included the view that yields on

Treasuries, being articially depressed

owing to lack of supply, may no longer
represent fair value. This raised the
question of possible replacement
measures of value. The alternative
benchmarks he cited included the
government agency mortgage bond
market (Ginnie Mae) as well as the
federal agency market (Fannie Mae),
Literature
Several studies have highlighted an
environment of lower liquidity in
government bond markets, with yield
levels falling below true risk-free rates.
Fleming (2000) highlighted the
implications of the shrinking market in
US Treasuries, caused by a reduction in
issuance, and the implications this caused
for benchmark interest rates.
1
This
Palgrave Macmillan Ltd 1479-179X/06 $30.00 Vol. 7, 1, 1721 Journal of Asset Management 17
An alternative bond relative value
measure: Determining a fair value
of the swap spread using Libor
and GC repo rates
Received: 28th June, 2005
Moorad Choudhry
is Head of Treasury at KBC Financial Products in London, and a Visiting Professor at the Department of Economics, London
Metropolitan University.
KBC Financial Products, 111 Old Broad Street, London EC2N 1FP, UK
Tel: +44 020 7614 6428; E-mail: [email protected]
Abstract During the 1990s, a reduced issuance in US Treasury, UK gilt and other
sovereign bond markets resulted in the government bond yield curve being viewed as a
potentially biased benchmark or indicator of fair value. This led to bond market
investors starting to use the interest-rate swap curve as their main reference benchmark
for valuation and analysis purposes. This reected practitioners views that continued
exclusive reliance on government yields may lead to inaccurate analysis. The
interest-rate swap curve is viewed as an acceptable alternative measurement
benchmark. It is also used as a comparator instrument, against which the yields on
government bonds can be measured. As the interest-rate swap market is considered to
be sufciently liquid and exposing users to inter-bank credit risk, considered a relatively
low level of default risk, the swap curve can be viewed to be trading at a fair value to
the theoretical zero-coupon yield curve. Such being the case, it is a key issue for
investors to have comfort that the swap curve is fairly valued at all times. This enables
them to use swap spreads over the government curve as a reasonable measure of the
liquidity and fair value of government bonds. This paper provides an expression for the
fair value for the swap spread, which is shown to be a function of the Libor rate and
the general collateral repo rate. This spread can then be used as an alternative
benchmark measure to assess relative value for all debt market instruments.
Keywords: asset-swap, Libor, fair value, repo, general collateral, swap spread
although these markets presented their
own issues with regard to liquidity and
accessibility. McCauley and Remolona
(2000) considered the supply and
demand factor determinant of
government yields and suggested that a
lack of supply, resulting in lower
liquidity, could render the true value of
such yields problematic. The issue was
considered by the Bank for International
Settlements (BIS), who investigated the
impact of squeezes on government bond
supply (see Jeanneau and Scott, 2001).
In a later article, Fleming (2001)
suggested that the interest-rate swap
curve could serve as an alternative
benchmark. This was considered for the
euro-denominated market as well as the
Treasury market by Schinasi et al. (2001),
who recommended the interest-rate swap
curve as an alternative benchmark to the
government curve.
Under these circumstances, bond
market investors, concerned with
identifying a fair value measure of
government yields as well as a relative
value measure for corporate bonds, need
to identify a fair value for the swap
spread.
Interest rate swaps
In setting fair value in the debt market,
the only plausible alternative to the
government yield curve in the euro and
sterling market appears to be the interest
rate swap. A conventional interest rate
swap is an over-the-counter derivative
contract between two parties, in which
one party makes xed-interest payments,
calculated on a notional amount, while
the other party makes oating-rate
interest payments. The xed rate is set at
the inception of the contract, and the
oating-rate is linked to an external
reference such as Libor during the life of
the swap.
2
The xed-rate payer in a swap is
conceptually long a oating-rate bond
(FRN), and short a xed coupon bond;
the opposite is true for the oating-rate
payer. When calculating the fair value
swap rate, the concept of the par swap is
used, which is a vanilla swap of zero net
present value. This means that the
present value of the cashows of the
xed-leg and oating-leg of the swap are
equal. The xed leg of the swap is
identical to the interest payments of a
bond with xed coupon equal to the
swap rate, while the oating leg is
identical to the cash ows of an FRN
with coupon set at the Libor rate. At the
start of the contract, the FRN and the
xed bond must be equal to the notional
swap value. In a par swap, therefore, the
xed leg is conceptually similar to a
xed coupon bond priced at par, with a
coupon equal to the swap rate. The
swap rate in a par swap is therefore
known as the par yield.
The observed difference between the
swap rate and the government bond
yield at any point on the yield curve
is known as the swap spread. The
swap spread is a measure of inter-bank
risk, given that the government curve
is assumed to be default free, and the
swap market is used by banks and
corporates to hedge their interest rate
risk. Traditionally, the swap spread was
taken to be the measure of corporate
default risk, with a corporate bond
asset-swap spread taken as an indicator
of the credit risk on that reference
bond. The swap market, however, is
now very large and liquid, and does
not suffer from illiquidity, even out to
long-dated maturities. There are also
no supply constraints in the swap
market, unlike for (say) long-dated gilts
or Treasuries, and the use of
collateralisation, margining, netting and
other measures has substantially reduced
counterparty risk. Government bond
markets, on the other hand, have
18 Journal of Asset Management Vol. 7, 1, 1721 Palgrave Macmillan Ltd 1479-179X/06 $30.00
Choudhry
Libor rate. Bidoffer spreads are ignored.
A market participant enters into the
following trades simultaneously:
a short sale of ex million of the
benchmark [5 or 10]-year semi-annual
paying government bond, which is
priced at par and therefore trading at
the par yield of rm per cent; the bond
is available for funding in the repo
market at the general collateral (GC)
rate
a six-month reverse repo, using the
cash proceeds from the bond sale; this
pays out the six-month GC repo rate
rg per cent;
a ex million [5 or 10]-year pay
oating/receive xed interest-rate swap,
where the trader receives six-month
Libor or rl and pays at the swap rate rs
per cent.
This sets up a zero-cost portfolio, which
comprises the following cash ows:
pay out of rm every six months
receipt of rg every six months
pay out of rs every six months
receipt of rl every six months
The net cash ow of the combined
portfolio during the time it is held is
given by
Port nom

