Financial Mathematics For Actuaries PDF

Financial Mathematics
for Actuaries
Chapter 1
Interest Accumulation and
Time Value of Money
1
Learning Objectives
1. Basic principles in calculation of interest accumulation
2. Simple and compound interest
3. Frequency of compounding
4. Effective rate of interest
5. Rate of discount
6. Present and future values of a single payment
2
1.1 Accumulation Function and Amount Function
• The sum of money borrowed is called the principal.
• The borrower compensates the lender by paying interest.
• At the end of the loan period the borrower pays the lender the
accumulated amount, which is equal to the sum of the principal
plus interest.
• We denote A(t) as the accumulated amount at time t, called the

amount function.
• A(0) is the initial principal and
I(t) = A(t) − A(t − 1) (1.1)
3
is the interest incurred from time t − 1 to time t, namely, in the tth
period.
• For the special case of an initial principal of 1 unit, we denote the

accumulated amount at time t by a(t), which is called the accumu-
lation function.
• If the initial principal is A(0) = k, then
A(t) = k × a(t).
4
1.2 Simple and Compound Interest
• While theoretically there are numerous ways of calculating the in-

terest, there are two methods which are commonly used in practice.
• These are the simple-interest method and the compound-interest

method.
• For the simple-interest method the interest earned over a period of

time is proportional to the length of the period.
• The interest incurred from time 0 to time t, for a principal of 1 unit,

is r × t, where r is the constant of proportion called the rate of
interest.
5
• Hence the accumulation function for the simple-interest method is
a(t) = 1 + rt, for t ≥ 0, (1.2)
and
A(t) = A(0)a(t) = A(0)(1 + rt), for t ≥ 0. (1.3)
• The most commonly used base is the year, in which case the term
annual rate of interest is used. We shall maintain this assump-
tion, unless stated otherwise.
Example 1.1: A person borrows $2,000 for 3 years at simple interest.

The rate of interest is 8% per annum. What are the interest charges for
years 1 and 2? What is the accumulated amount at the end of year 3?
6
Solution: The interest charges for years 1 and 2 are both equal to
2,000 × 0.08 = $160.
The accumulated amount at the end of year 3 is
2,000 (1 + 0.08 × 3) = $2,480.
• For the compound-interest method the accumulated amount over a

period of time is the principal for the next period.
• The accumulation function becomes
a(t) = (1 + r)t , for t = 0, 1, 2, · · · , (1.4)
and the amount function is
A(t) = A(0)a(t) = A(0)(1 + r)t , for t = 0, 1, 2, · · · . (1.5)
7
• Two remarks
— For the compound-interest method the accumulated amount at

the end of a year becomes the principal for the following year.
This is in contrast to the simple-interest method, for which the
principal remains unchanged through time.
— While (1.2) and (1.3) apply for t ≥ 0, (1.4) and (1.5) hold
only for integral t ≥ 0. There are alternative ways to define
the accumulation function for the compound-interest method
when t is not an integer.
Example 1.2: Solve the problem in Example 1.1 using the compound-
interest method.
8
Solution: The interest for year 1 is
2,000 × 0.08 = $160.
For year 2 the principal is
2,000 + 160 = $2,160,
so that the interest for the year is
2,160 × 0.08 = $172.80.
The accumulated amount at the end of year 3 is
2,000 (1 + 0.08)3 = $2,519.42.
9
• Compounding has the effect of generating a larger accumulated amount.
• When the interest rate is high, compounding the interest induces

the principal to grow much faster than the simple-interest method.
10
Table 1.1: Accumulated amount for a principal of $100
5% interest 10% interest

Year Simple Compound Simple Compound
interest ($) interest ($) interest ($) interest ($)
1 105.00 105.00 110.00 110.00

2 110.00 110.25 120.00 121.00
3 115.00 115.76 130.00 133.10
4 120.00 121.55 140.00 146.41
5 125.00 127.63 150.00 161.05
6 130.00 134.01 160.00 177.16
7 135.00 140.71 170.00 194.87
8 140.00 147.75 180.00 214.36
9 145.00 155.13 190.00 235.79
10 150.00 162.89 200.00 259.37
11
1.3 Frequency of Compounding
• Although the rate of interest is often quoted in annual term, the

interest accrued to an investment is often paid more frequently than
once a year.
• For example, a savings account may pay interest at 3% per year,

where the interest is credited monthly.
• The frequency of interest payment (also called the frequency of

compounding) makes an important difference to the accumulated
amount and the total interest earned. Thus, it is important to define
the rate of interest accurately.
• We use r(m) to denote the nominal rate of interest payable m

times a year. Thus, m is the frequency of compounding per year
12
and 1/m year is the compounding period or conversion period.
• Assume t (in years) is an integer multiple of 1/m, i.e., tm is an

integer representing the number of interest-conversion periods over
t years.
• The interest earned over the next 1/m year, i.e., the next interest-
conversion period, is
1 1 2
a(t) × r(m) × , for t = 0, , , · · · .
m m m
Thus,
µ ¶ " #
(m) (m)
1 a(0)r r r(m)
a = a(0) + = a(0) 1 + =1+ ,
m m m m
and in general
µ ¶ " #
(m)
1 r 1 2
a t+ = a(t) 1 + , for t = 0, , , · · · ,
m m m m
13
so that
" #
(m) mt
r 1 2
a(t) = 1 + , for t = 0, , ,···, (1.6)
m m m
and
" #
(m) mt
r 1 2
A(t) = A(0) 1 + , for t = 0, , , · · · . (1.7)
m m m
Example 1.3: A person deposits $1,000 into a savings account that

earns 3% interest payable monthly. How much interest will be credited in
the first month? What is the accumulated amount at the end of the first
month?
Solution: The rate of interest over one month is

1
0.03 × = 0.25%,
12
14
so that the interest earned over one month is
1,000 × 0.0025 = $2.50,
and the accumulated amount after one month is
1,000 + 2.50 = $1,002.50.
2
Example 1.4: $1,000 is deposited into a savings account that pays 3%
interest with monthly compounding. What is the accumulated amount
after two and a half years? What is the amount of interest earned over
this period?
Solution: The investment interval is 30 months. Thus, using (1.7), the

accumulated amount is
∙ ¸
0.03 30
1,000 1 + = $1,077.78.
12
15
The amount of interest earned over this period is
1,077.78 − 1, 000 = $77.78.
• When the loan period is not an integer multiple of the compounding

period, we may extend (1.6) and (1.7) to apply to any tm ≥ 0 (not
necessarily an integer).
• At the same nominal rate of interest, the more frequent the interest
is paid, the faster the accumulated amount grows. For example,
assuming the nominal rate of interest to be 5% and the principal
to be $1,000, the accumulated amounts after 1 year under several
different compounding frequencies are given in Table 1.2.
16
Table 1.2: Accumulated amount for a principal of $1,000
Frequency of Accumulated
interest payment m amount ($)
Yearly 1 1,050.00
Quarterly 4 1,050.95
Monthly 12 1,051.16
Daily 365 1,051.27
• When the compounding frequency m increases, the accumulated

amount tends to a limit.
• Let r̄ denote the nominal rate of interest for which compounding is

made over infinitely small intervals (i.e., m → ∞ so that r̄ = r(∞) ).
• We call this compounding scheme continuous compounding.
17
• For practical purposes, daily compounding is very close to continu-
ous compounding.
From the well-known limit theorem (see Appendix A.1) that

∙ ¸m
r̄
lim 1 + = er̄ (1.8)
m→∞ m
for any constant r̄, we conclude that, for continuous compounding, the
accumulation function (see (1.6)) is
∙ ¸mt ∙ µ ¶m ¸t
r̄ r̄
a(t) = m→∞
lim 1 + = lim 1+ = er̄t . (1.9)
m m→∞ m
• We call r̄ the continuously compounded rate of interest.
• Equation (1.9) provides the accumulation function of the continu-

ously compounding scheme at nominal rate of interest r̄.
18
1.4 Effective Rate of Interest
• The accumulated amount depends on the compounding frequency.

Comparing two investment schemes by just referring to their nomi-
nal rates of interest without taking into account their compounding
frequencies may be misleading.
• The measure called the effective rate of interest is often used to

compare investments with different periods of compounding.
• The annual effective rate of interest for year t, which we denote by

i(t), is the ratio of the amount of interest earned in a year, from time
t − 1 to time t, to the accumulated amount at the beginning of the
year (i.e., at time t − 1).
19
• It can be calculated by the following formula
I(t) A(t) − A(t − 1) a(t) − a(t − 1)
i(t) = = = . (1.10)
A(t − 1) A(t − 1) a(t − 1)
• For the simple-interest method, we have

(1 + rt) − (1 + r(t − 1)) r
i(t) = = ,
1 + r(t − 1) 1 + r(t − 1)
which decreases when t increases.
• For the compound-interest method with annual compounding (i.e.,

m = 1), we have (denoting r(1) = r)
(1 + r)t − (1 + r)t−1
i(t) = t−1
= r,
(1 + r)
which is the nominal rate of interest and does not vary with t.
20
• When m-compounding is used, the effective rate of interest is
" # " #
(m) tm (m) (t−1)m
r r
1+ − 1+ " #
(m) m
m m r
i(t) = " #(t−1)m = 1+ − 1, (1.11)
r(m) m
1+
m
which again does not vary with t.
• When m > 1, " #

(m) m
r
1+ − 1 > r(m) ,
m
so that the effective rate of interest is larger than the nominal rate
of interest.
• For continuous compounding, we have

exp(r̄t) − exp[r̄(t − 1)]
i(t) = = er̄ − 1, (1.12)
exp[r̄(t − 1)]
21
which again does not vary with t.
• As the effective rate of interest for the compound-interest method

does not vary with t, we shall simplify the notation and denote
i ≡ i(t).
Example 1.7: Consider two investment schemes A and B. Scheme A

offers 12% interest with annual compounding. Scheme B offers 11.5% in-
terest with monthly compounding. Calculate the effective rates of interest
of the two investments. Which scheme would you choose?
Solution: The effective rate of interest of Scheme A is equal to its

nominal rate of interest, i.e., 12%. The effective rate of interest of Scheme
B is ∙ ¸
0.115 12
1+ − 1 = 12.13%.
12
22
Although Scheme A has a higher nominal rate of interest, Scheme B offers
a higher effective rate of interest. Hence, while an investment of $100
in Scheme A will generate an interest of $12 after one year, a similar
investment in Scheme B will generate an interest of $12.13 over the same
period. Thus, Scheme B is preferred. 2
Another advantage of the effective rate of interest is that, for investments

that extend beyond one year the calculation of the accumulated amount
can be based on the effective rate without reference to the nominal rate.
Example 1.8: For the investment schemes in Example 1.7, calculate the
accumulated amount after 10 years on a principal of $1,000.
Solution: The accumulated amount after 10 years for Scheme A is
1,000 (1 + 0.12)10 = $3,105.85,
23
and that for Scheme B is
1,000 (1 + 0.1213)10 = $3,142.09.
2
Note that in the above example, the accumulated amount of Scheme B is
calculated without making use of its nominal rate.
• While i(t) defined in (1.10) is a 1-period effective rate, the concept

can be generalized to a n-period effective rate.
• We denote i(t, t + n) as the annual effective rate of interest in the

period t to t + n, where integral n > 1.
• The amount a(t) at time t compounded annually at the rate of i(t, t+

n) per year accumulates to a(t + n) at time t + n, where
a(t + n) = a(t)[1 + i(t, t + n)]n , (1.13)
24
which implies
" #1
a(t + n) n
i(t, t + n) = − 1. (1.14)
a(t)
• (1.10) can also be applied to intervals of less than one year. Suppose
∆t < 1, the effective rate of interest in the period t to t+∆t, denoted
by i(t, t + ∆t) is defined as
a(t + ∆t) − a(t)

i(t, t + ∆t) = , (1.15)
a(t)
which is an effective rate over a period of ∆t. Note that this rate is
not annualized.
25
1.5 Rates of Discount
• Some financial transactions have interest paid or deducted up-front.
• A popular way of raising a short-term loan is to sell a financial

security at a price less than the face value. Upon the maturity of
the loan, the face value is repaid. For example, a Treasury Bill is a
discount instrument.
• For a discount security, the shortfall between the sale price and the
face value is called the discount.
• The nominal principal (face value) has to be adjusted to take

account of the interest deducted, which affects the rate of interest,
as opposed to the rate of discount that is quoted.
26
• If the loan period is one year, the effective principal of the loan
after the interest is deducted is
Effective principal = Nominal principal × (1 − Rate of discount).
• The effective principal is A(0) and the nominal principal is A(1).
• If the quoted rate of discount is d, we have
A(0) = A(1)(1 − d), (1.16)
and
I(1) = A(1) − A(0) = A(1)d.
• The equivalent effective rate of interest i over the period of the dis-
count instrument is
A(1) − A(0) A(1)d d
i= = = , (1.17)
A(0) A(1)(1 − d) 1−d
27
from which we have
i
d= .
1+i
• Combining (1.4) and (1.17), we can see that
a(t) = (1 + i)t
" #t
d
= 1+
1−d
= (1 − d)−t ,
which is the accumulated value of 1 at time t at the rate of discount

d.
• When the loan period is less than 1 year, we should first calculate
the rate of interest over the period of the loan and then calculate
the effective rate of interest using the principle of compounding.
28
• Suppose the period of loan is 1/m year, we denote the nominal
rate of discount by d(m) .
• The nominal principal is A(1/m) and the interest deducted is

µ ¶ µ ¶
1 1 1
I =A × d(m) × ,
m m m
so that the effective principal is
µ ¶ µ ¶ µ ¶" (m)
#
1 1 1 d
A(0) = A −I =A 1− ,
m m m m
and the rate of interest charged over the (1/m)-year period as

µ ¶
1 d(m)
A − A(0) d(m)
m = m = .
A(0) (m) m−d (m)
d
1−
m
29
• Hence, the annualized equivalent nominal rate of interest is
(m) d(m) d(m)

r =m× = (m)
, (1.18)
m − d(m) d
1−
m
and the annual effective rate of interest is
" # " #m " #
(m) m (m) (m) −m
r d d
i= 1+ −1= 1+ −1= 1− − 1.
m m − d(m) m
(1.19)
• To compute the accumulation function a(t) for a discount instrument

with maturity of 1/m year (m > 1), we note
" #
(m) −mt
d 1 2
a(t) = (1 + i)t = 1 − , for t = 0, , , · · · . (1.20)
m m m
30
• From (1.19) we have
d(m) 1
−m
1− = (1 + i) , (1.21)
m
so that from (1.18) we conclude
1
(m)
r = (1 + i) d(m) .
m (1.22)
Example 1.11: The discount rate of a 3-month Treasury Bill is 6%

per annum. What is the annual effective rate of interest? What is the
accumulated value of 1 in 2 years?
Solution: The rate of interest charged for the 3-month period is

1
0.06 ×
4 = 1.52%.
1
1 − 0.06 ×
4
31
Therefore, the equivalent nominal rate of interest compounded quarterly
is
4 × 0.0152 = 6.08%,
and the annual effective rate of interest is
(1.0152)4 − 1 = 6.22%.
The accumulated value of 1 in 2 years is, using (1.20),

∙ ¸−8
0.06
a(2) = 1 − = 1.13.
4
Note that this can also be calculated as
(1 + i)2 = (1.0622)2 = 1.13.
32
1.6 Force of Interest
• If we divide i(t, t + ∆t) by ∆t, we obtain the rate of interest of the

investment per unit time in the interval (t, t + ∆t).
• The instantaneous rate, which is obtained when ∆t is infinitesimally

small, is
" #
i(t, t + ∆t) 1 a(t + ∆t) − a(t)
lim = lim
∆t→0 ∆t ∆t→0 ∆t a(t)
" #
1 a(t + ∆t) − a(t)
= lim
a(t) ∆t→0 ∆t
a0 (t)
= , (1.24)
a(t)
where a0 (t) is the derivative of a(t) with respect to t.
33
• Thus, we define
a0 (t)
δ(t) = , (1.25)
a(t)
which is called the force of interest.
• The force of interest is the instantaneous rate of increase of the accu-

mulated amount, a0 (t), as a percentage of the accumulated amount
at time t, a(t).
• Given a(t), the force of interest δ(t) can be computed using (1.25).
• Now we note that (1.25) can be written as

d ln a(t)
δ(t) = ,
dt
from which we have
Z t Z t
δ(s) ds = d ln a(s)
0 0
34
= ln a(s)]t0
= ln a(t) − ln a(0)
= ln a(t),
so that µZ t ¶
a(t) = exp δ(s) ds . (1.26)
0
• In the case when the force of interest is constant (not varying with
t), we denote δ(t) ≡ δ, and we have
a(t) = eδt . (1.27)
• Thus, if the force of interest is constant, it is equal to the continu-

ously compounded rate of interest, i.e., r̄ = δ.
• For the simple-interest method, we obtain

a0 (t) = r,
35
so that
r
δ(t) = , for t ≥ 0. (1.28)
1 + rt
• Hence, δ(t) decreases as t increases.
• For the compound-interest method, we have
a0 (t) = (1 + i)t ln(1 + i),
so that
(1 + i)t ln(1 + i)
δ(t) = t
= ln(1 + i). (1.29)
(1 + i)
• Thus, δ(t) does not vary with time and we write δ(t) ≡ δ so that
eδ = 1 + i.
36
Example 1.13: A fund accumulates at a simple-interest rate of 5%. An-
other fund accumulates at a compound-interest rate of 4%, compounded
yearly. When will the force of interest be the same for the two funds?
After this point, which fund will have a higher force of interest?
Solution: From (1.28), the force of interest of the simple-interest fund

at time t is
0.05
δ(t) = .
1 + 0.05t
From (1.29), the force of interest of the compound-interest fund is ln(1.04)
at any time. The two funds have the same force of interest when
0.05
= ln(1.04),
1 + 0.05t
i.e.,
0.05 − ln(1.04)
t= = 5.4967,
0.05 ln(1.04)
37
after which the force of interest of the simple-interest fund remains lower
than that of the compound-interest fund. 2
Example 1.14: If a fund accumulates at force of interest δ(t) = 0.02t,

find the annual effective rate of interest over 2 years and 5 years.
Solution: From (1.26), we have

µZ 2 ¶ ³ ´
a(2) = exp 0.02s ds = exp 0.01s2 ]20 = e0.04 .
0
We solve the equivalent annual effective rate of interest i from the equation
(compare this with (1.13))
(1 + i)2 = e0.04
to obtain
i = e0.02 − 1 = 2.02%.
38
Similarly, ³ ´
a(5) = exp 0.01s2 ]50 = e0.25 ,
so that
(1 + i)5 = e0.25
and
i = e0.05 − 1 = 5.13%.
2
39
1.7 Present and Future Values
• At the effective rate of interest i, a 1-unit investment today will

accumulate to (1 + i) units at the end of the year.
• The accumulated amount (1 + i) is also called the future value of

1 at the end of the year.
• Similarly, the future value of 1 at the end of year t is (1 + i)t .
• A (1/(1 + i))-unit payment invested today will accumulate to 1 unit

at the end of the year.
• Thus, 1/(1 + i) is called the present value of 1 to be paid at the

end of year 1.
• The present value of 1 to be paid at the end of year t is 1/(1 + i)t .
40
Example 1.15: Given i = 6%, calculate the present value of 1 to be
paid at (a) the end of year 1, (b) the end of year 5 and (c) 6.5 years.
Solution: (a) The present value of 1 to be paid at the end of year 1 is

1
= 0.9434.
1 + 0.06
The answers to (b) and (c) are, respectively,
1
= 0.7473
(1 + 0.06)5
and
1
= 0.6847.
(1 + 0.06)6.5
2
41
Example 1.16: An insurance agent offers a policy that pays a lump sum
of $50,000 five years later. If the rate of interest is 8%, how much would
you pay for the plan?
Solution: The fair amount to pay for this policy is the present value of
the lump sum, which is equal to
50,000
5
= $34,029.16.
(1.08)
2
Example 1.17: A person wants to accumulate $100,000 eight years from

today to sponsor his son’s education. If an investment plan offers him 8%
compounded monthly, what amount must he invest today?
Solution: We first calculate the effective rate of interest, which is equal
42
to ∙ ¸
0.08 12
1+ − 1 = 8.30%.
12
The amount required today is thus
100,000
8
= $52,841.16.
(1.083)
2
• We now denote
1
v= , (1.30)
1+i
which is the present value of 1 to be paid 1 year later.
• It is also called the discount factor.
• Combining (1.17) and (1.30), we obtain
d = iv, (1.31)
43
so that the present value of i is d.
• Also,
1 i
v+d= + = 1, (1.32)
1+i 1+i
which says that a unit payment at time 1 is the sum of its present
value and discount.
• Since a(1) = 1 + i, (1.30) implies

1
v= ,
a(1)
and, for a general time t, we denote v(t) as the present value of 1 to
be paid at time t.
• Then
1 1
v(t) = t
= , (1.33)
(1 + i) a(t)
44
which is the discount factor for payments at time t.
• For a general accumulation function a(·), the discount factor for

payments at time t is
1
v(t) = .
a(t)
Example 1.18: Find the sum of the present values of two payments
of $100 each to be paid at the end of year 4 and 9, if (a) interest is
compounded semiannually at the nominal rate of 8% per year, and (b)
the simple-interest method at 8% per year is used.
Solution: We first calculate the discount factors v(4) and v(9). For case
(a), the effective rate of interest is
(1.04)2 − 1 = 0.0816,
45
so that
1
v(4) = 4
= 0.7307
(1.0816)
and
1
v(9) = 9
= 0.4936.
(1.0816)
Hence, the present value of the two payments is
100(0.7307 + 0.4936) = $122.43.
For case (b), we have

1
v(4) = = 0.7576
1 + 0.08 × 4
and
1
v(9) = = 0.5814,
1 + 0.08 × 9
46
so that the present value of the two payments is
100(0.7576 + 0.5814) = $133.90.
• We now consider a payment of 1 at a future time τ . What is the

future value of this payment at time t > τ ?
• The answer to this question depends on how a new payment at a

future time accumulates with interest.
• Let us assume that any future payment starts to accumulate interest

following the same accumulation function as a payment made at time
0.
• As the 1-unit payment at time τ earns interest over a period of t − τ

until time t, its accumulated value at time t is a(t − τ ).
47
• However, if we consider a different scenario in which the 1-unit
amount at time τ has been accumulated from time 0 and is not
a new investment, what is the future value of this amount at time
t?
• To answer this question, we first determine the invested amount at

time 0, which is the present value of 1 due at time τ , i.e., 1/a(τ ).
The future value of this investment at time t is then given by
1 a(t)
× a(t) = .
a(τ ) a(τ )
• We can see that the future values of the two investments, i.e., a 1-unit
investment at time τ versus a 1-unit amount at time τ accumulated
from time 0, are not necessarily the same.
48
• However, they are equal if
a(t)
a(t − τ ) = (1.35)
a(τ )
for t > τ > 0.
• The compound-interest accumulation satisfies the condition (1.35).
• The principal at time τ accumulates interest at the rate of i per year,

whether the principal is invested at time τ or is accumulated from
the past. Specifically, we have
a(t) (1 + i)t t−τ

= τ
= (1 + i) = a(t − τ ).
a(τ ) (1 + i)
• In contrast, simple-interest accumulation does not satisfy condition

(1.35). The future value at time t of a unit payment at time τ is
49
a(t − τ ) = 1 + r(t − τ ). However, we have
a(t) 1 + rt 1 + rτ + r(t − τ ) r(t − τ )
= = = 1+ < 1+r(t−τ ) = a(t−τ ).
a(τ ) 1 + rτ 1 + rτ 1 + rτ
Example 1.19: Let a(t) = 0.02t2 + 1. Calculate the future value of an

investment at t = 5 consisting of a payment of 1 now and a payment of 2
at t = 3.
Solution: The future value at time 5 is
a(5) + 2 × a(2) = [0.02(5)2 + 1] + 2[0.02(2)2 + 1] = 3.66.
Example 1.20: Let δ(t) = 0.01t. Calculate the future value of an invest-
ment at t = 5 consisting of a payment of 1 now and a payment of 2 at
t = 2.
50
Solution: We first derive the accumulation function from the force of
interest, which is
µZ t ¶
a(t) = exp 0.01s ds = exp(0.005t2 ).
0
Thus, a(5) = exp(0.125) = 1.1331 and a(3) = exp(0.045) = 1.0460, from

which we obtain the future value of the investment as
a(5) + 2 × a(3) = 1.1331 + 2(1.0460) = 3.2251.
51
1.8 Equation of Value
• Consider a stream of cash flows occurring at different times. The

present value of the cash flows is equal to the sum of the present
values of each payment.
• Assume the payments are of values Cj occurring at time j = 0, 1, · · · , n.
• If the annual effective rate of interest is i with corresponding discount

factor v, the present value P of the cash flows is given by
n
X
P = Cj v j . (1.36)
j=0
• The future value F of the cash flows at time n is

⎛ ⎞
n
X n
X
F = (1 + i)n P = (1 + i)n ⎝ Cj v j ⎠ = Cj (1 + i)n−j . (1.37)
j=0 j=0
52
• We call these the equations of value.
Example 1.21: At the annual effective rate of interest i, when will an

initial principal be doubled?
Solution: This requires us to solve for n from the equation (note that
Cj = 0 for j > 0, and we let C0 = 1 and F = 2)
(1 + i)n = 2,
from which
ln(2)
n= .
ln(1 + i)
Thus, n is generally not an integer but can be solved exactly from the
above equation.
53
To obtain an approximate solution for n, we note that ln(2) = 0.6931 so
that
0.6931 i
n= × .
i ln(1 + i)
We approximate the last fraction in the above equation by taking i = 0.08
to obtain
0.6931 0.72
n' × 1.0395 = .
i i
Thus, n can be calculated approximately by dividing 0.72 by the effective
rate of interest. This is called the rule of 72. It provides a surprisingly
accurate approximation to n for a wide range of values of i. For example,
when i = 2%, the approximation gives n = 36 while the exact value is
35. When i = 14%, the approximate value is 5.14 while the exact value is
5.29. 2
Example 1.22: How long will it take for $100 to accumulate to $300 if
54
interest is compounded quarterly at the nominal rate of 6% per year?
Solution: Over one quarter, the interest rate is 1.5%. With C0 = $100,
Cj = 0 for j > 0, and F = $300, from (1.36) the equation of value is (n is
in quarters)
100(1.015)n = 300,
so that
ln(3)
n= = 73.79 quarters,
ln(1.015)
i.e., 18.45 years. 2
• The examples above concern only one payment. When multiple pay-
ments are involved, numerical methods may be required to calculate
n. However, analytical formulas are available when the payments
are level, and this will be discussed in the next chapter.
55
• A related problem is the solution of the rate of interest that will give
rise to a targeted present value or future value with corresponding
cash flows. The example below illustrates this point.
Example 1.24: A savings fund requires the investor to pay an equal

amount of installment each year for 3 years, with the first installment to
be paid immediately. At the end of the 3 years, a lump sum will be paid
back to the investor. If the effective interest rate is 5%, what is the amount
of the installment so that the investor can get back $10,000?
Solution: Let k be the installment. From (1.37), the equation of value

is
10,000 = k[(1.05)3 + (1.05)2 + 1.05] = 3.31k,
so that the installment is
10,000
k= = $3,021.03.
3.31
56
Table 1.3: Summary of accumulated value and present value formulas (all rates quoted per year)
Freq of Rate of Future value Present value

Accumulation conversion interest or of 1 at time t of 1 at time t Equations
method per year discount (in years): a(t) (in years): 1/a(t) in book Remarks
Compound 1 r (= i) (1 + r)t (1 + r)−t (1.4) also applies to

interest non-integer t > 0
h imt h i−mt
(m) r (m) r (m)
Compound m r 1+ m
1+ m
(1.6) m > 1 if compounding is
interest more frequent than annually
Compound ∞ r̄ er̄t e−r̄t (1.9) r̄ is the constant

interest force of interest δ
Compound 1 d (1 − d)−t (1 − d)t (1.4), (1.7)

discount
h i−mt h imt
d(m) d(m)
Compound m d(m) 1− m
1− m
(1.19) m > 1 for loan period
discount shorter than 1 year
Simple r 1 + rt (1 + rt)−1 (1.2)

interest
Simple d (1 − dt)−1 1 − dt (1.23)

discount
for Actuaries
Chapter 2
Annuities
Learning Objectives
1. Annuity-immediate and annuity-due
2. Present and future values of annuities
3. Perpetuities and deferred annuities
4. Other accumulation methods
5. Payment periods and compounding periods
6. Varying annuities
2
2.1 Annuity-Immediate
• Consider an annuity with payments of 1 unit each, made at the end

of every year for n years.
• This kind of annuity is called an annuity-immediate (also called

an ordinary annuity or an annuity in arrears).
• The present value of an annuity is the sum of the present values

of each payment.
Example 2.1: Calculate the present value of an annuity-immediate of

amount $100 paid annually for 5 years at the rate of interest of 9%.
Solution: Table 2.1 summarizes the present values of the payments as

well as their total.
3
Table 2.1: Present value of annuity
Year Payment ($) Present value ($)
1 100 100 (1.09)−1 = 91.74

2 100 100 (1.09)−2 = 84.17
3 100 100 (1.09)−3 = 77.22
4 100 100 (1.09)−4 = 70.84
5 100 100 (1.09)−5 = 64.99
Total 388.97
• We are interested in the value of the annuity at time 0, called the

present value, and the accumulated value of the annuity at time n,
called the future value.
4
• Suppose the rate of interest per period is i, and we assume the
compound-interest method applies.
• Let anei denote the present value of the annuity, which is sometimes
denoted as ane when the rate of interest is understood.
• As the present value of the jth payment is v j , where v = 1/(1 + i) is

the discount factor, the present value of the annuity is (see Appendix
A.5 for the sum of a geometric progression)
ane = v + v 2 + v 3 + · · · + v n
∙ ¸
1 − vn
= v×
1−v
1 − vn
=
i
1 − (1 + i)−n
= . (2.1)
i
5
• The accumulated value of the annuity at time n is denoted by snei
or sne .
• This is the future value of ane at time n. Thus, we have
sne = ane × (1 + i)n

(1 + i)n − 1
= . (2.2)
i
• If the annuity is of level payments of P , the present and future values

of the annuity are P ane and P sne , respectively.

amount $100 paid annually for 5 years at the rate of interest of 9% using
formula (2.1). Also calculate its future value at time 5.
6
Solution: From (2.1), the present value of the annuity is
" #
−5
1 − (1.09)
100 a5e = 100 × = $388.97,
0.09
which agrees with the solution of Example 2.1. The future value of the
annuity is
(1.09)5 × (100 a5e ) = (1.09)5 × 388.97 = $598.47.
Alternatively, the future value can be calculated as

" #
5
(1.09) − 1
100 s5e = 100 × = $598.47.
0.09
2
amount $100 payable quarterly for 10 years at the annual rate of interest
7
of 8% convertible quarterly. Also calculate its future value at the end of
10 years.
Solution: Note that the rate of interest per payment period (quarter)
is (8/4)% = 2%, and there are 4 × 10 = 40 payments. Thus, from (2.1)
the present value of the annuity-immediate is
" #
−40
1 − (1.02)
100 a40e0.02 = 100 × = $2,735.55,
0.02
and the future value of the annuity-immediate is
2735.55 × (1.02)40 = $6,040.20.
• A common problem in financial management is to determine the in-

stallments required to pay back a loan. We may use (2.1) to calculate
the amount of level installments required.
8
Example 2.4: A man borrows a loan of $20,000 to purchase a car at
annual rate of interest of 6%. He will pay back the loan through monthly
installments over 5 years, with the first installment to be made one month
after the release of the loan. What is the monthly installment he needs to
pay?
Solution: The rate of interest per payment period is (6/12)% = 0.5%.

