PENSION MATHEMATICS

with Numerical Illustrations

Second Edition

Howard E. Winklevoss, Ph.D., MAAA, EA

President
Winklevoss Consultants, Inc.

Published by

Pension Research Council

Wharton School of the University of Pennsylvania

and

University of Pennsylvania Press

Philadelphia
© Copyright 1977 (first edition) and 1993 (second edition) by the
Pension Research Council of the Wharton School of the University of Pennsyl-
vania

All rights reserved

Library of Congress Cataloging-in-Publication Data

Winklevoss, Howard E.
Pension mathematics with numerical illustrations / Howard E.
Winklevoss. -2nd ed.
p. em.
Includes bibliographical references and index.
ISBN 0-8122-3196-1
I. Pensions-Mathematics. 2. Pensions-Costs-Mathematics. 3. Pension
trusts-Accounting. I. Title.
HD7105.W55 1993
331.25'2-dc20 92-44652
CIP

Printed in the United States of America

Chapter 3
Basic Actuarial Functions

The purpose of this chapter is to introduce several actuarial

functions used in the development of pension mathematics
throughout the remainder of the book. The discussion begins
with the composite survival function and interest function, per-
haps the two most basic concepts in pension mathematics. Pen-
sion plan benefit functions are then presented, followed by a dis-
cussions of annuities, the latter representing a combination of
interest and survival functions.

COMPOSITE SURVIVAL FUNCTION

The composite survival function represents the probability

that an active plan participant survives in service for a given pe-
riod, based on all of the decrement rates to which the employee is
exposed. Whereas the probability of surviving one year in a sin-
gle-decrement environment is equal to the complement of the
rate of decrement, the probability of surviving one year in a mul-
tiple-decrement environment is equal to the product of such
complements for each applicable rate of decrement. The proba-
bility of an active participant aged x surviving one year is
p~T} = (1 - q~m») (1 - qi'») (1 - qi d ») (1 - qi r») , (3.1a)
or equivalently,
PY) = Plm ) pi') pld ) plr ). ([Link])
This same probability can be expressed in terms of multiple-
decrement probabilities:
PY) =1 - (qi m) + qi') + qid ) + ql'»). (3.2a)

31
32 Pension Mathematics

A common approximation for multiple-decrement probabilities

is illustrated for the mortality probability as follows:
ql m) '" q;m) (1 _ ~q;t»)(l _ ~q~d») (1 _ ~q~r»). (3.2b)
The probability of surviving in active service for n years is equal
to the product of successive one-year composite survival proba-
bilities:
n-I
n Px
(T) -- II p(T)
x+t· (3.2c)
1=0
Table 3-1 shows the probability of surviving to age 65 from
each attained age, based on nine different entry ages, under the
various rates of decrement set out in Chapter 2. Since the termi-
nation decrement includes a 5-year select period, probabilities
are given for each of the first five years subsequent to the nine
entry ages. The chance of reaching retirement, at least under the
illustrative assumptions, is quite low even up through attained
age 50, where the probability is 0.62. Since retirement-related
pension cost estimates are in direct proportion to the survival
probability (Le., the lower the probability the lower are such es-
timates), it is clear that discounting for the various decrements
reduces pension costs significantly. Obviously, the cost reduc-
tion is mitigated to the extent that a benefit is provided with each
type of decrement (e.g., a vested termination benefit, a death ben-
efit, or a disability benefit).
The calculation of survival probabilities in accordance with
the above definitions can be quite arduous. The probabilities,
however, are readily obtainable from a service table. This table
shows the hypothetical number of employees, from an arbitrary
initial number, who survive to each future age. The initial num-
ber, or radix, is generally taken to be some large value, such as
100,000 or 1,000,000, and IY) is the notation for the survivors at
age x. 1 The total number of employees leaving active service dur-
ing the year is denoted by d;T) and defined as
d;T) = IY)qY). (3.3a)

1Although not indicated by the symbol Ix(T), it is understood to represent

the number of survivors at age x who entered the plan at age y. At a later point
the more general symbollx\9 will be used, which makes the entry age variable
explicit.
3. Basic Actuarial Functions 33

TABLE 3-1
Probability of Surviving in Service to Age 65, .. -xp!}

Select Ages
Entry Attained
Age y y+l y+2 y+3 y+4 x Age
20 0.02 0.03 0.04 0.05 om 0.08 25
0.10 26
0.11 27
0.13 28
0.15 29
25 om 0.09 0.11 0.13 0.15 0.17 30
0.19 31
0.21 32
0.23 33
0.25 34
30 0.14 0.17 0.20 0.23 0.25 0.28 35
0.30 36
032 37
034 38
0.36 39
35 0.24 0.28 0.31 0.33 0.36 0.38 40
0.40 41
0.43 42
0.45 43
0.47 44
40 0.35 0.38 0.41 0.44 0.47 0.49 45
052 46
054 47
057 48
0.60 49
45 0.47 0.51 0.54 0.57 0.60 0.62 50
0.65 51
0.69 52
0.72 53
0.75 54
50 0.51 0.55 0.58 0.61 0.64 0.79 55
0.80 56
0.81 57
0.82 58
0.84 59
55 0.59 0.63 0.67 0.70 0.74 0.85 60
0.87 61
Q.89 62
0.92 63
0.96 64
60 0.74 0.79 0.84 0.89 0.94 1.00 65
34 Pension Mathematics

