Professional Documents
Culture Documents
RISK MANAGEMENT
& VALUE
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
Mondher Bellalah
Université de Cergy-Pontoise, France
World Scientific
NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TA I P E I • CHENNAI
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN-13 978-981-283-862-9
ISBN-10 981-283-862-7
Printed in Singapore.
DEDICATION
Mondher Bellalah
v
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
FOREWORD
This book covers the main aspects regarding derivatives, risk, and the role of
information and financial innovation in capital markets and in the banking
system. An analysis is provided regarding financial markets and financial
instruments and their role in the 2007–2008 financial crisis. This analysis
hopefully will be useful in avoiding or at least mitigating future financial
crises.
The book presents the principal concepts, the basics, the theory, and
the practice of virtually all types of financial derivatives and their use in risk
management. It covers simple vanilla options as well as structured products
and more exotic derivative transactions. Special attention is devoted to risk
management, value at risk, credit valuation, credit derivatives, and recent
pricing methodologies.
This book is not only useful for specific courses in risk management
and derivatives, but also is a valuable reference for users and potential
users of derivatives and more generally for those with risk management
responsibilities.
vii
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
FOREWORD
ix
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
FOREWORD
xi
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
FOREWORD
Both the trading of options and the theory of option pricing have long
histories. The first use of option contracts took place during the Dutch
tulip mania in the 17th century. Organized trading in calls and puts began
in London during the 18th century, but such trading was banned on several
occasions. The creation of the Chicago Board Options Exchange (CBOE)
in 1973 greatly encouraged the trading of options. Initially, trading took
place at the CBOE only in calls of 16 common stocks, but soon expanded
to many more stocks, and in 1977, put options were also listed. The great
success of option trading at the CBOE contributed to their trading in other
exchanges, such as the American, Philadelphia and Pacific Stock Exchanges.
Currently, daily option trading is a multibillion-dollar global industry.
The theory of option pricing has had a similar history that dates
to Bachelier (1900). Sixty-five years after Bachelier’s remarkable study,
Samuelson (1965) revisited the question of pricing a call. Samuelson
recognized that Bachelier’s assumption that the price of the underlying
asset follows a continuous random walk leads to negative asset prices, and
thus makes a correction by assuming a geometric continuous random walk.
Samuelson obtained a formula very similar to the Black–Scholes–Merton
formula, but discounted the cash flows of the call at the expected rate of
return of the underlying asset. The seminal papers of Black and Scholes
(1973) and Merton (1973) ushered in the modern era of derivatives.
This is a lucid textbook treatment of the principles of derivatives
pricing and hedging. At the same time, it is an exhaustively comprehensive
encyclopedia of the vast array of exotic options, fixed-income options,
corporate claims, credit derivatives and real options. Written by an expert
in the field, Mondher Bellalah’s comprehensive and rigorous book is an
indispensable reference on any professional’s desk.
xiii
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
xv
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
CONTENTS
Dedication v
Foreword by Edward C. Prescott vii
Foreword by Harry M. Markowitz ix
Foreword by James J. Heckman xi
Foreword by George M. Constantinides xiii
About the Author xv
xvii
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
Contents xix
Contents xxi
Contents xxiii
Contents xxv
Contents xxvii
7.4. The Hull and White Trinomial Model for Interest Rate
Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
7.5. Pricing Path-Dependent Interest Rate Contingent Claims
Using a Lattice . . . . . . . . . . . . . . . . . . . . . . . . . 355
7.5.1. The framework . . . . . . . . . . . . . . . . . . . . 355
7.5.2. Valuation of the path-dependent security . . . . . 358
7.5.2.1. Fixed-coupon rate security . . . . . . . . 358
7.5.2.2. Floating-coupon security . . . . . . . . . 359
7.5.3. Options on path-dependent securities . . . . . . . 359
7.5.3.1. Short-dated options . . . . . . . . . . . . 359
7.5.3.2. Long-dated options . . . . . . . . . . . . 359
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
Contents xxix
Contents xxxi
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Contents xxxiii
Contents xxxv
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
Appendix A: An Alternative Derivation of the Compound Option’s
Formula Using the Martingale Approach . . . . . . . . . . . 656
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664
Contents xxxvii
Contents xxxix
Contents xli
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 826
Appendix D: The Monte–Carlo Method and the Dynamics of
Asset Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . 829
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 830
Contents xliii
Appendix C: The Algorithm for the American Put with Dividends 862
Appendix D: The Algorithm for CBs with Call and Put Prices . . 864
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 867
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 867
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 872
Contents xlv
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 943
This page intentionally left blank
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Part I
1
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Chapter 1
Chapter Outline
This chapter is organized as follows:
3
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Introduction
In the last three decades, there has been a wave of financial innovations
and structural changes in the securities industry. The three main natural
questions which arise are:
What are the specific features of the “new” financial contracts?
Why are there so many financial contracts?
What are the fundamental reasons behind the proliferation of financial
assets?
A partial answer to such questions is given in the analysis of Miller
(1986), Merton (1988), and Ross (1989), among others. These financial
instruments are traded either in organized markets or in non-organized
markets, known as over-the-counter markets, or OTC markets. These
products are presented either in a straight forward form or in a package.
They can be used to create several combinations with different risk and
reward trade-offs.
Financial crisis, subprime and credit crunch in 2008–2009 are exacer-
bated by the use of derivatives in the areas of mortgages, credit and other
areas of finance.
the new forward rate will be set at a new equilibrium level. Forward rate
contracts are flexible and allow for customized hedges since all the terms
can be negotiated with the counterparty. However, each side of the contract
bears the risk that the other side defaults on the future commitments. That
is why futures contracts are often referred to as forward contracts.
price risk. The first contract, gas oil futures began trading in 1981. A Brent
crude futures contract was launched in 1988. A natural gas futures contract
is also traded. A network of Quote Vendors relay information on a real-
time basis to end users in several countries worldwide. The Brent crude
contract and the gas oil contract are used as benchmarks or price references
in trading.
progressively more expensive in the future. Hence, the oil price depends not
only on its quality but also on the delivery date range. The location of the
oil affects its price.
mechanism. The unit of trading is one or more lots of 1000 net barrels
(42,000 US gallons) of Brent crude oil. The contract specifies the current
pipeline export quality Brent blend as supplied at Sullom Voe. The contract
price is in US dollars and cents per barrel. The minimum price fluctuation
is one cent per barrel, which gives a tick value of US$10. All open contracts
are marked-to-market daily.
Options
The IPE offers American options contracts for Brent crude and gas oil
futures. Options enable companies to carry out several complex and hedging
techniques. The unit of trading is represented for IPE gas oil options is one
IPE gas oil futures contract. The contract price is in US dollars and cents
per tonne. The strike price increments are multiples of US$5 per tonne.
A minimum of 5 strike prices are listed for each contract month. Due to
futures style margining, option premiums are not paid or received at the
time of the transaction. Margins are received or paid each day according to
the changing value of the option. The total value to be paid or received is
only known when the position is closed. This is done by an opposing sale
or purchase, the exercise or the maturity of the option. The options can be
exercised into gas oil futures contracts.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Spread trading.
Forward cargoes can either be traded as single cargoes with an absolute
price agreed, or in spread trading. This latter case involves the simultaneous
purchase or sale of at least two cargoes and appears in different forms.
Inter-month spreads.
Spread trading in the Brent market appears as transactions of the difference
in price between Brent for delivery in different months using two Brent
cargoes.
When trader A buys an April/May spread from trader B, then A has
bought an April cargo from B and simultaneously sold a May cargo to B.
The inter-month spread is simply a position on the absolute level of
the backwardation or contango between the delivery months. In general, a
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
contango appears when prices are higher for more distant delivery months.
Backwardation is the reverse.
Inter-crude spreads.
It is possible to trade the differential against another crude oil as Dubai
or WTI. When trader A buys an April Brent-Dubai spread, he buys the
April Brent cargo and sells the April Dubai cargo. An inter-crude spread is
a position taken on the path of the difference in prices between the crude
oils. Trader A will gain if the price of the Brent strengthens relative to that
of Dubai.
Swaps.
The swap allows the producer or the consumer of crude oil and oil products
to lock in a price or a margin. The main participants are finance houses and
the trading departments of large oil companies. A producer can arrange
a swap for a given volume over a specified period at a price equating
to a “mean” market price over that period. At each agreed settlement
period, actual market prices for the agreed volume are compared to the
value of that volume under the specified price in the swap transaction.
When market prices are higher, the producer pays the swaps provider the
difference times the agreed volume. When market prices are lower, the swaps
provider pays the producer the difference times the agreed volume. In the
swap transaction, there is a physical exchange of oil, but a series of netted
transactions or contract for the differences.
1.4.2. Arbitrage
Arbitrage involves the simultaneous purchase of futures or physical com-
modities in one market against the sale of the same quantity of futures
or physical commodity in a different market. An arbitrage strategy is
often implemented to take advantage of differentials in the price of the
same instrument on different markets. Cocoa can be traded on CSCE in
dollars and LIFFE in pound sterling. The arbitrage price can be derived by
subtracting the CSCE price converted to pound sterling from the LIFFE
price. In practice, London cocoa sells at a premium over New York cocoa
because of a quality in the difference of cocoa.
The arbitrage price is affected by the forces of supply and demand and
by exchange rates. Arbitrage allows speculation on whether the premium
of London cocoa will increase or decrease over New York.
Trade house.
Buys from country of origin and assumes risks associated with transporting
and selling the product to buyers in consuming countries.
Processor.
Buys cocoa beans and/or produces cocoa liquor, powder, and butter.
Manufacturer.
Buys beans and/or sell the above products from the trade-houses and or
processors.
Speculator.
The use of the cocoa contract by managed futures funds, who tend to take
short-term positions. Institutional investors have a long-term view.
Trade house.
Buys from country of origin and assumes risks associated with transporting
and selling the product to buyers in consuming countries.
Roaster.
Buys green coffee and roasts it.
Manufacturer.
Buys beans and/or sell the above products from the trade-houses and or
processors.
Speculator.
The use of the Robusta coffee contract by managed futures funds, who
tend to take short-term positions and institutional investors, who have a
long-term view.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Trade house.
Buys from country of origin and assumes risks associated with transporting
and selling the product to buyers in consuming countries.
Manufacturer.
Buys either raw sugar in bulk for further refining or white sugar in clean
bags from both country of origin and trade-houses.
Speculator.
The managed funds are a vital part of daily volumes.
Country Index
Index options on stock indexes and stock index futures began trading
in the United States in 1983 with the introduction of the S&P 100 contract
on the Chicago Board Options Exchange. The 10 largest markets in the
FT-Actuaries World Index have listed options (See Table 1.1).
In these countries, index futures are also traded. In general, combined
options and futures volumes exceed trading in the underlying stocks.
Volume is concentrated in one-month contracts. The volume in options
with longer maturities takes place in the OTC. The OTC options market
began to develop in 1988.
Examples
The DOW JONES STOXXSM 50 AND DOW JONES
EURO SOXXSM 50
The specific features of the Dow Jones STOXXSM 50 index are as follows.
Dow Jones STOXXSM 50 composition: Basket of 50 highly liquid
European blue chips (16 countries), belonging to the main business sectors.
Calculation method: The index level is given by: I = 1000 (sum of
real-time market capitalization for each component stock/adjusted base
capitalization).
The index is calculated in real-time by STOXX Ltd.
Price quotation: The index is disseminated every 15 seconds by
ParisBourse.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
The specific features of the Dow Jones Euro STOXXSM 50 index are
as follows.
Dow Jones Euro STOXXSM 50 composition:
Basket of 50 highly liquid Euro zone (10 countries) blue chips, belonging
to the main business sectors.
Calculation method:
The index level is given by: I = 1000 (sum of real-time market
capitalization for each component stock/adjusted base capitalization). The
index is calculated in real-time by STOXX Ltd. Price quotation: The index
is disseminated every 15 seconds by ParisBourse (Table 1.2).
needs. Options on the currency futures have been traded since 1982. These
options are standardized contracts.
A coupon-paying bond.
This bond is often regarded as a portfolio of several cash flows (the coupons)
where each cash flow can be seen as a zero-coupon bond. Hence, a coupon-
paying bond can be viewed as a package of zero-coupon bonds.
The principal amount of a bond issue can decrease or amortize during
the life of the interest-sensitive instrument. The principal is paid back
gradually at a given rate and interest is paid on the amount of the principal
outstanding.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
A corporate bond.
This is a bond issued by a firm. The bond obliges the issuer to pay interest
rate charges and the principal amount according to a specified schedule.
If the bond is guaranteed by some assets of the issuer, it becomes a
mortgage bond. If the only guarantee is represented by the credibility of
the issuer, the bond is a debenture bond. Each bond issue is accompanied
with a document known as indenture. It specifies the main features of
the issue.
Some bonds are not redeemed before another class of debt. They are
referred to as junior or subordinated bonds. The higher priority claims are
referred to as senior bonds. A sinking fund provision is often inserted in the
bond indenture to describe the way bondholders will be paid.
Indexed bonds.
These bonds are useful when the operating profits of a corporation are
exposed to the fluctuations of an index, as with a commodity price like
oil, aluminium or inflation. The exposure risk can be partially hedged by
issuing bonds whose interest rate payments and/or principal repayment
is linked to the index, in such a way that the effective cost of debt is
reduced when there is an unfavorable movement in the price index, and is
increased to the benefit of the investors when the movement is favorable
to the firm. Such a bond issue can be split into parts: the bull and bear
tranches, so that investors can choose only one side of the risk exposure.
These bonds allow some investors to take risky positions which are not
directly available to them, or not allowed, on organized markets. Investors
are ready to pay a premium for these opportunities which is translated into
a reduced financing cost.
A convertible bond.
Entitles its holder the right to convert the bond into a certain number of
units of the equity of the issuing firm or into other bonds.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
An exchangeable bond.
It is similar to a convertible bond, with the exception that it gives its
holder the right to exchange the bonds for the equity of another company,
etc. When a corporation has a low credit rating and must implement a
large investment program to survive, it may well be too costly to issue
standard debt, while raising equity might dilute considerably the current
shareholders’ position. Then, warrants (bonds with attached warrants) and
convertibles become the only affordable financing instruments.
Stripped bonds.
The cash flows from treasury bonds can be separated into two assets: an
asset corresponding to the principal amount, (principal only, PO), and an
asset corresponding to interest rates, (interest only, IO). This type of bond re-
presents a stripped asset and is referred to as treasury-backed stripped. The
amount of principal, PO, represents the value of a zero coupon bond. The
amount of interest, IO, corresponds to a portfolio of zero coupon bonds.
The separation between the cash flows can eliminate the reinvestment risk. It
allows the investor to use different IO and PO in hedging strategies.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Strips.
Correspond to an instrument called “Separate Trading of Registered
Interest and Principal of Securities”. The separation between coupons and
principal of a bond allows the creation of artificial zero-coupon bonds of
longer maturities than would otherwise exist.
If you consider an entity that purchases several loans, pools them, and
uses them as collateral for issuance of a security, the security created is
referred to as a mortgage pass-through security. The security is guaranteed
by the GNMA, the FNMA, and the FHLMC.
Pass-through securities can be issued also by private entities. In this
case, they are referred to as conventional pass-throughs.
When mortgage loans are used in a pool to create a pass-through
security, they are said to be securitized. The process of creating the pass-
through security is known as the securitization of mortgage loans.
The FHLMC introduced in 1983, the collaterized mortgage obligations,
(CMO). Since an investor in a pass-through security is exposed to the total
pre-payment risk due to the pool of mortgage loans underlying the security,
it is possible to create three classes of bonds with different par values. This
can be done by indicating how the principal is distributed from the pass-
through security. In general, there are three classes: A, B, and C. This
mortgage-backed security refers to a CMO.
The total pre-payment risk for the CMO remains similar to that of
the mortgage loans. The stripped MBS becomes an attractive instrument
in managing portfolios of mortgage securities. It is possible to forecast the
prepayments from a pass-through security. Therefore, some pre-payment
benchmark conventions must be known. In general, the Standard pre-
payment model, PSA developed by the public securities association can be
used. This benchmark is expressed as a monthly series of annual constant
pre-payment rates, CPRs. The CPR is converted into a monthly pre-
payment rate, known as the single monthly mortality rate (SMM) where
SMM = 1 − (1 − CPR)1/12 .
The PSA model assumes that pre-payment rates will be low for newly
originated mortgages. The rate will speed up as the mortgages become
seasoned. For more details, see Fabozzi (1993).
margin. The margin varies with the variation in the futures price. The
futures contract is marked to market at the end of each trading day and is
subject to interim cash flows. The main difference between futures contracts
and forward contracts is that forward contracts are OTC instruments which
are nonstandardized and are subject to counter-party risk. There are several
traded interest rate futures contracts. Interest rate futures contracts are
traded on treasury bonds, notes, bills and on the LIBOR rate.
Interest rate futures options are traded on T-bond futures, T-note
futures, Eurodollar futures, etc.
where:
D = difference between the face value and the price of a bill maturing in
t days, known also as a dollar discount;
F = face or nominal value and
T = number of days remaining to maturity.
The treasury bill futures contract is quoted in terms of an index associated
to the yield as follows: Index = 100 − (Yd )(100).
Eurodollar futures
Eurodollar represent the liabilities of banks outside the United States of
America. The London Interbank offered rate, LIBOR is paid in Eurodollars.
The underlying asset of the Eurodollar futures contract is the three-month
Eurodollar. The contract is settled in cash.
the contract. If he decides to make delivery, the seller must deliver some
treasury bond chosen from the list of specific bonds published by the CBT.
The delivery process allows the seller of the futures contract to choose
from one of the acceptable deliverable treasury bonds.
The CBT uses conversion factors for the computation of the invoice
price of each deliverable treasury. This factor is determined before a con-
tract with a given settlement date begins trading and it remains constant.
The invoice price indicates the price paid by the buyer when the
treasury bond is delivered. It corresponds to the settlement futures price
plus accrued interest and is calculated as follows:
Accrued interest
Bond market prices are clean prices since they are quoted without any
accrued interest. The accrued interest corresponds to the amount of interest
since the payment of the last coupon. It is computed as follows:
Accrued interest = interest due in full period (N1 /N2 )
with
N1 = number of days since the last coupon date and
N2 = number of days between coupon payments.
The dirty price corresponds to the quoted clean price plus the accrued
interest. Upon delivery, the seller will deliver the bond which is cheapest
to deliver, also known as the cheapest to deliver (CTD). The seller must
compute the return to be earned from buying bonds and delivering them at
the settlement date. The return is computed using the price of the treasury
issue and the futures price for delivery. This return is referred to as the
implied repo rate. The CTD issue corresponds to the issue with the highest
implied repo rate since it gives the seller the highest return by buying and
delivering the issue.
The delivery process gives the contract seller some options.
The quality option, also known as the swap option, allows the seller to
choose among different acceptable treasury issues.
The timing option gives the seller the right to choose the exact time
during the delivery month to make delivery. The wildcard option allows
the seller to give a notice of intent to deliver up to 8 p.m. Chicago time
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
after the closing of the exchange (3:15 p.m.) on the date when the futures
settlement price is scheduled.
Since there are many bonds that can be delivered in the T-bond futures
contract, the CTD is that deliverable issue for which the following difference
is minimized:
The basis or the difference between the spot and futures prices is minimal
for the CTD bond or:
bit = Bci − ft CF i
where:
Bci = current price of the ith deliverable bond;
ft = bond futures settlement price and
CF i = conversion factor for the ith bond.
the bondholder can be different from the return anticipated when buying
the bond.
Pay-later options
For these options, the premium is paid upon exercise. They are contingent
options. In fact, the buyer has the obligation to pay upon exercise when the
option is in the money regardless of the amount by which the underlying
asset price exceeds the strike price.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Chooser options
They are on the holder, immediately after a pre-determined elapsed time,
to choose whether the option is to be a call or a put. There are two kinds
of chooser options: simple and complex choosers.
Asian options
Asian options have been in popular in the foreign exchange market,
interest rate and commodity markets. These financial innovations are
traded in OTC markets and allow investors to accomplish several hedging
strategies.
Examples of these options include commodity-linked bond contracts
and average currency options. Commodity-linked bond contracts give the
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
right to the holder to receive the average value of the underlying commodity
over a certain period or the nominal value of the bond, whichever is higher.
Barrier options
These belong to the family of path-dependent options. They are in life
when they knock-in and are extinguished when they knock-out. They are
sometimes referred to as knock-ins or knock-outs when the underlying asset
hits (or does not hit) the barrier.
The standard form of barrier options refers to European options which
appear or disappear (ins and outs) when the underlying asset reaches a
certain level known as the barrier. This barrier or knock-out level is set
below the strike price for the call and above it for the put. For example, an
in barrier option comes into existence whenever the underlying asset value
hits a specified level. The right to exercise an out barrier option is forfeited
when the barrier is hit.
Ratio options
These are options on the ratio of two asset prices, index levels, commodities,
etc. An example is given by the dollar-denominated European option on the
ratio of the Germain DAX stock index to the French CAC index.
Innovations in OTC options markets not only involve certain relations
between the underlying asset price and the strike price but also on the
number of time units for which a certain condition is satisfied. This
corresponds, for example, to financial assets which are traded within a
specified range.
Structured products with embedded digitals are much more interesting
than vanilla digitals. There are many types of range structures which may
be in the form of range binaries, at maturity range binaries, rebate range
binaries, mandarin collars, mega-premium options, limit binary options,
boundary options, corridors, wall options, mini-premium options, volatility
options, etc.
North America.
U.S index options trading appear on listed markets and OTC markets with
customized features.
Options are traded on SP 100, SP 500, MMI, SPMidCap, options on
small capitalization indexes, the NYSE Composite index.
Main information used concerns the average daily volume, Average
daily dollar volume (in millions) and Index level.
Options on SP are preferred by retail investors.
MidCap Options and options on SP 500 index attract the interest of
institutional money managers and pension funds.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Japan.
Options exist on Osaka Nikkei, options on TOPIX.
Japanese institutions often use for their long term options exposure or
customized strike prices fixed income securities with embedded index options.
Osaka Nikkei options are used by domestic institutional in short term
trading. Regulations by the Ministry of Finance prevent pension funds from
completely hedging their portfolios (hedging limit 50%).
Hedgers integrated their activities into equity risk management
systems.
Life insurance companies focus on using options for directional trading.
Offshore hedge funds use the Osaka Nikkei options to take outright short-
term trading positions.
The Government intervenes to support the market. Foreign institutions
act in the OTC market for different reasons:
They are restricted by regulation from trading listed options.
They do not want to incur the costs of rolling over.
Competition among dealers makes this market very competitive.
Sector options are popular in Japan. The following Table provides the
volume (number of contracts traded) in several countries for index options.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
American SE 40,985,108 33,137,709 123.68%
BM&F 89,965 0
Bourse de Montreal 336,544 961,650 35.00%
Chicago Board of Trade (CBOT) 762,007 263,629 289.05%
Chicago Board Options Exchange (CBOE) 136,679,303 110,822,096 123.33%
Chicago Mercantile Exchange (CME) 6,451,862 5,168,914 124.82%
International Securities Exchange (ISE) 40,886,923 23,979,352 170.51%
MexDer 35,989 0
New York Board of Trade (NYBOT) 181,215 110,079 164.62%
Options Clearing Corp. 0 0
Pacific SE 14,119,270 15,744,139 89.68%
Philadelphia SE 25,360,908 19,746,264 128.43%
Sao Paulo SE 1,589,765 1,600,461 99.33%
Asia Pacific
Australian SE 794,121 630,900 125.87%
BSE, The SE Mumbai 56,046 43 130339.53%
Hong Kong Exchanges 2,133,708 2,150,923 99.20%
Korea Exchange 2,521,557,274 2,837,724,956 88.86%
National Stock Exchange India 2,812,109 1,332,417 211.05%
Osaka SE 16,561,365 14,958,334 110.72%
SFE Corp. 523,428 585,620 89.38%
Singapore Exchange 247,388 289,361 85.49%
TAIFEX 43,824,511 21,720,084 201.77%
Tokyo SE 17,643 98,137 17.98%
2005 2004
Exchange Volume Traded
(Number of Contracts)
Americas
American SE 8,678,564 7,290,157
BM&F 6,344 16,485
Bourse de Montreal 650,186 336,544
Chicago Board of Trade (CBOT) 728,349 762,007
Chicago Board Options Exchange (CBOE) 192,536,695 136,679,303
Chicago Mercantile Exchange (CME) 15,106,187 6,451,862
International Securities Exchange (ISE) 4,464,094 83,358
MexDer 37,346 35,989
New York Board of Trade (NYBOT) 217,334 181,215
Options Clearing Corp. 0 0
Philadelphia SE 6,234,567 5,275,701
Sao Paulo SE 2,257,756 1,589,765
Asia Pacific
Australian SE 1,163,260 794,121
Bombay SE 100 NA
Hong Kong Exchanges 3,367,228 2,133,708
Korea Exchange 2,535,201,693 2,521,557,274
National Stock Exchange India 10,140,239 2,812,109
Osaka SE 24,894,925 16,561,365
SFE Corp. 680,303 523,428
Singapore Exchange 157,742 247,388
TAIFEX 81,533,102 43,824,511
Tokyo SE 20,004 17,643
Europe.
In Germany.
Listed DAX options are done on a screen-based system.
Major players in this market are the large U.S and Continental
investment banks.
In France.
Listed CAC 40 options trade on the French options market where trading is
dominated by locals taking speculative positions and by large investment
banks.
Institutional users are French insurance companies and fund managers.
Players seek leveraged exposures on the market.
Guaranteed funds on the CAC 40 issued by French banks are popular
among retail investors. CAC 40 options are used as part of these products.
Major participants in the OTC market are large U.S and European
investment banks.
United Kingdom.
The market is dominated by major international banks and brokers.
Short-term maturities have the most liquidity. End-users are mainly U.K
institutions for hedging and guaranteed funds.
In OTC markets, the volume is also high because of greater liquidity
in the longer-dated contracts. There is flexibility in expiration dates.
Switzerland.
Options are traded on the SOFFEX in an electronic screen system. Active
participants are major Swiss and American Banks. End users are a mixture
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Netherlands.
This market is dominated by locals who service a retail base.
Users are mainly pension funds who hedge equity portfolios.
The index must be compiled using a specific method.
The weighting of the index can overweight smaller, domestically ori-
ented stocks and underweight larger, more internationally oriented stocks.
For example, stocks can be weighted using a market capitalization and the
maximum weight of a stock in the index will not exceed 10%. This puts a
cap on some stocks.
The following Table gives the notional value (value traded of stocks),
the open interest (positions opened and still not unwind) and the option
premiums for several countries.
The following Tables provide different information for several markets
and instruments. The reader can compare the different markets and
instruments using these Tables (source: World Federation of Exchanges).
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD millions) (Nber of Contracts) (USD millions)
Americas
American SE 186,994,609 193,086,271 45,779 42,238 NA NA 4,709,107 7,652,680 NA NA
Boston Options Exchange 92,260,125 77,582,231 NA NA NA NA NA NA NA NA
Bourse de Montreal 12,265,461 10,032,227 68,947 54,904 1,583,405 1,346,141 732,202 554,076 2,212 1,645
Buenos Aires SE 49,235,173 92,386,767 NA NA 1,654,931 1,605,194 NA NA 456 547
Chicago Board Options Exchange (CBOE) 390,657,577 275,646,980 1,960,297 1,264,511 187,953,281 151,157,355 25,792,792 16,820,556 98,751 61,220
International Securities Exchnage (ISE) 583,749,099 442,387,776 NA NA NA NA NA NA NA NA
MexDer 448,120 135,931 829 208 0 2,030 62 49 NA NA
Options Clearing Corp. 0 0 NA NA 220,032,992 181,694,503 NA NA NA NA
Pacific SE 196,586,356 144,780,498 NA NA NA NA NA NA NA NA
Philadelphia SE 265,370,986 156,222,383 89,732 49,318 8,846,285 8,379,867 15,843,704 7,190,023 89,732 49,318
Sao Paulo SE 285,699,806 266,362,631 513,350 392,331 1,833,555 1,824,504 6,542,663 5,777,709 9,746 7,909
Asia Pacific
Australian SE 20,491,483 21,547,732 303,986 270,423 1,766,513 1,678,335 1,474,017 1,418,149 11,501 9,057
Hong Kong Exchanges 18,127,353 8,772,393 88,371 41,784 2,533,807 1,021,913 399,129 241,785 2,477 1,334
Korea Exchange 1,195 3,655 41 11 50 NA NA 103 NA 0
National Stock Exchange India 5,214,191 5,224,485 44,479 40,260 21,549 24,181 4,478,610 4,550,367 1,254 1,100
Osaka SE 753,937 1,206,987 NA NA 22,541 79,610 4,064 5,454 186 293
TAIFEX 1,089,158 1,018,917 32 79 2,797 3,959 45,088 126,245 31 161
Tokyo SE 190,876 201,798 21 33 39,428 11,906 NA NA 21 33
NA : Not Available
- : Not Applicable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
MexDer 3,000 19,400 21 85 0 3,400 62 17
Asia Pacific
Australian SE 693,653 490,233 8,645 5,872 124,307 78,289 5,194 3,308
Bursa Malaysia Derivatives 958 - 4 - 0 - NA -
Hong Kong Exchanges 102,010 13,069 655 77 4,260 1,750 9,382 2,170
National Stock Exchange India 100,285,737 68,911,754 857,436 510,701 642,395 464,559 82,217,305 56,491,871
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD millions) (Nber of Contracts) (USD millions)
Americas
American SE 16% 10,050,680 8,678,564 18,801 6,922 NA NA 123,559 122,714 NA NA
BM&F 126% 228,254 101,003 4,401 3,135 106,601 38,382 749 466 NA NA
Bourse de Montréal 108% 57,974 27,897 3,477 1,527 1,691 4,813 4,620 1,648 70 141
Chicago Board of Trade (CBOT) -24% 551,190 728,349 NA NA 21,815 26,794 NA NA NA NA
Chicago Board Options Exchange (CBOE) 45% 279,005,803 192,536,695 17,791,735 11,541,513 37,749,429 29,381,746 11,479,090 7,432,423 212,207 141,437
Chicago Mercantile Exchange (CME) 81% 27,295,611 15,106,187 6,005,296 3,295,855 1,527,059 1,226,413 2,666,446 1,457,075 NA NA
International Securities Exchnage (ISE) 84% 8,212,419 4,464,094 NA NA NA NA NA NA NA NA
MexDer 215% 117,568 37,346 23,110 5,048 9,965 3,493 909 459 NA NA
New York Board of Trade (NYBOT) -27% 159,209 217,334 NA NA 9,163 10,904 NA NA NA NA
Philadelphia SE 22% 7,625,523 6,236,922 NA NA NA NA NA NA NA NA
Sao Paulo SE -19% 1,818,764 2,257,756 4,303 2,773 146,377 185,895 531,001 357,506 4,303 2,773
Australian SE -1% 1,820,804 1,844,059 108,058 94,089 137,643 193,239 80,637 602,125 2,056 813
Hong Kong Exchanges 46% 4,915,263 3,367,228 578,927 304,789 303,988 225,654 1,067,221 728,417 NA NA
Korea Exchange -5% 2,414,422,955 2,535,201,693 41,205,406 34,652,198 3,468,456 3,299,722 NA 87,656,989 152,013 137,847
National Stock Exchange India 84% 18,702,248 10,140,239 141,111 60,025 154,919 85,370 5,440,629 2,749,463 2,811 1,022
Osaka SE 13% 28,231,169 24,894,925 NA NA 695,661 1,160,453 1,598,319 1,109,841 24,032 12,943
Singapore Exchange 146% 387,673 157,742 26,111 10,750 35,458 27,620 NA NA NA NA
TAIFEX 22% 99,507,934 81,533,102 21,492 20,903 612,589 790,814 16,849,126 15,559,660 21,496 40,207
Tokyo SE -8% 18,354 20,004 2,352 2,102 2,176 3,550 NA NA 116 156
Athens -4% 670,583 700,094 9,674 7,745 11,345 10,820 74,996 73,200 161 135
BME Spanish 25% 5,510,621 4,407,465 83,268 52,421 1,235,886 892,188 227,616 86,390 2,347 1,316
Borsa Italiana 9% 2,819,916 2,597,830 331,662 259,612 153,854 120,680 645,422 576,503 3,250 2,802
Eurex 45% 217,232,549 149,380,569 9,556,257 5,273,496 32,928,972 24,866,988 NA NA 246,120 140,841
JSE 2% 11,801,030 11,605,030 13,859 7,696 1,343,735 1,512,486 13,699 10,550 NA NA
OMX 11% 13,613,210 12,229,145 185,555 147,261 985,614 973,817 NA NA 20,879 13,001
Oslo Børs 156% 1,320,651 515,538 NA NA 44,194 21,405 19,409 NA 176 114
Tel Aviv SE 20% 75,539,100 63,133,416 1,427,043 964,607 436,345 341,242 12,917,880 9,640,727 15,827 11,084
Warsaw SE 27% 316,840 250,060 3,055 1,758 4,347 6,432 117,266 83,834 46 22
140%
120%
100%
National Stock Exchange India
80%
60%
Hong Kong Exchanges
40%
TAIFEX
20% Osaka SE
Australian SE
0%
Korea Exchange Tokyo SE
-20% 0 2 4 6 8 10
Exchange
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
BM&F 16,940,891 6,683,525 293,433 207,990 178,243 301,558 1,464,734 803,605
Bourse de Montréal 3,098,659 2,258,404 370,621 245,880 166,640 110,405 1,743,005 1,025,432
Chicago Board of Trade (CBOT) 28,730,906 26,679,733 NA 1,501,704 167,040 97,208 NA NA
Chicago Mercantile Exchange (CME) 470,196,436 378,748,159 29,270,013 22,578,526 47,144,863 41,786,549 145,708,814 122,479,477
MexDer 620,557 410,565 132,292 61,413 30,959 22,130 33,238 24,244
New York Board of Trade (NYBOT) 860,539 922,099 NA NA 71,698 92,485 NA NA
Asia Pacific
Australian SE 6,652,323 5,713,161 613,940 451,370 268,488 175,546 1,459,407 1,155,276
Bursa Malaysia Derivatives 1,628,043 1,111,575 21,153 13,210 24,621 17,814 NA NA
Hong Kong Exchanges 19,747,246 13,393,462 2,014,834 987,256 185,262 136,465 9,443,472 6,338,836
Korea Exchange 46,696,151 43,912,281 4,283,838 2,982,607 91,200 83,418 NA 13,557,429
National Stock Exchange India 70,286,227 47,375,214 515,354 279,775 307,761 234,624 18,792,431 12,771,115
Osaka SE 31,661,331 18,070,352 3,560,096 2,068,205 388,666 409,588 3,025,602 949,211
Singapore Exchange 31,200,243 21,725,170 1,660,847 1,068,947 499,159 411,558 NA NA
TAIFEX 13,930,545 10,104,645 519,019 688,666 66,980 63,667 16,864,405 8,464,444
Thailand Futures Exchange (TFEX) 198,737 - 2,595 - 7,601 - 111,214 -
Tokyo SE 14,907,723 12,786,102 2,074,924 1,510,707 369,690 385,914 NA NA
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD millions) (Nber of Contracts) (USD millions)
Americas
BM&F 10,554,948 3,052,800 11,195 20,940 2,354,423 697,304 12,853 9,855 NA NA
Bourse de Montréal 605,806 377,370 535,720 311,501 78,861 44,375 2,084 1,476 92 76
Chicago Board of Trade (CBOT) 9,424,628 6,534,587 NA 32,672,935 1,130,942 927,916 NA NA NA NA
Chicago Board Options Exchange (CBOE) 2,594 4,381 13 14 343 317 288 577 1 1
Chicago Mercantile Exchange (CME) 268,957,139 188,001,096 268,957,127 188,001,090 18,808,764 16,325,364 1,140,562 951,078 NA NA
Asia Pacific
Australian SE 206,853 247,790 156,487 188,719 59,544 54,132 382 425 NA NA
Singapore Exchange 8,700 0 7,091 0 8,700 0 NA 0 NA 0
Tokyo Financial Exchange 3,976,697 41,204 3,418,070 37,171 481,355 32,500 NA NA NA NA
NA : Not Available
- : Not Applicable
Americas
BM&F 180,822,732 143,655,871 7,353,654 5,538,228 9,784,628 7,332,556 555,046 486,397
Bourse de Montréal 16,702,302 11,157,298 14,770,015 9,209,807 393,078 331,916 825,430 724,190
Chicago Board of Trade (CBOT) 17,833,331 11,602,282 NA 58,011,410 414,975 455,444 NA NA
Chicago Mercantile Exchange (CME) 503,729,899 411,706,656 505,339,873 413,781,671 9,564,114 8,596,023 60,357,744 52,168,804
MexDer 267,450,231 104,339,918 26,564,227 10,348,810 44,058,415 21,205,907 85,227 48,626
Asia Pacific
Australian SE 22,860,491 18,199,674 19,823,462 15,665,366 902,397 760,267 250,184 236,344
Bursa Malaysia Derivatives 272,502 162,592 74,545 42,963 59,831 37,966 NA NA
Hong Kong Exchanges 14,043 25,181 2,171 3,877 1,532 1,477 752 1,229
Korea Exchange 615 3,308 187 622 NA NA NA 163
Singapore Exchange 3,573,665 2,890,729 2,915,805 2,466,068 288,215 415,431 NA NA
TAIFEX 40 217 138 310 0 0 72 217
Tokyo Financial Exchange 31,495,084 10,977,591 27,070,811 9,903,104 2,326,719 1,418,937 NA NA
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD m illions) (Nber of Contracts) (USD millions)
Americas
Bourse de Montréal 2,275 7 202 0 0 2 25 NA 0 NA
Buenos Aires SE 8,437 86,036 NA NA 0 293 NA NA 1 5
Chicago Board of Trade (CBOT) 95,737,966 89,888,554 NA 8,931,116 3,097,170 2,517,698 NA NA NA NA
Chicago Board Options Exchange (CBOE) 18,736 61,245 92 265 2,038 7,465 1,318 5,203 3 13
Asia Pacific
Australian SE 3,086,456 2,307,659 235,067 175,753 14,733 1,729 11,078 10,494 NA NA
Singapore Exchange 0 725 0 308 NA NA NA NA NA NA
Tokyo SE 2,060,624 1,699,037 NA 2,120,602 16,987 22,939 NA NA 4,306 3,222
Americas
BM&F 67,301 16,172 4,214 1,484 1,731 181 1,102 307
Bourse de Montréal 7,777,098 4,824,924 695,280 398,274 337,120 166,504 1,005,657 772,125
Chicago Board of Trade (CBOT) 512,163,874 446,065,592 NA 46,723,075 5,035,467 3,614,314 NA NA
MexDer 500,479 284,460 52,437 27,750 43,450 2,101 2,584 1,402
Philadelphia SE 10 - NA - 0 - 10 -
Asia Pacific
Australian SE 45,121,853 36,255,583 3,413,538 2,761,260 872,581 593,812 671,133 655,235
Bursa Malaysia Derivatives 28,181 27,068 771 715 0 150 NA NA
Hong Kong Exchanges 0 1,250 0 169 NA NA 0 50
Korea Exchange 10,346,884 11,223,811 1,180,451 1,208,118 112,652 81,407 NA 1,836,163
Singapore Exchange 1,427,462 1,241,852 116,352 105,758 40,186 27,645 NA NA
TAIFEX 40,675 2,887 6,745 1,045 258 22 51,878 2,348
Tokyo Financial Exchange 13,680 78,943 1,176 7,122 300 1,450 NA NA
Tokyo SE 12,149,979 9,844,617 10,357,258 8,881,026 131,772 116,664 NA NA
NA : Not Available
- : Not Applicable
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD m illions) (Nber of Contracts) (USD millions)
Americas
Bourse de Montréal 31,262 7,264 277 70 2,838 2,691 2,010 466 3 1
BM&F 10,525,832 6,850,041 44,173 36,604 927,188 799,576 30,110 28,340 NA NA
Chicago Mercantile Exchange (CME) 3,289,498 3,182,525 451,686 440,565 230,426 228,288 682,415 608,974 NA NA
MexDer 306 0 34 0 2 0 9 0 NA 0
New York Board of Trade (NYBOT) 44,322 35,970 NA NA 3,690 1,778 NA NA NA NA
Options Clearing Corp. 0 0 NA NA 10,602 17,330 NA NA NA NA
Philadelphia SE 131,508 159,748 149 166 10,476 17,213 6,370 8,861 149 166
NA : Not Available
- : Not Applicable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
BM&F 1,726,351 1,293,181 2,738,810 1,737,251 677,724 475,755 1,726,351 1,293,181
Buenos Aires SE 800 2,416 1 2 NA NA NA NA
Chicago Mercantile Exchange (CME) 110,338,043 81,105,391 13,399,645 9,798,906 1,098,880 711,360 65,453,858 53,154,207
ROFEX 17,936,247 12,932,275 NA NA 196,293 323,169 NA NA
MexDer 6,077,409 2,934,783 670,393 323,969 248,205 134,992 4,415 2,785
New York Board of Trade (NYBOT) 3,653,024 3,604,877 NA NA 149,595 127,497 NA NA
Asia Pacific
Australian SE 1,363 4,422 103 337 0 37 370 966
Korea Exchange 3,158,049 2,667,005 158,463 133,679 160,722 85,520 NA 633,614
Tokyo Financial Exchange 0 600 0 5 NA NA NA NA
NA : Not Available
- : Not Applicable
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of contracts) (USD millions) (Nber of contracts) (USD millions)
Americas
BM&F 177,719 195,103 194 194 12,541 5,999 1,354 1,560 NA NA
Chicago Board of Trade (CBOT) 21,861,340 16,353,965 NA 304,650 2,177,795 900,266 NA NA NA NA
Chicago Mercantile Exchange (CME) 2,010,226 943,377 67,569 34,006 307,489 116,431 470,806 389,526 NA NA
Mercado a Término de Buenos Aires 2,815,000 2,091,500 NA NA NA NA NA NA NA NA
New York Board of Trade (NYBOT) 11,662,056 8,663,470 NA 220,560 1,146,100 928,436 NA NA NA NA
NYMEX 54,468,396 38,002,895 NA 2,193,391 9,297,986 NA NA NA NA NA
ROFEX 34,815 59,475 NA NA 6,039 4,706 NA NA NA NA
Asia Pacific
Australian SE 10,683 558 380 72 21,264 369 488 49 NA NA
Tokyo Grain Exchange 27,262 27,101 NA 42 409 288 284 49 NA NA
NA : Not Available
- : Not Applicable
Americas
BM&F 1,318,203 1,073,471 12,436 10,106 63,964 50,996 219,847 214,293
Chicago Board of Trade (CBOT) 118,719,938 76,786,994 NA 1,293,074 2,821,951 1,732,853 NA NA
Chicago Mercantile Exchange (CME) 17,448,155 11,558,317 613,145 394,707 536,649 387,575 5,079,223 4,212,551
Mercado a Término de Buenos Aires 11,899,472 11,502,296 NA NA NA NA NA NA
New York Board of Trade (NYBOT) 28,233,129 24,486,440 NA 500,155 1,065,666 901,038 NA NA
NYMEX 178,929,185 166,608,642 NA 8,893,687 9,326,151 NA NA NA
ROFEX 116,937 118,973 NA NA 11,984 10,409 NA NA
Asia Pacific
Australian SE 185,349 36,481 3,321 1,160 55,600 18,010 12,295 6,150
Bursa Malaysia Derivatives 2,230,340 1,158,510 48,051 21,313 74,567 28,918 NA NA
Central Japan Com modity Exchange 9,019,416 33,179,422 NA 1,943,220 117,816 182,304 NA NA
Dalian Commodity Exchange 117,681,038 99,174,714 NA 622,949 1,154,982 482,979 NA NA
Korea Exchange 3,158,049 2,667,005 158,463 133,679 160,722 85,520 NA NA
Shanghai Futures Exchange 58,106,001 33,789,754 NA 515,274 196,219 154,723 NA NA
TAIFEX 35,027 0 2,206 0 44 0 12,724 0
Tokyo Grain Exchange 19,106,247 25,573,238 1,302,452 406,973 438,435 563,665 NA NA
Zhengzhou Commodity Exchange 46,298,117 28,472,570 NA 16,166 213,847 452,058 NA NA
NA : Not Available
- : Not Applicable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Summary
The last three decades have witnessed a proliferation of financial innova-
tions. Roughly speaking, financial innovations seem to belong to two classes.
First, there are the new securities and their markets such as traded
and OTC equity and interest rate derivative assets. Second, there are
dynamic trading strategies using these instruments. Traded derivative
assets are standardized contracts which are listed on options exchanges.
OTC derivative assets are tailor-made to the investor’s needs and are often
written by investment banks. Examples of classic or standard financial
assets and commodity contracts include forward rate contracts, futures
contracts, swaps, standard calls and puts, traded stock options, equity
warrants, covered warrants, options on equity indices, options on index
futures contracts, options on currency forwards or currency futures and
bond options. Futures and options market enable investors to manage
price risk. The market offers an environment that allows all users to
control the price risk. The prices of these financial instruments are fully
transparent because they are updated second by second as trading occurs.
Examples of commodity contracts are oil and cocoa. The oil market is
ultimately concerned with the transportation, processing, and storage of a
raw material. Crude oil is traded on world markets using the spot asset,
physical forward contracts, futures contracts, options on futures contracts,
swaps, warrants, etc. Price information can be obtained from oil and energy
pricing information such as Reuters, Bridge Telerate, Platt’s, etc. The
size and complexity of global crude oil trade are unique among physical
commodities. Worldwide crude oil trade in the last 30 years has gone
through revolutionary changes that have had large political and economic
impact adding to its uniqueness. Each crude oil from each field is unique in
quality. The trading instuments apply to some crudes including West Texas
Intermediate (WTI), Dubai, Alaska North Slope (ANS) and Brent blend.
Each of these crudes or blends define its specific oil market. However, the
markets are linked together through arbitrage. The Brent market includes
partial forward transactions, a futures contract traded in London at the
International Petroleum Exchange (IPE), options on this contract and swap
deals. The history of cocoa dates back to the 6th century with its origins in
the Amazon Basin. It was first brought to Europe in the 17th century
as a luxury drink. Market users include the international cocoa trade,
cocoa processors and chocolate manufacturers, managed futures funds,
institutional investors and options specialists. Full cocoa-related statistics
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Questions
1. What are the specific features of options?
2. What are the specific features of futures and forward contracts?
3. What are the trading characteristics of commodity contracts?
4. What are the specific features of the main instruments traded on the
International Petroleum Exchange?
5. Describe the specific features of the cocoa market.
6. Describe the specific features of equity options.
7. Describe the specific features of options on currency forwards and
futures.
8. Describe the specific features of bonds and bond options markets.
9. Provide some examples of simple and complex financial instruments.
10. Why there are so many new financial instruments?
11. What are the fundamental reasons behind the proliferation of financial
assets?
12. Why has the wave of financial innovation not stopped?
Exercises
1. Explain how an investor uses options.
• Options are easily bought and sold.
• Holders can sell or exercise their options at any time.
• Most options are traded on exchanges and/or on over the counter.
• At maturity, holders of physical options exercise into actual shares.
• Holders of cash-settled options choose to sell their options.
• They involve the purchaser in completing options counterparty documen-
tation.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
A call gives the buyer the right, but not the obligation to exercise,
and thereby receive in cash or physical delivery any amount by which the
underlying asset is above the strike price.
A put gives the buyer the right, but not the obligation to exercise,
and thereby receives in cash or physical delivery any amount by which the
underlying asset is below the strike price.
European options are only exercised at maturity.
American options are exercised at any time before maturity.
The seller, on the other hand, must post margin with the clearing house
as a performance bond or calculated by a formula based on the relationship
of the asset price and the strike price.
The seller may be required to deposit additional margin if his position
moves against him.
Options can be exercised in cash against the closing price of the
underlying asset (compared with the strike price).
They can be exercised also with the requirement of transferring actual
shares of stock.
Investors can contact brokers and the exchanges to obtain the current
specifications.
• Time duration;
• Good-Til Canceled (or open); and
• Opening only.
Appendix
Derivatives Markets in the World Before and During the
Financial Crisis
Stock Options, Index Options, Interest Rate and
Commodity Options and Futures Markets
Global overview
Several institutions produce information regarding futures and options
around the world.
Often, summary statistics on volume and open interest are given for
futures and index options.
Index options on stock indexes and index futures contracts begin
trading in the U.S in 1983. This has been facilitated with the introduction
of the SP 100 index contract on the Chicago Board Options Exchange.
Today, index futures are traded and are more liquid than index options.
North America.
U.S index options trading appear on listed markets and OTC markets with
customized features.
Options are traded on SP 100, SP 500, MMI, SPMidCap, options on
small capitalization indexes, the NYSE Composite index.
Main information used concerns the average daily volume, Average
daily dollar volume (in millions) and Index level.
Options on SP are preferred by retail investors.
MidCap Options and options on SP 500 index attract the interest of
institutional money managers and pension funds.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Japan.
Options exist on Osaka Nikkei, options on TOPIX.
Japanese institutions often use for their long term options exposure
or customized strike prices fixed income securities with embedded index
options.
Osaka Nikkei options are used by domestic institutional in short term
trading. Regulations by the Ministry of Finance prevent pension funds from
completely hedging their portfolios (hedging limit 50%).
Hedgers integrated their activities into equity risk management
systems.
Life insurance companies focus on using options for directional trading.
Offshore hedge funds use the Osaka Nikkei options to take outright short-
term trading positions.
The Government intervenes to support the market. Foreign institutions
act in the OTC market for different reasons:
They are restricted by regulation from trading listed options.
They do not want to incur the costs of rolling over.
Competition among dealers makes this market very competitive.
Sector options are popular in Japan. The following Table provides the
volume (number of contracts traded) in several countries for index options.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
American SE 40,985,108 33,137,709 123.68%
BM&F 89,965 0
Bourse de Montreal 336,544 961,650 35.00%
Chicago Board of Trade (CBOT) 762,007 263,629 289.05%
Chicago Board Options Exchange (CBOE) 136,679,303 110,822,096 123.33%
Chicago Mercantile Exchange (CME) 6,451,862 5,168,914 124.82%
International Securities Exchange (ISE) 40,886,923 23,979,352 170.51%
MexDer 35,989 0
New York Board of Trade (NYBOT) 181,215 110,079 164.62%
Options Clearing Corp. 0 0
Pacific SE 14,119,270 15,744,139 89.68%
Philadelphia SE 25,360,908 19,746,264 128.43%
Sao Paulo SE 1,589,765 1,600,461 99.33%
Asia Pacific
Australian SE 794,121 630,900 125.87%
BSE, The SE Mumbai 56,046 43 130339.53%
Hong Kong Exchanges 2,133,708 2,150,923 99.20%
Korea Exchange 2,521,557,274 2,837,724,956 88.86%
National Stock Exchange India 2,812,109 1,332,417 211.05%
Osaka SE 16,561,365 14,958,334 110.72%
SFE Corp. 523,428 585,620 89.38%
Singapore Exchange 247,388 289,361 85.49%
TAIFEX 43,824,511 21,720,084 201.77%
Tokyo SE 17,643 98,137 17.98%
2005 2004
Exchange Volume Traded
(Nber of Contracts)
Americas
American SE 8,678,564 7,290,157
BM&F 6,344 16,485
Bourse de Montreal 650,186 336,544
Chicago Board of Trade (CBOT) 728,349 762,007
Chicago Board Options Exchange (CBOE) 192,536,695 136,679,303
Chicago Mercantile Exchange (CME) 15,106,187 6,451,862
International Securities Exchange (ISE) 4,464,094 83,358
MexDer 37,346 35,989
New York Board of Trade (NYBOT) 217,334 181,215
Options Clearing Corp. 0 0
Philadelphia SE 6,234,567 5,275,701
Sao Paulo SE 2,257,756 1,589,765
Asia Pacific
Australian SE 1,163,260 794,121
Bombay SE 100 NA
Hong Kong Exchanges 3,367,228 2,133,708
Korea Exchange 2,535,201,693 2,521,557,274
National Stock Exchange India 10,140,239 2,812,109
Osaka SE 24,894,925 16,561,365
SFE Corp. 680,303 523,428
Singapore Exchange 157,742 247,388
TAIFEX 81,533,102 43,824,511
Tokyo SE 20,004 17,643
Europe.
In Germany.
Listed DAX options are done on a screen-based system.
Major players in this market are the large U.S and Continental
investment banks.
In France.
Listed CAC 40 options trade on the French options market where trading is
dominated by locals taking speculative positions and by large investment
banks.
Institutional users are French insurance companies and fund managers.
Players seek leveraged exposures on the market.
Guaranteed funds on the CAC 40 issued by French banks are popular
among retail investors. CAC 40 options are used as part of these products.
Major participants in the OTC market are large U.S and European
investment banks.
United Kingdom.
The market is dominated by major international banks and brokers.
Short-term maturities have the most liquidity. End-users are mainly U.K
institutions for hedging and guaranteed funds.
In OTC markets, the volume is also high because of greater liquidity
in the longer-dated contracts. There is flexibility in expiration dates.
Switzerland.
Options are traded on the SOFFEX in an electronic screen system. Active
participants are major Swiss and American Banks. End users are a mixture
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Netherlands.
This market is dominated by locals who service a retail base.
Users are mainly pension funds who hedge equity portfolios.
The index must be compiled using a specific method.
The weighting of the index can overweight smaller, domestically ori-
ented stocks and underweight larger, more internationally oriented stocks.
For example, stocks can be weighted using a market capitalization and the
maximum weight of a stock in the index will not exceed 10%. This puts a
cap on some stocks.
The following table gives the notional value (value traded of stocks),
the open interest (positions opened and still not unwind) and the option
premiums for several countries.
The following tables provide different information for several markets
and instruments. The reader can compare the different markets and
instruments using these tables (source: World Federation of Exchanges).
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD millions) (Nber of Contracts) (USD millions)
Americas
American SE 186,994,609 193,086,271 45,779 42,238 NA NA 4,709,107 7,652,680 NA NA
Boston Options Exchange 92,260,125 77,582,231 NA NA NA NA NA NA NA NA
Bourse de Montreal 12,265,461 10,032,227 68,947 54,904 1,583,405 1,346,141 732,202 554,076 2,212 1,645
Buenos Aires SE 49,235,173 92,386,767 NA NA 1,654,931 1,605,194 NA NA 456 547
Chicago Board Options Exchange (CBOE) 390,657,577 275,646,980 1,960,297 1,264,511 187,953,281 151,157,355 25,792,792 16,820,556 98,751 61,220
International Securities Exchnage (ISE) 583,749,099 442,387,776 NA NA NA NA NA NA NA NA
MexDer 448,120 135,931 829 208 0 2,030 62 49 NA NA
Options Clearing Corp. 0 0 NA NA 220,032,992 181,694,503 NA NA NA NA
Pacific SE 196,586,356 144,780,498 NA NA NA NA NA NA NA NA
Philadelphia SE 265,370,986 156,222,383 89,732 49,318 8,846,285 8,379,867 15,843,704 7,190,023 89,732 49,318
Sao Paulo SE 285,699,806 266,362,631 513,350 392,331 1,833,555 1,824,504 6,542,663 5,777,709 9,746 7,909
Asia Pacific
Australian SE 20,491,483 21,547,732 303,986 270,423 1,766,513 1,678,335 1,474,017 1,418,149 11,501 9,057
Hong Kong Exchanges 18,127,353 8,772,393 88,371 41,784 2,533,807 1,021,913 399,129 241,785 2,477 1,334
Korea Exchange 1,195 3,655 41 11 50 NA NA 103 NA 0
National Stock Exchange India 5,214,191 5,224,485 44,479 40,260 21,549 24,181 4,478,610 4,550,367 1,254 1,100
Osaka SE 753,937 1,206,987 NA NA 22,541 79,610 4,064 5,454 186 293
TAIFEX 1,089,158 1,018,917 32 79 2,797 3,959 45,088 126,245 31 161
Tokyo SE 190,876 201,798 21 33 39,428 11,906 NA NA 21 33
NA : Not Available
- : Not Applicable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
MexDer 3,000 19,400 21 85 0 3,400 62 17
Asia Pacific
Australian SE 693,653 490,233 8,645 5,872 124,307 78,289 5,194 3,308
Bursa Malaysia Derivatives 958 - 4 - 0 - NA -
Hong Kong Exchanges 102,010 13,069 655 77 4,260 1,750 9,382 2,170
National Stock Exchange India 100,285,737 68,911,754 857,436 510,701 642,395 464,559 82,217,305 56,491,871
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD millions) (Nber of Contracts) (USD millions)
Americas
American SE 16% 10,050,680 8,678,564 18,801 6,922 NA NA 123,559 122,714 NA NA
BM&F 126% 228,254 101,003 4,401 3,135 106,601 38,382 749 466 NA NA
Bourse de Montréal 108% 57,974 27,897 3,477 1,527 1,691 4,813 4,620 1,648 70 141
Chicago Board of Trade (CBOT) -24% 551,190 728,349 NA NA 21,815 26,794 NA NA NA NA
Chicago Board Options Exchange (CBOE) 45% 279,005,803 192,536,695 17,791,735 11,541,513 37,749,429 29,381,746 11,479,090 7,432,423 212,207 141,437
Chicago Mercantile Exchange (CME) 81% 27,295,611 15,106,187 6,005,296 3,295,855 1,527,059 1,226,413 2,666,446 1,457,075 NA NA
International Securities Exchnage (ISE) 84% 8,212,419 4,464,094 NA NA NA NA NA NA NA NA
MexDer 215% 117,568 37,346 23,110 5,048 9,965 3,493 909 459 NA NA
New York Board of Trade (NYBOT) -27% 159,209 217,334 NA NA 9,163 10,904 NA NA NA NA
Philadelphia SE 22% 7,625,523 6,236,922 NA NA NA NA NA NA NA NA
Sao Paulo SE -19% 1,818,764 2,257,756 4,303 2,773 146,377 185,895 531,001 357,506 4,303 2,773
Australian SE -1% 1,820,804 1,844,059 108,058 94,089 137,643 193,239 80,637 602,125 2,056 813
Hong Kong Exchanges 46% 4,915,263 3,367,228 578,927 304,789 303,988 225,654 1,067,221 728,417 NA NA
Korea Exchange -5% 2,414,422,955 2,535,201,693 41,205,406 34,652,198 3,468,456 3,299,722 NA 87,656,989 152,013 137,847
National Stock Exchange India 84% 18,702,248 10,140,239 141,111 60,025 154,919 85,370 5,440,629 2,749,463 2,811 1,022
Osaka SE 13% 28,231,169 24,894,925 NA NA 695,661 1,160,453 1,598,319 1,109,841 24,032 12,943
Singapore Exchange 146% 387,673 157,742 26,111 10,750 35,458 27,620 NA NA NA NA
TAIFEX 22% 99,507,934 81,533,102 21,492 20,903 612,589 790,814 16,849,126 15,559,660 21,496 40,207
Tokyo SE -8% 18,354 20,004 2,352 2,102 2,176 3,550 NA NA 116 156
Athens -4% 670,583 700,094 9,674 7,745 11,345 10,820 74,996 73,200 161 135
BME Spanish 25% 5,510,621 4,407,465 83,268 52,421 1,235,886 892,188 227,616 86,390 2,347 1,316
Borsa Italiana 9% 2,819,916 2,597,830 331,662 259,612 153,854 120,680 645,422 576,503 3,250 2,802
Eurex 45% 217,232,549 149,380,569 9,556,257 5,273,496 32,928,972 24,866,988 NA NA 246,120 140,841
JSE 2% 11,801,030 11,605,030 13,859 7,696 1,343,735 1,512,486 13,699 10,550 NA NA
OMX 11% 13,613,210 12,229,145 185,555 147,261 985,614 973,817 NA NA 20,879 13,001
Oslo Børs 156% 1,320,651 515,538 NA NA 44,194 21,405 19,409 NA 176 114
Tel Aviv SE 20% 75,539,100 63,133,416 1,427,043 964,607 436,345 341,242 12,917,880 9,640,727 15,827 11,084
Warsaw SE 27% 316,840 250,060 3,055 1,758 4,347 6,432 117,266 83,834 46 22
140%
120%
100%
National Stock Exchange India
80%
60%
Hong Kong Exchanges
40%
TAIFEX
20% Osaka SE
Australian SE
0%
Korea Exchange Tokyo SE
-20% 0 2 4 6 8 10
Exchange
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
BM&F 16,940,891 6,683,525 293,433 207,990 178,243 301,558 1,464,734 803,605
Bourse de Montréal 3,098,659 2,258,404 370,621 245,880 166,640 110,405 1,743,005 1,025,432
Chicago Board of Trade (CBOT) 28,730,906 26,679,733 NA 1,501,704 167,040 97,208 NA NA
Chicago Mercantile Exchange (CME) 470,196,436 378,748,159 29,270,013 22,578,526 47,144,863 41,786,549 145,708,814 122,479,477
MexDer 620,557 410,565 132,292 61,413 30,959 22,130 33,238 24,244
New York Board of Trade (NYBOT) 860,539 922,099 NA NA 71,698 92,485 NA NA
Asia Pacific
Australian SE 6,652,323 5,713,161 613,940 451,370 268,488 175,546 1,459,407 1,155,276
Bursa Malaysia Derivatives 1,628,043 1,111,575 21,153 13,210 24,621 17,814 NA NA
Hong Kong Exchanges 19,747,246 13,393,462 2,014,834 987,256 185,262 136,465 9,443,472 6,338,836
Korea Exchange 46,696,151 43,912,281 4,283,838 2,982,607 91,200 83,418 NA 13,557,429
National Stock Exchange India 70,286,227 47,375,214 515,354 279,775 307,761 234,624 18,792,431 12,771,115
Osaka SE 31,661,331 18,070,352 3,560,096 2,068,205 388,666 409,588 3,025,602 949,211
Singapore Exchange 31,200,243 21,725,170 1,660,847 1,068,947 499,159 411,558 NA NA
TAIFEX 13,930,545 10,104,645 519,019 688,666 66,980 63,667 16,864,405 8,464,444
Thailand Futures Exchange (TFEX) 198,737 - 2,595 - 7,601 - 111,214 -
Tokyo SE 14,907,723 12,786,102 2,074,924 1,510,707 369,690 385,914 NA NA
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD millions) (Nber of Contracts) (USD millions)
Americas
BM&F 10,554,948 3,052,800 11,195 20,940 2,354,423 697,304 12,853 9,855 NA NA
Bourse de Montréal 605,806 377,370 535,720 311,501 78,861 44,375 2,084 1,476 92 76
Chicago Board of Trade (CBOT) 9,424,628 6,534,587 NA 32,672,935 1,130,942 927,916 NA NA NA NA
Chicago Board Options Exchange (CBOE) 2,594 4,381 13 14 343 317 288 577 1 1
Chicago Mercantile Exchange (CME) 268,957,139 188,001,096 268,957,127 188,001,090 18,808,764 16,325,364 1,140,562 951,078 NA NA
Asia Pacific
Australian SE 206,853 247,790 156,487 188,719 59,544 54,132 382 425 NA NA
Singapore Exchange 8,700 0 7,091 0 8,700 0 NA 0 NA 0
Tokyo Financial Exchange 3,976,697 41,204 3,418,070 37,171 481,355 32,500 NA NA NA NA
NA : Not Available
- : Not Applicable
Americas
BM&F 180,822,732 143,655,871 7,353,654 5,538,228 9,784,628 7,332,556 555,046 486,397
Bourse de Montréal 16,702,302 11,157,298 14,770,015 9,209,807 393,078 331,916 825,430 724,190
Chicago Board of Trade (CBOT) 17,833,331 11,602,282 NA 58,011,410 414,975 455,444 NA NA
Chicago Mercantile Exchange (CME) 503,729,899 411,706,656 505,339,873 413,781,671 9,564,114 8,596,023 60,357,744 52,168,804
MexDer 267,450,231 104,339,918 26,564,227 10,348,810 44,058,415 21,205,907 85,227 48,626
Asia Pacific
Australian SE 22,860,491 18,199,674 19,823,462 15,665,366 902,397 760,267 250,184 236,344
Bursa Malaysia Derivatives 272,502 162,592 74,545 42,963 59,831 37,966 NA NA
Hong Kong Exchanges 14,043 25,181 2,171 3,877 1,532 1,477 752 1,229
Korea Exchange 615 3,308 187 622 NA NA NA 163
Singapore Exchange 3,573,665 2,890,729 2,915,805 2,466,068 288,215 415,431 NA NA
TAIFEX 40 217 138 310 0 0 72 217
Tokyo Financial Exchange 31,495,084 10,977,591 27,070,811 9,903,104 2,326,719 1,418,937 NA NA
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD m illions) (Nber of Contracts) (USD millions)
Americas
Bourse de Montréal 2,275 7 202 0 0 2 25 NA 0 NA
Buenos Aires SE 8,437 86,036 NA NA 0 293 NA NA 1 5
Chicago Board of Trade (CBOT) 95,737,966 89,888,554 NA 8,931,116 3,097,170 2,517,698 NA NA NA NA
Chicago Board Options Exchange (CBOE) 18,736 61,245 92 265 2,038 7,465 1,318 5,203 3 13
Asia Pacific
Australian SE 3,086,456 2,307,659 235,067 175,753 14,733 1,729 11,078 10,494 NA NA
Singapore Exchange 0 725 0 308 NA NA NA NA NA NA
Tokyo SE 2,060,624 1,699,037 NA 2,120,602 16,987 22,939 NA NA 4,306 3,222
Americas
BM&F 67,301 16,172 4,214 1,484 1,731 181 1,102 307
Bourse de Montréal 7,777,098 4,824,924 695,280 398,274 337,120 166,504 1,005,657 772,125
Chicago Board of Trade (CBOT) 512,163,874 446,065,592 NA 46,723,075 5,035,467 3,614,314 NA NA
MexDer 500,479 284,460 52,437 27,750 43,450 2,101 2,584 1,402
Philadelphia SE 10 - NA - 0 - 10 -
Asia Pacific
Australian SE 45,121,853 36,255,583 3,413,538 2,761,260 872,581 593,812 671,133 655,235
Bursa Malaysia Derivatives 28,181 27,068 771 715 0 150 NA NA
Hong Kong Exchanges 0 1,250 0 169 NA NA 0 50
Korea Exchange 10,346,884 11,223,811 1,180,451 1,208,118 112,652 81,407 NA 1,836,163
Singapore Exchange 1,427,462 1,241,852 116,352 105,758 40,186 27,645 NA NA
TAIFEX 40,675 2,887 6,745 1,045 258 22 51,878 2,348
Tokyo Financial Exchange 13,680 78,943 1,176 7,122 300 1,450 NA NA
Tokyo SE 12,149,979 9,844,617 10,357,258 8,881,026 131,772 116,664 NA NA
NA : Not Available
- : Not Applicable
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of Contracts) (USD m illions) (Nber of Contracts) (USD millions)
Americas
Bourse de Montréal 31,262 7,264 277 70 2,838 2,691 2,010 466 3 1
BM&F 10,525,832 6,850,041 44,173 36,604 927,188 799,576 30,110 28,340 NA NA
Chicago Mercantile Exchange (CME) 3,289,498 3,182,525 451,686 440,565 230,426 228,288 682,415 608,974 NA NA
MexDer 306 0 34 0 2 0 9 0 NA 0
New York Board of Trade (NYBOT) 44,322 35,970 NA NA 3,690 1,778 NA NA NA NA
Options Clearing Corp. 0 0 NA NA 10,602 17,330 NA NA NA NA
Philadelphia SE 131,508 159,748 149 166 10,476 17,213 6,370 8,861 149 166
NA : Not Available
- : Not Applicable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Americas
BM&F 1,726,351 1,293,181 2,738,810 1,737,251 677,724 475,755 1,726,351 1,293,181
Buenos Aires SE 800 2,416 1 2 NA NA NA NA
Chicago Mercantile Exchange (CME) 110,338,043 81,105,391 13,399,645 9,798,906 1,098,880 711,360 65,453,858 53,154,207
ROFEX 17,936,247 12,932,275 NA NA 196,293 323,169 NA NA
MexDer 6,077,409 2,934,783 670,393 323,969 248,205 134,992 4,415 2,785
New York Board of Trade (NYBOT) 3,653,024 3,604,877 NA NA 149,595 127,497 NA NA
Asia Pacific
Australian SE 1,363 4,422 103 337 0 37 370 966
Korea Exchange 3,158,049 2,667,005 158,463 133,679 160,722 85,520 NA 633,614
Tokyo Financial Exchange 0 600 0 5 NA NA NA NA
NA : Not Available
- : Not Applicable
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
Exchange Volume Traded Notional Value Open Interest Number of Trades Option Premium
(Nber of contracts) (USD millions) (Nber of contracts) (USD millions)
Americas
BM&F 177,719 195,103 194 194 12,541 5,999 1,354 1,560 NA NA
Chicago Board of Trade (CBOT) 21,861,340 16,353,965 NA 304,650 2,177,795 900,266 NA NA NA NA
Chicago Mercantile Exchange (CME) 2,010,226 943,377 67,569 34,006 307,489 116,431 470,806 389,526 NA NA
Mercado a Término de Buenos Aires 2,815,000 2,091,500 NA NA NA NA NA NA NA NA
New York Board of Trade (NYBOT) 11,662,056 8,663,470 NA 220,560 1,146,100 928,436 NA NA NA NA
NYMEX 54,468,396 38,002,895 NA 2,193,391 9,297,986 NA NA NA NA NA
ROFEX 34,815 59,475 NA NA 6,039 4,706 NA NA NA NA
Asia Pacific
Australian SE 10,683 558 380 72 21,264 369 488 49 NA NA
Tokyo Grain Exchange 27,262 27,101 NA 42 409 288 284 49 NA NA
NA : Not Available
- : Not Applicable
Americas
BM&F 1,318,203 1,073,471 12,436 10,106 63,964 50,996 219,847 214,293
Chicago Board of Trade (CBOT) 118,719,938 76,786,994 NA 1,293,074 2,821,951 1,732,853 NA NA
Chicago Mercantile Exchange (CME) 17,448,155 11,558,317 613,145 394,707 536,649 387,575 5,079,223 4,212,551
Mercado a Término de Buenos Aires 11,899,472 11,502,296 NA NA NA NA NA NA
New York Board of Trade (NYBOT) 28,233,129 24,486,440 NA 500,155 1,065,666 901,038 NA NA
NYMEX 178,929,185 166,608,642 NA 8,893,687 9,326,151 NA NA NA
ROFEX 116,937 118,973 NA NA 11,984 10,409 NA NA
Asia Pacific
Australian SE 185,349 36,481 3,321 1,160 55,600 18,010 12,295 6,150
Bursa Malaysia Derivatives 2,230,340 1,158,510 48,051 21,313 74,567 28,918 NA NA
Central Japan Com modity Exchange 9,019,416 33,179,422 NA 1,943,220 117,816 182,304 NA NA
Dalian Commodity Exchange 117,681,038 99,174,714 NA 622,949 1,154,982 482,979 NA NA
Korea Exchange 3,158,049 2,667,005 158,463 133,679 160,722 85,520 NA NA
Shanghai Futures Exchange 58,106,001 33,789,754 NA 515,274 196,219 154,723 NA NA
TAIFEX 35,027 0 2,206 0 44 0 12,724 0
Tokyo Grain Exchange 19,106,247 25,573,238 1,302,452 406,973 438,435 563,665 NA NA
Zhengzhou Commodity Exchange 46,298,117 28,472,570 NA 16,166 213,847 452,058 NA NA
NA : Not Available
- : Not Applicable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
References
Fabozzi, F (1993). Fixed Income Mathematics, US: Irwin.
Merton, RC (1998). Applications of option pricing theory: twenty five years later.
American Economic Review, 88(3), 323–345.
Miller, M (1986). Financial innovation: the last twenty years and the next. Journal
of Financial and Quantitative Analysis, 21 (December), 451–471.
Ross, S (1989). Financial markets, financial marketing and financial innovation.
Journal of Finance, 44(3), 541–556.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Chapter 2
Chapter Outline
This chapter is organized as follows:
67
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Introduction
The first of the modern commodity markets began trading a little over a
century ago. Today, futures markets are a direct development of traditional
agricultural markets, which were initially located in Chicago and London.
Chicago is the largest commodity trading center in the world. The
standardized futures markets such as the New York Merchantile Exchange
(Nymex), the International Petroleum Exchange (IPE), and the Singapore
International Monetary Exchange (Simex), or the forward markets like
dated Brent, Littlebrook Lottery, or the Russian Roulette have become
an important factor in the pricing of crude oil and refined products.
The futures price can be described using different parameters: the spot
price, the risk-less interest rate, the cost of carrying the stocks, and
the convenience yield. The convenience yield corresponds to a specific
interest rate of the commodity. Forward and futures contracts enable
firms to determine a price for future delivery. Forward and futures
prices can differ from the spot price of the commodity. However, as the
expiration date approaches, the forward, futures, and spot prices must
converge.
The cost of carry model corresponds to the relation between the futures
price and the spot price. It is the basis for the valuation of forward and
futures contracts. Futures prices and forward prices are often regarded as
being equivalent. However, this is true only if the risk-free interest rate
is constant or a known function of time. For the valuation of interest-
rate futures contracts, the theoretical futures price can be determined as
a function of the underlying asset price (the bond price), the coupon rate,
and the financing rate for borrowing and lending during a given period.
The fair price of a forward contract is given by the spot price plus the cost
of carry until the maturity date of the bond.
Futures markets date back to the medieval marketplaces, but they
developed in the United States in the 1800s in response to the nature of
agricultural products. In 1848, the Chicago Board of Trade became an
organized marketplace for grain transactions. Hedging is a price protection
that is used to minimize losses and to protect profits during the production,
storage, and marketing of commodities. Hedging is the strategy of taking a
position in the futures market as a temporary substitute for the purchase
or the sale of a commodity. In general, a perfect hedge is possible when
the “basis” or the relationship between the cash market and the futures
market is the same when the hedge is removed, as it was when the hedge
was implemented. A futures contract is, in general, liquidated by offsetting
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
strategies can be used for most of the derivative assets in this book, since
they can be implemented using options with any particular payoff.
(ST − F0 )eN r + F0 eN r = ST eN r
Since for a strategy in futures contracts no funds are invested, the result
ST eN r corresponds to the investment of F0 in the risk-free bond.