Let P be the monthly installment. As there are 5 × 12 = 60 payments,
from (2.1) we have
20,000 = P a60e0.005
" #
−60
1 − (1.005)
= P×
0.005
= P × 51.7256,
9
so that
20,000
P = = $386.66.
51.7256
2
• The example below illustrates the calculation of the required install-

ment for a targeted future value.
Example 2.5: A man wants to save $100,000 to pay for his son’s
education in 10 years’ time. An education fund requires the investors to
deposit equal installments annually at the end of each year. If interest of
7.5% is paid, how much does the man need to save each year in order to
meet his target?
Solution: We first calculate s10e , which is equal to

(1.075)10 − 1
= 14.1471.
0.075
10
Then the required amount of installment is
100,000 100,000
P = = = $7,068.59.
s10e 14.1471
2
11
2.2 Annuity-Due
• An annuity-due is an annuity for which the payments are made at

the beginning of the payment periods
• The first payment is made at time 0, and the last payment is made
at time n − 1.
• We denote the present value of the annuity-due at time 0 by änei (or

äne ), and the future value of the annuity at time n by s̈nei (or s̈ne ).
• The formula for äne can be derived as follows

äne = 1 + v + · · · + v n−1
1 − vn
=
1−v
1 − vn
= . (2.3)
d
12
• Also, we have
s̈ne = äne × (1 + i)n

(1 + i)n − 1
= . (2.4)
d
• As each payment in an annuity-due is paid one period ahead of the

corresponding payment of an annuity-immediate, the present value
of each payment in an annuity-due is (1+i) times of the present value
of the corresponding payment in an annuity-immediate. Hence,
äne = (1 + i) ane (2.5)
and, similarly,
s̈ne = (1 + i) sne . (2.6)
13
• As an annuity-due of n payments consists of a payment at time 0
and an annuity-immediate of n − 1 payments, the first payment of
which is to be made at time 1, we have
äne = 1 + an−1e . (2.7)
• Similarly, if we consider an annuity-immediate with n + 1 payments

at time 1, 2, · · ·, n + 1 as an annuity-due of n payments starting at
time 1 plus a final payment at time n + 1, we can conclude
sn+1e = s̈ne + 1. (2.8)
Example 2.6: A company wants to provide a retirement plan for an

employee who is aged 55 now. The plan will provide her with an annuity-
immediate of $7,000 every year for 15 years upon her retirement at the
14
age of 65. The company is funding this plan with an annuity-due of 10
years. If the rate of interest is 5%, what is the amount of installment the
company should pay?
Solution: We first calculate the present value of the retirement annuity.

This is equal to
" #
−15
1 − (1.05)
7,000 a15e = 7,000 × = $72,657.61.
0.05
This amount should be equal to the future value of the company’s install-
ments P , which is P s̈10e . Now from (2.4), we have
(1.05)10 − 1
s̈10e = −1
= 13.2068,
1 − (1.05)
so that
72,657.61
P = = $5,501.53.
13.2068
15
2.3 Perpetuity, Deferred Annuity and Annuity
Values at Other Times
• A perpetuity is an annuity with no termination date, i.e., n → ∞.
• An example that resembles a perpetuity is the dividends of a pre-

ferred stock.
• To calculate the present value of a perpetuity, we note that, as v < 1,

v n → 0 as n → ∞. Thus, from (2.1), we have
1
a∞e = . (2.9)
i
• For the case when the first payment is made immediately, we have,
from (2.3),
1
ä∞e = . (2.10)
d
16
• A deferred annuity is one for which the first payment starts some
time in the future.
• Consider an annuity with n unit payments for which the first pay-
ment is due at time m + 1.
• This can be regarded as an n-period annuity-immediate to start at

time m, and its present value is denoted by m| anei (or m| ane for short).
Thus, we have
m| ane = v m ane
∙ ¸
m 1 − vn
= v ×
i
v m − v m+n
=
i
(1 − v m+n ) − (1 − v m )
=
i
17
= am+ne − ame . (2.11)
• To understand the above equation, see Figure 2.3.
• From (2.11), we have
am+ne = ame + v m ane

= ane + v n ame . (2.12)
• Multiplying the above equations throughout by 1 + i, we have
äm+ne = äme + v m äne

= äne + v n äme . (2.13)
• We also denote v m äne as m| äne , which is the present value of a n-

payment annuity of unit amounts due at time m, m+1, · · · , m+n−1.
18
• If we multiply the equations in (2.12) throughout by (1 + i)m+n , we
obtain
sm+ne = (1 + i)n sme + sne

= (1 + i)m sne + sme . (2.14)
• See Figure 2.4 for illustration.
• It is also straightforward to see that
s̈m+ne = (1 + i)n s̈me + s̈ne

= (1 + i)m s̈ne + s̈me . (2.15)
• We now return to (2.2) and write it as
sm+ne = (1 + i)m+n am+ne ,
19
so that
v m sm+ne = (1 + i)n am+ne , (2.16)
for arbitrary positive integers m and n.
• How do you interpret this equation?
20
2.4 Annuities Under Other Accumulation Methods
• We have so far discussed the calculations of the present and future

values of annuities assuming compound interest.
• We now extend our discussion to other interest-accumulation meth-

ods.
• We consider a general accumulation function a(·) and assume that

the function applies to any cash-flow transactions in the future.
• As stated in Section 1.7, any payment at time t > 0 starts to accu-

mulate interest according to a(·) as a payment made at time 0.
• Given the accumulation function a(·), the present value of a unit

payment due at time t is 1/a(t), so that the present value of a n-
21
period annuity-immediate of unit payments is
n
X 1
ane = . (2.17)
t=1 a(t)
• The future value at time n of a unit payment at time t < n is

a(n − t), so that the future value of a n-period annuity-immediate
of unit payments is
n
X
sne = a(n − t). (2.18)
t=1
• If (1.35) is satisfied so that a(n − t) = a(n)/a(t) for n > t > 0, then

n
X n
X
a(n) 1
sne = = a(n) = a(n) ane . (2.19)
t=1 a(t) t=1 a(t)
• This result is satisfied for the compound-interest method, but not

the simple-interest method or other accumulation schemes for which
equation (1.35) does not hold.
22
Example 2.7: Suppose δ(t) = 0.02t for 0 ≤ t ≤ 5, find a5e and s5e .
Solution: We first calculate a(t), which, from (1.26), is

µZ t ¶
a(t) = exp 0.02s ds
0
= exp(0.01t2 ).
Hence, from (2.17),

1 1 1 1 1
a5e = + + + + = 4.4957,
e0.01 e0.04 e0.09 e0.16 e0.25
and, from (2.18),
s5e = 1 + e0.01 + e0.04 + e0.09 + e0.16 = 5.3185.
Note that a(5) = e0.25 = 1.2840, so that
a(5) a5e = 1.2840 × 4.4957 = 5.7724 6= s5e .
23
2
• Note that in the above example, a(n − t) = exp[0.01(n − t)2 ] and

a(n)
= exp[0.01(n2 − t2 )],
a(t)
so that a(n − t) 6= a(n)/a(t) and (2.19) does not hold.
Example 2.8: Calculate a3e and s3e if the nominal rate of interest is
5% per annum, assuming (a) compound interest, and (b) simple interest.
Solution: (a) Assuming compound interest, we have

1 − (1.05)−3
a3e = = 2.723,
0.05
and
s3e = (1.05)3 × 2.72 = 3.153.
24
(b) For simple interest, the present value is
3
X 3
X
1 1 1 1 1
a3e = = = + + = 2.731,
t=1 a(t) t=1 1 + rt 1.05 1.1 1.15
and the future value at time 3 is

3
X 3
X
s3e = a(3 − t) = (1 + r(3 − t)) = 1.10 + 1.05 + 1.0 = 3.150.
t=1 t=1
At the same nominal rate of interest, the compound-interest method gen-

erates higher interest than the simple-interest method. Therefore, the
future value under the compound-interest method is higher, while its
present value is lower. Also, note that for the simple-interest method,
a(3) a3e = 1.15 × 2.731 = 3.141, which is different from s3e = 3.150. 2
25
2.5 Payment Periods, Compounding Periods and
Continuous Annuities
• We now consider the case where the payment period differs from the
interest-conversion period.
Example 2.9: Find the present value of an annuity-due of $200 per

quarter for 2 years, if interest is compounded monthly at the nominal rate
of 8%.
Solution: This is the situation where the payments are made less
frequently than interest is converted. We first calculate the effective rate
of interest per quarter, which is
∙ ¸3
0.08
1+ − 1 = 2.01%.
12
26
As there are n = 8 payments, the required present value is
" #
−8
1 − (1.0201)
200 ä8e0.0201 = 200 × −1
= $1,493.90.
1 − (1.0201)
2
Example 2.10: Find the present value of an annuity-immediate of
$100 per quarter for 4 years, if interest is compounded semiannually at
the nominal rate of 6%.
Solution: This is the situation where payments are made more fre-
quently than interest is converted. We first calculate the effective rate of
interest per quarter, which is
∙ ¸1
0.06 2
1+ − 1 = 1.49%.
2
27
Thus, the required present value is
" #
−16
1 − (1.0149)
100 a16e0.0149 = 100 × = $1,414.27.
0.0149
2
• It is possible to derive algebraic formulas to compute the present

and future values of annuities for which the period of installment is
different from the period of compounding.
• We first consider the case where payments are made less frequently
than interest conversion, which occurs at time 1, 2, · · ·, etc.
• Let i denote the effective rate of interest per interest-conversion pe-

riod. Suppose a m-payment annuity-immediate consists of unit pay-
ments at time k, 2k, · · ·, mk. We denote n = mk, which is the
number of interest-conversion periods for the annuity.
28
• Figure 2.6 illustrates the cash flows for the case of k = 2.
• The present value of the above annuity-immediate is (we let w = v k )
v k + v 2k + · · · + v mk = w + w ∙
2
+ · · · +
¸
w m
1 − wm
= w×
1−w
∙ n¸
1 − v
= vk ×
1 − vk
1 − vn
=
(1 + i)k − 1
ane
= , (2.20)
ske
and the future value of the annuity is

ane
n sne
(1 + i) = . (2.21)
ske ske
29
• We now consider the case where the payments are made more fre-
quently than interest conversion.
• Let there be mn payments for an annuity-immediate occurring at

time 1/m, 2/m, · · ·, 1, 1+1/m, · · ·, 2, · · ·, n, and let i be the effective
rate of interest per interest-conversion period. Thus, there are mn
payments over n interest-conversion periods.
• Suppose each payment is of the amount 1/m, so that there is a

nominal amount of unit payment in each interest-conversion period.
• Figure 2.7 illustrates the cash flows for the case of m = 4.

(m)
• We denote the present value of this annuity at time 0 by anei , which
1
can be computed as follows (we let w = v )m
(m) 1 ³ 1 2
1+ 1
n
´
anei = vm + vm + ··· + v + v m + ··· + v
m
30
1
= (w + w2 + · · · + wmn )
m∙ ¸
1 1 − wmn
= w×
m" 1−w #
1 1 1 − vn
= vm × 1
m 1 − vm
" #
n
1 1−v
=
m (1 + i) m1 − 1
1 − vn
= , (2.22)
r(m)
where h i
1
(m)
r = m (1 + i) m −1 (2.23)
is the equivalent nominal rate of interest compounded m times per
interest-conversion period (see (1.19)).
31
• The future value of the annuity-immediate is
(m) (m)
snei = (1 + i)n anei
(1 + i)n − 1
=
r(m)
i
= (m) sne . (2.24)
r
• The above equation parallels (2.22), which can also be written as
(m) i
anei = ane .
r(m)
• If the mn-payment annuity is due at time 0, 1/m, 2/m, · · · , n − 1/m,

(m)
we denote its present value at time 0 by äne , which is given by
∙ ¸
(m) 1 (m) 1 1 − vn
äne = (1 + i) ane = (1 + i)
m m × (m)
. (2.25)
r
32
• Thus, from (1.22) we conclude
(m) 1 − vn d
äne = (m) = (m) äne . (2.26)
d d
• The future value of this annuity at time n is
(m) n (m) d
s̈ne = (1 + i) äne = s̈ne . (2.27)
d(m)
• For deferred annuities, the following results apply

(m) (m)
q| ane = v q ane , (2.28)
and
(m) q (m)
q| äne = v äne . (2.29)
Example 2.11: Solve the problem in Example 2.9 using (2.20).
33
Solution: We first note that i = 0.08/12 = 0.0067. Now k = 3 and
n = 24 so that from (2.20), the present value of the annuity-immediate is
" #
a24e0.0067 −24
1 − (1.0067)
200 × = 200 ×
s3e0.0067 (1.0067)3 − 1
= $1,464.27.
Finally, the present value of the annuity-due is
(1.0067)3 × 1,464.27 = $1,493.90.
2
Example 2.12: Solve the problem in Example 2.10 using (2.22) and
(2.23).
Solution: Note that m = 2 and n = 8. With i = 0.03, we have, from
34
(2.23)
(2)
√
r = 2 × [ 1.03 − 1] = 0.0298.
Therefore, from (2.22), we have
(2) 1 − (1.03)−8
=
a8e0.03 = 7.0720.
0.0298
As the total payment in each interest-conversion period is $200, the re-
quired present value is
200 × 7.0720 = $1,414.27.
• Now we consider (2.22) again. Suppose the annuities are paid con-
tinuously at the rate of 1 unit per interest-conversion period over
n periods. Thus, m → ∞ and we denote the present value of this
continuous annuity by āne .
35
• As limm→∞ r(m) = δ, we have, from (2.22),
1 − vn 1 − vn i
āne = = = ane . (2.30)
δ ln(1 + i) δ
• The present value of a n-period continuous annuity of unit payment

per period with a deferred period of q is given by
q| āne = v q āne = āq+ne − āqe . (2.31)
• To compute the future value of a continuous annuity of unit payment

per period over n periods, we use the following formula
(1 + i)n − 1 i
s̄ne = (1 + i)n āne = = sne . (2.32)
ln(1 + i) δ
• We now generalize the above results to the case of a general accu-

mulation function a(·). The present value of a continuous annuity
36
of unit payment per period over n periods is
Z n Z n µ Z t ¶
āne = v(t) dt = exp − δ(s) ds dt. (2.33)
0 0 0
• To compute the future value of the annuity at time n, we assume

that, as in Section 1.7, a unit payment at time t accumulates to
a(n − t) at time n, for n > t ≥ 0.
• Thus, the future value of the annuity at time n is

Z n Z n µZ n−t ¶
s̄ne = a(n − t) dt = exp δ(s) ds dt. (2.34)
0 0 0
37
2.6 Varying Annuities
• We consider annuities the payments of which vary according to an

arithmetic progression.
• Thus, we consider an annuity-immediate and assume the initial pay-

ment is P , with subsequent payments P + D, P + 2D, · · ·, etc., so
that the jth payment is P + (j − 1)D.
• We allow D to be negative so that the annuity can be either stepping

up or stepping down.
• However, for a n-payment annuity, P + (n − 1)D must be positive

so that negative cash flow is ruled out.
• We can see that the annuity can be regarded as the sum of the
following annuities: (a) a n-period annuity-immediate with constant
38
amount P , and (b) n − 1 deferred annuities, where the jth deferred
annuity is a (n − j)-period annuity-immediate with level amount D
to start at time j, for j = 1, · · · , n − 1.
• Thus, the present value of the varying annuity is

⎡ ⎤ ⎡ ⎤
n−1
X n−1
X n−j
⎣ j ⎦ ⎣
(1 − v
j )⎦
P ane + D v an−j e = P ane + D v
j=1 j=1 i
⎡ ³P ´ ⎤
n−1 j n
j=1 v − (n − 1)v
= P ane + D ⎣ ⎦
i
⎡ ³P ´ ⎤
n j n
j=1 v − nv
= P ane + D ⎣ ⎦
i
" n
#
ane − nv
= P ane + D . (2.35)
i
39
• For a n-period increasing annuity with P = D = 1, we denote its
present and future values by (Ia)ne and (Is)ne , respectively.
• It can be shown that

äne − nv n
(Ia)ne = (2.36)
i
and
sn+1e − (n + 1) s̈ne − n
(Is)ne = = . (2.37)
i i
• For an increasing n-payment annuity-due with payments of 1, 2, · · · , n
at time 0, 1, · · · , n − 1, the present value of the annuity is
(Iä)ne = (1 + i)(Ia)ne . (2.38)
• This is the sum of a n-period level annuity-due of unit payments and

a (n − 1)-payment increasing annuity-immediate with starting and
incremental payments of 1.
40
• Thus, we have
(Iä)ne = äne + (Ia)n−1e . (2.39)
• For the case of a n-period decreasing annuity with P = n and D =

−1, we denote its present and future values by (Da)ne and (Ds)ne ,
respectively.
• Figure 2.9 presents the time diagram of this annuity.
• It can be shown that

n − ane
(Da)ne = (2.40)
i
and
n(1 + i)n − sne
(Ds)ne = . (2.41)
i
41
• We consider two types of increasing continuous annuities. First,
we consider the case of a continuous n-period annuity with level
payment (i.e., at a constant rate) of τ units from time τ − 1 through
time τ .
• We denote the present value of this annuity by (Iā)ne , which is given

by
n
X Z τ n
X Z τ
(Iā)ne = τ v s ds = τ e−δs ds. (2.42)
τ =1 τ −1 τ =1 τ −1
• The above equation can be simplified to

äne − nv n
(Iā)ne = . (2.43)
δ
• Second, we may consider a continuous n-period annuity for which
the payment in the interval t to t + ∆t is t∆t, i.e., the instantaneous
rate of payment at time t is t.
42
¯ e , which is given
• We denote the present value of this annuity by (Iā)n
by Z n Z n n
¯ e= ā e − nv
(Iā)n tv t dt = te−δt dt = n . (2.44)
0 0 δ
• We now consider an annuity-immediate with payments following a
geometric progression.
• Let the first payment be 1, with subsequent payments being 1 + k

times the previous one. Thus, the present value of an annuity with
n payments is (for k 6= i)
n−1
X
2 n n−1
v + v (1 + k) + · · · + v (1 + k) = v [v(1 + k)]t
t=0
n−1
" #t
X 1+k
= v
t=0 1+i
43
⎡ Ã !n ⎤
1+k
⎢1 − ⎥
⎢ 1+i ⎥
⎢
= v⎢ ⎥
1+k ⎥
⎣ ⎦
1−
1+i
Ã !n
1+k
1−
1+i
= . (2.45)
i−k
Example 2.13: An annuity-immediate consists of a first payment of

$100, with subsequent payments increased by 10% over the previous one
until the 10th payment, after which subsequent payments decreases by 5%
over the previous one. If the effective rate of interest is 10% per payment
period, what is the present value of this annuity with 20 payments?
Solution: The present value of the first 10 payments is (use the second
44
line of (2.45))
100 × 10(1.1)−1 = $909.09.
For the next 10 payments, k = −0.05 and their present value at time 10
is µ ¶
0.95 10
1−
100(1.10)9 (0.95) × 1.1 = 1,148.64.
0.1 + 0.05
Hence, the present value of the 20 payments is
909.09 + 1,148.64(1.10)−10 = $1,351.94.
Example 2.14: An investor wishes to accumulate $1,000 at the end of

year 5. He makes level deposits at the beginning of each year for 5 years.
The deposits earn a 6% annual effective rate of interest, which is credited
45
at the end of each year. The interests on the deposits earn 5% effective
interest rate annually. How much does he have to deposit each year?
Solution: Let the level annual deposit be A. The interest received at

the end of year 1 is 0.06A, which increases by 0.06A annually to 5 × 0.06A
at the end of year 5. Thus, the interest is a 5-payment increasing annuity
with P = D = 0.06A, earning annual interest of 5%. Hence, we have the
equation
1,000 = 5A + 0.06A(Is)5e0.05 .
From (2.37) we obtain (Is)5e0.05 = 16.0383, so that
1,000
A= = $167.7206.
5 + 0.06 × 16.0383
2
46
2.7 Term of Annuity
• We now consider the case where the annuity period may not be an
integer. We consider an+ke , where n is an integer and 0 < k < 1.
We note that
1 − v n+k
an+ke =
i ³ ´
(1 − v n ) + v n − v n+k
=
"i #
k
n+k (1 + i) − 1
= ane + v
i
= ane + v n+k ske . (2.46)
• Thus, an+ke is the sum of the present value of a n-period annuity-

immediate with unit amount and the present value of an amount ske
47
paid at time n + k.
Example 2.15: A principal of $5,000 generates income of $500 at the

end of every year at an effective rate of interest of 4.5% for as long as
possible. Calculate the term of the annuity and discuss the possibilities of
settling the last payment.
Solution: The equation of value
500 ane0.045 = 5,000
implies ane0.045 = 10. As a13e0.045 = 9.68 and a14e0.045 = 10.22, the principal
can generate 13 regular payments. The investment may be paid off with
an additional amount A at the end of year 13, in which case
500 s13e0.045 + A = 5,000 (1.045)13 ,
48
which implies A = $281.02, so that the last payment is $781.02. Alter-
natively, the last payment B may be made at the end of year 14, which
is
B = 281.02 × 1.045 = $293.67.
If we adopt the approach in (2.46), we solve k from the equation
1 − (1.045)−(13+k)
= 10,
0.045
which implies
13+k 1
(1.045) = ,
0.55
from which we obtain
µ ¶
1
ln
k= 0.55 − 13 = 0.58.
ln(1.045)
49
Hence, the last payment C to be paid at time 13.58 years is
" #
0.58
(1.045) −1
C = 500 × = $288.32.
0.045
Note that A < C < B, which is as expected, as this follows the order
of the occurrence of the payments. In principle, all three approaches are
justified. 2
• Generally, the effective rate of interest cannot be solved analytically

from the equation of value. Numerical methods must be used for
this purpose.
Example 2.16: A principal of $5,000 generates income of $500 at the

end of every year for 15 years. What is the effective rate of interest?
50
Solution: The equation of value is
5,000
a15ei = = 10,
500
so that
1 − (1 + i)−15
a15ei = = 10.
i
A simple grid search provides the following results
i a15ei
0.054 10.10
0.055 10.04
0.056 9.97
A finer search provides the answer 5.556%. 2
• The Excel Solver may be used to calculate the effective rate of in-
terest in Example 2.16.
51
• The computation is illustrated in Exhibit 2.1.
• We enter a guessed value of 0.05 in Cell A1 in the Excel worksheet.
• The following expression is then entered in Cell A2:

(1 − (1 + A1)ˆ(−15))/A1, which computes a15ei with i equal to the
value at A1.
• We can also use the Excel function RATE to calculate the rate of
interest that equates the present value of an annuity-immediate to
a given value. Specifically, consider the equations
anei − A = 0 and änei − A = 0.
Given n and A, we wish to solve for i, which is the rate of interest

per payment period of the annuity-immediate or annuity-due. The
use of the Excel function RATE to compute i is described as follows:
52
Exhibit 2.1: Use of Excel Solver for Example 2.16
Excel function: RATE(np,1,pv,type,guess)
np = n,
pv = −A,
type = 0 (or omitted) for annuity-immediate, 1 for annuity-due
guess = starting value, set to 0.1 if omitted
Output = i, rate of interest per payment period of the annuity
• To use RATE to solve for Example 2.16 we key in the following:

“=RATE(15,1,-10)”.
53
for Actuaries
Chapter 3
Spot Rates, Forward Rates and
the Term Structure
1
Learning Objectives
1. Spot rate of interest
2. Forward rate of interest
3. Yield curve
4. Term structure of interest rates
2
3.1 Spot and Forward Rates of Interest
• We now allow the rate of interest to vary with the duration of the
investment.
• We consider the case where investments over different horizons earn

different rates of interest, although the principle of compounding
still applies.
• We consider two notions of interest rates, namely, the spot rate of

interest and the forward rate of interest.
• Consider an investment at time 0 earning interest over t periods. We

assume that the period of investment is fixed at the time of invest-
ment, but the rate of interest earned per period varies according to
the investment horizon.
3
• Thus, we define iSt as the spot rate of interest, which is the annualized
effective rate of interest for the period from time 0 to t.
• The subscript t in iSt highlights that the annual rate of interest varies
with the investment horizon.
• Hence, a unit payment at time 0 accumulates to
a(t) = (1 + iSt )t (3.1)
at time t.
• The present value of a unit payment due at time t is

1 1
= S t
. (3.2)
a(t) (1 + it )
• We define iFt as the rate of interest applicable to the period t − 1 to

t, called the forward rate of interest.
4
• This rate is determined at time 0, although the payment is due at
time t − 1 (thus, the use of the term forward).
• By convention, we have iS1 ≡ iF1 . However, iSt and iFt are generally
different for t = 2, 3, · · ·.
• See Figure 3.1.
• A plot of iSt against t is called the yield curve, and the mathemat-
ical relationship between iSt and t is called the term structure of
interest rates.
• The spot and forward rates are not free to vary independently of
each other.
• Consider the case of t = 2. If an investor invests a unit amount at

time 0 over 2 periods, the investment will accumulate to (1 + iS2 )2 at
5
time 2.
• Alternatively, she can invest a unit payment at time 0 over 1 period,

and enters into a forward agreement to invest 1 + iS1 unit at time 1
to earn the forward rate of iF2 for 1 period.
• This rollover strategy will accumulate to (1 + iS1 )(1 + iF2 ) at time 2.

The two strategies will accumulate to the same amount at time 2,
so that
(1 + iS2 )2 = (1 + iS1 )(1 + iF2 ), (3.3)
if the capital market is perfectly competitive, so that no arbitrage
opportunities exist.
• Equation (3.3) can be generalized to the following relationship con-
6
cerning spot and forward rates of interest
(1 + iSt )t = (1 + iSt−1 )t−1 (1 + iFt ), (3.4)
for t = 2, 3, · · · .
• We can also conclude that

(1 + iSt )t = (1 + iF1 )(1 + iF2 ) · · · (1 + iFt ). (3.5)
• Given iSt , the forward rates of interest iFt satisfying equations (3.4)
and (3.5) are called the implicit forward rates.
• The quoted forward rates in the market may differ from the implicit
forward rates in practice, as when the market is noncompetitive.
• Unless otherwise stated we shall assume that equations (3.4) and

(3.5) hold, so that it is the implicit forward rates we are referring to
in our discussions.
7
• From equation (3.4),
(1 + iSt )t
iFt = − 1. (3.6)
(1 + iSt−1 )t−1
Example 3.1: Suppose the spot rates of interest for investment horizons
of 1, 2, 3 and 4 years are, respectively, 4%, 4.5%, 4.5%, and 5%. Calculate
the forward rates of interest for t = 1, 2, 3 and 4.
Solution: First, iF1 = iS1 = 4%. The rest of the calculation, using (3.6),
is as follows
(1 + iS2 )2 (1.045)2
iF2 = S
−1= − 1 = 5.0024%,
1 + i1 1.04
S 3 3
(1 + i3 ) (1.045)
iF3 = S 2
−1= 2
− 1 = 4.5%
(1 + i2 ) (1.045)
8
and
S 4 4
(1 + i4 ) (1.05)
iF4 = S 3
−1= 3
− 1 = 6.5144%.
(1 + i3 ) (1.045)
2
Example 3.2: Suppose the forward rates of interest for investments in
year 1, 2, 3 and 4 are, respectively, 4%, 4.8%, 4.8% and 5.2%. Calculate
the spot rates of interest for t = 1, 2, 3 and 4.
Solution: First, iS1 = iF1 = 4%. From (3.5) the rest of the calculation
is as follows
h i1 √
S F F 2
i2 = (1 + i1 )(1 + i2 ) − 1 = 1.04 × 1.048 − 1 = 4.3992%,
h i1 1
iS3 = (1 + iF1 )(1 + iF2 )(1 + iF3 ) 3
−1 = (1.04×1.048×1.048) −1 = 4.5327% 3
and
h i1
iS4 = (1 + iF1 )(1 + iF2 )(1 + iF3 )(1 + iF4 ) 4
−1
9
1
= (1.04 × 1.048 × 1.048 × 1.052) − 1 4
= 4.6991%.
• We define the multi-period forward rate iFt,τ as the annualized rate

of interest applicable over τ periods from time t to t + τ , for t ≥ 1
and τ > 0, with the rate being determined at time 0.
• The following no-arbitrage relationships hold
(1+iFt,τ )τ = (1+iFt+1 )(1+iFt+2 ) · · · (1+iFt+τ ), for t ≥ 1, τ > 0, (3.7)
and
(1 + iSt+τ )t+τ = (1 + iSt )t (1 + iFt,τ )τ , for t ≥ 1, τ > 0. (3.8)
10
Example 3.3: Based on the spot rates of interest in Example 3.1,
calculate the multi-period forward rates of interest iF1,2 and iF1,3 .
Solution: Using (3.7) we obtain

F 12 1
iF1,2 = [(1 + iF2 )(1 + i3 )] − 1 = (1.050024 × 1.045) − 1 = 4.7509%.
2
Similarly, we have
1
iF1,3 = (1.050024 × 1.045 × 1.065144) 3 − 1 = 5.3355%.
We may also use (3.8) to compute the multi-period forward rates. Thus,
" #1 " #1
(1 + iS3 )3 2
(1.045) 3 2
iF1,2 = −1= − 1 = 4.7509%,
1 + iS1 1.04
and similarly,
" #1 " #1
(1 + iS4 )4 3
(1.05)4 3
iF1,3 = −1= − 1 = 5.3355%.
1 + iS1 1.04
11
3.2 The Term Structure of Interest Rates
• Empirically the term structure can take various shapes. A sample

of four yield curves of the US market are presented in Figure 3.4.
• The spot-rate curve on 1998/06/30 is an example of a nearly flat

term structure.
• On 2000/03/31 we have a downward sloping term structure. In

this case the forward rate drops below the spot rate.
• We have an upward sloping term structure on 2001/08/31.
• Unlike the case of a downward sloping yield curve, the forward rate
exceeds the spot rate when the yield curve is upward sloping.
12
Figure 3.4: Yield curves of the US market
• An upward sloping yield curve is also said to have a normal term
structure as this is the most commonly observed term structure
empirically.
• We have an inverted humped yield curve on 2007/12/31.
• Some questions may arise from a cursory examination of this sample

of yield curves. For example,
— How are the yield curves obtained empirically?