Total decrements from the active population equal the sum of

each separate decrement:
diT) = d;m) + di /) + did) + di r ) (3.3b)

= IY) (qi m) + qi') + qi d ) + qi r ». (3.3c)

Table 3-2 illustrates the concept of a service table for
1,000,000 entrants at age 20, based on the decrement assumptions
specified in Chapter 2. 2 The probability of an age-20 entrant
surviving in active service to age 65 is easily found from this
table:
(T) _ Ig) _ 24,448
6S-20P20 -/(T) - 000000 = 0.0244. (3.4a)
20 1, ,
Similarly, the probability of an employee age 40 surviving to age
65 would be
l
(T) = Ig = 24,448 = 0.3745. (3.4b)
2SP4Q (T) 65276
140 '

Although service tables are an important source of computa-

tional efficiency, even when working with high speed computers,
they are used only occasionally in presenting the theory of pen-
sion mathematics in this book.

INTEREST FUNCTION

The interest function is used to discount a future payment to

the present time. It plays a crucial role in determining pension
costs and, like the survival function of the previous section, it re-
duces such values. If i, is the interest rate assumed for the tth
year, the present value of one dollar due in n years is given by
1 (3.5a)
(1 + il) (1 + i2) ... (1 + in)
and, if h = iz =... = in , we have
1 (3.5b)
(1 + if

2To the extent that any rates are entry-age dependent. such as termination
rates, a separate service table is applicable to each entry age.
3. Basic Actuarial Functions 3S

TABLE 3-2
Service Table

x I}T) d~m) dY) d~d) dY) dY)

20 1,000,000 442 243,002 263 0 243,708

21 756,292 350 169,718 201 0 170,270
22 586,023 286 . 121,314 158 0 121,757
23 464,265 238 88,543 126 0 88,907
24 375,358 202 65,921 103 0 66,226
25 309,132 176 49,933 85 0 50,194
26 258,938 156 38,460 72 0 38,688
27 220,251 140 3O,OW 62 0 30,251
28 189,999 129 23,814 53 0 23,996
29 166,004 119 19,113 47 0 19,280
30 146,724 112 15,529 56 0 15,697
31 131,027 107 12,754 50 0 12,911
32 118,116 103 10,576 45 0 10,725
33 107,392 101 8$15 41 0 9,017
34 98,375 99 7,510 38 0 7,647
35 90,727 98 6,419 35 0 6,552
36 84,176 98 5,534 41 0 5,673
37 78,503 99 4,816 46 0 4,960
38 73,543 100 4,224 50 0 4,374
39 69,169 102 3,738 54 0 3,893
40 65,276 104 3,338 57 0 3,499
41 61,777 108 3,004 60 0 3,172
42 58,605 114 2,7Z7 69 0 2,910
43 55,695 123 2,491 76 0 2,690
44 53,006 133 2,290 83 0 2,506
45 50,499 144 2,121 89 0 2,354
46 48,145 156 1,969 94 0 2,219
47 45,926 169 1,841 99 0 2,108
48 43,818 181 1,721 107 0 2,009
49 41,808 194 1,616 115 0 1,925
50 39,884 206 1,517 121 0 1,845
51 38,039 219 1,424 127 0 1,769
52 36,270 230 1,335 135 0 1,700
53 34,570 241 1,244 142 0 1,628
54 32,942 252 1,159 148 0 1,559
55 31,383 267 0 156 0 423
56 30,960 286 0 166 0 452
57 30,508 305 0 182 0 487
58 30,020 326 0 203 0 529
59 29,491 350 0 235 0 585
60 28,907 377 0 281 0 659
61 28,248 405 0 348 0 753
62 27,495 433 0 436 0 869
63 26,626 459 0 549 0 1,008
64 25,618 485 0 685 0 1,170
65 24,448 0 0 0 24,448 24,448
36 Pension Mathematics

The following simplifying definition is used in connection with

the present value function:
v = _1_. (3.6)
(1 + i)
Thus, v n represents the present value of one dollar due in n years
at an annual compound rate of interest equal to i.
The interest function, vt, begins at a value of unity for t = 0
and approaches zero as t approaches infinity, provided i > O.
Also, v t is inversely related to i, taking on the value of unity for
i = 0 and approaching zero as i approaches infinity. Table 3-3 il-
lustrates the significance of the interest factor. In addition to
showing vt for the illustrative interest assumption of 8 percent,
the function is also evaluated for interest rates of 6 and 10 per-
cent. Since the interest factor is associated with each entrant's
potential future age, a period of about 70 years for entrants age 30
or younger, the interest discount function is tabulated at 5-year
intervals for 70 years.
The significance of the interest rate function is readily appar-
ent. For an age-30 entrant, v 65 - 30 is 0.07 at an 8 percent rate of
interest, 0.13 at 6 percent, and 0.04 at 10 percent. Retirement-re-
lated cost estimates are directly related to this factor and, since
the interest discount extends beyond retirement, the total effect
of the interest assumption is even greater than the effect of vr-y
alone.
Figure 3-1 shows the survival function from age x to age 65
and the interest function over this same age interval for an age-30
entrant under the model assumptions. This graph indicates that
the interest function, based on an 8 percent rate, is smaller than
the survival function at all ages, although this relationship may
not always hold. The product of the interest and survival func-
tions is frequently encountered in pension mathematics. Since
both functions are less than unity, their product is quite small
over most attained ages.