Another strategy can be constructed to give the same pay-off as the
preceding one. In fact, if f0 stands for the forward price at the end of day
0, then a strategy which consists in investing this amount in a risk-less
bond and an amount eN r in forward contracts also gives a final payoff
at T equal to ST eN r . Since the two strategies require an investment of
an amount F0 , (f0 ) and yield the same result, ST eN r , they must have the
same value in efficient capital markets. In the absence of profitable arbitrage
opportunities, the futures price F0 must be equal to the forward price f0 .
Hence, the proposed relation applies for both futures and forward prices
and we have F = f = SebT . Some examples are given below to illustrate
the use of this relationship in the determination of forward and futures
prices on some securities.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
F = 180e(0.07−0.06)0.25 = 180.45
0 = F + cT P − (P + rT P )
outlay or:
P rof it = P + rT P − (F + cT P )
0 = P + rT P − (F + cT P )
F = S + cp
In general, the futures contract is traded at a price which accounts for the
implicit options. Its price must satisfy the following relationship:
F = S + cp − O + e
where: F = futures price, S = spot price for the cheapest bond, cp = cost
of carry, O = value of the embedded options, and e = a white noise.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
The process for the pricing of a futures contract must account for the
following relationship:
F (f c) = S + cp − O + e
F = (S + cp − O)/(f c) + e
Backwardation: When spot prices are higher than long-term prices, any
hedge using a future maturity will be equivalent to a forward sale below the
spot price. This can lead to a loss if the market prices do not fall at the same
rate. Careful long-term analysis may provide good hedging opportunities.
Contango: When spot prices are lower than long-term prices, the producer
can sell the futures market at a higher price. So, he/she can fix his/her hedge
or future sales at a better price than the spot market. In this case, hedging
can generate profits if prices are not increasing at the same rate.
He/she expects a price rise and enters into a long futures contract to
protect his/her portfolio.
If the cash price of heating oil increases, the marketer will realize a
loss on the physical sale to customers because he/she will have to buy oil
at a higher price in the cash market to satisfy his/her sales commitment.
However, the cash loss would be offset by a futures profit on his/her futures
position.
If the oil price decreases, the resulting profit from the physical sale will
be offset by a loss on the futures market.
We give an example using IPE gasoil contracts and the Nymex heating-
oil contracts. IPE gasoil contracts are 100 tonnes and Nymex heating-oil
contracts are 1000 gallons. The relationship shows the trading of three
Nymex contracts for every four IPE contracts.
Besides, since IPE gasoil prices are quoted in $/tonne and Nymex
heating-oil prices are quoted in cents/gallon, a conversion factor must be
used. Assuming a specific gravity for gasoil of 0.845 kg/liter and since there
are 313 gallons of heating oil in a tonne, the conversion factor is 3.13.
Heating-oil arbitrage
Consider a trader who expects Nymex heating oil to move to a premium
over the IPE. He/she buys the Nymex heating-oil contracts and sells IPE
gasoil contracts.
The total profit of US$ 5.10/tonne comes from the change in the
differential regardless of the market direction (Table 2.4). Other types
of spreads can be implemented. The analysis can be extented to the
valuation of these contracts in the presence of information costs in
Appendices 1 and 2.
The financial crisis in 2008 reveals the importance of hedging strategies
in a risk framework.
The two following main relationships apply for the European and
American calls (c and C, respectively) and European and American puts
payoffs (p and P , respectively) at maturity:
where K is the strike price and T is the option’s maturity date. The
boundary space refers to the largest range of possible option prices before
expiration.
The call payoff shows that the lower the strike price, the higher is
the call value. This property can be shown using no-arbitrage arguments.
Hence, if you consider two calls with different strike prices, then the call
with the lower strike price must be at least equal to the call price with the
higher strike price.
It can also be shown that the call value must be at least equal to
the stock price minus the present value of the strike price or Ct ≥ St −
Ke−r(T −t) . This relationship holds good at any time before expiration.
Example: Consider a call trading at 3 when St = 105, K = 100, r = 7%,
and T = 0.5 years. These data violate the previous relationship and give
rise to an arbitrage opportunity because the call is undervalued. In this
case, an investor can implement the following strategy: sell the underlying
asset at 105, buy the option at −3, and buy the bond with the remaining
funds at −102.
The investor can exercise immediately the option, pay 100, return the
stock, and keep 2.
The investor can also wait for the maturity date. At this date, the bond
price is 102e0.07(0.5) = 105.6332.
If the stock price is 97 at expiration, the option expires worthless and
the investor pays 97 for the share to re-pay the obligation. The profit in
this case is 102e0.07(0.5) − 95 = 10.6332.
Even, if we consider other levels of the underlying asset, the investor
can re-pay with profit. Hence, the option value must be at least equal to
105 − 100e−0.07(0.5) = 8.435.
All the prices below this allow arbitrage profits.
The call price is an increasing function of time until expiration. In fact,
if we consider two calls with the same characteristics, except for maturity,
then the price of the call with a longer maturity must equal or exceed
the price of the call with a shorter maturity. If this principle is violated,
arbitrage can be implemented with risk-less profits.
In the absence of dividend or cash distributions to the underlying asset,
there is no reason to exercise a call on a non-dividend paying asset. In fact,
a call on a non-dividend paying asset is always worth more than its intrinsic
value St − K. Since before expiration, the call value must be at least worth
St − Ke−r(T −t) , exercising the call before expiration discards at least the
option time value, i.e., the difference between K and Ke−r(T −t) . Early
exercise is never optimal for an American call on a non-dividend paying
asset. Since European calls cannot be exercised before expiration, this
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
makes an equivalence between the European call price and the American
call price in the absence of distributions to the underlying asset.
pt ≥ Ke−r(T −t) − St
This relationship holds good at any time before expiration since the put
holder cannot exercise his/her put before expiration.
Since the actual price is 2.5, the investor can implement a trading
strategy to generate risk-less profits. In this case, he/she can buy the put at
2.5 and buy the stock at 95. The strategy can be implemented by borrowing
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
at 97.5 or (95 + 2.5) at 7% for six months. The net cash-flow of this strategy
is zero.
At maturity, the investor exercises his/her put option, (receives the
strike price of 100 and delivers the stock). He re-pays the borrowed amount
97e0.07(0.5). The net cash-flow from this strategy is 100.45511. This is
the risk-less arbitrage profit. Hence, with zero investment, the strategy
guaranteed a risk-less arbitrage profit. Therefore, the above inequality must
hold good.
The put price is an increasing function of time until expiration. In fact,
if we consider two puts with the same characteristics, except for maturity,
then the price of the put with a longer maturity must equal or exceed
the price of the put with the shorter maturity. If this principle is violated,
arbitrage can be implemented with risk-less profits.
The put price is a decreasing function of the strike price. In fact, if we
consider two American puts with the same characteristics, except for the
strike price, then the price of the put with a higher strike price must be
higher than the price of the put with the lower strike price.
If we consider two European puts with the same characteristics, except
for the strike price, then the price of the put with a higher strike price must
be higher than the price of the put with the lower strike price.
If we consider two American puts with the same characteristics, except
for the strike price, then the difference between their prices must be less
than the difference in their strike prices.
If we consider two European puts with the same characteristics, except
for the strike price, then the difference between their prices must be less
than the difference in the present value of their strike prices.
Ct ≥ St − Ke−r(T −t)
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
This equation shows that the higher the interest rates, the smaller is
the present value of the strike price Ke−r(T −t) . Hence, higher interest rates
give a higher value for the difference St − Ke−r(T −t).
pt ≥ Ke−r(T −t) − St
Since the put holder receives a maximum value of the strike price (a
potential cash inflow), the higher the present value of this cash inflow,
the higher is the put value. Therefore, the put must be higher for lower
interest rates. The above relationship can be used in this context to show
the presence of a profitable arbitrage opportunity when the put fails to
adjust to the changing interest rates.
date (one year, for example) is diagonal. For a bond, the graph of the profit
and loss is a straight line because the bond value is known with certainty
at maturity.
The call seller implements a short position in the call. The value of a short
call position at expiration is:
It is clear that the value of the short position is always negative. This is
because at initial time, the option seller receives the premium option.
ST = 110 and receives the strike price K = 100. Hence, the value of the
call position for the seller is −10 or:
CT − Ct = max[0, ST − K] − Ct
In the same way, if we denote the value of the sold call at time t by Ct ,
then the profit or loss on a short call position held until maturity is:
Ct − CT = Ct − max[0, ST − K]
(CT − Ct ) + (Ct − CT ) = 0
since
A trader buys, in general, a call when he/she expects the underlying asset
price to rise. He/she shorts a call when he/she expects a stability or a
decline in the /sheunderlying asset price.
If the put holder had paid 10 at time zero, his/her net profit is 5 or (15−10).
The previous analysis shows that the payoffs of calls and puts depend
on the position of the option’s underlying asset with respect to the strike
price. A trader buys, in general, a put when he/she expects the underlying
asset price to fall. He/she shorts a put when he/she expects a stability or
an appreciation in the underlying asset price.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
CT + PT = max[0, ST − K] + max[0, K − ST ]
spread is worth:
The cost of the long bull spreads is Pt (St , K1 , T ) − Pt (St , K2 , T ) with K1 <
K2 . At maturity, the bear spread with puts is worth:
and at the same date, the bear spread with puts is worth:
Example: Consider the following four legs of a transaction: a long call with
K1 = 90, a short call with K2 = 100, a long put with K2 = 100, and a
short put with K1 = 90.
At maturity, the box spread is worth:
or
8 − 0 + 2 − 0 = 10.
or
0 − 0 + 30 − 20 = 10.
Note that in all the cases, the box spread is worth 10. This corresponds
to the difference between the strike prices. Hence, the strategy appears
equivalent to a risk-less investment. Therefore, to avoid profitable arbitrage
opportunities, the box-spread value at time zero must be the discounted
value of the difference between the two strike prices. Hence, its initial cost
K2 −K1
must be (1+r) ( T −t) .
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
This shows that the box is worth (K2 − K1 ) at the maturity date. If its
value is less than the discounted value of (K2 − K1 ), then risk-less arbitrage
would be possible.
Consider the two following two relationships between the European
options c and p and the American options C and P :
These relations account for the value of the early exercise premium for calls
and puts with different strike prices. If the first condition was not satisfied,
then selling the American call and buying the European call (with a strike
price K2 ) and buying the American call and selling the European call
(with a strike price K1 ), would allow an immediate profit. If the American
call with a strike price K2 is not exercised before the maturity date, the
position produces a zero cash-flow at this date. If the call with a strike
price K2 is exercised, the option with a strike price K1 can be exercised to
generate a cash-flow (K2 − K1 ), which will be invested until the maturity
date T .
If the option with a strike price K2 is exercised before the maturity
date at a date t1 < T , the result at maturity is (K2 − K1 )er(T −t1 ) >
K2 − K1 . Tests of the box strategy for options traded on the Chicago Board
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Options Exchange (CBOE) over eight years reveal some violations of the
non-arbitrage condition. However, the profitable opportunities disappeared
when transaction costs were taken into account. Hence, the market is
globally efficient.
An investor short the butterfly sells the call with the lowest strike price
K1 , sells the call with the highest strike price K3 , and buys two calls with
intermediate strike price K2 . This strategy tends to be profitable when
the underlying asset at maturity is at the intermediate strike prices. At
maturity, the short butterfly spread with calls is worth:
A trader implements a long butterfly spread with calls when he/she expects
the underlying asset price to be relatively stable. He/she implements a short
butterfly spread with calls when he/she expects strong movements in the
underlying asset price.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
An investor short the butterfly sells the put with the lowest strike price
K1 , sells the put with the highest strike price K3 , and buys two puts with
an intermediate strike price K2 . At maturity, the short butterfly spread
with puts is worth:
A trader implements a long butterfly spread with puts when he/she expects
the underlying asset price to be relatively stable. He/she implements a short
butterfly spread with puts when he/she expects strong movements in the
underlying asset price.
with the highest strike price K4 . At maturity, the long condor with calls
is worth:
An investor short the condor sells the call with the lowest strike price
K1 , buys the call with a somewhat higher strike price K2 , buys the call
with the yet higher strike price K3 , and sells the calls with the highest
strike price K4 .
At maturity, the short condor with calls is worth:
An investor short the condor sells the put with the lowest strike price K1 ,
buys the put with a somewhat higher strike price K2 , buys the put with
the yet higher strike price K3 , and sells the put with the highest strike
price K4 . At maturity, the short condor with puts is worth:
A trader implements a long condor with puts when he/she expects the
underlying asset price to be relatively stable. He/she implements a short
condor with puts when he/she expects strong movements in the underlying
asset price.
ST + PT = ST + max[0, K − ST ]
The cost of the insured portfolio at initial time t corresponds to the sum of
the stock portfolio St and the put price Pt or (St + Pt ).
The profit (or loss) on an uninsured portfolio is simply the difference
between its final value and initial value or ST − St . The insured portfolio
has a superior performance only when:
max[0, K − ST ] − Pt − St > 0
This strategy pays when markets are down, as in the year 2008.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
cT − pT = max[0, ST − K] − max[0, K − ST ]
max[0, ST − St ] − max[0, St − ST ]
If the underlying asset price rises with respect to the initial level St ,
then ST > St and the call is worth ST − St . The put value is zero.
If the underlying asset price falls with respect to the initial level St ,
then ST < St , the call is worth zero and the put value is St − ST .
Note that this result is equivalent to that of the stock portfolio. Hence,
the value of a portfolio comprising a long call and a short put is equivalent
to that of the underlying asset or portfolio. This result is always verified
when the option’s strike price corresponds to the value of the underlying
asset when this strategy is implemented.
the option’s underlying asset. The synthetic underlying asset has the same
value and exhibits the same profit (loss) pattern as the underlying asset
(or stock).
The value of the synthetic underlying asset at time t can be written as
St = ct − pt + Ke−r(T −t) .
The investor invests an amount Ke−r(T −t) in risk-free bonds at time t.
This amount grows at the rate er(T −t) . At maturity, the value of the
investment in risk-free bonds is exactly K. This corresponds also to the
face value of the discount bonds at maturity.
At the option’s maturity date, the value of the portfolio comprising a
long call, a short put, and a certain amount K is:
cT − pT + K = max[0, ST − K] − max[0, K − ST ] + K
In the absence of an early exercise, this equality is also verified for American
options at expiration:
CT − PT + K = max[0, ST − K] − max[0, K − ST ] + K
pt = ct − St + Ke−r(T −t)
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
It stipulates that the put can be duplicated by a short position in the stock,
a long position in the call, and an investment in risk-free bonds paying the
strike price at the option’s common maturity date.
When at initial time t, St = K, then the call price must be higher than
the put price. In fact, the relationship can also be presented in the following
form: ct − pt = St + Ke−r(T −t). Since St = K, the right-hand side must be
positive. Therefore, ct − pt must be positive. Therefore, the call price must
be higher than the put price.
c − p = S − Ke−rT
where r stands for the risk-less interest rate and T is the option’s
maturity date.
A conversion is a strategy based on the above relationship. It can also
be written as: short a call + long a put + long the underlying asset = short
a synthetic underlying asset + long the underlying asset.
A reversal corresponds simply to a reverse conversion: long a call+short
a put+short the underlying asset = long a synthetic underlying asset+short
the underlying asset.
If we substitute the underlying asset by a synthetic underlying asset in
the conversion strategy for a different strike price, this eliminates the risks
associated with the variations of the underlying asset price and gives the
well-known strategy, the box spread. The box spread is simply a strategy
equivalent to borrowing or lending money for a certain period. For an anal-
ysis of financial markets, volume, volatility, and spreads, readers can refer
to Hong and Wang (2000), French (1980), and Gibbons and Hess (1981).
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Table 2.5. Results at the maturity date in three months of the strategy of selling 6,666
calls at 0.69 when volatility is at 20%.
Table 2.6. Results at the maturity date in three months of the strategy of selling 6,666
calls at 1.5 when volatility is at 47%.
No matter how far the underlying asset may fall by expiration, the calls he
sold short will still expire worthless and the most he can hope to collect on
them is what he initially sold them for 4,599.54, no matter what happens
to the stock, the T-bills will still yield 57.49.
We can calculate the up-side tolerance point for a short sale of 6,666 calls,
15.6986.
Below this stock price level, the premium initially received from the short
sale of calls and the interest income from his T-bills is sufficient to repay
the exercise value of the short calls.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Case 1: C = 0,69
Above this point, the exercise value of the calls becomes great enough to
cause a net loss.
The 15.6986 in the price at which the risk return line crosses down through
the center of the chart into the region representing losses.
Case 2: C = 1, 5
Above this point, the exercise value of the calls becomes great enough to
cause a net loss.
The 16.518 in the price at which the risk return line crosses down through
the center of the chart into the region representing losses.
Table 2.10. Results at the maturity date in three months of the strategy of selling calls
and buying the underlying stock for a hedge: delta stocks: 0.55 stocks.
Table 2.11. Results at the maturity date in three months of the strategy of selling calls
and buying the underlying stock for a hedge: delta stocks: 0.55 stocks.
The patterns of risk and return associated with this strategy shows that
with the stock unchanged or down, the investor’s profit would be the initial
proceeds from the sale plus interest from the T-bills, a total of 93,148.98.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Table 2.12. Results at the maturity date in three months of the strategy using leverge
in selling calls.
Table 2.13. Results at the maturity date in three months of the strategy using leverge
in selling calls.
2.13.2.2. Leverage in selling Call options (without holding the stocks): The
extreme case
Extreme case: Stock up by 100%
Consider the limiting case in which the stock doubles from 15 to 30, the
aggregate exercise value the call seller would have to pay totals (1,999,990).
Offset by the initial proceeds from sale of the calls and the interest on
the T-bills, the investor’s total loss is (1,906,849).
Table 2.16. Results at the maturity date in three months of the strategy of selling calls
and buying the underlying stock.
Table 2.17. Results at the maturity date in three months of the strategy of selling calls
and buying the underlying stock.
If an investor sells short a stock, another who already owns the stock
(IC) has to be willing to lend it to DC to sell.
First, if the lender (IC) suddenly requires (DC) to immediately return
the borrowed stock, forcing DCG to buy it in the open market, prices may
be far from favorable.
Second, If IC accepts to lend the stock, DC must put collateral to
guarantee the loan (apart from the margin DC must put with broker),
tying up some of the proceeds of the sale.
This creates an economic asymmetry between buying and short selling.
Buying stocks ties up funds that could otherwise be invested to earn
interest and although the proceeds from selling stocks one already owns can
be fully reinvested to earn interest only on a portion of the proceeds from
short-selling stocks can be reinvested.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Table 2.19. Short selling stocks (stock lending) Example: double the dividend payment.
In the US, New York Exchange rules plus-tick rule, permits short sales
to be executed only at prices representing a plus tick from the previous
different price.
If we sold the portfolio of stocks short, the loss is:
As the long stock has a downside tolerance point, the short stock has
an upside tolerance point.
Net percentage interest income from T-bills: offset by the resti-
tution of dividends/Holding period net interest yield = (Interest
income − dividends)
Value of T-bills
15.0025 = 15 + 0.0025
The short position shows a loss if the index is unchanged, rather than
a profit.
Performance is also degraded when the stock is down.
The question is how low the stock must fall before the position fails to
show a loss?
Holding period interest yield = (Interest income − dividends)/
Value of T-bills
The balance of the 100,000 risk capital that could be invested in T-bills
would be:
Remaining capital in T-bills = total capital − investment required
to buy calls
Valuation:
Using option valuation procedure and assuming a volatility of
30%, the interest rate at 5%, the call price is 0. 96.
15 = 15 × 1
The balance of the 100 000 Risk capital that could be invested in T-bills
would be:
Remaining capital in T-bills = total capital − investment required
to buy calls
$15.511 = 0.5110 + 15 =
If the stock rallies by 10%, from 15 to 16.5, each option will have an exercise
value of 1.5.
Exercise value of call = (stock − call’s strike) × contract multiplier
1.5 = (16.5 − 15) × 1
Aggregate Exercise value of position = Exercise value of call ×
number of calls
9,999 = 1.5 × 6,666
Including the interest income from T-bills, the investor’s total profit with
the stock up 10% is 6,592.57.
At maturity after 3 months:
If the asset price up by 10%:
Initial cost of calls (4,599)
Exercise value: 1.5 × 6,666×1 9,999
Profit from calls 5,400
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
Advantages: The advantages gained for these handicaps are that the
option buyer is assured that his investment cannot do any worse than a
known maximum loss equal to the costs of the options, minus any interest
earned on the leftover cash invested in T-bills.
Maximum loss in position = cost of calls − interest on T-bills
5,391.04 = 6,583.608 − 1, 192.575
Table 2.20. Results at the maturity date in three months of the strategy of buying calls
and holding the underlying stock.
Table 2.21. Leverage in buying call options (without selling the underlying).
145,000 = 100,000/(0.69 × 1)
Summary
A forward contract or a futures contract is an agreement between two
parties to buy or sell a specific asset at a specified price at a given time
in the future. Futures contracts are traded on an exchange and have
standardized features. They are settled on a daily basis while the forward
contracts are settled at the end of the contract. Besides, for most futures
contracts, delivery is never actually made. Futures markets are used for
hedging, speculation, and arbitrage motives. Futures and forward contracts
are priced using the cost of carry model. Petroleum futures contracts (or
other commodity contracts) can be used as specific hedges when they are
associated with a planned cash transaction. The benefit to a company using
petroleum products futures is to “lock in” profit margins and/or to protect
inventory against falling prices. When spot prices are higher than long-term
prices, any hedge using a future maturity will be equivalent to a forward sale
below the spot price. This can lead to a loss if market prices do not fall at the
same rate. When spot prices are lower than long-term prices, the producer
can sell on the futures market at a higher price. So, he/she can fix his/her
hedge or future sales at a better price than the spot market. Companies
using the physical oil market, for example, (or other commodities) can hedge
themselves against adverse price movements by taking an opposite position
on the futures or the forward market. The potential loss in the physical
market can be offset by an equivalent gain on the futures or the forward
market. The futures market offers a facility for hedging price risks. Hedging
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
with two calls, it is designed to profit from falling underlying asset prices.
This strategy with calls corresponds to the short position to the bull spread.
A box spread corresponds to a combination of a bull spreads with calls
and a bear spread with puts. It is implemented using two spreads with
two strike prices. A butterfly spread corresponds to a combination using
three calls with three strike prices K1 < K2 < K3 . The calls have the same
underlying asset and maturity date. A condor corresponds to a combination
using four calls with four strike prices K1 < K2 < K3 < K4 . The calls have
the same underlying asset and maturity date.
A ratio spread is a strategy involving two or more related options in a
given proportion. A trader can buy a call with a lower strike price and sell
a higher number of calls with a higher strike price. Portfolio insurance is
an investment-management technique that protects a portfolio from drops
in value. This technique proposes some simple concepts allowing to insure
a stock portfolio.
An investor who buys a European call and sells a European put on
the same underlying asset creates a position that exhibits the same payoff
pattern as the underlying asset. A portfolio with a long European call and a
short European put shows a profit (loss) pattern that can mimic the result
of the underlying asset. This chapter develops some of the most frequent
strategies used in the market place.
Questions
1. What are the main specific features of forward and futures markets?
2. What are the main pricing relationships for forward and futures
contracts?
3. What are the main trading motives in futures markets?
4. Provide some definitions for hedging, speculation, and arbitrage.
5. What are the main bounds on option prices?
6. Describe the simple trading strategies for options and their underlying
assets.
7. How can one implement a straddle?
8. How can one implement a strangle?
9. Describe option spreads in bull and bear strategies involving calls.
10. Describe option spreads in bull and bear strategies involving puts.
11. How can one implement butterfly strategies using put and call options?
12. How can one implement condor strategies using put and call options?
13. Describe ratio-spread strategies.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
14. How can one implement some combinations of options with bonds and
stocks and portfolio insurance strategies?
15. How can one implement portfolio insurance strategies?
16. What are the basic synthetic positions?
17. How can one implement a conversion?
18. How can one implement a reverse conversion?
19. What is the main characteristic of a box spread?
25.00%
20.00%
15.00%
P&L from buy put at
10.00%
Maturity "p"
5.00%
0.00%
%
20 %
29 %
38 %
7%
11 %
%
-5.00%
33
7
67
33
33
33
.6
.6
.6
.6
.
2.
3.
4.
5.
-6
-3
-2
-1
-10.00%
10.00%
5.00%
0.00%
1 7 13 19 25 31 37 43 49 55 61 67 73
-5.00%
P&L from sell put at
-10.00%
Maturity -p
-15.00%
-20.00%
-25.00%
-30.00%
The graphic shows P&L from the Table for buying and selling puts.
buy call
35.00%
30.00%
25.00%
20.00%
15.00%
10.00% buy call
5.00%
0.00%
-33.33%
-27.33%
-21.33%
-15.33%
-9.33%
-3.33%
2.67%
8.67%
14.67%
20.67%
26.67%
32.67%
38.67%
-5.00%
-10.00%
-15.00%
sell call
0.15
0.10
0.05
0.00
-33.33%
-27.33%
-21.33%
-15.33%
-9.33%
-3.33%
2.67%
8.67%
14.67%
20.67%
26.67%
32.67%
38.67%
-0.05
-0.10 sell call
-0.15
-0.20
-0.25
-0.30
-0.35
Table 2.2. Buy a put and hedge and sell a put and hedge.
3.1. Strategy of selling put and hedge: sell delta units of the
underlying
We win if the stock lies within a given range.
We loose if the stock is outside that range on the right or on the left.
10.00%
5.00%
0.00%
P&L , sell put, sell 0.5
-33.33%
-25.33%
-17.33%
-9.33%
-1.33%
6.67%
14.67%
22.67%
30.67%
38.67%
Stock -p-0.5S
-5.00%
-10.00%
-15.00%
Fig. 2.5. Selling a put and sell delta units of the underlying with delta equal 0.5.
October 10, 2009 10:14 spi-b708 9in x 6in b708-ch02
3.2. Strategy of buy put and hedge: buy delta units of the
underlying
15.00%
10.00%
5.00%
Buy put, buy 0.5S
p=0.5s
0.00%
-33.33%
-26.33%
-19.33%
-12.33%
-5.33%
1.67%
8.67%
15.67%
22.67%
29.67%
36.67%
-5.00%
-10.00%
Fig. 2.6. Buy a put and buy delta units of the underlying.
The investor wins if the stock lies outside a given range. The investor loses
if the stock is inside that range on the right or on the left.
4. Strategy of Buy Call, Sell Put, and Buy Call, Sell Put
and Hedge
Table 2.3. Buy call, sell put, and buy call, sell put and hedge.
50.00%
40.00%
30.00%
20.00%
10.00%
buy call, sell put,
0.00%
c-p
-10.00%
3%
26 %
36 %
7%
16 %
-2 3%
-1 %
7
67
33
33
.3
.6
.6
.6
3
6.
3.
3.
3.
-3
-20.00%
-3
-30.00%
-40.00%
-50.00%
0.00%
-33.33%
-26.33%
-19.33%
-12.33%
-5.33%
1.67%
8.67%
15.67%
22.67%
29.67%
36.67%
-1.00%
-2.00%
-4.00%
-5.00%
-6.00%
Stock price at maturity Sell call, buy put, −c + p Sell call, buy put and hedge
Stock price at maturity Sell call, buy put, −c + p Sell call, buy put and hedge
Stock price at maturity Sell call, buy put, −c + p Sell call, buy put and hedge
References
French, K (1980). Stock returns and the weekend effect. Journal of Financial
Economics, 8 (March), 55–69.
Gibbons, MR and P Hess (1981). Day of the week effects and asset returns.
Journal of Business, 54, 579–596.
Hong, H and J Wang (2000). Trading and returns under periodic market closures.
Journal of Finance, 55(1) (February), 297–354.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Chapter 3
Chapter Outline
This chapter is organized as follows:
1. Section 3.1 develops the basic strategies using calls and puts.
2. Section 3.2 illustrates several combined strategies.
3. Section 3.3 explains the way traders use option pricing models to
compute option prices and to estimate the market volatility.
Introduction
Using the definition of a standard or a plain vanilla option, it is evident that
the higher the underlying asset price, the greater the call’s value. When the
underlying asset price is much greater than the strike price, the current
option value is nearly equal to the difference between the underlying asset
price and the discounted value of the strike price. The discounted value of
the strike price is given by the price of a pure discount bond, maturing
at the same time as the option, with a face or nominal value equal to the
strike price. Hence, if the maturity date is very near, the call’s value (put’s
value) is nearly equal to the difference between the underlying asset price
and the strike price or zero. If the maturity date is very far, then the call’s
value is nearly equal to that of the underlying asset since the bond’s price
will be very low. The call’s value can not be negative and can not exceed
the underlying asset price. Options enable investors to customize cash-flow
patterns. We present some of the most common used option strategies,
141
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 143
Long the underlying asset: (1, 1): , Short the underlying asset:
(−1, −1): .
The symbols (−1, 0, 1) refer to a downward movement, (−1), a flat
position (0) or an upward movement (1). The risk-return trade-off of the
basic strategies can be represented using the different symbols.
Using the above notations, it is possible to construct the risk-reward
trade-off of any option strategy.
For example, long a call (0, 1) and short a put (1, 0) is equivalent to
long the underlying asset (1, 1). Also, short a call (0, −1) and long a put
(−1, 0) is equivalent to a short position in the underlying asset (−1, −1).
We give the basic synthetic positions when the options have the same
strike prices and maturity dates.
Long a synthetic underlying asset = long a call + short a put.
positions are often implemented by traders and market makers who follow
delta-neutral strategies. For example, when managing an option position,
buying a call or a put with the same strike price are two equivalent strategies
since when buying a call, the trader or the market maker hedges his
transaction by the sale of the underlying asset and when buying a put,
he hedges his transaction by the purchase of the underlying asset. Buying
the call and selling the put is equivalent to a long put with the same strike
price. This transaction enables the trader or market maker to make a direct
sale of the put since a position in a long call and a short put is equivalent
to a long position in the underlying asset.
Specific features: The potential gain is not limited but the potential loss
is limited to the option premium.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 145
Buy a call
20
18 18.1
16
14
13.1
12
10
profit
8 8.1
6
4
3.1
2
0
-2 1.9- 1.9- 1.9- 1.9- 1.9-
-4
-18 -13 -8 -3 2 7 12 17 22
cours du support
Fig. 3.1.
The risk-reward trade-off is inverted when selling calls (Table 3.2). The
results of the strategy are given in Fig. 3.2.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Break-even point A K +c
Maximal loss Not limited
Maximal gain Premium
4
2 1,9 1,9 1,9 1,9 1,9
0
-2
-3,1
-4
-6
profit
-8 -8,1
-10
-12
-13,1
-14
-16
-18 -18,1
-20
90 95 100 105 110 115 120 125 130
S
Short a call
90 −12 1, 9 100
95 −7 1, 9 100
100 −2 1, 9 100
105 3 1, 9 100
110 8 1, 9 100
115 13 −3, 1 −165
120 18 −8, 1 −430
125 23 −13, 1 −696
130 27 −18, 1 −961
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 147
In Fig. 3.2, the break-even point is given by the sum of the strike price
and the option premium.
Specific features: The potential gain is not limited and the potential loss
is limited to the option premium.
In Fig. 3.3, the break-even point is given by the algebraic sum of the
strike price and the option premium (Table 3.3).
40
37,3
35
30
27,3
25
20
profit
17,3
15
10
7,3
5
0
-2,7 -2,7 -2,7 -2,7 -2,7
-5
60 70 80 90 100 110 120 130 140
S
Long a put
Break-even point A K −p
Maximal loss Premium
Maximal gain Not limited
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 149
-2,3
-5
profit
-7,3
-10
-12,3
-15
-17,3
-20
80 85 90 95 100 105 110 115 120
S
Short a put
Break-even point A K −p
Maximal loss Not limited
Maximal Gain Premium
Expectations: The trader expects a high volatility until the maturity date.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 151
20
15
13.7 13.7
10
8.7 8.7
profit
5
3.7 3.7
0
-1.3 -1.3
Call
Put
-5
Straddle
-6.3
-10
80 85 90 95 100 105 110 115 120
S
Break-even point A S = K − (c + p)
B S = K + (c + p)
Maximal loss C (c + p) if S = K
Maximal gain K − (c + p), if S tends toward 0
Unlimited, if S is beyond the limits
Definition:
Specific features:
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 153
10
Call
6.3
Put
5 Straddle
1.3 1.3
0
-3.7 -3.7
profit
-5
-8.7 -8.7
-10
-13.7 -13.7
-15
-20
80 85 90 95 100 105 110 115 120
S
Definition: Buy a call with a strike price Kc and buy a put with a strike
price Kp where the Kp < Kc .
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Specific features
Notes: The trader buys the 105 call and the 95 put. The theoretical
prices of these options, respectively are 2.04 and 0.55, or a total of 2.58.
The quantity is 10, and the total cost of the strategy is 25.8.
20
Call
Put 17.4
Strangle
15
12.4
10
profit
7.4 7.4
2.4 2.4
-2.6
-5
85 90 95 100 105 110 115 120 125
S
Break-even point A S = Kp − (c + p)
B S = Kc + (c + p)
Maximal loss (c + p), if Kp < S < Kc
Maximal gain A Kp − (c + p), if S tends toward 0
B Unlimited, if S is higher
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 155
5
2.6
-2.4 -2.4
-5
profit
-7.4 -7.4
-10
-12.4
-15
Call
Put -17.4
Strangle
-20
85 90 95 100 105 110 115 120 125
S
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 157
The reader can make the specific comments by comparing this strategy
with the long strangle.
Definition: Buy an out-of-the money call and sell out-of-the money put
as in Table 3.10 (Fig. 3.9).
40
33,0
30
23,0
20
13,0
10
profit
3,0
0
-7,0 -7,0
-10 -7,0
-17,0
-20
-27,0
Call Out
-30
Put Out
Tunnel
-40
530 540 550 560 570 580 590 600 610
S
Fig. 3.9. Long a tunnel (Buy an out-of-the money call and sell an out-of-the
money put).
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Simulation:
40
30
27,0
20
17,0
10
7,0 7,0
profit
7,0
0
-3,0
-10
-13,0
-20
-23,0
Call Out
-30 Put Out
Tunnel -33,0
-40
530 540 550 560 570 580 590 600 610
Fig. 3.10. Short a tunnel (Sell an out-of-the money call and buy an out-of-the
money put).
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 159
Table 3.11.
80
60
40
20 11.1
profit
-8.9
-20
Call In
-40
Call Out
Spread
-60
490 510 530 550 570 590 610 630 650
S
Table 3.12. Bull spread with calls: S = 102, r = 5%, volatility = 20%,
and T = 100 days.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 161
80
60
40
20 11.0
0
profit
-9.0
-20
-40
Put Out
-60
Put In
Spread
-80
-100
490 510 530 550 570 590 610 630 650
S
The trader can sell the put spread by selling the higher strike price put
and buying the lower strike price put. The strategy is done with a credit.
60
40
20
8.5
0
profit
-20 - 11.5
-40
Call In
Call Out
-60
Spread
-80
490 510 530 550 570 590 610 630 650
S
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 163
100
Put Out
80 Put In
Spread
60
40
20 8.7
profit
-20 -113.
-40
-60
-80
490 510 530 550 570 590 610 630 650
S
20
10
5.5
0.5
0 0.5
-4.5 -4.5
profit
-10
-20
Call
-30 Call At
Call In
Butterfly
-40
80 85 90 95 100 105 110 115 120
S
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 165
40
Call
Call At
30
Call At
Call In
20 Condor
10
3.9
profit
-20
-30
-40
60 70 80 90 100 110 120 130 140
S
Variation Performance
S (%) Call out Call at Call at Call in Condor (%)
40
30
20
10 6.1 6.1
profit
-10 -3.9
Variation Performance
S (%) Call out Call at Call at Call in Condor (%)
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 167
The third method uses the relative variations in the underlying asset
prices.
Underlying asset, S + −
Strike price, K − +
Dividends − +
Interest rates + −
Maturity + +
Volatility + +
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 169
Table 3.21. Computation of the historical volatility for the CAC40 index.
0 11/07/03
−1 10/07/03 2929.09 −0.73% 0.00010550
−2 09/07/03 2950.56 0.71% 0.00001672
−3 08/07/03 2929.81 −0.61% 0.00008176
−4 07/07/03 2947.66 0.45% 0.00000229
−5 04/07/03 2934.48 −0.09% 0.00001459
−6 03/07/03 2936.98 0.94% 0.00004160
−7 02/07/03 2909.45 −1.18% 0.00021865
−8 01/07/03 2944.04 2.96% 0.00070764
−9 30/06/03 2858.26 −1.14% 0.00020654
−10 27/06/03 2891.04 −0.09% 0.00001495
−11 26/06/03 2893.64 0.91% 0.00003755
−12 25/06/03 2867.44 2.93% 0.00069115
−13 24/06/03 2784.76 0.81% 0.00002669
−14 23/06/03 2762.20 0.18% 0.00000125
−15 20/06/03 2757.10 0.63% 0.00001133
−16 19/06/03 2739.69 −0.44% 0.00005412
−17 18/06/03 2751.74 −0.39% 0.00004771
−18 17/06/03 2762.60 −1.20% 0.00022318
−19 16/06/03 2795.87 −0.45% 0.00005599
−20 13/06/03 2808.52 1.73% 0.00020624
−21 12/06/03 2760.27
Mean 0.30% 0.00276546
0.00014555
1.21% 1.21%
Volatility 23.05%
The estimated volatility using high and low prices is given by the
following formula:
(Ht − Bt )2
σ̂22 (t) ≡
4Ln(2)
where H refers to the high price and B refers to the low price. This
estimated volatility is 5.2 times more efficient than σ̂02 .
The estimated volatility using high and low prices is given by the
following formula:
(o) (f )
(St − St−1 )2 (Ht − Bt )2
σ̂32 (t) ≡ a + (1 − a) 0 < f < 1,
f 4Ln(2)(1 − f )
This estimated volatility is 6.2 times more efficient than σ̂02 when the
coefficient a = 0.17.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
σ̂ = f −1 [f (σ)] = f −1 (Call)
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 171
1.0
0.4
0.2
Delta
0.0
-0.6
-0.8
-1.0
75 80 85 90 95 100 105 110 115 120 125
S
Fig. 3.18. Evolution of the delta as a function of the underlying asset price.
or in the hedge ratio, as the underlying asset price changes The gamma is
given by the derivative of the hedge ratio with respect to the underlying
asset price. As such, it is an indication of the vulnerability of the hedge
ratio. Figure 3.19 shows the dynamics of the gamma for the call and the
put as a function of the underlying asset price.
The vega or lambda is a measure of the change in the option price
for a small change in the underlying asset’s volatility. The vega or lambda
measures the change in the option price for a change in the underlying’s
asset volatility. It is given by the first derivative of the option premium with
respect to the volatility parameter. Figure 3.20 shows the dynamics of the
vega for the call and the put as a function of the underlying asset price.
The theta measures the change in the option price as time elapses.
The theta measures the change in the option price as time elapses since
time decays presents a negative impact on option values. The theta is
given by the first partial derivative of the option premium with respect
to time.
Figures 3.21 and 3.22 shows the dynamics of the theta for the call and
the put as a function of the underlying asset price and the days to maturity.
A chapter is devoted to the use of these parameters in the monitor-
ing and the management of an option position. This is because of the
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
3.0%
2.5%
2.0%
Gamma
1.5%
1.0%
0.5%
0.0%
75 80 85 90 95 100 105 110 115 120 125
S
Fig. 3.19. Evolution of the gamma as a function of the underlying asset price.
30.0
25.0
20.0
Vega
15.0
10.0
0.0
75 80 85 90 95 100 105 110 115 120 125
S
Fig. 3.20. Evolution of the vega as a function of the underlying asset price.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 173
2.0
1.0
0.0
-2.0
Theta
-3.0
-4.0
-5.0
Theta θ(Call)
-6.0
-7.0
75 80 85 90 95 100 105 110 115 120 125
S
Fig. 3.21. Evolution of the theta as a function of the underlying asset price.
-5.0
-10.0
-15.0
Theta
-20.0
-25.0
-30.0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Days to matuirity
Fig. 3.22. Evolution of the theta as a function of the number of days until maturity.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
3.4. Summary
Since there are at least three maturity dates and three to sometimes
twelve strike prices, the fundamental question is to determine which
option to trade. The most well-known strategies in portfolio management
involve combinations of options. They include vertical spreads, calendar
spreads, diagonal spreads, ratio spreads, volatility spreads, and synthetic
contracts.
A vertical spread involves the purchase of an option and the sale of an
other with the same time to maturity and a different strike price. When
the strategy produces a cash-out flow, we say that the investor is long the
spread. When the strategy generates a cash-inflow, the investor is said short
the spread. The strategy can be implemented by calls or puts using different
strike prices.
A vertical bull spread is implemented when an at-the-money option is
bought and an out-of-the-money option is sold. A calendar-spread strategy
represents a position where the investor is long an option with a longer
term and short an option with a short term for the same strike price.
A diagonal spread involves the purchase of an option with a longer
term and the sale of an other with a short term where both options have
different strike prices.
A bullish diagonal spread is implemented when the purchased option
is at parity and the short option is out-of-the-money.
Volatility strategies are often based on the option implicit volatility.
Examples of volatility strategies include straddles and combinations.
A straddle corresponds to the purchase of a call and a put with the same
strike price and the same maturity date. A combination is a straddle for
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 175
which the options exercice prices are out-of-the money. When the strategy
implies a cash-outflow, the investor is long the strategy and when it involves
a cash-inflow, the investor is short the strategy.
A synthetic forward contract can be created by the purchase of a call
and the sale of a put with the same strike price. In this case, the investor
is long the synthetic forward contract.
The main difference between futures contracts and option contracts is
that the investor pays a premium for options but nothing to establish a
futures position.
The option price depends on the underlying asset price S, the strike
price K, the interest rate r, the time to maturity, T , the volatility σ, and
dividend payouts. The option maturity corresponds to the number of days
until expiration. It is often given in a fraction of a year or in days. The
dividends must be known or estimated before using an option pricing model.
Volatility is the most difficult parameter to estimate. It is a measure of risk
and is a fundamental element in the computation of the option premium. It
can be computed using historical prices of the underlying asset. This refers
to the historical volatility. It can be also calculated using the observed
option prices in the market place and an option pricing model. This refers
to the implicit volatility.
Using an option pricing model, for example the Black and Scholes
model, the call or model, it is possible to compute an option-implied
volatility. We have also introduced the main concepts governing the
management of a portfolio of options: the delta, the gamma, the theta, and
the vega. These parameters enable market participants and option users to
manage their portfolios of options.
Questions
1. Define the strategy of buying (selling) calls.
2. Define the strategy of buying (selling) puts.
3. Define the strategy of buying (selling) spreads.
4. Define the strategy of buying (selling) combinations.
5. What are the determinants of an option price?
6. Define the specific features of long (or short) a straddle.
7. Define the specific features of long (or short) a strangle.
8. Define the specific features of long (or short) a tunnel.
9. Define the specific features of long (or short) spreads.
10. Define the specific features of long (or short) butterfly.
11. Define the specific features of long (or short) a condor.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
CASE STUDIES
Case Studies — Equity Derivatives & Return
Enhancement: Options and Swaps
Case study 1: Combining Total Return Swaps and options
This case studies total return swaps, TRS, and call options to be imple-
mented between two institutions. IC has stocks in its inventory. DC does
not own the underlying stocks.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 177
The Total Return Receiver, TRS, DC, may have the option
to buy the asset.
Valuation and cash flows:
A TRS is initially structured so the net present value is zero to both parties.
As time progresses, the TRS gains and or losses value on each leg so one or
the other counterparty obtains a profit.
Benefits
– the greatest benefit is leverage and flexibility: parties do not transfer
actual ownership of assets.
– this does not induce transaction costs
– Investment return risk: IC conserves the asset in its balance sheet and
DC assumes the risk of capital losses since it guarantees any drop in
value.
– interest rate risk since Libor can change.
– similar to a classic position “long the stock” for DC (Short the stock for
IC)
– main difference: the contract is for difference
– DC pays Libor + or − and the difference in stocks depreciation
– DC receives capital gains and pays Libor or a fee: 1% for example.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 179
Table 2. Position in the stock if IC makes stock lending at Libor 5% Total Return
Swap, TRS, for IC.
Stock price
IC Receives from DC unchanged Stock + 10% Stock − 10%
5. General case for Total return swap: DC long the stock by entering
a TRS with IC for different Libor rate
General case means that instead of using a given stock price, we
use a percentage of the position that applies to any level of the
stock price.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Table 3.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 181
Table 4. (Continued).
Stock price/libor
– the higher the Libor, the higher the profits of IC (the higher the losses
of DC).
– best case is to implement the strategy at Libor 0%.
We see that:
– the profit/loss in % is linear for all rates. The profits and losses are
symmetric for a libor rate Libor = 0.
– for higher rates, IC wins more than DC and the profile is asymmetric.
– for example, IC receives a return of 33% on a position if the stock goes
down to 10. IC pays the same if the stock is up to 20.
– for higher rates, for example a Libor of 7%, for a stock value at 10, IC
return is 40.33%. For a stock value of 20.05, loss for IC is only 26.67% of
the position. In all cases, the profit of IC is the loss of DC.
– best case: zero Libor which also helps to hedge interest rate risks.
The graphic shows for different levels of the underlying stock, X-axis, the
return for IC in % (Y-axis) as a function of Libor rates set at 0% and 5%.
The best case is Libor 0% for both: IC and DC.
Otherwise, all other configurations make more profits to IC than to DC.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 183
TRS Libor 0%
TRS libor 5%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
10.00
10.90
11.80
12.70
13.60
14.50
15.40
16.30
17.20
18.10
19.00
19.90
20.80
-10.00%
-20.00%
-30.00%
-40.00%
-50.00%
Graph 1. For different levels of the underlying stock, X-axis, the return for IC in %
(Y-axis) as a function of Libor rates set at 0% and 5% from entering into the TRS with DC.
For DC: it can enter the swap by paying any fixed fee: for example 1%
since the total return swap allows DC to be long the stock “synthetically.
This long position allows DC to sell calls to the market or to third
parties and to use the “synthetic stock” for “hedging” purposes.
The position has to be “scaled” for the initial hedge. However, at maturity,
among exercise, the option must be cash-settled. Otherwise, for physical
settled options, DC has to buy the stock from the market among option
exercise.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
– Position of DC is like selling a put: sell a call and enter a total return
swap, TRS.
– Their position is like buying a put: buy a call and enter a total return
swap.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 185
Table 6. Results at the maturity date in three months of the strategy of Buying Calls
and entering a TRS (stock lending) with DC at LIBOR = premium: equivalent to buy
a put.
Stock
unchanged Stock + 10% Stock − 10%
Table 7. Results at the maturity date in three months of the strategy of selling calls
and entering a TRS (long stocks) at LIBOR = premium.
Stock
unchanged Stock + 10% Stock − 10%
Conclusion:
When the Libor equals the option premium, IC wins and DC loses in this
combined structures of selling calls and being “synthetically long the stock”
via the TRS.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Table 8. Results at the maturity date in three months of the strategy of selling calls
and entering a TRS with IC at LIBOR < premium.
Stock
unchanged Stock + 10% Stock − 10%
Table 9. Results at the maturity date in three months of the strategy of selling
calls and entering a TRS with IC at LIBOR < premium.
Stock
unchanged Stock + 10% Stock − 10%
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 187
If DC used TRS and options, the premium must be around two times the
Libor.
It must be also set at a level which is coherent with the stock price among
exercise.
For example, if the market is expected to go up by 10%, the premium must
be in line with this level.
P&L DCG
10000
5000
0
10.00
10.75
11.50
12.25
13.00
13.75
14.50
15.25
16.00
16.75
17.50
18.25
19.00
19.75
20.50
-5000
-10000
-15000
-20000
-25000
-30000
P&L DCG
Graph 2. For DC: Risk/return from Selling a call and enter a TRS (stock borrowing).
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Table 10. Results at the maturity date in three months of the strategy of buying
calls and entering a TRS with DC at LIBOR < premium.
Stock
1 call for 10% premium unchanged Stock + 10% Stock − 10%
The graph shows the profit and loss as a function of the underlying stock
price at maturity. Positive sign is a profit.
The maximum profit is limited to the premium 5000.
Loss can be unlimited if the stock falls.
DC is in the opposite postion of IC
For IC: Risk/return from Buying a call and (stock lending) entering
simultaneously into a total return swap with DC
Option premim = 2 times Libor
We use same assumptions as before, with stock Ema = 15, strike
price = 15, premium 0.1, Libor = 0.05, Stock position: 100,000.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 189
P&L ICD
30000
25000
20000
15000
10000
5000
0
10.00
10.75
11.50
12.25
13.00
13.75
14.50
15.25
16.00
16.75
17.50
18.25
19.00
19.75
20.50
-5000
-10000
P&L ICD
Graph 3. For IC: Risk/return from buying a call and enter a TRS (stock lending): Long
call + enter TRS.
DC: sells the call and uses the stock (receives premium, pays for difference)
This is a symmetric payoff for DC and IC
What IC wins, DC loses and vice versa.
Profit/loss for IC from Buying a call and entering simultaneously into a
total return swap with DC is as follows:
The graph shows the profit and Loss (vertical Axis) as a function of the
underlying stock price of Ema at maturity. Positive sign is a profit.
If the stock goes up, loss is reduced to the premium paid.
If the stock goes down, IC is protected and wins.
IC is protected against any downward movement in the stock price of
Emmar in the region from nearly 15 to 10 and less.
–The cost of this insurance is just 5 % of the notional amount.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 191
If IC enters a TRS with DC and buys a call, the profits and loss will be for
some levels of the stock as follows in following Table.
When the TRS is zero for IC, stock at 14.95 and they buy a call, their
return on the total transactions is slightly less than −10%.
When the TRS gives to IC a return of 33% and they buy the call at 10%
premium, the total impact is 23.33% return.
If IC does not enter this transaction, it looses at the stock price level at 10,
about 33% of the value of its stocks.
For IC: the above graph shows the impact of entering a TRS with DC and
buying a call from DC. The graphic shows the Total impact on the 30%
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
50.00%
40.00% Total impact on
30.00% the 30M position
20.00% from 10%
10.00% TRS+Call
0.00% initial position
-10.00%
%
%
7%
3%
3%
3%
17
27
37
-20.00%
-3
-3
-2
-1
-30.00%
-40.00%
Graph 4. The impact of entering a TRS with DC and buying a call from DC. X axis:
return on stocks, Y axis: return from the transaction, and Improves IC situation.
position from entering a TRS (for 10% of the stock position and Buying a
call).
This graph compares the return to IC without entering a TRS and buying
a call (initial position) with the return to IC by entering a TRS and buying
a call.
We see that this structure improves the downside risks of IC.
IF IC buys a call and enters TRS, the P&L is as follows.
The graph is generated using the simulations of the following
table.
Remark:
The graphic shows clearly that this structure improves the downside risks
of IC.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 193
They are more protected on the downside risk. If the market is up, they can
loose if the calls are exercised. This graph shows the total impact on the
position of IC from entreting a TRS, selling calls and using the proceeds of
sale invested in an asset that yields 5%.
0.00%
-16.67%
-15.67%
-14.67%
-13.67%
-12.67%
-11.67%
-10.67%
-9.67%
-8.67%
-7.67%
-6.67%
-2.00%
-4.00%
-6.00%
Graph 5. IF IC buys a call and enters TRS, the P&L is as follows generated using the
previous table.
Risk and return with options without a TRS: the case where we hold the
stock.
Case 2 — For DC: risk/return from selling a call and buying the stock
DC sells the call and buys the stock.
Position:
DC sells a call, holds the stocks and receives a premium of nearly 10%.
DC profits are limited to the premium received.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 195
Table 14. Impact on the position of IC of being short call, long call and entering a
total return swap.
TRS IC Short call Impact TRS and long CALL Total impact TRS and CALL
TRS IC Short call Impact TRS and long CALL Total impact TRS and CALL
DC losses can be unlimited as shown in the graph until 25 times their initial
payment shown in the graph. If the option is not exercised, the profit is 100%.
Case 3 — For DC: risk/return from selling a call and buying the stock
(hedging the stock using delta)
DC sells a call on Ema to IC or third party and implements a hedge by
buying delta units (0.55) of the underlying stock.
Position:
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 197
Table 15. Impact on the position of IC of short call, long call and entering a total
return swap.
TRS IC Short call Impact TRS and long CALL Total impact TRS and CALL
50.00%
40.00%
30.00%
20.00%
Total impact TRS-
10.00% CALL
0.00% inital position
-33.33%
-26.33%
-19.33%
-12.33%
-5.33%
1.67%
8.67%
15.67%
22.67%
29.67%
36.67%
-10.00%
-20.00%
-30.00%
-40.00%
Graph 6. Sell a call, enter TRS with DC and uses the proceeds of selling a call at 5%;
impact on total position of IC.
DC sells a call, holds the stocks in the delta proportion and receives a
premium of nearly 10%. DC profits are limited to the premium received.
DC losses are also limited since it is hedged.
35000.00
30000.00
25000.00
20000.00
15000.00
10000.00 ICD buy call
5000.00
0.00
10
11.1
12.1
13.2
14.2
15.3
16.3
17.4
18.4
19.5
20.5
-5000.00
-10000.00
-15000.00
Graph 7. Risk/return for IC from buying a call from DC. (or any call buyer: third party
for example or the market).
15000.00
10000.00
5000.00
0.00
10
11.2
12.4
13.6
14.8
16
17.2
18.4
19.6
20.8
-15000.00
-20000.00
-25000.00
Graph 8. Risk/return from selling a call and buying the stock (or any call buyer: third
party for example or the market).
P&L from the strategy of selling the call and holding the stock for different
levels of the underlying asset to DC.
Table 16 shows the P&L from the strategy of selling the call and holding
the stock for different levels of the underlying asset. In order to study the
impact of different strategies on the P&L, we consider the following Table.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 199
15000
10000
5000
-10000
-15000
Graph 9. Risk/return from selling a call and buying the stock as a hedge.
Table 16. P&L from the strategy of selling the call and holding the stock for different
levels of the underlying asset to DC.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 201
30.00%
20.00%
10.00%
0.55S-C
s-c
0.00%
C-0.55S
%
14 %
-2 3%
-2 3%
-1 3%
-9 %
20 %
26 %
32 %
38 %
7%
-3 %
3%
C-S
67
67
33
7
7
3
-10.00%
3
.6
.6
.6
.6
.6
.3
.3
2.
8.
3.
7.
1.
5.
-3
-20.00%
-30.00%
Column 3: P&L from the strategy long call, short delta stocks
Column 4: P&L from the strategy of selling the call at maturity and being
long the stocks on all the stock position of IC
Column 5: P&L from the strategy of selling the call 1/10 by IC at maturity
Column 6: P&L from the initial position of holding the stocks for IC (30%)
The important column is column 4 because it shows the impact on the
total position of IC.
Impact on IC:
Remark:
Selling calls improves the downside risk of IC.
If the option are exercised, this affects the upside.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 203
Table 17. Impact on IC overall position (30%) of selling calls on 3% of EMA. IC total
position and P&L from the strategy of short call on (1/10) of its position and long stocks
“−:sell, + buy”.
Short call,
long Delta
stocks: 0.55 S-C C-S IC total
delta hedge long call, position IC; sellcalls All stocks
(c-delta s) short stock, C-0.55S impact on all 1/10 sell call intial position
Short call,
long Delta
stocks: 0.55 S-C C-S IC total
delta hedge long call, position IC; sellcalls All stocks
(c-delta s) short stock, C-0.55S impact on all 1/10 sell call intial position
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 205
Table 18.
50.00%
40.00%
30.00%
20.00%
10.00% S-0.1C
0.00% initial position
-33.33%
-26.33%
-19.33%
-12.33%
-5.33%
1.67%
8.67%
15.67%
22.67%
29.67%
36.67%
-10.00%
-20.00%
-30.00%
-40.00%
20.00%
10.00%
0.00%
-33.33%
-29.00%
-24.33%
-19.67%
-15.00%
-10.33%
-5.67%
-1.00%
3.67%
8.33%
13.00%
total + 10%C-S
-10.00%
initial position
-20.00%
-30.00%
-40.00%
Graph 12. Total impact on IC overall position (30%) of buying calls on 3% of EMA.
Table 19.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 207
EXERCISES
Call options specific features
Exercise 1: Describe the following strategy: Buying call options is an
alternative to buying stocks — a known maximum loss but potentially
unlimited profit.
Buying calls can be a unique managerial alternative to buying the
underlying asset. (Another alternative is to short sell puts).
Investments in stocks cannot be accurately replicated by buying calls —
buying calls inevitably produces a unique, nonlinear pattern of returns.
In fact, call buying can create patterns that may more closely match
the true risk/return preferences of investors.
This position will be profitable only if the underlying asset is high
enough to give the calls sufficient exercise value to repay their initial cost.
This is called the upside break-even point.
With this strategy, the investment can not do any worse than a known
maximum loss equal to the cost of the options minus any interest earned
on the leftover cash invested in Treasury bills.
When the investor uses calls, his position will not be profitable unless
the underlying is high enough to give the options sufficient exercise value
to repay the initial cost.
When the investor chooses to buy calls he is inevitably making a
profoundly different kind of investment. He is engaging in a trade-off-in
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 209
However, the investor can loose the money paid for the option if the
asset finishes below or the strike price.
Exercise: Explain how buying put options is as an alternative to
short selling stocks.
Buying puts can be an alternative to short selling the stocks, however
the replication is not easy.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Example: Investor buy put contracts and finance the cost by borrowing at
a given rate. It is possible to construct a position in puts to replicate short
selling a stock and borrowing the funds required to buy the puts.
When an investor buys a put, he will break even when the stock declines
sufficiently to allow the exercise value of the put to repay the initial put
cost plus the cost of financing the purchase (of the put).
We can calculate the down-side break-even point for this put position.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 211
Below this point, the exercise value of the puts becomes great enough
to generate a loss.
By short selling put options, the investor subjects himself to losses on
the downside beneath the tolerance point limited only by the fact that the
underlying can not be less than zero.
But, this is less risk than he would have taken by buying a stock for
which selling a put is an explicit alternative.
Exercise: Explain leverage associated with selling put options.
Using put options:
– put options can be used to hedge a portfolio against declining stock prices
– put options are sometimes used to speculate on the downside asset
movement
– put allow the investor to participate in the downside of the market when
he can not sell the underlying or sell a futures
– the maximum loss is limited to the premium
– investors do not bear significant risk as that of a short position in the
underlying.
In this case, we put a collateral. This collateral can be invested at zero risk
in Cds or safe assets for the option’s maturity life.
Often, the maturity mostly used is 3 months.
This can yield a return of 4.5%.
We can get 40% of that income plus the premium if the option is not
exercised.
Exercise: How to implement the hedge when buying calls?
We sell the underlying assets.
If we do not have the stock, we sell it short.
Short selling is not allowed.
So we borrow the stock.
We have to pay the collateral.
This collateral can be invested to yield a return (riskless return) on a riskless
asset. We share this.
Time value
If call price is 3 and the intrinsic value is 2, then the time value is 1.
Remarks:
Investors using options as a substitute for stocks will receive a funding
benefit due to the lower outlay required to achieve equivalent exposure via
an option than with a straight equity (or bond).
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 213
The higher the interest rate in the currency of the underlying, the
higher this funding benefit becomes, and the higher the option price.
The increased price of the option will result in an increased premium.
Parameters
Asset price: 14, Interest rate: 5%, Volatility: 40%, Call premium: 3.68 for a
strike of 14, Put premium: 2.35 for a strike of 14.
Minimum trade size: 10 options,
Call exercise
– The call gives the investor upon exercise the difference, at any time before
maturity, between the underlying asset level (index level) and the strike
price.
EMA 14 call — cost 3.68
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 215
Table 1.
Type of
options Strike price % of Moneyness Issue price Gearing
Table 2.
Underlying expiry value Call expiry value Call P/L in $ Call P/L in
10 0 0 −100%
11 0 0 −100%
12 0 0 −100%
13 0 0 −100%
14 0 0 −100%
15 1 0 −100%
16 2 0 −100%
17 3 0 −100%
18 4 0.32 0.086
19 5 1.32 0.358
20 6 2.32 0.6304
30 16 12.32 3.3478
50 36 32.32 8.7826
60 46 42.32 11.52
When the stock price ends at 30, the P/L is about 12.32 and the profit
is nearly 12 times the initial investment in the option.
The calls settlement value or final exercise is given by the difference
between the asset price (index) closing level and the strike price.
When the underlying asset price (index) is higher than the strike price,
the call has a positive intrinsic value. Otherwise, the call expires worthless.
The total return is the underlying asset price (index) less strike price
less purchase price.
The break-even point = underlying asset price (index) above the strike
by the original purchase price of the option.
Table 3.
Type of
options Strike price % of moneyness Issue price Gearing
Table 4.
10 0 0 −100%
11 0 0 −100%
12 0 0 −100%
13 0 0 −100%
14 0 0 −100%
15 1 −0.2 −16%
16 2 0.8 −66%
17 3 1.8 1.5
18 4 2.8 2.33
19 5 3.8 3.16
20 6 4.8 4
30 16 14.8 12.33
50 36 34.8 29
60 46 58.8 49
Call exercise:
– The call gives the investor upon exercise the difference, at any time before
maturity, between the underlying asset level (index level) and the strike
price.
Trading Options and Their Underlying Asset: Risk Management in Discrete Time 217
Table 5.
Table 6.
Stock expiry value put expiry value Put P/L Put P/L in%
1 13 12 12
5 9 8 8
10 4 3 3
13 1 0 −0
14 0 0 −100%
15 0 0 −100%
16 0 0 −100%
17 0 0 −100%
References
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June) 301–320.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–654.
Cox, J, S Ross and M Rubinstein (1979). Option pricing: a simplified approach.
Journal of Financial Economics, 7, 229–263.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
October 9, 2009 15:36 spi-b708 9in x 6in b708-ch03
Part II
219
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
220
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
Chapter 4
Chapter Outline
This chapter is organized as follows:
1. Section 4.1 presents the Cox, Ross and Rubinstein (CRR) model for the
valuation of standard equity options (Cox et al., 1979).
2. Section 4.2 extends the standard binomial model of Cox et al. (1979), to
account for the effects of distributions to the underlying asset.
Introduction
This chapter deals with the valuation of derivative assets using the binomial
or the lattice approach. Ironically enough, however, the more complex
approach, namely the Black–Scholes (1973) one, was discovered before the
simple binomial approach. Even if the discrete-time approach is not always
computationally efficient, option valuation with the lattice approach is very
flexible. It can handle many situations where no analytical solutions are
possible. This is the case, for example, for the valuation of stock options
and index options, when there are several discrete cash payouts made by
the underlying asset, such as cash dividends on a stock.
Various numerical procedures have been proposed by researchers
for the pricing of derivative securities. These procedures include the
lattice approach, finite difference schemes, and the Monte-Carlo method,
among others. When pricing European options, these procedures are
asymptotically equivalent to closed-form solutions à la Black and Scholes
(1973). However, binomial methods may be more practical for the pricing
221
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
At the end of the year, the underlying stock can increase by 20%, from
40 to (40 × 1.2), or 48, as it can decrease by the same amount from 40 to
(40 × 0.8), or 32 as in Fig. 4.1.
uS = 48
S = 40
dS = 32
Fig. 4.2.
uS − HC = 32
S − HC
dS − HC = 32
(Cu − Cd ) = (8 − 0) = 0.
When the stock price increases, the value of the hedge portfolio is:
When the stock price decreases, the value of the hedge portfolio is:
Since the initial portfolio value is (S − HC), its final value must be
multiplied by the risk-less rate since it is a hedge portfolio. The value of
the hedged portfolio at the maturity date becomes R(S − HC). In order to
avoid risk-less arbitrage, we must have R(S − HC) = (uS − HCu ), which
gives:
S(R − u) + Hcu
C=
HR
S(u−d)
Since the value of H is given by H = Cu −Cd , then
(R − d) (u − R)
C = Cu + Cd R
(u − d) (u − d)
with
p uS
(1−p) dS
√
In a risk-neutral world, the expected value of S is Ser ∆t
. The expected
value can also be calculated as follows:
Simplifying by S gives:
√
er ∆t
= pu + (1 − p)d (4.3)
The variance of S over the same time interval ∆t is σ 2 S 2 ∆t, since the
variance of a random variable X is given by:
E(X 2 ) − E(X)2
Using Eqs. (4.3), (4.4), and u = 1/d, it is possible to show that the
following relationships are verified:
√ √ √
u = er ∆t
, d = er ∆t
, m = er ∆t
, p = (m − d)/(u − d) (4.5)
At each node, the underlying asset value can be written as Suj di−j ,
for j varying from 0 to i. The first index i corresponds to the period
and the second index j indicates the position. For example, when the
option’s maturity date is in one period, i = 1 and j varies from 0 to i, i.e.,
0 to 1. Using 0 for the lowest position at each period, when the underlying
asset value decreases, we have Su0 d1−0 = Sd. When it increases, we have
Su1 d1−1 = Su. The value of an European or an American option at each
pair (i, j) is denoted by Fi,j . The option price at time 0 can be computed
by a recursion starting from the maturity date T . The option price is given
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
where K is the strike price. Authors use E or K to denote the strike price.
We use both E and K for the strike price.
The option value at each node can be computed using the two
immediate successive nodes. The expected value must be discounted using
the risk-less rate as follows:
√
Fi,j = er ∆t
[pFi+1 ,j+1 +(1 − p)Fi+1,j ] (4.6)
for 0 ≤ i ≤ M − 1 and 0 ≤ j ≤ i.
Since the value of an American call option must be at least equal to its
intrinsic value, the following condition must be satisfied:
√
Fi,j = max[Suj di−j − K, er ∆t
(pFi+1,j+1 + (1 − p)Fi+1,j )] (4.7)
This model appears in Cox et al. (1979), Cox and Rubinstein (1985),
Boyle (1986, 1988) and Hull and White (1993) etc.
In the first step, the values of the model parameters must be computed:
√ √ √
u = e−σ ∆t = 1.1519, d = e−σ ∆t , = e−0.2 1/2 = 0.8681,
√ √
m = er ∆t = e−0.1 1/2 = 1.0732, p = (m − d)/(u − d) = 0.7227.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
For an initial value of S = 100, the two possible values in the next
period are:
These two values of the underlying asset price lead to three possible
values, as shown in Fig. 4.5:
Suu = 132.68
Su = 115,19
S = 100 Sud = 100
Sd = 86, 81
Sdd = 75.36
S2,2 = 132.68
S1,1 = 115,19
S0,0 = 100 S2,1 = 100
S1,0 = 86,81
S2,0 = 75.36
√
C1,1 = e−r ∆t
[pC2,2 + (1 − p)C2,1 ],
or
√
C1,0 = e−r ∆t
= [pC2,1 + (1 − p)C2,0 ], or
C1,0 = e−0,1(1/2) [0.7227(22.465) + (1 − 0.7227)0] = 0
Using the possible values in one period, what is the option price at
time 0?
Using the same formula, C0,0 is given by:
√
C0,0 = e−r ∆t
[pC1,1 + (1 − p)C1,0 ], or
C0,0 = e−0,1(1/2) [0.7227(0) + (1 − 0.7227)0] = 15.443
152.84
u²S
123.63
u
uS d
S
S 00 d u 100
dS
84.17 d
80.88 d ²S
65.41
C 22
C 11
C 21
C 00
C 10
C 20
P 22
P11
P 21
P 00
P10
P 20
Hence, the option price is 8.021. We can check that the put–call parity
theorem is verified.
0.5064 ∗ 52.84 + 0.4936 ∗ 0
C1,1 = max ; 23.63 = 26.105
1.025
0.5064 ∗ 0 + 0
C1,0 = max ; 0 =0
1.025
0.5064 ∗ 26.105
C0,0 = = 12.88
1.025
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
167.71
3
uS
141.16
u²S 118.81
118.81
uS uS
S 00 S dS
100 84.17
dS
84.17 70.85
d ²S
3
dS
59.63
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
178.1312
158.7055
141.3982 141.3982
125.9784 125.9784
112.2401 112.2401 112.2401
100 100 100
89.0947 89.0947 89.0947
79.3787 79.3787
70.7222 70.7222
63.0098
56.1384
0
0
0 0
1.2720 0
4.2282 2.6033 0
8.6380 7.3442 5.3282
13.3256 12.3506 10.9053
19.7110 19.7914
27.6249 29.2778
36.1603
43.8616
0
0
0 0
1.2720 0
4.3250 2.6033 0
8.9769 7.5423 5.3282
13.9195 12.7561 10.9053
20.7226 20.6213
29.2778 29.2778
36.9902
43.8616
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
78.1312
59.5354
43.0511 41.3982
29.7194 26.8082
19.7467 16.4963 12.2401
12.7191 9.8132 6.1581
5.6987 3.0982 0
1.5587 0
0 0
0
0
78.1312
59.5354
43.0511 41.3982
29.7194 26.8082
19.7467 16.4963 12.2401
12.7191 9.8132 6.1581
5.6987 3.0982 0
1.5587 0
0 0
0
0
142.5050
126.9644
113.1186 113.1186
100.7827 100.7827
89.7921 89.7921 89.7921
80 80 80
71.2758 71.2758 71.2758
63.5030 63.5030
56.5778 56.5778
50.4078
44.9107
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
0
0
2.4369 0
7.0283 4.9875
12.9178 11.8756 10.2079
19.3423 19.2016 19.1701
26.2863 27.0714 28.7242
34.0280 35.6672
41.7694 43.4222
48.7623
55.0893
0
0
2.4369 0
7.2265 4.9875
13.5117 12.2811 10.2079
20.8323 20.2132 20
28.7242 28.7242 28.7242
36.4970 36.4970
43.4222 43.4222
49.5922
55.0893
42.5050
27.7943
17.2083 13.1186
10.2801 6.6001
3.3206 0
3.4234 1.6706 0
0.8405 0 0
0 0
0 0
0
0
42.5050
27.7943
17.2083 13.1186
10.2801 6.6001
3.3206 0
3.4234 1.6706 0
0.8405 0 0
0 0
0 0
0
0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
Portfolio A: one call option plus a discount bond that will be worth K at
time T and
Portfolio B: one put option plus one share.
c − p ± (put–call parity) = R − D
are accounted for, it can be shown that a sufficient condition for optimal
exercise is:
with (Di +Ri ) ≥ 0. Generally, a put is exercised when it is in the money and
the call price is less than the cash amount. Formally, the put is exercised
at time ti if:
with (Di + Ri ) ≥ 0.
√ √
u = eσ ∆t
, u = e0.4 1/2 , or
1 er∆t − d
u = 1.1224, d= = 0.8909, p= = 0.5073
u u−d
S ∗ = S − D = 110 − 10 = 100
S1,1 = S ∗ u1 d0 + De−r((105/365)−(1/12))
S5,0 = S ∗ u0 d5 = 56.1386
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
S 5,5
S 4,4
S 5,4
S 3,3
S 4,3
S 2,2
S 3,2 S 5,3
S 1,1
S 2,1 S 4,2
S 00
S 5,2
S 3,1
S 1,0
S 4,1
S 2,0 S 5,1
S 3,0
S 4,0
S 5,0
p · C5,3 + q · C5,2
C4,2 = max ; max[0; S4,2 − K] = 0
er∆t
p · C5,2 + q · C5,1
C4,1 = max ; max[0; S4,1 − K] = 0
er∆t
p · C5,1 + q · C5,0
C4,0 = max ; max[0; S4,0 − K] = 0
er∆t
p · C4,4 + q · C4,3
C3,3 = max ; S3,3 − K
er∆t
= max[28.9563; 36.3603] = 36.3603
p · C4,3 + q · C4,2
C3,2 = max ; S3,2 − K = max[6.6813; 7.2024] = 7.2024
er∆t
p · C4,2 + q · C4,1
C3,1 = max ; S3,1 − K = 0,
er∆t
p · C4,1 + q · C4,0
C3,0 = max ; S 3,0 − K =0
er∆t
63.105
44.6586
36.3603 26.3979
21.8117
13.2805
7.2024 0
12.7437
3.6235 0
7.3019 0
0
1.8229
0 0
0
0
0
0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
p · C3,3 + q · C3,2
C2,2 = max ; S2,2 − K
er∆t
= max[21.8117; 20.8579] = 21.8117
p · C3,2 + q · C3,1
C2,1 = max ; max[0; S2,1 − K]
er∆t
= max[3.6235; 0] = 3.6235
p · C3,1 + q · C3,0
C2,0 = max ; max[0; S2,0 − K] = 0
er∆t
p · C2,2 + q · C2,1
C1,1 = max ; max[0; S 1,1 − K]
er∆t
= max[12.7437; 7.0377] = 12.7437
p · C2,1 + q · C2,0
C1,0 = max ; max[0; S 1,0 − K]
er∆t
= max[1.8229; 0] = 1.8229
p · C1,1 + q · C1,0
C0,0 = max ; max[0; S 0,0 − K]
er∆t
= max[7.3019; 0] = 7.3019
p · P5,3 + q · P5,2
P4,2 = max ; max[0; K − S4,2 ] = max[14.0461; 0] = 15
er∆t
p · P5,2 + q · P5,1
P4,1 = max ; max[0; K − S 4,1 ]
er∆t
= max[34.6672; 35.6212] = 35.6212
p · P5,1 + q · P5,0
P4,0 = max ; max[0; K − S 4,0 ]
er∆t
= max[51.0360; 51.9900] = 51.9900
0
0.6589
1.3486
4.2441
8.0076 2.76
10.0605
16.2201 15
17.0991
25.9052
24.9513
24.6367
35.6212
33.7211 44.2776
43.3236
51.9900
58.8614
p · P4,4 + q · P4,3
P3,3 = max ; max[0; K − S3,3 ]
er∆t
= max[0.6589; 0] = 0.6589
p · P4,3 + q · P4,2
P3,2 = max ; max[0; K − S3,2 ]
er∆t
= max[8.0076; 0] = 8.0076
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
p · P4,2 + q · P4,1
P3,1 = max ; max[0; K − S3,1 ]
er∆t
= max[24.9513; 15.9428] = 24.9513
p.P4,1 + q · P4,0
P3,0 = max ; max[0; K − S3,0 ]
er∆t
= max[43.3236; 34.3152] = 43.3236
p · P3,3 + q · P3,2
P2,2 = max ; max[0; K − S2,2 ]
er∆t
= max[4.2441; 0] = 4.2441
p · P3,2 + q · P3,1
P2,1 = max ; max[0; K − S2,1 ]
er∆t
= max[16.2201; 5.1203] = 16.2201
p · P3,1 + q · P3,0
P2,0 = max ; max[0; K − S2,0 ]
er∆t
= max[33.7211; 25.7414] = 33.7211
p · P2,2 + q · P2,1
P1,1 = max ; max[0; K − S1,1 ]
er∆t
= max[10.0605; 0] = 10.0605
p · P2,1 + q · P2,0
P1,0 = max ; max[0; K − S 1,0 ]
er∆t
= max[24.6367; 16.1075] = 24.6367
p · P1,1 + q · P1,0
P0,0 = max ; max[0; K − S 0,0 ]
er∆t
= max[17.0991; 5.2836] = 17.0991
178.1312
158.7055
151.3606 141.3982
135.8581 125.9784
122.0378 122.2025 112.2401
110 109.8797 100
98.8925 99.0571 89.0947
89.2584 79.3787
80.6846 70.7222
63.0098
56.1384
0
0
0 0
1.2720 0
4.2282 2.6033 0
8.6380 7.3442 5.3282
13.3256 12.3506 10.9053
19.7110 19.7914
27.6249 29.2778
36.1603
43.8616
0
0
0 0
1.2720 0
4.3250 2.6033 0
8.8801 7.5423 5.3282
13.7214 12.7561 10.9053
20.3171 20.6213
28.4479 29.2778
36.9902
43.8616
78.1312
59.5354
43.0511 41.3982
29.7194 26.8082
19.7467 16.4963 12.2401
12.7191 9.8132 6.1581
5.6987 3.0982 0
1.5587 0
0 0
0
0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
78.1312
59.5354
51.3606 41.3982
36.6880 26.8082
24.6554 22.2025 12.2401
15.8944 12.6840 6.1581
7.1430 3.0982 0
1.5587 0
0 0
0
0
Call price S = 18 S = 19 S = 20 S = 21
Call price S = 22 S = 25 S = 28 S = 30
Call price S = 18 S = 19 S = 20 S = 21
S = 22 S = 25 S = 28 S = 30
CRR
K σ = 0.244 σ = 0.304
CRR
K σ = 0.244 σ = 0.304
Summary
Cox et al. (1979) proposed the first discrete-time model for the pricing of
stock options. This binomial model is used for the valuation of options on
different underlying assets.
