THE LEHMAN BROTHERS

GUIDE TO EXOTIC CREDIT DERIVATIVES
T
H
E

L
E
H
M
A
N

B
R
O
T
H
E
R
S
G
U
I
D
E

T
O

E
X
O
T
I
C

C
R
E
D
I
T

D
E
R
I
V
A
T
I
V
E
S
lehman cover.qxd 10/10/2003 11:03 Page 1
Effective Structured Credit
Solutions for our Clients
With over seventy professionals
worldwide, Lehman Brothers gives you
access to top quality risk-management,
structuring, research and legal
expertise in structured credit. The team
combines local market knowledge with
global co-ordinated expertise.
Lehman Brothers has designed specific
solutions to our clients’ problems,
including yield-enhancement, capital
relief, portfolio optimisation, complex
hedging and asset-liability
management.
.
Credit Default Swaps
.
Portfolio Swaps
.
Credit Index Products
.
Repackagings
.
Default Baskets
.
Secondary CDO trading
.
Customised CDO tranches
.
Default swaptions
.
Credit hybrids
For further information please contact
your local sales representative or call:
London: Giancarlo Saronne
+44 20 7260 2745 gsaronne@lehman.com
New York: Mike Glover
+1 212 526 7090 mglover@lehman.com
Tokyo: Jawahar Chirimar
81-3-5571-7257 jchirima@lehman.com
Structured Credit Solutions
Product Innovation
All Rights Reserved. Member SIPC. Lehman Brothers International (Europe) is regulated by the Financial Services Authority. ©2003 Lehman Brothers Inc.
Leadership in Fixed Income
Research
Document1 06/10/2003 09:54 Page 1
The Lehman Brothers Guide to Exotic Credit Derivatives 1
The credit derivatives market has revolu-
tionised the transfer of credit risk. Its impact
has been borne out by its significant growth
which has currently achieved a market notion-
al close to $2 trillion. While not directly com-
parable, it is worth noting that the total
notional outstanding of global investment
grade corporate bond issuance currently
stands at $3.1 trillion.
This growth in the credit derivatives market
has been driven by an increasing realisation
of the advantages credit derivatives possess
over the cash alternative, plus the many new
possibilities they present to both credit
investors and hedgers. Those investors seek-
ing diversification, yield pickup or new ways
to take an exposure to credit are increasingly
turning towards the credit derivatives market.
The primary purpose of credit derivatives is
to enable the efficient transfer and repack-
aging of credit risk. In their simplest form,
credit derivatives provide a more efficient
way to replicate in a derivative format the
credit risks that would otherwise exist in a
standard cash instrument.
More exotic credit derivatives such as syn-
thetic loss tranches and default baskets cre-
ate new risk-return profiles to appeal to the
differing risk appetites of investors based on
the tranching of portfolio credit risk. In doing
so they create an exposure to default correla-
tion. CDS options allow investors to express
a view on credit spread volatility, and hybrid
products allow investors to mix credit risk
views with interest rate and FX risk.
More recently, we have seen a stepped
increase in the liquidity of these exotic credit
derivative products. This includes the devel-
opment of very liquid portfolio credit vehicles,
the arrival of a two-way correlation market in
customised CDO tranches, and the develop-
ment of a more liquid default swaptions mar-
ket. To enable this growth, the market has
developed new approaches to the pricing and
risk-management of these products.
As a result, this book is divided into two
parts. In the first half, we describe how exotic
structured credit products work, their ratio-
nale, risks and uses. In the second half, we
review the models for pricing and risk manag-
ing these various credit derivatives, focusing
on implementation and calibration issues.
Foreword
Authors
Dominic O'Kane
T. +44 207 260 2628
E. dokane@lehman.com
Marco Naldi
T. +1 212 526 1728
E. mnaldi@lehman.com
Sunita Ganapati
T. +1 415 274 5485
E. sganapati@lehman.com
Arthur Berd
T. +1 212 526 2629
E. arthur.berd@lehman.com
Claus Pedersen
T. +1 212 526 7775
E. cmpeders@lehman.com
Lutz Schloegl
T. +44 207 260 2113
E. luschloe@lehman.com
Roy Mashal
T. +1 212 526 7931
E. rmashal@lehman.com
guide.qxd 10/10/2003 11:15 Page 1
2 The Lehman Brothers Guide to Exotic Credit Derivatives
Contents
Foreword 1
Credit Derivatives Products
Market overview 3
The credit default swap 4
Basket default swaps 8
Synthetic CDOs 12
Credit options 23
Hybrid products 28
Credit Derivatives Modelling
Single credit modelling 31
Modelling default correlation 33
Valuation of correlation products 39
Estimating the dependency structure 43
Modelling credit options 47
Modelling hybrids 51
References 53
guide.qxd 10/10/2003 11:15 Page 2
The Lehman Brothers Guide to Exotic Credit Derivatives 3
Market overview
The credit derivatives market has changed
substantially since its early days in the late
1990s, moving from a small and highly eso-
teric market to a more mainstream market
with standardised products. Initially driven
by the hedging needs of bank loan man-
agers, it has since broadened its base of
users to include insurance companies,
hedge funds and asset managers.
The latest snapshot of the credit deriva-
tives market was provided in the 2003 Risk
Magazine credit derivatives survey. This sur-
vey polled 12 dealers at the end of 2002,
composed of all the major players in the
credit derivatives market. Although the
reported numbers cannot be considered
‘hard’, they can be used to draw fairly firm
conclusions about the recent direction of
the market.
According to the survey, the total market
outstanding notional across all credit deriva-
tives products was calculated to be $2,306
billion, up more than 50% on the previous
year. Single name CDS remain the most
used instrument in the credit derivatives
world with 73% of market outstanding
notional, as shown in Figure 1. This supports
our observation that the credit default mar-
ket has become more mainstream, focusing
on the liquid standard contracts. We believe
that this growth in CDS has been driven by
hedging demand generated by synthetic
CDO positions, and by hedge funds using
credit derivatives as a way to exploit capital
structure arbitrage opportunities and to go
outright short the credit markets.
An interesting statistic from the survey is
the relatively equal representation of North
American and European credits. The survey
showed that 40.1% of all reference entities
originate in Europe, compared with 43.8%
from North America. This is in stark con-
trast to the global credit market which has
a significantly smaller proportion of
European originated bonds relative to
North America.
The base of credit derivatives users has
been broadening steadily over the last few
years. We show a breakdown of the market
by end-users in Figure 2 (overleaf). Banks
still remain the largest users with nearly
50% share. This is mainly because of their
substantial use of CDS as hedging tools for
their loan books, and their active participa-
tion in synthetic securitisations. The hedg-
ing activity driven by the issuance of
synthetic CDOs (discussed later) has for
the first time satisfied the demand to buy
protection coming from bank loan hedgers.
Readers are referred to Ganapati et al (2003)
for a full discussion of the market impact.
Insurance companies have also become
an important player, mainly by investing in
investment-grade CDO tranches. As a result,
Credit Derivatives Products
Portfolio/
correlation
products
22%
Credit default
swaps
73%
Total return
swaps
1%
Credit linked
notes
3% Options and
hybrids
1%

Figure 1. Market breakdown by
instrument type
Source: Risk Magazine 2003 Credit Derivatives Survey
guide.qxd 10/10/2003 11:15 Page 3
4 The Lehman Brothers Guide to Exotic Credit Derivatives
the insurance share of credit derivatives
usage has increased to 14% from 9% the
previous year.
More recently, the growth in the usage of
credit derivatives by hedge funds has had a
marked impact on the overall credit deriva-
tives market itself, where their share has
increased to 13% over the year. Hedge
funds have been regular users of CDS espe-
cially around the convertible arbitrage strate-
gy. They have also been involved in many of
the ‘fallen angel’ credits where they have
been significant buyers of protection. Given
their ability to leverage, they have substan-
tially increased their volume of CDS con-
tracts traded, which in many cases has been
disproportionate to their absolute size.
Finally, in portfolio products, by which we
mean synthetic CDOs and default baskets,
the total notional for all types of credit
derivatives portfolio products was $449.4
billion. Their share has kept pace with the
growth of the credit derivatives market at
about 22% over the last two years. This is
not a surprise, since there is a fundamen-
tally symbiotic relationship between the
synthetic CDO and single name CDS mar-
kets, caused by dealers originating synthet-
ic tranches either by issuing the full capital
structure or hedging bespoke tranches.
Since this survey was published, the credit
derivatives market has continued to consoli-
date and innovate. The ISDA 2003 Credit
Derivative Definitions were another milestone
on the road towards CDS standardisation.
The year 2003 has also seen a significant
increase in the usage of CDS portfolio prod-
ucts. There has been a stepped increase in
liquidity for correlation products, with daily
two-way markets for synthetic tranches now
being quoted. The credit options market, in
particular the market for those written on
CDS, has grown substantially.
A number of issues still remain to be
resolved. First, there is a need for the gener-
ation of a proper term structure for credit
default swaps. The market needs to build
greater liquidity at the long end and, espe-
cially, the short end of the credit curve.
Greater transparency is also needed around
the calibration of recovery rates. Finally, the
issue of the treatment of restructuring
events still needs to be resolved. Currently,
the market is segregated along regional lines
in tackling this issue, but it is hoped that a
global standard will eventually emerge.
The credit default swap
The credit default swap is the basic building
block for most ‘exotic’ credit derivatives and
hence, for the sake of completeness, we set
out a short description before we explore
more exotic products.
A credit default swap (CDS) is used to trans-
fer the credit risk of a reference entity (corpo-
rate or sovereign) from one party to another.
In a standard CDS contract one party pur-
chases credit protection from the other party,
to cover the loss of the face value of an asset
following a credit event. A credit event is a
legally defined event that typically includes
Hedge
funds

13%
Insurance
14%
SPVs
5%
Banks
(synthetic
securitisation)
10%
Banks
(other)
38%
Reinsurance
10%
Corporates
3%
Third-party
asset managers
7%

Figure 2. Breakdown by end users
Source: Risk Magazine 2003 Credit Derivatives Survey.
guide.qxd 10/10/2003 11:15 Page 4
The Lehman Brothers Guide to Exotic Credit Derivatives 5
bankruptcy, failure to pay and restructuring.
Buying credit protection is economically
equivalent to shorting the credit risk. Equally,
selling credit protection is economically
equivalent to going long the credit risk.
This protection lasts until some specified
maturity date. For this protection, the pro-
tection buyer makes quarterly payments, to
the protection seller, as shown in Figure 3,
until a credit event or maturity, whichever
occurs first. This is known as the premium
leg. The actual payment amounts on the pre-
mium leg are determined by the CDS spread
adjusted for the frequency using a basis
convention, usually Actual 360.
If a credit event does occur before the
maturity date of the contract, there is a pay-
ment by the protection seller to the protec-
tion buyer. We call this leg of the CDS the
protection leg. This payment equals the dif-
ference between par and the price of the
assets of the reference entity on the face
value of the protection, and compensates the
protection buyer for the loss. There are two
ways to settle the payment of the protection
leg, the choice being made at the initiation of
the contract. They are:
Physical settlement – This is the most wide-
ly used settlement procedure. It requires the
protection buyer to deliver the notional
amount of deliverable obligations of the ref-
erence entity to the protection seller in
return for the notional amount paid in cash.
In general there are several deliverable obli-
gations from which the protection buyer can
choose which satisfy a number of character-
istics. Typically they include restrictions on
the maturity and the requirement that they
be pari passu – most CDS are linked to
senior unsecured debt.
If the deliverable obligations trade with dif-
ferent prices following a credit event, which
they are most likely to do if the credit event
is a restructuring, the protection buyer can
take advantage of this situation by buying
and delivering the cheapest asset. The pro-
tection buyer is therefore long a cheapest to
deliver option.
Cash settlement – This is the alternative to
physical settlement, and is used less fre-
quently in standard CDS but overwhelming-
ly in tranched CDOs, as discussed later. In
cash settlement, a cash payment is made by
the protection seller to the protection buyer
equal to par minus the recovery rate of the
reference asset. The recovery rate is calcu-
lated by referencing dealer quotes or
observable market prices over some period
after default has occurred.
Suppose a protection buyer purchases
five-year protection on a company at a CDS
spread of 300bp. The face value of the pro-
tection is $10m. The protection buyer
therefore makes quarterly payments ap-
proximately (we ignore calendars and day
count conventions) equal to $10m × 0.03
× 0.25 = $75,000. After a short period the
reference entity suffers a credit event.
Assuming that the cheapest deliverable
asset of the reference entity has a recovery
price of $45 per $100 of face value, the pay-
ments are as follows:
Contingent payment of loss on par
following a credit event (protection leg)


Protection
buyer

Protection
seller

Default swap spread
(premium leg)
Figure 3. Mechanics of a CDS
guide.qxd 10/10/2003 11:15 Page 5
6 The Lehman Brothers Guide to Exotic Credit Derivatives
■ The protection seller compensates the
protection buyer for the loss on the face
value of the asset received by the protec-
tion buyer and this is equal to $5.5m.
■ The protection buyer pays the accrued
premium from the previous premium
payment date to the time of the credit
event. For example, if the credit event
occurs after a month then the protection
buyer pays approximately $10m × 300bp
× 1/12 = $25,000 of premium accrued.
Note that this is the standard for corpo-
rate reference entity linked CDS.
For severely distressed reference entities,
the CDS contract trades in an up-front for-
mat where the protection buyer makes a
cash payment at trade initiation which pur-
chases protection to some specified maturi-
ty – there are no subsequent payments
unless there is a credit event in which the
protection leg is settled as in a standard
CDS. For a full description of up-front CDS
see O’Kane and Sen (2003).
Liquidity in the CDS market differs from
the cash credit market in a number of ways.
For a start, a wider range of credits trade in
the CDS market than in cash. In terms of
maturity, the most liquid CDS is the five-year
contract, followed by the three-year, seven-
year and 10-year. The fact that a physical
asset does not need to be sourced means
that it is generally easier to transact in large
round sizes with CDS.
Uses of a CDS
The CDS can do almost everything that cash
can do and more. We list some of the main
applications of CDS below.
■ The CDS has revolutionised the credit
markets by making it easy to short credit.
This can be done for long periods without
assuming any repo risk. This is very use-
ful for those wishing to hedge current
credit exposures or those wishing to take
a bearish credit view.
■ CDS are unfunded so leverage is possi-
ble. This is also an advantage for those
who have high funding costs, because
CDS implicitly lock in Libor funding to
maturity.
■ CDS are customisable, although devia-
tion from the standard may incur a liquid-
ity cost.
■ CDS can be used to take a spread view
on a credit, as with a bond.
■ Dislocations between cash and CDS pre-
sent new relative value opportunities.
This is known as trading the default
swap basis.
Evolution of CDS documentation
The CDS is a contract traded within the legal
framework of the International Swaps and
Derivatives Association (ISDA) master agree-
ment. The definitions used by the market for
credit events and other contractual details
have been set out in the ISDA 1999 document
and recently amended and enhanced by the
ISDA 2003 document. The advantage of this
standardisation of a unique set of definitions
is that it reduces legal risk, speeds up the con-
firmation process and so enhances liquidity.
Despite this standardisation of defini-
tions, the CDS market does not have a uni-
versal standard contract. Instead, there is a
US, European and an Asian market stan-
dard, differentiated by the way they treat a
restructuring credit event. This is the con-
sequence of a desire to enhance the posi-
guide.qxd 10/10/2003 11:15 Page 6
The Lehman Brothers Guide to Exotic Credit Derivatives 7
tion of protection sellers by limiting the
value of the protection buyer’s delivery
option following a restructuring credit
event. A full discussion and analysis of
these different standards can be found in
O’Kane, Pedersen and Turnbull (2003).
Determining the CDS spread
The premium payments in a CDS are
defined in terms of a CDS spread, paid peri-
odically on the protected notional until
maturity or a credit event. It is possible to
show that the CDS spread can, to a first
approximation, be proxied by either (i) a par
floater bond spread (the spread to Libor at
which the reference entity can issue a float-
ing rate note of the same maturity at a price
of par) or (ii) the asset swap spread of a
bond of the same maturity provided it
trades close to par.
Demonstrating these relationships relies
on several assumptions that break down in
practice. For example, we assume a com-
mon market-wide funding level of Libor, we
ignore accrued coupons on default, we
ignore the delivery option in the CDS, and
we ignore counterparty risk. Despite these
assumptions, cash market spreads usually
provide the starting point for where CDS
spreads should trade. The difference
between where CDS spreads and cash
LIBOR spreads trade is known as the
Default Swap Basis, defined as:
Basis = CDS Spread – Cash Libor Spread.
A full discussion of the drivers behind the
CDS basis is provided in O’Kane and
McAdie (2001). A large number of
investors now exploit the basis as a rela-
tive value play.
Determining the CDS spread is not the
same as valuing an existing CDS contract.
For that we need a model and a discussion of
the valuation of CDS is provided on page 32.
Funded versus unfunded
Credit derivatives, including CDS, can be
traded in a number of formats. The most
commonly used is known as swap format,
and this is the standard for CDS. This format
is also termed ‘unfunded’ format because
the investor makes no upfront payment.
Subsequent payments are simply payments
of spread and there is no principal payment
at maturity. Losses require payments to
be made by the protection seller to the
protection buyer, and this has counterparty
risk implications.
The other format is to trade the risk in the
form of a credit linked note. This format is
known as ‘funded’ because the investor has
to fund an initial payment, typically par. This
par is used by the protection buyer to pur-
chase high quality collateral. In return the pro-
tection seller receives a coupon, which may
be floating rate, ie, Libor plus a spread, or
may be fixed at a rate above the same matu-
rity swap rate. At maturity, if no default has
occurred the collateral matures and the
investor is returned par. Any default before
maturity results in the collateral being sold,
the protection buyer covering his loss and the
investor receiving par minus the loss. The
protection buyer is exposed to the default risk
of the collateral rather than the counterparty.
Traded CDS portfolio products
CDS portfolio products are products that
enable the investor to go long or short the
credit risk associated with a portfolio of CDS
in one transaction.
In recent months, we have seen the emer-
gence of a number of very liquid portfolio
products, whose aim is to offer investors a
diverse, liquid vehicle for assuming or hedg-
guide.qxd 10/10/2003 11:15 Page 7
8 The Lehman Brothers Guide to Exotic Credit Derivatives
ing exposure to different credit markets, one
example being the TRAC-X
SM
vehicle. These
have added liquidity to the CDS market and
also created a standard which can be used
to develop portfolio credit derivatives such
as options on TRAC-X.
The move of the CDS market from banks
towards traditional credit investors has greatly
increased the need for a performance bench-
mark linked directly to the CDS market. As a
consequence, Lehman Brothers has intro-
duced a family of global investment grade CDS
indices which are discussed in Munves (2003).
These consist of three sub-indices, a US
250 name index, a European 150 name index
and a Japanese 40 name index. All names
are corporates and the maturity of the index
is maintained close to five years. Daily pric-
ing of all 440 names is available on our
LehmanLive website.
Basket default swaps
Correlation products are based on redistribut-
ing the credit risk of a portfolio of single-
name credits across a number of different
securities. The portfolio may be as small as
five credits or as large as 200 or more credits.
The redistribution mechanism is based on the
idea of assigning losses on the credit portfo-
lio to the different securities in a specified pri-
ority, with some securities taking the first
losses and others taking later losses. This
exposes the investor to the tendency of
assets in the portfolio to default together, ie,
default correlation. The simplest correlation
product is the basket default swap.
A basket default swap is similar to a CDS,
the difference being that the trigger is the
nth credit event in a specified basket of ref-
erence entities. Typical baskets contain five
to 10 reference entities. In the particular case
of a first-to-default (FTD) basket, n=1, and it
is the first credit in a basket of reference
credits whose default triggers a payment to
the protection buyer. As with a CDS, the con-
tingent payment typically involves physical
delivery of the defaulted asset in return for a
payment of the par amount in cash. In return
for assuming the nth-to-default risk, the pro-
tection seller receives a spread paid on the
notional of the position as a series of regular
cash flows until maturity or the nth credit
event, whichever is sooner.
The advantage of an FTD basket is that it
enables an investor to earn a higher yield
than any of the credits in the basket. This is
because the seller of FTD protection is lever-
aging their credit risk.
To see this, consider that the fair-value
spread paid by a credit risky asset is deter-
mined by the probability of a default, times
the size of the loss given default. FTD bas-
kets leverage the credit risk by increasing the
probability of a loss by conditioning the pay-
off on the first default among several credits.
The size of the potential loss does not
increase relative to buying any of the assets
in the basket. The most that the investor can
lose is par minus the recovery value of the
FTD asset on the face value of the basket.
The advantage is that the basket spread
paid can be a multiple of the spread paid by
the individual assets in the basket. This is
shown in Figure 4 where we have a basket
of five investment grade credits paying an
average spread of about 28bp. The FTD bas-
ket pays a spread of 120bp.
More risk-averse investors can use default
baskets to construct lower risk assets: sec-
ond-to-default (STD) baskets, where n=2,
trigger a credit event after two or more assets
have defaulted. As such they are lower risk
second-loss exposure products which will
pay a lower spread than an FTD basket.
TRAC-X is a service mark of JPMorgan and Morgan Stanley
guide.qxd 10/10/2003 11:15 Page 8
The Lehman Brothers Guide to Exotic Credit Derivatives 9
The basket spread
One way to view an FTD basket is as a
trade in which the investor sells protection
on all of the credits in the basket with the
condition that all the other CDS cancel at
no cost following a credit event. Such a
trade cannot be replicated using existing
instruments. Valuation therefore requires
a pricing model. The model inputs in order
to determine the nth-to-default basket
spread are:
■ Value of n: An FTD (n=1) is riskier than an
STD (n=2) and so commands a higher
spread.
■ Number of credits: The greater the num-
ber of credits in the basket, the greater
the likelihood of a credit event, and so the
higher the spread.
■ Credit quality: The lower the credit quali-
ty of the credits in the basket, in terms of
spread and rating, the higher the spread.
■ Maturity: The effect of maturity de-
pends on the shape of the individual
credit curves.
■ Recovery rate: This is the expected
recovery rate of the nth-to-default asset
following its credit event. This has only a
small effect on pricing since a higher
expected recovery rate is offset by a
higher implied default probability for a
given spread. However, if there is a
default the investor will certainly prefer a
higher realised recovery rate.
■ Default correlation: Increasing default
correlation increases the likelihood of
assets to default or survive together. The
effect of default correlation is subtle and
significant in terms of pricing. We now
discuss this is more detail.
Baskets and default correlation
Baskets are essentially a default correlation
product. This means that the basket spread
depends on the tendency of the reference
assets in the basket to default together.
It is natural to assume that assets issued
by companies within the same country and
industrial sector should have a higher
default correlation than those within differ-
ent industrial sectors. After all, they share
the same market, the same interest rates
and are exposed to the same costs. At a
global level, all companies are affected by
the performance of the world economy.
We believe that these systemic sector risks
far outweigh idiosyncratic effects so
we expect that default correlation is
usually positive.
There are a number of ways to explain how
default correlation affects the pricing of
default baskets. Confusion is usually caused
by the term ‘default correlation’. The fact is




Lehman
Brothers












Basket
investor
Contingent payment
of par minus recovery
on FTD on $10m face value



120bp paid on $10m
until FTD or five-year
maturity, whichever
is sooner
Reference portfolio
Coca Cola 30bp
C. de Saint Gobain 30bp
Electricidade de Portugal 27bp
Hewlett Packard 29bp
Teliasonera 30bp
Figure 4. Five-year first to default (FTD)
basket on five credits. We show the five
year CDS spreads of the individual credits
guide.qxd 10/10/2003 11:15 Page 9
10 The Lehman Brothers Guide to Exotic Credit Derivatives
that if two assets are correlated, they will
not only tend to default together, they will
also tend to survive together.
There are two correlation limits in which a
FTD basket can be priced without resorting
to a model – independence and maximum
correlation.
■ Independence: Consider a five-credit
basket where all of the underlying credits
have flat credit curves. If the credits are
all independent and never become corre-
lated during the life of the trade, the nat-
ural hedge is for the basket investor to
buy CDS protection on each of the indi-
vidual names to the full notional. If a
credit event occurs, the CDS hedge cov-
ers the loss on the basket and all of the
other CDS hedges can be unwound at no
cost, since they should on average have
rolled down their flat credit curves. This
implies that the basket spread for
independent assets should be equal to
the sum of the spreads of the names in
the basket.
■ Maximum correlation: Consider the
same FTD basket but this time where the
default correlation is at its maximum. In
practice, this means that when any asset
defaults, the asset with the widest
spread will always default too. As a
result, the risk of one default is the same
as the risk of the widest spread asset
defaulting. Because an FTD is triggered
by only one credit event, it will be as risky
as the riskiest asset and the FTD basket
spread should be the widest spread of
the credits in the basket.
The best way to understand the behaviour
of default baskets between these two cor-
relation limits is to study the loss distribu-
tion for the basket portfolio. See page 33
for a discussion of how to model the loss
distribution.
We consider a basket of five credits with
spreads of 100bp and an assumed recovery
rate for all of 40%. We have plotted the loss
distribution for correlations of 0%, 20%, and
50% in Figure 5. The spread for an FTD bas-
ket depends on the probability of one or
more defaults which equals one minus the
probability of no defaults. We see that the
probability of no defaults increases with
increasing correlation – the probability of
credits surviving together increases – and
the FTD spread should fall.
The risk of an STD basket depends on the
probability of two or more defaults. As corre-
lation goes up from 0–20%, the probability of
two, three, four and five defaults increases.
This makes the STD spread increase.
The process for translating these loss dis-
tributions into a fair value spread requires a
model of the type described on page 39.
Essentially we have to find the basket
spread for which the present value of the
protection payments equals the present
value of the premium payments.
We should not forget that in addition to the
0
10
20
30
40
50
60
70
80
0 1 2 3 4 5
Number of defaults
P
r
o
b
a
b
i
l
i
t
y

(
%
)

ρ = 0% ρ = 20% ρ = 50%
Figure 5. Loss distribution for a
five-credit basket with 0%, 20% and
50% correlation
guide.qxd 10/10/2003 11:15 Page 10
The Lehman Brothers Guide to Exotic Credit Derivatives 11
protection leg, the premium leg of the
default basket also has correlation sensitivi-
ty because it is only paid for as long as the
nth default does not occur.
Using a model we have calculated the cor-
relation sensitivity of the FTD and STD spread
for the five-credit basket shown in Figure 6.
At low correlation, the FTD spread is close to
146bp, which is the sum of the spreads. At
high correlation, the basket has the risk of the
widest spread asset and so is at 30bp. The
STD spread is lowest at zero correlation since
the probability of two assets defaulting is low
if the assets are independent. At maximum
correlation the STD spread tends towards the
spread of the second widest asset in the bas-
ket which is also 30bp.
Applications
■ Baskets have a range of applications.
Investors can use default baskets to lever-
age their credit exposure and so earn a
higher yield without increasing their
notional at risk.
■ The reference entities in the basket are all
typically investment grade and so are
familiar to most credit analysts.
■ The basket can be customised to the
investors’ exact view regarding size,
maturity, number of credits, credit selec-
tion, FTD or STD.
■ Buy and hold investors can enjoy the
leveraging of the spread premium. This is
discussed in more detail later.
■ Credit investors can use default baskets
to hedge a blow-up in a portfolio of cred-
its more cheaply than buying protection
on the individual credits.
■ Default baskets can be used to express a
view on default correlation. If the
investor’s view is that the implied correla-
tion is too low then the investor should sell
FTD protection. If implied correlation is too
high they should sell STD protection.
Hedging default baskets
The issuers of default baskets need to
hedge their risks. Spread risk is hedged by
selling protection dynamically in the CDS
market on all of the credits in the default
basket. Determining how much to sell,
known as the delta, requires a pricing model
to calculate the sensitivity of the basket
value to changes in the spread curve of the
underlying credit.
Although this delta hedging should immu-
nise the dealer’s portfolio against small
changes in spreads, it is not guaranteed to be
a full hedge against a sudden default. For
instance, a dealer hedging an FTD basket
where a credit defaults with a recovery rate of
R would receive a payment of (1-R)F from the
protection seller, and will pay D(1-R)F on the
hedged protection, where F is the basket face
value and D is the delta in terms of percent-
age of face value. The net payment to the
protection buyer is therefore (1-D)(1-R)F.
0
20
40
60
80
100
120
140
0 10 20 30 40 50 60 70 80 90 100
Correlation (%)
B
a
s
k
e
t

s
p
r
e
a
d

(
b
p
)
FTD
STD

Figure 6. Correlation dependence of
spread for FTD and STD basket
guide.qxd 10/10/2003 11:15 Page 11
12 The Lehman Brothers Guide to Exotic Credit Derivatives
There will also probably be a loss on the
other CDS hedges. The expected spread
widening on default on the other credits in
the basket due to their positive correlation
with the defaulted asset will result in a loss
when they are unwound. The greater unwind
losses for baskets with higher correlations
will be factored into the basket spread.
One way for a default basket dealer to
reduce his correlation risk is by selling pro-
tection on the same or similar default bas-
kets. However this is difficult as it is usually
difficult to find protection buyers who select
the exact same basket as an investor.
The alternative hedging approach is for the
dealer to buy protection using default bas-
kets on other orders of protection. This is
based on the observation that a dealer who
is long first, second, third up to Mth order
protection on an M-credit basket has almost
no correlation risk, since this position is
almost economically equivalent to buying
full face value protection using CDS on all M
credits in the basket.
Figure 7 shows an example basket with
the delta and spread for each of the five
credits. Note that the deltas are all very
similar. This reflects the fact that all of the
assets have a similar spread. Differences
are mainly due to our different correlation
assumptions.
Hedgers of long protection FTD baskets
are also long gamma. This means that as the
spread of an asset widens, the delta will
increase and so the hedger will be selling
protection at a wider spread. If the spread
tightens, then the delta will fall and the
hedger will be buying back hedges at a
tighter level. So spread volatility can be ben-
eficial. This effect helps to offset the nega-
tive carry associated with hedged FTD
baskets. This is clear in the previous exam-
ple where the income from the hedges is
211bp, lower than the 246bp paid to the FTD
basket investor.
Different rating agencies have developed
their own model-based approaches for the
rating of default baskets. We discuss these
on page 39.
Synthetic CDOs
Synthetic collateralised debt obligations
(Synthetic CDOs) were conceived in 1997 as
a flexible and low-cost mechanism for trans-
ferring credit risk off bank balance sheets.
The primary motivation was the banks’
reduction of regulatory capital.
More recently, however, the fusion of cred-
it derivatives modelling techniques and
derivatives trading have led to the creation of
a new type of synthetic CDO, which we call
a customised CDO, which can be tailored to
the exact risk appetites of different classes
of investors. As a result, the synthetic CDO
has become an investor-driven product.
Overall, these different types of synthetic
CDO have a total market size estimated by
the Risk 2003 survey to be close to $500 bil-
lion. What is also of interest is that the deal-
er-hedging of these products in the CDS
market has generated a substantial demand
to sell protection, balancing the traditional
protection-buying demand coming from
bank loan book managers.
Figure 7. Default basket deltas for a
€10m notional five-year FTD basket on
five credits. The FTD spread is 246bp.
Reference entity CDS Spread Delta
Walt Disney 62bp 6.26m
Rolls Royce 60bp 6.55m
Sun Microsystems 60bp 6.87m
Eastman Chemical 60bp 7.16m
France Telecom 64bp 7.57m
guide.qxd 10/10/2003 11:15 Page 12
The Lehman Brothers Guide to Exotic Credit Derivatives 13
The performance of a synthetic CDO is
linked to the incidence of default in a portfo-
lio of CDS. The CDO redistributes this risk by
allowing different tranches to take these
default losses in a specific order. To see this,
consider the synthetic CDO shown in Figure
8. It is based on a reference pool of 100
CDS, each with a €10m notional. This risk is
redistributed into three tranches; (i) an equi-
ty tranche, which assumes the first €50m of
losses, (ii) a mezzanine tranche, which take
the next €100m of losses, and (iii) the senior
tranche with a notional of €850m takes all
remaining losses.
The equity tranche has the greatest risk
and is paid the widest spread. It is typically
unrated. Next is the mezzanine tranche
which is lower risk and so is paid a lower
spread. Finally we have the senior tranche
which is protected by €150m of subordina-
tion. To get a sense of the risk of the senior
tranche, note that it would require more than
25 of the assets in the 100 credit portfolio to
default with a recovery rate of 40% before
the senior tranche would take a principal
loss. Consequently the senior tranche is typ-
ically paid a very low spread.
The advantage of CDOs is that by chang-
ing the details of the tranche in terms of its
attachment point (this is the amount of sub-
ordination below the tranche) and width, it is
possible to customise the risk profile of a
tranche to the investor’s specific profile.
Full capital structure synthetics
In the typical synthetic CDO structured
using securitisation technology, the spon-
soring institution, typically a bank, enters
into a portfolio default swap with a Special
Purpose Vehicle (SPV). This is shown in
Figure 9 (overleaf).
The SPV typically provides credit protec-
tion for 10% or less of the losses on the
reference portfolio. The SPV in turn issues
notes in the capital markets to cash collat-
eralise the portfolio default swap with the
originating entity. The notes issued can
include a non-rated ‘equity’ piece, mezza-
nine debt and senior debt, creating cash lia-
bilities. The remainder of the risk, 90% or
more, is generally distributed via a senior
swap to a highly rated counterparty in an
unfunded format.
Reinsurers, who typically have AAA/AA rat-
ings, have traditionally had a healthy appetite
for this type of senior risk, and are the largest
participants in this part of the capital structure
– often referred to as super-senior AAAs or
super-senior swaps. The initial proceeds from
the sale of the equity and notes are invested
in highly rated, liquid assets.
If an obligor in the reference pool defaults,
the trust liquidates investments in the trust
and makes payments to the originating enti-
ty to cover default losses. This payment is
offset by a successive reduction in the equi-
ty tranche, then the mezzanine and finally the
super-seniors are called to make up losses.
See Ganapati et al (2001) for more details.
Mechanics of a synthetic CDO
When nothing defaults in the reference port-
folio of the CDO, the investor simply