SALARY FUNCTION

If a pension plan has benefits expressed in terms of salary, it

is necessary to develop salary-related notation and procedures
for estimating future salary. The current dollar salary for a par-
 3. Basic Actuarial Functions 37

TABLE 3-3
Compound Interest Function, vi
Interest Rate

6% 8% 10%

0 I.()OO) I.()OO) I.()OO)

5 0.7473 0.68lXi 0.6209
10 0.5584 0.4632 0.3855
15 0.4173 0.3152 0.2394
20 0.3118 0.2145 0.1486
25 0.2330 0.1460 0.0923
30 0.1741 0.0994 0.0573
35 0.1301 0.0676 0.0356
40 0.0972 0.0460 0.0221
45 0.0727 0.0313 0.0137
50 0.0543 0.0213 0.0085
55 0.0406 0.0145 0.0053
60 0.0303 0.0099 0.0033
65 0.0227 0.0067 0.0020
70 0.01&'! 0.0046 0.0013

FIGURE 3-1
Survival and Interest Functions from Age x to Age 65
ProbabilitylDollars
1.0 " T ' " " - - - - - - - - - - - - - - - - - - - - - - - - . . . . . . "... 1.0

0.9 0.9

us 0.8

0.7 0.7

0.6 0.6

us 05

0.4 0.4

U3 03

Q.2 02

Q.l 0.1

0.0 +------.---"T"""-----.---"T"""-----.---"T"""---+ 0.0

30 35 40 45 50 55 60 65
Age
38 Pension Mathematics

ticipant age x is denoted by sx, and Sx represents the cumulative

salary from entry age y up to, but not including, age x. 3 Thus, for
x> ywe have
x-I
Sx = I St· (3.7)
t= Y

In order to estimate the dollar salary at age x, based on the

employee's age-y salary, the following formula is used:
S = S (SS)x [(1 + J)(1 + P)] (x - y) (3.8a)
x y (SS)y ,
where
Sy = entry-age dollar salary
(SS)x = merit salary scale at age x
I = rate of inflation
P = rate of productivity reflected in the salary increases.

An age-y entrant's salary at age x can also be defined in terms

of the age-z salary (y < z < x):

Sx = Sz i:~: [(1 + J)(1 + pW- z). (3.8b)

If all of the salary increase assumptions were met from age y to

age z, the employee's salary at age z would be equal to

Sz = Sy i:~:[(1 + J)(1 + P)](z-y). ([Link])

Substituting (3.8c) for Sz in (3.8b) reduces the latter to (3.8a),

showing that Sx is identical, under the salary increase assump-
tions, whether derived from the entry age salary or the attained
age salary.
Table 3--4 shows the salary function per dollar of entry-age
salary based on the previously discussed merit scale, 1 percent

3rt is to be understood throughout the remainder of this book that Sx and

Sx are dependent on each participant's entry age y, despite the fact that y does
not appear in the symbol. The symbols Sx,y and Sx,y could be used to denote
these two functions, which have the virtue of making the entry age explicit.
These symbols are used to make the entry age variable explicit at a later point
where the equations require a summation over all combinations of x and y.
3. Basic Actuarial Functions 39

TABLE 3-4
Salary Function per Dollar of Entry Age Salary, Sx + Sy

Entry Age, Y

x 20 30 40 50 60
20 1000
21 1.097
22 1.203
23 1.317
24 1.442
25 1.575
26 1.721
'lJ 1K77
28 2.045
29 2.228
30 2.422 1000
31 2.632 1.087
32 2.859 1.180
33 3.100 1.280
34 3.360 1.387
35 3.636 1.501
36 3.934 1.624
'51 4.249 1.754
38 4.587 1.894
~ 4.948 2.043
40 5.328 2200 1000
41 5.736 2368 1.077
42 6.166 2546 1.157
43 6.625 2735 1.244
44 7.108 2935 1.334
45 7.619 3.146 1.430
46 8.160 3.369 1.532
'iI 8.733 3.605 1.639
48 9.334 3.854 1.752
49 9.969 4.116 1.871
50 10.632 4.389 1.996 1000
51 11.331 4.678 2.1'lJ 1.066
52 12.065 4.981 2.264 1.135
53 12.833 5.298 2.409 1.207
54 13.638 5.630 2.560 1.283
55 14.474 5.976 2.717 1.361
56 15.354 6.339 2.882 1.444
57 16.262 6.714 3.052 1.530
58 17.215 7.107 3.231 1.619
Cf! 18.203 7.515 3.417 1.712
60 19.226 7.938 3.609 1.808 1000
61 20.291 8.377 3.808 1.908 1.055
62 21.391 8.831 4.015 2.012 1.113
63 22.526 9.300 4.228 2119 1.172
64 23.695 9.782 4.447 2.229 1.232
40 Pension Mathematics

productivity factor, and 4 percent inflation rate, illustrated for

decennial entry ages from 20 through 60. The age-64 salary is 24
times greater than the age-20 starting salary. At the other ex-
treme, an age-60 entrant's initial salary is expected to increase by
23 percent by age 64. Retirement-related cost estimates are di-
rectly proportional to an employee's finalS-year average salary
under the model benefit formula used to illustrate pension costs
in this book; consequently, it is clear from Table 3-4 that the
growth in a participant's future salary can increase pension cost
estimates substantially. This is in contrast to the interest rate and
decrement probabilities, both of which have a decreasing effect
on pension cost estimates.