Questions
1. Describe the Cox et al. (1979) model for equity options for one period.
2. Describe the Cox et al. (1979) model for equity options for several
periods.
3. How can we implement a hedging strategy in this context?
4. What are the valuation parameters in the lattice approach for stock
prices?
5. How is an option priced in the lattice approach for stock prices?
6. What modifications are necessary to the standard lattice approach to
apply it to American options?
7. What are the effects of cash distributions on the stock price?
one of its two new values Su and Sd with probabilities p and (1 − p). When
u = 1/d, it can be shown that:
a−d √ √
p= , u = eσ ∆t
, d = e−σ ∆t
, a = er∆t .
u−d
The nature of the lattice of stock prices is completely specified and the
nodes correspond to Suj di−j for j = 0, 1, . . . , i. The option is evaluated by
starting at time T and working backward. Let us denote by Fi,j , the option
value at time t + i∆t when the stock price is Suj di−j . At time t + i∆t, the
option holder can choose to exercise the option and receives the amount
by which K (or S) exceeds the current stock price (or K) or wait. The
American call is given by:
Exercises
Example 1
Consider the valuation of European and American options in the following
context:
125.9784
118.9110 118.9110
112.2401 112.2401
105.9434 105.9434 105.9434
100 100 100
94.3900 94.3900 94.3900
89.0947 89.0947
84.0965 84.0965
79.3787
74.9256
0
0
0 0
0.6062 0
2.0783 1.2727 0
4.3771 3.7021 2.6720
6.9229 6.3844 5.6100
10.4965 10.4895
15.0736 15.9035
20.2055
25.0744
0
0
0 0
0.6062 0
2.1232 1.2727 0
4.5368 3.7964 2.6720
7.2092 6.5824 5.6100
10.9947 10.9053
15.9035 15.9035
20.6213
25.0744
33.4658
26.3942
19.7409 18.9110
14.0885 12.6559
9.6745 8.0460 5.9434
6.4389 4.9443 3.0878
2.9658 1.6043 0
0.8335 0
0 0
0
0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
33.4658
26.3942
19.7409 18.9110
14.0885 12.6559
9.6745 8.0460 5.9434
6.4389 4.9443 3.0878
2.9658 1.6043 0
0.8335 0
0 0
0
0
Example 2
Consider the valuation of European and American options in the following
context:
Underlying asset, S = 100, strike price K = 100, interest rate = 0.05,
volatility = 0.2, T = 5 months, N = 5, dividend = 10, and dividend date:
in 105 days. In this case, we have: p = 0.5217, d = 0.9439, u = 1.0594.
125.9784
128.8922 118.9110
122.1798 112.2401
115.8418 115.9246 105.9434
110 109.9397 100
104.2884 104.3712 94.3900
99.0344 89.0947
94.0777 84.0965
79.3787
74.9256
0
0
0 0
0.6062 0
2.0783 1.2727 0
4.3771 3.7021 2.6720
6.9229 6.3844 5.6100
10.4965 10.4895
15.0736 15.9035
20.2055
25.0744
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
0
0
0 0
0.6062 0
2.1232 1.2727 0
4.4919 3.7964 2.6720
7.1149 6.5824 5.6100
10.7967 10.9053
15.4877 15.9035
20.6213
25.0744
33.4658
26.3942
19.7409 18.9110
14.0885 12.6559
9.6745 8.0460 5.9434
6.4389 4.9443 3.0878
2.9658 1.6043 0
0.8335 0
0 0
0
0
33.4658
26.3942
28.8922 18.9110
22.5956 12.6559
16.6716 15.9246 5.9434
11.7393 10.3555 3.0878
6.4618 4.3712 0
2.2710 0
0 0
0
0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
Appendix
Simulations:
At the money call and put options using the Black–Scholes Model.
Stock price (S) Strike price (K) Interest rate (r) Maturity (t) Volatility
At the money call (Out of the Money) put options using the Black–Scholes
Model.
Stock price (S) Strike price (K) Interest rate (r) Maturity (t) Volatility
Out of the money call (In the Money) put options using the Black–Scholes
Model.
Stock price (S) Strike price (K) Interest rate (r) Maturity (t) Volatility
Last Trade
Xt: random T=3
Timestamp Close Return Mean Volatility m−1/2σ 2 numbers Days simulation VT= VT−V0
spi-b708
19/07/2007 92,00 0,00 Kurtosis 4,2988 −0,77629 92,072 100021 20,8743
20/07/2007 92,00 0,00 0,21629 92,1333 99954,3 −45,6531
23/07/2007 91,00 −0,0109 0,068 7,495 0,81376 92,1702 99914,3 −85,6764
24/07/2007 91,00 0,00 1,66880 92,2231 99857,1 −142,9266
26/07/2007 91,00 0,00 0,07546 92,1246 99963,8 −36,2167
27/07/2007 91,00 0,00 1,45689 92,21 99871,3 −128,7412
30/07/2007 91,00 0,00 1,44571 92,2093 99872 −127,9926
31/07/2007 91,00 0,00 −1,20244 92,0457 100049 49,4498
9in x 6in
01/08/2007 91,98 0,0108 −0,73838 92,0744 100018 18,3326
03/08/2007 91,50 −0,0052 0,34935 92,1415 99945,4 −54,5677
06/08/2007 91,50 0,00 −0,63400 92,0808 100011 11,3346
07/08/2007 92,10 0,0066 −1,86099 92,0051 100094 93,6258
08/08/2007 94,00 0,0206 0,60674 92,1574 99928,2 −71,8104
10/08/2007 92,00 −0,0213 −0,52731 92,0874 100004 4,1820
14/08/2007 92,00 0,00 −1,24315 92,0432 100052 52,1802
15/08/2007 92,02 0,0002 −1,22724 92,0442 100051 51,1130
b708-ch04
16/08/2007 92,00 −0,0002 −0,38141 92,0964 99994,4 −5,5982
17/08/2007 92,00 0,00 0,78272 92,1683 99916,4 −83,5978
255
(Continued)
September 10, 2009 14:41
St+h mean PerCentile VaR
256
92,12478 99816 184,1338
Last Trade
Xt: random T=3
Timestamp Close Return Mean Volatility m−1/2σ 2 numbers Days simulation VT= VT−V0
spi-b708
29/08/2007 94,00 0,00 −0,42598 92,0937 99997,4 −2,6107
30/08/2007 94,00 0,00 −0,02322 92,1185 99970,4 −29,6041
31/08/2007 92,54 −0,0155 0,94055 92,1781 99905,8 −94,1681
03/09/2007 92,54 0,00 −1,51721 92,0263 100071 70,5621
04/09/2007 92,50 −0,0004 −1,02602 92,0566 100038 37,6191
05/09/2007 92,70 0,0022 −0,33977 92,099 99991,6 −8,3890
06/09/2007 92,62 −0,0009 −1,09711 92,0522 100042 42,3864
9in x 6in
07/09/2007 92,70 0,0009 0,12028 92,1274 99960,8 −39,2197
10/09/2007 92,50 −0,0022 0,77114 92,1676 99917,2 −82,8218
11/09/2007 92,30 −0,0022 −0,49255 92,0896 100002 1,8517
12/09/2007 92,00 −0,0033 0,40264 92,1448 99941,9 −58,1378
13/09/2007 92,00 0,00 −0,15155 92,1106 99979 −21,0041
14/09/2007 92,00 0,00 −0,77919 92,0719 100021 21,0686
17/09/2007 91,98 −0,0002 −0,69984 92,0768 100016 15,7487
18/09/2007 92,00 0,0002 −0,35008 92,0983 99992,3 −7,6979
b708-ch04
19/09/2007 92,00 0,00 −1,21243 92,0451 100050 50,1198
20/09/2007 92,00 0,00 1,61324 92,2196 99860,8 −139,2075
(Continued)
September 10, 2009 14:41
Option Pricing: The Discrete-Time Approach for Stock Options
St+h mean PerCentile VaR
92,12478 99816 184,1338
Last Trade
Xt: random T=3
Timestamp Close Return Mean Volatility m−1/2σ 2 numbers Days simulation VT= VT−V0
spi-b708
24/09/2007 92,00 −0,0017 0,84954 92,1724 99911,9 −88,0730
25/09/2007 92,00 0,00 −0,43311 92,0932 99997,9 −2,1327
26/09/2007 91,98 −0,0002 1,74528 92,2278 99852 −148,0461
27/09/2007 93,00 0,0111 0,07017 92,1243 99964,1 −35,8621
01/10/2007 92,00 −0,0108 0,42573 92,1463 99940,3 −59,6846
02/10/2007 92,00 0,00 0,39824 92,1446 99942,2 −57,8435
03/10/2007 92,00 0,00 −0,83644 92,0683 100025 24,9072
9in x 6in
04/10/2007 92,98 0,0107 −0,36631 92,0973 99993,4 −6,6100
05/10/2007 92,00 −0,0105 1,84818 92,2342 99845,1 −154,9327
08/10/2007 92,00 0,00 −2,30586 91,9777 100123 123,4789
09/10/2007 92,10 0,0011 −0,15914 92,1101 99979,5 −20,4955
10/10/2007 92,00 −0,0011 −0,11042 92,1131 99976,2 −23,7604
11/10/2007 92,12 0,0013 −1,18806 92,0466 100048 48,4856
b708-ch04
257
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch04
References
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–654.
Boyle, P (1986). Option valuation using a three jump process. International
Options Journal, 3, 7–12.
Boyle, PP (1988). A lattice framework for option pricing with two state variables.
Journal of Financial and Quantitative Analysis, 23 (March), 1–12.
Briys, E, M Bellalah, F de Varenne and H Mai (1998). Options, Futures and Other
Exotics. New York: John Wiley and Sons.
Cox, J, S Ross and M Rubinstein (1979). Option pricing: a simplified approach,
Journal of Financial Economics, 7, 229–263.
Hull, J and A White (1993). Efficient procedures for valuing European and
American path dependent options. Journal of Derivatives, 1, Fall 1993,
21–31.
Hull, J, A White (1988). An analysis of the bias in option pricing caused by a
stochastic volatility. Advances in Futures and Options Research, 3, 29–61.
Hull, J (2000). Options, Futures, and Other Derivative Securities. New Jersey:
Prentice Hall International Editions.
Jarrow, RA and A Rudd (1983). Option Pricing. Homewood, IL: Irwin.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
Rendleman, RJ and BJ Barter (1980). The pricing of options on debts securities.
Journal of Financial and Quantitative Analysis, 15 (March), 11–24.
Rubinstein, M (1994). Implied binomial trees. Journal of Finance, No 3, 771–818.
Whaley, RE (1986). Valuation of American futures options: theory and empirical
tests. Journal of Finance, 41 (March), 127–150.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Chapter 5
Chapter Outline
This chapter is organized as follows:
Introduction
The price of a bond depends on the future coupons and the notional amount.
The price corresponds to the present value of all the future cash flows
259
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
While I and F occur at one time, the fixed amount A occurs at each
interest period for a given or a specified number of periods. A net cash flow
refers to the difference between receipts (income) and cash disbursements
(costs).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 261
F1 = I + rI = I(1 + r)
Fn = I(1 + r)n
or
1
I=F (5.2)
(1 + r)n
1 1 1 1
I=A +A 2
+A 3
+ ··· + A
(1 + r) (1 + r) (1 + r) (1 + r)n
or:
n
1 1 1 1 1
I=A + + + ··· + =A
(1 + r) (1 + r)2 (1 + r)3 (1 + r)n i=1
(1 + r)i
or:
−r 1 1
I =A −1
1+r (1 + r)n 1+r
n
1 (1 + r)n − 1 1 − (1 + r)−n
= =
(1 + r)i r(1 + r) n r
i=1
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 263
1
I=F
(1 + r)n
The discounting factor [r/((1 + r)n − 1)] is the sinking fund factor (SFF).
It is denoted by (A/F, r%, n). This equation allows the computation of the
uniform series A that starts at the end of period 1 and continues through the
period of a specified F . Equation (5.5) can also be written in the following
way:
The term in brackets refers to the USCAF. It is denoted by (F/A, r%, n).
When this factor is multiplied by a given uniform amount A, this gives the
future worth of the uniform series. The factors (a/b, r, n) in formulas (5.1)
to (5.6) allows one to find the value of (a) when (b) is given for a specified
interest rate at a given number of periods.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Applications
Using Eq. (5.1), when r = 6%, n = 8 years, formula (5.1) gives the factor:
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 265
F = 1000(1.12) = 1120
If the bank pays an interest computed for every six months, the future
value must account for the interest on the interest earned. When the annual
interest rate is 12%, this means that the bank will pay 6% interest two
times a year. In the presence of a 6% effective semi-annual interest rate,
the future value is: 1000(1 + 0.06)2 = 1123, 6. Hence, the effective annual
interest rate is 12.36% rather than 12%. The following relationship shows
the link between nominal interest rates and effective interest rates:
r = (1 + im )m − 1
or:
r = (1 + (12%/2))2 − 1 = 12.36%
where r is the effective interest rate per period, i is the nominal interest
rate per period, and m stands for the number of compounding periods.
This relation represents the effective interest rate equation. This equation
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Lim (1 + 1/h)h = e
h→∞
We have: r = ei − 1.
Example: If i = 20% (annual), the effective continuous rate is:
r = e0,2 − 1 = 22,1408%.
For more details, see Blanck and Tarquin (1989) and Bellalah (1991,
1998a, b).
where:
The coupon corresponds to the interest rate times, the nominal value
of the bond. In several countries, semi-annual coupons are paid every
six months. In other countries, coupons are paid annually. For semi-
annual coupon payments, the periodic interest rate used in the discounting
procedure must be the required yield divided by two. Using standard
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 267
formulas for the annuity, it is possible to write the bond price in the
following form:
1
1 − (1+r) n M
B = c +
r (1 + r)n
When calculating the time value of money, the following equality is often
used to facilitate the computations:
1
n
1 1 − (1+r) n
=
t=1
(1 + r)t r
where CFt refer to the cash flows, y to the yield and t denotes the number
of years from year 1 to n. The yield y is calculated in general using a trial
and error procedure.
1 120
2 120
3 120
4 120
5 1000
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 269
5.3.2. The CY
The CY gives an indication of the relation between the annual coupon
interest and the market bond price as:
the current coupons, their timing and the capital gain or loss from holding
the bond until maturity.
where
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 271
or:
C[(1 + r)n − 1]
Ip =
r
where the total coupon amount is given by the semi-annual coupon interest
times the number of periods or:
C[(1 + r)n − 1]
Interest on interest = − nC
r
This analysis assumes that re-investment of the coupons is done at
the YTM.
c c c M
B= + + ···+ +
(1 + y) (1 + y)2 (1 + y)n (1 + y)n
where c and M stand respectively for the coupon amount and the principal.
To compute the duration of a bond one needs to examine the sensitivity
of the bond price to the yield. This variation is given by:
dB −1 −2 −n −n
=c +c +· · ·+c +M
dy (1 + y)2 (1 + y)3 (1 + y)n+1 (1 + y)n+1
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
or:
dB 1 1c 2c nc nM
=− + + ··· + + (5.7)
dy (1 + y) (1 + y) (1 + y)2 (1 + y)n (1 + y)n
The term between brackets shows that each cash flow is weighted by
its “maturity”. If both sides of Eq. (5.7) are divided by the bond price P ,
we obtain the percentage price change for a small variation in y:
dB 1 1
=−
dy B (1 + y)
c 2c nc nM 1
× + + ···+ +
(1 + y) (1 + y)2 (1 + y)n (1 + y)n P
The term between brackets divided by the bond price is known as the
duration of Macaulay. The amounts of the coupons can be variable with
different amounts of ci. In this case, the bond duration can be written as:
c1 c2 c3 cn +M
(1+y)1 (1+y)2 (1+y)3 (1+y)n
D=1 +2 +3 + ··· + n
B B B B
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 273
Dp = w1 D1 + w2 D2 + +wn Dn
or:
Dp = Di wi
i
where:
P vi
wi =
i P vi
where
Dp : Macaulay duration for a portfolio of bonds;
N : number of bonds in the portfolio;
P vi : present value (market price) of the ith bond and
wi : market value of bond i divided by the market value of the portfolio.
So, we have:
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 275
The first term of this equation refers to the dollar price change based
on dollar duration. The second term corresponding to the second derivative
can be used as a proxy for the convexity of the price-yield relationship. The
dollar convexity of the bond is given by:
d2 B
Dollar convexity = 0.5
dy 2
dB = (dollar convexity)(dy)2
When the second derivative of the bond price is divided by the price,
this gives a measure of the percentage change in the bond price due to
convexity:
d2 B 1
Convexity =
dy 2 B
dB
= (convexity)(dy)2
B
In practice, using the second derivative of the bond price with respect
to y gives the following equation which is used to compute the convexity:
d2 B ct(t + 1)
n
n(n + 1)M
= +
dy 2 t=1
(1 + y)t+2 (1 + y)n+2
For more details, see Fabozzi (1996) and Briys et al. (1998).
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 277
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 279
C C C (C + 100)
Bn = + 2
+ 3
+ ···+
(1 + y1 ) (1 + y2 ) (1 + y3 ) (1 + yn )n
where
n1
(C + 100)
yn =
n−1 1
−1
Bn − C − t=1 (1+yt )t
These two strategies are equivalent if they give the same result over
the one-year investment horizon. The knowledge of the spot rates on a six-
month and a one-year Treasury bills allows the computation of the yield on
a six-month Treasury bill, six months from now or the forward rate.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
B = 100/(1 + Y2 )2
B(1 + Y1 )
B(1 + Y1 )(1 + f2 )
The result of investing in an asset that gives 100 in one year can be
written as:
or:
B = 100/[(1 + Y1 )(1 + f2 )]
The two strategies are equivalent for the investor if they lead to the
same result or:
f2 = [(1 + Y2 )2 /(1 + Y1 )] − 1
Example: When the six-month and the one-year Treasury bill rates are
respectively equal to 8.5% and 8.9%, then Y1 = 4,25%, Y2 = 4,45%, and
the implied forward rate is 4,65%, or:
f2 = [(1,0445)2/1,0425] − 1 = 4,65%
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 281
Issuer: Company X;
Nominal amount = 10, 000, 000;
Nominal value of each bond = 2000;
Number of bonds = 5000;
Issue price = 980, or 98% of the nominal amount;
Maturity date = 10 years;
Payment date = 01/01/1994;
Maturity date = 01/01/2004;
Interest rate = 10% and
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Coupon amounts: the first coupon of 2000 is paid the first January 1995
or (2000)(10%)(1) = 2000.
The same coupon amount is paid each year at the same date
Payments:
1000 bonds re-paid the first January 2000 at 100,1% of the nominal value
or 2002;
1000 bonds re-paid the first January 2001 at 100,3% of the nominal value
or 2006;
1000 bonds re-paid the first January 2002 at 100,5% of the nominal value
or 2010;
1000 bonds re-paid the first January 2003 at 100,7% of the nominal value
or 2014 and
1000 bonds re-paid the first January 2004 at 100,9% of the nominal value
or 2018.
Table 5.3 shows the re-payment of the issue for the borrower.
Table 5.3. The re-payment profile of the issue for the borrower: issue date, the first
January 1994.
5 1000 0 0 1000 5
5 1000 0 0 1000 5
5 1000 0 0 1000 5
5 1000 0 0 1000 5
5 1000 0 0 1000 5
5 1000 0 2002 (2002)103 3002 4
4 800 1 2006 (2006)103 2806 3
3 600 1 2010 (2010)103 2610 2
2 400 1 2014 (2014)103 2414 1
1 200 1 2018 (2018)103 2218 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 283
Table 5.4. The profile of cash flows for the bondholder (the investor lending money):
initial investment at time 0: 19,800 ((1980)(10)).
From the year 1995 to 1999, he pays each year 1,000,000 of coupons.
From year 6, (the year 2000), he pays the coupons and makes principal
re-payment until year 2004. At this latter date, he pays 200,000 of coupons
and re-pays 2,018,000 of the principal. The total amount is 2,218,000.
The profile of the cash flows for the bondholder depends on the payment
dates.
Consider an investor who buys the first January 1994, 10 bonds of the
firm X. The issuer makes payment for the first five bonds the first January
2001 and for the other five bonds the first January 2004. The profile of cash
flows for the bondholder is given in Table 5.4. With (6∗ ) number of bonds
alive at the end of the year in 103 (1)−(6).
fixed annual interest rate times the outstanding mortgage balance amount
at the beginning of the previous month. The second element corresponds
to the re-payment of a fraction of the outstanding mortgage balance or the
principal. The mortgage payment is defined in a way such that after the last
fixed monthly payment, the amount of the outstanding mortgage balance
is zero.
The first month, the mortgage balance corresponds to the interest rate
for the month on the 150,000 Euros borrowed or (10%/12) 150,000 = 1250
Euros. The monthly mortgage payment that represents re-payment of the
principal corresponds to the difference between the monthly mortgage
payment and the interest rate. The last monthly mortgage payment is
sufficient to pay off the remaining mortgage balance.
The following formula is used to compute the monthly mortgage
payment for a level-payment fixed-rate mortgage.
1 − (1+r)1
n
PV = A
r
where:
Using the above P V formula and the data in the example, we have n =
180, VA = 150,000 Euros, r = 0,1/12, the monthly mortgage payment is
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 285
Table 5.5. Amortization schedule for a level-payment fixed-rate mortgage using the
following parameters: term of loan n = 180, mortgage loan = 150,000 Euros, and
mortgage rate = 0.1/12.
1611,907677:
150000 = 1611,907677.
1
1− (1,00833)180
0,008333
Sw = B1 − B2
9.6% 9.5%
Firm A Commercial bank Firm B
LIBOR LIBOR
Fig. 5.1. An interest rate swap.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 287
where
C: semi-annual coupon;
Ti : time until the ith coupon payment where 1 ≤ i ≤ n;
ri : risk-less rate for ti and
N : the notional amount.
Since the value of B2 is based on a variable interest rate, its price after
each coupon date must be equal to the discounted value of the notional
amount and the last coupon. Hence, the value B2 between two payment
dates is:
B2 = N e−r1 t1 + C1 e−r1 t1
Example 2
Two firms A and B are engaged in a swap. The bank receives a commission
of 25 basis points on the principal amount denominated in dollars as in
Fig. 5.2.
Firm A pays in dollars and receives sterling. If the NA is 30 million
dollars and 20 million sterling, each year, A pays 2.4 million dollars (8%)
(30 million) and receives 2.2 million sterling (11%) (20 million). At the swap
term, firm A pays 30 million dollars of principal and 20 million sterling. The
swap allows A to convert a fixed-rate loan denominated in sterling, (11%
per year) into a fixed-rate loan denominated in dollars (8% per year). Firm
B converts a fixed-rate loan in dollars (7.75% per year) into a fixed-rate
loan in sterling (11% per year).
If, for example, the loan for A (in sterling) is at a rate of 11% and B
(in dollars) at 9%, the swap can be presented as in the Fig. 5.3.
The loan at 11% for A is converted into a dollar loan at 8% and the
loan for B in dollars at 9% is converted into a loan in sterling at 12.25%. If
the exchanged amounts are 20 million dollars and 30 million sterling, the
(dollars)8% (dollars)7.75%
Firm A Bank Firm B
11%(pound) 11%(pound)
(dollars)8% (dollars)9%
Firm A Bank Firm B
11%(sterling) 12.25%(sterling)
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 289
bank loses 3,00,000 dollars per year and gains 2,50,000 sterling. The bank
can implement a hedge against exchange rate risk by selling forward the
2,50,000 sterling against dollars.
Summary
The yield on an investment can be computed by determining the interest
rate that makes the present value of all future cash flows equal to the
initial price. The YTM is calculated in the same way as the internal rate of
return for an investor, holding the bond until maturity. In bond analysis, the
coupons received by the investor can be re-invested at a specified rate giving
rise to a dollar return from coupon interest and interest on interest. The
concept of duration is introduced by Macaulay (Bellalah et al., 1998) as a
proxy for the length of time a bond investment is standing. It is given by the
weighted average term-to-maturity of the cash flows of a bond. A portfolio
duration can be computed using the weighted average of the duration of
the bonds in the portfolio.
Several theories are developed to explain the shape of the yield
curve. The main theories explaining the behavior of interest rates are the
expectations theory and the market segmentation theory. The expectations
theory has several variations: the pure expectations theory, the liquidity
theory, and the preferred habitat theory. In these theories, the market
expectations about future short-term rates can explain the forward rates in
current long-term bonds.
The YTM, corresponds to the average annual rate of return expected
from the purchase of a bond. The YTM refers to a promised return rather
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
than an actual return. It can be computed using one of the following three
methods: the arithmetic mean, the geometric mean and the internal YTM.
The shape of the yield curve is affected by market expectations about
future interest rates. The relationship between the yield on zero-coupon
Treasury securities and maturity is referred to as the Treasury spot-rate
curve where the yield on a zero-coupon bond refers to the spot rate.
Several types of bonds and mortgage securities are issued in the market
place.
When determining the monthly mortgage payments, the formula for
the present value of an ordinary annuity is used. The mortgage payment for
each month is defined with respect to a level-payment fixed-rate mortgage.
A swap position can be seen as a package of forward or futures
contracts.
Since the parties in an interest swap agree to exchange future interest
payments with no upfront payment, this means that the present value of
the cash flows for the payments must be equal. This equivalence allows
the computation of the swap rate. Hence, the swap rate corresponds to an
interest rate that makes equal, the present value of the payments on the
fixed side with the present value of the payments on the floating rate side.
A swaption gives the right to assume a position in an underlying interest
rate swap with a given maturity. In swaptions, the right to pay the fixed
component is equivalent to the right to receive the floating component and
vice versa. Swaptions are offered as receiver swaptions and payer swaptions.
Receiver swaption gives the right to receive a fixed interest rate and payer
swaption gives the right to pay a fixed interest rate.
Using coupon paying bonds with different maturity dates in the
presence of different taxation rates (for capital gains), it is difficult to
observe the term structure of interest rates. Therefore, several techniques
are proposed to estimate the term structure of interest rates. Several
methods are proposed in the literature for the estimation of forward interest
rates. Empirical methods are continuous or discrete.
Questions
1. What are the different types of bonds?
2. What are the specific risks in bond investments?
3. What are the main concepts in the pricing of bonds?
4. What are the main measures that allow investors to calculate the yield
on bonds?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 291
5. What is duration?
6. What is convexity?
7. What are the main theories of interest rates?
8. What are the specific features of spot and forward interest rates?
9. How bonds are issued and redeemed?
10. What are the specific features of mortgage backed securities?
11. What are the specific features of swaps?
12. What are the main techniques used in the estimation models of the
term structure?
References
Blanck, LT and A Tarquin (1989). Engineering Economy. New York: Mc Graw-
Hill.
Bellalah, M (1991). Gestion Quantitative du Portefeuille et nouveaux marchs
financiers. Paris: Editions Nathan.
Bellalah, M (1998a). Gestion financire: diagnostic, valuation et choix des
investissements. Paris: Editions Economica.
Bellalah, M (1998b). Finance d’entreprise: stratgies et politiques financires. Paris:
Editions Economica.
Briys, E, M Bellalah et al. 1998. Options, Futures and exotic Derivatives, en
collaboration avec E. Briys, et al., John Wiley & Sons.
Cox, J, I Ingersoll and S Ross (1985a). An intertemporal General equilibrium
model of Asset prices. Econometrica, 53, 363–384.
Cox, J, I Ingersoll and S Ross (1985b). A Theory of the term structure of interest
rates. Econometrica, 53, 385–407.
Capie, F (1991). Major inflations in History. Vermont: Edward Elgar.
Carleton, W and I Cooper (1976). Estimation and uses of the term structure of
interest rates. Journal of Finance, 31, 1067–1083.
Fabozzi, F (1996). Bond Markets, Analysis and Strategies. New Jersey: Prentice
Hall.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch05
Chapter 6
Chapter Outline
This chapter is organized as follows:
1. Section 6.1 extends the standard binomial model of Cox et al. (1979)
for the valuation of interest-rate sensitive instruments. It develops the
Rendleman and Bartter (for details, refer to Bellalah et al., 1998) model
for the pricing of bonds and bond options.
2. Section 6.2 studies the Ho and Lee (1986) model for the valuation of
bonds and bond options.
3. Section 6.3 shows how to construct interest-rate trees and how to price
bonds and options.
4. Section 6.4 presents a simple derivation of the Black-Derman-Toy model.
5. Section 6.5 shows how to construct trinomial interest-rate trees for the
pricing of bonds and options.
Introduction
This chapter extends the basic lattice approach to the pricing of interest-
rate sensitive instruments and options in the presence of several distribu-
tions to the underlying asset. We are interested in the lattice approach
pioneered by Cox et al. (CRR) (1979). These authors proposed a binomial
model in a discrete-time setting for the valuation of options. Using the
risk-neutral framework, their approach is based on the construction of a
binomial lattice for stock prices. They applied the risk-neutral valuation
argument, pioneered by Black and Scholes (1973), which simply means
293
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
that one can value the option at hand as if investors were risk-neutral.
With an appropriate choice of the binomial parameters, they established a
convergence result of their model to that of Black and Scholes. Since then,
the CRR approach was extended and extensively used for the valuation of
many contingent claims and options.
Rendleman and Bartter (for details, refer to Bellalah et al., 1998) applied
this methodology to the pricing of options on debt instruments. They pro-
posed a binomial approach similar to that of the CRR for the pricing of
derivative assets. The bond’s value at a given node can be computed using the
immediate next two nodes since at a given node, the bond’s value will depend
on the future cash flows. The future cash flows correspond to the bond value
of one year from now and the coupon payments. The binomial model is based
on the recursive procedure starting from the last year and working backward
through the tree till the initial time. The value at each node is given by the
expected cash flows under the appropriate discount rate.
Ho and Lee (1986) proposed a model for the pricing of bonds and
options. However, their model allows for the possibility of negative interest
rates. Ritchken and Boenewan (1990) developed a simple approach to
eliminate the possibility of observing negative interest rates. This analysis is
extended by Pederson et al. (for details, refer to Bellalah et al., 1998). Bliss
and Ronn (for details, refer to Bellalah et al., 1998) and Hull and White
(1988) developed a trinomial model for the pricing of interest-rate sensitive
instruments. An alternative to the Ho and Lee model was proposed by
Black et al. (1990), and Hull and White (1993) among others. Black et al.
(1990) used a binomial tree to construct a one-factor model of the short
rate that fits the current volatilities of all discount bond yields as well as
the current term structure of interest rates.
Rubinstein (1994) developed a new method for inferring risk-neutral
probabilities or option prices from observed market prices. These probabil-
ities were used to infer a binomial tree by implementing a simple backward
recursive procedure. However, this approach is restricted to the European
options, and future research must be done with regard to the pricing of the
American options.
Rendleman and Bartter (for details, refer to Bellalah et al., 1998). This
model is used for the pricing of bonds and interest-rate instruments.1 The
dynamics of the interest rate are described by a two-state process with the
following parameters:
√
u = eσ dt
, d = 1/u, q = e(a−σθ)∆t , and p = (q − d)/(u − d)
To illustrate the Rendleman and Bartter (for details, refer to Bellalah et al.,
1998) model over a period of six years when the current interest rate is equal
to 11%, consider the following data:
a = 0, σ = 0.2, θσ = −0.03, ∆t = 1, n = 6.
36.52
29.90
24.48 24.48
20.04 20.04
16.41 16.41 16.41
13.43 13.43 13.43
11 11 11 11
9 9 9
7.37 7.37 7.37
6.03 6.03
4.94 4.94
4.04
3.31
Year
0 1 2 3 4 5 6
Fig. 6.1. Rendleman and Bartter model for the dynamics of interest rates.
1 In this model, the dynamics of the short-term rate are described by:
dr = (a − θσ)rdt + σrdz
where a, θ and σ are constants and dz is a Wiener process. The term (a − θσ) represents
the drift in the short-term rate.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
This operation is repeated till the end. The rate 3.31 is obtained by
4.04(1/1.2214).
with:
The bond price at any position is given by the expected values in an upstate
pBi+1,j+1 plus the expected value in a down state (1−p) Bi+1,j , discounted
to the present using the corresponding interest rate er ij ∆t. Using this
recursive procedure, the bond price at time 0 is 846,869 (Fig. 6.2).
Since the bond price at maturity is 1000, the price one period before
the maturity date is 804,580. It is calculated using the relation above for
1000
804,580
727,350 1000
712,44 887,930
734,740 850,100 1000
781,824 856,54 948,590
846,869 891,855 944,170 1000
947,400 970,45 991,550
1018,02 1013,19 1000
1055,8 1021,43
1062,34 1000
1041,97
1000
Year
0 1 2 3 4 5 6
(n+1)
Pi+1 (.) : up
(n)
Pi (.)
(n+1)
Pi (.) : down
This equation gives the bond price at a given node where the probability
is given by:
π = (r − d)/(u − d)
with:
r: return for one period, u return in an upstate, d return in a down state.
Ho and Lee (1986) showed that:
1 δT
h(T ) = for T ≥ 0 and h∗ (T ) =
π + (1 − π)δ T π + (1 − π)δ T
where the term δ corresponds to the spread between the two perturbation
functions.
The dynamics of interest rates is completely specified by π and δ.
Table 6.1. Discount function and bond prices in the Ho and Lee model.
P00 (3) = 0.9474, P00 (4) = 0.9296, and P00 (5) = 0.9119. Results are
reproduced in Table 6.1.
P00 (0) = 1, P00 (1) = 0.982, P00 (2) = 0.961, P00 (3) = 0.941, P00 (4) = 0.921,
and P00 (5) = 0.911. Results are reproduced in Table 6.2.
Table 6.2. Discount function and bond prices in the Ho and Lee model.
1
0.9958
0.9914
0.9905
0.9860
0.9878
0.9814
0.9708
0.9664
0.9611
function is: P00 (0) = 1, P00 (1) = 0.992, P00 (2) = 0.975, P00 (3) = 0.957,
P00 (4) = 0.939, and P00 (5) = 0.921, where P00 (1) is the price of a default-
free bond paying 1 in 3 months. Figure 6.4 shows the dynamics of the
Treasury bond.
At maturity, the option price is calculated using the following condition:
Using the recursive procedure, the put option price in period 3, position 1
is 4,477, or (0,5(0) + 0,5(9,173))0.976 = 4,477.
At period 2, the put price at the pair (2,0) is:
0 0
0
0
0
1.085
2.197
3.196 4.477 0
5.358
8.762
9.173
13.556
18.882
Fig. 6.5. Dynamics of the option price for the following parameters n = 4, δ = 0.99,
π = 0.5, and K = 980.
P(1) = 1.091135
P(1) = 1.04214 P(2) = 1.16448
P(1) = 0.97457 P(2) = 1.08026
P(2) = 0.96486
P(3) = 1.09826
P(3) = 0.9527 P(1) = 0.98202
P(1) = 0.90909 P(4) = 0.92531
P(2) = 0.94322
P(2) = 0.84168
P(1) = 0.93792
P(3) = 0.79383 P(2) = 0.87501
P(3) = 0.80063
P(4 )= 0.74880 P(1) = 0.88381
P(1) = 0.87712
P(5) = 0.69655 P(2) = 0.78154 P(2) = 0.76401
P(3) = 0.6945
P(4) = 0.60709 P(1) = 0.84413
P(2) = 0.70875
P(3) = 0.58366 P(1) = 0.79543
P(2) = 0.61885
Fig. 6.6. Possibility of negative interest rates in the Ho and Lee model.
Hence, in the first year, there are two possible rates. We specify the
following relationship between rising and falling rates as follows:
r 2,uu
r 1,u
r0 r 2,ud
r 1,d
r2,dd
In the second year, there are three possible values for the one-year forward
rates:
r2,uu , r2,ud , r2,dd
where:
r2,uu : one-year forward rate in the second year if the interest rate rises in
the first and the second year;
r2,ud : one-year forward rate in the second year if the interest rate rises in
the first year and falls in the second year or vice versa and
r2,dd : one-year forward rate in the second year if the interest rate falls in
the first and the second year.
The same relationship between rising and falling interest rates is
maintained. Hence, we have:
Let us denote simply by rt the one-year forward rate t years from now
if the rates decline.
The volatility of the one-year forward rate is equal to r0 σ. It is possible
to see this result by noting that e2σ is nearly equal to (1 + 2σ). In this case,
the volatility of the one-period forward rate can be written as:
V = 100
C = 4.5
r2, uu = ?
V = 99.748576
C = 5.4
r1, u = 4.7634%
V = 99.567 V = 100
C=0 C = 4.5
r0 = 4.5% r2, ud = ?
V = 100.5774
C = 4.5
r1, d = 3.9%
V = 100
C = 4.5
r2, dd = ?
Fig. 6.9. Computing the one-year forward rates for year 1 using a two-year 4.5% on-
the-run issue: first trial.
• Step 1: A value of r1 (the one-year forward rate, one year from now)
is set arbitrarily to 3.9%.
• Step 2: Since the one-year forward rate, if rates, rise corresponds to
r1 e2σ , then r1,u = r1,d e2σ = 4.7634% = 3.9%e(2×0.1) .
• Step 3: The bond’s value is computed for one year from now.
The two-year bond’s value is given by its maturity value (100) plus the
coupon payment at the same date (4.5), or 104.5.
When interest rates rise, the present value is Vu = 99.748576 =
104.5/1.047634.
When interest rates fall, the present value is Vd = 100.5774 =
104.5/1.039.
At time 0, the present value resulting from a rise in the interest rate is
computed as:
The present value resulting from a fall in the interest rate is computed as:
Since the observed market value of the bond at time 0 is 100, the
computed value of 100.15596 is not the correct one. This means that
the one-period forward interest rate used (r1 ) is not consistent with the
volatility assumption of 10% and the process used to generate the one-year
forward rate.
Hence, the 3.9% rate is low and a higher rate must be used in the same
procedure.
If we use an interest rate of 4% for r1 , this leads to a bond price equal
to 100 at time 0. Figure 6.10 is based on a second trial in the computation
of the interest rate.
• Step 1: A value of r1 (the one-year forward rate one year from now) is
set arbitrarily to 4%.
• Step 2: Since the one-year forward rate, if rates rise corresponds to
r1 e2σ , then r1,u = r1,d e2σ = 4.8856% = 4%e(2×0.1)
• Step 3: The bond’s value is computed one year from now.
V = 100
C = 4.5
r2, uu = ?
V = 99.6323
C = 4.5
r1, u = 4.8856%
V = 99.567 V = 100
C=0 C = 4.5
r0 = 4.5% r2, ud = ?
V = 100.46
C = 4.5
r1, d = 4%
V = 100
C = 4.5
r2, dd = ?
Fig. 6.10. Computing the one-year forward rates for year 1 using a two-year 4.5%
on-the-run issue: second trial.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
The two-year bond’s value is given by its maturity value (100) plus the
coupon payment at the same date (4.5), or 104.5.
When interest rates rise, the present value is Vu = 99.6323 =
104.5/1.048856.
When interest rates fall, the present value is Vd = 100.48076 =
104.5/1.04.
At time 0, the present value resulting from a rise in the interest rate
is computed as:
The present value resulting from a fall in the interest rate is computed as:
Since the observed market value of the bond at time 0 is 100, the
computed value of 100 is the correct one. This means that the one-year
forward interest rate used (r1 ) is consistent with the volatility assumption
of 10% and the process used to generate the one-year forward rate. Hence,
the current one-year forward rate is 4.5% and the forward rate one year
from now is 4%. It is possible to extend the analysis to three periods using
a 5% three-year bond. The same method can be used in the computation of
one-year forward rate, two years from now. This analysis allows to obtain
the value of r2 that leads to a bond value of 100. We let this as an exercise
for the interested reader.
The following example calibrates the BDT model to the current term
structure of zero-coupon rates and volatilities.
1 8 20
2 8.5 18
3 9 16
4 9.5 14
5 10 12
100
Su 4
Su 3 100
Su 3d
Su 2
Su2d 100
Su
Su 2d 2
S Sud
100
Sd Su d 2
Sd 2 Sd 3u
100
Sd 3
Sd 4
100
92.59259
100
This second step allows to build a two-period price tree. The second-year
bond prices at year one can be computed using the short rates at step one:
100
Su
84.945528 100
Sd
100
The standard relationships in the binomial models apply also in the BDT
model.
In fact, remember that in the standard binomial analysis, we have:
√ √
u = eσ T /N
, d = 1/u, u/d = e2σ T /N ,
√ √
ln(u/d) = 2σ T /N and σ = [1/2 T /N ] ln(u/d).
or
ru4
ru 3
ru 2 ru 3d
2
ru ru d
r rud ru 2d 2
rd rud 2
rd2 rd 3u
3
rd
rd 4
We can use the previous equations to determine the two values of the
interest rate (Fig. 6.12).
Since ru = rd e0.18(2) , this gives the following quadratic equation:
Solving this equation gives the following two rates: rd = 7.4%, ru = 10.60%.
Since these two rates are known, it is possible to use these rates to
compute the corresponding bond prices, i.e., 93.11 = (100/1.074) and
90.41 = (100/1.1060). The tree becomes:
100
90.41
84.11 100
93.11
100
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
10.60%
8% rud = rdu
7.4%
rdd
In order to find the rates at the end of period 2, we use the fact that
rates are log-normally distributed and that volatility is time dependent.
In this case, we have:
If the bond’s maturity is in two years, its yield must satisfy the following:
Su = 100/(1 + yu )2 , Sd = 100/(1 + yd )2
or
√ √
yu = (100/Su) − 1, yd = (100/Sd) − 1
We have: ln(yu /yd ) = 0.16 or ln(yu /yd ) = 0.32, which gives: yu /yd = e0.32 .
Using these last two equations, it is possible to obtain:
√
yu = (yu /yd )yd or yu = e0.32 ( (100/Sd) − 1).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
0.1(0.05 − r)∆t.
√
The standard deviation is 0.01 ∆t.
The interest-rate process is:
The dynamics of the interest rate can be modeled using a grid of equally
spaced rates. The probabilities corresponding to up, p1 , down, p2 , and “stay
the same”, p3 can be computed in a way to preserve the correct mean and
standard deviation at each node. The sum of these probabilities must be
equal to 1. Figure 6.13 describes a two-step tree where each interval is one
year. The spacing between different rates is 1.5%.
The initial value of r is 0.05. Since 0.1(0.05 − r) = 0 for the initial value
of r, the expected increase in the interest rate during the first period is
zero. The standard deviation is 0.01. The probabilities must verify:
p1 + p2 + p3 = 1.
The probabilities satisfy the following equation for the expected values:
or
r=
C r=
0.05
B r = 0.05
r=
A
r=
The probabilities also verify the following equation for the variance:
p1 + p2 + p3 = 1
0.065p1 + 0.05p2 + 0.035p3 = 0.05
0.065 p1 + 0.052 p2 + 0.0352p3 − 0.052 = 0.012
2
Hence:
For the second period, at the node A, the expected increase in the
interest rate is: 0.1(0.05 − 0.035) = 0.0015, so that the expected increase
in the interest rate at the end of the year is: 0.035 + 0.0015 = 0.0365. The
standard deviation is 0.01.
Using a system of three equations allows the computation of the
different probabilities. Hence, we have:
p1 + p2 + p3 = 1
(0.035 + 0.015)p1 + (0.035 + 0)p2 + (0.035 − 0.015)p3 = 0.0365
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
or
and
p1 + p2 + p3 = 1
0.05p1 + 0.035p2 + 0.02p3 = 0.0365
0.05 p1 + 0.0352p2 + 0.022p3 − 0.03652 = 0.012
2
is
For the second period, at the node B, the expected increase in the
interest rate is: 0.1(0.05 − 0.05) = 0.00, so that the expected increase in
interest rates at the end of the year is: 0.05 + 0.00 = 0.05. The standard
deviation is 0.01.
Using a system of three equations allows the computation of different
probabilities. Hence, we have:
p1 + p2 + p3 = 1
(0.05 + 0.015)p1 + (0.05 + 0)p2 + (0.05 − 0.015)p3 = 0.05
or
and
For the second period, at the node C, the expected increase in the
interest rate is: 0.1(0.05 − 0.065) = −0.0015, so that the expected increase
at the end of the year is: 0.065 − 0.0015 = 0.0635. The standard deviation
is 0.01.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
p 1 + p2 + p3 = 1
(0.065 + 0.015)p1 + (0.065 + 0)p2 + (0.065 − 0.015)p3 = 0.0635
or
and
Table 6.4. Pricing a call using the binomial and trinomial models.
Call price S = 18 S = 19 S = 20 S = 21
The maturity date is 15/06/2004. The following strike prices are used: 22,
23, 24, 25, 26, 27, 28, 29, and 30. The following dates and amounts of
dividends are available:
For 1999, 0.25; for 2000, 0.25; for 2001, 0.25; for 2002, 0.3; for 2003, 0.35
and for 2004, 0.35.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
Table 6.5. Pricing a call using the binomial and trinomial models.
S = 22 S = 25 S = 28 S = 30
Summary
Rendleman and Bartter (for details, refer to Bellalah et al., 1998) developed
a similar model for the pricing of interest-rate sensitive instruments. Ho and
Lee (1986) extended the binomial model for the valuation of interest-rate
options and bond options. This model presents some deficiencies. In fact,
the constraints imposed on movements of the entire discount function are
not sufficient to eliminate negative interest rates. This deficiency has been
recognized by Heath, et al. (1987), Pedersen et al. (for details, refer to
Bellalah et al., 1998), and Ritchken and Boenawen (1990) among others.
These authors proposed other specific models.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
Table 6.6. Pricing a call using the binomial and trinomial models.
S = 18 S = 19 S = 20 S = 21
Table 6.7. Pricing a call using the binomial and trinomial models.
S = 22 S = 25 S = 28 S = 30
Questions
1. Describe the Rendleman and Bartter model (1979) for interest-rate
sensitive instruments.
2. Describe the Ho and Lee model for interest rates and bond options.
3. Describe the binomial interest-rate trees and the log-normal random
walk.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
Appendix A
Ho and Lee model and binomial dynamics of bond prices
When the term structure is described by a binomial model, the same
dynamics applies for zero-coupon bonds P (N ) with a maturity date N
at initial time. When it remains (N − 1) periods, the discounting functions
(1)
in upstates and down states allow the computation of P1 (N − 1) and
P01 (N − 1).
This model is similar to the binomial model of CRR (1979) and
Rendleman and Bartter. Ho and Lee introduced two perturbation functions,
(n)
h(T ) and h ∗ (T ). The discount function at period n and state i is Pi (.T ).
The discounting function that prevents risk-less profitable arbitrage is
known as the implied forward discount function that is specified by
(n)
Fi (T ), or:
(n)
(n) (n+1) (n+1) [Pi (T + 1)]
Fi (T ) = Pi (T ) = Pi+1 (T ) = (n)
for T = 0, 1, . . .
[Pi (1)]
This equation gives the bond price at a given node where the probability
is given by:
π = (r − d)/(u − d)
(n)
(n+2) Pi (T + 2) h(T + 1)h∗ (T )
Pi+1 (T ) = (n)
Pi (2) h(1)
(n)
(n+2) Pi (T + 2) h∗ (T + 1)h(T )
Pi+1 (T ) = .
(n)
Pi (2) h∗ (1)
Elimination of h∗ gives:
1 δ
= +Γ
h(T + 1) h(T )
1 π(h(1) − 1)
h(1) = ; Γ= .
π + (1 − π)δ (1 − π)h(1)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
or
(n) P (T + n)h(T + n − 1)h(T + n − 2) · · · h(T )δ T (n−1)
Pi (T ) = . (A.3)
P (n)h(n − 1)h(n − 2) · · · h(1)
Equation (A.3) gives the discount function that can be applied at each time
step.
When T = 1, the bond price is:
(n) P (n)
ri (1) = ln + ln(πδ −n + (1 − π)) + i ln δ.
P (n + 1)
(n)
In the presence of a probability q, for each moment n, ri (1) follows a
binomial distribution with a mean µ and a variance, σ with:
or:
and:
Exercises
Example 1. Consider the valuation of European and American options in
the following context:
Underlying asset, S = 100, strike price K = 80, interest rate = 0.05,
volatility = 0.2, T = 5 months, and N = 5. In this case, we have: p =
0.5217, d = 0.9439, and u = 1.0594.
106.7726
100.7827
95.1288 95.1288
89.7921 89.7921
84.7547 84.7547 84.7547
80 80 80
75.5120 75.5120 75.5120
71.2758 71.2758
67.2772 67.2772
63.5030
59.9404
0
2.3202
5.8694 4.8712
9.9155 9.7921
14.0858 14.4154 15.2453
18.1946 18.7578 19.5842
22.8351 23.6581 24.4880
27.4820 28.3084
31.8929 32.7228
36.0812
40.0596
0
2.3202
6.0675 4.8712
10.4136 10.2079
15.2453 15.2453 15.2453
20 20 20
24.4880 24.4880 24.4880
28.7242 28.7242
32.7228 32.7228
36.4970
40.0596
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
6.7726
3.5187
1.8281 0
0.9498 0
0.4934 0 0
0.2564 0 0
0 0 0
0 0
0 0
0
0
6.7726
3.5187
1.8281 0
0.9498 0
0.4934 0 0
0.2564 0 0
0 0 0
0 0
0 0
0
0
References
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–654.
Black, F, E Derman and W Toy (1990). A one factor model of interest rates
and its application to treasury bond options. Financial Analysts Journal, 46
(January–February), 33–39.
Boyle, P (1986). Option valuation using a three jump process. International
Options Journal, 3, 7–12.
Boyle, PP (1988). A lattice framework for option pricing with two state variables.
Journal of Financial and Quantitative Analysis, 23 (March) 1–12.
Briys, E, M Bellalah, F de Varenne and H Mai (1998). Options, Futures and Other
Exotics. Chichester, UK: John Wiley and Sons.
Cox, J, S Ross and M Rubinstein (1979). Option pricing: a simplified approach.
Journal of Financial Economics, 7, 229–263.
Heath, D, R Jarrow and A Morton (1987). Bond pricing and the term structure
of interest rate: a new methodology for contingent claims valuation. Working
paper, Ithaca, NY: Cornell University (Revised edition, 1989).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch06
Ho, T and S Lee (1986). Term structure movements and pricing interest rate
contingent claims. Journal of Finance, 41, 1011–1029.
Hull, J (2000). Options, Futures, and Other Derivative Securities. NJ, USA:
Prentice Hall International Editions.
Hull, J and A White (1988). An analysis of the bias in option pricing caused by
a stochastic volatility. Advances in Futures and Options Research, 3, 29–61.
Hull, J and A White (1993). Efficient procedures for valuing European and
American path dependent options. Journal of Derivatives, 1 (Fall 1993),
21–31.
Jarrow, RA and A Rudd (1983). Option Pricing. Homewood, IL: Irwin.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
Omberg, E (1988). Efficient discrete time jump process models in option pricing.
Journal of Financial and Quantitative Analysis, 23(2), 161–174.
Rendleman, RJ and BJ Bartter (1980). The pricing of options on debts securities.
Journal of Financial and Quantitative Analysis, 15(March), 11–24.
Ritchken, P and K Boenawen (1990). On arbitrage free pricing of interest rate
contingent claims. Journal of Finance, 55(1), 259–264.
Rubinstein, M (1994). Implied binomial trees. Journal of Finance, 49(3), 771–818.
Sandmann, K (1993). The pricing of options with an uncertain interest rate: a
discrete time approach. Mathematical Finance, 3(April), 201–216.
Trigeorgis, L (1993). The nature of option interactions and the valuation of
investments with multiple real options. Journal of Financial and Quantitative
Analysis, 28, 1–20.
Whaley, RE (1986). Valuation of American futures options: theory and empirical
tests. Journal of Finance, 41(March), 127–150.
Yisong, T (1993). A modified lattice approach to option pricing. Journal of
Futures Markets, 13, 563–577.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
Chapter 7
Chapter Outline
This chapter is organized as follows:
1. Section 7.1 presents the lattice approach and the binomial model for the
valuation of equity and futures options.
2. Section 7.2 presents a simple extension of the lattice approach to account
for the effects of information costs.
3. Section 7.3 develops some important results regarding the binomial
model and the risk neutrality.
4. Section 7.4 presents the Hull and White’s interest-rate trinomial model
for the valuation of interest-rate derivatives.
5. Section 7.5 develops a simple context for the pricing of path-dependent
interest-rate contingent claims using a lattice.
Introduction
This chapter deals with the valuation of derivative assets using the
binomial or the lattice approach. The lattice approach was initiated by
Cox et al. (CRR) (1979). CRR approach considers the situation where
there is only a single underlying asset: the price of a non-dividend paying
stock. The time to maturity of the option is divided into several equal
327
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
small intervals during which the underlying asset price moves from its
initial value to one of the two new values up or down. In a risk-neutral
world, it is possible to obtain the values corresponding to an upward or
downward movements and the corresponding probabilities. The valuation
of options in a binomial framework starts at the maturity date since
at this date, the option payoff is known. Then, we proceed backward
through the binomial tree from the maturity to the initial time. In a risk-
neutral world, the option value at each time can be calculated as the
expected value at the maturity date discounted at the risk-less rate of
interest.
The lattice approach can be easily extended to account for the effects
of a continuous dividend yield. If a security pays a dividend yield, then the
expected return on the underlying asset is given by the difference between
the risk-less rate and the continuous dividend yield. The extension of the
lattice approach in the presence of discrete dividends to the valuation of
options on stocks paying a known dividend can be easily implemented. In
the presence of a discrete dividend, the pricing problem can be simplified
as in Hull (2000) by assuming that the implicit spot stock price has two
components: a part which is stochastic and a part which is the present value
of all future cash payments during the option’s life.
The lattice approach has also been used by several authors to model
the dynamics of the term structure of interest rates and to value bonds
and bond options. There have been many attempts and approaches to
describe yield-curve movements using a one-factor model. The approach
presented by Ho and Lee (1986), in the form of a binomial tree for discount
bonds, provides an exact fit to the current-term structure of interest
rates. Their model is interesting since it takes the market data such as
the current-term structure of interest rates as given. In this respect, it
is close to a binomial stock option pricing approach where the current
stock price is taken as an input to the model. Unlike most interest-
rate contingent claims models, this model uses full information on the
current-term structure. In fact, using an ingenious discrete-time approach
for pricing bonds and interest-rate contingent claims, Ho and Lee (1986)
succeeded in incorporating all information about the yield curve in their
model.
Hull and White (1993) presented a general numerical procedure
involving the use of trinomial trees for constructing one-factor models where
the short rate is Markovian and the models are consistent with initial
market data. Their procedure is efficient and provides a convenient way
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
or
pS 2 u2 + (1 − p)S 2 d2 − (Ser∆t )2 .
√ √
This last expression can be written as S 2 (er∆t (eσ ∆t + e−σ ∆t) −
1 − e2r∆t ).
2 3
Now, using the expansion of ex in series form as ex = 1+x+ x2 + x6 +· · ·
It is clear that when terms of order ∆t2 and higher are ignored, the variance
of the stock price is S 2 σ 2 ∆t. This shows that we have the appropriate
values for u, d, and p. The nature of the lattice of stock prices is completely
specified and the nodes correspond to:
In the same context, at time t + i∆t, the American call option value is:
At each node, at time t + i∆t, the American put value is given by,
dates between t and (t + i∆t) is accounted for using the following values of
the stock at different nodes: S(1 − δi )uj di−j for j = 0, 1, . . . , i.
7.1.4. Examples
In these examples, we use the following parameters for the valuation of a
European and an American put option on a stock paying a dividend of
2.05 in three months and a half: S ∗ = 40, S = 42, K = 45, r = 0.1,
N = 5, T = 5 months, ∆t = 1 month, and σ = 0.4. The √ first step is√the
calculation of the parameters u, d, a, and p using u = eσ ∆t , d = e−σ ∆t ,
a−d
p = u−d , q = 1 − p, a = er∆t . This gives u = 1.1224, d = 0.8909, a = 1.0084,
p = 0.5073, and q = 0.4927.
Using these parameters, it is possible to generate the dynamics of the
underlying asset. The values of the underlying asset at different nodes are:
S0,0 = 42;
S1,1 = 46.9136, S1,0 = 37.6556;
S2,2 = 52.4256, S2,1 = 42.0344, S2,0 = 33.7859;
S3,3 = 58.6107, S3,2 = 46.9475, S3,1 = 37.6893, S3,0 = 30.3403;
S4,4 = 63.4822, S4,3 = 50.3914, S4,2 = 40, S4,1 = 31.7515,
S4,0 = 25.2039 and
S5,5 = 71.2525, S5,4 = 56.5593, S5,3 = 44.8960, S5,2 = 35.6379,
S5,1 = 28.2289, and S5,0 = 22.4554.
The indices (i, j) correspond respectively to the period and the position.
The lowest position on a tree is indexed by zero. For example, the values
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
(τ −1)
S0,0 = S ∗ (u0 d0 ) + De−r 12 , S1,1 = S ∗ (u1 d0 ) + De−r
τ
12
(τ −1) (τ −2)
S1,0 = S ∗ (u0 d1 ) + De−r 12 , S2,2 = S ∗ (u2 d0 ) + De−r 12
(τ −2) (τ −2)
S2,1 = S ∗ (u1 d1 ) + De−r 12 , S2,0 = S ∗ (u0 d2 ) + De−r 12
(τ −3) (τ −3)
S3,3 = S ∗ (u3 d0 ) + De−r 12 , S3,0 = S ∗ (u0 d3 ) + De−r 12
When y > τ , we do not discount the dividends and the above values are
generated as follows: S4,4 = S ∗ (u4 d0 ), S4,3 = S ∗ (u3 d1 ), S4,0 = S ∗ (u0 d4 ),
and S5,0 = S ∗ (u0 d5 ).
For example, the values P4,4 and P4,3 are calculated as follows:
The price of the American put is 6.0633 when there are dividends. The
difference between the American price and the European price corresponds
to the early exercise premium. The lattice can be used to estimate the hedge
ratio ∆ from the nodes at time t + ∆t as:
F1,1 − F1,0
∆= .
Su − Sd
F2,2 − F2,0
∆=
Su2 − Sd2
R̄S − r = βS [R̄m − r] + λS − βS λm
where:
• t: current time;
• T : maturity date of the contract in years;
• (T − t): time remaining until the maturity of the contract in years;
• S: spot price of the asset;
• K: delivery price for a forward contract;
• f : value of a long position in a forward contract;
• F : forward price at time t;
• r: risk-free rate at time t for maturity T ;
• rf : risk-free rate at time t for maturity T in a foreign country and
• λS : information cost for the asset S.
The equivalence between these values of the two portfolios at each time
implies that
Re-call that when a contract is initiated, its forward price must be equal
to the delivery price, which is chosen in such a way that the contract’s value
is zero. The forward price F corresponds to the value of K for which, f = 0.
Using the last equation, the forward price F = Se(r+λS )(T −t) .
f + Ke−(r+λS )(T −t) = Se−q(T −t) or f = Se−q(T −t) − Ke−(r+λS )(T −t) .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
or
this market. This cost represents its minimal cost before acting in a given
market. It is in some way, its minimal return required before implementing
a given strategy.
The expected return on the stock can be written as:
The variance on the stock can be written for the same interval ∆t as:
√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, p= , a = e(r+λs )∆t .
u−d
At each node, at time t + i∆t, the American call option value is given by:
This terminal value gives the option values for the (N + 1) terminal
nodes. Then at each node, at time t + i∆t, the American put value is
given by:
In the same context, at time t + i∆t, the American call option value is
given by:
At each node, at time t + i∆t, the American put value is given by:
and
with (Di + Ri ) ≥ 0. When there are no dividends, the American put may
be exercised because interest can be earned on the exercisable proceeds of
the option. The put is exercised at time ti if:
a−d √ √
p= , u = eσ ∆t
, d = e−σ ∆t
, and a = e(r+λS )∆t .
u−d
The term λS appears because of the duplication portfolio. The nature
of the lattice of stock prices is completely specified and the nodes
correspond to:
When there is just one ex-cash income date τ , and k∆t ≤ τ ≤ (k+1)∆t,
then at time y, the value of the stochastic component S is given by:
where,
rh − d u − rh
pu = pd =
u − d u − d
C u = max[0, uS − K]
C d = max[0, dS − K]
T
and r = 1+ the risk-less rate over a single period and h = N periods.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
u = exp(σ(T /N )0.5 )
d = exp(−σ(T /N )0.5 )
q = (1/2)[1 + (µ/σ)(T /N )0.5 ).
Jarrow and Rudd (1983) showed that the first three parameters
for the stock’s return are consistent with the log-normal process. The
three moments are the mean µ(T /N ), the variance σ2 (T /N ), and the
skewness, which equal zero. However, the three moments for the log-normal
process using the CRR parameters are inconsistent with the corresponding
moments of the log-normal process. In fact, using the CRR parameters, the
moments are: the mean µ(T /N ), the variance σ2 (T /N ) − µ2 (T /N )2 , and
the skewness which equals 2µ[µ2 (T /N )3 − σ 2 (T /N )2 ]. This simple analysis
shows that the CRR parameters imply a level of skewness, which is different
from zero. The variance and the skewness of the binomial model of CRR
converge to the variance and the skewness of the log-normal process only
in the continuous time limit. This is the case since (T /N )2 and (T /N )3
become insignificant in comparison to (T /N ) as N tends to infinity. The
Jarrow and Rudd (1983) parameters are inconsistent with the risk-neutral
approach in discrete time. In fact, if one replaces the values of u and d from
Eqs. (7.4) and (7.5) in Eq. (7.3), we see that the probabilities and option
values depend on the preference parameter µ.
Consider the general form of the binomial model parameters:
with
We must have for the Jarrow and Rudd (1983) choice that:
u d
Ct = [pu Ct+h + pd Ct+h ]/rh (7.7)
where the different parameters are given in Eq. (7.6). We can write that:
u d
Ct = C(St , T − t), Ct+h = C(uSt , T − (t + h)), Ct+h = C(dSt , T − (t + h)).
By appropriate substitutions, we can write (7.7) as:
rh − exp[mh − σ(h0.5 )]
exp[mh + σ(h0.5 )] − exp[mh − σ(h0.5 )]
· C(exp[mh + σ(h0.5 ] · St , T − (t + h))
exp[mh + σ(h0.5 )] − rh
+
exp[mh + σ(h0.5 )] − exp[mh − σ(h0.5 )]
· C(exp[mh + σ(h0.5 ] · St , T − (t + h))
− rh C(St , T − t) = 0
1 2 2 ∂ 2 Ct ∂Ct ∂Ct
σ St 2 h + (logr)S ·h+ · h − (logr)Ct h + Z = 0
2 ∂ St ∂St ∂St
where Z contains all terms of higher order of h. Dividing this last equation
by h gives:
1 2 2 ∂ 2 Ct ∂Ct ∂Ct
σ St 2 h + (logr)S + − (logr)Ct + Z/h = 0.
2 ∂ St ∂St ∂t
The solution to this equation depends on the revealed preference
parameter contained in the term Z/h. However, when the number of sub-
intervals tend to infinity, the term Z/h goes to zero. This last equation
converges then to the Black and Scholes (1973) partial differential equation.
This equation is independent of m.
In this model, the volatility σ is a known constant and the functional form
of µ is known. The value of θt is unknown.
The short interest rate r corresponds to the continuously compounded
yield on a discount bond maturing at date ∆t. When the tree is constructed,
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
the values of r are equally spaced and have the following form: Tθ + j∆t.
Note that r0 is the current interest-rate value and j is an integer, which
may be positive or negative. Also, the values of ∆t are equally spaced with
a positive integer i. As shown in Hull and White (1990), the variables ∆t
and
√ ∆r must be √ chosen in such a way that ∆r lies in an interval between
σ
2 3∆t and 2σ ∆t.
We use the following notation:
(i, j): a node on the tree for the values of t = i∆t, and rj = r0 + j∆r
with i ≥ 2;
R(i): yield at time zero on a discount bond maturing at time i∆t;
µ(i, j): drift rate of r at node (i, j) with rj = r0 + j∆r and
(i,j)
Pk : for k = 1, 2, 3, the probabilities corresponding respectively to the
upper, middle, and lower branches emanating from node (i, j).
If the tree constructed up to time n∆t, (n ≥ 0) is consistent with the
observed R(i) and the interest rate r at time i∆t applies to the interval
time between i∆t and (i + 1)∆t, then the tree reflects the values of R(i)
for i ≤ (n + 1). Note that between times n∆t and (n + 1)∆t, the value
of θ(n∆t) must be chosen in such a way that the tree is consistent with
R(n + 2).
Once the value of θ(n∆t) is known, then it is possible to calculate the
drift rates µn,j for r at time n∆t using:
The three nodes from node (n, j) are: (n + 1, k + 1), (n + 1, k), and
(n + 1, k − 1), where k must be chosen in such a way that rk , (the value of
the interest rate reached by the middle branch) is very close to the expected
value of the interest rate, rj + µn,j ∆t.
Hull and White (1993) gave the following probabilities:
σ 2 ∆t η2 η
P1 (n, j) = 2
+ 2
+
2∆r 2∆r 2∆r
σ 2 ∆t η2
P2 (n, j) = 1 − 2
+
∆r ∆r2
σ 2 ∆t η2 η
P3 (n, j) = 2
+ −
2∆r 2∆r2 2∆r2
with
η = µn,j ∆t + (j − k)∆r.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
This general procedure for fitting a one-factor model of the short rate
to the initial yield curve using a trinomial interest-rate tree can be used
to test the effect of a wide range of assumptions about the interest-rate
process on the prices of interest-rate derivatives. However, much research
remains to be done to obtain better models for interest rates. Anyway, this
model also allows for negative interest rates.
pdd
Ψ(1, 0) (1+r d)
pu pud pd pdu
Ψ(2, 1) = × + ×
(1 + r) (1 + ru ) (1 + r) (1 + rd )
pud pdu
= Ψ(1, 1) × u
+ Ψ(1, 0) ×
(1 + r ) (1 + rd )
pu puu
Ψ(2, 0) = ×
(1 + r) (1 + ru )
puu
= Ψ(1, 1) . (7.12)
(1 + ru )
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
where l can be (j − 1) or j.
At each period, the security pays a cash flow C(i, j, k) calculated as a
percentage, c(i − 1, l) of the remaining principal in the previous period. The
remaining principal is fully paid at maturity, implying that a total payout
is equal to [1 + c(n − 1), l]F (n − 1, l, k) and P (n, j, k) = 0. The next step
is to store the principal at each node in terms of its Arrow-Debreu price at
each lattice node as follows:
F ∗ (0, 0) = F (0, 0)
1 − p(i − 1, 0)
F ∗ (i, 0) = F ∗ (i − 1, 0)[1 − a(i, 0)]
1 + r(i − 1, 0)
p(i − 1, i − 1)
F ∗ (i, i) = F ∗ (i − 1, i − 1)[1 − a(i, i)]
1 + r(i − 1, i − 1)
1 − p(i − 1, j)
F ∗ (i, j) = F ∗ (i − 1, j)[1 − a(i, j)]
1 + r(i − 1, j)
p(i − 1, j − 1)
+ F ∗ (i − 1, j − 1)[1 − a(i, j)]
1 + r(i − 1, j − 1)
for 0 < j < i. (7.14)
1 − p(i − 1, 0)
C ∗ (i, 0) = F ∗ (i − 1, 0)c(i − 1, 0)
1 + r(i − 1, 0)
p(i − 1, i − 1)
C ∗ (i, i) = F ∗ (i − 1, i − 1)c(i − 1, i − 1)
1 + r(i − 1, i − 1)
1 − p(i − 1, j)
C ∗ (i, j) = F ∗ (i − 1, j)c(i − 1, j)
1 + r(i − 1, j)
p(i − 1, j − 1)
+ F ∗ (i − 1, j − 1)c(i − 1, j − 1)
1 + r(i − 1, j − 1)
for 0 < j < i. (7.15)
1 − p(i − 1, 0)
P ∗ (i, 0) = F ∗ (i − 1, 0)a(i, 0)
1 + r(i − 1, 0)
p(i − 1, i − 1)
P ∗ (i, i) = F ∗ (i − 1, i − 1)a(i, i)
1 + r(i − 1, i − 1)
1 − p(i − 1, j)
P ∗ (i, j) = F ∗ (i − 1, j)a(i, j)
1 + r(i − 1, j)
p(i − 1, j − 1)
+ F ∗ (i − 1, j − 1)a(i, j)
1 + r(i − 1, j − 1)
for 0 < j < i. (7.16)
n−1
i n
V (0, 0) = [C ∗ (i, j) + P ∗ (i, j)] + [C ∗ (n, j) + F ∗ (n, j)] (7.17)
i=1 j=0 j=0
This method is exact and does not use any approximation. It is better
than the one presented in Hull and White (1993).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
v(n, j) = 1
v(i, j) = [v(i + 1, j)[1 − a(i + 1, j)] + c(i, j) + a(i + 1, j)]
1 − p(i, j)
× + [v(i + 1, j + 1)[1 − a(i + 1, j + 1)]
1 + r(i, j)
p(i, j)
+ c(i, j) + a(i + 1, j + 1)] × for i < n. (7.18)
1 + r(i, j)
This method is similar to that proposed in Hillard et al. (for more
details, refer to Bellalah et al., 1998). In the last period in the lattice, the
nodes are set to one. Each term in the brackets corresponds to the sum of
three values (the price scaled by the amortization rate, the cash flow and
the pre-paid principal) from the nodes in the next period. In general, for
all k values of the principal, we have:
When z = 3, denote the minimum by F0 (s, j), the average by F1 (s, j),
and the maximum by F2 (s, j).
At maturity, the value of the underlying security corresponding to the
three values is computed using Eq. (7.19).
Denote these payoffs by V0opt (s, j), V1opt (s, j) and V2opt (s, j).
The option payoffs corresponding to the values of the principal at the
previous node are computed using the quadratic interpolation method in
Hull and White (1993). The computation of V1opt (i, j) corresponding to
F1 (i − 1, j) needs the calculation of F1 (i, j) = F1 (i − 1, j)a(i, j).
Equation (7.22) that is used for the interpolation is given by:
Summary
The lattice approach for the pricing of options can be specified with respect
to stock options in the absence of payments to the underlying asset. The
option’s maturity date is divided into several reasonably small intervals of
time. During each time interval, the underlying asset price moves either
upward or downward. This movement in the stock price is binomial with
a probability p attached to an upward jump and a probability (1 − p) to
a downward movement. The parameters corresponding to the up, down,
and the probability are functions of the mean and variance of the rates
of return on S during the interval. The basic lattice approach suggested
by CRR considers the situation where there is only one state variable:
the price of a non-dividend paying stock. The time to maturity of the
option is divided into equal intervals. It is a simple matter to extend
this approach to account for the effects of a continuous dividend yield,
a discrete dividend, information costs, etc. The valuation of options in
a binomial model is easy to implement since we start at the maturity
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
date, where the payoff is known, and then proceed backward through the
tree. In a risk-neutral world, the option value at a given time can be
calculated as the expected value at the maturity date discounted at the
appropriate rate.
The valuation of European options is slightly different from that of
the American options. For the American options, the holder can exercise
his/her option at any instance before the maturity date. Therefore, at each
node, the option value must be at least equal to its intrinsic value. The
extension of the lattice approach in the presence of information costs and
a continuous dividend yield is simple.
This chapter shows that the consistency of the discrete-time binomial
option pricing model of Cox et al. (1979) with the risk-neutrality argument
depends heavily on the appropriate choice of its parameters. In fact, for a
certain choice of the parameters, the option prices can depend on investor
preferences. This preference dependence diminishes as the number of sub-
intervals become large. This dependence disappears only in the continuous
time limit when the binomial model converges to the Black and Scholes
(1973) model. Risk neutrality is obtained when the CRR model assumes
a very specific behavior regarding the price changes of the underlying
asset. Hence, in general, risk neutrality is not obtained in discrete time
in some cases.
Hull and White (1993) developed a numerical procedure that is based
on the use of trinomial trees for constructing one-factor models of interest
rates. The models are consistent with initial market data where the short
rate follows a Markovian process. The procedure adopted by Hull and White
(1993) is efficient and provides a convenient way of implementing several
other models in the literature.
This chapter presents the basic concepts and techniques underlying
the derivative assets pricing problem within the context of binomial models
and lattice approaches. The lattice approach is applied to the valuation
of European and American equity options when the underlying asset is
traded in a spot or in a futures market. The approach is extended to
the valuation of options by taking into account several cash distributions
to the underlying assets. It is convenient to note that lattice approaches
can be easily implemented and adapted to different derivative asset
payoffs. This approach is more pedagogical than the continuous time
approach. However, it takes some time to offer accurate option prices,
which is not a major handicap when there is no closed form or analytic
solutions.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
Questions
1. Describe the lattice approach and the binomial model for the valuation
of equity options.