— What determines the shape of the term structure?
— Why are upward sloping yield curves observed more often?
— Does the term structure have any useful information about the
real economy?
13
3.3 Present and Future Values Given the Term Structure
• We now consider the present and future values of an annuity given

the term structure.
• We continue to use the actuarial notations introduced in Chapter 2.
• The present value of a unit-payment annuity-immediate over n pe-

riods is
n
X 1
ane =
a(t)
t=1
Xn
1
= S t
t=1 (1 + it)
1 1 1
= S
+ S 2
+ · · · + S n
. (3.9)
(1 + i1 ) (1 + i2 ) (1 + in )
14
• The future value at time t of a unit payment at time 0 is
t
Y
a(t) = (1 + iSt )t = (1 + iFj ) = (1 + iF1 )(1 + iF2 ) · · · (1 + iFt ). (3.10)
j=1
• The present value of a unit payment due at time t is

1 1
= Qt F
, (3.11)
a(t) j=1 (1 + ij )
and the present value of a n-payment annuity-immediate can also

be written as
Xn
1
ane = (3.12)
t=1 a(t)
Xn
1
= Qt F
t=1 j=1 (1 + ij )
1 1 1
= F
+ F F
+ · · · + F F
.
(1 + i1 ) (1 + i1 )(1 + i2 ) (1 + i1 ) · · · (1 + in )
15
• The computation of the future value at time n of a payment due at
time t, where 0 < t < n, requires additional assumptions.
• We consider the assumption that a payment occurring in the

future earns the forward rates of interest.
• The future value at time n of a unit payment due at time t is
(1 + iFt,n−t )n−t = (1 + iFt+1 ) · · · (1 + iFn ), (3.13)
and the future value at time n of a n-period annuity-immediate is

"n−1 #
X
sne = (1 + iFt,n−t )n−t + 1
t=1
⎡ ⎛ ⎞⎤
n−1
X n
Y
= ⎣ ⎝ (1 + iFj )⎠⎦ + 1
t=1 j=t+1
= [(1 + iF2 )(1 + iF3 ) · · · (1 + iFn )] + [(1 + iF3 )(1 + iF4 ) · · · (1 + iFn )] + · · ·
16
· · · + (1 + iFn ) + 1. (3.14)
• From (3.14) we can see that

" n # " #
Y 1 1 1
sne = (1 + iFt ) × + + ··· + .
t=1
1 + iF1 (1 + iF1 )(1 + iF2 ) (1 + iF1 ) · · · (1 + iFn )
(3.16)
• Hence, Ã n !
Y
sne = (1 + iFt ) ane = a(n)ane . (3.17)
t=1
• An alternative formula to calculate sne using the spot rates of interest

is
sne = (1 + iSn )n ane
" n #
X 1
= (1 + iSn )n S t
. (3.18)
t=1 (1 + it )
17
of 1, 2, 3 and 4 years are, respectively, 4%, 4.5%, 4.5%, and 5%. Calculate
a4e , s4e , a3e and s3e .
Solution: From (3.9) we have

1 1 1 1
a4e = + + + = 3.5763.
1.04 (1.045)2 (1.045)3 (1.05)4
From (3.17) we obtain
s4e = (1.05)4 × 3.5763 = 4.3470.
Similarly,
1 1 1
a3e = + 2
+ 3
= 2.7536,
1.04 (1.045) (1.045)
and
s3e = (1.045)3 × 2.7536 = 3.1423.
18
2
Example 3.5: Suppose the 1-period forward rates of interest for in-
vestments due at time 0, 1, 2 and 3 are, respectively, 4%, 4.8%, 4.8% and
5.2%. Calculate a4e and s4e .
Solution: From (3.12) we have
1 1 1 1
a4e = + + +
1.04 1.04 × 1.048 1.04 × 1.048 × 1.048 1.04 × 1.048 × 1.048 × 1.052
= 3.5867.
As a(4) = 1.04 × 1.048 × 1.048 × 1.052 = 1.2016, we have

s4e = 1.2016 × 3.5867 = 4.3099.
Alternatively, from (3.14) we have
s4e = 1 + 1.052 + 1.048 × 1.052 + 1.048 × 1.048 × 1.052
= 4.3099.
19
2
• To further understand (3.17), we write (3.13) as (see (3.8))
(1 + iSn )n a(n)
(1 + iFt,n−t )n−t = S t
= . (3.19)
(1 + it ) a(t)
• Thus, the future value of the annuity is, from (3.14) and (3.19),
"n−1 #
X
sne = (1 + iFt,n−t )n−t + 1
t=1
n
Xa(n)
=
t=1 a(t)
n
X 1
= a(n)
t=1 a(t)
= a(n)ane . (3.20)
20
• In Chapter 2 we assume that the current accumulation function a(t)
applies to all future payments.
• Under this assumption, if condition (1.35) holds, equation (3.20) is

valid. However, for a given general term structure, we note that
(1 + iSn )n a(n)
a(n − t) = (1 + iSn−t )n−t 6= = ,
(1 + iSt )t a(t)
so that condition (1.35) does not hold.
• Thus, if future payments are assumed to earn spot rates of interest

based on the current term structure, equation (3.20) does not hold
in general.
of 1 to 5 years are 4%, and for 6 to 10 years are 5%. Calculate the present
21
value of an annuity-due of $100 over 10 years. Compute the future value
of the annuity at the end of year 10, assuming (a) future payments earn
forward rates of interest, and (b) future payments earn the spot rates of
interest as at time 0.
Solution: We consider the 10-period annuity-due as the sum of an

annuity-due for the first 6 years and a deferred annuity-due of 4 payments
starting at time 6. The present value of the annuity-due for the first 6
years is
" #
−6
1 − (1.04)
100 × ä6e0.04 = 100 ×
1 − (1.04)−1
= $545.18.
The present value at time 0 for the deferred annuity-due in the last 4 years
22
is
" #
−10 −6
1 − (1.05) 1 − (1.05)
100 × (ä10e0.05 − ä6e0.05 ) = 100 × −
1 − (1.05)−1 1 − (1.05)−1
= 100 × (8.1078 − 5.3295)
= $277.83.
Hence, the present value of the 10-period annuity-due is
545.18 + 277.83 = $823.01.
We now consider the future value of the annuity at time 10. Under as-
sumption (a) that future payments earn the forward rates of interest, the
future value of the annuity at the end of year 10 is, by equation (3.20),
(1.05)10 × 823.01 = $1,340.60.
23
Note that using (3.20) we do not need to compute the forward rates of
interest to determine the future value of the annuity, as would be required
if (3.16) is used.
Based on assumption (b), the payments at time 0, · · · , 4 earn interest at 5%

per year (the investment horizons are 10 to 6 years), while the payments
at time 5, · · · , 9 earn interest at 4% per year (the investment horizons are
5 to 1 years). Thus, the future value of the annuity is
100 × (s̈10e0.05 − s̈5e0.05 ) + 100 × s̈5e0.04 = $1,303.78.
Thus, equation (3.20) does not hold under assumption (b). 2
24
3.4 Accumulation Function and the Term Structure
• Equation (3.1) can be extended to any t > 0, which need not be an

integer.
• The annualized spot rate of interest for time to maturity t, iSt , is

given by
1
iSt = [a(t)] t − 1. (3.22)
• We may also use an accumulation function to define the forward

rates of interest.
• Let us consider the forward rate of interest for a payment due at

time t > 0, and denote the accumulation function of this payment
by at (·), where at (0) = 1.
25
• Now a unit payment at time 0 accumulates to a(t + τ ) at time t + τ ,
for τ > 0.
• On the other hand, a strategy with an initial investment over t pe-

riods and a rollover at the forward rate for the next τ periods will
accumulate to a(t)at (τ ) at time t + τ .
• By the no-arbitrage argument, we have
a(t)at (τ ) = a(t + τ ), (3.23)
so that
a(t + τ )
at (τ ) = . (3.24)
a(t)
• The annualized forward rate of interest in the period t to t + τ , iFt,τ ,
satisfies
at (τ ) = (1 + iFt,τ )τ ,
26
so that
1
iFt,τ = [at (τ )] τ − 1. (3.25)
• If τ < 1, we define the forward rate of interest per unit time (year)
for the fraction of a period t to t + τ as
1 at (τ ) − at (0) at (τ ) − 1
iFt,τ = × = . (3.26)
τ at (0) τ
• The instantaneous forward rate of interest per unit time at

time t is equal to iFt,τ for τ → 0, which is given by
at (τ ) − 1
lim iFt,τ = lim
τ →0 τ →0 τ" #
1 a(t + τ )
= lim × −1
τ →0 τ a(t)
" #
1 a(t + τ ) − a(t)
= lim
a(t) τ →0 τ
27
a0 (t)
=
a(t)
= δ(t). (3.27)
• Thus, the instantaneous forward rate of interest per unit time is

equal to the force of interest.
Example 3.7: Suppose a(t) = 0.01t2 +0.1t+1. Compute the spot rates
of interest for investments of 1, 2 and 2.5 years. Derive the accumulation
function for payments due at time 2, assuming the payments earn the
forward rates of interest. Calculate the forward rates of interest for time
to maturity of 1, 2 and 2.5 years.
Solution: Using (3.22), we obtain iS1 = 11%, iS2 = 11.36% and iS2.5 =
11.49%. Thus, we have an upward sloping spot-rate curve. To calculate
28
the accumulation function of payments at time 2 we first compute a(2) as
a(2) = 0.01(2)2 + 0.1(2) + 1 = 1.24.
Thus, the accumulation function for payments at time 2 is

a(2 + t) 0.01(2 + t)2 + 0.1(2 + t) + 1
a2 (t) = = = 0.0081t2 + 0.1129t + 1.
a(2) 1.24
Using the above equation, we obtain a2 (1) = 1.1210, a2 (2) = 1.2581 and
a2 (2.5) = 1.3327, from which we conclude iF2,1 = 12.10%,
1
iF2,2 = (1.2581) − 1 = 12.16%
2
and
1
iF2,2.5 = (1.3327) 2.5 − 1 = 12.17%.
Thus, the forward rates exceed the spot rates, which agrees with what
might be expected of an upward sloping yield curve. 2
29
• We can further establish the relationship between the force of in-
terest and the forward accumulation function, and thus the forward
rates of interest.

³R ´
t+τ
exp 0 δ(s) ds µZ t+τ ¶
a(t + τ )
at (τ ) = = ³R ´ = exp δ(s) ds ,
a(t) t
exp 0 δ(s) ds t
(3.28)
so that we can compute the forward accumulation function from the
force of interest.
Example 3.8: Suppose δ(t) = 0.05t. Derive the accumulation function

for payments due at time 2, assuming the payments earn the forward rates
of interest. Calculate the forward rates of interest for time to maturity of
1 and 2 years.
30
Solution: Using (3.26) we obtain
µZ 2+t ¶ h i
2 2
a2 (t) = exp 0.05s ds = exp 0.025(2 + t) − 0.025(2) = exp(0.025t2 +0.1t).
2
Thus, we can check that a2 (0) = 1. Now,
a2 (1) = exp(0.125) = 1.1331,
so that iF2,1 = 13.31%. Also, a2 (2) = exp(0.3) = 1.3499, so that

1
iF2,2 = (1.3499) − 1 = 16.18%.
2
• We now consider payments of C1 , C2 , · · · , Cn at time t1 < t2 < · · · <

tn , respectively.
• We wish to compute the value of these cash flows at any time t (≥ 0).
31
• For the payment Cj at time tj ≤ t, its accumulated value at time t
is Cj atj (t − tj ). On the other hand, if tj > t, the discounted value
of Cj at time t is Cj /at (tj − t).
• Thus, the value of the cash flows at time t is (see equation (3.24))
" # " # " #
X X 1 X a(t) X a(t)
Cj atj (t − tj ) + Cj = Cj + Cj
tj >t
at (tj − t) a(tj ) t
a(tj )
tj ≤t tj ≤t j >t
n
" #
X a(t)
= Cj
j=1
a(tj )
n
X
= a(t) Cj v(tj )
j=1
= a(t) × present value of cash flows.
(3.29)
32
• An analogous result can be obtained if we consider a continuous cash
flow.
• If C(t) is the instantaneous rate of cash flow at time t for 0 ≤ t ≤ n,

the value of the cash flow at time τ ∈ [0, n] is
Z τ Z n Z τ Z n
C(t) C(t)a(τ ) C(t)a(τ )
C(t)at (τ − t) dt + dt = dt + dt
0 τ aτ (t − τ ) 0 a(t) τ a(t)
Z n
C(t)
= a(τ ) dt
0 a(t)
Z n
= a(τ ) C(t)v(t) dt. (3.30)
0
Example 3.9: Suppose a(t) = 0.02t2 + 0.05t + 1. Calculate the value at

time 3 of a 1-period deferred annuity-immediate of 4 payments of $2 each.
You may assume that future payments earn the forward rates of interest.
33
Solution: We first compute the present value of the annuity. The
payments of $2 are due at time 2, 3, 4 and 5. Thus, the present value of
the cash flows is
" #
1 1 1 1
2× + + + .
a(2) a(3) a(4) a(5)
Now, a(2) = 0.02(2)2 + 0.05(2) + 1 = 1.18, and similarly we have a(3) =
1.33, a(4) = 1.52 and a(5) = 1.75. Thus, the present value of the cash
flow is
∙ ¸
1 1 1 1
2× + + + = 2 × 2.82866 = $5.6573,
1.18 1.33 1.52 1.75
and the value of the cash flow at time 3 is a(3) × 5.6573 = 1.33 × 5.6573 =
$7.5242. 2
Example 3.10: Suppose δ(t) = 0.02t. An investor invests in a fund

at the rate of 10t per period at time t, for t > 0. How much would she
34
accumulate in the fund at time 2? You may assume that future payments
earn the forward rates of interest.
Solution: The amount she invests in the period (t, t + ∆t) is 10t∆t,
which would accumulate to (10t∆t)at (2 − t) at time 2. Thus, the total
amount accumulated at time 2 is
Z 2
10tat (2 − t)dt.
0
From (3.28), we have

µZ 2 ¶ µZ 2 ¶
at (2 − t) = exp δ(s) ds = exp 0.02s ds .
t t
Now, we have
Z 2
0.02s ds = 0.01s2 ]2t = 0.01(2)2 − 0.01t2 ,
t
35
so that
at (2 − t) = exp(0.04 − 0.01t2 )
and
Z 2 Z 2
2
10tat (2 − t)dt = 10 te0.04−0.01t dt
0 0
Z 2
2
= 10e0.04 te−0.01t dt
t
10e0.04 ³ −0.01t2 2 ´
= −e ]0
0.02
10e0.04 (1 − e−0.04 )
=
0.02
= 20.4054.
Example 3.11: Suppose the principal is C and interest is earned at the
36
force of interest δ(t), for t > 0. What is the present value of the interest
earned over n periods.
Solution: As δ(t) is the instantaneous rate of interest per period at

time t, the amount of interest earned in the period (t, t + ∆t) is Cδ(t)∆t,
and the present value of this interest is [Cδ(t)∆t]v(t). Thus, the present
value of all the interest earned in the period (0, n) is
Z n
Cδ(t)v(t) dt.
0
Now, we have
Z n Z n µ Z t ¶
δ(t)v(t) dt = δ(t) exp − δ(s) ds dt
0 0 0
µ µ Z t ¶¶¸n
= − exp − δ(s) ds
0 0
µ Z 0 ¶ µ Z n ¶
= exp − δ(s) ds − exp − δ(s) ds
0 0
37
= 1 − v(n).
Hence, the present value of the interest earned is
C[1 − v(n)] = C − Cv(n),
which is the principal minus the present value of the principal redeemed
at time n. 2
38
for Actuaries
Chapter 4
Rates of Return
1
Learning Objectives
• Internal rate of return (yield rate)
• One-period rate of return of a fund: time-weighted rate of return

and dollar-weighted (money-weighted) rate of return
• Rate of return over longer periods: geometric mean rate of return

and arithmetic mean rate of return
• Portfolio return and return of a short-selling strategy
• Crediting interest: investment-year method and portfolio method
• Inflation: real rate of return
• Capital budgeting and project appraisal
2
4.1 Internal Rate of Return
• Consider a project with initial investment C0 . We assume the cash

flows occur at regular intervals.
• The project lasts for n years and the future cash flows are denoted
by C1 , · · · , Cn .
• We adopt the convention that cash inflows to the project (invest-

ments) are positive and cash outflows from the project (withdrawals)
are negative.
• We define the internal rate of return (IRR) (also called the yield
rate) as the rate of interest such that the sum of the present values
of the cash flows is equated to zero.
3
• Denoting the internal rate of return by y, we have
n
X Cj
j
= 0, (4.1)
j=0 (1 + y)
where j is the time at which the cash flow Cj occurs.
• This equation can also be written as

n
X Cj
C0 = − j
. (4.2)
j=1 (1 + y)
• Thus, the net present value of all future withdrawals (injections are
negative withdrawals) evaluated at the IRR is equal to the initial
investment.
Example 4.1: A project requires an initial cash outlay of $2,000 and

is expected to generate $800 at the end of year 1 and $1,600 at the end
4
of year 2, at which time the project will terminate. Calculate the IRR of
the project.
Solution: If we denote v = 1/(1 + y), we have, from (4.2)
2,000 = 800v + 1,600v 2 ,
or
5 = 2v + 4v 2 .
Dropping the negative answer from the quadratic equation, we have
√
−2 + 4 + 4 × 4 × 5
v= = 0.8956.
2×4
Thus, y = (1/v) − 1 = 11.66%. Note that v < 0 implies y < −1, i.e., the
loss is larger than 100%, which is precluded from consideration. 2
5
• There is generally no analytic solution for y in (4.1) when n > 2,
and numerical methods have to be used. The Excel function IRR
enables us to compute the answer easily. Its usage is described as
follows:
Excel function: IRR(values,guess)
values = an array of values containing the cash flows

guess = starting value, set to 0.1 if omitted
Output = IRR of cash flows
Example 4.2: An investor pays $5 million for a 5-year lease of a

shopping mall. He will receive $1.2 million rental income at the end of
each year. Calculate the IRR of his investment.
Solution: See Exhibit 4.1. 2
6
E hibit 4 1 IRR computation in Example 4.2
Exhibit 4.1: IRR t ti i E l 42
• If the cash flows occur more frequently than once a year, such as
monthly or quarterly, y computed from (4.1) is the IRR for the
payment interval.
• Suppose cash flows occur m times a year, the nominal IRR in an-
nualized term is m · y, while the annual effective rate of return is
(1 + y)m − 1.
Example 4.3: A cash outlay of $100 generates incomes of $20 after 4

months and 8 months, and $80 after 2 years. Calculate the IRR of the
investment.
Solution: If we treat one month as the interest conversion period, the

equation of value can be written as
20 20 80
100 = 4
+ 8
+ 24
,
(1 + y1 ) (1 + y1 ) (1 + y1 )
7
where y1 is the IRR on monthly interval. The nominal rate of return
on monthly compounding is 12y1 . Alternatively, we can use the 4-month
interest conversion interval, and the equation of value is
20 20 80
100 = + 2
+ 6
,
1 + y4 (1 + y4 ) (1 + y4 )
where y4 is the IRR on 4-month interval. The nominal rate of return

on 4-monthly compounding is 3y4 . The effective annual rate of return is
y = (1 + y1 )12 − 1 = (1 + y4 )3 − 1.
The above equations of value have to be solved numerically for y1 or y4 .
We obtain 1.0406% as the IRR per month, namely, y1 . The effective
annualized rate of return is then (1.010406)12 − 1 = 13.23%. Solving
for y4 with Excel, we obtain the answer 4.2277%. Hence, the annualized
effective rate is (1.042277)3 − 1 = 13.23%, which is equal to the effective
rate computed using y1 . 2
8
• When cash flows occur irregularly, we can define y as the annualized
rate and express all time of occurrence of cash flows in years.
• Suppose there are n + 1 cash flows occurring at time 0, t1 , · · ·, tn ,

with cash amounts C0 , C1 , · · ·, Cn . Equation (4.2) is rewritten as
n
X Cj
C0 = − t
, (4.3)
j=1 (1 + y)
j
which requires numerical methods for the solution of y.
• We may also use the Excel function XIRR to solve for y. Unlike IRR,
XIRR allows the cash flows to occur at irregular time intervals. The
specification of XIRR is as follows:
9
Excel function: XIRR(values,dates,guess)
values = Cj , an array of values containing the cash flows in (4.3)

dates = tj , an array of values containing the timing of the cash flows in (4.3)
guess = guess starting value, set to 0.1 if omitted
Output = IRR, value of annualized y satisfying (4.3)
Example 4.4: A project requires an outlay of $2.35 million in return

for $0.8 million after 9 months, $1 million after 15 months and $1 million
after 2 years. What is the IRR of the project?
Solution: Returns of the project occur at time (in years) 0.75, 1.25
and 2. We solve for v numerically from the following equation using Excel
Solver (see Exhibit 4.2)
235 = 80v 0.75 + 100v 1.25 + 100v 2
10
E hibit 4 1 IRR computation in Example 4.2
Exhibit 4.1: IRR t ti i E l 42
Figure 4.2: Solution of Example 4.4 using Solver

Figure 4.3: Solution of Example 4.4 using XIRR
to obtain v = 0.879, so that
1
y= − 1 = 13.77%,
0.879
which is the effective annualized rate of return of the project. Alterna-
tively, we may use the Excel function XIRR as shown in Exhibit 4.3. 2
• A project with no subsequent investment apart from the initial cap-

ital is called a simple project.
• For simple projects, (4.1) has a unique solution with y > −1, so that
IRR is well defined.
Example 4.5: A project requires an initial outlay of $8 million, gen-

erates returns of $50 million 1 year later, and requires $50 million to
terminate at the end of year 2. Solve for y in (4.1).
11
Solution: We are required to solve
8 = 50v − 50v 2 ,
which has v = 0.8 and 0.2 as solutions. This implies y has multiple solu-
tions of 25% and 400%. 2
12
4.2 One-Period Rate of Return
• We consider methods of calculating the return of a fund over a 1-

period interval. The methodology adopted depends on the data
available.
• We start with the situation where the exact amounts of fund with-
drawals and injections are known, as well as the time of their occur-
rence.
• Consider a 1-year period with initial fund amount B0 (equal to C0 ).

Cash flow of amount Cj occurs at time tj (in fraction of a year) for
j = 1, · · · , n, where 0 < t1 < · · · < tn < 1.
• Note that Cj are usually fund redemptions and new investments, and
do not include investment incomes such as dividends and coupon
13
payments.
• Denoting the fund value before and after the transaction at time tj
by BjB and BjA , respectively, we have BjA = BjB +Cj for j = 1, · · · , n.
• The difference between BjB and Bj−1 A

, i.e., the balance before the
transaction at time tj and after the transaction at time tj−1 , is due
to investment incomes, as well as capital gains and losses.
B
• Let the fund balance at time 1 be B1 , and define Bn+1 = B1 and
B0A = B0 (this notation will allow us to express the gross return as
(4.4) below). Figure 4.1 illustrates the time diagram.
• We now introduce two methods to calculate the 1-year rate of re-

turn: the time-weighted rate of return (TWRR) and the dollar-
weighted rate of return (DWRR).
14
• To compute the TWRR we first calculate the return over each subin-
terval between the occurrences of transactions by comparing the fund
balances just before the new transaction to the fund balance just af-
ter the last transaction.
• If we denote Rj as the rate of return over the subinterval tj−1 to tj ,

we have
BjB
1 + Rj = A , for j = 1, · · · , n + 1. (4.4)
Bj−1
• Then TWRR over the year, denoted by RT , is

⎡ ⎤
n+1
Y
RT = ⎣ (1 + Rj )⎦ − 1. (4.5)
j=1
• The TWRR requires data of the fund balance prior to each with-
drawal or injection. In contrast, the DWRR does not require this
15
information. It only uses the information of the amounts of the
withdrawals and injections, as well as their time of occurrence.
• In principle, when cash of amount Cj is injected (withdrawn) at

time tj , there is a gain (loss) of capital of amount Cj (1 − tj ) for the
remaining period of the year.
• Thus, the effective capital of the fund over the 1-year period, denoted
by B, is given by
n
X
B = B0 + Cj (1 − tj ).
j=1
P
• Denoting C = nj=1 Cj as the net injection of cash (withdrawal if
negative) over the year and I as the interest income earned over the
year, we have B1 = B0 + I + C, so that
I = B1 − B0 − C. (4.6)
16
Hence the DWRR over the 1-year period, denoted by RD , is
I B1 − B0 − C
RD = = P . (4.7)
B B0 + nj=1 Cj (1 − tj )
Example 4.6: On January 1, a fund was valued at 100k (1k = 1,000).

On May 1, the fund increased in value to 112k and 30k of new principal was
injected. On November 1, the fund value dropped to 125k, and 42k was
withdrawn. At the end of the year, the fund was worth 100k. Calculate
the DWRR and the TWRR.
Solution: As C = 30 − 42 = −12, there is a net withdrawal. From

(4.6), the interest income earned over the year is
I = 100 − 100 − (−12) = 12.
17
Cash flow 30 42
Time t, mm/dd 1/1 5/1 11/1 1/1
Value before t 112 125
Value after t 142 83
Subperiod 1 2 3
Figure 4.2: Illustration for Example 4.6

Hence, from (4.7), the DWRR is
12
RD = = 10.62%.
2 1
100 + × 30 − × 42
3 6
The fund balance just after the injection on May 1 is 112 + 30 = 142k,
and its value just after the withdrawal on November 1 is 125 − 42 = 83k.
From (4.4), the fund-value relatives over the three subperiods are
112
1 + R1 = = 1.120,
100
125
1 + R2 = = 0.880,
142
100
1 + R3 = = 1.205.
83
Hence, from (4.5), the TWRR is
RT = 1.120 × 0.880 × 1.205 − 1 = 18.76%.
18
2
• TWRR compounds the returns of the fund over subperiods after

purging the effects of the timing and amount of cash injections and
withdrawals.
• As fund managers have no control over the timing of fund injection

and withdrawal, the TWRR appropriately measures the performance
of the fund manager.
• DWRR is sensitive to the timing and amount of the cash flows.
• If the purpose is to measure the performance of the fund, the DWRR

is appropriate.
• It allows superior market timing to impact the return of the fund.
19
• For funds with frequent cash injections and withdrawals, the com-
putation of the TWRR may not be feasible. The difficulty lies in
the evaluation of the fund value BjB , which requires the fund to be
constantly marked to market.
• In some situations the exact timing of the cash flows may be difficult
to identify.
• In this situation, we may approximate (4.7) by assuming the cash

flows to be evenly distributed throughout the 1-year evaluation pe-
riod.
Pn
• Hence, we replace 1 − tj by its mean value of 0.5 so that j=1 Cj (1 −
tj ) = 0.5C, and (4.7) can be written as
I
RD '
B0 + 0.5C
20
I
=
B0 + 0.5(B1 − B0 − I))
I
= . (4.8)
0.5(B1 + B0 − I)
Example 4.7: For the data in Example 4.6, calculate the approximate
value of the DWRR using (4.8).
Solution: With B0 = B1 = 100, and I = 12, the approximate RD is

12
RD = = 12.76%.
0.5(100 + 100 − 12)
2
21
4.3 Rate of Return over Multiple Periods
• We now consider the rate of return of a fund over a m-year period.
• We first consider the case where only annual data of returns are
available. Suppose the annual rates of return of the fund have been
computed as R1 , · · · , Rm . Note that −1 ≤ Rj < ∞ for all j.
• The average return of the fund over the m-year period can be cal-
culated as the mean of the sample values. We call this measure the
arithmetic mean rate of return, denoted by RA , which is given
by
m
1 X
RA = Rj . (4.9)
m j=1
• An alternative is to use the geometric mean to calculate the average,
called the geometric mean rate of return, denoted by RG , which
22
is given by
⎡ ⎤1
m m
Y
RG = ⎣ (1 + Rj )⎦ − 1
j=1
1
= [(1 + R1 )(1 + R2 ) · · · (1 + Rm )] m − 1. (4.10)
Example 4.8: The annual rates of return of a bond fund over the last
5 years are (in %) as follows:
6.4 8.9 2.5 −2.1 7.2
Calculate the arithmetic mean rate of return and the geometric mean rate
of return of the fund.
Solution: The arithmetic mean rate of return is
RA = (6.4 + 8.9 + 2.5 − 2.1 + 7.2)/5
23
= 22.9/5
= 4.58%,
and the geometric mean rate of return is
1
RG = (1.064 × 1.089 × 1.025 × 0.979 × 1.072) − 15
1
= (1.246) 5 − 1
= 4.50%.
2
Example 4.9: The annual rates of return of a stock fund over the last
8 years are (in %) as follows:
15.2 18.7 −6.9 −8.2 23.2 −3.9 16.9 1.8
Calculate the arithmetic mean rate of return and the geometric mean rate
of return of the fund.
24
Solution: The arithmetic mean rate of return is
RA = (15.2 + 18.7 + · · · + 1.8)/8 = 7.10%
and the geometric mean rate of return is

1
RG = (1.152 × 1.187 × · · · × 1.018) − 1 = 6.43%.
8
• Given any sample of data, the arithmetic mean is always larger than
the geometric mean.
• If the purpose is to measure the return of the fund over the holding
period of m years, the geometric mean rate of return is the appro-
priate measure.
25
• The arithmetic mean rate of return describes the average perfor-
mance of the fund for one year taken randomly from the sample
period.
• If there are more data about the history of the fund, alternative mea-
sures of the performance of the fund can be used. The methodology
of the time-weighted rate of return in Section 4.2 can be extended
to beyond 1 period (year).
• Suppose there are n subperiods, with returns denoted by R1 , · · · ,

Rn , over a period of m years. Then we can measure the m-year
return by compounding the returns over each subperiod to form the
TWRR using the formula
⎡ ⎤1
n m
Y
RT = ⎣ (1 + Rj )⎦ − 1. (4.11)
j=1
26
• We can also compute the return over a m-year period using the IRR.
We extend the notations for cash flows in Section 4.2 to the m-year
period.
• Suppose cash flows of amount Cj occur at time tj for j = 1, · · · , n,

where 0 < t1 < · · · < tn < m. Let the fund value at time 0 and time
m be B0 and B1 , respectively. We treat −B1 as the last transaction,
i.e., fund withdrawal of amount B1 .
• The rate of return of the fund is calculated as the IRR which equates
the discounted values of B0 , C1 , · · · , Cn , and −B1 to zero.
• This is referred to as the DWRR over the m-year period. We denote

it as RD , which solves the following equation
n
X Cj B1
B0 + t
− m
= 0. (4.12)
j=1 (1 + RD ) (1 + RD )
j
27
• The example below concerns the returns of a bond fund. When a
bond makes the periodic coupon payments, the bond values drop
and the coupons are cash amounts to be withdrawn from the fund.
Example 4.10: A bond fund has an initial value of $20 million. The
fund records coupon payments in six-month periods. Coupons received
from January 1 through June 30 are regarded as paid on April 1. Likewise,
coupons received from July 1 through December 31 are regarded as paid
on October 1. For the 2-year period 2008 and 2009, the fund values and
coupon payments were recorded in Table 4.1.
28
Table 4.1: Cash flows of fund
Time Coupon received Fund value before

mm/dd/yy ($ millions) date ($ millions)
01/01/08 20.0
04/01/08 0.80 22.0
10/01/08 1.02 22.8
04/01/09 0.97 21.9
10/01/09 0.85 23.5
12/31/09 25.0
Calculate the TWRR and the DWRR of the fund.
Solution: As the coupon payments are withdrawals from the fund (the
portfolio of bonds), the fund drops in value after the coupon payments.
For example, the bond value drops to 22.0 − 0.80 = 21.2 million on April
1, 2008 after the coupon payments. Thus, the TWRR is calculated as
29
∙ ¸0.5
22 22.80 21.90 23.50 25.0
RT = × × × × −1 = 21.42%.
20 22.0 − 0.8 22.80 − 1.02 21.90 − 0.97 23.50 − 0.85
To calculate the DWRR we solve RD from the following equation

0.8 1.02 0.97 0.85 25
20 = + + + +
(1 + RD )0.25 (1 + RD )0.75 (1 + RD )1.25 (1 + RD )1.75 (1 + RD )2
= 0.8v + 1.02v 3 + 0.97v 5 + 0.85v 7 + 25v 8 ,
where v = (1+RD )−0.25 . We let (1+y)−1 = v and use Excel to obtain y =

4.949%, so that the annual effective rate of return is RD = (1.04949)4 −1 =
21.31%. 2
30
Figure 4.4: DWRR computation in Example 4.10
4.4 Portfolio Return
• We now consider the return of a portfolio of assets.
• Suppose a portfolio consists of N assets denoted by A1 , · · · , AN . Let

the value of asset Aj in the portfolio at time 0 be A0j , for j =
1, · · · , N.
• We allow A0j to be negative for some j, so that asset Aj is sold short

in the portfolio.
PN
• The portfolio value at time 0 is B0 = j=1 A0j . Let the asset values
P
at time 1 be A1j , so that the portfolio value is B1 = N
j=1 A1j .
• Denote RP as the return of the portfolio in the period from time 0
31
to time 1. Thus,
B1 − B0 B1
RP = = − 1.
B0 B0
• We define
A0j
wj = ,
B0
which is the proportion of the value of asset Aj in the initial portfolio,
so that
N
X
wj = 1,
j=1
and wj < 0 if asset j is short sold in the portfolio.
• We also denote
A1j − A0j A1j
Rj = = − 1,
A0j A0j
32
which is the rate of return of asset j. Thus,
B1
1 + RP =
B0
N
1 X
= A1j
B0 j=1
N
X A0j A1j
= ×
j=1 B0 A0j
N
X
= wj (1 + Rj ),
j=1
which implies
N
X
RP = wj Rj , (4.13)
j=1
so that the return of the portfolio is the weighted average of the

returns of the individual assets.
33
• Eq (4.13) is an identity, and applies to realized returns as well as
returns as random variables. If we take the expectations of (4.13),
we obtain
N
X
E(RP ) = wj E(Rj ), (4.14)
j=1
so that the expected return of the portfolio is equal to the weighted

average of the expected returns of the component assets.
• The variance of the portfolio return is given by

N
X N X
X N
Var(RP ) = wj2 Var(Rj ) + wh wj Cov(Rh , Rj ). (4.15)
j=1 h=1 j=1
| {z }
h6=j
• For example, consider a portfolio consisting of two funds, a stock

fund and a bond fund, with returns denoted by RS and RB , respec-
34
tively. Likewise, we use wS and wB to denote their weights in the
portfolio.
• Then, we have
E(RP ) = wS E(RS ) + wB E(RB ), (4.16)
and
Var(RP ) = wS2 Var(RS )+wB

