1.8K views

Uploaded by Shiv Shankar

A great book describing the finance models and models that shochatic mathematical calculations

- Finance
- 6fj3u.stochastic.calculus.for.Finance
- Cpp Quant Finance eBook 20131003
- Applied Quantitative Finance 5
- Financial Engineering
- Fundamental Models in Financial Theory.pdf
- Financial Modeling With Excel and VBA
- Mathematical Finance
- Computational Finance an Introductory Course With R
- The Mathematics Of Financial Derivatives
- Financial Analysis and Decision Making
- Financial Engineering Principles - Perry Beaumont
- Quantitative Financial Economics Stocks Bonds Foreign Exchange
- Discrete Models of Financial Markets
- Bouchaud Potters - Theory of Financial Risk and Derivative Pricing (2Ed CUP 2003)
- Algorithmic Trading & DMA: An Introduction to Direct Access and Trading Strategies
- A Buy-side Handbook
- The Handbook of Corporate Financial Risk Management
- Model Risk
- Financial Engineering

You are on page 1of 557

Hans Fllmer

Alexander Schied

Stochastic Finance

An Introduction in Discrete Time

De Gruyter

Mathematics Subject Classification 2010: Primary: 60-01, 91-01, 91-02; Secondary: 46N10, 60E15,

60G40, 60G42, 91B08, 91B16, 91B30, 91B50, 91B52, 91B70, 91G10, 91G20, 91G80, 91G99.

in the series De Gruyter Studies in Mathematics.

ISBN 978-3-11-021804-6

e-ISBN 978-3-11-021805-3

Library of Congress Cataloging-in-Publication Data

Fllmer, Hans.

Stochastic finance : an introduction in discrete time / by Hans

Fllmer, Alexander Schied. 3rd, rev. and extended ed.

p. cm.

Includes bibliographical references and index.

ISBN 978-3-11-021804-6 (alk. paper)

1. Finance Statistical methods.

2. Stochastic analysis.

3. Probabilities. I. Schied, Alexander.

II. Title.

HG176.5.F65 2011

332.011519232dc22

2010045896

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;

detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

2011 Walter de Gruyter GmbH & Co. KG, Berlin/New York

Typesetting: Da-TeX Gerd Blumenstein, Leipzig, www.da-tex.de

Printing and binding: Hubert & Co. GmbH & Co. KG, Gttingen

Printed on acid-free paper

Printed in Germany

www.degruyter.com

This third edition of our book appears in the de Gruyter graduate textbook series. We

have therefore included more than one hundred exercises. Typically, we have used the

book as an introductory text for two major areas, either combined into one course or

in two separate courses. The first area comprises static and dynamic arbitrage theory

in discrete time. The corresponding core material is provided in Chapters 1, 5, and 6.

The second area deals with mathematical aspects of financial risk as developed in

Chapters 2, 4, and 11. Most of the exercises we have included in this edition are

therefore contained in these core chapters. The other chapters of this book can be

used both as complementary material for the introductory courses and as basis for

special-topics courses.

In recent years, there has been an increasing awareness, both among practitioners

and in academia, of the problem of model uncertainty in finance and economics, often

called Knightian uncertainty; see, e.g., [259]. In this third edition we have put more

emphasis on this issue. The theory of risk measures can be seen as a case study how

to deal with model uncertainty in mathematical terms. We have therefore updated

Chapter 4 on static risk measures and added the new Chapter 11 on dynamic risk

measures. Moreover, in Section 2.5 we have extended the characterization of robust

preferences in terms of risk measures from the coherent to the convex case. We have

also included the new Sections 3.5 and 8.3 on robust variants of the classical problems

of optimal portfolio choice and efficient hedging.

It is a pleasure to express our thanks to all students and colleagues whose comments

have helped us to prepare this third edition, in particular to Aurlien Alfonsi, Gnter

Baigger, Francesca Biagini, Julia Brettschneider, Patrick Cheridito, Samuel Drapeau,

Maren Eckhoff, Karl-Theodor Eisele, Damir Filipovic, Zicheng Hong, Kostas Kardaras, Thomas Knispel, Gesine Koch, Heinz Knig, Volker Krtschmer, Christoph

Khn, Michael Kupper, Mourad Lazgham, Sven Lickfeld, Mareike Massow, Irina

Penner, Ernst Presman, Michael Scheutzow, Melvin Sim, Alla Slynko, Stephan Sturm,

Gregor Svindland, Long Teng, Florian Werner, Wiebke Wittm, and Lei Wu. Special thanks are due to Yuliya Mishura and Georgiy Shevchenko, our translators for the

Russian edition.

Berlin and Mannheim, November 2010

Hans Fllmer

Alexander Schied

Since the publication of the first edition we have used it as the basis for several

courses. These include courses for a whole semester on Mathematical Finance in

Berlin and also short courses on special topics such as risk measures given at the

Institut Henri Poincar in Paris, at the Department of Operations Research at Cornell

University, at the Academia Sinica in Taipei, and at the 8th Symposium on Probability

and Stochastic Processes in Puebla. In the process we have made a large number of

minor corrections, we have discovered many opportunities for simplification and clarification, and we have also learned more about several topics. As a result, major parts

of this book have been improved or even entirely rewritten. Among them are those on

robust representations of risk measures, arbitrage-free pricing of contingent claims,

exotic derivatives in the CRR model, convergence to the BlackScholes model, and

stability under pasting with its connections to dynamically consistent coherent risk

measures. In addition, this second edition contains several new sections, including a

systematic discussion of law-invariant risk measures, of concave distortions, and of

the relations between risk measures and Choquet integration.

It is a pleasure to express our thanks to all students and colleagues whose comments

have helped us to prepare this second edition, in particular to Dirk Becherer, Hans

Bhler, Rose-Anne Dana, Ulrich Horst, Mesrop Janunts, Christoph Khn, Maren

Liese, Harald Luschgy, Holger Pint, Philip Protter, Lothar Rogge, Stephan Sturm,

Stefan Weber, Wiebke Wittm, and Ching-Tang Wu. Special thanks are due to Peter

Bank and to Yuliya Mishura and Georgiy Shevchenko, our translators for the Russian edition. Finally, we thank Irene Zimmermann and Manfred Karbe of de Gruyter

Verlag for urging us to write a second edition and for their efficient support.

Berlin, September 2004

Hans Fllmer

Alexander Schied

graduate students in mathematics, and it may also be useful for mathematicians in

academia and in the financial industry. Our focus is on stochastic models in discrete

time. This limitation has two immediate benefits. First, the probabilistic machinery

is simpler, and we can discuss right away some of the key problems in the theory

of pricing and hedging of financial derivatives. Second, the paradigm of a complete

financial market, where all derivatives admit a perfect hedge, becomes the exception

rather than the rule. Thus, the discrete-time setting provides a shortcut to some of the

more recent literature on incomplete financial market models.

As a textbook for mathematicians, it is an introduction at an intermediate level, with

special emphasis on martingale methods. Since it does not use the continuous-time

methods of It calculus, it needs less preparation than more advanced texts such as

[99], [98], [107], [171], [252]. On the other hand, it is technically more demanding

than textbooks such as [215]: We work on general probability spaces, and so the text

captures the interplay between probability theory and functional analysis which has

been crucial for some of the recent advances in mathematical finance.

The book is based on our notes for first courses in Mathematical Finance which

both of us are teaching in Berlin at Humboldt University and at Technical University.

These courses are designed for students in mathematics with some background in

probability. Sometimes, they are given in parallel to a systematic course on stochastic

processes. At other times, martingale methods in discrete time are developed in the

course, as they are in this book. Usually the course is followed by a second course on

Mathematical Finance in continuous time. There it turns out to be useful that students

are already familiar with some of the key ideas of Mathematical Finance.

The core of this book is the dynamic arbitrage theory in the first chapters of Part II.

When teaching a course, we found it useful to explain some of the main arguments

in the more transparent one-period model before using them in the dynamical setting.

So one approach would be to start immediately in the multi-period framework of

Chapter 5, and to go back to selected sections of Part I as the need arises. As an

alternative, one could first focus on the one-period model, and then move on to Part II.

We include in Chapter 2 a brief introduction to the mathematical theory of expected

utility, even though this is a classical topic, and there is no shortage of excellent expositions; see, for instance, [187] which happens to be our favorite. We have three

reasons for including this chapter. Our focus in this book is on incompleteness, and

incompleteness involves, in one form or another, preferences in the face of risk and

uncertainty. We feel that mathematicians working in this area should be aware, at

viii

least to some extent, of the long line of thought which leads from Daniel Bernoulli via

von NeumannMorgenstern and Savage to some more recent developments which are

motivated by shortcomings of the classical paradigm. This is our first reason. Second,

the analysis of risk measures has emerged as a major topic in mathematical finance,

and this is closely related to a robust version of the Savage theory. Third, but not least,

our experience is that this part of the course was found particularly enjoyable, both by

the students and by ourselves.

We acknowledge our debt and express our thanks to all colleagues who have contributed, directly or indirectly, through their publications and through informal discussions, to our understanding of the topics discussed in this book. Ideas and methods

developed by Freddy Delbaen, Darrell Duffie, Nicole El Karoui, David Heath, Yuri

Kabanov, Ioannis Karatzas, Dimitri Kramkov, David Kreps, Stanley Pliska, Chris

Rogers, Steve Ross, Walter Schachermayer, Martin Schweizer, Dieter Sondermann

and Christophe Stricker play a key role in our exposition. We are obliged to many

others; for instance the textbooks [73], [99], [98], [155], and [192] were a great help

when we started to teach courses on the subject.

We are grateful to all those who read parts of the manuscript and made useful suggestions, in particular to Dirk Becherer, Ulrich Horst, Steffen Krger, Irina Penner,

and to Alexander Giese who designed some of the figures. Special thanks are due

to Peter Bank for a large number of constructive comments. We also express our

thanks to Erhan inlar, Adam Monahan, and Philip Protter for improving some of

the language, and to the Department of Operations Research and Financial Engineering at Princeton University for its hospitality during the weeks when we finished the

manuscript.

Berlin, June 2002

Hans Fllmer

Alexander Schied

Contents

vi

vii

Arbitrage theory

1.1 Assets, portfolios, and arbitrage opportunities . . .

1.2 Absence of arbitrage and martingale measures . . .

1.3 Derivative securities . . . . . . . . . . . . . . . . .

1.4 Complete market models . . . . . . . . . . . . . .

1.5 Geometric characterization of arbitrage-free models

1.6 Contingent initial data . . . . . . . . . . . . . . . .

1

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Preferences

2.1 Preference relations and their numerical representation

2.2 Von NeumannMorgenstern representation . . . . . .

2.3 Expected utility . . . . . . . . . . . . . . . . . . . . .

2.4 Uniform preferences . . . . . . . . . . . . . . . . . .

2.5 Robust preferences on asset profiles . . . . . . . . . .

2.6 Probability measures with given marginals . . . . . . .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

50

. 51

. 57

. 67

. 83

. 94

. 113

.

.

.

.

.

.

121

121

130

139

148

151

159

3.1 Portfolio optimization and the absence of arbitrage

3.2 Exponential utility and relative entropy . . . . . . .

3.3 Optimal contingent claims . . . . . . . . . . . . .

3.4 Optimal payoff profiles for uniform preferences . .

3.5 Robust utility maximization . . . . . . . . . . . .

3.6 Microeconomic equilibrium . . . . . . . . . . . .

.

.

.

.

.

.

3

3

7

16

27

33

37

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

175

4.1 Risk measures and their acceptance sets . . . . . . . . . . . . . . . . 176

4.2 Robust representation of convex risk measures . . . . . . . . . . . . . 186

4.3 Convex risk measures on L1 . . . . . . . . . . . . . . . . . . . . . . 199

Contents

4.4

4.5

4.6

4.7

4.8

4.9

II

5

Value at Risk . . . . . . . . . . . . . . . . . . . . . . .

Law-invariant risk measures . . . . . . . . . . . . . . .

Concave distortions . . . . . . . . . . . . . . . . . . . .

Comonotonic risk measures . . . . . . . . . . . . . . . .

Measures of risk in a financial market . . . . . . . . . .

Utility-based shortfall risk and divergence risk measures

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Dynamic hedging

259

5.1 The multi-period market model . . . . . . . . . .

5.2 Arbitrage opportunities and martingale measures

5.3 European contingent claims . . . . . . . . . . . .

5.4 Complete markets . . . . . . . . . . . . . . . . .

5.5 The binomial model . . . . . . . . . . . . . . . .

5.6 Exotic derivatives . . . . . . . . . . . . . . . . .

5.7 Convergence to the BlackScholes price . . . . .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

261

261

266

274

287

290

296

302

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

321

321

327

337

342

347

7 Superhedging

7.1 P -supermartingales . . . . . . . . . . . . . . . .

7.2 Uniform Doob decomposition . . . . . . . . . .

7.3 Superhedging of American and European claims

7.4 Superhedging with liquid options . . . . . . . . .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

354

354

356

359

368

6.1 Hedging strategies for the seller

6.2 Stopping strategies for the buyer

6.3 Arbitrage-free prices . . . . . .

6.4 Stability under pasting . . . . .

6.5 Lower and upper Snell envelopes

207

213

219

228

236

246

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Efficient hedging

380

8.1 Quantile hedging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380

8.2 Hedging with minimal shortfall risk . . . . . . . . . . . . . . . . . . 387

8.3 Efficient hedging with convex risk measures . . . . . . . . . . . . . . 396

9.1 Absence of arbitrage opportunities

9.2 Uniform Doob decomposition . .

9.3 Upper Snell envelopes . . . . . .

9.4 Superhedging and risk measures .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

404

404

412

417

424

xi

Contents

10.1 Local quadratic risk . . . . . . . . . . . . . . . . . . . . . . . . . . .

10.2 Minimal martingale measures . . . . . . . . . . . . . . . . . . . . . .

10.3 Variance-optimal hedging . . . . . . . . . . . . . . . . . . . . . . . .

428

428

438

449

456

11.1 Conditional risk measures and their robust representation . . . . . . . 456

11.2 Time consistency . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465

Appendix

A.1 Convexity . . . . . . . . . . . . . . . . . . . . . . . . .

A.2 Absolutely continuous probability measures . . . . . . .

A.3 Quantile functions . . . . . . . . . . . . . . . . . . . . .

A.4 The NeymanPearson lemma . . . . . . . . . . . . . . .

A.5 The essential supremum of a family of random variables

A.6 Spaces of measures . . . . . . . . . . . . . . . . . . . .

A.7 Some functional analysis . . . . . . . . . . . . . . . . .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

476

476

480

484

493

496

497

507

Notes

512

Bibliography

517

List of symbols

533

Index

535

Part I

Chapter 1

Arbitrage theory

a financial market. We consider a finite number of assets. Their initial prices at time

t D 0 are known, their future prices at time t D 1 are described as random variables

on some probability space. Trading takes place at time t D 0. Already in this simple

model, some basic principles of mathematical finance appear very clearly. In Section 1.2, we single out those models which satisfy a condition of market efficiency:

There are no trading opportunities which yield a profit without any downside risk.

The absence of such arbitrage opportunities is characterized by the existence of an

equivalent martingale measure. Under such a measure, discounted prices have the

martingale property, that is, trading in the assets is the same as playing a fair game.

As explained in Section 1.3, any equivalent martingale measure can be identified with

a pricing rule: It extends the given prices of the primary assets to a larger space of

contingent claims, or financial derivatives, without creating new arbitrage opportunities. In general, there will be several such extensions. A given contingent claim has

a unique price if and only if it admits a perfect hedge. In our one-period model, this

will be the exception rather than the rule. Thus, we are facing market incompleteness,

unless our model satisfies the very restrictive conditions discussed in Section 1.4. The

geometric structure of an arbitrage-free model is described in Section 1.5.

The one-period market model will be used throughout the first part of this book.

On the one hand, its structure is rich enough to illustrate some of the key ideas of the

field. On the other hand, it will provide an introduction to some of the mathematical methods which will be used in the dynamic hedging theory of the second part.

In fact, the multi-period situation considered in Chapter 5 can be regarded as a sequence of one-period models whose initial conditions are contingent on the outcomes

of previous periods. The techniques for dealing with such contingent initial data are

introduced in Section 1.6.

1.1

Consider a financial market with d C 1 assets. The assets can consist, for instance,

of equities, bonds, commodities, or currencies. In a simple one-period model, these

assets are priced at the initial time t D 0 and at the final time t D 1. We assume that

the i th asset is available at time 0 for a price i 0. The collection

D . 0 ; 1 ; : : : ; d / 2 RdCC1

is called a price system. Prices at time 1 are usually not known beforehand at time 0.

In order to model this uncertainty, we fix a measurable space .; F / and describe the

asset prices at time 1 as non-negative measurable functions

S 0; S 1; : : : ; S d

on .; F / with values in 0; 1/. Every ! 2 corresponds to a particular scenario

of market evolution, and S i .!/ is the price of the i th asset at time 1 if the scenario !

occurs.

However, not all asset prices in a market are necessarily uncertain. Usually there is

a riskless bond which will pay a sure amount at time 1. In our simple model for one

period, such a riskless investment opportunity will be included by assuming that

0 D 1 and

S0 1 C r

for a constant r, the return of a unit investment into the riskless bond. In most situations it would be natural to assume r 0, but for our purposes it is enough to require

that S 0 > 0, or equivalently that

r > 1:

In order to distinguish S 0 from the risky assets S 1 ; : : : ; S d , it will be convenient to

use the notation

S D .S 0 ; S 1 ; : : : ; S d / D .S 0 ; S /;

and in the same way we will write D .1; /.

At time t D 0, an investor will choose a portfolio

D . 0 ; / D . 0 ; 1 ; : : : ; d / 2 Rd C1 ;

where i represents the number of shares of the i th asset. The price for buying the

portfolio equals

d

X

D

i i :

i D0

S .!/ D

d

X

i S i .!/ D 0 .1 C r/ C S.!/;

i D0

depending on the scenario ! 2 . Here we assume implicitly that buying and selling

assets does not create extra costs, an assumption which may not be valid for a small

investor but which becomes more realistic for a large financial institution. Note our

convention of writing x y for the inner product of two vectors x and y in Euclidean

space.

this corresponds to taking out a loan such that we receive the amount j 0 j at t D 0

and pay back the amount .1 C r/j 0 j at time t D 1. If i < 0 for i 1, a quantity of

j i j shares of the i th asset is sold without actually owning them. This corresponds to

a short sale of the asset. In particular, an investor is allowed to take a short position

i < 0, and to use up the received amount i j i j for buying quantities j 0, j i,

of the other assets. In this case, the price of the portfolio D . 0 ; / is given by

D 0.

Remark 1.1. So far we have not assumed that anything is known about probabilities

that might govern the realization of the various scenarios ! 2 . Such a situation

is often referred to as Knightian uncertainty, in honor of F. Knight [176], who introduced the distinction between risk which refers to an economic situation in which

the probabilistic structure is assumed to be known, and uncertainty where no such

assumption is made.

}

Let us now assume that a probability measure P is given on .; F /. The asset

prices S 1 ; : : : ; S d and the portfolio values S can thus be regarded as random variables on .; F ; P /.

Definition 1.2. A portfolio 2 Rd C1 is called an arbitrage opportunity if 0

but S 0 P -a.s. and P S > 0 > 0.

Intuitively, an arbitrage opportunity is an investment strategy that yields with positive probability a positive profit and is not exposed to any downside risk. The existence

of such an arbitrage opportunity may be regarded as a market inefficiency in the sense

that certain assets are not priced in a reasonable way. In real-world markets, arbitrage

opportunities are rather hard to find. If such an opportunity would show up, it would

generate a large demand, prices would adjust, and the opportunity would disappear.

Later on, the absence of such arbitrage opportunities will be our key assumption. Absence of arbitrage implies that S i vanishes P -a.s. once i D 0. Hence, there is no

loss of generality if we assume from now on that

i > 0

for i D 1; : : : ; d .

Remark 1.3. Note that the probability measure P enters the definition of an arbitrage opportunity only through the null sets of P . In particular, the definition can be

formulated without any explicit use of probabilities if is countable. In this case,

we can simply apply Definition 1.2 with an arbitrary probability measure P such that

P ! > 0 for every ! 2 . Then an arbitrage opportunity is a portfolio with

0, with S .!/ 0 for all ! 2 , and such that S.!0 / > 0 for at least

}

one !0 2 .

The following lemma shows that absence of arbitrage is equivalent to the following

property of the market: Any investment in risky assets which yields with positive

probability a better result than investing the same amount in the risk-free asset must

be exposed to some downside risk.

Lemma 1.4. The following statements are equivalent.

(a) The market model admits an arbitrage opportunity.

(b) There is a vector 2 Rd such that

S .1 C r/ P -a.s.

Proof. To see that (a) implies (b), let be an arbitrage opportunity. Then 0 D

0 C . Hence,

S .1 C r/ S C .1 C r/ 0 D S:

Since S is P -a.s. non-negative and strictly positive with non-vanishing probability,

the same must be true of S .1 C r/ .

Next let be as in (b). We claim that the portfolio . 0 ; / with 0 WD is

an arbitrage opportunity. Indeed, D 0 C D 0 by definition. Moreover,

S D .1 C r/ C S, which is P -a.s. non-negative and strictly positive with

non-vanishing probability.

Exercise 1.1.1. On D !1 ; !2 ; !3 we fix a probability measure P with P !i >

0 for i D 1; 2; 3. Suppose that we have three assets with prices

0 1

1

D@2A

7

at time 0 and

0 1

1

S .!1 / D @ 3 A;

9

0 1

1

S.!2 / D @ 1 A;

5

1

1

S.!3 / D @ 5 A

10

Exercise 1.1.2. We consider a market model with a single risky asset defined on a

probability space with a finite sample space and a probability measure P that assigns strictly positive probability to each ! 2 . We let

a WD min S.!/

!2

and

b WD max S.!/:

!2

Show that the model does not admit arbitrage if and only if a < .1 C r/ < b.

Exercise 1.1.3. Show that the existence of an arbitrage opportunity implies the following seemingly stronger condition.

(a) There exists an arbitrage opportunity such that D 0.

Show furthermore that the following condition implies the existence of an arbitrage

opportunity.

(b) There exists 2 Rd C1 such that < 0 and S 0 P -a.s.

What can you say about the implication (a))(b)?

1.2

In this section, we are going to characterize those market models which do not admit

any arbitrage opportunities. Such models will be called arbitrage-free.

Definition 1.5. A probability measure P is called a risk-neutral measure, or a martingale measure, if

Si

; i D 0; 1; : : : ; d:

(1.1)

i D E

1Cr

Remark 1.6. In (1.1), the price of the i th asset is identified as the expectation of

the discounted payoff under the measure P . Thus, the pricing formula (1.1) can

be seen as a classical valuation formula which does not take into account any risk

aversion, in contrast to valuations in terms of expected utility which will be discussed

in Section 2.3. This is why a measure P satisfying (1.1) is called risk-neutral. The

connection to martingales will be made explicit in Section 1.6.

}

The following basic result is sometimes called the fundamental theorem of asset

pricing or, in short, FTAP. It characterizes arbitrage-free market models in terms of

the set

of risk-neutral measures which are equivalent to P . Recall that two probability measures P and P are said to be equivalent (P P ) if, for A 2 F , P A D 0 if

and only if P A D 0. This holds if and only if P has a strictly positive density

dP =dP with respect to P ; see Appendix A.2. An equivalent risk-neutral measure is

also called a pricing measure or an equivalent martingale measure.

Theorem 1.7. A market model is arbitrage-free if and only if P ;. In this case,

there exists a P 2 P which has a bounded density dP =dP .

We show first that the existence of a risk-neutral measure implies the absence of

arbitrage.

Proof of the implication ( of Theorem 1:7. Suppose that there exists a risk-neutral measure P 2 P . Take a portfolio 2 Rd C1 such that S 0 P -a.s.

and E S > 0. Both properties remain valid if we replace P by the equivalent

measure P . Hence,

D

d

X

i i

D

i D0

d

X

i D0

i S i

1Cr

DE

S

1Cr

> 0:

For the proof of the implication ) of Theorem 1.7, it will be convenient to introduce the random vector Y D .Y 1 ; : : : ; Y d / of discounted net gains:

Y i WD

Si

i ;

1Cr

i D 1; : : : ; d:

(1.2)

With this notation, Lemma 1.4 implies that the absence of arbitrage is equivalent to

the following condition:

For 2 Rd :

Y 0 P -a.s. H) Y D 0 P -a.s.

(1.3)

measure P is well-defined, and so P is a risk-neutral measure if and only if

E Y D 0:

(1.4)

E Y i , i D 1; : : : ; d . The assertion of Theorem 1.7 can now be read as follows:

Condition (1.3) holds if and only if there exists some P P such that E Y D 0,

and in this case, P can be chosen such that the density dP =dP is bounded.

Proof of the implication ) of Theorem 1:7. We have to show that (1.3) implies the

existence of some P P such that (1.4) holds and such that the density dP =dP

is bounded. We will do this first in the case in which

E jY j < 1:

Let Q denote the convex set of all probability measures Q P with bounded

densities dQ=dP , and denote by EQ Y the d -dimensional vector with components

EQ Y i , i D 1; : : : ; d . Due to our assumption jY j 2 L1 .P /, all these expectations

are finite. Let

C WD EQ Y j Q 2 Q;

Q WD Q1 C .1 /Q0 2 Q and

EQ1 Y C .1 /EQ0 Y D EQ Y ;

which lies in C .

Our aim is to show that C contains the origin. To this end, we suppose by way of

contradiction that 0 C . Using the separating hyperplane theorem in the elementary form of Proposition A.1, we obtain a vector 2 Rd such that x 0 for all

x 2 C , and such that x0 > 0 for some x0 2 C . Thus, satisfies EQ Y 0 for

all Q 2 Q and EQ0 Y > 0 for some Q0 2 Q. Clearly, the latter condition yields

that P Y > 0 > 0. We claim that the first condition implies that Y is P -a.s.

non-negative. This fact will be a contradiction to our assumption (1.3) and thus will

prove that 0 2 C .

To prove the claim that Y 0 P -a.s., let A WD Y < 0, and define functions

1

1

IA C IAc :

'n WD 1

n

n

We take 'n as densities for new probability measures Qn :

1

dQn

WD

'n ;

dP

E 'n

n D 2; 3; : : : :

0 EQn Y D

1

E Y 'n :

E 'n

E Y IY <0 D lim E Y 'n 0:

n"1

This proves the claim that Y 0 P -a.s. and completes the proof of Theorem 1.7

in case E jY j < 1.

If Y is not P -integrable, then we simply replace the probability measure P by

a suitable equivalent measure PQ whose density d PQ =dP is bounded and for which

Q jY j < 1. For instance, one can define PQ by

E

c

d PQ

D

dP

1 C jY j

for c WD E

1

1 C jY j

1

:

Recall from Remark 1.3 that replacing P with an equivalent probability measure does

not affect the absence of arbitrage opportunities in our market model. Thus, the first

10

part of this proof yields a risk-neutral measure P which is equivalent to PQ and whose

density dP =d PQ is bounded. Then P 2 P , and

dP

dP d PQ

D

dP

d PQ dP

is bounded. Hence, P is as desired, and the theorem is proved.

Remark 1.8. Note that neither the absence of arbitrage nor the definition of the class

P involve the full structure of the probability measure P , they only depend on the

class of nullsets of P . In particular, the preceding theorem can be formulated in a

situation of Knightian uncertainty, i.e., without fixing any initial probability measure

P , whenever the underlying set is countable.

}

Remark 1.9. Our assumption that asset prices S are non-negative implies that the

components of Y are bounded from below. Note however that this assumption was

not needed in our proof. Thus, Theorem 1.7 also holds if we only assume that S is

finite-valued and 2 Rd . In this case, the definition of a risk-neutral measure P

via (1.1) is meant to include the assumption that S i is integrable with respect to P

for i D 1; : : : ; d .

}

Example 1.10. Let P be any probability measure on the finite set WD !1 ; : : : ; !N

that assigns strictly positive probability pi to each singleton !i . Suppose that there

is a single risky asset defined by its price D 1 at time 0 and by the random variable

S D S 1 . We may assume without loss of generality that the values si WD S.!i / are

distinct and arranged in increasing order: s1 < < sN . According to Theorem 1.7,

this model does not admit arbitrage opportunities if and only if

Q S j PQ P D

.1 C r/ 2 E

N

X

pQi D 1 D .s1 ; sN /;

i D1

i D1

solve the linear equations

s1 p1 C C sN pN

D .1 C r/;

p1 C C pN

D 1:

If a solution exists, it will be unique if and only if N D 2, and there will be infinitely

many solutions for N > 2.

}

Exercise 1.2.1. On D !1 ; !2 ; !3 we fix a probability measure P with P !i >

0 for i D 1; 2; 3. Suppose that we have three assets with prices

0 1

1

@

D 2A

7

11

at time 0 and

0 1

1

S .!1 / D @ 3 A;

9

0 1

1

S.!2 / D @ 1 A;

5

1

1

S.!3 / D @ 5 A

13

at time 1. Show that this market model does not admit arbitrage and find all riskneutral measures. Note that this model differs from the one in Exercise 1.1.1 only in

}

the value of S 2 .!3 /.

Exercise 1.2.2. Consider a market model with one risky asset that is such that 1 > 0

and the distribution of S 1 has a strictly

positive density function f W .0; 1/ !

Rx

1 x D

.0; 1/. That is, P S

0 f .y/ dy for x > 0. Find an equivalent risk}

neutral measure P .

Remark 1.11. The economic reason for working with the discounted asset prices

X i WD

Si

;

1Cr

i D 0; : : : ; d;

(1.5)

is that one should distinguish between one unit of a currency (e.g. C) at time t D 0

and one unit at time t D 1. Usually people tend to prefer a certain amount today over

the same amount which is promised to be paid at a later time. Such a preference is

reflected in an interest r > 0 paid by the riskless bond: Only the amount 1=.1 C r/ C

must be invested at time 0 to obtain 1 C at time 1. This effect is sometimes referred to

as the time value of money. Similarly, the price S i of the i th asset is quoted in terms

of C at time 1, while i corresponds to time-zero euros. Thus, in order to compare

the two prices i and S i , one should first convert them to a common standard. This is

achieved by taking the riskless bond as a numraire and by considering the discounted

prices in (1.5).

}

Remark 1.12. One can choose as numraire any asset which is strictly positive. For

instance, suppose that 1 > 0 and P S 1 > 0 D 1. Then all asset prices can be

expressed in units of the first asset by considering

e

i WD

i

1

and

Si

;

S1

i D 0; : : : ; d:

a particular numraire. Thus, an arbitrage-free market model should admit a riskneutral measure with respect to the new numraire, i.e., a probability measure PQ

P such that

"

#

i

S

i

Q D EQ

; i D 0; : : : ; d:

S1

12

d PQ

S1

D

2

P

:

for

some

P

PQ D PQ

dP

E S 1

Indeed, if PQ lies in the set on the right, then

i

E S i

i

S

Q

D

D Q i ;

D

E

S1

E S 1

1

and so PQ 2 PQ . Reversing the roles of PQ and P then yields the identity of the two

sets. Note that

P \ PQ D ;

as soon as S 1 is not P -a.s. constant, because Jensens inequality then implies that

h

i

1

0

Q 1 C r > 1 C r

D

Q

D

E

1

S1

EQ S 1

and hence EQ S 1 > E S 1 for all PQ 2 PQ and P 2 P .

Let

V WD S j 2 Rd C1

denote the linear space of all payoffs which can be generated by some portfolio. An

element of V will be called an attainable payoff. The portfolio that generates V 2 V

is in general not unique, but we have the following law of one price.

Lemma 1.13. Suppose that the market model is arbitrage-free and that V 2 V can

be written as V D S D S P -a.s. for two different portfolios and . Then

D .

Proof. We have . / S D 0 P -a.s. for any P 2 P . Hence,

DE

. / S

1Cr

D 0;

due to (1.1).

By the preceding lemma, it makes sense to define the price of V 2 V as

.V / WD

if V D S,

(1.6)

13

Remark 1.14. Via (1.6), the price system can be regarded as a linear form on the

finite-dimensional vector space V . For any P 2 P we have

.V / D E

V i

;

1Cr

V 2 V:

larger space L1 .P / of P -integrable random variables. Since this space is usually infinite-dimensional, one cannot expect that such a pricing measure is in general

unique; see however Section 1.4.

}

We have seen above that, in an arbitrage-free market model, the condition S D 0

P -a.s. implies that D 0. In fact, one may assume without loss of generality that

S D 0 P -a.s.

H)

D 0;

(1.7)

for otherwise we can find i 2 0; : : : ; d such that i 0 and represent the i th asset

as a linear combination of the remaining ones

i D

1 X j j

i

and

Si D

j i

1 X j j

S :

i

j i

Definition 1.15. The market model is called non-redundant if (1.7) holds.

Exercise 1.2.3. Show that in a non-redundant market model the components of the

vector Y of discounted net gains are linearly independent in the sense that

Y D 0 P -a.s.

H)

D 0:

(1.8)

Show then that condition (1.8) implies non-redundance if the market model is arbitrage-free.

}

Exercise 1.2.4. Show that in a non-redundant and arbitrage-free market model the set

2 Rd C1 j D w and S 0 P -a.s.

is compact for any w > 0.

Definition 1.16. Suppose that the market model is arbitrage-free and that V 2 V is

an attainable payoff such that .V / 0. Then the return of V is defined by

R.V / WD

V .V /

:

.V /

14

Note that we have already seen the special case of the risk-free return

rD

S0 0

D R.S 0 /:

0

tainable payoffs Vk , then

R.V / D

n

X

k R.Vk /

kD1

Pn

kD1 k Vk

of non-zero at-

.Vk /

for k D Pn k

:

i D1 i .Vi /

Vk . As a particular case of the formula above, we have that

R.V / D

d

X

i i

i D0

R.S i /

for all non-zero attainable payoffs V D S (recall that we have assumed that all i

are strictly positive).

Proposition 1.17. Suppose that the market model is arbitrage-free, and let V 2 V be

an attainable payoff such that .V / 0.

(a) Under any risk-neutral measure P , the expected return of V is equal to the

risk-free return r

E R.V / D r:

(b) Under any measure Q P such that EQ jSj < 1, the expected return of V

is given by

dP

; R.V / ;

EQ R.V / D r cov

dQ

Q

where P is an arbitrary risk-neutral measure in P and covQ denotes the covariance with respect to Q.

Proof. (a): Since E V D .V /.1 C r/, we have

E R.V / D

E V .V /

D r:

.V /

cov.' ; R.V // D EQ ' R.V / EQ ' EQ R.V /

Q

D E R.V / EQ R.V / :

Using part (a) yields the assertion.

15

Theorem 1.7 to market models with an infinity of tradable assets S 0 ; S 1 ; S 2 ; : : : . We

assume that S 0 1 C r for some r > 1 and that the random vector

S.!/ D .S 1 .!/; S 2 .!/; : : : /

takes values in the space `1 of bounded real sequences. This space is a Banach space

with respect to the norm

kxk1 WD sup jx i j

for x D .x 1 ; x 2 ; : : : / 2 `1 .

i 1

A portfolio D . 0P

; / is chosen in such a way that D . 1 ; 2 ; : : : / is a sequence

1

1 , i.e.,

i

in the space `

i D1 j j < 1. We assume that the corresponding price system

0

D . ; / satisfies 2 `1 and 0 D 1. Clearly, this model class includes our

model with d C 1 traded assets as a special case.

Our first observation is that the implication ( of Theorem 1.7 remains valid, i.e.,

the existence of a measure P P with the properties

E kS k1 < 1

and

Si

1Cr

D i

implies the absence of arbitrage opportunities. To this end, suppose that is a portfolio

strategy such that

S 0 P -a.s. and E S > 0:

(1.9)

Then we can replace P in (1.9) by the equivalent measure P . Hence, cannot be an

arbitrage opportunity since

D

1

X

i D0

E

Si

1Cr

DE

S

1Cr

> 0:

Note that interchanging summation and integration is justified by dominated convergence, because

1

X

j i j 2 L1 .P /:

j 0 j C kS k1

i D0

The following example shows that the implication ) of Theorem 1.7, namely that

absence of arbitrage opportunities implies the existence of a risk-neutral measure,

may no longer be true for an infinite number of assets.

}

Example 1.19. Let D 1; 2; : : : , and choose any probability measure P which

assigns strictly positive probability to all singletons !. We take r D 0 and define a

16

i D 1; 2; : : : , by

8

<0 if ! D i,

i

S .!/ D 2 if ! D i C 1,

:

1 otherwise.

Let us show that this market model is arbitrage-free. To this end, suppose that D

. 0 ; / is a portfolio such that 2 `1 and such that S.!/ 0 for each ! 2 , but

such that 0. Considering the case ! D 1 yields

0

0 S .1/ D C

1

X

k D 1 1 :

kD2

0

0 S.!/ D C 2

i 1

1

X

k D C i 1 i i 1 i :

kD1

ki;i 1

It follows that 0 1 2 . But this can only be true if all i vanish, since we

have assumed that 2 `1 . Hence, there are no arbitrage opportunities.

However, there exists no probability measure P P such that E S i D i for

all i . Such a measure P would have to satisfy

1 D E S i D 2P i C 1 C

1

X

P k

kD1

ki;i C1

D 1 C P i C 1 P i

for i > 1. This relation implies that P i D P i C 1 for all i > 1, contra}

dicting the assumption that P is a probability measure and equivalent to P .

1.3

Derivative securities

In real financial markets, not only the primary assets are traded. There is also a large

variety of securities whose payoff depends in a non-linear way on the primary assets

S 0 ; S 1 ; : : : ; S d , and sometimes also on other factors. Such financial instruments are

usually called options, contingent claims, derivative securities, or just derivatives.

Example 1.20. Under a forward contract, one agent agrees to sell to another agent an

asset at time 1 for a price K which is specified at time 0. Thus, the owner of a forward

contract on the i th asset gains the difference between the actual market price S i and

17

the delivery price K if S i is larger than K at time 1. If S i < K, the owner loses

the amount K S i to the issuer of the forward contract. Hence, a forward contract

corresponds to the random payoff

C fw D S i K:

Example 1.21. The owner of a call option on the i th asset has the right, but not the

obligation, to buy the i th asset at time 1 for a fixed price K, called the strike price.

This corresponds to a payoff of the form

S i K if S i > K,

call

i

C

D .S K/ D

C

0

otherwise.

Conversely, a put option gives the right, but not the obligation, to sell the asset at time

1 for a strike price K. The corresponding random payoff is given by

K S i if S i < K,

put

i C

C D .K S / D

0

otherwise.

Call and put options with the same strike K are related through the formula

C call C put D S i K:

Hence, if the price .C call / of a call option has already been fixed, then the price

.C put / of the corresponding put option is determined by linearity through the putcall parity

K

:

(1.10)

.C call / D .C put / C i

1Cr

}

Example 1.22. An option on the value V D S of a portfolio of several risky assets

is sometimes called a basket or index option. For instance, a basket call would be of

the form .V K/C . The asset on which the option is written is called the underlying

asset or just the underlying.

}

Put and call options can be used as building blocks for a large class of derivatives.

Example 1.23. A straddle is a combination of at-the-money put and call options

on a portfolio V D S , i.e., on put and call options with strike K D .V /:

C D ..V / V /C C .V .V //C D jV .V /j:

Thus, the payoff of the straddle increases proportionally to the change of the price of

between time 0 and time 1. In this sense, a straddle is a bet that the portfolio price

will move, no matter in which direction.

}

18

C D .K jV .V /j/C ;

where K > 0 and where V D S is the price of a given portfolio or the value of

a stock index. Clearly, the payoff of the butterfly spread is maximal if V D .V /

and decreases if the price at time 1 of the portfolio deviates from its price at time 0.

Thus, the butterfly spread is a bet that the portfolio price will stay close to its present

value.

}

Exercise 1.3.1. Draw the payoffs of put and call options, a straddle, and a butterfly

spread as functions of its underlying.

}

Exercise 1.3.2. Consider a butterfly spread as in Example 1.24 and write its payoff

as a combination of

(a) call options,

(b) put options

on the underlying. As in the put-call parity (1.10), such a decomposition determines

the price of a butterfly spread once the prices of the corresponding put or call options

have been fixed.

}

Example 1.25. The idea of portfolio insurance is to increase exposure to rising asset

prices, and to reduce exposure to falling prices. This suggests to replace the payoff

V D S of a given portfolio by a modified profile h.V /, where h is convex and

increasing. Let us first consider the case where V 0. Then the corresponding

payoff h.V / can be expressed as a combination of investments in bonds, in V itself,

and in basket call options on V . To see this, recall that convexity implies that

Z x

h.x/ D h.0/ C

h0 .y/ dy

0

h0

WD h0C of h; see Appendix A.1. By

the arguments in Lemma A.19, the increasing rightcontinuous function h0 can be

represented as the distribution function of a positive Radon measure on 0; 1/:

h0 .x/ D .0; x/ for x 0. Recall that a positive Radon measure is a -additive

measure that assigns to each Borel set A 0; 1/ a value in .A/ 2 0; 1, which is

finite when A is compact. An example is the Lebesgue measure on 0; 1/. Using the

representation h0 .x/ D .0; x/, Fubinis theorem implies that

Z xZ

h.x/ D h.0/ C

.dz/ dy

0

0;y

Z

dy .dz/:

D h.0/ C .0/ x C

.0;1/

y j zyx

19

Z

0

.V K/C .dK/:

h.V / D h.0/ C h .0/ V C

.0;1/

(1.11)

}

in V D S itself, and in call options on V . It requires, however, that V is nonnegative

and that both h.0/ and h0 .0/ are finite. Also, it is sometimes more convenient to have

a development around the initial value V WD of the portfolio than to have a

development around zero. Corresponding extensions of formula (1.11) are explored

in the following exercise.

Exercise 1.3.3. In this exercise, we consider the situation of Example 1.25 without

insisting that the payoff V D S takes only nonnegative values. In particular, the

portfolio may also contain short positions. Let h W R ! R be a continuous function.

(a) Show that for convex h there exists a nonnegative Radon measure on R such

that the payoff h.V / can be realized by holding bonds, forward contracts, and a

mixture of call and put options on V :

h.V / D h. V / C h0 . V /.V V /

Z

Z

C

.V K/C .dK/ C

.1; V

.K V /C .dK/:

. V

;1/

Note that the put and call options occurring in this formula are out of the

money in the sense that their intrinsic value, i.e., their value when V is replaced by its present value V , is zero.

(b) Now let h be any twice continuously differentiable function on R. Deduce from

part (a) that

h.V / D h. V / C h0 . V /.V V /

Z

V

C 00

.V K/ h .K/ dK C

1

V

Example 1.26. A reverse convertible bond pays interest which is higher than that

earned by an investment into the riskless bond. But at maturity t D 1, the issuer may

convert the bond into a predetermined number of shares of a given asset S i instead

of paying the nominal value in cash. The purchase of this contract is equivalent to

the purchase of a standard bond and the sale of a certain put option. More precisely,

20

suppose that 1 is the price of the reverse convertible bond at t D 0, that its nominal

value at maturity is 1 C r,

Q and that it can be converted into x shares of the i th asset.

Q

Thus, the

This conversion will happen if the asset price S i is below K WD .1 C r/=x.

payoff of the reverse convertible bond is equal to

1 C rQ x.K S i /C ;

i.e., the purchase of this contract is equivalent to a risk-free investment of the amount

.1 C r/=.1

Q

C r/ with interest r and the sale of the put option x.K S i /C for the price

.rQ r/=.1 C r/.

}

Example 1.27. A discount certificate on V D S pays off the amount

C D V ^ K;

where the number K > 0 is often called the cap. Since

C D V .V K/C ;

buying the discount certificate is the same as purchasing and selling the basket call

option C call WD .V K/C . If the price .C call / has already been fixed, then the price

of C is given by .C / D .V / .C call /. Hence, the discount certificate is less

expensive than the portfolio itself, and this explains the name. On the other hand, it

}

participates in gains of only up to the cap K.

Example 1.28. For an insurance company, it may be desirable to shift some of its

insurance risk to the financial market. As an example of such an alternative risk

transfer, consider a catastrophe bond issued by an insurance company. The interest

paid by this security depends on the occurrence of certain special events. For instance,

the contract may specify that no interest will be paid if more than a given number of

insured cars are damaged by hail on a single day during the lifetime of the contract; as

a compensation for taking this risk, the buyer will be paid an interest above the usual

market rate if this event does not occur.

}

Mathematically, it will be convenient to focus on contingent claims whose payoff is

non-negative. Such a contingent claim will be interpreted as a contract which is sold

at time 0 and which pays a random amount C.!/ 0 at time 1. A derivative security

whose terminal value may also become negative can usually be reduced to a combination of a non-negative contingent claim and a short position in some of the primary

assets S 0 ; S 1 ; : : : ; S d . For instance, the terminal value of a reverse convertible bond

is bounded from below so that it can be decomposed into a short position in cash and

into a contract with positive value. From now on, we will work with the following

formal definition of the term contingent claim.

21

Definition 1.29. A contingent claim is a random variable C on the underlying probability space .; F ; P / such that

0C <1

P -a.s.

measurable with respect to the -field .S 0 ; : : : ; S d / generated by the assets, i.e., if

C D f .S 0 ; : : : ; S d /

for a measurable function f on Rd C1 .

So far, we have only fixed the prices i of our primary assets S i . Thus, it is not

clear what the correct price should be for a general contingent claim C . Our main

goal in this section is to identify those possible prices which are compatible with the

given prices in the sense that they do not generate arbitrage. Our approach is based

on the observation that trading C at time 0 for a price C corresponds to introducing

a new asset with the prices

d C1 WD C

and

S d C1 WD C:

(1.12)

Definition 1.30. A real number C 0 is called an arbitrage-free price of a contingent claim C if the market model extended according to (1.12) is arbitrage-free. The

set of all arbitrage-free prices for C is denoted .C /.

In the previous definition, we made the implicit assumption that the introduction

of a contingent claim C as a new asset does not affect the prices of primary assets.

This assumption is reasonable as long as the trading volume of C is small compared

to that of the primary assets. In Section 3.6 we will discuss the equilibrium approach

to asset pricing, where an extension of the market will typically change the prices of

all traded assets.

The following result shows in particular that we can always find an arbitrage-free

price for a given contingent claim C if the initial model is arbitrage-free.

Theorem 1.31. Suppose that the set P of equivalent risk-neutral measures for the

original market model is non-empty. Then the set of arbitrage-free prices of a contingent claim C is non-empty and given by

.C / D E

C

1Cr

P 2 P such that E C < 1 :

(1.13)

22

Proof. By Theorem 1.7, C is an arbitrage-free price for C if and only if there exists

an equivalent risk-neutral measure PO for the market model extended via (1.12), i.e.,

i

D EO

Si

1Cr

for i D 1; : : : ; d C 1.

Conversely, if C D E C =.1 C r/ for some P 2 P , then this P is also an

equivalent risk-neutral measure for the extended market model, and so the two sets in

(1.13) are equal.

To show that .C / is non-empty, we first fix some measure PQ P such that

Q

E C < 1. For instance, we can take d PQ D c.1 C C /1 dP , where c is the normalizing constant. Under PQ , the market model is arbitrage-free. Hence, Theorem 1.7

yields P 2 P such that dP =d PQ is bounded. In particular, E C < 1 and

C D E C =.1 C r/ 2 .C /.

Exercise 1.3.4. Show that the set .C / of arbitrage-free prices of a contingent claim

is convex and hence an interval.

}

The following theorem provides a dual characterization of the lower and upper

bounds

inf .C / WD inf .C / and sup .C / WD sup .C /;

which are often called arbitrage bounds for C .

Theorem 1.32. In an arbitrage-free market model, the arbitrage bounds of a contingent claim C are given by

C i

1Cr

P 2P

D max m 2 0; 1/ 9 2 Rd with m C Y

inf .C / D inf E

C

P -a.s.

1Cr

(1.14)

and

C i

1Cr

P 2P

D min m 2 0; 1 9 2 Rd with m C Y

sup .C / D sup E

C

P -a.s. :

1Cr

Proof. We only prove the identities for the upper arbitrage bound. The ones for the

lower bound are obtained in a similar manner; see Exercise 1.3.5. We take m 2 0; 1

and 2 Rd such that m C Y C =.1 C r/ P -a.s., and we denote by M the set of

23

get

C i

1Cr

P 2P

h C i

P 2 P ; E C < 1 D sup .C /;

sup E

1Cr

inf M sup E

(1.15)

Next we show that all inequalities in (1.15) are in fact identities. This is trivial if

sup .C / D 1. For sup .C / < 1, we will show that m > sup .C / implies m

inf M . By definition, sup .C / < m < 1 requires the existence of an arbitrage

opportunity in the market model extended by d C1 WD m and S d C1 WD C . That is,

there is .; d C1 / 2 Rd C1 such that Y C d C1 .C =.1 C r/ m/ is almost-surely

non-negative and strictly positive with positive probability. Since the original market

model is arbitrage-free, d C1 must be non-zero. In fact, we have d C1 < 0 as taking

expectations with respect to P 2 P for which E C < 1 yields

h

d C1 E

C i

m 0;

1Cr

and the term in parenthesis is negative since m > sup .C /. Thus, we may define

WD = d C1 2 Rd and obtain m C Y C =.1 C r/ P -a.s., hence m inf M .

We now prove that inf M belongs to M . To this end, we may assume without loss of

generality that inf M < 1 and that the market model is non-redundant in the sense of

Definition 1.15. For a sequence mn 2 M that decreases towards inf M D sup .C /, we

fix n 2 Rd such that mn Cn Y C =.1Cr/ P -almost surely. If lim infn jn j < 1,

there exists a subsequence of .n / that converges to some 2 Rd . Passing to the

limit yields sup .C / C Y C =.1 C r/ P -a.s., which gives sup .C / 2 M . But

this is already the desired result, since the following argument will show that the

case lim infn jn j D 1 cannot occur. Indeed, after passing to some subsequence

if necessary, n WD n =jn j converges to some 2 Rd with jj D 1. Under the

assumption that jn j ! 1, passing to the limit in

C

mn

C n Y

jn j

jn j.1 C r/

P -a.s.

whence D 0 by non-redundance of the model. But this contradicts the fact that

jj D 1.

Exercise 1.3.5. Prove the identity (1.14).

24

Remark 1.33. Theorem 1.32 shows that sup .C / is the lowest possible price of a

portfolio with

S C P -a.s.

Such a portfolio is often called a superhedging strategy or superreplication of C ,

and the identities for inf .C / and sup .C / obtained in Theorem 1.32 are often called

superhedging duality relations. When using , the seller of C would be protected

against any possible future claims of the buyer of C . Thus, a natural goal for the seller

would be to finance such a superhedging strategy from the proceeds of C . Conversely,

the objective of the buyer would be to cover the price of C from the sale of a portfolio

with

S C P -a.s.,

which is possible if and only if inf .C /. Unless C is an attainable payoff,

however, neither objective can be fulfilled by trading C at an arbitrage-free price, as

shown in Corollary 1.35 below. Thus, any arbitrage-free price involves a trade-off

between these two objectives.

}

For a portfolio the resulting payoff V D S, if positive, may be viewed as

a contingent claim, and in particular as a derivative. Those claims which can be

replicated by a suitable portfolio will play a special role in the sequel.

Definition 1.34. A contingent claim C is called attainable .replicable, redundant/ if

C D S P -a.s. for some 2 Rd C1 . Such a portfolio strategy is then called a

replicating portfolio for C .

If one can show that a given contingent claim C can be replicated by some portfolio

, then the problem of determining a price for C has a straightforward solution: The

price of C is unique and equal to the cost of its replication, due to the law of one

price. The following corollary also shows that the attainable contingent claims are in

fact the only ones which admit a unique arbitrage-free price.

Corollary 1.35. Suppose the market model is arbitrage-free and C is a contingent

claim.

(a) C is attainable if and only if it admits a unique arbitrage-free price.

(b) If C is not attainable, then inf .C / < sup .C / and

.C / D .inf .C /; sup .C //:

Proof. To prove part (a), note first that j.C /j D 1 if C is attainable. The converse

implication will follow from (b).

In order to prove part (b), note first that .C / is an interval due to Exercise 1.3.4.

To show that this interval is open, it suffices to exclude the possibility that it contains

25

one of its boundary points inf .C / and sup .C /. To this end, we use Theorem 1.32 to

get 2 Rd such that

inf .C / C Y

C

1Cr

P -a.s.

with 0 WD inf .C /, the strategy . 0 ; ; 1/ 2 Rd C2 is an arbitrage opportunity in the market model extended by d C1 WD inf .C / and S d C1 WD C . Therefore

inf .C / is not an arbitrage-free price for C . The possibility sup .C / 2 .C / is excluded by a similar argument.

Remark 1.36. In Theorem 1.32, the set P of equivalent risk-neutral measures can be

replaced by the set PQ of risk-neutral measures that are merely absolutely continuous

with respect to P . That is,

inf .C / D inf EQ

PQ 2PQ

C i

1Cr

and

sup .C / D sup EQ

PQ 2PQ

C i

;

1Cr

(1.16)

for any contingent claim C . To prove this, note first that P PQ , so that we get the

two inequalities and in (1.16). On the other hand, for PQ 2 PQ , P 2 P with

E C < 1, and " 2 .0; 1, the measure P" WD "P C .1 "/PQ belongs to P

Q C . Sending " # 0 yields the converse

and satisfies E" C D "E C C .1 "/E

inequalities.

}

Remark 1.37. Consider any arbitrage-free market model, and let C call D .S i K/C

be a call option on the i th asset with strike K > 0. Clearly, C call S i so that

call

C

E

i

1Cr

for any P 2 P . From Jensens inequality, we obtain the following lower bound:

E

C call

1Cr

C

C

Si

K

K

i

E

D

:

1Cr

1Cr

1Cr

Thus, the following universal bounds hold for any arbitrage-free market model:

i

K

1Cr

C

(1.17)

K

i

1Cr

C

K

:

1Cr

(1.18)

26

If r 0, then the lower bound in (1.17) can be further reduced to inf .C call /

. i K/C . Informally, this inequality states that the value of the right to buy the

i th asset at t D 0 for a price K is strictly less than any arbitrage-free price for C call .

This fact is sometimes expressed by saying that the time value of a call option is

non-negative. The quantity . i K/C is called the intrinsic value of the call option.

Observe that an analogue of this relation usually fails for put options: The left-hand

side of (1.18) can only be bounded by its intrinsic value .K i /C if r 0. If the

intrinsic value of a put or call option is positive, then one says that the option is in

the money. For i D K one speaks of an at-the-money option. Otherwise, the

option is out of the money.

}

In many situations, the universal arbitrage bounds (1.17) and (1.18) are in fact attained, as illustrated by the following example.

Example 1.38. Take any market model with a single risky asset S D S 1 such that

the distribution of S under P is concentrated on 0; 1; : : : with positive weights.

Without loss of generality, we may assume that S has under P a Poisson distribution

with parameter 1, i.e., S is P -a.s. integer-valued and

PS D k D

e 1

k

for k D 0; 1; : : : .

is arbitrage-free. We are going to show that the upper and lower bounds in (1.17)

are attained for this model by using Remark 1.36. To this end, consider the measure

PQ 2 PQ which is defined by its density

d PQ

D e ISD1 :

dP

We get

Q .S K/C D .1 K/C D . K/C ;

E

so that the lower bound in (1.17) is attained, i.e., we have

inf ..S K/C / D . K/C :

To see that also the upper bound is sharp, we define

e

gn .k/ WD e

I0 .k/ C .n 1/ e In .k/;

n

It is straightforward to check that

d PQn WD gn .S/ dP

k D 0; 1; : : :

27

K C

C

Q

:

En .S K/ D 1

n

By sending n " 1, we see that also the upper bound in (1.17) is attained

sup ..S K/C / D :

Furthermore, the put-call parity (1.10) shows that the universal bounds (1.18) for put

options are attained as well.

}

Exercise 1.3.6. We consider the market model from Exercise 1.1.2 and suppose that

a < .1 C r/ < b so that the model is arbitrage-free. Let C be a derivative that is

given by C D h.S/, where h 0 is a convex function. Show that

sup .C / D

h.b/ .1 C r/ a

h.a/ b .1 C r/

C

:

1Cr

ba

1Cr

ba

is given by C D h.S 1 /, where h 0 is a convex function. Derive the following

arbitrage bounds for C :

inf .C /

1.4

h. 1 .1 C r//

1Cr

and

sup .C /

h.x/ 1

h.0/

C lim

:

1 C r x"1 x

Our goal in this section is to characterize the particularly transparent situation in which

all contingent claims are attainable.

Definition 1.39. An arbitrage-free market model is called complete if every contingent claim is attainable.

The following theorem characterizes the class of all complete market models. It is

sometimes called the second fundamental theorem of asset pricing.

Theorem 1.40. An arbitrage-free market model is complete if and only if there exists

exactly one risk-neutral probability measure, i.e., if jP j D 1.

Proof. If the model is complete, then the indicator IA of each set A 2 F is an attainable contingent claim. Hence, Corollary 1.35 implies that P A D E IA

is independent of P 2 P . Consequently, there is just one risk-neutral probability

measure.

28

states that the set .C / of arbitrage-free prices is non-empty and given by

C

.C / D E

P 2 P such that E C < 1 :

1Cr

Since P has just one element, the same must hold for .C /. Hence, Corollary 1.35

implies that C is attainable.

We will now show that every complete market model has a finite structure and can

be reduced to a finite probability space. To this end, observe first that in every market

model the following inclusion holds for each P 2 P :

V D S j 2 Rd C1 L1 .; .S 1 ; : : : ; S d /; P /

(1.19)

L0 .; F ; P / D L0 .; F ; P /I

see Appendix A.7 for the definition of Lp -spaces. If the market is complete then all of

these inclusions are in fact equalities. In particular, F coincides with .S 1 ; : : : ; S d /

modulo P -null sets, and every contingent claim coincides P -a.s. with a derivative of

the traded assets. Since the linear space V is finite-dimensional, it follows that the

same must be true of L0 .; F ; P /. But this means that the model can be reduced to

a finite number of relevant scenarios. This observation can be made precise by using

the notion of an atom of the probability space .; F ; P /. Recall that a set A 2 F is

called an atom of .; F ; P /, if P A > 0 and if each B 2 F with B A satisfies

either P B D 0 or P B D P A .

Proposition 1.41. For p 2 0; 1, the dimension of the linear space Lp .; F ; P / is

given by

dim Lp .; F ; P /

D supn 2 N j 9 partition A1 ; : : : ; An of with Ai 2 F and P Ai > 0: (1.20)

Moreover, n WD dim Lp .; F ; P / < 1 if and only if there exists a partition of

into n atoms of .; F ; P /.

Proof. Suppose that there is a partition A1 ; : : : ; An of such that Ai 2 F and

P Ai > 0. The corresponding indicator functions IA1 ; : : : ; IAn can be regarded

as linearly independent vectors in Lp WD Lp .; F ; P /. Thus dim Lp n. Consequently, it suffices to consider only the case in which the right-hand side of (1.20)

is a finite number, n0 . If A1 ; : : : ; An0 is a corresponding partition, then each Ai is

an atom because otherwise n0 would not be maximal. Thus, any Z 2 Lp is P -a.s.

constant on each Ai . If we denote the value of Z on Ai by z i , then

ZD

n0

X

i D1

zi IAi

P -a.s.

29

Hence, the indicator functions IA1 ; : : : ; IAn0 form a basis of Lp , and this implies

dim Lp D n0 .

Since for a complete market model the inclusions in (1.19) are in fact equalities,

we have

dim L0 .; F ; P / D dim V d C 1;

with equality when the model is non-redundant. Together with Proposition 1.41, this

implies the following result on the structure of complete market models.

Corollary 1.42. For every complete market model there exists a partition of into

at most d C 1 atoms of .; F ; P /.

Example 1.43. Consider the simple situation where the sample space consists of

two elements ! C and ! , and where the measure P is such that

p WD P ! C 2 .0; 1/:

We assume that there is one single risky asset, which takes at time t D 1 the two

values b and a with the respective probabilities p and 1 p, where a and b are such

that 0 a < b:

*

p

H

S.! C / D b

HH

H

j

H

1p H

S.! / D a

Q S j PQ P D pQ b C .1 p/a

.1 C r/ 2 E

Q j pQ 2 .0; 1/ D .a; b/I

(1.21)

see also Example 1.10. In this case, the model is also complete: Any risk-neutral

measure P must satisfy

.1 C r/ D E S D p b C .1 p /a;

and this condition uniquely determines the parameter p D P ! C as

p D

.1 C r/ a

2 .0; 1/:

ba

directly verify completeness by showing that a given contingent claim C is attainable

if (1.21) holds. Observe that the condition

C.!/ D 0 S 0 .!/ C S.!/ D 0 .1 C r/ C S.!/

for all ! 2

30

is a system of two linear equations for the two real variables 0 and . The solution is

given by

D

C.! C / C.! /

ba

and

0 D

C.! /b C.! C /a

:

.b a/.1 C r/

.C / D D

C.! C / .1 C r/ a

C.! / b .1 C r/

C

:

1Cr

ba

1Cr

ba

..S K/C / D

bK

.b K/a

1

:

ba

ba

1Cr

(1.22)

Note that this price is independent of p and increasing in r, while the classical discounted expectation with respect to the objective measure P ,

p.b K/

C

D

;

E

1Cr

1Cr

is decreasing in r and increasing in p.

In this example, one can illustrate how options can be used to modify the risk of a

position. Consider the particular case in which the risky asset can be bought at time

t D 0 for the price D 100. At time t D 1, the price is either S.! C / D b D 120 or

S.! / D a D 90, both with positive probability. If we invest in the risky asset, the

corresponding returns are given by

R.S/.! C / D C20 %

or

R.S/.! / D 10 %:

the price of the call option is

.C / D

20

6:67

3

R.C / D

.S K/C .C /

.C /

R.C /.! C / D

20 .C /

D C200 %

.C /

31

or

0 .C /

D 100 %;

.C /

R.C /.! / D

according to the outcome of the market at time t D 1. Here we see a dramatic increase

of both profit opportunity and risk; this is sometimes referred to as the leverage effect

of options.

On the other hand, we could reduce the risk of holding the asset by holding a

combination

CQ WD .K S /C C S

of a put option and the asset itself. This portfolio insurance will of course involve an

additional cost. If we choose our parameters as above, then the put-call parity (1.10)

yields that the price of the put option .K S/C is equal to 20=3. Thus, in order to

hold both S and a put, we must invest the capital 100 C 20=3 at time t D 0. At time

t D 1, we have an outcome of either 120 or of 100 so that the return of CQ is given by

R.CQ /.! C / D C12:5 %

and

(A) D !1 ; !2 with r D

1

9

1 D 5

S 1 .!1 / D

(B) D !1 ; !2 ; !3 with r D

1 D 5

S 1 .!1 / D

(C) D !1 ; !2 ; !3 with r D

D

5

;

10

S.!1 / D

1

9

20

3

40

3

S 1 .!2 / D

49

:

9

20

;

3

1

9

20

;

3

S 1 .!2 / D

49

;

9

S 1 .!3 / D

10

:

3

!

;

S.!2 / D

20

3

80

9

!

;

S.!3 / D

40

9

80

9

!

:

Each of these models is endowed with a probability measure that assigns strictly positive probability to each element of the corresponding sample space .

(a) Which of these models are arbitrage-free? For those that are, describe the set

P of equivalent risk-neutral measures. For those that are not, find an arbitrage

opportunity.

(b) Discuss the completeness of those models that are arbitrage-free. For those that

are not complete find non-attainable contingent claims.

}

32

such that P !i > 0 for i D 1; 2; 3 and consider the market model with r D 0

and one risky asset with prices 1 D 1 and 0 < S 1 .!1 / < S 2 .!2 / < S 3 .!3 /. We

suppose that the model is arbitrage-free.

(a) Describe the following objects as subsets of three-dimensional Euclidean space

R3 :

(i) the set P of equivalent risk neutral measures;

(ii) the set PQ of absolutely continuous risk neutral measures;

(iii) the set of attainable contingent claims.

(b) Find an example for a non-attainable contingent claim.

(c) Show that the supremum

Q C

sup E

(1.23)

PQ 2PQ

(d) Let C be a contingent claim. Give a direct and elementary proof of the fact that

Q C is constant if and

the map that assigns to each PQ 2 PQ the expectation E

only if the supremum (1.23) is attained in some element of P .

}

Exercise 1.4.3. Let D !1 ; : : : ; !N be endowed with a probability measure P

such that P !i > 0 for i D 1; : : : ; N . On this probability space we consider a

market model with interest rate r D 0 and with one risky asset whose prices satisfy

1 D 1 and

0 < S 1 .!1 / < S 1 .!2 / < < S 1 .!N /:

Show that there are strikes K1 ; : : : ; KN 2 > 0 and prices C .Ki / such that the

corresponding call options .S 1 Ki /C complete the market in the following sense:

the market model extended by the risky assets with prices

i WD C .Ki 1 /

and

S i WD .S 1 Ki 1 /C ;

i D 2; : : : ; N 1;

}

such that P !i > 0 for i D 1; : : : ; N C 1.

(a) On this probability space we consider a non-redundant and arbitrage-free market model with d risky assets and prices 2 Rd C1 and S, where d < N .

Show that this market model can be extended by additional assets with prices

d C1 ; : : : ; N and S d C1 ; : : : ; S N in such a way that the extended market

model is arbitrage-free and complete.

33

8

< 1 for i D 1;

S 1 .!i / D 2 for i D 2;

:

3 for i D 3:

We suppose furthermore that the risk-free interest rate r is chosen such that

the model is arbitrage-free. Find a non-attainable contingent claim. Then find

an extended model that is arbitrage-free and complete. Finally determine the

}

unique equivalent risk-neutral measure P in the extended model.

1.5

The fundamental theorem of asset pricing in the form of Theorem 1.7 states that a

market model is arbitrage-free if and only if the origin is contained in the set

dQ

is bounded, EQ jY j < 1 Rd ;

Mb .Y; P / WD EQ Y Q P;

dP

where Y D .Y 1 ; : : : ; Y d / is the random vector of discounted net gains defined in

(1.2). The aim of this section is to give a geometric description of the set Mb .Y; P /

as well as of the larger set

M.Y; P / WD EQ Y j Q P; EQ jY j < 1 :

To this end, it will be convenient to work with the distribution

WD P Y 1

of Y with respect to P . That is,

is a Borel probability measure on Rd such that

R

d such that

jyj .dy/ < 1, we will call

If

is

a

Borel

probability

measure

on

R

R

y .dy/ its barycenter.

Lemma 1.44. We have

Z

Mb .Y; P / D Mb .

/ WD

y .dy/

;

is bounded, jyj .dy/ < 1 ;

d

and

Z

M.Y; P / D M.

/ WD

y .dy/

; jyj .dy/ < 1 :

34

Proof. If

is a Borel probability measure on Rd , then the RadonNikodym

derivative of with respect to

evaluated at the random variable Y defines a probability measure Q P on .; F /:

d

dQ

.!/ WD

.Y .!//:

dP

d

R

Clearly, EQ Y D y .dy/. This shows that M.

/ M.Y; P / and Mb .

/

Mb .Y; P /.

Conversely, if QQ is a given probability measure on .; F / which is equivalent to

P , then the RadonNikodym theorem in Appendix A.2 shows that the distribution

Q WD QQ Y 1 must be equivalent to

, whence M.Y; P / M.

/. Moreover,

Q

it follows from Proposition A.11 that the density d =d

Q

is bounded if d Q=dP

is

bounded, and so Mb .Y; P / Mb .

/ also follows.

By the above lemma, the characterization of the two sets Mb .Y; P / and M.Y; P /

is reduced to a problem for Borel probability measures on Rd . Here and in the sequel,

we do not need the fact that

is the distribution of the lower boundedR random vector Y

of discounted net gains; our results are true for arbitrary

such that jyj

.dy/ < 1;

see also Remark 1.9.

Definition 1.45. The support of a Borel probability measure on Rd is the smallest

closed set A Rd such that .Ac / D 0, and it will be denoted by supp .

The support of a measure can be obtained as the intersection of all closed sets A

with .Ac / D 0, i.e.,

\

A:

supp D

A closed

.Ac /D0

We denote by

.

/ WD conv.supp

/

X

n

n

D

k yk k 0;

k D 1; yk 2 supp

; n 2 N

kD1

kD1

. Thus, .

/ is the smallest convex set which

contains supp

; see also Appendix A.1.

Example 1.46. Take d D 1, and consider the measure

1

.1 C C1 /:

2

35

is equal to 1; C1 and so .

/ D 1; C1. A measure

is equivalent to

if and only if

D 1 C .1 /C1

for some 2 .1; C1/. Hence, Mb .

/ D M.

/ D .1; C1/.

The previous example gives the correct intuition, namely that one always has the

inclusions

Mb .

/ M.

/ .

/:

But while the first inclusion will turn out to be an identity, the second inclusion is usually strict. Characterizing M.

/ in terms of .

/ will involve the following concept.

Definition 1.47. The relative interior of a convex set C Rd is the set of all points

x 2 C such that for all y 2 C there exists some " > 0 with

x ".y x/ 2 C:

The relative interior of C is denoted ri C .

If the convex set C has non-empty topological interior int C , then ri C D int C , and

the elementary properties of the relative interior collected in the following remarks

become obvious. This applies in particular to the set .

/ if the non-redundance

condition (1.8) is satisfied. For the general case, proofs of these statements can be

found, for instance, in 6 of [221].

Remark 1.48. Let C be a non-empty convex subset of Rd , and consider the affine

hull aff C spanned by C , i.e., the smallest affine set which contains C . If we identify

aff C with some Rn , then the relative interior of C is equal to the topological interior

of C , considered as a subset of aff C Rn . In particular, each non-empty convex set

has non-empty relative interior.

}

Exercise 1.5.1. Let C be a non-empty convex subset of Rd and denote by C its

closure. Show that for x 2 ri C ,

x C .1 /y 2 ri C

(1.24)

In particular, ri C is convex. Moreover, show that the operations of taking the closure

or the relative interior of a convex set C are consistent with each other

ri C D ri C

and ri C D C :

(1.25)

}

36

After these preparations, we can now state the announced geometric characterization of the set Mb .

/. Note that the proof of this characterization relies on the

fundamental theorem of asset pricing in the form of Theorem 1.7.

Theorem 1.49. The set of all barycenters of probability measures

coincides

with the relative interior of the convex hull of the support of

. More precisely,

Mb .

/ D M.

/ D ri .

/:

Proof. In a first step, we show the inclusion ri .

/ Mb .

/. Suppose we are given

m 2 ri .

/. Let

Q denote the translated measure

.A/

Q

WD

.A C m/

where A C m WD x C m j x 2 A. Then Mb .

/

Q D Mb .

/ m, and analogous

identities hold for M.

/

Q and .

/.

Q It follows that there is no loss of generality in

assuming that m D 0, i.e., we must show that 0 2 Mb .

/ if 0 2 ri .

/.

We claim that 0 2 ri .

/ implies the following no-arbitrage condition:

If 2 Rd is such that y 0 for

-a.e. y, then y D 0 for

-a.e. y.

(1.26)

-a.e. y but

is contained in

the closed set y j y 0 but not in the hyperplane y j y D 0. We conclude

that y 0 for all y 2 supp

and that there exists at least one y 2 supp

such

that y > 0. In particular, y 2 .

/ so that our assumption m D 0 2 ri .

/

implies the existence of some " > 0 such that "y 2 .

/. Consequently, "y

can be represented as a convex combination

"y D 1 y1 C C n yn

of certain y1 ; : : : ; yn 2 supp

. It follows that

0 > " y D 1 y1 C C n yn ;

in contradiction to our assumption that y 0 for all y 2 supp

. Hence, (1.26)

must be true.

Applying the fundamental theorem of asset pricing in the form of Theorem 1.7

to WD Rd , P WD

, and to the random variable Y .y/ WD y, yields a prob

density d

=d

is bounded and which satisfies

Rability measure

R

whose

jyj

.dy/ < 1 and y

.dy/ D 0. This proves the inclusion ri .

/ Mb .

/.

Clearly, Mb .

/ M.

/. So the theorem will be proved if we can show the

inclusion M.

/ ri .

/. To this end, suppose by way of contradiction that

is such that

Z

Z

jyj .dy/ < 1 and m WD y .dy/ ri .

/:

37

Again, we may assume without loss of generality that m D 0. Applying the separating

hyperplane theorem in the form of Proposition A.1 with C WD ri .

/ yields some

2 Rd such that y 0 for all y 2 ri .

/ and y > 0 for at least one

y 2 ri .

/. We deduce from (1.24) that y 0 holds also for all y 2 .

/.

Moreover, y0 must be strictly positive for at least one y0 2 supp

. Hence,

y 0 for

-a.e. y 2 Rd

and

.y j y > 0/ > 0.

(1.27)

By the equivalence of

and , (1.27) is also true for instead of

, and so

Z

Z

m D y .dy/ D y .dy/ > 0;

in contradiction to our assumption that m D 0. We conclude that M.

/ ri .

/.

Remark 1.50. Note that Theorem 1.49 does not extend to the set

Z

Q

M .

/ WD

y .dy/

and

jyj .dy/ < 1 :

Already the simple case

WD 12 .1 C C1 / serves as a counterexample, because

here MQ .

/ D 1; C1 while ri .

/ D .1; C1/. In this case, we have an identity

between MQ .

/ and .

/. However, also this identity fails in general as can be seen

by considering the normalized Lebesgue measure
on 1; C1. For this choice one

finds MQ .
/ D .1; C1/ but .
/ D 1; C1.

}

From Theorem 1.49 we obtain the following geometric characterization of the absence of arbitrage.

Corollary 1.51. Let

be the distribution of the discounted price vector S=.1 C r/ of

the risky assets. Then the market model is arbitrage-free if and only if the price system

belongs to the relative interior ri .

/ of the convex hull of the support of

.

1.6

The idea of hedging contingent claims develops its full power only in a dynamic

setting in which trading may occur at several times. The corresponding discretetime theory is presented in Chapter 5. The introduction of additional trading periods

requires more sophisticated techniques than those we have used so far. In this section

we will introduce some of these techniques in an extended version of our previous

market model in which initial prices, and hence strategies, are contingent on scenarios.