Reference
pool


100 invest-
ment grade
names in
CDS format

€10m x 100
assets = 1bn
total notional




Senior
tranche

€850m
5bp


Equity

tranche €50m



Mezzanine
tranche €100m








Lehman
Brothers
200bp

1,500bp
Contingent payment
Figure 8. A standard synthetic CDO
guide.qxd 10/10/2003 11:15 Page 13
14 The Lehman Brothers Guide to Exotic Credit Derivatives
receives the Libor spread until maturity and
nothing else changes. Using the synthetic
CDO described earlier and shown in Figure
8, consider what happens if one of the ref-
erence entities in the reference portfolio
undergoes the first credit event with a 30%
recovery, causing a €7m loss.
The equity investor takes the first loss of
€7m, which is immediately paid to the orig-
inator. The tranche notional falls from €50m
to €43m and the equity coupon, set at
1500bp, is now paid on this smaller notion-
al. These coupon payments therefore fall
from €7.5m to 15% times €43m = €6.45m.
If traded in a funded format, the €3m
recovered on the defaulted asset is either
reinvested in the portfolio or used to reduce
the exposure of the senior-most tranche
(similar to early amortisation of senior
tranches in cash flow CDOs).
The senior tranche notional is decreased by
€3m to €847m, so that the sum of protect-
ed notional equals the sum of the collateral
notionals which is now €990m. This has no
effect on the other tranches.
This process repeats following each cred-
it event. If the losses exceed €50m then the
mezzanine investor must bear the subse-
quent losses with the corresponding reduc-
tion in the mezzanine notional. If the losses
exceed €150m, then it is the senior investor
who takes the principal losses.
The mechanics of a standard synthetic
CDO are therefore very simple, especially
compared with traditional cash flow CDO
waterfalls. This also makes them more easi-
ly modelled and priced.
The CDO tranche spread
The synthetic CDO spread depends on a
number of factors. We list the main ones and
describe their effects on the tranche spread.
■ Attachment point: This is the amount of
subordination below the tranche. The
higher the attachment point, the more
defaults are required to cause tranche
principal losses and the lower the tranche
spread.
■ Tranche width: The wider the tranche
for a fixed attachment point, the more
losses to which the tranche is exposed.
However, the incremental risk ascending


Reference portfolio
$1bn

notional

CDS spread
income
Equity notes

(unrated)

Senior notes
AAA



Special
purpose
vehicle

(SPV)

Credit
protection

Mezzanine notes
BBB/A







Sponsoring
bank

Super
senior swap
premium
$900m
super
senior
credit
protection




Highly
rated
counterparty

Subordinated
swap
premium

10% first
loss
subordinated
credit
protection

Proceeds
Issued
notes
Figure 9. The full capital structure synthetic CDO
guide.qxd 10/10/2003 11:15 Page 14
The Lehman Brothers Guide to Exotic Credit Derivatives 15
the capital structure is usually declining
and so the spread falls.
■ Portfolio credit quality: The lower the
quality of the asset portfolio, measured
by spread or rating, the greater the risk of
all tranches due to the higher default
probability and the higher the spread.
■ Portfolio recovery rates: The expected
recovery rate assumptions have only a
secondary effect on tranche pricing. This
is because higher recovery rates imply
higher default probabilities if we keep the
spread fixed. These effects offset each
other to first order.
■ Swap maturity: This depends on the
shapes of the credit curves. For upward
sloping credit curves, the tranche curve
will generally be upward sloping and so
the longer the maturity, the higher the
tranche spread.
■ Default correlation: If default correlation
is high, assets tend to default together
and this makes senior tranches more
risky. Assets also tend to survive togeth-
er making the equity safer. To understand
this more fully we need to better under-
stand the portfolio loss distribution.
The portfolio loss distribution
No matter what approach we use to gener-
ate it, the loss distribution of the reference
portfolio is crucial for understanding the risk
and value of correlation products. The port-
folio loss is clearly not symmetrically dis-
tributed: it is therefore informative to look at
the entire loss distribution, rather than sum-
marising it in terms of expected value and
standard deviation. We can use models of
the type discussed on page 33 to calculate
the portfolio loss distribution. We can expect
to observe one of the two shapes shown in
Figure 10. They are (i) a skewed bell curve; (ii)
a monotonically decreasing curve.
The skewed bell curve applies to the case
when the correlation is at or close to zero. In
this limit the distribution is binomial and the
peak is at a loss only slightly less than the
expected loss.
As correlation increases, the peak of the
distribution falls and the high quantiles
increase: the curves become monotonically
decreasing. We see that the probability of
larger losses increases and, at the same
time, the probability of smaller losses also
increases, thereby preserving the expected
loss which is correlation independent (for
further discussion see Mashal, Naldi and
Pedersen (2003)).
For very high levels of asset correlations
(hardly ever observed in practice), the distri-
bution becomes U-shaped. At maximum
default correlation all the probability mass is
located at the two ends of the distribution.
The portfolio either all survives or it all
defaults. It resembles the loss distribution
of a single asset.
0
5
10
15
20
25
30
35
40
0 5
1
0
1
5
4
7
Loss (%)
P
r
o
b
a
b
i
l
i
t
y

(
%
)
ρ = 0
ρ = 20%
ρ = 95%

Figure 10. Portfolio loss distribution
for a large portfolio at 0%, 20% and
95% correlation
guide.qxd 10/10/2003 11:15 Page 15
16 The Lehman Brothers Guide to Exotic Credit Derivatives
How then does the shape of the portfolio
loss distribution affect the pricing of tranch-
es? To see this we must study the tranche
loss distribution.
The tranche loss distribution
We have plotted in Figures 11–13 the loss dis-
tributions for a CDO with a 5% equity, 10%
mezzanine and 85% senior tranche for corre-
lation values of 20% and 50%. At 20% corre-
lation, we see that most of the portfolio loss
distribution is inside the equity tranche, with
about 14% beyond, as represented by the
peak at 100% loss. As correlation goes to
50% the probability of small losses increases
while the probability of 100% losses increas-
es only marginally. Clearly equity investors
benefit from increasing correlation.
The mezzanine tranche becomes more
risky at 50% correlation. As we see in Figure
12, the 100% loss probability jumps from
0.50% to 3.5%. In most cases mezzanine
investors benefit from falling correlation –
they are short correlation. However, the cor-
relation directionality of a mezzanine tranche
depends upon the collateral and the tranche.
In certain cases a mezzanine tranche with a
very low attachment point may be a long
correlation position.
Senior investors also see the risk of their
tranche increase with correlation as more
joint defaults push out the loss tail. This is
clear in Figure 13. Senior investors are short
correlation.
In Figure 14 we plot the dependence of the
value of different CDO tranches on correla-
tion. As expected, we clearly see that:
■ Senior investors are short correlation. If
correlation increases, senior tranches
fall in value.
■ Mezzanine investors are typically short
correlation, although this very much
depends upon the details of the tranche
and the collateral.
■ Equity investors are long correlation.
When correlations go up, equity tranches
go up in value.
In the process of rating CDO tranches, rat-
ing agencies need to consider all of these
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0
1
0
2
0
3
0
4
0
5
0
6
0
7
0
8
0
9
0
1
0
0
Mezzanine tranche loss (%)
P
r
o
b
a
b
i
l
i
t
y

(
%
)
ρ = 20% ρ = 50%
Figure 12. Mezzanine tranche loss
distribution for correlation of 20% and
50%. We have eliminated the zero loss
peak, which is about 86% in both cases
0
5
10
15
20
25
30
0
1
0
2
0
3
0
4
0
5
0
6
0
7
0
8
0
9
0
1
0
0
Equity tranche loss (%)
P
r
o
b
a
b
i
l
i
t
y

(
%
)
ρ = 20% ρ = 50%
Figure 11. Equity tranche loss
distribution for correlations of 20%
and 50%
guide.qxd 10/10/2003 11:15 Page 16
The Lehman Brothers Guide to Exotic Credit Derivatives 17
risk parameters and so have adopted model
based approaches. These are discussed on
page 43.
Customised synthetic CDO tranches
Customisation of synthetic tranches has
become possible with the fusion of deriva-
tives technology and credit derivatives.
Unlike full capital structure synthetics, which
issue the equity, mezzanine and senior parts
of the capital structure, customised synthet-
ics may issue only one tranche. There are a
number of other names for customised CDO
tranches, including bespoke tranches, and
single tranche CDOs.
The advantage of customised tranches is
that they can be designed to match exactly
the risk appetite and credit expertise of the
investor. The investor can choose the credits
in the collateral, the trade maturity, the
attachment point, the tranche width, the rat-
ing, the rating agency and the format (fund-
ed or unfunded). Execution of the trade can
take days rather than the months that full
capital structure CDOs require.
The basic paradigm has already been dis-
cussed in the context of default baskets. It
is to use CDS to dynamically delta-hedge
the first order risks of a synthetic tranche
and to use a trading book approach to
hedge the higher order risks. This is shown
in Figure 15 (overleaf).
For example, consider an investor who
buys a customised mezzanine tranche from
Lehman Brothers. We will then hedge it by
selling protection in an amount equal to the
delta of each credit in the portfolio via the
CDS market. The delta is the amount of
protection to be sold in order to immunise
the portfolio against small changes in
the CDS spread curve for that credit.
Each credit in the portfolio will have its
own delta.
Understanding delta for CDOs
For a specific credit in a CDO portfolio, the
delta is defined as the notional of CDS for
that credit which has the same mark-to-
market change as the tranche for a small
movement in the credit’s CDS spread
curve. Although the definition may be
straightforward, the behaviour of the delta
is less so.
One way to start thinking about delta is to
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0
1
0
2
0
3
0
4
0
Senior tranche loss (%)
P
r
o
b
a
b
i
l
i
t
y

(
%
)

ρ = 20%
ρ = 50%
Figure 13. Senior tranche loss
distribution for correlations of 20% and
50%. We have eliminated the zero loss
peak, which is greater than 96% in
both cases
–30
–20
–10
0
10
20
30
40
0
1
0
2
0
3
0
4
0
5
0
6
0
7
0
8
0
9
0
1
0
0
Correlation (%)
T
r
a
n
c
h
e

M
T
M

(
€ €
m
)
Equity
Mezzanine
Senior

Figure 14. Correlation dependence
of CDO tranches
guide.qxd 10/10/2003 11:15 Page 17
18 The Lehman Brothers Guide to Exotic Credit Derivatives
imagine a queue of all of the credits sorted
in the order in which they should default.
This ordering will depend mostly on the
spread of the asset relative to the other
credits in the portfolio and its correlation rel-
ative to the other assets in the portfolio. If
the asset whose delta you are calculating is
at the front of this queue, it will be most like-
ly to cause losses to the equity tranche and
so will have a high delta for the equity
tranche. If it is at the back of the queue then
its equity delta will be low. As it is most like-
ly to default after all the other asset, it will be
most likely to hit the senior tranche. As a
result the senior tranche delta will rise. This
framework helps us understand the direc-
tionality of delta.
The actual magnitude of delta is more dif-
ficult to quantify because it depends on the
tranche notional and the contractual
tranche spread, as well as the features of
the asset whose delta we are examining.
For example the delta for a senior tranche
to a credit whose CDS spread has widened
will fall due to the fact that it is more likely
to default early and hit the equity tranche,
and also because the CDS will have a high-
er spread sensitivity and so require a small-
er notional.
To show this we take an example CDO
with 100 credits, each $10m notional. It has
three tranches: a 5% equity, a 10% mezza-
nine and an 85% senior tranche. The asset
spreads are all 150bp and the correlation
between all the assets is the same.
The sensitivity of the delta to changing the
spread of the asset whose delta we are cal-
culating is shown in Figure 16. If the single
asset spread is less than the portfolio aver-
age of 150bp, then it is the least risky asset.
As a result, it would be expected to be the
last to default and so most likely to impact
the senior-most tranche. As the spread of
the asset increases above 150bp, it
becomes more likely to default before the
others and so impacts the equity or mezza-
nine tranche. The senior delta drops and the
equity delta increases.
In Figure 17 we plot the delta of the asset
versus its correlation with all of the other


Reference pool


100 investment grade
names in CDS format


$10m x 100 assets = $1bn



Bespoke tranche















Lehman
Brothers
Spread
Contingent
payment

∆ of CDS on Name 1

∆ of CDS on Name 2

∆ of CDS on Name 3

∆ of CDS on Name 100

∆ of CDS on Name 99

Investor

Figure 15. Delta hedging a synthetic CDO
guide.qxd 10/10/2003 11:15 Page 18
The Lehman Brothers Guide to Exotic Credit Derivatives 19
assets in the portfolio. These all have a cor-
relation of 20% with each other. If the asset
is highly correlated with the other assets it is
more likely to default or survive with the
other assets. As a result, it is more likely to
default en masse, and so senior and mezza-
nine tranches are more exposed. For low
correlations, if it defaults it will tend to do so
by itself while the rest of the portfolio tends
to default together. As a result, the equity
tranche is most exposed.
There is also a time effect. Through time,
senior and mezzanine tranches become
safer relative to equity tranches since less
time remains during which the subordina-
tion can be reduced resulting in principal
losses. This causes the equity tranche delta
to rise through time while the mezzanine
and senior tranche deltas fall to zero.
Building intuition about the delta is not triv-
ial. There are many further dependencies to
be explored and we intend to describe these
in a forthcoming paper.
Higher order risks
If properly hedged, the dealer should be
insensitive to small spread movements.
However, this is not a completely risk-free
position for the dealer since there are a num-
ber of other risk dimensions that have not
been immunised. These include correlation
sensitivity, recovery rate sensitivity, time
decay and spread gamma. There is also a
risk to a sudden default which we call the
value-on-default risk (VOD).
For this reason, dealers are motivated to
do trades that reduce these higher order
risks. The goal is to flatten the risk of the
correlation book with respect to these high-
er order risks either by doing the offsetting
trade or by placing different parts of the
capital structure with other buyers of cus-
tomised tranches.
Idiosyncratic versus systemic risk
In terms of how they are exposed to credit,
there is a fundamental difference between
equity and senior tranches. Equity tranches
are more exposed to idiosyncratic risk – they
incur a loss as soon as one asset defaults.
The portfolio effect of the CDO is only
expressed through the fact that it may take
several defaults to completely reduce the
equity notional. This implies that equity
investors should focus less on the overall
properties of the collateral, and more on try-
ing to choose assets which they believe will
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
0 10 20 30 40 50
Correlation with rest of portfolio (%)
Equity Mezzanine Senior

T
r
a
n
c
h
e

d
e
l
t
a
Figure 17. Dependency of tranche
delta on the asset’s correlation with the
rest of the portfolio
0
1
2
3
4
5
6
7
8
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
7
0
0
8
0
0
9
0
0
1
,
0
0
0
Asset spread (bp)
T
r
a
n
c
h
e

d
e
l
t
a Equity
Mezzanine
Senior

Figure 16. Dependency of tranche
delta on the spread of the asset
guide.qxd 10/10/2003 11:15 Page 19
20 The Lehman Brothers Guide to Exotic Credit Derivatives
not default. As a result we would expect
equity tranche buyers to be skilled credit
investors, able to pick the right credits for
the portfolio, or at least be able to hedge the
credits they do not like.
On the other hand, the senior investor has a
significant cushion of subordination to insu-
late them from principal losses until maybe
20 or more of the assets in the collateral have
defaulted. As a consequence, the senior
investor is truly taking a portfolio view and so
should be more concerned about the average
properties of the collateral than the quality of
any specific asset. The senior tranche is real-
ly a deleveraged macro credit trade.
Evolution of structures
Initially full capital structure synthetic CDOs
had almost none of the structural features
typically found in other securitised asset
classes and cash flow CDOs. It was only in
1999 that features that diverted cash flows
from equity to debt holders in case of cer-
tain covenant failures began entering the
landscape. The intention was to provide
some defensive mechanism for mezzanine
holders fearing that the credit cycle would
affect tranche performance. Broadly, these
features fit into two categories – ones that
build extra subordination using excess
spread, and others that use excess spread
to provide upside participation to mezza-
nine debt holders.
The most common example of structural
ways to build additional subordination is the
reserve account funding feature. Excess
spread (the difference between premium
received from the CDS portfolio and the
tranche liabilities) is paid into a reserve
account. This may continue throughout the
life of the deal or until the balance reaches a
predetermined amount. If structured to
accumulate to maturity, the equity tranche
will usually receive a fixed coupon through-
out the life of the transaction and any upside
or remainder in the reserve account at matu-
rity. If structured to build to a predetermined
level, the equity tranche will usually receive
excess interest only after the reserve
account is fully funded. More details are
provided in Ganapati and Ha (2002).
Other structures incorporated features to
share some of the excess spread with
mezzanine holders or to provide a step
up coupon to mezzanines if losses exceed-
ed a certain level or if the tranche was
downgraded. Finally, over-collateralisation
trigger concepts were adopted from cash
flow CDOs.
Principal protected structures
Investors who prefer to hold highly rated
assets can do so by purchasing CDO tranch-
es within a principal protected structure.
This is designed to guarantee to return the
investor’s initial investment of par. One par-
ticular variation on this theme is the Lehman
Brothers High Interest Principal Protection
with Extendible Redemption (HIPER). This is
typically a 10-year note which pays a fixed
coupon to the investor linked to the risk of a
CDO equity tranche.
This risk is embedded within the coupons
of the note such that each default causes a
reduction in the coupon size. However the
investor is only exposed to this credit risk for
a first period, typically five years, and the
coupon paid for the remaining period is
frozen at the end of year five. The coupon is
typically of the form:
In Figure 18 we show the cash flows
Coupon
Portfolio loss
Tranche size
× −

¸

1
]
1
8 1 0 % max ,
guide.qxd 10/10/2003 11:15 Page 20
The Lehman Brothers Guide to Exotic Credit Derivatives 21
assuming two credit events over the lifetime
of the trade. The realised return is depen-
dent on the timing of credit events. For a
given number of defaults over the trade
maturity, the later they occur, the higher the
final return.
Managed synthetics
The standard synthetic has been based on a
static CDO, ie, the reference assets in the
portfolio do not change. However, recently,
Lehman Brothers and a number of other
dealers have managed to combine the cus-
tomised tranche with the ability for an asset
manager or the issuer of the tranche to
manage the portfolio of reference entities.
This enables investors to enjoy all the bene-
fits of customised tranches and the benefits
of a skilled asset manager. The customis-
able characteristics include rating, rating
agency, spread, subordination, issuance for-
mat plus others.
The problem with this type of structure is
that the originator of the tranche has to fac-
tor into the spread the cost of substituting
assets in the collateral. Initially this was
based on the asset manager being told the
cost of substituting an asset using some
black-box approach.
More recently the format has evolved to
one where the manager can change the
portfolio subject to some constraints. One
example of such technology is Lehman
Brothers’ DYNAMO structure. The advan-
tage of this approach is that it frees the man-
ager to focus on the credits without having
to worry about the cost of substitution.
The other advantages of such a structure
for the asset manager are fees earned and
an increase in assets under management.
For investors the incentive is to leverage the
management capabilities of a credit asset
manager in order to avoid blow-ups in the
portfolio and so better manage downturns in
the credit cycle.
The CDO of CDOs
A recent extension of the CDO paradigm has
been the CDO of CDOs, also known as ‘CDO
squared’. Typically this is a mezzanine
‘super’ tranche CDO in which the collateral
is made up of a mixture of asset-backed
securities and several ‘sub’ tranches of syn-
thetic CDOs. Principal losses are incurred if
the sum of the principal losses on the under-
lying portfolio of synthetic tranches exceeds
the attachment point of the super-tranche.
Looking forward, we see growing interest in
synthetic-only portfolios.
Leveraging the spread premium
Market spreads paid on securities bearing
credit risk are typically larger than the
levels implied by the historical default rates
for the same rating. This difference, which
we call the spread premium, arises
because investors demand compensation
for being exposed to default uncertainty, as
well as other sources of risk, such as
spread movements, lack of liquidity or rat-
ings downgrades.
Portfolio credit derivatives, such as basket
default swaps and synthetic CDO tranches,
offer a way for investors to take advantage
of this spread premium. When an investor
Credit events
Credit window Coupons not reduced
by defaults after maturity
of credit window
100
100

guaranteed
Figure 18. The HIPER structure
guide.qxd 10/10/2003 11:15 Page 21
22 The Lehman Brothers Guide to Exotic Credit Derivatives
sells protection via a default basket or a
CDO tranche, the note issuer passes this on
by selling protection in the CDS market. This
hedging activity makes it possible to pass
this spread premium to the buyer of the
structured credit asset. For buy and hold
credit investors the spread premium paid
can be significant and it is possible to show,
see O’Kane and Schloegl (2003) for details of
the method, that under certain criteria, these
assets may be superior to single-name
credit investments.
Our results show that an FTD basket lever-
ages the spread premium such that the size
of the spread premium is much higher for
an FTD basket than it is for a single-credit
asset paying a comparable spread. This is
shown in Figure 19 where we see that an
FTD basket paying a spread of 350bp has
around 290bp of spread premium. Compare
this with a single-credit Ba3 asset also pay-
ing a spread close to 340bp. This has only
70bp of spread premium.
For an STD basket we find that the spread
premium is not leveraged. Instead, it is the
ratio of spread premium to the whole
spread which goes up. There are therefore
two conclusions:
1. FTD baskets leverage spread premium.
This makes them suitable for buy and
hold yield-hungry investors who wish to
be paid a high spread but also wish to
minimise their default risk.
2. STD baskets leverage the ratio of spread
premium to the market spread. This is
suitable for more risk-averse investors
who wish to maximise return per unit of
default risk.
We therefore see that default baskets can
appeal to a range of investor risk preferences.
CDO tranches exhibit a similar leveraging
of the premium embedded in CDS spreads.
The advantage of CDO of CDOs is that they
provide an additional layer of leverage to
the traditional CDO. This can make leverag-
ing the spread premium arguments even
more compelling.
The conclusion is that buy-and-hold corre-
lation investors are overcompensated for
their default risk compared with single-
name investors.
CDO strategies
Investors in correlation products should pri-
marily view them as buy and hold invest-
ments which allow them to enjoy the spread
premium. This is a very straightforward
strategy for mezzanine and senior investors.
However, for equity investors, there are a
number of strategies that can be employed
in order to dynamically manage the idiosyn-
cratic risk. We list some strategies below.
1. The investor buys CDO equity and
hedges the full notional of the 10 or so
worst names. The investor enjoys a sig-
nificant positive carry and at the same
time reduces his idiosyncratic default
risk. The investor may also sell CDS pro-
tection on the tightest names, using the
0
50
100
150
200
250
300
350
400
Ba3 FTD
Instrument
S
p
r
e
a
d

(
b
p
)
Actuarial spread Spread premium

Figure 19. Spread premium for an FTD
compared with a Ba3 single-name asset
guide.qxd 10/10/2003 11:15 Page 22
The Lehman Brothers Guide to Exotic Credit Derivatives 23
income to offset some of the cost of pro-
tection on the widest names.
2. The investor may buy CDO equity and
delta hedge. The net positive gamma
makes this trade perform well in high
spread volatility scenarios. By dynamical-
ly re-hedging, the investor can lock in this
convexity. The low liquidity of CDOs
means that this hedging must continue
to maturity.
3. The investor may use the carry from CDO
equity to over-hedge the whole portfolio,
creating a cheap macro short position.
While this is a negative carry trade, it can
be very profitable if the market widens
dramatically or if a large number of
defaults occur.
For more details see Isla (2003).
Credit options
Activity in credit options has grown sub-
stantially in 2003. From a sporadic market
driven mostly by one-off repackaging deals,
it has extended to an increasingly vibrant
market in both bond and spread options,
options on CDS and more recently options
on portfolios and even on CDO tranches.
This growth of the credit options market
has been boosted by declines in both
spread levels and spread volatility. The
reduction in perceived default risk has
made hedge funds, asset managers, insur-
ers and proprietary dealer trading desks
more comfortable with the spread volatility
risks of trading options and more willing to
exploit their advantages in terms of lever-
age and asymmetric payoff.
The more recent growth in the market for
options on CDS has also been driven by the
increased liquidity of the CDS market,
enabling investors to go long or short the
option delta amount.
Hedge funds have been the main growth
user of credit options, using them for credit
arbitrage and also for debt-equity strategies.
They are typically buyers of volatility, hedg-
ing in the CDS market and exploiting the
positive convexity. Asset managers seeking
to maximise risk-adjusted returns are
involved in yield-enhancing strategies such
as covered call writing. Bank loan portfolio
managers are beginning to explore default
swaptions as a cheaper alternative to buying
outright credit protection via CDS.
One source of credit optionality is the cash
market. Measured by market value weight,
5.6% of Lehman Brothers US Credit Index
and 54.7% of Lehman Brothers US High
Yield Index have embedded call or put
options. Hence, two strategies which have
been, and continue to be important in the
bond options market are the repack trade
and put bond stripping.
The repack trade
The first active market in credit bond
options was developed in the form of the
repack trade, spearheaded by Lehman
Brothers and several other dealers. Figure
20 (overleaf) shows the schematic of one
such transaction.
In a typical repack trade, Lehman Brothers
purchased $32,875,000 of the Motorola
(MOT) 6.5% 2028 debentures and placed
them into a Lehman Trust called CBTC. The
Trust then issued $25 par class A-1
Certificates to retail investors with a coupon
set at 8.375% – the prevailing rate for MOT in
the retail market at the time. Since the
8.375% coupon on the CBTC trust is higher
than the coupon on the MOT Bond, the CBTC
trust must be over-collateralised with enough
face value of MOT bonds to pay the 8.375%
coupon. An A-2 Principal Only (PO) tranche
captures the excess principal. Both class A-1
guide.qxd 10/10/2003 11:15 Page 23
24 The Lehman Brothers Guide to Exotic Credit Derivatives
and class A-2 certificates are issued with an
embedded call.
This call option was sold separately to
investors in a form of a long-term warrant.
The holder of the MOT call warrant has the
right but not the obligation to purchase the
MOT bonds from the CBTC trust beginning
on 19/7/07 and thereafter at the preset call
strike schedule. This strike is determined by
the proceeds needed to pay off the A-1 cer-
tificate at par plus the A-2 certificate at the
accreted value of the PO. Because retail
investors are willing to pay a premium for
the par-valued low-notional bonds of well-
known high quality issuers, the buyers of the
call warrant can use this structure to source
volatility at attractive levels.
Put bond stripping
According to Lehman Brothers’ US Credit
Index, bonds with embedded puts consti-
tute approximately 2.3% of the US credit
bond market, by market value. These bonds
grant the holder the right, but not the obli-
gation to sell the bond back to the issuer at
a predetermined price (usually par) at one or
more future dates.
This option can be viewed as an extension
option since by failing to exercise it, the bond
maturity is extended. In the past several
years, the market has priced these bonds as
though they matured on the first put date and
has not given much value to the extension
option. Recently, credit investors have
realised a way to extract this extension risk
premium via a put bond stripping strategy.
Essentially, an investor can buy the put
bond, and sell the call option to the first put
date at a strike price of par. Thus, the investor
has a long position in the bond coupled with a
synthetic short forward (long put plus short
call) with a maturity coinciding with the first
put date. He then hedges this position by
asset swapping the bond to the put date,
effectively eliminating all of the interest rate
risk and locking in the cheap volatility. Given
the small amount of outstanding put bonds,
this strategy has led to more efficient pricing
of the optionality in these securities.
Bond options
There are a variety of bond options traded in
the market. The two most important ones
for investors are:

CBTC
Series
2002 - 14

$25.515mm

A - 1 retail
tranche

6.50% + par










$32.875mm
MOT 6.50%
11/15/28

$7.36mm

A - 2 PO
tranche


MOT call
warrant
8.375% (25.515mm)
Residual principal
Option to purchase
MOT bond
Par
PV of CF

Market price

Premium
Figure 20. Mechanics of MOT 6.5% 15/11/28 repack transaction
guide.qxd 10/10/2003 11:15 Page 24
The Lehman Brothers Guide to Exotic Credit Derivatives 25
Price-based options: at exercise, the option
holder pays a fixed amount (strike price) and
receives the underlying bond – the payoff is
proportional to the difference between the
price of the bond and the strike price.
Examples of actively trading price options
are Brady bond options, corporate bond
options, CBTC call warrants and calls on put
bonds. See the illustration in Figure 21.
Spread-based options: at exercise, the
option holder pays an amount equal to the
value of the underlying bond calculated
using the strike spread and receives the
underlying bond – the payoff is proportion-
al to difference between the underlying
spread and the strike spread (to a first order
of approximation). Spread options can be
structured using spreads to benchmark
Treasury bonds, default swap spreads or
asset swap spreads.
The exercise schedule can be a single date
(European), multiple pre-specified dates
(Bermudan) or any date in a given range
(American). Currently, the most active trad-
ing occurs in short-term (less than 12
months to expiry) European-style price
options on bonds.
Two of the most common strategies using
price options on bonds are covered calls and
naked puts. They can be considered respec-
tively as limit orders to sell or buy the under-
lying bond at a predetermined price (the
option strike) on a predetermined day (the
option expiry date).
Covered call strategy: an investor who owns
the underlying bond sells an out-of-money
call on the same face value, receiving an
upfront premium. If the bond price on the
expiry date is greater than the strike, the
investor delivers the bonds and receives the
strike price. The option premium offsets the
investor’s loss of upside on the price. If the
price is less than the strike the investor keeps
the bonds and the premium.
Naked put strategy: an investor writes an
out-of-the-money put on a bond which he
does not own but would like to buy at a
lower price. If the bond price on the expiry
date is lower than the strike price, it is deliv-
ered to the investor. The option premium
compensates him for not being able to buy
the bond more cheaply in the market. If the
bond price is above the option strike price,
the investor earns the premium.
In both of these strategies, the main objec-
tive for the investor is to find a strike price at
which he is willing to buy or sell the under-
lying bond and which provides sufficient
premium to compensate for the potential
upside that he forgoes.
Default swaptions and callable CDS
An exciting development in the credit deriva-
tives markets in the past 12 months has
been the emergence of default swaptions.
These are options on credit default swaps.
The emerging terminology from this mar-



