Chapter 9

The term structure of interest rates

1. Introduction
Throughout this guide so far, we have only ever considered ‘the’ interest rate, that is
the one set by the monetary authorities, or that is allowed to change to clear the money
markets if the money supply is the instrument of choice. In reality, there are a large number
of interest rates, from those on debt that mature overnight to interest rates on debt that
mature up to 30 years in the future. This chapter will examine in more detail the links
between short-term and long-term interest rates, explaining why such links are important
and providing theories that explain the short-term–long-term interest rate relationship.

2. Aims
This chapter will introduce the relevance of a rich variety of interest rates in financial
and macroeconomic decisions. We will also discuss different term structure theories.

3. Learning outcomes
By the end of this chapter, and having completed the Essential reading and activities,
you should be able to:
• explain what a yield curve is and why it is important for policy makers
• list and discuss the different empirical regularities of the yield curve
• list and describe the three different theories of the term structure outlined in this
chapter, noting any differences and similarities and stating which theories can explain
which empirical facts
• describe the relationship between rates of return and bond prices and why this is so
important for the absence of arbitrage conditions on which the expectations hypoth-
esis focuses.

125
4. Reading advice
The main readings for this section are the chapters in the textbooks by Goodhart (1989),
Harris (1985) and especially Mishkin (2003). Chapter 7 of Mishkin should be read before
starting this part of the subject as it starts by giving a general introduction to the ideas
behind the term structure and then develops the relevant theory. The two entries in the
New Palgrave Dictionary by Malkiel and Mishkin are very useful but the chapter by Shiller
in the Handbook of Monetary Economics is difficult and should only be read if you feel
comfortable with the material.

5. Essential reading
Goodhart, C.A.E. Money, Information and Uncertainty. (London: Macmillan, 1989)
Chapter 11.

Harris, L. Monetary Theory. (New York; London: McGraw-Hill, 1985) Chapter 17.

Malkiel, B.G. ‘The term structure of interest rates’, in Newman, P., M. Milgate and
J. Eatwell (eds) The New Palgrave Dictionary of Money and Finance. (London:
Macmillan, 1994).

Mishkin, F.S. The Economics of Money, Banking and Financial Markets. (Boston,
Mass.; London: Addison Wesley, 2003) Chapter 7.

Mishkin, F.S. ‘The yield curve’, in Newman, P., M. Milgate and J. Eatwell (eds) The
New Palgrave Dictionary of Money and Finance. (London: Macmillan, 1994).

126
6. Further reading
Books

Shiller, R.J. ‘The term structure of interest rates’, in Friedman, B. and F. Hahn (eds)
Handbook of Monetary Economics. (Amsterdam: North-Holland, 1990).

Walsh, C.E. Monetary Theory and Policy. (Cambridge, Mass.: MIT Press, 2003)
Chapter 10.

Journal articles

Aksoy, Y. and H. Basso, ‘Liquidity, term spreads and monetary policy’, Economic
Journal, 124, 2014, 1234–1278.

Mankiw, N.G. and L.H. Summers ‘Do long-term interest rates overreact to short-term
interest rates?’, Brookings Papers on Economic Activity (1984) 1, pp.223–247.

McCallum, B.T. ‘Monetary policy and the term structure of interest rates’, National
Bureau of Economic Research working paper, w4938, (1994).

Shiller, R.J. ‘The volatility of long-term interest rates and expectations models of the
term structure’, Journal of Political Economy 87(5, Part 2) 1979, 1190–1219.

7. The yield curve

Governments issue a number of differently dated bonds, from short-term paper through
to very long-dated bonds that mature in a number of years in the future. Each bond will
pay a different rate of return, also known as yield to maturity. If we plot on a graph the
time to maturity of government debt on the horizontal axis and the yield to maturity on the
vertical axis, this will give us what is known as the yield curve. An example of such a curve
is shown in Figure 9.1.
Figure 9.1 shows that, for this example, government debt that matures in one month’s
time pays an (annualised) interest rate of 4%. One-year paper pays 6% and debt that ma-
tures in 10 years from now pays an annual rate of return of 9%.

Activity (9.1)

What does the yield curve look like in your country?