BENEFrr FUNCfION

The benefit function is used to determine the amount of ben-

efits paid at retirement, vested termination, disablement, and
death. This function, the interest function, and the survival func-
tion provide the basic components required to formulate pension
costs, as shown in subsequent chapters. In this section consider-
ation is given to the three most common types of benefit formulas
used with defined benefit pension plans.
The symbol b x denotes the annual benefit accrual during age
x to age x + 1 for an age-y entrant, and is referred to as the benefit
accrual function. The benefit accrual function can equal the for-
mula accruals or, as discussed in this section, some other defini-
tion of accruals, such as a portion of the participant's projected
retirement-age benefit. The accrued benefit, denoted by B x , is
equal to the sum of each attained age accrual up to, but not in-
cluding, age x. This function is called the accrued benefit func-
tion and is defined for x > y by
x-I
Bx =L bt • (3.9)
l=y

The convention of using lower and upper case letters, with the
upper case denoting a summation of lower case functional values,
was also used for the salary scale. 4

4The benefit functions, b x and B x , are dependent on the employee's entry

age y, suggesting the nOlation bx,y and Bx,y, which is not used here for simplic-
ity.
3. Basic Actuarial Functions 41

flat Dollar Unit Benefit

Under a flat dollar unit benefit formula, b x is equal to the

annual benefit payable per year of service. These values are
found to range from about $150 to $300 per year of service.
Although the attained age subscript is shown here, it is unneces-
sary since the accrual is independent of age. The accrued benefit
is a years-of-service multiple of the benefit accrual, that is,
Bx = (x-y)b x . (3.10)

Career Average

The career average benefit formula has the following defini-

tions for the benefit accrual and the accrued benefit functions at
age x:
bx = k sx, (3.11a)
Bx = k Sx, (3.l1b)
where k denotes the proportion of attained age salary provided as
an annual benefit accrual. The benefit functions under the career
average formula follow precisely the pattern of the attained age
salary and the cumulative salary discussed and illustrated previ-
ously.

Final Average

The final average benefit formula is somewhat more compli-

cated. Let n denote the number of years over which the partici-
pant's salary prior to retirement is to be averaged, and let k equal
the proportion of the average salary provided per year of service.
The projected retirement benefit, assuming retirement occurs at
the beginning of age r, is defined as
r~l

Br =k (r-y)t I St, (3.12a)

f=r-n

or more simply
Br = k(r-y).l(Sr-
n
Sr-n). (3.12b)
The attained age benefit accrual and accrued benefit func-
tions can be defined in several ways under this benefit formula.
42 Pension Marhematics

One approach is to defined B t according to the benefit formula

based on the participant's current salary average:
Bx = k (x- y)1.n (Sx - Sx _ n) , (3.13)
where n is the smaller of the years specified in the benefit for-
mula or x - y. The corresponding benefit accrual at age x can be
determined by the following basic relationship:
bx = Bx+l - Bx . (3.14a)
Substituting the definition of B x given in (3.13) and simplifYing,
we have:
bx = kl(Sx+l
n - Sx+l-n)
+ k~(x-y)[(SX+l - Sx+l-n) - (Sx - Sx-n)] (3.14b)

= k~(Sx+l - Sx+l-n) + k~(x -y)[sx - Sx-n]. (3.14c)

The first term on the right side of (3.14c) is the portion of the
benefit accrual earned in the current year based on the partici-
pant's current n-year salary average, while the second term repre-
sents the portion earned as a result of the increase in the n-year
salary average base. The increase in the salary base, which is rep-
resented by ~ (sx - sx-n) in (3.14c), is multiplied by years-of-ser-
vice to date. Therefore, the benefit accrual during age x includes
an implicit updating of the previous accruals which can cause bx
to be a steeply increasing function of x.
As will be shown later, steeply increasing benefit accrual and
accrued benefit functions produce even steeper pension cost
functions under some methods of determining pension costs and
liabilities, characteristics that may be undesirable. Consequent-
Iy, two modifications to benefit functions that are based on the
plan's formula have been developed for funding purposes in or-
der to mitigate this effect.
The constant dollar (CD) modification defines the benefit ac-
crual function as a pro rata share of the participant's retirement-
age projected benefit:

CDb -
x -
~
(r- y) ,
(y :0:; x < r) (3.15a)

a constant for all x. The accrued benefit function is a years-of-

service multiple of this constant:
3. Basic Actuarial Functions 43

CDB x = ~(X - y). (y ~ X ~ r) (3.15b)