In this context, we are going to characterize the absence of arbitrage strategies. The

38

results will be used as building blocks in the multiperiod setting of Part II; their study

can be postponed until Chapter 5.

Suppose that we are given a -algebra F0 F which specifies the information

that is available to an investor at time t D 0. The prices for our d C 1 assets at time

0 will be modelled as non-negative F0 -measurable random variables S00 ; S01 ; : : : ; S0d .

Thus, the price system D . 0 ; 1 ; : : : ; d / of our previous discussion is replaced

by the vector

S 0 D .S00 ; : : : ; S0d /:

The portfolio chosen by an investor at time t D 0 will also depend on the information available at time 0. Thus, we assume that

D . 0 ; 1 ; : : : ; d /

is an F0 -measurable random vector. The asset prices observed at time t D 1 will be

denoted by

S 1 D .S10 ; S11 ; : : : ; S1d /:

They are modelled as non-negative random variables which are measurable with respect to a -algebra F1 such that F0 F1 F . The -algebra F1 describes the

information available at time 1, and in this section we can assume that F D F1 .

A riskless bond could be included by taking S00 1 and by assuming S10 to be

F0 -measurable and P -a.s. strictly positive. However, in the sequel it will be sufficient

to assume that S00 is F0 -measurable, S10 is F1 -measurable, and that

P S00 > 0 and S10 > 0 D 1:

(1.28)

X ti WD

S ti

;

S t0

i D 1; : : : ; d; t D 0; 1;

Y D X1 X0

the vector of the discounted net gains.

Definition 1.52. An arbitrage opportunity is a portfolio such that S 0 0,

S 1 0 P -a.s., and P S 1 > 0 > 0.

By our assumption (1.28), any arbitrage opportunity D . 0 ; / satisfies

Y 0 P -a.s.

(1.29)

39

is equivalent to the existence of an arbitrage opportunity. This can be seen as in

Lemma 1.4.

The space of discounted net gains which can be generated by some portfolio is

given by

K WD Y j 2 L0 .; F0 ; P I Rd / :

Here, L0 .; F0 ; P I Rd / denotes the space of Rd -valued random variables which are

P -a.s. finite and F0 -measurable modulo the equivalence relation (A.23) of coincidence up to P -null sets. The spaces Lp .; F0 ; P I Rd / for p > 0 are defined in the

p

p

same manner. We denote by LC WD LC .; F1 ; P / the cone of all non-negative elep

p

ments in the space L WD L .; F1 ; P /. With this notation, the absence of arbitrage

opportunities is equivalent to the condition

K \ L0C D 0:

We will denote by

K L0C

the convex cone of all Z 2 L0 which can be written as the difference of some Y 2

K and some U 2 L0C .

The following definition involves the notion of the conditional expectation

EQ Z j F0

of a random variable Z with respect to a probability measure Q, given the -algebra

F0 F ; see Appendix A.2. If Z D .Z 1 ; : : : ; Z n / is a random vector, then

EQ Z j F0 is shorthand for the random vector with components EQ Z i j F0 , i D

1; : : : ; n.

Definition 1.53. A probability measure Q satisfying

EQ X ti < 1

for i D 1; : : : ; d and t D 0; 1

and

X0 D EQ X1 j F0

Q-a.s.

risk-neutral measures P which are equivalent to P .

Remark 1.54. The definition of a martingale measure Q means that for each asset

i D 0; : : : ; d , the discounted price process .X ti / tD0;1 is a martingale under Q with

respect to the -fields .F t / tD0;1 . The systematic discussion of martingales in a multiperiod setting will begin in Section 5.2. The martingale aspect will be crucial for the

theory of dynamic hedging in Part II.

}

40

As the main result of this section, we can now state an extension of the fundamental theorem of asset pricing in Theorem 1.7 to our present setting. In the context of

Section 1.2, where F0 D ;; , the following arguments simplify considerably, and

they yield an alternative proof of Theorem 1.7, in which the separation argument in

Rd is replaced by a separation argument in L1 .

Theorem 1.55. The following conditions are equivalent:

(a) K \ L0C D 0.

(b) .K L0C / \ L0C D 0.

(c) There exists a measure P 2 P with a bounded density dP =dP .

(d) P ;.

Proof. (d) ) (a): Suppose by way of contradiction that there exist both a P 2 P

and some 2 L0 .; F0 ; P I Rd / with non-zero payoff Y 2 K \ L0C . For large

enough c > 0, .c/ WD Ijjc will be bounded, and the payoff .c/ Y will still be

non-zero and in K \ L0C . However,

E .c/ Y D E .c/ E Y j F0 D 0;

which is the desired contradiction.

(a) , (b): It is obvious that (a) is necessary for (b). In order to prove sufficiency,

suppose that we are given some Z 2 .K L0C / \ L0C . Then there exists a random

variable U 0 and a random vector 2 L0 .; F0 ; P I Rd / such that

0 Z D Y U:

This implies that Y U 0, which, according to condition (a), can only happen

if Y D 0. Hence, also U D 0 and in turn Z D 0.

(b) ) (c): This is the difficult part of the proof. The assertion will follow by

combining Lemmas 1.57, 1.58, 1.60, and 1.68.

Remark 1.56. If is discrete, or if there exists a decomposition of in countable

many atoms of .; F0 ; P /, then the martingale measure P can be constructed by

applying the result of Theorem 1.7 separately on each atom. In the general case, the

idea of patching together conditional martingale measures would involve subtle arguments of measurable selection; see [67]. Here we present a different approach which

is based on separation arguments in L1 .P /. It is essentially due to W. Schachermayer

[233]; our version uses in addition arguments by Y. Kabanov and C. Stricker [164]. }

We start with the following simple lemma, which takes care of the integrability

condition in Definition 1.53.

41

Lemma 1.57. For the proof of the implication (b) ) (c) in Theorem 1:55, we may

assume without loss of generality that

E jX t j < 1 for t D 0; 1.

(1.30)

d PQ

WD c.1 C jX0 j C jX1 j/1

dP

where c is chosen such that the right-hand side integrates to 1. Clearly, (1.30) holds

for PQ . Moreover, condition (b) of Theorem 1.55 is satisfied by P if and only if it is

satisfied by the equivalent measure PQ . If P 2 P is such that the density dP =d PQ

is bounded, then so is the density

dP d PQ

dP

D

:

dP

d PQ dP

Therefore, the implication (b) ) (c) holds for P if and only if it holds for PQ .

From now on, we will always assume (1.30). Our goal is to construct a suitable

Z 2 L1 such that

dP

Z

WD

dP

E Z

defines an equivalent risk-neutral measure P . The following simple lemma gives a

criterion for this purpose, involving the convex cone

C WD .K L0C / \ L1 :

Lemma 1.58. Suppose c 0 and Z 2 L1 are such that

E Z W c

for all W 2 C .

Then:

(a) E Z W 0 for all W 2 C , i.e., we can take c D 0.

(b) Z 0 P -a.s.

(c) If Z does not vanish P -a.s., then

Z

dQ

WD

dP

E Z

defines a risk-neutral measure Q

P .

42

Proof. (a): Note that C is a cone, i.e., W 2 C implies that W 2 C for all 0.

This property excludes the possibility that E ZW > 0 for some W 2 C .

(b): C contains the function W WD IZ<0 . Hence, by part (a),

E Z D E Z W 0:

(c): For all 2 L1 .; F0 ; P I Rd / and 2 R we have Y 2 C by our

integrability assumption (1.30). Thus, a similar argument as in the proof of (a) yields

E Z Y D 0. Since is bounded, we may conclude that

0 D E Z Y D E E ZY j F0 :

As is arbitrary, this yields E ZY j F0 D 0 P -almost surely. Proposition A.12

now implies

EQ Y j F0 D

1

E ZY j F0 D 0 Q-a.s.,

E Z j F0

In view of the preceding lemma, the construction of risk-neutral measures Q

P

with bounded density is reduced to the construction of elements of the set

Z WD Z 2 L1 j 0 Z 1; P Z > 0 > 0; and E Z W 0 for all W 2 C :

In the following lemma, we will construct such elements by applying a separation

argument suggested by the condition

C \ L1C D 0;

which follows from condition (b) of Theorem 1.55. This separation argument needs

the additional assumption that C is closed in L1 . Showing that this assumption is indeed satisfied in our situation will be one of the key steps in our proof; see Lemma 1.68

below.

Lemma 1.59. Assume that C is closed in L1 and satisfies C \ L1C D 0. Then for

each non-zero F 2 L1C there exists some Z 2 Z such that E F Z > 0.

Proof. Let B WD F so that B \ C D ;. Since the set C is non-empty, convex

and closed in L1 , we may apply the HahnBanach separation theorem in the form of

Theorem A.57 to obtain a continuous linear functional ` on L1 such that

sup `.W / < `.F /:

W 2C

Since the dual space of L1 can be identified with L1 , there exists some Z 2 L1

such that `.F / D E F Z for all F 2 L1 . We may assume without loss of generality

that kZk1 1. By construction, Z satisfies the assumptions of Lemma 1.58, and

so Z 2 Z. Moreover, E F Z D `.F / > 0 since the constant function W 0 is

contained in C .

43

positive element Z under the assumptions of Lemma 1.59. After normalization, Z

will serve as the density of our desired risk-neutral measure P 2 P .

Lemma 1.60. Under the assumptions of Lemma 1:59, there exists Z 2 Z with

Z > 0 P -a.s.

Proof. As a first step, we claim that Z is countably convex: If .k /k2N is a sequence

of non-negative real numbers summing up to 1, and if Z .k/ 2 Z for all k, then

Z WD

1

X

k Z .k/ 2 Z:

kD1

Indeed, for W 2 C

1

X

jk Z .k/ W j jW j 2 L1 ;

kD1

E ZW D

1

X

k E Z .k/ W 0:

kD1

c WD supP Z > 0 j Z 2 Z:

We choose Z .n/ 2 Z such that P Z .n/ > 0 ! c. Then

Z WD

1

X

2n Z .n/ 2 Z

nD1

Z > 0 D

1

[

Z .n/ > 0:

nD1

Hence P Z > 0 D c.

In the final step, we show that c D 1. Then Z will be as desired. Suppose by way

of contradiction that P Z D 0 > 0, so that W WD IZ D0 is a non-zero element of

L1C . Lemma 1.59 yields Z 2 Z with E W Z > 0. Hence,

P Z > 0 \ Z D 0 > 0;

and so

1

P

.Z C Z / > 0 > P Z > 0 D c;

2

44

Thus, we have completed the proof of the implication (b) ) (c) of Theorem 1.55

up to the requirement that C is closed in L1 . Let us pause here in order to state

general versions of two of the arguments we have used so far. The first is known as

the HalmosSavage theorem.

Theorem 1.61. Let Q be a set of probability measures which are all absolutely continuous with respect to a given measure P . Suppose moreover that Q P in the

sense that Q A D 0 for all Q 2 Q implies that P A D 0. Then there exists a

countable subfamily QQ Q which satisfies QQ P . In particular, there exists an

equivalent measure in Q as soon as Q is countably convex in the following sense: If

of non-negative real numbers summing up to 1, and if Qk 2 Q

.k /k2N is a sequence

P

kD1 k Qk 2 Q.

Exercise 1.6.1. Prove Theorem 1.61 by modifying the exhaustion argument used in

the proof of Lemma 1.60.

}

An inspection of Lemmas 1.58, 1.59, and 1.60 shows that the particular structure of

C D .K L0C / \ L1 was only used for part (c) of Lemma 1.58. All other arguments

relied only on the fact that C is a closed convex cone in L1 that contains all bounded

negative functions and no non-trivial positive function. Thus, we have in fact proved

the following KrepsYan theorem, which was obtained independently in [266] and

[186].

Theorem 1.62. Suppose C is a closed convex cone in L1 satisfying

C L1

C

and

C \ L1C D 0:

Then there exists Z 2 L1 such that Z > 0 P -a.s. and E W Z 0 for all W 2 C.

Let us now turn to the closedness of our set C D .K L0C / \ L1 . The following

example illustrates that we cannot expect C to be closed without assuming the absence

of arbitrage opportunities.

Example 1.63. Let P be the Lebesgue measure on the Borel field F1 of D 0; 1,

and take F0 D ;; and Y .!/ D !. This choice clearly violates the no-arbitrage

condition, i.e., we have K \ L0C 0. The convex set C D .K L0C / \ L1 is a

proper subset of L1 . More precisely, C does not contain any function F 2 L1 with

F 1: If we could represent F as Y U for a non-negative function U , then it

would follow that

Y D F C U 1;

which is impossible for any . However, as we show next, the closure of C in L1

coincides with the full space L1 . In particular, C cannot be closed. Let F 2 L1 be

arbitrary, and observe that

Fn WD .F C ^ n/ I 1 ;1 F

n

45

.F C ^ n/ I 1 ;1 n2 Y:

n

closed, using a simplified version of Lemma 1.68 below. In this way, we obtain an

alternative proof of Theorem 1.7. In the general case we need some preparation. Let

us first prove a randomized version of the BolzanoWeierstra theorem. It yields a

simple construction of a measurable selection of a convergent subsequence of a given

sequence in L0 .; F0 ; P I Rd /.

Lemma 1.64. Let .n / be a sequence in L0 .; F0 ; P I Rd / with lim infn jn j < 1.

Then there exists 2 L0 .; F0 ; P I Rd / and a strictly increasing sequence . m / of

F0 -measurable integer-valued random variables such that

m .!/ .!/ ! .!/

for P -a.e. ! 2 .

Proof. Let .!/ WD lim infn jn .!/j, and define m WD m on the P -null set D

0

1. On < 1 we let 10 WD 1, and we define F0 -measurable random indices m

by

0

0

; m D 2; 3; : : : :

m WD inf n > m1 j jn j j

m

We use recursion on i D 1; : : : ; d to define the i th component i of the limit and to

i of random indices. Let

extract a new subsequence m

i D lim inf i i1 ;

m"1

i : Let

which is already defined if i D 1. This i can be used in the construction of m

i WD 1 and, for m D 2; 3; : : : ,

1

i

m

.!/

WD inf

ni 1 .!/

1

i 1

i

i

i

:

n .!/ > m1 .!/ and j i1 .!/ .!/j

n

m

Then m WD m

Y D Q Y

P -a.s.;

46

Remark 1.65. We could exclude this possibility by the following assumption of nonredundance:

Y D Q Y P -a.s. H) D Q P -a.s.

(1.31)

Under this assumption, we can immediately move on to the final step in Lemma 1.68.

}

Without assumption (1.31), it will be convenient to have a suitable linear space

N of reference portfolios which are uniquely determined by their payoff. The

construction of N ? is the purpose of the following lemma. We will assume that the

spaces L0 and L0 .; F0 ; P I Rd / are endowed with the topology of convergence in

P -measure, which is generated by the metric d of (A.24).

?

N WD 2 L0 .; F0 ; P I Rd / j Y D 0 P -a.s.;

N ? WD 2 L0 .; F0 ; P I Rd / j D 0 P -a.s. for all 2 N :

(a) Both N and N ? are closed in L0 .; F0 ; P I Rd / and, in the following sense,

invariant under the multiplication with scalar functions g 2 L0 .; F0 ; P /: If

2 N and 2 N ? , then g 2 N and g 2 N ? .

(b) If 2 N ? and Y D 0 P -a.s., then D 0, i.e., N \ N ? D 0.

(c) Every 2 L0 .; F0 ; P I Rd / has a unique decomposition D C ? , where

2 N and ? 2 N ? .

Remark 1.67. For the proof of this lemma, we will use a projection argument in

Hilbert space. Let us sketch a more probabilistic construction of the decomposition

D C ? . Take a regular conditional distribution of Y given F0 , i.e., a stochastic

kernel K from .; F0 / to Rd such that K.!; A/ D P Y 2 A j F0 .!/ for all

Borel sets A Rd and P -a.e. ! (see, e.g., 44 of [20]). If one defines ? .!/

as the orthogonal projection of .!/ onto the linear hull L.!/ of the support of the

measure K.!; /, then WD ? satisfies Y D 0 P -a.s., and any Q with the same

property must be P -a.s. perpendicular to L.!/. However, carrying out the details

of this construction involves certain measurability problems; this is why we use the

projection argument below.

}

Proof. (a): The closedness of N and N ? follows immediately from the metrizability

of L0 .; F0 ; P I Rd / (see Appendix A.7) and the fact that every sequence which

converges in measure has an almost-surely converging subsequence. The invariance

under the multiplication with F0 -measurable scalar functions is obvious.

(b): Suppose that 2 N \ N ? . Then taking WD in the definition of N ? yields

D jj2 D 0 P -a.s.

47

.!/ D 1 .!/ e1 C C d .!/ ed ;

where ei denotes the i th Euclidean unit vector, and where i .!/ is the i th component

of .!/. Consider ei as a constant element of L0 .; F0 ; P I Rd /, and suppose that

we can decompose ei as

ei D ni C ei?

(1.32)

Since by part (a) both N and N ? are invariant under the multiplication with F0 measurable functions, we can then obtain the desired decomposition of by letting

.!/ WD

d

X

i .!/ ni .!/

and

i D1

? .!/ WD

d

X

i D1

It remains to construct the decomposition (1.32) of ei . The constant ei is an element

of the space H WD L2 .; F0 ; P I Rd /, which becomes a Hilbert space if endowed

with the natural inner product

.; /H WD E ;

; 2 L2 .; F0 ; P I Rd /:

Observe that both N \ H and N ? \ H are closed subspaces of H , because convergence in H implies convergence in L0 .; F0 ; P I Rd /. Therefore, we can define

the corresponding orthogonal projections

0 W H ! N \ H

and

? W H ! N ? \ H:

Thus, letting ni WD 0 .ei / and ei? WD ? .ei / will be the desired decomposition

(1.32), once we know that ei D 0 .ei / C ? .ei /. To prove this, we need only show

that WD ei 0 .ei / is contained in N ? . We assume by way of contradiction that is

not contained in N ? \ H . Then there exists some 2 N such that P > 0 > 0.

Clearly,

Q WD I>0; jjc

is contained in N \ H for each c > 0. But if c is large enough, then 0 < E Q D

.;

Q /H , which contradicts the fact that is by construction orthogonal to N \ H .

After these preparations, we can now complete the proof of Theorem 1.55 by showing the closedness of C D .K L0C / \ L1 in L1 . This is an immediate consequence

of the following lemma, since convergence in L1 implies convergence in L0 , i.e.,

convergence in P -measure. Recall that we have already proved the equivalence of the

conditions (a) and (b) in Theorem 1.55.

48

Proof. Suppose Wn 2 .K L0C / converges in L0 to some W as n " 1. By passing

to a suitable subsequence, we may assume without loss of generality that Wn ! W

P -almost surely. We can write Wn D n Y Un for n 2 N ? and Un 2 L0C .

In a first step, we will prove the assertion given the fact that

lim inf jn j < 1

n"1

P -a.s.,

(1.33)

which will be established afterwards. Assuming (1.33), Lemma 1.64 yields F0 -measurable integer-valued random variables 1 < 2 < and some 2 L0 .; F0 ; P I Rd /

such that P -a.s. n ! . It follows that

Un D n Y Wn ! Y W DW U

P -a.s.,

(1.34)

Let us now show that A WD lim infn jn j D C1 satisfies P A D 0 as claimed

in (1.33). Let

n

when jn j > 0,

n WD jn j

otherwise,

e1

where e1 is the unit vector .1; 0; : : : ; 0/. Using Lemma 1.64 on the sequence .n /

yields F0 -measurable integer-valued random variables 1 < 2 < and some 2

L0 .; F0 ; P I Rd / such that P -a.s. n ! . The convergence of .Wn / implies that

Un

Wn

0 IA

D IA n Y

! IA Y P -a.s.

jn j

jn j

Hence, our assumption K \ L0C D 0 yields .IA / Y D 0. Below we will show that

IA 2 N ? , so that

D 0 P -a.s. on A.

(1.35)

On the other hand, the fact that jn j D 1 P -a.s. implies that jj D 1 P -a.s., which

can only be consistent with (1.35) if P A D 0.

It remains to show that IA 2 N ? . To this end, we first observe that each n

belongs to N ? since, for each 2 N ,

n D

1

X

kD1

In Dk

1

k D 0

jk j

P -a.s.

If in the proof of Lemma 1.68 Wn D n Y for all n, then U D 0 in (1.34), and

W D limn Wn is itself contained in K. We thus get the following lemma, which will

be useful in Chapter 5.

49

In fact, it is possible to show that K is always closed in L0 ; see [257], [233]. But

this stronger result will not be needed here.

As an alternative to the randomized BolzanoWeierstra theorem in Lemma 1.64,

we can use the following variant of Komlos principle of subsequences. It yields

a convergent sequence of convex combinations of a sequence in L0 .; F0 ; P I Rd /,

and this will be needed later on. Recall from Appendix A.1 the notion of the convex

hull

X

n

n

conv A D

i xi xi 2 A; i 0;

i D 1; n 2 N

i D1

i D1

Lemma 1.70. Let .n / be a sequence in L0 .; F0 ; P I Rd / such that supn jn j < 1

P -almost surely. Then there exists a sequence of convex combinations

n 2 convn ; nC1 ; : : :

which converges P -almost surely to some 2 L0 .; F0 ; P I Rd /.

Proof. We can assume without loss of generality that supn jn j 1 P -a.s.; otherwise

we consider the sequence Qn WD n = supn jn j. Then .n / is a bounded sequence

in the Hilbert space H WD L2 .; F0 ; P I Rd /. Since the closed unit ball in H is

weakly compact, the sequence .n / has an accumulation point 2 H ; note that

weak sequential compactness follows from the BanachAlaoglu theorem in the form

of Theorem A.63 and the fact that the dual H 0 of the Hilbert space H is isomorphic

to H itself. For each n, the accumulation point belongs to the L2 -closure Cn of

convn ; nC1 ; : : : , due to the fact that a closed convex set in H is also weakly

closed; see Theorem A.60. Thus, we can find n 2 convn ; nC1 ; : : : such that

E jn j2

1

:

n2

Remark 1.71. The original result by Komlos [178] is more precise: It states that for

any bounded sequence .n / in L1 .; F ; P I Rd / there is a subsequence .nk / which

satisfies a strong law of large numbers, i.e.,

N

1 X

nk

N "1 N

lim

kD1

Chapter 2

Preferences

by arbitrage arguments, without involving the preferences of economic agents. In an

incomplete model, such claims may carry an intrinsic risk which cannot be hedged

away. In order to determine desirable strategies in view of such risks, the preferences

of an investor should be made explicit, and this is usually done in terms of an expected

utility criterion.

The paradigm of expected utility is the theme of this chapter. We begin with a

general discussion of preference relations on a set X of alternative choices and their

numerical representation by some functional U on X. In the financial context, such

choices can usually be described as payoff profiles. These are defined as functions

X on an underlying set of scenarios with values in some set of payoffs. Thus we are

facing risk or even uncertainty. In the case of risk, a probability measure is given on

the set of scenarios. In this case, we can focus on the resulting payoff distributions.

We are then dealing with preferences on lotteries, i.e., on probability measures on

the set of payoffs.

In Sections 2.2 and 2.3 we discuss the conditions or axioms under which such a

preference relation on lotteries

can be represented by a functional of the form

Z

u.x/

.dx/;

where u is a utility function on the set of payoffs. This formulation of preferences on

lotteries in terms of expected utility goes back to D. Bernoulli [25]; the axiomatic theory was initiated by J. von Neumann and O. Morgenstern [209]. Section 2.4 characterizes uniform preference relations which are shared by a given class of functions u.

This involves the general theory of probability measures on product spaces with given

marginals which will be discussed in Section 2.6.

In Section 2.5 we return to the more fundamental level where preferences are defined on payoff profiles, and where we are facing uncertainty in the sense that no

probability measure is given a priori. L. Savage [232] clarified the conditions under

which such preferences on a space of functions X admit a representation of the form

U.X / D EQ u.X /

where Q is a subjective probability measure on the set of scenarios. We are going to

concentrate on a robust extension of the Savage representation which was introduced

51

and Rustichini [198]. Here the utility functional is of the form

U.X / D inf .EQ u.X / C .Q//:

Q2Q

It thus involves a whole class Q of probability measures Q, which are taken more

or less seriously according to their penalization .Q/. The axiomatic approach to

the robust Savage representation is closely related to the construction of coherent and

convex risk measures, which will be the topic of Chapter 4.

2.1

choice of an economic agent. If presented with two choices x; y 2 X, the agent

might prefer one over the other. This will be formalized as follows.

Definition 2.1. A preference order (or preference relation) on X is a binary relation

with the following two properties.

Asymmetry: If x y, then y x.

must hold.

Negative transitivity states that if a clear preference exists between two choices

x and y, and if a third choice z is added, then there is still a choice which is least

preferable (y if z y) or most preferable (x if x z).

Definition 2.2. A preference order on X induces a corresponding weak preference

order
defined by

x
y W y x;

and an indifference relation given by

x y W x
y and y
x:

Thus, x
y means that either x is preferred to y or there is no clear preference

between the two.

Remark 2.3. It is easy to check that the asymmetry and the negative transitivity of

are equivalent to the following two respective properties of
:

(a) Completeness: For all x; y 2 X, either y
x or x
y or both are true.

(b) Transitivity: If x
y and y
z, then also x
z.

52

Chapter 2 Preferences

Conversely, any complete and transitive relation
induces a preference order via

the negation of
, i.e.,

yx W

x y:

The indifference relation is an equivalence relation, i.e., it is reflexive, symmetric

and transitive.

}

Exercise 2.1.1. Prove the assertions in the preceding remark.

X ! R such that

y x U.y/ > U.x/:

(2.1)

Clearly, (2.1) is equivalent to

y x U.y/ U.x/:

Note that such a numerical representation U is not unique: If f is any strictly increasing function, then UQ .x/ WD f .U.x// is again a numerical representation.

Definition 2.5. Let be a preference relation on X. A subset Z of X is called

order dense if for any pair x; y 2 X such that x y there exists some z 2 Z with

x z y.

The following theorem characterizes those preference relations for which there exists a numerical representation.

Theorem 2.6. For the existence of a numerical representation of a preference relation

it is necessary and sufficient that X contains a countable, order dense subset Z. In

particular, any preference order admits a numerical representation if X is countable.

Proof. Suppose first that we are given a countable order dense subset Z of X. For

x 2 X, let

Z.x/ WD z 2 Z j z x and

Z.x/ WD z 2 Z j x z:

The relation x
y implies that Z.x/ Z.y/ and Z.x/ Z.y/. If the strict relation

x y holds, then at least one of these inclusions is also strict. To see this, pick

z 2 Z with x
z
y, so that either x z
y or x
z y. In the first case,

z 2 Z.x/nZ.y/, while z 2 Z.y/nZ.x/ in the second case.

Next, take any strictly positive probability distribution

on Z, and let

X

X

.z/

.z/:

U.x/ WD

z2Z.x/

z2Z.x/

53

By the above, U.x/ > U.y/ if and only if x y so that U is the desired numerical

representation.

For the proof of the converse assertion take a numerical representation U and let J

denote the countable set

J WD a; b j a; b 2 Q; a < b; U 1 .a; b/ ; :

For every interval I 2 J we can choose some zI 2 X with U.zI / 2 I and thus define

the countable set

A WD zI j I 2 J:

At first glance it may seem that A is a good candidate for an order dense set. However,

it may happen that there are x; y 2 X such that U.x/ < U.y/ and for which there is

no z 2 X with U.x/ < U.z/ < U.y/. In this case, an order dense set must contain

at least one z with U.z/ D U.x/ or U.z/ D U.y/, a condition which cannot be

guaranteed by A.

Let us define the set C of all pairs .x; y/ which do not admit any z 2 A with

y z x:

C WD .x; y/ j x; y 2 XnA; y x and z 2 A with y z x :

Then .x; y/ 2 C implies the apparently stronger fact that we cannot find any z 2 X

such that y z x: Otherwise we could find a, b 2 Q such that

U.x/ < a < U.z/ < b < U.y/;

so I WD a; b would belong to J, and the corresponding zI would be an element of

A with y zI x, contradicting the assumption that .x; y/ 2 C .

It follows that all intervals .U.x/; U.y// with .x; y/ 2 C are disjoint and nonempty. Hence, there can be only countably many of them. For each such interval

J we pick now exactly one pair .x J ; y J / 2 C such that U.x J / and U.y J / are the

endpoints of J , and we denote by B the countable set containing all x J and all y J .

Finally, we claim that Z WD A [ B is an order dense subset of X. Indeed, if x,

y 2 XnZ with y x, then either there is some z 2 A such that y z x, or

.x; y/ 2 C . In the latter case, there will be some z 2 B with U.y/ D U.z/ > U.x/

and, consequently, y
z x.

The following example shows that even in a seemingly straightforward situation, a

given preference order may not admit a numerical representation.

Example 2.7. Let be the usual lexicographical order on X WD 0; 1 0; 1, i.e.,

.x1 ; x2 / .y1 ; y2 / if and only if either x1 > y1 , or if x1 D y1 and simultaneously

x2 > y2 . One easily checks that is antisymmetric and negative transitive, and hence

a preference order. We show now that does not admit a numerical representation.

54

Chapter 2 Preferences

To this end, let Z be any order-dense subset of X. Then, for x 2 0; 1 there must

be some .z1 ; z2 / 2 Z such that .x; 1/
.z1 ; z2 /
.x; 0/. It follows that z1 D x

and that Z is uncountable. Theorem 2.6 thus implies that there cannot be a numerical

representation of the lexicographical order .

}

Definition 2.8. Let X be a topological space. A preference relation is called continuous if for all x 2 X

B.x/ WD y 2 X j y x and

B.x/ WD y 2 X j x y

(2.2)

Remark 2.9. Every preference order that admits a continuous numerical representation is itself continuous. Under some mild conditions on the underlying space X, the

converse statement is also true; see Theorem 2.15 below.

}

Example 2.10. The lexicographical order of Example 2.7 is not continuous: If

.x1 ; x2 / 2 0; 1 0; 1 is given, then

.y1 ; y2 / j .y1 ; y2 / .x1 ; x2 / D .x1 ; 1 0; 1 [ x1 .x2 ; 1;

which is typically not an open subset of 0; 1 0; 1.

two distinct points in X have disjoint open neighborhoods. In this case, all singletons

x are closed. Clearly, every metric space is a topological Hausdorff space.

Proposition 2.11. Let be a preference order on a topological Hausdorff space X.

Then the following properties are equivalent:

(a) is continuous.

(b) The set .x; y/ j y x is open in X X.

(c) The set .x; y/ j y x is closed in X X.

Proof. (a) ) (b): We have to show that for any pair

.x0 ; y0 / 2 M WD .x; y/ j y x

there exist open sets U; V X such that x0 2 U , y0 2 V , and U V M .

Consider first the case in which there exists some z 2 B.x0 / \ B.y0 / for the notation

B.x0 / and B.y0 / introduced in (2.2). Then y0 z x0 , so that U WD B.z/ and

V WD B.z/ are open neighborhoods of x0 and y0 , respectively. Moreover, if x 2 U

and y 2 V , then y z x, and thus U V M .

If B.x0 / \ B.y0 / D ;, we let U WD B.y0 / and V WD B.x0 /. If .x; y/ 2 U V ,

then y0 x and y x0 by definition. We want to show that y x in order to

55

negative transitivity, hence y0 y x0 . But then y 2 B.x0 / \ B.y0 / ;, and we

have a contradiction.

(b) ) (c): First note that the mapping .x; y/ WD .y; x/ is a homeomorphism of

X X. Then observe that the set .x; y/ j y x is just the complement of the open

set ..x; y/ j y x/.

(c) ) (a): Since X is a topological Hausdorff space, x X is closed in X X,

and so is the set

x X \ .x; y/ j y x D x y j y x:

Hence y j y x is closed in X, and its complement y j x y is open. The

same argument applies to y j y x.

Example 2.12. For x0 < y0 consider the set X WD .1; x0 [ y0 ; 1/ endowed

with the usual order > on R. Then, with the notation introduced in (2.2), B.y0 / D

.1; x0 and B.x0 / D y0 ; 1/. Hence,

B.x0 / \ B.y0 / D ;

despite y0 x0 , a situation we had to consider in the preceding proof.

the union of two disjoint and non-empty open sets. Assuming that X is connected

will rule out the situation occurring in Example 2.12.

Proposition 2.13. Let X be a connected topological space with a continuous preference order . Then every dense subset Z of X is also order dense in X. In particular,

there exists a numerical representation of if X is separable.

Proof. Take x, y 2 X with y x, and consider B.x/ and B.y/ as defined in (2.2).

Since y 2 B.x/ and x 2 B.y/, neither B.x/ nor B.y/ are empty sets. Moreover,

negative transitivity implies that X D B.x/ [ B.y/. Hence, the open sets B.x/

and B.y/ cannot be disjoint, as X is connected. Thus, the open set B.x/ \ B.y/

must contain some element z of the dense subset Z, which then satisfies y z x.

Therefore Z is an order dense subset of X.

Separability of X means that there exists a countable dense subset Z of X, which

then is order dense. Hence, the existence of a numerical representation follows from

Theorem 2.6.

Remark 2.14. Consider the situation of Example 2.12, where X WD .1; x0 [

y0 ; 1/, and suppose that x0 and y0 are both irrational. Then Z WD Q \ X is dense

in X, but there exists no z 2 Z such that y0 z x0 . This example shows that the

assumption of topological connectedness is essential for Proposition 2.13.

}

56

Chapter 2 Preferences

Theorem 2.15. Let X be a topological space which satisfies at least one of the following two properties:

For a proof we refer to [77], Propositions 3 and 4. For our purposes, namely for the

proof of the von NeumannMorgenstern representation in the next section and for the

proof of the robust Savage representation in Section 2.5, the following lemma will be

sufficient.

Lemma 2.16. Let X be a connected metric space with a continuous preference order

. If U W X ! R is a continuous function, and if its restriction to some dense

subset Z is a numerical representation for the restriction of to Z, then U is also a

numerical representation for on X.

Proof. We have to show that y x if and only if U.y/ > U.x/. In order to verify

the only if part, take x, y 2 X with y x. As in the proof of Proposition 2.13,

we obtain the existence of some z0 2 Z with y z0 x. Repeating this argument

yields z00 2 Z such that z0 z00 x. Now we take two sequences .zn / and .zn0 / in Z

with zn ! y and zn0 ! x. By continuity of , eventually

zn z0 z00 zn0 ;

and thus

U.zn / > U.z0 / > U.z00 / > U.zn0 /:

The continuity of U implies that U.zn / ! U.y/ and U.zn0 / ! U.x/, whence

U.y/ U.z0 / > U.z00 / U.x/:

For the proof of the converse implication, suppose that x, y 2 X are such that

U.y/ > U.x/. Since U is continuous,

U.x/ WD z 2 X j U.z/ > U.x/

and

U.y/ WD z 2 X j U.z/ < U.y/

are both non-empty open subsets of X. Moreover, U.y/ [ U.x/ D X. Connectedness of X implies that U.y/ \ U.x/ ;. As above, a repeated application of the

preceding argument yields z0 , z00 2 Z such that

U.y/ > U.z0 / > U.z00 / > U.x/:

57

Since Z is a dense subset of X, we can find sequences .zn / and .zn0 / in Z with

zn ! y and zn0 ! x as well as with U.zn / > U.z0 / and U.zn0 / < U.z00 /. Since U is

a numerical representation of on Z, we have

zn z0 z00 zn0 :

Hence, by the continuity of , neither z0 y nor x z00 can be true, and negative

transitivity yields y x.

2.2

Suppose that each possible choice for our economic agent corresponds to a probability

distribution on a given set of scenarios. Thus, the set X can be identified with a

subset M of the set M1 .S; S/ of all probability distributions on a measurable space

.S; S/. In the context of the theory of choice, the elements of M are sometimes

called lotteries. We will assume in the sequel that M is convex. The aim of this

section is to characterize those preference orders on M which allow for a numerical

representation U of the form

Z

U.

/ D u.x/

.dx/ for all

2 M,

(2.3)

where u is a real function on S .

Definition 2.17. A numerical representation U of a preference order on M is called

a von NeumannMorgenstern representation if it is of the form (2.3).

Any von NeumannMorgenstern representation U is affine on M in the sense that

U.

C .1 // D U.

/ C .1 /U./

for all

; 2 M and 2 0; 1. It is easy to check that affinity of U implies the

following two properties, or axioms, for a preference order on M. The first property

says that a preference

is preserved in any convex combination, independent of

the context described by another lottery
.

Definition 2.18. A preference relation on M satisfies the independence axiom if,

for all

, 2 M, the relation

implies

C .1 /
C .1 /

for all
2 M and all 2 .0; 1.

58

Chapter 2 Preferences

The independence axiom is also called the substitution axiom. It can be illustrated

by introducing a compound lottery, which represents the distribution

C .1 /

as a two-step procedure. First, we sample either lottery

or
with probability and

1 , respectively. Then the lottery drawn in this first step is realized. Clearly, this

is equivalent to playing directly the lottery

C .1 /
. With probability 1 ,

the distribution
is drawn and in this case there is no difference to the compound

lottery where is replaced by

. The only difference occurs when

is drawn, and

this happens with probability . Thus, if

then it seems reasonable to prefer the

compound lottery with

over the one with .

Definition 2.19. A preference relation on M satisfies the Archimedean axiom if for

any triple

there are , 2 .0; 1/ such that

C .1 /

C .1 /:

The Archimedean axiom derives its name from its similarity to the Archimedean

principle in real analysis: For every small " > 0 and each large x, there is some

n 2 N such that n " > x. Sometimes it is also called the continuity axiom, because

it can act as a substitute for the continuity of in a suitable topology on M. More

precisely, suppose that M is endowed with a topology for which convex combinations

are continuous curves, i.e.,

C .1 / converges to or

as # 0 or " 1,

respectively. Then continuity of our preference order in this topology automatically

implies the Archimedean axiom.

Remark 2.20. As an axiom for consistent behavior in the face of risk, the Archimedean axiom is less intuitive than the independence axiom. Consider the following

three deterministic distributions: yields 1000 C,
yields 10 C, and

is the lottery

where one dies for sure. Even for small 2 .0; 1/ it is not clear that someone would

prefer the gamble

C .1 /, which involves the probability of dying, over the

conservative 10 C yielded by
. Note, however, that most people would not hesitate

to drive a car for a distance of 50 km in order to receive a premium of 1000 C, even

though this might involve the risk of a deadly accident.

}

Our first goal is to show that the Archimedean axiom and the independence axiom

imply the existence of an affine numerical representation.

Theorem 2.21. Suppose that is a preference relation on M satisfying both the

Archimedean and the independence axiom. Then there exists an affine numerical representation U of . Moreover, U is unique up to positive affine transformations,

i.e., any other affine numerical representation UQ with these properties is of the form

UQ D a U C b for some a > 0 and b 2 R.

The affinity of a numerical representation does not always imply that it is also of

von NeumannMorgenstern form; see Exercise 2.2.1 and Example 2.26 below. In

59

two important cases, however, such an affine numerical representation will already

be of von NeumannMorgenstern form. This is the content of the following two

corollaries, which we state before proving Theorem 2.21. For the first corollary, we

need the notion of a simple probability distribution. This is a probability measure

PN of Dirac masses, i.e., there

exist x1 ; : : : ; xN 2 S and 1 ; : : : ; N 2 .0; 1 with i D1 i D 1 such that

N

X

i xi :

i D1

Corollary 2.22. Suppose that M is the set of all simple probability distributions on S

and that is a preference order on M that satisfies both the Archimedean and the independence axiom. Then there exists a von NeumannMorgenstern representation U .

Moreover, both U and u are unique up to positive affine transformations.

Proof. Let U be an affine numerical representation, which exists by Theorem 2.21.

We define u.x/ WD U.x /, for x 2 S. If

2 M is of the form

D 1 x1 C C

N xN , then affinity of U implies

U.

/ D

N

X

Z

i U.xi / D

u.x/

.dx/:

i D1

On a finite set S , every probability measure is simple. Thus, we obtain the following result as a special case.

Corollary 2.23. Suppose that M is the set of all probability distributions on a finite

set S and that is a preference order on M that satisfies both the Archimedean and

the independence axiom. Then there exists a von NeumannMorgenstern representation, and it is unique up to positive affine transformations.

For the proof of Theorem 2.21, we need the following auxiliary lemma. Its first assertion states that taking convex combination is monotone with respect to a preference

order satisfying our two axioms. Its second part can be regarded as an intermediate

value theorem for straight lines in M, and (c) is the analogue of the independence

axiom for the indifference relation .

Lemma 2.24. Under the assumptions of Theorem 2:21, the following assertions are

true.

(a) If

, then 7!

C .1 / is strictly increasing with respect to . More

precisely,

C .1 /

C .1 / for 0 < 1.

60

Chapter 2 Preferences

(b) If

and

, then there exists a unique 2 0; 1 with

C .1 /.

(c) If

, then

C .1 /
C .1 /
for all 2 0; 1 and all
2 M.

Proof. (a): Let
WD

C .1 /. The independence axiom implies that

C .1 / D . Hence, for WD =,

C .1 / D .1 /
C
.1 / C
D

C .1 /:

(b): Part (a) guarantees that is unique if it exists. To show existence, we need only

to consider the case

, for otherwise we can take either D 0 or D 1.

The natural candidate is

WD sup 2 0; 1 j

C .1 /:

If

C .1 / is not true, then one of the following two possibilities must

occur:

C .1 /; or

C .1 /:

(2.4)

In the first case, we apply the Archimedean axiom to obtain some 2 .0; 1/ such that

C .1 / C .1 /

D

C .1 /

(2.5)

C .1 / , which contradicts (2.5). If the second case in (2.4) occurs, the

Archimedean axiom yields some 2 .0; 1/ such that

.

C .1 // C .1 / D

C .1 / :

(2.6)

C .1 /. Part (a) and the fact that < imply that

C .1 /

C .1 /;

which contradicts (2.6).

(c): We must exclude both of the following two possibilities

C .1 / C .1 /

and

C .1 /

C .1 /
: (2.7)

; otherwise

the result is trivial. Let us assume that

; the case in which

is similar. Suppose that the first possibility in (2.7) would occur. The independence

axiom yields

C .1 / C .1 / D

61

C .1 / C .1 /

C .1 /

(2.8)

Using our assumption that the first possibilities in (2.7) is occurring, we obtain from

part (b) a unique 2 .0; 1/ such that, for any fixed ,

C .1 /
. C .1 / C .1 /
/ C .1 / C .1 /

D C .1 / C .1 /

C .1 /
;

where we have used (2.8) for replaced by in the last step. This is a contradiction.

The second possibility in (2.7) is excluded by an analogous argument.

Proof of Theorem 2:21. For the construction of U , we first fix two lotteries
and

with
and define

M.
; / WD

2 M j

I

the assertion is trivial if no such pair
exists. If

2 M.
; /, part (b) of

Lemma 2.24 yields a unique 2 0; 1 such that

C .1 /, and we put

U.

/ WD . To prove that U is a numerical representation of on M.
; /, we must

show that for ;

2 M.
; / we have U.

/ > U./ if and only if

. To prove

sufficiency, we apply part (a) of Lemma 2.24 to conclude that

U.

/
C .1 U.

// U./
C .1 U.// ;

Hence

. Conversely, if

then the preceding arguments already imply that

we cannot have U./ > U.

/. Thus, it suffices to rule out the case U.

/ D U./.

But if U.

/ D U./, then the definition of U yields

, which contradicts

.

We conclude that U is indeed a numerical representation of restricted to M.
; /.

Let us now show that M.
; / is a convex set. Take

; 2 M.
; / and 2 0; 1.

Then

C .1 /

C .1 /;

using the independence axiom to handle the cases
and

, and part (c)

of Lemma 2.24 for
and for

. By the same argument it follows that

Therefore, U.

C .1 // is well defined; we proceed to show that it equals

U.

/ C .1 /U./. To this end, we apply part (c) of Lemma 2.24 twice:

C .1 / .U.

/
C .1 U.

/// C .1 /.U./
C .1 U.///

D U.

/ C .1 /U./
C 1 U.

/ .1 /U./:

62

Chapter 2 Preferences

The definition of U and the uniqueness in part (b) of Lemma 2.24 imply that

U.

C .1 // D U.

/ C .1 /U./:

So U is indeed an affine numerical representation of on M.
; /.

In a further step, we now show that the affine numerical representation U on

M.
; / is unique up to positive affine transformations. So let UQ be another affine

numerical representation of on M.
; /, and define

UQ .

/ UQ ./

;

UO .

/ WD

UQ .
/ UQ ./

2 M. ; /:

UO . / D 1 D U. /. Hence, affinity of UO and the definition of U imply

UO .

/ D UO .U.

/ C .1 U.

/// D U.

/UO . / C .1 U.

//UO ./ D U.

/

for all

2 M. ; /. Thus UO D U .

Finally, we have to show that U can be extended as a numerical representation to

Q Q 2 M such that M. ;

Q /

the full space M. To this end, we first take ;

Q M. ; /.

By the arguments in the first part of this proof, there exists an affine numerical repQ /,

resentation UQ of on M. ;

Q and we may assume that UQ . / D 1 and UQ ./ D 0;

otherwise we apply a positive affine transformation to UQ . By the previous step of

the proof, UQ coincides with U on M. ; /, and so UQ is the unique consistent exQ /,

tension of U . Since each lottery belongs to some set M. ;

Q the affine numerical

representation U can be uniquely extended to all of M.

Remark 2.25. In the proof of the preceding theorem, we did not use the fact that

the elements of M are probability measures. All that was needed was convexity of

the set M, the Archimedean, and the independence axiom. Yet, even the concept of

convexity can be generalized by introducing the notion of a mixture space; see, e.g.,

[187], [115], or [151].

}

Let us now return to the problem of constructing a von NeumannMorgenstern representation for preference relations on distributions. If M is the set of all probability

measures on a finite set S , any affine numerical representation is already of this form,

as we saw in the proof of Corollary 2.23. However, the situation becomes more involved if we take an infinite set S . In fact, the following examples show that in this

case a von NeumannMorgenstern representation may not exist.

Exercise 2.2.1. Let M be the set of probability measures

on S WD 1; 2; : : : for

which U.

/ WD limk"1 k 2

.k/ exists as a finite real number. Show U is affine

and induces a preference order on M which satisfies both the Archimedean and the

independence axiom. Show next that U does not admit a von NeumannMorgenstern

representation.

}

63

Example 2.26. Let M be set the of all Borel probability measures on S D 0; 1, and

denote by
the Lebesgue measure on S . According to the Lebesgue decomposition

theorem, which is recalled in Theorem A.13, every

2 M can be decomposed as

D

s C

a ;

where

s is singular with respect to
, and

a is absolutely continuous. We define a

function U W M ! 0; 1 by

Z

U.

/ WD x

a .dx/:

It is easily seen that U is an affine function on M. Hence, U induces a preference

order on M which satisfies both the Archimedean and the independence axioms.

But cannot have a von NeumannMorgenstern representation: Since U.x / D 0

for all x, the only possible choice for u in (2.3) would be u 0. So the preference

relation would be trivial in the sense that

for all

2 M, in contradiction for

instance to U.
/ D 12 and U. 1 / D 0.

}

2

One way to obtain a von NeumannMorgenstern representation is to assume additional continuity properties of , where continuity is understood in the sense of

Definition 2.8. As we have already remarked, the Archimedean axiom holds automatically if taking convex combinations is continuous for the topology on M. This is

indeed the case for the weak topology on the set M1 .S; S/ of all probability measures

on a separable metric space S, endowed with the -field S of Borel sets. The space S

will be fixed for the rest of this section, and we will simply write M1 .S/ D M1 .S; S/.

Theorem 2.27. Let M WD M1 .S/ be the space of all probability measures on S

endowed with the weak topology, and let be a continuous preference order on M

satisfying the independence axiom. Then there exists a von NeumannMorgenstern

representation

Z

U.

/ D

u.x/

.dx/

are unique up to positive affine transformations.

Proof. Let Ms denote the set of all simple probability distributions on S . Since continuity of implies the Archimedean axiom, we deduce from Corollary 2.22 that

restricted to Ms has a von NeumannMorgenstern representation.

Let us show that the function u in this representation is bounded. For instance, if

u is not bounded from above, then there are x0 ; x1 ; : : : 2 S such that u.x0 / < u.x1 /

and u.xn / > n. Now let

1

1

n WD 1 p x0 C p xn :

n

n

64

Chapter 2 Preferences

Clearly,

n ! x0 weakly as n " 1. The continuity of together with the assumpp

tion that x1 x0 imply that x1

n for all large n. However, U.

n / > n for

all n, in contradiction to x1

n .

Suppose that the function u is not continuous. Then there exists some x 2 S

and a sequence .xn /n2N S such that xn ! x but u.xn / u.x/. By taking

a subsequence if necessary, we can assume that u.xn / converges to some number

a u.x/. Suppose that u.x/ a DW " > 0. Then there exists some m such that

ju.xn / aj < "=3 for all n m. Let

WD 12 .x C xm /. For all n m

U.x / D a C " > a C

2"

"

1

> .u.x/ C u.xm // D U.

/ > a C > U.xn /:

3

2

3

Therefore x

xn , although xn converges weakly to x , in contradiction to

the continuity of . The case u.x/ < a is excluded in the same manner.

Let us finally show that

Z

U.

/ WD u.x/

.dx/ for

2 M

defines a numerical representation of on all of M. Since u is bounded and continuous, U is continuous with respect to the weak topology on M. Moreover, Theorem A.38 states that Ms is a dense subset of the connected metrizable space M. So

the proof is completed by an application of Lemma 2.16.

The scope of the preceding theorem is limited insofar as it involves only bounded

functions u. This will not be flexible enough for our purposes. In the next section,

for instance, we will consider risk-averse preferences which are defined in terms of

concave functions u on the space S D R. Such a function cannot be bounded unless

it is constant. Thus, we must relax the conditions of the previous theorem. We will

present two approaches. In our first approach, we fix some point x0 2 S and denote

by B r .x0 / the closed metric ball of radius r around x0 . The space of boundedly

supported measures on S is given by

[

M1 .B r .x0 //

Mb .S/ WD

r>0

D

2 M1 .S/ j

.B r .x0 // D 1 for some r 0 :

Clearly, this definition does not depend on the particular choice of x0 .

Corollary 2.28. Let be a preference order on Mb .S/ whose restriction to each

space M1 .B r .x0 // is continuous with respect to the weak topology. If satisfies the

independence axiom, then there exists a von NeumannMorgenstern representation

Z

U.

/ D u.x/

.dx/

65

affine transformations.

Proof. Theorem 2.27 yields a von NeumannMorgenstern representation of the restriction of to M1 .B r .x0 // in terms of some continuous function ur W B r .x0 / !

R. The uniqueness part of the theorem implies that the restriction of ur to some

smaller ball B r 0 .x0 / must be a equal to ur 0 up to a positive affine transformation.

Thus, it is possible to find a unique continuous extension u W S ! R of ur 0 which defines a von NeumannMorgenstern representation of on each set M1 .B r .x0 //.

Our second variant of Theorem 2.27 includes measures with unbounded support,

but we need stronger continuity assumptions. Let be a continuous function with

values in 1; 1/ on the separable metric space S. We use as a gauge function and

define

Z

.x/

.dx/ < 1 :

M1 .S/ WD

2 M1 .S/

A suitable space of continuous test functions for measures in M1 .S/ is provided by

C .S/ WD f 2 C.S/ j 9 c W jf .x/j c

These test functions can now be used to define a topology on M1 .S/ in precisely

the same way one uses the set of bounded continuous function to define the weak

topology: A sequence .

n / in M1 .S/ converges to some

2 M1 .S/ if and only if

Z

Z

f d

n ! f d

for all f 2 C .S/.

To be rigorous, one should first define a neighborhood base for the topology and

then check that this topology is metrizable, so that it suffices indeed to consider the

convergence of sequences; the reader will find all necessary details in Appendix A.6.

We will call this topology the -weak topology on M1 .S/. If we take the trivial

case 1, C .S/ consists of all bounded continuous functions, and we recover the

standard weak topology on M11 .S/ D M1 .S/. However, by taking as some nonbounded function, we can also include von NeumannMorgenstern representations in

terms of unbounded functions u. The following theorem is a version of Theorem 2.27

for the -weak topology. Its proof is analogous to that of Theorem 2.27, and we leave

it to the reader to fill in the details.

Theorem 2.29. Let be a preference order on M1 .S/ that is continuous in the weak topology and satisfies the independence axiom. Then there exists a numerical

representation U of von NeumannMorgenstern form

Z

U.

/ D u.x/

.dx/

66

Chapter 2 Preferences

transformations.

So far, we have presented the classical theory of expected utility, starting with the

independence axiom and the Archimedean axiom. However, it is well known that in

reality people may not behave according to this paradigm.

Example 2.30 (Allais paradox). The so-called Allais paradox questions the descriptive aspect of expected utility by considering the following lotteries. Lottery

1 D 0:33 2500 C 0:66 2400 C 0:01 0

yields 2500 C with a probability of 0.33, 2400 C with probability 0.66, and draws a

blank with the remaining probability of 0.01. Lottery

1 WD 2400

yields 2400 C for sure. When asked, most people prefer the sure amount even

though lottery 1 has the larger expected value, namely 2409 C.

Next, consider the following two lotteries

2 and 2 :

and

2 , in accordance with

the expectations of 2 and

2 , which are 825 C and 816 C, respectively.

This observation is due to M. Allais [5]. It was confirmed by D. Kahnemann and

A. Tversky [165] in empirical tests where 82 % of interviewees preferred

1 over

1 while 83 % chose 2 rather than

2 . This means that at least 65 % chose both

1 1 and 2

2 . As pointed out by M. Allais, this simultaneous choice leads

to a paradox in the sense that it is inconsistent with the von NeumannMorgenstern paradigm. More precisely, any preference relation for which

1 1 and

2

2 are both valid violates the independence axiom, as we will show now. If the

independence axiom were satisfied, then necessarily

1 C .1 /2 1 C .1 /2 1 C .1 /

2

for all 2 .0; 1/. By taking D 1=2 we would arrive at

1

1

.

1 C 2 / .1 C

2 /

2

2

which is a contradiction to the fact that

1

1

.

1 C 2 / D .1 C

2 /:

2

2

Therefore, the independence axiom was violated by at least 65 % of the people who

were interviewed. This effect is empirical evidence against the von NeumannMorgenstern theory as a descriptive theory. Even from a normative point of view, there

are good reasons to go beyond our present setting, and this will be done in Section 2.5.

In particular, we will take a second look at the Allais paradox in Remark 2.72.

}

67

2.3

Expected utility

In this section, we focus on individual financial assets under the assumption that their

payoff distributions at a fixed time are known, and without any regard to hedging

opportunities in the context of a financial market model. Such asset distributions may

be viewed as lotteries with monetary outcomes in some interval on the real line. Thus,

we take M as a fixed set of Borel probability measures on a fixed interval S R. In

this setting, we discuss the paradigm of expected utility in its standard form, where the

function u appearing in the von NeumannMorgenstern representation has additional

properties suggested by the monetary interpretation. We introduce risk aversion and

certainty equivalents, and illustrate these notions with a number of examples.

Throughout this section, we assume that M is convex and contains all point masses

x for x 2 S. We assume also that each

2 M has a well-defined expectation

Z

m.

/ WD

x

.dx/ 2 R:

Remark 2.31. For an asset whose (discounted) random payoff has a known distribution

, the expected value m.

/ is often called the fair price of the asset. For

an insurance contract where

is the distribution of payments to be received by the

insured party in dependence of some random damage within a given period, the expected value m.

/ is also called the fair premium. Typically, actual asset prices and

actual insurance premiums will be different from these values. In many situations,

such differences can be explained within the conceptual framework of expected utility, and in particular in terms of risk aversion.

}

Definition 2.32. A preference relation on M is called monotone if

x > y implies x y .

The preference relation is called risk averse if for

2 M

m./

unless

D m./ .

which admit a von NeumannMorgenstern representation.

Proposition 2.33. Suppose the preference relation has a von NeumannMorgenstern representation

Z

U.

/ D u d

:

68

Chapter 2 Preferences

Then

(a) is monotone if and only if u is strictly increasing.

(b) is risk averse if and only if u is strictly concave.

Proof. (a): Monotonicity is equivalent to

u.x/ D U.x / > U.y / D u.y/

for x > y.

xC.1/y x C .1 /y

holds for all distinct x; y 2 S and 2 .0; 1/. Hence,

u.x C .1 /y/ > u.x/ C .1 /u.y/;

i.e., u is strictly concave. Conversely, if u is strictly concave, then Jensens inequality

implies risk aversion

Z

Z

x

.dx/ u.x/

.dx/ D U.

/

U.m./ / D u

with equality if and only if

D m./ .

Remark 2.34. In view of the monetary interpretation of the state space S, it is natural to assume that the preference relation is monotone. The assumption of risk

aversion is more debatable, at least from a descriptive point of view. In fact, there

is considerable empirical evidence that agents tend to switch between risk aversion

and risk seeking behavior, depending on the context. In particular, they may be risk

averse after prior gains, and they may become risk seeking if they see an opportunity

to compensate prior losses. Tversky and Kahneman [261] propose to describe such a

behavioral pattern by a function u of the form

for x c;

.x c/

(2.9)

u.x/ D

for x < c;

.c x/

where c is a given benchmark level, and their experiments suggest parameter values

around 2 and slightly less than 1. Nevertheless, one can insist on risk aversion from

a normative point of view, and in the sequel we explore some of its consequences. }

Definition 2.35. A function u W S ! R is called a utility function if it is strictly

concave, strictly increasing, and continuous on S. A von NeumannMorgenstern

representation

Z

U.

/ D

u d

(2.10)

69

interval .a; b S ; see Proposition A.4. Hence, the condition of continuity in the

preceding definition is only relevant if S contains its lower boundary point. Note that

any utility function u.x/ decreases at least linearly as x # inf S. Therefore, u cannot

be bounded from below unless inf S > 1.

From now on, we will consider a fixed preference relation on M which admits a

von NeumannMorgenstern representation

Z

U.

/ D u d

value theorem applied to the function u yields for any

2 M a unique real number

c.

/ for which

Z

u.c.

// D U.

/ D

u d

:

(2.11)

It follows that

c./

;

i.e., there is indifference between the lottery

and the sure amount of money c.

/.

Definition 2.36. The certainty equivalent of the lottery

2 M with respect to u is

defined as the number c.

/ of (2.11), and

%.

/ WD m.

/ c.

/

is called the risk premium of

.

When u is a utility function, risk aversion implies that

u.c.

// D U.

/ < U.m./ / D u.m.

//

for every lottery

with

m./ . Hence, monotonicity yields that

c.

/ < m.

/

for

m./ .

/ associated with a utility function is always nonnegative, and it is strictly positive as soon as the distribution

carries any risk.

Remark 2.37. The certainty equivalent c.

/ can be viewed as an upper bound for any

price of

which would be acceptable to an economic agent with utility function u.

Thus, the fair price m.

/ must be reduced at least by the risk premium %.

/ if one

wants the agent to buy the asset distribution

. Alternatively, suppose that the agent

holds an asset with distribution

. Then the risk premium may be viewed as the

amount that the agent would be ready to pay for replacing the asset by its expected

value m.

/.

}

70

Chapter 2 Preferences

1

X

2n 2n1

nD1

which may be viewed as the payoff distribution of the following game. A fair coin

is tossed until a head appears. If the head appears on the nth toss, the payoff will be

2n1 C. Up to the early 18th century, it was commonly accepted that the price of a

lottery should be computed as the fair price, i.e., as the expected value m.

/. In the

present example, the fair price is given by m.

/ D 1, but it is hard to find someone

who is ready to pay even 20 C. In view of this paradox, posed by Nicholas Bernoulli

in 1713, Gabriel Cramer and Daniel Bernoulli [25] independently introduced the idea

of determining an acceptable price as the certainty equivalent with respect to some

utility function. For the two utility functions

p

u1 .x/ D x and u2 .x/ D log x

proposed, respectively, by G. Cramer and by D. Bernoulli, these certainty equivalents

are given by

p

c1 .

/ D .2 2 /2 2:91 and c2 .

/ D 2;

and this is within the range of prices people are usually ready to pay. Note, however,

that for any utility function which is unbounded from above we could modify the

payoff in such a way that the paradox reappears.

R For example, we could replace the

payoff 2n by u1 .2n / for n 1000, so that u d

D C1. The choice of a utility

function that is bounded from above would remove this difficulty, but would create

others; see the discussion on pp. 7780.

}

Given the preference order on M, we can now try to determine those distributions in M which are maximal with respect to . As a first illustration, consider the

following simple optimization problem. Let X be an integrable random variable on

some probability space .; F ; P / with nondegenerate distribution

2 M. We assume that X is bounded from below by some number a in the interior of S. Which is

the best mix

X

WD .1
/X C
c

of the risky payoff X and the certain amount c, that also belongs to the interior of S?

If we evaluate X

by its expected utility E u.X

/ and denote by

the distribution

of X

under P , then we are looking for a maximum of the function f on 0; 1 defined

by

Z

f .
/ WD U.

/ D

u d

D E u..1 /X C c/ :

When u is a utility function, f is strictly concave and attains its maximum in a unique

point
2 0; 1.

71

(a) We have D 1 if E X c, and > 0 if c c.

/.

(b) If u is differentiable, then

D 1

E X c

D 0

c

and

E Xu0 .X /

:

E u0 .X /

f . / u.E X

/ D u..1 /E X C c/;

with equality if and only if D 1. It follows that D 1 if the right-hand side is

increasing in , i.e., if E X c.

Strict concavity of u implies

f . / E .1 /u.X / C u.c/

D .1 /u.c.

// C u.c/;

with equality if and only if 2 0; 1. The right-hand side is increasing in if

c c.

/, and this implies > 0.

(b): Clearly, we have D 0 if and only if the right-hand derivative fC0 of f

satisfies fC0 .0/ 0; see Appendix A.1 for the definition of fC0 and f0 . Note that the

difference quotients

u.X

/ u.X /

u.X

/ u.X /

D

.c X /

X

X

are P -a.s. bounded by

u0C .a ^ c/jc Xj 2 L1 .P /

and that they converge to

u0C .X /.c X /C u0 .X /.c X /

as # 0. By Lebesgues theorem, this implies

fC0 .0/ D E u0C .X /.c X /C E u0 .X /.c X / :

If u is differentiable, or if the countable set x j u0C .x/ u0 .x/ has

-measure 0,

then we can conclude

fC0 .0/ D E u0 .X /.c X / ;

72

Chapter 2 Preferences

c

E Xu0 .X /

:

E u0 .X /

f0 .1/ D u0 .c/E .X c/ u0C .c/E .X c/C :

If u is differentiable at c, then we can conclude

f0 .1/ D u0 .c/.c E X /:

This implies f0 .1/ < 0, and hence < 1, if and only if E X > c.

Exercise 2.3.1. As above, let X be a random variable with a nondegenerate distribution

2 M. Show that for a differentiable utility function u we have

m.

/ > c.

/ >

E u0 .X /X

:

E u0 .X /

(2.12)

}

Example 2.40 (Demand for a risky asset). Let S D S 1 be a risky asset with price

D 1 . Given an initial wealth w, an agent with utility function u 2 C 1 can invest a

fraction .1
/w into the asset and the remaining part
w into a risk-free bond with

interest rate r. The resulting payoff is

X

D

.1
/w

.S / C
w r:

The preceding proposition implies that there will be no investment into the risky asset

if and only if

S

:

E

1Cr

In other words, the price of the risky asset must be below its expected discounted

payoff in order to attract any risk averse investor, and in that case it will indeed be

optimal for the investor to invest at least some amount. Instead of the simple linear

profiles X

, the investor may wish to consider alternative forms of investment. For

example, this may involve derivatives such as maxS; K D K C .S K/C for some

threshold K. In order to discuss such non-linear payoff profiles, we need an extended

formulation of the optimization problem; see Section 3.3 below.

}

73

Example 2.41 (Demand for insurance). Suppose an agent with utility function u 2

C 1 considers taking at least some partial insurance against a random loss Y , with

0 Y w and P Y E Y > 0, where w is a given initial wealth. If insurance

of
Y is available at the insurance premium
, the resulting final payoff is given by

X

WD w Y C
.Y / D .1
/.w Y / C
.w /:

By Proposition 2.39, full insurance is optimal if and only if E Y . In reality,

however, the insurance premium will exceed the fair premium E Y . In this

case, it will be optimal to insure only a fraction
Y of the loss, with
2 0; 1/.

This fraction will be strictly positive as long as

<

E .w Y /u0 .w Y /

E Y u0 .w Y /

D

w

:

E u0 .w Y /

E u0 .w Y /

Since the right-hand side is strictly larger than E Y due to (2.12), risk aversion may

create a demand for insurance even if the insurance premium lies above the fair

price E Y . As in the previous example, the agent may wish to consider alternative forms of insurance such as a stop-loss contract whose payoff has the non-linear

}

structure .Y K/C of a call option.

Let us take another look at the risk premium %.

/ of a lottery

. For an approximate

calculation, we consider the Taylor expansion of a sufficiently smooth and strictly

increasing function u.x/ at x D c.

/ around m WD m.

/, and we assume that

has

finite variance var.

/. On the one hand,

u.c.

// u.m/ C u0 .m/.c.

/ m/ D u.m/ u0 .m/%.

/:

On the other hand,

Z

u.c.

// D u.x/

.dx/

Z

1

D

u.m/ C u0 .m/.x m/ C u00 .m/.x m/2 C r.x/

.dx/

2

1

u.m/ C u00 .m/ var.

/;

2

where r.x/ denotes the remainder term in the Taylor expansion of u. It follows that

%.

/

u00 .m.

//

1

var.

/ DW .m.

// var.

/:

0

2 u .m.

//

2

(2.13)

Thus, .m.

// is the factor by which an economic agent with von NeumannMorgenstern preferences described by u weighs the risk, measured by 12 var.

/, in order

to determine the risk premium he or she is ready to pay.

74

Chapter 2 Preferences

increasing function on S . Then

.x/ WD

u00 .x/

u0 .x/

Example 2.43. The following classes of utility functions u and their corresponding

coefficients of risk aversion are standard examples.

(a) Constant absolute risk aversion (CARA): .x/ equals some constant > 0.

Since .x/ D .log u0 /0 .x/, it follows that u.x/ D a b e x . Using an

affine transformation, u can be normalized to

u.x/ D 1 e x :

(b) Hyperbolic absolute risk aversion (HARA): .x/ D .1 /=x on S D .0; 1/

for some < 1. Up to affine transformations, we have

u.x/ D log x

for D 0,

1

x

for 0.

u.x/ D

Sometimes, these functions are also called CRRA utility functions, because their

relative risk aversion x.x/ is constant. Of course, these utility functions can

be shifted to any interval S D .a; 1/. The risk-neutral limiting case D 1

would correspond to an affine function u.

}

Exercise 2.3.2. Compute the coefficient of risk aversion for the S-shaped utility function in (2.9). Sketch the graphs of u and its risk aversion for
D 2 and D 0:9. }

Proposition 2.44. Suppose that u and uQ are two strictly increasing functions on S

which are twice continuously differentiable, and that and Q are the corresponding

ArrowPratt coefficients of absolute risk aversion. Then the following conditions are

equivalent:

(a) .x/ .x/

Q

for all x 2 S.

(b) u D F uQ for a strictly increasing concave function F .

(c) The respective risk premiums % and %Q associated with u and uQ satisfy %.

/

%.

/

Q

for all

2 M.

75

Proof. (a) ) (b): Since uQ is strictly increasing, we may define its inverse function,

w. Then F .t / WD u.w.t // is strictly increasing, twice differentiable, and satisfies

u D F u.

Q For showing that F is concave we calculate the first two derivatives of w

w0 D

1

;

uQ 0 .w/

Q

w 00 D .w/

1

:

uQ 0 .w/2

F 0 D u0 .w/ w 0 D

u0 .w/

>0

uQ 0 .w/

and

F 00 D u00 .w/.w 0 /2 C u0 .w/w 00

D

u0 .w/

.w/

Q

.w/

uQ 0 .w/2

(2.14)

0:

This proves that F is concave.

(b) ) (c): Jensens inequality implies that the respective certainty equivalents c.

/

and c.

/

Q

satisfy

Z

Z

u.c.

// D u d

D F uQ d

(2.15)

Z

F

uQ d

D F . u.

Q c.

///

Q

D u. c.

//:

Q

Hence, %.

/ D m.

/ c.

/ m.

/ c.

/

Q

D %.

/.

Q

(c) ) (a): If condition (a) is false, there exists an open interval O S such that

.x/

Q

> .x/ for all x 2 O. Let OQ WD u.O/,

Q

and denote again by w the inverse

of u.

Q Then the function F .t / D u.w.t // will be strictly convex in the open interval

OQ by (2.14). Thus, if

is a measure with support in O, the inequality in (2.15) is

reversed and is even strict unless

is concentrated at a single point. It follows that

%.

/ < %.

/,

Q

which contradicts condition (c).

As an application of the preceding proposition, we will now investigate the structure of those continuous and strictly increasing functions u on R whose associated

certainty equivalents have the following translation property:

c.

t / D c.

/ C t

for all

2 M and all t 2 R,

t of

2 M by t 2 R is defined by

Z

Z

g.x/

t .dx/ D g.x C t /

.dx/ for bounded measurable g.

(2.16)

76

Chapter 2 Preferences

t 2 M for all

2 M

and t 2 R.

Lemma 2.45. Suppose the certainty equivalent associated with a continuous and

strictly increasing function u W R ! R satisfies the translation property (2.16). Then

u belongs to C 1 .R/.

Proof. Let denote the Lebesgue measure on 0; 1. Then

Z

f .t / WD u.c. / C t / D u.c. t // D

1Ct

u d t D

u.y/ dy;

(2.17)

f 0 .t / D u.1 C t / u.t /:

(2.18)

hence f 2 C 2 .R/. Iterating the argument we get u 2 C 1 .R/.

The following proposition implies in particular that a utility function u that satisfies

the translation property (2.16) is necessarily a CARA utility function of exponential

type as in part (a) of Example 2.43.

Proposition 2.46. Suppose the certainty equivalent associated with a continuous and

strictly increasing function u W R ! R satisfies the translation property (2.16). Then

u has constant absolute risk aversion and is hence either linear or an exponential

function. More precisely, there are constants a 2 R and b; > 0 such that u.x/

equals one of the following three functions

8

x

<a be

a C bx

:

a C be x :

Proof. For t 2 R let u t .x/ WD u.x C t /, and denote by c t .

/ the corresponding

certainty equivalent. For

2 M we have

Z

Z

Z

u t ..c t .

// D u t d

D u.x C t /

.dx/ D u d

t D u.c.

t //

D u.c.

/ C t / D u t .c.

//:

It follows that c t .

/ D c.

/ for all t 2 R and

2 M. Therefore,

% t .

/ WD m.

/ c t .

/ D m.

/ c.

/ D %.

/

77

for all t 2 R. Since u is smooth by Lemma 2.45, we may apply Proposition 2.44

to conclude that the respective Arrow-Pratt coefficients t .x/ D u00t .x/=u0t .x/ and

.x/ D u00 .x/=u0 .x/ are equal for all t and x. But t .x/ D .x C t /, and so .x/

does not depend on x. When > 0, we see as in Example 2.43 that u is of the form

u.x/ D a be x . When D 0, u must be linear. And when < 0, we must have

u.x/ D a C be x .

Now we focus on the case in which u is a utility function and preferences have the

expected utility representation (2.10). In view of the underlying axioms, the paradigm

of expected utility has a certain plausibility on a normative level, i.e., as a guideline

of rational behavior in the face of risk. But this guideline should be applied with

care: If pushed too far, it may lead to unplausible conclusions. In the remaining part

of this section we discuss some of these issues. From now on, we assume that S is

unbounded from above, so that w C x 2 S for any x 2 S and w 0. So far, we

have implicitly assumed that the preference relation on lotteries reflects the views

of an economic agent in a given set of conditions, including a fixed level w 0

of the agents initial wealth. In particular, the utility function may vary as the level

of wealth changes, and so it should really be indexed by w. Usually one assumes

that uw is obtained by simply shifting a fixed utility function u to the level w, i.e.,

uw .x/ WD u.w C x/. Thus, a lottery

is declined at a given level of wealth w if and

only if

Z

u.w C x/

.dx/ < u.w/:

Let us now return to the situation of Proposition 2.39 when

is the distribution of an

integrable random variable X on .; F ; P /, which is bounded from below by some

number a in the interior of S . We view X as the net payoff of some financial bet, and

we assume that the bet is favorable in the sense that

m.

/ D E X > 0:

Remark 2.47. Even though the favorable bet X might be declined at a given level

w due to risk aversion, it follows from Proposition 2.39 that it would be optimal to

accept the bet at some smaller scale, i.e., there is some > 0 such that

E u.w C X / > u.w/:

On the other hand, it follows from Proposition 2.49 below that the given bet X becomes acceptable at a sufficiently high level of wealth whenever the utility function is

unbounded from above.

}

Sometimes it is assumed that some favorable bet is declined at every level of wealth.

The assumption that such a bet exists is not as innocent as it may look. In fact it has

rather drastic consequences. In particular, we are going to see that it rules out all

utility functions in Example 2.43 except for the class of exponential utilities.

78

Chapter 2 Preferences

Example 2.48. For any exponential utility function u.x/ D 1 e x with constant

risk aversion > 0, the induced preference order on lotteries does not at all depend

on the initial wealth w. To see this, note that

Z

Z

u.w C x/

.dx/ < u.w C x/ .dx/

is equivalent to

e x

.dx/ >

e x .dx/:

at every wealth level w

leads to a not quite plausible conclusion: At high levels of wealth, the agent would

reject a bet with huge potential gain even though the potential loss is just a negligible

fraction of the initial wealth.

Proposition 2.49. If the favorable bet

is rejected at any level of wealth, then the

utility function u is bounded from above, and there exists A > 0 such that the bet

1

WD .A C 1 /

2

is rejected at any level of wealth.

