Premium Taxes: 2% of gross premiums paid

Issue Expenses: $5,000 at the time of policy issue

Maintenance Expenses: $250 at the beginning of each year (including the

first year)

Termination Expenses: $1,000 at the time of death benefit payment

41 | P a g e
 a) Calculate the amount of the annual gross premium P.

b) Calculate the gross premium reserve at time 20.

c) Use the answer to the previous part, along with Equation (4.6), to calculate
the gross premium reserve for this policy at time 21.

d) Now suppose that the insurance company is considering replacing the sales
representatives at the company with commissioned brokers. Doing so would
eliminate the policy's Issue Expenses and Maintenance Expenses, but would
instead require the company to pay a commission to the broker. This
commission would be paid as y percent of the gross premiums paid, incurred
at the time of premium payment. (The Premium Tax Expenses and
Termination Expenses would be unchanged.) If, after making this
replacement, the total EPV of expenses remains the same, calculate y.

42 | P a g e
5. Adjustments for Non‐Annual Cases
The development presented in the previous chapters has assumed that all cash
flows happen on an annual basis. However, in reality, cash flows for life‐contingent
instruments such as life insurance and annuities often occur at various points within
a year. Payments may happen on a more frequent regular (e.g., monthly) basis, or at
any given time during the year as a result of a life‐contingent event, such as death
benefit payment.

This chapter presents an approach to adjusting the concepts and formulas discussed
earlier for these types of situations; these methods lend themselves easily to
spreadsheet implementation. A more thorough discussion of continuous models and
approximations thereof can be found in any standard life contingencies text.

Section 5.1 introduces the UDD (Uniform Distribution of Deaths) fractional age
assumption, and illustrates how it can be used to analyze life‐contingent cash flows
occurring more frequently than annually. Section 5.2 considers approximations
applicable to cash flows that may occur at any point within a period.

5.1. The Uniform Distribution of Deaths Assumption

The survival models used in this study note have been presented in the form of life
tables, such as Table 2.1. Most life tables are based on annual increments and thus
only contain information regarding the number of people living at integral ages.
However, in order to proceed with valuing cash flows occurring at intermediate
ages, information regarding survival probabilities at these ages will be needed.
Fractional age assumptions enable these calculations by positing a particular
survival model within each integral age. That is, while a life table provides values of
 x for integer values of x, a fractional age assumption provides a means for
determining  x for the non‐integral values of x.

The fractional age assumption used here is known as the Uniform Distribution of
Deaths (UDD) assumption. Under this assumption, for an integer x and t  [0,1) ,  x t
is defined as

 x t  (1  t )   x  t   x 1. (5.1)

Note that this assumption is simply linear interpolation of the  x function between
integer ages. Combining Equation (5.1) with Equations (2.2) and (2.3) yields the
following relationships, under the UDD assumption for integer x and t  [0,1) :

 x t (1  t )   x  t   x 1
t px    (1  t )  t  px  1  t (1  px )  1  t  qx . (5.2)
x x

43 | P a g e
 t qx  1 t px  1  1  t  qx   t  qx . (5.3)

It can also be shown that under the UDD assumption,

| qx  t  qx
s t (5.4)

for integer x and s, t , s  t  [0,1) . Note that since the UDD assumption is a piecewise
interpolation of the  x function between consecutive integral values of x, Equation
(5.4) is only directly applicable when calculating mortality probabilities within a
given year, not probabilities spanning multiple years. The bounds placed on x, s and
t in Equation (5.4) correspond to these restrictions. Proof of this relationship is left
as an exercise.