2
Var(RB )+2wS wB Cov(RS , RB ), (4.17)
where wS + wB = 1.
Example 4.11: A stock fund has an expected return of 0.15 and

variance of 0.0625. A bond fund has an expected return of 0.05 and
variance of 0.0016. The correlation coefficient between the two funds is
−0.2.
35
(a) What is the expected return and variance of the portfolio with 80%
in the stock fund and 20% in the bond fund?
(b) What is the expected return and variance of the portfolio with 20%
in the stock fund and 80% in the bond fund?
(c) How would you weight the two funds in your portfolio so that your
portfolio has the lowest possible variance?
Solution: For (a), we use (4.16) and (4.17), with wS = 0.8 and wB =
0.2, to obtain
E(RP ) = (0.8)(0.15) + (0.2)(0.05) = 13%

q
Var(RP ) = (0.8)2 (0.0625) + (0.2)2 (0.0016) + 2(0.8)(0.2)(−0.2) (0.0016)(0.0625)
= 0.03942.
36
√
Thus, the portfolio has a standard deviation of 0.03942 = 19.86%.
For (b), we do similar calculations, with wS = 0.2 and wB = 0.8, to
obtain E(RP ) = 7%, Var(RP ) = 0.002884 and a standard deviation of
√
0.002884 = 5.37%.
Hence, we observe that the portfolio with a higher weightage in stock has
a higher expected return but also a higher standard deviation, i.e., higher
risk.
For (c) we rewrite (4.17) as
Var(RP ) = wS2 Var(RS ) + (1 − wS )2 Var(RB ) + 2wS (1 − wS )Cov(RS , RB ).
To minimize the variance, we differentiate Var(RP ) with respect to wS to

obtain
2wS Var(RS ) − 2(1 − wS )Var(RB ) + 2(1 − 2wS )Cov(RS , RB ).
37
Equating the above to zero, we solve for wS to obtain
Var(RB ) − Cov(RS , RB )
wS =
Var(RB ) + Var(RS ) − 2Cov(RS , RB )
= 5.29%.
The expected return of this portfolio is 5.53%, its variance is 0.001410,

and its standard deviation is 3.75%, which is lower than the standard
deviation of the bond fund of 4%. Hence, this portfolio dominates the
bond fund, in the sense that it has a higher expected return and a lower
standard deviation.
Note that the fact that the above portfolio indeed minimizes the variance
can be verified by examining the second-order condition. 2
38
4.5 Short sales
• A short sale is the sale of a security that the seller does not own. It
can be executed through a margin account with a brokerage firm.
• The seller borrows the security from the brokerage firm to deliver to
the buyer.
• The sale is based on the belief that the security price will go down
in the future so that the seller will be able to buy back the security
at a lower price, thus keeping the difference in price as profit.
• Proceeds from the short sale are kept in the margin account, and
cannot be invested to earn income.
• The seller is required to place cash or securities into the margin

account. The initial percentage margin m is the percentage of
39
the proceeds of the short sold security that the seller must place into
the margin account.
• If P0 is the price of the security when it is sold short, the initial

deposit D is P0 m. The seller will earn interest from the deposit.
• At any point in time, there is a maintenance margin m∗ , which

is the minimum percentage of the seller’s equity in relation to the
current value of the security sold short.
• If the current security price is P1 , the equity E is P0 + D − P1 and

E/P1 must be larger than m∗ .
• If E/P1 falls below m∗ , the seller will get a margin call from the
broker instructing him to top up his margin account.
40
Example 4.12: A person sold 1,000 shares of a stock short at $20.
If the initial margin is 50%, how much should he deposit in his margin
account? If the maintenance margin is 30%, how high can the price go up
before there is a margin call?
Solution: The initial deposit is
1,000 × 20 × 0.5 = $10,000.
At price P1 , the percent of equity is

10,000 + 1,000(20 − P1 )
,
1,000P1
which must be more than 0.3. Thus, the maximum P1 is
30
= $23.08.
1.3
2
41
• To calculate the rate of return of a short sale strategy we note that
the capital is the deposit D in the margin account. The return
includes the interest earned in the deposit. The seller, however,
pays the dividend to the buyer if there is any dividend payout. The
net rate of return will thus take account of the interest earned and
the dividend paid.
Example 4.13: A person sells a stock short at $30. The stock pays a
dividend of $1 at the end of the year, after which the man covers his short
position by buying the stock back at $27. The initial margin is 50% and
interest rate is 4%. What is his rate of return over the year?
Solution: The capital investment per share is $15. The gross return
after one year is (30 − 27) − 1 + 15 × 0.04 = $2.6. Hence, the return over
42
the 1-year period is
2.6
= 17.33%.
15
Note that in the above calculation we have assumed that there was no
margin call throughout the year. 2
43
4.6 Crediting Interest: Investment-Year Method and
Portfolio Method
• A fund pools the investments of individual investors. The investors

may invest new money into the fund at any time.
• While the fund invests the aggregate of the investments, there is an

issue of how to credit interest to the individual investors’ accounts.
• A simple method is to credit the average return of the fund to all

investors. This is called the portfolio method.
• The portfolio method may not be equitable when the individual

investments are made at different times.
• For example, at a time when returns to securities are going up, new
investments are likely to achieve higher returns compared to old
44
investments. In this case, crediting the average return will not be
attractive to new investments.
• When security returns are declining, the portfolio method may be

crediting higher returns to new investors at the expense of the old
ones.
• Another method is to credit the accounts according to their year of

investment. This method is called the investment-year method.
• Under this method, new investments are credited the investment-

year rates of interest over a period of time, after which the investors
are credited the portfolio rate of interest.
• Table 4.2 gives an example. For simplicity of exposition, we assume

that investments are made at the beginning of the calendar year and
interests are credited at the end of the year.
45
Table 4.2: An example of investment-year rates and portfolio rates
Calendar year Investment-year rates (%) Portfolio rates (%)

of investment (Y ) iY1 iY2 iY3 iY +3
2004 5.6 5.6 5.7 5.9
2005 5.6 5.7 5.8 6.2
2006 5.8 5.9 6.0 6.3
2007 6.2 6.3 6.6
2008 6.7 6.4
2009 7.1
• In Table 4.2, investments are credited the investment-year rates in

the first 3 years. The calendar year of the investment is denoted by
Y , the rate of interest credited for the tth year of the investment
made in calendar year Y is iYt for t = 1, 2, 3, and iY is the portfolio
rate credited for calendar year Y .
46
• Using these notations, iYt = iY +t−1 for t > 3.
Example 4.14: A fund credits interest according to Table 4.2. Find

the total interest credited in the period 2007 through 2009 for (a) an
investment in 2002, (b) an investment in 2006, (c) an investment in 2007,
(d) an investment in 2007, withdrawn every year and reinvested in the
fund as new money. All investments, including reinvestments, are made
on January 1.
Solution: The last column of the table gives the portfolio rate of
interest in 2007 through 2009. For (a), the investment made in 2002 earns
the portfolio rate of interest in 2007 through 2009. Thus, the total return
over the 3-year period is
1.059 × 1.062 × 1.063 − 1 = 19.55%.
47
For (b), the investment made in 2006 earns the investment-year rates in
2007 and 2008 of 5.9% and 6.0%, respectively, and the portfolio rate in
2009 of 6.3%. The total return is
1.059 × 1.06 × 1.063 − 1 = 19.33%.
For (c), the investment made in 2007 earns the investment-year rates in
2007 through 2009 to obtain the total return
1.062 × 1.063 × 1.066 − 1 = 20.34%.
Finally, for (d) if an investment made in 2007 is withdrawn every year and
reinvested at the new money rate, the total return is
1.062 × 1.067 × 1.071 − 1 = 21.36%.
48
4.7 Inflation and Real Rate of Interest
• Inflation refers to the increase in the general price level of goods and
services.
• The rate of inflation is usually measured by the rate of increase of

a general price index such as the consumer price index, whole-sale
price index or gross domestic product deflator.
• We define the real rate of interest as the interest earned by an

investment after taking account of the erosion of purchasing power
due to inflation.
• We denote rI as the inflation rate, and rN as the nominal rate of

interest.
• The real rate of interest rR is then defined by the equation
49
1 + rN
1 + rR = , (4.18)
1 + rI
1 + rN rN − rI
rR = −1= . (4.19)
1 + rI 1 + rI
• If the rate of inflation rI is low, we conclude from (4.19) that
rR ' rN − rI , (4.20)
which says that the real rate of interest is approximately equal to

the nominal rate of interest minus the rate of inflation.
• Note that (4.18) and (4.19) are ex post relationships. They hold
empirically and do not refer to any theoretical relationship between
the three quantities.
50
• The economist Irving Fisher argued that the nominal rate of interest
ought to increase one for one with the expected rate of inflation.
• The so-called Fisher equation states that
rN = rR + E(rI ), (4.21)
where E(rI ) is the expected rate of inflation.
Example 4.15: A stock will pay a dividend of $0.50 two months from
now and the annual dividend is expected to grow at a rate of 4% per
annum indefinitely. If the stock is traded at $7.5 and inflation is expected
to be 2% per annum, what is the expected effective annual real rate of
return for an investor who purchases the stock?
Solution: Let v = 1/(1 + rN ), where rN is the nominal rate of return

prior to adjustment for inflation. The dividends are 0.5, 1.04(0.5), 1.042 (0.5), · · ·
51
at time of 2 months, 14 months, 26 months, · · ·, respectively from now.
Thus, the equation of value is
1 14 26
7.5 = 0.5v 6 + 1.04(0.5)v 12 + 1.042 (0.5)v 12 + · · ·
1
h i
2
= 0.5v 6 1 + 1.04v + (1.04v) + · · ·
1
0.5v 6
= .
1 − 1.04v
The above equation has no analytic solution, and we solve it using Excel
to obtain v = 0.898568 and rN = (1/0.898568) − 1 = 0.112882. Thus, the
real rate of return is
1 + rN 1.112882
rR = −1= − 1 = 9.1061%.
1 + rI 1.02
2
52
4.8 Capital Budgeting and Project Appraisal
• Capital budgeting and project appraisal refer to the managerial deci-

sion of investing in a project. We consider some methods of making
such decisions.
• One approach is called the IRR rule, which requires the manager to
calculate the IRR of the project. The IRR is then compared against
the required rate of return of the project, which is the minimum
rate of return on an investment needed to make it acceptable to a
business.
• We denote the required rate of return by RR . The IRR rule accepts

the project if IRR ≥ RR , .
• However, RR is often a difficult quantity to identify. It depends
53
on factors such as the market rate of interest, the horizon of the
investment and the risk of the project. In what follows, we shall
assume RR to be given and focus on the application of the budgeting
rules.
• An alternative approach is the net present value (NPV) rule.
• Consider a project with annual cash flows and adopt the notations
in Section 4.1. Discounting the future cash flows by RR , we rewrite
(4.2) and define the NPV of the project as
n
X Cj
NPV = − j
− C0 . (4.22)
j=1 (1 + RR )
• A positive NPV implies that the sum of the present values of cash
outflows created by the project exceeds the sum of the present values
54
of all investments. Thus, managers should invest in projects with
positive NPV.
• As we have seen in Section 4.1, the IRR may not be unique. When
there are multiple IRRs, the decision is ambiguous.
Example 4.16: Consider a 2-year project with an initial investment

of $1,000. A cash amount of $2,230 will be generated after 1 year, and
the project will be terminated with fund injection of $1,242 at the end of
year 2. What conclusion can be drawn regarding the acceptance of this
project?
Solution: The cash flow diagram is given in Figure 4.3. To calculate

the IRR we solve for the equation
1,000 − 2,230v + 1,242v 2 = 0,
55
where v = 1/(1 + y) and y is the IRR. Note that we have adopted the con-
vention of cash inflow being positive and outflow being negative. Solving
for the above equation we obtain y = 8% or 15%. For RR lying between
the two roots of y, the IRR rule cannot be applied. However, if RR is
below the minimum of the two solutions of y, say 6%, can we conclude
the project be accepted?
Evaluating the NPV at RR = 6%, we obtain
2,230 1,242
NPV = −1,000 + −
1.06 (1.06)2
= −$1.60.
Thus, the project has a negative NPV and should be rejected. Figure 4.4
plots the NPV of the project as a function of the interest rate. It shows
that the NPV is positive for 0.08 < RR < 0.15, which is the region of RR
for which the project should be accepted. Outside this region, the project
56
has a negative NPV and should be rejected. 2
• The NPV rule can be applied with varying interest rates that reflects
a term structure which is not flat.
• We can modify (4.22) to calculate the NPV, where the cash flow at
time j is discounted by the rate of interest Rj , such as
n
X Cj
NPV = − j
− C0 . (4.23)
j=1 (1 + Rj )
• If the IRR method is adopted, the term structure must be assumed

to be flat.
• While (4.22) computes the NPV of the project upon its completion,
it can be modified to compute the NPV of the project at a time prior
to its completion.
57
• Thus, for t ≤ n, we define the NPV of the project up to time t as
t
X Cj
NPV(t) = − j
− C0 . (4.25)
j=1 (1 + RR )
• If NPV(t) changes sign only once in the duration of the project, we

can calculate the value of t at which NPV(t) first becomes positive.
• This value of t is called the discounted payback period (DPP).

Formally, DPP is defined as
DPP = min {t : NPV(t) > 0}. (4.26)
• A capital budgeting criterion is to choose the project with the lowest

DPP.
58
Example 4.19: Consider the two projects in Example 4.17. Discuss
the choice of the two projects based on the DPP criterion.
Solution: For Project B, as the most substantial cash flow generated

is at the end of the project duration, the DPP of the project is 15 years.
For Project A, Table 4.3 summarizes the results for several required rates
of interest RR .
59
Table 4.3: NPV of Project A
RR t NPV(t) DPP
0.03 12 —4.5996 13
13 63.4955
0.04 13 —1.4352 14
14 56.3123
0.05 14 —10.1359 15
15 37.9658
0.06 15 —28.7751 16
16 10.5895
For required rates of interest of 3%, 4%, 5% and 6%, the DPP of Project
A are, respectively, 13, 14, 15 and 16 years. When the required rate
of interest is 6%, Project B has a negative NPV and the DPP is not
defined. When the required interest rate is 5%, Project B has the same
60
DPP as Project A, namely, 15 years. For other required rates of interest
considered, the DPP of Project A is shorter. 2
• If RR in (4.25) is set equal to zero, the resulting value of t at which

NPV(t) first becomes positive is called the payback period. Al-
though a project appraisal method may be based on the minimum
payback period, this method lacks theoretical rigor as the time value
of money is not taken into account.
61
for Actuaries
Chapter 5
Loans and Costs of Borrowing
1
Learning Objectives
1. Loan balance: prospective method and retrospective method
2. Amortization schedule
3. Sinking fund
4. Varying installments and varying interest rates
5. Quoted rate of interest and equivalent nominal rate of interest in

monthly rest
6. Flat rate loan and flat rate discount loan
7. Annual percentage rate, annual percentage yield, effective rate of

interest, and comparison rate of interest
2
5.1 Loan Balance: Prospective and Retrospective Methods
• Consider a loan with a fixed term of maturity, to be redeemed by a

series of repayments.
• If the repayments prior to maturity are only to offset the interests,

the loan is called an interest-only loan.
• If the repayments include both payment of interest and partial re-

demption of the principal, the loan is called a repayment loan.
• We consider two approaches to compute the balance of the loan: the

prospective method and the retrospective method.
• The prospective method is forward looking. It calculates the loan

balance as the present value of all future payments to be made.
3
• The retrospective method is backward looking. It calculates the
loan balance as the accumulated value of the loan at the time of
evaluation minus the accumulated value of all installments paid up
to the time of evaluation.
• Let the loan amount be L, and the rate of interest per payment
period be i. If the loan is to be paid back in n installments of an
annuity-immediate, the installment amount A is given by
L
A= . (5.1)
anei
• We also denote L = B0 , which is the loan balance at time 0.
• Immediately after the mth payment has been made the loan is re-
deemable by a n−m annuity-immediate. We denote the loan balance
4
after the mth installment by Bm , which is
Bm = A an−mei
L an−mei
= . (5.2)
anei
• To use the retrospective method, we first calculate the accumulated

loan amount at time m, which is L(1 + i)m .
• The accumulated value of the installments is Asmei .
• Thus, the loan balance can also be written as
L(1 + i)m − Asmei . (5.3)
• To show that the two methods are equivalent, we simplify (5.3).
5
Thus, we have
" #
n m
1−v (1 + i) − 1
Aanei (1 + i)m − Asmei = A (1 + i)m −
i i
" #
n−m
1−v
= A
i
= A an−mei .
Example 5.1: A housing loan of $400,000 was to be repaid over 20

years by monthly installments of an annuity-immediate at the nominal rate
of 5% per year. After the 24th payment was made, the bank increased
the interest rate to 5.5%. If the lender was required to repay the loan
within the same period, how much would be the increase in the monthly
installment. If the installment remained unchanged, how much longer
would it take to pay back the loan?
6
Solution: We first demonstrate the use of the prospective and retro-
spective methods for the calculation of the loan balance after the 24th
payment. From (5.1), the amount of the monthly installment is
400,000
A =
a240e0.05/12
400,000
=
151.525
= $2,639.82.
By the prospective method, after the 24th payment the loan would be
redeemed with a 216-payment annuity-immediate so that the balance is
A a216e0.05/12 = 2,639.82 × 142.241

= $375,490.
7
By the retrospective method, the balance is
µ ¶24
0.05
400,000 1 + − 2,639 s24e0.05/12 = 441,976.53 − 2,639.82 × 25.186
12
= $375,490.
Hence, the two methods give the same answer. After the increase in the
rate of interest, if the loan is to be repaid within the same period, the
revised monthly installment is
375,490 375,490
=
a216e0.055/12 136.927
= $2,742.26,
so that the increase in installment is $102.44. Let m be the remaining

number of installments if the amount of installment remained unchanged.
Thus,
ame0.055/12 = a216e0.05/12 = 142.241,
8
from which we solve for m = 230.88 ≈ 231. Thus, it takes 15 months
more to pay back the loan. 2
Example 5.2: A housing loan is to be repaid with a 15-year monthly

annuity-immediate of $2,000 at a nominal rate of 6% per year. After 20
payments, the borrower requests for the installments to be stopped for
12 months. Calculate the revised installment when the borrower starts
to pay back again, so that the loan period remains unchanged. What
is the difference in the interest paid due to the temporary stoppage of
installments?
Solution: The loan is to be repaid over 15 × 12 = 180 payments. After

20 payments, the loan still has 160 installments to be paid. Using the
prospective method, the balance of the loan after 20 payments is
2,000 × a160e0.005 = $219,910.
9
Note that if we calculate the loan balance using the retrospective method,
we need to compute the original loan amount. The full calculation using
the retrospective method is
2,000 a180e0.005 (1.005)20 − 2,000 s20e0.005 = 2,000 (130.934 − 20.979)

= $219,910.
Due to the delay in payments, the loan balance 12 months after the 20th
payment is
219,910 (1.005)12 = $233,473,
which has to be repaid with a 148-payment annuity-immediate. Hence,
the revised installment is
233,473 233,473
=
a148e0.005 104.401
= $2,236.31.
10
The difference in the interest paid is
2,236.31 × 148 − 2,000 × 160 = $10,973.
Example 5.3: A man borrows a housing loan of $500,000 from Bank

A to be repaid by monthly installments over 20 years at nominal rate of
interest of 4% per year. After 24 installments Bank B offers the man a loan
at rate of interest of 3.5% to be repaid over the same period. However,
if the man wants to re-finance the loan he has to pay Bank A a penalty
equal to 1.5% of the outstanding balance. If there are no other re-financing
costs, should the man re-finance the loan?
Solution: The monthly installment paid to Bank A is

500,000 500,000
=
a240e0.04/12 165.02
11
= $3,029.94.
The outstanding balance after paying the 24th installment is
3,029.94 a216e0.04/12 = 3,029.94 × 153.80

= $466,004.
If the man re-finances with Bank B, he needs to borrow
466,004 × 1.015 = $472,994,
so that the monthly installment is

472,994 472,994
=
a216e0.035/12 160.09
= $2,954.56.
As this is less than the installments of $3,029.94 he pays to Bank A, he

should re-finance. 2
12
5.2 Amortization
• If a loan is repaid by the amortization method, each installment is

first used to offset the interest incurred since the last payment.
• The remaining part of the installment is then used to reduce the

principal.
• Consider a loan to be repaid by a n-payment unit annuity-immediate.

The principal is anei and the interest incurred in the first payment
period is i anei = 1 − v n .
• Thus, out of the unit payment, 1 − v n goes to paying the interest

and the principal is reduced by the amount v n .
13
• The principal is then reduced to
n 1 − vn
anei − v = − vn
i
1 − v n (1 + i)
=
i
1 − v n−1
=
i
= an−1ei .
• Making use of this result we can construct an amortization sched-

ule that separates each installment into the interest and principal
components.
14
Table 5.1: Amortization of a loan of anei by a n-payment unit annuity-
immediate at effective interest rate of i per period
Interest Principal Outstanding

Time Installment payment payment balance
0 ane
1 1 i ane = 1 − v n vn ane − vn = an−1e
· · · · ·
· · · · ·
t 1 i an−t+1e = 1 − vn−t+1 vn−t+1 an−t+1e − v n−t+1 = an−te
· · · · ·
· · · · ·
n 1 i a1e = 1 − v v a1e − v = 0
Total n n − ane ane
15
Example 5.4: Construct an amortization schedule of a loan of $5,000
to be repaid over 6 years with a 6-payment annuity-immediate at effective
rate of interest of 6% per year.
Solution: The annual payment is

5,000 5,000
=
a6e0.06 4.9173
= $1,016.81.
Using Table 5.1 with installments of 1,016.81 in place of 1, we obtain the

amortization schedule in Table 5.2.
16
Table 5.2: Amortization of a loan of $5,000 by a 6-payment annuity-
immediate at effective interest rate of 6% per year
Interest Principal Outstanding

Year Installment payment payment balance
0 5,000.00
1 1,016.81 300.00a 716.81b 4,283.19c
2 1,016.81 256.99d 759.82 3,523.37
3 1,016.81 211.40 805.41 2,717.96
4 1,016.81 163.08 853.73 1,864.23
5 1,016.81 111.85 904.96 959.26
6 1,016.81 57.55 959.26 0.00
Total 6,100.86 1,100.86 5,000.00
The details of the computation in Table 5.2 are given as follows:

a
Interest incurred in the first period: 5,000 × 0.06 = $300.
b
Principal reduction with the first installment: 1,016.81 − 300 = $716.81.
17
c
Outstanding balance at the end of the first year: 5,000 − 716.81 =
$4,283.19.
d
Interest incurred in the second period: 4,283.19 × 0.06 = $256.99.
The rest of the computations follows similarly. 2
18
5.3 Sinking Fund
• A loan may also be served by payment of interest on a periodic basis,

with a lump sum equal to the original principal paid at the end of
the loan period.
• The borrower deposits an amount periodically into a sinking fund

so as to accumulate to the principal.
• Consider a loan of amount anei .
• By the sinking fund method, the borrower pays an amount i anei =

1 − v n each period to serve the interest.
• He deposits installments into a sinking fund. Suppose the sinking

fund accumulates interest at the rate of interest i.
19
• To accumulate to anei at time n, each sinking fund installment is
anei /snei = v n .
• The total amount the borrower has to pay each period is
(1 − v n ) + v n = 1,
which is the same as the amount paid in amortization.
• Table 5.3 presents the sinking fund schedule based on a loan of anei .
20
Table 5.3: Sinking fund schedule of a loan of anei by a n-payment unit
annuity-immediate at effective interest rate of i per period
Interest Sinking fund Sinking fund

Time Installment payment deposit balance
1 1 i ane = 1 − v n vn v n s1e = v n
2 1 i ane = 1 − v n vn v n s2e
· · · · ·
· · · · ·
t 1 i ane = 1 − v n vn v n ste
· · · · ·
· · · · ·
n 1 i ane = 1 − v n vn v n sne = ane
Total n n − nv n nv n
• The interest paid in Table 5.3 remains constant at 1 − v n , while the

interest paid in Table 5.1 using the amortization method decreases
with the time period t.
21
• This is due to the interest earned in the sinking fund. At time t,
the interest earned in the sinking fund over the payment period is
iv n st−1e = v n [(1 + i)t−1 − 1] = v n−t+1 − v n .
• We subtract this amount from the interest served to obtain the net
interest paid as
1 − v n − (v n−t+1 − v n ) = 1 − v n−t+1 ,
which is the same as the value in Table 5.1.
Example 5.5: Construct a sinking fund schedule for the loan in Ex-
ample 5.4.
Solution: The interest payment is 5,000 × 0.06 = $300, and the sinking
fund deposit is
5,000
= $716.81.
s6e0.06
22
Hence, the total annual installment is 300 + 716.81 = $1,016.81. The
sinking fund schedule is presented in Table 5.4.
Table 5.4: Sinking fund schedule of a loan of $5,000 by a 6-payment annuity-

immediate at effective interest rate of 6% per year
Interest Sinking fund Sinking fund Sinking fund

Year Installment payment deposit interest balance
1 1,016.81 300.00 716.81 0.00 716.81
2 1,016.81 300.00 716.81 43.01 1,476.63
3 1,016.81 300.00 716.81 88.60 2,282.04
4 1,016.81 300.00 716.81 136.92 3,135.77
5 1,016.81 300.00 716.81 188.15 4,040.73
6 1,016.81 300.00 716.81 242.44 5,000.00
Total 6,100.86 1,800.00 4,300.86 699.14
• Throughout the 6 years the sinking fund accumulates interest of

$699.14, which is the difference between the interest paid under
23
the sinking fund method and the interest paid using amortization,
namely, 1,800 − 1,100.86. 2
• We now consider the case where the sinking fund earns a rate of
interest different from that charged to the loan.
• Suppose the sinking fund earns interest at the rate of j per payment
period. For a loan of amount anei , to accumulate to the principal in
the sinking fund over n periods the periodic installment is
anei
.
snej
• Hence, the total installment is

Ã !
anei 1
i anei + = anei i + . (5.4)
snej snej
24
• When i = j, the sinking fund method and the amortization method
are the same, so that the monthly payment in (5.4) is equal to 1.
Hence, we have
1 1
i+ = . (5.5)
snei anei
• In (5.5), the right-hand side is the n-payment annuity-immediate to
pay back a loan of 1 by amortization, and the left-hand side states
that the annuity consists of payment of i for the interest and 1/snei
for the sinking fund.
• From (5.4) we can see that a loan of unit amount can be redeemed
over n periods by a sinking fund that charges rate of interest i to
the loan and credits rate of interest j to the sinking fund by periodic
payment of amount
1
i+ .
snej
25
• If this amount is used to redeem a loan of unit amount by the amor-
tization method at the rate of interest i∗ , then we have
1 1
=i+ . (5.6)
anei∗ snej
• We interpret i∗ as the interest rate for the amortization method that

is equivalent to the sinking fund method charging interest rate i to
the loan and crediting interest rate j to the sinking fund.
• If i > j, then snei > snej and we have

Ã !
1
1 = anei∗ i +
snej
Ã !
1
> anei∗ i +
snei
anei∗
= ,
anei
26
which implies anei > anei∗ . From this we conclude i∗ > i.
• Thus, if the interest charged to the loan is higher than the interest
credited to the sinking fund, the equivalent interest rate of the loan
redeemed by amortization is higher.
Example 5.6: A 5-year loan of $500 is to be repaid by a 5-payment

annuity-immediate using the sinking fund method. The loan charges 6%
interest and the sinking fund credits 4% interest. What is the annual
installment? If this installment is used to pay back a loan of the same
amount by amortization, what is the rate of interest?
Solution: The installment is

500
500 × 0.06 + = $122.31.
s5e0.04
We solve for i∗ in the equation
27
122.31a5ei∗ = 500
to obtain i∗ ≈ 7.11%. 2
Example 5.7: A loan is to be repaid by an 8-payment annual annuity-

immediate of $200 at interest of 5% for both the loan charge and the
sinking fund credit. What is the interest component of the 4th payment
in the amortization schedule? What is the interest credited to the sinking
fund at the end of year 5?
Solution: In the amortization schedule, the loan balance after the third
payment is
200a5e0.05 = $865.90.
28
Hence, in amortization the interest in the 4th payment is
865.90 × 0.05 = $43.30.
For the sinking fund method, we first calculate the loan amount, which is
200a8e0.05 = $1,292.64.
Thus, the sinking fund payment is
1,292.64
= $135.37.
s8e0.05
In the beginning of year 5, the sinking fund balance is
135.37s4e0.05 = $583.46,
so that the interest credited to the sinking fund at the end of year 5 is
583.46 × 0.05 = $29.17.
2
29
• We now consider a fund in which the interest on the sinking fund
deposits earns a reinvestment rate of interest different from the
sinking fund deposits.
• We assume the installments paid to the sinking fund generate inter-

est at the rate j, while the interest received is reinvested at the rate
of interest j 0 .
• To calculate the accumulated amount, we first consider the case of

a unit payment to the sinking fund at time 0. Interest of amount j
is paid at time 1, 2, · · ·, n. These interest payments earn interest at
the rate j 0 .
• The time diagram of the fund is given in Figure 5.1.
• The accumulated amount at time n is made up of the unit-payment
30
at time 0 and the n-payment annuity-immediate of amount j which
earns interest at rate j 0 . Thus, the accumulated value is 1 + jsnej 0 .
• If j = j 0 , this expression reduces to (1 + j)n .
• If an n-payment annuity-immediate of unit amount is paid to the

sinking fund, the accumulated value of this annuity is
Ã !
snej 0 − n
n + j(Is)n−1ej 0 =n+j . (5.9)
j0
Example 5.8: A loan of $1,000 at effective rate of interest of 6% per

year is to be redeemed over 5 years by a sinking fund that credits interest
at 5.5% with reinvestment rate of interest of 5%. Calculate the annual
installments of the annuity-immediate required to redeem the loan. What
is the equivalent rate of interest using the amortization method?
31
Solution: The interest payment per year is 1,000×0.06 = $60. Let X be
the annuity payments to the sinking fund, which, after 5 years accumulate
to
" Ã !#
h i s5e0.05 − 5
X 5 + 0.055(Is)4e0.05 = X 5 + 0.055
0.05
= 5.5782X.
Setting this to 1,000, we obtain X = 179.27. Hence, the annual installment

is 60+179.27 = $239.27. If this payment redeems the load by amortization,
the rate of interest i∗ solves the equation
239.27a5ei∗ = 1,000,
from which we obtain i∗ ≈ 6.29%. 2
32
5.4 Varying Installments and Varying Interest Rates
• Installments to repay a loan need not be level. Suppose a loan over n

periods at rate of interest i is to be paid back by a varying n-payment
annuity-immediate A1 , · · · , An . The loan amount L is given by
n
X
L= At v t . (5.10)
t=1
• The balance of the loan Bk−1 after the (k − 1)th payment is

n
X
Bk−1 = At v t−k+1 .
t=k
• Thus, the interest in the kth payment is iBk−1 and the principal is
reduced by the amount Ak − iBk−1 .
33
Example 5.9: A 20-year loan is to be repaid by installments of $100 at
the end of year 1, $200 at the end of year 2, and so on, increasing by $100
each year, up to $600 at the end of the sixth year. From then onwards,
a level installment of $600 will be paid. The rate of interest charged is
6%. Calculate the loan amount. What is the interest and the principal
reduction in the 3rd and the 12th payments?
Solution: The time diagram of the repayment is shown in Figure 5.3.