2. Describe the binomial model for the valuation of futures options.
3. Describe the extension of the lattice approach to account for the effects
of information costs.
4. Describe the Hull and White interest-rate trinomial model for the
valuation of interest-rate derivatives.
5. Describe the model of Dharan for the pricing of path-dependent interest-
rate contingent claims using a lattice.
References
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June 1987), 301–320.
Bellalah, M (1999a). The valuation of futures and commodity options with
information costs. Journal of Futures Markets, (September), 19, 645–664.
Bellalah, M (1999b). Les biais des modeles d’options revisites. Revue Francaise
de Gestion, (Juin) 94–100.
Bellalah, M (2000a). A risk management approach to options trading on the Paris
Bourse. Derivatives Strategy, 5(6), 31–33.
Bellalah, M (2000b). Valuation of American CAC 40 index options and wildcard
options. International Review of Economics and Finance.
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13(3),
1895–1927.
Bellalah, M and B Jacquillat (1995). Option valuation with information costs:
theory and tests. The Financial Review, 30(3), 617–635.
Bellalah, M and J-L Prigent (2001). Pricing standard and exotic options in the
presence of a finite mixture of Gaussian distributions. International Journal
of Finance, 13(3), 1975–2000.
Bellalah, M and F Selmi (2001). On the quadratic criteria for hedging under
transaction costs. International Journal of Finance, 13(3), 2001–2020.
Bellalah, M, JL Prigent and C Villa (2001a). Skew without skewness: asymmetric
smiles; information costs and stochastic volatilities. International Journal of
Finance, 13(2), 1826–1837.
Bellalah, M, Ma Bellalah and R Portait (2001b). The cost of capital in
international finance. International Journal of Finance, 13(3), 1958–1973.
Black, F (1976). Studies of stock price volatility changes. In Proc. of the 1976
Meeting of the Business Economics Statistics Section, American Statistical
Association, pp. 177–181.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
Cox, JC and SA Ross (1976). The valuation of options for alternative stochastic
processes. Journal of Financial Economics, 3, 145–166.
Cox, J, S Ross and M Rubinstein (1979). Option pricing: a simplified approach.
Journal of Financial Economics, 7, 229–263.
Dharan, VG (1997). Pricing path-dependent interest rate contingent claims using
a lattice. Journal of Fixed Income, 6(March), 41–49.
Harrison, JM and D Kreps (1979). Martingales and arbitrage in multiperiod
securities markets. Journal of Economic Theory, 20, 381–408.
Ho, I and S Lee (1986). Term structure movements and pricing interest rate
contingent claims. Journal of Finance, 41, 1011–1029.
Hull, J (2000). Options, Futures, and Other Derivative Securities. 4th Ed.
Englewood Cliffs, NJ: Prentice Hall.
Hull, J and A White (1990). Pricing interest rate derivative securities. Review of
Financial Studies, 3, 573–592.
Hull, J and A White (1993). Efficient procedures for valuing European and
American path dependent options. Journal of Derivatives, 1, 21–31 (Fall).
Jarrow, RA and A Rudd (1983). Option Pricing, Homewood, IL: Irwin. Merton R
(1973). Theory of rational option pricing. Bell Journal of Economics and
Management Science, 4, 141–183.
Lintner, J (1965). Security prices, risk and maximal gains from diversification.
Journal of Finance, 20, 587–616.
Markowitz, HM (1952). Portfolio selection. Journal of Finance, 7(1), 77–91.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
Merton, RC (1987). A simple model of capital market equilibrium with incomplete
information. Journal of Finance, 42(3), 483–510.
Nawalkha, SK and DR Chambers (1995). The binomial model and the risk
neutrality: some important details. Financial Review, 30(3), (August),
605–615.
Scholes, M (1998). Derivatives in a dynamic environment. American Economic
Review, 88(2), 350–370.
Shapiro, A (2000). The investor recognition hypothesis in a dynamic general
equilibrium: theory and evidence. Working Paper, New York University:
Stern School of Business.
Sharpe, WF (1964). Capital asset prices: a theory of market equilibrium under
conditions of risk. Journal of Finance, 19, 425–442.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch07
Part III
365
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
366
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
Chapter 8
Chapter Outline
This chapter is organized as follows:
1. Section 8.1 gives an overview of the option pricing theory in the
pre-Black–Scholes period.
2. Section 8.2 presents the story and the main results in the breakthrough
work of Black–Scholes for the pricing of derivative assets when the
underlying asset is traded in a spot market. It proposes the story until
the publication of the original formula.
3. Section 8.3 develops the foundations of the Black–Scholes–Merton
Theory.
4. Section 8.4 presents two alternative derivations of the Black–Scholes
model. The formula is derived using equilibrium market conditions and
arbitrage theory.
5. Section 8.5 reviews the main results in Black’s (1976) model for the
pricing of derivative assets when the underlying asset is traded on a
forward or a futures market. Some applications of the model are also
given.
6. Section 8.6 applies the capital-asset pricing model (CAPM) to the
valuation of forward contracts, futures, and commodity options.
7. Section 8.7 studies the “holes” in the Black–Scholes–Merton theory.
9. Appendix 8.A provides an approximation of the cumulative normal
distribution function.
367
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
Introduction
Numerous researchers have worked on building a theory of rational option
pricing and a general theory of contingent claims valuation. The story
began in 1900, when the French mathematician, Louis Bachelier, obtained
an option pricing formula. His model is based on the assumption that stock
prices follow a Brownian motion. Since then, numerous studies on option
valuation have blossomed. The proposed formulas involve one or more arbi-
trary parameters. They were developed by Sprenkle (1961), Ayres (1963),
Boness (1964), Samuelson (1965), Thorp and Kassouf (1967), and Chen
(1970) among others. The Black and Scholes (1973) formulation, hereafter
called as B–S, solved a problem, which has occupied the economists for at
least three-quarters of a century. This formulation represented a significant
breakthrough in attacking the option pricing problem. In fact, the Black–
Scholes theory is attractive since it delivers a closed-form solution to the
pricing of European options. Assuming that the option is a function of a
single source of uncertainty, namely the underlying asset price, and using
a portfolio which combines options and the underlying asset, Black and
Scholes constructed a risk-less hedge, which allowed them to derive an
analytical formula. This model provides a no-arbitrage value for European
options on shares. It is a function of the share price S, the strike price K,
the time to maturity T , the risk-free interest rate r, and the volatility of the
stock price, σ. This model involves only observable variables to the excep-
tion of volatility and it has become the benchmark for traders and market
makers. It also contributed to the rapid growth of the options markets by
making a brand new pricing technology available to market players.
About the same time, the necessary conditions to be satisfied by
any rational option pricing theory were summarized in Merton’s (1973)
theorems. The post-Black–Scholes period has seen many theoretical devel-
opments. The contributions of many financial economists to the extensions
and generalizations of Black–Scholes type models has enriched our under-
standing of derivative assets and their seemingly endless applications.
The first specific option pricing model for the valuation of futures
options is introduced by Black (1976). Black (1976) derived the formula
for futures and forward contracts and options under the assumption that
investors create risk-less hedges between options and the futures or forward
contracts. The formula relies implicitly on the capital asset pricing model
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
(CAPM). Futures markets are not different in principal from the market for
any other asset. The returns on any risky asset are governed by the asset
contribution to the risk of a well-diversified portfolio. The classic CAPM
is applied by Dusak (1973) in the analysis of the risk premium and the
valuation of futures contracts.
Black’s (1976) model is used in Barone-Adesi and Whaley (1987) for
the valuation of American commodity options. This model is referred to as
the BAW (1987) model. It is helpful, as in Smithson (1991), to consider the
Black–Scholes model within a family tree of option pricing models. This
allows the identification of three major tribes within the family of option
pricing models: analytical models, analytic approximations, and numerical
models. Each analytical tribe can be further divided into three distinct
lineages: precursors to the Black–Scholes model, extensions of the Black–
Scholes model, and generalizations of the Black–Scholes model.
This chapter presents in detail the basic theory of rational option
pricing of European options and its applications along the Black–Scholes
lines and its extensions by Black (1976) for options on futures, and
European commodity options. The question of dividends, stochastic interest
rates, and stochastic volatilities are left to other chapters since the main
concern in this chapter is about analytical models under the Black and
Scholes’ (1973) assumptions. There is usually a difference between model
values and options market prices. There are three possible reasons for the
difference between the theoretical value and the market price. The first is
that the model gives the correct value and the option price is out of line.
In this case, it may be possible to trade profitably using this model. The
second reason is that the wrong inputs are used in the formula. The main
input is the volatility of the underlying asset over the life of the option that
must be estimated. The third reason is that the formula is wrong because
of its assumptions. These three reasons can explain the difference between
model values and market prices.
where
where ρ is the average rate of growth of the share price and Z corresponds
to the degree of risk aversion. As it appears in this formula, the parameters
corresponding to the average rate of growth of the share price and the
degree of risk aversion must be estimated. This reduces considerably the
use of this formula. Sprenkle (1961) tried to estimate the values of these
parameters, but he was unable to do so.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
The main idea to derive the formula was as follows: If the stock had
an expected return equal to the interest rate, so would the option. Hence,
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
all the stock’s risk could be diversified away, so could all the option’s risk.
In other terms, if the stock’s beta were zero, the option’s beta would have
to be zero too.
Hence, the expected terminal option’s value must be discounted at
the constant interest rate. The Black and Scholes (1973) formula can be
obtained using Sprenkle’s formula by putting in the interest rate for the
expected return on the stock and putting in the interest rate again for the
discount rate for the option.
out how much money could one have made if he had bought the options
whose prices seemed lower than the formula values and sold the options
whose prices seemed higher than the formula values. When transaction
costs are ignored, this trading rule seemed to generate substantial profits.
The second test was to assume buying the underpriced options and selling
the overpriced options at the values given by the formula. Galai (for more
details, refer to Bellalah et al., 1998) tested the formula using listed options
traded in the Chicago Board Options Exchange (CBOE) and found profits
which are much larger than those found in the OTC market. He tested the
profitability of spreads that are kept neutral continuously. A neutral spread
corresponds to a strategy of buying an option and selling another option on
the same underlying asset. A neutral spread is maintained when the long or
short positions are changed for every change in the underlying asset price
and time to maturity. This strategy involves buying one contract of the
underpriced option and selling either more or less than one contract of the
overpriced option. This strategy seemed to generate a consistent profit if
transaction costs and other trading costs were ignored. Today, traders use
the formula so much that the market prices are usually close to formula
values even in the presence of a cash takeover.
RS − r = βS [Rm − r] (8.5)
where,
This model gives a general method for discounting future cash flows
under uncertainty. Denote the value of the option by C(S, t) as a function
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
of the underlying asset and time. To derive their valuation formula, B–S
assumed that the hedged position was continuously re-balanced in order to
remain risk-less. They found that the price of a European call or put must
verify a certain differential equation, which is based on the assumption that
the price of the underlying asset follows a geometric Wiener process:
∆S
= αdt + σ∆z (8.6)
S
where α and σ refer to the instantaneous rate of return and the standard
deviation of the underlying asset, and z refers to Brownian motion. The
relationship between an option’s beta and its underlying security’s beta is:
CS
βC = S βS (8.7)
C
where,
R̄S − r = βS [R̄m − r]
where R̄S is the expected return on the asset S and R̄m is the expected
return on the market portfolio. This equation may also be written as:
∆S
E = [r + βS (R̄m − r)]∆t (8.8)
S
Using the CAPM, the expected return on a call option should be:
∆C
E = [r + βC (R̄m − r)]∆t (8.9)
C
When substituting for the option’s elasticity from Eq. (8.7), Eq. (8.11)
becomes after transformation:
C(S, T ) = S − K, if S ≥ K
C(S, T ) = 0 if S < K
is solved using standard methods for the price of a European call, which is
found to be equal to:
with
S 1 √ √
d1 = ln + r + σ2 T σ T, d2 = d1 − σ T
K 2
C(S, t)
S−
(8.18)
∂c(S,t)
∂S
∆c(S, t)
∆S −
(8.19)
∂c(S,t)
∂S
Since the return to the equity in the hedged position is certain, it must be
equal to r∆t where r stands for the short-term interest rate. Hence, the
change in the equity must be equal to the value of the equity times r∆t, or:
2
1 2 2 ∂ c(S,t) ∂c(S,t)
2 σ S ∂S 2 + ∂t ∆t c(S, t)
− ∂c(S,t)
= S − ∂c(S,t) r∆t. (8.22)
∂S ∂S
Dropping the time and re-arranging gives the Black–Scholes partial differ-
ential equation:
1 2 2 ∂ 2 c(S, t) ∂c(S, t) ∂c(S, t)
σ S − rc(S, t) + + rS = 0. (8.23)
2 ∂S 2 ∂t ∂S
This partial differential equation must be solved under the boundary condi-
tions expressing the call’s value at maturity date: c(S, t∗ ) = max[0, St∗ − K]
where K is the option’s strike price.
For the European put, the above equation must be solved under the
following maturity date condition: P (S, t∗ ) = max[0, K − St∗ ]. To solve
this differential equation, under the call-boundary condition, Black–Scholes
made the following substitution:
∗ 2 σ2 S 1
c(S, t) = er(t−t ) y 2 r − ln − r − σ 2 (t∗ − t) ,
σ 2 K 2
2
2(t∗ − t) 1 2
− r − σ (8.24)
σ2 2
The solution to this problem is the solution to the heat transfer equation
given in Churchill (1963):
„ 1 (u+q√2s)σ2
«
“ ”
∞ 2 2
1 r− 1 σ2 − q2
y(u, s) = √ K e 2 −1 e dq.
2Π −u
√
2s
Substituting from this last equation in Eq. (8.24) gives the following solution
for the European call price with T = t∗ − t
8.4.4. Examples
Example 1: When the underlying asset S = 18, the strike price K = 15,
the short-term interest rate r = 10%, the maturity date T = 0.25, and the
volatility σ = 15%, the call price is calculated as follows.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
V (F, t, F, t∗ ) = 0 (8.26)
Equation (8.26) implies that the forward contract’s value is zero when the
contract is initiated and the contract price, K, is always equal to the current
futures price F (t, t∗ ). The main difference between a futures contract and a
forward contract is that a futures contract may be assimilated to a series of
forward contracts. This is because the futures contract is re-written every
day with a new contract price equal to the corresponding futures price.
Hence, when F rises, i.e., F > K, the forward contract has a positive
value and when F falls, F < K, then the forward contract has a negative
value. When the transaction takes place, the futures price equals the spot
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
price and the value of the forward contract equals the spot price minus the
contract price or the spot price:
V (F, t∗ , K, t∗ ) = F − K. (8.27)
c(F, t∗ ) = 0. (8.28)
Using the fact that the return to a hedged portfolio must be equal to the
risk-free interest rate and expanding ∆c(F, t) gives the following partial
differential equation:
∂c(F, t) 1 2 2 ∂ 2 c(F, t)
= rc(F, t) − σ F
∂t 2 ∂F 2
or
1 2 2 ∂ 2 c(F, t) ∂c(F, t)
σ F − rc(F, t) + =0 (8.30)
2 ∂F 2 ∂t
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
p − c = e−rT (K − F ). (8.33)
equal to the spot price. We now give some examples for the calculation of
option prices using Black’s formula.
Equation (8.35) shows that the expected change in the futures price is
proportional to the dollar “beta” of the futures price. The expected change
in the futures price may be zero, positive, or negative. Also, the expected
change in the futures price is zero when the covariance of the change in
the futures price with the market portfolio is zero, i.e., E(∆S) = 0 when
cov(∆S, R̃m ) = 0.
priced using a minor modification of the Black and Scholes (1973) option
pricing formula. In deriving expressions for the behavior of the futures
price, he assumed that both taxes and transaction costs are zero and that
the CAPM applies at each instance of time. Following Black (1976), we
assume that the fractional change in the futures price is distributed log-
normally, with a known constant variance rate, σ. We also assume that
all the parameters of the CAPM are constant through time. Under these
assumptions, the value of the futures commodity option, C(S, t), can be
written as a function of the underlying futures price and time. In this
context, it is possible to create a risk-less hedge by taking a long position
in the option and a short position in the futures contract with the same
transaction date. Black (1976) assumed that a continuously re-balanced
self-financing portfolio of the underlying futures contracts and the risk-
less asset can be constructed to duplicate the payoff of the futures option.
The relationship between a commodity option’s beta and its underlying
CS
security’s beta is given by βC = S C βS , where βc and βS refer
respectively, to the betas of the commodity option and its underlying
commodity contract. The expected return on a security in the context of
the CAPM is:
R̄S − r = βS [R̄m − r]
or
This equation can be written for the expected return on the spot asset and
the option as:
∆S
E = r∆t + aβS ∆t
S
∆C
E = r∆t + aβC ∆t.
C
Multiplying this last equation by C and substituting for βC gives:
Simplification gives:
1
CSS S 2 σ 2 + rSCS − rC + Ct = 0.
2
This equation is the Black and Scholes (1973) equation.
Since the value of a futures contract is zero, the equity in the position
is just the value of the option. In this context, the system:
1
E(∆C) = rSCS ∆t + aSβS CS ∆t + Ct ∆t + CSS S 2 σ 2 ∆t
2
E(∆C) = rC∆t + aSβS CS ∆t,
becomes:
1
E(∆C) = Ct ∆t + CSS S 2 σ 2 ∆t
2
E(∆C) = rC∆t,
which gives:
1
CSS S 2 σ 2 − rC + Ct = 0.
2
This equation is the Black (1976) equation.
1
CSS S 2 σ 2 − rC + Ct = 0. (8.37)
2
Let T be the maturity date of the call and K be its strike price. The
equation must be solved under the boundary condition at maturity:
C(S, T ) = S − K if S ≥ K
C(S, T ) = 0 if S < K.
There is a simple relationship between the future price and the spot price
F = SebT . The value of a European commodity call is:
with
F σ2
ln + T √
K 2
d1 = √ , d2 = d1 − σ T .
σ T
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
The term F N (d1 ) − KN (d2 ) shows that the expected value of the futures
call at expiration, is the expected difference between the futures price and
the strike price conditional upon the option being in-the-money times the
probability that it will be in-the-money. The term e−rT is the appropriate
discount factor by which the expected expiration value is brought to the
present. It is possible to use directly the put-call parity to write down the
put formula. Re-call that the put-call parity relationship between puts and
call with identical strike prices is an arbitrage-based relationship, which
holds regardless of the distribution of financial asset prices. More general
results about calls and puts with different strike prices can be written down
for both symmetric and asymmetric processes using the propositions in
Bates (1997).
with
Se−dT σ2
ln K + 2
T √
d1 = √ , d2 = d1 − σ T .
σ T
with
Se−dT σ2
ln K + 2
T √
d1 = √ , d2 = d1 − σ T .
σ T
8.7.6. Taxes
The existence of different tax rates for institutions and individuals can affect
option values. The exact rules used to restrict tax arbitrage will affect option
values. For example, index option positions are taxed, in general, partly at
short-term capital gains rates and partly at long-term capital gains rates.
This depends on whether the position has been closed out each year. If
several investors pay taxes on gains and cannot deduct losses, they want
to limit the volatilty of their positions and have the freedom to control the
timing of their gains and losses. This can affect the use of options and the
option values.
8.7.7. Dividends
Dividends reduce call-option values and increase put-option values. Several
formulas are proposed in the literature to handle dividends, but the exact
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
8.7.8. Takeovers
The Black–Scholes model assumes that the stock will continue trading for
the option’s life. If, for example, firm A takes over firm B through an
exchange of stock, options on B’s stock will expire early. However, the stock
value is higher than before. The premium of the tender offer will increase
the call value and reduce the put value.
Summary
“Because options are specialized and relatively unimportant financial
securities, the amount of time and space devoted to the development of
a pricing theory might be questioned”, said Professor Merton (1973), in
Bell Journal of Economics and Management Science. Thirty years ago, no
one could have imagined the changes that were about to occur in finance
theory and the financial industry. The seeds of change were contained in
option theory being conducted by the Nobel Laureates Black, Scholes, and
Merton. Valuing claims to future income streams is one of the central
problems in finance. The first known attempt to value options appeared
in Bachelier (1900) doctoral dissertation using an arithmetic Brownian
motion. This process amounts to negative asset prices. Sprenkle (1961) and
Samuelson (1965) used a geometric Brownian motion that eliminated the
occurence of negative asset prices. Samuelson and Merton (1969) proposed
a theory of option valuation by treating the option price as a function of
the stock price. They advanced the theory by realizing that the discount
rate must be determined in part by the requirement that investors hold
all the amounts of stocks and the option. Their final formula depends on
the utility function assumed for a “typical” investor. Several discussions
are done with Merton (1973) who was also working on option valuation.
Merton (1973) pointed out that assuming continuous trading in the option
or its underlying asset can preserve a hedged portfolio between the option
and its underlying asset. Merton was able to prove that in the presence
of a non-constant interest rate, a discount bond maturing at the option
expiration date must be used. Black and Scholes (1973) and Merton (1973)
showed that the construction of a risk-less hedge between the option and
its underlying asset, allows the derivation of an option pricing formula
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
Questions
1. What is wrong in Bachelier’s formula?
2. What is wrong in Sprenkle’s formula?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
N (x) = 1 − n(x)(a1 k + a2 k 2 + a3 k 3 ) when x 0
1 − n(−x) when x < 0
where
1
k= , a1 = 0.4361836, a2 = −0.1201676,
1 + 0.33267x
1 2
a3 = 0.9372980, and n(x) = √ e−x /2 .
2π
The next approximation provides the values of N (x) within six decimal
places of the true value.
N (x) = 1 − n(x)(a1 k + a2 k 2 + a3 k 3 + a4 k 4 + a5 k 5 ) when x 0
1 − n(−x) when x < 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
where
1
k= , a1 = 0.319381530, a2 = −0.356563782,
1 + 0.2316419x
a3 = 1.781477937, a4 = −1.821255978, and a5 = 1.330274429.
where
a b
a1 = , b1 =
2(1 − ρ2 ) 2(1 − ρ2 )
x1 = 0.24840615 y1 = 0.10024215
x2 = 0.39233107 y2 = 0.48281397
x3 = 0.21141819 y3 = 1.0609498
x4 = 0.033246660 y4 = 1.7797294
x5 = 0.00082485334 y5 = 2.6697604
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
M (a, b; ρ) = φ(a, b; ρ)
2. If a 0, b 0, and ρ 0, then:
3. If a 0, b 0, and ρ 0, then:
4. If a 0, b 0, and ρ 0, then:
where M (a, 0; ρ1 ) and M (a, 0; ρ2 ) are computed from the rules, where the
product of a, b, and ρ is negative, and:
(ρa − b)Sign(a) (ρb − a)Sign(b)
ρ1 = , ρ2 =
a2 − 2ρab + b2 a2 − 2ρab + b2
1 − Sign(a) × Sign(b) +1 when x 0
δ= , Sign(x) =
4 −1 when x < 0
References
Ayres, HF (1963). Risk aversion in the warrants market. Industrial Management
Review, 4 (Fall), 497–505.
Bachelier, L (1900). Theorie de la speculation, Ph.D. Thesis in Mathematics,
Annales de l’Ecole Normale Superieure, III.17, 21–86.
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June), 301–320.
Bates, DS (1997). Jumps and stochastic volatility: exchange rate processes
implicit in Deutsche mark options. Review of Financial Studies, 9, 69–107.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch08
Chapter 9
Chapter Outline
This chapter is organized as follows:
1. Section 9.1 gives some applications of the Black and Scholes (1973)
option pricing theory.
2. Section 9.2 presents the main applications of the Black’s (1976) model.
3. Section 9.3 develops the main results in Garman and Kohlhagen’s (1983)
model for the pricing of currency options.
4. Section 9.4 presents the main results in the models of Merton (1973)
and Barone-Adesi and Whaley (1987) model for the pricing of European
commodity and commodity futures options. Some applications of the
model are also proposed.
5. Section 9.5 compares the Black–Scholes world with the real world.
Introduction
Broadly speaking, there are four groups of equity options: traded options,
over-the-counter (OTC) options, equity warrants, and covered warrants.
Traded options are standardized contracts which are listed on options
exchanges. These options are not protected against dividend and their strike
prices and maturity dates are set by the exchange. Stock index options and
futures markets have experienced remarkable growth rates. Stock index
options are of the European or the American type and often involve cash
settlement procedure upon exercise. The Black and Scholes (1973) model
403
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
can be used in the valuation of options for which the underlying asset is
a fixed-income instrument. However, this model presents some limitations
in the valuation of interest rate options. The standard binomial models
can also be applied for the valuation of interest rate options. Since the
Black and Scholes formula is valid for a non-cash paying security, it can be
used for the pricing of a zero-coupon bond. The Black and Scholes (1973)
model is universally applied by market participants even though several
other alternative models exist. The main question is why the Black–Scholes
model successful and how to apply the model when the imperfections of the
real world loom large.
This chapter presents the theory of European options and its applica-
tions along the Black–Scholes lines and its extensions by Black (1976) for
options on futures, Garman and Kohlhagen (1983) for options on currencies
and indirectly by Merton (1973), and Barone-Adesi and Whaley (1987) for
European commodity and futures options. These models and especially the
Black’s (1976) model apply for commodity options. Commodity markets
can be traced back to the corn markets of the Middle Ages when farmers,
merchants, and end-users would all meet in a specific place. The term
“commodity” refers, today, to a variety of products, ranging from the
traditional agricultural crops to oil and financial instruments. Since the
main concern in this chapter is about analytical models under the Black–
Scholes (1973) assumptions, the question of dividends, stochastic interest
rates, and stochastic volatilities are discussed in other chapters.
Calls and puts are bought or sold in anticipation of future cash flows,
for defensive purposes, and speculative reasons. The investor must choose
the appropriate options to be bought or sold. Therefore, the question of the
management of an option position is as important as the question of option
valuation and strategies.
Over-the-counter (OTC) options are tailor-made to the investor’s
needs and are often written by investment banks.
Equity warrants are long-term options and are often traded in
securities markets rather than in option markets. When these options are
exercised, new shares are issued by the company.
Covered warrants are OTC long-term options issued by securities
houses. These equity options can be valued using the Black–Scholes
model. However, the following specificities of these instruments imply some
extensions of the Black–Scholes model.
First, these options are frequently traded on an asset, which distributes
dividends and they are of the American type, i.e., they can be exercised
before maturity.
Second, the assumed diffusion process may not represent reality since
equity prices may jump downward or upward in response to either bad or
good news.
Third, it is more dificult to justify a constant volatility for the
underlying asset when the option maturity is long. The same argument
applies for the risk-less interest rate. In this chapter, we restrict our analysis
to the assumptions of the Black–Scholes model, which will be relaxed
afterwards when studying the extensions and generalizations of the model.
Many strategies can be implemented with equity derivatives. These
strategies are obviously not specific to equities. They also apply to options
on other types of underlying assets.
given by:
where
B
σ2
ln K + r+ 2 T √
d1 = √ , and d2 = d1 − σ T
σ T
where
B: price of the bond;
K: strike price of the option;
T : time to maturity of the option;
σ: instantaneous standard deviation of the bond price and
r: spot rate on a risk-free investment with a maturity date T .
Using the put-call parity relationship, the put’s value is given by:
A collar: When an investor buys a cap (floor) and sells the floor (cap),
this strategy is known as the collar.
The cap, the floor, and the collar can be valued using standard formulas
for call and put options.
in the bonds are of the American type. We give an application of this model
to options on zero-coupon bonds and options on coupon-paying bonds
under the Black–Scholes assumptions. If the bond is a coupon-paying bond,
then the present value of all coupons due during the option’s life must be
substracted from the bond’s price.
Example
Consider a European call for which the underlying asset is a coupon bond
with the following characteristics:
The option has a strike price equal to 100 Euro and its maturity date is in
one year. The present value of coupon payments is 9.59 Euro, or:
Receiver swaption gives the right to receive a fixed interest rate and payer
swaption gives the right to pay a fixed interest rate.
The buyer of a receiver swaption benefits when the interest rate falls
because he/she is guaranteed to receive a fixed rate, which is higher than
the floating rate. If interest rates rise, the swaption can be ignored because
the buyer can get a higher fixed rate in the market. The seller of the receiver
swaption is obliged to pay the fixed rate and to receive the floating rate in
the swap context. Most swaptions are of the European style.
Interest-rate swap agreements are second-order derivatives. Interest-
rate swaps reflect an equilibrium rate that equates a floating-rate stream
of payments with a fixed-rate stream of payments at the present date.
A swaption price can be computed using an option pricing model where
the underlying market input is the rate on the swap. The forward rate for
the expiry date of the swaption can be used for a swaption with a European
exercise. The following formula is often used to determine forward/forward
rates:
(1 + ST )T
F = (T − t) −1
(1 + rt )
where:
F : forward swap rate;
S: spot swap rate;
r: deposit rate for time t (the time to swaption expiration) and
T : term of the swap in years.
where:
W : price of receivers/payers option;
x1 , x2 : the prices of assets 1 and 2;
σ1 (σ2 ): the volatility of assets 1 (asset 2);
ρ1,2 : correlation coefficient;
t: current date and
t∗ : expiration date.
This formula is derived in the last chapter (Chapter 8). Tompkins (for
more details, refer to Bellalah et al., 1998) proposed the following put-call
relationship for swaptions expressed in annual terms:
Payer swaption − receiver
T
(prevailing swap rate − swaption strike)
swaption =
t=i
(1 + rt )t
where:
r: discount rate for the time period t;
T : term of the interest rate swap and
t: first exchange date of coupons.
These options must have the same value as European options on the
spot currency since they are not exercised before the maturity date, at
which the futures price is equal to the spot price.
The similarity between forward and futures prices suggests the use of
Black’s model for the valuation of these options.
Currency futures and currency forwards are often used to hedge
currency risks. Options on currency futures are applied in the currency-risk
management. The basic strategies of buying and selling calls and puts,
the vertical, diagonal, calendar, and volatility spreads are also applied in
the currency futures options markets. They can also be used in portfolio
insurance.
Several other applications of currency futures and options are used
to manage currency risks. Examples include basket options, average rate
options, cylinder options, and many other exotic options.
Note that the main difference between these formulae and those of B–S
model for the pricing of equity options is that the foreign risk-free rate is
used in the adjustment of the spot rate. The spot rate is adjusted by the
known “dividend”, i.e., the foreign interest earnings, whereas the domestic
risk-free rate enters the calculation of the present value of the strike price
since the domestic currency is paid over on exercise.
Examples
Assume that the US dollar/sterling spot rate is 1.8, the time to maturity is
three months, the three-month dollar interest rate is 7%, and the sterling
interest rate is 10%. When the volatility is 20%, the option price is
6.3817, or:
Note that the value of N (d1 ) is discounted to the present using the foreign
interest rate. This is because this rate is assumed to correspond to a
continuous dividend stream on the underlying asset. In the same way, we
can calculate risk parameters of other options.
∗
f = Se(r−r )T
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
which means simply that the formula for the pricing of a European call on
a spot currency can be re-written as:
Merton (1973) showed indirectly that the European call price is:
9.5.1. Volatility
The historical volatility can be used as a proxy of the future volatility of
the underlying asset. For example, it is possible to consider the prices of the
asset every day for the last 200 days and compute the standard deviation
of log returns on this arbitrary time interval. This gives a good estimate of
volatility with a standard error of around 5%. There are several other ways
of predicting volatility. In any case, the assumption of a constant volatility
is violated in the real world.
two bounds coalesce and become equal to the Black–Scholes price. If the
trader could predict the total variance exactly and if prices do not jump,
the options can be priced exactly.
Summary
Stock index options and futures markets have experienced remarkable
growth rates. Stock index options are either of the European or the
American type and often involve cash settlement procedure upon exercise.
Stock index options are traded on the major indices around the world.
These options are of the European or American type. Options on the spot
index are cash-settled and there is no physical delivery of the underlying
index. The price of any financial asset is given by the present value of its
expected cash flows. Options on index futures require upon exercise the
exchange of a long position in the future contract for a call and a short
position in the future contract for the put. Options on currency forwards
are traded in the OTC market. This market is regarded as the major
market for currency options. The growth of the OTC market is due to
its flexibility, since many banks and financial institutions offer options with
tailor-made characteristics in order to match the clients’ needs. Garman and
Kohlhagen (1983) provided a formula for the valuation of foreign currency
options. These options are traded on the foreign exchange market, which
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Questions
1. What are the main applications of the Black and Scholes’ model?
2. What are the main applications of the Black’s model?
3. What is meant by the interest-rate parity theorem?
4. What are the main characteristics of currency options and their
markets?
5. What is the main difference between Black and Scholes’ model and
Black’s model?
6. What is the main difference between Black and Scholes’ model and
Garman and Kohlhagen’s model?
7. What is inappropriate in the derivation of Garman and Kohlhagen’s
model?
8. What do you think of the assumptions underlying Garman and
Kohlhagen’s model?
9. What are the main differences between futures and forward contracts?
10. How can we obtain the formulas in Black and Scholes’ model, Black’s
model, and Garman and Kohlhagen’s model using the formula in
Merton and BAW?
11. What are the main specificities of index options and their markets?
12. How the Black and Scholes model is adjusted for index options?
13. What are the implications of arbitrage for index option markets and
their assets?
14. How indexes are constructed?
15. What is the main difference between zero-coupon bonds and coupon-
paying bonds?
16. What are the different types of bonds ?
17. What are the main specificities of short-term options on long-term
bonds?
18. What are the main specificities of bond options?
19. How can we obtain the put-call parity relationship for futures options
from that of European spot options?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Applications of Black–Scholes Type Models 427
Black–Scholes Valuation: Bloomberg
Appendix
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
428 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Applications of Black–Scholes Type Models 429
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
430 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Applications of Black–Scholes Type Models 431
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
432 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Applications of Black–Scholes Type Models 433
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
434 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Applications of Black–Scholes Type Models 435
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
436 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
References
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June), 301–320.
Black, F (1976). The pricing of commodity contracts. Journal of Financial
Economics, 3, 167–179.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
Brennan, MJ and ES Schwartz (1990). Arbitrage in stock index futures. Journal
of Business, 63, 7–31.
Britten-Jones, M and AF Neuberger (1996). Arbitrage pricing with incomplete
markets. Applied Mathematical Finance, 3(4), 347–363 and 11–13.
Britten-Jones, M and AF Neuberger (1998). Welcome to the real world. Risk,
September 11–13.
Evnine, J and A Rudd (1985). Index options: the early evidence. Journal of
Finance, 40(3), 743–756.
Garman, M. and S Kohlhagen (1983). Foreign currency option values. Journal of
International Money and Finance, 2, 231–237.
Grabe, JO (1983). The pricing of call and put options on foreign exchange. Journal
of International Money and Finance, 2, 239–253.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch09
Chapter 10
Chapter Outline
This chapter is organized as follows:
1. In Section 10.1, option price sensitivities are presented and the formulas
are applied.
2. In Section 10.2, the Greek-letter risk measures are simulated for different
parameters. The question of monitoring and managing an option position
in real time is studied for the different risk measures with respect to an
option pricing model.
3. In Section 10.3, some of the characteristics of volatility spreads are
presented.
4. In Appendix A, we give the Greek-letter risk measures with respect to
the analytical models presented in Chapter 9.
5. In Appendix B, we show the relationship between some of these Greek-
letter risk measures.
6. In Appendix C, we provide a detailed derivation and demonstration of
the hedging parameters.
7. In Appendix D, we provide a detailed derivation of the Greek-letter risk
measures.
439
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Introduction
As seen during the financial crisis 2008–2009, risk appears as a major
concern for the financial system and defies all that has been written by
academics in this area. The sensitivity parameters or Greek-letter risk
measures are important in managing an option position.
The delta measures the absolute change in the option price with
respect to a small change in the price of the underlying asset. It is given by
the option’s partial derivative with respect to the underlying asset price. It
represents the hedge ratio, or the number of options to write in order to
create a risk-free portfolio.
Call buying involves the sale of a quantity delta of the underlying asset
in order to form the hedging portfolio.
Call selling involves the purchase of a quantity delta of the asset to
create the hedging portfolio.
Put buying requires the purchase of a quantity delta of the underlying
asset to hedge a portfolio.
Put selling involves the sale of delta stocks to create a hedged portfolio.
The delta varies from zero for deep out-of-the-money (OTM) options
to one for deep in-the-money (ITM) calls. This is not surprising, since by
definition, the delta is given by the first partial derivative of the option
price with respect to the underlying asset. For example, the value of a deep
ITM call is nearly equal to the intrinsic value for which the first partial
derivative with respect to the underlying asset is one.
Charm is a risk measure that clarifies the concept of carry in financial
instruments. The concept of carry refers to the expenses due to the
financing of a deferred delivery of commodities, currencies, or other assets
in financial contracts. Even though charm is used by market participants as
an ad hoc measure of how delta may change overnight, it is an important
measure of risk since it divides the theta into its asset-based constituents.
The gamma measures the change in delta, or in the hedge ratio, as the
underlying asset price changes. The gamma is the greatest for at-the-money
(ATM) options. It is nearly zero for deep ITM and deep OTM options.
Approximately, the gamma is to the delta what convexity is to duration.
The gamma is given by the derivative of the hedge ratio with respect to
the underlying asset price. As such, it is an indication of the vulnerability
of the hedge ratio. The gamma is very important in the management and
the monitoring of an option position. It gives rise to two other measures of
risk: speed and color.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Hence, the delta is ∆c = N (2.8017) = 0.997. This delta value means that
the hedge of the purchase of a call needs the sale of 0.997 units of the
underlying asset. When the underlying asset price rises by 1 unit, from 18
to 19, the option price rises from 3.3659 to approximately (3.3659 + 0.997),
or 4.3629. When the asset price falls by 1 unit, the option price changes
from 3.3659 to approximately (3.3659 − 0.997), or 2.3689.
10.1.2. Gamma
The option’s gamma corresponds to the option’s second partial derivative
with respect to the underlying asset or to the delta partial derivative with
respect to the asset price.
(0.997 − 0.0727), or 0.9243. Also, a rise in the asset price from 18 to 19,
yields a change in the delta from 0.997 to (0.997 + 0.0727), or 1. This means
that the option is deeply ITM, and its value is given by its intrinsic value
(S − K). The same arguments apply to put options. The call and the put
have the same gamma.
10.1.3. Theta
The option’s theta is given by the option’s first partial derivative with
respect to the time remaining to maturity.
∂c −Sσn(d1 )
Θc = = √ − rKe−rT N (d2 )
∂T 2 T
Using the same data as in the example above, we obtain:
When the time to maturity is shortened by 1% per year, the call’s price
decreases by 0.01 (1.1918), or 0.011918 and its price changes from 3.3659
to approximately (3.3659 − 0.01918), or 3.3467.
∂p Sσn(d1 )
Θp = =− √ + rKe−rT N (d2 )
∂T 2 T
or Θp = −0.2653 + 0.0058 = −0.2594
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Using the same reasoning, the put price changes from 0.0045 to
approximately (0.0045 − 0.0025), or 0.002.
10.1.4. Vega
The option’s vega is given by the option’s price derivative with respect to
the volatility parameter.
10.1.5. Rho
The call’s rho
The option’s rho is given by the option’s first partial derivative with respect
∂c
to interest rates. In the B–S model, the call’s rho is given by Rhoc = ∂r =
−rT
KT e N (d2 ).
Using the above data Rhoc = e−0.1(0.25) (0.996)(0.25) = 3.64.
The Rho does not affect the call and put prices in the same way. In
fact, a rise in the interest rate yields higher call price (positive Rho) and
reduces the put price (negative Rho).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
10.1.6. Elasticity
The call’s elasticity
For a call option, this measure is given by Elasticity = S ∆c
c =
S
c N (d1 ) or
using the above data:
18
Elasticity = 0.997 = 5.3317
3.3659
The elasticity shows the change in the option price when the underlying
asset price varies by 1%. Hence, a rise in the asset price by 1%, i.e., 0.18,
induces an increase in the call price by 5.33%. The put price decreases by
12%. Hence, when the asset price changes from 18 to 18.18, the call’s price
is modified from 3.3659 to approximately (3.3659 (1 + 5.33%)), or 3.545. In
the same way, the put price changes from 0.0045 to (0.0045(1 − 12%)), or
0.00396.
accounts. Such a model allows them to know the variations in option price
sensitivities and their risk exposure. With such quantities, the monitoring
and the management of option positions are more easily achieved. We
illustrate the management of an option position in real time using the model
proposed indirectly in Merton (1973) and derived afterwards in Black (1975)
and Barone-Adesi and Whaley (1987) for the valuation of European futures
options. First, option prices are simulated and the sensitivity parameters
are calculated. Second, we study the risk-management problem in real time
with respect to option price sensitivities.
time to maturity. However, the gamma and the theta are more important
on near maturities.
Table 10.5 gives ITM call prices (S = 100, K = 90) for different levels
of the volatility parameter. When the delta is equal to 1, the gamma is
nearly equal to zero. Also, the vega is nearly nil and the theta is weak.
Table 10.6 gives the same information as Table 10.5, except that
calculations are done for ATM options (S = K = 100).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
to maturity. For an ATM put, there is more theta on short maturities and
more vega on longer maturities.
Table 10.11 gives OTM put prices when the volatility varies from 0.05
to 0.4.
When the volatility is respectively equal to 0.05 and 0.1, the put price
is nil and so are the sensitivity parameters as well.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
0 0.05 −0.00 0 0 0
0.01 0.10 −0.00 0 0.01 0.01
0.43 0.20 −0.09 0.02 0.08 0.34
1.50 0.30 −0.17 0.02 0.13 0.80
2.91 0.40 −0.23 0.02 0.15 1.22
Table 10.12 shows ATM put prices (S = K = 100) for different levels
of the volatility, parameter. Note that the put price, the delta (in absolute
value), the vega, and the theta are increasing functions of the volatility
parameter.
Note that the negative sign for the delta concerns only the put option.
The delta is −0.5 for an ATM put, −1 for a deep ITM put and 0 for a
deep OTM put.
The call delta is often assimilated to the hedge ratio. Since the
underlying asset delta is 1 and that of an ATM call is 0.5, the hedge ratio
is (1/0.5) or 2/1. Hence, we need two ATM calls to hedge the sale of the
underlying asset. Note that the delta is calculated in practice using the
observed volatility or the implicit volatility.
Delta-neutral hedging requires the adjustment of the option position
according to the variations in the delta. When buying or selling a call (put),
the investor must sell or buy (buy or sell) delta units of the underlying asset
to represent a hedged portfolio.
In practice, the hedged portfolio is adjusted nearly continuously to
account for the variations in the delta’s value. An initially hedged position
must be re-balanced by buying and selling the underlying asset as a function
of the variations in the delta through time.
The delta changes as the value of the underlying asset, the volatility,
the interest rate, and the time to maturity are modified.
Table 10.13 shows how to adjust a hedged portfolio in order to preserve
main characteristics of delta-neutral strategies.
It is important to note that delta-neutral hedging strategies do not
protect completely the option position against the variations in the
volatility parameter (Table 10.14).
It is also important to note that delta-neutral hedging does not protect
the option position against the variations in the time remaining to maturity.
The adjustment of the position when the time to maturity changes can be
done as explained in Table 10.15.
Note that the deltas are additive. For example, when buying two calls
having respectively, a delta of 0.2 and 0.7, the investor must sell 0.9 units
of the underlying asset in a delta-neutral strategy.
Long a call Delta increases: short more S Delta decreases: buy more S
Short a call Delta increases: buy more S Delta decreases: sell more S
Long a put Delta decreases: sell more S Delta increases: buy more S
Short a put Delta decreases: buy more S Delta increases: sell more S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Long a call
ITM Delta rises: sell more S
ATM Delta nonmodified
OTM Delta decreases: buy more S
Short a call
ITM Delta rises: buy more S
ATM Delta nonmodified
OTM Delta decreases: sell more S
Long a put
ITM Delta rises: buy more S
ATM Delta nonmodified
OTM Delta decreases: sell more S
Short a put
ITM Delta rises: sell more S
ATM Delta nonmodified
OTM Delta decreases: buy more S
he/she will lose money since he/she will sell the underlying asset at a lower
price. When the delta is positive, the investor is over-hedged with respect
to delta-neutral strategies.
When the delta (in monetary unit) is negative, the investor is shorting
the underlying asset. If the underlying asset price rises, the investor loses
money since he/she adjusts his/her position by buying more units of the
underlying asset. However, when the asset price falls, he/she makes a profit
since he/she pays less for the underlying asset.
When a portfolio is constructed by buying (selling) the securities and
derivative assets vj, the portfolio value is given by:
P = n1 v1 + n2 v2 + n3 v3 + · · · + nj vj
where nj stands for the numbers of units of the assets bought or sold.
The delta’s position, or its partial derivative with respect to the
securities and derivative assets is:
∂v1 ∂v2 ∂v3 ∂vj
∆position = + n2 + n3 + · · · + nj
∂S ∂S ∂S ∂S
= n1 ∆1 + n2 ∆2 + n3 ∆3 + · · · + nj ∆j
When the delta is constant, the gamma is zero. The gamma varies when
the market conditions change. The gamma is the highest for an ATM option
and decreases either side when the option gets ITM or OTM. The gamma
of an ATM option rises significantly when the volatility decreases and the
option approaches its maturity date.
When the gamma is positive, an increase in the underlying asset price
yields a higher delta. The adjustment of the position entails the sale of
more units of the underlying asset. When the asset price falls, the delta
decreases and the adjustment of the option position requires the purchase
of more units of the underlying asset. Since the adjustment is done in the
same direction as the changes in the market direction, this monitoring of
an option position with a positive gamma is easily done.
When the gamma is negative, an increase in the underlying asset price
reduces the delta. The adjustment of a delta-neutral position needs the
purchase of more units of the underlying asset. When the asset price
decreases, the delta rises. The adjustment of the option position requires
the sale of more units of the underlying asset. Hence, the adjustment of the
position implies a re-balancing against the market direction which produces
some losses (Table 10.16).
In general, the option’s gamma is a decreasing function of the time to
maturity. The longer is the time to maturity, the weaker is the gamma and
vice versa.
When the option approaches its maturity date, the gamma varies
significantly (Table 10.17).
The variations in the gamma for ITM, ATM and OTM options is
explained below in Table 10.18.
The management of an option position with a positive gamma is
simple. When the market rises, the investor becomes long and must sell
some quantity of the underlying asset to re-establish his/her delta-neutral
position. This produces a gain.
Long options Positive Market up: sell more S Easy adjustment, yields profits
Long options Positive Market down: buy more S Easy adjustment, yields profits
Short options Negative Market up: buy more S Difficult adjustment, yields losses
Short options Negative Market down: sell more S Easy adjustment, yields losses
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Gamma Near zero High for a shorter T Near zero for a near T
Gamma Near zero Stable for a longer T Near zero for a near T
Effect Weak Gamma is fundamental Weak
When the underlying asset decreases, the investor becomes short and
must buy more units of the underlying asset to re-establish his/her delta-
neutral position. This yields a profit.
The management of an option position with a negative gamma is
more difficult when the underlying asset’s volatility is high. When the
market rises, the investor becomes short and must buy some quantity
of the underlying asset to re-establish his/her delta-neutral position. This
produces a loss. When the underlying asset decreases, the investor becomes
long and must sell more units of the underlying asset to re-establish his/her
delta-neutral position. This yields a loss.
In general, one should be careful when adopting a positive gamma since
the option position loses from its theta when the market is not volatile. In
this context, it is better to have a negative gamma. However, when the
market is volatile, a position with a positive gamma allows profits, since
the adjustment requires buying the underlying asset when the market falls
and selling it when the market rises. The gamma of an option position with
several assets is given by:
∂∆position
Γposition =
∂S
∂∆1 ∂∆2 ∂∆3 ∂∆3
= n1 + n2 + n3 + · · · + nj
∂S ∂S ∂S ∂S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Examples
A theta of $1000 means that the option buyer pays $1000 each day for the
holding of an option position. This amount profits to the option writer.
The theta remains until the last day of trading. When a position shows a
positive gamma, its theta is negative. In general, a high gamma induces
a high theta and vice versa. For example, when Γ = 1500, theta may be
$10,000, i.e., a loss of $10,000 each day for the option position. This loss is
compensated by the profits on the positive gamma since the adjustments
of the position imply selling (buying) more units of the underlying asset
when the market rises (falls).
Table 10.19. The option value and the theta for an ATM option.
The impact of the vega on ATM option is the highest and is summarized
in Table 10.22.
When the vega is $50,000, this means that a rise in the volatility by 1%
produces a profit of $500. However, when the volatility decreases by 1%,
this implies a loss of $500.
When the vega is $50,000, an increase in the volatility by 1% implies a
loss of $500 and a decrease by 1% yields a profit of $500.
The vega of an option position is given by:
∂v1 ∂v2 ∂v3 ∂vj
Vegaposition = n1 − +n2 − +n3 − +· · ·+n1 − (7)
∂σ ∂σ ∂σ ∂σ
Summary
It is important to monitor the variations in a derivative asset price with
respect to its determinants or the parameters, which enter the option
formula. These variations are often known as Greek-letter risk measures.
The most widely used measures are known as the delta, charm, gamma,
speed, color, theta, vega, rho, and elasticity.
The delta shows the absolute change in the option price with respect
to a small variation in the underlying asset price. Charm corresponds to
the partial derivative of the delta with respect to time.
The gamma gives the change in the delta, or in the hedge ratio as the
underlying asset price changes.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
∂∆c 1
∆c = N (d1 ), Γc = = √ n(d1 )
∂S Sσ T
∂c Sσn(d1 )
Θc = = √ − rKe−rT N (d2 )
∂T 2 T
∂c √ ∂c
vc = = S T n(d1 ), Rhoc = = KT e−rT N (d2 ).
∂σ ∂r
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
∂∆p 1
∆p = ∆c − 1, Γp = = √ n(d1 )
∂S Sσ T
∂p Sσn(d1 )
Θp = =− √ + rKe−rT N (−d2 )
∂T 2 T
∂p √ ∂p
vp = = S T n(d1 ), Rhop = = −KT e−rT N (−d2 ).
∂σ ∂r
∂∆c e−rT
∆c = e−rT N (d1 ), Γc = = √ n(d1 )
∂S Sσ T
∂c Se−rT σn(d1 )
Θc = =− √ + rSe−rT N (d1 ) − rKe−rT N (d2 )
∂T 2 T
∂c √
vc = = Se−rt T n(d1 )
∂σ
∂∆p 1
∆p = ∆c − e−rT , Γp = = √ n(d1 )
∂S Sσ T
∂p Sσe−rT n(d1 )
Θp = − = √ − rSe−rT N (−d1 ) + rKe−rT N (−d2 )
∂T 2 T
∂p √
vp = = S T n(d1 )
∂σ
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
∗
−r ∗ T ∂∆c e−r T
∆c = e N (d1 ), Γc = = √ n(d1 )
∂S Sσ T
∗
∂c ∗ Se−r σn(d1 )
Θc = = r∗ Se−r T N (d1 − rKe−rT N (d2 )) − √
∂T 2 T
∂c ∗
Rhoc = ∗
= −T Se−r T N (d1 )
∂r
∂c ∗ √
vc = = Se−r T T n(d1 )
∂σ
∗
−r ∗ T ∂∆p e−r T
∆p = e [N (d1 − 1)], Γp = = √ n(d1 )
∂S Sσ T
∗
∂p ∗ Sσe−r T n(d1 )
Θp = = −r∗ Se−r T N (−d1 ) + rKe−rT N (−d2 ) − √
∂T 2 T
∂p ∗ √
vp = = Se−r T T n(d1 )
∂σ
∂p ∗
Rho = ∗ = T Se−r T N (−d1 )
∂r
∂∆c e(b−r)
∆ = e(b−r) N (d1 ), Γc = = √ n(d1 )
∂S Sσ T
∂p Se(b−r)T n(d1 )
Θc = = (r − b)Se(b−r)T N (d1 ) − rKe−rT N (d2 ) − √
∂T 2 T
∂c ∗ √ ∂c
vc = = Se−r T T n(d1 ), Rhoc = = T Se(b−r)T N (d1 ).
∂σ ∂b
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
∆p = −e(b−r)T [N (−d1 ) + 1]
∂∆p e(b−r)T
Γp = = √ n(d1 )
∂S Sσ T
∂p Se(b−r) σN (d2 )
Θp = = Se(b−r)N (−d1 ) + rKe−rT N (−d2 ) − √
∂T 2 T
∂p √
vp = = Se(b−r)T T n(d1 )
∂σ
∂p
Rho = = −T Se(b−r)T N (−d2 )
∂b
Using the definitions of the delta, gamma, and theta, the Merton and
Barone-Adesi and Whaley (1987) equation can be written as:
1 2 2 ∂ 2 c(S, t) ∂c(S, t)
σ S + bS − rc(S, t) + Θ = 0
2 ∂S 2 ∂S
or
1
rc(S, t) = Θ + bS∆ + S 2 σ 2 Γ.
2
For a delta-neutral position, the following relationship applies:
1 2 2
rc(S, T ) = S σ Γ+Θ
2
• Si : price of an asset i;
• C: value of the contract;
• σi : volatility of the underlying asset Si ;
• ρi,j : correlation coefficient between assets Si and Si ;
• r: instantaneous rate of return on accounting commodity and
• ri : instantaneous rate of return on the asset i.
with Hi = ∂∆ i
∂t .
The proof of this result relies on the replication equation which states
that:
N
Si ∆i = C
i=1
e−rT = (1 + τ T )−1
or
−rT = − ln(1 + τ T )
hence,
1
r= ln(1 + τ T )
T
and
Nj
T =
base × 100
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
∂C ∂d1 ∂d2
∆= = e(b−r)T N (d1 ) + Se(b−r)T N (d1 ) − Ke−rT N (d2 )
∂S ∂S ∂S
with:
1 d21 1 d22
N (d1 ) = √ e− 2 , N (d2 ) = √ e− 2
2Π 2Π
Since
∂d1 ∂d2 1
= = √
∂S ∂S Sσ T
then:
∂C 1 d21 1
= e(b−r)T N (d1 ) + Se(b−r)T √ e− 2 √
∂S 2Π Sσ T
1 d2
2 1
− Ke−rT √ e 2 √
2Π Sσ T
or
∂C 1 d21 1
= e(b−r)T N (d1 ) + e(b−r)T √ e− 2 √
∂S 2Π σ T
1 d2
1 S 1
−Ke−rT √ e(− 2 +ln K +bT ) √
2Π Sσ T
Hence, we have:
∂C 1 d21
= e(b−r)T N (d1 ) + √ e(b−r)T − 2
∂S σ 2ΠT
1 S d21 K
− √ eln K e((b−r)T − 2 )
σ 2ΠT S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
∆ = e(b−r)T N (d1 ).
∂P
∆= = Ke−rT N (−d2 )(−d2 ) − e(b−r)T N (−d1 )
∂S
− Se(b−r)T N (−d1 )(−d1 )
or
1 d2
2
∆ = −e(b−r)T N (−d1 ) + Ke−rT √ e(− 2 ) (−d2 )
2π
1 d2
1
− Se−(b−r)T √ e(− 2 ) (−d1 )
2π
e −rT
d22 d2
1
+√ Ke(− 2 ) (−d2 ) − SebT e(− 2 ) (−d1 )
2π
e−rT
(− d22 ) d2
1
− √ Ke 2 − SebT e(− 2 )
Sσ 2πT
and
1
−(−d1 ) = (−d2 ) = − √
Sσ T
which is also:
S 1 2
d21 = d22 + 2 ln + 2 b − σ T + σ2 T
K 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
S
or d21 = d22 + 2 ln K + 2bT
∂C ∂d1 ∂d2
v= = Se(b−r)T N (d1 ) − Ke−rT N (d2 )
∂σ ∂σ ∂σ
which is equivalent to:
∂C 1 −d2 −
d2
1
= Se(b−r)T √ e 2
∂σ 2Π σ
d2
−rT 1 − 22 −d2 √
− Ke √ e − T
2Π σ
or
d2
∂C (b−r)T 1 − 22 S
− ln K −bT d2
= Se √ e e e −
∂σ 2Π σ
d2
1 2 −d2 √
−K e−rT √ e − 2 − T
σ 2Π σ
Hence, we have:
∂C Ke−rT −
d2
2 −d2 Ke−rT −
d2
2 d2 √
= √ e 2 − √ e 2 − − T
∂σ 2Π σ 2Π σ
Finally, we obtain:
d2
√
1 − 2
v = e−rT √ e 2 T
2Π
∂P
v=
∂σ
This can be written as:
∂P
v= = Ke−rT N (−d2 )(−d2 ) − Se(b−r)T N (−d1 )(−d1 )
∂σ
or
∂P e−rT −
d2
2 d2 √ 1 −
d2
1 d2
v= =K√ e 2 − T − Se(b−r)T √ e 2
∂σ 2π σ 2π σ
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
and
„
d2
«
∂P 1 d2
− 21 d2 √ − 22 −ln S
−bT d2
= √ e−rT − SebT e
K
v= Ke − T
∂σ 2π σ σ
Hence,
2
∂P Ke−rT d2 d2 √ 1 d2
d2
v= = √ e2 − T − e− 2
∂σ 2π σ σ
Finally, we have:
−rT −
d2
1 T
v = Ke e 2
2π
We have used the following results in the derivation.
∂d1 d2 ∂d2 ∂d1 √ d2 √
− = , − = − T = − T.
∂σ σ ∂σ ∂σ σ
b = r − rf .
Hence:
∂C
ρ= = Se−rf T N (d1 )(d1 ) + T Ke−rT N (d2 ) − Ke−rT N (d2 )(d2 )
∂r
which is equivalent to:
d2
∂C 1 1 T
ρ= = T Ke−rT N (d2 ) + Se−rf T √ e 2 √
∂r 2Π σ T
d2
1 2 T
−Ke−rT √ e − 2 √
2Π σ T
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
or
∂C 1 −
d2
1 T
ρ= = T Ke−rT N (d2 ) + Se−rf T √ e 2 √
∂r 2Π σ T
d2
1 1 T
− Se−rT √ e − 2 e(r−rf )T √
2Π σ T
Finally, we obtain:
and
S
d1 = d21 − 2 ln − 2bT
K
which is
S
d1 = d21 − 2 ln − 2(r − rf )T
K
We also use the result:
„ «
d2 d2 d2
− 22
S
− 21 +ln K +(r−rf )T S 1
e =e = e − 2 e(r−rf )T
K
Hence, we have:
∂P
= Ke−rT N (−d2 )(−d2 ) + Se−rf T N (−d1 ) − Se(b−r)T N (−d1 )(−d1 )
∂rf
which is equivalent to:
2 d2
∂P −rf T e−rT d
− 22 bT − 21
= Se N (−d1 ) + √ Ke (−d2 ) − Se e (−d1 )
∂rf 2π
„ «
2 d2 d2
e−rT d
− 22 bT − 21 bT
S
− 22 −ln K −bT
+ √ Ke − Se e − Se e
σ 2πT
Hence, the Rho is given by:
or
or
√ √
d21 = d22 + 2(d1 − σ T )σ T + σ2 T
and
√
d21 = d22 + 2d1 σ T − 2σ 2 T + σ2 T
which is
√
d21 = d22 + 2d1 σ T − σ2 T
or
S
d21 = d22 + 2 ln + 2bT + σ 2 T − σ2 T
K
and
S
d21 = d22 + 2 ln + 2bT
K
Using this last expresion for d21 , the following relation:
„ «
d2
d22
1 − 21
√ e−rT Se (d1 )ebT − Ke (d2 )
2
2π
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
2π
K −
d2
2 K −
d22 1 σ
= √ e−rT e 2 ((d1 ) − (d2 )) = √ e−rT e 2 √
2π 2π 2 T
2 2πT
∂P
θ= = −rKe−rT N (−d2 ) + Ke−rT N (−d2 )(−d2 )
∂T
− (b − r)Se(b−r)T N (−d1 ) − Se(b−r)T N (−d1 )(−d1 )
or
1 σ
(−d2 ) = −(−d1 ) + √
2 T
or
∂P
= T Ke−rT (N (−d2 ) − 1)
∂r
d2
e−rT
2 d2
Ke − 2 (−d2 ) − SebT e− 2 (−d1 )
1
+√
2π
d2
e−rT T
− d22 2 S
+ √ Ke 2 − SebT e − 2 −ln K −bT
σ 2πT
Finally, we obtain:
∂P
v= = T Ke−rT (N (d2 ) − 1).
∂σ
or
d2
1 d2
2
∂C Se(b−r)T e− 2 e−rT e− 2
=− √ − erT N (d2 ) + √
∂K Kσ 2ΠT σ 2ΠT
S
d21 = d22 + 2 ln + 2bT
K
we obtain:
„ «
d2 S
− 21 −ln −bT d2
2
(b−r)T
e−rT e− 2
K
∂C S e e −rT
=− √ −e N (d2 ) + √
∂K K σ 2ΠT σ 2ΠT
or
∂C
= −erT N (d2 ).
∂K
∂P ∂(−d2 )
= e−rT N (−d2 ) + Ke−rT N (−d2 )
∂K ∂K
∂(−d1 ) 1 d2
2 1
− Se−(b−r)T N (−d1 ) + Ke−rT √ e− 2 √
∂K 2Π Kσ T
d2
S 1 d2 S 1 1 S
√ e(b−r)T √ e − 2 −ln K −bT
1
− √ e(b−r)T √ e− 2 −
Kσ T 2Π Kσ T 2Π
Hence, we have:
∂P
= e−rT N (−d2 ),
∂K
or
∂P
= −e−rT (N (d2 ) − 1)
∂K
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Questions
1. What is the option’s delta? Provide a simple derivation of this param-
eter in the context of analytical models.
2. What is the charm?
3. What is the gamma? Provide a simple derivation of this parameter in
the context of analytical models.
4. What does spread mean?
5. What does color mean?
6. What does theta mean? Provide a simple derivation of these parameters
in the context of analytical models.
7. What does vega mean? Provide a simple derivation of this parameter
in the context of analytical models.
8. What does rho mean?
9. What does elasticity mean?
10. Why the knowledge of these Greek-letter risk measures is important?
11. How does the delta change in response to the changes in the option
valuation parameters?
12. How does the gamma change in response to the changes in the option
valuation parameters?
13. How does the theta change in response to the changes in the option
valuation parameters?
14. How does the vega change in response to the changes in the option
valuation parameters?
15. How a hedged portfolio is adjusted in response to the changes in the
underlying asset price?
16. How a hedged portfolio is adjusted in response to the changes in the
volatility parameter?
17. How a hedged portfolio is adjusted in response to the changes in the
time to maturity?
18. What are the main characteristics of volatility spreads?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Applications of Option Pricing Models 479
Graphique
Appendix
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
480 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Applications of Option Pricing Models 481
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
482 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Applications of Option Pricing Models 483
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
484 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Applications of Option Pricing Models 485
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
486 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Applications of Option Pricing Models 487
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
488 Derivatives, Risk Management and Value
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
References
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June), 301–320.
Black, F (1975). The Pricing of Complex Options and Corporate Liabilities.
Chicago, IL: Graduate School of Business, University of Chicago.
Garman, M (1992). Spread the load. Risk, 5(11), 68–84.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch10
Part IV
491
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
492
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Chapter 11
Chapter Outline
This chapter is organized as follows:
1. Section 11.1 introduces continuous time stochastic processes for the
dynamics of asset prices. In particular, the Wiener process, the
generalized Wiener process, and the Ito process are presented and
applied to stock prices.
2. In Section 11.2, Ito’s lemma is constructed and several of its applications
are provided.
3. In Section 11.3, we intoduce the concepts of arbitrage, hedging, and
replication in connection with the application of Ito’s lemma. This allows
the derivation of the partial differential equation governing the prices of
derivative assets.
4. In Section 11.4, forward and backward equations are presented and some
applications are given. In particular, we give the density of the first
passage time and the density for the maximum or minimum of diffusion
processes.
5. In Section 11.5, a general arbitrage principle is provided.
6. In Section 11.6, we introduce some concepts used in discrete hedging.
7. Appendix A is a mathematical introduction to diffusion processes.
8. Appendix B gives the main properties of the conditional expectation
operator.
9. Appendix C reminds readers regarding the Taylor series formula.
493
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Introduction
This chapter explain how assets are priced with complete and incomplete
information. Incomplete information reflects billiquidity, and lack of trans-
parency. This can lead to a loss of confidence, as was the case of the finan-
cial crisis in 2008. The reader who is unfamiliar with mathematics can skip
this part of the book. Most financial models describing the dynamics of price
changes, interest rate changes, exchange rate variations, bond price changes,
and derivative asset dynamics among other things, present a term known as a
Wiener process. This process is a particular type of a general class of stochastic
processes known as Markov stochastic processes. A stochastic process can
be defined either in a simple way, as throughout this chapter, or in a more
mathematical sense, as in Appendix A. Our presentation is at the same time
intuitive and rigorous in order to allow the understanding of the necessary
tools in continuous time finance. These tools are not as complicated as an
uninformed reader might think. Using the definition of a stochastic process
enables us to define the standard Brownian motion and the Ito process. The
Ito process allows the construction of stochastic integrals and the definition
of Ito’s theorem or what is commonly known as Ito’s lemma. This lemma can
be obtained using Taylor’s series expansions or a more rigorous mathematical
approach. In both cases, some applications of this lemma to the dynamics of
asset and derivative asset prices and returns are provided. The introduction
of the concepts of arbitrage, replication, and the hedging argument, which
are the basic concepts in finance, allows the derivation of a partial differential
equation for the pricing of derivative assets. This equation first appeared in
Black and Scholes (1973) and Merton (1973). These authors introduced the
arbitrage theory of contingent claim pricing and, using Ito’s theorem, showed
that a continuously revised hedge between a contingent claim and its under-
lying asset is perfect. Since then, Ito’s theorem, the Black and Scholes hedge
portfolio, and the concepts of arbitrage and replicating portfolios have been
used by many researchers in continuous and discrete time finance. The basic
equivalent results in the theory of option pricing in a discrete time setting
were obtained by Cox et al. (1979) and Rendleman and Barter (1979). They
showed that option values calculated with discrete time models converge to
option values obtained by continuous time models. In other words, theoretical
work on convergence shows that some discrete time processes converge to
continuous time processes. For example, in the context of binomial models, an
option can be perfectly hedged using the underlying asset, and Ito’s theorem
can be implemented when constructing the hedging portfolio for infinitesimal
time intervals.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Some important questions regarding the use of Ito’s lemma and the
perfect hedge can be studied. The main question is whether a continuously
revised hedge is perfect over each revision interval or only when cumulating
the hedging error to zero over a large number of revision intervals. In all such
cases, these basic arguments lead to a partial differential equation, which
must be satisfied by the prices of derivative assets. This can be derived
using one of the two definitions of Ito’s lemma: the definition given in the
mathematical literature or simply the one obtained by an extension of the
Taylor series.
Using the independence property, it follows from this last equation that
the change W (T ) − W (0), is normally distributed with a √ mean of zero, a
variance of N ∆t = T , and a standard deviation equal to T . This is the
basic Wiener process with a zero mean or drift rate and a unit variance
rate. A mean or a drift rate of zero means that the future change is equal
to the current change. A variance rate of one means that the change at time
T is 1 × T .
Example: Consider a variable W following a Wiener process, starting at
W (0) = 20 (in years). This variable will attain in one year, a value which
is normally distributed with a mean of 20 and a standard deviation of 1. In
two years, its value follows
√ a similar type of distribution with mean 20 and a
standard deviation of 2. In n years, its value follows the same distribution
with mean 20 and a variance of n. What happens if the interval ∆t gets
very small, i.e., tends to zero. When ∆t gets close to zero, the analogous to
Eq. (11.1) is:
√
dW = ξ dt. (11.3)
E(Mt | Fs ) = Ms (11.4)
The dependence of both the expected drift rate and the variance rate on
X and time t, is the main difference from the generalized Wiener process.
This process has been extensively used in the finance literature, especially
for modeling stock price dynamics.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Ito process
We now give the mathematical definition of an Ito process. Consider a
probability space (Ω, Ftt≥0 , P ) with a filtration and (Mt )t≥T an (F )t -
Brownian motion.
An Ito process is a process (Mt )0≤t≤T having its values in R for which
for all t ≤ T :
t t
Xt = X0 + Ks ds + Hs dWs (11.9)
0 0
This process for stock prices, also known as the geometric Brownian motion
can be written in a discrete time setting as:
∆S √
= µ∆t + σξ ∆t (11.11)
S
where ξ is a random sample from a normal distribution with a zero mean
and a unit standard deviation. When the variance rate of return of the
stock price is zero, the expected drift in S over ∆t is:
dS
dS = µSdt or = µdt (11.12)
S
√
deviation σ ∆t, or:
∆S √
∼ N (µ∆t, σ ∆t) (11.13)
S
or
∆S = 100(0.005384 + 0.068639ξ).
or
1 2
ln(Sti ) ∼ N µ − σ (ti − t) + ln(St ), σ (ti − t) .
2
Example: If S = 100, the expected return is 15% per annum and the
volatility is 30% per annum, the distribution of ln(Sti ) in six months is:
6
ln(Sti ) ∼ N ln(100) + [0.15 − 0.5(0.09)](0.5), 0.3
12
or
then
1 2 σ
α∼N µ− σ , . (11.16)
2 (ti − t)
Example: What is the distribution of the actual rate of return over two
years for a stock having an expected return of 15% per annum and a
volatility of 30%?
The distribution is normal with a mean of 10.5%; (15% − 0.09
2 ), and a
0.3
√
standard deviation of 2 i.e., 21.21%.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
or
√
∆x = a(x, t)∆t + bξ ∆t.
In the limit, when ∆x and ∆t are close to zero, we cannot ignore, as before
the term in ∆x2 since it is equal to ∆x2 = b2 ξ 2 ∆t + terms in higher order
in ∆t. In this case, the term in ∆t cannot be neglected. Since the term ξ
is normally distributed with a zero mean, E(ξ) = 0 and a unit variance,
E(ξ 2 ) − E(ξ)2 = 1, then E(ξ 2 ) = 1 and E(ξ)2 ∆t is ∆t. The variance of
ξ 2 ∆t is of order ∆t2 and consequently, as ∆t approaches zero, ξ 2 ∆t becomes
certain and equals its expected value, ∆t. In the limit, Eq. (11.20) becomes:
∂f ∂f 1 ∂ 2f
df = ∆x + ∆t + b2 dt. (11.21)
∂x ∂t 2 ∂x2
dF
Applying Ito’s lemma, we obtain: dF = 4X 3 dX + 6X 2 dt since dt = 0. This
gives:
t t
dF = 4X 3 (τ )dX(τ ) + 6X 2 (τ )d(τ ) = X 4 (t)
0 0
and we have,
t t
1 4 3
X 3 (τ )dX(τ ) = X (t) − X 2 (τ )dτ.
0 4 2 0
In this case:
or
t t t t
3 2 2 3
2τ X(τ )dX(τ )+ 3τ X (τ )+ τ dτ = dF = t3 X 2 (t)−03 X 2 (0)
0 0 0 0
3
since 0 X 2 (0) = 0. Hence, we have:
t t t
3 3 2 1 2
τ X(τ )dX(τ ) + τ X (τ ) + τ 3 dτ = t3 X 2 (t).
0 2 0 2 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
or
t
1 1 4
τ 3 dτ = τ
2 0 8
then:
t t
1 3 2 3 1
τ 3 X(τ )dX(τ ) = t X (t) − τ 2 X 2 (τ )dτ − τ 4 .
0 2 2 0 8
t
Solution: When F = X 5 (t), then 0 dF = X 5 (t) − X 5 (0) = X 5 (t) since
X(0) = 0.
Applying Ito’s lemma to the function F gives:
dF = 5X 4 dX + 10X 3dt.
Hence:
t t
4
5X (τ )dX(τ ) + 10X 3 (τ )dτ = X 5 (t)
0 0
and
t t
1 5
X 4 (τ )dX(τ ) + 2 X 3 (τ )dτ = X (t)
0 0 5
or
t t
1
X (τ )dX(τ ) = X 5 (t) − 2
4
X 3 (τ )dτ
0 5 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
dS = µSdt + σSdW.
where by definition,
t
dX, Xt = Hs2 ds.
0
t t
1
Wt2 = W02 + 2Ws dWs + 2dX, Xs .
0 2 0
Since f (x) = (x2 ) = 2x, f (x) = (x2 ) = 2 and dX, Xs = ds, then:
t t t
1
Wt2 = 2Ws dWs + 2ds = 2 Ws dWs + t.
0 2 0 0
t t
Since W0 = 0 and 0 ds = t, then Wt2 − t = 2 0 Ws dWs .
t 2
Since the expected value of 0 Ws ds is finite, then (Ws2 − t) is a
martingale.
t
St = S0 + Ss [µSdS + σdWs ] (11.29)
0
t t
dSs 1 −1 2 2
ln(St ) = ln(S0 ) + + σ Ss ds. (11.30)
0 Ss 2 0 Ss2
t t
1
Yt = Y0 + µ − σ2 ds + σdWs . (11.31)
0 2 0
t t
This result is straightforward since 0 ds = t, 0 dWs = Wt − W0 with
W0 = 0. So, we have Yt = ln(St ) = ln(S0 ) + µ − 12 σ 2 t + σWt .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
This is the explicit formula for the underlying asset price when its
dynamics are given by the stochastic differential equation (11.29). The
following theorem which is stated without proof, shows that the solution to
Eq. (11.29) is unique.
Theorem: When (Wt )t≥0 is a Brownian motion, there is a unique Ito
process (St)0≤t≤T for which, for every t ≤ T :
t
St = S0 + Ss [µSdS + σdWs ] (11.32)
0
1 2
This process is given by St = S0 e(µ− 2 σ )t+σdWt . This process is used in the
derivation of the Black and Scholes formula and is often referred to as the
Black–Scholes process.