(r- y)
While the constant dollar modification is applicable to a career
average benefit formula, it has no effect on a flat dollar unit bene-
fit formula.
The constant percent (CP) modification defines b x as a con-
stant percent of salary. The appropriate percentage of attained
age salary is found by dividing the projected benefit by the cumu-
lative projected salary of the age-y entrant:

(y ~ x < r) (3.16a)

CPB x = B r Sx. (y ~ x ~ r) (3.16b)

Sr
The constant percent modification has no effect on a career aver-
age benefit formula, just as the constant dollar version has no ef-
fect on a flat dollar benefit formula.
Table 3-5 shows the benefit accrual and accrued benefit
functions under the final average benefit formula used in this
book (i.e., 1.5 percent of final 5-year average salary per year of
service) for an age-30 entrant. The constant dollar and constant
percent versions are illustrated along with the unmodified ver-
sion of this benefit formula. In each case, the benefit functions
are expressed as a percentage of the employee's projected retire-
ment benefit, with retirement assumed to occur at the beginning
of age 65. The constant percent version has only a minor effect as
compared to the unmodified benefit functions; however, the con-
stant dollar version produces significant differences. This modi-
fication allocates a constant 2.86 percent, or 1/35, of the projected
benefit per year of service. The constant dollar version develops
an accrued benefit equal to 50 percent of the projected benefit
when one-half of the expected career has been completed, while
the allocation under the constant percent is only 25 percent at
this point, and the unmodified version is 18 percent. Thus, even
though all of the accrued benefit functions begin at zero and at-
tain B r by age r, the modifications, and particularly the constant
dollar version, cause the intermediate values to be different.
Figures 3-2 and 3-3 show these relationships graphically.
44 Pension Mathematics

TABLE 3-S
Benefit Accrual and Accrued Benefit Functions
Expressed as a Percent of the Projected Benefit

llnwod;fi,d Can rtgnt Percent Con51ant Dollgr

Age b. Bx CDb CDB CPb CPB
x x x x

30 0.32 0.00 0.67 0.00 2.86 0.00

31 0.35 0.32 0.73 0.67 2.86 2.86
32 0.38 0.67 0.79 1.41 2.86 5.71
33 0.41 1.06 0.86 220 2.86 857
34 0.45 1.47 0.93 3.06 2.86 11.43
35 0.58 1.92 1.01 4.00 2.86 14.29
36 0.66 249 1.09 5.01 2.86 17.14
37 0.75 3.15 1.18 6.10 2.86 20.00
38 0.84 3.90 1.28 7.28 2.86 22.86
39 0.95 4.74 1.38 8.56 2.86 25.71
40 1.07 5.70 1.48 9.93 2.86 2857
41 1.19 6.76 1.59 11.42 2.86 31.43
42 1.33 7.95 1.71 13.01 2.86 34.29
43 1.47 9.28 1.84 14.73 2.86 37.14
44 1.63 10.75 1.98 16.57 2.86 40.00
45 [Link] 12.39 212 18.54 2.86 42.86
46 1.99 14.19 2.27 20.66 2.86 45.71
47 2.18 16.18 2.43 22.93 2.86 48.57
48 2.39 18.36 260 25.36 2.86 51.43
49 2.62 20.75 2n 27.95 2.86 54.29
50 2.86 23.37 296 30.73 2.86 57.14
51 3.11 26.23 3.15 33.68 2.86 60.00
52 3.38 29.34 3.35 36.83 2.86 62.86
53 3.66 32.71 357 40.19 2.86 65.71
54 3.96 36.38 3.79 43.76 2.86 68.57
55 4.28 40.34 4.02 47.55 2.86 71.43
56 4.61 44.62 4.27 51.57 2.86 74.29
57 4.96 49.23 4.52 55.84 2.86 n.14
58 5.32 54.19 4.79 60.36 2.86 [Link]
59 5.71 5951 5.06 65.15 2.86 82.86
60 6.10 65.22 5.35 70.21 2.86 85.71
61 6.51 71.32 5.64 75.56 2.86 88.57
62 6.95 n.83 5.95 81.20 2.86 91.43
63 7.38 84.78 6.26 87.15 2.86 94.29
64 7.84 92.16 6.59 93.41 2.86 97.14
65 100.00 100.00 100.00
3. Basic Actuarial Functions 45

FIGUREJ-2
Attained Age Benefit Accrual Functions
as 8 Percent of the Projected Retirement Benefit
Percent
8 ""T"------------------------,.. 8
7 7

6 6

5 5

4 4

3 3

2 2

0 0
30 35 40 45 50 55 60 65
Age

F1GUREJ-3
Attained Age Accrued Benem Functions
lIS a Percent of the Projected Retirement Benefit

Percent
100 100

90 90

80 80

70 70

60 60

50 50

40 40

30 30

20 20

10 10

0 0
30 35 40 45 50 55 60 65
Age
46 Pension Mathematics

ANNUITY FUNcrlONS

Annuities represent a combination of the survival and inter-

est functions. Most annuities are based on the mortality-only
survival function, and the material below reflects this emphasis.
However, this section also defines a temporary employment-
based annuity which uses the composite survival function, since
this annuity is required for some funding methods.

