Derivatives Mondar Bellalah PDF

DERIVATIVES,
RISK MANAGEMENT
& VALUE
September 14, 2009 9:35 spi-b708 9in x 6in b708-fm
This page intentionally left blank

DERIVATIVES,
RISK MANAGEMENT
& VALUE
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
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.
DERIVATIVES, RISK MANAGEMENT & VALUE

Copyright © 2010 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.
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
Typeset by Stallion Press

Email: enquiries@stallionpress.com
Printed in Singapore.
Sandhya - Derivatives, Risk Management.pmd 1 1/11/2010, 5:41 PM

DEDICATION
I dedicate this book to the President of the Tunisian Republic, his

Excellency Mr. Zine El Abidine Ben Ali, in recognition of his continuous and
pivotal support for Science and its men in Tunisia. Over the last twenty-plus
years, the scientific achievements of Tunisian researchers in various fields
were of high importance. I do believe that the distinction of the scientific
research in Tunisia, in relation to other countries in the Euro-Mediterranean
region, is due to the particular efforts of Mr. Ben Ali.
In the field of Finance, the scientific community and I in Tunisia
were lucky to benefit from his particular attention. Indeed, by placing the
International Finance Conferences that I organized and headed under his
high patronage, they gained a remarkable international reputation. With
this opportunity, Ph.D. students, college professors and professionals were
able to communicate with professors and experts from the best U.S. uni-
versities and institutions, alongside Nobel laureates such as H. Markowitz
and J. Heckman.
Another important insight of Mr. Ben Ali’s role is related to the setting
of national awards to strengthen the mechanics of the scientific research at
undergraduate and graduate levels. Such awards boost one’s willingness to
improve the Tunisian economy and its ability to meet the new challenges
posed by the international context. Furthermore, the creation of Ben Ali’s
Chair for the dialogue of civilizations and religions in 2001 had a key role in
the enrichment of knowledge and human values in a multi-religions context.
In addition, it is considered as the Mecca of researchers from all over the
world who are involved in bringing new approaches to make people closer.
Mondher Bellalah
v

FOREWORD
by Edward C. Prescott (Arizona State University; Federal Reserve Bank of

Minneapolis)
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

FOREWORD
by Harry M. Markowitz (University of California, San Diego)
Herein follows a remarkable volume, suitable as both a textbook and a

reference book. Mondher Bellalah starts with an introduction to options
and basic hedges built from specific options. He then presents an accessible
account of the formulae used in valuing options. This account includes
historically important formulae as well as the currently most used results
of Black–Scholes, Merton and others. Bellalah then proceeds to the main
task of the volume, to show how to value an endless assortment of exotic
options.
Mondher Bellalah is to be congratulated for this tour de force of the
field.
ix

FOREWORD
by James J. Heckman (University of Chicago and University College Dublin)
Mondher Bellalah offers a lucid and comprehensive introduction to the

important field of modern asset pricing. This field has witnessed a
remarkable growth over the past 50 years. It is an example of economic
science at its best where theory meets data, and shapes and improves
on reality. Economic theory has suggested a variety of new and “exotic”
financial instruments to spread risk. Created from the minds of theorists
and traders guided by theory, these instruments are traded in large volume
and now define modern capital markets.
Bellalah offers a step-by-step introduction to this evolving theory
starting from its classical foundations. He takes the reader to the frontier
by systematically building up the theory. His examples and intuition are
splendid and the formal proofs are clearly stated and build on each other.
I strongly recommend this book to anyone seeking to gain a deep
understanding of the intricacies of asset pricing.
xi

FOREWORD
by George M. Constantinides (University of Chicago)
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

ABOUT THE AUTHOR
After obtaining his Ph.D. in Finance in 1990 at France’s leading University

Paris-Dauphine, Mondher Bellalah began his career both as a Professor of
Finance (HEC, INSEAD, University of Maine, and University of Cergy-
Pontoise) and as an international consultant and portfolio manager. He start-
ed out as a market maker on the Paris Bourse, before being put in charge of
BNP’s financial engineering research team as Head of Derivatives and Struc-
tured Products. Dr. Mondher has acted as an advisor to various leading
financial institutions, including BNP, Rothschild Bank, Euronext, Houlihan
Lokey Howard & Zukin, Associés en Finance, the NatWest, Central Bank of
Tunisia,DubaiHolding,etc.,andhasbeenChiefRiskOfficer,ManagingDirector
inAlternativeInvestments,HeadofCapitalMarketsandHeadofTrading.
Dr. Mondher has also enjoyed a distinguished academic career as a
tenured Professor of Finance at the University of Cergy-Pontoise in Paris for
about 20 years. During this time, he has authored more than 14 books and 150
articles in leading academic and professional journals, and was awarded the
Turgot Prize for the best French-language book on risk management in 2005.
English-language books co-written/co-edited by Dr. Mondher include
Options, Futures and Exotic Derivatives: Theory, Application and Practice
published by John Wiley in 1998, and Risk Management and Value:
Valuation and Asset Pricing published by World Scientific in 2008. His
French-language books include Quantitative Portfolio Management and
New Financial Markets; Options, Futures and Risk Management; and Risk
Management and Classical and Exotic Derivatives.
Dr. Mondher is an associate editor of the International Journal of
Finance, Journal of Finance and Banking and International Journal of
Business, and has been published in leading academic journals including
Financial Review, Journal of Futures Markets, International Journal of
Finance, International Journal of Theoretical and Applied Finance as well
as in the Harvard Business Review.
xv

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
PART I. FINANCIAL MARKETS AND

FINANCIAL INSTRUMENTS: BASIC
CONCEPTS AND STRATEGIES 1
CHAPTER 1. FINANCIAL MARKETS, FINANCIAL

INSTRUMENTS, AND FINANCIAL CRISIS 3
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1. Trading Characteristics of Commodity Contracts:
The Case of Oil . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.1. Fixed prices . . . . . . . . . . . . . . . . . . . . . . 7
1.1.2. Floating prices . . . . . . . . . . . . . . . . . . . . 8
1.1.3. Exchange of futures for Physical (EFP) . . . . . . 8
1.2. Description of Markets and Instruments: The Case of the
International Petroleum Exchange . . . . . . . . . . . . . . 8
1.3. Characteristics of Crude Oils and Properties of Petroleum
Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.1. Specific features of some oil contracts . . . . . . . 9
1.3.2. Description of Markets and Trading Instruments:
The Brent Market . . . . . . . . . . . . . . . . . . 11
xvii
xviii Derivatives, Risk Management and Value
1.4. Description of Markets and Trading Instruments: The Case

of Cocoa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4.1. How do the futures and physicals market work? . . 14
1.4.2. Arbitrage . . . . . . . . . . . . . . . . . . . . . . . 14
1.4.3. How is the ICCO price for cocoa beans calculated? 14
1.4.4. Information on how prices are affected by changing
economic factors? . . . . . . . . . . . . . . . . . . . 15
1.4.5. Cocoa varieties . . . . . . . . . . . . . . . . . . . . 15
1.4.6. Commodities — Market participants: The case of
cocoa, coffee, and white sugar . . . . . . . . . . . . 15
1.5. Trading Characteristics of Options: The Case of Equity
Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.1. Options on equity indices . . . . . . . . . . . . . . 17
1.5.2. Options on index futures . . . . . . . . . . . . . . 17
1.5.3. Index options markets around the world . . . . . . 18
1.5.4. Stock Index Markets and the underlying indices in
Europe . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6. Trading Characteristics of Options: The Case of Options on
Currency Forwards and Futures . . . . . . . . . . . . . . . . 21
1.7. Trading Characteristics of Options: The Case of Bonds and
Bond Options Markets . . . . . . . . . . . . . . . . . . . . . 22
1.7.1. The specific features of classic interest rate
instruments . . . . . . . . . . . . . . . . . . . . . . 22
1.7.2. The specific features of mortgage-backed securities 25
1.7.3. The specific features of interest rate futures, options,
bond options, and swaps . . . . . . . . . . . . . . . 26
1.8. Simple and Complex Financial Instruments . . . . . . . . . 31
1.9. The Reasons of Financial Innovations . . . . . . . . . . . . 34
1.10. Derivatives Markets in the World: Stock Options, Index
Options, Interest Rate and Commodity Options and Futures
Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.10.1. Global overview . . . . . . . . . . . . . . . . . . . 36
1.10.2. The main indexes around the world: a historical
perspective . . . . . . . . . . . . . . . . . . . . . . 36
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Contents xix
CHAPTER 2. RISK MANAGEMENT, DERIVATIVES

MARKETS AND TRADING STRATEGIES 67
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.1. Introduction to Commodity Markets: The Case of Oil . . . 70
2.1.1. Oil futures markets . . . . . . . . . . . . . . . . . . 70
2.1.2. Oil futures exchanges . . . . . . . . . . . . . . . . 70
2.1.3. Delivery procedures . . . . . . . . . . . . . . . . . 70
2.1.4. The long-term oil market . . . . . . . . . . . . . . 71
2.2. Pricing Models . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.2.1. The pricing of forward and futures oil contracts . . 71
2.2.1.1. Relationship to physical market . . . . . 71
2.2.1.2. Term structure of prices . . . . . . . . . 72
2.2.2. Pricing swaps . . . . . . . . . . . . . . . . . . . . . 72
2.2.3. The pricing of forward and futures commodity
contracts: General principles . . . . . . . . . . . . 72
2.2.3.1. Forward prices and futures prices: Some
definitions . . . . . . . . . . . . . . . . . 73
2.2.3.2. Futures contracts on commodities . . . . 74
2.2.3.3. Futures contracts on a security with no
income . . . . . . . . . . . . . . . . . . . 74
2.2.3.4. Futures contracts on a security with a
known income . . . . . . . . . . . . . . . 74
2.2.3.5. Futures contracts on foreign currencies . 75
2.2.3.6. Futures contracts on a security with a
discrete income . . . . . . . . . . . . . . 75
2.2.3.7. Valuation of interest rate futures
contracts . . . . . . . . . . . . . . . . . . 76
2.2.3.8. The pricing of future bond contracts . . 77
2.3. Trading Motives: Hedging, Speculation, and Arbitrage . . . 78
2.3.1. Hedging using futures markets . . . . . . . . . . . 78
2.3.1.1. Hedging: The case of cocoa . . . . . . . 79
2.3.1.2. Hedging: The case of oil . . . . . . . . . 79
2.3.1.3. Hedging: The case of petroleum products
futures contracts . . . . . . . . . . . . . 80
2.3.1.4. The use of futures contracts by petroleum
products marketers, jobbers, consumers,
and refiners . . . . . . . . . . . . . . . . 82
xx Derivatives, Risk Management and Value
2.3.2. Speculation using futures markets . . . . . . . . . 84

2.3.3. Arbitrage and spreads in futures markets . . . . . 84
2.4. The Main Bounds on Option Prices . . . . . . . . . . . . . . 85
2.4.1. Boundary conditions for call options . . . . . . . . 86
2.4.2. Boundary conditions for put options . . . . . . . . 86
2.4.3. Some relationships between call options . . . . . . 86
2.4.4. Some relationships between put options . . . . . . 88
2.4.5. Other properties . . . . . . . . . . . . . . . . . . . 89
2.5. Simple Trading Strategies for Options and their Underlying
Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.5.1. Trading the underlying assets . . . . . . . . . . . . 90
2.5.2. Buying and selling calls . . . . . . . . . . . . . . . 91
2.5.3. Buying and selling puts . . . . . . . . . . . . . . . 93
2.6. Some Option Combinations . . . . . . . . . . . . . . . . . . 94
2.6.1. The straddle . . . . . . . . . . . . . . . . . . . . . 94
2.6.2. The strangle . . . . . . . . . . . . . . . . . . . . . 94
2.7. Option Spreads . . . . . . . . . . . . . . . . . . . . . . . . . 95
2.7.1. Bull and bear spreads with call options . . . . . . 95
2.7.2. Bull and bear spreads with put options . . . . . . 96
2.7.3. Box spread . . . . . . . . . . . . . . . . . . . . . . 96
2.7.3.1. Definitions and examples . . . . . . . . . 96
2.7.3.2. Trading a box spread . . . . . . . . . . . 98
2.8. Butterfly Strategies . . . . . . . . . . . . . . . . . . . . . . . 99
2.8.1. Butterfly spread with calls . . . . . . . . . . . . . . 99
2.8.2. Butterfly spread with puts . . . . . . . . . . . . . . 100
2.9. Condor Strategies . . . . . . . . . . . . . . . . . . . . . . . . 100
2.9.1. Condor strategy with calls . . . . . . . . . . . . . . 100
2.9.2. Condor strategy with puts . . . . . . . . . . . . . . 101
2.10. Ratio Spreads . . . . . . . . . . . . . . . . . . . . . . . . . . 102
2.11. Some Combinations of Options with Bonds and Stocks . . . 103
2.11.1. Covered call: short a call and hold the underlying
asset . . . . . . . . . . . . . . . . . . . . . . . . . . 103
2.11.2. Portfolio insurance . . . . . . . . . . . . . . . . . . 103
2.11.3. Mimicking portfolios and synthetic instruments . . 104
2.11.3.1. Mimicking the underlying asset . . . . . 104
2.11.3.2. Synthetic underlying asset: Long call plus
a short put and bonds . . . . . . . . . . 104
2.11.3.3. The synthetic put: put-call parity
relationship . . . . . . . . . . . . . . . . 105
Contents xxi
2.12. Conversions and Reversals . . . . . . . . . . . . . . . . . . . 106

2.13. Case study: Selling Calls (Without Holding the Stocks/as an
Alternative to Short Selling Stocks/the Idea of Selling Calls
is Also an Alternative to Buying Puts) . . . . . . . . . . . . 107
2.13.1. Data and assumptions . . . . . . . . . . . . . . . . 107
2.13.1.1. Selling calls (without holding the stock) 107
2.13.1.2. Comparing the strategy of selling calls
(with a short portfolio of stocks): the
extreme case . . . . . . . . . . . . . . . . 109
2.13.1.3. Selling calls (holding the stock) . . . . . 110
2.13.2. Leverage in selling call options (without holding the
stocks) . . . . . . . . . . . . . . . . . . . . . . . . . 110
2.13.2.1. Selling Call options (without holding the
stocks) . . . . . . . . . . . . . . . . . . . 111
2.13.2.2. Leverage in selling Call options (without
holding the stocks): The extreme case . 113
2.13.2.3. Selling calls using leverage (and holding
the stock) . . . . . . . . . . . . . . . . . 113
2.13.3. Short sale of the stocks without options . . . . . . 113
2.14. Buying Calls on EMA . . . . . . . . . . . . . . . . . . . . . 116
2.14.1. Buying a call as an alternative to buying the stock:
(also as an alternative to short sell put options) . . 116
2.14.1.1. Data and assumptions . . . . . . . . . . 116
2.14.1.2. Pattern of risk and return . . . . . . . . 116
2.14.2. Compare buying calls (as an alternative to portfolio
of stocks) . . . . . . . . . . . . . . . . . . . . . . . 117
2.14.2.1. Risk return in options . . . . . . . . . . 118
2.14.3. Example by changing volatility to 20% . . . . . . . 120
2.14.3.1. Data and assumptions: . . . . . . . . . . 120
2.14.3.2. Compare buying calls (as an alternative
to portfolio of stocks.) . . . . . . . . . . 121
2.14.3.3. Leverage in buying call options (without
selling the underlying) . . . . . . . . . . 123
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Case Study: Comparisons Between put and Call Options . . . . . . 128
1. Buying Puts and Selling Puts Naked . . . . . . . . . . . . . 128
1.1. Buying puts . . . . . . . . . . . . . . . . . . . . . . 128
1.2. Selling puts . . . . . . . . . . . . . . . . . . . . . . 129
xxii Derivatives, Risk Management and Value
2. Buying and Selling Calls . . . . . . . . . . . . . . . . . . . . 131

2.1. Buying calls . . . . . . . . . . . . . . . . . . . . . . 131
2.2. Selling a call . . . . . . . . . . . . . . . . . . . . . 132
3. Strategy of Buying a Put and Hedge and Selling a Put and
Hedge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.1. Strategy of selling put and hedge: sell delta units of
the underlying . . . . . . . . . . . . . . . . . . . . 134
3.2. Strategy of buy put and hedge: buy delta units of
the underlying . . . . . . . . . . . . . . . . . . . . 135
4. Strategy of Buy Call, Sell Put, and Buy Call, Sell Put and
Hedge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5. Strategy of Buy Call, Sell Put: Equivalent to Holding the
Underlying . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6. Strategy of Buy Call, Sell Put and Hedge: Reduces Profits
and Reduces Losses . . . . . . . . . . . . . . . . . . . . . . . 138
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
CHAPTER 3. TRADING OPTIONS AND THEIR

UNDERLYING ASSET: RISK MANAGEMENT
IN DISCRETE TIME 141
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
3.1. Basic Strategies and Synthetic Positions . . . . . . . . . . . 142
3.1.1. Options and synthetic positions . . . . . . . . . . . 142
3.1.2. Long or short the underlying asset . . . . . . . . . 144
3.1.3. Long a call . . . . . . . . . . . . . . . . . . . . . . 144
3.1.4. Short call . . . . . . . . . . . . . . . . . . . . . . . 145
3.1.5. Long a put . . . . . . . . . . . . . . . . . . . . . . 147
3.1.6. Short a put . . . . . . . . . . . . . . . . . . . . . . 148
3.2. Combined Strategies . . . . . . . . . . . . . . . . . . . . . . 150
3.2.1. Long a straddle . . . . . . . . . . . . . . . . . . . . 150
3.2.2. Short a straddle . . . . . . . . . . . . . . . . . . . 152
3.2.3. Long a strangle . . . . . . . . . . . . . . . . . . . . 153
3.2.4. Short a strangle . . . . . . . . . . . . . . . . . . . 156
3.2.5. Long a tunnel . . . . . . . . . . . . . . . . . . . . . 157
3.2.6. Short a tunnel . . . . . . . . . . . . . . . . . . . . 158
3.2.7. Long a call bull spread . . . . . . . . . . . . . . . . 159
3.2.8. Long a put bull spread . . . . . . . . . . . . . . . . 159
Contents xxiii
3.2.9. Long a call bear spread . . . . . . . . . . . . . . . 161

3.2.10. Selling a put bear spread . . . . . . . . . . . . . . 162
3.2.11. Long a butterfly . . . . . . . . . . . . . . . . . . . 162
3.2.12. Short a butterfly . . . . . . . . . . . . . . . . . . . 163
3.2.13. Long a condor . . . . . . . . . . . . . . . . . . . . 163
3.2.14. Short a condor . . . . . . . . . . . . . . . . . . . . 164
3.3. How Traders Use Option Pricing Models: Parameter
Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
3.3.1. Estimation of model parameters . . . . . . . . . . 167
3.3.1.1. Historical volatility . . . . . . . . . . . . 167
3.3.1.2. Implied volatilities and option pricing
models . . . . . . . . . . . . . . . . . . . 170
3.3.2. Trading and Greek letters . . . . . . . . . . . . . . 170
3.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
PART II. PRICING DERIVATIVES AND THEIR

UNDERLYING ASSETS IN A DISCRETE-TIME
SETTING 219
CHAPTER 4. OPTION PRICING: THE DISCRETE-

TIME APPROACH FOR STOCK OPTIONS 221
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
4.1. The CRR Model for Equity Options . . . . . . . . . . . . . 222
4.1.1. The mono-periodic model . . . . . . . . . . . . . . 222
4.1.2. The multiperiodic model . . . . . . . . . . . . . . . 224
4.1.3. Applications and examples . . . . . . . . . . . . . 226
4.1.3.1. Applications of the CRR model within
two periods . . . . . . . . . . . . . . . . 226
4.1.3.2. Other applications of the binomial model
of CRR for two periods . . . . . . . . . . 228
4.1.3.3. Applications of the binomial model of
CRR for three periods . . . . . . . . . . 231
4.1.3.4. Examples with five periods . . . . . . . 234
xxiv Derivatives, Risk Management and Value
4.2. The Binomial Model and the Distributions to the Underlying

Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
4.2.1. The Put-Call parity in the presence of several
cash-distributions . . . . . . . . . . . . . . . . . . . 237
4.2.2. Early exercise of American stock options . . . . . . 237
4.2.3. The model . . . . . . . . . . . . . . . . . . . . . . 238
4.2.4. Simulations for a small number of periods . . . . . 238
4.2.5. Simulations in the presence of two dividend dates . 246
4.2.6. Simulations for different periods and several
dividends: The general case . . . . . . . . . . . . . 246
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Appendix: The Lattice Approach . . . . . . . . . . . . . . . . . . . 249
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
CHAPTER 5. CREDIT RISKS, PRICING BONDS,

INTEREST RATE INSTRUMENTS, AND THE TERM
STRUCTURE OF INTEREST RATES 259
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
5.1. Time Value of Money and the Mathematics of Bonds . . . . 260
5.1.1. Single payment formulas . . . . . . . . . . . . . . . 261
5.1.2. Uniform-series present worth factor (USPWF) and
the capital recovery factor (CRF) . . . . . . . . . . 262
5.1.3. Uniform-series compound-amount factor (USCAF )
and the sinking fund factor (SFF ) . . . . . . . . . 263
5.1.4. Nominal interest rates and continuous compounding 265
5.2. Pricing Bonds . . . . . . . . . . . . . . . . . . . . . . . . . . 266
5.2.1. A coupon-paying bond . . . . . . . . . . . . . . . . 266
5.2.2. Zero-coupon bonds . . . . . . . . . . . . . . . . . . 267
5.3. Computation of the Yield or the Internal Rate of Return . . 268
5.3.1. How to measure the yield . . . . . . . . . . . . . . 268
5.3.2. The CY . . . . . . . . . . . . . . . . . . . . . . . . 269
5.3.3. The YTM . . . . . . . . . . . . . . . . . . . . . . . 269
5.3.4. The YTC . . . . . . . . . . . . . . . . . . . . . . . 270
5.3.5. The potential yield from holding bonds . . . . . . 270
5.4. Price Volatility Measures: Duration and Convexity . . . . . 271
5.4.1. Duration . . . . . . . . . . . . . . . . . . . . . . . 271
Contents xxv
5.4.2. Duration of a bond portfolio . . . . . . . . . . . . 273

5.4.3. Modified duration . . . . . . . . . . . . . . . . . . 273
5.4.4. Price volatility measures: Convexity . . . . . . . . 274
5.5. The Yield Curve and the Theories of Interest Rates . . . . . 275
5.5.1. The shapes of the yield curve . . . . . . . . . . . . 276
5.5.2. Theories of the term structure of interest rates . . 276
5.5.2.1. The pure expectations theory . . . . . . 276
5.6. The YTM and the Theories of the Term Structure of Interest
Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
5.6.1. Computing the YTM . . . . . . . . . . . . . . . . 277
5.6.2. Market segmentation theory of the term structure 278
5.7. Spot Rates and Forward Interest Rates . . . . . . . . . . . . 279
5.7.1. The theoretical spot rate . . . . . . . . . . . . . . 279
5.7.2. Forward rates . . . . . . . . . . . . . . . . . . . . . 279
5.8. Issuing and Redeeming Bonds . . . . . . . . . . . . . . . . . 281
5.9. Mortgage-Backed Securities: The Monthly Mortgage
Payments for a Level-Payment Fixed-Rate Mortgage . . . . 283
5.10. Interest Rate Swaps . . . . . . . . . . . . . . . . . . . . . . 286
5.10.1. The pricing of interest rate swaps . . . . . . . . . . 286
5.10.2. The swap value as the difference between the prices
of two bonds . . . . . . . . . . . . . . . . . . . . . 286
5.10.3. The valuation of currency swaps . . . . . . . . . . 287
5.10.4. Computing the swap . . . . . . . . . . . . . . . . . 289
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
CHAPTER 6. EXTENSIONS OF SIMPLE BINOMIAL

OPTION PRICING MODELS TO INTEREST RATES AND
CREDIT RISK 293
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
6.1. The Rendleman and Bartter Model (for details, refer to
Bellalah et al., 1998) for Interest-Rate Sensitive Instruments 294
6.1.1. Using the model for coupon-paying bonds . . . . . 296
6.2. Ho and Lee Model for Interest Rates and Bond Options . . 297
6.2.1. The binomial dynamics of the term structure . . . 297
6.2.2. The binomial dynamics of bond prices . . . . . . . 298
xxvi Derivatives, Risk Management and Value
6.2.3. Computation of bond prices in the Ho and Lee

model . . . . . . . . . . . . . . . . . . . . . . . . . 298
6.2.4. Option pricing in the Ho and Lee model . . . . . . 299
6.2.5. Deficiency in the Ho and Lee model . . . . . . . . 302
6.3. Binomial Interest-Rate Trees and the Log-Normal Random
Walk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
6.4. The Black-Derman-Toy Model (BDT) . . . . . . . . . . . . 308
6.4.1. Examples and applications . . . . . . . . . . . . . 309
6.5. Trinomial Interest-Rate Trees and the Pricing of Bonds . . 313
6.5.1. The model . . . . . . . . . . . . . . . . . . . . . . 313
6.5.2. Applications of the binomial and trinomial models 316
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Appendix A: Ho and Lee model and binomial dynamics
of bond prices . . . . . . . . . . . . . . . . . . . . . . . . . . 321
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
CHAPTER 7. DERIVATIVES AND PATH-DEPENDENT

DERIVATIVES: EXTENSIONS AND GENERALIZATIONS
OF THE LATTICE APPROACH BY ACCOUNTING FOR
INFORMATION COSTS AND ILLIQUIDITY 327
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
7.1. The Standard Lattice Approach for Equity Options: The
Standard Analysis . . . . . . . . . . . . . . . . . . . . . . . 329
7.1.1. The model for options on a spot asset with any pay
outs . . . . . . . . . . . . . . . . . . . . . . . . . . 329
7.1.2. The model for futures options . . . . . . . . . . . . 330
7.1.3. The model with dividends . . . . . . . . . . . . . . 330
7.1.3.1. A known dividend yield . . . . . . . . . 331
7.1.3.2. A known proportional dividend yield . . 331
7.1.3.3. A known discrete dividend . . . . . . . . 332
7.1.4. Examples . . . . . . . . . . . . . . . . . . . . . . . 332
7.1.4.1. The European put price with dividends 333
7.1.4.2. The American put price with dividends 333
7.2. A Simple Extension to Account for Information Uncertainty
in the Valuation of Futures and Options . . . . . . . . . . . 338
Contents xxvii
7.2.1. On the valuation of derivatives and information

costs . . . . . . . . . . . . . . . . . . . . . . . . . . 338
7.2.2. The valuation of forward and futures contracts in
the presence of information costs . . . . . . . . . . 340
7.2.2.1. Forward, futures, and arbitrage . . . . . 340
7.2.2.2. The valuation of forward contracts in the
absence of distributions to the underlying
asset . . . . . . . . . . . . . . . . . . . . 340
presence of a known cash income to the
underlying asset . . . . . . . . . . . . . . 341
presence of a known dividend yield to the
underlying asset . . . . . . . . . . . . . . 341
7.2.2.5. The valuation of stock index futures . . 342
7.2.2.6. The valuation of Forward and futures
contracts on currencies . . . . . . . . . . 342
7.2.2.7. The valuation of futures contracts on
silver and gold . . . . . . . . . . . . . . 343
7.2.2.8. The valuation of Futures on other
commodities . . . . . . . . . . . . . . . . 343
7.2.3. Arbitrage and information costs in the lattice
approach . . . . . . . . . . . . . . . . . . . . . . . 343
7.2.4. The binomial model for options in the presence of a
continuous dividend stream and information costs 346
known dividend yield and information costs . . . . 347
discrete dividend stream and information costs . . 347
7.2.7. The binomial model for futures options in the
presence of information costs . . . . . . . . . . . . 347
7.2.8. The lattice approach for American options with
information costs and several cash distributions . . 348
7.2.8.1. The model . . . . . . . . . . . . . . . . . 348
7.3. The Binomial Model and the Risk Neutrality: Some
Important Details . . . . . . . . . . . . . . . . . . . . . . . . 349
7.3.1. The binomial parameters and risk neutrality . . . 349
7.3.2. The convergence argument . . . . . . . . . . . . . 352
xxviii Derivatives, Risk Management and Value
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
PART III. OPTION PRICING IN A

CONTINUOUS-TIME SETTING: BASIC
MODELS, EXTENSIONS AND APPLICATIONS 365
CHAPTER 8. EUROPEAN OPTION PRICING

MODELS: THE PRECURSORS OF THE BLACK–
SCHOLES–MERTON THEORY AND HOLES
DURING MARKET TURBULENCE 367
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
8.1. Precursors to the Black–Scholes Model . . . . . . . . . . . . 369
8.1.1. Bachelier formula . . . . . . . . . . . . . . . . . . . 369
8.1.2. Sprenkle formula . . . . . . . . . . . . . . . . . . . 370
8.1.3. Boness formula . . . . . . . . . . . . . . . . . . . . 371
8.1.4. Samuelson formula . . . . . . . . . . . . . . . . . . 371
8.2. How the Black–Scholes Option Formula is Obtained . . . . 372
8.2.1. The short story . . . . . . . . . . . . . . . . . . . . 372
8.2.2. The differential equation . . . . . . . . . . . . . . . 373
8.2.3. The derivation of the formula . . . . . . . . . . . . 373
8.2.4. Publication of the formula . . . . . . . . . . . . . . 374
8.2.5. Testing the formula . . . . . . . . . . . . . . . . . 374
8.3. Financial Theory and the Black–Scholes–Merton Theory . . 375
8.3.1. The Black–Scholes–Merton theory . . . . . . . . . 375
8.3.2. Analytical formulas . . . . . . . . . . . . . . . . . 376
Contents xxix
8.4. The Black–Scholes Model . . . . . . . . . . . . . . . . . . . 377

8.4.1. The Black–Scholes model and CAPM . . . . . . . 377
8.4.2. An alternative derivation of the Black–Scholes
model . . . . . . . . . . . . . . . . . . . . . . . . . 380
8.4.3. The put-call parity relationship . . . . . . . . . . . 382
8.4.4. Examples . . . . . . . . . . . . . . . . . . . . . . . 383
8.5. The Black Model for Commodity Contracts . . . . . . . . . 386
8.5.1. The model for forward, futures, and option contracts 386
8.5.2. The put-call relationship . . . . . . . . . . . . . . . 388
8.6. Application of the CAPM Model to Forward and Futures
Contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
8.6.1. An application of the model to forward and futures
contracts . . . . . . . . . . . . . . . . . . . . . . . 389
8.6.2. An application to the derivation of the commodity
option valuation . . . . . . . . . . . . . . . . . . . 390
8.6.3. An application to commodity options and
commodity futures options . . . . . . . . . . . . . 393
8.7. The Holes in the Black–Scholes–Merton Theory and the
Financial Crisis . . . . . . . . . . . . . . . . . . . . . . . . . 394
8.7.1. Volatility changes . . . . . . . . . . . . . . . . . . . 394
8.7.2. Interest rate changes . . . . . . . . . . . . . . . . . 395
8.7.3. Borrowing penalties . . . . . . . . . . . . . . . . . 396
8.7.4. Short-selling penalties . . . . . . . . . . . . . . . . 396
8.7.5. Transaction costs . . . . . . . . . . . . . . . . . . . 396
8.7.6. Taxes . . . . . . . . . . . . . . . . . . . . . . . . . 396
8.7.7. Dividends . . . . . . . . . . . . . . . . . . . . . . . 396
8.7.8. Takeovers . . . . . . . . . . . . . . . . . . . . . . . 397
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398
Appendix A. The Cumulative Normal Distribution Function . . . 399
Appendix B. The Bivariate Normal Density Function . . . . . . . 400
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
CHAPTER 9. SIMPLE EXTENSIONS AND

APPLICATIONS OF THE BLACK–SCHOLES TYPE
MODELS IN VALUATION AND RISK MANAGEMENT 403
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
xxx Derivatives, Risk Management and Value
9.1. Applications of the Black–Scholes Model . . . . . . . . . . . 404

9.1.1. Valuation and the role of equity options . . . . . . 404
9.1.2. Valuation and the role of index options . . . . . . 405
9.1.2.1. Analysis and valuation . . . . . . . . . . 405
9.1.2.2. Arbitrage between index options and
futures . . . . . . . . . . . . . . . . . . . 406
9.1.3. Valuation of options on zero-coupon bonds . . . . 407
9.1.4. Valuation and the role of short-term options on
long-term bonds . . . . . . . . . . . . . . . . . . . 408
9.1.5. Valuation of interest rate options . . . . . . . . . . 409
9.1.6. Valuation and the role of bond options: the case of
coupon-paying bonds . . . . . . . . . . . . . . . . . 410
9.1.7. The valuation of a swaption . . . . . . . . . . . . . 411
9.2. Applications of the Black’s Model . . . . . . . . . . . . . . . 413
9.2.1. Options on equity index futures . . . . . . . . . . 413
9.2.2. Options on currency forwards and options
on currency futures . . . . . . . . . . . . . . . . . . 414
9.2.2.1. Options on currency forwards . . . . . . 414
9.2.2.2. Options on currency futures . . . . . . . 414
9.2.3. The Black’s model and valuation of interest
rate caps . . . . . . . . . . . . . . . . . . . . . . . 415
9.3. The Extension to Foreign Currencies: The Garman and
Kohlhagen Model and its Applications . . . . . . . . . . . . 416
9.3.1. The currency call formula . . . . . . . . . . . . . . 416
9.3.2. The currency put formula . . . . . . . . . . . . . . 416
9.3.3. The interest-rate theorem and the pricing of
forward currency options . . . . . . . . . . . . . . 417
9.4. The Extension to Other Commodities: The Merton,
Barone-Adesi and Whaley Model, and Its Applications . . . 420
9.4.1. The model . . . . . . . . . . . . . . . . . . . . . . 420
9.4.2. An application to portfolio insurance . . . . . . . . 421
9.5. The Real World and the Black–Scholes Type Models . . . . 422
9.5.1. Volatility . . . . . . . . . . . . . . . . . . . . . . . 422
9.5.2. The hedging strategy . . . . . . . . . . . . . . . . . 422
9.5.3. The log-normal assumption . . . . . . . . . . . . . 422
9.5.4. A world of finite trading . . . . . . . . . . . . . . 423
9.5.5. Total variance . . . . . . . . . . . . . . . . . . . . 423
9.5.6. Black–Scholes as the limiting case . . . . . . . . . 423
9.5.7. Using the model to optimize hedging . . . . . . . . 424
Contents xxxi
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
CHAPTER 10. APPLICATIONS OF OPTION PRICING

MODELS TO THE MONITORING AND THE
MANAGEMENT OF PORTFOLIOS OF DERIVATIVES
IN THE REAL WORLD 439
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
10.1. Option-Price Sensitivities: Some Specific Examples . . . . . 441
10.1.1. Delta . . . . . . . . . . . . . . . . . . . . . . . . . 441
10.1.2. Gamma . . . . . . . . . . . . . . . . . . . . . . . . 442
10.1.3. Theta . . . . . . . . . . . . . . . . . . . . . . . . . 443
10.1.4. Vega . . . . . . . . . . . . . . . . . . . . . . . . . . 444
10.1.5. Rho . . . . . . . . . . . . . . . . . . . . . . . . . . 444
10.1.6. Elasticity . . . . . . . . . . . . . . . . . . . . . . . 445
10.2. Monitoring and Managing an Option Position in Real Time 445
10.2.1. Simulations and analysis of option price sensitivities
using Barone-Adesi and Whaley model . . . . . . . 446
10.2.2. Monitoring and adjusting the option position
in real time . . . . . . . . . . . . . . . . . . . . . . 451
10.2.2.1. Monitoring and managing the delta . . . 451
10.2.2.2. Monitoring and managing the gamma . 454
10.2.2.3. Monitoring and managing the theta . . . 457
10.2.2.4. Monitoring and managing the vega . . . 458
10.3. The Characteristics of Volatility Spreads . . . . . . . . . . . 459
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
Appendix A: Greek-Letter Risk Measures in Analytical Models . . 461
A.1. B–S model . . . . . . . . . . . . . . . . . . . . . . 461
A.2. Black’s Model . . . . . . . . . . . . . . . . . . . . . 462
A.3. Garman and Kohlhagen’s model . . . . . . . . . . 463
A.4. Merton’s and Barone-Adesi and Whaley’s model . 463
Appendix B: The Relationship Between Hedging Parameters . . . 464
Appendix C: The Generalized Relationship Between the Hedging
Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Appendix D: A Detailed Derivation of the Greek Letters . . . . . . 479
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
xxxii Derivatives, Risk Management and Value
PART IV. MATHEMATICAL FOUNDATIONS

OF OPTION PRICING MODELS IN A
CONTINUOUS-TIME SETTING: BASIC
CONCEPTS AND EXTENSIONS 491
CHAPTER 11. THE DYNAMICS OF ASSET PRICES

AND THE ROLE OF INFORMATION: ANALYSIS AND
APPLICATIONS IN ASSET AND RISK MANAGEMENT 493
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
11.1. Continuous Time Processes for Asset Price Dynamics . . . 495
11.1.1. Asset price dynamics and Wiener process . . . . . 495
11.1.2. Asset price dynamics and the generalized Wiener
process . . . . . . . . . . . . . . . . . . . . . . . . 497
11.1.3. Asset price dynamics and the Ito process . . . . . 497
11.1.4. The log-normal property . . . . . . . . . . . . . . . 499
11.1.5. Distribution of the rate of return . . . . . . . . . . 500
11.2. Ito’s Lemma and Its Applications . . . . . . . . . . . . . . . 501
11.2.1. Intuitive form . . . . . . . . . . . . . . . . . . . . . 501
11.2.2. Applications to stock prices . . . . . . . . . . . . . 505
11.2.3. Mathematical form . . . . . . . . . . . . . . . . . . 505
11.2.4. The generalized Ito’s formula . . . . . . . . . . . . 508
11.2.5. Other applications of Ito’s formula . . . . . . . . . 510
11.3. Taylor Series, Ito’s Theorem and the Replication Argument 513
11.3.1. The relationship between Taylor series and Ito’s
differential . . . . . . . . . . . . . . . . . . . . . . 513
11.3.2. Ito’s differential and the replication portfolio . . . 514
11.3.3. Ito’s differential and the arbitrage portfolio . . . . 515
11.3.4. Why are error terms neglected? . . . . . . . . . . . 517
11.4. Forward and Backward Equations . . . . . . . . . . . . . . 518
11.5. The Main Concepts in Bond Markets and the General
Arbitrage Principle . . . . . . . . . . . . . . . . . . . . . . . 519
11.5.1. The main concepts in bond pricing . . . . . . . . . 519
11.5.2. Time-dependent interest rates and information
uncertainty . . . . . . . . . . . . . . . . . . . . . . 521
11.5.3. The general arbitrage principle . . . . . . . . . . . 522
11.6. Discrete Hedging and Option Pricing . . . . . . . . . . . . . 523
11.6.1. Discrete hedging . . . . . . . . . . . . . . . . . . . 523
Contents xxxiii
11.6.2. Pricing the option . . . . . . . . . . . . . . . . . . 525

11.6.3. The real distribution of returns and the
hedging error . . . . . . . . . . . . . . . . . . . . . 526
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
Appendix A: Introduction to Diffusion Processes . . . . . . . . . . 528
Appendix B: The Conditional Expectation . . . . . . . . . . . . . 529
Appendix C: Taylor Series . . . . . . . . . . . . . . . . . . . . . . . 529
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
CHAPTER 12. RISK MANAGEMENT: APPLICATIONS

TO THE PRICING OF ASSETS AND DERIVATIVES IN
COMPLETE MARKETS 535
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536
12.1. Characterization of Complete Markets . . . . . . . . . . . . 536
12.2. Pricing Derivative Assets: The Case of Stock Options . . . . 538
12.2.1. The problem . . . . . . . . . . . . . . . . . . . . . 538
12.2.2. The PDE method . . . . . . . . . . . . . . . . . . 539
12.2.3. The martingale method . . . . . . . . . . . . . . . 541
12.3. Pricing Derivative Assets: The Case of Bond Options and
Interest Rate Options . . . . . . . . . . . . . . . . . . . . . 546
12.3.1. Arbitrage-free family of bond prices . . . . . . . . 546
12.3.2. Time-homogeneous models . . . . . . . . . . . . . 547
12.3.3. Time-inhomogeneous models . . . . . . . . . . . . 550
12.4. Asset Pricing in Complete Markets: Changing Numeraire
and Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
12.4.1. Assumptions and the valuation context . . . . . . 551
12.4.2. Valuation of derivatives in a standard
Black–Scholes–Merton economy . . . . . . . . . . . 552
12.4.3. Changing numeraire and time: The martingale
approach and the PDE approach . . . . . . . . . . 554
12.5. Valuation in an Extended Black and Scholes Economy
in the Presence of Information Costs . . . . . . . . . . . . . 560
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Appendix A: The Change in Probability and the Girsanov Theorem 564
xxxiv Derivatives, Risk Management and Value
Appendix B: Resolution of the Partial Differential Equation for

a European Call Option on a Non-Dividend Paying Stock
in the Standard Context . . . . . . . . . . . . . . . . . . . . 565
Appendix C: Approximation of the Cumulative Normal Distribution 571
Appendix D: Leibniz’s Rule for Integral Differentiation . . . . . . . 572
Appendix E: Pricing Bonds: Mathematical Foundations . . . . . . 573
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
CHAPTER 13. SIMPLE EXTENSIONS AND

GENERALIZATIONS OF THE BLACK–SCHOLES TYPE
MODELS IN THE PRESENCE OF INFORMATION COSTS 583
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583
13.1. Differential Equation for a Derivative Security on a Spot
Asset in the Presence of a Continuous Dividend Yield and
Information Costs . . . . . . . . . . . . . . . . . . . . . . . 584
13.2. The Valuation of Securities Dependent on Several Variables
in the Presence
of Incomplete Information: A General Method . . . . . . . . 585
13.3. The General Differential Equation for the Pricing of
Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
13.4. Extension of the Risk-Neutral Argument in the Presence of
Information Costs . . . . . . . . . . . . . . . . . . . . . . . 589
13.5. Extension to Commodity Futures Prices within Incomplete
Information . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
13.5.1. Differential equation for a derivative security
dependent on a futures price in the presence
of information costs . . . . . . . . . . . . . . . . . 590
13.5.2. Commodity futures prices . . . . . . . . . . . . . . 592
13.5.3. Convenience yields . . . . . . . . . . . . . . . . . 592
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Appendix A: A General Equation for Derivative Securities . . . . . 594
Appendix B: Extension to the Risk-Neutral Valuation Argument . 596
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
Contents xxxv
PART V. EXTENSIONS OF OPTION PRICING

THEORY TO AMERICAN OPTIONS AND
INTEREST RATE INSTRUMENTS IN A
CONTINUOUS-TIME SETTING: DIVIDENDS,
COUPONS AND STOCHASTIC INTEREST RATES 613
CHAPTER 14. EXTENSION OF ASSET AND RISK

MANAGEMENT IN THE PRESENCE OF
AMERICAN OPTIONS: DIVIDENDS, EARLY EXERCISE,
AND INFORMATION UNCERTAINTY 615
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616
14.1. The Valuation of American Options: The General Problem 618
14.1.1. Early exercise of American calls . . . . . . . . . . . 618
14.1.2. Early exercise of American puts . . . . . . . . . . . 620
14.1.3. The American put option and its critical stock price 623
14.2. Valuation of American Commodity Options and Futures
Options with Continuous Distributions . . . . . . . . . . . . 627
14.2.1. Valuation of American commodity options . . . . . 627
14.2.3. Valuation of American futures options . . . . . . . 632
14.3. Valuation of American Commodity and Futures Options with
Continuous Distributions within Information Uncertainty . 634
14.3.1. Commodity option valuation with information costs 634
14.3.2. Simulation results . . . . . . . . . . . . . . . . . . 638
14.4. Valuation of American Options with Discrete
Cash-Distributions . . . . . . . . . . . . . . . . . . . . . . . 640
14.4.1. Early exercise of American options . . . . . . . . 641
14.4.2. Valuation of American options with dividends . . . 642
14.5. Valuation of American Options with Discrete Cash
Distributions within Information Uncertainty . . . . . . . . 645
14.5.1. The model . . . . . . . . . . . . . . . . . . . . . . 645
14.6. The Valuation Equations for Standard and Compound
Options with Information Costs . . . . . . . . . . . . . . . . 648
14.6.1. The pricing of assets under incomplete information 650
14.6.2. The valuation of equity as a compound option . . 650
xxxvi Derivatives, Risk Management and Value
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
Appendix A: An Alternative Derivation of the Compound Option’s
Formula Using the Martingale Approach . . . . . . . . . . . 656
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664
CHAPTER 15. RISK MANAGEMENT OF BONDS AND

INTEREST RATE SENSITIVE INSTRUMENTS IN THE
PRESENCE OF STOCHASTIC INTEREST RATES AND
INFORMATION UNCERTAINTY: THEORY AND TESTS 667
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 667
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668
15.1. The Valuation of Bond Options and Interest Rate Options . 669
15.1.1. The problems in using the B–S model for
interest-rate options . . . . . . . . . . . . . . . . . 669
15.1.2. Sensitivity of the theoretical option prices to
changes in factors . . . . . . . . . . . . . . . . . . 670
15.2. A Simple Non-Parametric Approach to Bond Futures
Option Pricing . . . . . . . . . . . . . . . . . . . . . . . . . 670
15.2.1. Canonical modeling and option pricing theory . . . 671
15.2.2. Assessing the distribution of the underlying futures
price . . . . . . . . . . . . . . . . . . . . . . . . . . 672
15.2.3. Transforming actual probabilities into risk-neutral
probabilities . . . . . . . . . . . . . . . . . . . . . . 672
15.2.4. Qualitative comparison of Black and canonical
model values . . . . . . . . . . . . . . . . . . . . . 673
15.3. One-Factor Interest Rate Modeling and the Pricing of Bonds:
The General Case . . . . . . . . . . . . . . . . . . . . . . . . 673
15.3.1. Bond pricing in the general case: The arbitrage
argument and information costs . . . . . . . . . . . 673
15.3.2. Pricing callable bonds within information
uncertainty . . . . . . . . . . . . . . . . . . . . . . 676
15.4. Fixed Income Instruments as a Weighted Portfolio of
Power Options . . . . . . . . . . . . . . . . . . . . . . . . . 676
15.5. Merton’s Model for Equity Options in the Presence
of Stochastic Interest Rates: Two-Factor Models . . . . . . 678
Contents xxxvii
15.5.1. The model in the presence of stochastic interest

rates . . . . . . . . . . . . . . . . . . . . . . . . . . 679
15.5.2. Applications of Merton’s model . . . . . . . . . . . 680
15.6. Some Models for the Pricing of Bond Options . . . . . . . . 681
15.6.1. An extension of the Ho-Lee model for bond options 681
15.6.2. The Schaefer and Schwartz model . . . . . . . . . 683
15.6.3. The Vasicek (1977) model . . . . . . . . . . . . . . 683
15.6.4. The Ho and Lee model . . . . . . . . . . . . . . . . 684
15.6.5. The Hull and White model . . . . . . . . . . . . . 685
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
Appendix A: Government Bond Futures and Implicit Embedded
Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
A.1. Criteria for the CTD . . . . . . . . . . . . . . . . . 688
A.2. Yield changes . . . . . . . . . . . . . . . . . . . . . 688
A.3. The value for a futures position . . . . . . . . . . . 690
A.4. Parallel yield shift . . . . . . . . . . . . . . . . . . 691
A.5. Relative yield shift . . . . . . . . . . . . . . . . . . 692
Appendix B: One-Factor Fallacies for Interest Rate Models . . . . 692
B.1. The models in practice . . . . . . . . . . . . . . . . 693
B.2. Spreads between rates . . . . . . . . . . . . . . . . 693
Appendix C: Merton’s Model in the Presence of Stochastic
Interest Rates . . . . . . . . . . . . . . . . . . . . . . . . . . 694
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
CHAPTER 16. MODELS OF INTEREST RATES,

INTEREST-RATE SENSITIVE INSTRUMENTS, AND
THE PRICING OF BONDS: THEORY AND TESTS 703
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 704
16.1. Interest Rates and Interest-Rate Sensitive Instruments . . . 705
16.1.1. Zero-coupon bonds . . . . . . . . . . . . . . . . . . 705
16.1.2. Term structure of interest rates . . . . . . . . . . . 705
16.1.3. Forward interest rates . . . . . . . . . . . . . . . . 706
16.1.4. Short-term interest rate . . . . . . . . . . . . . . . 707
16.1.5. Coupon-bearing bonds . . . . . . . . . . . . . . . . 707
16.1.6. Yield-to-Maturity (YTM) . . . . . . . . . . . . . . 708
16.1.7. Market conventions . . . . . . . . . . . . . . . . . . 709
xxxviii Derivatives, Risk Management and Value
16.2. Interest Rates and the Pricing of Bonds . . . . . . . . . . . 710

16.2.1. The instantaneous interest rates under certainty . 710
16.2.2. The instantaneous interest rate under uncertainty 711
16.3. Interest Rate Processes and the Pricing of Bonds and Options 712
16.3.1. The Vasicek model . . . . . . . . . . . . . . . . . . 713
16.3.2. The Brennan and Schwartz model . . . . . . . . . 713
16.3.3. The CIR model . . . . . . . . . . . . . . . . . . . . 713
16.3.4. The Ho and Lee model . . . . . . . . . . . . . . . . 714
16.3.5. The HJM model . . . . . . . . . . . . . . . . . . . 714
16.3.6. The BDT model . . . . . . . . . . . . . . . . . . . 716
16.3.8. Fong and Vasicek model . . . . . . . . . . . . . . . 718
16.3.9. Longstaff and Schwartz model . . . . . . . . . . . . 718
16.4. The Relative Merits of the Competing Models . . . . . . . . 718
16.5. A Comparative Analysis of Term Structure Estimation
Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
16.5.1. The construction of the term structure and coupon
bonds . . . . . . . . . . . . . . . . . . . . . . . . . 721
16.5.2. Fitting functions and estimation procedure . . . . 722
16.6. Term Premium Estimates From Zero-Coupon Bonds: New
Evidence on the Expectations Hypothesis . . . . . . . . . . 724
16.7. Distributional Properties of Spot and Forward Interest Rates:
USD, DEM, GBP, and JPY . . . . . . . . . . . . . . . . . . 726
16.7.1. Interest rate levels . . . . . . . . . . . . . . . . . . 728
16.7.2. Interest rate differences and log differences . . . . 728
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 731
Appendix A: An Application of Interest Rate Models to Account for
Information Costs: An Exercise . . . . . . . . . . . . . . . . 732
A.1. An application of the HJM model in the presence of
information costs . . . . . . . . . . . . . . . . . . . 732
A.1.1. The forward rate equation . . . . . . . . 732
A.1.2. The spot rate process . . . . . . . . . . . 733
A.1.3. The market price of risk . . . . . . . . . 734
A.1.4. Relationship between risk-neutral forward
rate drift and volatility . . . . . . . . . . 735
A.1.5. Pricing derivatives . . . . . . . . . . . . 735
A.2. An application of the Ho and Lee model in the
presence of information cost . . . . . . . . . . . . . 736
Contents xxxix
Appendix B: Implementation of the BDT Model with Different

Volatility Estimators . . . . . . . . . . . . . . . . . . . . . . 737
B.1. The BDT model . . . . . . . . . . . . . . . . . . . 737
B.2. Estimation results . . . . . . . . . . . . . . . . . . 738
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 739
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 740
PART VI. GENERALIZATION OF OPTION

PRICING MODELS AND STOCHASTIC VOLATILITY 743
CHAPTER 17. EXTREME MARKET MOVEMENTS,

RISK AND ASSET MANAGEMENT: GENERALIZATION
TO JUMP PROCESSES, STOCHASTIC VOLATILITIES,
AND INFORMATION COSTS 745
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
17.1. The Jump-Diffusion and the Constant Elasticity of Variance
Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747
17.1.1. The jump-diffusion model . . . . . . . . . . . . . . 747
17.1.2. The constant elasticity of variance diffusion (CEV)
process . . . . . . . . . . . . . . . . . . . . . . . . 748
17.2. On Jumps, Hedging and Information Costs . . . . . . . . . 749
17.2.1. Hedging in the presence of jumps . . . . . . . . . . 750
17.2.2. Hedging the jumps . . . . . . . . . . . . . . . . . . 752
17.2.3. Jump volatility . . . . . . . . . . . . . . . . . . . . 753
17.3. On the Smile Effect and Market Imperfections in the
Presence of Jumps and Incomplete Information . . . . . . . 754
17.3.1. On smiles and jumps . . . . . . . . . . . . . . . . . 754
17.3.2. On smiles, jumps, and incomplete information . . 759
17.3.3. Empirical results in the presence of jumps and
incomplete information . . . . . . . . . . . . . . . 760
17.4. Implied Volatility and Option Pricing Models: The Model
and Simulation Results . . . . . . . . . . . . . . . . . . . . . 763
17.4.1. The valuation model . . . . . . . . . . . . . . . . . 763
17.4.3. Model calibration and the smile effect . . . . . . . 766
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
xl Derivatives, Risk Management and Value
CHAPTER 18. RISK MANAGEMENT DURING

ABNORMAL MARKET CONDITIONS: FURTHER
GENERALIZATION TO JUMP PROCESSES,
STOCHASTIC VOLATILITIES, AND INFORMATION COSTS 771
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 771
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
18.1. Option Pricing in the Presence of a Stochastic Volatility . . 774
18.1.2. Stein and Stein model . . . . . . . . . . . . . . . . 775
18.1.3. The Heston model . . . . . . . . . . . . . . . . . . 777
18.1.4. The Hoffman, Platen, and Schweizer model . . . . 778
18.1.5. Market price of volatility risk . . . . . . . . . . . . 781
18.1.6. The market price of risk for traded assets . . . . . 782
18.2. Generalization of Some Models with Stochastic Volatility
and Information Costs . . . . . . . . . . . . . . . . . . . . . 782
18.2.1. Generalization of the Hull and White (1987) model 782
18.2.2. Generalization of Wiggins’s model . . . . . . . . . 784
18.2.3. Generalization of Stein and Stein’s model . . . . . 785
18.2.4. Generalization of Heston’s model . . . . . . . . . . 786
18.2.5. Generalization of Johnson and Shanno’s model . . 788
18.3. The Volatility Smiles: Some Standard Results . . . . . . . . 788
18.3.1. The smile effect in stock options and index options 788
18.3.2. The smile effect for bond and currency options . . 789
18.3.3. Volatility smiles: Empirical evidence . . . . . . . . 790
18.4. Empirical Results Regarding Information Costs and Option
Pricing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
18.4.1. Information costs and option pricing:
The estimation method . . . . . . . . . . . . . . . 791
18.4.2. The asymmetric distortion of the smile . . . . . . . 792
18.4.3. Asymmetric Smiles and information costs in
a stochastic volatility model . . . . . . . . . . . . . 793
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796
Contents xli
PART VII. OPTION PRICING MODELS AND

NUMERICAL ANALYSIS 799
CHAPTER 19. RISK MANAGEMENT, NUMERICAL

METHODS AND OPTION PRICING 801
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801
19.1. Numerical Analysis and Simulation Techniques: An
Introduction to Finite Difference Methods . . . . . . . . . . 803
19.1.1. The implicit difference scheme . . . . . . . . . . . 804
19.1.2. Explicit difference scheme . . . . . . . . . . . . . . 806
19.1.3. An extension to account for information costs . . . 807
19.2. Application to European Options on Non-Dividend Paying
Stocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
19.2.1. The analytic solution . . . . . . . . . . . . . . . . . 808
19.2.2. The numerical solution . . . . . . . . . . . . . . . . 808
19.2.3. An application to European calls on non-dividend
paying stocks in the presence of information costs 810
19.3. Valuation of American Options with a Composite Volatility 811
19.3.1. The effect of interest rate volatility on the index
volatility . . . . . . . . . . . . . . . . . . . . . . . 811
19.3.2. Valuation of index options with a composite
volatility . . . . . . . . . . . . . . . . . . . . . . . 812
19.3.3. Numerical solutions and simulations . . . . . . . . 813
19.4. Simulation Methods: Monte–Carlo Method . . . . . . . . . 817
19.4.1. Simulation of Gaussian variables . . . . . . . . . . 818
19.4.2. Relationship between option values and simulation
methods . . . . . . . . . . . . . . . . . . . . . . . 818
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 819
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 820
Appendix A: Simple Concepts in Numerical Analysis . . . . . . . . 820
Appendix B: An Algorithm for a European Call . . . . . . . . . . 822
Appendix C: The Algorithm for the Valuation of American
Long-term Index Options with a Composite Volatility . . . 823
xlii Derivatives, Risk Management and Value
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 826
Appendix D: The Monte–Carlo Method and the Dynamics of
Asset Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . 829
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 830
CHAPTER 20. NUMERICAL METHODS AND PARTIAL

DIFFERENTIAL EQUATIONS FOR EUROPEAN AND
AMERICAN DERIVATIVES WITH COMPLETE AND
INCOMPLETE INFORMATION 833
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 833
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 834
20.1. Valuation of American Calls on Dividend-Paying Stocks . . 835
20.1.1. The Schwartz model . . . . . . . . . . . . . . . . . 835
20.2. American Puts on Dividend-Paying Stocks . . . . . . . . . . 837
20.2.1. The Brennan and Schwartz model . . . . . . . . . 837
20.3. Numerical Procedures in the Presence of Information Costs:
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . 839
20.3.1. Finite difference methods in the presence of
information costs . . . . . . . . . . . . . . . . . . . 839
20.3.2. An application to the American put using explicit
or implicit finite difference methods . . . . . . . . 841
20.4. Convertible Bonds . . . . . . . . . . . . . . . . . . . . . . . 842
20.4.1. Specific features of CB . . . . . . . . . . . . . . . . 842
20.4.2. The valuation equations . . . . . . . . . . . . . . . 843
20.4.4. Simulations . . . . . . . . . . . . . . . . . . . . . . 847
20.5. Two-Factor Interest Rate Models and Bond Pricing within
Information Uncertainty . . . . . . . . . . . . . . . . . . . . 847
20.6. CBs Pricing within Information Uncertainty . . . . . . . . . 850
20.6.1. The pricing of CBs . . . . . . . . . . . . . . . . . . 850
20.6.2. Specific call and put features . . . . . . . . . . . . 851
20.6.3. The pricing of CBs in two-factor models within
information uncertainty . . . . . . . . . . . . . . . 851
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 852
Appendix A: A Discretizing Strategy for Mean-Reverting Models . 853
Appendix B: An Algorithm for the American Call with Dividends 861
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
PART VIII. EXOTIC DERIVATIVES 875
CHAPTER 21. RISK MANAGEMENT: EXOTICS AND

SECOND-GENERATION OPTIONS 877
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 877
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 877
21.1. Exchange Options . . . . . . . . . . . . . . . . . . . . . . . 879
21.2. Forward-Start Options . . . . . . . . . . . . . . . . . . . . . 880
21.3. Pay-Later Options . . . . . . . . . . . . . . . . . . . . . . . 882
21.4. Simple Chooser Options . . . . . . . . . . . . . . . . . . . . 884
21.5. Complex Choosers . . . . . . . . . . . . . . . . . . . . . . . 885
21.6. Compound Options . . . . . . . . . . . . . . . . . . . . . . . 886
21.6.1. The call on a call in the presence of a cost of carry 887
21.6.2. The put on a call in the presence of a cost of carry 888
21.6.3. The formula for a call on a put in the presence of a
cost of carry . . . . . . . . . . . . . . . . . . . . . 888
21.6.4. The put on a put in the presence of a cost of carry 889
21.7. Options on the Maximum (Minimum) . . . . . . . . . . . . 889
21.7.1. The call on the minimum of two assets . . . . . . . 891
21.7.2. The call on the maximum of two assets . . . . . . 892
21.7.3. The put on the minimum (maximum) of two assets 892
21.8. Extendible Options . . . . . . . . . . . . . . . . . . . . . . . 893
21.8.1. The valuation context . . . . . . . . . . . . . . . . 893
21.8.2. Extendible calls . . . . . . . . . . . . . . . . . . . . 894
21.9. Equity-Linked Foreign Exchange Options and Quantos . . . 896
21.9.1. The foreign equity call struck in foreign currency . 898
21.9.2. The foreign equity call struck in domestic currency 898
21.9.3. Fixed exchange rate foreign equity call . . . . . . . 899
21.9.4. An equity-linked foreign exchange call . . . . . . . 900
21.10. Binary Barrier Options . . . . . . . . . . . . . . . . . . . . . 901
21.10.1. Path-independent binary options . . . . . . . . . . 902
xliv Derivatives, Risk Management and Value
21.10.1.1. Standard cash-or-nothing options . . . . 902

21.10.1.2. Cash-or-nothing options with shadow
costs . . . . . . . . . . . . . . . . . . . . 903
21.10.1.3. Standard asset-or-nothing options . . . . 904
21.10.1.4. Asset-or-nothing options with shadow
costs . . . . . . . . . . . . . . . . . . . . 905
21.10.1.5. Standard gap options . . . . . . . . . . . 906
21.10.1.6. Gap options with shadow costs . . . . . 908
21.10.1.7. Supershares . . . . . . . . . . . . . . . . 908
21.11. Lookback Options . . . . . . . . . . . . . . . . . . . . . . . 908
21.11.1. Standard lookback options . . . . . . . . . . . . . 909
21.11.2. Options on extrema . . . . . . . . . . . . . . . . . 909
21.11.2.1. On the maximum . . . . . . . . . . . . . 909
21.11.2.2. On the minimum . . . . . . . . . . . . . 910
21.11.3. Limited risk options . . . . . . . . . . . . . . . . . 910
21.11.4. Partial lookback options . . . . . . . . . . . . . . . 911
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913
Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 914
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 914
CHAPTER 22. VALUE AT RISK, CREDIT RISK,

AND CREDIT DERIVATIVES 917
Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 917
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 917
22.1. VaR and Riskmetrics: Definitions and Basic Concepts . . . 919
22.1.1. The definition of risk . . . . . . . . . . . . . . . . . 920
22.1.2. VaR: Definition . . . . . . . . . . . . . . . . . . . . 920
22.2. Statistical and Probability Foundation of VaR . . . . . . . . 921
22.2.1. Using percentiles or quantiles to measure market
risk . . . . . . . . . . . . . . . . . . . . . . . . . . 922
22.2.2. The choice of the horizon . . . . . . . . . . . . . . 922
22.3. A More Advanced Approach to VaR . . . . . . . . . . . . . 923
22.4. Credit Valuation and the Creditmetrics Approach . . . . . . 927
22.4.1. The portfolio context of credit . . . . . . . . . . . 927
22.4.2. Different credit risk measures . . . . . . . . . . . . 927
22.4.3. Stand alone or single exposure risk calculation . . 928
22.4.4. Differing exposure type . . . . . . . . . . . . . . . 928
Contents xlv
22.5. Default and Credit-Quality Migration in the Creditmetrics

Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 929
22.5.1. Default . . . . . . . . . . . . . . . . . . . . . . . . 929
22.5.2. Credit-quality migration . . . . . . . . . . . . . . . 929
22.5.3. Historical tabulation and recovery rates . . . . . . 930
22.6. Credit-Quality Correlations . . . . . . . . . . . . . . . . . . 930
22.7. Portfolio Management of Default Risk in the Kealhofer,
McQuown and Vasicek (KMV) Approach . . . . . . . . . . 932
22.7.1. The model of default risk . . . . . . . . . . . . . . 932
22.7.2. Asset market value and volatility . . . . . . . . . . 933
22.8. Credit Derivatives: Definitions and Main Concepts . . . . . 933
22.8.1. Forward contracts . . . . . . . . . . . . . . . . . . 933
22.8.2. The structure of credit-default instruments . . . . 934
22.8.2.1. Total return swaps . . . . . . . . . . . . 934
22.8.2.2. Credit-default swaps . . . . . . . . . . . 934
22.8.2.3. Basket default swaps . . . . . . . . . . . 935
22.8.2.4. Credit-default exchange swap . . . . . . 935
22.8.2.5. Credit-linked notes (CLNs) . . . . . . . 936
22.9. The Rating Agencies Models and the Proprietary Models . 936
22.9.1. The rating agencies models . . . . . . . . . . . . . 936
22.9.2. The proprietary models . . . . . . . . . . . . . . . 938
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 940
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 941
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 943
September 10, 2009 14:41 spi-b708 9in x 6in b708-ch01
Part I
Financial Markets and Financial Instruments: Basic

Concepts and Strategies
1
2
Chapter 1
FINANCIAL MARKETS, FINANCIAL

INSTRUMENTS, AND FINANCIAL CRISIS
Chapter Outline
This chapter is organized as follows:
1. Section 1.1 presents the trading characteristics of commodity contracts.

The analysis concerns mainly oil markets.
2. Section 1.2 studies the main trading characteristics of commodity
markets and instruments. The analysis concerns the instruments in the
International Petroleum Exchange.
3. Section 1.3 develops the characteristics of crude oils and the properties
of petroleum products.
4. Section 1.4 presents the trading characteristics of another commodity:
Cocoa.
5. Section 1.5 illustrates the main trading characteristics of options.
The analysis pertains mainly to equity options.
6. Section 1.6 studies the trading characteristics of options on currency
forwards and futures.
7. Section 1.7 develops the trading characteristics of options. The analysis
pertains mainly to bonds and bond options markets.
8. Section 1.8 provides examples of simple and complex financial instru-
ments.
9. Section 1.9 provides the main reasons behind financial innovations.
3
4 Derivatives, Risk Management and Value
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.
What are derivatives?

A derivative is a generic term to encompass all financial transactions,
which are not directly traded in the primary physical market. It refers to
a financial instrument to manage a given risk. The term includes forwards,
futures, options, commodity contracts, etc.
What is a forward contract

They are the simplest and most basic hedging instruments. A forward con-
tract is an agreement between two parties to set the price today for a trans-
action that will not be completed until a specified date in the future. An
example is a forward contract for $1 million, to be delivered in six months,
at a price of 5.30 for a dollar. These terms obligate the seller of the contract
to deliver $1 million on the company’s account, for the price set today. On
the other hand, the buyer has no alternative than to accept delivery under
the terms of the contract. The only possibility for the buyer to cancel the
contract at a later date, is to enter into a reverse forward contract, with the
same bank or another institution, but at the risk of a loss (or a gain) since
Financial Markets, Financial Instruments, and Financial Crisis 5
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.
What is a futures contract?

A future is an exchange-traded contract between a buyer and seller and
the clearinghouse of a futures exchange to buy or sell a standard quantity
and quality of a commodity at a specified future date and price. The
clearinghouse acts as a counterpart in all transactions and is responsible for
holding traders’ surety bonds to guarantee that transactions are completed.
Like forward contracts, futures contracts are used to lock in the interest
rate, exchange rate, or commodity price. But, futures markets are organized
in such a way that the risk of default is always completely eliminated. This
is possible by trading futures contracts on an organized exchange with a
clearinghouse which steps in between a buyer and a seller, each time a deal
is struck in the pit. The clearinghouse adopts the position of the buyer to
every seller, and of the seller to every buyer, i.e., the clearinghouse keeps a
zero net position. This means that every trader in the futures markets has
obligations only to the clearinghouse, and has strong expectations that the
clearinghouse will maintain its side of the bargain as well. The credibility
of the system is maintained through the requirements of margin and daily
settlements. The main purpose of the margin is to provide a safeguard
to ensure that traders will perform their obligations. It is usually set to
the maximum loss a trader can experience in a normal trading day. Daily
settlements or marking to market just consists in a transfer of cash from
one account to another. The elimination of default risk has a cost: contracts
are standardized in order to bring liquidity to the market, there are only
a few financial assets which are traded on futures markets and they do
not necessarily correspond to the risk to be hedged. Therefore, there is no
perfect hedge with futures contracts. The hedgers keep what is called a
basis risk and a correlation risk which cannot be fully eliminated.
What are standard options?

Options are more flexible than forwards and futures in the sense they
provide the buyer with the protection needed, and leave him/her with the
full benefits associated with a favorable development of the asset price,

interest rate, or exchange rate. This nice feature has a price: the option
premium. On the contrary, forward and futures prices are set at a level
such that the initial price of the contract is exactly zero. A standard or a
vanilla option is a security that gives its holder the right to buy or sell the
underlying asset within a specified period of time, at a given price, called
the strike price, striking price or strike price and no obligation to deliver.
A European option can be exercised only on the last day of the contract,
called the maturity date or the expiration date. An American option can
be exercised at any time during the contract’s life.
What are commodity contracts?

Commodity contracts are traded around the world. One of the main
examples is oil. Oil has become one of the biggest commodity market in
the world. Oil trading evolved from a primarily physical activity into a
financial market. The physical oil market trades different types of crude
oil and refined products. Prior to 1973, oil trading was a marginal activity
and the industry was dominated by large integrated oil companies. The
structure of the industry changed in the 1970s with the nationalization of
the interests in major oil companies in the Middle East. The driving force
behind rapid growth in oil trading is explained by the huge variability in
the price of oil. Market participants are exposed to the risk of very large
changes in the value of any oil. The emergence of the 15-day Brent market
in the 1980s results mainly from the economic features of international
trade in oil and the de-integration of the industry in the 1970s. As a
consequence of nationalization in the OPEC region, the major companies
lost many of the concessions which had provided them with equity oil. The
1979 crisis and de-integration created the necessary conditions for a market
to merge. The major development in the late 1970s and early 1980s was
the emergence of two systems of oil price determination. There was OPEC
fixing at that time a price for a marker crude (Arabian Light) and a market
for non-OPEC crudes in which prices were subjected to the pressures of
economic forces. The developments spurred significant growth in market
activity leading to the emergence of new trading instruments. By the end
of 1985, the world market entered a new crisis. The oil shocks ended up with
compromises that changed important features of the petroleum world. The
international petroleum exchange, IPE, was established by representatives
from 28 countries in order to offer the industry the means to manage the
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.
1.1. Trading Characteristics of Commodity Contracts: The

Case of Oil
The oil market is ultimately concerned with the transportation, processing,
and storage of a raw material. Since oil is a liquid, it requires specialized
handling facilities for transportation, processing, and storage. These ele-
ments represent the basic building blocks for the physical oil market. The
behavior of prices is influenced by the fundamental forces of demand and
supply. The demand of oil depends on the state of the global economy. It is
closely linked to the growth of the economy. The oil industry is not properly
integrated. In general, oil producers maximize their output, subject to the
technical constraints of the field. Since operating costs are lower than the
sunk capital costs, oil is produced until its price reaches very low levels.
Oil is often viewed as a highly political commodity. The threat of supply
disruption remains real and political forces play an important role in the
oil market.
1.1.1. Fixed prices

Outright prices.
A contract for the sale of a cargo of oil must stipulate the basic price,
the guaranteed quality, and agreed price adjustment for quality deviations,
availability date range, etc. These factors describe the elements of the price
of a cargo of oil. Gasoil is sold in Europe at x dollars per metric tonne,
based on a specific gravity. The important qualities for crude oil are gravity,
measured in API degrees, metals content and sulfur. Crude is in general
traded in US dollars per barrel. In Europe, oil products are generally sold
in US dollars per metric tonnes. Timing can have an impact on pricing when
the market is in backwardation or contango. Backwardation corresponds to
a situation in which the price of the commodity available on a prompt
basis is higher than that for deferred delivery. Contango corresponds to a
situation in which a commodity is cheapest in the prompt position and gets
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.
Official selling prices.

Until the mid-1980s, traded crude oil, except that of US origin was priced
at an official selling price, OSP. Even if an official price is quoted for crude
oil, the price actually paid by a refiner is set in general at a premium or a
discount to the OSP.
1.1.2. Floating prices

As oil prices became volatile, there was increasing uncertainty about the
value of oil at the time it was to be loaded or discharged. As the oil market
moved away from fixed prices, oil prices reflected the market value of the
oil at the time of moving the oil. The growth of the forward market, futures
markets, swaps, and options markets developed the pricing mechanisms.
1.1.3. Exchange of futures for Physical (EFP)

Exchange of futures for physical (EFP) provide a method of pricing a cargo
of oil at a differential to the futures market. The buyer and the seller utilize
existing futures positions that match their exposure on the physical oil
market.
The buyer of a physical cargo transfers ownership to the seller (of the
cargo) of a certain number of futures contracts, equivalent to the volume of
the cargo of oil. The value of the futures contract is used to calculate the
price of the physical oil. The seller becomes long futures contracts and the
buyer short futures contracts at the agreed level.
1.2. Description of Markets and Instruments: The Case

of the International Petroleum Exchange
Crude oil trade is a key nexus between the two main centers of activity:
upstream exploration and production and downstream refining, and mar-
keting. In this context, the price of crude oil results from the interaction
between the signals provided by product markets and the revenue objectives
of producers. The growth of the international spot market in crude oil
and the ensuing transformation in the structure of oil markets explain

the way oil is priced today. The growth in the international spot market
during the early 1980s induced the emergence of a market discovery system
driven by marginal spot trading, which replaced administered selling prices.
Wide swings in prices have fostered the growth of forward and futures
markets as well as several risk-management tools. The values of oil grades
of crude oil depend on the refined products that can be made from
individual grades. Each refined product resulting from a barrel of crude
oil has its own separate markets under the law of supply and demand.
The upgrading technologies maximize the product output from a barrel of
crude oil. The starting point of the process is the distillation of the crude
oil. This involves heating the crude oil to gradually higher temperatures
giving different types of hydrocarbons. The cracking process allows to break
lighter gasoline and gas oil fractions out of heavy gas oil and certain kinds
of residue. Refiners and oil-market participants rely on detailed assays of
actual cargoes in order to determine the specific features of an individual
crude oil. A refiner must evaluate transportation alternatives and the
price dynamics of the market. There are more than 100 crude oils in
international trade.
1.3. Characteristics of Crude Oils and Properties

of Petroleum Products
Petroleum or crude oil can be described as a viscous brown to black liquid
mixture. Petroleum contains a hydrocarbon mixture and non-hydrocarbon
compounds such as sulfur, nitrogen, and oxygen compounds.
1.3.1. Specific features of some oil contracts

The IPE Brent Crude futures contract is one of the most important energy
price indicators in the world. This contract represents the critical part of
the Brent Blend complex which represents the benchmark for two thirds of
the world’s internationally traded crude oil.
Brent crude futures

There is no maximum price fluctuation imposed upon Brent crude futures.
The contract can be settled in cash against a physical price index calculated
by the IPE. It can be settled with physical delivery through the EFP
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.
Gas oil futures

The IPE gas oil contract is a benchmark for the physical market in Europe
and beyond. This contract is used as a price basis for most middle distillate
spot trades in northwest Europe. Companies can use this contract to
evaluate arbitrage, storage, and investment opportunities.
Natural gas futures

The natural gas futures contract was launched in January 1997. Since the
launch, many changes to the contract have been made as the industry has
uncovered new needs and oportunities. The IPE natural gas futures (NBP)
contracts are traded through the IPE automated energy trading system
(ETS) or by the EFPs. The contract size corresponds to a minimum of 5 lots
of 1000 therms per lot of natural gas per day during the contract period.
The contract price is in Sterling and in pence per therm. If not closed out
at expiry, contracts obligate delivery or taking delivery on each day in that
contract period of the number of lots remaining open upon expiry.
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.
1.3.2. Description of Markets and Trading Instruments:

The Brent Market
Dated Brent and 15-day Brent
Two types of transactions can be done in the physical Brent blend market.
The first, known as “dated Brent” cargo is a conventional spot transaction
which refers to the sale of a specific cargo. The cargo is either available
in a specific loading slot or is loaded and in transit to some destination.
The second, known as “15-day Brent” is a forward deal, which refers to a
standard parcel that will be made available by the seller on an unspecified
day of the relevant month. Oil is sold f.o.b. (insurance, freight, and ocean
losses are the buyer’s responsibility), but demurrage at the terminal is the
seller’s. For 15-day Brent, the contract is a standard telex and there is no
exchange to match sellers and buyers. The clearing of the market involves
all participants. The clearing consists of two different operations book-outs
and the seller’s nominations. A book-out is simply an agreement between
some participants to cancel their contracts with a cash settlement procedure
for the difference between an agreed reference price and the contract price.
The contracts which are not cleared by a book out cancellation are cleared
through the nomination process. Sellers through the forward market serve
15-day notices to buyers of cargoes for the relevant month. The 15-day
market reveals the buyer’s uncertainty regarding the exact date of delivery
and it is characterized by a lack of perfect price/volume transparency.
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
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.
The box trade.

This strategy is implemented by taking a position on the movement of the
relative backwardation (or contango) between two crude oil markets. The
trader sells and buys simultaneously two inter-crude spreads. The strategy
involves the trading of four cargoes. For example, a Brent-Dubai box spread
involves the simultaneous purchase of a Brent cargo and the sale of a Dubai
cargo for the same delivery month and the purchase of a Dubai cargo for
another delivery month.
The IEP Brent futures contract.

The International Petroleum Exchange of London (IEP) trades Brent
futures contract on cargoes of 500,000 barrels. The physical base of the
New York Mercantile Exchange (NYMEX) contract is pipeline scheduling
at Cushing Oklahoma of 1000 barrel batches.
IPE Brent contracts represent two contracts: one between the buyer
and the clearing house and one between the seller and the clearing house.
The clearing house is the International Commodities Clearing House Ltd
(ICCH).
Exchange of futures for physical (EFPs).

An EFP is a physical link between the IPE Brent futures contract and
North Sea spot market on the 15-day market. This can be used to exchange
an IPE position for a spot cargo. It represents, for example, the exchange
of a futures market position of 500 IPE lots, (i.e., 500,000 barrels), for a
15-day cargo. It can be viewed as a spread between forward Brent and

futures Brent.
IPE Brent options.

The IPE’s Brent American options contract was launched on 11 May 1989
trading lots of 1000 barrels of Brent blend. The IPE Brent futures contract
is the underlying asset. The option is exercised, i.e., transferred to a Brent
futures contract at any time before maturity. A call gives its holder the right
to buy the underlying futures contract at a strike price defined in multiples
of 50 cents per barrel. A put gives its holder the right to sell the underlying
futures contract at a strike price defined also in multiples of 50 cents
per barrel. Options are also traded on Brent delivery month spreads. The
over-the-counter (OTC) market represents a series of personalized bilateral
trades and provides tailor-made options on deals of any size. This market
is used by several large financial institutions.
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. Description of Markets and Trading Instruments: The

Case of Cocoa
The International Cocoa Organization (ICCO) was established in 1973
to administer the first International Cocoa Agreement, that of 1972 and
its successor Agreements, of 1975, 1980, 1986, and 1993. For further
information, the reader can refer to library@icco.org. The Agreements
were concluded among the governments of cocoa producing and consuming

countries, under the auspices of the United Nations. As on 13 July 1998,
the membership of the ICCO comprised a total of 41 members.
1.4.1. How do the futures and physicals market work?

There are large differences between physical and futures prices. The cocoa
trade is based on the actuals market and the futures market. The actuals
market is known as the physical market, the spot market or the cash market.
Futures contracts are traded in lots of 10 tonnes. They represent a
commitment to deliver or receive the quantity of cocoa implied by the
contract at the expiry of the contract term. Any cocoa that has passed tests
of quality and bean size through the terminal markets’ grading process can
be tendered against contracts. The buyer who takes delivery of cocoa from a
terminal market would usually obtain material close to the minimum quality
necessary to pass the market’s grading test. In physical contracts, the prices
tend to be higher because of a control of the specification of the material.
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.
1.4.3. How is the ICCO price for cocoa beans calculated?

Is the ICCO price for cocoa beans related to the grade of cocoa?
The ICCO prices for cocoa beans are not related to a specific grade of
cocoa but to the prices on the London and New York Terminal markets.
At LIFFE, the London Terminal market, and at the CSCE, the
New York Terminal market, different grades of cocoa are deliverable against
contracts. However, each lot of cocoa is sampled and graded by Exchange

graders.
1.4.4. Information on how prices are affected by changing

economic factors?
Cocoa price movements can be separated into three categories: long-term,
intermediate-term, and short-term. Long-term cocoa price fluctuations are
induced by the links between the rate of new planting, production, stocks,
and prices. Intermediate-term cocoa price fluctuations, consumption and
stocks represent the response of the cocoa industry to annual variations
in world cocoa production. Short-term cocoa price fluctuations reflect
alternating tides of bullish and bearish speculative enthusiasm in the world’s
cocoa markets.
1.4.5. Cocoa varieties

The names Criollo, Forastero, and Trinitario refer to the three main types
or groups of populations of Theobroma cacao, the cocoa tree.
The world cocoa market distinguishes between two broad categories of
cocoa beans: “fine or flaviour” cocoa beans, and “bulk” or ordinary cocoa
beans. Fine or flaviour cocoa beans are produced from Criollo or Trinitario
cocoa-tree varieties, while bulk cocoa beans come from Forastero trees.
In 1998, the top producing countries are Cote d’Ivoire, Ghana, Indone-
sia, Brazil, Nigeria, Cameroon, and Malaysia. The top grindings countries
in the world are the Netherlands, United States, Germany, Cote d’Ivoire,
Brazil, United Kingdom, and France.
1.4.6. Commodities — Market participants: The case

of cocoa, coffee, and white sugar
We describe some specific features regarding the cocoa, coffee, and white
sugar contracts.
The case of cocoa

Producer in country of origin.
There are different systems of marketing the crop depending on the
country.
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.
The case of coffee

Grower in country of origin.
There are different systems of marketing the crop depending on the country.
Trade house.
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.
The case of White Sugar.

Producer in country of origin.
There is in general in each country a central sugar marketing organization
that negotiates all domestic and international sales.
Trade house.
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.
1.5. Trading Characteristics of Options: The Case

of Equity Options
This section describes the specific features of options markets. A call gives
the right to its holder to buy the underlying asset at a given price within or
at a specified period of time. A put option gives the right to the buyer to
sell the underlying asset at the striking price within or at a specified period
of time.
Equity warrants are long-term options traded often in securities markets
rather than in option markets. Covered warrants are (OTC) long-term
options issued by securities houses.
1.5.1. Options on equity indices

These options are traded on the major indices around the world. Options
on the spot index are cash-settled, i.e., there is no physical delivery of the
underlying index.
1.5.2. Options on index futures

These options require upon exercise a long (a short) position in the future
contract for a call (a put) in the same contract.
Table 1.1. Countries in the FT-actuaries world with

listed index options.
Country Index
USA S&P 500 and S&P 100 since 1983

Japan Nikkei 225 since 1989
France CAC 40 since 1988
Germany DAX since 1991
Switzerland SMI since 1988
Canada TSE 35 since 1987
the Netherlands EOE since 1978
Australia All ordinaries since 1983
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.
1.5.3. Index options markets around the world

In North America, options available are traded on several indices: S&P
100, S&P 500, Major Market, S&P MidCap, NYSE Composite, Value
Line, Torento 35, etc. Several listed options exist on indices like Value
Line (PHX), National OTC (PHX) indexes, etc. In Japan, the main
option contract is the Osaka Nikkei. Options exist also on the Singapore
International Monetary Exchange (SIMEX), the Tokyo stock price index
(TOPIX), etc. The OTC activity appears on the FT-Actuaries Japan Index.
In Europe, several options are listed on European indices.
In Germany, listed DAX options trade on the Deutsche Terminbourde
Bourse (DTB) on a screen-based system. Trading in options started in 1991.
Investors use also the OTC market for DAX options with maturities greater
than three months.
In France, CAC 40 options are traded on the Paris-Bourse. The OTC
market is used mainly by large US and European investment banks. In the
United Kingdom, FT-SE options are traded on the London International
Financial Futures Exchange (LIFFE). The market is dominated by major

international banks and UK brokers.
The OTC market is mainly used by global investment banks and the
major UK brokers.
In Switzerland, listed SMI options are traded on the Swiss Options and
Financial Futures Exchange (SOFFEX) in an electronic-screen system. In
the Netherlands, the EOE options market is mainly used by locals and some
US banks and Dutch pension funds. In Spain, the main participants in the
IBEX-35 options market are the US investment banks.
1.5.4. Stock Index Markets and the underlying indices

in Europe
The CAC 40 Index
The CAC 40 index is computed using 40 stocks in the French market. It
does not account for the distributions of dividends. The following formula
is used:

1000 N i=1 qi,t Si,t
It =
Kt CA0
where:
t: instant at which the index is computed;
N : the number of stocks used (40);
qi,t : number of shares of stock i, at date t;
Si,t : the price of stock i at date t;
CA0 : market capitalization of the sample used in the reference date
(December 1987) and
Kt : an adjustment coefficient at date t.
The STOXXSM 50 and EURO STOXXSM

The Dow Jones STOXXSM and Dow Jones EURO STOXSM 50 indices
are European indices launched on 26 February 1998 by STOXX limited.
These indices provide a representative picture of European equity market
performance. These indices are composed of 50 industrial, commercial, and
financial blue chips. These indices are available on all major information
networks and are disseminated every 15 seconds.
The methodology in constructing these indices is based upon a matrix
approach that begins with 80% of the investible universe.
Futures on the Dow Jones STOXXSM and Dow Jones EURO

STOXXSM 50 indices allow fund managers to insulate the asset value of
their European equity portfolios regardless of their performance benchmark.
This can be achieved by selling futures contracts in proportion to the
sensitivity of these portfolios to fluctuations in the Dow Jones STOXXSM
and Dow Jones EURO STOXXSM 50 indices. Selling derivatives allow
market participants to stabilize their portfolios. Anticipating a rise (a fall)
in prices, investors can buy (sell) index futures and sell at higher (lower)
prices at a later date.
Theoretical values of index futures can be calculated at any time
using the prices of the equity baskets represented by the Dow Jones
STOXXSM and Dow Jones EURO STOXXSM 50 indices. The theoretical
value corresponds to the value of the equity basket plus the basis, i.e., the
cost of purchasing the index portfolio components, less dividends.
Options on Dow Jones STOXXSM and Dow Jones EURO STOXXSM
50 indices can be used as short hedging instruments, through the purchase
of puts.
Derivatives on Dow Jones STOXXSM and Dow Jones EURO
STOXXSM 50 indices give investors the tools to pursue simple strategies
based on their expectations of market movements. Investors can buy call
(put) options or sell put (call) options. Options can be used in arbitrage
transactions allowing strategies to be pursued based on comparative
fluctuations between equity markets within the Euro zone and the different
countries.
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.
Table 1.2. Contract specifications.
Contract specifications Dow Jones STOXXSM and Dow Jones EURO

STOXXSM futures
Underlying index Dow Jones STOXXSM and Dow Jones EURO
STOXXSM indices
Trading unit Contract valued at EURO 10 times the index
quoted in the future
Price quotation Index without decimal
Minimum price fluctuation (tick) 1 index point equivalent to EURO 10
Contract month 3 spot months, 2 quarterly contract months of
the March, June, September, and December
cycle
Last trading month 3rd Friday of the contract month at 12:00 p.m.
First trading day First trading day following the last trading day
of the previous contract month
Settlement/exercise Cash settlement, expiration settlement
price = arithmetic mean (with 2 decimals) of
each index value calculated and displayed
between 11:50 a.m. and 12:00 p.m.
(41 values)
Margin Margin information can be obtained from the
MATIF/MONEP information department
Trading hours NSC day session: 8:00 a.m.–5:00 p.m.
NSC evening session: 5:05 p.m.–10:00 p.m
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).
1.6. Trading Characteristics of Options: The Case

of Options on Currency Forwards and Futures
These options are traded in the OTC market. The growth of the OTC
market is due to its flexibility. In fact, many banks and financial institutions
offer options with tailor-made characteristics in order to match the clients
needs. Options on the currency futures have been traded since 1982. These
options are standardized contracts.
1.7. Trading Characteristics of Options: The Case of Bonds

and Bond Options Markets
There are several types of bonds and bond options traded in organized
and OTC markets. These financial instruments correspond, for example, to
zero-coupon bonds, bonds with call provisions, putable bonds, convertible
bonds, bonds with warrants attached, exchangeable bonds, etc.
1.7.1. The specific features of classic interest rate

instruments
Zero-coupon bonds.
These bonds are bonds with no periodic coupon payments. The interest due
to the bond holder is given by the difference between the maturity value
and the purchase price. This class of bonds is referred to as zero-coupon
bonds and its price is given by the present value of the maturity value.
In the mathematics of bonds, continuous compounded interest rates
and/or discrete compounded interest rates can be used.
For continuous trading in derivatives, interest rate is often continuously
compounded using the factor e−rT to discount US$ 1 payable in T years at
a rate r.
In this case, the value of a zero-coupon bond is computed by discounting
its maturity value at this factor. When interest is accumulated annually
for T years, discretely compounded interest rate is computed by using
(1/(1 + r )T ). The equivalence between the two fomulas appears when
r = log(1 + r ).
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.
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.
Bonds with specific features.

A bond with a call provision gives the right to the issuer to call the issue
before the specified redemption date. A bond with a put provision gives its
holder the right to put the bond back to the issuer at a fixed price.
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.
A bond with an attached warrant.

It is simply a package comprising the bond and a warrant. Most of these
bonds are Eurobonds issued in international capital markets.
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.
Floating Rate Notes, FRN.

The value of an FRN depends mainly on the coupon date payment. The
coupon is often determined as a mean of the interest rates applied to three-
month treasury bills. A risk premium is added to the mean rate to account
for the risk of the issuer.
Floating Rate Bonds, FRB.

The coupon payments are indexed with reference to a variable interest rate
index as the rate on the three-month treasury bills or the rate on 30-year
treasury bonds. Several floaters show implicit embedded provisions which
have the specific features of call and put options. For example, the provision
of the type Floor and Ceiling specifies a minimum coupon rate of x% and
a maximum coupon rate of y%.
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.
In 1985, some American Treasury bonds are traded in the forms of

separate trading of registered interest and principal of securities (STRIPS).
In the mortgage market, bonds are also traded in the forms of STRIPS
as PO STRIP (payment of the principal) and IO STRIP (payment of
interest rates).
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.
1.7.2. The specific features of mortgage-backed securities

These securities are linked to the financial crisis in 2007–2008 and mainly
to the subprime. The mortgage is a pledge of real estate which is used to
secure the payment of a loan originated for the purchase of a real property.
The lender, known as the mortgagee, has the right to foreclose on the loan
and to seize the property if the borrower (mortgagor) does not satisfy his
contracted obligations.
A mortgage loan is specified by the interest rate of the loan, the
number of years to maturity, and the frequency of payments. The mortgage
instrument represents an instrument which is guaranteed by a real asset,
a land, a building, etc. Mortgages can be divided into different classes
according to the nature of the asset used as a guarantee.
Mortgages represent the underlying collateral of mortgage-backed
securities. When bonds are guaranteed by the shares of a firm, they are
referred to as collateral trust bonds. When the bond is guaranteed by a
building for example or other assets of the issuer, it is a mortgage security.
If a firm has 100 securities and uses a guarantee of 30, it can issue 30 of
mortgage bonds.
In the United States, the Federal National Mortgage Association
(FNMA), the Government National Mortgage Association, GNMA, and the
Federal Home Loan Mortgage Corporation, (FHLMC) play an important
role in the mortgage market. The FNMA introduces mortgage-backed-
securities (MBS), which are created by pooling mortgage loans and using
this pool of mortgage loans as collateral for the security. The cash flow of
an MBS is a function of the cash flows of the underlying mortgage pool.
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).
1.7.3. The specific features of interest rate futures, options,

bond options, and swaps
Interest rate futures contracts
A futures contract is an agreement between a buyer or a seller and an
established exchange to take or make delivery of a given commodity at a
specified price at a given delivery or settlement date. An investor can be
long (a buyer) or short (a seller). Each investor must deposit an initial
margin before trading futures contracts. The margin used to guarantee
the transactions can attain a minimum level known as the maintenance
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.
Treasury bill futures

The underlying asset of this contract is a short-term debt obligation. The
treasury bill is quoted in the cash market in terms of the annualized yield
on a bank discount basis:

D 360
Yd =
T t
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.
Treasury bond futures

Treasury bond futures contracts are traded on several exchanges. The
underlying asset of the futures contract traded on the Chicago Board of
Trade is 100,000 par value of a hypothetical 20-year, 8% coupon-bond. The
futures price is quoted in terms of par being 100. The seller of the futures
contract can unwind his position before the maturity date by buying back
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:
Invoice price = Contract size (settlement price of the futures contract

× times conversion factor) + accrued interest.
The term accrued interest can be defined 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
after the closing of the exchange (3:15 p.m.) on the date when the futures
settlement price is scheduled.
Treasury bond futures

The CBT created in 1975 the first financial futures contract a futures for
mortgage-backed securities. These securities are issued by the Government
National Mortgage Association (GNMA). The underlying asset of a trea-
sury bond futures contracts on the CBT is a 15-year T-bond with a coupon
rate of 8%. This rate has changed since then.
The invoice price received by the party with a short position in the
contract is given by the bond futures settlement price which multiplies the
delivery factor for the bond to be delivered plus the accrued interest.
For each deliverable bond, there is a delivery factor which is calculated
with respect to the coupon rate and the time to maturity of that bond.
For example, the conversion factor for a bond with coupon rate rc and
a maturity in m years is:
2m
rc /2 1
CF = +
(1 + 0.04)j (1 + 0.04)2m
j=1
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:
Quoted bond price − settlement futures price (C. factor).
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.
Forward rate agreements

A forward rate agreement (FRA) allows a company to reduce interest rate
exposure by locking into a rate of interest. In this contract, the parties agree
to exchange, at some future date, interest payments on the notional amount
of the contract. The buyer of an FRA contract agrees to pay interest at a
specified rate and to receive interest at a floating rate that prevails at a

future date T .
Interest rate swaps

An interest rate swap is an agreement between two counter-parties to
exchange periodic interest payments. These interest payments are deter-
mined with reference to a pre-determined principal amount known as the
notional principal amount. In general, one party, the fixed rate payer, agrees
to pay the other party fixed-interest payments with a given frequency at
some specified dates. The other party, the floating rate payer, agrees to pay
some interest rate payments that vary according to a reference rate. The
London Inter-bank Offered Rate, LIBOR, is often used as the reference rate.
Risks in bond investments

The buyer of a bond faces different risks: an interest rate risk, a re-
investment risk, a default or credit risk, an inflation risk, an exchange risk,
a liquidity risk, etc.
The interest rate risk

The variations in interest rates modify the bond price. A higher interest
rate leads a lower bond value and a lower interest rate produces a higher
bond value.
The re-investment risk

The return for a bond buyer comes from the perceived interest (the
coupons), the capital gain (variation in the bond price), and the interest
from the placement of the coupons.
The credit and default risk

This risk accounts for the possibility of the borrower to honor his liabilities:
payments of coupons and principal at their exact timing. The credit crunch
in 2008 is largely due to this risk.
The risk to call

The issuer can insert a provision in the debt contract that allows him to
buy back his bonds before the maturity date. In this case, the return for
the bondholder can be different from the return anticipated when buying
the bond.
The inflation risk

The variations in inflation rates affect the return from holding the bond. All
bonds are expressed in nominal terms. The difference between a nominal
return and the inflation rate gives the return in real terms.
The exchange rate risk

The price of a bond denominated in a foreign currency is affected by the
changes in exchange rates. These rates can affect significantly the return
from holding the bond.
The liquidity risk

The liquidity risk reflects the difficulty in selling the bond at a given market
price. This risk can be measured by the spread observed in the market
place. The higher the spread, the greater is the liquidity risk. The quality
of a bond is denoted by a given letter or rating. Rating agencies like Moody
and Standard & Poor give their rating to show the risks associated with
investments in bonds. The passage from a letter A to B or C and D reflects
a higher risk. The risk premium is higher for bonds of type B than type A.
This situation characterises most Islamic bonds or sukuks for which
there is often no secondary markets.
1.8. Simple and Complex Financial Instruments

Forward-start options
These options give 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 at the money call with a different time to
maturity with no additional cost?
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.
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.
Options on the minimum or the maximum

of two or more risky assets
These options may to be useful in the pricing of a wide variety of contingent
claims, traded assets, and financial instruments whose values depend on
extreme values. Examples include discount option bonds, compensation
plans, risk-sharing contracts, collateralized loans, and growth opportunities
among other contracts.
Two-color rainbow options

They refer not only to options on the maximum (minimum) of two assets,
but also to all options whose pay-off depend on two or more underlying
assets: options delivering the best of two assets and cash, spread options,
portfolio options, dual-strike options, etc.
Options with extendible maturities

They include any financial contract with provisions concerning a re-
scheduling of payments and a re-negotiation of terms.
There are many types of exotic and second generation options which
take different forms. They include path-dependent options, lookbacks, par-
tial lookbacks, Asian options, shout options, binaries or digitals, knockouts
or barriers, ladder, and cliquet options among other things.
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
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.
Complex digitals or binaries

In their complex forms, complex digitals or binaries may be presented
in different forms: compound digitals, boolean digitals, and corridors.
Compound digitals obey the same principle as compound options and take
different forms: quanto digitals, barriered digitals, and options on digitals.
1.9. The Reasons of Financial Innovations

Financial engineers are working on the design and the valuation of
these financial instruments and the new strategies for portfolio and risk
management.
The main questions are:
Why there are so many new financial instruments?

Why has the wave of financial innovation not stopped?
The most common explanation often advanced by market authorities

is that financial markets are incomplete and that these instruments allow
us to complete these markets. However, as noted by Ross (1989), this
explanation seems awkward. In fact, the success in valuing these financial
instruments comes from the fact that they are regarded as contingent claims
or derivative securities which are spanned by the underlying assets on which
they are traded and a riskless bond. This allows the derivation of simple
valuation formulas in complete markets.
In reality, markets can never be fully complete but with regard to
the price determination, it is often assumed that markets are complete.
Otherwise, the pricing of these instruments would be a difficult task.
According to Ross (1989), there are two dominant features which
contribute to the wave of financial innovation: the role of institutions and
the role of marketing. These reasons complement the arguments by Miller
(1986) and Merton (1988). Miller’s (1986) analysis is based on the role of
taxes and regulations in the innovation process.
In his analysis, Miller considers taxes as a source of much of the
motivation for financial innovation. Merton (1988) proposes a detailed
analysis of the production function underlying the innovation in derivative
securities markets. He puts the accent on the role of transaction costs.
The analysis by Ross (1989) ignores production costs and is interested
mainly in the role of agency costs and marketing costs, which help to shape
the form of the new institutional features. Agency costs and restrictions may
arise from monitoring and the regulatory environment. They may result
either from the natural needs of market relations between institutions and
participants or may be imposed by the government.
The first reason in Ross’s analysis is that financial markets become
institutional markets since institutions are the most significant participants
in these markets. This does not mean that financial markets are solely
markets where institutions operate, but rather markets where institutions
are significant forces.
Financial institutions range from transparent through translucent to

opaque. In this classification, a mutual fund is regarded as a transparent
institution and an insurance company is seen as an opaque institution.
A pension fund is regarded as a translucent institution. Institutions can
be regarded as financial market players whose activities are dominated by
agency relations.
The second reason in Ross’s analysis of financial innovation is the
role of marketing. In perfect and frictionless markets, selling a financial
instrument is costless. In reality, the less familiar and the more esoteric
the financial instrument, the more costly it is to sell. When the states of
nature are exogenously specified, each security can be defined by its pay-
offs corresponding to the different states. Since uncertainty remains about
these pay-offs, new states are generated and are not yet spanned. This
uncertainty may be “nearly” spanned. In complete markets, marketing
can “explain” the pay-offs in a view where the marginal cost equals the
marginal benefit from a transaction. This view of complete markets allows
the pricing of financial instruments with a great accuracy. It recognizes the
existence of a marketing cost for a new financial instrument or strategy.
This instrument or strategy, corresponds in the begining to the needs of
some institutions or retail clients. In its mature phase, marketing costs
are reduced since the financial instrument or strategy becomes a standard
commodity. Buying or selling securities which are standardized and trade
in well functioning markets with large volume induce nearly no marketing
costs. This is not the case for the tailored and low-grade securities.
Ross makes a distinction between marketed and non-marketed securities
rather than between high- and low-grade securities. Stocks, for example,
are low-grade securities which trade in well organized and competitive
makets. Financial futures are examples of low-grade innovations which
have evolved into low-cost well-traded commodities. This evolution is
costly and the ultimate success relies on the ability to standardize the
financial product and to sustain a sufficient volume of trade to justify the
initial costs.
In Ross’s model, the existence of new financial instruments and
strategies and the marketing process are based on the cost structure of
the marketing networks and distribution channels. It is the institutional
structure of contracts and incentives that allows the process of financial
engeneering to continue. Hence, it seems that institutional markets and
financial marketing are central to the understanding of financial innova-
tions. For more details, see Ross (1989).
1.10. Derivatives Markets in the World: Stock Options,

Index Options, Interest Rate and Commodity
Options and Futures Markets
1.10.1. 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.
1.10.2. The main indexes around the world: a historical

perspective
The first options traded on indexes can be traced back to US (SP 500 and SP
100 in 1983), Japan (Nikkei 225 in 1989), UK (FT-SE 100 in 1984), France
(CAC 40, 1989), Germany (DAX, 1991), Switzerland, (SMI, 1980), Canada
(TSE 35, 1987), Netherlands (EOE, 1978), Australia (All Ordinaries, 1983),
...
Options volume in listed markets is mostly concentrated in one month
contracts in all markets. For most options, volume with longer maturities
take place in OTC markets.
In the OTC market, trading began early in 1988. Several investors buy
long-term puts to implement portfolio insurance strategies.
Today, dealers run large OTC options books. This can reduce or
eliminate risk in the market.
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.
SP 100 are the mostly traded contracts in the U.S.

SP 500 are the have the greatest open interest in the U.S.
Hundred billions of dollars are traded.
Institutional use of index options:
Covered call writing: a call is sold and the underlying asset is held. Long
index put strategy and collar positions, which is preferred by institutions.
The collar can lead to a skew in index options implied volatilities: out of
the money puts have higher volatilities than calls. Options are available on
National OTC (PHX) indexes.
Stock index markets in North America.

The SP 500 index fluctuated in a band. The move gives a volatility in
a range of 10%–25%. We can represent a monthly volatility for the year.
With its heavier dose of cyclical stocks, the DJIA has been outperforming
for some years the broader market.
We can compute historical volatility and implied volatility from at the
money options. We should compute the spread.
The following Tables shows the volume (number of contracts traded)
in several countries.
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.
Stock index options

2004 2003
Exchange Volume Traded
(Number of Contracts) 2003
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%
Europe, Africa, Middle East

Athens Derivatives Exchange 941,387 1,388,985 67.78%
BME Spanish Exchanges 2,947,529 2,981,593 98.86%
Borsa Italiana 2,220,807 2,505,351 88.64%
Copenhagen SE 1,299 8,440 15.39%
Eurex 117,779,232 108,504,301 108.55%
Euronext 99,607,852 103,986,651 95.79%
JSE South Africa 11,268,763 10,505,417 107.27%
OMX Stockholm SE 8,947,439 6,371,381 140.43%
Oslo Bors 681,783 543,090 125.54%
Tel Aviv SE 36,915,103 29,353,595 125.76%
Warsaw SE 124,392 153,106 81.25%
Wiener Börse 40,855 27,680 147.60%
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%
Total 3,137,482,893 3,357,354,658 93.45%

2005 2004
(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
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

Athens Derivatives Exchange 700,094 941,387
BME Spanish Exchanges 4,407,465 2,947,529
Borsa Italiana 2,597,830 2,220,807
Eurex 149,380,569 117,779,232
Euronext.liffe 70,228,310 99,607,852
JSE 11,473,116 11,303,311
OMX 12,229,145 8,947,439
Oslo Børs 515,538 695,672
Tel Aviv SE 63,133,416 36,915,103
Warsaw SE 250,060 78,752
Wiener Börse 37,127 40,855
Total 3,203,028,688 3,028,651,872

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.
Stock index markets in France.

– An interesting development in the CAC 40 futures is the distribution of
open interest across various months.
– Institutions have led to move into the quarterly contracts to eliminate
the chore of rolling on a monthly basis.
The lack of a developed stock-borrowing market can reduce trading in

futures.
Professional traders can use the futures to hedge OTC options. To
hedge collars traders can be short futures.
Arbitrageurs (short stock/long futures) can unwind easily their positions.
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
of short-term speculators and international institutions looking for long

exposures.
The OTC market is important because there is a need for longer-term
strategies on the SMI from pension funds.
Zero premium collars are very popular.
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).
DERIVATIVES - 3.1 STOCK OPTIONS
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

Athens Derivatives Exchange 17,194 21,729 52 60 1,297 2,004 396 397 3 2
BME Spanish Exchanges 12,425,979 10,915,227 27,775 20,605 2,748,562 2,411,628 75,313 65,136 1,067 633
Borsa Italiana 16,056,751 12,439,716 91,803 67,776 1,964,411 1,646,014 475,942 442,151 2,771 1,979
Budapest SE 350 176 5 6 NA NA 6 8 NA NA
Eurex 272,543,052 255,918,793 964,097 752,434 52,069,011 53,312,606 NA NA 59,286 38,740
Euronext.liffe 155,552,010 264,714,188 603,265 618,732 45,341,415 55,353,971 3,272,555 2,728,180 32,141 70,685
JSE 5,751,832 2,539,526 312 153 916,339 564,302 2,835 1,733 NA NA
OMX 64,514,641 57,138,563 69,691 57,580 8,418,826 7,404,092 NA NA 27,306 15,434
Oslo Børs 5,811,946 3,325,368 NA NA 616,315 364,265 34,135 NA 643 321
RTS SE 10,727,870 7,281,162 11,453 2,797 1,431,028 433,158 150,940 113,317 NA NA
Warsaw SE 10,988 4,372 98 29 162 413 5,501 2,642 4 1
Wiener Börse 1,053,298 816,032 5,385 4,609 116,063 76,166 NA NA 230 165
Total 2,653,601,416 2,311,714,514 - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.2 STOCK FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

Exchange Volume Traded Notional Value Open Interest Number of Trades
(Nber of Contracts) (USD millions) (Nber of Contracts)
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

Athens Derivatives Exchange 2,476,487 1,431,514 5,543 3,160 116,576 124,815 285,982 167,715
BME Spanish Exchanges 21,229,811 18,813,689 43,266 31,708 1,649,184 1,921,717 139,441 119,499
Borsa Italiana 7,031,974 5,957,674 49,636 41,798 41,319 58,071 56,774 66,605
Budapest SE 919,426 740,396 9,052 7,842 65,015 24,936 92,618 81,468
Eurex 35,589,089 77,802 203,038 NA 1,459,509 58,107 NA NA
Euronext.liffe 29,515,726 12,158,093 344,198 64,062 1,489,169 467,117 22,948 21,006
JSE 69,671,751 24,469,988 26,288 10,223 12,027,716 1,535,839 392,154 177,766
OMX 8,459,165 5,659,823 6,128 NA 1,764,492 1,387,095 NA NA
Oslo Børs 3,626,036 1,796,570 3,502 2,516 268,572 126,266 NA NA
Warsaw SE 112,824 172,828 782 845 1,122 2,928 87,999 130,674
Wiener Börse 12,371 23,748 180 331 1,339 2,448 NA NA
Total 279,730,018 140,736,581 - - - - - -
DERIVATIVES - 3.3 STOCK INDEX OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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
2006 Option Trading Volume Growth: Asia

160%
Singapore Exchange
Rate of Growth (annual)
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
DERIVATIVES - 3.4 STOCK INDEX FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

Exchange Volume Traded Notional Value Open Interest Num ber of Trades
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
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

BME Spanish Exchanges 8,007,257 6,081,276 1,012,015 615,976 86,067 75,608 2,889,255 1,993,832
Borsa Italiana 5,697,622 4,875,301 1,041,826 777,839 15,470 26,348 3,763,954 2,966,677
Budapest SE 1,879,064 529,563 7,313 5,222 66,747 4,307 303,992 182,057
Eurex 270,134,951 184,495,160 18,565,389 10,851,303 2,790,632 2,166,815 NA NA
Euronext.liffe 72,135,006 56,092,515 6,318,763 4,154,454 1,166,209 1,027,559 18,101,967 13,122,326
JSE 15,506,101 10,663,676 398,761 224,904 296,485 289,601 301,306 445,755
OMX 24,374,765 20,259,026 329,352 NA 551,421 504,687 NA NA
Oslo Børs 2,437,118 562,591 15,616 8,245 56,943 13,665 22,816 NA
Tel Aviv SE 32,474 13,460 589 219 2,682 2,315 219 71
Warsaw SE 6,257,203 5,167,111 59,920 34,864 72,706 30,348 2,121,215 1,437,611
Wiener Börse 154,521 104,677 13,533 6,981 17,046 13,260 NA NA
Total 1,166,606,884 881,260,593 - - - - - -
DERIVATIVES - 3.5 SHORT TERM INTEREST RATE OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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

Euronext.liffe 92,985,715 79,482,008 104,878,071 89,052,387 10,367,389 9,586,715 65,325 76,311 NA NA
OMX 95,000 - NA - 67,000 - NA - NA -
Total 386,818,080 277,741,236 - - - - - - - -
NA : Not Available
- : Not Applicable
DERIVAT IVES - 3.6 SHORT TERM INTEREST RATE FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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
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
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

Budapest SE 2,500 1,390 12 3 0 500 5 16
Eurex 767,458 688,831 937,064 833,748 48,307 37,838 NA NA
Euronext.liffe 296,008,444 248,662,893 341,274,218 280,316,062 6,092,072 5,242,458 32,413,840 25,668,450
JSE 667 0 NA NA 63 0 NA NA
OMX 8,170,853 6,315,805 NA NA 526,914 345,833 NA NA
Total 1,349,704,857 970,390,236 - - - - - -

DERIVATIVES - 3.7 LONG TERM INTEREST RATE OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
(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

Eurex 76,328,806 58,551,836 10,870,919 8,449,133 1,786,810 1,405,446 NA NA NA NA
JSE 2,785 4,831 NA 11 NA NA NA 79 NA NA
Total 177,246,085 152,599,930 - - - - - - - -
DERIVATIVES - 3.8 LONG TERM INTEREST RATE FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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
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

BME Spanish Exchanges 15 46 2 6 1 2 8 22
Budapest SE 2,500 - 12 - 0 - 5 -
Eurex 654,119,660 599,621,461 92,905,934 85,843,727 3,796,014 3,357,373 NA NA
Euronext.liffe 23,245,504 19,078,373 4,356,744 3,468,410 360,521 292,141 2,099,645 2,002,722
JSE 8,947 10,362 NA NA 63 0 NA NA
OMX 4,354,311 3,097,742 NA NA 184,780 140,258 NA NA
Tel Aviv SE 25,005 - 562 - 651 - 1,985 -
Warsaw SE 12,875 32,362 431 1,028 50 58 164 484
Total 1,271,406,293 1,131,707,505 - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.9 CURRENCY OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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

Budapest SE 1,022,457 258,000 1,323 251 25,500 86,700 1,050 209 NA NA
Euronext.liffe 733,039 403,957 9,056 4,728 52,150 42,240 17,712 23,871 126 133
Tel Aviv SE 7,447,717 6,937,575 74,820 69,802 224,904 217,476 335,782 270,799 1,456 1,597
Total 23,225,941 17,835,080 - - - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.10 CURRENCY FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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

Athens Derivatives Exchange 84 21,844 7 1,692 0 80 3 3,861
Budapest SE 10,857,327 7,742,408 14,535 10,698 301,032 406,942 30,281 19,760
Euronext.liffe 8,807 7,435 216 176 1,043 518 1,221 1,510
Turkish Derivatives Exchange 4,598,416 1,603,797 NA 1,663 170,431 134,063 NA NA
Warsaw SE 3,144 6,216 34 65 68 84 2,579 5,184
Total 158,359,064 113,926,650 - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.11 COMMODITY OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
(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

Budapest SE 832 40 13.42 0 260 95 29 3 NA NA
Euronext.liffe 727,190 444,754 271 226 136,475 60,129 9,257 7,059 21 11
ICE Futures 138,129 118,476 NA NA 23,987 5,832 NA NA NA NA
JSE 512,518 451,885 1,898,026 337,671 48,568 57,950 52,749 40,655 NA NA
London Metal Exchange 8,412,350 8,184,187 NA 468,446 1,007,248 757,837 NA NA 6,716 4,397
Total 102,858,516 75,536,786 - - - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.12 COMMODITY FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

Americas
BM&F 1,318,203 1,073,471 12,436 10,106 63,964 50,996 219,847 214,293
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
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

Budapest SE 8,750 778 140 9 1,093 601 1,856 189
Euronext.liffe 9,124,195 8,054,116 119,436 85,794 449,829 419,333 1,257,639 906,230
ICE Futures 92,582,921 41,936,609 NA NA 1,389,618 642,753 NA NA
JSE 1,436,155 1,335,964 1,864,750 15,158,450 43,295 51,295 205,430 169,767
London Metal Exchange 78,527,839 70,444,665 7,146,569 4,045,775 1,515,663 2,411,870 NA NA
Total 794,164,463 637,958,959 - - - - - -
NA : Not Available
- : Not Applicable
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
are published in the “Quarterly Bulletin of Cocoa Statistics” in the ICCO

publications. For example, the estimate of world cocoa-bean production for
the 1996/1997 cocoa year is 2,695,000 tonnes, down 20,000 tonnes from
the figure in the June 1997 Newsletter. World grindings of cocoa beans
in 1996/1997 were estimated at 2,815,000 tonnes, representing an increase
of 12,000 tonnes compared with the previous forecast. The information
concerns the gross crop, the net crop, grindings, surplus/deficit, total stocks,
and free stocks. The cocoa futures contract was originally launched in
1928. The cocoa traded options contract was launched in 1987 as a means
of offering market participants even greater flexibility and choice in their
underlying activities. These contracts are traded in London (LIFFE). Index
options on stock indices 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. There are several types of bonds
and bond options traded in organized and OTC markets. They include
zero-coupon bonds, bonds with call provisions, putable bonds, convertible
bonds, bonds with warrants attached, exchangeable bonds, etc. 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. Financial assets may appear in non-
standard fashion, i.e., they can be tailor-made and their pay-offs may be
path-dependent or path-independent. A path-dependent contingent claim is
an option whose pay-off depends on the history of the underlying asset price.
In general, an upward movement of the underlying asset price followed by
a downward movement is different from a downward movement followed by
an upward movement. This is a main property of path-dependent options.
For path-independent contingent claims, an upward movement of the
underlying asset price followed by a downward movement is equivalent to a
downward movement followed by an upward movement. Examples of non-
standard financial assets include forward start options, pay-later options,
chooser options, options on the minimum or the maximum of several
assets, two-color rainbow options, options with extendible maturities, ratio
options, exotic options, barrier options, Asian options, partial and full
lookback options and more sturctured products with embedded digitals
such as rebate range binaries, mandarin collars, mega-premium options,
and limit binary options. Several authors proposed different explanations
for the development of these markets and the proliferation of the new
financial instruments. For example, according to Ross (1989), the existence

of new financial instruments and strategies and the marketing process are
based on the cost structure of the marketing networks and distribution
channels. It is the institutional structure of contracts and incentives that
allows the process of financial engineering to continue. Hence, it seems
that institutional markets and financial marketing are central to the
understanding of financial innovations.
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.
2. Explain why institutions and individuals use options.

Institutional and individual investors use options to achieve an astonish-
ingly broad spectrum of goals, such as hedging, arbitrage and speculation.
Options can be used in several strategies:
• Aggressive strategies
• Leveraged strategies
• Protect an existing portfolio
• Combinations where cash spent is recouped by interest. This prevents
from putting capital at risk, etc.
3. Explain exchange traded or listed options and the role

of the clearing house.
Options are exchange traded contracts (or OTC contracts) for making eco-
nomic commitments based on the shifting values of stock prices, indices, etc.
The clearing house interposes itself in all transactions as the buyer to
every seller and the seller to every buyer, so every party is free to liquidate
his position at any time by making an offset closing transaction.
A committee charged with developing new financial instruments can
submit the proposal. Then, we wait for the approval.
4. Provide historical reasons for the development of the

option market.
The idea of stock options was borne in 1972.
The idea of index options and futures was born in 1977.
After the success of the initial ‘covered market’ (exercise against
shares), banks issued options without having an underlying corporate to
provide the hedge.
New issues of options can also be cash settled at maturity.
The market today provides a range of exposures on most of the world’s
significant equity markets.
In many markets, local authorities are actively fostering options devel-
opment.
The market today provides options which are highly liquid. For options
and warrants, the market provides liquid and less liquid warrants and exotic
warrants in emerging markets.
5. What is the appropriate definition of an option?

Options can be calls or puts.
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.
6. How is an option exercised?

An option entitles the right and not the obligation to buy or sell the
underlying. This right has value.
A call entitles the right to the holder to buy the underlying asset.
A put entitles the right to the holder to sell the underlying asset.
Options give the holder the right to buy or sell a specific asset at a
fixed price on or before a given expiry date.
If the right is exercised at any time, this is an American type option.
If the right is exercised at maturity, this is a European type option.
The value in cash (received or stock) corresponds to the exercise
value.
Options with exercise value are said to be in the money.
Options with no exercise value are out of the money.
Options for which the strike price is equal to the underlying asset price
is at the money.
7. What happens for buyers and sellers among exercise?

For the option buyer, exercise is a right, not an obligation.
Sellers have an unconditional obligation to respond whenever the buyer
chooses to exercise. It may seem that the buyer has all the advantages, and
that the seller assumes nothing but liabilities. That is why the buyer has
to pay the seller for the option.
The buyer must pay cash to the seller for the option’s full price.
Since this is the maximum amount the buyer can lose in a transaction,
he is not required to pay any additional security as margin.
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.
8. How can options be used?

• Options can be used as an alternative means of implementing different
strategies investors execute directly in the stock market, but with
enhanced performance and reduced transaction costs.
• Options can be used to structure unique patterns of risks and returns
that would have been impossible without them.
The cost of an option is significantly less than the price of the

underlying asset.
This allows for leverage (or Gearing), of the option.
(The underlying asset of the option is a single stock, a basket of stocks,
an equity index, a currency, etc.)
Options can be exercised and settled physically in return for the
physical shares.
Options can be exercised and settled in cash for an amount equal to
their intrinsic value.
9. How is each option contract specified?

In selecting the contract that best suits the investment applications,
investors can sort through a variety financial instruments: calls, puts,
European and American options.
Each contract is specified by:
• A contract multiplier: the value times which the contract price is

multiplied to determine its total value;
• Minimum fluctuation: the smallest permissible increment of price change;
• Expiration terms dates for expiration;
• Trading hours;
• Position limits: the maximum number of contracts the exchange will

permit an investor to control.
Investors can contact brokers and the exchanges to obtain the current
specifications.
10. Describe cash and margin requirements.

Buyers pay the full dollar value of their contracts.
The buyer is never required to deposit additional funds if the position
moves against him.
The option seller is obliged to pay the difference between the option’s
strike price and the underlying, a good-faith deposit of cash or securities
ensure eventual performance. This is the case for cash setteled options.
For traditional stock market trading, the term ‘margin’ suggests a down
payment on the full value of securities purchased, with the brokerage firm
loaning the investor the balance.
For options, the exact amount of margining to be deposited must be
determined.
Margins can also be different between speculators and hedgers.
The margin, for example, for stock (index) options may be a % of the
underlying stock (index) plus the option price.
Margin requirements can be recalculated each day on a mark-to-market
basis.
If subsequent calculations show higher requirements, the seller must
deposit additional margins.
11. Should investors pay transaction costs?

Any time investors trade securities they pay two types of transaction costs.
First, they pay an explicit commission to a broker for executing and
clearing the trade.
Second, they pay an implicit market impact cost because their bids
will inevitably drive prices higher when they wish to buy and their offers
will drive prices marginally lower when they wish to sell.
Commissions are negotiable between the investor and the broker.
Like stocks, options commissions are charged on a one way basis.
12. What about tax treatment?

Profits and losses from trading can be treated as long term capitalgains or
losses and short term capital gains or losses.
13. What about trading orders?

In many ways, trading options is just like trading stocks. Trading orders
most commonly used reflect:
• Time duration;
• Good-Til Canceled (or open); and
• Opening only.
14. What are price specifications?

Can include:
• Limit order (maximum purchase price);

• Discretion or limit order;
• Delta;
• Market order;
• Market on close order; and
• Market if touched.
15. What are contingencies?

• Contingent: a contingent order is in effect when a specified condition is
satisfied;
• Stop order; and
• Stop limit.
16. What are special instructions?

• Immediate or cancel;
• All or none; or
• Fill or Kill.
17. Explain cancellations.

• Straight cancel;
• If nothing done cancel; or

• Cancel former order.
18. How are contracts exercised and assignments conveyed?

Orders to exercise must be tendered in writing to the exchanges by a
member firm no later than the close of trading on the day of the exercise.
Brokerage firms may apply earlier cutoff times for receipt of oral
exercise instructions from their customers.
At expiration, long customer positions in the money are automatically
exercised.
Once a contract has been exercised, the clearing corporation assigns it
by random lottery among the universe of member brokerage firms carrying
matching short positions.
Assignment notices are generally conveyed to customers before the
opening of the trading on the market day following the exercise date.
19. Explain the world of floor traders.

Floor traders are of two basic breeds: market makers and floor/brokers.
Floor brokers act as agents executing orders in the crowd on behalf of
others.
They earn their livelihoods by collecting commissions on the trades
they execute. Their income is determined by the volume of transactions
they complete.
Market makers put their own capital at risk in the trading. Their
only source of income is the profit they can derive from their trading and
their only limit is the risk they are willing to bear.
Before the exchanges admit a new trader as a member, they investigate
his background and administer a test.
20. What are some floor strategies?

The scalpers: They exploit the fact that the price of any traded asset is
quoted as a two sided market comprised of the highest bid and the lowest
offer.
They will simultaneously make bids and offers, indifferent to whether

they end up buying or selling. Their concern is that whenever they buy,
they buy on the bid side of the market, and whenever they sell, they sell
on the offer side of the market, thus their profit earned is the differential
between the two.
The shooter: He is another type of market maker who tries to make
purchases at the bid and sales at the offer. Unlike the scalper, he is willing
to inventory positions in anticipation of market moves. The shooter is in
the game for the big score.
The spreader: He seeks out and exploits minute inefficiencies in the
pricing structure of the options markets.
21. Describe a day of trading and how the exchange works.

The clearing houses accept as only firms that demonstrate substantial
financial strength and business integrity members.
They maintain elaborate safeguards against defaults, including special
funds to be used in the event of losses, to which member firms must
contribute.
The process of determining the opening price is an unstructured
negotiation that begins several minutes before the official opening.
Market makers provide the prices.
Exchange employees called pit observers report the transactions to
terminal operators who disseminate the transactions to quotation services
around the world.
Discipline in the pit is provided by pit observers who monitor trading
activity for accuracy and fairness throughout the day.
Public orders are handled by floor brokers.
Market continuity is provided by the presence of competing market
makers.
In other markets, a staff of exchange employees maintains a public
book of limit orders.
Customer market orders can be put in computer that randomly
assigns them to participating market makers and reports the trades
instantaneously. New technologies are used and are integrated into the stock
exchange trading process.
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.
The main indexes around the world: a historical perspective

The first options traded on indexes can be traced back to US (SP 500 and
SP 100 in 1983), Japan (Nikkei 225 in 1989), UK (FT-SE 100 in 1984), France
(CAC 40, 1989), Germany (DAX, 1991), Switzerland, (SMI, 1980), Canada
(TSE 35,1987),Netherlands(EOE,1978),Australia(AllOrdinaries,1983),.. .
Options volume in listed markets is mostly concentrated in one month
contracts in all markets. For most options, volume with longer maturities
take place in OTC markets.
In the OTC market, trading began early in 1988. Several investors buy
long-term puts to implement portfolio insurance strategies.
Today, dealers run large OTC options books. This can reduce or
eliminate risk in the market.
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.
SP 100 are the mostly traded contracts in the U.S.