Option
buyer




Lehman
Brothers
1.13% premium
Right to buy F
7.25% 11 at 100.76

Figure 21. Three month price call
option on F 7.25% 25/10/11, struck at
100.76% price.
guide.qxd 10/10/2003 11:15 Page 25
26 The Lehman Brothers Guide to Exotic Credit Derivatives
ket is that protection calls (option to buy pro-
tection) are called payer default swaptions.
Protection puts (option to sell protection) are
called receiver default swaptions.
Unlike price options on bonds, the exercise
decision for default swaptions is based on
credit spread alone. As a result, they are
essentially a ‘pure’ credit product, with pricing
being mostly driven by CDS spread volatility.
Default swaptions give investors the oppor-
tunity to express views on the future level and
variability of default swap spreads for a given
issuer. They can be traded outright or embed-
ded in callable CDS. The typical maturity of
the underlying CDS is five years but can range
from one–10 years, and the time to option
expiry is typically three months to one year.
Payer default swaption
The option buyer pays a premium to the
option seller for the right but not the obliga-
tion to buy CDS protection on a reference
entity at a predetermined spread on a future
date. Payer default swaptions can be struc-
tured with or without a provision for knock
out at no cost if there is a credit event
between trade date and expiry date. If the
knock out provision is included in the swap-
tion, the option buyer who wishes to main-
tain protection over the entire maturity range
can separately buy protection on the under-
lying name until expiry of the swaption.
The relevant scenarios for this investment
are complementary to the ones in the case
of the protection put. If spreads tighten by
the expiry date, the option buyer will not
exercise the right to buy protection at the
strike and the option seller will keep the
option premium.
Receiver default swaption
In a receiver default swaption, the option
buyer pays a premium to the option seller
for the right, but not the obligation, to sell
CDS protection on a reference entity at a
predetermined spread on a future date. This
spread is the option strike.
We do not need to consider what happens
if the reference entity experiences a credit
event between trade date and expiry date as
they would never exercise the option in this
case. As a result, there is no need for a
knockout feature for receiver default swap-
tions. Consider the following example.
Lehman Brothers pays 1.20% for an at-
the-money receiver default swaption on
five-year GMAC, struck at the current five
year spread of 265bp and with three
months to expiry. The investor is short the
option. From the investor’s perspective, the
relevant scenarios are:
■ If five-year GMAC trades above 265bp in
three months, Lehman does not exercise,
as they can sell protection for a higher
spread in the market. The investor has
realised an option premium of 1.20% in a
quarter of a year.
■ If five-year GMAC trades at 238bp in three
months, the trade breaks even. (1.20% up
front option premium equals the payoff of
(265bp–238bp)=27bp times the five-year
PV01 of 4.39). If five-year GMAC trades
below 238bp in three months, the loss on
the exercise of the option will be greater
than the upfront premium and the investor
will underperform on this trade.
Hedging default swaptions
Dealers hedge these default swaptions using
a model of the type discussed on page 49.
The underlying in a default swaption is the for-
ward CDS spread from the option expiry date
to the maturity date of the CDS. Theoretically,
a knock-out payer swaption should be delta
guide.qxd 10/10/2003 11:15 Page 26
The Lehman Brothers Guide to Exotic Credit Derivatives 27
hedged with a short protection CDS to the
final maturity of the underlying CDS and a long
protection CDS to the default swaption expiry
date. This combination will produce a synthet-
ic forward CDS that knocks out at default
before the forward date. In practice, swap-
tions with expiry of 1 year and less are hedged
only with CDS to final maturity due to a lack of
liquidity in CDS with short maturities. We have
summarised the key features of these differ-
ent swaption types in Figure 22.
Callable default swaps
In addition to default swaptions, there is a
growing interest in callable default swaps.
These are a combination of plain vanilla CDS
with an embedded short receiver swaption
position. The seller of a callable default
swap is long credit exposure but this expo-
sure can be terminated by the option buyer
at some strike spread on a future date.
Consider an example.
Lehman Brothers buys five-year GMAC
protection, callable in one year, for 315bp
from an investor. The assumed current mid-
market spread for five-year GMAC protec-
tion is 265bp.
If the four-year GMAC spread in one year
is less than the strike spread of 315bp, then
Lehman Brothers will exercise the option
and so cancel the protection, enabling us to
buy protection at the lower market spread.
The investor therefore has earned 315bp for
selling five-year protection on GMAC for
one year.
If the four-year GMAC spread in one year
is greater than 315bp, the contract contin-
ues and the investor continues to earn
315bp annually.
From the perspective of the option seller,
the callable default swap has a limited MTM
upside compared with plain vanilla CDS. The
additional spread of 315bp–260bp=55bp in
this example compensates the option seller
for the lost potential upside.
Selling protection in callable default swap
is equivalent to a covered call strategy on
underlying issuer spreads and is particularly
suitable as a yield-enhancement technique
for asset managers and insurers.
Credit portfolio options
Starting in mid-2003 market participants
have been able to trade in portfolio options
whose underlying asset is the TRAC-X North
America portfolio with 100 credits. Liquidity
is also growing in the European version.
The rationale for options based on TRAC-X
is that the portfolio effect will reduce the
option volatility and make it easier for deal-
ers to hedge. From an investor perspective it
presents a way to take a macro view on
spread volatility.
We are now seeing investors trading both
at-the-money and out-of-money puts and
calls to maturities extending from three to
nine months. The contracts are typically
traded with physical delivery. If the TRAC-X
portfolio spread is wider than the strike
level on the expiry date, the holder of the
Product Payer default Receiver default
swaption swaption
Description Option to buy Option to sell
protection protection
Exercised if CDS spread at CDS spread at
expiry > strike expiry < strike
Credit view Short credit Long credit
forward forward
Knockout May trade with Not relevant
or without
Figure 22. Default swaption types
guide.qxd 10/10/2003 11:15 Page 27
28 The Lehman Brothers Guide to Exotic Credit Derivatives
payer default swaption will exercise the
option and lock in the portfolio protection
at more favourable levels. Conversely, if the
TRAC-X spread is tighter than the strike, the
holder of the receiver swaption will benefit
from exercising the option and realising the
MTM gain.
Investors can monetise a view on the future
range of market spreads by trading bearish
spread (buying at-the-money receiver swap-
tion and selling farther out-of-money receiver
swaption) or bullish spread (buying ATM payer
swaption and selling farther out-of-money
payer swaption) strategies. Other strategies
include expressing views on spread changes
over a given time horizon by trading calendar
spreads (buying near maturity options and
selling farther maturity options).
Finally, because the TRAC-X spread is less
subject to idiosyncratic spread spikes, and
because of the existing two-way markets
with varying strikes, investors can express
their views on the direction of changes in
the macro level of spread volatility by trading
straddles, ie, simultaneously buying payer
and receiver default swaptions as a way to
go long volatility while being neutral to the
direction of spread changes.
Hybrid products
Hybrid credit derivatives are those which
combine credit risk with other market risks
such as interest rate or currency risk.
Typically, these are credit event contingent
instruments linked to the value of a deriva-
tives payout, such as an interest rate swap
or an FX option.
There are various motivations for entering
into trades which have these hybrid risks.
Below, we give an overview of the economic
rationale for different types of structures. We
discuss the modelling of hybrid credit deriva-
tives in more detail on page 51.
Clean and perfect asset swaps
One important theme is the isolation of the
pure credit risk component in a given
instrument. For example, a European CDO
investor may wish to access USD collateral
without incurring any of the associated cur-
rency risks.
Cross-currency asset swaps are the tradi-
tional mechanism by which credit investors
transform foreign currency fixed-rate bonds
into local currency Libor floaters. This has
the benefit that it substantially reduces the
currency and interest rate risk, converting
the bond from an FX, interest rate and cred-
it play into an almost pure credit play.
However, the currency risk has not been
completely removed. First, note that a cross
currency asset swap is really two trades: (i)
purchase of a foreign currency asset; and (ii)
entry into a cross-currency swap. In the case
of a European investor purchasing a dollar
asset, the investor receives Euribor plus a
spread paid in euros.
As long as the underlying dollar asset
does not default during the life of the asset
swap there is no currency risk to the
investor. However, if the asset does
default, the investor loses the future dollar
coupons and principal of the asset, just
receiving some recovery amount which is
paid in dollars on the dollar face value. As
the cross-currency swap is not contingent,
meaning that the payments on the swap
contract are unaffected by any default of
the asset, the investor is therefore obliged
to either continue the swap or to unwind it
at the market value with a swap counter-
party. This unwind value can be positive or
negative – the investor can make a gain or
loss – depending on the direction of move-
ments in FX and interest rates since the
trade was initiated.
The risk is significant. We have modelled
guide.qxd 10/10/2003 11:15 Page 28
The Lehman Brothers Guide to Exotic Credit Derivatives 29
the distribution of the swap MTM at
default, shown in Figure 23, for a five year
euro-dollar cross-currency swap using mar-
ket calibrated parameters. The downside
risk is significant with possible swap relat-
ed losses comparable in size to the loss on
the defaulted credit.
As a result, the basic cross-currency
asset swap has a default contingent inter-
est rate and currency risk. This can be
removed through the use of a hybrid. Two
variations exist.
1. In the clean asset swap, Lehman Brothers
takes on the default contingent swap
unwind value risk.
2. In the perfect asset swap, Lehman
Brothers takes on the default contingent
swap unwind value risk and guarantees
(quantoes) the recovery rate of the default-
ed asset in the investor’s base currency.
Both structures have featured widely in the
CDO market where they have been used to
immunise mixed-currency high-yield bond
portfolios against currency risk in order to
allow the structure to qualify for the desired
rating from the rating agency.
However, they can also be used by
investors to convert foreign currency assets
into their base currency. For example,
European investors can use the perfect asset
swap to take advantage of the often higher
spread levels which exist for the same cred-
its when denominated in US dollars.
The cost of removing this default contin-
gent swap MTM risk and paying the recov-
ery rate in the investors domestic currency
depends upon a number of factors includ-
ing the volatility of the FX rate, the credit
quality of the reference credit, the shape of
the Libor curves in both currencies, interest
rate volatilities in both currencies and the
correlations between rates, FX and credit.
The cost can be amortised over the life of
the trade as a reduction in the asset swap
spread paid. It is interesting to note that the
reduction in the perfect asset swap spread
may not be significant given that the two FX
risks to the swap MTM and the recovery rate
are actually partially offsetting.
Counterparty risk hybrids
The mitigation of counterparty risk gives
rise to another type of hybrid. Consider an
investor who enters into an interest rate
swap with a credit risky counterparty.
Suppose also that this counterparty
defaults with a recovery rate of R. If the
MTM of the swap is negative from the
viewpoint of the investor, the swap is
unwound at market value. This means that
in an MTM framework the investor incurs
no loss from the default event; the swap
could be replaced by one with more advan-
tageous terms at zero cost. However, if the
value is positive, the investor loses a frac-
tion of this MTM.
A way to mitigate the counterparty risk is
therefore to buy protection on the counter-
0
1
2
3
4
5
6
–58 –28 2 32 63 93
P
r
o
b
a
b
i
l
i
t
y

(
%
)
Swap mark-to-market value at default (%)
Figure 23. Modelled distribution of the
swap MTM at default as a percentage of
face value for a five-year euro-dollar swap
guide.qxd 10/10/2003 11:15 Page 29
30 The Lehman Brothers Guide to Exotic Credit Derivatives
party, where the contingent payout is linked
to the replacement cost of the swap. As we
discuss on page 51, the investor is effec-
tively purchasing an interest rate swaption
which is automatically exercised upon a
default event. Clearly, similar structures can
be constructed to provide credit protection
for other payouts.
Yield enhancement
In the current interest rate environment with
very low short term rates and steep yield
curves, coupled with the more mature credit
derivatives market, there is increasing
investor interest in credit-linked notes with
more exotic coupons. These include paying a
spread over a Constant Maturity Swap (CMS)
rate or a particular inflation index.
Hedge cost mitigation
The final class of application concerns the
pure hedging of credit contingent FX or
interest rate risks, such as those faced by
an international corporation in the course of
its business activities. For example, one
way to reduce the cost of credit hedging
could be to purchase default protection
linked to an FX rate being above or below a
specified threshold.
Companies looking to hedge interest rate
or FX risk may be concerned with the cost of
outright hedging using vanilla derivatives. In
this case, a hedge which knocks out on the
default of a reference credit can provide an
adequate hedge while significantly decreas-
ing costs. Clearly, the hedger is implicitly
taking a bullish view on the reference credit.
guide.qxd 10/10/2003 11:15 Page 30
The Lehman Brothers Guide to Exotic Credit Derivatives 31
To be able to price and risk-manage credit
derivatives, we need a framework for valu-
ing credit risk at a single issuer level, and at
a multi-issuer level. The growth of the credit
derivatives market has created a need for
more powerful models and for a better
understanding of the empirical evidence
needed to calibrate these models. In this
section we will present a detailed overview
of modelling approaches from a practical
perspective, ie, we will discuss models,
implementation and calibration.
Single credit modelling
The world of credit modelling is divided into
two main approaches, called structural and
reduced-form. In the structural approach, the
default is characterised as the consequence
of some event such as a company’s asset
value being insufficient to cover a repayment
of debt. Such models are usually extensions
of Merton’s 1974 model that used a contin-
gent claims analysis for modelling default.
Structural models are generally used to say
at what spread corporate bonds should
trade based on the internal structure of the
company. They therefore require information
about the balance sheet of the company and
can be used to establish a link between pric-
ing in the equity and debt markets. However,
they are limited in a number of ways includ-
ing the fact that they generally lack the flex-
ibility to fit exactly a given term structure of
spreads; and they cannot be easily extended
to price complex credit derivatives.
In the reduced-form approach, the credit
process is modelled directly via the proba-
bility of the credit event itself. Reduced-form
models also generally have the flexibility to
refit the prices of a variety of credit instru-
ments of different maturities. They can also
be extended to price more exotic credit
derivatives. It is for these reasons that they
are used for credit derivatives pricing.
The hazard rate approach
The most widely used reduced-form
approach is based on the work of Jarrow
and Turnbull (1995), who characterise a cred-
it event as the first event of a Poisson count-
ing process which occurs at some time t
with a probability defined as
ie, the probability of a default occurring with-
in the time interval [t, t+dt) conditional on
surviving to time t, is proportional to some
time dependent function λ(t), known as the
hazard rate, and the length of the time inter-
val dt. Over a finite time period T, it is possi-
ble to show that the probability of surviving
is given by
The expectation is taken under the risk-neu-
tral measure. A common assumption is that
the hazard rate process is deterministic. By
extension, this assumption also implies that
the hazard rate is independent of interest
rates and recovery rates.
Pricing model for CDS
The breakeven spread in a CDS is the spread
at which the present values (PV) of premium
and protection legs are equal, ie
Premium PV = Protection PV.
1
1
]
1

¸

,
_

¸
¸


T
Q
ds s E T Q
0
) ( exp ) , 0 ( λ
dt t t dt t ) ( ] | Pr[ λ τ τ > + ≤
Credit Derivatives Modelling
guide.qxd 10/10/2003 11:15 Page 31
32 The Lehman Brothers Guide to Exotic Credit Derivatives
To determine the spread we therefore need
to be able to value the protection and pre-
mium legs. It is important to take into
account the timing of the credit event
because this can have a significant effect on
the present value of the protection leg and
also the amount of premium paid on the pre-
mium leg. Within the hazard rate approach
we can solve this timing problem by condi-
tioning on the probability of defaulting with-
in each small time interval [s, s+ds], given by
Q(0,s)λ(s)ds, then paying (1–R) and discount-
ing this back to today at the risk-free rate.
Assuming that the hazard rate and risk free
rate term structures are flat, we can write
the value for the protection leg as
The value of the premium leg is the PV of the
spread payments which are made to default
or maturity. If we assume that the spread S
on the premium leg is paid continuously, we
can write the present value of the premium
leg as
Equating the protection and premium legs
and solving for the breakeven spread gives
This relationship is known as the credit tri-
angle because it is a relationship between
three variables where knowledge of any two
is sufficient to calculate the third. It basical-
ly states that the spread paid per small time
interval exactly compensates the investor
for the risk of default per small time interval.
Within this model the interest rate depen-
dency drops out.
Given a CDS which has a flat spread curve
at 150bp, and assuming a 50% recovery
rate, the implied hazard rate is 0.015 divided
by 0.5, which implies a 3% hazard rate. The
implied one-year survival probability is there-
fore exp(–0.03)=97.04%. For two years it is
exp(–0.06)=94.18%, and so on.
Valuation of a CDS position
The value of a CDS position at time t follow-
ing initiation at time 0 is the difference
between the market implied value of the
protection and the cost of the premium pay-
ments, which have been set contractually at
S
C
. We therefore write
MTM(t) = ± (Protection PV –Premium PV),
where the sign is positive for a long protec-
tion position and negative for a short protec-
tion position. If the current market spread is
given by S(t) then the MTM can be written as
where the RPV01 is the risky PV01 which is
given by
and where
An investor buys $10m of five-year protec-
tion at 100bp. One year later, the credit
trades at 250bp. Assuming a recovery rate
S t
R 1

( )

λ .
RPV t
e
r
r T t
01
1
( )

( )
+ ( )
− + ( ) − ( ) λ
λ
,
− S t S t MTM )) 0 ( ) ( ( ) ( × RPV01
) 1 ( R S − λ

+

+ −
+ −
T T r
s r
r
e S
ds e S
0
) (
) (
) 1 (
λ
λ
λ

+
− −

+ −
+ −
T T r
s r
r
e R
ds e R
0
) (
) (
) 1 )( 1 (
) 1 (
λ
λ
λ
λ
λ
guide.qxd 10/10/2003 11:15 Page 32
The Lehman Brothers Guide to Exotic Credit Derivatives 33
of 40%, the value is given by substituting,
r=3.0%, R=40%, S(t)=0.025, S(0)=0.01 and
t=4 into the above equation to give
λ=4.17% and an MTM value of $521,661.
This is a simple yet fairly accurate model
which works quite well when the interest
rate and credit curves are flat. When this is
not the case, it becomes necessary to use
bootstrapping techniques to build a full term
structure of hazard rates. This may be
assumed to be piecewise flat or piecewise
linear. For a description of such a model see
O’Kane and Turnbull (2003).
Default probabilities
The default probabilities calculated for
pricing purposes can be quite different
from those calculated from historical
default rates of assets with the same rat-
ing. These real-world default probabilities
are generally much lower. The reason for
this is that the credit spread of an asset
contains not just a compensation for pure
default risk; it also depends on the mar-
ket’s risk aversion expressed through a risk
premium, as well as on supply-and-
demand imbalances.
One should also comment on the market’s
use of Libor as a risk-free rate in pricing.
Pricing theory shows that the price of a
derivative is the cost of replicating it in a risk-
free portfolio using other securities. Since
most market dealers are banks which fund
close to Libor, the cost of funding these
other securities is also close to Libor. As a
consequence it is the effective risk-free rate
for the derivatives market.
Calibrating recovery rates
The calibration of recovery rates presents a
number of complications for credit deriva-
tives. Strictly speaking, the recovery rate
used in the pricing of credit derivatives is the
expected recovery rate following a credit
event where the expectation is under the
risk-neutral measure.
Such expectations are only available from
price information, and the problem in credit
is that given one price, it is difficult to sepa-
rate the probability of default from the recov-
ery rate expectation.
The market standard is therefore to revert to
rating agency default studies for estimates of
recovery rates. These typically show the aver-
age recovery rate by seniority and type of
credit instrument, and usually focus on a US
corporate bond universe. Adjustments may
be made for non-US corporate credits and for
certain industrial sectors.
Problems with rating agency recovery
statistics include the fact that they are back-
ward looking and that they only include the
default and bankruptcy credit events –
restructuring is not included. In their favour,
one should say that, as they represent the
price of the defaulted asset as a fraction of
par some 30 days after the default event,
they are similar to the definition of the recov-
ery value in a CDS.
Recent work (Altman et al 2001) shows that
there is a significant negative correlation
between default and recovery rates. One
way to incorporate this effect is to assume
that recovery rates are stochastic. The stan-
dard approach is to use a beta distribution.
Modelling default correlation
By modelling correlation products, we mean
modelling products whose pricing depends
upon the joint behaviour of a set of credit
assets. These include default baskets and
synthetic CDOs. As a result of the growth in
usage of these products, this is an area of
pricing which has recently gained a lot of
attention, and in which we have seen a num-
ber of significant modelling developments.
guide.qxd 10/10/2003 11:15 Page 33
34 The Lehman Brothers Guide to Exotic Credit Derivatives
In this section we will describe some of
these current models, show how they can
be applied to the valuation of baskets and
CDOs, and towards the end discuss model
calibration issues.
Modelling joint defaults
The valuation of default-contingent instru-
ments calls for the modelling of default
mechanisms. As discussed earlier, a classical
dichotomy in credit models distinguishes
between a ‘structural approach’, where
default is triggered by the market value of the
borrower’s assets (in terms of debt plus equi-
ty) falling below its liabilities, and a ‘reduced-
form approach’, where the default event is
directly modelled as an unexpected arrival.
Although both the structural and the reduced-
form approaches can in principle be extended
to the multivariate case, structural models
calibrated to market-implied default probabili-
ties (often called ‘hybrid’ models) have gained
favour among practitioners because of their
tractability in high dimensions.
If we think of defaults as generated by
asset values falling below a given boundary,
then the probabilities of joint defaults over a
specified horizon must follow from the joint
dynamics of asset values: consistent with
their descriptive approach of the default
mechanism, multivariate structural models
rely on the dependence of asset returns in
order to generate dependent default events.
This is shown in Figure 24 where we have
simulated 1,000 pairs of asset returns mod-
elled as normally distributed random vari-
ables, for two firms i and j for two different
asset return correlations of 10% and 90%.
The vertical and horizontal lines represent
default thresholds for firms i and j respec-
tively. Clearly we see that the probability of
both i and j defaulting, represented by the
number of points in the bottom left quad-
rant defined by the default thresholds,
increases as the asset return correlation
increases. Therefore asset correlation
leads to default correlation.
Although the pay-offs of multi-credit
default-contingent instruments such as nth-
to-default baskets and synthetic loss tranch-
es cannot be statically replicated by trading
in a set of single-credit contracts, the cur-
rent market practice is to value correlation
products using standard no-arbitrage argu-
ments. It follows that the valuation of these
multi-credit exposures boils down to the
computation of (risk-neutral) expectations
over all possible default scenarios.
A number of different hybrid frameworks
have been proposed in the literature for mod-
elling correlated defaults and pricing multi-
name credit derivatives. Hull and White (2001)
generate dependent default times by diffus-
ing correlated asset values and calibrating
default thresholds to replicate a set of given
marginal default probabilities. Multi-period
extensions of the one-period CreditMetrics
paradigm are also commonly used, even if
they produce the undesirable serial indepen-
dence of the realised default rate.
1
While most
multi-credit models require simulation, the
-4
-3
-2
-1
0
1
2
3
4
5
-4 -2 0 2 4
-4
-3
-2
-1
0
1
2
3
4
5
-4 -2 0 2 4
C
A
C
B
C
A
C
B
V
A
V
B
V
B
V
A
Both firms default
10% Correlation 90% Correlation
Figure 24. Scatterplot of 1,000 simulat-
ed asset returns with 0% and 90% corre-
lation. Default thresholds are also shown
guide.qxd 10/10/2003 11:15 Page 34
The Lehman Brothers Guide to Exotic Credit Derivatives 35
need for accurate and fast computation of
greeks has pushed researchers to look for
modelling alternatives. Finger (1999) and,
more recently, Gregory and Laurent (2003)
show how to exploit a low-dimensional factor
structure and conditional independence to
obtain semi-analytical solutions.
Asset and default event correlation
Default event correlation (DEC) measures
the tendency of two credits to default joint-
ly within a specified horizon. Formally, it is
defined as the correlation between two
binary random variables that indicate
defaults, ie
where p
A
and p
B
are the marginal default
probabilities for credits A and B, and P
AB
is
the joint default probability. Of course, p
A
, p
B
and p
AB
all refer to a specific horizon. Notice
that default event correlation increases lin-
early with the joint probability of default and
is equal to zero if and only if the two default
events are independent. Its limits are not
–100% to +100% but are actually a function
of the marginal probabilities themselves.
Default event correlations are the funda-
mental drivers in the valuation of multi-name
credit derivatives. Unfortunately, the scarcity
of default data makes joint default probabili-
ties, and thus default event correlations,
very hard to estimate directly. As a result,
market participants rely on alternative meth-
ods to calibrate the frequency of joint
defaults within their models. Hybrid models,
such as the simulation approach and the
semi-analytical framework described below,
use the dependence among asset returns to
generate joint defaults, therefore avoiding
the need for a direct estimation of joint
default probabilities.
Default-time simulation
Monte Carlo models generally aim at the
generation of default paths, where each
path is simply a list of default times for each
of the credits in the reference portfolio
drawn at random from the joint default dis-
tribution. Knowing the time and identity of
each default event allows for a precise valu-
ation of any multi-credit product, no matter
how complex the contractual specification
of the payoff.
In an influential paper, Li (2000) presents a
simple and computationally inexpensive
algorithm for simulating correlated defaults.
His methodology builds on the implicit
assumption that the multivariate distribution
of default times and the multivariate distri-
bution of asset returns share the same
dependence structure, which he assumes to
be Gaussian and is therefore fully charac-
terised by a correlation matrix.
For valuation purposes, we need to sample
from the multivariate distribution of default
times under the risk-neutral probability mea-
sure. In this case, it is common practice to
back out the marginal distributions of default
times, which we will denote with F
1
, F
2
,…, F
d
,
from single-credit defaultable instruments
(such as CDS). We then join these marginal
distributions with a correlation matrix, which
according to the stated assumptions repre-
sents the correlation matrix of the asset
returns of the reference credits. Since asset
returns are not directly observable, it is com-
mon practice to proxy asset correlations
using equity correlations. Towards the end of
( ) ( )
B B A A
B A AB
p p p p
p p p
DEC
− −

1 1
1
Finger (2000) offers an excellent comparison of several
multivariate models in terms of the default distributions that
they generate over time when calibrated to the same
marginals and first-period joint default probabilities.
guide.qxd 10/10/2003 11:15 Page 35
36 The Lehman Brothers Guide to Exotic Credit Derivatives
this section we will discuss whether this
seems to be a reasonable approximation.
The high-dimensionality of multi-credit
instruments means that it is not possible to
use market prices to obtain the full depen-
dence structure. Instead, practitioners gen-
erally estimate the necessary correlations
using historical returns, implicitly relying on
the extra assumption that the correlations
among asset returns remain unchanged
when we move from the objective to the
pricing probability measure.
In the bivariate case, the joint distribution
function of default times and is simply
given by
where N
2,ρ
denotes the bivariate cumulative
standard Normal with correlation ρ, and N
–1
indicates the inverse of a cumulative stan-
dard Normal. The extension to the d-dimen-
sional case is immediate.
Simulating default times from this distribu-
tion is straightforward. With d reference
credits and correlation matrix Σ we have the
following algorithm:
1. Choleski decomposition of the correla-
tion matrix to simulate a multivariate
Normal random vector with cor-
relation Σ.
2. Transform the vector into the unit hyper-
cube using
3. Translate U into the corresponding
default times vector t using the inverse of
the marginal distributions:
The simulation algorithm is illustrated in
Figure 25. It is easy to verify that τ has the
given marginals and a Normal dependence
structure fully characterised by the correla-
tion matrix Σ.
Once we know how to sample from the
risk-neutral distribution of default times, it is
straightforward to price a correlation trade.
Generally speaking, the valuation of a given
correlation product always boils down to the
computation of an expectation of the form,
( ) ( ) ( ) ( ) ( )
d d d
u F u F u F
1
2
1
2 1
1
1 2 1
,..., , ,..., ,
− − −
τ τ τ τ
( ) ( ) ( ) ( )
d
x N x N x N U ,..., ,
2 1

d
R X ∈
(y))), (F (x)),N (F (N N y) x, P(
2
1
1
1
, 2 2 1
− −
< <
ρ
τ τ


0
0.2

0.4

0.6

0.8

1.0
Cumulative normal

0 2 4 6 8 10
0
0.2

0.4

0.6

0.8


Time (years)

Cumulative default probability
4 6
x
2 0 2 4 6
1.0
Figure 25. Mapping a normal random variable to a default time
guide.qxd 10/10/2003 11:15 Page 36
The Lehman Brothers Guide to Exotic Credit Derivatives 37
E[f (τ)] where τ = (τ
1

2
,...,τ
d
) represents the
vector of default times and f is a function
describing the discounted cash flows (both
positive and negative) associated with the
instrument under consideration (see, eg,
Mashal and Naldi (2002b)).
In summary, Monte Carlo simulation
allows for accurate valuations and risk mea-
surements of multi-credit payoffs, even
when complex path-dependencies such as
reserve accounts, interest coverage or col-
lateralisation tests are involved. This is
because both the exact default times and
the identity of the defaulters are known on
every simulated path. Moreover, we can
easily expand the set of variables to be
treated as random. For example, Frye (Risk,
2003) argues that modelling stochastic
recoveries and allowing for negative
correlation between recovery and default
rates is essential for a proper valuation of
credit derivatives.
The precision and flexibility of this
approach, however, come at the cost of
computational speed. The basic problem of
using simulation is that defaults are rare
events, and a large number of simulation
paths are usually required to achieve a suf-
ficient sampling of the probability space.
There are ways to improve the situation.
Useful techniques include antithetic sam-
pling, importance sampling and the use of
low-discrepancy sequences. This problem
becomes particularly significant when we
turn our attention to the calculation of sen-
sitivity measures, since for a reasonable
number of paths the simulation noise can
be similar to or greater than the price
change due to the perturbation of the input
parameter. Once again, techniques exist to
alleviate the problem, but it is hard to
achieve precise hedge ratios in a reason-
able amount of time. The question then
arises whether some other numerical
approach can be used.
A semi-analytical approach
The recent development of correlation trad-
ing and the associated need for fast compu-
tation of sensitivities have generated a great
deal of interest in semi-analytical models. To
enjoy the advantages of fast pricing, one
needs to impose more structure on the
problem. One way to do this is to rely on
two basic simplifications:
1. assume a one-factor correlation structure
for asset returns, and
2. discretise the time-line to allow for a finite
set of dates at which defaults can hap-
pen. These are chosen with a resolution
to provide a sufficient level of accuracy.
In particular, 1) makes it possible to com-
pute the risk-neutral loss distribution of the
reference portfolio for any given horizon,
while 2) is needed to price an instrument
knowing only the loss distribution of the
reference portfolio at a finite set of dates.
While these two assumptions are suffi-
cient to price plain vanilla portfolio swaps,
derivatives structures with more complex
path-dependencies may also require that
we approximate the payoff function f(τ).
Even in this case, it is usually possible to
obtain reasonable approximations, so that
the benefit of precise sensitivities can be
retained at a relatively low cost. Moreover,
the error can be controlled by comparing
the analytical solution to a Monte Carlo
implementation.
We now discuss how to exploit a one-factor
model to construct the risk neutral loss distri-
bution. Later on, we will show how to use a
sequence of loss distributions at different
horizons to price synthetic loss tranches.
guide.qxd 10/10/2003 11:15 Page 37
38 The Lehman Brothers Guide to Exotic Credit Derivatives
A one-factor model
Let us start by assuming that the asset
return of the ith issuer between today and a
given horizon is described by a standard
Normal random variable A
i
with mean 0 and
standard deviation of 1, and is of the form
where Z
1
,Z
2
,...,Z
D
are N(0,1) distributed
independent random variables. The variable
Z
M
describes the asset returns due to a
common market factor, while Z
i
models the
idiosyncratic risk of the ith issuer, and β
i
stands for the correlation of the asset return
of issuer i with the market. The asset return
correlation between asset i and asset j is
given by β
i
β
j
.
Default will occur if the realised asset
return falls below a given threshold.
Mathematically, the ith issuer defaults in the
event that A
i
< C
i
. Given that A
i
is N(0,1) dis-
tributed, it is possible to write
and calibration of the marginal probabilities
is no more complicated than inverting the
cumulative normal function.
In the single-issuer case, this is merely an
exercise in calibration, as the default thresh-
olds are chosen to reproduce default proba-
bilities which re-price market instruments.
The concept of asset returns becomes mean-
ingful when studying the joint behaviour of
more than one credit. In this case, the returns
are assumed to be described by a multivari-
ate Normal distribution. Using the threshold
levels determined before, it is possible to
obtain the probabilities of joint defaults.
With a general correlation structure, the
calculation of joint default probabilities
becomes computationally intensive, making
it necessary to resort to Monte Carlo simu-
lation. However, substantial simplifications
can be achieved by imposing more structure
on the model.
The advantage of this one-factor setup is
that, conditional on Z
M
, the asset returns are
independent. This makes it easy to compute
conditional default probabilities. Conditional
on the market factor Z
M
, an asset defaults if
The conditional default probability p
i
(Z
M
) of
an individual issuer i is therefore given by
If we assume that the loss on default for
each issuer is the same unit (consistent with
all assets having the same seniority), u, then
building up the portfolio loss distribution can
be done iteratively by adding assets to the
portfolio. The process is as follows:
1. Beginning with asset one, there are two
outcomes on the loss distribution: a loss
of zero with a probability 1–p
1
(Z
M
) and a
loss of u with a probability p
1
(Z
M
).
2. Adding a second asset, we adjust each of
the previous losses. The zero loss peak
requires that the new asset survives and
so has a probability (1–p
1
(Z
M
))(1–p
2
(Z
M
)).
A loss u corresponds with the previous
zero loss probability times the probability
that asset two defaults plus the previous
u loss probability times the probability
asset two does not default. We arrive at a

,
_

¸
¸

2
1
) (
i
M i i
M i
Z C
N Z p
β
β
2
1
i
M i i
i
Z C
Z
β
β



( ) ) (
1
i i i i
p N C C N p


i i M i i
Z Z A
2
1 β β − +
guide.qxd 10/10/2003 11:15 Page 38
The Lehman Brothers Guide to Exotic Credit Derivatives 39
probability ( p
1
(Z
M
)+p
2
(Z
M
)–2p
1
(Z
M
)p
2
(Z
M
)).
A loss of 2u can only occur if the previous
loss of u is multiplied by the probability
that asset two also defaults to give a
probability of p
1
(Z
M
)p
2
(Z
M
).
3. We continue adding on to the portfolio
until all of the assets have been includ-
ed. We then repeat with a different value
of Z
M
and integrate the loss distribu-
tions over the N(0,1) distributed market
factor Z
M
.
The advantage of this approach is that there
is no simulation noise. Other algorithms
exist, including Fast Fourier techniques
which may be more efficient.
Both the times-to-default model and the
one-factor model described above are
Gaussian copula models. This means that as
long as both are calibrated to the same
marginals, and as long as both use the same
(one-factor) correlation matrix, both models
should generate exactly the same prices.
This is an important point – it means that
models can be categorised by their copula
specification alone. It also means that differ-
ent products may be priced consistently
using different default mechanisms, as long
as they use the same copula.
Valuation of correlation
products
We now explain in detail how these models
may be applied to the pricing of correlation
products. We start with the default basket.
Pricing default baskets
A default basket is inherently an idiosyncrat-
ic product in the sense that the identity of
the defaulted asset must be known. One
approach is therefore to use the times-to-
default model via a Monte Carlo implemen-
tation. The generation of default times
within this framework has been described in
detail above. Given that we know when each
asset in the portfolio defaults in each simu-
lation path, to calculate the fair-value spread
we proceed as follows:
1. Sort default times in ascending order and
denote the nth time to default as τ
n
(i),
where i is the label of the defaulted asset.
2. Calculate the PV of a 1bp coupon stream
paid to time τ*=min[τ
n
(i),T] where T is the
basket maturity.
3. If τ
n
(i) < T then calculate the PV =
B(τ
n
(i))(1–R(i)), where B is the Libor dis-
count factor and R(i) is the recovery rate for
asset i. Otherwise the protection PV = 0.
4. Average both the premium leg PV of 1bp,
known as the basket PV01, and the pro-
tection leg PV over all paths.
5. Divide the protection leg by the basket
PV01 to get the fair-value spread.
This approach is simple to implement and
the size of default baskets, typically m=5–10
assets means that pricing is quite fast in
Monte Carlo. Monte Carlo is also flexible
enough to enable you to introduce stochas-
tic recovery rates, perhaps drawn from a
beta distribution. It is also quite straightfor-
ward to introduce alternative copulas, see
Mashal and Naldi (2002a).
An analytical approach is also possible here.
The main constraint is to build the basket n-
to-default probabilities while retaining the
identity of the defaulted assets. See Gregory
and Laurent (2003) for a full discussion.
Rating models for default baskets
Rating agencies have developed a number
of models for rating default baskets. For
example Moody’s has adopted a Monte
Carlo-based extension of the asset value
approach, in which an asset value is simu-
guide.qxd 10/10/2003 11:15 Page 39
40 The Lehman Brothers Guide to Exotic Credit Derivatives
lated to multiple periods for each of the
assets in the basket. These asset values
may be correlated internally in order to
induce a default correlation. If the asset
value falls below the default threshold at
the future period, the asset defaults and a
recovery rate is drawn from a beta distribu-
tion. The recovery amount and asset value
can be correlated. The model is calibrated
using historical default statistics and the
assigned rating is linked to the expected
loss of the basket to the trade maturity.
S&P has recently switched its approach
from a weakest-link approach which assigns
an FTD the rating of the lowest-rated entity
in the basket, to one based on a Monte Carlo
model, which uses the same framework as
its CDO Evaluator model.
Pricing synthetic loss tranches
We would now like to consider in detail
how we would set about valuing a loss
tranche. One approach is to use the times
to default model via a Monte Carlo imple-
mentation as discussed earlier. Given that
we know when each asset in the portfolio
defaults in each simulation path, we would
proceed as follows:
1. Calculate the present value of all principal
losses on the protection leg in each path.
2. Calculate the PV of 1bp on the premium
leg taking care to reduce the notional of
the tranche following defaults which
cause principal losses in that path.
3. Average both the 1bp premium leg PV,
known as the tranche PV01, and the pro-
tection leg PV over all paths.
4. Divide the protection PV by the tranche
PV01 to compute the breakeven tranche
spread.
This approach is simple to implement but
can be computationally slow if there are a
large number of assets in the portfolio.
Another way is to use a semi-analytical
approach which relies on the fact that a
standard synthetic loss tranche can be
priced directly from its loss distribution.
To see this, consider a portfolio CDS on a
tranche defined by the attachment point K
d
and the upper boundary K
u
, expressed as
percentages of the reference portfolio
notional. This is N
port
so that the tranche
notional is given by
The maturity of the portfolio swap is T, and
for each time t≤T we denote by L(t) the
cumulative portfolio loss up to time t. The
tranche loss is therefore given by
Note that this is similar to an option style
payoff. Indeed it is possible to think of CDO
tranches as options on the portfolio loss
amount.
The two legs of the swap can be priced in
the same way as a CDS if we introduce a
tranche ‘default probability’ P(t) defined as
where we use the risk-neutral (pricing) mea-
sure for taking the expectation.
Assuming that the credit, recovery rate and
interest rates processes are independent,
the contingent leg is therefore given by
Protection Leg PV =
tranche
tranche
Q
N
t L E
t P
)] ( [
) (
0

] 0 , ) ( max[ ] 0 , ) ( max[ ) (
u d tranche
K t L K t L t L − − −
) (
d u port tranche
K K N N −
guide.qxd 10/10/2003 11:15 Page 40
The Lehman Brothers Guide to Exotic Credit Derivatives 41
Where B(0,u) is the Libor discount factor
from today, time 0 to time u. We can dis-
cretise the integral by introducing the grid
points 0=t
0
<t
1
<…<t
K
=T where greater
accuracy is obtained by having a higher
number of grid points. In practice we
would generally assume that monthly
intervals would be sufficient. There are a
number of ways of evaluating an integral
on this interval, the simplest being first
order differences:
Protection Leg PV =
To value the premium leg, we denote the
contractual spread on the tranche by s, and
the coupon payment dates as
0=T
0
<T
1
<…<T
k
=T. The accrual factor for
the ith payment is denoted by ∆
i
. The PV of
the premium leg is then given by
Premium Leg PV=
The PV of the tranche from the perspective
of the investor who is receiving the spread is
therefore given by
Tranche PV = Premium Leg PV – Protection
Leg PV.
As a result we can value a standard syn-
thetic CDO if we have the loss distribution at
the set of future dates t and T. One way to
do this is to use the one-factor model
described above to calculate a loss distribu-
tion at each of the future dates required.
Care must be taken to ensure that the
default threshold is re-calibrated at each
horizon date such that the marginal distribu-
tion is correctly recovered.
This approach can be more efficient than
Monte Carlo since it is uses the loss distri-
bution directly and there are fast ways of
calculating this. To emphasise this point,
one can actually exploit the large number
of assets in a synthetic CDO to derive a
closed-form analytical solution to calculat-
ing the loss distribution of a portfolio.
An asymptotic approximation
It is possible to exploit the high dimension-
ality of the CDO to derive a closed form
model for analysing CDO tranches. In fact
we can use this approach to obtain results
which differ from the exact approach by
1–5% for a real CDO.
To begin with, let us assume that the port-
folio is homogeneous, ie, that each asset’s β
and default probability are the same. Hence,
all assets have the same default threshold C.
As a result, the conditional default probabili-
ty of any individual issuer in the reference
portfolio is given by
If we also assume that the loss exposure to
each issuer is of the same notional amount
u, the probability that the percentage loss L
of the portfolio is ku is equal to the proba-
bility that exactly k of the m issuers default,

,
_

¸
¸

2
1
) (
β
β
M
M
Z C
N Z p
.

