Chapter 11

1 Term structure of interest rates

1.1 The term structure of interest rates
Throughout this course we have made use of the simplifying assumption that the discount factor
over time interval [0, t] is

Pt = e−δt = (1 + i)−t (b.t.u. = years)

for constant δ and i. Clearly P (0, t) represents the price % at time 0 of a zero coupon bond
redeemable at par at time t, calculated at a constant yield to maturity.
The main question that arises is how these yields to maturity evolve over time. In Chapter
13, we had 3 cases of the behaviour of yields as functions of the term. We said in summary
that

• If i(p) > (1 − t1 ) g, then the yield to maturity is a decreasing function of the term

• If i(p) < (1 − t1 ) g, then the yield to maturity is an increasing function of the term

• If i(p) = (1 − t1 ) g, then the yield to maturity is constant (flat yield) independent of the
term

The variation by term of interest rates (read yields to maturity) is often referred to as the term
structure of interest rates. In this chapter, we shall focus on various types of interest rates i.e
spot rates, forward rates and yields to maturity and how they are connected. We shall then
try and revisit the term structure of interest rates from another angle, a belief on the market
that the term structure of interest rates may have nothing to do with the relationship between
i(p) and (1 − t1 ) g but by some facts about the market. We now begin by defining spot rates,
which in part represents the yield to maturity of zero-coupon bonds.

1.2 Spot Rates

Definition 14.2.1 (Spot Rate) The n-year zero rate (spot rate) is the rate of interest earned on
an investment that starts today and lasts for n years, where all the interest and principal is
realized at the end of n years. It is a measure of the average interest rate over the period from
now until n years’ time.

1.2.1 Annually compounded spot rates

Let Pt denote the price of a zero coupon bond that matures at time t. Then the annually
compounded spot rate yt satisfies

Pt = (1 + yt )−t ⇒ Pt−1 = (1 + yt )t

These are merely the yields on the corresponding zero coupon bonds expressed as a constant
force of interest and constant effective annual rate respectively.

1
 Example: The prices for the zero coupon bonds of various terms are as follows:
1 year = $94%, 5 years = $70%, 10 years = $47%, 15 years = $30%
$x% means $x per $100 nominal.
Calculate the spot rates for these terms and sketch a graph of these rates as a function of
the term.

2 Solution
The spot rates for the various terms can be found from the equations of value: 1 year:
100 (1 + y1 )−1 = 94 ⇒ y1 = 6.4% or (1 + y1 )−1 = 0.94
5 years: 100 (1 + y5 )−5 = 70 ⇒ y5 = 7.4%
10 years: 100 (1 + y10 )−10 = 47 ⇒ y10 = 7.8%
15 years: 100 (1 + y15 )−15 = 30 ⇒ y15 = 8.4%

The graph above is an example of a yield curve. It isn’t always spot rates that are plotted
in a yield curve. It might instead be redemption yields (yields to maturity) or forward rates.
Our observation for this example is that the term structure of spot rates is an increasing trend.
Since rates of interest differ according to the term of the investment, in general ys ̸= yt for

2
s ̸= t. Every fixed-interest investment may be regarded as a combination of (perhaps notional)
zero coupon bonds. For example, a bond paying coupons of D every year for n years, with a
final redemption payment of R at time n may be regarded as a combined investment of n zero
coupon bonds with maturity value D, with terms of 1 year, 2 years, ..., n years, plus a zero
coupon bond of nominal value R with term n years.

Defining vyt = (1 + yt )−1 , the price of the bond is:

A = D (P1 + P2 + . . . Pn ) + RPn = D vy1 + vy22 + · · · + vynn + Rvynn .

This is actually a consequence of "no arbitrage"; the portfolio of zero coupon bonds has the
same payouts as the fixed-interest bond, and the prices must therefore be the same.
The variation by term of interest rates is often referred to as the term structure of interest
rates. The curve of spot rates {yt } is an example of a yield curve

Example
Calculate the price of a five year fixed interest security, redeemable at par, with 6%
annual
coupons if the annual term structure of interest rates is: 7%, 7 4 %, 7 2 %, 7 4 %, 8%, . . .
1 1 3

Solution: The price per $100 nominal is
given by the equation:

2 3 4 5 8
P = 6 v7 % + v7.25 + v7.5% + v7.75% + v8% + 100v8%
⇒ P = 6 × 4.0314 + 100 × 0.6806 = $92.25
The gross redemption yield for this security is the value of i that solves 92.25 = 6a5i +100v 5 .
Solving we get i = 0.0793786 ≈ 7.94%

2.0.1 Continuous time spot rates

The continuous time spot rate is the force of interest that is equivalent to an effective rate of
interest equal to the spot rate.
Let Pt be the price of a unit zero coupon bond of term t. Then the t-year spot force of
interest is Yt where:
Pt = e−t.Yt ⇒ Yt = − 1t ln (Pt )
This is also called the continuously compounded spot rate of interest or the continuous-time
spot rate. Yt and its corresponding discrete annual rate yt are connected in the same way as δ
and i; an investment of R1 for t years at discrete spot rate yt accumulates to (1 + yt )t ; at the
continuous time rate it accumulates to et.Yt ; these must equal, so yt = eYt − 1 or Yt = ln (1 + yt )

Example
The prices of zero coupon bonds of various terms are as follows: 1 year = $94%5 years = $70%
years = $47%15 years = $30%
Calculate Y10

Solution: Since Yt = − 1t ln (Pt ) and since P10 = 0.47, we have Y10 = − 10

1
ln 0.47 = 7.55%

3
3 Forward interest rates
The word forward has previously been used in 2 different aspects. We defined a forward rate
agreement as some form of agreement by two parties to pay cash to each other in future at
different interest rates (usually a fixed rate versus a floating rate) calculated on some nominal
amount. We then defined a forward contract as a derivative giving two parties the right and
obligation to trade a predetermined underlying security, at a predetermined price (the forward
price) and at a predetermined future date (the maturity date). We now add another concept
using the word forward. Be careful not to confuse these.
Definition 14.3.1 (Forward interest rate) A forward interest rate, often referred to as a
forward rate, is an interest rate at which we can agree now to borrow or lend over a specified
future time period. In the bond markets, a forward interest rate is the interest rate implied by
current zero-coupon rates for a specified future time period. Forward rates can be expressed as
continuously or discretely compounded rates.