Proof. We have assumed that X is bounded from below, i.e.,

is concentrated on

a; 1/ for some a < 0, where a is in the interior of S. Moreover, we can choose

b > 0 such that

.B/

Q

WD

.B \ a; b/ C b .B/

..b; 1//

is still favorable. Since u is increasing, we have

Z

Z

u.w C x/

.dx/

Q

u.w C x/

.dx/ < u.w/

for any w 0, i.e., also the lottery

Q is rejected at any level of wealth. It follows that

Z

Z

u.w C x/ u.w/

.dx/

Q

<

u.w/ u.w C x/

.dx/:

Q

0;b

a;0/

Let us assume for simplicity that u is differentiable; the general case requires only

minor modifications. Then the previous inequality implies

u0 .w C b/ mC .

/

Q < u0 .w C a/ m .

/;

Q

where

Q WD

mC .

/

Z

x

.dx/

Q

>

0;b

a;0

.x/

.dx/

Q

DW m .

/;

Q

79

Q is favorable. Thus,

Q

m .

/

u0 .w C b/

<

DW < 1

0

C

u .w jaj/

m .

/

Q

for any w, hence

u0 .x C n.jaj C b// < n u0 .x/

for any x in the interior of S . This exponential decay of the derivative implies

u.1/ WD limx"1 u.x/ < 1. More precisely, if A WD n.jaj C b/ for some n,

then

1 Z xC.kC1/A

X

u0 .y/ dy

u.1/ u.x/ D

xCkA

kD0

1 Z

X

<

u0 .z C .k C 1/A/ dz

xA

kD0

1

X

.kC1/n

u0 .z/ dz

xA

kD0

n

.u.x/ u.x A//:

1 n

u.1/ u.x/ < u.x/ u.x A/;

i.e.,

1

.u.1/ C u.x A// < u.x/

2

for all x such that x A 2 S .

Example 2.50. For an exponential utility function u.x/ D 1ex , the bet defined

in the preceding lemma is rejected at any level of wealth as soon as A > 1 log 2. }

Suppose now that the lottery

2 M is played not only once but n times in a row.

For instance, one can think of an insurance company selling identical policies to a

large number of individual customers. More precisely, let .; F ; P / be a probability space supporting a sequence X1 ; X2 ; : : : of independent random variables with

common distribution

. The value of Xi will be interpreted as the outcome of the

i th drawing of the lottery

. The accumulated payoff of n successive independent

repetitions of the financial bet X1 is given by

Zn WD

n

X

i D1

Xi ;

80

Chapter 2 Preferences

and we assume that this accumulated payoff takes values in S; this is the case if, e.g.,

S D 0; 1/.

Remark 2.51. It may happen that an agent refuses the single favorable bet X at any

level of wealth but feels tempted by a sufficiently large series X1 ; : : : ; Xn of independent repetitions of the same bet. It is true that, by the weak law of large numbers, the

probability

n

h1X

i

Xi < m.

/ "

P Zn < 0 D P

n

i D1

(for " WD m.

/) of incurring a cumulative loss at the end of the series converges to 0

as n " 1. Nevertheless, the decision of accepting n repetitions is not consistent with

the decision to reject the single bet at any wealth level w. In fact, for Wk WD w C Zk

we obtain

E u.Wn / D E E u.Wn1 C Xn / j X1 ; : : : ; Xn1

Z

DE

u.Wn1 C x/

.dx/

< E u.Wn1 / < < u.w/;

i.e., the bet described by Zn should be rejected as well.

Let us denote by

n the distribution of the accumulated payoff Zn . The lottery

n / D n m.

/, the certainty equivalent c.

n /, and the associated risk premium %.

n / D n m.

/ c.

n /. We are interested in the asymptotic

behavior of these quantities for large n. Kolmogorovs law of large numbers states

that the average outcome n1 Zn converges P -a.s. to the constant m.

/. Therefore, one

might guess that a similar averaging effect occurs on the level of the relative certainty

equivalents

c.

n /

(2.19)

cn WD

n

and of the relative risk premiums

%.

n /

D m.

/ cn :

n

Does cn converge to m.

/, and is there a successive reduction of the relative risk

premiums %n as n grows to infinity? Applying our heuristic (2.13) to the present

situation yields

%n WD

1

1

.m.

n // var.

n / D .n m.

// var.

/:

2n

2

Thus, one should expect that %n tends to zero only if the ArrowPratt coefficient .x/

becomes arbitrarily small as x becomes large, i.e., if the utility function is decreasingly risk averse. This guess is confirmed by the following two examples.

%n

81

function with conR utility

x

.dx/ < 1. Then,

stant risk aversion > 0 and assume that

is such that e

with the notation introduced above,

Z

e

x

n .dx/ D E

n

h Y

Xi

Z

D

x

n

.dx/ :

i D1

n is given by

Z

n

c.

n / D log e x

.dx/ D n c.

/:

It follows that cn and %n are independent of n. In particular, the relative risk premiums

are not reduced if the lottery is drawn more than once.

}

The second example displays a different behavior. It shows that for HARA utility

functions the relative risk premiums will indeed decrease to 0. In particular, the lottery

n will become attractive for large enough n as soon as the price of the single lottery

is less than m.

/.

Example 2.53. Suppose that

is a non-degenerate lottery concentrated on .0; 1/,

and that u is a HARA utility function of index 2 0; 1/. If > 0 then u.x/ D 1 x

and c.

n / D E .Zn / 1= , hence

cn D

c.

n /

DE

n

1

Zn

n

1=

< m.

/:

1

Zn :

log cn D log c.

n / log n D E log

n

Thus, we have

1

u.cn / D E u Zn

n

1

ZnC1 D E Xk j ZnC1

nC1

for k D 1; : : : ; n C 1;

1

1

ZnC1 D E

Zn ZnC1 :

nC1

n

(2.20)

82

Chapter 2 Preferences

is non-degenerate, we get

Zn ZnC1

u.cnC1 / D E u E

n

1

> E E u Zn ZnC1

n

D u.cn /;

i.e., the relative certainty equivalents are strictly increasing and the relative risk premiums %n are strictly decreasing. By Kolmogorovs law of large numbers,

1

Zn ! m.

/ P -a.s.

n

(2.21)

is concentrated on "; 1/

for some " > 0 if D 0),

1

lim inf u.cn / E lim inf u Zn

D u.m.

//;

n

n"1

n"1

hence

lim cn D m.

/

n"1

and

lim %n D 0:

n"1

is given by 2 .c.

/; m.

//. At initial wealth w D 0,

the agent would decline a single bet. But, in contrast to the situation in Remark 2.51, a

series of n repetitions of the same bet would now become attractive for large enough n,

since c.

n / D ncn > n for

n n0 WD mink 2 N j ck > < 1:

1

1

nC1

n

where AnC1 D .ZnC1 ; ZnC2 ; : : : /. This means that the stochastic process n1 Zn ,

n D 1; 2; : : : , is a backwards martingale, sometimes also called reversed martingale.

In particular, Kolmogorovs law of large numbers (2.21) can be regarded as a special case of the convergence theorem for backwards martingales; see part II of 20

in [20].

}

Exercise 2.3.3. Investigate the asymptotics of cn in (2.19) for a HARA utility func}

tion u.x/ D 1 x with < 0.

83

2.4

Stochastic dominance

fixed utility function u. In this section, we focus on the question whether one distribution is preferred over another, regardless of the choice of a particular utility function.

For simplicity, we take S D R as the set of possible payoffs. Let M be the set of

all

2 M1 .R/ with well-defined and finite expectation

Z

m.

/ D x

.dx/:

Recall from Definition 2.35 that a utility function on R is a strictly concave and strictly

increasing function u W R ! R. Since each concave function u is dominated

R by an

affine function, the existence of m.

/ implies the existence of the integral u d

as

an extended real number in 1; 1/.

Definition 2.55. Let and

be lotteries in M. We say that the lottery

is uniformly

preferred over and we write

<uni

if

Z

u d

u d

Thus,

<uni holds if and only if every risk-averse agent will prefer

over ,

regardless of which utility function the agent is actually using. In this sense,

<uni

expresses a uniform preference for

over . Sometimes, <uni is also called second

order stochastic dominance; the notion of first order stochastic dominance will be

introduced in Definition 2.67.

Remark 2.56. The binary relation <uni is a partial order on M, i.e., <uni satisfies the

following three properties:

Reflexivity:

<uni

for all

2 M.

Transitivity:

<uni and <uni
imply

<uni
.

Antisymmetry:

<uni and <uni

imply

D .

The first two properties are obvious, the third is derived in Remark 2.58. Moreover,

<uni is monotone and risk-averse in the sense that

y <uni x for y x, and

m./ <uni

for all

2 M.

Note, however, that <uni is not a weak preference relation in the sense of Definition 2.2,

since it is not complete, see Remark 2.3.

}

84

Chapter 2 Preferences

statement

<uni . One of them needs the notion of a stochastic kernel on R. This is

a mapping

Q W R ! M1 .R/

such that x 7! Q.x; A/ is measurable for each fixed Borel set A R. See Appendix A.3 for the notion of a quantile function, which will be used in condition (e).

Theorem 2.57. For any pair

; 2 M the following conditions are equivalent:

(a)

<uni .

R

R

(b) f d

f d for all increasing concave functions f .

(c) For all c 2 R

.c x/

.dx/

.c x/C .dx/:

and , then

Z

F .x/ dx

F .x/ dx

1

for all c 2 R.

1

and , then

Z

q .s/ ds

0

q .s/ ds

for 0 < t 1.

(f) There exists a probability space .; F ; P / with random variables X and X

having respective distributions

and such that

E X j X X

P -a.s.

(g) There exists a stochastic kernel Q.x; dy/ on R such that Q.x; / 2 M and

m.Q.x; // x for all x and such that D

Q, where

Q denotes the measure

Z

Q.A/ WD Q.x; A/

.dx/ for Borel sets A R.

Below we will show the following implications between the conditions of the theorem:

(e) (d) (c) (b) (a) (H (g) (H (f):

(2.22)

The difficult part is the proof that (b) implies (f). It will be deferred to Section 2.6,

where we will prove a multidimensional variant of this result; cf. Theorem 2.94.

85

(d) , (c): By Fubinis theorem,

Z c Z

Z c

F .y/ dy D

.dz/ dy

1

1

Z Z

.1;y

Izyc dy

.dz/

D

Z

D

.c z/C

.dz/:

(c) , (b): Condition (b) implies (c) because f .x/ WD .c x/C is concave and

increasing. In order to prove the converse assertion, we take an increasing concave

function f and let h WD f . Then h is convex and decreasing, and its increasing

right-hand derivative h0 WD h0C can be regarded as a distribution function of a nonnegative Radon measure on R,

h0 .b/ D h0 .a/ C ..a; b/

see Appendix A.1. As in (1.11):

for a < b;

.z x/C .dz/

for x < b:

.1;b

Z

Z

h d

D h.b/

..1; b/ h0 .b/ .b x/C

.dx/

.1;b

.z x/C

.dx/ .dz/

.1;b

Z

h.b/

..1; b/ h0 .b/ .b x/C .dx/

Z

Z

C

.z x/C .dx/ .dz/

Z

D

.1;b

h d C h.b/

..1; b/ ..1; b/:

.1;b

R

R

Taking b " 1 yields f d

f d. Indeed, the convex decreasing function h

decays at most linearly, and the existence of first moments for

and implies that

b

..1; b/ ! 0 and b..1; b/ ! 0 for b " 1.

(a) , (b): That (b) implies (a) is obvious. For Rthe proof of the

R converse implication,

choose any utility function u0 for which both u0 d

and u0 d are finite. For

instance, one can take

x e x=2 C 1 if x 0;

u0 .x/ WD p

x C 1 1 if x 0.

86

Chapter 2 Preferences

u .x/ WD f .x/ C .1 /u0 .x/

is a utility function. Hence,

Z

Z

Z

Z

f d

D lim u d

lim u d D f d:

"1

"1

(f) ) (g): By considering the joint distribution of X and X , we may reduce our

setting to the situation in which D R2 and where X and X are the respective

projections on the first and second coordinates, i.e., for ! D .x; y/ 2 D R2 we

have X .!/ D x and X .!/ D y. Let Q.x; dy/ be a regular conditional distribution

of X given X , i.e., a stochastic kernel on R such that

P X 2 A j X .!/ D Q.X .!/; A/

for all Borel sets A R and for P -a.e. ! 2 (see, e.g., Theorem 44.3 of [20] for an

existence proof). Clearly, D

Q. Condition (f) implies that

Z

X .!/ E X j X .!/ D y Q.X .!/; dy/ for P -a.e. ! 2 .

Hence, Q satisfies

Z

y Q.x; dy/ x

for

-a.e. x.

By modifying Q on a

-null set (e.g., by putting Q.x; / WD x there), this inequality

can be achieved for all x 2 R.

(g) ) (a): Let u be a utility function. Jensens inequality applied to the measure

Q.x; dy/ implies

Z

u.y/ Q.x; dy/ u.m.Q.x; /// u.x/:

Hence,

Z Z

u d D

Z

u.y/ Q.x; dy/

.dx/

u d

;

Remark 2.58. Let us note some consequences of the preceding theorem. First, taking

in condition (b) the increasing concave function f .x/ D x yields

m.

/ m./ if

<uni ,

i.e., the expectation m./ is increasing with respect to <uni .

87

and are such that

Z

Z

.c x/C

.dx/ D .c x/C .dx/ for all c.

Then we have both

<uni and <uni

, and condition (d) of the theorem implies that

the respective distribution functions satisfy

Z c

Z c

F .x/ dx D

F .x/ dx for all c.

1

1

D , i.e., a measure

2 M is

uniquely determined by the integrals .c x/C

.dx/ for all c 2 R. In particular,

}

<uni is antisymmetric.

The following proposition characterizes the partial order <uni considered on the set

of all normal distributions N.m; 2 /. Recall that the standard normal distribution

N.0; 1/ is defined by its density function

1

2

'.x/ D p e x =2 ;

2

x 2 R:

Z x

'.y/ dy; x 2 R:

.x/ D

1

More generally, the normal distribution N.m; 2 / with mean m 2 R and variance

2 > 0 is given by the density function

p

.x m/2

exp

;

2 2

2 2

1

x 2 R:

Q Q 2 / if

Proposition 2.59. For two normal distributions, we have N.m; 2 / <uni N.m;

2

2

and only if both m m

Q and Q hold.

Proof. In order to prove necessity, note that N.m; 2 / <uni N.m;

Q Q 2 / implies that

Z

Z

2 2

e mC =2 D e x N.m; 2 /.dx/ e x N.m;

Q Q 2 /.dx/

Q

D e mC

m

2

Q 2 =2

1 2

1

m

Q Q 2 ;

2

2

which gives m m

Q by letting # 0 and 2 Q 2 for " 1.

88

Chapter 2 Preferences

Q D 0. Note that the distribution

2 / is given by .x= /. Since ' 0 .x/ D x'.x/,

function of N.0;

Z c

Z c

x

c

d

x

x

'

dx D

2 dx D '

> 0:

d 1

1

Note that interchanging differentiation and integration

is justified by dominated conRc

vergence. Thus, we have shown that 7! 1 .x= / dx is strictly increasing for

all c, and N.0; 2 / <uni N.0; Q 2 / follows from part (d) of Theorem 2.57.

Now we turn to the case of arbitrary expectations m and m.

Q Let u be a utility

function. Then

Z

Z

Z

2

2

u dN.m; / D u.m C x/ N.0; /.dx/ u.m

Q C x/ N.0; 2 /.dx/;

because m m.

Q Since x 7! u.m

Q C x/ is again a utility function, we obtain from the

preceding step of the proof that

Z

Z

Z

2

2

Q C x/ N.0; Q /.dx/ D u dN.m;

Q Q 2 /;

u.m

Q C x/ N.0; /.dx/ u.m

Q Q 2 / follows.

and N.m; 2 / <uni N.m;

Remark 2.60. Let us indicate an alternative proof for the sufficiency part of Proposition 2.59 that uses condition (g) instead of (d) in Theorem 2.57. To this end, we define

a stochastic kernel by Q.x; / WD N.x C m

Q m; O 2 /, where O 2 WD Q 2 2 > 0.

Then m.Q.x; // D x C m

Q m x and

Q m; O 2 / D N.m C m

Q m; 2 C O 2 / D N.m; Q 2 /;

N.m; 2 / Q D N.m; 2 / N.m

Q Q 2 / follows.

where denotes convolution. Hence, N.m; 2 / <uni N.m;

<uni for lotteries with the same

expectation. A multidimensional version of this result will be given in Corollary 2.95

below.

Corollary 2.61. For all

, 2 M the following conditions are equivalent:

(a)

<uni and m.

/ D m./.

R

R

(b) f d

f d for all .not necessarily increasing/ concave functions f .

R

R

(c) m.

/ m./ and .x c/C

.dx/ .x c/C .dx/ for all c 2 R.

(d) There exists a probability space .; F ; P / with random variables X and X

having respective distributions

and such that

E X j X D X

P -a.s.

89

that m.Q.x; // D x for all x 2 S, such that D

Q.

Proof. (a) ) (e): Condition (g) of Theorem 2.57 yields a stochastic kernel Q such

that D

Q and m.Q.x; // x. Due to the assumption m.

/ D m./, Q must

satisfy m.Q.x; // D x at least for

-a.e. x. By modifying Q on the

-null set where

m.Q.x; // < x (e.g. by putting Q.x; / WD x there), we obtain a kernel as needed

for condition (e).

(e) ) (b): Since

Z

f .y/ Q.x; dy/ f .m.Q.x; /// D f .x/

by Jensens inequality, we obtain

Z

Z Z

Z

f d D

f .y/ Q.x; dy/

.dx/ f d

:

(b) ) (c): Just take the concave functions f .x/ D .x c/C , and f .x/ D x.

(c) ) (a): Note that

Z

Z

C

x

.dx/ c C c

..1; c/:

.x c/

.dx/ D

.c;1/

The existence of m.

/ implies that c

..1; c/ ! 0 as c # 1. Hence, we deduce

from the second condition in (c) that m.

/ m./, i.e., the two expectations are in

fact identical. Now we can apply the following put-call parity (compare also (1.10))

Z

Z

C

.c x/

.dx/ D c m.

/ C .x c/C

.dx/

to see that our condition (c) implies the third condition of Theorem 2.57 and, thus,

<uni .

(d) , (a): Condition (d) implies both m.

/ D m./ and condition (f) of Theorem 2.57, and this implies our condition (a). Conversely, assume that (a) holds. Then

Theorem 2.57 provides random variables X and X having the respective distributions

and such that E X j X X . Since X and X have the same mean,

this inequality must in fact be an almost-sure equality, and we obtain condition (d).

Let us denote by

Z

var.

/ WD

.x m.

//2

.dx/ D

2 M.

x 2

.dx/ m.

/2 2 0; 1

90

Chapter 2 Preferences

and be two lotteries in M such that m.

/ D m./ and

/ var./.

In the financial context, comparisons of portfolios with known payoff distributions

often use a mean-variance approach based on the relation

<

m.

/ m./ and var.

/ var./.

and , we have seen that the relation

< is equivalent

to

<uni . Beyond this special case, the equivalence typically fails as illustrated by

the following example and by Proposition 2.65 below.

Example 2.62. Let

be the uniform distribution on the interval 1; 1, so that

m.

/ D 0 and var.

/ D 1=3. For we take D p1=2 C .1 p/2 . With

the choice of p D 4=5 we obtain m./ D 0 and 1 D var./ > var.

/. However,

Z

Z

C

C

1

1

1

x .dx/ D 0;

D

x

.dx/ >

16

2

2

so

<uni does not hold.

Remark 2.63. Let

and be two lotteries in M. We will write

<con if

Z

Z

f d

f d for all concave functions f on R.

(2.23)

Note that

<con implies that m.

/ D m./, because both f .x/ D x and fQ.x/ D

x are concave. Corollary 2.61 shows that <con coincides with our uniform partial

order <uni if we compare two measures which have the same mean. The partial order

<con is sometimes called concave stochastic order. It was proposed in [226] and [227]

to express the view that

is less risky than . The inverse relation

<bal defined by

Z

Z

f d

f d for all convex functions f on R

(2.24)

is sometimes called balayage order or convex stochastic order.

Definition 2.64. A real-valued random variable Y on some probability space

.;F ;P / is called log-normally distributed with parameters 2 R and 0 if

it can be written as

Y D exp. C X /;

(2.25)

where X has a standard normal law N.0; 1/.

91

Clearly, any log-normally distributed random variable Y on .; F ; P / takes P a.s. strictly positive values. Recall from above the standard notations ' and for the

density and the distribution function of the standard normal law N.0; 1/. We obtain

from (2.25) the distribution function

log y

; 0 < y < 1;

PY y D

and the density

.y/ D

1

log y

'

I.0;1/ .y/

y

(2.26)

formula

1

E Y p D exp p C p 2 2 :

2

In particular, the law

of Y has the expectation

1

m.

/ D E Y D exp C 2

2

and the variance

var.

/ D exp.2 C 2 /.exp. 2 / 1/:

Proposition 2.65. Let

and

Q be two log-normal distributions with parameters

.; / and .;

Q /,

Q respectively. Then

<uni

Q holds if and only if 2 Q 2 and

1 2

C 2 Q C 12 Q 2 .

Proof. First suppose that 2 Q 2 and m.

/ m.

/.

Q We define a kernel Q.x; /

as the law of x exp. C Z/ where Z is a standard normal random variable. Now

suppose that

is represented by (2.25) with X independent of Z, and let f denote a

bounded measurable function. It follows that

Z

2

2 1=2

f d.

Q/ D Ef .e CX e

CZ / D Ef .e C

C. C / U /;

where

X C Z

U Dp

2 C 2

is also N.0;

p 1/-distributed. Thus,

Qpis a log-normal distribution with parameters

. C
; 2 C 2 /. By taking WD Q 2 2 and
WD Q , we can represent

Q as

Q D

Q. With this parameter choice,

2

1

D Q D log m.

/

Q log m.

/ . Q 2 2 / :

2

2

92

Chapter 2 Preferences

<uni

Q follows from condition (g) of

Theorem 2.57.

As to the converse implication, the inequality m.

/ m.

/

Q is already clear. To

prove 2 Q 2 , let WD

log1 and Q WD

Q log1 so that D N.; 2 /

and Q D N.;

Q Q 2 /. For " > 0 we define the concave increasing function f" .x/ WD

log." C x/. If u is a concave increasing function on R, the function u f" is a concave

and increasing function on 0; 1/, which can be extended to a concave increasing

function v" on the full real line. Therefore,

Z

Z

u d D lim

"#0

Z

v" d

lim

"#0

v" d

Q D

u d :

Q

(2.27)

Remark 2.66. The inequality (2.27) shows that if D N.; 2 /, Q D N.;

Q Q 2 / and

and

Q denote the images of and Q under the map x 7! e x , then

<uni

Q implies

<uni .

Q However, the converse implication <uni Q )

<uni

Q fails, as can be

seen by increasing Q until m.

/

Q > m.

/.

}

Because of its relation to the analysis of the BlackScholes formula for option

prices, we will now sketch a second proof of Proposition 2.65.

Second proof of Proposition 2:65. Let

Ym;

2

WD m exp X

2

E .Ym; c/C D m .dC / c .d /

with d D

log xc 12 2

I

see Example 5.56 in Chapter 5. Calculating the derivative of this expectation with

respect to > 0, one finds that

d

d

E .Ym; c/C D

.m .dC / c .d // D x '.dC / > 0I

d

d

see (5.43) in Chapter 5. The law

m; of Ym; satisfies m.

m; / D m for all > 0.

Condition (c) of Corollary 2.61 implies that

m; is decreasing in > 0 with respect

to <uni and hence also with respect to <con , i.e.,

m; <con

m;Q if and only if .

Q

For two different expectations m and m,

Q simply use the monotonicity of the function

93

Z

u d

m; D Eu. m exp. X 2 =2//

Eu. m

Q exp. X 2 =2//

Z

u d

m;

Q Q ;

provided that m m

Q and 0 < .

Q

The partial order <uni was defined in terms of integrals against increasing concave

functions. By taking the larger class of all concave functions as integrands, we arrived

at the partial order <con defined by (2.23) and characterized in Corollary 2.61. In

the remainder of this section, we will briefly discuss the partial order of stochastic

dominance, which is induced by increasing instead of concave functions:

Definition 2.67. Let

and be two arbitrary probability measures on R. We say that

<mon if

Z

Z

f d

f d for all bounded increasing functions f 2 C.R/.

Stochastic dominance is sometimes also called first order stochastic dominance. It

is indeed a partial order on M1 .R/: Reflexivity and transitivity are obvious, and antisymmetry follows, e.g., from the equivalence (a) , (b) below. As will be shown by

the following theorem, the relation

<mon means that the distribution

is higher

than the distribution . In our one-dimensional situation, we can provide a complete

proof of this fact by using elementary properties of distribution functions. The general

version of this result, given in Theorem 2.96, will require different techniques.

Theorem 2.68. For

; 2 M1 .R/ the following conditions are equivalent:

(a)

<mon .

(b) The distribution functions of

and satisfy F .x/ F .x/ for all x.

(c) Any pair of quantile functions for

and satisfies q .t / q .t / for a.e. t 2

.0; 1/.

(d) There exists a probability space .; F ; P / with random variables X and X

with distributions

and such that X X P -a.s.

(e) There exists a stochastic kernel Q.x; dy/ on R such that Q.x; .1; x/ D 1

and such that D

Q.

In particular,

<mon implies

<uni .

94

Chapter 2 Preferences

..1; x/ can be written as

Z

F .x/ D 1 I.x;1/ .y/

.dy/:

It is easy to construct a sequence of increasing continuous functions with values in

0; 1 which increase to I.x;1/ for each x. Hence,

Z

Z

I.x;1/ .y/

.dy/ I.x;1/ .y/ .dy/ D 1 F .x/:

(b) , (c): This follows from the definition of a quantile function and from Lemma

A.17.

(c) ) (d): Let .; F ; P / be a probability space supporting a random variable U

with a uniform distribution on .0; 1/. Then X WD q .U / and X WD q .U / satisfy

X X P -almost surely. Moreover, it follows from Lemma A.19 that they have the

distributions

and .

(d) ) (e): This is proved as in Theorem 2.57 by using regular conditional distributions.

(e) ) (a): Condition (e) implies that x y for Q.x; /-a.e. y. Hence, if f is

bounded and increasing, then

Z

Z

f .y/ Q.x; dy/ f .x/ Q.x; dy/ D f .x/:

Therefore,

Z Z

f d D

Z

f .y/ Q.x; dy/

.dx/

f d

:

Finally, due to the equivalence (a) , (b) above and the equivalence (a) , (d) in

Theorem 2.57,

<mon implies

<uni .

Remark 2.69. It is clear from conditions (d) or (e) of Theorem 2.68 that the set of

bounded, increasing, and continuous functions in Definition 2.67 can be replaced by

the set of all increasing functions for which the two integrals make sense. Thus,

<mon for

; 2 M implies

<uni , and in particular m.

/ m./. Moreover,

}

condition (d) shows that

<mon together with m.

/ D m./ implies

D .

2.5

In this section, we discuss the structure of preferences for assets on a more fundamental level. Instead of assuming that the distributions of assets are known and that

preferences are defined on a set of probability measures, we will take as our basic

objects the assets themselves. An asset will be viewed as a function which associates real-valued payoffs to possible scenarios. More precisely, X will denote a set

95

that no a priori probability measure is given on .; F /. In other words, we are facing

uncertainty instead of risk.

We assume that X is endowed with a preference relation . In view of the financial

interpretation, it is natural to assume that is monotone in the sense that

Y X

Under a suitable condition of continuity, we could apply the results of Section 2.1 to

obtain a numerical representation of . L. J. Savage introduced a set of additional

axioms which guarantee that there is a numerical representation of the special form

Z

(2.28)

U.X / D EQ u.X / D u.X.!// Q.d!/ for all X 2 X

where Q is a probability measure on .; F / and u is a function on R. The measure

Q specifies the subjective view of the probabilities of events which is implicit in the

preference relation . Note that the function u W R ! R is determined by restricting

U to the class of constant functions on .; F /. Clearly, the monotonicity of is

equivalent to the condition that u is an increasing function.

Definition 2.70. A numerical representation of the form (2.28) will be called a Savage

representation of the preference relation .

Remark 2.71. Let

Q;X denote the distribution of X under the subjective measure Q. Clearly, the preference order on X given by (2.28) induces a preference

order on

MQ WD

Q;X j X 2 X

with von NeumannMorgenstern representation

Z

UQ .

Q;X / WD U.X / D EQ u.X / D

i.e.,

u d

Q;X ;

Z

UQ .

/ D

u.x/

.dx/ for

2 MQ .

On this level, Section 2.3 specifies the conditions on UQ which guarantee that u is a

(strictly concave and strictly increasing) utility function.

}

Remark 2.72. Even if an economic agent with preferences would accept the view

that scenarios ! 2 are generated in accordance to a given objective probability

measure P on .; F /, the preference order on X may be such that the subjective measure Q appearing in the Savage representation (2.28) is different from the

objective measure P . Suppose, for example, that P is Lebesgue measure restricted

96

Chapter 2 Preferences

to D 0; 1, and that X is the space of bounded right-continuous increasing functions on 0; 1. Let

P;X denote the distribution of X under P . By Lemma A.19,

every probability measure on R with bounded support is of the form

P;X for some

X 2 X, i.e.,

Mb .R/ D

P;X j X 2 X:

Suppose the agent agrees that, objectively, X 2 X can be identified with the lottery

on Mb .R/ with numerical representation

U .

P;X / WD U.X /:

This does not imply that U satisfies the assumptions of Section 2.2; in particular, the

preference relation on Mb .R/ may violate the independence axiom. In fact, the agent

might take a pessimistic view and distort P by putting more emphasis on unfavorable

scenarios. For example, the agent could replace P by the subjective measure

Q WD 0 C .1 /P

for some 2 .0; 1/ and specify preferences by a Savage representation in terms of u

and Q. In this case,

Z

U .

P;X / D EQ u.X / D u d

Q;X

D u.X.0// C .1 /EP u.X /

Z

D u.X.0// C .1 / u d

P;X :

Note that X.0/ D `.

P;X / for

`.

/ WD inf.supp

/ D supa 2 R j

..1; a// D 0 ;

where supp

is the support of

. Hence, replacing P by Q corresponds to a nonlinear distortion on the level of lotteries:

D

P;X is distorted to the lottery

D

Q;X given by

D `./ C .1 /

;

and the preference relation on lotteries has the numerical representation

Z

U .

/ D u.x/

.dx/ for

2 Mb .R/.

Let us now show that such a subjective distortion of objective lotteries provides a

possible explanation of the Allais paradox. Consider the lotteries

i and i , i D 1; 2,

described in Example 2.30. Clearly,

1 D

1

and

1 D 0 C .1 /1 ;

97

while

2 D 0 C .1 /

2

and

1 D 0 C .1 /1 :

2 /, and for > 9=2409

we obtain U .

1 / > U .1 /, in accordance with the observed preferences 2

2

and

1 1 described in Example 2.30.

For a systematic discussion of preferences described in terms of a subjective distortion of lotteries we refer to [173]. In Section 4.6, we will discuss the role of distortions

in the context of risk measures, and in particular the connection to Yaaris dual theory

of choice under risk [265].

}

Even in its general form (2.28), however, the paradigm of expected utility has a

limited scope as illustrated by the following example.

Example 2.73 (Ellsberg paradox). You are faced with a choice between two urns,

each containing 100 balls which are either red or black. In the first urn, the proportion

p of red balls is know; assume, e.g., p D 0:49. In the second urn, the proportion pQ

is unknown. Suppose that you get 1000 C if you draw a red ball and 0 C otherwise.

In this case, most people would choose the first urn. Naturally, they make the same

choice if you get 1000 C for drawing a black ball and 0 C for a red one. But this

behavior is not compatible with the paradigm of expected utility: For any subjective

probability pQ of drawing a red ball in the second urn, the first choice would imply

p > p,

Q the second would yield 1 p > 1 p,

Q and this is a contradiction.

}

For this reason, we are going to make one further conceptual step beyond the Savage representation before we start to prove a representation theorem for preferences

on X. Instead of a single measure Q, let us consider a whole class Q of measures

on .; F /. Our aim is to characterize those preference relations on X which admit a

representation of the form

U.X / D inf EQ u.X / :

Q2Q

(2.29)

This may be viewed as a robust version of the paradigm of expected utility: The

agent has in mind a whole collection of possible probabilistic views of the given set

of scenarios and takes a worst-case approach in evaluating the expected utility of a

given payoff.

It will be convenient to extend the discussion to the following framework where

payoffs can be lotteries. Let X denote the space of all bounded measurable functions

on .; F /. We are going to embed X into a certain space XQ of functions XQ on

.; F / with values in the convex set

Mb .R/ D

2 M1 .R/ j

.c; c/ D 1 for some c 0

98

Chapter 2 Preferences

fined as the convex set of all those stochastic kernels X.!;

dy/ from .; F / to R for

which there exists a constant c 0 such that

Q

X.!;

c; c/ D 1

for all ! 2 .

In economics, the elements of XQ are sometimes called acts or horse race lotteries; see,

for example, [187]. The space X can be embedded into XQ by virtue of the mapping

Q

X 3 X 7! X 2 X:

(2.30)

In this way, X can be identified with the set of all XQ 2 XQ for which the measure

XQ .!; / is a Dirac measure. A preference order on X defined by (2.29) clearly extends

to XQ by

Z Z

Q D inf

UQ .X/

u.y/ XQ .!; dy/ Q.d!/ D inf EQ u.

Q XQ /

(2.31)

Q2Q

Q2Q

Z

u.

/

Q

D u d

;

2 Mb .R/:

Remark 2.74. Restricting the preference order on XQ obtained from (2.31) to the

Q

constant maps X.!/

D

for

2 Mb .R/, we obtain a preference order on Mb .R/,

and on this level we know how to characterize risk aversion by the property that u is

strictly concave.

}

Example 2.75. Let us show how the Ellsberg paradox fits into our extended setting,

and how it can be resolved by a suitable choice of the set Q. For D 0; 1 define

XQ 0 .!/ WD p 1000 C .1 p/0 ;

and

ZQ i .!/ WD 1000 Ii .!/ C 0 I1i .!/;

i D 0; 1:

Take

Q WD q 1 C .1 q/0 j a q b

with a; b 0; 1. For any increasing function u, the functional

Q WD inf EQ u.

Q XQ /

UQ .X/

Q2Q

satisfies

UQ .XQi / > UQ .ZQ i /;

i D 0; 1;

as soon as a < p < b, in accordance with the preferences described in Example 2.73.

}

99

Let us now formulate those properties of a preference order on the convex set

Q

X which are crucial for a representation of the form (2.31). For XQ ; YQ 2 XQ and

2 .0; 1/, (2.31) implies

Q XQ / C .1 / EQ u.

Q YQ / /

UQ . XQ C .1 /YQ / D inf . EQ u.

Q2Q

Q C .1 /UQ .YQ /:

UQ .X/

In contrast to the Savage case Q D Q, we can no longer expect equality, except

for the case of certainty YQ .!/

. If XQ YQ , then UQ .XQ / D UQ .YQ /, and the lower

bound reduces to UQ .XQ / D UQ .YQ /. Thus, satisfies the following two properties:

Uncertainty aversion: If XQ ; YQ 2 XQ are such that XQ YQ , then

XQ C .1 /YQ
XQ

Q ZQ

2 Mb .R/, and 2 .0; 1 we have

Certainty independence: For XQ ; YQ 2 X,

XQ YQ

Q

XQ C .1 /ZQ YQ C .1 /Z:

Remark 2.76. In order to motivate the term uncertainty aversion, consider the situation of the preceding example. Suppose that an agent is indifferent between the

choices ZQ 0 and ZQ 1 , which both involve the same kind of Knightian uncertainty. For

2 .0; 1/, the convex combination YQ WD ZQ 0 C.1/ZQ 1 , which is weakly preferred

to both ZQ 0 and ZQ 1 in the case of uncertainty aversion, takes the form

1000 C .1 /0 for ! D 1,

YQ .!/ D

0 C .1 /1000 for ! D 0,

i.e., uncertainty is reduced in favor of risk. For D 1=2, the resulting lottery

YQ .!/ 12 .1000 C 0 / is independent of the scenario !, i.e., Knightian uncertainty

is completely replaced by the risk of a coin toss.

}

Remark 2.77. The axiom of certainty independence extends the independence axiom for preferences on lotteries to our present setting, but only under the restriction

that one of the two contingent lotteries XQ and YQ is certain, i.e., does not depend on

the scenario ! 2 . Without this restriction, the extended independence axiom would

lead to the Savage representation in its original form (2.28); see Exercise 2.5.3 below.

Q As an exThere are good reasons for not requiring full independence for all ZQ 2 X.

Q

Q

Q An agent

Q

ample, take D 0; 1 and define X .!/ D ! , Y .!/ D 1! , and Z D X.

Q

Q

may prefer X over Y , thus expressing the implicit view that scenario 1 is somewhat

more likely than scenario 0. At the same time, the agent may like the idea of hedging

against the occurrence of scenario 0, and this could mean that the certain lottery

1

1 Q

. Y C ZQ /./ .0 C 1 /

2

2

100

Chapter 2 Preferences

1 Q

Q

. X C ZQ /./ X./;

2

thus violating the independence assumption in its unrestricted form. In general, the

role of ZQ as a hedge against scenarios unfavorable for YQ requires that YQ and ZQ are

not comonotone, i.e.,

9 !; 2 W

Q

Q

YQ .!/ YQ ./; Z.!/

Z./:

(2.32)

Thus, the wish to hedge would still be compatible with the following enforcement of

certainty independence, called

Comonotonic independence: For XQ ; YQ ; ZQ 2 XQ and 2 .0; 1

XQ YQ

Q

XQ C .1 /ZQ YQ C .1 /Z:

(2.33)

whenever YQ and ZQ are comonotone in the sense that (2.32) does not occur.

The consequences of requiring comonotonic independence will be analyzed in Exercise 2.5.4 below. It is also relevant to weaken the axiom of certainty independence

and to require instead

Weak certainty independence: if for XQ ; YQ 2 XQ and for some 2 M1;c .S/ and

2 .0; 1 we have XQ C .1 / YQ C .1 /, then

XQ C .1 /

YQ C .1 /

for all

2 Mb .S/.

The consequences of requiring weak certainty independence will be analyzed in Theorem 2.88 below.

}

Q The set Mb .R/

From now on, we assume that is a given preference order on X.

Q

will be regarded as a subset of X by identifying a constant function ZQ

with its

value

2 Mb .R/. We assume that possesses the following properties:

Uncertainty aversion.

Certainty independence.

Q

Q Moreover, is

Monotonicity: If YQ .!/
X.!/

for all ! 2 , then YQ
X.

compatible with the usual order on R, i.e., y x if and only if y > x.

Q If

Continuity: The following analogue of the Archimedean axiom holds on X:

Q

Q

Q

Q

Q

Q

Q

X ; Y ; Z 2 X are such that Z Y X, then there are ; 2 .0; 1/ with

ZQ C .1 /XQ YQ ZQ C .1 /XQ :

Moreover, for all c > 0 the restriction of to M1 .c; c/ is continuous with

respect to the weak topology.

101

Let us denote by

M1;f WD M1;f .; F /

the class of all set functions Q W F ! 0; 1 which are normalized to Q D 1

and which are finitely additive, i.e., Q A [ B D Q A C Q B for all disjoint

A; B 2 F ; see Appendix A.6. By EQ X we denote the integral of X with respect

to Q 2 M1;f ; see Appendix A.6. With M1 DM1 .; F / we denote the -additive

members of M1;f , that is, the class of all probability measures on .; F /. Note that

the inclusion M1 M1;f is typically strict as is illustrated by Example A.53.

Theorem 2.78. Consider a preference order on XQ satisfying the four properties of

uncertainty aversion, certainty independence, monotonicity, and continuity.

(a) There exists a strictly increasing function u 2 C.R/ and a convex set Q

M1;f .; F / such that

Z

Q

Q

Q

u.x/ X .; dx/

U .X / D min EQ

Q2Q

(b) If the induced preference order on X, viewed as a subset of XQ as in (2.30),

satisfies the following additional continuity property

X Y and Xn % X

H)

Xn Y

(2.34)

then the set functions in Q are in fact probability measures, i.e., each Q 2 Q is

-additive. In this case, the induced preference order on X has the robust Savage

representation

U.X / D min EQ u.X / for X 2 X

Q2Q

with Q M1 .; F /.

Remark 2.79. Even without its axiomatic foundation, the robust Savage representation is highly plausible as it stands, since it may be viewed as a worst-case approach

to the problem of model uncertainty. This aspect will be of particular relevance in our

discussion of risk measures in Chapter 4.

}

The proof of Theorem 2.78 needs some preparation.

Q the axiom of certainty indeWhen restricted to Mb .R/, viewed as a subset of X,

pendence is just the independence axiom of the von NeumannMorgenstern theory.

Thus, the preference relation on Mb .R/ satisfies the assumptions of Corollary 2.28,

and we obtain the existence of a continuous function u W R ! R such that

Z

u.

/

Q

WD u.x/

.dx/

(2.35)

102

Chapter 2 Preferences

to positive affine transformations. The second part of our monotonicity assumption

implies that u is strictly increasing. Without loss of generality, we assume u.0/ D 0

and u.1/ D 1.

Remark 2.80. In view of the representation (2.35), it follows as in (2.11) that any

/ 2 R for which

c./ :

Thus, if X 2 X is defined for XQ 2 XQ as X.!/ WD c.XQ .!//, then the first part of our

monotonicity assumption yields

XQ X ;

(2.36)

and so the preference relation on XQ is uniquely determined by its restriction to X.

}

Lemma 2.81. There exists a unique extension UQ of the functional uQ in (2.35) as a

Q

numerical representation of on X.

Proof. For XQ 2 XQ let c > 0 be such that XQ .!; c; c/ D 1 for all ! 2 . Then

Q XQ .!// u.

Q c/

u.

Q c / u.

for all ! 2 ,

c XQ c :

We will show below that there exists a unique 2 0; 1 such that

XQ .1 /c C c :

(2.37)

Once this has been achieved, the only possible choice for UQ .XQ / is

UQ .XQ / WD u..1

Q

/c C c / D .1 /u.

Q c / C u.

Q c /:

Q

This definition of UQ provides a numerical representation of on X.

The proof of the existence of a unique 2 0; 1 with (2.37) is similar to the proof

of Lemma 2.24. Uniqueness follows from the monotonicity

>

H)

.1 /c C c .1 /c C c ;

(2.38)

(2.35). Now we let

WD sup 2 0; 1 j XQ .1 /c C c :

103

XQ .1 /c C c

.1 /c C c XQ :

(2.39)

(2.40)

In the case (2.39), our continuity axiom yields some 2 .0; 1/ for which

XQ .1 /c C c C .1 /c D .1 /c C c

where D C .1 / > , in contradiction to the definition of .

If (2.40) holds, then the same argument as above yields 2 .0; 1/ with

c C .1 /c XQ :

By our definition of there must be some 2 .; / with

XQ
.1 /c C c c C .1 /c ;

where the second relation follows from (2.38). This, however, is a contradiction.

Remark 2.82. Note that the proof of the preceding lemma relies only on the assumptions of continuity and monotonicity of . Certainty independence and uncertainty

aversion are not needed.

}

Via the embedding (2.30), Lemma 2.81 induces a numerical representation U of

on X given by

(2.41)

U.X / WD UQ .X /:

The following proposition clarifies the properties of the functional U and provides the

key to a robust Savage representation of the preference order on X.

Proposition 2.83. Given u of (2.35) and the numerical representation U on X constructed via Lemma 2:81 and (2.41), there exists a unique functional W X ! R

such that

U.X / D .u.X // for all X 2 X,

(2.42)

and such that the following four properties are satisfied:

104

Chapter 2 Preferences

Proof. Denote by Xu the space of all X 2 X which take values in a compact subset of the range u.R/ of u. Clearly, Xu coincides with the range of the non-linear

transformation X 3 X 7! u.X /. Note that this transformation is bijective since u is

continuous and strictly increasing due to our assumption of monotonicity. Thus, is

well-defined on Xu via (2.42). We show next that this has the four properties of the

assertion.

Monotonicity is obvious. For positive homogeneity on Xu , it suffices to show that

.
X / D
.X / for X 2 Xu and
2 .0; 1. Let X0 2 X be such that u.X0 / D X.

We define ZQ 2 XQ by

ZQ WD
X0 C .1
/0 :

By (2.36), ZQ Z where Z is given by

Z.!/ D c.
X0 .!/ C .1
/0 /

D u1 .
u.X0 .!// C .1
/u.0//

D u1 .
u.X0 .!///;

where we have used our convention u.0/ D 0. It follows that u.Z/ D
u.X0 / D
X,

and so

Q

.
X / D U.Z/ D UQ .Z/:

(2.43)

As in (2.37), one can find 2 Mb .R/ such that X0 . Certainty independence

implies that

ZQ D
X0 C .1
/0
C .1
/0 :

Hence,

Q D u.

UQ .Z/

Q

C .1
/0 / D
u./

Q

D
U.X0 / D
.X /:

This shows that is positively homogeneous on Xu .

Since the range of u is an interval, we can extend from Xu to all of X by positive

homogeneity, and this extension, again denoted , is also monotone and positively

homogeneous.

Let us now show that is cash invariant. First note that

.1/ D

u.

Q x/

.u.x//

D

D1

u.x/

u.x/

for any x such that u.x/ 0. Now take X 2 X and z 2 R. By positive homogeneity,

we may assume without loss of generality that 2X 2 Xu and 2z 2 u.R/. Then there

are X0 2 X such that 2X D u.X0 / as well as z0 ; x0 2 R with 2z D u.z0 / and

2.X / D u.x0 /. Note that X0 x0 . Thus, certainty independence yields

1

1

ZQ WD .X0 C z0 / .x0 C z0 / DW

:

2

2

105

Q D U.

/ D 1 u.x0 / C 1 u.z0 / D .X / C z:

UQ .Z/

2

2

On the other hand, the same reasoning which lead to (2.43) shows that

Q D .X C z/:

UQ .Z/

As to concavity, we need only show that . 12 X C 12 Y / 12 .X / C 12 .Y / for

X; Y 2 Xu , by Exercise 2.5.2 below. Let X0 ; Y0 2 X be such that X D u.X0 / and

Y D u.Y0 /. If .X / D .Y /, then X0 Y0 , and uncertainty aversion gives

1

ZQ WD .X0 C Y0 / X0 ;

2

which by the same arguments as above yields

1

1

1

Q

Q

U .Z/ D X C Y .X / D ..X / C .Y //:

2

2

2

The case in which .X / > .Y / can be reduced to the previous one by letting z WD

.X / .Y /, and by replacing Y by Yz WD Y C z. Cash invariance then implies that

1

1

1

1

1

X C Y C z D X C Yz

2

2

2

2

2

1

..X / C .Yz //

2

1

1

D ..X / C .Y // C z:

2

2

A functional W X ! R satisfying the properties of monotonicity, concavity,

positive homogeneity, and cash invariance is sometimes called a coherent monetary

utility functional. The functional .X / WD .X / is called a coherent risk measure.

Functionals of this type will be studied in detail in Chapter 4.

Exercise 2.5.1. Let W X ! R be a functional that satisfies the properties of monotonicity and cash invariance stated in Proposition 2.83. Show that is Lipschitz continuous on X with respect to the supremum norm k k, i.e.,

j.X / .Y /j kX Y k for all X; Y 2 X.

. 12 .X C Y // 12 .X / C 12 .Y / for all X; Y 2 X. Use Exercise 2.5.1 to show that

is concave.

}

106

Chapter 2 Preferences

Let us now show that a function with the four properties established in Proposition 2.83 can be represented in terms of a family of set functions in the class M1;f .

Proposition 2.84. A functional W X ! R is monotone, concave, positively homogeneous, and cash invariant if and only if there exists a set Q M1;f such that

.X / D inf EQ X ;

Q2Q

X 2 X:

Moreover, the set Q can always be chosen to be convex and such that the infimum

above is attained, i.e.,

.X / D min EQ X ;

Q2Q

X 2 X:

Proof. The necessity of the four properties is obvious. Conversely, we will construct

for any X 2 X a finitely additive set function QX such that .X / D EQX X and

.Y / EQX Y for all Y 2 X. Then

.Y / D min EQ Y

Q2Q0

for all Y 2 X

(2.44)

convex hull Q WD conv Q0 .

To construct QX for a given X 2 X, we define three convex sets in X by

B WD Y 2 X j .Y / > 1;

C1 WD Y 2 X j Y 1;

and

C2 WD Y 2 X Y

X

:

.X /

The convexity of C1 and C2 implies that the convex hull of their union is given by

C WD conv.C1 [ C2 / D Y1 C .1 /Y2 j Yi 2 Ci and 2 0; 1 :

Since Y 2 C is of the form Y D Y1 C .1 /Y2 for some Yi 2 Ci and 2 0; 1,

.Y / . C .1 /Y2 / D C .1 /.Y2 / 1;

and so B and C are disjoint. Let X be endowed with the supremum norm kY k WD

sup!2 jY .!/j. Then C1 , and hence C , contains the unit ball in X. In particular, C

has non-empty interior. Thus, we may apply the separation argument in the form of

Theorem A.55, which yields a non-zero continuous linear functional ` on X such that

c WD sup `.Y / inf `.Z/:

Y 2C

Z2B

107

Since C contains the unit ball, c must be strictly positive, and there is no loss of

generality in assuming c D 1. In particular, `.1/ 1 as 1 2 C. On the other hand,

any constant b > 1 is contained in B, and so

`.1/ D lim `.b/ c D 1:

b#1

Hence, `.1/ D 1.

If A 2 F then IAc 2 C1 C , which implies that

`.IA / D `.1/ `.IAc / 1 1 D 0:

By Theorem A.51 there exists a finitely additive set function QX 2 M1;f .; F /

such that `.Y / D EQX Y for any Y 2 X.

It remains to show that EQX Y .Y / for all Y 2 X, with equality for Y D X.

By the cash invariance of , we need only consider the case in which .Y / > 0. Then

Yn WD

Y

1

C 2 B;

.Y /

n

EQX Y

D lim EQX Yn 1:

.Y /

n"1

On the other hand, X=.X / 2 C2 C yields the inequality

EQX X

c D 1:

.X /

We are now ready to complete the proof of the first main result in this section.

Proof of Theorem 2:78. (a): By Remark 2.80, it suffices to consider the induced preference relation on X once the function u has been determined. According to

Lemma 2.81 and the two Propositions 2.83 and 2.84, there exists a convex set Q

M1;f such that

U.X / D min EQ u.X /

Q2Q

(b): The assumption (2.34) applied to X 1 and Y b < 1 gives that any

sequence with Xn % 1 is such that Xn b for large enough n. We claim that

this implies that U.Xn / % u.1/ D 1. Otherwise, U.Xn / would increase to some

number a < 1. Since u is continuous and strictly increasing, we may take b such that

a < u.b/ < 1. But then U.Xn / > U.b/ D u.b/ > a for large enough n, which is a

contradiction.

108

Chapter 2 Preferences

n An D

lim min Q An D lim U.IAn / D 1:

n"1 Q2Q

n"1

But this means that each Q 2 Q satisfies limn Q An D 1, which is equivalent to the

-additivity of Q.

The continuity assumption (2.34), required for all Xn 2 X, is actually quite strong.

In a topological setting, our discussion of risk measures in Chapter 4 will imply the

following version of the representation theorem.

Proposition 2.85. Consider a preference order as in Theorem 2:78. Suppose that

is a Polish space with Borel field F and that (2.34) holds if Xn and X are continuous. Then there exists a class of probability measures Q M1 .; F / such that the

induced preference order on X has the robust Savage representation

U.X / D min EQ u.X / for continuous X 2 X.

Q2Q

Proof. As in the proof of Theorem 2.78, the continuity property of implies the

corresponding continuity property of U , and hence of the functional in (2.42). The

result follows by combining Proposition 2.83, which reduces the representation of U

to a representation of , with Proposition 4.27 applied to the coherent risk measure

WD .

Now we consider an alternative setting where we fix in advance a reference measure P on .; F /. In this context, X will be identified with the space L1 .; F ; P /,

and the representation of preferences will involve measures which are absolutely continuous with respect to P . Note, however, that this passage from measurable functions to equivalence classes of random variables in L1 .; F ; P /, and from arbitrary

probability measures to absolutely continuous measures, involves a certain loss of

robustness in the face of model uncertainty.

Theorem 2.86. Let be a preference relation as in Theorem 2:78, and assume that

X Y

whenever X D Y P -a.s.

U.X / D inf EQ u.X / ;

Q2Q

X 2 X;

with respect to P , if and only if satisfies the following condition of continuity from

above:

Y X and Xn & X P -a.s.

H)

Y Xn

109

U.X / D min EQ u.X / ;

Q2Q

X 2 X;

with respect to P , if and only if satisfies the following condition of continuity from

below:

X Y and Xn % X P -a.s.

H)

Xn Y

Proof. As in the proof of Theorem 2.78, the continuity property of implies the

corresponding continuity property of U , and hence of the functional in (2.42). The

results follow by combining Proposition 2.83, which reduces the representation of

U to a representation of , with Corollary 4.37 and Corollary 4.38 applied to the

coherent risk measure WD .

In the following two exercises we explore the impact of replacing the axiom of

certainty independence by stronger requirements as discussed in Remark 2.77.

Exercise 2.5.3. Show that the following conditions are equivalent:

(a) The preference relation satisfies the following unrestricted independence axiom on XQ ,

Independence: For XQ ; YQ ; ZQ 2 XQ and 2 .0; 1 we have

XQ YQ

Q

XQ C .1 /ZQ YQ C .1 /Z:

(c) The set Q has exactly one element Q 2 M1;f , and so U.X / D .u.X // admits

the Savage representation

U.X / D EQ u.X / ;

X 2 X:

X.!/ X.! 0 /Y .!/ Y .! 0 / 0 for all pairs .!; ! 0 / 2 .

Show that the following conditions are equivalent:

(a) The preference relation satisfies the axiom of comonotonic independence

(2.33).

(b) The functional is comonotonic in the sense that .X C Y / D .X / C .Y /

whenever X; Y 2 X are comonotone.

}

110

Chapter 2 Preferences

Now we discuss what happens if we replace the assumption of certainty independence by weak certainty independence as introduced in Remark 2.77. We thus assume

from now on that is a preference relation on XQ satisfying the following conditions.

Uncertainty aversion.

Monotonicity.

Continuity.

Exercise 2.5.5. Show that the restriction of to Mb .R/ satisfies the independence

axiom of von NeumannMorgenstern theory and hence admits a von NeumannMorgenstern representation

Z

(2.45)

u.

/

Q

WD u.x/

.dx/;

2 Mb .R/;

with a continuous and strictly increasing function u W R ! R.

For simplicity we will assume for the rest of this section that the function u in (2.45)

has an unbounded range u.R/ containing zero. The assumption of an unbounded

range is satisfied automatically if the restriction of to Mb .R/ is risk averse and u

hence concave. Lemma 2.81 and Remark 2.82 imply that the numerical representation

uQ in (2.45) admits a unique extension UQ W XQ ! R that is a numerical representation

Q We define again

of on all of X.

U.X / WD UQ .X / for X 2 X.

As in Remark 2.80, we see that XQ X when X 2 X is defined as X.!/ WD c.XQ .!//

with

Z

1

c.

/ D u

u d

2 Mb .R/. In particular, the preference

relation on XQ is uniquely determined by its restriction to X. We now analyze the

structure of U in analogy to Proposition 2.83. Our next result shows that U is again of

the form U.X / D .u.X // for a functional W X ! R. However, may no longer

be positively homogeneous, since we have replaced certainty independence with weak

certainty independence.

Proposition 2.87. Under the above assumptions, there exists a unique functional

W X ! R such that

UQ .X / D .u.

Q XQ //

Q

for all XQ 2 X,

(2.46)

111

Proof. Let Xu denote the set of all X 2 X that take values in a compact subset of the

range u.R/ of u. Since u is strictly increasing, we can define on Xu via

.X / WD UQ .u1 .X/ /;

X 2 Xu :

Then we have

Q

Q X//;

UQ .XQ / D UQ .c.XQ / / D .u.c.XQ /// D .u.

(2.47)

assumption.

We now prove that is cash invariant on Xu . To this end, we assume first that

u.R/ D R and take X 2 X and some z 2 R. We then let X0 WD u1 .2X /,

z0 WD u1 .2z/, and y WD u1 .0/. Taking a > 0 such that a X0 .!/ a for each

!, we see as in (2.37) that there exists 2 0; 1 such that

1

1

1

1

1

X0 C y .a C y / C

.a C y / D

C y ;

2

2

2

2

2

2

where

D a C .1 /a . Using weak certainty independence, we may replace

y by z0 and obtain 12 .X0 C z0 / 12 .

C z0 /. Hence, from (2.47),

1

1

1

1

u.X0 / C u.z0 / D uQ X0 C z0

2

2

2

2

1

1

1

1

D UQ X0 C z0 D UQ

C z0

2

2

2

2

1

1

1

1

Q

C u.z0 /:

D uQ

C z0 D u.

/

2

2

2

2

.X C z/ D

Cash invariance now follows from u.z0 / D 2z and the fact that

1

1

1

1

1

1

u.

/

Q

D .u.

/

Q

C u.y // D uQ

C y D UQ

C y

2

2

2

2

2

2

1

1

1

1

D UQ X0 C y D u.X0 / C u.y/ D .X /:

2

2

2

2

Here we have again applied (2.47). If u.R/ is not equal to R it is sufficient to consider

the cases in which u.R/ contains 0; 1/ or .1; 0 and to work with positive or negative quantities X and z, respectively. Then the preceding argument establishes the

cash invariance of on the spaces of positive or negative bounded measurable functions, and can be extended by translation to the entire space of bounded measurable

functions.

112

Chapter 2 Preferences

This is enough due to Exercise 2.5.2. Let X0 WD u1 .X / and Y0 WD u1 .Y / and

suppose first that .X / D .Y /. Then X0 Y0 and uncertainty aversion implies

that ZQ WD 12 X0 C 12 Y0 X0 . Hence, by using (2.47),

1

Q UQ .X0 / D .X / D 1 .X / C 1 .Y /:

.X C Y / D UQ .Z/

2

2

2

.Yz / D .X /. Hence,

1

1

1

1

1

1

1

.X CY / C z D .X CYz / .X /C .Yz / D .X /C .Y /C z:

2

2

2

2

2

2

2

2

1

cash invariance is sometimes called a concave monetary utility functional, and WD

is called a convex risk measure. In Chapter 4 we will derive various representation

results for convex risk measures. In particular it will follow from Theorem 4.16 that

every concave monetary utility functional W X ! R is of the form

.X / D

Q2M1;f

X 2 X;

(2.48)

Exercise 2.5.6. Prove the representation (2.48) for the case in which X is the set of

all functions X W ! R on a finite set . To this end, one can either use biduality

(Theorem A.62) or, more directly, a separation argument as given in Proposition A.1.

}

Combining Proposition 2.87 with the representation (2.48) yields the final result of

this section:

Theorem 2.88. Consider a preference order on XQ satisfying the four properties

of uncertainty aversion, weak certainty independence, monotonicity, and continuity.

Assume moreover that the function u in (2.45) has an unbounded range u.R/. Then

there exists a penalty function W M1;f ! R [ C1 that is bounded from below

such that

Z

Q

EQ

u.x/ XQ .; dx/ C .Q/ ; XQ 2 X;

UQ .XQ / D min

Q2M1;f

is a numerical representation of .

113

2.6

In this section, we study the construction of probability measures with given marginals. In particular, this will yield the missing implication in the characterization of

uniform preference in Theorem 2.57, but the results in this section are of independent

interest. We focus on the following basic question: Suppose

1 and

2 are two

probability measures on S, and is a convex set of probability measures on S S;

when does contain some

which has

1 and

2 as marginals?

The answer to this question will be given in a general topological setting. Let S be

a Polish space, and let us fix a continuous function on S with values in 1; 1/. As

in Section 2.2 and in Appendix A.6, we use as a gauge function in order to define

the space of measures

Z

.x/

.dx/ < 1

M1 .S/ WD

2 M1 .S/

and the space of continuous test functions

C .S/ WD f 2 C.S/ j 9 c W jf .x/j c

The

Z

M1 .S/ 3

7! f d

is a continuous mapping for all f 2 C .S/; see Appendix A.6 for details. On the

product space S S, we take the gauge function

.x; y/ WD

.x/ C

.y/;

and define the corresponding set M1 .S S /, which will be endowed with the -weak

topology.

Theorem 2.89. Suppose that M1 .S S/ is convex and closed in the -weak

topology, and that

1 ,

2 are probability measures in M1 .S/. Then there exists some

1 and

2 if and only if

Z

Z

Z

f1 d

1 C f2 d

2 sup .f1 .x/ C f2 .y// .dx; dy/ for all f1 ; f2 2 C .S/.

Theorem 2.89 is due to V. Strassen [255]. Its proof boils down to an application of

the HahnBanach theorem; the difficult part consists in specifying the right topological setting. First, let us investigate the relations between M1 .S S/ and M1 .S/. To

this end, we define mappings

i W M1 .S S / ! M1 .S/;

i D 1; 2,

114

Chapter 2 Preferences

Z

Z

Z

Z

f d.2 / D f .y/ .dx; dy/;

f d.1 / D f .x/ .dx; dy/ and

for all f 2 C .S/.

Lemma 2.90. 1 and 2 are continuous and affine mappings from M1 .S S/ to

M1 .S/.

Proof. Suppose that n converges to in M1 .S S/. For f 2 C .S/ let f .x; y/ WD

f .x/. Clearly, f 2 C .S S /, and thus

Z

Z

Z

Z

f d.1 n / D f d n ! f d D f d.1 /:

Therefore, 1 is continuous, and the same is true of 2 . Affinity is obvious.

Now, let us consider the linear space

E WD

j

; 2 M1 .S/; ; 2 R

R

spanned by M1 .S/. For D

2 E the integral f d against a function

f 2 C .S/ is well-defined and given by

Z

Z

Z

f d D f d

f d:

R

In particular, 7! f d is linear functional Ron E, so we

R can regard C .S/ as a

subset of the algebraic dual E of E. Note that f d D f d Q for all f 2 C .S/

implies D ,

Q i.e., C .S/ separates the points of E. We endow E with the coarsest

topology .E; C .S// for which all maps

Z

E 3 7! f d; f 2 C .S/;

are continuous; see Definition A.58. With this topology, E becomes a locally convex

topological vector space.

Lemma 2.91. Under the above assumptions, M1 .S/ is a closed convex subset of E,

and the relative topology of the embedding coincides with the -weak topology.

Proof. The sets of the form

U" .I f1 ; : : : ; fn / WD

Z

Z

Q 2 E fi d fi d Q < "

n

\

i D1

115

with 2 E, n 2 N, fi 2 C .S/, and " > 0 form a base of the topology .E; C .S//.

Thus, if U E is open, then every point

2 U \ M1 .S/ possesses some neighborhood U" .

I f1 ; : : : ; fn / U . But U" .

I f1 ; : : : ; fn / \ M1 .S/ is an open neighborhood of

in the -weak topology. Hence, U \ M1 .S/ is open in the -weak

topology. Similarly, one shows that every open set V M1 .S/ is of the form

V D U \M1 .S/ for some open subset U of E. This shows that the relative topology

M1 .S/ \ .E; C .S// coincides with the -weak topology.

Moreover, M1 .S/ is an intersection of closed subsets of E

Z

Z

M1 .S/ D 2 E

1 d D 1 \

2E

f d 0 :

f 2C .S/

f 0

Next, let E 2 denote the product space E E. We endow E 2 with the product

topology for which the sets U V with U; V 2 .E; C .S// form a neighborhood

base. Clearly, E 2 is a locally convex topological vector space.

Lemma 2.92. Every continuous linear functional ` on E 2 is of the form

Z

Z

`.1 ; 2 / D f1 d1 C f2 d2

for some f1 ; f2 2 C .S/.

Proof. By linearity, ` is of the form `.1 ; 2 / D `1 .1 / C `2 .2 /, where `1 .1 / WD

`.1 ; 0/ and `2 .2 / WD `.0; 2 /. By continuity of `, the set

V WD `1 ..1; 1//

is open in E 2 and contains the point .0; 0/. Hence, there are two open neighborhoods

U1 ; U2 E such that .0; 0/ 2 U1 U2 V . Therefore,

0 2 Ui `1

i ..1; 1//

for i D 1; 2,

i ..1; 1//. It follows that the `i are continuous at 0,

which in view of their linearity implies continuity everywhere on E.R Finally, we may

conclude from Proposition A.59 that each `i is of the form `i ./ D fi di for some

fi 2 C .S/.

The proof of the following lemma uses the characterization of compact sets for

the -weak topology that is stated in Corollary A.47. It is here that we need our

assumption that S is Polish.

116

Chapter 2 Preferences

N

H WD .1 ; 2 / j 2

is a closed convex subset of E 2 .

Proof. It is enough to show that H is closed in M1 .S/2 WD M1 .S/ M1 .S/,

because Lemma 2.91 implies that the relative topology induced by E 2 on M1 .S/2

coincides with the product topology for the -weak topology. This is a metric topology by Corollary A.45. So let .

n ; n / 2 H , n 2 N, be a sequence converging

to some .

; / 2 M1 .S/2 in the product topology. Since both sequences .

n /n2N

and .n /n2N are relatively compact for the -weak topology, Corollary A.47 yields

functions i W S ! 1; 1, i D 1; 2, such that sets of the form Kki WD i k ,

k 2 N, are relatively compact in S and such that

Z

Z

sup 1 d

n C sup 2 dn < 1 :

n2N

n2N

n and 2 n D n . Hence, if we

let .x; y/ WD 1 .x/ C 2 .y/, then

Z

Z

Z

sup d n D sup

1 d

n C 2 dn < 1 :

n2N

n2N

this claim, let li 2 N be such that

li sup

.x/ :

x2Kki

Then, since

1,

2

1

[ Kk.1Cl

Kk2 ;

k Kk1 Kk.1Cl

1/

2/

and the right-hand side is a relatively compact set in S S . It follows from Corollary A.47 that the sequence .
n /n2N is relatively compact for the -weak topology.

Any accumulation point
of this sequence belongs to the closed set . Moreover,

has marginal distributions

and , since the projections i are continuous according

to Lemma 2.90. Hence .

; / 2 H .

Proof of Theorem 2:89. Let

1 ,

2 2 M1 .S/ be given. Since H is closed and

convex in E 2 by Lemma 2.93, we may apply Theorem A.57 with B WD .

1 ;

2 /

and C WD H : We conclude that .

1 ;

2 / H if and only if there exists a linear

functional ` on E 2 such that

`.

1 ;

2 / >

sup

`.1 ; 2 / D sup `.1 ; 2 /:

.1 ;2 /2H

117

We will now use Theorem 2.89 to deduce the remaining implication of Theorem 2.57. We consider here a more general, d -dimensional setting. To this end,

let x D .x 1 ; : : : ; x d / and y D .y 1 ; : : : ; y d / be two d -dimensional vectors. We will

say that x y if x i y i for all i. A function on Rd is called increasing, if it is

increasing with respect to the partial order .

d

RTheorem 2.94. Suppose

1 and

2 are Borel probability measures on R with

jxj

i .dx/ < 1 for i D 1; 2. Then the following assertions are equivalent:

R

R

(a) f d

1 f d

2 for all increasing concave functions f on Rd .

(b) There exists a probability space .; F ; P / with random variables X1 and X2

having distributions

1 and

2 , respectively, such that

E X2 j X1 X1

P -a.s.

Z

y Q.x; dy/ x for all x 2 Rd

Rd

2 D

1 Q.

Proof. (a) ) (b): We will apply Theorem 2.89 with S WD Rd and with the gauge

functions .x/ WD 1 C jxj and .x; y/ WD .x/ C .y/. We denote by Cb .Rd / the

set of bounded and continuous functions on Rd . Let

Z

Z

2 M1 .Rd Rd /

WD

yf .x/
.dx; dy/ xf .x/
.dx; dy/ :

f 2Cb .Rd /

Each single set of the intersection is convex and closed in M1 .Rd Rd /, because

the functions g.x; y/ WD yf .x/ and g.x;

Q

y/ WD xf .x/ belong to C .Rd Rd / for

f 2 Cb .S/. Therefore, itself is convex and closed.

Suppose we can show that contains an element P that has

1 and

2 as marginal

distributions. Then we can take WD Rd Rd with its Borel -algebra F , and let

X1 and X2 denote the canonical projections on the first and the second components,

respectively. By definition, Xi will have the distribution

i , and

E E X2 j X1 f .X1 / D E X2 f .X1 / E X1 f .X1 /

By monotone class arguments, we may thus conclude that

E X2 j X1 X1

so that the assertion will follow.

P -a.s.

118

Chapter 2 Preferences

It remains to prove the existence of P . To this end, we will apply Theorem 2.89

with the set defined above. Take a pair f1 ; f2 2 C .Rd /, and let

fQ2 .x/ WD infg.x/ j g is concave, increasing, and dominates f2 :

Then fQ2 is concave, increasing, and dominates f2 . In fact, fQ2 is the smallest function

with these properties. We have

Z

Z

Z

Z

f1 d

1 C f2 d

2 f1 d

1 C fQ2 d

2

Z

.f1 C fQ2 / d

1

sup .f1 .x/ C fQ2 .x// DW r0 :

x2Rd

We will establish the condition in Theorem 2.89 for our set by showing that for

r < r0 we have

Z

r < sup .f1 .x/ C f2 .y//
.dx; dy/:

Z

d

z WD 2 M1 .R /

x .dx/ z

and

Z

g2 .z/ WD sup

f2 d 2 z :

and 2 2 z2 , then

1 C .1 /2 2 z1 C.1/z2

for 2 0; 1. Therefore, g2 is concave, and we conclude that g2 fQ2 (recall

that fQ2 is the smallest increasing and concave function dominating f2 ). Hence, r <

f1 .z/ C g2 .z/ for some z 2 Rd , i.e., there exists some 2 z such that the product

measure WD z satisfies

Z

Z

r < f1 .z/ C f2 d D .f1 .x/ C f2 .y// .dx; dy/:

But D z 2 .

(b) ) (c): This follows as in the proof of the implication (f) ) (g) of Theorem 2.57

by using regular conditional distributions.

(c) ) (a): As in the proof of (g) ) (a) of Theorem 2.57, this follows by an application of Jensens inequality.

119

By the same arguments as for Corollary 2.61, we obtain the following result from

Theorem 2.94.

d

RCorollary 2.95. Suppose

1 and

2 are Borel probability measures on R such that

jxj

i .dx/ < 1, for i D 1; 2. Then the following conditions are equivalent:

R

R

(a) f d

1 f d

2 for all concave functions f on Rd .

(b) There exists a probability space .; F ; P / with random variables X1 and X2

having distributions

1 and

2 , respectively, such that

E X2 j X1 D X1

P -a.s.

Z

y Q.x; dy/ D x for all x 2 Rd

.i.e., Q is a mean-preserving spread/ and such that

2 D

1 Q.

We conclude this section with a generalization of Theorem 2.68. Let S be a Polish

space which is endowed with a preference order . We will assume that is continuous in the sense of Definition 2.8. A function on S will be called increasing if it is

increasing with respect to .

Theorem 2.96. For two Borel probability measures

1 and

2 on S , the following

conditions are equivalent.

R

R

(a) f d

1 f d

2 for all bounded, increasing, and measurable functions f

on S.

(b) There exists a probability space .; F ; P / with random variables X1 and X2

having distributions

1 and

2 , respectively, such that X1 X2 P -a.s.

(c) There exists a kernel Q on S such that

2 D

1 Q and

Q.x; y j x y / D 1

for all x 2 S.

Proof. (a) ) (b): We will apply Theorem 2.89 with the gauge function 1, so

that M1 .S/ is just the space M1 .S/ of all Borel probability measures on S with the

usual weak topology. Then 2 which is equivalent to taking WD 1. Let

M WD .x; y/ 2 S S j x
y:

This set M is closed in S S by Proposition 2.11. Hence, the portmanteau theorem

in the form of Theorem A.39 implies that the convex set

WD
2 M1 .S S/ j
.M / D 1

120

Chapter 2 Preferences

fQ2 .x/ WD supf2 .y/ j x y:

Then fQ2 is bounded, increasing, and dominates f2 . Therefore, if f1 2 Cb .S/,

Z

Z

Z

Z

f1 d

1 C f2 d

2 f1 d

1 C fQ2 d

2

Z

.f1 C fQ2 / d

1

sup .f1 .x/ C fQ2 .x//

x2S

xy

Z

sup .f1 .x/ C f2 .y// D sup .f1 .x/ C f2 .y// .dx; dy/:

xy

Hence, all assumptions of Theorem 2.89 are satisfied, and we conclude that there

exists a probability measure P 2 with marginals

1 and

2 . Taking WD S S

and Xi as the projection on the i th coordinate finishes the proof of (a) ) (b).

(b) ) (c) follows as in the proof of Theorem 2.57 by using regular conditional

distributions.

(c) ) (a) is proved as the corresponding implication of Theorem 2.68.

Chapter 3

In Section 3.1, we discuss the problem of constructing a portfolio which maximizes

the expected utility of the resulting payoff. The existence of an optimal solution is

equivalent to the absence of arbitrage opportunities. This leads to an alternative proof

of the fundamental theorem of asset pricing, and to a specific choice of an equivalent martingale measure defined in terms of marginal utility. Section 3.2 contains

a detailed case study describing the interplay between exponential utility and relative

entropy. In Section 3.3, the optimization problem is formulated for general contingent

claims. Typically, optimal profiles will be non-linear functions of a given market portfolio, and this is one source of the demand for financial derivatives. Section 3.6 introduces the idea of market equilibrium. Prices of risky assets will no longer be given in

advance; they will be derived as equilibrium prices in a microeconomic setting, where

different agents demand contingent claims in accordance with their preferences and

with their budget constraints.