Straight Life Annuity

If retirement benefits cease upon death, the annuity is called

a straight life annuity and its present value, assuming an annual
benefit of one dollar payable at the beginning of age x, is given by
~

ax = I IPim) V t. (3.17a)
t = 0

The infinity sign is used as the upper limit of the summation for
simplicity, since IP}m) becomes zero beyond some advanced age. s
A special case of the life annuity is when the interest rate is
zero, in which case we have, simply, one plus the curtate life ex-
pectancy (i.e., based on whole years only) at age x. This is ex-
pressed notationally as

ex = [t~O m
IPi )] - 1. (31.7b)

Period Certain Life Annuity

It is not uncommon to find an n-year period certain life annu-

ity used as the basis for distributing pension benefits. During the
certain period, benefits are payable whether or not the annuitant
is alive. This type of annuity is a combination of an n-year period
certain annuity plus an n-year deferred life annuity:
n~l ] (m)
"-
a x :it1 =[ I v
I
+ nPx V
n"
a x +n
t = 0

5An approximation for the present value of an annuity payable m times a

year, with payments at the beginning of each period. is found by subtracting
(m - 1)l2m from the annuity payable annually. Since retirement benefits are
paid monlhly. 11/24 should be sublracted from the above annuity 10 approxi-
mate a monthly payment of 1/12 of a dollar. The annual-pay annuity is assumed
throughout this book for simplicity.
3. Basic Actuarial Functions 47

(3.18)
These equations introduce notations deserving comment. The
bar over the subscript x:nJ signifies that the annuity is paid until
the last status fails, where the two statuses represent (1) the life of
the plan member and (2) the n-year period. Thus, the period cer-
tain life annuity is technically a last survivor annuity. An annuity
discussed subsequently is ax :lil without the bar. This annuity is
known as an n-year temporary life annuity which pays until the
first of the two statuses fails and is technically a joint annuity. Fi-
nally, (3.18) uses two other symbols: ani indicating an n-year pe-
riod certain annuity and nlax denoting the present value of an n-
year deferred life annuity.

Joint and Survivor Annuity

Another type of annuity is known as the joint and survivor

annuity. The term "joint" suggests that the payment amount is
based on more than one status, and the term "survivor" suggests
that it pays as least some amount until the last status fails. For
example, a 50 percent joint and survivor annuity pays one dollar
annually while both statuses are alive (usually husband and wife,
but not necessarily restricted to couples), and reduces to 50 cents
after the first death. Let x denote the age of the plan member, z
the joint annuitant's age, and k the portion of the annual benefit
paid to the survivor after the first death, regardless of who dies
first. The lOOk percent joint and survivor annuity may be repre-
sented as

kaxz = L v'Lplm),p1m) + k ,plm) (1 - ,p1m»

, =0
+ k ,p1m) (1 - ,plm»]. (3.19a)
The first term inside the brackets represents a payment of $1 if
both x and z are alive at time t, the second term represents a pay-
ment of $k if only x is alive, while the third term represents a
payment of $k if only z is alive.
A widely used variation of this annuity is known as a contin-
gent joint and survivor annuity. Under this form, the annuity
benefit is reduced only if the plan member is the first to die. The
48 Pension Mathematics

survivor's benefit might be any portion, with one-half and two-

thirds representing choices usually available. A lOOk percent
contingent joint and survivor annuity may be expressed as
~

\ixz = I V '[tP~m)tP~m) + tplm) (1 - tP~m»)

I 1::::0

+ k tP~m) (1 - tplm»)], (3.19b)

where the 1 under the x subscript stipulates that, if x is the first to
die, only k dollars are continued to z. The bracketed expression
represents a payment of $1 if both x and z are alive at time t, a
payment of $1 if x is alive and z is not alive, and a payment of $k
if z is alive and x is not alive. This expression reduces to
~

k
a"Xl -_ L'" v'[p(m)
/ x
+ k t p(m)
Z
_ k I p(m)p(m)]
x t l . (3.19c)
1 1=0

In this form, the bracketed term represents a payment of $1 to x

regardless of whether or not z is alive, a payment of $k to z re-
gardless of whether or not x is alive, and since this would result
in a total payment of $(1 + k) in the event both are alive in year ~
$k is subtracted if both are alive at time t.

Refund Annuities

Pension plans that require employee contributions frequently

provide death benefits in retirement equal to the excess, if any, of
the employee's accumulated contributions at retirement date
over the cumulative benefits received up to the time of death. If
the difference is paid in a lump sum, the annuity is termed a
modified cash refund annuity, whereas if the difference is paid by
continuing the benefit payments to a named beneficiary, the an-
nuity is termed a modified installment refund annuity.6
Let C r denote the employee's accumulated contributions at
retirement and n' denote such contributions per dollar of retire-
ment benefit (i.e., n' = Cr + B r ). The present value at age r of the
modified cash refund annuity may be written as

6Cash refund and installment annuities guarantee a return of the annuity's

purchase cost. The term modified is used in connection with pension plans to
indicate that the guarantee involves a return of accumulated employee contri-
butions.
3. Basic Actuarial Functions 49

MCR ar =I V'tpJ",l[l + vq~':\maxl(n' - t -1),0j]. (3.20)