Example 5.1: Calculate the probability that a person currently age 70:

a) Dies prior to reaching age 70.7

b) Dies between the ages of 70.2 and 70.7

c) Dies between the ages of 70.7 and 73.5

Solution:

a) 0.7 q70  0.7q70  (0.7)(0.02041)  0.01429

b) | q70  0.5 q70  (0.5)(0.02041)  0.01021

0.2 0.5

c) The quantity to be calculated here is | q70 . Since this probability

0.7 2.8

encompasses a period spanning multiple ages, Equation (5.4) cannot be used

directly. However, Equations (2.4) and (5.1) can be utilized to calculate the
desired probability:

 70.7   73.5 [(1  0.7) 70  (0.7) 71 ]  [(1  0.5) 73  (0.5) 74 ]

| q70 
0.7 2.8 
 70  70
[(1  0.7)(78,522)  (0.7)(76,919)]  [(1  0.5)(73,350)  (0.5)(71,378)]

78,522
 0.06413
The UDD assumption can be very useful for cases involving non‐annual cash flows.
Specifically, this assumption can be used to modify an annual life table to provide
information regarding mortality on shorter time intervals, such as quarterly or
monthly. As an illustration, consider using Equation (5.1) to produce a quarterly
mortality table from Table 2.1, an annual life table. Table 5.1 gives an excerpt of the
resulting quarterly life table.

44 | P a g e
 Table 5.1 – Quarterly Life Table Under UDD Assumption

x lx x lx x lx
  42.25 96,707 44.75 96,270
40 97,033 42.5 96,667 45 96,222
40.25 96,999 42.75 96,626 45.25 96,170
40.5 96,964 43 96,586 45.5 96,117
40.75 96,930 43.25 96,543 45.75 96,065
41 96,895 43.5 96,499 46 96,012
41.25 96,858 43.75 96,456 46.25 95,954
41.5 96,821 44 96,412 46.5 95,896
41.75 96,784 44.25 96,365 46.75 95,837
42 96,747 44.5 96,317  

All of the formulas previously developed for annual life‐contingent cash flows are
easily adaptable to situations involving cash flows of other frequencies. Thus, life‐
contingent cash flows of a given frequency can easily be analyzed, once the
appropriate life table has been created; this is illustrated by the following examples.

Example 5.2: A person currently age 42 purchases a 3‐year term insurance, with a
death benefit of $200,000 payable at the end of the quarter of death. Level
premiums are payable at the beginning of each quarter that the policy is in force.
Using Table 5.1 and an annual effective interest rate of i = 6%, use the Equivalence
Principle to calculate the quarterly benefit premium for this policy.

Solution: Calculating the benefit premium requires determining the EPV of both the
benefit premiums and the benefits paid; in both cases, Equation (3.7) can be used.
Denoting the quarterly benefit premium by P,

EPV (benefit premiums) = P + P 0.25 p42 v 0.25  P 0.5 p42 v 0.5    P 2.75 p42 v 2.75
   42.25   1 
0.25
  42.5   1 
0.5
  44.75   1  
2.75

 P 1 +            
   42   1.06    42   1.06    42   1.06  
 11.06447 P, and

45 | P a g e
EPV (benefits paid) = 200, 000 0.25 q42 v 0.25 + 200, 000 |
0.25 0.25 q42 v 0.5    200, 000 |
2.75 0.25 q42 v3
  42.25   1 0.25   42.25   42.5   1 0.5
 200, 000 1    +  
  42   1.06    42   1.06 
    1 
3

    44.75 45    
  42   1.06  
= 985.33.

Then Equation (4.1) yields a quarterly benefit premium of P  89.05.

Example 5.3: (Extension of Example 5.2) Calculate the benefit reserve at time 1.25
for the policy described in Example 5.2.

Solution: Equation (4.3) can be used to calculate the benefit reserve as follows:

1.25 V  EPV1.25 (future benefits to be paid)  EPV1.25 (future premiums to be collected),

where the EPV at time 1.25 of the future benefits to be paid is

EPV (future benefits) = 200, 000  q

0.25 43.25 v 0.25 + |
0.25 0.25 q43.25 v 0.5    1.5 | 0.25 q43.25 
  43.5   1 
0.25
  43.5   43.75   1 
0.5

 200, 000 1    +  

  43.25   1.06    43.25   1.06 
    1 
1.75

    44.75 45    
  43.25   1.06  
= 625.95,

and the EPV at time 1.25 of the future premiums is

EPV (benefit premiums) = 89.05 + (89.05) 0.25 p43.25 v 0.25  (89.05) 0.5 p43.25 v 0.5
   (89.05) 1.5 p43.25 v1.5
   43.5   1 
0.25
  1 
0.5

 89.05 1 +      43.75   
   43.25   1.06    43.25   1.06 
  44.75   1  
1.5

     
  43.25   1.06  
 596.14,

so that 1.25 V  625.95  596.14  29.81.