3.0.1 Discrete time forward rates

The discrete time forward rate, ft,r , is the annual interest rate agreed at time 0 for an investment
made at time t > 0 for a period of r years.
That is, if an investor agrees at time 0 to invest $100 at time t for r years, the accumulated
investment at time t + r is:
100 (1 + ft,r )r
Therefore the forward rate, ft,r is a measure of the average interest rate between times t and
t + r as agreed to now.
Forward rates, spot rates and zero coupon bonds are all connected. The accumulation at
time t of an investment of 1 at time 0 is (1 + yt )t . If we agree at time 0 to invest the amount
(1 + yt )t at time t for r years, we will earn an annual rate of ft,r . So we know that $1 invested
for t + r years will accumulate to (1 + yt )t (1 + ft,r )r . But we also know from the (t + r) spot
rates that $1 invested for t + r years accumulates to (1 + yt+r )t+r , and we also know from the
zero coupon bond prices that R1 invested for t + r years accumulates to Pt+r −1
. Hence we know
that:
(1 + yt )t (1 + ft,r )r = (1 + yt+r )t+r = Pt+r
−1

from which we find

r (1 + yt+r )t+r Pt
(1 + ft,r ) = t =
(1 + yt ) Pt+r
so that the full term structure may be determined given the spot rates, the forward rates or
the zero coupon bond prices.
The connection between the spot rates and the forward rates can be represented on a time
line.

4
 Accumulating payments from time 0 to time t + r using the spot rate yt+r is equivalent to
first accumulating to time t using yt , and then accumulating from time t to time t + r using
the forward rate ft,r .
One period forward rates are of particular interest. The one- period forward rate at time t
(agreed at time 0 ) is denoted ft = ft,1 . Clearly f0 = y1 . Comparing an amount of R1 invested
for t years at the spot rate yt , and the same investment invested 1 year at a time with proceeds
reinvested at the appropriate one-year rate, we have
(1 + yt )t = (1 + f0 ) (1 + f1 ) (1 + f2 ) . . . (1 + ft−1 )
The one - year forward rate, ft , is therefore the rate of interest from time t to time t + 1. It
can be expressed in terms of spot rates:
t+1
(1 + ft ) = (1+y t+1 )
(1+yt )t
Example
If the n year spot rates can be approximated by the function 0.09 − 0.03e−0.1n , calculate the
one year forward rate at time 10
Solution
We have yn = 0.09 − 0.03e−0.1n and we want to find f10 .
11
(1 + y11 )11 (1 + 0.09 − 0.03 × e−0.1×11 )
Now (1 + f10 ) = = = 1.09057.
(1 + y10 )10 (1 + 0.09 − 0.03 × e−0.1×10 )10

Therefore f10 = 0.09057 = 9.06%.

3.0.2 Continuous time forward rates

The continuous time forward rate Ft,r is the force of interest equivalent to the annual forward
rate of interest ft,r .
A R1 investment of duration r, starting at time t, agreed at time 0 ≤ t accumulates using
the annual forward rate of interest to (1 + ft,r )r at time t + r.
Using the equivalent forward force of interest the same investment accumulates to er.Ft,r .
Hence the annual rate and continuous-time rate are related as:
ft,r = eFt,r − 1
The relationship between the continuous time spot rate and forward rates may be derived
by considering the accumulation of $1 at a continuous time spot rate of Yt for t years, followed
by the continuous time forward rate Ft,r for r years.
Compare this with an investment of $1 at a continuous time spot rate of Yt+r for t + r years.

5
The two investments are equivalent, so the accumulated values must be the same. Hence:

etYt · er.Ft,r = e(t+r)Yt+r ⇒ t.Yt + rFt,r = (t + r)Yt+r

(t + r)Yt+r − tYt
⇒ Ft,r =
r
Also, using Yt = − 1t log Pt , then we have
 
1 Pt
Ft,r = log .
r Pt+r
Once again we represent the connection between the continuous time spot and forward rates
on a time line.

Example
The prices of zero coupon bonds of various terms are as follows: 1 year = $94%5 years = $70%
years = $47%15 years = $30% Calculate F5,10 .
 