The loan amount L is
" #
−6
5.2124 − 6(1.06)
100(Ia)6e + 600v 6 a14e = 100 × + 3,931.56
0.06
= $5,569.22.
To calculate the balance after the second installment, we use the retro-
spective method to obtain
B2 = 5,569.22(1.06)2 − 100(1.06) − 200 = $5,951.57.
34
Note that the loan amount has actually increased, as the initial payments
are not sufficient to cover the interest. This is the case of a negative
amortization. The interest in the third payment is
0.06 × 5,951.57 = $357.09.
The third installment of $300 is not sufficient to pay for the interest, and
there is an increase of loan balance by $57.09 to $6,008.66.
To split the 12th installment into interest and principal, we first calculate
the loan balance after the 11th payment using the prospective method,
which is
B11 = 600a9e
= $4,081.02.
Thus, the interest in the 12th payment is 4,081.02 × 0.06 = $244.86, so

that the principal reduction is 600 − 244.86 = $355.14. 2
35
• For the sinking fund method in which the loan charges rate of interest
i and the sinking fund credits interest at the rate j, the interest in
each installment is iL.
• If the sinking fund contribution at time t is St for t = 1, · · · , n, then

n
X
St (1 + j)n−t = L, (5.11)
t=1
and the tth installment is
At = St + iL.
Example 5.10: Suppose a 20-year loan of $1,000 at rate of interest

6% is to be redeemed by the sinking fund method which credits sinking
fund deposits at 5.5%. If the sinking fund deposit increases by 5% in each
36
payment, what is the amount of the sixth installment of this loan? What
is the total amount of installments to repay this loan?
Solution: Let S be the first sinking fund deposit. The tth sinking
fund deposit is S(1.05)t−1 for t = 1, · · · , 20. As the deposits accumulate
to $1,000 in 20 years, we have
20
X
1,000 = S(1.05)t−1 (1.055)20−t
t=1
20 µ ¶
S(1.055)20 X 1.05 t
=
1.05 t=1 1.055
20
X
= 2.7788S (0.9953)t
t=1
= 52.91S,
37
from which we solve for S = $18.90. Thus, the 6th installment is
1,000 × 0.06 + 18.90(1.05)5 = $84.12.
The total amount paid is

19
X
1,000 × 0.06 × 20 + 18.90 (1.05)t = $1,824.95.
t=0
For comparison, the total payments using the amortization method is

1,000
20 × = $1,743.69.
a20e0.06
2
Example 5.11: A man borrows a housing loan of $300,000 from a bank

for 20 years, to be paid back by monthly installments. In the first year
the bank charges an interest rate of 2.5%, followed by the second year of
38
3%. From then onwards, the interest rate will be 5%. All rates quoted
are nominal per year. The bank calculates installments as if the loan is
to be repaid over the remaining term at the ongoing rate. Thus, the first
year installment assumes rate of interest of 2.5% for 20 years.
(a) Calculate the installments in the first, second and third year.
(b) What are the interests paid in the first and the second year?
Solution: (a) The first-year payment is

300,000 300,000
=
a240e0.025/12 188.71
= $1,589.71.
Using the prospective method, the loan balance after 12 payments is
1,589.71a228e0.025/12 = $288,290
39
so that the revised installment is
288,290 288,290
=
a228e0.03/12 173.63
= $1,660.38.
Repeating the same calculations, the balance after 24 payments is

1,660.38 a216e0.03/12 = $276,858. The revised installment is
276,858
= $1,946.41.
a216e0.05/12
For (b), the total amount of payments in the first year is 12 × 1,589.71 =
$19,076.52, and the reduction in principal is 300,000 − 288,290 = $11,710.
Thus, the interest served is 19,076−11,710 = $7,366. Likewise, the interest
served in the second year is 12×1,660.38−(288,290−276,858) = $8,492.56.
2
40
5.5 Comparison of Borrowing Costs
• The cost of a loan depends on the quoted rate of interest and the
way the interest is calculated.
• For example, the balance of a loan may be computed on a periodic

rest basis, where the period is specified in the loan agreement.
• This means that interest is charged on the outstanding balance at the

beginning of the period for the whole period, although installments
may be made prior to the end of the period.
• To illustrate, we consider an annual-rest versus a monthly-rest loan.
• An annual-rest loan charges interest for the whole year, even though
installments may be made monthly throughout the year.
41
Example 5.12: A man is borrowing a housing loan of $500,000 from
a bank. The quoted rate of interest is 6.5% per annum. Calculate the
monthly installments in arrears if the loan is based on (a) annual rest,
and (b) monthly rest. Compare the results for a 15-year loan and a 20-
year loan.
Solution: For the annual-rest loan, we first calculate the annual pay-
ment, which, for the 15-year loan, is
500,000 500,000
=
a15e0.065 9.4027
= $53,176.22.
The monthly installment is

53,176.22
= $4,431.35.
12
42
For the monthly-rest loan, the monthly installment is
500,000 500,000
=
a180e0.065/12 114.796
= $4,355.55.
For the 20-year loan, the monthly installments for the annual-rest and
monthly-rest loans are, respectively, $3,781.52 and $3,727.87. 2
Example 5.13: A bank offers two options of housing loans. Option A

is annual-rest at a quoted rate of 6.25% per annum. Option B is monthly-
rest at 6.5% per annum. Both options require repayment by monthly
installments in arrears. If a 10-year loan is required, which option should
be preferred? What is your answer for a 20-year loan?
Solution: We consider an arbitrary loan amount of $500,000. For a 10-

year loan, the monthly installments of the annual-rest and monthly-rest
43
loans are, respectively, $5,728.40 and $5,677.40. Thus, the monthly-rest
loan is preferred. For a 20-year loan, the installments for the annual-rest
and monthly-rest loans are, respectively, $3,706.75 and 3,727.85. Thus,
the annual rest loan is preferred. 2
• When loans are paid back by installments of the same payment in-
terval, they can be compared by converting the quoted rates to an
equivalent nominal rate with compounding frequency equal to the
payment frequency.
• We can convert the quoted rate of interest for the loan on annual
rest to an equivalent nominal rate (ENR) of interest based
on monthly rest.
• Consider a principal of amount L. For a n-year annual-rest loan at

quoted rate r to be repaid by monthly installments, the installment
44
amount is
L
. (5.12)
12aner
• We denote the annualized ENR on monthly-rest by r∗ , so that the

monthly installment
L
(5.23)
a12ner∗ /12
is equal to (5.12). This implies
12aner = a12ner∗ /12 , (5.14)
from which we can solve for r∗ numerically.
Example 5.14: For the loans in Example 5.13, calculate the ENR of
the annual-rest loan for a 10-year and 20-year loan.
45
Solution: For the 10-year loan, the ENR is 6.70%. This can be checked
by verifying the equation
12a10e0.0625 = a120e0.0670/12 .
Similarly, we can verify that the ENR of the 20-year annual-rest loan is
6.43%. Thus, the ENR of the 10-year annual-rest loan is higher than the
quoted rate of the monthly-rest loan, which shows that the monthly-rest
loan is preferred. On the other hand, the ENR of the 20-year annual-rest
loan is lower than the quoted rate of the monthly rest loan, which means
that the annual-rest loan should be preferred. This is consistent with the
answer in Example 5.13. 2
• It should be noted that the ENR r∗ defined in (5.14) is actually r(12)

as defined in Chapter 1.
46
• If the installment is made m times a year, we can modify (5.14) to
define the ENR accordingly, and we shall denote it simply by r(m) .
• Note that the ENR is a nominal rate; it is not an effective rate as

defined in Chapter 1.
47
5.6 Flat Rate Loan and Flat Rate Discount Loan
• A flat-rate loan charges interest on the original principal for the

whole term of the loan, even though the loan is progressively repaid
by periodic installments.
• For a n-year loan of amount L charged on a quoted flat rate of

interest of r per annum, the monthly installment A is computed by
L(1 + rn)
A= . (5.15)
12n
• To compute the ENR of this loan, we solve for r(12) from the equation
Aa12ner(12) /12 = L,
or, equivalently,
12n
a12ner(12) /12 = . (5.16)
1 + rn
48
Example 5.15: A bank charges a car loan at 3% per annum flat. If the
loan period is 5 years, and the repayments are by monthly installments in
arrears, what is the ENR? What is the effective rate of interest?
Solution: Using (5.16), we solve for r(12) in

12 × 5
a60er(12) /12 = = 52.1739
1 + 0.03 × 5
to obtain r(12) = 5.64%. The effective rate of interest is
µ ¶
0.0564 12
1+ − 1 = 5.79%.
12
2
• If a loan is to be repaid by monthly installments in advance, i.e., by

an annuity-due, to solve for the ENR, we modify (5.16) to
12n
ä12ner(12) /12 = . (5.17)
1 + rn
49
• A loan with such repayments is called a flat-rate discount loan.
Example 5.16: Solve the problem in Example 5.15, assuming now the
monthly installments are to be paid in advance.
Solution: We solve for r(12) /12 from
ä60er(12) /12 = 52.1739,
to obtain r(12) /12 = 0.004876. Thus the ENR is r(12) = 5.84% and the
effective rate of interest is (1.004876)12 −1 = 6.00%. Hence we can see that
the effective cost of the payments-in-advance loan is about 21 basis points
higher than that of the payments-in-arrears loan, although the quoted
rates of interest are the same. 2
Example 5.17: A car buyer is offered two options of car loan. Option
A charges interest at 4% flat with monthly installments paid in arrears.
50
Option B charges 3.75% flat with monthly installments paid in advance. A
loan of $50,000 is required, and the man is considering whether to borrow
for 2 years or 5 years. Which option should he prefer?
Solution: For a 2-year loan, Option A requires installments of
50,000(1 + 2 × 0.04)
= $2,250.00,
2 × 12
while Option B requires
50,000(1 + 2 × 0.0375)
= $2,239.58.
2 × 12
Similarly, the installments for the 5-year loans of Option A and Option B
are, respectively, $1,000.00 and $989.58. As Option B has a lower quoted
rate of interest, its monthly repayment is always lower than that of Option
A with the same loan period.
51
We solve the ENR of the loans using (5.16) and (5.17). The ENR of Option
A for the 2- and 5-year loans are, respectively, 7.5% and 7.42%. Similarly,
the ENR of Option B for the 2- and 5-year loans are, respectively, 7.68%
and 7.23%. Thus, based on the ENR, Option A is preferred if a 2-year
loan is required, while Option B is preferred if a 5-year loan is required.
In contrast, if we compare the loans by the monthly installments, we will
erroneously conclude that Option B is preferred due to its lower monthly
payment. 2
• When the payment frequencies of different loans are not the same,
the ENR is not a valid measure to compare the loans’ actual costs.
The appropriate measure is the effective rate of interest computed
from the ENR. Below is an example.
52
Example 5.18: A bank offers a loan of $100,000 for 2 years under
three different options. Option A charges 6.75% per annum on monthly
rest, with repayment installments made monthly. Option B charges 5.50%
per annum on annual rest, with repayment installments made quarterly.
Option C charges interest flat at 3.5% in year 1 and flat at 4.00% in year
2. The principal is repaid by equal monthly installments, and the interest
in each year is repaid by equal monthly installments in that year. If all
payments are made in arrears, which option is preferred?
Solution: Option A is straightforward. The effective rate of interest is

∙ ¸12
0.0675
1+ − 1 = 6.96%.
12
For Option B, the quarterly installment is
100,000
= $13,540.59,
4a2e0.055
53
and we solve for r∗ from the equation
13,540.59 a8er∗ = 100,000
to obtain r∗ = 0.01812. Hence, the ENR in quarterly rest is 4 × 0.01812 =

7.25%. The effective rate of interest for Option B is
(1.01812)4 − 1 = 7.45%.
For Option C, the monthly installment for the principal is

100,000
= $4,166.67.
24
Therefore, the total monthly installment for the first year is
100,000 × 0.035
4,166.67 + = $4,458.34
12
54
while that for the second year is
100,000 × 0.040
4,166.67 + = $4,500.00.
12
We solve for r∗ from the equation
4,458.34 a12er∗ + 4,500.00 a12er∗ (1 + r∗ )−12 = 100,000
to obtain r∗ = 0.005856, so that the ENR in monthly rest is 12×0.005856 =

7.03%. The effective rate of interest of Option C is
(1.005856)12 − 1 = 7.26%.
Hence, Option A has the lowest effective rate of interest, and Option B
has the highest effective rate of interest. 2
55
5.7 Borrowing Costs and Reference Rates
• The reference rates are required by regulators in communicating the

borrowing costs to the clients.
• The term Annual Percentage Rate (APR) is commonly used in

Canada, Hong Kong, UK and US.
• The equivalence of the APR in the academic literature is r(m) , and

in particular, for m = 12. Australia uses the term Comparison
Rate, which takes account of most fee costs.
• In Singapore, the term used is the Effective Interest Rate (EIR),

which is actually the Equivalent Nominal Rate defined in Section
5.5. The loan balance in early exit is determined by the so-called
Rule of 78.
56
• For a n-year loan of amount L with k installments made, the Rule
of 78 determines the balance B after the kth payment as
B = Principal + Interest — Installments Paid — Interest Rebate,
where Interest is the total interest charged over the whole term of
the loan, and the Interest Rebate, denoted by R, is
(12n − k)(12n − k + 1)Lrn

R= . (5.18)
12n(12n + 1)
• Hence, B is given by
B = L(1 + rn) − Ak − R, (5.19)
where A is the monthly installment given in (5.15).
57
Example 5.19: A 6-year car loan charges flat rate at 2.5% per annum,
to be repaid by a monthly annuity-due. Calculate the ENR in monthly
rest of the loan. If the borrower decides to redeem early after 12, 24, 36,
48 and 60 payments, calculate the ENR in monthly rest charged over these
periods.
Solution: The ENR in monthly rest is the value of r∗ that satisfies the
equation (see (5.17)):
72
ä12ner(12) /12 = = 62.6087,
1 + 6 × 0.025
from which we obtain r∗ = 4.853%. If the loan is redeemed after k pay-
ments, the ENR in monthly rest is r∗ that solves
⎡ ⎤
⎢k−1 ⎥
⎢X A ⎥ B
L=⎢ µ ¶ ⎥+ ,
⎣
t=0 r∗ t ⎦ µ r ∗ ¶k−1
1+ 1+
12 12
58
where A and B are given in (5.15) and (5.19), respectively. Solving the
above equation for various values of k we obtain the following results:
k 12 24 36 48 60
r∗ (%) 5.369 5.085 4.972 4.906 4.867
If the loan is redeemed at maturity, the annual effective rate of interest is

(1 + 0.04853/12)12 − 1 = 4.9624%. However, if the loan is redeemed after
1 year, the effective rate of interest is (1 + 0.05369/12)12 − 1 = 5.5503%.
In the case the loan involves a further penalty for early redemption, the
difference in the borrowing costs may be even higher. 2
59
for Actuaries
Chapter 6
Bonds and Bond Pricing
1
Learning Objectives
1. Types, features and risks of bond investments
2. Formulas for pricing a bond
3. Construction of bond amortization schedules
4. Pricing a bond between two coupon-payment dates
5. Callable bonds and their pricing approaches
6. Price of a bond under a nonflat term structure
2
6.1 Basic Concepts
• A bond is a contract/certificate of debt in which the issuer promises

to pay the holder a definite sequence of interest payments for a spec-
ified period of time, and to repay the loan at a specified terminal
date (called the maturity or redemption date).
• There are many risks involved in investing in bonds, including
Interest-rate risk: Bond investors receive interest payments periodically

before the bond redemption date. Coupon interest payments are usually
constant during the life of the bond. After the investor purchased the
bond, if the prevailing interest rate rises, the price of the bond will fall as
the bond is less attractive. This is called interest-rate risk.
3
Default risk: Default risk is the risk that the bond issuer is unable
to make interest payments and/or redemption repayment. Based on the
bond issuer’s financial strength, bond rating agencies provide a rating (a
measure of the quality and safety) of the bond. Bonds are classified by
rating agencies into two categories: investment-grade bonds (a safer
class) and junk bonds (a riskier class).
Reinvestment risk: For coupon-paying bonds, investors receive inter-

est payments periodically before the redemption date. The reinvestment
of these interest payments (sometimes referred to as interest-on-interest)
depends on the prevailing interest-rate level at the time of reinvestment.
Zero-coupon bonds do not have reinvestment risk.
Call risk: The issuer of a callable bond has the right to redeem the
bond prior to its maturity date at a preset call price under certain con-
ditions. These conditions are specified at the bond issue date, and are
4
known to the investors. The issuer will typically consider calling a bond
if it is paying a higher coupon rate than the current market interest rate.
Callable bonds often carry a call protection provision. It specifies a
period of time during which the bond cannot be called.
Inflation risk: Inflation risk arises because of the uncertainty in the

real value (i.e., purchasing power) of the cash flows from a bond due to
inflation. Inflation-indexed bonds are popular among investors who do
not wish to bear inflation risk.
Other risks in investing in bonds include: market risk, which is the risk
that the bond market as a whole declines; event risk, which arises when
some specific events occur; liquidity risk, which is the risk that investors
may have difficulty finding a counterparty to trade.
5
6.2 Bond Evaluation
• The following features of a bond are often agreed upon at the issue
date.
Face Value: Face value, denoted by F , also known as par or principal

value, is the amount printed on the bond.
Redemption Value: A bond’s redemption value or maturity value,

denoted by C, is the amount that the issuer promises to pay on the re-
demption date. In most cases the redemption value is the same as the face
value.
Time to Maturity: Time to Maturity refers to the length of time before

the redemption value is repaid to the investor.
6
Coupon Rate: The coupon rate, denoted by r, is the rate at which
the bond pays interest on its face value at regular time intervals until the
redemption date.
• We only consider the financial mathematics of default-free bonds.
• We denote n as the number of coupon payment periods from the

date of purchase (or the settlement date) to the maturity date,
P as the purchase price, and i as the market rate of interest (called
the yield rate).
• The yield rate reflects the current market conditions, and is deter-
mined by the market forces, giving investors a fair compensation in
bearing the risks of investing in the bond.
• We assume that r and i are measured per coupon-payment period.
7
Thus, for semiannual coupon bonds, r and i are the rate of interest
per half-year.
• The price of a bond is the sum of the present values of all coupon
payments plus the present value of the redemption value due at
maturity.
• We assume that a coupon has just been paid, and we are interested
in pricing the bond after this payment. Figure 6.1 illustrates the
cash-flow pattern of a typical coupon bond.
• We shall assume that the term structure is flat, so that cash flows
at all times are discounted at the same yield rate i.
• Thus, the fair price P of the bond is given by the following basic
price formula
P = (F r)ane + Cv n , (6.1)
8
where the interest and annuity functions are calculated at the yield
rate i.
Example 6.1: The following shows the results of a government bond

auction:
Type of Bond Government bond
Issue Date February 15, 2010
Maturity Date February 15, 2040
Coupon Rate 4.500% payable semiannually
Yield Rate 4.530% convertible semiannually
Assume that the redemption value of the bond is the same as the face
value, which is $100. Find the price of the bond.
Solution: This is a 30-year bond and we can use the basic price formula
(6.1) with
F = 100,
9
C = 100,
0.045
r = = 0.0225,
2
0.0453
i = = 0.02265,
2
n = 60,
to calculate the price of the bond, which is
P = (F r)ane + Cv n
= 2.25a60e0.02265 + 100(1.02265)−60
= $99.51.
Example 6.2: The following shows the information of a government

bond traded in the secondary market.
10
Type of Bond Government bond
Issue Date June 15, 2005
Date of Purchase June 15, 2009
Maturity Date June 15, 2020
Coupon Rate 4.2% payable semiannually
Yield Rate 4.0% convertible semiannually
Assume that the redemption value of the bond is the same as the face
value, which is $100. Find the purchase price of the bond immediately
after its 8th coupon payment on June 15, 2009.
Solution: This is a 15-year bond at issue but there are only 22 remaining
coupon-payment periods after its 8th coupon payment on June 15, 2009.
We use the basic price formula (6.1) with
F = 100,
C = 100,
11
0.042
r = = 0.021,
2
0.040
i = = 0.020,
2
n = 22,
to obtain
P = (F r)ane + Cv n
= 2.1a22e0.02 + 100(1.02)−22
= $101.77.
Example 6.4: Show that

g
P = K + (C − K),
i
12
where K = Cv n is the present value of the redemption payment and
g = F r/C is the modified coupon rate.
Solution: We start with (6.1) to obtain
P = (F r)ane + Cv n
µ ¶
1 − vn
= (F r) +K
i
µ ¶
1 − vn
= Cg +K
i
g
= (C − Cv n ) + K
i
g
= (C − K) + K. (6.2)
i
This formula is called the Makeham formula. 2
13
6.3 Bond Amortization Schedule
• From the basic bond price formula, we have:
P = (F r)ane + Cv n
= (F r)ane + C(1 − iane )
= C + (F r − Ci)ane , (6.3)
which is called the premium/discount formula because
P − C = (F r − Ci)ane (6.4)
represents the bond premium (when it is positive) or the bond
discount (when it is negative).
• In other words, if the selling price of a bond is larger than its re-
demption value, the bond is said to be traded at a premium of value
P − C = (F r − Ci)ane .
14
• On the other hand, if the selling price of a bond is less than its
redemption value, the bond is said to be traded at a discount of
amount C − P = (Ci − F r)ane .
• In most cases the redemption value is the same as the face value
(i.e., C = F ), so that the bond is traded at
Premium: P − F = F (r − i)ane , if r > i,
Par: P − F = 0, if r = i,
Discount: F − P = F (i − r)ane , if i > r.
• In the bond premium situation the bondholder pays more than the
face value. The premium represents an amount that the investor
will not receive at maturity.
• This paid-in-advance premium (which is an asset for the bondholder)

will be refunded (amortized) periodically from the coupon payments
over the life of the bond.
15
• We consider the effective interest method for the amortization
of a bond.
• As an illustrative example, we consider a $1,000 face value (same as

the redemption value) 3-year bond with semiannual coupons at the
rate of 5% per annum. The current required rate of return i for the
bond is 4% convertible semiannually. The price of the bond is
P = (1,000 × 0.025)a6e0.02 + 1,000(1.02)−6
= $1,028.01,
and the premium is $28.01.
• In each half-year period a $25 coupon payment is received. This

amount is larger than the interest earned at the yield rate based
on the book value, which is $1,028.01 × 0.02 = $20.56 for the first
half-year.
16
• The remaining part of the coupon payment is then used to reduce the
unamortized premium. We can construct a bond premium amorti-
zation schedule based on this principle. The result is shown in Table
6.1.
Table 6.1: A bond premium amortization schedule
Half- Coupon Effective Amortized amount Book

year payment interest earned of premium value
0 1,028.01a
1 25.00b 20.56c 4.44d 1,023.57e
2 25.00 20.47 4.53 1,019.04
3 25.00 20.38 4.62 1,014.42
4 25.00 20.29 4.71 1,009.71
5 25.00 20.19 4.81 1,004.90
6 25.00 20.10 4.90 1,000.00
Total 150.00 121.99 28.01
17
The details of the computation are as follows:
a
The purchase price is the beginning balance of the book value.
b
The coupon payment is 1,000(0.05/2) = $25.00.
c
Effective interest earned in the first half-year is 1,028.01(0.04/2) =
$20.56.
d
Amortized amount of premium in the first half-year is 25.00 − 20.56 =
$4.44.
e
The book value after amortization is 1,028.01 — 4.44 = $1,023.57.
• The bond amortization schedule is useful to bondholders for account-

ing and taxation purposes. In some countries the bond interest in-
come is taxable. In the above example, however, the $25 coupon
payment received each period should not be fully taxable.
• In the bond discount situation the bondholder purchases the bond
18
for less than the face value. The discount represents an amount that
must be paid by the issuer to the investor at the time of maturity.
• The discount (a form of pre-paid interest) will be amortized period-

ically from the coupon payments over the life of the bond.
• We revisit the illustrative example. For the $1,000 face value 3-

year bond with semiannual coupons at the rate of 5% per annum,
we now assume that the required rate of return is 6% convertible
semiannually. The purchase price of this bond is $972.91, and the
discount is $27.09.
• We construct a bond discount amortization schedule in Table 6.2.

The computational steps follow closely those of Table 6.1.
19
Table 6.2: A bond discount amortization schedule
Half- Coupon Effective Amortized amount Book

year payment interest earned of discount value
0 972.91
1 25.00 29.19 —4.19 977.10
2 25.00 29.31 —4.31 981.41
3 25.00 29.44 —4.44 985.86
4 25.00 29.58 —4.58 990.43
5 25.00 29.71 —4.71 995.15
6 25.00 29.85 —4.85 1,000.00
Total 150.00 177.09 —27.09
From equation (6.4), we obtain

Premium: P − C = C(g − i)ane , if g > i,
Discount: C − P = C(i − g)ane , if i > g,
where g = (F r)/C is the modified coupon rate as defined in Example 6.4.
20
• A general bond amortization schedule is given in Table 6.3.
Table 6.3: Formulas for a general bond amortization schedule
Half- Coupon Effective Amortized Book

year payment interest earned amount value
0 P = C + C(g − i)ane
1 Cg i[C + C(g − i)ane ] C(g − i)v n C + C(g − i)an−1e
2 Cg i[C + C(g − i)an−1e ] C(g − i)v n−1 C + C(g − i)an−2e
· · · · ·
· · · · ·
t Cg i[C + C(g − i)an−t+1e ] C(g − i)v n−t+1 C + C(g − i)an−te
· · · · ·
· · · · ·
n Cg i[C + C(g − i)a1e ] C(g − i)v C + C(g − i)a0e = C
Total nCg nCg − C(g − i)ane C(g − i)ane
21
Example 6.5: Find the price of a $1,000 face value 15-year bond with
redemption value of $1,080 and coupon rate of 4.32% payable semiannu-
ally. The bond is bought to yield 5.00% convertible semiannually. Show
the first 5 entries, as well as entries of the 20th and 30th half-year periods,
of its bond amortization schedule.
Solution: Using the bond price formula (6.3), we have
P = C + (F r − Ci)ane
= 1,080 + (21.6 − 27.0)a30e0.025
= $966.98.
Entries of the amortization schedule, using formulas in Table 6.3, can be

computed as in Table 6.4.
22
Table 6.4: Amortization schedule for Example 6.5
Half- Coupon Effective Amortized Book

year payment interest earned amount value
0 966.98
1 21.60 24.17 −2.57 969.55
2 21.60 24.24 −2.64 972.19
3 21.60 24.30 −2.70 974.89
4 21.60 24.37 −2.77 977.67
· · · · ·
20 21.60a 25.72b −4.12c 1,032.74d
· · · · ·
30 21.60 26.87 −5.27 1,080.00
23
The details of the computation are as follows:
a
The modified coupon rate is g = F r/C = 2% per half-year period.
Coupon payment is Cg = 1,080(0.02) = $21.60, and it is the same as
F r = 1,000(0.0216) = $21.60, where r is 0.0432/2 = 2.16% per half-year
period.
b
Effective interest earned in the 20th half-year is 0.025[1,080+1,080(0.02−
0.025)a11e0.025 ] = $25.72.
c
Amortized amount in the 20th half-year is 1,080(0.02−0.025)(1.025)−11 =
−$4.12. A negative amortized amount represents a discount situation. Al-
ternatively, this is equal to 21.60 — 25.72.
d
The ending balance (after the 20th coupon payments) of the book value
is 1,080 + 1,080(0.02 − 0.025)a10e0.025 = $1,032.74.
The rest of the computations follows similarly.
2
24
6.4 Valuation between Coupon-Payment Dates
• The pricing formulas discussed are applicable to a bond at its issue

date or at a date immediately after a coupon payment.
• Bond transactions may occur any time before maturity.
• We now consider the pricing formula of a bond traded between

coupon-payment dates.
• In practice the market price of a bond is stated as a percentage (e.g.,

90.125, or 90.125%) of its face value, which is called the quoted
price (or the clean price).
• When a bond is purchased between the coupon-payment dates, inter-

est is earned by the seller of the bond from the last coupon-payment
date, which is referred to as the accrued interest.
25
• The purchaser of the bond has to pay the accrued interest to the
seller since the seller will not be entitled to any amount of the next
coupon payment.
• The accrued interest is added to the quoted price to determine the

purchase price (also called the dirty price or the invoice price),
i.e.,
Purchase price = quoted price + accrued interest. (6.5)
• Let us denote k as the time immediately after the kth coupon pay-
ment.
• We consider a bond traded in between two coupon-payment dates

(say, between time k and k + 1) at time k + t, where t measures the
length of time since the last coupon payment as a fraction of the
time between k and k + 1.
26
• While there are different market practices in defining t, we adopt
the actual/actual day count convention and define t as
Number of days since time k

t= , (6.6)
Number of days between k and k + 1
so that 0 < t < 1.
• We denote the bond price at time k and k + 1 by Pk and Pk+1 ,

respectively.
• These prices are the bond values after the coupon payments. For
an investor who purchased the bond after the coupon payment at
time k and held it till time k + 1, the value of her investment in the
period is illustrated in Figure 6.2.
• If the yield rate remains unchanged from time k to k + 1, we have
27
the following relationship
Pk+t = Pk (1 + i)t = (Pk+1 + F r)(1 + i)−(1−t) , (6.7)
where i is the yield rate in the coupon-payment period.
• We decompose Pk+t into two components. In common practice, the

simple-interest method is used to calculate the accrued interest at
time k + t, denoted by Ik+t .
• Using the simple-interest formula, we have
Ik+t = t(F r). (6.8)
The book value (quoted price) at time t, denoted by Bk+t , is then given
by
Bk+t = Pk+t − Ik+t . (6.9)
28
Example 6.6: Assume that the coupon dates for the government
bond in Example 6.2 are June 15 and December 15 of each year. On
August 18, 2009, an investor purchased the bond to yield 3.8% convertible
semiannually. Find the purchase price, accrued interest and the quoted
price of the bond at the date of purchase, based on a face value of 100.
Solution: We have i = 0.038/2 = 1.9%, so that the values of Pk and

Pk+1 are
Pk = 2.1a22e0.019 + 100(1.019)−22 = 103.5690,

Pk+1 = 2.1a21e0.019 + 100(1.019)−21 = 103.4368.
The number of days between June 15, 2009 and August 18, 2009 is 64
and the number of days between the two coupon dates in 2009 is 183.
Therefore t = 64/183. Hence,
64
Pk+t = Pk (1 + i)t = 103.5690(1.019) 183 = 104.2529,
29
or equivalently,
1−t − 119
Pk+t = v (Pk+1 + F r) = (1.019) 183 (103.4368 + 2.1) = 104.2529.
The accrued interest is
Ik+t = t(F r) = 0.7344,
and the quoted price (book value) is:
Bk+t = Pk+t − Ik+t = 104.2529 − 0.7344 = 103.5185.
• The Excel function PRICE can be used to compute the quoted price
of a bond between coupon-payment dates. Its specification is as
follows:
30
Excel function: PRICE(smt,mty,crt,yld,rdv,frq,basis)
smt = settlement date

mty = maturity date
crt = coupon rate of interest per annum
yld = annual yield rate
rdv = redemption value per 100 face value
frq = number of coupon payments per year
basis = day count, 30/360 if omitted (or set to 0) and actual/actual if set to 1
Output = quoted price of bond at settlement date per 100 face value
• The default for the input “basis” is the 30/360 convention. Under
this convention, every month has 30 days and every year has 360
days.
31
Exhibit 6.1: Excel solution of Example 6.6
6.5 Callable Bonds
• Callable bonds are bonds that can be redeemed by the issuer prior
to the bonds’ maturity date.
• Investors whose bonds are called are paid a specified call price,
which was fixed at the issue date of the bond.
• The call price may be the bond’s face value, or it may be a price
somewhat higher. The difference between the call price and the face
value is called the call premium.
• If a bond is called between coupon dates, the issuer must pay the
investor accrued interest in addition to the call price.
• Investors of callable bonds often require a higher yield to compensate

for the call risk as compared to non-callable bonds.
32
• Some callable bonds offer a call protection period. The issuer is
not allowed to call the bond before the ending date of the protection
period. The first call date is the date after which the bond is fully
callable.
• Theoretically, there is an optimal call date for the issuer to maximize

the call benefit.
• However, in addition to the call benefit, there are many other factors
(such as transaction costs, possible negative impact on the compa-
ny’s reputation and competition, etc.) affecting the decision of the
issuer to make a call. Therefore, it is difficult to predict when the
issuer will call.
• Here we consider a defensive pricing approach for the investor. Un-

der this approach, an investor assumes that the issuer will call the
33
bond at a date which will maximize the call benefit.
• If a callable bond is traded at a discount and the call price is fixed

at the bond’s redemption value, the optimal call date for the issuer
is at maturity.
• On the other hand, if a callable bond is traded at a premium and

the call price is fixed at the bond’s redemption value, the optimal
call date for the issuer is the first call date.
• In general, when the call price is not a constant but varies in a pre-
fixed relation with the possible call dates, it is not easy to determine
the issuer’s optimal call date.
• In this situation, we can apply the strategy in which the investor

pays the lowest price among the prices calculated by assuming all
possible call dates.
34
Example 6.7: Assume that the bond in Table 6.1 is callable and
the first call date is the date immediately after the 4th coupon payment.
Find the price of the bond for the yield to be at least 4% compounded
semiannually.
Solution: Treating the first call date as the maturity date of the bond,
the price of the bond is 25a4e0.02 + 1,000(1.02)−4 = $1,019.04. The prices
of the bond assuming other possible call dates, namely, after the 5th and
6th coupon payments, are, respectively,
25a5e0.02 + 1,000(1.02)−5 = $1,023.57,
and
25a6e0.02 + 1,000(1.02)−6 = $1,028.01.
Thus, the price of the bond is the lowest among all possible values, i.e.,
$1,019.04. 2
35
Example 6.8: Consider a $1,000 face value 15-year bond with coupon
rate of 4.0% convertible semiannually. The bond is callable and the first
call date is the date immediately after the 15th coupon payment. Assume
that the issuer will only call the bond at a date immediately after the nth
coupon (15 ≤ n ≤ 30) and the call price (i.e., redemption value) is
(
1,000, if 15 ≤ n ≤ 20,
C=
1,000 + 10(n − 20), if 20 < n ≤ 30.
Find the price of the bond if the investor wants to achieve a yield of at
least 5% compounded semiannually.
Solution: First, we compute the prices of the bond assuming all possible
call dates, with r = 0.02, i = 0.025, F = 1,000 and the value of C following
the given call price formula. The results are given in Table 6.5.
36
Table 6.5: Results for Example 6.8
n Redemption value C Price