SP 500 are the have the greatest open interest in the U.S.
Hundred billions of dollars are traded.
Institutional use of index options:
Covered call writing: a call is sold and the underlying asset is held, Long
index put strategy and collar positions, which is preferred by institutions.
The collar can lead to a skew in index options implied volatilities: out of
the money puts have higher volatilities than calls. Options are available on
National OTC (PHX) indexes.
Stock index markets in North America:

The SP 500 index fluctuated in a band. The move gives a volatility in
a range of 10%–25%. We can represent a monthly volatility for the year.
With its heavier dose of cyclical stocks, the DJIA has been outperforming
for some years the broader market.
We can compute historical volatility and implied volatility from at the
money options. We should compute the spread.
The following Tables shows the volume (number of contracts traded)
in several countries.
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.
Stock index options

2004 2003
(Nber of Contracts) 2003
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%
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%

Athens Derivatives Exchange 941,387 1,388,985 67.78%
BME Spanish Exchanges 2,947,529 2,981,593 98.86%
Borsa Italiana 2,220,807 2,505,351 88.64%
Copenhagen SE 1,299 8,440 15.39%
Eurex 117,779,232 108,504,301 108.55%
Euronext 99,607,852 103,986,651 95.79%
JSE South Africa 11,268,763 10,505,417 107.27%
OMX Stockholm SE 8,947,439 6,371,381 140.43%
Oslo Bors 681,783 543,090 125.54%
Tel Aviv SE 36,915,103 29,353,595 125.76%
Warsaw SE 124,392 153,106 81.25%
Wiener Börse 40,855 27,680 147.60%
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%
Total 3,137,482,893 3,357,354,658 93.45%

2005 2004
(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
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

Athens Derivatives Exchange 700,094 941,387
BME Spanish Exchanges 4,407,465 2,947,529
Borsa Italiana 2,597,830 2,220,807
Eurex 149,380,569 117,779,232
Euronext.liffe 70,228,310 99,607,852
JSE 11,473,116 11,303,311
OMX 12,229,145 8,947,439
Oslo Børs 515,538 695,672
Tel Aviv SE 63,133,416 36,915,103
Warsaw SE 250,060 78,752
Wiener Börse 37,127 40,855
Total 3,203,028,688 3,028,651,872

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.
Stock index markets in France:

– An interesting development in the CAC 40 futures is the distribution of
open interest across various months.
– Institutions have led to move into the quarterly contracts to eliminate
the chore of rolling on a monthly basis.
The lack of a developed stock-borrowing market can reduce trading in

futures.
Professional traders can use the futures to hedge OTC options. To
hedge collars traders can be short futures.
Arbitrageurs (short stock/long futures) can unwind easily their positions.
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
of short-term speculators and international institutions looking for long

exposures.
The OTC market is important because there is a need for longer-term
strategies on the SMI from pension funds.
Zero premium collars are very popular.
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).
DERIVATIVES - 3.1 STOCK OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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

Athens Derivatives Exchange 17,194 21,729 52 60 1,297 2,004 396 397 3 2
BME Spanish Exchanges 12,425,979 10,915,227 27,775 20,605 2,748,562 2,411,628 75,313 65,136 1,067 633
Borsa Italiana 16,056,751 12,439,716 91,803 67,776 1,964,411 1,646,014 475,942 442,151 2,771 1,979
Budapest SE 350 176 5 6 NA NA 6 8 NA NA
Eurex 272,543,052 255,918,793 964,097 752,434 52,069,011 53,312,606 NA NA 59,286 38,740
Euronext.liffe 155,552,010 264,714,188 603,265 618,732 45,341,415 55,353,971 3,272,555 2,728,180 32,141 70,685
JSE 5,751,832 2,539,526 312 153 916,339 564,302 2,835 1,733 NA NA
OMX 64,514,641 57,138,563 69,691 57,580 8,418,826 7,404,092 NA NA 27,306 15,434
Oslo Børs 5,811,946 3,325,368 NA NA 616,315 364,265 34,135 NA 643 321
RTS SE 10,727,870 7,281,162 11,453 2,797 1,431,028 433,158 150,940 113,317 NA NA
Warsaw SE 10,988 4,372 98 29 162 413 5,501 2,642 4 1
Wiener Börse 1,053,298 816,032 5,385 4,609 116,063 76,166 NA NA 230 165
Total 2,653,601,416 2,311,714,514 - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.2 STOCK FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

Exchange Volume Traded Notional Value Open Interest Number of Trades
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

BME Spanish Exchanges 21,229,811 18,813,689 43,266 31,708 1,649,184 1,921,717 139,441 119,499
Borsa Italiana 7,031,974 5,957,674 49,636 41,798 41,319 58,071 56,774 66,605
Budapest SE 919,426 740,396 9,052 7,842 65,015 24,936 92,618 81,468
Eurex 35,589,089 77,802 203,038 NA 1,459,509 58,107 NA NA
Euronext.liffe 29,515,726 12,158,093 344,198 64,062 1,489,169 467,117 22,948 21,006
JSE 69,671,751 24,469,988 26,288 10,223 12,027,716 1,535,839 392,154 177,766
OMX 8,459,165 5,659,823 6,128 NA 1,764,492 1,387,095 NA NA
Oslo Børs 3,626,036 1,796,570 3,502 2,516 268,572 126,266 NA NA
Warsaw SE 112,824 172,828 782 845 1,122 2,928 87,999 130,674
Wiener Börse 12,371 23,748 180 331 1,339 2,448 NA NA
Total 279,730,018 140,736,581 - - - - - -
DERIVATIVES - 3.3 STOCK INDEX OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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
2006 Option Trading Volume Growth: Asia

160%
Singapore Exchange
Rate of Growth (annual)
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
DERIVATIVES - 3.4 STOCK INDEX FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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
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
Hong Kong Exchanges 19,747,246 13,393,462 2,014,834 987,256 185,262 136,465 9,443,472 6,338,836
Osaka SE 31,661,331 18,070,352 3,560,096 2,068,205 388,666 409,588 3,025,602 949,211
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

BME Spanish Exchanges 8,007,257 6,081,276 1,012,015 615,976 86,067 75,608 2,889,255 1,993,832
Borsa Italiana 5,697,622 4,875,301 1,041,826 777,839 15,470 26,348 3,763,954 2,966,677
Budapest SE 1,879,064 529,563 7,313 5,222 66,747 4,307 303,992 182,057
Eurex 270,134,951 184,495,160 18,565,389 10,851,303 2,790,632 2,166,815 NA NA
Euronext.liffe 72,135,006 56,092,515 6,318,763 4,154,454 1,166,209 1,027,559 18,101,967 13,122,326
JSE 15,506,101 10,663,676 398,761 224,904 296,485 289,601 301,306 445,755
OMX 24,374,765 20,259,026 329,352 NA 551,421 504,687 NA NA
Oslo Børs 2,437,118 562,591 15,616 8,245 56,943 13,665 22,816 NA
Tel Aviv SE 32,474 13,460 589 219 2,682 2,315 219 71
Warsaw SE 6,257,203 5,167,111 59,920 34,864 72,706 30,348 2,121,215 1,437,611
Wiener Börse 154,521 104,677 13,533 6,981 17,046 13,260 NA NA
Total 1,166,606,884 881,260,593 - - - - - -
DERIVATIVES - 3.5 SHORT TERM INTEREST RATE OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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

Euronext.liffe 92,985,715 79,482,008 104,878,071 89,052,387 10,367,389 9,586,715 65,325 76,311 NA NA
OMX 95,000 - NA - 67,000 - NA - NA -
Total 386,818,080 277,741,236 - - - - - - - -
NA : Not Available
- : Not Applicable
DERIVAT IVES - 3.6 SHORT TERM INTEREST RATE FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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
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
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

Budapest SE 2,500 1,390 12 3 0 500 5 16
Eurex 767,458 688,831 937,064 833,748 48,307 37,838 NA NA
Euronext.liffe 296,008,444 248,662,893 341,274,218 280,316,062 6,092,072 5,242,458 32,413,840 25,668,450
JSE 667 0 NA NA 63 0 NA NA
OMX 8,170,853 6,315,805 NA NA 526,914 345,833 NA NA
Total 1,349,704,857 970,390,236 - - - - - -

DERIVATIVES - 3.7 LONG TERM INTEREST RATE OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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

Eurex 76,328,806 58,551,836 10,870,919 8,449,133 1,786,810 1,405,446 NA NA NA NA
JSE 2,785 4,831 NA 11 NA NA NA 79 NA NA
Total 177,246,085 152,599,930 - - - - - - - -
DERIVATIVES - 3.8 LONG TERM INTEREST RATE FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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
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
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

BME Spanish Exchanges 15 46 2 6 1 2 8 22
Budapest SE 2,500 - 12 - 0 - 5 -
Eurex 654,119,660 599,621,461 92,905,934 85,843,727 3,796,014 3,357,373 NA NA
Euronext.liffe 23,245,504 19,078,373 4,356,744 3,468,410 360,521 292,141 2,099,645 2,002,722
JSE 8,947 10,362 NA NA 63 0 NA NA
OMX 4,354,311 3,097,742 NA NA 184,780 140,258 NA NA
Tel Aviv SE 25,005 - 562 - 651 - 1,985 -
Warsaw SE 12,875 32,362 431 1,028 50 58 164 484
Total 1,271,406,293 1,131,707,505 - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.9 CURRENCY OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
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

Budapest SE 1,022,457 258,000 1,323 251 25,500 86,700 1,050 209 NA NA
Euronext.liffe 733,039 403,957 9,056 4,728 52,150 42,240 17,712 23,871 126 133
Tel Aviv SE 7,447,717 6,937,575 74,820 69,802 224,904 217,476 335,782 270,799 1,456 1,597
Total 23,225,941 17,835,080 - - - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.10 CURRENCY FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

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

Athens Derivatives Exchange 84 21,844 7 1,692 0 80 3 3,861
Budapest SE 10,857,327 7,742,408 14,535 10,698 301,032 406,942 30,281 19,760
Euronext.liffe 8,807 7,435 216 176 1,043 518 1,221 1,510
Turkish Derivatives Exchange 4,598,416 1,603,797 NA 1,663 170,431 134,063 NA NA
Warsaw SE 3,144 6,216 34 65 68 84 2,579 5,184
Total 158,359,064 113,926,650 - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.11 COMMODITY OPTIONS
2006 2005 2006 2005 2006 2005 2006 2005 2006 2005
(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

Budapest SE 832 40 13.42 0 260 95 29 3 NA NA
Euronext.liffe 727,190 444,754 271 226 136,475 60,129 9,257 7,059 21 11
ICE Futures 138,129 118,476 NA NA 23,987 5,832 NA NA NA NA
JSE 512,518 451,885 1,898,026 337,671 48,568 57,950 52,749 40,655 NA NA
London Metal Exchange 8,412,350 8,184,187 NA 468,446 1,007,248 757,837 NA NA 6,716 4,397
Total 102,858,516 75,536,786 - - - - - - - -
NA : Not Available
- : Not Applicable
DERIVATIVES - 3.12 COMMODITY FUTURES
2006 2005 2006 2005 2006 2005 2006 2005

Americas
BM&F 1,318,203 1,073,471 12,436 10,106 63,964 50,996 219,847 214,293
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
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

Budapest SE 8,750 778 140 9 1,093 601 1,856 189
Euronext.liffe 9,124,195 8,054,116 119,436 85,794 449,829 419,333 1,257,639 906,230
ICE Futures 92,582,921 41,936,609 NA NA 1,389,618 642,753 NA NA
JSE 1,436,155 1,335,964 1,864,750 15,158,450 43,295 51,295 205,430 169,767
London Metal Exchange 78,527,839 70,444,665 7,146,569 4,045,775 1,515,663 2,411,870 NA NA
Total 794,164,463 637,958,959 - - - - - -
NA : Not Available
- : Not Applicable
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
RISK MANAGEMENT, DERIVATIVES MARKETS

AND TRADING STRATEGIES
Chapter Outline
1. Section 2.1 gives an overview of futures markets and the trading

mechanisms in these markets.
2. Section 2.2 presents the main pricing relationships for forward and
futures contracts.
3. Section 2.3 develops the main trading motives in futures markets. Sev-
eral examples explain strategies with reference to hedging, speculation,
and arbitrage.
4. Section 2.4 studies the main bounds on option prices.
5. Section 2.5 illustrates some simple trading strategies for options and
their underlying assets.
6. Section 2.6 presents some option combinations involving straddles and
strangles.
7. Section 2.7 illustrates some option spreads in bull and bear strategies
involving calls and puts.
8. Section 2.8 develops butterfly strategies using put and call options.
9. Section 2.9 presents Condor strategies using put and call options.
10. Section 2.10 studies ratio-spread strategies.
11. Section 2.11 illustrates some combinations of options with bonds and
stocks and portfolio insurance strategies.
12. Section 2.12 studies conversion and reversal strategies.
67
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
Risk Management, Derivatives Markets and Trading Strategies 69
it with a futures contract in the opposite direction. Transactions are done

on margin in the futures market. The margin represents a small fraction of
the value of the contract. It varies by the type of commodity.
Option prices depend on factors that affect the main elements in
computing their prices. These factors concern the volatility in the financial
markets, the level of interest rates, the option’s maturity date, the exercise
price, dividends on the underlying asset, etc. When a derivative asset is
about to expire, it is relatively a simple matter to compute its value. In
fact, at the option maturity date, the holder can either exercise it or let it
expire. Hence, at this date, the option price is given only by the position
of the underlying asset price with respect to the option strike price. This
position defines the option payoff at this date. At any time, there is a market
price for the derivative asset. Models are used to compute the option price
at any time. But, this does not mean that the market option price must be
equal to the model price. The difference represents the mispricing. Options
can be either European or American. A European style-derivative asset
cannot be exercised before its maturity date T . An American contract can
be exercised at any time t before the maturity date T . At maturity, the
European and American derivatives have identical values because it is the
last moment to exercise or let expire a contract. A call is in-the-money when
the underlying asset price is higher than the strike price. It is out-of-the-
money, if the underlying asset price is lower than the strike price. The call is
at-the-money if the underlying asset price is equal to the strike price. A put
is in-the-money when the underlying asset price is lower than the strike
price. It is out-of-the-money if the underlying asset price is higher than the
strike price. The put is at-the-money if the underlying asset price is equal
to the strike price. These definitions apply at maturity and at each instant
before expiration. When a long or a short position is initiated in a derivative
contract, the profit or loss is known when an opposite transaction is done
or when the option is at its maturity date. The profit or loss is computed
with respect to the purchase price or the sale price. It is possible to analyze
the profits and losses on options positions by looking at the initial price
in the transaction and the maturity price of the derivative contract. This
allows the computation of the profit and loss, P&L. This chapter presents
in detail the basic theory of commodities, futures, and forward markets.
It illustrates the specific features of these markets and the main pricing
relationships. It develops the main trading strategies in options markets.
In particular, strategies involving calls and puts, straddles, strangles,
conversions and reversals, and the box spread are studied. These trading
strategies can be used for most of the derivative assets in this book, since
they can be implemented using options with any particular payoff.
2.1. Introduction to Commodity Markets: The Case of Oil

The main commodity exchanges that currently trade oil futures contracts
are the Nymex, IPE of London, and the Simex.
2.1.1. Oil futures markets

A futures contract is successful when it is based on a volatile price, a
standard quality specification, and a wide range of participants in the
market. Volatile prices are necessary because they induce investors to
use futures markets. In fact, the prime function of a futures market is
to provide a hedging mechanism for the related industry. The standard
quality specification will attract the whole sector of the industry. The overall
volume traded by market participants give a good guide to show how liquid
a contract is. A second measure of a successful contract is open interest,
i.e., the total number of outstanding bought-and-sold contracts at the close
of each trading day. Commercial traders are companies whose business
involves handling the physical commodity, i.e., hedgers. Non-commercial
traders are mainly financial companies, i.e., speculators.
2.1.2. Oil futures exchanges

The most successful oil futures market is the Nymex. The success of this
market is largely due to the importance of its West Texas Intermediate
(WTI) crude contract.
The IPE of London trades only energy contracts. Unlike the Nymex,
the IPE has now chosen the cash settlement procedure rather than physical
delivery for the Brent crude contract. However, there is a physical delivery
option. The Simex offers an oil futures contract.
2.1.3. Delivery procedures

In general, futures contracts are almost never delivered since participants
prefer to close out (or roll over) their positions before the last trading day
for each delivery month. The main reason is that futures contracts are either
traded in conjunction with a position in the physical market (hedging) or
used in a speculative strategy. Physical delivery ensures that the futures
prices remain anchored to the underlying physical market. The physical

delivery of the standardized commodity specified in the futures contract or
its cash equivalent are not the only delivery mechanisms. Other procedures
are used like the exchange for physical (EFP), exchange for swaps (EFS),
and alternative delivery procedure (ADP).
2.1.4. The long-term oil market

The standardized futures markets such as the Nymex, the IPE, and the
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 emergence of a long-term oil market allows
one to meet specific requirements in the industry. This market provides new
opportunities for speculators. The participants in this new over-the-counter
(OTC) market are highly ranked oil companies, banks, and traders.
2.2. Pricing Models

The futures price, F can be described using different parameters: the spot
price S, the risk-less interest rate r, the cost of carrying the stocks b, and
the convenience yield cy. The cy represents short- to medium-term effects
related to physical supply and demand unbalance. It can be measured
using futures prices. For the case of oil, this yield indicates the intrinsic
oil interest rate.
2.2.1. The pricing of forward and futures oil contracts

Forward and futures contracts enable firms to determine a price for future
delivery. Most forward and futures contracts are traded for a few months
ahead. In the refined products markets, trading extends to a year. The
futures contract, Nymex and WTI can be traded up to four years ahead.
2.2.1.1. Relationship to physical market

Forward and futures prices can differ from the spot price of the commodity.
However, as the expiration date approaches, the prices must converge.
However, there are occasional squeezes on forwards and futures. When
the nearby contract is at a premium to the later month, the market is
in backwardation. When the nearby month is at a discount, the market is
in contango.
2.2.1.2. Term structure of prices

Changes in demand and in stock levels held by the industry affect forward
and futures prices. There is a limit on the size of the contango imposed
by the cash and carry arbitrage. In fact, buying the oil for one month,
paying for it, moving it into storage, insuring it, and delivering it back
to the market one month later at profit represent the cash and carry
arbitrage.
Example. A trader implements the following transactions:
• Buys heating oil at 60 cents/gallon in March

• Sells heating oil at 63 cents/gallon in April
In March, he takes delivery of the heating oil futures at 60 cents/gallon

He pays storage costs for 6 weeks, at 2.50 cents/gallon
The interest costs and product losses are 0.25 cents/gallon
The total cost in March is 62.75 cents/gallon
In April, he delivers the heating oil at 63 cents/gallon
The net profit is 0.25 cents/gallon
2.2.2. Pricing swaps

The price of a swap can be determined using the arbitrage relationships
between the swap and the forward or the futures markets.
A swap agreement can be replicated by a position in a portfolio of
futures or forward contracts.
Example. Consider a swap agreement between a producer of oil and a
swap provider. The swap allows the producer to sell a specified quantity
of crude oil at a fixed price over a period of one year. The swap can be
reproduced as a portfolio of short futures or forward contracts on the same
volume of oil for each delivery month.
2.2.3. The pricing of forward and futures commodity

contracts: General principles
The proposed relation between the futures price and the spot price, F =
SebT is useful. Re-call that F is the future price, S is the spot price, b
corrresponds to the carrying cost, and T is the time to maturity. This
relationship applies to futures prices and forward prices as well.
2.2.3.1. Forward prices and futures prices: Some definitions

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. The proof of this equivalence is based on a roll-
over strategy proposed in Cox et al. (1981). In this strategy, the investor
buys every day a specific amount of futures so that he/she holds er futures
contracts at the end of the first day of trading (initial time, day 0), e2r
futures contracts at the end of the day 1, e3r futures contracts at the end of
the day 2, etc., and eir futures contracts at the end of the day i − 1, where
r corresponds to the daily interest rate.
Since at the beginning of day i, the investor has eir contracts in his/her
position, the position on that day shows a profit (loss) of (Fi − Fi−1 )eir .
When this amount is invested until the day N corresponding to the
contract’s maturity date, where the number of days i is between 0 and N ,
this amount will be (Fi − Fi−1 )eir e(N −i)r = (Fi − Fi−1 )eN r .
The sum of the amounts of profit (loss) from the day 0 until day N turns
out to be (FN − F0 )eN r . However, since at the contract’s maturity date,
the futures price FN is equal to the spot price, ST , the terminal value of
this investment strategy is (FN − F0 )eN r = (ST − F0 )eN r . Using a portfolio
which corresponds to this strategy and an investment of an amount F0 in
a risk-free bond, gives the following payoff at time T :
(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.
2.2.3.2. Futures contracts on commodities

When the cost of carrying an asset refers to the storage costs of a
commodity such as silver or gold and when these costs are proportional
to the commodity price, the futures price is given by F = Se(r+a)T , where
“a” stands for the storage costs.
2.2.3.3. Futures contracts on a security with no income

When there are no distributions from the underlying asset, the cost of carry
is equal to the risk-less interest rate. The future price is given by F = SerT .
More generally, this relation represents the forward or futures price F as a
function of the spot price S. It applies to the valuation of forward or futures
contracts on a security that provides no income.
Example. Consider the valuation of a forward contract on a non-dividend
paying stock. Suppose the maturity date is in three months, the current
asset price is 100, and the three-month risk-free rate is 7% per annum.
In this case, S = 100, T = 0.25 year, r = 0.07, so the futures or forward
price is 101.765 or F = 100e(0.07)0.25 = 101.765.
2.2.3.4. Futures contracts on a security with a known income

For dividend-paying assets, the cost of carrying the stocks is given by the
difference between the risk-less rate and the dividend yield, d. This gives the
following relation F = Se(r−d)T . This relationship gives the forward price
F as a function of the spot price S for a forward contract on a security that
provides a known dividend yield.
Example. Consider the valuation of a three-month forward or futures
contract on a security that provides a continuous dividend yield of 5% per
annum. Suppose that the current asset price is 100 and the risk-free rate is
7% per annum. In this case, S = 100, T = 0.25 year, r = 0.07, and d = 5%,
so the futures or forward price is 100.501: F = 100e(0.07−0.05)0.25 = 100.501.
contract on the CAC 40 stock index. The index provides a continuous
dividend yield of 4% per annum. Suppose that the current index price
is 1000 and the risk-free rate is 7% per annum. In this case, S = 1000,
T = 0.25 year, r = 0.07, d = 4% per annum, so the futures or forward price
is 1007.528 F = 1000e(0.07−0.04)0.25 = 1007.528.
2.2.3.5. Futures contracts on foreign currencies

The cost of carrying a foreign currency is given by the difference between
the domestic risk-less rate and the foreign risk-less rate, r∗ . This gives
the following relation between the futures price and the spot price of the
∗
currency: F = Se(r−r )T . This relationship also gives the forward price F
(or foreign exchange rate F ) as a function of the spot price S for a forward
contract on a currency. This is often known in international finance as the
interest-rate parity theorem.

contract on a foreign currency. If the spot price is 180, the domestic risk-
free rate is 6% per annum and the foreign risk-free rate is 7% per annum,
then: S = 180, T = 0.25 year, r = 0.07, and r∗ = 6% per annum. The
future or forward price is 180.45.
F = 180e(0.07−0.06)0.25 = 180.45
2.2.3.6. Futures contracts on a security with a discrete income

When the cost of carrying the commodity is not a constant proportional
rate, the formulae discussed above must be slightly modified. In the
case of stocks paying known dividends, coupon-bearing bonds and some
commodities for which there are storage costs, the formula for future prices
becomes F = (S − I)erT . where I is the discounted value of the cash-flow
between t and t∗ . It is positive when it corresponds to an income and is
negative when it refers to a cost.
Example. Consider the valuation of a one-year forward contract on a two-

year bond. The two-year bond’s price is 800, the delivery price is 820,
and two coupons of 50 will be paid in 6 and 12 months, respectively. The
risk-less interest rate is 8% per annum for 6 months and 9% per annum,
for 12 months. In order to apply the formula, the value of I must be
discounted to the present at an appropriate interest rate. In this case, I is
given by:
I = 50e−0.08(0.5) + 50e−0.09(1) = 48.039 + 45.696 = 93.735
and the forward price is: F = (800 − 93.735)e0.09(1) = 706.265e0.09(1) =

772.777.
Example. Consider now the valuation of a one-year forward contract on a

stock with a price equal to 100. When the dividend is two and the interest
paid at the end of the year is 10% per annum, the present value of dividends
is I = 2e−0.01(1) = 1.8096 and the forward price is given by F = (100 −
1.8096)e0.1(1) = 108.516.
Example. Consider the valuation of a one-year futures contract on gold. If

the cost of carry is 3 per ounce paid at the end of the year, the spot price
is 500, and the risk-free rate is 10% per annum, the value of I is given by
I = 3e−0.1(1) = 2.7145 and the futures price is F = (500 + 2.7145)e0.1(1) =
555.585.
2.2.3.7. 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. We denote these by:
P: bond price in the cash market;
F: futures price;
T: time to maturity (the delivery date);
r: financing cost or rate and
c: coupon rate divided by the market bond price, known also as
the current yield.
Consider the following strategy: sell a futures contract at F , buy the
bond at P , and borrow an amount P at the rate r until the date T . At the
contract’s delivery date, the investor receives F plus the accrued interest
cT P . He must re-pay the loan P and the interest-rate charges, rT P . The
profit is given by the difference between the amount received and the outlay
or: P rof it = F + cT P − (P + rT P ). At equilibrium, in an efficient market,
the fair futures price must be:
0 = F + cT P − (P + rT P )
or: F = P (1 + T (r − c)). This price allows to avoid a cash and carry

arbitrage. Consider now the following strategy: buy a futures contract at
F , sell (short) the bond at P , and lend (invest) an amount P at the rate r
until the date T . At the contract’s delivery date, the investor pays F plus
the accrued interest cT P . He/she receives P and the interest earned, rT P .
The profit is given by the difference between the amount received and the
outlay or:
P rof it = P + rT P − (F + cT P )
At equilibrium, in an efficient market, the fair futures price must be:
0 = P + rT P − (F + cT P )
Hence: F = P (1 + T (r − c)). If this equation is not satisfied, then a reverse

cash and carry arbitrage can be implemented. The difference (r−c) refers to
the net financing cost or the cost of carry. The futures price is at a discount
with respect to the cash price (F < P ) when (c > r). The futures price is
at a premium with respect to the cash price (F > P ) when (c < r). The
futures price is equal to the cash price (F = P ) when (c = r).
Using the cash and carry arbitrage and the reverse cash and carry
arbitrage, it is possible to derive the theoretical fair price of a forward or
a futures contract. Our analysis assumes that there is only one deliverable
bond. However, in practice, futures contracts on treasury bonds are issued
on a number of deliverable issues and the futures price tracks in general,
the price of the bond which is the cheapest to deliver. Since the cheapest to
deliver is unknown before the delivery date, the futures price must account
for the price of the quality option: F = P + P (r − c) — quality option
premium. Since the exact delivery date is also unknown and the seller has
delivery options, the fair futures price must be:
F = P + P (r − c) — delivery option premiums
2.2.3.8. The pricing of future bond contracts

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:
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.
The process for the pricing of a futures contract must account for the
following relationship:
F (f c) = S + cp − O + e
Hence, its theoretical fair price must be:
F = (S + cp − O)/(f c) + e
2.3. Trading Motives: Hedging, Speculation, and Arbitrage

Futures markets are used for hedging, speculation, and arbitrage motives.
The main question for the producer is to implement the appropriate hedging
strategies in response to the changes in backwardation or contango.
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.
2.3.1. Hedging using futures markets

Companies using the physical oil market 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 in the futures or the forward market. In practice, the
hedge is rarely perfect.
The futures market offers a facility for hedging price risks. Hedging
price risk can be regarded as a trading operation allowing one to transform
a less acceptable risk into a more acceptable risk by engaging in an offsetting
transaction in a similar commodity under roughly the same terms as the
original transaction. In this spirit, a futures purchase, which is equal and
opposite to the physical trading contract is made with an idea that any
loss in the first transaction will be compensated by an equal gain in the
offsetting operation.
2.3.1.1. Hedging: The case of cocoa

The selling hedge
A company holding stocks of a commodity can protect itself against the
risk that the value of the unsold products will depreciate if the commodity
price falls. This risk is offset by a forward sale of the same tonnage on the
futures market. The hedge is based on the assumption that futures prices
decline as physical prices fall.
The buying hedge

Manufacturers of chocolate, exporters selling cocoa do not actually possess,
and manufacturers who fix selling prices and current contracts for future
delivery based on the current costs, etc. can implement a buying hedge.
These agents would suffer if the costs of the raw material rise. Therefore,
they can purchase the required tonnage on the futures market using a long
hedge. When the price falls by the time of shipment, the hedger will lose on
the futures market and profit from a cheaper purchase on the spot market.
The hedging strategy needs some funds to cover for the security deposits
and margins and any commission paid to brokers.
2.3.1.2. Hedging: The case of oil

Example: A short hedge
An oil trader who buys a cargo of physical oil while oil is in transit, can
protect himself/herself, by selling an equivalent volume of oil futures or
forward contracts. When the physical cargo is sold, the hedger can lift
his/her hedge by buying back the futures contract.
Example: A short hedge

Consider a company A buying 500,000 barrel cargo of crude oil for
$28/barrel when futures price is $28.50/barrel. The manager decides to
use the futures market to implement a hedging strategy by selling futures
contracts. A week later, the manager sells the physical cargo for $27.00/
barrel. He/she decides to lift the hedge by buying back the futures contracts
at their market price of $27.40/barrel (Table 2.1).
The profit on the futures position is larger than the loss on the physical
market. This result can be explained by this basis. It corresponds to the
differential between the futures and the spot price.
Table 2.1. Short hedge using futures contracts.
Period Physical market $/bl Futures market $/bl
t=0 bought at 28 futures sold at 28.50

t = +6 sold at 27.00 futures bought at 27.40
loss −1.00 profit +1.10
Table 2.2. Long hedge using futures contracts.
Period Physical market $/tonne Futures market $/tonne
t=1 sold at 200 futures bought at 199

t=3 bought at 210 futures sold at 208
loss −10 profit +9
Example: A long hedge

Consider a company B which is short of oil in the spot market. A
gasoil distributor agrees to sell oil to a customer at a fixed price for
some months ahead. The manager decides to use the futures market to
implement a hedging strategy by buying futures contracts in order to
protect himself/herself against a rise in price (Table 2.2).
This result can also be explained by this basis. It corresponds to the
differential between the futures and the spot price.
2.3.1.3. Hedging: The case of petroleum products futures contracts

When the futures markets tend to move parallel to the spot (cash) market,
a hedge can be implemented by buying or selling futures contracts. This is
possible because a loss due to an adverse price change in the cash market
can be offset by a gain in the futures market. Also, a loss in the futures
market can be offset by a gain in the cash market.
Using a Short Hedge

A short (selling) hedge is implemented when a decline in cash market prices
is expected. This is the case for a refiner or a distributor who is holding an
inventory. The short hedger forgoes the opportunity of additional profits in
a rising cash market.
Hedging by a petroleum marketer — Hedge of a fixed price purchase

commitment
Consider a marketer who is committed to buy a certain quantity of a
product at a given price. He/she expects a price decline and enters into
a short futures contract to protect his/her normal gross profit margin.
If the cash price of heating oil declines, the futures price can also decline.
Hence, the marketer will realize a loss or less of a gain on physical sale and
a profit on his/her futures position.
If the oil price increases, the resulting profit from the physical sale will
be offset by a loss on the futures market. In both cases, the marketer will
have protected his/her normal gross profit margin.
Hedging by a petroleum refiner

Consider a refiner who has not a firm sales commitment. Since there is a
time lag between the time the refiner purchased crude oil and delivers the
refined oil to the consumer, the price of oil can decline during this time
period. He/she can implement a hedge by selling futures contracts.
If the oil price declines, the resulting loss in the value of the inventory
will be offset by a gain on the futures market. In the same way, if the oil
price increases, the loss on the futures market will be offset by an increase
in inventory value.
Using a long hedge

A long (buying) hedge is implemented when a rise in cash-market prices is
expected. This is the case for a consumer of petroleum products.
The long hedger gives up the opportunity of increased profits should
the price of the physical asset decline. The hedger forgoes this opportunity
for the protection of his/her operating margins.
Hedging by a petroleum marketer — Hedge of a fixed-price

sale commitment
Consider a marketer who enters into long-term contracts to deliver products
to customers in the futures at a fixed price. His/her profit margin will be
at risk since the oil prices may increase before delivery. The marketer can
implement a hedge against price risk (a rise in the price in excess of the
contracted sale price) by buying futures contracts for the month of delivery.
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.
Hedging by a petroleum consumer to protect against

rising prices
Consider a petroleum consumer who expects an increase in petroleum
product prices. He/she can implement a hedge by buying futures contracts.
If the oil price increases, the consumer will have to pay the higher
market price at the time of purchase. The increased cost can be offset by
a gain on the futures market. In the same way, if the oil price decreases,
the gain on the physical transaction will be offset by a loss on the futures
market.
2.3.1.4. The use of futures contracts by petroleum products

marketers, jobbers, consumers, and refiners
Example A: Hedge by a petroleum products marketer
against a fixed-price purchase commitment
This example shows how a marketer can maintain his/her normal gross
profits in the physical market, regardless of the dynamics of the oil prices.
A marketer is commited to buy a fixed quantity at 90 cents per gallon.
He/she expects a price decline and implements a hedge by selling futures
contracts. In September, the marketer enters into a contract with a company
to supply 420,000 gallons of heating oil for delivery the following December.
If the current price of US$ 1 per gallon is expected to prevail at delivery,
the marketer expects a profit of 10 cents. The marketer can protect this
profit margin against a decline in oil prices by December. He/she can sell
10 heating oil contracts at US$ 1 per gallon for delivery in December (the
contract is equivalent to 42,000 gallons).
If the cash price declines to 80 cents, the marketer will realize a loss of
10 cents (90–80) on the physical sale.
However, a profit of 20 cents is achieved in the futures market since the

short futures contract is sold at US$ 1 and the investor bought it back at
80 cents.
The net effect of 10 cents corresponds to the original expected profit of
10 cents.
If oil prices rise, the resulting profit from the physical sale will be offset
by an equivalent loss on the futures market.
Example B: Hedge by a petroleum products marketer

against a fixed-price sales commitment
A marketer is commited to sell to customers a fixed quantity at US$ 1.30
per gallon. In June, the marketer enters into a sales contract to deliver
420,000 gallons of heating oil in December.
He/she expects a price decline from US$ 1.3 to US$ 1 in December. The
profit margin will be eroded if the marketer is forced to buy the heating oil
at a price in excess of the contract price of US$ 1.3.
The marketer can protect his/her position against the price risk by
buying 10 futures contracts at US$ 1 for delivery in December.
In December, he/she can accept the delivery or close out the position
by selling an identical futures contract.
If the oil price is US$ 1.4 per gallon, the marketer will lose 10 cents
on the physical market (to honor his/her sales contract price of 1.3). The
marketer can close his/her futures long position by selling a futures contract
at 1.4. A profit of 40 cents is achieved in the futures market since the long
futures contract is bought at US$ 1 and it is sold at US$ 1.4.
The net effect of 30 cents corresponds to the original expected profit of
30 cents.
If oil prices decrease to 90 cents, the resulting profit from the physical
sale is 40 cents, (selling at 1.3 minus cost of 90 cents).
Example C: Hedge of an existing asset (inventory)

position by a refiner
This example shows how a refiner can protect his/her inventory from an
erosion in value when prices fall, using the futures market.
A refiner has a risk resulting from the time lag between the time of
buying crude oil and the time of delivery to the consumer. When the oil price
declines during this time period, the value of his/her inventory will decline.
Consider a refiner who holds 420,000 gallons of uncommitted refined

inventory at an average cost of 90 cents per gallon at first in September.
He/she expects to sell his/her inventory in the cash market in December.
Expecting a possible decline below 90 cents, he/she sells 10 futures contracts
maturing in December at US$ 1.
At maturity, if the price declines to 80 cents, he/she loses 10 cents on
his/her inventory. When closing his/her futures position, he/she will realize
a gain of 20 cents (US$ 1 less 80 cents). This reduces the basis of his/her
inventory to 70 cents. In this context, he/she can sell the inventory in the
cash market at a lower cash price, and preserving a profit margin. If the oil
prices increase, this would be offset by a corresponding loss on the futures
position.
2.3.2. Speculation using futures markets

The main objective of speculators is profit. Speculators take on the risk
which hedgers try to lay off. Speculators hold onto their positions for a
very short time. They are sometimes in-and-out of the market several times
a day.
Example: Speculation on a price rise

Consider a speculator who expects prices to rise and buys consequently
IPE Brent futures contracts. A week later, the speculator closes out his/her
position in the futures market before the prices can fall back again. The
two trades are regarded as purely speculative because there is no physical
transaction in the spot market (Table 2.3).
2.3.3. Arbitrage and spreads in futures markets

Arbitrage keeps prices in line since the arbitrageur buys the asset in one
market and sells it in other market. When prices move out of line, the
arbitrageur buys the under-priced asset in one market and sells the over-
priced asset in another market.
Table 2.3. Speculating on a price rise.
Period Physical market $/bl Futures market $/bl
t=0 nothing — futures bought at 26.50

t = +6 nothing — futures sold at 25.00
— profit +1.50
Table 2.4. Heating-oil arbitrage.
Nymex Price Price IPE Price Price

Period heating oil cts/gall $/tonne gasoil $/tonne differential
t=0 Buy 3 at 62 170 Sell 4 at 170.10 −0.10

t = +1 Sell 3 at 61.50 169 Buy 4 at 164 +5.00
Loss −0.50 −1 Profit +6.10 +5.10
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.
2.4. The Main Bounds on Option Prices

The option value before maturity is a function of five parameters: the
price of the underlying asset, the strike price, the risk-free rate of interest,
the movements in the underlying asset prices (volatility), and the time
remaining to maturity. Option prices move at each instant of time as a
reaction to the changes in the above parameters. Since we know the option
pay-off at maturity, it is possible to determine some main relationships
before the maturity date. Hence, we can give minimum and maximum values
of calls and puts as well as the behavior of the option price in response to
the changing parameter values.
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:
CT = cT = max[0, ST − K], PT = pT = max[0, K − ST ]
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.
2.4.1. Boundary conditions for call options

When the call holder exercises his/her option, he/she receives the difference
between the underlying asset price and the strike price.
When the strike price is zero and the maturity date is infinite, the call
can be exercised with zero cost. This gives the call holder the right to receive
the underlying asset with zero cost. The value of this option must be equal
to that of its underlying asset. Hence, the call price must be between these
two limits.
2.4.2. Boundary conditions for put options

When the put holder exercises his/her option at maturity, he/she receives
the difference between the strike price and the underlying asset price. This
is the minimum value for the put.
When the underlying asset price tends to become zero, the put holder
can receive at maximum the value of the strike price. Hence, the maximum
value of the American put, which can be exercised at any time before
maturity must be the option’s strike price. However, the maximum value
for a European put must be the present value of the strike price since the
put holder must wait until expiration to exercise his option. Hence, the
minimum put price must be max[0, K − ST ]. The maximum price for an
American put is K and for a European put is Ke−r(T −t) , where t is the
current time and r is the risk-free interest rate.
2.4.3. Some relationships between call options

We define some main relationships which hold good for European and
American calls.
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
makes an equivalence between the European call price and the American
call price in the absence of distributions to the underlying asset.
Example: Consider a call when St = 110, K = 100, r = 7%, and T = 0.5

years. In this case, the call’s intrinsic value is 10, St −K. But, the call market
price must be at least Ct ≥ St −Ke−r(T −t) ≥ 110−100e−0.07(0.5) ≥ 13.4395.
The investor can exercise immediately the option of throwing away at least
the difference K − Ke−r(T −t) = 100 − 100e−0.07(0.5) = 3.435.
The investor discards also the difference between 13.4395−10 = 3.4395.
2.4.4. Some relationships between put options

We define some main relationships, which hold good for European and
American puts.
The put payoff shows that the put must be worth at expiration the
difference between the strike price and the underlying asset value.
The put payoff reveals that before maturity, the put value must be
worth at least the difference between the strike price and the underlying
asset value. This is because the American put holder can exercise his/her
option at any instant before maturity. It can also be shown that the
European put value must be at least equal to the present value of
the strike price minus the underlying asset price. Since the American
put value must be at least equal to K − St because of the possibility
of an early exercise, the European put value must verify the following
relationship:
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.
Example: Consider a European put trading at 2.5 when St = 95, K = 100,

r = 7%, and T = 0.5 years. Using this information, the put value satisfying
the following relationship: pt ≥ Ke−r(T −t) − St = 100e−0.07(0.5) − 95 =
1.5605, must be worth 1.5605.
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
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.
2.4.5. Other properties

Interest rate and option prices
Call prices are an increasing function of the interest rate. This is not the
case for put options.
Interest rate and call prices

The price of a European or an American call must satisfy the following
relationship:
Ct ≥ St − Ke−r(T −t)
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).
Interest rate and put prices

The price of a European put must satisfy the following relationship:
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.
Risk and option prices

Call and put prices are increasing functions of the risk of the underlying
asset.
Risk is measured by the standard deviation or the volatility of the
underlying asset’s returns.
With 25 years of experience with financial markets, I have learnt that
options are to be bought, not sold.
2.5. Simple Trading Strategies for Options and their

Underlying Assets
2.5.1. Trading the underlying assets
The value of any option underlying asset such as a stock, a default-free
bond, a futures or a forward contract is determined in a market place. In
general, there are two prices quoted: the bid and the ask price. These two
prices define a spread. In theory, there is only one price for each asset, but
in practice, these two prices are observed in financial markets. An investor
can buy (long position) or sell (short position) a financial instrument. It is
possible to represent the gain or loss from a transaction at a given date in
the future. If one buys a stock at 100 and sells it in a year at 110, the profit
is 10. If one buys a bond at 90 and sells it in a year at 100, the profit is 10.
The graph of the profit and loss as a function of the stock price at a given
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.
2.5.2. Buying and selling calls

Buying a call gives the right to the option holder to pay the strike price
K at maturity and to receive the value of an underlying asset ST . He/she
can also let the option expire. In this case, the call option is worth zero.
The positive difference between (ST − K) refers to the intrinsic value or
exercise value. Hence, at maturity, the value of a European or an American
call is given by CT = cT = max[0, ST − K]. This represents the value of a
long call position at expiration, where CT and cT refer respectively to the
American and European call values.
The profit for the option buyer represents exactly the loss of the option
seller and vice versa.
Example: Consider a call with K = 100 when ST = 90. At maturity, the

call buyer does not exercise his/her option because he/she pays 100 and
receives an underlying asset that is worth 90. He/she allows the option to
expire to avoid losing 10 since the option is worthless.
ST − K = 90 − 100 = −10 < 0
Hence, the payoff for the call buyer is:
CT = max[0, ST − K] = max[0, 90 − 100] = max[0, −10] = 0
However, if ST = 110, this gives an immediate payoff of 10 exercise, or:
CT = max[0, ST − K] = max[0, 110 − 100] = max[0, 10] = 10
The call seller implements a short position in the call. The value of a short
call position at expiration is:
−CT = −cT = −max[0, ST − K]
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.
Example: Consider a call with K = 100 when ST = 110. In this case,

upon exercise by the call holder, the option seller delivers a stock worth
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] = − max[0, 110 − 100] = −10
As we compute the payoffs at maturity, we can determine the profit or loss

P&L from holding long and short positions in options. If we denote the
value of the purchased call at time t or the cost of the call by Ct , then the
profit or loss on a long call position held until maturity is:
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]
Example: Consider a call with K = 100 when ST = 110. The investor

paid 10 for this call at time t = 0. In this case, if at the maturity date T ,
ST ≤ 110, the option buyer loses his/her option premium. However, when
100 < ST < 110, the call holder loses less than the purchase price 10. When
ST = 105, the call holder loses 5. In fact, he/she receives 5 upon exercise,
i.e., (105 − 100), coupled with the 10 paid for the option, his/her net loss
is 5. If ST = 110, he/she receives 10 upon exercise coupled with the 10 paid
for the option, gives a zero profit.
It is important to note that, in any case, the profits from the buyer and
the seller of the option sum up always to zero since the gains for the buyer
are the losses of the seller and vice versa. If an investor buys and sells calls
on the same underlying asset for the same maturity date, then the profits
and losses from the two operations are equal to zero at the call’s maturity
date. In fact, derivatives markets are zero-sum games because they do not
lead to net profits and losses. What the buyer wins is lost by the seller and
vice versa.
If one computes the sum of gains and losses from the long and short
call positions, the result is zero. In fact:
(CT − Ct ) + (Ct − CT ) = 0
since
(max[0, ST − K] − Ct ) + (Ct − max[0, ST − K]) = 0

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.
2.5.3. Buying and selling puts

Buying a put gives the right to the option holder to sell at maturity the
underlying asset ST at the strike price K. He/she can also let the option
expire. In this case, the put option is worth zero. The positive difference
between (K − ST ) refers to the intrinsic value or an exercise value. Hence,
at maturity, the value of a European or an American call is given by PT =
pT = max[0, K − ST ]. The value of a short put position at expiration is
−PT = −pT = − max[0, K − ST ].
Example: Consider a put with K = 100 when ST = 105. At maturity, the

put buyer does not exercise his/her option because he/she receives 100 and
delivers an underlying asset that is worth 105. He/she allows the option to
expire to avoid losing 5 since the option is worthless. Hence, the payoff for
the put buyer is:
PT = max[0, K − ST ] = max[0, 100 − 105] = max[0, −5] = 0
If ST = 100, this gives an immediate payoff of zero upon exercise, or:
PT = max[0, K − ST ] = max[0, 100 − 100] = max[0, 0] = 0
The holder of the put receives nothing upon exercise when ST ≥ K.

However, if ST < K, the put value is positive for any value of the stock.
If ST = 85, we have:
PT = max[0, K − ST ] = max[0, 100 − 85] = max[0, 15] = 15
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.
2.6. Some Option Combinations

It is possible to use calls or puts in different simple strategies. We can
combine calls and puts in order to create some option combinations that
are discussed below.
2.6.1. The straddle

The straddle is one of the simplest option strategies. An investor buying a
straddle buys simultaneously a call and a put on the same underlying asset
with the same strike price and maturity date.
An investor selling a straddle sells simultaneously a call and a put on
the same underlying asset with the same strike price and maturity date.
Example: Consider a call and a put on the same underlying asset and
maturity date. The call costs 10 and the put’s value is 8. The strike price
is 100. The P&L of the straddle corresponds to the combined profits and
losses from buying or selling both options.
Let us denote the current option prices by Ct and Pt and the option
prices at maturity by CT and PT . At maturity, the long straddle position
is worth:
CT + PT = max[0, ST − K] + max[0, K − ST ]
The cost of the long straddle is Ct + Pt . The maximum loss of

this strategy corresponds to the costs necessary to its implementation.
A trader buys, in general, a straddle when he/she expects erratic
movements in the underlying asset price. This refers to a very volatile
market for the underlying asset. At maturity, the short straddle position
is worth:
−CT − PT = − max[0, ST − K] − max[0, K − ST ]
A trader shorts a straddle when he/she expects a stability in the underlying

asset price. The cost of the short straddle is −Ct − Pt .
2.6.2. The strangle

The strangle is one of the simplest option strategies. An investor buying
a strangle (long a strangle) buys simultaneously a call and a put on the
same underlying asset with different strike prices K1 and K2 at the same
maturity date. The call with a strike price K1 and the put with a strike
price K2 are both out-of-the money with K1 > K2 . An investor selling

a strangle (short a strangle) sells simultaneously a call and a put on the
same underlying asset with two strike prices. The only difference with the
straddle is the presence of two different strike prices.
The long strangle is worth at maturity:
CT (ST , K1 , T ) + PT (ST , K2 , T ) = max[0, ST − K1 ] + max[0, K2 − ST ]
The cost of the long strangle is Ct (St , K1 , T ) + Pt (St , K2 , T ). A trader

buys, in general, a strangle when he/she expects erratic movements in the
underlying asset price. The short strangle is worth at maturity:
−CT (ST , K1 , T ) − PT (ST , K2 , T ) = − max[0, ST − K1 ] − max[0, K2 − ST ]
The cost of the short strangle is −Ct (St , K1 , T ) − Pt (St , K2 , T ). A trader

sells, in general, a strangle when he/she expects some stability in the
underlying asset price.
2.7. Option Spreads

2.7.1. Bull and bear spreads with call options
A spread consists in buying an option and selling another option. A bull
spread corresponds to a combination of options with the same underlying
asset and the same maturity, but different strike prices. When it is
implemented with two calls, it is designed to have a profit from a rise in
the underlying asset price. This strategy limits the risks and the potential
of profits.
The bull spread buyer buys an in-the-money call and sells an out-of-
the-money call. At maturity, the long bull spread is worth:
CT (ST , K1 , T ) − CT (ST , K2 , T ) = max[0, ST − K1 ] − max[0, ST − K2 ]
The cost of the long bull spread is Ct (St , K1 , T ) − Ct (St , K2 , T ).

A bear spread corresponds to a combination of options with the same
underlying asset and the same maturity, but different strike prices. When it
is implemented 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. The bear spread buyer buys the call with the higher strike
price and sells the call with a lower strike price. At maturity, the long bear
spread is worth:
−CT (ST , K1 , T ) + CT (ST , K2 , T )

= − max[0, ST − K1 ] + max[0, ST − K2 ]
The cost of the long bear spread is −Ct (St , K1 , T ) + Ct (St , K2 , T ).

A trader implements a bull spread with calls when he/she expects the
underlying asset price to rise. He/she implements a bear spread with calls
when he/she expects the underlying asset price to fall.
2.7.2. Bull and bear spreads with put options

A bull spread corresponds to a combination of puts with the same
underlying asset and the same maturity, but different strike prices. When
it is implemented with two puts, the trader buys a put with a lower strike
price and sells a put with a higher strike price. The bear spread trader buys
a put with a higher strike price and sells a put with a lower strike price. At
maturity, the bull spread is worth:
PT (ST , K1 , T ) − PT (ST , K2 , T ) = max[0, K1 − ST ] − max[0, K2 − ST ]
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:
−PT (ST , K1 , T ) + PT (ST , K2 , T ) = − max[0, K1 − ST ] + max[0, K2 − ST ]
The cost of the bear spread is −Pt (St , K1 , T ) + Pt (St , K2 , T ).

A trader implements a bull spread with puts when he/she expects the
underlying asset price to rise. He/she implements a bear spread with puts
when he/she expects the underlying asset price to fall.
2.7.3. Box spread

2.7.3.1. Definitions and examples
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. Since at maturity, the long bull spread is worth:
CT (ST , K1 , T ) − CT (ST , K2 , T ) = max[0, ST − K1 ] − max[0, ST − K2 ]

and at the same date, the bear spread with puts is worth:
−PT (ST , K1 , T ) + PT (ST , K2 , T ) = − max[0, K1 − ST ] + max[0, K2 − ST ]
then at maturity, the box spread is worth:
CT (ST , K1 , T ) − CT (ST , K2 , T ) − PT (ST , K1 , T ) + PT (ST , K2 , T )
This is equal to:
max[0, ST − K1 ] − max[0, ST − K2 ] − max[0, K1 − ST ] + max[0, K2 − ST ]
This strategy is “neutral” since it produces a return equal to the risk-less

rate of interest.
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:
max[0, ST − 90] − max[0, ST − 100] + max[0, 100 − ST ] − max[0, 90 − ST ]
When the underlying asset ST = 98, the payoff is:
max[0, 98 − 90] − max[0, 98 − 100] + max[0, 100 − 98] − max[0, 90 − 98]
or
8 − 0 + 2 − 0 = 10.
When the underlying asset ST = 70, the payoff is:
max[0, 70 − 90] − max[0, 70 − 100] + max[0, 100 − 70] − max[0, 90 − 100]
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) .
2.7.3.2. Trading a box spread

The box-spread strategy can be implemented with options on spot or
options on futures. In the following discussion, c(K1 ), c(K2 ), p(K1 ), and
p(K2 ) denote respectively, the prices of calls and puts with strike prices K1
and K2 with K2 > K1 .
Consider a portfolio corresponding to the following strategy:
a long bullish call spread: buy c(K1 ) and sell c(K2 )
a long bearish put spread: buy p(K2 ) and sell p(K1 )
This strategy is a box spread which costs c(K1 ) − c(K2 ) − p(K1 ) + p(K2 ).
The following non-arbitrage condition must be satisfied:
c(K1 ) − c(K2 ) − p(K1 ) + p(K2 ) ≤ (K2 − K1 )e−rT
At the maturity date, the result of the strategy is always (K2 − K1 ). In

fact, the pay-off of each option is:
max[ST − K1 , 0] − max[ST − K2 , 0] − max[K1 − ST , 0] + max[K2 − ST , 0]
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 :
C(K1 ) − c(K1 ) ≥ C(K2 ) − c(K2 ), P (K2 ) − p(K2 ) ≥ P (K1 ) − p(K1 )
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
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.
2.8. Butterfly Strategies

2.8.1. Butterfly spread with calls
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. An investor long the butterfly buys the call with the lowest
strike price K1 , buys the call with the highest strike price K3 , and sells two
calls with an intermediate strike price K2 . At maturity, the long butterfly
spread is worth:
CT (ST , K1 , T ) − 2CT (ST , K2 , T ) + CT (ST , K3 , T )

= max[0, ST − K1 ] − 2 max[0, ST − K2 ] + max[0, ST − K3 ]
The cost of a long butterfly spread is:
Ct (St , K1 , T ) − 2Ct (St , K2 , T ) + Ct (St , K3 , T )
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:
−CT (ST , K1 , T ) + 2CT (ST , K2 , T ) − CT (ST , K3 , T )

= − max[0, ST − K1 ] + 2 max[0, ST − K2 ] − max[0, ST − K3 ]
The cost of the short butterfly spread with calls is:
−Ct (St , K1 , T ) + 2Ct (St , K2 , T ) − Ct (St , K3 , T )
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
2.8.2. Butterfly spread with puts

A butterfly spread corresponds to a combination using three puts with three
strike prices K1 < K2 < K3 . The calls have the same underlying asset and
maturity date. An investor long the butterfly buys the put with the lowest
strike price K1 , buys the put with the highest strike price K3 , and sells two
puts with an intermediate strike price K2 . At maturity, the long butterfly
spread with puts is worth:
PT (ST , K1 , T ) − 2PT (ST , K2 , T ) + PT (ST , K3 , T )

= max[0, K1 − ST ] − 2 max[0, K2 − ST ] + max[0, K3 − ST ]
The cost of a long butterfly spread with puts is:
pt (St , K1 , T ) − 2Pt (St , K2 , T ) + Pt (St , K3 , T )
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:
−PT (ST , K1 , T ) + 2PT (ST , K2 , T ) − PT (ST , K3 , T )

= − max[0, K1 − ST ] + 2 max[0, K2 − ST ] − max[0, K3 − ST ]
The cost of the short butterfly spread is:
−Pt (St , K1 , T ) + 2Pt (St , K2 , T ) − Pt (St , K3 , T )
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
2.9. Condor Strategies

2.9.1. Condor strategy with calls
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. An investor long the condor buys the call with the
lowest strike price K1 , sells the call with a somewhat higher strike price
K2 , sells the call with the yet higher strike price K3 , and buys the calls
with the highest strike price K4 . At maturity, the long condor with calls
is worth:
CT (ST , K1 , T ) − CT (ST , K2 , T ) − CT (ST , K3 , T ) + CT (ST , K4 , T )

= max[0, ST − K1 ] − max[0, ST − K2 ] − max[0, ST − K3 ]
+ max[0, ST − K4 ]
The cost of a long condor with calls is:
Ct (St , K1 , T ) − Ct (St , K2 , T ) − Ct (St , K3 , T ) + Ct (St , K4 , T )
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:
−CT (ST , K1 , T ) + CT (ST , K2 , T ) + CT (ST , K3 , T ) − CT (ST , K4 , T )

= − max[0, ST − K1 ] + max[0, ST − K2 ] + max[0, ST − K3 ]
− max[0, ST − K4 ]
The cost of a short condor with calls is −Ct (St , K1 , T ) + Ct (St , K2 , T ) +

Ct (St , K3 , T ) − Ct (St , K4 , T ).
A trader implements a long condor with calls when he/she expects the
underlying asset price to be relatively stable. He/she implements a short
condor with calls when he/she expects strong movements in the underlying
asset price.
2.9.2. Condor strategy with puts

A condor corresponds to a combination using four puts with four strike
prices K1 < K2 < K3 < K4 . The puts have the same underlying asset and
maturity date. An investor long the condor buys the put with the lowest
strike price K1 , sells the put with a somewhat higher strike price K2 , sells
the put with the yet higher strike price K3 , and buys the put with the
highest strike price K4 .
At maturity, the long condor with puts is worth:
PT (ST , K1 , T ) − PT (ST , K2 , T ) − PT (ST , K3 , T ) + PT (ST , K4 , T )

= max[0, K1 − ST ] − max[0, K2 − ST ] − max[0, K3 − ST ]
+ max[0, K4 − ST ]
The cost of a long condor with puts is:
Pt (St , K1 , T ) − Pt (St , K2 , T ) − Pt (St , K3 , T ) + Pt (St , K4 , T )
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:
−PT (ST , K1 , T ) + PT (ST , K2 , T ) + PT (ST , K3 , T ) − PT (ST , K4 , T )

= − max[0, K1 − ST ] + max[0, K2 − ST ] + max[0, K3 − ST ]
− max[0, K4 − ST ]
The cost of a short condor with puts is:
−Pt (St , K1 , T ) + Pt (St , K2 , T ) + Pt (St , K3 , T ) − Pt (St , K4 , T )
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.
2.10. Ratio Spreads

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. A 2:1 ratio spread corresponds,
for example, to a strategy in which the trader buys two options and sells
an option. It is possible to use options with different maturities.
A spread based on options with different times to maturity is referred to
as a calendar spread. The investor can implement different combinations of
options with different strikes and time to maturity in order to construct
a wide range of profit and loss profiles. These strategies can also be
implemented in connection with the underlying assets and in particular
with stocks, bonds, foreign currencies, commodities, etc.
Using simultaneously options and their underlying assets allows one to

adjust payoff patterns to fit the attitude of investors toward the risk-return
profile.
A trader implements ratio spreads with calls and puts in different
contexts according to his/her future expectations about the underlying
asset price and his/her risk-return profile.
2.11. Some Combinations of Options with Bonds

and Stocks
2.11.1. Covered call: short a call and hold
the underlying asset
This strategy is implemented by selling the call and buying a certain
quantity of the underlying asset. This strategy corresponds to a covered
strategy since the investor owns the underlying asset. This asset covers the
obligation inherent in selling the call. This strategy enhances income.
2.11.2. Portfolio insurance

Portfolio insurance is an investment-management technique that protects a
portfolio from drops in value. This technique proposes some simple concepts
allowing one to insure a stock portfolio. The strategy can be implemented
using options, futures contracts, and other financial products. Consider, for
example, a well-diversified portfolio of stocks. Portfolio insurance can be
implemented in its basic form by buying a put on the owned portfolio
of assets. At maturity or the horizon date, the value of the insured
portfolio corresponds to the sum of the stock portfolio ST and the put
PT written on this portfolio. The value of the insured portfolio can be
written as:
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.
2.11.3. Mimicking portfolios and synthetic instruments

A trader can use European options in connection with other instruments to
create some specific payoff patterns at expiration. Two main relationships
are often used: mimicking portfolios and synthetic instruments.
A mimicking portfolio: It shows the same results (profits and losses) as
the instrument it mimics, but it might not have the same value.
A synthetic instrument: It exhibits the same results and value as the
instrument it synthetically replicates.
2.11.3.1. Mimicking the underlying asset

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. At maturity, the payoff of this position is:
cT − pT = max[0, ST − K] − max[0, K − ST ]
The initial cost of this position is ct − pt . Assume that at time t, K = St .

In this setting, the value of the portfolio comprising a long call and a short
put is obtained by replacing K with St in the previous equality:
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.
2.11.3.2. Synthetic underlying asset: Long call plus

a short put and bonds
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.
Using a risk-free bond (as a proxy for investing a certain amount K at the
risk-free interest rate r), it is possible to create a portfolio that synthesizes
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
At maturity, if ST > K, the call’s value is ST − K. The put value is

zero. The value of the portfolio is simply that of the call and the bond or
ST − K + K = ST .
At maturity, if ST < K, the call’s value is zero and the put’s value
is K − ST . The value of the portfolio simply corresponds to that of the
short position in the put and the bond or −(K − ST ) + K = ST . Hence, as
discussed before, the value of a portfolio comprising a long call and a short
put and the bond is equivalent to that of the underlying asset.
Buying a call, selling a put, and investing in a risk-free discount bond
is a strategy equivalent to an investment in the option’s underlying asset.
2.11.3.3. The synthetic put: put-call parity relationship

The put-call parity relationship stipulates simply that buying a call, selling
a put, and investing in a risk-free discount bond is a strategy equivalent
to an investment in the option’s underlying asset. Hence, using three of
these four instruments allows one to synthesize the fourth instrument. The
put-call parity relationship is often presented in the following form:
pt = ct − St + Ke−r(T −t)
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.
2.12. Conversions and Reversals

A strategy can be implemented by going long a call and short a put. If
the underlying asset price is above the strike price at the option’s maturity
date, the put is worthless and the call’s value corresponds to the intrinsic
value. The position will behave exactly as the value of the underlying asset.
However, if the underlying asset price is below the strike price, the call is
worthless and the put’s value is its intrinsic value. The position will again
behave exactly as the value of the underlying asset.
Buying the call and selling the put is a position equivalent to buying the
underlying asset. More generally, the following no-arbitrage between a call
c, a put p, the underlying asset price S, and the strike price K must hold:
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).
2.13. Case study: Selling Calls (Without Holding the Stocks/

as an Alternative to Short Selling Stocks/the Idea of
Selling Calls is Also an Alternative to Buying Puts)
2.13.1. Data and assumptions
We consider a Risk capital of 100,000 to invest for 3 months.
When the stock is at 15 and the strike price at 15, using option valuation
procedure and assuming a volatility of 20%, the interest rate at 5%, the
call price is 0.69.
If we change the volatility to 45%, the option price is 1.5.
The results show a profit in each scenario.
2.13.1.1. Selling calls (without holding the stock)

The investor (DC) sells (short) 6,666 calls.
This is a covered position because its aggregate underlying value is no
greater than the risk capital that would have gone into buying stocks
instead of selling calls: 100,000.
Aggregate underlying value = stock price × number of options sold
× contract multiplier
100,000 = 15 × 6, 666 × 1
The short sale of 6,666 calls would result in proceeds of 4,599.54, which
would be invested in T-bills during the holding period:
Proceeds of call short sale = call price × number calls × multiplier
4, 599.54 = 0.69 × 6, 666 × 1
The pattern of risk return associated with this strategy is as follows:
If the stock price is unchanged or down at expiration, the investment profit
is 4,657.03.
If the stock is up by 10%, from 15 to 16.5, each option has an exercise
value of 1.5 for an aggregate exercise value of 9,999.
Offset by the initial proceeds of the short sale and income from T-bills, the
investor’s loss is 4,542.05. Recall, if the investor sold a portfolio of stocks
short, his loss was 9,750.
Comparison with selling short the stock:

— When the investor short sells calls, with the stock unchanged, he
experiences a profit of 4,657.03 and this is his maximum gain.
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%.
Stock unchanged: 15 Stock + 10%: 16.5 Stock − 10%: 13,5
Initial proceeds from 4,599.54 4,599.54 4,599.54

calls: 6,666 × 0.69
Exercise value of calls 0 (9,999) 0
1.5 × 6,666
Profit (loss) from calls 4,599.54 (5,399.46) 4,599.54
Interest income on 57.49 57.49 57.49
T-bills:
5% × 4,599.54 × 0.25
Total profit (loss) 4,657.03: maximum (4,542.05) 4,657.03: maximum
gain gain
Return on risk capital 4,65% (4,54%) 4,65%
Remark: No initial capital to use.
Profit if the stock is less than 15.6986, otherwise losses appear.
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%.
Stock unchanged: 15 Stock + 10%: 16.5 Stock − 10%:13,5
Initial proceeds from 9,999 9,999 9,999

calls: 6,666 × 1.5
1.5 × 6,666
Profit (loss) from calls 9,999 (0) 9,999
Interest income 124,98 124,98 124,98
on T-bills:
5% × 9,999 × 0.25
Total profit (loss) 10,123: maximum 124,98 10,123: maximum
gain gain
Remark: No initial capital to use.
Profit in each scenario if call sold at this price of 1.5.
Profit in every scenario for IC.
If the stock is higher than 16.518, losses appear.
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.
Case 1: C = 0,69
Proceeds per contract = (proceeds from calls + T-bills)

/(Number contracts × mulplier)
0,6986 = (4,599.54 + 57.49)/(6,666 × 1)
Upside tolerance point = Proceeds per contract + strike price
15.6986 = 0.6986 + 15
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
Proceeds per contract = (proceeds from calls + T-bills)

/(Number contracts × mulplier)
1,5187 = (9,999 + 124,98)/(6,666 × 1)
Upside tolerance point = Proceeds per contract + strike price
16.518 = 1,5187 + 15
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.
2.13.1.2. Comparing the strategy of selling calls (with a short portfolio of

stocks): the extreme case
Comparing covered short calls with a short portfolio of stocks, the short
calls are less risky.
The Extreme case: In theory, there is no limit to how far the stock might
rise over a given holding period, but as an extreme example, let’s see what
would happen if the stock doubled from 15 to 30.
Table 2.7. Extreme case: Stock up by 100%.
Proceeds from short sale of stocks 100,000,

Repurchase of stocks (200,000),
Loss from stocks (100,000),
Interest income on T-bills 5% × 100,000 × 0.25 1,250
Total loss (98,750)
Table 2.8. Stock up by 100%: position in the short calls: case 1.
Proceeds from sale of calls 6,666 × 0.69 4,599.54

Exercise value of calls 6,666 × 15 × 1 (99,990)
Loss from calls (95,390.46)
Interest income on T-bills 5% × 4, 599.54 × 0.25 57.49
Total loss (95,332.97)
Table 2.9. Stock up by 100%: position in the short calls: case 2.
Proceeds from sale of calls 6,666 × 1.5 9,990

Exercise value of calls 6,666 × 15 × 1 (99,990)
Loss from calls (90,000)
Interest income on T-bills 5% × 9,990 × 0.25 124.875
Total loss (89,875)
Position in calls: shows a net loss of (89,875).
The portfolio would show a loss of 100,000, offset by 1,250 in interest

income − a net loss of 98,750. The position with 6,666 calls, fares
somewhat better.
The investor must pay an aggregate exercise value of 99,990, but this is
offset by the initial proceeds from the calls and the interest income — the
net loss is (95,332.97), an improvement of 3,417.03. (98,750-95,332.97)
2.13.1.3. Selling calls (holding the stock)

The investor sells short 6,666 calls.
What must be the position in the stock:
To cover the position, we buy Delta units of the underlying asset.
Assume that the call’s delta is about 0.55 and that the option is on one
share.
At time zero, we sell the calls and we buy: 0.55×6,666×15 = 55,000 stocks.
2.13.2. Leverage in selling call options (without holding the

stocks)
If the investor of the previous example were willing to lower his upside
tolerance point in exchange for higher profits with the stock unchanged or
down, he could sell a larger number of calls.
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.
Stock Unchanged Stock + 10% Stock − 10%
Total profit (loss) on options 4,657.03 (4,542.05) 4,657.03

Portfolio of Stock’s ending value 55,000 60,500 49,500
(55,000 × 1,1)
Profit (loss) from stocks 0 5,500 (5,500)
Total profit (loss) from 4,657.03 958 (843)
options and stocks
Return on risk capital (in stocks) 8,46% 1,74% (1,53%)
= 4,657.03/55,000 = 958/55,000 = (843)/55,000
If we account for dividends of 1%, about 549 dollars, all the results may become positive.
and buying the underlying stock for a hedge: delta stocks: 0.55 stocks.
Stock Unchanged Stock + 10% Stock − 10%

Portfolio of Stock’s ending value 55,000 60,500 49,500
(55,000 X1,1)
Total profit (loss) from 9,999 5,500 4,499
options and stocks
Return on risk capital (in stocks) 18% 10% 8,2%
= 9,999/55,000 = 5,500/55,000 4,499/55,000
2.13.2.1. Selling Call options (without holding the stocks)

If the investor wished to commit all of his 100,000 dollars to margining
short calls he could sell:
Net requirement = 5%of stock price × contract multiplier
0.75 = 5% × 15 × 1
Maximum contracts to Margin = Total Margin deposit/Net requirement per
contract
133, 333 = 100,000/0.75
At 0.69, the proceeds from selling 133,333 calls would be:
Proceeds of call short sale = call price × number × contract multiplier
91,999 = 0.69 × 133,333 × 1
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.
Table 2.12. Results at the maturity date in three months of the strategy using leverge
in selling calls.
Stock unchanged Stock + 10% Stock − 10%
Initial proceeds from 91,999 91,999 91,999

calls: 133,333 × 0.69
1.5 × 133,3333
Profit (loss) from calls 91,999 (100,000) 91,999
Interest income on 1,149,98 1,149,98 1,149,98
T-bills:
5% × 91,999 × 0.25
Total profit (loss) 93,148.98 (98,850) 93,148.98
Return: Profit/initial 1,012 = 93,148.98 (1,074) = (98,850 1.012 = (93,148.98
proceeds /91,999 /91,999) /91,999)
Table 2.13. Results at the maturity date in three months of the strategy using leverge
in selling calls.
Initial proceeds from calls: 199,999 199,999 199,999

133,333 × 1.5
1.5 × 133,3333
Interest income on T-bills: 2,499.98 2,499.98 2,499.98
5% × 199,999 × 0.25
Total profit (loss) 202,498 2,499,98 202,498
Return: Profit/initial 101,2% = 202,498 1,02% = (2,499.98 101,2% = (202,498
proceeds /199,999 /199,999) /199,999)
This will be the maximum profit (as other short positions).

But, if the stock rises, he is exposed to unlimited losses.
If the stock rises 10%, the aggregate exercise value is (199,999).

Offset by the initial proceeds from the sale of calls and the interest on the
TB, the investor total loss is (98,850).
The investor loss can be more than the 100,000 marging deposit.
Losses in short calls are almost limitless. (Unlike the case with short puts,
in which losses are limited by the fact that the stock cannot trade below
zero.)
Table 2.14. Case 1. Leverage in selling Call options (without holding

the stocks): The extreme case stock up by 100%.

Exercise value of calls 133,3333 × 15 × 1 (1,999,990)
Loss from calls (1,907,999)
Interest income on T-bills 5% × 91,999 × 0.25 1,149.875
Total loss (1,906,849)
Table 2.15. Case 2. Leverage in selling Call options (without holding

the stocks): The extreme case stock up by 100%: call: 1.5.

Exercise value of calls 133,3333 × 15 × 1 (1,999,995)
Loss from calls (1,799,996)
Interest income on T-bills 5% × 199,999 × 0.25 2,499.9875
Total loss (1,7907,496)
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).
2.13.2.3. Selling calls using leverage (and holding the stock)

The investor sells short 133,3333 calls.
What must be the position in the stock?
To cover the position, we buy Delta units of the underlying asset.
Assume that the call’s delta is about 0.55 and that the option is on one
share.
At time zero, we sell the calls and we buy stocks: 0.55 × 133,3333 ×15 =
1,099,997.
2.13.3. Short sale of the stocks without options

An investor (DC) buys stocks in anticipation of a rise in the market.
An investor (DC) can sell stocks short in anticipation of a decline.
and buying the underlying stock.
Total profit on 93,148.98 (98,850) 93,148.98

options (loss)
Portfolio of Stock’s 1,099,997 1,209,996 989,997
ending value 0.55 × 133,3333 × 0.55 × 133,3333 × 0.55 × 133,3333 ×
15(1,099,997) 15(1,099,997) × 1,1 15(1,099,997) × 0,9
Profit (loss) from 0 109,999 (109,999)
stocks
Total profit (loss) 93,148.98 11,149 (16,851)
from options
and stocks
Return on options 100% − 100% 100%
Return on stocks 0 10% (−10%)
and buying the underlying stock.

Portfolio of Stock’s 1,099,997 1,209,996 989,997
ending value
0.55 × 133,3333 × 0.55 × 133,3333 ×
15(1,099,997) × 1 15 × 1.1
Total profit (loss) from 93,148.98 109,999 90,0000
options and stocks
Return on options 100% 0% 100%
Return on stocks 0 10% (−10% )
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.
Table 2.18. Short selling stocks (stock lending).
Stocks Stock unchanged Stock + 10% Stock − 10%
Short sale stocks 100,000 100,000 100,000

Repurchase of stocks 100,000 110,000 90,000
Profit(loss) from stocks 0 (10,000) 10,000
Interest income from T-bills 1,250 1,250 1,250
(5%100, 000 × 0.25)
Restitution of dividends: 1% (1,000) (1,000) (1,000)
Total profit or loss 250 (9,750) 10,250
Table 2.19. Short selling stocks (stock lending) Example: double the dividend payment.
Stocks Stock unchanged Stock + 10% Stock − 10%
Short sale stocks 100,000 100,000 100,000

Repurchase of stocks 100,000 110,000 90,000
Profit(loss) from stocks 0 (10,000) 10,000
Interest income from T-bills 1,250 1,250 1,250
(5%100,000 × 0.25)
Restitution of dividends: 1% (2,000) (2,000) (2,000)
Total profit or loss (750) (10,750) 9,250
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
0.25% = (1, 250 − 1, 000)/100, 000 = 0.0025
The upside tolerance point is:

Upside tolerance point = stock price + holding period net interst yield.
15.0025 = 15 + 0.0025
If restitution of dividend payments through expiration is greater than

the interest income from the reinvestment of the proceeds of the short sale,
the investor will show a loss if the stock is unchanged.
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
0.75% = (1, 250−2, 000)/100, 000 = −0.0075
The upside tolerance point is:

Downside breakeven point = stock price + holding period
interst yield
14.9925 = 15 − 0.0075
A position of $ 100 millions.
How options can be used as alternatives to a direct investment
in stocks?
2.14. Buying Calls on EMA

2.14.1. Buying a call as an alternative to buying the stock:
(also as an alternative to short sell put options)
Buying calls can be a unique managerial alternative to buying stocks.
Another alternative is to short sell put options.
2.14.1.1. Data and assumptions

An investor with 100,000 to invest for 3 months in the call. When the stock
is at 15, he can either buy the stock or invest in T-bills. This leads to
symmetrical risk and return pattern, profit or loss dollar for dollar plus
income from dividends.
2.14.1.2. Pattern of risk and return

Consider the pattern of risk and return with a purchase of 3 month call
option with a strike price of 15. We selected at the money so that at
expirations, appreciation in the underlying will be reflected in the exercise
value of calls.
To scale the position in calls to the investor Risk capital of 100,000, we
begin by calculating the value of one contract.
The stock price times the multiplier.
Call’s underlying value = asset price × contract multiplier

= 15 × 1
Number of calls to buy:
To determine the number of calls to buy, we divide the risk capital of
100,000 by call’s underlying value of 15:
Contracts required = risk capital: call’s underlying value

6,666 = 100000: 15
Valuation:
Using option valuation procedure and assuming a volatility of
30%, the interest rate at 5%, the call price is 0.98746.
Assuming options can be purchased at theoretical value, the
investor would have to spend to buy 6,666 contracts:
Investment required to buy = call price × number × contract

multiplier
$6 583.608 = 0.98764 × 6,666 × 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
93,416,391 = 100,000 − 6,583
Strategy: invest an amount in options and the balance in T-bills.

Buy calls, strike 15, stock initially at 15 at 0.98746.
Invest: 93,416,391 in 3 month TB, Hold for 3 months, Collect interest.
2.14.2. Compare buying calls (as an alternative to portfolio

of stocks)
Here we compare buying calls as an alternative to portfolio of stocks.
At maturity After 3 months:
If the asset price does not change:
Initial cost of calls (6,583.6080)
Exercise value: 0
Loss from calls (6,583.6080)
Interest income on T-bills

(5% on 93,416,391 for 3 months: 1,167.704)
Total loss: (5,415.903)
If we examine the outcomes of investing directly in stock, the
return is zero.
Return in stock: 0
2.14.2.1. Risk return in options

If the investor uses options to take a position in the market, the position is
not profitable unless the asset is high enough to give options sufficient
exercise value to repay their initial cost.
The upside break even point can be computed as follows.
Cost per contract = (invt to buy calls-T-bill income)

/(number of contracts × multiplier contract)
= (6,583.608 − 1,167.704)/(6,666 × 1)
0.8124 = 5, 415.904/6,666
Upside break-even point = Net cost per contract + call’s strike

price
$15.8124 = 0, 8124 + 15
The Upside break-even point is $15.8124.
Each call at this point would has an exercise value of 0,8124.
The aggregate exercise value of the position would be 5,415.455 just above the
amount required to repay the cost less the offsetting treasury bill income:
0,8124 × 6, 666 = 5,415.455
Exercise value of call = stock price upside — option strike price ×

contract multiplier
0, 8124 = (15.8124 − 15) × 1

Aggregate exercise value of position = exercise value of call ×
number of calls in position
5, 415.455 = 0, 8124 × 6, 666

If the stock rallies beyond the upside break even point, the position will
show a healthy profit.
For example, 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 4,584.26.
At maturity after 3 months:
If the asset price up by 10%:
Initial cost of calls (6,583.6080)
Exercise value: 1.5 × 6,666 × 1 9,999
Profit from calls 3,415.4
(5% on 93,416,391 for 3 months: 1,167.704) 1,167.704
Total profit: 4,584.26
Return from options and T-bills: 4,58%
This profit is less than a direct investment in the stock:
Initial cost of stocks = 100,000
Sale of stocks = 110,000
Profit = 10,000
Return from the stock: 10%
When the investor established his position in stocks, he earned a profit of
10,000.
With options, he can not break even until the stock rises to $15.8124.
With stocks, his profit with the stock up by 10% was 10,000, with options
only 4,584.26.
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, 415.904 = 6583.608 − 1, 167.704
This example is done to replicate a position in the stock.
When the investor chooses to buy options, he is making a different kind of

investment.
The tradeoff is: in exchange of a known maximum loss (the option
premium), he sacrifies performance.
The tradeoff is equitable.
This is a managerial choice driven by the attitude toward the pattern of
risk and returns.
2.14.3. Example by changing volatility to 20%

2.14.3.1. Data and assumptions:
An investor with 100,000 to invest for 3 months in the call.
When the stock is at 15.
He can either buy the stock or invest in T-bills. This leads to symmetrical
risk and return pattern, profit or loss dollar for dollar plus income from
dividends.
Pattern of risk and return

Consider the pattern of risk and return with a purchase of 3 month call
option with a strike price of 15.
We selected at the money so that at expiration any appreciation in the
underlying will be reflected in the exercise value of calls.
To scale the position in calls to the investor Risk capital of 100,000, we
begin by calculating the value of one contract.
The stock price times the multiplier
Valuation:
Using option valuation procedure and assuming a volatility of
30%, the interest rate at 5%, the call price is 0. 96.
Call’s underlying value = asset price × contract multiplier
15 = 15 × 1
Number of calls to buy:

Contracts required = Risk capital: call’s underlying value
6,666 = 100,000: 15
Assuming options can be purchased at theoretical value, the investor would

have to spend to buy 6,666 contracts:
Investment required to buy = call price × number × contract

multiplier
$4,599 = 0.69 × 6,666 × 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
95,400.46 = 100,000 − 4,599
Strategy: Buying calls and investing balance in T-bills

Buy calls, strike 15, stock initially at 15 at 0.69
Invest: 95,400.46 in 3 month TB
2.14.3.2. Compare buying calls (as an alternative to portfolio of stocks.)

Here we compare buying calls as an alternative to portfolio of stocks.
At maturity: After 3 months:
If the asset price does not change:
Initial cost of calls (4,599)
Exercise value: 0
Loss from calls (4,599)
(5% on 95,400.46 for 3 months: 1,192.575)
Total loss: (3,406.425)
Return from investment is stocks: 0%
Return using call options:
If the investor uses options to take a position in the market, the position is
not profitable unless the asset is high enough to give options sufficient
exercise value to repay their initial cost.
The upside break even point can be computed as follows.
Cost per contract = (invt to buy calls − T-bill income)/(number
of contracts × multiplier contract)
0.5110 = (4,599 − 1,192.575)/(6,666 × 1)

Upside break-even point = net cost per contract + call’s strike

price
$15.511 = 0.5110 + 15 =
The Upside break-even point is $15.511.

Each call at this point would has an exercise value of 0,511.
The aggregate exercise value of the position would be 3,406.42 just above
the amount required to repay the cost less the offsetting treasury bill
income:
0,511 × 6, 666 = 3,406.42
Exercise value of call = stock price upside-option strike price ×

contract multiplier
0,5111 = (15.511 − 15) × 1
Aggregate exercise value of position = exercise value of call ×

number of calls in position:
3, 406.42 = 0, 511 × 6, 666
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:
Initial cost of calls (4,599)
Exercise value: 1.5 × 6,666×1 9,999
Profit from calls 5,400

(5% on 95,400.46 for 3 months: 1,192.575) 1,192.575
Total profit: 6,592.57
Return from options and T-bills: 6,59%
This profit is less than a direct investment in the stock:
Initial cost of stocks = 100,000
Sale of stocks = 110,000
Profit = 10,000
Return from stock: 10%
– When the investor established his position in stocks he earned a profit

of 10,000.
– With options, he cannot break even until the stock rises to $15.511.
– With stocks, his profit with the stock up by 10% was 10,000, with
options only 6,592.57.
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
This example is done to replicate a position in the stock.

When the investor chooses to buy options, he is making a different kind
of investment. The tradeoff is: in exchange of a known maximum loss (the
option premium), he sacrifies performance. The tradeoff is equitable
This is a managerial choice driven by the attitude toward the pattern of
risk and returns.
Buy calls, strike 15, stock initially at 15 at 0.69
Invest: 95,400.46 in 3 month TB
2.14.3.3. Leverage in buying call options (without selling the underlying)

What happens if the investor devotes the risk capital to buy calls?
With this capital, the investor can buy 145,000.
Table 2.20. Results at the maturity date in three months of the strategy of buying calls
and holding the underlying stock.
Investment required to buy (4,599) (4,599) (4,599)

calls: 0.69 × 6,666 × 1
Exercise value of calls 0 9,999 (1.5 × 6,666) 0
Profit (loss) from calls 0 5,400 0
Interest income on T-bills: 1,192.575 1,192.575 1,192.575
5% on 95,400.46
Total profit (loss) (3,406.425) 6,592.57 (3,406.425)
Table 2.21. Leverage in buying call options (without selling the underlying).
Asset level unchanged Stock + 10% Stock − 10%
Initial cost of calls (100,000) (100,000) (100,000)

Exercise value of calls 0 217,500 0
1.5 × (145, 000calls)
Profit (loss) from calls (100,000) 117,500 (100,000)
Total profit and loss (100,000) 117,500 (100,000)
Return/initial calls (100%) 117.5% (100%)

Number of contracts = risk capital/(cost per contract × contract multiplier)
145,000 = 100,000/(0.69 × 1)
If the asset declines or is unchanged, the calls expire worthless.

The investor will lose 100,000.
In the previous example, with the purchase of 6,666 contracts, the loss was
3,406.
This reflects the smaller initial cost of the position and the interest income
from T-bill. He can no longer afford not to buy.
If the asset rises by 10%, from 15 to 16.5, the investor profit will be 117,500.
With options, the investor has the benefit of an assured maximum loss- no
matter how much the stock declines, the loss can never be greater than the
price paid for options.
Using the procedure developed before, the break even point is:
Upside break-even point = net cost per contract + call’s strike price
$15.69 = 0.69 + 15.

Table 2.22. Hedging: buying call options (selling the underlying).
Asset level Unchanged Stock + 10% Stock − 10%
Total profit and loss from calls (100,000) 117,500 (100,000)

Initial stock position Sell: 1,196,250 1,196,250 1,196,250
0.55 × 15 × 145,000
Ending stock position 1,196,250 1,315,875 1,076,625
Profit and loss from stock (000) (119,625 ) 119,625
position
Total profit and loss from calls (100,000) (2,125) 19,625
and stocks
The investor is assured of a maximum loss of 100,000 with the possibility

of upside rewards that could potentially be greater than those of other
strategies with other instruments.
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
price risk can be regarded as a trading operation allowing to transform a less

acceptable risk into a more acceptable risk by engaging into an offsetting
transaction in a similar commodity under roughly the same terms as the
original transaction. Futures markets are often used for speculation. The
main objective for speculators is to accomplish profit. Speculators take on
the risk, which hedgers try to lay off. Speculators hold onto their positions
for a very short time. They are sometimes in and out of the market several
times a day. Arbitrage keeps prices in line since the arbitrageur buys the
assets in one market and sells it in other market. When prices move out
of line, the arbitrageur buys the under-priced asset in one market and sells
the over-priced asset in another market.
The option value before maturity is a function of five parameters: the
price of the underlying asset, the strike price, the risk-free rate of interest,
the movements in the underlying asset prices (volatility), and the time
remaining to maturity.
The value of any option underlying asset such as a stock, a default-free
bond, or a futures or a forward contract is determined in a market place.
In general, there are two prices quoted: the bid and the ask price. These
two prices define a spread. In theory, there is only one price for each asset,
but in practice, two prices are observed in financial markets.
Buying a call gives the right to the option holder to pay the strike price
K at maturity and to receive the value of an underlying asset ST . Buying a
put gives the right to the option holder to sell at maturity, the underlying
asset ST at the strike price K. The most frequent strategies consist in
buying and selling calls and puts with or without using the underlying
asset. Several other strategies can be implemented as a function of the
risk-reward profile. An investor selling a straddle sells simultaneously a call
and a put on the same underlying asset, with the same strike price and
maturity date.
The strangle is one of the simplest option strategies. An investor buying
a strangle (long a strangle) buys simultaneously a call and a put on the
same underlying asset with different strike prices K1 and K2 and at the
same maturity date. The call with a strike price K1 and the put with a
strike price K2 are both out-of-the money with K1 > K2 .
A spread consists in buying an option and selling another option. A bull
spread corresponds to a combination of options with the same underlying
asset and the same maturity, but different strike prices. A bear spread
corresponds to a combination of options with the same underlying asset
and the same maturity, but different strike prices. When it is implemented
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.
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?
CASE STUDY: COMPARISONS BETWEEN PUT AND

CALL OPTIONS
1. Buying Puts and Selling Puts Naked
We consider a strategy of buying puts using the same data as with calls for
the underlying assets.
The put price premium is 6% of the underlying asset price.
1.1. Buying puts

Allows to cover a position in the underlying asset. The investor pays a
premium of let’s say 6 % and hedges the exposure to the underlying asset’s
movements. If the market goes down, the investor can earn money.
P&L from buy put at Maturity "p"

30.00%
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%
Fig. 2.1. Buying a put.

1.2. Selling puts

The investor receives the premium of 6 %. He wins if the market goes up:
maximum gain limited to the option premium received. The investor loses
money if the market goes down.
P&L from sell put at Maturity -p
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%
Fig. 2.2. Selling a put.
The graphic shows P&L from the Table for buying and selling puts.
Table 2.1. P&L at maturity for different price of the underlying.
P&L from buy put P&L from sell put

Stock price at maturity at maturity, p at maturity, −p
−33.33% 10 27.33% −27.33%

−32.33% 10.15 26.33% −26.33%
−31.33% 10.3 25.33% −25.33%
−30.33% 10.45 24.33% −24.33%
−29.33% 10.6 23.33% −23.33%
−28.33% 10.75 22.33% −22.33%
−27.33% 10.9 21.33% −21.33%
−26.33% 11.05 20.33% −20.33%
−25.33% 11.2 19.33% −19.33%
−24.33% 11.35 18.33% −18.33%
−23.33% 11.5 17.33% −17.33%
−22.33% 11.65 16.33% −16.33%
−21.33% 11.8 15.33% −15.33%
−20.33% 11.95 14.33% −14.33%
−19.33% 12.1 13.33% −13.33%
−18.33% 12.25 12.33% −12.33%
−17.33% 12.4 11.33% −11.33%
(Continued)
Table 2.1. (Continued ).

−16.33% 12.55 10.33% −10.33%

−15.33% 12.7 9.33% −9.33%
−14.33% 12.85 8.33% −8.33%
−13.33% 13 7.33% −7.33%
−12.33% 13.15 6.33% −6.33%
−11.33% 13.3 5.33% −5.33%
−10.33% 13.45 4.33% −4.33%
−9.33% 13.6 3.33% −3.33%
−8.33% 13.75 2.33% −2.33%
−7.33% 13.9 1.33% −1.33%
−6.33% 14.05 0.33% −0.33%
−5.33% 14.2 −0.67% 0.67%
−4.33% 14.35 −1.67% 1.67%
−3.33% 14.5 −2.67% 2.67%
−2.33% 14.65 −3.67% 3.67%
−1.33% 14.8 −4.67% 4.67%
−0.33% 14.95 −5.67% 5.67%
0.67% 15.1 −6.00% 6.00%
1.67% 15.25 −6.00% 6.00%
2.67% 15.4 −6.00% 6.00%
3.67% 15.55 −6.00% 6.00%
4.67% 15.7 −6.00% 6.00%
5.67% 15.85 −6.00% 6.00%
6.67% 16 −6.00% 6.00%
7.67% 16.15 −6.00% 6.00%
8.67% 16.3 −6.00% 6.00%
9.67% 16.45 −6.00% 6.00%
10.67% 16.6 −6.00% 6.00%
11.67% 16.75 −6.00% 6.00%
12.67% 16.9 −6.00% 6.00%
13.67% 17.05 −6.00% 6.00%
14.67% 17.2 −6.00% 6.00%
15.67% 17.35 −6.00% 6.00%
16.67% 17.5 −6.00% 6.00%
17.67% 17.65 −6.00% 6.00%
18.67% 17.8 −6.00% 6.00%
19.67% 17.95 −6.00% 6.00%
20.67% 18.1 −6.00% 6.00%
21.67% 18.25 −6.00% 6.00%
22.67% 18.4 −6.00% 6.00%
23.67% 18.55 −6.00% 6.00%
24.67% 18.7 −6.00% 6.00%
25.67% 18.85 −6.00% 6.00%
26.67% 19 −6.00% 6.00%
27.67% 19.15 −6.00% 6.00%
(Continued)

28.67% 19.3 −6.00% 6.00%

29.67% 19.45 −6.00% 6.00%
30.67% 19.6 −6.00% 6.00%
31.67% 19.75 −6.00% 6.00%
32.67% 19.9 −6.00% 6.00%
33.67% 20.05 −6.00% 6.00%
34.67% 20.2 −6.00% 6.00%
35.67% 20.35 −6.00% 6.00%
36.67% 20.5 −6.00% 6.00%
37.67% 20.65 −6.00% 6.00%
38.67% 20.8 −6.00% 6.00%
39.67% 20.95 −6.00% 6.00%
40.67% 21.1 −6.00% 6.00%
41.67% 21.25 −6.00% 6.00%
42.67% 21.4 −6.00% 6.00%
2. Buying and Selling Calls

A strategy of buying and selling calls shows the following P&L on the same
stocks.
2.1. Buying calls

The investor pays a premium of 10 %. He wins if the market goes up.
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%
Fig. 2.3. Buying a call.

2.2. Selling a call

The investor receives a premium of 10 %. He wins if the market goes down.
He loses if the market goes up.
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
Fig. 2.4. Selling a call.
3. Strategy of Buying a Put and Hedge and Selling a Put

and Hedge
Table 2.2. Buy a put and hedge and sell a put and hedge.

−33.33% 10 −14.00% 14.00%

−32.33% 10.15 −13.40% 13.40%
−31.33% 10.3 −12.80% 12.80%
−30.33% 10.45 −12.20% 12.20%
−29.33% 10.6 −11.60% 11.60%
−28.33% 10.75 −11.00% 11.00%
−27.33% 10.9 −10.40% 10.40%
−26.33% 11.05 −9.80% 9.80%
−25.33% 11.2 −9.20% 9.20%
−24.33% 11.35 −8.60% 8.60%
−23.33% 11.5 −8.00% 8.00%
−22.33% 11.65 −7.40% 7.40%
−21.33% 11.8 −6.80% 6.80%
−20.33% 11.95 −6.20% 6.20%
−19.33% 12.1 −5.60% 5.60%
−18.33% 12.25 −5.00% 5.00%
−17.33% 12.4 −4.40% 4.40%
(Continued)

−16.33% 12.55 −3.80% 3.80%

−15.33% 12.7 −3.20% 3.20%
−14.33% 12.85 −2.60% 2.60%
−13.33% 13 −2.00% 2.00%
−12.33% 13.15 −1.40% 1.40%
−11.33% 13.3 −0.80% 0.80%
−10.33% 13.45 −0.20% 0.20%
−9.33% 13.6 0.40% −0.40%
−8.33% 13.75 1.00% −1.00%
−7.33% 13.9 1.60% −1.60%
−6.33% 14.05 2.20% −2.20%
−5.33% 14.2 2.80% −2.80%
−4.33% 14.35 3.40% −3.40%
−3.33% 14.5 4.00% −4.00%
−2.33% 14.65 4.60% −4.60%
−1.33% 14.8 5.20% −5.20%
−0.33% 14.95 5.80% −5.80%
0.67% 15.1 5.73% −5.73%
1.67% 15.25 5.33% −5.33%
2.67% 15.4 4.93% −4.93%
3.67% 15.55 4.53% −4.53%
4.67% 15.7 4.13% −4.13%
5.67% 15.85 3.73% −3.73%
6.67% 16 3.33% −3.33%
7.67% 16.15 2.93% −2.93%
8.67% 16.3 2.53% −2.53%
9.67% 16.45 2.13% −2.13%
10.67% 16.6 1.73% −1.73%
11.67% 16.75 1.33% −1.33%
12.67% 16.9 0.93% −0.93%
13.67% 17.05 0.53% −0.53%
14.67% 17.2 0.13% −0.13%
15.67% 17.35 −0.27% 0.27%
16.67% 17.5 −0.67% 0.67%
17.67% 17.65 −1.07% 1.07%
18.67% 17.8 −1.47% 1.47%
19.67% 17.95 −1.87% 1.87%
20.67% 18.1 −2.27% 2.27%
21.67% 18.25 −2.67% 2.67%
22.67% 18.4 −3.07% 3.07%
23.67% 18.55 −3.47% 3.47%
24.67% 18.7 −3.87% 3.87%
25.67% 18.85 −4.27% 4.27%
26.67% 19 −4.67% 4.67%
(Continued)

27.67% 19.15 −5.07% 5.07%

28.67% 19.3 −5.47% 5.47%
29.67% 19.45 −5.87% 5.87%
30.67% 19.6 −6.27% 6.27%
31.67% 19.75 −6.67% 6.67%
32.67% 19.9 −7.07% 7.07%
33.67% 20.05 −7.47% 7.47%
34.67% 20.2 −7.87% 7.87%
35.67% 20.35 −8.27% 8.27%
36.67% 20.5 −8.67% 8.67%
37.67% 20.65 −9.07% 9.07%
38.67% 20.8 −9.47% 9.47%
39.67% 20.95 −9.87% 9.87%
40.67% 21.1 −10.27% 10.27%
41.67% 21.25 −10.67% 10.67%
42.67% 21.4 −11.07% 11.07%
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.
P&L , sell put, sell 0.5 Stock -p-0.5S
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.
3.2. Strategy of buy put and hedge: buy delta units of the
underlying
Buy put, buy 0.5S p=0.5s
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.
buy call sell put and

Stock price at maturity buy call, sell put, c−p hedge, c − p = stocks
−33.33% 10 −37.33% −5.67%

−32.33% 10.15 −36.33% −5.62%
−31.33% 10.3 −35.33% −5.57%
−30.33% 10.45 −34.33% −5.52%
−29.33% 10.6 −33.33% −5.47%
−28.33% 10.75 −32.33% −5.42%
−27.33% 10.9 −31.33% −5.37%
−26.33% 11.05 −30.33% −5.32%
−25.33% 11.2 −29.33% −5.27%
−24.33% 11.35 −28.33% −5.22%
−23.33% 11.5 −27.33% −5.17%
−22.33% 11.65 −26.33% −5.12%
−21.33% 11.8 −25.33% −5.07%
−20.33% 11.95 −24.33% −5.02%
(Continued)

−19.33% 12.1 −23.33% −4.97%

−18.33% 12.25 −22.33% −4.92%
−17.33% 12.4 −21.33% −4.87%
−16.33% 12.55 −20.33% −4.82%
−15.33% 12.7 −19.33% −4.77%
−14.33% 12.85 −18.33% −4.72%
−13.33% 13 −17.33% −4.67%
−12.33% 13.15 −16.33% −4.62%
−11.33% 13.3 −15.33% −4.57%
−10.33% 13.45 −14.33% −4.52%
−9.33% 13.6 −13.33% −4.47%
−8.33% 13.75 −12.33% −4.42%
−7.33% 13.9 −11.33% −4.37%
−6.33% 14.05 −10.33% −4.32%
−5.33% 14.2 −9.33% −4.27%
−4.33% 14.35 −8.33% −4.22%
−3.33% 14.5 −7.33% −4.17%
−2.33% 14.65 −6.33% −4.12%
−1.33% 14.8 −5.33% −4.07%
−0.33% 14.95 −4.33% −4.02%
0.67% 15.1 −3.33% −3.97%
1.67% 15.25 −2.33% −3.92%
2.67% 15.4 −1.33% −3.87%
3.67% 15.55 −0.33% −3.82%
4.67% 15.7 0.67% −3.77%
5.67% 15.85 1.67% −3.72%
6.67% 16 2.67% −3.67%
7.67% 16.15 3.67% −3.62%
8.67% 16.3 4.67% −3.57%
9.67% 16.45 5.67% −3.52%
10.67% 16.6 6.67% −3.47%
11.67% 16.75 7.67% −3.42%
12.67% 16.9 8.67% −3.37%
13.67% 17.05 9.67% −3.32%
14.67% 17.2 10.67% −3.27%
15.67% 17.35 11.67% −3.22%
16.67% 17.5 12.67% −3.17%
17.67% 17.65 13.67% −3.12%
18.67% 17.8 14.67% −3.07%
19.67% 17.95 15.67% −3.02%
20.67% 18.1 16.67% −2.97%
21.67% 18.25 17.67% −2.92%
22.67% 18.4 18.67% −2.87%
23.67% 18.55 19.67% −2.82%
24.67% 18.7 20.67% −2.77%
(Continued)

25.67% 18.85 21.67% −2.72%

26.67% 19 22.67% −2.67%
27.67% 19.15 23.67% −2.62%
28.67% 19.3 24.67% −2.57%
29.67% 19.45 25.67% −2.52%
30.67% 19.6 26.67% −2.47%
31.67% 19.75 27.67% −2.42%
32.67% 19.9 28.67% −2.37%
33.67% 20.05 29.67% −2.32%
34.67% 20.2 30.67% −2.27%
35.67% 20.35 31.67% −2.22%
36.67% 20.5 32.67% −2.17%
37.67% 20.65 33.67% −2.12%
38.67% 20.8 34.67% −2.07%
39.67% 20.95 35.67% −2.02%
40.67% 21.1 36.67% −1.97%
41.67% 21.25 37.67% −1.92%
42.67% 21.4 38.67% −1.87%
5. Strategy of Buy Call, Sell Put: Equivalent to Holding

the Underlying
buy call, sell put ,c-p
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%
Fig. 2.7. Buy a call and sell a put.

6. Strategy of Buy Call, Sell Put and Hedge: Reduces

Profits and Reduces Losses
buy call sell put and hedge c-p=stcocs
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%

-3.00%
hedge c-p=stcocs
-4.00%
-5.00%
-6.00%
Fig. 2.8. Buy a call and sell a put and hedge.
Table 2.4. Sell call, buy put and hedge.
Stock price at maturity Sell call, buy put, −c + p Sell call, buy put and hedge
−33.33% 10 37.33% 5.67%

−32.33% 10.15 36.33% 5.62%
−31.33% 10.3 35.33% 5.57%
−30.33% 10.45 34.33% 5.52%
−29.33% 10.6 33.33% 5.47%
−28.33% 10.75 32.33% 5.42%
−27.33% 10.9 31.33% 5.37%
−26.33% 11.05 30.33% 5.32%
−25.33% 11.2 29.33% 5.27%
−24.33% 11.35 28.33% 5.22%
−23.33% 11.5 27.33% 5.17%
−22.33% 11.65 26.33% 5.12%
−21.33% 11.8 25.33% 5.07%
−20.33% 11.95 24.33% 5.02%
−19.33% 12.1 23.33% 4.97%
−18.33% 12.25 22.33% 4.92%
−17.33% 12.4 21.33% 4.87%
−16.33% 12.55 20.33% 4.82%
−15.33% 12.7 19.33% 4.77%
−14.33% 12.85 18.33% 4.72%
−13.33% 13 17.33% 4.67%
−12.33% 13.15 16.33% 4.62%
(Continued)
−11.33% 13.3 15.33% 4.57%

−10.33% 13.45 14.33% 4.52%
−9.33% 13.6 13.33% 4.47%
−8.33% 13.75 12.33% 4.42%
−7.33% 13.9 11.33% 4.37%
−6.33% 14.05 10.33% 4.32%
−5.33% 14.2 9.33% 4.27%
−4.33% 14.35 8.33% 4.22%
−3.33% 14.5 7.33% 4.17%
−2.33% 14.65 6.33% 4.12%
−1.33% 14.8 5.33% 4.07%
−0.33% 14.95 4.33% 4.02%
0.67% 15.1 3.33% 3.97%
1.67% 15.25 2.33% 3.92%
2.67% 15.4 1.33% 3.87%
3.67% 15.55 0.33% 3.82%
4.67% 15.7 −0.67% 3.77%
5.67% 15.85 −1.67% 3.72%
6.67% 16 −2.67% 3.67%
7.67% 16.15 −3.67% 3.62%
8.67% 16.3 −4.67% 3.57%
9.67% 16.45 −5.67% 3.52%
10.67% 16.6 −6.67% 3.47%
11.67% 16.75 −7.67% 3.42%
12.67% 16.9 −8.67% 3.37%
13.67% 17.05 −9.67% 3.32%
14.67% 17.2 −10.67% 3.27%
15.67% 17.35 −11.67% 3.22%
16.67% 17.5 −12.67% 3.17%
17.67% 17.65 −13.67% 3.12%
18.67% 17.8 −14.67% 3.07%
19.67% 17.95 −15.67% 3.02%
20.67% 18.1 −16.67% 2.97%
21.67% 18.25 −17.67% 2.92%
22.67% 18.4 −18.67% 2.87%
23.67% 18.55 −19.67% 2.82%
24.67% 18.7 −20.67% 2.77%
25.67% 18.85 −21.67% 2.72%
26.67% 19 −22.67% 2.67%
27.67% 19.15 −23.67% 2.62%
28.67% 19.3 −24.67% 2.57%
29.67% 19.45 −25.67% 2.52%
30.67% 19.6 −26.67% 2.47%
31.67% 19.75 −27.67% 2.42%
32.67% 19.9 −28.67% 2.37%
33.67% 20.05 −29.67% 2.32%
34.67% 20.2 −30.67% 2.27%
(Continued)
35.67% 20.35 −31.67% 2.22%

36.67% 20.5 −32.67% 2.17%
37.67% 20.65 −33.67% 2.12%
38.67% 20.8 −34.67% 2.07%
39.67% 20.95 −35.67% 2.02%
40.67% 21.1 −36.67% 1.97%
41.67% 21.25 −37.67% 1.92%
42.67% 21.4 −38.67% 1.87%
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.
Chapter 3
TRADING OPTIONS AND THEIR

UNDERLYING ASSET: RISK MANAGEMENT
IN DISCRETE TIME
Chapter Outline
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
which apply to options on a spot asset, to options on a futures contract

and in general to options with any particular payoff. These strategies are
illustrated with respect to the diagram payoffs or the expected return and
risk trade-off of standard options. The understanding of option strategies
is based on the use of synthetic positions. This chapter provides several
illustrations of the main strategies provided in the previous chapter. In
particular, we develop the basic strategies and synthetic positions: long
or short the underlying asset, long a call, long a put, and short a put.
Then, we present some combinations and more elaborated strategies as:
long a straddle, short a straddle, long or short a strangle, long a tunnel,
short a tunnel, long a call or put bull spread, long a call or a put bear
spread, long or short a butterfly, long or short a condor, etc. Finally, we
show how traders and market participants use option pricing models and
estimate model parameters. We introduce the concepts of Greek letters
or the sensitivities of the option price or position with respect to some
parameters.
3.1. Basic Strategies and Synthetic Positions

This section develops the main option strategies and synthetic positions.
3.1.1. Options and synthetic positions

Synthetic positions can be constructed by options on spot assets, options
on futures contracts, and their underlying assets. If we use the symbol 0
to denote a horizontal line, the symbol −1 for the slope under 0 and the
symbol 1 for the slope above 0, then it is possible to represent the diagram
pay offs of a long call by (0, 1), a short call by (0, −1), a long put by (−1, 0),
and a short put by (1, 0).
Adopting this notation for the basic option pay offs, it is possible to
construct all the synthetic positions as well as most elaborated diagram
strategies using this representation. We denote by:
S, (F ): price of the underlying asset, which may be a spot asset, S (or a
futures contract F );
K: strike price;
C: call price and
P : put price.
We use the following symbols: 0, : 1, : −1.

Trading Options and Their Underlying Asset: Risk Management in Discrete Time 143
The results of the basic strategies can be represented as follows:

Long a call: (0, 1): , Short a call: (0, −1): , Long a put: (−1, 0):
, Short a put: (1, 0): .
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.
(1, 1) = (0, 1) + (1, 0)
Short a synthetic underlying asset = short a call + long a put.
(−1, −1) = (0, −1) + (−1, 0)
Long a synthetic call = long the underlying asset + long a put.
(0, 1) = (1, 1) + (−1, 0)
Short a synthetic call = short the underlying asset + short a put.
(0, −1) = (−1, −1) + (1.0)
Long a synthetic put = short the underlying asset + long a call.
(−1, 0) = (−1, 1) + (0, 1)
Short a synthetic put = long the underlying asset + short a call.
(1, 0) = (1, 1) + (0, −1)
The knowledge of synthetic positions is necessary for market partici-

pants since it allows the implementation of hedged positions. Hedged
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.
3.1.2. Long or short the underlying asset

The risk-return profile for a position which is long or short the underlying
asset (for example a futures contract) shows unlimited profit or loss. If we
put on a horizontal line the underlying asset price and on a vertical line
the profit or loss, the payoff to a long or a short position in the underlying
asset can be easily represented. If the asset price rises or falls by one point,
the profit or loss will be of the same amount.
3.1.3. Long a call

Expectations: The trader expects a rising market and (or) a high volatility
until the maturity date.
Definition: Buy a call, c with a strike price K.
Specific features: The potential gain is not limited but the potential loss
is limited to the option premium.
Buying the call at 1.9, reveals the risk-reward profile as shown in

Fig. 3.1 at expiration.
If S = 111.9 at maturity; (110+1.9), the profit is zero. This is the break-
even point of the position. Beyond this level, the profit is not limited. The
maximum loss or performance corresponds to 1.9 or 100% (Table 3.1). In
Fig. 3.1, the break-even point is given by the sum of the strike price and
the option premium.
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.
Table 3.1. Long a call: S = 102, r = 5%, volatility = 20%,

and T = 100 days.
Type Point Value
Break-even point A K+c

Maximal loss c
Maximal gain Not limited, if S > K
3.1.4. Short call

Expectations: The trader expects a falling market and (or) a lower
volatility until the maturity date.
Definition: Sell a call, c with a strike price K.
Specific features: The potential gain is limited to the perceived premium

and the potential loss is not limited.
The risk-reward trade-off is inverted when selling calls (Table 3.2). The
results of the strategy are given in Fig. 3.2.
Table 3.2. Short a call: S = 102, r = 5%,

volatility = 20%, T = 100 days.
Type Point Value
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
Fig. 3.2. Short a call.
Short a call
Strike price = 110

Premium = 1, 9
Profit for a multiple of 10: 19
Break-even point = 111.9
S Variation (%) Call Performance (%)
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
In Fig. 3.2, the break-even point is given by the sum of the strike price
and the option premium.
3.1.5. Long a put

Expectations: The trader expects a falling market and (or) a higher
volatility until the maturity date.
Definition: Buy a put p with a strike price K
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
Fig. 3.3. Long a put.

Long a put
Strike price = 100

Prime = 2, 7
Profit for a multiple of 10:27
Break-even point = 97, 28
S Variation (%) Put Performance (%)
60 −41 37, 3 1371

70 −31 27, 3 1003
80 −22 17, 3 635
90 −12 7, 3 268
100 −2 −2, 7 −100
110 8 −2, 7 −100
120 18 −2, 7 −100
130 27 −2, 7 −100
140 37 −2, 7 −100
Table 3.3. Long a Put: S = 102, r = 5%,

volatility = 20%, and T = 100 days.
Type Point Value
Break-even point A K −p
Maximal loss Premium
Maximal gain Not limited
3.1.6. Short a put

Expectations: The trader expects a stable and (or) a rising market.
Definition: Sell a put p with a strike price K.
Specific features: The potential gain is limited to the option premium

and the potential loss is unlimited (Fig. 3.4).
2,7 2,7 2,7 2,7 2,7
-2,3
-5
profit
-7,3
-10
-12,3
-15
-17,3
-20
80 85 90 95 100 105 110 115 120
S
Fig. 3.4. Short a put.
Short a put
Strike price 100

Premium 2.7
Profit 27
Break-even point 97.28
S Variation (%) Put Performance (%)
80 −22 −17.3 −635

85 −17 −12.3 −451
90 −12 −7.3 −268
95 −7 −2.3 −84
100 −2 2.7 100
105 3 2.7 100
110 8 2.7 100
115 13 2.7 100
120 18 2.7 100
Table 3.4. Short a put: S = 102, r = 5%,

volatility = 20%, and T = 100 days.
Type Point Value
Break-even point A K −p
Maximal loss Not limited
Maximal Gain Premium
Figure 3.3 represents in a certain way the opposite of the risk-reward

profile in Fig. 3.2. The profit is limited when the underlying asset price
increases and the risk is unlimited when the underlying asset price is
decreasing (Table 3.4).
3.2. Combined Strategies

This section illustrates several combined strategies involving call and put
options.
3.2.1. Long a straddle

This strategy is perfect during the 2008 financial crisis.
Expectations: The trader expects a high volatility until the maturity date.
Definition: Buy a call, c and simultaneously buy a put, p on the same

underlying (for the same maturity date and the same strike price).
Specific features: The initial investment is important since the investor

buys simultaneously the call and the put.
• The loss is limited to the initial cost (c and p).

• The maximum potential gain is not limited when the market goes up
or down.
Buying a straddle needs a simultaneous purchase of call and a put with

the same strike price for the same maturity (Fig. 3.5). When the put is
worthless, the call is deep-in-the money. When the call is worthless, the put
is in-the-money (Table 3.5).
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
Fig. 3.5. Buying a Straddle.
Long a call Long a put Strategy
Strike price 100 100

Prime 4.5 1.8 6.3
Cost 45 18 63
Break-even point 104.50 98.18 5
S Variation (%) Call Put Straddle Performance (%)
80 −22 −4.5 18.2 13.7 216

85 −17 −4.5 13.2 8.7 137
90 −12 −4.5 8.2 3.7 58
95 −7 −4.5 3.2 −1.3 −21
100 −2 −4.5 −1.8 −6.3 −100
105 3 0.5 −1.8 −1.3 −21
110 8 5.5 −1.8 3.7 58
115 13 10.5 −1.8 8.7 137
120 18 15.5 −1.8 13.7 216
Table 3.5. Long a straddle.
Type Point Value
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
Notes: The strike price is chosen according to the trader expectations

about the future market direction.
Simulation: Underlying asset S = 102, Interest rate r (%) = 5%, Volatility

(%) = 20%, and Maturity (in days) = 50.
3.2.2. Short a straddle

Expectations: The trader expects a low volatility until the maturity date.
Definition:
• Sell a call c and simultaneously.

• Sell a put, p on the same underlying for the same maturity date and the
same strike price.
Specific features:
• The initial revenue is limited to the option premiums.

• The loss is not limited when the market goes up or down.
• The maximum potential gain is limited to the initial premium
(c and p).
When the underlying asset price is expected to be in a specified interval

at maturity, the trader can sell simultaneously a call and a put. The profit
is limited to the premium received and the risk may be unlimited (Fig. 3.6).
If the underlying asset is not expected to move much either side, the
investor can sell the put and the call. The maximum profit at expiration is
obtained when S is in a given interval (Table 3.6).
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
Fig. 3.6. Short a straddle.
Table 3.6. Shorting a straddle.
S Variation (%) Call Put Straddle Performance (%)
80 −22 4.5 −18.2 −13.7 −216

85 −17 4.5 −13.2 −8.7 −137
90 −12 4.5 −8.2 −3.7 −58
95 −7 4.5 −3.2 1.3 21
100 −2 4.5 1.8 6.3 100
105 3 −0.5 1.8 1.3 21
110 8 −5.5 1.8 −3.7 −58
115 13 −10.5 1.8 −8.7 −137
120 18 −15.5 1.8 −13.7 −216
3.2.3. Long a strangle

Expectations: The trader expects a high volatility during the options’ life.
Definition: Buy a call with a strike price Kc and buy a put with a strike
price Kp where the Kp < Kc .
Specific features
• This strategy costs less than the straddle.

• The maximum loss is limited to the initial cost of (c + p).
• The net result is a profit only when the market movement is important
(Fig. 3.7). In this example, the market must increase by 5%, (107,
49 − 102)/102 or decrease by 9%, (92, 41 − 102)/102 (Table 3.7).
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
Fig. 3.7. Profit (per unit) of a long strangle strategy.
Table 3.7. Long a strangle.
Type Point Value
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
Table 3.8. Profit (per unit) of a long strangle strategy.
S Variation (%) Call Put Strangle Performance (%)
85 −17 −2 9.5 7.4 287

90 −12 −2 4.5 2.4 93
95 −7 −2 −0.5 −2.6 −100
100 −2 −2 −0.5 −2.6 −100
105 3 −2 −0.5 −2.6 −100
110 8 3 −0.5 2.4 93
115 13 8 −0.5 7.4 287
120 18 13 −0.5 12.4 480
125 23 18 −0.5 17.4 674
The two break-even points are computed as follows:
• 105 + (2.04 + 0.55) = 107.59 or a variation of 5.38%.

• 95 − (2.04 + 0.55) = 92.41 or a variation of −9.40%.
If the underlying asset price is between the two strike prices at

expiration, the maximum loss is reduced to the initial cost of 25.8. The
net result is a loss, if the underlying asset price is between the two break-
even points, 92.41 and 107.59. This loss is less than the initial cost. However,
if the underlying asset price is above the break-even points, on either side,
the trader benefits from the leverage effect (Table 3.8). For example, if
the underlying asset price is 90 at expiration, or a variation of 12%, the
net result is 93%. If the underlying asset goes up by 18% to attain a level
of 120, the net profit of 12.4, compared to 2.58, represents a performance
of 480%.
Simulation: The parameters used in the simulation are: S = 102, r = 5%,

Volatility = 20%, and Maturity = 50 days.
Long a call Long a put
Strike price 105 95

Premium 2.04 0.55
Cost for 10 20.4 5.5
Break-even point 107.59 92.41
3.2.4. Short a strangle

Expectations: The trader expects a low volatility during the options’ life.
Definition: Sell a call with a strike price Kc and Sell a put with a strike
price Kp where the Kp < Kc (Fig. 3.8).
Specific features:
• The maximum gain is limited to the initial premium of (c + p).
• The strategy can show a loss (Table 3.9).
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
Fig. 3.8. Short a strangle.
Table 3.9. Short a strangle.
S Variation (%) Call Put Strangle Performance (%)
85 −17 2 −9.5 −7.4 −287

90 −12 2 −4.5 −2.4 −93
95 −7 2 0.5 2.6 100
100 −2 2 0.5 2.6 100
105 3 2 0.5 2.6 100
110 8 −3 0.5 −2.4 −93
115 13 −8 0.5 −7.4 −287
120 18 −13 0.5 −12.4 −480
125 23 −18 0.5 −17.4 −674
The reader can make the specific comments by comparing this strategy
with the long strangle.
3.2.5. Long a tunnel

Expectations: The trader expects a high volatility during the options’ life.
Definition: Buy an out-of-the money call and sell out-of-the money put
as in Table 3.10 (Fig. 3.9).
Table 3.10. Long a tunnel.
10 Long a call Short a put

Premium 22.3 15.3 7
Cost for 10 223 153 −70
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).
Call out- Put out-

Underlying Variation of-the of-the Performance
asset S (%) money money Tunnel (%)
530 −5 −22.3 −4.7 −27 −386

540 −4 −22.3 5.3 −17 −243
550 −2 −22.3 15.3 −7 −100
560 0 −22.3 15.3 −7 −100
570 2 −22.3 15.3 −7 −100
580 4 −12.3 15.3 3 43
590 5 −2.3 15.3 13 186
600 7 7.7 15.3 23 329
610 9 17.7 15.3 33 471
Simulation:
3.2.6. Short a tunnel

This is the opposite of the previous strategy (Fig. 3.10) (Table 3.11).
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).
Underlying Variation Call out Put out Tunnel Performance

asset (%) (%)
530 −5 22.3 4.7 27 386

540 −4 22.3 −5.3 17 243
550 −2 22.3 −15.3 7 100
560 0 22.3 −15.3 7 100
570 2 22.3 −15.3 7 100
580 4 12.3 −15.3 −3 −43
590 5 2.3 −15.3 −13 −186
600 7 −7.7 −15.3 −23 −329
610 9 −17.7 −15.3 −33 −471
Table 3.11.
Q = 10 Short a call Long a put

Premium 22.3 15.3 7.0
Cost 223 153 −70
3.2.7. Long a call bull spread

A strategy can be implemented by buying a call with a lower strike price
and selling a call with a higher strike price (Fig. 3.11).
If the underlying asset price is below the lower strike price at expiration,
the maximum loss is limited to the difference between the two option
premiums.
If the underlying asset price is above the higher strike price at
expiration, the lower strike price call is worth the intrinsic value. This
strategy shows a limited profit (a loss) (Table 3.12).
Long a Bull Spread with Calls for the following parameters: S = 102,
r = 5%, volatility = 20%, T = 100 days.
3.2.8. Long a put bull spread

Expectations: Buying a put spread is equivalent to buying the higher
strike price put and selling the lower strike price put (Fig. 3.12).
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
Fig. 3.11. Buying a bull spread with calls.
Table 3.12. Bull spread with calls: S = 102, r = 5%, volatility = 20%,
and T = 100 days.
S Variation (%) Call in Call out Spread Performance (%)
490 −14 −19.9 11 −8.9 −100

510 −11 −19.9 11 −8.9 −100
530 −7 −19.9 11 −8.9 −100
550 −4 −19.9 11 −8.9 −100
570 0 −19.9 11 −8.9 −100
590 3 0.1 11 11.1 125
610 7 20.1 −9 11.1 125
630 10 40.1 −29 11.1 125
650 14 60.1 −49 11.1 125
If the underlying asset is around the lower strike price at maturity,

the higher strike price put is worth the intrinsic value and the lower
strike price is worthless. The maximum profit is given by the difference
between the two option premiums. The strategy is done with a debit
(Table 3.13).
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
Fig. 3.12. Buying a bull spread with puts.
Table 3.13. Buying a bull spread with puts.
S Variation (%) Put out Put in Spread Performance (%)
490 −14 66 −75 −9 82

510 −11 46 −55 −9 82
530 −7 26 −35 −9 82
550 −4 6 −15 −9 82
570 0 −14 5 −9 82
590 3 −14 25 11 −100
610 7 −14 25 11 −100
630 10 −14 25 11 −100
650 14 −14 25 11 −100
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.
3.2.9. Long a call bear spread

The reader can refer to the previous chapter for more details and make the
necessary comments (Fig. 3.13) (Table 3.14).
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
Fig. 3.13. Selling a call bear spread.
Table 3.14. Selling a call bear spread.
S Variation (%) Call in Call out Spread Performance (%)
490 −14 18.8 −10.3 8.5 100

510 −11 18.8 −10.3 8.5 100
530 −7 18.8 −10.3 8.5 100
550 −4 18.8 −10.3 8.5 100
570 0 18.8 −10.3 8.5 100
590 4 −1.2 −10.3 −11.5 −134
610 7 −21.2 9.7 −11.5 −134
630 11 −41.2 29.7 −11.5 −134
650 14 −61.2 49.7 −11.5 −134
3.2.10. Selling a put bear spread

Refer Fig. 3.14 and Table 3.15. This is best understood from Fig. 3.14 and
Table 3.15. (see below).
3.2.11. Long a butterfly

Expectations: The reader can refer to the previous chapter for more
details and make the necessary comments (Fig. 3.15) (Table 3.16).
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
Fig. 3.14. Selling a put bear spread.
Table 3.15. Selling a put bear spread.
S Variation (%) Put out Put in Spread Performance (%)
490 −14 −65.1 73.8 8.7 −77

510 −11 −45.1 53.8 8.7 −77
530 −7 −25.1 33.8 8.7 −77
550 −4 −5.1 13.8 8.7 −77
570 0 14.9 −6.2 8.7 −77
590 4 14.9 −26.2 −11.3 100
610 7 14.9 −26.2 −11.3 100
630 11 14.9 −26.2 −11.3 100
650 14 14.9 −26.2 −11.3 100
3.2.12. Short a butterfly

details and make the necessary comments (Table 3.17).
3.2.13. Long a condor

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
Fig. 3.15. Long a butterfly.
Table 3.16. Long a butterfly.
S Variation (%) Call out Call at Call in Butterfly Performance (%)
80 −22 −12.7 9 −0.7 −4.5 −20

85 −17 −12.7 9 −0.7 −4.5 −20
90 −12 −12.7 9 −0.7 −4.5 −20
95 −7 −7.7 9 −0.7 0.5 2
100 −2 −2.7 9 −0.7 5.5 25
105 3 2.3 −1 −0.7 0.5 2
110 8 7.3 −11 −0.7 −4.5 −20
115 13 12.3 −21 4.3 −4.5 −20
120 18 17.3 −31 9.3 −4.5 −20
3.2.14. Short a condor

Table 3.17. Short a butterfly.
S Variation (%) Call out Call at Call in Butterfly Performance (%)
80 −22 12.7 −9 0.7 4.5 20

85 −17 12.7 −9 0.7 4.5 20
90 −12 12.7 −9 0.7 4.5 20
95 −7 7.7 −9 0.7 −0.5 −2
100 −2 2.7 −9 0.7 −5.5 −25
105 3 −2.3 1 0.7 −0.5 −2
110 8 −7.3 11 0.7 4.5 20
115 13 −12.3 21 −4.3 4.5 20
120 18 −17.3 31 −9.3 4.5 20
40
Call
Call At
30
Call At
Call In
20 Condor
10
3.9
profit
-10 -6.1 -6.1
-20
-30
-40
60 70 80 90 100 110 120 130 140
S
Fig. 3.16. Long a condor.
Table 3.18. Long a condor.
Variation Performance
S (%) Call out Call at Call at Call in Condor (%)
60 −43 −16.5 8.2 2.9 −0.7 −6.1 −22

70 −33 −16.5 8.2 2.9 −0.7 −6.1 −22
80 −24 −16.5 8.2 2.9 −0.7 −6.1 −22
90 −14 −16.5 8.2 2.9 −0.7 −6.1 −22
100 −5 −6.5 8.2 2.9 −0.7 3.9 14
110 5 3.5 −1.8 2.9 −0.7 3.9 14
120 14 13.5 −11.8 −7.1 −0.7 −6.1 −22
130 24 23.5 −21.8 −17.1 9.3 −6.1 −22
140 33 33.5 −31.8 −27.1 19.3 −6.1 −22
40
30
20
10 6.1 6.1
profit
-10 -3.9
-20 Call Out

Call At
Call At
-30
Call In
Condor
-40
60 70 80 90 100 110 120 130 140
S
Fig. 3.17. Short a condor.
Table 3.19. Short a condor.
Variation Performance
S (%) Call out Call at Call at Call in Condor (%)
60 −43 16.5 −8.2 −2.9 0.7 6.1 22

70 −33 16.5 −8.2 −2.9 0.7 6.1 22
80 −24 16.5 −8.2 −2.9 0.7 6.1 22
90 −14 16.5 −8.2 −2.9 0.7 6.1 22
100 −5 6.5 −8.2 −2.9 0.7 −3.9 −14
110 5 −3.5 1.8 −2.9 0.7 −3.9 −14
120 14 −13.5 11.8 7.1 0.7 6.1 22
130 24 −23.5 21.8 17.1 −9.3 6.1 22
140 33 −33.5 31.8 27.1 −19.3 6.1 22
3.3. How Traders Use Option Pricing Models:

Parameter Estimation
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 the
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.
3.3.1. Estimation of model parameters

We explain briefly how the model parameters are obtained when trading
options. The interest rates must be estimated during the option’s life.
Volatility 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
also be calculated using the observed option prices in the market place and
an option pricing model. This refers to the implicit volatility. The effect of
the parameters on the option value can be appreciated using Table 3.20.
There are at least two ways to estimate the volatility: the historical
volatility and the implied volatility.
3.3.1.1. Historical volatility

The return on an asset can be computed using three measures for the
variations of its price.
The first direct measure of return is given by the difference between
the asset prices at two dates as:
(1)
Ri,t = Si,t+1 − Si,t
The second measure allows the computation of a compound return as:

(2)
Ri,t = Log(Si,t+1 ) − Log(Si,t )
The third method uses the relative variations in the underlying asset
prices.
(3) Si,t+1 − Si,t

Ri,t =
Si,t
Table 3.20. The effect of the parameters on

option prices.
Parameters Call Put
Underlying asset, S + −
Strike price, K − +
Dividends − +
Interest rates + −
Maturity + +
Volatility + +
Historical volatility can be estimated using historical returns as follows:

Nj
τ =1 (Rt−τ − R̄)
2
σ̂H
2
(t, Nj ) =
Nj − 1
where the mean return is given by,

Nj
1
R̄ = Rt−τ
Nj τ =1
where Nj corresponds to the number of observations. In general, historical

volatilities are computed using a period of nearly 20 days for short-term
options and a period of 250 trading days for long-term options.
Example: Consider the computation of the historical volatility for the

CAC40 (Table 3.21). Computations are done on 11 July 2003. The volatility
is computed for a 20-day period using closing prices for the CAC40 over the
past 21 days (between lines −1 and −21). The logarithmic formula giving
the compound return is used in the computation of the return. For example,
the relative variation of prices on 7 July 2003 is computed using the prices
of the 4 and 7 July as: 0.45% = Ln(2947,66) − Ln(2934,48) where Ln(.)
corresponds to the Neperian logarithm.
Other methods can be used in the computation of historical volatilities.