3.1

Let us consider the one-period market model of Section 1.1 in which d C 1 assets are

priced at time 0 and at time 1. Prices at time 0 are given by the price system

D . 0 ; / D . 0 ; 1 ; : : : ; d / 2 RdCC1 ;

prices at time 1 are modeled by the price vector

S D .S 0 ; S / D .S 0 ; S 1 ; : : : ; S d /

consisting of non-negative random variables S i defined on some probability space

.; F ; P /. The 0th asset models a riskless bond, and so we assume that

0 D 1 and

S0 1 C r

D . 0 ; / D . 0 ; 1 ; : : : ; d / 2 Rd C1

122

where i represents the amount of shares of the i th asset. Such a portfolio requires

an initial investment and yields at time 1 the random payoff S.

Consider a risk-averse economic agent whose preferences are described in terms

of a utility function u,

Q and who wishes to invest a given amount w into the financial

market. Recall from Definition 2.35 that a real-valued function uQ is called a utility

function if it is continuous, strictly increasing, and strictly concave. A rational choice

of the investors portfolio D . 0 ; / will be based on the expected utility

E u.

Q S/

(3.1)

of the payoff S at time 1, where the portfolio satisfies the budget constraint

w:

(3.2)

Thus, the problem is to maximize the expected utility (3.1) among all portfolios

2 Rd C1 which satisfy the budget constraint (3.2). Here we make the implicit

assumption that the payoff S is P -a.s. contained in the domain of definition of the

utility function u.

Q

In a first step, we remove the constraint (3.2) by considering instead of (3.1) the

expected utility of the discounted net gain

S

D Y

1Cr

earned by a portfolio D . 0 ; /. Here Y is the d -dimensional random vector with

components

Si

i ; i D 1; : : : ; d:

Yi D

1Cr

For any portfolio with < w, adding the risk-free investment w would

lead to the strictly better portfolio . 0 Cw ; /. Thus, we can focus on portfolios

which satisfy D w, and then the payoff is an affine function of the discounted

net gain

S D .1 C r/. Y C w/:

Moreover, for any 2 Rd there exists a unique numraire component 0 2 R such

that the portfolio WD . 0 ; / satisfies D w.

Let u denote the following transformation of our original utility function u:

Q

u.y/ WD u..1

Q

C r/.y C w//:

Note that u is again a utility function, and that CARA and (shifted) HARA utility

functions are transformed into utility functions in the same class.

123

unconstrained problem of maximizing the expected utility E u. Y / among all

2 Rd such that Y is contained in the domain D of u.

Assumption 3.1. We assume one of the following two cases:

(a) D D R. In this case, we will admit all portfolios 2 Rd , but we assume that u

is bounded from above.

(b) D D a; 1/ for some a < 0. In this case, we only consider portfolios which

satisfy the constraint

Y a P -a.s.,

and we assume that the expected utility generated by such portfolios is finite,

i.e.,

E u. Y / < 1 for all 2 Rd with Y a P -a.s.

Remark 3.2. Part (a) of this assumption is clearly satisfied in the case of an exponential utility function u.x/ D 1 e x . Domains of the form D D a; 1/ appear,

for example, in the case of (shifted) HARA utility functions u.x/ D log.x b/ for

b < a and u.x/ D 1 .x c/ for c a and 0 < < 1. The integrability assumption

in (b) holds if E jY j < 1, because any concave function is bounded from above by

an affine function.

}

In order to simplify notations, let us denote by

S.D/ WD 2 Rd j Y 2 D P -a.s.

the set of admissible portfolios for D. Clearly, S.D/ D Rd if D D R. Our aim is

to find some 2 S.D/ which is optimal in the sense that it maximizes the expected

utility E u. Y / among all 2 S.D/. In this case, will be an optimal investment

strategy into the risky assets. Complementing with a suitable numraire component

0 yields a portfolio D . 0 ; / which maximizes the expected utility E u.

Q S/

under the budget constraint D w. Our first result in this section will relate the

existence of such an optimal portfolio to the absence of arbitrage opportunities.

Theorem 3.3. Suppose that the utility function u W D ! R satisfies Assumption 3:1.

Then there exists a maximizer of the expected utility

E u. Y / ;

2 S.D/;

if and only if the market model is arbitrage-free. Moreover, there exists at most one

maximizer if the market model is non-redundant in the sense of Definition 1:15.

124

Proof. The uniqueness part of the assertion follows immediately from the strict concavity of the function 7! E u. Y / for non-redundant market models. As to

existence, we may assume without loss of generality that our model is non-redundant.

If the non-redundance condition (1.8) does not hold, then we define a linear space

N Rd by

N WD 2 Rd j Y D 0 P -a.s.:

Clearly, Y takes P -a.s. values in the orthogonal complement N ? of N . Moreover,

the no-arbitrage condition (1.3) holds for all 2 Rd if and only if it is satisfied for

all 2 N ? . By identifying N ? with some Rn , we arrive at a situation in which the

non-redundance condition (1.8) is satisfied and where we may apply our result for

non-redundant market models.

If the model admits arbitrage opportunities, then a maximizer of the expected

utility E u. Y / cannot exist: Adding to some non-zero 2 Rd for which

Y 0 P -a.s., which exists by Lemma 1.4, would yield a contradiction to the

optimality of , because then

E u. Y / < E u.. C / Y / :

From now on, we assume that the market model is arbitrage-free. Let us first consider the case in which D D a; 1/ for some a 2 .1; 0/. Then S.D/ is compact.

In order to prove this claim, suppose by way of contradiction that .n / is a sequence in

S.D/ such that jn j ! 1. By choosing a subsequence if necessary, we may assume

that n WD n =jn j converges to some unit vector 2 Rd . Clearly,

n Y

a

lim

D0

n"1 jn j

n"1 jn j

Y D lim

P -a.s.,

In the next step, we show that our assumptions guarantee the continuity of the

function

S.D/ 3 7! E u. Y / ;

which, in view of the compactness of S.D/, will imply the existence of a maximizer

of the expected utility. To this end, it suffices to construct an integrable random variable which dominates u. Y / for all 2 S.D/. Define 2 Rd by

i WD 0 _ max i < 1:

2S.D/

Y D

S

S

0 ^ min 0 :

1Cr

1Cr

0 2S.D/

Note that Y is bounded from below by and that there exists some 2 .0; 1

such that < jaj. Hence 2 S.D/, and so E u. Y / < 1. Applying

125

Lemma 3.4 below first with b WD and then with b WD 0 ^ min 0 2S.D/ 0

shows that

S

0

< 1:

E u

0 ^ min

1Cr

0 2S.D/

This concludes the proof of the theorem in case D D a; 1/.

Let us now turn to the case of a utility function on D D R which is bounded from

above. We will reduce the assertion to a general existence criterion for minimizers of

lower semicontinuous convex functions on Rd , given in Lemma 3.5 below. It will be

applied to the convex function h./ WD E u. Y / . We must show that h is lower

semicontinuous. Take a sequence .n /n2N in Rd converging to some . By part (a) of

Assumption 3.1, the random variables u.n Y / are uniformly bounded from below,

and so we may apply Fatous lemma:

lim inf h.n / D lim inf E u.n Y / E u. Y / D h./:

n"1

n"1

By our non-redundance assumption, h is strictly convex and admits at most one

minimizer. We claim that the absence of arbitrage opportunities is equivalent to the

following condition:

lim h. / D C1

"1

(3.3)

This is just the condition (3.4) required in Lemma 3.5. It follows from (1.3) and

(1.8) that a non-redundant market model is arbitrage-free if and only if each non-zero

2 Rd satisfies P Y < 0 > 0. Since the utility function u is strictly increasing

and concave, the set Y < 0 can be described as

Y < 0 D

lim u. Y / D 1

"1

for 2 Rd .

lim E u. Y / D 1;

"1

because u is bounded from above. This observation proves that the absence of arbitrage opportunities is equivalent to the condition (3.3) and completes the proof.

Lemma 3.4. If D D a; 1/, b < jaj, 0 < 1, and X is a non-negative random

variable, then

E u.X b/ < 1

H)

E u.X / < 1:

126

u.X / u.0/

u.X b/ u.b/

u.X / u.0/

:

X 0

X 0

X b .b/

Multiplying by X shows that u.X / can be dominated by a multiple of u.X b/ plus

some constant.

Lemma 3.5. Suppose h W Rd ! R [ C1 is a convex and lower semicontinuous

function with h.0/ < 1. Then h attains its infimum provided that

lim h. / D C1

"1

(3.4)

Moreover, if h is strictly convex on h < 1, then also the converse implication holds:

the existence of a minimizer implies (3.4).

Proof. First suppose that (3.4) holds. We will show below that for c > inf h the level

sets x j h.x/ c of h are bounded and hence compact. Once the compactness of

the level sets is established, it follows that the set

\

x 2 Rd j h.x/ c

x 2 Rd j h.x/ D inf h D

c>inf h

Suppose c > inf h is such that the level set h c is not compact, and take

a sequence .xn / in h c such that jxn j ! 1. By passing to a subsequence

if necessary, we may assume that xn =jxn j converges to some non-zero . For any

> 0,

xn

D lim inf h

xn C 1

0

h./ lim inf h

jxn j

jxn j

jxn j

n"1

n"1

h.0/

cC 1

lim inf

jxn j

jxn j

n"1

D h.0/:

Thus, we arrive at a contradiction to condition (3.4). This completes the proof of the

existence of a minimizer under assumption (3.4).

In order to prove the converse implication, suppose that the strictly convex function

h has a minimizer x but that there exists a non-zero 2 Rd violating (3.4), i.e., there

exists a sequence .n /n2N and some c < 1 such that n " 1 but h.n / c for

all n. Let

xn WD
n x C .1
n /n

127

the compactness of the Euclidean unit sphere centered in x , we may assume that xn

converges to some x. Then necessarily jx x j D 1. As n diverges, we must have

that n ! 1. By using our assumption that h.n / is bounded, we obtain

h.x/ lim inf h.xn / lim . n h.x / C .1 n /h.n // D h.x /:

n"1

n"1

Thus, (3.4) must hold if the strictly convex function h takes on its infimum.

Remark 3.6. Note that the proof of Theorem 3.3 under Assumption 3.1 (a) did not

use the fact that the components of Y are bounded from below. The result remains

true for arbitrary Y .

}

We turn now to a characterization of the solution of our utility maximization

problem for continuously differentiable utility functions.

Proposition 3.7. Let u be a continuously differentiable utility function on D such

that E u. Y / is finite for all 2 S.D/. Suppose that is a solution of the utility

maximization problem, and that one of the following two sets of conditions is satisfied:

Then

u0 . Y / jY j 2 L1 .P /;

and the following first-order condition holds:

E u0 . Y / Y D 0:

(3.5)

Proof. For 2 S.D/ and " 2 .0; 1 let " WD " C .1 "/ , and define

" WD

u." Y / u. Y /

:

"

" % u0 . Y / . / Y

as " # 0.

Note that our assumptions imply that u.Y / 2 L1 .P / for all 2 S.D/. In particular,

we have 1 2 L1 .P /, so that monotone convergence and the optimality of yield

that

(3.6)

0 E " % E u0 . Y / . / Y as " # 0.

In particular, the expectation on the right-hand side of (3.6) is finite.

128

deduce from (3.6) by letting WD that

E u0 . Y / Y 0

for all in a small ball centered in the origin of Rd . Replacing by shows that

the expectation must vanish.

Remark 3.8. Let us comment on the assumption that the optimal is an interior

point of S.D/:

(a) If the non-redundance condition (1.8) is not satisfied, then either each or none of

the solutions to the utility maximization problem is contained in the interior of

S.D/. This can be seen by using the reduction argument given at the beginning

of the proof of Theorem 3.3.

(b) Note that Y is bounded from below by in case has only non-negative

components. Thus, the interior of S.D/ is always non-empty.

(c) As shown by the following example, the optimal need not be contained in the

interior of S.D/ and, in this case, the first-order condition (3.5) will generally

fail.

}

Example 3.9. Take r D 0, and let S 1 be integrable but unbounded. We choose

D D a; 1/ with a WD 1 , and we assume that P S 1 " > 0 for all " > 0.

Then S.D/ D 0; 1. If 0 < E S 1 < 1 then Example 2.40 shows that the optimal

investment is given by D 0, and so lies in the boundary of S.D/. Thus, if u is

sufficiently smooth,

E u0 . Y / Y D u0 .0/.E S 1 1 / < 0:

The intuitive reason for this failure of the first-order condition is that taking a short

position in the asset would be optimal as soon as E S 1 < 1 . This choice, however,

is ruled out by the constraint 2 S.D/.

}

Proposition 3.7 yields a formula for the density of a particular equivalent riskneutral measure. Recall that P is risk-neutral if and only if E Y D 0.

Corollary 3.10. Suppose that the market model is arbitrage-free and that the assumptions of Proposition 3:7 are satisfied for a utility function u W D ! R and an

associated maximizer of the expected utility Eu. Y /. Then

u0 . Y /

dP

D

dP

Eu0 . Y /

defines an equivalent risk-neutral measure.

(3.7)

129

that its expectation vanishes. Hence, we may conclude that P is an equivalent riskneutral measure if we can show that P is well-defined by (3.7), i.e., if u0 . Y / 2

L1 .P /. Let

for D D a; 1/,

u0 .a/

c WD supu0 .x/ j x 2 D and jxj j j

0 .j j/ for D D R,

u

which is finite by our assumption that u is continuously differentiable on all of D.

Thus,

0 u0 . Y / c C u0 . Y /jY j IjY j1 ;

and the right-hand side has a finite expectation.

Remark 3.11. Corollary 3.10 yields an independent and constructive proof of the

fundamental theorem of asset pricing in the form of Theorem 1.7: Suppose that the

model is arbitrage-free. If Y is P -a.s. bounded, then so is u. Y /, and the measure

P of (3.7) is an equivalent risk-neutral measure with a bounded density dP =dP . If

Y is unbounded, then we may consider the bounded random vector

YQ WD

Y

;

1 C jY j

which also satisfies the no-arbitrage condition (1.3). Let Q be a maximizer of the

expected utility E u. YQ / . Then an equivalent risk-neutral measure P is defined

through the bounded density

u0 . Q YQ /

dP

WD c

;

dP

1 C jY j

where c is an appropriate normalizing constant.

u.x/ D 1 e x

with constant absolute risk aversion > 0. The requirement that E u. Y / is finite

is equivalent to the condition

E e Y < 1

for all 2 Rd .

If is a maximizer of the expected utility, then the density of the equivalent riskneutral measure P in (3.7) takes the particular form

e Y

dP

D

:

dP

E e Y

130

if and only if WD is a minimizer of the moment generating function

Z. / WD E e

Y ;

2 Rd ;

of Y . In Corollary 3.25 below, the measure P will be characterized by the fact that

it minimizes the relative entropy with respect to P among the risk-neutral measures

in P ; see Definition 3.20 below.

}

3.2

In this section we give a more detailed study of the problem of portfolio optimization

with respect to a CARA utility function

u.x/ D 1 e x

for > 0. As in the previous Section 3.1, the problem is to maximize the expected

utility

E u. Y /

of the discounted net gain Y earned by an investment into risky assets. The key

assumption for this problem is that

E u. Y / > 1

for all 2 Rd .

(3.8)

Recall from Example 3.12 that the maximization of E u. Y / is reduced to the

minimization of the moment generating function

Z.
/ WD E e

Y ;

2 Rd ;

which does not depend on the risk aversion . The key assumption (3.8) is equivalent

to the condition that

(3.9)

Z.
/ < 1 for all
2 Rd .

Throughout this section, we will always assume that (3.9) holds. But we will not need

the assumption that Y is bounded from below (which in our financial market model

follows from assuming that asset prices are non-negative); all results remain true for

general random vectors Y ; see also Remarks 1.9 and 3.6.

Lemma 3.13. The condition (3.9) is equivalent to

E e jY j < 1 for all > 0.

131

Proof. Clearly, the condition in the statement of the lemma implies (3.9). To prove

P

the converse assertion, take a constant c > 0 such that jxj c diD1 jx i j for x 2 Rd .

By Hlders inequality,

d

d

h

X

i Y

i

jY i j

E e cd jY j 1=d :

E e jY j E exp c

i D1

i D1

In order to show that the i th factor on the right is finite, take
2 Rd such that

i D cd and
j D 0 for j i. With this choice,

i

E e cd jY j E e

Y C E e

Y ;

which is finite by (3.9).

Definition 3.14. The exponential family of P with respect to Y is the set of measures

P

j
2 Rd

defined via

e

Y

dP

D

:

dP

Z.
/

Example 3.15. Suppose that the risky asset S 1 has under P a Poisson distribution

with parameter > 0, i.e., S 1 takes values in 0; 1; : : : and satisfies

P S 1 D k D e

k

;

k

k D 0; 1; : : : :

a Poisson distribution

with parameter e

. Hence, the exponential family of P generates the family of all

Poisson distributions.

}

Example 3.16. Let Y have a standard normal distribution N.0; 1/. Then (3.9) is

satisfied, and the distribution of Y under P

is equal to the normal distribution N. ; 1/

with mean and variance 1.

}

Remark 3.17. Two parameters and 0 in Rd determine the same element in the

exponential family of P if and only if . 0 / Y D 0 P -almost surely. It follows

that the mapping

7! P

Y D 0 P -a.s.

H)

D 0:

(3.10)

}

132

In the sequel, we will be interested in the barycenters of the members of the exponential family of P with respect to Y . We denote

m.
/ WD E

Y D

1

E Y e

Y ;

Z.
/

2 Rd :

The next lemma shows that m.
/ can be obtained as the gradient of the logarithmic

moment generating function.

Lemma 3.18. Z is a smooth function on Rd , and the gradient of log Z at
is the

expectation of Y under P

.r log Z/.
/ D E

Y D m.
/:

Moreover, the Hessian of log Z at
equals the covariance matrix .covP .Y i ; Y j //i;j

of Y under the measure P

@2

log Z.
/ D cov.Y i ; Y j / D E

Y i Y j E

Y i E

Y j :

@
i @
j

P

In particular, log Z is convex.

Proof. Observe that

x

@e

i

x

exp.1 C j
j/ jxj:

@
i D jx j e

Hence, Lemma 3.13 and Lebesgues dominated convergence theorem justify the interchanging of differentiation and integration (see the differentiation lemma in [21],

16, for details).

The following corollary summarizes the results we have obtained so far. Recall

from Section 1.5 the notion of the convex hull ./ of the support of a measure on

Rd and the definition of the relative interior ri C of a convex set C .

Corollary 3.19. Denote by

WD P Y 1 the distribution of Y under P . Then the

function

7!
m0 log Z.
/

takes on its maximum if and only if m0 is contained in the relative interior of the

convex hull of the support of

, i.e., if and only if

m0 2 ri .

/:

In this case, any maximizer
satisfies

m0 D m.
/ D E

Y :

In particular, the set m.
/ j
2 Rd coincides with ri .

/. Moreover, if the

non-redundance condition (3.10) holds, then there exists at most one maximizer
.

133

Proof. Taking YQ WD Y m0 reduces the problem to the situation where m0 D 0. Applying Theorem 3.3 with the utility function u.z/ D 1 e z shows that the existence

of a maximizer
of logZ is equivalent to the absence of arbitrage opportunities.

Corollary 3.10 states that m.
/ D 0 and that 0 belongs to Mb .

/, where Mb .

/ was

defined in Lemma 1.44. An application of Theorem 1.49 completes the proof.

It will turn out that the maximization problem of the previous corollary is closely

related to the following concept.

Definition 3.20. The relative entropy of a probability measure Q with respect to P

is defined as

dQ

log

if Q

P ,

E dQ

dP

dP

H.QjP / WD

C1

otherwise.

Remark 3.21. Jensens inequality applied to the strictly convex function h.x/ D

x log x yields

dQ

H.QjP / D E h

h.1/ D 0;

(3.11)

dP

with equality if and only if Q D P .

}

Example 3.22. Let be a finite set and F be its power set. Every probability Q on

.; F / is absolutely continuous with respect to the uniform distribution P . Let us

denote Q.!/ WD Q ! . Clearly,

X

X

Q.!/

Q.!/ log

Q.!/ log Q.!/ C log jj:

D

H.QjP / D

P .!/

!2

!2

The quantity

H.Q/ WD

!2

is usually called the entropy of Q. Observe that H.P / D log jj, so that

H.QjP / D H.P / H.Q/:

Since the left-hand side is non-negative by (3.11), the uniform distribution P has

maximal entropy among all probability distributions on .; F /.

}

Example 3.23. Let

D N.m; 2 / denote the normal distribution with mean m and

Q Q 2 /

variance 2 on R. Then, for

Q D N.m;

d

Q

.x m/2

.x m/

Q 2

;

C

.x/ D exp

d

Q

2 Q 2

2 2

and hence

1 Q 2

Q 2

Q 2

1 mm

H.

j

/

Q

D

log 2 1 C

:

2 2

2

134

is the unique minimizer of the relative entropy

H.QjP / among all probability measures Q with EQ Y D E

Y .

Theorem 3.24. Let m0 WD m.P

0 / for some given 0 2 Rd . Then, for any probability measure Q on .; F / such that EQ Y D m0 ,

H.QjP / H.P

0 jP / D 0 m0 log Z. 0 /;

and equality holds if and only if Q D P

0 . Moreover, 0 maximizes the function

m0 log Z. /

over all 2 Rd .

Proof. Let Q be a probability measure on .; F / such that EQ Y D m0 . We show

first that for all 2 Rd

H.QjP / D H.QjP

/ C m0 log Z. /:

(3.12)

P . Otherwise

dQ dP

dQ e

Y

dQ

D

D

;

dP

dP

dP

dP

Z.
/

and taking logarithms and integrating with respect to Q yields (3.12).

Since H.QjP

/ 0 according to (3.11), we get from (3.12) that

H.QjP /
m0 log Z.
/

(3.13)

for all
2 Rd and all measures Q such that EQ Y D m0 . Moreover, equality holds

in (3.13) if and only if H.QjP

/ D 0, which is equivalent to Q D P

. In this case,

must be such that m.
/ D m0 . In particular, for any such

H.P

jP / D
m0 log Z.
/:

Thus,
0 maximizes the right-hand side of (3.13), and P

0 minimizes the relative

entropy on the set

M0 WD Q j EQ Y D m0 :

But the relative entropy H.QjP / is a strictly convex functional of Q, and so it can

have at most one minimizer in the convex set M0 . Thus, any
with m.
/ D m0

induces the same measure P

0 .

Taking m0 D 0 in the preceding theorem yields a special equivalent risk-neutral

measure in our financial market model, namely the entropy-minimizing risk-neutral

measure. Sometimes it is also called the Esscher transform of P . Recall our assumption (3.9).

135

Corollary 3.25. Suppose the market model is arbitrage-free. Then there exists a

unique equivalent risk-neutral measure P 2 P which minimizes the relative entropy

H.PO jP / over all PO 2 P . The density of P is of the form

dP

e

Y

;

D

dP

E e

Y

where
denotes a minimizer of the moment generating function E e

Y of Y .

Proof. This follows immediately from Corollary 3.19 and Theorem 3.24.

By combining Theorem 3.24 with Remark 3.17, we obtain the following corollary.

It clarifies the question of uniqueness in the representation of points in the relative

interior of .P Y 1 / as barycenters of the exponential family.

Corollary 3.26. If the non-redundance condition (3.10) holds, then

7! m.
/

is a bijective mapping from Rd to ri .P Y 1 /.

Remark 3.27. It follows from Corollary 3.19 and Theorem 3.24 that for all m 2

ri .P Y 1 /

min

EQ Y Dm

2Rd

(3.14)

Here, the right-hand side is the FenchelLegendre transform of the convex function

}

log Z evaluated at m 2 Rd .

The following theorem shows that the variational principle (3.14) remains true for

all m 2 Rd , if we replace min and max by inf and sup.

Theorem 3.28. For m 2 Rd

H.QjP / D sup
m log Z.
/:

inf

EQ Y Dm

2Rd

The proof of this theorem relies on the following two general lemmas.

Lemma 3.29. For any probability measure Q,

H.QjP / D

sup

Z2L1 .;F

.EQ Z log E e Z /

;P /

D supEQ Z log E e Z j e Z 2 L1 .P /:

The second supremum is attained by Z WD log dQ

if Q

P .

dP

(3.15)

136

Proof. We first show in (3.15). To this end, we may assume that H.QjP / < 1.

For Z with e Z 2 L1 .P / let P Z be defined by

eZ

dP Z

D

:

dP

E e Z

Then P Z is equivalent to P and

log

dP Z

dQ

dQ

C

log

D log

:

dP

dP

dP Z

H.QjP / D H.QjP Z / C EQ Z log E e Z :

Since H.QjP Z / 0 by (3.11), we have proved that H.QjP / is larger than or equal

to both suprema on the right of (3.15).

To prove the reverse inequality, consider first the case Q 6

P . Take Zn WD nIA

where A is such that Q A > 0 and P A D 0. Then, as n " 1,

EQ Zn log E e Zn D n Q A ! 1 D H.QjP /:

Now suppose that Q

P with density ' D dQ=dP . Then Z WD log ' satisfies

e Z 2 L1 .P / and

H.QjP / D EQ Z log E e Z :

For the first identity we use an approximation argument. Let Zn D .n/_.log '/^n.

We split the expectation E e Zn according to the two sets ' 1 and ' < 1.

Using monotone convergence for the first integral and dominated convergence for the

second yields

E e Zn ! E e log ' D 1:

Since x log x 1=e, we have 'Zn 1=e uniformly in n, and Fatous lemma

yields

lim inf EQ Zn D lim inf E 'Zn E ' log ' D H.QjP /:

n"1

n"1

lim inf.EQ Zn log E e Zn / H.QjP /;

n"1

137

Remark 3.30. The preceding lemma shows that the relative entropy is monotone with

respect to an increase of the underlying -algebra: Let P and Q be two probability

measures on a measurable space .; F /, and denote by H.QjP / their relative entropy. Suppose that F0 is a -field such that F0 F and denote by H0 .QjP / the

relative entropy of Q with respect to P considered as probability measures on the

smaller space .; F0 /. Then the relation L1 .; F0 ; P / L1 .; F ; P / implies

H0 .QjP / H.QjP /I

}

Lemma 3.31. For all 0, the set

WD ' 2 L1 .; F ; P / j ' 0; E ' D 1; E ' log '

is weakly sequentially compact in L1 .; F ; P /.

Proof. Let Lp WD Lp .; F ; P /. The set of all P -densities,

D WD ' 2 L1 j ' 0; E ' D 1 ;

is clearly convex and closed in L1 . Hence, this set is also weakly closed in L1 by

Theorem A.60. Moreover, Lemma 3.29 states that for ' 2 D

E ' log ' D sup .E Z ' log E e Z /:

Z2L1

In particular,

' 7! E ' log '

is a weakly lower semicontinuous functional on D, and so is weakly closed. In

addition, is bounded in L1 and uniformly integrable, due to the criterion of de

la Valle Poussin; see, e.g., Lemma 3 in 6 of Chapter II of [251]. Applying the

DunfordPettis theorem and the Eberleinmulian theorem as stated in Appendix A.7

concludes the proof.

Proof of Theorem 3:28. In view of Theorem 3.24 and inequality (3.13) (whose proof

extends to all m 2 Rd ), it remains to prove that

inf

EQ Y Dm

(3.16)

2Rd

/, where

WD P Y 1 . The right-hand side

of (3.16) is just the FenchelLegendre transform at m of the convex function log Z

and, thus, denoted .log Z/ .m/.

138

/ of the

convex hull of the support of

. Proposition A.1, the separating hyperplane theorem,

yields some 2 Rd such that

m > sup x j x 2 .

/ sup x j x 2 supp

:

By taking n WD n, it follows that

n m log Z. n / n m

sup y ! C1

as n " 1.

y2supp

/.

It remains to prove (3.16) for m 2 .

/n ri .

/ with .log Z/ .m/ < 1. Recall

from (1.25) that ri .

/ D ri .

/. Pick some m1 2 ri .

/ and let

1

1

mn WD m1 C 1

m:

n

n

Then mn 2 ri .

/ by (1.24). By the convexity of .log Z/ , we have

n1

1

.log Z/ .m1 / C

.log Z/ .m/

lim sup.log Z/ .mn / lim sup

n

n

n"1

n"1

(3.17)

D .log Z/ .m/:

We also know that to each mn there corresponds a
n 2 Rd such that

mn D E

n Y

and

H.P

n jP / D .log Z/ .mn /:

(3.18)

lim sup H.P

n jP / D lim sup.log Z/ .mn / .log Z/ .m/ < 1:

n"1

n"1

In particular, H.P

n jP / is uniformly bounded in n, and Lemma 3.31 implies that

after passing to a suitable subsequence if necessary the densities dP

n =dP converge

weakly in L1 .; F ; P / to a density '. Let dP1 D ' dP . By the weak lower

semicontinuity of

dQ

7! H.QjP /;

dP

which follows from Lemma 3.29, we may conclude that H.P1 jP / .log Z/ .m/.

The theorem will be proved once we can show that E1 Y D m. To this end, let

WD supn .log Z/ .mn /, which is a finite non-negative number by (3.17). Taking

Z WD IjY jc jY j

139

E

n jY j IjY jc log E exp.jY jIjY jc /

for all n 1.

Note that the rightmost expectation is finite due to condition (3.9) and Lemma 3.13.

By taking large so that = < "=2 for some given " > 0, and by choosing c such

that

"

log E exp.jY jIjY jc / <

;

2

we obtain that

sup E

n jY j IjY jc ":

n1

But

E

n jY j IjY j<c ! E1 jY j IjY j<c

by the weak convergence of dP

n =dP ! dP1 =dP , and so taking " # 0 yields

m D lim E

n Y D E1 Y ;

n"1

as desired.

3.3

E u.X /

under a given budget constraint in a broader context. The random variables X will

vary in a general convex class X L0 .; F; P / of admissible payoff profiles. In

the setting of our financial market model, this will allow us to explain the demand for

non-linear payoff profiles provided by financial derivatives.

In order to formulate the budget constraint in this general context, we introduce a

linear pricing rule of the form

.X / D E X D E 'X

where P is a probability measure on .; F /, which is equivalent to P with density

'. For a given initial wealth w 2 R, the corresponding budget set is defined as

B WD X 2 X \ L1 .P / j E X w:

(3.19)

Maximize E u.X / among all X 2 B:

(3.20)

Note, however, that we will need some extra conditions which guarantee that the

expectations E u.X / make sense and are bounded from above.

140

Remark 3.32. In general, our optimization problem would not be well posed without

the assumption P P . Note first that it should be rephrased in terms of a class

X of measurable functions on .; F / since we can no longer pass to equivalence

classes with respect to P . If P is not absolutely continuous with respect to P then

there exists A 2 F such that P A > 0 and P A D 0. For X 2 L1 .P / and

c > 0, the random variable XQ WD X C c IA would satisfy E XQ D E X and

E u.XQ / > E u.X / . Similarly, if P A > 0 and P A D 0 then

XO WD X C c

c

I

P A A

would have the same price as X but higher expected utility. In particular, the expectations in (3.20) would be unbounded in both cases if X is the class of all measurable

functions on .; F / and if the function u is not bounded from above.

}

Remark 3.33. If a solution X with E u.X / < 1 exists then it is unique, since

B is convex and u is strictly concave. Moreover, if X D L0 .; F ; P / or X D

L0C .; F ; P / then X satisfies

E X D w

since E X < w would imply that X WD X C w E X is a strictly better

choice, due to the strict monotonicity of u.

}

Let us first consider the unrestricted case X D L0 .; F; P / where any finite random variable on .; F; P / is admissible. The following heuristic argument identifies

a candidate X for the maximization of the expected utility. Suppose that a solution

X exists. For any X 2 L1 .P / and any
2 R,

X

WD X C
.X E X /

satisfies the budget constraint E X

D w. A formal computation yields

d

E u.X

/

d

D0

D E u0 .X /.X E X /

0D

D E u0 .X /X E XE u0 .X / '

D E X.u0 .X / c '/

where c WD E u0 .X / . The identity

E X u0 .X / D c E X '

for all bounded measurable X implies u0 .X / D c ' P -almost surely. Thus, if we

denote by

I WD .u0 /1

141

the inverse function of the strictly decreasing function u0 , then X should be of the

form

X D I.c '/:

We will now formulate a set of assumptions on our utility function u which guarantee that X WD I.c '/ is indeed a maximizer of the expected utility, as suggested

by the preceding argument.

Theorem 3.34. Suppose u W R ! R is a continuously differentiable utility function

which is bounded from above, and whose derivative satisfies

lim u0 .x/ D C1:

x#1

(3.21)

X WD I.c '/ 2 L1 .P /:

Then X is the unique maximizer of the expected utility E u.X / among all those

X 2 L1 .P / for which E X E X . In particular, X solves our optimization

problem (3.20) for X D L0 .; F ; P / if c can be chosen such that E X D w.

Proof. Uniqueness follows from Remark 3.33. Since u is bounded from above, its

derivative satisfies

lim u0 .x/ D 0;

x"1

in addition to (3.21). Hence, .0; 1/ is contained in the range of u0 , and it follows that

I.c '/ is P -a.s. well-defined for all c > 0.

To show the optimality of X D I.c '/, note that the concavity of u implies that

for any X 2 L1 .P /

u.X / u.X / C u0 .X /.X X / D u.X / C c '.X X /:

Taking expectations with respect to P yields

E u.X / E u.X / C c E X X :

Hence, X is indeed a maximizer in the class X 2 L1 .P / j E X E X .

absolute risk aversion > 0. In this case,

y

1

I.y/ D log :

142

It follows that

1

c

1

E I.c '/ D log E ' log '

c

1

1

D log H.P jP /;

where H.P jP / denotes the relative entropy of P with respect to P ; see Definition 3.20. Hence, the utility maximization problem can be solved for any w 2 R if

and only if the relative entropy H.P jP / is finite. In this case, the optimal profile is

given by

1

1

X D log ' C w C H.P jP /;

E u.X / D 1 exp.w H.P jP //;

corresponding to the certainty equivalent

wC

1

H.P jP /:

Let us now return to the financial market model considered in Section 3.1, and let P

be the entropy-minimizing risk-neutral measure constructed in Corollary 3.25. The

density of P is of the form

e Y

;

'D

E e Y

where 2 Rd denotes a maximizer of the expected utility E u. Y / ; see Example 3.12. In this case, the optimal profile takes the form

X D Y C w D

S

;

1Cr

is determined by the budget constraint D w. Thus, the optimal profile is given

by a linear profile in the given primary assets S 0 ; : : : ; S d : No derivatives are needed

at this point.

}

In most situations it will be natural to restrict the discussion to payoff profiles which

are non-negative. For the rest of this section we will make this restriction, and so the

utility function u may be defined only on 0; 1/. In several applications we will also

use an upper bound on payoff profiles given by an F -measurable random variable

W W ! 0; 1 . We include the case W C1 and define the convex class of

admissible payoff profiles as

X WD X 2 L0 .P / j 0 X W P -a.s.:

143

Thus, our goal is to maximize the expected utility E u.X / among all X 2 B where

the budget set B is defined in terms of X and P as in (3.19), i.e.,

B D X 2 L1 .P / j 0 X W P -a.s. and E X w :

We first formulate a general existence result:

Proposition 3.36. Let u be any utility function on 0; 1/, and suppose that W is

P -a.s. finite and satisfies E u.W / < 1. Then there exists a unique X 2 B which

maximizes the expected utility E u.X / among all X 2 B.

Proof. Take a sequence .Xn / in B with E Xn w and such that E u.Xn /

converges to the supremum of the expected utility. Since supn jXn j W < 1

P -almost surely, we obtain from Lemma 1.70 a sequence

XQ n 2 convXn ; XnC1 ; : : :

Q Clearly, every XQn

of convex combinations which converge almost-surely to some X.

is contained in B. Fatous lemma implies

E XQ lim inf E XQn w;

n"1

coefficients in 0 summing up to 1. Hence,

u.XQ n /

m

X

Pm

n

i D1 i Xni

in u.Xni /;

i D1

E u.XQ n / inf E u.Xm / :

mn

By dominated convergence,

E u.XQ / D lim E u.XQ n / ;

n"1

and the right-hand side is equal to the supremum of the expected utility.

Remark 3.37. The argument used to prove the preceding proposition works just as

well in the following general setting. Let U W B ! R be a concave functional on a

set B of random variables defined on a probability space .; F ; P / and with values

in Rn . Assume that

144

there exists a random variable W 2 L0C .; F ; P / with jX i j W < 1 P -a.s.

for each X D .X 1 ; : : : ; X n / 2 B,

supX2B U.X / < 1,

Then there exists an X 2 B which maximizes U on B, and X is unique if U is

strictly concave. As a special case, this includes the utility functionals

Q2Q

profiles, where u is a utility function on Rn and Q is a set of probability measures

equivalent to P ; see Section 2.5.

}

We turn now to a characterization of the optimal profile X in terms of the inverse

of the derivative u0 of u in case where u is continuously differentiable on .0; 1/. Let

a WD lim u0 .x/ 0

x"1

and

x#0

We define

I C W .a; b/ ! .0; 1/

as the continuous, bijective, and strictly decreasing inverse function of u0 on .a; b/,

and we extend I C to the full half axis 0; 1 by setting

0

for y b,

C

I .y/ WD

(3.22)

C1 for y a.

With this convention, I C W 0; 1 ! 0; 1 is continuous.

Remark 3.38. If u is a utility function defined on all of R, the function I C is the

inverse of the restriction of u0 to 0; 1/. Thus, I C is simply the positive part of

the function I D .u0 /1 . For instance, in the case of an exponential utility function

u.x/ D 1 e x , we have a D 0, b D , and

1

y

D .I.y//C ; y 0:

(3.23)

log

I C .y/ D

}

Theorem 3.39. Assume that X 2 B is of the form

X D I C .c '/ ^ W

for some constant c > 0 such that E X D w. If E u.X / < 1 then X is the

unique maximizer of the expected utility E u.X / among all X 2 B.

145

v.y; !/ WD

sup

.u.x/ xy/

(3.24)

0xW .!/

defined for y 2 R and ! 2 . Clearly, for each ! with W .!/ < 1 the supremum

above is attained in a unique point x .y/ 2 0; W .!/, which satisfies

x .y/ D 0

x .y/ D W .!/

u0 .x/ < y

0

u .x/ > y

for all x 2 .0; W .!//.

follows that

x .y/ D I C .y/ ^ W .!/;

or

X D x .c '/ on W < 1.

(3.25)

If W .!/ D C1, then the supremum in (3.24) is not attained if and only if u0 .x/ >

y for all x 2 .0; 1/. By our convention (3.22), this holds if and only if y a and

hence I C .y/ D C1. But our assumptions on X imply that I C .c '/ < 1 P -a.s.

on W D 1, and hence that

X D x .c '/

P -a.s. on W D 1.

(3.26)

u.X / X c ' D v.c'; /

P -a.s.

u.X / c 'X u.X / c 'X

P -a.s.

E u.X / E u.X / C c E X X E u.X / :

Hence, X maximizes the expected utility on B. Uniqueness follows from Remark 3.33.

In the following examples, we study the application of the preceding theorem to

CARA and HARA utility functions. For simplicity we consider only the case W

1. The extension to a non-trivial bound W is straightforward.

146

(3.23)

y '

1

1 y '

D h

;

'I C .y '/ D ' log

where h.x/ D .x log x/ . Since h is bounded by e 1 , it follows that 'I C .y '/

belongs to L1 .P / for all y > 0. Thus,

1 h y ' i

g.y/ WD E I C .y '/ D E h

y

unique c with g.c/ D w. The corresponding profile

X WD I C .c '/

maximizes the expected utility E u.X / among all X 0. Let us now return to

the special situation of the financial market model of Section 3.1, and take P as the

entropy-minimizing risk-neutral measure of Corollary 3.25. Then the optimal profile

X takes the form

X D . Y K/C ;

where is the maximizer of the expected utility E u. Y / , and where K is given

by

1

c

1

c

1

1

K D log log E e Y D log C H.P jP /:

Note that X is a linear combination of the primary assets only in the case where

Y K P -almost surely. In general, X is a basket call option on the attainable

asset w C .1 C r/ Y 2 V with strike price w C .1 C r/K. Thus, a demand for

derivatives appears.

}

Example 3.41. If u is a HARA utility function of index 2 0; 1/ then u0 .x/ D

x 1 , hence

1

I C .y/ D y 1

and

I C .y '/ D y 1 ' 1 :

In the logarithmic case D 0, we assume that the relative entropy H.P jP / of P

with respect to P is finite. Then

X D

dP

w

Dw

'

dP

E log X D log w C H.P jP /:

147

If 2 .0; 1/ and

then the unique optimal profile is given by

and the maximal value of expected utility is equal to

E u.X / D

1

w .E ' 1 /1 :

Exercise 3.3.1. In the context of Example 3.41, compute the maximal value of expected utility when ' has a log-normal distribution.

}

The following corollary gives a simple condition on W which guarantees the existence of the maximizer X in Theorem 3.39.

Corollary 3.42. If E u.W / < 1 and if 0 < w < E W < 1, then there exists

a unique constant c > 0 such that

X D I C .c '/ ^ W

satisfies E X D w. In particular, X is the unique maximizer of the expected

utility E u.X / among all X 2 B.

Proof. For any 2 .0; 1/,

y 7! I C .y/ ^

is a continuous decreasing function with limy"b I C .y/ ^ D 0 and I C .y/ ^ D

for all y u0 ./. Hence, dominated convergence implies that the function

g.y/ WD E I C .y '/ ^ W ;

is continuous and decreasing with

lim g.y/ D 0 < w < E W D lim g.y/:

y"1

y#0

exists a unique c with g.c/ D w, and Theorem 3.39 yields the optimality of the

corresponding X .

Let us now extend the discussion to the case where preferences themselves are

uncertain. This additional uncertainty can be modelled by incorporating the choice of

a utility function into the description of possible scenarios; for an axiomatic discussion

see [172]. More precisely, we assume that preferences are described by a measurable

148

continuously differentiable on .0; 1/. For each ! 2 , the inverse of u0 .; !/ is

extended as above to a function

I C .; !/ W 0; 1 ! 0; 1:

Using exactly the same arguments as above, we obtain the following extension of

Corollary 3.42 to the case of random preferences:

Corollary 3.43. If E u.W; / < 1 and if 0 < w < E W < 1, then there exists

a unique constant c > 0 such that

X .!/ W D I C .c '.!/; !/ ^ W .!/

is the unique maximizer of the expected utility

Z

E u.X; / D u.X.!/; !/ P .d!/

among all X 2 B.

3.4

So far, we have discussed the structure of asset profiles which are optimal with respect

to a fixed utility function u. Let us now introduce an optimization problem with

respect to the uniform order <uni as discussed in Section 2.4. The partial order <uni can

be viewed as a reflexive and transitive relation on the space of financial positions

X WD L1C .; F ; P /

by letting

X <uni Y W

X <uni

Y

(3.27)

where

X and

Y denote the distributions of X and Y under P . Note that X <uni

Y <uni X if and only if X and Y have the same distribution; see Remark 2.58. Thus,

the relation <uni is antisymmetric on the level of distributions but not on the level of

financial positions.

Let us now fix a position X0 2 X such that E X0 < 1, and let us try to minimize the cost among all positions X 2 X which are uniformly at least as attractive

as X0

Minimize E X among all X <uni X0 .

149

In order to describe the minimal cost and the minimizing profile, let us denote by F'

and FX0 the distribution functions and by q' and qX0 quantile functions of ' and X0 ;

see Appendix A.3.

Theorem 3.44. For any X 2 X such that X <uni X0 ,

Z

E X

0

(3.28)

0; 1/ defined by

f .x/ WD qX0 .1 F' .x//

if x is a continuity point of F' , and by

f .x/ WD

1

F' .x/ F' .x/

F' .x/

F' .x/

qX0 .1 t / dt

otherwise.

The proof will use the following lemma, which yields another characterization of

the relation <uni .

Lemma 3.45. For two probability measures

and on R, the following conditions

are equivalent:

(a)

<uni .

(b) For all decreasing functions h W .0; 1/ ! 0; 1/,

Z

h.t /q .t / dt

0

h.t /q .t / dt;

(3.29)

and .

(c) The relation (3.29) holds for all bounded decreasing functions h W .0; 1/ !

0; 1/.

Proof. The relation

<uni is equivalent to

Z

q .t / dt

0

q .t / dt

see Theorem 2.57. The implication (c) ) (a) thus follows by taking h D I.0;t . For the

proof of (a) ) (b), we may assume without loss of generality that h is left-continuous.

150

Then there exists a positive Radon measure on .0; 1 such that h.t / D .t; 1/.

Fubinis theorem yields

Z

1Z y

h.t / q .t / dt D

0

q .t / dt .dy/

0

0

Z 1Z y

q .t / dt .dy/

0

0

1

h.t / q .t / dt:

D

0

Proof of Theorem 3:44. Using the first HardyLittlewood inequality in Theorem A.24,

we see that

Z

1

E X D E X'

q' .1 t / qX .t / dt;

0

where qX is a quantile function for X. Taking h.t / WD q' .1 t / and using Lemma 3.45 thus yields (3.28).

Let us now turn to the identification of the optimal profile. Note that the function

f defined in the assertion satisfies

f .q' / D E

g j q'

(3.30)

j q' denotes the conditional

expectation with respect to q' under the Lebesgue measure on .0; 1/. Let us show

that X D f .'/ satisfies X <uni X0 . Indeed, for any utility function u

Z

u.f .q' // dt

0

0

u.qX0 .1 t // dt D

u.qX0 .t // dt

D E u.X0 / ;

where we have applied Lemma A.19 and Jensens inequality for conditional expectations. Moreover, X attains the lower bound in (3.28)

E X D E f .'/ ' D

f .q' .t // q' .t / dt

0

D

0

due to (3.30).

qX0 .1 t / q' .t / dt D

Z

0

151

Remark 3.46. The solution X has the same expectation under P as X0 . Indeed,

(3.30) shows that

Z 1

Z 1

f .q' .t // dt D

qX0 .1 t / dt D E X0 :

}

E X D E f .'/ D

0

Remark 3.47. The lower bound in (3.28) may be viewed as a reservation price for

X0 in the following sense. Let X0 be a financial position, and let X be any class of

financial positions such that X 2 X is available at price .X /. For a given relation

on X [ X0 ,

R .X0 / WD inf.X / j X 2 X; X
X0

is called the reservation price of X0 with respect to X, , and
.

If X is the space of constants with .c/ D c, and if the relation
is of von

NeumannMorgenstern type with some utility function u, then R .X0 / reduces to

the certainty equivalent of X0 with respect to u; see (2.11).

In the context of the optimization problem (3.20), where

X
X0 W E u.X / E u.X0 / ;

the reservation price is given by E X , where X is the utility maximizer in the

budget set defined by w WD E X0 .

In the context of the financial market model of Chapter 1, we can take X as the

space V of attainable claims with

V
X0 W V X0 P -a.s.

and .V / D for V D S . In this case, the reservation price R .X0 / coincides

with the upper bound sup .X0 / of the arbitrage-free prices for X0 ; see Theorem 1.32.

}

3.5

In this section we consider the optimal investment problem for an economic agent

who, as in Section 2.5, is averse against both risk and Knightian uncertainty. Under the

assumptions of Theorem 2.78 (b), the preferences of such an agent can be described

by the following robust utility functional

inf EQ u.X / ;

Q2Q

(3.31)

assume that such a robust utility functional is given and that all measures in Q are

152

also assume that all payoff profiles X are nonnegative and, for simplicity, that the

utility function u is defined and finite on 0; 1/. As in Section 3.3, the budget set for

a given initial capital w > 0 is defined as

B WD X 2 L1C .P / j E X w:

The problem of robust utility maximization can thus be stated as follows,

maximize inf EQ u.X / over all X 2 B.

Q2Q

(3.32)

In the sequel, we will assume that Q is a convex set, which can be done without loss

of generality. Moreover, we will assume throughout this section that Q is equivalent

to P in the following sense:

P A D 0

Q A D 0 for all Q 2 Q.

(3.33)

Clearly, our problem (3.32) would not be well-posed without the implication ).

Note that (3.33) implies that every measure Q 2 Q is absolutely continuous with

respect to P but not necessarily equivalent to P .

When Q consists of the single element Q P , we have seen in Section 3.3 that

the solution involves the RadonNikodym derivative dP =dQ. In our present situation, however, not every Q 2 Q needs to be equivalent to P . The RadonNikodym

derivative dP =dQ will therefore be understood in the sense of the Lebesgue decomposition; see Theorem A.13. With the convention x0 WD C1 for x > 0 it can be

expressed as

dQ 1

dP

P -a.s.

D

dQ

dP

It will turn out in Theorem 3.56 below that the following concept can be a key for

solving problem (3.32).

Definition 3.48. A measure Q0 2 Q is called a least favorable measure with respect

to P if the density D d P =d Q0 satisfies

Q0 t D inf Q t

Q2Q

Not every set Q is such that it admits a least-favorable measure with respect to a

given measure P . But here are two examples in which least-favorable measures can

be determined explicitly.

Example 3.49. Let Y be a measurable function on .; F /, and denote by

its law

under P . For

given, let

Q WD Q

P j Q Y 1 D :

153

The interpretation behind the set Q is that an investor has full knowledge about the

pricing measure P but is uncertain about the true distribution P of market prices and

has only the weak information that a certain functional Y of the stock price has the

distribution under P . Define Q0 as follows via its RadonNikodym derivative:

dQ0

d

D

.Y /:

dP

d

=d.Y / is the same for all

}

Q 2 Q. Hence, Q0 is a least favorable measure.

Example 3.50. For 2 .0; 1/ and a given probability measure P P let

dQ

1

:

Q

WD Q

P

dP

In Chapter 4, this set will play an important role as the maximal representing set for the

risk measure AV@R

. When ' WD dP =dP satisfies P ' >
1 > 0 and admits

a continuous and strictly increasing distribution function F' .x/ WD P ' x , then

Q

admits a least-favorable measure with respect to P , which is given by

dQ0

1

'

D

;

dP

' _ q' .t

/

where q' D F'1 is the (unique) quantile function of ' and t

is the unique solution

of the equation

Z

t

q' .t

/.t

1 C
/ D

q' .t / dt:

0

This will be proved in Corollary 8.28. In this case, one finds that

D

dP

D
.' _ q' .t

//:

dQ0

Remark 3.51. The existence of a least-favorable measure of a set Q with respect to

P is closely related to the concept of submodularity of the set function

c.A/ WD sup Q A ;

A2F;

Q2Q

which will be discussed in Sections 4.6 and 4.7. The HuberStrassen theorem states

that a set Q that is weakly compact with respect to a Polish topology on admits a

least favorable measure with respect to any other probability measure P Q if and

only if the set function c is submodular

c.A [ B/ C c.A \ B/ c.A/ C c.B/;

A; B 2 F :

154

We refer to Huber and Strassen [154] and Lembcke [195]. Example 3.50 is a special

case of this situation since it will follow from Proposition 4.75 that the associated set

function is submodular.

}

Let us now show that we always have Q0 P if Q satisfies a mild closure

condition.

Lemma 3.52. Suppose that dQ=d P j Q 2 Q is closed in L1 .P /. Then every

least favorable measure Q0 is equivalent to P .

Proof. Due to our assumption (3.33) and the closedness of dQ=d P j Q 2 Q, we

may apply the HalmosSavage theorem in the form of Theorem 1.61 and obtain a

measure Q1 P in Q. We get

1 D Q0 < 1 D lim Q0 t D lim inf Q t Q1 < 1 :

t"1 Q2Q

t"1

Q0 .

We have the following characterization of least favorable measures.

Proposition 3.53. For a measure Q0 2 Q with Q0 P and WD d P =dQ0 , the

following conditions are equivalent:

(a) Q0 is a least favorable measure for P .

(b) If f W .0; 1 ! R is decreasing and infQ2Q EQ f ./ ^ 0 > 1, then

inf EQ f ./ D EQ0 f ./ :

Q2Q

sup EQ g./ D EQ0 g./ :

Q2Q

Ig .QjP / WD

Z

g

dQ

dP

dP

such that Ig .QjP / is finite for some Q 2 Q.

Proof. (a) , (b): According to the definition, Q0 is a least favorable measure if and

only if Q0 t Q t for all t 0 and each Q 2 Q. By Theorem 2.68, this

is equivalent to the fact that, for all Q 2 Q, Q0 1 stochastically dominates Q 1

in the sense of Definition 2.67 . Hence, if f is bounded, then the equivalence of (a) and

155

(b) follows from Theorem 2.68. If f is unbounded and infQ2Q EQ f ./^0 > 1,

then assertion (b) holds for fN WD .N / _ f ^ 0 where N 2 N. Thus, for all Q 2 Q,

EQ fN ./ EQ0 fN ./ EQ0 f ./ ^ 0 > 1 :

By sending N to infinity, it follows that EQ f ./ ^ 0 EQ0 f ./ ^ 0 for every

Q 2 Q. After using a similar argument for 0 _ f ./, we get

EQ f ./ D EQ f ./ _ 0 C EQ f ./ ^ 0 EQ0 f ./

for all Q 2 Q.

(b) ) (d): Clearly, Ig .QjP / is well-defined and larger than g.1/ for each Q

P due to Jensens inequality. Now take Q1 2 Q with Ig .Q1 jP / < 1, and denote

0 .x/ the right-hand derivative of g at x 0. Suppose first that g 0 is bounded.

by gC

C

0 .x/.y x/, we have

Since g.y/ g.x/ gC

Ig .Q1 jP / Ig .Q0 jP /

Z

Z

dQ

0

gC

. 1 /

1

dP

f ./ dQ1

dQ0

dP

dP

f ./ dQ0 ;

R

0

where

f .x/ WD gC

.1=x/ is a bounded decreasing function. Therefore f ./ dQ1

R

f ./ dQ0 , and Q0 minimizes Ig . jP / on Q.

0 , let

denote

To prove the assertion for convex functions g with unbounded gC

Q

dQ

under

P

.

Then

the

preceding

step

of

the

proof

implies

in

particular

the law

of

dP

R

R

that .x c/C

RQ .dx/ .x c/C

Q0 .dx/ for all c 2 R and each Q 2 Q.

Since in addition x

Q .dx/ D 1 for all Q 2 Q, the result now follows from the

equivalence (b) , (c) of Corollary 2.61.

(d) ) (b): It is enough to prove (b)Rfor continuous bounded decreasing functions

x

f . For such a function f let g.x/ WD 1 f .1=t / dt. Then g is convex. For Q1 2 Q

we let Q t WD tQ1 C .1 Rt /Q0 and h.t / RWD Ig .Q t jP /. The right-hand derivative of

h satisfies 0 h0C .0/ D f ./ dQ1 f ./ dQ0 , and the proof is complete.

Remark 3.54. Let us discuss the connection between least-favorable measures and

the statistical test theory for composite hypotheses, extending the standard Neyman

Pearson theory as outlined in Appendix A.4. In a composite hypothesis testing problem, one tests a hypothesis P against a null hypothesis Q and allows both P and Q

to be sets of probability measures. Our situation here corresponds to the special case

in which P consists of the single element P and only the null hypothesis Q is composite. As in Remark A.32 one looks for a randomized statistical test 0 W ! 0; 1

that maximizes the power E among all randomized tests with a given significance level. For a composite null hypothesis Q, the significance of a randomized test

156

maximize E

over all

2 R with sup EQ

;

(3.34)

Q2Q

where 2 .0; 1/ is a given significance level and R denotes the class of all randomized tests (see Appendix A.4). This problem can be solved in a straightforward

manner as soon as Q admits a least-favorable measure Q0 with respect to P . Indeed,

choose 2 0; 1 and c > 0 such that

0

satisfies EQ0

yields that

0

D . Since

WD I

Dc C I

>c

0

sup EQ

0

D EQ0

D :

Q2Q

particular EQ0 , and so Theorem A.31 implies E E 0 , because

0 is an optimal randomized test for the standard problem of testing the hypothesis

P against the null hypothesis Q0 . Therefore 0 is a solution of (3.34). Note also

that the likelihood of a type 1 error for the test 0 is maximized by Q0 . The fact that

this is the case for every significance level is remarkable, and it explains why Q0 is

called a least-favorable measure.

}

The next proposition prepares for the solution of the robust utility maximization

problem (3.32).

Proposition 3.55. Let Q0 P be a least favorable measure and D d P =dQ0 .

(a) For any X 2 B there exists XQ 2 B such that

inf EQ u.XQ / inf EQ u.X /

Q2Q

Q2Q

(b) Every solution X of (3.32) is of the form X D f ./ for a deterministic

decreasing function f W .0; 1/ ! 0; 1/.

Proof. (a) We need to construct a decreasing function f 0 such that E f ./

w and

inf EQ u.f .// inf EQ u.X / :

(3.35)

Q2Q

Q2Q

qY .t / a quantile function of a random variable Y with respect to the probability measure Q0 .

157

if F is continuous at t ,

qX .1 F .t //

R F

.t/

f .t / WD

1

q .1 s/ ds otherwise.

F

.t/F

.t/ F

.t/ X

(3.36)

h j q , where is the Lebesgue

measure and h.t / WD qX .1 t /; see Exercise 3.4.1. Hence, Jensens inequality for

conditional expectations and Lemma A.23 show that

Z t

Z 1

inf EQ u.X / EQ0 u.X / D

u.qX .t // dt D

u.h.t // dt

Q2Q

0

0

Z 1

Z 1

u.E

h j q .t // dt D

u.qf ./ .1 t // dt

(3.37)

0

Q2Q

where we have used Proposition 3.53 in the last step. Thus, f satisfies (3.35).

It remains to show that f ./ satisfies the capital constraint. To this end, we first

use the lower HardyLittlewood inequality in Theorem A.24

Z 1

q .t / qX .1 t / dt :

(3.38)

w E X D EQ0 X

0

h j q .t / D f .q .t //. We then get

Z 1

Z 1

q .t / qX .1 t / dt D

q .t / f .q .t // dt

(3.39)

0

0

D EQ0 f ./ D E f ./ :

Thus, f is as desired.

(b) Now suppose X solves (3.32). If X is not Q0 -a.s. ./-measurable, then

Y WD EQ0 X j must satisfy

EQ0 u.Y / > EQ0 u.X / ;

(3.40)

the proof of part (a) yields that

E fQ./ D EQ0 fQ./ EQ0 Y D EQ0 X w;

and by (3.37) and (3.40),

inf EQ u.fQ.// EQ0 u.Y / > EQ0 u.X / inf EQ u.X / ;

Q2Q

Q2Q

and can hence be written as a (not yet necessarily decreasing) function of .

158

of yet another solution in B. Clearly, we must have

R 1E X D w D E f ./ .

Thus, (3.38) and (3.39) yield that EQ0 X D 0 q .t / qX .1 t / dt . But then

the only if part of the lower HardyLittlewood inequality together with the ./measurability of X imply that X is a decreasing function of .

We can now state and prove the main result of this section.

Theorem 3.56. Suppose that Q admits a least favorable measure Q0 P . Then

the robust utility maximization problem (3.32) is equivalent to the standard utility

maximization problem with respect to Q0 , i.e., to (3.32) with Q replaced by Q0 .

More precisely, X 2 B solves the robust problem (3.32) if and only if it solves the

standard problem for Q0 , and the values of the corresponding optimization problems

are equal, whether there exists a solution or not:

sup inf EQ u.X / D sup EQ0 u.X / ;

X2B Q2Q

X2B

Proof. Proposition 3.55 implies that in solving the robust utility maximization problem (3.32) we may restrict ourselves to strategies whose terminal wealth is a decreasing function of . By Propositions 3.53, the robust utility of a such a terminal

wealth is the same as the expected utility with respect to Q0 . On the other hand,

taking Q0 WD Q0 in Proposition 3.55 implies that the standard utility maximization

problem for Q0 also requires only strategies whose terminal wealth is a decreasing

function of . Therefore, the two problems are equivalent, and Theorem 3.56 is

proved.

Theorem 3.56 states in particular that the robust utility maximization problem

(3.32) has a solution if and only if the standard problem for Q0 has a solution, and we

refer to Theorem 3.39 (with W D C1) for a sufficient condition.

Example 3.57. Consider the situation of Example 3.50, where for 2 .0; 1/ a leastfavorable measure of Q

with respect to P is given by

1

'

dQ0

D

:

dP

' _ q' .t

/

Here ' D dP =dP satisfies P ' > 1 > 0 and admits a continuous and strictly

increasing quantile function q' with respect to P , and t

is the solution of a certain

nonlinear equation. Let us also assume for the sake of concreteness that u.x/ D 1 x

is a HARA utility function with risk aversion 2 .0; 1/ and that E ' =.1 / < 1.

159

It was shown in Example 3.41 that the standard utility maximization problem under

P has the solution

1

X D w.E ' 1 /1 ' 1 :

Note that X is large for small values of ', that is, for low-price scenarios.

Now let us replace the single probability measure P by the entire set Q

. By

Theorem 3.56, the corresponding robust utility maximization problem will be solved

by

1

XQ D w.E 1 /1 1 :

where

D

dP

D
.' _ q' .t

//:

dQ0

It follows that

XQ D c.X ^ r/

where r > 0 and c > 1 are certain constants. Thus, the effect of robustness is here

that the optimal payoff profile X is cut off at a certain threshold. That is, one gives

up the opportunity for very high profits in low-price scenarios in favor of enhanced

returns in all other scenarios.

}

3.6

Microeconomic equilibrium

The aim of this section is to provide a brief introduction to the theory of market equilibrium. Prices of assets will no longer be given in advance. Instead, they will be

derived from first principles in a microeconomic setting where different agents demand asset profiles in accordance with their preferences and with their budget constraints. These budget constraints are determined by a given price system. The role of

equilibrium prices consists in adjusting the constraints in such a way that the resulting

overall demand is matched by the overall supply of assets.

Consider a finite set A of economic agents and a convex set X L0 .; F ; P / of

admissible claims. At time t D 0, each agent a 2 A has an initial endowment whose

discounted payoff at time t D 1 is described by an admissible claim

Wa 2 X;

The aggregated claim

W WD

a 2 A:

X

Wa

a2A

is also called the market portfolio. Agents may want to exchange their initial endowment Wa against some other admissible claim Xa 2 X. This could lead to a new

allocation .Xa /a2A if the resulting total demand matches the overall supply.

160

Definition 3.58. A collection .Xa /a2A X is called a feasible allocation if it satisfies the market clearing condition

X

Xa D W P -a.s.

(3.41)

a2A

The budget constraints will be determined by a linear pricing rule of the form

.X / WD E ' X ;

X 2 X;

where ' is a price density, i.e., an integrable function on .; F / such that ' > 0 P a.s. and E jWa j ' < 1 for all a 2 A. To any such ' we can associate a normalized

price measure P ' P with density 'E ' 1 .

Remark 3.59. In the context of our one-period model of a financial market with

d risky assets S 1 ; : : : ; S d and a risk-free asset S 0 1 C r, P ' is a risk-neutral

measure if the pricing rule is consistent with the given price vector D . 0 ; /,

where 0 D 1. In this section, the pricing rule will be derived as an equilibrium

price measure, given the agents preferences and endowments. In particular, this will

amount to an endogenous derivation of the price vector . In a situation where the

structure of the equilibrium is already partially known in the sense that it is consistent

with the given price vector , the construction of a microeconomic equilibrium yields

a specific choice of a martingale measure P , i.e., of a specific extension of from

the space V of attainable payoffs to a larger space of admissible claims.

}

The preferences of agent a 2 A are described by a utility function ua . Given the

price density ', an agent a 2 A may want to exchange the endowment Wa for an

'

admissible claim Xa which maximizes the expected utility

E ua .X /

among all X in the agents budget set

Ba .'/ WD X 2 X j E 'jX j < 1 and E ' X E ' Wa

D X 2 X j X 2 L1 .P ' / and E ' X E ' Wa :

'

In this case, we will say that Xa solves the utility maximization problem of agent

a 2 A with respect to the price density '. The key problem is whether ' can be

'

chosen in such a way that the requested profiles Xa , a 2 A, form a feasible allocation.

Definition 3.60. A price density ' together with a feasible allocation .Xa /a2A is

called an ArrowDebreu equilibrium if each Xa solves the utility maximization problem of agent a 2 A with respect to ' .

161

Thus, the price density ' appearing in an ArrowDebreu equilibrium decentralizes the crucial problem of implementing the global feasibility constraint (3.41). This

is achieved by adjusting the budget sets in such a way that the resulting demands respect the market clearing condition, even though the individual demand is determined

without any regard to this global constraint.

Example 3.61. Assume that each agent a 2 A has an exponential utility function

with parameter a > 0, and let us consider the unconstrained case

X D L0 .; F ; P /:

In this case, there is a unique equilibrium, and it is easy to describe it explicitly. For

a given pricing measure P P such that Wa 2 L1 .P / for all a 2 A, the utility

maximization problem for agent a 2 A can be solved if and only if H.P jP / < 1,

and in this case the optimal demand is given by

Xa D

1

1

log ' C wa C

H.P jP /

a

a

where

wa WD E Wa I

see Example 3.35. The market clearing condition (3.41) takes the form

W D

X

1

1

log ' C

wa C H.P jP /

a2A

X 1

1

:

D

(3.42)

a2A

' D

e W

;

E e W

(3.43)

E jWa je W < 1;

a 2 AI

this condition is satisfied if, e.g., the random variables Wa are bounded from below.

Define P P via (3.43). Then

H.P jP / D E W log E e W < 1;

162

and the optimal profile for agent a 2 A with respect to the pricing measure P takes

the form

.W E W /:

(3.44)

Xa D wa C

a

Since

X

wa D E W ;

a2A

.Xa /a2A

the allocation

is feasible, and so we have constructed an ArrowDebreu

equilibrium. Thus, the agents share the market portfolio in a linear way, and in inverse

proportion to their risk aversion.

Let us now return to our financial market model of Section 3.1. We assume that

the initial endowment of agent a 2 A is given by a portfolio a 2 Rd C1 so that the

discounted payoff at time t D 1 is

Wa D

a S

;

1Cr

a 2 A:

.0 ; /. The optimal claim for agent a 2 A in (3.44) takes the form

S

;

Xa D a C

a

1Cr

a D

i

DE

Si

1Cr

for i D 1; : : : ; d .

Thus, we could have formulated the equilibrium problem within the smaller space

X D V of attainable payoffs, and the resulting equilibrium allocation would have been

the same. In particular, the extension of X from V to the general space L0 .; F ; P /

of admissible claims does not create a demand for derivatives in our present example.

}

From now on we assume that the set of admissible claims is given by

X D L0C .; F ; P /;

and that the preferences of agent a 2 A are described by a utility function ua W

0; 1/ ! R which is continuously differentiable on .0; 1/. In particular, the initial

endowments Wa are assumed to be non-negative. Moreover, we assume

P Wa > 0 0 for all a 2 A

and

E W < 1:

(3.45)

163

A function ' 2 L1 .; F ; P / such that ' > 0 P -a.s. is a price density if

E ' W < 1I

note that this condition is satisfied as soon as ' is bounded, due to our assumption

(3.45). Given a price density ', each agent faces exactly the optimization problem

discussed in Section 3.3 in terms of the price measure P ' P . Thus, if .Xa /a2A

is an equilibrium allocation with respect to the price density ' , feasibility implies

0 Xa W , and so it follows as in the proof of Corollary 3.42 that

Xa D IaC .ca ' /;

a 2 A;

(3.46)

with positive constants ca > 0. Note that the market clearing condition

X

X

W D

Xa D

IaC .ca ' /

a2A

a2A

will determine ' as a decreasing function of W , and thus the optimal profiles Xa

will be increasing functions of W .

Before we discuss the existence of an ArrowDebreu equilibrium, let us first illustrate the structure of such equilibria by the following simple examples. In particular,

they show that an equilibrium allocation will typically involve non-linear derivatives

of the market portfolio W .

Example 3.62. Let us consider the constrained version of the preceding example

where agents a 2 A have exponential utility functions with parameters a > 0.

Define

w WD supc j W c P -a.s. 0;

and let P be the measure defined via (3.43). For any agent a 2 A such that

wa WD E Wa

.E W w/;

a

(3.47)

Xa D wa C

.W E W /

a

satisfies Xa 0 P -a.s. Thus, if all agents satisfy the requirement (3.47) then the

unrestricted equilibrium computed in Example 3.61 is a forteriori an ArrowDebreu

equilibrium in our present context. In this case, there is no need for non-linear derivatives of the market portfolio.

If some agents do not satisfy the requirement (3.47) then the situation becomes

more involved, and the equilibrium allocation will need derivatives such as call options. Let us illustrate this effect in the simple setting where there are only two agents

164

not. For c 0, we define the measure P c P in terms of the density

1 1 W

e

on W c;

c

' WD Z11 W

on W c;

Z2 e

where is given by (3.42), and where the constants Z1 and Z2 are determined by the

continuity condition

log Z2 log Z1 D c.1 /

and by the normalization E ' c D 1. Note that P 0 D P with P as in (3.43).

Consider the equation

c

E .W c/C D w2c WD E c W2 :

2

(3.48)

Both sides are continuous in c. As c increases from 0 to C1, the left-hand side

decreases from 2 E W to 0, while w2c goes from w20 < 2 E W to E 1 W2 >

0. Thus, there exists a solution c of (3.48). Let us now check that

X2c WD

.W c/C ;

2

X1c WD W X2c

defines an equilibrium allocation with respect to the pricing measure P c . Clearly, X1c

and X2c are non-negative and satisfy X1c CX2c D W . The budget condition for agent 2

is satisfied due to (3.48), and this implies the budget condition

E c X1c D E c W w2c D w1c

for agent 1. Both are optimal since

Xac D IaC .
a ' c /

with

1 WD 1 Z1

and

2 WD 2 Z2 e c :

Thus, agent 2 demands 2 shares of a call option on the market portfolio W with

strike c, agent 1 demands the remaining part of W , and so the market is cleared.

In the general case of a finite set A of agents, the equilibrium price measure PO has

the following structure. There are levels 0 WD c0 < < cN D 1 with 1 N jAj

such that the price density 'O is given by

'O D

1 i W

e

Zi

for i D 1; : : : ; N , where

i WD

on W 2 ci 1 ; ci

X 1 1

;

a

2Ai

165

and where Ai .i D 1; : : : ; N / are the increasing sets of agents which are active at

the i th layer in the sense that Xa > 0 on W 2 .ci 1 ; ci . At each layer .ci 1 ; ci ,

the active agents are sharing the market portfolio in inverse proportions to their risk

aversion. Thus, the optimal profile XO a of any agent a 2 A is given by an increasing

piecewise linear function in W , and thus it can be implemented by a linear combination of call options with strikes ci . More precisely, an agent a 2 Ai takes i =a

shares of the spread

.W ci 1 /C .W ci /C ;

i.e., the agent goes long on a call option with strike ci 1 and short on a call option

}

with strike ci .

Example 3.63. Assume that all agents a 2 A have preferences described by HARA

utility functions so that

1

IaC .y/ D y 1 a ; a 2 A

with 0 a < 1. For a given price density ', the optimal claims take the form

1

(3.49)

with constants ba > 0. If a D for all a 2 A, then the market clearing condition

(3.41) implies

X

X

1

Xa D

ba ' 1 ;

W D

a2A

a2A

'

' D

1 1

;

W

Z

where Z is the normalizing constant, and so the agents demand linear shares of the

market portfolio W . If risk aversion varies among the agents then the structure of the

equilibrium becomes more complex, and it will involve non-linear derivatives of the

market portfolio. Let us number the agents so that A D 1; : : : ; n and 1 n .

Condition (3.49) implies

Xi D di Xni

with some constants di , and where

i WD

1 n

1 i

a convex increasing function of Xn . In equilibrium, Xn is a concave function of W

determined by the condition

n

X

di Xni D W;

(3.50)

i D1

166

' D

1 n 1

:

X

Z n

p

A D 1; 2 with u1 .x/ D x and u2 .x/ D log x, i.e., 1 D 12 and 2 D 0; see

Example 2.38. The solutions of (3.50) can be parameterized with c 0 such that

p

p

p

X2c D 2 c W C c c 2 0; W

and

X1c D W X2c :

The corresponding price density takes the form

'c D

1

1

p

p ;

Z.c/ W C c c

where Z.c/ is the normalizing constant. Now assume that W 1 2 L1 .P /, and let

P 1 denote the measure with density W 1 .E W 1 /1 . As c increases from 0 to

1, E c X2c increases continuously from 0 to E 1 W , while E c W2 goes continuously from E 0 W2 > 0 to E 1 W2 < E 1 W ; here we use our assumption that

P Wa > 0 0 for all a 2 A. Thus, there is a c 2 .0; 1/ such that

E c X2c D E c W2 ;

and this implies that the budget constraint is satisfied for both agents. With this choice

of the parameter c, .X1c ; X2c / is an equilibrium allocation with respect to the pricing

measure P c : Agent 2 demands the concave profile X2c , agent 1 demands the convex

profile X1c , both in accordance with their budget constraints, and the market is cleared.

}

Let us now return to our general setting, and let us prove the existence of an Arrow

Debreu equilibrium. Consider the following condition:

W

0

0

lim sup x ua .x/ < 1 and E ua

< 1; a 2 A:

(3.51)

jAj

x#0

Remark 3.64. Condition (3.51) is clearly satisfied if

u0a .0/ WD lim u0a .x/ < 1;

x#0

a 2 A:

(3.52)

assume

E W a 1 < 1; a 2 A;

in addition to our assumption E W < 1.

167

Theorem 3.65. Under assumptions (3.45) and (3.51), there exists an ArrowDebreu

equilibrium.

In a first step, we are going to show that an equilibrium allocation maximizes a

suitable weighted average

U

.X / WD

a E ua .Xa /

a2A

of the individual utility functionals over all feasible allocations X D .Xa /a2A . The

weights are non-negative, and without loss of generality we can assume that they are

normalized so that the vector
WD .
a /a2A belongs to the convex compact set

X

a D 1 :

D
2 0; 1A

a2A

In a second step, we will use a fixed-point argument to obtain a weight vector and a

corresponding price density such that the maximizing allocation satisfies the individual budget constraints.

Definition 3.66. A feasible allocation .Xa /a2A is called
-efficient for
2 if it

maximizes U

over all feasible allocations.

In view of (3.46), part (b) of the following lemma shows that the equilibrium

allocation .Xa /a2A in an ArrowDebreu

P 1 equilibrium is
-efficient for the vector

1 /

1

D .c ca a2A , where c WD a ca . Thus, the existence proof for an Arrow

Debreu equilibrium is reduced to the construction of a suitable vector
2 .