rs rm
2

rl rg
2

(1)
where nom is the notional value equal to
ex million.
It can be shown that the present value
of the portfolio cash ows during its life
is given by
PV
n

i=1
x
2
(rs rm)

1
rz
i
2

i
E
0
n

i=1
x
2
(rl
i
rg
i
)

1
s
i
2

i
(2)
where n is the maturity (in years) of the
experienced low liquidity and supply
constraints, leading to inverted curves,
causing some commentators to suggest
that the government yields have traded
below the true risk-free level; see
Fleming (2000, 2001). In other words,
the government curve may on occasion
be overvalued, whereas the swap curve
can be regarded as lying at fair value.
The assumption that the swap curve is
always at fair value, whereas the
government curve at times may not be,
is open to some debate. This is not least
because the swap curve is a function of
market makers bidoffer swap rates and,
in the event of a signicant shock or
market correction, the book position and
supply of certain prices will inuence the
swap curve, affecting its use as a
benchmark.
3
If this assumption is made,
however, one is then in a position to
make an assessment of the swap spread
and, by taking its fair value, of the
government curve fair value as well.
Assuming that the swap rate is at fair
value, one can use the extent of the
swap spread itself to determine whether
the government curve is at fair value; if
the swap spread differs from that
suggested by the swap rate fair value, this
indicates a government curve that is
overvalued.
The next section considers an
alternative expression for the fair value
for the swap spread, using an approach
that enables practitioners to use actual
market rates in the form of Libor and
repo rates.
Swap spread fair value
We now consider the calculation of the
fair value for the swap spread. This is
done using a zero-cost portfolio trading
approach, consisting of the benchmark
government bond (paying xed coupon
on a semi-annual basis), the six-month
government repo rate and the six-month
Palgrave Macmillan Ltd 1479-179X/06 $30.00 Vol. 7, 1, 1721 Journal of Asset Management 19
An alternative bond relative value measure
bond and swap, and x is the nominal
value as before. The term rz
i
is the
theoretical i-year risk-free spot interest
rate.
The rst term in the expression is the
present value of the cash ows of the
bond and the xed leg of the swap. The
second term is the present value of the
expected difference between the oating
rate of the swap and the repo rate. These
are not known with certainty, as they
will be set by the Libor xing every six
months and the repo rate as this is rolled
over at the same time and for the same
term. As these cash ows cannot be
determined at the inception of the trade,
they are discounted at the rate of s
i
,
which is the spot rate rz
i
plus an
additional element t, where t represents
the risk factor arising from the change in
the spread rl rg. The additional amount
t is user determined. So one has
s
i
rz
i
t (3)
It is assumed that the swap rate curve
and the benchmark government bond are
fairly valued. This being so, the
expression at (2) must be equal to zero,
because the portfolio set up by the trader
is a zero-cost transaction. This enables us
to set
n