15 1,000 938.09
16 1,000 934.72
17 1,000 931.44
18 1,000 928.23
19 1,000 925.11
20 1,000 922.05
21 1,010 925.03
22 1,020 927.79
23 1,030 930.34
24 1,040 932.69
25 1,050 934.85
26 1,060 936.82
27 1,070 938.62
28 1,080 940.25
29 1,090 941.71
30 1,100 943.02
Thus, the minimum price is $922.05, which is the price a defensive investor
is willing to pay for a yield of at least 5% compounded semiannually.
37
6.6 Bond Pricing under a General Term Structure
• The basic price formula in equation (6.1) assumes a flat term struc-
ture such that the yield rate for all cash flows (coupons and redemp-
tion payment) is the same.
• In the case when the term structure is not flat the pricing formula
has to be modified.
• Following the principle that the fair price of the bond is the present
value of the cash flows, we can compute the price of the bond as the
sum of the present values of the coupon and redemption payments.
• We consider a bond with n coupon payments of amounts C1 , · · · , Cn ,

and redemption at time n with redemption value C. Let iSj be the
38
spot rate of interest for cash flows due in j years and r be the rate
of interest per coupon payment.
• For the case of an annual level-coupon bond, the price P is given by

n
X Cj C
P = S j
+ S )n
j=1 (1 + ij ) (1 + in
n
X 1 C
= Fr S j
+ S n
. (6.10)
j=1 (1 + ij ) (1 + in )
• For a semiannual level-coupon bond, the pricing formula is

n
X 1 C
P = Fr ⎛ + ⎛ ⎞n . (6.11)
S ⎞j S
in
j=1 ij
⎝1 + 2 ⎠ ⎝1 + 2 ⎠
2 2
39
Example 6.9: Consider a $100 face value bond with coupon rate of
4.0% per annum paid semiannually. The spot rates of interest in the
market are given by
j (year) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
iSj (%) 3.0 3.0 3.5 3.5 4.0 4.0 4.5 4.5 5.0 5.0
Find the price of the bond if it matures in (a) 3 years, and (b) 5 years.
Solution: For (a), the price of the bond is

∙ ¸
1 1 1 1 1 1 100
P = 2 + 2
+ 3
+ 4
+ 5
+ 6
+
1.015 1.015 1.0175 1.0175 1.02 1.02 1.026
= 100.0608.
Similarly, for (b), the price of the bond is
∙ ¸
1 1 1 1 1 100
P = 2 + + · · · + + +
1.015 1.0152 1.02258 1.0259 1.02510 1.02510
= 95.9328.
40
Note that we have a premium bond for the case of (a), with the price
higher than the redemption value. This is due to the fact that the spot
rates of interest are less than the coupon rate of interest for cash flows of
maturity less than 3 years.
On the other hand, for the case of (b) we have a discount bond, with the
price lower than the redemption value. As the spot rates of interest for
cash flows of maturity of more than 3 years are higher than the coupon
rate, the present value contributions of the cash flows beyond 3 years are
lower, causing the price of the bond to drop.
Hence, when the term structure is not flat the premium/discount rela-
tionship of a bond with respect to its redemption value may vary with the
time to maturity even for bonds with the same coupon rate of interest. 2
41
for Actuaries
Chapter 7
Bond Yields and the Term Structure
1
Learning Objectives
1. Yield to maturity, yield to call and par yield
2. Realized compound yield and horizon analysis
3. Estimation of the yield curve: bootstrap method and least squares

method
4. Estimation of the instantaneous forward rate and the term structure
5. Models of the determination of the term structure
2
7.1 Some Simple Measures of Bond Yield
• Current yield is the annual dollar amount of coupon payment(s)

divided by the quoted (clean) price of the bond.
• It is the annual coupon rate of interest divided by the price of the

bond per unit face value.
• The current yield does not adequately reflect the potential gain of
the bond investment.
• The nominal yield is the annual amount of coupon payment(s)

divided by the face value, or simply the coupon rate of interest per
annum. Like the current yield, the nominal yield is not a good
measure of the potential return of the bond.
3
7.2 Yield to Maturity
• Given the yield rate applicable under the prevailing market con-
ditions, we can compute the bond price using one of the pricing
formulas in Chapter 6.
• In practice, however, the yield rate or the term structure are not
observable, while the transaction price of the bond can be observed
from the market.
• Given the transaction price, we can solve for the rate of interest
that equates the discounted future cash flows (coupon payments and
redemption value) to the transaction price.
• This rate of interest is called the yield to maturity (or the yield
to redemption), which is the return on the bond investment if the
4
investor holds the bond until it matures, assuming all the entitled
payments are realized.
• The yield to maturity is indeed the internal rate of return of the

bond investment.
• For a n-year annual coupon bond with transaction price P , the yield
to maturity, denote by iY , is the solution of the following equation:
n
X 1 C
P = Fr j
+ n
. (7.1)
j=1 (1 + iY ) (1 + iY )
• In the case of a n-year semiannual coupon bond, we solve iY from

the equation (the coupon rate r is now per half-year)
2n
X 1 C
P = Fr µ ¶j + µ ¶2n . (7.2)
j=1 iY iY
1+ 1+
2 2
5
• To calculate the solutions of equations (7.1) and (7.2) numerical
methods are required. The Excel Solver may be used for the com-
putation.
Example 7.1: A $1,000 par value 10-year bond with redemption value
of $1,080 and coupon rate of 8% payable semiannually is purchased by an
investor at the price of $980. Find the yield to maturity of the bond.
Solution: The cash flows of the bond in this example are plotted in
Figure 7.1. We solve for i from the following equation:
980 = 40a20ei + 1,080(1 + i)−20
to obtain i = 4.41%, so that the yield to maturity iY is 8.82% per annum

convertible semiannually. 2
6
Example 7.2: The following shows the information of a bond traded
in the secondary market:
Type of Bond Non-callable government bond

Issue Date March 10, 2002
Maturity Date March 10, 2012
Face Value $100
Redemption Value $100
Coupon Rate 4% payable semiannually
Assume that the coupon dates of the bond are March 10 and September
10 of each year. Investor A bought the bond on the issue date at a price
of $105.25. On January 5, 2010, Investor B purchased the bond from A
at a purchase (dirty) price of $104.75. Find (a) the yield to maturity of
Investor A’s purchase on March 10, 2002, (b) the realized yield to Investor
A on the sale of the bond, and (c) the yield to maturity of Investor B’s
purchase on January 5, 2010.
7
Solution: (a) We need to solve for i in the basic price formula
P = (F r)anei + Cv n ,
which, on the issue date, is
105.25 = 2a20ei + 100(1 + i)−20 .
The numerical solution of i is 1.69% (per half-year). Thus, the yield to

maturity is 3.38% per annum convertible semiannually.
(b) Investor A received the 15th coupon payment on September 10, 2009
and there are 117 days between September 10, 2009 and the sale date of
January 5, 2010. The number of days between the two coupon payments,
namely, September 10, 2009 and March 10, 2010 is 181. Thus, we need to
compute i in the following equation:
117
105.25 = 2a15ei + 104.75(1 + i)−15 181 ,
8
the solution of which is 1.805% (per half-year). Thus, the realized yield is
3.61% per annum convertible semiannually.
(c) Investor B will receive the next 5 coupon payments starting on March
10, 2010. There are 64 days between the purchase date and the next
coupon date. Investor B’s return i per half-year is the solution of the
following equation:
h i 64
−4
104.75 = 2ä5ei + 100(1 + i) (1 + i)− 181
h i 117
−5
= 2a5ei + 100(1 + i) (1 + i) 181 ,
which is 1.18%. Thus, the yield to maturity is 2.36% per annum convert-
ible semiannually. 2
• If a bond is callable prior to its maturity, a commonly quoted mea-

sure is the yield to call, which is computed in the same way as the
yield to maturity, with the following modifications:
9
— the redemption value in equations (7.1) and (7.2) is replaced
by the call price,
— the maturity date is replaced by the call date.
• An investor will be able to compute a schedule of the yield to call

as a function of the call date (which also determines the call price),
and assess her investment over the range of possible yields.
• Although the Excel Solver can be used to calculate the solution of

equations (7.1) and (7.2), the computation can be more easily done
using the Excel function YIELD, the specification of which is given
as follows:
10
Excel function: YIELD(smt,mty,crt,prc,rdv,frq,basis)

mty = maturity date
prc = quoted (clean) price of the bond per 100 face value
rdv = redemption value per 100 face value
basis = day count, 30/360 if omitted and actual/actual if set to 1
Output = yield to maturity of the bond
• Exhibit 7.1 illustrates the use of the function YIELD to solve Example
7.2.
• We first enter the dates March 10, 2002, January 5, 2010 and March
10, 2012 into Cells A1 through A3, respectively.
11
• In Cell A4, the YIELD function is entered, with settlement date A1,
maturity date A3 and price of bond 105.25.
• The output is the yield to maturity of Investor A’s purchase on

March 10, 2002, i.e., 3.38%, which is the answer to Part (a).
• Part (b) cannot be solved by the YIELD function, as the bond was
not sold on a coupon-payment date.
• For Part (c), note that the input price of the bond required in the
YIELD function is the quoted price.
• To use the YIELD function, we first compute the quoted price of the
bond on January 5, 2010, which is 104.75 − 2 × 117
181
= 103.4572.
• The answer to Part (c) is shown in Cell A5 to be 2.36%.
12
7.3 Par Yield
• Given the prevailing spot-rate curve and that bonds are priced ac-
cording to the existing term structure, the yield to maturity iY is
solved from the following equation (for an annual coupon bond):
n
X Xn
1 C 1 C
Fr j
+ n
= Fr ³ ´j + S n , (7.3)
j=1 (1 + iY ) (1 + iY ) j=1 1 + iSj
(1 + in )
which is obtained from equations (6.10) and (7.1).
• Hence, iY is a nonlinear function averaging the spot rates iSj , j =

1, · · · , n.
• As an averaging measure of the spot rates, however, iY has some

disadvantages.
13
— First, no analytic solution of iY exists.
— Second, iY varies with the coupon rate of interest, even for
bonds with the same maturity.
• To overcome these difficulties, the par yield may be used, which is

defined as the coupon rate of interest such that the bond is traded
at par based on the prevailing term structure.
• Denoting the par yield by iP and setting F = C = 100, we have

n
X 1 100
100 = 100 iP ³ ´j + S n, (7.4)
j=1 1 + iS (1 + in )
j

³ ´−n
1 − 1 + iSn
iP = P ³ ´−j . (7.5)
n S
j=1 1 + ij
14
• More generally, if the bond makes level coupon payments at time
t1 , · · · , tn , and the term structure is defined by the accumulation
function a(·), the par yield is given by
1 − [a(tn )]−1
iP = Pn −1
j=1 [a(tj )]
1 − v(tn )
= Pn . (7.6)
j=1 v(tj )
• Table 7.2 illustrates the par yields computed from two different spot-
rate curves: an upward sloping curve and a downward sloping curve.
15
Table 7.2: Par yields of two term structures
Case 1 Case 2
n iSn iP iSn iP
1 3.50 3.50 6.00 6.00
2 3.80 3.79 5.70 5.71
3 4.10 4.08 5.40 5.42
4 4.40 4.37 5.10 5.14
5 4.70 4.64 4.80 4.86
6 5.00 4.91 4.50 4.58
7 5.30 5.18 4.20 4.30
8 5.60 5.43 3.90 4.02
9 5.90 5.67 3.60 3.74
10 6.20 5.91 3.30 3.46
11 6.50 6.13 3.00 3.18
12 6.80 6.34 2.70 2.89
16
• The par yield is more a summary measure of the existing term struc-
ture than a measure of the potential return of a bond investment.
• On the other hand, the yield to maturity is an ex ante measure of

the return of a bond.
• If the bond is sold before it matures, or if the interest-on-interest

is different from the current yield rate, the ex post return of the bond
will be different.
• We now consider the evaluation of the return of a bond investment

taking account of the possibility of varying interest rates prior to
redemption as well as sale of the bond before maturity.
17
7.4 Holding-Period Yield
• For various reasons, investors often sell their bonds before maturity.
• The holding-period yield, also called the realized compound

yield or the total return, is often computed on an ex post basis
to evaluate the average return of the investment over the holding
period of the bond.
• This methodology can also be applied to assess the possible return of

the bond investment over a targeted horizon under different interest-
rate scenarios.
• This application, called horizon analysis, is a useful tool for active

bond management.
18
• We denote the holding-period yield of a bond by iH .
• We first discuss the simple case of a one-period holding yield.
• Suppose a bond is purchased at time t − 1 for Pt−1 . At the end of

the period the bondholder receives a coupon of F r and then sells
the bond for Pt .
• The holding yield over the period t − 1 to t, denoted by iH , is then

given by
(Pt + F r) − Pt−1
iH = . (7.7)
Pt−1
Example 7.4: A $1,000 face value 3-year bond with semiannual coupons
at 5% per annum is traded at a yield of 4% per annum convertible semi-
annually. If interest rate remains unchanged in the next 3 years, find the
holding-period yield at the third half-year period.
19
Solution: Using the basic price formula, the prices of the bond after the
second and third coupon payments are, respectively, P2 = 1,019.04 and
P3 = 1,014.42. Therefore, the holding-period yield for the third half-year
period is
(1,014.42 + 25.00) − 1,019.04
iH = = 2%.
1,019.04
2
• Example 7.4 illustrates that when the yield rate is unchanged after a
bond investment until it is sold prior to maturity, the holding-period
yield in that period is equal to the yield to maturity.
• However, when the prevailing yield rate fluctuates, so will the bond’s
holding-period yield.
• The computation of the one-period holding yield can be extended to

multiple periods.
20
• To calculate the n-period holding yield over n coupon-payment peri-
ods, we need to know the interest earned by the coupons when they
are paid.
• Let P0 be the beginning price of the bond, Pn be the ending price of

the bond and V be the accumulated value of the n coupons at time
n.
• The n-period holding yield iH is then the solution of the equation
P0 (1 + iH )n = Pn + V, (7.8)
so that the annualized n-period holding yield is

∙ ¸ n1
Pn + V
iH = − 1. (7.9)
P0
21
Example 7.5: Consider a $1,000 face value 5-year non-callable bond
with annual coupons of 5%. An investor bought the bond at its issue date
at a price of $980. After receiving the third coupon payment, the investor
immediately sold the bond for $1,050. The investor deposited all coupon
income in a savings account earning an effective rate of 2% per annum.
Find the annualized holding-period yield iH of the investor.
Solution: We have P0 = 980 and P3 = 1,050. The accumulated

interest-on-interest of the coupons, V , is given by
V = (1,000 × 0.05)s3e0.02 = 50 × 3.0604 = 153.02.
Thus, the annualized 3-period holding yield iH is the solution of
980(1 + iH )3 = 1,050 + 153.02 = 1,203.02,
22
which implies
∙ ¸1
1,203.02 3
iH = − 1 = 7.0736%.
980
2
Example 7.6: Consider two bonds, A and B. Bond A is a 10-year 2%

annual-coupon bond, and Bond B is a 3-year 4% annual-coupon bond.
The current spot-rate curve is flat at 3%, and is expected to remain flat
for the next 3 years. A fund manager assumes two scenarios of interest-
rate movements. In Scenario 1, the spot rate increases by 0.25 percentage
point each year for 3 years. In Scenario 2, the spot rate drops to 2% next
year and remains unchanged for 2 years. If the manager has an investment
horizon of 3 years, what is her recommended strategy under each scenario?
You may assume that all coupons and their interests are reinvested to earn
the prevailing one-year spot rate.
23
Solution: We first compute the current bond prices. For Bond A, the
current price is
2a10e0.03 + 100(1.03)−10 = 91.4698,
and for Bond B, its current price is
4a3e0.03 + 100(1.03)−3 = 102.8286.
Under Scenario 1, the price of Bond A after 3 years is
2a7e0.0375 + 100(1.0375)−7 = 89.3987,
and the accumulated value of the coupons is 2 × (1.0325 × 1.035 + 1.035 +

1) = 6.2073. Thus, the holding-period yield of Bond A over the 3 years is
∙ ¸1
89.3987 + 6.2073 3
− 1 = 1.4851%.
91.4698
24
Under Scenario 2, Bond A is traded at par in year 3 with the accumulated
value of the coupons being 2 × (1.02 × 1.02 + 1.02 + 1) = 6.1208. Thus,
the holding-period yield is
∙ ¸1
100 + 6.1208 3
− 1 = 5.0770%.
91.4698
On the other hand, Bond B matures in 3 years so that its holding-period
yield under Scenario 1 is
" #1
100 + 4 × (1.0325 × 1.035 + 1.035 + 1) 3
− 1 = 3.0156%,
102.8286
while under Scenario 2 its holding-period yield is
" #1
100 + 4 × (1.02 × 1.02 + 1.02 + 1) 3
− 1 = 2.9627%.
102.8286
Thus, Bond B is the preferred investment under Scenario 1, while Bond
A is preferred under Scenario 2. 2
25
7.5 Discretely Compounded Yield Curve
• We have so far used the prevailing term structure to price a bond or

compute the net present value of a project, assuming the spot-rate
curve is given.
• In practice, spot rates of interest are not directly observable in the

market, although they can be estimated from the observed bond
prices.
• We now discuss the estimation of the spot rates of interest, which

are assumed to be compounded over discrete time intervals.
• We consider the estimation of the spot rates of interest convertible

semiannually using a series of semiannual coupon bonds.
26
• The simplest way to estimate the spot-rate curve is by the boot-
strap method. This method requires the bond-price data to follow
a certain format.
• In particular, we assume that the coupon-payment dates of the bonds

are synchronized and spaced out 6 months apart. The following
example illustrates the use of the bootstrap method.
Example 7.7: Table 7.3 summarizes a series of semiannual coupon

bonds with different time to maturity, coupon rate of interest r (in percent
per annum) and price per 100 face value. Using the given bond data,
estimate the spot rate of interest iSt over t years, for t = 0.5, 1, · · · , 6.
27
Table 7.3: Bond price data
Price per
Maturity (yrs) Coupon rate r (%) 100 face value
0.5 0.0 98.41
1.0 4.0 100.79
1.5 3.8 100.95
2.0 4.5 102.66
2.5 2.5 98.53
3.0 5.0 105.30
3.5 3.6 101.38
4.0 3.2 99.83
4.5 4.0 102.83
5.0 3.0 98.17
5.5 3.5 100.11
6.0 3.6 100.24
28
Solution: We first compute the spot rate of interest for payments due in
half-year, which can be obtained from the first bond. Equating the bond
price to the present value of the redemption value (there is no coupon),
we have
100
98.41 = ,
iS0.5
1+
2
which implies
∙ ¸
100
iS0.5 = 2 × − 1 = 2 × 0.016157 = 3.231%.
98.41
For the second bond, there are two cash flows. As a coupon is paid at
time 0.5 year and the spot rate of interest of which has been computed,
we obtain the following equation of value for the bond:
2 102
100.79 = +Ã ! ,
S 2
1.016157 i1
1+
2
29
from which we obtain iS1 = 3.191%. Similarly, for the third bond the
equation of value involves iS0.5 , iS1 and iS1.5 , of which only iS1.5 is unknown
and can be solved from the equation. Thus, for the bond with k semiannual
coupons, price Pk and coupon rate rk per half-year, the equation of value
is
X k
1 100
Pk = 100 rk +⎛ .
⎛ S ⎞j S ⎞k
j=1 ij ik
⎝1 + 2 ⎠ ⎝1 + 2 ⎠
2 2
These equations can be used to solve sequentially for the spot rates. Figure
7.2 plots the spot-rate curve for maturity of up to 6 years for the data given
in Table 7.3. 2
• Although the bootstrap method is simple to use, there are some data
requirements that seriously limit its applicability:
30
1
— the data set of bonds must have synchronized coupon-payment
dates
— there should not be any gap in the series of bonds in the time
to maturity
• We now consider the least squares method, which is less demand-

ing in the data requirement.
• We assume a set of m bonds for which the coupon-payment dates

are synchronized, and denote the prices of these bonds per 100 face
value by Pj , their coupon rate by rj per half-year and their time to
maturity by nj half-years, for j = 1, · · · , m.
31
• We denote
1
vh = ⎛ ⎞h ,
S
ih
⎝1 + 2 ⎠
2
which is the discount factor for payments due in h half-years. Thus,
the equation of value for the jth bond is
nj
X
Pj = 100 rj vh + 100vnj .
h=1
• If the last payment (redemption plus coupon) of the bonds occur in

M periods (i.e., M is the maximum of all nj for j = 1, · · · , m), the
pricing equations of the bonds can be written as
Pj = Cj1 v1 + Cj2 v2 + · · · + CjM vM , j = 1, · · · , m, (7.10)
32
in which Cjh are known cash-flow amounts and vh are the unknown
discount factors.
• Thus, in equation (7.10) we have a multiple linear regression

model with M unknown coefficients v1 , · · · , vM and m observations
(Pj is the dependent variable and Cj1 , · · · , CjM are the indepen-
dent variables, for j = 1, · · · , m).
• We can solve for the values of vh using the least squares method,
and subsequently obtain the values of iSh . 2
33
7.6 Continuously Compounded Yield Curve
• We now introduce the estimation of the term structure assuming

that interest is credited based on continuous compounding.
• We denote the current time by 0, and use iSt to denote the contin-
uously compounded spot rate of interest for payments due at time
t.
• As shown in equation (3.27), if we consider the limiting value of the

forward rate of interest iFt,τ for τ approaching zero, we obtain the
instantaneous forward rate, which is equal to the force of interest,
i.e.,
lim iFt,τ = δ(t).
τ →0
34
• An important method to construct the yield curve based on the
instantaneous forward rate is due to Fama and Bliss (1987).
• If we denote the current price of a zero-coupon bond with unit face

value maturing at time t by P (t), we have
1 h i
S
P (t) = v(t) = = exp −t it , (7.11)
a(t)
and
a0 (t) v 0 (t) P 0 (t)
δ(t) = =− =− . (7.12)
a(t) v(t) P (t)
• If we had a continuously observable sequence of zero-coupon bond
prices P (t) maturing at time t > 0, the problem of recovering the
spot rates of interest is straightforward using equation (7.11).
• In practice, this sequence is not available.
35
• The Fama-Bliss method focuses on the estimation of the instanta-
neous forward rates, from which the spot-rate curve can be com-
puted.
• Recall that ∙Z t ¸
a(t) = exp δ(u) du . (7.13)
0
• From (7.11) and (7.13), we conclude that

h i ∙Z t ¸
exp t iSt = exp δ(u) du ,
0
so that Z t
1
iSt = δ(u) du, (7.14)
t 0
which says that the continuously compounded spot rate of interest
is an equally-weighted average of the force of interest.
36
• Thus, if we have a sequence of estimates of the instantaneous forward
rates, we can compute the spot-rate curve using (7.14).
• To apply the Fama-Bliss method, we assume that we have a sequence

of bonds with possibly irregularly spaced maturity dates. We assume
that the force of interest between two successive maturity dates is
constant.
• Making use of the equations of value for the bonds sequentially in

increasing order of the maturity, we are able to compute the force of
interest over the period of the data. The spot rates of interest can
then be calculated using equation (7.14).
• Although this yield curve may not be smooth, it can be fine tuned
using some spline smoothing method.
37
Example 7.8: You are given the bond data in Table 7.4. The jth bond
matures at time tj with current price Pj per unit face value, which equals the
redemption value, for j = 1, · · · , 4 and 0 < t1 < t2 < t3 < t4 . Bonds 1 and 2
have no coupons, while Bond 3 has two coupons, at time t∗31 with t1 < t∗31 < t2
and at maturity t3 . Bond 4 has three coupons, with coupon dates t∗41 , t∗42 and
t4 , where t2 < t∗41 < t3 and t3 < t∗42 < t4 .
Table 7.4: Bond data for Example 7.8
Coupon per Bond

Bond face value Coupon date Maturity price
1 0 — t1 P1
2 0 — t2 P2
3 C3 t∗31 , t1 < t∗31 < t2 t3 P3
C3 t3
4 C4 t∗41 , t2 < t∗41 < t3 t4 P4
C4 t∗42 , t3 < t∗42 < t4
C4 t4
38
It is assumed that the force of interest δ(t) follows a step function taking
constant values between successive maturities, i.e., δ(t) = δ i for ti−1 ≤
t < ti , i = 1, · · · , 4 with t0 = 0. Estimate δ(t) and compute the spot-rate
curve for maturity up to time t4 .
Solution: As Bond 1 is a zero-coupon bond, its equation of value is

h i ∙ Z t1 ¸
P1 = exp −t1 iSt1 = exp − δ(u) du = exp [−δ 1 t1 ] ,
0
so that
1
δ 1 = − ln P1 ,
t1
which applies to the interval (0, t1 ). Now we turn to Bond 2 (which is
again a zero-coupon bond) and write down its equation of value as
∙ Z t2 ¸
P2 = exp − δ(u) du = exp [−δ 1 t1 − δ 2 (t2 − t1 )] .
0
39
Solving the above equation for δ 2 , we obtain
1
δ2 = − [ln P2 + δ 1 t1 ]
t2 − t1
∙ ¸
1 P2
= − ln .
t2 − t1 P1
For Bond 3, there is a coupon payment of amount C3 at time t∗31 , with
t1 < t∗31 < t2 . Thus, the equation of value for Bond 3 is
∙ Z t3 ¸ " Z t∗ #
31
P3 = (1 + C3 ) exp − δ(u) du + C3 exp − δ(u) du
0 0
= (1 + C3 ) exp [−δ 1 t1 − δ 2 (t2 − t1 ) − δ 3 (t3 − t2 )] + C3 exp [−δ 1 t1 − δ 2 (t∗31 − t1 )] ,

" #
∗
1 P3 − C3 exp[−δ 1 t1 − δ 2 (t31 − t1 )]
δ3 = − ln
t3 − t2 (1 + C3 )P2
" #
∗
1 P3 − P1 C3 exp[−δ 2 (t31 − t1 )]
= − ln .
t3 − t2 (1 + C3 )P2
40
Going through a similar argument, we can write down the equation of
value for Bond 4 as
P4 = (1+C4 )P3∗ exp [−δ 4 (t4 − t3 )]+P2 C4 exp [−δ 3 (t∗41 − t2 )]+P3∗ C4 exp [−δ 4 (t∗42 − t3 )] ,
where
P3∗ = exp [−δ 1 t1 − δ 2 (t2 − t1 ) − δ 3 (t3 − t2 )] ,
so that
P4 − P2 C4 exp [−δ 3 (t∗41 − t2 )]
(1+C4 ) exp [−δ 4 (t4 − t3 )]+C4 exp [−δ 4 (t∗42 − t3 )] = ∗
.
P3
Thus, the right-hand side of the above equation can be computed, while
the left-hand side contains the unknown quantity δ 4 . The equation can be
solved numerically for the force of interest δ 4 in the period (t3 , t4 ). Finally,
41
using (7.14), we obtain the continuously compounded spot interest rate as
⎧
⎪
⎪ δ1, for 0 < t < t1 ,
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ δ 1 t1 + δ 2 (t − t1 )
⎪
⎪
⎪
⎪
⎪
, for t1 ≤ t < t2 ,
⎪
⎨ t
iSt = ⎪
⎪
⎪ δ 1 t1 + δ 2 (t2 − t1 ) + δ 3 (t − t2 )
⎪
⎪
⎪
, for t2 ≤ t < t3 ,
⎪
⎪ t
⎪
⎪
⎪
⎪
⎪
⎪ δ 1 t1 + δ 2 (t2 − t1 ) + δ 3 (t3 − t2 ) + δ 4 (t − t3 )
⎪
⎪
⎩ , for t3 ≤ t < t4 .
t
2
• Thus, the instantaneous forward rate of interest can be computed us-

ing the bond data, from which the spot-rate curve can be computed
by equation (7.14).
42
7.7 Term Structure Models
• Empirically the yield curve can take various shapes.
• It will be interesting to examine the factors determining the shapes

of the yield curve, how the yield curve evolves over time, and whether
the shape of the yield curve has any implications for the real economy
such as the business cycle.
• There are several approaches in defining the term-structure models.