Solution Since Ft,r = 1r ln PPt+r , therefore F5,10 = 10
1 1
ln 0.7


t
ln (P5 P15 ) = 10 0.3
= 8.4%

3.0.3 Instantaneous forward rates

In the discrete time case, we considered the special case r = 1 and obtained the annual forward
rates ft,1 =ft . In
 the continuous time case we are interested in the case r = 0. But then since
Ft,r = 1r ln PPt+r
t
then r cannot equal 0 . It then only makes sense to consider the case r → 0.
Definition 14.3.2 (Instantaneous forward rate) The instantaneous forward rate Ft is Ft =
limr→0 Ft,r .
The instantaneous forward rate may broadly be thought of as the forward force of interest
applying in the instant of time t → t + ∆t. Therefore mathematically we can calculate the
instantaneous forward rate as follows:
 
1 Pt
Ft = limr→0 r log Pt+r
log Pt −log Pt+r
= limr→0 r
log Pt+r −log Pt
= − limr→0 r
= − dtd log Pt
= − P1t dtd Pt

6
We also find, by integrating both sides of (14.3.3) and using the fact that P0 = 1, that
− 0t Fs ds
which looks similar to the expression
R
Pt = e R
t
v(t) = e− 0 δ(s)ds that you studied earlier in the course.
As you can see, Pt is equivalent to v(t) and Fs is equivalent to δ(s)

Example
The price at time 0, Pt , of a zero-coupon bond for 2 < t < 4 is given by the equation:
Pt = (100 − 2t2 ) %.
Calculate the instantaneous forward rate F3 .

4 Solution
We use Ft = − P1t dtd Pt . Now d
P
dt t
= −4t and thus Ft = 4t
100−2t2
and F3 = 12
82
= 0.1463 = 14.63%

4.0.1 The case of term structure over time

We have described the initial term structure, where everything is fixed at time 0 . In practice
the term structure varies rapidly over time, and the 5-year spot rate tomorrow may be quite
different from the 5-year spot rate today. In more sophisticated treatments we model the change
in term structure over time.
In this case all the variables we have used, i.e
Pt yt ft,r Yt Ft,r
need another argument, v, say, to give the "starting point". For example, yv,t would be
the t-year discrete spot rate of interest applying at time v; Fv,t,r would be the force of interest
agreed at time v, applying to an amount invested at time v + t for the r-year period to time
v + t + r.

5 Example
Using the above notation, if y0,5 = 6%, y5,5 = 7 12 %, F0,5,5 = 7% and F5,5,5 = 8 14 , calculate the
value of a ten-year zero coupon bond issued at time zero

6 Solution
 
We want to calculate P0,10 = P10 . We have F5,5 = 1
5
ln P5
P10
= 7% and P5 = (1 + y5 )−5 =
(1 + y0,5 )−5 = 1.06− 5.
Therefore P10 = 1.065 ×e
1
0.35 = 0.526584 = 52.66%

7
7 Theories of the term structure of inter- est rates
7.0.1 Why interest rates vary over time
The prevailing interest rates in investment markets usually vary depending on the time span
of the investments to which they relate. This variation determines the term structure of the
interest rates.
The variation arises because the interest rates that lenders expect to receive and borrowers
are prepared to pay are influenced by the following factors which are not normally constant
over time.

7.1 - Supply and demand

Interest rates are determined by market forces, i.e. the interaction between borrowers and
lenders. If cheap finance is easy to obtain or if there is little demand for finance, this will push
interest rates down.

7.2 - Base rates

In many countries there is a central bank that sets a base rate of interest which provides
a reference point for other interest rates. For example, an interest rate in the UK may be
expressed as the Bank of England’s base rate plus 4%. Investors will have a view on how this
rate is likely to move in the future.

7.3 - Interest rates in other countries

The interest rates in a particular country will also be influenced by the cost of borrowing in
other countries because major investment institutions have the alternative of borrowing from
abroad.

7.4 - Expected future inflation

Lenders will expect the interest rates they obtain to outstrip inflation. So periods of high
inflation tend to be associated with high interest rates.
- Tax rates If tax rates are high, interest rates may also be high because investors will require
a certain level of return after tax.

• Risk associated with changes in interest rates

In general, rates of interest tend to increase as the term increases because of the risk of loss
due to a change in interest rates is greater for longer-term investments.

7.4.1 The theories

Some examples of typical (spot rate) yield curves are given below.
Yield

8
 Term in years
Figure 1: Decreasing Yield Curve In Figure 1 the long-term bond yields are lower than
for the short-term bonds. Since price is a decreasing function of yield, an interpretation is
that long-term bonds are more expensive than short-term bonds. There are several possible
explanations - for example it is possible that investors believe that they will get a higher overall
return from long-term bonds, and the higher demand for long-term bonds has pushed up the
price, which is equivalent to pushing down the yield, compared with short-term bonds. Other
explanations for different yield curve shapes are given below.
In Figure 2 the long-term bonds are higher yielding (or cheaper) than the short-term bonds.

9
 Figure 2: Increasing Yield Curve
In Figure 3 the short-term bonds are generally cheaper than the long bonds, but the very
short rates (with terms less than 1 year) are lower than the 1-year rates.

10
 Figure 3: Humped Yield Curve
The three most popular explanations (although we don’t go into any detail) for the fact
that interest rates vary according to the term of the investment are:

1. Expectations Theory

2. Liquidity Preference

3. Market Segmentation

7.4.2 Expectations Theory

The key assumption behind the expectations theory is that investors do not prefer bonds of one
maturity over another, so that they will not hold any quantity of a bond if its expected return
is less than that of a bond with a different maturity. Therefore bonds are perfect substitutes.
The key assumptions of the expectations theory are:

• bonds of different maturities are perfect substitutes i.e demand for one type of bond
increases as the price of the other increases. - investors are risk neutral i.e there is no risk

11
 premium required for long term bonds so investors are attracted to bonds that offer high
yields without worrying about levels of risk.