Lemma 3.67. (a) For any
2 there exists a unique
-efficient allocation .Xa

/a2A .

(b) A feasible allocation .Xa /a2A is
-efficient if and only if it satisfies the first

order conditions

a u0a .Xa / ';

(3.53)

with respect to some price density '. In this case, .Xa /a2A coincides with

.Xa

/a2A , and the price density can be chosen as

'

WD max
a u0a .Xa

/:

a2A

maximizes E ua .X / over all X 2 X such that

E '

X E '

Xa

:

(3.54)

168

Proof. (a): Existence and uniqueness follow from the general argument in Remark

3.37 applied to the set B of all feasible allocations and to the functional U

. Note

that

U

.X / max E ua .W /

a2A

for any feasible allocation, and that the right-hand side is finite due to our assumption

(3.45). Moreover, by dominated convergence, U

is indeed continuous on B with

respect to P -a.s. convergence.

(b): Let us first show sufficiency. If X D .Xa /a2A is a feasible allocation satisfying

the first order conditions, and Y D .Ya /a2A is another feasible allocation then

X

U

.X / U

.Y / D

a E ua .Xa / ua .Ya /

a2A

a2A

h X

X i

Xa

Ya

D 0;

E '

a2A

a2A

using concavity of ua in the second step and the first order conditions in the third.

This shows that X is
-efficient.

Turning to necessity, consider the
-efficient allocation .Xa

/a2A for
2 and

another feasible allocation .Xa /a2A . For " 2 .0; 1, let Ya" WD "Xa C .1 "/Xa

.

Since .Ya" /a2A is feasible,
-efficiency of .Xa

/a2A yields

0

1X

a E ua .Ya" / ua .Xa

/

"

a2A

1X

a E u0a .Ya" /.Ya" Xa

/

"

a2A

X

a E u0a .Ya" /.Xa Xa

/ :

D

(3.55)

a2A

Let us first assume (3.52); in part (d) of the proof we show how to modify the

argument under condition (3.51). Using dominated convergence and (3.52), we may

let " # 0 in the above inequality to conclude

X

a2A

E 'a

Xa

E 'a

Xa

E '

W ;

a2A

where

'a

WD
a u0a .Xa

/:

(3.56)

169

is a price density since by (3.52)

0 < '

max a u0a .0/ j a 2 A < 1:

Take a feasible allocation .Xa /a2A such that

X

'a

Xa D '

W I

(3.57)

a2A

T .!/ WD mina j 'a

.!/ D '

.!/:

In view of (3.56), we see that

X

E 'a

Xa

D E '

W :

(3.58)

a2A

D '

on Xa

> 0, which is equivalent to the first order condition

(3.53) with respect to '

.

(c): In order to show optimality of Xa

, we may assume without loss of generality

that P Xa

> 0 > 0, and hence a > 0. Thus, the first order condition with respect

to '

takes the form

Xa

D IaC .
1

a ' /;

due to our convention (3.22). By Corollary 3.42, Xa

solves the optimization problem

for agent a 2 A under the constraint

E '

X E '

Xa

:

(d): If (3.52) is replaced by (3.51), then we first need an additional argument in

order to pass from (3.55) to (3.56). Note first that by Fatous lemma,

X

X

a E u0a .Ya" / Xa

a lim inf E u0a .Ya" / Xa

lim inf

"#0

a2A

a2A

"#0

a E u0a .Xa

/ Xa :

a2A

WD max sup x u0a .x/ < 1

a2A 0<x1

by (3.51), we have xu0a .x/ C xu0a .1/ .1 C x/ for all x 0. This implies

u0a .Xa /Xa V WD .1 C W / 2 L1 .P /;

(3.59)

170

and also

u0a .Ya" / Xa

u0a ..1 "/ Xa

/Xa

.1 "/1 V;

since Ya" .1 "/Xa

. Thus, dominated convergence implies

E u0a .Ya" / Xa

! E u0a .Xa

/ Xa

;

" # 0;

By (3.59), we have

'a

Xa

WD a u0a .Xa

/ Xa

2 L1 .P /:

Hence E '

W < 1 follows by taking in (3.56) a feasible allocation .Xa /a2A

which is as in (3.57). We furthermore get (3.58), which yields as in part (b) the first

order conditions (3.53).

It remains to show that '

is integrable in order to conclude that '

is a price

density. Our assumption (3.51) implies

W

2 L1 .P /;

(3.60)

F WD max u0a

jAj

a2A

and so it is enough to show that F '

. Since Xa

D IaC .'

= a /, feasibility and

a 1 imply

X

W

IaC .'

/ jAj max IaC .'

/;

a2A

a2A

hence

/

a2A

b2A

//

D '

on

/ D IaC0 .'

/ :

a2A

ArrowDebreu equilibrium. Note that for each 2 the -efficient allocation

.Xa

/a2A and the price density '

would form an ArrowDebreu equilibrium if

E '

Wa D E '

Xa

for all a 2 A.

(3.61)

If this is not the case, then we can replace
by the vector g.
/ D .ga .
//a2A defined

by

1

E '

.Wa Xa

/ ;

ga .
/ WD
a C

E V

where V is given by (3.59). Note that g.
/ 2 : Since the first order conditions

(3.53) together with (3.59) imply

E '

Xa

D
a E u0a .Xa

/ Xa

a E V ;

171

P

we have ga .
/ 0, and a ga .
/ D 1 follows by feasibility. Thus, we increase

the weights of agents which were allocated less than they could afford. Clearly, any

fixed point of the map g W ! will satisfy condition (3.61) and thus yield an

ArrowDebreu equilibrium.

Proof of Theorem 3:65. (a): The set is convex and compact. Thus, the existence of

a fixed point of the map g W ! follows from Brouwers fixed point theorem as

soon as we can verify that g is continuous; see, for instance, Corollary 16.52 in [3]

for a proof of Brouwers fixed point theorem. Suppose that the sequence .
n /

converges to
2 . In part (c) we show that Xn WD X

n and 'n WD '

n converge P a.s. to X

and '

, respectively. We will show next that we may apply the dominated

convergence theorem, so that

lim E 'n Wa D E '

Wa

n"1

and

lim E 'n Xn D E '

X

n"1

and this will prove the continuity of g. To verify the assumptions of the dominated

convergence theorem, note that

Wa 'n W 'n W F;

where F is as in (3.60). Moreover,

W F jAjF IW jAj C max u0a .1/ W 2 L1 .P /:

a2A

(b): By our convention (3.22), the map f W 0; 1 ! 0; 1 defined by

X

f . ; y/ D

IaC . 1

a y/

a2A

strictly decreasing on .a. /; b. // where

a. / WD max lim a u0a .x/ 0

a2A x"1

and

a2A

each w 2 .0; 1/ there exists exactly one solution y

2 .a. /; b. // of the equation

f . ; y

/ D w:

Recall that 0; 1 can be regarded as a compact topological space. To see that y

172

nk of f . nk ; y/ D w converge to some limit y1 2

a. /; b. /. By continuity of f ,

f . ; y1 / D lim f . nk ; yk / D w;

k"1

.

(c): Recall that

Xa

D IaC .
1

a ' /

(3.62)

W D

Xa

D f .
; '

/:

a2A

Thus, '

n converges P -a.s. to '

as
n !
due to part (b), and so X

n converges P a.s. to X

due to (3.62). This completes the proof in (a) that the map g is continuous.

Remark 3.68. In order to simplify the exposition, we have restricted the discussion

of equilibrium prices to contingent claims with payoff at time t D 1. We have argued

in terms of discounted payoffs, and so we have implicitly assumed that the interest

rate r has already been fixed. From an economic point of view, also the interest rate

should be determined by an equilibrium argument. This requires an intertemporal

extension of our setting, which distinguishes between deterministic payoffs y at time

t D 0 and nominal contingent payoffs Y at time t D 1. Thus, we replace X D L0C

by the space

Y WD Y D .y; Y / j y 2 0; 1/; Y 2 L0C :

A pricing rule is given by a linear functional on Y of the form

.Y / WD '0 y C E ' Y ;

where '0 2 .0; 1/ and ' is a price density as before. Any such price system specifies

an interest rate for transferring income from time t D 0 to time t D 1. Indeed,

comparing the forward price c E ' for the fixed amount c to be delivered at time

1 with the spot price c '0 for the amount c made available at time 0, we see that the

implicit interest rate is given by

1Cr D

'0

:

E '

Ua .Y / D ua;0 .y/ C Eua;1 .Y /

173

with smooth utility functions ua;0 and ua;1 , then we can show along the lines of the

preceding discussion that an ArrowDebreu equilibrium exists in this extended set

ting. Thus, we obtain an equilibrium allocation .Y a /a2A and an equilibrium price

system D .'0 ; ' / such that each Y a maximizes the functional Ua in the agents

budget set determined by an initial endowment in Y and by the pricing rule . In particular, we have then specified an equilibrium interest rate r . Normalizing the price

system to '0 D 1 and defining P as a probability measure with density ' =E ' ,

we see that the price at time t D 0 of a contingent claim with nominal payoff Y 0

at time t D 1 is given as the expectation

Y

E

1 C r

of the discounted claim with respect to the measure P .

Let us now extend the discussion to situations where agents are heterogeneous not

only in their utility functions but also in their expectations. Thus, we assume that the

preferences of agent a 2 A are described by a Savage functional of the form

Ua .X / WD EQa ua .X / ;

where Qa is a probability measure on .; F / which is equivalent to P . In addition

to our assumption

(3.63)

lim sup x u0a .x/ < 1; a 2 A;

x#0

we assume that

EQa W < 1

and

W

EQa u0a

< 1;

jAj

a 2 A:

(3.64)

As before, a feasible allocation .Xa /a2A together with a price density ' is called

an ArrowDebreu equilibrium if each Xa maximizes the functional Ua on the budget

set of agent a 2 A, which is determined by ' .

Theorem 3.69. Under assumptions (3.45), (3.63), and (3.64), there exists an Arrow

Debreu equilibrium.

Proof. For any
2 , the general argument of Remark 3.37 yields the existence of

a
-efficient allocation .Xa

/a2A , i.e., of a feasible allocation which maximizes the

functional

X

a Ua .Xa /

U

.X / WD

a2A

Ua .Xa

/ D E 'a ua .Xa

/ ;

174

.Xa

/a2A can be viewed as a
-efficient allocation in the model where agents have

random utility functions of the form

uQ a .x; !/ D ua .x/ 'a .!/;

while their expectations are homogeneous and given by P . In view of Corollary 3.43,

it follows as before that X

satisfies the first order conditions

1

Xa

D IaC .
1

a 'a ' /;

a 2 A;

with

'

D max
a u0a .Xa

/ 'a ;

a2A

and that Xa

satisfies

Ua .Xa

/ E ua .Ya / 'a Ua .Ya /

for all Ya in the budget set of agent a 2 A. The remaining arguments are essentially

the same as in the proof of Theorem 3.65.

Chapter 4

In this chapter, we discuss the problem of quantifying the risk of a financial position. As in Chapter 2, such a position will be described by the corresponding payoff

profile, that is, by a real-valued function X on some set of possible scenarios. In a

probabilistic model, specified by a probability measure on scenarios, we could focus

on the resulting distribution of X and try to measure the risk in terms of moments or

quantiles. Note that a classical measure of risk such as the variance does not capture

a basic asymmetry in the financial interpretation of X: Here it is the downside risk

that matters. This asymmetry is taken into account by measures such as Value at Risk

which are based on quantiles for the lower tail of the distribution, see Section 4.4 below. Value at Risk, however, fails to satisfy some natural consistency requirements.

Such observations have motivated the systematic investigation of measures of risk that

satisfy certain basic axioms.

From the point of view of an investor, we could simply turn around the discussion

of Chapter 2 and measure the risk of a position X in terms of the loss functional

L.X/ D U.X /:

Here U is a utility functional representing a given preference relation on financial

positions. Assuming robust preferences, we are led to the notion of robust shortfall

risk defined by

L.X/ D sup EQ `.X / ;

Q2Q

probability measures. The results of Section 2.5 show how such loss functionals can

be characterized in terms of convexity and monotonicity properties of the preference

relation. In particular, a financial position could be viewed as being acceptable if the

robust shortfall risk of X does not exceed a given bound.

From the point of view of a supervising agency, however, a specific monetary purpose comes into play. In this perspective a risk measure is viewed as a capital requirement: We are looking for the minimal amount of capital which, if added to the

position and invested in a risk-free manner, makes the position acceptable. This monetary interpretation is captured by an additional axiom of cash invariance. Together

with convexity and monotonicity, it singles out the class of convex risk measures.

These measures can be represented in the form

.X / D sup.EQ X .Q//;

Q

176

where is a penalty function defined on probability measures on . Under the additional condition of positive homogeneity, we obtain the class of coherent risk measures. Here we are back to the situation in Proposition 2.84, and the representation

takes the form

.X / D sup EQ X ;

Q2Q

The axiomatic approach to such monetary risk measures was initiated by P. Artzner,

F. Delbaen, J. Eber, and D. Heath [12], and it will be developed in the first three sections. In Section 4.4 we discuss some coherent risk measures related to Value at Risk.

These risk measures only involve the distribution of a position under a given probability measure. In Section 4.5 we characterize the class of convex risk measures which

share this property of law-invariance. Section 4.6 discusses the role of concave distortions, and in Section 4.7 the resulting risk measures are characterized by a property

of comonotonicity. In Section 4.8 we discuss risk measures which arise naturally in

the context of a financial market model. In Section 4.9 we analyze the structure of

monetary risk measures which are induced by our notion of robust shortfall risk.

4.1

X W ! R where X.!/ is the discounted net worth of the position at the end

of the trading period if the scenario ! 2 is realized. The discounted net worth

corresponds to the profits and losses of the position and is also called the P&L. Our

aim is to quantify the risk of X by some number .X /, where X belongs to a given

class X of financial positions. Throughout this section, X will be a linear space

of bounded functions containing the constants. We do not assume that a probability

measure is given on .

Definition 4.1. A mapping W X ! R is called a monetary risk measure if it satisfies

the following conditions for all X; Y 2 X:

Monotonicity: If X Y , then .X / .Y /.

Cash invariance: If m 2 R, then .X C m/ D .X / m.

The financial meaning of monotonicity is clear: The downside risk of a position is

reduced if the payoff profile is increased. Cash invariance is also called translation

invariance or translation property. It is motivated by the interpretation of .X / as a

capital requirement, i.e., .X / is the amount which should be added to the position X

in order to make it acceptable from the point of view of a supervising agency. Thus,

if the amount m is added to the position and invested in a risk-free manner, the capital

requirement is reduced by the same amount. In particular, cash invariance implies

.X C .X // D 0;

(4.1)

177

and

.m/ D .0/ m for all m 2 R.

For most purposes it would be no loss of generality to assume that a given monetary

risk measure satisfies the condition of

Normalization: .0/ D 0.

For a normalized risk measure, cash invariance is equivalent to cash additivity, i.e., to

.X C m/ D .X / C .m/. In some situations, however, it will be convenient not to

insist on normalization.

Remark 4.2. We are using the convention that X describes the worth of a financial position after discounting. For instance, the discounting factor can be chosen as

1=.1 C r/ where r is the return of a risk-free investment. Instead of measuring the

risk of the discounted position X, one could consider directly the nominal worth

XQ D .1 C r/X:

The corresponding risk measure .

Q XQ / WD .X / is again monotone. Cash invariance

is replaced by the following property:

.

Q XQ C .1 C r/m/ D .

Q XQ / m;

(4.2)

manner. Conversely, any Q W X ! R which is monotone and satisfies (4.2) defines a

monetary measure of risk via .X / WD ..1

Q C r/X /.

}

Lemma 4.3. Any monetary risk measure is Lipschitz continuous with respect to the

supremum norm k k:

j.X / .Y /j kX Y k:

Proof. Clearly, X Y C kX Y k, and so .Y / kX Y k .X / by monotonicity

and cash invariance. Reversing the roles of X and Y yields the assertion.

From now on we concentrate on monetary risk measures which have an additional

convexity property.

Definition 4.4. A monetary risk measure W X ! R is called a convex risk measure

if it satisfies

Consider the collection of possible future outcomes that can be generated with the

resources available to an investor: One investment strategy leads to X, while a second

strategy leads to Y . If one diversifies, spending only the fraction
of the resources

on the first possibility and using the remaining part for the second alternative, one

178

idea that diversification should not increase the risk. This idea becomes even clearer

when we note that, for a monetary risk measure, convexity is in fact equivalent to the

weaker requirement of

Exercise 4.1.1. Prove that a monetary risk measure is quasi-convex if and only if it is

convex.

}

Exercise 4.1.2. Show that if is convex and normalized, then

.
X /
.X/ for 0
1,

.
X /
.X/ for
1.

Definition 4.5. A convex risk measure is called a coherent risk measure if it satisfies

i.e., .0/ D 0. Under the assumption of positive homogeneity, convexity is equivalent

to

This property allows to decentralize the task of managing the risk arising from a collection of different positions: If separate risk limits are given to different desks, then

the risk of the aggregate position is bounded by the sum of the individual risk limits.

In many situations, however, risk may grow in a non-linear way as the size of the

position increases. For this reason we will not insist on positive homogeneity. Instead,

our focus will be mainly on convex measures of risk.

Exercise 4.1.3. Let be a normalized monetary risk measure on X. Show that any

two of the following properties imply the remaining third.

Convexity.

Positive homogeneity.

Subadditivity.

A WD X 2 X j .X / 0

of positions which are acceptable in the sense that they do not require additional capital. The class A will be called the acceptance set of . The following two propositions summarize the relations between monetary risk measures and their acceptance

sets.

179

Proposition 4.6. Suppose that is a monetary risk measure with acceptance set A WD

A .

(a) A is non-empty, closed in X with respect to the supremum norm k k, and

satisfies the following two conditions:

infm 2 R j m 2 A > 1:

X 2 A; Y 2 X, Y X

H)

Y 2 A.

(4.3)

(4.4)

.X / D infm 2 R j m C X 2 A:

(4.5)

(d) is positively homogeneous if and only if A is a cone. In particular, is coherent if and only if A is a convex cone.

Proof. (a): Properties (4.3) and (4.4) are straightforward, and closedness follows from

Lemma 4.3.

(b): Cash invariance implies that for X 2 X,

infm 2 R j m C X 2 A D infm 2 R j .m C X / 0

D infm 2 R j .X / m

D .X /:

(c): A is clearly convex if is a convex measure of risk. The converse will follow

from Proposition 4.7 together with (4.7).

(d): Clearly, positive homogeneity of implies that A is a cone. The converse

follows as in (c).

Conversely, one can take a given class A X of acceptable positions as the primary object. For a position X 2 X, we can then define the capital requirement as the

minimal amount m for which m C X becomes acceptable

A .X / WD infm 2 R j m C X 2 A:

(4.6)

A D :

(4.7)

Proposition 4.7. Assume that A is a non-empty subset of X which satisfies (4.3) and

(4.4). Then the functional A has the following properties:

180

(b) If A is a convex set, then A is a convex risk measure.

(c) If A is a cone, then A is positively homogeneous. In particular, A is a coherent risk measure if A is a convex cone.

(d) A is a subset of AA , and A D AA holds if and only if A is k k-closed in X.

Proof. (a): It is straightforward to verify that A satisfies cash invariance and monotonicity. We show next that A takes only finite values. To this end, fix some Y in the

non-empty set A. For X 2 X given, there exists a finite number m with m C X > Y ,

because X and Y are both bounded. Then

A .X / m D A .m C X / A .Y / 0;

and hence A .X / m < 1. Note that (4.3) is equivalent to A .0/ > 1. To

show that A .X / > 1 for arbitrary X 2 X, we take m0 such that X C m0 0 and

conclude by monotonicity and cash invariance that A .X / A .0/ C m0 > 1.

(b): Suppose that X1 ; X2 2 X and that m1 ; m2 2 R are such that mi C Xi 2 A. If

2 0; 1, then the convexity of A implies that .m1 C X1 / C .1 /.m2 C X2 / 2 A.

Thus, by the cash invariance of A ,

0 A . .m1 C X1 / C .1 /.m2 C X2 //

D A . X1 C .1 /X2 / . m1 C .1 /m2 /;

and the convexity of A follows.

(c): As in the proof of convexity, we obtain that A . X / A .X / for 0 if

A is a cone. To prove the converse inequality, let m < A .X /. Then m C X A and

hence m C X A for 0. Thus m < A . X /, and (c) follows.

(d): The inclusion A AA is obvious, and Proposition 4.6 implies that A is

k k-closed as soon as A D AA . Conversely, assume that A is k k-closed. We have

to show that X A implies that A .X / > 0. To this end, take m > kXk. Since A

is k k-closed and X A, there is some 2 .0; 1/ such that m C .1 /X A.

Thus,

0 A . m C .1 /X / D A ..1 /X / m:

Since A is a monetary risk measure, Lemma 4.3 shows that

jA ..1 /X / A .X /j kX k:

Hence,

A .X / A ..1 /X / kXk .m kXk/ > 0:

Exercise 4.1.4. Let A be a non-empty subset of X which satisfies (4.3) and (4.4).

Show that A is k k-closed if and only if A satisfies the following closure property: if

X C m 2 A for all m > 0 then X 2 A.

}

181

In the following examples, we take X as the linear space of all bounded measurable

functions on some measurable space .; F /, and we denote by M1 D M1 .; F / the

class of all probability measures on .; F /.

Example 4.8. Consider the worst-case risk measure max defined by

max .X / D inf X.!/

!2

for all X 2 X.

The value max .X / is the least upper bound for the potential loss which can occur in

any scenario. The corresponding acceptance set A is given by the convex cone of all

non-negative functions in X. Thus, max is a coherent risk measure. It is the most

conservative measure of risk in the sense that any normalized monetary risk measure

on X satisfies

.X / inf X.!/ D max .X /:

!2

max .X / D sup EQ X ;

(4.8)

Q2Q

mapping W Q ! R with supQ .Q/ < 1, which specifies for each Q 2 Q some

floor .Q/. Suppose that a position X is acceptable if

EQ X .Q/

for all Q 2 Q.

The set A of such positions satisfies (4.3) and (4.4), and it is convex. Thus, the

associated monetary risk measure D A is convex, and it takes the form

.X / D sup ..Q/ EQ X /:

Q2Q

.X / D sup .EQ X .Q//;

(4.9)

Q2M1

for Q 2 Q and .Q/ D C1 otherwise. Note that is a coherent risk measure if

.Q/ D 0 for all Q 2 Q.

}

Example 4.10. Consider a utility function u on R, a probability measure Q 2 M1 ,

and fix some threshold c 2 R. Let us call a position X acceptable if its certainty

182

equivalent is at least c, i.e., if its expected utility EQ u.X / is bounded from below

by u.c/. Clearly, the set

A WD X 2 X j EQ u.X / u.c/

is non-empty, convex, and satisfies (4.3) and (4.4). Thus, A is a convex risk measure.

As an obvious robust extension, we can define acceptability in terms of a whole class

Q of probability measures on .; F /, i.e.,

\

X 2 X j EQ u.X / u.cQ /;

A WD

Q2Q

with constants cQ such that supQ2Q cQ < 1. The corresponding risk measures will

be studied in more detail in Section 4.9.

}

Example 4.11. Suppose now that we have specified a probabilistic model, i.e., a probability measure P on .; F /. In this context, a position X is often considered to be

acceptable if the probability of a loss is bounded by a given level
2 .0; 1/, i.e., if

P X < 0
:

The corresponding monetary risk measure V@R

, defined by

V@R

.X / D infm 2 R j P m C X < 0
;

is called Value at Risk at level
. Note that it is well defined on the space L0 .; F ; P /

of all random variables which are P -a.s. finite, and that

V@R

.X / D E X C 1 .1
/ .X /;

(4.10)

of the distribution function of N.0; 1/. Clearly, V@R

is positively homogeneous,

but in general it is not convex, as shown by Example 4.46 below. In Section 4.4, Value

at Risk will be discussed in detail. In particular, we will study some closely related

coherent and convex risk measures.

}

Exercise 4.1.5. Compute V@R

.X / when X is

(a) uniform,

(b) log-normally distributed, i.e., X D e ZCm with Z N.0; 1/ and m; 2 R. }

Example 4.12. As in Example 4.11, we fix a probability measure P on .; F /. For

Q 0,

an asset with payoff XQ 2 L2 D L2 .; F ; P /, price .XQ /, and variance 2 .X/

the Sharpe ratio is defined as

E XQ .XQ /.1 C r/

E X

D

;

.X /

.XQ /

183

that we find the position X acceptable if the Sharpe ratio is bounded from below by

some constant c > 0. The resulting functional c on L2 defined by (4.6) for the class

Ac WD X 2 L2 j E X c .X /

is given by

c .X / D E X C c .X /:

It is sometimes called the mean-standard deviation risk measure. It is cash invariant

and positively homogeneous, and it is convex since . / is a convex functional on

L2 . But c is not a monetary risk measure, because it is not monotone. Indeed, if

X D e Z and Z is a random variable with normal distribution N.0; 2 /, then X 0

but

p

2

2

2

c .X / D e =2 C ce =2 e 1

becomes positive for large enough . Note, however, that (4.10) shows that c .X /

coincides with V@R

.X / if X is Gaussian and if c D 1 .1 / with 0 <

1=2. Thus, both c and V@R

have all the properties of a coherent risk measure if

restricted to a Gaussian subspace XQ of L2 , i.e, a linear space consisting of normally

distributed random variables. But neither c nor V@R

can be coherent on the full

space L2 , since the existence of normal random variables on .; F ; P / implies that

X will also contain random variables as considered in Example 4.46.

}

Example 4.13. Let u W R ! R be a strictly increasing continuous function. For

X 2 X WD L1 .; F ; P / we consider the certainty equivalent of the law of X under

P as a functional of X by setting

c.X / WD u1 .E u.X / /:

Then .X / WD c.X / is monotone,

X Y

H)

.X / .Y /:

If is also cash invariant, and hence a monetary risk measure, then Proposition 2.46

shows that u is either linear or a function with exponential form: u.x/ D a C be x

or u.x/ D a be x for constants a 2 R and b; > 0. In the linear case we have

.X / D E X :

In the first exponential case is of the form

.X / D

1

log E e X :

184

.X / D

1

log E e X

(4.11)

and called the entropic risk measure for reasons that will become clear in Example 4.34. There (and in Exercise 4.1.6 below) we will also see that is a convex risk

measure.

}

Exercise 4.1.6. Let u W R ! R be a strictly increasing continuous function and let

be defined as in Example 4.13. Show that is quasi-convex,

.
X C .1
/Y / .X / _ .Y /

for 0 1,

when u is concave and conclude that the entropic risk measure in (4.11) is a convex

risk measure.

}

Example 4.14. Let c W F ! 0; 1 be any set function which is normalized and

monotone in the sense that c.;/ D 0; c./ D 1, and c.A/ c.B/ if A B. For

instance, c can be given by c.A/ WD .P A / for some probability measure P and

an increasing function

W 0; 1 ! 0; 1 such that .0/ D 0 and .1/ D 1. The

Choquet integral of a bounded measurable function X 0 with respect to c is defined

as

Z 1

Z

c.X > x/ dx:

X dc WD

0

R

If c is a probability measure, Fubinis theorem implies that X dc coincides with

the usual integral. In theR general case,Rthe ChoquetR integral is a nonlinear

functional

R

of X, but we still have
X dc D
X dc and .X C m/ dc D X dc C m for

constants
; m 0. If X 2 X is arbitrary, we take m 2 R such that X C m 0 and

get

Z

.X C m/ dc m D

.c.X > x/ 1/ dx C

m

0

the definition of the Choquet integral by putting

Z

.c.X > x/ 1/ dx C

X dc WD

1

Z

Z

X dc D X dc

c.X > x/ dx

0

Z

and

Z

.X C m/ dc D

X dc C m

Z

Z

Y dc X dc

185

for Y X.

Z

.X / WD

.X / dc;

(4.12)

characterize these risk measures in terms of a property called comonotonicity. We

will also show that is convex, and hence coherent, if and only if c is submodular or

2-alternating, i.e.,

c.A \ B/ C c.A [ B/ c.A/ C c.B/ for A; B 2 F .

In this case, admits the representation

.X / D max EQ X ;

Q2Qc

(4.13)

where Qc is the core of c, defined as the class of all finitely additive and normalized set

functions Q W F ! 0; 1 such that Q A c.A/ for all A 2 F ; see Theorem 4.94.

}

Exercise 4.1.7. Let P be a probability measure on .; F / and fix n 2 N. For X 2 X

let X1 ; : : : ; Xn be independent copies of X and set

.X / WD E min.X1 ; : : : ; Xn / :

The functional W X ! R is sometimes called MINVAR. Show that is a coherent

risk measure on X. Show next that fits into the framework of Example 4.14. More

precisely, show that can be represented as a Choquet integral,

Z

.X / D

.X / dc;

where the set function c is of the form c.A/ D .P A/ for a concave increasing

function W 0; 1 ! 0; 1 such that .0/ D 0 and .1/ D 1.

}

In the next two sections, we are going to show how representations of the form

(4.8), (4.13), (4.9), or (4.12) for coherent or convex risk measures arise in a systematic

manner.

186

4.2

measure P is fixed on the measurable space .; F /. Let X denote the space of

all bounded measurable functions on .; F /. Recall that X is a Banach space if

endowed with the supremum norm k k. As in Section 2.5, we denote by M1 WD

M1 .;F / the set of all probability measures on .;F / and by M1;f WDM1;f .;F /

the set of all finitely additive set functions Q W F ! 0; 1 which are normalized to

Q D 1. By EQ X we denote the integral of X with respect to Q 2 M1;f ; see

Appendix A.6. We do not assume that a probability measure on .; F / is given a

priori.

If is a coherent risk measure on X, then we are in the context of Proposition 2.84,

i.e., the functional defined by .X / WD .X / satisfies the four properties listed in

Proposition 2.83. Hence, we have the following result:

Proposition 4.15. A functional W X ! R is a coherent risk measure if and only if

there exists a subset Q of M1;f such that

.X / D sup EQ X ;

X 2 X:

(4.14)

Q2Q

Moreover, Q can be chosen as a convex set for which the supremum in (4.14) is attained.

Our first goal in this section is to obtain an analogue of this result for convex risk

measures. Applied to a coherent risk measure, it will yield an alternative proof of

Proposition 4.15, which does not depend on the discussion in Chapter 2, and it will

provide a description of the maximal set Q in (4.14). Our second goal will be to obtain

criteria which guarantee that a risk measure can be represented in terms of -additive

probability measures.

Let W M1;f ! R [ C1 be any functional such that

inf

Q2M1;f

.Q/ 2 R:

and cash invariant on X, and these three properties are preserved when taking the

supremum over Q 2 M1;f . Hence,

.X / WD

Q2M1;f

.0/ D

inf

Q2M1;f

.Q/:

(4.15)

187

The functional will be called a penalty function for on M1;f , and we will say that

is represented by on M1;f .

Theorem 4.16. Any convex risk measure on X is of the form

.X / D

Q2M1;f

X 2 X;

(4.16)

min .Q/ WD sup EQ X for Q 2 M1;f .

X2A

Moreover, min is the minimal penalty function which represents , i.e., any penalty

function for which (4.15) holds satisfies .Q/ min .Q/ for all Q 2 M1;f .

Proof. In a first step, we show that

.X /

for all X 2 X.

Q2M1;f

To this end, recall that X 0 WD .X / C X 2 A by (4.1). Thus, for all Q 2 M1;f

min .Q/ EQ X 0 D EQ X .X /:

From here, our claim follows.

For X given, we will now construct some QX 2 M1;f such that

.X / EQX X min .QX /;

which, in view of the previous step, will prove our representation (4.16). By cash

invariance it suffices to prove this for X 2 X with .X / D 0. Moreover, we may

assume without loss of generality that .0/ D 0. Then X is not contained in the

nonempty convex set

B WD Y 2 X j .Y / < 0:

Since B is open due to Lemma 4.3, we may apply the separation argument in the form

of Theorem A.55. It yields a non-zero continuous linear functional ` on X such that

`.X / inf `.Y / DW b:

Y 2B

that 1 C Y 2 B for any > 0. Hence,

`.X / `.1 C Y / D `.1/ C `.Y /

which could not be true if `.Y / < 0.

188

Our next claim is that `.1/ > 0. Since ` does not vanish identically, there must

be some Y such that 0 < `.Y / D `.Y C / `.Y /. We may assume without loss of

generality that kY k < 1. Positivity of ` implies `.Y C / > 0 and `.1 Y C / 0.

Hence `.1/ D `.1 Y C / C `.Y C / > 0.

By the two preceding steps and Theorem A.51, we conclude that there exists some

QX 2 M1;f such that

EQX Y D

`.Y /

`.1/

for all Y 2 X.

min .QX / D sup EQX Y sup EQX Y D

Y 2A

Y 2B

b

:

`.1/

On the other hand, Y C " 2 B for any Y 2 A and each " > 0. This shows that

min .QX / is in fact equal to b=`.1/. It follows that

EQX X min .QX / D

1

.b `.X // 0 D .X /:

`.1/

Finally, let be any penalty function for . Then, for all Q 2 M1;f and X 2 X

.X / EQ X .Q/;

and hence

.Q/ sup .EQ X .X //

X2X

(4.17)

X2A

min .Q/:

Thus, dominates min .

Remark 4.17. (a) If we take D min in (4.17), then all inequalities in (4.17) must

be identities. Thus, we obtain an alternative formula for the minimal penalty

function min :

min .Q/ D sup .EQ X .X //:

(4.18)

X2X

(b) Note that min is convex and lower semicontinuous for the total variation distance on M1;f as defined in Definition A.50, since it is the supremum of affine

continuous functions on M1;f .

189

determines min :

min .Q/ D sup EQ X for all Q 2 M1;f .

X2A

This follows from the fact that X 2 A implies " C X 2 A for all " > 0.

Remark 4.18. Equation (4.18) shows that the penalty function min corresponds to

the FenchelLegendre transform, or conjugate function, of the convex function on

the Banach space X. More precisely,

min .Q/ D .`Q /;

(4.19)

.`/ D sup .`.X / .X //;

X2X

an alternative proof of Theorem 4.16. First note that, by Theorem A.51, X 0 can be

identified with the space ba WD ba.; F / of finitely additive set functions with finite

total variation. Moreover, is lower semicontinuous with respect to the weak topology .X; X 0 /, since any set c is convex, strongly closed due to Lemma 4.3,

and hence weakly closed by Theorem A.60. Thus, the general duality theorem for

conjugate functions as stated in Theorem A.62 yields

D ;

where denotes the conjugate function of , i.e.,

.X / D sup .`.X / .`//:

(4.20)

`2ba

In a second step, using the arguments in the second part of the proof of Theorem 4.16,

we can now check that monotonicity and cash invariance of imply that ` 0 and

`.1/ D 1 for any ` 2 X 0 D ba such that .`/ < 1. Identifying ` with Q 2 M1;f

and using equation (4.19), we see that (4.20) reduces to the representation

.X / D

Q2M1;f

due to the BanachAlaoglu theorem stated in Theorem A.63, and so the upper semicontinuous functional Q 7! EQ X min .Q/ attains its maximum on M1;f .

}

190

The representation

.X / D sup EQ X ;

X 2 X;

(4.21)

Q2Q

4.15, is a particular case of the representation theorem for convex risk measures, since

it corresponds to the penalty function

0

if Q 2 Q

.Q/ D

C1 otherwise.

The following corollary shows that the minimal penalty function of a coherent risk

measure is always of this type.

Corollary 4.19. The minimal penalty function min of a coherent risk measure takes

only the values 0 and C1. In particular,

.X / D max EQ X ;

Q2Qmax

X 2 X;

Qmax WD Q 2 M1;f j min .Q/ D 0;

and Qmax is the largest set for which a representation of the form (4.21) holds.

Proof. Recall from Proposition 4.6 that the acceptance set A of a coherent risk measure is a cone. Thus, the minimal penalty function satisfies

min .Q/ D sup EQ X D sup EQ X D min .Q/

X2A

X2A

for all Q 2 M1;f and
> 0. Hence, min can take only the values 0 and C1.

Exercise 4.2.1. Let be a coherent risk measure on X and assume that admits a

representation

.X / D sup EQ X

Q2Q

with some class Q of probability measures on .; F /. Show that is additive, i.e.,

.X C Y / D .X / C .Y /

for all X; Y 2 X,

if and only if the class Q reduces to a single probability measure Q, i.e., is simply

the expected loss with respect to Q.

}

191

The penalty function arising in the representation (4.15) is not unique, and it is

often convenient to represent a convex risk measure by a penalty function that is not

the minimal one. For instance, the minimal penalty function may be finite for certain finitely additive set functions while another is concentrated only on probability

measures as in the case of Example 4.8. Another situation of this type occurs for risk

measures which are constructed as the supremum of a family of convex risk measures.

Proposition 4.20. Suppose that for every i in some index set I we are given a convex

risk measure i on X with associated penalty function i . If supi 2I i .0/ < 1 then

.X / WD sup i .X /;

X 2 X;

i 2I

is a convex risk measure that can be represented with the penalty function

.Q/ WD inf i .Q/;

i 2I

Q 2 M1;f :

Proof. The condition .0/ D supi 2I i .0/ < 1 implies that takes only finite

values. Moreover,

sup .EQ X i .Q//

.X / D sup

i 2I Q2M1;f

EQ X inf i .Q/ ;

sup

Q2M1;f

i 2I

In the sequel, we are particularly interested in those convex measures of risk which

admit a representation in terms of -additive probability measures. Such a risk measure can be represented by a penalty function which is infinite outside the set

M1 WD M1 .; F /:

.X / D sup .EQ X .Q//:

(4.22)

Q2M1

In this case, one can no longer expect that the supremum above is attained. This is

illustrated by Example 4.8 if X does not take on its infimum.

A representation (4.22) in terms of probability measures is closely related to certain

continuity properties of . We first examine a necessary condition of continuity from

above.

Lemma 4.21. A convex risk measure which admits a representation (4.22) on M1

is continuous from above in the sense that

Xn & X

H)

.Xn / % .X /:

(4.23)

192

Moreover, continuity from above is equivalent to the Fatou property of lower semicontinuity with respect to bounded pointwise convergence: If .Xn / is a bounded sequence in X which converges pointwise to X 2 X, then

.X / lim inf .Xn /:

(4.24)

n"1

Proof. First we show (4.24) under the assumption that has a representation in terms

of probability measures. Dominated convergence implies that EQ Xn ! EQ X

for each Q 2 M1 . Hence,

.X / D sup lim EQ Xn .Q/

Q2M1

n"1

n"1 Q2M1

n"1

In order to show the equivalence of (4.24) and (4.23), let us first assume (4.24). By

monotonicity, .Xn / .X / for each n if Xn & X , and so .Xn / % .X / follows.

Now we assume continuity from above. Let .Xn / be a bounded sequence in X

which converges pointwise to X. Define Ym WD supnm Xn 2 X. Then Ym decreases

to X. Since .Xn / .Yn / by monotonicity, condition (4.23) yields that

lim inf .Xn / lim .Yn / D .X /:

n"1

n"1

The following theorem gives a strong sufficient condition which guarantees that

any penalty function for is concentrated on the set M1 of probability measures.

This condition is continuity from below rather than from above; we will see a class

of examples in Section 4.9.

Theorem 4.22. For a convex risk measure on X, the following two conditions are

equivalent:

(a) is continuous from below in the sense that

Xn % X pointwise on

H)

(b) The minimal penalty function min (and hence every other penalty function representing ) is concentrated on the class M1 of probability measures, i.e.,

min .Q/ < 1

H)

Q is -additive.

In particular we have

.X / D max .EQ X min .Q//;

Q2M1

X 2 X;

193

Lemma 4.23. Let be a convex risk measure on X which is represented by the

penalty function on M1;f , and consider the level sets

c WD Q 2 M1;f j .Q/ c;

inf

Q2M1;f

.Q/.

For any sequence .Xn / in X such that 0 Xn 1, the following two conditions are

equivalent:

(a) .
Xn / ! .
/ for each
1.

(b) infQ2c EQ Xn ! 1 for all c > .0/.

Proof. (a) ) (b): In a first step, we show that for all Y 2 X

inf EQ Y

Q2c

c C .
Y /

(4.25)

c .Q/ EQ Y . Y /;

and dividing by yields (4.25).

Now consider a sequence .Xn / which satisfies (a). Then (4.25) shows that for all

1

c C . Xn /

c C .0/

D1

:

n"1

n"1 Q2c

(b) ) (a): Clearly, for all n

. / . Xn / D

Q2M1;f

the right-hand side for which

.Q/ 1 . / D 1 C .0/ DW c:

Hence, for all n

. Xn / D sup .EQ Xn .Q//:

Q2c

and so (a) follows.

194

The proof of our theorem will also rely on Dinis lemma, which we recall here for

the convenience of the reader.

Lemma 4.24. On a compact set, a sequence of continuous functions fn increasing to

a continuous function f converges even uniformly.

T

Proof. For " > 0, the compact sets Kn WD fn f " satisfy n Kn D ;, hence

Kn0 D ; for some n0 .

Proof of Theorem 4:22. To prove the implication (a) ) (b), recall that Q is -additive if and

S only if Q An % 1 for any increasing sequence of events An 2 F

such that n An D . Thus, our claim is implied by the implication (a) ) (b) of

Lemma 4.23 if we take Xn WD IAn .

We now prove the implication (b) ) (a) of our theorem. Suppose Xn % X pointwise on . We need to show that .Xn / & .X /. By cash invariance, we may assume

without loss of generality that Xn 0 for all n. As in the proof of the implication (a)

) (b) of Lemma 4.23, we see that

.Xn / D

max .EQ Xn min .Q// D max .EQ Xn min .Q//; (4.26)

Q2M1;f

Q2c

that c M1 implies that

EQ Xn ! EQ X uniformly in Q 2 c .

(4.27)

Together with the representation (4.26), this will imply the desired convergence

.Xn / ! .X /.

To prove (4.27), recall first from Appendix A.6, and in particular from Definition

A.50, that M1;f belongs to the larger vector space ba WD ba.; F / of all finitely

additive set functions

W F ! R with finite total variation k

kvar . In fact, ba can be

identified with the topological dual of the Banach space X with respect to k k; see

Theorem A.51. Since we can write

M1;f D

2 ba j

./ D 1 and

.A/ 0 for all A 2 F ;

it is clear that M1;f is a bounded and weak closed set in ba. Hence M1;f is weak

compact by the BanachAlaoglu theorem (see Theorem A.63). Since min is weak

lower semicontinuous on M1;f as supremum of the weak continuous maps Q 7!

EQ Y with Y 2 A , the level set c is also weak compact.

After these preparations, we can now prove (4.27). Clearly, the functions `n .Q/ WD

EQ Xn form a decreasing sequence of weak continuous functions on c . Moreover, when c M1 , we even have `n .Q/ & `.Q/ WD EQ X for each Q 2 c .

By the established compactness of c , (4.27) thus follows from Dinis lemma.

195

Remark 4.25. Let be a convex risk measure which is continuous from below. Then

is also continuous from above, as can be seen by combining Theorem 4.22 and

Lemma 4.21.

}

Exercise 4.2.2. Show that for a convex risk measure on X the following two conditions are equivalent:

(a) is continuous from below.

(b) satisfies the following Lebesgue property: .Xn / ! .X / whenever .Xn /

is a bounded sequence in X which converges pointwise to X .

}

Exercise 4.2.3. Show that for a convex risk measure on X the following two conditions are equivalent:

(a) is continuous from below.

(b) For every c > .0/, the coherent risk measures

c .X / WD sup EQ X

Q2c

M1 .; F /, and fix some threshold c 2 R. As in Example 4.10, we suppose that

a position X is acceptable if its expected utility EQ u.X / is bounded from below

by u.c/. Alternatively, we can introduce the convex increasing loss function `.x/ D

u.x/ and define the convex set of acceptable positions

A WD X 2 X j EQ `.X / x0 ;

where x0 WD u.c/. Let WD A denote the convex risk measure induced by A.

In Section 4.9, we will show that is continuous from below, and we will derive a

formula for its minimal penalty function.

}

Let us now continue the discussion in a topological setting. More precisely, we

will assume for the rest of this section that is a separable metric space and that F

is the -field of Borel sets. As before, X is the linear space of all bounded measurable functions on .; F /. We denote by Cb ./ the subspace of bounded continuous

functions on , and we focus on the representation of convex risk measures viewed

as functionals on Cb ./.

Proposition 4.27. Let be a convex risk measure on X such that

.Xn / & . / for any sequence .Xn / in Cb ./ that increases to a constant > 0.

(4.28)

196

.X / D max .EQ X .Q//

for X 2 Cb ./.

(4.29)

Q j E Q D EQ on Cb ./:

.Q/ WD inf min .Q/

Q

(4.30)

Q2M1

Proof. Let min be the minimal penalty function of on M1;f . We show that for any

Q < 1 there exists Q 2 M1 such that E Q X D EQ X for all

QQ with min .Q/

Q

X 2 Cb ./. Take a sequence .Yn / in Cb ./ which increases to some Y 2 Cb ./,

and choose > 0 such that Xn WD 1 C .Yn Y / 0 for all n. Clearly, .Xn / satisfies

condition (a) of Lemma 4.23, and so EQQ Xn ! 1, i.e.,

EQQ Yn % EQQ Y :

This continuity property of the linear functional EQQ on Cb ./ implies, via the

DaniellStone representation theorem as stated in Appendix A.6, that it coincides on

Cb ./ with the integral with respect to a -additive measure Q. Taking as in (4.30)

gives the result.

Remark 4.28. If is compact then any convex risk measure admits a representation

(4.29) on the space Cb ./ D C./. Indeed, if .Xn / is a sequence in Cb ./ that

increases to a constant
, then this convergence is even uniform by Lemma 4.24.

Since is Lipschitz continuous on C./ by Lemma 4.3, it satisfies condition (4.28).

Alternatively, we could argue as in Remark 4.18 and apply the general duality theorem for the FenchelLegendre transform to the convex functional on the Banach

space C./. Just note that any continuous functional ` on C./ which is positive and

normalized is of the form `.X / D EQ X for some probability measure Q 2 M1 ;

see Theorem A.48.

}

Definition 4.29. A convex risk measure on X is called tight if there exists an increasing sequence K1 K2 of compact subsets of such that

.
IKn / ! .
/

for all 1.

Proposition 4.30. Suppose that the convex risk measure on X is tight. Then (4.28)

holds and the conclusion of Proposition 4:27 is valid. Moreover, if is a Polish space

and is a penalty function on M1 such that

.X / D sup .EQ X .Q//

for X 2 Cb ./,

Q2M1

then the level sets c D Q 2 M1 j .Q/ c are relatively compact for the weak

topology on M1 .

197

Proof. First we show (4.28). Suppose Xn 2 Cb ./ are such that Xn %
> 0. We

may assume without loss of generality that is normalized. Convexity and normalization guarantee that condition (4.28) holds for all
> 0 as soon as it holds for all

c where c is an arbitrary constant larger than 1. Hence, the cash invariance of

implies that there is no loss of generality in assuming Xn 0 for all n. We must

show that .Xn / .
/ C 2" eventually, where we take " 2 .0;
1/.

By assumption, there exists a compact set KN such that

..
"/IKN / .
"/ C " D .
/ C 2":

By Dinis lemma as recalled in Lemma 4.24, there exists some n0 2 N such that

" Xn on KN for all n n0 . Finally, monotonicity implies

.Xn / ..
"/IKN / .
/ C 2":

To prove the relative compactness of c , we will show that for any " > 0 there

exists a compact set K" such that for all c > .0/

inf Q K" 1 ".c C .0/ C 1/:

Qc

Q2

characterization of weakly compact sets in M1 , as stated in Theorem A.42. We fix a

countable dense set !1 ; !2 ; : : : and a complete metric which generates the

topology of . For r > 0 we define continuous functions ri on by

ri .!/ WD 1

.!; !i / ^ r

:

r

The function ri is dominated by the indicator function of the closed metric ball

B r .!i / WD ! 2 j .!; !i / r:

Let

Xnr .!/ WD max ri .!/:

i n

Clearly, Xnr is continuous and satisfies 0 Xnr 1 as well as Xnr % 1 for n " 1.

According to (4.25), we have for all
> 0

inf Q

Q2c

n

h [

i D1

i

c C .
Xnr /

B r .!i / inf EQ Xnr

:

Q2c

Now we take
k WD 2k =" and rk WD 1=k. The first part of this proof and (4.28) yield

the existence of nk 2 N such that

.
k Xnrkk / .
k / C 1 D
k C 1;

198

and thus

sup Q

nk

h \

Q2c

i D1

i cC1

nB rk .!i /

D "2k .c C 1/:

k

We let

K" WD

nk

1 [

\

B rk .!i /:

kD1 i D1

Q K" D 1 Q

nk

1 \

h [

nB rk .!i /

kD1 i D1

1

1

X

"2k .c C 1/

kD1

D 1 ".c C 1/:

The reader may notice that K" is closed, totally bounded and, hence, compact. A short

proof of this fact goes as follows: Let .xj / be a sequence in K" . We must show that

.xj / has a convergent subsequence. Since K" is covered by B rk .!1 /; : : : ; B rk .!nk /

for each k, there exists some ik nk such that infinitely many xj are contained in

B rk .!ik /. A diagonalization argument yields a single subsequence .xj 0 / which for

each k is contained in some B rk .!ik /. Thus, .xj 0 / is a Cauchy sequence with respect

to the complete metric and, hence, converging to some element ! 2 .

Remark 4.31. Note that the representation (4.29) does not necessarily extend from

Cb ./ to the space X of all bounded measurable functions. Suppose in fact that

is compact but not finite, so that condition (4.28) holds as explained in Remark 4.28.

There is a finitely additive Q0 2 M1;f which does not belong to M1 ; see Example A.53. The proof of Proposition 4.27 shows that there is some QQ 2 M1 such that

the coherent risk measure defined by .X / WD EQ0 X coincides with EQQ X

for X 2 Cb ./. But does not admit a representation of the form

.X / D sup .EQ X .Q//

for all X 2 X.

Q2M1

.Q/ EQ0 X EQ X

for Q 2 M1 and any X 2 X, hence .Q/ D 1 for any Q 2 M1 .

4.3

199

In the sequel, we fix a probability measure P on .; F / and consider risk measures

such that

.X / D .Y / if X D Y P -a.s.

(4.31)

Note that only the nullsets of P will matter in this section.

Lemma 4.32. Let be a convex risk measure that satisfies (4.31) and which is represented by a penalty function as in (4.15). Then .Q/ D C1 for any Q 2

M1;f .; F / which is not absolutely continuous with respect to P .

Proof. If Q 2 M1;f .; F / is not absolutely continuous with respect to P , then there

exists A 2 F such that Q A > 0 but P A D 0. Take any X 2 A , and define

Xn WD X n IA . Then .Xn / D .X /, i.e., Xn is again contained in A . Hence,

.Q/ min .Q/ EQ Xn D EQ X C n Q A ! 1

as n " 1.

In view of (4.31), we can identify X with the Banach space L1 WD L1 .; F ; P /.

Let us denote by

M1 .P / WD M1 .; F ; P /

the set of all probability measures on .; F / which are absolutely continuous with

respect to P . The following theorem characterizes those convex risk measures on L1

that can be represented by a penalty function concentrated on probability measures,

and hence on M1 .P /, due to Lemma 4.32.

Theorem 4.33. Suppose W L1 ! R is a convex risk measure. Then the following

conditions are equivalent:

(a) can be represented by some penalty function on M1 .P /.

(b) can be represented by the restriction of the minimal penalty function min to

M1 .P /

.X / D

sup

X 2 L1 :

(4.32)

Q2M1 .P /

(d) has the following Fatou property: for any bounded sequence .Xn / which converges P -a.s. to some X,

.X / lim inf .Xn /:

n"1

200

(f) The acceptance set A of is weak closed in L1 , i.e., A is closed with

respect to the topology .L1 ; L1 /.

Proof. The implication (b) ) (a) is obvious, and (a) ) (c) , (d) follows as in

Lemma 4.21, replacing pointwise convergence by P -a.s. convergence.

(c) ) (e): We have to show that C WD c is weak closed for c 2 R. To

this end, let Cr WD C \ X 2 L1 j kX k1 r for r > 0. If .Xn / is a sequence

in Cr converging in L1 to some random variable X, then there is a subsequence that

converges P -a.s., and the Fatou property of implies that X 2 Cr . Hence, Cr is

closed in L1 , and Lemma A.65 implies that C WD c is weak closed.

(e) ) (f) is obvious.

(f) ) (b): We fix some X 2 L1 and let

mD

sup

. EQ X min .Q/ /:

(4.33)

Q2M1 .P /

In view of Theorem 4.16, we need to show that m .X / or, equivalently, that

m C X 2 A . Suppose by way of contradiction that m C X A . Since the nonempty convex set A is weak closed by assumption, we may apply Theorem A.57 in

the locally convex space .L1 ; .L1 ; L1 // with C WD A and B WD m C X . We

obtain a continuous linear functional ` on .L1 ; .L1 ; L1 // such that

WD inf `.Y / > `.m C X / DW > 1:

Y 2A

(4.34)

0. To show this, fix Y 0 and note that . Y / .0/ for 0, by monotonicity.

Hence Y C .0/ 2 A for all 0. It follows that

1 < < `. Y C .0// D `.Y / C `..0//:

Taking " 1 yields that `.Y / 0 and in turn that Z 0. Moreover, P Z > 0 > 0

since ` is non-zero. Thus,

Z

dQ0

WD

dP

E Z

defines a probability measure Q0 2 M1 .P /. By (4.34), we see that

min .Q0 / D sup EQ0 Y D

Y 2A

:

E Z

However,

EQ0 X C m D

`.m C X /

D

<

D min .Q0 /;

E Z

E Z

E Z

.X /.

201

The theorem shows that any convex risk measure of L1 that is continuous from

above arises in the following manner. We consider any probabilistic model Q 2

M1 .P /, but these models are taken more or less seriously as described by the penalty

function. Thus, the value .X / is computed as the worst case, over all models

Q 2 M1 .P /, of the expected loss EQ X , but reduced by .Q/. In the following

example, the given model P is the one which is taken most seriously, and the penalty

function .Q/ is proportional to the deviation of Q from P , measured by the relative

entropy.

Example 4.34. Consider the penalty function W M1 .P / ! .0; 1 defined by

.Q/ WD

1

H.QjP /;

h

dQ i

H.QjP / D EQ log

dP

is the relative entropy of Q 2 M1 .P / with respect to P ; see Definition 3.20. The

corresponding entropic risk measure is given by

.X / D

sup

Q2M1 .P /

EQ X

1

H.QjP / :

The variational principle for the relative entropy as stated in Lemma 3.29 shows that

EQ X

1

1

H.QjP / log E e X ;

and the upper bound is attained by the measure with the density e X =E e X .

Thus, the entropic risk measure takes the form

.X / D

1

log E e X :

Note that is in fact the minimal penalty function representing , since Lemma 3.29

implies

1

1

min .Q/ D sup EQ X log E e X D H.QjP /:

X2L1

A financial interpretation of the entropic risk measure in terms of shortfall risk will be

discussed in Example 4.114.

}

Exercise 4.3.1. Show that the entropic risk measure converges to the worst-case

risk measure for " 1 and to the expected loss under P for # 0.

}

202

The following corollary characterizes those convex risk measures on L1 that satisfy the property of continuity from below, which is stronger than continuity from

above.

Corollary 4.35. For a convex risk measure on L1 , the following conditions are

equivalent:

(a) is continuous from below: Xn % X H) .Xn / & .X /.

(b) satisfies the Lebesgue property: .Xn / ! .X / whenever .Xn / is a bounded

sequence in L1 which converges P -a.s. to X .

(c) The minimal penalty function min is concentrated on M1 .P /, i.e., min .Q/ <

1 implies Q 2 M1 .P /.

In particular we have

.X / D

max

Q2M1 .P /

X 2 L1 ;

Proof. The equivalence of conditions (a) and (b) was shown in Exercise 4.2.2. The

equivalence of conditions (a) and (c) follows from Theorem 4.22 and Lemma 4.32.

Exercise 4.3.2. Show that the three conditions in Corollary 4.35 are equivalent to the

following fourth condition:

(d) For each c 2 R, the level set c WD Q j min .Q/ c is contained in M1 .P /,

and the corresponding set of densities,

dQ

Q 2 c ;

dP

is weakly compact in L1 .; F ; P /.

Hint: Use Lemma 4.23 and the DunfordPettis theorem (Theorem A.67).

function satisfying g.1/ < 1 and the superlinear growth condition g.x/=x ! C1

as x " 1. Associated with it is the g-divergence

h dQ i

;

Ig .QjP / WD E g

dP

Q 2 M1 .P /:

(4.35)

The g-divergence Ig .QjP / quantifies the deviation of the hypothetical model Q from

the reference measure P . Thus,

g .Q/ WD Ig .QjP /;

Q 2 M1 .P /;

203

g .X / WD sup .EQ X Ig .QjP //

(4.36)

QP

is sometimes called divergence risk measure. Note that, for g.x/ D 1 x log x, g is

just the entropic risk measure discussed in Example 4.34. Divergence risk measures

will be discussed in more detail in Section 4.9.

}

Exercise 4.3.3. Show that the risk measure in (4.36) is continuous from below and

that g .Q/ D Ig .QjP /, Q 2 M1 .P /, is its minimal penalty function. In particular,

the supremum in (4.36) is in fact a maximum.

Hint: Using Exercise 4.3.2 can be helpful.

}

Theorem 4.33 takes the following form for coherent risk measures; the proof is the

same as the one for Corollary 4.19.

Corollary 4.37. A coherent risk measure on L1 can be represented by a set Q

M1 .P / if and only if the equivalent conditions of Theorem 4:33 are satisfied. In this

case, the maximal representing subset of M1 .P / is given by

Qmax WD Q 2 M1 .P / j min .Q/ D 0:

Let us also state a characterization of those coherent risk measures on L1 which

are continuous from below.

Corollary 4.38. For a coherent risk measure on L1 the following properties are

equivalent:

(a) is continuous from below: Xn % X H) .Xn / & .X /.

(b) satisfies the Lebesgue property: .Xn / ! .X / whenever .Xn / is a bounded

sequence in L1 which converges P -a.s. to X .

(c) We have Qmax M1 .P /.

(d) The set of densities

dQ

Q 2 Qmax

dP

In this case, the representation

.X / D max EQ X ;

Q2Qmax

X 2 L1 ;

204

We now give three examples of coherent risk measures which will be studied in

more detail in Section 4.4.

Example 4.39. In our present context, where we require condition (4.31), the worstcase risk measure takes the form

max .X / WD ess inf X D infm 2 R j X C m 0 P -a.s.:

One can easily check that max is coherent and satisfies the Fatou property. Moreover,

1

the acceptance set of max is equal to the positive cone L1

C in L , and this implies

min .Q/ D 0 for any Q 2 M .P /. Thus,

1

max .X / D

sup

EQ X :

Q2M1 .P /

Note however that the supremum on the right cannot be replaced by a maximum as

soon as .; F ; P / cannot be reduced to a finite model. Indeed, in that case there

exists X 2 L1 such that X does not attain its essential infimum, and so there can

be no Q 2 M1 .P / such that EQ X D ess inf X D max .X /. In this case, the

preceding corollary shows that max is not continuous from below.

}

Example 4.40. Let Q

be the class of all Q 2 M1 .P / whose density dQ=dP is

bounded by 1=
for some fixed parameter
2 .0; 1/. The corresponding coherent

risk measure

AV@R

.X / WD sup EQ X

(4.37)

Q2Q

will be called the Average Value at Risk at level
. This terminology will become clear

in Section 4.4, which contains a detailed study of AV@R

. By taking g.x/ WD 0 for

x

1 and g.x/ WD C1 for x >

1 , one sees that AV@R

falls into the class of divergence risk measures as introduced in Example 4.36. It follows from Exercise 4.3.3

that Q

is equal to the maximal representing subset of AV@R

, that AV@R

is continuous from below, and that the supremum in (4.37) is actually attained. An explicit

construction of the maximizing measure will be given in the proof of Theorem 4.52.

}

Example 4.41. We take for Q the class of all conditional distributions P j A such

that A 2 F has P A >
for some fixed level
2 .0; 1/. The coherent risk measure

induced by Q,

WCE

.X / WD sup E X j A j A 2 F ; P A >
;

(4.38)

is called the worst conditional expectation at level
. We will show in Section 4.4

that it coincides with the Average Value at Risk of Example 4.40 if the underlying

probability space is rich enough.

}

205

Let be a convex risk measure with the Fatou property. We now consider the situation in which admits a representation in terms of equivalent probability measures

Q P , i.e.,

.X / D sup .EQ X min .Q//;

X 2 L1 :

(4.39)

Q P

We will show in the next theorem that this property can be characterized by the following concept of sensitivity, which is sometimes also called relevance. It formalizes

the idea that should react to every nontrivial loss at a sufficiently high level.

Definition 4.42. A convex risk measure on L1 is called sensitive with respect to

P if for every nonconstant X 2 L1 with X 0 there exists
> 0 such that

.
X/ > .0/.

Theorem 4.43. For a convex risk measure with the Fatou property, the following

conditions are equivalent:

(a) admits the representation (4.39) in terms of equivalent probability measures.

(b) is sensitive with respect to P .

(c) For every A 2 F with P A > 0 there exists
> 0 such that .
IA / > .0/.

(d) For every A 2 F with P A > 0 there exists Q 2 M1 .P / with Q A > 0 and

min .Q/ < 1.

(e) There exists Q P with min .Q/ < 1.

Proof. Throughout the proof we will assume for simplicity that is normalized in the

sense that .0/ D 0. This can be done without loss of generality.

(a) ) (b): Take X 2 L1

C with E X > 0. By (4.39) there exists Q P with

min

.Q/ < 1 and

.
X/
EQ X min .Q/:

The right-hand side is strictly positive as soon as

>

min .Q/

;

EQ X

The implications (b) ) (c) and (c) ) (d) are both obvious.

(d) ) (e): For every c > 0 the level set c WD Q 2 M1 .P / j min .Q/ c

is nonempty due to our assumption .0/ D 0. We will show that c contains some

Q P . We show first the following auxiliary claim:

For any A 2 F with P A > 0 there exists Q 2 c with Q A > 0.

(4.40)

206

Indeed, (d) implies that there is Q0 2 M1 .P / with Q0 A > 0 and min .Q0 / < 1.

Now we take Q1 2 c=2 and let Q" WD "Q0 C .1 "/Q1 for 0 < " < 1. We clearly

have Q" A > 0 and

c

min .Q" / " min .Q1 / C .1 "/ < c

2

for sufficiently small " > 0. This implies (4.40).

We now apply the HalmosSavage theorem in the form of Theorem 1.61. It yields

the existence of Q 2 c with Q P if we can show that c is countably convex.

real numbers

To show countable convexity, let .k /k2N be a sequence of nonnegative

P

k

summing up to 1 and take Qk 2 c for k 2 N. We define Q WD 1

kD1 k Q . Then

Q 2 M1 .P / and

min .Q/ D sup EQ X D sup

X2A

1

X

kD1

1

X

X2A kD1

k sup EQk X D

X2A

k EQk X

1

X

k min .Qk / c:

kD1

(e) ) (a): By the representation (4.32) in Theorem 4.33 it is sufficient to show that

.X / D

sup

Q2M1 .P /

Q P

for any given X 2 L1 . To this end, we take > 0 and choose Q1 2 M1 .P / such

that

EQ1 X min .Q1 / > .X/ :

Then we take Q0 P with min .Q0 / < 1, which exists due to (e). When letting

Q" WD "Q0 C .1 "/Q1 we have Q" P for all " 2 .0; 1/ and min .Q" /

" min .Q1 / C .1 "/ min .Q0 /. Hence,

EQ" X min .Q" / ".EQ0 X min .Q0 //

C .1 "/.EQ1 X min .Q1 //;

and this is larger that .X / when " is sufficiently small.

Condition (e) in Theorem 4.43 is of course satisfied when min .P / < 1. This is

the case for the worst conditional expectation WCE (Example 4.41), Average Value

at Risk AV@R (Example 4.40), the divergence risk measures (Example 4.36), and in

particular the entropic risk measure (Example 4.34). These, and many other risk measures, are therefore sensitive and admit a representation (4.39) in terms of equivalent

risk measures.

207

Remark 4.44. In analogy to Remark 4.18, the implication (e) ) (a) in the Representation Theorem 4.33 can be viewed as a special case of the general duality in

Theorem A.62 for the FenchelLegendre transform of the convex function on L1 ,

combined with the properties of a monetary risk measure. From this general point of

view, it is now clear how to state representation theorems for convex risk measures on

the Banach spaces Lp .; F ; P / for 1 p < 1. More precisely, let q 2 .1; 1 be

such that p1 C q1 D 1, and define

dQ

q

2 Lq :

M1 .P / WD Q 2 M1 .P /

dP

A convex risk measure on Lp is of the form

.X / D

sup

Q2M1q .P /

.EQ X .Q//

if and only if it is lower semicontinuous on Lp , i.e., the Fatou property holds in the

form

}

Xn ! X in Lp H) .X / lim inf .Xn /:

n"1

4.4

Value at Risk

consists in specifying a quantile of the distribution of X under the given probability

measure P . For 2 .0; 1/, a -quantile of a random variable X on .; F ; P / is any

real number q with the property

PX q

and

P X < q ;

and the set of all
-quantiles of X is an interval qX .
/; qXC .
/, where

qX .t / D supx j P X < x < t D infx j P X x t

is the lower and

qXC .t / D infx j P X x > t D supx j P X < x t

is the upper quantile function of X; see Appendix A.3. In this section, we will focus

on the properties of qXC .
/, viewed as a functional on a space of financial positions X.

Definition 4.45. Fix some level
2 .0; 1/. For a financial position X , we define its

Value at Risk at level
as

V@R

.X / WD qXC .
/ D qX

.1
/ D infm j P X C m < 0
:

(4.41)

208

.X / is the smallest amount of capital which, if added

to X and invested in the risk-free asset, keeps the probability of a negative outcome

below the level . However, Value at Risk only controls the probability of a loss; it

does not capture the size of such a loss if it occurs. Clearly, V@R

is a monetary risk

measure on X D L0 , which is positively homogeneous; see also Example 4.11. The

following example shows that the acceptance set of V@R

is typically not convex,

and so V@R

is not a convex risk measure. Thus, V@R

may penalize diversification

instead of encouraging it.

Example 4.46. Consider an investment into two defaultable corporate bonds, each

with return rQ > r, where r 0 is the return on a riskless investment. The discounted

net gain of an investment w > 0 in the i th bond is given by

w

in case of default,

Xi D w. rQ r/

otherwise.

1Cr

If a default of the first bond occurs with probability p
, then

i

h

w. rQ r/

< 0 D P 1st bond defaults D p
:

P X1

1Cr

Hence,

w. rQ r/

< 0:

1Cr

This means that the position X1 is acceptable in the sense that is does not carry a

positive Value at Risk, regardless of the possible loss of the entire investment w.

Diversifying the portfolio by investing the amount w=2 into each of the two bonds

leads to the position Y WD .X1 C X2 /=2. Let us assume that the two bonds default

independently of each other, each of them with probability p. For realistic r,

Q the

probability that Y is negative is equal to the probability that at least one of the two

bonds defaults: P Y < 0 D p.2 p/. If, for instance, p D 0:009 and
D 0:01

then we have p <
< p.2 p/, hence

w

rQ r

V@R

.Y / D 1

:

2

1Cr

V@R

.X1 / D

Typically, this value is close to one half of the invested capital w. In particular, the

acceptance set of V@R

is not convex. This example also shows that V@R may

strongly discourage diversification: It penalizes quite drastically the increase of the

probability that something goes wrong, without rewarding the significant reduction

of the expected loss conditional on the event of default. Thus, optimizing a portfolio

with respect to V@R

may lead to a concentration of the portfolio in one single asset

with a sufficiently small default probability, but with an exposure to large losses. }

209

Exercise 4.4.1. Let .Yn / be a sequence of independent and identical distributed random variables in L1 .; F ; P /. Show that

V@R

n

1 X

Yi ! E Y1 as n " 1

i D1

for any
2 .0; 1/. Choose the common distribution in such a way that convexity is

violated for large n, i.e.,

V@R

n

1 X

Yi > V@R

.Y1 /:

i D1

In the remainder of this section, we will focus on monetary measures of risk which,

in contrast to V@R

, are convex or even coherent on X WD L1 . In particular, we are

looking for convex risk measures which come close to V@R

. A first guess might

be that one should take the smallest convex measure of risk, continuous from above,

which dominates V@R

. However, since V@R

itself is not convex, the following

proposition shows that such a smallest V@R

-dominating convex risk measure does

not exist.

Proposition 4.47. For each X 2 X and each
2 .0; 1/,

V@R

.X / D min.X / j is convex, continuous from above, and V@R

:

Proof. Let q WD V@R

.X / D qXC .
/ so that P X < q
. If A 2 F satisfies

P A >
, then P A \ X q > 0. Thus, we may define a measure QA by

QA WD P j A \ X q :

It follows that EQA X q D V@R

.X /.

Let Q WD QA j P A >
, and use this set to define a coherent risk measure

via

.Y / WD sup EQ Y :

Q2Q

.X /. Hence, the assertion will follow if we can show that

.Y / V@R

.Y / for each Y 2 X. Let " > 0 and A WD Y V@R

.Y / C ".

Clearly P A > , and so QA 2 Q. Moreover, QA A D 1, and we obtain

.Y / EQA Y V@R

.Y / ":

Since " > 0 is arbitrary, the result follows.

For the rest of this section, we concentrate on the following risk measure which

is defined in terms of Value at Risk, but does satisfy the axioms of a coherent risk

measure.

210

given by

Z

1

AV@R

.X / D

V@R .X / d:

0

Sometimes, the Average Value at Risk is also called the Conditional Value at

Riskor the expected shortfall, and one writes CV@R

.X / or ES

.X /. These

terms are motivated by formulas (4.44) and (4.42) below, but they are potentially misleading: Conditional Value at Risk might also be used to denote the Value at Risk

with respect to a conditional distribution, and expected shortfall might be understood as the expectation of the shortfall X . For these reasons, we prefer the term

Average Value at Risk. Note that

Z

1

AV@R

.X / D

qX .t / dt

0

by (4.41). In particular, the definition of AV@R

.X / makes sense for any X 2

L1 .; F ; P / and we have, in view of Lemma A.19,

Z 1

AV@R1 .X / D

qXC .t / dt D E X :

0

.X / when X is

(a) uniform,

(b) normally distributed,

(c) log-normally distributed, i.e., X D e ZCm with Z N.0; 1/ and m; 2 R.

Recalling the results from Example 4.11 and Exercise 4.1.5, compare the behavior of

V@R

.X / and AV@R

.X / as the parameter increases from 0 to 1.

}

Remark 4.49. Theorem 2.57 shows that the partial order <uni on probability measures

on R with finite mean can be characterized in terms of Average Value at Risk

<uni AV@R

.X / AV@R

.X /

and .

lim V@R

.X / D ess inf X D infm j P X C m < 0 0:

#0

which is the worst-case risk measure on L1 introduced in Example 4.39. Recall that

it is continuous from above but in general not from below.

}

211

AV@R

.X / D

1

1

E .q X /C q D

inf .E .r X /C
r/:

r2R

(4.42)

Z

Z

1

1 1

1

C

C

.qqX .t // dt q D

qX .t / dt D AV@R

.X /:

E .qX / q D

0

0

This proves the first identity. The second one follows from Lemma A.22.

Theorem 4.52. For
2 .0; 1, AV@R

is a coherent risk measure which is continuous from below. It has the representation

AV@R

.X / D max EQ X ;

Q2Q

X 2 X;

(4.43)

where Q

is the set of all probability measures Q

P whose density dQ=dP is

P -a.s. bounded by 1=
. Moreover, Q

is equal to the maximal set Qmax of Corollary 4:37.

Proof. Since Q1 D P , the assertion is obvious for
D 1. For 0 <
< 1, consider

the coherent risk measure

.X / WD supQ2Q EQ X . First we assume that we

are given some X < 0. We define a measure PQ P by d PQ =dP D X=E X . Then

.X / D

E X

Q ' j 0 ' 1; E ' D
:

supE

Clearly, the condition E ' D
on the right can be replaced by E '
. Thus,

we can apply the NeymanPearson lemma in the form of Theorem A.31 and conclude

that the supremum is attained by

'0 D IX <q C IXDq

for a
-quantile q of X and some 2 0; 1 for which E '0 D
. Hence,

.X / D

E X Q

1

E '0 D E X '0 :

, we conclude that

.X / D max EQ X D EQ0 X

Q2Q

1

.E X I X < q q
C qP X < q /

1

D E .q X /C q

D AV@R

.X /;

212

where we have used (4.42) in the last step. This proves (4.43) for X < 0. For arbitrary

X 2 L1 , we use the cash invariance of both

and AV@R

.

It remains to prove that Q

is the maximal set of Corollary 4.37. This follows from

Exercise 4.3.3 and Example 4.40, but we also give a different argument here. To this

end, we show that

sup .EQ X AV@R

.X // D C1

X2X

for Q Q

. We denote by ' the density dQ=dP . There exist
0 2 .0;
/ and

k > 1=
0 such that P ' ^ k 1=
0 > 0. For c > 0 define X .c/ 2 X by

X .c/ WD c.' ^ k/I'1=0 :

Since

PX

.c/

1

< 0 D P ' 0

0 < ;

we have V@R

.X .c/ / D 0, and (4.42) yields that

AV@R

.X

.c/

1

c

1

.c/

/ D E X D E ' ^ kI ' 0 :

EQ X

.c/

1

D c E ' ' ^ kI ' 0

1

c

0 E ' ^ kI ' 0 :

.X .c/ / becomes arbitrarily

large as c " 1.

Remark 4.53. The proof shows that for 2 .0; 1/ the maximum in (4.43) is attained

by the measure Q0 2 Q

, whose density is given by

dQ0

1

D .IX <q C IXDq /;

dP

where q is a -quantile of X, and where is defined as

0

if P X D q D 0,

WD

P X<q

otherwise.