We shall examine the theories of the term structure in terms of the
1-period holding-period yield introduced in Section 7.4.
• For simplicity of exposition, we assume investments in zero-coupon

bonds and relate different measures of returns to the bond prices.
43
1
1
(n)
• We denote iH as the 1-period holding-period (from time 0 to 1)
yield of a bond maturing at time n. The current time is 0 and the
price of a bond at time t with remaining n periods to mature is
denoted by Pt (n).
• We introduce the new notation t iSτ to denote the spot rate of interest
at time t for payments due τ periods from t (i.e., due at time t + τ ).
• Thus, for t > 0, t iSτ is a random variable (at time 0).
• Suppose an investor purchased a n-period bond at time 0 and held

it for 1 period, the bond price at time 1 is P1 (n − 1), so that the
1-period holding-period yield from time 0 to 1 is
(n) P1 (n − 1) − P0 (n)
iH = . (7.15)
P0 (n)
44
(n)
• Note that P1 (n − 1) is a random variable at time 0, and so is iH .
Furthermore, t iSτ is given by
" #1
S 1 τ
t iτ = − 1. (7.16)
Pt (τ )
• If we assume that all market participants are risk neutral (they

neither avoid nor love risks) and that they have no preference for
the maturities of the investments, then their only criterion for the
selection of an investment is its expected return.
• This assumption leads to the pure expectations hypothesis, which

states that the expected 1-period holding-period yields for bonds of
all maturities are equal, and thus equal to the 1-period spot rate of
interest (which is the 1-period holding-period yield of a bond ma-
turing at time 1).
45
(n) (n)
• If we denote E[iH ] as the expected value of iH at time 0, the pure
expectations hypothesis states that
(n)
E[iH ] = iS1 , for n = 1, · · · . (7.17)
• However, while the 1-period holding-period yield for a 1-year zero-

coupon bond is known at time 0 (i.e., iS1 ), it is unknown for bonds
with longer maturities. Thus, the bonds with longer maturities in-
volve uncertainty and may have higher risks.
• Investors may demand higher expected return for bonds with matu-
rities longer than a year, which is called the risk premium.
• If the risk premium is constant for bonds of all maturities, we have

the expectations (or constant premium) hypothesis, which
states that
(n)
E[iH ] = iS1 + , for n = 2, · · · , (7.18)
46
where is a positive constant independent of n.
• However, if investors prefer short-maturity to long-maturity assets

due to their better liquidity, then the expected return for bonds with
longer maturities must be higher to compensate the investors.
• This leads to the liquidity premium hypothesis, which states

that
(n)
E[iH ] = iS1 + (n) , for n = 2, · · · , (7.19)
(n)
where the risk premium for a n-period bond increases with n,
so that (2) ≤ (3) ≤ · · ·.
• Some theorists argue that investors do not necessarily prefer short-

maturity assets to long-maturity assets. Thus, the risk premiums of
bonds with different maturities (n) may not be a monotonic function
47
of n, but may depend on other covariates w(n) , which are possibly
time varying.
• This is called the market segmentation hypothesis, for which

we have
(n)
E[iH ] = iS1 + (w(n) ), for n = 2, · · · , (7.20)
so that the risk premium (·) is a function of the specific asset mar-
ket.
• Finally, the preferred habitat hypothesis states that institutions

do not have a rigid targeted maturity class of assets to invest, but
will be influenced by the returns expected of assets with different
maturities. Thus, bonds with similar maturities will be close substi-
tutes of each other.
• We now examine the implications of the prevailing term structure
48
for the future movements of interest rates under the aforementioned
term-structure models.
(2)
• Consider the 1-period holding-period yield of a 2-period bond, iH ,
which is given by
(2) P1 (1) − P0 (2) P1 (1)
iH = = − 1. (7.21)
P0 (2) P0 (2)
• If we consider the spot rate of interest over the 2-period horizon, iS2 ,
we have
³ ´ 1 P1 (1) 1 ³ ´³ ´
S 2 (2) S
1 + i2 = = × = 1 + iH 1 + 1 i1 . (7.22)
P0 (2) P0 (2) P1 (1)
• Using equation (3.4) to rewrite the left-hand side of the above equa-
tion, we have
³ ´³ ´ ³ ´³ ´
(2) S
1+ iH 1+ 1 i1 = 1+ iS1 1+ iF2 . (7.23)
49
• As rates of interest are small, to the first-order approximation the
above equation can be written as
(2)
iH + 1 iS1 = iS1 + iF2 . (7.24)
• Now we take expectations on both sides of the equation. If we

(n)
adopt the pure expectations hypothesis, we have E[iH ] = iS1 , so
that equation (7.24) implies
E[1 iS1 ] = iF2 , (7.25)
which says that the expected value of the future 1-period spot rate
is equal to the prevailing 1-period forward rate for that period.
• This statement is called the unbiased expectations hypothesis,

which is itself implied by the pure expectations hypothesis. Note
50
that equation (7.25) can be generalized to
E[t iSτ ] = iFt,τ , for 0 < t, τ , (7.26)
so that the prevailing forward rates have important implications for

the expected future values of the spot rates at any time t over any
horizon τ .
• We have seen that if the term structure is upward sloping, the for-
ward rate of interest is higher than the spot rate of interest.
• Thus, under the pure expectations hypothesis, an upward sloping

yield curve implies that the future spot rate is expected to be higher
than the current spot rate. However, if long-term bonds command
a risk premium, an upward sloping yield curve may not imply that
the spot rate is expected to rise.
51
for Actuaries
Chapter 8
Bond Management
Learning Objectives
1. Macaulay duration and modified duration
2. Duration and interest-rate sensitivity
3. Convexity
4. Some rules for duration calculation
5. Asset-liability matching and immunization strategies
6. Target-date immunization and duration matching
7. Redington immunization and full immunization
8. Cases of nonflat term structure
2
8.1 Macaulay Duration and Modified Duration
• Suppose an investor purchases a n-year semiannual coupon bond for

P0 at time 0 and holds it until maturity.
• As the amounts of the payments she receives are different at different

times, one way to summarize the horizon is to consider the weighted
average of the time of the cash flows.
• We use the present values of the cash flows (not their nominal values)
to compute the weights.
• Consider an investment that generates cash flows of amount Ct at

time t = 1, · · · , n, measured in payment periods. Suppose the rate
of interest is i per payment period and the initial investment is P .
3
• We denote the present value of Ct by PV(Ct ), which is given by
Ct
PV(Ct ) = t
. (8.1)
(1 + i)
and we have n
X
P = PV(Ct ). (8.2)
t=1
• Using PV(Ct ) as the factor of proportion, we define the weighted

average of the time of the cash flows, denoted by D, as
n
" #
X PV(Ct )
D = t
t=1 P
Xn
= twt , (8.3)
t=1
where
PV(Ct )
wt = . (8.4)
P
4
P
• As wt ≥ 0 for all t and nt=1 wt = 1, wt are properly defined weights
and D is the weighted average of t = 1, · · · , n.
• We call D the Macaulay duration, which measures the average

period of the investment.
• The value computed from (8.3) gives the Macaulay duration in terms
of the number of payment periods.
• If there are k payments per year and we desire to express the duration
in years, we replace t in (8.3) by t/k. The resulting value of D is
then the Macaulay duration in years.
Example 8.1: Calculate the Macaulay duration of a 4-year annual

coupon bond with 6% coupon and a yield to maturity of 5.5%.
5
Solution: The present values of the cash flows can be calculated us-
ing (8.1) with i = 5.5%. The computation of the Macaulay duration is
presented in Table 8.1.
Table 8.1: Computation for Example 8.1
t Ct PV(Ct ) wt twt
1 6 5.6872 0.0559 0.0559
2 6 5.3907 0.0530 0.1060
3 6 5.1097 0.0502 0.1506
4 106 85.5650 0.8409 3.3636
Total 101.7526 1.0000 3.6761
The price of the bond P is equal to the sum of the third column, namely
101.7526. Note that the entries in the fourth column are all positive and
sum up to 1. The Macaulay duration is the sum of the last column, which
6
is 3.6761 years. Thus, the Macaulay duration of the bond is less than its
time to maturity of 4 years. 2
Example 8.2: Calculate the Macaulay duration of a 2-year semiannual

coupon bond with 4% coupon per annum and a yield to maturity of 4.8%
compounded semiannually.
Solution: The cash flows of the bond occur at time 1, 2, 3 and 4 half-
years. The present values of the cash flows can be calculated using (8.1)
with i = 2.4% per payment period (i.e., half-year). The computation of
the Macaulay duration is presented in Table 8.2.
7
Table 8.2: Computation for Example 8.2
t Ct PV(Ct ) wt twt
1 2 1.953 0.0198 0.0198
2 2 1.907 0.0194 0.0388
3 2 1.863 0.0189 0.0568
4 102 92.768 0.9419 3.7676
Total 98.491 1.0000 3.8830
The price of the bond is equal to the sum of the third column, namely
98.491. The Macaulay duration is the sum of the last column, namely
3.8830 half-years, which again is less than the time to maturity of the
bond of 4 half-years. The Macaulay duration of the bond can also be
stated as 3.8830/2 = 1.9415 years. 2
8
• The formula of the Macaulay duration can be extended to the case
when the cash flows occur at irregular intervals. In such case, i
will be the rate of interest per base period, e.g., per year, and the
discount factor (1 + i)t may be applied to any non-integral value of
t (years).
• The Macaulay duration computed will then be expressed in terms

of the number of base periods (years).
• Consider a bond with face value (also the redemption value) F ,

coupon rate r per payment, and time to maturity of n payment
periods. The rate of interest i applicable to (8.3) is the yield to
maturity per coupon-payment period.
9
• Now Ct is equal to F r for t = 1, · · · , n − 1 and Cn = F r + F . Thus,
from (8.3) we have
" n #
1 XtF r nF
D = +
P t=1 (1 + i)t (1 + i)n
" n
#
1 X
= Fr PV(t) + F nv n . (8.5)
P t=1
• From (6.1) we have P = (F r)ane + F v n . Hence, the Macaulay

duration of the bond is (in terms of the number of payment periods)
Pn
Fr PV(t) + F nv n
t=1
D =
(F r)ane + F v n
r(Ia)ne + n v n
= n
. (8.6)
r ane + v
10
Example 8.3: Calculate the Macaulay duration of the bonds in Exam-
ples 8.1 and 8.2 using equation (8.6).
Solution: For Example 8.1, r = 6%, i = 5.5% and n = 4. Thus,

a4e = 3.5052 and we use (2.36) to obtain (Ia)4e = 8.5285. Now using (8.6)
we have
0.06 × 8.5294 + 4(1.055)−4
D= −4
= 3.6761 years,
0.06 × 3.5052 + (1.055)
which is the answer in Example 8.1. For Example 8.2, we have r = 2%,
i = 2.4% and n = 4. Thus, a4e = 3.7711 and (Ia)4e = 9.3159. Hence, we
have
0.02 × 9.3159 + 4(1.024)−4
D= = 3.8830 half-years.
0.02 × 3.7711 + (1.024)−4
2
11
• While the Macaulay duration was originally proposed to measure
the average horizon of an investment, it turns out that it can be
used to measure the price sensitivity of the investment with respect
to interest-rate changes.
• To measure this sensitivity we consider the derivative dP/di. As

the price of the investment P drops when interest rate i increases,
dP/di < 0.
• We consider (the negative of) the percentage change in the price of

the investment per unit change in the rate of interest, i.e., −(dP/di)/P .
• Using (8.1) and (8.2), this quantity is given by
12
n
1 dP 1 X (−t)Ct
− = −
P di P t=1 (1 + i)t+1
Xn
1 tCt
=
P (1 + i) t=1 (1 + i)t
n
" #
1 X PV(Ct )
= t
1 + i t=1 P
D
= . (8.7)
1+i
• We define
∗ D
D = , (8.8)
1+i
and call it the modified duration, which is always positive and
measures the percentage decrease of the value of the investment per
unit increase in the rate of interest.
13
• Note that in (8.8) i is the rate of interest per payment period, while
the Macaulay duration D can be stated in terms of years or the
number of payment periods.
Example 8.4: Calculate the modified duration of the bonds in Exam-

ples 8.1 and 8.2.
Solution: For Example 8.1, we have
∗ 3.6761
D = = 3.4845 years.
1.055
Thus, the bond drops in value by 3.4845% per 1 percentage point increase
(not percentage increase) in interest rate per year. However, as the bond
price and interest rate relationship is nonlinear, this statement is only
correct approximately and applies to the current rate of interest of 5.5%.
14
For Example 8.2, we have
3.8830
D∗ = = 3.7920 half-years.
1.024
Thus, the bond drops in value by 3.7920% per 1 percentage point increase
in the rate of interest per half-year. 2
15
• In (8.3), we may replace t by kt∗ , in which t∗ denotes the time of the
occurrence of the cash flow in years and k is the number of payments
per year.
• Then we have
" #
1 dP k X ∗ PV(Ct∗ )
− = t
P di 1 + i t∗ P
= kD∗ , (8.9)
where D∗ is the modified duration in years (as t∗ are in years) and

the summation is over all occurrences of cash flows.
• Excel provides the function DURATION to compute the Macaulay du-

ration and the function MDURATION to compute the modified dura-
tion. The bond is assumed to be redeemable at par. The specifica-
tions of these functions are given as follows:
16
Excel functions: DURATION/MDURATION(smt,mty,crt,yld,frq,basis)

mty = maturity date
yld = annualized bond yield
basis = day count, 30/360 if omitted (or set to 0) and actual/actual if set to 1
Output = Macaulay/modified duration of the bond in years
• Exhibit 8.1 illustrates Examples 8.1 through 8.4. We have arbitrarily

fixed the settlement date to be January 1, 2001, and the maturity
date is then entered based on the given time to maturity of the bond.
17
8.2 Duration for Price Correction
• We now consider the use of the modified duration to approximate
the price change of a bond when the rate of interest changes.
• We denote P (i) as the price of a bond when the yield to maturity is

i per coupon-payment period.
• When the rate of interest changes to i +∆i, the bond price is revised
to P (i + ∆i).
• While the bond price can be re-calculated at the rate of interest i+∆i
using one of the pricing formulas in Chapter 6, an approximation is
available using the modified duration.
• For a continuous function f (x) with first- and second-order deriv-

atives, the function evaluated at x + ∆x, i.e., f (x + ∆x), can be
18
approximated by Taylor’s expansion as follows (see Appendix A.8):
df (x) 1 d2 f (x) 2
f (x + ∆x) ≈ f (x) + ∆x + 2
(∆x) .
dx 2 dx
• Thus, if we expand the bond price P (i+∆i) using Taylor’s expansion

up to the first-order derivative, we obtain
dP (i)
P (i + ∆i) ≈ P (i) + ∆i
" di
Ã ! #
1 dP (i)
= P (i) 1 − − ∆i
P (i) di
= P (i) (1 − D∗ ∆i) . (8.10)
• Hence, we can use the modified duration to obtain a linear approx-

imation to the revised bond price with respect to a change in the
rate of interest.
19
• Note that in (8.10), as i is per coupon-payment period, D∗ and ∆i
should also be measured in coupon-payment period.
• However, we may also express D∗ in years, in which case ∆i is the

change in the rate of interest per annum.
Example 8.5: A 10-year semiannual coupon bond with coupon rate of

7% is selling to yield 6.5% per year compounded semiannually. What is
the bond price if the yield changes to (a) 6%, and (b) 6.7%, compounded
semiannually?
Solution: We use the basic formula (6.1) with r = 3.5%, i = 3.25%

and n = 20, to obtain
P (0.0325) = 103.6348.
20
Similarly, we compute the bond price at the new rates of interest i = 3%
and 3.35%, to obtain
P (0.03) = 107.4387
and
P (0.0335) = 102.1611.
Thus, the bond price increases by 3.8039 when the yield per half-year
drops by 0.25 percentage point, and it decreases by 1.4737 when the yield
per half-year increases by 0.1 percentage point.
We may also approximate the price change using the modified duration.
At the rate of interest of 3.25% we have (Ia)20e = 137.306, so that from
(8.6) the Macaulay duration is
0.035 × 137.306 + 20(1.0325)−20
D =
1.036348
= 14.8166 half-years,
21
and hence the modified duration is
14.8166
D∗ = = 14.3502 half-years.
1.0325
The annual yield rate decreases from 6.5% to 6% when ∆i = −0.0025 (per
half-year). Thus, from (8.10), we have
P (0.03) ≈ 103.6348[1 − 14.3502(−0.0025)] = 107.3528 < 107.4387.
Similarly, if the annual yield rate increases from 6.5% to 6.7%, we have
∆i = 0.001 (per half-year), so that
P (0.0335) ≈ 103.6348[1 − 14.3502(0.001)] = 102.1476 < 102.1611.
These results are quite close to the exact results obtained above, although
it can be seen that the approximate values are less than the exact values
in both cases. Thus, using (8.10), approximate values of the price of the
22
bond can be computed with small changes in the interest rate without
using the bond price formulas. 2
• Figure 8.2 illustrates the application of (8.10). The relationship
between the bond price and the rate of interest is given by the curve,
which is convex to the origin.
• Equation (8.10) approximates the bond price using the straight line
which is tangent to the point (i, P (i)) with a negative slope of
−P (i)D∗ .
• Note that due to the convexity of the relationship between the in-
terest rate and the bond price, the correction based on the modified
duration always under-approximates the exact price.
• This point is illustrated by Example 8.5. To improve the approxi-

mation, we may take into account the convexity of the relationship.
23
8.3 Convexity
• To obtain a better approximation for the bond price, we apply Tay-
lor’s expansion in (8.10) to the second order, giving
dP (i) 1 d2 P (i) 2
P (i + ∆i) ≈ P (i) + ∆i + 2
(∆i)
" di
Ã 2 di! Ã ! #
2
1 dP (i) 1 d P (i) 2
= P (i) 1 − − ∆i + 2
(∆i) .
P (i) di 2P (i) di
(8.11)
• Now we define the convexity of the bond as

1 d2 P (i)
C= × 2
, (8.12)
P (i) di
so that (8.11) becomes
∙ ¸
1
P (i + ∆i) ≈ P (i) 1 − D∗ ∆i + C(∆i)2 . (8.13)
2
24
• For the investment with price given in (8.1) and (8.2), we have
n
d2 P (i) X (t + 1)tCt
= , (8.14)
di2 t=1 (1 + i)t+2
so that the convexity is

1 d2 P (i)
C = ×
P (i) di2
n
1 X (t + 1)tCt
=
P (i) t=1 (1 + i)t+2
Xn
1 (t + 1)tCt
=
P (i)(1 + i)2 t=1 (1 + i)t
Xn
1
= 2
(t + 1)t PV(Ct ). (8.15)
P (i)(1 + i) t=1
• For a bond investment, Ct ≥ 0 for all t, so that C > 0, verifying the

convexity relationship.
25
• Thus, the correction term C(∆i)2 /2 in (8.13) is always positive,
which compensates for the under-approximation in (8.10).
Example 8.6: Revisit Example 8.5 and approximate the bond prices
with convexity correction.
Solution: We calculate the convexity using (8.15) to obtain
∙ ¸
1 2 × 1 × 3.5 3 × 2 × 3.5 21 × 20 × 103.5
C = + + · · · +
(103.6348)(1.0325)2 1.0325 (1.0325)2 (1.0325)20
= 260.9566.
Thus, the approximate bond prices are

P (0.03) ≈ 107.3528 + (103.6348)(0.5)(260.9566)(−0.0025)2 = 107.4373,
and
P (0.0335) ≈ 102.1476 + (103.6348)(0.5)(260.9566)(0.001)2 = 102.1612.
26
8.4 Some Rules for Duration
• We summarize some useful rules for duration.
Rule 1: The Macaulay duration D of a bond is always less than or equal

to its time to maturity n. Equality holds only for a zero-coupon bond.
Rule 2: Holding the time to maturity n of a bond constant, when the

coupon rate of interest r decreases, the Macaulay duration D increases.
Rule 3: Other things being equal, when the yield to maturity i decreases,
the Macaulay duration D increases.
Rule 4: For a level perpetuity, the modified duration D∗ is equal to 1/i.
Rule 5: For a level annuity of n payments, the modified duration is

1 n
D∗ = − n
. (8.16)
i (1 + i) [(1 + i) − 1]
27
This rule can be proved by direct differentiation of the price of the annuity,
which is
1 − vn
P (i) = ane = .
i
Rule 6: The modified duration D∗ of a coupon bond with coupon rate of
r per payment, n payments to maturity and yield to maturity of i is
∗1 (1 + i) + n(r − i)
D = − . (8.17)
i (1 + i) [((1 + i)n − 1) r + i]
Rule 7: For a coupon bond selling at par, the modified duration is

" #
∗ 1 1
D = 1− n
. (8.18)
i (1 + i)
Rule 8: Holding other things constant, a bond’s duration D usually in-

creases with its time to maturity n.
28
• For premium and par bonds, the relationship is monotonic so that
D always increases with n.
• For deep-discount bonds, however, D may increase with n for bonds

with short maturity and then decreases with increases in maturity.
Figure 8.3 illustrates this phenomenon, where the prevailing yield
curve is flat at 8%.
• Suppose a portfolio of bonds is constructed from M bonds, with

durations D1 , · · · , DM .
• Let the bond values be P1 , · · · , PM , so that their total is P =

PM
j=1 Pj .
• Define wj = Pj /P as the weight of Bond j in the portfolio, then

the duration DP of the portfolio is the weighted average of the bond
29
18
16
14
Macaulay duration (years)
12
10
4
Coupon rate = 2%
2 Coupon rate = 8%
Coupon rate = 10%
0
0 10 20 30 40 50
Time to maturity (years)
durations, i.e.,
M
X
DP = wj Dj . (8.19)
j=1
• This result is very useful for bond portfolio management when a

portfolio with a certain duration is required.
Example 8.7: A bond manager has a choice of two bonds, A and B.

Bond A is a 4-year annual coupon bond with coupon rate of 6%. Bond B
is a 2-year annual coupon bond with coupon rate of 4%. The current yield
to maturity in the market is 5.5% per annum for all maturities. How does
the manager construct a portfolio of $100 million, consisting of bonds A
and B, with a Macaulay duration of 2.5 years?
Solution: From Example 8.1, we know that Bond A has a duration of

3.6761 years. We compute the duration of Bond B as 1.9610 years. Let w
30
be the proportion of investment in Bond A. Thus, from (8.19) we have
3.6761w + 1.9610(1 − w) = 2.5,
so that
2.5 − 1.9610
w= = 31.43%.
3.6761 − 1.9610
Hence the portfolio should consist of $31.43 million of Bond A and $68.57
million of Bond B. 2
31
8.5 Immunization Strategies
• Financial institutions are often faced with the problem of meeting a

liability of a given amount some time in the future.
• We consider a liability of amount V to be paid T periods later.
• A simple strategy to meet this obligation is to purchase a zero-

coupon bond with face value V , which matures at time T .
• This strategy is called cash-flow matching.
• When cash-flow matching is adopted, the obligation is always met,

even if there is fluctuation in the rate of interest.
• However, zero-coupon bonds of the required maturity may not be

available in the market.
32
• Immunization is a strategy of managing a portfolio of assets such
that the business is immune to interest-rate fluctuations.
• For the simple situation above, the target-date immunization

strategy may be adopted.
• This involves holding a portfolio of bonds that will accumulate in

value to V at time T at the current market rate of interest.
• The portfolio, however, should be constructed in such a way that its

Macaulay duration D is equal to the targeted date of the liability T .
• Suppose the current yield rate is i, the current value of the portfolio
of bonds, denoted by P (i), must be
V
P (i) = T
. (8.20)
(1 + i)
33
• If the interest rate remains unchanged until time T , this bond portfo-
lio will accumulate in value to V at the maturity date of the liability.
• If interest rate increases, the bond portfolio will drop in value. How-
ever, the coupon payments will generate higher interest and com-
pensate for this.
• On the other hand, if interest rate drops, the bond portfolio value
goes up, with subsequent slow-down in accumulation of interest.
• Under either situation, as we shall see, the bond portfolio value will
finally accumulate to V at time T , provided the portfolio’s Macaulay
duration D is equal to T .
• We consider the bond value for a one-time small change in the rate
of interest.
34
• If interest rate changes to i + ∆i immediately after the purchase
of the bond, the bond price becomes P (i + ∆i) which, at time T ,
accumulates to P (i + ∆i)(1 + i + ∆i)T if the rate of interest remains
at i + ∆i.
• We approximate (1 + i + ∆i)T to the first order in ∆i to obtain

(apply Taylor’s expansion to f (i) = (1 + i)T )
(1 + i + ∆i)T ≈ (1 + i)T + T (1 + i)T −1 ∆i. (8.21)
• Using (8.10) and (8.21) we have

h i
T ∗ T T −1
P (i+∆i)(1+i+∆i) ≈ P (i)(1−D ∆i) (1 + i) + T (1 + i) ∆i .
• However, as D∗ = D/(1+i) and T = D, the above equation becomes
35
h i
T D ∗ D D−1
P (i + ∆i)(1 + i + ∆i) ≈ P (i) (1 + i) − D ∆i(1 + i) + D(1 + i) ∆i
= P (i)(1 + i)D
= V. (8.22)
Example 8.8: A company has to pay $100 million 3.6761 years from
now. The current market rate of interest is 5.5%. Demonstrate the funding
strategy the company should adopt with the 6% annual coupon bond in
Example 8.1. Consider the scenarios when there is an immediate one-time
change in interest rate to (a) 5%, and (b) 6%.
Solution: From equation (8.20), the current value of the bond should be
100
= $82.1338 million.
(1.055)3.6761
36
From Example 8.1, the bond price is 101.7526% of the face value and
the Macaulay duration is 3.6761 years, which is the target date for the
payment. Hence, the bond purchased should have a face value of
82.13375
= $80.7191 million.
1.017526
At the end of year 3, the accumulated value of the coupon payments is
80.7191 × 0.06s3e0.055 = $15.3432 million,
and the bond price is (the bond will mature in 1 year with a 6% coupon
payment and redemption payment of 80.7191)
80.7191 × 0.06 + 80.7191
= $81.1017 million.
1.055
Thus, the bond price plus the accumulated coupon values at time 3.6761
years is
(81.1017 + 15.3432)(1.055)0.6761 = $100 million.
37
Suppose interest rate drops to 5% immediately after the purchase of the
bond, the accumulated coupon value 3 years later is
80.7191 × 0.06s3e0.05 = $15.2680 million,
and the bond price at year 3 is

80.7191(1.06)
= $81.4879 million.
1.05
The total of the bond value and the accumulated coupon payments at
time 3.6761 years is
(81.4879 + 15.2680)(1.05)0.6761 = $100 million.
On the other hand, if the interest rate increases to 6% immediately after

the purchase of the bond, the accumulated coupon value 3 years later is
80.7191 × 0.06s3e0.06 = $15.4186 million,
38
and the bond price at year 3 is 80.7191 (this is a par bond with yield
rate equal to coupon rate). Thus, the total of the bond value and the
accumulated coupon payments at time 3.6761 years is
(80.7191 + 15.4186)(1.06)0.6761 = $100 million.
Thus, for an immediate one-time small change in interest rate, the bond
accumulates to the targeted value of $100 million at 3.6761 years, and the
business is immunized. 2
Example 8.9: A company has to pay $100 million 4 years from now.
The current market rate of interest is 5.5%. The company uses the 6%
annual coupon bond in Example 8.1 to fund this liability. Is the bond
sufficient to meet the liability when there is an immediate one-time change
in interest rate to (a) 5%, and (b) 6%?
39
Solution: As the target date of the liability is 4 years and the Macaulay
duration of the bond is 3.6761 years, there is a mismatch in the durations
and the business is not immunized. To fund the liability in 4 years, the
value of the bond purchased at time 0 is
100
4
= $80.7217 million ,
(1.055)
and the face value of the bond is
80.7217
= $79.3313 million.
1.017526
If interest rate drops to 5%, the asset value at year 4 is
79.3313 × 0.06s4e0.05 + 79.3313 = $99.8470 million,
so that the liability is under-funded. On the other hand, if the interest
rate increases to 6%, the asset value at year 4 is
79.3313 × 0.06s4e0.06 + 79.3313 = $100.1539 million,
40
so that the liability is over-funded. 2
• Figure 8.4 describes the working of the target-date immunization

strategy.
• If a financial institution has multiple liability obligations to meet,

the manager may adopt cash-flow matching to each obligation.
• This is a dedication strategy in which the manager selects a port-

folio of bonds (zero-coupon or coupon bonds) to provide total cash
flows in each period to match the required obligations.
• The manager may also consider the liability obligations as a whole

and construct a portfolio to fund these obligations with the objec-
tive of controlling for the interest-rate risk. A commonly adopted
strategy is duration matching.
41
• We assume a financial institution has a stream of liabilities L1 , L2 , · · · , LN
to be paid out at various times in the future.
• It will fund these liabilities with assets generating cash flows A1 , A2 , · · · , AM

at various times in the future.
• We assume that the rate of interest i is flat for cash flows of all
maturities and applies to both assets and liabilities.
• We denote
M
X
PV(assets) = PV(Aj ) = VA , (8.23)
j=1
and
N
X
PV(liabilities) = PV(Lj ) = VL . (8.24)
j=1
42
• We denote the Macaulay durations of the assets and liabilities by
DA and DL , respectively.
• The duration matching strategy involves constructing a portfolio of

assets such that the following conditions hold:
1. VA ≥ VL
2. DA = DL .
• Condition 2 ensures that, to the first-order approximation, the asset-

liability ratio will not drop when interest rate changes. This result
can be deduced as follows:
43
dVA dVL
µ ¶ VL − VA
d VA di di
=
di VL VL2
Ã !
VA 1 dVA 1 dVL
= −
VL VA di VL di
VA
= (DL − DA )
VL (1 + i)
= 0. (8.25)
Example 8.10: A financial institution has to pay $1,000 after 2 years

and $2,000 after 4 years. The current market interest rate is 10%, and the
yield curve is assumed to be flat at any time. The institution wishes to
immunize the interest rate risk by purchasing zero-coupon bonds which
mature after 1, 3 and 5 years. One member in the risk management team
of the institution, Alan, devised the following strategy:
44
• Purchase a 1-year zero-coupon bond with a face value of $44.74,
• Purchase a 3-year zero-coupon bond with a face value of $2,450.83,
• Purchase a 5-year zero-coupon bond with a face value of $500.00.
(a) Find the present value of the liability. (b) Show that Alan’s portfo-
lio satisfies the conditions of the duration matching strategy. (c) Define
surplus S = VA − VL , calculate S when there is an immediate one-time
change of interest rate from 10% to (i) 9%, (ii) 11%, (iii) 15%, (iv) 30%
and (v) 80%. (d) Find the convexity of the portfolio of assets and the
portfolio of liabilities at i = 10%.
Solution: (a) The present value of the liabilities is
VL = 1,000 (1.1)−2 + 2,000 (1.1)−4 = $2,192.47.
45
For (b), the present value of Alan’s asset portfolio is
VA = 44.74 (1.1)−1 + 2,450.83 (1.1)−3 + 500.00 (1.1)−5 = $2,192.47.
The Macaulay duration of the assets and liabilities can be calculated using
equation (8.3) to give
h i
−1 −3 −5
DA = 1 × 44.74 (1.1) + 3 × 2,450.83 (1.1) + 5 × 500.00 (1.1) /2,192.47
= 3.2461 years,
h i
−2 −4
DL = 2 × 1,000 (1.1) + 4 × 2,000 (1.1) /2,192.47
= 3.2461 years.
Since VA = VL and DA = DL , the conditions of the duration matching

strategy are met for Alan’s portfolio.
46
For (c), when there is an immediate one-time shift in interest rate from
10% to 9%, using equation (8.2), we have
VA = 44.74 (1.09)−1 + 2,450.83 (1.09)−3 + 500.00 (1.09)−5 = $2,258.50,

VL = 1,000 (1.09)−2 + 2,000 (1.09)−4 = $2,258.53,
S = 2,258.50 − 2,258.53 = −$0.03.
We repeat the above calculations for interest rate of 11%, 15%, 30% and
80%. The results are summarized as follows:
i VA VL S
0.09 2,258.50 2,258.53 −0.03
0.10 2,192.47 2,192.47 0.00
0.11 2,129.05 2,129.08 −0.03
0.15 1,898.95 1,899.65 −0.70
0.30 1,284.61 1,291.97 −7.36
0.80 471.55 499.16 −27.61
47
For (d), using equation (8.15), the convexity of the assets is
2 × 1 × 44.74 (1.1)−1 + 4 × 3 × 2,450.83 (1.1)−3 + 6 × 5 × 500.00 (1.1)−5
CA =
(1.1)2 × 2,192.47
= 11.87,
and the convexity of the liabilities is

h i
−2 −4
CL = 3 × 2 × 1,000 (1.1) + 5 × 4 × 2,000 (1.1) /[(1.1)2 × 2,192.47]
= 12.17.
• It should be noted that the duration matching strategy is based on

the first-order approximation. To improve the strategy, we may take
into account the convexity of the asset and liability portfolios.
48
• To protect the asset-liability ratio from dropping when interest rate
changes, the Redington immunization strategy, named after the
British actuary Frank Redington, imposes the following three con-
ditions for constructing a portfolio of assets:
1. VA ≥ VL
2. DA = DL
3. CA > CL .
Example 8.11: For the financial institution in Example 8.10, a risk

consultant, Alfred, recommended the following strategy:
49
• Purchase a 5-year zero-coupon bond with a face value of $660.18.
(a) Show that Alfred’s portfolio satisfies the three conditions of the Red-
ington immunization strategy. (b) Define surplus S = VA − VL , calculate
S when there is an immediate one-time change of interest rate from 10%
to (i) 9%, (ii) 11%, (iii) 15%, (iv) 30% and (v) 80%.
Solution: (a) The present value of Alfred’s asset portfolio is
VA = 154.16 (1.1)−1 + 2,186.04 (1.1)−3 + 660.18 (1.1)−5 = $2,192.47.
The Macaulay duration of the assets and liabilities can be calculated using
equation (8.3) to give
h i
−1 −3 −5
DA = 1 × 154.16 (1.1) + 3 × 2,186.04 (1.1) + 5 × 660.18 (1.1) /2,192.47
= 3.2461 years,
50
h i
−2 −4
DL = 2 × 1,000 (1.1) + 4 × 2,000 (1.1) /2,192.47
= 3.2461 years.
Furthermore, using equation (8.15), we get

2 × 1 × 154.16 (1.1)−1 + 4 × 3 × 2,186.04 (1.1)−3 + 6 × 5 × 660.18 (1.1)−5
CA =
(1.1)2 × 2,192.47
= 12.1704,
h i
−2 −4
CL = 3 × 2 × 1,000 (1.1) + 5 × 4 × 2,000 (1.1) /[(1.1)2 × 2,192.47]
= 12.1676.
Since VA = VL , DA = DL and CA > CL , the conditions of the Redington

immunization strategy are met for Alfred’s strategy.
For (b), when there is an immediate one-time shift in interest rate from
51
VA = 154.16 (1.09)−1 + 2,186.04 (1.09)−3 + 660.18 (1.09)−5 = $2,258.53,

VL = 1,000 (1.09)−2 + 2,000 (1.09)−4 = $2,258.53,
S = 2,258.53 − 2,258.53 = 0.
i VA VL S
0.09 2,258.53 2,258.53 0.00
0.10 2,192.47 2,192.47 0.00
0.11 2,129.08 2,129.08 0.00
0.15 1,899.64 1,899.65 −0.02
0.30 1,291.40 1,291.97 −0.57
0.80 495.42 499.16 −3.74
2
52
• Under certain conditions, it is possible to construct a portfolio of
assets such that the net-worth position of the financial institution is
guaranteed to be non-negative in any positive interest rate environ-
ment.
• A full immunization strategy is said to be achieved if under any

one-time shift of interest rate from i0 to i,
S(i) = VA (i) − VL (i) ≥ 0, for i > 0.
• We consider the example of a single liability of amount L to be paid

TL periods later.
53
• Full immunization strategy involves funding the liability by a portfo-
lio of assets which will produce two cash inflows. The first inflow of
amount A1 is located at time T1 , which is ∆1 periods before time TL .
The second inflow of amount A2 is at time T2 , which is ∆2 periods
after time TL .
• Figure 8.5 illustrates these three cashflows. It should be noted that

all the values of i0 , i, A1 , A2 , L, ∆1 , ∆2 , TL , T1 and T2 are positive,
and ∆1 is not necessarily equal to ∆2 .
• In this particular example, the conditions for constructing a portfolio

of assets under the full immunization strategy are:
1. VA = VL
2. DA = DL .
54
Figure 8.5
• The above conditions can be rewritten as
1. A1 (1 + i0 )−T1 + A2 (1 + i0 )−T2 = L(1 + i0 )−TL
2. T1 A1 (1 + i0 )−T1 + T2 A2 (1 + i0 )−T2 = TL L(1 + i0 )−TL .
Example 8.12: For the financial institution in Examples 8.10 and 8.11,
an actuary, Albert, constructed the following strategy:
• Purchase a 5-year zero-coupon bond with a face value of $1,100.00.
(a) Show that Albert’s portfolio satisfies the conditions of the full immu-
nization strategy. (b) Define surplus S = VA − VL , calculate S when there
55
is an immediate one-time change of interest rate from 10% to (i) 9%, (ii)
11%, (iii) 15%, (iv) 30% and (v) 80%.
Solution: (a) The present value of Albert’s asset portfolio is

VA = 454.55 (1.1)−1 + 1,459.09 (1.1)−3 + 1,100.00 (1.1)−5 = $2,192.47.
Let A1 and A2 be the amount of 1-year and 3-year zero-coupon bonds that
are needed to fully immunize the first liability of L = 1,000. Note that
T1 = 1, TL = 2 and T2 = 3. The two conditions for the full immunization
strategy require
A1 (1.1)−1 + A2 (1.1)−3 = 1,000(1.1)−2
(1)A1 (1.1)−1 + (3)A2 (1.1)−3 = (2)1,000(1.1)−2 .
Solving the above system of equations, we get A1 = 454.55 and A2 =
550.00. Next, let A∗1 and A∗2 be the amounts of 3-year and 5-year zero-
coupon bonds that would be needed to fully immunize the second liability
56
L = 2,000. Note that now T1 = 3, TL = 4 and T2 = 5. The two conditions
for the full immunization strategy require
A∗1 (1.1)−3 + A∗2 (1.1)−5 = 2,000(1.1)−4

(3)A∗1 (1.1)−3 + (5)A∗2 (1.1)−5 = (4)1,000(1.1)−4 .
Solving the above system of equations, we get A∗1 = 909.09 and A∗2 =
1, 100.00. The combined asset portfolio consists of a 1-year zero-coupon
bond with a face value of $454.55, a 3-year zero-coupon bond with a face
value of ($550.00 + $909.09) = $1,459.09 and a 5-year zero-coupon bond
with a face value of $1,100.00. This is indeed Albert’s asset portfolio,
which satisfies the full immunization conditions.
For (b), when there is an immediate one-time shift in interest rate from
VA = 454.55 (1.09)−1 + 1,459.09 (1.09)−3 + 1,100.00 (1.09)−5 = $2,258.62,
57
VL = 1,000 (1.09)−2 + 2,000 (1.09)−4 = $2,258.53,
S = 2,258.62 − 2,258.53 = 0.09.
i VA VL S
0.09 2,258.62 2,258.53 0.09
0.10 2,192.47 2,192.47 0.00
0.11 2,129.17 2,129.08 0.09
0.15 1,901.53 1,899.65 1.88
0.30 1,310.04 1,291.97 18.07
0.80 560.93 499.16 61.76
58
• To compare the duration matching, Redington and full immuniza-
tion strategies, the results of Examples 8.10, 8.11 and 8.12 are plot-
ted in Figure 8.6.
59
35
Full Immunization
25
15
5
Surplus
‐5
Redington Immunization
‐15 Duration Matching
‐25
‐35
0% 10% 20% 30% 40% 50% 60% 70% 80%
Interest Rate
8.6 Some Shortcomings of Duration Matching
• In classical duration matching, the term structure is assumed to be

flat.
• This limitation can be relaxed to allow for a term structure that is

not flat, resulting in a more general definition of duration.
• Furthermore, we may consider different changes in the interest rate

depending on the maturity of the cash flow. In other words, we may
allow the shift in the term structure to be non-parallel.
• The characteristics of the assets and liabilities change through time.