46 | P a g e
Example 5.4: (Extension of Example 5.3) Using the result of Example 5.3, calculate
the benefit reserve at time 1.5 for the policy described in Example 5.2.

Solution: Equation (4.4) can be adapted for use on a quarterly basis in order to
perform this calculation:

(1.25V  P)(1  i )0.25  0.25 q43.25 B  0.25 p43.25 1.5V

(29.81  89.05)(1.06)0.25  (0.00045)(200, 000)  (0.99955)1.5V
1.5 V  30.62.

Note that the premium, interest rate and probabilities were all adapted to quarterly
values.

Thus, all of the concepts and formulas derived in previous chapters can be adapted
to a given payment frequency. The UDD assumption can be used as shown above to
calculate mortality probabilities, enabling the analysis of life‐contingent cash flows.

5.2. Approximating the Continuous Case

The previous section discussed adjustments that can be made to analyze life‐
contingent cash flows occurring on bases more frequent than annual. This can be
very useful for payments such as premiums that may be made on a monthly or
quarterly basis. However, this may not be adequate for all types of life‐contingent
cash flows. For example, death benefits for life insurance policies are often assumed
to occur at the time of death of the insured. These cases (sometimes referred to as
continuous cases) require additional methods or assumptions in order to value the
associated cash flows.

One possible approximation that is commonly used is to assume that certain cash
flows – in particular, those that might happen at any point within a given period –
occur precisely in the middle of the period in question. Thus, it might be assumed,
for example, that life insurance death benefits (and any expenses associated with
their payment) are incurred mid‐period, whereas premiums and other expenses are
still payable at the beginning of the period. This approximation can be applied in
annual or more frequent periods, depending on the nature of the problem; here it
will be assumed that the length of the period corresponds to the premium paying
frequency.

With respect to finding the EPV of payments assumed to occur mid‐period, the only
modification necessary is to change the length of time for which these payments are
discounted with interest. This is illustrated through the following examples.

Example 5.5: (Extension of Example 5.2) Repeat Example 5.2, assuming now that
death benefits are payable at the moment of death, approximated by a mid‐quarter
payment in the quarter in which the insured dies. Again, find the quarterly benefit

47 | P a g e
premium for this policy, assuming premiums are payable at the beginning of each
quarter.

Solution: The EPV of the benefit premiums is unchanged from Example 5.2. The EPV
of the benefits is:

EPV (benefits paid) = 200, 000  0.25 q42 v 0.250.125 + |

0.25 0.25 q42 v 0.50.125    |
2.75 0.25 q42 v 30.125 
   1 
0.125
    1 
0.375

 200, 000 1  42.25    +  42.25 42.5   

  42   1.06    42   1.06 
    1 
2.875

    44.75 45    
  42   1.06  
= 992.53.

Then Equation (4.1) yields

EPV (benefit premiums) = EPV (benefits paid)
11.06447 P = 992.53

so that the quarterly benefit premium is P  89.70.

The death benefits in this example were discounted with interest for one‐half period
less time (half a quarter or 0.125 years in this case) than was the case in Example
5.2; this is the only change necessary to accommodate the continuous death benefit
assumption.

Example 5.6: (Extension of Example 5.5) Repeat Example 5.5, additionally

assuming now that expenses are incurred as described below. Calculate the gross
premium for this policy.

Premium Taxes: 2% of gross premiums paid

Maintenance Expenses: $50 at the beginning of each quarter

Termination Expenses: $1,000 at the time of death benefit payment

Solution: The EPV of the benefits paid are unchanged from the previous example.
Let G denote the quarterly gross premium. The EPV of the gross premiums can be
found (in a similar manner to the previous example) to be 11.06447 G. The EPV of
the expenses is:

48 | P a g e
EPV (expenses paid) = 0.02G 1  0.25 p42 v 0.25  0.5 p42 v 0.5    2.75 p42 v 2.75 
 50  1  0.25 p42 v 0.25  0.5 p42 v 0.5    2.75 p42 v 2.75 
 1, 000  q v 0.125 +
0.25 42 |
0.25 0.25 q42 v 0.375    |
2.75 0.25 q42 v 2.875 
 0.02(11.06447)G  50(11.06447)  4.96
 0.22128G  558.18.