P XDq

AV@R

.X / WCE

.X /

E X j X V@R

.X /

V@R

.X /;

(4.44)

213

where WCE

is the coherent risk measure defined in (4.38). Moreover, the first two

inequalities are in fact identities if

P X qXC .
/ D
;

(4.45)

Proof. If P A , then the density P j A with respect to P is bounded by 1= .

Therefore, Theorem 4.52 implies that AV@R

dominates WCE

. Since

P X V@R

.X / " > ;

we have

WCE

.X / E X j X V@R

.X / " ;

and the second inequality follows by taking the limit " # 0. Moreover, (4.42) shows

that

AV@R

.X / D E X j X V@R

.X /

as soon as (4.45) holds.

Remark 4.55. We will see in Corollary 4.68 that the two coherent risk measures

AV@R

and WCE

coincide if the underlying probability space is rich enough. If

this is not the case, then the first inequality in (4.44) may be strict for some X; see [2].

Moreover, the functional

E X j X V@R

.X /

does not define a convex risk measure. Hence, the second inequality in (4.44) cannot

reduce to an identity in general.

}

Remark 4.56. We have seen in Proposition 4.47 that there is no smallest convex risk

measure dominating V@R

. But if we restrict our attention to the class of convex

risk measures that dominate V@R

and only depend on the distribution of a random

variable, then the situation is different. In fact, we will see in Theorem 4.67 that

AV@R

is the smallest risk measure in this class, provided that the underlying probability space is rich enough. In this sense, Average Value at Risk can be regarded as

the best conservative approximation to Value at Risk.

}

4.5

Clearly, V@R

and AV@R

only involve the distribution of a position under the

given probability measure P . In this section we study the class of all risk measures

which share this property of law-invariance.

214

Definition 4.57. A monetary risk measure on X D L1 .; F ; P / is called lawinvariant if .X / D .Y / whenever X and Y have the same distribution under P .

Throughout this section, we assume that the probability space .; F ; P / is rich

enough in the sense that it supports a random variable with a continuous distribution.

This condition is satisfied if and only if .; F ; P / is atomless; see Proposition A.27.

Remark 4.58. Any law-invariant monetary risk measure is monotone with respect

to the partial order <mon introduced in Definition 2.67. More precisely,

<mon

H)

.X / .X /;

and . To prove this, let q

and q be quantile functions for

and and take a random variable U with a uniform

distribution on .0; 1/. Then XQ WD q .U / q .U / DW XQ by Theorem 2.68, and

XQ and XQ have the same distribution as X and X by Lemma A.19. Hence, law}

invariance and monotonicity of imply .X / D .XQ / .XQ / D .X /.

We can now formulate our first structure theorem for law-invariant convex risk

measures.

Theorem 4.59. Let be a convex risk measure and suppose that is continuous from

above. Then is law-invariant if and only if its minimal penalty function min .Q/

under P when Q 2 M1 .P /. In this case,

depends only on the law of 'Q WD dQ

dP

has the representation

Z 1

.X / D sup

qX .t /q'Q .t / dt min .Q/ ;

(4.46)

Q2M1 .P /

Z 1

min

.Q/ D sup

qX .t /q'Q .t / dt

X2A

D sup

X2L1

Z

1

0

(4.47)

For the proof, we will need the following lemma. Here and in the sequel we will

write X XQ to indicate that the two random variables X and XQ have the same law.

Lemma 4.60. For two random variables X and Y ,

Z 1

Q ;

qX .t /qY .t / dt D max E XY

0

Q

X

X

215

prove the converse inequality, take a random variable U with a uniform distribution on .0; 1/ such that Y D qY .U / P -a.s. Such a random variable U exists by

Lemma A.28 and our assumption that the underlying probability space is atomless.

Since XQ WD qX .U / X by Lemma A.19, we obtain

Z 1

Q

qX .t /qY .t / dt;

E X Y D E qX .U /qY .U / D

0

and hence .

Proof of Theorem 4:59. Suppose first that is law-invariant. Then X 2 A implies

that XQ 2 A for all XQ X . Hence,

min .Q/ D sup E X 'Q D sup sup E XQ 'Q

X2A

X2A X

X

Q

D sup

X2A

where we have used Lemma 4.60 in the last step. It follows that min .Q/ depends

only on the law of 'Q . In order to check the second identity in (4.47), note that

XQ WD X C .X / belongs to A for any X 2 L1 and that qX .X / is a quantile

Q

function for X.

Conversely, let us assume that min .Q/ depends only on the law of 'Q . Let us write

QQ Q to indicate that 'Q and 'QQ have the same law. Then Lemma 4.60 yields

.X / D

sup

Q2M1 .P /

sup

Q2M1 .P / Q

Q

Q

sup

Q2M1 .P /

Z

1

0

qX .t /q'Q .t / dt min .Q/ :

Exercise 4.5.1. For 0 < < 1, start from the definition of Average Value at Risk as

dQ

1

AV@R

.X / WD sup EQ X j Q 2 M1 .P / and

P -a.s. ;

dP

check that the conditions of Theorem 4.59 are satisfied, and deduce from (4.46) that

the representation

Z

1

AV@R

.X / D

V@R .X / d

0

holds.

216

a position X 2 L1 is acceptable if E u.X / c, where c is a given constant in the

interior of u.R/. We have seen in Example 4.10 that the corresponding acceptance set

induces a convex risk measure . Clearly, is law-invariant, and it will be shown in

Proposition 4.113 that is continuous from below and, hence, from above. Moreover,

the corresponding minimal penalty function can be computed as

min

1

.Q/ D inf

>0

cC

0

` .
q'Q .t // dt ;

where

` .y/ D sup .xy C u.x// D sup .xy `.x//

x2R

x2R

u.x/; see Theorem 4.115.

}

The following theorem clarifies the crucial role of the risk measures AV@R

: they

can be viewed as the building blocks for law-invariant convex risk measures on L1 .

Recall that we assume that .; F ; P / is atomless.

Theorem 4.62. A convex risk measure is law-invariant and continuous from above

if and only if

.X / D

sup

2M1 ..0;1/

Z

.0;1

AV@R

.X /

.d
/ min .

/ ;

where

min

(4.48)

Z

.

/ D sup

X2A

.0;1

AV@R

.X /

.d
/:

Proof. Clearly, the right-hand side of (4.48) defines a law-invariant convex risk measure that is continuous from above. Conversely, let be law-invariant and continuous

from above. We will show that for Q 2 M1 .P / there exists a measure

2 M1 ..0; 1/

such that

Z 1

Z

qX .t /q' .t / dt D

AV@Rs .X /

.ds/;

0

.0;1

where ' WD 'Q D dQ

dP

qX .t / D V@R1t .X / and q' .t / D q'C .t / for a.e. t 2 .0; 1/,

Z

qX .t /q' .t / dt D

0

217

Since q'C is increasing and right-continuous, we can write q'C .t / D .1 t; 1/ for

some positive locally finite measure on .0; 1. Moreover, the measure

given by

Z 1

Z 1

Z

t .dt / D

.s; 1/ ds D

q'C .s/ ds D E ' D 1:

.0;1

Thus,

Z

1

0

qX .t /q' .t / dt D

V@R t .X /

0

Z

D

.0;1

Z

D

1

s

.t;1

.ds/ dt

s

V@R t .X / dt

.ds/

(4.49)

AV@Rs .X /

.ds/:

.0;1

Conversely,

for any probability measure

on .0; 1, the function q defined by q.t / WD

R

1

.ds/ can be viewed as a quantile function of the density ' WD q.U / of

s

.1t;1

a measure Q 2 M1 .P /, where U has a uniform distribution on .0; 1/. Altogether,

we obtain a one-to-one correspondence between laws of densities ' and probability

measures

on .0; 1.

Theorem 4.62 takes the following form for coherent risk measures.

Corollary 4.63. A coherent risk measure is continuous from above and law-invariant if and only if

Z

.X / D sup

2M .0;1

AV@R

.X /

.d
/

Remark 4.64. In the preceding results of this section, it was assumed that is lawinvariant and continuous from above and that the underlying probability space is atomless. Under the additional regularity assumption that L2 .; F ; P / is separable, it can

be shown that every law-invariant convex risk measure is continuous from above;

see [161].

}

Randomness of a position is reduced in terms of P if we replace the position by its

conditional expectation with respect to some -algebra G F . Such a reduction of

randomness is reflected by a convex risk measure if it is law-invariant.

Corollary 4.65. Assume that is a convex risk measure which is continuous from

above and law-invariant. Then is monotone with respect to the binary relation <uni

introduced in (3.27):

Y <uni X

H)

.Y / .X /;

218

for Y; X 2 X. In particular,

.E X j G / .X /;

for X 2 X and any -algebra G F , and

.E X / D .0/ E X .X /:

Proof. The first inequality follows from Theorem 4.62 combined with Remark 4.49

The second inequality is a special case of the first one, since E X jG <uni X according to Theorem 2.57. The third follows from the second by taking G D ;; .

Recall from Theorem 2.68 that

<mon implies

<uni . Thus, the preceding conclusion for convex risk measures is stronger than the one of Remark 4.58 for monetary

risk measures.

Remark 4.66. If G1 G2 F are -algebras, then

.E X j Gn / ! .E X j G1 / as n " 1,

S

where is as in Corollary 4.65 and G1 D . n Gn /. Indeed, Doobs martingale

convergence theorem (see, e.g., Theorem 19.1 in [20]) states that E X j Gn !

E X j G1 P -a.s. as n " 1. Hence, the Fatou property and Corollary 4.65 show

that

.E X j G1 / D lim E X j Gn

n"1

n"1

.E X j G1 /:

is the

best conservative approximation to V@R

in the class of all law-invariant convex

risk measures which are continuous from above.

Theorem 4.67. AV@R

is the smallest law-invariant convex measure of risk which

is continuous from above and dominates V@R

.

Proof. That AV@R

dominates V@R

was already stated in (4.44). Suppose now

that is another law-invariant convex risk measure which dominates V@R

and

which is continuous from above. We must show that for a given X 2 X

.X / AV@R

.X /:

(4.50)

219

By cash invariance, we may assume without loss of generality that X > 0. Take

" > 0, and let A WD X V@R

.X / " and

Y WD E X j XIAc D X IAc C E X j A IA :

Since Y > qXC .
/ C " E X j A on Ac , we get P Y < E X j A D 0. On the

other hand, P Y E X j A P A >
, and this implies that V@R

.Y / D

E X j A . Since dominates V@R

, we have .Y / E X j A . Thus,

.X / .Y / D E X j X V@R

.X / " ;

by Corollary 4.65. Taking " # 0 yields

.X / E X j X V@R

.X / :

If the distribution of X is continuous, Corollary 4.54 states that the conditional expectation on the right equals AV@R

.X /, and we obtain (4.50). When the distribution of

X is not continuous, we denote by D the set of all points x such that P X D x > 0

and take any bounded random variable Z 0 with a continuous distribution. Such a

random variable exists due to our assumption that .; F ; P / is atomless. Note that

Xn WD X C n1 ZIX 2D has a continuous distribution. Indeed, for any x,

X

P X D y; Z D n.x y/ D 0:

P Xn D x D P X D x; X D C

y2D

Moreover, Xn decreases to X. The inequality (4.50) holds for each Xn and extends

to X by continuity from above.

Corollary 4.68. AV@R

and WCE

coincide under our assumption that the probability space is atomless.

Proof. We know from Corollary 4.54 that WCE

.X / D AV@R

.X / if X has a continuous distribution. Repeating the approximation argument at the end of the preceding proof yields WCE

.X / D AV@R

.X / for each X 2 X.

4.6

Concave distortions

Z

.X / WD AV@R

.X /

.d /;

(4.51)

which appear in the Representation Theorem 4.62 for law-invariant convex risk measures. We are going to characterize these risk measures in two ways, first as Choquet integrals with respect to some concave distortion of the underlying probability

measure P , and then, in the next section, by a property of comonotonicity.

220

Again, we will assume throughout this section that the underlying probability space

.; F ; P / is atomless. Since AV@R

is coherent, continuous from below, and lawinvariant, any mixture for some probability measure

on .0; 1 has the same properties. According to Remark 4.50, we may set AV@R0 .X / D ess inf X so that we

can extend the definition (4.51) to probability measures

on the closed interval 0; 1.

However, will only be continuous from above and not from below if

.0/ > 0,

because AV@R0 is not continuous from below.

Our first goal is to show that .X / can be identified with the Choquet integral

of the loss X with respect to the set function c .A/ WD .P A /, where is the

concave function defined in the following lemma. Choquet integrals were introduced

in Example 4.14, and the risk measure MINVAR of Exercise 4.1.7 provides a first

example for a risk measure arising as the Choquet integral of a set function c . Recall

0 ; see

that every concave function admits a right-continuous right-hand derivative C

Proposition A.4.

Lemma 4.69. The identity

0

C .t /

s 1

.ds/;

0 < t < 1;

(4.52)

.t;1

on 0; 1 and

increasing concave functions

W 0; 1 ! 0; 1 with .0/ D 0 and .1/ D 1.

Moreover, we have .0C/ D

.0/.

Proof. Suppose first that

is given and is defined by

is concave and increasing on .0; 1. Moreover,

Z

1

.0C/ D

Z

.t / dt D

.0;1

1

s

1

0

It <s1 dt

.ds/ D

..0; 1/ 1:

Hence, we may set .0/ WD 0 and obtain an increasing concave function on 0; 1.

0 .t / is a decreasing right-continuous function on

Conversely, if is given, then C

0

.0; 1/ and can be written as C .t / D ..t; 1/ for some locally finite positive measure

on .0; 1. We first define

on .0; 1 by

.dt / D t .dt /. Then (4.52) holds and, by

Fubinis theorem,

Z

1Z

..0; 1/ D

0

Hence, setting

.0/ WD

.0;1

It <s .ds/ dt D 1

.0C/ 1:

on 0; 1.

221

on 0; 1, let be the concave function

defined in Lemma 4:69. Then, for X 2 X,

Z 1

qX .t / 0 .1 t / dt

(4.53)

.X / D .0C/AV@R0 .X / C

0

. .P X > x / 1/ dx C

D

1

.P X > x / dx:

0

.X / D qX .1 /, we get as in (4.49) that

Z 1

Z

AV@R

.X /

.d / D

qX .t / 0 .1 t / dt:

0

.0;1

Hence, we obtain the first identity. For the second one, we will first assume X 0.

Then

Z 1

qXC .t / D supx 0 j FX .x/ t D

Z 1Z 1

Z 1

0

qX .t / .1 t / dt D

IFX .x/1t 0 .t / dt dx

0

.1 FX .x// dx

Ry

since 0 0 .t / dt D . .y/ .0C//Iy>0 . This proves the second identity for

X 0, since .0C/ D

.0/ and ess sup X D AV@R0 .X /. If X 2 L1 is

arbitrary, we consider X C C , where C WD ess inf X . The cash invariance of

yields

Z 1

.P X > x C / dx

C C .X / D

Z

.P X > x / dx C

D

C

.P X > x / dx

0

. .P X > x / 1/ dx C

DC C

1

.P X > x / dx:

0

is itself of the form where

D

t

1

^ 1 D .t ^ /:

.t / D

Thus, we obtain yet another representation of AV@R

:

Z

1 1

AV@R

.X / D

P X > x ^ dx for X 2 L1

}

C.

0

222

on 0; 1 such that the coherent risk

measure MINVAR introduced in Exercise 4.1.7 is of the form (4.51).

}

Exercise 4.6.2. Suppose that there exists a set A 2 F with 0 < P A < 1 such

that .IA / D .IA /. Use the representation (4.53) to deduce that

D 1 , i.e.,

.X / D E X for all X 2 L1 . More generally, show that

D 1 if there exists

a nonconstant X 2 L1 such that .X / D .X /.

}

Corollary 4.72. If

.0/ D 0 in Theorem 4.70, then

Z 1

qX .'.t // dt;

.X / D

0

Proof. Due to Lemma A.15, the distribution of ' under the Lebesgue measure has the

distribution function and hence the density 0 . Therefore

Z

qX .'.t // dt D

0

qX .t /

.t / dt D

qX .1 t /

.t / dt;

where we have used Lemma A.23 in the last step. An application of Theorem 4.70

concludes the proof.

Let us continue with a brief discussion of the set function c .A/ D

.P A /.

and .1/ D 1. The set function

c .A/ WD

.P A /;

.0/ D 0

A2F;

is called the distortion of the probability measure P with respect to the distortion

function .

Definition 4.74. A set function c W F ! 0; 1 is called monotone if

c.A/ c.B/ for A B

and normalized if

c.;/ D 0

and

c./ D 1:

c.A [ B/ C c.A \ B/ c.A/ C c.B/:

Clearly, any distortion c is normalized and monotone.

223

Proposition 4.75. Let c be the distortion of P with respect to the distortion function . If is concave, then c is submodular. Moreover, if the underlying probability space is atomless, then also the converse implication holds.

Proof. Suppose first that is concave. Take A; B 2 F with P A P B . We

must show that c WD c satisfies

c.A/ c.A \ B/ c.A [ B/ c.B/:

This is trivial if r D 0, where

r WD P A P A \ B D P A [ B P B :

For r > 0 the concavity of

c.A/ c.A \ B/

c.A [ B/ c.B/

:

PA PA \ B

PA [ B PB

Multiplying both sides with r gives the result.

Now suppose that c D c is submodular and assume that .; F ; P / is atomless.

By Exercise A.1.1 below it is sufficient to show that .y/ . .x/ C .z//=2 whenever 0 x z 1 and y D .x C z/=2. To this end, we will construct two sets

A; B F such that P A D P B D y, P A \ B D x, and P A [ B D z.

Submodularity then gives .x/ C .z/ 2 .y/ and in turn the concavity of .

In order to construct the two sets A and B, take a random variable U with a uniform

distribution on 0; 1, which exists by Proposition A.27. Then

A WD 0 U y

and

B WD z y U z

are as desired.

Let us now recall the notion of a Choquet integral, which was introduced in Example 4.14.

Definition 4.76. Let c W F ! 0; 1 be any set function which is normalized and

monotone. The Choquet integral of a bounded measurable function X on .; F /

with respect to c is defined as

Z

.c.X > x/ 1/ dx C

X dc WD

1

0

Note that the Choquet integral coincides with the usual integral as soon as c is a

-additive probability measure; see also Lemma 4.97 below. With this definition,

Theorem 4.70 allows us to identify the risk measure as the Choquet integral of the

loss with respect to a concave distortion c of the underlying probability measure P .

224

on 0; 1, let be the concave distortion

function defined in Lemma 4:69, and let c denote the distortion of P with respect

to . Then, for X 2 L1 ,

Z

.X / D

.X / dc :

Combining Corollary 4.77 with Theorem 4.62, we obtain the following characterization of law-invariant convex risk measures in terms of concave distortions:

Corollary 4.78. A convex risk measure is law-invariant and continuous from above

if and only if

Z

.X / D sup

.X / dc min . / ;

where the supremum is taken over the class of all concave distortion functions

Z

min

. / WD sup

.X / dc :

and

X2A

The following series of exercises should be compared with Exercises 4.1.7 and

4.6.1.

Exercise 4.6.3. For 1 consider the concave distortion function

Show that for 2 N the corresponding risk measure

Z

MAXVAR .X / WD .X / dc

.x/

WD x .

identically distributed (i.i.d.) random variables for which max.Y1 ; : : : ; Y / X. }

1

Show that for 2 N the corresponding risk measure

Z

MAXMINVAR .X / WD .X / dc Q

has the property that MAXMINVAR .X / D E Y1 , if Y1 ; : : : ; Y are i.i.d. random variables for which max.Y1 ; : : : ; Y / min.X1 ; : : : ; X /, when X1 ; : : : ; X

are independent copies of X.

}

1

Show that for 2 N the corresponding risk measure

Z

MINMAXVAR .X / WD .X / dc O

has the property that MINMAXVAR .X / D E min.Y1 ; : : : ; Y / , if Y1 ; : : : ; Y

are i.i.d. random variables for which max.Y1 ; : : : ; Y / X .

}

225

maximal representing set Q M1 .P / for the coherent risk measure .

Theorem 4.79. Let

be a probability measure on 0; 1, and let be the corresponding concave function defined in Lemma 4:69. Then can be represented as

.X / D sup EQ X ;

Q2Q

Z 1

dQ

qZ .s/ ds

satisfies

Q WD Q 2 M1 .P / Z WD

dP

t

.1 t / for t 2 .0; 1/ :

Proof. The risk measure is coherent and continuous from above. By Corollary 4.37, it can be represented by taking the supremum of expectations over the set

Qmax D Q 2 M1 .P / j min .Q/ D 0. Using (4.47) and Theorem 4.70, we see that

a measure Q 2 M1 .P / with density Z D dQ=dP belongs to Qmax if and only if

Z 1

qX .s/qZ .s/ ds .X /

0

(4.54)

Z

1

.0C/AV@R0 .X / C

qX .s/

.1 s/ ds

for all X 2 L1 . For constant random variables X t , we have qX D It;1 a.e., and

so we obtain

Z 1

Z 1

0

qZ .s/ ds .0C/ C

.1 s/ ds D .1 t /

t

Qmax

Q . For the proof of the converse inclusion, we

show that the density Z of a fixed measure Q 2 Q satisfies (4.54) for any given

X 2 L1 . We may assume without loss of generality that X 0. Let be the

positive finite measure on 0; 1 such that qXC .s/ D .0; s/. Using Fubinis theorem

and the definition of Q , we get

Z 1

Z 1

Z

qX .s/qZ .s/ ds D

qZ .s/ ds .dt /

0

0;1

.1 t / .dt /

0;1

0

Z

.1 s/

.dt / ds;

0;s

226

Corollary 4.80. In the context of Theorem 4.79, the following conditions are equivalent:

(a) is continuous from below.

(b)

.0/ D 0.

(c) .X / D maxQ2Q EQ X for all X 2 L1 .

If these equivalent conditions are satisfied, then the maximum in (c) is attained by

the measure QX 2 Q with density dQX =dP D f .X /, where f is the decreasing

function defined by

f .x/ WD 0 .FX .x//

if x is a continuity point of FX , and by

f .x/ WD

1

FX .x/ FX .x/

FX .x/

.t / dt

FX .x/

dQ 1

Q D Q

P P

<uni
.

dP

0 1

(4.55)

Proof. The equivalence of conditions (a) and (c) has already been proved in Corollary 4.38. If (b) holds, then is continuous from below, due to Theorem 4.52

and monotone convergence. Let us now show that condition (a) is not satisfied if

WD

.0/ > 0. In this case, we can write

D AV@R0 C .1 /0 ;

where

0 WD

. j.0; 1/. Then 0 is continuous from below since

0 .0/ D 0, but

AV@R0 is not, and so does not satisfy (a); see Remark 4.50.

Let us now prove the remaining assertions. Since .0C/ RD

.0/ D 0, Ra measure

1

1

Q with density Z D dQ=dP belongs to Q if and only if t qZ .s/ ds t 0 .1

0 .1 t / is a quantile function for the law of

0 under
,

s/ ds for all t . Since

part (e) of Theorem 2.57 implies (4.55). The problem of identifying the maximizing

measure QX is hence equivalent to minimizing E ZX under the constraint that Z

is a density function such that P Z 1 <uni
. 0 /1 . Let us first assume that

X 0. Then it follows from Theorem 3.44 that f .X / minimizes E YX among

all Y 2 L1C such that P Y 1 <uni
. 0 /1 . Moreover, Remark 3.46 shows that

R1

E f .X / D 0 0 .t / dt D 1, and so ZX WD f .X / 0 is the density of an optimal

probability measure QX 2 Q . If X is not positive, then we may take a constant

c such that X C c 0 and apply the preceding argument. The formula for f then

follows from the fact that FXCc .X C c/ D FX .X /.

227

gate function as

'.x/ WD sup . .y/ xy/;

x 0:

y20;1

dQ

satisfies E .Z x/C '.x/ for x 0 : }

Q D Q 2 M1 .P / Z WD

dP

Remark 4.81. As long as we are interested in a law-invariant risk assessment, we can

represent a financial position X 2 L1 by its distribution function FX or, equivalently,

by the function

GX .t / WD 1 FX .t / D P X > t :

If we only consider positions X with values in 0; 1 then their proxies GX vary in

the class of right-continuous decreasing functions G on 0; 1 such that G.1/ D 0 and

G.0/ 1. Due to Theorem 4.70, a law-invariant coherent risk measure induces a

functional U on the class of proxies via

Z

U.GX / WD .X / D

.GX .t // dt:

0

Since is increasing and concave, the functional U has the form of a von Neumann

Morgenstern utility functional on the probability space given by Lebesgue measure on

the unit interval 0; 1. As such, it can be characterized by the axioms in Section 2.3,

and this is the approach taken in Yaaris dual theory of choice [265]. More generally, we can introduce a utility function u on 0; 1 with u.0/ D 0 and consider the

functional

Z 1

.GX .t // du.t /

U.GX / WD

0

introduced by Quiggin [216]. For u.x/ D x this reduces to the dual theory, for

.x/ D x we recover the classical utility functionals

Z

U.GX / D

GX .t / du.t /

0

u.t / dGX .t /

D

0

D E u.X /

discussed in Section 2.3.

228

4.7

In many situations, the risk of a combined position X C Y will be strictly lower than

the sum of the individual risks, because one position serves as a hedge against adverse

changes in the other position. If, on the other hand, there is no way for X to work

as a hedge for Y then we may want the risk simply to add up. In order to make

this idea precise, we introduce the notion of comonotonicity. Our main goal in this

section is to characterize the class of all convex risk measures that share this property

of comonotonicity.

As in the first two sections of this chapter, we will denote by X the linear space of

all bounded measurable functions on the measurable space .; F /.

Definition 4.82. Two measurable functions X and Y on .; F / are called comonotone if

.X.!/ X.! 0 //.Y .!/ Y .! 0 // 0

(4.56)

.X C Y / D .X / C .Y /

whenever X; Y 2 X are comonotone.

Lemma 4.83. If is a comonotonic monetary risk measure on X, then is positively

homogeneous.

Proof. Note that .X; X / is a comonotone pair. Hence .2X / D 2.X /. An iteration

of this argument yields .rX / D r.X / for all rational numbers r 0. Positive

homogeneity now follows from the Lipschitz continuity of ; see Lemma 4.3.

We will see below that every comonotonic monetary risk measure on X arises as

the Choquet integral with respect to a certain set function on .; F /. In the sequel,

c W F ! 0; 1 will always denote a set function that is normalized and monotone; see

Definition 4.74. Unless otherwise mentioned, we will not assume that c enjoys any

additivity properties. Recall from Definition 4.76 that the Choquet integral of X 2 X

with respect to c is defined as

Z 1

Z

Z 0

.c.X > x/ 1/ dx C

c.X > x/ dx:

X dc D

1

The proof of the following proposition was already given in Example 4.14.

Proposition 4.84. The Choquet integral of the loss,

Z

.X / WD .X / dc;

is a monetary risk measure on X which is positively homogeneous.

229

rX W .0; 1/ ! R of the increasing function GX .x/ WD 1 c.X > x/, taken in the

sense of Definition A.14, is called a quantile function for X with respect to c.

If c is a probability measure, then GX .x/ D c.X x/. Hence, the preceding

definition extends the notion of a quantile function given in Definition A.20. The

following proposition yields an alternative representation of the Choquet integral in

terms of quantile functions with respect to c.

Proposition 4.86. Let rX be a quantile function with respect to c for X 2 X. Then

Z

Z 1

X dc D

rX .t / dt:

0

rX C m a.e. for all m 2 R and each quantile function rXCm of X C m. Thus, we

may assume without loss of generality that X 0. In this case, Remark A.16 and

Lemma A.15 imply that the largest quantile function rXC is given by

Z 1

C

IGX .x/t dx:

rX .t / D supx 0 j GX .x/ t D

0

Z

1Z 1

rX .t / dt D

0

IGX .x/t dx dt

.1 GX .x// dx

Z

X dc:

when applied to a continuous distortion of a probability measure as defined in Definition 4.73.

Corollary 4.87. Let c .A/ D .P A / be the distortion of the probability measure

P with respect to the continuous distortion function . If ' is an inverse function for

the increasing function in the sense of Definition A.14, then the Choquet integral

with respect to c satisfies

Z

X dc D

qX .1 '.t // dt;

0

230

infx j .x/ > t . Thus, we can compute the lower quantile function of X with

respect to c

rX .t / D infx 2 R j 1 c .X > x/ t

D infx 2 R j .P X > x / 1 t

D infx 2 R j P X > x ' C .1 t /

D qX .1 ' C .1 t //:

Next note that ' C .t / D '.t / for a.e. t . Moreover, ' has the continuous distribution

function under the Lebesgue measure, and so we can replace qX by the arbitrary

quantile function qX .

Theorem 4.88. A monetary risk measure on X is comonotonic if and only if there

exists a normalized monotone set function c on .; F / such that

Z

.X / D .X / dc; X 2 X:

In this case, c is given by c.A/ D .IA /.

The preceding theorem implies in view of Corollary 4.77 that all mixtures

Z

D

AV@R

.d /

0;1

are comonotonic. We will see in Theorem 4.93 below that these are in fact all convex

risk measures that are law-invariant and comonotonic. The proof of Theorem 4.88

requires a further analysis of comonotone random variables.

Lemma 4.89. Two measurable functions X and Y on .; F / are comonotone if and

only if there exists a third measurable function Z on .; F / and increasing functions

f and g on R such that X D f .Z/ and Y D g.Z/.

Proof. Clearly, X WD f .Z/ and Y WD g.Z/ are comonotone for given Z, f , and g.

Conversely, suppose that X and Y are comonotone and define Z by Z WD X C Y .

We show that z WD Z.!/ has a unique decomposition as z D x C y, where .x; y/ D

.X.! 0 /; Y .! 0 // for some ! 0 2 . Having established this, we can put f .z/ WD x

and g.z/ WD y. The existence of the decomposition as z D x C y follows by taking

x WD X.!/ and y WD Y .!/, so it remains to show that these are the only possible

values x and y. To this end, let us suppose that X.!/ C Y .!/ D z D X.! 0 / C Y .! 0 /

for some ! 0 2 . Then

X.!/ X.! 0 / D .Y .!/ Y .! 0 //;

231

and comonotonicity implies that this expression vanishes. Hence x D X.! 0 / and

y D Y .! 0 /.

Next, we check that both f and g are increasing functions on Z./. So let us

suppose that

X.!1 / C Y .!1 / D z1 z2 D X.!2 / C Y .!2 /:

This implies

X.!1 / X.!2 / .Y .!1 / Y .!2 //:

Comonotonicity thus yields that X.!1 /X.!2 / 0 and Y .!1 /Y .!2 / 0, whence

f .z1 / f .z2 / and g.z1 / g.z2 /. Thus, f and g are increasing on Z./, and it is

straightforward to extend them to increasing functions defined on R.

Lemma 4.90. If X; Y 2 X is a pair of comonotone functions, and rX , rY , rXCY are

quantile functions with respect to c, then

rXCY .t / D rX .t / C rY .t /

for a.e. t .

Proof. Write X D f .Z/ and Y D g.Z/ as in Lemma 4.89. The same argument as

in the proof of Lemma A.23 shows that f .rZ / and g.rZ / are quantile functions for

X and Y under c if rZ is a quantile function for Z. An identical argument applied to

the increasing function h WD f C g shows that h.rZ / D f .rZ / C g.rZ / is a quantile

function for X CY . The assertion now follows from the fact that all quantile functions

of a random variable coincide almost everywhere, due to Lemma A.15.

Remark 4.91. Applied to the special case of quantile function with respect to a probability measure, the preceding lemma yields that V@R

and AV@R

are comonotonic.

}

Proof of Theorem 4:88. We already know from Proposition 4.84 that the Choquet integral of the loss is a monetary risk measure. Comonotonicity follows by combining

Proposition 4.86 with Lemma 4.90.

Conversely, suppose now that is comonotonic. Then is positively homogeneous

according to Lemma 4.83. In particular we have .m/ D m for m 0. Thus, we

obtain a normalized

monotone set function by letting c.A/ WD .IA /. Moreover,

R

c .X / WD .X / dc is a comonotonic monetary risk measure on X that coincides

with on indicator functions: .IA / D c.A/ D c .IA /. Let us now show that

and c coincide on simple random variables of the form

XD

n

X

xi IAi ;

xi 2 R; Ai 2 F :

i D1

Since these random variables are dense in L1 , Lemma 4.3 will then imply that D

c . In order to show that c .X / D .X / for X as above, we may assume without

232

can write

cash invariance,

we may also assume X 0, i.e., xn 0. Thus, weS

P

X D niD1 bi IBi , where bi WD xi xi C1 0, xnC1 WD 0, and Bi WD ikD1 Ak .

P

Note that bi IBi and bk IBk is a pair of comonotone functions. Hence, also k1

i D1 bi IBi

and bk IBk are comonotone, and we get inductively

.X / D

n

X

bi .IBi / D

i D1

n

X

bi c .IBi / D c .X /:

i D1

Remark 4.92. The argument at the end of the preceding proof shows that the Choquet

integral of a simple random variable

XD

n

X

xi IAi

with x1 xn xnC1 WD 0

i D1

Z

X dc D

n

n

X

X

.xi xi C1 /c.Bi / D

xi .c.Bi / c.Bi 1 //;

i D1

where B0 WD ; and Bi WD

i D1

Si

kD1 Ak

for i D 1; : : : ; n.

So far, we have shown that comonotonic monetary risk measures can be identified

with Choquet integrals of normalized monotone set functions. Our next goal is to

characterize those set functions that induce risk measures with the additional property

of convexity. To this end, we will first consider law-invariant risk measures. The

following result shows that the risk measures AV@R

may be viewed as the extreme

points in the convex class of all law-invariant convex risk measures on L1 that are

comonotonic.

Theorem 4.93. On an atomless probability space, the class of risk measures

Z

.X / WD AV@R

.X /

.d
/;

2 M1 .0; 1/;

is precisely the class of all law-invariant convex risk measures on L1 that are comonotonic. In particular, any convex risk measure that is law-invariant and comonotonic is also coherent and continuous from above.

Proof. Comonotonicity of follows from Corollary 4.77 and Theorem 4.88. Conversely, let us assume that is a law-invariant

convex risk measure that is also coR

monotonic. By Theorem 4.88, .X / D .X / dc for c.A/ WD .IA /. The lawinvariance of implies that c.A/ is a function of the probability P A , i.e., there

233

.P A /. Note that IA[B and IA\B is a pair of comonotone functions for all A; B 2

F . Hence, comonotonicity and subadditivity of imply

c.A \ B/ C c.A [ B/ D .IA\B / C .IA[B / D .IA\B IA[B /

D .IA IB /

(4.57)

c.A/ C c.B/:

Proposition 4.75 thus implies that is concave. Corollary 4.77 finally shows that the

Choquet integral with respect to c can be identified with a risk measure , where

Now we turn to the characterization of all comonotonic convex risk measures on X.

Recall that, for a positively homogeneous monetary risk measure, convexity is equivalent to subadditivity. Also recall that M1;f WD M1;f .; F / denotes the set of all

finitely additive normalized set functions Q W F ! 0; 1, and that EQ X denotes

the integral of X 2 X with respect to Q 2 M1;f , as constructed in Theorem A.51.

Theorem 4.94. For the Choquet integral with respect to a normalized monotone set

function c, the following conditions are equivalent:

R

(a) .X / WD .X / dc is a convex risk measure on X.

R

(b) .X / WD .X / dc is a coherent risk measure on X.

(c) For Qc WD Q 2 M1;f j Q A c.A/ for all A 2 F ,

Z

X dc D max EQ X for X 2 X.

Q2Qc

In this case, Qc is equal to the maximal representing set Qmax for .

Before giving the proof of this theorem, let us state the following corollary, which

gives a complete characterization of all comonotonic convex risk measures, and a

remark concerning the set Qc in part (c), which is usually called the core of c.

Corollary 4.95. A convex risk measure on X is comonotonic if and only if it arises

as the Choquet integral of the loss with respect to a submodular, normalized, and

monotone set function c. In this case, c is given by c.A/ D .IA /, and has the

representation

.X / D max EQ X ;

Q2Qc

representing set Qmax .

234

R

Proof. Theorems 4.88 and 4.94 state that .X / WD .X / dc is a comonotonic coherent risk measure, which can be represented as in the assertion, as soon as c is a submodular, normalized, and monotone set function. Conversely, any comonotonic convex risk measure is coherent and arises as the Choquet integral of c.A/ WD .IA /,

due to Theorem 4.88. Theorem 4.94 then gives the submodularity of c.

Remark 4.96. Let c be a normalized, monotone, submodular set function. Theorem

4.94 implies in particular that the core Qc of c is non-empty. Moreover, c can be

recovered from Qc :

c.A/ D max Q A for all A 2 F .

Q2Qc

property that c.An / ! 0 for any decreasing sequence .An / of events such that n An D ;, then this property is shared by any

that Q is -additive. Thus, the corresponding coherent risk

Q 2 Qc , and it follows

R

measure .X / D .X / dc admits a representation in terms of -additive probability

measures. It follows by Lemma 4.21 that is continuous from above.

}

The proof of Theorem 4.94 requires some preparations. The assertion of the following lemma is not entirely obvious, since Fubinis theorem may fail if Q 2 M1;f

is not -additive.

Lemma 4.97. For

R X 2 X and Q 2 M1;f , the integral EQ X is equal to the

Choquet integral X dQ.

P

Proof. It is enough to prove the result for X 0. Suppose first that X D niD1 xi IAi

is as in Remark 4.92. Then

Z

n

i

n

h [

i X

X

X dQ D

.xi xi C1 /Q

Ak D

xi Q Ai D EQ X :

i D1

kD1

i D1

take Ronly finitely many values, and by using the Lipschitz continuity of both EQ

and dQ with respect to the supremum norm.

Lemma 4.98. Let A1 ; : : : ; An be a partition of into disjoint measurable sets, and

suppose that the normalized monotone set function c is submodular. Let Q be the

probability measure on F0 WD .A1 ; : : : ; An / with weights

Q Ak WD c.Bk / c.Bk1 /

for B0 WD ; and Bk WD

k

[

Aj ; k 1:

(4.58)

j D1

R

P

Then X dc EQ X for all F0 -measurable X D niD1 xi IAi , and equality holds

if the values of X are arranged in decreasing order: x1 xn .

235

Proof.

Clearly, it suffices to consider only the case X 0. Then Remark 4.92 implies

R

X dc D EQ X R as soon as the values of X are arranged in decreasing order.

Now we prove X dc EQ X for arbitrary F0 -measurable X 0. To this end,

note that any permutation of 1; : : : ; n induces a probability measure Q on F0

by applying the definition of Q to the re-labeled partitionRA.1/ ; : : : ; A.n/ . If is a

permutation such that x.1/ x.n/ , then we have X dc D EQ
X , and so

the assertion will follow if we can prove that EQ
X EQ X . To this end, it is

enough to show that EQ X EQ X if is the transposition of two indices i and

i C 1 which are such that xi < xi C1 , because can be represented as a finite product

of such transpositions.

Note next that

EQ X EQ X D xi .Q Ai Q Ai /

C xi C1 .Q Ai C1 Q Ai C1 /:

(4.59)

B0

WD ; and

Bk

WD

k

[

A.j / ;

k D 1; : : : ; n:

j D1

Q Ai C Q Ai C1 D Q A.i/ C Q A.iC1/ D c.BiC1 / c.Bi1 /

D c.Bi C1 / c.Bi 1 / D Q Ai C Q Ai C1 :

(4.60)

c.Bi / C c.Bi /, due to the submodularity of c. Thus,

Q AiC1 D c.Bi C1 / c.Bi / c.Bi / c.Bi1 / D Q A.i/ D Q Ai C1 :

Using (4.59), (4.60), and our assumption xi < xi C1 thus yields EQ X EQ X .

Proof of Theorem 4:94. (a) , (b): According to Proposition 4.84, the property of

positive homogeneity is shared by all Choquet integrals, and the implication (b) )

(a) is obvious.

(b) ) (c): By Corollary 4.19, .X / D maxQ2Qmax EQ X , where Q 2 M1;f

belongs to Qmax if and only if

Z

EQ X .X / D X dc for all X 2 X.

(4.61)

We will now show thatRthis set Qmax coincides with the set Qc . If Q 2 Qmax then,

in particular, Q A IA dc D c.A/ for all A 2 F . Hence Q 2 Qc . Conversely,

236

suppose Q 2 Qc . If X 0 then

Z

Z

Z 1

c.X > x/ dx

X dc D

0

Q X > x dx D EQ X ;

(c) ) (b) is obvious.

(b) ) (d): This follows precisely as in (4.57).

(d) ) (b): We have to show that the Choquet integral is subadditive. By Lemma

4.3, it is again enough to prove this for random variables which only take finitely many

values. Thus, let A1 ; : : : P

; An be a partition

Pof into finitely many disjoint measurable

sets. Let us write X D i xi IAi , Y D i yi IAi , and let us assume that the indices

i D 1; : : : ; n are arranged such that x1 C y1 xn C yn . Then the probability

measure Q constructed in Lemma 4.98 is such that

Z

Z

Z

.X C Y / dc D EQ X C Y D EQ X C EQ Y X dc C Y dc:

But this is the required subadditivity of the Choquet integral.

4.8

In this section, we will consider risk measures which arise in the financial market

model of Section 1.1. In this model, d C 1 assets are priced at times t D 0 and t D 1.

Prices at time 1 are modelled as non-negative random variables S 0 ; S 1 ; : : : ; S d on

some probability space .; F ; P /, with S 0 1 C r. Prices at time 0 are given by

a vector D .1; /, with D . 1 ; : : : ; d /. The discounted net gain of a trading

strategy D . 0 ; / is given by Y , where the random vector Y D .Y 1 ; : : : ; Y d / is

defined by

Si

i for i D 1; : : : ; d .

Yi D

1Cr

As in the previous two sections, risk measures will be defined on the space L1 D

; P /. A financial position X can be viewed as riskless if X 0 or, more

generally, if X can be hedged without additional costs, i.e., if there exists a trading

strategy D . 0 ; / such that D 0 and

L1 .; F

XC

S

D X C Y 0 P -a.s.

1Cr

A0 WD X 2 L1 j 9 2 Rd with X C Y 0 P -a.s. :

(4.62)

237

coherent risk measure. Moreover, 0 is sensitive in the sense of Definition 4:42 if and

only if the market model is arbitrage-free. In this case, 0 is continuous from above

and can be represented in terms of the set P of equivalent risk-neutral measures

0 .X / D sup E X :

(4.63)

P 2P

Proof. The fact that 0 is a coherent risk measure follows from Proposition 4.7. If

the model is arbitrage-free, then Theorem 1.32 yields the representation (4.63), and it

follows that 0 is sensitive and continuous from above.

Conversely, suppose that 0 is sensitive, but the market model admits an arbitrage

opportunity. Then there are 2 Rd and " > 0 such that 0 Y P -a.s. and

A WD Y " satisfies P A > 0. It follows that Y "IA 0, i.e., "IA is

acceptable. However, the sensitivity of 0 implies that

0 ."IA / D "0 .IA / > 0 .0/ D 0;

where we have used the coherence of 0 , which follows from fact that A0 is a cone.

Thus, we arrive at a contradiction.

There are several reasons why it may make sense to allow in (4.62) only strategies

that belong to a proper subset S of the class Rd of all strategies. For instance, if the

resources available to an investor are limited, only those strategies should be considered for which the initial investment in risky assets is below a certain amount. Such

a restriction corresponds to an upper bound on . There may be other constraints.

For instance, short sales constraints are lower bounds on the number of shares in the

portfolio. In view of market illiquidity, the investor may also wish to avoid holding

too many shares of one single asset, since the market capacity may not suffice to resell

the shares. Such constraints will be taken into account by assuming throughout the

remainder of this section that S has the following properties:

0 2 S.

S is convex.

AS WD X 2 L1 j 9 2 S with X C Y 0 P -a.s.

(4.64)

Moreover, we will assume from now on that

infm 2 R j m 2 AS > 1:

(4.65)

238

S .X / WD AS .X / D infm 2 R j m C X 2 AS

is a convex risk measure on L1 . Note that (4.65) holds, in particular, if S does not

contain arbitrage opportunities in the sense that Y 0 P -a.s. for 2 S implies

P Y D 0 D 1.

Remark 4.100. Admissibility of portfolios is a serious restriction; in particular, it

prevents unhedged short sales of any unbounded asset. Note, however, that it is consistent with our notion of acceptability for bounded claims in (4.64), since X C Y 0

implies Y kXk.

}

Two questions arise: When is S continuous from above, and thus admits a representation (4.32) in terms of probability measures? And, if such a representation

exists, how can we identify the minimal penalty function Smin on M1 .P /? In the

case S D Rd , both questions were addressed in Proposition 4.99. For general S,

only the second question has a straightforward answer, which will be given in Proposition 4.102. As can be seen from the proof of Proposition 4.99, an analysis of the

first question requires an extension of the arbitrage theory in Chapter 1 for the case

of portfolio constraints. Such a theory will be developed in Chapter 9 in a more general dynamic setting, and we will address both questions for the corresponding risk

measures in Corollary 9.32. This result implies the following theorem for the simple

one-period model of the present section:

Theorem 4.101. In addition to the above assumptions, suppose that the market model

is non-redundant in the sense of Definition 1:15 and that S is a closed subset of Rd

such that the cone j 2 S; 0 is closed. Then S is sensitive if and only if

S contains no arbitrage opportunities. In this case, S is continuous from above and

admits the representation

(4.66)

EQ X sup EQ Y :

S .X / D sup

Q2M1 .P /

2S

In the following proposition, we will explain the specific form of the penalty function in (4.66). This result will not require the additional assumptions of Theorem

4.101.

Proposition 4.102. For Q 2 M1 .P /, the minimal penalty function Smin of S is

given by

Smin .Q/ D sup EQ Y :

2S

239

2 S by admissibility. If X 2 AS , there exists 2 S such that X Y P -almost

surely. Thus,

EQ X EQ Y sup EQ Y

2S

for any Q 2 M1 .P /. Hence, the definition of the minimal penalty function yields

Smin .Q/ sup EQ Y :

(4.67)

2S

bounded since is admissible. Moreover,

Xk C Y D . Y k/ IY k 0;

so that Xk 2 AS . Hence,

Smin .Q/ EQ Xk D EQ . Y / ^ k ;

and so Smin .Q/ EQ Y by monotone convergence.

Exercise 4.8.1. Show that the identity

Smin .Q/ D sup EQ Y

2S

that Y is P -a.s. bounded. We thus obtain the representation

EQ X sup EQ Y

max

S .X / D

Q2M1;f .P /

2S

Remark 4.103. Suppose that S is a cone. Then the acceptance set AS is also a

cone, and S is a coherent measure of risk. If S is continuous from above, then

Corollary 4.37 yields the representation

S .X / D

sup EQ X

max

Q2QS

max

D Q 2 M1 .P / j Smin .Q/ D 0. It follows from

in terms of the non-empty set QS

Proposition 4.102 that for Q 2 M1 .P /

max

Q 2 QS

if and only if EQ Y 0

for all 2 S.

max

If S is sensitive, then the set S cannot contain any arbitrage opportunities, and QS

contains the set P of all equivalent martingale measures whenever such measures

max can be described as the set of absolutely continuous

exist. More precisely, QS

supermartingale measures with respect to S; this will be discussed in more detail in

the dynamical setting of Chapter 9.

}

240

Let us now relax the condition of acceptability in (4.64). We no longer insist that

the final outcome of an acceptable position, suitably hedged, should always be nonnegative. Instead, we only require that the hedged position is acceptable in terms of a

given convex risk measure A with acceptance set A. Thus, we define

AN WD X 2 L1 j 9 2 S; A 2 A with X C Y A P -a.s. :

(4.68)

A WD AN :

From now on, we assume that

> 1;

(4.69)

Proposition 4.104. The minimal penalty function min for is given by

min .Q/ D Smin .Q/ C min .Q/;

(4.70)

where Smin is the minimal penalty function for S and min is the minimal penalty

function for A .

Proof. We claim that

N

X 2 L1 j .X / < 0 X S C A j X S 2 AS ; A 2 A A:

(4.71)

X S WD X A 2 AS . Next, if X S 2 AS then X S C Y 0 for some 2 S.

N

Hence, for any A 2 A, we get X S C A C Y A 2 A, i.e., X WD X S C A 2 A.

In view of (4.71), we have

EQ X

sup

sup

sup EQ X S A

X S 2AS A2A

XW .X/<0

sup EQ X sup EQ X :

N

X2A

X2A

But the left- and rightmost terms are equal, and so we get that

min .Q/ D sup EQ X

N

X2A

sup

sup EQ X S A

X S 2AS A2A

241

Remark 4.105. It can happen that the sum of two penalty functions as on the righthand side of (4.70) is infinite for every Q 2 M1 .P /. In this case, the condition (4.69)

will be violated and the risk measure does not exist. Consider, for example, the

situation of a complete market model without trading constraints. Then S .X / D

E X , where P is the unique equivalent risk-neutral measure. That is,

Smin .Q/ D

0

when Q D P

C1 otherwise.

. Then min .Q/ C Smin .Q/ is infinite for Q P , and

min .P / C Smin .P / is finite (and in fact zero) if and only if

dP

1

dP

P -a.s.

Exercise 4.8.2. Suppose that condition (4.69) is satisfied. Show that the risk measure

can also be obtained as

.X / D inf.X X S / j X S 2 AS

and as

.X / D inf1 ..X Z/ C S .Z//:

Z2L

we now define in a general context.

Definition 4.106. For two convex risk measures 1 and 2 on L1 ,

1 2 .X / WD inf1 .1 .X Z/ C 2 .Z//;

Z2L

X 2 L1 ;

Exercise 4.8.3. Let 1 and 2 be two convex risk measures on L1 and assume that

1 2 .0/ D inf1 .1 .Z/ C 2 .Z// > 1:

Z2L

(b) Show that WD 1 2 is a convex risk measure on L1 .

242

(c) Show that the minimal penalty function min of is equal to the sum of the

respective minimal penalty functions 1min and 2min of 1 and 2 :

min .Q/ D 1min .Q/ C 2min .Q/

for Q 2 M1;f .

(d) Show that is continuous from below as soon as 1 is continuous from below.

}

For the rest of this section, we consider the following case study, which is based

on [45]. Let us fix a finite class

Q0 D Q1 ; : : : ; Qn

of equivalent probability measures Qi P such that jY j 2 L1 .Qi /; as in [45], we

call the measures in Q0 valuation measures. Define the sets

B WD X 2 L0 j EQi X exists and is 0; i D 1; : : : ; n

(4.72)

and

B0 WD X 2 B j EQi X D 0 for i D 1; : : : ; n :

Note that

B0 \ L0C D 0;

(4.73)

Qi P .

As the initial acceptance set, we take the convex cone

A WD B \ L1 :

(4.74)

suitable hedge is defined as in (4.68)

AN WD X 2 L1 j 9 2 Rd with X C Y 2 B :

Let us now introduce the following stronger version of the no-arbitrage condition

K \ L0C D 0, where K WD Y j 2 Rd :

K \ B D K \ B0 :

(4.75)

In other words, there is no portfolio 2 Rd such that the result satisfies the valuation

inequalities in (4.72) and is strictly favorable in the sense that at least one of the

inequalities is strict.

Note that (4.75) implies the absence of arbitrage opportunities:

K \ L0C D K \ B \ L0C D K \ B0 \ L0C D 0;

243

where we have used (4.73) and B \L0C D L0C . Thus, (4.75) implies, in particular, the

existence of an equivalent martingale measure, i.e., P ;. The following proposition

may be viewed as an extension of the fundamental theorem of asset pricing. Let us

denote by

n

n

X

X

i > 0;

i Qi

i D 1

R WD

i D1

i D1

the class of all representative models for the class Q0 , i.e., all mixtures such that

each Q 2 Q0 appears with a positive weight.

Proposition 4.107. The following two properties are equivalent:

(a) K \ B D K \ B0 .

(b) P \ R ;.

Proof. (b) ) (a): For V 2 K \ B and R 2 R, we have ER V 0. If we can

choose R 2 P \ R then we get ER V D 0, hence V 2 B0 .

(a) ) (b): Consider the convex set

C WD ER Y j R 2 R Rd I

we have to show that C contains the origin. If this is not the case then there exists

2 Rd such that

x 0 for x 2 C ,

(4.76)

and

x > 0

for some x 2 C ;

ER V 0

for all R 2 R,

ER V > 0, hence V K \ B0 , in contradiction to our assumption (a).

We can now state a representation theorem for the coherent risk measure correN It is a special case of Theorem 4.110 which will be

sponding to the convex cone A.

proved below.

Theorem 4.108. Under assumption (4.75), the coherent risk measure WD AN corresponding to the acceptance set AN is given by

.X / D

sup

P 2P \R

E X :

244

Q

P with jY j 2 L1 .Q/; as in [45], we call them stress test measures. In addition

to the valuation inequalities in (4.72), we require that an admissible position passes a

stress test specified by a floor

.Q/ < 0 for each Q 2 Q1 .

Thus, the convex cone A in (4.74) is reduced to the convex set

A1 WD A \ B1 D L1 \ .B \ B1 /;

where

B1 WD X 2 L0 j EQ X .Q/ for Q 2 Q1 :

Let

AN 1 WD X 2 L1 j 9 2 Rd with X C Y 2 B \ B1

denote the resulting acceptance set for positions combined with a suitable hedge.

Remark 4.109. The analogue

K \ .B \ B1 / D K \ B0

(4.77)

of our condition (4.75) looks weaker, but it is in fact equivalent to (4.75). Indeed, for

X 2 K \ B we can find " > 0 such that X1 WD "X satisfies the additional constraints

EQ X1 .Q/ for Q 2 Q1 .

Since X1 2 K \ B \ B1 , condition (4.77) implies X1 2 K \ B0 , hence X D 1" X1 2

}

K \ B0 , since K \ B0 is a cone.

Let us now identify the convex risk measure 1 induced by the convex acceptance

set AN 1 . Define

X

R1 WD

.Q/ Q
.Q/ 0;

.Q/ D 1 R

Q2Q

Q2Q

X

.Q/.Q/

.R/ WD

Q2Q

for R D

245

Theorem 4.110. Under assumption (4.75), the convex risk measure 1 induced by

the acceptance set AN 1 is given by

1 .X / D

.E X C .P //;

sup

(4.78)

P 2P \R1

C1

for Q P \ R1 ,

1 .Q/ WD

.Q/ for Q 2 P \ R1 .

Proof. Let denote the convex risk measure defined by the right-hand side of (4.78),

and let A denote the corresponding acceptance set

A WD X 2 L1 j E X .P / for all P 2 P \ R1 :

It is enough to show A D AN 1 .

(a): In order to show AN 1 A , take X 2 AN 1 and P 2 P \ R1 . There exists

2 Rd and A1 2 A1 such that X C Y A1 . Thus,

E X C Y E A1 .P /;

due to P 2 R1 . Since E Y D 0 due to P 2 P , we obtain E X .P /,

hence X 2 A .

(b): In order to show A AN 1 , we take X 2 A and assume that X AN 1 . This

/ with components

means that the vector x D .x1 ; : : : ; xN

xi WD EQi X .Qi /

does not belong to the convex cone

N

C WD .EQi Y /i D1;:::;N C y j 2 Rd ; y 2 RN

CR ;

show that C is closed. Thus, there exists 2 RN such that

x < inf xI

x2C

(4.79)

N

see Proposition

P A.1. Since C RC , we obtain
i 0 for i D 1; : : : ; N , and we

may assume i
i D 1 since
0. Define

R WD

N

X

i D1

i Qi 2 R1 :

246

Since C contains the linear space of vectors .EQi V /i D1;:::;N with V 2 K, (4.79)

implies

ER V D 0 for V 2 K,

hence R 2 P . Moreover, the right-hand side of (4.79) must be zero, and the condition

x < 0 translates into

ER X < .R/;

contradicting our assumption X 2 A .

(c): It remains to show that C is closed. For 2 Rd we define y./ as the vector

in RN with coordinates yi ./ D EQi Y . Any x 2 C admits a representation

x D y./ C z

?

with z 2 RN

C and 2 N , where

N WD 2 Rd j EQi Y D 0 for i D 1; : : : ; N ;

and

N ? WD 2 Rd j D 0 for all 2 N :

Take a sequence

xn D y.n / C zn ;

n D 1; 2; : : : ;

N

with n 2 N ? and zn 2 RN

C , such that xn converges to x 2 R . If lim infn jn j < 1,

then we may assume, passing to a subsequence if necessary, that n converges to

2 Rd . In this case, zn must converge to some z 2 RN

C , and we have x D y./Cz 2

C . Let us now show that the case limn jn j D 1 is in fact excluded. In that case,

n WD .1 C jn j/1 converges to 0, and the vectors n WD n n stay bounded. Thus,

we may assume that n converges to 2 N ? . This implies

C:

n"1

n"1

Since 2 N ? and jj D limn jn j D 1, we obtain y./ 0. Thus, the inequality

EQi ./ Y D yi ./ 0

holds for all i and is strict for some i, in contradiction to our assumption (4.75).

4.9

measures

In this section, we will establish a connection between convex risk measures and the

expected utility theory of Chapter 2.

247

Suppose that a risk-averse investor assesses the downside risk of a financial position

X 2 X by taking the expected utility E u.X / derived from the shortfall X , or

by considering the expected utility E u.X / of the position itself. If the focus is

on the downside risk, then it is natural to change the sign and to replace u by the

function `.x/ WD u.x/. Then ` is a strictly convex and increasing function, and

the maximization of expected utility is equivalent to minimizing the expected loss

E `.X / or the shortfall risk E `.X / . In order to unify the discussion of both

cases, we do not insist on strict convexity. In particular, ` may vanish on .1; 0,

and in this case the shortfall risk takes the form

E `.X / D E `.X / :

Definition 4.111. A function ` W R ! R is called a loss function if it is increasing

and not identically constant.

Let us return to the setting where we consider monetary risk measures defined on

the class X of all bounded measurable functions on some given measurable space

.; F /. Let us fix a probability measure P on .; F /. For a given loss function `

and an interior point x0 in the range of `, we define the following acceptance set:

A WD X 2 X j E `.X / x0 :

(4.80)

A D X 2 X j E u.X / y0 ;

(4.81)

where u.x/ D `.x/ and y0 D x0 . The acceptance set A satisfies (4.3) and (4.4).

By part (a) of Proposition 4.7 it induces a monetary risk measure given by

.X / D infm 2 R j E `.X m/ x0

D infm 2 R j E u.X C m/ y0 :

(4.82)

This risk measure satisfies (4.31) and hence can be regarded as a monetary risk measure on L1 . When ` is convex or u concave, is a convex risk measure. It is

normalized when x0 D `.0/.

Exercise 4.9.1. Let ` be a strictly increasing continuous loss function. Suppose that

the risk measure associated to ` via (4.82) satisfies

.X / .Y /

E `.X / E `.Y /

Hint: Apply Proposition 2.46.

For the rest of this section, we will only consider convex loss functions.

248

Definition 4.112. The convex risk measure in (4.82) is called utility-based shortfall

risk measure.

Proposition 4.113. The utility-based shortfall risk measure is continuous from below. Moreover, the minimal penalty function min for is concentrated on M1 .P /,

and can be represented in the form

.X / D

max

Q2M1 .P /

(4.83)

Proof. We have to show that is continuous from below. Note first that z D .X / is

the unique solution to the equation

E `.z X / D x0 :

(4.84)

Indeed, that z D .X / solves (4.84) follows by dominated convergence, since the

finite convex function ` is continuous. The solution is unique, since ` is strictly increasing on .`1 .x0 / "; 1/ for some " > 0.

Suppose now that .Xn / is a sequence in X which increases pointwise to some

X 2 X. Then .Xn / decreases to some finite limit R. Using the continuity of ` and

dominated convergence, it follows that

E `..Xn / Xn / ! E `.R X / :

But each of the approximating expectations equals x0 , and so R is a solution to (4.84).

Hence R D .X /, and this proves continuity from below. Since satisfies (4.31), the

representation (4.83) follows from Theorem 4.22 and Lemma 4.32.

Let us now compute the minimal penalty function min .

Example 4.114. For an exponential loss function `.x/ D e x , the minimal penalty

function can be described in terms of relative entropy, and the resulting risk measure

coincides, up to an additive constant, with the entropic risk measure introduced in

Example 4.34. In fact,

.X / D infm 2 R j E e .mCX/ x0 D

1

.log E e X log x0 /:

In this special case, the general formula (4.18) for min reduces to the variational

formula for the relative entropy H.QjP / of Q with respect to P

1

log x0

min .Q/ D sup EQ X log E e X

X2X

D

1

.H.QjP / log x0 /I

249

see Lemma 3.29. Thus, the representation (4.83) of is equivalent to the following

dual variational identity:

log E e X D

Q2M1 .P /

of the FenchelLegendre transform or conjugate function ` of the convex function `

defined by

` .z/ WD sup . zx `.x/ /:

x2R

Theorem 4.115. For any convex loss function `, the minimal penalty function in the

representation (4.83) is given by

1

dQ

x0 C E `

; Q 2 M1 .P /:

(4.85)

min .Q/ D inf

dP

>0

In particular,

dQ

1

x0 C E `

;

EQ X inf

dP

>0

.X / D

max

Q2M1 .P /

X 2 L1 :

To prepare the proof of Theorem 4.115, we summarize some properties of the functions ` and ` as stated in Appendix A.1. First note that ` is a proper convex function, i.e., it is convex and takes some finite value. We denote by J WD .` /0C its

right-continuous derivative. Then, for x; z 2 R,

xz `.x/ C ` .z/

(4.86)

Lemma 4.116. Let .`n / be a sequence of convex loss functions which decreases

pointwise to the convex loss function `. Then the corresponding conjugate functions

`n increase pointwise to ` .

Proof. It follows immediately from the definition of the FenchelLegendre transform

that each `n is dominated by ` , and that `n .z/ increases to some limit `1 .z/. We

have to prove that `1 D ` .

The function z 7! `1 .z/ is a lower semicontinuous convex function as the increasing limit of such functions. Moreover, `1 is a proper convex function, since it is

dominated by the proper convex function ` . Consider the conjugate function `

1 of

`1 . Clearly, `1 `, since `1 ` and since ` D ` by Proposition A.6. On

the other hand, we have by a similar argument that `

1 `n for each n. By taking

D ` .

D

`,

which

in

turn

gives

`

n " 1, this shows `

1

1

250

(a) ` .0/ D infx2R `.x/ and ` .z/ `.0/ for all z.

(b) There exists some z1 2 0; 1/ such that

` .z/ D sup .xz `.x//

for z z1 .

x0

(c)

` .z/

z

! 1 as z " 1.

(b): Let N WD z 2 R j ` .z/ D `.0/. We show in a first step that N ;.

Note that convexity of ` implies that the set S of all z with zx `.x/ `.0/ for all

x 2 R is non-empty. For z 2 S we clearly have ` .z/ `.0/. On the other hand,

` .z/ `.0/ by (a).

Now we take z1 WD sup N . It is clear that z1 0. If z > z1 and x < 0, then

xz `.x/ xz1 `.x/ ` .z1 / `.0/;

where the last inequality follows from the lower semicontinuity of ` . But ` .z/ >

`.0/, hence

sup .xz `.x// < ` .z/:

x<0

(c): For z z1 ,

` .z/=z D sup .x `.x/=z/

x0

by (b). Hence

` .z/

xz 1;

z

where xz WD supx j `.x/ z. Since ` is convex, increasing, and takes only finite

values, we have xz ! 1 as z " 1.

Proof of Theorem 4:115. Fix Q 2 M1 .P /, and denote by ' WD dQ=dP its density.

First, we show that it suffices to prove the claim for x0 > `.0/. Otherwise we can find

some a 2 R such that `.a/ < x0 , since x0 was assumed to be an interior point of

Q

`.R/. Let `.x/

WD `.x a/, and

Q XQ / x0 :

AQ WD XQ 2 X j E `.

Then AQ D X a j X 2 A, and hence

sup EQ XQ D sup EQ X C a:

Q

Q A

X2

X2A

(4.87)

251

Q

The convex loss function `Q satisfies the requirement `.0/

< x0 . So if the assertion is

established in this case, we find that

1

1

.x0 C E `Q .
'/ / D inf .x0 C E ` .
'/ / C aI

>0

>0

Q

XQ 2A

here we have used the fact that the FenchelLegendre transform `Q of `Q satisfies

`Q .z/ D ` .z/ C az. Together with (4.87), this proves that the reduction to the

case `.0/ < x0 is indeed justified.

For any
> 0 and X 2 A, (4.86) implies

X' D

1

1

.X /.
'/ .`.X / C ` .
'//:

1

1

.E `.X / C E ` . '// .x0 C E ` . '//:

X2A

min .Q/ inf

>0

1

.x0 C E ` .
'//

(4.88)

in case where min .Q/ < 1. This will be done first under the following extra conditions:

There exists 2 R such that `.x/ D inf ` for all x .

(4.89)

(4.90)

(4.91)

Note that these assumptions imply that ` .0/ < 1 and that J.0C/ . Moreover,

J.z/ increases to C1 as z " 1, and hence so does `.J.z//. Since

` .z/ `.0/ > x0

for all z,

(4.92)

lim `.J.z// x0 < lim.`.J.z// C ` .z// D lim zJ.z/ D 0:

z#0

z#0

z#0

These facts and the continuity of J imply that for large enough n there exists some

n > 0 such that

E `.J.
n '/I'n / D x0 :

Let us define

X n WD J.
n '/I'n :

252

Then X n is bounded and belongs to A. Hence, it follows from (4.86) and (4.92) that

min .Q/ EQ X n

1

E I'n J.
n '/.
n '/

n

1

D

E .`.X n / C ` .
n '// I'n

n

1

D

.x0 `.0/ P ' > n C E ` .
n '/I'n /

n

x0 `.0/

:

n

Since we assumed that min .Q/ < 1, the decreasing limit
1 of
n must be strictly

positive. The fact that ` is bounded from below allows us to apply Fatous lemma

min .Q/ lim inf

n"1

1

.x0 `.0/ P ' > n C E ` .
n '/I'n /

n

1

.x0 C E ` .
1 '/ /:

1

This proves (4.88) under the assumptions (4.89), (4.90), and (4.91).

If (4.89) and (4.90) hold, but J is not continuous, then we can approximate the

upper semicontinuous function J from above with an increasing continuous function

JQ on 0; 1/ such that

Z z

JQ .y/ dy

`Q .z/ WD ` .0/ C

0

satisfies

` .z/ `Q .z/ ` ..1 C "/z/ for z 0.

Let `Q WD `Q denote the FenchelLegendre transform of `Q . Since ` D ` by

Proposition A.6, it follows that

x

Q

`.x/

`.x/:

1C"

`

Therefore,

Q

AQ WD X 2 X j E `.X

/ x0 .1 C "/X j X 2 A DW A" :

253

Q we get that

Since we already know that the assertion holds for `,

dQ

dQ

1

1

x0 C E `

inf

x0 C E `Q

inf

dP

dP

>0

>0

D sup EQ X

Q

X2A

sup EQ X

X2A"

By letting " # 0, we obtain (4.88).

Finally, we remove conditions (4.89) and (4.90). If ` .z/ D C1 for some z,

then z must be an upper bound for the slope of `. So we will approximate ` by a

sequence .`n / of convex loss functions whose slope is unbounded. Simultaneously,

we can handle the case where ` does not take on its infimum. To this end, we choose

a sequence n # inf ` such that n `.0/ < x0 . We can define, for instance,

1

`n .x/ WD `.x/ _ n C .e x 1/C :

n

Then `n decreases pointwise to `. Each loss function `n satisfies (4.89) and (4.90).

Hence, for any " > 0 there are "n such that

1 > min .Q/ nmin .Q/

1

.x0 C E `n .
"n '/ / "

"n

for each n,

where nmin .Q/ is the penalty function arising from `n . Note that `n % ` by

Lemma 4.116. Our assumption min .Q/ < 1, the fact that

inf `n .z/ `n .0/ D `.0/ > x0 ;

z2R

and part (c) of Lemma 4.117 show that the sequence .
"n /n2N must be bounded away

from zero and from infinity. Therefore, we may assume that
"n converges to some

" 2 .0; 1/. Using again the fact that `n .z/ `.0/ uniformly in n and z, Fatous

lemma yields

min .Q/ C " lim inf

n"1

1

1

.x0 C E `n .
"n '/ / " .x0 C E ` .
" '/ /:

"

n

Example 4.118. Take

`.x/ WD

1 p

px

if x 0,

otherwise,

254

` .z/ WD

1 q

qz

if z 0,

C1 otherwise,

where q D p=.p 1/ is the usual dual coefficient. We may apply Theorem 4.115 for

any x0 > 0. Let Q 2 M1 .P / with density ' WD dQ=dP . Clearly, min .Q/ D C1

if ' Lq .; F ; P /. Otherwise, the infimum in (4.85) is attained for

px0 1=q

:

Q D

E ' q

Hence, we can identify min .Q/ for any Q

P as

dQ q 1=q

:

pmin .Q/ D .px0 /1=p E

dP

Taking the limit p # 1, we obtain the case `.x/ D x C where we measure the risk in

terms of the expected shortfall. Here we have

dQ

min

:

}

1 .Q/ D x0

dP 1

Together with Proposition 4.20, Theorem 4.115 yields the following result for risk

measures which are defined in terms of a robust notion of bounded shortfall risk. Here

it is convenient to define ` .1/ WD 1.

Corollary 4.119. Suppose that Q is a family of probability measures on .; F /, and

that `, ` , and x0 are as in Theorem 4:115. We define a set of acceptable positions by

A WD X 2 X j EP `.X / x0 for all P 2 Q :

Then the corresponding convex risk measure can be represented in terms of the penalty

function

1

dQ

x0 C inf EP `

; Q 2 M1 .; F /;

.Q/ D inf

dP

P 2Q

>0

where dQ=dP is the density appearing in the Lebesgue decomposition of Q with

respect to P as in Theorem A.13.

Example 4.120. In the case of Example 4.114, the corresponding robust problem in

Corollary 4.119 leads to the following entropy minimization problem: For a given Q

and a set Q of probability measures, find

inf H.QjP /:

P 2Q

Note that this problem is different from the standard problem of minimizing H.QjP /

with respect to the first variable Q as it appears in Section 3.2.

}

255

0

if z D 1,

` .z/ WD

C1 otherwise.

Therefore, .Q/ D 1 if Q P , and .X / D E X . If Q is a set of probability

measures, the robust risk measure of Corollary 4.119 is coherent, and it is given

by

}

.X / D sup EP X :

P 2Q

parametric family P for 2 . In each model P , the expected utility of a random

variable X 2 X is E u.X / , where u W R ! R is a given utility function. In

a Bayesian approach, one would choose a prior distribution

on . In terms of

F. Knights distinction between risk and uncertainty, we would now be in a situation

of model risk. Risk neutrality with respect to this model risk would be described by

the utility functional

Z

U.X / D

E
u.X /

.d /I

model risk aversion, we could choose another utility function uO W R ! R and consider

the utility functional UO .X / defined by

Z

u.

O UO .X // D u.E

O u.X / /

.d /:

Show that UO is quasi-concave, i.e.,

UO .X C .1 /Y / UO .X / ^ UO .Y /

for X; Y 2 L1 and 0 1.

O

D 1 e x , the utility functional UO .X / takes the form

O

U .X / D .u.X // for a convex risk measure . Then compute the minimal penalty

function in the robust representation of (here you may assume that is a finite set).

Discuss the limit " 1.

}

We now explore the relations between shortfall risk and the divergence risk measures introduced in Example 4.36. To this end, let g W 0; 1! R [ C1 be a lower

semicontinuous convex function satisfying g.1/ < 1 and the superlinear growth

condition g.x/=x ! C1 as x " 1. Recall the definition of the g-divergence,

h dQ i

;

Ig .QjP / WD E g

dP

Q 2 M1 .P /;

(4.93)

256

g .X / WD sup .EQ X Ig .QjP //;

X 2 L1 :

(4.94)

QP

We have seen in Exercise 4.3.3 that is continuous from below and that Ig . jP /

is its minimal penalty function, so the supremum in (4.94) is actually a maximum.

The following representation for g extends the corresponding result for AV@R in

Lemma 4.51, where g D 1 I.1=;1/ .

Theorem 4.122. Let g .y/ D supx>0 .xy g.x// be the FenchelLegendre transform of g. Then

g .X / D inf .E g .z X / z/;

z2R

X 2 L1 :

(4.95)

The proof of Theorem 4.122 is based on Theorem 4.115. In fact, we will see that,

in some sense, the two representation (4.85) and (4.95) are dual with respect to each

other. We prepare the proof with the following exercises. The first one concerns a

nice and sometimes very useful property of convex functions.

Exercise 4.9.3. If h is a convex function on 0; 1/, then

.x; y/ 7! xh

y

x

function satisfying g.1/ < 1 and the superlinear growth condition g.x/=x ! C1

as x " 1. For > 0 let g

.x/ WD g.x= /. Then . ; x/ 7! g

.x/ is convex by

Exercise 4.9.3. Let

.Q/ D Ig .QjP / be the corresponding g

-divergence. Show

that . ; Q/ 7!

.Q/ is a convex functional and that

h. / WD

.Q// if > 0,

C1

otherwise,

and h be as in Exercise 4.9.4. Our aim is to compute

h.1/. The idea is to use Theorem 4.115 so as to identify the FenchelLegendre transform h of h. To this end, we first observe that ` WD g satisfies the assumptions of

Theorem 4.115. Next, ` D g D g by Proposition A.6. Hence, Theorem 4.115

257

yields that

f .x/ WD infm 2 R j E g .m X / x

h dQ i

D max

EQ X inf
x C E g

dP

>0

Q2M1 .P /

D inf

min

>0 Q2M1 .P /

.EQ X C
x C

.Q//

D inf . x C h. // D h .x/;

>0

for all x in the interior of g .R/, which coincides with the interior of dom f . Exercise 4.9.4 hence yields h.1/ D h .1/ D supx .x f .x//. We have seen in the

proof of Proposition 4.113 that x D E g .f .x/ X / whenever x belongs

to the interior of g .R/. Hence,

h.1/ D sup .E g .f .x/ X / f .x//;

x2R

and the assertion follows by noting that the range of f contains all points to the left

of kX k1 x0 , where x0 is the lower bound for all points in which the right-hand

derivative of g is strictly positive.