These methods use opening prices, closing prices, opening and closing
prices, high and low prices during a trading period, etc. The estimated
volatility using closing prices is given by the following formula:
(f ) (f )
σ̂02 (t) = (St − St−1 )2
where (f ) refers to closing prices.

The estimated volatility using opening prices is given by:
(o) (o)
σ̂02 (t) = (St+1 − St )2
where (o) refers to opening prices.

The estimated volatility using opening and closing prices is given by:
(o) (f )
(St − St−1 )2 (f ) (o)
(S − St )2
σ̂12 (t) ≡ + t 0<f <1
2f 2(1 − f )
This estimated volatility is two times more efficient than σ̂02 .

Table 3.21. Computation of the historical volatility for the CAC40 index.
No Date CAC40 Rt = Ln(St ) − Ln(St−1 ) (Rt − R)2
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.
3.3.1.2. Implied volatilities and option pricing models

Using an option pricing model, (for example the Whaley, 1987; Black and
Scholes, 1973 model; Merton, 1973 or Barone-Adesi), the call price can
be computed using the different parameters: the underlying asset price S,
the strike price K, the interest rate r, the time to maturity T and the
volatility σ. Using the market price, and the inputs S, K, r, and T , the
following relationship is used to compute the implied volatility:
f (σ) = Call market price.
The implied volatility is estimated by,
σ̂ = f −1 [f (σ)] = f −1 (Call)
Several numerical approximations can be used in the computation of

the solution to the function F (x) = 0. It is possible to aggregate different
implicit volatilities each day to get a significant number.
Hence, we can compute an arithmetic mean of the implied standard
deviations (AISD) as:
N (t)
(1) 1
σAISD (t) = σIMP (t, j)
N (t) j=1
We can compute a weighted average of implied standard deviations (WISD).
3.3.2. Trading and Greek letters

It is important for traders to follow the changes in their portfolios of options.
It is not sufficient to buy and sell options because the risk inherent to these
operations must be managed. The management is facilitated by the use of
an option pricing model and some of its partial derivatives, often known as
Greek-letter risk measures. The most widely used measures are known as
the delta, gamma, vega, and theta.
The delta shows the absolute change in the option price with respect
to a small variation in the underlying asset price. 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. Figure 3.18 shows the dynamics of the delta for the call and
the put as a function of the underlying asset price.
The gamma gives the change in the delta, or in the hedge ratio as the
underlying asset price changes. The gamma measures the change in delta,
1.0
0.8 Delta (Call)

0.6
0.4
0.2
Delta
0.0
-0.2 Delta (Put)

-0.4
-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
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
Vega (Call or Put)

5.0
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.
2.0
1.0
0.0
-1.0 Theta θ(Put)
-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.
importance of the sensitivity parameters in the management of portfolios

of derivatives.
Figure 3.19 shows the shape of the option’s gamma as a function of the
underlying asset price. The gamma is the same for call and put options.
Figure 3.20 shows the shape of the option’s vega as a function of the
Figure 3.21 shows the shape of the option’s theta as a function of the
Figure 3.22 shows the shape of the option’s theta as a function of the
number of days until maturity.
It is important for any head of trading to compare the notional amount
of the position with respect to the capital of the institution.
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
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.
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.
1. Stock lending case: Entering a total return swap, TRS, between

IC and DC
Definition
A total return swap (TRS) is a bilateral financial transaction where the
counterparties swap the total return of a single asset in exchange for
periodic cash flows, a floating rate as Libor + or − a basis point spread
and a guarantee against capital losses.
TRS is similar to a plain vanilla swap, except the deal is structured
such that the total return (capital appreciation/depreciation) is exchanged,
rather than just the cash flows. A key feature of a TRS is that the parties
do not transfer actual ownership of the assets. This allows greater
flexibility and reduced up front capital to execute a trade.
IC is the Total return Payer:
– Own reference asset

– Has lower cost financing
– Pays total return of assets
– Receives Libor + or − spread
– Receives payment to offset any capital losses
– Transfers away asset return risk
DC is the Total Return Receiver (TRS):
– Does not own reference asset

– Has higher cost of financing
– receives total return of assets
– Pays Libor + or − spread
– Pays for any capital losses
– Takes on asset return risk
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.
At each payment date, payments received by:
DC: the Total return Receiver:

– Price appreciation since the last fixing
At each payment date, payments received by:

IC: the Total return Payer:
– Price depreciation since the last fixing
– Libor + or − spread
Benefits
– the greatest benefit is leverage and flexibility: parties do not transfer
actual ownership of assets.
– this does not induce transaction costs
Risks of total return swaps, TRS:
– 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.
2. Enter into a Total Return Swap (TRS)

Position of IC: sells the stock (as stock lending, “lease”)
Position of DC: buy the stock (synthetically)
If the stock is up:

DC receives price appreciation and pays Libor + or − %,
If the stock is down:

DC pays price depreciation and pays Libor + or − %,
DC is in the opposite side to IC.

Table 1. Total Return Swap, TRS, for IC.
IC Receives from DC Stock unchanged Stock + 10% Stock − 10%
Initial value of stock 100,000 100,000 100,000

Stocks value at maturity S1 100 000 110 000 90 000
Profit (loss) without the 0 10,000 (10,000)
TRS with DC
Enter TRS: Profit from Libror 0,000 0,000 0,000
(0% × 100, 000)
IC receives from and pays to 0 (10,000) 10,000
DC for the stock
IC profit and loss with DC 0,000 (10,000) 10,000
No obstacle to enter a TRS swap.

This structure allows IC:
– to keep its portfolio

– to enhance returns
3. Position in the stock if IC makes ‘stock lending’ at Libor 0%

Total return swap: Base case: $ 100 M
Conclusion: We can compare:
– Profit (loss) without the swap TRS between DC and IC
– profit and loss with the swap TRS between IC and DC.
It is clear that this is a zero sum game: what IC wins, DC loses and
Vice-versa.
It is ‘neutral’ for IC to enter the TRS with Rate = 0 from a financial
standpoint.
In fact, there is an equal probability that the stock goes up or goes down.
(win or lose 10,000)
But, potential upside is hidden . . .
Position
– 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.
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%
Initial value of stock 100,000 100,000 100,000

Stocks value at maeturity S1 100 000 110 000 90 000
Profit (loss) without 0 10,000 (10,000)
the TRS with DC
Enter TRS with DC: 5,000 5,000 5,000
Profit from Libror
(5% × 100, 000) = R
IC receives from and 0 (10,000) 10,000
pays to DC for the stock
IC profit and loss 5,000 (5,000) 15,000
with DC (opportunity cost)
Capital for transactions: difference to pay or receive at the

maturity date (each quarter) No obstacle to implement the TRS.
4. Position in the stock if IC makes stock lending at Libor: 5% Total

return swap
We can compare the profit and loss for two scenario with and without Libor.
IC wins more than it loses by entering this transaction.
Position for DC:
– similar to a classic position “long the stock synthetically”
– main difference: the contract is for the difference
– pays libor + or − and the difference in stocks losses
– receives capital gains
Capital for transactions: difference to pay or receive at the
maturity date (each quarter) No obstacle.
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.
Table 3.
Stock Return at Libor 0% Return at Libor 5%
10.00 33.33% 38.33%

15.10 −0.67% 4.33%
20.80 −38.67% −33.67%
Table 4. TRS between IC and DC for different Libor Rates: position.

rate
Stock price 0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00%
10.00 33.33% 34.33% 35.33% 36.33% 37.33% 38.33% 39.33% 40.33%
10.15 32.33% 33.33% 34.33% 35.33% 36.33% 37.33% 38.33% 39.33%
10.30 31.33% 32.33% 33.33% 34.33% 35.33% 36.33% 37.33% 38.33%
10.45 30.33% 31.33% 32.33% 33.33% 34.33% 35.33% 36.33% 37.33%
10.60 29.33% 30.33% 31.33% 32.33% 33.33% 34.33% 35.33% 36.33%
10.75 28.33% 29.33% 30.33% 31.33% 32.33% 33.33% 34.33% 35.33%
10.90 27.33% 28.33% 29.33% 30.33% 31.33% 32.33% 33.33% 34.33%
11.05 26.33% 27.33% 28.33% 29.33% 30.33% 31.33% 32.33% 33.33%
11.20 25.33% 26.33% 27.33% 28.33% 29.33% 30.33% 31.33% 32.33%
11.35 24.33% 25.33% 26.33% 27.33% 28.33% 29.33% 30.33% 31.33%
11.50 23.33% 24.33% 25.33% 26.33% 27.33% 28.33% 29.33% 30.33%
11.65 22.33% 23.33% 24.33% 25.33% 26.33% 27.33% 28.33% 29.33%
11.80 21.33% 22.33% 23.33% 24.33% 25.33% 26.33% 27.33% 28.33%
11.95 20.33% 21.33% 22.33% 23.33% 24.33% 25.33% 26.33% 27.33%
12.10 19.33% 20.33% 21.33% 22.33% 23.33% 24.33% 25.33% 26.33%
12.25 18.33% 19.33% 20.33% 21.33% 22.33% 23.33% 24.33% 25.33%
12.40 17.33% 18.33% 19.33% 20.33% 21.33% 22.33% 23.33% 24.33%
12.55 16.33% 17.33% 18.33% 19.33% 20.33% 21.33% 22.33% 23.33%
12.70 15.33% 16.33% 17.33% 18.33% 19.33% 20.33% 21.33% 22.33%
12.85 14.33% 15.33% 16.33% 17.33% 18.33% 19.33% 20.33% 21.33%
13.00 13.33% 14.33% 15.33% 16.33% 17.33% 18.33% 19.33% 20.33%
13.15 12.33% 13.33% 14.33% 15.33% 16.33% 17.33% 18.33% 19.33%
13.30 11.33% 12.33% 13.33% 14.33% 15.33% 16.33% 17.33% 18.33%
13.45 10.33% 11.33% 12.33% 13.33% 14.33% 15.33% 16.33% 17.33%
13.60 9.33% 10.33% 11.33% 12.33% 13.33% 14.33% 15.33% 16.33%
13.75 8.33% 9.33% 10.33% 11.33% 12.33% 13.33% 14.33% 15.33%
13.90 7.33% 8.33% 9.33% 10.33% 11.33% 12.33% 13.33% 14.33%
14.05 6.33% 7.33% 8.33% 9.33% 10.33% 11.33% 12.33% 13.33%
14.20 5.33% 6.33% 7.33% 8.33% 9.33% 10.33% 11.33% 12.33%
14.35 4.33% 5.33% 6.33% 7.33% 8.33% 9.33% 10.33% 11.33%
14.50 3.33% 4.33% 5.33% 6.33% 7.33% 8.33% 9.33% 10.33%
14.65 2.33% 3.33% 4.33% 5.33% 6.33% 7.33% 8.33% 9.33%
14.80 1.33% 2.33% 3.33% 4.33% 5.33% 6.33% 7.33% 8.33%
14.95 0.33% 1.33% 2.33% 3.33% 4.33% 5.33% 6.33% 7.33%
15.10 −0.67% 0.33% 1.33% 2.33% 3.33% 4.33% 5.33% 6.33%
15.25 −1.67% −0.67% 0.33% 1.33% 2.33% 3.33% 4.33% 5.33%
15.40 −2.67% −1.67% −0.67% 0.33% 1.33% 2.33% 3.33% 4.33%
(Continued)
Table 4. (Continued).
15.55 −3.67% −2.67% −1.67% −0.67% 0.33% 1.33% 2.33% 3.33%

15.70 −4.67% −3.67% −2.67% −1.67% −0.67% 0.33% 1.33% 2.33%
15.85 −5.67% −4.67% −3.67% −2.67% −1.67% −0.67% 0.33% 1.33%
16.00 −6.67% −5.67% −4.67% −3.67% −2.67% −1.67% −0.67% 0.33%
16.15 −7.67% −6.67% −5.67% −4.67% −3.67% −2.67% −1.67% −0.67%
16.30 −8.67% −7.67% −6.67% −5.67% −4.67% −3.67% −2.67% −1.67%
16.45 −9.67% −8.67% −7.67% −6.67% −5.67% −4.67% −3.67% −2.67%
16.60 −10.67% −9.67% −8.67% −7.67% −6.67% −5.67% −4.67% −3.67%
16.75 −11.67% −10.67% −9.67% −8.67% −7.67% −6.67% −5.67% −4.67%
16.90 −12.67% −11.67% −10.67% −9.67% −8.67% −7.67% −6.67% −5.67%
17.05 −13.67% −12.67% −11.67% −10.67% −9.67% −8.67% −7.67% −6.67%
17.20 −14.67% −13.67% −12.67% −11.67% −10.67% −9.67% −8.67% −7.67%
17.35 −15.67% −14.67% −13.67% −12.67% −11.67% −10.67% −9.67% −8.67%
17.50 −16.67% −15.67% −14.67% −13.67% −12.67% −11.67% −10.67% −9.67%
17.65 −17.67% −16.67% −15.67% −14.67% −13.67% −12.67% −11.67% −10.67%

17.80 −18.67% −17.67% −16.67% −15.67% −14.67% −13.67% −12.67% −11.67%
17.95 −19.67% −18.67% −17.67% −16.67% −15.67% −14.67% −13.67% −12.67%
18.10 −20.67% −19.67% −18.67% −17.67% −16.67% −15.67% −14.67% −13.67%
18.25 −21.67% −20.67% −19.67% −18.67% −17.67% −16.67% −15.67% −14.67%
18.40 −22.67% −21.67% −20.67% −19.67% −18.67% −17.67% −16.67% −15.67%
18.55 −23.67% −22.67% −21.67% −20.67% −19.67% −18.67% −17.67% −16.67%
18.70 −24.67% −23.67% −22.67% −21.67% −20.67% −19.67% −18.67% −17.67%
18.85 −25.67% −24.67% −23.67% −22.67% −21.67% −20.67% −19.67% −18.67%
19.00 −26.67% −25.67% −24.67% −23.67% −22.67% −21.67% −20.67% −19.67%
19.15 −27.67% −26.67% −25.67% −24.67% −23.67% −22.67% −21.67% −20.67%
19.30 −28.67% −27.67% −26.67% −25.67% −24.67% −23.67% −22.67% −21.67%
19.45 −29.67% −28.67% −27.67% −26.67% −25.67% −24.67% −23.67% −22.67%
19.60 −30.67% −29.67% −28.67% −27.67% −26.67% −25.67% −24.67% −23.67%
19.75 −31.67% −30.67% −29.67% −28.67% −27.67% −26.67% −25.67% −24.67%
19.90 −32.67% −31.67% −30.67% −29.67% −28.67% −27.67% −26.67% −25.67%
20.05 −33.67% −32.67% −31.67% −30.67% −29.67% −28.67% −27.67% −26.67%
20.20 −34.67% −33.67% −32.67% −31.67% −30.67% −29.67% −28.67% −27.67%
20.35 −35.67% −34.67% −33.67% −32.67% −31.67% −30.67% −29.67% −28.67%
20.50 −36.67% −35.67% −34.67% −33.67% −32.67% −31.67% −30.67% −29.67%
20.65 −37.67% −36.67% −35.67% −34.67% −33.67% −32.67% −31.67% −30.67%
20.80 −38.67% −37.67% −36.67% −35.67% −34.67% −33.67% −32.67% −31.67%
TRS between IC and DC for different Libor Rates: position

expressed in % to apply to any amount. The Table shows the
proft and loss to IC from entering a total return swap for
different levels of the stock at maturity in three months.
For example, when the Libor is 0 %, the stock price is 10, the
return from the TRS for IC is 33%. If the stock is 20, the
Libor is 0, the return is − 33%.
Table 5. TRS between IC and DC for different Libor Rates: position in %.
Stock price/libor
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00%

10.00 33.33% 34.33% 35.33% 36.33% 37.33% 38.33% 39.33% 40.33%
10.15 32.33% 33.33% 34.33% 35.33% 36.33% 37.33% 38.33% 39.33%
14.95 0.33% 1.33% 2.33% 3.33% 4.33% 5.33% 6.33% 7.33%
15.10 −0.67% 0.33% 1.33% 2.33% 3.33% 4.33% 5.33% 6.33%
15.25 −1.67% −0.67% 0.33% 1.33% 2.33% 3.33% 4.33% 5.33%
19.90 −32.67% −31.67% −30.67% −29.67% −28.67% −27.67% −26.67% −25.67%
20.05 −33.67% −32.67% −31.67% −30.67% −29.67% −28.67% −27.67% −26.67%
20.20 −34.67% −33.67% −32.67% −31.67% −30.67% −29.67% −28.67% −27.67%
Simulations are done for different levels of Libor from 0 to 7%.

TRS between IC and DC: Simulations for different Libor rates
from 1% to 7%, for different levels of the stock price from 10 to 21
The simulations show that:
– 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.
TRS between IC and DC for different Libor Rates: P&L for IC

is symmetric to DC.
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.
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.
It is an appropriate solution to enter a TRS for IC with DC at 0%.

Any rate above 0%, generates a profit to IC.
This solution is equivalent to lending the stock to DC.
DC has a position in “long the stock”.
Recommendation to IC to implement the TRS:

For IC: it can enter the swap by receiving any fee: for
example 1% since the total return swap allows IC to:
– to keep its portfolio composition

– to enhance returns: by the perceived fee from DC: 1% for example. By
fixing the fee, this allows to hedge immediately against interest rate risks.
– to reduce transaction costs in buying and selling the stocks
Recommendation to DC to implement the TRS:
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.
Case study 2: Combining options and a Total Return Swap

1. Entering a Total Return Swap and IC Buys a call:
If DC sells a call to IC and enters a total return swap:
– 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.
For IC: Buying a call and ‘selling a stock’ is equivalent to a position in

Buying a Put. This strategy offers protection againt market falls.
For DC: Selling a call and entering a TRS (long stock) is equivalent to
selling a Put (synthetic). This strategy does not “hedge” DC against market
falls. The hedge requires that DC sells the stock in the proportion of delta.
2. Risk/Return from Buying Calls and entering a TRS with DC:

General case
Consider a position in stock of 100,000.
I made calculations for three months, interest rate 5%, Volatility 20%, etc.
If the option is worth 0.69, and the stock is 15, it represents 4.6% of the
underlying.
If the option premium equals Libor, we have the following Table.
Example 1. LIBOR = Option Premium = 4.6%
For IC:
This shows a profit in each scenario for IC.
They are protected against downside risk and enhance their returns.
In fact, this is because their postion is similar to buying a put “an insurance
contract”.
Position of DC:
– similar to a classic position “long a stock”, short call: as sell a

put.
– we pay libor + or − and the difference in stocks losses.
– we receive capital gains.
– we receive the option premium.
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%
Investment required to buy calls: (4,600) (4,600) (4,600)

(0.69/15) × 100, 000
Profit (loss) from calls (4,600) 5,400 (4,600)
IC profit and loss with DC in stocks 0 (10,000) 10,000
(opportunity cost)
Libor: 4.6% × 100, 000 4,600 4,600 4,600
Total profit for the option 0 0 10,000
purchase and entering a TRS swap
Return with respect to $100,000 0 0 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
Sell calls: (0.69/15) × 100, 000 4,600 4,600 4,600

Profit (loss) from calls 4,600 (5,400) 4,600
DC profit and loss with IC in stocks 0 10,000 (10,000)
Libor: 4.6% × 100, 000 (4,600) 4,600 (4,600)
Total profit for the option 0 0 (10,000)
sale and entering a TRS swap
Return with respect to 100,000 0 0 (10%)
Example 2. LIBOR = Option Premium = 4.6 %
FOR DC: the opposite position

This shows a profit & Loss which is the opposite to IC.
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.
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
Sell calls: (1.2/15) × 100, 000 8,000 8,000 8,000

Profit (loss) from calls 8,000 (2,000) 8,000
DC profit and loss with IC in stocks 0 10,000 (10,000)
Libor: 5% × 100, 000 (5,000) (5,000) (5,000)
Total profit for the option sold 3,000 3,000 (7,000)
and entering a TRS swap
Return with respect to 100,000 3% 3% (7%)
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
sell calls: (1.5/15) × 100, 000 10,000 10,000 10,000

DC profit and loss with IC in stocks 0 10,000 (10,000)
Libor: 5% × 100, 000 (5,000) (5,000) (5,000)
Total profit for the option 5,000 5,000 (5,000)
sold and entering a TRS swap
Return with respect to 100,000 5% 5% (5%)
Example 3. Option premium set higher than Libor

LIBOR = 5%, Option Premium = 8%
For DC:
This is not a symmetric profile.
Example 4: Option premium = 2 times LIBOR,
LIBOR = 5%, Option Premium = 10% instead of 8%.
For DC:
This is a symmetric profile.
Conclusion: When the option premium is set at a level = 2 times
Libor, the profile becomes symmetric between IC and DC.
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.
– This is a kind of insurance against downside risk.

– This postion is similar to nuying a put.
– Pay nearly equivalent of libor 5% to buy the protection against downside
risk.
General cases and P&L

For DC: Risk/return from selling a call and (stock borrowing) entering
simultaneously into a total return swap with IC
Option premim = 2 times Libor
We use same assumptions as before, with stock = 15, strike price = 15,
premium 0.1, Libor = 0.05, Stock position: $100,000.
Position:
– we pay libor + or − and the difference in stocks losses.

– we receive capital gains.
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).
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%
Buy calls: (1.5/15) × 100, 000 (10,000) (10,000) (10,000)

Profit (loss) from calls (10,000) (0,000) (10,000)
IC profit and loss with in stocks with 0 (10,000) 10,000
Libor: 5% × 100, 000 5,000 5,000 5,000
Total profit for the option (5,000) (5,000) 5,000
sold and entering a TRS swap
Return with respect to 100,000 (5%) (5%) 5%
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
– DC sells protection to IC against any downward movement in the stock

price of EMA.
– DC receives the premium (about 5% of the notional amount in this case).
– If the stock goes down, DC pays IC the losses.
– This postion is similar to selling a put.
– DC receives nearly equivalent of Libor 5% to sell the protection against
downside risk to IC.
Case 2: strategy of buying calls and entering a TRS with DC

For IC: option premium = 2 times LIBOR
Conclusion: This is a symmetric profile.
When IC wins, DC loses the same amount and vice-versa.
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.
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.
Total impact on the 30% position from entering a TRS

For IC: Total impact on the 30% position from entering a TRS
(for 10% of the stock position and Buying a call from DC)
For IC: Total impact on the 30% position from entering a TRS (for 10%
of the stock position and Buying a call from DC on the 3% of the stocks.)
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:
For IC: Total impact on the 30% position from entering a TRS (for 10% of
the stock position and Buying a call from DC on the 3% of the stocks.
Table 11. Total TRS IC & DC.
IC long call IC postion

TRS IC call impact TRS + long CALL
33% −10.00% 23.33%

32% −10.00% 22.33%
31% −10.00% 21.33%
30% −10.00% 20.33%
29% −10.00% 19.33%
28% −10.00% 18.33%
27% −10.00% 17.33%
26% −10.00% 16.33%
25% −10.00% 15.33%
24% −10.00% 14.33%
23% −10.00% 13.33%
22% −10.00% 12.33%
21% −10.00% 11.33%
20% −10.00% 10.33%
19% −10.00% 9.33%
18% −10.00% 8.33%
17% −10.00% 7.33%
16% −10.00% 6.33%
15% −10.00% 5.33%
14% −10.00% 4.33%
13% −10.00% 3.33%
12% −10.00% 2.33%
11% −10.00% 1.33%
10% −10.00% 0.33%
9% −10.00% −0.67%
8% −10.00% −1.67%
7% −10.00% −2.67%
6% −10.00% −3.67%
5% −10.00% −4.67%
4% −10.00% −5.67%
3% −10.00% −6.67%
2% −10.00% −7.67%
1% −10.00% −8.67%
0% −10.00% −9.67%
−1% −9.33% −10.00%
−2% −8.33% −10.00%
−3% −7.33% −10.00%
−4% −6.33% −10.00%
−5% −5.33% −10.00%
−6% −4.33% −10.00%
−7% −3.33% −10.00%
−8% −2.33% −10.00%
−9% −1.33% −10.00%
−10% −0.33% −10.00%
(Continued)
IC long call IC postion

TRS IC call impact TRS + long CALL
−11% 0.67% −10.00%

−12% 1.67% −10.00%
−13% 2.67% −10.00%
−14% 3.67% −10.00%
−15% 4.67% −10.00%
−16% 5.67% −10.00%
−17% 6.67% −10.00%
−18% 7.67% −10.00%
−19% 8.67% −10.00%
−20% 9.67% −10.00%
−21% 10.67% −10.00%
−22% 11.67% −10.00%
−23% 12.67% −10.00%
−24% 13.67% −10.00%
−25% 14.67% −10.00%
−26% 15.67% −10.00%
−27% 16.67% −10.00%
−28% 17.67% −10.00%
−29% 18.67% −10.00%
−30% 19.67% −10.00%
−31% 20.67% −10.00%
−32% 21.67% −10.00%
−33% 22.67% −10.00%
−34% 23.67% −10.00%
−35% 24.67% −10.00%
−36% 25.67% −10.00%
−37% 26.67% −10.00%
−38% 27.67% −10.00%
−39% 28.67% −10.00%
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%
Table 12. Impact TRS and long a call.
Stock price/Libor TRS IC call impact TRS + long CALL
10.00 33% −10.00% 23.33%

10.15 32% −10.00% 22.33%
14.95 0% −10.00% −9.67%
15.10 −1% −9.33% −10.00%
15.25 −2% −8.33% −10.00%
19.90 −33% 22.67% −10.00%
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.
General case further: IC — TRS and sells a call

Sell a call, enter TRS and uses the proceeds of selling a call at 5%; impact
on total position of IC.
It is clear that the total impact on the position of IC is clear.
Table 13. IC buys a call and enters TRS.
12.50 6667 −6667 6.67% −16.67% −16.00%

12.55 6333 −6333 6.33% −16.33% −15.70%
12.60 6000 −6000 6.00% −16.00% −15.40%
12.65 5667 −5667 5.67% −15.67% −15.10%
12.70 5333 −5333 5.33% −15.33% −14.80%
12.75 5000 −5000 5.00% −15.00% −14.50%
12.80 4667 −4667 4.67% −14.67% −14.20%
12.85 4333 −4333 4.33% −14.33% −13.90%
12.90 4000 −4000 4.00% −14.00% −13.60%
12.95 3667 −3667 3.67% −13.67% −13.30%
13.00 3333 −3333 3.33% −13.33% −13.00%
13.05 3000 −3000 3.00% −13.00% −12.70%
13.10 2667 −2667 2.67% −12.67% −12.40%
13.15 2333 −2333 2.33% −12.33% −12.10%
13.20 2000 −2000 2.00% −12.00% −11.80%
13.25 1667 −1667 1.67% −11.67% −11.50%
13.30 1333 −1333 1.33% −11.33% −11.20%
13.35 1000 −1000 1.00% −11.00% −10.90%
13.40 667 −667 0.67% −10.67% −10.60%
13.45 333 −333 0.33% −10.33% −10.30%
13.50 0 0 0.00% −10.00% −10.00%
13.55 −333 333 −0.33% −9.67% −9.70%
13.60 −667 667 −0.67% −9.33% −9.40%
13.65 −1000 1000 −1.00% −9.00% −9.10%
13.70 −1333 1333 −1.33% −8.67% −8.80%
13.75 −1667 1667 −1.67% −8.33% −8.50%
13.80 −2000 2000 −2.00% −8.00% −8.20%
13.85 −2333 2333 −2.33% −7.67% −7.90%
13.90 −2667 2667 −2.67% −7.33% −7.60%
13.95 −3000 3000 −3.00% −7.00% −7.30%
14.00 −3333 3333 −3.33% −6.67% −7.00%
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%.
Case Study 2: Stock holding (or buying)

DC: Stock disposal scenario
– DC can buy the stock from the market, or from IC or uses IC stocks and
shares profit from market making activities or fees from management
positions.
– DC sells calls to IC or to third parties.
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%
-8.00% Total impact on the 30M

position from 10% TRS+Call
-10.00%
initial position
-12.00%
-14.00%
-16.00%
-18.00%
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 1 — For IC: risk/return from buying a call from DC

(or any call buyer : third party for example or the market)
This graph concerns IC or any call buyer.
The graph shows the profit and loss as a function of the underlying stock
price at maturity. Positive sign is a profit.
If IC buys a call and believes the stock is going up, it will realize unlimited
profits.
IC buys a call expecting the stock to rise and pays to DC a premium of
nearly 10%.
IC profits can be unlimited as shown in the graph until 35 times their initial
payment.
Case 2 — For DC: risk/return from selling a call and buying the stock
DC sells the call and buys the stock.
Position:
– we pay for the stock.

DC sells a call, holds the stocks and receives a premium of nearly 10%.
DC profits are limited to the premium received.
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
33% 10.00% 43.33% −29.00%

32% 10.00% 42.33% −28.10%
31% 10.00% 41.33% −27.20%
30% 10.00% 40.33% −26.30%
29% 10.00% 39.33% −25.40%
28% 10.00% 38.33% −24.50%
27% 10.00% 37.33% −23.60%
26% 10.00% 36.33% −22.70%
25% 10.00% 35.33% −21.80%
24% 10.00% 34.33% −20.90%
23% 10.00% 33.33% −20.00%
22% 10.00% 32.33% −19.10%
21% 10.00% 31.33% −18.20%
20% 10.00% 30.33% −17.30%
19% 10.00% 29.33% −16.40%
18% 10.00% 28.33% −15.50%
17% 10.00% 27.33% −14.60%
16% 10.00% 26.33% −13.70%
15% 10.00% 25.33% −12.80%
14% 10.00% 24.33% −11.90%
13% 10.00% 23.33% −11.00%
12% 10.00% 22.33% −10.10%
11% 10.00% 21.33% −9.20%
10% 10.00% 20.33% −8.30%
9% 10.00% 19.33% −7.40%
8% 10.00% 18.33% −6.50%
7% 10.00% 17.33% −5.60%
6% 10.00% 16.33% −4.70%
5% 10.00% 15.33% −3.80%
4% 10.00% 14.33% −2.90%
3% 10.00% 13.33% −2.00%
2% 10.00% 12.33% −1.10%
1% 10.00% 11.33% −0.20%
0% 10.00% 10.33% 0.70%
−1% 9.33% 8.67% 1.53%
−2% 8.33% 6.67% 2.33%
−3% 7.33% 4.67% 3.13%
−4% 6.33% 2.67% 3.93%
−5% 5.33% 0.67% 4.73%
−6% 4.33% −1.33% 5.53%
−7% 3.33% −3.33% 6.33%
−8% 2.33% −5.33% 7.13%
−9% 1.33% −7.33% 7.93%
(Continued)
−10% 0.33% −9.33% 8.73%

−11% −0.67% −11.33% 9.53%
−12% −1.67% −13.33% 10.33%
−13% −2.67% −15.33% 11.13%
−14% −3.67% −17.33% 11.93%
−15% −4.67% −19.33% 12.73%
−16% −5.67% −21.33% 13.53%
−17% −6.67% −23.33% 14.33%
−18% −7.67% −25.33% 15.13%
−19% −8.67% −27.33% 15.93%
−20% −9.67% −29.33% 16.73%
−21% −10.67% −31.33% 17.53%
−22% −11.67% −33.33% 18.33%
−23% −12.67% −35.33% 19.13%
−24% −13.67% −37.33% 19.93%
−25% −14.67% −39.33% 20.73%
−26% −15.67% −41.33% 21.53%
−27% −16.67% −43.33% 22.33%
−28% −17.67% −45.33% 23.13%
−29% −18.67% −47.33% 23.93%
−30% −19.67% −49.33% 24.73%
−31% −20.67% −51.33% 25.53%
−32% −21.67% −53.33% 26.33%
−33% −22.67% −55.33% 27.13%
−34% −23.67% −57.33% 27.93%
−35% −24.67% −59.33% 28.73%
−36% −25.67% −61.33% 29.53%
−37% −26.67% −63.33% 30.33%
−38% −27.67% −65.33% 31.13%
−39% −28.67% −67.33% 31.93%
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:
– we implement a hedge by buying delta units of the stock

– we receive the option premium
Table 15. Impact on the position of IC of short call, long call and entering a total
return swap.
33% 10.00% 43.33% −29.00%

15% 10.00% 25.33% −12.80%
11% 10.00% 21.33% −9.20%
7% 10.00% 17.33% −5.60%
4% 10.00% 14.33% −2.90%
3% 10.00% 13.33% −2.00%
2% 10.00% 12.33% −1.10%
1% 10.00% 11.33% −0.20%
0% 10.00% 10.33% 0.70%
−1% 9.33% 8.67% 1.53%
−2% 8.33% 6.67% 2.33%
−7% 3.33% −3.33% 6.33%
−11% −0.67% −11.33% 9.53%
−15% −4.67% −19.33% 12.73%
−39% −28.67% −67.33% 31.93%
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.
Case study — General cases for any initial capital for

different strategies
We use the return (P&L in %)
ICD buy call
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).
DCG sells call and buy stock
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
DCG sells call and

-5000.00
buy stock
-10000.00
-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.
Column 1: stock price at maturity

Column 2: P&L from the strategy of selling the call at maturity
Delta hedge 0.55
15000
10000
5000
0 Delta hedge 0.55

10
11.1
12.1
13.2
14.2
15.3
16.3
17.4
18.4
19.5
20.5
-5000
-10000
-15000
Graph 9. Risk/return from selling a call and buying the stock as a hedge.
Column 3: P&L from the strategy long the stock

Column 4: P&L from the strategy of selling the call at maturity
Column 1: stock price at maturity
S-C: P&L from the strategy short call, long the stock.
C-S: P&L from the strategy long call, short the stock.
0.55 S-C: P&L from the strategy short call, hedge: long 0.55 units
of the stock.
−0.55 S + C: P&L from the strategy long call, hedge: short 0.55
units of the stock.
Each graph shows the P&L of the strategy for a given level of the underlyng
stock in percentage.
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.
Table 16 shows the P&L from the strategy of selling the call and holding the
stock for different levels of the underlying asset by implementing a hedge:
long delta stocks.
It shows also the P&L from being long a call, short the stocks.
It gives IC total position of short call on 1/10 of its position and long stocks.
Study the impact on different strategies:
Table 16. P&L from the strategy of selling the call and holding the stock for different
levels of the underlying asset to DC.
P&L from Sell Profit&Loss from P&L, short call

Stock price call at Maturity Stock at T long Stock
at maturity “−c” s s−c
10 10.00% −33.33% −23.33% −0.10

10.15 10.00% −32.33% −22.33% −0.10
10.1 10.00% −31.33% −21.33% −0.10
10.45 10.00% −30.33% −20.33% −0.10
10.1 10.00% −29.33% −19.33% −0.10
10.75 10.00% −28.33% −18.33% −0.10
10.1 10.00% −27.33% −17.33% −0.10
11.05 10.00% −26.33% −16.33% −0.10
11.1 10.00% −25.33% −15.33% −0.10
11.35 10.00% −24.33% −14.33% −0.10
11.1 10.00% −23.33% −13.33% −0.10
11.65 10.00% −22.33% −12.33% −0.10
11.1 10.00% −21.33% −11.33% −0.10
11.95 10.00% −20.33% −10.33% −0.10
12.1 10.00% −19.33% −9.33% −0.10
12.25 10.00% −18.33% −8.33% −0.10
12.1 10.00% −17.33% −7.33% −0.10
12.55 10.00% −16.33% −6.33% −0.10
12.1 10.00% −15.33% −5.33% −0.10
12.85 10.00% −14.33% −4.33% −0.10
13 10.00% −13.33% −3.33% −0.10
13.15 10.00% −12.33% −2.33% −0.10
13.1 10.00% −11.33% −1.33% −0.10
13.45 10.00% −10.33% −0.33% −0.10
13.1 10.00% −9.33% 0.67% −0.10
13.75 10.00% −8.33% 1.67% −0.10
13.1 10.00% −7.33% 2.67% −0.10
14.05 10.00% −6.33% 3.67% −0.10
14.1 10.00% −5.33% 4.67% −0.10
14.35 10.00% −4.33% 5.67% −0.10
14.1 10.00% −3.33% 6.67% −0.10
14.65 10.00% −2.33% 7.67% −0.10
14.1 10.00% −1.33% 8.67% −0.10
14.95 10.00% −0.33% 9.67% −0.10
15.1 9.33% 0.67% 10.00% −0.09
15.25 8.33% 1.67% 10.00% −0.08
15.1 7.33% 2.67% 10.00% −0.07
15.55 6.33% 3.67% 10.00% −0.06
15.1 5.33% 4.67% 10.00% −0.05
15.85 4.33% 5.67% 10.00% −0.04
16 3.33% 6.67% 10.00% −0.03
(Continued)
P&L from Sell Profit&Loss from P&L, short call

Stock price call at Maturity Stock at T long Stock
at maturity “−c” s s−c
16.15 2.33% 7.67% 10.00% −0.02

16.1 1.33% 8.67% 10.00% −0.01
16.45 0.33% 9.67% 10.00% 0.00
16.1 −0.67% 10.67% 10.00% 0.01
16.75 −1.67% 11.67% 10.00% 0.02
16.1 −2.67% 12.67% 10.00% 0.03
17.05 −3.67% 13.67% 10.00% 0.04
17.1 −4.67% 14.67% 10.00% 0.05
17.35 −5.67% 15.67% 10.00% 0.06
17.1 −6.67% 16.67% 10.00% 0.07
17.65 −7.67% 17.67% 10.00% 0.08
17.1 −8.67% 18.67% 10.00% 0.09
17.95 −9.67% 19.67% 10.00% 0.10
18.1 −10.67% 20.67% 10.00% 0.11
18.25 −11.67% 21.67% 10.00% 0.12
18.1 −12.67% 22.67% 10.00% 0.13
18.55 −13.67% 23.67% 10.00% 0.14
18.1 −14.67% 24.67% 10.00% 0.15
18.85 −15.67% 25.67% 10.00% 0.16
19 −16.67% 26.67% 10.00% 0.17
19.15 −17.67% 27.67% 10.00% 0.18
19.1 −18.67% 28.67% 10.00% 0.19
19.45 −19.67% 29.67% 10.00% 0.20
19.1 −20.67% 30.67% 10.00% 0.21
19.75 −21.67% 31.67% 10.00% 0.22
19.1 −22.67% 32.67% 10.00% 0.23
20.05 −23.67% 33.67% 10.00% 0.24
20.2 −24.67% 34.67% 10.00% 0.25
20.35 −25.67% 35.67% 10.00% 0.26
20.2 −26.67% 36.67% 10.00% 0.27
20.65 −27.67% 37.67% 10.00% 0.28
20.2 −28.67% 38.67% 10.00% 0.29
20.95 −29.67% 39.67% 10.00% 0.30
21.2 −30.67% 40.67% 10.00% 0.31
21.25 −31.67% 41.67% 10.00% 0.32
21.4 −32.67% 42.67% 10.00% 0.33
In order to study the impact of different strategies on the P&L, we consider

the following Table.
Column 1: short call, long delta stocks at maturity

Column 2: P&L from the strategy long call, short stock at maturity
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%
Graph 10. P&L from different strategies.
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.
Impact on IC overall position (30%) of buying calls on 3%

of EMA
If IC buys calls and sells the stock, the graph provides the P&L for all the
positions in stocks after implementing the strategy. It improves the upside
potential.
Table 17 gives the P&L for IC from different structures: Impact on all
position from sell call on 1/10, (buy call on 1/10).
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
−8.33% 23.33% 8.33% −32.33% 1.00% −33.33%

−7.78% 22.33% 7.78% −31.33% 1.00% −32.33%
−7.23% 21.33% 7.23% −30.33% 1.00% −31.33%
−6.68% 20.33% 6.68% −29.33% 1.00% −30.33%
−6.13% 19.33% 6.13% −28.33% 1.00% −29.33%
−5.58% 18.33% 5.58% −27.33% 1.00% −28.33%
−5.03% 17.33% 5.03% −26.33% 1.00% −27.33%
−4.48% 16.33% 4.48% −25.33% 1.00% −26.33%
−3.93% 15.33% 3.93% −24.33% 1.00% −25.33%
−3.38% 14.33% 3.38% −23.33% 1.00% −24.33%
−2.83% 13.33% 2.83% −22.33% 1.00% −23.33%
−2.28% 12.33% 2.28% −21.33% 1.00% −22.33%
−1.73% 11.33% 1.73% −20.33% 1.00% −21.33%
−1.18% 10.33% 1.18% −19.33% 1.00% −20.33%
−0.63% 9.33% 0.63% −18.33% 1.00% −19.33%
−0.08% 8.33% 0.08% −17.33% 1.00% −18.33%
0.47% 7.33% −0.47% −16.33% 1.00% −17.33%
1.02% 6.33% −1.02% −15.33% 1.00% −16.33%
1.57% 5.33% −1.57% −14.33% 1.00% −15.33%
2.12% 4.33% −2.12% −13.33% 1.00% −14.33%
2.67% 3.33% −2.67% −12.33% 1.00% −13.33%
3.22% 2.33% −3.22% −11.33% 1.00% −12.33%
3.77% 1.33% −3.77% −10.33% 1.00% −11.33%
4.32% 0.33% −4.32% −9.33% 1.00% −10.33%
4.87% −0.67% −4.87% −8.33% 1.00% −9.33%
5.42% −1.67% −5.42% −7.33% 1.00% −8.33%
5.97% −2.67% −5.97% −6.33% 1.00% −7.33%
6.52% −3.67% −6.52% −5.33% 1.00% −6.33%
7.07% −4.67% −7.07% −4.33% 1.00% −5.33%
7.62% −5.67% −7.62% −3.33% 1.00% −4.33%
8.17% −6.67% −8.17% −2.33% 1.00% −3.33%
8.72% −7.67% −8.72% −1.33% 1.00% −2.33%
9.27% −8.67% −9.27% −0.33% 1.00% −1.33%
9.82% −9.67% −9.82% 0.67% 1.00% −0.33%
9.70% −10.00% −9.70% 1.60% 0.93% 0.67%
9.25% −10.00% −9.25% 2.50% 0.83% 1.67%
8.80% −10.00% −8.80% 3.40% 0.73% 2.67%
8.35% −10.00% −8.35% 4.30% 0.63% 3.67%
7.90% −10.00% −7.90% 5.20% 0.53% 4.67%
(Continued)
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
7.45% −10.00% −7.45% 6.10% 0.43% 5.67%

7.00% −10.00% −7.00% 7.00% 0.33% 6.67%
6.55% −10.00% −6.55% 7.90% 0.23% 7.67%
6.10% −10.00% −6.10% 8.80% 0.13% 8.67%
5.65% −10.00% −5.65% 9.70% 0.03% 9.67%
5.20% −10.00% −5.20% 10.60% −0.07% 10.67%
4.75% −10.00% −4.75% 11.50% −0.17% 11.67%
4.30% −10.00% −4.30% 12.40% −0.27% 12.67%
3.85% −10.00% −3.85% 13.30% −0.37% 13.67%
3.40% −10.00% −3.40% 14.20% −0.47% 14.67%
2.95% −10.00% −2.95% 15.10% −0.57% 15.67%
2.50% −10.00% −2.50% 16.00% −0.67% 16.67%
2.05% −10.00% −2.05% 16.90% −0.77% 17.67%
1.60% −10.00% −1.60% 17.80% −0.87% 18.67%
1.15% −10.00% −1.15% 18.70% −0.97% 19.67%
0.70% −10.00% −0.70% 19.60% −1.07% 20.67%
0.25% −10.00% −0.25% 20.50% −1.17% 21.67%
−0.20% −10.00% 0.20% 21.40% −1.27% 22.67%
−0.65% −10.00% 0.65% 22.30% −1.37% 23.67%
−1.10% −10.00% 1.10% 23.20% −1.47% 24.67%
−1.55% −10.00% 1.55% 24.10% −1.57% 25.67%
−2.00% −10.00% 2.00% 25.00% −1.67% 26.67%
−2.45% −10.00% 2.45% 25.90% −1.77% 27.67%
−2.90% −10.00% 2.90% 26.80% −1.87% 28.67%
−3.35% −10.00% 3.35% 27.70% −1.97% 29.67%
−3.80% −10.00% 3.80% 28.60% −2.07% 30.67%
−4.25% −10.00% 4.25% 29.50% −2.17% 31.67%
−4.70% −10.00% 4.70% 30.40% −2.27% 32.67%
−5.15% −10.00% 5.15% 31.30% −2.37% 33.67%
−5.60% −10.00% 5.60% 32.20% −2.47% 34.67%
−6.05% −10.00% 6.05% 33.10% −2.57% 35.67%
−6.50% −10.00% 6.50% 34.00% −2.67% 36.67%
−6.95% −10.00% 6.95% 34.90% −2.77% 37.67%
−7.40% −10.00% 7.40% 35.80% −2.87% 38.67%
−7.85% −10.00% 7.85% 36.70% −2.97% 39.67%
−8.30% −10.00% 8.30% 37.60% −3.07% 40.67%
−8.75% −10.00% 8.75% 38.50% −3.17% 41.67%
−9.20% −10.00% 9.20% 39.40% −3.27% 42.67%
Table 18.
IC total position all stocks

IC; sellcalls, 1/10 sell call Impact on all intial position
1.00% −32.33% −33.33%

1.00% −31.33% −32.33%
1.00% −20.33% −21.33%
1.00% −19.33% −20.33%
1.00% −18.33% −19.33%
1.00% −17.33% −18.33%
1.00% −16.33% −17.33%
1.00% −0.33% −1.33%
1.00% 0.67% −0.33%
0.93% 1.60% 0.67%
0.83% 2.50% 1.67%
0.73% 3.40% 2.67%
−0.07% 10.60% 10.67%
−0.67% 16.00% 16.67%
−1.17% 20.50% 21.67%
−3.27% 39.40% 42.67%
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%
Graph 11. Impact on IC overall position (30%) of selling calls on 3% of EMA.

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.
Sell call impact on IC

(no hedge) All stocks Buy call impact:
impact on all IC position IC: 1/10 sell call intial position sell stock
−32.33% 1.00% −33.33% −31.00%

−32.00% 1.00% −33.00% −30.70%
−31.33% 1.00% −32.33% −30.10%
−30.67% 1.00% −31.67% −29.50%
−30.00% 1.00% −31.00% −28.90%
−29.33% 1.00% −30.33% −28.30%
−28.67% 1.00% −29.67% −27.70%
−28.00% 1.00% −29.00% −27.10%
−27.33% 1.00% −28.33% −26.50%
−26.67% 1.00% −27.67% −25.90%
−26.00% 1.00% −27.00% −25.30%
−25.33% 1.00% −26.33% −24.70%
−24.67% 1.00% −25.67% −24.10%
−24.00% 1.00% −25.00% −23.50%
−23.33% 1.00% −24.33% −22.90%
−22.67% 1.00% −23.67% −22.30%
−22.00% 1.00% −23.00% −21.70%
−21.33% 1.00% −22.33% −21.10%
−20.67% 1.00% −21.67% −20.50%
−20.00% 1.00% −21.00% −19.90%
−19.33% 1.00% −20.33% −19.30%
−18.67% 1.00% −19.67% −18.70%
−18.00% 1.00% −19.00% −18.10%
−17.33% 1.00% −18.33% −17.50%
(Continued)

−16.67% 1.00% −17.67% −16.90%

−16.00% 1.00% −17.00% −16.30%
−15.33% 1.00% −16.33% −15.70%
−14.67% 1.00% −15.67% −15.10%
−14.00% 1.00% −15.00% −14.50%
−13.33% 1.00% −14.33% −13.90%
−12.67% 1.00% −13.67% −13.30%
−12.00% 1.00% −13.00% −12.70%
−11.33% 1.00% −12.33% −12.10%
−10.67% 1.00% −11.67% −11.50%
−10.00% 1.00% −11.00% −10.90%
−9.33% 1.00% −10.33% −10.30%
−8.67% 1.00% −9.67% −9.70%
−8.00% 1.00% −9.00% −9.10%
−7.33% 1.00% −8.33% −8.50%
−6.67% 1.00% −7.67% −7.90%
−6.00% 1.00% −7.00% −7.30%
−5.33% 1.00% −6.33% −6.70%
−4.67% 1.00% −5.67% −6.10%
−4.00% 1.00% −5.00% −5.50%
−3.33% 1.00% −4.33% −4.90%
−2.67% 1.00% −3.67% −4.30%
−2.00% 1.00% −3.00% −3.70%
−1.33% 1.00% −2.33% −3.10%
−0.67% 1.00% −1.67% −2.50%
0.00% 1.00% −1.00% −1.90%
0.67% 1.00% −0.33% −1.30%
1.30% 0.97% 0.33% −0.67%
1.90% 0.90% 1.00% 0.00%
2.50% 0.83% 1.67% 0.67%
3.10% 0.77% 2.33% 1.33%
3.70% 0.70% 3.00% 2.00%
4.30% 0.63% 3.67% 2.67%
4.90% 0.57% 4.33% 3.33%
5.50% 0.50% 5.00% 4.00%
6.10% 0.43% 5.67% 4.67%
6.70% 0.37% 6.33% 5.33%
7.30% 0.30% 7.00% 6.00%
7.90% 0.23% 7.67% 6.67%
8.50% 0.17% 8.33% 7.33%
9.10% 0.10% 9.00% 8.00%
9.70% 0.03% 9.67% 8.67%
(Continued)
Table 19. (Continued)

10.30% −0.03% 10.33% 9.33%

10.90% −0.10% 11.00% 10.00%
11.50% −0.17% 11.67% 10.67%
12.10% −0.23% 12.33% 11.33%
12.70% −0.30% 13.00% 12.00%
13.30% −0.37% 13.67% 12.67%
13.90% −0.43% 14.33% 13.33%
14.50% −0.50% 15.00% 14.00%
15.10% −0.57% 15.67% 14.67%
15.70% −0.63% 16.33% 15.33%
16.30% −0.70% 17.00% 16.00%
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
exchange for the assurance of a known maximum loss he loses performance.

The tradeoff is economically equitable.
Exercise: Explain how leverage is associated with buying call
options.
Options can be used as an alternative to a direct investment in the
stock.
They can also be used for leverage.
The investor can devote his risk capital to buy call options.
With options, the investor has the benefit of an assured maximum loss:
no matter how much the stock declines, the loss can never be greater than
the price paid for the options.
For this position, the maximum loss is realized at and below the strike
price. Above the strike price, the risk/return line raises diagonally. The
dollar amount of profit and loss is increased to reflect leverage.
Exercise: Explain selling calls.
– Selling (short) calls is an alternative to short selling the
stocks
Selling calls is an alternative to buying puts for an investor who
anticipated a falling market.
– Leverage is associated with selling call options

– Options may be part of a defensive strategy: they can be held in
conjunction with cash as an alternative to stock. The cash provides
protection in the event of poor market performance. The calls give
exposure to the upside of the market.
– Options can be used to participate in the upside of the market without
having the full amount of necessary cash available to gain exposure.
– Potential loss is limited to the premium paid. This allows exposure to a
substantial amount of the underlying asset while only risking a proportion
of the exposure obtained.
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.
Short positions in stocks must be margined and do not have to

be financed. They generate a positive cash flow that can be invested in
Treasury bills while the stocks are held short.
Put options although they are geared to return a profit when the
underlying asset falls, require an initial cash investment when they are
purchased.
To come as close as possible to constructing a position in put options
to replicate short selling a portfolio of stocks, we must borrow the funds
required to buy the puts.
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.
Exercise: Explain the strategy of selling put options.

Selling puts (short sale) is an alternative to buying stocks: the
investor receives the assurance of a known maximum gain but a
potentially enormous loss.
Buying calls can be a unique managerial alternative to buying the

underlying asset. Another alternative is to short sell puts.
Selling puts can be an alternative to buying calls for an investor who
anticipated a rising market.
By short-selling puts, the investor subjects himself to losses on the
downside beneath the tolerance point limited only by the fact that the
underlying stock can’t be priced less than zero.
A short put position is covered if its aggregate underlying value is no
greater than the amount of cash that would have been required to buy the
portfolio.
If an investor created an uncovered position, or nacked, in short puts,
his loss in ‘unlimited’.
We can compute downside tolerance point when the investor sells puts.
Above a given stock price, the premium initially received by the investor
from the short sale and the interest income from his T. Bills are sufficient
to repay the exercise value of the short puts.
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.
Exercise: Explain how options positions are hedged.

Hedging the option using the underlying asset:
Buy a call: sell the underlying
Sell a call: buy the underlying
Buy a put: buy the underlying
Sell a put: sell the underlying
Hedging the option using an option:

– Buy a call and sell a put is a strategy equivalent to holding the underlying
asset
To be completely hedged: sell the underlying
– Sell a call and buy a put is a strategy equivalent to selling the underlying
asset
To be completely hedged: buy the underlying
Exercise: How to implement the hedge when selling calls?

We buy the underlying assets
Or we borrow the stock.
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.
Exercise: What are the classic measures in option valuation?

Time value and instrisic value:
When the stock is 16 and the strike price is 14, the instrinsic value is 2, the
call price may be 3. It corresponds to the cash obtained for an immediate
exercise of the option.
Time value
If call price is 3 and the intrinsic value is 2, then the time value is 1.
The interest advantage

In our example, the investor is gaining exposure to an underlying asset
worth 16 for an outlay of 3.
He is able to place 13, or (16 −3) on deposit for the six month until
the options maturity. If the rate of interest is 10%, he will receive a benefit
of 13 × 0.5 × (0.1) = 0.65. This can be compared with total time value of 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).
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.
Exercise: What are the effect of parameters on option valuation?

Call price depend on underlying, strike, dividends, interest, time and
volatility.
– When the asset price goes up, the call price is higher.
– The call holder has a funding benefit over the equivalent exposure in the
underlying asset.
– The purchase of the put involves expenditure, whereas the alternative
short stock position provides a positive cash flow and interest income.
– The option price depends on the forward price of the underlying.
This forward price depends also on interest rates.
– The higher the interest rates, the greater the benefit of holding a cash
deposit and agreeing to purchase the asset at a fixed price at an agreed
date in the future.
– Higher interest rates result in higher forward prices.
Calls require a low initial outlay and show the ability to place residual
cash on deposit. Hence, the call increases in value when the interest rate
rises.
In contrast, the holder of a short asset or forward position, will benefit
from a funding credit, and higher interest rates will increase the size of his
credit. As this is not paid to put holders, their value will decrease as rates
rise.
– Future dividends reduce the call cost and increase the cost of the put.
Exercise: What is the option delta?

It is the amount by which the option price changes for a corresponding
change in the underlying asset price.
Exercise: What is short sale of stocks?

The New York stock exchange’s plus-tick rule requires that any short sale
of stock be executed only on a plus-tick or zero plus tick from the previous
trade price.
Exercise: What is a synthetic call?

– buy a certain amount of the underlying asset and borrow a given amount
Exercise: What is a synthetic put?

– short sell a certain amount of the underlying asset and lend a given
amount
Exercise: What is a basket of option?

A basket option can be exercised into a defined basket of shares, rather
than a single stock.
They can be structured to gain sector, or country specific exposure.
Combinations of single stocks will in general be more expensive than
an option on a basket of the same stocks, because the volatility exhibited
by a basket of stocks will normally be lower than the average volatility of
the component companies. This can be explained by imperfect correlation
between stocks.
– It is seldom to find options on all stocks
Case Study: Calls and Puts on EMA

Example: A call on EMA
Issuer: DFGCV
Issue Size: 100,000 calls
Style: American: Exercisable for cash (or physical) at any time
Maturity: 2 years
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
Table 1.
Type of
options Strike price % of Moneyness Issue price Gearing
call 14 At the money: 3.68 3.8 × (14/3.68)

strike price equal
asset price
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.
Example: If we change the maturity to 3 months

Parameters:
Asset price: 14, Interest rate: 5%
Volatility: 40%, Call premium: 1.2 for a strike of 14
Put premium: 1 for a strike of 14
Minimum trade size: 10 options,
Table 3.
Type of
options Strike price % of moneyness Issue price Gearing
call 14 At the money: 1.2 11.66 × (14/1.2)

strike price equal
asset price
Table 4.
Underlying Call P/L

expiry value Call expiry value In dollars 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 −0.2 −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.
Emaar 14 call — cost 1.2

Leverage: 49 times
Exercise: Example of a put option

Issuer: DC
Issue Size: 100,000 puts
Style: American or European style: exercisable for physical or cash at any
time
Underlying asset price level, (Index level): 14
Maturity: 3 months
Table 5.
Type of option Strike price % of moneyness premium gearing

put 14 At the money 1% 14 X
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%
Put option exercise

– The put gives the investor upon exercise the difference in the currency
specified, at any time before maturity, between the strike price and the
underlying asset (index) level.
EMA put — cost 1

The option settlement value or final exercise is given by the difference
between the strike price and underlying (index) closing level.
When the underlying (index) is less than the strike price, the put has
a positive intrinsic value. Otherwise, the put expires worthless.
The total return is strike price less underlying (index) level less purchase
price.
The break-even point = Underlying asset (index) below the strike by
the original Purchase price of the put option.
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.

Part II
Pricing Derivatives and their Underlying Assets in a

Discrete-Time Setting
219
220
Chapter 4
OPTION PRICING: THE DISCRETE-TIME

APPROACH FOR STOCK OPTIONS
Chapter Outline
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
of American options and complex derivative securities. In fact, these

approaches can handle more complex situations such as cash dividend
payouts.
In this chapter, we are interested in the lattice approach pioneered by
Cox et al. (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 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–Scholes. Since then, the CRR approach was extended and
extensively used for the valuation of many contingent claims and options.
4.1. The CRR Model for Equity Options

Cox et al. (1979), proposed the first discrete-time model for the pricing of
stock options. Rendleman and Bartter (1980), developed a similar model
for the pricing of interest rate sensitive instruments.
4.1.1. The mono-periodic model

To illustrate the foundations of the binomial model, consider the following
data:
• Underlying asset price: S = 40;

• Strike price: K = 40;
• Riskless interest rate: r = 10% or R = 1 + r = 1.1 and
• Time to maturity: 1 year.
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
or u = (1 + 0.2) = 1.2 and d = (1 − 0.2) = 0.8
Fig. 4.1. One period binomial model.

Option Pricing: The Discrete-Time Approach for Stock Options 223
max (o, uS − K) = (0.48 − 40) = 8 = Cu

C=
max (o, dS − K) = (0.32 − 40) = 0 = Cd
Fig. 4.2.
uS − HC = 32
S − HC
dS − HC = 32
Fig. 4.3. Dynamics of the hedge portfolio.
The dynamics of the option is nearly similar to that of the underlying

asset. The call option price at the maturity date is given by the greater of
zero and the intrinsic value. As in Fig. 4.2, the option price can go up to
Cu or down to Cd .
It is possible to construct an initial hedge portfolio using the underlying
asset S and a certain number H of options as (S — HC ). If this portfolio
hedges the investor against risk, it must lead to the same result at the
maturity date as in Fig. 4.3.
The possible change in the underlying asset price at maturity is
given by:
(uS − dS) = 48 − 32 = S(u − d) = 16, or 40(1.2 − 0.8) = 16.
The possible change in the option price at maturity is given by:
(Cu − Cd ) = (8 − 0) = 0.
Using this information, it is possible to compute the number H as

follows:
H = S(u − d)/(Cu − Cd ) = 40(1.2 − 0.8)/(8 − 0) = 16/8 = 2.
When the stock price increases, the value of the hedge portfolio is:
uS − HCu = 1.2(40) − 2(8) = 48 − 16 = 32.
When the stock price decreases, the value of the hedge portfolio is:
dS − HCd = 0.8(40) − 2(0) = 32.
What is the option price at time 0 in this simple binomial model?

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)
This is the option price in a mono-periodic binomial model.
Example: Using the following data:
Cu = 8, Cd = 0, u = 1.2, d = 0.8, R = 1.1
The option price is:

(1.1 − 0.8) (1.2 − 1.1)
C= 8 +0 1.1 = (6 + 0)/1.1 = 5.4545
(1.2 − 0.8) (1.2 − 0.8)
The call price can also be written as:
C = [pCu + (1 − p)Cd ]/R
with
p = (R − d)/(u − d) and (1 − p) = (u − R)/(u − d)
where p refers to the probability of an increase in the underlying asset price.
4.1.2. The multiperiodic model

This simple mono-periodic model can be repeated N times to construct the
multiperiodic binomial option pricing model. Time to maturity T is divided
into N intervals of length ∆t, where the underlying asset price increases
from S to uS or decreases from S down to dS with a probability p and
(1 − p) as in Fig. 4.4.
p uS
(1−p) dS
Fig. 4.4. Dynamics of the underlying asset price.
√
In a risk-neutral world, the expected value of S is Ser ∆t
. The expected
value can also be calculated as follows:
pSu + (1 − p)Sd (4.1)
The equality between the two expected values gives:

√
Ser ∆t
= pSu + (1 − p)Sd (4.2)
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
Calculating the variance and simplifying gives:
σ 2 ∆t = pu2 + (1 − p)d2 − [pu + (1 − p)d]2 (4.4)
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
by its expected future value discounted to the present at the appropriate

risk-less rate. At maturity, the payoff from an European call is:
FN,j = max[0, Suj dN −j − K]
The payoff from an European put:
FN,j = max[0, K − Suj dN −j ]
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)
The value of an American put option must satisfy the following

condition:
√
Fi,j = max[K − Suj di−j , er ∆t
(pFi+1,j+1 + (1 − p)Fi+1,j )] (4.8)
This model appears in Cox et al. (1979), Cox and Rubinstein (1985),
Boyle (1986, 1988) and Hull and White (1993) etc.
4.1.3. Applications and examples

4.1.3.1. Applications of the CRR model within two periods
Consider the following data for the pricing of an European call:
S = 100, K = 100, T = 1 year, N = 2, σ = 0.2, r = 0.1.
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.
For an initial value of S = 100, the two possible values in the next
period are:
Su = 100(1.1519) = 115.19, Sd = 100(0.8681) = 86.81.
These two values of the underlying asset price lead to three possible
values, as shown in Fig. 4.5:
Suu = 115.19(1.1519) = 132.68, Sud = 115.19(0.8681) = 100

and Sdd = 86.81(0.8681) = 75.36.
Using the index representation, (i, j), we have (Fig. 4.6):

In the above representation, S0,0 refers to the initial time 0 and the
lowest position 0. S1,0 corresponds to the period 1 and the lowest position 0.
S1,1 refers to the period 1 and the first position after 0, i.e., 1. The dynamics
of the option price are given in Fig. 4.7.
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
C2,2 = 132.68 − 100 = 32.68

C1,1 = ?
C0,0 = ? C2,1 = 100 − 100 = 0
C1,0 = ?
C2,0 = 75.36 − 100 = 0
Fig. 4.7. Dynamics of the option price.

The option value can be computed by using a simple application of the

following formula:
√
Fi,j = e−r ∆t
[pFi+1,j+1 + (1 − p)Fi+1,j ].
The option value at node (1,1), i.e., C1,1 is given by:
√
C1,1 = e−r ∆t
[pC2,2 + (1 − p)C2,1 ],
or
C1,1 = e−0,1(1/2) [0.7227(32.68) + (1 − 0.7227)0] = 22.465.
The option value at node (1, 0), i.e. C1,0 is:
√
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
The option value at time 0 is 15.443. The same method is used by

Rendleman and Bartter (1979) for the valuation of interest rate options.
4.1.3.2. Other applications of the binomial model

of CRR for two periods
Consider the following data for the valuation of European and American
call and put options:
S = 100, K = 100, r = 5%, σ = 30%, N = 2, T = 1 year.

Computation of the European call price

Using the above date, the dynamics of the underlying asset are given by:
152.84
u²S
123.63
u
uS d
S
S 00 d u 100
dS
84.17 d
80.88 d ²S
65.41
The dynamics of the option price are given by:
C 22
C 11
C 21
C 00
C 10
C 20
In this case, option values are given by:
C2,2 = 52.84, C1,1 = 0, C2,0 = 0

0.5064 ∗ 52.84 + 0 ∗ (1 − 0.5064)
C1,1 = = 26.105, C1,0 = 0
1.025
0.5064 ∗ 26.105 + 0
C0,0 = = 12.897.
1.025
Hence, the option price is 12.897.

Computation of the European put price

The dynamics of the put price are given by:
P 22
P11
P 21
P 00
P10
P 20
Option prices are given by:
P2,2 = 0, P2,1 = 0, P2,0 = 100 − 65.41 = 34.59

0.5064 ∗ 0 + 0
P1,1 = =0
1.025
0.5064 ∗ 0 + 0.4936 ∗ 34.59
P1,0 = = 16.657
1.025
0 + 0.4936 ∗ 16.657
P0,0 = = 8.021
1.025
Hence, the option price is 8.021. We can check that the put–call parity
theorem is verified.
P = C + Ke−rt − S = 12.897 + 100e−0.05∗1 − 100 = 8.02
Computation of the American call price

Option prices are computed as:

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
Computation of the American put price

Option prices are computed as:

0.5064 ∗ 0 + 0.4936 ∗ 0
P1,1 = max ; 100 − 80.88 = 0
1.025

0.5064 ∗ 0 + 0.4936 ∗ 34.59
P1,0 = max ; 100 − 80.88 = 19.12
1.025
0.5064 ∗ 0 + 0.4936 ∗ 19.12
P0,0 = = 9.2
1.025
4.1.3.3. Applications of the binomial model of CRR for three periods

call and put options: S = 100, K = 100, r = 5%, σ = 30%, N = 3, T = 1
year.
Computation of the European call price

√
Using the above data, we have u = eσ ∆t
with N ∆t = T . Hence,
1 √ 1 1
∆t = = 0.33, u = e0.30 0.33
= 1.1881, d= = = 0.8417
3 u 1.1881
er∆t − d e0.05∗0.33 − 0.8417

p= = = 0.5050
u−d 1.1881 − 0.8417
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
The option payoff at maturity is:
C3,3 = 167.71 − 100 = 67.71, C3,2 = 118.81 − 100 = 18.81

C3,1 = 84.17 − 100 = 0, C3,0 = 59.63 − 100 = 0.
Before maturity, option prices are computed as:

0.5050 ∗ 67.71 + 0.495 ∗ 18.81
C2,2 = = 42.78
1.017
0.5050 ∗ 18.81 + 0.495 ∗ 0
C2,1 = = 9.34
1.017
0.5050 ∗ 0 + 0.495 ∗ 0
C2,0 = =0
1.017
0.5050 ∗ 42.78 + 0.495 ∗ 9.34
C1,1 = = 25.79
1.017
0.5050 ∗ 9.34 + 0.495 ∗ 0
C1,0 = = 4.64
1.017
0.5050 ∗ 25.79 + 0.495 ∗ 4.64
C0,0 = = 15.06
1.017
The option price is C0,0 = 15.06.
The option payoff at maturity is given by:
P3,3 = 100 − 167.71 = 0, P3,2 = 100 − 118.81 = 0

P3,1 = 100 − 84.17 = 15.83
P3,0 = 100 − 59.63 = 40.37
Before maturity, option prices are given by:

0.5050 ∗ 0 + 0.495 ∗ 0
P2,2 = =0
1.017
0.5050 ∗ 0 + 0.495 ∗ 15.83
P2,1 = = 7.70
1.017
0.5050 ∗ 15.83 + 0.495 ∗ 40.37
P2,0 = = 27.51
1.017
0.5050 ∗ 0 + 0.495 ∗ 7.70
P1,1 = = 3.75
1.017
0.5050 ∗ 7.70 + 0.495 ∗ 27.51
P1,0 = = 17.21
1.017
0.5050 ∗ 3.75 + 0.495 ∗ 17.21
P0,0 = = 10.24
1.017
We can check the put–call parity relationship in this context:
P = C + Ke−rt − S = 15.06 + 100e−0.05∗1 − 100 = 10.2
Computation of the American call price

American call option prices are computed as follows:

0.5050 ∗ 27.71 + 0.495 ∗ 18.81
C2,2 = max ; 141.16 − 100 = 42.78
1.017

0.5050 ∗ 18.81 + 0.495 ∗ 0
C2,1 = max ; 100 − 100 = 9.34
1.017

0.5050 ∗ 0 + 0.495 ∗ 0
C2,0 = max ; 0 =0
1.017

0.5050 ∗ 42.78 + 0.495 ∗ 9.34
C1,1 = max ; 118.8 − 100 = 25.79
1.017

0.5050 ∗ 9.34 + 0.495 ∗ 0
C1,0 = max ; 84.17 − 100 = 4.64
1.017
0.5050 ∗ 25.79 + 0.495 ∗ 4.64
C0,0 = = 15.06
1.017
Computation of the American put price

American put prices are computed as follows:

0.5050 ∗ 0 + 0.495 ∗ 0
P2,2 = max ; 100 − 141.16 = 0
1.017

0.5050 ∗ 0 + 0.495 ∗ 15.83
P2,1 = max ; 100 − 100 = 7.70
1.017

0.5050 ∗ 15.83 + 0.495 ∗ 40.37
P2,0 = max ; 100 − 70.85 = 29.15
1.017

0.5050 ∗ 0 + 0.495 ∗ 7.70
P1,1 = max ; 100 − 118.8 = 3.75
1.017

0.5050 ∗ 7.70 + 0.495 ∗ 29.15
P1,0 = max ; 100 − 84.17 = 18.01
1.017
0.5050 ∗ 3.75 + 0.495 ∗ 18.01
P0,0 = = 10.63
1.017
4.1.3.4. Examples with five periods

Example 1. Consider the valuation of European and American options in
the following context:
Underlying asset, S = 100, strike price K = 100, interest rate = 0.1,
volatility = 0.4, T = 5 months, and N = 5. In this case, we have p = 0.5073,
d = 0.8909, and u = 1.1224.
Dynamics of the underlying asset price
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
The valuation of European put options
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
The valuation of American put options
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
The valuation of European call options
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
The valuation of American call options
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

volatility = 0.4, T = 5 months, and N = 5. In this case, we have p = 0.5073,
d = 0.8909, and u = 1.1224.
Dynamics of the underlying asset price
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
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
Valuation of American put options
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
4.2. The Binomial Model and the Distributions

to the Underlying Assets
Let us denote by:
C(.), c(.): prices of American and European call options and
P (.), p(.): prices of American and European put options.
4.2.1. The Put-Call parity in the presence of several

cash-distributions
The traditional put–call parity can be easily adapted to account for the
cash income paid (or received) at the end of each month on the underlying
stock. Consider the two following portfolios:
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.
At the maturity date, both portfolios are worth max(K, ST ). If the

options are European, the portfolios must have identical values at time
t : [c + Ke−r(T −t) = p + S].
The two most basic forms of arbitrage are the conversion and the reverse
conversion or reversal. Since the value of the stock must be discounted by
the amount of the cash received, R, or paid, D, the conversion ought to net
a profit when:
c − p + (− call parity or + put parity) > R − D
The reverse is profitable when:
c − p − (− call parity or + call parity) < R − D
These relationships must hold independently from the option-valuation

model. To prevent risk-less free arbitrage in efficient markets, we must have:
c − p ± (put–call parity) = R − D
4.2.2. Early exercise of American stock options

The argument of Merton (1973) and Jarrow and Rudd (1983) can be
extended to early exercise of American calls when there are several cash
payments at different dates ti. When the amounts of cash distributions
are accounted for, it can be shown that a sufficient condition for optimal
exercise is:
(Di + Ri ) > K[1 − e−r(ti+1−ti) ]
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:
(Di + Ri ) < K[1 − e−r(ti+1−ti) ]
with (Di + Ri ) ≥ 0.
4.2.3. The model

When there is just one ex-cash income date τ , during the option’s life and
k ∆ t ≤ τ ≤ (k + 1)∆t, then at time x, the value of the random component
S is:
S ∗ (x) = S(x) when x > τ

S ∗ (x) = S(x) − (Di + Ri )e−r(τ −x) when x ≤ τ
Assume σ∗ is the constant volatility of S ∗ . Using the parameters p, u,

and d, at time t + i∆t, the nodes on the tree define the stock prices:
If i∆t < τ : S ∗ (t)uj di−j + (Di + Ri )e−r(τ −i∆t) j = 0, 1, . . . , i

∗ j i−j
If i∆t ≥ τ : S (t)u d j = 0, 1, . . . , i
4.2.4. Simulations for a small number of periods

Example 1. Applications of the CRR model for five periods with dividends
call and put options: S = 110, K = 115, r = 10%, σ = 40%, N = 0.5,
t = 5 months, ∆t = 1 month. date of dividend: 105 days, dividend amount:
D = 10.
The European call price

Using the above data, the dynamics of the underlying asset are computed
using:
√ √
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
S0,0 = S ∗ u0 d0 + De−r(105/365) = 109.7164
S1,1 = S ∗ u1 d0 + De−r((105/365)−(1/12))
= 100(1.1224) + 10De−r((105/365)−(1/12)) = 122.0377
S1,0 = S ∗ u0 d1 + De−r((105/365)−(1/12)) = 98.8925
S2,2 = S ∗ u2 d0 + De−r((105/365)−(2/12)) = 135.8579
S2,1 = S ∗ u1 d1 + De−r((105/365)−(2/12)) = 109.8797
S2,0 = S ∗ u0 d2 + De−r((105/365)−(2/12)) = 89.2586
S3,3 = S ∗ u3 d0 + De−r((105/365)−(3/12)) = 151.3603
S3,2 = S ∗ u2 d1 + De−r((105/365)−(3/12)) = 122.2024
S3,1 = S ∗ u1 d2 + De−r((105/365)−(3/12)) = 99.0572
S3,0 = S ∗ u0 d3 + De−r((105/365)−(3/12)) = 80.6848
S4,4 = S ∗ u4 d0 = 158.7050, S4,3 = S ∗ u3 d1 = 125.782
S4,2 = S ∗ u2 d2 = 100.00, S4,1 = S ∗ u1 d3 = 79.3788
S4,0 = S ∗ u0 d4 = 63.0100, S5,5 = S ∗ u5 d0 = 178.1305
S5,4 = S ∗ u4 d1 = 141.3979, S5,3 = S ∗ u3 d2 = 112.2400
S5,2 = S ∗ u2 d3 = 89.0948, S5,1 = S ∗ u1 d4 = 70.7224
S5,0 = S ∗ u0 d5 = 56.1386
The dynamics of the underlying asset are represented in the follow-

ing way.
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
The option’s maturity value is computed as:
C5,5 = max[0; S5,5 − K] = 63.1305, C5,4 = max[0; S5,4 − K] = 26.3979

C5,3 = max[0; S5,3 − K] = 0, C5,2 = max[0; S5,2 − K] = 0
C5,1 = max[0; S5,1 − K] = 0, C5,0 = max[0; S5,0 − K] = 0
The American call option price is computed as:

p · C5,5 + q · C5,4
C4,4 = max ; V I
er∆t
= max[44.6586; S4,4 − K] = max[44.6586; 43.7050]
= 44.6586 where VI stands for the option intrinsic value.

p · C5,4 + q · C5,3
C4,3 = max ; S4,3 − K
er∆t
= max[13.2805; 10.9782] = 13.2805

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
American call option prices are computed as follows:
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

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
The American put price

Using the above data, the American put price is computed as follows at
different nodes.
P5,5 = max[0; K − S5,5 ] = 0, P5,4 = max[0; K − S5,4 ] = 0,

P5,3 = max[0; K − S5,3 ] = 2.76
P5,2 = max[0; K − S5,2 ] = 25.9052, P5,1 = max[0; K − S5,1 ] = 44.2776
P5,0 = max[0; K − S5,0 ] = 58.8614
Before maturity, the option price is computed as:

p · P5,5 + q · P5,4
P4,4 = max ; max[0; K − S4,4 ] = 0
er∆t

p · P5,4 + q · P5,3
P4,3 = max ; max[0; K − S4,3 ]
er∆t
= max[1.3486; 0] = 1.3486

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
Option prices are reported in the following figure.
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

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


volatility = 0.4, T = 5 months, N = 5, dividend = 10, and dividend
date = 105. In this case, we have:
p = 0.5073, d = 0.8909, and u = 1.1224.

Dynamics of the underlying asset for five periods
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
The valuation of American put options
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
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
4.2.5. Simulations in the presence of two dividend dates

We consider the valuation of a stock option using the binomial model.
The option is priced as of 15/06/2002. The maturity date is 15/06/2004.
The following strike prices are used: 22, 23, 24, 25, 26, 27, 28, 29, 30. The
following dates and amounts of dividends are available:
For 2003, 0.35 and for 2004, 0.35. We used a historical simulation to
estimate the volatility parameter. Dividends are distributed at the end
of May each year. The interest rate is 5%. The annualized volatility is
between 45% and 50%. In this analysis, a dividend rate is used by dividing
the dividend amount by the initial underlying asset price. The annualized
volatility is 45% (Tables 4.1–4.4).
We consider the same valuation problem, except that the volatility
used is 50%.
4.2.6. Simulations for different periods and several

dividends: The general case
Tables 4.5 and 4.6 present the simulation results of the model proposed
for American long-term call and put values by taking into account the
magnitude of cash distributions and their timing. Simulations are run as
of 11/05/2002 for a stock price of 423. Each share entitles the holder on
05/06/2002, a net dividend of 12.
The following dates and amounts for the cash distributions are
retained: 24/05/2002: −3.08, 24/06/2002: −3.08, 24/07/2002: −2.99,
24/08/2002: −2.98, 24/09/2002: −2.99, 24/10/2002: −2.82, 24/11/2002:
−2.82, 24/12/2002: −2.82.
Table 4.1. Simulations in the presence of two

dividends using the binomial model.
Call price S = 18 S = 19 S = 20 S = 21
K = 22 3.49 4.09 4.71 5.33

K = 23 3.25 3.76 4.38 5.00
K = 24 3.03 3.50 4.05 4.67
K = 25 2.82 3.29 3.76 4.33
K = 26 2.60 3.07 3.55 4.03
K = 27 2.38 2.86 3.33 3.80
K = 28 2.19 2.64 3.11 3.58
K = 29 2.06 2.42 2.89 3.37
K = 30 1.93 2.26 2.68 3.15

Call price S = 22 S = 25 S = 28 S = 30
K = 22 5.95 8.13 10.38 12.05

K = 23 5.61 7.67 9.90 11.46
K = 24 5.28 7.21 9.44 10.94
K = 25 4.95 6.80 8.98 10.47
K = 26 4.62 6.47 8.52 10.01
K = 27 4.29 6.14 8.06 9.55
K = 28 4.06 5.81 7.66 9.09
K = 29 3.84 5.47 7.33 8.63
K = 30 3.62 5.14 6.99 8.23

Call price S = 18 S = 19 S = 20 S = 21
K = 22 3.99 4.61 5.23 5.86

K = 23 3.71 4.30 4.92 5.54
K = 24 3.51 4.00 4.60 5.23
K = 25 3.30 3.78 4.29 4.92
K = 26 3.10 3.58 4.06 4.60
K = 27 2.90 3.38 3.86 4.34
K = 28 2.70 3.18 3.66 4.14
K = 29 2.49 2.97 3.45 3.93
K = 30 2.36 2.77 3.25 3.73

S = 22 S = 25 S = 28 S = 30
K = 22 6.48 8.68 10.94 12.56

K = 23 6.17 8.24 10.50 12.01
K = 24 5.85 7.80 10.06 11.56
K = 25 5.54 7.41 9.61 11.12
K = 26 5.23 7.10 9.17 10.68
K = 27 4.90 6.79 8.73 10.23
K = 28 4.60 6.47 8.34 9.79
K = 29 4.42 6.16 8.03 9.35
K = 30 4.21 5.84 7.72 8.96
These amounts correspond to interest rates of 8.75% for options

maturing in June, 8.5% for options maturing in September, and 8% for
the maturity date of December. For example, when the stock price equals
423, the interest rate is 8.75%, the cash amount is 3.08.
Cox et al. (1979) indicates the prices obtained from a modified
version of the CRR model. The following dates and amounts for the
cash distributions are retained: 24/05/2002: −3.08, 24/06/2002: −3.08,
24/07/2002: −2.99, 24/08/2002: −2.98, 24/09/2002: −2.99, 24/10/2002:
−2.82, 24/11/2002: −2.82, 24/12/2002: −2.82.
Table 4.5 shows the American call prices for different strike prices
varying from 360 to 460. The volatility parameter takes two values 24.4%
and 30.4%. Table 4.6 presents the American put prices for the same
parameters as in Table 4.5. The number of iterations used for the CRR
Table 4.5. Simulations of the American long-term

equity call values: S = 423, K = 360 to 460, r = 0.08,
D = 12, T = 233 days, number of iterations: 233, and
dividend date: 25 days.
CRR
K σ = 0.244 σ = 0.304
360 86.11 94.01

380 70.88 80.70
400 57.22 68.58
420 45.29 57.91
440 35.14 48.50
460 26.82 40.29
Table 4.6. Simulations of the American long-term

equity option values.
CRR
K σ = 0.244 σ = 0.304
360 5.52 13.71

380 9.62 19.74
400 15.50 27.12
420 23.42 36.10
440 33.49 46.51
460 45.77 58.33
model corresponds to the number of days to maturity (233). The dividend

is paid in 25 days.
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?
Appendix: The Lattice Approach

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 N equal intervals of
length ∆t during which the stock price moves from its initial value S to
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:
Fi,j = max[Suj d i−j − K, e−r∆t(pFi+1,j+1 + (1 − p)Fi+1,j )]
The American put is given by:
Fi,j = max[K − Suj di−j , e−r∆t(pFi+1,j+1 + (1 − p)Fi+1,j )]
The extension of the lattice approach to the valuation of American

options on stocks paying a known cash income is as follows. When there is
only one cash income at date, τ , between k∆t and (k + 1)∆t, it is possible
to design trees where the number of nodes at time ∆t is always (i + 1).
The analysis which parallels that in Hull (2000) and Briys et al. (1998)
can be simplified 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.
Exercises
Example 1
Consider the valuation of European and American options in the following
context:

volatility = 0.2, T = 5 months, and N = 5. In this case, we have: p =
0.5217, d = 0.9439, and u = 1.0594.
Dynamics of the underlying asset for five periods 133.4658
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
Valuation of an European put option
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
Valuation of an American put option
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
Valuation of an European call option
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
Valuation of an American call option
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:
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.
Dynamics of the underlying asset for five periods 133.4658
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
Valuation of European put options
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
Valuation of American put options
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
Valuation of European call options
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
Valuation of American call options
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
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
100 100 0,05 0,25 20%

d1 d2 N(d1) DELTA N(d2)
0,175000 0,075000 0,569460 0,529893
Put Price Call Price C−P S−Kexp(−rt)
3,3728 4,61499713 1,242220 1,242220
At the money call (Out of the Money) put options using the Black–Scholes
Model.
120 100 0,05 0,25 20%

d1 d2 N(d1) N(d2)
1,998216 1,898216 0,977153 0,971166
21,242220 21,242220
0,1060 21,34818885
Out of the money call (In the Money) put options using the Black–Scholes
Model.
80 100 0,05 0,25 20%

d1 d2 N(d1) N(d2)
−2,056436 −2,156436 0,019870 0,015525
−18,757780 −18,757780
18,8142 0,056423801
Implementing Monte Carlo method for a stock:

Computing return, volatility, generating random numbers, assuming an
initial portfolio value of 100,000, percentile is 99,816 and Value at Risk
is 184.

1
S(t + h) = S(t) exp µ − σ h + σε (h)
2
September 10, 2009 14:41
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
Option Pricing: The Discrete-Time Approach for Stock Options

11/07/2007 92,30 0,00 0,0062 0,00 −1,32735 92,038 100058 57,8275
12/07/2007 92,00 −0,0033 0,70690 92,1636 99921,5 −78,5194
13/07/2007 92,00 0,00 V0 0,51740 92,1519 99934,2 −65,8259
16/07/2007 92,00 0,00 100000,00 0,02069 92,1212 99967,5 −32,5468
17/07/2007 92,00 0,00 −0,33920 92,099 99991,6 −8,4271
18/07/2007 92,00 0,00 Skewness 0,0692 1,53959 92,2151 99865,7 −134,2772
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
256
92,12478 99816 184,1338
Last Trade
Xt: random T=3
20/08/2007 92,00 0,00 0,82779 92,1711 99913,4 −86,6164

22/08/2007 92,00 0,00 −2,78022 91,9484 100155 155,3205
23/08/2007 92,00 0,00 0,35179 92,1417 99945,3 −54,7312
Derivatives, Risk Management and Value

24/08/2007 92,20 0,0022 −0,01442 92,1191 99969,8 −30,1938
27/08/2007 93,80 0,0174 0,16635 92,1302 99957,7 −42,3065
28/08/2007 94,00 0,0021 −1,04287 92,0556 100039 38,7491
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
92,12478 99816 184,1338
Last Trade
Xt: random T=3
21/09/2007 92,16 0,0017 −0,16898 92,1095 99980,2 −19,8356
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
References
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,
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.
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.
Chapter 5
CREDIT RISKS, PRICING BONDS, INTEREST

RATE INSTRUMENTS, AND THE TERM
STRUCTURE OF INTEREST RATES
Chapter Outline
1. Section 5.1 is an introduction to the main concepts in discounting and

factors.
2. Section 5.2 develops the main concepts for the pricing of bonds.
3. Section 5.3 presents some simple measures to calculate the yield on
bonds.
4. Section 5.4 develops the main concepts in the analysis of bonds:
duration and convexity.
5. Section 5.5 is an introduction to the yield curve and the theories of
interest rates.
6. Section 5.6 presents a simple analysis of the yield to maturity and the
theories of the term structure of interest rates.
7. Section 5.7 reviews the specific features of spot and forward interest
rates.
8. Section 5.8 studies the way bonds are issued and redeemed.
9. Section 5.9 presents a simple analysis of mortgage-backed securities.
10. Section 5.10 is an introduction to swaps.
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
discounted to the present at the appropriate discount rate. The yield on

a bond can be determined by calculating the interest rate that makes the
present value of all future cash flows equal to the initial bond price. The
management of a portfolio of bonds needs the knowledge of the main yield
measures: the current yield (CY), yield to maturity (YTM), and the yield
to call (YTC).
The duration of a bond is a measure of the bond price volatility.
The concept of duration is introduced by Macaulay (1938) as a proxy
for the length of time a bond investment is outstanding. The yield
curve corresponds to the graphical relationship between the YTM of
a security and the corresponding maturity date. This curve is often
constructed using the maturities and the observed yield on Treasury
securities.
An interest rate swap is an agreement between two counter-parties to
exchange periodic interest payments. These payments are computed with
reference to a pre-determined amount known as the notional principal
amount. In general, one party, the fixed rate payer, agrees to pay the
other party fixed-interest payments with a given frequency at some specified
dates. The other party, the floating rate payer, agrees to pay some interest
rate payments that vary according to a reference rate. The London Inter-
bank Offered Rate (LIBOR), is often used as the reference rate. The risk
that one of the parties does not respect his/her obligation in the swap
agreement refers to the default risk or the counter-party risk.
The credit crunch in 2008 was associated with debt and bonds.
5.1. Time Value of Money and the Mathematics of Bonds

We use the following symbols:
I: value or sum of money at the present (a present sum);

F : value or sum of money in the future (future sum);
A: series of equal end-of-period amounts of money (a uniform series);
n: number of periods and
r: interest rate per interest period.
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).
Credit Risks, Pricing Bonds, Interest Rate Instruments, and Term Structure 261
5.1.1. Single payment formulas

Simple interest is computed using the principal amount by ignoring interest
accrued in previous interest periods. Simple interest is computed using the
formula: Interest = I(n)r. Example: When I = 100, r = 10%, and n = 1
year, the formula shows an interest of 10, or 100(0,1)(1).
The compound interest is used when more than one interest period is
used.
When an amount I is invested at time 0, it becomes in one period (a
year), F1 or:
F1 = I + rI = I(1 + r)
At the end of the second period, the amount accumulated corresponds to

the amount accumulated after the first period plus the interest from the
end of period 1 to the end of period 2, or:
F2 = F1 + F1 r = I(1 + r) + I(1 + r)r = I(1 + 2r + r2 ) = I(1 + r)2
It is straight forward to generalize this formula for n periods as:
Fn = I(1 + r)n
or
F = I(1 + r)n (5.1)
The term (1 + r)n is often referred to as the single-payment compound-

amount factor (SPCAF). It is denoted by (F/I, r%, n). It provides the
future value F of an initial investment I, after n periods (years) for a given
interest rate r. Formula (5.1) can be used to generate the present worth I
of a future amount F after n years at the rate r as:
1
I=F (5.2)
(1 + r)n
The term 1/(1 + r)n is referred to as the single-payment present worth

factor (SPPWF). It is denoted by (I/F, r%, n). When there is a single
payment (or receipt), the formulas allow the derivation of future and present
values.
5.1.2. Uniform-series present worth factor (USPWF)

and the capital recovery factor (CRF)
The present worth of uniform series can be computed by assimilating each
given A as a future worth F in formula (5.2). The present value is:
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
Multiplying both sides by (1/(1 + r)) gives:

I 1 1 1 1
=A + + + · · · +
(1 + r) (1 + r)2 (1 + r)3 (1 + r)4 (1 + r)n+1
The difference between the last two equations gives:

1 1 1
I −1 =A −
1+r (1 + r)n+1 (1 + r)
or:

−r 1 1
I =A −1
1+r (1 + r)n 1+r
Dividing by −r/(1 + r) yields:

n
((1 + r)n − 1) 1
I=A =A (5.3)
r(1 + r)n (1 + r)i
i=1
The term or factor:

n
1 (1 + r)n − 1 1 − (1 + r)−n
= =
(1 + r)i r(1 + r) n r
i=1
is the USPWF. It is denoted by (I/A, r%, n). It allows the computation of

the present worth I of an equivalent uniform annual series A that starts at
the end of the first period (year 1), for n years.
Equation (5.3) can be written as:

n
r(1 + r)n r
A=I = I (1 + r)i = (5.4)
((1 + r)n − 1) 1 − (1 + r)−n
i=1

r(1+r)n
The term ((1+r)n −1 corresponds to the capital-recovery factor (CRF).
It is denoted by (A/I, r%, n). It provides the equivalent uniform annual
worth A over n years of a given amount I at the rate r.
5.1.3. Uniform-series compound-amount factor (USCAF )

and the sinking fund factor (SFF )
Re-call that the present worth is given by:
1
I=F
(1 + r)n
If I is substituted in Eq. (5.4), this yields:

1 r(1 + r)n
A=F
(1 + r)n ((1 + r)n − 1)
or:
n
(1 + r)i r
A=F =F −1 (5.5)
i=1
(1 + r)n (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:
((1 + r)n − 1) (1 + r)n

F =A = A
n (5.6)
r i=1 (1 + r)
i
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.
Table 5.1. Computations with reference to standard notations.
To find (a) Appropriate Formula

for a given (b) factor Equation (5.1) to (5.6)
I, F Simple (I/F, r, n): SPPWF Formula

h (5.1): i
1
discounting I = F (1+r)n
h i
1
(1+r)n
F, I Simple (F/I, r, n): SPCAF Formula (5.2):

capitalization (1 + r)n F = I(1 + r)n
I, A Factor (I/A, r, n): USPWF Formula (5.3):
h i
((1+r)n −1)
providing I I =A r(1+r)n
h i
1−(1+r)−n
as a function =A r
of an equivalent
annuity.
It is denoted
sometimes by
P
I =A n 1
i=1 (1+r)i
A, I Factor of an annuity (A/I, r, n): CRF Formula
h (5.4): n i
r(1+r)
equivalent to I A = I (1+r)n −1)
Pn h i
r
i=1 (1 + r)i = J 1−(1+r) −n
A, F SFF (A/F, r, n): SFF Formula (5.5):

Pn h i
(1+r)i r
i=1 (1+r)n A=F ((1+r)n −1)
F, A Factor providing (F/A, r, n): USCAF Formula

h (5.6):n i
((1+r) −1)
the future value F =A r
of a constant annuity
(1+r)n
Pn i
i=1 (1+r)
Table 5.1 provides the necessary formulas.
Applications
Using Eq. (5.1), when r = 6%, n = 8 years, formula (5.1) gives the factor:
(I/F, 6%, 8) = 1/(1 + 0.06)8 = 0.6274126
When F = 1000, the initial value I is 627.4126.

Using Eq. (5.3), when r = 6%(8%, 11%), n = 11(20, 14) years,

formula (5.3) gives the value of the uniform series A:
(I/A, 6%, 11) = (1 + 0, 06)11 − 1/0.06(1 + 0, 06)1 1 = 7.886869
(I/A, 8%, 20) = (1.08)2 0 − 1/0, 08(1, 08)20 = 9.81815
(I/A, 11%, 14) = (1.11)1 4 − 1/0, 11(1, 11)14 = 6.981866.
Using Eqs. (5.4), (5.5), (5.6), we obtain:
(A/I, 6%, 5) = 0, 06(1, 06)5/(1, 06)5 − 1 = 0, 2373966
(A/F, 6%, 5) = 0, 06/(1, 06)5 − 1 = 0, 1773966
(F/A, 6%, 5) = (1, 06)5 − 1/0, 06 = 5, 637087.
5.1.4. Nominal interest rates and continuous compounding

A nominal interest rate is the usual interest rate which accounts for the
effects of inflation. In fact, the real rate plus inflation define the nominal
interest rate. When an individual deposits an amount of 1000 in a bank
account, for an interest rate of 12% per year (compounded annually), the
future worth is:
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
can be approximated by: r + 1 = (1 + mim ) or r = m(im ) where r is the

effective global interest rate. When the number of compounding periods m
increases, m approaches infinity, and we must use continuous compounding.
Re-call the definition of the natural logarithm base e:
Lim (1 + 1/h)h = e
h→∞
When im = (i/m) = (1/h), (m = hi) in Eq. (5.1), we have:
Lim r = Lim (1 + i/m)m − 1 = Lim [(1 + 1/h)h ]i − 1

m→∞ m→∞ m→∞
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).
5.2. Pricing Bonds

5.2.1. A coupon-paying bond
The price of a bond depends on the future coupons and the notional amount.
The price of a bond is given by the present value of all the future cash
flows discounted to the present as follows:

n
c M
B= +
t=1
(1 + r)t (1 + r)n
where:
B: bond price at time 0;

c: coupon payments;
r: periodic interest rate;
M : par or maturity value of the bond and
n: number of periods.
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
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
The interest rate used in the discounting procedure is referred to as the

required yield. The following example illustrates the use of the formula in
the computation of the bond price.
Example: Consider the pricing of a bond in the following context.
Time to maturity = 10 years, coupon rate = 10%, maturity value of the
bond = 1000, and periodicity of coupons = 20. The annual coupon is 100
or 10% (1000). The semi-annual coupon is 50 or (100/2). Since there are 20
coupons and M is 1000, we need the required yield to compute the bond
price. If the yield on a comparable bond (with the same characteristics and
risk) is 11%, then the required yield for six months is 5.5% or (11%/2).
Using this discount rate, the present value of the coupons is 597,519 or:
c({1 − [1/(1 + r)n ]}/r) = 50({1 − [1/(1 + 0,055)20 ]}/0,055)
The present value of the nominal amount is 342,298 or 1000/(1,055)20.

Hence, the bond price is 940,247 or (597,519 + 342,298).
5.2.2. Zero-coupon bonds

A zero-coupon bond pays no periodic coupons. However, the investor gains
interest from the difference between the purchase price and the maturity
value. Using zero coupons, the bond pricing formula becomes:
M
B=
(1 + r)n
It shows that the zero-coupon bond price is given by the discounting of its
final value to the present.
5.3. Computation of the Yield or the Internal

Rate of Return
5.3.1. How to measure the yield
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. It is calculated from the following equality:

n
CFt
B=
t=1
(1 + y)t
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.
Example: Consider a financial instrument which offers the following annual

payments in Table 5.2.
Suppose that the price of this instrument is 931. What is the yield or
the internal rate of return on this instrument?
If we use an interest rate of 10%, we have:
120(3,169) + 1000(0,6209) = 1001,18 Euros.
If we use an interest rate of 12%, we get:
120(3,0373) + 1000(0,5674) = 931 Euros.
This computation is done with reference to the following identity:

n
1 (1 − (1 + y)−n )
P = =
t=1
(1 + y)t y
Table 5.2. The promised annual payments for a financial

instrument.
Number of years from now Promised annual payments
1 120
2 120
3 120
4 120
5 1000
Hence, the internal rate of return, y is 12%. In general, people annualize

interest rates multiplying by the frequency of payments per year. If, for
example, the semi-annual interest rate is 4%, then the annual interest rate
is 8% and vice versa. This is not always correct because interest can also
be earned when compounding the interest. To obtain the effective annual
yield, which is associated with a given periodic interest rate, the following
formula can be used:
Effective annual yield = (1 + periodic interest rate)m − 1
where m stands for the frequency of payments each per year.
Periodic interest rate = (1 + effective annual yield)1/m − 1
The management of a portfolio of bonds requires the knowledge of the

main yield measures: CY, YTM, and YTC.
Very high yield bonds are regarded as junk bonds.
5.3.2. The CY
The CY gives an indication of the relation between the annual coupon
interest and the market bond price as:
CY = Annual coupon payments/market bond price
5.3.3. The YTM

The YTM is calculated in the same way as the internal rate of return for
an investor holding the bond until maturity.
For a semi-annual coupon paying bond, the YTM is obtained by solving
the following equation:

n
C M
B= +
t=1
(1 + y)t (1 + y)n
or:

{1 − [1/(1 + y)n ]} M
B=C +
y (1 + y)n
where B is the bond price, C is the semi-annual coupon, y is half of the
YTM, M is the maturity value of the bond and n the number of periods.
For a semi-annual coupon paying bond, it is sufficient to double the
interest rate or the discount rate to get the YTM. The YTM accounts for
the current coupons, their timing and the capital gain or loss from holding
the bond until maturity.
5.3.4. The YTC

Some bonds have embedded options. These options allow the issuer to call
the bond at some specified points in time. The cash flows for the calculation
of the yield to call correspond to those appearing before the first call date.
Hence, the yield to call can be calculated as the interest rate that would
make the present value of the cash flows, if the bond is held until the first
call date equal the bond price. The following formula is used to calculate
the YTC:

n
C CP
B= +
t=1
(1 + y)t (1 + y)n
where
B: the bond price;

y: one-half the yield to call;
c: semiannual coupon interest;
n: the number of periods until the first call and
CP : the call price used to redeem the bond.
5.3.5. The potential yield from holding bonds

Investors can calculate the potential yield from holding a bond. This is done
using the potential sources of dollar return which must be converted in a
yield measure. An investor holding a bond portfolio can expect to receive a
given return from the coupons perceived, the capital gain or loss and from
the re-investment of the periodic coupons.
The coupons received by the investor can be re-invested at a given rate,
giving rise to a dollar return from coupon interest and interest on interest.
This component of return can be calculated using the formula for the
future value of an annuity.
The interest on interest can be an important source of potential return
for an investor holding the bond.
Coupon interest and interest on interest

= Coupons + interest on interest
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:
Total coupon amount = nC
The formula for Ip corresponds to the future value of an annuity where

C corresponds to the amount of the annuity, r is the annual interest rate
divided by the number of payments per year and n corresponds to the
number of periods.
It is possible to compute just the interest on interest by subtracting
from the Ip formula the amount of the total coupon interest as:
C[(1 + r)n − 1]
Interest on interest = − nC
r
This analysis assumes that re-investment of the coupons is done at
the YTM.
5.4. Price Volatility Measures: Duration and Convexity

5.4.1. Duration
The duration of a bond is a measure of the bond price volatility.
The concept of duration is introduced as a proxy for the length of time,
a bond investment is outstanding. It is given by the weighted average term-
to-maturity of the cash flows of a bond. It is often computed using the first
derivative of the bond price with respect to the yield y. Re-call that the
bond price is given by the following relationship:
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
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
Each coupon payment date (including the principal) is multiplied by

the present value of the cash flow in period t. The formula for the bond
duration multiplies the present value of each cash flow in period t by the
period, when the cash flow is expected to be received. The resulting amount
is divided by the total present value of the cash flow of the bond using the
prevailing yield to maturity. The formula for the duration can be written as:
1c 2c
(1+y)1
+ (1+y)2
+ ···+ nc
(1+y)n
+ nM
(1+y)n
Dm =
B
or:

n tc nM
t=1 (1+y)t + (1+y)n
Dm =
B
When the Macaulay duration (in periods) is divided by the number

of payments per year (k = 2 for semi-annual-pay bonds), this gives the
Macaulay duration in years:
Macaulay duration (in years) = Macaulay duration (in periods)/k

5.4.2. Duration of a bond portfolio

A portfolio duration can be computed using the weighted average of the
duration of the bonds in the portfolio as:
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.
5.4.3. Modified duration

Substituting for Macaulay duration for a bond gives:

dB 1 1
=− (Macaulay duration) (5.8)
dy B (1 + y)
This relationship allows the definition of the Modified duration:
Macaulay duration
Modified duration =
(1 + y)
where y is one-half the YTM.
Replacing this expression in the previous relationship gives:

dB 1
= −Modified duration
dy B
It is possible to show that the approximate percentage change in price is
given by:
= −(1/(1 + y)) × Macaulay duration × Yield change
= −Modified duration × Yield change

There is a relationship between duration and the bond price volatility.

In fact, using a Taylor series of the price function, it is possible to show
that:
Approximate percentage change in price

= −(1/(1 + y)) × Macaulay duration × Yield change
where y corresponds to one-half the YTM.

The modified duration is given by:
Modified duration = Macaulay duration/(1 + y)
which can also be written as:
Approximate percentage change in price

= −Modified duration × Yield change
The modified duration can be used to approximate the percentage

change in the bond price per basis point change.
The dollar price change per 100 dollars of par value can be computed as:
Approximate dollar price change

= −(Modified duration) × (Initial price) × (Yield change)
The dollar duration is given by:
Dollar duration = (Modified duration) (Initial price)
So, we have:
Approximate dollar price change = −(Dollar duration) (Yield change)
5.4.4. Price volatility measures: Convexity

It is possible to use a Taylor series to approximate the price change of a
bond for a given change in the required yield by the following relationship:
dB d2 B
dB = dy + 0.5 2 (dy 2 ) +
dy dy
where is an error term. When both sides of this equality are divided by
the bond price, this allows to express the percentage price change as:

dB 1 d2 B 2 1
dB = dy + 0.5 2 (dy ) +
dy B dy B B
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
The approximate change in the bond price due to convexity is:
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
This refers also to convexity.

The percentage price change for the bond, which results from its
convexity can be written as:
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).
5.5. The Yield Curve and the Theories of Interest Rates

The yield curve corresponds to the graphical relationship between the
YTM of a security and the corresponding maturity date. This curve is
often constructed using the maturities and the observed yield on Treasury
securities.
5.5.1. The shapes of the yield curve

Different shapes of the yield curve are observed over time.
When the yield increases with maturity, the relationship corresponds
to an upward sloping or normal yield curve.
When the yield decreases with maturity, the relationship corresponds
to a downward sloping or an inverted yield curve.
When the yield increases and then decreases with maturity, this
pattern indicates a humped yield curve. When the yield is constant for
all maturities, the yield curve is flat (see Capie, 1991).
5.5.2. Theories of the term structure of interest rates

5.5.2.1. The pure expectations theory
This theory asserts that the entire term structure is explained by the
market’s expectations of the future short-term rates of different maturities.
Hence, a rising term structure can be explained by the fact that the market
participants expect a rise in short-term rates in the future. An inverted
term structure can be explained by the fact that the market participants
expect a decline in short-term rates in the future. A flat term structure can
be explained by the fact that the market participants expect constant and
stable short-term rates in the future.
The pure expectations theory is based on the assumption that
forward rates are affected systematically by the expected future short-term
rates. The main drawback of the pure expectations theory is that it does
not account for the risks inherent in investing in bonds. By assuming that
forward rates are perfect predictors of future interest rates, then the bond
future prices would be certain. It ignores the uncertainty about the bond’s
price at the end of the investment horizon as well as the re-investment risk.
The first interpretation of the pure expectations theory, according
to Cox et al. (1985a, b), is that the expected return for any investment
horizon is the same regardless of the maturity. The second interpretation
given by the local expectations theory assumes that the return over
very short-term intervals of time must be equal to the short-term risk-
less rate of interest. The third interpretation of the pure expectations
theory is that the return from a roll-over short-term strategy is equiv-
alent to holding a zero-coupon bond with a maturity equal to that of
the investment horizon. This is referred to as the return-to-maturity
expectations.
Biased expectations theories

The liquidity theory and the preferred habitat theory, referred to
as biased expectations theories, assume that other factors affect the
forward rates.
The liquidity theory: According to this theory, forward interest
rates account for interest rate expectations and a liquidity premium or
a risk premium. Investors expect to hold bonds for longer maturity dates
for higher risk premiums. Hence, the presence of a risk premium implies
that the implied forward rates will not be an unbiased estimate of the
expectations of future interest rates. The shape of an upward sloping of
the yield curve can reflect expectations about a rise, a stable or a fall in
the expected future rates, where the liquidity premium can increase fast
enough to conserve an upward-sloping yield curve.
The preferred habitat theory: This theory shows that the expec-
tations of future interest rates account for a risk premium where this
premium (positive or negative) does not necessarily rise uniformly with
maturity. According to this theory, for a given imbalance between the
supply and demand for funds within a maturity range, investors will
not be reluctant to shift their portfolios out of their preferred matu-
rity sector. Investors require a yield premium to move out of their
preferred sector. This theory can then explain different shapes of the
yield curve.
Market segmentation theory: This theory supports the “preferred
habitat” and explains the yield curve shape by asset and liability manage-
ment constraints. This theory assumes that investors are not willing to shift
from one maturity to another, to take advantages from the differences in
expectations and forward rates. Hence, the yield curve is explained by the
supply and demand within each maturity sector (see Carleton and Cooper,
1976).
5.6. The YTM and the Theories of the Term Structure

of Interest Rates
5.6.1. Computing the YTM
The YTM, refers to the average annual rate of return expected from the
purchase of a bond. This is a promised return rather than an actual return.
The YTM can be computed using one of the following three methods: the
arithmetic mean, the geometric mean, and the internal YTM.
The arithmetic mean YTM

This method computes the arithmetic mean of the expected future rates
of return on a bond. It uses the expected rates of return over the next
years. The arithmetic mean corresponds to the sum of the expected returns
on the bond over its life, which is divided by the number of years until
maturity. The resulting figure corresponds to the arithmetic mean YTM or
the expected average annual percentage increase in the capital invested.
This method assumes that the dollar investment in the bond remains
constant.
The geometric mean YTM

This method adds 1.00 to each expected annual return on the bond. All
the converted returns are multiplied and the nth root of the product is
computed, where n refers to the number of years until maturity. Subtracting
one from the root gives the geometric mean YTM. This method is based
on the assumption that all profits are re-invested by buying more bonds.
The geometric mean YTM reflects in this context, the average percentage
increase in the capital over the bond’s life.
The internal YTM

The equality between the current bond market price and the discounted
value of all cash flows of the investment gives an internal YTM. This method
assumes the re-investment of coupons at the internal yield without any
re-investment risk.
5.6.2. Market segmentation theory of the term structure

The market segmentation theory assumes that the market is composed of
extremely risk averse investors whose objective is portfolio “immunization”.
Immunization is achieved when the effective maturity of assets is perfectly
matched up with the effective maturity of liabilities. In this theory, the
survival of an institution is an objective function. The yield of each
segment in the market corresponds to the intersection of supply and
demand.
5.7. Spot Rates and Forward Interest Rates

5.7.1. The theoretical spot rate
The theoretical spot rate for an nth six-month period can be calculated
using the following equation:
C C C (C + 100)
Bn = + 2
+ 3
+ ···+
(1 + y1 ) (1 + y2 ) (1 + y3 ) (1 + yn )n
where
Bn : price of a coupon Treasury security with n periods to maturity (per

100 dollar of value);
C: semi-annual coupon for the coupon Treasury security with n periods
to maturity (per 100 dollar of par value) and
yt : t = 1, . . . , (n − 1): known theoretical spot rates.
This equation can also be written as:
  n1
(C + 100)
yn = 
n−1 1
 −1
Bn − C − t=1 (1+yt )t
When yn is multiplied by two, this gives the theoretical spot rate on a

bond-equivalent basis.
5.7.2. Forward rates

Consider the two following investment opportunities:
Strategy 1: Invest in a one-year Treasury bill,

Strategy 2: Invest in a six-month Treasury bill and buy another six-month
Treasury bill at the maturity date of the first Treasury bill.
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.
An investor buying a one-year Treasury bill maturing in a year with a

maturity value of 100, pays a price B:
B = 100/(1 + Y2 )2
where Y2 represents one-half the bond-equivalent yield of the one-year

spot rate.
If the investor buys the six-month Treasury bill, the value of his
investment in six months would be:
B(1 + Y1 )
where Y1 represents one-half the bond-equivalent yield of the six-month

spot rate.
In the same context, if we denote by f2 one-half the forward rate on a
six month Treasury bill available six months from now, then the P dollars
in one year becomes:
B(1 + Y1 )(1 + f2 )
The result of investing in an asset that gives 100 in one year can be
written as:
B(1 + Y1 )(1 + f2 ) = 100
or:
B = 100/[(1 + Y1 )(1 + f2 )]
The two strategies are equivalent for the investor if they lead to the
same result or:
100/(1 + Y2 )2 = 100/[(1 + Y1 )(1 + f2 )]
Hence, the value of the implied forward rate is given by:
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%
Hence, the forward rate for a six-month instrument on a bond-

equivalent basis is 9,3% or 2(4,65%). The formula can be generalized to
compute implicit forward rates from theoretical spot rates as follows:
1/t
(1 + yn+t )n+t
fnt = −1
(1 + y n )n
where
fnt : forward rate n periods from now for t periods,
yn : semi-annual spot rate.
The implied forward rate on a bond-equivalent basis is obtained by

multiplying fnt by two.
5.8. Issuing and Redeeming Bonds

Example: Consider a firm issuing 1000 bonds. The nominal value of each
bond is 1000, and the principal amount of the issue is 1,000,000. If the
quoted bond price is 100, this means 100% of 1000. If the interest rate falls,
the bond price may be 102, i.e., 102% of 1000, or 1020. If the interest rate
rises, the bond price can become 98, i.e., 98% of 1000, or 980. If the interest
rate is 12% and the coupon is paid the first January each year, an investor
buying this bond in France the first May, must pay 1000 plus accrued
interest for four months, i.e., 40, 4% of 1000, or 1040. If the coupon is paid
each six months, the first July the investor receives 60, corresponding to
the 40 paid to the seller and the interest 20 corresponding to two months.
The amount of the issue corresponds to the number of bonds (coupures)
times the nominal value of each bond. The issue price corresponds to the
price paid when buying the bond.
Application: Consider the following bond issue:
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
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;
or 2006;
or 2010;
or 2014 and
or 2018.
Table 5.3 shows the re-payment of the issue for the borrower.
10000 = 500 × 200
At the time of the issue, the issuer receives 9,900,000 or ((5000)(1980)).
Table 5.3. The re-payment profile of the issue for the borrower: issue date, the first
January 1994.
Years 1 to 10 Years 1 to 10 Years 1 to 10
(1) Number (2) The (3) The (6) Number

of bonds total number of (5) The of bonds
alive in the amount of bonds paid (4) The annu- alive at the
beginning of coupon “dead” at amount of ities in end of the
the year payments the end the payments 103 year in
in 103 in 103 year in 103 in 103 (2)+ (4) 103 (1) − (3)
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
Table 5.4. The profile of cash flows for the bondholder (the investor lending money):
initial investment at time 0: 19,800 ((1980)(10)).
Years 1 to 10 Years 1 to 10 Years 1 to 10

(1) Number (3) The
of bonds (2) The total number of
alive in the amount of bonds paid (4) The
beginning of coupon “dead” at amount of (5) The
the year payments in the end the payments annuities in
in 103 103 year in 103 in 103 103 (2) + (4) (6∗ )
10 2000 (10)200 0 0 2000 10

10 2000 0 0 2000 10
10 2000 0 0 2000 10
10 2000 0 0 2000 10
10 2000 0 0 2000 10
10 2000 0 0 2000 10
10 2000 5 10,030 5(2006) 12,030 5
5 1000 (5)200 0 0 1000 5
5 1000 0 0 1000 5
5 1000 5 10,090 5(2018) 11,090 0
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).
5.9. Mortgage-Backed Securities: The Monthly Mortgage

Payments for a Level-Payment Fixed-Rate Mortgage
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.
The monthly mortgage payment is due in the beginning of each month. It
consists of two elements. The first corresponds to interest of 1/12th of the
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.
Example: Consider a mortgage loan of 150,000 Euros with maturity in

15 years (180 months), the mortgage rate is 10%. Table 5.5 shows the
amortization schedule for a level-payment fixed-rate mortgage for the period
of 180 months. Each monthly mortgage payment comprises interest and re-
payment of principal.
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:
P V : present value of an annuity or the original mortgage balance;

A: amount of the annuity (monthly mortgage payment);
n: number of periods or months and
r: periodic interest rate (annual interest rate/12).
The formula corresponds to the present value of an ordinary annuity

formula.
The equality can also be written as:
Amount due to be paid

Monthly payment =
Present value of an annuity of 1 Euro/month
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
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.
Beginning Monthly Ending

mortgage mortgage Interest Principal mortgage
Month balance payment for month re-payment balance (2)−(5)
1 1, 50, 000 1611.907677 1250 361.9076766 1, 49, 638.0923

2 1, 49, 638.0923 1611.907677 1246.984103 364.9235739 1, 49, 273.1687
3 1, 49, 273.1687 1611.907677 1243.943073 367.9646036 1, 48, 905.2041
4 1, 48, 905.2041 1611.907677 1240.876701 371.0309753 1, 48, 534.1732
5 1, 48, 534.1732 1611.907677 1237.784776 374.1229001 1, 48, 160.0503
6 1, 48, 160.0503 1611.907677 1234.667086 377.240591 1, 47, 782.8097
7 1, 47, 782.8097 1611.907677 1231.523414 380.3842626 1, 47, 402.4254
8 1, 47, 402.4254 1611.907677 1228.353545 383.5541314 1, 47, 018.8713
9 1, 47, 018.8713 1611.907677 1225.157261 386.7504158 1, 46, 632.1209
10 1, 46, 632.1209 1611.907677 1221.934341 389.973336 1, 46, 242.1475
11 1, 46, 242.1475 1611.907677 1218.684563 393.2231138 1, 45, 848.9244
12 1, 45, 848.9244 1611.907677 1215.407703 396.4999731 1, 45, 452.4244
13 1, 45, 452.4244 1611.907677 1212.103537 399.8041395 1, 45, 052.6203
14 1, 45, 052.6203 1611.907677 1208.771836 403.1358407 1, 44, 649.4845
15 1, 44, 649.4845 1611.907677 1205.412371 406.495306 1, 44, 242.9892
16 1, 44, 242.9892 1611.907677 1202.02491 409.8827669 1, 43, 833.1064
17 1, 43, 833.1064 1611.907677 1198.60922 413.2984566 1, 43, 419.8079
18 1, 43, 419.8079 1611.907677 1195.165066 416.7426104 1, 43, 003.0653
... ...
74 1, 13, 834.9611 1611.907677 948.624676 663.2830006 1, 13, 171.6781
75 1, 13, 171.6781 1611.907677 943.0973176 668.8103589 1, 12, 502.8678
76 1, 12, 502.8678 1611.907677 937.523898 674.3837786 1, 11, 828.484
77 1, 11, 828.484 1611.907677 931.9040332 680.0036434 1, 11, 148.4803
78 1, 11, 148.4803 1611.907677 926.2373361 685.6703404 1, 10, 462.81
86 1, 05, 500.4334 1611.907677 879.1702784 732.7373982 1, 04, 767.696
87 1, 04, 767.696 1611.907677 873.0641334 738.8435431 1, 04, 028.8525
88 1, 04, 028.8525 1611.907677 866.9071039 745.0005727 1, 03, 283.8519
89 1, 03, 283.8519 1611.907677 860.6987658 751.2089108 1, 02, 532.643
90 1, 02, 532.643 1611.907677 854.4386915 757.468985 1, 01, 775.174
... ...
175 9395.51514 1611.907677 78.2959595 1533.611717 7861.903423
176 7861.903423 1611.907677 65.51586186 1546.391815 6315.511609
177 6315.511609 1611.907677 52.6292634 1559.278413 4756.233195
178 4756.233195 1611.907677 39.63527663 1572.2724 3183.960795
179 3183.960795 1611.907677 26.53300663 1585.37467 1598.586126
180 1598.586126 1611.907677 13.32155105 1598.586126 0
1611,907677:
 
 150000  = 1611,907677.
1
1− (1,00833)180
0,008333
It is assumed in the above analysis of fixed-rate mortgages, that the

level homeowner would not pay any fraction of the mortgage balance before
the fixed date. We refer to the payments made in excess of the fixed principal
payments as pre-payments. The amount and the timing of the pre-payment
are uncertain and define the pre-payment risk.
5.10. Interest Rate Swaps

5.10.1. The pricing of interest rate swaps
Example: Consider the following interest rate swap between two firms A
and B in the presence of a commercial bank (Fig. 5.1).
The bank pays the six-month LIBOR rate to the firm A and receives
a fixed rate of 9.6%. The bank pays to firm B a fixed rate of 9.5% and
receives the LIBOR. A notional amount No is used. The bank benefits from
a commission of 0.10% per year on the notional amount. If the LIBOR rate
is 11%, firm B pays the bank: 1/2(0.11 − 0.095) = 0.0075 No and the bank
pays A: 1/2(0.11 − 0.096) = 0.0070 No.
The bank gains 0.005 No for six months. This swap can be assimilated
to two default bonds since the bank borrows from A at the six-month
LIBOR and lends to A at a rate of 9.6%.
5.10.2. The swap value as the difference between the prices

of two bonds
Let us denote by Sw, the value of the swap for the bank, B1 the value of
bond paying a coupon at 9.6% per year, and by B2 the value of a bond
paying the six-month LIBOR. The value of the swap for the bank is:
Sw = B1 − B2
9.6% 9.5%
Firm A Commercial bank Firm B
LIBOR LIBOR
Fig. 5.1. An interest rate swap.
The value of the bond B1 is given by:

n
B1 = Ce−ri ti + N e−rn tn
i=1
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
where C1 is the certain coupon at date t1 .
The swap as a series of forward contracts

A swap position can be seen as a package of forward or futures contracts.
In fact, firm A has agreed to pay 9.6% and receive six-month LIBOR.
Assuming a 100 million dollars NO, A has agreed to buy the six-month
LIBOR for 48 million dollars. This represents a six-month forward contract,
where A agrees to pay 48 million in exchange for delivery of six-month
LIBOR. An interest rate swap can be assimilated to a portfolio of forward
contracts. In fact, firm A receives each six month an amount equal to 0.5 No
(LIBOR−0.096) where the LIBOR rate corresponds to the previous period.
This is a forward contract on the six-month LIBOR rate.
5.10.3. The valuation of currency swaps

Example 1
Consider a firm A borrowing 6,00,000 units of currency A at a fixed rate of
8% for five years. The firm must convert this in a currency B.
A firm B enters into a five-year swap with firm A at a rate of 9.5% for
an amount of 2,200,000 of currency B. The principal amount is exchanged
for an exchange rate equal to 3.666, (the day of the swap implementation).
At each payment date, firm A pays B interest charges in the currency of B

using a rate of 9.5% on the principal amount of 2,200,000 units of currency
B. The firm B pays A at a rate of 8% on the principal amount of 6,00,000
units of currency A. Each six months, A pays 1,04,500 units of currency B,
or (9.5%) (2,200,000)/2, and B pays 88,000 units of currency B, or (8%)
(2,200,000)/2. At the swap term, A pays 2,200,000 units of currency B to
B and B pays 6,00,000 units of currency A to A. This swap operation is
equivalent to a position in a currency forward contract.
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)
Fig. 5.2. Foreign currency swap.
(dollars)8% (dollars)9%
Firm A Bank Firm B
11%(sterling) 12.25%(sterling)
Fig. 5.3. The foreign currency swap.

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.
5.10.4. Computing the swap

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.
Each cash-flow in a swap must be discounted at a unique theoretical
spot rate. This spot rate is obtained from forward rates. The following
example illustrates the procedure for the computation of the swap rate.
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
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?
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.