Example: The Black and Scholes (1973) model for the valuation of a
European options uses two assets: a risky asset with a price St at time
t and a risk-less asset Bt at time t. The dynamics of the risk-less asset or
bond are given by the following ordinary differential equation:
where r stands for the risk-less interest rate. The bond’s value at time 0,
B0 = 1 in a way such that Bt = ert . The dynamics of the risky asset or
stock are given by the following stochastic differential equation:
∂f ∂f 1 ∂2f
∆f = ∆xi + ∆t + ∆xi ∆xj
i
∂xi ∂t 2 i j ∂xi ∂xj
∂2f
+ ∆xi ∆t + · · · (11.36)
∂xi ∂t
√
Equation (11.35) can be discretized as follows ∆xi = ai ∆t+bi i ∆ zi where
the term i corresponds to a random sample from a standardized normal
distribution. The terms i and j reflecting the Wiener processes present
a correlation coefficient ρi,j . It is possible to show that when the time
interval tends to zero, in the limit, the term ∆x2i = b2i dt and the product
∆xi ∆xj = bi bj ρi,j dt. Hence, in the limit, when the time interval becomes
close to zero, Eq. (11.36) can be written as:
∂f ∂f 1 ∂2f
df = dxi + dt + bi bj ρi,j dt.
∂xi ∂t 2 ∂xi ∂xj
i i j
This gives the generalized version of Ito’s lemma. Using Eq. (11.35) gives:
∂f ∂f 1 ∂ 2f ∂f
df = ai + + bi bj ρij + bi dzi
∂xi ∂t 2 ∂xi ∂xj ∂xi
i i j
(11.37)
When the function f (.) has second-order partial derivatives in x and first-
order partial derivatives in t, which are continuous in (x, t), then the
generalized Ito’s lemma is:
t
∂f
f (t, Xt1 , . . . , Xtn ) = f (0, X01, . . . , X0n ) + (s, Xs1 , . . . , Xsn )ds
0 ∂s
n t
∂f
+ (s, Xs1 , . . . , Xsn )dXsi
i=1 0
∂x i
n
1 t ∂ 2f
+ (s, Xs1 , . . . , Xsn )dX i , Xjs
2 i,j=1 0 ∂xi ∂xj
(11.38)
with
p
p
dXsi = Ksi ds + Hsi,j dWsj , dX i , X j s = Hsi,m Hsj,m ds
j=1 m=1
In the financial literature, the notation is more compact than that used
in the mathematical literature. The term dX i , Xjs corresponds to the
changes in the instantaneous covariance terms Cov(dXsi , dXsj ).
since X(0) = 0.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
1
dF = nX n−1 dX + n(n − 1)X n−2 dt.
2
Hence:
t t
1
dF = nX n−1 (τ )dX(τ ) + n(n − 1)X n−2 (τ )dτ = X n (t)
0 0 2
⇐⇒
t
1 1
X n−1 (τ )dX(τ ) + (n − 1)X n−2 (τ )dτ = X n (t)
0 2 n
⇐⇒
t t
n−1 1 n − 1 n−2
X (τ )dX(τ ) = X n (t) − X (τ )dτ.
0 n 0 2
t
1 m n n − 1 t m n−2
τ m X n−1 (τ )dX(τ ) = t X (t) − t X (τ )dτ
0 n 2 0
m t m−1 n
− t X (τ )dτ.
n 0
t
dF = tm X n (t) − tm X n (0) = tmX n (t)
0
since X(0) = 0.
Using Ito’s lemma gives:
1
dF = ntm X n−1 dX + n(n − 1)X n−2dt + mtm−1 X n (t)dt.
2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Hence, we have:
t t t
1
dF = n τ m X n−1 (τ )dX(τ ) + n(n − 1) τ m X n−2 (τ )dτ
0 0 2 0
t
+m τ m−1 X n (τ )dτ = tm X n(t)
0
⇐⇒
t
1
τ m X n−1 (τ )dX(τ ) + (n − 1)τ m X n−2 (τ )dτ
0 2
t
m 1
+ τ m−1 X n (τ )dτ = tm X n (t)
n 0 n
⇐⇒
t t
m n−1 1 m n n − 1 m n−2
τ X (τ )dX(τ ) = t X (t) − τ X (τ )dτ
0 n 0 2
m t m−1 n
− τ X (τ )dτ
n 0
Example: We consider a function f (t) which is continuous and bounded
on the interval [0, t]. Using the integration by parts, we want to show that:
t t
f (τ )dX(τ ) = f (t)X(t) − X(t)df (τ )
0 0
Therefore,
t t
f (τ )dX(τ ) + X(τ )df (τ ) = f (t)X(t)
0 0
dS
= µdt + σdW (t) (11.39)
S
where the last term appears because dS 2 is of order dt. The last term in
Eq. (11.40) appears because the term (dS)2 is of order dt.
Omberg (1991) makes a decomposition of the last term in Eq. (11.40)
into its expected value and an error term. This allows one to establish a
link between Taylor’s series (dc) and Ito’s differential dcI as:
∂c ∂c 1 ∂2c
dc = dS + dt + σ2 S 2 dW 2 + de(t)
∂S ∂t 2 ∂2S
where
∂c ∂c 1 ∂ 2c
dcI = dS + dt + σ 2 S 2 dt
∂S ∂t 2 ∂ 2S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
and
1 ∂2c
de(t) = σ 2 S 2 dW 2 − dt .
2 ∂2S
The standard case: In general, the payoff of any derivative asset can
be created by an investment of a certain amount in the underlying risky
asset and another amount in risk-free discount bonds. Also, the payoff of a
derivative asset can be created using the discount bond, some options, and
the underlying asset. The portfolio which duplicates the payoff of the asset
is called the replicating portfolio. When using Ito’s lemma, the error term
de(t) is often neglected and, the equation for the option is approximated
only by the term dcI . The quantity dcI is replicated by QS units of the
∂c
underlying
asset an amount of cash Qc with QS = ∂S and Qc =
2 and
1 ∂c 1 ∂ c 2 2
r ∂S + 2 ∂ 2 S σ
S where r stands for the risk-free rate of return.
Hence, the dynamics of the replicating portfolio are given by:
∂c
dΠR = dS + rQc dt (11.42)
∂S
must be at least
∂c
dΠR = dS + (r + λ)Qc dt
∂S
with
1 ∂c 1 ∂ 2c 2 2
Qc = + σ S (11.43)
r+λ ∂S 2 ∂ 2S
where ΠR refers to the replicating portfolio. This shows that the required
return must cover at least the costs necessary for constructing the replicat-
ing portfolio plus the risk-free rate. In fact, when constructing a portfolio,
some money is spent and a return for this must be required. Hence, there
must be a minimal cost and a minimal return required for investing in
information at the aggregate market level. For this reason, the required
return must be at least λ plus the risk-less rate.
c = Qs S + Qc .
or
∂c 1 ∂c 1 ∂2c 2 2
c= S+ + σ S (11.44)
∂S r ∂t 2 ∂2S
and
1 ∂2c 2 2 ∂c ∂c
σ S + rc − r S+ = 0.
2 ∂2S ∂S ∂t
This equation is often referred to, in financial economics, as the Black–
Scholes–Merton partial differential
∂c equation. Note that the value of the
replicating portfolio is ΠR = ∂S S + Qc .
It is possible to implement a hedged position by buying the derivative
asset and selling delta units of the underlying asset:
∂c
ΠH = c − S = Qc (11.45)
∂S
where the subscript H refers to the hedged portfolio.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
c = QS S + Qc
or
∂c 1 ∂c 1 ∂ 2c
c= S+ + σ2 S 2 (11.46)
∂S (r + λ) ∂t 2 ∂ 2S
and
1 ∂2c 2 2 ∂c ∂c
σ S + (r + λ)c − (r + λ) S + = 0.
2 ∂2S ∂S ∂t
costs: cost λc on the option market and a cost λS on the underlying asset
market. In this case, we obtain the following more general equation as in
Bellalah and Jacquillat (1995) and Bellalah (1999):
1 2 2 ∂2c ∂c ∂c
σ S 2 + (r + λc )c − (r + λS ) S+ = 0. (11.47)
2 ∂ S ∂S ∂t
and
τ
de(t) = 0 for τ > 0.
0
These equations can be solved under the condition f (S, t0 ; S0 , t0 ) = δS0 (S),
i.e., at the initial time, S is equal to S0 and δS0 is the Dirac measure at S0 .
It is defined by δS0 (S)([a, b]) = 1, if S0 is ∈ [a, b] and zero else where.
Since we are interested only in time-homogeneous processes for which
σ = σ(S) and µ = µ(S), two types of constraints are sometimes imposed to
obtain an absorbing barrier and a reflecting barrier when the drift is finite.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
An absorbing barrier means simply that once the process attains a certain
value, this value will be conserved for all subsequent instants. A reflecting
barrier means that when the process hits a certain level, he/she will return
from the direction from which it comes.
When we constrain f (S1+ , t) to take the value zero, then S1 is an
absorbing barrier from above (+). The intuition of this result is that when
S = S1 at t1 , then S will conserve this value for all the subsequent instants
t after t1 . When we constrain the term,
1 ∂ 2 f (S, t)σ2 (S)
− [µ(S)f (S, t)] (11.52)
2 ∂S 2
where N [.] stands for the density of the standard normal distribution.
When the market bond price is set equal to this equality, this gives the
YTM or internal rate of return.
Duration
If we derive the expresion of the bond price with respect to the YTM, this
gives the slope of the price/yield curve,
N
dV
= −(T − t)P e−y(T −t) − (ti − t)Ci e−y(ti −t) .
dy
i=1
−1 dV
. (11.55)
V dy
If the discrete compounded rate is used, this quantity refers to the modified
duration. Duration is often computed for small movements in the yield in
order to examine the change in the price of the bond.
Convexity
We denote by δy, the change in the yield y. Using a Taylor’s series expansion
of V gives:
dV 1 dV 1 d2 V
= δy + [δy]2 + · · · (11.56)
V V dy 2V dy 2
N
d2 V 2 −y(T −t)
= (T − t) P e − (ti − t)2 Ci e−y(ti −t) . (11.57)
dy 2 i=1
1 d2 V
Convexity corresponds simply to V dy 2 .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
dB
= (r(t) + λB )B.
dt
dB
= r(t)B.
dt
For a coupon-paying bond, when coupons C(t)dt are paid dB in the time
interval [t, t + dt], then the holdings change by an amount dt + C(t) dt.
Again, if investors pay sunk costs to get informed about the bond and
the coupons, then we have:
dB
+ C(t) dt = (r(t) + λB )B. (11.58)
dt
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
where (rt )t≥0 is the spot-rate process and Et is the conditional expectation
in a risk-neutral measure P .
The price of a zero-coupon bond at time t with maturity T is:
T
P (t, T ) = Et exp − rs ds .
t
Rogers (1997) showed that it is a good solution to keep the same model
all the time and to express the new assets in terms of this same model.
To have another look on interest rate modeling, Rogers (1997) assumed a
reference probability P̃ with respect to which the risk-neutral probability
has a density:
dP
ρt =
dP̃ Ft
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
The final expression for the bond price corresponds to a different approach
in the modeling of interest rates called by Rogers (1997), as the “potential
approach”.
S = ex (11.62)
with
σ2 1
δx = µ− δt + σφδt 2 (11.63)
2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Π = V − ∆S. (11.64)
2
σ φ S ∂S 2 , which can also be written as 12 σ 2 S 2 ∂∂SV2 + 12 (φ2 − 1)σ 2 S 2 ∂∂SV2 .
The second term in this equation corresponding to the hedging error
is random. The hedging errors at each re-hedge can add up giving the
total hedging error. The latter has a zero mean and a standard deviation
1
of O(δt 2 ).
∂V 1 ∂ 2V ∂V
+ σ2S 2 + (r + λS )S − (r + λV )V + δt(. . .) = 0. (11.69)
∂t 2 ∂S 2 ∂S
When information costs are zero, the first term reduces to the Black–
Scholes equation. The second term is a correction to allow for the imperfect
hedge. Solving Eqs. (11.67) and (11.69) iteratively as in Willmott (1998),
the option price must satisfy.
∂V 1 ∂2V ∂V
+ σ2 S 2 2
+ (r + λS )S − (r + λV )V
∂t 2 ∂S ∂S
1 ∂ 2V
+ δt(µ − r)(r − µ − σ2 )S 2 =0 (11.70)
2 ∂S 2
∂V 1 ∂ 2V
∆= + δt(µ − r + λS + σ2 )S .
∂S 2 ∂S 2
This equation shows the growth rate of the asset and information costs
on the option and its underlying asset. The second derivative terms in
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
1 2 ∂ 2V 2
δΠ − rΠδt = S (ψ − σi2 δt).
2 ∂S 2
In an economy with information costs, a factor reflecting these costs must
be added to the interest rate r.
Summary
This chapter contains the basic and general material for the dynamics of
financial assets and derivative asset prices in a continuous time framework.
The presentation is made as simple as possible. The aim is to allow non-
familiar readers with these concepts to follow without difficulties the basic
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Questions
1. How the dynamics of asset prices are modeled in the literature?
2. What is a Wiener process or a Brownian motion?
3. What is the martingale property?
4. What is an Ito process?
5. What is the log-normal property?
6. Define the simple form and the generalized Ito’s formula.
7. What do you understand from the replication argument?
8. Why error terms are neglected?
9. What are the main concepts in bond markets?
10. What stipulates the general arbitrage principle?
11. What are the specific features of diffusion processes?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Definition: A stochastic process X(t) for which the changes in its values
over successive intervals are random, independent, and homogeneous is said
to have no “memory”. The process has no derivative and its path can be
represented by a continuous curve. The Wiener-Levy process is the most
regular process among the processes for which the changes are independant
and homogeneous.
E(E(X | B)) = X.
1 1
f (x + h) = f (x) + f (x)h + f (x)h2 + · · · + f n (x)hn .
2 n!
If the function f and its derivatives up to order n exist in the same region,
then using Taylor series, we have:
1 1
f (x + h) = f (x) + f (x)h + f (x)h2 + · · · + f (n−1) (x)h(n−1)
2 (n − 1)!
1 n ∗ n
+ f (x )h
n!
Exercises
Exercise: Find the values of u(w, t) and v(w, t) where dW (t) = udt+vdx(t)
in the presence of the following functions for W (t).
W (t) = X β (t)
W (t) = 1 + tα + etX(t)
W (t) = f (t)α X β (t)
with αβ ∈ N ∗ .
Solution: Ito’s lemma can be used for a function W (X(t), t):
∂W ∂W 1 ∂ 2W
dW = dX + dt + dt.
∂X ∂t 2 ∂X 2
For the first function W (t) = X β (t), we have:
β(β − 1) β−2 β−1 β(β − 1) β−2
dW = βX β−1 dX + X dt = βW β dX + W β dt.
2 2
Hence,
β−1 β(β − 1) β−2
u(W, t) = βW β ; v(W, t) = W β .
2
For the second function W (t) = 1 + tα + etX(t) , we have:
1
dW = tetX(t)dX + αtα−1 + X(t)etX(t) dt + t2 etX(t) dt
2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
or
etX(t) = W (t) − 1 − tα
hence:
∂f 1α(α − 1) β α−2
dW = αf β X α−1 dX + βX α f β−1 dt + f X dt
∂t 2
α−2
W (t) W (t) ∂f α(α − 1) W (t)
dW = α dX + β dt + dt.
X(t) f (t) ∂t 2 f (t)
Hence,
W (t)
u(W, t) = α = αf β X α−1
X(t)
and
α−2
W (t) ∂f α(α − 1) W (t)
v(W, t) = β +
f (t) ∂t 2 f (t)
with
W (t)
X α (t) =
f β (t)
Solution: Re-call that Ito’s lemma for a function f (S) is given by:
∂f ∂f 1 ∂2f
df = σS dX + µS + σ2 S 2 2 dt
∂S ∂S 2 ∂S
or
∂f ∂ 2f
= AeS , = AeS
∂S ∂S 2
S S 1 2 2
df = Ae σSdX + Ae µS + σ S dt.
2
The application of the lemma to g(S) gives:
∂g
= nS n−1 eS
n
∂S
∂ 2g
= ((nS n−1 )2 + n(n − 1)S n−2 )eS
n
∂S 2
hence,
dg = nS n−1 eS σSdX
n
n−1 S n 1 2 2 n−1 2 n−2 S n
+ SnS e µ + σ S ((nS ) + n(n − 1)S )e dt
2
then
n Sn n Sn 1 2 n 2 n Sn
dg = nS e σdX + nS e µ + σ ((nS ) + n(n − 1)S )e dt
2
or
then
n Sn 1 2 n
dg = nS e σdX + µ + σ (nS + n − 1) dt.
2
References
Bellalah, M (1999). The valuation of futures and commodity options with
information costs. Journal of Futures Markets, 19 (September), 645–664.
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13(3),
1895–1927.
Bellalah, M and B Jacquillat (1995). Option valuation with information costs:
theory and tests. The Financial Review, 30(3) (August), 617–635.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Bellalah, M and J-L Prigent (2001). Pricing standard and exotic options in the
presence of a finite mixture of Gaussian distributions. International Journal
of Finance, 13(3), 1975–2000.
Bellalah, M and F Selmi (2001). On the quadratic criteria for hedging under
transaction costs. International Journal of Finance, 13(3), 2001–2020.
Bellalah, M, JL Prigent and C Villa (2001a). Skew without skewness: asymmetric
smiles; information costs and stochastic volatilities. International Journal of
Finance, 13(2), 1826–1837.
Bellalah, M, Ma Bellalah and R Portait (2001b). The cost of capital in
international finance. International Journal of Finance, 13(3), 1958–1973.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659. Risk Premiums (1973). Journal
of Political Economy, 1387–1404.
Cox, J, S Ross and M Rubinstein (1979). Option pricing: a simplified approach.
Journal of Financial Economics, 7, 229–263.
Merton, R (1973). Theory of rational option pricing. Bell Journal of Economics
and Management Science, 4, 141–183.
Merton, RC (1987). A simple model of capital market equilibrium with incomplete
information. Journal of Finance, 42(3), 483–510.
Omberg, E (1991). On the theory of perfect hedging. Advances in Futures and
Options Research, 5, 1–29.
Rendleman, RJ and BJ Barter (1979). The pricing of options on debts securities.
Journal of Financial and Quantitative Analysis, 15 (March) 11–24.
Rogers, C (1997). One for all. Risk, 10(3) (March) 56–59.
Willmott, P (1998). Derivatives. New York: John Wiley and Sons.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch11
Chapter 12
Chapter Outline
This chapter is organized as follows:
1. In Section 12.1, we give some definitions and characterize complete
markets.
2. In Section 12.2, equity options are priced with respect to the par-
tial differential equation method and the martingale approach. Both
methods are applied to the valuation of equity options.
3. In Section 12.3, bond options and interest rate options are studied and
valued. Several models are proposed for the dynamics of interest rates.
4. In Section 12.4, the main techniques are proposed for the pricing of assets
in complete markets using the change of numeraire and time.
5. In Section 12.5, the main results in Section 12.4 are extended to account
for the effects of information uncertainty.
6. Appendix A presents the change in probability and the Girsanov
theorem.
7. Appendix B gives in great detail the resolution of the partial differential
equation under the appropriate condition for a European call option.
8. Appendix C gives two approximations of the cumulative normal distri-
bution function.
9. Appendix D states Leibniz’s rule for integral differentiation.
535
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Introduction
Modern finance allows to quantify risk and its remuneration. The pioneering
papers of Arrow (1953) and Merton (1973) assumed that markets are
complete. Prices of contingent claims are presented in the form of solu-
tions to partial differential equations (PDE). Prices are also represented
as conditional expectation of functionals of stochastic processes. These
representations provide solutions to the main pricing problems in modern
finance. The pricing of derivative assets is usually based upon two methods
which use the same basic arguments. The first method involves the reso-
lution of a PDE under the appropriate boundary conditions corresponding
to the derivative asset’s payoffs. This is often referred to as the Black–
Scholes method. The second method uses the martingale method, initiated
by Harrison and Kreps (1979) and Harrison and Pliska (1981), where the
current price of any financial asset is given by its discounted future payoffs
under the appropriate probability measure. The probability is often referred
to as the risk-neutral probability. Both methods are illustrated in detail for
the pricing of European call options. Black and Scholes (1973) and Merton
(1973) provide the PDE for derivatives and its solutions. The probabilistic
method known also as the martingale method, initiated by Cox and Ross
(1976), Harrison and Kreps (1979), and Harrison and Pliska (1981) allows
to compute the prices of derivatives under a risk-neutral probability, under
which the discounted price of any financial claim is a martingale. It must
be clear that both methods lead to the same results. Application of the
Feynman–Kac formula allows a switch from price as a solution of a PDE to
a probability representation. Unfortunately, for most problems in financial
economics, and in particular for the pricing of American options, there is
often no closed form solutions and option prices must be approximated.
Therefore, financial economists often resort to numerical techniques.
Vt (Φ) = St E ∗ [h | ST | Ft ] for t : 0, . . . , T.
where
d “ 2
”
1 − x2
N (d) = √ e dx
2Π −∞
where:
d 2
1 x
N (d) = √ exp − dx
2Π −∞ 2
given by:
or
The option price in the Black and Scholes (1973) economy can
be computed using its discounted expected terminal value under the
appropriate probability P ∗ as:
ct = E ∗ [e−r(T −t) h | Ft ]
with h(ST − K)+ = f (ST ). The option price at time t can be expressed as
a function of time and the underlying asset price. In fact,
ct = E ∗ [e−r(T −t) h | Ft ]
or
∗ ∗ 1 2
ct = E ∗ e−r(T −t) f St e−r(T −t) eσ(WT −Wt )− 2 σ (T −t) |Ft .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
∗ 1 2
Since for all t, St = S0 eσWt − 2 σ t , and at the option’s maturity date
∗ 1 2
ST = S0 eσWT − 2 σ T the ratio of the stock price between two dates is:
ST ∗ ∗ 1 2
= e[σ(WT −Wt )− 2 σ (T −t)]
St
h(y) −y2
Ep∗ [h(Y )] = √ e 2 dy,
2Π
since
√
H(t, S) = e−r(T −t) Ep∗ G( T − t )Y
where
√ √ σ2
G( T − t)Y = f Seσ( T −t)Y +(r− 2 )(T −t)
1 2
and Y follows N (0, 1) with density √12Π e− 2 y under P ∗ .
If one replaces the call’s pay off function, then using Eq. (12.1) gives:
σ2 ∗ ∗ +
H(t, S) = E ∗ e−r(T −t) Se(r− 2 )(T −t)+σ(WT −Wt ) − K
or
√ σ2 +
H(t, S) = E ∗ Se(σY τ − 2 τ ) − Ke−rτ (12.2)
where τ = T − t.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
with
1, if Z ≥ K
IZ≥K =
0 else
or
√ 1 K
σY τ − σ2 τ ≥ ln .
2 S
Hence
K
ln S − r − 12 σ 2 τ
Y ≥ √
σ τ
or
Y ≥ −d2 or Y + d2 ≥ 0.
or
∞ √ 2 1 1 2
H(t, S) = Se(σY τ − σ2 τ )
− Ke−rt √ e(− 2 y ) dy
−d2 2Π
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
The second integral is equal to −Ke−rτ N (d2 ). Using the change in variable
√ √
z = y + σ τ or y = z − σ τ , the first integral can be written as:
√
d2 +σ τ √ √ √ 2
2 1 1
Se(−σ(z−σ τ ) τ − σ2 τ )
√ e(− 2 (z−σ τ ) ) dz
−∞ 2Π
Simple computation of this integral gives exactly SN (d1 ). The sum of both
integrals corresponds to the Black–Scholes formula:
with
S S
ln K + r + 12 σ 2 (T − t) ln K + r − 12 σ 2 (T − t)
d1 = √ , d2 = √ .
σ T −t σ T −t
then
(µ − r − λS )
Θt = ,
σ
there is a probability P ∗ equivalent to P under which Wt∗ is a standard
Brownian motion. Hence, under P ∗ , we deduce from this last equation that
St∗ is a martingale and:
∗ ∗ ∗ 1 2
St = S0 exp σWt − σ t .
2
h(Y ) −y2
Ep∗ [h(Y )] = √ e 2 dy,
2Π
since
√
H(t, S) = e−(r+λc )(T −t) Ep∗ [G( T − t)Y ]
where
√ √ σ2
G( T − t)Y = f S exp σ( T − t)Y + (r + λS ) − (T − t)
2
1 2
and Y follows N (0, 1) with density √1 e− 2 y under P ∗ .
2Π
where a, b, and σ are positive constants. This Gaussian model allows for
the possibility of negative interest rates.
Consider a security paying a continuous dividend at a rate (h, rt , t)
whose payoff is a function of the interest rate r, FT = f (rT ) at time T .
The price process Ft of this security has the representation Ft = v(rt , t)
where v is solution to the following PDE:
∂v 1 ∂ 2v ∂v
(r, t) + σ 2 2 (r, t) + (a − br) (r, t) − rv(r, t) + h(r, t) = 0,
∂t 2 ∂r ∂r
(12.6)
where
1
n(t, T ) = (1 − e−b(T −t) )
b
and
T T
σ2 2
m(t, T ) = n (u, T )du − a n(u, T )du
2 t t
or
2 σ2
vH (t, T ) = (1 − e−2b(T −t) )(1 − e−b(H−t) )2
2b3
The formula for the call on a bond is given by:
with
ln (B(t, H)/B(t, T )) − ln K ± 12 vH
2
(t, T )
d1,2 (t, T ) =
vH (t, T )
for every t less than T and H. This formula does not show the coefficient
a and accounts for the bond volatility.
where a, b, and σ are positive constants. This process does not allow for
negative interest rates because of the square root in the diffusion process.
The price process of a standard European option Ft = v(rt , t) must satisfy
the following partial differential equation:
∂v 1 ∂2v ∂v
(r, t) + σ2 r 2 (r, t) + (a − br) (r, t) − rv(r, t) + h(r, t) = 0,
∂t 2 ∂r ∂r
under the boundary condition:
v(r, T ) = f (r, T ).
CIR (1985) provide the following closed-form solutions for the price of a
zero-coupon bond.
2a Γebr/2
m(t, T ) = 2 ln
σ Γ cosh Γr + 12 b sinh Γr
and
sinh Γr
n(t, T ) = ,
Γ cosh Γr + 12 b sinh Γr
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
with
τ = T − t, 2Γ = (b2 + 2σ 2 ).
They give also solutions in closed forms for options on zero-coupon and
coupon-bearing bonds. Longstaff (1990) proposes closed-form formulas for
European options on yields in the context of the CIR (1985) model. The
yield on a zero-coupon bond is a linear function of the short-term rate in
the CIR model since:
The yield at time t for a zero-coupon bond with a maturity τ for a current
level rt = r is Ỹ (t, τ ). The payoff of a yield European call is given by:
CTY = (Ỹ (rT , τ ) − K)+ , where K is the fixed level of the yield.
where m, n, and p are known functions. In this model, the yield of the bond
is a non-linear function of the short-term rate.
the Vasicek (1977) and the CIR (1985) model. When β = 0, this gives a
generalized Vasicek model:
where the different parameters a(t), b(t) and σ(t) are deterministic
functions.
Under the risk-neutral probability Q, the dynamics of return on zero-
coupon bonds in the absence of default are given by:
dP (t, T )
= r(t)dt − σp (t, T )dW1 (t)
P (t, T )
dSt
= r(t)dt + σS ρdW1 (t) + 1 − ρ2 dW2 (t)
St
dSt
= rdt + σS dW2 (t)
St
Using arbitrage arguments, Black and Scholes derived the following
PDE:
∂C 1 ∂ 2C ∂C
+ σS2 S 2 2 + rS − rC = 0 (12.12)
∂t 2 ∂S ∂S
Consider a replicating portfolio V consisting of a long position in the
∂C
option and a short position
∂C in ∂S units of the underlying asset. The
portfolio
results from investing ∂S in the risky asset and an amount C − S ∂C ∂S
in the risk-free asset. Its dynamics are given by:
∂C ∂C
dV = C − S rdt + dS (12.13)
∂S ∂S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
∂C 1 ∂ 2C
+ σS2 f 2 2 − rC = 0 (12.15)
∂t 2 ∂f
∂C
dV = rCdt + df
∂f
since
S0 2
f0 τ∗ σS
ln K + 2 ln K + r+ 2 T
√ = √
τ∗ σS T
A simple comparison of Eqs. (12.16) and (12.12) reveals that the option
is priced as if the interest rate were zero. Besides, the volatility is replaced
by 1 and the maturity T by τ ∗ . This result comes from the changes in
numeraires but, in practice, the interest rate enters the cost of carry and
the time change comprises the volatility of the forward contract.
We introduce these concepts and make the analogy with partial differential
equations approach.
Change of numeraire
A numeraire corresponds in general to a process X(t) for which there
exists a probability measure QX which is defined by its Radon–Nikodym
derivative with respect to Q as:
in such a way that the relative price of the asset compared with the
numeraire X is a QX martingale. In the above expression B(t) corresponds
to the capitalization factor and Q is the risk-neutral probability.
In fact, as we know, in the martingale approach, the price of any
derivative asset with a payoff h(T ) can be computed using the absence
of arbitrage opportunities argument. This allows the computation of the
price of any
contingent
claim under the risk-neutral probability Q as
h(T )
V0 = EQ B(T ) , where EQ corresponds to the conditional expectation
under the probability Q.
This conditional expectation can also be written as:
h(T ) h(T )
EQ = X(0)EQX
B(T ) X(T )
dSt
= rdt + σS dW2 (t)
St
In the presence of stochastic interest rates, the option price can be com-
puted as:
ST − K
CS = EQ IST ≥K
B(T )
∂CS 1 ∂ 2 CS 1 ∂ 2 CS ∂ 2 CS
+ σS2 S 2 2
+ σp2 P 2 2
− ρσS σP SP
∂t 2 ∂S 2 ∂P ∂S∂P
∂CS ∂CS
+ rS + rP − rCS = 0 (12.18)
∂S ∂P
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
∂CS ∂CS
CS = S +P
∂S ∂P
∂CS 1 ∂ 2 CS 1 2 2 ∂ 2 CS ∂ 2 CS
+ σS2 S 2 + σ P − ρσS σP SP =0 (12.19)
∂t 2 ∂S 2 2 p ∂P 2 ∂S∂P
Now, a change of variables can be made using the bond price P (t, T )
as a new numeraire. We denote by Ft the price of a forward contract on
St
the underlying asset S for delivery at time T , Ft = P (t,T )
and by CF the
price of a European call on the forward contract CF = CPS .
Using this new change of variables, Eq. (12.19) can be written as:
∂CF 1 ∂ 2 CF
+ σF (t, T )2 F 2 =0 (12.20)
∂t 2 ∂F 2
dQS ST P (0, T )
=
dQP S0 P (T, T )
ST P (T ,T )
where
2
t
M t = x2i (u)du.
i=1 0
St
The quantities P (t,T )
and P (t,T
S
)
are respectively QP and QS martin-
gales and satisfy
t
St P (t, T )
ln = ln = [σS2 + σp (u, T )2 + 2ρσp (u, T )σS ]du
P (t, T ) t St t 0
with
t
∗
τ = τ (T ) = [σS2 + σp (u, T )2 + 2ρσp (u, T )σA ]du.
0
We can use Eq. (12.21) to show the existence of two standard one-
dimensional Brownian motions Y P and Y S under the appropriate prob-
abilities QP and QS such that
St S0 Y p − 1 τ (t) P (t, T ) P (0, T ) A 1
= e τ (t) 2 , = eYτ (t) − τ (t).
P (t, T ) P (0, T ) At A0 2
Using these two last expressions and Eq. (12.22), the European call price is:
CS = S0 EQA IY S ≤ln S0 + τ − KP (0, T )EQS IY P ≥ln S0 + τ ∗
τ∗ KP (0,T ) 2 τ∗ KP (0,T ) 2
under the appropriate terminal condition CF = max[F −K, 0]. The equation
gives the price of a European option of a forward contract with a strike price
K, a maturity date τ ∗ and a unit volatility for the underlying asset. The
solution as given by the martingale approach (12.23) can also be stated as
the solution to the standard Black–Scholes–Merton (1973) equation:
S0
CS = P (0, T )CF (F0 ) = P (0, T )C , τ ∗ , 1, 0
P (0, T )
The above analysis shows that the change of numeraire absorbs the
stochastic character in the pricing of stock options when interest rates are
random. This allows to transform the economy into a new economy in which
the forward volatility is unity.
dSt
= (r + λS )dt + σS dW2 (t)
St
∂C 1 ∂2C ∂C
+ σS2 S 2 2 + (r + λS )S − (r + λc )C = 0 (12.24)
∂t 2 ∂S ∂S
The difference between Eqs. (12.26) and (12.25) is the partial differential
Eq. (12.24). The solution for a European call is:
τ = σS2 t, τ ∗ = σS2 T.
∂C 1 ∂ 2C
+ σS2 f 2 2 − (r + λc )C = 0 (12.27)
∂t 2 ∂f
This last equation corresponds to the extended Black (1976) equation in the
presence of information costs. It gives the value of an option on a forward
contract f . The dynamics of the replicating portfolio can be written as:
∂C
dV = C(r + λc )dt + df
∂f
gives the dynamics of C. When Cf = Ce(r+λc )(T −t) , this change of variables
accounts for the following expression:
∂Cf 1 2 2 ∂ 2 Cf
+ σA f =0 (12.28)
∂t 2 ∂f 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
∂Cf
The dynamics of the replicating portfolio are given by dVf = ∂f df and
Ito’s lemma gives:
∂Cf 1 2 2 ∂ 2 Cf ∂Cf
dVf = + σS f 2
dt + df
∂t 2 ∂f ∂f
∂Cf 1 ∂ 2 Cf
+ f2 = 0. (12.29)
∂τ 2 ∂f 2
since
S 2
f0 τ∗ σS
ln K + 2 ln K
0
+ r + λS + 2 T
√ = √
τ∗ σS T
A simple comparison of Eqs. (12.29) and (12.24) reveals that the option is
priced as if the interest rate were zero and there are no information costs
in the economy. Besides the volatility is replaced by 1 and the maturity T
by τ ∗ .
The effect of this change of numeraire can be appreciated using the Black–
Scholes context for the pricing of a European call option on an asset S in
the presence of a stochastic interest rate. Let us use the following dynamics
for the underlying asset
dSt
= (r + λS )dt + σS dW2 (t)
St
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
CS = max[S − K, 0]
ST P (T ,T )
Summary
This chapter contains the basic material for the pricing of derivative assets
in a continuous-time framework. The presentation is made as simple as
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Questions
1. Provide a definition of a complete market.
2. How equity options are priced with respect to the partial differential
equation method? and the martingale approach?
3. How equity options are priced with respect to the martingale approach?
4. How bond options and interest rate options are valued?
5. Describe the main techniques for the pricing of assets in complete
markets using the change of numeraire and time.
6. Describe the resolution of the partial differential equation under the
appropriate condition for a European call option.
t
then Wt∗ = Wt + 0
Θs ds is a standard Brownian motion.
under the following terminal condition which must be satisfied by the call
price at its maturity date c(S, t∗ ) = max[0, St∗ − K]. Let us try a change
of variables and postulate that the solution is of the following form:
We can search for the functions f , u1 (S, t) and u2 (S, t). Assume that
∂f
−rf (t)y(u1 , u2 ) + y(u1 , u2 ) = 0
∂t
or
∂f
rf (t) = .
∂t
∗
Hence, the function f (t) is given by f (t) = er(t−t ) . We denote by
2
1 2 2 ∂u1
a2 = σ S .
2 ∂S
is equal to zero if
1 ∂2y ∂u22 ∂u2 ∂u2
a = σ2 S 2
2
+ rS +
2 ∂u21 ∂S 2 ∂S ∂t
Assuming that
∂u2 ∂u22
= =0
∂S ∂S 2
u2 = −a2 (t − t∗ )
or
√ a S
u1 = 2 ln + b(t)
σ K
which is equivalent to
1 √ √ a
aσ 2 + b (t) + r 2 =0
2 σ
Re-call that the solution to the heat transfer equation is given by:
+∞ 2
1 (− 2u
ε
)
y(u1 , u2 ) = √ U0 (ε)e 2 dε (B.11)
2Πu2 −∞
or
S σk
ln =√ .
K 2a
Hence
( √σk )
S = Ke 2a .
u√
1
The term u2 2
is given by:
√
2
u1 a 2 S σ ∗ 1
√ = ln + − r (t − t ) √
u2 2 σ K 2 2a(t − t∗ )
or
S
σ2
u1 ln K + 2
− r (t − t∗ )
√ =− √
u2 2 σ t∗ − t
and we obtain the following result:
√
σ a 2 S σ2 √ √
√ ln + − r (t − t∗ ) + qa 2 t∗ − t
a 2 σ K 2
2
S σ √
= ln + − r (t − t∗ ) + qσ t∗ − t
K 2
If we denote this last term by
2
S σ √
d = ln + − r (t − t∗ ) + qσ t∗ − t
K 2
and substitute these values in the integral solution of the heat equation, we
obtain:
∞
K ln( S )+( σ2 −r)(t−t∗ )+qσ√t∗ −t q2
y(u, t) = √ e K 2 − 1 e(− 2 ) dq (B.14)
2Π d
Equation (B.14) can also be written as:
∗
Ker(t−t )
y(u, t) = √
2Π
∞ ∞
√ q2 S σ2 ∗ q2
e(qσ t−t ) e(− 2 ) dq e( 2 −r)(t−t ) −
∗
× e(− 2 ) dq
d K d
(B.15)
1 √ 2 1 √ 2
q + σ t∗ − t + q − σ t∗ − t = q2 + σ2 (t∗ − t)
2 2
√∗
and letting p = q + σ t − t the integral can be written as:
√
−d+σ t∗ −t √
q2
)(−qσ t∗ −t)
e(− 2 dq
−∞
−d 2
(− q2 − √ qσ )
= e ∗ t −t
dq
−∞
2
−d √
( σ2 )(t∗ −t) 1
t∗ −t)2 )
=e e(− 2 (q+σ dq
−∞
which is equivalent to
2
√
−d+σ t∗ −t
( σ2 )(t∗ −t) 1 2
e e(− 2 p ) dp.
∞
∗ σ2 2
−r)(t−t∗ )] [ σ2 (t−t∗ )] ∗
y(u, t) = er(t−t ) e[( 2 e SN (d1 ) − Ker(t−t ) N (d2 )
with
S S
σ2 r−σ2
ln K
+ 2
+ r (t∗ − t) ln K
+ 2
(t∗ − t)
d1 = √ , d2 = √ .
σ t∗ − t σ t∗ − t
is often used in option pricing. Two approximations are provided for this
function. The first approximation has a precision of 10−3. If x ≥ 0, then
with coefficients
the formula is
1
N (x) = 1 − (1 + a1 x + a2 x2 + a3 x3 + a4 x4 )−4
2
if x < 0, then N (−x) = 1 − N (−x).
The second approximation has a precision of 10−7 .
If x ≥ 0, then with coefficients
the formula is
1 x2
N (x) = 1 − √ e 2 (b1 c + b2 c2 + b3 c3 + b4 c4 + b5 c5 ).
2π
If the function F and its derivatives are continuous in t in the interval [A, B]
and x takes all its values in the interval [a, b], then the derivative of this
integral is given by:
B
∂I(x) ∂F (x, t)
= dt + F (x, B)B (x) − F (x, A)A (x)
∂x A ∂x
If A and/or B are infinite, then this rule is applied only if the absolute
value of the derivative of F with respect to t is less than a certain . value
C(t) for all x in [a, b] and t in [A, B]. Also, the indefinite integral . C(t)dt
must be convergent in the interval [A, B].
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Expectations Hypotheses
The local expectations hypothesis (L-EH) of Cox et al. (1981) or the risk-
neutral expectations hypothesis shows that the current bond price is equal
to its expected value discounted at the current short-term rate.
Under the actual probability measure P , the bond price is given by:
RT
B(t, T ) = EP e− t ru du | Ft , ∀ t ∈ [0, T ]
1 RT
= EP e t ru du | Ft , ∀ t ∈ [0, T ],
B(t, T )
or
or in an equivalent form:
t t
rt = r0 + µu du + σu · dWu , ∀ t ∈ [0, T ]
0 0
Proposition: When the short rate follows an Ito process under P, then for
any martingale measure P ∗ = P α , the process r satisfies under P α :
The implications of the above results is that for any probability measure
P ∗ = P γ equivalent to P , the bond price can be defined by:
RT α
B(t, T ) = EP ∗ e− t ru du | FtW , ∀ t ∈ [0, T ]
Using Eq. (E.3), the bond price satisfies under the actual probability P :
This result indicates that the instantaneous returns from holding a bond
are in general different from the short-term rate by an additional term
reflecting the risk premium or the market price for risk.
Using the process for the short-term interest rate and a probablity
measure P ∗ the initial term structure is determined by the following
formula:
RT
B(0, T ) = EP ∗ e− 0 ru du , ∀ T ∈ [0, T ∗ ]
Exercises
Consider the following dynamics of a share price
Solution
First method: Application of Ito’s Lemma
∂f ∂f 1 ∂2f
df = dS + dt + (dS)2
∂S ∂t 2 ∂S 2
∂f ∂f ∂f 1 ∂2f 2
df = AdX + B dt + dt + A (dX)2
∂S ∂S ∂t 2 ∂S 2
with
or
Hence,
∂f ∂f ∂f 1 ∂ 2f
df = A dX + B + + A2 2 dt
∂S ∂S ∂t 2 ∂S
∂f ∂f 1 ∂ 2f
f (S + δS, t + δt) = f (S, t) + δS + δt + (δS)2
∂S ∂t 2 ∂S 2
∂ 2f 1 ∂2f
+ δSδt + · · · (dt)2 + · · ·
∂S∂t 2 ∂t2
∂f ∂f ∂f 1 ∂2f
δf = σSδX + µSδt + δt + (δS)2
∂S ∂S ∂t 2 ∂S 2
1 ∂2f 1 ∂ 2f
+ δtδS + (δt)2 + · · ·
2 ∂S∂t 2 ∂t2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
∂f ∂f 1 ∂ 2f 2 3
δf = (AδX + Bδt) + δt + A (δX)2 + O(δt 2 )
∂S ∂t 2 ∂S 2
3
where the terms of order O(δt 2 ) and smaller can be neglected.
When δt → 0, we can replace δX by dX and (δX)2 by dt to obtain the
following stochastic differential equation that must be satisfied by f (S, t):
∂f ∂f ∂f 1 2 ∂2f
df = A dX + B + + A dt
∂S ∂S ∂t 2 ∂S 2
2. The function g(S) with a zero drift, but non-zero variance must satisfy
the following equation:
∂g ∂g 1 2 ∂ 2g
dg = A dX + B + A dt
∂S ∂S 2 ∂S 2
with the restriction that g(S) has a zero drift and Var(g(S))
= 0 i.e.,
∂g 1 ∂ 2g
B + A2 2 = 0,
∂S 2 ∂S
or
∂2g 2B ∂g
+ 2 =0
∂S 2 A ∂S
with A2
= 0. It is possible to find a solution to the last equation.
Let us denote by
∂2g ∂g
= y et = y.
∂S 2 ∂S
Hence, we have
2B
y (S) + y = 0.
A2
dy 2B(S, t)
= 2 d(S, t),
y A (S, t)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Exercise
Consider the following dynamics for the stocks S1 and S2 :
dS1 = µ1 S1 dt + σ1 S1 dX1
dS2 = µ2 S2 dt + σ2 S2 dX2
Solution
The application of Taylor’s theorem over a short-time interval gives:
∂f ∂f 1 ∂ 2f
f (S1 , δS1 , S2 + δS2 ) = f (S) + δS1 + δS2 + (δS1 )2 2
∂S1 ∂S2 2 ∂S1
∂ 2f 1 ∂ 2f
+ δS1 δS2 + (δS2 )2 2 + · · ·
∂S1 ∂S2 2 ∂S2
When the terms in δS1 and δS2 are substituted in this equation and the
3
terms of O(δt 2 ) and smaller are neglected, we obtain:
∂f ∂f ∂f ∂f
δf = σ1 S1 δX1 + σ2 S2 δX2 + µ1 S1 δt + µ2 S2 δt
∂S1 ∂S2 ∂S1 ∂S2
1 ∂2f 1 ∂2f
+ (σ1 )2 (S1 )2 2 (δX1 )2 + (σ2 )2 (S2 )2 2 (δX2 )2
2 ∂S1 2 ∂S2
∂2f 3
+ σ1 σ2 S1 S2 δX1 δX2 + O(δt 2 )
∂S1 ∂S2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Exercise
Consider the following dynamics of the underlying asset:
dS = µSdt + σSdX.
f (S) = log(S n ) n ∈ N.
Solution
When Ito’s lemma is applied for a function f (S), we have:
∂f 1 ∂2f ∂f
df = dS + (dS)2 + dt
∂S 2 ∂S 2 ∂t
Since we have the following partial derivatives then:
∂f ∂f 1 ∂2f 1
= 0, =n , 2
= −n 2
∂t ∂S S ∂S S
(dS)2 = µ2 S 2 dt2 + 2µS 2 σdtdX + σ 2 S 2 (dX)2
(dt)2 = 0; dtdX = 0; (dX)2 = dt
Hence,
∂f ∂f 1 ∂ 2f
df = µS dt + σS dX + σ2 S 2 2 dt
∂S ∂S 2 ∂S
∂f ∂f 1 2 2 ∂ 2f
df = σS dX + µS + σ S dt
∂S ∂S 2 ∂S 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Replacing gives:
1 2
df = nσdX + n µ − σ dt.
2
Since this stochastic differential equation for log(S n ) has constant coeffi-
cients, S satisfies a log-normal random walk.
References
Arrow, KJ (1953). Le rôle des valeurs boursières pour la rèpartition la meilleur
des risques. International Colloquium on Econometrics, 1952, CNRS, Paris.
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13(3),
1895–1927.
Black, F (1976). The pricing of commodity contracts. Journal of Financial
Economics, 79(3), 167–179.
Bellalah, M and JL Prigent (2001). Pricing standard and exotic options in the
presence of a finite mixture of Gaussian Distributions. International Journal
of Finance, 13(3), 1975–2000.
Bellalah, M and F Selmi (2001). On the quadratic criteria for hedging under
transaction costs. International Journal of Finance, 13(3), 2001–2020.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
Bellalah, M, Ma Bellalah and R Portait (2001). The cost of capital in international
finance. International Journal of Finance, 13(3), 1958–1973.
Bellalah, M, JL Prigent and C Villa (2001a). Skew without Skewness: asymmetric
smiles; information costs and stochastic volatilities. International Journal of
Finance, 13(2), 1826–1837.
Cox, JC and SA Ross (1976). The valuation of options for alternative stochastic
processes. Journal of Financial Economics, 3, 145–66.
Cox, J, J Ingersoll and S Ross (1981). A reexamination of traditional hypothesis
about the term structure of interest rates. Journal of Finance, 36, 769–799.
Cox, J, J Ingersoll and S Ross (1985). A theory of the term structure of interest
rates. Econometrica 53, 385–407.
Debreu, G (1954). Representation of a preference ordering by a numerical
function. In Decision Processes, R Thrall, C Coombs and R Davis (eds.),
pp. 159–165, New York: Wiley.
El Karouin N and H Geman (1993). A probabilistic approach to the valuation of
general floating rate notes with an application to interest swaps. Unpublished
working paper.
Harrison, JM and D Kreps (1979). Martingales and arbitrage in multiperiod
securities markets. Journal of Economic Theory, 20, 381–408.
Harrison, JM and S Pliska (1981). Martingales and stochastic integrals in the
theory of continuous trading. Stochastic Processes and their applications, 11,
215–260.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch12
Chapter 13
Chapter Outline
This chapter is organized as follows:
Introduction
This chapter develops a general context for the analysis and valuation
of options and futures contracts in the presence of one or several state
variables and information uncertainty. We provide a simple derivation of the
differential equation for a derivative security on a spot asset in the presence
of a continuous dividend yield and information costs. Then, we extend this
583
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
analysis to account for several state variables. This allows the derivation
of the valuation equation for securities dependent on several variables in
the presence of incomplete information. When a variable does not indicate
the price of a traded security, its market price of risk corresponds to the
market price of risk of a traded security, whose price is a function only on
the value of the variable and time. The value of the market price of risk
of the variable is the same at each instant of time. We also show, how to
extend the risk-neutral argument in the presence of information costs and
how to apply this analysis for the valuation of commodity futures within
incomplete information.
dS = µSdt + σSdz
where the drift term µ and the volatility σ are constants. Using Ito’s lemma
for the function f (S, t) gives:
∂V dV 1 ∂ 2V 2 2 ∂V
dV = µS + + σ S dt + σSdz.
dS dt 2 ∂S 2 ∂S
∂V
Π = −V + S.
dS
Over a short time interval, the change in the portfolio value can be
written as:
∂V 1 ∂ 2V 2 2
∆Π = − − σ S ∆t.
∂t 2 ∂S 2
µ − r − λ = γσ. (13.3)
The excess return over the risk-free rate in the presence of shadow costs on
a security corresponds to its market price of risk multiplied by its volatility.
When γ > 0, the expected return on an asset is higher than the risk-free
rate plus information costs.
When γ = 0, the expected return on an asset is exactly the risk free
rate plus information costs.
When γ < 0, the expected return on an asset is less than the risk free
rate plus information costs.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
When a variable does not indicate the price of a traded security, its
market price of risk corresponds to the market price of risk of a traded
security whose price is a function only on the value of the variable and
time. The value of the market price of risk of the variable is the same at
each instant of time. In fact, we can show as shown by Hull (for details,
refer to Bellalah et al., 1998) that two traded securities depending on the
same asset must have the same price of risk, i.e., that Eq. (13.2) must be
verified. Consider the following dynamics for a variable θ, which is not a
tradable asset:
dθ
= µ(θ, t)dt + s(θ, t)dz.
θ
We denote this by V1 and V2 , respectively the prices of two derivative
securities as a function of θ and t. The dynamics of these derivatives can
be written as:
dV1
= µ1 dt + σ1 dz (13.4)
V1
dV2
= µ1 dt + σ2 dz. (13.5)
V2
These two processes can be written in discrete time as:
∆V1 = µ1 V1 ∆t + σ1 V1 ∆z (13.6)
∆V2 = µ2 V2 ∆t + σ2 V2 ∆z. (13.7)
Π = σ2 V2 V1 − σ1 V1 V2 .
∆Π = σ2 V2 ∆V1 − σ1 V1 ∆V2 .
Using Eqs. (13.6) and (13.7), the change in the portfolio value can be
written as:
∆Π = µ1 σ2 V1 V2 − µ2 σ1 V1 V2 ∆t.
Since the portfolio Π is instantaneously risk less, it must earn the risk free
rate plus information costs on both markets. Hence, we must have:
µ1 σ2 − µ2 σ1 = (r + λ1 )σ2 − (r + λ2 )σ1
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
or
µ1 − (r + λ1 ) µ2 − (r + λ2 )
= . (13.8)
σ1 σ2
θi = mi θi dt + si θi dzi
where dzi are Wiener processes. The terms mi and si correspond to the
expected growth rate and the volatility of the θi with (i = 1, . . . , n). The
price process for a derivative security that depends on the variables θi can
be written as:
n
dV
= µdt + σi dzi (13.9)
V i=1
Equation (13.10) shows that the expected excess return on the security
(option) in the presence of shadow costs depends on γi and σi .
When γi σi > 0, a higher return is required by investors to get
compensated for the risk arising from the variable θi .
When γi σi < 0, a lower return is required by investors to get
compensated for the risk arising from θi .
where ρi, k stands for the correlation coefficient between the variables θi
and θk .
We show in Appendix A, how to obtain a similar equation in the
presence of incomplete information. In this context, the equation becomes:
∂V ∂V 1 ∂2V
+ θi (mi −γi si )+ ρi,k si sk θi θk = (r+λ)V. (13.11)
∂t i
∂θi 2 ∂θi ∂θk
i,k
q + m − r − λ = γs or m − γs = r + λ − q.
mi − r − λi = γi si or mi − γi si = r + λi .
This result shows that a change in the expected growth rate of the state
variable from θi to (mi − γi si) is equivalent to using an expected return
from the security equal to the risk-less rate plus shadow costs of incomplete
information.
For the case of a dividend-paying traded asset, we have
qi + mi − r − λi = γi si or mi − γi si = r + λi − qi .
This result shows that a change in the expected growth rate of the state
variable from θi to (mi − γi si) is equivalent to using an expected return
(including continuous dividends at a rate q) from the security equal to the
risk-less rate plus shadow costs of incomplete information. This analysis
allows the pricing of any derivative security as the value of its expected
payoff discounted to the present at the risk-free rate plus the information
cost on this security or:
where the drift rate in θi corresponds to (mi − γi si). When the interest rate
r is stochastic, it is considered as the other underlying state variables.
In this case, the drift rate in r becomes γr sr where γr refers to the
market price of risk related to r. The term sr indicates its volatility. In this
case, the pricing of a derivative is given by the discounting of its terminal
payoff at the average value of r as:
∂F
σF F = σS = σSeα(T −t) = σF.
∂S
Hence, the volatility of the futures price is equal to the volatility of the spot
price and σF = σ. Now, consider the following dynamics for the futures
price:
dF = µF F dt + σF dz.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
Since the derivative asset price f (F, t) is a function of the futures price F
and time t, using Ito’s lemma gives:
∂f ∂f 1 ∂2f 2 2 ∂f
f= µF F + + σ F dt + σF dz.
∂F ∂t 2 ∂F 2 ∂F
∆F = µF F ∆t + σF ∆z
and
∂f ∂f 1 ∂ 2f 2 2 ∂f
∆f = − µF F + + 2
σ F ∆t + σF ∆z
∂F ∂t 2 ∂F ∂F
√
where ∆z = ∆t.
Hence, we have:
∂f 1 ∂2f 2 2
∆W = − − σ F ∆t.
∂t 2 ∂F 2
Since this change in value is risk free, it must earn the risk free rate plus
information costs or:
∂f 1 ∂2f 2 2
− − σ F ∆t = −(r + λf )f ∆t
∂t 2 ∂F 2
∂f 1 ∂ 2f 2 2
+ σ F = (r + λf )f.
∂t 2 ∂F 2
This equation must be satisfied by a derivative security dependent on
a futures price in the presence of information uncertainty.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
F = Ê[ST ]. (13.16)
Equation (13.16) shows that the futures price corresponds to the expected
spot price in a risk-neutral world. If (γσ) is constant and the drift µ is a
function of time then:
µ − γσ = r + λ + g − y. (13.18)
y minus the storage costs u. Hence, the convenience yield must satisfy the
relationship:
y = g + r + λ − µ + γσ.
µ − γσ = r + λ + g.
Summary
This chapter develops a general context for the pricing of derivative assets.
First, we derive the differential equation for a derivative security on a
spot asset in the presence of a continuous dividend yield and information
costs.
Second, we provide the valuation of securities dependent on several
variables in the presence of incomplete information.
Third, we propose the general differential equation in the same context.
Fourth, we show how to extend the risk-neutral argument in the
presence of information costs.
Fifth, we extend the analysis to the valuation of commodity futures
contracts within incomplete information.
Questions
1. Provide a simple derivation of the differential equation for a derivative
security on a spot asset in the presence of a continuous dividend yield
and information costs.
2. Describe the valuation of securities dependent on several variables in the
presence of incomplete information.
3. How can one extend the risk-neutral argument in the presence of
information costs?
4. How can one price commodity futures contracts within incomplete
information?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
θi = mi θi dt + si θi dzi
where the growth rate mi and the volatility si can be functions of any of
the n variables and time.
Let us denote respectively by,
where
∂Vj ∂Vj 1 ∂ 2f
µj Vj = + m i θi + ρi,k si sj θi θj ∂θk (A.2)
∂t i
∂θi 2 ∂θi
i,k
∂fj
σij fj = si θi . (A.3)
∂θi
In this context, it is possible to construct a portfolio using the (n+1) traded
securities. We denote by aj , the amount of the jth security in the portfolio
Π so that Π = j aj Vj , where the aj is chosen in a way to eliminate the
stochastic components of the returns. Using Eq. (A.1), we have:
aj σij fj = 0 (A.4)
j
for i between 1 and n. The instantaneous return from this portfolio can be
written as dΠ = j aj µj Vj dt, where the cost of constructing this portfolio
is j aj Vj .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
If this portfolio is riskless, it must earn the risk-less rate plus informa-
tion costs corresponding to each asset in the portfolio:
aj µj fj = aj fj (r + λj ) (A.5)
j j
or
µj − r − λj = γi σij (A.8)
i
∂Vj ∂Vj 1 ∂ 2 Vj
+ θi (mi − γi si ) + ρik si sk θi θk = (r + λj )Vj .
∂t i
∂θi 2 ∂θi ∂θk
i,k
∂V ∂Vj 1 ∂2V
+ θi (mi − γi si ) + ρik si sk θi θk = (r + λ)V (A.9)
∂t ∂θi 2 ∂θi ∂θk
i i,k
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
q̂ = r + λi − mi + γi si .
The values of θˆi and θi must be equal and the following differential equation
must be verified:
∂V ∂V 1 ∂ 2V
+ θˆi (r + λi − q̂i ) + ρi,k si sk θˆi θˆk = (r + λ)V.
∂t ∂ θˆi 2 ∂ θˆi ∂ θˆk
i i,k
(r + λi − q̃) = mi − γi si ,
Exercises
Exercise 1
1) Can you verify that the following terms are solutions of the extended
Black–Scholes equation derived by Bellalah (1999)?
1.a) V (S, t) = S
1.b) V (S, t) = e(r+λv )t
where r refers to the risk-less interest rate and λv corresponds to the
information cost regarding the security V .
Solution
Re-call that the extended Black–Scholes equation in the presence of shadow
costs of incomplete information can be written as:
∂V 1 ∂2V ∂V
+ σ2 S 2 + (r + λs )S − (r + λv )V = 0
∂t 2 ∂S 2 ∂S
1.a) We can compute the partial derivatives with respect to S and replace
in the PDE:
∂V ∂V ∂ 2V
= 0, = 1, = 0.
∂t ∂S ∂S 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
(r + λs )S − (r + λv )V = 0
or
1.b) For the second function, the calculations of the partial derivatives gives
∂V ∂V ∂ 2V
= (r + λv )e(r+λv )t , =0 and = 0.
∂t ∂S ∂S 2
(r + λv )e(r+λv )t − (r + λv )e(r+λv )t = 0
Solution
1) When V = A(S) is replaced in the extended Black–Scholes equation,
this gives:
1 2 2 ∂ 2A ∂A
σ S + (r + λs )S − (r + λA )A = 0. (13.19)
2 ∂S 2 ∂S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
b) When we have
1 2 1
σ = 0, r + λA = 0, r + λs − σ2 = 0.
2 2
In this case, we compute the ∆ of Eq. (13.20) is defined by:
∆ = b2 − 4ac.
Hence,
2
1 1
∆= r + λs − σ2 + 4 σ2 (r + λA )
2 2
or
1
∆ = (r + λs )2 + σ4 − σ2 (r + λs ) + 2σ2 r + 2σ2 λa
4
which is equivalent to
1 4
∆= σ + r2 + 2rλs + λ2s − σ 2 r − λs σ 2 + 2σ 2 r + 2σ 2 λA
4
or
1 4
∆= σ + r2 + λ2s + σ2 r + 2rλs + 2σ 2 λA − λS σ 2
4
or
1
∆ = r2 + (2λs + σ 2 )r + 2σ 2 λA + λ2s + σ 4 − λs σ 2 .
4
So, we have
2
2 σ2 2 σ2
∆ = r + 2 λs + r + 2λA σ + λs − (13.21)
2 2
b.1) When
λs = λA ⇐⇒ ∆ = 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
∀ r ∈ −{r1 = r2 }, ∆>0
V (S, t) = HS n1 + KS n2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
b.1.2) When
σ2
r = − λs + ; ∆ = 0.
2
−r − λs + 12 σ 2 −r + λs 1
n1 = n2 = n = = +
σ2 σ2 2
as
1
r = −λs + σ2
2
Hence,
1 1
n= + =1
2 2
b.2) It is the case when λA > λs , ∆ < 0 and Eq. (13.21) does not have
a solution and ∆ > 0, since the equation has the same sign as r2 . In this
case, we have:
√
−(r + λs ) + 12 σ 2 + ∆
m1 = ,
σ2
√
−(r + λs ) + 12 σ 2 − ∆
m2 = and V (S, t) = LS m1 + GS m2
σ2
b.3) When
λA < λs ; ∆ > 0.
V (S, t) = NS α1 + MS α2
−(r + λs − 12 σ 2 )
β=
σ2
or
σ2
− −λs − 2
− |σ| 2(λs − λA ) + λs − 12 σ 2
β=
σ2
which is equivalent to:
−σ 2 − |σ| 2(λs − λA ) 2(λs − λA )
β=− =1+
σ2 |σ|
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
and finally:
V (S, t) = ES β
where E is a constant.
b.3.3) When
σ2
r2 = −λs − + |σ| 2(λs − λA ), ∆=0
2
Hence:
−2(r + λs ) + σ 2 2(λs − λA )
n = 0 or n = =− +1
σ2 σ2
2(λs −λA )
V (S, t) = C + DS − σ2
+1
∂C 1 ∂ 2B ∂B
B + σ2 S 2 C + (r + λs )SC − (r + λv )B(S)C(t) = 0.
∂t 2 ∂S 2 ∂S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
∂V ∂C ∂V ∂B ∂2V ∂ 2B
= B(S) , = C(t) , = C(t) .
∂t ∂t ∂S ∂S ∂S 2 ∂S 2
This gives
1 ∂C 1 1 2 2 ∂2B ∂B
=− σ S + (r + λs )S − (r + λv )B(S) .
C ∂t B 2 ∂S 2 ∂S
and finally
1 2 2 1
σ n + r + λs − σ 2 n − (r + λv − k) = 0 (13.22)
2 2
First case
If
1 1 2 1 2
r + λs − σ2 = 0; r= σ − λs ; σ = 0, r + λv − k = 0,
2 2 2
then
1
2 2(r + λv − k) 2 2
σ 2 − λs + λv − k
n = = .
σ2 σ2
Since
2 12 σ 2 + λv − λs − k
n1 = −n2 , n1 =
|σ|
then we have,
Second case
If
1 2 1
r − k − λv = 0, r = k − λv , σ = 0 and r + λs − σ2 = 0,
2 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
Third case
If
1 σ2
r + λs − σ2 = 0, k − r − λv = 0, = 0
2 2
then
2
1 1
∆ = r2 + 2 λs + σ2 r + λs σ2 + 2σ 2 (λv − k)
2 2
∆ = 2σ 2 (λs − λv − k).
First situation
If
∆ < 0; λs − λv − k < 0;
∆ has no roots and when ∆ > 0, Eq. (13.20) has two solutions n1 and n2
where:
√ √
−(r + λs − 12 σ 2 ) − ∆ −(r + λs − 12 σ 2 ) + ∆
n1 = , n2 = .
σ2 σ2
The solution is given by:
Second situation
If
∆ = 0 ⇐⇒ λs = λv + k ⇐⇒ λs − λv = k,
σ2
−(r + λs − 2 )
m1 = = 1.
σ2
Third situation
∆ > 0 ⇐⇒ λs > λv + k.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
and
σ2
√ σ2
√
−((r + λs ) − 2 ) + ∆ −((r + λs ) − 2 ) − ∆
α= , β= .
σ2 σ2
The solution is given by
V (S, t) = ekt (M S α1 )
where M is a constant.
When r = a2 , the ∆ = 0 and the solution is:
2(λs − λv − k)
γ = 1− .
|σ|
V (S, t) = NS γ ekt
where N is a constant.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
Exercise 3
When a change of time variable is used, the equation can be reduced to the
∂2 u
diffusion equation, C(τ ) ∂u
∂τ = ∂x2 , when C(τ ) > 0.
Consider the extended Black–Scholes equation with a constant volatil-
ity and interest rate. Can this equation be reduced to the diffusion equation
in this case?
Solution
We can try a change of variable of the form u(x, τ ) = v(x, τ̃ ), where
τ̃ = F (τ ) corresponds to a certain function of τ . In this case, we have:
∂u ∂u dF (τ )
= .
∂τ ∂ τ̃ dτ
The partial differential equation becomes
dF (τ ) ∂v(x; τ̃ ) ∂ 2 v(x, τ̃ )
C(τ ) = .
dτ ∂ τ̃ ∂x2
If you choose the function F (τ ) such that:
dF (τ ) dS
C(τ ) = 1, where F (τ ) =
dτ C(S)
∂v ∂ 2v
= .
∂ τ̃ ∂x2
Now, it is possible to solve the extended Black–Scholes equation:
∂V 1 ∂2V 2 2 ∂V
+ σ S + (r(t) + λs (t))S − (r(t) + λv (t))V = 0.
∂t 2 ∂S 2 ∂S
We use the following transformation
∂V ∂v ∂V ∂v ∂x E ∂v ∂2V E ∂v E ∂ 2v
=E , = = , 2
=− 2 + 2 2.
∂t ∂t ∂S ∂x ∂S S ∂x ∂S S ∂x S ∂x
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
This leads to
∂v 1 2 ∂ 2v ∂v ∂v
+ σ (t) 2
− + (r(t) + λs (t)) − (r(t) + λv (t))v = 0
∂t 2 ∂x ∂x ∂x
⇐⇒
2 ∂v ∂ 2v 2 ∂v 2
+ + (r(t) + λs (t) − 1) 2 − (r(t) + λv (t))V = 0.
σ 2 (t) ∂t ∂x2 σ (t) ∂x σ 2 (t)
2 ∂v(x, t) ∂
= v(X, τ )
σ 2 (t) ∂t ∂τ
i.e.,
t
1
τ =− σ 2 (s)ds.
2 0
If we set
and
2
1 2 2
β= (r(t) + λs (t) + 1 + 2 (λs (t) + λv (t))
4 σ2 σ
we obtain,
∂w ∂2w
= .
∂τ ∂x2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch13
Exercise 4
For the following problem:
∂u ∂ 2u
= ; −∞ < x < ∞, τ > 0 with u(x, 0) = u0 (x) > 0
∂τ ∂x2
u(x, τ ) > 0 ∀ τ.
This result can be used to show that an option will always have a positive
value.
Solution
The solution to this general problem:
∂u ∂ 2u
= ; −∞ < x < ∞, τ > 0
∂τ ∂x2
with
is
∞
1 −(x2 −s)2
u(x, τ ) = √ u0 (S)e 4τ ds if u0 (x) > 0,
2 Πτ −∞
then
−(x2 −s)2
u0 (s)e 4τ > 0 and u(x, τ ) > 0.
∂V 1 ∂ 2V 2 2 ∂V
+ 2
S σ + (r + λs )S − (r + λv )V = 0
∂t 2 ∂S ∂S
References
Bellalah, M (1999). The valuation of futures and commodity options with
information costs. Journal of Futures Markets, 19 (September) 645–664.
Briys, E, M Bellalah et al. (1998). Options, Futures and Exotic Derivatives. En
collaboration avec E. Briys, et al., John Wiley & Sons.
Cox, JC, JE Ingersoll and SA Ross (1985). A theory of the term structure of
interest rates. Econometrica, 53, 385–407.
Garman, M (1976). A general theory of asset valuation under diffusion state
processes, Working Paper, No. 50, Berkeley: University of California.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Part V
613
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
614
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Chapter 14
Chapter Outline
This chapter is organized as follows:
1. Section 14.1 is an introduction to the general context for the pricing
of American options with and without distributions to the underlying
asset.
2. Section 14.2 studies the valuation of American spot and futures options
in the context of a constant proportional rate.
3. Section 14.3 deals with the valuation of American spot and futures
options in the context of a constant proportional rate within incomplete
information.
4. Section 14.4 studies the valuation of American options when there are
discrete distributions to the underlying asset.
5. Section 14.5 deals with the valuation of American options when there
are discrete distributions to the underlying asset within incomplete
information.
6. Section 14.6 is devoted to the valuation of compound options within
incomplete information.
7. Appendix A presents an alternative derivation of the compound option’s
formula using the martingale approach.
615
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Introduction
There are several extensions of the basic Black and Scholes (1973) model.
We have already seen the first extension by Black (1976) who takes into
account the specificities of futures contracts. The second extension was
made by Garman and Kohlhagen (1983) who derive analytical valuation
formulas for European options on currencies. In the same year, Grabbe
(1983) implemented a similar approach to that used in Black (1976) to
provide the value of options on foreign currencies. Merton (1973) indirectly
and Barone-Adesi and Whaley (1987), hereafter Barone-Adesi and Whaley
(1987) provided the values of European commodity options and commodity
futures options. All these models apply only to European options. Since,
all the proposed analytical models deal with the pricing of the European
options in the absence of discrete distributions to the underlying assets, the
extensions of the analytical models to the valuation of American options
are proposed in this chapter.
The two important extensions and contributions to the literature on the
valuation of American options are those of Merton (1973) and Geske (1979).
The pricing of American options is first analyzed by Merton (1973) who
showed the difficulties in obtaining closed-form solutions when there are
discrete distributions to the underlying asset. However, he provided closed-
form solutions for American options when the time to maturity is infinite.
Geske (1979) provided analytical formulas for the valuation of options on
options or compound options. He used a valuation by duplication technique
for the pricing of American options. A compound option can be defined as
an option on the firm’s equity. For a levered equity firm (a firm with debt in
its capital structure), equity can be seen as a call on the value of the assets.
This was first noted by Black and Scholes (1973); B–S, who considered
the option on equity as an option on the value of the firm’s asset. Geske’s
approach is interesting since it allows the valuation of American options
and includes the question of dividend.
American option pricing models have been proposed by several authors.
However, given the difficulties in obtaining closed-form solutions, it became
quite a natural way to resort to analytical approximation models, binomial
methods, and numerical techniques. When there are continuous distribu-
tions to the underlying asset, the literature on the valuation of American
options provides some analytical approximation formulas. Barone-Adesi
and Whaley (1987) presented simple analytic approximations for the pricing
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 617
Extension of Asset and Risk Management in the Presence of American Options 619
C(0, τ, K) = 0 (14.4)
C(S, 0, K) = max[0, S − K] (14.5)
C(S, τ, K) ≥ max[0, S − K] (14.6)
Equation (14.4) shows that the option is worthless when the underlying
asset is zero for any time to maturity τ . Equation (14.5) indicates that
the option price is equal to the greater of zero and its intrinsic value at the
option maturity date. Equation (14.6) is an arbitrage condition. It indicates
the American option value at any time during the option’s life. It shows
also that at each instant τ , there is a positive probability of early exercise.
This implies that there exists a certain level I(τ ) of the underlying asset
price such that for each S > I(τ ), the option is worth more “dead” than
in “life”. Since the value of an immediate exercise is (S − K), the structure
of the problem implies the additional condition C(I(τ ), τ, K) = I(τ ) −
K = h, where C(I(τ ), τ, K) satisfies the partial differential equation for
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 621
P (∞, τ, K) = 0 (14.7)
P (S, 0, K) = max[0, K − S] (14.8)
P (S, τ, K) ≥ max[0, K − S] (14.9)
Equation (14.7) shows that the put is worthless when the underlying asset
price tends to infinity. This result is rather intuitive. In fact, the put intrinsic
value is given by the difference between a constant (the strike price) and
the infinite underlying asset price. Equation (14.8) is standard and gives
the put value at the option’s maturity date. Equation (14.9) shows that
it is sometimes optimal to exercise the put before its maturity date. This
happens when the underlying asset price tends to zero. In this situation,
it is interesting to exercise the put as soon as possible to benefit from the
investment of the strike price until the option’s maturity date. From the
analyses of McKean (1969), Samuelson (1972), and Merton (1973), there
is no-closed form solution for a finite-life put when the above boundary
conditions are applied to the partial differential equation. However, for an
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 623
• the put ceases to depend on time when the critical stock price is reached
and
• the result that an American put corresponds to the value of a European
put plus the early exercise premium, Bunch and Johnson (2000) provide
a formula for the critical price.
where T is the maturity date and K is the strike price. The first factor in the
integral corresponds to a discount factor. The second factor corresponds to
the payoff among exercise. The third factor f is the first-passage probability,
i.e., the probability that the stock price declines from its value S to the
critical value Sc for the first time t. The maximization concerns all possible
functions Sc (τ ) where τ = T −t. For values of the underlying asset less than
the critical price, the put value is simply its intrinsic value. For an infinite
time to maturity, the critical asset price is constant and can be determined
using the first-order condition. The first-passage probability is provided by
Feller (1971).
First, define:
1 S 1 2
Z= log + r− σ t
σ St 2
where σ corresponds to the volatility of the underlying asset.
Second, define:
1 S 1
a= log + r − σ2 t
σ Sc 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
This term can be seen as a standardized measure of how the put is from
early exercise. When γ = σ2r2 = 1, so that a is constant, Eq. (14.10) requires
an expression for the first-passage time of a standardized Brownian motion.
Feller (1971) shows that the following density function must be used:
1
f (x)d(x) = √ e−1/(2x) dx.
2πx3
The American put value can be computed using the appropriate density
obtained via the transformation x = at2 so that
a 2
f (t)d(t) = √ e−a /(2t)
dt
2πt3
in Eq. (14.10). It is possible to choose the constant in this last equation to
normalize f (t), so the integral of f (t) from zero to infinity is one. Since the
critical price is constant in this case, we have:
∞ 2
K − Sc ae−rt e−a /(2t)
P = √ dt.
2π 0 t3/2
A similar result is obtained by Merton (1973). In fact, the Black and Scholes
(1973) partial differential equation,
∂P ∂P 1 ∂2P
= rP − rS − σ2S 2 2 (14.12)
∂t ∂S 2 ∂S
simplifies to an ordinary equation for an infinite maturity.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 625
∂P 1 ∂ 2P
rP − rS − σ2 S 2 2 (14.13)
∂S 2 ∂S
we can show that Eq. (14.11) satisfies Eq. (14.13). Equation (14.11) can be
written as:
2
Sc Sc
P =K −S . (14.14)
S S
where,
S
log Sc + (r − 12 σ 2 )t
d2 (S, Sc , t) = √ .
σ t
The second term corresponds to the early exercise premium. Bunch and
Johnson (2000) shows that:
Sc 2 √
= e−(r+(1/2)σ )τ −gσ τ (14.17)
K
where,
σ2 K
g=± 2 log 2r where, x = . (14.18)
√
α
x log xe−α(r+(1/2)σ2 )2 τ /(2σ2 ) Sc
The function g has typically a value about 1.5. Equation (14.17) has the
appropriate asymptote for very large τ when,
1 1+γ 1
g = √ log − r + σ2 τ . (14.19)
τ γ 2
It is important to know, when g = 0 and the corresponding time τ0 by
setting g = 0 in Eq. (14.19):
log 1+γ
γ
τ0 ≈ . (14.20)
r + 12 σ 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 627
The analysis in Bunch and Johnson (2000) shows that for typical puts, the
values of g are between one and two. Equation (14.21) shows that when
r = 0, there is no reason to early exercise, so Sc should be zero. In fact,
when the interest rate is zero in Eq. (14.21), g tends to infinity and the
critical stock price is worthless.