• shape of yield curve is determined by investors’ expectations of future interest rates

• upward slopping yield curve means short term interest rates are expected to rise in the
future. Downward slopping yield curve means short term interest rates are expected to
fall in the future. Flat yield curve means short term interest rates will remain unchanged
in the future

According to the expectations theory, since investors are indifferent to bonds of different matu-
rities giving the same yield and bonds are perfect substitutes, then the following two strategies
are similar and give the same yield

1. Buy a one year bond, hold it for one year and re-invest the proceeds in another one year
bond one year from now

2. Buy a two year bond and hold it for two years.

Therefore according to the Expectations theory, (1 + y2 )2 = (1 + y1 ) (1 + f1 ), i.e according to

the expectations theory, the interest on long term bonds should be equal to the geometric
average of short term interest rate.

8 Example
Suppose that the one year bond yield y1 = 9% and the expected one year bond yield is f1,1 =
f1 = 11% during year 2 . According to√the Expectations Hypothesis theory, the interest on the
long term (2 year) bond equals y2 = 1.09 × 1.11 − 1 = 0.999545 = 9.999545%. In this case
the yield curve is upward slopping as in Figure 2 .
Note that, we may also use the arithmetic mean to approximate the geometric mean in the
expectations theory argument. If we had used the arithmetic mean, then we would have got
y2 = 12 (y1 + f1 ) = 10%.
The Expectations theory also believes that interest rates of different maturities tend to move
together over time. Suppose the current one year yield is y1 . Then by the Expectations theory
(by using the arithmetic mean for our arguments) we have
1
y2 ≈ (y1 + f2 ) and
2
1
y3 ≈ (y1 + f2 + f3 )
3
If y1 goes up, so does y2 and y3 . So the Expectations theory says that interest rates for different
maturities tend to move together over time

12
8.0.1 Liquidity Preference
Like the Expectations Theory, the liquidity preference theory says that long term bonds are
an average of expected future short term rates but the liquidity preference theory believes that
investors must be compensated for holding long term bonds by a risk premium. The higher the
risk the higher the premium. The key assumptions of the liquidity preference theory are that:

• bonds of different maturities are substitutes but not perfect.

• investors are risk averse

• investors will demand a Liquidity Premium for long term bonds because of interest rate
risk

• the liquidity premium increases with term

Assume that one year rates over the next 5 years are expected to be y1 = 5%, f1 = 6%, f2 =
7%, f3 = 8% and f4 = 9%. Also assume that the Liquidity premiums for 1-5 year bonds are
0%, 0.25%, 0.5%, 0.75% and 1%. Then we have
√ 1
y2 = 1.05 × 1.06 − 1 + 0.25% = 0.0574882 = 5.74882% ≈ (5% + 6%)+
2
0.25% = 5.75%
1
y3 ≈ (5% + 6% + 7%) + 0.5% = 6.5%
3
1
y4 ≈ (5% + 6% + 7% + 8%) + 0.75% = 7.25%
4
1
y5 ≈ (5% + 6% + 7% + 8% + 9%) + 1% = 8.0%
5
As a result, liquidity preference theory also believes that interest rates
for different maturities tend to move together over time

8.0.2 Market Segmentation

Bonds of different terms are attractive to different investors, who will choose assets that are
similar in term to their liabilities. The liabilities of banks, for example, a very short term
(investors may withdraw a large proportion of the funds at very short notice); hence banks
invest in very short-term bonds. Many pension funds have liabilities that are very long term,
so pension funds are more interested in the longest dated bonds. The demand for bonds will
therefore differ for different reasons.
The supply of bonds will also vary by term, as governments, and companies’s strategies may
not correspond to the investor’s requirements.
Remember that governments and companies issue bonds because they need to borrow money,
not because they are kind hearted and want to give investors something to invest in. More
bonds will be supplied if more money needs to be borrowed. This will put downward pressure
on prices.
The market segmentation hypothesis argues that the term structure emerges from these
different forces of supply and demand.

13
 Market segmentation does not explain why interest rates tend to move together over time
i.e it predicts movements of interest rates but not comovements.

8.1 Yields to maturity

The yield to maturity for a coupon paying bond (also called the redemption yield) has been
defined as the effective rate of interest at which the discounted value of the proceeds of a bond
equal the price. It is widely used, but has the disadvantage that ie depends on the coupon rate
of the bond, and therefore does not give a simple model of the relationship between term and
yield.

8.2 Par yields

The n-year par yield represents the coupon per $1 nominal that would be payable on a bond
with term n years, which would give the bond a current price under the current term structure
of $1 per $1 nominal, assuming the bond is redeemed at par.
In other words,
The n-year par yield for annual coupon bonds is the coupon rate on a bond redeemed at
par in n years time that is priced at par under the current term structure.
The n-year par yield C solves the equation:
n
!
X
CPk + Pn or 1 = C vy1 + vy22 + vy33 + · · · + vynn + 1.vynn


1=
k=1
The par yields give an alternative measure of the relationship between the yield and term of
investments. The difference between the par yield rate and the spot rate is called the "coupon
bias". Clearly one may also define par yields for semiannual coupon bonds, or any other coupon
payment structure. (par yield = annual coupon rate)
Example: Given annually compounded spot rates
k 1 2 3 4
yk 9% 10% 9.5% 9%
Calculate:
(a) The price of a zero coupon bond redeemable in 3 years time.
(b) The price of an annual coupon bond redeemable in 3 years time at 120% and paying
coupons at rate 11% per annum.
(c) The 4 year par yield for annual coupon bonds.
P3 = (1 + y3 )−3 = (1.095)−3
Solution: (a)
= 76.1654%
(b) Price = 11 (1 + y1 )−1 + 11 (1 + y2 )−2
+131 (1 + y3 )−3
= 11(1.09)−1 + 11(1.10)−2 + 131(76.1654%)
= 118.9593 per 100 nominal.
(c) Solve for C :
1 =C(1.09)−1 + C(1.10)−2 + C(1.095)−3
+ (1 + C)(1.09)−4
⇒ C =9.07214% p.a.