i=1
x
2
(rs rm)

1
rz
i
2

i
E
0
n

i=1
x
2
(rl
i
rg
i
)

1
s
i
2

i
(4)
One can now set the swap spread in
terms of the expected spread between
the Libor and GC repo rates. This is
done by rearranging (4) and moving
(rs rm) outside the summation,
shown as
rs rm
E
0
n

i=1

(rl
i
rg
i
)/

1
s
i
2

i=1

1/

1
rz
i
2

i

(5)
That is, the fair value of the swap spread
is given by the expected spread between
the Libor rate and the repo rate. By
setting a risk factor for this spread t as
zero, the fair value of the swap spread is
given by the weighted average of future
expectations of the spread during the
term of the swap. As the Libor and repo
rates can be observed with transparent
accuracy in the market, adopting these as
determinants of the fair-value swap rate
carries less uncertainty for investors.
Conclusion
It has been shown that, under certain
assumptions of the swap rate and the
future spread of the six-month Libor rate
and six-month repo rate, the fair value of
the swap spread is given by the
expectation of this spread. The spread
was given as (5) above.
Market practitioners must set the level
of the risk premium factor. Under
normal market conditions, as one
moves to a market correction or market
shock, the swap spread will widen.
Therefore, as the swap rate must be
higher when markets are experiencing
downside movement, the risk premium
will be negative. This gives a higher
swap spread at the time of market
downside correction.
On the basis discussed, the swap
spread may be used as an alternative
benchmark measure to assess relative
value for all debt market instruments.
That is, by determining where the
market-observed swap spread is relative
to that suggested by the fair value swap
spread, one can observe the actual
over-valuation of benchmark government
bonds as captured by the government
yield curve.
Disclaimer
The views, thoughts and opinions expressed in this
paper represent those of Moorad Choudhry in his
20 Journal of Asset Management Vol. 7, 1, 1721 Palgrave Macmillan Ltd 1479-179X/06 $30.00
Choudhry
3. Events such as the LTCM crisis or the emerging
market fall outs of 1997 and 1998, for example.
References
Fleming, M. (2000) The Benchmark US Treasury
Market: Recent Performance and Possible
Alternatives, Federal Reserve Bank of New York:
Economic Policy Review, no. 1, April.
Fleming, M. (2001) Financial Market Implications of
the Federal Debt Paydown, Federal Reserve Bank of
New York: Staff Report, no. 120, May.
Jeanneau, S. and Scott, R. (2001) Anatomy of a
Squeeze, BIS Quarterly Review, June, 323.
McCauley, R. and Remonola, E. (2000) Size and
Liquidity of Government Bond Markets, BIS
Quarterly Review, November, 528.
Schinasi, G., Kramer, C. and Todd Smith, R. (2001)
Financial Implications of the Shrinking Supply of
US Treasury Securities, International Monetary
Fund, 20th March.
individual private capacity, and should not be taken to
be the views of KBC Financial Products or KBC Bank
N.V., or of Moorad Choudhry as an ofcer, employee
or representative of KBC Financial Products or KBC
Bank N.V.
Notes
1. New issuance of US Treasuries has subsequently
risen again, including a reversion to issuance of
30-year securities that the Federal Reserve had
called a halt to in the mid-1990s. The requirement
for alternative benchmarks remains, however,
particularly during times of market correction or
illiquidity.
2. The typical sterling swap pays semi-annual interest.
Only the net difference between the two interest
amounts is exchanged. The notional or principal
amount is not exchanged.
Palgrave Macmillan Ltd 1479-179X/06 $30.00 Vol. 7, 1, 1721 Journal of Asset Management 21
An alternative bond relative value measure