This equation applies to American options, European options, and the early
exercise premium. This premium is given by the difference in value between
the American and European option values, i.e., c = C(S, T ) − c(S, T ) and
p = P (S, T ) − p(S, T ). Using this notation, Eq. (14.22) can be re-written
for the early exercise premium as:
2r 2b
Let M = σ2 , N = σ2 , τ = T − t. Multiplying Eq. (14.23) by ( σ22 ) gives:
Extension of Asset and Risk Management in the Presence of American Options 629
Now, define the early exercise premium as (S, k) = k(τ )f (S, k). Hence,
∂ 2 (S, t) ∂2f ∂(S, t) ∂f
=k and = kτ f + kkτ .
∂ 2S ∂ 2S ∂τ ∂k
with
S∗
A2 = 1 − e(b−r)T N (d1 (S ∗ ))
q2
1 M
q2 = −(N − 1) + (N − 1)2 + 4
2 k
2r 2b
N= , M= , k = 1 − e−rT .
σ2 σ2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
In this formula, S ∗ stands for the critical commodity price, which can be
determined iteratively using the following system:
S ∗ [1 − e(b−r)T N (d1 (S ∗ )]
S ∗ − K = c(S ∗ , T ) + .
q2
The value of S ∗ corresponds to the value of S above which the call’s value
is equal to its exercisable proceeds, (S − K). This value can be determined
by a Newton–Raphson procedure or using the efficient algorithm presented
by Barone-Adesi and Whaley (1987). When b is less than r, the American
call value is given by the above formula. Otherwise, when b is greater or
equal to r, C(S, T ) = c(S, T ) since the call will never be exercised early.
In the same context, the American commodity option value P (S, T ) is
given by:
q1
S
P (S, T ) = p(S, T ) + A1 when S > S ∗
S∗
P (S, T ) = K − S when S ≤ S ∗
with,
−S ∗
A1 = 1 − e(b−r)T N (−d1 (S ∗ ))
q1
1 M
q1 = −(N − 1) − (N − 1) + 4 2 ,
2 k
2r 2b
N= , M= , k = 1 − e−rT .
σ2 σ2
The critical underlying commodity price is given by an iterative procedure
from the following equation
Extension of Asset and Risk Management in the Presence of American Options 631
S c C S∗
S c C S∗
S p P S∗
S p P S∗
with
F∗
A2 = 1 − e−rT N (d1 (F ∗ ))
q2
1 M 2r
q2 = 1+ 4 , M = 2 , k = 1 − e−rT .
2 k σ
with,
F∗
A2 = 1 − e(b−r)T N (d1 (F ∗ ))
q2
1 M
q2 = −(N − 1) + (N − 1)2 + 4
2 k
2r 2b
N= , M= , k = 1 − e−rT .
σ2 σ2
The price of an American call futures option C(F, T ), is equal to the price
of a European futures option c(F, T ), plus a term corresponding to the
probability of early exercise A2 (F/F ∗ )q2 . The critical futures price, F ∗ , is
calculated using the following equation:
∗ ∗ F ∗ 1 − e(b−r)T N (d1 (F ∗ )
F − K = c(F , T ) + .
q2
When the futures price is below F ∗ , the American option price is given
by the European price plus the early exercise premium. When the futures
price is above F ∗ , the American option price is given by the intrinsic value,
(F − K).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 633
In the absence of carrying costs, the formula for the American futures
put is:
with,
−F ∗∗
A1 = 1 − erT N (−d1 (F ∗ ))
q1
1 M 2r
q1 = 1− 4 , M = 2 , k = 1 − e−rT .
2 k σ
When there are carrying costs, the formula for the American put is:
with,
−F ∗
A1 = 1 − e(b−r)T N (−d1 (F ∗ ))
q1
1 M
q1 = −(N − 1) − (N − 1) + 4 2 ,
2 k
2r 2b
N= , M= , k = 1 − e−rT .
σ2 σ2
When the futures price is above the critical price, the American futures
option price is given by the sum of the European price and the early exercise
premium A1 (F/F ∗ )q1 . When the futures price is above the critical futures
price, the American futures option value is equal to its intrinsic value,
(K − F ).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
S N (d1 ) N (d2 ) c C
S N (−d1 ) N (−d2 ) p P
Extension of Asset and Risk Management in the Presence of American Options 635
futures contracts and the risk-less asset can be constructed to duplicate the
payoff of the futures option. The absence of costless arbitrage oppotunities
implies the following relationship between the futures or the forward price
and its underlying asset: F = Se(b+λS )T where F is the current forward
price, T is the option’s maturity date, b is the constant proportional cost of
carrying the commodity, and λS is the information cost on the spot asset.
When a hedged position is constructed and “continuously” re-balanced,
using limiting arguments as in Omberg (1991), yields:
1 2 2
σ S CSS + (b + λS )SCS − (r + λC )C + Ct = 0.
2
When λS and λC are set equal to zero, this equation collapses to that in
Barone-Adesi and Whaley (1987).
Let T be the maturity date of the call and K be its strike price. This
equation must be solved under the call boundary condition at maturity.
The value of a European commodity call by Bellalah (1999) is:
with
S 1 2 √ √
d1 = ln + b + σ + λS T σ T, d2 = d1 − σ T
K 2
and where N (·) is the cumulative normal density function.
When λS and λC are equal to zero and b = r, this formula is the
same as that in Black and Scholes.1 If besides, the cost of carrying the
or
ˆ ˜
R̄S − r − λS = aβS with a = R̄m − r − λm .
This equation can be written for the expected return on the spot asset and the option as:
„ «
∆S
E = (r + λS )∆t + aβS ∆t
S
„ «
∆C
E = (r + λC )∆t + aβC ∆t
C
Multiplying this last equation by C and substituting for βC gives:
with
F 1 √ √
d1 = ln + σ2 T σ T, d2 = d1 − σ T .
K 2
The solution for a European futures put option in the same context is:
The term FN (d1 ) − KN (d2 ) shows that the expected value of the futures
call at expiration, is the expected difference between the futures price
and the strike price conditional upon the option being in-the-money times
the probability that it will be in-the-money. The term e−(r+λC )T is the
appropriate discount factor within a framework of incomplete information
by which the expected expiration value is brought to the present. The
following equation (with information costs) is the analogous of that as given
by Black (1976)2 :
1
CSS S 2 σ 2 − (r + λC )C + Ct = 0.
2
Extension of Asset and Risk Management in the Presence of American Options 637
with
S∗
A2 = (1 − e(b+λS −r−λC )T N (d1 (S ∗ )))
q2
1 M
q2 = −(N − 1) + (N − 1)2 + 4
2 k
2(r + λC ) 2(b + λS )
N= , M= , k = 1 − e−(r+λC )T .
σ2 σ2
becomes:
1
E(∆C) = Ct ∆t + CSS S 2 σ2 ∆t
2
E(∆C) = (r + λC )C∆t,
which gives:
1
CSS S 2 σ 2 − (r + λC )C + Ct = 0.
2
If λC = 0, this equation is the Black equation with information costs.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
with:
−S ∗∗
A1 = (1 − e(b+λS −r−λC )T N (−d1 (S ∗∗ )))
q1
1 M 2(r + λC )
q1 = −(N − 1) − (N − 1) + 4 2 , N= ,
2 k σ2
2(b + λS )
M= , k = 1 − e−(r+λC )T .
σ2
3 The above solutions are written as in Barone-Adesi and Whaley (1987). They can be
re-written as a function of the futures price as given by Whaley (1986) using the cost-of
carry model: the relationship between the futures price and the spot price.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 639
Table 14.7. Simulations of Futures call (put) option values using Black’s model, CBlack ,
(PBlack ) and the proposed model, CM , (PM ) for the following parameters. K = 100,
T = 0.25, σ = 0.2, r = 0.08, and b = 0.
Panel A, F CBlack CM,λC =0.01 CM,λC =0.05 PBlack PM,λC =0.01 PM,λC =0.05
Table 14.8. Comparisons of European and American Futures call option values
using the proposed models, CEuropean , CAmerican , CBAW for the following
parameters. K = 100, T = 0.25, σ = 0.2, r = 0.08, b = 0, and λC = 0.01.
Extension of Asset and Risk Management in the Presence of American Options 641
Table 14.9. Comparisons of European and American Futures put option values
using the proposed models, PEuropean , PAmerican , PBAW for the following
parameters. K = 100, T = 0.25, σ = 0.2, r = 0.08, b = 0, and λC = 0.01.
Extension of Asset and Risk Management in the Presence of American Options 643
where P is the stock price cum-dividend. The price dynamics of the stock
S are given by the stochastic differential equation:
(a) The purchase of a European call having a strike price K and a maturity
date T ;
(b) The purchase of a European call with a strike price St∗ and a maturity
date (t − ) and
(c) The sale of a European call option on the option defined in (a) with a
strike price (St∗ + D − K) and a maturity date (t − ).
c(St∗ , T − t, K) = St∗ + D − K.
The option described in (c/) can be priced using the compound option
formula proposed in Geske (1979). Its value is given by:
(y−r)T t
cc = Se N2 a1 , b1 , − Ke−rT N2
T
ti
× a2 , b2 , − (St∗ + D − K)e−rtN1 (b2 )
T
where,
1 2 √ √
a1 = ln(S/K) + y + σ T σ T, a2 = a 1 − σ T
2
1 √ √
b1 = ln(S/St∗ ) + y + σ2 t σ t, b2 = b1 − σ t.
2
Extension of Asset and Risk Management in the Presence of American Options 645
Simulations
The formula obtained by Whaley (1981) is used to generate call option
prices. Using the following parameters:
Strike price, K = 100, Ex-dividend date, t = 6 months, Time to
maturity, T = 1 year, Interest rate for 6 months, r = 0.04, Volatility of
the stock, (6 months), σ = 0.2, and Dividend, D = 5. Option prices are
reported in Table 14.10.
The different applications presented for European options can be used
with American options. Besides, the compound option approach apply to
the valuation of wildcard options, some exotic and complex options. These
applications will be studied in this book.
dividend, S, is:
c(St∗ , T − t, E) = St∗ + D − K.
The option described in (c/) can be priced using:
((y−r−(λC −λS ))T t
cc = Se N 2 a1 , b 1 ,
T
−(r+λC )T ti
− Ke N 2 a2 , b 2 , − (St∗ + D − K)e−(r+λC )t N1 (b2 )
T
with:
S 1 √ √
a1 = ln + y + σ 2 + λS T σ T, a2 = a1 − σ T
K 2
S 1 2 √ √
b1 = ln + y + σ + λS t σ t, b2 = b 1 − σ t
St∗ 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 647
+ De−(r+λC )t N1 (b2 ).
Table 14.11. Simulations of option values for the continuous-time model and the
discrete time model using the following parameters. S = 175, r = 0.1, D = 1.5, T = 30,
t = 24, σ = 0.32, λc = 0, and λs = 0.
Table 14.12. Simulations of option values for the continuous-time model and the
discrete time model using the following parameters. S = 175, r = 0.1, D = 1.5, T = 30,
t = 24, σ = 0.32, λc = 0.001, and λs = 0.01.
Extension of Asset and Risk Management in the Presence of American Options 649
Table 14.13. Simulations of option values for the continuous-time model and the
discrete time model using the following parameters. S = 175, r = 0.1, D = 1.5, T = 30,
t = 24, σ = 0.32, λc = 0.05, and λs = 0.1.
Table 14.14. Simulations of option values for the continuous-time model and the discrete
time model using the following parameters. S = 175, r = 0.1, D = 1.5, T = 30, t = 24,
σ = 0.32, λc = 0.1, and λs = 0.05.
S(V, T ) = VT − M if VT ≥ M
S(V, T ) = 0 if VT < M
is solved using standard methods for the price of a European call, which is
found to be equal to:
with:
V 1 √ √
d1 = ln + r + σ 2 + λV T σ T d2 = d1 − σ T
M 2
and where N1 (·) is the univariate cumulative normal density function.
Extension of Asset and Risk Management in the Presence of American Options 651
for the current time. The return on the firm’s assets follows the stochastic
differential equation:
dV /V = αv dt + σv dzv
where αv and σv refer to the instantaneous rate of return and the standard
deviation of the return of the firm per unit time, and dzv is a Brownian
motion. Using the definition of the call C(V, t), its return can be described
by: dC/C = αc dt + σc dzc where αc and σc refer to the instantaneous rate of
return and the standard deviation of the return on the call per unit time,
and dzc is a Brownian motion. Using Ito’s lemma as before, the dynamics
of the call can be expressed as:
1
dC = Cvv σv2 V 2 dt + Cv dV + Ct dt.
2
It is possible to create a risk-less hedge with two securities, between the
firm and a call to get the partial differential equation:
1 2 2
σ V Cvv + (r + λv )V Cv − (r + λC )C + Ct = 0 (14.28)
2 v
where λv in an information cost relative to the firm’s value. At the option’s
maturity date, t = T0 , the value of the call option on the firm’s stock must
satisfy the following condition: CT0 = max[ST0 − K, 0] where K stands for
the strike price. Since the stock is viewed as an option on the value of the
firm, re-call the value of ST0 :
S(VT0 , T0 ) = VT0 e−(λS −λV )(T −T0 ) N1 (d1 ) − M e−(r+λS )(T −T0 ) N1 (d2 )
with:
V 1
d1 = ln + r + σ2 + λV (T − T0 ) σ (T − T0 )
M 2
d2 = d1 − σ (T − T0 ).
With:
V0 1 2
h = ln + r + λv − σv T0 σv T 0
V̄ 2
V0 1 √
k = ln + r + λv − σv2 T σv T
M 2
where:
V̄ 1 2 √
k(V̄ ) = ln + r + λv − σv T σv T
M 2
and N2 x, y, TT0 is the bivariate cumulative normal distribution with
upper integral limits x and y and TT0 is the correlation coefficient.
A first special case is obtained when information costs regarding the
call are equal to information costs for the stocks, i.e., λc = λs . In this
case, the investors suffer sunk costs to get informed about the equity and
the assets of the firm. The costs regarding the equity and the firm’s cash
flows reflect the agency costs and the asymmetric information costs. The
formula is:
√ T0
C0 = V0 e−(λc −λv )T N2 h + σv T0 , k + σv T ,
T
−(r+λc )T T0
− Me N2 h, k, − Ke−(r+λc )T0 N1 (h).
T
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 653
Table 14.15. Simulation and comparison of the Black and Scholes model and our
model for the values of European equity options. M = 100, r = 0.08, T = 0.25,
and σv = 0.4.
70 0 0 0 0
80 0.9800 0.9829 0.9775 0.9780
90 4.1206 4.1189 4.1103 4.0984
100 8.9163 8.9085 8.8941 8.8641
110 15.6302 15.6119 15.5912 15.5340
120 23.8003 23.7660 23.7409 23.6489
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
λc = 0% λc = 2% λc = 1% λc = 1%
λs = 0% λs = 2% λs = 1% λs = 2%
V0 λv = 0% λv = 2% λv = 2% λv = 2%
reported for the Black and Scholes model. The difference between the two
models depends on the magnitude of information costs. Since the Black and
Scholes model overvalues call option prices, our model reduces the amount
of this mispricing bias.
Gives the simulation results for the compound option formula with
information costs and the Geske’s compound call formula for the following
parameters: K = 20, M = 100, r = 0.08, T = 0.25, T0 = 0.125, and
σv = 0.4. The parameters used for information costs in Table 14.2 are:
Case a: (λc = λs = λv = 0%);
Case b: (λc = λs = λv = 2%);
Case c: (λc = λs = 1%, λv = 2%) and
Case d: (λc = 1%, λs = λv = 2%).
In case (a), we have exactly the same values as those generated by
the formula given by Geske (1979). This case is a benchmark for the
comparisons of our results. The table shows that the compound option price
is an increasing function of the firm’s assets V . This result is independent
of the values attributed to information costs. Note also that the compound
option price is an increasing function of the information costs regarding the
firm’s assets, λv . When λv is fixed, this allows the study of the effects of the
other information costs on the option value. In this case, the option price
seems to be a decreasing function of the two information costs λc and λs .
When comparing cases (b) and (c) on one hand and the cases (c) and (d)
on the other hand, we observe this decreasing feature.
Summary
The option pricing literature has evolved since the work of Black–Scholes
(1973) and Merton (1973). Even if models now exist for the valuation of
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 655
Questions
1. What is the general problem in the valuation of American options?
2. When American calls are exercised?
3. When American puts are exercised?
4. When American futures calls and puts are exercised?
Extension of Asset and Risk Management in the Presence of American Options 657
But,
√
T0 ξ+(r+λV −1/2σ 2 )T0 ]
VT0 = V0 eσ
where√
the distribution of ξ is the standard Gaussian law. Letting g(x) =
2
V0 eσ T0 x+(r+λV −1/2σ )T0 ] , then:
−(r+λC )T0
C0 = e (e−(λS −λV )(T −T0 ) g(x)N (d1 (x))
2
−(r+λS )(T −T0 ) e−x /2
− Me N (d2 (x)) − K) √ dx
2π
with:
V
√
ln M
0
+ σ T0 x + (r + λV )T + 1/2σ 2 T − σ2 T0
d1 (x) =
σ (T − T0 )
d2 (x) = d1 (x) − σ (T − T0 ).
Exercises
Exercise
Show that the following bounds apply for the valuation of European call
options on an underlying asset S for a maturity date T .
1. Prove that C ≤ S.
2. Prove that C ≥ max[S − Ee(r+λs )(T −t) , 0].
3. Show that:
Solution
1. To show that the call value is less than the underlying asset price, we
construct the following portfolio, Π which comprises the underlying asset
S and the call C:
Π=S−C
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Π(T ) = S − max(S − E, 0) ≥ 0
Hence, the absence of arbitrage opportunities in efficient markets
implies that:
Π(t) = S − C ≥ 0.
Since C ≥ 0, then:
C ≥ max S − e−(r+λs +λc )(T −t),0
3. We construct a portfolio with two calls with two different strike prices
on the same underlying asset,
Π = C1 − C2 .
Extension of Asset and Risk Management in the Presence of American Options 659
Finally, we have:
Exercise
Show that the following bounds and relationships apply for the valuation
of European put options on an underlying asset S with a maturity date T .
Solution
The arbitrage argument can be used to prove the different results.
1. To show that the put value is less than the discounted value of the
strike price, we construct the following portfolio, Π which comprises the
underlying asset S and the put P :
Π = P − E at the maturity date, t = T , the value of the portfolio can
be written as:
Π(T ) = max(E − S, 0) − E ≤ 0.
Hence,
Π(T ) = S + max(E − S, 0) ≥ E.
Hence,
Π = P2 − P1 .
This gives:
0 ≤ Π(T ) ≤ E2 − E1 .
Exercise
We consider a call C(S, t) and a put option P (S, t) on the same underlying
asset S and the same strike price. The maturity date is T .
1. Since, each of these options satisfies the extended Black–Scholes equation,
can you prove that a portfolio with a long call and a short put verifies also
the Black–Scholes equation?
2. What are the boundary and final conditions that must be satisfied for
this portfolio?
Solution
Since the call and the put satisfy the extended Black–Scholes equation, we
have:
∂C 1 ∂ 2C 2 2 ∂C
+ σ S + (r + λs )S − (r + λc )C = 0 (14.29)
∂t 2 ∂S 2 ∂S
∂P 1 ∂2P 2 2 ∂P
+ σ S + (r + λs )S − (r + λp )P = 0. (14.30)
∂t 2 ∂S 2 ∂S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 661
(r + λv )V = (r(C − P ) + λc C − λp P ).
This can be written as:
r(C − P ) + λv (C − P ) = r(C − P ) + λc C − λp P
or,
λv (C − P ) = λc C − λp P.
Hence,
λv (C − P ) = λc (C − P ) + λp (C − P ) + λc P − λp C
= (λ+ λp )(C − P ) + λc P − λp C
By identification, the information cost on the portfolio corresponds to the
sum of both costs on the call and the put option,
λc + λp = λv
Hence, λc P = λp C then:
λc C
= .
λp P
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
λc P = λp C.
when,
S → +∞.
Exercise
We look for a random walk followed by a European option V (S, t). The
extended Black–Scholes equation can be used to simplify the equation
for dV .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 663
Solution
Consider the following dynamics of the underlying asset.
dS = µSdt + σSdX.
Exercise
We consider the extended Black equation for the pricing of futures options
and the Black–Scholes equation with a constant continuous dividend yield
D = r. How futures options are priced when we know the value of an option
with the same payoff for the spot asset?
Solution
We denote by W (F, t) the value of an option as a function of the futures
price F and time t. This option must satisfy the following equation:
∂W 1 ∂ 2W 2 2
+ σ F − (r + λw )W = 0.
∂t 2 ∂F 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
References
Abramowitz, M and IA Stegun (1972). Handbook of Mathematical Functions
with Formulas, Graphs, and Mathematical Tables. Washington, D.C.: U.S.
Government Printing Office.
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June), 301–320.
Bellalah, M. (1999). The valuation of futures and commodity options with
information costs. Journal of Futures Markets, 19 (September), 645–664.
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13(3),
1895–1927.
Bellalah, M and B Jacquillat (1995). Option valuation with information costs:
theory and tests. The Financial Review, 30(3) (August), 617–635.
Bellalah, M and JL Prigent (2001). Pricing standard and exotic options in the
presence of a finite mixture of Gaussian distributions. International Journal
of Finance, 13(3), 1975–2000.
Bellalah, M and F Selmi (2001). On the quadratic criteria for hedging under
transaction costs. International Journal of Finance, 13(3), 2001–2020.
Bellalah, M, JL Prigent and C Villa (2001a). Skew without skewness: asymmetric
smiles; information costs and stochastic volatilities. International Journal of
Finance, 13(2), 1826–1837.
Bellalah, M, Ma Bellalah and R Portait (2001b). The cost of capital in
international finance. International Journal of Finance, 13(3), 1958–1973.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch14
Extension of Asset and Risk Management in the Presence of American Options 665
Chapter 15
Chapter Outline
This chapter is organized as follows:
1. Section 15.1 develops the main concepts for the valuation of bond options
and interest-rate options.
2. Section 15.2 presents a simple non-parametric approach to bond futures
option pricing.
3. Section 15.3 provides one-factor interest-rate modeling and the pricing
of bonds in a general case by accounting for the effects of information
uncertainty.
4. Section 15.4 shows how fixed income instruments can be valued as a
weighted portfolio of power options.
5. Section 15.5 develops in detail the Merton’s model for equity options in
the presence of stochastic interest rates.
6. Section 15.6 provides some models for the pricing of bond options.
7. Appendix A presents in detail an analysis of Government bond futures
and their implicit embedded options.
8. Appendix B shows how one-factor interest-rate models are used in
practice.
9. Appendix C presents in detail the derivation of Merton’s model in the
presence of stochastic interest rates.
667
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
Introduction
Several authors have derived alternative formulas to the basic Black and
Scholes (1973) model (B–S) for the pricing of stock options, index options,
bond options, and foreign currency options, when interest rates are not
constant. However, until 1989, all the proposed models, except Merton
(1973), used the assumption of a constant free rate prevailing during the
option’s life, i.e., the effects of the interest rate’s variance and covariance
with the underlying asset’s return on the option’s price were precluded from
models other than that of Merton (1973). Options on the Chicago Board of
Trades’ (CBOT) Treasury bond futures were introduced in October 1982.
It is well known that interest-rate options are more difficult to value than
stock options, currency options, index options, and most futures options.
The Black (1976) model applied to fixed-income futures options is based
on the assumption of a known constant interest rate, while long-term rates
and their associated note and bond futures prices are uncertain. The most
popular alternatives to the Black model depend in general on an ad hoc
dynamic assumption about the dynamics of the short-term interest rate,
which in turn constrains their comovements with long-term interest rates.
One-factor term structure models are widely used in the pricing of interest-
rate derivatives that cannot be accomodated by Black’s (1976) model. In
these models, changes in the yields of all maturities are perfectly correlated,
at least instantaneously. This is the case for affine one-factor term structure
models as the Ho and Lee (1986), Hull and White (1990), and Cox et al.
(1985). In these models, the yield on each zero-coupon bond is a linear
function of the short-term interest rate. This implicit assumption affects
the volatility of forward interest rates and option prices. The standard
models of interest rates proposed by Vasicek (1977), Cox et al. (1985),
Brennan and Schwartz (1977) etc., have two major disadvantages. First,
they have a significant number of parameters which are not observable.
Second, they are not often consistent with the current term structure of
interest rates. This chapter presents some models for the analysis and
valuation of derivative assets whose values depend on stochastic interest
rates. Since Merton’s model represents a convenient starting point for most
of the complete option pricing models, it deserves a special treatment in
this chapter. We present specific models for the valuation of bonds and
contingent claims whose values depend directly on the dynamics of interest
rates. Positions in government bond futures have several implicit embedded
options. These positions corresponding to several specifications of the day
(delivery date and end-of-month options), the time of delivery (wildcard
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
rate options. The third assumption regarding a constant variance for prices
is also inappropriate. Therefore, it is appropriate to use models which
account for the yield curve and do not allow arbitrage opportunities: yield
curve option pricing model and arbitrage for option pricing models. Option
pricing models use six parameters: the current price of the underlying bond,
the strike price, the short-term risk-free rate, the coupon rate, the time
to maturity, and the expected interest rate volatility. This last factor is
unknown and must be estimated. This parameter can be estimated by
assuming that the option is priced correctly and to imply the interest rate
volatility from the option pricing model.
where F is the current underlying futures price and X is the option’s strike
price. This model and any other parametric model, determine implicit esti-
mates of both actual and risk-neutral probabilities. By contrast, canonical
valuation gives the user the possibility to specify a particular assessment
about the actual distribution on the underlying asset at expiration. This
provides the basis for risk-neutral probabilities. Hence, the futures price
growth rate until T does not have to be normal or to conserve the same
distribution for each possible current futures price, as it is assumed in
Black’s model.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
Stutzer and Chowdhury (1999) used the history of bond prices to form a
catalog of histograms of futures price growth rates. This allows to determine
non-normal distributions of the futures price growth rate, which varies with
the current futures price.
solving Eq. (15.2) are known as the canonical probability distribution Π̂(h),
h = 1, . . . , H. The canonical model value of a call is computed as:
H
Call = Π̂(h) max[Ph (T, F ) − X, 0]e−rT (15.3)
h=1
Π = V1 − ∆V2 (15.4)
It is possible to apply Ito’s lemma to find the change in the portfolio value
over a short interval of time dt.
∂V1 ∂V1 1 ∂ 2 V1
dΠ = dt + dr + w2 dt
∂t ∂r 2 ∂r2
∂V2 ∂V2 1 2 ∂ 2 V2
−∆ dt + dr + w dt (15.5)
∂t ∂r 2 ∂r2
Since this equation has two unknowns, and the right and left-hand
sides differ by the maturity, this means that we can write this without the
maturity. In this case, we have:
2
∂V
∂t
+ 12 w2 ∂∂rV2 − (r + λV )V
∂V
= a(r, t).
∂r
∂V 1 ∂ 2V ∂V
+ w2 2 + (u − γw) − (r + λV )V = 0 (15.6)
∂t 2 ∂r ∂r
where λV stands for the information cost in the bond market and γ(r, t)
corresponds to the market price of risk. Since the zero-coupon bond price
at maturity T is V (r, T ; T ) = 1. This gives the final condition that must be
imposed in the search for the solution to this equation.
In the presence of a coupon-paying bond, the coupons must be
integrated in the equation as follows:
∂V 1 ∂ 2V ∂V
+ w2 2 + (u − γw) − (r + λV )V + C(r, t) = 0
∂t 2 ∂r ∂r
If the coupon is paid discretely, then the following condition must be
satisfied:
V (r, t− +
c ; T ) = V (r, tc ; T ) + C(r, tc ),
Using the previous bond pricing equation, this may be written as:
∂V ∂V
dV = w dX + wγ + (r + λV )V dt
∂r ∂r
or in an equivalent way:
∂V
dV − (r + λV )V dt = w (dX + γdt) (15.7)
∂r
The term dX in this equation reveals that this portfolio is not riskless.
The deterministic term in γ in the right-hand side can be seen as the excess
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
return above the risk-free rate. The extra risk is compensated by an extra
γdt per unit of extra risk, dX. The solution of the bond pricing equation
can be seen as the expected value of all cash flows where the expectation is
taken with respect to the risk-neutral variable rather than the real variable.
In fact, the drift in the equation does not correspond to the drift of the real
spot rate u, but rather to the drift of a “risk-neutral spot rate”. The drift of
this rate is (u − γw) and its dynamics are given by dr = (u − γw)dt + wdX.
∂V 1 ∂2V ∂V
+ w2 2 + (u − γw) − (r + λV )V = 0
∂t 2 ∂r ∂r
where λV corresponds to the information cost related to the debt market.
At maturity, the callable bond price converges to V (r, T ) = 1. At a coupon-
payment date, the following relationship is applied: If the coupon is paid
discretely, then the following condition must be satisfied:
V (r, t− +
c ) = V (r, tc ) + C
where t− +
c refers to the instant just before the coupon payment and tc refers
to the instant just after the coupon payment. This condition reflects the
jump in the bond price at the coupon date. If the call price is Cp , then the
bond price must satisfy the following condition V (r, t) ≤ Cp .
where
c: fixed swap coupon;
Ra : forward rate for one year and
Rb : two-year rate (in a year).
where
S: underlying asset price;
K: strike price;
r: risk-less rate to t and
d: continuous dividend yield or the risk-less foreign rate.
The generalization for any power n gives:
√ √
S n exp(−t(r − n(r − d + σ2 /2) − (nσ2 )/2))N [x − σ t + nσ t ] (15.8)
Using Taylor’s expansion of e−Y (T −t) with Eq. (15.8) gives an iteration
of the following sum for exponentials:
inf
1/n!(−Y τ )n exp(−t(r − n(r − d − σ2 /2) − (nσ2 )/2))
n=0
where τ = T − t.
The first four terms of this expression are:
1
exp(−rt) + (−Y τ ) exp(−dt) + (−Y τ )2 exp(−t(−r + 2d − σ 2 ))
2
1
+ ((−Y τ )3 exp(−t(−2r + 3d − 3σ2 )), . . .
6
Using the Black framework for r = d gives:
with
τ
−rτ 1
P (τ ) = e and στ2 = [σ 2 + δ 2 − 2ρσδ]du.
τ 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
of negative interest rates. In this case, the Ho-Lee model can be written as:
√
dr = θ(t)dt + σ rdz.
and
√
dr = (θ(t) + a(t)(b − r))dt + σ rdz
σ(B(t, T2 ) − B(t, T1 ))
.
F (T2 − T1 )
In this model, B(0, T ) can be obtained using the current term structure
of volatilities and A(0, T ) from the term structure of interest rates. Hull and
White (1990) showed that:
Hull and White showed that the price of a European call option
maturing at time T on a discount bond maturing at time s is:
with
2
log PP21X v T
σ
h= + , v 2 = (B(0.τ ) − B(0, T ))2 dτ
v 2 0 δB(0, τ )/δτ
where P1 and P2 refer to the prices of discount bonds maturing at times τ
and T , respectively.
Bond prices
The time t price of a discount bond maturing in T is given by:
and
[1 − e−κ(T −t) ]
B(t, T ) =
κ
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
with
[(B(t,T )−T +t)(κ2 θ−0.5σ2 )]
A(t, T ) = e κ2 −σ2 B(t,T )2 /4κ .
where:
1 P (t, τ ) σ 2 [1 − e−2κ(T −t) ]
h= ln + 0.5σp , σp = B(t, τ )
σp P (t, T )X 2κ
The European put price is:
dr = θ(t)dt + σdz.
and
P (0, T )
ln A(t, T ) = ln − (T − t)δ ln P (0, t)/δt − 0.5σ2 t(T − t)2 .
P (0, t)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
European options
The European call price is:
where:
1 P (t, τ )
h= ln + 0.5σp , σp = σ(τ − T )T − t]
σp P (t, T )X
Bond prices
The time t price of a discount bond maturing in T is given by:
and
[1 − e−κ(T −t) ]
B(t, T ) =
κ
with
[(B(t,T )−T +t)(κ2 θ−0.5σ2 )]
ln A(t, T ) = e κ2 −σ2 B(t,T )2 /4κ
where
σ2
ν(T, t)2 = .
2κ3 (e−κT − e−κT )2 (e2κT − 1)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
where:
1 P (0, τ )
h= ln + 0.5ν(T, τ ).
ν(T, t) P (0, T )X
Summary
Several models are proposed in the finance literature for the estimation of
the option fair price. Most models are based on an arbitrage or risk-less
valuation argument. The most well-known model is the Black and Scholes
(1973) model. The most comprehensive empirical studies of the CBOT bond
futures options reveal that the Black’s model of futures options suffers from
moneyness bias, which is similar to the one documented for B–S model
of stock index options. Stutzer and Chowdhury (1999) used a canonical
model to value CBOT bond futures without any assumption implying
a specific parametric form for the underlying futures price distribution.
They observed that Black’s model underpriced in-the-money calls relative
to others and that the implied volatilities are inversely related to the
strike price. This mispricing bias does not seem to appear in the canonical
model. Hart (1997) showed that a fixed income instrument, which is either
convex or concave with respect to the interest rate can be viewed as a
weighted portfolio of power options on interest rates through a polynomial
transformation of the option payout. This allows the pricing of swaps (on
Eurodollar futures) in a Black and Scholes context. If swaps are viewed
as bonds, the swap convexity can be valued using Black’s (1976) model.
In this chapter, the model developed by Merton (1973) for the pricing
of stock options in the context of stochastic interest rates is presented in
great detail. This model represents a starting point towards the theory of
option pricing when interest rates are uncertain. The main ideas in Merton’s
model were used later by many authors for the pricing of a large number
of interest-rate sensitive claims. The literature on the pricing of bonds and
bond options is concerned mainly with the stochastic process describing
the dynamics of interest rates. Since there are several approaches for the
modeling of term structure dynamics, we give the main results in the work
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
Questions
1. What are the different categories of government assets?
2. What is a Treasury security?
3. What is the main property of a bond?
4. What are the different measures of yields?
5. Which of the models presented in this chapter is appropriate for the
pricing of short-term options on long-term bonds?
6. What are the valuation parameters in Merton’s model?
7. What are the valuation parameters in Heath, Jarrow, and Morton’s
model?
8. What are the specificities of the Heath, Jarrow, and Morton’s model?
time to maturity and different strike prices. They defined simple creteria
for the CTD under yield changes, quantified the value for a short position,
and presented the European call and put options values on the bond yield.
where
t: value date for each cash bond settlement;
t : delivery date of the futures contract;
P (t): clean price of the bond;
AI(t): interest accrued at time t;
n: number of days between the cash and the futures settlement dates;
Crec : coupon received in the holding period (per face value) and
CF : conversion factor. This criteria is used in the determination of the
CTD.
The implied futures price for each bond is defined by:
P (t)
IF P = (A.2)
CF
where
IF P : implied futures price;
P (t): clean price of the bond and
CF : conversion factor.
where y refers to the bond yield. The CTD bond is determined by comparing
the B of various bonds for each of the N bonds in the basket using the
variable:
∆BN = BN − BCT D .
The change in the yield (referred to as Switchpar (N ) which makes the bond
N the CTD is equivalent to:
When Eq. (A.1) is used to get the first derivative of BN with respect to dy,
this gives:
dBN dPN dF
= (1 + m/36,000) − (A.5)
dy dy dy
Using Eqs. (A.3) and (A.5), Dabansens and Bento (1997) gave the
following expression for the yield switch for bond N :
∆BN
Switchpar (N ) =
(1 + m/36,000)
× [M DN (t)PN (t) − (CFN /CFCT D )M DCT D (t)PCT D (t)]−1
In this context, the futures price related to this yield shift (F utSwpar
(N )) is:
PCT D M DCT D
F (t) − F utSwpar (N ) = (1 + m/36,000)Switchpar (N )
CFCT D
Let us consider in a second step the case of a relative yield shift. Assuming
no change in the yields of other bonds, the yield change for the bond
N , Switchrel (N ) can be calculated using a similar condition as in the
previous case:
The relative yield switch for bond N to become the CTD is:
∆BN
Switchrel (N ) =
(1 + m/36,000)(M DN (t)PN (t))
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
where Ts refers to the time when the CTD switches from the original CTD
bond to bond N from the basket. In this context, the expression of the
profit at futures settlement is given by:
and
PN (TS )M DN (TS )CFCT D
XN =
CFN
in order to write the profit as:
the profit:
+ Switchpar (N ) − y)dy
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
with:
The value of the delivery switch option related to bond N is given by:
∞
Optpar (N ) = −Krel PSN (SprN = S)
sprN O+Switchrel (N )
interest rates will be higher than in reality. Hence, it is risky to use one-
factor models in the pricing of instruments other than the ones on which
the model is calibrated. Using market data, Klaassen et al. (1998) showed
when the deficiencies of one-factor term structure models lead to large
errors.
where ρ(r1 , r2 ) refers to the correlation between the two rates and σ
corresponds to the volatility. This relation shows the effect of the correlation
between zero-coupon rates on the standard deviation of a forward rate.
Denoting by ∆ = t2 − t1 and deriving σ(ft1 t2 ) with respect to ρ gives:
dS
= αdt + σdW (C.1)
S
where
α: instantaneous expected return on the common stock and
σ 2 : instantaneous variance of return, restricted to be a known function of
time.
dP
= µ(τ )dt + δ(τ )dq(t, τ ) (C.2)
P
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
where
P (τ ): price of a discounted loan with maturity τ satisfying P (0) = 1;
µ(τ ): instantaneous expected return;
δ 2 (τ ): instantaneous variance with δ(0) = 0 and
dq(t, τ ): a standard Gauss–Wiener process.
and
dq(t, τ )dq(t, T ) = ρτ T dt
Using the properties of stochastic calculus, the values of (dS)2 , (dP )2 , and
(dS)(dP ) are given by:
Substituting from Eqs. (C.1) and (C.2), Eq. (C.3) is re-written as:
∂C ∂C ∂C
dC = [αSdt + σSdW ] + [µP dt + δP dq] + dτ
∂S ∂P ∂τ
2 2 2
1 ∂ C 2 2 ∂ C 2 2 ∂ C
+ σ S dt + δ P dt + 2 σδρSP dt
2 ∂ 2S ∂ 2P ∂S∂S
or in a simple form:
with
1 ∂C ∂C ∂C 1 ∂ 2C
β= (αS) + (µP ) − + σ2 S 2
C ∂S ∂P ∂τ 2 ∂ 2S
2
1 ∂2C 2 2 ∂ C
+ δ P + σδρSP
2 ∂2P ∂S∂S
S ∂C S ∂C
γ=σ , η=δ
C ∂S C ∂S
Now consider a strategy where the weights wj = wj∗ are chosen so that
the stochastic terms in Eq. (C.7) affecting dW and dq are always zero.
The expected return on this strategy must be zero since it requires zero
investment. Hence,
w1∗ (α − µ) + w2∗ (β − µ) = 0
w1∗ σ + w2∗ γ = 0 (C.8)
−w1∗ δ + w2∗ (η − δ) = 0
Using Eq. (C.6) and re-arranging terms, Eq. (C.11) can be re-written as:
2 2
1 2 2 ∂2C ∂ C 2 2 ∂ C ∂C
0= σ S + δ P + 2σδρSP −
2 ∂2S ∂2P ∂S∂S ∂τ
(C.12)
S
Applying Ito’s lemma for x = KP (τ ) and replacing each partial derivatives
gives:
1 −S 1 2S 2 2
dx = dS + dP + (δP ) dt − σSδpηdt
KP KP 2 2 KP KP 2
or
1 −S S 1
dx = dS + 2
dP + (δ)2 dt − σSδP ηdt
KP KP KP KP
1 −S
dx = [αSdt + σSdz] + [µ(τ )P dt + δ(τ )P dq]
KP KP 2
S
+ [(δ)2 − σδpη]dt
KP
1 σS −S −Sδ S S
dx = αSdt + dz + µdt + dq + (δ)2 dt − σδηdt
KP KP KP KP KP KP
S
This last equation can be written using the definition of x = KP
or
dx
= αdt + σdz − µdt − δdq + δ 2 dt − σδηdt
x
dx
= (α − µ + δ 2 − σδη)dt + σdz − δdq
x
h(x, τ ; K) = C(S, P, τ, K)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
1 2 2
ν x h11 − h2 = 0, h(0, τ, K) = 0, h(x, 0, K) = max[0, x − 1]
2
We denote by
τ
T = ν 2 (s)ds
−0
and
y(x, T ) = h(x, τ )
The solution presented in Merton (1973) for the call price is:
1
y(x, T ) = [xerf c(h1 ) − erf c(h2 )] (C.13)
2
where
ln(x) + 12 T ln(x) − 12 T
h1 = − √ , h2 = − √
2T 2T
and
τ
T = [σ 2 + δ 2 − 2ρσδ]du
0
with
τ
1
P (τ ) = e−rτ and στ2 = [σ 2 + δ 2 − 2ρσδ]du.
τ 0
References
Briys, E, M Bellalah et al. (1998). Options, Futures and Exotic Derivatives. En
collaboration avec E. Briys, et al., John Wiley & Sons.
Black, F (1995). Interest rates as options. Journal of Finance, 50 (December),
1411–1416.
Black, F (1976). The pricing of commodity contracts. Journal of Financial
Economics, 3, 167–179.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
Brennan, MJ and E Schwartz (1977). Saving bonds, retractable bonds, and
callable bonds. Journal of Financial Economics, 5, 67–88.
Cox, J, J Ingersoll and S Ross (1985). A theory of the term structure of interest
rates. Econometrica, 53, 385–407.
Dabansens, F and F Bento (1997). Swiching on to bonds. Risk, 10(1), (January),
68–73.
Hart, I (1997). Unifying theory. Risk, 10(2) (February), 54–55.
Hart, I and M Ross (1994). Striking continuity. Risk, 7(6), 50–51.
Hull, J and A White. Coming to terms. Black holes.
Ho and Lee (1986). Term structure movements and pricing interest rate contingent
claims. Journal of Finance, 41, 1011–1029.
Hull, J and A White (1990). Pricing interest rate derivative securities. Review of
Financial Studies, 3, 573–592.
Jamshidian, F (1989). An exact bond option formula. Journal of Finance,
44 (March), 205–209.
Klaassen, P, E Van Leewen and B Schreurs (1998). One-factor fallacies. Risk,
11 (December) 1998, 56–59.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch15
Chapter 16
Chapter Outline
This chapter is organized as follows.
1. Section 16.1 provides simple examples of interest-rate sensitive instru-
ments and explains the main concepts in bond pricing.
2. Section 16.2 studies interest rates and the pricing of bonds under
certainty and uncertainty.
3. Section 16.3 develops some standard models for the pricing of bonds and
bond options.
4. Section 16.4 discusses the relative merits of the competing models.
5. Section 16.5 provides a comparative analysis of term structure estimation
models.
6. Section 16.6 presents some new evidence on the expectations hypothesis.
It gives some term premium estimates from zero-coupon bonds.
7. Section 16.7 studies the distributional properties of spot and forward
interest rates using the following currencies: USD, DEM, GBP, and JPY.
8. Appendix A provides some applications of interest-rate models to
account for the effects of information costs.
9. Appendix B shows how to implement the Black–Derman and Toy model
(BDT) with different volatility estimators.
703
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
Introduction
The fixed income market refers to the global financial market where various
interest-rate sensitive products are traded. The management of interest-rate
risk refers to the control of changes in value of the cash flows due to the
changes in interest rates. Market conventions can vary from one country
to another. The Treasury bill or T-bill is issued in general for a short-term
period and corresponds to a discount bond for which the buyer receives
the face value at maturity with no coupons. The pricing of bonds and
interest-rate options needs a specification of the dynamics of interest rates
under certainty and uncertainty. Several interest-rate models are based on
the spot rates where the spot interest rate is the underlying state variable.
This is the case for the Vasicek (1977) model, Hull and White (1987, 1988,
1990, 1993), Cox et al. (1985) and so on. These models require as inputs
different parameters in the description of the possible future paths of the
spot rate. This implies the search for parameter values which causes the
calculated zero-coupon bond prices to be close to the market prices. Many
other popular spot rates can be expressed in a simple way under the Heath,
Jarrow and Morton (1992) (HTM) model. The Ho and Lee (1986) model
is useful in the pricing of interest-rate options with respect to the observed
initial term structure of interest rates. However, the model is based on a
simplifying assumption, that all rates along the yield curve fluctuate to the
same degree. This assumption is not supported in practice. Hull and White
(1992) defines a yield-curve-based interest-rate model as a model for the
dynamics of the current term structure of interest rates. The model should
allow a correct valuation of bonds giving a price equivalent to the market
price. Hull and White (1992) compared three models describing the process
for the short rate: Ho and Lee (1986), Black, Derman and Toy (1990) BDT,
and Heath, Jarrow and Morton (1990). Empirical work by Fama (1984a, b,
1986), Fama and Bliss (1987) and Froot (1989) document term premiums at
the short end of the term structure. Longstaff (1990) finds that even short-
term premiums may be simply a function of the time-varying nature of bond
returns. Fama (1984b) and Fama and Bliss (1987) find that forward rates
are poor forecasters of future short-term interest rates. However, forecasts
improve for longer forecast horizons. This chapter presents some of the
popular models of the short-term interest rate in a continuous-time setting.
Several authors have shown the existence of an arbitrage free family of bond
prices associated with a given short-term rate process. In the presence of a
one-dimensional diffusion process, it is possible to obtain analytic results or
closed-form solutions for zero-coupon bonds and some European options.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
1 − B(t, T ) 1
= − 1.
B(t, T ) B(t, T )
1
Y (t, T ) = − ln[B(t, T )], ∀ t ∈ [0, T ). (16.1)
T −t
The term structure of interest rates or the yield curve describes the
relationship between YTM and the maturity T . Given the dynamics of
the YTM, it is possible to show that the bond price process is determined
by the following formula:
This allows to compute the implied instantaneous forward interest rate as:
∂ ln[B(t, T )]
f (t, T ) = − . (16.5)
∂T
The instantaneous forward rate is also viewed as a limit case of a forward
rate f (t, T, H) that prevails at t for risk-less operations over the time
interval [T, H]. In the same way, using two zero-coupon bonds with different
maturities T and H, the discounting factor contains the forward rate as:
B(t, H)
= e−f (t,T,H)(H−T ) , ∀ t ≤ T ≤ H.
B(t, T )
ln B(t, T ) − ln B(t, H)
f (t, T, H) = . (16.6)
H −T
1
B(t, T ) = , ∀ t ≤ T.
(1 + ra (t, T ))T −t
In the same context, the forward actuarial rate prevailing at t for the
interval [T, H] must satisfy the following relationship:
dBt = rt Bt dt
m
Bc (t) = cj B(t, Tj ). (16.8)
j=1
A real bond pays in general a fixed coupon c and re-pays the principal
amount N . Hence, the last cash flow is cm = (c + N ). The total return on
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
m
c N
Bc (0) = + .
j=1
(1 + Ỹc (0))j (1 + Ỹc (0))m
m
rc N N
Bc (0) = + .
j=1
(1 + Ỹc (0)) j (1 + Ỹc (0))m
When coupon rate rc = Ỹc (0), the bond price is priced at par since its price
equals its face value.
The bond is priced at a discount (below par) when Bc (0) < N or
rc < Ỹc (0). The bond is priced at a premium (above par) when Bc (0) > N
or rc > Ỹc (0). For continuous compounding, using the current market price
of the bond shows that the corresponding YTM, Yc (0) satisfy:
m
Bc (0) = ce−jYc (0) + N e−mYc (0) .
j=1
For the case of zero-coupon bonds, the knowledge of its YTM allows the
computation of its price in a discrete setting as:
1
B(0, m) = .
(1 + Ỹ (0, m))m
The bond price does not account for the coupon at time i since we use
the price Bc (i) after the coupon at time i has been paid. The continuously
compounded YTM at time
on a coupon bond with “m” cash flows can be calculated using the formula:
Bc (t) = cj e−Yc (t)(Tj −t) . (16.10)
Tj >t
The bond price does not account for the coupon at time t since we use the
price Bc (t) after the coupon at time t has been paid. Note that the bond
price moves inversely to its YTM. In general, the decrease in yields raises
bond prices more than the same increase lowers bond prices. This reflects
the convexity.
where:
rc : coupon rate;
N : bond face value and
Ŷc (0): an annualized interest rate with no compounding.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
rc N N (1 − rc /Ŷc (0))
Bc (0) = + .
Ŷc (0) (1 + Ŷc (0)/2)2m
In this context, the YTM on a bond at time “i” can be found using the
following relationship:
2m−1
rc N/2 (1 + rc /2)N
Bc (i) = + ,
j=i+1 (1 + Ŷc (i)/2)j (1 + Ŷc (i)/2)2m
after the ith coupon payment. To obtain the compounded annualized yield,
or the effective annual yield, the following equality is used:
In the context of certainty, when the interest rates are known over the
period, the function F is given at each instant by:
Using this equality and the fact that F (t, t) = 1, allows one to show
that there is an interest rate r(t) such that in the absence of arbitrage
opportunities, the amount F (t, T ) is given by:
RT
F (t, T ) = e( t
r(s)ds)
. (16.12)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
So,
T
1
R(t, T ) = r(s)ds . (16.13)
(T − t) t
or
Ru
P (t, u) = EP ∗ (e− t
rs ds
| Ft ). (16.15)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
EP ∗ (X | Ft ) = E(XLT | Ft )/Lt ,
with
γσ
b∗ = b −
a
and
Ŵt = Wt + γt.
where a and σ are positive and b is a real number. In this context, the
process q(t) takes the form q(t) = −α r(t) where α is a real number.
Consequently, the model shows the same drawbacks as that of Ho and Lee
(1986), namely, the possibility of negative interest rates. A more interesting
model is obtained by HJM when σ(.) is a function of forward rate f (t, u)’s
maturity (u − t). For example, when σ(u − t) = σ0 e(−a(u−t)) with a > 0
and σ0 > 0, then the model proposed in Vasicek and studied in Hull and
White is obtained.
Example:
2 t
f (t, T ) = f (0, T ) + σ t T − + σ Ŵ (t) .
2
In the HJM model, the price of a European bond is given by:
0 ≤ t ≤ t∗ ≤ T.
For more details and other applications of the model to the pricing of an
entire book of options and volatility estimation, see Heath et al. (1992),
and Spindel (1992).
where:
r: short-term interest rate;
θ: drift factor which is a function of time;
α: reversion rate or a mean reverting factor;
σ: standard deviation and
dW : “white noise”.
This model is a simple extension of the Vasicek model. Since, the drift
factor is a function of time, this model guarantees consistency with the
initial term structure of interest rates. The mean reverting factor allows
the long-term rates to show lower volatility than short-term rates. This
specificity is confirmed in practice and represents an improvement on the
Ho and Lee (1986) and the Vasicek (1977) models. The volatility function
in this model reflects the fact that short-term interest rates fluctuate more
than long-term rates because of the reversion parameter in the spot rates.
Using this model, it is possible to obtain an analytic solution for the pricing
of European options. The price of a European call on a discount bond at
time zero is given by:
with
1 P (0, T ) 1
d1 = ln + v(t∗ , T ), d2 = d1 − v(t∗ , T )
v(t∗ , T ) P (0, t∗ )K 2
1 2 −αT
v(t, T ) = σ (e − e−αt )(e2αT − 1)
2α3
where:
t∗ : option maturity date;
K: strike price;
T: maturity date of the underlying discount bond;
P (t1 , t2 ): price at time t1 of a bond maturing at time t2 and
v(t, T ): volatility as a function of the reversion rate.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
The reader can note the analogy between the Hull and White and the
HJM formulas.
and the square root of the volatility of the spot rate are:
√
dσ = b(σ̄ − σ)dt + c σdX2 .
This formulation allows the derivation of some simple pricing formulas for
interest-rate sensitive instruments.
and
√
dy = a(ȳ − y)dt + ydX2
dr = θ(t)dt + σdz.
The process for the short rate in the BDT model is given by:
σ (t)
d log r = θ(t) + log r dt + σ(t)dz.
σ(t)
This model assumes that interest-rate changes are log-normal. The process
for the short rate in the Hull and White (1990) model is:
where:
f: forward rate;
a: reversion rate;
m: forward rate reverts to this mean m;
σ: volatility rate and
z: a standard normal variate.
The models of spot interest-rate processes are written using the same
notation employed for the estimation of discount price functions. Each fitted
function gives a discount bond pricing equation Dt = e−rt t . This equation
allows the computation of the spot rate for maturity t. In this context, the
polynomial model is given by:
Dt = a0 + a1 t1 + a2 t2 + a3 t3 + a4 t4 + a5 t5 .
Dt = e−rt t
with
1 − e−a3 t
rt = a0 + (a1 + a2 ) − a2 e−a3 t .
a3 t
1 − e−a1 t
Dt = A(t)e−Bt a0 with B(t) = .
a1
a2 a2
ln(A(t)) = (B(t) − t) a2 − 32 − 3 (B(t))2 .
2a1 4a1
Dt = A(t)e−Bt a0
with:
2a1 a2
a2
2we0.5(a1 +w)t
A(t) = h 3 , h= ,
((a1 + w)(ewt − 1) + 2w
2(ewt − 1)
B(t) = , w = a21 + a23 .
((a1 + w)(ewt − 1) + 2w
The simulations conducted show that the Vasicek–Fong model (VF6) is the
best of the models above since, it provides fast and reliable fit for the full
simulated term structure.
an interpolation method for forward rates and returns from existing one-
through five-year bond issues. Froot (1989) extracted premiums on different
fixed-income securities from survey expectations. Dhillon and Lasser (1998)
estimated term-structure premiums from a yield curve, consisting of zero-
coupon stripped US Treasury securities. Using the stripped Treasury bond
yield curve, they found a monotonically increasing term premium. This
evidence is not consistent with the expectations hypothesis. They found
that the current forward rates can be used to forecast both short-term and
long-term interest rates. However, Fama (1984a,b) and Fama and Bliss
(1987) found that short-term rates cannot be forecasted using forward
rates. Dhillon and Lasser (1998) used quarterly US coupon strip prices
from August 1986 to May 1997. Coupon strips existed at every 3-month
interval. The quarterly return from time t to t + 31 on a zero-coupon bond
with τ months to maturity at time t is given by:
B(τ − 1)t+1
Rτt+1 = ln (16.20)
Bτt
where:
Bτt : price at time t of a zero-discount bond maturing at month t + τ and
B(τ − 1)t+1 : price of a zero-discount bond at quarter t + 1.
The premium in the quarterly return is given by the difference between the
quarterly return Rτt+1 and the quarterly spot rate,
Bτt+1 = exp(−St+1 − F 2t − · · · − F τt )
where:
B0,t : price of a pure discount bond paying $1 at t;
rt : money market rate for a loan of an identical maturity and
αo,t : accrual factor.
For USD, DM, and JPY, this factor is
30
αi−1,i = .
360
For GBP, this factor is
ti − ti−1
αi−1,i = .
365
The bootstrap method is applied to obtain the prices of discount bonds
implied by the swap market as follows:
t−1
1 − st i=1 αi−1,i B0,i
B0,t =
1 + st αt−1,t
where:
B0,t : price of a pure discount bond that pays $1 at time t;
st : swap rate and
αi−1,i : accrual factor which refers to a certain number of days.
30
The accrual factor used in the study is 360 for USD, DEM, and Yen.
For the GBP, the accrual factor is
(ti − ti−1 )
αi−1,i = .
365
The above equation allows the computation of zero bond prices by the
bootstrap method.
Using these prices, it is straight forward to estimate the implied
annualized yields as:
−α0,t
1
R0,t = −1
B0,t
(ti − ti−1 )
αi−1,i = .
365
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
The one-year forward rates are estimated using the following equation:
−αt,t+12m
B0,t
f0,t,t+12m = − 1.
B0,t+12m
where:
T : number of observations;
h: a smoothing parameter and
K(.): Gaussian Kernel.
The estimated probability densities of spot and forward interest-rate
levels show that interest rates are not generated from a normal or a log-
normal distribution. They are generated from a mixture of distributions
with different means and standard deviations.
The tests of Lekkos (1999) show that the most characteristic feature
of interest rates is the high kurtosis. The results indicate that neither the
normal distribution nor the log-normal is adequate for the description of the
distribution of interest rates. The most characteristic feature of interest-rate
changes is the high kurtosis, which seems to be higher than three.
A mixture of two normal distributions seems to give a better fit to the
data than the normal or the log-normal distribution.
Lekkos (1999) assumes a mixture of normal distributions with N
mixing components. The probability density function of each observation
is described by:
N
f (∆rt ; Π, σ, µ) = Πi gi (∆rt ; σi , µi ) (16.23)
i=1
ϑ = (µ1 , µ2 , . . . , µN , σ12 , . . . , σN
2
, Π1 , Π2 , . . . , ΠN )
Πs gs (∆rt ; σs , µs )
P (s | ∆rt ) = (16.25)
f (∆rt ; Π, σ, µ)
T
s = 1
Π P (s | ∆rt ) s = 1, 2, 3, . . . , N (16.26)
T t=1
T
1
µ
s = P (s | ∆rt )∆rt s = 1, 2, 3, . . . , N (16.27)
s
TP t=1
T
1
σs = s )2 .
P (s | ∆rt )(∆rt − µ (16.28)
T Pˆist=1
This system can be solved via an iterative procedure. Given initial values
of P (s | ∆r ) for t = 1, . . . , T can be evaluated. These can be inserted
in Eqs. (16.26) to (16.28) to produce revized parameter estimates. This
procedure stops when some convergence criteria is satisfied. The EM
algorithm is useful for the estimation of ϑ̂, but does not give the number of
component distributions N . This number can be computed empirically.
We denote by Li (ϑ) the log-likelihood function when N = i.
Lekkos (1999) compares two specifications with i and j components
using:
Π1 g1 (∆rt ; σ1 , µ1 )
P (s = 1 | ∆r ) = (16.30)
f (∆rt ; Π, σ, µ)
gives the probability that each ∆rt comes from the first component
distribution. He finds that all interest rate changes have nearly a zero mean
and a very small variance. He shows that only rare events produce large
changes in interest rates, which in turn create the second component that
is responsible for excess kurtosis and skewness.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
Summary
Identification of the stochastic process describing the dynamics of the
options underlying asset is an important key in asset valuation. When the
movements of interest rates are well described by an interest-rate model,
this allows the derivation of accurate bond and option prices. Theoretical
interest-rate models depend on the knowledge of statistical properties of
the “stylized features” of the term structure. The identification of the
stochastic properties of the dynamics of interest rates is of great interest
in theory and practice. The statistical properties of the term structure of
interest rates are fundamental to the development of theoretical interest-
rate models. Each interest-rate model has its merits and dismerits. The
use of a one-factor model offers the ease of computation (the speed). The
use of a multifactor model has an advantage in hedging the yield curve.
Hence, from a practical point of view a trade-off must be done: more
robust management of yield curve risk against speed. The choice between
the two approaches depends on the situation in which a decision must
be made. What is certain is that the choice of the model depends on
the users’ needs. Ferguson and Raymar (1998) study several popular term
structure estimation models and the methodological issues in estimating
the term structure. They consider six methods to derive a cross-sectional
discount function. They generate a “true forward rate curve” to obtain
the associated spot curve and discount function. Lekkos (1999) present
an in-depth analysis of the distributional properties of interest rates. He
conducts an indirect test of the validity of the assumptions in interest rate
models of Vasicek (1977), Cox et al. (1985), Ho and Lee (1986), Hull and
White (1990), Black et al. (1990) and Black and Karasinski (1991). The
differences between these models result from the assumptions imposed on
the stochastic process. The statistical analysis in Lekkos (1999) shows that
the kernel density estimation method provides estimates of the probability
densities of interest-rate levels. The density estimates show a great degree
of multimodality. He rejects the normal and the log-normal distributions
of interest rates because of the excess kurtosis in all interest rates. He
finds that a better fit to the data is achieved using a mixture of two
normal distributions. The models of Vasicek (1977), Ho and Lee (1986)
and Hull and White (1990) assume that the distribution of interest-rate
differences is normal with a constant volatility. BDT (1990) and Black
and Karasinski (1991) assume that the distribution of interest rates is log-
normal. The tests of Lekkos (1999) show that the most characteristic feature
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
of interest rates is the high kurtosis. The results indicate that neither the
normal distribution nor the log-normal is adequate for the description of
the distribution of interest rates. A mixture of two normal distributions
seems to give a better fit to the data, than the normal or the log-normal
distribution. Bali and Karagozoglu (1999) use different estimators in the
pricing of interest rate sensitive options with the single BDT (1990) model.
The BDT model is often implemented with a constant volatility estimators
(historical or implied volatility). These volatilities estimators do not allow
the volatility of the short rate to vary over time. The choice of a class of
volatility estimation model for short rates is crucial since, the short-term
rate drives the changes in the term structure in the BDT (1990) model.
∂
F (t; T ) = − log Z(t; T ) + λ.
∂T
r(t) = F (t; t) − λ.
Wilmott (1998) shows how the HJM approach is slow in the pricing of
derivatives. We will reproduce their argument here in the presence of
information costs. Consider the current time t∗ and assume that the whole
forward rate curve is known today, F (t∗ , T ). The spot rate can be written
for any time t as:
t
r(t) = F (t; t) − λ = F (t∗ ; t) + dF (s; t) − λ.
t∗
Π = Z(t; T1 ) − ∆Z(t; T2 ).
Over a short interval of time, the change in the portfolio’s value can be
written as:
dΠ = dZ(t; T1 ) − ∆dZ(t; T2 )
or
σ(t, T1 )Z(t; T1 )
∆=
σ(t, T2 )Z(t; T2 )
then the portfolio is hedged and is risk-free. Hence, the return on this
portfolio must be the risk-less rate plus the information costs, which are
necessary to get informed about the market and the arbitrage opportunities.
In this context, we have:
where from Eq. (A.3), the value of the forward rate volatility ν(t, T ) is
given by
∂
ν(t, T ) = − σ(t, T )
∂T
and the drift rate is computed as:
T
∂ 1 2 ∂
σ (t, T ) − µ(t, T ) = ν(t, T ) ν(t, s)ds − µ(t, T ).
∂T 2 t ∂T
Since in a risk-neutral world the value of µ(t, T ) = r(t) + λ, the drift of
risk-neutral forward rate curve is linked to the volatility by:
T
m(t, T ) = ν(t, T ) ν(t, s)ds. (A.5)
t
Multi-factor HJM
Several authors develop multifactor models for the pricing of derivative
assets. If the risk-neutral forward rate curve follows a stochastic differential
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
where the different processes are dXi are uncorrelated. In this case, the
drift term is written as:
n
T
m(t, T ) = νi (t, T ) νi (t, s)ds.
i=1 t
∂Z 1 ∂ 2Z ∂Z
+ c2 2 + η(t) − (r + λZ )Z = 0 (A.6)
∂t 2 ∂r ∂r
with the condition that the zero-coupon bond price at maturity equals
to one:
Z(r, T ; T ) = 1.
The drift in the Ho and Lee model must fit the yield curve at time t∗ .
Hence, the forward rate can be written as:
T
1
F (t∗ ; T ) = r(t∗ ) − c2 (T − t∗ )2 + η(s)ds
2 t∗
and so
∂F (t∗ ; t)
η(t) = + c2 (t − t∗ ).
∂t
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
For any time later than t∗ , the forward rate can be written as:
T
1
F (t; T ) = r(t) − c2 (T − t)2 + η(s)ds.
2 t
where:
ln r(t): natural logarithm of the short rate at t;
∆ ln r(t) = ln r(t + θ) − ln r(t): change in the log interest rate;
θ: length of the time interval;
δ0 (t), δ1 (t): time-varying parameters to be estimated;
σ(t): conditional volatility of log interest rate changes at t;
W (t): a standard Brownian motion and
∆W (t): a random variable normally distributed with zero mean and
a variance θ.
The evolution of the short rate in a discrete time can be approximated
using a binomial representation:
ln r(t + θ) − ln r(t)
√
[δ0 (t) − δ1 (t) ln r(t)]θ + σ(t)√θ; with probability 1/2
= . (B.1)
[δ0 (t) − δ1 (t) ln r(t)]θ − σ(t) θ; with probability 1/2
In this context, the term δ1 (t) can be seen as a measure of the speed of
mean reversion in the log rate levels. The drift of the logarithm of the
short rate
√
[ln r(t + θ)up − ln r(t)] = [δ0 (t) − δ1 (t) ln r(t)]θ + σ(t) θ (B.1a)
or
√
[ln r(t + θ)down − ln r(t)] = [δ0 (t) − δ1 (t) ln r(t)]θ − σ(t) θ. (B.1b)
The difference between these two last equations gives the sensitivity formula
of the BDT’s model. The formula gives at time t + θ, the spread of two log
interest rates as a function of the volatility.
where Pi and P̂i correspond to the actual and estimated prices of the
options. This statistic depends on the scale of the option price. A lower value
of MSE shows that the estimated option prices are close to actual prices.
They use also the Theil inequality coefficient (TIC) which is invariant to
scale:
1 n 2
n i=1 (Pi − P̂i )
TIC =
, 0 ≤ T IC ≤ 1. (B.3)
1 n 2+ 1 n 2
n P
i=1 i n P̂
i=1 i
This statistic is between zero and one where zero corresponds to a perfect fit.
Bali and Karagozoglu (1999) shows that the time series volatility estimates
lead to more accurate predictions of option prices than the moving average
models. In particular, the GARCH and the integrated GARCH models
give more accurate representation of the volatility structure than moving
average models.
Questions
1. Can you describe with simple examples some interest rate sensitive
instruments?
2. How can we price bonds under certainty and uncertainty?
3. How to develop some standard models for the pricing of bonds and bond
options?
4. What are the relative merits of the competing models?
5. Can you provide a comparative analysis of term structure estimation
models?
6. What are the distributional properties of spot and forward interest rates?
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
References
Ait-Sahalia, Y (1996). Testing continuous-time models of the spot interest rate.
Review of Financial Studies, 9, 385–426.
Bali, T and A Karagozoglu (1999). Implementation of the BDT model with
different volatility estimators. The Journal of Fixed Income, 8, March, 25–33.
Black, F (1976). The pricing of commodity contracts. Journal of Financial
Economics, 3, 167–179.
Black, F and P Karasinski (1991). Bond pricing and option pricing when short
rates are lognormal. Financial Analysts Journal, 47, 52–59.
Black, F, E Derman and W Toy (1990). A one factor model of interest rates and
its application to treasury bond options. Financial Analysts Journal, 46(1),
January–February, 33–39.
Brennan, MJ and ES Schwartz (1982). The case for convertibles. Chase Financial
Quarterly, 1(3), Spring, 27–46.
Cox, J, J Ingersoll and S Ross (1985). A theory of the term structure of interest
rates. Econometrica, 53, 385–407.
Dhillon, US and DJ Lasser 1998. Term premium estimates from zero-coupon
bonds: new evidence on the expectations hypothesis. Journal of Fixed
Income, 8 (June), 53–58.
Fama, EF (1984a). The information in the term structure. Journal of Financial
Economics, 13, 509–528.
Fama, EF (1984b). Premiums in bond returns. Journal of Financial Economics,
13, 529–546.
Fama, EF (1986). Term premiums and default premiums in money markets.
Journal of Financial Economics, 17, 175–196.
Fama, EF and RR Bliss (1987). The information in long maturities forward rates.
American Economic Review, 77, 680–692.
Ferguson, R and S Raymar (1998). A comparative analysis of several popular term
structure estimation models. Journal of Fixed Income, 8 (March), 17–33.
Froot, KA (1989). New hope for the expectations hypothesis of the term structure
of interest rates. Journal of Finance, 44, 283–305.
Heath, D, R Jarrow and A Morton (1989). Contingent claims valuation with a
random evolution of interest rates. Review of Futures Markets, 9(1), 54–76.
Heath, D, R Jarrow and A Morton (1992). Bond pricing and the term structure
of interest rates: a new methodology for contingent claims valuation.
Econometrica, 60, 77–105.
Ho, T and S Lee (1986). Term structure movements and pricing interest rate
contingent claims. Journal of Finance, 41, 1011–1029.
Hull, J and A White (1987). The pricing of options on assets with stochastic
variables. The Journal of Finance, 42, 281–300.
Hull, J and A White (1988). An analysis of the bias in option pricing caused by
a stochastic volatility. Advances in Futures and Options Research, 3, 29–61.
Hull, J and A White (1990). Pricing interest rate derivative securities. Review of
Financial Studies, 3, 573–592.
Hull and White (1992).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
Hull, J and A White (1993). Efficient procedures for valuing European and
American path dependent options. Journal of Derivatives, 1, Fall, 21–31.
Lekkos, I (1999). Distributional properties of spot and forward interest rates:
USD, DEM, GBP, and JPY. Journal of Fixed Income, 8, March, 35–54.
Leong, K (1998). Model choice. Risk, September, 19–22.
Longstaff, FA (1990). Time varying term premiums and traditional hypothesis
about the term structure. Journal of Finance, 45, 1307–1314.
Nelson, C and A Siegel (1987). Parsimonious modeling of yield curves. Journal
of Business, 60(4), 475–489.
Spindel, M (1992). Easier done than said. Risk, 5(9), October, 77–80.
Vasicek, O (1977). A equilibrium characterization of the term structure. Journal
of Financial Economics, 5, 177–188.
Vasicek, O and G Fong (1982). Term structure modeling using exponential splines.
Journal of Finance, 37(2), 339–348.
Willmott, P (1998). Derivatives, Chicheslec: John Wiley and Sons.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch16
Part VI
743
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
744
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
Chapter 17
Chapter Outline
This chapter is organized as follows:
1. Section 17.1 presents briefly the main results in the Merton’s (1976)
jump-diffusion model and the Cox and Ross (1976) constant elasticity
of variance model.
2. Section 17.2 develops the main results regarding the pricing and hedging
of options in the presence of jumps and information costs.
3. Section 17.3 reviews the models for the valuation of options within the
presence of jumps and information costs. It also calibrates the model to
market data and shows some properties of the smile.
4. Section 17.4 studies implied volatility functions and option pricing
models.
Introduction
Since the path-breaking contribution by Black and Scholes (1973) was
published, many papers have tried to relax its most stringent assumptions.
In particular, the introduction of a stochastic volatility has been considered
by many authors. Much of the known bias reported in empirical studies
based on the B–S formula has something to do with this assumption.
The biases reported in Rubinstein (1994) for stock options, Melino and
Turnbull (1990), and Knoch (1992) for options on other underlying assets,
745
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
seem to be more pronounced for foreign currency options. These biases are
not surprising since the model assumes a log-normal distribution for the
underlying asset with known mean and variance.
Using S&P 500 options, Dumas et al. (1998) studied the predictive
and hedging performance of a deterministic volatility function option
pricing model. They found that some option pricing models were not
better than an ad hoc procedure that merely smooths Black and Scholes
(1973) implied volatilities. They concluded that “simpler is better”. Using
a different approach, we investigate the performance of a simple model for
the valuation of options within a context of “implied liquidity premiums”.
Rubinstein (1994) documented the main following bias when testing
the Black–Scholes (1973) model.
• When pricing out of the money calls, (puts), Black–Scholes model tends
to overvalue (undervalue) these options with respect to market values.
• When pricing in the money calls (puts), Black–Scholes model tends to
undervalue (overvalue) these options with respect to market values.
• The degree of mispricing is a function of the option’s moneyness.
Dumas et al. (1998) documented a smile effect and specified four dif-
ferent models regarding the estimation of the volatility function. However,
when measuring the prediction errors, they used an “Ad hoc” strawman as
a proxy for a benchmark. Re-calling the fact that several market makers
smoothed the implied volatility relation across exercise prices to price
options, they fit the Black and Scholes model to the observed options prices.
This method allows them to operationalize the practice. This procedure is
internally inconsistent since the Black and Scholes model depends on the
assumption of a constant volatility. This “ad hoc” strawman procedure is
tested in Dumas et al. (1998), who found that it performs marginally better
than other models. The fact that asset prices do not move continuously
but rather jump from time to time, led Cox and Ross (1976) to price
options for alternative stochastic processes. In the same way, Merton (1976)
used a combination of a jump process and a diffusion process. The simple
analytic formulas given by Bellalah and Jacquillat (1995) and Bellalah
(1999) explained some of the biases reported in the literature, and in
particular, the smile effect.
We present a simple option pricing model when markets can make sud-
den jumps. The option value depends upon the probability and magnitude
of jumps and a continuous volatility. The model is useful in explaining the
smile effect. Using the market prices of at least two options on the same
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
underlying asset and maturity with different strike prices, the model can
be used to extract the market implied volatility and information regarding
the implied jumps.
The model can be applied to hedging strategies for different strike prices
and can be used for the valuation of different types of options. It can also
be used in the identification of mispriced options. Some simulations are
run with and without shadow costs of incomplete information. The option
valuation model proposed in Bellalah and Prigent (2001) and Bellalah
(2002) might help to understand why the Black and Scholes model, leads
to theoretical prices which are systematically biased and why implied
volatilities differ from one strike price to another for the same underlying
asset. For the analysis of information costs and valuation, we can refer to
Bellalah (2001), Bellalah et al. (2001a), Bellalah et al. (2001b), Bellalah
and Prigent (2001), Bellalah and Selmi (2001) and so on.
with
S
S
ln K
+ r + 12 σ 2 (T ) ln + r − 12 σ 2 (T )
K
d1 = √ , d2 = √
σ T σ T
where:
n ln(1 + h) n 2
r = r − γh + , σ 2 = σ 2 + σ ,
T T j
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
where
γ: rate at which jumps occur;
h: average jump size measured as a proportional increase in the
stock price and
σj2 : variance in the distribution of jumps.
For non-familiar readers, this is an introductive definition of the Poisson
process.
Also, when
σ 2 = δ 2 S θ−2 .
When θ < 2, this variance is a decreasing function of the asset price. When
θ = 2, the instantaneous variance of returns is δ2 and the process reduces
to that used in Black and Scholes (1973). The variance is independent of
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
the asset price level. The formula presented in Cox and Ross (1976) for the
valuation of a call at any instant of time is:
∞
S E 1
C=S g G +
n=0
n+1 n+1 2−θ
−rT S 1 E
−Ke g + G
n+1 2−θ n+1
where,
2re−rT (2−θ) [e−rT (2−θ) − 1]
S =S 2−θ
δ 2 (2 − θ)
−rT (2−θ)
2−θ 2r(e − 1)
E =E
δ 2 (2 − θ)
where the gamma density function,
e−x xm−1
g(x, m) =
Γ(m)
and
∞
G(x, m) = g(y, m)dy
x
where K is the strike price, r is the risk-less interest rate, and T is the
time to expiration. This model can be easily simulated. In fact, given
a daily variance of return and a value of θ, δ can be chosen to satisfy
σ 2 = δ 2 S θ−2 .