− ∆
n
j
j j j tranche
T B T P sN
1
) , 0 ( )) ( 1 (



K
i
i i i tranche
t P t P t B N
1
1
)) ( ) ( )( , 0 (

T
tranche
u dP u B N
0
) ( ) , 0 (
guide.qxd 10/10/2003 11:15 Page 41
42 The Lehman Brothers Guide to Exotic Credit Derivatives
which is given by the simple binomial
This distribution becomes computationally
intensive for large values of m, but we can
use methods of varying sophistication to
approximate it. One very simple and sur-
prisingly accurate method is the so-called
‘large homogeneous portfolio’ (LHP)
approximation, originally due to Vasicek
(1987). Since the asset returns, conditional
on Z
M
are independent and identically dis-
tributed, by the law of large numbers, the
fraction of issuers defaulting will tend to
the conditional probability of an asset
defaulting p(Z
M
). As a result, the condition-
al loss is directly linked to the value of the
market factor Z
M
which itself is normally
distributed. We can then write the probabil-
ity of the portfolio loss being less than or
equal to some loss threshold K as
where N(x) is the cumulative normal func-
tion. More involved calculations show that
the distribution of L actually converges to
this limit as m tends to infinity.
We can use the portfolio distribution to
derive the loss distributions of individual
tranches. If L(K
1
,K
2
) is the percentage loss
of the mezzanine tranche with attachment
point K
1
and upper loss threshold K
2
, then
this can be written as a function of the loss
on the reference portfolio. As a fraction of
the tranche notional this is given by
If K<1, then
This equation shows that we can easily derive
the loss distribution of the tranche from that
of the reference portfolio. This is discontinu-
ous at the edges owing to the probability of
the portfolio loss falling outside the interval,
and this discontinuity becomes more pro-
nounced when the tranche is narrowed. We
can further compute the expected loss of the
tranche analytically. For more details see
O’Kane and Schloegl (2001). We then arrive at
This is only a one-period approach.
However, it can easily be extended to multi-
ple periods as described earlier.
Correlation skew
It is now possible to observe tradable
tranche spreads for different levels of
seniority. If we attempt to imply out the mar-
ket correlation using a simple Gaussian cop-
ula model fitted to observed market tranche
spreads, we can observe a skew in the aver-
age correlation as a function of the width
and attachment point of the tranche. This
skew may be the consequence of a number
of factors such as the assumption of inde-
pendence of default and recovery rates. It
may also be due to our incorrect specifica-
tion of the dependence structure, as
( ) [ ]
( ) ( ) ( ) ( )
1 2
2
2
1
2
2
1
1
2
2 1
1 , , 1 , ,
,
K K
C K N N C K N N
K K L E

− − − − − − −

− −
β β
( ) ( )
1 2 1 2 1
, K K K K L K K K L − + ≤ ⇔ ≤
( )
( ) ( )
1 2
2 1
2 1
0 , max 0 , max
,
K K
K L K L
K K L

− − −

[ ] [ ] ) (
1
K p N K L P

− ≤
[ ]
k m
M
k
M
Z C
N
Z C
N
k
m
Z ku L P

,
_

¸
¸

,
_

¸
¸


,
_

¸
¸

,
_

¸
¸

2
2
1
1
1
|
β
β
β
β
guide.qxd 10/10/2003 11:15 Page 42
The Lehman Brothers Guide to Exotic Credit Derivatives 43
explained later. Supply and demand imbal-
ances also play a role.
Rating agency models for CDOs
Different rating agencies have their own
models for rating CDO tranches. All of them
attempt to capture the risks of CDOs in
terms of asset quality, recovery rates,
default correlation and structural features.
Moody’s standard rating model for CDO
tranches is a multinomial extension of the
Binomial Expansion Technique (BET) model.
To capture default correlation, the portfolio
is represented by a lower number of inde-
pendent assets, known as the diversity
score. Roughly speaking, this is the number
of independent assets which have the same
width of loss distribution as the actual CDO
reference portfolio of correlated assets. The
diversity score is determined using a lookup
table and is based on the incremental effect
of having groups of assets in the same
industry classification.
After calibrating to historical default data,
the model is able to generate an expected
loss for a tranche which takes into account
the subordination. This expected loss is then
mapped to a rating category.
In S&P’s ratings methodology, a Monte Carlo
simulation is used to derive a probability loss
distribution for the underlying collateral pool
based on the total principal balance of the
portfolio. Each corporate asset is assigned a
default probability based on S&P’s historical
default studies, dependent on its rating and
maturity. Corporate sectors are assumed to
have a correlation of 30% within a given
industry and 0% between industry sectors.
From these inputs and the par amounts of
each asset, a default probability distribution is
created. Unlike Moody’s which rates on the
basis of expected loss, S&P rates on the basis
of the probability of incurring a loss.
Estimating the
dependency structure
Several well-known multivariate models,
including the ones described earlier in this
chapter, rely on the assumption that the
dependence structure (or ‘copula’) of asset
returns is Normal. The widespread use of
the Normal dependence structure, which is
fully characterised by a correlation matrix,
is certainly related to its simplicity. It
remains to be seen, however, whether this
assumption is supported by empirical evi-
dence. A number of recent studies have
shown that the joint behaviour of equity
returns is better described by a ‘fat-tailed’
Student-t copula than by a Normal copula,
and that correlations are therefore not suf-
ficient to appropriately characterise their
dependence structure.
2
The first goal of
this section is to apply the same kind of
analysis to asset returns, and test the null
hypothesis of Gaussian dependence
versus the alternative of ‘joint fat tails’.
Of course, we face a major obstacle when
attempting to estimate the dependence
structure of asset returns: asset values are
not directly observable. In fact, the use of
unobservable underlying processes is one
of several criticisms that the structural
approach has received over the years. Given
the lack of observable asset returns, it has
become customary to proxy the asset
dependence with equity dependence, and
to estimate the parameters governing the
joint behaviour of asset returns from equity
return series.
3
However, the use of equity
returns to infer the joint behaviour of asset
2
See, for example, Mashal and Naldi (2002a) and Mashal
and Zeevi (2002).
3
Fitch Ratings (2003) have recently published a special
report describing their methodology for constructing port-
folio loss distributions: it is based on a Gaussian copula
parameterised by equity correlations.
guide.qxd 10/10/2003 11:15 Page 43
44 The Lehman Brothers Guide to Exotic Credit Derivatives
returns is often criticised on the grounds of
the different leverage of assets and equity.
The second goal of this section is to shed
some light on the magnitude of the error
induced by using equity data as a proxy for
asset returns.
To provide a plausible answer to these
questions, we first need to ‘back out’ asset
values from observable data. One way to
estimate the market value of a company’s
assets is to implement a univariate struc-
tural model. Such a procedure is at the
heart of KMV’s CreditEdge™, a popular
credit tool that first computes a measure of
distance-to-default and then maps it into a
default probability (EDF™) by means of a
historical analysis of default frequencies.
4
In
what follows, we summarise recent work
by Mashal, Naldi and Zeevi (2002), who use
the asset value series generated by KMV’s
model to study the dependence properties
of asset returns.
Methodology
A key observation in modelling and testing
dependencies is that any d-dimensional
multivariate distribution can be specified
via a set of d marginal distributions that are
‘knitted’ together using a copula function.
Alternatively, a copula function can be
viewed as ‘distilling’ the dependencies that
a multivariate distribution attempts to cap-
ture, by factoring out the effect of the
marginals. Copulas have many important
characteristics that make them a central
concept in the study of joint dependencies,
see, eg, the recent survey paper by
Embrechts et al. (2003).
A particular copula that plays a crucial role
in our study is given by the dependence
structure underlying the multivariate
Student-t distribution. The Gaussian distri-
bution lies at the heart of most financial
models and builds on the concept of corre-
lation; the Student-t retains the notion of
correlation but adds an extra parameter
into the mix, namely, the degrees-of-free-
10
1
10
2
10
3
10
4
10
5
10
6
10
0
10
1
10
2
10
3
Null DoF
T
e
s
t

S
t
a
t
i
s
t
i
c
DJIA Asset Returns Sensitivity Analysis
pvalue = 0.01
pvalue = 0.0001
Figure 26. DJIA portfolio: asset and
equity returns test statistics as
functions of null hypothesis for DoF
10
1
10
2
10
3
10
4
10
5
10
6
10
0
10
1
10
2
10
3
Null DoF
T
e
s
t

S
t
a
t
i
s
t
i
c
DJIA Equity Returns Sensitivity Analysis
pvalue = 0.01
pvalue = 0.0001
4
Copyright © 2000-2002 KMV LLC. All rights reserved. KMV
and the KMV logo are registered trademarks of KMV LLC.
CreditEdge and EDF are trademarks of KMV LLC.
guide.qxd 10/10/2003 11:15 Page 44
The Lehman Brothers Guide to Exotic Credit Derivatives 45
dom (DoF). The latter plays a crucial role in
modelling and explaining extreme co-
movements in the underlyings.
Moreover, it is well known that the
Student-t distribution is very ‘close’ to the
Gaussian when the DoF is sufficiently large
(say, greater than 30); thus, the Gaussian
model is nested within the t-family. The
same statement holds for the underlying
dependence structures, and the DoF param-
eter effectively serves to distinguish the two
models. This suggests how empirical stud-
ies might test whether the ubiquitous
Gaussian hypothesis is valid or not. In par-
ticular, these studies would target the
dependence structure rather than the distri-
butions themselves, thus eliminating the
effect of marginal returns that would ‘con-
taminate’ the estimation problem in the lat-
ter case. To summarise, the t-dependence
structure constitutes an important and quite
plausible generalisation of the Gaussian
modelling paradigm, which is our main moti-
vation for focusing on it.
With this in mind, the key question that we
now face is how to estimate the parameters
of the dependence structure. Mashal, Naldi
and Zeevi (2002) describe a methodology
which can be used to estimate the parame-
ters of a t-copula without imposing any
parametric restriction on the marginal distri-
butions of returns. They also construct a
likelihood ratio statistic to test the hypothe-
sis of Gaussian dependence, and compare
the dependence structures of asset and
equity returns to evaluate the common prac-
tice of proxying the former with the latter. In
the remainder of this section we summarise
their empirical findings.
Empirical results
For the purpose of this study, asset and
equity values are both obtained from KMV’s
database. The reader should keep in mind,
however, that equity values are observable,
while asset values have been ‘backed out’
by means of KMV’s implementation of a uni-
variate Merton model. We apply our analysis
to a portfolio of 30 credits included in the
Dow Jones Industrial Average and use daily
data covering the period from 31/12/00 to
8/11/02. The reader is referred to Mashal,
Naldi and Zeevi (2003) for more examples
using high yield credits and different sam-
pling frequencies.
Following the methodology mentioned
above, we estimate the number of degrees-
of-freedom (DoF) of a t-copula without
imposing any structure on the marginal dis-
tributions of returns. Then, using a likelihood
ratio test statistic, we perform a sensitivity
analysis for various null hypotheses of the
underlying tail dependence, as captured by
the DoF parameter. The two horizontal lines
in Figure 26 represent significance levels of
99% and 99.99%; a value of the test statis-
tic falling below these lines corresponds to a
value of DoF that is not rejected at the
respective significance levels.
The minimal value of the test statistic is
achieved at 12 DoF for asset returns and at
13 DoF for equity returns. In both cases, we
can reject any value of the DoF parameter
outside the range [10,16] with 99% confi-
dence; in particular, the null assumption of
a Gaussian copula (DoF=∞) can be rejected
with an infinitesimal probability of error.
Also, the point estimates of the asset
returns’ DoF lies within the non-rejected
interval for the equity returns’ DoF, and vice
versa, indicating that the two are essential-
ly indistinguishable from a statistical signif-
icance viewpoint. Moreover, the difference
between the joint tail behaviour of a 12- and
a 13-DoF t-copula is negligible in terms of
any practical application.
guide.qxd 10/10/2003 11:15 Page 45
46 The Lehman Brothers Guide to Exotic Credit Derivatives
Figure 27 reports the point estimates of
the DoF for asset and equity returns in the
DJIA basket, as well as for three subsets
consisting of the first, middle, and last 10
credits (in alphabetical order). The similari-
ties between the joint tail dependence (as
measured by the DoF) of asset and equity
returns are quite striking.
5
Next, we compare the remaining parame-
ters that define a t-copula, ie, the correlation
coefficients. Using a robust estimator based
on Kendall’s rank statistic
6
, we compute the
two 30x30 correlation matrices from asset
and equity returns. The maximum absolute
difference (element-by-element) is 4.6%,
and the mean absolute difference is 1.1%,
providing further evidence of the similarity
of the two dependence structures.
Example: synthetic loss tranche
The models described earlier in this chapter
can be modified to account for a fat-tailed
dependence structure of asset returns. Here
we analyse the impact of a non-Normal
assumption on the expected discounted loss
(EDL) of a portfolio loss tranche. We focus on
EDL because this measure relates both to the
agency rating (when computed under real-
world probabilities) and to the fair compensa-
tion for the credit exposure (when computed
under risk-neutral probabilities).
We consider a five-year deal with a refer-
ence portfolio of 100 credits, each with $1m
notional. We assume i) uniform recovery
rates of 35%, for every credit in the reference
portfolio; ii) 1% yearly hazard rate for each
reference credit; iii) 20% asset correlation
between every pair of credits; iv) a risk-free
curve flat at 2%. Using a default-time simula-
tion, Figure 28 compares the expected dis-
counted losses for several tranches under the
two alternative assumptions of Gaussian
dependence and t dependence with 12 DoF.
The results show the significant impact that
the (empirically motivated) consideration of
tail dependence has on the distribution of
losses across the capital structure: expected
losses are clearly redistributed from the junior
to the senior tranches, as a consequence of
the increased volatility of the overall portfolio
loss distribution. Notice that even larger dif-
ferences can be observed if one compares
higher moments or tail measures of the
tranches’ loss distributions.
The LHP with tail dependence
It is possible to incorporate a Student-t cop-
ula into the LHP model discussed above. To
do so, we must change the distribution for
the asset returns to be multivariate Student-
t distributed where we denote the DoF
parameter with ν and retain the one-factor
correlation structure. This gives
where Wis an independent random variable
2
1
/
i i i
i
Z
A
W
β β ε
ν
+ −

5
The range of accepted DoF is very narrow in each case,
exhibiting similar behaviour to that displayed in Figure 26.
6
See Lindskog (2000).
Figure 27. Maximum likelihood
estimates of DoF for DJIA portfolios
Portfolio Asset returns Equity returns
DoF DoF
30-credit DJIA 12 13
First 10 credits 8 9
Middle 10 credits 10 10
Last 10 credits 9 9
guide.qxd 10/10/2003 11:15 Page 46
The Lehman Brothers Guide to Exotic Credit Derivatives 47
following a chi-square distribution with ν
degrees of freedom. This is the simplest
possible way to introduce tail dependence
via a Student-t copula function.
Note that this is no longer a ‘factor model’ in
the sense that the asset return is composed
of two independent and random factors. The
fact that both the market and idiosyncratic
terms ‘see’ the same value of W means that
they are no longer independent.
For this model, we have been able to cal-
culate a closed-form solution for the densi-
ty of the portfolio loss distribution, and
show in O’Kane and Schloegl (2003) that it
possesses the same tail dependent proper-
ties as described above, ie, a widening of
senior spreads and a reduction in equity
tranche spreads.
Summary
Our empirical investigation of the depen-
dence structure of asset returns sheds
some light on the two main issues that
were raised at the beginning of this section.
First, the assumption of Gaussian depen-
dence between asset returns can be reject-
ed with extremely high confidence in favour
of an alternative ‘fat-tailed dependence.’
Multivariate structural models that rely on
the normality of asset returns will generally
underestimate default correlations, and
thus undervalue junior tranches and over-
value senior tranches of multi-name credit
products. A fat-tailed dependence of asset
returns will produce more accurate joint
default scenarios and more accurate valua-
tions. Second, the dependence structures
of asset and equity returns appear to be
strikingly similar. The KMV algorithm that
produces the asset values used in our anal-
ysis is nothing other than a sophisticated
way of de-leveraging the equity to get to the
value of a company’s assets. Therefore, the
popular conjecture that the different lever-
age of assets and equity will necessarily
create significant differences in their joint
dynamics seems to be empirically unfound-
ed. From a practical point of view, these
results represent good news for practition-
ers who only have access to equity data for
the estimation of the dependence parame-
ters of their models.
Modelling credit options
We separate credit options into options on
bonds and options on default swaps.
Pricing options on bonds
Options on corporate bonds are naturally
divided into three groups according to how
the exercise price is specified. The option
can be struck on price, yield or credit
spread. The exercise price is constant for
options struck on price, but for options
struck on yield it depends on the time to
maturity of the underlying bond and is found
from a standard yield-to-maturity calculation.
Obviously, for European options there is no
difference between specifying a strike price
and a strike yield.
Bond options struck on spread are differ-
ent. For credit spread options the exercise
price depends both on the time to maturity
Figure 28. Expected discounted loss,
100K paths, standard errors in parenthesis
Tranche Normal copula t copula DoF=12 Pctg
(%) EDL (std err %) EDL (std err %) diff
0-5 $2,256,300 (0.14) $2,012,200 (0.23) -11
5-10 $533,020 (0.63) $601,630 (0.66) 13
10-15 $146,160 (1.37) $221,120 (1.06) 51
15-20 $41,645 (1.70) $90,231 (1.62) 117
guide.qxd 10/10/2003 11:15 Page 47
48 The Lehman Brothers Guide to Exotic Credit Derivatives
of the bond and on the term structure of
interest rates at the exercise time. A credit
spread strike is commonly specified as a
yield spread to a Treasury bond or interest
rate swap, or as an asset swap spread.
For short-dated European options with
price (or yield) strike on long-term bonds,
the well-known Black-Scholes formula goes
a long way but is not recommended
beyond this limited universe. An important
problem with Black-Scholes is that it does
not properly account for the pull-to-par of
the bond price. This problem can be solved
by a yield diffusion model where the bond’s
yield is the underlying stochastic process.
It is relatively easy to build a lattice for a
lognormal yield, say, and price the option
by standard backwards induction tech-
niques. The lattice approach also allows for
easy valuation of American/Bermudan exer-
cise. Yield diffusion models can be con-
structed to fit forward bond prices but
would usually assume constant yield
volatility which is inconsistent with empiri-
cal evidence.
The problems of pull-to-par of price and
non-constant volatility can be solved by a full
term structure model such as Black-Derman-
Toy, Black-Karasinski or Heath-Jarrow-
Morton. These models were designed for
default-free interest rates but can be applied
analogously to credit risky issuers. Instead of
calibrating the model to Libor rates, the
model should be calibrated to an issuer-spe-
cific credit curve.
It is rarely possible to calibrate volatility
parameters because of the lack of liquid
bond option prices. Usually it is more appro-
priate to base volatility parameters on histor-
ical estimates. Naldi, Chu and Wang (2002)
and Berd and Naldi (2002) present a multi-
factor framework for modelling the stochas-
tic components of corporate bond returns.
The framework can be used to derive
price/yield volatilities from return volatilities
and tend to give more robust estimates than
direct estimation.
Extending interest rate models to mod-
elling risky rates only works if we assume
either zero recovery on default, or assume
that recovery is paid as a fraction of the mar-
ket value at default. However, creditors have
a claim for return of full face value in
bankruptcy, so it is necessary to specifically
model the recovery at default as a fraction of
face value. This is especially important for
lower credit quality issuers where the bonds
trade on price rather than yield.
For high quality issuers, the above
approaches are less controversial for pric-
ing options struck on price or yield. For
these issuers, interest rate volatility is the
main driver of price action. Volatility param-
eters should therefore be related to
implied volatilities from interest rate swap-
tion markets but must also incorporate the
negative correlation between credit
spreads and interest rates (see Berd and
Ranguelova (2003)) which can cause yield
volatility on corporate bonds to be signifi-
cantly lower than comparable interest rate
swaption volatility.
When pricing spread options, it is impor-
tant to specifically take into account the
default risk. In the next section we discuss


Default
swaption
expiry date
T
CDS
maturity
date
T
M

Default
swaption
settlement

T
S
Default swap cash flows if exercised into
Figure 29. Receiver default swaption
guide.qxd 10/10/2003 11:15 Page 48
The Lehman Brothers Guide to Exotic Credit Derivatives 49
how to price options to buy or sell protection
through CDS. Bond options struck on credit
spread can be priced in similar fashion.
Pricing default swaptions
The growing market in default swaptions has
led to a demand for models to price these
products. First, let us clarify terminology. An
option to buy protection is a payer swaption
and an option to sell protection is a receiver
swaption. This terminology is analogous to
that used for interest rate swaptions.
Black’s formulas for interest rate swap-
tions can be modified to price European
default swaptions. Consider a European
payer swaption with option expiry date T and
strike spread K. The contract is to enter into
a long protection CDS from time T to T
M
, and
it knocks out if default occurs before T (see
Figure 29). Conditional on surviving to time
T, the option payoff is
where the PV01
T
is the value at T of a risky
1bp annuity to time T
M
or default, and S
T
is
the market spread observed at T on a CDS
with maturity T
M
.
Ignoring the maximum, this is the stan-
dard payoff calculation for a forward start-
ing CDS. See page 32 for a discussion of
CDS pricing.
From finance theory we know that for any
given security, say B, that only makes a pay-
ment at T, there is a probability distribution
of the spread S
T
such that for any other
security, say A, that also only makes a pay-
ment at T, the ratio A
0
/B
0
of today’s values of
the securities is equal to the expectation of
the ratio A
T
/B
T
of the security payments at T.
The result is valid even if B
T
can be 0 as long
as A
T
= 0 when B
T
= 0. The states where B
T
= 0 are simply ignored in that case and the
distribution of S
T
will be such that the proba-
bility that B
T
= 0 is 0. (See Harrison and
Kreps (1979) for details.)
To use this result we first let A be a securi-
ty that at T pays 0 if default has occurred and
otherwise pays the upfront cost (as of T) of a
zero-premium CDS with the same maturity
as the CDS underlying the swaption. We let
B be a security that at T pays 0 if default has
occurred and otherwise pays PV01
T
. With
these definitions the ratio A
T
/B
T
is equal to
the spread S
T
if default has not occurred at T,
otherwise A
T
/B
T
and S
T
are not defined. The
distribution of S
T
should then be such that
E[S
T
] = A
0
/B
0
where the probability of
default before T is put to 0. A
0
/B
0
is the T-for-
ward spread, denoted F
0
, for the underlying
CDS where the forward contract knocks out
if default occurs before T.
The next step is to let A be the swaption,
in which case A
T
is 0 if default happens
before T and PS
T
otherwise. The value of the
swaption today is then
If we make the assumption that log(S
T
) is
normally distributed with variance σ
2
T, corre-
sponding to the spread following a log-nor-
mal process with constant volatility σ, then
with the requirement E(S
T
) = F
0
(the forward
spread), we have determined the distribution
of S
T
to be used to find E[max{S
T
–K,0}]. It is
easy to calculate this expectation and we
arrive at the Black formula
( ), ) ( ) ( 01
2 1 0 0 0
d N K d N F PV PS ⋅ − ⋅ ⋅
T d d
T
T K F
d σ
σ
σ

+

1 2
2
0
1

2 / ) / log(
¦ ¦ [ ] 0 , max 01
0 0 0
K S E PV
B
PS
E B PS
T
T
T
− ⋅
1
]
1

¸

¦ ¦ 0 , max 01 K S PV PS
T T T
− ⋅
guide.qxd 10/10/2003 11:15 Page 49
50 The Lehman Brothers Guide to Exotic Credit Derivatives
where N is the standard normal distribution
function. The Black formula for a receiver
swaption is found analogously. (See Hull and
White (2003) and Schonbucher (2003) for
more details.)
It is important that the forward spread, F
0
,
and PV01
0
are values under the knockout
assumption. Calculation of F
0
and PV01
0
should be done using a credit curve that has
been calibrated to the current term structure
of CDS spreads.
To value a payer swaption that does not
knock out at default, we must add the value
of credit protection from today until option
maturity. Under a non-knockout payer swap-
tion, a protection payment is not made until
option maturity. The value of the credit pro-
tection is therefore less than the upfront
cost of a zero-premium CDS that matures
with the option. The knockout feature is not
relevant for receiver swaptions, as they will
never be exercised after default.
The last step in valuing the swaption is to
find the volatility σ to be used in the Black
formula. An estimate of σ can be made from
a time series of CDS spreads. Examination
of CDS spread time series also reveals that
the log-normal spread assumption often is
inappropriate. It is not uncommon for CDS
spreads to make relatively large jumps as
reaction to firm specific news. However cal-
ibrating a mixed jump and Brownian process
to spread dynamics is not easy.
To price default swaptions with
Bermudan exercise we could construct a
lattice for the forward spread, but criticism
analogous to that for using a yield diffusion
model to price bond options applies.
Instead, we recommend using a stochastic
hazard rate model.
To illustrate the use of the Black formula,
consider a payer swaption on Ford Motor
Credit with option maturity 20/12/2003 and
swap maturity 20/9/2008. The valuation
date is 28/8/2003. The strike is 260, which
is the five year CDS spread on the valuation
date. Our trading desk quoted the swaption
at 2.29%/2.52%, which means that $10m
notional can be bought for $252,000 and
sold for $229,000. These are prices of non-
knockout options. The fair value of protec-
tion until swaption maturity given the credit
curve on the valuation date is 0.211%, the
PV01 used in the Black formula is 4.027 and
the forward spread is 274.7bp. These num-
bers imply that the bid/offer volatilities
quoted by our desk are approximately 75%
and 85%.
Interest rates and credit risk
One way to model credit spread dynamics is
via a hazard rate process. This provides a
consistent framework for modelling spreads
of many different maturities. The differences
between directly modelling the credit
spread and modelling the hazard rate are
similar to the differences between mod-
elling the yield of a bond and using a term
structure model as discussed above.
A stochastic hazard rate model is natural-
ly combined with a term structure model to
produce a unified model that can, at least in
theory, be used to price both bond options
and credit spread options. However, a num-
ber of practical complications arise in get-
ting such an approach to work. Since bond
and the CDS markets have their own
dynamics, usually not implying the same
issuer curves, it may not be appropriate to
naively price all options in the same cali-
brated model without properly adjusting for
the basis between CDS and bonds.
Also, calibration of the volatility parame-
ters of the hazard rate process, when possi-
ble, is less than straightforward, especially
when calibrating to bond options struck on
guide.qxd 10/10/2003 11:15 Page 50
The Lehman Brothers Guide to Exotic Credit Derivatives 51
price or yield, or to bonds with embedded
options. Interest rate volatility parameters
should be calibrated separately using liquid
prices of interest rate swaptions. These
parameters are then taken as given when
determining the hazard rate volatility param-
eters by fitting to time series estimates or
calibrating to market prices.
Finally, it is necessary to determine the cor-
relation between interest rates and hazard
rates. When estimating correlations it is
important not to use yield spread or OAS
(option adjusted spread) as a proxy for the
hazard rate, because such spread measures
do not properly take into account the risk of
default. Using OAS is especially a problem for
long maturity bonds with high default proba-
bilities and high recovery rates. For such
bonds there can be a significant decrease in
OAS when interest rates increase even if haz-
ard rates remain unchanged. Correlations
should be estimated directly using bond
prices or CDS spreads.
Modelling hybrids
The behaviour of hybrid credit derivatives is
driven by the joint evolution of credit
spreads and other market variables such as
interest and exchange rates, which are com-
monly modelled as diffusion processes. As
a consequence, the reduced form approach
is the natural framework for pricing and
hedging these products.
We illustrate the main ideas via the exam-
ple of default protection on the MTM of an
interest rate swap. Suppose an investor
has entered into a receiver swap with fixed
rate k with a credit risky counterparty. If the
MTM of the receiver swap, RS
t
is positive
to the investor at the time of default, this is
paid by the protection seller. If RS
t
is nega-
tive, the investor receives nothing, so that
the payoff at default is max(RS
t
,0). This is
an option to enter into a receiver swap with
fixed rate k for the remaining life of the
original trade at default. For simplicity, we
assume that default can only take place at
times t
i
. If B denotes the price process of
the savings account, then computing the
expected discounted cash flows gives the
value V
0
for the price of the default protec-
tion, where
The interpretation of this equation is that the
value of default protection is a probability
weighted strip of receiver swaptions, where
each swaption is priced conditional on
default happening at t
i
.
Representing the default protection via a
strip of swaptions is a very useful framework
for developing intuition around the pricing
and helps us understand the importance of
the shape of the interest rate curve. In terms
of volatility exposure, the protection seller
has sold swaptions and is therefore short
interest rate volatility.
If the rate and credit process are correlat-
ed, then there will also be a spread volatility
dependence. However, under the assump-
tion of independence, the volatility of the
credit spread does not enter into the valua-
tion, only the default probabilities.
A strip decomposition, as we have shown
above, is the basic building block for most
hybrid credit derivatives. A cross-currency
swap for example could be dealt with in
exactly the same way, with the exchange
rate as an additional state variable. For this
instrument, the exchange rate exposure is of
prime importance.
In terms of tractability, it is important to be
( )
( )
[ ]
0
1
max , 0
i
n
t
i i
i i
RS
V E t P t
B t
τ τ