127
 9%

Yield to maturity
6%

4%

1 month 1 year 10 years

Time to maturity

Figure 9.1:

8. Why is the yield curve of importance to policy-makers?

Central Banks, when setting interest rates, only set one interest rate, normally at the
very short end of the spectrum. In the US, the Federal Reserve sets overnight rates. How-
ever, investment decisions and aggregate demand in the economy will tend to depend on
long-term interest rates since firms will compare the rate of return on investment projects
that accrue over the entire life of the project (15 years plus), to the alternative of buying
equivalent dated bonds maturing in 15 years’ time or so. If aggregate demand depends
on long-term interest rates and the monetary authorities set short-term rates but wish to
affect aggregate demand, they will need to know the relationship between short-term and
long-term rates of return (i.e. they need to know the shape of the yield curve).
But why then do Central Banks not just set long-term rates of return and allow the
market to determine the rates of return along the rest of the yield curve? If the authorities
set long-term rates, they may be faced with large capital losses due to fluctuations of bond
prices in the market. To avoid the possibility of facing large price variations, the authorities
instead set the rate of return on short-dated paper, on which the possibility of capital gains/
losses is very small or non-existent. However, not only is the yield curve important for
analysing the effects of monetary policy on the economy, it is also important because it
gives information on expected future inflation, as will be explained below.

9. Bond prices and the rate of return

There are, in general, two types of bonds a government can issue. These are:

• Coupon bonds. Typically, coupon bonds promise to pay a fixed sum of money, the
coupon, every period, whether it is every month, quarter or year, and also promise to
pay the holder of the bond its face value on maturity. For example, a ten-year bond
may be bought from the government for £100 that promises to pay £5 (the coupon

128
 payment) every year for the next ten years. In the final year, both the last £5 coupon
and the maturity value of the bond, £100, are paid.

• Discount bonds. Discount bonds do not offer any interest or coupon payments. In-
stead, they are sold at a ‘discount’ and pay a larger amount on maturity. For example,
a ten-year discount bond may be sold today for £60 that pays no coupons at all but
pays £100 on maturity, ten years from now.

Consider a coupon bond that pays C every year from now to infinity. Such bonds, without
a maturity date, are called perpetuities or Consols (the UK term). The price of this bond,
P, will be the future cash flow, discounted back to today by the interest rate, R, assumed
constant here for convenience.
!
C C C C 1 C
P= + + +··· = = . (9.1)
1 + R (1 + R) 2 (1 + R) 3 1 + R 1 − 1/(1 + R) R

There is then a negative relationship between the price of a bond and the interest rate. This
was discussed in Chapter 3 when examining Keynes individual money demand function.
(For bonds that do have a maturity date, the relationship between bond prices and the
interest rate, although still negative, is not as simple as (9.1). Intuitively, if the interest rate
increased, people would be willing to pay less for a bond that pays a fixed coupon payment
each period. If a perpetuity bond that paid a coupon of £5 had a price of £100, this implies
the market interest rate is 5%. If the market interest rate doubled to 10%, no investor would
be willing to pay more than £50 for the same bond since the £5 coupon on a £50 bond is
10%. Hence the bond price and interest rate are negatively related.

10. Empirical regularities of the term structure

There are three features of the term structure that any theory of the yield curve should
be able to explain:

1. Rates of return on short- and long-dated bonds move together over time.

2. When short rates are low, the yield curve is likely to be upward sloping and vice
versa.

3. Yield curves generally have a persistent upward slope.

We now examine a number of theories that try to explain these features.

11. The expectations hypothesis

The expectations hypothesis of the yield curve links the rate of return on short-term
bonds to that on long-term bonds, essentially by assuming that short and long-dated bonds

129
are perfect substitutes. Consider someone who wishes to save a fixed amount of money
for n periods. She could either buy long-dated bonds now, at date t, that pay t Rt+n per
cent per year. The left-hand subscript denotes the time when the bond is bought and the
right-hand subscript denotes when the bond will mature. Alternatively, she could buy a
bond that matures at date t + 1 (a one-period bond) paying a rate of return t rt+1 . When
this matures she will buy another one-period bond (at date t + 1 that matures at date t + 2)
paying a rate of return t+1 rt+2 and will continue doing this until date t + n. If the return from
a portfolio made up of long-dated bonds is not identically equal to the expected return from
continually rolling over one-period debt, then arbitrage opportunities will be exploited to
bring the two returns together.
Suppose for example that a two-year bond paid a rate of return of 4% per year. If the
rate of return on a one-year bond was 3% today and expected to remain at 3% next year,
investors will rush in to buy the longer-dated bond as it pays a higher rate of return. The
increased demand for the two period bonds will push the price up and, as explained above,
will push the rate of return down from 4%. Also, the reduction in demand for one-period
bonds will push the price down, causing the one-period rate of return to increase from 3%.
In equilibrium, under the expectations hypothesis, the total return from each portfolio must
be equal to avoid arbitrage opportunities. Using the same notation as above, but noting that
at date t when the saving decision is made, all future one-period rates are not known and
have to be estimated, then the discount bond reads:

(1 + t Rt+n )n = (1 + t rt+1 )(1 + t+1 rt+2

e
)(1 + t+2 rt+3
e
) · · · (1 + t+n−1 rt+n
e
). (9.2)

The left-hand side is the total return from holding n period bonds until they mature. The
right-hand side is the total expected return from holding and continually rolling over one-
period bonds, n times. The rates of return on all bonds bought at date t + 1 onwards are
not known at date t, hence the e (expectation) superscript. Taking logs of (9.2), noting that
ln(1 + X) ≈ X for small X, gives:
n · t Rt+n = t rt+1 + t+1 rt+2 + t+2 rt+3 + . . . + t+n−1 rt+n
e e e

t+n−1
1 X e
⇒ t Rt+n = sr . (9.3)
n s=t s+1

This is one of the main results of the expectations hypothesis; the long-term interest rate is
an average of the current and all future expected short-term interest rates. The expectations
hypothesis does have a number of criticisms, however:
• It cannot explain the empirical fact that the yield curve has a persistent upward slope.
If the long rate is an average of current and expected future short rates, this can only
be explained if the short rate increases through time. This clearly is not the case.

• If the long rate is an average of current and future short rates, the long rate must be a
smoother series (when plotted through time) than the short rate. This is not true; the
long rate is just as volatile.

130
 • In order to avoid arbitrage opportunities and keep total returns equal, if the rate of
return on long-term bonds is greater than the return on short bonds, then holding
long-dated bonds must be accompanied by a capital loss (i.e. the price of long-term
bonds must fall). If the price of long-term bonds falls, the rate of return must increase
still further in the next period. Putting this another way: if the long rate is higher than
the short rate, the long rate must increase. This does not happen in reality.

The expectations hypothesis and expected inflation

If the long rate is an average of the current and expected future short rates, then an
upward-sloping yield curve suggests that the short rate is expected to increase in the future
(see (9.3)). If the real interest rate is constant then the increase in expected future nominal
interest rates must be associated with an increase in expected future inflation, as per the
Fisher equation. The greater the slope of the yield curve, the more short-term rates are
expected to rise and so the faster is inflation expected to increase.

12. The segmentation hypothesis

Whereas the expectations hypothesis assumes short- and long-term bonds are perfect
substitutes so the decision as to whether to hold long- or short-dated debt depends entirely
on expected returns, the segmentation hypothesis assumes short- and long-term bonds are
not substitutes in any way. The rate of return on m period bonds will not depend on the
market for, or the return on, m − j period bonds at all. Instead, its rate of return will
depend entirely on the demand for and supply of credit that matures in m periods’ time. If
the demand for m period bonds increased, caused by more people wanting to save for m
periods, then the price of m period bonds will increase and the rate of return will fall. The
markets for m, m + 1, m + 2, etc. period debt are said to be segmented.
Whereas the expectations hypothesis could not explain the persistent upward sloping
nature of the yield curve, this feature can be easily explained if we use the segmentation
hypothesis. If people generally prefer to hold short-dated debt (i.e. save by buying short-
term bonds), then this will cause the price of short-term bonds to be high, relative to long-
term bonds, and so the rate of return on short-term debt will then be lower than the long-
term rate, implying an upward-sloping yield curve.

Activity (9.2)

Why do you think people would prefer to hold short-term, rather than long-term,
debt?
(Hint: think of the desire to lock up money in long-term bonds and the risk of
capital gains or losses when such holdings have to be liquidated.)

131
 Despite being able to explain persistent upward-sloping yield curves, the segmenta-
tion hypothesis cannot explain the fact that interest rates move together since the markets
are completely segmented. Also, the theory cannot explain why yield curves are upward
(downward)-sloping when short rates are low (high).