The above equation shows that the dynamics of the underlying asset price
corresponds to the standard diffusion process dS = µSdt + σSdX plus a
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
Π = V (S, t) − ∆S.
The change in the portfolio’s value over a small interval of time can be
written as:
∂V 1 ∂ 2V ∂V
dΠ = + σ2 S 2 2 dt + − ∆ dS
∂t 2 ∂S ∂S
+ [V (JS, t) − V (S, t) − ∆(J − 1)S]dq
where P (J) corresponds to the probability density function for the jump
size J. In the absence of jumps γ and information costs λS and λV ,
this equation becomes the standard Black and Scholes (1973) equation.
It is possible to obtain a closed-form solution to this equation when the
logarithm of J is normally distributed with a volatility σ . In this case, the
European option price is given by:
∞
1 −γ (T −t)
e (γ (T − t))n VBS (S, t; λS , λV , σn , rn ).
n=1
n!
with
nσ2
k = E[J − 1], γ = γ(1 + k), σn2 = σ2 +
T −t
and
n log(1 + k)
rn = r + λs − γk +
T −t
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
where VBS corresponds to the standard Black and Scholes (1973) price
in the absence of jumps within a context of incomplete information. This
formula shows that the option price corresponds to a sum of individual
prices where it is assumed that there are n jumps and where each price is
weighted by the probability of the occurrence of n jumps before the option
maturity date.
d = γE[(J − 1)2 ] + σ 2 .
In the absence of information costs and jumps, this equation reduces to the
classic Black–Scholes equation. When the volatility σ is zero, (the case of
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
∂V + 1 +2 2 ∂ 2 V + ∂V +
+ σ S + (r + λS ) − (r + λV )V + + γ − (V − − V + ) = 0.
∂t 2 ∂S 2 ∂S
The same methodology leads to the following equation for the value V − of
the option in the presence of the volatility σ− :
∂V − 1 −2 2 ∂ 2 V − ∂V −
+ σ S + (r + λS ) − (r + λV )V − + γ + (V + − V − ) = 0.
∂t 2 ∂S 2 ∂S
σ(τ ) = σ − + (σ + − σ − )e−ντ
with
τ : time since the last jump in volatility and
ν: a decay parameter.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
The smile effect and the S&P 500 index in the presence of jumps
Consider the implied volatilities on June 21, 1991 for the European-style
July S&P 500 index options expiring in 28 days. Table 17.1 shows the
implied volatilities and the deltas of S&P calls and puts using the Black
and Scholes (1973) model.
Table 17.1 is reproduced from Derman et al. (1991). Note that the
− sign refers to the put’s delta and the sign + refers to the call’s delta.
It is important to note that options with strike prices below the index
price or out-of-the-money (OTM) puts with low deltas are traded at higher
implied volatilities than options with strike prices above the asset price
which correspond to OTM calls with low deltas. The presence of different
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
implied volatilities for different strike prices refers to the well-known smile.
This may be viewed as an “anomaly” in the Black–Scholes model since
when using their formula, one must adjust the volatility as the strike price
changes. Besides, the fact that implied volatilities seem to be higher for puts
than calls may be a “strange” result. In fact, if market participants believe
that the underlying asset is driven by a continuous random walk, then the
volatility must be independent of the strike price. This strange result can
be explained by the fact that market participants expect an occasionally
sharp downward jump in the underlying asset price. If it were the case,
then OTM puts could show a higher probability of paying off than OTM
calls. In this case, the smile can be explained by a jump-diffusion process.
This process corresponds to a continuous diffusion which is accompanied
occasionally by a jump. The use of the Black and Scholes (1973) model
assumes that all future variations in the underlying asset value is attributed
to the continuous diffusion and none to the discontinuous jump.
The jump-diffusion process is defined by a diffusion volatility and
a probability and magnitude for the discontinuous jump. The diffusion
volatility characterizes the continuous diffusion. A small probability of a
jump of the underlying asset price in the direction of the strike price can
affect the value of an OTM option. In the presence of such process, at least
two options are used to extract information about the implied volatility
and the implied jump.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
d(1 − w) = wu.
Using the model for calls when the index can jump up
Consider the trading of index options when the current index level is 100,
r = 10%, and T = 1. Table 17.2 gives market prices and the corresponding
implied volatilities.
Using the model needs the knowledge of the upward jump and the
diffusion volatility, which are consistent with these prices. Using a simple
algorithm and formula, as in Eq. (17.1), it is possible to verify that the
upward jump u = 171.8% and the diffusion volatility is σ = 10%. The
implied probability of an upward jump to 271.8 is 1%. The downward jump
probability to 98.26 is 99%. The subsequent diffusion volatility is 10%. Note
that the Black–Scholes implied volatility in Table 17.2, σBlack−Scholes =
12.1% is higher than the implied diffusion volatility σ = 10%. Using
Eq. (17.1), Table 17.3. gives the details of the computation of call values
after the jump.
Knowing the values of w, σ, u, and d that fit the data, it is possible
to use Eq. (17.1) to value options on the same underlying and maturity
for different strike prices. Table 17.4 gives some option values and their
corresponding deltas using the model.
∗
The value of σBlack−Scholes is the implied Black–Scholes volatility
corresponding to the model value. It corresponds to the implied volatility
necessary to have the Black–Scholes formula reproduce the model price.
Note that the OTM call deltas are less than those predicted by the
Table 17.4. A comparison between Black–Scholes and model call values and
deltas in the presence of jumps.
∆ ∗
σBlack−Scholes ∆Black−Scholes
Strike Call (in %) (in %) (in %)
90 18.3 90 10.7 97
100 10.26 80 11.2 82
110 4.83 47 12.1 52
120 2.41 18 14.1 29
130 1.67 6.2 17.3 19
140 1.46 3.2 21.0 15
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
Using the model for puts when the index can jump down
Consider the trading of index options using the values as shown in
Table 17.1. The OTM puts with strike price of 370 and 365 are used to fit
the model for different levels of jump probabilities w. Table 17.5 gives the
implied percentage jump and the implied diffusion volatility using Eq. (17.1)
and the two option prices.
Table 17.5 shows for example that a 2% probability of an immediate
down jump in the underlying asset by 18.6% and a subsequent diffusion
of 12.9% can exactly reproduce the put values with strike prices 370
and 365. Table 17.6 shows the Black–Scholes implied volatilities and the
corresponding prices for the model in the presence of jumps for a jump
probability of 5%. Table 17.6 also gives the market prices and Black–Scholes
implied volatilities of Table 17.1. For example, with a 5% probability of a
10.7% downward jump, the model fits the values of OTM puts. The model
can be implemented in trading by choosing from the possible jumps that fit
the initial two options by using the trader perception of what is a reasonable
jump probability and magnitude.
1 31.4 13.1
2 18.6 12.9
3 14.3 12.7
4 12.1 12.5
5 10.7 12.2
6 9.8 12.0
7 9.1 11.8
8 8.6 11.6
9 8.2 11.4
10 7.8 11.2
11 7.5 11.0
12 7.2 10.7
13 7.0 10.5
14 6.8 10.2
15 6.7 9.9
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
Model σ Market σ
Strike (in %) (in %) Option Model–price Market–price
d(1 − w) = wu.
with
S 1 2 √
d1 = ln + r + σ + λS T σ T (17.2)
K 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
and
√
d2 = d1 − σ T (17.3)
with
S 1 2 √ √
d1 = ln + b + σ + λS T σ T, d2 = d1 − σ T .
K 2
When λS and λC are equal to zero and b = r, this formula is the same as
that in Black and Scholes.
For an index option, b = r − δ, where δ is the distribution rate. For a
foreign currency option, b = r − r∗, where r∗ is the foreign interest rate.
For a commodity, b = r − δ, where δ is the convenience yield.
The price of a European put with a strike K is:
with
S 1 √ √
d1 = ln + b + σ 2 + λS T σ T, d2 = d1 − σ T .
K 2
3 58.71 19.73
4 46.87 19.51
5 39.73 19.28
6 34.94 19.04
7 31.49 18.80
8 28.90 18.55
9 26.85 18.29
10 25.19 18.03
11 23.82 17.76
12 22.67 17.47
13 21.67 17.19
14 20.79 16.89
15 20.03 16.58
w d σdiffusion
Table 17.10. Comparison between Black and Scholes and model prices.
a possible (and reliable) way to extract information costs using option prices.
Recent evidence in 2008 and 2009 seems to confirm the results.
Example
Assume that the strike price K is in an interval [aS0 , bS0 ] with 0 < a <
1 < b, then the amount of mispricing can be appreciated with respect to
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
the position of the observed underlying asset price S0 with respect to the
strike price and the degree of moneyness of the option.
For the call option, define, for example, the function fc (S0 , K) as:
2
aS0
if S0 > K, then fc (S0 , K) = −α0 1 − SK0 with (1−a) 2 > α0 and
2
if S0 < K, then fc (S0 , K) = β0 1 − SK0 with β0 > 0.
The parameter δ0 affects the prices of ITM puts and the parameter γ0
affects OTM put prices. The call option price in the Black–Scholes model
is given by:
which is also:
with
S0 1 2 √
d1 (σ0 , gc (S0 , K)) = ln + r+ σ T σ T
gc (S0 , K) 2
√
d2 (σ0 , gc (S0 , K)) = d1 − σ T
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
which is also:
The model corrects for both biases with respect to the Black–Scholes model.
as follows:
σK,t
RvK,t = for K = −5, . . . , 5.
σ0,t
Summary
The financial crisis reveals the failure of markets and institutions in risk
management and asset pricing. The main reason for this is that models
ignored rare and extreme events as well as the importance of information.
The Black–Scholes model is the most simple and successful model in the
theory of option pricing. Many authors, however, have tried to relax some
of its assumptions in order to explain its well-known biases. These biases
are shown to be more pronounced for foreign exchange options than for
stock options. The extensions of the B–S model include constant elasticity
of variance processes and jump-diffusion processes to explain some of the
option biases. This chapter presents recent developments along these lines,
especially, the models of Merton (1976) and the Cox and Ross (1976). From
a theoretical point of view, these models are rather interesting. They point
out the difficulty of pricing assets in an incomplete market. From a practical
point of view, the use of these models are very limited given the burden of
parameter estimation to implement them. We develop a simple model for
the valuation of options in the presence of jumps and information costs.
The model is an extension of the models of Derman et al. (1991) and
Bellalah (1999). Our model has the potential to explain the smile effect.
It is calibrated to market data and allows an implicit estimation of the
magnitude of information costs. While our methodology and our model are
applied only to index options, they can be used in different option markets.
Following the approaches in Black and Scholes (1973), this chapter analyzed
and applied an option pricing model with liquidity premiums. The concept
of liquidity premiums refer to the amount of mispricing in the Black–Scholes
model. This can be defined by a function which depends on the spread
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
between the underlying asset price and the option strike price. In fact, it
is well known that the mispricing in the Black–Scholes model depends on
the degree of parity of the option. These premiums defined with respect
to the option parity can be justified on the grounds of the risk-reward
trade off as in capital asset pricing models. When “liquidity” premiums
are ignored, the model reduces to the Black–Scholes model for valuing
European options. Since the model presents an additional parameter with
respect to the Black–Scholes model, it may be easily calibrated to market
prices and may explain biases observed in Black–Scholes type models. The
simulation results for a constant volatility show that the model prices when
compared to Black–Scholes prices, do not show a “smile” of volatility. This
result is very important. In fact, the smile observed for the Black–Scholes
model for different strike prices is a surprising result since the volatility is
associated with the stock price rather than the strike price.
Questions
1. What are the specificities of the jump-diffusion model?
2. What are the specificities of the constant elasticity of variance model?
3. What is the empirical evidence regarding the volatility smiles?
4. Describe briefly the main results in the Merton’s (1976) jump-diffusion
model.
5. Describe the main results in the Cox and Ross (1976) constant elasticity
of variance model.
6. Describe briefly the main results regarding the pricing and hedging of
options in the presence of jumps and information costs.
7. Describe briefly the main results in the models for the valuation of
options in the presence of jumps and information costs.
8. Describe briefly the main results regarding implied volatility functions
and option pricing models.
References
Barone-Adesi, G and RE Whaley (1987). Efficient analytic approximation of
American option values. Journal of Finance, 42 (June), 301–320.
Bellalah, M (1999). The valuation of futures and commodity options with
information costs. Journal of Futures Markets, 19 (September), 645–664.
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13(3),
1895–1927.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch17
Chapter 18
Chapter Outline
This chapter is organized as follows:
1. Section 18.1 presents the Hull and White (1987) model, which is one
of the simplest models of option pricing with stochastic volatilities.
It develops the main results of the Stein and Stein (1991) model.
A generalization of derivative asset pricing models to a context of
stochastic volatilities within complete and incomplete markets is pre-
sented. In particular, the main results in the models of Heston (1993)
and Hoffman et al. (1992) are reviewed. It is important to note that an
incomplete market is simply a market in which there is not a unique
equivalent martingale measure or risk-neutral probability for the asset
price.
2. Section 18.2 develops a framework for option pricing in the presence of
a stochastic volatility and information costs.
3. Section 18.3 is devoted to the theory of option pricing biases and in
particular the “smile effect”.
4. Section 18.4 presents some empirical evidence regarding option pricing
within information uncertainty and stochastic volatility.
771
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
Introduction
Volatility is an important parameter in option pricing theory. Black and
Scholes (1973) proposed an option-valuation equation under the assumption
of a constant volatility in a complete market without frictions. Engle (1982)
developed a discrete-time model, to show that the volatility depends on
its previous values. Many papers have tried to relax the most stringent
assumptions in the Black–Scholes (B–S) theory and in particular, the
constant volatility. Much of the known bias reported in empirical studies
based on the B–S formula has something to do with this assumption. The
biases reported in Rubinstein (1985, 1994) for stock options, Melino and
Turnbull (1990, 1991) and Knoch (1992) for options on other underlying
assets, seem to be more pronounced for foreign currency options. These
biases are not surprising since the model assumes a log-normal distribution
for the underlying asset with known mean and variance. The stochastic
volatility problem has been examined by several authors. For example,
Hull and White (1987), Wiggins (1987), and Johnson and Shanno (1987)
studied the general case in which the instantaneous variance of the stock
price follows some geometric processes. Scott (1989) and Stein and Stein
(1991) used an arithmetic volatility in the study of option pricing. All
these models describe (with precision) the effects of the volatility on
the options prices. Stein and Stein (1991) and Heston (1993) proposed a
dynamic approach for the volatility, which is represented by an Ornstein–
Ulhenbeck. It is difficult to find an analytic solution for the stochastic
volatility option-pricing problem. Merton (1987) proposed a capital-asset
pricing model in the presence of the shadow costs of incomplete information.
Bellalah (1990) applied the Merton (1987) model to the valuation of
options under incomplete information. Bellalah and Jacquillat (1995) and
Bellalah (1999) derived the Black and Scholes (1973) equation in the
context of Merton’s (1987) model. They obtained another version of the
Black and Scholes equation within information uncertainty. Options can be
valued in the context of Merton’s (1987) “Simple model of capital market
equilibrium with incomplete information”. We provide the derivation of the
partial differential equation for options in the presence of shadow costs of
incomplete information and stochastic volatility. We illustrate our approach
by specific applications and show the dependancy of the option price on
information and stochastic volatility. As in Platen and Schweizer (1992),
we show that the investor’s choice of the minimal equivalent martingale
measure is not changing, but the process of the price of the asset depends
on incomplete information. Bellalah et al. (2001) carried out an empirical
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
where St stands for the stock price at date t, νt = σt2 is the instantaneous
variance, and Wt1 and Wt2 are Brownian motions correlated with a
correlation coefficient ρ.
In this context, the pricing of a call option implies the construction of a
risk-less portfolio containing the option, the stock, and a second call option
with the same strike price but a different time to maturity.
Let c(t, St ) be the value of the first call option which depends on time t
and the stock price St . This approach, which was also used by Johnson
and Shanno (1987), Scott (1987), and Wiggins (1987), yields a partial
differential equation for the option price. However, the solution to this
equation is not unique unless the price function for the second call is known.
To obtain a unique solution, Hull and White (1987) made the additional
assumptions that the two Wiener processes are independent and that the
variance has no systematic risk. This convenient assumption implies that
the volatility risk does not entail any risk premium. With these additional
assumptions, it is possible to obtain a unique option price, computed as
the expectation of the discounted terminal pay off under a risk-neutral
probability measure. The terminal boundary condition at the maturity date
T is c(T, S) = ST − K, if ST ≥ K and c(T, S) = 0 otherwise.
Using the additional assumption that the variance ν is not influenced
by the stock price and setting
T
∗ 1
νt,T = νs ds
(T − t) t
the value of a European call option when the volatility is uncorrelated with
the stock price is:
∞
∗
∗
c(t, St , σt ) =
2
cB−S t, St , νt,T dF νt,T | St , σt2 (18.3)
0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
where cB−S is the usual B–S option price corresponding to the variance,
∗ ∗
νt,T and F is the conditional distribution of νt,T under the risk-neutral
∗
probability measure, given St and σt . The value of cB−S (t, St , νt,T
2
) is
given by:
∗
cB−S t, St , νt,T = St N (d1 ) − Ke−rT N (d2 )
S ∗
ln K + r + 12 νt,T (T − t) ∗ (T − t).
d1 = , d2 = d1 − νt,T
∗
νt,T (T − t)
equation:
where µ and σ stand for the expected instantaneous return and variance,
respectively and W1 is a Wiener process.
Let the dynamics of the volatility be governed by an arithmetic
Ornstein–Uhlenbeck process, with a tendency to revert back to a long-run
average level θ as follows:
This process has been used by many researchers in modeling and studying
the empirical properties of volatility. Examples include Poterba and
Summers (1986), Stein (1989), and Merville and Pieptea (1989) among
others. From Eqs. (18.4) and (18.5), it is possible to obtain an explicit
closed-form solution for the stock price distribution. Setting P0 equal to
one and denoting by S0 (P, t) the time t stock price distribution when the
stock price drift µ = 0, the solution given by Stein and Stein (1991) is:
1 −3 ∞ 1 t iη ln(P )
S0 (P, t) = P 2 η2 + e dη.
2π η=−∞ 4 2
When the stock has a non-zero drift, µ different from zero, the solution is:
where the integral I(.) is calculated using Eqs. (18.3) to (18.5) in Stein and
Stein (1991). The European call price is:
∞
−rt
C0 (P, t) = e [P − K]S(P, t | δ, r, κ, θ)dP.
P =K
Using this model implies the choice of reasonable values for the
parameters δ, κ, and θ. Results of the empirical studies by Stein (1989)
and Merville and Pieptea (1989) show that reasonable values are θ = 0.3,
δ = 16, and κ = 0.4. Simulations of the Stein and Stein (1991) model show
that stochastic volatility exerts an upward influence on all option prices.
Also, the stochastic volatility is more important for OTM or in-the-money
(ITM) options than for ATM options, i.e., the implied volatilities in the
context of this model exhibit a U-shape as the strike price is varied. Hence,
an ATM option has the lowest implied volatility, and this volatility rises in
either direction as the strike price moves.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
where p1 and p2 are probabilities which can be calculated using the following
formula:
iφ ln(k)
1 1 ∞ e fj (x, ν, T, φ)
pj (x, ν, T, ln(K)) = + Re dφ
2 π 0 iφ
for j = 1, 2, with fj (x, ν, T, φ)
given by Eqs. (18.16) and (18.17) in Stein and Stein (1991). The probabil-
ities in Eq. (18.7) multiplying the asset price S and the strike price K in
Eq. (18.6) must be calculated to obtain the option price.
Following Merton (1973a) and Ingersoll (1990), Heston (1993) incor-
porated stochastic interest rates in his option pricing model and applied
it to options on bonds and options on foreign currencies. This is possible
if one modifies the dynamics of the asset to allow time dependence in the
volatility of the spot asset, σs (t):
dSt = µs Sdt + σs (t) ν(τ )SdWt1 .
It is convenient to note that variances of the spot asset and the bond are
given by the same variable ν(t). In this context, the valuation equation is
given by Eq. (18.22) in Heston (1993a). The model can also be applied to
options on foreign currencies. In fact, when S(t) stands for the dollar price
of a foreign currency and the dynamics of the foreign price of a foreign
discount bond, F (t, T ), are given by the following equation:
dF (t, T ) = µP F (t, T )dt + σP (t) ν(τ )F (t, T )dWt2
then Eq. (18.26) in Heston (1993) can be used for the pricing of European
currency options. The model can be used for the valuation of stock options,
bond options, and currencyoptions. It is interesting since it links most biases
in option prices to the dynamics of the spot price and the distribution
of spot returns. With a proper choice of its parameters, the stochastic
volatility model seems to be flexible and promising in the description of
option prices. However, the main drawback of this model is the number of
parameters to be estimated.
It may be noted that such a model goes back to Merton (1971, 1973, 1990).
In this formulation, an option or a contingent claim is a random variable
of the form c(XT ). Within this general formulation, there is no unique
equivalent martingale measure for the stock price process. This leads to
the problem of pricing in incomplete markets. To overcome this problem,
and to be able to compute option prices, Hoffman et al. (1992) supposed
that there is a “small” investor who should use the minimal equivalent
martingale measure. This allows them to decompose the space of all those
assets compatible with the given stock is a direct sum of two subspaces:
purely traded assets and totally non-tradable assets. The flexibility and
generality of the model give rise to stochastic differential equations that
do not have explicit solutions. Hoffman et al. used numerical methods to
compute option prices and simulate the performance of hedging strategies
under various possible scenarios. Such a model is rich from a theoretical
point of view but is “poor” from a practical perspective, since it is rather
difficult to implement by market participants. The main drawback is the
number of parameters to be estimated.
dS = µSdt + σSdW1
The two processes dW1 and dW2 are Brownian motions with a correlation
coefficient ρ. The functions p(S, σ, t) and q(S, σ, t) are specified in a way
that fits the dynamics of the volatility over time. Hence, the derivative
asset price V (S, σ, t) can be a new source of randomness that cannot be
easily hedged away. The pricing of options in this context needs the search
for two hedging contracts. The first is the underlying asset. The second can
be an option that allows a hedge against volatility risk. Following the same
logic as in the original Black–Scholes model (1973), consider a portfolio
comprising a long position in the option V , a short position of ∆ units of
the underlying asset, and a short position of −∆1 units of an other option
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
pi = V − ∆S − ∆1 V1 . (18.8)
Over a short interval of time dt, applying Ito’s lemma for the functions S, σ,
and t gives the change in the value of this portfolio as:
∂V 1 ∂ 2V ∂ 2V 1 ∂2V
dΠ = + σ2 S 2 2 + ρσqS + q 2 2 dt
∂t 2 ∂S ∂S∂σ 2 ∂σ
∂V1 1 2 2 ∂ V1
2
∂ 2 V1 1 2 ∂ 2 V1
− ∆1 + σ S + ρσqS + q dt
∂t 2 ∂S 2 ∂S∂σ 2 ∂σ 2
∂V ∂V1 ∂V ∂V1
+ − ∆1 ∆ dS + − ∆1 dσ.
∂S ∂S ∂σ ∂σ
All the sources of randomness in the portfolio value resulting from dS can
be eliminated by setting the quantity before dS equal to zero, or ∂V∂S
−∆−
∆1 ∂V
∂S
1
= 0 and also by setting the quantity before dσ equal to zero, or
∂V ∂V1
∂σ
− ∆1 ∂σ
= 0.
After eliminating the stochastic terms, the terms in dt must yield the
deterministic return as in a B–S “hedge” portfolio. Hence, the instantaneous
return on the portfolio must be the risk-free rate plus information costs on
each asset in the portfolio as shown in Bellalah (1999). This gives:
∂V 1 ∂2V ∂ 2V 1 2 ∂ 2V
dΠ = dt + σ2 S 2 dt + ρσSq dt + q dt
∂t 2 ∂S 2 ∂S∂σ 2 ∂σ 2
∂V1 1 2 2 ∂ 2 V1 ∂ 2 V1 1 2 ∂ 2 V1
− ∆1 dt + σ S dt + ρσSq dt + q dt
∂t 2 ∂S 2 ∂S∂σ 2 ∂σ 2
= [(r + λV )V − (r + λS )∆S − (r + λV1 )∆1 V1 ]dt.
2 2 2
∂V
∂t + 12 σ 2 S 2 ∂∂SV2 + ρσSq ∂S∂σ
∂ V
+ 12 q 2 ∂∂σV2 + (r + λS )S ∂V
∂S − (r + λV )V
∂V
∂σ
2 2 2
∂V1
∂t + 12 σ 2 S 2 ∂∂SV21 + ρσSq ∂S∂σ
∂ V1
+ 12 q 2 ∂∂σV21 + (r + λS )S ∂V
∂S − (r + λV1 )V1
1
= ∂V1
.
∂σ
Since the two options differ by their strikes, payoffs, and maturities, this
implies that both sides of the equation given above are independent of the
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
contract type. Since both sides are functions of the independent variables
S, σ, and t, we have:
∂V 1 ∂ 2V ∂2V 1 2 ∂ 2V
+ σ2 S 2 + ρσSq + q
∂t 2 ∂S 2 ∂S∂σ 2 ∂σ 2
∂V ∂V
+ (r + λS )S − (r + λV )V = −(p − δq) ,
∂S ∂σ
for a function δ(S, σ, t) referred to as the market price for risk or volatility
risk. This equation can also be written as
∂V 1 ∂2V ∂2V 1 ∂ 2V ∂V
+ σ 2 S 2 2 + ρσSq + q 2 2 + (r + λS )S
∂t 2 ∂S ∂S∂σ 2 ∂σ ∂S
∂V
+ (p − δq) − (r + λV )V = 0. (18.9)
∂σ
∂V ∂V
This equation shows two hedge ratios ∂S
and ∂σ
. The term (p − δq) is
known as the risk-neutral drift rate.
dΠ − [(r + λV )V − (r + λS )∆S]dt
∂V 1 ∂ 2V ∂2V 1 2 ∂ 2V ∂V
= + σ2 S 2 + ρσqS + q + (r + λS )S
∂t 2 ∂S 2 ∂S∂σ 2 ∂σ 2 ∂S
∂V ∂V
− (r + λV )V ] dt + dσ = q (δdt + dW2 ).
∂σ ∂σ
This results from Eqs. (18.8) and (18.9). The term dW2 represents a unit
of volatility risk. There are δ units of extra return, given by dt for each unit
of volatility risk.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
∂V 1 ∂ 2V ∂V
+ σ2 S 2 + (µ − δS σ)S − (r + λV )V = 0. (18.10)
∂t 2 ∂S 2 ∂S
If the asset is traded, then V = S must be a solution to Eq. (18.10)
Substituting V = S in the Eq. (18.10) gives
(µ − δS σ)S − (r + λS )S = 0.
The market price of risk for a traded asset in the presence of information
costs δS = µ−(r+λ
σ
S)
. Substituting δS in (21) gives the following equation:
∂V 1 ∂ 2V ∂V
+ σ2 S 2 2 + (r + λS )S − (r + λV )V = 0.
∂t 2 ∂S ∂S
This is the B–S equation in the presence of information costs.
where St denotes the stock price at time t, νt = σt2 its instantaneous vari-
ance, r the risk-less interest rate, which is assumed to be constant, and λBt
is the shadow cost related to Bt . W 1 and W 2 are Brownian motions under
P , they are independent νt has no systematic risk. This yields a unique
option price which can be computed as the (conditional) expectation of the
discounted terminal payoff under a risk-neutral probability measure P̃ . Put
differently, P̃ is obtained from P by means of a Girsanov transformation
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
such that:
with
case, to:
∂V 1 ∂ 2V ∂2V 1 ∂ 2V
+ σt2 St2 2 + ρσt2 θSt + θ2 σt2
∂t 2 ∂St ∂St ∂σt 2 ∂σt2
∂V ∂V
+ (r + λSt )St + (f (σt ) − δθσt ) − (r + λV )V = 0.
∂St ∂σt
As in Wiggins (1987), we can write the following equation:
The Weiner processes dW1 and dW2 are uncorrelated. When Eq. (18.8) is
applied in this context, we have:
∂V 1 ∂ 2V 1 ∂ 2V ∂V
+ σt2 St2 2 + k 2 + (r + λSt )St
∂t 2 ∂St 2 ∂σt2 ∂St
∂V
+ [−(σt − θ) − δk] − (r + λV )V = 0.
∂σt
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
When δ = 0 or a constant, this equation has a solution with the same form
as given in Stein and Stein (1991). The solution depends on information
costs of V and the underlying asset S. The option price has the following
form:
∞
V = e−(r+λV )T [St − K]H(St, t | , r + λSt , k, θ)dSt
St =K
with H(St , t) is the price distribution of the underlying asset at the time t
with a non-zero drift of St .
with ρ the correlation coefficient between dWt1 and dWt2 . In this case, the
value of any option V (St , νt , t) must satisfy the following partial differential
equation:
∂V ∂ 2V ∂2V 2 1∂ V ∂V
2
1
+ ρσt νt St + νt St2 2 + σt νt + (r + λSt )St
∂t ∂St∂νt 2 ∂St 2 ∂νt2 ∂St
√ ∂V
+ [κ(θ − νt ) − δσt νt ] − (r + λV )V = 0 (18.12)
∂νt
V (S, νt , t) = max(o, S − K)
V (0, νt , t) = 0
∂V
(∞, νt , t) = 1
∂St
∂V ∂V ∂V
(r + λSt )St + κ(θ) − (r + λV )V + =0
∂St ∂νt ∂t
V (S, ∞, t) = S.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
with P (t, t+τ ) = e−(r+λSt )τ , the price at time t of a unit discount bond that
matures at time t + τ . The first term of the right-hand side of the solution
V (S, νt , t) is the present value of the underlying asset upon optimal exercise.
The second term is the present value of the strike price. Both of these
terms must satisfy the Eq. (18.9). It is convenient to write them in terms
of the logarithm, x = ln(S). By substituting the solution of Eq. (18.10) in
Eq. (18.9), P1 and P2 must satisfy the following equation:
∂Pj 1 ∂2V ∂ 2V 1 ∂ 2 Pj
+ νt + ρσt νt St + σt2 νt
∂t 2 ∂x ∂St ∂νt 2 ∂νt2
∂Pj √ ∂Pj
+ (r + λSt + uj νt ) + (a − bj νt ) =0
∂x ∂νt
√
for j = 1, 2, where u1 = 1/2, u2 = −1/2, a = κθ, b1 = (κ − ρσt ) νt + δσt ,
√
and b2 = κ νt + δσt .
Following the same resolution method as followed in Heston (1993), we
obtain the solution of the characteristic function:
when
a 1 − gedτ
C(τ ; φ) = i(r + λSt )φτ + (bj − iρσt φ + d)τ − 2 ln
σt2 1−g
bj − iρσt φ + d 1 − edτ
D(τ ; φ) =
σt2 1 − gedτ
and
bj − iρσt φ + d
g= , d= (iρσt φ − bj )2 − σt2 (2iuj φ − φ2 ).
bj − iρσt φ − d
By inverting the characteristic functions fj , we obtain the desired proba-
bilities:
−iφ ln K ∗
1 1 ∞ (e ) fj (x, νt , T ; φ)
Pj (x, νt , t; ln K) = + Re dφ
2 π 0 iφ
with dWt1 dWt2 = ρdt. When Eq. (18.8) is applied to this model, we obtain:
∂V 1 ∂ 2V ∂2V 1 ∂2V
+ σt2 St2α + ρσt3 Stα σσt + σt2 σσ2t
∂t 2 ∂St 2 ∂S∂σt 2 ∂σt2
∂V ∂V
+ (r + λSt )St + (µσt − δσσt )σt − (r + λV )V = 0
∂St ∂σt
Johnson and Shanno (1987), supposed that the risk premium of the
µ
volatility is zero. Consequently, we have: δ = σσσt .
t
where:
n: the number of bid-ask price quotes sampled on a prior day;
COBS : the observed call price and
CBS ()ISD : the theoretical B–S call price computed with the ISD parameter.
Theoretical B–S prices are obtained with the ISD we computed and
which is estimated from one-day lagged price observation. We measure
option moneyness relative to the price deviations. Option moneyness is
calculated as (Ke−rt − SADJ )/Ke−rt, where SADJ is the dividend-adjusted
cash price and Ke−rt , the discounted strike price. Note that negative
(positive) moneyness corresponds to ITM (OTM) call options with low
(high) strike prices. We see that the B–S model undervalues strongly
low OTM calls and overvalues slightly high OTM calls. To compute the
information cost-adjusted B–S option price model, we estimate the ISD,
the option information cost λC , and the underlying information cost λS ,
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
CMj (ISD , IλC , IλS ) is the call price of the proposed model computed with
the ISD parameter which includes the implied information costs parameters,
and the cost of carrying the commodity. This call price is calculated for
any option in a given current day’s sample. We want to go further into the
analysis of the differences between the observed price and the theoretical
price that could be computed by the B–S model or the information cost
model. We use the mean of absolute forecast error (MAE) and the mean
absolute percentage forecast error (MAPE). Table 18.1 gives the results
obtained by the MAE and MAPE tests. These tests pertain to both call
and put options.
We clearly see that the information cost model proposed by Bellalah
(1999) performs better than the B–S model.
Call Put
In this case, Renault and Touzi (1996) have shown that the shape of the
volatility structure with respect to the moneyness of the option is symmetric
when the returns innovations and the volatility are uncorrelated. In the
presence of information costs without stochastic volatility, the value of a
European call is given by the formula given in Bellalah (1999). It can be
said that:
(1) The option price is equal to the standard B–S price with a new risk-less
rate equal to (r + λS ) multiplied by the discount factor e(−(λC −λS ))τ .
(2) In the same way, the option price is equal to the standard Black and
Scholes price with a new stock price equal to SeλS τ multiplied by the
discount factor e−λC τ . Longstaff (1995) documented the evidence of
implicit stock prices greater than by about half percent in mean. Even
though Longstaff (1995) still computed B–S implicit stock prices, it is
clear that this argument can be extended to Hull and White (1987)
pricing:
T
−λC τ BS λS τ 1
C=e E C Se , σu du
2
.
T −t t
when one computes B–S implicit volatilities with S (ignoring the implied
S taken into account by the option market). They show that the shape
and the order of magnitude of resulting skewness, is conformable to well-
documented empirical evidence.
Summary
Jarrow and Eisenberg (1991) and Stein and Stein (1991) built their models
on the assumption that volatility is uncorrelated with the spot asset. They
obtained solutions that looked like an average of B–S values over different
paths of volatility. The absence of this correlation implies that the model
cannot capture important skewness effects. Heston (1993) used a new
technique to derive a closed-form solution for a European call option in
a stochastic volatility context. His model allows for arbitrary correlation
between the spot asset returns and the volatility. Also, he introduced
stochastic interest rates and applied the model to the pricing of bond
options and foreign currency options. Hoffman et al. (1992) considered a
very general diffusion model for asset prices allowing the description of
stochastic and past-dependent volatilities. Since their model implied an
incomplete market, they were unable to get analytical solutions. They
used stochastic numerical methods for the study of option prices and
hedging strategies. When the underlying asset dynamics are believed to
be stochastic, Hull and White (1987), Stein and Stein (1991), and Heston
(1993) among others, proved that the implied volatility may vary with
the option’s strike price. Simulations of their models show that a plot of
theoretical implied against strike prices display a U-shaped curve, (smile
effect). Although there have been several models to explain the strike price
bias and the smile effect, only little empirical work has been done by Shastri
and Wethyavivom (1987), Fung and Hsieh (1991) and Xu and Taylor (1994).
This chapter presents recent developements along these lines, especially, the
models of Merton (1976), the Cox and Ross (1976), Hull and White (1988),
Stein and Stein (1991), Heston (1993), and Hofman et al. (1992). From a
theoretical point of view, these models are rather interesting. They point
out the difficulty of pricing assets in an incomplete market. From a practical
point of view, the use of these models is very limited given the burden of
parameter estimation to implement them. We develop a general context for
the valuation of options with stochastic volatility and information costs.
The shadow costs are integrated in the investor’s portfolio wealth process
in the same vein as in Merton (1987), Bellalah and Jacquillat (1995), and
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
Bellalah (1999). We carry out an empirical test for the valuation model of
options in the presence of information costs proposed in Bellalah (1999).
We examine a way of extending the B–S model in order to take into
consideration the biases implied by the nonintegration of information costs
in the valuation model. We test the statistical performance of the models
used. We give an explanation of the skewness observed in the smile on
the basis of our model. We find that taking into account the information
costs in the option valuation formula, enhances the accuracy of the
valuation.
Questions
1. How can we proceed to price options in a stochastic economy?
2. What are the basic concepts behind Stein and Stein’s model?
3. What are the basic concepts behind Heston’s model?
4. What are the basic concepts behind HPS’s model?
5. What are the basic concepts behind Heston’s model?
6. What are the theoretical reasons for the existence of the volatility smiles?
7. What is the empirical evidence regarding the volatility smiles?
References
At-Sahalia, Y and A Lo (1998). Nonparametric estimation of state-price densities
implicit in financial asset prices. Journal of Finance, 53, 499–548.
At-Sahalia, Y and A Lo (2000). Nonparametric risk management and implied
risk-aversion. Journal of Econometrics, 94, 9–51.
Bates, DS (1997). Jumps and stochastic volatility: exchange rate processes
implicit in Deutsche mark options. Review of Financial Studies, 9, 69–107.
Bellalah, M (1990). Quatre essais pour l’valuation des options sur indices et sur
contrats terme d’indice. Doctorat de l’Universite de Paris-Dauphine.
Bellalah, M (1999). The valuation of futures and commodity options with
information costs. Journal of Futures Markets, 19 (September), 645–664.
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13(3),
1895–1927.
Bellalah, M and B Jacquillat (1995). Option valuation with information costs:
theory and tests. The Financial Review, 30(3), 617–635.
Bellalah, M and J-L Prigent (2001). Pricing standard and exotic options in the
presence of a finite mixture of Gaussian distributions. International Journal
of Finance, 13(3), 1975–2000.
Bellalah, M and F Selmi (2001). On the quadratic criteria for hedging under
transaction costs. International Journal of Finance, 13(3), 2001–2020.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
Bellalah, M and WU Zhen (2002). A model for market closure and international
portfolio management within incomplete information. International Journal
of Theoretical and Applied Finance, 5(5), 479–495.
Bellalah, M, JL Prigent and C Villa (2001a). Skew without skewness: asymmetric
smiles; information costs and stochastic volatilities. International Journal of
Finance, 13(2), 1826–1837.
Bellalah, M, Ma Bellalah and R Portait (2001b). The cost of capital in
international finance. International Journal of Finance, 13(3), 1958–1973.
Black, F (1976). The pricing of commodity contracts. Journal of Financial
Economics, 3, 167–179.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
Cox, JC and SA Ross (1976). The valuation of options for alternative stochastic
processes. Journal of Financial Economics, 3, 145–166.
Day, ET and CM Lewis (1992) (in press). Stock market volatility and the
information content of stock index options. Journal of Econometrics,
52, 115–128.
Engle, RF (1982). Autoregressive conditional heteroskedasticity with estimates
of the variance of U.K. inflation. Econometrica, 50, 987–1008.
Geske, R (1979). A note on an analytical valuation formula for unprotected
American call options with known dividends. Journal of Financial Eco-
nomics, 7, 375–380.
Ghysels, E, A Harvey and E Renault (1996). Stochastic volatility. In
Statistical Methods in Finance, CR Rao and GS Muddala (eds.), pp. 119–191.
North-Holland, Amsterdam: Elsevier.
Heston, S (1993). A closed-form solution for options with stochastic volatility
with applications to bond and currency options. The Review of Financial
Studies, 6, 327–343.
Hofmann, N, E Platen and M Schweizer (1992). Option pricing under incomplete-
ness, Mathematical Finance, 2(3), (July), 153–187.
Hull, J and A White (1987). The pricing of options on assets with stochastic
variables. The Journal of Finance, 42, 281–300.
Hull, J and A White (1988). An analysis of the bias in option pricing caused by
a stochastic volatility. Advances in Futures and Options Research, 3, 29–61.
Jackwerth, JC and M Rubinstein (1996). Recovering probability distributions
from contemporaneous security prices. Journal of Finance, 51, 1611–1631.
Jarrow, RA and LK Eisenberg (1991). Option pricing with random volatilities in
complete markets. Working Paper, Federal Reserve Bank of Atlanta.
Johnson, H and D Shanno (1987). Option pricing when the variance is changing.
Journal of Financial and Quantitative Analysis, 22, 143–151.
Knoch, HJ (1992). The pricing of foreign currency options with stochastic
volatilities. PhD dissertation, Yale School of Organization and Management.
Longstaff, FA (1995). Option pricing and Martingale restriction. Review of
Financial Studies, 8, 1091–1124.
Melino, A and SM Turnbull (1990). Pricing foreign currency options with
stochastic volatility. Journal of Econometrics, 45, 239–265.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch18
Part VII
799
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
800
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
Chapter 19
Chapter Outline
This chapter is organized as follows:
Introduction
The pricing of derivative assets is usually based upon two methods, which
use the same basic arguments. The first method involves the resolution of
801
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
In this chapter, we “mix” some of the ideas underlying these models with
the observed behavior of the volatility and interest rates for the valuation of
long-term options. The analysis is based on the results of numerous studies,
where it is shown that the variance of a portfolio of assets is simultaneously
a function of the variance of the firms cash flows and the interest-rate
variance. Hence, it is possible to provide a decomposition of the market
portfolio volatility into two components. The first is specific to the basket
of stocks. The second is linked to the variance of interest rates. We study
the implications of this decomposition on the pricing of index options and
index futures options and their values in a model with a composite volatility
and stochastic interest rates. In particular, the volatility of interest rates
may trigger an early exercise of American stock index options.
Using the principal results in the literature, a model is derived for
the valuation of American long-term index options and index futures
options. The model is a two state, with the values of the underlying index
and interest rates as state variables. We develop a stable and convergent
numerical scheme for the solutions for American index options and index
futures options in this context.
∂ 2c 1
= 2 [c(i + 1, j) + c(i − 1, j) − 2c(i, j)] (19.5)
∂S 2 h
∂c
The partial derivative with respect to time, ∂t
, can be approximated at
point (i, j) by the forward difference:
∂c 1
= [c(i, j + 1) − c(i, j)] (19.6)
∂t k
The computation of the different partial derivatives is a first step in
numerical analysis. The second step requires to replace these partial
derivatives by their corresponding formulas in the extended Black–Scholes
PDE. Hence, for i = 1 to (N − 1) and j = 0 to (M − 1), the PDE can be
written as:
1 2 2 2 1
rc(i, j) = σ i h 2 [c(i + 1, j) + c(i − j, j) + c(i − 1, j) − 2c(i, j)]
2 h
1
+ rih [c(i + 1, j) − c(i − 1, j)]
2h
1
+ [c(i, j + 1) + c(i, j)] (19.7)
k
This system can be re-written as:
c(S, T ) = max[0, ST − K]
for i = 1, 2, 3, . . . , N .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
∂c 1
= [c(i + 1, j + 1) − c(i − 1, j + 1)] (19.11)
∂S 2h
for the option partial derivative with respect to the state variable:
∂ 2c 1
= 2 [c(i + 1, j + 1) − c(i − 1, j + 1) − 2c(i, j + 1)] (19.12)
∂S 2 h
and
∂c 1
= [c(i, j + 1) − c(i, j)] (19.13)
∂t k
for the option partial time derivative. As before, these partial derivatives
must be replaced in the PDE in order to obtain the following system:
with
1 1 1 2 2
∗
ai = − rik + σ i k
(1 + rk)
2 2
1
b∗i = 1 + σ 2 i2 k
(1 + rk)
1 1 1 2 2
rik + σ i k
ci =
(1 + rk) 2 2
1 2 2 ∂ 2c ∂c ∂c
σ S + rS + − rc = 0
2 ∂S 2 ∂S ∂t
under the following terminal or boundary condition c(S, T ) =
max[0, ST − K].
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
1 2 2 ∂ 2c ∂c ∂c
σ S 2
+ (r + λS )S + − (r + λc )c = 0
2 ∂S ∂S ∂t
under the following terminal condition which must be satisfied by the call
price at its maturity date c(S, t∗ ) = max[0, St∗ − K].
The valuation of a put option consists in finding a solution to the
following PDE:
1 2 2 ∂ 2p ∂p ∂p
σ S + (r + λS )S + − (r + λp )p = 0
2 ∂S 2 ∂S ∂t
under the following terminal condition which must be satisfied by the put
price at its maturity date p(S, t∗ ) = max[0, K − St∗ ].
The appendix is proposed as an exercise to provide the analytic solution
to this problem.
or a backward difference:
∂c 1
= [c(i, j) − c(i − 1, j)].
∂S h
∂c 1
= [c(1 + i, j) − c(i − 1, j)]
∂S 2h
∂2 c
The term ∂S 2 is approximated by:
∂ 2c 1 ∂c(S + h, T ) ∂c(S, T )
= − (19.16)
∂S 2 h2 ∂S ∂S
∂c
When the partial derivative of ∂S
is replaced, Eq. (19.16) gives:
∂ 2c 1
= 2 [c(i + 1, j) + c(i − 1, j) − 2c(i, j)]. (19.17)
∂S 2 h
∂c
The partial time derivative of ∂t
can be approximated by:
∂c 1
= [c(i, j) − c(i, j − 1)] (19.18)
∂t k
1 2 2 2 1
0= σ i h 2 [c(i + 1, j) + c(i − 1, j) − 2c(i, j)]
2 h
1
+ rih [c(i + 1, j) − c(i − 1, j)]
2h
1
+ [c(i, j) − c(i, j − 1)] − rc(i, j) (19.19)
k
system:
1 2 2 ∂ 2c ∂c ∂c
σ S + (r + λS )S + − (r + λc )c = 0
2 ∂S 2 ∂S ∂t
where:
The dynamics of the spot interest rates are given by the familiar square
root process, which has correlation ρ with dzS , i.e., cov(dzS , dzr ) = ρdt. In
the Brenner et al. model, (1987) the effect of the volatility of interest rates
is explicitly taken into account. The dynamics of the index price are given
by the following equation:
where δ stands for the dividend yield on the index price, σI is the specific
index volatility, σr is the interest-rate volatility and ν is a coefficient
which transmits the effect of interest-rates volatility to the index volatility.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
where dzS has correlation ρ with dzr , i.e., cov(dzS , dzr ) = ρdt.
This formulation is similar in spirit to that by Brenner et al. (1987). In
the above formulation, αS stands for the expected instantaneous relative
price change of the underlying commodity asset. When there are no
arbitrage opportunities, the assumption of a constant proportional cost of
C(S, r, T ) = (S − K)+
C(S, r, t) = (S − K)+
∂C(S, r, t)
=1 (19.27)
∂S
The last condition is the usual “smooth fit” principle. An American stock
index put option with a strike price K and a maturity date T must satisfy
the PDE Eq. (19.26) subject to the following boundary conditions:
P (S, r, T ) = (K − S)+
P (S, r, t) = (K − S)+
∂P (S, r, t)
= −1 (19.28)
∂S
are run. The numerical scheme is a θ scheme of the implicit type for which
θ = 1/2. It is centered in space and in time. It is unconditionally stable and
convergent.
Numerical solutions
The time to maturity (T − t) is divided into N time intervals of length
k. The option value is calculated at time (s − k) in a recursive way as
a function of its value at instant s with t ≤ s ≤ T . The instant s = T
corresponds to the option’s maturity date. The underlying index price and
the interest rate are divided into M intervals of size h. The state variables
are considered within the intervals [0, S∗] and [0, r∗]. Note that the larger
the values of M and N , the closer is the numerical solution of the discrete
system to the real solution of the PDE. Hence, using:
S = ih for 0 ≤ i ≤ M
r = jh for 0 ≤ j ≤ M
t = nk for 0 ≤ n ≤ N (19.29)
∂U n (i, j) U n (i + 1, j) − U n (i − 1, j)
=
∂S 2h
∂U n (i, j) U n (i, j + 1) − U n (i, j − 1)
=
∂r 2h
∂ 2 U n (i, j) U n (i − 1, j) − 2U n(i, j) + U n (i + 1, j)
2
=
∂S h2
∂ 2 U n (i, j) U n (i, j − 1) − 2U n(i, j) + U n (i, j + 1)
=
∂r 2 h2
U n (i + 1, j + 1) − U n (i − 1, j + 1)
∂ 2 U n (i, j) − U n(i + 1, j − 1) + U n (i − 1, j − 1)
=
∂S∂r 4h2
∂U n (i, j) U n (i, j) − U n − 12 (i, j)
=
∂t k/2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
+ 4U n (i, j) − 2U n (i − 1, j) + U n (i + 1, j − 1) − 2U n (i, j − 1)
+ U n (i − 1, j − 1)
3 Itis possible to show that this system of (M − 1)2 equations may be solved in two steps,
given a value of n. For each i, it presents (M − 1) equations with (M + 1) unknowns
for 0 ≤ i ≤ M . After calculating the values U n∗ (i + 1, j − 1), the second step implies
to solve a system of (M − 1) equations with (M + 1) unknowns for 0 ≤ j ≤ M . This
method must be repeated N times for 0 ≤ j ≤ M .
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
1 2 2 ∂2U ∂U ∂U ∂U
σ S + κµ + bS + =0 (19.30)
2 I ∂S 2 ∂r ∂S ∂t
This PDE is also discretized by the Crank–Nicholson scheme, centered
in the space, except for the term ∂U ∂r , which is treated explicitly and
“decentered inside the scheme”. Hence, all the terms are of the second
order, except this latter term, which is of a first order in time and of a
second order in space. The following discretization is used:
n
∂U 3U (i, 0) − 4U n (i, 1) + U n (i, 2) H∗r U n−1 (i, 0)
= =
∂r 2h 2h
which gives:
1 2 1
1 − A δS − B HS U n (i, 0)
2 4
1 1 1
= 1 + A δS2 + B HS + C H∗r U n−1 (i, 0)
2 4 2
with
σI2 (i2 k) (bih)k κµk
A = , B = , C = .
2 h h
A detailed algorithm is provided in the appendix for the pricing of
American call and put options in this context. The algorithm can also
be adapted for the pricing of other derivatives like interest-rate options,
warrants, etc.
The impact of the composite volatility and stochastic interest rates
seem to be significant on American at the money call and put option
values. This effect is more important for put options and may trigger
an early exercise of index puts. The effect reported here is greater
than that in standard models and is less than that in BCS. While the
negligible effect in standard models is intuitive, the effect in the BCS
model may be due to the process of interest rates chosen, allowing for
negative interest rates. Also, the numerical scheme presented here is more
efficient than those presented in previous studies. Empirical tests can
be conducted in order to test if the parameters are statistically stable
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
and if option prices are close to the market prices. For more details, see
Bellalah (2003).
2Πσ
The law of X is known as the standard normal distribution.
When a random variable Y = µ + σX is of the Gaussian type, it
follows a normal distribution with a parameter µ as the mean and σ 2 as
the variance.
It is often denoted by:
N (µ, σ 2 ).
z = µ + σg.
dS = rSdt + σSdz
Under the risk-neutral probability, the option value is given by its expected
value discounted to the present.
The option price can be computed by simulating the risk-neutral
random walk for the underlying asset. We start by today’s asset value S0
and continue until maturity. This provides a realization of the underlying
price path. We use this realization to compute the option payoff. Then,
we make similar realizations and compute the average payoff over all
realizations. The present value of this average provides the option value.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
1
d(log S) = r − σ 2 dt + σdz
2
over a short timestep, this can be written as:
1 √
S(t + δt) = S(t) + δS = S(t) exp r − σ 2 δt + σ δtφ
2
This expression is exact. Simulations are provided in the appendix.
Summary
This chapter contains the basic material for the pricing of derivative assets
in a continuous-time framework. The presentation is made as simple as
possible in order to enable uninformed readers to understand the main
derivations.
First, we present in detail the search for an analytic formula for
European call option within the PDE method.
Second, we illustrate in detail the martingale method for the derivation
of a European call formula.
Third, we apply finite difference methods for the valuation of European
call options.
Fourth, we present a model for the valuation of American options with
a composite volatility. The option-pricing literature has been concerned
with modeling stochastic interest rates and stochastic volatility without
an explicit treatment for long-term stock index options and index futures
options. In this chapter, we present a specific model for the valuation of
these options since the effect of interest rates on the long run is probably
more important than on the short-term options. This model is motivated by
the results of empirical and theoretical studies regarding the market index,
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
Questions
1. What is an implicit difference scheme?
2. What is an explicit difference scheme?
3. What is the solution method for the valuation of European calls on non-
dividend paying stocks?
4. Describe the solution method for the valuation of European calls on
non-dividend paying stocks in the presence of information costs.
5. What is a composite volatility?
6. What are the main principles of simulation methods?
7. Describe the Monte–Carlo method.
∂u ∂ 2u
= (A.1)
∂t ∂S 2
The function u(S, t) depends on time and the underlying asset value S. The
option’s pay off is known at the option’s maturity date. The problem is to
find the value of the function u(S, t), which satisfies the following system
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
of three conditions:
∂u ∂2u
= for 0 ≤ S ≤ L
∂t ∂S 2
u(0, t) = u(L, t) = 0 for t ≥ 0
u(S, 0) = u0 (S) for 0 ≤ S ≤ L
The first condition corresponds to the heat transfer equation for which the
underlying asset lies between 0 and a specified value.
The second condition is a boundary or a limit condition. It shows that
the option value is zero when the underlying asset price is zero.
The third condition represents a terminal or a maturity condition. It
gives the option price at the maturity date.
This general scheme takes different forms according to the values attributed
to θ with 0 ≤ θ ≤ 1.
b(1) = 1 + rk + σ2 k c(1) = − 12 rk − 12 σ 2 k
For i = 2 to n − 1
ai = 12 rik − 12 σ 2 i2 k
bi = 1 + σ2 i2 k + rk
ci = − 12 rik − 12 σ 2 i2 k
di = u(i, j − 1)
end (for i = 2 to n − 1)
a(n) = −1, b(n) = 1, c(n) = 0, d(n) = h
This is the procedure for the matrix inversion:
For i = 1 to n
u(i) = c(i)/[b(i) − u(i − 1)a(i)]
g(i) = (d(i) − g(i − 1)a(i)]/[b(i) − u(i − 1)a(i)]
end (for i = 1to n).
u(n, j) = g(n)
For i = n − 1 down to 1
u(i, j)g(i) − u(i)u(i + 1, j)
end for
End (for j = 1 to m).
For i = 1, M − 1
a(0) = 0, b(0) = 1, c(0) = 0, d(0) = U (i, 0)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
For j = 1, M − 1
Use gg = G, f f = F , ee = E, dd = D, cc = C, bb = B, and aa = A
with
a(j) = 14 ee − 12 bb
b(j) = 1 + bb
c(j) = − 14 ee − 12 bb
d(j) = 14 cc(U (i + 1, j + 1) − U (i + 1, j − 1) − U (i − 1, j + 1)
+U (i − 1, j − 1)) − a(j)U (i, j − 1) − c(j)U (i, j + 1)
+(f f − bb − 2aa)U (i, j) + 12 dd + aa U (i + 1, j)
+ aa − 12 dd U (i − 1, j)
End For j = 1, M − 1
a(M ) = 0, b(M ) = 1, c(M ) = 0, d(M ) = ih
(For a put, d(M ) becomes:
if (K − ih) > 0, then d(M ) = K − ih Else d(M ) = 0 End if).
The matrix is inverted again and option values are generated for a nil
interest rate: for i = 0, M , U (i, 0) = x(i), End.
Now, we can generate option values for a fixed j. This the second step.
For j = 1, M − 1
a(0) = 0, b(0) = 1, c(0) = 0, d(0) = 0
(For a put, a(0) = 0, b(0) = 1, c(0) = 0, d(0) = K) .
For i = 1, M − 1
jhk
gg = 1 + 2
√
(σI +νσr jh)2 (i2 k)
aa = 2gg
dd = (bih)k
hgg
a(i) = ( 14 dd − 12 )aa
b(i) = 1 + aa
c(i) = − 41 dd − 12 aa
d(i) = U (i, j) + c(i)U (i + 1, j)(b(i) − 1)U (i, j) + a(i)U (i − 1, j)
End For i = 1, M − 1
a(M ) = b(M − 1) + 4a(M − 1)
b(M ) = c(M − 1) − 3a(M − 1)
c(M ) = 0
d(M ) = d(M − 1) − 2ha(M − 1)
(For a put, the modification is a(M ) = 0, b(M ) = 1, c(M ) = 0, d(M ) = 0).
The matrix must be inverted again for the third time: for i =
0, M, U (i, j) = x(i) End, For i = 1, M − 1, U (i, j) = uu(i, M ) End, End
For j = 1, M − 1
Since options are of the American type:
For i = 0, M
For j = 0, M
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
Exercises
Exercise 1
Consider the following function F (X(t)) = eX(t)
Show that
t
1 t X(τ )
eX(τ ) dX(τ ) = eX(t) − 1 − e dτ.
0 2 0
Solution:
1
dF (X(t)) = eX(t) dX(t) + eX(t) dt
2
Since:
t
dF = eX(t) − eX(0) = eX(t) − 1
0
we have,
t t
1
eX(τ ) dX(τ ) = eX(t) − 1 − eX(τ ) dτ
0 2 0
Exercise 2
Consider the following function:
1. Show that:
t
X 2 (t) 1
X(τ )dX(τ ) = − t
0 2 2
2. Show that:
t t
b t 2 τ X(τ )
b τ eX(τ ) dX(τ ) + τ e dτ + b X(τ )eτ X(τ )dτ = etX(t) − 1
0 2 0 0
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
3. Deduce that:
t
(2aX(τ ) + bτ eτ X(τ ) )dX(τ ) = b(eX(τ ) − 1) + aX 2 (t) − at − b
0
t
τ X(τ ) τ2
× e X(τ ) + dτ
0 2
Solution:
1. Let us denote by G(X(t)) = aX 2 (t).
Hence,
dG = 2aX(t)dX(t) + adt
or,
t
dG = a(X 2 (t) − X 2 (0)) = aX 2 (t)
0
since X 2 (0) = 0.
We can compute the quantity,
t t t
dG = 2aX(τ )dX(τ ) + a dt
0 0 0
or,
t t
dG = 2aX(τ )dX(τ ) + at
0 0
Hence, we have:
t
aX 2 (t) = 2aX(τ )dX(τ ) + at
0
⇐⇒
t
2
aX (t) − at = 2a X(τ )dX(τ )
0
then,
t
X 2 (t) t
X(τ )dX(τ ) = −
0 2 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
1
dH = btetX(t) dX(t) + bX(t)etX(t) + bt2 etX(t) dt
2
or,
t
dH = b(etX(t) − e0X(0) ) = betX(t) − b
0
Since we have:
t t t
τ X(τ ) τ X(τ ) τ2
dH = b τe dX(τ ) + b e X(τ ) + dτ
0 0 0 2
then,
t
t
τ2
b(etX(t) − 1) = b τ eτ X(τ ) dX(τ ) + b
eτ X(τ ) X(τ ) + dτ
0 0 2
t t t 2
τ τ X(τ )
etX(t) − 1 = τ eτ X(τ ) dX(τ ) + eτ X(τ )X(τ )dτ + e dτ
0 0 0 2
tX(t) tX(t) t2
dF = (2aX(t) + bte )dX(t) + a + be X(t) + dt
2
or
t
dF = aX(t)2 + b(etX(t) − 1)
0
and
t t t
τ X(τ )
dF = (2aX(τ ) + bτ e )dX(τ ) + adt
0 0 0
t
τ X(τ ) τ2
+ be X(τ ) + dτ
0 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
t
X(t)2 t2
X(t)dX(τ ) = −
0 2 2
Hence,
t
2aX(τ )dX(τ ) = aX(t)2 − at
0
and
t t
τ2
b τ eτ X(τ ) dX(τ ) = b(etX(t) − 1) − b eτ X(τ ) X(τ ) + dτ
0 0 2
t
(2aX(t) + bτ eτ X(τ ) )dX(τ ) = b(etX(t) − 1) + aX(t)2 − at − b
0
t
τ2
× eτ X(τ ) X(τ ) + dτ
0 2
Table D.1.
References
Bellalah, M (2003). Valuation of long term options. International Journal of
Finance.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–654.
Boyle, P (1976). Rates of return as random variables. Journal of Risk and
Insurance, 43 (December), 694–711.
Boyle, P (1986). Option valuation using a three jump process. International
Options Journal, 3, 7–12.
Boyle, PP (1988). A lattice framework for option pricing with two state variables.
Journal of Financial and Quantitative Analysis, 23 (March) 1–12.
Brenner, M, G. Courtadon and M Subrahmanyam (1987). The valuation of stock
index options, Solomon Center Working Paper, 414 (June).
Cox, J, J Ingersoll and S Ross (1985). A theory of the term structure of interest
rates. Econometrica, 53, 385–407.
Fama, E and KR French (1993). Common risk factors in the returns on stocks
and bonds. Journal of Financial Economics, 33, 3–56.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch19
Chapter 20
Chapter Outline
This chapter is organized as follows:
1. Section 20.1 presents a numerical solution to the valuation of an
American call option on a dividend-paying stock.
2. Section 20.2 develops a numerical solution for the pricing of an
American put option on a dividend-paying stock.
3. Section 20.3 provides some numerical procedures in the presence
of information costs. An application is given for an American put
option.
4. Section 20.4 presents a numerical solution to the valuation of an
American convertible bond (CB) with many embedded call and put
options.
5. Section 20.5 shows how to apply two-factor interest rate models in the
pricing of bonds within information uncertainty.
6. Section 20.6 is devoted to CB pricing within information uncertainty.
7. Appendix A develops an improved finite difference approach to fitting
the initial term structure. It applies the model in Vetzal to the valuation
of European and American-style options.
8. Appendix B provides a detailed algorithm for the valuation of American
calls when there are several dividends.
9. Appendix C presents a detailed algorithm for the valuation of American
puts in the same context.
833
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Introduction
Financial economists often resort to numerical methods. They particularly
use finite difference methods to solve partial differential equations (PDEs)
that must be satisfied by the prices of derivative securities. Indeed, these
methods are a powerful tool in option pricing when there are no closed-form
solutions.
The finite difference method consists in discretizing the PDE and
the boundary conditions using either a forward or a backward difference
approximation scheme. The resulting system is then solved iteratively.
This gives the derivative asset price at each instant of time as a function
of different levels of the underlying asset price. It is possible to classify
the main approaches in pricing interest-rate contingent claims into two
classes. The first specifies the evolution of some smaller number of points
on the yield curve and incorporates time-dependent parameters into the
model. This allows to meet the consistency with observed initial curve.
This approach is based on the work of Black et al. (BDT) (1990),
Hull and White (HW), (1990a), Jamshidian (1991), and Black and
Karasinski (1991).
The second approach is based on the work of Heath et al. (HJM),
(1992). This method takes all the term structure as a model input and
specifies the dynamics in an arbitrage framework. This method is, by
definition, consistent with the current yield curve and need not augment
the model with time-dependent parameters. But, this approach is hard
to implement for American-style claims. Brennan and Schwartz (1977)
presented a numerical solution for the valuation of American put options
when there are discrete distributions to the underlying asset. The CB is
a security-paying periodic coupon. It is more complex than the warrant
and involves a dual option. It gives the right to the bondholder to convert
the bond into common stocks and provides the issuing firm the right to
call the bond for redemption. Hull and White (1990b, 1993) proposed
a technique (explicit finite difference method), which is only applicable
to a specific interest-rate model, in order to implement a time-dependent
parameter approach. However, explicit finite difference methods are prone
to stability problems unless certain conditions regarding the size of the
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 835
time step are satisfied. Besides, the HW method requires that the model be
mean reverting after the transformation of the model to one with constant
volatility. However, this condition may not be true for some parametric
values and it is difficult to vary time step sizes so as to match the dates
of cash flows. It is well known that implicit difference methods are quite
flexible and unconditionally stable. When implementing implicit difference
methods as shown in Uhrig and Walter (1996), there are no needs in
transforming the model to one with a constant volatility. Vetzal (1998)
proposed a discretizing strategy for mean-reverting models.
where d stands for the dividend amount and T − and T + refer to the instants
just before and just after the underlying asset goes ex-dividend.
The first condition gives the call payoff at the maturity date. The
second condition shows that the call is worthless when the underlying asset
price is zero. The third condition indicates that the call value cannot be
less than its immediate exercise value when the underlying asset goes ex-
dividend. This condition characterizes the existence of a certain level of the
underlying asset, for which the option value (with dividends) is equal to its
value upon exercise. This is the critical underlying asset price corresponding
to the situation where the option intrinsic value is above C(S, T + ). More
generally, this situation needs the use of the following condition on the
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
This condition must be satisfied for a sufficiently high level of the underlying
asset price.
Si = ih for i = 0 to n
Tj = jk for j = 0 to m.
The option price u(S, T ) can be written as C(S,T ) = C(Si , Tj ) = C(ih, jk).
The partial derivative with respect to time, ∂C ∂t , can be approximated
at the point (i, j) by the difference:
∂C [C(i, j) − C(i, j − 1)]
= .
∂t k
The partial derivative with respect to the asset price can be approximated
at the point (i, j) by the difference:
∂C [C(i + 1, j) − C(i − 1, j)]
= .
∂S 2h
2
The term ∂∂SC2 can be approximated at the (i, j) point by:
∂ 2C [C(i + 1, j) − 2C(i, j) + C(i − 1, j)]
= .
∂S 2 h2
If we replace these partial derivatives with their values in the B–S PDE, we
obtain:
with
1 1 1 1
ai = rik − σ2 i2 k, bi = 1 + σ2 i2 k + rik, and ci = − rik − σ2 i2 k.
2 2 2 2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 837
Boundary condition given in Eq. (20.2) for the American call is approxi-
mated by:
K
C(i, 0) = ih − K if i ≥
h
K
C(i, 0) = 0 if i < .
h
Boundary condition given in Eq. (20.3), C(0, t) = 0, is approximated by:
C(0, j) = 0 for j = 0, 1, . . . , m.
This condition shows that just before the stock goes ex-dividend
(instant t− ), the put value must be equal to the greater of the intrinsic
value and the put price when the stock is ex-dividend, (the instant t+ ).
This problem has also no analytic solution and numerical methods must
be used.
ai P (i − 1, j) + bi P (i, j) + ci P (i + 1, j) = P (i, j − 1)
with
1 1
ai = rik − σ2 i2 k (20.13)
2 2
bi = 1 + σ 2 i2 k + rk
1 1
ci = − rik − σ2 i2 k
2 2
for i = 1 to n − 1 and j = 1 to m − 1.
The boundary condition given by Eq. (20.11) for each value of j is
approximated by P (n−1, j)−P (n, j) = 0 for j = 1 to m. Equations (20.13)
and (20.11) represent a set of n linear equations with (n + 1) unknowns
u(i, j) for i = 0 to n. The use of condition given in Eq. (20.7) allows
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 839
This system can be solved by inverting the matrix to give all possible values
of the option price at each instant j as a function of the values at an instant
before. Appendix C provides a detailed algorithm corresponding to this
model.
∂V ∂V 1 ∂ 2V
+ (r + λS )S + σ2S 2 = (r + λV )V. (20.14)
∂t ∂S 2 ∂S 2
Using a finite number of equally spaced times between the present time
t and the option’s maturity date T , we have: ∆t = (TN−t) , where a total of
(N + 1) times are considered from t, t + ∆t, t + 2∆t, . . . , T . In the same way,
we consider a finite number of equally spaced stock prices from 0 to M ,
where ∆S = SM max
with stock prices taking the values 0, ∆S, 2∆S, . . . , Smax .
This allows a diagramatic representation for the different values of the state
variable and time variable. Each point in the grid (i, j) corresponds to time
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
t + i∆t and a stock price j∆S. Consider now the valuation of an option Vi,j
at time i and position j.
Each interior point (i, j) can be approximated by its partial derivative
using a forward difference approximation:
∂V Vi,j+1 − Vij
= (20.15)
∂S ∆S
or a backward difference approximation:
∂V Vi,j − Vi,j−1
= . (20.16)
∂S ∆S
It is also possible to use a symmetrical approximation using two space steps
by averaging the two partial derivatives:
∂V Vi,j+1 − Vi,j−1
= . (20.17)
∂S 2∆S
The time partial derivative can be approximated using a forward difference
approximation to make a link between time t + i∆t and t + (i + 1)∆t:
∂V Vi+1,j − Vij
= . (20.18)
∂t ∆t
Using Eq. (20.16), the backward difference at the node (i, j + 1) is given by
Vi,j+1 −Vi,j 2
∆S
. The term ∂∂ 2VS can be approximated at node (i, j) by:
Vi,j+1 −Vi,j V −V
∂ 2V ∆S
− i,j ∆Si,j−1
=
∂ 2S ∆S
or
∂2V Vi,j+1 − Vi,j−1 − 2Vi,j
= . (20.19)
∂2S ∆2 S
Since S = j∆S, replacing Eqs. (20.17), (20.18) and (20.19) in Eq. (20.14)
gives:
Vi+1,j − Vij Vi,j+1 − Vi,j−1
+ (r + λS )j∆S
∆t 2∆S
1 Vi,j+1 + Vi,j−1 − 2Vij
+ σ2 j 2 ∆S 2 = (r + λV )Vi,j
2 ∆S 2
for j = 1, 2, 3, . . . , M − 1 and i = 0, 1, . . . , N − 1.
This last equation can be re-written as:
Numerical Methods and PDEs for European and American Derivatives 841
where
1 1
aj = (r + λS )j∆t − σ2 j 2 ∆t, bj = 1 + σ 2 j 2 ∆t + (r + λV )∆t,
2 2
1 1
cj = − (r + λS )j∆t − σ2 j 2 ∆t.
2 2
This is the general method for the B–S PDE in the presence of information
uncertainty. This method corresponds to the implicit finite difference
method. The implicit method gives a relationship among three values of
the option at time (t + i∆t) and a value at time (t + (i + 1)∆t). The four
values are respectively,
When the underlying asset price is zero, the put price corresponds to its
strike price or:
When the underlying asset price tends to infinity, the put price approaches
zero or
If the option value is less than its intrinsic value, then early exercise
is optimal at time (T − 1∆t). In this case, the option value is set equal to
(K − j∆S). A similar test is done at all the other nodes.
where
1 1 1 2 2
a∗j = − (r + λS )j∆t + σ j ∆t ,
1 + (r + λV )∆t 2 2
1
b∗j = [1 − σ2 j 2 ∆t],
1 + (r + λV )∆t
and
1 1 1
c∗j = (r + λS )j∆t + σ2 j 2 ∆t .
1 + (r + λV )∆t 2 2
Numerical Methods and PDEs for European and American Derivatives 843
The bondholder has the choice at each call date to receive the call price
CP (t) or the conversion value, C(V, t). Therefore, the value of the called
bond VIC (V, t) must satisfy the following condition:
and at any time during the call period, the bond’s value cannot exceed the
call price, i.e.,
Nc W (V, t) ≤ V. (20.30)
Equation (20.31) indicates that the bond value is zero when the firm is
worthless.
W (0, t) = 0. (20.31)
Equation (20.32) shows that the CB value is less than the value of an
equivalent straight bond and the maximum number of shares in which it
can be converted:
Numerical Methods and PDEs for European and American Derivatives 845
Equation (20.37) must be applied at each dividend date where the instants
just before and just after are denoted respectively by t− and t+ .
Equation (20.38) must be applied at each coupon date when the bond is
not currently callable. The instants just before and just after are denoted
also by t− and t+ , respectively.
W (V, t− ) = W (V − I, t+ ) + i (20.38)
Equation (20.39) must be applied at each coupon date when the bond is
currently callable.
Vi = ih for i = 0 to n
τj = jk for j = 0 to m.
with
1 1 1 1
ai = rik − σ2 i2 k, bi = 1 + σ2 i2 k + rk, ci = − rik − σ2 i2 k
2 2 2 2
for i = 1 to n − 1 and j = 1 to m.
For a given j, the system given in Eq. (20.40) gives (n − 1) equations with
(n + 1) unknowns. (when i varies from 0 to n).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
W (0, j) = 0 for j = 0 to m
The discretization of Eq. (20.34) gives:
Since the bond can be called before the maturity date, the value of W (i, j) is
not defined for a certain i greater than a certain value q, given by q = CP(j)
zh .
On a dividend date, Eq. (20.37) is approximated to:
D D
W i − , jD = zih if W i − , jD ≥ zV
h h
D
W (i, jD ) = zih if W i − , jD ≤ zih.
h
On a coupon date, Eq. (20.38) is approximated to:
I
W (i, jc ) = W i − , jc + I (20.43)
h
and Eq. (20.39) is approximated to
I I
W (i, jc ) = W i − , jc + I for W i − , jc + I ≤ CP (jc )
h h
I
W (i, jc ) = CP (jc ) for W i − , jc + I > CP (jc )
h
A detailed algorithm is given in Appendix D.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 847
20.4.4. Simulations
Using the algorithm in Appendix D, the CB price is simulated with the
following data:
Par value of the bond: 40, semi-annual coupon: 1.1, quaterly dividend: 1.1,
firm variance rate: 0.0012 per month, Risk-free rate: 0.0057 per month. The
bond is not callable for 5 years; it is callable at 43 (plus accrued interest)
for the next 5 years, at 42 for the next 5 years and at 41 for the last 5 years.
Using 200 iterations, h = V200 max
= 2.5, (T − t) = 240 months and a time
step of one month, Table 20.1 provides the prices for different levels of the
underlying asset V .
where:
∂Z 1 ∂2Z ∂ 2Z 1 ∂ 2Z
H(Z) = + w2 2 + ρwq + q2 2
∂t 2 ∂r ∂r∂l 2 ∂l
∂Z ∂Z1 ∂Z2
− ∆1 − ∆2 = 0.