1
1

1
¸ ]

guide.qxd 10/10/2003 11:15 Page 51
52 The Lehman Brothers guide to exotic credit derivatives
able to calculate the conditional expecta-
tions appearing in the strip decomposition.
A further consequence of the hazard rate
method is that we do not actually have to
condition on the realisation of the default
time itself, conditioning instead on the reali-
sation of the hazard rate process. This is par-
ticularly advantageous for Monte Carlo
simulation, as one does not have to explicit-
ly simulate default times and is a useful vari-
ance reduction technique.
The parameters needed for pricing
hybrids are essentially volatility and depen-
dence parameters. Calibrating volatilities
for interest rates and FX is relatively
straightforward. Credit spread volatilities
are somewhat more involved. Until recent-
ly, we have had to rely on estimates of his-
torical volatilities; now the growing options
markets are starting to make the calibration
of implied spread volatilities feasible.
Determining the correct dependence
structure between credit spreads and the
other market variables is the main challenge
in modelling hybrids. The simplest approach
is to work within a diffusion setting where
spreads and interest rates/FX are correlated.
Such a model will not necessarily generate
levels of dependence representative of peri-
ods of market stress where investment
grade defaults are likely. As a starting point,
however, it appears to be reasonable.
Even within this framework, the effect of
correlation on the valuation of a hybrid
instrument can be marked. At this stage of
the market it is safe to say that this correla-
tion is a ‘realised’ parameter as opposed to
an implied one, ie pricing and hedging
must proceed on the basis of a view on cor-
relation founded on historical estimates.
Going forward, it will be interesting to see
to what extent a market in implied correla-
tions will develop via standardised hybrid
credit derivatives.
guide.qxd 10/10/2003 11:15 Page 52
The Lehman Brothers Guide to Exotic Credit Derivatives 53
Altman E, Resti A and Sirone A, 2001
Analyzing and explaining default
recovery rates
Report submitted to ISDA, Stern School of
Business, New York University, December
Berd A and Naldi M, 2002
New estimation options for the Lehman
Brothers risk model
Quantitative Credit Research Quarterly,
Lehman Brothers, September
Berd A and Ranguelova E, 2003
The co-movement of interest rates and
spreads: implications for credit investors
High Grade Research, Lehman Brothers,
Embrechts P, Lindskog F and
McNeal A, 2003
Modelling dependence with copulas and
applications to risk management
In Handbook of Heavy Tailed Distributions in
Finance, Elsevier, New York, pages 329–384
Finger C, 1999
Conditional approaches for CreditMetrics
portfolio distributions
RiskMetrics Group, New York
Finger C, 2000
A comparison of stochastic default
rate models
RiskMetrics Group, New York
Fitch Ratings, 2003
Default correlation and its effect on
portfolios of credit risk
Structured Finance, Credit Products Special
Report, February
Frye J, 2003
A false sense of security
Risk August, pages 95–99
Ganapati S, Ha P and O’Kane D, 2001
Synthetic CDOs – how are they structured?
how do they work?
CDO Monthly Update, Lehman Brothers,
November
Ganapati S and Ha P, 2002
Evolution of synthetic arbitrage CDOs
Structured Credit Strategies, Lehman
Brothers, November
Ganapati S, Berd A, Ha P and
Ranguelova E, 2003
The synthetic CDO bid and basis conver-
gence – which sector is next?
Structured Credit Strategies, Lehman
Brothers, March
Gregory J and Laurent J-P, 2003
I will survive
Risk June, pages 103–107
Harrison J and Kreps D, 1979
Martingales and arbitrage in multiperiod
securities markets
Journal of Economic Theory 20, pages
381–408
Hull J, 2000
Options, futures and other derivatives
Fourth edition, Prentice Hall
Hull J and White A, 2001
Valuing credit default swaps II: modelling
default correlations
Journal of Derivatives 8, pages 12–22
Hull J and White A, 2003
The valuation of credit default swap options
Working paper, University of Toronto
Isla L, 2003
Hedged CDO equity strategies
Fixed Income Research, Lehman Brothers,
September
References
guide.qxd 10/10/2003 11:15 Page 53
54 The Lehman Brothers Guide to Exotic Credit Derivatives
Jarrow R and Turnbull S, 1995
Pricing derivatives on financial securities
subject to credit risk
Journal of Finance 50, pages 53–85
Li D, 2000
On default correlation: a copula
function approach
Journal of Fixed Income 9, pages 43–54
Lindskog F, 2000
Linear correlation estimation
Working paper, RiskLab, ETH Zurich
Mashal R and Naldi M, 2002a
Extreme events and default baskets
Risk June, pages 119–122
Mashal R and Naldi M, 2002b
Beyond CADR: searching for value in the
CDO market
Quantitative Credit Research Quarterly,
Lehman Brothers, September 2002
Mashal R, Naldi M and Pedersen C, 2003
Leverage and correlation risk of synthetic
loss tranches
Quantitative Credit Research Quarterly,
Lehman Brothers, April
Mashal R, Naldi M and Zeevi A, 2002
The dependence structure of asset returns
Quantitative Credit Research Quarterly.
Lehman Brothers, December
Mashal R, Naldi M and Zeevi A, 2003
Extreme events and multi-name
credit derivatives
In Credit Derivatives The Definitive Guide,
Risk, London
Mashal R and Zeevi A, 2002
Beyond correlation: extreme co-movements
between financial assets
Working paper, Columbia University,
New York
Munves D, Berd A, O’Kane D and
Desclee R, 2003
The Lehman Brothers CDS indices
Fixed Income Research, Lehman Brothers,
September
Naldi M, Chu K and Wang G, 2002
The new Lehman Brothers credit risk model
Quantitative Credit Research Quarterly,
Lehman Brothers, May
O’Kane D and McAdie R, 2001
Explaining the basis: cash versus
default swaps
Lehman Brothers, March
O’Kane D, Pedersen C and Turnbull S, 2003
Valuing the restructuring clause in CDS
Lehman Brothers, June
O’Kane D and Schloegl L, 2001
Modelling credit: theory and practice
Lehman Brothers, February
O’Kane D and Schloegl L, 2003
Leveraging the spread premium with
correlation products
Quantitative Credit Research Quarterly,
Lehman Brothers, July
O’Kane D and Sen S, 2003
Upfront credit default swaps
Quantitative Credit Research Quarterly,
Lehman Brothers, July
O’Kane D and Turnbull S, 2003
Valuation of credit default swaps
Quantitative Credit Research Quarterly,
Lehman Brothers, June
Schonbucher P, 2003
A note on survival measures and the pricing
of options on credit default swaps
Working paper, ETH Zurich
Vasicek O, 1987
Probability of loss on loan portfolio
Working paper, KMV Corporation
guide.qxd 10/10/2003 11:15 Page 54
The Lehman Brothers Guide to Exotic Credit Derivatives 55
Approved by Lehman Brothers International
(Europe), which is authorised and regulated by the
Financial Services Authority. Rules made for the
protection of investors under the UK Financial
Services Act may not apply to investment business
conducted at or from an office outside the UK.
Published by Incisive Risk Waters Group,
Haymarket House, 28–29 Haymarket,
London SW1Y 4RX
© 2003 Lehman Brothers and Incisive RWG Ltd
All rights reserved. No part of this publication may
be reproduced, stored in or introduced into any
retrieval system, or transmitted, in any form or by
any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior written
permission of the copyright owners.
Risk Waters Group and Lehman Brothers and the
author have made every effort to ensure the
accuracy of the text; however, neither they nor any
company associated with the publication can
accept legal or financial responsibility for
consequences, which may arise from errors,
omissions, or any opinions given.
guide.qxd 10/10/2003 11:15 Page 55
56 The Lehman Brothers Guide to Exotic Credit Derivatives
Notes
guide.qxd 10/10/2003 11:15 Page 56
The Lehman Brothers Global Fixed Income Research
Group has been consistently recognised as a top-ranked
research department.
.
For the fourth year in a row, the Group was ranked #1
in the Institutional Investor 2003 All-America Fixed
Income Research Team Survey
.
Institutional Investor ranked Lehman Brothers #1 for
Fixed Income Research in 10 out of the last 14 years
.
Lehman Brothers website, LehmanLive - winner of 8
“Best Site” categories in the Euromoney.com Awards
.
LehmanLive - #1 dealer website on the Bond.Hub portal
.
Dominant global marketshare in Fixed Income Indices
and Index Analytics - available on our POINT
TM
platform
We have an established tradition of providing the highest
quality research services to our clients. With over 375
dedicated professionals worldwide covering the whole
spectrum of Fixed Income products, we provide strategic
advice and analysis to our institutional clients.
All of our research and analysis is accessible through our
website, LehmanLive.
For further information please contact the Global Fixed
Income Research Group or visit www.lehmanlive.com
New York: Ravi Mattu
+1 212 526 0206 rmattu@lehman.com
Europe: Robert Campbell
+44-(0)20-7260-2639 rcampbel@lehman.com
Prafulla Nabar
+44-(0)20-7256-4480 pnabar@lehman.com
Asia: Hua He
+81-3-5571-7215 hhe@lehman.com
Award-winning website
www.lehmanlive.com
In-depth product
analysis
Weekly commentary
across asset classes
All Rights Reserved. Member SIPC. Lehman Brothers International (Europe) is regulated by the Financial Services Authority. ©2003 Lehman Brothers Inc.
An Established Tradition
of Research Excellence
Document1 06/10/2003 09:59 Page 1
THE LEHMAN BROTHERS
GUIDE TO EXOTIC CREDIT DERIVATIVES
T
H
E

L
E
H
M
A
N

B
R
O
T
H
E
R
S
G
U
I
D
E

T
O

E
X
O
T
I
C

C
R
E
D
I
T

D
E
R
I
V
A
T
I
V
E
S
lehman cover.qxd 10/10/2003 11:03 Page 1

Effective Structured Credit Solutions for our Clients
Structured Credit Solutions

With over seventy professionals worldwide, Lehman Brothers gives you access to top quality risk-management, structuring, research and legal expertise in structured credit. The team combines local market knowledge with global co-ordinated expertise. Lehman Brothers has designed specific solutions to our clients’ problems, including yield-enhancement, capital relief, portfolio optimisation, complex hedging and asset-liability management.

Product Innovation

. . . . . . . . .
Leadership in Fixed Income Research

Credit Default Swaps Portfolio Swaps Credit Index Products Repackagings Default Baskets Secondary CDO trading Customised CDO tranches Default swaptions Credit hybrids

For further information please contact your local sales representative or call:
London: Giancarlo Saronne
+44 20 7260 2745 gsaronne@lehman.com

New York: Mike Glover
+1 212 526 7090 mglover@lehman.com

Tokyo:

Jawahar Chirimar
81-3-5571-7257 jchirima@lehman.com

All Rights Reserved. Member SIPC. Lehman Brothers International (Europe) is regulated by the Financial Services Authority. ©2003 Lehman Brothers Inc.

Foreword
The credit derivatives market has revolutionised the transfer of credit risk. Its impact has been borne out by its significant growth which has currently achieved a market notional close to $2 trillion. While not directly comparable, it is worth noting that the total notional outstanding of global investment grade corporate bond issuance currently stands at $3.1 trillion. This growth in the credit derivatives market has been driven by an increasing realisation of the advantages credit derivatives possess over the cash alternative, plus the many new possibilities they present to both credit investors and hedgers. Those investors seeking diversification, yield pickup or new ways to take an exposure to credit are increasingly turning towards the credit derivatives market. The primary purpose of credit derivatives is to enable the efficient transfer and repackaging of credit risk. In their simplest form, credit derivatives provide a more efficient way to replicate in a derivative format the credit risks that would otherwise exist in a standard cash instrument. More exotic credit derivatives such as synAuthors Dominic O'Kane T. +44 207 260 2628 E. dokane@lehman.com Marco Naldi T. +1 212 526 1728 E. mnaldi@lehman.com Sunita Ganapati T. +1 415 274 5485 E. sganapati@lehman.com Arthur Berd T. +1 212 526 2629 E. arthur.berd@lehman.com

thetic loss tranches and default baskets create new risk-return profiles to appeal to the differing risk appetites of investors based on the tranching of portfolio credit risk. In doing so they create an exposure to default correlation. CDS options allow investors to express a view on credit spread volatility, and hybrid products allow investors to mix credit risk views with interest rate and FX risk. More recently, we have seen a stepped increase in the liquidity of these exotic credit derivative products. This includes the development of very liquid portfolio credit vehicles, the arrival of a two-way correlation market in customised CDO tranches, and the development of a more liquid default swaptions market. To enable this growth, the market has developed new approaches to the pricing and risk-management of these products. As a result, this book is divided into two parts. In the first half, we describe how exotic structured credit products work, their rationale, risks and uses. In the second half, we review the models for pricing and risk managing these various credit derivatives, focusing on implementation and calibration issues.

Claus Pedersen T. +1 212 526 7775 E. cmpeders@lehman.com Lutz Schloegl T. +44 207 260 2113 E. luschloe@lehman.com Roy Mashal T. +1 212 526 7931 E. rmashal@lehman.com

The Lehman Brothers Guide to Exotic Credit Derivatives 1

Contents Foreword Credit Derivatives Products Market overview The credit default swap Basket default swaps Synthetic CDOs Credit options Hybrid products 3 4 8 12 23 28 1 Credit Derivatives Modelling Single credit modelling Modelling default correlation Valuation of correlation products Estimating the dependency structure Modelling credit options Modelling hybrids 31 33 39 43 47 51 53 References 2 The Lehman Brothers Guide to Exotic Credit Derivatives .

hedge funds and asset managers. The survey Figure 1. We believe that this growth in CDS has been driven by hedging demand generated by synthetic CDO positions. The base of credit derivatives users has been broadening steadily over the last few years. Market breakdown by instrument type Credit default swaps 73% Credit linked notes 3% Options and hybrids 1% Portfolio/ correlation products 22% Total return swaps 1% Source: Risk Magazine 2003 Credit Derivatives Survey showed that 40. Banks still remain the largest users with nearly 50% share. they can be used to draw fairly firm conclusions about the recent direction of the market. The latest snapshot of the credit derivatives market was provided in the 2003 Risk Magazine credit derivatives survey. This is mainly because of their substantial use of CDS as hedging tools for their loan books.1% of all reference entities originate in Europe. An interesting statistic from the survey is the relatively equal representation of North American and European credits. Readers are referred to Ganapati et al (2003) for a full discussion of the market impact. The hedging activity driven by the issuance of synthetic CDOs (discussed later) has for the first time satisfied the demand to buy protection coming from bank loan hedgers. According to the survey. and by hedge funds using credit derivatives as a way to exploit capital structure arbitrage opportunities and to go outright short the credit markets. mainly by investing in investment-grade CDO tranches. Single name CDS remain the most used instrument in the credit derivatives world with 73% of market outstanding notional. it has since broadened its base of users to include insurance companies. We show a breakdown of the market by end-users in Figure 2 (overleaf). Insurance companies have also become an important player. This survey polled 12 dealers at the end of 2002. Initially driven by the hedging needs of bank loan managers. This is in stark contrast to the global credit market which has a significantly smaller proportion of European originated bonds relative to North America.306 billion. The Lehman Brothers Guide to Exotic Credit Derivatives 3 . as shown in Figure 1. moving from a small and highly esoteric market to a more mainstream market with standardised products. and their active participation in synthetic securitisations. This supports our observation that the credit default market has become more mainstream. Although the reported numbers cannot be considered ‘hard’. up more than 50% on the previous year. the total market outstanding notional across all credit derivatives products was calculated to be $2. composed of all the major players in the credit derivatives market. compared with 43. As a result. focusing on the liquid standard contracts.8% from North America.Credit Derivatives Products Market overview The credit derivatives market has changed substantially since its early days in the late 1990s.

the total notional for all types of credit derivatives portfolio products was $449. Given their ability to leverage. to cover the loss of the face value of an asset following a credit event. has grown substantially. the insurance share of credit derivatives usage has increased to 14% from 9% the previous year. the short end of the credit curve. especially. Finally. Breakdown by end users Reinsurance SPVs 5% 10% Insurance 14% Banks (synthetic securitisation) 10% Hedge funds 13% Third-party Corporates asset managers 3% 7% Banks (other) 38% Source: Risk Magazine 2003 Credit Derivatives Survey. but it is hoped that a global standard will eventually emerge. in particular the market for those written on CDS. First.4 billion. which in many cases has been disproportionate to their absolute size. The credit options market. The year 2003 has also seen a significant increase in the usage of CDS portfolio products. in portfolio products. Currently. This is not a surprise. A number of issues still remain to be resolved. the credit derivatives market has continued to consolidate and innovate. the market is segregated along regional lines in tackling this issue. they have substantially increased their volume of CDS contracts traded. the issue of the treatment of restructuring events still needs to be resolved. More recently. Their share has kept pace with the growth of the credit derivatives market at about 22% over the last two years.Figure 2. since there is a fundamentally symbiotic relationship between the synthetic CDO and single name CDS markets. There has been a stepped increase in liquidity for correlation products. Greater transparency is also needed around the calibration of recovery rates. for the sake of completeness. we set out a short description before we explore more exotic products. by which we mean synthetic CDOs and default baskets. with daily two-way markets for synthetic tranches now being quoted. caused by dealers originating synthet- ic tranches either by issuing the full capital structure or hedging bespoke tranches. A credit default swap (CDS) is used to transfer the credit risk of a reference entity (corporate or sovereign) from one party to another. They have also been involved in many of the ‘fallen angel’ credits where they have been significant buyers of protection. In a standard CDS contract one party purchases credit protection from the other party. The market needs to build greater liquidity at the long end and. A credit event is a legally defined event that typically includes 4 The Lehman Brothers Guide to Exotic Credit Derivatives . there is a need for the generation of a proper term structure for credit default swaps. the growth in the usage of credit derivatives by hedge funds has had a marked impact on the overall credit derivatives market itself. The credit default swap The credit default swap is the basic building block for most ‘exotic’ credit derivatives and hence. The ISDA 2003 Credit Derivative Definitions were another milestone on the road towards CDS standardisation. where their share has increased to 13% over the year. Hedge funds have been regular users of CDS especially around the convertible arbitrage strategy. Finally. Since this survey was published.

The protection buyer therefore makes quarterly payments approximately (we ignore calendars and day count conventions) equal to $10m × 0. The actual payment amounts on the premium leg are determined by the CDS spread adjusted for the frequency using a basis convention. Typically they include restrictions on the maturity and the requirement that they be pari passu – most CDS are linked to senior unsecured debt. a cash payment is made by the protection seller to the protection buyer equal to par minus the recovery rate of the reference asset. The recovery rate is calculated by referencing dealer quotes or observable market prices over some period after default has occurred. as shown in Figure 3. The protection buyer is therefore long a cheapest to deliver option. until a credit event or maturity. which they are most likely to do if the credit event is a restructuring. For this protection. selling credit protection is economically equivalent to going long the credit risk. as discussed later. Equally. to the protection seller.000. If a credit event does occur before the maturity date of the contract. Suppose a protection buyer purchases five-year protection on a company at a CDS spread of 300bp. It requires the protection buyer to deliver the notional amount of deliverable obligations of the reference entity to the protection seller in return for the notional amount paid in cash. the payments are as follows: Figure 3. This protection lasts until some specified maturity date. In general there are several deliverable obligations from which the protection buyer can choose which satisfy a number of characteristics. the choice being made at the initiation of the contract. In cash settlement.03 × 0. Mechanics of a CDS Protection buyer Default swap spread (premium leg) Protection seller Contingent payment of loss on par following a credit event (protection leg) The Lehman Brothers Guide to Exotic Credit Derivatives 5 . Cash settlement – This is the alternative to physical settlement.bankruptcy. They are: Physical settlement – This is the most widely used settlement procedure. usually Actual 360. and compensates the protection buyer for the loss. This payment equals the difference between par and the price of the assets of the reference entity on the face value of the protection. and is used less frequently in standard CDS but overwhelmingly in tranched CDOs. failure to pay and restructuring. We call this leg of the CDS the protection leg.25 = $75. the protection buyer can take advantage of this situation by buying and delivering the cheapest asset. If the deliverable obligations trade with different prices following a credit event. there is a payment by the protection seller to the protection buyer. Buying credit protection is economically equivalent to shorting the credit risk. the protection buyer makes quarterly payments. After a short period the reference entity suffers a credit event. This is known as the premium leg. Assuming that the cheapest deliverable asset of the reference entity has a recovery price of $45 per $100 of face value. The face value of the protection is $10m. whichever occurs first. There are two ways to settle the payment of the protection leg.

the most liquid CDS is the five-year contract. For severely distressed reference entities.■ The protection seller compensates the protection buyer for the loss on the face value of the asset received by the protection buyer and this is equal to $5. because CDS implicitly lock in Libor funding to maturity. although deviation from the standard may incur a liquidity cost. The definitions used by the market for credit events and other contractual details have been set out in the ISDA 1999 document and recently amended and enhanced by the ISDA 2003 document. sevenyear and 10-year. ■ CDS can be used to take a spread view on a credit. This can be done for long periods without assuming any repo risk. This is also an advantage for those who have high funding costs. if the credit event occurs after a month then the protection buyer pays approximately $10m × 300bp × 1/12 = $25. For a start. a wider range of credits trade in the CDS market than in cash. Note that this is the standard for corporate reference entity linked CDS. Instead. ■ The CDS has revolutionised the credit markets by making it easy to short credit. 6 The Lehman Brothers Guide to Exotic Credit Derivatives .5m. For example. the CDS market does not have a universal standard contract. The advantage of this standardisation of a unique set of definitions is that it reduces legal risk. ■ CDS are unfunded so leverage is possible. We list some of the main applications of CDS below. the CDS contract trades in an up-front format where the protection buyer makes a cash payment at trade initiation which purchases protection to some specified maturity – there are no subsequent payments unless there is a credit event in which the protection leg is settled as in a standard CDS. ■ The protection buyer pays the accrued premium from the previous premium payment date to the time of the credit event. For a full description of up-front CDS see O’Kane and Sen (2003).000 of premium accrued. ■ CDS are customisable. The fact that a physical asset does not need to be sourced means that it is generally easier to transact in large round sizes with CDS. In terms of maturity. as with a bond. European and an Asian market standard. Liquidity in the CDS market differs from the cash credit market in a number of ways. followed by the three-year. speeds up the confirmation process and so enhances liquidity. This is known as trading the default swap basis. This is the consequence of a desire to enhance the posi- Uses of a CDS The CDS can do almost everything that cash can do and more. Despite this standardisation of definitions. differentiated by the way they treat a restructuring credit event. This is very useful for those wishing to hedge current credit exposures or those wishing to take a bearish credit view. Evolution of CDS documentation The CDS is a contract traded within the legal framework of the International Swaps and Derivatives Association (ISDA) master agreement. there is a US. ■ Dislocations between cash and CDS present new relative value opportunities.

and we ignore counterparty risk. Traded CDS portfolio products CDS portfolio products are products that enable the investor to go long or short the credit risk associated with a portfolio of CDS in one transaction. Determining the CDS spread is not the same as valuing an existing CDS contract. Funded versus unfunded Credit derivatives. paid periodically on the protected notional until maturity or a credit event. which may be floating rate. we have seen the emergence of a number of very liquid portfolio products. A large number of investors now exploit the basis as a relative value play. we ignore accrued coupons on default. A full discussion of the drivers behind the CDS basis is provided in O’Kane and McAdie (2001). At maturity. For example. A full discussion and analysis of these different standards can be found in O’Kane. we assume a common market-wide funding level of Libor. For that we need a model and a discussion of the valuation of CDS is provided on page 32. Pedersen and Turnbull (2003). Libor plus a spread. be proxied by either (i) a par floater bond spread (the spread to Libor at which the reference entity can issue a floating rate note of the same maturity at a price of par) or (ii) the asset swap spread of a bond of the same maturity provided it trades close to par. Losses require payments to be made by the protection seller to the protection buyer. we ignore the delivery option in the CDS. the protection buyer covering his loss and the investor receiving par minus the loss. cash market spreads usually provide the starting point for where CDS spreads should trade. ie. Determining the CDS spread The premium payments in a CDS are defined in terms of a CDS spread. The protection buyer is exposed to the default risk of the collateral rather than the counterparty. Subsequent payments are simply payments of spread and there is no principal payment at maturity. can be traded in a number of formats. and this has counterparty risk implications. to a first approximation. Despite these assumptions. including CDS. This format is known as ‘funded’ because the investor has to fund an initial payment. defined as: Basis = CDS Spread – Cash Libor Spread. The other format is to trade the risk in the form of a credit linked note.tion of protection sellers by limiting the value of the protection buyer’s delivery option following a restructuring credit event. It is possible to show that the CDS spread can. and this is the standard for CDS. Demonstrating these relationships relies on several assumptions that break down in practice. liquid vehicle for assuming or hedg- The Lehman Brothers Guide to Exotic Credit Derivatives 7 . This format is also termed ‘unfunded’ format because the investor makes no upfront payment. The difference between where CDS spreads and cash LIBOR spreads trade is known as the Default Swap Basis. whose aim is to offer investors a diverse. This par is used by the protection buyer to purchase high quality collateral. or may be fixed at a rate above the same maturity swap rate. Any default before maturity results in the collateral being sold. The most commonly used is known as swap format. In return the protection seller receives a coupon. In recent months. typically par. if no default has occurred the collateral matures and the investor is returned par.

To see this. The portfolio may be as small as five credits or as large as 200 or more credits. These consist of three sub-indices. Basket default swaps Correlation products are based on redistributing the credit risk of a portfolio of singlename credits across a number of different securities.ing exposure to different credit markets. Daily pricing of all 440 names is available on our LehmanLive website. As with a CDS. FTD baskets leverage the credit risk by increasing the probability of a loss by conditioning the payoff on the first default among several credits. A basket default swap is similar to a CDS. consider that the fair-value spread paid by a credit risky asset is determined by the probability of a default. times the size of the loss given default. In return for assuming the nth-to-default risk. The simplest correlation product is the basket default swap. This exposes the investor to the tendency of assets in the portfolio to default together. Lehman Brothers has introduced a family of global investment grade CDS indices which are discussed in Munves (2003). where n=2. ie. This is because the seller of FTD protection is leveraging their credit risk. As such they are lower risk second-loss exposure products which will pay a lower spread than an FTD basket. 8 The Lehman Brothers Guide to Exotic Credit Derivatives . with some securities taking the first losses and others taking later losses. whichever is sooner. The advantage of an FTD basket is that it enables an investor to earn a higher yield than any of the credits in the basket. one example being the TRAC-XSM vehicle. the protection seller receives a spread paid on the notional of the position as a series of regular cash flows until maturity or the nth credit event. a US 250 name index. a European 150 name index and a Japanese 40 name index. The size of the potential loss does not increase relative to buying any of the assets in the basket. The most that the investor can lose is par minus the recovery value of the FTD asset on the face value of the basket. All names are corporates and the maturity of the index is maintained close to five years. the contingent payment typically involves physical delivery of the defaulted asset in return for a payment of the par amount in cash. The advantage is that the basket spread paid can be a multiple of the spread paid by the individual assets in the basket. The FTD basket pays a spread of 120bp. More risk-averse investors can use default baskets to construct lower risk assets: second-to-default (STD) baskets. This is shown in Figure 4 where we have a basket of five investment grade credits paying an average spread of about 28bp. Typical baskets contain five to 10 reference entities. As a consequence. default correlation. These have added liquidity to the CDS market and also created a standard which can be used to develop portfolio credit derivatives such as options on TRAC-X. trigger a credit event after two or more assets have defaulted. and it is the first credit in a basket of reference credits whose default triggers a payment to the protection buyer. n=1. The move of the CDS market from banks towards traditional credit investors has greatly increased the need for a performance benchmark linked directly to the CDS market. the difference being that the trigger is the nth credit event in a specified basket of reference entities. In the particular case TRAC-X is a service mark of JPMorgan and Morgan Stanley of a first-to-default (FTD) basket. The redistribution mechanism is based on the idea of assigning losses on the credit portfolio to the different securities in a specified priority.

However. the greater the likelihood of a credit event. if there is a default the investor will certainly prefer a higher realised recovery rate. It is natural to assume that assets issued by companies within the same country and industrial sector should have a higher default correlation than those within different industrial sectors. ■ Number of credits: The greater the number of credits in the basket. Confusion is usually caused by the term ‘default correlation’. the higher the spread. ■ Credit quality: The lower the credit quality of the credits in the basket. ■ Default correlation: Increasing default correlation increases the likelihood of assets to default or survive together. We show the five year CDS spreads of the individual credits 120bp paid on $10m until FTD or five-year maturity. The effect of default correlation is subtle and significant in terms of pricing. The fact is The Lehman Brothers Guide to Exotic Credit Derivatives 9 . they share the same market.Figure 4. There are a number of ways to explain how default correlation affects the pricing of default baskets. all companies are affected by the performance of the world economy. Baskets and default correlation Baskets are essentially a default correlation product. whichever is sooner Lehman Brothers Basket investor ■ Maturity: The effect of maturity depends on the shape of the individual credit curves. At a global level. the same interest rates and are exposed to the same costs. in terms of spread and rating. This has only a small effect on pricing since a higher expected recovery rate is offset by a higher implied default probability for a given spread. We believe that these systemic sector risks far outweigh idiosyncratic effects so we expect that default correlation is usually positive. Five-year first to default (FTD) basket on five credits. Valuation therefore requires a pricing model. and so the higher the spread. Such a trade cannot be replicated using existing instruments. ■ Recovery rate: This is the expected recovery rate of the nth-to-default asset following its credit event. We now discuss this is more detail. This means that the basket spread depends on the tendency of the reference assets in the basket to default together. The model inputs in order to determine the nth-to-default basket spread are: ■ Value of n: An FTD (n=1) is riskier than an STD (n=2) and so commands a higher spread. After all. de Saint Gobain 30bp Electricidade de Portugal 27bp Hewlett Packard 29bp Teliasonera 30bp The basket spread One way to view an FTD basket is as a trade in which the investor sells protection on all of the credits in the basket with the condition that all the other CDS cancel at no cost following a credit event. Contingent payment of par minus recovery on FTD on $10m face value Reference portfolio Coca Cola 30bp C.

the asset with the widest spread will always default too. The process for translating these loss distributions into a fair value spread requires a model of the type described on page 39. We should not forget that in addition to the that if two assets are correlated. As correlation goes up from 0–20%. See page 33 for a discussion of how to model the loss distribution. ■ Independence: Consider a five-credit basket where all of the underlying credits have flat credit curves. 20% and 50% correlation 80 70 Probability (%) ρ = 0% ρ = 20% ρ = 50% 60 50 40 30 20 10 0 0 1 2 3 4 Number of defaults 5 default correlation is at its maximum.Figure 5. the CDS hedge covers the loss on the basket and all of the other CDS hedges can be unwound at no cost. four and five defaults increases. it will be as risky as the riskiest asset and the FTD basket spread should be the widest spread of the credits in the basket. We consider a basket of five credits with spreads of 100bp and an assumed recovery rate for all of 40%. This implies that the basket spread for independent assets should be equal to the sum of the spreads of the names in the basket. This makes the STD spread increase. We have plotted the loss distribution for correlations of 0%. We see that the probability of no defaults increases with increasing correlation – the probability of credits surviving together increases – and the FTD spread should fall. and 50% in Figure 5. If the credits are all independent and never become correlated during the life of the trade. the risk of one default is the same as the risk of the widest spread asset defaulting. they will not only tend to default together. ■ Maximum correlation: Consider the same FTD basket but this time where the 10 The Lehman Brothers Guide to Exotic Credit Derivatives . In practice. three. The spread for an FTD basket depends on the probability of one or more defaults which equals one minus the probability of no defaults. the probability of two. Loss distribution for a five-credit basket with 0%. If a credit event occurs. since they should on average have rolled down their flat credit curves. The risk of an STD basket depends on the probability of two or more defaults. 20%. Essentially we have to find the basket spread for which the present value of the protection payments equals the present value of the premium payments. they will also tend to survive together. Because an FTD is triggered by only one credit event. There are two correlation limits in which a FTD basket can be priced without resorting to a model – independence and maximum correlation. The best way to understand the behaviour of default baskets between these two correlation limits is to study the loss distribution for the basket portfolio. the natural hedge is for the basket investor to buy CDS protection on each of the individual names to the full notional. this means that when any asset defaults. As a result.

Investors can use default baskets to leverage their credit exposure and so earn a higher yield without increasing their notional at risk. it is not guaranteed to be a full hedge against a sudden default. 140 120 100 80 60 40 20 0 0 Basket spread (bp) FTD STD 10 20 30 40 50 60 70 80 90 100 Correlation (%) The Lehman Brothers Guide to Exotic Credit Derivatives 11 . At high correlation. ■ Credit investors can use default baskets to hedge a blow-up in a portfolio of credits more cheaply than buying protection on the individual credits. Applications ■ Baskets have a range of applications. which is the sum of the spreads. If implied correlation is too high they should sell STD protection. the FTD spread is close to 146bp. number of credits. If the investor’s view is that the implied correlation is too low then the investor should sell FTD protection.protection leg. ■ Buy and hold investors can enjoy the leveraging of the spread premium. ■ Default baskets can be used to express a view on default correlation. The STD spread is lowest at zero correlation since the probability of two assets defaulting is low if the assets are independent. the premium leg of the default basket also has correlation sensitivity because it is only paid for as long as the nth default does not occur. Using a model we have calculated the correlation sensitivity of the FTD and STD spread for the five-credit basket shown in Figure 6. where F is the basket face value and D is the delta in terms of percentage of face value. maturity. known as the delta. At low correlation. Spread risk is hedged by selling protection dynamically in the CDS market on all of the credits in the default basket. For instance. the basket has the risk of the widest spread asset and so is at 30bp. a dealer hedging an FTD basket where a credit defaults with a recovery rate of R would receive a payment of (1-R)F from the protection seller. requires a pricing model to calculate the sensitivity of the basket value to changes in the spread curve of the underlying credit. At maximum correlation the STD spread tends towards the spread of the second widest asset in the basket which is also 30bp. This is discussed in more detail later. The net payment to the protection buyer is therefore (1-D)(1-R)F . Figure 6. ■ The reference entities in the basket are all typically investment grade and so are familiar to most credit analysts. Although this delta hedging should immunise the dealer’s portfolio against small changes in spreads. ■ The basket can be customised to the investors’ exact view regarding size. FTD or STD. Correlation dependence of spread for FTD and STD basket Hedging default baskets The issuers of default baskets need to hedge their risks. credit selection. Determining how much to sell. and will pay D(1-R)F on the hedged protection.