13. Preferred habitat theory

The preferred habitat theory lies in between the expectations and segmentations hy-
potheses. It assumes bonds are neither fully substitutable nor non-substitutable. Instead,
people have a preferred bond maturity they wish to hold (as in the segmentation hypoth-
esis) but will be willing to move to another bond maturity if the gains from doing so are
significant (so exploiting excess arbitrage opportunities as in the expectations hypothesis).
The preferred habitat theory can be represented as (9.4):
t+n−1
1 X e
t Rt+n = sr + t kn . (9.4)
n s=t s+1
This is exactly the same as the expectations hypothesis, (9.3), except for the inclusion of
a term premium, t kn . If n period bonds were not the bond of choice, then people would
need an extra rate of return in order to encourage them to hold such debt. In this case
t kn would be positive. If people generally tend to prefer short-dated debt then the term
premium will be a monotonic function of maturity; in order to hold longer and longer dated
debt, an increasing term premium must be offered. When combined with the ‘expectations
hypothesis’ component, (9.4) can explain the persistent upward slope of the yield curve,
along with the other empirical regularities: the co-movement of short and long rates, and the
fact the yield curve tends to be upward (downward)-sloping when short rates are low (high).
However, using the term premium to fix the problems of the expectations and segmentations
hypotheses, is arguably a not a proper solution. A theory of the term premium should
be provided, rather than simply assumed in order to make the data consistent with the
‘fixed’ theory. For a recent paper on a theory of term premium that builds upon maturity
transformation risk faced by the banks, see Aksoy and Basso (2014).

14. A reminder of your learning outcomes

By the end of this chapter, and having completed the Essential reading and activities,
you should be able to:
• explain what a yield curve is and why it is important for policy makers
• list and discuss the different empirical regularities of the yield curve
• list and describe the three different theories of the term structure outlined in this
chapter, noting any differences and similarities and stating which theories can explain
which empirical facts

132
 • describe the relationship between rates of return and bond prices and why this is so
important for the absence of arbitrage conditions on which the expectations hypoth-
esis focuses.

15. Sample examination questions

Section A
Specify whether the following statement is true, false or uncertain. Explain your answer
in a short paragraph.

1. ‘Under the segmentation hypothesis, the slope of the yield curve tells us nothing
about expected inflation.’

Section B
2. Empirical fact: yield curves tend to have an especially steep upward slope when short
rates are low and a downward slope when short-term interest rates are high. Can you
explain this fact with the theories of the term structure analysed in this chapter?

3. Assuming that the expectations hypothesis is the correct theory of the term structure,
calculate the interest rates for one to five year bonds and plot the resulting yield
curves when one year interest rates over the next five years look as follows:

(a) 5%, 7%, 7%, 7%, 7%.

(b) 5%, 4%, 4%, 4%, 4%.

How would your yield curves change if people preferred shorter-term bonds over
longer-term bonds?

4. If the yield curve was steeply upward sloping, what is the market predicting about
the movements of future short-term interest rates? What might the market predict
about the inflation rate in the future?

5. How can the preferred habitat theory explain the appearance of downward-sloping
yield curves if t kn , the term premium, is positive?

16. Feedback to Sample examination questions

Section B
3. From the expectations hypothesis, and in particular (9.3), we can calculate the 2, 3,
4 and 5-year interest rates from the one year rates given in the question.

133
(a) Using (9.3), we see that:

2 year rate, t Rt+2 = 1/2(5 + 7) = 6% per year

3 year rate, t Rt+3 = 1/3(5 + 7 + 7) = 6.33% per year

The 4- and 5- year rates can be calculated in a similar way to obtain t Rt+4 = 6.5%
per year and t Rt+5 = 6.6% per year. The yield curve is therefore upward sloping
(notice that an upward-sloping yield curve suggests interest rates will rise in the
future; 5% to 7%).
(b) Again using (9.3), we can calculate the 2, 3, 4 and 5-year interest rates to ob-
tain t Rt+2 = 4.5% per year, t Rt+3 = 4.33% per year, t Rt+4 = 4.25% per year
and t Rt+5 = 4.2% per year. The yield curve is therefore downward sloping (a
downward-sloping yield curve suggests interest rates will fall; 5% to 4%).
If people preferred shorter-term bonds then they would have to receive an in-
creased rate of return on longer-term debt (over and above that implied by the
expectations hypothesis) in order to encourage them to hold such assets. In
case (a), the yield curve would be even steeper and in part (b), the yield curve
will also be rotated anticlockwise, so becoming more flat, or perhaps becoming
positively sloped if the term premium was large enough.

134