I; 0

This formulation assumes that the lump sum payment, if any, will
be made at the end of the year of death.
The modified installment refund annuity is equal to the sum
of an n '-year annuity certain plus an n '-year deferred life annuity
(Le., a period certain life annuity, where the certain period is de-
termined by the ratio of employee contributions at retirement to
annual benefit payments):
(3.21)
Table 3-6 shows numerical values for the annuities discussed
up to this point under alternative interest rate, mortality rate, and
attained-age assumptions. Life expectancies at age 55, 65, and 70
are shown in Section I. A 25 percent change in the underlying
mortality rate is seen to affect this statistic by less than 25 per-
cent. Section II of Table 3-6 shows annuity certain values, with
the last row displaying an annuity certain for a period equal to an
age-65 individual's life expectancy. Note that this value is greater
than an age-65 life annuity shown in Section V. Figure
3-4 is useful in reasoning through this relationship. The area for
both annuity representations is, of course, equal. Since their re-
spective annuity values involve the product of each payment (or
expected payment in the case of the life annuity) times vi, for t
ranging from zero to 45 (110 - 65), the life annuity payment
stream involves much smaller values of vi beyond age 65 + e65
than the values of vi used with the annuity certain.
Section III of Table 3-6 shows the very substantial difference
between a last survivor annuity and a temporary life annuity,
both involving lO-year periods. Section IV illustrates the rela-
tively small impact of providing a 100 percent contingent benefit
versus a 50 or 75 percent contingency. Likewise, a lO-year differ-
ence in the age of the contingent annuitant has little impact on
the annuity value.
Section V shows life annuity values for ages 55, 65, and 70. A
25 percent change in the underlying mortality rate has only
slightly less impact than a 2 percentage point change in the inter-
est rate. Finally, the modified cash refund and modified install-
ment refund annuities are shown in Section VI of the table.
Since employee contributions at retirement are assumed to be
50 Pension Mathematics

equal to 5 times annual benefits, the installment refund annuity is

equal to a 5-year and age-65 last survivor annuity.

TABLE3~

Annuity Values Under Alternative Interest and Mortality Assumptions

Interest Rates: 60;0 8% 10%

Type of Mortality as %
Annuity of GAM-I971: 100% 75% 100% 125% 100%
e55 24.95 22.21 20.21
Life
1 e65 17.00 14.61 1291
Expectancy
e70 13.57 11.41 9.89

a51 4.47 4.31 4.17

aiOl 7.80 7.25 6.76

II.
Annuity
Certain
aiSl 10.29 9.24 8.37

ae;] 10.13 9.85 9.12 8.50 8.27

Last Survivor a 65 ,iOI 10.55 9.&J 9.34 8.98 8.36

1II. and Temporary
Life Annuities a65,iol 6.98 6.69 6.51 6.35 6.10
so ••
a65,60 11.09 10.22 9.65 9.17 8.53
7S ••
Contingent Joint a65,6Q 11.78 10.71 10.18 9.72 8.94
IV. and Survivor 100 ••
Annuity a65' 60 12.46 11.21 10.70 10.26 9.35
so .•
a65'55 11.46 10.43 9.89 9.44 8.70
so ••
a65,65 10.75 10.02 9.41 8.90 8.36
a55 12.24 10.90 10.45 10.06 9.10
v. Life Annuity a65 9.73 9.24 8.60 [Link] 7.71
070 8.35 8.23 7.52 6.95 6.84
Modified Cash and MCR··
a 65 9.93 9.39 8.&J 8.32 7.90
VI. Instaliment Refund
MlR··
Annuities a65 9.92 9.38 8.78 8.31 7.88

n' = Cr + B r = 5 for modified cash and installment refund annuities.

3. Basic ACllIarial Functions 51

FlGURE3-4
Ufe Annuity vs. Annuity Certain for Life Expectancy

Dollars
12 "T"""----------------------...,.. 12
1.0 ;,.;:::- -, AnnualooPayments 1.0
/ Under Qe;,

D.8 0.8

0.6 0.6
Annual Expected .•
/ Payments Under Q 6S
0.4 0.4

02 02

0.0 +-................,.....,........ .....,I""""I"'".....,.....:;;::;:::;:::;:""- 0.0

T'"'"""T""",.....!T'"""I'""'T"""'I-.-............
65 70 75 80 85 90 95 100
Age

Temporary Annuities

Temporary annuities, as will be shown in later chapters, are

required for some actuarial cost methods. In this case, consider-
ation is given to an employment-based annuity, subject to multi-
ple decrements, rather than tbe typical retirement-related annuity
involving only mortality. As noted earlier, the temporary annu-
ity bas x:rn as its subscript, indicating that the payments cease at
the end of n years or, if sooner, at the time the status x fails. In
addition, a superscript T is added to the annuity symbol to signify
a multiple-decrement environment. Equation (3.22) sets forth
the basic definition of this annuity:
n-I
.• T (T) ,
Q x :1il = I tPX V . (3.22)
I = 0