26m 6. these different types of synthetic CDO have a total market size estimated by the Risk 2003 survey to be close to $500 billion. Synthetic CDOs Synthetic collateralised debt obligations (Synthetic CDOs) were conceived in 1997 as a flexible and low-cost mechanism for transferring credit risk off bank balance sheets. 12 The Lehman Brothers Guide to Exotic Credit Derivatives . More recently. second. If the spread tightens. the fusion of credit derivatives modelling techniques and derivatives trading have led to the creation of a new type of synthetic CDO. Different rating agencies have developed their own model-based approaches for the rating of default baskets. Note that the deltas are all very similar. Overall. However this is difficult as it is usually difficult to find protection buyers who select the exact same basket as an investor. The primary motivation was the banks’ reduction of regulatory capital. which we call a customised CDO. This is clear in the previous example where the income from the hedges is 211bp. Figure 7. Differences are mainly due to our different correlation assumptions. The expected spread widening on default on the other credits in the basket due to their positive correlation with the defaulted asset will result in a loss when they are unwound. This effect helps to offset the negative carry associated with hedged FTD baskets. Figure 7 shows an example basket with the delta and spread for each of the five credits. Default basket deltas for a €10m notional five-year FTD basket on five credits. As a result. This means that as the spread of an asset widens. What is also of interest is that the dealer-hedging of these products in the CDS market has generated a substantial demand to sell protection. balancing the traditional protection-buying demand coming from bank loan book managers. which can be tailored to the exact risk appetites of different classes of investors. lower than the 246bp paid to the FTD basket investor. This is based on the observation that a dealer who is long first.57m Hedgers of long protection FTD baskets are also long gamma. the delta will increase and so the hedger will be selling protection at a wider spread. The greater unwind losses for baskets with higher correlations will be factored into the basket spread. then the delta will fall and the hedger will be buying back hedges at a tighter level. The FTD spread is 246bp. Reference entity Walt Disney Rolls Royce Sun Microsystems Eastman Chemical France Telecom CDS Spread 62bp 60bp 60bp 60bp 64bp Delta 6. One way for a default basket dealer to reduce his correlation risk is by selling protection on the same or similar default baskets. third up to Mth order protection on an M-credit basket has almost no correlation risk. The alternative hedging approach is for the dealer to buy protection using default baskets on other orders of protection.55m 6. So spread volatility can be beneficial. however. since this position is almost economically equivalent to buying full face value protection using CDS on all M credits in the basket.There will also probably be a loss on the other CDS hedges.87m 7.16m 7. We discuss these on page 39. the synthetic CDO has become an investor-driven product. This reflects the fact that all of the assets have a similar spread.

This payment is offset by a successive reduction in the equity tranche. Consequently the senior tranche is typically paid a very low spread. Next is the mezzanine tranche which is lower risk and so is paid a lower spread. each with a €10m notional.500bp Contingent payment The performance of a synthetic CDO is linked to the incidence of default in a portfolio of CDS. mezzanine debt and senior debt. which assumes the first €50m of losses. Finally we have the senior tranche which is protected by €150m of subordination. This is shown in Figure 9 (overleaf). 90% or more. To see this. The remainder of the risk. and are the largest participants in this part of the capital structure – often referred to as super-senior AAAs or super-senior swaps. The CDO redistributes this risk by allowing different tranches to take these default losses in a specific order. This risk is redistributed into three tranches. Reference pool 100 investment grade names in CDS format €10m x 100 assets = 1bn total notional Full capital structure synthetics In the typical synthetic CDO structured using securitisation technology. creating cash liabilities. and (iii) the senior tranche with a notional of €850m takes all remaining losses. The equity tranche has the greatest risk and is paid the widest spread. the trust liquidates investments in the trust and makes payments to the originating entity to cover default losses. Reinsurers. (i) an equity tranche. it is possible to customise the risk profile of a tranche to the investor’s specific profile. The SPV in turn issues notes in the capital markets to cash collateralise the portfolio default swap with the originating entity. Lehman Brothers 200bp 1. note that it would require more than 25 of the assets in the 100 credit portfolio to default with a recovery rate of 40% before the senior tranche would take a principal loss. It is based on a reference pool of 100 CDS. consider the synthetic CDO shown in Figure 8. To get a sense of the risk of the senior tranche. the sponsoring institution. then the mezzanine and finally the super-seniors are called to make up losses. See Ganapati et al (2001) for more details. The SPV typically provides credit protection for 10% or less of the losses on the reference portfolio. It is typically unrated. is generally distributed via a senior swap to a highly rated counterparty in an unfunded format. A standard synthetic CDO 5bp Senior tranche €850m Mezzanine tranche €100m Equity tranche €50m ordination below the tranche) and width. who typically have AAA/AA ratings. which take the next €100m of losses. The notes issued can include a non-rated ‘equity’ piece. the investor simply The Lehman Brothers Guide to Exotic Credit Derivatives 13 . (ii) a mezzanine tranche.Figure 8. typically a bank. have traditionally had a healthy appetite for this type of senior risk. liquid assets. The advantage of CDOs is that by changing the details of the tranche in terms of its attachment point (this is the amount of sub- Mechanics of a synthetic CDO When nothing defaults in the reference portfolio of the CDO. If an obligor in the reference pool defaults. enters into a portfolio default swap with a Special Purpose Vehicle (SPV). The initial proceeds from the sale of the equity and notes are invested in highly rated.

However. the more defaults are required to cause tranche principal losses and the lower the tranche spread. ■ Tranche width: The wider the tranche for a fixed attachment point.45m. The tranche notional falls from €50m to €43m and the equity coupon. The higher the attachment point. consider what happens if one of the reference entities in the reference portfolio undergoes the first credit event with a 30% recovery. This process repeats following each credit event. causing a €7m loss. This has no effect on the other tranches. The mechanics of a standard synthetic CDO are therefore very simple. The senior tranche notional is decreased by €3m to €847m. ■ Attachment point: This is the amount of subordination below the tranche. the €3m recovered on the defaulted asset is either reinvested in the portfolio or used to reduce the exposure of the senior-most tranche (similar to early amortisation of senior tranches in cash flow CDOs). If the losses exceed €50m then the mezzanine investor must bear the subsequent losses with the corresponding reduction in the mezzanine notional. The CDO tranche spread The synthetic CDO spread depends on a number of factors.5m to 15% times €43m = €6. If the losses exceed €150m. If traded in a funded format.Figure 9. set at 1500bp. These coupon payments therefore fall from €7. The full capital structure synthetic CDO Highly rated counterparty Super senior swap premium Sponsoring bank $900m super senior credit protection Credit protection Subordinated swap premium 10% first loss subordinated credit protection CDS spread income Special purpose vehicle (SPV) Issued notes Senior notes AAA Mezzanine notes BBB/A Proceeds Equity notes (unrated) Reference portfolio $1bn notional receives the Libor spread until maturity and nothing else changes. which is immediately paid to the originator. the more losses to which the tranche is exposed. then it is the senior investor who takes the principal losses. We list the main ones and describe their effects on the tranche spread. Using the synthetic CDO described earlier and shown in Figure 8. the incremental risk ascending 14 The Lehman Brothers Guide to Exotic Credit Derivatives . The equity investor takes the first loss of €7m. so that the sum of protected notional equals the sum of the collateral notionals which is now €990m. especially compared with traditional cash flow CDO waterfalls. is now paid on this smaller notional. This also makes them more easily modelled and priced.

the capital structure is usually declining and so the spread falls. ■ Portfolio credit quality: The lower the quality of the asset portfolio, measured by spread or rating, the greater the risk of all tranches due to the higher default probability and the higher the spread. ■ Portfolio recovery rates: The expected recovery rate assumptions have only a secondary effect on tranche pricing. This is because higher recovery rates imply higher default probabilities if we keep the spread fixed. These effects offset each other to first order. ■ Swap maturity: This depends on the shapes of the credit curves. For upward sloping credit curves, the tranche curve will generally be upward sloping and so the longer the maturity, the higher the tranche spread. ■ Default correlation: If default correlation is high, assets tend to default together and this makes senior tranches more risky. Assets also tend to survive together making the equity safer. To understand this more fully we need to better understand the portfolio loss distribution.

the portfolio loss distribution. We can expect to observe one of the two shapes shown in Figure 10. They are (i) a skewed bell curve; (ii) a monotonically decreasing curve. The skewed bell curve applies to the case when the correlation is at or close to zero. In this limit the distribution is binomial and the peak is at a loss only slightly less than the expected loss. As correlation increases, the peak of the distribution falls and the high quantiles increase: the curves become monotonically decreasing. We see that the probability of larger losses increases and, at the same time, the probability of smaller losses also increases, thereby preserving the expected loss which is correlation independent (for further discussion see Mashal, Naldi and Pedersen (2003)). For very high levels of asset correlations (hardly ever observed in practice), the distribution becomes U-shaped. At maximum default correlation all the probability mass is located at the two ends of the distribution. The portfolio either all survives or it all defaults. It resembles the loss distribution of a single asset. Figure 10. Portfolio loss distribution for a large portfolio at 0%, 20% and 95% correlation
40 35 Probability (%) 30 25 20 15 10 5 0 0 5 10 15 Loss (%) 47 ρ = 95% ρ=0 ρ = 20%

The portfolio loss distribution
No matter what approach we use to generate it, the loss distribution of the reference portfolio is crucial for understanding the risk and value of correlation products. The portfolio loss is clearly not symmetrically distributed: it is therefore informative to look at the entire loss distribution, rather than summarising it in terms of expected value and standard deviation. We can use models of the type discussed on page 33 to calculate

The Lehman Brothers Guide to Exotic Credit Derivatives 15

Figure 11. Equity tranche loss distribution for correlations of 20% and 50%
30 25 Probability (%) 20 15 10 5 0 0 Equity tranche loss (%) 100 10 20 30 40 50 60 70 80 90 ρ = 20% ρ = 50%

Figure 12. Mezzanine tranche loss distribution for correlation of 20% and 50%. We have eliminated the zero loss peak, which is about 86% in both cases
4.0 3.5 Probability (%) 3.0 2.5 2.0 1.5 1.0 0.5 0.0 10 20 30 40 50 60 70 80 90 Mezzanine tranche loss (%) 100 0 ρ = 20% ρ = 50%

lation, we see that most of the portfolio loss distribution is inside the equity tranche, with about 14% beyond, as represented by the peak at 100% loss. As correlation goes to 50% the probability of small losses increases while the probability of 100% losses increases only marginally. Clearly equity investors benefit from increasing correlation. The mezzanine tranche becomes more risky at 50% correlation. As we see in Figure 12, the 100% loss probability jumps from 0.50% to 3.5%. In most cases mezzanine investors benefit from falling correlation – they are short correlation. However, the correlation directionality of a mezzanine tranche depends upon the collateral and the tranche. In certain cases a mezzanine tranche with a very low attachment point may be a long correlation position. Senior investors also see the risk of their tranche increase with correlation as more joint defaults push out the loss tail. This is clear in Figure 13. Senior investors are short correlation. In Figure 14 we plot the dependence of the value of different CDO tranches on correlation. As expected, we clearly see that: ■ Senior investors are short correlation. If correlation increases, senior tranches fall in value. ■ Mezzanine investors are typically short correlation, although this very much depends upon the details of the tranche and the collateral. ■ Equity investors are long correlation. When correlations go up, equity tranches go up in value. In the process of rating CDO tranches, rating agencies need to consider all of these

How then does the shape of the portfolio loss distribution affect the pricing of tranches? To see this we must study the tranche loss distribution.

The tranche loss distribution
We have plotted in Figures 11–13 the loss distributions for a CDO with a 5% equity, 10% mezzanine and 85% senior tranche for correlation values of 20% and 50%. At 20% corre-

16 The Lehman Brothers Guide to Exotic Credit Derivatives

risk parameters and so have adopted model based approaches. These are discussed on page 43.

Customised synthetic CDO tranches
Customisation of synthetic tranches has become possible with the fusion of derivatives technology and credit derivatives. Unlike full capital structure synthetics, which issue the equity, mezzanine and senior parts of the capital structure, customised synthetics may issue only one tranche. There are a number of other names for customised CDO tranches, including bespoke tranches, and single tranche CDOs. The advantage of customised tranches is that they can be designed to match exactly the risk appetite and credit expertise of the investor. The investor can choose the credits in the collateral, the trade maturity, the attachment point, the tranche width, the rating, the rating agency and the format (funded or unfunded). Execution of the trade can take days rather than the months that full capital structure CDOs require. The basic paradigm has already been discussed in the context of default baskets. It is to use CDS to dynamically delta-hedge the first order risks of a synthetic tranche and to use a trading book approach to hedge the higher order risks. This is shown in Figure 15 (overleaf). For example, consider an investor who buys a customised mezzanine tranche from Lehman Brothers. We will then hedge it by selling protection in an amount equal to the delta of each credit in the portfolio via the CDS market. The delta is the amount of protection to be sold in order to immunise the portfolio against small changes in the CDS spread curve for that credit. Each credit in the portfolio will have its own delta.

Figure 13. Senior tranche loss distribution for correlations of 20% and 50%. We have eliminated the zero loss peak, which is greater than 96% in both cases
0.8 0.7 Probability (%) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 10 20 30 Senior tranche loss (%) 40 0 ρ = 20% ρ = 50%

Figure 14. Correlation dependence of CDO tranches
40
Tranche MTM (€m) €

30 20 10 0 –10 –20 –30 0 10

Equity Mezzanine Senior

20

30

40

50

60

70

80

90

Correlation (%)

Understanding delta for CDOs
For a specific credit in a CDO portfolio, the delta is defined as the notional of CDS for that credit which has the same mark-tomarket change as the tranche for a small movement in the credit’s CDS spread curve. Although the definition may be straightforward, the behaviour of the delta is less so. One way to start thinking about delta is to

The Lehman Brothers Guide to Exotic Credit Derivatives 17

100

then it is the least risky asset. It has three tranches: a 5% equity. a 10% mezzanine and an 85% senior tranche. each $10m notional. as well as the features of the asset whose delta we are examining. If it is at the back of the queue then its equity delta will be low. As it is most likely to default after all the other asset. it becomes more likely to default before the others and so impacts the equity or mezzanine tranche. If the single asset spread is less than the portfolio average of 150bp. and also because the CDS will have a higher spread sensitivity and so require a smaller notional. For example the delta for a senior tranche to a credit whose CDS spread has widened will fall due to the fact that it is more likely to default early and hit the equity tranche. The sensitivity of the delta to changing the spread of the asset whose delta we are calculating is shown in Figure 16. it will be most likely to hit the senior tranche. The asset spreads are all 150bp and the correlation between all the assets is the same. As a result the senior tranche delta will rise. This framework helps us understand the directionality of delta. If the asset whose delta you are calculating is at the front of this queue. This ordering will depend mostly on the spread of the asset relative to the other credits in the portfolio and its correlation relative to the other assets in the portfolio. The actual magnitude of delta is more difficult to quantify because it depends on the tranche notional and the contractual tranche spread. it would be expected to be the last to default and so most likely to impact the senior-most tranche. To show this we take an example CDO with 100 credits. Delta hedging a synthetic CDO ∆ of CDS on Name 1 ∆ of CDS on Name 2 ∆ of CDS on Name 3 Lehman Brothers Investor Spread ∆ Contingent payment ∆ of CDS on Name 99 ∆ of CDS on Name 100 Reference pool 100 investment grade names in CDS format $10m x 100 assets = $1bn Bespoke tranche imagine a queue of all of the credits sorted in the order in which they should default. As a result. it will be most likely to cause losses to the equity tranche and so will have a high delta for the equity tranche. As the spread of the asset increases above 150bp. The senior delta drops and the equity delta increases. In Figure 17 we plot the delta of the asset versus its correlation with all of the other 18 The Lehman Brothers Guide to Exotic Credit Derivatives .Figure 15.

There are many further dependencies to be explored and we intend to describe these in a forthcoming paper. This implies that equity investors should focus less on the overall properties of the collateral.0 3. Building intuition about the delta is not trivial. The portfolio effect of the CDO is only expressed through the fact that it may take several defaults to completely reduce the equity notional.0 2. and so senior and mezzanine tranches are more exposed.0 0. Dependency of tranche delta on the spread of the asset 8 7 Tranche delta 6 5 4 3 2 1 0 Equity Mezzanine Senior Asset spread (bp) Idiosyncratic versus systemic risk In terms of how they are exposed to credit.0 6. recovery rate sensitivity. There is also a time effect. These all have a correlation of 20% with each other. If the asset is highly correlated with the other assets it is more likely to default or survive with the other assets.0 4. The goal is to flatten the risk of the correlation book with respect to these higher order risks either by doing the offsetting trade or by placing different parts of the capital structure with other buyers of customised tranches.0 0 10 20 30 40 50 Correlation with rest of portfolio (%) Equity Mezzanine Senior Higher order risks If properly hedged. As a result.0 7. These include correlation sensitivity.0 8. dealers are motivated to do trades that reduce these higher order risks. This causes the equity tranche delta to rise through time while the mezzanine and senior tranche deltas fall to zero. Equity tranches are more exposed to idiosyncratic risk – they incur a loss as soon as one asset defaults. Through time.assets in the portfolio. For low correlations. However. this is not a completely risk-free position for the dealer since there are a number of other risk dimensions that have not been immunised. if it defaults it will tend to do so by itself while the rest of the portfolio tends to default together. and more on trying to choose assets which they believe will Figure 17. the equity tranche is most exposed. Figure 16.0 9. For this reason.000 100 200 300 400 500 600 700 800 900 0 . As a result. senior and mezzanine tranches become safer relative to equity tranches since less time remains during which the subordination can be reduced resulting in principal losses. The Lehman Brothers Guide to Exotic Credit Derivatives 19 Tranche delta 1.0 1. the dealer should be insensitive to small spread movements. time decay and spread gamma. it is more likely to default en masse. Dependency of tranche delta on the asset’s correlation with the rest of the portfolio 10. There is also a risk to a sudden default which we call the value-on-default risk (VOD).0 5. there is a fundamental difference between equity and senior tranches.

0 Tranche size   In Figure 18 we show the cash flows 20 The Lehman Brothers Guide to Exotic Credit Derivatives . This is designed to guarantee to return the investor’s initial investment of par. More details are provided in Ganapati and Ha (2002). the senior investor has a significant cushion of subordination to insulate them from principal losses until maybe 20 or more of the assets in the collateral have defaulted. If structured to accumulate to maturity. or at least be able to hedge the credits they do not like. Excess spread (the difference between premium received from the CDS portfolio and the tranche liabilities) is paid into a reserve account. However the investor is only exposed to this credit risk for a first period. It was only in 1999 that features that diverted cash flows from equity to debt holders in case of certain covenant failures began entering the landscape. If structured to build to a predetermined level. and others that use excess spread to provide upside participation to mezzanine debt holders. over-collateralisation trigger concepts were adopted from cash flow CDOs. typically five years. On the other hand. Evolution of structures Initially full capital structure synthetic CDOs had almost none of the structural features typically found in other securitised asset classes and cash flow CDOs. and the coupon paid for the remaining period is frozen at the end of year five. The most common example of structural ways to build additional subordination is the reserve account funding feature. will usually receive a fixed coupon throughout the life of the transaction and any upside or remainder in the reserve account at maturity. Other structures incorporated features to share some of the excess spread with mezzanine holders or to provide a step up coupon to mezzanines if losses exceeded a certain level or if the tranche was downgraded. the senior investor is truly taking a portfolio view and so should be more concerned about the average properties of the collateral than the quality of any specific asset. As a consequence. able to pick the right credits for the portfolio. Finally. As a result we would expect equity tranche buyers to be skilled credit investors. these features fit into two categories – ones that build extra subordination using excess spread. the equity tranche Principal protected structures Investors who prefer to hold highly rated assets can do so by purchasing CDO tranches within a principal protected structure. This may continue throughout the life of the deal or until the balance reaches a predetermined amount. This risk is embedded within the coupons of the note such that each default causes a reduction in the coupon size. This is typically a 10-year note which pays a fixed coupon to the investor linked to the risk of a CDO equity tranche. One particular variation on this theme is the Lehman Brothers High Interest Principal Protection with Extendible Redemption (HIPER). The intention was to provide some defensive mechanism for mezzanine holders fearing that the credit cycle would affect tranche performance. The senior tranche is really a deleveraged macro credit trade.not default. the equity tranche will usually receive excess interest only after the reserve account is fully funded. Broadly. The coupon is typically of the form:  Portfolio loss  Coupon = 8% × max 1 − .

offer a way for investors to take advantage of this spread premium. as well as other sources of risk. The advantage of this approach is that it frees the manager to focus on the credits without having to worry about the cost of substitution. such as basket default swaps and synthetic CDO tranches. Initially this was based on the asset manager being told the cost of substituting an asset using some black-box approach. When an investor The Lehman Brothers Guide to Exotic Credit Derivatives 21 . Principal losses are incurred if the sum of the principal losses on the underlying portfolio of synthetic tranches exceeds the attachment point of the super-tranche. The other advantages of such a structure for the asset manager are fees earned and an increase in assets under management. The CDO of CDOs A recent extension of the CDO paradigm has been the CDO of CDOs. the reference assets in the portfolio do not change. Looking forward. spread. ie. The HIPER structure 100 guaranteed Credit events Managed synthetics The standard synthetic has been based on a static CDO. Leveraging the spread premium Market spreads paid on securities bearing credit risk are typically larger than the levels implied by the historical default rates for the same rating. This difference. we see growing interest in synthetic-only portfolios. lack of liquidity or ratings downgrades. also known as ‘CDO squared’. which we call the spread premium. Typically this is a mezzanine ‘super’ tranche CDO in which the collateral is made up of a mixture of asset-backed securities and several ‘sub’ tranches of synthetic CDOs. arises because investors demand compensation for being exposed to default uncertainty. However. the higher the final return. The realised return is dependent on the timing of credit events. One example of such technology is Lehman Brothers’ DYNAMO structure. For a given number of defaults over the trade maturity. For investors the incentive is to leverage the management capabilities of a credit asset manager in order to avoid blow-ups in the 100 Credit window Coupons not reduced by defaults after maturity of credit window portfolio and so better manage downturns in the credit cycle. such as spread movements. The customisable characteristics include rating.assuming two credit events over the lifetime of the trade. More recently the format has evolved to one where the manager can change the portfolio subject to some constraints. Portfolio credit derivatives. This enables investors to enjoy all the benefits of customised tranches and the benefits of a skilled asset manager. recently. Lehman Brothers and a number of other dealers have managed to combine the customised tranche with the ability for an asset manager or the issuer of the tranche to manage the portfolio of reference entities. The problem with this type of structure is that the originator of the tranche has to factor into the spread the cost of substituting assets in the collateral. Figure 18. rating agency. the later they occur. issuance format plus others. subordination.

using the 22 The Lehman Brothers Guide to Exotic Credit Derivatives . it is the ratio of spread premium to the whole spread which goes up. The advantage of CDO of CDOs is that they provide an additional layer of leverage to the traditional CDO. For an STD basket we find that the spread premium is not leveraged. This makes them suitable for buy and hold yield-hungry investors who wish to be paid a high spread but also wish to minimise their default risk. 2. For buy and hold credit investors the spread premium paid can be significant and it is possible to show. The conclusion is that buy-and-hold correlation investors are overcompensated for their default risk compared with singlename investors. We therefore see that default baskets can appeal to a range of investor risk preferences. The investor enjoys a significant positive carry and at the same time reduces his idiosyncratic default risk. FTD baskets leverage spread premium. We list some strategies below. The investor buys CDO equity and hedges the full notional of the 10 or so worst names. This is a very straightforward strategy for mezzanine and senior investors. This has only 70bp of spread premium. Our results show that an FTD basket leverages the spread premium such that the size of the spread premium is much higher for an FTD basket than it is for a single-credit asset paying a comparable spread. There are therefore two conclusions: Spread (bp) CDO strategies Investors in correlation products should primarily view them as buy and hold investments which allow them to enjoy the spread premium. Spread premium for an FTD compared with a Ba3 single-name asset 400 350 300 250 200 150 100 50 0 FTD Instrument Spread premium Actuarial spread Ba3 1. see O’Kane and Schloegl (2003) for details of the method. STD baskets leverage the ratio of spread premium to the market spread. The investor may also sell CDS protection on the tightest names. This can make leveraging the spread premium arguments even more compelling. This is suitable for more risk-averse investors who wish to maximise return per unit of default risk. CDO tranches exhibit a similar leveraging of the premium embedded in CDS spreads. that under certain criteria.Figure 19. for equity investors. these assets may be superior to single-name credit investments. Compare this with a single-credit Ba3 asset also paying a spread close to 340bp. However. there are a number of strategies that can be employed in order to dynamically manage the idiosyncratic risk. the note issuer passes this on by selling protection in the CDS market. This is shown in Figure 19 where we see that an FTD basket paying a spread of 350bp has around 290bp of spread premium. sells protection via a default basket or a CDO tranche. This hedging activity makes it possible to pass this spread premium to the buyer of the structured credit asset. Instead. 1.

Hedge funds have been the main growth user of credit options. Hence.5% 2028 debentures and placed them into a Lehman Trust called CBTC. For more details see Isla (2003).7% of Lehman Brothers US High Yield Index have embedded call or put options. The more recent growth in the market for options on CDS has also been driven by the increased liquidity of the CDS market. 5. the CBTC trust must be over-collateralised with enough face value of MOT bonds to pay the 8. enabling investors to go long or short the option delta amount. The reduction in perceived default risk has made hedge funds. it can be very profitable if the market widens dramatically or if a large number of defaults occur. While this is a negative carry trade. Bank loan portfolio managers are beginning to explore default swaptions as a cheaper alternative to buying outright credit protection via CDS. This growth of the credit options market has been boosted by declines in both spread levels and spread volatility. it has extended to an increasingly vibrant market in both bond and spread options. 2. Since the 8. hedging in the CDS market and exploiting the positive convexity. By dynamically re-hedging. asset managers. Asset managers seeking to maximise risk-adjusted returns are involved in yield-enhancing strategies such as covered call writing. options on CDS and more recently options on portfolios and even on CDO tranches. In a typical repack trade. creating a cheap macro short position. Lehman Brothers purchased $32.375% – the prevailing rate for MOT in the retail market at the time. Both class A-1 The Lehman Brothers Guide to Exotic Credit Derivatives 23 . Credit options Activity in credit options has grown substantially in 2003.875. and continue to be important in the bond options market are the repack trade and put bond stripping.income to offset some of the cost of protection on the widest names. The Trust then issued $25 par class A-1 Certificates to retail investors with a coupon set at 8. They are typically buyers of volatility.000 of the Motorola (MOT) 6. insurers and proprietary dealer trading desks more comfortable with the spread volatility risks of trading options and more willing to exploit their advantages in terms of leverage and asymmetric payoff. two strategies which have been. The investor may use the carry from CDO equity to over-hedge the whole portfolio. One source of credit optionality is the cash market. The net positive gamma makes this trade perform well in high spread volatility scenarios. Measured by market value weight. The repack trade The first active market in credit bond options was developed in the form of the repack trade. From a sporadic market driven mostly by one-off repackaging deals.375% coupon on the CBTC trust is higher than the coupon on the MOT Bond.375% coupon. spearheaded by Lehman Brothers and several other dealers. using them for credit arbitrage and also for debt-equity strategies. 3.6% of Lehman Brothers US Credit Index and 54. the investor can lock in this convexity. The investor may buy CDO equity and delta hedge. Figure 20 (overleaf) shows the schematic of one such transaction. An A-2 Principal Only (PO) tranche captures the excess principal. The low liquidity of CDOs means that this hedging must continue to maturity.

Figure 20. Mechanics of MOT 6.5% 15/11/28 repack transaction 8.2 PO tranche MOT call warrant Market price Option to purchase MOT bond and class A-2 certificates are issued with an embedded call.515mm) Par $32. the market has priced these bonds as though they matured on the first put date and has not given much value to the extension option.50% 11/15/28 6. Essentially. bonds with embedded puts constitute approximately 2. Bond options There are a variety of bond options traded in the market. This strike is determined by the proceeds needed to pay off the A-1 certificate at par plus the A-2 certificate at the accreted value of the PO. Put bond stripping According to Lehman Brothers’ US Credit Index. The two most important ones for investors are: 24 The Lehman Brothers Guide to Exotic Credit Derivatives . Recently. In the past several years. but not the obligation to sell the bond back to the issuer at a predetermined price (usually par) at one or more future dates. He then hedges this position by asset swapping the bond to the put date. and sell the call option to the first put date at a strike price of par. effectively eliminating all of the interest rate risk and locking in the cheap volatility.36mm A . Because retail investors are willing to pay a premium for the par-valued low-notional bonds of wellknown high quality issuers.1 retail tranche $7. the bond maturity is extended. This option can be viewed as an extension option since by failing to exercise it. the investor has a long position in the bond coupled with a synthetic short forward (long put plus short call) with a maturity coinciding with the first put date.875mm MOT 6. this strategy has led to more efficient pricing of the optionality in these securities.3% of the US credit bond market. the buyers of the call warrant can use this structure to source volatility at attractive levels. Thus.375% (25. The holder of the MOT call warrant has the right but not the obligation to purchase the MOT bonds from the CBTC trust beginning on 19/7/07 and thereafter at the preset call strike schedule. Given the small amount of outstanding put bonds. by market value.515mm A .50% + par CBTC Series 2002 . These bonds grant the holder the right. credit investors have realised a way to extract this extension risk premium via a put bond stripping strategy. an investor can buy the put bond.14 Residual principal PV of CF Premium $25. This call option was sold separately to investors in a form of a long-term warrant.

If the bond price is above the option strike price.25% 25/10/11. Spread options can be structured using spreads to benchmark Treasury bonds. 1. the option holder pays a fixed amount (strike price) and receives the underlying bond – the payoff is proportional to the difference between the price of the bond and the strike price. If the bond price on the expiry date is greater than the strike. receiving an upfront premium. Three month price call option on F 7.13% premium Option buyer Lehman Brothers Right to buy F 7. it is delivered to the investor.25% 11 at 100.76 investor’s loss of upside on the price.Price-based options: at exercise. multiple pre-specified dates (Bermudan) or any date in a given range (American). the investor earns the premium. Default swaptions and callable CDS An exciting development in the credit derivatives markets in the past 12 months has been the emergence of default swaptions. They can be considered respectively as limit orders to sell or buy the underlying bond at a predetermined price (the option strike) on a predetermined day (the option expiry date). These are options on credit default swaps. corporate bond options. The emerging terminology from this mar- The Lehman Brothers Guide to Exotic Credit Derivatives 25 . the investor delivers the bonds and receives the strike price. Examples of actively trading price options are Brady bond options. See the illustration in Figure 21. the most active trading occurs in short-term (less than 12 months to expiry) European-style price options on bonds. Spread-based options: at exercise. The exercise schedule can be a single date (European). struck at 100. Naked put strategy: an investor writes an out-of-the-money put on a bond which he does not own but would like to buy at a lower price. The option premium offsets the Figure 21. Two of the most common strategies using price options on bonds are covered calls and naked puts. the option holder pays an amount equal to the value of the underlying bond calculated using the strike spread and receives the underlying bond – the payoff is proportional to difference between the underlying spread and the strike spread (to a first order of approximation). Currently. If the price is less than the strike the investor keeps the bonds and the premium. the main objective for the investor is to find a strike price at which he is willing to buy or sell the underlying bond and which provides sufficient premium to compensate for the potential upside that he forgoes. In both of these strategies.76% price. If the bond price on the expiry date is lower than the strike price. Covered call strategy: an investor who owns the underlying bond sells an out-of-money call on the same face value. The option premium compensates him for not being able to buy the bond more cheaply in the market. CBTC call warrants and calls on put bonds. default swap spreads or asset swap spreads.

a knock-out payer swaption should be delta Receiver default swaption In a receiver default swaption. the exercise decision for default swaptions is based on credit spread alone. the option buyer who wishes to maintain protection over the entire maturity range can separately buy protection on the underlying name until expiry of the swaption. (1. If spreads tighten by the expiry date.39). ■ If five-year GMAC trades at 238bp in three months. Payer default swaptions can be structured with or without a provision for knock out at no cost if there is a credit event between trade date and expiry date. The relevant scenarios for this investment are complementary to the ones in the case of the protection put. there is no need for a knockout feature for receiver default swaptions. for the right. Hedging default swaptions Dealers hedge these default swaptions using a model of the type discussed on page 49. they are essentially a ‘pure’ credit product.20% in a quarter of a year. the option buyer will not exercise the right to buy protection at the strike and the option seller will keep the option premium. The investor is short the option. the loss on the exercise of the option will be greater than the upfront premium and the investor will underperform on this trade. They can be traded outright or embedded in callable CDS. As a result. This spread is the option strike. Unlike price options on bonds. Consider the following example. and the time to option expiry is typically three months to one year. Lehman Brothers pays 1. We do not need to consider what happens if the reference entity experiences a credit event between trade date and expiry date as they would never exercise the option in this case. to sell CDS protection on a reference entity at a predetermined spread on a future date.ket is that protection calls (option to buy protection) are called payer default swaptions. Protection puts (option to sell protection) are called receiver default swaptions. The typical maturity of the underlying CDS is five years but can range from one–10 years. As a result. From the investor’s perspective. Payer default swaption The option buyer pays a premium to the option seller for the right but not the obligation to buy CDS protection on a reference entity at a predetermined spread on a future date. Theoretically. the option buyer pays a premium to the option seller 26 The Lehman Brothers Guide to Exotic Credit Derivatives . Default swaptions give investors the opportunity to express views on the future level and variability of default swap spreads for a given issuer. If five-year GMAC trades below 238bp in three months. If the knock out provision is included in the swaption. The investor has realised an option premium of 1. but not the obligation.20% for an atthe-money receiver default swaption on five-year GMAC. with pricing being mostly driven by CDS spread volatility. The underlying in a default swaption is the forward CDS spread from the option expiry date to the maturity date of the CDS. Lehman does not exercise. as they can sell protection for a higher spread in the market.20% up front option premium equals the payoff of (265bp–238bp)=27bp times the five-year PV01 of 4. the relevant scenarios are: ■ If five-year GMAC trades above 265bp in three months. the trade breaks even. struck at the current five year spread of 265bp and with three months to expiry.