14
8.3 Interest rate risk
Traded financial instruments are priced by the market in categories that define a common
effective annual rate of interest. Within a category, the yields on individual instruments at any
given time do vary, but generally only by a very small amount from the market average for that
category.
For examples of common categories of investments consider:

• long term annuities ( > 20 years)

• short term, high coupon bonds (<5 years)

• very short term zero coupon bonds ( < 1 year)

• overnight deposits, etc.

Depending on the type of analyses required a category could be broad or specific.

The rate of interest offered on a bank account is also determined to some extent within a
market for this product.
These observations illustrate that prices of contracts are sensitive to changes in the level of
the rate of interest offered in the market.
In many financial markets (e.g. the various categories of currency or futures markets) the
change in yield can be rapid over a short time period. On the other hand, interest rates offered
on bank accounts are usually constant for a number of months at a time.
The ideas explored in this section are more applicable to slowly moving markets using a
constant rate of interest to value future cash flows independent of their term. Both these con-
ditions are very restrictive and not at all realistic in the modern financial world. Nevertheless,
they form a starting point for further development.

8.3.1 Duration/discounted mean term

Consider the pricing of a stream of positive cash flows at constant force of interest δ :

15
 The duration or (Fredrick Macaulay) discounted mean term is defined to be:

PV′ (δ)
DMT(δ) = −
P V (δ)
" m Z T #
1 X
= Ck tk e−δtk + tρ(t)e−δt dt
PV(δ) k=1 0

In the case of discrete payments only,

m 
Ck e−δtk
X 
DMT(δ) = tk
k=0
PV(δ)
Xm
= wk tk
k=0

16
where m
Ck e−δtk X
wk = and wk = 1
PV(δ) k=0

It follows that DMT represents a weighted average time, where the weights are determined by
the discounted values of the cash flows occurring at each payment date.
The DMT is a measure of sensitivity of the PV to "small" changes in δ : Using a first order
Taylor approximation:
∆ PV(δ) ≈ PV′ (δ)∆δ
∆ PV(δ)
⇒ ≈ − DMT(δ)∆δ
PV(δ)
So the relative change or proportional (%) change in the value of PV due to a change in force
of interest used is determined by the value of DMT(δ).

9 Note:
1. The DMT is only useful as a measure of sensitivity to change if ∆δ is "small". 2. For many
financial instruments, rather than using the force of interest, it is more natural to refer
to the effective annual rate i or even i(2) . This is true particularly for securities paying
annual or semiannual coupons. In this case we often refer to MODIFIED DURATION or
EFFECTIVE DURATION as

PV′ (i) PV′ i(2)

− or −
PV(i) PV (i(2) )
McCutcheon & Scott use the term volatility for:

PV′ (i)
Vol(i) = −
PV(i)

(This term has a very different meaning in the context of financial mathematics.)
Our previous approximation then becomes

∆PV(i)
≈ − Vol(i)∆i
PV(i)

To connect derivatives relative to different variables:

1. did = dδ

d d
di dδ
= v dδ

2. dδ
d
= dv

d d
dδ dv
= −v dv

3. did = dv .

d d
di dv
= −v 2 dv

It follows that
d d
dδ
PV di
PV
DMT(δ) = − = −(1 + i)
PV PV
= (1 + i) Vol(i)

17
10 Examples:
1. A zero coupon bond

PV(δ) = e−δn ∴ P V ′ (δ) = −ne−δn ∴ DMT(δ) = n

PV(i) = (1 + i)−n ∴ P V ′ (i) = −n(1 + i)−n−1 ∴ Vol(i) = nv

1. A discretely payable level annuity

1 − vn 1 − e−δn
PV(δ) = a n = = δ
−δn
i
δ

e − 1 −δn
δ
ne e −1 − 1−e e
∴ PV′ (δ) =
(eδ − 1)2
(+nv n ) i − (1 − v n ) (1 + i)
=
 i2 
1−v n
i − nv n
( 1+i )
=−
i
ä n − nv n
=− = −(Ia) n , .
i
(Ia) n (Ia) n |
∴ DM T = and ∴ Vol = v .
an an

1. A perpetuity:

1
PV = a ∞ =
i
1 1
∴ PV′ (i) = −
2
∴ Vol(i) =
i i
1+i 1
⇒ DMT = = .
i d
1. A continuously payable level annuity:

18
 1 − vn 1 − e−δn
 
PV = ā n = =
δ δ
(+nv ) δ − (1 − v n ) · 1
n
∴ PV′ (δ) = 2
 1−vn δ n 
− nv
=− δ
δ
ān̄| − nv n
 
=−
δ
¯
= −(Iā) n ·
¯ n
(Iā) ¯ n
(Iā)
∴ DMT = , and Vol = v .
ā n ā n

1. Annuity-due:

1 − vn
PV = ä n i =
d
h i
a n −nv n a n −nv n
Exercise: PV′ (δ) = − d
and DMT(δ) = 1−v n

1. Security with annual coupons:

PV = Ca n + Rv n = P
∴ P V ′ (δ) = − C(Ia) n | + Rnv n
 
   
C R
∴ DMT = (Ia) n + nv n
P P
n
ä n − nv
  
C n

= + P − Ca n
P i P
1 − vn
 
(C/P )  n
 n
= ä n − nv + P −C
i P i
 
(C/P )  n n
 C/P
= ä n − nv + nv + n 1 −
i i
   
C/P C/P
= ä n + 1 − n
i i
"A kind of weighted average."
Defining y = C/P · · · the running yield ,

∂ 1

DMT = ä n − n < 0
∂y i
Therefore DMT decreases with increasing running yield.