Chapter 6
EXTENSIONS OF SIMPLE BINOMIAL OPTION

PRICING MODELS TO INTEREST RATES AND
CREDIT RISK
Chapter Outline
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
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.
6.1. The Rendleman and Bartter Model (for details, refer

to Bellalah et al., 1998) for Interest-Rate Sensitive
Instruments
The CRR (1979) approach can be applied to the valuation of interest-
rate sensitive instruments. It can be presented in the form developed by
Extensions of Simple Binomial Option Pricing Models to Interest Rates 295
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.
Using these parameters, we have: u = 1.2214, d = 0.8187, q = 1.03045,

and p = 0.5257.
Figure 6.1 describes the dynamics of interest rates in this context.
Starting from an interest rate of 11%, the next rate in an upstate
13.43 is given by 11 times u or 11(1.2214). The rate 16.41% is given by
13.43 times u, or 13.43(1.2214). In each upstate, this operation is repeated
until the rate 36.52% given by 29.9(1.2214) is obtained at the end. Starting
from a rate of 11%, the rate 9% is obtained from the product of 11 by d,
or 11(1/1.2214). The rate 7.36 corresponds to 9 times d, or 9(1/1.2214).
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.
This operation is repeated till the end. The rate 3.31 is obtained by
4.04(1/1.2214).
6.1.1. Using the model for coupon-paying bonds

Consider a 8.5%, six-year bond with a maturity value equal to 1000 paying
an annual coupon of 85.
In the first step, the bond price must be computed at each node in the
binomial tree.
The bond value can be calculated using the following relationship:
Bij = [pBi+1,j+1 + (1 − p)Bi+1,j + coupon)]/er ij ∆t
with:
rij : interest rate at position j and time i;

Bij : corresponding to the bond price at state j and time i; and
p: the probability corresponding to an upstate.
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
Fig. 6.2. Dynamics of the bond price.

the bond value as follows:
[0,5257(1000) + (1 − p)(1000) + 85)]/e0,299(1) = 804,58
The same formula can be applied until the initial time 0.
6.2. Ho and Lee Model for Interest Rates

and Bond Options
Ho and Lee (1986) assumed the existence of a zero-coupon bond for each
(n)
maturity, n, (n = 0, 1, 2, . . .). They denoted by Pi (T ) the equilibrium
(n)
price of a zero-coupon bond at time n and state i. Hence, the price Pi (.)
is a function relating each bond price with its corresponding maturity date
or a discount function. The price of a bond maturing instantaneously is 1 or:
(n)
Pi (0) = 1 for all i, n
The price of a long-lived bond is nearly zero since:

(n)
lim Pi (T ) = 0 for all i, n
T →∞
6.2.1. The binomial dynamics of the term structure

At time zero, the discount function is observed and is denoted by:
(0)
P (.) = P0 (.) = 1.
(1) (1)
At time 1, the discount function is specified by two P1 (0) and P0 (0)
for an upstate and a down state, respectively. The discount function at the
second period between two dates 1 and 2 leads to two possible functions.
(1) (2) (2)
Hence, at time 1, P1 (.) leads to the functions P2 (.) and P1 (.) at time 2.
(n)
We denote the discount function at time n by Pi (.) after i ups and (n − i)
downs, at period (n + 1) between the dates n and (n + 1) (Fig. 6.3).
Given the discount function P (T ), the yield curve can be written as:
r(T ) = −ln(P (T ))/T
where r(T ) is the continuous yield for a bond maturing in T .
(n+1)
Pi+1 (.) : up
(n)
Pi (.)
(n+1)
Pi (.) : down
Fig. 6.3. The discount function in the Ho-Lee model.

6.2.2. The 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 an
initial time. When it remains at (N − 1) periods, the discounting functions
(1)
in upstates and down states allow the computation of P1 (N − 1) and
1
P0 (N − 1).
This model is similar to the binomial model of CRR (1979) and
Rendleman and Bartter (for details, refer to Bellalah et al., 1998). Ho
and Lee (1986) introduced two perturbation functions h(T ) and h∗ (T ).
(n)
The discount function at period n and state i is P1 (.T ). The details are
provided in the appendix of this chapter. They developed a non-arbitrage
condition, that ensures that:
πh(T ) + (1 − π)h∗ (T ) = 1 for n, i > 0
where π corresponds to the implied binomial probability. Hence, we have:

(n) (n+1) (n+1) (n)
Pi (T ) = πPi+1 (T − 1) + (1 − π)Pi (T − 1) Pi (1)
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.
6.2.3. Computation of bond prices in the Ho

and Lee model
Example 1: Consider the following parameters in the Ho and Lee model:
n = 4, δ = 0.994, π = 0.5, and t = 365 days. The term structure is
specified by the following discount function observed at initial time and
applying for periods 1 to 5: P00 (0) = 1, P00 (1) = 0.9826, P00 (2) = 0.9651,
Table 6.1. Discount function and bond prices in the Ho and Lee model.
Period T Pij (k) 0 1 2 3 4 5
P00 (K) 1 0.98260 0.96510 0.94740 0.92960 0.9119

P10 (K) 1 0.97923 0.95837 0.93752 0.91687 —
P11 (K) 1 0.98514 0.96997 0.95460 0.93921 —
P20 (K) 1 0.9757 0.95164 0.9278 — —
P21 (K) 1 0.98164 0.96316 0.94477 — —
P22 (K) 1 0.98756 0.97482 0.96198 — —
P30 (K) 1 0.97235 0.94520 — — —
P31 (K) 1 0.97822 0.95664 — — —
P32 (K) 1 0.98412 0.96823 — — —
P33 (K) 1 0.99006 0.97995 — — —
P40 (K) 1 0.96915 — — — —
P41 (K) 1 0.97500 — — — —
P42 (K) 1 0.98088 — — — —
P43 (K) 1 0.98680 — — — —
P44 (K) 1 0.99276 — — — —
P00 (3) = 0.9474, P00 (4) = 0.9296, and P00 (5) = 0.9119. Results are
reproduced in Table 6.1.
Example 2: Consider the following parameters to compute the discount

functions in the Ho and Lee model: n = 4, δ = 0.994, π = 0.5, and t =
365 days. The present term structure is specified by the following discount
function:
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.
6.2.4. Option pricing in the Ho and Lee model

Consider the pricing of an option on an interest-rate instrument, with a
maturity T and a final payoff X(n, i) such that: C(T, i) = X(n, i) 0 ≤ i ≤ T .
The underlying asset may be a bond, an interest rate, or a futures
contract. The option price at moment n and position i can be located
between a lower value L and an upper bound U . The possible prices of
the option at instant n and position i satisfy the following relationship:
L(n, i) ≤ C(n, i) ≤ U (n, i).
The option holder receives at instant n and state i, the payoff X(n, i)
for i in the interval 1 ≤ n < T . Ho and Lee (1986) showed that a hedged
portfolio comprising options and zero-coupon bonds allowed the elimination
Table 6.2. Discount function and bond prices in the Ho and Lee model.
Period Pij (k) 0 1 2 3 4 5
P00 (K) 1 0.98200 0.96100 0.94100 0.92100 0.91100

P10 (K) 1 0.97567 0.95248 0.92941 0.91653 —
P11 (K) 1 0.98155 0.96604 0.94634 0.93886 —
P20 (K) 1 0.9732 0.9468 0.9309 — —
P21 (K) 1 0.97917 0.95832 0.94786 — —
P22 (K) 1 0.98508 0.96992 0.96513 — —
P30 (K) 1 0.96991 0.95069 — — —
P31 (K) 1 0.97576 0.96220 — — —
P32 (K) 1 0.98165 0.97385 — — —
P33 (K) 1 0.98758 0.98564 — — —
P40 (K) 1 0.9772 — — — —
P41 (K) 1 0.98313 — — — —
P42 (K) 1 0.98907 — — — —
P43 (K) 1 0.99504 — — — —
P44 (K) 1 1 — — — —
of profitable arbitrage opportunities and leads to the following equation:
C(n, i) = [π{C(n + 1, i + 1) + X(n + 1, i + 1)}

(n)
+ (1 − π){C(n + 1, i) + X(n + 1, i)}]Pi (1)
(n)
where Pi (1) is the price of a zero-coupon bond at the node (n, i)
and π corresponds to the implied binomial probability. The option price
must satisfy the condition: C(T − 1, i) = max[L(T − 1, i), min(C ∗ (T − 1, i),
U (T − 1, i)].
The application of this model to the pricing of any type of interest-
contingent claim requires the estimation of the probability π and the spread
δ between the two perturbations functions.
First, the discount function at the time of pricing must be estimated.
Then, the parameters π and δ are estimated using a non-linear procedure
like that in Ho and Lee (1986) or Whaley (1986).
The estimation approach uses observed contingent claim prices and a
pricing model in order to imply the parameters in the same way as we
calculate an implied volatility for stock options.
Example 3: Consider the pricing of a one-year put option on a Treasury

bill. The option’s strike price is 980. The three-month T-bill has a face
value of 1000 euros. The term structure is defined by the following
parameters: n = 4, δ = 0.99, π = 0.5, and ∆t = 3 months. The discount
1
0.9958
0.9914
0.9905
0.9860
0.9878
0.9814
0.992 0.9761 0.9806

0.9779
0.9716
0.9708
0.9664
0.9611
Fig. 6.4. Dynamics of the Treasury bond.
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:
P = max[980 − 1000P 4j(1), 0] for j = 0, 1, 2, 3, 4.
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:
(0.5(4,477) + 0,5(13,556))0,9716 = 8,761.
As in Fig. 6.5, the option price at initial time is 3.196.

Ritchken and Boenawen (1990) showed that the Ho and Lee model
does not completely prevent the possibility of negative interest rates. These
(n)
negative rates disappear when the following constraint is used: Pn (1) < 1.
To illustrate this point, consider the following parameters: δ = 0.9,
π = 0.5, and n = 4. The observed initial term structure of interest rates for
the following five periods is specified by:
r0 (1) = 9.531%, r0 (2) = 8.6178%, r0 (3) = 7.696%, r0 (4) = 7.2321%,

r0 (5) = 7.2321%, which gives: p(1) = e−r0 (1).1 , p(2) = e−r0 (2).2 , p(3) =
e−r0 (3).3 , p(4) = e−r0 (4).4 , and p(5) = e−r0 (5).5
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.
Figure 6.6 reveals the possibility of negative interest rates in the Ho

and Lee model.
6.2.5. Deficiency in the Ho and Lee model

Ho and Lee model presents a deficiency. 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), and Ritchken and Boenawen (1990) among others. These
authors generalized the Ho and Lee model and provided the necessary
adjustments in order to obtain an economically meaningful model. Ritchken
and Boenawen (1990) showed that the restrictions imposed by Ho and Lee
imply that bonds are priced by a model characterized by a “risk neutral”
probability parameter π and an “interest-rate spread”, δ. However, the
evolution of the discount function, namely the default-free zero-coupon
bond prices, may not be bounded in the interval. To show this, they
developed an example where π = 0.4, and δ = 0.8. They generated prices of
pure discount bonds at all the vertices and found that some prices exceeded
one. This indicates the presence of negative interest rates in the lattice. In
fact, if one modifies their δ from 0.8 to 0.9 and generates bond prices,
it is clear that some bond prices exceed one. To avoid negative interest
rates, the constraint Pnn (1) < 1, must be added for each time period in the
lattice.
6.3. Binomial Interest-Rate Trees and the Log-Normal

Random Walk
Consider a binomial interest-rate tree where the interest rate moves up (u)
or down (d). The initial interest rate r0 corresponds to the current one-year
interest rate when the length of each period is one year. It also corresponds
to the one-year forward rate.
The initial rate or the one-year forward rate can take on two possible
values in the next period with the same probability of occurring. Hence,
one rate results from a rise in rates and the other results from a fall in
rates. Assume that the dynamics of interest rates are specified by a log-
normal random walk process with a given volatility (Fig. 6.7). For the sake
of clarity, let us denote respectively by:
• σ: assumed volatility for the one-year forward rates;

• r1,u : one-year forward rate, one year from now if rates rise and
• r1,d : one-year forward rate, one year from now if rates fall.
Hence, in the first year, there are two possible rates. We specify the
following relationship between rising and falling rates as follows:
r1,u = r1,d e2σ .

r 2,uu
r 1,u
r0 r 2,ud
r 1,d
r2,dd
Fig. 6.7. Dynamics of interest rates.
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:
r2,ud = r2,dd e2σ and r2,uu = r2,dd e4σ
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:
(re2σ − r)/2 which is nearly equal to (r + 2rσ − r)/2 or σr.
The process generating the interest-rate tree or the forward rates

implies that volatility is measured with respect to the current level of rates.
For example, if σ = 20% and the one-year rate is 3%, then the volatility
of the one-year forward rate is 20% (3%) or 6% or 60 basis points. Hence,

when the current one-year rate is 10%, the volatility of the one-year forward
rate would be 10% × 20% or 200 basis points. 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 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 until the initial
time. The value at each node is given by the expected cash flows under the
appropriate discount rate. The one-year forward rate at this node must be
used. We denote these values by:
Vu : value of the bond if the one-year interest rate rises;

Vd : value of the bond if the one-year interest rate falls;
C: amount of the coupon payment and
r: one-year forward rate at the node where valuation is sought.
Hence, the value of the bond at a given node is:
Bond value = p(Vu + c)/(1 + r− ) + (1 − p)(Vd + c)/(1 + r− )
How to construct a binomial interest-rate tree?

Consider the valuation of a two-year bond with a coupon rate of 4.5%
when σ is 10%. Figure 6.8 shows the cash flows at each node. The initial
interest rate is equal to 4.5%.
Figure 6.9 shows how to calculate the one-year forward rates for the
first year by a trial-and-error method.
The model is based on an iterative process that allows the computation
of the forward rates r1,d and r1,d , which are consistent with the volatility
assumption and the observed market value of the bond. This can be done
in different steps as shown below.
Vu + c: cash flow in a high state
One-year rate at a node

where the bond price is V
calculated at the rate r-
Vd + c: cash flow in a low state
Fig. 6.8. Computing the bond value at a given node.

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:
(Vu + c)/1.045 = (99.748576 + 4.5)/1.045 = 99.7594.
The present value resulting from a fall in the interest rate is computed as:
(Vd + c)/1.045 = (100.5774 + 4.5)/1.045 = 100.5525.

Using an equal probability, the value at time 0 is:
Value = 0.5(99.113 + 100.021) = 100.15596.
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.
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:
(Vu + c)/1.045 = (99.6323 + 4.5)/1.045 = 99.6481.
The present value resulting from a fall in the interest rate is computed as:
(Vd + c)/1.045 = (100.48076 + 4.5)/1.045 = 100.46.
Using an equal probability, the value at time 0 is:
Value = 0.5(99.6481 + 100.46) = 100.
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.
6.4. The Black-Derman-Toy Model (BDT)

This model is consistent with the observed term structure of interest rates.
The volatility of the short rate is time dependent. The continuous process
of the short rate is given by:
d ln r = [θ(t) + (δσ(t)/δt)/σ(t)ln(r)]dt + σ(t)dz
where (δσ(t)/δt)/σ(t) corresponds to the speed of mean-reversion.

The following example calibrates the BDT model to the current term
structure of zero-coupon rates and volatilities.
6.4.1. Examples and applications

Example 4: Consider the following sample current data (Table 6.3):
Consider the following tree for bond prices (Fig. 6.11).
The price of a zero-coupon bond maturing in one year is given by:
92.59259 = [100(0.5) + 100(0.5)]/1.08.
Table 6.3. Current data for zero-coupon bonds.
Years to maturity Zero-coupon rates (in %) Zero-coupon volatilities (in %)
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
Fig. 6.11. Binomial tree for bond prices.

This allows the description of a one-period price tree:

100
92.59259
100
The price today of a two-year zero-coupon bond maturing in two years

can be computed as:
84.945528 = 100/1.0852 = 100/1.177225.
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 following relationship must be verified:
84.945528 = [0.5Su + 0.5Sd ]/1.08.
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).
In the BDT model, rates follow a log-normal distribution, so we have:

√
σN = [1/2 T /N ] ln(ru /rd ) = 0.18
or
σN = [1/2] ln(ru /rd ) = 0.18 and Sd = 100/(1 + ru ), Su = 100/(1 + rd ).

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
Fig. 6.12. Binomial tree for interest rates.
Substituting these expressions, it gives:
84.945528 = [0.5{100/(1 + ru )} + 0.5{100/(1 + rd )}]/1.08.
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:
84.945528 = [0.5{100/(1 + rd e0.18(2) )} + 0.5{100/(1 + rd )}]/1.08.
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
The next step gives the following exhibit.

ruu
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:
0.5 ln(ruu/rud) = 0.5 ln(rud/rdd) or rdd = r2 ud/ruu.
As before, the price today of a bond maturing in three years is:
77.22 = 100/(1 + 0.09)3
Using the risk-neutral valuation principle, we have:
Suu = 100/(1 + ruu), Sud = 100(1 + rud), Sdd = 100/(1 + rdd)

Su = [0.5 Suu + 0.5 Sdd]/(1 + 0.1060),
Sd = [0.5 Sdd + 0.5 Sud]/(1 + 0.074),
77.22 = [0.5 Su + 0.5 Sd]/(1 + 0.08).
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).
In this case, Su can be written as:

√
Su = 100/[1 + e0.32 ( (100/Sd) − 1)]2 .
This equation can be solved using the Newton–Raphson algorithm. The

solution gives the values of Su, Sd, rdd, rud, and ruu.
6.5. Trinomial Interest-Rate Trees and the Pricing

of Bonds
6.5.1. The model
There are three possible movements: up, down, and stay the same. Consider
an initial short-term interest rate r of 5% per annum (r = 0.05). The
standard deviation is 0.01 per year. The drift pulls the interest rate back
to its level of 5% at a rate of 10% per year. Over a short-term interval, the
expected increase in the interest rate is written as:
0.1(0.05 − r)∆t.
√
The standard deviation is 0.01 ∆t.
The interest-rate process is:
dr = 0.1(0.05 − r)dt + 0.01dz.
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:
(0.05 + 0.015)p1 + (0.05 + 0)p2 + (0.05 − 0.015)p3 = 0.05
or
0.065p1 + 0.05p2 + 0.035p3 = 0.05.

r=
C r=
0.05
B r = 0.05
r=
A
r=
Fig. 6.13. The construction of the trinomial tree.
The probabilities also verify the following equation for the variance:
0.0652p1 + 0.052 p2 + 0.0352p3 − 0.052 = 0.012.
The solution to the following system gives the appropriate probabilities:
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:
p1 = 0.222222; p2 = 0.555556, and p3 = 0.222222.
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
or
0.05p1 + 0.035p2 + 0.02p3 = 0.0365
and
0.052 p1 + 0.0352 p2 + 0.022 p3 − 0.03652 = 0.012 .
The solution to the following three equations is:
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
p1 = 0.277222; p2 = 0.545556, and p3 = 0.177222
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
0.065p1 + 0.05p2 + 0.035p3 = 0.05
and
0.0652p1 + 0.052 p2 + 0.0352p3 − 0.052 = 0.012 .
The solution to these three equations is:
p1 = 0.222222, p2 = 0.555556, p3 = 0.222222.
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.
Using a system of three equations allows the computation of different

probabilities:
p 1 + p2 + p3 = 1
(0.065 + 0.015)p1 + (0.065 + 0)p2 + (0.065 − 0.015)p3 = 0.0635
or
0.08p1 + 0.065p2 + 0.05p3 = 0.0635
and
0.082 p1 + 0.0652p2 + 0.052 p3 − 0.06352 = 0.012 .
The solution to these three equations is:
p1 = 0.177222, p2 = 0.545556, and p3 = 0.277222.
Derivative securities can be valued using the recursive method through

the trinomial tree in the same way, as in the binomial model. This model
applies also to discount bonds. Using bond prices, it is also possible to
determine the complete yield curve at any given node of the trinomial
lattice. It is convenient to note that the tree is used to model a function
or the short rate r, f (r) with the same volatility as r in the small time
interval. When the volatility of r is constant, r can be modeled directly.
When the volatility of r is proportional to r, as it is the case in the log-
normal model, f (r) = log(r) can be modeled directly. When the volatility
of r is proportional to rα , f (r) can be r(1−α) .
The trinomial model proposed above is based on the following branch-
ing process: up, down, and stay the same. This is not the only way to
construct the trees. It is also possible to construct a trinomial tree as
illustrated in the following figure (Fig. 6.14).
The trinomial model allows the user to specify future volatility of
the short rate, the current volatility of spot rates, and the volatility
associated with the current-term structure of interest rates. It is important
to remember that this trinomial approach is a variation of finite difference
methods.
6.5.2. Applications of the binomial and trinomial models

We consider the valuation of a stock option using the binomial model
and the trinomial model (Table 6.4). The option is priced on 15/06/2002.
Fig. 6.14. Possible branching in a trinomial tree.
Table 6.4. Pricing a call using the binomial and trinomial models.
Call price S = 18 S = 19 S = 20 S = 21
K = 22 Binomial 2.99 3.50 4.02 4.54

Trinomial 2.99 3.47 4.00 4.54
K = 23 Binomial 2.73 3.22 3.73 4.25
Trinomial 2.77 3.21 3.70 4.23
K = 24 Binomial 2.55 2.94 3.45 3.96
Trinomial 2.54 2.98 3.42 3.93
K = 25 Binomial 2.37 2.75 3.16 3.68
Trinomial 2.32 2.76 3.20 3.64
K = 26 Binomial 2.19 2.57 2.95 3.39
Trinomial 2.16 2.54 2.97 3.41
K = 27 Binomial 2.01 2.39 2.77 3.15
Trinomial 2.01 2.35 2.75 3.19
K = 28 Binomial 1.83 2.21 2.59 2.97
Trinomial 1.85 2.20 2.54 2.97
K = 29 Binomial 1.68 2.03 2.41 2.79
Trinomial 1.70 2.04 2.39 2.75
K = 30 Binomial 1.58 1.86 2.23 2.61
Trinomial 1.57 1.89 2.23 2.58
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.
S = 22 S = 25 S = 28 S = 30
K = 22 Binomial 5.08 7.02 9.05 10.58

Trinomial 5.09 6.99 9.08 10.54
K = 23 Binomial 4.77 6.60 8.57 10.01
Trinomial 4.77 6.58 8.58 10.02
K = 24 Binomial 4.48 6.18 8.14 9.47
Trinomial 4.46 6.18 8.09 9.51
K = 25 Binomial 4.19 5.77 7.71 9.03
Trinomial 4.16 5.78 7.68 9.01
K = 26 Binomial 3.91 5.46 7.29 8.60
Trinomial 3.86 5.47 7.28 8.55
K = 27 Binomial 3.62 5.17 6.88 8.18
Trinomial 3.63 5.16 6.87 8.14
K = 28 Binomial 3.36 4.88 6.46 7.75
Trinomial 3.41 4.85 6.48 7.74
K = 29 Binomial 3.17 4.59 6.15 7.34
Trinomial 3.18 4.55 6.16 7.34
K = 30 Binomial 2.99 4.31 5.86 6.92
Trinomial 2.96 4.28 5.85 6.94
We use a historical simulation to estimate the volatility parameter

(Table 6.5). The interest rate is 5%. The annualized volatility is between
45% and 50%. In this analysis, a dividend rate is used by dividing the
dividend amount by the initial underlying asset price (Table 6.6). The
annualized volatility is 45% (Table 6.7).
The annualized volatility is 50%.
The reader can compare the differences between both these models.
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.
S = 18 S = 19 S = 20 S = 21
K = 22 Binomial 3.48 4.00 4.53 5.07

Trinomial 3.45 3.97 4.52 5.07
K = 23 Binomial 3.20 3.73 4.26 4.79
Trinomial 3.23 3.69 4.23 4.77
K = 24 Binomial 2.99 3.46 3.98 4.51
Trinomial 3.02 3.47 3.93 4.48
K = 25 Binomial 2.82 3.22 3.71 4.24
Trinomial 2.81 3.26 3.71 4.19
K = 26 Binomial 2.65 3.04 3.44 3.96
Trinomial 2.60 3.05 3.50 3.95
K = 27 Binomial 2.48 2.87 3.27 3.69
Trinomial 2.45 2.83 3.28 3.74
K = 28 Binomial 2.31 2.70 3.10 3.49
Trinomial 2.30 2.66 3.07 3.52
K = 29 Binomial 2.14 2.53 2.92 3.32
Trinomial 2.16 2.51 2.87 3.31
K = 30 Binomial 1.97 2.36 2.76 3.15
Trinomial 2.01 2.36 2.72 3.10
S = 22 S = 25 S = 28 S = 30
K = 22 Binomial 5.62 7.60 9.62 11.13

Trinomial 5.63 7.57 9.65 11.12
K = 23 Binomial 5.33 7.19 9.19 10.58
Trinomial 5.33 7.17 9.16 10.62
K = 24 Binomial 5.04 6.79 8.78 10.12
Trinomial 5.03 6.79 8.73 10.13
K = 25 Binomial 4.77 6.38 8.36 9.70
Trinomial 4.74 6.40 8.33 9.64
K = 26 Binomial 4.49 6.09 7.96 9.28
Trinomial 4.44 6.10 7.94 9.24
K = 27 Binomial 4.22 5.81 7.55 8.87
Trinomial 4.19 5.80 7.55 8.84
K = 28 Binomial 3.94 5.53 7.15 8.47
Trinomial 3.98 5.50 7.17 8.45
K = 29 Binomial 3.72 5.25 6.86 8.06
Trinomial 3.76 5.21 6.86 8.07
K = 30 Binomial 3.54 4.98 6.57 7.66
Trinomial 3.55 4.92 6.56 7.68
Trigeorgis (1993) applied the methodology to the valuation of invest-

ments with multiple real options and to the pricing of managerial flexibility
implicit in investment opportunities.
Hull and White (1988,1993) provided a modified binomial lattice model
for the valuation of stock options and interest-rate options.
Boyle (1986) proposed a trinomial option-pricing model in which the
stock price can move either upwards or downwards or stay unchanged at a
given time period. In another paper, Boyle (1988) showed how a five-jump,
three-dimensional lattice can be used for the valuation of options on two
underlying assets.
Omberg (1988) studied a family of discrete-time jump processes and
applied a Gauss-Hermite quadrature technique to derive the prices of
options on options or compound options.
Yisong (1993) modified Boyle’s approach by presenting a general
methodology that can be applied to any multidimensional lattice approach.
He proposed a modified approach to the selection of lattice parameters
including probabilities and jumps using additional restrictions.
Sandmann (1993) developed a model for the pricing of European
options under the assumption of a stochastic interest rate in a discrete
time setting. He used a combination of the binomial model for a stock with
a binomial model for the spot interest rate.
Derivative securities can be valued using the recursive method through
the trinomial tree in the same way, as in the binomial model. This model
also applies to discount bonds. Using bond prices, it is also possible to
determine the complete yield curve at any given node of the trinomial
lattice.
The trinomial model proposed by Hull and White allows the user to
specify future volatility of the short rate, the current volatility of spot
rates, and the volatility associated with the current, term structure of
interest rates. It is important to remember that this trinomial approach
is a variation of finite difference methods.
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.
4. Describe the Black-Derman-Toy model.

5. Describe the trinomial interest-rate model and the pricing of bonds.
6. What are the specific features of the Ho and Lee approach in the
description of the term structure of interest rates?
7. What are the specific features of the Ho and Lee approach for the
valuation of interest-rate-dependent contingent claims?
8. What are the deficiencies in the Ho and Lee model?
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)]
The two functions h(T ) and h∗ (T ) ensure that:
(n+1) (n) (n)

Pi+1 (T ) = [Pi (T + 1)/Pi (1)]h(T ) (A.1)
(n+1) (n) (n)
Pi (T ) = [Pi (T + 1)/Pi (1)]h∗ (T ) (A.2)
h(0) = h∗ (0) = 1
The non-arbitrage condition ensures that:
πh(T ) + (1 − π)h∗ (T ) = 1 for n, i > 0

where π corresponds to the implied binomial probability. Hence, we have:
(n) (n+1) (n+1) (n)

Pi (T ) = [πPi+1 (T − 1) + (1 − π)Pi (T − 1)]Pi (1).
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, and d: return in a

down state.
(n)
Consider the discount function Pi (T ) at state i and time n. Using
Eqs. (A.1) and (A.2), we have:
(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)
The independence condition (reflecting the fact that an upward

movement followed by a downward movement is equivalent to a downward
movement followed by an upward movement) allows to write:
h(T + 1)h∗ (T )h∗ (1) = h∗ (T + 1)h(T )h(1).
Elimination of h∗ gives:
h(T + 1)[1 − πh(T )][1 − πh(1)] = (1 − π)h(1)h(T )[1 − πh(T + 1)].
Simplifying for T = 1 gives:
1 δ
= +Γ
h(T + 1) h(T )
where the constants δ and Γ are defined in a way such that:
1 π(h(1) − 1)
h(1) = ; Γ= .
π + (1 − π)δ (1 − π)h(1)
The term δ corresponds to the spread between the two perturbation

functions. Ho and Lee showed that:
1 δT
h(T ) = for T ≥ 0 and h∗ (T ) = .
π + (1 − π)δ T π + (1 − π)δ T
The dynamics of interest rates is completely specified by π and δ. Using
Eqs. (A.1) and (A.2), it is possible to obtain the discount function for any
moment as a function of the initial value, or:
P (T + n)h∗ (T + n − 1)h∗ (T + n − 2) · · · h∗ (T + i)h(T + i − 1) · · · h(T )

Pin (T ) =
P (n)h∗ (n − 1)h∗ (n − 2) · · · h∗ (i)h(i − 1) · · · h(1)
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 + 1)δ n−i

Pi (T ) = .
P (n)(π + (1 − π)δ n )
(n) (n) (n)
And the one-period interest rate ri (1) is: ri (1) = − ln Pi (1) or
(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:
µ = ln[P (n)/P (n + 1)] + ln(πδ −n + (1 − π)) + nq ln δ
or:
µ = ln[P (n)/P (n + 1)] + ln(πδ −(1−q)n + (1 − π)δ qn )
and:
σ = nq(1 − q)(ln δ)2
Do not forget that notional amount of debt has to be monitored

carefully after what happened in 2008 with the credit crunch.
Exercises
volatility = 0.2, T = 5 months, and N = 5. In this case, we have: p =
0.5217, d = 0.9439, and u = 1.0594.
Dynamics of the underlying asset for five periods
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
Valuation of European put option
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
Valuation of American put option
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
Valuation of European call option
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
Valuation of American call option
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, 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.
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.
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).
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.
American path dependent options. Journal of Derivatives, 1 (Fall 1993),
21–31.
Jarrow, RA and A Rudd (1983). Option Pricing. Homewood, IL: Irwin.
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.
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.
Chapter 7
DERIVATIVES AND PATH-DEPENDENT

DERIVATIVES: EXTENSIONS AND
GENERALIZATIONS OF THE LATTICE
APPROACH BY ACCOUNTING FOR
INFORMATION COSTS AND ILLIQUIDITY
Chapter Outline
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
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
Derivatives and Path-Dependent Derivatives 329
of implementing models already suggested in the literature. In this respect,

their contribution is more a numerical one than a financial one. This chapter
covers binomial and more general lattice approaches for the pricing of equity
and interest-rate dependent claims.
Dharan (1997) developed a simple framework for the valuation of path-
dependent interest-rate claims such as index-amortizing swaps or mortgage-
backed securities with a simple pre-payment function for which there is
no analytical solution. Dharan (1997) showed how to construct a lattice
to value a mortgage-backed security when the pre-payment function is
linear. The different models are illustrated in detail using several numerical
examples.
7.1. The Standard Lattice Approach for Equity Options:

The Standard Analysis
We re-call the standard lattice approach here in order to allow its extension
to account for the effects of information costs on the pricing of derivatives.
7.1.1. The model for options on a spot asset

with any pay outs
The lattice approach can be introduced by first looking at a stock option
whose underlying asset does not pay any dividend. Let T be the option’s
maturity date, which is divided into N reasonably small intervals of length
∆t, so that T = N ∆t. In this world, the expected return on the underlying
asset in time ∆t is r∆t. The variance of this underlying asset on the same
interval is σ2 ∆t. For the valuation of derivative assets, re-call that the
expected value of the underlying asset in a risk-neutral world is Ser∆t . In
this context, we can write the equality between this expected value and the
one given by the binomial model as:
pSu + (1 − p)Sd = Ser∆t or pu + (1 − p)d = er∆t (7.1)
Since the variance of a variable X is given by E(X 2 ) − E(X)2 , we can

write the equality between this variance and S 2 σ 2 ∆t:
S 2 σ 2 ∆t = S 2 (pu2 + (1 − p)d2 ) − S 2 (pu + (1 − p)d)2
or
σ 2 ∆t = pu2 + (1 − p)d2 − (pu + (1 − p)d)2 (7.2)

If we use Eqs. (7.1), (7.2) and suppose that u = d1 , then we have:

√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, and p = , a = er∆t .
u−d
The expected return on the stock can be written as pSu + (1 − p)Sd =

r∆t
Se . The variance on the stock can be written for the same interval ∆t as:
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:
Suj di−j where j = 0, 1, . . . , i.
The option is evaluated by starting at time T and working backward.
7.1.2. The model for futures options

Merton (1973), Black (1976), and Barone-Adesi and Whaley (1987) among
others, showed that futures contracts, stock index options, and currency
options may be assimilated to options on a stock that pays a continuous
dividend. In a risk-neutral economy, the expected return on an asset paying
a continuous dividend yield δ is (r − b) so that we can write e(r−b)∆t =
pu + (1 − p)d and a = e(r−b)∆t. Hence, the model for futures options is
completely specified using the following equations:
√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, p= , and a = e(r−b)∆t.
u−d
For a futures contract, r = b so that a = 1.
7.1.3. The model with dividends

We study three cases: a known dividend yield, a known proportional
dividend yield, and a discrete dividend.
7.1.3.1. A known dividend yield

The extension of the lattice approach in the presence of a continuous
dividend yield is simple. If a security pays a dividend yield q, then the
expected return on the underlying asset is (r − q). In this case, we can
write the equality between the expected value of the underlying asset and
the one given by the binomial model as pSu + (1 − p)Sd = Se(r−q)∆t or
pu + (1 − p)d = e(r−q)∆t .
In this case, using the same procedure as before, it can be shown that
the parameters for the binomial model are:
√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, p= , a = e(r−q)∆t .
u−d
The value of a European contingent claim Fi,j at time (t + i∆t) when
the underlying asset is Suj di−j can be calculated using the following
equation:
Fi,j = e(−r)∆t [(pFi+1,j+1 + (1 − p)Fi+1,j )]
In the same context, at time t + i∆t, the American call option value is:
Fi,j = max[Suj di−j − K, e−r∆t(pFi+1,j+1 + (1 − p)Fi+1,j )]
At each node, at time t + i∆t, the American put value is given by,
Fi,j = max[K − Suj di−j , e−r∆t(pFi+1,j+1 + (1 − p)Fi+1,j )].
It is convenient to note that these formulas apply to index options,

currencies, and futures contracts.
7.1.3.2. A known proportional 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 is as follows.
Assume that a known proportional dividend yield δ is to be paid at a
certain date. When there is only one dividend at the time (t + i∆t), the
nodes correspond to the stock prices Suj di−j for j = 0, 1, . . . , i where the
time (t + i∆t) is prior to the stock going ex-dividend.
√ √
a−d
The values of u, d, and p are u = eσ ∆t , d = e−σ ∆t , and p = u−d
respectively. When the time (t + i∆t) is after the underlying stock goes
ex-dividend, then the nodes give the following prices S(1 − δ)uj di−j for
j = 0, 1, . . . , i. The same analysis applies when there are several dividends.
In this case, the total dividend yield δi corresponding to all ex-dividend
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.
7.1.3.3. A known discrete dividend

In general, the dividend amount D is known, rather than the dividend yield.
Assume that there is only one dividend date, τ , between an instant k∆t
and (k + 1)∆t. When there is just one ex-cash income date τ , during the
option’s life, then the nodes on the tree at time (t + i∆t) are Suj di−j for
i ≤ k with j = 0, 1, 2, . . . , i. The nodes on the tree at time i = (k + 1) are:
(Suj di−j − D)u and (Suj di−j − D)d for j = 0, 1, 2, . . . , i. The analysis can
be simplified as before.
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
of the underlying asset in the nodes following nodes are generated as

follows:
(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 ).
7.1.4.1. The European put price with dividends

The values are computed by starting at the maturity date. At this date,
the possible option values are: P5,5 = 0, P5,4 = 0, S5,3 = 0.104, P5,2 =
9.3621, P5,1 = 16.7111, and P5,0 = 22.5446.
Using the recursive procedure, the European put values are: P4,4 = 0,
P4,3 = 0.0508, P4,2 = 4.6266, P4,1 = 12.8751, P4,0 = 19.4226, P3,3 = 0.0248,
P3,2 = 2.2861, P3,1 = 8.6183, P3,0 = 15.9673, P2,2 = 1.1294, P2,1 = 5.3610,
P2,0 = 12.1375, P1,1 = 3.1876, and P1,0 = 8.6274. The European put price
with dividends is equal to 5.8190.
For example, at the nodes (5,0) and (4,0), the European put price is
computed as: P5,0 = max[0, K − S5,0 ], P4,0 = [pP5,1 + qP5,0 ]/a.
7.1.4.2. The American put price with dividends

For the put, we have Pi,j = max[[pPi+1,j+1 + qPi+1,j ]/a; max[0, K − Si,j ]]
The values are computed by starting at the maturity date. At this
date, the possible option values are P5,5 = 0, P5,4 = 0, S5,3 = 0.104, P5,2 =
9.3621, P5,1 = 16.7111, and P5,0 = 22.5446. These values are the same
for both European and American options. Using the recursive procedure,
the European put values are P4,4 = 0, P4,3 = 0.0508, P4,2 = 5, P4,1 =
13.2485, P4,0 = 19.7961, P3,3 = 0.0248, P3,2 = 2.4685, P3,1 = 8.9887,
P3,0 = 16.3377, P2,2 = 1.2186, P2,1 = 5.6337, P2,0 = 12.5047, P1,1 = 3.3657,
and P1,0 = 8.9441.
For example, the values P4,4 and P4,3 are calculated as follows:
P4,4 = max(0.0508; max[0; 45 − 50.3914]) = max(0.0508; 0) = 0.0508.

P4,3 = max(4.6266; max[0; 45 − 40]) = max(4.6266; 5) = 5.
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
We can obtain a more accurate estimate at time t − 2∆t by assuming

a stock price S at this time. In this case, the ∆ is
F2,2 − F2,0
∆=
Su2 − Sd2
The Γ is given by:
F2,2 −F2,1 F2,1 −F2,0

Su2 −S − S−Sd2
Γ= 1 2 2
2 (Su − Sd )
For an introduction to the Greek letters and their use by market

participants, the reader can refer to Chapter 3. The Greek-letter delta
corresponds to the option partial derivative with respect to the underlying
asset price. The gamma indicates the partial derivative of the delta with
respect to the underlying asset price. The theta indicates the option
partial derivative with respect to time. The vega indicates the option
partial derivative with respect to the volatility parameter. The following
Tables 7.1–7.8 show the simulations of binomial option prices using
150 periods for different parameters. The reader can see at the same time the
sensitivities of the option price to different parameters (the Greek letters).
These parameters are provided for illustrative purposes.
The binomial model can also be used for the pricing of foreign currency
options. In this case, the cost of carry or the risk-free rate is replaced by
the difference between the domestic risk-less rate r and the foreign risk-less
rate r∗ . In this case, the underlying asset price S indicates the exchange
rate (Table 7.9).
Table 7.1. Simulations of European binomial call prices,

S = 100, K = 100, t = 22/12/2003, T = 22/12/2004, r = 2%,
and σ = 20%.
S Price Delta Gamma Vega Theta
80 1.43986 0.18415 −0.00000 0.21391 −0.00656

85 2.54367 0.24220 0.06668 0.26153 −0.00827
90 4.17276 0.38248 −0.00000 0.34631 −0.01107
95 6.29601 0.46056 0 0.37528 −0.01239
100 8.91144 0.61818 −0.00000 0.39070 −0.01336
105 12.06935 0.69165 0 0.37547 −0.01345
110 15.63786 0.75817 0 0.34667 −0.01313
115 19.55770 0.81605 0 0.30750 −0.01246
120 23.76469 0.86445 0 0.26200 −0.01154
Table 7.2. Simulations of American binomial call prices,

S = 100, K = 100, t = 22/12/2003, T = 22/12/2004, r = 2%,
and σ = 20%.
80 1.43986 0.18415 −0.00000 0.21391 −0.00656

85 2.54367 0.24220 0.06668 0.26153 −0.00827
90 4.17276 0.38248 −0.00000 0.34631 −0.01107
95 6.29601 0.46056 0 0.37528 −0.01239
100 8.91144 0.61818 −0.00000 0.39070 −0.01336
105 12.06935 0.69165 0 0.37547 −0.01345
110 15.63786 0.75817 0 0.34667 −0.01313
115 19.55770 0.81605 0 0.30750 −0.01246
120 23.76469 0.86445 0 0.26200 −0.01154
Table 7.3. Simulations of European binomial put prices,

S = 100, K = 100, t = 22/12/2003, T = 22/12/2004, r = 2%,
and σ = 20%.
80 19.45436 −0.81585 0 0.21391 −0.00119

85 15.55817 −0.75780 0.06668 0.26153 −0.00289
90 12.18726 −0.61752 0 0.34631 −0.00570
95 9.31050 −0.53944 −0.00000 0.37528 −0.00702
100 6.92594 −0.38182 0 0.39070 −0.00799
105 5.08385 −0.30835 −0.00000 0.37547 −0.00808
110 3.65236 −0.24183 −0.00000 0.34667 −0.00776
115 2.57220 −0.18395 −0.00000 0.30750 −0.00709
120 1.77919 −0.13555 −0.00000 0.26200 −0.00617
Table 7.4. Simulations of American binomial put prices,

S = 100, K = 100, t = 22/12/2003, T = 22/12/2004, r = 2%,
and σ = 20%.
80 20.32877 −0.88286 0.01314 0.14678 −0.00244

85 16.16444 −0.79424 0.05407 0.23133 −0.00417
90 12.58896 −0.65043 0.00726 0.32805 −0.00657
95 9.58087 −0.55677 0.00338 0.36977 −0.00777
100 7.10987 −0.43411 0.03640 0.39021 −0.00857
105 5.19837 −0.31985 0.00356 0.37831 −0.00853
110 3.72566 −0.24764 0.00126 0.34991 −0.00807
115 2.61920 −0.18740 0.00100 0.31041 −0.00731
120 1.80907 −0.13757 0.00057 0.26443 −0.00633
Table 7.5. Simulations of European binomial call prices,

S = 100, K = 100, t = 22/12/2003, T = 22/06/2004, r = 2%,
and σ = 20%.
80 0.36079 0.08693 0 0.09055 −0.00526

85 0.93264 0.16894 −0.00000 0.15281 −0.00903
90 2.00036 0.28849 −0.00000 0.21932 −0.01322
95 3.70780 0.43730 0 0.26800 −0.01658
100 6.11625 0.59557 −0.00000 0.27901 −0.01794
105 9.25379 0.67066 0.06878 0.26813 −0.01807
110 12.96246 0.80000 0 0.21954 −0.01604
115 17.11629 0.85123 0.04165 0.18706 −0.01475
120 21.61974 0.92542 0 0.12028 −0.01144
Table 7.6. Simulations of American binomial call prices,

S = 100, K = 100, t = 22/12/2003, T = 22/06/2004, r = 2%,
and σ = 20%.
80 0.36079 0.08693 0 0.09055 −0.00526

85 0.93264 0.16894 −0.00000 0.15281 −0.00903
90 2.00036 0.28849 −0.00000 0.21932 −0.01322
95 3.70780 0.43730 0 0.26800 −0.01658
100 6.11625 0.59557 −0.00000 0.27901 −0.01794
105 9.25379 0.67066 0.06878 0.26813 −0.01807
110 12.96246 0.80000 0 0.21954 −0.01604
115 17.11629 0.85123 0.04165 0.18706 −0.01475
120 21.61974 0.92542 0 0.12028 −0.01144
Table 7.7. Simulations of European binomial put prices,

S = 100, K = 100, t = 22/12/2003, T = 22/06/2004, r = 2%,
and σ = 20%.
80 19.36306 −0.91307 0 0.09055 0.00017

85 14.93492 −0.83106 0 0.15281 −0.00360
90 11.00263 −0.71151 −0.00000 0.21932 −0.00779
95 7.71007 −0.56270 0 0.26800 −0.01115
100 5.11852 −0.40443 0 0.27901 −0.01252
105 3.25606 −0.32934 0.06878 0.26813 −0.01265
110 1.96473 −0.20000 −0.00000 0.21954 −0.01061
115 1.11857 −0.14877 0.04165 0.18706 −0.00933
120 0.62201 −0.07458 −0.00000 0.12028 −0.00602
Table 7.8. American binomial — CRR, put S = 100, K = 100,

t = 22/12/2003, T = 22/06/2004, r = 2%, and σ = 20%.
80 20.00779 −0.97909 0.01600 0.02403 −0.00077

85 15.33467 −0.87409 0.01205 0.11948 −0.00456
90 11.24360 −0.73921 0.00890 0.20431 −0.00862
95 7.85080 −0.58226 0.00764 0.26240 −0.01179
100 5.20147 −0.45140 0.03865 0.27868 −0.01303
105 3.30000 −0.33283 0.06361 0.26892 −0.01297
110 1.98701 −0.20271 0.00120 0.22094 −0.01083
115 1.13111 −0.14941 0.04099 0.18764 −0.00943
120 0.62757 −0.07518 0.00028 0.12100 −0.00609
Table 7.9. CRR binomial call currency price, S = 1, K = 1,

t = 07/02/2003, T = 07/02/2004, r = 3%, r∗ = 4%, and
σ = 20%.
0.96 0.05371 0.45561 0.50475 0.00356 0.00008

0.97 0.05826 0.46131 0.49915 0.00374 0.00008
0.98 0.06288 0.46131 0.49921 0.00377 0.00009
0.99 0.06749 0.46131 0.49921 0.00380 0.00009
1 0.07210 0.53768 0.42285 0.00382 0.00009
1.01 0.07748 0.53768 0.42285 0.00385 0.00008
1.02 0.08286 0.53768 0.42285 0.00388 0.00008
1.03 0.08823 0.53768 0.42285 0.00391 0.00008
1.04 0.09361 0.60551 0.35503 0.00394 0.00008
7.2. A Simple Extension to Account for Information

Uncertainty in the Valuation of Futures and Options
We can extend the previous analysis to account for the effects of information
costs. Information costs can be used in the valuation of futures and options.
For introduction, we refer to Appendices A and B. 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.
7.2.1. On the valuation of derivatives

and information costs
An important question in financial economics is how frictions affect
equilibrium in capital markets since in a world of costly information, some
investors will have incomplete information. Merton (1987) developed a
simple model of capital market equilibrium with incomplete information,
CAPMI. The CAPMI model can explain several anomalies in financial
markets.
Merton (1987) advanced the investor recognition hypothesis (IRH) in a
mean-variance model. This assumption explains the portfolio formation of
informationally constrained investors (ICI). The IRH in Merton’s context
states that investors buy and hold only those securities about which they
have enough information. Merton (1987) adapted the rational framework
of the static CAPM to account for incomplete information. The premise
in Merton’s (1987) model and Shapiro’s (2000) extension is that the costs
of gathering and processing data lead some investors to focus on stocks
with high visibility and also to entrust a portion of their wealth to money
managers employed by pension plans.
In this context, a trading strategy shaped by real-world information
costs should incorporate an investment in well-known, visible stocks, and
an investment delegated to professional money managers. In this theory,
an investor considers only the stocks visible to him/her, i.e., those about
which he/she has sufficient information to implement the optimal portfolio
re-balancing. In general, information about larger firms is likely to be
available at a lower cost. The claim that large firms are more widely
known is consistent with the empirical evidence that large firms have
more shareholders as in Merton (1987). For these reasons, it is important
to account for information costs in the pricing of assets and derivatives.
The works of Markowitz (1952), Sharpe (1964), and Lintner (1965) on

the capital asset pricing model provided the general equilibrium model of
asset prices under uncertainty. This model represents a fundamental tool in
measuring the risk of a security under uncertainty. The CAPM model can
be applied to the valuation of options and futures contracts.
Merton’s model is a two-period model of CAPM in an economy where
each investor has information about only a subset of the available securities.
The main assumption in the Merton’s model is that an investor includes
an asset S in his/her portfolio only if he/she has some information about
the first and second moment of the distribution of its returns. In this
model, information costs have two components: the costs of gathering and
processing data for the analysis and the valuation of the firm and its
assets, and the costs of information transmission from an economic agent
to another. Merton’s model may be stated as follows:
R̄S − r = βS [R̄m − r] + λS − βS λm
where:
• R̄S : the equilibrium-expected return on security S;

• R̄m : the equilibrium-expected return on the market portfolio;
• R: one plus the risk-less rate of interest, r;
• βS = cov( R̃S /R̃m )
var(R̃m )
: the beta of security S;
• λS : the equilibrium aggregate “shadow cost” for the security S and
• λm : the weighted average shadow cost of incomplete information over all
securities in the market place.
The CAPM of Merton (1987), referred to as the CAPMI is an extension

of the standard CAPM to a context of incomplete information. Note
that when λm = λS = 0, this model reduces to the standard CAPM of
Sharpe (1964).
Since the publication of the pioneering papers by Black and Scholes
(1973) and Merton (1973), three industries have blossomed: an exchange
industry in derivatives, an OTC industry in structured products, and an
academic industry in derivative research. As it appears in Scholes (1998),
derivative instruments provide low-cost solutions to investor problems than
that of competing alternatives. Differences in information are important
in both financial and real markets (see the models in Bellalah 1999a,b;
Bellalah and Jacquillat, 1995).
7.2.2. The valuation of forward and futures contracts

in the presence of information costs
7.2.2.1. Forward, futures, and arbitrage
We denote respectively by:
• 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.
For an introduction to information costs and their use in the valuation

of derivatives, we can refer to Bellalah and Jacquillat (1995) and Bellalah
(2000a,b, 2001). It is important to make a difference between the price
and value of a contract. The forward price of a contract corresponds to its
delivery price that would make its value equal to zero. When a contract
is initiated, the delivery price is fixed equal to the forward price in such a
way that f = 0 and K = F . Bellalah shows that these costs of information
explain to a large extent the financial crisis of 2008 because of a lack of
information transmission between economic agents.
7.2.2.2. The valuation of forward contracts in the absence

of distributions to the underlying asset
The underlying asset may be a non-dividend paying stock or a non-coupon-
bearing bond. Consider an investment in two portfolios.
Portfolio A corresponds to a long position in a forward contract and a
cash amount equal to Ke−r(T −t) .
Portfolio B contains the underlying security of the forward contract.
At maturity, cash can be used to pay for the security. However, before
buying the security, an investor pays information costs and therefore needs
an additional return of λS before investing in the asset.
To implement arbitrage, the investor must be informed about the
markets and pays information costs. Hence, arbitrage considerations imply
that the real discount rate must be e−(r+λS )(T −t) rather than e−r(T −t) .
The equivalence between these values of the two portfolios at each time
implies that
f + Ke−(r+λS )(T −t) = S or f = S − Ke−(r+λS )(T −t) .
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) .
7.2.2.3. The valuation of forward contracts in the presence

of a known cash income to the underlying asset
The underlying asset may be a dividend-paying stock or a coupon-bearing
bond. Consider an investment in two portfolios.
Portfolio A corresponds to a long position in a forward contract and a
cash amount equal to Ke−r(T −t) .
Portfolio B contains the underlying security of the forward contract
plus borrowings of an amount I corresponding to the known income.
Since income can be used to re-pay the loan, portfolio B has the same
value as one security or portfolio A at T . However, before using the security,
an investor pays information costs and therefore needs an additional return
of λS before investing in the security. Both portfolios must have the same
initial value and f +Ke−(r+λS )(T −t) = S −I or f = S −I −Ke−(r+λS )(T −t) .
As before, the forward price F can be computed as the value of K for
which, f = 0, or F = (S − I)e(r+λS )(T −t) .
7.2.2.4. The valuation of forward contracts in the presence

of a known dividend yield to the underlying asset
The underlying asset may be a stock index or a currency. A known dividend
yield corresponds to an income expressed as a percentage of the security
price. Consider an investment in two portfolios A and B. Portfolio A is
conserved and portfolio B contains an amount e−q(T −t) of the security with
income re-invested. Hence, the fraction of the security in B will grow as a
result of the dividends so that at maturity, one security is held.
At T , portfolios A and B have the same value and this must be true at
time t to give:
f + Ke−(r+λS )(T −t) = Se−q(T −t) or f = Se−q(T −t) − Ke−(r+λS )(T −t) .
As before, the forward price F can be computed as the value of K for

which f = 0, or F = Se(r−q+λS )(T −t) .
7.2.2.5. The valuation of stock index futures

A stock index is constructed using some fixed number of stocks. The stocks
can have equal weights or weights that change over time. The stock index
can be seen as the price of an asset that pays dividends. In general, it
is a reasonable approximation for some indices to assume that they pay
a continuous dividend yield. In this case, it is possible to use arbitrage
arguments as before to show that the futures price must satisfy the following
relationship:
F = Se(r−q+λS )(T −t) .
7.2.2.6. The valuation of Forward and futures contracts on currencies

Currency forward and futures contracts can be analyzed using arbitrage
arguments. Consider as before two portfolios A and B. The first portfolio A
corresponds to a long position in a forward contract plus an amount of cash,
which is equal to Ke−r(T −t) . Portfolio B contains an amount e−rf (T −t) of
the foreign currency. The value of the two portfolios at time T is equivalent
to one unit of the foreign currency. Portfolios A and B must have the same
value at time t to give:
f + Ke−(r+λS )(T −t) = Se−rf (T −t)
or
f = Se−rf (T −t) − Ke−(r+λS )(T −t) .
As before, the forward price F or the forward exchange rate can be

computed as the value of K for which f = 0 in this last equation, or:
F = Se(r−rf +λS )(T −t) .
This corresponds to the interest-rate parity theorem in international

finance in the presence of information uncertainty. These last equations are
identical to those for stock indices, where q is replaced by rf . This result
reveals that a foreign currency is similar to a security paying a continuous
dividend yield. The latter corresponds to the foreign risk-free interest rate.
7.2.2.7. The valuation of futures contracts on silver and gold

The underlying assets of these contracts are held by several investors for
investment purposes. This is not the case for several other commodities.
When storage costs are neglected, silver and gold can be seen as securities
paying no income. In this case, the futures price must satisfy the following
relationship:
F = Se(r+λS )(T −t)
where λS corresponds to information costs on the security. We denote by

G the present value of the storage costs. In the same context, we have
F = (S + G)e(r+λS )(T −t) .
When storage costs are proportional to the commodity price, then
F = Se(r+g+λS )(T −t) where g corresponds to the storage cost per annum.
7.2.2.8. The valuation of Futures on other commodities

Several commodities are held by investors for some reasons other than
investment purposes. This is the case for commodities held in inventory
because of their consumption values. In this case, investors conserve the
commodity and do not buy futures contracts because they cannot be
consumed. This is one of the reasons given to explain the decrease in futures
prices for longer maturities. In general, holders of commodity positions
feel some benefits from holding physical commodities such as the ability
to keep a production process running. These benefits are not available
for the holder of a futures contract. The benefits are referred to as the
convenience yield (cy). In the presence of information uncertainty, the cy
can be defined as:
F ecy(T −t) = (S + G)e(r+λS )(T −t) .
When storage costs are expressed as a constant proportion, g, then y is

defined so that: F ecy(T −t) = Se(r+g+λS )(T −t). or F = Se(r+g−cy+λS )(T −t) .
When a commodity is held only for investment purposes, the cy is zero
and we have F = (S + G)e(r+λS )(T −t) or F = Se(r+g+λS )(T −t) .
7.2.3. Arbitrage and information costs in the lattice

approach
The lattice approach can be introduced with reference to a stock option in
the absence of payments to the underlying asset.
Let T be the option’s maturity date which is divided into N reasonably

small intervals of length, so that T = N ∆t. During each time interval, the
stock price moves either upwards from S to Su or downwards from S to dS.
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 u, d, and p are functions of the mean and variance of the rates
of return on S during the interval. The basic lattice approach as 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 N equal intervals of length ∆t during which the stock price
moves from its initial value S to one of the two new values Su and Sd with
probabilities p and (1 − p), respectively.
It is a simple matter to extend the binomial approach to account for
information costs. The acquisition of information and its dissimination
are central activities in finance, and especially in capital markets. Merton
(1987) provided a model of CAPMI to provide some insights into the
behavior of security prices. From Merton’s model (1987), it appears that
taking into account the effect of incomplete information on the equilibrium
price of an asset is similar to applying an additional discount rate to this
asset’s future cash flows. In fact, the expected return on the asset is given
by the appropriate discount rate that must be applied to its future cash
flows. In fact, in a risk-neutral world, it is possible to show that we have the
appropriate value for u, d, and p. In this world, the expected return on the
underlying asset in time ∆t is (r + λS )∆t. The variance of this underlying
asset on the same interval is σ2 ∆t.
The term λS indicates the shadow cost of incomplete information for
the asset S. Consider the previous analysis in the presence of an information
cost λ. This cost can be seen as a shadow cost or a marginal cost reflecting
the return required by market participants to include some assets in their
portfolios. It is well known that there are thousands of financial assets
in financial markets. However, investors decide to include only some of
these assets in their portfolios. They suffer “a sunk cost” in collecting data,
analysing financial products and implementing financial models. These
costs are necessary in the choice of financial assets and the implementation
of some strategies. Therefore, investors need to be compensated by a
return corresponding to these costs as it appears in Merton’s (1987) model.
Consider, for example, a financial institution using a given market. If the
costs of portfolio selection and models conception, etc. are computed, then
it can require at least a return of say, for example λ = 3%, before acting in
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:
pSu + (1 − p)Sd = Se(r+λS )∆t .
The variance on the stock can be written for the same interval ∆t as:
pS 2 u2 + (1 − p)S 2 d2 − (Se(r+λS )∆t )2 .
The last expression can be written as:

√ √
S 2 (e(r+λS )∆t (eσ ∆t
+ e−σ ∆t
) − 1 − e2(r+λS )∆t ).
For the valuation of derivative assets, the expected value of the

underlying asset in a risk-neutral world of Merton (1987) is Se(r+λS )∆t .
In this context, we can write the equality between this expected value and
the one given by the binomial model: pSu + (1 − p)Sd = Se(r+λS )∆t or:
pu + (1 − p)d = e(r+λS )∆t .
If we use the above equations and assume that u = d1 , then we have:
√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, p= , a = e(r+λs )∆t .
u−d
In a risk-neutral world, the option value at time T −∆t can be calculated

as the expected value at the maturity date T discounted at (r + λc ) for the
time ∆t. In the same way, the values of the derivative security can be
calculated at time T − 2∆t as the expected value at time T − ∆t discounted
at (r + λc ) for the time ∆t and so on. If we define the value of a contingent
claim as Fi,j at time t + i∆t when the underlying asset is Suj di−j , then the
value of a European option with information costs can be computed using
Fi,j = e−(r+λc )∆t [(pFi+1,j+1 + (1 − p)Fi+1,j )].
At each node, at time t + i∆t, the American call option value is given by:
Fi,j = max[Suj di−j − K, e−(r+λc )∆t (pFi+1,j+1 + (1 − p)Fi+1,j )].

At maturity, the value of a European put on a non-dividend paying

asset is:
FN,j = max[K − Suj dN −j , 0]
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:
Fi,j = max[K − Suj di−j , e−(r+λc )∆t (pFi+1,j+1 + (1 − p)Fi+1,j )].
7.2.4. The binomial model for options in the presence

of a continuous dividend stream and
information costs
The extension of the lattice approach in the presence of information costs
and a continuous dividend yield is simple. If a security pays a dividend yield
q, then the expected return on the underlying asset is (r + λc − q). In this
case, we can write the equality between the expected value of the underlying
asset and the one given by the binomial model as: pSu + (1 − p)Sd =
Se(r+λS −q)∆t or: pu + (1 − p)d = e(r+λS −q)∆t . In this case, using the same
procedure as before, it can be shown that the parameters for the binomial
model are:
√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, p= , and a = e(r+λs −q)∆t .
u−d
The value of a European contingent claim Fi,j at time t + i∆t when
the underlying asset is Suj di−j , in the presence of information uncertainty
can be calculated using the following equation:
Fi,j = e−(r+λc )∆t [(pFi+1,j+1 + (1 − p)Fi+1,j )].
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:


of a known dividend yield and information costs
The lattice or binomial approach can be easily modified when the underly-
ing asset pays a known dividend yield, δ.
At an instant t + i∆t, prior to the ex-dividend date, stock prices are
given by: Suj di−j for j = 0, 1, . . . , i. At an instant (t + i∆t), just after the
ex-dividend date, stock prices are given by:
S(1 − d)uj di−j for j = 0, 1, . . . , i.

of a discrete dividend stream and information costs
When there is just one ex-cash income date τ , during the option’s life and
k∆t ≤ τ ≤ (k + 1)∆t, then at time y, the value of the stochastic component
S is given by:
S ∗ (y) = S(y) when y > τ,
and
S ∗ (y) = S(y) − Di e−(r+λc )(τ −y) when y ≤ τ.
Assume σ ∗ is the constant volatility of S ∗ . Using the parameters p, u,

If i∆t < τ : S ∗ (t)uj di−j + De−(r+λc )(τ −i∆t) j = 0, 1, . . . , i
If i∆t ≥ τ : S ∗ (t)uj di−j j = 0, 1, . . . , i.
7.2.7. The binomial model for futures options

in the presence of information costs
In a risk-neutral economy, the expected return on an asset paying a
continuous dividend yield q is (r − q).
In this context, the equations:
√ √ a−d
u = eσ ∆t
, d = e−σ ∆t
, p= ,
u−d
are used, except for a, which must be re-written as: a = e(r−q+λc )∆t . For a
futures, contract, r + λc = b and a = 1. When information costs are equal
to zero, then r = b and a = 1.
7.2.8. The lattice approach for American options

with information costs and several
cash distributions
7.2.8.1. The model
When the amounts of payments are accounted for, it can be shown that a
sufficient condition for optimal exercise is:
(Di + Ri ) > K[1 − e−(r+λS )(ti+1 −ti ) ]
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:
(Di + Ri ) < K[1 − e−(r+λS )(ti+1 −ti ) ]
with (Di +Ri ) ≥ 0. It is a simple matter to extend this approach to account

for information uncertainty. When u = 1/d, it can be shown that:
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:
Suj di−j where 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:
The discounting factor is adjusted by information costs in the option

market to reflect sunk costs paid in this market. The American put is
given by:

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:
S ∗ (y) = S(y) when y > τ

S ∗ (y) = S(y) − (Di + Ri)e−(r+λS )(τ −y) when y ≤ τ
Assume σ ∗ is the constant volatility of S ∗ . Using the parameters p, u,

If i∆t < τ : S ∗ (t)uj di−j + (Di + Ri)e−(r+λS )(τ −i∆t) j = 0, 1, . . . , i
If i∆t ≥ τ : S ∗ (t)uj di−j j = 0, 1, . . . , i.
7.3. The Binomial Model and the Risk Neutrality: Some

Important Details
Nawalkha and Chambers (1995) re-examine the consistency of the binomial
option pricing model with the risk-neutrality argument of Cox and Ross
(1976). They show that risk neutrality in discrete time is a consequence of
a specific choice of binomial parameters by Cox et al. (CRR) (1979).
7.3.1. The binomial parameters and risk neutrality

The original discrete-time binomial model of CRR, can be presented as
follows. Consider at time t = 0, a call option with a maturity date T .
Let T be divided into N number of sub-intervals. Denote the current time
T T
by t = N (N − 1). At the current time, the option is N periods from the
expiration date. In the next period, the stock S goes up to uS or down to dS
with u ≥ d. The probability of an upward movement is q. The probability
of a downward movement is (1 − q).
As shown by CRR (1979), the call price at the current time is:
C = [pu C u + pd C d ]/rh (7.3)
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.
The binomial parameters specified by CRR are:
u = exp(σ(T /N )0.5 )
d = exp(−σ(T /N )0.5 )
q = (1/2)[1 + (µ/σ)(T /N )0.5 ).
In this expression, the term µ indicates the preference parameter.

In this model, the probability q is between 0 and 1, so −σ(T /N )0.5 <
µ(T /N )0.5 < σ(T /N )0.5 . Since the expressions of u and d do not contain µ,
the risk-neutral probabilities pu and pd and the call prices C u and C d
are independent of preferences. The choice of binomial parameters is not
unique. For example, Jarrow and Rudd (1983) specify u, d, and q as:
u = exp[µ(T /N ) + σ(T /N )0.5 ] (7.4)

0.5
d = exp[µ(T /N ) − σ(T /N ) ] (7.5)
q = (1/2).
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:
u = exp[m(T /N ) + σ(T /N )0.5 ]

d = exp[m(T /N ) − σ(T /N )0.5 ] (7.6)
q = (1/2)[1 + ((µ − m)/σ)(T /N )0.5 ]
with
−σ(T /N )0.5 < (µ − m)(T /N ) < σ(T /N )0.5 .
This choice of the binomial parameters leads to the following three

moments over a discrete subinterval: a mean µ(T /N ), a variance σ 2 (T /N )−
(µ− m)2 (T /N )2 and the skewness which equals 2(µ− m)[(µ− m)2 (T /N )3 −
σ 2 (T /N )2 ]. Again, the terms (T /N )2 and (T /N )3 become insignificant in
comparison to (T /N ) as N tends to infinity.
In this case, the variance and the skewness of the binomial process
implied by these parameters converge to the variance and skewness of the
log-normal process in the continuous time limit.
Equation (7.6) and other equations given under Eq. (7.6) imply that
the probabilities and option prices depend upon m. Investors may disagree
about the preference parameter µ and agree on m.
In the CRR model, the parameter m = 0.
In the Jarrow and Rudd (1983) model, the parameter µ = m.
A unique equivalent probability measure à la Harrison and Kreps (1979)
exists and implies that the stock price discounted at the risk-less rate is a
martingale under this measure. In this setting, the risk-neutral probabilities
must be greater than zero in Eq. (7.3). Hence, we must have for the CRR
choice that:
exp[−σ(T /N )0.5 ] < rh < exp[σ(T /N )0.5 ].
We must have for the Jarrow and Rudd (1983) choice that:
exp[µ(T /N ) − σ(T /N )0.5 ] < rh < exp[µ(T /N ) + σ(T /N )0.5 ].
We must have for the revealed preference parameters in Nawalkha and

Chambers (1995) choice that:
exp[m(T /N ) − σ(T /N )0.5 ] < rh < exp[m(T /N ) + σ(T /N )0.5 ].
The analysis shows that when m = µ, the binomial model is

inconsistent with the risk-neutrality argument of Cox and Ross (1976)
and the model is preference dependent. When m differs from µ, the risk
neutrality can still hold because investors are allowed to disagree with µ.
In this case, different values of m can generate different option prices in
discrete time. These limitations disappear in the continuous time limit. In
fact, when N tends to infinity, the Black and Scholes (1973) formula is
obtained. Hence, the binomial model is independent of the parameter m

only in the continuous time limit.
7.3.2. The convergence argument

We use the binomial equation for the general choice of binomial parameters.
The call price at time t between 0 and T is given by:
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
Using Taylor series expansions of C(exp[mh + σ(h0.5 )]St , T − (t + h))

and C(exp[mh − σ(h0.5)]St , T − (t + h)) around the point (St , (T − t)) gives:
C(exp[mh + σ(h0.5 )]St , T − (t + h)) = Ct

∂Ct
+ [exp[mh + σ(h0.5 )] − 1]St
∂St
1 ∂ 2 Ct ∂Ct
+ [exp[mh + σ(h0.5 )] − 1]2 St2 2 + h + ···+ (7.8)
2 ∂ St ∂t
A similar expression can be obtained for C(exp[mh − σ(h0.5 )]St , T −

(t + h)) with −σ(h0.5 ) instead +σ(h0.5 ). Using Taylor series for:
exp[mh + σ(h0.5 )] = 1 + [mh + σ(h0.5 )]

1
+ [mh + σ(h0.5 )]2 + · · · + (7.9)
2
A similar expression can be obtained for exp[mh − σ(h0.5 )]. A Taylor

series expansion of rh gives:
1
rh = exp(h · logr) = 1 + h · logr + (h · logr)2 + · · · + (7.10)
2
Substituting Eqs. (7.8), (7.9), (7.10) in Eq. (A2) gives:
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.
7.4. The Hull and White Trinomial Model for Interest

Rate Options
Hull and White (1993) presented a general numerical procedure involving
the use of trinomial trees for constructing one-factor models, which are
consistent with initial market data where the short rate follows a Markovian
process. Their procedure is efficient and provides a convenient way of
implementing models already suggested in the literature. It should be noted
in passing that a one-period trinomial tree is somehow equivalent to a
standard two-period binomial tree. For example, they studied the case
where the process for r(t) has the general form considered by Hull and
White (1990), also called the extended Vasicek model:
dr = µ[θ, r, t]dt + σdW (t)
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,
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:
µn,j = µ[θ(n∆t), r0 + j∆r, n∆t)]. (7.11)
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.
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.
7.5. Pricing Path-Dependent Interest Rate Contingent

Claims Using a Lattice
Dharan (1997) proposed a framework for the pricing of path-dependent
interest rate derivatives such as index-amortizing swaps or mortgage-
backed securities with a simple pre-payment function for which there is
no analytical solution. He showed how to construct a lattice to value a
mortgage-backed security when the pre-payment function is linear. This
section develops the models using numerical examples.
7.5.1. The framework

Let us denote the interest rate by r. The interest rate goes up to ru with
a probability pu or down to rd with a probability pd = 1 − pu . In this case,
the interest goes up to ru u with a probability pu u, etc. The superscripts,
u and d, represent the sequence of up and down movements through the
two-period lattice.
If we denote by Ψ(i, j) the Arrow-Debreu price at a node corresponding
to the ith period and jth node from the bottom, counting from zero, then:
pd pu pd pdd
Ψ(0, 0) = 1, Ψ(1, 0) = (1+r) , Ψ(1, 1) = (1+r) , Ψ(2, 0) = (1+r) × (1+r d) , =
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 )
Using Eq. (7.12), it is possible to construct a lattice of Arrow-Debreu

prices using forward induction. The Arrow-Debreu prices at a node is a
summary measure of all paths leading to that node.
Consider a path-dependent interest rate derivative. Its original prin-
cipal is F and its maturity date is T with T = n∆t. The principal is
amortized at a given rate a(i, j) that is node dependent. When there are
k paths arriving at a node, the principal corresponding to each path is
denoted by F (i, j, k). The security pays the difference between the principal
in the previous period and its currently amortized value (representing the
prepayments for a mortgage-backed security.
We denote this difference by P (i, j, k):
P (i, j, k) = F (i − 1, l, k) − F (i, l, k) = a(i, j)F (i − 1, l, k) (7.13)
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)
Equation (7.14) use some new notations.

Let us denote by r(i, j) and p(i, j), the upward jump in the spot rate and
its probability, respectively and by F ∗ (i, j) the Arrow-Debreu price of the
principal payments at node (i, j). The second and third lines in Eq. (7.14)
represent the values for the nodes located at the lower and upper edges of
the lattice. The values for interior nodes are given by the last line.
The forward induction procedure begins at date 1. C ∗ (i, j) are cal-

culated as follows.
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)
The values of P ∗ (i, j) at similar nodes are computed as follows:
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)
At time 0, the value of the security V (0, 0) corresponds to the

sum of the Arrow-Debreu prices of all cash flows until maturity. It is
computed as:
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).
Assuming an initial principal of 1$, the induction method for these

prices is:
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:
V (i, j, k) = v(i, j)F (i, j, k) (7.19)
7.5.2. Valuation of the path-dependent security

Dharan (1997) developed two examples in the illustration of Eqs. (7.14)–
(7.17). The first is a fixed-rate mortgage corresponding to a mortgage-
backed security where the underlying pool pays a fixed coupon. The second
example is an adjustable-rate mortgage corresponding to a mortgage-
backed security based on a pool paying a floating-rate coupon.
The examples are based on a specific model for interest rates and the
following simple pre-payment function (simlar to that in Hull and White
(1993)):

 0 r(i, j) ≥ 0.05,




0.05
a(i, j) = 0, 45 −1 0.03 < r(i, j) < 0.05,

 r(i, j)



0.3 r(i, j) ≤ 0.03
7.5.2.1. Fixed-coupon rate security

The original face value of the security is 100 dollars and the value of
coupon is 5%.
In this case, it is possible to construct a tree for four periods where

the cash flows at each node are the sum of the prepayments, the coupons
paid, and the remaining principal at maturity. The different cash flows are
discounted to the present.
Equations (7.14)–(7.16) are used to compute at each node the Arrow-
Debreu prices of the principal, coupons, and prepayments, i.e., F ∗ (i, j),
C ∗ (i, j), and P ∗ (i, j).
7.5.2.2. Floating-coupon security

The floating-rate coupon is assumed to be the spot interest rate prevailing
in the previous period. Equations (7.14) and (7.16) are used to build the
lattice. Equation (7.17) gives the security price.
7.5.3. Options on path-dependent securities

Equation (7.18) is used to compute the value of the underlying security for
an original principal of one dollar. It is necessary to get the possible values
of the principal at maturity. Dharan (1997) proposed two cases: short-dated
options for which the number of steps is less than 12 and long-dated options
where the difference.
7.5.3.1. Short-dated options

Equation (7.19) gives the correct values of the underlying security for each
value of the principal at a particular node. The option payoff is given by:
V opt (s, j, k) = max(V (s, j, k) − X, 0) (7.20)
where V opt (s, j, k) corresponds to the option value at maturity s and X is

the strike price.
The American option price is computed using:
V opt (i, j, k) = max[(V (i, j, k) − X)

V opt (i + 1, j + 1, k)p(i, j) + V opt (i + 1, j, k)(1 − p(i, j))
(7.21)
1 + r(i, j)
7.5.3.2. Long-dated options

Let us denote by z the number of values of the principal stored at each
node.
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:
V1opt − V0opt (F1 − F0 )(F1 − F1 )

V1opt = V0opt + (F1 − F0 ) × +
F1 − F0 (F2 − F0 )

V2opt − V1opt V opt − V0opt
× − 1 .
F2 − F1 F1 − F0
The present value option is obtained by repeating the interpolation at

each step and discounting the resulting value.
This model and several of its applications to the pricing of path-
dependent claims are provided by Dharan (1997).
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
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.
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
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.
Cox, JC and SA Ross (1976). The valuation of options for alternative stochastic
processes. Journal of Financial Economics, 3, 145–166.
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
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.
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, 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.

Part III
Option Pricing in a Continuous-Time Setting: Basic

Models, Extensions and Applications
365
366
Chapter 8
EUROPEAN OPTION PRICING

MODELS: THE PRECURSORS OF THE
BLACK–SCHOLES–MERTON THEORY AND
HOLES DURING MARKET TURBULENCE
Chapter Outline
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
10. Appendix 8.B provides an approximation of the bivariate normal

density function.
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
European Option Pricing Models 369
(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.
8.1. Precursors to the Black–Scholes Model

The story began in 1900 with a doctoral dissertation at the Sorbonne in
Paris, France, in which Louis Bachelier gave an analytical valuation formula
for options.
8.1.1. Bachelier formula

Using an arithmetic Brownian motion for the dynamics of share prices and
a normal distribution for share returns, Bachelier obtained the following
formula for the valuation of a European call option on a non-dividend

paying stock:

S −K S−K √ K −S
c(S, T ) = SN √ − KN √ + σ Tn √ (8.1)
σ T σ T σ T
where
S: underlying common stock price;

K: option’s strike price;
T: option’s time to maturity;
σ: instantaneous standard deviation of return;
N (.): cumulative normal density function and
n(.): density function of the normal distribution.
As pointed out by Merton (1973) and Smith (1976), this formulation

allows for both negative security and option prices and does not account
for the time value of money. Sprenkle (1961) re-formulated the option
pricing problem by assuming that the dynamics of stock prices are log-
normally distributed. By introducing a drift in the random walk, he ruled
out negative security prices and allowed risk aversion. By letting asset prices
to have multiplicative, rather than additive fluctuations, the distribution
of the option’s underlying asset at maturity is log-normal rather than
normal.
8.1.2. Sprenkle formula

Sprenkle (1961) derived the following formula:
c(S, T ) = SeρT N (d1 ) − (1 − Z)KN (d2 )

S 2
S σ2

(8.2)
ln K + ρ + σ2 T ln K + ρ− 2 T
d1 = √ , d2 = √
σ T σ T
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.
8.1.3. Boness formula

Boness (1964) presented an option pricing formula accounting for the
time value of money through the discounting of the terminal stock price
using the expected rate of return to the stock. The option pricing formula
proposed is:
c(S, T ) = SN (d1 ) − e−ρT KN (d2 )

S 2
S σ2

(8.3)
ln K + ρ + σ2 T ln K + ρ− 2 T
d1 = √ , d2 = √
σ T σ T
where ρ is the expected rate of return to the stock.
8.1.4. Samuelson formula

Samuelson (1965) allowed the option to have a different level of risk from
the stock. Defining ρ as the average rate of growth of the share price and w
as the average rate of growth of the call’s value, he proposed the following
formula:
c(S, T ) = Se(ρ−w)T N (d1 ) − e−wT KN (d2 )
S 2
S σ2

(8.4)
ln K + ρ + σ2 T ln K + ρ− 2 T
d1 = √ , d2 = √ .
σ T σ T
Note that all the proposed formulas show one or more arbitrary parameters,
depending on the investors’ preferences towards risk or the rate of return
on the stock. 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. Black and Scholes (1973) used a formula that was
derived in Thorp and Kassouf (1967), who presented an empirical formula
for warrants. This formula determines the ratio of shares of stocks needed
to construct a hedged position. This position is constructed by buying an
asset and selling another. However, Thorp and Kassouf did not realize that
in equilibrium, the expected return on a hedged position must be the return
on a risk-less asset. This concept is due to Black and Scholes as we will see
in the derivation of their model.
8.2. How the Black–Scholes Option Formula is Obtained

8.2.1. The short story
The main idea behind the Black–Scholes formula is the existence of a
relationship among a call price, its underlying asset, the volatility, the
strike price, the maturity date, and the interest rate. This relationship
indicates how much the option value will change when the underlying
asset changes by a small amount within a short time. Assume that the
option price increases by $0.5 when the underlying asset increases by $1
and that the option price decreases by $0.5 when the underlying asset
decreases by $1. In this context, it is possible to create a hedged position
by selling two options and buying one round lot of stock. For small
changes in the underlying asset price, the losses on one side will be nearly
offset by gains on the other side. Hence, at first, a hedged position is
created by selling two options and holding the underlying asset. When the
underlying asset price increases, the position shows a loss on the option
and a profit on the underlying asset. When the underlying asset price
decreases, the position shows a gain on the option and a loss on the
underlying asset. A neutral hedge can be maintained by modifying the
position in the option, in the stock, or in both. The principle leading to
the option formula is that a hedged position should yield an amount equal
to the short-term interest rate on close-to-risk-less securities. A “reverse
hedge” can also be implemented to generate the Black and Scholes (1973)
formula by selling short the stock and buying the option (in the right
ratio).
The formula can also be derived by assuming that a neutral spread
must earn the interest rate. In fact, selling an option and buying another
option on the same underlying asset in the right ratio is a neutral spread.
The formula can be obtained even without the assumption of hedging or
spreading. In this case, a comparison is done between a long stock position
with a long option position that has the same action as the stock. In the
above example, the investor compares a long position of one round lot of
stock and the purchase of twooption contracts. These two positions show
the same movements for very small changes in the stock prices, so their
returns must differ only by an amount equal to the interest rate times the
difference in the total values of both positions. At equilibrium, investors
are indifferent between the two positions. This leads to the same valuation
formula.
8.2.2. The differential equation

The notion of equilibrium in the market for risky assets implies that riskier
securities must have higher expected returns, or investors will not hold
them — except that investors do not count the part of the risk they are
able to diversify away.
Black (1976) applied the CAPM to the valuation of warrants in 1969.
The method used at that time was based on the discounted expected
value of the warrant at expiration. This method has two problems. First,
the warrant price depends on the stock’s expected return. Second, an
appropriate discount rate must be chosen.
One key to solve this problem is to write the warrant formula as a
function of the stock price and other factors. This approach was adopted
by Black (1998) and Samuelson and Merton (1969). Black’s unpublished
formula shows that the expected return on a warrant depends on the risk
of the warrant in the same way that a common stock’s expected return
depends on its risk.
Black used the CAPM to write down how the discount rate for a
warrant varies with time and the stock price. This gave a differential
equation for the warrant formula. Black did not recognize the equation
as a version of the “heat equation”. Therefore, he did not write down its
solution. Besides, he did not note that the warrant value did not depend
on the stock’s expected return or any other asset’s expected return.
8.2.3. The derivation of the formula

In 1969, Scholes and Black started working on the option problem at
Massachusetts Institute of Technology (MIT). They concentrated on the
fact that the option price depends on the volatility rather than the expected
return. They assumed that the stock’s expected return was equal to the
constant interest rate. This assumption is equivalent to the fact that the
stock’s beta is zero, so all of its risk can be diversified away. They assumed
also that the stock’s volatility was constant and the terminal stock price
“fits” into a log-normal distribution. Sprenkle used the same assumptions,
except that he allowed the stock to have any constant expected return. The
problem was to find the present value of the option rather than its expected
terminal value.
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,
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.
8.2.4. Publication of the formula

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) worked separately on
the application of the formula to the valuation of risky corporate bonds
and common stock. Black and Scholes gave an early version of their paper
at a conference on capital market theory during summer 1970. Black and
Scholes sent their article to the Journal of Political Economy, the Review
of Economics and Statistics and it was rejected even without a review.
The paper was sent again in 1971 to the Journal of Political Economy after
accounting for the comments and suggestions by Merton, Miller, and Fama.
The final draft of the paper dated May 1972 appeared in the May/June 1973
issue of the Journal of Political Economy. However, a paper on the results of
some empirical tests appeared in the May 1972 issue of Journal of Finance.
8.2.5. Testing the formula

The formula was first tested on warrants. Black and Scholes estimated
the volatility of the stock of each of a group of companies with warrants.
Black, Scholes, and Merton found the best to buy be National General
new warrants. They bought a bunch of these warrants. The formula and
the volatility estimates were always based on the information at hand.
Using the premiums received by a broker’s option-writing customers in the
over-the-counter (OTC) option market for a period of several years, some
trading rules were tested by Black and Scholes. The formula was used to find
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.
8.3. Financial Theory and the Black–Scholes–Merton

Theory
8.3.1. The Black–Scholes–Merton theory
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 regardless of investor’s risk
preferences. The main intuition behind the risk-free hedge is simple.
Consider an at-the-money European call giving the right to its holder to
buy one unit of the underlying asset in one month at a strike price of $100.
Assume that the final asset price is either 105 or 95. An investor selling a
call on the unit of the asset will receive either 5 or 0. In this context, selling
two calls against each unit of the asset will create a terminal portfolio value
of 95. The certain terminal value of this portfolio must be equal today to
the discounted value of 95 at the risk-less interest rate. If this rate is 1%, the
present value is (95/1.01). The current option value is (100 − (95/1.01))/2.
If the observed market price is above (or below) the theoretical price, it is
possible to implement an arbitrage strategy by selling the call and buying
(selling) a portfolio comprising a long position in a half unit of the asset

and a short position in risk-free bonds. The Black–Scholes–Merton model
is the continuous-time version of this example. The theory assumes that
the underlying asset follows a geometric Brownian motion and is based on
the construction of a risk-free hedge between the option and its underlying
asset. This implies that the call payout can be duplicated by a portfolio
consisting of the asset and risk-free bonds. In this theory, the option value
is the same for a risk-neutral investor and a risk-averse investor. Hence,
options can be valued in a risk-neutral world, i.e., expected returns on all
assets are equal to the risk-free rate of interest.
8.3.2. Analytical formulas

The main and general result in the Black–Scholes–Merton theory is that if
a risk-free hedge can be implemented using the option and its underlying
asset, then risk-neutral valuation may be applied. This means that the
theory applies in the presence of simple and complex options payouts. The
Black–Scholes and Merton formula for a European call follows directly from
the work of Sprenkle and Samuelson. In a risk-neutral world, all assets show
an expected rate of return equal to the risk-free interest rate. This does not
mean that all assets have the same expected rate of price appreciation. If
the asset income which may be a dividend, a coupon, etc. is modeled as a
constant continuous proportion of the asset price, then the expected rate
of price appreciation must be equal to the interest rate less than the cash
distribution rate. The theory covers a wide range of underlying assets.
For non-dividend-paying stock options: When there are no dividends

on the underlying stock, the expected price appreciation rate of the stock
is the risk-free interest rate.
For constant-dividend-yield stock options: When stocks pay divi-

dends at a constant and a continuous dividend yield, Merton (1973) derives
the option valuation formula. In his formula, the cost of carrying the
underlying asset corresponds to the difference between the risk-free rate
and the stock’s dividend-yield rate.
Futures options: In a risk-neutral world with constant interest rates, the

expected rate of price appreciation on a futures contract is zero. This is
because the futures contract does not involve any cash outlay.
Foreign currency options: Garman and Kohlhagen (1983) value options

on foreign currency. The expected rate of price appreciation of a foreign
currency equals the domestic rate of interest less than the foreign rate of
interest.
8.4. The Black–Scholes Model

Under the following assumptions, the value of the option will depend only
on the price of the underlying asset S, time t, and on other variables that are
assumed constants. These assumptions or “ideal conditions” as expressed
by Black–Scholes are the following:
• The option is European;

• The short-term interest rate is known;
• The underlying asset follows a random walk with a variance rate pro-
portional to the stock price. It pays no dividends or other distributions;
• There is no transaction costs and short selling is allowed, i.e., an investor
can sell a security that he/she does not own and
• Trading takes place continuously and the standard form of the capital
market model holds at each instant. The main attractions of the Black–
Scholes model are that their formula is a function of “observable”
variables and that the model can be extended to the pricing of any
type of option.
8.4.1. The Black–Scholes model and CAPM

The CAPM of Sharpe (1964) can be stated as follows:
RS − r = βS [Rm − r] (8.5)
where,
RS : equilibrium-expected return on security S;

Rm : equilibrium-expected return on the market portfolio;
r: 1 + the risk-less rate of interest and
βS = cov( R̃S /R̃m )
var(R̃ )
: the beta of security S, that is the covariance of the return
m
on this security with the return on the market portfolio,
divided by the variance of market return.
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
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,
βc : the option’s beta;

βS : the stock’s beta;
C: the option value;
CS : the first derivative of the option with respect to its underlying asset.
It is also the hedge ratio or the option’s delta in a covered position.
According to the CAPM, the expected return on a security should be:
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
Multiplying (8.8) and (8.9) by S and C yields:
E(∆S) = [rS + SβS (R̄m − r)]∆t (8.10)

E(∆C) = [rC + CβC (R̄m − r)]∆t. (8.11)
When substituting for the option’s elasticity from Eq. (8.7), Eq. (8.11)
becomes after transformation:
E(∆C) = [rC + SCS βS (R̄m − r)]∆t (8.12)
Assuming a hedged position is constructed and “continuously” re-balanced,

and since ∆C is a continuous and differentiable function of two variables,
it is possible to use Taylor’s series expansion to expand ∆C.
1
∆C = CSS (∆S)2 + CS ∆S + Ct ∆t (8.13)
2
This is just an extension of simple results to get Ito’s lemma. Taking
expectations of both sides of Eq. (8.13) and replacing ∆S, we obtain:
1 2 2
E(∆C) = σ S CSS ∆t + CS E(∆S) + Ct ∆t (8.14)
2
Replacing the expected value of ∆S from Eq. (8.8) gives,
1 2 2
E(∆C) = σ S CSS ∆t + CS [rS + SβS (R̄m − r)]∆t + Ct ∆t (8.15)
2
Combining Eqs. (8.11) and (8.15) and re-arranging yields:
1 2 2
σ S CSS + rSCS − rC + Ct = 0. (8.16)
2
This partial differential equation corresponds to the Black–Scholes valua-
tion equation. Let T be the maturity date of the call and E be its strike
price. Equation (8.16) subject to the boundary condition at maturity:
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:
C(S, T ) = SN (d1 ) − Ke−rT N (d2 ) (8.17)
with

S 1 √ √
d1 = ln + r + σ2 T σ T, d2 = d1 − σ T
K 2
and where N (.) is the cumulative normal density function.

It may be shown that Eq. (8.16) applies to both European and

American options.
8.4.2. An alternative derivation of the Black–Scholes model

Assuming that the option price is a function of the single source of
uncertainty, namely stock price and time to maturity, c(S, t) and that over
“short”-time intervals, ∆t, a hedged portfolio consisting of the option, the
underlying asset, and a risk-less security can be formed, where portfolio
weights are chosen to eliminate “market risk” Black–Scholes expressed the
expected return on the option in terms of the option price function and
its partial derivatives. In fact, following Black–Scholes, it is possible to
1
create a hedged position consisting of a sale of ∂c(S,t) options against one
[ ∂S ]
share of stock long. If the stock

price changes by a small amount ∆S, the
option changes by an amount ∂c(S,t)
∂S ∆S. Hence, the change in value in the
1
long position (the stock) is approximately offset by the change in
[ ∂c(S,t)
∂S ]
options. This hedge can be maintained continuously so that the return
on the hedged position becomes completely independent of the change in
the underlying asset value, i.e., the return on the hedged position becomes
certain.
The value of equity in a hedged position, containing a stock purchase
1
and a sale of ∂c(S,t) options is:
[ ∂S ]
C(S, t)
S−
(8.18)
∂c(S,t)
∂S
Over a short interval ∆t, the change in this position is:
∆c(S, t)
∆S −
(8.19)
∂c(S,t)
∂S
where, ∆c(S, t) is given by c(S + ∆S, t + ∆t) − c(S, t).

Using stochastic calculus for ∆c(S, t) gives,
∂c(S, t) 1 ∂ 2 c(S, t) 2 2 ∂c(S, t)

∆c(S, t) = ∆S + σ S ∆t + ∆t. (8.20)
∂S 2 ∂S 2 ∂t
The change in the value of equity in the hedged position is found by

substituting ∆c(S, t) from Eq. (8.20) in Eq. (8.19).
2

1 2 2 ∂ c(S,t) ∂c(S,t)
2
σ S ∂S 2 + ∂t ∆t
− ∂c(S,t)
. (8.21)
∂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
Using this substitution, the differential equation becomes:

∂y ∂ 2y
= .
∂t ∂S 2
This differential equation is the heat transfer equation in physics. The
boundary condition is re-written as y(u, 0) = 0, if u < 0 otherwise,
„ 1 uσ2 «
2
r− 1 σ2
y(u, 0) = K e 2 −1 .
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
c(S, T ) = SN (d1 ) − Ke−rT N (d2 )

“ ”
ln( K
S 2
)+ r+ σ2 T √
d1 = √
σ T
, d2 = d1 − σ T where N (.) is the cumulative normal
density function given by:
d “ 2
”
1 − x2
N (d) = √ e dx.
2Π −∞
It is important to note that the option value is independent of its underlying

asset-expected return. This may sound rather strange. One intuitive way
to account for this is to say the expected return on the stock is already
embedded into the stock price itself. It is also worth noting that the
option price rises when the asset price, the time to maturity,

the
interest
∂c(S,t)
rate, and the variance increase. The partial derivative ∂S , which is
equal to N (d1 ) gives the ratio of the underlying asset to options in the
hedged
position.
It also refers to what is known as the option’s delta. Since
S ∂c(S,t)
c(S,t) ∂S is always greater than one, the option is more volatile than
its underlying asset. The value of the put option can be obtained from that
of the call option using the put-call parity relationship.
8.4.3. The put-call parity relationship

The put-call parity relationship can be derived as follows. Consider a
portfolio A which comprises a call option with a maturity date t∗ and a
discount bond that pays K dollars at the option’s maturity date. Consider
also a portfolio B, with a put option and one share.
The value of portfolio A at maturity is max[0, St∗ − K] + K =
max[K, St∗ ]. The value of portfolio B at maturity is max[0, K −St∗ ]+St∗ =
max[K, St∗ ]. Since both these portfolios have the same value at maturity,
they must have the same initial value at time t, otherwise arbitrage will
be profitable. Therefore, the following put-call relationship must hold

∗
ct − pt = St − Ke−r(t −t) with t∗ − t = T .
If this relationship does not hold, then arbitrage would be profitable.
∗
In fact, suppose for example, that ct − pt > St − Ke−r(t −t) .
At time t, the investor can construct a portfolio by buying the put
and the underlying asset and selling the call. This strategy yields a result
equal to ct − pt − St . If this amount is positive, it can be invested at the
risk-less rate until the maturity date t∗ , otherwise it can be borrowed at
the same rate for the same period. At the option maturity date, the options
will be in-the-money or out-of-the money according to the position of the
underlying asset St∗ with respect to the strike price K. If St∗ > K, the
call is worth its intrinsic value. Since the investor sold the call, he/she
is assigned on this call. He/she will receive the strike price, delivers the
stock, and closes his/her position in the cash account. The put is worthless.
∗
Hence, the position is worth K + er(t −t) [ct − pt − St ] > 0. If ST < K, the
put is worth its intrinsic value. Since the investor is long the put, he/she
exercises his/her option. He/she will receive upon exercise the strike price,
delivers the stock, and closes his/her position in the cash account. The call
∗
is worthless. Hence, the position is worth K + er(t −t) [ct − pt − St ] > 0.
In both cases, the investor makes a profit without an initial cash outlay.
This is a risk-less arbitrage, which must not exist in efficient markets.
Therefore, the above put-call parity relationship must hold good. Using
this relationship, the European put option value is given by:
p(S, T ) = −SN (−d1 ) + Ke−rT N (−d2 ) (8.25)

S 2

ln K + r + σ2 T √
d1 = √ , d2 = d1 − σ T ,
σ T
where N (.) is the cumulative normal density function. We illustrate by the

following examples, the application of the Black and Scholes (1973) model
for the determination of call and put prices.
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.
First, we compute the discounted value of the strike price,
Ke−rT = 15e−0.1(0.25) = 14.6296.
Second, the values of d1 and d2 are calculated as:

1 18 1 2
d1 = √ ln + 0.1 + 0.15 0.25
0.15 0.25 15 2
0.21013
= = 2.8017
0.075
√
d2 = 2.8017 − 0.5 0.25 = 2.7267.
Substituting d1 and d2 in the call price formula gives:
C = 18N (2.8017) − 15e−0.1(0.25) N (2.7267).
Using the approximation of the cumulative normal distribution in the points

2.8017 and 2.7267, the call price is 3.3659 or:
C = 18(0.997) − 14.6296(0.996) = 3.3659
The following (Tables 8.1–8.4) provide simulation results for a European

call and put prices using the Black–Scholes model. The tables also provide
the Greek letters.
Table 8.1. Simulations of Black and Scholes put prices,

S = 100, K = 100, t = 22/12/2003, T = 22/12/2004, r = 2%,
and σ = 20%.
80 19.44832 −0.81955 0.01676 0.21155 −0.00575

85 15.56924 −0.72938 0.01967 0.28240 −0.00769
90 12.17306 −0.62762 0.02105 0.34124 −0.00931
95 9.29821 −0.52216 0.02086 0.37895 −0.01035
100 6.94392 −0.42055 0.01933 0.39156 −0.01069
105 5.07582 −0.32847 0.01692 0.38031 −0.01038
110 3.63657 −0.24939 0.01412 0.35019 −0.00955
115 2.55742 −0.18449 0.01130 0.30789 −0.00838
120 1.76806 −0.13331 0.00871 0.26002 −0.00707
Table 8.2. Simulations of Black and Scholes call prices,

S = 100, K = 100, t = 22/12/2003, T = 22/12/2004, r = 2%,
and σ = 20%.
80 1.43382 0.18045 0.01676 0.21155 −0.00575

85 2.55474 0.27062 0.01967 0.28240 −0.00769
90 4.15856 0.37238 0.02105 0.34124 −0.00931
95 6.28371 0.47784 0.02086 0.37895 −0.01035
100 8.92943 0.57945 0.01933 0.39156 −0.01069
105 12.06132 0.67153 0.01692 0.38031 −0.01038
110 15.62208 0.75061 0.01412 0.35019 −0.00955
115 19.54292 0.81551 0.01130 0.30789 −0.00838
120 23.75356 0.86669 0.00871 0.26002 −0.00707
Table 8.3. Simulations of Black and Scholes call prices,

S = 100, K = 100, t = 22/12/2003, T = 22/06/2004, r = 2%,
and σ = 20%.
80 0.36459 0.07582 0.01332 0.08172 −0.00442

85 0.93156 0.15729 0.02070 0.14559 −0.00791
90 1.99540 0.27358 0.02656 0.21252 −0.01158
95 3.70489 0.41284 0.02900 0.26201 −0.01429
100 6.12966 0.55640 0.02763 0.27966 −0.01526
105 9.24535 0.68669 0.02332 0.26377 −0.01439
110 12.95409 0.79246 0.01781 0.22359 −0.01218
115 17.12243 0.87053 0.01245 0.17279 −0.00939
120 21.61673 0.92356 0.00807 0.12322 −0.00668
Table 8.4. Simulations of Black and Scholes put prices,

S = 100, K = 100, t = 22/12/2003, T = 22/06/2004, r = 2%,
and σ = 20%.
80 19.36686 −0.92418 0.01332 0.08172 −0.00442

85 14.93383 −0.84271 0.02070 0.14559 −0.00791
90 10.99767 −0.72642 0.02656 0.21252 −0.01158
95 7.70716 −0.58716 0.02900 0.26201 −0.01429
100 5.13193 −0.44360 0.02763 0.27966 −0.01526
105 3.24762 −0.31331 0.02332 0.26377 −0.01439
110 1.95636 −0.20754 0.01781 0.22359 −0.01218
115 1.12471 −0.12947 0.01245 0.17279 −0.00939
120 0.61900 −0.07644 0.00807 0.12322 −0.00668
8.5. The Black Model for Commodity Contracts

Using some assumptions similar to those used in deriving the original
B–S option formula, Black (1976) presented a model for the pricing of
commodity options and forward contracts.
8.5.1. The model for forward, futures,

and option contracts
In this model, the spot price S(t) of an asset or a commodity is the price
at which an investor can buy or sell it for an immediate delivery at current
time, t. This price may rise steadily, fall, and fluctuate randomly.
The futures price F (t, t∗ ) of a commodity can be defined as the price
at which an investor agrees to buy or sell at a given time in the future, t∗ ,
without putting up any money immediately. When t = t∗ , the futures price
is equal to the spot price.
A forward contract is a contract to buy or sell at a price that stays
fixed until the maturity date, whereas the futures contract is settled every
day and re-written at the new futures price.
Following Black (1976), let v be the value of the forward contract,
u be the value of the futures contract, and c be the value of an option
contract. Each of these contracts is a function of the futures price F (t, t∗ )
as well as other variables. So, we can write at an instant t, the values of
these contracts, respectively as V (F, t), u(F, t), and c(F, t). The value of
the forward contract also depends on the price of the underlying asset, K
at time t∗ and can be written V (F, t, K, t∗ ). It is important to distinguish
between the price and the value of the contract. The futures price is the price
at which a forward contract presents a zero current value. It is written as:
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
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)
At maturity, the value of a commodity option is given by the maximum

of zero and the difference between the spot price and the contract price.
Since at this date, the futures price equals the spot price, it follows that if
F ≥ K, then c(F, t∗ ) = F − K otherwise,
c(F, t∗ ) = 0. (8.28)
In order to value commodity contracts and commodity options, Black

(1976) assumed that:
• The futures price changes are distributed log-normally with a constant

variance rate σ 2 ;
• All the parameters of the CAPM are constant through time and
• There are no transaction costs and no taxes.
Under these assumptions, 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.

Let ∂c(F,t)
∂F be the weight affected to the short position in the futures
contract, which is the derivative of c(F, t) with respect to F . The change
in the hedged position may be written as:

∂c(F, t)
∆c(F, t) − ∆F. (8.29)
∂F
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
Denoting T = t∗ − t and using Eqs. (8.28) and (8.30), the value of a

commodity option is:
c(F, T ) = e−rT [F N (d1 ) − KN (d2 )]

F σ2 (8.31)
ln K + T √
d1 = √ 2 , d2 = d1 − σ T .
σ T
It is convenient to note that the commodity option’s value is the same as
the value of an option on a security paying a continuous dividend. The rate
of distribution is equal to the stock price times the interest rate. If F e−rT
is substituted in the original formula derived by Black and Scholes, the
result is exactly the above formula. In the same context, the formula for
the European put is:
p(F, T ) = e−rT [−F N (d1 ) + KN (−d2 )]

F σ2 (8.32)
ln K + T √
d1 = √ 2 , d2 = d1 − σ T
σ T
where N (.) is the cumulative normal density function given by:
d “ 2”
1 −x
N (d) = √ e 2 dx.
2Π −∞
The value of the put option can be obtained directly from the put-call
parity.
8.5.2. The put-call relationship

The put-call parity relationship for futures options is:
p − c = e−rT (K − F ). (8.33)
This relationship can be explained as follows. Consider a portfolio where

the investor is long a future contract, long a put on the future contract,
and short a call with the same time to maturity and strike price. Note that
the combination of the call and the put is equivalent to a short synthetic
future. At expiration, the payoff is given by the difference between the
options strike prices and the current futures price. Hence, the current value
of this portfolio must be equal to the present value of this difference. Since
these options are European, they have the same cash flows as options on
the spot asset. This is because at the maturity date, the futures price is
equal to the spot price. We now give some examples for the calculation of
option prices using Black’s formula.
8.6. Application of the CAPM Model to Forward

and Futures Contracts
The major difficulty arising from the application of the CAPM to the
risk-return relationship on a capital asset comes from the definition of the
apropriate rate of return. More specifically, the CAPM cannot be directly
applied to the futures contract because the initial value of the contract
is zero.
8.6.1. An application of the model to forward

and futures contracts
Following the work of Dusak (1973), when applying the classic CAPM, the
percentage change in the futures price is used as a candidate since it can
always be computed. However, the percentage change in the futures price
cannot be interpreted as a rate of return comparable to the R̃S value, since
the holder of the position does not invest current resources in the contract. It
can be rather seen as a risk premium, (R̃S −r) on the underlying spot assets.
In fact, the buyer of the futures contract takes the risk and has no capital
on his/her own invested. Hence, he/she earns no interest on the capital.
The CAPM equilibrium conditions can then be re-stated by saying
that the expected return on any asset S, is written as E(R̃S ) = r(1 − βS ) +
βS (E(R̃m )) where E(R̃S ) can be represented in terms of periods 0 and 1
prices for the asset S as: E(SS1 −S
0
0)
.
The equilibrium risk-return relationship on asset S can then be
expressed as S0 = E(S1 ) − βS (E(R̃m ) − r)S0 , where S0 is the price of
asset S in the start of period and S1 is the asset’s price at the end
of the period. In present-value form, the above equation is equivalent to
S0 = E(S1 )−βS(1+r)
(E(R̃m )−r)S0
.
On the other hand, the current price for asset S under an agreement to
buy the asset at time 0 but with payment deferred to time 1 is S0 (1 + r).
This can be seen as the current futures price for payment of the spot
price in a period. The term E(S1 ) is interpreted as the spot price expected
to prevail at time 1. Multiplying the above equation by (1 + r) gives:
(1 + r)S0 = E(S1 ) − βS (E(R̃m ) − r)S0 .

Setting F0 = (1 + r)S0 and re-arranging the terms gives:

E(S1 ) − F0
= βS (E(R̃m ) − r).
S0
This equation can be interpreted as expressing the risk premium on the
underlying spot asset as the change in the futures price divided by the
period zero spot price. The analysis in Black (1976) can be implemented in
this context. In fact, let us re-write the CAPM in the following form:
E(R̃S ) − r = βS [E(R̃m ) − r].
Writing S0 and S1 as the start and end-of-period prices of an asset i and

using the definition of βi , the CAPM can be written as:

cov S1S−S 0
, R̃
(S1 − S0 ) 0
m
E −r = [E(R̃m ) − r]. (8.34)
S0 var(R̃m )
If we multiply by S0 , we obtain the expected dollar return on asset i as:
cov((S1 − S0 ), R̃m )
E(S1 − S0 ) − rS0 = [E(R̃m ) − r]. (8.35)
var(R̃m )
The price S0 can be set equal to zero since the futures contact’s value at
this time is zero. Ignoring daily limits, we set S1 equal to ∆S. In fact, if we
apply this equation to a futures contract, the change in the futures price
over the period is E(∆S) = cov(∆S, R̃m )
var(R̃m )
[E(R̃m ) − r].
Equation (8.34) can be written for the futures prices as:
E(∆S) = β ∗ [E(R̃m ) − r]. (8.36)
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.
8.6.2. An application to the derivation of the commodity

option valuation
Black (1976) showed that in the absence of interest-rate uncertainty, a
European commodity option on a futures (or a forward) contract can be
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
R̄S − r = aβS with a = [R̄m − r].
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:
E(∆C) = rC∆t + aSβS CS ∆t.
Taking the expected value and replacing E(∆S) gives:

1
E(∆C) = rSCS ∆t + aSβS CS ∆t + Ct ∆t + CSS S 2 σ 2 ∆t.
2
Making the equality between this equation and
E(∆C) = rC∆t + aSβS CS ∆t.
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:
C(F, T ) = e−rT [F N (d1 ) − KN (d2 )]
with
F σ2
ln + T √
K 2
d1 = √ , d2 = d1 − σ T .
σ T
The solution for a European futures put is:
P (F, T ) = e−rT [−F N (−d1 ) + KN (−d2 )].
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).
8.6.3. An application to commodity options and commodity

futures options
A commodity call gives the right to its holder to buy a specific commodity
at a specified price within a specified period of time. A commodity put
gives the right to its holder to buy a specific commodity at a specified
price within a pre-determined period of time. A commodity futures option
is an option on the futures contract having a commodity as an underlying
asset.
The commodity may be a precious metal such as either silver or gold.
It may be a financial instrument such as a common stock, a treasury
bond, or a foreign currency. For example, if the commodity option is
written on a foreign currency, the option refers to a currency option. If
the commodity option is written on a stock index, the option is an index
option.
Since all the analytical models for European options presented in this
chapter can be obtained by modifying the parameters in the Merton (1973)
and BAW (1987) model, all the applications presented here are also true
for this model.
For example, when the option’s underlying asset is an index, which is
constructed to pay continuous dividends, this version is a particular case of
the Merton’s and BAW’s model for commodity options.
When the continuous dividend yield is d, the formula for European

commodity options is re-written for index options with b = (r − d). For a
European index call, the formula is:
c(S, T ) = Se−dT N (d1 ) − Ke−rT N (d2 ) (8.38)
with

Se−dT σ2
ln K + 2
T √
d1 = √ , d2 = d1 − σ T .
σ T
For a European index put, the formula becomes:
p(S, T ) = −Se−dT N (−d1 ) + Ke−rT N (−d2 )
with

Se−dT σ2
ln K + 2
T √
d1 = √ , d2 = d1 − σ T .
σ T
8.7. The Holes in the Black–Scholes–Merton Theory and

the Financial Crisis
Black (1998) examined the assumptions of the Black and Scholes (1973)
model and suggested some modifications to improve the model.
In the original formula, it is assumed that:
• the volatility of the underlying asset is known;

• investors can either borrow or lend at a single interest rate;
• the short seller of the underlying asset can use the proceeds of the sale
even if they are used as collateral;
• there are no transaction costs;
• there are no dividends and no taxes and
• there are no takeovers or other events that end the life of the option
early.
8.7.1. Volatility changes

In practice, volatility is not constant. It can change and affect especially
far-out-of-the-money options.
For example, if the formula is computed for an out-of-the-money option

with a volatility of 20% when S = 28 and K = 40, the option price is 0.0088.
Doubling the volatility to 40%, this can give a new option value of 0.46,
which is more than 50 times higher than in the first case.
In this case, it is possible to assign probability estimates to various
volatility figures and to use the probabilities to weigh the resulting option
values. For example, if there is a 0.5 chance that the volatility will be
0.2 and a 0.5 chance that it will be 0.4, then the option value will be
0.23. This procedure increases out-of-the-money option values. It is true
that volatility changes in an unexplainable way, but there is a relationship
between the stock price and volatility. In general, the stock price and
the volatility change in opposite directions, i.e., when the stock price
increases, the volatility decreases and vice versa. Cox and Ross (1976)
used different formulas to explain the deviations between model values and
option prices by accounting for this fact. The Cox and Ross (1976) formula
gives lower values for out-of-the-money options than the Black–Scholes
formula. Merton (1976) accounted for jumps to increase the relevant values
of both out-of-the-money and in-the-money options. His formula decreases
at the same time the values of at-the-money-options.
Jumps affect the underlying asset price and can be viewed as momen-
tary large increases in the volatility of the underlying asset. The formula
handles only jumps and does not account for stock-price related changes in
volatility. The changing character of volatility does not lead to a “close-to-
riskless” hedge. Consider an initial position in which the investor is short
two calls and long one round lot of the underlying asset. If the stock moves
by $1 and the call by $0.5, the position is protected against stock-price
movement but not against changes in volatility. In practice, it is impossible
to diversify away this risk so that investors do not care about it.
8.7.2. Interest rate changes

As volatility changes over time, interest rates also move with a main
difference so that interest rates can be observed. Merton (1973) has shown
that the random character of interest rates can be accounted for by
substituting the interest rate on a zero-coupon bond and a maturity equal
to the option expiration for the short-term interest rate in the formula. This
is possible when the volatility is constant. The effect of changing interest
rates on option values do not seem as great as the effects of changes in
volatility.
8.7.3. Borrowing penalties

In general, investors borrow at a higher rate than the rate at which they
can lend. High-borrowing rates may increase option values. This is because
options can provide leverage that can substitute for borrowing.
8.7.4. Short-selling penalties

Short-selling leads the investor to borrow the stock and to put up a cash
collateral with the person who lends the stock. This cash does not produce
an interest. In general, professional option traders find that the penalties to
writing naked options do not affect them. However, there are often penalties
on short selling of the stock.
Since buying a put option is equivalent to selling stock short, penalties
on short selling of stock can increase the prices of put options.
8.7.5. Transaction costs

Transaction costs paid in the form of brokerage charges and clearing charges
can affect the hedging strategy. It is not possible to maintain a neutral hedge
continuously, by changing the ratio of option position to stock position as
the parameters change. Stock prices can also jump without a chance for
trades to take place. This makes the strategy impossible to maintain a
neutral hedge.
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
solution, needs the knowledge of how the amount of future unknown

dividends depends on the factors affecting the stock price.
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
regardless of investors’ risk preferences. The main attractions of the Black–

Scholes model are that their formula is a function of “observable” variables
and that the model can be extended to the pricing of any type of option.
Using some assumptions similar to those used in deriving the original
B–S option formula, Black (1976) presented a model for the pricing of
commodity options and forward contracts. Black (1976) showed that in the
absence of interest-rate uncertainty, a European commodity option on a
futures (or a forward) contract can be 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 instant of
time. This chapter presented in detail the basic concepts and techniques
underlying rational derivative asset pricing in the context of analytical
European models along the lines of Black and Scholes (1973), Black (1976),
and Merton (1973). First, an overview of the analytical models proposed
by the precursors is given. Second, the simple model of Black and Scholes
(1973) is derived in detail for the valuation of options on spot assets and
some of its applications are presented. Third, the Black model, which
is an extension of the Black–Scholes model for the valuation of futures
contracts and commodity options, is analyzed. Also, applications of the
model are proposed. Fourth, the basic limitations of the Black–Scholes–
Merton theory are studied and the models are applied to the valuation
of several financial contracts. The Black–Scholes hedge works in the real,
discrete, and frictionful world when the hedger uses the correct volatility
of the prices at which he/she actually trades and when the asset prices do
not jump too much. The assumptions of the Britten-Jones and Neuberger
(1996) model provided a framework in which a trader can avoid jumps and
in which total variance can be estimated perfectly. The model transforms
the question of pricing and hedging options into how well investors can
predict the total variance of returns of the associated hedging strategy.
The Black–Scholes formula gives a rough approximation to the formula
investors would use, if they knew how to account for the above factors.
Modifications of the Black–Scholes formula can move it to the hypothetical
perfect formula.
Questions
1. What is wrong in Bachelier’s formula?
2. What is wrong in Sprenkle’s formula?
3. What is wrong in Boness’s formula?