Part II

Dynamic hedging

Chapter 5

Here we will work in a multiperiod setting, where the stochastic price fluctuation of

a financial asset is described as a stochastic process in discrete time. Portfolios will

be successively readjusted, taking into account the information available at each time.

In its weakest form, market efficiency requires that such dynamic trading strategies

should not create arbitrage opportunities. In Section 5.2 we show that an arbitragefree model is characterized by the existence of an equivalent martingale measure.

Under such a measure, the discounted price processes of the traded assets are martingales, that is, they have the mathematical structure of a fair game. In Section 5.3 we

introduce European contingent claims. These are financial instruments whose payoff

at the expiration date depends on the behavior of the underlying primary assets, and

possibly on other factors. We discuss the problem of pricing such contingent claims

in a manner which does not create new arbitrage opportunities. The pricing problem

is closely related to the problem of hedging a given claim by using a dynamic trading

strategy based on the primary assets. An ideal situation occurs if any contingent claim

can be perfectly replicated by the final outcome of such a strategy. In such a complete

model, the equivalent martingale measure P is unique, and derivatives are priced in

a canonical manner by taking the expectation of the discounted payoff with respect

to the measure P . Section 5.5 contains a simple case study for completeness, the

binomial model introduced by Cox, Ross, and Rubinstein. In this context, it is possible to obtain explicit pricing formulas for a number of exotic options, as explained

in Section 5.6. In Section 5.7 we pass to the limiting diffusion model of geometric

Brownian motion. Using a suitable version of the central limit theorem, we are led

to the general BlackScholes formula for European contingent claims and to explicit

pricing formulas for some exotic options such as lookback options and the up-and-in

and up-and-out calls.

The general structure of complete models is described in Section 5.4. There it will

become clear that completeness is the exception rather than the rule: Typical market

models in discrete time are incomplete.

5.1

Throughout this chapter, we consider a market model in which d C 1 assets are priced

at times t D 0; 1; : : : ; T . The price of the i th asset at time t is modelled as a nonnegative random variable S ti on a given probability space .; F ; P /. The random

262

to a -algebra F t F . One should think of F t as the class of all events which are

observable up to time t . Thus, it is natural to assume that

F0 F1 FT :

(5.1)

Definition 5.1. A family .F t / tD0;:::;T of -algebras satisfying (5.1) is called a filtration. In this case, .; F ; .F t / tD0;:::;T ; P / is also called a filtered probability space.

To simplify the presentation, we will assume that

F0 D ;; and

F D FT :

(5.2)

Let .E; E/ be a measurable space. A stochastic process with state space .E; E/ is

given by a family of E-valued random variables on .; F ; P / indexed by time. In

our context, the typical parameter sets will be 0; : : : ; T or 1; : : : ; T , and the state

space will be some Euclidean space.

Definition 5.2. A stochastic process Y D .Y t / tD0;:::;T is called adapted with respect

to the filtration .F t / tD0;:::;T if each Y t is F t -measurable. A stochastic process Z D

.Z t / tD1;:::;T is called predictable with respect to .F t / tD0;:::;T if each Z t is F t1 measurable.

Note that in our definition predictable processes start at t D 1 while adapted processes are also defined at t D 0. In particular, the asset prices S D .S t / tD0;:::;T form

an adapted stochastic process with values in Rd C1 .

Definition 5.3. A trading strategy is a predictable Rd C1 -valued process

D . 0 ; / D . t0 ; t1 ; : : : ; td / tD1;:::;T :

The value ti of a trading strategy corresponds to the quantity of shares of the

i

is the

i asset held during the t th trading period between t 1 and t . Thus, ti S t1

th

i

i

amount invested into the i asset at time t 1, while t S t is the resulting value at

time t . The total value of the portfolio t at time t 1 is

th

t S t1 D

d

X

i

ti S t1

:

i D0

t St D

d

X

i D0

ti S ti :

263

The predictability of expresses the fact that investments must be allocated at the

beginning of each trading period, without anticipating future price increments.

Definition 5.4. A trading strategy is called self-financing if

t S t D tC1 S t

for t D 1; : : : ; T 1.

(5.3)

Intuitively, (5.3) means that the portfolio is always rearranged in such a way that its

present value is preserved. It follows that the accumulated gains and losses resulting

from the asset price fluctuations are the only source of variations of the portfolio value

tC1 S tC1 t S t D tC1 .S tC1 S t /:

(5.4)

through summation over (5.4) that

t S t D 1 S 0 C

t

X

k .S k S k1 /

for t D 1; : : : ; T .

kD1

Here, the constant 1 S 0 can be interpreted as the initial investment for the purchase

of the portfolio 1 .

Example 5.5. Often it is assumed that the 0th asset plays the role of a locally riskless

bond. In this case, one takes S00 1 and one lets S t0 evolve according to a spot rate

r t 0: At time t , an investment x made at time t 1 yields the payoff x.1 C r t /.

Thus, a unit investment at time 0 produces the value

S t0 D

t

Y

.1 C rk /

kD1

0

at time t . An investment in S is locally riskless if the spot rate r t is known beforehand at time t 1. This idea can be made precise by assuming that the process r is

predictable.

}

Without assuming predictability as in the preceding example, we assume from now

on that

S t0 > 0 P -a.s. for all t .

This assumption allows us to use the 0th asset as a numraire and to form the discounted price processes

X ti WD

S ti

;

S t0

t D 0; : : : ; T; i D 0; : : : ; d:

units of the numraire. As explained in Remark 1.11, discounting allows comparison

of asset prices which are quoted at different times.

264

trading strategy is given by

V0 WD 1 X 0

and

V t WD t X t

for t D 1; : : : ; T:

G0 WD 0

and

t

X

G t WD

kD1

Clearly,

Vt D t X t D

t St

;

S t0

so V t can be interpreted as the portfolio value at the end of the t th trading period

expressed in units of the numraire asset. The gains process

Gt D

t

X

k .Xk Xk1 /

kD1

reflects, in terms of the numraire, the net gains which have accumulated through the

trading strategy up to time t . For a self-financing trading strategy , the identity

t S t D 1 S 0 C

t

X

k .S k S k1 /

(5.5)

kD1

remains true if all relevant quantities are computed in units of the numraire. This is

the content of the following simple proposition.

Proposition 5.7. For a trading strategy the following conditions are equivalent:

(a) is self-financing.

(b) t X t D tC1 X t for t D 1; : : : ; T 1.

P

(c) V t D V0 C G t D 1 X 0 C tkD1 k .Xk Xk1 / for all t .

Proof. By dividing both sides of (5.3) by S t0 it is seen that condition (b) is a reformulation of Definition 5.4. Moreover, (b) holds if and only if

tC1 X tC1 t X t D tC1 .X tC1 X t / D tC1 .X tC1 X t /

for t D 1; : : : ; T 1, and this identity is equivalent to (c).

265

0

t0 D . tC1 t / X t

tC1

for t D 1; : : : ; T 1.

(5.6)

Since

10 D V0 1 X0 ;

(5.7)

the entire process 0 is determined by the initial investment V0 and the d -dimensional

process . Consequently, if a constant V0 and an arbitrary d -dimensional predictable

process are given, then we can use (5.7) and (5.6) as the definition of a predictable

process 0 , and this construction yields a self-financing trading strategy WD . 0 ; /.

In dealing with self-financing strategies , it is thus sufficient to focus on the initial

}

investment V0 and the d -dimensional processes X and .

Remark 5.9. Different economic agents investing into the same market may choose

different numraires. For example, consider the following simple market model in

which prices are quoted in euros ( C) as the domestic currency. Let S 0 be a locally

riskless C-bond with the predictable spot rate process r 0 , i.e.,

S t0 D

t

Y

.1 C rk0 /;

kD1

and let S 1 describe the price of a locally riskless investment into US dollars ($). Since

the price of this $-bond is quoted in C, the asset S 1 is modeled as

S t1 D U t

t

Y

.1 C rk1 /;

kD1

where r 1 is the spot rate for a $-investment, and U t denotes the price of 1$ in terms

of C, i.e., U t is the exchange rate of the $ versus the C. While it may be natural

for European investors to take S 0 as their numraire, it may be reasonable for an

American investor to choose S 1 . This simple example explains why it may be relevant

to check which concepts and results of our theory are invariant under a change of

numraire; see, e.g., the discussion at the end of Section 5.2.

}

Exercise 5.1.1. Consider a market model with two assets which are modeled as usual

by the stochastic process S D .S 0 ; S 1 / that is adapted to the filtration .F t / tD0;:::;T .

Decide which of the following processes are predictable and which in general are

not.

(i) t D IS 1 >S 1

t

t 1

(ii) 1 D 1 and t D IS 1

1

t 1 >S t 2

for t 2;

266

(iv) t D IS 1 >S 1 ;

t

1

t 1 <S0

5.2

for t 2.

profit with positive probability but without any downside risk.

Definition 5.10. A self-financing trading strategy is called an arbitrage opportunity

if its value process V satisfies

V0 0;

VT 0 P -a.s.,

and

P VT > 0 > 0:

The existence of such an arbitrage opportunity may be regarded as a market inefficiency in the sense that certain assets are not priced in a reasonable way. In this

section, we will characterize those market models which do not allow for arbitrage

opportunities. Such models will be called arbitrage-free. The following proposition

shows that the market model is arbitrage-free if and only if there are no arbitrage opportunities for each single trading period. Later on, this fact will allow us to apply the

results of Section 1.6 to our multi-period model.

Proposition 5.11. The market model admits an arbitrage opportunity if and only if

there exist t 2 1; : : : ; T and 2 L0 .; F t1 ; P I Rd / such that

.X t X t1 / 0

P -a.s.;

and

(5.8)

Proof. To prove necessity, take an arbitrage opportunity D . 0 ; / with value process V , and let

t WD mink j Vk 0 P -a.s., and P Vk > 0 > 0 :

Then t T by assumption, and either V t1 D 0 P -a.s. or P V t1 < 0 > 0. In the

first case, it follows that

t .X t X t1 / D V t V t1 D V t

P -a.s.

F t1 -measurable, and

.X t X t1 / D .V t V t1 / IV t 1 <0 V t1 IV t 1 <0 :

The expression on the right-hand side is non-negative and strictly positive with a positive probability, so (5.8) holds.

267

Now we prove sufficiency. For t and as in (5.8), define a d -dimensional predictable process by

if s D t ,

s WD

0 otherwise.

Via (5.7) and (5.6), uniquely defines a self-financing trading strategy D . 0 ; /

with initial investment V0 D 0. Since the corresponding value process satisfies VT D

.X t X t1 /, the strategy is an arbitrage opportunity.

Exercise 5.2.1. Let V be the value process of a self-financing strategy in an arbitragefree market model. Prove that the following two implications hold for all t 2 0; : : : ;

T 1 and A 2 F t with P A > 0.

P V tC1 V t 0 j A D 1

H)

P V tC1 V t D 0 j A D 1;

P V tC1 V t 0 j A D 1

H)

P V tC1 V t D 0 j A D 1:

space .; F ; .F t /; Q/ is called a martingale if M is adapted, satisfies EQ jM t j <

1 for all t , and if

Ms D EQ M t j Fs for 0 s t T .

(5.9)

For each time s and for each horizon t > s, the conditional expectation of the future

gain M t Ms is zero, given the information available at s.

Exercise 5.2.2. Let M D .M t / tD0;:::;T be an adapted process on .; F ; .F t /; Q/

such that EQ jM t j < 1 for all t . Show that the following conditions are equivalent:

(a) M is a martingale.

(b) M t D EQ M tC1 j F t for 0 t T 1.

(c) There exists F 2 L1 .; FT ; Q/ such that M t D EQ F j F t for t D 0; : : : ; T ,

that is, M arises as a sequence of successive conditional expectations.

}

Exercise 5.2.3. Let QQ be a probability measure on .; F ; .F t // that is absolutely

continuous with respect to Q. Show that the density process

d QQ

Z t WD

; t D 0; : : : ; T;

dQ F t

is a martingale with respect to Q.

Whether or not a given process M is a martingale depends on the underlying probability measure Q. If we wish to emphasize the dependence of the martingale property

of M on a particular measure Q, we will say that M is a Q-martingale or that M is

a martingale under the measure Q.

268

if the discounted price process X is a (d -dimensional) Q-martingale, i.e.,

EQ X ti < 1

and

Xsi D EQ X ti j Fs ;

0 s t T; i D 1; : : : ; d:

to the original measure P on FT . The set of all equivalent martingale measures is

denoted by P .

The following result is a version of Doobs fundamental systems theorem for

martingales. It states that a fair game admits no realistic gambling system which

produces a positive expected gain. Here, Y denotes the negative part Y ^ 0 of a

random variable Y .

Theorem 5.14. For a probability measure Q, the following conditions are equivalent:

(a) Q is a martingale measure.

(b) If D . 0 ; / is self-financing and is bounded, then the value process V of

is a Q-martingale.

(c) If D . 0 ; / is self-financing and its value process V satisfies EQ VT < 1,

then V is a Q-martingale.

(d) If D . 0 ; / is self-financing and its value process V satisfies VT 0 Q-a.s.,

then EQ VT D V0 .

Proof. (a) ) (b): Let V be the value process of a self-financing trading strategy

D . 0 ; / such that there is a constant c such that j i j c for all i. Then

jV t j jV0 j C

t

X

kD1

T 1,

EQ V tC1 j F t D EQ V t C tC1 .X tC1 X t / j F t

D V t C tC1 EQ X tC1 X t j F t

D Vt ;

where we have used that tC1 is F t -measurable and bounded.

(b) ) (c): We will show the following implication:

If EQ V t < 1 then EQ V t j F t1 D V t1 .

Since EQ VT < 1 by assumption, we will then get

EQ VT1 D EQ EQ VT j FT 1 EQ VT < 1;

(5.10)

269

due to Jensens inequality for conditional expectations. Repeating this argument will

yield EQ V t < 1 and EQ V t j F t1 D V t1 for all t . Since V0 is a finite constant, we will also get EQ V t D V0 , which together with the fact that EQ V t < 1

implies V t 2 L1 .Q/ for all t . Thus, the martingale property of V will follow.

To prove (5.10), note first that EQ V t j F t1 is well defined due to our assumption

.a/

.a/

EQ V t < 1. Next, let t WD t Ij t ja for a > 0. Then t .X t X t1 / is

.a/

.a/

and EQ t .X t X t1 / j F t1 D 0. Hence,

.X t X t1 / 2 L1 .Q/

.a/

.a/

D EQ V t Ij t ja t

.X t X t1 / j F t1

.X t X t1 / j F t1

D V t1 Ij t ja :

By sending a " 1, we obtain (5.10).

(c) ) (d): By (5.2), every Q-martingale M satisfies

M0 D EQ MT j F0 D EQ MT :

(d) ) (a): To prove that X ti 2 L1 .Q/ for given i and t , consider the deterministic

j

process defined by si WD Ist and s WD 0 for j i. By Remark 5.8, can be

complemented with a predictable process 0 such that D . 0 ; / is a self-financing

strategy with initial investment V0 D X0i . The corresponding value process satisfies

VT D V0 C

T

X

s .Xs Xs1 / D X ti 0:

sD1

EQ X ti D EQ VT D V0 D X0i ;

(5.11)

i I A for

Condition (a) will follow if we can show that EQ X ti I A D EQ X t1

given t , i, and A 2 F t1 . To this end, we define a d -dimensional predictable process

j

by is WD Is<t C IAc IsDt and s WD 0 for j i . As above, we take a predictable

0 such that D .0 ; / is a self-financing strategy with initial investment

process

VQ0 D X0i . Its terminal value is given by

VQT D VQ0 C

T

X

sD1

i

s .Xs Xs1 / D X ti IAc C X t1

IA 0:

270

X0i D VQ0 D EQ VQT

i

D EQ X ti I Ac C EQ X t1

I A :

i I A.

By comparing this identity with (5.11), we conclude that EQ X ti I A D EQ X t1

Remark 5.15. (a) Suppose that the objective measure P is itself a martingale measure, so that the fluctuation of prices may be viewed as a fair game. In this case, the

preceding proposition shows that there are no realistic self-financing strategies which

would generate a positive expected gain. Thus, the assumption P 2 P is a strong

version of the so-called efficient market hypothesis. For a market model containing

a locally risk-less bond, this strong hypothesis would imply that risk-averse investors

would not be attracted towards investing into the risky assets if their expectations are

consistent with P ; see Example 2.40.

(b) The strong assumption P 2 P implies, in particular, that there is no arbitrage

opportunity, i.e., no self-financing strategy with positive expected gain and without

any downside risk. Indeed, Theorem 5.14 implies that the value process of any selffinancing strategy with V0 0 and VT 0 satisfies E VT D V0 , hence VT D 0

P -almost surely. The assumption that the market model is arbitrage-free may be

viewed as a much milder and hence more flexible form of the efficient market hypothesis.

}

We can now state the following dynamic version of the fundamental theorem of

asset pricing, which relates the absence of arbitrage opportunities to the existence of

equivalent martingale measures.

Theorem 5.16. The market model is arbitrage-free if and only if the set P of all

equivalent martingale measures is non-empty. In this case, there exists a P 2 P

with bounded density dP =dP .

Proof. Suppose first that there exists an equivalent martingale measure P . Then it

follows as in Remark 5.15 (b) that the market model in which the probability measure

P is replaced by P is arbitrage-free. Since the notion of an arbitrage opportunity

depends on the underlying measure only through its null sets and since these are common for the two equivalent measures P and P , it follows that also the original market

model is arbitrage-free.

Let us turn to the proof of the converse assertion. For t 2 1; : : : ; T , we define

K t WD .X t X t1 / j 2 L0 .; F t1 ; P I Rd / :

(5.12)

271

K t \ L0C .; F t ; P / D 0

(5.13)

holds for all t . Note that (5.13) depends on the measure P only through its null sets.

Condition (5.13) allows us to apply Theorem 1.55 to the t th trading period. For t D

T we obtain a probability measure PQT P which has a bounded density d PQT =dP

and which satisfies

EQ T XT XT 1 j FT 1 D 0:

Now suppose that we already have a probability measure PQ tC1 P with a bounded

density d PQtC1 =dP such that

EQ tC1 Xk Xk1 j Fk1 D 0

for t C 1 k T:

(5.14)

The equivalence of PQtC1 and P implies that (5.13) also holds with P replaced by

PQ tC1 . Applying Theorem 1.55 to the t th trading period yields a probability measure

PQ t with a bounded F t -measurable density Z t WD d PQt =d PQtC1 > 0 such that

EQ t X t X t1 j F t1 D 0:

Clearly, PQt is equivalent to P and has a bounded density, since

d PQtC1

d PQt

d PQt

D

dP

dP

d PQtC1

is the product of two bounded densities. Moreover, if t C1 k T , Proposition A.12

and the F t -measurability of Z t D d PQt =d PQtC1 imply

EQ tC1 .Xk Xk1 /Z t j Fk1

EQ t Xk Xk1 j Fk1 D

EQ tC1 Z t j Fk1

D EQ tC1 Xk Xk1 j Fk1

D 0:

Hence, (5.14) carries over from PQtC1 to PQt . We can repeat this recursion until finally

P WD PQ1 yields the desired equivalent martingale measure.

Clearly, the absence of arbitrage in the market is independent of the choice of the

numraire, while the set P of equivalent martingale measures generally does depend

on the numraire. In order to investigate the structure of this dependence, suppose

that the first asset S 1 is P -a.s. strictly positive, so that it can serve as an alternative

numraire. The price process discounted by S 1 is denoted by

0

St

S t2

S td

S t0

0

1

d

Y t D .Y t ; Y t ; : : : ; Y t / WD

;

1;

;

:

:

:

;

X t ; t D 0; : : : ; T:

D

S t1

S t1

S t1

S t1

272

Let PQ be the set of equivalent martingale measures for Y . Then PQ ; if and only

if P ;, according to Theorem 5.16 and the fact that the existence of arbitrage

opportunities is independent of the choice of the numraire.

Proposition 5.17. The two sets P and PQ are related via the identity

d PQ

XT1

D

for

some

P

2

P

:

PQ D PQ

dP

X01

Proof. The process X t1 =X01 is a P -martingale for any P 2 P . In particular,

E XT1 =X01 D 1, and the formula

XT1

d PQ

D

dP

X01

defines a probability measure PQ which is equivalent to P . Moreover, by Proposition A.12,

1

EQ Y t j Fs D 1 E Y t X t1 jFs

Xs

1

D 1 E X t j Fs

Xs

D Y s:

Hence, PQ is an equivalent martingale measure for Y , and it follows that

d PQ

XT1

PQ PQ

D

for

some

P

2

P

:

dP

X01

Reversing the roles of X and Y yields the identity of the two sets.

Remark 5.18. Unless XT1 is P -a.s. constant, the two sets P and PQ satisfy

P \ PQ D ;:

}

Exercise 5.2.4. Let X t WD X t1 be the P -a.s. strictly positive discounted price process

of a risky asset. The corresponding returns are

X t X t1

;

RQ t WD

X t1

so that

X t D X0

t

Y

t D 1; : : : ; T;

.1 C RQ k /:

kD1

273

(a) Show that X is a P -martingale when the .RQ t / are independent and integrable

random variables with E RQ t D 0.

(b) Now give necessary and sufficient conditions on the .RQ t / such that X is a P martingale.

(c) Construct an example in which X is a martingale but the .RQ t / are not independent.

}

Exercise 5.2.5. Let Z1 ; : : : ; ZT be independent standard normal random variables on

.; F ; P /, and let F t be the -field generated by Z1 ; : : : ; Z t , where t D 1; : : : ; T .

We also let F0 WD ;; . For constants X01 > 0, i > 0, and mi 2 R we now define

the discounted price process of a risky asset as the following sequence of log-normally

distributed random variables,

X t1 WD X01

t

Y

e i Zi Cmi ;

t D 0; : : : ; T:

(5.15)

i D1

Construct an equivalent martingale measure for X 1 under which the random variables

}

X t1 have still a log-normal distribution.

Exercise 5.2.6. For a square-integrable random variable X on .; F ; P / and a algebra F0 F , the conditional variance of X given F0 is defined as

var.X jF0 / WD E .X E X j F0 /2 j F0 :

Show that

var.X jF0 / D E X 2 j F0 .E X j F0 /2

and that

var.X / D E var.X jF0 / C var.E X j F0 /:

Exercise 5.2.7. Let Y1 and Y2 be jointly normal random variables with mean 0, variance 1, and correlation % 2 .1; 1/. That is, the joint distribution of .Y1 ; Y2 / has the

density

'.y1 ; y2 / D

1

1

.y 2 Cy 2 2%y1 y2 /

p

e 2.1%2 / 1 2

;

2 1 %2

.y1 ; y2 / 2 R2 :

(b) Compute the conditional variance var.Y2 jY1 /.

(c) For constants m; 2 R compute E e Y2 Cm j Y1 .

274

log-normal price process in analogy to (5.15)

X t1

WD

X01

t

Y

e i Zi Cmi ;

t D 0; : : : ; 2;

(5.16)

i D1

(a) Compute the conditional expectation E X21 j X11 .

(b) Construct an equivalent martingale measure for the price process in (5.16) when

}

the filtration is the one generated by the process X 1 .

Exercise 5.2.9. Let X0 ; X1 ; : : : describe the discounted prices of a risky asset in a

market model with infinite time horizon that is modeled on a filtered probability space

.; .F t / tD0;1;::: ; P /. Suppose that every market model X0 ; : : : ; XT with finite time

horizon T 2 N is arbitrage-free.

(a) Show that there exists a sequence .PT /T D1;2::: of probability measures such that

PT is defined on .; FT /, is equivalent to P on FT , and such that the restriction

of PT to FT 1 equals PT1 , i.e., PT A D PT1 A for all A 2 FT 1 .

arises as the restriction to F of a

(b) Can you give conditions under which PTS

T

measure P that is defined on F1 WD . t0 F t /?

Hint: You may choose a setting in which one can apply the Kolmogorov extension theorem.

}

5.3

A key topic of mathematical finance is the analysis of derivative securities or contingent claims, i.e., of certain assets whose payoff depends on the behavior of the

primary assets S 0 ; S 1 ; : : : ; S d and, in some cases, also on other factors.

Definition 5.19. A non-negative random variable C on .; FT ; P / is called a European contingent claim. A European contingent claim C is called a derivative of the

underlying assets S 0 ; S 1 ; : : : ; S d if C is measurable with respect to the -algebra

generated by the price process .S t / tD0;:::;T .

A European contingent claim has the interpretation of an asset which yields at time

T the amount C.!/, depending on the scenario ! of the market evolution. T is called

the expiration date or the maturity of C . Of course, maturities prior to the final trading

period T of our model are also possible, but unless it is otherwise mentioned, we will

assume that our European contingent claims expire at T . In Chapter 6, we will meet

another class of derivative securities, the so-called American contingent claims. As

long as there is no risk of confusion between European and American contingent

275

claims, we will use the term contingent claim to refer to a European contingent

claim.

Example 5.20. The owner of a European call option has the right, but not the obligation, to buy an asset at time T for a fixed price K, called the strike price. This

corresponds to a contingent claim of the form

C call D .STi K/C :

Conversely, a European put option gives the right, but not the obligation, to sell the

asset at time T for a strike price K. This corresponds to the contingent claim

C put D .K STi /C :

Example 5.21. The payoff of an Asian option depends on the average price

i

WD

Sav

1 X i

St

jT j

t2T

instance, an average price call with strike K corresponds to the contingent claim

call

i

WD .Sav

K/C ;

Cav

put

i C

WD .K Sav

/ :

Cav

Average price options can be used, for instance, to secure regular cash streams against

exchange rate fluctuations. For example, assume that an economic agent receives at

each time t 2 T a fixed amount of a foreign currency with exchange rates Sti . In

this case, an average price put option may be an efficient instrument for securing the

incoming cash stream against the risk of unfavorable exchange rates.

An average strike call corresponds to the contingent claim

i C

.STi Sav

/ ;

i

STi /C :

.Sav

An average strike put can be used, for example, to secure the risk from selling at time

T a quantity of an asset which was bought at successive times over the period T . }

276

Example 5.22. The payoff of a barrier option depends on whether the price of the

underlying asset reaches a certain level before maturity. Most barrier options are

either knock-out or knock-in options. A knock-in option pays off only if the barrier B

is reached. The simplest example is a digital option

1 if max0tT S ti B;

C dig WD

0 otherwise,

which has a unit payoff if the price processes reaches a given upper barrier B > S0i .

Another example is the down-and-in put with strike price K and lower barrier BQ < S0i

which pays off

Q

.K STi /C if min0tT S ti B,

put

Cd&i WD

0

otherwise.

A knock-out barrier option has a zero payoff once the price of the underlying asset

reaches the predetermined barrier. For instance, an up-and-out call corresponds to the

contingent claim

call

Cu&o WD

0

otherwise;

}

see Figure 5.1. Down-and-out and up-and-in options are defined analogously.

S01

K

T

Figure 5.1. In one scenario, the payoff of the up-and-out call becomes zero because the

stock price hits the barrier B before time T . In the other scenario, the payoff is given by

.ST K/C .

277

Example 5.23. Using a lookback option, one can trade the underlying asset at the

maximal or minimal price that occurred during the life of the option. A lookback call

has the payoff

STi min S ti ;

0tT

max S ti STi :

0tT

The discounted value of a contingent claim C when using the numraire S 0 is given

by

H WD

C

:

ST0

We will call H the discounted European claim or just the discounted claim associated with C . In the remainder of this text, H will be the generic notation for the

discounted payoff of any type of contingent claim.

The reader may wonder why we work simultaneously with the notions of a contingent claim and a discounted claim. From a purely mathematical point of view, there

would be no loss of generality in assuming that the numraire asset is identically equal

to one. In fact, the entire theory to be developed in Part II can be seen as a discretetime stochastic analysis for the d -dimensional process X D .X 1 ; : : : ; X d / and its

stochastic integrals

t

X

k .Xk Xk1 /

kD1

of predictable d -dimensional processes . However, some of the economic intuition would be lost if we would limit the discussion to this level. For instance, we

have already seen the economic relevance of the particular choice of the numraire,

even though this choice may be irrelevant from the mathematicians point of view.

As a compromise between the mathematicians preference for conciseness and the

economists concern for keeping track explicitly of economically relevant quantities,

we develop the mathematics on the level of discounted prices, but we will continue

to discuss definitions and results in terms of undiscounted prices whenever it seems

appropriate.

From now on, we will assume that our market model is arbitrage-free or, equivalently, that

P ;:

278

if there exists a self-financing trading strategy whose terminal portfolio value coincides with C , i.e.,

C D T S T P -a.s.

Such a trading strategy is called a replicating strategy for C .

Clearly, a contingent claim C is attainable if and only if the corresponding discounted claim H D C =ST0 is of the form

H D T X T D VT D V0 C

T

X

t .X t X t1 /;

tD1

we will say that the discounted claim H is attainable, and we will call a replicating strategy for H . The following theorem yields the surprising result that an attainable discounted claim is automatically integrable with respect to every equivalent

martingale measure. Note, however, that integrability may not hold for an attainable

contingent claim prior to discounting.

Theorem 5.25. Any attainable discounted claim H is integrable with respect to each

equivalent martingale measure, i.e.,

E H < 1

for all P 2 P .

Moreover, for each P 2 P the value process of any replicating strategy satisfies

Vt D E H j Ft

P -a.s. for t D 0; : : : ; T .

Proof. This follows from VT D H 0 and the systems theorem in the form of

Theorem 5.14.

Remark 5.26. The identity

V t D E H j F t ;

t D 0; : : : ; T;

appearing in Theorem 5.25 has two remarkable implications. Since its right-hand side

is independent of the particular replicating strategy, all such strategies must have the

same value process. Moreover, the left-hand side does not depend on the choice of

P 2 P . Hence, V t is a version of the conditional expectation E H j F t for every

}

P 2 P . In particular, E H is the same for all P 2 P .

279

Theorem 5.25 states that

0 C

t S t D St E

F t ; t D 0; : : : ; T;

ST0

P -a.s. for all P 2 P and for every replicating strategy . In particular, the initial

investment which is needed for a replication of C is given by

C

1 S 0 D S00 E

:

}

ST0

Let us now turn to the problem of pricing a contingent claim. Consider first a

discounted claim H which is attainable. Then the (discounted) initial investment

1 X 0 D V0 D E H

(5.17)

needed for the replication of H can be interpreted as the unique (discounted) fair

price of H . In fact, a different price for H would create an arbitrage opportunity.

For instance, if H could be sold at time 0 for a price Q which is higher than (5.17),

then selling H and buying the replicating portfolio yields the profit

Q 1 X 0 > 0

at time 0, although the terminal portfolio value VT D T X T suffices for settling the

claim H at maturity T . In order to make this idea precise, let us formalize the idea of

an arbitrage-free price of a general discounted claim H .

Definition 5.28. A real number H 0 is called an arbitrage-free price of a discounted claim H , if there exists an adapted stochastic process X d C1 such that

X0d C1 D H ;

X td C1 0

for t D 1; : : : ; T 1,

and

(5.18)

XTd C1 D H;

and such that the enlarged market model with price process .X 0 ; X 1 ; : : : ; X d ; X d C1 /

is arbitrage-free. The set of all arbitrage-free prices of H is denoted by .H /. The

lower and upper bounds of .H / are denoted by

inf .H / WD inf .H /

and

sup .H / WD sup .H /:

at which H can be traded at time 0 without introducing arbitrage opportunities into

the market model: If H is sold for H , then neither buyer nor seller can find an

280

investment strategy which both eliminates all the risk and yields an opportunity to

make a positive profit. Our aim in this section is to characterize the set of all arbitragefree prices of a discounted claim H .

Note that an arbitrage-free price H is quoted in units of the numraire asset. The

amount that corresponds to H in terms of currency units prior to discounting is equal

to

C WD S00 H ;

and C is an (undiscounted) arbitrage-free price of the contingent claim C WD ST0 H .

Theorem 5.29. The set of arbitrage-free prices of a discounted claim H is non-empty

and given by

.H / D E H j P 2 P and E H < 1 :

(5.19)

inf .H / D inf E H

P 2P

and

sup .H / D sup E H :

P 2P

find an equivalent martingale measure PO for the market model extended via (5.18).

PO must satisfy

O X i j F t for t D 0; : : : ; T and i D 1; : : : ; d C 1.

X ti D E

T

O H . Thus, we obtain the incluIn particular, PO belongs to P and satisfies H D E

sion in (5.19).

Conversely, if H D E H for some P 2 P , then we can define the stochastic

process

X td C1 WD E H j F t ; t D 0; : : : ; T;

which satisfies all the requirements of (5.18). Moreover, the same measure P is

clearly an equivalent martingale measure for the extended market model, which hence

is arbitrage-free. Thus, we obtain the identity of the two sets in (5.19).

To show that .H / is non-empty, we first fix some measure PQ P such that

Q

E H < 1. For instance, we can take d PQ D c.1 C H /1 dP , where c is the

normalizing constant. Under PQ , the market model is arbitrage-free. Hence, Theorem 5.16 yields P 2 P such that dP =d PQ is bounded. In particular, E H < 1

and hence E H 2 .H /.

The formula for inf .H / follows immediately from (5.19) and the fact that .H /

;. The one for sup .H / needs an additional argument. Suppose that P 1 2 P is such

that E 1 H D 1. We must show that for any c > 0 there exists some 2 .H /

with > c. To this end, let n be such that Q WD E 1 H ^ n > c, and define

X td C1 WD E 1 H ^ n j F t ;

t D 0; : : : ; T:

281

X d ; X d C1 /, which hence is arbitrage-free. Applying the already established fact that

the set of arbitrage-free prices of any contingent claim is nonempty to the extended

market model yields an equivalent martingale measure P for .X 0 ; : : : ; X d ; X d C1 /

such that E H < 1. Since P is also a martingale measure for the original market

model, the first part of this proof implies that WD E H 2 .H /. Finally, note

that

D E H E H ^ n D E XTd C1 D X0d C1 D Q > c:

Hence, the formula for sup .H / is proved.

Example 5.30. In an arbitrage-free market model, we consider a European call option C call D .ST1 K/C with strike K > 0 and with maturity T . We assume that

the numraire S 0 is the predictable price process of a locally riskless bond as in Example 5.5. Then S t0 is increasing in t and satisfies S00 1. For any P 2 P ,

Theorem 5.29 yields an arbitrage-free price call of C call which is given by

call D E

C call

ST0

D E

XT1

K

ST0

C

:

call can be bounded from below as follows:

C

K

call

1

E XT 0

ST

C

K

D S01 E

(5.20)

ST0

.S01 K/C :

In financial language, this fact is usually expressed by saying that the value of the

option is higher than its intrinsic value .S01 K/C , i.e., the payoff if the option were

exercised immediately. The difference of the price call of an option and its intrinsic

value is often called the time-value of the European call option; see Figure 5.2. }

Example 5.31. For a European put option C put D .K ST1 /C , the situation is more

complicated. If we consider the same situation as in Example 5.30, then the analogue

of (5.20) fails unless the numraire S 0 is constant. In fact, as a consequence of the

put-call parity, the time value of a put option whose intrinsic value is large (i.e., the

option is in the money) usually becomes negative; see Figure 5.3.

}

Our next aim is to characterize the structure of the set of arbitrage-free prices of a

discounted claim H . It follows from Theorem 5.29 that every arbitrage-free price H

282

S01

Figure 5.2. The typical price of a call option as a function of S01 is always above the

options intrinsic value .S01 K/C .

inf .H / D inf E H and

P 2P

sup .H / D sup E H :

P 2P

We also know that inf .H / and sup .H / are equal if H is attainable. The following

theorem shows that also the converse implication holds, i.e., H is attainable if and

only if inf .H / D sup .H /.

Theorem 5.32. Let H be a discounted claim.

(a) If H is attainable, then the set .H / of arbitrage-free prices for H consists of

the single element V0 , where V is the value process of any replicating strategy

for H .

(b) If H is not attainable, then inf .H / < sup .H / and

.H / D .inf .H /; sup .H //:

Proof. The first assertion follows from Remark 5.26 and Theorem 5.29.

To prove (b), note first that

.H / D E H j P 2 P ; E H < 1

is an interval because P is a convex set. We will show that .H / is open by constructing for any 2 .H / two arbitrage-free prices L and O for H such that L < < .

O

To this end, take P 2 P such that D E H . We will first construct an equivalent

O H > E H . Let

martingale measure PO 2 P such that E

U t WD E H j F t ;

t D 0; : : : ; T;

283

S01

Figure 5.3. The typical price of a European put option as a function of S01 compared to

the options intrinsic value .K S01 /C .

so that

H D U0 C

T

X

.U t U t1 /:

tD1

K t \ L1 .P /, where

K t WD .X t X t1 / j 2 L0 .; F t1 ; P I Rd / :

By Lemma 1.69, K t \ L1 .P / is a closed linear subspace of L1 .; F t ; P /.

Therefore, Theorem A.57 applied with B WD U t U t1 and C WD K t \ L1 .P /

yields some Z 2 L1 .; F t ; P / such that

supE W Z j W 2 K t \ L1 .P / < E .U t U t1 / Z < 1:

From the linearity of K t \ L1 .P / we deduce that

E W Z D 0

for all W 2 K t \ L1 .P /,

(5.21)

E .U t U t1 / Z > 0:

(5.22)

ZO WD 1 C Z E Z j F t1

can be taken as the density d PO =dP D ZO of a new probability measure PO P .

284

O H D E H ZO

E

D E H C E E H j F t Z E E H j F t1 E Z j F t1

D E H C E U t Z E U t1 Z

> E H ;

O H 5 E H <

where we have used (5.22) in the last step. On the other hand, E

3

O H will yield the desired arbitrage-free price larger than if we

1. Thus, O WD E

have PO 2 P .

Let us prove that PO 2 P . For k > t , the F t -measurability of ZO and Proposition A.12 yield that

O Xk Xk1 j Fk1 D E Xk Xk1 j Fk1 D 0:

E

For k D t , (5.21) yields E .X t X t1 / Z j F t1 D 0. Thus, it follows from

E ZO j F t1 D 1 that

O X t X t1 j F t1

E

D E .X t X t1 /.1 E Z j F t1 / j F t1 C E .X t X t1 /Z j F t1

D 0:

Finally, if k < t then P and PO coincide on Fk . Hence

O Xk Xk1 j Fk1 D E Xk Xk1 j Fk1 D 0;

E

and we may conclude that PO 2 P .

It remains to construct another equivalent martingale measure PL such that

L H < E H D :

L WD E

(5.23)

d PO

d PL

WD 2

;

dP

dP

which defines a probability measure PL P , because the density d PO =dP is

bounded from above by 5=3 and below by 1/3. PL 2 P is then obvious as is (5.23).

Remark 5.33. So far, we have assumed that a contingent claim is settled at the terminal time T . A natural way of dealing with an FT0 -measurable payoff C0 0 maturing

at some time T0 < T is to apply our results to the corresponding discounted claim

H0 WD

C0

ST00

285

in the market model with the restricted time horizon T0 . Clearly, this restricted model

is arbitrage-free. An alternative approach is to invest the payoff C0 at time T0 into the

numraire asset S 0 . At time T , this yields the contingent claim

C WD C0

ST0

ST00

H D

C

C0

D 0

0

ST

ST0

intuitively clear that these two approaches for determining the arbitrage-free prices of

C0 should be equivalent. A formal proof must show that the set .H / is equal to the

set

.H0 / WD E0 H0 j P0 2 P0 and E0 H0 < 1

of arbitrage-free prices of H0 in the market model whose time horizon is T0 . Here,

P0 denotes the set of measures P0 on .; FT0 / which are equivalent to P on FT0 and

which are martingale measures for the restricted price process .X t / tD0;:::;T0 . Clearly,

each P 2 P defines an element of P0 by restricting P to the -algebra FT0 . In

fact, Proposition 5.34 below shows that every element in P0 arises in this way. Thus,

the two sets of arbitrage-free prices for H and H0 coincide, i.e.,

.H / D .H0 /:

It follows, in particular, that H0 is attainable if and only if H is attainable.

Proposition 5.34. Consider the situation described in Remark 5:33 and let P0 2 P0

be given. Then there exists some P 2 P whose restriction to FT0 is equal to P0 .

Proof. Let PO 2 P be arbitrary, and denote by ZT0 the density of P0 with respect to

the restriction of PO to the -algebra FT0 . Then ZT0 is FT0 -measurable, and

dP WD ZT0 d PO

defines a probability measure on F . Clearly, P is equivalent to PO and to P , and it

coincides with P0 on FT0 . To check that P 2 P , it suffices to show that X t X t1

is a martingale increment under P for t > T0 . For these t , the density ZT0 is F t1 measurable, so Proposition A.12 implies that

O X t X t1 j F t1 D 0:

E X t X t1 j F t1 D E

286

Example 5.35. Let us consider the situation of Example 5.30, where the numraire

S 0 is a locally riskless bond. Remark 5.33 allows us to compare the arbitrage-free

prices of two European call options C0 D .ST10 K/C and C D .ST1 K/C with

the same strikes and underlyings but with different maturities T0 < T . As in Example 5.30, we get that for P 2 P

E

0

C

ST0

C

1

1

FT0 0 ST0 K E

FT0

ST0

ST0

ST0

(5.24)

C0

:

ST00

price of C0 is lower than the price of C

E

C

ST0

E

C0

:

ST00

This argument suggests that the price of a European call option should be an increasing function of the maturity.

}

Exercise 5.3.1. Let Yk .k D 1; : : : ; T / be independent identically distributed random

variables in L1 .; F ; P /, and suppose that the Yk are not P -a.s. constant and satisfy

E Yk D 0. Let furthermore

X t WD X0 C

t

X

Yk ;

t D 0; : : : ; T:

kD1

and F t D .Y1 ; : : : ; Y t / for t D 1; : : : ; T . We now enlarge the filtration by adding

insider information of the terminal value XT . That is, we consider the enlarged

filtration

FQn WD .Fn [ .XT //:

(a) Show that X is no longer a P -martingale with respect to .FQt /.

(b) Prove that the process

XQ t WD X t

t1

X

kD0

1

.XT Xk /:

T k

287

(c) The insider information of the terminal value XT implies the existence of selffinancing strategies with positive expected profit. Construct a strategy that

maximizes the expected profit

T

hX

t .X t X t1 /

tD1

within the class of all .FQt /-predictable strategies with j t j 1 P -a.s. for all t .

5.4

Complete markets

We have seen in Theorem 5.32 that any attainable claim in an arbitrage-free market

model has a unique arbitrage-free price. Thus, the situation becomes particularly

transparent if all contingent claims are attainable.

Definition 5.36. An arbitrage-free market model is called complete if every contingent claim is attainable.

Complete market models are precisely those models in which every contingent

claim has a unique and unambiguous arbitrage-free price. However, in discrete time,

only a very limited class of models enjoys this property. The following characterization of market completeness is sometimes called the second fundamental theorem of

asset pricing.

Theorem 5.37. An arbitrage-free market model is complete if and only if there exists

exactly one equivalent martingale measure. In this case, the number of atoms in

.; FT ; P / is bounded from above by .d C 1/T .

Proof. If the model is complete, then H WD IA for A 2 FT is an attainable discounted

claim. It follows from the results of Section 5.3 that the mapping P 7! E H D

P A is constant over the set P . Hence, there can be only one equivalent martingale

measure.

Conversely, if jP j D 1, then the set .H / of arbitrage-free prices of every discounted claim H has exactly one element. Hence, Theorem 5.32 implies that H is

attainable.

To prove the second assertion, note first that the asserted bound on the number of

atoms in FT holds for T D 1 by Corollary 1.42. We proceed by induction on T . Suppose that the assertion holds for T 1. By assumption, any bounded FT -measurable

random variable H 0 can be written as

H D VT 1 C T .XT XT 1 /;

288

where both VT 1 and T are FT 1 -measurable and hence constant on each atom A of

.; FT 1 ; P /. It follows that the dimension of the linear space L1 .; FT ; P jA/

is less than or equal to d C 1. Thus, Proposition 1.41 implies that .; FT ; P jA/

has at most d C 1 atoms. Applying the induction hypothesis concludes the proof.

Below we state additional characterizations of market completeness. Denote by Q

the set of all martingale measures in the sense of Definition 5.13. Then both P and Q

are convex sets. Recall that an element of a convex set is called an extreme point of

this set if it cannot be written as a non-trivial convex combination of members of this

set.

Property (d) in the following theorem is usually called the predictable representation property, or the martingale representation property, of the P -martingale X.

Theorem 5.38. For P 2 P the following conditions are equivalent:

(a) P D P .

(b) P is an extreme point of P .

(c) P is an extreme point of Q.

(d) Every P -martingale M can be represented as a stochastic integral of a d dimensional predictable process

M t D M0 C

t

X

kD1

Q1 ; Q2 2 Q, then Q1 and Q2 are both absolutely continuous with respect to P . By

defining

1

Pi WD .Qi C P /; i D 1; 2;

2

we thus obtain two martingale measures P1 and P2 which are equivalent to P .

Hence, P1 D P2 D P and, in turn, Q1 D Q2 D P .

(c) ) (b): This is obvious since P Q.

(b) ) (a): Suppose that there exists a PO 2 P which is different from P . We

will show below that in this case PO can be chosen such that the density d PO =dP is

bounded by some constant c > 0. Then, if " > 0 is less than 1=c,

dP 0

d PO

WD 1 C " "

dP

dP

defines another measure P 0 2 P different from P . Moreover, P can be represented

as the convex combination

P D

1

" O

PC

P 0;

1C"

1C"

289

which contradicts condition (b). Hence, P must be the unique equivalent martingale

measure.

It remains to prove the existence of PO 2 P with a bounded density d PO =dP

if there exists some PQ 2 P which is different from P . Then there exists a set

A 2 FT such that P A PQ A . We enlarge our market model by introducing the

additional asset

Xtd C1 WD PQ A j F t ; t D 0; : : : ; T;

and we take P instead of P as our reference measure. By definition, PQ is an equivalent martingale measure for .X 0 ; X 1 ; : : : ; X d ; X d C1 /. Hence, the extended market

model is arbitrage-free, and Theorem 5.16 guarantees the existence of an equivalent

martingale measure PO such that the density d PO =dP is bounded. Moreover, PO must

be different from P , since P is not a martingale measure for X d C1 :

X0d C1 D PQ A P A D E XTd C1 :

(a) ) (d): The terminal value MT of a P -martingale M can be decomposed into

the difference of its positive and negative parts

MT D MTC MT :

MTC and MT can be regarded as two discounted claims, which are attainable by

Theorem 5.37. Hence, there exist two d -dimensional predictable process C and

such that

T

X

k .Xk Xk1 / P -a.s.

MT D V0 C

kD1

for two non-negative constants V0C and V0 . Since the value processes

V t WD V0 C

t

X

k .Xk Xk1 /

kD1

M t D E MTC MT j F t D V tC V t :

This proves that the desired representation of M holds in terms of the d -dimensional

predictable process WD C .

(d) ) (a): Applying our assumption to the martingale M t WD P A j F t shows

that H D IA is an attainable contingent claim. Hence, it follows from the results of

Section 5.3 that the mapping P 7! P A is constant over the set P . Thus, there

can be only one equivalent martingale measure.

290

WD 1; C1T D ! D .y1 ; : : : ; yT / j yi 2 1; C1

and denote by Y t .!/ WD y t , for ! D .y1 ; : : : ; yT /, the projection on the t th coordinate. As filtration we take F0 WD ;; and F t WD .Y1 ; : : : ; Y t / for t D 1; : : : ; T .

We consider a financial market model with two assets such that the discounted price

process X t WD X t1 D S t1 =S t0 is of the form

X t D X0 exp

t

X

. k Yk C mk /

kD1

for a constant X0 > 0 and two predictable processes . t / and .m t /. We suppose that

0 jm t j < t for all t .

(a) Show that when P is a probability measure on for which P ! > 0 for all

!, then there exists a unique equivalent martingale measure P .

(b) By using the binary structure of the model, and without using Theorem 5.37,

prove the following martingale representation result. If P is as in (a), every

P -martingale M can be represented as

M t D M0 C

t

X

k .Xk Xk1 /

t D 0; : : : ; T ,

kD1

k D

5.5

Mk Mk1

:

Xk Xk1

A complete financial market model with only one risky asset must have a binary tree

structure, as we have seen in Theorem 5.37. Under an additional homogeneity assumption, this reduces to the following particularly simple model, which was introduced by Cox, Ross, and Rubinstein in [58]. It involves the riskless bond

S t0 WD .1 C r/t ;

t D 0; : : : ; T;

R t WD

S t S t1

S t1

in the t th trading period can only take two possible values a; b 2 R such that

1 < a < b:

291

Thus, the stock price jumps from S t1 either to the higher value S t D S t1 .1 C b/ or

to the lower value S t D S t1 .1 C a/. In this context, we are going to derive explicit

formulas for the arbitrage-free prices and replicating strategies of various contingent

claims.

Let us construct the model on the sample space

WD 1; C1T D ! D .y1 ; : : : ; yT / j yi 2 1; C1 :

Denote by

Y t .!/ WD y t

the projection on the

t th

for ! D .y1 ; : : : ; yT /

(5.25)

1 C Y t .!/

a

1 Y t .!/

Cb

D

R t .!/ WD a

2

2

b

if Y t .!/ D 1,

if Y t .!/ D C1.

S t WD S0

t

Y

.1 C Rk /;

kD1

where the initial value S0 > 0 is a given constant. The discounted price process takes

the form

t

Y

1 C Rk

St

X t D 0 D S0

:

1Cr

St

kD1

As filtration we take

F t WD .S0 ; : : : ; S t / D .X0 ; : : : ; X t /;

t D 0; : : : ; T:

F t D .Y1 ; : : : ; Y t / D .R1 ; : : : ; R t /

for t D 1; : : : ; T ;

F WD FT coincides with the power set of . Let us fix any probability measure P on

.; F / such that

P ! > 0 for all ! 2 .

(5.26)

Such a model will be called a binomial model or a CRR model. The following theorem

characterizes those parameter values a; b; r for which the model is arbitrage-free.

Theorem 5.39. The CRR model is arbitrage-free if and only if a < r < b. In

this case, the CRR model is complete, and there is a unique martingale measure

P . The martingale measure is characterized by the fact that the random variables

R1 ; : : : ; RT are independent under P with common distribution

r a

P R t D b D p WD

; t D 1; : : : ; T:

ba

292

price process is a martingale under Q, i.e.,

1 C R tC1

Q-a.s.

X t D EQ X tC1 j F t D X t EQ

Ft

1Cr

for all t T 1. This identity is equivalent to the equation

r D EQ R tC1 j F t D b Q R tC1 D b j F t C a .1 Q R tC1 D b j F t /;

i.e., to the condition

Q R tC1 D b j F t .!/ D p D

r a

ba

for Q-a.e. ! 2 .

But this holds if and only if the random variables R1 ; : : : ; RT are independent under

Q with common distribution Q R t D b D p . In particular, there can be at most

one martingale measure for X.

If the market model is arbitrage-free, then there exists an equivalent martingale

measure P . The condition P P implies

p D P R1 D b 2 .0; 1/;

which holds if and only if a < r < b.

Conversely, if a < r < b then we can define a measure P P on .; F / by

P ! WD .p /k .1 p /T k > 0

where k denotes the number of occurrences of C1 in ! D .y1 ; : : : ; yT /. Under P ,

Y1 ; : : : ; YT , and hence R1 ; : : : ; RT , are independent random variables with common

distribution P Y t D 1 D P R t D b D p , and so P is an equivalent martingale measure.

From now on, we consider only CRR models which are arbitrage-free, and we

denote by P the unique equivalent martingale measure.

Remark 5.40. Note that the unique martingale measure P , and hence the valuation of any contingent claim, is completely independent of the initial choice of the

objective measure P within the class of measures satisfying (5.26).

}

Let us now turn to the problem of pricing and hedging a given contingent claim C .

The discounted claim H D C =ST0 can be written as

H D h.S0 ; : : : ; ST /

for a suitable function h.

293

V t D E H j F t ;

t D 0; : : : ; T;

V t .!/ D v t .S0 ; S1 .!/; : : : ; S t .!//;

where the function v t is given by

S1

ST t

; : : : ; xt

:

v t .x0 ; : : : ; x t / D E h x0 ; : : : ; x t ; x t

S0

S0

(5.27)

Proof. Clearly,

S tC1

ST

; : : : ; St

V t D E h S0 ; S1 ; : : : ; S t ; S t

Ft :

St

St

Each quotient S tCs =S t is independent of F t and has under P the same distribution

as

s

Y

Ss

D

.1 C Rk /:

S0

kD1

Since V is characterized by the recursion

VT WD H

and

V t D E V tC1 j F t ;

t D T 1; : : : ; 0;

vT .x0 ; : : : ; xT / D h.x0 ; : : : ; xT /;

(5.28)

O

v t .x0 ; : : : ; x t / D p v tC1 .x0 ; : : : ; x t ; x t b/

where

aO WD 1 C a

and

bO WD 1 C b:

Example 5.42. If H D h.ST / depends only on the terminal value ST of the stock

price, then V t depends only on the value S t of the current stock price

V t .!/ D v t .S t .!//:

Moreover, the formula (5.27) for v t reduces to an expectation with respect to the

binomial distribution with parameter p

!

T

t

X

T t

T tk O k

b /

v t .x t / D

h.x t aO

.p /k .1 p /T tk :

k

kD0

294

!

T

X

T

T k O k

h.S0 aO

.p /k .1 p /T k :

b /

.H / D v0 .S0 / D

k

kD0

For h.x/ D .x K/C =.1Cr/T or h.x/ D .K x/C =.1Cr/T , we obtain explicit formulas for the arbitrage-free prices of European call or put options with strike price K.

For instance, the price of H call WD .ST K/C =.1 C r/T is given by

!

T

X

1

call

T k O k

C T

.S0 aO

}

b K/

.p /k .1 p /T k :

.H / D

k

.1 C r/T

kD0

M t WD max Ss ;

0st

0 t T;

instance, H can be an up-and-in or up-and-out barrier option or a lookback put. Then

the value process of H is of the form

V t D v t .S t ; M t /;

where

h S

i

M

T t

T t

:

v t .x t ; m t / D E h x t

; mt _ xt

S0

S0

This follows from (5.27) or directly from the fact that

Su

MT D M t _ S t max

;

tuT S t

where max tuT Su =S t is independent of F t and has the same law as MT t =S0

under P . The same argument works for options that depend on the minimum of the

stock price such as lookback calls or down-and-in barrier options.

}

Exercise 5.5.1. For an Asian option depending on the average price

Sav WD

1 X

St

jT j

t2T

X

Ss :

A t WD

s2T ; st

Show that the value process V t of the Asian option is a function of S t , A t , and t . }

295

Let us now derive a formula for the hedging strategy D . 0 ; / of our discounted

claim H D h.S0 ; : : : ; ST /.

Proposition 5.44. The hedging strategy is given by

t .!/ D t .S0 ; S1 .!/; : : : ; S t1 .!//;

where

t .x0 ; : : : ; x t1 / WD .1 C r/

O

v t .x0 ; : : : ; x t1 ; x t1 b/

:

O

x t1 b x t1 aO

t .!/.X t .!/ X t1 .!// D V t .!/ V t1 .!/:

(5.29)

In this equation, the random variables t , X t1 , and V t1 depend only on the first

t 1 components of !. For a fixed t , let us define ! C and ! by

! WD .y1 ; : : : ; y t1 ; 1; y tC1 ; : : : ; yT /:

Plugging ! C and ! into (5.29) shows

O C r/1 X t1 .!// D V t .! C / V t1 .!/

t .!/ .X t1 .!/ b.1

O C r/1 X t1 .!// D V t .! / V t1 .!/:

t .!/ .X t1 .!/ a.1

Solving for t .!/ and using our formula (5.28) for V t , we obtain

t .!/ D .1 C r/

V t .! C / V t .! /

D t .S0 ; S1 .!/; : : : ; S t1 .!//:

X t1 .!/.bO a/

O

Remark 5.45. The term t may be viewed as a discrete derivative of the value

function v t with respect to the possible stock price changes. In financial language,

a hedging strategy based on a derivative of the value process is often called a Delta

hedge.

}

Remark 5.46. Let H D h.ST / be a discounted claim which depends on the terminal

value of S by way of an increasing function h. For instance, h can be the discounted

payoff function h.x/ D .x K/C =.1 C r/T of a European call option. Then

v t .x/ D E h.x ST t =S0 /

is also increasing in x, and so the hedging strategy satisfies

t .!/ D .1 C r/t

O v t .S t1 .!/ a/

v t .S t1 .!/ b/

O

0:

O

S t1 .!/ b S t1 .!/ aO

In other words, the hedging strategy for H does not involve short sales of the risky

asset.

}

296

call option has the form

C

S

T

K :

ST0

}

5.6

Exotic derivatives

The recursion formula (5.28) can be used for the numeric computation of the value

process of any contingent claim. For the value processes of certain exotic derivatives

which depend on the maximum of the stock price, it is even possible to obtain simple

closed-form solutions if we make the additional assumption that

aO D

1

;

bO

where aO D 1 C a and bO D 1 C b. In this case, the price process of the risky asset is

of the form

S t .!/ D S0 bO Z t .!/

where, for Yk as in (5.25),

Z0 WD 0

and

Z t WD Y1 C C Y t ;

t D 1; : : : ; T:

P ! WD

1

D 2T ;

jj

! 2 :

Under the measure P , the random variables Y t are independent with common distribution P Y t D C1 D 12 . Thus, the stochastic process Z becomes a standard

random walk under P . Therefore,

t

if t C k is even,

2t t Ck

2

(5.30)

P Zt D k D

0

otherwise.

The following lemma is the key to numerous explicit results on the distribution of

Z under the measure P ; see, e.g., Chapter III of [110]. For its statement, it will be

convenient to assume that the random walk Z is defined up to time T C 1; this can

always be achieved by enlarging our probability space .; F /. We denote by

M t WD max Zs

0st

297

P MT k and ZT D k l D P ZT D k C l ;

and

P MT D k and ZT D k l D 2

kCl C1

P ZT C1 D 1 C k C l :

T C1

Proof. Let

.!/ WD inft 0 j Z t .!/ D k ^ T:

For ! D .y1 ; : : : ; yT / 2 we define .!/ by .!/ D ! if .!/ D T and by

.!/ D .y1 ; : : : ; y.!/ ; y.!/C1 ; : : : ; yT /

otherwise, i.e., if the level k is reached before the deadline T . Intuitively, the two

trajectories .Z t .!// tD0;:::;T and .Z t ..!/// tD0;:::;T coincide up to .!/, but from

then on the latter path is obtained by reflecting the original one on the horizontal axis

at level k; see Figure 5.4.

kCl

kl

Let Ak;l denote the set of all ! 2 such that ZT .!/ D k l and MT k. Then

is a bijection from Ak;l to the set

MT k and ZT D k C l ;

298

which coincides with ZT D k Cl, due to our assumption l 0. Hence, the uniform

distribution P must assign the same probability to Ak;l and ZT D k C l, and we

obtain our first formula.

The second formula is trivial in case T C k C l is not even. Otherwise, we let

j WD .T C k C l/=2 and apply (5.30) together with part one of this lemma

P MT D k; ZT D k l

D P MT k; ZT D k l P MT k C 1; ZT D k l

D P ZT D k C l P ZT D k C l C 2

!

!

T

T T

T

D2

2

j

j C1

!

T C 1 2j C 1 T

;

D 2T

T C1

j C1

and this expression is equal to the right-hand side of our second formula.

Formula (5.30) will change if we replace the uniform distribution P by our martingale measure P , described in Theorem 5.39

P Zt D k D

.p /

t Ck

2

.1 p /

t k

2

t Ck

2

if t C k is even,

otherwise.

Lemma 5.48 (Reflection principle for P ). For all k 2 N and l 2 N0 ,

P MT k; ZT D k l D

1 p l

P ZT D k C l

p

p k

P ZT D k l ;

D

1 p

and

P MT D k; ZT D k l

1 1 p l k C l C 1

P ZT C1 D 1 C k C l

D

p

p

T C1

p k k C l C 1

1

D

P ZT C1 D 1 k l :

1 p 1 p

T C1

299

T CZT

T ZT

dP

D 2T .p / 2 .1 p / 2 :

dP

P ! D .p /k .1 p /T k

to each ! D .y1 ; : : : ; yT / 2 which contains exactly k components with yi D C1.

But for such an ! we have ZT .!/ D k .T k/ D 2k T , and our formula follows.

From the density formula, we get

P MT k and ZT D k l

D 2T .p /

T Ckl

2

.1 p /

T Clk

2

P MT k and ZT D k l :

Applying the reflection principle and using again the density formula, we see that the

probability term on the right is equal to

P ZT D k C l D 2T .p /

T CkCl

2

.1 p /

T kl

2

P ZT D k C l ;

which gives the first identity. The proof of the remaining ones is analogous.

Example 5.49 (Up-and-in call option). Consider an up-and-in call option with payoff

call

Cu&i

D

0

otherwise,

where B > S0 _ K denotes a given barrier, and where K > 0 is the strike price. Our

aim is to compute the arbitrage-free price

call

.Cu&i

/D

1

call

E Cu&i

:

.1 C r/T

Clearly,

call

D E .ST K/C I max S t B

E Cu&i

0tT

D E .ST K/ I ST B

C E .ST K/C I max S t B; ST < B :

0tT

The first expectation on the right can be computed explicitly in terms of the binomial

distribution. Thus, it remains to compute the second expectation, which we denote

by I . To this end, we may assume without loss of generality that B lies within the

300

range of possible asset prices, i.e., there exists some k 2 N such that B D S0 bO k .

Then, by Lemma 5.48,

X

E .ST K/C I MT k; ZT D k l

I D

l1

l1

l1

p k

P ZT D k l

1 p

p k

X

Q C P ZT D k l

D

.S0 bO kl K/

bO 2k

1p

l1

p k B 2

Q C I ST < BQ ;

E .ST K/

D

1 p

S0

where

S 2

S2

0

and BQ WD 0 :

KQ D K bO 2k D K

B

B

Hence, we obtain the formula

1

call

E .ST K/C I ST B

.Cu&i

/D

.1 C r/T

p k B 2

C

Q I ST < BQ :

E .ST K/

C

1 p

S0

Both expectations on the right now only involve the binomial distribution with parameters p and T . They can be computed as in Example 5.42, and so we get the explicit

formula

call

/

.Cu&i

1

D

.1 C r/T

X

nk

nD0

p k B 2

C

1 p

S0

T

X

!

T

T n

Q C .p /T n .1 p /n

.S0 bO T 2n K/

nDnk C1

!

T

T n

;

}

Example 5.50 (Up-and-out call option). Consider an up-and-out call option with payoff

0

if max0tT S t B;

call

Cu&o D

C

otherwise,

.ST K/

301

where K > 0 is the strike price and B > S0 _ K is an upper barrier for the stock

price. As in the preceding example, we assume that B D S0 .1 C b/k for some k 2 N.

Let

C call WD .ST K/C

denote the corresponding plain vanilla call, whose arbitrage-free price is given by

.C call / D

1

E .ST K/C :

.1 C r/T

call

call

Since C call D Cu&o

C Cu&i

, we get from Example 5.49 that

call

call

/ D .C call / .Cu&i

/

.Cu&o

1

D

E .ST K/C I ST < B

.1 C r/T

p k B 2

C

Q

Q

E

.S

K/

I

S

<

B

:

T

T

1 p

S0

Example 5.49.

}

Exercise 5.6.1. Derive a formula for the arbitrage-free price of a down-and-in put

option with payoff

0

if min0tT S t > B;

put

Cd&i D

C

otherwise,

.K ST /

where K > 0 is the strike price and B < S0 is a lower barrier for the stock price.

Then compute the price of the option for the following specific parameter values:

T D 3;

S0 D 100;

a D 0:1;

r D 0:05;

B D 70;

K D 90:

Example 5.51 (Lookback put option). A lookback put option corresponds to the contingent claim

put

Cmax

WD max S t ST I

0tT

put

see Example 5.23. In the CRR model, the discounted arbitrage-free price of Cmax is

given by

1

put

.Cmax

/D

E max S t S0 :

T

0tT

.1 C r/

302

E

T

X

max S t D S0

bO k P MT D k :

0tT

kD0

P MT D k D

P MT D k; ZT D k l

l0

X

l0

1 p k k C l C 1

P ZT C1 D 1 k l

1 p 1 p

T C1

1 p k 1

E ZT C1 I ZT C1 1 k :

1 p 1 p T C 1

put

.Cmax

/ C S0

T

p k

X

S0

Ok

D

E ZT C1 I ZT C1 1 k :

b

1 p

.1 C r/T .1 p /.T C 1/

kD0

As before, one can give explicit formulas for the expectations occurring on the right.

}

Exercise 5.6.2. Derive a formula for the price of a lookback call option with payoff

ST min S t :

0tT

5.7

In practice, a huge number of trading periods may occur between the current time

t D 0 and the maturity T of a European contingent claim. Thus, the computation

of option prices in terms of some martingale measure may become rather elaborate.

On the other hand, one can hope that the pricing formulas in discrete time converge

to a transparent limit as the number of intermediate trading periods grows larger and

larger. In this section, we will formulate conditions under which such a convergence

occurs.

Throughout this section, T will not denote the number of trading periods in a fixed

discrete-time market model but rather a physical date. The time interval 0; T will

T 2T

; N ; : : : ; NNT , and the date kT

be divided into N equidistant time steps N

N will correspond to the k th trading period of an arbitrage-free market model. For simplicity, we

303

will assume that each market model contains a riskless bond and just one risky asset.

In the N th approximation, the risky asset will be denoted by S .N / , and the riskless

bond will be defined by a constant interest rate rN > 1.

The question is whether the prices of contingent claims in the approximating market

models converge as N tends to infinity. Since the terminal values of the riskless bonds

should converge, we assume that

lim .1 C rN /N D e rT ;

N "1

where r is a finite constant. This condition is in fact equivalent to the following one:

lim N rN D r T:

N "1

.N /

Let us now consider the risky assets. We assume that the initial prices S0 do

.N /

D S0 for some constant S0 > 0. The prices Sk.N /

not depend on N , i.e., S0

are random variables on some probability space .N ; F .N / ; PN /, where PN is a

risk-neutral measure for each approximating market model, i.e., the discounted price

process

Sk.N /

.N /

; k D 0; : : : ; N;

Xk WD

.1 C rN /k

is a PN -martingale with respect to the filtration Fk.N / WD .S1.N / ; : : : ; Sk.N / /. Our

remaining conditions will be stated in terms of the returns

.N /

.N /

Rk

WD

Sk

.N /

Sk1

.N /

Sk1

k D 1; : : : ; N:

.N /

.N /

dent under PN and satisfy

.N /

1 < N Rk N ;

are indepen-

k D 1; : : : ; N;

lim N D lim N D 0:

N "1

N "1

.N /

Second, we assume that the variances varN .Rk / under PN are such that

2

N

N

1 X

WD

var.Rk.N / / ! 2 2 .0; 1/:

T

N

kD1

The following result can be regarded as a multiplicative version of the central limit

theorem.

304

.N /

Theorem 5.52. Under the above assumptions, the distributions of SN under PN

converge weakly to the log-normal distribution with parameters log S0 C rT 12 2 T

p

and T , i.e., to the distribution of

1 2

ST WD S0 exp WT C r T ;

(5.31)

2

where WT has a centered normal law N.0; T / with variance T .

Proof. We may assume without loss of generality that S0 D 1. Consider the Taylor

expansion

1

log.1 C x/ D x x 2 C .x/ x 2

(5.32)

2

where the remainder term is such that

j.x/j .; /

for 1 < x ,

.N /

SN

N

Y

.N /

.1 C Rk /;

kD1

this yields

.N /

log SN

N

X

.N /

Rk

kD1

1 .N / 2

.Rk / C N ;

2

where

jN j .N ; N /

N

X

.Rk.N / /2 :

kD1

R .N / D r , and it follows that

Since PN is a martingale measure, we have EN

N

k

jN j .N ; N /

EN

N

X

2

.var.Rk.N / / C rN

/ ! 0:

kD1

to the Dirac measure 0 . Slutskys theorem, as stated in Appendix A.6, asserts that it

suffices to show that the distributions of

ZN WD

N

X

kD1

N

X

1

.N /

Rk.N / .Rk.N / /2 DW

Yk

2

kD1

305

converge weakly to the normal law N.rT 12 2 T; 2 T /. To this end, we will check

that the conditions of the central limit theorem in the form of Theorem A.37 are satisfied.

Note that

1 2

.N /

! 0

max jYk j N C N

2

1kN

for N WD jN j _ jN j, and that

1 2

1

2

ZN D N rN . N

T C N rN

/ ! r T 2 T:

EN

2

2

Finally,

var.ZN / ! 2 T;

N

N

X

.N /

p2

EN

jRk jp N

kD1

N

X

.N /

EN

.Rk /2 ! 0:

kD1

Remark 5.53. The assumption of independent returns in Theorem 5.52 can be relaxed. Instead of Theorem A.37, we can apply a central limit theorem for martingales

under suitable assumptions on the behavior of the conditional variances

.n/

var. Rk j Fk1 /I

N

Example 5.54. Suppose the approximating model in the N th stage is a CRR model

with interest rate

rT

rN D

;

N

.N /

and with returns Rk , which can take the two possible values aN and bN ; see Section 5.5. We assume that

aO N D 1 C aN D e

p

T =N

and

bON D 1 C bN D e

p

T =N

p

N rN ! 0;

p

p

N aN ! T ;

p

p

N bN ! T

as N " 1, (5.33)

306

we have aN < rN < bN for large enough N . Theorem 5.39 yields that the N th

model is arbitrage-free and admits a unique equivalent martingale measure PN . The

measure PN is characterized by

.N /

PN Rk

D bN DW pN

D

rN aN

;

bN aN

1

D :

lim pN

2

N "1

.N /

R

Moreover, EN

k

N

X

kD1

D rN , and we get

2

2

2

var.Rk.N / / D N.pN

bN C .1 pN

/aN

rN

/ ! 2 T

N

the risky assets terminal value. In each approximating model, this corresponds to a

contingent claim

.N /

C .N / D f .SN /:

Corollary 5.55. If f is bounded and continuous, the arbitrage-free prices of C .N /

calculated under PN converge to a discounted expectation with respect to a lognormal distribution, which is often called the BlackScholes price. More precisely,

lim

N "1

EN

C .N /

.1 C rN /N

D e rT E f .ST /

e rT

D p

2

f .S0 e

(5.34)

p

T yCrT 2 T =2

/e y

2 =2

dy;

1

This convergence result applies in particular to the choice f .x/ D .K x/C corresponding to a European put option with strike K. Since the put-call parity

EN

.N /

.SN K/C

.1 C rN /N

D

EN

.N /

.K SN /C

.1 C rN /N

C S0

K

.1 C rN /N

holds for each N , the convergence (5.34) is also true for a European call option with

the unbounded payoff profile f .x/ D .x K/C .

307

Example 5.56 (BlackScholes formula for the price of a call option). The limit of the

.N /

arbitrage-free prices of C .N / D .SN K/C is given by v.S0 ; T /, where

e rT

v.x; T / D p

2

.xe

p

T yCrT 2 T =2

K/C e y

2 =2

dy:

1

y

x

C .r 12 2 /T

log K

DW d .x; T / DW d :

p

T

x

p

C .r C 12 2 /T

log K

dC WD dC .x; T / WD d .x; T / C T D

;

p

T

Rx

2

and let us denote by .x/ D .2/1=2 1 e y =2 dy the distribution function of

the standard normal distribution. Then

Z C1

p

x

2

e .y T / =2 dy e rT K.1 .d //;

v.x; T / D p

2 d

and we arrive at the BlackScholes formula for the price of a European call option

with strike K and maturity T

v.x; T / D x .dC .x; T // e rT K.d .x; T //:

See Figure 5.5 for the plot of the function v.x; t /.

(5.35)

}

Remark 5.57. For fixed x and T , the BlackScholes price of a European call option

increases to the upper arbitrage bound x as " 1. In the limit # 0, we obtain the

lower arbitrage bound .x e rT K/C ; see Remark 1.37.

}

The following proposition gives a criterion for the convergence (5.34) in case f is

not necessarily bounded and continuous. It applies in particular to f .x/ D .x K/C ,

and so we get an alternative proof for the convergence of call option prices to the

BlackScholes price.

Proposition 5.58. Let f W .0; 1/ ! R be measurable, continuous a.e., and such

that jf .x/j c .1 C x/q for some c 0 and 0 q < 2. Then

.N /

EN

f .SN

/ ! E f .ST /;

308

2K

0

K

t

T

Figure 5.5. The BlackScholes price v.x; t / of a European call option .ST K/C plotted

as a function of the initial spot price x D S0 and the time to maturity t.

N

Y

.N /

.SN /2 D log

log EN

.N /

kD1

N

X

.N /

.var.1 C Rk / C EN

1 C Rk 2 /

N

.N /

log.var.Rk / C .1 C rN /2 /

kD1

2

2

T C 2N rN C N rN

C cQ

N

N

X

2 2

.var.Rk.N / / C 2jrN j C rN

/

kD1

Q Thus,

.N / 2

.SN

/ < 1:

sup EN

N

With this property established, the assertion follows immediately from Theorem 5.52

and the Corollaries A.46 and A.47, but we also give the following more elementary

proof. To this end, we may assume that q > 0, and we define p WD 2=q > 1. Then

.N / p

.N / 2

jf .SN

/j c p sup EN

.1 C SN

/ < 1;

sup EN

N

Lemma 5.59. Suppose .