19
Example:
A zero coupon bond is redeemable at par after 10 years. The bond is valued at i = 10% per
annum, the market rate used for these bonds. If the market rate drops to 9% per annum, the
value of the bond will increase. The approximate proportional change is:

∆PV(i) 1
≈ −(10) · (−1%)
PV(i) 1 + 10%
= 9.091% (Vol = 9.091)
Compare this approximation to the actual values:
PV(i = 10%) = 100(1.1)−10 = 38.55433
PV(i = 9%) = 100(1.09)−10 = 42.24108
∆PV
∴ = 9.562%
PV
We see that the use of Vol is a rough guide to the percentage change.

Example:
A bond redeemable at par in 10 years time pays annual coupons at a rate of 8% per annum. Cal-
culate the DMT for this bond at each of the effective rates of interest: 5%, 10%, 15% per annum.

Solution: We can avoid the evaluation of (Ia)n̄| functions by using the formula on the
previous page:
Yield 5% 10% 15% (TVM)
Price 123.17% 87.71% 64.87%
DMT 7.54 7.04 6.52 (years)

10.0.1 The effect of a change in i on bonds of different DMT

Both the volatility and the discounted mean term provide a measure of the average "life" of
an investment. This is important when considering the effect of changes in interest rates on
investment portfolios since an investment with a longer term will in general be affected more
drastically by a change in interest rates than an investment with a shorter term.

Example:
How will the price of a conventional gilt that is redeemable at par with an annual coupon of
3% be affected if future rates of interest over all terms increase from 7% to 8%, if the term of
the gilt is (a) 5 years and (b) 25 years. Comment on your results.

Solution: Using the formula P = 3a n + 100v n with n = 5 and n = 25, we find that

• The price of the 5 year stock would fall from $83.60 to $80.04, i.e. a fall of 4.3% - The
price of the 25 year stock would fall from $53.39 to $46.63, i.e. a fall of 12.7%

20
 The change in interest rate has a greater effect on the longer 25 year stock, which has a
DMT of 14.9 years (based on 7% interest), than it has on the shorter 5 year stock, which has
a DMT of 4.7 years.
From (14.7.5) we also have
∆PV(i)
≈ − Vol(i)∆i
PV(i)
This formula corresponds with our example because if volatility (or DMT) increases then the
proportional change in the present value also increases. Thus for a small change in interest
rates a stock with a larger DMT will have a larger proportional present value change (than that
for a smaller DMT stock).
Roughly speaking, a change in interest rates has the same effect on the present value of a
cashflow series as it would have on a zero coupon bond with the same discounted mean term
or volatility.

Example:
A speculator buys large quantities of a fixed-interest security when she expects interest rates to
fall, with the intention of selling in a short time to realize a profit. At present she is choosing
between the following two secure fixed-interest stocks:
Security 1 which bears interest at 5% per annum payable annually in arrear and is repayable
at par in five years time.
Security 2 which bears interest at 11% per annum payable annually in arrear and is repayable
at par in six years time.
These stocks always have the same gross yield per annum, at present 10%. Which should
she buy in order to obtain the larger capital gain on a small fall in interest rates? (Ignore
taxation.) Solution: Calculate the DMT of each stock:

P = 5a + 100v 5 %

= 81.05%
Security 1:
 
5/81.05 5/81.05
∴ DMT = ä5 + 1 − 5
10% 10%
= 4.488

Security 2: P = 11a6 + 100v 6 %

= 104.36%
∴ DMT = 4.725 ∴ Choose security 2.

21
10.0.2 Convexity
Recall that for our general discrete payment stream we have
m
X
PV(i) = Ck v tk
k=0
Xm
P V ′ (i) = − Ck tk v tk +1
k=0
Xm
∴ P V ′′ (i) = + Ck tk (tk + 1) v tk +2
k=0

11 Definition Convexity
′′
The convexity of the payment stream is: Conv(i) = PPVV (i)
(i)

Using Taylor’s approximation to second order we have:

∆PV (∆i)2
≈ − Vol(i)∆i + Conv(i)
PV 2

Examples
1. PV(i) = (1 + i)−n (Zero coupon bond)

⇒ P V ′ (i) = −n(1 + i)−n−1

⇒ P V ′′ (i) = +n(n + 1)(1 + i)−n−2
∴ Conv(i) = n(n + 1)v 2

1. PV(i) = a∞ = 1
i

⇒ PV′ (i) = − i12

⇒ PV′′ (i) = 2
i3
⇒ Conv(i) = i22
3.

22
 0
PV(i) = 2000v 5 + 4000v 10 = 2784.02
PV′ (i) = − (5)(2000)v 6 + (10)(4000)v 11
 

= −19664.50
PV′′ (i) = (30)(2000)v 7 + (110)(4000)v 12
= 170987.05
19664.50
∴ Vol(i) = = 7.0633
2784.02
170987.05
Conv(i) = = 61.4173
2784.02
∆PV ∼ (∆i)2
∴ = −7.0633∆i + 61.4173
PV 2
Conclude that a change of +1% in the value of i will lower the PV by ≈ 6.756%. Using actual
values: PV(i = 11%) = 2595.64
∆PV
∴ = −6.766%
PV
In this case the approximation is quite accurate.