4. What is wrong in Samuelson’s formula?
5. What are the main differences between the Black and Scholes model
and the precursors models?
6. How can we obtain the put-call parity relationship for options on spot
assets?
7. How can we obtain the put-call parity relationship for options on futures
contracts?
8. Is the Black model appropriate for the valuation of derivative assets
whose values depend on interest rates? Justify your answer.
9. Is there any difference between a futures price and the value of a futures
contract?
10. What are the holes in the Black–Scholes–Merton theory?
Appendix A. The Cumulative Normal Distribution

Function
The following approximation of the cumulative normal distribution function
N (x) produces values to within 4-decimal place accuracy.
x
1
N (x) = √ exp(−z 2 /2)dz
2π −∞

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
where
1
k= , a1 = 0.319381530, a2 = −0.356563782,
1 + 0.2316419x
a3 = 1.781477937, a4 = −1.821255978, and a5 = 1.330274429.
Appendix B. The Bivariate Normal Density Function

1 1 2 2
F (x, y) = exp − (x − 2ρxy + y ) .
2π 1 − ρ2 2π(1 − ρ2 )
The cumulative bivariate normal density function

The standardized cumulative normal function gives the probability that a
specified random variable is less than a and that another random variable
is less than b when the correlation between the two variables is ρ. It is
given by:
a b 2
1 x − 2ρxy + y 2
M (a, b; ρ) = exp − .
2π 1 − ρ2 −∞ −∞ 2π(1 − ρ2 )
This following approximation produces values of M (a, b; ρ) to within

six decimal places accuracy.
5 5
1 − ρ2
φ(a, b; ρ) = xi xj f (yi , yj ),
π i=1 j=1
where
f (yi , yj ) = exp[a1 (2yi − a1 ) + b1 (2yj − b1 ) + 2ρ(yi − a1 )(yj − b1 )]
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
If the product of a, b, and ρ is nonpositive, we must compute the cumulative

bivariate normal probability by applying the following rules:
1. If a 0, b 0, and ρ 0, then:
M (a, b; ρ) = φ(a, b; ρ)
M (a, b; ρ) = N (a) − φ(a, −b; −ρ)
M (a, b; ρ) = N (b) − φ(−a, b; −ρ)
M (a, b; ρ) = N (a) + N (b) − 1 + φ(−a, −b; ρ).
In cases where the product of a, b, and ρ is positive, compute the cumulative

bivariate normal function as:
M (a, b; ρ) = M (a, 0; ρ1 ) + M (b, 0; ρ2 ) − δ
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.
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.
Black, F (1976). The pricing of commodity contracts. Journal of Financial

Economics, 3, (January/March), 167–179.
Black, F (1998). The holes in Black–Scholes, Risk (September), 6–8.
Boness, AJ (1964). Elements of a theory of stock option value. Journal of Political
Economy, 72 (April), 163–175.
Britten-Jones, M and A Neuberger (1996). Arbitrage pricing with incomplete
markets. Applied Mathematical Finance, 3(4), 347–363 and 11–13.
Chen, A (1970). A model of warrant pricing in a dynamic market. Journal of
Finance, 25, 1041–1060.
Churchill, RV (1963). Fourier Series and Boundary Value Problems. 2nd Ed.
New York: McGraw–Hill.
Dusak, K (1973). Futures trading and investor returns: an investigation of com-
modity market risk premiums. Journal of Political Economy, 81, 1387–1406.
Garman, M and S Kohlhagen (1983). Foreign currency option values. Journal of
International Money and Finance, 2, 231–237.
Merton, R (1973). Theory of Rational Option Pricing. Bell Journal of Economics
Merton, RC (1976). Option pricing when underlying stock returns are discontin-
uous. Journal of Financial Economics, 3, 125–144.
Samuelson, PA (1965). Rational theory of warrant pricing. Industrial Management
Review, 6, 13–31.
Samuelson, P and RC Merton (1969). A complete model of warrant pricing that
maximises utility. Industrial Management Review, 10, 17–46.
Sharpe, WF (1964). Capital asset prices: a theory of market equilibrium under
conditions of risk. Journal of Finance, 19, 425–442.
Smith, CW (1976). Option pricing: A review. Journal of Financial Economics,
3(January/March 1976), 3–52.
Smithson, C (1991). Wonderful life. Risk, 4(9), (October), 50–51.
Sprenkle, C (1961). Warrant prices as indications of expectations. Yale Economic
Essays, 1, 179–232.
Thorp, E and S Kassouf (1967). Beat the Market. New York: Random House.
Chapter 9
SIMPLE EXTENSIONS AND APPLICATIONS OF

THE BLACK–SCHOLES TYPE MODELS IN
VALUATION AND RISK MANAGEMENT
Chapter Outline
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
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.
9.1. Applications of the Black–Scholes Model

Equity options can be used in several ways in portfolio management.
Buying or selling options involves the payment or the receipt of the option
premium at the initial time when the transaction is done.
9.1.1. Valuation and the role of equity options

Since the option payoff is asymmetric, this gives rise to an asymmetric
distribution of returns. Hence, options can be used in portfolio management
to structure the distribution of expected returns. The best-known strategies
in portfolio management involve combinations of options. They include
vertical spreads, calendar spreads, diagonal spreads, ratio spreads, volatility
spreads, and synthetic contracts. The main difference between futures
contracts and option contracts is that the investor pays a premium for
options and nothing to establish a futures position.
Applications of Black–Scholes Type Models 405
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.
9.1.2. Valuation and the role of index options

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.
9.1.2.1. Analysis and valuation

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, i.e., of a weighted average of prices of the stocks that constitutes the
index.
There are several weighting schemes. The most commonly used is the
market capitalization where each equity price is weighted by the market
capitalization of the firm, i.e., the number of shares times the share price.
Two alternative methods are sometimes used: equal weighting and
price weighting. The last two methods assign greater relative weight to
small-company constituents than do capitalization-weighted indices. Index
options are also sold in the OTC market as OTC warrants. In this case, they
refer to long-term options on the spot index. Because they are traded on
OTC markets, they are subject to credit risk. When the option underlying
index is constructed to pay continuous dividends, the index price is adjusted
by the discounted value of the continuous dividend yield. The appropriately
adjusted Black–Scholes (1973) version when the continuous dividend yield
d corresponds to the following formula for a European index call is:
c(S, T ) = Se−dT N (d1 ) − Ke−rT N (d2 )

−dT 2

ln SeK + r + σ2 T √
d1 = √ , d2 = d1 − σ T
σ T
where N (.) is the cumulative normal density function given by:
d “ 2”
1 −x
N (d) = √ e 2 dx
2Π −∞
The European index put formula is given by:
p(S, T ) = −Se−dT N (−d1 ) + Ke−rT N (−d2 )

−dT 2

ln SeK + r + σ2 T √
d1 = √ , d2 = d1 − σ T
σ T
9.1.2.2. Arbitrage between index options and futures

It is convenient to note that the same strategies for stock options can also be
used in portfolio management with index options. Also, these options can
be used in asset allocation and portfolio insurance. Since these intruments
are based on the same underlying index, their prices must be interrelated.
If this is not the case, the relative mispricing should instantaneously
disappear given the variety of cross-market strategies. These strategies

include arbitrage between index options and index futures, and between
index futures and the stocks comprising the index. Many researchers have
studied these arbitrages which imply that significant deviations from prices
dictated by the relevant market interrelationships should disappear. It is
often found that there are some violations of the no-arbitrage bounds.
However, when taking into account the transaction costs, the future price
lies often between the no-arbitrage bounds.
Even if all tests and published studies are in favor of market efficiency
and integrated markets, it is often reported that relative mispricing does
exist between index options and futures contracts. For more details, see
for example, Evnine and Rudd (1985) and Brennan and Schwartz (1990),
among others. On the other hand, despite the controversy about index
arbitrage and program trading, these financial intruments are benefical to
stock portfolio managers and institutional investors. Before the emergence
of these contracts, market participants cannot hedge and control the market
risk of their portfolios. Even if there is some evidence that trading in index
futures increases cash-market volatility, arbitrage activities via program
trading may cause prices to adjust more rapidly to new information. This
helps to keep the movements of index futures price and the stock index more
synchronous. The deviations of futures prices from their “fair” value result
from various considerations including imperfect substitutability between
spot and futures markets, the speed with which information is incorporated
in prices in the different markets, and market imperfections including
transaction costs and regulatory constraints, among other things.
9.1.3. Valuation of options on zero-coupon bonds

The Black and Scholes (1973) model 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.
When using the Black and Scholes formula, the underlying asset is the bond
price. The bond price can be observed in the market place, as it can be
calculated by discounting its maturity value at the appropriate risk-free
rate. The underlying bond price can have a maturity of five years for
example and the option time to maturity is in five months. The underlying
asset (the bond) has, in principle, a maximum value. It is given by the

sum of the coupon payments plus the maturity value. If the bond price
is greater than this maximum value, this means that interest rates are
negative. Re-call that the bond price is higher, if higher is the interest rate.
Or, in the Black and Scholes model for stock options, the underlying stock
does not have a maximum value. Hence, applying the Black–Scholes model
for an interest rate option can lead to nonsensical option prices. Besides,
the Black–Scholes model assumes that interest rates are constant during
the option’s life. This is clearly an inappropriate assumption for interest
rate options since interest rates change every day and affect the option
price. Finally, the Black–Scholes model assumes a constant variance for the
underlying asset. This assumption is inappropriate for interest rates since
the bond-price volatility declines as the bond approaches the maturity date.
In fact, the bond price tends to reach its face value at the maturity date.
When there is no coupon payment, the Black–Scholes model is applied as
follows for bond calls:
c(B, T ) = BN (d1 ) − Ke−rT N (d2 )

B 2

ln K + r + σ2 T √
d1 = √ , d2 = d1 − σ T
σ T
The Black–Scholes model is applied as follows for bond puts:
p(B, T ) = −BN (−d1 ) + Ke−rT N (−d2 )

B 2

ln K + r + σ2 T √
d1 = √ , d2 = d1 − σ T
σ T
9.1.4. Valuation and the role of short-term options

on long-term bonds
Short-term options on long-term bonds are often traded on OTC markets.
These options may be of the European or the American type. There is a
traded option on the Chicago Board Options Exchange (CBOE), which is
based upon the yield to maturity on a portfolio of bonds. The yield to
maturity is driven by the changes in the term structure of interest rates.
The Black–Scholes model is sometimes used to price short-term
European options on zero-coupon bonds. In this context, the call’s value is
given by:
c(B, T ) = BN (d1 ) − Ke−rT N (d2 )
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:
p(B, T ) = −BN (−d1 ) + Ke−rT N (−d2 )

B 2

ln K + r + σ2 T √
d1 = √ , d2 = d1 − σ T
σ T
However, other European models, which are extensions or generalizations of
the Black–Scholes model like Merton’s (1973) model, are more appropriate
for the pricing of these options.
9.1.5. Valuation of interest rate options

Interest rate options are often used in the management of interest-rate
risk in the same way as equity options. A direct implication is that option
strategies for equity options apply directly to interest rate options.
The most common and specific strategies based on short-term interest
rate options are caps and floors. These strategies place either a cap on
the future level of interest rates on a floating instrument or a floor on the
interest rate receivable on deposits.
A cap: It is an option strategy which protects from a rise in interest rates
and allows a profit when interest rates are falling.
A floor: It is an option strategy which protects from a decrease in interest

rates at the time when the deposit rate is re-set.
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.
9.1.6. Valuation and the role of bond options: the case

of coupon-paying bonds
The price of any financial asset is given by the present value of its expected
cash flows. The first step in determining the bond’s price is to determine its
cash flows, i.e., the periodic coupon interest payments until the maturity
date and the par value at maturity. Since the bond price is given by
the present value of the cash flows, its price is given by adding all the
discounted future payments at the appropriate interest rates. Some bonds
do not make any periodic coupon payments and the interest due to the
difference between the maturity value and the purchase price given by the
bond holder. This class of bonds is referred to as zero-coupon bonds. It is
convenient to note that there are several types of bonds: bonds with call
provisions, putable bonds, convertible bonds, bonds with warrants attached,
exchangeable bonds, etc.
A bond with a call provision gives the right to the issuer to call the
issue before the specified redemption date. The call price is different from
par and is specified at the bond issue.
A bond with a put provision gives its holder the right to put the
bond back to the issuer at a fixed price. It is a putable bond.
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. This number is often called the rate of conversion which is specified
when the bond is issued.
A bond with an attached warrant is simply a package comprising
the bond and a warrant. It allows the holder to purchase the equity of the
issuing firm. Most of these bonds are Eurobonds issued in international
capital markets.
An exchangeable bond is similar to a convertible bond, with the
exception that it gives its holder the right to exchange the bonds for the
equity of other company. The call or the put provision in these bonds
can be valued using the Black–Scholes model. However, the model is not
appropriate if there are many call or put dates and if the embedded 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:
• Bond’s price = 96 Euro;

• One year interest rate = 10%;
• Time to maturity = 10 years;
• Volatility of the bond’s price = 8%;
• Coupon payments = 5 Euros in 3 months and 9 months;
• Three-month interest rate = 8% and
• Nine-month interest rate = 8.5%.
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:
5e−0.25x0.08 + 5e−0.75x0.085 = 4.9 + 4.69 = 9.59.
Applying the Black–Scholes formula gives:

B = 96 − 9.59 = 86.41
86.41 Euro,
1
d1 = 0.080 ln 100 + 0.1 + 0.0032
d2 = d1 − 0.08 × 1 = −0.6158
c = 86.41N (−0.5358) + 100e−0.1 N (−0.6158)
or c = 1.25 Euro.
It is convenient to note the “incoherence” with this model since it assumes

constant interest rates and at the same time a stochastic bond price.
9.1.7. The valuation of a swaption

A swaption: It is 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.
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.
Swaption pricing is based on a model allowing the computation of the value

of the option to exchange one asset for another as given by the formula in
Margrabe (1978):
W (x1 , x2 , t) = x1 N (d1 ) − x2 N (d2 )
(ln(x1 /x2 ) + 0.5σ2 (t∗ − t))

d1 =
(t∗ − t)

d2 = d1 − v (t∗ − t)
σ2 = σ12 + σ22 − 2σ1 σ2 ρ1,2

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.
9.2. Applications of the Black’s Model

9.2.1. Options on equity index futures
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. Hence, a call is exercised into a long position in
the future contract and a put is exercised into a short position in the
same contract. The underlying futures contract does not require a physical
delivery but is rather settled in cash. The amount received corresponds to
the difference between the current and the future level of the underlying
index. In this context, the futures contract is regarded as an agreement to
either pay or receive a cash payment based upon the difference between the
current and the future values of a specified index.
Options on index futures are often treated as options on an asset paying
a continuous stream of dividends, regardless of whether the underlying spot
index pays a continuous or a discrete dividend.
Since the assumptions used by Black are similar to those in Black–
Scholes, some of them are also questionable, such as the constant volatility
and the certainty of interest rates. In fact, the nonstationarity of volatility

causes some problems in the pricing of options on index futures. As we
will see later, some extensions on the Black’s model are more appropriate
for the valuation of these options. However, when these options are of
the European type, Black’s model is often used in the pricing of these
options.
The various strategies applied with options on individual assets can be
used as well for options on index futures.
Index futures and options on index futures are often used in asset
allocation and portfolio insurance. Asset allocation refers to the structuring
of a multiasset portfolio with respect to the type and the weighing scheme
of asset classes. Strategic asset allocation is the construction of a portfolio
such that long-run objectives are attained when different classes of assets
are transacted at their long-run equilibrium values.
Tactical asset allocation, also known as market timing involves short-
term allocations toward rising markets and away from falling markets.
Portfolio insurance refers to a group of techniques that insure a portfolio
against falling in value below a certain specified level, the floor level.
This level does not eliminate the potential profits from a rise in the asset
value.
9.2.2. Options on currency forwards and options

on currency futures
9.2.2.1. Options on currency forwards
They 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.
9.2.2.2. Options on currency futures

These options have been traded since 1982. These options are standardized
contracts and can be inflexible. They are priced off the underlying futures
contract. When exercised, the call buyer receives a long position in one
futures contract marked-to-market at the current price. In the same way,
when exercised, the put holder receives a short position in one futures
contract.
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.
9.2.3. The Black’s model and valuation of interest rate caps

An interest rate cap: It is defined as an agreement or a contractual
arrangement where the seller known as the grantor is obliged to pay cash
to the buyer whenever the interest rate exceeds or is less than a pre-specified
agreed level at some future time. When the grantor pays cash to the cap
holder, this latter’s net position is equivalent to borrowing at a fixed rate
at this specified level. Hence, an interest rate cap can be seen as an option
where the holder pays a premium upfront. The well-known forms of interest
rate cap agreements are the floor and ceiling agreements. The floor holder
can establish a minimum level for his/her floating-rate deposits over a given
period. If at future dates the interest rate falls below the floor rate, the
seller makes good the holder’s interest income shortfall. However, if rates
are higher than the floor rate, the buyer receives nothing, but has the
possibility to place his/her deposit at a higher market rate. The ceiling
agreement gives the right to the buyer to establish a maximum interest-rate
level for borrowing over a given period. If rates turn to be higher than the
ceiling rate, the buyer receives cash to exactly offset the additional interest-
rate charge due because of higher rates. However, if rates fall in the future,
the holder can borrow at a rate lesser than the ceiling rate. Therefore, the
Black’s model can be used for the pricing of the caps. The model requires
five variables: time to maturity, the price of the underlying futures, the
strike price or cap level, the risk-free rate for the option maturity, and the
volatility.
9.3. The Extension to Foreign Currencies: The Garman

and Kohlhagen Model and its Applications
Garman and Kohlhagen (1983) provided a formula for the valuation of
foreign currency options. These options are traded on the foreign exchange
market, which is fundamentally an inter-bank market where transactions
are conducted over the telecommunications system. The foreign exchange
market called also the FX market operates internationally 24 hours a
day where the major participants are commercial banks around the world
and treasury departments of large companies. As in other markets, the
various participants search for hedging exchange risks, speculation, and
the implementation of arbitrage strategies. Foreign currency options satisfy
some of the needs of these participants and the important volume of
transactions implies the use of an option pricing model. A simple and an
interesting analytic model is provided by Garman and Kohlhagen (1983).
Foreign currency options are priced along the lines of Black and Scholes
(1973) and Merton (1973). Specifically, Garman and Kohlhagen (1983) and
Grabbe (1983) presented models for currency options, which are based on
the assumption that a risk-less hedge portfolio can be formed by investing
in foreign bonds, domestic bonds, and the option.
9.3.1. The currency call formula

Using the same assumptions as in the Black and Scholes (1973) model,
Garman and Kohlhagen (1983) presented the following formula for a
European currency call:
∗
c(S, T ) = Se−r T N (d1 ) + Ke−rT N (d2 )
S 2
ln K + (r − r∗ + σ2 )T √
d1 = √ , d2 = d1 − σ T
σ T
9.3.2. The currency put formula

The formula for a European currency put is:
∗
p(S, T ) = −Se−r T
N (−d1 ) − Ke−rT N (−d2 )
S σ2
ln + (r − r∗ + )T √
K 2
d1 = √ , d2 = d1 − σ T
σ T
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:
C = 180e−(0.1)(0.25)N (d1 ) − 180e−(0.07)(0.25) N (d2 )

√
d1 = [ln(180/180) + (0.06 − 0.10 + 0.5(0.2)2 )0.25]/0.2 0.25 = −0.05
√
d2 = d1 − 0.2 0.25 = −0.15
N (d1 ) = 0.4801, N (d2) = 0.4404
C = 84.284 − 77.896 = 6.3817.
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.
9.3.3. The interest-rate theorem and the pricing of forward

currency options
The interest-rate parity theorem states that the forward rate is equal to the
spot rate compounded by the differential between the foreign and domestic
interest rates. Using continuously compounded interest rates, the forward
exchange rate is:
∗
f = Se(r−r )T
which means simply that the formula for the pricing of a European call on
a spot currency can be re-written as:
c(F, T ) = e−rT [f N (d1 ) − KN (d2 )]

2
f
ln K + σ2 T √
d1 = √ , and d2 = d1 − σ T
σ T
The formula for the European currency put is:
p(F, T ) = e−rT [−f N (−d1 ) + KN (−d2 )]

2
f
ln K + σ2 T √
d1 = √ , and d2 = d1 − σ T
σ T
where all the parameters have the same meaning as before, except for the
spot exchange rate S, which is replaced by the forward exchange rate f .
Note that the interest-rate differential is not explicitly taken into account
in the above formula. This is because all the available information about
spot rates and the interest-rate differential is integrated in the forward
exchange rate via the interest-rate parity theorem. The following tables
provide simulation results for option prices using the Garman-Kohlhagen
model (Tables 9.1–9.4). The tables also give the Greek letters. The reader
can make comments about the values of the Greek letters.
These tables provide the risk matrix for managing and monitoring
options positions. The head of trading and the asset risk officer must
continuously control these risk parameters.
Table 9.1. Simulations of Garman-Kohlhagen call prices. S = 1,

K = 1, t = 07/02/2003, T = 07/02/2004, r = 3%, r ∗ = 4% and
σ = 20%.
0.96 0.05384 0.43109 0.52931 0.00364 0.00008

0.97 0.05815 0.45081 0.50961 0.00370 0.00008
0.98 0.06266 0.47042 0.49003 0.00376 0.00009
0.99 0.06737 0.48984 0.47063 0.00380 0.00009
1 0.07227 0.50904 0.45145 0.00383 0.00009
1.01 0.07736 0.52798 0.43253 0.00386 0.00008
1.02 0.08264 0.54661 0.41391 0.00387 0.00008
1.03 0.08810 0.56492 0.39562 0.00388 0.00008
1.04 0.09375 0.58286 0.37769 0.00387 0.00008
Table 9.2. Simulations of Garman-Kohlhagen call prices. S = 1.1,

K = 1, t = 07/02/2003, T = 07/02/2004, r = 3%, r ∗ = 4% and
σ = 20%.
1.06 0.10315 0.61071 0.34986 0.00385 0.00008

1.07 0.10988 0.62921 0.33137 0.00382 0.00008
1.08 0.11680 0.64714 0.31345 0.00378 0.00007
1.09 0.12392 0.66449 0.29611 0.00373 0.00007
1.10 0.13123 0.68123 0.27937 0.00368 0.00007
1.11 0.13872 0.69736 0.26325 0.00362 0.00007
1.12 0.14639 0.71287 0.24775 0.00355 0.00006
1.13 0.15423 0.72774 0.23289 0.00347 0.00006
1.14 0.16224 0.74198 0.21865 0.00339 0.00006
Table 9.3. Simulations of Garman-Kohlhagen put prices. S = 1,

K = 1, t = 07/02/2003, T = 07/02/2004, r = 3%, r ∗ = 4% and
σ = 20%.
0.96 0.10195 −0.52960 0.52931 0.00364 0.00011

0.97 0.09666 −0.50987 0.50961 0.00370 0.00011
0.98 0.09156 −0.49027 0.49003 0.00376 0.00011
0.99 0.08666 −0.47084 0.47063 0.00380 0.00011
1 0.08195 −0.45164 0.45145 0.00383 0.00011
1.01 0.07744 −0.43271 0.43253 0.00386 0.00011
1.02 0.07311 −0.41407 0.41391 0.00387 0.00011
1.03 0.06896 −0.39577 0.39562 0.00388 0.00011
1.04 0.06500 −0.37783 0.37769 0.00387 0.00011
Table 9.4. Simulations of Garman-Kohlhagen put prices and the Greek

letters. S = 1.1, K = 1, t = 07/02/2003, T = 07/02/2004, r = 3%,
r ∗ = 4% and σ = 20%.
1.06 0.05904 −0.34997 0.34986 0.00385 0.00011

1.07 0.05519 −0.33148 0.33137 0.00382 0.00011
1.08 0.05155 −0.31354 0.31345 0.00378 0.00011
1.09 0.04810 −0.29620 0.29611 0.00373 0.00011
1.10 0.04484 −0.27945 0.27937 0.00368 0.00011
1.11 0.04177 −0.26332 0.26325 0.00362 0.00010
1.12 0.03887 −0.24781 0.24775 0.00355 0.00010
1.13 0.03615 −0.23294 0.23289 0.00347 0.00010
1.14 0.03358 −0.21870 0.21865 0.00339 0.00010
9.4. The Extension to Other Commodities: The Merton,

Barone-Adesi and Whaley Model, and Its Applications
The model presented in Barone-Adesi and Whaley (1987) is a direct
extension of the models presented by Black and Scholes (1973), Merton
(1973), and Black (1976).
9.4.1. The model

The absence of risk-less arbitrage opportunities imply that the following
relationship exists between the futures contract, F , and the price of
its underlying spot commodity, S: F = SebT , where T is the time to
expiration and b is the cost of carrying the commodity. When the underlying
commodity dynamics are given by:
dS
= αdt + σdW
S
where α is the expected instantaneous relative price change of the commod-
ity and σ is its standard deviation, then the dynamics of the futures price
are given by the following differential equation:
dF
= (α − b)dt + σdW
F
Assuming that a hedged portfolio containing the option and the underlying
commodity can be constructed and adjusted continuously, the partial
differential equation that must be satisfied by the option price, c, is:

1 2 2 ∂ 2 c(S, t) ∂c(S, t) ∂c(S, t)

σ S − rc(S, t) + bS + =0
2 ∂S 2 ∂S ∂t
This equation first appeared indirectly in Merton (1973).
When the cost of carry b is equal to the risk-less interest rate, this
equation reduces to that of the equation in B–S (1973) model.
When the cost of carry is zero, this equation reduces to that of the
equation given in Black (1976).
When the cost of carry is equal to the difference between the domestic
and the foreign interest rate, this equation reduces to that in Garman and
Kohlhagen (1983).
It is convenient to note that the short-term interest rate r, and the cost
of carrying the commodity, b, are assumed to be constant and proportional
rates.
Using the terminal boundary condition: c(S, T ) = max[0, ST − K]
Merton (1973) showed indirectly that the European call price is:
c(S, T ) = Se(b−r)T N (d1 ) − Ke−rT N (d2 )

S σ2
ln + (b + )T √
K 2
d1 = √ , and d2 = d1 − σ T
σ T
Using the boundary condition for the put p(S, T ) = max[0, K − ST ], the
European put price is given by:
p(S, T ) = −Se(b−r)T N (−d1 ) + Ke−rT N (−d2 )

S 2

ln K + b + σ2 T √
d1 = √ , d2 = d1 − σ T
σ T
The call formula provides the composition of the asset-bond portfolio that
mimics exactly the call’s payoff. A long position in a call can be replicated
by buying e(b−r)T N (d1 ) units of the underlying asset and selling N (d2 )
units of risk-free bonds, each unit with strike price Ke−rT . When the asset
price varies, the units invested in the underlying aset and risk-free bonds
will change. Using a continuous re-balancing of the portfolio, the pay* outs
will be identical to those of the call. The same strategy can be used to
duplicate the put’s payoff.
9.4.2. An application to portfolio insurance

Dynamic portfolio insurance strategies are based on a dynamic replication
argument. The nontrading of the long-term index put options in the 1980s
led stock portfolio managers to create their own insurance using a dynamic
re-balancing portfolio-containing stocks and risk-free bonds. The portfolio
weights in a dynamically re-balancing portfolio are determined using the
put option formula:
p(S, T ) = −Se(b−r)T N (−d1 ) + KN (−d2 )

S σ2
ln + (b + )T √
K 2
d1 = √ and d2 = d1 − σ T
σ T
The objective of portfolio insurance is to create an “insured” portfolio whose
pay outs mimic the portfolio Se(b−r)T + p. The strategy is equivalent to:
Se(b−r)T + p = Se(b−r)T − Se(b−r)T N (−d1 ) + Ke−rT N (−d2 )

or Se(b−r)T N (d1 ) + Ke−rT N (−d2 ), in this context, a dynamically insured

portfolio shows an investment of Se(b−r)T N (d1 ) units of stocks and
N (−d2 ) units of risk-free bonds. When the stock price increases, funds
are transferred from bonds to stocks and vice versa.
9.5. The Real World and the Black–Scholes Type Models

The Black and Scholes (1973) type 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.
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.
9.5.2. The hedging strategy

In the Black–Scholes (1973) theory, the more frequently the hedge is
adjusted, the more precise is the hedge. This assumes that the volume of
trading will also increase without limit as the portfolio is re-balanced more
frequently. The model seems to break down when frictions are introduced
into the trading process.
9.5.3. The log-normal assumption

This model assumes that the log returns are normally distributed. However,
for several underlying assets, returns seem to be fat-tailed. Besides, this
model assumes that the dynamics of the underlying asset are almost
continuous, however prices tend to move discretely and jump. Despite its
simplifying assumptions, the model seems to give “good” results because
it represents the limiting case for option price bounds that exist in more
general models.
9.5.4. A world of finite trading

Britten-Jones and Neuberger (1998) developed a model of finite trading
to analyze trading costs, hedging strategies, and the effect of price jumps.
Consider the pricing and the hedging of a short position in a European call
option on a non-dividend paying asset. The model assumes that the trader
can trade when he/she wants and not continuously as in the Black–Scholes
model. Denote the net prices by P0 , P1 , . . . , PN (after taking account of
market impact, spreads, commissions, etc.) used by the trader to re-balance
the hedge. The price P0 corresponds to the initial price when the option is
sold and PN is the price at the option maturity. The trader increases the
hedge when prices increase and reduces the hedge when the prices decrease.
Pi+1 is the total cost of buying (ask price plus commissions) when Pi+1 > Pi
and it represents the net revenue from selling (bid price minus commissions)
when Pi+1 < Pi . The model assumes that jumps in prices are less than some
amount d with: d ≥ log PPi−1 i
with i = 1, . . . , N .
9.5.5. Total variance

2
Pi
The total variance in this model is given by: ν = ΣN
1 log Pi−1 . This total
variance is a function of transaction prices.
9.5.6. Black–Scholes as the limiting case

Britten-Jones and Neuberger (1996) have shown that when there is an upper
bound on total variance and when asset prices do not jump too much, it
is possible to place an upper bound on the call price C(ν, d). This upper
bound depends on the total variance and the maximum jump size. In the
same way, it is possible to place a lower bound on the option price C(ν, d).
If the trader has a view that the jump size will not exceed d and the total
variance will not be higher than ν, then he/she can sell the option at C(ν, d)
confident that at worst he/she will break even. When d tends to zero, the
option price in this model tends to be the Black–Scholes price. If the trader
is confident that the total variance will lie somewhere between ν and ν and
the maximum jump will be less than d, then it is possible to place bounds
on the option price C(ν, d) and C(ν, d). These two bounds allow the trader
to buy the option for less than the lower bound and to sell it for more than
the upper bound. In this limit, when ν is known and d is very small, the
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.
9.5.7. Using the model to optimize hedging

How often a hedge should be re-balanced for a sold call?
This question can be turned into an empirical issue about forecasting
the total variance of returns. Britten-Jones and Neuberger (1998) give an
example to illustrate the answer to this question.
Consider a trader who believes that the volatility of the asset price is
15%. Transaction costs are five basis points.
If the hedge is re-balanced 10 times for a three-month maturity call, this
leads to low transaction costs. Selling an option at a volatility of 15.3% can
lead to profits. However, because of infrequent re-balancing, this situation
can lead to losing money. The trader can be sure to make money at 99% if
the option priced at three standard deviations away from the mean is sold
on a volatility of 23%. If the position is re-balanced about 300 times, the
investor can make money when the option is sold at a volatility of 18.3%.
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
is fundamentally an inter-bank market where transactions are conducted

over the telecommunications system. The foreign exchange market also
called the FX market operates internationally 24 hours a day where the
major participants are commercial banks around the world and treasury
departments of large companies. Currency options were traded on the spot
currency for the first time in 1982 at the Philadelphia Stock Exchange.
Since this date, currency options are traded on many other financial places.
However, the trading on the OTC market seems to be more important.
The strategies discussed for stock options also apply to currency options
and currency futures options. This chapter presented in detail the basic
concepts and techniques underlying rational derivative asset pricing in the
context of the analytical European models along the lines of Black–Scholes
and Merton. First, some applications of the Black–Scholes (1973) model
are provided. Second, applications of the Black (1976) model are presented.
Third, applications of the Garman and Kohlhagen model are presented for
the valuation of currency options. Fourth, the Merton (1973) and Barone-
Adesi and Whaley (1987) model is proposed for the valuation of European
commodity contracts, commodity options, and commodity futures options.
Some applications of the model are given. Note that this model reduces to
the models in Black–Scholes, Black and Garman and Kohlhagen for some
values of its parameters. Since all these models (except Arone-Adesi and
Whaley (1987)) are interested in the valuation of European style options
in a continuous time framework, (without discrete distributions to the
underlying asset), a natural extension of these models must introduce the
possibility of an early exercise and discrete distributions. However, before
making some extensions of the basic analytical models, it is useful to study
in this simple context, the option price sensitivities and the use of these
Greek-risk measures in the monitoring and the management of an option
position. The question of managing an option position is as important as
some issues regarding the option pricing. The Black–Scholes hedge works
in the real, discrete, and frictionful world when the hedger uses the correct
volatility of the prices at which he/she actually trades and when the
asset prices do not jump too much. The assumptions of the Britten-Jones
and Neuberger (1998) model provide a framework in which a trader can
avoid jumps and in which total variance can be estimated perfectly. The
model transforms the question of pricing and hedging options into how well
investors can predict the total variance of returns of the associated hedging
strategy.
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?
Black–Scholes Valuation: Bloomberg
Appendix
References
Economics, 3, 167–179.
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
Grabe, JO (1983). The pricing of call and put options on foreign exchange. Journal
of International Money and Finance, 2, 239–253.

Chapter 10
APPLICATIONS OF OPTION PRICING MODELS

TO THE MONITORING AND THE
MANAGEMENT OF PORTFOLIOS OF
DERIVATIVES IN THE REAL WORLD
Chapter Outline
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
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.
Applications of Option Pricing Models 441
Speed is given by the gamma’s derivative with respect to the under-

lying asset price.
Color is given by the gamma’s derivative with respect to the time
remainning to maturity.
The theta measures the change in the option price as time elapses
since time decays presents a negative impact on option values. Theta is
given by the first partial derivative of the option premium with respect to
time.
The vega or lambda measures the change in the option price for a
change in the underlying asset’s volatility. It is given by the first derivative
of the option premium with respect to the volatility parameter.
The knowledge of the “true” option price is not sufficient for the
monitoring and the management of an option position. Therefore, it
is important to know the option-price sensitivities with respect to the
parameters entering the option formula. We begin our discussion with the
delta. In this chapter, we show how to calculate some of these parameters
within the context of each analytical model presented in Chapter 9. Also, we
develop some examples to show how to use the Greek-letter risk measures
in the monitoring and the management of an option position in response
to the changing market conditions.
10.1. Option-Price Sensitivities: Some Specific Examples

10.1.1. Delta
The delta is given by the option’s first partial derivative with respect to
the underlying asset price. It represents the hedge ratio in the context of
the Black–Scholes model (B–S model).
The call’s delta

The call’s delta is given by ∆c = N (d1 ). The use of this formula requires
the computation of d1 given by:
S

ln K + r + 12 σ 2 T
d1 = √
σ T
Example. Let the underlying asset price 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 option’s delta is given by ∆c = N (d1 ).
To apply this formula, the calculation of d1 is:

1 18 1 2
d1 = √ ln + 0.1 + 0.5 0.25 = 2.8017
0.15 0.25 15 2
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.
The put’s delta

The put’s delta has the same meaning as the call’s delta. It is also given
by the option’s first derivative with respect to the underlying asset price.
When selling (buying) a put option, the hedge needs selling (buying) delta
units of the underlying asset. The put’s delta is given by: ∆p = ∆c − 1 =
0.0997 − 1 = −0.003. The hedge ratio is −0.003. When the underlying
asset price rises from 18 to 19, the put price changes from 0.0045 to
approximately (0.0045 − 0.003), or 0.0015. When it falls from 18 to 17, the
put price rises from 0.0045 to approximately (0.0045 + 0.003), or 0.0075.
Appendix A provides the derivation of the Greek letters in the context
of analytical models. Appendix D provides a detailed derivation of these
parameters.
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.
The call’s gamma

∂∆c 1√
In the B–S model, the call’s gamma is given by Γc = ∂S
= Sσ T
n(d1 )
1 2
with n(d1 ) = √12π e− 2 d1 . Using the same data as in the example, n(d1 ) =
1 2
√ 1 e− 2 (2.8017) = 0.09826and Γc = 18(0.15) 1√
0.09826 = 0.0727.
6.2831 0.25
When the underlying asset price is 18 and its delta is 0.997, a fall in the asset
price by 1 unit yields a change in the delta from 0.997 to approximately
(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.
The put’s gamma

∂∆p 1√
The put’s gamma is given by Γp = ∂S
= Sσ T
n(d1 ) with n(d1 ) =
√1 e − 12 d21 1√
2π
or Γp = 18(0.15) 0.25
0.09826
= 0.0727.
When the asset price changes by one unit, the put price changes by the
delta amount and the delta changes by an amount equal to the 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.
The call’s theta

In the B–S model, the theta is given by:
∂c −Sσn(d1 )
Θc = = √ − rKe−rT N (d2 )
∂T 2 T
Using the same data as in the example above, we obtain:
Θc = −0.2653 − 1.4571 = −1.1918
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.
The put’s theta

In the B–S model, the put’s theta is given by:
∂p Sσn(d1 )
Θp = =− √ + rKe−rT N (d2 )
∂T 2 T
or Θp = −0.2653 + 0.0058 = −0.2594
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.
The call’s vega

∂c
√
In the B–S model, the call’s vega is√ given by vc = ∂σ = S T n(d1 )
or using the above data vc = 18 0.25(0.09826) = 0.88434. Hence,
when the volatility rises by 1 point, the call price increases by 0.88434.
The increase in volatility by 1% changes the option price from 3.3659
to (3.3659 + 1%(0.88434)), or 3.37474. In the same context, the put’s
vega is equal to the call’s vega. The put price changes from 0.0045 to
(0.0045 + 1%(0.88434)), or 0.0133434. When the volatility falls by 1%, the
call’s price changes from 3.3659 to (3.3659 − 1%(0.88434)), or 3.36156 .
In the same way, the put price is modified from 0.0045 to approximately
(0.0045 − 1%(0.88434)), or zero since option prices cannot be negative.
The put’s vega

In the B–S model, the put’s vega is given by:
∂p √
vp = = S T n(d1 )
∂σ
√
or vp = 18 0.25(0.09826) = 0.88434 and it has the same meaning as the
call’s vega.
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).
The put’s rho

In the B–S model, the put’s rho is given by:
∂p
Rhop = = −KT e−rT N (−d2 )
∂r
or Rhop = −15e−0.1(0.25)(0.996)(0.25) = −3.64.
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.
The put’s elasticity

The put’s elasticity is given by:
∆p S
Elasticity = S = [N (d1 ) − 1]
p p
18
or Elasticity = 0.0045 [0.997 − 1] = −12.
The knowledge of the variations in these parameters is fundamental
for the monitoring and the management of an option position. Appendix B
provides the relationship between these hedging parameters.
10.2. Monitoring and Managing an Option Position

in Real Time
Since option prices change in an unpredictible way in response to the
changes in market conditions, traders, market makers, and all option users
must rely upon some model to monitor the evolution of their profit and loss
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.
10.2.1. Simulations and analysis of option price

sensitivities using Barone-Adesi and Whaley model
Re-call that the commodity call price and the commodity futures call price
in the context of Merton’s (1973) and Barone-Adesi and Whaley’s (1987)
model is given by:
c(S, T ) = Se(b−r)T N (d1 ) − Ke−rT N (d2 )

S
ln K + b + 12 σ 2 T √
d1 = √ , d2 = d1 − σ T
σ T
where b stands for the cost of carrying the underlying commodity.
By the put-call parity relationship or by a direct derivation, the put’s
value is given in the same context by:
p(S, T ) = −Se(b−r)T N (−d1 ) + Ke−rT N (−d2 )

S
ln K + b + 12 σ 2 T √
d1 = √ , d2 = d1 − σ T
σ T
where all the parameters have the same meaning as before.
For a non-dividend paying asset, b = r. For a dividend-paying asset,
b = r − d where d stands for the dividend yield.
For a currency option, b = r − r∗ , where r∗ stands for the foreign
risk-less rate.
Tables 10.1 to 10.12 simulate option values and sensitivity parameters
for calls and puts in the context of the above model.
Sensitivity parameters for call options

Table 10.1 gives call prices, delta, gamma, theta, and vega when the
underlying commodity price varies from 75 to 110 in steps of 5. For example,
Table 10.1. Changing the underlying asset prices for Euro-

pean calls. σ = 0.2, r = 0.08, b = 0.10, T = 0.25, and K = 90.
Call Asset Delta Gamma Theta Vega
0.23 75 0.07 0.02 0.05 0.47

0.88 80 0.21 0.03 0.11 1.10
2.42 85 0.42 0.04 0.17 1.71
5.07 90 0.64 0.04 0.17 1.94
8.76 95 0.82 0.03 0.13 1.78
13.17 100 0.93 0.02 0.08 1.46
17.44 105 0.97 0.01 0.04 1.05
22.37 110 0.99 0.01 0.02 0.89

pean calls. σ = 0.2, r = 0.08, b = 0.10, T = 0.25, and K = 100.
Call Asset Delta Gamma Theta Vega
0.01 75 0.01 0 0.01 0.06

0.09 80 0.03 0.01 0.03 0.25
0.40 85 0.10 0.02 0.08 0.71
1.23 90 0.24 0.03 0.14 1.37
2.92 95 0.44 0.04 0.19 1.94
5.64 100 0.64 0.04 0.19 2.16
8.87 105 0.79 0.03 0.15 1.89
13.11 110 0.90 0.02 0.10 1.56
when the volatility σ = 20%, r = 8%, b = 10%, and T = 3 months (0.25

year), the price of an ATM call for K = 90 is 5.07.
The call has a delta of 0.64, a gamma of 0.04, a theta of 0.17, and a
vega of 1.94. Note that an ATM call has more theta and vega than an ITM
and an OTM call.
Using the same data except for the strike price, which is modified from
90 to 100, Table 10.2 shows that an ATM call (K = 100, S = 100) has more
gamma, theta, and vega than ITM, and OTM calls.
Table 10.3 shows call prices and sensitivity parameters for European
calls when the time to maturity varies from 0.05 to 0.75 year.
Note that the call price, the vega, and the theta increase with the time
to maturity. However, the delta falls when the time to maturity is longer.
Table 10.4 gives call prices and sensitivity parameters for ATM calls
when (S = K = 100) and the time to maturity varies from 0.05 to a year.
Note that the call price, its delta, and vega are increasing functions of the
Table 10.3. Changing time to maturity for European calls.

σ = 0.2, r = 0.08, b = 0.10, S = 100, and K = 90.
Call Maturity Delta Gamma Vega Theta
10.47 0.05 0.99 0 0 0.83

11.00 0.10 0.97 0.01 0.02 1.03
12.71 0.25 0.91 0.02 0.08 1.34
15.44 0.50 0.88 0.01 0.14 1.45
17.98 0.75 0.88 0.01 0.19 1.46
Table 10.4. Changing time to maturity for European calls.

σ = 0.2, r = 0.08, b = 0.10, S = 100, and K = 100.
Call Maturity Delta Gamma Vega Theta
2.04 0.05 0.55 0.08 0.09 2.09

3.05 0.10 0.58 0.06 0.12 2.09
5.32 0.25 0.62 0.04 0.20 2.05
8.36 0.50 0.67 0.03 0.26 1.98
11.04 0.75 0.71 0.02 0.31 1.91
13.53 1.00 0.74 0.02 0.34 1.84
Table 10.5. Changing the volatility for European calls. T = 0.25,

r = 0.08, b = 0.10, S = 100, and K = 90.
Call Volatility Delta Gamma Vega Theta
12.29 0.05 1 0 0 0.76

12.29 0.10 1 0.025 0.01 0.78
12.71 0.20 0.92 0.01 0.08 1.34
13.78 0.30 0.85 0.01 0.13 2.18
15.19 0.40 0.78 0.01 0.15 3.02
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).
Table 10.6. Changing the volatility for European calls,

T = 0.25, r = 0.08, b = 0.10, S = 100, and K = 100.
Call Volatility Delta Gamma Vega Theta
2.69 0.05 0.85 0.09 0.12 0.94

3.46 0.10 0.70 0.06 0.17 1.28
5.32 0.20 0.62 0.03 0.19 2.05
7.26 0.30 0.60 0.02 0.19 2.84
9.21 0.40 0.59 0.01 0.20 3.64

pean puts. σ = 0.2, r = 0.08, b = 0.10, T = 0.25, and K = 90.
Put Asset Delta Gamma Vega Theta
13.04 75 −0.89 0.02 0.05 0.83

8.61 80 −0.81 0.03 0.11 0.96
5.04 85 −0.61 0.04 0.16 1.01
2.55 90 −0.38 0.04 0.17 0.88
1.12 95 −0.26 0.03 0.13 0.61
0.43 100 −0.08 0.02 0.08 0.34
Sensitivity parameters for put options

Table 10.7 gives put prices and the sensitivity parameters when: σ = 0.2,
r = 0.08, b = 0.1, T = 0.25, and K = 90.
For an ATM put, (K = 90, S = 90), the gamma and the vega are
important.
The put’s theta increases when the option tends to parity and decreases
afterwards.
The same behavior applies for the put’s gamma and vega.
Table 10.8 gives the same information as Table 10.7, except for the
strike price which is changed from 90 to 100. For ITM puts, the delta
approaches −1, the gamma and vega are not important, and the theta is
very weak.
Table 10.9 gives OTM put prices (S = 100, K = 90) and the put price
sensitivities when the time to maturity varies from 0.05 to a year. The
values of the delta and gamma are weak. However, there is more vega and
theta for longer maturities.
Table 10.10 shows the same information for ATM put prices (S = K =
100). Note that the deltas and gammas are decreasing functions of the time

pean puts. σ = 0.2, r = 0.08, b = 0.10, T = 0.25, and K = 100.
22.65 75 −1.00 0 0 0.79

17.69 80 −0.97 0.01 0.02 0.85
12.94 85 −0.91 0.02 0.07 0.97
8.69 90 −0.73 0.03 0.14 1.09
5.26 95 −0.48 0.04 0.19 1.12
2.84 100 −0.38 0.04 0.19 0.98
Table 10.9. Changing time to maturity for European puts.

σ = 0.2, r = 0.08, b = 0.10, S = 100, and K = 90.
0.01 0.05 −0.01 0 0 0.07

0.08 0.10 −0.04 0.01 0.02 0.22
0.43 0.25 −0.08 0.02 0.08 0.34
0.91 0.5 −0.12 0.01 0.14 0.35
1.23 0.75 −0.13 0.01 0.19 0.32
1.45 1.00 −0.13 0.01 0.22 0.30

σ = 0.2, r = 0.08, b = 0.10, S = 100, and K = 100.
1.54 0.05 −0.45 0.09 0.09 1.85

2.04 0.10 −0.42 0.06 0.12 1.40
2.84 0.25 −0.38 0.04 0.14 0.98
3.43 0.50 −0.34 0.03 0.27 0.74
3.71 0.75 −0.31 0.02 0.31 0.62
3.83 1.00 −0.28 0.02 0.34 0.54
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.

T = 0.25, r = 0.08, b = 0.10, S = 100, and K = 90.
Put Volatility Delta Gamma Vega Theta
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. Changes in volatility, TM puts. T = 0.25, r = 0.08,

b = 0.10, S = 100, and K = 100.
Put Volatility Delta Gamma Vega Theta
0.21 0.05 −0.16 0.10 0.12 0.23

0.98 0.10 −0.30 0.07 0.17 0.55
2.84 0.20 −0.38 0.04 0.19 0.97
4.78 0.30 −0.41 0.03 0.19 1.34
6.73 0.40 −0.41 0.02 0.20 1.69
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.
10.2.2. Monitoring and adjusting the option position

in real time
10.2.2.1. Monitoring and managing the delta
The call’s delta is between zero and one and the put’s delta is between
0 and −1. When the delta is 0.5, the call price rises (falls) by 0.5 point
for each increase (decrease) in the asset price by 1 point. A delta of 0.5
corresponds to an ATM call.
For a deep ITM call, the variation in the asset price by one unit implies
an equivalent variation in the option price. The delta of an OTM call is
almost zero and of a deep ITM call is almost 1.
Since the put’s delta lies in the interval −1 and 0, a rise in the
underlying asset price implies a fall in the put price and vice versa.
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.
Table 10.13. Adjustment of the hedged portfolio as a function of the

Options Delta hedging when S rises Delta hedging when S falls
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
Table 10.14. Adjustment of a hedged position when the volatility changes.
Options Volatility increases Volatility decreases

Long a call
ITM Delta decreases: buy more S Delta increases: sell more S
ATM Delta nonadjusted Delta nonadjusted
OTM Delta decreases: sell more S Delta decreases: buy more S
Short a call
ITM Delta decreases: re-sell of S Delta increases: buy more S
OTM Delta increases: buy more S Delta decreases: sell more S
Long a put
ITM Delta decreases: re-sell of S Delta increases: buy more S
OTM Delta increases: buy more S Delta decreases: sell more S
Short a put
ITM Delta increases: buy more S Delta increases: sell more S
OTM Delta decreases: sell more S Delta decreases: buy more S
Table 10.15. Adjustment of a hedged position as a function of time.
Options Adjustment of the hedged position as a function of time
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
OTM Delta decreases: sell more S
Long a put
ITM Delta rises: buy more S
OTM Delta decreases: sell more S
Short a put
ITM Delta rises: sell more S
OTM Delta decreases: buy more S
Option market-makers implement often delta-neutral hedging strate-

gies in order to maintain a nil delta (in monetary unit). When the delta of
an option position is positive, this means that the market-maker is long the
underlying asset. If the asset price rises, he/she makes a profit since he/she
will be able to sell it at a higher price. However, if the asset price decreases,
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
Delta-neutral hedging is convenient for an investor who does not have

prior expectations about the market direction. However, if the investor
expects a rising market, he/she can have a positive delta, i.e., long the
underlying asset, so he/she can sell at a higher price when the market
effectively rises.
If the investor expects a down market, he/she can have a negative delta,
i.e., short the underlying asset, so he/she can buy it at a lower price when
the market effectively goes down.
10.2.2.2. Monitoring and managing the gamma

The gamma is given by the second derivative of the option price with respect
to the underlying asset price. A high value of gamma (either positive or
negative) shows a higher risk for an option position. The gamma shows
what the option gains (loses) in delta when the underlying asset price
rises (falls).
For example, when the option’s gamma is 4 and the option’s delta is
zero, an increase by 1 point in the underlying asset price allows the option
to gain 4 points in delta, i.e., the delta is equal to 4.
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.
Table 10.16. The adjustment of a hedged postion and the Gamma.
Position Gamma Adjustment of the position Effect on the position
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
Table 10.17. The variations in gamma and the time to maturity.
Longer T Short T Near T
Gamma Low High High for ATM

Gamma Low High Very low for OTM
Effect on position Low Easy adjustment Delta is very sensitive to S
Effect on position Low Easy adjustment Gamma is used with care
Table 10.18. Effect of the gamma on an option position.
OTM ATM ITM
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
For delta-neutral strategies, a positive gamma allows profits when the

market conditions change rapidly and a negative gamma produces losses in
the same context.
10.2.2.3. Monitoring and managing the theta

The theta is given by the option’s partial derivative with respect to the
remaining time to maturity. As maturity date approaches, the option loses
value. The theta is often expressed as a function of the number of points
lost each day. A theta of 0.4, means that the option loses $0.4 in value when
the maturity date is reduced by one day.
In general, the gamma and the theta are of opposite signs. A high
positive gamma is associated with a high negative theta and vice versa.
By analogy with the gamma, as a high gamma is an indicator of a
high risk associated with the underlying asset price, a high theta is an
indicator of a high exposure to the passage of time. An ATM option with
a short maturity loses value much more than a corresponding option on a
longer term. The theta of an ATM option is often higher than that of an
equivalent ITM or an OTM option having the same maturity date. The
option buyer loses the theta value and the option writer “gains” the theta
value (Table 10.19).
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.
Longer maturity Shorter maturity Near maturity
Loss in time value Low High Very high

Effect on position Needs passive monitoring Profit for seller profit for seller
Effect on position Needs passive monitoring Loss for buyer loss for buyer
When the Γ = −1500 for an option position, theta may be 10,000

i.e., a gain of $10,000 each day. However, the position implies a loss on
the underlying asset since the adjustments are done against the market
direction when the gamma is negative.
The theta of an option position is:

∂v1 ∂v2 ∂v3 ∂vj
Θposition = n1 − + n2 − + n3 − + · · · + n1 −
∂t ∂t ∂t ∂t
10.2.2.4. Monitoring and managing the vega

The vega is given by the option’s derivative with respect to the volatility
parameter. It shows the induced variation in the option price when the
volatility varies by 1%.
The vega is always positive for call and put options since the option
price is an increasing function of the volatility parameter. A vega of 0.6
means that an increase in the volatility by 1% increases the option price
by 0.6. For a fixed time to maturity, the vega of an ATM option is higher
than that of an ITM or an OTM option.
Since all the option pricing parameters are observable, except the
volatility, buying (selling) options is equivalent to buying (selling) the
volatility.
When monitoring an option position, a trade off must be realized
between the gamma and the vega. Buying options and hence having a
positive gamma is easy to manage. However, when the implicit volatility
falls, the investor must adopt one of the two following strategies.
He/she can either preserve a positive gamma, if he/she thinks that the
loss due to a decrease in volatility will be compensated by adjusting the
gamma in the market direction.
He/she can sell the volatility (options) and re-establish a position with
a negative gamma.
In this case, the losses due to the adjustments of the delta must
be sufficient to compensate for the decrease in volatility (Tables 10.20
and 10.21).
Table 10.20. Effect of the volatility on a portfolio of options.
Options Volatility Effect
Long Long Profit (loss) when the volatility rises (falls)

Short Short Loss (profit) when the volatility rises (falls)
Table 10.21. Effect of the vega with respect to time to maturity.
Longer maturity Shorter maturity Near maturity
Vega High Low Implicit volatility

depends on other factors
Effect Very sensitive Little sensitivity Implicit volatility
depends on other factors
Table 10.22. Effect of the vega on the option price.
OTM ATM ITM
Vega depends on T For a given T , vega is higher Depends on T

Option position Low High Low
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
10.3. The Characteristics of Volatility Spreads

As a simple standard option, a spread is also characterized by its delta,
gamma, theta, and vega. These sensitivity parameters allow the investor
to manage his/her, option positions as a consequence of the changes in
the market conditions. The implementation of strategies based on volatility
spreads, implies often the use of delta-neutral strategies to be able to predict
the variations in the market conditions.
When the changes in the underlying asset value give more value to
the spread, the gamma is positive. On the other direction, the spread’s
gamma exhibits is negative when the variations in the underlying asset
price reduce the spread value. Since the effects of these changes in the
underlying asset price and the time to maturity operate in the opposite
side, a spread with a positive gamma shows a negative theta and vice versa.
Table 10.23. Characteristics of volatility spreads.
Strategy, Position ∆ Γ Θ Vega
Short a call ratio spread 0 + − +

Short a put spread 0 + − +
Long a straddle 0 + − +
Long a strangle 0 + − +
Short a butterfly 0 + − +
Long call ratio spread 0 − + −
Long a put ratio spread 0 − + −
Short a straddle 0 − + −
Short a strangle 0 − + −
Long a butterfly 0 − + −
Table 10.23 summarizes the effect of the sensitivity parameters on various

spread strategies.
The investor can implement delta-neutral strategies when he/she has
no prior anticipations as to where the market is going. However, he/she can
resort to bullish and bearish spreads when he/she is confident about the
market timing. This leads him/her to be long or short the underlying asset.
When the options used are overvalued, (according to the investor), for
instance, when the implied volatility is high (with respect to the historical
volatility and its normal level), the investor can sell some puts if the market
rises and some calls if the market falls.
When the options are undervalued, (according to the investor), for
instance, when the implied volatility is low (with respect to the historical
volatility and its normal level), the investor can be long calls and puts when
the market falls.
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.
Color corresponds to the gamma’s derivative with respect to the time

remaining to maturity.
The theta measures the change in the option price as time elapses.
The vega or lambda is a measure of the change in the option price for
a small change in the underlying asset’s volatility.
This chapter presented the main Greek-letter risk measures, i.e.,
the delta, the gamma, the theta, and the vega in the context of the
European analytical models. These risk measures are simulated for different
parameters, which enter the option formulas. Then the magnitude of these
risk measures is appreciated in connection with the management and the
monitoring of an option position.
The knowledge of the changes in these risk parameters is necessary for
the management of an option position and the determination of the profits
and losses associated with the portfolio. To put it differently, the pricing of a
European call option can be viewed as requiring inputs (the underlying asset
and the Treasury bill) and a production technology (the hedge portfolio and
the Greek-letter risk measures). In a B–S world, by tracking continuously
the hedge ratio (being delta-neutral), the investor makes sure that the
duplicating portfolio does mimic the call option, namely does “produce” the
option. In the course of doing so, the investor controls his/her production
costs and protects his/her mark up on the option. However, these risk
measures depend on the theoretical model used for the valuation and the
management of the option position. This is why, one could call such a
risk measure as a technological risk. This position must be adjusted nearly
continuously, in response to the changes in the market conditions.
Appendix A: Greek-Letter Risk Measures

in Analytical Models
A.1. B–S model
Call sensitivity parameters
∂∆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
Put sensitivity parameters
∂∆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
A.2. Black’s Model

The option sensitivity parameters in the Black’s model are presented as
follows.
∂∆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
A.3. Garman and Kohlhagen’s model

∗
−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
A.4. Merton’s and Barone-Adesi and Whaley’s model

∂∆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
∆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
Appendix B: The Relationship Between Hedging

Parameters
Using the definitions of the delta, gamma, and theta, the B–S equation can
be written as:

∂c(S, t) ∂c(S, t) 1 2 2 ∂ 2 c(S, t)
= rc(S, t) − rS − σ S
∂t ∂S 2 ∂S 2
or
1
−Θ = −rc(S, t) + rS∆ + σ2 S 2 Γ
2
or
1
rc(S, t) = Θ + rS∆ + σ2 S 2 Γ.
2
For a delta-neutral position, the following relationship applies:
1
rc(S, t) = Θ + σ2 S 2 Γ
2
Using the definitions of the hedging parameters, the Black equation can
be written as:

∂c(F, t) 1 2 2 ∂ 2 c(F, t)
= rc(F, t) − σ F
∂t 2 ∂S 2
or
1
Θ = rc(F, t) − σ2 F 2 Γ.
2
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
Appendix C: The Generalized Relationship Between

the Hedging Parameters
We denote respectively by:
• 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.
If we denote by Hi , the charm asociated with the asset i, the theta

given in Garman (1992) is:
N

Θ= S i Hi
i=1
with Hi = ∂∆ i
∂t .
The proof of this result relies on the replication equation which states
that:
N

Si ∆i = C
i=1
This equation shows that an outright purchase of the security is equivalent

to the replicating strategy, which consists of buying or selling the amount
dictated by the delta.
The generalized B–S equation can also be expressed as a function of

the greek-letter risk measures as follows:
N N N
1
σi σj ρij Si Sj Γij + (r − ri )Si ∆i + Θ = rC
2 i=1 j=1 i=1
The use of charm allows to write a potentially more separable equation

for derivative assets:
N N N
1
σi σj ρij Si Sj Γij + Si [Hi − ri ∆i ] = 0
2 i=1 j=1 i=1
Appendix D: A Detailed Derivation of the Greek Letters

Re-call that the call option formula is given by:
C = Se(b−r)T N (d1 ) − Ke−rT N (d2 )

S
ln K + (b + 12 σ 2 )T √
d1 = √ , d2 = d1 − σ T
σ T
where r stands for the continuous interest rate, τ is the discrete interest
rate, and b is the cost of carry.
The following relationship applies between interest rates:
1
e−rT =
1 + τT
which is equivalent to
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
The derivation of the call’s delta
∂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
The following relationship is used to obtain the desired result:

√ √
d22 = (d1 − σ T )2 = d21 − 2d1 σ T + σ 2 T

√ S 1 2 S
d1 σ T = ln + b + σ T, d22 = d21 − 2 ln − 2bT
K 2 K
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
The call’s delta is given by:
∆ = e(b−r)T N (d1 ).
The derivation of the put’s delta

Re-call that the put’s formula is:
P = Ke−rT N (−d2 ) − Se(b−r)T N (−d1 )
with N (−d2 ) = −(N (d2 ) − 1) and N (−d1 ) = −(N (d1 ) − 1)

The delta is given by:
∂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
The following results are used in the computations.

d2 d2

d2
d2

1 S 1
(− 2 ) bT (− 1 ) − +ln +bT bT −
Ke 2 − Se e 2 = Ke 2 K − Se e 2
and
1
−(−d1 ) = (−d2 ) = − √
Sσ T
Hence, the put’s delta is:
∆ = −e(b−r)T N (−d1 ) or ∆ = e(b−r)T N (d1 ) − e(b−r)T .

The derivation of the call’s gamma

The gamma can be computed as follows:
∂∆ ∂d1
Γ= or Γ = e(b−r)T N (d1 ) ,
∂S ∂S
1 d2
1 1
Γ = √ e 2 e(b−r)T √
2Π Sσ T
So, we obtain:
1 −d2
1
Γ= √ e(b−r)T e 2 .
σS 2ΠT
The derivation of the put’s gamma

The gamma is given by:

∂∆ 1 −
d2
1 1
Γ= = −e(b−r)T √ e 2 − √
∂S 2π Sσ T
or
d2

1 − 1
Γ= √ e 2 e(b−r)T .
Sσ 2πT
The derivation of the call’s Vega

The following equalities are used to simplify the computations. The partial
derivatives with respect to the volatility are:
√ √ √
S
∂d1 σ2 T T − T ln K + T b + 12 σ 2 T √ d1 d2
= 2
= T− =−
∂σ σ T σ σ
∂d2 ∂d1 √ d1 d2 √
= − T =− =− − T
∂σ ∂σ σ σ
The following equality applies:

√ 2 √
d21 = d2 + σ T = d22 + 2d2 σ T + σ 2 T
which is also:

S 1 2
d21 = d22 + 2 ln + 2 b − σ T + σ2 T
K 2
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
The derivation of the put’s vega
∂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
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
The derivation of the call’s Rho with respect to r

The rho corresponds to the option partial derivative with respect to the
interest rate:
∂C
ρ= .
∂r
When the cost of carry is given by the difference between the domestic
and foreign interest rate:
b = r − rf .
The call formula for a currency option becomes:
C = Se−rf T N (d1 ) − Ke−rT N (d2 )
Hence:
∂C
ρ= = Se−rf T N (d1 )(d1 ) + T Ke−rT N (d2 ) − Ke−rT N (d2 )(d2 )
∂r
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
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:
ρmon = T Ke−rT N (d2 ).
The following equalities are used to obtain the desired result.

T √ 2
d1 = √ = d22 = d1 − σ T
σ T
or
√
d1 = d21 − 2d1 σ T + σ2 T
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
The derivation of the call’s Rho with respect to rf

∂C
ρdev =
∂rf
ρdev = −T Se−rf T N (d1 ) + Se−rf T N (d1 )(d1 ) − Ke−rT N (d2 )(d2 )

d2 d2
1 1 T 1 2 T
+ Se−rf T √ e − 2 √ − KerT √ e − 2 √
2Π σ T 2Π σ T
or
ρdev = −T Se(b−r)T N (d1 )

Hence, we have:
ρdev = −T Se−rf T N (d1 ).
Remark: For a futures contract, b = 0, and the call formula can be

written as:
C = e−rT (SN (d1 ) − XN (d2 ))
The derivation of the put’s Rho with respect to rf
∂P
= Ke−rT N (−d2 )(−d2 ) + Se−rf T N (−d1 ) − Se(b−r)T N (−d1 )(−d1 )
∂rf
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:
ρdev = −ST e−rf T (N (d1 ) − 1)
The following relation is used in the derivation:

T
d2 = −d1 = − √ .
σ T
The derivation of the call’s theta

The call’s theta is computed as:
∂C
θ= = (b − r)Se(b−r)T N (d1 ) + Se(b−r)T N (d1 )(d1 )
∂T
+ rKe−rT N (d2 ) − Ke−rT N (d2 )(d2 )
This is equivalent to:

d2

1 − 1
θ = (b − r)Se(b−r)T N (d1 ) + rKe−rT N (d2 ) + Se(b−r)T √ e 2 (d1 )
2π
d2
1 2
−Ke−rT √ e − 2 (d2 )
2π
or
θ = (b − r)Se(b−r)T N (d1 ) + rKe−rT N (d2 )

1
d21 d2
2

+ √ e−rT Se 2 (d1 )ebT − Ke − 2 (d2 )
2π
The following results are used in the computations.
√ S
b + 12 σ 2 σ T − 12 √σT ln K + b + 12 σ 2 T

d1 =
σ2 T
and
1 σ
d2 = d1 − √
2 T
We use also the fact that:
√ √
d21 = (d2 + σ T )2 = d22 + 2d2 σ T + σ 2 T
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
can be written as:

„
d2
«

1 − 21 −ln S
−bT d2
2
−
√ e−rT (d1 )erT (d2 )
K
Se − Ke 2
2π

K −
d2
2 K −
d22 1 σ
= √ e−rT e 2 ((d1 ) − (d2 )) = √ e−rT e 2 √
2π 2π 2 T
Hence, the call’s theta is given by:

Kσ −rT −
d2
2
θ = (b − r)Se(b−r)T N (d1 ) + rKe−rT N (d2 ) + √ e e 2
2 2πT
The derivation of the put’s theta

The put’s theta is given by:
∂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
θ = (b − r)Se(b−r)T (N (−d1 ) − 1) + rKe−rT (N (−d2 ) − 1)

2 d2
e−rT d
− 22 bT − 21
+√ Ke (−d2 ) − Se e (−d1 )
2πT
„ «
d2 d2
S
− 22 −ln K −bT e−rT 2
bT
− Se e (−d1 ) + √ Ke − 2 (−d2 ) − (−d1 ))
2πT
The following relation is used in the derivation:
1 σ
(−d2 ) = −(−d1 ) + √
2 T
Hence, the theta is given by:
θ = (b − r)Se(b−r)T (N (d1 ) − 1) + rKe−rT (N (d2 ) − 1)

d2
σ 2
+ √ Xe−rT e − 2
2 2πT
The derivation of the put’s rho

The put’s Rho is given by:
∂P
= −T Ke−rT N (−d2 ) + Ke−rT N (−d2 )(−d2 )
∂r
− Se(b−r)T N (−d1 )(−d1 )
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
The call’s partial derivative with respect

to the strike price
The computation of the partial derivative with respect to K gives:
∂C ∂d1 ∂d2
= Se(b−r)T N (d1 ) − e−rT N (d2 ) − Ke−rT N (d2 )
∂K ∂K ∂K
with
ln S − ln K + (b + 12 σ 2 )T √
d1 = √ , d2 = d1 − σ T
σ T
The partial derivatives of d1 and d2 are:
∂d1 1 ∂d2 1
=− √ , =− √
∂K Kσ T ∂K Kσ T
Hence, we have:
∂C 1 d2
1 1
= −Se(b−r)T √ e− 2 √ − e−rT N (d2 )
∂K 2Π Kσ T
1 d22 1
+ Ke−rT √ e− 2 √
2Π Kσ T
or
d2
1 d2
2
∂C Se(b−r)T e− 2 e−rT e− 2
=− √ − erT N (d2 ) + √
∂K Kσ 2ΠT σ 2ΠT
Using the following relationship:
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
The put’s partial derivative with respect to the strike price

The partial derivative of the put price with respect to K can be computed
in the same way.
∂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
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
13. How does the theta change in response to the changes in the option
14. How does the vega change in response to the changes in the option
15. How a hedged portfolio is adjusted in response to the changes in the
underlying asset price?
volatility parameter?
time to maturity?
18. What are the main characteristics of volatility spreads?
Graphique
Appendix
References
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.

Part IV
Mathematical Foundations of Option Pricing Models

in a Continuous-Time Setting: Basic Concepts and
Extensions
491
492
Chapter 11
THE DYNAMICS OF ASSET PRICES AND THE

ROLE OF INFORMATION: ANALYSIS AND
APPLICATIONS IN ASSET AND RISK
MANAGEMENT
Chapter Outline
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
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.
The Dynamics of Asset Prices and the Role of Information 495
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.
11.1. Continuous Time Processes for Asset

Price Dynamics
The dynamics of asset prices are often represented as a function of a Wiener
process or what is also known as Brownian motion.
11.1.1. Asset price dynamics and Wiener process

The Wiener process has some interesting properties and can be introduced
with respect to a change in a variable W over a small interval of time t.
Wiener process or Brownian motion

Let W denote a variable following a Wiener process and ∆W a change in
its value over a small interval of time ∆t. The relation between ∆W and
∆t is given by the following equation:
√
∆W = ξ ∆t (11.1)
where ξ is a random sample from a normal distribution having a zero mean

and a unit standard deviation. If one takes two reasonably short intervals of
time, then the values of ∆W are independant. Using these properties, it is
clear that ∆W has √also a normal distribution with a zero mean, a standard
deviation equal to ∆t, and a variance of ∆t.
Now, if one considers the change in W over a longer time period [0, T ],
composed of N periods of length ∆t, i.e., T = N ∆t, then the change in W ,
from W (0) to W (T ), or W (T ) − W (0), over this period of time is equal to
the sum of changes over shorter periods. Hence, one can write:
N
√
W (t) − W (0) = ξi ∆t (11.2)
i=1
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)
The martingale property and the Brownian motion

The notion of martingale is useful in financial models, particularily, when
analyzing the concept of arbitrage. A martingale can be defined as follows.
Consider a probability space (Ω, F, P ) and a filtration (Ft )t≥0 . An adapted
family (Mt )t≥0 of integrable random variables having a finite mean is a
martingale when for all s ≤ t, then we have:
E(Mt | Fs ) = Ms (11.4)
where E(.) stands for the mathematical conditional expectation operator.

Appendix B gives the main properties of the conditional expectation
operator. The notion of martingale asserts that the best approximation
of Mt , given all the available information Fs , is Ms . In terms of financial
markets, this means that the best way to predict futures prices is to use the
current prices. Hence, using current information is equivalent to using all
the historical information, since only the most recent information matters.
Using this definition, it must be clear that when (Mt )t≥0 is a martingale,
then E(Mt ) = E(M0 ). The following result is advanced without proof.
If (Mt )t≥0 is an (Ft )t — Brownian standard motion, then (Wt )t is an
1 2
(Ft )t — martingale, Wt2 — t is an (Ft )t — martingale, and e(σWt − 2 σ t)
is an (Ft )t — martingale.
The first and second properties characterize the standard Brownian

motion. The third property is useful when studying the dynamics of
financial asset prices. In fact, as will be shown later, the price of a stock is
often written as:
1 2
St = S0 e(σWt − 2 σ t) . (11.5)
11.1.2. Asset price dynamics and the generalized

Wiener process
For a variable X, a generalized Wiener process can be expressed as:
dX = adt + bdW (11.6)
where a and b are constants.

This process shows the dynamics of the variable X in terms of time
and dW . The first term, adt, called the deterministic term, means that the
expected drift rate of X is ‘a’ per unit time. The second term, bdW , called
the stochastic component, shows the variability or the noise added to the
dynamics of X. This noise is given by b times the Wiener process. When
the stochastic component is zero, dX = adt, or dX dt
= a. This is equivalent
to X = X0 + at. Hence, the value of X at any time is given by its initial
value X0 plus its drift multiplied by the length of the time period. Now,
it is possible to write the equivalent of Eq. (11.6) using Eq. (11.1) for a
longer ∆t:
√
∆X = a∆t + bξ dt. (11.7)
∆X has a normal distribution, its mean is a∆t, its

Hence, as before, since √
standard deviation is b ∆t, and its variance is b2 ∆t.
11.1.3. Asset price dynamics and the Ito process

An Ito process for a variable X can be written as follows:
dX = a(X, t)dt + b(X, t)dW. (11.8)
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.
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
where Kt and Ht are stochastic processes adapted to Ft for which the

integral corresponding to the second-order moment is finite. It will be shown
later that the second integral in the above expression is a martingale.
The dynamics of stock prices

The dynamics of the stock price S are represented by the following Ito
process with a drift rate, µS, and a variance rate σ 2 S 2 :
dS = µSdt + σSdW. (11.10)
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
so that S = S0 eµt . When the variance rate is not zero and σ2 S 2 ∆t is

the variance of the actual change in S during ∆t, the dynamics of the
stock price are given by the expected instantaneous increase in S plus
its instantaneous variance times the noise dW . This discrete time version
says that the proportional return on the stock S, over a short period of
time, is given by an expected return µ∆t and a stochastic return σξ∆dt.
Hence, ∆S S is normally distributed with a mean µ∆t and a standard
√
deviation σ ∆t, or:
∆S √
∼ N (µ∆t, σ ∆t) (11.13)
S
Some numerical examples

Example: Consider the XYZ stock price characterized by an expected
return of 14% per annum and a standard deviation or volatility of 35% per
annum. The initial stock price is 100 . Using Eq. (11.13), the dynamics of
this stock price are given by:
dS
= 0.14dt + 0.35dW (11.14)
S
or for a small interval,
dS √
∆t : = 0.14∆t + 0.35ξ ∆t.
S
If the time interval ∆t is 2 weeks (or 0.03846 year), then the price
increase is given by
√
∆S = 100[0.14(0.03846) + 0.35ξ 0.03846]
or
∆S = 100(0.005384 + 0.068639ξ).
The price increase is a random sample from a normal distribution with

a mean of 0.538 and a volatility of 6.86.
Example: Consider the Y Z stock price, having an expected return of 20%
per annum and a standard deviation or volatility of 25% per annum. Over
a time interval of 3 days, or 0.008219178 per year, ∆S
S
is normal with,
∆S
∼ N (0.00164, 0.0226)
S
11.1.4. The log-normal property

Using the previous example, since the change in the underlying asset
price
between
time t and time ti , is normally distributed, with a mean
1 2
µ − 2 σ (ti − t) and a variance σ 2 (ti − t), we have

1
ln(Sti ) − ln(St ) ∼ N µ − σ2 (ti − t), σ (ti − t) (11.15)
2
or

1 2
ln(Sti ) ∼ N µ − σ (ti − t) + ln(St ), σ (ti − t) .
2
Hence, Sti has a log-normal distribution.
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
ln(Sti ) ∼ N [4.6576, 0.212].
11.1.5. Distribution of the rate of return

Let α to be a continuously compounded rate of return. What is the
distribution of α?
At a future date ti , the stock price can be written as:
Sti = St eα(ti −t)

1 Sti
and α = ti −t
ln St
. Using the log-normal property, i.e.,

Sti 1 2
ln ∼N µ − σ (ti − t), σ (ti − t)
St 2
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%.
11.2. Ito’s Lemma and Its Applications

Financial models are rarely described by a function that depends on a
single variable. In general, a function which is itself a function of more than
one variable is used. Ito’s lemma, which is the fundamental instrument
in stochastic calculus, allows such functions to be differentiated. First,
we derive Ito’s lemma with reference to simple results using Taylor series
approximations. We then give a more rigorous definition of Ito’s theorem.
The formula for Taylor series is given in Appendix C.
11.2.1. Intuitive form

Let f be a continuous and differentiable function of a variable x.
If ∆x is a small change in x, then using Taylor series, the resulting
change in f is given by:

df 1 d2 f 2 1 d3 f
∆f ∼ ∆x + ∆x + ∆x3 + · · · (11.17)
dx 2 dx2 6 dx3
If f depends on two variables x and y, then Taylor series expansion
of ∆f is:

∂f ∂f 1 ∂ 2f 2 1 ∂ 2f
∆f ∼ ∆x + ∆y + ∆x + ∆y 2
∂x ∂y 2 ∂x2 2 ∂y 2
2
∂ f
+ ∆x∆y + · · · (11.18)
∂y∂y
In the limit case, when ∆x and ∆y are close to zero, Eq. (11.18) becomes:

∂f ∂f
∆f ∼ dx + dy. (11.19)
∂x ∂y
Now, if f depends on two variables x and t in lieu of x and y, the analogous
to Eq. (11.18) is,

∂f ∂f 1 ∂ 2f 2 1 ∂ 2f
∆f ∼ ∆x + ∆t + ∆x + ∆t2
∂x ∂t 2 ∂x2 2 ∂t2
2
∂ f
+ ∆x∆t + · · · (11.20)
∂x∂t
Consider a derivative security, f (x, t), whose value depends on time and on
the asset price x. Assuming that x follows the general Ito process,
dx = a(x, t)dt + b(x, t)dW

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
This is exactly Ito’s lemma.

Substituting a(x, t)dt + b(x, t)dW for dx, Eq. (11.21) gives:

∂f ∂f 1 ∂2f ∂f
df = a+ + b2 dt + bdW. (11.22)
∂x ∂t 2 ∂x2 ∂x
Example: Let us denote by X(t) a standard Brownian motion. Using

X 4 (t), show that the following integral is equal to:
t t
1 3
X (τ )dX(τ ) = X 4 (t) −
3
X 2 (τ )dτ.
0 4 2 0
Solution: We apply Ito’s lemma for a function F (X(t)):

dF dF 1 d2 F
dF = dX + dt + dt.
dX dt 2 dX 2
Let, F (X(t)) = X 4 (t) and note that,

t
dF = [F (t)]t0 .
0

t
dF (X(t)) = X 4 (t) − X 4 (0) = X 4 (t) as X 4 (0) = 0.
0
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

t t
4X 3 (τ )dX(τ ) + 6X 2 (τ )d(τ )
0 0
t
6 1
= X 3 (τ )dX(τ ) + X 2 (τ )d(τ ) = X 4 (t)
0 4 4
and we have,
t t
1 4 3
X 3 (τ )dX(τ ) = X (t) − X 2 (τ )dτ.
0 4 2 0
Example: Using Ito’s lemma, show the following result:

t t
3 1
τ 3 X(τ )dX(τ ) = t3 X 2 (t) − τ 2 X 2 (τ )dτ − τ 4
0 2 0 8
or, Ito’s lemma will be applied for a function F (X(t), t) = t3 X 2 (t):

dF dF 1 d2 F
dF = dX + + dt.
dX dt 2 dX 2
In this case:
dF = 2t3 X(t)dX + (3t2 X 2 (t) + t3 )dt
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
or, re-arranging the formula, we obtain the following result:

t t t
1 3 2 3 1
τ 3 X(τ )dX(τ ) = t X (t) − τ 2 X 2 (τ )dτ − τ 3 dτ
0 2 2 0 2 0
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
Example: Use Ito’s lemma to show that:

t t
1 5
X 5 (τ )dX(τ ) = X (t) − 2 X 3 (τ )dτ.
0 5 0
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
11.2.2. Applications to stock prices

Example: Apply Ito’s lemma to a function f (S, t) when the dynamics of
stock prices are described by the following stochastic equation:
dS = µSdt + σSdW.
Equation (11.22) gives:

∂f ∂f 1 ∂2f 2 2 ∂f
df = µS + + σ S dt + σSdW.
∂x ∂t 2 ∂S 2 ∂t
(11.23)
Example: Apply Ito’s lemma to derive the process of f = ln(S).

First calculate the derivatives,
2
∂f 1 ∂ f 1 ∂f
= ; 2
= − 2; = 0.
∂S S ∂S S ∂t
Then from Ito’s lemma, one obtains

∂f ∂f 1 ∂ 2f 2 2 ∂f
df = µS + + σ S dt + σSdW
∂x ∂t 2 ∂S 2 ∂S
or

1
df = µ − σ2 dt + σdW.
2
This last equation shows
f follows a generalized Wiener process with
that,
a constant drift of µ − 12 σ 2 and a variance rate of σ 2 .
11.2.3. Mathematical form

The following theorem gives Ito’s formula.
Theorem: If (Xt)0≤t≤T is an Ito process, i.e.,
t t
Xt = X 0 + Ks ds + Hs dWs (11.24)
0 0
and f is a continuous function with second-order continuous derivatives,

then:
t
1 t
f (Xt ) = f (X0 ) + f (Xs )dXs + f (Xs )dX, Xs (11.25)
0 2 0
where by definition,
t
dX, Xt = Hs2 ds.
0
The first integral is given by:

t t t
f (Xs )dXs = f (Xs )Ks ds + f (Xs )Hs dWs .
0 0 0
The second integral is given by:

t t
f (Xs )dX, Xs = f (Xs )Hs2 ds.
0 0
Hence, Ito’s formula is:

t
f (Xt ) = f (X0 ) + f (Xs )Ks ds
0
t
1
+ f (Xs )Hs dWs + f (Xs )Hs2 ds. (11.26)
0 2
More generally, if f (x, t) has first-order continuous partial derivatives in t

and continuous second-order derivatives in x, then:
t t
f (t, Xt ) = f (0, X0 ) + fs (s, Xs )ds + fx (s, Xs )dXs
0 0
t
1
+ fxx (s, Xs )dX, Xs . (11.27)
2 0
Note that if we put Ks = 0 and Hs = 1, in the Ito process, i.e.,

t t
Xt = X0 + Ks ds + Hs dWs (11.28)
0 0
then it reduces to Xt = Wt , which is simply the Brownian motion, since

W0 = 0.
Example: In this example, Ito’s formula is applied to the dynamics of the

squared Brownian motion, Wt2 . When f (x) = x2 and Xt = Wt , (the case
where Ks = 0 and Hs = 1), we have:
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.
Example: This application of Ito’s lemma concerns the calculation of an

explicit solution to the process describing the dynamics of stock prices. Let
us look for the solutions to (St )t≥0 for the following equation:
t
St = S0 + Ss [µSdS + σdWs ] (11.29)
0
which is often written in the form dSt = St (µdt + σdWt ).

Let Yt = ln(St ), where St is solution to the preceding equation. Since
St follows an Ito process with Ks = µSs and Hs = σSs , application of Ito’s
formula to f (S) = ln(S) gives:
t t
dSs 1 −1 2 2
ln(St ) = ln(S0 ) + + σ Ss ds. (11.30)
0 Ss 2 0 Ss2
Since f (S) = ln(S), f (S) = S1 , f (S) = −1

S 2 , and the term corresponding to
the equivalence of dX, Xs is the instantaneous variance of dSs , Var(dSs ) =
σ 2 Ss2 ds. Substituting Eq. (11.29) in Eq. (11.30) and simplifying gives:
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 .
Applying the exponential to Yt gives the solution to Eq. (11.29)

1 2
St = S0 e(µ− 2 σ )t+σdWt .
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:
dBt = rBt dt (11.33)
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:
dSt = µSt dt + σSt dWt (11.34)
where µ and σ are constants and Wt is a standard Brownian motion.

As we have shown, the solution to Eq. (11.34) is:
1 2
St = S0 e(µ− 2 σ )t+σdWt
where, S0 is the initial asset price at time 0.
11.2.4. The generalized Ito’s formula

Ito’s formula can be generalized to the case where the function depends on
several Ito processes, which are expressed in terms of standard independent
Brownian motions. This is the multidimensional Ito’s formula or the vector

form of Ito’s lemma. The formula is useful in deriving interest rate models
and models of derivative asset pricing with several state variables.
Generalization of Ito’s lemma: The first form: The generalization of

Ito’s lemma is useful for a function that depends on n stochastic variables
xi , where i varies from 1 to n. Consider the following dynamics for the
variables xi ,
dxi = ai dt + bi dzi (11.35)
Using Taylor series expansion of f gives:
∂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)
Generalization of Ito’s lemma: The second form: Consider Wt =

(Wt1 , . . . , Wtp ) where (Wti )t≥0 are independent standard Brownian motions
and Wt is a p-dimensional Brownian motion. The mathematical

expression of Ito’s formula is introduced with respect to n Ito processes
(Xt1 , Xt2 , . . . , Xtn )
t p
t
Xti = X0i + Ksi ds + Hsij dXsi .
0 i=1 0
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 ).
11.2.5. Other applications of Ito’s formula

t
1 n − 1 t n−2
X n−1 (τ )dX(τ ) = X n (t) − X (τ )dτ
0 n 2 0
Solution: When F = X n (t), n ∈ N ∗ then

t
dF = X n (t) − X n(0) = X n (t)
0
since X(0) = 0.
Using Ito’s lemma gives:
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
Solution: When F = X n(t), where n, m ∈ N ∗ then
t
dF = tm X n (t) − tm X n (0) = tmX n (t)
0
since X(0) = 0.
1
dF = ntm X n−1 dX + n(n − 1)X n−2dt + mtm−1 X n (t)dt.
2
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
Solution: Consider the following function F = f (t)X(t). In this case, we

have
t
dF = f (t)X(t) − f (0)X(0) = f (t)X(t)
0
since X(0) = 0 and by Ito’s lemma:

dF = f dX + Xdf.
Therefore,
t t
f (τ )dX(τ ) + X(τ )df (τ ) = f (t)X(t)
0 0
which can be written as:

t t
f (τ )dX(τ ) = f (t)X(t) − X(τ )df (τ ).
0 0
11.3. Taylor Series, Ito’s Theorem and the Replication

Argument
Let us denote by c(S, t) the option value at time t as a function of the
underlying asset price S and time t. Assume that the underlying asset
price follows a geometric Brownian motion:
dS
= µdt + σdW (t) (11.39)
S
where µ and σ 2 correspond, respectively to the instantaneous mean and the

variance of the rate of return of the stock.
11.3.1. The relationship between Taylor series

and Ito’s differential
Using Taylor series differential, it is possible to express the price change of
the option over a small interval of time [t, t + dt] as:

∂c ∂c 1 ∂2c
dc = dS + dt + (dS)2 (11.40)
∂S ∂t 2 ∂2S
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
which can be written as the sum of two components corresponding to the

Ito’s differential dcI and an error term de(t):
dc = dcI + de(t) (11.41)
where

∂c ∂c 1 ∂ 2c
dcI = dS + dt + σ 2 S 2 dt
∂S ∂t 2 ∂ 2S
and

1 ∂2c
de(t) = σ 2 S 2 dW 2 − dt .
2 ∂2S
11.3.2. Ito’s differential and the replication portfolio

A replication argument is often used in financial theory. It means simply
that in complete markets, the payoff of an asset can be created or duplicated
using some other assets. A combination of these assets provides a similar
payoff as that of the original asset.
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
where ΠR refers to the replicating portfolio.
An extension to account for information costs: Consider, for example,

a financial institution using a given market. If the costs of portfolio selection
and models conception, etc., are computed, then it can require at least a
return of say, for example λ = 3%, before acting in this market. This cost
is in some sense the minimal cost before acting in a certain market. It
represents somehow, the minimal return required before implementing a
given strategy. For the analysis of information costs and valuation, we can
refer to Bellalah (1999, 2000, 2001), Bellalah et al. (2001a,b), Bellalah and
Prigent (2001), Bellalah and Selmi (2001), and soon. If you consider the
above replicating strategy, then the returns from the replicating portfolio
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.
11.3.3. Ito’s differential and the arbitrage portfolio

If one uses arbitrage arguments, then the option value must be equal to the
value of its replicating portfolio.
The standard analysis: Using arbitrage arguments, we must have
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.
A hedged position or portfolio is a portfolio whose return at equilibrium

must be equal in theory to the short-term risk-free rate of interest. This is
the main contribution of Black–Scholes (1973) to the pricing of derivative
assets. Merton (1973) uses the same argument as Black–Scholes (1973) by
implementing the concept of self-financing portfolio. This portfolio is also
constructed by buying the option and selling the replicating portfolio or
vice versa. The condition on the self-financing portfolio is:

∂c
ΠA = c − S − Qc = 0
∂S
where S refers to the self-financing portfolio. The ommited error term in

the above analysis, de(t), can reflect a replication error, a hedging error or
an arbitrage error. It can have different interpretations. The term de(t) is
neglected or omitted when the revision of the portfolio is done to allow for
the replicating portfolio to be self-financing. When this term is positive, this
may refer to an additional cash that must be put in the portfolio. When it
is negative, a withdrawal of cash from the portfolio is possible.
An extension to account for information costs: In the same way, the

previous analysis can be extended to acount for information costs. In this
context, we must have:
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
This equation corresponds to an extended version of the well-known Black–

Scholes–Merton partial differential equation accounting for the effect of
information costs. For the sake of simplicity, we assume that information
costs are equal in both markets: the option market and the underlying
asset market. Or in practice, institutions and investors support these costs
on both markets. Therefore, a more suitable analysis must account for two
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
11.3.4. Why are error terms neglected?

Now, we give an answer to the following question:
If each portfolio revision is self-financing and the hedging error de(t)
tends to zero as the interval of time becomes extremely small, is there a
mathematical or an economic “rationale” in ignoring the error term?
In the Black and Scholes (1973) theory, the term de(t) is ignored
because it is not correlated with the underlying asset price in the context of
the capital asset pricing model (CAPM). In this context, it is regarded as a
diversifiable risk. This justification is referred to as the “equilibrium option
pricing theory”. However, if one uses the Black and Scholes theory with
respect to the implementation of Ito’s lemma, then ignoring the error term,
refers to a pure-arbitrage result. Omberg (1991) proposed two explanations
for pure arbitrage results with respect to the two following assumptions H1
and H2 .
According to H1 , the error term is of order o(dt), which is a higher
order than dt:
H1 : de(t) = o(dt). (11.48)
According to H2 , the error term is of order Odt:
H2 : de(t) = O(dt) (11.49)
and
τ
de(t) = 0 for τ > 0.
0
In H1 , the error is neglected because in the limit, the hedging error

disappears more quickly than the risk-free return. This is because the risk-
free return is of order dt. In this context, the replicating portfolio is perfect
because the smallness of the interval justifies the disparition of the hedging
error.
In H2 , the price dynamics are represented by the Taylor’s series

differential rather than Ito’s differential. In this case, cumulation of hedging
errors over several time intervals is necessary.
Merton (1973) considered that the error term is O(dt) and neglected it
in the context of the law of large numbers.
Omberg (1991) showed that the hedging error is zero over each revision
interval in two cases. The first case corresponds to the limit of the binomial
discrete option pricing model of Cox et al. (1979). The second case concerns
a stochastic revision strategy, which succeeds with a certain probability
of one-over-one revision interval. However, it remains unproved that the
strategy succeeds over a high number of revision intervals.
For other cases, the hedging error over one revision interval is of the
same order as the risk-free return and cannot be eliminated.
11.4. Forward and Backward Equations

When the asset price dynamics are described by the following Markov
diffusion process:
dSt = µ(S, t)dt + σ(S, t)dWt
the probability density function for S at time t conditional on St0 = S0 ,

denoted by f (S, t; S0 , t0 ), satisfies the partial differential equations of
motions, which are the backward and the forward Kolmogorov equations.
The backward Kolmogorov equation is given by:

1 2 ∂ 2f ∂f ∂f
σ (S0 , t0 ) + µ(S0 , t0 ) + = 0. (11.50)
2 ∂S 2 ∂S0 ∂t0
The forward Kolmogorov or Fokker–Planck equation is given by:

1 ∂2f ∂f ∂f
[σ 2 (S0 , t0 )f ] − [µ(S, t)f ] + = 0. (11.51)
2 ∂S 2 ∂S0 ∂t
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.
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
to zero at S = S1 , then S1 is a reflecting barrier.

It is convenient to note that when σ(S1 ) = 0, then the value at which σ
is zero represents a natural barrier. When µ(S1 ) is zero, we have a natural
absorbing barrier and when µ(S, t) is strictly positive (or negative), we have
a natural reflecting barrier from above (or below).
Example: When µ(S) = µ and σ(S) = σ, then S is a Brownian motion

with drift. It is possible to verify that the solution to Eqs. (11.50) and
(11.51) is:

S − x0 − µ(T − t0 )
f (S, t; x0 , t0 ) = N √ (11.53)
σ T − t0
where N [.] stands for the density of the standard normal distribution.
11.5. The Main Concepts in Bond Markets

and the General Arbitrage Principle
This section presents the main concepts in bond markets and the general
arbitrage principle.
11.5.1. The main concepts in bond pricing

The yield to maturity (YTM)
Consider a zero-coupon bond B(t, T ) at time t maturing in T years. The
present value of one-dollar received at time T is B(t, T ) = e−y(T −t) . Hence,
the YTM is given by y = − log B
T −t . The price of a coupon-paying bond
corresponds to the present value of all its cash flows N + 1: N coupons
Ci and principal payment P as follows:

N

V = P e−y(T −t) + Ci e−y(ti −t) . (11.54)
i=1
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
Macaulay duration is given by:
−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
Convexity is often computed for large movements in the yield in order to

examine the change in the price of the bond with respect to yield.
The dollar convexity is defined with respect to the bond price as:
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 .
11.5.2. Time-dependent interest rates and information

uncertainty
When the spot interest rate is only a known function of time, then B = B(t).
Consider the change in the value of a zero-coupon bond paying 1 at t = T
over a time step dt from t to (t + dt). The change in the zero-coupon bond
price can be written as dBdt
. In this case, this return must be equal to the
return from a bank deposit r(t) or dB dt
= r(t)B.
However, if this equality is set by arbitrage considerations, then
investors must suffer sunk costs to get informed about these opportunities.
In fact, it is well known in practice to find only one interest rate for each
maturity. This is more difficult to assert in an international context. There
will be also some risks related to the re-investments of the coupons and
investors must also pay about future investment opportunities. Merton
(1987) showed that information costs are specific to each market and are
paid when investors want to get informed about investment opportunities.
In this case, they require that the return on the bond must be equal to
the risk-less rate plus an “additional return” corresponding to information
costs. Hence, we have
dB
= (r(t) + λB )B.
dt
In the standard literature, the additionel return λB = 0 and we have:
dB
= r(t)B.
dt
The price of the zero coupon satisfyingRthis equation in the presence of an

“additional return” λB is B(t; T ) = e− t (r(τ )+λB )dτ .
T
In the standard literature, the additional return λB = 0 and we have:

RT
B(t; T ) = e− t
(rτ )dτ
.
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
We obtain the standard case when λB = 0. The solution to this differential

equation is given by:

RT T RT
B(t) = e− t
(r(τ )+λB )dτ
1+ C(t )e t
(r(τ )+λB )dτ
dt
t (11.59)
V (t− +
c ) = V (tc ) + Cc .
In the standard case, the solution becomes:

RT T RT
− r(τ )dτ rτ dτ
B(t) = e t 1+ C(t )e t dt (11.60)
t
11.5.3. The general arbitrage principle

Rogers (1997) considered the general principles of financial modeling in the
light of his 1997 approach to interest rate modeling. The general arbitrage
principle shows that an asset with price YT at time T , is represented at
time t < T by:

T
Yt = Et exp − rs ds YT (11.61)
t
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
on the collection Ft of events. In this context, the bond prices can be

written as:

T
P (t, T ) = Et exp − rs ds
t

T
= Ẽt ρT exp − rs ds ρt = Ẽt [ζT ] ζt
t
where ζt is a state-price density process defined by:

t
ζt = exp − rs ds ρt .
0
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”.
11.6. Discrete Hedging and Option Pricing

The Black–Scholes model is based on the assumption of continuous hedging.
This is impossible in practice since hedging is done in a discrete way.
In general, re-hedging is done regularly at times separated by a constant
interval referred to as the length of the hedging period, δt. This period can
vary from a few minutes to some weeks depending on the market activity.
It is very short for a market maker and longer for a trader.
11.6.1. Discrete hedging

It is possible to use Taylor’s series expansions to find a better hedge than
the Black–Scholes. This hedge results from the use of a certain number of
the underlying assets that minimizes the variance of the hedged portfolio.
This allows also to find an option-adjusted value.
Following the analysis in Willmott (1998), consider the discrete model
for the underlying asset,
S = ex (11.62)
with

σ2 1
δx = µ− δt + σφδt 2 (11.63)
2
where φ is a random variable drawn from a standardized normal distribution

1
and φδt 2 corresponds to the Wiener process in the continuous time version.
Now, consider the construction of a hedged portfolio by a long position
in the derivative security and a short position in ∆ units of the underlying
asset S,
Π = V − ∆S. (11.64)
If we determine the right hedge, it is relatively a simple matter to price the

option.
Using Taylor’s series expansion for the portfolio gives:
1 3
δΠ = δt 2 A1 (φ, ∆) + δtA2 (φ, ∆) + δt 2 A3 (φ, ∆)
+ δt2 A4 (φ, ∆) + · · · (11.65)
where A1 is given by:

∂V
A1 (φ, ∆) = σφS −∆ ,
∂S
and A2 is,

∂V ∂V 1 1 ∂ 2V
A2 (φ, ∆) = +S −∆ µ + σ2 (φ2 − 1) + σ2 φ2 S 2
∂t ∂S 2 2 ∂S 2
In this context, the ∆ must be chosen in such a way so as to minimize
the variance of the change in the portfolio’s value. The option is valued
by setting the expected return on the portfolio equal to the risk-less rate
plus information costs. Using Eq. (11.65), the variance of the change in the
portfolio value can be computed as:
Var[δΠ] = E[δΠ2 ] − (E[δΠ])2 (11.66)
The value of ∆ minimizing the variance is found using:

∂
Var[δΠ] = 0
∂∆
Hence, the optimal ∆ is:
∂V
∆= + δt(. . .) (11.67)
∂S
The first term is the Black–Scholes hedge ratio in a context of continuous
trading. The second term indicates a reduction in risk.
The leading-order random term in the portfolio is given by

1 2 2 2 ∂2V 2 2
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 ).
11.6.2. Pricing the option

The determination of the right hedge ratio allows the computation of the
option price. Since the expected return on the discretely hedged portfolio
must be equal to the risk-less rate plus the information costs, we have:

1
E[δΠ] = (r + λV )δt + (r + λV )2 δt2 + · · · V
2

1
− (r + λS )δt + (r + λS )2 δt2 + · · · ∆S. (11.68)
2
If we substitute Eqs. (11.64) and (11.65) in Eq. (11.68), one obtains:
∂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
and the “better” delta is given by:
∂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
Eq. (11.70) correspond to a constant times the squared value of S. Hence,

the corrected option price needs the adjustment of the volatility and using
the value σ ∗ with:

δt
σ ∗ = σ 1 + 2 (µ − r)(r − µ − σ2 ) .
2σ
11.6.3. The real distribution of returns

and the hedging error
Several studies have shown that the empirical distribution of the underlying
assets have a higher peak and fatter tails than the normal distribution.
Hence, how this distribution affects the hedging arguments and the hedging
errors?
Consider the following return for the underlying asset at time δt δS
S = ψ,
where the distribution of the random variable ψ is determined empirically.
Now, following the analysis in Willmott (1998), we study the change in
the value of the hedged portfolio under the assumption that the option
component obeys the Black–Scholes equation with an implied volatility
of σi . The random return in excess of the risk-free rate for the hedged
portfolio is:

∂V
δΠ − rΠδt = S ∆ − (rδt − ψ)
∂S
1 ∂ 2V 2
+ S2 (ψ − σi2 δt) + · · · (11.71)
2 ∂S 2
If a delta hedge is implemented as in Black–Scholes, this gives:
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
methods in continuous time finance. Each concept is proposed in two

forms: an intuitive version and a more rigorous mathematical version.
The Wiener and Ito processes are used to model the dynamics of asset
prices. Ito’s lemma is proposed to differentiate a function of one and
several stochastic variables. It is illustrated through several examples in
different contexts. The Kolmogorov forward and backward equations are
presented as well as the first passage time and the maximum (minimum)
of a diffusion process. These tools allow the pricing of standard options
and more complex derivative assets. The principal of arbitrage simply
states that financial assets having identical characteristics must trade at
the same price. If this principle is not respected, then selling the high-
priced asset and buying the low-priced asset allows a risk-free profit. This
principle is used to determine the fair price of a security or a derivative
asset. The analysis by Omberg (1991) reveals that the hedging error
is zero over each revision interval in the following two cases. The first
case represents the limit of the classical binomial discrete option pricing
model. The second case corresponds to a stochastic revision strategy, which
succeeds with a certain probability of one-over-one revision interval. For
other cases, the hedging error over one revision interval is of the same
order as the risk-free return and cannot be eliminated. This chapter is
necessary for the understanding of the basic techniques behind the theory
of rational option pricing in a continuous time framework. However, it is
not necessary for the use and applications of all the formulas presented in
this book.
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?
Appendix A: Introduction to Diffusion Processes

The notion of a stochastic process can be introduced with respect to the
notion of vector of variables. Using the notations:
Ω: the space of all possible states ω;

Ft : σ-algebra defined over Ω; a class of partitions of Ω.
X(ω) is said to be a random variable when it is a measurable application

from (Ω, Ft ) to R.
A vector of random variables X(ω) = [X1 (ω), . . . , Xn (ω)] is a measur-
able application from (Ω, Ft ) into Rn .
The notion of a vector of random variables is similar to that of n
ordinary variables defined on the same probability space.
A stochastic process is the extension of the notion of a vector of
variables when the number of elements becomes infinite. It is a family
of random variables, Xt (ω), t ∈ T when the index varies in a finite or
an infinite group. When ω = ω0 , X(ω0 , t) is a function of t called a
realization or a path of the process. When t = t0 , X(ω, t0 ) is a vector of
variables.
A stochastic process will be denoted by X(t).
Definition: A continuous time stochastic process having its values in a

space E with a tribe, Ft is a family of random variables, (Xt )t≥0 is defined
on the probability space (Ω, Ft ; P ) taking its values in (E, Ft ).
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.
Definition: A filtration (Ft )t≥0 is an increasing family of sub-tribes of Ft

in the probability space (Ω, Ft , P ).
An example of stochastic processes often used in continuous time

finance when deriving models for the valuation of financial assets is the
Brownian motion.
Appendix B: The Conditional Expectation

Consider a probability space (Ω, Ft ; P ) and let B be is a sub-tribe of F .
The following theorem allows the definition of the conditional expectation.
Theorem: For any integrable random variable X, there is a unique

associated random variable Y such that for all element in B,
E(X1(B) ) = E(Y 1(B) )
Y is known as the conditional expectation of X given B, or E(X/B).

The conditional expectation operator obeys the following properties:
If X is B-mesurable, then E(X | B) = X.
E(E(X | B)) = X.
For any random variable Z, measurable with respect to B, E(ZE(X | B)) =

E(ZX).
E(αX + µY | B) = αE(X | B) + µE(Y | B).
If X > 0 then E(X | B) ≥ 0.

If C is a sub-tribe of B, then E(E(X | B) | C) = E(X | C).
If X is independant of B, then E(X | B) = E(X).
Appendix C: Taylor Series

If a function f has derivatives in the region (x, x+h), then the development
of this function around x gives:
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!
where x∗ is in the region [x, x + h].

A function of two variables x and y is represented in Taylor series

expansions by:

∂f (x, y)
f (x + h, y + k) = f (x, y) + h
∂x

∂f (x, y) 1 ∂ 2 f (x, y)
+ k+ h2
∂y 2 ∂x2
2
1 ∂ 2 f (x, y) 2 ∂ f (x, y)
+ k + hk
2 ∂y 2 ∂x∂y
n
1 1 ∂ 2 f (x, y) 1 ∂ 2 f (x, y)
+···+ h k .
n! 2 ∂x 2 ∂y
Similar expressions can be derived for functions of three or more variables.
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
or
etX(t) = W (t) − 1 − tα
hence:
dW = t(W (t) − 1 − tα )dX + (αtα−1 + X(t)(W (t) − 1 − tα ))dt

1
+ t2 (W (t) − 1 − tα )dt
2
and finally, we have u(W, t) = t(W (t) − 1 − tα ),

1
v(W, t) = αtα−1 + X(t) + t2 (W (t) − 1 − tα ) .
2
For the third function, W (t) = f (t)β X α (t), we have
∂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)
Exercise: Consider an underlying asset S whose price follows a log-normal

random walk.
1. Apply Ito’s lemma for the following functions f (S) and g(S):
f (S) = AeS + B, g(S) = eS

n
where A and B are constants.

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
(nS n )2 + n(n − 1)S n = nS n (nS n + n − 1)
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.
1895–1927.
theory and tests. The Financial Review, 30(3) (August), 617–635.
presence of a finite mixture of Gaussian distributions. International Journal
of Finance, 13(3), 1975–2000.
Finance, 13(2), 1826–1837.
international finance. International Journal of Finance, 13(3), 1958–1973.
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.
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.
Rogers, C (1997). One for all. Risk, 10(3) (March) 56–59.
Willmott, P (1998). Derivatives. New York: John Wiley and Sons.

Chapter 12
RISK MANAGEMENT: APPLICATIONS TO THE

PRICING OF ASSETS AND DERIVATIVES IN
COMPLETE MARKETS
Chapter Outline
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
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.
12.1. Characterization of Complete Markets

In financial markets, there are two classes of financial assets: securities and
derivative assets. Securities correspond fundamentaly to common stocks
and bonds.
Derivative assets are contingent claims characterized by their interme-

diate and final payoffs. There are several definitions of a complete market.
The idea of complete markets is proposed first by Arrow (1953) and
Applications to the Pricing of Assets and Derivatives in Complete Markets 537
Debreu (1954). They defined a complete market with respect to state

securities or state contingent claims.
A state security or an Arrow-Debreu security is a security which

pays off one dollar if and only if a given state of the nature occurs. A state
of nature or a possible state in the economy is said to be an insurable state
when it is possible to construct a portfolio of assets which has a non-zero
return in that state. In this economy, the price of each traded asset at the
begining of the period is ps . For an economy where every state is insurable,
a price vector can be completely determined with unique state prices. This
implies the absence of arbitrage opportunities.
A contingent claim is attainable, if there is a strategy that gives at the

derivative asset’s maturity date the same value as the contingent claim
terminal payoff. Hence, a complete market can be defined as a market
in which all the contingent claims are attainable, i.e., all the contingent
claims can be reached or obtained when implementing a replication or a
duplication strategy.
A complete market can be defined with respect to the concept of a viable

financial market. A viable financial market is a market where there is no
profitable risk-less arbitrage. The absence of risk-less arbitrage or arbitrage
opportunities means that a strategy which is implemented at the initial time
with a zero-cost must have a nil final or terminal value. It is important
to note that there is a relationship between the notion of arbitrage and
the martingale property of security prices. The martingale property for
security prices means simply that the best estimation of the future price
is the last information. Hence, when historical data of security prices are
used to predict the future price, only the most recent information matters,
i.e., the last price. This concept defines also that of an efficient market.
In mathematical terms, a financial market is viable if and only if there is
a probability P ∗ which is equivalent to a probability P , under which the
discounted asset prices have the martingale property. This is done under
the theorems of the change in probability and in particular, the Girsanov
theorem. (See Appendix A for more detail).
A viable financial market is complete, if and only if there is a probability

P ∗ equivalent to the probability P under which the martingale property is
satisfied by security prices. It can be shown that this probability P ∗ is
unique.
If one considers a contingent claim specified by its final payoff h, which

is of the form h = (ST − K)+ for a European call and h = (K − ST )+
for a European put, then an attainable strategy Φ simulates or duplicates
the option price when its payoff at the maturity date T is equal to h, or
VT (Φ) = h. The sequence of the random derivative asset prices between the
initial time 0 and the option’s maturity date T is a martingale under the
unique probability P ∗ . Hence, we have: V0 (Φ) = E ∗ [VT (Φ)] and V0 (Φ) =
E ∗ [h | ST ]. This gives the general result in complete markets:
Vt (Φ) = St E ∗ [h | ST | Ft ] for t : 0, . . . , T.
If at the initial time a derivative asset is sold at its expected price,

E ∗ [h | ST ], then an investor following a replication strategy can obtain the
exact payoff h at time T . He is said to follow a full hedging strategy.
12.2. Pricing Derivative Assets: The Case

of Stock Options
Since each financial asset is specified by its intermediate and terminal
payoffs, option pricing consists in finding the fair price at the initial time
when the derivative asset is bought or sold. There is a unique approach for
the pricing of derivative assets.
12.2.1. The problem

The value of each option is given by its expected terminal and intermediate
payoffs discounted to the present. However, there are two methods for the
pricing of options. The first was initiated by Black and Scholes (1973) and
Merton (1973). The second is the martingale approach due to Harrison and
Kreps (1979) and Harrison and Pliska (1981). The first method, known as
the Black–Scholes method, is based on the resolution of the following PDE:

1 2 2 ∂2c ∂c ∂c
σ S 2 + rS + − rc = 0
2 ∂ S ∂S ∂t
under the appropriate boundary conditions. The second is based on the
use of martingale techniques. In either the first or the second approach, the
boundary conditions are the same. These conditions refer to the appropriate
payoff conditions corresponding to the value of European and American
contingent claims. For a European call, the final payoff is given by c(S, T ) =
max[0, ST − K] where K is the option strike price. For a European put, the
final payoff is given by p(S, T ) = max[0, K − ST ] where K is the option

strike price. For an American call, the additional following condition must
be satisfied by the PDE C(S, t) ≥ max[0, St − K] where K is the option
strike price. The difference from the condition for a European call is that the
American call holder can exercise his option at each instant. This condition
indicates that at each instant, the American call value must be at least equal
to or greater than the intrinsic value, which corresponds to the value of a
European call at maturity. For an American put, the additional following
condition must be satisfied since the put holder can exercise his option
at each instant P (S, t) ≥ max[0, K − ST ] where K is the option strike
price. This condition shows that at each instant, the American put value
must be at least equal to the intrinsic value. In the presence of dividend
distributions, some other conditions must be imposed. All these conditions
apply in the absence of dividends. When there are distridutions to the
underlying asset, there is in general no explicit solutions to these problems
and numerical methods are often used. First, we illustrate the PDE method
for the valuation of European call options. Second, we illustrate the use of
the martingale approach for the pricing of European calls. The reader can
verify that the price of the call is the same under both methods. Third,
since it is difficult to get closed form solutions for American options with and
without distributions to the underlying asset, financial economists often use
numerical methods. Therefore, we develop the principles of these techniques
in the last section of this chapter.
12.2.2. The PDE method

The standard analysis
Consider the search for the solution of the following PDE:

1 ∂2c 2 2 ∂c ∂c
σ S + rS + − rc = 0
2 ∂2S ∂S ∂t
for a European call with the payoff c(S, T ) = max[0, ST − K] where K is

the option strike price. Using the appropriate change of variables, it can be
shown that the call price is:
c = SN (d1 ) − Ke−r(T −t) N (d2 )

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
where
d “ 2
”
1 − x2
N (d) = √ e dx
2Π −∞
A detailed resolution of this system is given in Appendix B. The payoff

of a European put option is p(S, T ) = max[0, K −ST ]. Using the appropriate
change of variables, it can be shown that the put price is:
p = −SN (−d1 ) + Ke−r(T −t) N (−d2 )

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
The Analysis in the presence of information costs

Consider the search for the solution of the following PDE:

1 ∂ 2c ∂c ∂c
σ 2 S 2 + (r + λS )S + − (r + λc )c = 0
2 ∂ 2S ∂S ∂t
for a European call with the payoff c(S, T ) = max[0, ST − K] where K is

the option strike price.
Using the appropriate change of variables, it can be shown that the call
price under information uncertainty is given by:
c = Se−(λS −λc )(T −t) N (d1 ) − Ke−(r+λc )(T −t) N (d2 )

S
ln K + r + λS + 12 σ 2 (T − t)
d1 = √ ,
σ T −t
S
ln K + r + λS − 12 σ 2 (T − t)
d2 = √
σ T −t
where:
d 2
1 x
N (d) = √ exp − dx
2Π −∞ 2
The payoff of a European put option is p(S, T ) = max[0, K − ST ]. Using

the appropriate change of variables, it can be shown that the put price is
given by:
p = Se−(λS −λc )(T −t) N (−d1 ) + Ke−(r+λc )(T −t) N (−d2 )
This model appears in Bellalah (2001).
12.2.3. The martingale method

The Black and Scholes (1973) pricing of options requires first the knowledge
of the probability under which the asset price St is a martingale.
The standard analysis

In the context of the Black–Scholes model, there is an equivalent probability
to P under which the discounted expected stock price, St∗ = e−rt St is a
martingale. In fact, if one uses the stochastic differential equation for the
stock price, we have:
dSt∗ = −re−rt St dt + e−rt dSt
or
dSt∗ = St∗ [(µ − r)dt + σdWt ].

If we put the change in variables Wt∗ = Wt + µ−r σ t then dSt∗ =
∗ ∗ (µ−r)
St σdWt . Using the Girsanov theorem, Θt = σ , there is a probability
P ∗ equivalent to P under which (Wt∗ )0≤t≤T is a standard Brownian motion.
Hence, under P ∗ , we deduce from Eq. (12.12) that St∗ is a martingale and
∗ 1 2
St∗ = S0∗ eσWt − 2 σ t
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 .
∗ 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
Since the random stock price St is Ft measurable and (WT∗ − Wt∗ ) is

independent of Ft under P ∗ , it can be shown that ct = H(t, St ) with:
∗ ∗ σ2
H(t, S) = E ∗ e−r(T −t) f Ser(T −t) eσ(WT −Wt )− 2 (T −t) (12.1)
Underthe probability P ∗ , the quantity√(WT∗ − Wt∗ ) follows a Gaussian

law, N (0, (T − t). When (WT∗ − Wt∗ ) = Y T − t and Y follows a N (0, 1),
then
∞
σ2
√ 1 −y2
H(t, S) = e−r(T −t) f Se(r− 2 )(T −t)+σy T −t √ e 2 dy
−∞ 2Π
When Y follows N (0, 1), under P ∗ , then
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.
Equation (12.2) is equivalent to Eq. (12.1) where f (z) is replaced by

(zT − K)+ . Now, using the notation,
S
ln K + r + 12 σ 2 τ √
d1 = √ , d2 = d1 − σ τ
σ τ
we consider again the equation

σ2 ∗ ∗ +
H(t, S) = E Se(r− 2 )(T −t)+σ(WT −Wt ) − K
Using the following lemma
E[(Z − K)+ ] = E[[(Z − K)]IZ≥K ]
with

1, if Z ≥ K
IZ≥K =
0 else
the condition (Z ≥ K) is equivalent to:

√
τ − 12 σ2 τ )
Se(σY ≥K
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.
Using this remark, H(t, S) can be written as:

√ σ2
H(t, S) = E Se σY τ − 2 τ − Ke−rτ IY +d2 ≥0
or
∞ √ 2 1 1 2
H(t, S) = Se(σY τ − σ2 τ )
− Ke−rt √ e(− 2 y ) dy
−d2 2Π

d2 (−σY √τ − σ2 τ ) 1 1 2
H(t, S) = Se 2 − Ke−rt √ e(− 2 y ) dy
−∞ 2Π
This integral can be written as the difference between two integrals

d2 √ 2 1 1 2
H(t, S) = Se(−σY τ − σ2 τ )
√ e(− 2 y ) dy
−∞ 2Π
d2
1 1 2
− Ke−rτ √ e(− 2 y ) dy
−∞ 2Π
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:
H(t, S) = SN (d1 ) − Ke−rτ N (d2 ) (12.3)
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
The analysis in the presence of information costs

It is possible to use the martingale method for the pricing of derivative
claims in a Black–Scholes context in the presence of information uncer-
tainty. For the analysis of information costs and valuation, we can refer to
Bellalah (2001), Bellalah et al. (2001a,b), Bellalah and Prigent (2001) and
Bellalah and Selmi (2001) etc. In this context, we can show that there is
an equivalent probability to P under which the discounted expected value
of the underlying asset given by St∗ = e−(r+λS )t St is a martingale. In this
expression, the term λi refers to information costs.
Consider the stochastic differential equation for the underlying asset:
dSt∗ = −(r + λS )e−(r+λS )t St dt + e−(r+λS )t dSt
or the following dynamics:
dSt∗ = St∗ [(µ − r − λS )dt + σdWt ]
If the following change in variables is used

(µ − r − λS )
Wt∗ = Wt + t
σ
then
dSt∗ = St∗ σdWt∗ .
Using the Girsanov theorem,
(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
The option price in the Black–Scholes economy can be computed using

its discounted expected terminal value under the appropriate probability
P ∗ as
ct = E ∗ [e−(r+λc )(T −t) h/Ft ] with h(ST − K)+ = f (ST )
The option price at time t for a maturity date is given by:

ct = E ∗ e−(r+λc )(T −t) f St e−(r+λS )(T −t)

∗ ∗ 1 2
× exp σ(WT − Wt ) − σ (T − t) Ft
2
∗ 1 2
Since for all t, we have: St = S0 e[σWt − 2 σ t] and the
value of the underlying
asset at the option’s maturity date is ST = S0 exp σWT∗ − 12 σ 2 T , the ratio
of the stock price between two dates is given by:

ST ∗ ∗ 1 2
= exp σ(WT − Wt ) − σ (T − t)
St 2
It can be shown that the option value is given by ct = H(t, St ) with:

∗ ∗ σ2
H(t, S) = E ∗ e−(r+λc )(T −t) f Se(r+λS )(T −t) eσ(WT −Wt )− 2 (T −t) (12.4)
Under√the probability P ∗ , the quantity WT∗ −√Wt∗ follows a Gaussian

law, N (0, t). When the difference WT∗ − Wt∗ = Y T − t and Y follows an
N (0, 1), then
∞
σ2
√ 1 −y2
H(t, S) = e−(r+λc )(T −t) f Se(r+λS − 2 )(T −t)+σy T −t √ e 2 dy
−∞ 2Π
When Y follows N (0, 1), under P ∗ , then
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
12.3. Pricing Derivative Assets: The Case of Bond Options

and Interest Rate Options
The value of a zero-coupon bond at maturity is B(T, T ) = 1.
12.3.1. Arbitrage-free family of bond prices

The concept of arbitrage is central to the valuation of financial assets.
Musiela (1997) gives the following definition for the arbitrage family of
bond prices.
Definition: Consider a process r defined on a filtered probability space

(Ω, F, P ). An arbitrage-free family of bond prices relative to the interest rate
r when B(T, T ) = 1 for every T in [0, T ∗ ], and when there is a probability
P ∗ on (Ω, F) equivalent to P such that the relative bond price
Z ∗ (t, T ) = B(t, T )/Bt , ∀ t ∈ [0, T ]
is a martingale under P ∗ . The probability P ∗ is a martingale measure for

the family B(t, T ). Hence, for any P ∗ of an arbitrage-free family of bond
prices, we have:
RT
B(t, T ) = EP ∗ e− t ru du | Ft , ∀ t ∈ [0, T ] (12.5)
The reader can refer to Appendix E for more details.
12.3.2. Time-homogeneous models

The Vasicek (1977) model
The diffusion process proposed in the Vasicek (1977) model is a mean-
reverting version of the Ornstein–Uhlenbeck process. In this model, the
short-term interest rate follows the following dynamics:
drt = (a − brt )dt + σdWt∗
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)
The terminal boundary condition is:
v(r, T ) = f (r) (12.7)
When h = 0 and f (r) = 1, the price of a zero-coupon bond is:
B(t, T ) = v(rt , t, T ) = em(t,T )−n(t,T )rt , (12.8)

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
It is possible to verify this result using Eq. (12.8) with m(T, T ) =

n(T, T ) = 0. Using the PDE and seperating the terms in r gives the
following system:
nt (t, T ) = bn(t, T ) − 1, n(T, T ) = 0,

1 (12.9)
mt (t, T ) = an(t, T ) − σ2 n2 (t, T ), m(T, T ) = 0
2
which leads to the above expressions. We can check that we have:
dB(t, T ) = B(t, T )(rt dt + σn(t, T )dWt∗ )
where the bond price volatility equals b(t, T ) = σn(t, T ).

Using the expression (12.9) for the bond price, the bond’s yield is:
n(t, T )rt − m(t, T )

Y (t, T ) = ,
T −t
Since this yield is an affine function and therefore the categories of similar
models are known as affine models of the term structure.
Jamshidian (for more details, refer to Bellalah et al., 1998) gives a
closed-form solution for a European call on a bond (with and without
coupons) with a H-maturity in the context of this model using:

Ct = B(t, T )EQ (ξη − K)+ | Ft (12.10)
B(t,H)
where η = B(t,T )
;
K: the strike price;
Q: a probability measure equivalent to P ∗ and
ξ: a random variable with a variance:
T
2 2
vH (t, T ) = |b(t, T ) − b(t, H)| du
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:
Ct = B(t, H)N (d1 (t, T )) − KB(t, T )N (d2 (t, T ))
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.
The Cox, Ingersoll and Ross (CIR) (1985) model

In their general equilibrium approach, CIR (1985) use the familiar square
root process for the dynamics of interest rates:
√
drt = (a − brt )dt + σ rt dWt∗
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
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:
Y (t, t + τ ) = Ỹ (rt , τ ) = m̃(τ ) + ñ(τ )rt ,
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.
The Longstaff model

Longstaff (1989) uses the following dynamics for the short-term rate:
√ √
drt = a(b − c rt )dt + σ rt dWt∗ (12.11)
referred to as a double square root process. In this model, the price of a
zero-coupon bond is given by:
√
B(t, T ) = v(rt , t, T ) = em(t,T )−n(t,T )rt −p(t,T ) rt
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.
12.3.3. Time-inhomogeneous models

It is important to note that the Vasicek (1977) and the CIR (1985) models
are special cases of the following process:
drt = a(b − crt )dt + σrtβ dWt∗
where β is a constant between zero and one.

Hull and White use the following dynamics for the short-term interest
rate:
drt = (a(t) − b(t)rt )dt + σ(t)rtβ dWt∗
where β is a positive constant, W ∗ a Brownian motion and a, b, and σ

are locally bounded functions. This model appears as a generalization of
the Vasicek (1977) and the CIR (1985) model. When β = 0, this gives a
generalized Vasicek model:
drt = (a(t) − b(t)rt )dt + σ(t)dWt∗ .
When β = 0.5, this gives a generalized CIR model:

√
drt = (a(t) − b(t)rt )dt + σ(t) rt dWt∗
12.4. Asset Pricing in Complete Markets: Changing

Numeraire and Time
This section introduces two mathematical instruments: the change of
numeraire and the change of time. These tools are very effective in solving
derivative asset pricing problems.
12.4.1. Assumptions and the valuation context

Consider an economy in which transactions take place instantaneously
without transactions costs and taxes. The interest rate follows a Gaussian
process. The short-term risk-less interest rate r(t) under the risk-neutral
probability Q is governed by the following stochactic differential equation
dr(t) = a(t)[b(t) − r(t)]dt + σ(t)dW1 (t)
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 )
where P (t, T ) is the price of a zero-coupon bond at time t for a maturity

date T .
The volatility in the process describing the dynamics of zero-coupon
bond prices is given by:
T u
σp (t, T ) = σ(t) exp − a(s)ds du
t t
Consider the value at time t of a portfolio which corresponds to an

investment of one dollar at time t = 0 at the rate r(t):
Rt
r(u)du
B(t) = e 0
This portfolio corresponds to a capitalization factor. Consider the

following dynamics for a risky asset St :
dSt
= r(t)dt + σS ρdW1 (t) + 1 − ρ2 dW2 (t)
St
under the risk-neutral probability Q. The coefficient ρ ∈ [−1, 1] introduces

a correlation between the risky asset and the interest rate. The absence of
arbitrage opportunities gives 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 probabil-

ity Q.
12.4.2. Valuation of derivatives in a standard

Black–Scholes–Merton economy
Consider the dynamics of the option’s underlying stock
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

∂C 1 2 2 ∂2C ∂C
dV = + σS S 2
dt + dS (12.14)
∂t 2 ∂S ∂S
The solution for a European call is:
C(S0 , T, σS , r) = S0 N (d1 ) − Ke−rt N (d2 )

σ2

ln SK0 + r + 2S T √
d1 = √ = d2 + σS T
σS T
Now, consider the following change of variables in Eq. (12.12)
ft = St er(T −t) , Cf = Cer(T −t)
The two changes of variables correspond to a change of numeraire giving

respectively the value of a forward contract as a function of its underlying
asset and the cost of carry and the value of a forward call as a function of
a spot option.
Consider also the following time change in Eq. (12.12) τ = σS2 t and
τ = σS2 T .
∗
The change of underlying asset f in Eq. (12.12) allows to write this

equation as:
∂C 1 ∂ 2C
+ σS2 f 2 2 − rC = 0 (12.15)
∂t 2 ∂f
Equation (12.15) corresponds to the Black (1976) equation providing the

value of an option on a forward contract f . A simple comparison between
Eqs. (12.15) and (12.12) reveals that the term rS ∂C
∂S
has disappeared. This
is because the forward positions cost nothing to initiate and to adjust.
Therefore, the dynamics of the replicating portfolio can be written as:
∂C
dV = rCdt + df
∂f
Using Ito’s lemma

∂C 1 2 2 ∂ 2C ∂C
dV = + σA f dt + df
∂t 2 ∂f 2 ∂f
gives the dynamics of C in Eq. (12.15).

When Cf = Cer(T −t) , this change of variables gives:

∂Cf 1 2 2 ∂ 2 Cf
+ σA f =0
∂t 2 ∂f 2
In the same way, no initial investment is required for the call on a
forward contract, since it represents a forward option. In this case, the
∂C
dynamics of the replicating portfolio are given by dVf = ∂ff df and Ito’s
lemma gives:

∂Cf 1 2 2 ∂ 2 Cf ∂Cf
dVf = + σS f 2
dt + df
∂t 2 ∂f ∂f
It is possible to use a change of variables for τ and τ ∗ to simplify the
equation to the following form:
∂Cf 1 ∂ 2 Cf
+ f2 =0 (12.16)
∂τ 2 ∂f 2
Equation (12.16) is useful for the valuation of a forward option on a forward
contract maturing in τ ∗ for which the volatility is equal to one. In this
case, the solution to Eq. (12.16) for a forward call on a forward contract is
given by:
Cf (f0 , τ ∗ ) = C(f0 , τ ∗ , 1, 0) = f0 N (d1 ) − KN (d2 )
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.
12.4.3. Changing numeraire and time: The martingale

approach and the PDE approach
The change of numeraire and the change of time appear in several papers
including Merton (1973), Jamshidian (for details, refer to Bellalah et al.,
1998), El Karoui and Geman (1993) etc. In a martingale approach, the
change of numeraire and time is useful from a computational point of view.
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:
dQX X(T ) B(0)

=
dQ X(0) B(T )
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 )
As an example, consider the price of a risk-less zero-coupon bond

P (t, T ) as a numeraire in an economy with stochastic interest rates. Using
this numeraire, it is possible to define for its process a new probability QP
by giving its Radon–Nikodym derivative with respect to Q as:
dQP P (T, T ) B(0)

=
dQ P (0, T ) B(T )
which is also equal to dQ

P
1 1
dQ
= P (0,T ) B(T )
since by definition, the prices
P (T, T ) and B(0) are equal to 1. Using the two previous equations, it is
evident that:

h(T )
EQ = P (0, T )EQX [h(T )]
B(T )
The effect of this change of numeraire can be appreciated using the

Black–Scholes context for the pricing of a European call on an asset S in
the presence of a stochastic interest rate. We use the following dynamics
for the underlying asset:
dSt
= rdt + σS dW2 (t)
St
and for the short-term interest rate:
dr(t) = a(t)[b(t) − r(t)]dt + σ(t)dW1 (t)
In the presence of stochastic interest rates, the option price can be com-
puted as:

ST − K
CS = EQ IST ≥K
B(T )
The indicator function IE for E is the real-valued random variable

defined by IE (ω) = 1 for all ω ∈ E and zero otherwise.
Equivalently, under the probability QP , the value of the call can be
written as:

ST K
CS = P (0, T )EQP − I ST ≥K (12.17)
P (T, T ) P (T, T ) P (T ,T )
where the zero-coupon bond value is by definition equal to one at maturity.

Equation (12.17) allows us to avoid the stochastic character of interest
rates. A comparison can be done with respect to the PDE approach. In
the presence of stochastic interest rates, the call price is a function of the
underlying asset S, the interest rate r and time t. It is also possible to
consider the option price as a function of the prices of the underlying asset,
the price of zero-coupon bonds, P and time. In this case, it is possible
to show that the European call price must satisfy the following partial
∂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
This equation must be solved under the terminal condition CS =

max[S − K, 0]. Since the call price CA is first-degree homogeneous in the
underlying asset and the discount bond, or
∂CS ∂CS
CS = S +P
∂S ∂P
Equation (12.18) can be written, after simplification as:
∂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
where the instantaneous variance of the forward contract in this stochastic

interest rate economy is
σF (t, T )2 = σS2 + σP (t, T )2 + 2ρσS σP (t, T )
When the interest rate is constant, the correlation coefficient is zero,

and the bond’s price volatility is also zero. In this case, the volatility of
the forward contract is equivalent to that of the underlying spot asset
or σF (t, T ) = σS . The option payoff corresponds to that of a European
option on a forward contract CF = max[F − K, 0]. As it is shown,
the change of numeraire P under the martingale approach amounts to a
change of variables in the partial differential equation method. The main
difference between this economy and the Black–Scholes economy appears
in the presence of stochastic interest rates. The intuition of the change in
numeraire appears in Merton’s (1973) paper for the pricing of a stock option
in the presence of stochastic interest rates. Merton (1973) uses a change
of variables KPS(t,T
t
) . The new probability Q
P
corresponds to a “forward-
neutral” probability. It is possible to simplify the analysis by considering
the underlying asset itself as a numeraire. In this case, a new probability
measure QS can be defined such that:
dQS ST P (0, T )
=
dQP S0 P (T, T )
where, as before, the value of P (T, T ) is equal to one by definition. When

this change of probability is used in the right-hand side of Eq. (12.17), the
European call price is given by:

CS = S0 E Q I P (T ,T ) ≤K − KP (0, T )E Q I ST ≥K
S P
ST P (T ,T )
The change of time

The main idea behind the change of numeraire in the partial differential
equation approach is to operate an appropriate change of variables that
facilitates the search of a solution to the price of a derivative contract in
the standard Merton (1973) approach. The main idea behind the change of
numeraire in the martingale approach is to express under an appropriate
probability measure, the relative price of a security as a martingale.
Consider a continuous P -martingale defined by the following stochastic
dMt = x1 (t)dW1 (t) + x2 (t)dW2 (t)
where the coefficients x1 and x2 are deterministic functions.

In this context, it is possible to show the existence of a unique one-
dimentional standard Brownian motion Y such that:
2
t
Mt = xi (u)dWi (u) = YMt (12.21)
i=1 0
where
2
t
M t = x2i (u)du.
i=1 0
Consider again the pricing of a European call in the presence of

stochastic interest rates. Using two changes of numeraire, the call price is:

CS = S0 E Q I P (T ,T ) ≤K − KP (0, T )E Q I ST ≥K
S P
(12.22)
ST P (T ,T )
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
Now, it is possible to introduce the time change τ defined as

t t
2 2
τ (t) = [σS + σp (u, T ) + 2ρσp (u, T )σS ]du = σF (u, T )2 du
0 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
The computation of this equation is straightforward. Hence, the European

option price is:
   
τ∗ τ∗
ln KPS(0,T
0
) + 2
ln S0
KP (0,T ) − 2
CS = S0 N  √  − KP (0, T )N  √ 
τ∗ τ∗
(12.23)
We can compare this martingale approach with the PDE approach.

Consider the following change of variable in Eq. (12.20)
t
τ (t) = σF (u, T )2
0
In this case, the forward option price CF as a function of F and τ satisfies

the equation:
∂CF 1 ∂ 2 CF
+ F2 =0
∂τ 2 ∂F 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 )
where C is the extended Black–Scholes equation defined by:
C(S0 , T, σS , r) = S0 N (d1 ) − Ke−rtN (d2 )

σ2
ln SK0 + r + 2S T √
d1 = √ = d2 + σS T .
σS 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.
12.5. Valuation in an Extended Black and Scholes Economy

in the Presence of Information Costs
Consider the dynamics of the option’s underlying stock
dSt
= (r + λS )dt + σS dW2 (t)
St
where λS corresponds to the shadow cost of incomplete information on

asset S. We have shown by arbitrage arguments that the extended Black–
Scholes equation in the presence of information costs is given by:
∂C 1 ∂2C ∂C
+ σS2 S 2 2 + (r + λS )S − (r + λc )C = 0 (12.24)
∂t 2 ∂S ∂S
The dynamics of this portfolio are given by:

∂C ∂C
dV = C(r + λc )dt − S (r + λc )dt + dS (12.25)
∂S ∂S

∂C 1 ∂2C ∂C
dV = + σS2 S 2 2 dt + dS (12.26)
∂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:
C(S0 , T, σS , r) = S0 e−(λS −λc )t N (d1 ) − Ke−(r+λc )t N (d2 )

σ2

ln SK0 + r + λS + 2S T √
d1 = √ = d2 + σS T
σS T
Now, consider the following change of variables:
ft = St e(r+λS )(T −t) , Cf = Ce(r+λc )(T −t)
Consider also the time change in Eq. (12.24)
τ = σS2 t, τ ∗ = σS2 T.
The change of underlying asset f in Eq. (12.24) gives:
∂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
Using Ito’s lemma

∂C 1 2 2 ∂ 2C ∂C
dV = + σA f + df
∂t 2 ∂f 2 ∂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
∂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
It is possible to use a change of variables for τ and τ ∗ to simplify Eq. (12.28)

to the following form:
∂Cf 1 ∂ 2 Cf
+ f2 = 0. (12.29)
∂τ 2 ∂f 2
This last equation is useful for the valuation of a forward option on a

forward contract maturing in τ ∗ for which the volatility is equal to one. In
this case, the solution for a forward call on a forward contract is given by:
Cf (f0 , τ ∗ ) = C(f0 , τ ∗ , 1, 0) = f0 N (d1 ) − KN (d2 ).
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 ∗ .
Changing numeraire and time

We know that:

h(T )
EQ = P (0, T )EQX [h(T )]
B(T )
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
It is possible to show that the European call price in the presence of

information costs must satisfy the following partial differential equation:
∂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
+ (r + λS )S + (r + λP )P − (r + λc )CS = 0. (12.30)
∂S ∂P
This equation must be solved under the following terminal condition
CS = max[S − K, 0]
Equation (12.30) can be written, after simplification as Eqs. (12.19) and

(12.20). When the change of probability is used as before, the European
call price is given by:

CS = S0 E Q I P (T ,T ) ≤K − KP (0, T ) E Q I ST ≥K
S P
ST P (T ,T )
The change of time

Consider a continuous P -martingale and the Eqs. (12.21) to (12.23). Using
the same procedure, the solution for a call as given by the martingale
approach can also be stated as the solution to the standard Black–Scholes–
Merton (1973)

S0
CS = P (0, T )CF (F0 ) = P (0, T )C , τ ∗ , 1, 0
P (0, T )
where C is the extended Black–Scholes equation
C(S0 , T, σS , r) = S0 e−(λS −λc )t N (d1 ) − Ke−(r+λc )t N (d2 )

σ2
ln SK0 + (r + λS + 2S )T √
d1 = √ = d2 + σS T
σS 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.
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 partial differential equation method.
Second, we illustrate in detail the martingale method for the derivation
of a European call formula.
Third, we develop the foundations of some interest rate models. Fourth,
we present two mathematical tools, the change of numeraire and time
change, in order to facilitate the pricing of options. These two transfor-
mations have an economic significance and can simplify considerably some
valuation problems. While this chapter is necessary for the understanding
of the basic techniques of option pricing theory, it is not necessary for the
use and applications of all the formulas presented in this book.
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.
Appendix A: The Change in Probability and the

Girsanov Theorem
The equivalent probability
Consider a certain probability space (Ω, F, P) in the presence of a
probability P , and a probability Q which is continuous with respect to P .
The following theorem gives the equivalent probability.
Theorem: A probability Q is strictly continuous with respect to probability
P, if and only if there exists a random variable Z taking on positive values
on (Ω, F) such that for all A in

F : Q(A) = Z(w)dP (w).
A
where Z corresponds to the density of Q with respect to P . It is often

denoted by dQ
dP
. These two probabilities P and Q are equivalent when they
are continuous with respect to each other.
The Girsanov theorem

Consider the probability space (Ω, F, F , P). If (Θt )0≤T is an adapted
process such that:
T
Θ2s ds < ∞
0
and the following process (Lt )0≤T is a martingale

t t
2
Lt = exp − Θs dWs − Θs ds
0 0
t
then Wt∗ = Wt + 0
Θs ds is a standard Brownian motion.
Appendix B: Resolution of the Partial Differential

Equation for a European Call Option
on a Non-Dividend Paying Stock
in the Standard Context
The problem is to search for the solution to the following partial differential
equation:

1 2 2 ∂ 2c ∂c ∂c
σ S + rS + − rc = 0 (B.1)
2 ∂S 2 ∂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]. Let us try a change
of variables and postulate that the solution is of the following form:
c(S, t) = f (t)y(u1 , u2 ) (B.2)
where f (t) and y(u1 , u2 ) are unknown functions.