These are a combination of plain vanilla CDS with an embedded short receiver swaption position. The rationale for options based on TRAC-X is that the portfolio effect will reduce the option volatility and make it easier for dealers to hedge. enabling us to buy protection at the lower market spread. there is a growing interest in callable default swaps. This combination will produce a synthetic forward CDS that knocks out at default before the forward date. The investor therefore has earned 315bp for selling five-year protection on GMAC for one year. Default swaption types Product Description Exercised if Credit view Knockout Payer default swaption Option to buy protection CDS spread at expiry > strike Short credit forward May trade with or without Receiver default swaption Option to sell protection CDS spread at expiry < strike Long credit forward Not relevant hedged with a short protection CDS to the final maturity of the underlying CDS and a long protection CDS to the default swaption expiry date. the contract continues and the investor continues to earn 315bp annually. In practice. The seller of a callable default swap is long credit exposure but this exposure can be terminated by the option buyer at some strike spread on a future date. callable in one year. The additional spread of 315bp–260bp=55bp in this example compensates the option seller for the lost potential upside. then Lehman Brothers will exercise the option and so cancel the protection. Credit portfolio options Starting in mid-2003 market participants have been able to trade in portfolio options whose underlying asset is the TRAC-X North America portfolio with 100 credits. From an investor perspective it presents a way to take a macro view on spread volatility. swaptions with expiry of 1 year and less are hedged only with CDS to final maturity due to a lack of liquidity in CDS with short maturities. From the perspective of the option seller. for 315bp from an investor. If the TRAC-X portfolio spread is wider than the strike level on the expiry date. We are now seeing investors trading both at-the-money and out-of-money puts and calls to maturities extending from three to nine months. Liquidity is also growing in the European version. Consider an example. We have summarised the key features of these different swaption types in Figure 22. the callable default swap has a limited MTM upside compared with plain vanilla CDS. The assumed current midmarket spread for five-year GMAC protection is 265bp. If the four-year GMAC spread in one year is less than the strike spread of 315bp. The contracts are typically traded with physical delivery.Figure 22. If the four-year GMAC spread in one year is greater than 315bp. Selling protection in callable default swap is equivalent to a covered call strategy on underlying issuer spreads and is particularly suitable as a yield-enhancement technique for asset managers and insurers. Lehman Brothers buys five-year GMAC protection. The Lehman Brothers Guide to Exotic Credit Derivatives 27 . the holder of the Callable default swaps In addition to default swaptions.

payer default swaption will exercise the option and lock in the portfolio protection at more favourable levels. interest rate and credit play into an almost pure credit play. these are credit event contingent instruments linked to the value of a derivatives payout. Typically. and because of the existing two-way markets with varying strikes. We discuss the modelling of hybrid credit derivatives in more detail on page 51. the investor loses the future dollar coupons and principal of the asset. Cross-currency asset swaps are the traditional mechanism by which credit investors transform foreign currency fixed-rate bonds into local currency Libor floaters. meaning that the payments on the swap contract are unaffected by any default of the asset. we give an overview of the economic rationale for different types of structures. ie. This has the benefit that it substantially reduces the currency and interest rate risk. if the asset does default. As the cross-currency swap is not contingent. the holder of the receiver swaption will benefit from exercising the option and realising the MTM gain. such as an interest rate swap or an FX option. However. just receiving some recovery amount which is paid in dollars on the dollar face value. There are various motivations for entering into trades which have these hybrid risks. As long as the underlying dollar asset does not default during the life of the asset swap there is no currency risk to the investor. Clean and perfect asset swaps One important theme is the isolation of the pure credit risk component in a given instrument. note that a cross currency asset swap is really two trades: (i) purchase of a foreign currency asset. In the case of a European investor purchasing a dollar asset. Conversely. converting the bond from an FX. because the TRAC-X spread is less subject to idiosyncratic spread spikes. 28 The Lehman Brothers Guide to Exotic Credit Derivatives . a European CDO investor may wish to access USD collateral without incurring any of the associated currency risks. We have modelled Hybrid products Hybrid credit derivatives are those which combine credit risk with other market risks such as interest rate or currency risk. Other strategies include expressing views on spread changes over a given time horizon by trading calendar spreads (buying near maturity options and selling farther maturity options). simultaneously buying payer and receiver default swaptions as a way to go long volatility while being neutral to the direction of spread changes. For example. investors can express their views on the direction of changes in the macro level of spread volatility by trading straddles. This unwind value can be positive or negative – the investor can make a gain or loss – depending on the direction of movements in FX and interest rates since the trade was initiated. The risk is significant. the investor is therefore obliged to either continue the swap or to unwind it at the market value with a swap counterparty. if the TRAC-X spread is tighter than the strike. Finally. First. However. Investors can monetise a view on the future range of market spreads by trading bearish spread (buying at-the-money receiver swaption and selling farther out-of-money receiver swaption) or bullish spread (buying ATM payer swaption and selling farther out-of-money payer swaption) strategies. Below. the investor receives Euribor plus a spread paid in euros. and (ii) entry into a cross-currency swap. the currency risk has not been completely removed.

if the value is positive. FX and credit. the swap is unwound at market value. 2. It is interesting to note that the reduction in the perfect asset swap spread may not be significant given that the two FX risks to the swap MTM and the recovery rate are actually partially offsetting. Both structures have featured widely in the CDO market where they have been used to immunise mixed-currency high-yield bond portfolios against currency risk in order to allow the structure to qualify for the desired rating from the rating agency. However. the shape of the Libor curves in both currencies. However. for a five year euro-dollar cross-currency swap using market calibrated parameters. Consider an investor who enters into an interest rate swap with a credit risky counterparty. Lehman Brothers takes on the default contingent swap unwind value risk and guarantees (quantoes) the recovery rate of the defaulted asset in the investor’s base currency. The cost of removing this default contingent swap MTM risk and paying the recovery rate in the investors domestic currency depends upon a number of factors including the volatility of the FX rate. European investors can use the perfect asset swap to take advantage of the often higher spread levels which exist for the same credits when denominated in US dollars. the investor loses a fraction of this MTM. A way to mitigate the counterparty risk is therefore to buy protection on the counter- The Lehman Brothers Guide to Exotic Credit Derivatives 29 . Modelled distribution of the swap MTM at default as a percentage of face value for a five-year euro-dollar swap 6 Probability (%) 5 4 3 2 1 0 –58 –28 2 32 63 93 Swap mark-to-market value at default (%) correlations between rates. As a result. This means that in an MTM framework the investor incurs no loss from the default event. The cost can be amortised over the life of the trade as a reduction in the asset swap spread paid. In the perfect asset swap. they can also be used by investors to convert foreign currency assets into their base currency. If the MTM of the swap is negative from the viewpoint of the investor. Suppose also that this counterparty defaults with a recovery rate of R. For example. the basic cross-currency asset swap has a default contingent interest rate and currency risk. Two variations exist. the credit quality of the reference credit. The downside risk is significant with possible swap related losses comparable in size to the loss on the defaulted credit. shown in Figure 23. 1. the swap could be replaced by one with more advantageous terms at zero cost.the distribution of the swap MTM at default. Lehman Brothers takes on the default contingent swap unwind value risk. interest rate volatilities in both currencies and the Figure 23. In the clean asset swap. Counterparty risk hybrids The mitigation of counterparty risk gives rise to another type of hybrid. This can be removed through the use of a hybrid.

coupled with the more mature credit derivatives market. For example. the hedger is implicitly taking a bullish view on the reference credit. Clearly. 30 The Lehman Brothers Guide to Exotic Credit Derivatives . similar structures can be constructed to provide credit protection for other payouts. Hedge cost mitigation The final class of application concerns the pure hedging of credit contingent FX or interest rate risks. These include paying a spread over a Constant Maturity Swap (CMS) rate or a particular inflation index. In this case. Companies looking to hedge interest rate or FX risk may be concerned with the cost of outright hedging using vanilla derivatives. there is increasing investor interest in credit-linked notes with more exotic coupons. Clearly. a hedge which knocks out on the default of a reference credit can provide an adequate hedge while significantly decreasing costs.party. one way to reduce the cost of credit hedging could be to purchase default protection linked to an FX rate being above or below a specified threshold. such as those faced by an international corporation in the course of its business activities. Yield enhancement In the current interest rate environment with very low short term rates and steep yield curves. where the contingent payout is linked to the replacement cost of the swap. As we discuss on page 51. the investor is effectively purchasing an interest rate swaption which is automatically exercised upon a default event.

the default is characterised as the consequence of some event such as a company’s asset value being insufficient to cover a repayment of debt. who characterise a credit event as the first event of a Poisson counting process which occurs at some time t with a probability defined as Single credit modelling The world of credit modelling is divided into two main approaches. and the length of the time interval dt. Reduced-form models also generally have the flexibility to refit the prices of a variety of credit instru- Pr[τ ≤ t + dt | τ > t ] = λ (t )dt ie. Structural models are generally used to say at what spread corporate bonds should trade based on the internal structure of the company. and they cannot be easily extended to price complex credit derivatives. this assumption also implies that the hazard rate is independent of interest rates and recovery rates. They therefore require information about the balance sheet of the company and can be used to establish a link between pricing in the equity and debt markets. They can also be extended to price more exotic credit derivatives. the credit process is modelled directly via the probability of the credit event itself. implementation and calibration. and at a multi-issuer level. Such models are usually extensions of Merton’s 1974 model that used a contingent claims analysis for modelling default. Pricing model for CDS The breakeven spread in a CDS is the spread at which the present values (PV) of premium and protection legs are equal. However. known as the hazard rate. The growth of the credit derivatives market has created a need for more powerful models and for a better understanding of the empirical evidence needed to calibrate these models. called structural and reduced-form. It is for these reasons that they are used for credit derivatives pricing. In this section we will present a detailed overview of modelling approaches from a practical perspective. ie. ments of different maturities. A common assumption is that the hazard rate process is deterministic. The hazard rate approach The most widely used reduced-form approach is based on the work of Jarrow and Turnbull (1995). In the structural approach.Credit Derivatives Modelling To be able to price and risk-manage credit derivatives. it is possible to show that the probability of surviving is given by   T  Q(0. By extension. t+dt) conditional on surviving to time t. they are limited in a number of ways including the fact that they generally lack the flexibility to fit exactly a given term structure of spreads. we need a framework for valuing credit risk at a single issuer level. T ) = E Q exp − ∫ λ ( s )ds      0    The expectation is taken under the risk-neutral measure. the probability of a default occurring within the time interval [t. we will discuss models. The Lehman Brothers Guide to Exotic Credit Derivatives 31 . Over a finite time period T. ie Premium PV = Protection PV. In the reduced-form approach. is proportional to some time dependent function λ(t).

Assuming that the hazard rate and risk free rate term structures are flat. the implied hazard rate is 0.s)λ(s)ds. then paying (1–R) and discounting this back to today at the risk-free rate. S = λ (1 − R ) This relationship is known as the credit triangle because it is a relationship between three variables where knowledge of any two is sufficient to calculate the third. the credit trades at 250bp. which implies a 3% hazard rate. where the sign is positive for a long protection position and negative for a short protection position.18%. we can write the value for the protection leg as T for the risk of default per small time interval.04%. Within the hazard rate approach we can solve this timing problem by conditioning on the probability of defaulting within each small time interval [s. If we assume that the spread S on the premium leg is paid continuously.03)=97. It basically states that the spread paid per small time interval exactly compensates the investor λ= S (t ) . For two years it is exp(–0. Valuation of a CDS position The value of a CDS position at time t following initiation at time 0 is the difference between the market implied value of the protection and the cost of the premium payments. We therefore write MTM(t) = ± (Protection PV –Premium PV).06)=94. given by Q(0.5. If the current market spread is given by S(t) then the MTM can be written as (1 − R) ∫ λe −( r +λ ) s ds = 0 λ (1 − R)(1 − e − ( r +λ )T ) r+λ The value of the premium leg is the PV of the spread payments which are made to default or maturity.To determine the spread we therefore need to be able to value the protection and premium legs. Assuming a recovery rate 32 The Lehman Brothers Guide to Exotic Credit Derivatives . One year later. and so on. and assuming a 50% recovery rate. s+ds]. which have been set contractually at SC.015 divided by 0. It is important to take into account the timing of the credit event because this can have a significant effect on the present value of the protection leg and also the amount of premium paid on the premium leg. Given a CDS which has a flat spread curve at 150bp. we can write the present value of the premium leg as MTM (t ) = ( S (t ) − S (0)) × RPV01 where the RPV01 is the risky PV01 which is given by S ∫ e −( r +λ ) s ds = 0 T S (1 − e r +λ − ( r +λ )T ) Equating the protection and premium legs and solving for the breakeven spread gives (1 − e ( RPV 01(t ) = and where − r + λ )(T − t ) (r + λ) ). 1− R An investor buys $10m of five-year protection at 100bp. Within this model the interest rate dependency drops out. The implied one-year survival probability is therefore exp(–0.

the cost of funding these other securities is also close to Libor. this is an area of pricing which has recently gained a lot of attention. S(0)=0. This may be assumed to be piecewise flat or piecewise linear. we mean modelling products whose pricing depends upon the joint behaviour of a set of credit assets. These real-world default probabilities are generally much lower. and in which we have seen a number of significant modelling developments.01 and t=4 into the above equation to give λ=4.025. This is a simple yet fairly accurate model which works quite well when the interest rate and credit curves are flat. The market standard is therefore to revert to rating agency default studies for estimates of recovery rates.of 40%. Pricing theory shows that the price of a derivative is the cost of replicating it in a riskfree portfolio using other securities. one should say that. r=3. as well as on supply-anddemand imbalances. R=40%. Adjustments may be made for non-US corporate credits and for certain industrial sectors. it also depends on the market’s risk aversion expressed through a risk premium. One way to incorporate this effect is to assume that recovery rates are stochastic. For a description of such a model see O’Kane and Turnbull (2003). they are similar to the definition of the recovery value in a CDS. Such expectations are only available from price information. As a consequence it is the effective risk-free rate for the derivatives market. expected recovery rate following a credit event where the expectation is under the risk-neutral measure. Recent work (Altman et al 2001) shows that there is a significant negative correlation between default and recovery rates. Default probabilities The default probabilities calculated for pricing purposes can be quite different from those calculated from historical default rates of assets with the same rating. The standard approach is to use a beta distribution. the recovery rate used in the pricing of credit derivatives is the The Lehman Brothers Guide to Exotic Credit Derivatives 33 . As a result of the growth in usage of these products. Modelling default correlation By modelling correlation products. S(t)=0.17% and an MTM value of $521.0%. it becomes necessary to use bootstrapping techniques to build a full term structure of hazard rates. The reason for this is that the credit spread of an asset contains not just a compensation for pure default risk. These include default baskets and synthetic CDOs. and usually focus on a US corporate bond universe. Since most market dealers are banks which fund close to Libor. it is difficult to separate the probability of default from the recovery rate expectation. the value is given by substituting.661. Problems with rating agency recovery statistics include the fact that they are backward looking and that they only include the default and bankruptcy credit events – restructuring is not included. One should also comment on the market’s use of Libor as a risk-free rate in pricing. These typically show the average recovery rate by seniority and type of credit instrument. Strictly speaking. as they represent the price of the defaulted asset as a fraction of par some 30 days after the default event. and the problem in credit is that given one price. Calibrating recovery rates The calibration of recovery rates presents a number of complications for credit derivatives. When this is not the case. In their favour.

Hull and White (2001) generate dependent default times by diffusing correlated asset values and calibrating default thresholds to replicate a set of given marginal default probabilities.1 While most multi-credit models require simulation. even if they produce the undesirable serial independence of the realised default rate. Clearly we see that the probability of both i and j defaulting. a classical dichotomy in credit models distinguishes between a ‘structural approach’. where default is triggered by the market value of the borrower’s assets (in terms of debt plus equity) falling below its liabilities. If we think of defaults as generated by asset values falling below a given boundary. Although the pay-offs of multi-credit default-contingent instruments such as nthto-default baskets and synthetic loss tranches cannot be statically replicated by trading in a set of single-credit contracts.Figure 24. multivariate structural models rely on the dependence of asset returns in order to generate dependent default events. and towards the end discuss model calibration issues. the 34 The Lehman Brothers Guide to Exotic Credit Derivatives . It follows that the valuation of these multi-credit exposures boils down to the computation of (risk-neutral) expectations over all possible default scenarios. This is shown in Figure 24 where we have simulated 1. and a ‘reducedform approach’. As discussed earlier. A number of different hybrid frameworks have been proposed in the literature for modelling correlated defaults and pricing multiname credit derivatives. the current market practice is to value correlation products using standard no-arbitrage arguments. where the default event is directly modelled as an unexpected arrival. The vertical and horizontal lines represent default thresholds for firms i and j respectively. Scatterplot of 1. structural models calibrated to market-implied default probabilities (often called ‘hybrid’ models) have gained favour among practitioners because of their tractability in high dimensions. Multi-period extensions of the one-period CreditMetrics paradigm are also commonly used. for two firms i and j for two different asset return correlations of 10% and 90%. Therefore asset correlation leads to default correlation. increases as the asset return correlation increases. represented by the number of points in the bottom left quadrant defined by the default thresholds. Default thresholds are also shown 10% Correlation VB CA 5 4 3 3 90% Correlation VB CA 5 4 2 1 0 -4 -2 -1 -2 -3 -4 0 2 2 VA 4 -4 -2 1 0 -1 -2 -3 -4 0 2 VA 4 CB CB Both firms default In this section we will describe some of these current models. Although both the structural and the reducedform approaches can in principle be extended to the multivariate case. show how they can be applied to the valuation of baskets and CDOs. then the probabilities of joint defaults over a specified horizon must follow from the joint dynamics of asset values: consistent with their descriptive approach of the default mechanism.000 pairs of asset returns modelled as normally distributed random variables.000 simulated asset returns with 0% and 90% correlation. Modelling joint defaults The valuation of default-contingent instruments calls for the modelling of default mechanisms.

1 Finger (2000) offers an excellent comparison of several multivariate models in terms of the default distributions that they generate over time when calibrated to the same marginals and first-period joint default probabilities. Default-time simulation Asset and default event correlation Default event correlation (DEC) measures the tendency of two credits to default jointly within a specified horizon. we need to sample from the multivariate distribution of default times under the risk-neutral probability measure. and PAB is the joint default probability. use the dependence among asset returns to generate joint defaults. and thus default event correlations. As a result. it is common practice to back out the marginal distributions of default times. In this case. market participants rely on alternative methods to calibrate the frequency of joint defaults within their models. pA.Fd. it is defined as the correlation between two binary random variables that indicate defaults.…. Its limits are not –100% to +100% but are actually a function of the marginal probabilities themselves. the scarcity of default data makes joint default probabilities. where each path is simply a list of default times for each of the credits in the reference portfolio drawn at random from the joint default distribution. it is common practice to proxy asset correlations using equity correlations. In an influential paper.F2. more recently. from single-credit defaultable instruments (such as CDS). Towards the end of DEC = p AB − p A p B p A (1 − p A ) p B (1 − p B ) where pA and pB are the marginal default probabilities for credits A and B. Of course. which we will denote with F1. which according to the stated assumptions represents the correlation matrix of the asset returns of the reference credits. Finger (1999) and. very hard to estimate directly. The Lehman Brothers Guide to Exotic Credit Derivatives 35 . Since asset returns are not directly observable. ie Monte Carlo models generally aim at the generation of default paths. We then join these marginal distributions with a correlation matrix. pB and pAB all refer to a specific horizon. Notice that default event correlation increases linearly with the joint probability of default and is equal to zero if and only if the two default events are independent. Li (2000) presents a simple and computationally inexpensive algorithm for simulating correlated defaults.need for accurate and fast computation of greeks has pushed researchers to look for modelling alternatives. Unfortunately. therefore avoiding the need for a direct estimation of joint default probabilities. no matter how complex the contractual specification of the payoff. For valuation purposes. Gregory and Laurent (2003) show how to exploit a low-dimensional factor structure and conditional independence to obtain semi-analytical solutions. Hybrid models. Default event correlations are the fundamental drivers in the valuation of multi-name credit derivatives. such as the simulation approach and the semi-analytical framework described below. His methodology builds on the implicit assumption that the multivariate distribution of default times and the multivariate distribution of asset returns share the same dependence structure. Knowing the time and identity of each default event allows for a precise valuation of any multi-credit product. which he assumes to be Gaussian and is therefore fully characterised by a correlation matrix. Formally.

Simulating default times from this distribution is straightforward. Transform the vector into the unit hypercube using U = (N (x1 ). Generally speaking. τ 2 < y) = N 2 . where N2. With d reference credits and correlation matrix Σ we have the following algorithm: 36 The Lehman Brothers Guide to Exotic Credit Derivatives . Fd−1 (u d )) The simulation algorithm is illustrated in Figure 25.8 0. In the bivariate case..6 0.0 0. F2−1(u 2 ).4 0.2 0 Cumulative default probability 2 0 x 2 4 6 0 2 4 6 Time (years) 8 10 this section we will discuss whether this seems to be a reasonable approximation. N (x d )) 3. Once we know how to sample from the risk-neutral distribution of default times.. implicitly relying on the extra assumption that the correlations among asset returns remain unchanged when we move from the objective to the pricing probability measure.. The high-dimensionality of multi-credit instruments means that it is not possible to use market prices to obtain the full dependence structure.6 0.8 0. the joint distribution function of default times and is simply given by 1.τ d ) = (F1−1(u1).0 0.τ 2 . P( τ 1 < x. Choleski decomposition of the correlation matrix to simulate a multivariate Normal random vector X ∈ R d with correlation Σ. it is straightforward to price a correlation trade.2 0 6 4 Cumulative normal 1. practitioners generally estimate the necessary correlations using historical returns.4 0. N (x 2 ) .ρ denotes the bivariate cumulative standard Normal with correlation ρ. 2... Instead.... The extension to the d-dimensional case is immediate.... Translate U into the corresponding default times vector t using the inverse of the marginal distributions: τ = (τ 1 . the valuation of a given correlation product always boils down to the computation of an expectation of the form.N −1(F2(y))). and N–1 indicates the inverse of a cumulative standard Normal. ρ (N −1(F1(x)).Figure 25. Mapping a normal random variable to a default time 1. It is easy to verify that τ has the given marginals and a Normal dependence structure fully characterised by the correlation matrix Σ..

Monte Carlo simulation allows for accurate valuations and risk measurements of multi-credit payoffs.E[f (τ)] where τ = (τ1. For example. We now discuss how to exploit a one-factor model to construct the risk neutral loss distribution.τ2. Moreover. Even in this case.. come at the cost of computational speed. assume a one-factor correlation structure for asset returns.. 1) makes it possible to compute the risk-neutral loss distribution of the reference portfolio for any given horizon. interest coverage or collateralisation tests are involved. This problem becomes particularly significant when we turn our attention to the calculation of sensitivity measures.. Useful techniques include antithetic sampling. while 2) is needed to price an instrument knowing only the loss distribution of the reference portfolio at a finite set of dates. techniques exist to alleviate the problem. There are ways to improve the situation.. and 2. since for a reasonable number of paths the simulation noise can be similar to or greater than the price change due to the perturbation of the input parameter. Later on. it is usually possible to obtain reasonable approximations. The question then arises whether some other numerical approach can be used. however. One way to do this is to rely on two basic simplifications: 1. These are chosen with a resolution to provide a sufficient level of accuracy. A semi-analytical approach The recent development of correlation trading and the associated need for fast computation of sensitivities have generated a great deal of interest in semi-analytical models. To enjoy the advantages of fast pricing. The precision and flexibility of this approach. the error can be controlled by comparing the analytical solution to a Monte Carlo implementation. In particular. In summary. Once again. importance sampling and the use of low-discrepancy sequences. we can easily expand the set of variables to be treated as random. and a large number of simulation paths are usually required to achieve a sufficient sampling of the probability space. we will show how to use a sequence of loss distributions at different horizons to price synthetic loss tranches. Frye (Risk. The Lehman Brothers Guide to Exotic Credit Derivatives 37 . even when complex path-dependencies such as reserve accounts. derivatives structures with more complex path-dependencies may also require that we approximate the payoff function f(τ). While these two assumptions are sufficient to price plain vanilla portfolio swaps.τd) represents the vector of default times and f is a function describing the discounted cash flows (both positive and negative) associated with the instrument under consideration (see. The basic problem of using simulation is that defaults are rare events. but it is hard to achieve precise hedge ratios in a reasonable amount of time. discretise the time-line to allow for a finite set of dates at which defaults can happen. 2003) argues that modelling stochastic recoveries and allowing for negative correlation between recovery and default rates is essential for a proper valuation of credit derivatives. eg. This is because both the exact default times and the identity of the defaulters are known on every simulated path. so that the benefit of precise sensitivities can be retained at a relatively low cost. Mashal and Naldi (2002b)). Moreover. one needs to impose more structure on the problem.

Given that Ai is N(0. A loss u corresponds with the previous zero loss probability times the probability that asset two defaults plus the previous u loss probability times the probability asset two does not default.A one-factor model Let us start by assuming that the asset return of the ith issuer between today and a given horizon is described by a standard Normal random variable Ai with mean 0 and standard deviation of 1. the calculation of joint default probabilities If we assume that the loss on default for each issuer is the same unit (consistent with all assets having the same seniority). Using the threshold levels determined before. In the single-issuer case. and is of the form Ai = β i Z M + 1 − β i2 Z i where Z1. substantial simplifications can be achieved by imposing more structure on the model. this is merely an exercise in calibration. and βi stands for the correlation of the asset return of issuer i with the market.. The variable ZM describes the asset returns due to a common market factor. The zero loss peak requires that the new asset survives and so has a probability (1–p1(ZM))(1–p2(ZM)). However. The process is as follows: 1. 2. making it necessary to resort to Monte Carlo simulation. Adding a second asset.. Conditional on the market factor ZM.. This makes it easy to compute conditional default probabilities. then building up the portfolio loss distribution can be done iteratively by adding assets to the portfolio. an asset defaults if Zi ≤ Ci − β i Z M 1 − β i2 The conditional default probability pi (ZM) of an individual issuer i is therefore given by C − β Z i M pi ( Z M ) = N  i  1− β 2 i      pi = N (C i ) ⇒ C i = N ( pi ) −1 and calibration of the marginal probabilities is no more complicated than inverting the cumulative normal function.. With a general correlation structure.1) distributed independent random variables. while Zi models the idiosyncratic risk of the ith issuer. The asset return correlation between asset i and asset j is given by βiβj. In this case. the returns are assumed to be described by a multivariate Normal distribution. Beginning with asset one. u. it is possible to write becomes computationally intensive. the asset returns are independent. conditional on ZM. Mathematically. The advantage of this one-factor setup is that. there are two outcomes on the loss distribution: a loss of zero with a probability 1–p1(ZM) and a loss of u with a probability p1(ZM).Z2. The concept of asset returns becomes meaningful when studying the joint behaviour of more than one credit. the ith issuer defaults in the event that Ai < Ci. Default will occur if the realised asset return falls below a given threshold.1) distributed. as the default thresholds are chosen to reproduce default probabilities which re-price market instruments. we adjust each of the previous losses. it is possible to obtain the probabilities of joint defaults.ZD are N(0. We arrive at a 38 The Lehman Brothers Guide to Exotic Credit Derivatives .

Calculate the PV of a 1bp coupon stream paid to time τ*=min[τn(i). both models should generate exactly the same prices. known as the basket PV01. perhaps drawn from a beta distribution. see Mashal and Naldi (2002a). If τn(i) < T then calculate the PV = B(τn(i))(1–R(i)). 5. We start with the default basket. It is also quite straightforward to introduce alternative copulas. where i is the label of the defaulted asset. The main constraint is to build the basket nto-default probabilities while retaining the identity of the defaulted assets. 3. See Gregory and Laurent (2003) for a full discussion. within this framework has been described in detail above. A loss of 2u can only occur if the previous loss of u is multiplied by the probability that asset two also defaults to give a probability of p1(ZM)p2(ZM). The generation of default times Rating models for default baskets Rating agencies have developed a number of models for rating default baskets. to calculate the fair-value spread we proceed as follows: 1. Both the times-to-default model and the one-factor model described above are Gaussian copula models. typically m=5–10 assets means that pricing is quite fast in Monte Carlo. One approach is therefore to use the times-todefault model via a Monte Carlo implementation. This is an important point – it means that models can be categorised by their copula specification alone.1) distributed market factor ZM. as long as they use the same copula. Given that we know when each asset in the portfolio defaults in each simulation path. Divide the protection leg by the basket PV01 to get the fair-value spread. and as long as both use the same (one-factor) correlation matrix. We continue adding on to the portfolio until all of the assets have been included. Otherwise the protection PV = 0. Average both the premium leg PV of 1bp. Sort default times in ascending order and denote the nth time to default as τn(i). 2. Pricing default baskets A default basket is inherently an idiosyncratic product in the sense that the identity of the defaulted asset must be known. in which an asset value is simu- The Lehman Brothers Guide to Exotic Credit Derivatives 39 .T ] where T is the basket maturity. This approach is simple to implement and the size of default baskets. An analytical approach is also possible here. For example Moody’s has adopted a Monte Carlo-based extension of the asset value approach. where B is the Libor discount factor and R(i) is the recovery rate for asset i. 4. Monte Carlo is also flexible enough to enable you to introduce stochastic recovery rates. Valuation of correlation products We now explain in detail how these models may be applied to the pricing of correlation products. We then repeat with a different value of ZM and integrate the loss distributions over the N(0. It also means that different products may be priced consistently using different default mechanisms. The advantage of this approach is that there is no simulation noise.probability (p1(ZM)+p2(ZM)–2p1(ZM)p2(ZM)). Other algorithms exist. and the protection leg PV over all paths. This means that as long as both are calibrated to the same marginals. including Fast Fourier techniques which may be more efficient. 3.

and for each time t≤T we denote by L(t) the cumulative portfolio loss up to time t. This approach is simple to implement but Note that this is similar to an option style payoff.0] Pricing synthetic loss tranches We would now like to consider in detail how we would set about valuing a loss tranche. The tranche loss is therefore given by Ltranche (t ) = max[L(t ) − K d . and the protection leg PV over all paths. Average both the 1bp premium leg PV. the contingent leg is therefore given by Protection Leg PV = 40 The Lehman Brothers Guide to Exotic Credit Derivatives . 4. Calculate the PV of 1bp on the premium leg taking care to reduce the notional of the tranche following defaults which cause principal losses in that path. consider a portfolio CDS on a tranche defined by the attachment point Kd and the upper boundary Ku. To see this. 3. The recovery amount and asset value can be correlated. Divide the protection PV by the tranche PV01 to compute the breakeven tranche spread. recovery rate and interest rates processes are independent. we would proceed as follows: 1. can be computationally slow if there are a large number of assets in the portfolio. If the asset value falls below the default threshold at the future period. known as the tranche PV01. the asset defaults and a recovery rate is drawn from a beta distribution. Assuming that the credit. expressed as percentages of the reference portfolio notional. This is Nport so that the tranche notional is given by N tranche = N port ( K u − K d ) The maturity of the portfolio swap is T. S&P has recently switched its approach from a weakest-link approach which assigns an FTD the rating of the lowest-rated entity in the basket. which uses the same framework as its CDO Evaluator model. The model is calibrated using historical default statistics and the assigned rating is linked to the expected loss of the basket to the trade maturity. One approach is to use the times to default model via a Monte Carlo implementation as discussed earlier. 2.0] − max[L(t ) − K u . Given that we know when each asset in the portfolio defaults in each simulation path. The two legs of the swap can be priced in the same way as a CDS if we introduce a tranche ‘default probability’ P(t) defined as Q E 0 [ Ltranche (t )] N tranche P(t ) = where we use the risk-neutral (pricing) measure for taking the expectation. These asset values may be correlated internally in order to induce a default correlation. Another way is to use a semi-analytical approach which relies on the fact that a standard synthetic loss tranche can be priced directly from its loss distribution. to one based on a Monte Carlo model. Indeed it is possible to think of CDO tranches as options on the portfolio loss amount. Calculate the present value of all principal losses on the protection leg in each path.lated to multiple periods for each of the assets in the basket.

the conditional default probability of any individual issuer in the reference portfolio is given by To value the premium leg. time 0 to time u. This approach can be more efficient than Monte Carlo since it is uses the loss distribution directly and there are fast ways of calculating this. the simplest being first order differences: Protection Leg PV = thetic CDO if we have the loss distribution at the set of future dates t and T. N tranche ∑ B(0. In practice we would generally assume that monthly intervals would be sufficient.   If we also assume that the loss exposure to each issuer is of the same notional amount u. the probability that the percentage loss L of the portfolio is ku is equal to the probability that exactly k of the m issuers default. To emphasise this point. that each asset’s β and default probability are the same. We can discretise the integral by introducing the grid points 0=t0<t1<…<tK =T where greater accuracy is obtained by having a higher number of grid points. One way to do this is to use the one-factor model described above to calculate a loss distribution at each of the future dates required. In fact we can use this approach to obtain results which differ from the exact approach by 1–5% for a real CDO. Hence. T j ) j =1 n The PV of the tranche from the perspective of the investor who is receiving the spread is therefore given by Tranche PV = Premium Leg PV – Protection Leg PV.N tranche ∫ B (0. ie. To begin with. The accrual factor for the ith payment is denoted by ∆i. one can actually exploit the large number of assets in a synthetic CDO to derive a closed-form analytical solution to calculating the loss distribution of a portfolio. Care must be taken to ensure that the default threshold is re-calibrated at each horizon date such that the marginal distribution is correctly recovered. and the coupon payment dates as 0=T0<T1<…<Tk=T. u )dP(u ) 0 T Where B(0. The Lehman Brothers Guide to Exotic Credit Derivatives 41 . As a result we can value a standard syn- C − βZ M p(Z M ) = N   1− β 2    . As a result. There are a number of ways of evaluating an integral on this interval. let us assume that the portfolio is homogeneous. we denote the contractual spread on the tranche by s. t i )( P (t i ) − P(t i −1 )) i =1 K An asymptotic approximation It is possible to exploit the high dimensionality of the CDO to derive a closed form model for analysing CDO tranches. all assets have the same default threshold C.u) is the Libor discount factor from today. The PV of the premium leg is then given by Premium Leg PV= sN tranche ∑ ∆ j (1 − P(T j )) B(0.

we can observe a skew in the average correlation as a function of the width and attachment point of the tranche. This is discontinuous at the edges owing to the probability of the portfolio loss falling outside the interval. Since the asset returns. it can easily be extended to multiple periods as described earlier. and this discontinuity becomes more pronounced when the tranche is narrowed. but we can use methods of varying sophistication to approximate it. One very simple and surprisingly accurate method is the so-called ‘large homogeneous portfolio’ (LHP) approximation. the fraction of issuers defaulting will tend to the conditional probability of an asset defaulting p(ZM). However. As a fraction of 42 The Lehman Brothers Guide to Exotic Credit Derivatives .K2) is the percentage loss of the mezzanine tranche with attachment point K1 and upper loss threshold K2. We can use the portfolio distribution to derive the loss distributions of individual tranches.which is given by the simple binomial the tranche notional this is given by  m  C − β ZM P[L = ku | Z ]=   N  k     1− β 2   1 − N  C − β Z M   1− β 2       m−k     k L(K1 . originally due to Vasicek (1987). − 1 − β 2 − N 2 − N −1 (K 2 ). We can further compute the expected loss of the tranche analytically.C . by the law of large numbers. the conditional loss is directly linked to the value of the market factor ZM which itself is normally distributed. It may also be due to our incorrect specification of the dependence structure. If we attempt to imply out the market correlation using a simple Gaussian copula model fitted to observed market tranche spreads. For more details see O’Kane and Schloegl (2001). 0) K 2 − K1 If K<1. as where N(x) is the cumulative normal function. then this can be written as a function of the loss on the reference portfolio. P[L ≤ K]= N − p −1 ( K ) [ ] Correlation skew It is now possible to observe tradable tranche spreads for different levels of seniority. If L(K1. then L(K1 . K 2 ) = max(L − K1 . This skew may be the consequence of a number of factors such as the assumption of independence of default and recovery rates. 0)− max (L − K 2 . K 2 )≤ K ⇔ L ≤ K1 + K (K 2 − K1 ) This equation shows that we can easily derive the loss distribution of the tranche from that of the reference portfolio. We then arrive at E [L (K 1 . conditional on ZM are independent and identically distributed. K 2 )] = This distribution becomes computationally intensive for large values of m. − 1 − β 2 K 2 − K1 ( ) ( ) This is only a one-period approach.C . More involved calculations show that the distribution of L actually converges to this limit as m tends to infinity. We can then write the probability of the portfolio loss being less than or equal to some loss threshold K as N 2 − N −1 (K 1 ). As a result.

