Duration is only
so endurable
Duration has dominated bond fund management for
nearly a hundred years yet, unsurprisingly, it is only
now being seen as outdated. There is another way of
measuring the risk of a bond portfolio and generating
greater alpha, and that is to use forward rates instead
BY RUSS OXLEY, HEAD
OF RATES, IGNIS ASSET
MANAGEMENT
Government bonds are the
most basic assets traded
on the capital markets and
yet frequently they are not
seen as a major source of
alpha within an investment
portfolio. This is the result
of many bond managers
using an outdated duration-based
approach that
significantly impairs their
alpha generation potential.
Applying modern portfolio
theory means an alternative
approach is possible.
The theory behind bond
portfolio management has
a long history dating back
to the 1920s, predating the
development of modern
portfolio theory by Harry
Markowitz in the early
1950s. It even predates the
large scale bond issuance of
the US federal government.
The challenge for an
investor back then was to
measure the interest rate risk
of a bond, how the bond’s
value would be affected
by a change in interest
rates. There was no coherent
portfolio management
theory, and mean-variance
optimisation had not yet
been invented. Frederick
Macaulay proposed a solution:
duration, defined as
the measure of how much
the price of a bond will
change for a given change
in its interest rate.
l Inventing duration
To understand this, consider
a ten-year zero-coupon
bond. The interest rate on
the bond can be viewed
as being made up of a
sequence of ten one-year
rates. To understand the
price effect of a change in
the ten-year rate, Macaulay
made a natural assumption:
that the ten one-year
rates all changed by the
same amount.
Not surprisingly, therefore,
he concluded that the
price of a ten-year bond
changes ten times as much
as the price of a one-year
bond for a given interest
rate change – it has duration
of ten years. The fact
that the price change is
proportional to the time to
maturity of the bond motivates
the term ‘duration’.
This concept of duration
has dominated bond fund
management since the ’20s,
and has remained largely
unchanged and unchallenged
by the great theo-
retical advances in portfolio
theory that have occurred
since Markowitz got the
ball rolling in the ’50s.
Markowitz’s theory, and
subsequent refinements,
has had a profound impact
on the way fund managers
run equity portfolios.
Amazingly, however, the
key result of the theory has
had little effect on bond
portfolio managers.
This key result is that a
fund manager’s ability to
generate alpha depends not
only on their skill but also on
the number of uncorrelated
decisions they take using
that skill. The more uncorrelated
decisions taken, the
greater is the potential for
alpha generation.
l Inconsistent usage
The fundamental contradiction
is that the use of duration
to measure risk exposures
is completely inconsistent
with Markowitz’s
principles. Using duration
to measure the risk exposure
of a bond portfolio
removes the possibility of
generating extra alpha from
making a greater number
of uncorrelated decisions
across the yield curve.
To understand this,
FIXED INCOME FORWARD RATES
-10
-15
BarCap £ non-gilts 5+ Yr
-20
IBOXX £ non-gilts 5-10 Yr
Nov ’06 May ’07 Nov May ’08 Nov
Source: Morningstar
JANUARY 2010 [www.portfolio-adviser.com] PORTFOLIO ADVISER
%
Forward rates
P’mance of gilt vs non-gilt indices - 3 years
25
20
15
10
5
0
-5
IBOXX £ gilts 5-10 Yr
May ’09
Performance of gilt vs non-gilt indices
3m 6m 1yr 3yr 5yr 10yr
BarCap £ non-gilts 5+yr 8.69 20.47 20.17 1.28 3.53 5.23
IBOXX £ gilts 5-10 yr 2.01 0.72 11.67 7.49 6.39 –
IBOXX £ non-gilts 5-10yr 7.71 21.85 16.59 1.27 2.93 –
ML UK gilts 5-10 yr 2.04 0.73 11.76 7.51 6.40 6.34
MSCI World
Source: Morningstar
6.49 11.35 13.27 -3.47 2.77 -1.49
P’mance of gilt vs non-gilt indices - 1 year
25
20
15
10
%
5
0
-5
IBOXX £ gilts 5-10 Yr
-10
Nov ’08 Jan ’09 Mar
Source: Morningstar
BarCap £ non-gilts 5+ Yr
May
Jul
Sep
Nov
IBOXX £ non-gilts 5-10 Yr
Nov
43