11.0.1 Interpreting convexity as "spread"

In Subsection (14.7.1), the DMT is shown to be a weighted average time:
m
X Ck v tk
DMT = wk tk where wk = .
k=1
PV

Using the idea of statistical variance we may also define a "weighted variance" or spread about
the weighted average by:
m
X
Spread = wk (tk − DMT)2
k=1
m
X
wk t2k − 2tk DMT + DMT2


=
k=1
Xm m
X m
X
2
= wk t2k − 2DMT wk tk + DMT wk
k=1 k=0 k=0
m
!
X
= wk t2k − (DMT)2
k=1
2
P V ′′ (δ) PV′ (δ)

= −
PV PV

23
 d d
=v
di dδ
d2
  
dv d d d
⇒ 2 = +v
di di dδ di dδ
2
2 d 2 d
As = −v + v
dδ dδ 2 
d d2
= v2 − + 2
dδ dδ
PV′′ (i) PV′ (δ) PV′′ (δ)
 
2
∴ Conv(i) = =v − +
PV PV PV
Hence:
Conv(i) = v 2 spread + DMT2 + DMT
 

It follows that if two payment streams have the same DMT (or Vol) at rate i, then the stream
with larger convexity is the one with a higher "spread" of payment dates about the DMT.
Using the approximation:

∆PV (∆i)2
≈ − Vol(i)∆i + Conv(i)
PV 2
we see that for two streams with the same volatility:
∆i < 0 ⇒ the percentage increase in PV is more for the stream with higher Conv.
∆i > 0 ⇒ the percentage decrease in PV is less for the stream with higher Conv.
The stream with the higher convexity is then a better investment option from the point of
view of interest rate risk.

11.0.2 Portfolio immunization

One of the key factors a manager responsible for the investment of a fixed interest portfolio
will be concerned about is how the portfolio would be affected if there was a change in interest
rates and, in particular, whether such a movement might compromise the ability of the fund to
meet its liabilities.
We will now look at the technique of immunization, which is a method of minimizing the
risks relating to interest rates.
Suppose an institution holds assets of value VA , to meet liabilities of VL . Since both VA and
VL represent the discounted value of future cashflows, both are sensitive to the rate of interest.
We assume that the institution is healthy at time 0 so that currently VA ≥ VL .
If VA > VL , then we say that there is a surplus in the fund which is equal to VA − VL .
If VA < VL then the fund is in deficit.
If the rates of interest fall, both VA and VL will increase. If rates of interest rise then both
will decrease. We are concerned with the risk that following a downward movement in interest
rates the value of assets increases by less than the value of liabilities, or that, following an
upward movement in interest rates the value of assets decreases by more than the values of the
liabilities.
In other words, for a fund currently in surplus we are concerned that after a movement in
interest rates the fund moves into deficit.

24
 In order to examine the impact of interest rate movements on different cashflow sequences we
will use changes in the yield to maturity to represent changes in the underlying term structure.
This is approximately (but not exactly) the same as assuming a constant movement of similar
magnitude in the one-period forward rates.
Consider now an investment portfolio with associated future cash income stream and future
cash outflow stream. In this section we treat the two streams as separate positive cash flows
with present values at effective rate of interest i denoted by

PVA (i) · · · income stream (Assets)

PVL (i) · · · outflow stream (Liabilities)

Hence, for the entire transaction, 1

N P V (i) = PVA (i) − PVL (i).

Definition: The portfolio is immunised against small changes in the effective rate i iff:
(i) N P V (i) = 0
(ii) Any small change in i (either positive or negative) leads to a positive N P V .
Note: We have assumed that the effective rate i is independent of the term, i.e. that a flat
term structure prevails. So the above definition refers to immunisation against a small parallel
shift in the level of a flat term structure.

11.0.3 Redington’s Theorem

An investment portfolio is immunised at rate of interest i0 against small changes in the effective
rate i if each of the following conditions hold:
(i) PVA (i0 ) = PVL (i0 )
(ii) VolA (i0 ) = VolL (i0 )
(iii) ConvA (i0 ) > ConvL (i0 )
Proof: (i) ⇒ NPV (i0 ) = 0

PV′A (i0 ) PV′L (i0 )

(ii) ⇒ = ⇒ PV′A (i0 ) = PV′L (i0 ) ⇒ N P V ′ (i0 ) = 0
PVA (i0 ) PVL (i0 )
PV′′A (i0 ) PV′′L (i0 )
(iii) ⇒ > ⇒ PV′′A (i0 ) > PV′′L (i0 ) ⇒ NPV′′ (i0 ) > 0
PVA (i0 ) PVL (i0 )
1
In this context, N P V is sometimes referred to as the SURPLUS at rate i. So, NPV(i) has a
local minimum at i.

25
 Clearly this implies that the portfolio is suitably immunised. Note: Clearly we need only
check:
PVA (i0 ) = P VL (i0 ) , PV′A (i0 ) = PV′L (i0 ) , and PV′′A (i0 ) > PV′′L (i0 )

Example:
Assume that a fund must make payments of $50 000 at the end of the 6 th and 8 th years
respectively. The current term structure is flat with i = 7% per annum.
(a) Show how an immunisation to a small parallel shift in the term structure can be achieved
by purchasing an appropriately chosen combination of a 5 year zero coupon bond and a 10 year
zero coupon bond.
(b) Demonstrate that the above fund is immunised against a parallel shift of 0.5% in the
current term structure by calculating the surplus for i = 6.5% per annum and i = 7.5% per
annum.
Solution:

26
 PVL (7%) = 50000 v 6 + v 8 = 62418

PV′L (7%) = −50000 6v 7 + 8v 9 = −404398

(a)
PV′′L (7%) = 50000 (6)(7)v 8 + (8)(9)v 10

= 3052277.
Note that
404398
DMTL (δ7% ) = × (1.07) = 6.932 years,
62418
and that the assets have been chosen to have a "larger spread" about t = 6.932 than the
liabilities. This seems to be intuitively obvious, but the formula for the spread given in Section
3.10 involves the weights wk which cannot be intuitively assessed. This condition must be
verified mathematically.
For the asset stream:
PVA (7%) = P v 5 + Qv 10
= 0.71299P + 0.50835Q
′
P VA (7%) = − P 5v 6 + Q10v 11
 

= −[3.3317P + 4.7509Q]
′′
P VA (7%) = P 30v 7 + Q110v 12
= 18.6825P + 48.8413Q.
Using Redington’s conditions:
    
0.71299 0.50835 P 62418
=
3.3317 4.7509 Q 404398
⇒ P = R53710 and Q = R47454

27
Final check: P VA′′ (7%) = 3321152 > 3052250 and so the fund is indeed immunised.
   
1 1
PVA 6 % − PVL 6 %
2 2
=64482 − 64478 = 4 > 0
   
1 1
PVA 7 % − PVL 7 %
2 2
=60437 − 60433 = 4 > 0

12 Example:
An investor purchases a fixed interest security paying 12 yearly coupons in arrear at 10% per
annum. The stock will be redeemed at 110% in 20 years time. The investors ITR = 25%.
There is no capital gains tax.
(a) Calculate the price paid by the investor to obtain a net yield of 10% per annum effective.
(b) Calculate the volatility of the security at 10% per annum effective.
(c) The investor has two liability obligations of equal amount with the second liability due
in 10 years. The present value of these two liabilities is equal to the value of the investors
holding in the above security.
(i) What is the maturity of the first liability if the volatility of the liability portfolio is the
same as the volatility of the asset portfolio?
(ii) Calculate the convexity of the liability portfolio.
(iii) Without any further calculations discuss the likely outcome of assessing Redington’s
conditions to determine if the investors position is immunised against small movements in the
effective rate.

13 Solution:
(2)
PV = 100P = (1 − 25%)10a20 10% + 110v 20
⇒ P = 81.7608%
i(2)
(b) To calculate P V ′ (i), change to 1
2
yearly period using rate j = 2
. Then

i(2)
 
′ d′
P V (i) = P V (j)
di 2

28
  2
i(2)
Now 1 + 2
= 1 + i. Differentiating with respect to i :

i(2) d i(2)
   
2 1+ · =1
2 di 2
d i(2)
 
1 1
∴ =  = 1
di 2 2 1+ i
(2)
2(1 + i) 2
2

As PV (j) = (1 − 25%)5a40j + 110v@j 40

 
′ 1 41
⇒ −P V (j) = (3.75)(Ia)40j j + (110)(40)v@j
1+j
At j = 4.88088% ⇒ −P V ′ (j) = 904.5711 + 623.5950
1
∴ −P V ′ (i) = v@j (−P V ′ (j))
2
= 728.5246
728.5246
Vol(i) = = 8.9104
81.7608
40
As PV(j) = (1 − 25%)5a40|j + 110v@j
(c) (i)

X = amount of each liability per 100 nominal of asset held.

PVL = X v t + v 10 = 81.7608

−P VL′ (i) = X tv t+1 + 10v 11 = 728.5246

PV′L tv t+1 + 10v 11

∴ VolL (i) = − = = 8.9104
PVL v t + v 10
Rearranging this equation:

v t+1 ((8.9104)(1.1) − t) = v 11 (10 − (8.9104)(1.1))

⇒ v t (9.80144 − t) = 0.07655

This nonlinear equation for t must be solved numerically: Clearly t < 9.80144
Guess: t ≈ 9.5 ⇒ LHS = 0.12189.

29
 Interpolating between (9.80144, 0) and (9.5, 0.12189) gives t = 9.61213. (⇒ LHS = 0.07574).
Interpolating between (9.5, 0.12189) and (9.61213, 0.07574) gives t = 9.61016(⇒ LHS =
0.07654) OK.
t+2 12
(ii) ConvL (i) = t(t+1)vvt +v+110v
10 = 87.5273
(iii) As PVA = PVL and VolA = VolL at i = 10% we need to check ConvA > 87.5273 at i =
10%. This is likely to be the case as the spread of assets about DMT = 8.9104 × 1.1 = 9.80144
is greater than for liabilities.

13.0.1 Problems with Immunisation

In practice there are difficulties with implementing an immunisation strategy based on these
principles. For example, the method requires continuous rebalancing of portfolios to keep
the asset and liability volatilties equal. The asset portfolio required to provide Redington
immunisation normally depends on the initial interest rate. Once the interest rates have moved
away from the initial rate, it may be necessary to "rebalance" the portfolio so that it is once
again immunised at the new rate. This throws a spanner into the practical application of the
technique except in very simple situations.
Other limitations of immunisation include:

• There may be options or other uncertainties in the assets or in the liabilities, making the
assessment of the cashflows approximate rather than known.

• Assets may not exist to provide the necessary overall asset volatility to match the liability
volatility.

• The theory relies upon small changes in interest rates. The fund may not be protected
against large changes.

• The theory assumes a flat yield curve and requires the same change in interest rates at
all terms. In practice this is rarely the case.

• Immunisation removes the likelihood of making large profits.

Despite these problems, immunisation theory remains an important consideration in the selec-
tion of assets.
In practice, actuaries making investment decisions are aware of Redington’s theory in a
general sense. For example, they are aware of the consequences of investing "long" (i.e. holding
assets with a higher DMT than the liabilities), but they would not normally apply the theory
directly. A more open-ended technique called asset-liability modelling is often used instead.

30