Then Equation (4.2) yields:

EPV (gross premiums) = EPV (benefits and expenses paid)
11.06447 G  992.53  (0.22128)G  558.18
G = 143.01.

Note here that the expenses associated with the death benefit payment occur at the
same time as the benefit payment, and hence are placed at mid‐quarter. All other
expenses were still assumed to occur at the beginning of each quarter.

This idea of approximating a continuous case by a mid‐period payment can be

utilized with any of the formulas previously developed, and with any frequency of
period (annual, quarterly, monthly, etc.)

5.3. Exercises
(Assume an interest rate of i = 6% and that mortality is governed by Table 2.1 and
the UDD fractional age assumption unless otherwise specified.)

Exercise 5.1: Show that that under the UDD assumption, s | t qx  t  qx for integer x and
s, t , s  t  [0,1) .

Exercise 5.2: Calculate the probability that a person currently age 85:

a) Dies between the ages of 86.2 and 88.6

b) Lives to at least age 89.3

Exercise 5.3: A person currently age 41 purchases a 2‐year term insurance, with a
death benefit of $100,000 payable at the end of the quarter of death. Level
premiums are payable at the beginning of each quarter that the policy is inforce.

a) Using Table 5.1 and an interest rate of i = 6%, calculate the quarterly benefit
premium for this policy.

b) Assuming that expenses are identical to those given Exercise 4.2, calculate
the quarterly gross premium for this policy.

c) Calculate the benefit reserve at time 0.75 for this policy.

49 | P a g e
 d) Using Equation (4.6) and the result from the previous part, calculate the
benefit reserve at time 1 for this policy.

Exercise 5.4*:

a) Using the UDD assumption, augment Table 2.1 to create a quarterly life table
for ages 0, 0.25, 0.5, … , 110.75, 111.

b) A 25‐year term life insurance policy issued to a person age 40 has a death
benefit of $250,000 payable at the end of the quarter of death. Calculate the
quarterly benefit premium for this policy.

c) Repeat the previous part, assuming now that the death benefit is payable at
the moment of death (premiums are still payable at the beginning of each
quarter). Use the mid‐quarter payment approximation for the death benefit
payment.

50 | P a g e
 References
[DHW09] David C. M. Dickson, Mary R. Hardy, and Howard R. Waters. Actuarial
Mathematics for Life Contingent Risks. Cambridge University Press,
2009.

[CHL11] Robin J. Cunningham, Thomas N. Herzog, and Richard L. London.

Models for Quantifying Risk. ACTEX Publications, fourth edition, 2011.

[DV07] James W. Daniel and Leslie J. F. Vaaler. Mathematical Interest Theory.

Pearson Prentice Hall, 2007.

[Bro08] Samuel A. Broverman. Mathematics of Investment and Credit. ACTEX

Publications, fourth edition, 2008.

[Kel08] Stephen G. Kellison. The Theory of Interest. McGraw‐Hill / Irwin, third

edition, 2008.

[Klu12] Stuart A. Klugman, ed. Understanding Actuarial Practice. Chicago:

Society of Actuaries, 2012.

[HT09] Robert V. Hogg and Elliot A. Tanis. Probability and Statistical

Inference. Prentice Hall, eighth edition, 2009.

[NCHS] U.S. Centers for Disease Control and Prevention / National Center for
Health Statistics
(http://www.cdc.gov/nchs/products/life_tables.htm)

[BGHJN97] Newton L. Bowers, Hans U. Gerber, James C. Hickman, Donald A.

Jones, Cecil J. Nesbitt. Actuarial Mathematics. Society of Actuaries,
second edition, 1997.

51 | P a g e