The value of n is often set at r - x in the context of pension math-

ematics' resulting in an annuity running from the participant's
attained age x up to, but not including, retirement age.
The withdrawal decrement may have a unique effect on this
annuity, as shown in Table 3-7. Generally one would anticipate
52 Pension Mathematics

the present value of an annuity running to age 65 to become

smaller at older ages. However, because of the withdrawal
decrement, the value of the annuity for most entry ages reaches a
maximum at some age in between the employee's entry age and
retirement age. This maximum for an age-20 entrant is 8.18 at
ages 41 and 42, with younger and older ages having smaller val-
ues. For example, at age 20, the 45-year annuity takes on a value
of only 4.00.
An important variation of ii;:~ is represented by' ii;:r-xl, the
superscript s denoting that the annuity is salary-based, as defined
by
r-1
T ----, = ,,!..L
'a x:r-XI P (T) V
oo

L t-x x
t-x
. (3.23)
t = x Sx

This formula represents the present value of an employee's future

salary from age x to age r, per unit of salary at age x.
Table 3-8 replicates Table 3-7 based on the annuity defined
in (3.23). This annuity, as one would expect, develops larger at-
tained age values than the unit-based annuity given in Table 3-7.
Whereas the unit-based annuity for an age-20 entrant reaches a
maximum value of 8.18 at age 41, the salary-based annuity reaches
a maximum of 14.35 at age 36. By age 64, however, both annuity
values are equal to unity. As was the case for the annuity values
shown in Table 3-7, the values given in Table 3-8 for the same at-
tained age, but for different entry ages, are affected to some ex-
tent by the select period of the termination assumption and the
elimination of termination rates after the first early retirement
qualification age.
Figure 3-5 provides a graphical comparison of the annuities
specified by equations (3.22) and (3.23) for the age-30 entrant.
This graph shows that these annuities can take on unique shapes
during an employee's career.
3. Basic Actuarial Functions 53

TABLE 3-7

Present Value of a Temporary Employment-Based Ufe

Annuity from Age.r to Age 65, ;;:.. _... 1

Entry Age, y
x 20 30 40 50 60
20 4.00
21 4.29
22 4.58
23 4.88
24 5.19
25 5.49
26 5.79
27 6.08
28 6.36
29 6.63
30 6.88 6J17
31 7.11 6.59
32 732 7.03
33 751 7.38
34 7.67 7.63
35 7.81 7.81
36 7.93 7.93
37 8.02 8.02
38 8.09 8.09
39 8.14 8.14
40 8.17 8.17 755
41 8.18 8.18 7.83
42 8.18 8.18 8.00
43 8.16 8.16 8.08
44 8.12 8.12 8.09
45 8.07 8.07 8.07
46 8.00 8.00 8.00
47 7.93 7.93 7.93
48 7.84 7.84 7.84
49 7.73 7.73 7.73
50 7.62 7.62 7.62 7.09
51 7.49 7.49 7.49 7.01
52 735 735 735 6.86
53 7.19 7.19 7.19 6.67
54 7.01 7.01 7.01 6.44
55 6.80 6.80 6.80 6.17
56 634 634 6.34 5.86
57 5.85 5.85 5.85 5.50
58 5.32 5.32 5.32 5.09
59 4.74 4.74 4.74 4.64
60 4.12 4.12 4.12 4.12 3.84
61 3.44 3.44 3.44 3.44 3.30
62 270 270 270 270 264
63 1.89 1.89 1.89 1.89 1.87
64 1.00 1.00 1.00 1.00 1.00
54 Pension Mathematics

TABLE 3-8
Present Value of a Temporary Employment-Based and

Salary-Based Life Annuity from Age x to Age 65, 'jj:.. -xl

Entry Age, y

x 20 30 40 50 60

20 6.68
21 7.39
22 8.13
23 8.87
24 9.61
25 10.33
26 11.01
27 11.65
28 12.24
29 12.75
30 13.20 11.42
31 13.58 12.47
32 13.88 13.27
33 14.11 13.84
34 14.25 14.18
35 14.34 14.34
36 14.35 14.35
37 14.31 14.31
38 14.22 14.22
39 14.07 14.07
40 13.89 13.89 1272
41 13.66 13.66 13.01
42 13.40 13.40 13.08
43 13.12 13.12 12.98
44 12.81 12.81 12.77
45 12.49 12.49 12.49
46 12.15 12.15 12.15
47 11.79 11.79 11.79
48 11.42 11.42 11.42
49 11.04 11.04 11.04
50 10.65 10.65 10.65 9.75
51 10.25 10.25 10.25 9.44
52 9.83 9.83 9.83 9.~
53 9.40 9.40 9.40 8.62
54 8.95 8.95 8.95 8.14
55 8.49 8.49 8.49 7.63
56 7.72 7.72 7.72 7.08
57 6.95 6.95 6.95 6.50
58 6.16 6.16 6.16 5.89
59 5.35 5.35 5.35 5.23
60 4.54 4.54 4.54 4.54 4.22
61 3.70 3.70 3.70 3.70 3.55
62 283 283 283 283 2.77
63 1.94 1.94 1.94 1.94 1.92
64 1.00 1.00 1.00 1.00 1.00
3. Basic Actuarial Functions 55

FIGURE 3-5
Unit·Based and Salary-B_d Temporary Annuity Values from Age ... to Age 65
Dollars
16 -r--------------------------r 16
14 14

12 12

10 10

8 8

6 6

4 4

2 2

0 a
30 35 40 45 50 55 60 65
Age