∂r ∂r ∂r
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 849
This expression of the market price of risk is obtained because the consol
bond corresponds to a tradable security. In the presence of two state
variables; the spot rate and the consol bond, the valuation equation can
be written as:
∂Z 2 q 2 ∂Z
H (Z) = (γr w − u) − l − (r + λl )l + 2
∂r l ∂l
or
∂Z 1 ∂ 2Z ∂2Z 1 ∂ 2Z ∂Z
+ w2 2 + ρwq + q 2 2 + (u − γr w)
∂t 2 ∂r ∂r∂l 2 ∂l ∂r
2
q ∂Z
+ l2 − (r + λl )l + 2 − (r + λZ )Z = 0.
l ∂l
∂V 1 ∂ 2V ∂V
+ σ2 S 2 2 + (rS − d(S, t)) − rV ≤ 0. (20.46)
∂t 2 ∂S ∂S
The maturity condition corresponds to the value of the CB at this date,
which is scaled to one: V (S, T ) = 1. At each coupon date, the following
condition must be satisfied:
V (S, t− +
c ) = V (S, tc ) + C.
Since the bond can be converted into q shares of the issuer, its value must
be higher than V ≥ qS. The bond’s value just before maturity is given
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 851
by max(nS, 1). Besides, for high values of the underlying asset, i.e., when
S → ∞, the CB value approaches the conversion value, or V (S, t) ∼ nS.
When the underlying asset price tends to zero, the CB price corresponds
to the present value of the future coupons and principal:
V (0, t) = e−(r+λv )(T −t) + Ce−(r+λv )(tc −t) .
dS = µSdt + σSdX1
Using Ito’s lemma, it is possible to show that the dynamics of the CB price
are given by:
∂V ∂V ∂V
dV = dt + dS + dr
∂t ∂S ∂r
2
1 2 2∂ V ∂ 2V 2
2∂ V
+ σ S + 2ρσSw +w dt.
2 ∂S 2 ∂S∂r ∂r2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Summary
This chapter introduces the reader to the application of finite difference
methods to the pricing of American options. First, the numerical methods
are applied to the valuation of American call options when there are several
dividends. Second, the numerical methods are applied to the valuation of
American puts in the same context. Note that in both cases, there are no
analytical solutions in the literature. Third, we apply numerical methods for
the valuation of options in the presence of incomplete information. Fourth,
the numerical methods are illustrated for the valuation of American CBs
when there are several dividend dates, coupon dates, and implicit call and
put options. A CB on a stock gives its holder the right to receive coupons
at regular intervals and to get the principal at the bond’s maturity date.
Besides, the holder has the right to convert his/her bond into a specified
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 853
number of shares. Hence, the bond price must lay between its conversion
value and its straight value. The conversion value corresponds to the market
price of stock times the conversion ratio. Corporate bonds (CBs) correspond
to corporate securities that give the holder the right to forgo coupon/or
principal payments and convert to a pre-determined number of shares of
common stocks instead. In its simpler form, a CB can be viewed as a hybrid
security consisting of a straight bond and a call on the underlying equity.
Bond issues contain several optionality features like the possibility of early
conversion, callability by the issuer, putability by the holder etc. The pricing
of CBs needs in general, a simultaneous pricing of the equity and fixed-
income components. Several approaches have been proposed to account for
default risk in CB pricing. Practitioners account for credit risk in CBs by
introducing an effective credit spread in CB valuation tools. These spreads
are simple approximations based on the credit spread of a straight bond
conditioned for the hybrid nature of the CB. The Hull and White (1990b)
trinomial model uses an explicit method and is prone to stability problems.
To circumvent these concerns, Vetzal (1998) developed a simple two-point
upstream technique in the presence of an implicit scheme. The method can
be introduced after reviewing the Hull and White (1990b) trinomial model
and standard finite approaches. In each case, a detailed algorithm is given
in the appendix to illustrate the determination of the critical levels of the
underlying asset price corresponding to an optimal pre-mature exercise.
Fifth, we study the pricing of bonds in a two-factor interest rate model
within information uncertainty. Sixth, we study the valuation of CBs within
information uncertainty.
When the market price for interest-rate risk is zero, the price of an
interest rate-contingent claim u satisfies the following PDE:
1 2
σ urr + κ(θ − r)ur + ut − ru = 0 (A.1)
2
un+1 n+1
i+1 − 2ui + un+1
i−1 un+1 n+1
i+1 − ui−1
urr = , ur = ,
(∆r)2 2∆r
un+1
i − uni
ut = (A.2)
∆t
where
1 σ 2 ∆t κ(θ − r)∆t σ 2 ∆t
pi,i−1 = 2
− , pi,i = 1 − , and
2 (∆r) ∆r (∆r)2
1 σ 2 ∆t κ(θ − r)∆t
pi,i+1 = − (A.4)
2 (∆r)2 ∆r
When the ps are positive and sum to one, this explicit scheme is stable.
However, it is not easy to ensure that the ps are all positive. In fact, if r
is very low, the upward drift can be strong enough to lead to negative pi, i−1 .
When r is very high, the downward drift can lead to negative probabilities
pi, i+1 . This leads to the trinomial approach. When the spatial derivatives
ur,r and ur are approximated at time n rather than n + 1, this gives the
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 855
In this system, the first and second lines try to match the first and second
moments of the change in the state variable r over the small interval of
time. The term k(θ − r)∆t corresponds to the expected mean. The term
σ 2 ∆t + [k(θ − r)∆t]2 corresponds to the second moment. The third line
shows that the probabilities sum to one. This is a system of three equations
and with three unknowns which can be solved by very simple methods.
If the term [k(θ − r)∆t]2 is dropped, the solution to Eq. (A.7) is equal to
the ps in Eq. (A.4). This means that the second moment is approximated
by the variance and the error involved is of order (∆t)2 . For very low values
of the interest rate, rl , the branching in the lattice can be modified so that r
remains constant, goes up by ∆r or by 2∆r. This truncates the computation
domain since r will not be less than rl . Using a set of equations analogous
to Eq. (A.7), Vetzal (1998) showed that the probabilities that match the
first two moments are given by the following system:
In the same way, when the interest become very high at a level, rM , the
lattice can be modified in order to constraint r to that level, or to decrease
it by ∆r or 2∆r. In this case, the following system is used at this level to
compute the probabilities:
In sum, the Hull and White (1990b, 1993, 1994a,b, 1996) methods determine
the risk-neutral probabilities so as to match the first moments of the change
in the interest rate over the time interval ∆t. Their methods can account for
additional state variables and match initial yield curves and term structures
of interest-rate volatilities using time-dependent parameters. However, their
approach is based on an explicit method and the advantages of implicit
methods are known.
Vetzal (1998) considered a grid for the state variable from r1 to rM , uses
a standard method for all the interior points from r2 to uM−1 and applied
the following approach for the end points. For r1 , two Taylor expansions
around u(r1 ) are considered. The first is for u(r1 + ∆r). The second is for
u(r1 + 2∆r):
∆r2
u(r1 + ∆r) = u(r1 ) + ur ∆r + urr
2
(2∆r2 )
u(r1 + 2∆r) = u(r1 ) + ur 2∆r + urr . (A.10)
2
Equation (A.11) can be re-written as:
u(r1 + 2∆r) − 2u(r1 + ∆r) + u(r1 )
urr =
(∆r)2
−u(r1 + 2∆r) + 4u(r1 + ∆r) − 3u(r1 )
ur = . (A.11)
2∆r
At the level rM , two second-order expansions are used for u(rM − ∆r) and
u(rM − 2∆r) as follows:
(∆r)2
u(rM − ∆r) = u(rM ) − ur ∆r + urr
2
(2∆r)2
u(rM − 2∆r) = u(rM ) − ur 2∆r + urr (A.12)
2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 857
in Eqs. (A.11) and (A.13) at time step n rather than (n + 1). Hence, at the
minimum value of the interest rate, we have:
where:
1 σ2 ∆t 3κ(θ − r1 )∆t
c1,1 = 1 + r1 ∆t − − ,
2 (∆r)2 ∆r
σ 2 ∆t 2κ(θ − r1 )∆t
c1,2 = 2
− ,
(∆r) ∆r
1 σ2 ∆t κ(θ − r1 )∆t
c1,3 =− − (A.19)
2 (∆r)2 ∆r
The main advantage in the discretization in Vetzal (1998) is that this
approach does not impose a specific boundary condition at r1 . The classic
implicit numerical scheme given by Eqs. (A.5) and (A.6) is used for the
interior points and at the maximum value of r:
where:
1 σ 2 ∆t κ(θ − rM )∆t
cM,M−2 =− + ,
2 (∆r)2 ∆r
σ 2 ∆t 2k(θ − rM )∆t
cM,M−1 = − ,
(∆r)2 ∆r
1 σ2 ∆t 3κ(θ − rM )∆t
cM,M = 1 + rM ∆t − − .
2 (∆r)2 ∆r
This implicit scheme does not require a specific boundary condition at
the end point. This implicit scheme is quite flexible with regard to grid
construction. The implicit and the Crank–Nicolson versions of the scheme
are flexible with respect to the choice of the time. This can be done in such
a way to match cash-flow dates exactly near the option’s maturity date.
This flexibility does not exist in the HW models. This method shares with
the HW, the feature of avoiding the specification of boundary conditions
for models with mean reversion, but it can be applied to several situations
where the HW method cannot. The simulations conducted by Vetzal (1998)
show that the Crank–Nicolson scheme is much more accurate than the
HW method. For example, the maximum absolute error for the discount
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 859
function of the final grid values is almost 20 times smaller than the HW
method.
An un−1 = Bn un
un−2 = A−1 −1
n−1 Bn−1 An−1 Bn−1 u
n−1
= A−1 −1 n
n−1 Bn−1 An Bn u . (A.24)
This method is much faster than that proposed in Uhrig and Walter (1996).
Numerical results
Vetzal (1998) used three one-factor models corresponding to Eqs. (20.21)
and (20.73) with respectively, β = 0, β = 12 and β = 1. He also used
three EGARCH stochastic volatility extensions of these univariate models
as shown in Anderson and Lund (1997) and Vetzal (1997):
(1)
dr κ(θ − r) σrβ dBt
= dt + (1)
(2) (A.25)
d ln σ α + δ ln σ γ(ρdBt + 1 + ρ2 dBt )
where v = ln σ.
Hull and White (1993), Uhrig and Walter (1996), and Vetzal (1997)
showed that single-factor models give similar values for European bond
options, except for deep-out-of-the-money options.
For European-style options, the adjustment in Dybvig (1989) can be
used. The sample problem involves the valuation of a 5-year option on
a 10-year coupon bond using weekly time steps. The Crank–Nicolson
method is used. The results seem to be similar for single-factor models.
Stochastic volatility models produce somewhat higher values for deep-out-
of-the-money options. The impact of stochastic volatility depends on the
parameter, β.
Vetzal (1998) proposed a two-point upstream discretization strategy
for mean-reverting models. This method improves on the HW trinomial
lattice in several ways. Vetzal (1998) proposed an implicit scheme and also
the Crank–Nicolson scheme which offer superior stability and time-step
flexibility.
He also derived European option values consistent with an initial term
structure along the lines of Dybvig (1989) in a PDE context.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 861
For i = 0 to n, C(i, 0) = ih − K
if ih − K < 0, then C(i, 0) = 0, End (for i = 0 to n)
For i = 1 to n do
c(i)
w(i) = (b(i)−w(i−1)a(i))
,
(d(i)−g(i−1)a(i))
g(i) = (b(i)−w(i−1)a(i))
Here, the last elements in the system are calculated. We generate and
print all the unknowns at each instant of time using the Gauss method for
the inverted matrix.
For i = n − 1 down to 1, C(i, j) = g(i) − w(i)u(i + 1, j), End
For i = 0 to n write C(i, j).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
A special treatment is done for the determination of the critical asset price
corresponding to an optimal early exercise.
For j = 1 to m,
a(1) = 0, b(1) = 1 + σ 2 k + rk, c(1) = − 12 rk − 12 σ 2 k, and d(1) =
P (1, j − 1)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 863
For i = 2 to n − 1 do
a(i) = 12 rik − 12 σ 2 i2 k;
b(i) = 1 + σ 2 i2 k + rk;
c(i) = − 21 rik − 12 σ 2 i2 k, and
d(i) = P (i, j − 1).
For i = 1 to n do:
End for (i = 1 to n)
P (n, j) = g(n).
Here, the last elements in the system are calculated.
We generate all the unknowns at each instant of time.
For i = n − 1 down to 1, P (i, j) = g(i) − w(i)u(i + 1, j), End (for i =
n − 1 down to 1).
For i = 0 to n write P (i, j) end for.
A separate treatment is done for the determination of the critical asset price
corresponding to an early exercise.
For i = 0 to n do H(i) = P (i, j) − K + ih end for
k = 0,
For i = 1 to n do
if H(i) ≤ 0 then P (i, j) = K − ih, k = k + 1 end if
if k = 1 then Sc = (i − 1)h + hH(i − 1)/H(i − 1) − H(i) end if, write j, Sc
End (for i = 1 to n).
Initialization
For i = 1 to np1, enter d1(i) the length of period i in months, z(i),
For k = 1 to np2, enter d2(k), CP (k)(CP (0): no call)
nv = nvv; nvt = nvv; h = V nV
max
.
For i = 0 to nV ,
V (s) = ih
if z(1)v(s) ≥ P then W (i, j) = z(1)Vs else if vs ≥ P then W (i, j) = Vs
W (nv) = g(nv)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 865
To search for the value of z(i) corresponding to a given month, the following
treatment is done.
dt = 0, imp1 = 0,
Repeat imp1 = imp1 + 1;
dt = dt + d1(imp1)until(j ≤ dt) and
zz = z(imp1).
To search for the call price CP (j) corresponding to a given month j, the
following treatment is done.
dt = 0, imp2 = 0,
dt = dt + d2(imp2)until(j ≤ dt)
CPP = CP (imp2)
nv = nvv
nvt1 = trunc(CPP/zh) + 1
End for (i = 1 to nv − 1)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 867
Questions
1. Why are numerical methods used in asset pricing?
2. What is a finite difference scheme?
3. What is an implicit scheme?
4. What is an explicit scheme?
5. What are the main characteristics of CBs?
Exercises
Exercise 1
Consider the extended B–S equation for the pricing of an option with a
certain payoff at time T .
1. Can you reduce the extended B–S equation using the following transfor-
mations?
2τ
S = Eex , t=T − , V (S, t) = Ev(x, τ )
σ2
2. Can you reduce the extended B–S equation using the following transfor-
mations?
3. Can you give the new payoff and illustrate the method for the pricing of
a European call?
Solution
1. The extended B–S equation for the pricing of a European call option can
be written as:
∂V 1 ∂2V 2 2 ∂V
+ σ S + (r + λs )S − (r + λv )V = 0 (20.47)
∂t 2 ∂S 2 ∂S
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
V (0, t) = 0 whenS → ∞
lim V (S, t) = S
S→∞
V (S, T ) = max(S − E, 0)
For the first transformation, the partial derivative with respect to time
gives:
∂V ∂Ev(x, τ )
=
∂t ∂t
with
σ2 S
τ= (T − t), x = Log
2 E
or
∂V ∂v S σ2 σ2
=E Log , T − t
∂t ∂t E 2 2
∂V ∂v ∂x ∂v 1 E ∂v
= = =
∂S ∂x ∂S ∂x S S ∂x
and
∂V ∂v ∂τ σ 2 ∂v
= =− E
∂t ∂τ ∂t 2 ∂τ
The second partial derivative gives:
Numerical Methods and PDEs for European and American Derivatives 869
σ 2 ∂v 1 2
2∂ v 1 2 ∂v
− + Eσ 2
+ − σ + r + λs − (r + λv )v(x, τ ) = 0
2 ∂τ 2 ∂x 2 ∂x
⇐⇒
∂v ∂2v 2 ∂v 2
− + 2 + (r + λs ) 2 − 1 − 2 (r + λv )v(x, τ ) = 0
∂τ ∂x σ ∂x σ
⇐⇒
∂ 2v 2 ∂v 2 ∂v
+ (r + λs )1 − (r + λv )v(x, τ ) =
∂x2 σ2 ∂x σ 2 ∂τ
Let
2
r=k
σ2
hence,
∂v ∂ 2v 2λs ∂v 2
= + k+ 2 −1 − k + 2 λv v(x, τ )
∂τ ∂x2 σ ∂x σ
with
v(x, τ ) → 0
when
x → −∞
because:
V (0, t) = 0 and S → 0; Log S → −∞
then
x → −∞ and
V (S, t) ∼ S when S → ∞ since
S = Eex ; v(x, τ ) ∼ ex
because:
2τ
S = Eex → ∞ since ex → ∞ since S = Eex , t = T − σ2 we have for
t = T, τ = 0:
V (S, T ) = Ev(x, 0) = max(Eex − E, 0) = max(E(ex − 1), E, 0)
and so:
v(x, 0) = max(ex − 1, 0)
2.
2λs
2α + k − 1 + = 0.
σ2
It is possible to choose the u term by selecting the value:
2λs 2λv
β = α2 + α k + 2 − 1 − k 2 .
σ σ
We can solve the system of two equations for α and β.
2λs
2α + k − 1 + =0
σ2
2λs 2λv
α2 + αk − k + α + α 2 2 = β
σ σ
hence, we have:
2λs
1−k− σ2
α=
2
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 871
or
1 k λs
α= − −
2 2 σ2
and
2
1 2λs 2
β= k+1+ 2 + 2 (λs + λv ).
4 σ σ
When λu = λv , we have the following equation:
∂u ∂ 2u
=
∂τ ∂x2
with boundary conditions:
and
v(x, τ ) → 0 as x → −∞
and
u(x, τ ) → 0 as x → −∞
v(x, τ ) ∼ ex as ex → ∞
we will have:
u(x, τ ) ∼ e−αx−βτ ,
ex ∼ e(−1−α)x−βτ
1 2λs
⇐⇒ u(x, τ ) ∼ e 2 (1+k+ σ2 )x as ex → ∞
with also
and
v(x, 0) = max(ex − 1, 0)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
this gives:
1 2λs 1 2λs
e−αx v(x, 0) = max e 2 (k+1+ σ2 )x − e− 2 (k−1+ σ2 )x , 0
⇐⇒
u(x, 0) = max e
1
2x e
1
2 (k+1+ 2λ2s )x
σ
1 2λs
− e− 2 (k−1+ σ2 )x , 0
⇐⇒
1
x 1 2λs
u(x, 0) = max 2e sh
2 k + 2 x ,0
2 σ
References
Anderson, TG and J Lund (1987). Estimating continuous-time stochastic volatil-
ity models of the short-term interest rate. Journal of Econometrics, 77,
343–377.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
Black, F and P Karasinski (1991). Bond and option pricing when short rates are
lognormal. Financial Analysts Journal, 47, July–August, 52–59.
Black, F, E Derman and W Toy (1990). A one factor model of interest rates
and its application to treasury bond options. Financial Analysts Journal, 46,
January–February, 33–39.
Brennan, MJ and E Schwartz (1977). Saving bonds, retractable bonds, and
callable bonds. Journal of Financial Economics, 5, 67–88.
Courtadon, G (1982). The pricing of options on default-free bonds. Journal of
Financial and Quantitative Analysis, 17, 75–100.
Cox, J, J Ingersoll and S Ross (1985). A theory of the term structure of interest
rates. Econometrica, 53, 385–407.
Dybvig, PH (1989). Bond and bond option pricing based on the current term
structure. Working Paper, Washington University at Saint Louis.
Heath, R Jarrow and A Morton (1992). Bond pricing and the term structure
of interest rates: a new methodology for contingent claims valuation.
Econometrica, 60, 77–105.
Hull, J and A White (1990a). Pricing interest rate derivative securities. Review
of Financial Studies, 3, 573–592.
Hull, J and A White (1990b). Valuing derivative securities using the explicit
finite difference method. Journal of Financial and Quantitative Analysis, 25,
85–100.
Hull, J and A White (1993). One-factor interest rate models and the valuation
of interest rate derivative securities. Journal of Financial and Quantitative
Analysis, 28, 235–254.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch20
Numerical Methods and PDEs for European and American Derivatives 873
Part VIII
Exotic Derivatives
875
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
876
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
Chapter 21
Chapter Outline
This chapter is organized as follows:
1. In Section 21.1, the option to exchange one risky asset for another is
analyzed and valued.
2. In Section 21.2, forward-start options are analyzed.
3. In Section 21.3, pay-later options are studied and valued.
4. In Section 21.4, simple chooser options are analyzed.
5. In Section 21.5, complex chooser options are studied.
6. In Section 21.6, compound options are introduced and valued.
7. In Section 21.7, options on the minimum and options on the maximum
of two assets are studied.
8. In Section 21.8, extendible options are studied and valued.
9. In Section 21.9, equity-linked foreign exchange options and quantos are
analyzed and valued.
10. In Section 21.10, binary barrier options are studied.
11. In Section 21.11, lookback options are studied.
Introduction
The theory for pricing an option to exchange one risky asset for another
was proposed by Margrabe (1978). This theory grew out of the Black and
Scholes (1973) and Merton (1973) models. The option to exchange one risky
asset for another is implicit in some common financial arrangements.
877
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
• a long position in an asset or nothing call that pays the underlying asset
price at maturity when this asset price is higher than K and
• a short position in a cash or nothing call that pays K if the underlying
asset price is higher than K.
The value of the asset or nothing call is given by the first part of the
Black and Scholes (1973) formula. The value of the cash or nothing call is
given by the second part of the Black and Scholes (1973) formula. The sum
of the two parts corresponds to the Black–Scholes–Merton formula.
This chapter studies the main pricing relationships for the following
options: the option to exchange one risky asset for another, forward-
start options, pay-later options, simple chooser options, complex chooser
options, compound options, options on the minimum and options on the
maximum of two assets, extendible options, equity-linked foreign exchange
options, binary barrier options, and lookback options. For the analysis of
information costs and valuation, we can refer to Bellalah (2001), Bellalah
et al. (2001a, b), Bellalah and Prigent (2001), Bellalah and Selmi (2001)
and so on.
This option is worth at most S1 , and has a positive value, so one can write:
0 ≤ cexchange (S1 , S2 , T ) ≤ S1 .
∂c
It is possible to construct the following hedged portfolio, by selling c1 = ∂S1
∂c
units of asset 1 and buying, −c2 = − ∂S 2
units of asset 2, or:
∂c ∂c
c− S1 − S2 = 0.
∂S1 ∂S2
The value of this option is:
Note that when σ 2 = σ12 and σ2 = 0, this solution reduces to the Black–
Scholes equation. Margrabe (1978) showed that the usual put-call parity
relationship holds for the options to exchange one asset for another. In
particular, he showed that the following relationship applies for European
options and their underlying assets,
C(S1 , S2 , t) − C(S2 , S1 , t) + S2 = S1 .
Table 21.1 provides simulation results for options values as well as the Greek
letters. The reader can make some comments about these parameters. There
are two deltas and two gammas with respect to the first asset and the second
underlying asset.
C(St , St , T ) = St C(1, 1, T ).
When we account for the payout ratio, the current value of the forward
option is:
with:
−T
−T
Sd Sd
log KR −T + 12 σ 2 T log KR −T + 12 σ 2 t
x= √ , y= √
σ T σ t
When there is a cost of carry b, the value of a standard chooser is:
√
Cs = Se(b−r)T N (x ) − Ke−rT N (x − σ T ) − Se(b−r)T N (−y )
√
+ e−rT KN (−y + σ t)
with:
S S
ln K + b + 12 σ 2 T ln K + bT + 12 σ 2 t
x = √ , y = √ .
σ T σ t
Table 21.5 provides simulations of chooser option values and the corre-
sponding Greek letters.
given by:
The formula presented in Rubinstein (1991) for the complex chooser is:
√
cc = Sd−T1 N (x, y1 , ρ1 ) − K1 r−T1 N (x − σ t, y1 − σ T1 , ρ1 )
√
− Sd−T2 N (−x, −y2 , ρ2 ) + K2 r−T2 N (−x + σ t, −y2 + σ T2 , ρ2 )
with:
Sd−t
t t ln Scr r−t 1 √
ρ1 = , ρ2 = , x= √ + σ t
T1 T2 σ t 2
1 Sd−Ti 1
yi = √ ln + σ2 )Ti for, i = 1, 2
σ Ti Ki r−Ti 2
paying off the bondholders and giving the residual value (if any) to
stockholders, the bondholders have given the stockholders the option to
buy back the assets at the debt maturity date. In this context, a call on the
firm’s stock is a compound option, C(S, t) = f (g(V, t), t) where t stands
for the current time. If we assume that the return on the firm follows a
given diffusion process, then changes in the value of the call can be given
as a function of the changes in the firm’s value and time. The valuation of
options in this context is standard since a risk-less hedge can be constructed
by choosing an appropriate mixture of the firm and call options on the firm’s
stock. Hence, changes in the call’s value are expressed as a function of the
changes in the firm’s value and time.
with:
V 1 2 √ √
y1 = ln + b + σ v t1 σv t1 , y2 = y1 − σv t1 ,
Vcr 2
V 1 2
z1 = ln + b + σv T2 σv T2 , z2 = z1 − σv T2
K1 2
+ K2 e−rt1 N (−y2 )
cB−S (Vcr , K1 , T2 − T1 ) = K2 .
− K2 e−rt1 N (−y2 ).
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
+ K2 e−rt1 N (y2 )
PB−S (Vcr , K1 , T2 − t1 ) = K2 .
with:
S1
ln + b1 − b2 + 12 σ22 τ
S2 ln SK1 + b1 + 12 σ12 τ
d= √ , y1 = √
σ τ σ1 τ
ln SK2 + b2 + 12 σ22 τ
y2 = √ , σ2 = σ12 + σ22 − 2ρ12 σ1 σ2
σ2 τ
ρ1 σ = σ1 − ρσ2 , ρ2 σ = σ2 − ρσ1
If the strike is zero and there is no cost of carry, the formula for a call on
the minimum reduces to:
with:
S1
ln S2
+ 12 σ 2 τ √
d11 = √ , d22 = d11 − σ τ .
σ τ
where cE (S1 , S2 , 1, τ ) stands for the price of an option to exchange one unit
of asset S2 for one unit of asset S1 . This formula is also given in Margrabe
(1978).
where C(S2 , K, τ ) stands for the price of a European call on asset S1 , with
a strike price K and a maturity date τ .
The value of the call on the maximum of two assets can also be
written as:
√
cmax (S1 , S2 , K, τ ) = S1 e(b−r)τ N (y1 , d, ρ1 ) + S2 e(b−r)τ N (y2 , −d + σ τ , ρ2 )
√ √
− Ke−rτ [1 − N (−y1 + σ1 τ , −y2 + σ2 τ , ρ)].
The value of the put on the maximum, pmax (S1 , S2 , K, τ ) is given by Stulz
(1982) as:
Table 21.11. Simulations of the values of a call on the minimum of two assets. S1 = 100,
t = 11/01/2002, T = 11/01/2003, r = 4%, σ1 = 20%, S2 = 100, σ2 = 30%, and
ρ = 0.8%.
Table 21.12. Simulations of the values of a put on the minimum of two assets. S1 = 100,
t = 11/01/2002, T = 11/01/2003, r = 4%, σ1 = 20%, S2 = 100, σ2 = 30%, and
ρ = 0.8%.
Tables 21.11 and 21.12 provide simulation results for the option price as
well as the delta and the gamma with respect to the first asset and the
second asset.
1 2 2 ∂2V ∂V ∂V
σ S 2 + rS − rV + = 0.
2 ∂ S ∂S ∂t
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
i.e., the option holder can choose among three payoffs: the intrinsic value
(S − K1 ), zero, or the difference between the premium A and a standard
European call with a strike price K2 and a maturity date (T2 − T1 ). This
may also be written as:
C(I2 , K2 , T2 − T1 ) = I2 − K1 + A.
CE (S, K1 , T1 , K2 , T2 , A)
= C(S, K1 , T1 ) + SN2 (γ1 , γ2 , −∞, γ3 , ρ)
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
− K2 e−rT2 N (γ1 − σ 2 T1 , γ2 − σ 2 T1 , −∞, γ3 − σ 2 T2 , ρ)
− SN (γ1 , γ4 ) + K1 e−rT1 N (γ1 − σ 2 T1 , γ4 − σ 2 T1 )
− Ae−rT1 N (γ1 − σ 2 T1 , γ2 − σ 2 T1 )
where:
σ2 S
γ1 = + r+
ln T1 σ 2 T1 ,
2 I2
S σ2
γ2 = ln + r+ T1 σ 2 T1 ,
I1 2
S σ2
γ3 = ln + r+ T2 σ 2 T2 ,
K2 2
S σ2
γ4 = ln + r+ T1 σ 2 T1 ,
K1 2
and
T1
ρ=
T2
with:
N2 (a, b, c, d, ρ): the cumulative probability of the standard bi-variate nor-
mal density with correlation coefficient ρ for the region
[a, b]x[c, d];
N (a, b): the cumulative probability of the standard normal density
in the region [a, b] and
C(S, K1 , T1 ): the value of a standard call option in a Black and Scholes
context.
In the presence of a cost of carry b, the formula for the extendible call
is given by:
CE (S, K1 , T1 , K2 , T2 , A)
= C(S, K1 , T1 ) + Se(b−r)T2 N2 (γ1 , γ2 , −∞, γ3 , ρ)
− K2 e−rT2 N (γ1 − σ 2 T1 , γ2 − σ 2 T1 , −∞, γ3 − σ 2 T2 , ρ)
− Se(b−r)T1 N (γ1 , γ4 ) + K1 e−rT1 N (γ1 − σ 2 T1 , γ4 − σ 2 T1 )
− Ae−rT1 N (γ1 − σ 2 T1 , γ2 − σ 2 T1 )
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
where the interest rate r is replaced by the cost of carry b in the following
formulas:
S σ2
γ1 = ln + b+ T1 σ 2 T1 ,
I2 2
S σ2
γ2 = ln + b+ T1 σ 2 T1 ,
I1 2
S σ2
γ3 = ln + b+ T2 σ 2 T2 ,
K2 2
S σ2
γ4 = ln + b+ T1 σ 2 T1 ,
K1 2
and
T1
ρ=
T2
with:
N2 (a, b, c, d, ρ): the cumulative probability of the standard bi-variate nor-
mal density with correlation coefficient ρ for the region
[a, b]x[c, d];
N (a, b): the cumulative probability of the standard normal density
in the region [a, b] and
C(S, K1 , T1 ): the value of a standard call option in the Black and Scholes
context.
Tables 21.13 to 21.15 provide simulation results for the values of options
and the Greek letters.
Table 21.13. Simulation results for the values of extendible calls. S = 100,
K = 100, t = 11/01/2003, T = 11/01/2004, r = 4%, σ = 20%, extendible maturity
date = 11/06/2004, K2 = 110, and additional premium = 4.
Table 21.14. Simulation results for the values of extendible calls. S = 110, K1 = 100,
t = 28/12/2003, T = 27/12/2004, r = 2%, σ = 20%, extendible date = 27/12/2003,
K2 = 105, and A = 5.
Table 21.15. Simulation results for the values of extendible calls. S = 110, K1 = 100,
t = 28/12/2003, T = 27/12/2004, r = 2%, σ = 30%, extendible date = 27/12/2003,
K2 = 105, and A = 5.
C1∗ = X ∗ max[S ∗ − K , 0]
where S ∗ is the equity price in the currency of the investor’s country and K
is a foreign currency amount. The spot exchange rate expressed in domestic
currency of a unit of foreign currency, X ∗ , stands in front of the payoff
to show that the latter must be converted into domestic currency. The
domestic currency value of this call option is:
√
C1 = S X(d)−T N (d1 ) − K X(r∗ )−T N (d1 − σS T )
−T
S (d) √ 1 √
d1 = log ∗ −T
σS T + σS T
K (r ) 2
As stated before, for the continuous compounding of interest rates, the term
∗
(d)−T must be replaced by e−dT , the term (r∗ )−T by e−r T , and the term
(r)−T by e−rT .
where X̄ is the rate at which the conversion will be made. It can be written
in reciprocal units as
Reiner (for details, refer to Bellalah et al., 1998) gave the value of this
option in foreign currency as:
−T √
rd −(ρS X σS σX )T −T
C3 = X̄X S e N (d3 ) − K (r) N (d3 − σS T )
r∗
−T
S (d) √ 1 √
d3 = log − ρ σ
S X S Xσ T σS T + σS T .
K (r )
∗ −T 2
For the continuous compounding of interest rates, the term (d)−T must be
∗
replaced by e−dT , the term (r∗ )−T by e−r T , and the term (r)−T by e−rT .
Table 21.16 gives the results for the different models: Black–Scholes
(B–S), Garman–Kohlhagen (G–K), foreign equity/foreign strike (FE/FS),
foreign equity/domestic strike (FE/DS), fixed-rate foreign-equity, (FR/
FE), equity-linked foreign-exchange, (FL/FE), and equity-linked-foreign-
exchange (EL/FE).
Table 21.16.
B–S S K r d σS
G–K X K r r∗ σX
FE/FS SX K X r∗ d σS
FE/DS SX K r d σS X
rd ρS X σS σX
FL/FE S X̄ K X̄ r r∗
e σS
rd ρS X σS σX
EL/FE SX KS r∗
e d σX
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
Tables 21.17 to 21.20 provide simulations results for the values of foreign
currency options in different contexts. We also provide the Greek letters.
This allows the reader to make some comments.
Table 21.17. Simulations of foreign equity call struck in foreign currency. S = 100,
K = 100, t = 28/12/2003, T = 27/12/2004, r = 2%, σ = 30%, and (domestic/
foreign) = 1, 1.
Table 21.18. Simulations of foreign equity call struck in foreign currency. S = 100,
K = 100, t = 28/12/2003, T = 27/12/2004, r = 2%, σ = 30%, and S ∗ = 1.
Table 21.19. Simulations of foreign equity call struck in foreign currency. S = 0.95,
K = 100, t = 28/12/2003, T = 27/12/2004, r = 2%, and σ = 30%.
Table 21.20. Simulations of foreign equity put struck in foreign currency. S ∗ = 0.95,
K = 100, t = 28/12/2003, T = 27/12/2004, r = 2%, and σ = 30%.
ccon = 0 if St ≤ K
ccon = A else.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
pcon = 0 if St ≥ K
pcon = A if else.
where
S
log K + b − 12 σ 2 T
x = √ .
σ T
given by:
√
C = e(λS −λ+d)t φSN (φx) − φKe−(λ+r)t N (φx − φσ t)
S
log K + (r + λS − d)t 1 √
x= √ + σ t
σ t 2
with φ = 1 for a call and −1 for a put.
The first term in this formula, e(λS −λ+d)t φSN (φx), corresponds to the
present value of the underlying asset price conditional
√ upon exercising the
option. The second term, φKe−(λ+r)t N (φx − φσ t), indicates the present
value of the strike price times the probability of exercising the option.
The value of a cash-or-nothing call in the presence of shadow costs is
given by:
√
ccon = Ae(λS −λ)t N (x − σ t).
The value of the asset-or-nothing put is given by the present value of the
underlying asset price conditional upon exercising the put or:
Tables 21.21–21.24 provide some simulations of option values and the Greek
letters. The reader can make some comments regarding the evolution of
these parameters.
cgap = 0 if St ≤ K
cgap = St − A else.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
The “gap” refers to the difference (A−K). Note that the payoff of a gap
call corresponds to the difference between the payoffs of an asset-or-nothing
call and a cash-or-nothing call. Therefore, its value is given by:
√
cgap = Sd−t N (x) − AR−t N (x − σ t).
This formula is like that of a standard call except for the cash amount
sometimes replacing the strike price.
The pay-off of a gap option put is:
pgap = 0 if St ≥ K
pgap = St − A else.
Note that the payoff of a gap put corresponds to the difference between
the payoffs of an asset-or-nothing put and a cash-or-nothing put. Hence, its
value is given by:
√
pgap = −Sd−t N (−x) + AR−t N (−x + σ t).
This formula is like that of a standard put except for the cash amount
sometimes replacing the strike price.
Gap options can be defined with respect to two different strike
prices K1 and K2 . The call’s payoff is zero if S ≤ K1 and is S − K2
if S > K1 . The call’s payoff is zero, if S ≥ K1 and is K2 − S if
S < K1 . Using the analysis in Rubinstein and Reiner (1991), the call
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
formula is:
21.10.1.7. Supershares
S
The payoff from a supershare option is 0, if KL > S > KH and KL
otherwise. In this setting, the formula for a supershare option is:
Se(b−r)t
csupershare = [N (d1 ) − N (d2 )]
KL
(21.1)
log KSL + b + 12 σ 2 t log KSH + b + 12 σ 2 t
d1 = √ and d2 = √ .
σ t σ t
Option prices are computed at time 0 and option contracts are assumed to
have been initiated at time T0 ≤ 0.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
with
1 S0 σ2
d = √ ln + rT + T .
σ T m0T0 2
with:
1 S0 σ2
d = √ ln + rT + T .
σ T MT00 2
Note that these options correspond to the ordinary options as in Black–
Scholes formulas plus another term corresponding to the specificities of
their payoffs.
with
1 S0 σ2
dm = √ ln + rT + T .
σ T m 2
These options are issued in foreign exchange markets and also in stock index
markets.
Summary
The option to exchange one risky asset for another is analyzed and valued.
The identification of this option allows the pricing of several financial
contracts. This concept is useful in the valuation of complex options such
as options on the minimum or the maximum of several assets. Margrabe
provided valuation formulas for exchange options giving the right to
exchange one risky asset for another. The Black–Scholes–Merton formula
appears as a particular case of the Margrabe general formula. Forward-start
options provide an answer to the following question: how much can one pay
for the opportunity to decide after a known time in the future, known as
the “grant date”, to obtain an ATM call with a given time to maturity with
no additional cost? Pay-later options provide a certain insurance against
large one-way price movements and are traded on stock indices, foreign
currencies and other commodities. The buyer of pay-later options has the
obligation to exercise his/her option when it is in the money and to pay
the premium. A chooser allows its holder at the “choice date” to trade this
claim for either a call or a put. The claim is a regular chooser when the call
and the put have identical strike prices and time to maturity. The claim is a
“complex chooser” when the call and the put have different strike prices or
time to maturities. A complex chooser option is defined in the same way as
the simple chooser except that either the strike prices or (and) the time to
maturities for the call and the put are different. A compound option is an
option whose underlying asset is an option. Since an option may be either a
call or a put, we may find four types of compound options: a call on a call, a
call on a put, a put on a put, and a put on a call. Consider a levered firm for
which the debt corresponds to pure discount bonds maturing in some years
with a certain face value. Under the standard assumptions of liquidating
the firm in some years, paying off the bondholders and giving the residual
value (if any) to stockholders, the bondholders have given the stockholders
the option to buy back the assets at the debt maturity date. In this context,
a call on the firm’s stock is a compound option. If we assume that the return
on the firm follows a given diffusion process, then changes in the value of
the call can be given as a function of changes in the firm’s value and time.
The valuation of options in this context is standard. Changes in the call’s
value are expressed as a function of the changes in the firm’s value and
time. The chapter also presents a framework for the analysis and valuation
of forward-start options, pay-later options, simple chooser and complex
chooser options and several other forms of compound options as options
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
Questions
1. What is an exchange option?
2. What are the main applications of this concept?
3. How information costs affect the pricing procedure? What are the
definitions of the following options: forward-start options and pay-later
options?
4. What are the definitions of simple chooser options and complex choosers
options?
5. What are the payoffs of compound options?
6. What are the payoffs of options on the minimum and options on the
maximum of two assets?
7. What are the payoffs of extendible options?
8. What are the payoffs of equity-linked foreign exchange options and
quantos?
9. What are the payoffs of binary barrier options?
10. What are the payoffs of lookback options?
References
Bellalah, M (2001). Market imperfections; information costs and the valuation
of derivatives: some general results. International Journal of Finance, 13,
1895–1927.
Bellalah, M, and J-L Prigent (2001). Pricing standard and exotic options in the
presence of a finite mixture of Gaussian distributions. International Journal
of Finance, 13(3), 1975–2000.
Bellalah, M and F Selmi (2001). On the quadratic criteria for hedging under
transaction costs. International Journal of Finance, 13(3), 2001–2020.
Bellalah, M, Prigent JL and C Villa (2001a). Skew without skewness: asymmetric
smiles; information costs and stochastic volatilities. International Journal of
Finance, 13(2), 1826–1837.
Bellalah, M, Ma Bellalah and R Portait (2001b). The cost of capital in
international finance. International Journal of Finance, 13(3), 1958–1973.
Briys, E, M Bellalah et al. (1998). Options, Futures and Exotic Derivatives. En
collaboration avec E. Briys, et al., John Wiley & Sons.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch21
Chapter 22
Chapter Outline
This chapter is organized as follows:
1. Section 22.1 presents the definition of the value at risk (VaR) concept,
the risk measurement framework and risk metrics.
2. Section 22.2 studies the statistical and probability foundation of the VaR
concept.
3. Section 22.3 develops a more advanced approach to the VaR concept.
4. Section 22.4 concerns credit valuation.
5. Section 22.5 studies default and credit-quality migration.
6. Section 22.6 develops credit-quality correlations.
7. Section 22.7 studies portfolio management of credit risk in the Kealhofer,
McQuown and Vasicek (KMV) approach.
8. Section 22.8 is devoted to credit derivatives.
9. Section 22.9 is about models developed by rating agencies and propri-
etary models.
Introduction
Measuring the risks for a financial market participant or a financial
institution has become the main focus of modern finance theory. The
interest in measuring market risk and monitoring the positions is a
consequence of securitization and the need to measure performance. Value-
at-Risk (VaR) is a measure of the maximum potential change in the value
917
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
The VaR of the position in USD is determined in a second step. The VaR is
given by 1.65 times the standard deviation. It is approximately equal to the
market value of the position times the estimated volatility. If the estimated
volatility is 0.55%, then:
Hence, in 95% of the time, the firm will not lose more than 907,500 over
the next day.
It is important to note that the total risk of the bond must account for
the return on the Euro/USD exchange rate and the return on the 10-year
European bond. Assume the correlation is equal to −0.3. The total risk of
the position is:
2
VaR = σIrate + σF2 X + 2ρIrate, F X σIrate σF X
or
VaR = 0.9572 + 0.90752 + 2(−0.3)(0.957)(0.9075).
or
(rt − µt )
probability r˜t = < −1.65 = 5%,
σt
then
allow the different participants to unwind their positions in one day. This
approximation is mainly valid for market-trading activities.
V˜1 = V0 er̃
V0 − V˜1 .
V0 (1 − er̃ ).
When r̃ is small, er̃ is nearly equal to (1 + r̃). In this case, VaR is nearly
V0 r̃. Riskmetrics gives an estimation of r̃.
Example: Consider a portfolio with a current mark-to-market value V0 =
USD700 million. The determination of the VaR requires first the one-day
forecast of the mean. J-P Morgan assumed that this mean is zero over one
day. Then, we must calculate the standard deviation of the returns of the
portfolio. If the return on the portfolio is distributed conditionally normal,
then: assumed that the change in value of the portfolio is approximated by
its delta. The other greeks can also be used to appreciate the change in
value.
The second approach involves creating a large number of possible rate
scenarios and revaluing the portfolio under each scenario. VaR is defined in
this context as the fifth percentile of the distribution of the value changes.
More than one VaR model is used in practice since practitioners have
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
A 6.945 0.79 20
B 6.254 1 21
C 7.537 2.8 22
D 1.507 0.27 23
E 3.868 2 24
F 0.987 2.35 25
G 3.305 1 26
H 0.180 1 27
I 5.770 2 28
J 1.165 2.34 29
K 0.000 1 20
L 2.518 1 20
M 4.018 1.33 21
N 5.248 1 22
O 3.103 0.5 23
P 1.951 0.59 24
Q 3.844 1 25
R 1.902 0.5 26
S 0 2.47 20
T 4.702 2.11 20
Risk-free asset 35.198 0 0
Table 22.4. VaR of the portfolio for an horizon of 292 days (1 year).
spi-b708
T5 0.108 0.089 0.130 0.144 1 0.234 0.087 0.166 0.047 0.152 0.295 0.159 0.130 0.069 0.087 0.095 0.064 0.076 0.273 0.136 0
T6 0.167 0.137 0.2 0.222 0.234 1 0.146 0.280 0.079 0.255 0.493 0.268 0.219 0.118 0.148 0.162 0.110 0.130 0.458 0.230 0
T7 0.064 0.051 0.074 0.082 0.087 0.146 1 0.126 0.041 0.115 0.214 0.121 0.1 0.057 0.070 0.076 0.054 0.063 0.201 0.103 0
T8 0.119 0.098 0.142 0.158 0.166 0.280 0.126 1 0.072 0.226 0.428 0.236 0.194 0.107 0.132 0.145 0.099 0.117 0.403 0.201 0
T9 0.037 0.028 0.040 0.044 0.047 0.079 0.041 0.072 1 0.072 0.126 0.075 0.064 0.041 0.047 0.051 0.039 0.043 0.119 0.064 0
T10 0.108 0.089 0.130 0.144 0.152 0.255 0.115 0.226 0.072 1 0.429 0.238 0.196 0.109 0.134 0.147 0.101 0.119 0.406 0.2 0
T11 0.191 0.161 0.234 0.336 0.295 0.493 0.214 0.428 0.126 0.429 1 0.479 0.391 0.207 0.261 0.288 0.192 0.229 0.831 0.4 0
T12 0.114 0.093 0.136 0.151 0.159 0.268 0.121 0.236 0.075 0.238 0.479 1 0.270 0.150 0.185 0.203 0.139 0.164 0.559 0.268 0
T13 0.094 0.076 0.111 0.123 0.130 0.219 0.1 0.194 0.064 0.196 0.391 0.270 1 0.137 0.167 0.182 0.128 0.149 0.485 0.234 0
T14 0.055 0.041 0.060 0.065 0.069 0.118 0.057 0.107 0.041 0.109 0.207 0.150 0.137 1 0.111 0.120 0.091 0.104 0.270 0.135 0
T15 0.066 0.051 0.075 0.082 0.087 0.148 0.070 0.132 0.047 0.134 0.261 0.185 0.167 0.111 1 0.150 0.111 0.129 0.347 0.171 0
9in x 6in
T16 0.072 0.056 0.082 0.090 0.095 0.162 0.076 0.145 0.051 0.147 0.288 0.203 0.182 0.120 0.150 1 0.126 0.150 0.392 0.193 0
T17 0.051 0.038 0.055 0.061 0.064 0.110 0.054 0.099 0.039 0.101 0.192 0.139 0.128 0.091 0.111 0.126 1 0.119 0.276 0.141 0
T18 0.060 0.045 0.066 0.072 0.076 0.130 0.063 0.117 0.043 0.119 0.229 0.164 0.149 0.104 0.129 0.150 0.119 1 0.336 0.170 0
T19 0.189 0.160 0.232 0.259 0.273 0.458 0.201 0.403 0.119 0.406 0.831 0.559 0.485 0.270 0.347 0.392 0.276 0.336 1 0.537 0
T20 0.099 0.080 0.117 0.129 0.136 0.230 0.103 0.201 0.064 0.2 0.4 0.268 0.234 0.135 0.171 0.193 0.141 0.170 0.537 1 0
T21 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
b708-ch22
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
credit-rating category. The expected loss for each forward non-default credit
rating is given by:
Senior secured 65 20
Senior unsecured 45 18
Senior subordinated 35 16
Subordinated 25 19
Junior subordinated 15 10
1 2 3 5 7 10 20 30
(in %) (in %) (in %) (in %) (in %) (in %) (in %) (in %)
The study of the histories of the rating changes and defaults reveal
the existence of correlations in credit-rating changes. Creditmetrics use the
following formula to compute the average default correlation from the data:
σ2
N (µ−µ 2) − 1 σ2
ρ= ∼
(N − 1) (µ − µ2 )
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
where:
N : number of names covered in the data;
µ: average default rate over the years in the study;
σ: standard deviation of the default rates observed from year to year.
consider, for example, a firm with a market value of 100 million Dollars
and a debt of 50 million Dollars (maturing in one year) in the presence of
a given volatility, then it is possible to represent the range of possible asset
values and their frequencies in a diagram. The diagram representing the
future firm asset value and the frequency distribution will show a default
point at 50 million (the left-hand side of the frequency distribution). Hence,
the default risk of a company can be derived from the behavior of its asset
values and its liabilities.
study are: AAA, AA, A, BBB, BB, B, and CCC. In general, the higher
the rating, the lower the default probability and vice versa. Besides, the
lower an obligor’s original rating, the shorter the time it takes to observe
a default. Using default ratios, the study shows that default rates over a
one-year horizon exhibit a high degree of volatility. The default patterns
seem to share broad similarities across all pools. This result suggests that
S&P rating standards are consistent over time. Using transition analysis,
each one-year transition matrix shows all rating movements between letter
categories from the begining through the end of the year. Rating transition
ratios give useful information to investors and credit professionals. The one-
year rating transition ratios by rating category reveal that higher ratings
are long lived. The S&P study assumes that the rating transition rates
follow a first-order Markov process. This allows to model cumulative default
rates over several horizons. Rating transtion matrices are constructed to
produce stressed default rates. Multiyear transitions are also constructed for
periods of 2 through 15 years using the same methodology as for single-year
transitions. This allows to compute average transition matrices whose ratios
represent the historical incidence of the ratings. The study also reveals that
for example, 10-year transition ratios are less reliable than their one-year
counterparts. The analysis reveals an increase in corporate defaults in 1998.
In general, recoveries are estimated on the basis of the prices the defaulted
securities fatch at some time after the default event. The data corresponds
to 533 S&P’s rated straight-debt issues that defaulted in the period
1981–1997.
We denote by:
The methodology uses the fact that recoveries are based on the ultimate
values yielded by the completion of the bankruptcy process. Table 22.9
reproduced from the study of S&P summarizes the main findings.
with:
Nobs : the number of observations;
SA: the simple average;
SD: the standard deviation;
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
on the risk of the assets, and on the firm’s liabilities. The risk of the assets
is given by the standard deviation of the annual percentage change in the
asset value. The methodology of KMV Corporation looks for a default point
or the asset value at which the firm will default. The firm defaults when
its market net worth is zero. A ratio of default risk referred to as distance
default, compares the market net worth to the size of a standard deviation
move in the asset value:
Distance default = (market value assets − default point)
.
(market value assets − asset volatility)
The ratio says that a firm is n-standard deviation away from default.
When the probability distribution is known, the default probability can
be computed directly. KMV Corporation proposes a model of default
probability, Creditmonitor which computes the EDF. Expected default
frequency is the probability of default during some coming years. The
determination of the defaut probability of a firm is done in three steps.
Summary
The VaR corresponds to a number indicating the potential change in the
future value of a given portfolio. In the process of calculating the VaR, the
manager must specify the horizon for the calculation as well as the “degree
of confidence” chosen. Value at Risk calculations can also be done without
resorting to the standard deviation. Non-linear positions correspond, for
example, to a portfolio of options. The VaR of a portfolio of options can
be determined using the “greeks”. The basic method uses an option pricing
model to obtain the delta. This delta is used to determine the amount
of a market factor that must be held to compensate for a change in the
underlying asset price. The present value of the delta hedge position in
the underlying is included in the determination of the portfolio variance.
This method is efficient only for very small changes in the underlying asset
price. In fact, the delta is a linear measure only for very small changes
in the underlying asset price over very small intervals of time. Since the
VaR is concerned with the effects of large changes in the underlying asset
price, the linearity may lead to an inappropriate assesement of market-risk
measures. The estimation method is improved when the second derivatives
of the delta (the gamma) is used in the risk measure. Since the option
price function is nonlinear for different prices of the underlying asset, a
risk measure including gamma may also lead to an inaccurate measure of
market risk for significant changes in the underlying asset price. However,
the simultaneous use of delta and gamma can improve risk estimation.
A credit derivative can be seen as “any instrument that enables the
trading/management of credit risk in isolation from the other types of risk
associated with an underlying asset. These instruments may include: credit-
default products, credit-spread products, total-return products, basket
products, and credit-linked notes. Banks are the principal actors in the
market of credit risk. They represent the buyers, sellers, and intermediaries
of credit derivatives. Bankers are the principal buyers of credit protection.
However, any firm or institution with a portfolio subject to credit risk can
use credit derivatives. This is the case for firms with concentrated portfolios
who can use credit derivatives to manage credit lines. Sellers of credit
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
References
Bellalah, M and M Lavielle (2003). A decomposition of empirical distributions
with applications to the valuation of derivative assets. Multinational Finance
Journal, 83, 1871–1887.
Black, F and M Scholes (1973). The pricing of options and corporate liabilities.
Journal of Political Economy, 81, 637–659.
Garman, M (1997). Taking VaR to pieces. Risk, 10(10), October, 70–71.
Garman, M (1996a). Making VaR proactive. Financial Engineering Associates,
Working Paper.
Garman, M (1996b). Making VaR more flexible. Derivatives Strategy, April,
52–53.
Garman, M (1996c). Improving on VaR. Risk, 9(5), May, 61–63.
Hoppe, R (1998). VaR and the unreal world. Risk, 11, July, 45–50.
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch22
Index
algorithm, 313, 630, 638, 729, 730, Black–Scholes, 221, 222, 254, 339,
757, 762, 766, 801, 810, 816, 820, 367–369, 372, 375–377, 379–381,
822, 823, 833, 834, 837, 839, 846, 384, 394, 395, 397–399, 403–406,
847, 853, 861, 862, 864 408–411, 413, 422–425, 427, 441,
American option, 6, 10, 13, 50, 51, 508, 515, 516, 523–526, 536, 538,
98, 105, 222, 225, 234, 235, 244, 541, 544, 545, 552, 556, 557, 560,
249, 250, 252, 294, 324, 333, 348, 562, 563, 567, 583, 588, 594, 596,
359, 361, 380, 536, 539, 615–618, 597, 609, 611, 642, 654, 655,
621, 622, 627, 628, 632, 634, 637, 660–664, 678, 679, 714, 746, 752,
640–642, 645, 655, 656, 799, 801, 755, 757–762, 764, 765, 767, 768,
802, 811, 817, 819, 852, 880 772, 773, 779, 783, 787, 792, 794,
arbitrage, 10, 14, 20, 40, 46, 49, 60, 799, 802, 805, 835, 878–880, 887,
67, 72, 73, 76–78, 84, 85, 87, 89, 90, 893, 900, 903, 909, 913
97–99, 106, 125–127, 224, 237, 298, bond, 3, 5, 22–34, 46–48, 51, 67, 68,
300, 321, 340, 342, 343, 367, 368, 73, 75–77, 87, 90, 91, 102–106, 126,
375, 382, 383, 393, 396, 406, 407, 128, 141, 212, 237, 259, 260, 266,
416, 420, 426, 493, 494, 496, 267, 269–283, 286, 287, 289–291,
515–517, 519, 521, 522, 527, 537, 293–301, 303, 305–313, 316, 318,
546, 547, 552, 555, 560, 573, 617, 320–323, 328, 340, 341, 353, 354,
619, 621, 635, 643, 656, 658–660, 374, 376, 382, 393, 395, 397, 404,
669, 670, 673, 674, 686, 704, 710, 407–411, 416, 421, 422, 426, 494,
711, 714, 720, 734, 777, 802, 812, 508, 514, 519–523, 527, 535, 536,
834, 843, 849, 852, 878, 883 546–551, 555–557, 564, 573, 575,
asset pricing, 339, 367, 368, 398, 425, 617, 619, 650, 655, 667–677,
509, 517, 551, 767, 768, 771–773, 679–692, 694–696, 703–718,
777, 779, 793, 795, 867 721–727, 731–736, 739, 777,
811–813, 833, 834, 842–847,
binary barrier, 877, 879, 901, 914 849–854, 860, 864, 878, 879, 886,
binomial models, 221–224, 228, 231, 887, 913, 921, 927, 928, 930,
237, 246–249, 293, 294, 298, 305, 932–935, 941
310, 318, 320, 321, 327, 329, 331, bond futures, 27, 29, 667, 668,
334, 345–347, 349–352, 360–362, 670–673, 686, 687
404, 407, 494, 617, 622 bond option, 3, 22, 26, 46–48, 293,
bivariate normal, 368, 400, 401 297, 318, 320, 328, 410, 426, 535,
943
September 10, 2009 14:46 spi-b708 9in x 6in b708-Index
546, 564, 667–669, 681, 683, 686, delta, 53, 110, 111, 113, 134, 135, 144,
703, 712, 713, 739, 777, 778, 789, 170, 171, 175, 184, 196, 197, 199,
793, 795, 860 201–204, 213, 334–337, 378, 382,
bond yield, 294, 672, 687–689, 725 384, 385, 418, 419, 440–443,
bounds on options, 67, 85, 127 446–461, 464, 465, 467, 468, 478,
515, 525, 526, 670, 721, 754, 755,
CAPM, 338, 339, 367, 369, 373, 377, 757, 760, 761, 781, 852, 878,
378, 387, 389–391, 398, 517, 634 880–885, 887, 889, 890, 893, 897,
choosers, 32, 47, 877–879, 884–887, 901, 902, 905–907, 912, 924, 940
913, 914 delta hedging, 452, 781
combined strategies, 141, 150 derivatives, 4, 20, 22, 34, 36, 46, 56,
commodity futures, 393, 403, 425, 67, 69, 70, 92, 170, 171, 174, 176,
446, 583, 584, 590, 592, 593, 616, 221, 222, 271, 275, 294, 316, 320,
627, 628, 655, 656 327, 329, 334, 338–340, 345, 355,
composite volatility, 801, 803, 811, 356, 361, 362, 367, 368, 378, 380,
812, 816, 819, 820 382, 387, 398, 399, 405, 412, 425,
compound options, 33, 320, 615–617, 439–444, 454, 457, 458, 460, 461,
642, 644, 645, 648, 650, 653, 654, 466, 469, 471, 476, 477, 493–495,
656, 877–879, 885–887, 889, 913, 501, 505, 506, 509, 510, 514–516,
914 524–529, 535–538, 544, 546, 551,
conditional expectation, 493, 496, 552, 555, 558, 563, 565, 566, 572,
522, 529, 536, 552, 555, 782 574, 575, 579, 583–591, 593, 594,
condor, 67, 100–102, 127, 142, 596–598, 609, 620, 622, 625, 650,
163–166, 175 668, 681, 687, 689, 692, 698, 699,
contingent claims, 32, 34, 47, 222, 733, 735, 737, 750, 771, 779, 799,
294, 300, 321, 327, 328, 331, 345, 801, 802, 804–806, 808–810,
346, 355, 362, 368, 494, 536–538, 814–817, 819–821, 833, 834, 836,
552, 555, 588, 668, 779, 784, 834, 838–840, 842, 854, 857, 868, 875,
854, 859, 860, 884 917, 918, 933, 934, 940, 941
credit crunch, 4, 30, 260, 323 diffusion process, 405, 493, 518, 527,
credit risk, 30, 259, 293, 406, 680, 528, 547, 549, 587, 594, 704,
853, 917, 918, 920, 927, 928, 940 746–749, 755, 767, 775, 786, 887,
credit valuation, 917, 918, 927 913
crude oil, 3, 6–10, 12, 13, 46, 68, 71, discounting factors, 263, 348, 706
72, 79, 81, 83 distributions, 19, 35, 40, 48, 60, 74,
currency, 3, 21, 22, 31, 32, 46, 48, 51, 87, 88, 221, 237, 246, 248, 249, 293,
75, 213, 217, 287, 288, 330, 334, 310, 323, 339, 340, 348, 361, 367,
337, 341, 342, 377, 393, 403, 369, 370, 373, 376, 377, 384, 388,
414–418, 424–426, 446, 471, 655, 393, 399, 404, 425, 495–500, 509,
668–670, 687, 746, 760, 772, 777, 519, 524, 526, 535, 539, 571,
778, 789, 790, 795, 896, 898, 899, 615–622, 627, 634, 640, 642, 643,
901, 902, 920 645, 652, 655–657, 669, 671–673,
currency options, 32, 330, 334, 377, 686, 691, 692, 703, 720, 726,
393, 403, 416, 417, 424–426, 446, 728–732, 739, 746, 748, 750, 760,
471, 655, 668, 670, 746, 760, 772, 772, 773, 775, 776, 778, 783, 786,
777, 778, 789, 790, 795, 901 788, 789, 793, 817–819, 834, 837,
September 10, 2009 14:46 spi-b708 9in x 6in b708-Index
Index 945
845, 886, 888, 891, 900, 919, 921, 518, 527, 553, 554, 557, 559–563,
922, 924, 927, 933, 939 573, 590, 592, 635, 637, 646, 668,
dividends, 19, 20, 69, 75, 76, 111, 115, 676–678, 681, 682, 692–694, 703,
116, 120, 166, 167, 175, 213, 221, 704, 706, 707, 715, 716, 719, 721,
238, 246–248, 317, 328, 330, 331, 722, 724–728, 731–737, 739, 755,
333, 334, 341, 342, 348, 369, 376, 794, 804, 805, 808, 834, 840,
377, 393, 394, 396, 397, 404–406, 877–883, 913, 914, 918, 927–929,
413, 539, 584, 589, 615, 618–622, 933
627, 641, 642, 645, 655, 791, 802, futures, 3–10, 12–14, 16–22, 26–31,
833–835, 852, 861, 862, 864 35, 36, 40, 46–49, 56, 60, 67–84, 90,
98, 103, 125–127, 142, 144, 152,
early exercise, 87, 88, 98, 105, 237, 175, 211, 213, 226, 250, 259–266,
334, 425, 615, 618–620, 622–624, 268, 270, 271, 276–278, 287, 289,
626–629, 632, 633, 641–643, 655, 290, 294, 299, 305, 316, 320, 327,
803, 816, 820, 841, 842, 862, 863 328, 330, 331, 338–340, 342–344,
embedded call, 833 347, 361, 362, 367–369, 376, 377,
equity option, 3, 17, 48, 221, 222, 386–393, 397–399, 403–407, 409,
249, 329, 361, 362, 403–405, 409, 410, 413–415, 420, 422, 424–426,
535, 564, 653, 667, 678, 896, 898 446, 473, 496, 500, 521, 536, 537,
European option, 6, 33, 50, 51, 98, 573, 583, 584, 590–593, 615–617,
104, 221, 234, 235, 237, 244, 250, 619, 627, 628, 632–636, 638–642,
252, 294, 320, 324, 333, 345, 361, 655, 656, 663, 664, 667, 668,
367–369, 377, 380, 388, 393, 404, 670–673, 676, 680, 686–690, 692,
408, 415, 508, 549, 550, 557, 559, 704, 705, 707, 711, 712, 715, 716,
560, 616, 628, 637, 642, 645, 662, 719, 724–726, 737–739, 755, 773,
678, 681, 684–686, 704, 717, 751, 790, 802, 803, 813, 817, 819, 851,
768, 774, 777, 778, 790, 791, 793, 878, 880, 885, 889, 913, 919, 920,
795, 799, 807, 817, 860, 879, 880 923, 924, 930, 933, 939–941
extendible, 32, 47, 877, 879, 893–895, futures options, 27, 327, 330, 347,
897, 914 362, 368, 376, 388, 391, 393, 403,
404, 415, 425, 426, 446, 615, 616,
financial innovations, 3, 4, 32, 34, 35, 627, 628, 632–634, 656, 663, 664,
46, 48, 802 667, 670, 671, 673, 686, 738, 803,
financial instruments, 3, 4, 22, 24, 31, 819
32, 34, 35, 46, 48, 49, 51, 90, 268,
393, 404, 407, 440, 680, 681, 918,
gamma, 170–172, 174, 175, 334–337,
920
fixed income, 37, 57, 667, 676, 686, 384, 385, 418, 419, 440–443,
446–451, 454–461, 464, 465, 469,
704
478, 670, 749, 880–885, 887, 889,
forward start, 47
890, 893, 897, 901, 902, 905–907,
forward, 3–6, 8, 9, 11, 13, 21, 27, 29,
912, 940
31, 46–48, 67–69, 71–79, 90,
125–127, 175, 213, 259, 261, 276, Girsanov theorem, 535, 537, 541, 545,
277, 279–281, 287–291, 303–308, 564, 565, 574
321, 340–342, 356, 357, 367, 368, Greek letters, 142, 170, 334, 384, 418,
386, 387, 389, 390, 398, 412, 414, 419, 442, 466, 880, 885, 886, 889,
415, 417, 418, 424, 426, 493, 507, 896, 901, 905, 906, 912, 919
September 10, 2009 14:46 spi-b708 9in x 6in b708-Index
implied volatility, 37, 57, 167, 170, lattice approach, 221, 222, 249, 250,
175, 300, 460, 526, 671, 673, 686, 293, 320, 327–329, 331, 343, 344,
692, 732, 745–747, 754–758, 760, 346, 348, 360–362
761, 763, 766–768, 773, 776, 788, lookback, 32, 47, 875, 877, 879, 908,
790, 793, 795 909, 911, 912, 914
incomplete information, 338, 339,
344, 494, 560, 583–585, 588–590, market conditions, 367, 441, 445, 455,
593, 596, 615, 636, 650, 655, 669, 457, 459, 461, 688, 771
687, 747, 752, 759, 760, 772, 833, market volatility, 141, 407
849, 852 martingale approach, 535, 538, 539,
incomplete market, 767 554, 555, 557–560, 563, 564, 615
index options, 18, 36, 37, 47, 49, 56, martingale measure, 547, 574, 771,
57, 221, 330, 394, 396, 403, 772, 778, 779
405–407, 424, 426, 640, 642, 668, Merton model, 376, 772
673, 686, 728, 754, 756, 758, 760, monitoring, 34, 171, 418, 425,
767, 788, 790, 801–803, 812, 820, 439–441, 445, 446, 451, 454, 455,
823 457, 458, 461, 917
information costs, 85, 327, 329, Monte–Carlo simulation, 927
338–341, 343–348, 360–362, 514, mortgage backed securities, 291
516, 521, 524–526, 540, 544,
560–563, 583–586, 588–591, normal distributions, 310, 367, 369,
593–597, 617, 634–639, 646, 648, 370, 373, 384, 399, 495, 497–500,
650–654, 656, 661, 673–676, 687, 509, 519, 524, 526, 535, 571, 643,
703, 732–734, 736, 745, 747, 749, 652, 669, 672, 726, 728, 729, 731,
751–753, 759, 762, 763, 767, 768, 732, 746, 772, 818, 819, 886, 888,
771, 773, 780, 782, 783, 785, 786, 891, 921, 922
791–796, 807, 810, 820, 833, 839, numerical analysis, 799, 801, 803,
849, 850, 852, 879, 903, 914 805, 820
September 10, 2009 14:46 spi-b708 9in x 6in b708-Index
Index 947
numerical procedure, 221, 328, 353, 771–775, 777, 778, 788, 789, 791,
361, 833, 839 799, 801, 802, 819, 834, 932, 933,
numerical schemes, 801–804, 806, 938–940
814–816, 821, 822, 855, 858, 859 option strategies, 94, 126, 141–143,
409
oil markets, 3, 6–9, 12, 46, 71, 78, 125 options markets, 3, 8, 17–19, 22, 33,
option, 3–6, 8, 10, 13, 17–22, 24, 40, 46, 48, 49, 55, 60, 69, 348, 368,
26–28, 31–33, 36, 37, 40, 41, 46–53, 369, 374, 415, 426, 453, 516, 517,
55–57, 60, 61, 67, 69, 70, 77, 85–96, 646, 751
98, 99, 102–107, 110, 111, 113, 114,
116–129, 141–144, 147, 148, 150, parametric approach, 667, 670
152–154, 156, 157, 159, 160, partial differential equation, 353, 379,
166–168, 170, 171, 174–177, 381, 387, 420, 493–495, 515, 516,
183–188, 194, 196, 202, 208–217, 518, 535, 536, 549, 555–558, 561,
221–238, 240–246, 248–254, 270, 563–568, 588, 609, 619, 621, 624,
293, 294, 297, 299–302, 316, 318, 628, 651, 698, 735, 772, 774, 786,
320, 324, 325, 327–334, 338, 339, 799, 801–803, 833, 834
343–351, 353, 359–362, 367–383, path dependent options, 32, 33, 47,
386–399, 403–418, 420–426, 829
439–446, 448, 449, 451–462, 466, pay later, 31, 47, 875, 877–879,
471, 478, 494, 508, 513–518, 882–884, 913, 914
523–527, 535, 536, 538–542, 545, portfolio insurance, 36, 56, 67, 103,
546, 549, 550, 552–554, 556, 557, 127, 128, 406, 414, 415, 421
559–565, 568, 569, 572, 583, 588, pricing biases, 771
611, 612, 615–623, 627–651, pricing bonds, 259, 266, 328, 573
653–657, 659–664, 667–671, 673, pricing derivatives, 538, 546, 735, 779
676–688, 690–696, 698, 703, 704, pricing theory, 367, 368, 397, 403,
710, 712–717, 719, 720, 728, 731, 517, 564, 671, 772, 932, 933
732, 737–739, 745–747, 751–760, probability, 178, 224, 296, 298, 300,
762–768, 771–783, 786, 788–796, 303, 307, 308, 322, 323, 344,
799, 801–814, 816–821, 823, 825, 349–351, 355, 356, 360, 393, 395,
829, 833–842, 844, 852, 853, 400, 401, 496, 498, 518, 522,
858–861, 867, 875, 877–889, 527–529, 535–538, 541, 542,
891–896, 898–914, 918, 919, 925, 544–548, 551, 552, 555–558,
932–934, 938–941 563–565, 573–575, 619, 621, 623,
option combinations, 67, 94 628, 632, 636, 655, 669, 672, 673,
option pricing, 141, 142, 161, 167, 691, 692, 711, 712, 714, 728–731,
170, 175, 221, 224, 294, 299, 320, 738, 746, 750–752, 754–759, 764,
328, 349, 361, 365, 367–371, 375, 771, 774, 775, 778, 782, 794,
391, 397, 398, 403, 412, 416, 425, 802, 818, 895, 896, 903, 904,
439, 458, 491, 494, 517, 518, 523, 917–922, 924, 929, 932, 933,
527, 538, 564, 572, 613, 616, 617, 936–939, 941
627, 642, 653–655, 667, 668, 670, probability foundation, 917, 921
671, 686, 714, 719, 720, 743, 745,
746, 754, 759, 763, 767, 768, quantos, 33, 877, 896, 914
September 10, 2009 14:46 spi-b708 9in x 6in b708-Index
ratio spread, 102, 103, 127, 174, 404, term structure, 790
460 theta, 170, 171, 173–175, 334–337,
rating, 24, 31, 917, 918, 927–932, 936, 384, 385, 418, 419, 440, 441, 443,
937 446–451, 456–461, 464, 465, 473,
real world, 403, 404, 422, 439, 592 475, 478, 670, 729, 730, 811,
replication, 209, 421, 465, 493, 494, 882–885, 887, 889, 890, 897, 901,
513, 514, 516, 527, 537, 538, 879, 902, 905–907, 912
934 trading, 3, 5–7, 9–13, 17, 18, 21, 22,
risk management, 34, 37, 57, 67, 141, 25–28, 36, 37, 40, 46–48, 51–57, 60,
403, 493, 535, 615, 667, 721, 767, 67–71, 73, 78, 85, 87, 88, 90, 98,
771, 801, 877, 927, 940 126, 127, 141, 167, 168, 170, 174,
risk measures, 170, 425, 439–441, 460, 338, 374, 375, 377, 397, 407, 418,
461, 466, 478, 927, 940 421–423, 425, 457, 524, 678, 720,
risk neutral, 303, 327, 349, 351, 361, 756, 758, 784, 918, 923, 940
589, 735, 771, 773–775, 778, trading mechanisms, 67
781–783, 792–794
trinomial trees, 293, 313, 314, 316,
risk neutral probability, 303, 771, 774,
317, 320, 328, 353, 361
775, 778, 782, 793, 794
risk parameters, 417, 418, 461, 919
valuation, 34, 68, 74–76, 85, 107, 117,
smile effect, 746, 754, 766, 767, 771, 120, 177, 212, 213, 221, 222, 228,
773, 788, 789, 795 231, 234–238, 244–246, 249–253,
speculation, 14, 49, 67, 78, 84, 287, 293, 294, 305, 312, 316, 318,
125–127, 416 320, 321, 324, 325, 327–329, 331,
spot assets, 46, 142, 329, 388–391, 332, 338–343, 345, 358, 360–362,
398, 399, 557, 583, 584, 590, 593, 367–374, 376, 378, 379, 390,
635, 656, 663, 664, 777, 778, 789, 397–399, 403–405, 407–411,
795 414–416, 424, 425, 427, 446, 461,
spot options, 426, 553, 664 478, 508, 514, 528, 535, 539, 544,
stochastic interest rates, 320, 369, 546, 551, 552, 554, 560, 562, 564,
404, 555–558, 562, 667, 668, 678, 583–585, 592, 593, 596, 615–618,
681, 686, 777, 795, 801–803, 816, 620, 622, 627, 628, 632, 634,
819, 851 637–640, 642, 643, 645, 648, 650,
stochastic process, 493, 494, 498, 528, 654–657, 659, 667–669, 671, 680,
536, 686, 711, 712, 715, 731, 746 684, 686, 687, 692, 704, 713, 716,
stochastic volatility, 369, 404, 745, 718, 720, 731, 737, 739, 745–747,
771–778, 782, 785, 788, 789, 749, 754, 763, 765–768, 772, 773,
793–795, 802, 819, 860, 922 778, 779, 792, 795, 796, 801, 803,
straddles, 67, 69, 94, 95, 126, 127, 807, 808, 810–812, 817, 819, 820,
142, 150–154, 174, 175, 460, 884 823, 833–835, 837, 839, 840, 843,
subprime crisis, 4 845, 850, 852, 853, 859, 860, 875,
synthetic positions, 128, 142, 143 878, 879, 887, 889, 891, 893, 896,
913, 917–919, 927, 928
Taylor series, 274, 352, 353, 379, value at risk, 254, 917, 920, 940
493–495, 501, 509, 513, 518, 520, vega, 170–172, 174, 175, 334–337,
523, 524, 529, 530, 775, 857 384, 385, 418, 419, 441, 444,
September 10, 2009 14:46 spi-b708 9in x 6in b708-Index
Index 949
446–451, 458–461, 469, 470, 478, yield curve, 259, 260, 275–277, 289,
670, 882–885, 887, 889, 890, 897, 290, 297, 316, 320, 328, 355, 520,
901, 902, 905–907, 912 669, 670, 687, 704–706, 714, 715,
720–722, 725, 731, 736, 737, 739,
Wiener process, 295, 378, 493–497, 834, 856
505, 509, 524, 527, 585, 587, 590, yield to maturity, 259, 260, 272, 408,
643, 695, 748, 774, 776, 784, 811, 519, 680, 681, 705, 722
851