We begin by calculating the different partial derivatives in the partial
differential equation with respect to time and to the underlying asset price.
The partial derivative ∂c

∂t is given by:

∂c ∂f ∂y ∂u1
= y(u1 , u2 ) + f (t)
∂t ∂t ∂u1 ∂t

∂y ∂u2
+ f (t) (B.3)
∂u2 ∂t
∂c
The partial derivative ∂S is given by:

∂c ∂y ∂u1 ∂y ∂u2
= f (t) + (B.4)
∂S ∂u1 ∂S ∂u2 ∂S
∂2 c
The partial derivative ∂S 2 is:
2 2
∂2c ∂ 2y ∂u1 ∂y ∂u1
= f (t) 2 +
∂S 2 ∂u1 ∂S ∂u1 ∂S 2
2 2
∂2y ∂u2 ∂y ∂u2
+ f (t) 2 +
∂u2 ∂S ∂u2 ∂S 2

∂2y ∂u1 ∂u2
+ 2f (t) (B.5)
∂u1 ∂u2 ∂S ∂S
These quantities for the partial derivatives are substituted in the partial
differencial Eq. (B.5)
2
1 2 2 ∂2y ∂u1 ∂2y ∂u2 ∂u1
0 = σ S f (t) + 2
2 ∂u21 ∂S ∂u1 ∂u2 ∂S ∂S
2 2 2 2
∂y ∂u1 ∂ y ∂u2 ∂y ∂u2
+ + +
∂u1 ∂S 2 ∂u22 ∂S ∂u2 ∂S 2

∂y ∂u1 ∂y ∂u2
+ rSf (t) +
∂u1 ∂S ∂u2 ∂S

∂f
− rf (t)y(u1 , u2 ) + y(u1 , u2 )
∂t

∂y ∂u1 ∂y ∂u2
+ f (t) + (B.6)
∂u1 ∂t ∂u2 ∂t
This last equation can be solved, if we refer to the heat transfer equation

∂y 1 ∂2y
= (B.7)
∂u2 2 ∂ 2 u1
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
Simplifying by f (t) and re-writing the partial differential equation to

identify the different terms, it appears that the following quantity:
2 2
∂ y 1 2 2 ∂u1
σ S
∂u21 2 ∂S
2
∂y 1 2 2 ∂u2 ∂u2 ∂u2
+ σ S + rS +
∂u2 2 ∂S ∂S ∂t
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
it is possible to show that when u2 (t) = −a2 then
u2 = −a2 (t − t∗ )
The Black–Scholes equation is reduced to:

2
1 2 2 ∂u1 ∂u1 ∂u1
σ S + rS + =0 (B.8)
2 ∂S ∂S ∂t
Using the expresion for a2 , we have:

∂u1 √ a
= 2
∂S σS
or

√ a S
u1 = 2 ln + b(t)
σ K
Replacing this in Eq. (B.8) gives:

2
1 2 2 1 √ a
√ a 1
σ S − 2 + b (t) + rS 2 =0
2 S σ σ S
1 √ √ a
aσ 2 + b (t) + r 2 =0
2 σ
Since the term b (t) is equal to:

√

1 √ √ a a 2 σ2
b (t) = aσ 2 − r 2 = −r
2 σ σ 2
then the term b(t) is given by

√

a 2 σ2
b(t) = − r (t − t∗ ).
σ 2
This allows to write the value of u1 (t) as:

2
√ a S σ ∗
u1 (t) = 2 ln + − r (t − t ) (B.9)
σ K 2
This analysis concerns the partial differential equation. Now, we turn to

the call’s maturity condition. Since at the option’s maturity date, the call
value can be written as c(S, t∗ ) = f (t)y(u1 , u2 ) and since

∂y ∂ 2y
=
∂u2 ∂u22
then, we can write the maturity condition as:

√
∗ ∗ ∗ a 2 S
c(S, t ) = y[u1 (S, t ), u2 (S, t )] = y ln ,0 (B.10)
σ K
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 −∞
where the limit as u2 approaches zero of y(ε, u2 ) is equal to U0 (ε).

The following condition must be satisfied by the option price function
at the maturity date
( √kσ )
c(S, t∗ ) = y(k, 0) = K e 2a − 1
when k is positive or 0, if k is negative with

√
2a S
k= ln
σ K
or

S σk
ln =√ .
K 2a
Hence
( √σk )
S = Ke 2a .
When this condition is applied, Eq. (B.11) becomes

u1

1 1 2 1 ε2
y(u1 , u2 ) = √ K σ r σ 2 e(u−ε) − 1) e(− 2 ) dε (B.12)
2Πu2 −∞ 2 2
If we make a change in variables, Eq. (B.12) can be written as:

∞
√

1 σ q2
y(u1 , u2 ) = √ √ K √ e(u1 +q 2s)
− 1 e(− 2 ) dq (B.13)
2Π −u1 / 2s a 2
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)
If we make a change of variables and set p = −q. Equation (B.15) can be

written as:
∗
Ker(t−t ) S σ2 ∗
y(u, t) = √ e( 2 −r)(t−t )
2Π K
−d −qσ
√ q2 ∗
× e t∗ −t e(− 2 ) dq − Ker(t−t ) N (d2 ) (B.16)
−∞
Using the fact that
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
−∞
2
√
−d+σ t∗ −t
( σ2 )(t∗ −t) 1 2
e e(− 2 p ) dp.
∞
After these computations, Eq. (B.16) becomes:
∗ σ2 2
−r)(t−t∗ )] [ σ2 (t−t∗ )] ∗
y(u, t) = er(t−t ) e[( 2 e SN (d1 ) − Ker(t−t ) N (d2 )
The European call price is given by:

∗
c(S, t) = 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
Appendix C: Approximation of the Cumulative

Normal Distribution
The cumulative normal distribution given by
d
1 x2
N (d) = √ e(− 2 )
dx
2Π −∞
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
a1 = 0.196854, a2 = 0.115194, a3 = 0.000344, a4 = 0.019527
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
p = 0.0.2316419, b1 = 0.319381530, b2 = −0.356563782,

b3 = 1.781477937, b4 = −1.821255978,
1
b5 = 1.330274429, c=
(1 + px)
the formula is
1 x2
N (x) = 1 − √ e 2 (b1 c + b2 c2 + b3 c3 + b4 c4 + b5 c5 ).
2π
Appendix D: Leibniz’s Rule for Integral Differentiation

Consider the following integral
B(x)
I(x) = F (x, t)dt
A(x)
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].
Appendix E: Pricing Bonds: Mathematical Foundations

Using the definition in Section 12.3, for any P ∗ of an arbitrage-free family
of bond prices, we have:
RT
B(t, T ) = EP ∗ e− t ru du | Ft , ∀ t ∈ [0, T ] (E.1)
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 ]
According to the return-to-maturity expectations hypothesis (RTM-EH),

the return from holding a bond to maturity is given by the return expected
from a roll-over strategy of a series of a single-period bonds or:
1 RT
= EP e t ru du | Ft , ∀ t ∈ [0, T ],
B(t, T )
According to the yield-to-maturity expectations hypothesis (YTM-EH), or

the unbiased expectations hypothesis, the yield from holding a bond to
maturity is given by the yield expected from a roll-over strategy of a series
of a single-period bonds or:

T
B(t, T ) = exp −EP ru du | Ft , ∀ t ∈ [0, T ]
t
This formula can also be written as:

T
1
Y (t, T ) = EP ru du | Ft ,
T −t t
or
f (t, T ) = EP (rT | Ft ), ∀ t ∈ [0, T ]
In this context, the forward interest rate, f (t, T ) is an unbiased estimate of

the future short-term interest rate. The Ito process is often used to model
the dynamics of the short-term interest rate. When the dynamics of r is

given by:
drt = µt dt + σt · dWt , r0 > 0
or in an equivalent form:
t t
rt = r0 + µu du + σu · dWu , ∀ t ∈ [0, T ]
0 0
the underlying probability is the actual probability.

To construct the martingale probability, re-call that any probability Q
equivalent to P is given by the Radon–Nikodym derivative
.
dQ
= ET ∗ αu · dWu = ηT ∗ , P − a.s. (E.2)
dP 0
for some process where:

. t t
1 2
ηt = Et αu · dWu = exp αu · dWu − | αu | du
0 0 2 0
(See Musiela, 1997 for details).

Using the Girsanov theorem, the process:
t
Wtα = Wt − αu du, ∀ t ∈ [0, T ∗]
0
follows a d-dimensional Brownian motion under P α which denotes the

probability measure in Eq. (E.2). In this context, Musiela (1997) gives the
following proposition:
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 :
drt = (µt + σt · αt )dt + σt · dWtα
and there exists a process bα (t, T ) such that:
dB(t, T ) = B(t, T )(rt dt + bα (t, T ) · dWtα ) (E.3)

Hence, the bond price is given by:

t t
1 2
B(t, T ) = B(0, T )Bt exp bα (u, T ) · dWuα − |bα (u, T )| du
0 2 0
The corollary of the above proposition is that in the presence of two

probability measures equivalent to P , when the bond price is given by
Eq. (E.1), then:
RT R T∗ 1
R T∗ 2
B(t, T ) = EP α̃ e− t ru du e t (αu −α̃u )·dWu − 2 t |αu −α̃u | du |Ft .
α̃
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 :
dB(t, T ) = B(t, T )((rt − αt · bγ (t, T ))dt + bα (t, T ) · dWt ).
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
dS = A(S, t)dX + B(S, t)dt
where the functions A(S, t) and B(S, t) depend on S and t.

Consider a derivative security f (S, t).
1. Apply Ito’s Lemma or Taylor’s Theorem to the function f (S, t).
2. Can you choose the values of A and B so that a function g(S) has a zero
drift, but non-zero variance?
Solution
First method: Application of Ito’s Lemma
∂f ∂f 1 ∂2f
df = dS + dt + (dS)2
∂S ∂t 2 ∂S 2
we will replace dS by its equivalent indicated in the previous equation.
∂f ∂f ∂f 1 ∂2f 2
df = AdX + B dt + dt + A (dX)2
∂S ∂S ∂t 2 ∂S 2
with
(dS)2 = A2 dX 2 + B 2 (dt)2 + 2ABdXdt
or
dX 2 ∼ dt, (dt)2 = 0; dXdt = 0
Hence,

∂f ∂f ∂f 1 ∂ 2f
df = A dX + B + + A2 2 dt
∂S ∂S ∂t 2 ∂S
Second method: Application of Taylor’s theorem

Taylor’s theorem can be applied to a function f (S + δS, t + δt) over a small
time step. It can be written as:
∂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
Since δS is given by δS = σSδX + µSδt, it can be replaced to give:
∂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
We denote by A = σS and B = µS. In this case, δf can be written as:
∂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

∂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
To solve the following equation
dy 2B(S, t)
= 2 d(S, t),
y A (S, t)
let (S, t) = x hence

2B(x)
Log|y| = − dx + K
A2 (x)
We can denote by

2B(x)
F (x) = − dx
A2 (x)
2B
and y = eK eF (x) . The solution to this problem can be found when A2 is
independent of time.
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
The correlation coefficient between the two processes is denoted by ρ.

What is the stochastic differential equation that must be satisfied by
the function f (S1 , S2 )? Taylor’s Theorem can be used to find the change in
the function f over short-time intervals.
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
When the term δt → 0, δt becomes equivalent to dt, δX1 becomes equivalent

to dX1 , δX2 becomes equivalent to dX2 and the term (δX1 )2 becomes
equivalent to dt, and finally (δX2 )2 becomes equivalent to dt; δX1 δX2 dX2
becomes equivalent to ρdt. This gives the stochastic differential equation
that must be satisfied by f (S1 , S2 )

∂f ∂f ∂f ∂f
df = σ1 S1 dX1 + σ2 S2 dX2 + µ1 S1 + µ2 S2
∂S1 ∂S1 ∂S1 ∂S2

1 ∂ 2f ∂2f 1 ∂ 2f
+ σ12 S12 2 + ρσ1 σ2 S1 S2 + σ22 S22 2 dt
2 ∂S1 ∂S1 ∂S2 2 ∂S2
Exercise
Consider the following dynamics of the underlying asset:
dS = µSdt + σSdX.
Find the stochastic differential equation that must be satisfied by
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
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.
1895–1927.
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, 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
Finance, 13(2), 1826–1837.
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 S Pliska (1981). Martingales and stochastic integrals in the
theory of continuous trading. Stochastic Processes and their applications, 11,
215–260.
Longstaff, FA (1990). Time varying term premiums and traditional hypothesis

about the term structure. Journal of Finance, 45, 1307–1314.
Musiela, M (1997). Martingale Methods in Financial Modelling: theory and
Applications. Berlin: Springer.
Vasicek, O (1977). A equilibrium characterization of the term structure. Journal
of Financial Economics, 5, 177–188.

Chapter 13
SIMPLE EXTENSIONS AND GENERALIZATIONS

OF THE BLACK–SCHOLES TYPE MODELS IN
THE PRESENCE OF INFORMATION COSTS
Chapter Outline
1. Section 13.1 provides 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. Section 13.2 presents the valuation of securities dependent on several
variables in the presence of incomplete information.
3. Section 13.3 provides the general differential equation for the pricing of
derivatives.
4. Section 13.4 gives the extension of the risk-neutral argument in the
presence of information costs.
5. Section 13.5 provides an extension of the analysis to commodity futures
prices in the presence of information costs.
6. Appendix A provides a general equation for derivative securities.
7. Appendix B gives an extension to the risk-neutral valuation argument.
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
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.
13.1. Differential Equation for a Derivative Security

on a Spot Asset in the Presence of a Continuous
Dividend Yield and Information Costs
We denote by f the price of a derivative security on a stock with
a continuous dividend yield q. The dynamics of the underlying asset are
given by:
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
It is possible to construct a portfolio Π by holding a position in the

derivative security and a certain number of units of the underlying asset:
∂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
Over the same time interval, dividends are given by qS ∂V

∂S ∆t.
Simple Extensions and Generalizations 585
We denote by ∆W , the change in the wealth of the portfolio holder.


∂V 1 ∂ 2f 2 2 ∂V
∆W = − − σ S + qS ∆t.
∂t 2 ∂S 2 dS
Since this change is independent of the Wiener process, the portfolio is
instantaneously risk less and must earn the risk-free rate plus information
costs or:

∂V 1 ∂2f 2 2 ∂V ∂V
− − 2
σ S + qS ∆t = −(r + λV )V ∆t + (r + λS )S ∆t
∂t 2 ∂S ∂S ∂S
where λi refers to these costs. This gives:
∂V ∂V 1 ∂ 2V 2 2
+ (r + λS − q)S + σ S = (r + λV )V. (13.1)
∂t ∂S 2 ∂S 2
Equation (13.1) must be satisfied by the derivative security in the presence
of information costs and a continuous dividend yield.
13.2. The Valuation of Securities Dependent on Several

Variables in the Presence of Incomplete Information:
A General Method
When a variable does not indicate the price of a traded security, the pricing
of derivatives must account for the market price of risk. The market price
of risk γ for a traded security is given by:
µ−r−λ
γ= (13.2)
σ
where µ indicates the expected return from the security.
This equation can also be written as:
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
When γ < 0, the expected return on an asset is less than the risk free
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)
It is possible to construct a portfolio Π which is risk free using σ2 V2 of the

first derivative security and −σ1 V1 of the second derivative security.
Π = σ2 V2 V1 − σ1 V1 V2 .
The change in the value of this portfolio can be written as:
∆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
or
µ1 − (r + λ1 ) µ2 − (r + λ2 )
= . (13.8)
σ1 σ2
The term µ−(r+λ)

σ
must be the same for all securities that depend on time
and the variable θ. It is also possible to show that V1 and V2 must depend
positively on θ. Since, the volatility of V1 is σ1 , it is possible to use Ito’s
lemma for σ1 to obtain:
∂V1
σ1 V1 = sθ .
∂θ
Hence, when V1 is positively related to the variable θ, the σ1 is positive
and corresponds to the volatility of V1 . But, when f1 is negatively related
to the variable θ, σ1 is negative and Eq. (13.4) can be written as:
dV1
= µ1 dt + (−σ1 )(−dz).
V1
This indicates that the volatility is −σ1 rather than σ1 . The result in
Eq. (13.3) can be generalized to n state variables.
Consider n variables, which are assumed to follow Ito diffusion pro-
cesses, where for each state variable i between 1 and n, we have:
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
where µ corresponds to the expected return from the security and σi is

its volatility. The volatility of V is σi when all the underlying variables
except θi are kept fixed. This result is obtained directly using an extension
of the generalized version of Ito’s lemma in its discrete form. We show in
Appendix A that:
n

µ − r − λV = γi σi (13.10)
i=1
where γi indicates the market price of risk for the variable θi .

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 .
13.3. The General Differential Equation for the Pricing

of Derivatives
We denote by
• θi : value of ith state variable;
• mi : expected growth in ith state variable;
• γi : market price of risk of ith state variable;
• si : volatility of ith state variable;
• r: instantaneous risk-free rate and
• λi : shadow cost of incomplete information of ith state variable,
where i takes the values from 1 to n.
Garman (1976) and Cox et al. (1985) have shown that the price of any
contingent claim must satisfy the following partial differential equation:
∂V ∂V 1 ∂2f
+ θi (mi − γi si ) + ρi,k si sk θi θk = rf
∂t i
∂θi 2 ∂θi ∂θk
i,k
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
i,k
In the presence of a single state variable, θ, Eq. (13.11) becomes:

∂V ∂V 1 ∂ 2V
+θ (m − γs) + s2 θ2 2 = (r + λ)V. (13.12)
∂t ∂θ 2 ∂ θ
For a non-dividend paying security, the expected return and volatility
must satisfy: m − r − λ = γs and m − γs = r + λ.
In this case, Eq. (13.12) becomes the extended Black–Scholes equation
in the presence of information costs.
For a dividend-paying security at a rate q, we have:
q + m − r − λ = γs or m − γs = r + λ − q.
In this case, Eq. (13.12) becomes (13.1)

∂V ∂V 1 ∂ 2V
+ (r + λs − q)S + s2 S 2 2 = (r + λV )V.
∂t ∂S 2 ∂ S
13.4. Extension of the Risk-Neutral Argument

in the Presence of Information Costs
We know that the market price of risk is given by
µ−r−λ
γ= or µ − r − λ = γσ.
σ
Appendix B shows how to price a derivative as if the world were risk neutral.
This is possible when the expected growth rate of each state variable is
(mi − γi si ) rather than mi .
For the case of a non-dividend paying traded asset, using Eq. (13.5),
we have
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:
f = e−(r+λV )(T −t) Ê[fT ] (13.13)
where VT corresponds to the security’s payoff at maturity T and Ê is to the

expectation operator in a risk-neutral economy. This refers to an economy
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:
V = Ê[e−(r̄+λV )(T −t) VT ] (13.14)
where r̄ corresponds to the average risk-free rate between current time t

and maturity T .
13.5. Extension to Commodity Futures Prices

within Incomplete Information
Consider the pricing of a long position in a commodity forward contract
with delivery price K and maturity T .
13.5.1. Differential equation for a derivative security

dependent on a futures price in the presence
of information costs
Assume that the relationship between the futures price F and the spot price
S is given by F = Seα(T −t) , where α depends only on time. In this case, if
the volatility of the underlying asset is constant, then the volatility of F is
constant and equal to that of S. The dynamics of the spot asset are given
by dS = µSdt + σSdz, where µ is the instantaneous expected return, σ is
the instantaneous volatility, and dz is an increment of a Wiener process.
Using Ito’s lemma, the volatility of the futures price σF is given by:
∂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.
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
Consider the value of a portfolio comprising the derivative security and a

position in a certain number of the underlying futures contracts. The total
change in wealth for the portfolio holder over a short time interval can be
written as:
∂f
∆W = ∆F − ∆f.
∂F
The discrete versions of the previous equations for dF and df can be
written as:
∆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
and this gives
∂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.
13.5.2. Commodity futures prices

We denote by S the spot price of the commodity, F , the futures price, µ
the growth rate of the commodity price, σ its volatility, and Γ its market
price of risk. Using the extension of the risk-neutral valuation principle, the
price is given by
V = e−(r+λV )(T −t) Ê[ST − K] or f = e−(r+λf )(T −t) (Ê[ST ] − K)

(13.15)
where Ê(.) refers to the expected value in a risk-neutral economy.
The forward or futures price F corresponds to the value of K that
makes the value of the contract f equal to zero in Eq. (13.15). So, we have:
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:
Ê[ST ] = E[ST ]e−γσ(T −t)
where E corresponds to the real expectations or the expectations in the

real world. Using Eq. (13.16), we have:
F = E[ST ]e−γσ(T −t) . (13.17)
If γ = 0, the futures price is an unbiased estimate of the expected spot

price. However, if γ > 0, the futures price corresponds to a downward-
biased estimate of the expected spot price. If γ < 0, the futures price is a
downward-biased estimate of the expected spot price.
13.5.3. Convenience yields

When the drift or the expected growth in the commodity price is constant,
we have E(ST ) = Seµ(T −t) , Using Eq. (13.17), we have F = Se(µ−γσ)(T −t) .
This last equation is consistent with the cost of carry model, when the
convenience yield y satisfies the relationship:
µ − γσ = r + λ + g − y. (13.18)
This equation shows that the commodity can be assimilated to a traded

security paying a continuous dividend yield equal to the convenience yield
y minus the storage costs u. Hence, the convenience yield must satisfy the
relationship:
y = g + r + λ − µ + γσ.
When the convenience yield is zero, we have
µ − γσ = r + λ + g.
This result shows that some commodities can be assimilated to traded

securities paying negative dividend yields, which are equal to storage costs.
The main derivations in this chapter appear in Bellalah (1999).
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?
Appendix A: A General Equation for Derivative Securities

Consider a derivative security whose price depends on n state variables
and time t. The security can be priced under the standard Black–Scholes
assumptions. The state variables are assumed to follow Ito diffusion
processes, where for each state variable i between 1 and n, we have:
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,
• Vj : price of the jth traded security for j between 1 and n + 1;

• r: risk-free rate;
• λj : information cost for the jth traded security for j between 1 and n+ 1
and
• ρi,k : correlation coefficient between dzi and dzk .
Since, the (n + 1) traded securities depend on θi , then using Ito’s lemma,

we have:

fj = µj Vj dt + σi,j Vj dzi (A.1)
i
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 .
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

aj fj (µj − r − λj ) = 0. (A.6)
j
Equations (A.4) and (A.6) are consistent only if:

fj (µj − r + λj ) = γi σij fj (A.7)
i
or

µj − r − λj = γi σij (A.8)
i
where for γi , i is between 1 and n.

Using Eqs. (A.2) and (A.3) and replacing in Eq. (A.7) gives the
following equation:
∂Vj ∂Vj 1 ∂ 2 Vj ∂Vj

+ m i θi + ρik si sk θi θk − (r + λj )Vj = γi si θi .
∂t ∂θi 2 ∂θi ∂θk ∂θi
i i,k i
This last equation reduces to:
∂Vj ∂Vj 1 ∂ 2 Vj
+ θi (mi − γi si ) + ρik si sk θi θk = (r + λj )Vj .
∂t i
i,k
Hence, any security f contingent on the state variables θi and time

must satisfy the following second-order differential equation:
∂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
Appendix B: Extension to the Risk-Neutral Valuation

Argument
When the variable θi is not a traded asset, it is possible to assume the
existence of a traded asset θ˜i paying a continuous dividend q̂ where:
q̂ = r + λi − mi + γi si .
The values of θî and θi must be equal and the following differential equation
must be verified:
∂V ∂V 1 ∂ 2V
+ θî (r + λi − q̂i ) + ρi,k si sk θî θˆk = (r + λ)V.
∂t ∂ θî 2 ∂ θî ∂ θˆk
i i,k
This equation is independent of risk preferences. Since
(r + λi − q̃) = mi − γi si ,
the derivative security can be valued in a risk-neutral economy if the drift

term in θi is modified from mi to 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
Hence, the equation becomes:
(r + λs )S − (r + λv )V = 0
or
V (S, t) = S and, in this case λv = λs .
then V (S, t) = S verify B-S equation.
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
This leads to the extended Black–Scholes equation:
(r + λv )e(r+λv )t − (r + λv )e(r+λv )t = 0
then V (S, t) = e(r+λv )t verify the extended Black–Scholes equation.
Exercise 2: The Black–Scholes Model in the presence

Can you find the most general solution to the extended Black–Scholes
equation with each of the following forms:
(1) V (S, t) = A(S)

(2) V (S, t) = B(S)C(t)
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
We can try a solution of the form A(S) = S n because the powers of S in

the equation match the order of the derivatives. This gives:
1 2 2
σ S (n(n − 1))S n−2 + (r + λs )nS n−1 S − (r + λA )S n = 0
2
⇐⇒
1 2 n
σ S (n(n − 1)) + (r + λs )nS n − (r + λA )S n = 0
2
⇐⇒
1 2
σ n(n − 1) + (r + λs )n − (r + λA ) = 0
2
⇐⇒
1 2 2 1
σ n + (r + λs − σ2 )n − (r + λA ) = 0 (13.20)
2 2
Equation (13.20) is of order 2 in n if and only if 12 σ 2 = 0. The following
cases must be studied.
a) When
1 2
σ = 0; (r + λA ) = 0.
2
Hence,
1 2
r = −λA and r + λs = σ .
2
This gives
1 2 2 1
σ n − (r + λA ) = 0 ⇐⇒ σ2 n2 = (r + λA )
2 2
For this equation, we have:
2(r + λA )
n2 =
σ2
and the roots are

2(r + λA ) 2(r + λA )
n1 = and n2 = − .
|σ| |σ|
The general solution for V (S, t): V (S, t) = CS n1 + BS −n1 , where C
and B are constants.
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
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
We can compute ∆ for Eq. (13.21) as follows:

2
σ 2 1
∆ = λs + − 2λA σ 2 − λs − σ2 = 2σ 2 (λs − λA )
2 2
b.1) When
λs = λA ⇐⇒ ∆ = 0
In this case, Eq. (13.21) has a double solution:

σ2
r1 = r2 = − λs + < 0.
2
b.1.1)
∀ r ∈ −{r1 = r2 }, ∆>0
and Eq. (13.20) has two solutions:

2 2 2
− r + λs + σ2 + r + λs + σ2 + 2σ 2 (r + λs )
n1 =
σ2
or

σ2 2 2
− r + λs + 2 + r + λs + σ2
n1 =
σ2
which is also equivalent to:
σ2 σ2
−r − λs + 2
+ |r + λs + 2
|
1 =
σ2
In fact,

1 2 1
if r + λs + σ = r + λs + σ2 then n1 = 1.
2 2
or, when

r + λs + 1 σ2 = −r − λs − 1 σ2
2 2
then,

−2 (r + λs ) − r + λs − 12 σ 2 − |r + λs + 12 σ 2 |
n1 = and n2 = .
σ2 σ2
Hence,
−2 (r + λs )
n2 = 1 or n2 = .
σ2
Finally, the solution can be written as:
V (S, t) = HS n1 + KS n2
where H and K are arbitrary constants or

−2(r+λs ) σ2
(S, t) = HS + KS σ2 , ∀ r ∈ − r1 = − λs + .
2
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
and the result is:
V (S, t) = BS; B = constant.
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
where G and L are arbitrary constants.

b.3) When
λA < λs ; ∆ > 0.
Equation (13.20) has two solutions:

2

− λs + σ2 − |σ| 2(λs − λA )
r1 = <0
1

σ2
r2 = − λs + + |σ| 2(λs − λA ) < 0
2
for this, we will have the following results.
b.3.1) For the interval:
] − ∞, r1 [∪]r2 , + ∞[, ∆>0
In this case, Eq. (13.20) has two solutions:

√ √
− r + λs − 12 σ 2 + ∆ − r + λs − 12 σ 2 − ∆
α1 = , α2 = .
σ2 σ2
Hence:
V (S, t) = NS α1 + MS α2
where N and M are arbitrary constants.

b.3.2) When
σ2
r1 = −λs − − |σ| 2(λs − λA ), ∆ = 0.
2
−(r + λs − 12 σ 2 )
β=
σ2
or

σ2
− −λs − 2
− |σ| 2(λs − λA ) + λs − 12 σ 2
β=
σ2

−σ 2 − |σ| 2(λs − λA ) 2(λs − λA )
β=− =1+
σ2 |σ|
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
then Eq. (13.20) has a double solution:

r + λs − 12 σ 2 2(λs − λA )
g 1 = g2 = g = − 2
=1− .
σ |σ|
Hence, we have V (S, t) = F S g , where F is a constant.
c) When (r + λA ) = 0 then r = −λA .

Using Eq. (13.20), we have:

1 2 1
n σ n + r + λs − σ2 = 0.
2 2
Hence:
−2(r + λs ) + σ 2 2(λs − λA )
n = 0 or n = =− +1
σ2 σ2
and finally, we have:
2(λs −λA )
V (S, t) = C + DS − σ2
+1
where C and D are constants.
2) Show that when V (S, t) = B(S)C(t) is replaced in the extended Black

and Scholes equation, this gives:
∂C 1 ∂ 2B ∂B
B + σ2 S 2 C + (r + λs )SC − (r + λv )B(S)C(t) = 0.
∂t 2 ∂S 2 ∂S
In fact, since B(S) is two times derivable, we have:
∂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
Note that the left-hand side C1 ∂C

∂t of the equation is a function of time t.
The right-hand side,

1 1 2 2 ∂2B ∂B
− σ S + (r + λs )S − (r + λv )B(S)
B 2 ∂S 2 ∂S
is a function of S.
Both sides of the equation must be equal to a constant,
1 ∂C C
= = k.
C ∂t C
Therefore, we have:
LogC(t) = kt and C(t) = C0 ekt .
Using the right-hand side of the equation, we have:

1 1 2 2 ∂ 2B ∂B
σ S + (r + λs )S − (r + λv )B(S) = k
B 2 ∂S 2 ∂S
or
1 2 2 ∂ 2B ∂B
σ S + (r + λs )S − (r + λv )B(S) = −Bk
2 ∂S 2 ∂S
1 2 2 ∂2B ∂B
σ S 2
+ (r + λs )S − (r + λv − k)B(S) = 0.
2 ∂S ∂S
It is possible to try the following form B(S) = S n which gives:
1 2 n
σ S n(n − 1) + (r + λs )S n n − (r + λv − k)S n = 0
2
or
1 2
σ n(n − 1) + (r + λs )n − (r + λv − k) = 0
2
and finally

1 2 2 1
σ n + r + λs − σ 2 n − (r + λv − k) = 0 (13.22)
2 2
Now, we can look for the solutions to this quadratic equation.
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
The quadratic equation for n has two roots n1 and n2 :

2(r + λv − k) 2(r + λv − k)
n1 = and n2 = −
|σ| |σ|
Since

2 12 σ 2 + λv − λs − k
n1 = −n2 , n1 =
|σ|
then we have,
V (S, t) = (A1 S n1 + A2 S n2 )ekt
for two arbitrary constants A1 and A2 .
Second case
If
1 2 1
r − k − λv = 0, r = k − λv , σ = 0 and r + λs − σ2 = 0,
2 2
then the quadratic equation for n becomes:

1 2 1
n σ n + k − λv − σ 2 + λ s = 0
2 2
and the solutions n1 or n2 are given by:
−2(k − λv + λs ) + σ 2 −2
n1 = 0 or n2 = = 2 (k − λv + λs ) + 1.
σ2 σ
The final solution is:

V (S, t) = (D1 + D2 S n2 )ekt
where D1 and D2 are arbitrary constants.
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).
In this case, three situations must be analyzed.
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:
V (S, t) = (ES n1 + F S n2 )ekt
with F and E are constants.

Second situation
If
∆ = 0 ⇐⇒ λs = λv + k ⇐⇒ λs − λv = k,
then ∆ shows a double solution

σ2
r1 = r2 = r = − λs + .
2
Hence: Consider in this second situation, the case: ∀r ∈ −r1 ; ∆ > 0. In

this case, Eq. (13.20) has two distinct solutions:

2
− r + λs − 12 σ 2 − r + λs − 12 σ 2 − 2σ 2 (λs − r − 2λv )
m1 =
σ2

2
− r + λs − 12 σ 2 + r + λs − 12 σ 2 − 2σ 2 (λs − r − 2λv )
m2 = .
σ2
Finally, the solution is given by:
(S, t) = (GS m1 + HS m2 )e(λs −λv )t
where G and H are constants.

Consider in this second situation the case: when

σ2
r1 = r2 = − λs + ; ∆ = 0.
2
Equation (13.20) shows a double solution.
σ2
−(r + λs − 2 )
m1 = = 1.
σ2
Hence, the final solution is V (S, t) = ISe kt = IS 1 e(λs −λv )t , where I is a

constant.
Third situation
∆ > 0 ⇐⇒ λs > λv + k.
In this situation, the ∆ of the equation shows two distinct roots:

σ2
a1 = − λs + − |σ| 2(λs − λv − k)
2

σ2
a2 = − λs + + |σ| 2(λs − λv − k)
2
and
∀ r ∈] − ∞, a1 [∪]a2 , +∞[ ⇐⇒ ∆ > 0.
Equation (13.20) has two distinct roots given by:
σ2
√ σ2
√
−((r + λs ) − 2 ) + ∆ −((r + λs ) − 2 ) − ∆
α= , β= .
σ2 σ2
The solution is given by
V (S, t) = (LS α + JS β )ekt
where L and J are constants.

When r = a1 , the ∆ = 0, there is a double solution:

2(λs − λv − k)
α1 = 1 + .
|σ|
Hence, 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|
Hence, the solution is given by
V (S, t) = NS γ ekt
where N is a constant.
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)
then this leads to the following equation:
∂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
S = Eex and V (S, t) = Ev(x, t)
and compute the following partial derivatives:
∂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
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)
Now, we can choose a new time variable τ such that:
2 ∂v(x, t) ∂
= v(X, τ )
σ 2 (t) ∂t ∂τ
i.e.,
t
1
τ =− σ 2 (s)ds.
2 0
The equation is reduced to:

∂v ∂2v 2 ∂v 2
= + (r(t) + λs (t)) − 1 − (r(t) + λv (t)) 2 v.
∂τ ∂x2 σ 2 (t) ∂x σ (t)
If we set
v(x, t) = eαx+βτ w(x, τ )
and choose the values of:

1 2
α=− (r(t) + λs (t) − 1
2 σ2
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
Exercise 4
For the following problem:
∂u ∂ 2u
= ; −∞ < x < ∞, τ > 0 with u(x, 0) = u0 (x) > 0
∂τ ∂x2
it is possible to show that
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
u(x, 0) = u0 (x) > 0 and u ⇐⇒ 0 as |x| ⇐⇒ ∞
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.
For an option with a positive payoff, we want to solve the following

extended Black–Scholes equation:
∂V 1 ∂ 2V 2 2 ∂V
+ 2
S σ + (r + λs )S − (r + λv )V = 0
∂t 2 ∂S ∂S
under the condition that V (S, T ) > 0.

We apply the same transformation as before and in particular, we apply

the first transformation to obtain:

∂v ∂2v 2 ∂v 2
= 2
+ 2 (r + λs ) − 1 − (r + λv ) 2 v
∂τ ∂x σ (t) ∂x σ (t)
and
1
v(x, 0) = V (Eex , T ) > 0.
E
2
The second transformation gives ∂u ∂ u
∂τ = ∂x2 with initial data u(x, 0) =
e−αx v(x, 0) > 0. Hence, we show that u(x, τ ) > 0 leads to v(x, τ ) =
eαx+βτ u(x, τ ) > 0 and V (S, t) = V (Eex , T − σ2τ2 ) = Ev(x, t) > 0.
Finally, the option value is always positive and the result is independent
of the term λv .
References
information costs. Journal of Futures Markets, 19 (September) 645–664.
Briys, E, M Bellalah et al. (1998). Options, Futures and Exotic Derivatives. En
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.
Part V
Extensions of Option Pricing Theory to American

Options and Interest Rate Instruments in a
Continuous-Time Setting: Dividends, Coupons and
Stochastic Interest Rates
613
614
Chapter 14
EXTENSION OF ASSET AND RISK

MANAGEMENT IN THE PRESENCE OF
AMERICAN OPTIONS: DIVIDENDS, EARLY
EXERCISE, AND INFORMATION UNCERTAINTY
Chapter Outline
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
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
Extension of Asset and Risk Management in the Presence of American Options 617
of American options on commodity futures contracts. These formulas are

accurate and computationally more efficient than compound-option pricing
models, binomial models, and finite difference methods. The approach
relies on a quadratic approximation method similar to that used by
McMillan (1986) for the valuation of the American put option on a non-
dividend-paying stock. However, the formulas apply only for a constant
proportional rate.
When there are discrete distributions to the underlying asset, the
valuation of American options is more complex. In this context, models were
proposed by several authors including Roll (1977), Geske (1979), Whaley
(1981), and Whaley (1986) for call options. Put formulas were proposed
by Johnson (1983), Geske and Johnson (1984), Geske and Shastri (1985),
Blomeyer (1986), and Barone-Adesi and Whaley (1987), among others.
Most of these models, if not all, are based on the concept of compound
options. The compound option or an option on an option has been studied
in a context of complete information by several authors. The concept of
an option on an option is important in the study of several opportunities
with a sequential nature where some of them are available only if earlier
opportunities are undertaken. Black and Scholes (1973), Black and Cox
(1976), Galai and Masulis (1976), and Geske (1979) show that several
corporate liabilities may be considered as options. In a context of complete
information, they studied the pricing of a firm’s common stock and bonds
by considering the stock as an option on the firm’s value. They showed
that corporate investment opportunities may be analyzed as options and
compound options. However, their analysis does not account for information
uncertainty. We use arbitrage arguments to derive the formula in a Black
and Scholes (1973) economy. Such a formula might be applied to the
valuation of equity in the capital structure of the firm. The information
uncertainty about the firm and its cash flows reflects the agency costs and
the asymmetric information problems. By assuming the stock as an option
on the value of the firm, the value of the call as a compound option can
be derived as a function of the firm’s value by accounting for information
costs and the effects of leverage. We present two alternative derivations
for the value of the firm’s stock as a compound option in the presence of
information costs. 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) etc.
14.1. The Valuation of American Options:

The General Problem
Two problems arise in the valuation of American options. The first is
associated with the possibility of early exercise. The second is linked to the
distributions to the underlying asset. The nonexistence of a put call parity
theorem for American options implies a distinct treatment for call and put
options. The problem is different according to the pattern of distributions to
the underlying asset. In fact, if the dividend stream is continuous, it may not
induce the optimal early exercise of American options. However, when the
dividends are discrete, this may induce early exercise of American options.
14.1.1. Early exercise of American calls

We deal with the main reasons behind rational exercise of American options.
Exercise without distributions

When there are no distributions to the underlying asset, there is no incentive
to exercise an American call option before its maturity date. Hence, the
value of an American call is equal to that of a European call. This result
is due to Theorems 1 and 2 of Merton (1973) where it is shown that in the
absence of dividend payments, an American call will never be exercised prior
to expiration. In this context, the value of an American call is equivalent to
that of a European call when the interest rates are constant. The intuition of
this result is simple. In the absence of dividends, the option is worth more
“alive” than “dead” because of its time value. Killing the option would
mean pocketing its intrisic value while losing its speculative. The investor
is better off in selling the option rather than killing it.
Early exercise with continuous distributions

When dividends are paid continuously at a constant rate of d dollars per
unit time, and when the interest rate is constant, a sufficient condition for
no premature exercise is that:
d
K> (14.1)
r
This condition shows that the call strike price must be greater than the
ratio of the continuous dividend rate to the short-term risk-less rate.
Early exercise with discrete distributions

When the amounts of dividends per share dj are paid at different dates τj ,
for j = 1 to n, are known and the interest rate is constant, then a sufficient
condition for no premature exercise is:
p
P (τj )
K> d(τj ) (14.2)
1 − P (τn )
j=1
where P (τ ) is the price of a discount bond paying one-Dollar in τ years.

This condition shows that the net present value of future dividends
must be less than the present value of earnings from investing the strike
price for τ periods. The intuition of this result is simple. In fact, if the
losses from dividends are less than the gains from investing the required
funds to exercise the option and hold the stock, the option is exercised.
When the dividend per share is D(S, t), the B–S differential equation is
slightly modified. In fact, the instantaneous rate of return is no longer µSdt
but rather µ−D(S,t)
S
dt. This gives the following partial differential equation,
which was formulated first by Merton (1973):

1 2 2 ∂ 2C ∂C ∂C
σ S + [rS − D] + − rC = 0 (14.3)
2 ∂ 2S ∂S ∂t
where C(S, τ, K) is the American call value. This partial differential
equation must satisfy the following boundary conditions:
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
0 ≤ S ≤ I(τ ). Since I(τ ) is an unknown function of time, the structure of

the problem is complex. Indeed the boundary conditions are semi-infinite
and time dependent. The value of I(τ ) must be determined as part of the
solution. As defined by Samuelson (1972) and Merton (1973), this is rather
a difficult problem since the unknown function I(τ ) must be determined
from the behavior of the call’s holder, who maximizes the option value at
each instant, by adding a condition of regularity.
To make this point explicit, consider a function f (S, τ ; K, I(τ )), which
is a solution to this problem for a given I(τ ), or:
C(S, τ, K) = max(I) f (S, τ ; K, I).
The optimal I(τ ) is independent of the current underlying asset price,

and the following condition, which is “high contact” at the boundary
must be accounted for ∂C(I(τ ∂S
),τ,K)
= 1. This means that the option’s
partial derivative with respect to the underlying asset price is equal to one.
Intuitively, this condition corresponds to the point of tangency between
the call function and its intrinsic value, since only in-the-money options
are exercised. The proof of this condition is relatively a simple matter. In
fact, for the function f (x, I), (differentiable and concave with respect to its
second argument), the total derivative with respect to I along the boundary
x = I is given by:
df dh
= = f1 (I, I) + f2 (I, I).
dI dI
When the function has a maximum for a level I = I ∗ , f2 (x, I ∗ ) = 0.
df
Since, dI = dh ∗ ∗
dI = f1 (I , I ) and h = I − K, so f1 (I , I ) =
∗ ∗
dh
dI = 1. This is the proof that the derivative must be equal to one. The
solution to this problem gives the value of the American call when there
are dividends. Samuelson (1972) and Merton (1973) analyzed this “free
boundary” problem and did not provide solutions for a finite-life option.
However, a solution for an infinite-life option (a perpetual warrant) was
provided by Merton (Eq. (46), p. 172).
14.1.2. Early exercise of American puts

The valuation of American calls is simpler than the valuation of American
puts with and without discrete cash distributions to the underlying asset.
While European and American calls have the same value when there are no
distributions to the underlying asset, this result does not hold for European
and American puts. While there is no incentive to exercise an American call

option before its maturity date in the absence of dividends, there is always
a probability of pre-mature exercise of an American put. This is due to the
first arbitrage condition, according to which the value of an American put
must be greater than its intrinsic value at each instant, and second, to the
fact that the value of a European put with an infinite time to maturity or
a perpetual put is zero.
Exercise without distributions

Since there is no put-call parity relationship for American options,
American puts must be treated separately. The problem with the rational
pricing of European or American put options is somewhat different from
that of European or American calls. This results from the work of Merton
(1973) who showed that an American put option can be exercised early
even in the absence of distributions to the underlying asset. The American
put must satisfy the Black and Scholes (1973) or Merton’s (1973) partial

1 2 2 ∂ 2P ∂P ∂P
σ S + rS + − rP = 0
2 ∂ 2S ∂S ∂t
under the following boundary conditions:
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
infinite time to maturity, a closed-form solution is given by Merton (1973),

(Eq. (52), p. 174). The structure of this problem implies the existence of
a certain level of the underlying asset I(t), referred to as the critical asset
price level, for which the exercise depends on the maximization behavior of
the put’s holder. The determination of this critical level implies adding the
following condition on the put’s derivative with respect to the underlying
∗
asset ∂P (I ∂t
, ∞, K)
= −1. This condition is the Samuelson’s “high contact”
boundary condition. The analysis of this problem (the free boundary
problem) by Samuelson and McKean (for details, refer to Bellalah et al.,
1998) and Merton (1973) allows the derivation of the put’s value when the
time to maturity is infinite.
Early exercise with continuous distributions

When there are continuous distributions to the underlying asset, the pricing
of American puts is rather difficult. However, an interesting approach was
proposed by Barone-Adesi and Whaley (1987) for the valuation of American
options. This approach is referred to as the quadratic approximation
method.
Early exercise with discrete distributions

When there are no dividends, the American put option may be exercised
early. In fact, since interest income can be earned on the exercisable
proceeds of the option when exercised, this is sufficient to justify early
exercise. When the American option holder delays exercise, this means that
he forgoes the interest income on the exercisable proceeds.
When there are dividends, the American put option holder must
compare the effect of dividends on the put value and the interest income.
If the holder does not exercise his put, he forgoes the interest income. If he
exercises before the ex-dividend date, he will not profit from the increase
in the exercisable proceeds when the underlying stock goes ex-dividend.
There is a trade-off between the interest income and the dividend. Hence,
between dividend dates, the option holder is always in a dilemma. Some
interesting approximations of American values are given in the literature.
These approximations include, Johnson (1983), Geske and Johnson (1984),
Geske and Shastri (1985), Blomeyer (1986), and Barone-Adesi and Whaley
(for details, refer to Bellalah et al., 1998) among others. However, most of
them work less well than binomial models and finite difference methods.
14.1.3. The American put option and its critical

stock price
We provide an expression for the critical stock price for the American put.
The put price is expressed in an integral form which involves first passage
probabilities. Using the fact that:
• 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.
The first-passage approach and the perpetual put option case

The American put price P is exercised when the stock price S hits the
critical stock price Sc . When the stock price and its critical value have no
discontinuities, we can write:
T
P = max e−rt (K − Sc )f dt, (S Sc ) (14.10)
Sc 0
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
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
When γ = 1, a is constant: a = σ1 log(S/Sc )

Using the Laplace transform Table of Abramowitz and Stegun (1972,
p. 1026), the result is
∞
k 2 √
√ e−k /(4t) e−st dt = e−k s .
0 2πt 3
When s = r and k 2 = a2 , we have:

√ S
P = (K − Sc ) exp −( 2r/σ) log
Sc

S
P = (K − Sc ) (14.11)
Sc
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.
When γ = 1, by substituting Eq. (14.11) into:
∂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
Since in Eq. (14.11) ∂P 2 ∂P

∂S = −(K − Sc )Sc /S , we have ∂S = −(Sc /S) .
2
Equation (14.14) shows that the American put price corresponds to

the exercise price times the expected discount factor plus the stock price
times the hedge ratio. This is similar to a European put given in Black and
Scholes (1973) as:
p = Ke−rT N (−d2 ) − SN (−d1 ). (14.15)
It is possible to obtain expressions when γ is different from one. Bunch and

Johnson (2000) show that:
γ
Sc
P = (K − Sc ) (14.16)
S
Equations (14.11) and (14.16) do not contain normal density functions.
This is the exact result and the exact density is
2
log SSc e−a /(2t)
f (t) = √ .
2πσt3/2
Now, we consider the more difficult case of finite-lived puts.
The critical stock price function for a finite time to maturity

When the time to maturity is finite, the critical stock price is no longer
constant. The first-passage probabilities may be estimated using the tangent
approximation:
log SS + SṠc t −a2 /(2t)

f= √ c e
2πσt3/2
where the dot indicates the time derivative. The variables Sc and a are
general functions of t. Bunch and Johnson (2000) derive an expression for
the critical stock price using the tangent approximation:

√
γ e−β2 τ −α1 τ
Sc /X = +
1+γ 1+γ
where
σa a2 3 ar
β2 = + 2 −r− −
2t 2t 2t σt
and α1 is a positive constant.
When S = Sc , we have ∂P ∂τ
= 0.
It is possible to use the equation providing a relationship between the
American put, P and the European put p as:
T
P =p+ rKe−rtN (−d2 (S, Sc , t))dt
0
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
The exact expression can be obtained when g = 0 in Eq. (14.18) to:

√
2 α
2 log γ(1+γ)σ 2τ
τ0 = 2
α
0

σ (1 + γ) 1 − 4 (1 + γ)
This equation can be solved iteratively. Bunch and Johnson (2000)

shows that
2
A 1 γ
α=1− , where A = .
1 + (1+γ)
2
γ 2τ 2 1+γ
4
The approximation for g corresponding to a small τ is:

σ2
g = log (14.21)
4er2 τ /α
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.
14.2. Valuation of American Commodity Options

and Futures Options with Continuous
Distributions
Most options written on commodities and commodity futures are of
the American type. Hence, an early exercise premium is embedded
in American call and put prices. Analytical solutions for the Ameri-
can option pricing problems with several dividends have not yet been
found.
The quadratic approximation method used by Barone–Adesi and
Whaley (1987), is computationally efficient. It provides an accurate,
inexpensive method for the valuation of American call and put options
traded on commodities and commodity futures.
14.2.1. Valuation of American commodity options

The following analysis applies to commodity options for which the cost
of carrying the underlying commodity is a constant proportional rate, to
commodity futures options and to stock options. The model used here is
an extension of the Merton (1973) and Barone–Adesi and Whaley’s (1987)

model to American options. The pricing of American options involves the
valuation of the early exercise premium attached to the possibility of early
exercise. It is possible to show that, under certain conditions, American
options should be exercised early. In fact, when the cost of carrying the
underlying commodity, b, is less than the risk-less interest rate of interest
r, and the option becomes deep-in-the money, the values of N (d1 ) and
N (d2 ) approach 1. The European call value tends toward the quantity
Se(b−r)T − Ke−rT .
Since an American option may be exercised for its intrinsic value,
(S − K), this value may be higher than Se(b−r)T − Ke−rT .
When b = r, there is no possibility of early exercise for an American
call option. This argument does not hold for American put options since
for puts, as shown by Merton (1973), there is always some probability of
early exercise.
Since the cost of carrying any futures position is nil, i.e., b = 0, then
all formulas proposed here can be used to price commodity futures options
by substituting the futures price F for the commodity price S and setting
b = 0. Also, when the option’s underlying asset is a non-dividend paying
stock, b = r, and the formulas can be applied to stock options.
European and American option values must satisfy the following partial
differential equation,
1 2 2 ∂ 2 C(S, t) ∂C(S, t) ∂C(S, t)

σ S + bS + − rC(S, t) = 0. (14.22)
2 ∂ 2S ∂S ∂t
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:
1 2 2 ∂ 2 (S, t) ∂(S, t) ∂(S, t)

σ S + bS + − r(S, t) = 0. (14.23)
2 ∂2S ∂S ∂t
2r 2b
Let M = σ2 , N = σ2 , τ = T − t. Multiplying Eq. (14.23) by ( σ22 ) gives:
∂ 2 (S, t) ∂(S, t) M ∂(S, t)

S2 + NS − − M (S, t) = 0. (14.24)
∂ 2S ∂S r ∂τ
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
When substituting into Eq. (14.24), we obtain:

∂2f ∂f kτ k ∂f
S2 + NS − Mf 1 + 1+ = 0.
∂ 2S ∂S rk f ∂k
If you choose k(τ ) = 1 − e−rτ , this equation becomes,

∂ 2f ∂f Mf ∂f
S2 + NS − − M (1 − k) = 0. (14.25)
∂2S ∂S k ∂k
Now, the quadratic approximation

is applied to Eq. (14.25) in order ∂fto
eliminate the term M (1 − k) ∂f∂k
. In fact, as τ approaches 0, (∞), ∂k
approaches 0 (k approaches 1). Hence, Eq. (14.25) becomes a second-order
differential equation

2 ∂ 2f ∂f Mf
S + NS − = 0.
∂2S ∂S k
The above equation presents two linearly independent solutions of the

form f (S) = a1 S q1 + a2 S q2 .
Solving this equation, the American call option value presented by
Barone-Adesi and Whaley (1987) is:
q2
S
C(S, T ) = c(S, T ) + A2 when S < S ∗
S∗
C(S, T ) = S − K when S ≥ S ∗
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
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
S ∗ [1 − e(b−r)T N (−d1 (S ∗ ))]

K − S ∗ = p(S ∗ , T ) − .
q1

Tables 14.1 and 14.2 provide the values of European and American call
options and the critical underlying asset price level for different parameters.
Tables 14.3 and 14.4 provide the values of European and American
put options and the critical underlying asset price level for different
parameters.
Table 14.1. European and American

call option prices. K = 100, r = 0.08,
T = 0.25, b = 0.04, and σ = 0.2.
S c C S∗
90 0.84 0.849 214.952

100 4.439 4.431 214.952
110 11.451 11.662 214.952
120 20.891 20.898 214.952

call option prices. K = 100, r = 0.08,
T = 0.25, b = 0.08, and σ = 0.2.
S c C S∗
90 0.698 0.705 120.323

100 3.909 3.934 120.323
110 10.737 10.823 120.323
120 19.748 20.009 120.323

put option prices. S = 100, r = 0.08,
T = 0.25, b = −0.04, and σ = 0.2.
S p P S∗
90 9.765 10.180 87.438

100 3.455 3.544 87.438
110 0.777 0.798 87.438
120 0.112 0.118 87.438

put option prices. K = 100, r = 0.08,
T = 0.25, b = 0.08, and σ = 0.2.
S p P S∗
90 10.500 10.565 83.501

100 3.909 3.921 83.501
110 0.935 0.938 83.501
120 0.144 0.145 83.501
14.2.3. Valuation of American futures options

From the analysis by Barone-Adesi and Whaley (1987), American futures
call option formula when there are no carrying costs is:
C(F, T ) = c(F, T ) + A2 (F/F ∗ )q2 when F < F ∗

C(F, T ) = F − K when F ≥ F ∗
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
When there are carrying costs, the formula becomes:
C(F, T ) = c(F, T ) + A2 (F/F ∗ )q2 when F < F ∗

C(F, T ) = F − K when F ≥ F ∗
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).
In the absence of carrying costs, the formula for the American futures
put is:
P (F, T ) = p(F, T ) + A1 (F/F ∗ )q1 when F > F ∗

P (F, T ) = K − F when F ≤ F ∗
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:
P (F, T ) = p(F, T ) + A1 (F/F ∗ )q1 when F > F ∗

P (F, T ) = K − F when F ≤ F ∗
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
The critical futures price corresponding to an optimal early exercise is

calculated using the following equation:

∗ ∗ F ∗ 1 − e(b−r)T N (−d1 (F ∗ ))
K − F = p(F , T ) − .
q1
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 ).
Table 14.5. European and American futures call prices.

K = 100, r = 0.08, T = 0.25, b = 0, and σ = 0.2.
S N (d1 ) N (d2 ) c C
90 0.1624 0.1417 0.4360 0.4402

100 0.5199 0.4801 3.900 3.9251
110 0.8428 0.8173 10.7600 10.8380
120 0.9710 0.9635 19.7750 20.0250
Table 14.6. European and American futures put prices.

K = 100, r = 0.08, T = 0.25, b = 0, and σ = 0.2.
S N (−d1 ) N (−d2 ) p P
90 0.8522 0.8375 10.2300 10.6500

100 0.5199 0.4801 3.9087 4.0199
110 0.1826 0.1571 0.9607 0.9807
120 0.0364 0.0289 0.1714 0.1795

Tables 14.5 and 14.6 provide the values of European and American call and
put futures options for different parameters.
The difference between European and American call option prices
correspond to the early exercise premium attached to American options.
14.3. Valuation of American Commodity and Futures

Options with Continuous Distributions
within Information Uncertainty
This section is devoted to the valuation of American commodity options
within information uncertainty.
14.3.1. Commodity option valuation with information costs

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 of Merton (1987),
CAPMI, are constant through 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 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:
C(S, T ) = Se((b−r−(λC −λS ))T ) N (d1 ) − Ke−(r+λC )T N (d2 )
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
1 In fact, to obtain this equation, the CAPMI can be written as:

ˆ ˜
R̄S − r − λS = βS R̄m − r − λm ,
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:
E(∆C) = (r + λC )C∆t + aSβS CS ∆t.

commodity is zero, this equation is equivalent to that in Black. Using the

futures price F = Se(b+λS )T instead of the spot price, the equation becomes
for a European futures call as:
C(F, T ) = e−(r+λC )T (F N (d1 ) − KN (d2 ))
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:
P (F, T ) = e−(r+λC )T (−F N (−d1 ) + KN (−d2 ))
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
Taking the expected value b and replacing E(∆S) gives:

1
E(∆C) = (r + λS )SCS ∆t + aSβS CS ∆t + Ct ∆t + CSS S 2 σ2 ∆t.
2
Making the equality between this equation, and
E(∆C) = (r + λC )C∆t + aSβS CS ∆t,
and simplifying gives:

1
CSS S 2 σ2 + (r + λS )SCS − (r + λC )C + Ct = 0.
2
If λS = λC = 0, this equation is the Black and Scholes equation.
2 Itis possible to use the previous footnote to see this result. In fact, 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) = (r + λS )SCS ∆t + aSβS CS ∆t + Ct ∆t + CSS S 2 σ2 ∆t
2
E(∆C) = (r + λC )C∆t + aSβS CS ∆t,
This valuation equation applies to forward contracts, European, and

American options. The American commodity option value CA (S, T ) is
given by:
q2
S
CA (S, T ) = C(S, T ) + A2 when S < S ∗
S∗
CA (S, T ) = S − K when S ≥ S ∗
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

from the following equation:

∗ ∗ S ∗ 1 − e(b+λS −r−λC )T N (d1 (S ∗ ))
S −K =C (S , T ) + .
q2
The American commodity option value PA (S, T ) is:

q1
S
PA (S, T ) = P (S, T ) + A1 when S > S ∗∗
S∗
PA (S, T ) = K − S when S ≤ S ∗∗
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.
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

from the following equation3 :
S ∗ (1 − e(b+λS −r−λC )T N (−d1 (S ∗ )))

K − S ∗ = P (S ∗ , T ) − .
q1
A similar algorithm as the one developed by Barone-Adesi and Whaley

(1987) can be used to determine the critical underlying price.
14.3.2. Simulation results

Table 14.7 presents a sensitivity analysis of the theoretical European futures
call and put values as given by Black’s model and the model derived in the
previous section. The calculations are based on two alternative assumptions
regarding information costs, 1% and 5%. When F = K = 100, T = 0.25,
σ = 0.2, r = 0.08, and b = 0, results show that an increase in information
costs from 1% to 5% reduces call values respectively from 3.8990 to 3.8602.
The simulations show that model prices are less than Black’s prices for out-
of-the-money and in-the-money calls and are nearly equal to Black’s prices
for at-the-money calls. This result shows that our model explains at least
the relative mispricing of out and at-the-money calls, which is an advantage
with respect to Black’s model. It is important to note that the differences in
option values increase with the interaction between information costs and
the other option valuation parameters.
When the information costs vary from 1% to 5%, a decrease is observed
in put values. The simulations performed indicate that model prices are
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.
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
80 0 0 0 19.6611 19.6203 19.4549

90 0.6963 0.6353 0.6310 10.4365 10.3128 10.3112
100 3.9087 3.8990 3.8602 3.9087 3.8990 3.8602
110 10.7427 10.7358 10.6290 0.9487 0.9483 0.9482
120 19.7554 19.7260 19.5297 0.1414 0.1370 0.1363
Panel B T = 0.25 σ = 0.2 r = 0.12
80 0 0 0 19.4461 19.4249 19.2912

90 0.6820 0.6810 0.6767 10.4356 10.4112 10.3106
100 3.8698 3.8602 3.8217 3.8698 3.8602 3.8217
110 10.6356 10.6290 10.5237 0.9412 0.9408 0.9393
120 19.5487 19.5297 19.3354 0.1497 0.1493 0.1476
Panel C T = 0.5 σ = 0.2 r = 0.08
80 0 0 0 18.1417 18.0566 17.6991

90 1.6913 1.6833 1.6520 11.3091 11.2433 11.1226
100 5.4161 5.3890 5.2823 5.4161 5.3870 5.2823
110 11.7325 11.6939 11.4235 2.1246 2.1139 2.0916
120 19.9157 19.8562 19.4630 0.6999 0.6961 0.6916
Panel D T = 0.5 σ = 0.4 r = 0.08
80 2.6940 2.6806 2.6275 21.9098 21.8005 21.3688

90 6.1331 6.1025 5.9817 15.7410 15.6625 15.3524
100 10.8051 10.7512 10.5383 10.8151 10.7512 10.5383
110 16.7917 16.7080 16.6123 7.1838 7.1480 7.0165
120 19.9557 19.8562 19.4630 4.6800 4.6567 4.5645
nearly equivalent to Black’s prices for at and out-of-the money puts.

However, model prices are less than those reported in Black’s model for in-
the-money puts. Since our model eliminates the overvaluation bias of these
puts, this is an advantage with respect to Black’s model. The calculations
performed are based on three alternative assumptions regarding information
costs for European and American call options when the futures price varies
from 90 to 120. Table 14.8 shows that our model prices are most of the time
less or equal to Barone-Adesi and Whaley’s model prices for call options.
Table 14.9 provides the theoretical values for European and American
futures put options.
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.
Panel, A, F CEuropean CAmerican FCritical CBAW
90 0.6950 0.7407 121.770 0.70

100 3.8000 3.9380 121.758 3.93
110 10.7427 10.8158 121.778 10.81
120 19.7500 20.0270 121.770 20.02
Panel B T = 0.25 σ = 0.2 r = 0.08 λC = 0.02
90 0.6340 0.6419 121.629 0.70

100 3.8800 3.9201 121.612 3.93
110 10.7000 10.8186 121.629 10.81
120 19.6700 20.0230 121.624 20.02
Panel C T = 0.25 σ = 0.2 r = 0.08 λC = 0.05
90 0.6310 0.6478 121.3529 0.70

100 3.8602 3.910 121.3110 3.93
110 10.6290 10.8056 121.3244 10.81
120 19.5297 20.0183 121.3240 20.02
Panel D T = 0.5 σ = 0.2 r = 0.08 λC = 0.01
90 1.5830 1.61087 128.240 1.72

100 5.3800 5.46840 128.290 5.48
110 11.6930 11.9060 128.136 11.90
120 19.8560 20.3413 121.139 20.34
Panel E T = 0.5 σ = 0.2 r = 0.08 λC = 0.02
90 1.5750 1.61220 127.920 1.72

100 5.3600 5.4639 127.929 5.48
110 11.6300 11.8896 127.935 11.90
120 19.7500 20.3423 127.922 20.34
Panel F T = 0.5 σ = 0.2 r = 0.08 λC = 0.05
90 1.6520 1.6265 127.5502 1.72

100 5.2823 5.4647 127.5533 5.48
110 11.4235 11.8724 127.5430 11.90
120 19.4630 20.3218 127.5470 20.34
14.4. Valuation of American Options with Discrete

Cash-Distributions
Most of the traded stock and index options around the world may be exer-
cised before their expiration dates and are therefore of the American type.
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.
Panel, A, F PEuropean PAmerican FCritical PBAW
90 10.3128 10.6440 86.780 10.58

100 3.8990 4.0130 86.132 3.93
110 0.9584 0.9850 86.770 0.94
120 0.1710 0.1800 86.710 0.15
Panel B T = 0.25 σ = 0.2 r = 0.08 λC = 0.02
90 10.1870 10.6310 86.821 10.58

100 3.8890 4.00600 86.190 3.93
110 0.9450 0.98420 86.817 0.94
120 0.1370 0.17950 86.756 0.15
Panel C T = 0.25 σ = 0.2 r = 0.08 λC = 0.05
90 10.3112 10.6050 86.9420 10.58

100 3.8602 3.99600 86.3589 3.93
110 0.9482 0.98000 86.940 0.94
120 0.1363 0.17900 86.889 0.15
Panel D T = 0.5 σ = 0.2 r = 0.08 λC = 0.01
90 11.2433 11.7240 82.251 11.48

100 5.3890 5.5846 82.144 5.48
110 2.1139 2.21400 82.190 2.15
120 0.6961 0.7757 82.228 0.70
Panel E T = 0.5 σ = 0.2 r = 0.08 λC = 0.02
90 11.0870 11.69200 82.3570 11.48

100 5.3620 5.5656 82.2556 5.48
110 2.1232 2.20680 82.299 2.15
120 0.7325 0.77000 82.336 0.70
Panel F T = 0.5 σ = 0.2 r = 0.08 λC = 0.05
90 11.1226 11.6400 82.6640 11.48

100 5.2823 5.5209 82.5700 5.48
110 2.0916 2.1843 82.614 2.15
120 0.6916 0.77100 82.640 0.70
14.4.1. Early exercise of American options

Early exercise of these options is often induced by the payment of dividends
on the underlying asset. For stock options, it may be optimal to exercise
a call on the last cum-dividend day to receive the dividend. In this case,
the adapted versions of the Black–Scholes model proposed by Roll (1977),

Geske (1979), and Whaley (1981) can be used.
For an index option, when the dividends are distributed fairly evenly
over time, B–S type models may be used. However, when the dividends
tend to be clustered, early exercise may be triggered at many points in
time and B–S type models can not be applied. Another potential dividend-
induced reason of early exercise is due to asset carrying costs. When these
costs are different from the risk-less interest rate, early exercise is also
possible. Hence, for many reasons, the B–S model does not price adequately
American commodity and futures options, since by definition, the model is
“reserved” to European options on non-dividend paying stocks.
Harvey and Whaley (1992) showed that ad-hoc valuation procedures
sometimes produce large pricing errors because of the discrete and seasonal
pattern of dividends on some indexes like the S&P 100 index portfolio.
The most commonly used methods are the B–S model adjusted for
dividends because of its ease of computation and the American-style option
pricing approximations with a constant proportional dividend yield. These
include the quadratic approximation of Barone-Adesi and Whaley (1987)
and the Cox et al. (1979) binomial method under the assumption of a
constant proportional dividend yield rate. The appropriate approach for
the valuation of American options when there are discrete dividends is
based on the compound option approach. Several models are proposed in
this context. Given the contribution of the compound option approach to
the pricing of American options, this approach is presented in detail.
14.4.2. Valuation of American options with dividends

The valuation of American stock options is equivalent to the valuation
of a portfolio which contains a commodity option on a stock with some
distributions. These distributions are reflected in the assumed proportional
rate of carrying the stock, y, and the amount of cash dividend paid once a
year. For the sake of simplicity, let us use the context in Roll (1977), Geske
(1979), and Whaley (1981) and assume that:
• Investors can borrow and lend without restrictions at the short-term

instantaneous risk-less rate r during the option’s life T .
• The stock pays a cash income at some dates ti , (ti < T ).
• The stock pays a dividend D and the ex-dividend instant is t,
(t < T ).
• The stock price net of the discounted dividend, S, is given by:
Sτ = Pτ − De−r(t−τ ) for τ < t

Sτ = Pτ for τ ≥ t
where P is the stock price cum-dividend. The price dynamics of the stock
S are given by the stochastic differential equation:
dS/S = µdt + σdz
where µ and σ stand respectively for the instantaneous expected rate of

return and the standard deviation of the stock price return and dz is
a standard Wiener process. We can show that there exists some finite
ex-dividend stock price St above (under) which the early exercise of an
American call (put) may be optimal. We call this value of St , St∗ , which
is easily calculated by an iterative procedure. Once this critical value is
determined, the valuation by duplication technique can be implemented.
Consider the following portfolio of options:
(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 − ).
The contingent payoff of this portfolio of options is identical to that of an

American call. In a perfect capital market, the absence of costless arbitrage
opportunities ensures that the American call value is identical to that of
this portfolio. The American call value must be equal to the algebraic sum
of the three options in the portfolio. The option described in (a), Ca , can
be valued using the commodity option formula. Its value is given by:
ca = Se(y−r)T N1 (a1 ) − Ke−rT N1 (a2 )

S 1 √ √
a1 = ln + y + σ2 T σ T , a2 = a1 − σ T
K 2
where N1 (·) stands for the cumulative normal distribution.

The option described in (b/), Cb can be priced using an extension

of Merton’s (1973) formula for which the strike price is St∗ . Its value is
given by:
Cb = Se(y−r)t N (b1 ) − St∗ e−rt N (b2 )

∗ 1 2 √ √
b1 = ln(S/St ) + y + σ t σ t, b2 = b1 − σ t
2
where St∗ is the solution to the following equation,
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
Where N2 (. . .) is the bivariate cumulative normal density function with

upper integral limits a and b and a correlation coefficient ρ. Since the value
of the American call is equivalent to the algebraic sum of the three options
in the portfolio, we have: C = ca + cb − cc .
Using the properties of the bivariate cumulative normal density func-
tion, little algebra allows the derivation of the following formula for the
American call on a stock,

t
C = S e(y−r)t N1 (b1 ) + e(y−r)T N2 a1 , −b1 , −
T

−rt −rT t
− K e N1 (b2 ) + e N2 a2 , −b2 , − + De−rtN1 (b2 )
T
Table 14.10. American call prices in the presence of

dividends.
S Scr Scr C(S, T, K)
82 77.196 123.5818 3.8050

85 80.196 123.5818 4.8175
87 82.196 123.5818 5.5758
90 85.196 123.5818 6.8389
92 87.196 123.5818 7.7645
95 90.196 123.5818 9.2759
97 92.196 123.5818 10.3636
100 95.196 123.5818 12.1113
102 97.196 123.5818 13.3506
105 100.196 123.5818 15.3155
107 102.196 123.5818 16.6922
110 105.196 123.5818 18.8509
112 107.196 123.5818 20.3486
115 110.196 123.5818 22.6759
117 112.196 123.5818 24.2774
120 115.196 123.5818 26.7476
122 117.196 123.5818 28.4363
125 120.196 123.5818 31.0255
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.
14.5. Valuation of American Options with Discrete Cash

Distributions within Information Uncertainty
14.5.1. The model
We use a similar context as that in Roll (1977), Geske (1979), and Whaley
(1981) and assume besides that the stock price net of the discounted
dividend, S, is:
Sτ = Pτ − De−(r+λS )(t−τ ) for τ < t

Sτ = Pτ for τ ≥ t
where P is the stock price cum-dividend and λS is the information cost

regarding the underlying asset price.
If we account for information costs as given by Bellalah (1999), then
the cost of carry model is F = Se(y+λS )T where F is the current forward
price. The option Ca is given by:
ca = Se((y−r−(λC −λS ))T N1 (a1 ) − Ke−(r+λC )T N1 (a2 )

S 1 √
a1 = ln + y + λS + σ2 T σ T
K 2
√
a2 = a1 − σ T
where λC stands for information cost on the option market.

The option described in (b/), Cb can be priced using Bellalah’s (1999)
formula:
Cb = Se(y−r−(λC −λS ))t N (b1 ) − St∗ e−(r+λC )t N (b2 )

S 1 2 √
b1 = ln ∗ + y + σ + λ S t σ t,
St 2
√
b2 = b1 − σ t
where St∗ is the solution to the following equation:
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
where N2 (. . .) is the bivariate cumulative normal density function with

upper integral limits a and b and a correlation coefficient ρ.
The American call is given by:

((y−r−(λC −λS ))t ((y−r−(λC −λS ))T t
C=S e N1 (b1 ) + e N2 a1 , −b1 , −
T

t
− K e−(r+λC )t N1 (b2 ) + e−(r+λC )T N2 a2 , −b2 , −
T
+ De−(r+λC )t N1 (b2 ).

Table 14.11 gives the computation of the American call value referred to
as call, the option ca , the option cb , the option cc , the CRR price, the
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.
Strike Call ca cb cc CRR ca + cb − cc S∗ Call-CRR
100 76.03 74.42 74.25 72.65 75.81 76.03 100.02 0.22

105 71.06 69.46 69.11 67.51 70.85 71.06 105 0.21
110 66.09 64.51 63.96 62.37 65.89 66.09 110 0.20
115 61.13 59.55 59.07 57.49 60.93 61.13 115 0.19
120 56.16 54.59 53.65 52.07 55.97 56.16 120 0.19
125 51.19 49.63 48.48 46.92 51.01 51.19 125 0.18
130 46.22 44.67 43.32 41.76 46.05 46.23 130 0.17
135 41.26 39.72 38.49 36.94 41.09 41.26 135 0.16
140 36.30 34.79 32.99 31.47 36.16 36.30 140 0.14
145 31.37 29.90 27.87 26.39 31.25 31.37 145 0.11
150 26.50 25.11 22.85 21.46 26.42 26.50 150 0.07
155 21.78 20.52 18.05 16.80 21.77 21.78 154.99 0.01
160 17.31 16.23 13.64 12.56 17.37 17.31 159.99 −0.05
165 13.25 12.38 9.78 8.92 13.39 13.25 164.99 −0.13
170 9.72 9.07 6.63 5.98 9.91 9.72 169.99 −0.19
175 6.82 6.37 4.22 3.77 7.03 6.82 174.99 −0.20
180 4.56 4.27 2.52 2.23 4.77 4.56 179.99 −0.21
185 2.91 2.74 1.41 1.24 3.10 2.91 184.99 −0.18
190 1.77 1.68 0.74 0.64 1.92 1.77 189.99 −0.14
195 1.03 0.98 0.36 0.31 1.13 1.03 194.99 −0.10
200 0.57 0.55 0.16 0.14 0.64 0.57 200 −0.06
240 0 0 0 0 0 0 240 −0
t = 24, σ = 0.32, λc = 0.001, and λs = 0.01.
100 76.14 74.56 74.34 72.76 75.94 76.14 100.02 0.19

120 56.427 54.73 53.73 52.19 56.11 56.27 120 0.16
140 36.41 34.93 33.06 31.57 36.30 36.41 140.01 0.11
145 31.48 30.04 27.94 26.49 31.39 31.48 145 0.09
150 26.61 25.25 22.92 21.55 26.56 26.61 150 0.05
155 21.89 20.65 18.11 16.87 21.90 21.89 154.99 −0.01
160 17.42 16.35 13.67 12.61 17.49 17.42 159.99 −0.07
165 13.34 12.49 9.83 8.97 13.50 13.34 164.99 −0.15
170 9.80 9.16 6.66 6.02 10.01 9.80 169.99 −0.34
175 6.89 6.44 4.24 3.80 7.10 6.89 174.99 −0.21
180 4.62 4.33 2.53 2.25 4.84 4.62 179.99 −0.22
185 2.95 2.78 1.42 1.25 3.14 2.95 184.99 −0.19
190 1.80 1.71 0.74 0.65 1.95 1.80 189.99 −0.14
195 1.05 1 0.36 0.32 1.16 1.05 194.99 −0.10
200 0.58 0.56 0.17 0.14 0.65 0.58 200 −0.06
240 0 0 0 0 0 0 240 −0
algebric sum of the three options (ca + cb − cc ), the critical underlying

asset price, and the difference between our call formula and the benchmark.
Table 14.12 uses the same data except for information costs. Information
costs are set equal to λS = 0.01 and λC = 0.001. Table 14.13 uses the
same parameters except for the information costs which are set equal to
λS = 0.1 and λC = 0.05. The results reveal that the difference between
our model prices and the CRR prices are very small. Information costs
offer a simple way to calibrate model prices to market data as shown in
Bellalah and Jacquillat (1995) and Bellalah (1999, 2001). If we invert the
information costs as in Table 14.14, we see that the mispricing bias in
inverted. Table 14.14 gives the results for the different parameter values,
except for information costs, which are set respectively equal to λS = 0.05
and λC = 0.1.
14.6. The Valuation Equations for Standard

and Compound Options with Information Costs
We denote by S(V, t) the value of the option as a function of the underlying
asset and time.
t = 24, σ = 0.32, λc = 0.05, and λs = 0.1.
100 76.92 75.55 74.95 73.58 76.92 76.92 100.02 −0

120 57.12 55.79 54.31 52.98 57.17 57.12 120 −0.05
140 37.32 36.06 33.59 32.33 37.43 37.32 140.01 −0.10
145 32.40 31.18 28.45 27.23 32.53 32.40 145 −0.13
150 27.54 26.38 23.40 22.25 27.70 27.54 150 −0.16
155 22.80 21.75 18.55 17.51 23.01 22.80 154.99 −0.21
160 18.29 17.39 14.08 13.18 18.56 18.29 159.99 −0.26
165 14.15 13.43 10.15 9.42 14.47 14.15 164.99 −0.32
170 10.52 9.97 6.90 6.36 10.86 10.52 169.99 −0.34
175 7.49 7.11 4.41 4.04 7.81 7.49 174.99 −0.32
180 5.09 4.85 2.65 2.41 5.40 5.09 179.99 −0.30
185 3.31 3.16 1.49 1.34 3.56 3.31 184.99 −0.25
190 2.05 1.97 0.78 0.70 2.25 2.05 189.99 −0.19
195 1.22 1.18 0.38 0.34 1.35 1.22 194.99 −0.13
200 0.69 0.67 0.18 0.16 0.78 0.69 200 −0.08
240 0 0 0 0 0 0 240 −0
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.
100 76.10 74.52 74.25 72.67 75.90 76.10 100.02 0.20

120 56.36 54.85 53.73 52.22 56.22 56.36 120 0.13
140 36.63 35.21 33.15 31.72 36.57 36.63 140.01 0.06
145 31.73 30.35 28.03 26.66 31.70 31.73 145 0.03
150 26.89 25.59 23.04 21.74 26.90 26.89 150 −0.01
155 22.18 21 18.24 17.07 22.25 22.18 154.99 −0.07
160 17.71 16.70 13.82 12.81 17.85 17.71 159.99 −0.13
165 13.63 12.82 9.93 9.12 13.84 13.63 164.99 −0.21
170 10.07 9.46 6.75 6.15 10.32 10.07 169.99 −0.25
175 7.11 6.69 4.31 3.89 7.37 7.11 174.99 −0.25
180 4.80 4.53 2.58 2.31 5.05 4.80 179.99 −0.24
185 3.09 2.93 1.45 1.29 3.30 3.09 184.99 −0.21
190 1.90 1.81 0.76 0.67 2.07 1.90 189.99 −0.16
195 1.12 1.07 0.37 0.33 1.23 1.12 194.99 −0.11
200 0.63 0.60 0.17 0.15 0.70 0.63 200 −0.07
240 0 0 0 0 0 0 240 −0
14.6.1. The pricing of assets under incomplete information

Using the same assumptions as in the seminal paper of Black and Scholes
and the additional assumption of information uncertainty on the option
S(V, t) and its underlying asset, V , it is possible to construct a hedged
position which contains one share of stock long and S1V options short. The
term SV is the first derivative of the option with respect to its underlying
asset. Since the return in the hedged position is certain, the return must
be equal to (r + λV )∆t for the underlying asset and (r + λS )∆t for the
option where λV and λS refer respectively to the information costs on the
underlying asset and the option. The differential equation for the value of
the option:
1 2 2
σ V SV V + (r + λV )V SV − (r + λS )S + St = 0. (14.26)
2
Let T be the maturity date of the call and M be its strike price.
Equation (14.26) subject to the following boundary condition at maturity:
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:
S(V, T ) = V e−(λS −λV )T N1 (d1 ) − M e−(r+λS )T N1 (d2 ) (14.27)
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.
14.6.2. The valuation of equity as a compound option

Following Geske (1979), consider a levered firm for which the debt
corresponds to pure discount bonds maturing in T years with a face
value M . Under the standard assumptions of liquidating the firm in T
years, payingoff 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. The return on the firm’s assets follows the stochastic
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 ).
It is convenient to note that Eq. (14.28) is more difficult to solve than

Eq. (14.26) because its boundary condition is a function of the solution
to Eq. (14.26). At date T0 , the value of the firm making the holder of
the call on the stock indifferent with regard to the exercise decision is
solution to ST0 − K = 0 where ST0 is given by the last formula. Following
the methodology in Geske (1979), the compound call option value with
information costs is:

√ T0
−(λc −λv )T0 −(λs −λv )(T −T0 )
C0 = V0 e e N2 h + σv T0 , k + σv T ,
T

T0
− M e−(r+λs)T e−(λc −λs )(T0 ) N2 h, k, . − Ke−(r+λc )T0 N1 (h).
T
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
The value V̄ is determined by the following equation:

ST0 − K = V̄ e−(λs −λv )(T −T0 ) N1 (k(V̄ ) + σv T − T0 )
− M e−(r+λs)(T −T0 ) N1 (k(V̄ )) − K =
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
The value V̄ is determined by the following equation:

ST0 − K = V̄ e−(λc −λv )(T −T0 ) N1 (k + σv T − T0 )
− M e−(r+λc)(T −T0 ) N1 (k) − K = 0.
A second special case of the compound option formula is obtained when

the incurred information costs are equal to a same value λ for the stock,
the firm’s value and the call. In this context, the compound option formula
becomes:

√ T0
C0 = V0 N2 h + σv T0 , k + σv T ,
T

−(r+λ)T T0
− Me N2 h, k, − Ke−(r+λ)T0 N1 (h)
T
where the value V̄ is determined from the following equation:

ST0 − K = V̄ N1 k + σv T − T0 − M e−(r+λ)(T −T0 ) N1 (k) − K = 0.
If the information cost is zero, this compound option pricing formula

becomes the one as given by Geske (1979). The following tables present
the simulation results of the above models. Table 14.15 gives the call equity
values using the Black and Scholes model and our model. The following
parameters are used for the simulations: M = 100, r = 0.08, T = 0.25,
and σv = 0.4. The following information costs are used to generate option
values: (λs = λv = 0) which correspond to the Black and Scholes case, and
(λs = 0.1%, λv = 1%), (λs = 0%, λv = 1%), and (λs = 0.1%, λv = 3%).
The results show that in all cases our option values are less than those
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.
λs = 0.1%, λs = 0%, λs = 0.1%,

V λs = λv = 0 λv = 1% λv = 1% λv = 3%
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
Table 14.16. Simulation and comparisons of European equity values as compound

options in the presence of information costs using Geske’s model and our model.
K = 20, M = 100, r = 0.08, T = 0.25, T0 = 0.125, and σv = 0.4.
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%
110 6.82 7.13 7.16 7.14

120 15.17 15.65 15.70 15.67
130 26.52 27.16 27.25 27.20
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
European and American call and put options on a variety of underlying

assets and commodities including common stocks, bonds, agricultural
futures contracts, and financial futures, there is still no-known analytical
solutions for options written on commodities that have known discrete
cash payments during the option’s life. Bunch and Johnson (2000) use
the first-passage probability approach to derive exact expressions for the
critical stock price function and for the American put price. The analysis
considers the infinite or perpetual put and the finite case. The derivation
is based on the fact that exercise occurs when the interest rate effect is
exactly offset by the volatility effect. A similar result can be obtained,
if we account for the effects of shadow costs of incomplete information
since these costs are added to the interest rate. The method in Bunch and
Johnson (2000) can be extended to currency options, options on futures, to
exotic options, etc. This chapter presents in detail the basic concepts and
techniques underlying rational pricing of American options. This is done in
the context of analytical European models along the lines of Black–Scholes
(1973), Merton (1973), Black (1976), Garman and Kohlhagen (for details,
refer to Bellalah et al., 1998), and Barone-Adesi and Whaley (1987). Our
attention is focused on the question of dividend and the privilege feature
of early exercise and information uncertainty.
When there are no distributions to the underlying asset, there is no
incentive to exercise an American call option before its maturity date.
Hence, the value of an American call is equal to that of a European call.
The valuation of American calls is simpler than the valuation of American
puts with and without discrete cash distributions to the underlying asset.
While European and American calls have the same value when there
are no distributions to the underlying asset, this result does not hold
for European and American puts. The nonexistence of a put-call parity
theorem for American options imply a specific treatment for put options.
The pricing of American puts is rather a difficult task, even in the absence
of distributions to the underlying asset. However, some interesting results
are presented in the absence of dividends and when there is a continuous
dividend rate. Most options written on commodities and commodity futures
are of the American type. Hence, an early exercise premium is embedded
in American call and put prices. Analytical solutions for the American
option pricing problems with several dividends have not yet been found. The
quadratic approximation method used by Barone-Adesi and Whaley (1987)
is computationally efficient. It provides an accurate, inexpensive method for
the valuation of American call and put options traded on commodities and
commodity futures. In the case of discrete distributions to the underlying

asset, the formulas proposed in the literature are not so efficient as those
proposed for American calls. The valuation of American stock options is
equivalent to the valuation of a portfolio which contains a commodity
option on a stock with some distributions. The general problem of valuing
American options is analyzed in three different contexts: when there are no
distributions to the underlying asset, when there is a constant proportional
distribution rate, and in the presence of discrete cash distributions. The
analysis covers American options on spot assets and American futures
options. The main results reported in the literature regarding the pricing
of American calls are reviewed. Some models are presented in detail and
simulations are run.
Information costs play a central role in the analysis and the valuation
of financial assets, firms and their cash flows. We present an arbitrage
argument to derive an option pricing formula within a context of infor-
mation uncertainty. The formula collapses to that in Black and Scholes in
the absence of these costs. By analogy between the option theory and the
assets in the capital structure of the firm, the formula can be used to value
equity in a levered firm. The compound option formulas derived in this
chapter shed light on the effects of leverage and information uncertainty in
the valuation of corporate assets. Since several corporate liabilities can be
valued with the compound option approach, our results are useful in the
pricing of the capital structure of the firm.
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?
Appendix A: An Alternative Derivation of the Compound

Option’s Formula Using the Martingale
Approach
Using similar arguments as before, we obtain:
C0 = e−(r+λC )T0 EQ [CT0 ].

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 ).
Then by applying standard integral calculation (change of variables for

example), we deduce the result.
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:
0 ≤ C1 − C2 ≤ (E2 − E1 )e−(r+λc1 +λc2 )(T −t) .
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
At the maturity date T , the value of the portfolio is given by:
Π(T ) = S − max(S − E, 0) ≥ 0
Hence, the absence of arbitrage opportunities in efficient markets
implies that:
Π(t) = S − C ≥ 0.
Therefore, we must have:

S ≥ C.
2. As before, we can construct the following portfolio, Π which comprises

the underlying asset S and the call C:
Π = S − C.
Note that Π(T ) ≤ E.

The absence of arbitrage opportunities in efficient markets implies that
Π(t) = S − C ≤ Ee−(r+λs +λc )(T −t) .
Hence, we have:
C ≥ S − e−(r+λs +λc )(T −t) .
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 .
When t = T , the value of the portfolio at maturity is given by,
Π(T ) = max(S − E1 , 0) − max(S − E2 , 0)
this allows to write:

0 ≤ Π(T ) ≤ E2 − E1 .
implies that
0 ≤ Π(T ) ≤ (E2 − E1 )e−(r+λc1 +λc2 )(T −t)
Finally, we have:
0 ≤ C1 − C2 ≤ (E2 − E1 )e−(r+λc1 +λc2 )(T −t) .
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 .
1. P ≤ Ee−(r+λp )(T −t) .

2. P ≥ Ee−(r+λp +λs )(T −t) − S.
3. 0 ≤ P2 − P1 ≤ (E2 − E1 )e−(r+λp1 +λp2 )(T −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.

implies that:
Π(T ) = P − Ee−(r+λp +λs )(T −t) ≤ 0.
Hence,
P ≤ Ee−(r+λp +λs )(T −t) .
2. To prove the desired result, we construct a portfolio which comprises the

put and the underlying asset Π(T ) = P + S.
At the option’s maturity date t = T and the portfolio value is given by:
Π(T ) = S + max(E − S, 0) ≥ E.
The absence of arbitrage opportunities in efficient markets implies that:
Π(t) = S + P ≥ Ee−(r+λp +λs )(T −t) .

Hence,
P ≥ Ee−(r+λp +λs )(T −t) − S.

3. We construct a portfolio comprising two puts with two different strike
prices,
Π = P2 − P1 .
At the option’s maturity date t = T , the portfolio value is given by:

Π(T ) = max(E2 − S, 0) − max(E1 − S, 0).
This gives:
0 ≤ Π(T ) ≤ E2 − E1 .
The absence of arbitrage opportunities in efficient markets implies that:
0 ≤ Π(T ) ≤ (E2 − E1 )e−(r+λp1 +λp2 )(T −t)

hence,
0 ≤ P2 − P1 ≤ (E2 − E1 )e−(r+λp1 +λp2 )(T −t) .
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
The difference between the two equations gives:

∂C ∂P 1 ∂ 2C ∂2P
− + − σ2 S 2
∂t ∂t 2 ∂S 2 ∂S 2

∂C ∂P
+ (r + λs )S − − (r + λc )C + (r + λp )P = 0 (14.31)
∂S ∂S

∂C ∂P 1 ∂ 2C ∂ 2P
− + − σ2 S 2
∂t ∂t 2 ∂S 2 ∂S 2

∂C ∂P
+ (r + λs )S − − (r(C − P ) + λc C − λp P ) = 0.
∂S ∂S
We denote by V (S, t) = C(S, t) − P (S, t). In this case, the portfolio with
a long call and a short put satisfies the extended Black–Scholes equation.
Hence, we have:
∂V 1 ∂ 2V 2 2 ∂V
+ σ S + (r + λs )S − (r + λv )V = 0
∂t 2 ∂S 2 ∂S
where,
(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
The equation can be written as:

∂V 1 ∂2V 2 2 ∂V
+ σ S + (r + λs )S − (r + λv + λp )V = 0
∂t 2 ∂S 2 ∂S
with,
λc P = λp C.
Hence, the portfolio
V (S, t) = C(S, t) − P (S, t)
satisfies the extended Black–Scholes equation.

2. We can show that the boundary boundary conditions for the portfolio
are,
∂V
V (0, t) = C(0, t) − P (0, t) and − (r + λv + λp )V = 0
∂t
then
1 ∂V
= r + λc + λp
V ∂t
and, we have:
V (0, t) = −Ee−(r+λc +λp )(T −t) .
Another boundary condition is:
V (S, t) = C(S, t) − P (S, T )
when,
S → +∞.
The terminal condition at the maturity date T is:
V (S, T ) = C(S, T ) − P (S, T )

= max(S − E, 0) − max(E − S, 0) = S − E.
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 .
Solution
Consider the following dynamics of the underlying asset.
dS = µSdt + σSdX.
Using Ito’s lemma for the function V (S, t) gives:

∂V ∂V 1 ∂ 2V 2 2
dV = dS + + σ S dt. (14.32)
∂S ∂t 2 ∂S 2
According to the extended Black–Scholes equation, we have
∂V 1 ∂ 2V 2 2 ∂V
+ σ S = (r + λv )V − (r + λs )S = 0. (14.33)
∂t 2 ∂S 2 ∂S
Replacing Eq. (14.33) in Eq. (14.32) gives:
∂V ∂V
dV = dS + (r + λv )V dt − (r + λs )S dt.
∂S ∂S
Hence:

∂V ∂V ∂V
dV = dS + r V − S dt + λv V − λs S dt
∂S ∂S ∂S
or,
∂V ∂V
dV = (r + λv )V dt + σS dX + (µ − (r + λs ))S dt
∂S ∂S
and finally:

∂V ∂V
dV = σS dX + (r + λv )V + (µ − rλs ) S dt.
∂S ∂S
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
We denote by V (S, t) the value of a spot option whose value depends on

the underlying spot asset S, time t, and a continuous dividend yield D. The
value of this option must satisfy the extended Black–Scholes equation:
∂V 1 ∂ 2V 2 2 ∂V
+ 2
σ S + (r + λs − D)S − (r + λv )V = 0.
∂t 2 ∂S ∂S
When D = r, the first equation becomes:
∂W 1 ∂ 2W 2 2
+ σ F − (D + λw )W = 0
∂t 2 ∂F 2
and the second equation becomes:
∂V 1 ∂2V 2 2 ∂V
+ 2
σ S + λs S − (D + λv )V = 0.
∂t 2 ∂S ∂S
Both equations are identical only when λs = 0 and λw = λv .
This means that the knowledge of the method to price the spot option
allows to price the futures option when the asset pays a dividend equal to
the interest rate by accounting for the effect of information uncertainty.
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.
Bellalah, M. (1999). The valuation of futures and commodity options with
1895–1927.
theory and tests. The Financial Review, 30(3) (August), 617–635.
Bellalah, M and JL Prigent (2001). Pricing standard and exotic options in the
of Finance, 13(3), 1975–2000.
Finance, 13(2), 1826–1837.

Black, F and J Cox (1976). Valuing corporate securities: some effects of bond
indenture conditions. Journal of Finance, 31 (May), 351–367.
Blomeyer, EC (1986). An analytic approximation for the American put price
for options on stocks with dividends. Journal of Financial and Quantitative
Analysis, 21 (June), 229–233.
Bunch, DS and H Johnson (2000). The American put option and its critical stock
price. Journal of Finance, 55(5), 2333–2357.
Feller, W (1971). An Introduction to Probability Theory and Its Applications.
New York: John Wiley.
Galai, D and R Masulis (1976). The option pricing model and the risk factor of
stock. Journal of Financial Economics, 3, 53–81.
Geske, R (1979). The valuation of compound options. Journal of Financial
Economic, 7, 375–380.
Geske, R and HE Johnson (1984). The American put option valued analytically.
Journal of Finance, 39 (December), 1511–1524.
Geske, R and K Shastri (1985). Valuation by approximation: a comparison of
alternative option valuation techniques. Journal of Financial and Quantita-
tive Analysis, 20 (March), 45–71.
Grabe, JO (1983). The pricing of call and put options on foreign exchange. Journal
of International Money and Finance, 2, 239–253.
Harrison, JM and S Pliska (1981). Martingales and stochastic integrals in the
theory of continuous trading. Stochastic Processes and Their Applications,
11, 215–260.
Harvey, C and R Whaley (1992). Dividends and the S-P 100 index option
valuation. Journal of Futures Markets, 12(2), 123–137.
Johnson, HE (1983). An analytic approximation to the American put price.
Journal of Financial and Quantitative Analysis, 18(March), 141–148.
McKean, Jr (1969). Stochastic Integrals. New York: Academic press.
McMillan, L (1986). Analytic approximation for the American put option.
Advances in Futures and Options Research, 1, 119–139.
Merton, R (1987). An equilibrium market model with incomplete information.
Omberg (1991). On the theory of perfect hedging. Advances in Futures and
Options Research, 5, 1–29.
Roll, R (1977). An analytical valuation formula for unprotected American call

options with known dividends. Journal of Financial Economics, 5, 251–258.
Samuelson, P (1972). The Collected Scientific Papers of Paul Samuelson. RC
Merton (ed.). Cambridge, MA: MIT Press.
Whaley, R (1981). On the valuation of American call options on stocks with
known dividends. Journal of Financial Economics, 9, 207–211.
tests. Journal of Finance, 41 (March), 127–150.
Chapter 15
RISK MANAGEMENT OF BONDS AND

INTEREST RATE SENSITIVE INSTRUMENTS IN
THE PRESENCE OF STOCHASTIC INTEREST
RATES AND INFORMATION UNCERTAINTY:
THEORY AND TESTS
Chapter Outline
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
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
Risk Management of Bonds and Interest Rate Sensitive Instruments 669
options), the specific bond in a deliverable basket (the delivery switch

options) etc. The nominal short-term interest rate cannot be negative
since people can hold currency at a zero nominal rate. The real interest
rate can be negative since low-risk investment opportunities do not lead
necessarily to positive returns. In the same context, the inflation rate
can also be negative during periods of great depressions. However, at the
microeconomic or the macroeconomic levels, all the costs of information
relative to immaterial investments among others are ignored in financial
economics. If we admit that the nominal short-term rate is the “shadow
real interest rate, plus a “shadow” cost of incomplete information and the
expected inflation, or zero, then the nominal short-term rate can be seen
as an option as in Black (1995).
15.1. The Valuation of Bond Options and Interest

Rate Options
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.
15.1.1. The problems in using the B–S model

for interest-rate options
To illustrate the problems in using this model, consider a six-month
European call on a six-year zero-coupon bond. The bond’s value at maturity
is 100. The strike price is 130. Since this bond’s value never exceeds 100,
the option will not be exercised and its value must be zero. However, if
you consider a given interest rate and volatility, this option has some value
in the Black and Scholes (1973) model. Since the model is based on an
assumption of a log-normal distribution for the underlying zero-coupon
bond, this means that there is a probability that the asset takes on any
positive value. This means that the bond price can be above 100. This
situation is also possible in the presence of negative interest rates. In the
same vein, the value of a coupon-paying bond cannot be greater than the
sum of the coupon payments plus the maturity value. Hence, the use of
such values can lead to nonsensical option prices. The B–S model assumes
also that the short-term interest rate is constant. However, a change in the
short-term interest rate modifies the rates along the yield curve. Hence, it
is inappropriate to assume constant interest rates in the pricing of interest
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.
15.1.2. Sensitivity of the theoretical option prices

to changes in factors
The shape of the curve between the theoretical call option price and the
underlying bond is convex. The change in the call price with respect to a
small change in the underlying bond price refers to the option delta. The
lambda refers to the percentage change in the call price with respect to the
percentage change in the underlying bond price. For example, a lambda
of 1.25 shows that the call price will change by 1.25% for a change in the
underlying bond price by 1%. The measure of convexity for call options is
defined by the gamma or the change in delta with respect to a change in the
underlying bond price. The theta is given by the change in the option price
with respect to a decrease in the time to expiration. The vega also known as
the kappa gives the change in the option price with respect to a 1% change
in the expected interest rate volatility. As we define the modified duration
of a bond as an indicator of its price sensitivity to variations in interest
rates, it is possible to define the modified duration for an option as follows:
modified duration for an option = modified duration of the underlying
asset×(delta)×(price of the underlying asset option price). It is also possible
to use for interest-rate options, the put-call parity relationship which can
be written for a coupon-bearing bond as: put = call + present value of the
strike price + present value of the coupons − price of the underlying bond.
15.2. A Simple Non-Parametric Approach to Bond

Futures Option Pricing
Options on the CBOTs’ 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’s 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. Stutzer and Chowdhury (1999) proposed a combination of the risk-
neutral valuation framework and the flexibility of practitioner methods in
the pricing of options. A canonical valuation model is a simplified, risk-
neutral valuation method allowing the model user to specify an individual
assessment of the distribution of the underlying asset at the option maturity
date. This distribution is used to estimate risk-neutral probabilities needed
to value the option. Stutzer and Chowdhury (1999) showed that the
canonical model predicts that in the historically typical range of bond
futures prices, the Black’s model implied volatility of in-the-money calls
must be higher than of other calls. They showed that this pattern is
more pronounced for short-term options. They found that the canonical
model predicts that implied volatilities should, in general, be much higher
when the futures price is near historic lows. The cannical model seems to
outperform the Black’s model.
15.2.1. Canonical modeling and option pricing theory

Black’s model can be used for European futures options in the following
form:
Call = e−rT [F N (d1 ) − KN (d2 )] (15.1)

F 1 2
log K + 2 σ T √
d1 = √ , d2 = d1 − σ T
σ T
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.
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.
15.2.2. Assessing the distribution of the underlying

futures price
Daily closing prices or the CBOT bond futures are used for the year
1996. The ratio of two prices is recorded as the price relative or gross
“return” RhF . The T-ahead futures price is denoted by Ph (T, F ) = F RhF .
Assigning equal probabilities gives a simple non-parametric estimate of the
T -ahead probability distribution of the futures price, conditional on the
current futures price, F . The N -month-ahead probability distributions are
illustrated by histograms. The results show substantial skewness of the
returns distribution when the current futures price is relatively low, i.e.,
when bond yields are high. This means that the probability of unusually
large, positive increases in F is higher when the starting F is relatively
low. Also, the results show a higher volatility associated with the returns
in periods of low futures prices, i.e., high bond yields. These properties are
absent in the Black’s model.
15.2.3. Transforming actual probabilities into risk-neutral

probabilities
A canonical risk-neutral probability distribution is a distribution satisfying
the following martingale constraint: the risk-neutral expected value of the
time T -ahead futures price must equal the current futures price. For each
combination of T and F , the following convex problem is solved:
H

Π̂ = arg P max − Π(h) log Π(h) (15.2)
h Π(h)=1
h=1
subject to the martingale constraint requiring that F equals the risk-neutral

probability Π(h) weighted average of the T -ahead futures price Ph (T, F ):

Π(h)Ph (T, F ) = F
h
The maximand in Eq. (15.2) corresponds to the Shannon entropy of

the risk-neutral distribution. The risk-neutral probabilities obtained by
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
15.2.4. Qualitative comparison of Black and canonical

model values
The results show that the Black’s model has a tendency to relatively and
absolutely underprice in-the-money calls. Also, the relative mispricing of
in-the-money calls persists even when using an implied volatility. 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.
15.3. One-Factor Interest Rate Modeling and the Pricing

of Bonds: The General Case
The interest rate that applies to the shortest possible deposit is known as
the spot interest rate. The dynamics of the spot rate can be modeled by the
following stochastic differential equation dr = u(r, t)dt + w(r, t)dX where
the functional forms u(r, t) and w(r, t) can be specified in different contexts.
15.3.1. Bond pricing in the general case: The arbitrage

argument and information costs
Consider the pricing of an interest-rate sensitive instrument V (r, t, T ) when
the interest rate is stochastic. Since the interest rate is not a traded asset,
the pricing of bonds is different from the pricing of traded assets. If V (r, t, T )
refers to the price of a bond, then the implementation of a hedging strategy
in a B–S context needs the use of another bond with a different maturity
as a hedging instrument. Consider the following portfolio comprising a long
position in a bond V1 (r, t, T ) and a short position in ∆ units of the bond

V2 (r, t, T )
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
In order to construct a hedged portfolio, we have to eliminate the risk

component in the dr terms. In this case, the hedge ratio must be equal to
∆ = ∂V∂r
1
/ ∂V
∂r
2
. Using arbitrage arguments as in the B–S context, the return
on the hedged portfolio must be equal to the risk-free rate or the spot rate
plus information cost on each market (or an asset). This gives:

∂V1 1 ∂ 2 V1 ∂V1 ∂V2 ∂V2 1 2 ∂ 2 V2
dΠ = + w2 − + w dt
∂t 2 ∂r2 ∂r ∂r ∂t 2 ∂r2

∂V1 ∂V2
= (r + λV1 )V1 dt − (r + λV2 ) V2 dt
∂r ∂r
Collecting the terms in V1 and V2 and re-arranging this equation gives:

2 2
∂V1
∂t + 12 w2 ∂∂rV21 − (r + λV1 )V1 ∂V2
∂t + 12 w2 ∂∂rV22 − (r + λV2 )V2
∂V1
= ∂V2
.
∂r ∂r
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
It is possible to write the function a(r, t) as follows:
a(r, t) = w(r, t)γ(r, t) − u(r, t)

In this context, the bond pricing equation is given by:
∂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 ),
where tc refers to the coupon-payment date, t− c the instant just before

the coupon payment, and t+ c the instant just after the coupon payment.
This condition reflects the jump in the bond price at the coupon date. The
difference between the bond price before and after the coupon date
corresponds to the coupon payment. In order to explain the market price
of risk, consider an investor who holds an unhedged position in a bond
maturing in T years. The change in the bond value over a small interval of
time dt is given by:

∂V ∂V 1 2 ∂ 2V ∂V
dV = w dX + + w +u dt
∂r ∂t 2 ∂r2 ∂r
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
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.
15.3.2. Pricing callable bonds within information

uncertainty
A callable bond is a bond with a call provision that allows the issuer to
call back on specified dates for a given amount of the issue. The price of a
callable bond satisfies the following equation:
∂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 .
15.4. Fixed Income Instruments as a Weighted Portfolio

of Power Options
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
tansformation 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. There
is a relationship between swaps, bonds, and the forward rate agreements
(FRAs). A plain vanilla swap of the type “receive fixed, pay floating” can
be assimilated to a long position in a coupon-paying bond and a short

position in a floating-rate note. It can also be viewed as a series of FRAs
where the holder of the “receive fixed” position pays the difference between
the observed three-month LIBOR rate and the fixed swap coupon at each
re-set date.
Consider a two-year swap, starting one year forward. In the presence
of annual re-sets, the value of the swap in one year is the sum of the cash
flows multiplied by their respective forward discount functions:
C exp(−Ra ) + (C + 100) exp(−2Rb )
where
c: fixed swap coupon;
Ra : forward rate for one year and
Rb : two-year rate (in a year).
This value of a two-year swap in one year is equivalent to the value

of a bond paying an annual coupon. It is also equivalent to the value of
two zero-coupon bonds at time one year, one paying c in two years and the
other paying (c + 100) in three years.
Consider one of the two zero-coupon bonds (the second), and denote
by Vt the value of the cash flow in t years as: V1 = X exp(−2Rb ) where
X = (c + 100). This relationship is exponentially decreasing and represents
the value of an option as a function of the rate Rb , which is the underlying
asset. This reasoning also applies for the first bond. More generally, for
a forward contract on a zero-coupon bond (with maturity T , yield Y for
purchase at t), the value of the bond at time t is an exponentially decreasing
function of its yield: e−Y (T −t) per dollar face value. In the above example,
T = 3, t = 1, and Y is the forward rate which will give the value of the
bond at t = 1.
Using Taylor’s expansion, Hart (1997) provided the exponential curve
as a weighted portfolio of simpler curves of increasing power. This portfolio
can then be valued using a series of Black and Scholes power options using
the results in Hart and Ross (1994). The value of a European power 2 call is:
√ √
S 2 exp(t(r − 2d + σ 2 ))N [x + σ t ] − K exp(t(−r))N [x − σ t ]
S
ln K + (r − d)t 1 √
x= √ + σ t
σ t 2
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:
exp(−rt) + (−Y τ ) exp(−rt) + 1/2(Y τ )2 exp(−t(r − σ2 ))

1
+ (−Y τ )3 exp(−t(r − 3σ 2 )), . . .
6
This result gives the value of a forward zero-coupon swap in a Black–Scholes
economy with volatility σ and Eurodollar-implied yield Y that includes a
convexity. This result is similar to the time value of an option. This formula
can be used for a range of 5–15 terms.
15.5. Merton’s Model for Equity Options in the Presence

of Stochastic Interest Rates: Two-Factor Models
Consider the basic case of a European option with no payouts to the
underlying asset. Assume that there are no transaction costs that trading
takes place continuously and that no restrictions are imposed on borrowing
and short selling. Assume also that the borrowing rate is equal to the
lending rate.
15.5.1. The model in the presence of stochastic

interest rates
Let C(S, P, τ, K) be the option price function depending on the stock price
S, a bond price P , the strike price K, and the time remaining to maturity
τ with τ = T − t, where T is the maturity date and t is the current time.
The solution presented in Merton (1973) for the call price is:
1
y(x, T ) = [xerf c(h1 ) − erf c(h2 )]
2
where
ln(x) + 12 T ln(x) − 12 T
h1 = − √ , h2 = − √
2T 2T
and
τ
T = [σ2 + δ 2 − 2ρσδ]du
0
and where erf c(.) is the error complement function

h
2 w2
erf c(h) = 1 − √ e− 2 dw.
2Π 0
In the special case when r = 0, σ2 = 1, K = 1, Eq. (15.21) is identical

to the Black–Scholes equation. The call price in the context of Merton’s
model can be written in a B–S form as:
c = SN (d1 ) − P (τ )KN (d2 )

S
ln K − ln(P ) + 12 στ2 τ √
d1 = √ , d2 = d1 − στ τ
στ τ
In the same context, the put price is given by:
p = −SN (−d1 ) + P (τ )KN (−d2 )

S
ln K − ln(P ) + 12 στ2 τ √
d1 = √ , d2 = d1 − στ τ
στ τ
with
τ
−rτ 1
P (τ ) = e and στ2 = [σ 2 + δ 2 − 2ρσδ]du.
τ 0
15.5.2. Applications of Merton’s model

Options on yields of short-term financial instruments are traded in the
over-the-counter (OTC) market. These financial products are in the form
of bank deposits, certificates of deposit, Treasury bills, commercial papers
and so on.
Options on bank deposit rates and on treasury bill yields

Treasury securities are backed by the faith and credit of the government.
They are regarded as having no credit risk. The interest rates on Treasury
securities are often used as a benchmark in national and international
capital markets. There are two categories of government assets: discount
and coupon assets. Treasury bills are securities issued with maturities of a
year or less as discount securities, which do not make periodic payments.
All the assets having a longer maturity are issued as coupon securities.
When the maturity is between 2 and 10 years, the issued security is
a note. When the maturity is greater than 10 years, the Treasury security
is a bond. Re-call that a fundamental property of a bond is that its price
moves in the opposite side with respect to the required yield, i.e., when the
required yield increases, the present value of the par value and the coupon
decreases and vice versa. In general, the price yield relationship is convex.
The yield on any investment corresponds to the interest rate which makes
the present value of the cash flows equal to the cost of the investment.
The yield is simply an implicit interest rate. In practice, two measures of
yield are used: the current yield and the yield to maturity. The current
yield gives the relationship between annual coupon interest and the market
price. It is simply given by the ratio of the annual dollar coupon interest
to the market price. The yield to maturity is the implicit interest rate that
makes the present value of the future cash flows until the bond’s maturity
date equal to the bond’s price. Merton’s (1973) model can be applied to
the valuation of these options and is more appropriate than the Black and
Scholes (1973) model since interest rates are stochastic.
Short-term options on long-term bonds

Transactions on short-term options on long-term bonds take place often
in the OTC market. Some options are also traded on organized markets.
These options may be either European or American. A European short-term
option on a long-term bond is traded on the Chicago Board Options
Exchange (CBOE). This option is similar to the option on yield of short-

term financial instruments traded on the CBOE and is cash settled. The
interest rate corresponds to the average of the yields to maturity of the
two most recently auctioned 7-year Treasury notes and two most recently
auctioned 10-year Treasury notes and also the two most recently auctioned
30-year Treasury bonds. The average of the yield to maturity of these six
instruments gives the interest rate on which the option is priced. This option
can be priced using Merton’s model which is more appropriate than the B–S
model in the context of stochastic interest rates.
15.6. Some Models for the Pricing of Bond Options

Hull and White (for details, refer to Bellalah et al., 1998) developed a
unifying context for the pricing of all interest-rate dependent securities
based on the term structure of interest rates and their volatilities. Bond
options are often priced using a version of the B–S model. Caps, collars,
and floors are often priced by assuming that forward interest rates are log-
normal. The main difference between the dynamics of bonds and stocks
is that the bond price tends to reach its face value at maturity. This is
not the case for the stock price. European options can be valued using
the Black’s (1976) model by assuming that forward interest rates are log-
normal. In this case, volatility is a decreasing function of the option time
to maturity. It is important to recognize that when forward bond prices
are log-normal, this does not mean that forward interest rates are also log-
normal. The standard models of interest rates proposed by Vasicek (1977),
Cox et al. (1985), Brennan and Schwartz (1977) and so on 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.
15.6.1. An extension of the Ho-Lee model for bond options

Ho and Lee model allows the pricing of several interest rate derivatives in a
consistent way. The main drawbacks of this model are that actual changes in
interest rates are normally distributed and that all spot interest rates and
forward interest rates are equally variable. However, these disadvantages
can be overcome as shown in Heath et al. (for details, refer to Bellalah
et al., 1998), Jamshidian (1989), and Hull and White (1990). In fact, the
short-term interest rate process can be adjusted to eliminate the possibility
of negative interest rates. In this case, the Ho-Lee model can be written as:
√
dr = θ(t)dt + σ rdz.
It is also possible to introduce mean reversion in the Ho-Lee model

using some possible models suggested in Hull and White (for details, refer
to Bellalah et al., 1998). The two possible models accounting for mean
reversion are:
dr = (θ(t) + a(t)(b − r))dt + σdz
and
√
dr = (θ(t) + a(t)(b − r))dt + σ rdz
where b is constant and a is a function of time.

Hull and White (1990) derived simple formulas for the pricing of
European interest rate options. They considered the following dynamics
for the short-term interest rate r:
dr = (θ(t) + a(t)(b − r))dt + σdz,
dr = (φ(t) + a(t)r)dt + σdz.
We denote by A(t, T )e−B(t,T )r the price at time t of a discount bond

with a maturity T . The bond price volatility is σB(t, T ). For a period
between T1 and T2 , the volatility of a forward rate F is given by:
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:
(B(0, T ) − B(0, t)) (δ 2 B(0, t)/δt2 )

B(t, T ) = , a(t) = −
(δB(0, t)/δt) (δB(0, t)/δt)
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:
call = P1 N (h) − XP2 N (h − v)

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.
15.6.2. The Schaefer and Schwartz model

Schaefer and Schwartz (for details, refer to Bellalah et al., 1998) assumed
that the bond’s price volatility is proportional to the bond’s duration.
Schaefer and Schwartz (for details, refer to Bellalah et al., 1998) developed
a modified B–S model for the pricing of bond options. In their model, the
price volatility of a bond increases with duration. The call price is given by:
c = Se(b−r)T N (d1 ) − Xe−rT N (d2 )

S
ln X + (b + 12 σ 2 T ) √
d1 = √ , d2 = d1 − σ τ ,
σ τ
σ0
σ = D(KS (α−1) ), K =
D.S (α−1)
where:
D: duration of the bond after the option’s maturity date;
D∗ : duration of the bond today;
α: a constant which may be equal to 0.5 and
σ0 : the observed price volatility of the bond.
15.6.3. The Vasicek (1977) model

The dynamics of interest rates in Vasicek (1977) model are given by:
dr = κ(θ − r)dt + σdz
Bond prices
The time t price of a discount bond maturing in T is given by:
P (t, T ) = A(t, T )e−B(t,T )r(t)
and
[1 − e−κ(T −t) ]
B(t, T ) =
κ
with
[(B(t,T )−T +t)(κ2 θ−0.5σ2 )]
A(t, T ) = e κ2 −σ2 B(t,T )2 /4κ .
Prices of European options

The European call price is:
c = P (t, τ )N (h) − XP (t, T )N (h − σp )
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:
p = XP (t, τ )N (−h + σp ) − P (t, τ )N (−h)
When κ = 1.5%, θ = 2%, σ = 8%, r = 5%, P = 100, X = 100, t = 0,

T = 2, and τ = 3, we have: B(t, T ) = 1.97029776, A(t, T ) = 1.00778006,
and P (t, T ) = 0.91323235. For the valuation of options, we have: σp =
33.48%, h = 97.98%, B(t, τ ) = 2.993350121, A(t, τ ) = 1.38806545, and
P (t, τ ) = 1.19869799. The call price is 32.632 and the put price is 4.085.
When κ = 1.5%, θ = 2%, σ = 8%, r = 5%, P = 100, X = 100, t = 0,
T = 2, and τ = 2, we have: B(t, T ) = 1.97029776, A(t, T ) = 1.00778006,
and P (t, T ) = 0.91323235. For the valuation of options, we have: σp =
33.48%, h = 16.74%, B(t, τ ) = 1.97029776, A(t, τ ) = 1.00778006, and
P (t, τ ) = 0.91323235. The call price is 12.141 and the put price is 12.141.
15.6.4. The Ho and Lee model

The dynamics of the spot rate in the Ho and Lee (1986) model are
given by:
dr = θ(t)dt + σdz.
In this model, bond prices are given by:
P (t, T ) = A(t, T )e−r(t)(T −t)
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)
European options
The European call price is:
c = P (t, τ )N (h) − XP (t, T )N (h − σp )
where:
1 P (t, τ )
h= ln + 0.5σp , σp = σ(τ − T )T − t]
σp P (t, T )X
The European put price is:
p = XP (t, T )N (h − σp ) − P (t, τ )N (h)
15.6.5. The Hull and White model

The dynamics of interest rates in this model are given by:

θ(t)
dr = κ − r dt + σdz
κ
θ(t)
where κ
is a time-dependent mean-reversion level.
Bond prices
The time t price of a discount bond maturing in T is given by:
P (t, T ) = A(t, T )e−B(t,T )r(t)
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)
Prices of European options

The European call price with maturity date T on a zero-coupon bond
maturing in t is:
c = P (0, τ )N (h) − XP (0, T )N (h − ν(T, t))
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
of Vasicek and Cox et al. We also present an interesting model of interest

rates, in Heath et al. which accounts for several reasons of the term structure
movements. Much work is currently done on the term structure modeling.
The introduction of information costs in the analysis of interest rates and
the valuation of financial assets is equivalent to applying an additional
discount rate for the computation of the present value of future risky
cash flows. The shadow costs are introduced in Merton’s (1987) model of
capital market equilibrium with incomplete information. The concept of
information costs is used to illustrate the limiting case of money demand
behavior (the liquidity trap) within an option framework. When the short-
term rate is close to zero, this context allows to illustrate the effect of
the “currency” option embedded into nominal interest rates on savings,
investments, and the yield curve. Our analysis can be extended to develop
formal models of interest rates and derivatives.
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?
Appendix A: Government Bond Futures and Implicit

Embedded Options
Positions in government bond futures have several implicit embedded
options. These options corresponding to several specifications of the day
(delivery date and end-of-month options), the time of delivery (wildcard
options), the specific bond in a deliverable basket (the delivery switch
options) etc. Dabansens and Bento (1997) approximated the delivery
switch option as a simple combination of European calls and puts for which
the underlying asset corresponds to the cheapest to deliver (CTD) bond
yield and the deliverable bond yield spread. These options present the same
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.
A.1. Criteria for the CTD

The CTD is defined as the bond with the lowest basis net of carry B or
the lowest implied futures price (IFP). The basis net of carry corresponds
to the cost of a cash and carry trade on the bond: B = basis-carry
B = (P (t) + AI(t))(1 + m/36,000) − [Crec + Al(t ) + F (t)CF ] (A.1)
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.
A.2. Yield changes

When market conditions change, this makes one bond referred to as the new
CTD cheaper than the original CTD. This produces CTD switches. A new
bond becomes the CTD when the difference between its B and that of the
original CTD tends to become zero. Let us consider in a first step the case
of parallel yield shift and write the modified duration M D of a bond as:
dP = −P.M D.dy (A.3)

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:
dBN (Switchpar (N )) = −∆BN (A.4)
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:
dBN (Switchrel (N )) = −∆BN .
The relative yield switch for bond N to become the CTD is:
∆BN
Switchrel (N ) =
(1 + m/36,000)(M DN (t)PN (t))
A.3. The value for a futures position

How the holder of a cash and carry position on the CTD bond profits from
a switch in the CTD?
By construction, we have:
IF PN (TS ) = IF PCT D (TS ),
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:
P rofN = max (0, CFCT D (dIF PCT D − dIF PN ))

CFCT D
= max 0, dPCT D − dPN
CFN
where: dIF PN , (dIF PCT D ): change in IFP of bond N (or the CTD)
between the switch and futures settlement and dPN , (dPCT D ): changes in
bond price (or the CTD) between the switch and futures settlement.
We denote by dyN , (dyCT D ), the yield change of bond N , (CTD)
between switch and futures settlement. Using modified duration for first-
order changes of bond price allows to write the profit as:

P rofN = max 0, −PCT D (TS )M DCT D (TS )dyCT D

PN (TS )M DN (TS )dyN CFCT D
+
CFN
We simplify the notation and define:
ACT D = PCT D (TS )M DCT D (TS )
and
PN (TS )M DN (TS )CFCT D
XN =
CFN
in order to write the profit as:
P rofN = max[0, −ACT D .dyCT D + XN .dyN ].
Now, it is possible to answer the following question: What is the fair

value of this switch under different market scenarios? The values of delivery
options can be approximated under a parallel yield shift and a relative
yield shift.
A.4. Parallel yield shift

Under a parallel yield shift, the time of switch to CTD is TSpar dyCT D = dyN
and the profit is:
P rofN = max(0, Kpar .dyN )
with a strike price:
Kpar = −ACT D (TSpar ) + XN (TSpar )
Let us denote by:

Optpar (N ): the value of the delivery switch option related to bond N ; and
P yN (.): the probability distribution function of yN .
The value of the option is studied in two cases.
In the first case, if:
Kpar ≥ 0(XN ≥ ACT D )
the profit:
P rofN = Kpar max(0, dyN ), dyN ≥ 0
is positive when dyN > 0 or:
yN ∈ (yN 0 + Switchpar (N ), +∞),
where yN 0 corresponds to the present yield of bond N . In this case, the

option value is given by:
∞
Optpar (N ) = Kpar P (yN = y)(y−yN 0 −Switchpar (N ))dy
yN 0 +Switchpar (N )
In the second case, if:
Kpar ≤ 0(XN ≤ ACT D ), P rofN = −Kpar max(0, −dyN )
then the option value is given by:

yN 0 +Switchpar (N )
Optpar (N ) = −Kpar P yN (yN = y)(yN 0
0
+ Switchpar (N ) − y)dy
A.5. Relative yield shift

Under relative yield shift, the time of switch to CTD is TSrel dyCT D = 0
and yN represents the yield spread between bond N and the CTD. The
profit is:
P rofN = Krel max(0, dSprN )
with:
yN − yCDT = SprN , Krel = XN (TSrel )
The value of the delivery switch option related to bond N is given by:
∞
Optpar (N ) = −Krel PSN (SprN = S)
sprN O+Switchrel (N )
× (SprN − SprN O − Switchrel (N ))ds
where PSN (.) is the probability distribution of SprN .

This framework allows the valuation of delivery switch options in cases
of both parallel and non-parallel yield shifts. The above analysis can be
implemented to quantify the cheapness of a futures contract relative to the
CTD. It can also be used in the selection of a cash and carry or a reverse
cash and carry trade with a bond other than the CTD. For further details,
see the original paper by Dabansens and Bento (1997).
Appendix B: One-Factor Fallacies for Interest

Rate Models
One-factor term structure models are widely used in the pricing of interest
rate derivatives that cannot be accommodated by Black’s (1976) model. In
these models, changes in 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.
Using time series data, for the estimation of model parameters, the
volatility of forward interest rates will be underestimated, which in turn,
yield to lower option prices. Besides, when parameters are estimated by
calibrating the model to market data, implied volatilities of zero-coupon
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.
B.1. The models in practice

Since it is nearly impossible to obtain the equality between model
prices and market prices, (a perfect fit), model parameters are chosen
to minimize a certain measure of discrepancy between model and market
prices. This is referred to as calibration. It is often observed that model
parameters resulting from the calibration on prices of caps and floors
differ significantly from those obtained from the calibration on swaptions.
In fact, using one-factor models, Klaassen et al. (1998) found that the
cap/floor volatility curve lies about 10% higher than the swaption volatility
curve. This difference seems to persist over time. This difference can be
understood with reference to spreads between rates and the calculations of
correlations.
B.2. Spreads between rates

Caps, floors, and swaptions are viewed as options on forward interest rates.
Klaassen et al. (1998) showed that a forward interest rate can be written
as a weighted difference between two zero-coupon interest rates to account
for the fact that volatility depends on the correlation between zero-coupon
rates. Let us denote by f t1 t2 the forward rate between two instants as:
er1 t1 eft1 t2 (t2 − t1 ) = er2 t2
where r1 , (r2 ) refers to the continuously compounded zero-coupon rate for

maturity r1 , (r2 ). This forward rate can be written as a weighted average
between two zero-coupon rates:
r2 t2 − r1 t1
ft1 t2 = (B.1)
t2 − t1
In this context, the volatility of the forward rate is given by:

t21 σ(r1 )2 + t22 σ(r2 )2 − 2t1 t2 ρ(r1 , r2 )σ(r1 )σ(r2 )
σ(ft1 t2 ) = (B.2)
t2 − t 1
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:
δσ(ft1 ,t2 ) t1 (t1 + ∆) σ(r1 )σ(r2 )

=− (B.3)
δρ(r1 , r2 ) ∆2 σ(ft1 ,t2 )
Klaassen et al. (1998) estimated correlations between zero-coupon interest

rates from daily movements of Deutschmark swap rates for the period 1993–
1998. They found that when volatilities of zero-coupon rates equal their
empirical values, the volatilities of forward rates (and option prices) are
severely underestimated by one-factor models. When market prices of caps,
floors, and swaptions are used in the calibration, model parameters must be
chosen such that the forward-rate volatility implied by the option price is
matched by the model very closely. Hence, volatilities of zero-coupon rates
must be chosen higher than that they are in reality. The analysis reveals
how the implicit assumption of perfect correlation between zero-coupon
interest rates in one-factor models affects the pricing of options.
Appendix C: Merton’s Model in the Presence of Stochastic

Interest Rates
Let C(S, P, τ, K) be the option price function depending on the stock price
S, a bond price P , the strike price K, and the time remaining to maturity
τ with τ = T − t, where T is the maturity date and t is the current
time. The stock price dynamics are represented by the stochastic differential
equation:
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.
The bond price dynamics are given by the following equation:
dP
= µ(τ )dt + δ(τ )dq(t, τ ) (C.2)
P
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.
Assume that there is no serial correlation among the returns on any of

the assets:
dq(s, τ )dq(t, T ) = 0 for s#t;

dq(s, τ ))dW (t) = 0 for s#t
and
dq(t, τ )dq(t, T ) = ρτ T dt
where ρτ T may be less than 1 for τ #t.

When the interest rate is constant over time, this means that δ = 0,
µ = r, and p(t) = e−rτ .
Assuming that investors agree on the values of (δ, σ), Ito’s lemma gives
the change in the option price over time.