Even if the prevailing rate of interest remains unchanged, the du-
rations of the assets and liabilities change over time due to time
decay.
60
• Thus, the portfolio has to be re-balanced periodically to keep the
durations matched.
• We have assumed that the cash flows of the assets and liabilities
are fixed, and there is no uncertainty in their timing and value. For
some cases, however, cash flows may be contingent on some events.
• For cash flows that are contingent on some uncertain events, the tra-
ditional duration measure will not be applicable. Improved methods
such as the option-adjusted duration (also called effective du-
ration or stochastic duration) should be considered.
• Conceptually, the effective duration is defined as the percentage drop

in the price of the asset per unit increase in the rate of interest, i.e.,
−(dP/di)/P .
61
• When the cash flows are contingent on some random events, how-
ever, equation (8.7) cannot be used. The effective duration can be
numerically estimated using the formula
" #
P (i + ∆i) − P (i − ∆i)
− , (8.27)
2(∆i)P (i)
where the prices at perturbed interest rates i + ∆i and i − ∆i are

calculated numerically taking account of the embedded options of
the asset.
62
8.7 Duration under a Nonflat Term Structure
• We consider a nonflat term structure defined by the sequence of spot

rates iSt , for t = 1, · · · , n.
• Denoting i = (iS1 , · · · , iSn )0 as the vector of the spot rates and P (i)
as the price of the asset under the current term structure, we have
n
X Ct
P (i) = S t
. (8.28)
t=1 (1 + it )
• Let ∆ = (∆1 , · · · , ∆n )0 denote the vector of shifts in the spot rates

so that the new term structure is
i + ∆ = (iS1 + ∆1 , · · · , iSn + ∆n )0 , (8.29)
63
and the price of the asset under the new term structure is
n
X Ct
P (i + ∆) = S t. (8.30)
t=1 (1 + it + ∆t )
• If, however, ∆t = ∆ for t = 1, · · · , n, then the term structure has a

parallel shift, and (8.28) becomes
n
X Ct
P (i + ∆) = S t. (8.31)
t=1 (1 + it + ∆)
• Using the first-order approximation in Taylor’s expansion

1 1 t∆
S t ≈ S t
− S t+1 ,
(1 + it + ∆) (1 + it ) (1 + it )
we re-write (8.31) as
n
X Xn
Ct tCt
P (i + ∆) ≈ S t
− ∆ S t+1 ,
t=1 (1 + it ) t=1 (1 + it )
64
which implies
n
X
tCt
P (i + ∆) − P (i) ≈ −∆ S t+1
t=1 (1 + it )
n
" #
X t
= −∆ S
PV(Ct ), (8.32)
t=1 1 + it
where
Ct
PV(Ct ) = S t
. (8.33)
(1 + it )
• Thus, we conclude
" # n
" #
1 P (i + ∆) − P (i) X t PV(Ct )
− lim = S
P (i) ∆→0 ∆ t=1 1 + it P (i)
n
X
= Wt , (8.34)
t=1
65
where " #
t PV(Ct )
Wt = S
.
1 + it P (i)
• The Fisher-Weil duration, denoted by DF , is defined as
n
" n
#
X PV(Ct ) X
DF = t = twt , (8.35)
t=1 P (i) t=1
where
PV(Ct )
wt = .
P (i)
• The Fisher-Weil duration is a generalization of the Macaulay dura-
tion, with the present values of cash flows computed using a nonflat
term structure.
• Equation (8.34) is a generalization of the modified duration of equa-

tion (8.7).
66
Example 8.13: Bond A is a 2-year annual coupon bond with coupon
rate of 3%. Bond B is a 5-year annual coupon bond with coupon rate
of 5.5%. You are given that iSt = 4.2%, 4.2%, 4.5%, 4.7% and 4.8%, for
t = 1, · · · , 5, respectively. Compute the Fisher-Weil duration of the two
bonds, as well as the price sensitivity measure in (8.34). Also, calculate the
Fisher-Weil duration of a portfolio with equal weights in the two bonds.
Solution: For Bond A, we have cash flows C1 = 3 and C2 = 103. Using

the given term structure, we obtain PV(C1 ) = 2.879 and PV(C2 ) = 94.864,
so that the price of Bond A is 2.879 + 94.864 = 97.743. Consequently, we
obtain W1 = 0.028 and W2 = 1.863, so that the price sensitivity measure
of Bond A as given in equation (8.34) is 1.891 (i.e., Bond A drops in value
by 1.891% per 1 percentage point parallel increase in the term structure).
Similarly, the Fisher-Weil duration of the bond is computed as 1.971 years.
Similar calculations can be performed for Bond B, and the results are
67
shown in Table 8.3, with a Fisher-Weil duration of 4.510 years and a price
sensitivity of 4.305 (i.e., Bond B drops in value by 4.305% per 1 percentage
point parallel increase in the term structure). To compute the Fisher-Weil
duration of the portfolio, we take the weighted average of the Fisher-Weil
durations of the bonds to obtain 0.5(1.971 + 4.510) = 3.241 years.
A B
t PV(Ct ) Wt twt PV(Ct ) Wt twt
1 2.879 0.028 0.029 5.278 0.049 0.051
2 94.864 1.863 1.941 5.066 0.094 0.098
3 4.820 0.134 0.140
4 4.577 0.169 0.177
5 83.454 3.858 4.044
Sum 97.743 1.891 1.971 103.194 4.305 4.510
2
68
Example 8.14: Compute the prices of the bonds in Example 8.13
under the following term structures (with the years to maturity of the
bonds remaining unchanged):
t 1 2 3 4 5
iSt of Case 1 4.6% 4.8% 5.5% 6.1% 6.4%
iSt of Case 2 3.7% 3.7% 4.0% 4.2% 4.3%
Comment on the use of equation (8.34) for the price changes of these
bonds.
Solution: We note that the spot rate of interest iSt in Case 1 increases
by the amount of 0.004, 0.006, 0.01, 0.014 and 0.016, for t = 1, 2, · · · , 5,
respectively. Thus, the shift in the term structure is not parallel, but the
average increase is 1 percentage point. Using equation (6.10), the com-
puted prices of Bonds A and B are, respectively, 96.649 and 96.655. Thus,
the price of Bond A drops 1.119% and that of Bond B drops 6.337%. These
69
figures contrast the values of 1.891% and 4.305%, respectively, predicted
by equation (8.34). The discrepancy exists due to the fact that the shift
in the term structure is not parallel.
Case 2 represents a parallel shift in the spot rates of −0.5 percentage point
for all time to maturity. Using equation (6.10), the prices of Bonds A and
B are found to be 98.674 and 105.447, respectively. Thus, Bonds A and
B increase in value by 0.952% and 2.183%, respectively. These values are
quite close to the changes predicted by equation (8.34) (i.e., half of the
tabulated values, namely, 0.946% and 2.153%, respectively). 2
70
8.8 Passive versus Active Bond Management
• A bond fund may adopt a passive or active strategy.
• A passive strategy adopts a nonexpectational approach, without an-

alyzing the likely movements of the market. Immunization, indexing
and buy-and-hold are passive bond management strategies.
• An active bond management strategy may involve some form of

interest-rate forecasting. Example 7.6 illustrates the use of horizon
analysis to enhance the performance of a fund.
• A broader active management framework would take a quantitative

approach in assessing the value of a bond, taking into account all em-
bedded options and structures of the bond, and includes assessment
of the sector of the bond issuer and its credit profile.
71
for Actuaries
Chapter 9
Stochastic Interest Rates
1
Learning Objectives
• Deterministic scenarios of interest rates
• Random-scenario model
• Independent lognormal model
• Autoregressive lognormal model
• Dynamic term structure model
• Illustration of stochastic interest rate approach via an application
2
9.1 Deterministic Scenarios of Interest Rates
• We denote it , for t = 2, 3, · · ·, as the 1-period future spot rate ap-

plicable from time t − 1 to time t. For completeness we denote
i1 = iS1 , which is known at time 0.
• Unlike the forward rates iFt , which are determined by the current
term structure, it for t > 1 is unknown at time 0. However, scenarios
may be constructed to model the possible values of it .
• A deterministic interest rate scenario is a sequence of pre-specified

1-period rates it applicable in the future.
• The US 1990 Standard Valuation Law requires cash-flow testing un-

der seven prescribed interest rate scenarios. These scenarios are
3
similar to those used in the New York Regulation 126 and they are
now commonly known as the NY7 scenarios. See Table 9.1.
Table 9.1: The NY7 interest rate scenarios
Scenario Interest rate movement

1 Base scenario Level
2 Gradual increase Uniformly increasing over 10 years at 0.5% per year, and then level
3 Up-down Uniformly increasing over 5 years at 1% per year, then
uniformly decreasing over 5 years at 1% per year to the original level
at the end of 10 years, and then level
4 Pop-up An immediate increase of 3%, and then level
5 Gradual decrease Uniformly decreasing over 10 years at 0.5% per year, and then level
6 Down-up Uniformly decreasing over 5 years at 1% per year, then
uniformly increasing over 5 years at 1% per year to the original level
at the end of 10 years, and then level
7 Pop-down An immediate decrease of 3%, and then level
4
Example 9.1: Given that the current 1-year spot rate is 6%, (a) find it
for t = 1, · · · , 12 under the “up-down” scenario of NY7, and (b) compute
a12e under this interest rate scenario.
Solution: (a) The interest rates under the “up-down” scenario are given
in the following table:
t 1 2 3 4 5 6 7 8 9 10 11 12
it 6% 7% 8% 9% 10% 11% 10% 9% 8% 7% 6% 6%
(b) Using equation (3.12) (with iFt replaced by it ) and the above interest
rate scenario, we obtain a12e = 7.48. 2
5
9.2 Random-Scenario Model
• A random-scenario model is a collection of specified plausible interest

rate scenarios.
• The modeler, however, needs to state the probability distribution of

these scenarios. Both the choice of scenarios and the corresponding
probability assignment require personal judgement, which should
reflect the modeler’s view on the future interest-rate environment of
the economy.
Example 9.2: Consider the following random interest rate scenario

model.
6
Scenario Probability i1 i2 i3 i4 i5
1 0.1 3.0% 2.0% 2.0% 1.5% 1.0%
2 0.6 3.0% 3.0% 3.0% 3.5% 4.0%
3 0.3 3.0% 4.0% 5.0% 5.0% 5.0%
Find the mean, the variance and the standard deviation of a(5), 1/a(5),
a5e , ä5e , s5e and s̈5e under this model.
Solution: Note that

n−1
X 1
äne =
t=0 a(t)
n−1
X 1
= 1+
t=1 a(t)
n−1
X 1
= 1+ Qt . (9.1)
t=1 j=1 (1 + ij )
Using (3.10) through (3.12), (9.1), (3.14) and (3.15) (with iFt replaced by
7
it ), we can compute the values of a(5), 1/a(5), a5e , ä5e , s5e and s̈5e under
each scenario. For example, under the first scenario, we have
a(5) = (1.03)(1.02)(1.02)(1.015)(1.01) = 1.0986,

1 1
= = 0.9103,
a(5) 1.0986
1 1 1
a5e = + +· · ·+ = 4.6855,
(1.03) (1.03)(1.02) (1.03)(1.02)(1.02)(1.015)(1.01)
1 1
ä5e = 1 + + ··· + = 4.7753,
(1.03) (1.03)(1.02)(1.02)(1.015)
s5e = 1 + (1.01) + · · · + (1.01)(1.015)(1.02)(1.02) = 5.1474,
and
s̈5e = (1.01)+(1.01)(1.015)+· · ·+(1.01)(1.015)(1.02)(1.02)(1.03) = 5.2459.
8
We repeat the calculations for the other two scenarios and summarize the
results in the following table.
Scenario Probability a(5) 1/a(5) a5e ä5e s5e s̈5e

1 0.1 1.0986 0.9103 4.6855 4.7753 5.1474 5.2459
2 0.6 1.1762 0.8502 4.5630 4.7128 5.3670 5.5433
3 0.3 1.2400 0.8064 4.4466 4.6402 5.5141 5.7541
Therefore,
" #
1
E = (0.1 × 0.9103) + (0.6 × 0.8502) + (0.3 × 0.8064) = 0.8431,
a(5)
and
" #
1
Var = 0.1 × (0.9103 − 0.8431)2 + 0.6 × (0.8502 − 0.8431)
a(5)
+ 0.3 × (0.8064 − 0.8431)2
= 0.00089.
9
The mean, the variance and the standard deviation of other variables can
be computed similarly. The results are given in the following table.
a(5) 1/a(5) a5e ä5e s5e s̈5e

Mean 1.1876 0.8431 4.5403 4.6973 5.3892 5.5768
Variance 0.00170 0.00089 0.00505 0.00173 0.01082 0.02105
Std Dev 0.0412 0.0298 0.0711 0.0416 0.1040 0.1451
2
10
9.3 Independent Lognormal Model
• The Independent Lognormal Model assumes that 1 + it are in-

dependently lognormally distributed with parameters μ and σ 2 . In
other words, ln(1 + it ) follows a normal distribution with mean μ
and variance σ 2 (see Appendix A.14).
• Thus,
ln(1 + it ) ∼ N(μ, σ 2 ). (9.2)
• The mean and variance of the lognormal random variable (1 + it )

are, respectively,
μ+ 12 σ 2
E (1 + it ) = e , (9.3)
and ³ ´³ ´
2μ+σ2 σ2
Var(1 + it ) = e e −1 . (9.4)
11
• We now consider the mean and variance of a(n), 1/a(n), s̈ne , ane ,
sne and äne under the independent lognormal interest rate model.
• We shall assume all 1-period spot rates to be random, including the

rate for the first period i1 .

n
Y
a(n) = (1 + it ).
t=1
• Thus,
n
X
ln [a(n)] = ln(1 + it ). (9.5)
t=1
• Therefore, ln [a(n)] is normal with mean nμ and variance nσ 2 , im-

plying a(n) is lognormal with parameters nμ and nσ 2 .
12
• Using (9.3) and (9.4), we have
nμ+ n σ2
E [a(n)] = e 2 , (9.6)
and ³ ´³ ´
2nμ+nσ 2 nσ2
Var [a(n)] = e e −1 . (9.7)
• For the distribution of (1+it )−1 , under the lognormal model we have
h i
−1
ln (1 + it ) = − ln(1 + it ),
so that h i
−1
ln (1 + it ) ∼ N(−μ, σ 2 ), (9.8)
and (1 + it )−1 follows a lognormal distribution with parameters −μ
and σ 2 .
13
Yn
1
= (1 + it )−1 ,
a(n) t=1
so that " #
1 n 2
E = e−nμ+ 2 σ , (9.9)
a(n)
and " #
1 ³ ´³ ´
−2nμ+nσ 2 nσ 2
Var = e e −1 . (9.10)
a(n)
• The statistical properties of annuities (say, s̈ne , ane , sne and äne )
under the lognormal interest rate assumption are fairly complex.
• We present and discuss some applications of their mean and variance

formulas without going through the proof.
14
1 2
• We first define rs such that 1 + rs = E(1 + it ) = eμ+ 2 σ , implying
μ+ 12 σ 2
rs = e − 1. (9.11)
• Furthermore, we let
js = 2rs + rs2 + vs2 , (9.12)
where ³ ´³ ´
2μ+σ 2 σ2
vs2 = Var(1 + it ) = e e −1 . (9.13)
• The mean of s̈ne is

⎡ ⎤
³ ´ Xn Y
t
E s̈ne = E ⎣ (1 + in−j+1 )⎦
t=1 j=1
n Y
X t
= E(1 + in−j+1 )
t=1 j=1
15
n
X
= (1 + rs )t
t=1
= s̈ners , (9.14)
where s̈ners is s̈ne evaluated at the rate rs .
• Similarly, the variance of s̈ne is

Ã ! Ã !
³ ´ js + rs + 2 2js + 2 ³ ´2
Var s̈ne = s̈nejs − s̈ners − s̈ners . (9.15)
js − rs js − rs
• Next, we consider ane . We define ra such that (1 + ra )−1 =

−1 −μ+ 12 σ2
E(1 + it ) = e , implying
μ− 12 σ 2
ra = e − 1. (9.16)
• Furthermore, we let
2
ja = e2(μ−σ ) − 1. (9.17)
16
• Similar
³ ´
to (9.14) and (9.15), we can show that the mean of ane is
E ane = anera and
Ã ! Ã !
³ ´ ja + ra + 2 2ra + 2 ³ ´2
Var ane = aneja − anera − anera .
ra − ja ra − ja
(9.19)
• Furthermore, we have
³ ´ ³ ´
E äne = 1 + E an−1e , (9.20)
³ ´ ³ ´
E sne = 1 + E s̈n−1e . (9.21)
³ ´ ³ ´
Var äne = Var an−1e , (9.22)
and ³ ´ ³ ´
Var sne = Var s̈n−1e . (9.23)
17
Example 9.3: Consider a lognormal interest rate model with parameters
μ = 0.04 and σ 2 = 0.016. Find the mean and variance of a(5), 1/a(5), s̈5e ,
a5e , s5e and ä5e under this model.
Solution: We apply equations (9.6) through (9.10) to obtain
E [a(5)] = 1.27125,
Var [a(5)] = 0.13460,
" #
1
E = 0.85214,
a(5)
" #
1
Var = 0.06058.
a(5)
For s̈5e , using expressions (9.11) through (9.13), we have
1
rs = e0.04+ 2 (0.016) − 1 = 0.04917,
³ ´³ ´
vs2 = e 2(0.04)+0.016
e 0.016
− 1 = 0.01775,
18
js = 2rs + rs2 + vs2 = 0.11851,
s̈5ers = 5.78773,
s̈5ejs = 7.08477.
Applying equations (9.14) and (9.15), we obtain

³ ´
E s̈5e = 5.78773,
³ ´
Var s̈5e = 1.26076.
For a5e , using expressions (9.16) and (9.17), we have
0.04− 12 (0.016)
ra = e − 1 = 0.03252,
ja = e2(0.04−0.016) − 1 = 0.04917,
a5era = 4.54697,
a5eja = 4.33942.
19
Applying equations (9.18) and (9.19), we obtain
³ ´
E a5e = 4.54697,
³ ´
Var a5e = 0.72268.
Finally, employing formulas (9.20) through (9.23), we obtain

³ ´ ³ ´
E ä5e = 1 + E a4e = 1 + 3.69483 = 4.69483,
³ ´ ³ ´
Var ä5e = Var a4e = 0.40836,
³ ´ ³ ´
E s5e = 1 + E s̈4e = 1 + 4.51648 = 5.51648,
³ ´ ³ ´
Var s5e = Var s̈4e = 0.64414.
20
9.4 Autoregressive Model
• Let Yt = ln(1 + it ). Under the independent lognormal model, we

have
Yt = ln(1 + it ) ∼ N(μ, σ 2 ), (9.24)
where there are no correlations between Yt and Yt−k for all k ≥ 1.
• We now consider interest rate models that allow some dependence

among Yt ’s, say,
Corr(Yt , Yt−k ) 6= 0, for some k ≥ 1, (9.25)
while maintaining the lognormal assumption.
• We consider the class of first-order autoregressive models, denoted
21
by AR(1). The AR(1) model has the form
Yt = c + φYt−1 + et , (9.26)
where c is the constant intercept and φ is the autoregressive para-

meter and et are independently and identically distributed normal
random variates each with mean zero and variance σ 2 .
• For |φ| < 1, the correlation structure of the interest-rate process

{Yt } is
Corr(Yt , Yt−k ) = φk , for k = 1, 2, · · · , (9.27)
which does not vary with t.
• The mean and variance expressions of a(n), 1/a(n), ane , äne , sne and
s̈ne under an AR(1) interest rate model are very complex.
22
• Thus, we consider a stochastic simulation approach to obtain the
empirical distributions illustrated in Table 9.2.
Table 9.2: A simulation procedure of ane for the AR(1) process
Step Procedure
1 Using historical interest rate data, estimate the parameters (c, φ, σ 2 ) in the
AR(1) model.
2 Draw random normal numbers e1 , e2 , · · · , en (with mean zero and variance σ2 )
from a random number generator.
3 Compute Y1 , Y2 , · · · , Yn using the AR(1) equation (9.26) and setting the initial value
Y0 = Y , where Y is the sample average of the observed data.
4 Convert Yt ’s to an interest rate path (i1 , i2 , · · · , in ) using the relationship it = eYt − 1.
5 Compute the ane function under the simulated interest rate path in Step 4.
6 Repeat Steps 2 to 4 m times. Note that the random normal variates in Step 2 are
redrawn each time and we have a different simulated interest rate path in each
replication.
23
Example 9.4: Consider an AR(1) interest-rate model with parameters
c = 0.03, φ = 0.6 and σ 2 = 0.001. Using the stochastic simulation method
described in Table 9.2 (with m = 1, 000 and Y0 = 0.06), find the mean
and variance of a(10), 1/a(10), a10e , ä10e , s10e and s̈10e under this model.
Solution: We follow the simulation steps in Table 9.2 and obtain the
empirical distributions (histograms) of the six functions, which are plotted
in Figure 9.1. The empirical means and variances are as follows:
a(10) 1/a(10) a10e ä10e s10e s̈10e

Mean 2.1656 0.4836 6.8382 7.3546 14.5018 15.6674
Variance 0.2224 0.0110 0.5678 0.4383 3.5101 5.3524
2
24
Figure 9.1: Histograms of simulated results in Example 9.4
9.5 Dynamic Term Structure Model
• There are two popular approaches to model the dynamic term struc-
ture: no-arbitrage approach and equilibrium modeling ap-
proach.
• The no-arbitrage approach focuses on perfectly fitting the term

structure at a point in time to ensure that no arbitrage opportunities
exist in the financial markets. This approach is important for pricing
derivatives.
• The equilibrium modeling approach focuses on modeling the

instantaneous spot rate of interest, called the short rate, and use
affine models to derive the yields at other maturities under various
assumptions of the risk premium.
25
9.6 An Application: Guaranteed Investment Income
• We consider a hypothetical guaranteed income fund offered by a

bank with the following features:
• It is a closed-end fund. The bank sells only a fixed number

of units (say, 10,000 units with $1,000 face amount per unit) in
the initial public offering (IPO). No more units will be issued
by the bank after the IPO. Redemption prior to the maturity
date is not allowed.
• Investment period is 9 years.
• At maturity, the bank will return 100% of the face amount to
the investors.
26
• At the end of each of the nine years, earnings from the fund
will be distributed to investors. In addition, the bank provides
a guarantee on the minimum rate of return each year according
to a fixed schedule:
Guaranteed rate
At the end of (as % of initial investment)
Year 1 3%
Year 2 4%
Year 3 4%
Year 4 5%
Year 5 5%
Year 6 5%
Year 7 5%
Year 8 5%
Year 9 5%
Thus, if the fund earns less than the guaranteed rate in any
27
year, the bank has to top up the difference.
Example 9.5: Consider the above guaranteed income fund. Assume

that the rates of return for the fund in the next nine years are: 2%, 3%,
4%, 5%, 6%, 5%, 4%, 3%, 2%, respectively. Compute the cost of the
guarantee to the bank (i.e., the present value of the top-up amounts).
Solution: Let Gt be the guaranteed rate for year t offered by the bank,
it be the rate of return earned by the fund for year t, and Ut be the top-
up amount (per $1,000 face amount) by the bank to honor the guarantee
payable at the end of year t. Note that
(
0, if it ≥ Gt ,
Ut = (9.28)
1,000(Gt − it ), if it < Gt .
Using equation (9.28), we compute Ut for each year. The results are given
as follows.
28
t 1 2 3 4 5 6 7 8 9
it 2% 3% 4% 5% 6% 5% 4% 3% 2%
Gt 3% 4% 4% 5% 5% 5% 5% 5% 5%
Ut 10 10 0 0 0 0 10 20 30
The cost of the guarantee C is the present value of the Ut ’s. Thus,
9
X Ut
C = Qt
t=1 j=1 (1 + ij )
10 10 0
= + + + ···
1.02 (1.02)(1.03) (1.02)(1.03)(1.04)
= 62.98. (9.30)
• The bank needs to charge an up-front premium to cover the cost of

the guarantee. We now apply the stochastic interest rate approach
29
to examine the distribution of C. This information is useful to the
bank’s management for setting the premium.
• For illustration, we assume that the fund invests only in government

bonds and the returns of the fund follow an independent lognormal
interest rate model with parameters μ = 0.06 and σ 2 = 0.0009, i.e.,
ln(1 + it ) ∼ N(0.06, 0.0009). (9.30)
• We use stochastic simulation to produce a series of future interest

rate movements. For each of the simulated path, using equations
(9.28) and (9.29), a realization of C can be computed. We repeat the
simulation experiment m times (say, m = 5,000) and the empirical
distribution (histogram) of C can be obtained.
• The time lag between the bank’s filing of the required fund doc-
uments to the regulator for approval and the actual IPO is often
30
lengthy. The bank has to determine the premium rate well before
the IPO date. Therefore, in the simulation, the first-period rate of
return i1 is not known at time 0.
• The simulation study is carried out and the empirical distribution

(histogram) of C is plotted in Figure 9.2.
31
Figure 9.2: Empirical distribution (histogram) of the cost of the guarantee (C)

Financial Mathematics For Actuaries PDF

Uploaded by

Copyright:

Available Formats

Financial Mathematics For Actuaries PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Financial Mathematics For Actuaries PDF

Uploaded by

Copyright:

Available Formats

What are the two main methods for calculating interest?

What are the two main methods for calculating interest?

What are the accumulation and amount functions?

What are the accumulation and amount functions?

Financial Mathematics

1. Basic principles in calculation of interest accumulation

2. Simple and compound interest

4. Eﬀective rate of interest

6. Present and future values of a single payment

• The sum of money borrowed is called the principal.

• The borrower compensates the lender by paying interest.

• We denote A(t) as the accumulated amount at time t, called the

• A(0) is the initial principal and

I(t) = A(t) − A(t − 1) (1.1)

• For the special case of an initial principal of 1 unit, we denote the

• If the initial principal is A(0) = k, then

• While theoretically there are numerous ways of calculating the in-

• These are the simple-interest method and the compound-interest

• For the simple-interest method the interest earned over a period of

• The interest incurred from time 0 to time t, for a principal of 1 unit,

a(t) = 1 + rt, for t ≥ 0, (1.2)

A(t) = A(0)a(t) = A(0)(1 + rt), for t ≥ 0. (1.3)

Example 1.1: A person borrows $2,000 for 3 years at simple interest.

2,000 × 0.08 = $160.

The accumulated amount at the end of year 3 is

2,000 (1 + 0.08 × 3) = $2,480.

• For the compound-interest method the accumulated amount over a

• The accumulation function becomes

a(t) = (1 + r)t , for t = 0, 1, 2, · · · , (1.4)

and the amount function is

A(t) = A(0)a(t) = A(0)(1 + r)t , for t = 0, 1, 2, · · · . (1.5)

— For the compound-interest method the accumulated amount at

2,000 × 0.08 = $160.

For year 2 the principal is

2,000 + 160 = $2,160,

so that the interest for the year is

2,160 × 0.08 = $172.80.

The accumulated amount at the end of year 3 is

2,000 (1 + 0.08)3 = $2,519.42.

• When the interest rate is high, compounding the interest induces

5% interest 10% interest

1 105.00 105.00 110.00 110.00

• Although the rate of interest is often quoted in annual term, the

• For example, a savings account may pay interest at 3% per year,

• The frequency of interest payment (also called the frequency of

• We use r(m) to denote the nominal rate of interest payable m

• Assume t (in years) is an integer multiple of 1/m, i.e., tm is an

Example 1.3: A person deposits $1,000 into a savings account that

Solution: The rate of interest over one month is

1,000 × 0.0025 = $2.50,

and the accumulated amount after one month is

1,000 + 2.50 = $1,002.50.

Solution: The investment interval is 30 months. Thus, using (1.7), the

1,077.78 − 1, 000 = $77.78.

• When the loan period is not an integer multiple of the compounding

• When the compounding frequency m increases, the accumulated

• Let r̄ denote the nominal rate of interest for which compounding is

• We call this compounding scheme continuous compounding.

From the well-known limit theorem (see Appendix A.1) that