3 Fitch Ratings (2003) have recently published a special report describing their methodology for constructing portfolio loss distributions: it is based on a Gaussian copula parameterised by equity correlations. Supply and demand imbalances also play a role. a default probability distribution is created. The widespread use of the Normal dependence structure.explained later. Estimating the dependency structure Several well-known multivariate models. In S&P’s ratings methodology.3 However. and to estimate the parameters governing the joint behaviour of asset returns from equity return series. the portfolio is represented by a lower number of independent assets. whether this assumption is supported by empirical evidence. is certainly related to its simplicity. however. the use of unobservable underlying processes is one of several criticisms that the structural approach has received over the years. S&P rates on the basis of the probability of incurring a loss. which is fully characterised by a correlation matrix. Moody’s standard rating model for CDO tranches is a multinomial extension of the Binomial Expansion Technique (BET) model. default correlation and structural features. A number of recent studies have shown that the joint behaviour of equity returns is better described by a ‘fat-tailed’ Student-t copula than by a Normal copula. In fact. for example. The Lehman Brothers Guide to Exotic Credit Derivatives 43 . Mashal and Naldi (2002a) and Mashal and Zeevi (2002).2 The first goal of this section is to apply the same kind of analysis to asset returns. including the ones described earlier in this chapter. Roughly speaking. Corporate sectors are assumed to have a correlation of 30% within a given industry and 0% between industry sectors. and that correlations are therefore not sufficient to appropriately characterise their dependence structure. The diversity score is determined using a lookup table and is based on the incremental effect of having groups of assets in the same industry classification. the model is able to generate an expected loss for a tranche which takes into account the subordination. this is the number of independent assets which have the same width of loss distribution as the actual CDO reference portfolio of correlated assets. Of course. it has become customary to proxy the asset dependence with equity dependence. From these inputs and the par amounts of each asset. This expected loss is then mapped to a rating category. a Monte Carlo simulation is used to derive a probability loss distribution for the underlying collateral pool based on the total principal balance of the portfolio. After calibrating to historical default data. rely on the assumption that the dependence structure (or ‘copula’) of asset returns is Normal. dependent on its rating and maturity. It remains to be seen. recovery rates. All of them attempt to capture the risks of CDOs in terms of asset quality. the use of equity returns to infer the joint behaviour of asset See. Each corporate asset is assigned a default probability based on S&P’s historical default studies. and test the null hypothesis of Gaussian dependence versus the alternative of ‘joint fat tails’. 2 Rating agency models for CDOs Different rating agencies have their own models for rating CDO tranches. we face a major obstacle when attempting to estimate the dependence structure of asset returns: asset values are not directly observable. Given the lack of observable asset returns. To capture default correlation. known as the diversity score. Unlike Moody’s which rates on the basis of expected loss.

The Gaussian distribution lies at the heart of most financial models and builds on the concept of correlation. the recent survey paper by Embrechts et al. Such a procedure is at the heart of KMV’s CreditEdge™. the degrees-of-free- 44 The Lehman Brothers Guide to Exotic Credit Derivatives . Naldi and Zeevi (2002). a popular credit tool that first computes a measure of distance-to-default and then maps it into a default probability (EDF™) by means of a historical analysis of default frequencies. One way to estimate the market value of a company’s assets is to implement a univariate structural model. the Student-t retains the notion of correlation but adds an extra parameter into the mix. CreditEdge and EDF are trademarks of KMV LLC.01 10 0 10 1 10 2 10 Null DoF 3 10 4 10 5 10 6 in our study is given by the dependence structure underlying the multivariate Student-t distribution. Alternatively.01 10 0 10 1 10 2 10 Null DoF 3 10 4 10 5 10 6 DJIA Equity Returns Sensitivity Analysis 3 10 Methodology A key observation in modelling and testing dependencies is that any d-dimensional multivariate distribution can be specified via a set of d marginal distributions that are ‘knitted’ together using a copula function.4 In what follows. A particular copula that plays a crucial role 4 Copyright © 2000-2002 KMV LLC. (2003). All rights reserved. Copulas have many important characteristics that make them a central concept in the study of joint dependencies.0001 10 1 pvalue = 0. eg. To provide a plausible answer to these questions. a copula function can be viewed as ‘distilling’ the dependencies that a multivariate distribution attempts to capture. KMV and the KMV logo are registered trademarks of KMV LLC. The second goal of this section is to shed some light on the magnitude of the error induced by using equity data as a proxy for asset returns.0001 10 1 pvalue = 0. by factoring out the effect of the marginals. Figure 26. DJIA portfolio: asset and equity returns test statistics as functions of null hypothesis for DoF DJIA Asset Returns Sensitivity Analysis 3 10 10 Test Statistic 2 pvalue = 0. Test Statistic 10 2 pvalue = 0. we summarise recent work by Mashal.returns is often criticised on the grounds of the different leverage of assets and equity. we first need to ‘back out’ asset values from observable data. who use the asset value series generated by KMV’s model to study the dependence properties of asset returns. namely. see.

it is well known that the Student-t distribution is very ‘close’ to the Gaussian when the DoF is sufficiently large (say. while asset values have been ‘backed out’ by means of KMV’s implementation of a univariate Merton model. Naldi and Zeevi (2003) for more examples using high yield credits and different sampling frequencies. Then. Moreover. the Gaussian model is nested within the t-family. With this in mind. we can reject any value of the DoF parameter outside the range [10. which is our main motivation for focusing on it. They also construct a likelihood ratio statistic to test the hypothesis of Gaussian dependence. Moreover. we estimate the number of degreesof-freedom (DoF) of a t-copula without imposing any structure on the marginal distributions of returns. these studies would target the dependence structure rather than the distributions themselves. In particular. The Lehman Brothers Guide to Exotic Credit Derivatives 45 . The reader should keep in mind. a value of the test statistic falling below these lines corresponds to a value of DoF that is not rejected at the respective significance levels. thus. Empirical results For the purpose of this study.dom (DoF). asset and equity values are both obtained from KMV’s database. the t-dependence structure constitutes an important and quite plausible generalisation of the Gaussian modelling paradigm. The same statement holds for the underlying dependence structures. using a likelihood ratio test statistic. the difference between the joint tail behaviour of a 12. The minimal value of the test statistic is achieved at 12 DoF for asset returns and at 13 DoF for equity returns. and compare the dependence structures of asset and equity returns to evaluate the common practice of proxying the former with the latter. however.99%. as captured by the DoF parameter. Naldi and Zeevi (2002) describe a methodology which can be used to estimate the parameters of a t-copula without imposing any parametric restriction on the marginal distributions of returns. that equity values are observable. in particular. the null assumption of a Gaussian copula (DoF=∞) can be rejected with an infinitesimal probability of error.and a 13-DoF t-copula is negligible in terms of any practical application. This suggests how empirical studies might test whether the ubiquitous Gaussian hypothesis is valid or not. we perform a sensitivity analysis for various null hypotheses of the underlying tail dependence. the point estimates of the asset returns’ DoF lies within the non-rejected interval for the equity returns’ DoF. thus eliminating the effect of marginal returns that would ‘contaminate’ the estimation problem in the latter case. The two horizontal lines in Figure 26 represent significance levels of 99% and 99. Following the methodology mentioned above. and the DoF parameter effectively serves to distinguish the two models. We apply our analysis to a portfolio of 30 credits included in the Dow Jones Industrial Average and use daily data covering the period from 31/12/00 to 8/11/02. indicating that the two are essentially indistinguishable from a statistical significance viewpoint. Also. and vice versa. To summarise. The reader is referred to Mashal.16] with 99% confidence. In the remainder of this section we summarise their empirical findings. In both cases. The latter plays a crucial role in modelling and explaining extreme comovements in the underlyings. Mashal. greater than 30). the key question that we now face is how to estimate the parameters of the dependence structure.

Figure 28 compares the expected discounted losses for several tranches under the two alternative assumptions of Gaussian dependence and t dependence with 12 DoF . Notice that even larger differences can be observed if one compares higher moments or tail measures of the tranches’ loss distributions. Maximum likelihood estimates of DoF for DJIA portfolios Portfolio 30-credit DJIA First 10 credits Middle 10 credits Last 10 credits Asset returns Equity returns DoF DoF 12 8 10 9 13 9 10 9 Figure 27 reports the point estimates of the DoF for asset and equity returns in the DJIA basket. EDL because this measure relates both to the agency rating (when computed under realworld probabilities) and to the fair compensation for the credit exposure (when computed under risk-neutral probabilities). The LHP with tail dependence It is possible to incorporate a Student-t copula into the LHP model discussed above. Ai = β i Z + 1 − β i2 ε i W /ν where W is an independent random variable 46 The Lehman Brothers Guide to Exotic Credit Derivatives . the correlation coefficients. iii) 20% asset correlation between every pair of credits. as well as for three subsets consisting of the first. This gives Example: synthetic loss tranche The models described earlier in this chapter can be modified to account for a fat-tailed dependence structure of asset returns. Here we analyse the impact of a non-Normal assumption on the expected discounted loss (EDL) of a portfolio loss tranche. The similarities between the joint tail dependence (as measured by the DoF) of asset and equity returns are quite striking. The maximum absolute difference (element-by-element) is 4. ie.5 Next. Using a robust estimator based on Kendall’s rank statistic6. we must change the distribution for the asset returns to be multivariate Studentt distributed where we denote the DoF parameter with ν and retain the one-factor correlation structure. iv) a risk-free curve flat at 2%.1%. The results show the significant impact that the (empirically motivated) consideration of tail dependence has on the distribution of losses across the capital structure: expected losses are clearly redistributed from the junior to the senior tranches. exhibiting similar behaviour to that displayed in Figure 26. each with $1m notional. providing further evidence of the similarity of the two dependence structures. ii) 1% yearly hazard rate for each reference credit. we compute the two 30x30 correlation matrices from asset and equity returns. To do so. We consider a five-year deal with a reference portfolio of 100 credits. and the mean absolute difference is 1. as a consequence of the increased volatility of the overall portfolio loss distribution. we compare the remaining parameters that define a t-copula. Using a default-time simulation.6%. and last 10 credits (in alphabetical order). 6 See Lindskog (2000). for every credit in the reference portfolio. We focus on 5 The range of accepted DoF is very narrow in each case.Figure 27. We assume i) uniform recovery rates of 35%. middle.

300 (0. and show in O’Kane and Schloegl (2003) that it possesses the same tail dependent properties as described above.06) $90. but for options struck on yield it depends on the time to maturity of the underlying bond and is found from a standard yield-to-maturity calculation. Bond options struck on spread are different. yield or credit spread.630 (0. The fact that both the market and idiosyncratic terms ‘see’ the same value of W means that they are no longer independent. The option can be struck on price.645 (1. for European options there is no difference between specifying a strike price and a strike yield. Obviously. The KMV algorithm that produces the asset values used in our analysis is nothing other than a sophisticated way of de-leveraging the equity to get to the value of a company’s assets. Pricing options on bonds Options on corporate bonds are naturally divided into three groups according to how the exercise price is specified.012. Note that this is no longer a ‘factor model’ in the sense that the asset return is composed of two independent and random factors.14) $533.70) t copula DoF=12 EDL (std err %) $2.62) Pctg diff -11 13 51 117 Summary Our empirical investigation of the dependence structure of asset returns sheds some light on the two main issues that were raised at the beginning of this section. Therefore.following a chi-square distribution with ν degrees of freedom. 100K paths. the assumption of Gaussian dependence between asset returns can be rejected with extremely high confidence in favour of an alternative ‘fat-tailed dependence.231 (1. Second. ie. For this model. Figure 28. For credit spread options the exercise price depends both on the time to maturity The Lehman Brothers Guide to Exotic Credit Derivatives 47 . these results represent good news for practitioners who only have access to equity data for the estimation of the dependence parameters of their models. From a practical point of view.37) $41.256.200 (0.66) $221. the popular conjecture that the different leverage of assets and equity will necessarily create significant differences in their joint dynamics seems to be empirically unfounded.’ Multivariate structural models that rely on the normality of asset returns will generally underestimate default correlations.160 (1. standard errors in parenthesis Tranche Normal copula (%) EDL (std err %) 0-5 5-10 10-15 15-20 $2. and thus undervalue junior tranches and overvalue senior tranches of multi-name credit products. a widening of senior spreads and a reduction in equity tranche spreads. This is the simplest possible way to introduce tail dependence via a Student-t copula function. the dependence structures of asset and equity returns appear to be strikingly similar.23) $601.63) $146. A fat-tailed dependence of asset returns will produce more accurate joint default scenarios and more accurate valuations. First. Expected discounted loss.020 (0. Modelling credit options We separate credit options into options on bonds and options on default swaps. we have been able to calculate a closed-form solution for the density of the portfolio loss distribution.120 (1. The exercise price is constant for options struck on price.

It is relatively easy to build a lattice for a lognormal yield. it is important to specifically take into account the default risk.of the bond and on the term structure of interest rates at the exercise time. For these issuers. A credit spread strike is commonly specified as a yield spread to a Treasury bond or interest rate swap. An important problem with Black-Scholes is that it does not properly account for the pull-to-par of the bond price. It is rarely possible to calibrate volatility parameters because of the lack of liquid bond option prices. the model should be calibrated to an issuer-specific credit curve. For high quality issuers. interest rate volatility is the main driver of price action. The problems of pull-to-par of price and non-constant volatility can be solved by a full term structure model such as Black-DermanToy. Naldi. Extending interest rate models to modelling risky rates only works if we assume either zero recovery on default. Figure 29. However. creditors have a claim for return of full face value in bankruptcy. Chu and Wang (2002) and Berd and Naldi (2002) present a multifactor framework for modelling the stochastic components of corporate bond returns. Yield diffusion models can be constructed to fit forward bond prices but would usually assume constant yield volatility which is inconsistent with empirical evidence. Instead of calibrating the model to Libor rates. or assume that recovery is paid as a fraction of the market value at default. In the next section we discuss 48 The Lehman Brothers Guide to Exotic Credit Derivatives . For short-dated European options with price (or yield) strike on long-term bonds. Usually it is more appropriate to base volatility parameters on historical estimates. These models were designed for default-free interest rates but can be applied analogously to credit risky issuers. say. or as an asset swap spread. This problem can be solved by a yield diffusion model where the bond’s yield is the underlying stochastic process. This is especially important for lower credit quality issuers where the bonds trade on price rather than yield. When pricing spread options. Black-Karasinski or Heath-JarrowMorton. The lattice approach also allows for easy valuation of American/Bermudan exercise. Receiver default swaption Default swap cash flows if exercised into Default Default swaption swaption expiry date settlement T TS CDS maturity date TM The framework can be used to derive price/yield volatilities from return volatilities and tend to give more robust estimates than direct estimation. and price the option by standard backwards induction techniques. so it is necessary to specifically model the recovery at default as a fraction of face value. the above approaches are less controversial for pricing options struck on price or yield. Volatility parameters should therefore be related to implied volatilities from interest rate swaption markets but must also incorporate the negative correlation between credit spreads and interest rates (see Berd and Ranguelova (2003)) which can cause yield volatility on corporate bonds to be significantly lower than comparable interest rate swaption volatility. the well-known Black-Scholes formula goes a long way but is not recommended beyond this limited universe.

Consider a European payer swaption with option expiry date T and strike spread K. The distribution of ST should then be such that E[ST ] = A0/B0 where the probability of default before T is put to 0. With these definitions the ratio AT /BT is equal to the spread ST if default has not occurred at T. (See Harrison and Kreps (1979) for details.0}]  BT  If we make the assumption that log(ST) is normally distributed with variance σ2T. First. let us clarify terminology. in which case AT is 0 if default happens before T and PST otherwise. say A.0}].0} S where the PV01T is the value at T of a risky 1bp annuity to time TM or default. Bond options struck on credit spread can be priced in similar fashion. Pricing default swaptions The growing market in default swaptions has led to a demand for models to price these products. we have determined the distribution of ST to be used to find E[max{ST –K. This terminology is analogous to that used for interest rate swaptions. the ratio A0/B0 of today’s values of the securities is equal to the expectation of the ratio AT /BT of the security payments at T. The value of the swaption today is then  PS  PS 0 = B0 E  T  = PV 010 ⋅ E [max{ST − K . say B. See page 32 for a discussion of CDS pricing. The states where BT = 0 are simply ignored in that case and the distribution of ST will be such that the probability that BT = 0 is 0. the option payoff is PST = PV 01T ⋅ max{ T − K . Black’s formulas for interest rate swaptions can be modified to price European default swaptions. The next step is to let A be the swaption. then with the requirement E(ST) = F0 (the forward spread).) To use this result we first let A be a security that at T pays 0 if default has occurred and otherwise pays the upfront cost (as of T) of a zero-premium CDS with the same maturity as the CDS underlying the swaption. It is easy to calculate this expectation and we arrive at the Black formula PS 0 = PV 010 ⋅ (F0 ⋅ N (d1 ) − K ⋅ N (d 2 ) ) . that only makes a payment at T. Conditional on surviving to time T. corresponding to the spread following a log-normal process with constant volatility σ. We let B be a security that at T pays 0 if default has occurred and otherwise pays PV01T. there is a probability distribution of the spread ST such that for any other security. Ignoring the maximum.how to price options to buy or sell protection through CDS. d1 = log( F0 / K ) + σ 2T / 2 σ T d 2 = d1 − σ T The Lehman Brothers Guide to Exotic Credit Derivatives 49 . The contract is to enter into a long protection CDS from time T to TM. A0/B0 is the T-forward spread. for the underlying CDS where the forward contract knocks out if default occurs before T. From finance theory we know that for any given security. otherwise AT /BT and ST are not defined. The result is valid even if BT can be 0 as long as AT = 0 when BT = 0. An option to buy protection is a payer swaption and an option to sell protection is a receiver swaption. this is the standard payoff calculation for a forward starting CDS. that also only makes a payment at T. denoted F0. and it knocks out if default occurs before T (see Figure 29). and ST is the market spread observed at T on a CDS with maturity TM.

which is the five year CDS spread on the valuation date. A stochastic hazard rate model is naturally combined with a term structure model to produce a unified model that can. The differences between directly modelling the credit spread and modelling the hazard rate are similar to the differences between modelling the yield of a bond and using a term structure model as discussed above. a protection payment is not made until option maturity. is less than straightforward. which means that $10m notional can be bought for $252. Since bond and the CDS markets have their own dynamics. The Black formula for a receiver swaption is found analogously. To illustrate the use of the Black formula. when possible. The last step in valuing the swaption is to find the volatility σ to be used in the Black formula.211%. Calculation of F0 and PV010 should be done using a credit curve that has been calibrated to the current term structure of CDS spreads. It is not uncommon for CDS spreads to make relatively large jumps as reaction to firm specific news.52%. we recommend using a stochastic hazard rate model. but criticism analogous to that for using a yield diffusion model to price bond options applies. Interest rates and credit risk One way to model credit spread dynamics is via a hazard rate process. The value of the credit protection is therefore less than the upfront cost of a zero-premium CDS that matures with the option. (See Hull and White (2003) and Schonbucher (2003) for more details. calibration of the volatility parameters of the hazard rate process. a number of practical complications arise in getting such an approach to work.027 and the forward spread is 274. Also. it may not be appropriate to naively price all options in the same calibrated model without properly adjusting for the basis between CDS and bonds.000. especially when calibrating to bond options struck on 50 The Lehman Brothers Guide to Exotic Credit Derivatives . The knockout feature is not relevant for receiver swaptions. be used to price both bond options and credit spread options. To price default swaptions with Bermudan exercise we could construct a lattice for the forward spread. as they will never be exercised after default. at least in theory. However. the PV01 used in the Black formula is 4. The strike is 260. we must add the value of credit protection from today until option maturity. The valuation date is 28/8/2003. and PV010 are values under the knockout assumption.000 and sold for $229. Instead. These are prices of nonknockout options.29%/2. consider a payer swaption on Ford Motor Credit with option maturity 20/12/2003 and swap maturity 20/9/2008. These numbers imply that the bid/offer volatilities quoted by our desk are approximately 75% and 85%. However calibrating a mixed jump and Brownian process to spread dynamics is not easy. usually not implying the same issuer curves. The fair value of protection until swaption maturity given the credit curve on the valuation date is 0.where N is the standard normal distribution function. Examination of CDS spread time series also reveals that the log-normal spread assumption often is inappropriate. Under a non-knockout payer swaption. F0. This provides a consistent framework for modelling spreads of many different maturities. Our trading desk quoted the swaption at 2. An estimate of σ can be made from a time series of CDS spreads.7bp.) It is important that the forward spread. To value a payer swaption that does not knock out at default.

A strip decomposition. we assume that default can only take place at times ti. because such spread measures do not properly take into account the risk of default. under the assumption of independence. this is paid by the protection seller.0). where each swaption is priced conditional on default happening at ti. Representing the default protection via a strip of swaptions is a very useful framework for developing intuition around the pricing and helps us understand the importance of the shape of the interest rate curve. it is necessary to determine the correlation between interest rates and hazard rates. as we have shown above. For such bonds there can be a significant decrease in OAS when interest rates increase even if hazard rates remain unchanged. If the MTM of the receiver swap. RSt is positive to the investor at the time of default. only the default probabilities.price or yield. These parameters are then taken as given when determining the hazard rate volatility parameters by fitting to time series estimates or calibrating to market prices. or to bonds with embedded options. As a consequence. it is important to be Modelling hybrids The behaviour of hybrid credit derivatives is driven by the joint evolution of credit spreads and other market variables such as interest and exchange rates. In terms of tractability. This is The Lehman Brothers Guide to Exotic Credit Derivatives 51 . then computing the expected discounted cash flows gives the value V0 for the price of the default protection. which are commonly modelled as diffusion processes. For simplicity. the protection seller has sold swaptions and is therefore short interest rate volatility. If B denotes the price process of the savings account. so that the payoff at default is max(RSt. Correlations should be estimated directly using bond prices or CDS spreads. We illustrate the main ideas via the example of default protection on the MTM of an interest rate swap. A cross-currency swap for example could be dealt with in exactly the same way. then there will also be a spread volatility dependence. where  max (RSt . Using OAS is especially a problem for long maturity bonds with high default probabilities and high recovery rates. However. 0 )  n i V0 = ∑ E  τ = ti  P [τ = ti ] B (ti )   i =1   The interpretation of this equation is that the value of default protection is a probability weighted strip of receiver swaptions. Suppose an investor has entered into a receiver swap with fixed rate k with a credit risky counterparty. with the exchange rate as an additional state variable. If RSt is negative. For this instrument. Interest rate volatility parameters should be calibrated separately using liquid prices of interest rate swaptions. If the rate and credit process are correlated. When estimating correlations it is important not to use yield spread or OAS (option adjusted spread) as a proxy for the hazard rate. is the basic building block for most hybrid credit derivatives. the investor receives nothing. an option to enter into a receiver swap with fixed rate k for the remaining life of the original trade at default. In terms of volatility exposure. the exchange rate exposure is of prime importance. the reduced form approach is the natural framework for pricing and hedging these products. the volatility of the credit spread does not enter into the valuation. Finally.

we have had to rely on estimates of historical volatilities. Credit spread volatilities are somewhat more involved. Such a model will not necessarily generate levels of dependence representative of periods of market stress where investment grade defaults are likely. conditioning instead on the realisation of the hazard rate process. A further consequence of the hazard rate method is that we do not actually have to condition on the realisation of the default time itself. it will be interesting to see to what extent a market in implied correlations will develop via standardised hybrid credit derivatives. As a starting point. 52 The Lehman Brothers guide to exotic credit derivatives . as one does not have to explicitly simulate default times and is a useful variance reduction technique. At this stage of the market it is safe to say that this correlation is a ‘realised’ parameter as opposed to an implied one. Determining the correct dependence structure between credit spreads and the other market variables is the main challenge in modelling hybrids. however. The simplest approach is to work within a diffusion setting where spreads and interest rates/FX are correlated. Calibrating volatilities for interest rates and FX is relatively straightforward. Going forward. Until recently. it appears to be reasonable. Even within this framework. ie pricing and hedging must proceed on the basis of a view on correlation founded on historical estimates. now the growing options markets are starting to make the calibration of implied spread volatilities feasible. The parameters needed for pricing hybrids are essentially volatility and dependence parameters. the effect of correlation on the valuation of a hybrid instrument can be marked. This is particularly advantageous for Monte Carlo simulation.able to calculate the conditional expectations appearing in the strip decomposition.

November Ganapati S and Ha P 2002 . New York University. 1979 Martingales and arbitrage in multiperiod securities markets Journal of Economic Theory 20. I will survive Risk June. pages 12–22 Hull J and White A. pages 95–99 Ganapati S. 2001 Valuing credit default swaps II: modelling default correlations Journal of Derivatives 8. Lehman Brothers. November Ganapati S. Ha P and Ranguelova E. pages 381–408 Hull J. Lehman Brothers. Prentice Hall Hull J and White A. Lehman Brothers. Credit Products Special Report. 1999 Conditional approaches for CreditMetrics portfolio distributions RiskMetrics Group. Berd A. Resti A and Sirone A. December Berd A and Naldi M. 2000 A comparison of stochastic default rate models RiskMetrics Group. New York Finger C. pages 103–107 Harrison J and Kreps D. 2000 Options. 2001 Analyzing and explaining default recovery rates Report submitted to ISDA. New York. McNeal A. Ha P and O’Kane D. 2003 Default correlation and its effect on portfolios of credit risk Structured Finance. University of Toronto Isla L. Embrechts P Lindskog F and . September Berd A and Ranguelova E. March Gregory J and Laurent J-P 2003 . Evolution of synthetic arbitrage CDOs Structured Credit Strategies. futures and other derivatives Fourth edition. New York Fitch Ratings. Stern School of Business. Elsevier. 2003 The co-movement of interest rates and spreads: implications for credit investors High Grade Research. 2003 The synthetic CDO bid and basis convergence – which sector is next? Structured Credit Strategies. 2003 The valuation of credit default swap options Working paper. Lehman Brothers. 2002 New estimation options for the Lehman Brothers risk model Quantitative Credit Research Quarterly. September The Lehman Brothers Guide to Exotic Credit Derivatives 53 .References Altman E. Lehman Brothers. February Frye J. 2003 Hedged CDO equity strategies Fixed Income Research. 2003 Modelling dependence with copulas and applications to risk management In Handbook of Heavy Tailed Distributions in Finance. Lehman Brothers. pages 329–384 Finger C. 2003 A false sense of security Risk August. 2001 Synthetic CDOs – how are they structured? how do they work? CDO Monthly Update.

2002b Beyond CADR: searching for value in the CDO market Quantitative Credit Research Quarterly.Jarrow R and Turnbull S. Berd A. pages 53–85 Li D. Columbia University. 1987 Probability of loss on loan portfolio Working paper. Lehman Brothers. RiskLab. Pedersen C and Turnbull S. 2000 Linear correlation estimation Working paper. Lehman Brothers. 2000 On default correlation: a copula function approach Journal of Fixed Income 9. July O’Kane D and Turnbull S. 2003 Upfront credit default swaps Quantitative Credit Research Quarterly. ETH Zurich Vasicek O. September 2002 Mashal R. May O’Kane D and McAdie R. September Naldi M. KMV Corporation 54 The Lehman Brothers Guide to Exotic Credit Derivatives . Lehman Brothers. February O’Kane D and Schloegl L. December Mashal R. Naldi M and Zeevi A. O’Kane D and Desclee R. pages 43–54 Lindskog F. Lehman Brothers. 2003 Valuation of credit default swaps Quantitative Credit Research Quarterly. 2002 The new Lehman Brothers credit risk model Quantitative Credit Research Quarterly. 2001 Modelling credit: theory and practice Lehman Brothers. London Mashal R and Zeevi A. June O’Kane D and Schloegl L. 2002a Extreme events and default baskets Risk June. 2001 Explaining the basis: cash versus default swaps Lehman Brothers. March O’Kane D. Lehman Brothers. Naldi M and Zeevi A. 2003 Leveraging the spread premium with correlation products Quantitative Credit Research Quarterly. Risk. Lehman Brothers. Lehman Brothers. 2002 Beyond correlation: extreme co-movements between financial assets Working paper. Chu K and Wang G. A note on survival measures and the pricing of options on credit default swaps Working paper. Naldi M and Pedersen C. July O’Kane D and Sen S. 2002 The dependence structure of asset returns Quantitative Credit Research Quarterly. pages 119–122 Mashal R and Naldi M. June Schonbucher P 2003 . 2003 Extreme events and multi-name credit derivatives In Credit Derivatives The Definitive Guide. ETH Zurich Mashal R and Naldi M. 2003 Valuing the restructuring clause in CDS Lehman Brothers. 2003 Leverage and correlation risk of synthetic loss tranches Quantitative Credit Research Quarterly. 1995 Pricing derivatives on financial securities subject to credit risk Journal of Finance 50. New York Munves D. 2003 The Lehman Brothers CDS indices Fixed Income Research. April Mashal R. Lehman Brothers.

recording. or any opinions given. or transmitted. neither they nor any company associated with the publication can accept legal or financial responsibility for consequences. London SW1Y 4RX © 2003 Lehman Brothers and Incisive RWG Ltd All rights reserved. omissions. stored in or introduced into any retrieval system. Rules made for the protection of investors under the UK Financial Services Act may not apply to investment business conducted at or from an office outside the UK. in any form or by any means. Risk Waters Group and Lehman Brothers and the author have made every effort to ensure the accuracy of the text. Published by Incisive Risk Waters Group.Approved by Lehman Brothers International (Europe). or otherwise. Haymarket House. No part of this publication may be reproduced. electronic. which may arise from errors. mechanical. however. without the prior written permission of the copyright owners. The Lehman Brothers Guide to Exotic Credit Derivatives 55 . photocopying. which is authorised and regulated by the Financial Services Authority. 28–29 Haymarket.

Notes 56 The Lehman Brothers Guide to Exotic Credit Derivatives .

With over 375 dedicated professionals worldwide covering the whole spectrum of Fixed Income products.An Established Tradition of Research Excellence Weekly commentary across asset classes The Lehman Brothers Global Fixed Income Research Group has been consistently recognised as a top-ranked research department. Lehman Brothers International (Europe) is regulated by the Financial Services Authority. .com All Rights Reserved. In-depth product analysis For further information please contact the Global Fixed Income Research Group or visit www. .lehmanlive.Hub portal Dominant global marketshare in Fixed Income Indices and Index Analytics .com Awards LehmanLive . All of our research and analysis is accessible through our website.com New York: Ravi Mattu +1 212 526 0206 Europe: Robert Campbell +44-(0)20-7260-2639 Prafulla Nabar +44-(0)20-7256-4480 Asia: Hua He +81-3-5571-7215 rmattu@lehman. LehmanLive.com For the fourth year in a row.com pnabar@lehman. the Group was ranked #1 in the Institutional Investor 2003 All-America Fixed Income Research Team Survey Institutional Investor ranked Lehman Brothers #1 for Fixed Income Research in 10 out of the last 14 years Lehman Brothers website. .com rcampbel@lehman.#1 dealer website on the Bond. Award-winning website www. . . LehmanLive .lehmanlive.com hhe@lehman. Member SIPC.available on our POINTTM platform We have an established tradition of providing the highest quality research services to our clients.winner of 8 “Best Site” categories in the Euromoney. ©2003 Lehman Brothers Inc. we provide strategic advice and analysis to our institutional clients. .

Sign up to vote on this title
UsefulNot useful