∂C ∂C ∂C
dC = dS + dP + dτ
∂S ∂P ∂τ

2 2 2
1 ∂ C 2 ∂ C 2 ∂ C
+ (dS) + (dP ) + 2 dSdP (C.3)
2 ∂ 2S ∂2P ∂S∂S
Using the properties of stochastic calculus, the values of (dS)2 , (dP )2 , and
(dS)(dP ) are given by:
(dS)2 = (αSdt + σSdW )2 = σ2 S 2 dt (C.4)

(dP )2 = (µP dt + δP dq)2 = δ 2 P 2 dt (C.5)
(dS)(dP ) = (αSdt + σSdW )(µP dt + δP dq)
= σδSP (dW )(dq) = δρσSP dt
where ρ corresponds to the instantaneous correlation coefficient between

the returns on the stock and on the bond.
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

2
∂C ∂C ∂C 1 ∂ C
dC = (αS) + (µP ) − + σ2 S 2
∂S ∂P ∂τ 2 ∂2S
2 2
1 ∂ C ∂ C
+ δ2P 2 + σδρSP dt
2 ∂ P
2 ∂S∂S

∂C ∂C
+ σSdW + δP dq
∂S ∂P
or in a simple form:
dC = βCdt + γCdW + ηCdq (C.6)
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
The expression for β represents the instantaneous expected return on the

option. Consider now a hedged portfolio with the underlying asset, the
option and risk-less bonds where the portfolio weights w1 , w2 , and w3
are chosen to eliminate market risk. The aggregate investment in the
portfolio y is zero when investors are allowed to use proceeds from short
sales and to borrow without restrictions to finance long positions, so
that: w1 + w2 + w3 = 0. In this context, the instantaneous dollar return
dy, may be written as:

dS dP
dy = w1 + w2 [βdt + γdW + ηdq] + [−w1 − w2 ]
S P
or
dy = w1 [αdt + σdW ] + w2 [βdt + γdW + ηdq] + [−w1 − w2 ][µdt + δdq]
Re-arranging these terms yields:
dy = [w1 (α − µ) + w2 (β − µ)]dt + (w1 σ + w2 γ)dW

+ [ηw2 − (w1 + w2 )δ]dq (C.7)
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
This linear system presents a solution, if and only if:

(β − µ) γ (δ − η)
= = (C.9)
(α − µ) σ δ
If Eq. (C.9) holds, then:
γ η
= 1− (C.10)
σ δ
This implies from the definition of γ and Eq. (C.6) that:

S ∂C P ∂C
=1−
C ∂S C ∂P
or

∂C ∂C
C =S +P .
∂S ∂P
Using Eq. (C.17) gives:
γ(α − µ)
(β − µ) =
σ
Re-call the definitions of β and γ in (C.6), we have:

1 2 2 ∂ 2C 1 2 2 ∂ 2C
0= σ S + δ P
2 ∂ 2S 2 ∂ 2P
2
∂ C ∂C
+ σδρSP + (µS)
∂S∂S ∂S

∂C ∂C
+ (µP ) − − µC (C.11)
∂P ∂τ
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)
This is a second-order linear partial differential equation of the parabolic

type. The price of any option in the Merton’s economy must satisfy
Eq. (C.12). In particular, the price of a European call must satisfy this
equation and the following boundary conditions: C(0, P, τ, K) = 0 and
C(S, 1, 0, K) = max[0, S − K].
Using the change in variables, x = KPS(τ ) , let us re-write the change in
variables as follows:
S u
x= = ;
KP (τ ) v
dS
= αdt + σdz
S
and
dP
= µ(τ )dt + δ(τ )dq(t, τ ).
P
The system can also be written in a matrix form.
Taking the partial derivatives with respect to u, v, uu, vv, uv, gives

∂f 1 1 ∂f −1 1
= = , = 2 = ,
∂u v P (τ ) ∂u∂v v P (τ )2
2 2
∂ f 2u 2S ∂f −u −S ∂ f
= = , = 2 = , = 0.
∂2v v KP (τ ) ∂v v P (τ )2 ∂ 2u
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
If we substitute for dS and dP in this last equation, we have:
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
dx = αxdt + σxdz − µxdt − δxdq + δ 2 xdt − xσδηdt
or
dx
= αdt + σdz − µdt − δdq + δ 2 dt − σδηdt
x
Isolating the drift terms gives:
dx
= (α − µ + δ 2 − σδη)dt + σdz − δdq
x
It is clear that the expected instantaneous return on x is a function of S

and P and that the variance ν 2 (τ ) = σ 2 + δ 2 + 2σδη depends only on τ . It
is possible to obtain a solution to this last equation using:
h(x, τ ; K) = C(S, P, τ, K)
and using (H, x) in lieu of (H, S) to get the following system:
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 )
This last system can be written as

1 2
x y11 − y2 = 0, y(0, T ) = 0, h(x, 0) = max[0, x − 1]
2
Using the inverse change of variables and simplifying gives:

t
S
C(S, P, τ, K) = KP (τ )y ; ν 2 (s)ds
KP (τ ) −0
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
and where erf c(.) is the error complement function.

h
2 w2
erf c(h) = 1 − √ e− 2 dw
2Π 0
In the special case, when r = 0, σ2 = 1, K = 1, Eq. (C.13) is identical to

the B–S equation. The call price in the context of Merton’s model can be
written in a B–S form as:
c = SN (d1 ) − P (τ )KN (d2 )

S
ln K − ln(P ) + 12 στ2 τ √
d1 = √ , d2 = d1 − στ τ
στ τ
In the same context, the put price is given by:
p = −SN (−d1 ) + P (τ )KN (−d2 )

S
ln K − ln(P ) + 12 στ2 τ √
d1 = √ , d2 = d1 − στ τ
στ τ
with
τ
1
P (τ ) = e−rτ and στ2 = [σ 2 + δ 2 − 2ρσδ]du.
τ 0
References
Black, F (1995). Interest rates as options. Journal of Finance, 50 (December),
1411–1416.
Brennan, MJ and E Schwartz (1977). Saving bonds, retractable bonds, and
callable bonds. Journal of Financial Economics, 5, 67–88.
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.
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.

Merton, R (1987). An equilibrium market model with incomplete information.
Stutzer, M and M Chowdhury (1999). A simple non-parametric approach to bond
futures option pricing. Journal of Fixed Income, 8(4), 67–76.
Chapter 16
MODELS OF INTEREST RATES,

INTEREST-RATE SENSITIVE INSTRUMENTS,
AND THE PRICING OF BONDS:
THEORY AND TESTS
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
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.
Models of Interest Rates, Interest-Rate Sensitive Instruments 705
16.1. Interest Rates and Interest-Rate Sensitive

Instruments
16.1.1. Zero-coupon bonds
A zero-coupon bond or a discount bond of maturity “T ” refers to a financial
security, which pays one unit of cash (a dollar for example) at a future
prespecified date T . Hence, the bond’s nominal value (principal or face
value) is one dollar. We denote by B(t, T ) the price at time t of a zero-
coupon bond of maturity T . At the maturity date, B(T, T ) = 1. Since the
maximum value of this bond is unity, the bond trades at a discount (a value
less than the principal).
16.1.2. Term structure of interest rates

Consider a zero-coupon bond with a maturity date T less than T ∗ . The
simple rate of return from holding the bond until the maturity date can be
calculated as the difference between the final value and the current value is
divided by the current value or:
1 − B(t, T ) 1
= − 1.
B(t, T ) B(t, T )
Using continuous compounding, the equivalent rate of return or the

yield to maturity (YTM) on a bond is given by:
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:
B(t, T ) = e−(Y (t,T ))(T −t) , ∀ t ∈ [0, T ]. (16.2)
In this expression, the bond price is related to its maturity by a discount

function. At the initial time, the term structure of interest rates may be
described by the current bond prices B(0, T ) or by the initial yield curve:
B(0, T ) = e−Y (0,T )T , ∀ T ∈ [0, T ∗]. (16.3)

This term structure is estimated in practice using different instruments

and methodologies. The yield curve exhibit different shapes. It may be flat,
upward sloping, downward sloping, or humped.
16.1.3. Forward interest rates

Let us denote by f (t, T ) the forward interest rate at date t < T for
risk-free operations (borrowing and lending) at date T . The instantaneous
continously compounded forward rate corresponds to a “non observable”
interest rate over very short-time intervals [T, T +dT ] as viewed from time t.
HJM used an exogenous specification of a family of forward rates f (t, T )
and has defined the bond prices as:

T
B(t, T ) = exp − f (t, u)du , ∀ t ∈ [0, T ]. (16.4)
t
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 )
This expression gives the forward rate as:
ln B(t, T ) − ln B(t, H)
f (t, T, H) = . (16.6)
H −T
Since, the strategy of investing at t in T -maturity bonds is equivalent to the

strategy of lending over the interval [t, T ], it is clear that Y (t, T ) = f (t, t, T ).
In real markets, interest rates are quoted on an actuarial basis. For example,
the actual rate referred to also as the effective rate ra (t, T ) at t for maturity
T is given by:
(1 + ra (t, T ))T −t = ef (t,t,T )(T −t) = eY (t,T )(T −t) , ∀t ≤ T

where the standard bond price is given by:
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:
(1 + ra (t, T, H))H−T = exp[f (t, T, H)(H − T )] = B(t, T )/B(t, H).
16.1.4. Short-term interest rate

The short-term interest rate or the instantaneous interest rate rt is often
modeled as a price process of a risk-free asset:
t
Bt = exp ru du , ∀ t ∈ [0, T ∗ ] (16.7)
0
where the function B(t) solves the differential equation:
dBt = rt Bt dt
with the initial condition B0 = 1. Since Bt corresponds to the cash

accumulated up to t using one unit cash at time 0, and rolling over successive
periods, the process is also known as the accumulation factor or a savings
account.
16.1.5. Coupon-bearing bonds

A coupon-bearing bond or coupon-paying bond BC (T ) gives its holder
different cash-flows (c1 , c2 , . . . , cm ) at different dates (T1 , T2 , . . . , Tm ).
The price of this bond is obtained by discounting all the future cash flows
at the appropriate rates.
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
a coupon-bearing bond is uncertain because of the re-investment risk since

interest rates are not constant. Bonds with different coupons and time to
maturity are in general not directly comparable. To overcome this difficulty,
the concept of YTM is used.
16.1.6. Yield-to-Maturity (YTM)

The price of a bond with fixed equal coupons and an uncertain YTM at
time 0, Ỹc (0) is given by:
m
c N
Bc (0) = + .
j=1
(1 + Ỹc (0))j (1 + Ỹc (0))m
Using the coupon rate c = rc N , the bond price is:
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
In a continuous setting, the initial price of a zero-coupon bond with

maturity T and a YTM, Y (0, T ) is: B(0, T ) = e−Y (0,T )T . The discretely
compounded YTM at time i < m, Ỹc (i) on a coupon bond with “m” cash
flows can be calculated using the formula:

m
cj
Bc (i) = . (16.9)
j=i+1
(1 + Ỹc (i))j−i
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
t < Tm , Yc (t) = Yc (t; c1 , . . . , cm , T1 , . . . , Tm )
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.
16.1.7. Market conventions

Market conventions can vary from one country to another. In the United
States, the US Treasury issues bonds, notes and bills. 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 T -bonds and T -notes are coupon paying securities, which
differ in general by the maturity date. The maturity for T -bonds is more
than 10 years and it is longer than that of T -notes. A m-year government
bond pays coupons semi-annually at time Tj = 0.5j, for j = 1, 2, . . . , 2m.
The YTM on a government bond, Ŷc (0) called also a bond equivalent yield
is calculated from a bond price as follows:
2m−1
rc N/2 (1 + rc /2)N
Bc (0) = + , (16.11)
j=1 (1 + Ŷc (0)/2))j (1 + Ŷc (0)/2)2m
where:
rc : coupon rate;
N : bond face value and
Ŷc (0): an annualized interest rate with no compounding.
This formula can also be written as:
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:
Ŷce (0) = (1 + Ŷc (0)/2)2 − 1.
16.2. Interest Rates and the Pricing of Bonds

The pricing of bonds and interest-rate options needs a specification of the
dynamics of interest rates under certainty and uncertainty.
16.2.1. The instantaneous interest rates under certainty

When an investor borrows $1 at time t, he must pay F (t, T ) at time “T ”
where T is the maturity date of the debt. This amount corresponds to an
average interest rate R(t, T ) which applies over the period of [t, T ]. It is
given by:
F (t, T ) = eR(t,T )(T −t) .
In the context of certainty, when the interest rates are known over the
period, the function F is given at each instant by:
F (t, s) = F (t, u)F (u, s) for all t < u < s.
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)
So,

T
1
R(t, T ) = r(s)ds . (16.13)
(T − t) t
Consider a zero-coupon bond, which is a bond paying $1 with certainty at

its maturity date, P (T, T ) = 1. The price of this bond at time t, P (T, t)
in the context of certainty (in an economy with no risk, where all future
interest rates are known) is given by:
RT
P (t, T ) = e(− t
r(s)ds)
. (16.14)
In a context of uncertainty, the future interest rates, R(u, T ) are unknown

and the models of interest rates must be established to find the relationships
between these rates.
16.2.2. The instantaneous interest rate under uncertainty

Under uncertainty, the instantaneous interest rate r(t) is a stochastic
process between times t and t + dt. If we consider a risk-less asset, then
its price is given by:
Rt
B(0, t) = e(− 0
r(s)ds)
.
Denote by H the following assumption.

Consider a process P (t, u)0≤t≤u satisfying the boundary condition
P (u, u) = 1. Then, as for the processes of stock prices, it can be shown,
with no arbitrage opportunities, that there is a probability P ∗ equivalent
to the probability P under which the process
Rt
P̂ (t, u) = e(− 0
r(s)ds)
P (tu)
is a Martingale for each u in time interval [0, T ].

This is an interesting assumption since, under the new probability P ∗ ,
we have:
Ru
P̂ (t, u) = EP ∗ (P̂ (u, u) | Ft ) = EP ∗ (e− 0
rs ds
| Ft )
or
Ru
P (t, u) = EP ∗ (e− t
rs ds
| Ft ). (16.15)
If one compares this formula with formula (16.14), we notice that

the prices of zero-coupon bonds depend only on the process P (t, u)0≤t≤u
under the probability P ∗ . Assumption H allows the identification of the
probability density of P ∗ , denoted, LT , with respect to P . In this context,
it is possible to show that there exists a stochastic process q(t)0≤t≤T such
that for all t in the interval [0, T ]
Rt Rt
q(s)dWs − 12 q 2 (s)ds)
Lt = e(− 0 0 .
Using the property:
EP ∗ (X | Ft ) = E(XLT | Ft )/Lt ,
it is possible to show that the price at time t of a zero-coupon bond with a

maturity date u is:
u u u
1
P (t, u) = E exp − r(s)ds + q(s)dWs − q 2 (s)ds | Ft .
t t 2 t
(16.16)
The probability is often referred to as the risk-neutral probability.
16.3. Interest Rate Processes and the Pricing

of Bonds and Options
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 pats of the spot rate. This implies the
search for parameter values, which cause calculated zero-coupon bond prices
to be close to market prices. Many other popular spot rates can be expressed
in a simple way under the Heath et al. (1992) model. Once these problems
are solved, the option value can be calculated using the appropriate
boundary conditions. A European bond option with a maturity date on a
zero-coupon bond with a maturity date τ is an option characterized by its
terminal payoff. For a call option, this payoff is given by c = (P (τ, T )−K)+
For a put option, this payoff is given by c = (K − P (τ, T ))+.
16.3.1. The Vasicek model

The presentation here uses Eqs. (16.15) and (16.16). Vasicek (1977) used
the following process for the dynamics of the interest rate r(t)
dr(t) = a[b − r(t)]dt + σdWt (16.17)
where a, b and σ are constants. If we assume that the process q(t) is a

constant equal to −γ, then:
dr(t) = a[b∗ − r(t)]dt + σdŴt ,
with
γσ
b∗ = b −
a
and
Ŵt = Wt + γt.
The process in the Eq. (16.17) can be shown as an Ornstein–Uhlenbeck

process. The valuation of bond options in this context, can be easily done
since, the Ornstein–Uhlenbeck process is a Gaussian process.
16.3.2. The Brennan and Schwartz model

The model proposed in Brennan and Schwartz (1982) is a two-factor model
where the dynamics of the short-rate are given by:
dr = (a1 + b1 (r − l))dt + σ1 rdX1 .
The long-term rate dynamics are given by:
dl = l(a2 + b2 r + c2 l)dt + σ2 ldX2 .
The drift terms have mean-reversion features. The parameters have to be

selected using empirical data. However, this model can allow for infinite
interest rates.
16.3.3. The CIR model

In the context of this model, the dynamics of the instantaneous interest
rate are described by the following equation:

dr(t) = [a − br(t)]dt + σ r(t)dWt
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.
16.3.4. The Ho and Lee model

The Ho and Lee (1986) model corresponds to a new generation of interest-
rate option pricing model. This model exists in a discrete-time version and
in a continuous time version. The continuous-time version of the model
can be written as: dr = θ(t) + σdW . This model belongs to the class
of term structure option models which try to model the behavior of the
entire yield curve. These models eliminate in general risk-less arbitrage
opportunities that arise from inconsistencies between the model and the
observed yield curve. The Ho and Lee (1986) one factor model is based
on an implicit assumption that bond prices of all maturities are perfectly
correlated. Besides, it assumes a constant volatility regardless of the interest
rate level. This allows for the existence of negative interest rates with a
positive probability. These deficiencies are avoided in the HJM model. In
this model, the initial yield curve is endogenous since it is derived from
the specified interest rate process. Therefore, the parameters driving the
interest rate process must be chosen in a way to fit the initial yield curve.
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 the rates
along with the yield curve fluctuate to the same degree. This assumption
is not supported in practice.
16.3.5. The HJM model

The main drawbacks of the Vasicek (1977) and the CIR (1985) models are
that these models do not account for the observed term structure of interest
rates. Ho and Lee (1986) proposed an interesting discrete-time model, which
is appropriate for the description of the whole term structure of interest
rates. The same ideas were used in a continuous-time version of this model
by Heath et al. (1989), hereafter, HJM. The HJM, interest-rate models are
based on a new approach to interest-rate option pricing. As in the Black–
Scholes (B–S) model, the HJM model requires the underlying asset (the
initial term structure) and a measure of its volatility as the only inputs.
The HJM model uses all the available information in the term structure.
The model accounts for several reasons of the term structure movements.
It uses a methodology, which is often applied in multifactor models of the

term structure risk. The HJM, model is not based on the assumption of
perfect correlations between bond prices of all maturities and does not
allow for negative interest rates. The main features of the HJM model is
the use of an exogenous approach for the initial yield curve, the multifactor
modeling and the volatility functions. In this model, the initial yield curve
is exogenously given, which eliminates the parameter estimation. Since,
the model is very general, it contains the standard models of Ho and Lee
(1986), Vasicek (1977), Hull and White (1990), and Cox et al. (1985) as
special cases. The HJM model uses volatility functions to conform the
empirical volatility behavior by making the volatility of a given forward
rate as a decreasing function of the time to its effective date. This type
of models forces the long-term rates to fluctuate less than the short-term
rates. The use of the volatility functions allow it to overcome the problem of
negative interest rates. In the continuous-time model of HJM, the returns of
zero-coupon bonds of different maturities are not perfectly correlated and
volatility functions are directly calculated from data of changes in the term
structure of interest rates. Since, forward rates are more stable than the
prices of zero-coupon bonds, they are used in a two-factor model. The two
factors are the changing level and the changing slope of the term structure.
The two stochastic processes corresponding to these two factors prevent the
perfect correlation between bond prices or forward rates. The model has
many similarities with the B–S model since, it only needs the knowledge
of the underlying term structure and its associated volatilities. The use of
multiple volatility functions gives the model a certain flexibility since, it can
accomodate the volatility of the term structure resulting from changes in
the level, the slope and the curvature of the term structure. In this model,
option values depend only on s. This gives a situation, which is analogous
to the B–S. More formally, denoted by f (t, u) the forward interest rate, i.e.,
the rate at which an investor can contract at date t to borrow and lend for
a short period at a future date u. The initial forward rate curve is taken
from market data and used as an input in the HJM model. The volatility
of each forward rate is specified in the model. The forward rate for date u
can change over the instant of time from t to (t + dt) in the following
way f (t + dt) = f (t, u) + σ(.)dW + drif t(.)dt. The volatility function
σ(.) of each forward rate describes the dynamics of the term structure.
When σ(.) is a constant, all the forward rates have the same volatility and
the entire forward curve is submitted to parallel shocks. In this context,
the HJM model is the continuous time limit of the Ho and Lee model.
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:
C(t) = P (t, T )N (h) − P (t, t∗ )KN (h − q)

P (t,T ) 1 2 (16.18)
ln KP (t,t ) + 2 q
∗ √
h= , q = σ(T − t∗ ) t∗ − t
q
and where P (t, T ) is the bond with an exercise price K and a maturity t∗
with:
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).
16.3.6. The BDT model

The Black, Derman and Toy model (1990), can be written as:
dr
= −α(t)dt + σ(t)dW.
r
In the BDT model, the long-term yields are assumed to reflect the
expectations of the market regarding the future short-term rates. In this
context, if the market expects future short rates to be high (low), then there
is a tendency for current long-term yields to be higher (lower). The model
gives in a valuation formula current expected long-term yields, using the
expected future short rates. It also offers a sensitivity formula, that predicts
the expected changes in the current long-term yields when the short rate
changes.

The Hull and White one-factor model (1990) tries to overcome some of the
main shortcomings in the Ho and Lee (1986) model. The dynamics of the
short-term interest rate are given by:
dr = [θ(t) − αr]dt + σ(t)dW
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:
c = P (0, T )N (d1 ) − KP (0, t∗ )N (d2 )
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.
The reader can note the analogy between the Hull and White and the
HJM formulas.
16.3.8. Fong and Vasicek model

The Fong and Vasicek model proposed in Fong and Vasicek (1991) is a
two-factor model where the dynamics of the risk-adjusted spot rate are
given by:
√
dr = a(r̄ − r)dt + σdX1
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.
16.3.9. Longstaff and Schwartz model

The Longstaff and Schwartz model proposed in Longstaff and Schwartz
(1992) is a two-factor model where the dynamics of the risk-adjusted
variable is:
√
dX = a(x̄ − x)dt + xdX1
and
√
dy = a(ȳ − y)dt + ydX2
where the spot rate is given by r = cx + dy.
16.4. The Relative Merits of the Competing Models

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 et al. (1990). The process for the short rate in the
Ho and Lee (1986) model is given by:
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:
dr = (θ(t) − φ(t)r)dt + σ(t)dz.
This model assumes that interest-rate changes are normally distributed.

HJM, shows that the model for the short rate is completely determined
using the term structure of interest rates and the volatilities of all forward
rates. The function θ(t) can provide a perfect fit to the current term
structure of interest rates. In the Ho and Lee model, the volatility of the
term structure is described by the constant σ. Hence, the model ignores
the differences between the volatility of short and long-term spot rates. In
practice, long-term spot rates are less volatile than short spot rates.
In the BDT model, the volatility of the term structure is described
by σ(t). Hence, the model can provide a perfect fit to the current term
structure of spot rate volatilities. In the Hull and White model, the volatility
structure is described by two functions of time: σ(t) and φ(t). The second
function provides an additional degree of freedom with respect to the BDT
model. Mean reversion comes from the fact that there is a “link” between
current long-rate volatilities and future short-rate volatilities. It refers to
the tendency for short rates to be pulled back to some long-run average over
time. Mean reversion explains also the fact, that long-rate volatilities are
less than short-rate volatilities. When the interest rates are high, it is more
probable that they decrease. When they are low, it is more probable that
they increase. In economic theory, when interest rates rise, the economy
slows down since, loans are less demanded. When interest rates fall, the
economy heats up since loans are more demanded. Mean reversion exists
in the BDT model and Hull and White model, but not in the Ho and Lee
model. Hull and White find that their model is appropriate for the pricing of
at-the-money options. The log-normal assumption for interest rates seems
to be better than the normal assumption since it does not allow for negative
interest rates. However, the analytic tractability of the option pricing model
can be lost in this context. It is also possible to use the normal assumption
and adjust the volatilities of out-of-the-money and in-the-money options,

in order to account for the skewness of the interest rate distribution. Leong
(1998) reviews the relative merits of the competing models in the pricing of
interest rate sensitive instruments. He uses six criteria, which are necessary
in the choice between different models. The first criteria concerns arbitrage-
free pricing. In this context, a correct model must at least price correctly,
an instrument relative to the prices of the other related instruments. In
this sense, to avoid arbitrage opportunities, the model must preserve the
put-call parity and must be consistent with the observed term structure of
interest rates. When comparing Hull and White’s model and HJM model,
Leong (1998) finds that both models price options “correctly” with respect
to the initial yield curve. In fact, these models preserve the put-call parity.
When comparing Black’s (1976) model to the HJM model, HJM shows that
Black’s model is inconsistent when used in the pricing of caps of different
terms because of the use of different volatilities. However, this same problem
appears as well when using the HJM model. In general, dealers adopt a
volatility matrix approach to pricing and portfolio revaluation. This matrix
is based on different estimates for options with different time to maturities.
This practical approach leads to inconsistencies and incoherence regardless
of the model used. The second criteria corresponds to the computation
speed. Traders, market makers, and risk managers resort to option pricing
models to price the deals, to re-value on a mark-to-market basis, a trading
book, to calibrate the models to observed data, to manage the risks of a
trading position, and so on. Since, the model may be used several times
during a trading day, it must respond to a certain computation speed.
The main drawbacks of the HJM two factor model is that it is too slow.
In general, the speed problem in deal pricing is translated into serious
problems in model calibration. Therefore, as a book of interest-rate sensitive
instruments using a two-factor model may or may not be viable. The slower
computation speed causes the two-factor models to be at a disadvantage
with respect to one-factor models regardless of the state of the technology.
Leong (1998) finds that the pricing of a five-year cap on Sun Sparc two
station using the two-factor HJM model is 17 times as long as for a one-
factor model. The third criteria concerns the hedge effectiveness or the
model’s ability to suggest hedge ratios, which contribute to reduce the risk
exposure of the model user. The multifactor approach allows the model to
explain more of the variance in yield curve changes. When compared to a
one-factor approach, the multifactor appraoch suggests hedges, which are
more robust to nonparallel yield curve changes. The one-factor approach
ignores in general the lack of perfect correlations between different rates

in the yield curve. Since two-factor models compute two deltas for each
instrument, and lead to using two hedging instruments, this refinement
can lead to better hedge recommandations. However, the two-factor HJM
model may not be perfect in hedging yield curve risk. The fourth criteria
refers to the marginal utility. The “marginal utility of complexity” refers
to the practical benefit that can be achieved by adding another dimension
of complexity. A “good” model can do practically an equivalent task with
the least complexity. The fifth criteria refers to versatility and coherence. A
versatile model is able to price several instruments with the same analytical
framework. This framework is fundamental for integrated risk management.
The sixth criteria concerns with the fitting error. The theoretical value
generated by a model must be close to the market value. Hence, a good
model must be easily calibrated to reflect the general market level.
16.5. A Comparative Analysis of Term Structure

Estimation Models
Ferguson and Raymar (1998) used the bond prices to estimate six dis-
count functions: a six-degree polynomial, four-and-six-parameter Vasicek
and Fong (1982) models (VF4 and VF6); three four-parameter analytic
techniques of Nelson and Siegel, NS (1987); the Vasicek (1977) bond
price equation and the Cox et al., CIR (1985) bond equation. They
compared the six methods with respect to accuracy in a simulation
framework. They found that a simple OLS bond price application of the six
parameter Vasicek–Fong (1982) model is the most accurate and robust. The
polynomial estimator is accurate only when bond maturities are sampled at
uniform intervals. They show that the Nelson–Siegel model is also reliable
and accurate, and that the Vasicek (1977) and Cox et al. (1985) models are
not “as good” as the other methods.
16.5.1. The construction of the term structure

and coupon bonds
Ferguson and Raymar (1998) generated realistic term structures using a
forward rate generator (the following equation) to specify a set of cross-
sectional rate changes in the log of the forward rate:
√
∆ ln(f ) = a[m − ln(f )]∆t + σ ∆tz̃ (16.19)
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 study uses three-month time intervals over a 30 year horizon.

Hence, a single trial corresponds to 120 random draws z allowing the
definition of a forward rate curve. The initial forward rate f0 is 7% and
σ = a = 15%. The values of m are set to 7%, 10% and 4% to get respectively
flat, upward, and downward sloping spot curves. They generate for each
yield curve (shape), 50 forward rate curves. Using a forward curve, it is
simple to compute the discount function, the spot rate curve and the coupon
bond prices.
16.5.2. Fitting functions and estimation procedure

Using a set of 15 bond prices, Ferguson and Raymar (1998) estimated a
continuous discount function out to 30 years by each of six fitting functions.
They used a discount curve to get “fitted prices” for the bonds for which
they calculated simulated prices. Then, they minimize the sum of the
squared errors over the 15 true and fitted bond prices to get the parameters
of the continuous discount functions. Ferguson and Raymar (1998) used a
constrained OLS for three of the estimates: a six-degree polynomial and
four-and-six-parameter Vasicek-Fong (1982) models (VF4 and VF6). Non-
linear least square methods are used in the other estimation approaches.
The parameters of the polynomial and Vasicek-Fong models are estimated
subject to a constraint to price an instantaneously maturing $1 face value
discount bond at $1. The six term structure estimates or fitting functions
are defined as follows.
Following Ferguson and Raymar (1998), we have denoted by:
Dt : present value at time 0 of a bond that matures at time t;

ai : parameters used to fit the term structure in the presence
of the constraint that D0 = 1;
R: yield to maturity (YTM) on the longest bond used when fitting
the term structure and
rt : spot rate for a bond with a maturity date t.
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 .
The constraint equation in this context corresponds to a0 = 1. This model

is standard.
The Vasicek–Fong (1982) component functions are specified by VF
(1982) as appropriate. The Vasicek–Fong (1982) model (VF6) is given by:
Dt = a0 + a1 e−Rt + a2 e−2Rt + a3 e−3Rt + a4 e−4Rt + a5 e−5Rt .
The constraint equation in this context corresponds to Σai = 1.

The Vasicek–Fong (1982) model (VF4) is given by:
Dt = a0 + a1 e−Rt + a2 e−2Rt + a3 e−3Rt .
The constraint equation in this context corresponds to Σai = 1.

The Nelson–Siegel (1987) model is given by:
Dt = e−rt t
with

1 − e−a3 t
rt = a0 + (a1 + a2 ) − a2 e−a3 t .
a3 t
The constraint equation in this context is endogenous.

This model is presented in terms of the spot rates and the discount
function.
The Vasicek (1977) model is given by:
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

The short-rate process used is:
da0 = a1 (a2 − a0 )d3 t + a3 dz.
The CIR (1985) model is given by:
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 short-rate process used in this context is:
√
da0 = a1 (a2 − a0 )dt + a3 a0 dz.
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.
16.6. Term Premium Estimates From Zero-Coupon Bonds:

New Evidence on the Expectations Hypothesis
The concept of a liquidity premium implicit in the term structure of interest
rates was first proposed by Hicks (1946). Empirical work by Fama (1984a,
1984b, 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) found
that forward rates are poor forecasters of future short-term interest rates.
However, forecasts improve for longer forecast horizons. Dhillon and Lasser
(1998) reexamined the presence of term premiums in the term structure
and the issue of forecasting future interest rates from current forward rates.
They used a unique data set of zero-coupon stripped Treasury securities.
Since, Treasury bond issues have different characteristics regarding the
maturities and the coupons, Fama (1984) constructs portfolios of bonds
for maturities longer than a year. Fama and Bliss (1987) implemented
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,
St+1 : P τt+1 = Rτt+1 − St+1 .
The price of a zero-coupon maturing in τ months at any time t is

given by:
Bτt+1 = exp(−St+1 − F 2t − · · · − F τt )
where F τt corresponds to the forward rate for quarter t + τ observed at

time t. This forward rate can be determined using:

B(τ − 1)t
F τt = ln .
Bτt
The expectations hypothesis suggests that forward rates contain forecasts

of the future spot rates. The following two time series regressions are used
to determine whether the current forward-spot differential (F τt − St+1 )
represent a predictor of either the future quarterly premium P τt+1 or the

future change in spot rates
(St+τ − St+1 ): P τt+1 = α1 + β1 (F τt − St+1 ) + εt+1

(16.21)
St+τ − St+1 = α2 + β2 (F τt − St+1 ) + nt+τ −1 .
When β1 is positive and significant, then forward rates contain information

about the premium in three months. Also, when β2 is positive and
significant, then the forward rate at time t is a good predictor of the future
quarterly spot rate at time t + τi − 3.
Results show strong evidence for liquidity premiums in the term struc-
ture as well as a monotonically increasing relationship between liquidity
premium and term to maturity.
Results show that current forward rates can be used to forecast
quarterly interest rates. This evidence is in contrast with the findings in
Fama and Bliss (1987) where it is shown that short-term rates cannot be
forecast using forward rates.
16.7. Distributional Properties of Spot and Forward

Interest Rates: USD, DEM, GBP, and JPY
Using the Kernel density estimation method, Lekkos (1999) estimates the
distributions of spot and forward interest rates in levels and differences.
He studies normal and log-normal distributions as well as a mixture of
two distributions in the characterization of the distribution of interest-rate
changes. The database comprises daily money market and swap market
rates of the USD, the DEM, the GBP, and the Yen. Lekkos (1999) uses
money market and swaps market data to infer the prices and the yields
of zero-coupon bonds. He uses the bootstrap method to get daily spot and
forward term structures of interest rates from the rates and maturities in the
database. Lekkos (1999) uses money market and swaps market data to infer
the prices and yields of zero-coupon bonds. Money market rates data (short
rates with less than one year to maturity) are used to price zero-coupon
bonds. Swap market data are used for longer maturities. Money market
rates corresponding to the rates at which banks lend or borrow money for
three, six, and 12-months. The discount bond price for maturities upto one
year are estimated using:
1
B0,t = (16.22)
1 + rt αo,t
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
where the accrual factors referring to bonds are:
(ti − ti−1 )
αi−1,i = .
365
The one-year forward rates are estimated using the following equation:
−αt,t+12m
B0,t
f0,t,t+12m = − 1.
B0,t+12m
The Kernel method used by Ait–Sahalia (1996) and Lekkos (1999)

allows the identification of the “true” non-parametric distribution of
interest rates. The estimated density is constructed by centering around
each observation of a kernel function K(u) and averaging the values of the
kernel function at any given rate. Lekkos (1999) uses the following estimator
for the density function:
T

r − rt
fˆ(r) = (T h)−1 K
t=1
h
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.
16.7.1. Interest rate levels

The data shows for most spot and forward rates, the presence of multi-
modality. This reveals that interest rates are not generated from a univariate
distribution (normal or log-normal). They are generated from a mixture of
distributions with different means and volatilities. These shapes can be
explained by regime shifts or jumps in interest rates.
Similar patterns are observed in Bellalah and Prigent (2002) for stock
and index options.
16.7.2. Interest rate differences and log differences

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. Black et al. (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
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
where Π = (Π1 , Π2 , . . . , ΠN −1 ) represent the N − 1 independent mixing

proportions of the mixture and are such that:
N
−1
0 < Πi < 1, ΠN = 1 − Πi
i=1
and µ = (µ1 , . . . , µN ) and σ = (σ12 , . . . , σN

2
).
For the case of a mixture of normal distributions, the probability
density function of the ith component distribution is given by:

1 −(∆rt − µi )2
gi (∆rt ; σi , µi ) = √ exp .
2Πσi 2σi2
The parameter vector of the mixture,
ϑ = (µ1 , µ2 , . . . , µN , σ12 , . . . , σN
2
, Π1 , Π2 , . . . , ΠN )
can be estimated by maximizing the log-likelihood function:

T

L(ϑ) = log f (∆rt , ϑ). (16.24)
t=1
Since the maximization of the log-likelihood function L(ϑ) is not evident

with standard methods, the parameter vector was estimated via the
expectation maximization (EM) algorithm proposed by Dempster et al.
(1977). This algorithm is also used in Bellalah and Lavielle (2002). The EM
algorithm assumes the presence of the components in a fixed proportion
in the mixture. Then, it determines the posterior probability of that

observation which belongs to the component distribution s as:
Πs gs (∆rt ; σs , µs )
P (s | ∆rt ) = (16.25)
f (∆rt ; Π, σ, µ)
The maximum likelihood estimate of ϑ̂ is a solution to the following system

of non-linear equations:
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 Pîst=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:
log Λij = log Li (ϑ) − log Lj (ϑ) for i ≤ j. (16.29)
He makes inferences about which distribution is more likely to have

generated in each observation. Given ϑ, when the number N = 2, then:
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.
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
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.
Appendix A: An Application of Interest Rate Models

to Account for Information Costs:
An Exercise
As an exercise, this section introduces a parameter reflecting information
costs in interest-rate models. This allows a presentation of the old formulas
in a new form.
A.1. An application of the HJM model in the presence

The HJM Model is different from standard models since it starts with a
model for the whole forward rate curve.
A.1.1. The forward rate equation

We denote by:
F (t; T ): forward rate curve at time t;
Z(t; T ): price of a zero-coupon bond at t maturing at T and
λ: Information cost regarding the zero-coupon bond.
The price of a zero-coupon bond is given by:
RT
Z(t; T ) = e− t
(F (t;s)+λ)ds
. (A.1)
Consider the following dynamics for zero-coupon bond prices
dZ(t; T ) = µ(t, T )Z(t; T )dt + σ(t, T )Z(t; T )dX. (A.2)

Since the price of a zero-coupon bond maturing instantaneously is one, then

σ(t, t) = 0. Using (A.1), we have:
∂
F (t; T ) = − log Z(t; T ) + λ.
∂T
If we differentiate this equality and substitute from (A.2), this gives

the dynamics of the forward curve:

∂ 1 2 ∂
dF (t; T ) = σ (t; T ) − µ(t, T ) dt − σ(t; T )dX. (A.3)
∂T 2 ∂T
A.1.2. The spot rate process

The spot rate corresponds to the forward rate for a maturity equal to the
current date:
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∗
Using Eq. (A.3) gives:

t t
∗ ∂σ(s, t) ∂µ(s, t) ∂σ(s, t)
r(t) = F (t ; t)+ σ(s, t) − ds− dX(s)−λ.
t∗ ∂t ∂t t∗ ∂t
Differentiating this last expression with respect to time t gives:

t 2
∂F (t∗ ; t) ∂µ(t, s) ∂ 2 σ(s, t) ∂σ(s, t)
dr = − + σ(s, t) +
∂t ∂s s=t t∗ ∂t2 ∂t

∂ 2 µ(s, t) t 2
∂ σ(s, t) ∂σ(t, s)
− ds − dX(s) dt − dX.
∂t2 t∗ ∂t2 ∂s s=t
It is important to note that the underlined term depends on the history

of σ and the stochastic increments dX. This model makes the dynamics of
the spot rate non-Markov.
A.1.3. The market price of risk

In the HJM context, the changes in the forward rate curve are perfectly
correlated. Therefore, it is possible to hedge one bond with an other bond
maturing at a different date. An initial portfolio can be constructed using
a long position in a bond maturing in T1 and a short position in another
bond maturing in T2 as follows:
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
dΠ = Z(t; T1 )(µ(t, T1 )dt + σ(t, T1 )dX)

− ∆Z(t; T2 )(µ(t, T2 )dt + σ(t, T2 )dX).
When ∆ is chosen in a way such that:
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:
µ(t, T1 ) − (r(t) + λ) µ(t, T2 ) − (r(t) + λ)

= . (A.4)
σ(t, T1 ) σ(t, T2 )
When both sides are independent of the maturity T , we have:
µ(t, T ) = (r(t) + λ + γ(t)σ(t, T )
where γ(t) stands for the market price of risk.

Real and risk neutral

In a risk-neutral world, the return on any traded asset must be the rate
r(t) + λ rather than µ. This implies that the dynamics of all zero-coupon
bonds are given by:
dZ(t; T ) = (r(t) + λ)Z(t; T )dt + σ(t, T )Z(t; T )dX.
A.1.4. Relationship between risk-neutral forward rate

drift and volatility
It is possible to write the dynamics of the forward rate in a risk-neutral
world as:
dF (t; T ) = m(t, T )dt + ν(t, T )dX
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
A.1.5. Pricing derivatives

Since the HJM model is characterized by its non-Markov nature, it is not
possible to obtain a finite-dimensional partial differential equation for a
derivative price. The model can be simulated or implemented via a tree
structure.
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
equation in N -dimensions, then the dynamics of the forward rate can be

written as:
n

dF (t, T ) = m(t, T )dt + νi (t, T )dXi
i=1
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
A.2. An application of the Ho and Lee model

in the presence of information cost
The dynamics of the spot rate in the Ho and Lee model are given by:
dr = η(t)dt + cdX where c is a constant. The prices of zero-coupon bonds
satisfy the following equation:
∂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.
In this context, the zero-coupon bond price is given by:

T
1 2 3
Z(r, t; T ) = exp c (T − t) − η(s)(T − s)dS − (T − t)(r + λZ ) .
6 t
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
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
Hence, it is possible to show that:
dF (t; T ) = c2 (T − t)dt + cdX
Appendix B: Implementation of the BDT Model

with Different Volatility Estimators
Bali and Karagozoglu’s (1999) model both the instantaneous and time
properties of the interest rate in pricing Eurodollar futures options. When
testing the predictive power of the volatility estimation models, they show
that the forecasting ability of a moving average volatility estimators is
inferior to that of the time series volatility models. They provide evidence
on the sensitivity of derivatives pricing to volatility estimators in predicting
option prices.
B.1. The BDT model

In the BDT’s model, the long-term yields are assumed to reflect the
expectations of the market regarding the future short-term rates. In this
context, if the market expects future short rates to be high (low), then
there is a tendency for current long-term yields to be higher (lower).
The model gives in a valuation formula current expected long-term yields
using the expected future short rates. The model also offers a sensitivity
formula that predicts the expected changes in the current long-term yields
when the short rate changes. Bali and Karagozoglu (1999) determine the
value of the Eurodollar options within the BDT’s model using the yield
curve and the volatility curve, referred to as the term structure of interest
rates. Since BDT’s model assumes that changes in short-term interest rates
are log-normal, then the evolution of the short rate in a discrete time is
described by:
∆ ln r(t) = [δ0 (t) − δ1 (t) ln r(t)]θ + σ(t)∆W (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
µ(r, t) = δ0 (t) − δ1 (t) ln r(t)
is added to the log interest rate in upward or downwoard movements of the

log interest rate.
Using Eq. (B.1), ln r(t) will be one period later:
√
[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.
B.2. Estimation results

Bali and Karagozoglu (1999) uses data daily on Eurodollar spot rates,
futures and futures options from 1987 to 1996. They use a rolling regression
procedure and generate one-step-ahead forecasts by estimating the time-

varying drifts parameters. They construct time-varying forecasts of the
variance of log interest-rate changes using the above models. The time
varying parameters are estimated by a maximum likelihood estimation
technique. The valuation formulas are used to generate a binomial tree
of expected future short rates using at the same time the yield curve and
each of the four volatility curves as inputs. They compare the estimated
prices with the actual values and employ the mean square error for forecast
evaluation MSE:
n
1
MSE = (Pi − P̂i )2 (B.2)
n i=1
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?
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 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.
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.
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
Hull, J and A White (1987). The pricing of options on assets with stochastic
variables. The Journal of Finance, 42, 281–300.
Hull and White (1992).
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.
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 and G Fong (1982). Term structure modeling using exponential splines.
Willmott, P (1998). Derivatives, Chicheslec: John Wiley and Sons.

Part VI
Generalization of Option Pricing Models and

Stochastic Volatility
743
744
Chapter 17
EXTREME MARKET MOVEMENTS, RISK AND

ASSET MANAGEMENT: GENERALIZATION TO
JUMP PROCESSES, STOCHASTIC VOLATILITIES,
AND INFORMATION COSTS
Chapter Outline
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
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
Extreme Market Movements, Risk and Asset Management 747
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.
17.1. The Jump-Diffusion and the Constant Elasticity

of Variance Models
This section presents Merton’s (1976) model, which is a combination of a
jump process and a diffusion process. It also develops the results in the Cox
and Ross (1976) constant elasticity of variance diffusion model.
17.1.1. The jump-diffusion model

Merton (1976) used a combination of a jump process and a diffusion process
assuming that after each jump in the underlying asset price, a diffusion
process is used. By constraining the jumps in such a way that they are
distributed log-normally, Merton (1976) presented the following formula
for the pricing of European call options,
∞
1 −γ(1+h)T
C= e [γ(1 + h)T ]n [SN (d1 ) − Ke−r T N (d2 )]
n=1
n!
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
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.
Definition: The Poisson process

If (Nt )t≥0 is a Poisson process with an intensity γ, then for all t > 0, the
random variable Nt satisfies
1 −γt
P (Nt = n) = e (γt)n .
n!
In particular,
E(Nt ) = γt, Var(Nt ) = (Nt2 ) − (Nt )2 = γt.
Also, when
s > 0, (sNt ) = e(γt(s−1)) .
17.1.2. The constant elasticity of variance diffusion

(CEV) process
The family of CEV processes are described by the following stochastic
θ
dS = µSdt + δS 2 dW
with µ, θ, and δ > 0 and W is a Wiener process.

θ
In the above expression, δS 2 is the instantaneous variance of the stock
price where θ is the elasticity of this variance with respect to S.
In this equation, the instantaneous variance of the return σ 2 is given by:
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
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 .
17.2. On Jumps, Hedging and Information Costs

It is well known that sudden movements in asset prices appear at random
discrete times in financial markets. In general, when there is a jump in
the underlying asset price, it is very difficult to implement a hedge in the
standard sense. Financial asset prices can be modeled by a jump-diffusion
process, which corresponds to the standard diffusion plus a jump component
as follows:
dS = µSdt + σSdX + (J − 1)Sdq.
The above equation shows that the dynamics of the underlying asset price
corresponds to the standard diffusion process dS = µSdt + σSdX plus a
Poisson component. The term dq corresponds to the Poisson process, which

can be defined as follows:

0 with probability 1 − γdt
dq =
1 with probability γdt.
This means that in the presence of a jump, dq = 1, otherwise it is equal

to zero. The probability of a jump in q over the interval dt is γdt, where
γ refers to the intensity of the Poisson process. When there is a jump,
dq = 1 and S goes immediately to JS, where J refers to the jump size.
The jump size J can be a constant or a random variable. If J is random,
it is often assumed that it is drawn from a distribution with a probability
density function P (J) and that is independent of the Poisson process and
the Brownian motion. The random walk in the logarithm of the underlying
asset follows from the dynamics of S:

1 2
d(log S) = µ − σ dt + σdX + (log J)dq.
2
This represents a jump-diffusion version of Ito.
17.2.1. Hedging in the presence of jumps

As in the original Black and Scholes (1973) context, it is possible to
construct a portfolio with a long position in the derivative security and
a short position in ∆ units of the underlying asset:
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
This represents also a jump-diffusion version of Ito. In the absence of a

jump at time t, dq = 0 and the elimination of risk can be done using the
standard ∆ = ∂V
∂S . When there is a jump, dq = 1 and the portfolio changes
by an amount O(1) that cannot be diversified away. In order to implement
the hedging argument for the diffusion, consider the change in the portfolio
over a small time interval (with no stochastic term dS):

∂V 1 ∂ 2V ∂V
dΠ = + σ2 S 2 2 dt + V (JS, t) − V (S, t) − (J − 1)S dq.
∂t 2 ∂S ∂S
The above equation shows that the value of the portfolio in a deterministic
context comprises a “deterministic component” and a “non-deterministic
jump” component. Merton (1976) showed that in the absence of a
correlation between the jump component and the “market portfolio”, the
diversifiable risk should not be paid for. Hence, in the absence of market
price for risks (jumps are not priced), the expected return on the above
portfolio must be the risk-less rate plus the information costs. Following
this approach, this gives the following equation:
∂V 1 ∂2V ∂V
+ σ2 S 2 + (r + λS )S − (r + λV )V
∂t 2 ∂S 2 ∂S
∂V
+ γE[V (JS, t) − V (S, t)] − γ SE[(J − 1)] = 0
∂S
where λS and λV correspond to the information costs on the underlying
asset market and the option market, respectively. The expectation in this
equation is taken with respect to the jump size J and can be written as
follows:

E[x] = xP (J)dJ
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
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.
17.2.2. Hedging the jumps

It is possible to “hedge” the portfolio in the presence of jumps by choosing
“the hedge ratio” that minimizes the variance of the hedged portfolio as
shown in Willmott (1998). In this context, the change in the portfolio value
with any ∆ can be written as:

∂V
dΠ = − ∆ dS + (−∆(J − 1)S + V (JS, t) − V (S, t))dq + · · ·
∂S
The risk in this change of the portfolio value can be computed as:
2
∂V
Var[dΠ] = − ∆ σ 2 S 2 dt
∂S
+ γE[(−∆(J − 1)S + V (JS, t) − V (S, t))2 ]dt + · · ·
Differentiating with respect to ∆ and setting the resulting equation equal

to zero, gives the ∆ that minimizes the variance:
γE[(J − 1)(V (JS, t) − V (S, t))] + σ 2 S ∂V
∂S
∆=
γSE[(J − 1)2 ] + σ 2 S
Under this minimizing strategy, the discounted expectation of the option
must satisfy the following equation:

∂V 1 ∂ 2V ∂V σ2
+ σ2 S 2 2 + S µ− (µ + γk − (r + λS )) − (r + λV )V
∂t 2 ∂S ∂S d

J −1
+ γE (V (JS, t) − V (S, t)) 1 − (µ + γk − r + λS ) = 0,
d
where
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
no diffusion), the ∆ is given by
E[(J − 1)(V (JS, t) − V (S, t))]

∆=
SE[(J − 1)2 ]
and the equation becomes:

∂V ∂V
+ µS − (r + λV )V
∂t ∂S

J −1
+ γE (V (JS, t) − V (S, t)) 1 − (µ + γk − (r + λS )) = 0.
d
17.2.3. Jump volatility

Volatility can jump from one state to another from time to time. Consider
the volatility in one of the two states σ− or σ + with σ− < σ + . The volatility
can jump from lower to higher value in the presence of a Poisson process
with intensity γ + or γ − in one direction or the other. As given in Willmott
(1998), if a hedge portfolio is constructed and real expectations are taken,
then the return on the portfolio must be equal to the risk-less rate plus
information costs. This leads to the following equation for the value V of
the option V + 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
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
Jump volatility with exponential decay

After a jump, the volatility shows an exponential decay and can be
presented as follows:
σ(τ ) = σ − + (σ + − σ − )e−ντ
with
τ : time since the last jump in volatility and
ν: a decay parameter.
The volatility is described in the presence of a Poisson process with intensity

γ and can jump from its present level to σ + . In this context, the option
price V (S, t, τ ) is a solution to the following equation:
∂V ∂V 1 ∂2V ∂V
+ + σ(τ )2 S 2 + (r + λS )S − (r + λV )V
∂t ∂τ 2 ∂S 2 ∂S
+ γ(V (S, t, 0) − V (S, t, τ )) = 0.
17.3. On the Smile Effect and Market Imperfections

in the Presence of Jumps and Incomplete
Information
This section develops a simple option pricing model when markets can
make sudden 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.
17.3.1. On smiles and jumps

Using the market prices of at least two options on the same 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 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
Table 17.1. Implied volatilities and the deltas of S&P

calls and puts using the Black and Scholes (1973) model
when the index was at 377.25, r = 0.059, and the
dividend yield is 2%. The index forward price is 378.33.
Strike Type σImplied (in %) ∆ (in %)
345 put 21.5 −5.7

350 put 19.3 −6.8
355 put 18.4 −10.1
360 put 17.3 −14.5
365 put 16.2 −20.5
370 put 15.3 −29.2
375 put 13.3 −39.8
380 call 13.6 46.1
385 call 12.3 31.0
390 call 12.5 19.5
395 call 12.5 11.0
400 call 13.1 6.4
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.
Valuing options when markets can jump

Consider a simple model. The underlying asset price at time zero today
is S. In the next instant, the underlying asset price can jump up by u%
to Su with probability w or down by d% to Sd with probability (1 − w).
The probability w is expected to be close to zero or one. This means that
either a jump up or a jump down predominates. After the first jump, the
underlying asset will diffuse with volatility σ as in the Black and Scholes
(1973) model. No other jumps will occur. The value of any security in this
model can be computed as the average of its payoffs over the scenarios
where the undelying asset jumps up or down. Hence, the option value is
given by:
option = wBS(Su , K, σ, r, δ, t) + (1 − w)BS(Sd , K, σ, r, δ, t) (17.1)
where BS(S, K, σ, r, δ, t) is the formula in Black and Scholes (1973) and δ

refers to the continuous dividend yield. The values used for the underlying
asset are:
Su = S(1 + u), Sd = S(1 − d).
The current value of the underlying asset also corresponds to an average

value after a jump up and a jump down. Hence, the jump up and the jump
down are related by:
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.
Table 17.2. Estimation of the implied

probability of an upward jump and a diff-
usion volatility. Market call prices and
implied volatilities in the presence of an
upward jump for the following parameters:
S = 100, r = 0.1, δ = 0, and T = 1.
Strike Call σBlack−Scholes
110 4.83 12.1%

120 2.41 14.09%
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 17.3. Model values in the presence of jumps.
Strike Index–jump BS–call Probability Model–contribution
110 271.8 171.8 0.01 1.72

110 98.26 3.14 0.99 3.11
Total 4.83
120 271.8 162.71 0.01 1.63
120 98.26 0.79 0.99 0.78
Total 2.41
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
Black–Scholes model. The model seems to be insensitive to the exact value

of the jump probability w.
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.
Table 17.5. Estimation of the implied jump magnitude and diffusion

volatilities that match the S&P puts struck at 365 and 370.
Jump–probability, w Implied–jump–down, d Implied–volatility

(in %) (in %) (in %)
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
Table 17.6. Black–Scholes implied-volatilities for 5% jump probability.
Model σ Market σ
Strike (in %) (in %) Option Model–price Market–price
345 20.6 21.5 put 0.46 0.56

350 19.7 19.3 put 0.68 0.63
355 18.5 18.4 put 0.97 0.94
360 17.4 17.3 put 1.37 1.38
365 16.2 16.2 put 2.00 2.00
370 15.3 15.3 put 3.00 3.00
375 14.6 13.3 put 4.54 4.00
380 14.1 13.6 call 5.10 4.88
385 13.8 12.3 call 3.06 2.50
390 13.5 12.5 call 1.68 1.38
395 13.3 12.2 call 0.85 0.69
400 13.2 13.1 call 0.39 0.38
17.3.2. On smiles, jumps, and incomplete information

We develop a simple option pricing model when markets can make sudden
jumps in the presence of incomplete information. We build on Derman et al.
(1991) modeling of jumps on the underlying asset and combine it with
Bellalah (1999) approach to include information costs. The same approach
can be extended to allow the estimation of implied information costs from
market data. The option value is given by:
option = wBS(Su , K, σ, r, b, λs , λc , t) + (1 − w)BS(Sd , K, σ, r, b, λs , λc , t)
where BS(S, K, σ, r, b, λs , λc , t) is the formula given in Bellalah (1999).

The values used for the underlying asset are:
Su = S(1 + u), Sd = S(1 − d).
The jump up and the jump down are related by:
d(1 − w) = wu.
The price of a European call with a strike K is:
C(S, T ) = Se−(λC −λS )T N (d1 ) − Ke−(r+λC )T N (d2 )
with

S 1 2 √
d1 = ln + r + σ + λS T σ T (17.2)
K 2
and
√
d2 = d1 − σ T (17.3)
where N (.) is the univariate cumulative normal density function.

When λS and λC are set equal to zero, this equation collapses to that
in shown Barone–Adesi and Whaley (1987). In this case, the value of a
European commodity call is:
C(S, T ) = Se((b−r−(λC −λS ))T ) N (d1 ) − Ke−(r+λC )T N (d2 )
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:
P (S, T ) = −Se((b−r−(λC −λS ))T ) N (−d1 ) + Ke−(r+λC )T N (−d2 ) (17.4)
with

S 1 √ √
d1 = ln + b + σ 2 + λS T σ T, d2 = d1 − σ T .
K 2
17.3.3. Empirical results in the presence of jumps

and incomplete information
The smile and the jumps
Consider the implied volatilities on a given day for the European-style July
S&P index options expiring with a given maturity. Table 17.7 shows the
implied volatilities and the deltas of S&P calls and puts using the Black–
Scholes (1973) model. The option maturity date is in March 2001, the index
level is 1264.74, the risk-less interest rate is 5.81%, and the dividend yield
is 1.17%. The calibration has been made using the 1200 and the 1250 put
options as they correspond to the most liquid options given the maturity
we considered. The results are given in Tables 17.8 and 17.9. In Table 17.8,
the results are based on direct application of the Derman et al. (1991)
Table 17.7. Implied volatilities and the deltas of S&P

calls and puts using the Black–Scholes (1973) model.
Strike Type σimplied (in %) ∆ (in %)
950 put 35.49 −0.37

975 put 34.84 −0.72
1025 put 34.72 −2.26
1050 put 32.51 −3.68
1100 put 31.84 −8.44
1125 put 30.38 −11.97
1150 put 29.80 −16.31
1175 put 29.07 −21.45
1200 put 28.21 −27.29
1250 put 25.71 −40.45
1275 call 24.28 52.42
1300 call 24.57 45.56
1325 call 23.65 38.96
1350 call 22.47 32.77
1375 call 22.02 27.13
1400 call 21.20 22.10
Table 17.8. Parameter estimates using the Derman

et al. (1991) methodology.
w (in %) d (in %) σdiffusion (in %)
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
Table 17.9. Parameter estimates using the Derman

et al. (1991) methodology with endogenous w parameter.
w d σdiffusion
15.51% 19.67% 16.43%

Table 17.10. Comparison between Black and Scholes and model prices.
Market Black–Scholes Model Market σ Model σ

Strike Type price price price (in %) (in %)
950 put 1.875 0.10 0.813 35.49 31.20

975 put 2.625 0.21 1.587 34.84 31.90
1025 put 5.750 0.87 4.512 34.72 32.84
1050 put 6.375 1.59 6.746 32.51 32.98
1100 put 12.000 4.60 12.630 31.84 32.38
1125 put 14.500 7.28 16.194 30.38 31.64
1150 put 18.875 11.05 20.186 29.80 30.66
1175 put 24.000 16.13 24.719 29.07 29.49
1200 put 30.000 22.73 30.012 28.21 28.22
1250 put 44.125 41.07 44.140 25.71 25.71
1275 call 54.000 54.00 54.704 24.28 24.60
1300 call 43.625 42.98 41.540 24.57 23.62
1325 call 32.375 33.70 30.549 23.65 22.78
1350 call 22.500 26.03 21.727 22.47 22.06
1375 call 15.875 19.80 14.930 22.02 21.46
1400 call 10.250 14.84 9.907 21.20 20.96
methodology, i.e., the w parameter value has been explicitly chosen. We

give the results for a set of reasonable values, starting from w = 3%, as the
algorithm was unable to achieve convergence for values less than this figure.
In Table 17.9, the w value is endogenously determined, i.e., we let the algo-
rithm to calculate the parameter values (w, δ, and σ diffusion), which best
fit the market prices used for calibration. In the remainder, we decided to
restrict ourselves to this approach. Table 17.10 gives a comparison between
the market price, the Black and Scholes price, and the model price, and
between the model-implied diffusion volatility and the Black and Scholes
implicit volatility. The input value of sigma for the Black and Scholes
formula has been estimated using the 1250 at-the-money (ATM) put.
Introducing information costs

We introduce information costs in the Derman et al. (1991) methodology.
We considered information costs both on the option market (λC ) and the
underlying asset (λS ) and run simulations for different cost levels (from 1%
to 5%). However, due to space considerations, we restrict our presentation in
Fig. 17.1 to the most significant results. Information cost levels which give the
best fitting (λS = 1%) and λC = 2%) are very close to Merton’s estimates
although we use a radically different approach. Thus, we can view the model as
a possible (and reliable) way to extract information costs using option prices.
Recent evidence in 2008 and 2009 seems to confirm the results.
17.4. Implied Volatility and Option Pricing Models:

The Model and Simulation Results
The “liquidity premium” can be defined as an amount f (S0 , K), which
is added to (subtracted from) the classical option payoffs according to the
position of the initial underlying asset price with respect to the strike price.
The liquidity functions may be written as gc (S0 , K) = K + fc (S0 , K) for
a call and gp (S0 , K) = K + fp (S0 , K) for a put option, where fc (S0 , K)
and fp (S0 , K) stand for the amount of money subtracted or added to the
option price to account for the mispricing, S0 is the observed underlying
asset price, and K is the strike price.
17.4.1. The valuation model

Assumption
The function fc (S0 , K) must satisfy the following assumptions:
• if S0 > K, then −K < fc (S0 , K) ≤ 0 and fc (S0 , K) is nondecreasing
in K;
• if S0 < K, then fc (S0 , K) > 0 and fc (S0 , K) is nondecreasing in K and
• fc (S0 , K) is differentiable on R+∗2 and fc (K, K) = 0.
The function fp (S0 , K) must satisfy the following assumptions:
• if S0 > K, then fp (S0 , K) > 0 and fp (S0 , K) is nonincreasing in K;
• if S0 < K then −K < fp (S0 , K) ≤ 0 and fp (S0 , K) is nonincreasing
in K;

∂fp (S0 ,K)
• ∂K < 1 (which implies that gp (S0 , K) is increasing in K) and
• fp (S0 , K) is differentiable on +∗2
and fp (K, K) = 0.
The functions gc (S0 , K) and gp (S0 , K) satisfy the main properties as in
the standard Black and Scholes case: they are positive, both nondecreasing
in K and gc (K, K) = gp (K, K) = K.
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
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.
Then, we have gc (S0 , K) > 0 and is nondecreasing in K.

The parameter α0 affects the prices of in-the-money (ITM) calls and the
parameter β0 affects OTM call prices. The parameters α0 and β0 represent
the coefficients that give the desired shape of the smile. When these
parameters are zero, this situation corresponds to the Black–Scholes case.
For the put option, the following function fp (S0 , K) is defined as:
2
If S0 > K, then fp (S0 , K) = γ0 1 − SK0 with γ0 > 0 and sufficiently small
∂f (S ,K)
in order to have p ∂K0 < 1 and
2
If S0 < K, then fp (S0 , K) = −δ0 1 − SK0 with δ0 > 0 and small enough
∂f (S ,K)
to have p ∂K0 < 1 and (1− bS0
1 2 > δ0 (with b < 3).
)
b
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:
cBS (K) = e−rT EQ [(ST − K)+ ]
where the mathematical expectation EQ is taken with respect to the risk-

neutral probability Q. The call option price accounting for the liquidity
premium is given by:
cLP (K) = e−rT EQ [(ST − gc (S0 , K))+ ]
which is also:
cLP (K) = S0 N (d1 ) − gc (S0 , K)e−rT N (d2 )
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
When the liquidity premium is accounted for in the valuation of

European put options, the put formula is given by:
pLP (K) = e−rT EQ [(ST − gp (S0 , K))+ ]
which is also:
pLP (K) = −S0 N (−d1 ) + gp (S0 , K)e−rT N (−d2 ).
The model corrects for both biases with respect to the Black–Scholes model.

Using Black and Scholes model and our model with “liquidity premiums”,
Tables 17.11 and 17.12 develop simulation results for European call and put
options. Table 17.11 presents call prices for different levels of the underlying
asset price varying from 95 to 105. For illustrative purposes, the table
compares the Black and Scholes call price, cBS with the vertical premium
call price, cLP . The spread between the two models is reported as Dif
to reflect the amount of overvaluation or undervaluation in the Black and
Scholes model.
Tables 17.12 shows similar information for the values of European
put option prices for the same set of parameters. The option’s time to
maturity is one year from the 20th March 1999 to the 19th March 2000.
The strike price is 100, the risk-less interest rate is 0.1, and the volatility
parameter is 0.2. The vertical premium function is defined for the call using
Table 17.11. Simulations of European call prices for

BS and our model using: T = 1, r = 0.1, σ = 0.2,
K = 100, and α0 = β0 = 0.75.
S CBS CLP Dif, 10−3
95 9.86275014 9.86180506 −0.94508

96 10.5069582 10.506331 −0.62721
97 11.1703759 11.1700108 −0.36513
98 11.8523936 11.8522259 −0.16763
99 12.5523797 12.5523365 −0.00000
100 13.2696859 13.2696859 0
101 14.0036514 14.0036962 0
102 14.7536068 14.7537871 0.18032
103 15.5188787 15.5192863 0.40765
104 16.298793 16.2995203 0.72727
105 17.0926787 17.0938178 1.13906
Table 17.12. Simulations of European put prices for

BS and our model using: T = 1, r = 0.1, σ = 0.2,
K = 100, and γ0 = δ0 = 0.75.
S PBS PLP Dif, 10−3
95 5.34649194 5.345741 0.751

96 4.99070003 4.990241 0.459
97 4.65411775 4.653872 0.246
98 4.33613563 4.336032 0.104
99 4.03612149 4.036097 0.025
100 3.75342774 3.753428 0
101 3.48739318 3.487415 −0.022
102 3.23734858 3.237429 −0.081
103 3.00262046 3.002789 −0.168
104 2.7825348 2.782811 −0.0277
105 2.57642053 2.576820 −0.400
α0 = β0 = 0.75. For the put option, we use γ0 = δ0 = 0.75. The above

parameters are used for illustrative purposes, but other parameters can
also be used. These parameters can be adjusted to obtain any desired form
of a smile. In fact, if we modify these parameters, we change the shape of
the smile curve. The simulation results show also that the model prices are
very close to Black and Scholes prices for call and put options because of
the chosen parameters. The sign of spread between the two models depends
on the way it is calculated. It can be either positive or negative according
to the degree of parity and to the way the difference is calculated between
the two models. The difference reflects the amount of overvaluation or
undervaluation reported in empirical tests of the Black and Scholes type
models.
17.4.3. Model calibration and the smile effect

The Black and Scholes model and market prices are used to estimate
the implied volatility. The implicit volatility for short-term and long-term
options is calculated using an iterative procedure. The algorithm given by
Bellalah and Jacquillat (1995) is used for this purpose. Each day, implied
volatilities are aggregated with respect to the degree of the options parity.
Hence, we obtain each day, 11 average implied volatilities corresponding
to different degrees of parity. To obtain an idea about the index volatility
estimates, we construct an implied ratio of volatility. This ratio is defined
as follows:
σK,t
RvK,t = for K = −5, . . . , 5.
σ0,t
By construction, this ratio is equal to one for at the money (ATM)

options. For each day, the mean volatility is calculated. Then, the volatilities
are aggregated. The ratio of volatilities is constructed by dividing the
implied volatility for a trade by the volatility of ATM options. This smile
is an evidence against the Black and Scholes model. This smile is different
from that reported in Dumas et al. (1998). This result is an evidence that
the smile effect is different from one market to another for index options.
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
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
1895–1927.

of Finance, 13(3), 1975–2000.
Finance, 13(2), 1826–1837.
Derman, E, A Bergier and I Kani (1991). Valuing Index Options When Markets
Can Jump. Quantitative Research Notes (July). New York: Goldman Sachs.
Dumas, B, J Fleming and R Whaley (1998). Implied volatility functions: empirical
tests. Journal of Finance, 53, 2059–2106.
Knoch, H-J (1992). The pricing of foreign currency options with stochastic
volatilities. PhD. dissertation, Yale School of Organization and Management.
Melino, A and SM Turnbull (1990). Pricing foreign currency options with
stochastic volatility. Journal of Econometrics, 45, 239–265.
Rubinstein, M (1994). Implied binomial trees. Journal of Finance, 3, 771–818.
Willmott P (1998). Derivatives. New York: John Wiley and Sons.

Chapter 18
RISK MANAGEMENT DURING ABNORMAL

MARKET CONDITIONS: FURTHER
GENERALIZATION TO JUMP PROCESSES,
STOCHASTIC VOLATILITIES,
AND INFORMATION COSTS
Chapter Outline
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
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
Risk Management During Abnormal Market Conditions 773
test of the valuation model of options in the presence of information

costs developed by Bellalah (1999). They found that the inclusion of the
information costs in the option formula, increases the precision of the
option price. Due to these costs and stochastic volatility, they proposed
an explanation of the skewness observed in the smile even if there is no
skewness in the real distribution. We know how the Black–Scholes (1973)
option pricing model has generated non-negligible specification errors with
respect to the option-market observed data. Model prices as a function
of the strike price are systematically diverging from the observed option
prices. This feature in term of implied volatility is often called the “smile
effect”, where the so-called smile refers to the U-shaped pattern of implied
volatilities across different strike prices. Nevertheless, the smile can be
asymmetric. This skewness effect can often be described as the addition
of monotonic curve to the standard symmetric smile. The smile becomes
sometimes a smirk on a skew, since it appears more or less lopsided:
the so-called skewness effect. In general, the implied volatility curve has
its minimum for out-of-the-money (OTM). Jackwerth and Rubinstein
(1996) have linked the smile shape and the risk-neutral density as one.
They concluded that observed smiles translate into a left-skewed highly
leptokurtic risk-neutral distribution for the future underlying asset price.
Moreover, At-Sahalia and Lo (1998) have pointed out that the presence in
option prices of an implied volatility smile, whereby OTM put options are
more expensive than at-the-money (ATM) options, directly translate into a
negatively skewed option implied risk-neutral density. Indeed, an answer to
the question, “How to explain the smile (its shape)?” could be translated to,
“how to explain the shape of the risk-neutral density?” It is straightforward
to assume that both distributions (after the moments of the second orders)
are the same under the risk-neutral and the objective distributions. Renault
(1997) pointed out that the asymmetry of the implied volatility curves is
best characterized with reference to a benchmark model, which produces a
symmetric curve. When the volatility is stochastic as in the Hull and White
(1987) model, 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. Based on
this model, we show that information costs can produce an asymmetric
smile (skew pattern) even if the objective distribution has no skewness. 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) and Bellalah and Selmi (2001) and so on.
18.1. Option Pricing in the Presence of a Stochastic

Volatility
Hull and White (1988) considered the following model for the dynamics of
the stock price and its volatility, which is defined as a separate stochastic
variable:
dSt = µ(St , σt , t)St dt + σt St dWt1 (18.1)
dνt = (σt , t)νt dt + δt (σt , t)νt dWt2 (18.2)
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
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)
It is important to note that the moments of F cannot be determined

∗
in general, except in special cases when γ and δ are constants, and νt,T
is an integral over log-normal variables. In this very particular situation,
using a Taylor’s expansion of Eq. (18.3), Hull and White (1988) provided
the following solution:
c(t, St , σt2 ) = SN (d1 ) − Ke−rtN (d2 )

1 h 1 1
+ S (T − t)N (d1 )(d1 d2 − 1) 2σ e − h − 2 −
4
2 h 2
1

+ SN (d1 ) (d1 d2 − 3)(d1 d2 − 1) − (d21 + d22 )
6
1 σ6

+ σ6 3 e3h − (9 + 18h)eh + (8 + 24h + 18h2 + 6h3 )
8 3h
with h = 2 t.
18.1.2. Stein and Stein model

Stein and Stein (1991) studied stock price distributions when prices follow
diffusion processes with a varying volatility parameter. They obtained
interesting results regarding option pricing with stochastic volatilities and
the relationship between this parameter and the nature of fat tails in stock
price distributions. Their model is more general than that of Hull and
White for the following two reasons. First, Hull and White used Taylor’s
series expansion to solve explicitly for the option price, about the point
where the volatility is non stochastic, i.e., δ = 0. Second, it is not clear
that the expansion provides a good approximation to option prices when
the value of δ is far from zero. Using Stein and Stein’s notations, let the
dynamics of the stock price P , to be represented by the familiar following
equation:
dSt = µP dt + σP dW1 (18.4)
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:
dσ = −δ(σ − θ)dt + κdW2 . (18.5)
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:
S(P, t) = e−µt S0 (P e−µt , t)
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.
18.1.3. The Heston model

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. The
dynamics of the spot asset at time t are described by the following diffusion
equation:

dSt = µSdt + ν(τ )SdWt1 .
The volatility dynamics are governed by the Ornstein–Uhlenbeck process

used by Stein and Stein (1991).

d ν(τ ) = −β ν(τ )dt + δdWt2 . (18.6)
Using Ito’s lemma and standard arbitrage arguments, Heston (1993)

showed that the price of a European call is given by:
c(S, ν, t) = Sp1 − KP (t, T )p2 (18.7)
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 .
This equation is satisfied by discount bond prices in the model of Cox

et al. (1985a, b). The dynamics for the bond price P (t, T ), at time t, for a
maturity date T , are specified by the following equation:

dP (t, T ) = µP P (t, T )dt + σP (t) ν(t)P (t, T )dWt2 .
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.
18.1.4. The Hoffman, Platen, and Schweizer model

Hoffman et al. (HPS) (1992) provided an approach to option pricing which
allows the specification of general patterns of volatility behavior. The
approach combines the use of a high dimensional Markovian model with
stochastic numerical methods in an incomplete market. The assumption
of an incomplete market implies that there is no unique equivalent
martingale measure or risk-neutral probability for the underlying stock
price dynamics. When the volatility is both stochastic and past dependent,
the following multi-dimensional process was used by Hoffman et al.
(1992):
n

dXti = ai (t, Xt )dt + bij (t, Xt )dWtj for i = 0 to m and j = 1 to n.
i=1
In this formulation, the risk-less asset is represented by X 0 and the

dynamics of the underlying asset (the stock) are given by X 1 . The other
components of X can be used to model other assets. For example, they
could model the stochastic volatility and its dependence on the past.
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.
The valuation model

The pricing of derivative securities in the presence of a random volatility
needs the use of two processes: one for the underlying asset and one for the
volatility. Consider the following dynamics for the underlying asset:
dS = µSdt + σSdW1
and the following process for the volatility:
dσ = p(S, σ, t)dt + q(S, σ, t)dW2 .
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
with value V1 (S, σ, t):
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.
Isolating the terms in V and V1 gives:
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
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.
18.1.5. Market price of volatility risk

Suppose the investor holds only the option V , which is hedged only by the
underlying asset S in the following portfolio Π = V − ∆S. Over a short
interval of time dt, the change in the value of this portfolio can be written as:

∂V 1 ∂ 2V ∂ 2V 1 ∂2V
dΠ = + σ2 S 2 2 + ρσqS + q 2 2 dt
∂t 2 ∂S ∂S∂σ 2 ∂σ

∂V ∂V
+ − ∆ dS + dσ.
∂S ∂σ
In the standard delta hedging, the coefficient of dS is zero and we have:
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.
18.1.6. The market price of risk for traded assets

In the B–S analysis, the hedging portfolio is constructed using the option
and its underlying tradable asset. Consider the construction of a portfolio
as before using two options V and V1 with different characteristics. In this
case, if the underlying asset is not a tradable security, the initial portfolio
value would be Π = V − ∆1 V1 . Using the same methodology, as before,
gives the following equation:
∂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.
18.2. Generalization of Some Models with Stochastic

Volatility and Information Costs
18.2.1. Generalization of the Hull and White (1987) model
Consider the following model:
dBt = (r + λBt )Bt dt (18.11)

dSt = µ(St , σt , t)St dt + σt St dWt1
dνt = γ(σt , t)νt dt + δ(σt , t)νt dWt2
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
such that:
dBt = (r + λBt )Bt dt

dSt = (r + λSt )St dt + σt St dW̃t1
dνt = γ(σt , t)νt dt + δ(σt , t)νt dW̃t2
under P̃ , where W̃ 1 , W̃ 2 are independent Brownian motions under P̃ . The
risk-neutral dynamics of the bond and the underlying asset are used in
Bellalah (1999). The portfolio value would be Π = V − ∆S − ∆ Bt with ∆
units of the bond. When we apply the methodology of the previous section
to this model, Eq. (18.8) gives:
∂V 1 ∂ 2V ∂ 2V 1 ∂2V ∂V
+ νt St2 2 + ρσt3 St ξ + ξ 2 νt2 2 + (r + λSt )St
∂t 2 ∂St ∂S∂νt 2 ∂νt ∂St
∂V ∂V
+ (γ − δξ)νt + (r + λBt )Bt − (r + λV )V = 0
∂νt ∂Bt
with W̃ 1 , W̃ 2 independent Brownian motions under the probability
P̃ (ρ = 0) and λSt is the information cost of the security St . The investor
paid the shadow cost λSt , if he/she does not know the asset. Also λBt is
the information cost of the bond Bt . We suppose that δ = 0, the option
price is then given by:

Bt

V (t, St ) = Ẽ (ST − K) | Ft = e−(r+λBt )(T −t) Ẽ (ST − K)+ | Ft
+
BT
To obtain a more specific form for V , we use the additional assumption of
the independence of W 1 , W 2 (the instantaneous variance ν is not influenced
by the stock price S). Setting:
T
1
ν t,T = νs ds.
T −t t
The conditional distribution of SSTt under P̃ , given ν t,T , is log-normal with
parameters (r + λBt )(T − t) and ν T −t . This allows to write V as:
∞
V (t, St , σt ) =
2
uBS (t, St , ν t,T )dF (ν t,T | St , σt2 )
0
where VBS denotes the usual Black–Scholes (1973) price corresponding to

the variance ν t,T and F is the conditional distribution under P̃ of ν t,T
given St and σt2 . This is equivalent to writing:
∞
V (t, St , σt ) =
2
VBS (t, St , ν t,T )h(ν t,T | St , σt2 )dν t,T
0
with
VBS (ν) = St N (d1 ) − Xe−(r+λSt )(T −t) N (d2 )

log(St /K) + (r + λSt + ν/2)(T − t)
d1 = , d2 = d1 − ν(T − t).
ν(T − t)
When µ = 0 and as in Hull and White (1987) we have:

√
1 S T − tN (d1 )(d1 d2 − 1)
V (S, σt2 ) = VBS (ν) +
2 4σ3
4 k
2σ (e − k − 1)
× −σ 4
k2
√
1 S T − tN (d1 )[(d1 d2 − 1)(d1 d2 − 3) − (d21 + d22 )]
+
6 8σ5
3k
e − (9 + 18k)ek + (8 + 24k + 18k 2 + 6k 3 )
× σ6 + ··· ,
3k3
1 x2
with k = ξ 2 (T − t), N (x) = √ e− 2 .
2π
18.2.2. Generalization of Wiggins’s model

Under the assumption of the continuous trading, without frictions, in a
complete market, Wiggins (1987) used the following dynamics for the asset
and the volatility:
dSt = µ(St , σt , t)St dt + σt St dWSt

dσt = f (σt )dt + θσt dWσt
with dWSt , dWσt are processes of Wiener, the correlation coefficient

between stock returns and volatility movements is ρdt = dWSt dWσt and
(dP/P )(dSt /St ) = 0. The instantaneous rate of return on the hedge
portfolio P is:
dP/P = wdV /V + (1 − w)dSt /St
with w the fraction invested in the contingent claim V and (1 − w) the

fraction invested in the stock S. Equation (18.8) is equivalent, in this
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:
∂V 1 ∂2V ∂2V 1 ∂2V ∂V

+ σt2 St2 + ρσt2 θSt + θ2 σt2 + (r + λSt )St
∂t 2 ∂St2 ∂S∂σt 2 ∂σt2 ∂St
∂V
+ [f (σt ) − (µ − r − λSt )ρθ + Φ(.)θσt (1 − ρ2 )] − (r + λV )V = 0.
∂σt
We conclude that the market price of risk affects the term given by Wiggins
(1987), Φ(.) = (µP −r−λP )/σP . This term is the expected excess return per
unit risk, or the market price of risk, for the hedge portfolio. It represents
the return-to-risk tradeoff required by investors for bearing the volatility
risk of the stock.
δσt − (µ − r − λSt )ρ
Φ(.) = .
σt (1 − ρ2 )
The market price of risk depends on the information cost of the stock and
the stochastic volatility.
18.2.3. Generalization of Stein and Stein’s model

In this model, the stock price dynamics are given by the following process:
dSt = µ(St , σt , t)St dt + σt St dW1
The volatility follows an Ornstein–Uhlenbeck process:
dσt = (σt − θ)dt + kdW2 .
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
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 .
18.2.4. Generalization of Heston’s model

The underlying asset and the volatility follow the diffusion process:
√
dSt = µ(St , σt , t)St dt + νt St dWt1
√
dνt = κ(θ − νt )dt + σt νt dWt2
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
Under the same assumption as in Heston (1993), it is possible to obtain

solution to Eq. (18.9). This solution depends on information costs λSt . In
fact, a European call with a strike price K and maturing at time T , satisfies
the Eq. (18.9) subject to the following boundary conditions:
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.
By analogy with the Black–Scholes (1973) formula, Heston (1993) gave a

solution of the form:
V (S, νt , t) = SP 1 − KP (t, T )P2 (18.13)
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:
fj (x, νt , t; φ) = exp[C(T − t; φ) + D(T − t; φ)νt + ixφ]
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 fj (x, νt , T ; φ) = eiφx .

18.2.5. Generalization of Johnson and Shanno’s model

We consider the following model:
dSt = µSt St dt + σt Stα dWt1
dσt = µσt σt dt + σt σσt dWt2
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
18.3. The Volatility Smiles: Some Standard Results

18.3.1. The smile effect in stock options and index options
Stein and Stein (1991) presented a model for stock price distributions
when the volatility is driven by an Ornstein–Uhlenbeck process (AR1).
They applied the results obtained from the model to option pricing with
stochastic volatilities and to the analysis of the relationships between
stochastic volatilities and the nature of fat tails in stock price distributions.
In order to simulate their model, they estimated the necessary parameters
from data used in Stein (1989) and Merville and Pieptea (1989) concerning
individual stocks and the S&P 100 index. When comparing option prices,
they calculated option values using their model and the B–S model based
on the implied volatility associated with their model. Their results showed
that a stochastic volatility exerts an upward influence on all option prices
and that stochastic volatility is more important for away-from-the-money
options, that is, implied volatilities corresponding to new option prices,
exhibit a U-shape, as the strike price is changed. The implied volatility is
the lowest for ATM options and rises as the strike price moves in either
direction. For away-from-the-money options, the “mixing distribution”,
they used the U-shape in implied volatilities. They found that the overall
impact on option prices is economically significant, especially for OTM
options. For some parametric values, their model prices were 11.7% more
than B–S model’s, and even larger effects are observed with cheaper options.
18.3.2. The smile effect for bond and currency options

In order to explain strike biases in the B–S model, Heston (1993) proposed
an option pricing model allowing for arbitrary correlation between volatility
and spot asset returns when volatility follows an Ornstein-Uhlenbeck
process as in Stein and Stein (1991). His model can be applied to bond
options and currency options. When examining the effect of stochastic
volatilities on option prices in comparison to the B–S model, he used a
B–S model with a volatility parameter that matches the variance of the
spot return over the option life. The parameters used correspond roughly
to those estimated by Knoch (1992) for Yen and Deutschemark currency
options. The model links the biases of the B–S model to the dynamics
of the spot price and its distribution. Heston showed that B–S prices
are virtually identical to his model prices for ATM options, explaining
some of the empirical support to the B–S model. Also correlation between
volatility and the spot price is necessary in explaining skewness that
affects the pricing of ITM options relative to OTM options. In fact, a
positive correlation coefficient results in high variance, which spreads the
right tail of the probability density and results in a thin left tail in the
distribution.
This increases OTM call prices and decreases ITM call prices relative
to the B–S model with a comparable volatility. A negative correlation
coefficient has the opposite effects, i.e., it decreases OTM call prices
relative to ITM call prices with respect to B–S model. Hence, without
this correlation, stochastic volatility does not change skewness and affects
only the kurtosis. This latter affects prices of near-the-money calls rel-
ative to far-from-the-money calls. The model seems to impart not only
the strike price bias but also other biases reported when testing B–S
model.
Heston (1993) presented an option pricing model consistent with
empirical biases reported in the B–S formula. The model, based on a
log-gamma process, explains in particular, the smile effect with respect
to the distribution’s skewness. Model simulations show that option prices
are similar to those of B–S for ATM options. However, Heston’s model
assigns higher prices to OTM options and lower prices to ITM options
when compared to B–S prices. The differences in prices are economi-
cally significant. The strike biases are similar to the skewness-related
biases in models where stochastic volatility is correlated with stock
returns.
18.3.3. Volatility smiles: Empirical evidence

The expected volatility of an underlying asset can be inferred from the
option price. Implied volatilities can be taken for different time to maturities
and different strike prices. Hence, using option prices, a matrix of implied
volatilities with rows ordered by the strike price and columns ordered by
time to maturities can be constructed. The rows of the implied volatility
matrix may provide information about the term structure of expected
future volatility. Poterba and Summers (1986), Stein (1989), Franks and
Schwartz (1991), and Heynan et al. (1994) studied the term structure
of implied volatilities using only two times to maturity. The time series
studies done by Merville and Pieptea (1989) and Day and Lewis (1992) used
one implied volatility per day and ignored the term structure effects since
time to maturity varies from day-to-day. Stein (1989) used two daily time
series on implied volatilities for the S&P 100 index options over the period
1983–1987. Based on the assumption that the volatility is mean reverting,
he concluded that long-maturity options tend to overreact to changes in
the implied volatility of short maturity options. This conclusion has been
disputed by other authors who showed that overreaction depends on the
model used to represent changes in the volatility. Xu and Taylor (1994)
presented and illustrated methods for estimating this term structure from
one row of the implied volatility matrix corresponding to nearest-the-money
options. They modeled the term structure of expected volatility and its
time series properties and use spot currency options on the British pound,
Deutschmark, Japanese Yen, and Swiss Franc quoted against the US dollar
in their empirical work. The study concerned the period from January 1985
to November 1989. The main results appeared in Xu and Taylor (1994).
When examining the relation between short-term and long-term implied
volatilities for the European option exchange index and Philips stock, the
principal result in Heynen, Kemma, and Vorst study is that the major
determinant for the specification of the term structure of implied volatility
relations is the level of the unconditional volatility. This latter study
seems correctly specified in the case of the EGARCH(1,1) stock return
volatility model. The Xu and Taylor’s study presented the following results.
First, implied volatilities vary significantly across maturities. Second, the
direction of the term structure of implied volatilities changes (up or down)
nearly once every two or three months. Third, the variations in expectations
regarding long-term volatility are significant although they are more slowly
than those regarding short-term volatility.
18.4. Empirical Results Regarding Information Costs

and Option Pricing
Option prices are extracted from a CD-ROM of the SBF-BOURSE DE
PARIS. The data concern the long maturity option quotations, the strike
prices, the maturity, and the CAC 40 quotations during October 1998.
Bellalah et al. (2002) used intra-day data for both quotations. These options
are European type options written on the CAC 40 index. Our study is based
on both calls and puts. Their maturity is six months. The interest rate is
taken from DATASTREAM. We choose the Euro Interbank Offered Rate
(EURIBOR) six months as the risk-free interest rate. The maturity chosen
is March 1999 and the minimum number of purchasable options is 50. The
EURIBOR interest rate is a daily rate.
18.4.1. Information costs and option pricing: The

estimation method
We use the B–S model to estimate the implied standard deviation (ISD)
through the simultaneous equations procedure. This procedure gives the
implied values of our parameter. The simultaneous equations procedure
estimates the value of the parameter, ISD, which minimizes the following
sum of squares:
n
min [COBS j − CBS j (ISD )]2
ISD
j=1
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 ,
through the simultaneous equations procedure, which gives us the implied

values of our parameters. The simultaneous equations procedure estimates
the values of the parameters, ISD, λC , and λS , which minimizes the
following sum of squares:
n

min [COBS j − CMj (ISD, IλC , IλS )]2 .
ISD,IλC ,IλS
j=1
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.
18.4.2. The asymmetric distortion of the smile

The prediction of the B–S model produces systematic deviations from
the observed price (“volatility smiles”) corresponds to a “gap” in the
formula of option valuation as noticed by Black’s (1976) model prices
as a function of the strike prices are diverging from the observed option
prices, so that the residuals for low strike prices tend to have different
signs from the residuals for high strike prices. Following At Sahalia
and Lo (2000), the risk-neutral density is the relevant density only for
risk-neutral investors. If investors were not indifferent to risk, then the
Table 18.1. Comparison between different models: B–S

and Bellalah’s model.
Call Put
MAE MAPE MAE MAPE
Black–Scholes 15,1903 0,1874 27,3895 0,4207

Bellalah 4,9669 0,0692 7,6109 0,0725
corresponding subjective probability could be different. The fundamen-

tal relation can be expressed as: subjective probabilities = risk-neutral
probabilities risk aversion adjustment coefficient. In the context of the
Hull and White (1987) model, Renault and Touzi (1996) proved that
observed symmetric smiles represent an evidence against the hypothesis
that log-asset returns are homoskedastic. Black (1976) pointed out that
“perhaps the most obvious causal relation runs from changes in the
value of the firm to stock returns and volatility changes. A drop in
the value of the firm will cause a negative return on its stock, and
will usually increase the leverage effect of the stock (. . .) that rise in
the debt-equity ratio will surely mean a rise in the volatility of the
stock”. This is the so-called leverage effect which is usually captured by a
negative correlation coefficient between the innovations of the two factors:
the returns and its volatility. The coupon-bond option model of Geske
(1979), the constant elasticity of variance model of Cox and Ross (1976),
and stochastic volatility models in incomplete market (with imperfect
negative correlation) of Heston (1993) should reproduce these facts (see
Bates (1997)). In this last case, both the objective and the risk-neutral
distributions of stock returns are leptokurtic (by the stochastic volatility
assumption) and skewed (due to the correlation between stock returns
and volatility). However, as pointed out by Ghysels et al. (1996) “it is
important to be cautious about tempting associations: stochastic implied
volatility and stochastic volatility; asymmetry in stocks and skewness
in the smile”.
18.4.3. Asymmetric Smiles and information costs

in a stochastic volatility model
Based on Renault and Touzi (1996) results, let the data-generating process
used in this sub-section be defined on a probability space (Ω, F, P ). The
underlying asset price process S is described by:
dS
= µ(t, S, σ)dt + σdW1 (t)
S
σ 2 = α(t, σ)dt + βdW2 (t)
where W = (W1 , W2 ) is a standard bi-dimensional Brownian motion. We

denote by r the instantaneous interest rate supposed to be constant, so that
the price of a zero-coupon bond maturing at time T is given by e−r(T −t) .
Let C be the price process of a European call option on the underlying
asset S with strike price K and maturity T . We introduce the variable

x = ln(S/Ke−r(T −t)). The call option is said to be ITM, if x > 0, OTM,
if x < 0, ATM forward, if x = 0 and ATM if x = r(T − t). Following Hull
and White (1987), we impose the assumption of non-systematic volatility
risk. The risk-neutral data generating bi-variate process is given by:
dS
= µ(t, S, σ)dt + σdW̃1 (t)
S
σ 2 = α(t, σ)dt + βdW̃2 (t)
where W̃ = (W̃ 1 , W̃ 2 ) is a standard bi-dimensional Brownian motion under

the risk-neutral probability with W̃2 = W2 . The Hull and White (1987)
formula is given by:
T
BS 1
C(S, σ ) = E C
2
S, σu du
2
T −t t
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
The main interest of this generalization is to use Renault (1997) results,

who formally proved that while the random feature of the volatility implies
the existence of a volatility smile, a very small discrepancy between S and
the implied S (say 0.1%) may explain a sensible skewness in the smile
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
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.
1895–1927.
of Finance, 13(3), 1975–2000.
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.
Finance, 13(2), 1826–1837.
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.
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
Melino, A and SM Turnbull (1990). Pricing foreign currency options with
stochastic volatility. Journal of Econometrics, 45, 239–265.
Merton, RC (1971). Optimum consumption and portfolio rules in a continuous

time model. Journal of Economic Theory, 3, 373–413.
Merton, R (1987). An equilibrium market model with incomplete information:
Renault, E (1997). Econometric models of option pricing errors. Working Paper,
GREMAQ.
Renault, E and N Touzi (1996). Option hedging and implied volatilities in a
stochastic volatility model. Mathematical Finance, 6, 279–302.
Rubinstein, M (1985). Nonparametric tests of alternative option pricing models.
Rubinstein, M (1994). Implied binomial trees. Journal of Finance, 3, 771–818.
Scott, L (1987). Option pricing when the variance changes randomly: theory, esti-
mation and an application. Journal of Financial and Quantitative Analysis,
22 (December), 419–437.
Stein, J (1989). Overreactions in the option markets. Journal of Finance,
44, 1011–1023.
Stein, EM and JC Stein (1991). Stock price distributions with stochastic volatility:
an analytic approach. The Review of Financial Studies, 4, 727–752.
Wiggins, JB (1987). Option values under stochastic volatility. The Journal of
Financial Economics, 19, 351–372.
Part VII
Option Pricing Models and Numerical Analysis
Pricing options with specific features becomes a difficult task in a Black–

Scholes world. When there is no analytic solution, we can use numerical
methods in the pricing of derivatives. For the sake of clarity, we write two
chapters in order to explain the basic tools used in numerical analysis. The
analysis can be followed by any student with a level which is equivalent to
MBA. Chapter 18 develops the basic concepts and applications of numerical
methods in option pricing. Chapter 19 concerns numerical methods and
partial differential equations for European and American options and
different underlying assets.
799
800
Chapter 19
RISK MANAGEMENT, NUMERICAL METHODS

AND OPTION PRICING
Chapter Outline
1. In Section 19.1, we introduce numerical analysis and simulation tech-

niques. The implicit and explicit difference numerical schemes are
applied to the discretization of the Black and Scholes (1973) partial
differential equation (PDE).
2. In Section 19.2, a numerical solution to the valuation of European call
options on a non-dividend-paying stock is presented.
3. In Section 19.3, we present a model for the valuation of American options
with a composite volatility and stochastic interest rates.
4. In Section 19.4, we introduce simulation techniques and in particular,
the Monte–Carlo method.
5. Appendix A presents simple concepts in numerical analysis and the heat
transfer equation.
6. Appendix B provides an algorithm for the valuation of a European call.
7. Appendix C provides the algorithm for the valuation of American long-
term index options with a composite volatility.
8. Appendix D presents the Monte–Carlo method and the dynamics of asset
prices.
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
a partial differential equation 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, which was 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.
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. A brief presentation of
these methods is given in this chapter. Finite difference methods approx-
imate each partial derivative in the above equation by its corresponding
formula. There are three main numerical schemes in the approximation of
the partial derivative: the implicit difference scheme, the explicit difference
scheme, and a weighted average of these schemes known as the Crank–
Nicolson numerical scheme.
Options on the spot-equity index, futures, and options on futures can be
ranked among the most remarkable financial innovations and the securities
markets have witnessed. These options and their underlying assets are
subject to market imperfections, which interact to make cash and carry
arbitrage with equity index futures far from riskless. These imperfections
include the difficulty of shorting stocks, the tracking errors, the execution
risk, and the fact that the underlying index is not traded (only its individual
constituents). These reasons make difficult the implementation of a risk-free
portfolio.
Since these securities are based on the same underlying stock index,
their prices must be related. If their prices do not obey the inter-market
relationships, then the relative mispricing, often documented in empirical
studies, should be instantaneously corrected given the high degree of
sophistication of market participants. To circumvent some difficuties, it
is a common place in research and practice to price index options using
either European formulas or American-style approximation methods. Under
the assumption that the underlying index pays dividends at a constant
proportional rate, it seems that index options are severely mispriced. The
development of option pricing models with stochastic interest rates and
stochastic volatility explains some of the bias observed in empirical tests.
Risk Management, Numerical Methods and Option Pricing 803
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.
19.1. Numerical Analysis and Simulation Techniques:

An Introduction to Finite Difference Methods
Several problems in financial economics do not have closed-form solutions.
Though, it is always possible to get some simulation results using numerical
analysis and simulation techniques. The non existence of analytic solutions
to some partial differential equation under their appropriate boundary
conditions leads to the choice of an appropriate numerical scheme. The
reader can refer to the appendix for a short overview of finite difference
methods and their applications to the heat transfer equation.
Consider, for example, the discretization of the following PDE using
finite difference methods:
1 2 2 ∂2c ∂c ∂c
σ S + rS + − rc = 0 (19.1)
2 ∂S 2 ∂S ∂t
The discretization of the price function c(S, t) representing the option price
can be done with respect to the state variable S and the time variable t.
The method consists in dividing the underlying asset price S into N sub-
intervals of length ∆S and the time variable (T − t) into M sub-intervals
of length ∆t. The time step is denoted by ∆t = k and the state step
as ∆S = h. Since the time variable is finite and corresponds to the option

maturity, the state variable must also be finite. Therefore, the highest value
of the underlying asset will be denoted by
SH (T − t)
SH and ∆S = and ∆t = .
N M
Starting from zero, then gives (N + 1) state steps and (M + 1) time
steps. This information can be represented on a simple diagram with
(N + 1)(M + 1) points. Finite difference methods approximate each partial
derivative in Eq. 19.1 by its corresponding formula. There are three main
numerical schemes in the approximation of the partial derivative: the
implicit difference scheme, the explicit difference scheme, and a weighted
average of these schemes known as the Crank–Nicolson numerical scheme.
19.1.1. The implicit difference scheme

The approximation of the partial derivatives in the Eq. (19.1) can be done
using either a forward difference or a backward difference. As denote by
c(i, j), the value of the option c(S, t) at position i and time j, where i refers
to the space index and j indicates the time variable.
Using a forward difference, the value of the option at each interior point
in the grid can be approximated by:
∂c 1
= [c(i + 1, j) − c(i, j)] (19.2)
∂S h
This partial derivative approximates the option value at node (i, j). The
difference between the point (i + 1) and i corresponds to the value of h.
Using a backward difference, the value of the option at each interior point
in the grid can be approximated as:
∂c 1
= [c(i, j) − c(i − 1, j)] (19.3)
∂S h
On an average of the above two approximations can be used to obtain:
∂c 1
= [c(i + 1, j) − c(i − 1, j)] (19.4)
∂S 2h
The difference between the point (i − 1) and (i + 1) corresponds to the value
∂2c
of 2h. The term ∂S 2 in the PDE can be approximated at the (i, j) point by
the following relation:

∂2c 1 1 1
= [c(i + 1, j) − c(i, j)] − [c(i, j) − c(i − 1, j)]
∂S 2 h h h
which is equal to:
∂ 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(i, j + 1) = ai c(i − 1, j) − c(i − 1, j) + bi c(i, j) + ci c(i + 1, j) (19.8)



 1 1

 ai = rik − σ2 i2 k

 2 2

bi = 1 + σ 2 i2 k + rk





 1 1
ci = − rik − σ 2 i2 k
2 2
This system must be solved with the appropriate conditions.
For a European call option on a non-dividend paying stock, the terminal
or maturity condition is:
c(S, T ) = max[0, ST − K]
This boundary condition can be approximated by:
c(i, M ) = max[0, ih − K] (19.9)
for i = 1, 2, 3, . . . , N .
Equation (19.9) with j = (M − 1) gives (N − 1) simultaneous

equations:
ai c(i − 1, M − 1) + bi c(i, M − 1) + ci c(i + 1, M − 1) = c(i, M ) (19.10)
for, i = 1, 2, 3, . . . , N − 1, which can be solved for (N − 1) unknowns: c(1,

M − 1), c(2, M − 1), . . . , c(N − 1, M − 1), where the c(i, M ) are known and
given by Eq. (19.9).
Solving this system provides option prices at time zero as a function
of different levels of the underlying asset, c(1, 0), c(2, 0), . . . , c(M − 1, 0).
The implicit difference method gives a relationship among three different
values of the option at time t + jk, which means the values c(i −
1, j), c(i, j), and c(i + 1, j) and one option value at time t + (j + 1)k
or c(i, j + 1). The above simple system can be solved by any method
of matrix inversion since it needs the solution of a simple tri-diagonal
matrix.
19.1.2. Explicit difference scheme

It is possible to solve the same PDE using a simpler numerical scheme: the
explicit difference scheme. In this case, we can approximate the different
partial derivatives in the PDE by:
∂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:
c(i, j) = a∗i c(i − 1, j + 1) + b∗i c(i, j + 1) + c∗i c(i + 1, j + 1) (19.14)

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
This system can be solved by inverting a tri-diagonal matrix. The value

of an option can be computed by imposing on this system the appropriate
boundary conditions. The explicit difference scheme gives a relationship
between one value of the option c(i, j) at time t+jk and three values at time
t + (j + 1)k, i.e., c(i − 1, j + 1), c(i, j + 1), and c(i + 1, j + 1).
19.1.3. An extension to account for information costs

Consider, for example, the discretization of the following PDE using finite
difference methods:
1 2 2 ∂ 2c ∂c ∂c
σ S + (r + λS )S + − (r + λc )c = 0 (19.15)
2 ∂S 2 ∂S ∂t
The discretization of the price function c(S, t) representing the option price
can be done with respect to the state variable S and the time variable t.
The method consists in dividing the underlying asset price S into N
sub-intervals of length ∆S and the time variable (T −t) into M sub-intervals
of length ∆t.
The analysis is similar to the method proposed above.
19.2. Application to European Options on Non-Dividend

Paying Stocks
Consider the valuation of a European call on a non-dividend paying stock
in the presence of information costs. The call price must satisfy the
following PDE:

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].
An additional condition must be imposed to reflect the fact that, for

high values of the underlying asset price, the partial derivative of the call
price approaches one,
∂c
lim = 1.
S→∞ ∂S
19.2.1. The analytic solution

The valuation of a call option consists in finding a solution to the
following PDE:

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.
19.2.2. The numerical solution

The partial derivative with respect to the stock price can be approximated
using a forward difference:

∂c(S, T ) 1
= [c(S + h, T ) − c(S, T )]
∂S h
or a backward difference:

∂c(S, T ) 1
= [c(S, T ) − c(S − h, T )]
∂S h
At points (i, j), the option partial derivative is approximated by the forward
difference:
∂c 1
= [c(i + 1, j) − c(i, j)].
∂S h
or a backward difference:
∂c 1
= [c(i, j) − c(i − 1, j)].
∂S h
A better approximation of the option value c(i, j) at point (i, j)

considers two steps h in the underlying asset price and divides
by two:
∂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
The option price can be determined at each instant j as a function of

its value at an instant (j − 1). Replacing the partial derivatives by their
approximations in the PDE gives:
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
for i = 1 to (N − 1) and j = 0 to (M − 1).

Multiplying by −k and gathering the terms in c(i − 1, j),
c(i, j), c(i + 1, j), and c(i, j − 1) gives the following tri-diagonal
system:
c(i, j − 1) = ai c(i − 1, j) + bi c(i, j) + ci c(i + 1, j) (19.20)



 1 1

 ai = rik − σ2 i2 k




2 2
2 2
bi = 1 + σ i k + rk





 1 1 2 2

ci = − rik − σ i k
2 2
The terminal or maturity condition for a European call option on a non-

dividend paying stock c(S, T ) = max[0, ST − K] is approximated by:
c(i, M ) = max[0, ih − K] for i = 1, 2, 3, . . . , N.
This corresponds to u(i, 0) = ih − K, when ih is higher than K, otherwise

it is equal to zero. For sufficiently higher values of the underlying asset, the
condition on the option’s partial derivative can be approximated by:
c(n, j) − c(n − 1, j) = h for j = 0, . . . , M.
The resulting system gives (N − 1) linear equations and (N + 1) unknowns

u(i, j). Since the option value at expiration is given, it is possible to generate
at each time step, all the time prices by inverting the resulting linear system
algorithm.
19.2.3. An application to European calls on non-dividend

paying stocks in the presence of information costs
Consider the valuation of a European call on a non-dividend paying stock
in the presence of information costs. The call price must satisfy the
following PDE:

1 2 2 ∂ 2c ∂c ∂c
σ S + (r + λS )S + − (r + λc )c = 0
2 ∂S 2 ∂S ∂t
under the following terminal or boundary condition c(S, T ) =

max[0, ST − K].
An additional condition must be imposed to reflect the fact that, for
high values of the underlying asset price, the partial derivative of the call
price approaches one. The same analysis can be applied.
19.3. Valuation of American Options

with a Composite Volatility
The starting points are the common factors in the returns on stocks and
bonds. There is a large body of empirical work dealing with the relationships
among stock prices, interest rates, and inflation (See for example, Copeland
and Stapleton, 1985; Peterson and Peterson, for more details, refer to
Bellalah et al., 1998).
19.3.1. The effect of interest rate volatility

on the index volatility
Fama and French (1993) identified common risk factors in the returns on
stocks and bonds. They showed that stock returns have shared variation
due to the stock market factors and that they are associated with bond
returns through shared variation in bond market factors. It is possible
to decompose the variance of the market portfolio into a proportion ϑ
corresponding to the variance of cash flows and a proportion (1−ϑ) resulting
from variations in interest rates.
We briefly review the models by Ramaswamy and Sundaresan RS
(1985) and Brenner et al. (1987), and Bellalah (2003). In the RS model,
the dynamics of the index price, S, are given by the following equation:
dS = [αS − δS]dt + σS SdzS (19.21)
where:
• α: the instantaneous expected return on S;

• δ: the dividend yield on the stock index;
• σS : the instantaneous expected standard deviation of returns on S and
• dzS : the increment of a Wiener process.
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:
dS = [αS − δS]dt + (σI + νσr )SdzS (19.22)
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.
The dynamics of the interest rate r are given by an Ornstein–Uhlenbeck

process where cov(dzS , dzr ) = ρdt.
19.3.2. Valuation of index options with a

composite volatility
Using the main results in the previous section, the model is based on several
assumptions.
Using the L-EH is equivalent to assuming that the expected instanta-
neous return on a coupon-paying bond (independent of its time to maturity)
is equal to the spot rate r(t) on a bond maturing instantaneously, i.e.,
Et [dB/B] = r(t)dt (19.23)
where Et is the mathematical expectation conditional on all the available

information at time t.
The dynamics of the short interest rates are described by the familiar
square-root process:1
√
dr(t) = κ[µ − r]dt + σr rdzr (19.24)
In this expression, κ[µ − r] corresponds to the mean reverting drift pulling

the short interest rate towards its long-term value µ, where κ defines the
speed of the adjustment. It is convenient to note that the square-root
process is more suited than the Ornstein–Uhlenbeck process for some well-
known reasons and because it does not allow for negative interest rates. The
choice of this process rather than that used by Brenner et al. (1987) is based
on the fact that the critics addressed to arbitrage models as the Vasicek
(1977) model do not apply to the CIR inter-temporal general equilibrium
term structure model.
The dynamics of the stock index or the underlying commodity are
described by the following equation:
√
dS = (αS − δS)dt + (σS + νσr r)SdzS (19.25)
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
1 This process was used in Cox et al. (1985) among others.

carry implies that F = SebT , where F is the futures price, b = r − δ is the

cost of carrying the commodity and T is the time to maturity. For a non-
dividend-paying asset, b = r and for a dividend-paying asset, b = r − δ. The
√
volatility of the underlying index is given by σS + νσr r. It is composed
of a specific volatility σS , the interest-rate volatility σr and a coefficient ν,
which transmits the effect of interest rates volatility to the index volatility.
Let us denote by U (S, r, t), the option value as a function of the two
underlying state variables and time. Applying Ito’s lemma and using the
standard hedging arguments, it is possible to construct a locally risk-less
portfolio with a risk-less bond and the underlying stock index. At equilib-
rium, the expected rate of return on this portfolio under the L-EH must be
the short risk-less interest rate and, under the free boundary formulation,
the option price must obey the following PDE in the continuation region:
1 √ ∂2U 1 √ √ ∂ 2U
(σS + νσr r)2 S 2 2 + ρσr S(σS + νσr r) r
2 ∂S 2 ∂S∂r
2
1 ∂ U ∂U ∂U ∂U
+ rσr2 2 + bS + κ(µ − r) − rU + = 0 (19.26)
2 ∂r ∂S ∂r ∂t
To find solutions for an American stock index call option with a strike
price K and a maturity date T , the PDE shown in Eq. (19.26) must be
solved under the following boundary conditions:
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
19.3.3. Numerical solutions and simulations

The PDE is discretized in a grid with respect to the two space-variables S,
r and time t. The Crank–Nicholson scheme is used and some simulations
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)
The option value is represented by a 3-dimensional array, U (S, r, t) =

U (ih, jh, nk) = U n (i, j). At each time step, s = T − nk, the first and the
second derivatives of S and r and the time derivative, in the PDE (19.26)
are approximated using the central differences:
∂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
If we replace the derivatives by their values,

the PDE (19.26) can be

approximated for each instant s = T − n − 12 k at points S = ih and
r = jh by the following system2 :

1 1 1 1 1
1 − AδS2 − Bδr2 − CHr HS − DHS − EHr U n (i, j)
2 2 8 4 4

1 2 1 2 1 1 1
= Aδ + Bδ − CHr HS + DHS + EHr + F U n−1 (i, j)
2 S 2 r 8 4 4
with:
√
(σI + νσr jh)2 (i2 k) σr2 jk
A= , B=
2G 2hG
√
ρσr i(σI jh + νσr jh)k (bih)k
C= , D=
hG hG
κ(µ − jh)k 1 − ( 12 jhk)
E= , F =
hG G
1
G = 1 + jhk, 1 ≤ i ≤ M − 1 and 1 ≤ j ≤ M − 13
2
Until now, the Alternating Direction Implicit Method (ADI) scheme
shows no particular problems. The approximation of the boundary condi-
tion when the interest rate is zero poses a serious problem and makes the
difference between our method and the numerical schemes reported in the
2 For ease of exposition, the following operators H , H , H H , δ 2 , δ 2 , δ 2 δ 2 are

S r S r S r S r
used with:
HS Hr U n (i, j) = HS [U n (i, j + 1) − U n (i, j − 1)] = U n (i + 1, j + 1)

− U n (i − 1, j + 1) − U n (i + 1, j − 1) + U n (i − 1, j − 1)
2 n+1
δS U (i, j) = U n+1 (i + 1, j) − 2U n+1 (i, j) + U n+1 (i − 1, j)
δr2 U n+1 (i, j) = U n+1 (i, j + 1) − 2U n+1 (i, j) + U n+1 (i, j − 1)

2 2 n
δS δr U (i, j) = U n (i + 1, j + 1) − 2U n (i, j + 1) + U n (i − 1, j + 1) − 2U n (i + 1, j)
+ 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 .
financial literature. The main difficulty consists in preserving the properties

of the numerical scheme, i.e., the tri-diagonal structure of the matrix, the
stability, and the speed of convergence.
When the interest rate r is zero, i.e., when j = 0, the PDE becomes:
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
and if option prices are close to the market prices. For more details, see
Bellalah (2003).
19.4. Simulation Methods: Monte–Carlo Method

Option prices are computed using the expected value of the terminal payoff.
Since, this value is known, it can be simulated. Boyle (1976, 1986, 1988)
developed a simulation method using the distribution of the underlying
asset at the option’s maturity date. When the process specifying the
distribution and the future movements in the underlying asset is known,
then it is a simple matter to simulate its values. Each time a simulation is
conducted, the computer generates a terminal value of the financial asset.
Repeating the simulation 1000 times gives the distribution of terminal asset
values. The distribution is used to extract the expected terminal asset value.
The simulation can be conducted as follows. Given a random variable
with a lax µdx, if we draw n times on a computer the values X1 , X2 , . . . , Xn
for a high value of n such that Xn follows the same law µdx, and the
sequence (Xn )n≥1 is a sequence of independent random variables, then
by the law of large numbers, the derivative price F can be expressed as
follows:

N
1
lim f (Xn ) = f (x)µ(dx)
N →∞ N
n=1
The Monte–Carlo method can be implemented on a computer as follows.

Construct a sequence of numbers (Un )n≥1 in order to correspond to a
uniform sequence of independent random variables on the interval [0, 1].
Then search for a function to which (u1 , u2 , . . . , up ) corresponds to
F (u1 , u2 , . . . , up ) such that the law of the random variable F (u1 , u2 , . . . , up )
is the known law of µ(dx). The sequence of random independent variables
(Xn )n≥1 with
Xn = F [U(n−1) p+1 , . . . , Unp ]
follows the law µ.

To generate the sequence (Un )n≥1 with a computer, we use the random
function which gives a pseudo-random number between 0 and 1 or an integer
in a fixed interval. The function random or Rand(.) is often available on
computers. The main drawback of the Monte–Carlo method is that it is
cumbersome in the valuation of American options. The method is often
used in the valuation of European options.
19.4.1. Simulation of Gaussian variables

A random variable X is a Gaussian variable with a zero mean and a unit
variance when its density function is given by:
“ ”
1 x2
n(x) = √ e 2
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 ).
If there are two random uniform independent variables (U1 , U2 ) on the

interval [0, 1], then the cos(2πU2 ) −2 ln(U1 ) follows a normal distribution
with a zero mean and a unit variance.
In order to simulate a Gaussian law with mean µ and a variance σ2 ,
we just use the change in variables:
z = µ + σg.
For example, the following instuction may be used in a program written

in Turbo Pascal to generate the Gaussian law:

Gaussian = µ + σ cos (2πRandom) −2 log (Random)
19.4.2. Relationship between option values

and simulation methods
The risk-neutral random walk for the underlying asset S is given by:
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.
The method can be implemented by generating random numbers from

a standardized normal distribution. At each timestep, the asset price is
generated using these random increments. For example, we can use the
following dynamics:
√
δS = rSδt + σS δtφ
where φ is taken from a standardized normal distribution. The latest value

for S is used to compute δS and hence the next value for S.
It is possible for the log-normal random walk to use a simpler method
with reference to the equation:

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,
as opposed to a single stock and by the relative mispricing of long-term

index options.
We extend the classic (ADI) based on the Crank–Nicholson scheme
in 2-space dimensions and provide an efficient, stable, and convergent
algorithm for the pricing of these options. Some numerical receipts are
used to reduce the computation time. The algorithm reduces to solving
tri-diagonal systems and may be used to handle other complex problems
in financial economics. The solution method is quite general to be applied
to any option valuation problem in the presence of 2-state variables. The
results indicate the significant effect of interest rates and the volatility on
the pricing and the early exercise of long-term index options.
Fifth, we develop some simulation methods and in particular, the
Monte–Carlo Method.
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.
Appendix A: Simple Concepts in Numerical Analysis

The heat transfer equation
Some equations of the parabolic type, used in the pricing of derivatives,
can be transformed to the heat transfer equation. In its simplest form, this
equation can be written as:
∂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
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.
Some simple numerical schemes for the heat transfer equation

The following analysis shows how to find a numerical solution for the heat
transfer equation. Consider a function u(S, t) for which the underlying asset
S is divided into ih and time corresponds to jk, where h is the state step
and k is the time step.
The indexes (i, j) vary in the plane (S, t). Assume that the discrete
option values u(i, j) are known for all space indexes i at each instant of
time j, where j takes the values 0, 1, . . . , j.
The discretization of the heat transfer equation needs first the compu-
tation of the partial derivatives. The time derivative can be approximated
using the finite decentered difference as:
(j)
u(i, j + 1) − u(i, j) ∂u
= + o(k)
k ∂t i
This approximation is exact to the first order.
The derivative with respect to the space variable or the underlying
asset can be approximated using the following finite centered difference:
2 (j)
u(i, j + 1) − 2u(i, j) + u(i − 1, j) ∂ u
= + o(h2 ) (A.2)
h2 ∂S 2 i
This gives a numerical scheme, often referred to as the explicit scheme.
k
u(i, j + 1) = u(i, j) + [u(i + 1, j) − 2u(i, j) + u(i − 1, j)] (A.3)
h2
When k = o(h), the explicit numerical scheme is accurate to the first order.
Equation (A.2) can be applied at the level (j + 1) rather than j. In this

case, using the following two equations:
(j+1)
u(i, j + 1) − u(i, j) ∂u
= + o(k) (A.4)
k ∂t i
2 (j+1)
u(i + 1, j + 1) − 2u(i, j + 1) + u(i − 1, j + 1) ∂ u
2
= + o(h2 )
h ∂S 2 i
(A.5)
gives a totally implicit numerical scheme:

k
u(i, j + 1) = u(i, j) + [u(i + 1, j + 1) − 2u(i, j + 1) + u(i − 1, j + 1)]
h2
(A.6)
The well-known Crank–Nicolson scheme can be obtained using an average

of the numerical schemes in Eqs. (A.3) and (A.6):
k
u(i, j + 1) = u(i, j) + [u(i + 1, j)
2h2
− 2u(i, j) + u(i − 1, j) + u(i + 1, j + 1)]
− 2u(i, j + 1) + u(i − 1, j + 1)] (A.7)
This numerical scheme is also of the implicit type. It is a simple form of a

more general scheme, where θ is between 0 and 1. In this case, we have:
k
u(i, j + 1) = u(i, j) + {(1 − θ)[u(i + 1, j) − 2u(i, j) + u(i − 1, j)]
2h2
+ θ[u(i + 1, j + 1) − 2u(i, j + 1) + u(i − 1, j + 1)]} (A.8)
This general scheme takes different forms according to the values attributed
to θ with 0 ≤ θ ≤ 1.
When θ = 1, this scheme is of the implicit type.

When θ = 0, this scheme is of the explicit type.
When θ = 12 , this scheme is of the Crank–Nicolson type.
Appendix B: An Algorithm for a European Call

For j = 1, . . . , M
a(1) = 0,
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).
Appendix C: The Algorithm for the Valuation of American

Long-term Index Options with a Composite
Volatility
Read: K, σI , ν, σr , ρ, d, κ, µ, h, k
For i = 0, M
For j = 0, M, U (i, j) = 0
if (ih − K) > 0, then U (i, j) = ih − K, End if
End for M
(For a put (ih − K)is replaced by (K − ih))
For n = 1, nn (time to maturity)
For i = 1, M − 1
a(0) = 0, b(0) = 1, c(0) = 0, d(0) = U (i, 0)
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).
Solve the system (M, a, b, c, d, x) by inverting the matrix.

For j = 0, M , U U (i, j) = x(j), End
End For i = 1, M − 1.
This is the end of the first step. Before beginning the second step,
a special treatment is done for j = 0.
a(0) = 0, b(0) = 1, c(0) = 0, and d(0) = 1
(For a put, d(0) becomes d(0) = K)
For i = 1, M − 1
σI2 (i2 k)
aa = 2
(bih)k
bb = h
cc = κµkh
a(i) = − 21 (aa − 12 )bb
b(i) = 1 + aa
c(i) = − 12 (aa + 12 bb)
d(i) = −a(i)U (i − 1, 0) + (1 − aa − 32 cc)U (i, 0) − c(i)U (i + 1, 0)
+ 12 cc(−4U (i, 1) + U (i, 2))
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, a(M ) = 0, b(M ) = 1, c(M ) = 0, d(M ) = 0)
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
if U (i, j) < (ih − K) then U (i, j) = ih − K End if End for i,

End for j,
End for n = 1, nn
(For a put, the modification is:
if U (i, j) < (K − ih) then U (i, j) = K − ih End if).
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:
F (X(t)) = aX 2 (t) + betX(t) with a, b > 0
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
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
2. We denote by H(X(t)) = betX(t) , so:

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
The computation of the first integral gives:

t t t
τ 2 τ X(τ )
τ eτ X(τ ) dX(τ ) = etX(t) − 1 − eτ X(τ )X(τ )dτ + e dτ
0 0 0 2
Multiplying this equation by b, we deduce the result for the second question.
3. Since F (X(t)) = aX 2 + betX(t) , we have:

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
Using the answer for the first question, we obtain:
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
This gives the desired result:
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
Appendix D: The Monte–Carlo Method and the Dynamics

of Asset Prices
Consider the following data for the application of the Monte–Carlo
method. Initial time: 15/06/2002, maturity date: 15/06/2003. The following
parameters are used:
Initial asset price at time 0: 200, drift 1%, volatility = 40%, interest
rate = 3%, and time step = 0.01.
The Table D.1 provide the simulations using the above parameters.
We report the results for the first, second, and third simulations, Sim 1 to
Sim 3. We calculate the values of the calls and the puts for different strike
prices. The underlying asset is simulated using the Monte–Carlo method
for a drift of 5%. For each realization, the final stock is provided in the
last row. We provided the option payoff, the mean of all the payoffs over
all simulations. The present values of the means correspond to the option
values. This method is suitable for path-dependent options.
Table D.1.
Asset 200 S mean Time Sim 1 Sim 2 Sim 3
Drift 1% 200.00 0.00 200.00 200.00 200.00

Volatility 40% 199.34 0.01 197.88 209.89 210.36
Time-step 0.01 198.78 0.02 189.84 214.57 207.77
Interest Rate 3% 198.99 0.03 204.55 209.43 201.61

198.94 0.04 208.89 215.69 200.50
198.71 0.05 228.25 213.42 203.37
June 2003 200.08 0.06 223.95 219.91 207.10

199.67 0.07 230.12 220.61 212.48
199.13 0.08 228.91 212.36 225.35
231.38 3.61 134.49 255.98 148.51
Strike 1 290.13 CALL Payoff

Mean 55.52
PV 49.82
Strike 2 118.42 PUT Payoff
Mean 13.31
PV 11.94
Strike 3 623 CALL Payoff
Mean 9.38
PV 8.42
Strike 4 80 PUT Payoff
Mean 4
PV 3.59
References
Bellalah, M (2003). Valuation of long term options. International Journal of
Finance.
Boyle, P (1976). Rates of return as random variables. Journal of Risk and
Insurance, 43 (December), 694–711.
Brenner, M, G. Courtadon and M Subrahmanyam (1987). The valuation of stock
index options, Solomon Center Working Paper, 414 (June).
Fama, E and KR French (1993). Common risk factors in the returns on stocks
and bonds. Journal of Financial Economics, 33, 3–56.
Harrison, JM and DM Kreps (1979). Martingales and arbitrage in multiperiod

security 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.
Ramaswamy, K and S Sundaresan (1985). The valuation of options on futures
contracts. Journal of Finance, 5, 1319–1341.

Chapter 20
NUMERICAL METHODS AND PARTIAL

DIFFERENTIAL EQUATIONS FOR EUROPEAN
AND AMERICAN DERIVATIVES WITH
COMPLETE AND INCOMPLETE INFORMATION
Chapter Outline
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
10. Appendix D gives a detailed algorithm for the valuation of American

CBs with several call and put provisions in the presence of dividends
and coupon payments.
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
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.
20.1. Valuation of American Calls on Dividend-Paying

Stocks
20.1.1. The Schwartz model
Schwartz (1977) assumed that the Black–Scholes (1973) equation applies
between dividend dates:

1 2 2 ∂ 2C ∂C ∂C
σ S + rS + − rC = 0 (20.1)
2 ∂S 2 ∂S ∂t
He used the following boundary conditions to solve for the American

call value when there are dividends:
C(S, 0) = max[0, S − K] (20.2)

C(0, t) = 0 (20.3)
C(S, T + ) = max[0, S − K, C(S − d, T − )] (20.4)
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
option’s derivative with respect to the underlying asset price:

∂C(S, τ )
lim =1 (20.5)
S→∞ ∂S
This condition must be satisfied for a sufficiently high level of the underlying
asset price.

Consider a sub-division of the state variable into h equally spaced units of
the underlying asset and the time variable into k units of time, or:
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:
ai C(i − 1, j) + bi C(i, j) + ci C(i + 1, j) = C(i, j − 1) (20.6)
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
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.
At a dividend date, condition (20.4) is approximated by:

+ d − d −
C(i, j ) = C i − , j for C i − , j ≥ ih − K
h h

d
C(i, j + ) = ih − K for C i − , j − ≤ ih − K.
h
For sufficiently high values of the underlying asset, Eq. (20.5) is
approximated by C(i, j) − C(i − 1, j) = h for j = 0 to m. For each value
of j, there is a system of (n − 1) linear equations with (n + 1) unknowns.
Using Eqs. (20.3) and (20.5) gives a system with (n + 1) equations and
(n + 1) unknowns. This system can be solved by inverting the matrix to
give all possible values of the option price at each instant j.
Appendix B presents a detailed algorithm corresponding to this model.
20.2. American Puts on Dividend-Paying Stocks

20.2.1. The Brennan and Schwartz model
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 valuation of the put is given by the solution
to the B–S PDE under the following conditions:
P (S, T ) = max[0, K − S] (20.7)

P (S, t) ≥ max[0, K − S] (20.8)
P (S, t) ≥ 0 (20.9)
P (S, t) ≤ K (20.10)

∂P (S, t)
lim = 0. (20.11)
S→∞ ∂S
Equation (20.7) corresponds to the put value at the maturity date,

which is simply greater than zero and the intrinsic value. Equation (20.8)
shows that the American put value must be greater than its exercise value
at each instant. Equations (20.9) and (20.10) provide respectively, the
minimum and the maximum price for a put option. Equation (20.11) results
from Eqs. (20.9) and (20.10) and the convexity of the option’s price. On a
dividend date, the following condition must be satisfied:
P (S, t− ) = max[K − S, P (S − Dt , t+ )]. (20.12)
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.

Consider the discretization of the state variable (the stock) space into h
small, equally spaced units and the time variable (time) into k small units
of time. Also, we will use a new time variable, τ = T − t, instead of t, the
calendar time. The discretization of the asset price and the time to maturity
can be written as Si = ih for i = 0 to n, τj = jk for j = 0 to m. The put
price P (S, τ ) is approximated by P (Si , τj ) = P (ih, jk). Approximating the
option partial derivatives by their values and replacing in the differential
equation gives the following system:
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
the solution of P (i, j) as a function of P (i, j − 1). Equation (20.8) is

approximated by P (i, j) ≥ K − ih for i = 0 to n and the solution
must satisfy this condition. The value of the underlying asset for which
this inequality becomes a strict equality provides the critical asset price,
Sc = ic h, corresponding to an optimal exercise policy. At a dividend date,
Eq. (20.12) is approximated by

d d
P (i, j − ) = P i − , j + for P i − , j + > K − ih,
h h

d
P (i, j − ) = K − ih for P i − , j + < K − ih.
h
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.
20.3. Numerical Procedures in the Presence of Information

Costs: Applications
Finite difference methods allow the valuation of derivatives by solving the
differential equation numerically under the appropriate conditions.
20.3.1. Finite difference methods in the presence

Consider for example, the pricing of an option on a non-dividend paying
stock. The B–S differential equation in the presence of information uncer-
tainty is written as:
∂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
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:
aj Vi,j−1 + bj Vi,j + cj Vi,j+1 = Vi+1,j (20.20)

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,
Vi,j−1 , Vi,j , Vi,j+1 , and Vi+1,j .
20.3.2. An application to the American put using explicit

or implicit finite difference methods
Consider now the pricing of an American put option on a non-dividend
paying stock. Re-call that the put’s payoff at maturity T is max[K −j∆S, 0],
where K stands for the strike price. This terminal condition can be
approximated by:
VN,j = max[K − j∆S, 0] for j = 0, 1, 2, . . . , M. (20.21)
When the underlying asset price is zero, the put price corresponds to its
strike price or:
Vi,0 = K for i = 0, 1, 2, . . . , N. (20.22)
When the underlying asset price tends to infinity, the put price approaches
zero or
Vi,M = 0, for i = 0, 1, 2, . . . , N. (20.23)
Using Eq. (20.20) with i = (N −1) gives (M −1) simultaneous equations:
aj VN −1,j−1 + bj VN −1,j + cj VN −1,j+1 = VN,j (20.24)
for j varying from 1 to (M −1). Besides, since VN −1,0 = K and VN −1,M = 0,

we can solve the (M − 1) Eq. in (20.24) for the (M − 1) unknowns
VN −1,1 , . . . , VN −1,M−1 . To account for the possibility of an early exercise,
each value of VN −1,j must be compared with the option’s intrinsic value
K − j∆S.
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.
Explicit versus implicit finite difference methods

∂V ∂2V
It is possible to approximate the partial derivatives ∂S and ∂S 2 at point
(i, j) on the grid as follows:
∂V Vi+1,j+1 − Vi+1,j−1 ∂ 2V Vi+1,j+1 − Vi+1,j−1 − 2Vi+1,j

= , = .
∂S 2∆S ∂S 2 2∆S 2
In this case, Eq. (20.20) can be written as:
Vi,j = a∗j Vi+1,j−1 + b∗j Vi+1,j + c∗j Vi+1,j+1
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
This approximation refers to the explicit finite difference method.

The explicit method gives a relationship between one value of the option
at time (t + i∆t) and three values at time (t + (i + 1)∆t). The four values
are respectively fi,j , fi+1,j−1 , fi+1,j , and fi+1,j+1 .
20.4. Convertible Bonds

20.4.1. Specific features of CB
Following Brennan and Schwartz (1977), we use the following notations:
V (t): market value of the firm’s securities;

W (V, t): market value of a CB with par value $1000;
CP (t): call price at time t at which the bonds may be called for
redemption;
B(V, t): value of an otherwise identical bond with no conversion provi-

sion and
D(t): dividend payment to the common stocks.
Suppose that there are N c CB and N0 shares before conversion. We denote
by q(t), the number of shares into which a bond can be converted at time
t, I the coupon payments at each payment date, and i = NIc , the periodic
coupon payment per bond.
Since each CB can be converted into q(t) shares, the conversion value,
C(V, t) is given by:
q(t)V (t)
C(V, τ ) = = z(t)V (t) (20.25)
[N0 + Nc q(t)]
q(t)
with z(t) = [N0 +N c q(t)]
. Since an optimal conversion strategy implies that
the value of the unconverted bond is at least equal to the conversion value,
the following arbitrage condition must be satisfied:
W (V, t) ≥ C(V, t) (20.26)
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:
VIC (V, t) = max[CP (t), C(V, t)]. (20.27)
Moreover, at time t = t∗ , when the bond becomes callable (because in

practice bonds are not called until after a certain period), its value must
satisfy
W (V, t∗ ) = C(V, t∗ ) if C(V, t∗ ) ≥ CP (t∗ )
and at any time during the call period, the bond’s value cannot exceed the
call price, i.e.,
W (V, t) ≤ CP (t). (20.28)
20.4.2. The valuation equations

The CB can be valued using the B–S PDE:

1 2 2 ∂ 2W ∂W ∂W
σ V + rV + − rW = 0 (20.29)
2 ∂V 2 ∂V ∂t
under the appropriate boundary conditions. Equation (20.30) shows that

the total value of the bonds is less than the firm’s value. This is because
the firm’s value is equal to that of its stocks and bonds.
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:
W (V, t) ≤ B(V, t) + z ∗ (t)V (20.32)
where z ∗ (t) is the maximum value of z(τ ) for t ∈ (t, T ).

Equation (20.33) illustrates the option offered to the bondholder for
conversion.
W (V, t) > C(V, t) = z(t)V. (20.33)
Equation (20.34) indicates the pay off of the CB at the maturity date.
W (V, T ) = z(T )V, z(T )V ≥ 1000

1000
W (V, T ) = 1000, 1000Nc ≤ V ≤ (20.34)
z(T )
V
W (V, T ) = , V ≤ 1000Nc.
Nc
Equation (20.35) shows the constraint on the call price during the call
period. It indicates that the CB price cannot exceed the call price, otherwise
the issuer will call back the bonds,
W (V, t) ≤ CP (t). (20.35)
Equation (20.36) is a high-contact condition, which applies for a sufficiently

high value of V .

∂W (V, t)
lim = z(t). (20.36)
S→∞ ∂V
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+ .
W (V, t− ) = max[W (V − D, t+ ), z(t− )V ]. (20.37)
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.
W (V, t− ) = min[W (V − I, t+ ) + i, CP (t− )] (20.39)
Brennan and Schwartz (1977) presented a numerical solution for the

valuation of American CBs when there are discrete distributions to the
underlying asset and call and put provisions.

Since there is no closed-form solution to the B–S differential equation under
the above boundary conditions, numerical methods are useful in solving
such problems. Using a time variable τ = T − t, instead of the calendar
time t, the discretization of the underlying asset price and the time to
maturity is:
Vi = ih for i = 0 to n
τj = jk for j = 0 to m.
The CB is a solution to the B–S PDE which is approximated by:
ai W (i − 1, j) + bi W (i, j) + ci W (i + 1, j) = W (i, j − 1) (20.40)
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).
Equation (20.31) is written as:
W (0, j) = 0 for j = 0 to m
The discretization of Eq. (20.34) gives:
W (i, 0) = z(0)V = zhi for zhi ≥ P

P
W (i, 0) = P for P ≤ hi ≤
z
W (i, 0) = V = hi for hi ≤ P
where P stands for the par value of the CB.
Equation (20.36) is approximated by:
W (n, j) − W (n − i, j)
= z. (20.41)
h
Using the above system and Eq. (20.41), the values of W (i, j) can be
determined in a recursive manner from W (i, j − 1) since all the W (i, 0)
are given for all values of i. At a call date, Eq. (20.36) is replaced by
Eq. (20.35), or
W (i, j) ≤ CP (j). (20.42)
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.
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 .
20.5. Two-Factor Interest Rate Models and Bond Pricing

within Information Uncertainty
We denote at time t by Z(r, l, t; T ) the price of a zero-coupon bond as a
function of the spot interest rate r, the long interest rate l, and the maturity
date T . The dynamics of the spot rate are specified by: dr = udt + wdX1 ,
where the parameters u and w can depend on r and t. The dynamics of the
long rate are given by dl = pdt + qdX2 , where the parameters p and q can
depend on l and t.
Consider a portfolio with a long position in Z(r, l, t; T ) and two short
positions in two zero-coupon bonds with different maturities T1 and T2 .
The initial portfolio value is given by:
Π = Z(r, l, t; T ) − ∆1 Z(r, l, t; T1 ) − ∆2 Z(r, l, t; T2 ).
The change in this portfolio’s value can be written as:

∂Z ∂Z1 ∂Z2
(H(Z) − ∆1 H(Z1 ) − ∆2 H(Z2 ))dt + − ∆1 − ∆2 dr
∂r ∂r ∂r

∂Z ∂Z1 ∂Z2
+ − ∆1 − ∆2 dl (20.44)
∂l ∂l ∂l
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
Table 20.1. Prices for different levels of the underlying asset V .
u(0) = 0 u(1) = 1.960000 u(2) = 3.230000 u(3) = 3.910000

u(4) = 4.740000 u(5) = 5.560000 u(6) = 6.370000 u(7) = 7.180000
u(8) = 7.990000 u(9) = 8.800000 u(10) = 9.610000 u(11) = 10.420000
u(12) = 11.230000 u(13) = 12.040000 u(14) = 12.850000 u(15) = 13.660000
u(16) = 14.480000 u(17) = 15.300000 u(18) = 16.140000 u(19) = 17.020000
u(20) = 17.940000 u(21) = 18.950000 u(22) = 20.070000 u(23) = 21.330000
u(24) = 22.730000 u(25) = 24.290000 u(26) = 25.960000 u(27) = 27.690000
u(28) = 29.400000 u(29) = 31.000000 u(30) = 32.420000 u(31) = 33.600000
u(32) = 34.520000 u(33) = 35.190000 u(34) = 35.640000 u(36) = 36.080000
u(37) = 36.160000 u(38) = 36.190000 u(39) = 36.210000 u(40) = 36.214000
u(41) = 36.215000 u(42) = 36.215800 u(43) = 36.215800 u(44) = 36.215870
u(45) = 36.215875 u(46) = 36.215875 u(47) = 36.215875 u(48) = 36.215875
u(49) = 36.215875 u(50) = 36.215875 u(51) = 36.215875 u(52) = 36.215875
u(53) = 36.215875 u(54) = 36.215875 u(55) = 36.215875 u(56) = 36.215875
u(57) = 36.215875 u(58) = 36.215875 u(59) = 36.215760 u(60) = 36.215876
u(61) = 36.215876 u(62) = 36.215878 u(63) = 36.215870 u(64) = 36.215880
u(65) = 36.215890 u(66) = 36.215916 u(67) = 36.215951 u(68) = 36.216000
u(69) = 36.216100 u(70) = 36.216300 u(71) = 36.216600 u(72) = 36.217100
u(73) = 36.217900 u(74) = 36.219100 u(75) = 36.220900 u(76) = 36.223600
u(77) = 36.227500 u(78) = 36.233000 u(79) = 36.240600 u(80) = 36.250000
u(81) = 36.260000 u(82) = 36.280000 u(83) = 36.300000 u(84) = 36.320000
u(85) = 36.360000 u(86) = 36.390000 u(87) = 36.440000 u(88) = 36.480000
u(89) = 36.540000 u(90) = 36.590000 u(91) = 36.660000 u(92) = 36.720000
u(93) = 36.790000 u(94) = 36.860000 u(95) = 36.920000 u(96) = 36.990000
u(97) = 37.070000 u(98) = 37.140000 u(99) = 37.210000 u(100) = 37.280000
u(101) = 37.350000 u(102) = 37.430000 u(104) = 37.580000 u(105) = 37.660000
u(106) = 37.740000 u(107) = 37.830000 u(108) = 37.920000 u(109) = 38.000000
u(109) = 38.000000 u(110) = 38.090000 u(111) = 38.180000 u(112) = 38.270000
u(113) = 38.360000 u(114) = 38.440000 u(115) = 38.530000 u(116) = 38.610000
u(117) = 38.690000 u(118) = 38.760000 u(119) = 38.840000 u(120) = 38.910000
u(121) = 38.980000 u(122) = 39.050000 u(123) = 39.120000 u(124) = 39.190000
u(125) = 39.260000 u(126) = 39.330000 u(127) = 39.410000 u(128) = 39.490000
u(129) = 39.570000 u(130) = 39.650000 u(131) = 39.740000 u(132) = 39.830000
u(133) = 39.930000 u(134) = 40.030000 u(135) = 40.130000 u(136) = 40.240000
u(137) = 40.340000 u(138) = 40.450000 u(139) = 40.560000 u(140) = 40.660000
u(141) = 40.760000 u(142) = 40.860000 u(143) = 40.950000 u(144) = 41.040000
u(145) = 41.120000 u(146) = 41.200000 u(147) = 41.280000 u(148) = 41.350000
u(149) = 41.420000 u(150) = 41.500000 u(151) = 41.570000 u(152) = 41.640000
u(152) = 41.640000 u(153) = 41.710000 u(154) = 41.780000 u(155) = 41.850000
u(156) = 41.920000 u(157) = 41.990000 u(158) = 42.060000 u(159) = 42.130000
u(160) = 42.200000 u(161) = 42.270000 u(162) = 42.340000 u(163) = 42.410000
u(164) = 42.490000 u(165) = 42.560000 u(166) = 42.630000 u(167) = 42.710000
u(168) = 42.780000 u(169) = 42.790000 u(170) = 42.810000 u(171) = 42.850000
u(172) = 42.960000 u(173) = 43.260000 u(174) = 43.490000 u(175) = 43.750000
u(176) = 43.990000 u(177) = 44.250000 u(178) = 44.490000 u(179) = 44.750000
u(180) = 44.990000 u(181) = 45.250000 u(182) = 45.490000 u(183) = 45.750000
u(184) = 46.000000 u(185) = 46.250000 u(186) = 46.500000 u(187) = 46.750000
u(188) = 47.000000 u(189) = 47.250000 u(190) = 47.500000 u(191) = 47.750000
u(192) = 48.000000 u(193) = 48.250000 u(194) = 48.500000 u(195) = 48.750000
u(196) = 49.000000 u(197) = 49.250000 u(198) = 49.500000 u(199) = 49.750000
It is possible to construct a hedged portfolio using the appropriate ∆1 and

∆2 that allow the coefficients of dr and dl to be zero in Eq. (20.44).
This allows the construction of a risk-free portfolio which earns the
risk-free rate and the return corresponding to information costs. This leads
to the following system of three equations:
∂Z ∂Z1 ∂Z2
− ∆1 − ∆2 =0
∂l ∂l ∂l
H (Z) − ∆1 H (Z1 ) − ∆2 L (Z2 ) = 0
with H (Z) = H(Z) − (r + λZ )Z, where λZ corresponds to the shadow
cost of incomplete information corresponding to the bond market. Using a
matrix notation, this system can be written as:
 
H (Z) H (Z1 ) H (Z2 )
 
M = ∂Z/∂r ∂Z1 /∂r ∂Z2 /∂r
∂Z/∂l ∂Z1 /∂l ∂Z2 /∂l
with the additional condition that det(M) = 0.
Using this matrix M, it is possible to write:
∂Z ∂Z
H (Z) = (γr w − u)+ (γl q − p) .
∂r ∂l
The standard arbitrage arguments allow the derivation of the following
equation for the bond price:
∂Z 1 ∂ 2Z ∂ 2Z 1 ∂2Z ∂Z
+ w2 2 + ρwq + q 2 2 + (u − γr w)
∂t 2 ∂r ∂r∂l 2 ∂l ∂r
∂Z
+ (p − γl q) − (r + λZ )Z = 0 (20.45)
∂l
where γr and γl correspond respectively to the market prices of risk.
The long-term interest rate refers to the yield on a consol bond, i.e., a
fixed coupon-paying bond for which the maturity is infinite. The yield on a
consol bond paying a coupon of $1 each year until infinity is given by l = C1o .
In this context, the pricing equation becomes:
∂Co ∂Co
H (Co ) + 1 = (γr w − u) + (γl q − p)
∂r ∂l
This equation is similar to Eq. (20.45), where the additional term 1
corresponds to the coupon payments. Replacing 1l in this equation gives:
q2
p − γl q = l2 − (r + λl )l + .
l2
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
20.6. CBs Pricing within Information Uncertainty

20.6.1. The pricing of CBs
This analysis assumes that the coupon is known. Since the CB price is a
function of the underlying asset price S and time t, it can be written as
V = V (S, t). Consider a portfolio with a long position in the CB and a short
position in δ units of the underlying asset. Using hedging arguments, it is
possible to show that the change in the portfolio’s value over a short-time
interval is:
∂V ∂V 1 ∂ 2V ∂V
dΠ = dt + dS + σ 2 S 2 dt − dS.
∂t ∂S 2 ∂S 2 ∂S
Risk can be eliminated from the portfolio by choosing ∆ = ∂V ∂S
. As before,
the return on the risk-free portfolio must be equal to the risk-less rate plus
information costs paid by the investor in both markets. This gives:
∂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
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) .
20.6.2. Specific call and put features

The provision that allows the bond issuer to purchase back the bond at a
specified call price corresponds to a call feature. The provision that allows
the bond holder to return the bond to the issuing company corresponds to
a put feature. In the pricing of callable and puttable corporate CBs, the
following two conditions must be imposed V (S, t) ≤ CP and V (S, t) ≥ PP ,
where CP refers to the call price and PP to the put price.
20.6.3. The pricing of CBs in two-factor models

within information uncertainty
In the presence of stochastic interest rates, the bond price can be written
as a function of the underlying asset price, interest rates, and time as
V = V (S, r, t). In general, the following dynamics are used for the under-
lying asset:
dS = µSdt + σSdX1
and interest rates:
dr = u(r, t)dt + w(r, t)dX2
where the two Wiener processes are correlated by ρ(r, S, t) with:
E[dX1 dX2 ] = ρdt.
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
In this context, the pricing of the CB needs the construction of a portfolio

with a long position in the CB and two short positions in the underlying
asset and zero-coupon bonds. The short side justifies the use of two deltas
because of the two risks linked to the underlying asset price and the interest
rate risk Π = V − ∆1 S − ∆2 Z. The appropriate choice of the hedging ratios
allows the elimination of risk from this portfolio with

∂V ∂V ∂Z
∆1 = and ∆2 = .
∂S ∂r ∂r
Putting together the terms in T1 and T2 , re-arranging and imposing the
condition that the return from a hedged portfolio must be equal to the
risk-less rate plus information costs on the corresponding markets gives
∂V 1 ∂2V ∂2V 1 2 ∂2V ∂V

+ σ2 S 2 2
+ ρσSw + w 2
+ (r + λS )S
∂t 2 ∂S ∂S∂r 2 ∂r ∂S
∂V
+ (u − γw) − (r + λV )V = 0
∂r
The presence of information costs is justified by the costs suffered by
investors to implement arbitrage because arbitrage is not costless. In fact,
not all investors can implement risk-less arbitrage. This is only possible for
investors who are informed about the presence of arbitrage opportunities.
When the interest rate is constant, this equation reduces to the extended
B–S equation.
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
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.
Appendix A: A Discretizing Strategy for Mean-Reverting

Models
A simple two-point upstream technique in the presence
of an implicit scheme
Vetzal (1998) illustrated the methods using the one-factor Vasicek (1977)
model.
dr = κ(θ − r)dt + σdB
with k: the speed of adjustment, and

θ: reversion level.
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
The price of a pure discount bond, u(r, t, T ) maturing in T can be found

by solving Eq. (A.1) under the terminal condition u(r, t, T ) = 1.
The solution can be found by constructing a grid for values of r over
some interval [rmin , rmax ] with an M evenly spaced points r1 , r2 , . . . , rM .
The distance between these points is ∆r.
If we denote by uni , the value of u at the ith grid point at time
step n, then a central weighted explicit scheme can be used to approximate
Eq. (20.51) as follows:
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 ∆t corresponds to the length of the time step between n and n + 1.

The ru term in Eq. (A.1) is sometimes denoted by ri uni . Substituting
in Eq. (A.1) and re-arranging gives:
uni (1 + r∆t) = pi,i−1 un+1 n+1

i−1 + pi,i ui + pi,i+1 un+1
i+1 (A.3)
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
following implicit numerical scheme:

1 σ 2 ∆t κ(θ − r)∆t
Ci,i−1 = − − (A.5)
2 (∆r)2 ∆r
where

σ 2 ∆t 1 σ 2 ∆t κ(θ − r)∆t
Ci,i = 1 + + ri ∆t, Ci,i+1 = − − . (A.6)
(∆r)2 2 (∆r)2 ∆r
It is well known that the implicit scheme is unconditionally stable since
there is no constraint on the Cs to be positive. The system is tridiagonal
and can be easily solved. The basic idea in the modified trinomial HW
approach can be presented as follows. When the grid point of r closest to
its expected value after another time step corresponds to the current grid
point, i.e., if the absolute value of the expected change in the interest rate
over a small interval of time is less than ∆r 2 , a trinomial lattice can be
constructed by solving the following system for the ps:
p̂1,i−1 (−∆r) + p̂i,i (0) + p̂i,i+1 (∆r) = κ(θ − r)∆t

p̂1,i−1 (−∆r)2 + p̂i,i (0)2 + p̂i,i+1 (∆r)2 = σ2 ∆t + [κ(θ − r)∆t]2
p̂i,i−1 + p̂i,i + p̂i,i+1 = 1 (A.7)
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:
p̂1,1 (0) + p̂1,2 (∆r) + p̂1,3 (2∆r) = κ(θ − r)∆t

p̂1,1 (0) + p̂1,2 (∆r)2 + p̂1,3 (2∆r)2 = σ 2 ∆t + [κ(θ − r)∆t]2
2
p̂1,1 + p̂1,2 + p̂1,3 = 1. (A.8)

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:
p̂M,M−2 (−2∆r) + p̂M,M−1 (−∆r) + p̂M,M (0) = κ(θ − r)∆t

p̂M,M−2 (−2∆r)2 + p̂M,M−1 (−∆r)2 + p̂M,M (0)2 = σ2 ∆ + κ(θ − r)∆t
p̂M,M−2 + p̂M,M−1 + p̂M,M = 1. (A.9)
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
Equation (A.12) can be re-written as:

u(rM − 2∆r) − 2u(rM − ∆r) + u(rM )
urr =
(∆r)2
−u(rM − 2∆r) − 4u(rM − ∆r) + 3u(rM )
ur = . (A.13)
2∆r
This analysis implies that an explicit method can be used as follows.
For the interior points r2 , . . . , rM−1 , the explicit scheme in Eqs. (A.3)
and (A.4) can be used. For the minimal level of r, the discrete analogs of
the spatial derivatives in Eq. (A.10) at time n + 1 is substituted into the
first PDE. This gives the following system:
un1 (1 + r1 ∆t) = p1,1 un+1

1 + p1,2 un+1
2 + p1,3 un+1
3 (A.14)
where the probabilities are given by:

1 σ2 ∆t 3k(θ − r1 )∆t
p1,1 = 1 + −
2 (∆r)2 ∆r
σ 2 ∆t 2k(θ − r1 )∆t
p1,2 = 2
+
(∆r) ∆r
2
1 σ ∆t 3k(θ − r1 )∆t
p1,3 = − . (A.15)
2 (∆r)2 ∆r
In the same way, at the maximum value of r, the discrete analogs of
Eq. (A.12) at time n+1 for the spatial derivatives is substituted in Eq. (A.1)
to obtain:
unM (1 + rM ∆t) = pM,M−2 un+1 n+1 n+1

M−2 + pM,M−1 uM−1 + pM,M uM (A.16)
where the probabilities are given by:

1 σ2 ∆t κ(θ − rM )∆t
pM,M−2 = − ,
2 (∆r)2 ∆r

1 σ2 ∆t 2κ(θ − rM )∆t
pM,M−1 = − − ,
2 (∆r)2 ∆r

1 σ 2 ∆t 3κ(θ − rM )∆t
pM,M = 1 + − . (A.17)
2 (∆r)2 ∆r
The proposed approach in Vetzal (1998) becomes identical to that of
HW when the time interval tends to reach zero. Using Taylor’s series, which
allows the construction of an implicit scheme by evaluating the derivatives
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:
c1,1 un1 + c1,2 un2 + c1,3 un3 = un+1

1 (A.18)
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:
cM,M−2 unM−2 + cM,M−1 unM−1 + cM,M unM = un+1

M (A.20)
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
function of the final grid values is almost 20 times smaller than the HW
method.
Fitting the initial term structure

Black et al. (1990), Black and Karasinski (1991), Jamshidian (1991), and
Hull and White (1993, 1994a,b, 1996) among others proposed the use of
a time-dependent parameter in the calibration of interest-rate models to a
given term structure. Their models are inappropriate for implicit schemes
or the Crank–Nicolson scheme.
European-style securities can be valued using an adjustment proposed
in Theorem 1 in Dybvig (1989).
Valuation of American-style options

Consider the following dynamics for the spot risk-free rate:
dr = κ(θ − r)dt + σrβ dB (A.21)
When β = 0, this process corresponds to Vasicek (1977) model.

When β = 1, this process corresponds to Courtadon (1982) model.
When β = 12 , this process is the Cox et al. (1985) model.
In this context, the option price must satisfy the following PDE:
1 2 2β
σ r urr + [k(θ − r) − Φ(r)σrβ ]ur + ut − ru = 0 (A.22)
2
where Φ(r) is the market price of interest-rate risk.
The pricing of contingent claims is based on the choice of a given Φ(r)
or θ as a function of time. Vetzal (1998) used the following PDE for the
pricing of contingent claims:
1 2 2β
σ r urr + [κ(θ − r) − Φ(r)σrβ ]ur + ut − [r + rb (t)]u = 0
2
un−2 = A−1 −1
n−1 Bn−1 An−1 Bn−1 u
n−1
= A−1 −1 n
n Bn An Bn u . (A.23)
In a Crank–Nicolson numerical scheme, the solution un at time n is obtained

by solving a set of linear equations of the form:
An un−1 = Bn un
where A, B are square matrices of size M .

In the same way:
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 k, α, θ, δ, γ, and ρ are parameters.

Contingent claim prices must satisfy the following PDE:
1 2 2β
[σ r urr + 2ργσrβ urv + γ 2 uvv ] + [κ(θ − r]ur
2
+ [α + δv]uv + ut − [r + rb (t)]u = 0 (A.26)
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.
Appendix B: An Algorithm for the American Call

with Dividends
For each level Si = ih of the underlying asset, the terminal boundary
condition is written as:
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)
When the asset price is zero, the option is worthless.

For j = 1 to m, C(0, j) = 0, End.
The following system corresponds to the inversion of a tri-diagonal matrix
by the Gauss method.
For j = 1 to m, a(1) = 0, b(1) = 1 + σ 2 k + rk,
c(1) = − 21 rk − 12 σ 2 k,
d(1) = C(1, j − 1),
For i = 2 to n − 1
a(i) = 12 rik − 12 σ 2 i2 k,
b(i) = 1 + σ2 i2 k + rik,
c(i) = − 21 rik − 12 σ 2 i2 k,
d(i) = C(i, j − 1)
End, a(n) = −1, b(n) = 1, c(n) = 0, d(n) = h
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))
End, C(n, j) = g(n).
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).
A special treatment is done for the determination of the critical asset price
corresponding to an optimal early exercise.
For i = 0 to n do H(i) = C(i, j) + K − ih end

k = 0, For i = 1 to n do
if h(i) ≤ 0 then C(i, j) = ih − K, k = k + 1 end if
if k = 1 then S ∗ = (i − 1)h + hH(i − 1)/H(i − 1) − H(i) end if
End (for i = 1 to n).
At a dividend date, the following treatment is done.
If j is in J2 then k = int(d/h) where int(.) corresponds to the integer
part of a number.
For i = 0 to k − 1, v(i) = 0 end for
For i = k to n if C(i − k, j) < (ih − K) then v(i) = ih − K else
v(i) = C(i − k, j)
end if end for
For i = 0 to n do C(i, j) = v(i) end for
For i = 0 to n write C(i, j) end for, end if End.
Appendix C: The Algorithm for the American Put

with Dividends
For each level ih of the underlying asset, the terminal boundary condition
is written as:
For i = 0 to n, P (i, 0) = K − ih
if K − ih < 0 then P (i, 0) = 0, End (for i = 0 to n)
When the asset price is zero, the put value is equal to the strike price.
For j = 1 to m, P (0, j) = K, End.
For each time step, the system must be solved using for example,
the Gauss method.
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)
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).
End for (i = 2 to n − 1):
a(n) = −1, b(n) = −1, c(n) = 0, d(n) = 0
For i = 1 to n do:
w(i) = c(i)/(b(i) − w(i − 1)a(i)),

g(i) = (d(i) − g(i − 1)a(i))/(b(i) − w(i − 1)a(i)),
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).
For a dividend date, the following treatment is done. If j is in J1, then

k = int(d/h) where int(.) corresponds to the integer part of a number.
For i = 0 to k − 1, v(i) = 0 end for

For i = k to n if P (i − k, j) < (K − ih) then v(i) = K − ih else v(i) =
P (i − k, j)
end if end for.
For i = 0 to n do P (i, j) = v(i) end for
For i = 0 to n write P (i, j) end for, end if end for, End.
Appendix D: The Algorithm for CBs with Call

and Put Prices
To run the program, you enter V max, the bond price P , the volatility σ,
the interest rate r, the number of months until the maturity date, nm,
the maximum number of steps for the underlying asset, nV , the number of
periods where z changes, np1, the number of periods where the call price
changes, np2, the amounts of dividends, dv, the dates of dividends, dd, the
coupon amounts, Ic, the coupon dates, dc, the call price vector, CP (k),
and the length of the call period, d2(k).
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
The procedure to be used to invert the matrix is:

c(0) d(0)
w(0) = ; g(0) = .
b(0) b(0)
For i = 1 to nv do
w(i) = c(i)/(b(i) − w(i − 1)a(i))

g(i) = (d(i) − g(i − 1)a(i))/(b(i) − w(i − 1)a(i))
End for (i = 1 to n),
W (nv) = g(nv)
For i = nv − 1 down to 0 do;

W (i) = g(i) − w(i)u(i + 1),
End for
For j = 1 to nm (time until maturity)
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,
Repeat imp2 = imp2 + 1
dt = dt + d2(imp2)until(j ≤ dt)
CPP = CP (imp2)
If CPP = 0 then for i = 1 to nvt to nvv, W (i) = zhi,
nv = nvv
nvt = nvv end for (i = 1 to nvt to nvv)

Else
nvt1 = trunc(CPP/zh) + 1
for i = nvt to nvt1, W (i) = zhi end,

nvt = nvt1, nv = nvt, end
The equation is discretized as follows:
For i = 1 to (nv − 1) do
1 1 1 1
a(i) = ri − σ2 i2 , b(i) = 1 + σ2 i2 + r, c(i) = − ri − σ 2 i2 ,
2 2 2 2
d(i) = W (i, j − 1).
End for (i = 1 to nv − 1)
The boundary condition when i = 0 is:
a(0) = 0, b(0) = 1, c(0) = 0, d(0) = 0

The boundary condition when i = nV is:
a(nV ) = −1b(nV ) = 1, c(nV ) = 0, d(nV ) = hzz.
If CPP = 0 (no call) solve the tridiagonal system (a, b, c, d, w, nv)

else
Repeat
nv = nv − 1, a(nv) = 0, b(nv) = 1, c(nv) = 0, d(nv) = CPP
solve the tri-diagonal system (a, b, c, d, w, nv)
until W (nv − 1, j) ≤ Cpp
End if (CPP = 0)
For the dividend, the following treatment is done:
j
k = trunc( dd )
if (ddk − j = 0) then:
For i = 0 to nv, uu(i) = W (i, j) End for (i = 0 to nv)
For i = 0 to nv,
rk = i − ddh , k = trunc(rk)
If k > 0, then uu(k) ≥ (k + 1 − rk)uu(k) + (rk − k)uu(k + 1)
if uu(k) ≥ (zih) then W (i, j) = uuk else W (i, j) = zih end else
W (i, j) = uu(i) end if end for end if
The treatment for the coupons is as follows:
j
k = trunc( dc )
If (dck − j) = 0
for i = 0 to nvuu(i) = W (i, j) end for
for i = 0 to nv
Ic
rk = i −
h
k = trunc(rk)
if k > 0,
then uu(k) = (k + 1 − rk)uu(k) + (rk − k)uu(k + 1)
if (uuk + Ic ≤ CCP ) or (CPP = 0) then,
W (i, j) = uu(k) + Ic else W (i, j) = CPP end
Else W (i, j) = uu(i)

end if
end for (i = 0 to nv)
End if
End for (j = 1 to nm)
End.
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?
v = eα+βτ x u(x, τ ); ∀α, β
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
under the following boundary condition:
V (0, t) = 0 whenS → ∞
lim V (S, t) = S
S→∞
and the terminal condition:
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
The partial derivative with respect to the underlying asset gives:
∂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:
∂2V ∂2v ∂2x E ∂v E ∂ 2v

= = − +
∂S 2 ∂x2 ∂S 2 S 2 ∂x S 2 ∂x2
with substitution in Eq. (20.47) we have:
2
σ 2 ∂v 1 ∂ v ∂v ∂v
− E + Eσ 2 − + (r + λs )E (r + λv )Ev(x, τ ) = 0
2 ∂τ 2 ∂x2 ∂x ∂x
⇐⇒

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.
v(x, τ ) = eαx+βτ u(x, τ )

Using the results in Eq. (20.47) gives:

∂v ∂ 2v 2λs ∂v 2
= + k+ 2 −1 − k + 2 λv v
∂τ ∂x2 σ ∂x σ
or
∂v ∂u
= βeαx+βτ u(x, τ ) + eαx+βτ
∂τ ∂τ
and
∂v ∂u
= αeαx+βτ u(x, τ ) + eαx+βτ
∂x ∂x
also
∂ 2V 2 αx+βτ αx+βτ ∂u αx+βτ ∂u
2
αx+βτ ∂ u
= α e u(x, τ ) + αe + αe + e .
∂x2 ∂x ∂x ∂x2
Hence, we have the following equation:

∂u ∂ 2u 2λs ∂u
= + 2α + k + − 1
∂τ ∂x2 σ2 ∂x

2λs 2λv
+ α2 + α k + 2 − 1 − k − β − 2 u.
σ σ
∂u
We can eliminate the term ∂x
by choosing α such that:
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
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:
u(x, τ ) = e−αx−βτ v(x, τ )
and
v(x, τ ) → 0 as x → −∞
and
u(x, τ ) → 0 as x → −∞
when (−α < 0) this means α > 0

σ2
⇐⇒ (r + λs ) <
2
when
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
u(x, 0) = e−αx−βτ v(x, 0) = e−αx v(x, 0)
and
v(x, 0) = max(ex − 1, 0)
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 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.
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.
Hull, J and A White (1994a). Numerical procedures for implementing term-

structure models I: single-factor models. Journal of Derivatives, 2, Fall, 7–16.
Hull, J and A White (1994b). Numerical procedures for implementing term-
structure models II: two-factor model. Journal of Derivatives, 2, Winter,
37–48.
Hull, J and A White (1996). Using Hull-White interest rate trees. Journal of
Derivatives, 3, 26–36.
Jamshidian, F (1991). Bond and option valuation in the Gaussian interest rate
model. Research in Finance, 9, 131–170.
Schwartz, E (1977). The valuation of warrants: implementing a new approach.
Uhrig, M and U Walter (1996). A new numerical approach for fitting the initial
yield curve. Journal of Fixed Income, 5, 82–90.
Vetzal, KR (1997). Stochastic volatility movements in short-term interest rates,
and bond option values. Journal of Banking and Finance, 21, 169–196.
Vetzal, KR (1998). An improved finite difference approach to fitting the initial
term structure. Journal of Fixed Income, 7, 62–81.

Part VIII
Exotic Derivatives
Non-standard derivatives, exotic derivatives, or second-generation options

are traded mainly in the OTC market. Several examples are provided
in this part. They include exchange options, pay-later options, options
on the minimum (the maximum), barrier options, lookback options, etc.
We present the main formulas for the valuation and the applications of
these instruments. Several numerical examples are provided to illustrate
the values of these contracts and their hedging parameters. The constraint
of volume leads to the development of the chapters. We have planned to
publish a whole book on these options.
875
876
Chapter 21
RISK MANAGEMENT: EXOTICS AND

SECOND-GENERATION OPTIONS
Chapter Outline
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
Forward-start options are options which give an answer to the

following question: how much can one pay for the opportunity to decide
after a known elapsed time in the future to get an at-the-money (ATM)
call with another time to maturity?
This opportunity is not accompanied by an additional cost.
Pay-later options are options for which the premium is paid upon
exercise. They offer some protection against sharp price movements without
tying up capital. These options are contingent options since the buyer has
the obligation to pay upon exercise, when the option is in-the-money (ITM)
regardless of the amount by which the underlying asset price exceeds the
strike price. As noted by Turnbull and Wakeman (1991), these options
reveal some hedging difficulties since the pay off is discontinuous. For
example, the delta is negative for an out-of-the-money (OTM) call when
the option is close to maturity.
Chooser options allow 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.
Rubinstein (1991) used the compound option framework to value chooser
options. The holder of a chooser has the right to decide at some future date
whether the option is a call or a put.
Compound options take the form of a call on a call, a call on a put,
a put on a call, or a put on a put. The underlying call or put may be either
a standard or an exotic option. Geske (1979) presented the formula for a
call on a call and Rubinstein (1991) generalized this result to include a put
on a call, a call on a put, and a put on a put.
An important extension of the Black–Scholes–Merton theory corre-
sponds to the compound-option valuation theory developed by Geske
(1979).
Complex choosers can be valued using the concept of an option on
an option, or a compound option. An option on an option is an option for
which the underlying asset is an option.
When the total value of a firm is given by the market value of its stocks
and bonds, Black and Scholes (1973) showed that corporate stocks in a
levered firm are regarded as a call with a strike price equal to the payment
to be made to bondholders. However, since bonds are denominated in real
terms, the payment to bondholders is uncertain, and consequently so is the
call’s strike price. Hence, it is not possible to apply in a straightforward
way the arbitrage argument for the valuation of an option with an indexed
strike price. Besides, to value such options, it is necessary to infer how an
asset which hedges against changes in the strike price should be valued.
Risk Management: Exotics and Second-Generation Options 879
The valuation by replication technique can be applied in a static

context. In fact, several financial contracts can be regarded as portfolios
of standard options. It is important to note that a standard option can be
regarded as an exotic option.
To understand this point, consider a portfolio which comprises:
• 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.
21.1. Exchange Options

The first important development along the Black–Scholes–Merton theory
was by Margrabe (1978). Margrabe derived a valuation formula for an
exchange option, i.e. the right to exchange one risky asset for another. The
Black–Scholes–Merton formula is a special case of the Margrabe formula. In
fact, if the call is ITM at expiration, the option holder exchanges risk-free
bonds for the risky asset.
Following Margrabe (1978), let S1 and S2 be the prices of two assets 1
and 2, respectively. Let cexchange (S1 , S2 , T ) be the value at time T of a
European option exchange option with a maturity date t∗ . The option gives
the holder the right to receive the difference (S1 − S2 ) when exercised, or
nothing if not exercised. Its payoff is given by:
cexchange (S1 , S2 , t∗ ) = max(0, S1 − S2 ), where T = t∗ − t.

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:
cexchange (S1 , S2 , t) = S1 N (d1 ) − S2 N (d2 )

S1 1 1 √
d1 = ln + σ 2 )T √ , d2 = d1 − σ T
S2 2 σ T
where
σ2 = σ12 + σ22 − 2ρ1,2 σ1 σ2 .
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,
cexchange (S1 , S2 , t) − cexchange (S2 , S1 , t) + S2 = S1 .
For American options, the following relationship must be satisfied.
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.
21.2. Forward-Start Options

Forward-start options are options which give an answer to the following
question: how much can one pay for the opportunity to decide after a known
time t in the future, known as the “grant date”, to obtain an ATM call with
time to matutity t with no additional cost?
Table 21.1. Simulations of exchange options values. S1 = 110, t = 11/01/2003,

T = 11/01/2004, r = 4%, σ1 = 20%, S2 = 120, σ2 = 30%, and ρ = 0.5%.
S1 Price Delta 1 Gamma 1 Delta 2 Gamma 2
105.60 24.31492 −0.56932 0.01030 0.66342 0.00837

106.70 23.69484 −0.55799 0.01024 0.66342 0.00837
107.80 23.08719 −0.54673 0.01017 0.66342 0.00837
108.90 22.49189 −0.53555 0.01010 0.66342 0.00837
110 21.90884 −0.52444 0.01001 0.66342 0.00837
111.10 21.33796 −0.51343 0.00992 0.66342 0.00837
112.20 20.77913 −0.50252 0.00983 0.66342 0.00837
113.30 20.23225 −0.49172 0.00972 0.66342 0.00837
114.40 19.69718 −0.48103 0.00962 0.66342 0.00837
Following Rubinstein (1991a), we denote by:

St : unknown underlying asset value after time t;
d: 1 plus the known payout rate and
C(S, K, T ): value of a call with a time to maturity T .
Since this option is homogeneous of degree 1 in the underlying asset
price and the strike price, the value of a forward-starting ATM call can be
written as:
C(St , St , T ) = St C(1, 1, T ).
When we account for the payout ratio, the current value of the forward
option is:
C(St , St , T ) = Sd−t C(1, 1, T ).
Since all uncertainty is resolved once the underlying asset price is

observed, after time t, then C(1, 1, T ) is known in advance and corresponds
to the current value of an ATM call. This option can be easily replicated in a
simple buy-and-hold strategy by holding C(1, 1, T ) shares from the current
time to the grant date. It is convenient to note that the above results can be
easily generalized to allow the granting of options, which are proportionally
either ITM or OTM, i.e., by introducing a constant β in the call payoff,
C(St , βSt , T ).
Consider the case where the strike is set equal to a constant α times the
asset price after a certain time t. When α = 1, the option will start at ATM.
When α < 1, the call (put) will start (1 − α) percent ITM (OTM). When
α > 1, the call (put) will start (α − 1) percent OTM (ITM). In the presence
of a cost of carry b, the forward-start call formula is:
c = Se−(b−r)t [e−(b−r)(T −t) N (d1 ) − αe−r(T −t) N (d2 )]

1
ln α + b + 12 σ 2 (T − t)
d1 = , d2 = d1 − σ (T − t).
σ (T − t)
In the same context, the put formula is:
p = Se−(b−r)t [e−(b−r)(T −t) N (d1 ) − αe−r(T −t) N (−d2 )]

1
ln α + b + 12 σ 2 (T − t)
d1 = , d2 = d1 − σ (T − t)
σ (T − t)
Table 21.2 provides some simulations of forward-start options values.
21.3. Pay-Later Options

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 ITM and to pay the premium. The exercise takes
place regardless of the importance of the difference between the underlying
asset price and the strike price, i.e., the amount by which the option
is ITM.
Table 21.2. Simulations of forward-start call values. S = 100, t = 11/01/2003,

T = 05/06/2003, r = 4%, σ = 20%, and forward-start date = 06/03/2003.
96 4.29928 0.04478 0 0.18910 0.02625

97 4.34406 0.04478 −0.00000 0.19107 0.02653
98 4.38885 0.04478 −0.00000 0.19304 0.02680
99 4.43363 0.04478 −0.00000 0.19501 0.02707
100 4.47841 0.04478 0 0.19698 0.02735
101 4.52320 0.04478 −0.00000 0.19895 0.02762
102 4.56798 0.04478 0 0.20092 0.02789
103 4.61277 0.04478 0 0.20289 0.02817
104 4.65755 0.04478 0 0.20486 0.02844
Following Turnbull (1989), we denote by:

ST : price of the underlying asset at the option’s maturity
date;
F : current forward rate;
cT : option premium paid at the option’s maturity date and
r∗ : foreign interest rate.
At the option’s maturity date, the pay-later European call’s pay off is:
cpayl (ST , 0, K) = (ST − K − cT )1ST >K .
The option pays out ST − K − cT when ST > K, otherwise it has a zero

payoff. Applying standard arbitrage arguments, the value of the pay-later
European call option is given by:
∗
cpayl (ST , T, K) = Se−r T N (d1 ) − (K + cT )e−rT N (d2 )
S
ln K + r − r∗ + 12 σ 2 T √
d1 = √ , d2 = d1 − σ T .
σ T
The value of the pay-later European put option is given by:
∗
ppayl (S, T, K) = −Se−r T N (−d1 ) + (K − pT )e−rT N (−d2 )
S
ln K + r − r∗ + 12 σ 2 T √
d1 = √ , d2 = d1 − σ T
σ T
Tables 21.3 and 21.4 provide some simulations of pay-later options values.
The amount to be paid is denoted by A.
Table 21.3. Pay-later call values. S = 100, K = 100, t = 05/01/2003,

T = 05/01/2004, r = 3%, σ = 20%, and A = 10.
96 2.93255 0.32979 0.02153 0.39208 0.01300

97 3.26276 0.35132 0.02169 0.40454 0.01351
98 3.61441 0.37301 0.02180 0.41593 0.01400
99 3.98763 0.39481 0.02183 0.42620 0.01445
100 4.38255 0.41664 0.02179 0.43530 0.01488
101 4.79919 0.43843 0.02168 0.44319 0.01528
102 5.23751 0.46011 0.02152 0.44986 0.01564
103 5.69740 0.48163 0.02130 0.45530 0.01598
104 6.17871 0.50292 0.02103 0.45950 0.01627
Table 21.4. Pay-later put values. S = 100, K = 100, t = 05/01/2003,

T = 05/01/2004, r = 3%, σ = 20%, and A = 10.
96 13.67278 −0.67021 0.02153 0.39208 0.00422

97 13.00300 −0.64868 0.02169 0.40454 0.00473
98 12.35464 −0.62699 0.02180 0.41593 0.00522
99 11.72787 −0.60519 0.02183 0.42620 0.00568
100 11.12279 −0.58336 0.02179 0.43530 0.00611
101 10.53943 −0.56157 0.02168 0.44319 0.00651
102 9.97775 −0.53989 0.02152 0.44986 0.00687
103 9.43764 −0.51837 0.02130 0.45530 0.00720
104 8.91895 −0.49708 0.02103 0.45950 0.00750
21.4. Simple Chooser Options

A chooser is a contingent claim that allows its holder at a certain date,
known as 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. Hence,
the chooser is neither a call nor a put. Chooser options allow the holder,
immediately after a pre-determined elapsed time, to choose whether the
option is to be either a call or a put. This is the principal idea on which it
is based on a standard chooser. The payoff of the standard chooser is:
csimple = max[C ∗ (K, T − t), P ∗ (K, T − t); t],
where (T − t) is the time to maturity and t < T .

Since the buyer has the right to choose between a call and a put before
the chooser’s maturity date, the value of the chooser must lie between the
value of a standard option and a straddle. Following Rubinstein (1991), we
use the notation:
S ∗ : the unknown value of the underlying asset after the elapsed time t;
R: 1 plus the risk-less interest rate and
d: 1 plus the pay out rate.
The value of a standard chooser is:
√
Cs = Sd−T N (x) − KR −T N (x − σ T ) − Sd−T N (−y)
√
+ R−T KN (−y + σ t)
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.
21.5. Complex Choosers

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 complex chooser implies the choice at a
future date t, known as the “choice date”, a call or a put with a strike price
K1 or K2 and a time to maturity (T1 − t) or (T2 − t). In this spirit, the
complex chooser cannot be assimilated to a package of standard options
and is identified to a compound option. The payoff of a complex chooser is
Table 21.5. Simple chooser. S = 100, K = 100, t = 11/01/2003, T = 11/01/2004,

r = 4%, σ = 20%, and choose date = 09/06/2003.
96 12.56897 0.08711 0.05233 0.62680 0.00971

97 12.65713 0.13944 0.05119 0.62893 0.01031
98 12.79734 0.19063 0.04982 0.62848 0.01087
99 12.98848 0.24045 0.04826 0.62554 0.01140
100 13.22917 0.28870 0.04655 0.62024 0.01189
101 13.51787 0.33525 0.04472 0.61272 0.01235
102 13.85290 0.37997 0.04279 0.60316 0.01277
103 14.23244 0.42276 0.04079 0.59174 0.01314
104 14.65458 0.46355 0.03875 0.57866 0.01348
given by:
ccomplex−chooser = max[C ∗ (K1 , T1 ), P ∗ (K2 , T2 ); t]
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
where Scr is solution to the following equation:

Scr d−(T1 −t) N (z1 ) − K1 r−T1 −t N (z1 − σ (T1 − t)) − Scr d−(T2 −t) N (−z2 )

+ K2 r−(T2 −t) N (−z2 + σ (T2 − t)) = 0
and where z1 and z2 are given by:

1 Scr d−(Ti −t) 1 2
zi = √ ln + σ (Ti − t)
σ Ti − t Ki R−(Ti −t) 2
where N (a, b, ρ) is the bi-variate normal distribution function.

Table 21.6 provides simulation results for complex choosers and the
corresponding Greek letters.
21.6. Compound Options

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.
Following Geske (1979), consider a levered firm for which the debt
corresponds to pure discount bonds maturing in T years with a face value
K1 . Under the standard assumptions of liquidating the firm in T years,
Table 21.6. Complex chooser values. S = 100, t = 11/01/2003, r = 4%,

σ = 20%, call strike = 100, and put strike = 70, call maturity = 11/01/2004,
put maturity = 09/02/2004, and choose date = 11/04/2000.
96 8.56897 0.45984 0.02829 0.48553 0.01192

97 8.65713 0.48814 0.02919 0.48073 0.01252
98 9.79734 0.51732 0.02922 0.47348 0.01311
99 9.98848 0.54654 0.02858 0.46429 0.01369
100 10.22917 0.57512 0.02748 0.45367 0.01421
101 10.51787 0.60260 0.02607 0.44210 0.01468
102 11.85290 0.62867 0.02451 0.42999 0.01509
103 12.23244 0.65318 0.02291 0.41765 0.01543
104 12.65458 0.67609 0.02134 0.40530 0.01570
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.
21.6.1. The call on a call in the presence of a cost of carry

The payoff from a call on a call is:
Ccall = max[cB−S (V, K1 , T2 ) − K2 ],
where cB−S (V, K1 , T2 ) is the Black–Scholes formula with a strike K1 and a

time to maturity T2 . K1 indicates the strike price of the underlying option
and K2 is the strike price for the option on the option. Following Geske
(1979), the formula for a call on a call is:

t1 −rT2 t1
Ccall = V e (b−r)T2
N z1 , y1 , − K1 e N z2 , y2 ,
T2 T2
− K2 e−rt1 N (y2 )
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
where N (A, B, ρ) stands for the bi-variate cumulative normal distribution.

In this analysis, t1 is the time to maturity on the option and T2 is the time
to maturity on the underlying option.
21.6.2. The put on a call in the presence of a cost of carry

The formula for a put on a call corresponds to the following payoff:
Pcall = max[K2 − cB−S (V, K1 , T2 )]
The solution for a put on a call is given by:

−rT2 t1 t1
Pcall = −K1 e N z2 , −y2 , − −Ve (b−r)T2
N z1 , −y1 , −
T2 T2
+ K2 e−rt1 N (−y2 )
where the value Vcr is determined using the following equation:
cB−S (Vcr , K1 , T2 − T1 ) = K2 .
21.6.3. The formula for a call on a put in the presence

of a cost of carry
The formula for a call on a put corresponds to the following payoff:
cput = max[pB−S (V, K1 , T2 ) − K2 , 0].
The solution for a call on a put is given by:

−rT2 t1 t1
cput = K1 e N −z2 , −y2 , −Ve (b−r)T2
N −z1 , −y1 ,
T2 T2
− K2 e−rt1 N (−y2 ).
21.6.4. The put on a put in the presence of a cost of carry

The formula for a put on a put corresponds to the following payoff:
Pput = max[K2 − pB−S (V, K1 , T2 ), 0].
The solution for a put on a put is given by:

t1 −rT2 t1
Pput = V e (b−r)T2
N −z1 , y1 , − − K1 e N −z2 , y2 , −
T2 T2
+ K2 e−rt1 N (y2 )
where the value Vcr is determined using the following equation:
PB−S (Vcr , K1 , T2 − t1 ) = K2 .
Tables 21.7 to 21.10 provide simulations of compound option values and

the corresponding Greek letters using different parameters.
21.7. Options on the Maximum (Minimum)

Stulz (1982) and Johnson and Shanno (1987) derived valuation formulas
for options on the maximum and the minimum of two or more risky assets.
It is important to note that several exchange-traded futures contracts can
be valued using the formula for the option on the minimum.
Table 21.7. Simulations of the prices of a call on a call. S = 100,

K1 = 100, t = 11/01/2003, T = 11/01/2004, r = 4%, σ = 20%, compound
date = 10/03/2003, and K2 = 3.
96 2.31715 0.48657 0.05833 0.23593 0.00354

97 2.80488 0.54490 0.05687 0.25010 0.00420
98 3.35064 0.60177 0.05418 0.26101 0.00492
99 3.95296 0.65595 0.05039 0.26847 0.00568
100 4.60917 0.70634 0.04573 0.27253 0.00647
101 5.31551 0.75207 0.04045 0.27339 0.00728
102 6.06738 0.79252 0.03480 0.27143 0.00810
103 6.85953 0.82732 0.02905 0.26718 0.00891
104 7.68638 0.85637 0.02344 0.26120 0.00970
Table 21.8. Simulations of the prices of a call on a put. S = 100, K1 = 100,

t = 11/01/2003, T = 11/01/2004, r = 4%, σ = 20%, compound date = 10/03/2003,
and K2 = 3.
96 4.74495 −0.45490 0.02438 0.32804 0.00364

97 4.29053 −0.43052 0.02506 0.32180 0.00364
98 3.86038 −0.40546 0.02566 0.31401 0.00358
99 3.45517 −0.37980 0.02613 0.30471 0.00349
100 3.07550 −0.35367 0.02642 0.29397 0.00336
101 2.72183 −0.32725 0.02649 0.28190 0.00319
102 2.39444 −0.30076 0.02632 0.26861 0.00300
103 2.09342 −0.27444 0.02588 0.25425 0.00278
104 1.81859 −0.24856 0.02518 0.23901 0.00256
Table 21.9. Simulations of the prices of a put on a put. S = 100, K1 = 100,

and K2 = 15.
96 6.73448 0.42139 −0.01279 −0.33921 −0.00631

97 7.15562 0.40859 −0.01413 −0.34947 −0.00657
98 7.56401 0.39446 −0.01510 −0.35728 −0.00679
99 7.95832 0.37936 −0.01575 −0.36279 −0.00699
100 8.33760 0.36361 −0.01612 −0.36621 −0.00716
101 8.70122 0.34749 −0.01627 −0.36771 −0.00731
102 9.04879 0.33123 −0.01623 −0.36752 −0.00742
103 9.38018 0.31500 −0.01604 −0.36582 −0.00751
104 9.69543 0.29896 −0.01574 −0.36280 −0.00757
Table 21.10. Simulations of the prices of a put on a call. S = 100, K1 = 100,

and K2 = 15.
96 0.75775 −0.17692 0.03620 0.11211 −0.00062

97 0.58159 −0.14072 0.03033 0.09928 −0.00048
98 0.44135 −0.11039 0.02494 0.08602 −0.00037
99 0.33123 −0.08545 0.02016 0.07303 −0.00028
100 0.24589 −0.06529 0.01603 0.06086 −0.00021
101 0.18060 −0.04926 0.01254 0.04983 −0.00015
102 0.13128 −0.03672 0.00967 0.04013 −0.00011
103 0.09446 −0.02704 0.00735 0.03181 −0.00008
104 0.06729 −0.01969 0.00551 0.02485 −0.00006
21.7.1. The call on the minimum of two assets

Following Stulz (1982), let S1 and S2 stand, respectively for the prices of
two risky assets. At maturity, the payoff of a European call on the minimum
of these two assets is
cmin = max[min(S1 , S2 ) − K, 0].
The price of a European call on the minimum of S1 and S2 , with a

maturity date T and a strike price K, denoted by cmin (S1 , S2 , K, T − t) is
equal to the value of a self-financing portfolio which has the same value as
the option at date T .
Stulz (1982) provided the appropriate formulas for the valuation of
these options. In the presence of a cost of carry b1 for asset 1 and b2 for
asset 2, the formula for the pricing of a call on the minimum of two assets is:
cmin (S1 , S2 , K, τ ) = S1 e(b1 −r)τ N (y1 , −d, −ρ1 )

√
+ S2 e(b2 −r)τ N (y2 , d − σ τ , −ρ2 )
√ √
− Ke−rτ N (y1 − σ1 τ , y2 − σ2 τ , ρ)
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
and where N (α, β, ρ) is the bi-variate cumulative normal distribution, where

α, β are the upper limits of integration and ρ is the correlation coefficient.
We can write in a compact form, the price of a European call on the
minimum of two assets as:
cmin (S1 , S2 , K, τ ) = S1 e(b1 −r)τ N (β1 , β2 , ρc ) + S2 e(b2 −r)τ N (α1 , α2 , ρc )

− Ke−rτ N (γ1 , γ2 , ρ12 ).
If the strike is zero and there is no cost of carry, the formula for a call on
the minimum reduces to:
cmin (S1 , S2 , 0, τ ) = S1 − cE (S1 , S2 , 1, τ ) = S1 − N (d11 ) + S2 N (d22 )

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).
21.7.2. The call on the maximum of two assets

As for ordinary options, it is possible to obtain some parity relationships
between options on the minimum, the maximum, the underlying asset, and
the interest rate. At maturity, the payoff of a European call on the maximum
of two assets is:
cmax (S1 , S2 , K, τ ) = max[max(S1 , S2 ) − K, 0].
The value of this option is given by:
cmax (S1 , S2 , K, τ ) = C(S1 , K, τ ) − cmin (S1 , S2 , 0, τ ) + C(S2 , K, τ )
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)].
21.7.3. The put on the minimum (maximum) of two assets

Let pmin (S1 , S2 , K, τ ) be the price of a European put on the minimum of
two assets S1 and S2 . Its price must satisfy the following relationship:
pmin (S1 , S2 , K, τ ) = e−rτ K − cmin (S1 , S2 , 0, τ ) + cmin (S1 , S2 , K, τ ).
The value of the put on the maximum, pmax (S1 , S2 , K, τ ) is given by Stulz
(1982) as:
pmax (S1 , S2 , K, τ ) = e−rτ K − cmax (S1 , S2 , 0, τ ) + cmax (S1 , S2 , K, τ ).

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%.
S1 Price Delta1 Gamma 1 Delta 2 Gamma 2
96 2.58960 0.16282 0.00395 0.11297 0.00033

97 2.75440 0.16677 0.00365 0.11297 0.00033
98 2.92299 0.17041 0.00336 0.11297 0.00033
99 3.09510 0.17378 0.00305 0.11297 0.00033
100 3.27041 0.17683 0.00273 0.11297 0.00033
101 3.44862 0.17956 0.00241 0.11297 0.00033
102 3.62939 0.18196 0.00209 0.11297 0.00033
103 3.81242 0.18405 0.00177 0.11297 0.00033
104 3.99737 0.18583 0.00146 0.11297 0.00033
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%.
S1 Price Delta 1 Gamma 1 Delta 2 Gamma 2
96 12.14544 −0.47349 0.01145 −0.42858 0.01083

97 11.67764 −0.46203 0.01131 −0.42858 0.01083
98 11.22122 −0.45072 0.01115 −0.42858 0.01083
99 10.77603 −0.43957 0.01099 −0.42858 0.01083
100 10.34191 −0.42858 0.01083 −0.42858 0.01083
101 9.91871 −0.41775 0.01066 −0.42858 0.01083
102 9.50625 −0.40709 0.01049 −0.42858 0.01083
103 9.10437 −0.39660 0.01032 −0.42858 0.01083
104 8.71290 −0.38628 0.01014 −0.42858 0.01083
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.
21.8. Extendible Options

21.8.1. The valuation context
The valuation of these options is realized in the Black–Scholes context.
Following Longstaff (1990), the valuation equation that must be
satisfied by the option price, V (S, t) is:
1 2 2 ∂2V ∂V ∂V
σ S 2 + rS − rV + = 0.
2 ∂ S ∂S ∂t
Let CE (S, K1 , T1 , K2 , T2 , A) be the current value of an extendible call as

a function of the two strike prices K1 and K2 , times to maturity T1 and
T2 , the underlying asset S, and the premium A to be paid in the event of
extension.
At date T1 , the call’s payoff is:
CE (S, K1 , T1 , K2 , T2 , A) = max(0, C(S, K2 , T2 − T1 ) − A, S − K1 )
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:
CE = max{max[0, C(S, K2 , T2 − T1 ) − A], max[0, S − K1 ]}.
This payoff function corresponds to a maximum of two risky payoffs:

the payoff of a standard call and that of a call on a call. The payoff looks
like that of an option on the maximum of two assets.
When A is positive, there is some critical value of S at T1 denoted by
I1 below which the option is not extended, and another critical value I2
above which the option is again not extended. Hence, an extension occurs
when S is in the interval [I1 , I2 ].
At T1 , the value of I1 is a solution to the equation: C(I1 , K2 , T2 − T1 ) =
A and I1 must lie between A and A + K2 e−r(T2 −T1 ) .
When A = 0, I1 = 0 and when I1 ≥ K1 , the extension privilege is
worthless.
A sufficient condition for I1 to be less than K1 is:
A < K2 − K2 e−r(T2 −T1 ) .
At T1 , the value of I2 is given by the solution to the equation:
C(I2 , K2 , T2 − T1 ) = I2 − K1 + A.
21.8.2. Extendible calls

The value of the extendible call as given by Longstaff is:
CE (S, K1 , T1 , K2 , T2 , A)
= C(S, K1 , T1 ) + SN2 (γ1 , γ2 , −∞, γ3 , ρ)

− 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 )
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.
21.9. Equity-Linked Foreign Exchange Options

and Quantos
As shown in Garman and Kohlhagen (1983), the Black and Scholes (1973)
formula for stock options applies to the valuation of options on currencies
where the foreign interest rate replaces the dividend yield. When an investor
wants to link a strategy in a foreign stock and a currency, he/she can use
at least four different types of options: a foreign equity option struck in
foreign currency, a foreign equity option struck in domestic currency, fixed
exchange-rate foreign equity options also known as quanto options, or an
equity-linked foreign exchange option. These different types of options are
analyzed and valued in this section.
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.
96 8.22621 0.54988 0.01885 0.50571 0.01001

97 8.77647 0.56873 0.01851 0.50771 0.01024
98 9.34548 0.58724 0.01841 0.50854 0.01046
99 9.93290 0.60537 0.01774 0.50823 0.01066
100 10.53836 0.62311 0.01732 0.50681 0.01086
101 11.16147 0.64043 0.01687 0.50434 0.01104
102 11.80182 0.65731 0.01640 0.50083 0.01120
103 12.45896 0.67370 0.01592 0.49638 0.01136
104 13.13242 0.68963 0.01543 0.49102 0.01150
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.
80 1.49102 0.18434 0.01674 0.24246 −0.00473

85 2.63070 0.27411 0.01951 0.32142 −0.00628
90 4.24971 0.37487 0.02080 0.38670 −0.00757
95 6.38402 0.47896 0.02056 0.42850 −0.00839
100 9.03184 0.57920 0.01905 0.44261 −0.00867
105 12.15924 0.67006 0.01673 0.43046 −0.00844
110 15.71041 0.74831 0.01400 0.39742 −0.00779
115 19.61847 0.81275 0.01125 0.35076 −0.00687
120 23.81495 0.86383 0.00873 0.29766 −0.00583
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.
88 7.61027 0.43251 0.01370 0.43886 −0.00755

93 10.19395 0.50656 0.01304 0.47311 −0.00810
99 13.17356 0.57618 0.01206 0.49186 −0.00841
104 16.52056 0.63997 0.01090 0.49591 −0.00849
110 20.20043 0.69713 0.00965 0.48709 −0.00837
115.50 24.17575 0.74740 0.00841 0.46784 −0.00808
121 28.40893 0.79094 0.00722 0.44074 −0.00767
126.50 32.86391 0.82812 0.00612 0.40832 −0.00716
132 37.50725 0.85953 0.00514 0.37282 −0.00660
21.9.1. The foreign equity call struck in foreign currency

The payoff of a foreign equity call struck in foreign currency is:
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
where σS is the volatility of S . 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 .
21.9.2. The foreign equity call struck in domestic currency

The payoff of a foreign equity option struck in domestic currency is
C2∗ = max[S ∗ X ∗ − K, 0],
where K is the domestic currency amount.

For the foreign option writer, the payoff is

C2∗ = max[S ∗ − KX ∗ , 0], where X = 1/X.
X corresponds to the exchange rate quoted at the price of a unit of

domestic currency in terms of the foreign currency. This pay-off corresponds
to that of an option to exchange one asset (K units of our currency) for
another asset (a share of stock). Its value is:
√
C2∗ = S (d)−T N (d2 ) − KX (r)−T N (d2 − σS X T );
−T
S (d) √ 1 √
d2 = log −T
σS X T + σS X T , and
KX (r) 2

σ(S X ) = σS2 + σX
2 − 2ρ σ σ
S X S X
where ρS X is the correlation coefficient between the rates of return on S

and X .
The domestic value of this option in the same context is:
√
C2 = S X(d)−T N (d2 ) − K(r)−T N (d2 − σS X T )

S X(d)−T √ 1 √
d2 = log −T
σS X T + σS X T
K(r) 2

σ(S X) = σS2 + σX
2 − 2ρ σ σ = σ .
S X S X (S X )
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 .
21.9.3. Fixed exchange rate foreign equity call

The payoff of a fixed exchange rate foreign equity call, known as a quanto is:
C3∗ = X̄ max[S ∗ − K , 0] = max[S ∗ X̄ − K, 0]
where X̄ is the rate at which the conversion will be made. It can be written
in reciprocal units as
C3∗ = X̄X ∗ max[S ∗ − K , 0].
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
The domestic value of this option is:

−T √
rd
C3 = X̄ S e−(ρS X σS σX )T N (d3 )a − K (r)−T N (d3 − σS T )
r∗

S (d)−T √ 1 √
d3 = log − ρS X σS σX 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 .
21.9.4. An equity-linked foreign exchange call

The payoff of an equity-linked foreign exchange call is:
C4∗ = S ∗ max[X ∗ − K, 0].
It can also be written as:

1
C4∗ = S ∗ max[1 − KX ∗ − K, 0] = KS ∗ max − X ∗ , 0 .
K
The foreign value of this call option is given by:

−T
rd √
C4∗
= S (d) −T
N (d4 ) − KS X
e−(ρS X σS σX )T N (d4 − σX T )
r∗

X(r∗ )−T √ 1 √
d4 = log + ρS X σS σX T σX T + σX T .
K(r)−T 2
The domestic value of this call option is:

−T
rd √
C4 = S X(d)−T N (d4 ) − K S ∗ e−(ρS X σS σX )T N (d4 − σX T ).
r
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.
Type Asset Strike Rate Distribution σ
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
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.
21.10. Binary Barrier Options

Following Rubinstein and Reiner (1991), we denote by:
• R: 1 plus the risk-less interest rate r;

• d: 1 plus the instantaneous payout rate;
• H: barrier level;
• S(τ ): price of the underlying asset after elapsed time τ ;
• St : price of the underlying asset at expiration t;
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.
80 4.31208 0.32902 0.01598 0.30619 −0.01255

85 6.15817 0.40989 0.01633 0.35417 −0.01453
90 8.41102 0.49121 0.01606 0.39151 −0.01607
95 11.06549 0.57012 0.01528 0.41644 −0.01710
100 14.10374 0.64441 0.01416 0.42862 −0.01760
105 17.49871 0.71261 0.01282 0.42886 −0.01761
110 21.21759 0.77387 0.01138 0.41875 −0.01719
115 25.22464 0.82786 0.00993 0.40031 −0.01642
120 29.48371 0.87471 0.00854 0.37569 −0.01540
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.
80 3.92008 0.29911 0.01453 0.27835 −0.01141

85 5.59834 0.37263 0.01484 0.32197 −0.01321
90 7.64638 0.44655 0.01460 0.35591 −0.01461
95 10.05954 0.51829 0.01389 0.37858 −0.01555
100 12.82158 0.58583 0.01287 0.38965 −0.01600
105 15.90791 0.64783 0.01166 0.38987 −0.01601
110 19.28872 0.70352 0.01035 0.38068 −0.01563
115 22.93149 0.75260 0.00903 0.36392 −0.01493
120 26.80338 0.79519 0.00777 0.34154 −0.01400
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%.
80 3.72407 0.28415 0.01380 0.26444 −0.01084

85 5.31842 0.35400 0.01410 0.30587 −0.01255
90 7.26406 0.42422 0.01387 0.33812 −0.01388
95 9.55656 0.49238 0.01320 0.35965 −0.01477
100 12.18051 0.55654 0.01223 0.37017 −0.01520
105 15.11252 0.61544 0.01108 0.37038 −0.01521
110 18.32428 0.66834 0.00983 0.36165 −0.01484
115 21.78492 0.71497 0.00858 0.34573 −0.01418
120 25.46321 0.75543 0.00738 0.32446 −0.01330
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%.
80 20.84295 −0.66585 0.01380 0.26444 −0.01084

85 17.68730 −0.59600 0.01410 0.30587 −0.01255
90 14.88293 −0.52578 0.01387 0.33812 −0.01388
95 12.42544 −0.45762 0.01320 0.35965 −0.01477
100 10.29938 −0.39346 0.01223 0.37017 −0.01520
105 8.48139 −0.33456 0.01108 0.37038 −0.01521
110 6.94315 −0.28166 0.00983 0.36165 −0.01484
115 5.65379 −0.23503 0.00858 0.34573 −0.01418
120 4.58208 −0.19457 0.00738 0.32446 −0.01330
• η and φ: binary variables taking the value 1 or −1, and

• K: the strike price.
21.10.1. Path-independent binary options

21.10.1.1. Standard cash-or-nothing options
In their simplest forms, the payoff of a binary call is nothing when the
underlying asset terminal price, St , is below the strike price, K and is a
predetermined amount, A, if the underlying asset terminal price is above
the strike price. The payoff of the binary call is:
ccon = 0 if St ≤ K
ccon = A else.
The payoff of a binary put is nothing when the underlying asset

terminal price, St , is above the strike price, K and is a pre-determined
amount, A, if the underlying asset terminal price is below the strike
price.
The payoff of the binary put is:
pcon = 0 if St ≥ K
pcon = A if else.
These options can be valued with respect to classic formulas by Black–

Scholes. The values of standard options are given by:
√
C = φSd−t N (φx) − φKR −t N (φx − φσ t),

Sd−t
log KR −t 1 √
x= √ + σ t
σ t 2
with φ = 1 for a call and −1 for a put.
The Black–Scholes formula comprises two parts. The first term,
d−t φSN (φx), corresponds to the present value of the underlying asset price
conditional
√ upon exercising the option. The second term, φKR −t N (φx −
φσ t), refers to the present value of the strike price times the probability
√ option. The value of a cash-or-nothing call is ccon =
of exercising the
AR −t N (x − σ t). The value of a cash-or-nothing put is:
√
pcon = AR −t N (−x + σ t).
In the presence of a cost of carry b, the value of a cash-or-nothing call

is ccon = Ae−rT N (x ).
The value of a cash-or-nothing put is:
pcon = Ae−rT N (−x )
where
S

log K + b − 12 σ 2 T
x = √ .
σ T
21.10.1.2. Cash-or-nothing options with shadow costs

We denote by λS and λ, the information costs associated with S and
the option. The values of standard options with information costs are
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 a cash-or-nothing put is:

√
pcon = Ae(λS −λ)t N (−x + σ t).
21.10.1.3. Standard asset-or-nothing options

These binary options are similar to cash-or-nothing options except that
in their payoff, the pre-determined amount is replaced by the terminal
asset value. The payoff of the asset-or-nothing call is caon = 0, if
St ≤ K caon = St else.
The value of this option is given by the present value of the underlying
asset price conditional upon exercising the call or: caon = Sd−t N (x).
The payoff of the asset-or-nothing put is nothing when the underlying
asset terminal price, St , is above the strike price, and is the terminal asset
price, St if the underlying asset terminal price is below the strike price
paon = 0, if St ≥ K paon = St else.
The value of this option is given by the present value of the underlying
asset price conditional upon exercising the put or: paon = Sd−t N (−x).
In the presence of a cost of carry b, the value of an asset-or-nothing call
is caon = Se(b−r)t N (x ).
The value of an asset-or-nothing put is: paon = Se(b−r)t N (−x ), where
S

log K + b + 12 σ 2 t
x = √ .
σ t
21.10.1.4. Asset-or-nothing options with shadow costs

The value of the asset-or-nothing call is given by the present value of the
underlying asset price conditional upon exercising the call or:
√
caon = S0 e(λS −λ+d)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:
paon = S0 e(λS −λ−d)t N (−x).
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.
Table 21.21. Cash or nothing options. S = 100, t = 17/12/2003, T = 18/12/2004,

r = 2%, σ = 20%, and barrier = 140.
Barrier Price Delta Gamma Vega Theta
100 0.49005 0.01950 −0.00002 −0.00390 0.00011

101 0.47065 0.01947 0.00002 −0.00294 0.00008
102 0.45150 0.01940 0.00007 −0.00198 0.00005
103 0.43262 0.01929 0.00012 −0.00104 0.00003
104 0.41406 0.01913 0.00016 −0.00011 0.00000
105 0.39585 0.01893 0.00021 0.00078 −0.00002
110 0.31099 0.01742 0.00039 0.00474 −0.00013
120 0.17803 0.01290 0.00058 0.00909 −0.00025
130 0.09350 0.00829 0.00054 0.00917 −0.00025
140 0.04577 0.00477 0.00041 0.00708 −0.00019
Table 21.22. Cash or nothing options. S = 100, t = 05/01/2003, T = 05/01/2004,

r = 3%, σ = 20%, barrier = 90.
96 0.62605 0.01851 −0.00059 −0.01027 −0.00019

97 0.64456 0.01793 −0.00061 −0.01097 −0.00021
98 0.66247 0.01732 −0.00063 −0.01160 −0.00023
99 0.67978 0.01669 −0.00064 −0.01216 −0.00025
100 0.69647 0.01605 −0.00065 −0.01265 −0.00027
101 0.71253 0.01541 −0.00065 −0.01307 −0.00029
102 0.72794 0.01476 −0.00065 −0.01341 −0.00030
103 0.74271 0.01411 −0.00065 −0.01369 −0.00032
104 0.75683 0.01346 −0.00064 −0.01390 −0.00033
Table 21.23. Asset or nothing options. S = 100, t = 05/01/2003, T = 05/01/2004,

r = 3%, σ = 20%, and barrier = 110.
96 32.04676 2.17828 0.05551 1.09187 0.04444

97 34.22618 2.23380 0.05098 1.03383 0.04330
98 36.46076 2.28478 0.04629 0.96919 0.04195
99 38.74601 2.33107 0.04150 0.89846 0.04039
100 41.07730 2.37257 0.03664 0.82219 0.03864
101 43.44986 2.40920 0.03175 0.74100 0.03672
102 45.85890 2.44095 0.02686 0.65550 0.03465
103 48.29956 2.46781 0.02202 0.56637 0.03243
104 50.76702 2.48984 0.01724 0.47427 0.03009
Table 21.24. Asset or nothing options. S = 100, t = 17/12/2003, T = 18/12/2004,

r = 2%, σ = 20%, and barrier = 140.
Barrier Price Delta Gamma Vega Theta
100 57.94737 2.52913 0.02286 0.00194 −0.00001

101 55.99851 2.52674 0.02768 0.09883 −0.00267
102 54.05440 2.51958 0.03248 0.19597 −0.00534
103 52.11964 2.50773 0.03723 0.29265 −0.00799
104 50.19862 2.49134 0.04189 0.38822 −0.01062
105 48.29551 2.47053 0.04646 0.48204 −0.01319
110 39.17721 2.30761 0.06657 0.90634 −0.02483
120 23.92946 1.78728 0.08763 0.40286 −0.03839
130 13.39969 1.21145 0.08329 0.41011 −0.03846
140 6.98079 0.73818 0.06478 0.12724 −0.03062
150 3.42846 0.41329 0.04380 0.77475 −0.02094
In the presence of a cost of carry, equal to the difference between the

domestic interest rate r and the foreign interest rate, r∗ , these options can
be priced in the same way. Table 21.25 gives the values of these options and
the associated Greek letters.
21.10.1.5. Standard gap options

Gap options are structured to give the following payoff for a call:
cgap = 0 if St ≤ K
cgap = St − A else.
Table 21.25. Digital FX-cash or nothing options. S = 1, t = 07/02/2003,

T = 07/02/2004, r = 3%, r ∗ = 4%, σ = 20%, and barrier = 1.5.
0.96 0.00843 0.12559 0.34624 0.00253 0.00843

0.97 0.00969 0.14020 0.31487 0.00279 0.00969
0.98 0.01110 0.15593 0.28279 0.00306 0.01110
0.99 0.01266 0.17278 0.24997 0.00334 0.01266
1 0.01438 0.19078 0.21641 0.00363 0.01438
1.01 0.01629 0.20992 0.18211 0.00393 0.01629
1.02 0.01839 0.23020 0.14707 0.00425 0.01839
1.03 0.02069 0.25162 0.11130 0.00457 0.02069
1.04 0.02320 0.27415 0.07482 0.00489 0.02320
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
formula is:
cgap = Se(b−r)tN (d1 ) − K2 e−rt N (d2 )

log KS1 + b + 12 σ 2 t √
d1 = √ , d2 = d1 − σ t.
σ t
The put formula is:
pgap = K2 e−rtN (−d2 ) − Se(b−r)tN (−d1 )

log KS1 + b + 12 σ 2 t √
d1 = √ , d2 = d1 − σ t.
σ t
21.10.1.6. Gap options with shadow costs

The value of a gap call is given by:
√
cgap = e(λS −λ−d)t SN (x) − Ae−(λ+r)t N (x − σ t).
The value of the gap put is:

√
pgap = −e(λS −λ−d)t SN (−x) + Ae−(λ+r)t N (−x + σ t).
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
21.11. Lookback Options

We use the following notations for the maximum and the minimum over
the interval [t1 , t2 ]:
Mtt12 = max{Ss /s ∈ [t1 , t2 ]}

mtt21 = min{Ss /s ∈ [t1 , t2 ]}.
Option prices are computed at time 0 and option contracts are assumed to
have been initiated at time T0 ≤ 0.
21.11.1. Standard lookback options

Since the payoff of a standard European lookback call at the maturity date
is given by ST − mTT0 , its current value is:
√
C = S0 N (d ) − e−rT m0T0 N (d − σ T )
2 −( 2r2 )
σ S σ 2r √
+ e−rT N −d + T − erT N (−d )
0
S0
2r m0T0 σ
with

1 S0 σ2
d = √ ln + rT + T .
σ T m0T0 2
Since the payoff of a standard lookback put is (MT00 − ST ), its current

value is:
√
P = −S0 N (−d ) + e−rT MT00 N (−d + σ T )
2 −( 2r2 )

σ S σ √
+ e−rT N d − (2r/σ) T + erT N (d )
0
S0 −
2r MT00
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.
21.11.2. Options on extrema

21.11.2.1. On the maximum
The payoff of a call on the maximum at maturity T is (MT00 − K)+ . When
K ≥ MT00 , the current call price is:
√
CM = S0 N (d) − e−rT KN (d − σ T )
2 −( 2r2 )
−rT σ S0 σ 2r √ rT
+e S0 − N d− T + e N (d)
2r K σ

1 S0 σ2
d= √ ln + rT + T .
σ T K 2
When K < MT00 , the call’s value is:

√
CM = e−rT (MT00 − K) + S0 N (d ) − e−rT MT00 N (d − σ T )
2 −( 2r2 )
−rT σ S0 σ
2r √ rT
+e S0 − N d − T + e N (d )
2r MT00 σ
21.11.2.2. On the minimum

The payoff of a put on the minimum at the maturity date T is (K − m0T0 )+ .
When K < m0T0 , the put’s price is:
√
p = −S0 N (−d) + e−rT KN (−d + σ T )
2 −( 2r2 )
σ S σ 2r √
+ e−(rT T − erT N (−d)
0
S0 N −d +
2r K σ
When K ≥ m0T0 , the put’s value is:

√
p = e−rT (K − m0T0 ) − S0 N (−d ) + e−rT m0T0 N (−d + σ T )
2 −( 2r2 )
−rT σ S0 σ
2r √ rT
+e S0 − N −d − T − e N (−d ) .
2r m0T0 σ
21.11.3. Limited risk options

When m and K are constant, then the payoff of a limited risk call at T is:
(ST − K)+ 1(MTT ≤m) .

0
The call’s current value is zero when MT00 > m.

When MT00 ≤ m, the call’s current value is:
√ √
Clr = S0 [N (d) − N (dm )] − e−rT K[N (d − σ T ) − N (dm − σ T )]
(−2r/σ2 )
S0 r√ √
+m N 2dm − d − 2 T −σ T
m σ

2r √ S0 2r
− N dm − +σ T −e −rT
K e− σ 2
σ m

r√ r√
× N 2dm − d − 2 T − N dm − 2 T
σ σ
with

1 S0 σ2
dm = √ ln + rT + T .
σ T m 2
The payoff of a limited risk put is (K − ST )+ 1(mTT ≥m) .

0
When m0T0 < m, the put is worthless.
When m0T0 ≥ m, the put’s current value is:
p = −S0 [N (−d) − N (−dm )]

√ √
+ e−rT K[N (−d + σ T ) − N (−dm + σ T )]
(−2r/σ2 )
S0 2r √ √
−m N −2dm + d + T +σ T
m σ
− 2r2 +1
r√ √ −rT S0 σ
− N − dm + 2 T +σ T −e K
σ m

2r √ 2r √
× N − 2dm + d + T − N − dm + T .
σ σ
These options are issued in foreign exchange markets and also in stock index
markets.
21.11.4. Partial lookback options

The payoff of this option is (ST − ηmTT0 )+ with η > 1.
The current value of a partial lookback call is:

ln(η) ln(η) √
Cpl = S0 N d − √ − ηe−rT m0T0 N d − √ − σ T
σ T σ T
2 −( 2 )
2r
σ S0 σ ln(η) 2r √
+ e−rT ηS0 × N −d
− √ + T
2r m0T0 σ T σ

rT ( σ2r2 ) ln(η)
−e η N −d − √ .
σ T
The payoff of a partial lookback put is (ηMTT0 − ST )+ with 0 < η < 1.

When η = 1, these options become standard lookback options.
The current value of a partial lookback put is:

ln(η) −rT ln(η) √
p = −S0 N −d + √ − ηe MT0 N d + √ + σ T
0
σ T σ T
2 ( 2r2 )
−rT σ S0 σ
ln(η) 2r √
−e ηS0 N d + √ − T
2r MT00 σ T σ

2r ln(η)
− erT η ( σ2 ) N d + √
σ T
When η = 1, these options become standard lookback options. Tables 21.26

and 21.27 provide simulations values of lookback option values and the
Greek letters for different parameters. The reader can compare the evolution
of the different values.
Table 21.26. Simulations values of standard lookback calls. S = 100, t = 27/12/2003,

T = 27/12/2004, r = 2%, σ = 20%, and historical minimum for call values = 80.
80 12.69868 0.15888 0.04822 0.55516 −0.01517

85 14.07381 0.38520 0.04043 0.55539 −0.01516
90 16.47870 0.56967 0.03155 0.50224 −0.01370
95 19.69548 0.71022 0.02317 0.42028 −0.01144
100 23.51395 0.81140 0.01616 0.33046 −0.00897
105 27.75571 0.88085 0.01081 0.24686 −0.00668
110 32.28247 0.92664 0.00697 0.17669 −0.00477
115 36.99416 0.95584 0.00436 0.12199 −0.00328
120 41.82216 0.97392 0.00266 0.08168 −0.00219
Table 21.27. Simulations values of standard lookback calls. S = 100, t = 27/12/2003,

T = 27/12/2004, r = 2%, σ = 20%, and historical minimum = 90.
80 12.69868 0.15873 −0.00000 0.55516 −0.01517

85 13.49235 0.15873 −0.00000 0.58986 −0.01611
90 14.28601 0.15888 0.04293 0.62455 −0.01706
95 15.59962 0.36200 0.03687 0.62826 −0.01716
100 17.84962 0.53253 0.02989 0.58207 −0.01588
105 20.86480 0.66806 0.02305 0.50555 −0.01377
110 24.47425 0.77083 0.01702 0.41656 −0.01132
115 28.52559 0.84569 0.01213 0.32846 −0.00891
120 32.89347 0.89836 0.00837 0.24952 −0.00675
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
on the minimum and options on the maximum of two assets, extendible

options, equity-linked foreign exchange options and quantos, binary-barrier
options and lookback options.
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
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
of Finance, 13(3), 1975–2000.
Bellalah, M, Prigent JL and C Villa (2001a). Skew without skewness: asymmetric
Finance, 13(2), 1826–1837.

Geske, R (1979). A note on an analytical valuation formula for unprotected
American call options with known dividends. Journal of Financial Economics
7, 375–380.
Johnson, H and D Shanno (1987). Option pricing when the variance is changing.
Journal of Financial and Quantitative Analysis, 22, 143–151.
Margrabe, W (1978). The value of an option to exchange one asset for another.
Journal of finance 33, 177–186.
Rubinstein, M (1991a). Pay now, choose later. Risk, 4(2), February, 44–47.
Rubinstein, M (1991b). Double trouble. Risk, 4 (December–January), 53–56.
Rubinstein, M (1991c). Options for the undecided. Risk, 4 (April), 70–73.
Rubinstein, M (1991d). Somewhere over the Rainbow. Risk, 4(10) (November),
63–66.
Rubinstein, M and E Reiner (1991). Breaking down the barriers. Risk, 4(8), 28–35.
Stulz, RM (1982). Options on the minimum or maximum of two risky assets:
analysis and applications. Journal of Financial Economics, 10, 161–185.
Turnbull, SM and LM Wakeman (1991). A quick algorithm for pricing European
average options. Journal of Financial and Quantitative Analysis, 26(3),
September, 377–389.

Chapter 22
VALUE AT RISK, CREDIT RISK, AND

CREDIT DERIVATIVES
Chapter Outline
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
of a portfolio of a financial institution or financial instruments with a given

probability over a pre-specified horizon.
In the same vein, it is important to appreciate credit risk of a financial
institution with respect to the appropriate tools. Credit valuation appears
today as a fundamental subject in research and practice. Credit-risk
analysis and credit valuation need some preliminary definitions of the
basic concepts. Credit exposure refers to the amount, which is subject to
changes in value upon a change in credit quality or a loss in the event of
default.
The average shortfall corresponds to the expected loss given that a loss
occurs or exceeds a given level.
The counterparty refers to the partner in a transaction in which each
side takes broadly comparable credit risk to the other.
Credit scoring refers, in general, to the estimation of the relative
likelihood of default for a firm.
The current exposure corresponds to the amount it would cost to
replace a transaction now if the counterparty defaults.
The default probability refers to the likelihood that an obligor will
encounter credit distress within a given period.
Credit-quality migration refers to the possibility that an obligor with a
certain credit rating migrates to any other credit rating by the risk horizon.
Credit derivatives are financial instruments which isolate credit risk.
They facilitate the trading of credit risk, its transfer, and hedging. There
are different categories of credit derivatives: forwards, swaps, options, and
some building blocks combining some of these main products.
Credit derivatives are often presented in the form of three classes of
instruments: total return swaps (total rate of return swaps, loan swaps,
or credit swaps), credit-default instruments and credit-spread instruments.
Total return swaps are conceived to transfer the credit risk to the
counterparty. Credit-default instruments give a certain payoff upon the
occurrence of a default event. They are often in the form of a credit-
default swap or default options. Credit-spreads instruments are often
in the form of forward or option contracts on credit-sensitive assets.
Credit derivatives allow a re-structuring of the risk/return profiles of
credits and permit investors to access new markets. The investors have
the possibility to structure and to optimize the risk-adjusted perfor-
mance of their liabilities by diversifying them among several markets and
instruments.
Value at Risk, Credit Risk, and Credit Derivatives 919
22.1. VaR and Riskmetrics: Definitions and Basic Concepts

The VaR concept gives an answer to the following question:
How much the investor will lose with x% probability over a specified period
of time?
Riskmetrics is a set of tools, allowing the users to estimate their
exposure to market risk under the “VaR framework”. The market risk
corresponds to the potential changes in the value of a position as a
consequence of the changes in market prices. JP Morgan developed the
Riskmetrics of VaR methodologies and published them in a technical
document via Internet.
The integration of VaR in modern financial management requires that
all positions to be market-to-market and needs the estimation of future
variability of the market value. VaR is used by dealers, non-financial
corporations, institutional investors, bank and securities firm regulators,
and securities and exchange commissions. In 1995, the Basle Committee on
banking supervision proposed allowing banks to determine their market-
risk capital requirements by implementing the bank’s VaR model. In this
case, the committee specifies the parameters to load into the VaR model.
Full-valuation models are based on re-valuing the portfolio on different
scenarios. These scenarios can be generated using historical simulation,
distributions if returns generated from a set of variance–covariance matrixes
etc. This process is known as stress testing. These methods account for the
whole distribution of returns instead of a single VaR number. However, they
are time consuming.
The second difference between VaR approaches is how market move-
ments are estimated. Riskmetrics assumes at the beginning conditional
normality to estimate market movement. However, this approach was
refined and accounts now for higher order moments of the distribution of
returns (kurtosis and leptokurtosis).
Many practitioners think that the VaR number can be used to
aid managers in the understanding of their risk position. In practice,
the management of an individual trader’s book position requires more
careful considerations of the risk parameter sensitivities than the single
VaR number. This is important for the management of the option-price
sensitivities or Greek letters.
The European Union’s Capital Adequacy Directive, recognizes VaR
models as a valid model for the determination of capital requirements for
foreign-exchange risks and other market-risk capital requirements.
22.1.1. The definition of risk

Risk is often defined as the degree of uncertainty regarding future returns.
This global definition of risk can be extended to define different kinds
of risks according to the source of the underlying uncertainty. The
operational risk indicates the possible errors in settling transactions or in
instructing payments. The credit risk refers to the potential loss resulting
from the inability of a firm to fund its illiquid assets. Market risk
refers to the deviations of future earnings due to the changes in market
conditions.
22.1.2. VaR: Definition

Value-at-Risk is a measure of the maximum potential change in the value
of a portfolio of a financial institution or financial instruments with a given
probability over a pre-specified horizon. The VaR concept gives an answer
to the following question:
How much the investor will lose with x% probability over a specified
period of time?
The following two examples are adapted from Riskmetrics. If an
investor estimates that there is a 95% chance that the Euro/USD exchange
rate will not fall by more than 2% of its current value over the next
day, he/she can determine the maximum potential loss on, for example,
USD100 million Euro/USD position.
Example 1. Consider a USD-based firm which holds an 105 million

FX position. The manager wants to calculate the VaR over a one-day
horizon. He/she thinks that there is a 5% chance that the loss will be
higher than what VaR projected. The exposure to market risk must be
determined in a first step. The exposure corresponds to the market value
of the position in the investor base currency (the US firm). If the foreign
exchange rate is 1.05 /USD, the market value of the position is USD100
million or (105/1.05).
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:
FX risk = 100 million (1.65)(0.55%) = 907.500 Dollars.

Hence, in 95% of the time, the firm will not lose more than 907,500 over
the next day.
Example 2. Consider now a USD-based firm which holds an 105 million

position in a 10-year government bond. The manager wants to calculate the
VaR over a one-day horizon. He/she thinks that there is a 5% chance of
understating the realized loss. In a first step, the exposure to market risk
must be determined. The exposure corresponds to the interest-rate risk on
the bond and the FX risk resulting from the Euro. The market value of
the position is still USD100 million, which is at risk to the two market risk
factors (interest rates, exchange rates etc).
The estimation of the interest-rate risk in a second step needs the

calculation of the standard deviation on a 10-year European bond in Euro.
Suppose this standard deviation is equal to 0.58%. In this case, we have:
Interest-rate risk = 100 million Dollars (1.65)(0.58%) = 957,000 Dollars.
The estimation of the FX risk is given by:
FX risk = 100 million Dollars (1.65)(0.55) = 907,500 Dollars.
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).
22.2. Statistical and Probability Foundation of VaR

Is the distribution of return constant over time?
The time series reveals volatility clustering since periods of large returns
are clustered and distinct from periods of small returns, which are also
clustered. This shows clearly a change in variances referred to as het-
eroscedasticity. Researchers investigate alternative modeling methods other
than the normal distribution. These models use either unconditional (time-
independent) or conditional distributions of returns (time-dependent).
The first class corresponds, for example, to the normal distribution,

finite-variance symmetric and asymmetric stable paretian distributions etc.
The second class of models corresponds to GARCH and stochastic volatility
models for example.
22.2.1. Using percentiles or quantiles to measure

market risk
The percentile or quantile corresponds to a magnitude (the dollar amount
at risk) and is given by the following formula for a continuous probability
distribution:
α
p= f (r)dr;
−∞
where f (r) corresponds to the probability density function. The fifth

percentile is the value such that 95% of the observations lie above it. If
we define r˜t as r˜t = (rtσ−µ
t
t)
, then r˜t is normal with mean 0 and a unit
variance.
Example. Suppose an investor wants to find the 5% percentile of rt under
the normal distribution. Since
probability (r˜t < −1.65) = 5%,
or

(rt − µt )
probability r˜t = < −1.65 = 5%,
σt
then
probability (rt < −1.65σt + µt ) = 5%.
This equation says that there is a 5% probability that an observed return at

time t is less than −1.65 times its standard deviation plus its mean. When
µt = 0, we obtain the classic result for short-term horizon VaR calculation:
Probability (rt < −1.65σt) = 5%.
22.2.2. The choice of the horizon

Several models consider a horizon of one day and 95% to 99% confidence
interval in the measurement of the amount of risk for the institution. This
horizon assumes implicitly that markets and assets are very liquid and
Table 22.1. VaR of a single asset at the 1% level.
Alpha 1−Alpha Sigma t/t − 1

(in %) (in %) Z−alpha (in %) V (t − 1) VaR
1 99 2.3263 10 100 23.263

1 99 2.3263 11 100 25.59
1 99 2.3263 12 100 27.916
1 99 2.3263 13 100 30.242
1 99 2.3263 15 100 34.895
1 99 2.3263 20 100 46.527
1 99 2.3263 25 100 58.159
Table 22.2. VaR of a single asset at the 5% level.
Alpha 1−Alpha Sigma t/t − 1

(in %) (in %) Z−alpha (in %) V (t − 1) VaR
5 95 1.6449 10 100 16.449

5 95 1.6449 11 100 18.093
5 95 1.6449 12 100 19.738
5 95 1.6449 13 100 21.383
5 95 1.6449 15 100 24.673
5 95 1.6449 20 100 32.897
5 95 1.6449 25 100 41.121
allow the different participants to unwind their positions in one day. This
approximation is mainly valid for market-trading activities.
The correlation between financial prices and aggregation

Several VaR models use the historic correlations among the different
risk factors with respect to the main results in modern portfolio theory.
However, the historic correlations are unstable and other measures can be
used. For more details, see Garman (1996b), Hoppe (1998), and Bellalah
and Lavielle (2003).
The following Tables 22.1 and 22.2 provide the VaR of a single asset at
the 1% level.
22.3. A More Advanced Approach to VaR

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. VaR calculations can also be done without resorting

to the standard deviation. Consider a manager who wants to compute the
VaR of a portfolio over one day with a 5% chance that the actual loss in
the portfolios value is higher than the VaR estimate. In this case, the VaR
calculation is performed in four steps.
The first step determines the current value of the portfolio V0 on a mark-
to-market basis.
The second step determines the future value of the portfolio V1 = V0 er̃ ,
where r̃ refers to the portfolio’s return over the horizon.
The third step estimates the one-day return on the portfolio, r̃ in a way
such that there is a 5% chance that the actual return will be less than r̃,
i.e., probability (r < r̃) = 5%.
The fourth step determines the portfolios future “worse case” value,
V˜1 as:
V˜1 = V0 er̃
In this context, the VaR estimate is:
V0 − V˜1 .
This VaR estimate can also be written as:
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
selected an approach based on their specific needs. The models implemented

differ on the way changes in the values of a portfolio are estimated as a
reaction to market movements. They also differ on the way the potential
market movements are estimated.
Basically, there are two approaches to estimate the change in the
portfolios value as a result of market movements: the analytic approach
and the simulation approach.
Tables 22.3 to 22.5 illustrate the computation of VaR for a portfolio of
21 assets.
Non-linear positions correspond, for example, to a portfolio of opti-
ons. Riskmetrics provides two approaches to compute VaR of non-linear
Table 22.3. Weights, drifts, and volatility for a portfolio with 21

assets.
Weights Drift Volatility

Stocks (in %) (in %) (in %)
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).
Alpha 1−Alpha Z−alpha VaR Horizon time (days)
5% 95% 1.6449 8% 292

September 10, 2009 14:41
926
Table 22.5. Correlation matrix for the portfolio
T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15 T16 T17 T18 T19 T20 T21
T1 1 0.086 0.116 0.114 108 0.167 0.064 0.119 0.037 0.108 0.191 0.114 0.094 0.055 0.066 0.072 0.051 0.060 0.189 0.099 0
Derivatives, Risk Management and Value

T2 0.086 1 0.096 0.094 0.089 0.137 0.051 0.098 0.028 0.089 0.161 0.093 0.076 0.041 0.051 0.056 0.038 0.045 0.160 0.080 0
T3 0.116 0.096 1 0.137 0.130 0.2 0.074 0.142 0.040 0.130 0.234 0.136 0.111 0.060 0.075 0.082 0.055 0.066 0.232 0.117 0
T4 0.114 0.094 0.137 1 0.144 0.222 0.082 0.158 0.044 0.144 0.336 0.151 0.123 0.065 0.082 0.090 0.061 0.072 0.259 0.129 0
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
positions. The first is an analytical approximation. The second is a

structured Monte–Carlo simulation.
22.4. Credit Valuation and the Creditmetrics Approach

22.4.1. The portfolio context of credit
The need for a quantitative portfolio approach to credit risk management
allows the study of concentration risk. This risk refers to additional
risk resulting from higher exposure to one or several correlated obligors.
A portfolio credit-risk methodology as the one in Creditmetrics allows to
capture simultaneously the benefits of diversification and concentration
risks and provides an efficient risk-based capital allocation process.
If we consider a bond rated BBB, which matures in n periods, then
at the end of the year, the bond stays at BBB, the issuer defaults or it
migrates up or down to one of the other categories. Hence, the probabilities
that this bond will end up in one of the other categories in a period
allow the computation of the bond price under each of the possible rating
scenarios. The new present value of the bond can then be calculated
from the remaining cash flows under its new ratings. The discount rate
is obtained from the forward zero curve, which is different for each rating
category. The knowledge of the probabilities or likelihoods for the bond to
be in a given rating category and the values of the bond in these categories
allow the determination of the distribution of value of the bond in one
period.
22.4.2. Different credit risk measures

The two measures used in creditmetrics in the appreciation of credit
risk are the standard deviation and the percentile level. These measures
reflect the potential losses from the same portfolio distribution. The first
measure (determination of the standard deviation) needs the computation
of the mean value for the portfolio by multiplying the values with the
corresponding probabilities and then adding the resulting values. This
allows the computation of the standard deviation. The second measure is a
specified percentile level which indicates the lowest value that the portfolio
will acheive 1% of the time (the first percentile). The likelihood that the
actual portfolio value is less than this number is only 1%.
The above analysis and the proposed concepts apply to other types of
exposures like loans, letters of credit, swaps, and forward contracts.
22.4.3. Stand alone or single exposure risk calculation

The procedure used in creditmetrics for the computation of the credit risk
for a single or stand-alone exposure is based on three steps.
The first step: Credit-rating migration

This step estimates changes in value due to up (down) grades and default.
The likelihood of credit-rating migration is conditioned on the senior
unsecured credit rating of the obligor. A transition matrix is conceived
using public rating migration data. This allows the determination of the
likelihoods of migration to any credit-quality state in a given period.
The second step: Valuation

The values at the risk horizon are determined for the credit-quality states.
When the credit-quality migration corresponds to a default case, the likely
residual value is a function on the seniority class of the debt. When the
credit-quality migration is in an other category, the forward zero curves for
each rating category are used to re-value the bond’s remaining cash flows.
The third step: Credit-risk estimation

Using the likelihoods and values, it is possible to calculate the risk estimate:
the standard deviation or the percentile level.
22.4.4. Differing exposure type

Creditmetrics determines the credit risk for market-driven instruments like
swaps and forward contracts. The value of a swap is given by the difference
between two components. The first corresponds to the forward risk-free
value of the swap cash flows. This component is the same for all forward
credit rating states. The second component corresponds to the loss expected
on a swap, resulting from a default net of recoveries by the counterparty
on all the cash flows after the risk horizon. The difference between this
second component and the first one allows the re-valuation of the swap.
The re-valuation of the swap is based on the following formula: value of the
swap in a period (with rating R) = risk-free value in one period expected
loss in period 1 through maturity (with rating R), where R refers to any
credit-rating category. The expected loss for each forward non-default credit
rating is given by:
Expected loss (with rating R) = average exposure (from period 1 to

maturity).
Probability of default in period 1 through maturity (with rating R) (1 −

recovery fraction).
The average exposure calculation is time consuming. The probability

of default for each rating category between year 1 and the maturity
corresponds to the expected loss calculation. It is calculated using a
transition matrix. This method allows the computation of the swap value
in each of the non-default credit-rating categories. In the case of default
during the risk interval, the expected loss in the defaulted state is given
by: expected loss (case of default) = expected exposure (in the first period)
(1 − recovery fraction).
This expression assumes that the risk interval is very short (one year
for example).
22.5. Default and Credit-Quality Migration in the

Creditmetrics Approach
22.5.1. Default
Credit-rating agencies assign an alphabetic or numeric label to rating
categories. They defined a default event with reference to missed interest
and principal payments. The likelihood of credit distress is defined in
Creditmetrics with respect to default rates. They use credit ratings as an
indication of the chance of default and credit-rating migration likelihood.
They consider that the firm-encountered credit distress even in a context
when only the subordinated debt realized a default. The methodology in
J.P Morgan assumes that the senior credit rating is the most indicative of
encountering credit distress. Filling probabilities of default with a transition
matrix Creditmetrics uses historical default studies to obtain transition
matrices, which comprise one-year default rates.
22.5.2. Credit-quality migration

The value of a firm and its assets changes suggest changes in credit quality.
As a firm moves towards bankruptcy, the value of equity falls. Since, the
credit rating of the firm is given, it is possible to work backward to the

“threshold” in asset value that delimits default. In the approach used by
JP Morgan, the firm default model uses the default likelihood to place a
threshold below which default appears. The methodology uses the rating
migration probabilities to define thresholds above which the firm would
upgrade or downgrade from its current credit rating. The rating migration
probabilities are represented by a transition matrix. This matrix is a square
table of probabilities which reflect the likelihood of migrating to an other
rating category in a period given the obligator’s present credit rating.
22.5.3. Historical tabulation and recovery rates

Creditmetrics uses a technique to model different volatilities of credit-
quality migration conditioned on the actual credit rating. Hence, each row
in the transition matrix allows the description of a volatility of credit-
rating changes, which is unique to that row’s initial credit rating. Some
rating agencies publish tables of cumulative default likelihood over longer
holding periods. It is possible to use a cumulative default-rate table to get
an implied transition matrix which better replicates the default history. In
this spirit, Creditmetrics uses a transition matrix to model credit-rating
migrations. This matrix can be constructed by using a least squares fit
to the cumulative default rates. The Markov process is used to model the
proceess of default. Using the Markov process, it is possible to generate a
cumulative default rates matrix from an imputed transition matrix. In the
event of default, the estimation of recovery rates is not an easy task. The
bond market prices is an efficient way the future realized liquidation values.
The examination of the recovery statistics by seniority class shows
that the subordinated classes are appreciably different from one another
in their recovery realizations. However, there are no statistically significant
difference between secured and unsecured senior debt.
Table 22.6 provides the one-year transition matrix. Table 22.7 gives the
recovery rates for different types of debt. Table 22.8 gives the credit spreads
for different ratings and years.
22.6. Credit-Quality Correlations

The estimation of default correlations is not an easy task.
Table 22.6. Input data 1 year transition matrix.
AAA AA A BBB BB B CCC Default Sum

(in %) (in %) (in %) (in %) (in %) (in %) (in %) (in %) (in %)
AAA 90 8.05 0.60 0.05 0.20 0 0 1.10 100

AA 0.70 91.14 6.30 0.40 0.10 0.20 0.01 1.15 100
A 0.08 2.32 90 4 0.70 0.30 1 1.60 100
BBB 0.03 0.29 6 85 5 1.10 0.09 2.49 100
BB 0.02 0.09 0.34 5.66 84 6.15 0.80 2.94 100
B 0 0.09 0.11 0.30 6.34 86 3.50 3.66 100
CCC 0.17 0 0.24 1 1.99 8.50 70 18.10 100
Table 22.7. Recovery rates.
Mean (%) Standard Deviation (%)
Senior secured 65 20
Senior unsecured 45 18
Senior subordinated 35 16
Subordinated 25 19
Junior subordinated 15 10
Table 22.8. Credit spreads.
1 2 3 5 7 10 20 30
(in %) (in %) (in %) (in %) (in %) (in %) (in %) (in %)
AAA 0.12 0.17 0.22 0.27 0.32 0.37 0.42 0.47

AA 0.42 0.47 0.52 0.57 0.62 0.67 0.72 0.77
A 0.52 0.77 0.82 0.87 0.92 0.97 1.02 1.07
BBB 0.62 1.07 1.12 1.17 1.22 1.27 1.32 1.37
BB 0.72 1.37 1.42 1.47 1.52 1.57 1.62 1.67
B 0.82 1.67 1.72 1.77 1.82 1.87 1.92 1.97
CCC 0.92 1.97 2.02 2.07 2.12 2.17 2.22 2.27
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 )
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.
The estimation of default correlations is a first step towards the esti-

mation of the joint likelihood of any possible combination of credit-quality
outcomes. When the credit rating system uses n states (AAA, . . . ,default),
then between two obligators, there exists n(n) possible joint states for which
the likelihoods can be estimated. The estimation of joint credit-quality
migration likelihoods can be done using credit ratings time series across
several firms. This estimation method assumes that all firms with a given
credit rating are identical.
22.7. Portfolio Management of Default Risk in the

Kealhofer, McQuown and Vasicek (KMV) Approach
Corporate bonds and liabilities are subject to default risk. The default risk
is in general less than 0.5% for a typical high-grade borrower. This risk
cannot be hedged away. However, it can be shifted and someone must bear
it in the end. Portfolio theory and quantitative methods have been used to
compute the amount of risk reduction attainable through diversification
for a portfolio of equity. This theory is widely used by practitioners
who developed techniques for computing the asset attributes, which are
fundamental for an actual portfolio management tool. These developments
have not been implemented for debt portfolios. KMV developed these
methods in its practice with commercial banks in order to measure
diversification and to maximize return in debt portfolios.
22.7.1. The model of default risk

When a lender acquires a corporate note, it is as if he/she is engaged in two
transactions. The first is buying a debt obligation. The second is the sale of
a put option to the borrower. In fact, when the firm’s assets are less than
the face value of the debt, the borrower “puts” the assets to the lender and
uses the cash proceeds to pay the note. Hence, the situation of the lender
can be represented by a portfolio with two assets: long a risk-free bond and
short a default option. The probability of default can be determined using
option pricing theory and mainly the volatility of the firm’s assets. If you
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.
22.7.2. Asset market value and volatility

The firm’s equity behavior can be also derived from the firm’s asset values.
The position of stockholders can be viewed as a call on the firm’s asset value,
where the strike price is given by the face value of debt obligations. If at the
debt’s maturity date, the firm’s value is higher than the amount of liabilities,
stockholders exercise their calls by paying off their obligations. Otherwise,
they default. When the market value changes, this induces changes in the
value of liabilities depending on the degree of seniority. When the asset
value falls to a critical level, the probability of default increases and the
market value of the liabilities decreases. In general, when the asset value
falls, the volatility of equity increases. Option pricing theory can be used
to infer the volatility and asset value.
22.8. Credit Derivatives: Definitions and Main Concepts

22.8.1. Forward contracts
Forward contracts on bonds can be either cash-settled or physical-settled.
The cash flows for this forward agreement commits the buyer to buy a
given bond at a specified future date at a pre-determined price specified at
contract origination (time t = 0). The agreement can specify this instead of
using the price, the bond’s spread over a treasury asset or a benchmark will
be used. In this operation, there are two maturity dates: the maturity of the
forward agreement and the maturity of the reference bond. The maturity
of the forward agreement is, in general, shorter than that of the reference
bond. Since, in general, the default risk is borne by the buyer of the forward
contract on the spread, he/she will pay at the maturity date the following
quantity: (spread in forward agreement − spread at maturity) duration
(notional amount). When there is a credit event, the transaction is marked
to market and unwound.
22.8.2. The structure of credit-default instruments

Credit-default instruments dissociate the risk of default on credit obli-
gations. They are often presented in the form of either credit-default
swaps or credit-default options. In general, when there is a credit event
by the reference credit, then according to the terms of a credit-default
swap, the bank pays the counterparty an agreed default payment. The
counterparty pays a periodic fee and benefits from the protection of the
risk of default of the reference credit. This credit derivative structure is
based on the replication of the total performance of the underlying credit
asset (a reference bond, a loan). The swap is done between an investor and
the bank. The investor assumes the risks of the reference bond. The bank
passes through all payments of the bond and in return, the investor makes
a payment akin to a funding cost. The transaction is based on a notional
amount. The current bond price is used to compute the settlement value
under the transaction. The investor receives interest payments and pays a
money-market interest rate plus or minus a specified margin.
22.8.2.1. Total return swaps

In a total return swap, the rate payer makes periodic payments to another
party (the total return payer). He/she receives the total return less the
principal and interest payments plus or minus the price changes of the
reference asset. The total return swap is often used in the swap structures
on default risk. The two parties in a total return swap define at origination
the initial value P0 of the reference asset and agree on the reference rate.
At any time between t = 0 and the maturity date corresponding to the
settlement dates, the asset receiver obtains the cash flows from the reference
asset. He/she pays a certain amount fixed with reference to the reference
rate. At the maturity date of the contract, the value of the reference asset
PT is used. If this value is greater than P0 , the asset receiver gets the
difference (PT − P0 ), otherwise, he/she pays the difference (P0 − PT ).
The credit swap can also be based directly on a spread. In this case, the
asset receiver pays at maturity the difference in the spread of the reference
asset over a treasury security with a comparable maturity at origination
and at maturity.
22.8.2.2. Credit-default swaps

This derivative contract allows one party (the protection seller) to receive
fixed periodic payments from the protection buyer. The payments are in
return of making a single contingent payment covering losses with respect

to a reference asset following a default or an other specified “credit event”.
The main idea behind credit-default swaps is that they strip off the
default risk of some reference assets. This risk will be traded separately.
By implementing a credit-default swap, the protection seller earns an
investment income and the protection buyer hedges the risk of default
on the reference asset. This contract allows investors to hedge credits
without implementing costly strategies consisting of buying and selling cash
securities and loans.
The following example is adapted from a sponsor’s statement by
Barclays capital.
Example: Consider an investor A who gains customized access through

a bank B to a corporate bond by selling 3-year default protection to the
bank B on the bond. The investor A receives a fixed premium of 120 bp per
annum and agrees to make a credit event payment if the borrower defaults
on the bond. In the credit event, the swap terminates and the investor A
pays the bank B, the notional times the percentage fall from par of the
bond. The investor A can also settle the swap by buying the bond from the
bank B at par.
22.8.2.3. Basket default swaps

An investor can sell default protection on several assets. In a first-to-default
basket-default swap, the protection seller assumes the default risk on a
basket of bonds by agreeing to compensate market losses on the first asset
in the basket to default.
22.8.2.4. Credit-default exchange swap

It is possible to swap a default risk on an asset for that of an other asset. In a
credit-default exchange swap, both parties act simultaneously as protection
buyers and protection sellers.
Example: Consider two institutions A and B. A trades the default risk of

a loan it holds for that of a complementary loan held by B. A lays off the
default risk on loan A in return for assuming default risk on loan B. If a
reference credit experiences a credit event then the protection seller must
make a credit event payment to the protection buyer. This can terminate
the trade. The trade could continue with the protection buyer in the rest of
transaction paying an agreed rate. Parties do not make periodic payments

and swap just the contingent payments when default risks are perfectly
matched.
22.8.2.5. Credit-linked notes (CLNs)

Credit linked notes are associated with the credit performance of the
underlying assets. A principal protected note protects a pre-set portion
of principal. A principal-linked structure pays enhanced fixed coupons and
can redeem principal at a rate associated with the credit performance of
reference assets.
Credit-default notes allow investors either to buy or sell default
protection on reference credits.
22.9. The Rating Agencies Models and the Proprietary

Models
22.9.1. The rating agencies models
The models of default risk developed by rating agencies are based on the
current rating and the time to maturity of the obligation. These models
determine the cumulative risk of default (the total default probability over
a given period) and/or the marginal risk of default (the change in default
probability over some periods). In a first special report by the Standard
and Poor’s website, we find that corporate defaults rose sharply in 1998.
Their database contain a vast collection of statistics on default and rating
migration behavior on CD-ROM under the trade name CreditP roT M .
The data used corresponds to the issuer credit ratings that reflect S&Ps
opinion of a company’s overall capacity to pay its obligations. This opinion
is based on the obligor’s ability to meet its financial commitments on a
timely based. This indicates in general the likelihood of default regarding
the firm’s financial obligations. The definition of default corresponds to the
first occurrence of a payment default on any financial obligation. Preferred
stocks are not considered as financial obligations since a missed preferred
stock dividend cannot be equated to a default. The studies of default by
S&P are based on groupings called static pools. A static pool is constructed
at the first day of each year covered by the study and followed from that
point on. All obligors are followed year to year within each pool. All
of Standard and Poor’s default studies reveal a well-defined correlation
between credit quality and default remoteness. The rating letters in the
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:
Pd : the prices at the end of the default month, (referring to a default

event),
Pe : prices (just) preceding liquidation or emergence from out-of-court
settlement or Chapter 11 re-organization, (referring to an emergence
event).
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;
Table 22.9. Recoveries shortly after default.
Seniority — ranking Nobs SA, Pd Sd, Pd W A, Pd CV , Pd
Senior — secured 65 58.52 22.27 58.11 0.38

Senior — unsecured 180 49.60 26.51 53.66 0.53
Subordinated 144 38.29 25.23 36.86 0.66
Junior — subordinated 144 35.30 22.29 36.41 0.63
Total 533 43.77 25.81 43.93 0.59
W A: the weighted average by issue size in Dollars and

CV : the coefficient of variation.
Table 22.9 shows that investors who liquidate a position in defaulted

securities (shortly after default) expect to recover, on average, about
44 Cents on the Dollar. This means that on average, creditors receive about
40 Cents on the Dollar. Table 22.9 also shows a certain predictable degree of
variation across seniority classes. The study shows that average recoveries
at default and emergency are higher, the higher the seniority rankings of
the issues used. It also reveals that excess returns exhibit much uncertainty
at default and emergence and that uncertainty is in general higher, the
lower the seniority of the debt used.
22.9.2. The proprietary models

The default probability can be calculated as the probability that asset
values will be lesser than the value of the claims on the firm’s assets. In this
spirit, KMV Corporation developed the expected default frequency model
(EDF). This model needs an estimation of the market asset values, the
volatility of the assets and the market value of the liabilities. The volatility
can be implied from an option pricing model. Using the asset value, the
volatility and the cumulative liabilities, it is possible to calculate the default
risk of the firm. This model determines the default probability using the
distance in volatility between the asset value and the point at which the
asset value will be less than the liabilities. The EDF model uses large
databases of firms in the computation of historical default frequencies. This
approach allows the estimation of expected and unexpected default losses
and derives default within a volatility framework. Default risk corresponds
to the uncertainty surrounding the firm’s ability to service its obligations.
The default probability of a firm depends on the market value of its assets,
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.
The first step

The asset value and its volatility are estimated using the asset market value,
the volatility of equity, and the book value of liabilities. Using an option
pricing based approach where equity is viewed as a call on the underlying
assets with a strike price equal to the book value of the firm’s liabilities, it
is possible to estimate the value of the firm and the volatility of its assets.
The second step

The distance to default is calculated using the asset value and its volatility,
and the book value of liabilities. The default probability depends on the
current value and the distribution of the asset value, the volatility, the
expected rate of growth in the asset value, the horizon, and the level of
the default point. If the future distribution of asset values is known, the
default probability and the EDF correspond to the likelihood that the final
asset value be below the default point.
The third step

The default probability is computed using historical data on default and
bankruptcy frequencies. A frequency table is generated to link the likelihood
of default to different levels of distance default.
Creditmetrics models the process of value changes resulting from

changes in credit quality. It gives risk management information which is
tailored to the name, industry, and sector concentrations. The details and
data are freely available on Internet.
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
protection are motivated by the desire to enhance their risk-adjusted return

on capital. Banks searching for higher yields are using credit derivatives
written on low-rated assets.
Straight credit-default swaps seem to be the most widely used of all
credit derivative products.
The pricing of credit derivatives and credit instruments depends on
the default probability of the reference asset, the expected recovery rate,
and the nature of the exposure. The pricing approaches concern individual
transactions and portfolios.
The pricing of individual transactions or single transactions concerns
the loss exposure, the default probability, the recovery rate, and the
correlations between these features in order to predict future credit losses.
The pricing in a portfolio approach is focused on the correlations
between individual exposures, which means that we must account for the
joint default probabilities (correlations between default) and the correlation
between loss exposure and recovery rates.
The proprietary models are based on an option approach to default in
which the equity of a levered firm is equivalent to a call on the net asset
value of the firm. This approach is initiated by Black and Scholes (1973).
This approach considers the position of debtholders as a combination of a
long position in a bond plus a short position in a put on the firm’s assets.
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.
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 14:46 spi-b708 9in x 6in b708-Index
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
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,
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
heat equation, 373, 570 information uncertainty, 338, 342,

hedge ratio, 170, 171, 334, 378, 346, 348, 521, 535, 544, 583, 591,
440–442, 452, 460, 461, 524, 525, 617, 634, 645, 650, 655, 656, 664,
674, 720, 752, 781 667, 676, 771, 772, 833, 839, 841,
847, 850, 851, 853
hedging, 4, 10, 20, 24, 32, 37, 40, 49,
interest rate trees, 293, 303, 313, 320,
57, 60, 67, 68, 70, 78–82, 85, 125,
355
127, 183, 196, 211, 249, 372, 396,
398, 416, 422–425, 439, 440, 445, interest rates, 777, 782, 791, 793, 795
452–454, 464, 465, 493–495, Itô lemma, 379, 493–495, 501–505,
516–518, 523, 525–527, 538, 673, 507, 509–512, 514, 517, 527,
721, 731, 745–747, 749, 750, 752, 530–532, 553, 554, 561, 562, 575,
754, 768, 779, 781, 782, 795, 813, 576, 579, 584, 587, 590, 591, 594,
850, 852, 875, 878, 918 651, 663, 674, 695, 699, 777, 780,
813, 851
historical volatility, 37, 57, 167–169,
175, 422, 460
jump process, 320, 745–747
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
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
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,
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

Derivatives Mondar Bellalah PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Derivatives Mondar Bellalah PDF

Uploaded by

Copyright:

Available Formats

DERIVATIVES,

This page intentionally left blank

British Library Cataloguing-in-Publication Data

DERIVATIVES, RISK MANAGEMENT & VALUE

Typeset by Stallion Press

Sandhya - Derivatives, Risk Management.pmd 1 1/11/2010, 5:41 PM

I dedicate this book to the President of the Tunisian Republic, his

This page intentionally left blank

by Edward C. Prescott (Arizona State University; Federal Reserve Bank of

This page intentionally left blank

by Harry M. Markowitz (University of California, San Diego)

Herein follows a remarkable volume, suitable as both a textbook and a

This page intentionally left blank

by James J. Heckman (University of Chicago and University College Dublin)

Mondher Bellalah oﬀers a lucid and comprehensive introduction to the

This page intentionally left blank

by George M. Constantinides (University of Chicago)

This page intentionally left blank

ABOUT THE AUTHOR

After obtaining his Ph.D. in Finance in 1990 at France’s leading University

This page intentionally left blank

PART I. FINANCIAL MARKETS AND

CHAPTER 1. FINANCIAL MARKETS, FINANCIAL

xviii Derivatives, Risk Management and Value

1.4. Description of Markets and Trading Instruments: The Case

CHAPTER 2. RISK MANAGEMENT, DERIVATIVES

xx Derivatives, Risk Management and Value

2.3.2. Speculation using futures markets . . . . . . . . . 84

2.12. Conversions and Reversals . . . . . . . . . . . . . . . . . . . 106

xxii Derivatives, Risk Management and Value

2. Buying and Selling Calls . . . . . . . . . . . . . . . . . . . . 131

CHAPTER 3. TRADING OPTIONS AND THEIR

3.2.9. Long a call bear spread . . . . . . . . . . . . . . . 161

PART II. PRICING DERIVATIVES AND THEIR

CHAPTER 4. OPTION PRICING: THE DISCRETE-

xxiv Derivatives, Risk Management and Value

4.2. The Binomial Model and the Distributions to the Underlying

CHAPTER 5. CREDIT RISKS, PRICING BONDS,

5.4.2. Duration of a bond portfolio . . . . . . . . . . . . 273

CHAPTER 6. EXTENSIONS OF SIMPLE BINOMIAL

xxvi Derivatives, Risk Management and Value

6.2.3. Computation of bond prices in the Ho and Lee

CHAPTER 7. DERIVATIVES AND PATH-DEPENDENT

7.2.1. On the valuation of derivatives and information

xxviii Derivatives, Risk Management and Value

PART III. OPTION PRICING IN A

CHAPTER 8. EUROPEAN OPTION PRICING

8.4. The Black–Scholes Model . . . . . . . . . . . . . . . . . . . 377

CHAPTER 9. SIMPLE EXTENSIONS AND

xxx Derivatives, Risk Management and Value

9.1. Applications of the Black–Scholes Model . . . . . . . . . . . 404

CHAPTER 10. APPLICATIONS OF OPTION PRICING

xxxii Derivatives, Risk Management and Value

PART IV. MATHEMATICAL FOUNDATIONS

CHAPTER 11. THE DYNAMICS OF ASSET PRICES

11.6.2. Pricing the option . . . . . . . . . . . . . . . . . . 525

CHAPTER 12. RISK MANAGEMENT: APPLICATIONS

xxxiv Derivatives, Risk Management and Value

Appendix B: Resolution of the Partial Diﬀerential Equation for

CHAPTER 13. SIMPLE EXTENSIONS AND

PART V. EXTENSIONS OF OPTION PRICING

CHAPTER 14. EXTENSION OF ASSET AND RISK

xxxvi Derivatives, Risk Management and Value