N /N 2N is a sequence of probability measures on R converging weakly to

. If f is a measurable and

-a.e. continuous function on R such

309

that

Z

c WD sup

jf jp d

N < 1

N 2N

then

Z

f d

N !

f d

:

a bounded and

-a.e. continuous function for each k > 0. Clearly,

Z

Z

Z

f d

N D fk d

N C .f k/C d

N :

Due to part (e) of the portmanteau

R theorem in the form of Theorem A.39, the first

integral on the right converges to fk d

as N " 1. Let us consider the second term

on the right

Z

Z

Z

1

c

C

.f k/ d

N

f d

N p1 f p1 f d

N p1 ;

k

k

f >k

uniformly in N . Hence,

Z

Z

Z

fk d

N lim inf f d

N

fk d

D lim

N "1

N "1

f d

N

lim sup

fk d

C

N "1

fk d

%

c

k p1

f d

, and convergence follows.

Let us now continue the discussion of the BlackScholes price of a European call

option where f .x/ D .x K/C . We are particularly interested how it depends on the

various model parameters. The dependence on the spot price S0 D x can be analyzed

via the x-derivatives of the function v.t; x/ appearing in the BlackScholes formula

(5.35). The first derivative

.x; t / WD

@

v.x; t / D .dC .x; t //

@x

(5.36)

is called the options Delta; see Figure 5.6. In analogy to the formula for the hedging

strategy in the binomial model obtained in Proposition 5.44, .x; t / determines the

Delta hedging portfolio needed for a replication of the call option in continuous

time, as explained in (5.45) below.

The Gamma of the call option is given by

.x; t / WD

@2

1

@

.x; t / D 2 v.x; t / D '.dC .x; t // p I

@x

@x

x t

(5.37)

310

1

2K

0

K

t

T

Figure 5.6. The Delta .x; t / of the BlackScholes price of a European call option.

p

2

see Figure 5.7. Here '.x/ D 0 .x/ D e x =2 = 2 stands as usual for the density

of the standard normal distribution. Large Gamma values occur in regions where the

Delta changes rapidly, corresponding to the need for frequent readjustments of the

Delta hedging portfolio. Note that is always strictly positive. It follows that v.x; t /

is a strictly convex function of its first argument.

2K

0

K

t

T

Exercise 5.7.1. Prove the formulas (5.36) and (5.37) for Delta and Gamma of a European call option.

}

311

jv.x; t / v.y; t /j jx yj:

Thus, the total change of the option values is always less than a corresponding change

in the asset prices. On the other hand, the strict convexity of x 7! v.x; t / together

with (A.1) yields that for t > 0 and z > y

v.y; t / v.0; t /

v.y; t /

v.z; t / v.y; t /

>

D

zy

y0

y

and hence

zy

v.z; t / v.y; t /

>

:

v.y; t /

y

xy

v.x; t / v.y; t /

<

v.y; t /

y

for x < y. Thus, the relative change of option prices is larger in absolute value than

the relative change of asset values. This fact can be interpreted as the leverage effect

for call options; see also Example 1.43.

}

Another important parameter is the Theta

.x; t / WD

@

x

v.x; t / D p '.dC .x; t // C Kr e rt .d .x; t //I

@t

2 t

(5.38)

see Figure 5.8. The fact > 0 corresponds to our general observation, made in Example 5.35, that arbitrage-free prices of European call options are typically increasing

functions of the maturity.

Exercise 5.7.2. Prove the formula (5.38) for the Theta of a European call option.

Then show that the parameters , , and are related by the equation

1

.x; t / D rx .x; t / C 2 x 2 .x; t / r v.x; t /

2

when t > 0.

(5.39)

}

Equation (5.39) implies that, for .x; t / 2 .0; 1/ .0; 1/, the function v solves the

partial differential equation

@v

@2 v

@v

1

D rx

C 2 x 2 2 rv;

@t

@x

2

@x

often called the BlackScholes equation. Since

v.x; t / ! f .x/ D .x K/C

as t # 0,

(5.40)

(5.41)

v.x; t / is a solution of the Cauchy problem defined via (5.40) and (5.41). This fact is

not limited to call options, it remains valid for all reasonable payoff profiles f .

312

2K

0

K

t

T 0

Figure 5.8. The Theta .x; t /.

c.1 C x/p for some c; p 0, and define

Z 1

p

e rt

2

2

f .xe t yCrt t=2 /e y =2 dy;

u.x; t / WD e rt E f .S t / D p

2 1

where S t D x exp. W t C rt 2 t =2/ and W t has law N.0; t / under P . Then

u solves the Cauchy problem defined by the BlackScholes equation (5.40) and the

initial condition lim t#0 u.x; t / D f .x/, locally uniformly in x.

The proof of Proposition 5.61 is the content of the next exercise.

Exercise 5.7.3. In the context Proposition 5.61, use the formula (2.26) for the density

of a log-normally distributed random variable to show that

Z 1

log y rt C 2 t =2 log x

1

f .y/ dy;

p '

p

E f .S t / D

y t

t

0

p

2

where '.x/ D e x =2 = 2. Then verify the validity of (5.40) by differentiating

under the integral. Use the bound jf .x/j c.1 C x/p for some c; p 0 to justify

the interchange of differentiation and integration and to verify the initial condition

}

lim t#0 u.x; t / D f .x/.

Recall that the BlackScholes price v.S0 ; T / was obtained as the expectation of the

discounted payoff e rT .ST K/C under the measure P . Thus, at a first glance, it

may come as a surprise that the Rho of the option,

%.x; t / WD

@

v.x; t / D Kt e rt .d .x; t //;

@r

(5.42)

313

is strictly positive, i.e., the price is increasing in r; see Figure 5.9. Note, however, that

the measure P depends itself on the interest rate r, since E e rT ST D S0 . In a

simple one-period model, we have already seen this effect in Example 1.43.

2K

0

K

t

T

The parameter is called the volatility. As we have seen, the BlackScholes price

of a European call option is an increasing function of the volatility, and this is reflected

in the strict positivity of

V .x; t / WD

p

@

v.x; t / D x t '.dC .x; t //I

@

(5.43)

see Figure 5.10. The function V is often called the Vega of the call option price, and

the functions , , , %, and V are usually called the Greeks (although vega does

not correspond to a letter of the Greek alphabet).

Exercise 5.7.4. Prove the respective formulas (5.42) and (5.43) for Rho and Vega of

a European call option. Then derive formulas for the options Vanna,

@

@V

@2 v

D

D

;

@x@

@

@x

and the options Volga, which is also called Vomma,

@2 v

@V

:

D

2

@

@

Let us conclude this section with some informal comments on the dynamic picture

behind the convergence result in Theorem 5.52 and the pricing formulas in Example 5.56 and Proposition 5.58. The constant r is viewed as the interest rate of a

riskfree savings account

S t0 D e rt ; 0 t T:

314

2K

0

K

t

T 0

Figure 5.10. The Vega V .x; t /.

The prices of the risky asset in each discrete-time model are considered as a contin.N /

.N /

WD Sk.N / at the dates t D kT

uous process SQ .N / D .SQ t /0tT , defined as SQ t

N ,

and by linear interpolation in between. Theorem 5.52 shows that the distributions of

.N /

SQ t converge for each fixed t weakly to the distribution of

1

S t D S0 exp W t C r 2 t ;

2

(5.44)

where W t has a centered normal distribution with variance t . In fact, one can prove

convergence in the much stronger sense of a functional central limit theorem: The

laws of the processes SQ .N / , considered as C 0; T -valued random variables on

.N ; F .N / ; PN /, converge weakly to the law of a geometric Brownian motion S D

.S t /0tT , where each S t is of the form (5.44), and where the process W D.W t /0tT

is a standard Brownian motion or Wiener process. A Wiener process is characterized

by the following properties:

W0 D 0 almost surely,

t 7! W t is continuous,

W t1 W t0 ; : : : ; W tn W tn1

are independent and have normal distributions N.0; ti ti 1 /;

see, e.g., [171]. This multiplicative version of a functional central limit theorem follows as above if we replace the classical central limit theorem by Donskers invariance

principle; for details see, e.g., [99]. Sample paths of Brownian motion and geometric

Brownian motion can be found in Figures 5.11 and 5.12.

315

0.5

0.5

0.5

mathematical finance. In order to describe the model more explicitly, we denote by

W D .W t /0tT the coordinate process on the canonical path space D C 0; T ,

defined by W t .!/ D !.t /, and by .F t /0tT the filtration given by F t D .Ws I s

t /. There is exactly one probability measure P on .; FT / such that W is a Wiener

process under P , and it is called the Wiener measure. Let us now model the price

process of a risky asset as a geometric Brownian motion S defined by (5.44). The

discounted price process

X t WD

St

2

D S0 e W t t=2 ;

rt

e

0 t T;

316

E X t j Fs D Xs E e .Wt Ws /

2 .ts/=2

D Xs

that property.

As in discrete time, uniqueness of the equivalent martingale measure implies completeness of the model. Let us sketch the construction of the replicating strategy for

a given European option with reasonable payoff profile f .ST /, for example a call

option with strike K. At time t the price of the asset is S t .!/, the remaining time to

maturity is T t , and the discounted price of the option is given by

V t .!/ D e rt u.S t .!/; T t /;

where u is the function defined in Proposition 5.61. The process V D .V t /0tT can

be viewed as the value process of the trading strategy D . 0 ; / defined by

t D .S t ; T t /;

t0 D e rt u.S t ; T t / t X t ;

(5.45)

where D @u=@x is the options Delta. Indeed, if we view as the number of shares

in the risky asset S and 0 as the number of shares in the riskfree savings account

S t0 D e rt , then the value of the resulting portfolio in units of the numraire is given

by

V t D t X t C t0 D e rt . t S t C t0 S t0 /:

The strategy replicates the option since

VT WD lim e rt u.S t ; T t / D e rT f .ST / D

t"T

f .ST /

;

ST0

due to Proposition 5.61. Moreover, its initial cost is given by the BlackScholes price

Z

p

e rT 1

2

2

f .xe T yCrT T =2 /e y =2 dy:

V0 D u.S0 ; T / D e rT E f .ST / D p

2 1

It remains to show that the strategy is self-financing in the sense that changes in

the portfolio value are only due to price changes in the underlying assets and do not

require any additional capital. To this end, we use Its formula

dF .W t ; t / D

1 @2 F

@F

@F

C

.W t ; t / d W t C

.W t ; t / dt

@x

2 @x 2

@t

for a smooth function F , see, e.g., [171] or, for a strictly pathwise approach, [117].

Applied to the function F .x; t / D exp. x C rt 2 t =2/, it shows that the price

process S satisfies the stochastic differential equation

dS t D S t d W t C rS t dt:

(5.46)

317

Thus, the infinitesimal return dS t =S t is the sum of the safe return r dt and an additional noise term with zero expectation under P . The strength of the noise is

measured by the volatility parameter . Similarly, we obtain

dX t D X t d W t D e rt .dS t rS t dt /:

(5.47)

F .x; t / D e rt u.exp.x C rt 2 t =2/; T t /

and using (5.46), we obtain

d V t D e rt

1

@u

@2 u @u

.S t ; t / dS t C e rt 2 S t2 2

ru .S t ; t / dt:

@x

2

@x

@t

The BlackScholes partial differential equation (5.40) shows that the term in parenthesis is equal to rS t @u=@x, and we obtain from (5.47) that

d Vt D

@u

.S t ; t / dX t D t dX t :

@x

More precisely,

V t D V0 C

s dXs ;

0

where the integral with respect to X is defined as an It integral, i.e., as the limit of

non-anticipating Riemann sums

X

ti .X tiC1 X ti /

ti 2Dn ; ti t

along an increasing sequence .Dn / of partitions of the interval 0; T ; see, e.g., [117].

Thus, the It integral can be interpreted in financial terms as the cumulative net gain

generated by dynamic hedging in the discounted risky asset as described by the hedging strategy . This fact is an analogue of property (c) in Proposition 5.7, and in this

sense D . 0 ; / is a self-financing trading strategy in continuous time. Similarly,

we obtain the following continuous-time analogue of (5.5), which describes the undiscounted value of the portfolio as a result of dynamic trading both in the undiscounted

risky asset and the riskfree asset

Z t

Z t

rt

s dSs C

s0 dSs0 :

e V t D V0 C

0

Perfect replication also works for exotic options C.S/ defined by reasonable functionals C on the path space C 0; T , due to a general representation theorem for such

functionals as It integrals of the underlying Brownian motion W or, via (5.47), of the

318

that the arbitrage-free prices of the options C.S .N / /, computed as discounted expectations under the measure PN , converge to the discounted expectation

e rT E C.S/

under the Wiener measure P .

On the other hand, the discussion in Section 5.6 suggests that the prices of certain

exotic contingent claims, such as barrier options, can be computed in closed form as

the BlackScholes price for some corresponding payoff profile of the form f .ST /.

This is illustrated by the following example, where the price of an up-and-in call is

computed in terms of the distribution of the terminal stock price under the equivalent

martingale measure.

Example 5.62 (BlackScholes price of an up-and-in call option). Consider an upand-in call option

call

Cu&i

.S/ D

0

otherwise,

where B > S0 _ K denotes a given barrier, and where K > 0 is the strike price. As

approximating models we choose the CRR models of Example 5.54. That is, we have

interest rates

rT

rN D

N

and parameters aN and bN defined by

aO N D 1 C aN D e

p

T =N

and

bON D 1 C bN D e

p

T =N

for some given > 0. Applying the formula obtained in Example 5.49 yields

call

EN

Cu&i

.S .N / /

.N /

.N /

.SN

K/C I SN

BN

D EN

p kN B 2

N

.N /

.N /

N

EN

.SN

KQ N /C I SN

< BQ N ;

C

1 pN

S0

where

l pN

Bm

kN D

p log

S0

T

k , B

2

O kN Q

is the smallest integer k such that B S0 bON

N WD S0 bN , BN WD S0 =BN D

kN

S0 aO N

, and

S 2

0

2kN

DK

:

KQ N D K bON

BN

319

Then we have

BN & B; BQ N % BQ D

S02

;

B

and

S 2

0

KQ N % KQ D K

:

B

.N /

EN

.SN

.N /

K/C I SN

.N /

BN D EN

.SN

.N /

K/C I SN

B

! E .ST K/C I ST B ;

due to Proposition 5.58. Combining the preceding argument with the fact that

.N /

PN KQ N SN KQ ! 0

.N /

EN

.SN

.N /

.N /

.N /

KQ N /C I SN < BQ N D EN

.SN KQ N /C I SN < BQ

Q C I ST < BQ :

! E .ST K/

Next we note that for constants c; d > 0

1

2c

cx 2 C 1 e dx

D

log dx

d;

d

x#0 x

e 1 cx 2

lim

B 2r2 1

p kN

N

!

:

1 pN

S0

Thus, we may conclude that the arbitrage-free prices

1

E C call .SQ .N / /

.1 C rN /N N u&i

in the N th approximating model converge to

B 2r2 C1

rT

C

C

Q

Q

E .ST K/ I ST < B :

E .ST K/ I ST B C

e

S0

The expectations occurring in this formula are integrals with respect to a log-normal

distribution and can be explicitly computed as in Example 5.56. Moreover, our limit

is in fact equal to the BlackScholes price of the up-and-in call option: The functional

call

. / is continuous in each path in C 0; T whose maximum is different from the

Cu&i

value B, and one can show that these paths have full measure for the law of S uncall

der P . Hence, Cu&i

. / is continuous P S 1 -a.e., and the functional version of

Proposition 5.58 yields

call Q .N /

call

Cu&i

.S / ! E Cu&i

.S/ ;

EN

so that our limiting price must coincide with the discounted expectation on the right.

}

320

Remark 5.63. Let us assume, more generally, that the price process S is defined by

S t D S0 e W t Ct ;

0 t T;

for some 2 R. Applying Its formula as in (5.46), we see that S is governed by the

stochastic differential equation

dS t D S t d W t C bS t dt

with b D C 12 2 . The discounted price process is given by

X t D S0 e W t C.r/t D S0 e Wt

2 t=2

the measure P P defined by the density

dP

2

D e

WT

T =2 :

dP

In fact, P is the unique equivalent martingale measure for X. We can now repeat

the arguments above to conclude that the cost of perfect replication for a contingent

claim C.S/ is given by

}

e rT E C.S/ :

Even in the context of simple diffusion models such as geometric Brownian motion,

however, completeness is lost as soon as the future behavior of the volatility parameter

is unknown. If, for instance, volatility itself is modeled as a stochastic process, we

are facing incompleteness. Thus, the problems of pricing and hedging in discrete-time

incomplete markets as discussed in this book reappear in continuous time. Other versions of the invariance principle may lead to other classes of continuous-time models

with discontinuous paths, for instance to geometric Poisson or Lvy processes. Discontinuity of paths is another important source of incompleteness. In fact, this has

already been illustrated in this book, since discrete-time models can be regarded as

stochastic processes in continuous time, where jumps occur at predictable dates.

Chapter 6

So far, we have studied European contingent claims whose payoff is due at a fixed

maturity date. In the case of American options, the buyer can claim the payoff at any

time up to the expiration of the contract.

First, we take the point of view of the seller, whose aim is to hedge against all

possible claims of the buyer. In Section 6.1, this problem is solved under the assumption of market completeness, using the Snell envelope of the contingent claim.

The buyer tries to choose the best date for exercising the claim, contingent on the

information available up to that time. Since future prices are usually unknown, a formulation of this problem will typically involve subjective preferences. If preferences

are expressed in terms of expected utility, the choice of the best exercise date amounts

to solving an optimal stopping problem. In the special case of a complete market

model, any exercise strategy which maximizes the expected payoff under the unique

equivalent martingale measure turns out to be optimal even in an almost sure sense.

In Section 6.3, we characterize the set of all arbitrage-free prices of an American

contingent claim in an incomplete market model. This involves a lower Snell envelope

of the claim, which is analyzed in Section 6.5, using the fact that the class of equivalent

martingale measures is stable under pasting. This notion of stability under pasting is

discussed in Section 6.4 in a general context, and in Section 6.5 we point out its

connection with the time-consistency of dynamic risk measures. This connection will

be discussed systematically in Chapter 11. The results on lower Snell envelopes can

also be regarded as a solution to the buyers optimal stopping problem in the case

where preferences are described by robust Savage functionals. Moreover, these results

will be used in the theory of superhedging of Chapter 7.

6.1

Throughout this chapter we will continue to use the setting described in Section 5.1.

We start by introducing the Doob decomposition of an adapted process and the notion

of a supermartingale.

Proposition 6.1. Let Q be a probability measure on .; FT /, and suppose that Y is

a stochastic process that is adapted to the filtration .Ft / tD0;:::;T and satisfies Y t 2

L1 .Q/ for all t . Then there exists a unique decomposition

Y D M A;

(6.1)

322

is predictable. The decomposition (6.1) is called the Doob decomposition of Y with

respect to the probability measure Q.

Proof. Define A by

A t A t1 WD EQ Y t Y t1 j F t1 for t D 1; : : : ; T .

(6.2)

with the required properties must satisfy (6.2), so the uniqueness of the decomposition

follows.

Definition 6.2. Let Q be a probability measure on .; FT / and suppose that Y is an

adapted process such that Y t 2 L1 .Q/ for all t . Denote by Y D M A the Doob

decomposition of Y .

(a) Y is called a Q-supermartingale if A is increasing.

(b) Y is called a Q-submartingale if A is decreasing.

Clearly, a process is a martingale if and only if it is both a supermartingale and a

submartingale, i.e., if and only if A 0. The following exercise gives equivalent

characterizations of the supermartingale property of a process Y .

Exercise 6.1.1. Let Y be an adapted process with Y t 2 L1 .Q/ for all t . Show that

the following conditions are equivalent:

(a) Y is a Q-supermartingale.

(b) Ys EQ Y t j Fs for 0 s t T .

(c) Y t1 EQ Y t j F t1 for t D 1; : : : ; T .

(d) Y is a Q-submartingale.

}

T t we have Y tCs D 0 Q-a.s. on Y t D 0.

We now return to the market model introduced in Section 5.1. An American option,

or American contingent claim, corresponds to a contract which is issued at time 0 and

which obliges the seller to pay a certain amount C 0 if the buyer decides at time

to exercise the option. The choice of the exercise time is entirely up to the buyer,

except that the claim is automatically exercised at the expiration date of the claim.

The American contingent claim can be exercised only once: It becomes invalid as

soon as the payoff has been claimed by the buyer. This concept is formalized as

follows:

Definition 6.3. An American contingent claim is a non-negative adapted process C D

.Ct / tD0;:::;T on the filtered space .; .F t / tD0;:::;T /.

323

For each t , the random variable C t is interpreted as the payoff of the American

contingent claim if the claim is exercised at time t . The time horizon T plays the role

of the expiration date of the claim. The possible exercise times for C are not limited

to fixed deterministic times t 2 0; : : : ; T ; the buyer may exercise the claim in a way

which depends on the scenario ! 2 of the market evolution.

Definition 6.4. An exercise strategy for an American contingent claim C is an FT measurable random variable taking values in 0; : : : ; T . The payoff obtained by

using is equal to

C .!/ WD C.!/ .!/; ! 2 :

Example 6.5. An American put option on the i th asset and with strike K > 0 pays

the amount

put

C t WD .K S ti /C

if it is exercised at time t . The payoff at time t of the corresponding American call

option is given by

C tcall WD .S ti K/C :

Clearly, the American call option is out of the money (i.e., has zero payoff) if the

corresponding American put is in the money (i.e., has non-zero payoff). It is therefore a priori clear that the respective owners of C put and C call will usually exercise

their claims at different times. In particular, there will be no put-call parity for American options.

}

Similarly, one defines American versions of most options mentioned in the examples of Section 5.3. Clearly, the value of an American option is at least as high as the

value of the corresponding European option with maturity T .

Remark 6.6. It should be emphasized that the concept of American contingent claims

can be regarded as a generalization of European contingent claims: If C E is a European contingent claim, then we can define a corresponding American claim C A by

0

if t < T ,

A

Ct D

(6.3)

E

if t D T .

C

}

Example 6.7. A Bermuda option can be exercised by its buyer at each time of a

predetermined subset T 0; : : : ; T . For instance, a Bermuda call option pays the

amount .S ti K/C if it is exercised at some time t 2 T . Thus, a Bermuda option

is a financial instrument between an American option with T D 0; : : : ; T and a

European option with T D T , just as Bermuda lies between America and Europe;

hence the name Bermuda option. A Bermuda option can be regarded as a particular

}

American option C that pays the amount C t D 0 for t T .

324

The process

Ht D

Ct

;

S t0

t D 0; : : : ; T;

with C . As far as the mathematical theory is concerned, the discounted American

claim H will be the primary object. For certain examples it will be helpful to keep

track of the numraire and, thus, of the payoffs C t prior to discounting.

In this section, we will analyze the theory of hedging American claims in a complete

market model. We will therefore assume throughout this section that the set P of

equivalent martingale measures consists of one single element P

P D P :

Under this assumption, we will construct a suitable trading strategy that permits the

seller of an American claim to hedge against the buyers discounted claim H . Let us

first try to characterize the minimal amount of capital U t which will be needed at time

t 2 0; : : : ; T . Since the choice of the exercise time is entirely up to the buyer, the

seller must be prepared to pay at any time t the current payoff H t of the option. This

amounts to the condition U t H t . Moreover, the amount U t must suffice to cover

the purchase of the hedging portfolio for the possible payoffs Hu for u > t . Since the

latter condition is void at maturity, we require

UT D HT :

At time T 1, our first requirement on UT 1 reads UT 1 HT 1 . The second

requirement states that the amount UT 1 must suffice for hedging the claim HT in

case the option is not exercised before time T . Due to our assumption of market

completeness, the latter amount equals

E HT j FT 1 D E UT j FT 1 :

Thus,

UT 1 WD HT 1 _ E UT j FT 1

is the minimal amount that fulfills both requirements. Iterating this argument leads to

the following recursive scheme for U t :

UT WD HT ;

U t WD H t _ E U tC1 j F t for t D T 1; : : : ; 0.

(6.4)

Definition 6.8. The process U P WD U defined by the recursion (6.4) is called the

Snell envelope of the process H with respect to the measure P .

325

Example 6.9. Let H E be a discounted European claim. Then the Snell envelope

with respect to P of the discounted American claim H A associated with H E via

(6.3) satisfies

U tP D E HTA j F t D E H E j F t :

Thus, U is equal to the value process of a replicating strategy for H E .

.; FT / and for any adapted process H that satisfies the following integrability condition:

(6.5)

H t 2 L1 .Q/ for t D 0; : : : ; T .

In our finite-time setting, this condition is equivalent to

EQ max jH t j < 1:

tT

For later applications, the following proposition is stated for a general measure Q.

Proposition 6.10. Let H be an adapted process such that (6.5) holds. Then the Snell

envelope U Q of H with respect to Q is the smallest Q-supermartingale dominating

H : If UQ is another Q-supermartingale such that UQ t H t Q-a.s. for all t , then

Q

UQ t U t Q-a.s. for all t .

Q

U Q is indeed a supermartingale. If UQ is another supermartingale dominating H , then

Q

UQ T HT D UT . We now proceed by backward induction on t . If we already know

Q

that UQ t U t , then

Q

UQ t1 EQ UQ t j F t1 EQ U t j F t1 :

Q

Q

UQ t1 H t1 _ EQ U t j F t1 D U t1 ;

Proposition 6.10 illustrates how the seller can (super-) hedge a discounted American claim H by using the Doob decomposition

U tP D M t A t ;

t D 0; : : : ; T;

increasing, and .A t / tD1;:::;T is predictable. Since we assume the completeness of

326

the market model, Theorem 5.38 yields the representation of the martingale M as the

stochastic integral of a suitable d -dimensional predictable process

M t D U0P C

t

X

k .Xk Xk1 /;

t D 0; : : : ; T:

(6.6)

kD1

It follows that

M t U tP H t

for all t .

0

such that D . 0 ; / becomes a self-financing

trading strategy with initial investment U0P , we obtain a (super-) hedge for H ,

namely a self-financing trading strategy whose value process V satisfies

Vt Ht

for all t .

(6.7)

Thus, U tP may be viewed as the resulting capital at each time t if we use the selffinancing strategy , combined with a refunding scheme where we withdraw suc

cessively the amounts defined by the increments of A. In fact, U tP is the minimal

investment at time t for which one can purchase a hedging strategy such that (6.7)

holds. This follows from our next result.

Then there exists a d -dimensional predictable process such that

u

X

U tP C

k .Xk Xk1 / Hu

(6.8)

kDtC1

satisfies (6.8) in place of U tP is such that

UQ t U tP

P -a.s.

time t up to maturity.

Proof. Clearly, U P satisfies (6.8) for as in (6.6). Now suppose that UQ t is F t measurable, that Q is predictable, and that

Vu WD UQ t C

u

X

kDtC1

P for some u. Since our market model is

by assumption, so assume VuC1 UuC1

complete, Theorem 5.37 implies that Q is bounded. Hence, we get

E VuC1 Vu j Fu D E QuC1 .XuC1 Xu / j Fu D 0

P -a.s.

327

It follows that

P

j Fu D UuP :

Vu D E VuC1 j Fu Hu _ E UuC1

6.2

In this section, we take the point of view of the buyer of an American contingent

claim. Thus, our aim is to optimize the exercise strategy. It is natural to assume that

the decision to exercise the claim at a particular time t depends only on the market

information which is available at t . This constraint can be formulated as follows:

Definition 6.12. A function W ! 0; 1; : : : ; T [ C1 is called a stopping time

if D t 2 F t for t D 0; : : : ; T .

In particular, the constant function t is a stopping time for fixed t 2 0; : : : ; T .

Exercise 6.2.1. Show that a function W ! 0; 1; : : : ; T [ C1 is a stopping

time if and only if t 2 F t for each t . Show next that, if and are two

stopping times, then the following functions are also stopping times:

^ ;

_ ;

. C / ^ T:

Example 6.13. A typical example of a non-trivial stopping time is the first time at

which an adapted process Y exceeds a certain level c

.!/ WD inft 0 j Y t .!/ c:

In fact,

t D

t

[

Ys c 2 F t

sD0

for t D 0; : : : ; T . This example also illustrates the role of the value C1 in Definition 6.12: We have .!/ D C1 if, for this particular !, the criterion that triggers

is not met for any t 2 0; : : : ; T .

}

Definition 6.14. For any stochastic process Y and each stopping time we denote by

Y the process stopped in

Y t .!/ WD Y t^.!/ .!/ for ! 2 and for all t 2 0; : : : ; T .

It follows from the definition of a stopping time that Y is an adapted process if Y

is. Informally, the following basic theorem states that a martingale cannot be turned

into a favorable game by using a clever stopping strategy. This result is often called

Doobs stopping theorem or the optional sampling theorem. Recall that we assume

F0 D ;; .

328

Theorem 6.15. Let M be an adapted process such that M t 2 L1 .Q/ for each t .

Then the following conditions are equivalent:

(a) M is a Q-martingale.

(b) For any stopping time the stopped process M is a Q-martingale.

(c) EQ M^T D M0 for any stopping time .

Proof. (a) ) (b): Note that

M tC1

M t D .M tC1 M t / I>t :

M t j F t D EQ M tC1 M t j F t I>t D 0:

EQ M tC1

(b) ) (c): This follows simply from the fact that the expectation of M t is constant

in t .

(c) ) (a): We need to show that if t < T , then

EQ MT I A D EQ M t I A

(6.9)

t if ! 2 A,

.!/ WD

T if ! A.

We obtain that

M0 D EQ MT ^ D EQ M t I A C EQ MT I Ac :

Using the constant stopping time T instead of yields that

M0 D EQ MT D EQ MT I A C EQ MT I Ac :

Subtracting the latter identity from the previous one yields (6.9).

Exercise 6.2.2. Let Y D M A be the Doob decomposition with respect to Q of an

adapted process Y with Y t 2 L1 .Q/ (t D 0; : : : ; T ), and let be a stopping time.

Show that Y D M A is the Doob decomposition of Y .

}

Corollary 6.16. Let U be an adapted process such that U t 2 L1 .Q/ for each t . Then

the following conditions are equivalent:

(a) U is a Q-supermartingale.

(b) For any stopping time , the stopped process U is a Q-supermartingale.

329

Proof. If U D M A is the Doob decomposition of U , then Exercise 6.2.2 implies that U D M A is the Doob decomposition of U . This observation and

Theorem 6.15 yield the equivalence of (a) and (b).

Let us return to the problem of finding an optimal exercise time for a discounted

American claim H . We assume that the buyer chooses the possible exercise times

from the set

T WD j is a stopping time with T

of all stopping times which do not take the value C1. Assume that the aim of the

buyer is to choose a payoff from the class H j 2 T which is optimal in the sense

that it has maximal expectation. Thus, the problem is

Maximize E H among all 2 T .

(6.10)

The analysis of the optimal stopping problem (6.10) does not require any properties of

the underlying market model, not even the absence of arbitrage. We may also drop the

positivity assumption on H : All we have to assume is that H is an adapted process

which satisfies

(6.11)

H t 2 L1 .; F t ; P / for all t .

This relaxed assumption will be useful in Chapter 9, and it allows us to include the

interpretation of the optimal stopping problem in terms of the following utility maximization problem:

Remark 6.17. Suppose the buyer uses a preference relation on X WD H j 2 T

which can be represented in terms of a Savage representation

U.H / D EQ u.H /

where Q is a probability measure on .; F /, and u is a measurable or continuous

function; see Section 2.5. Then a natural goal is to maximize the utility U.H / among

all 2 T . This is equivalent to the optimal stopping problem (6.10) for the transformed process HQ t WD u.H t /, and with respect to the measure Q instead of P . This

utility maximization problem is covered by the discussion in this section as long as

HQ t 2 L1 .Q/ for all t . In Remark 6.49 we will discuss the problem of maximizing

the more general utility functionals which appear in a robust Savage representation.

}

Under the assumption (6.11), we can construct the Snell envelope U WD U P of H

with respect to P , i.e., U is defined via the recursive formula

UT WD HT

and

U t WD H t _ E U tC1 j F t ;

min WD mint 0 j U t D H t :

t D T 1; : : : ; 0:

330

Note that min T since UT D HT . As we will see in the following theorem, min

maximizes the expectation of H among all 2 T . In other words, min is a solution

to our optimal stopping problem (6.10). Similarly, we let

.t/

min WD minu t j Uu D Hu ;

which is a member of the set

T t WD 2 T j t :

The following theorem uses the essential supremum of a family of random variables

as explained in Appendix A.5.

Theorem 6.18. The Snell envelope U of H satisfies

U t D E H .t / j F t D ess sup E H j F t :

min

2T t

In particular,

U0 D E Hmin D sup E H :

2T

U t E U j F t E H j F t :

Therefore,

U t ess sup E H j F t :

2T t

min

in turn implied by the identity

U t D E U .t / j F t :

(6.12)

min

Us.t/ WD Us^ .t / ;

min

.t/

and fix some s between t and T . Then Us > Hs on min > s. Hence, P -a.s. on

.t/

min > s

.t/

.t/

.t/

.t/

.t/

Thus, U .t/ is a martingale from time t on

.t/

min

.t/

D E UsC1 j Fs .

331

It follows that

.t/

.t/

E U .t / j F t D E UT j F t D U t

min

D Ut :

Definition 6.19. A stopping time 2 T is called optimal .with respect to P / if

E H D sup E H :

2T

In particular, min is an optimal stopping time in the sense of this definition. The

following result implies that min is in fact the minimal optimal stopping time.

Proposition 6.20. A stopping time 2 T is optimal if and only if H D U P -a.s.,

and if the stopped process U is a martingale. In particular, any optimal stopping

time satisfies min .

Proof. First note that 2 T is optimal if it satisfies the two conditions of the assertion,

because then Theorem 6.18 implies that

sup E H D U0 D E UT D E U D E H :

2T

For the converse implication, we apply the assumption of optimality, the fact that

H U , and the stopping theorem for supermartingales to obtain that

U0 D E H E U U0 ;

so that all inequalities are in fact equalities. It follows in particular that H D U P almost surely. Moreover, the identity E U D U0 implies that the stopped process

U is a supermartingale with constant expectation U0 , and hence is a martingale.

In general, there can be many different optimal stopping times. The largest optimal

stopping time admits an explicit description: It is the first time before T for which the

Snell envelope U loses the martingale property

max WD inft 0 j E U tC1 U t j F t 0 ^ T

D inft 0 j A tC1 0 ^ T:

Here, A denotes the increasing process obtained from the Doob decomposition of U

under P .

Theorem 6.21. The stopping time max is the largest optimal stopping time. Moreover, a stopping time is optimal if and only if P -a.s. max and U D H .

332

that U D M A is the Doob decomposition of U for any stopping time .

Thus, U is a martingale if and only if A D 0, because A is increasing. Therefore,

U is a martingale if and only if max , and so the second part of the assertion

follows from Proposition 6.20. It remains to prove that max itself is optimal, i.e., that

Umax D Hmax . This is clear on the set max D T . On the set max D t for t < T

one has A t D 0 and A tC1 > 0. Hence,

E U tC1 U t j F t D .A tC1 A t / D A tC1 < 0 on max D t .

Thus, U t > E U tC1 j F t and the definition of the Snell envelope yields that U t D

H t _ E U tC1 j F t D H t on max D t .

Let us now return to our complete financial market model, where Ht is the discounted payoff of an American contingent claim. Thus, an optimal stopping strategy

for H maximizes the expected payoff E H . But a stopping time turns out to be the

best choice even in a pathwise sense, provided that it is optimal with respect to the

unique equivalent martingale measure P in a complete market model. In order to

explain this fact, let us first recall from Section 6.1 the construction of a perfect hedge

of H from the sellers perspective. Let

UP D M A

Since P is the unique equivalent martingale measure in our model, the martingale

M has the representation

M t D U0P C

t

X

k .Xk Xk1 /;

t D 0; : : : ; T;

kD1

the self-financing strategy constructed from and the initial investment U0P . Since

M dominates H , this yields a perfect hedge of H from the perspective of the seller: If

the buyer exercises the option at some stopping time , then the seller makes a profit

M H 0. The following corollary states that the buyer can in fact meet the

value of the sellers hedging portfolio, and that this happens if and only if the option

is exercised at an optimal stopping time with respect to P . In this sense, U0P can

be regarded as the unique arbitrage-free price of the discounted American claim H .

Corollary 6.22. With the above notation,

H M D U0P C

X

k .Xk Xk1 /;

kD1

and equality holds P -almost surely if and only if is optimal with respect to P .

333

Proof. At time ,

H UP D M A M :

Moreover, by Theorem 6.21, both H D UP and A D 0 hold P -a.s. if and only

if is optimal with respect to P .

Let us now compare a discounted American claim H to the corresponding discounted European claim HT , i.e., to the contract which is obtained from H by restricting the exercise time to be T . In particular, we are interested in the relation

between American and European put or call options. Let

V t WD E HT j F t

denote the amount needed at time t to hedge HT . Since our market model is complete,

V t can also be regarded as the unique arbitrage-free price of the discounted claim HT

at time t . From the sellers perspective, U tP plays a similar role for the American

option. It is intuitively clear that an American claim should be more expensive than the

corresponding European one. This is made mathematically precise in the following

statement.

Proposition 6.23. With the above notation, U tP

dominates H , then U P and V coincide.

Proof. The first statement follows immediately from the supermartingale property

of U P

U tP E UTP j F t D E HT j F t D V t :

Next, if the P -martingale V dominates H , then it also dominates the corresponding

Snell envelope U P by Proposition 6.10. Thus V and U P must coincide.

Remark 6.24. The situation in which V dominates H occurs, in particular, when the

process H is a P -submartingale. This happens, for instance, if H is obtained by

applying a convex function f W Rd ! 0; 1/ to the discounted price process X.

Indeed, in this case, Jensens inequality for conditional expectations implies that

E f .X tC1 / j F t f .E X tC1 j F t / D f .X t /:

Example 6.25. The discounted payoff of an American call option C tcall D .S t1 K/C

is given by

K C

call

1

:

Ht D Xt 0

St

Under the hypothesis that S t0 is increasing in t , (5.24) states that

call

j F t H tcall

E H tC1

P -a.s. for t D 0; : : : ; T 1.

334

In other words, H call is a submartingale, and the Snell envelope U P of H call coincides with the value process

K C

1

XT 0

Vt D E

Ft

ST

of the corresponding European call option with maturity T . In particular, we have

U0P D V0 , i.e., the unique arbitrage-free price of the American call option is equal to

its European counterpart. Moreover, Theorem 6.21 implies that the maximal optimal

stopping time with respect to P is given by max T . This suggests that, in a

complete model, an American call should not be exercised before maturity.

}

put

different, because the argument in (5.24) fails unless S 0 is decreasing. If S 0 is an

increasing bond, then the time value

1 C

0 .K ST /

W t WD S t E

F t .K S t1 /C

ST0

of a European put .K ST1 /C typically becomes negative at a certain time t , corresponding to an early exercise premium W t ; see Figure 5.3. Thus, the early exercise

premium is the surplus which an owner of the American put option would have over

the value of the European put .K ST1 /C .

The relation between the price of a put option and its intrinsic value can be illustrated in the context of the CRR model. With the notation of Section 5.5, the price

process of the risky asset S t D S t1 can be written as

S t D S0 t

for t WD

t

Y

.1 C Rk /

kD1

and with the constant S0 0. Recall that the returns Rk can take only two possible

values a and b with 1 < a < b, and that the market model is arbitrage-free if and

only if the riskless interest rate r satisfies a < r < b. In this case, the model is

complete, and the unique equivalent martingale measure P is characterized by the

fact that it makes R1 ; : : : ; RT independent with common distribution

P Rk D b D p D

Let

.x/ WD sup E

2T

put

r a

:

ba

.K x /C

.1 C r/

(6.13)

and decreasing function in x. Let us assume that r > 0 and that the parameter a is

335

strictly negative. A trivial situation occurs if the option is far out of the money in

the sense that

K

;

x

.1 C a/T

because then S t D x t K for all t , and the payoff of C put is always zero. In

particular, .x/ D 0. If

K

x

(6.14)

.1 C b/T

then S t D x t K for all t , and hence

K

x D K x:

.x/ D sup E

.1 C r/

2T

In this case, the price of the American put option is equal to its intrinsic value .Kx/C

at time t D 0, and an optimal strategy for the owner would simply consist in exercising

the option immediately, i.e., there is no demand for the option in the regime (6.14).

Now consider the case

K

Kx<

.1 C a/T

of a put option which is at the money or not too far out of the money. For large

put

enough t > 0, the probability P C t > 0 of a non-zero payoff is strictly positive,

while the intrinsic value .K x/C vanishes. It follows that the price .x/ is strictly

higher than the intrinsic value, and so it is not optimal for the buyer to exercise the

option immediately.

Summarizing our observations, we can say that there exists a value x with

K

x < K

.1 C b/T

such that

.x/ D .K x/C

for x x ,

.x/ D 0

and

Remark 6.27. In the context of an arbitrage-free CRR model, we consider a discounted American claim H whose payoff is determined by a function of time and of

the current spot price, i.e.,

Ht D h t .S t /

for all t .

336

S0

Figure 6.1. The price of an American put option as a function of S0 compared to the

options intrinsic value .K S0 /C .

Clearly, this setting includes American call and put options as special cases. By using

the same arguments as in the derivation of (5.28), we get that the Snell envelope U P

of H is of the form

U tP D u t .S t /;

t D 0; : : : ; T;

uT .x/ D hT .x/

and

O .1 p //:

and bO D 1 C b. Thus, the space 0; T 0; 1/ can be decomposed into the two

regions

Rc WD .t; x/ j u t .x/ > h t .x/ and

and the minimal optimal stopping time min can be described as the first exit time of

the space time process .t; S t / from the continuation region Rc or, equivalently, as the

first entrance time into the stopping region Rs

min D mint 0 j .t; S t / Rc D mint 0 j .t; S t / 2 Rs :

Exercise 6.2.3. Consider a market model with two assets and an American contingent

claim. The development of the discounted price process X WD X 1 and the discounted

American claim is described by the following diagram.

337

X2 D 9, H2 D 4

X1 D 8, H1 D 1:5 HH

HH

X0 D 5, H0 D 1

HH

H

H

X2 D 6, H2 D 1

HH

H

H X1 D 4, H1 D 0

HH

HH

H

H X2 D 3, H2 D 0

The buyer of the American claim uses a probability measure P that assigns equal

probability to each of the possible scenarios. Find an optimal stopping strategy that

maximizes E H over 2 T . What would be an optimal stopping time

p if the buyer

p

uses the utility function u.x/ D x and thus aims at maximizing E H ? Then

show that the market model admits a unique risk-neutral measure P and compute

}

the corresponding Snell envelope U P .

Exercise 6.2.4. Let

H tK WD

.K S t /C

.1 C r/t

be the discounted payoff of an American put option with strike K in a market model

with one risky asset S D .S t / tD0;:::T and a riskless asset S t0 D .1 C r/t , where

K the minimal optimal stopping time of the buyers problem

r > 0. We denote by min

K

to maximize E H over 2 T .

0

K K P -a.s. when K K 0 .

(a) Show that min

min

K D 0 P -a.s.

(b) Show that ess infK0 min

(c) Use (b) and the fact that F0 D ;; to conclude that there exists K0 0 such

K D 0 P -a.s. for all K K .

}

that min

0

6.3

Arbitrage-free prices

In this section, we drop the condition of market completeness, and we develop the

notion of an arbitrage-free price for a discounted American claim H in a general

incomplete framework. The basic idea consists in reducing the problem to the determination of the arbitrage-free price for the payoff H which arises from H by fixing

the exercise strategy . The following remark explains that H can be treated like the

discounted payoff of a European contingent claim, whose set of arbitrage-free prices

is given by

(6.15)

.H / D E H j P 2 P ; E H < 1 :

338

at time t < T can be regarded as a discounted European claim HQ E maturing at T .

HQ E is obtained from HQ t by investing at time t the payoff S t0 HQ t into the numraire,

i.e., by buying HQ t shares of the 0th asset, and by considering the discounted terminal

value of this investment:

1

HQ E D 0 . ST0 HQ t / D HQ t :

ST

In the case of our discounted American claim H which is payed off at the random

time , we can either apply this argument to each payoff

HQ t WD H IDt D H t IDt ;

or directly use a stopping time version of this argument. We conclude that H can

be regarded as a discounted European claim, whose arbitrage-free prices are given by

(6.15).

}

Now suppose that H is offered at time t D 0 for a price 0. From the buyers

point of view there should be at least one exercise strategy such that the proposed

price is not too high in the sense that 0 for some 0 2 .H /. From

the sellers point of view the situation looks different: There should be no exercise

strategy 0 such that the proposed price is too low in the sense that < 0 for all

0 2 .H 0 /. By adding the assumption that the buyer only uses stopping times in

exercising the option, we obtain the following formal definition.

Definition 6.29. A real number is called an arbitrage-free price of a discounted

American claim H if the following two conditions are satisfied:

The price is not too high in the sense that there exists some 2 T and

0 2 .H / such that 0 .

The price is not too low in the sense that there exists no 0 2 T such that

< 0 for all 0 2 .H 0 /.

inf .H / WD inf .H /

and

sup .H / WD sup .H /:

Recall from Remark 6.6 that every discounted European claim H E can be regarded

as a discounted American claim H A whose payoff is zero if H A is exercised before T ,

and whose payoff at T equals H E . Clearly, the two sets .H E / and .H A / coincide, and so the two Definitions 5.28 and 6.29 are consistent with each other.

Remark 6.30. It follows from the definition that any arbitrage-free price for H

must be an arbitrage-free price for some H . Hence, (6.15) implies that D E H

339

for some P 2 P . Similarly, we obtain from the second condition in Definition 6.29

that infP 2P E H for all 2 T . It follows that

sup inf E H sup sup E H for all 2 .H /.

2T P 2P

(6.16)

2T P 2P

In particular,

sup E H

2T

is the unique arbitrage-free price of H if P is the unique equivalent martingale measure in a complete market model, and so Definition 6.29 is consistent with the results

of the Section 6.1 and 6.2.

}

Exercise 6.3.1. Show that in every arbitrage-free market model and for any discounted American claim H ,

inf sup E H < 1;

(6.17)

P 2P 2T

Our main goal in this section is to characterize the set .H /, and to identify the

upper and lower bounds in (6.16) with the quantities sup .H / and inf .H /. We will

work under the simplifying assumption that

H t 2 L1 .P /

(6.18)

U tP D ess sup E H j F t :

2T t

2T P 2P

P 2P 2T

P 2P

In fact, a similar relation also holds for the lower bound in (6.16)

2T

P 2P

P 2P

2T

(6.19)

P 2P

The proof that the above interchange of infimum and supremum is indeed justified

under assumption (6.18) is postponed to the next section; see Theorem 6.45.

Theorem 6.31. Under condition (6.18), the set of arbitrage-free prices for H is a

real interval with endpoints

inf .H / D inf sup E H D sup inf E H

P 2P 2T

2T P 2P

340

and

sup .H / D sup sup E H D sup sup E H :

P 2P 2T

2T P 2P

Moreover, .H / either consists of one single point or does not contain its upper

endpoint sup .H /.

Proof. Let be a stopping time which is optimal with respect to a given P 2 P .

Then U0P D E H D sup 0 2T E H 0 , and consequently U0P 2 .H /.

Together with the a priori bounds (6.16), we obtain the inclusions

U0P j P 2 P .H / a; b;

(6.20)

where

a WD sup inf E H and

2T

b WD sup sup E H :

P 2P

2T P 2P

a D inf sup E H D inf U0P

and

P 2P

P 2P 2T

b D sup U0P :

P 2P

Together with (6.20), this yields the identification of inf .H / and sup .H / as a and b.

Now we claim that U0P j P 2 P is an interval, which, in view of the preceding

step, will prove that .H / is also an interval. Take P0 ; P1 2 P and define P 2 P

by P WD P1 C .1 /P0 for 0 1. By Theorem 6.18, f ./ WD U0P is the

supremum of the affine functions

7! E H D E1 H C .1 /E0 H ;

2T:

Thus, f is convex and lower semicontinuous on 0; 1, hence continuous; see part (a)

of Proposition A.4. Since P is convex, this proves our claim.

It remains to exclude the possibility that b belongs to .H / in case a < b. Suppose

by way of contradiction that b 2 .H /. Then there exist O 2 T and PO 2 P such that

O HO D b D sup sup E H :

E

2T P 2P

In particular, PO attains the supremum of E HO for P 2 P . Theorem 5.32 implies that the discounted European claim HO is attainable and that E HO is in fact

independent of P 2 P . Hence,

O HO D inf E HO sup inf E H ;

b D E

P 2P

2T P 2P

341

Comparing the previous result with Theorem 5.32, one might wonder whether

.H / contains its lower bound if inf .H / < sup .H /. At a first glance, it may

come as a surprise that both cases

inf .H / 2 .H /

and

inf .H / .H /

Example 6.32. Consider a complete market model with T D 2, defined on some

probability space .0 ; G0 ; P0 /. This model will be enlarged by adding two external

states ! C and ! , i.e., we define WD 0 ! C ; ! and

1

P0 !0 ; !0 2 0 :

2

We assume that this additional information is revealed at time 2. The enlarged financial market model will then be incomplete, and the corresponding set P of equivalent

martingale measures satisfies

P .!0 ; ! / WD

where Pp is determined by Pp 0 ! C D p. Consider the discounted American

claim H defined as

2 if ! D .!0 ; ! C /,

H0 0; H1 1; and H2 .!/ WD

0 if ! D .!0 ; ! /.

Clearly, .H / 1; 2/. On the other hand, 2 2 is an optimal stopping time for

Pp if p > 12 , while 1 1 is optimal for p 12 . Hence,

.H / D 1; 2/;

and the lower bound inf .H / D 1 is an arbitrage-free price for H . Now consider the

discounted American claim HQ defined by HQ t D H t for t D 0; 2 and by HQ 1 0. In

this case, we have

.HQ / D .0; 2/:

}

Theorem 6.31 suggests that an American claim H which admits a unique arbitragefree price should be attainable in an appropriate sense. Corollary 6.22, our hedging

result in the case of a complete market, suggests the following definition of attainability.

Definition 6.33. A discounted American claim H is called attainable if there exists

a stopping time 2 T and a self-financing trading strategy whose value process V

satisfies P -a.s.

V t H t for all t , and V D H .

The trading strategy is called a hedging strategy for H .

342

If H is attainable, then a hedging strategy protects the seller not only against those

claims H which arise from stopping times . The seller is on the safe side even if

the buyer would have full knowledge of future prices and would exercise H at an

arbitrary FT -measurable random time . For instance, the buyer even could choose

such that

H D max H t :

0tT

In fact, we will see in Remark 7.12 that H is attainable in the sense of Definition 6.33

if and only if V t H t for all t and V D H for some FT -measurable random

time .

If the market model is complete, then every American claim H is attainable. Moreover, Theorem 6.11 and Corollary 6.22 imply that the minimal initial investment

needed for the purchase of a hedging strategy for H is equal to the unique arbitragefree price of H . In a general market model, every attainable discounted American

claim H satisfies our integrability condition (6.18) and has a unique arbitrage-free

price which is equal to the initial investment of a hedging strategy for H . This follows from Theorem 5.25. In fact, the converse implication is also true.

Theorem 6.34. For a discounted American claim H satisfying (6.18), the following

conditions are equivalent:

(a) H is attainable.

(b) H admits a unique arbitrage-free price .H /, i.e., .H / D .H /.

(c) sup .H / 2 .H /.

Moreover, if H is attainable, then .H / is equal to the initial investment of any

hedging strategy for H .

The equivalence of (b) and (c) is an immediate consequence of Theorem 6.31. The

remainder of the proof of Theorem 6.34 is postponed to Remark 7.10 because it requires the technique of superhedging, which will be introduced in Section 7.

6.4

In this section we define the pasting of two equivalent probability measures at a given

stopping time. This operation will play an important role in the analysis of lower

and upper Snell envelopes as developed in Section 6.5. In particular, we will prepare

for the proof of the minimax identity (6.19), which was used in the characterization

of arbitrage-free prices of an American contingent claim. Let us start with a few

preparations.

Definition 6.35. Let be a stopping time. The -algebra of events which are observable up to time is defined as

F WD A 2 F j A \ t 2 F t for all t :

343

F D A 2 F j A \ D t 2 F t for all t

and conclude that F coincides with F t if t . Finally show that F F when

is a stopping time with .!/ .!/ for all ! 2 .

}

The following result is an addendum to Doobs stopping theorem; see Theorem

6.15:

Proposition 6.36. For an adapted process M in L1 .Q/ the following conditions are

equivalent:

(a) M is a Q-martingale.

(b) EQ M j F D M^ for all 2 T and all stopping times .

Proof. (a) ) (b): Take a set A 2 F and let us write

EQ M I A D EQ M I A \ C EQ M I A \ > :

Condition (b) will follow if we may replace M by M in the rightmost expectation.

To this end, note that

A \ D t \ > D A \ D t \ > t 2 F t :

Thus, since the stopped process M is a martingale by Theorem 6.15,

EQ M I A \ > D

T

X

EQ MT I A \ D t \ >

tD0

T

X

EQ M t I A \ D t \ >

tD0

D EQ M I A \ > :

(b) ) (a): This follows by taking t and s t .

Exercise 6.4.2. Let Z be the density process of a probability measure QQ that is absolutely continuous with respect to Q; see Exercise 5.2.3. Show that for a stopping

time , we have QQ

Q on F with density given by

d QQ

D EQ ZT j F D ZT ^ :

dQ F

344

Exercise 6.4.3. Show that for a stopping time , a random variable Y 2 L1 .; F ; Q/,

and t 2 0; : : : ; T ,

EQ H j F D EQ H j F t Q-a.s. on D t .

(6.21)

}

We next state the following extension of Theorem 6.18. It provides the solution to

the optimal stopping problem posed at any stopping time T .

Proposition 6.37. Let H be an adapted process in L1 .; F ; Q/, and define for

2T

T WD 2 T j :

Then the Snell envelope U Q of H satisfies Q-a.s.

UQ D ess sup EQ H j F ;

2T

./

Q

min WD mint j H t D U t :

Definition 6.38. Let Q1 and Q2 be two equivalent probability measures and take

2 T . The probability measure

Q A WD EQ1 Q2 A j F ;

Q

A 2 FT ;

The monotone convergence theorem for conditional expectations guarantees that QQ

is indeed a probability measure and that

EQQ Y D EQ1 EQ2 Y j F

for all FT -measurable Y 0. Note that QQ coincides with Q1 on F , i.e.,

EQQ Y D EQ1 Y

ZT

d QQ

D

;

dQ1

Z

where Z is the density process of Q2 with respect to Q1 .

345

Proof. For Y 0,

EQQ Y D EQ1 EQ2 Y j F

i

h 1

EQ1 Y ZT j F

D EQ1

Z

i

hZ

T

D EQ1

Y ;

Z

where we have used the martingale property of Z and the fact that Z > 0 Q1 -almost

surely. The equivalence of QQ and Q1 follows from ZT > 0 Q1 -almost surely.

Lemma 6.40. For Q1 Q2 , let QQ be their pasting in 2 T . Then, for all stopping

times and FT -measurable Y 0,

EQQ Y j F D EQ1 EQ2 Y j F _ j F :

Proof. If ' 0 is F -measurable, then 'I
is F \ F -measurable. Hence,

EQQ Y'I D EQ1 EQ2 Y j F 'I

D EQ1 EQ1 EQ2 Y j F j F 'I

D EQQ EQ1 EQ2 Y j F j F 'I ;

where we have used the fact that QQ coincides with Q1 on F . On the other hand,

EQQ Y'I > D EQ1 EQ2 EQ2 Y j F ' j F I >

D EQQ EQ2 Y j F 'I > :

It follows that

EQQ Y j F D EQ1 EQ2 Y j F j F I
C EQ2 Y j F I>
;

and this coincides with the right-hand side of the asserted identity.

Definition 6.41. A set Q of equivalent probability measures on .; F / is called stable if, for any Q1 ; Q2 2 Q and 2 T , also their pasting in is contained in Q.

The condition of stability in the preceding definition is sometimes also called fork

convexity, m-stability, or stability under pasting. For the purposes of this book, the

most important example of a stable set is the class P of all equivalent martingale

measures, but in Section 6.5 we will also discuss the connection between stable sets

and dynamic risk measures.

Proposition 6.42. P is stable.

346

stopping theorem in the form of Proposition 6.36 and Lemma 6.40 applied with Y WD

X ti 0 and s yield that for s t

Q X t j Fs D E1 E2 X t j F _s j Fs D E1 X _s j Fs D Xs :

E

Q X ti D X i < 1,

It follows in particular that each component X ti is in L1 .PQ / since E

0

concluding the proof of PQ 2 P .

We conclude this section by an alternative characterization of stable sets. It will be

used in Section 11.2. Suppose that 2 T takes at most one value t 2 0; : : : ; T that

is different from T . Then there exists a set B 2 F t such that D t IB C T IB c . It

follows that the pasting QQ of two equivalent probability measures Q1 and Q2 in is

given by

Q A WD EQ1 Q2 A j F t IB C I

; A 2 FT :

(6.22)

Q

A\B c

This observation can be used to give the following characterization of stable sets.

Proposition 6.43. A set Q of equivalent probability measures is stable if and only

if for any t 2 0; : : : ; T and B 2 F t the probability measure QQ defined in (6.22)

belongs again to Q.

Proof. We have already seen that QQ 2 Q when Q is stable. For the proof of the

converse implication, let 2 T be a stopping time and take Q1 ; Q2 2 Q. We define

recursively QQ T WD Q1 and

QQ t1 A WD EQQ t Q2 A j F t1 I Dt 1 C IA\ t 1

for t D T; : : : ; 1. Then QQ 0 2 Q by assumption. We claim that QQ 0 coincides with

the pasting of Q1 and Q2 in , and this will prove the assertion. To verify our claim,

note that the densities of QQ t with respect to Q1 satisfy the recursion

d QQ t1

d QQ t ZT

D

I Dt 1 C I t 1 ;

dQ1

dQ1 Z t1

where .Z t / is the density process of Q2 with respect to Q1 . But this implies that

d QQ t1

ZT

D

I

C I <t 1 ;

dQ1

Z t 1

and so our claim QQ 0 D QQ follows from Lemma 6.39.

347

6.5

Our main goal in this section is to provide a proof of the minimax identity (6.19),

that was used in the characterization of the set of arbitrage-free prices of an American contingent claim. The techniques and results which we develop here will help

to characterize the time-consistency of dynamic coherent risk measures and they will

also be needed in Chapter 7. Moreover, they can be interpreted in terms of an optimal stopping problem for general utility functionals which appear in a robust Savage

representation of preferences on payoff profiles. Let us now fix a set Q of equivalent

probability measures and an adapted process H such that

H t 2 L1 .Q/

Q

inf sup EQ H D inf U0 < 1;

Q2Q 2T

Q2Q

assume that

Q is stable.

Definition 6.44. The lower Snell envelope of H is defined as

#

Q

U t WD ess inf U t D ess inf ess sup EQ H j F t ;

Q2Q

Q2Q

t D 0; : : : ; T:

2T t

"

Q2Q

2T t

t D 0; : : : ; T:

Q2Q

We will first study the lower Snell envelope. The following minimax theorem

states that the essential infimum and the essential supremum occurring in the definition of U # may be interchanged if Q is stable. Applied at t D 0 and combined with

Proposition 6.42, this gives the identity (6.19), which was used in our characterization

of the arbitrage-free prices of H .

Theorem 6.45. The lower Snell envelope of H satisfies

U t# D ess sup ess inf EQ H j F t for each t .

2T t

Q2Q

In particular,

#

Q2Q 2T

2T Q2Q

(6.23)

348

of the next theorem, which solves the following optimal stopping problem that is

formulated with respect to the nonadditive expectation operator infQ2Q EQ :

maximize inf EQ H among all 2 T .

Q2Q

t WD minu t j Uu# D Hu :

Then, P -a.s.,

U t D ess inf EQ H t j F t :

(6.24)

Q2Q

In particular,

#

2T Q2Q

Q2Q

Lemma 6.47. Suppose that we are given Q1 ; Q2 2 Q, a stopping time 2 T , and

a set B 2 F . Let QQ 2 Q be the pasting of Q1 and Q2 in the stopping time

WD IB C T IB c :

Then the Snell envelopes associated with these three measures are related as follows:

Q

P -a.s.

Q

2T

EQ2 H j F _ D EQ2 H j F IB C H IB c :

Hence, Lemma 6.40 yields that

EQQ H j F D EQ2 H j F IB C EQ1 H j F IB c :

Moreover, whenever 1 ; 2 2 T , then

Q WD 1 IB C 2 IB c

is also a stopping time in T . Thus,

Q

2T

2T

(6.25)

349

F and

O

UQk & ess inf UQ D U# :

O

Q2Q

O

O

Q2Q

Q

Q

Proof. For Q1 ; Q2 2 Q, B WD U 1 > U 2 , take QQ 2 Q as in Lemma 6.47. Then

Q

(6.26)

O

WD UQ j QO 2 Q and QO D Q on F

is such that U# D ess inf . Moreover, (6.26) implies that is directed downwards,

and the second part of Theorem A.33 states the existence of the desired sequence

.Qk / Q. The proof for the essential supremum is analogous.

Q

each Q 2 Q, so that holds in (6.24). For the proof of the converse inequality, note

that

Q

for Q 2 Q.

t minu t j UuQ D Hu DW t

It was shown in Theorem 6.18 that tQ is the minimal optimal stopping time after

time t and with respect to Q. It was also shown in the proof of Theorem 6.18 that the

Q

stopped process .U Q / t is a Q-martingale from time t on. In particular,

U tQ D EQ UQt j F t

for all Q 2 Q.

(6.27)

such that UQt k decreases to U#t . We obtain

EQ H t j F t D EQ U#t j F t D EQ lim UQt k j F t

k"1

k"1

D lim

k"1

k"1

U tQk

U t# :

Q

Q

Q

Q

Here we have used that H t U t k U t 1 and EQ jUt 1 j D EQ1 jU t 1 j < 1

together with dominated convergence in the third step, the fact that Qk D Q on

F t F t in the fourth, and (6.27) in the fifth identity.

350

Remark 6.49. Suppose the buyer of an American option uses a utility functional of

the form

inf EQ u.Z/ ;

Q2Q

viewed as a robust Savage representation of a preference relation on discounted asset

payoffs; see Section 2.5. Thus, the aim of the buyer is to maximize the utility

inf EQ u.H /

Q2Q

utility maximization problem can be solved with the results developed in this section,

provided that the set Q is a stable set of equivalent probability measures. Indeed,

assume

HQ t WD u.H t / 2 L1 .Q/ for all t and each Q 2 Q,

and let U Q be the Snell envelope of HQ t with respect to Q 2 Q. Theorem 6.46 states

that the generalized optimal stopping problem is solved by the stopping time

Q2Q

i.e.,

#

inf sup EQ u.H / D U0 D inf EQ u.H / :

Q2Q 2T

Q2Q

"

Q

U t WD ess sup U t D ess sup ess sup EQ H j F t ;

Q2Q

2T t

t D 0; : : : ; T:

Q2Q

sup EQ jH t j < 1

for all t .

Q2Q

"

Q

U0 D sup U0 sup sup EQ jH j < 1:

Q2Q

2T Q2Q

Our main result on upper Snell envelopes states that, for stable sets Q, the upper

Snell envelope U " satisfies a recursive scheme that is similar to the one for ordinary

Snell envelopes. In contrast to (6.4), however, it involves the nonadditive conditional

expectation operators ess supQ EQ j F t .

351

"

"

"

t D T 1 : : : ; 0: (6.28)

Q2Q

"

Q

Q

U t D ess sup U t D H t _ ess sup EQ U tC1 j F t :

Q2Q

(6.29)

Q2Q

Next, we fix Q 2 Q and denote by Q tC1 .Q/ the set of all QO 2 Q which coincide

with Q on F tC1 . According to Lemma 6.48, there are Qk 2 Q tC1 .Q/ such that

Qk

"

Q1

Q1

j D EQ1 jU tC1

j < 1 combined with

U tC1 % U tC1 . The fact that EQ jU tC1

monotone convergence for conditional expectations shows that

"

Q

ess sup EQ U tC1

j F t ess sup EQ U tC1

j Ft

Q2Q

Q2Q

D ess sup

ess sup

O

Q2Q Q2Q

t C1 .Q/

O

Q

EQ U tC1 j F t

k

Q

ess sup lim inf EQ U tC1 j F t

k"1

Q2Q

(6.30)

"

Q2Q

In particular, all inequalities are in fact identities. Together with (6.29) we obtain the

recursive scheme for U " .

The following result shows that the nonadditive conditional expectation operators

ess supQ EQ j F t associated with a stable set Q enjoy a consistency property that

is similar to the martingale property for ordinary conditional expectations.

Theorem 6.51. Let Q be a set of equivalent probability measures and

"

V t WD ess sup EQ H j F t ;

t D 0; : : : ; T;

Q2Q

"

V" D ess sup EQ V" j F

Q2Q

for ; 2 T with .

352

V" D

T

X

tD0 Q2Q

T

X

tD0 Q2Q

D ess sup EQ H j F ;

Q2Q

Proof of Theorem 6:51. By Remark 6.52,

V" D ess sup EQ H j F D ess sup EQ EQ H j F j F :

Q2Q

Q2Q

"

The proof that the right-hand side is equal to ess supQ2Q EQ V j F is done by

first noting that V " is equal to the upper Snell envelope of the process H t given by

HT D H and H t D 0 for t < T . Then the same argument as in (6.30) applies. All

one has to do is to replace t C 1 by .

Remark 6.53. Let us conclude this section by pointing out the connection between

stability under pasting and the time-consistency of dynamic coherent risk measures.

Let

.Y / WD sup EQ Y ; Y 2 L1 .P /;

Q2Q

measures equivalent to P . In the context of a dynamic financial market model, it is

natural to update the initial risk assessment at later times t > 0. If one continues to

use Q as a basis to compute the risk but takes into account the available information,

one is led to consider the conditional risk measures

t .Y / D ess sup EQ Y j F t ;

t D 0; : : : ; T:

(6.31)

Q2Q

a dynamic risk measure is called time-consistent or dynamically consistent if

s . t .Y // D s .Y /

for 0 s t T .

(6.32)

When the set Q in (6.31) is a stable set of equivalent probability measures, then Theorem 6.51 implies immediately the time consistency (6.32). The following converse of

this statement, and hence a converse of Theorem 6.51, will be given in Theorem 11.22:

353

if . t / is a dynamically consistent sequence of conditional coherent risk measures satisfying certain regularity assumptions, then there exists a stable set Q of equivalent

probability measures such that (6.32) holds. An extension of dynamic consistency to

dynamic convex risk measures will be given in Section 11.2.

Note that Theorem 6.51 shows that in (6.32) the deterministic times s and t can

even be replaced by stopping times when Q is stable.

}

Chapter 7

Superhedging

The idea of superhedging is to find a self-financing trading strategy with minimal initial investment which covers any possible future obligation resulting from the sale of

a contingent claim. If the contingent claim is not attainable, the proof of the existence

of such a superhedging strategy requires new techniques, and in particular a new

uniform version of the Doob decomposition. We will develop this theory for general

American contingent claims. In doing so, we will also obtain new results for European contingent claims. In the first three sections of this chapter, we assume that our

market model is arbitrage-free or, equivalently, that the set of equivalent martingale

measures satisfies

P ;:

In the final Section 7.4, we discuss liquid options in a setting where no probabilistic

model is fixed a priori. Such options may be used for the construction of specific

martingale measures, and also for the purpose of hedging illiquid exotic derivatives.

7.1

P -supermartingales

sup E H t < 1 for all t :

(7.1)

P 2P

Our aim in this chapter is to find the minimal amount of capital Ut that will be needed

at time t in order to purchase a self-financing trading strategy whose value process

satisfies Vu Hu for all u t . In analogy to our derivation of the recursive scheme

(6.4), we will now heuristically derive a formula for U t . At time T , the minimal

amount needed is clearly given by

UT D HT :

At time T 1, a first requirement is to have UT 1 HT 1 . Moreover, the amount

UT 1 must suffice to purchase an FT 1 -measurable portfolio T such that T XT

HT almost surely. An informal application of Theorem 1.32, conditional on FT 1 ,

shows that

UT 1 ess sup E HT j FT 1 :

P 2P

355

Hence, the minimal amount UT 1 is equal to the maximum of HT 1 and this essential

supremum. An iteration of this argument yields the recursive scheme

UT D HT

and

P 2P

identify U as the upper Snell envelope

"

P 2P

2T t

P 2P

of H with respect to the stable set P , where U P denotes the Snell envelope of H

with respect to P . In the first three sections of this chapter, we will in particular give

a rigorous version of the heuristic argument above.

Note first that condition (7.1) implies that

P 2P

P 2P 2T

where we have used the identification of the upper bound sup .H / of the arbitragefree prices of H given in Theorem 6.31. It will turn out that the following definition

applies to the upper Snell envelope if we choose Q D P .

Definition 7.1. Suppose that Q is a non-empty set of probability measures on

.; FT /. An adapted process is called a Q-supermartingale if it is a supermartingale

with respect to each Q 2 Q. Analogously, we define the notions of a Q-submartingale

and of a Q-martingale.

In Theorem 5.25, we have already encountered an example of a P -martingale,

namely the value process of the replicating strategy of an attainable discounted European claim.

Theorem 7.2. The upper Snell envelope U " of H is the smallest P -supermartingale

that dominates H .

Proof. For each P 2 P the recursive scheme (6.28) implies that P -a.s.

"

"

"

U t H t _ E U tC1 j F t E U tC1 j F t :

"

Since U0 is a finite constant due to our integrability assumption (7.1), induction on

"

t shows that U t is integrable with respect to each P 2 P and hence is a P -supermartingale dominating H .

356

Chapter 7 Superhedging

"

If UQ is another P -supermartingale which dominates H , then UQ T HT D UT .

"

for some t , then

Moreover, if UQ tC1 U

tC1

"

UQ t H t _ E UQ tC1 j F t H t _ E U tC1 j F t :

Thus,

"

"

UQ t H t _ ess sup E U tC1 j F t D U t ;

P 2P

For European claims, Theorem 7.2 takes the following form.

Corollary 7.3. Let H E be a discounted European claim such that

sup E H E < 1:

P 2P

Then

"

V t WD ess sup E H E j F t ;

t D 0; : : : ; T;

P 2P

Remark 7.4. Note that the proof of Theorem 7.2 did not use any special properties

of the set P . Thus, if Q is an arbitrary set of equivalent probability measures, the

process U defined by the recursion

UT D HT

and

Q2Q

7.2

The aim of this section is to give a complete characterization of all non-negative P supermartingales. It will turn out that an integrable and non-negative process U is a

P -supermartingale if and only if it can be written as the difference of a P -martingale

N and an increasing adapted process B satisfying B0 D 0. This decomposition

may be viewed as a uniform version of the Doob decomposition since it involves

simultaneously the whole class P . It will turn out that the P -martingale N has a

special structure: It can be written as a stochastic integral of the underlying process X, which defines the class P . On the other hand, the increasing process B is

only adapted, not predictable as in the Doob decomposition with respect to a single

measure.

357

Theorem 7.5. For an adapted, non-negative process U , the following two statements

are equivalent:

(a) U is a P -supermartingale.

(b) There exists an adapted increasing process B with B0 D 0 and a d -dimensional

predictable process such that

U t D U0 C

t

X

k .Xk Xk1 / B t

kD1

Proof. First, we prove the easier implication (b) ) (a). Fix P 2 P and note that

VT WD U0 C

T

X

k .Xk Xk1 / UT 0:

kD1

Moreover, for P 2 P

E U tC1 j F t D E V tC1 B tC1 j F t V t B t D U t ;

and so U is a P -supermartingale.

The proof of the implication (a) ) (b) is similar to the proof of Theorem 5.32. We

must show that for any given t 2 1; : : : ; T , there exist t 2 L0 .; F t1 ; P I Rd /

and R t 2 L0C .; F t ; P / such that

U t U t1 D t .X t X t1 / R t :

This condition can be written as

U t U t1 2 K t L0C .; F t ; P /;

where K t is as in (5.12). There is no loss of generality in assuming that P is itself

a martingale measure. In this case, U t U t1 is contained in L1 .; F t ; P / by the

definition of a P -supermartingale. Assume that

U t U t1 C WD .K t L0C .; F t ; P // \ L1 .P /:

Absence of arbitrage and Lemma 1.68 imply that C is closed in L1 .; F t ; P /. Hence,

Theorem A.57 implies the existence of some Z 2 L1 .; F t ; P / such that

WD sup E Z W < E Z .U t U t1 / DW < 1:

(7.2)

W 2C

1.58 implies that such a random variable Z must be non-negative and must satisfy

E .X t X t1 / Z j F t1 D 0:

(7.3)

358

Chapter 7 Superhedging

In fact, we can always modify Z such that it is bounded from below by some " > 0

and still satisfies (7.2). To see this, note first that every W 2 C is dominated by a term

of the form t .X t X t1 /. Hence, our assumption P 2 P , the integrability of W ,

and an application of Fatous lemma yield that

E W E t .X t X t1 / lim inf E Ij t jc t .X t X t1 / 0:

c"1

Thus, if we let Z " WD " C Z, then Z " also satisfies E Z " W 0 for all W 2 C. If

we chose " small enough, then E Z " .U t U t1 / is still larger than 0; i.e., Z " also

satisfies (7.2) and in turn (7.3). Therefore, we may assume from now on that our Z

with (7.2) is bounded from below by some constant " > 0.

Let

Z t1 WD E Z j F t1 ;

and define a new measure PQ P by

Z

d PQ

:

WD

dP

Z t1

We claim that PQ 2 P . To prove this, note first that Xk 2 L1 .PQ / for all k, because

the density d PQ =dP is bounded. Next, let

Z

F

'k WD E

k ; k D 0; : : : ; T:

Z t1

If k t , then 'k1 D 'k ; this is clear for k > t , and for k < t it follows from

E Z j F t1

'k D E

Fk D 1:

Z t1

Thus, for k t

Q Xk Xk1 j Fk1 D

E

1

'k1

D E Xk Xk1 j Fk1

D 0:

If k D t , then (7.3) yields that

Q Xk Xk1 j Fk1 D

E

Hence PQ 2 P .

1

Z t1

E .X t X t1 / Z j F t1 D 0:

359

Since PQ 2 P , we have E

Q E

Q U t U t1 j F t1 Z t1

0 E

Q .U t U t1 / Z t1

D E

D E .U t U t1 / Z

D :

This, however, contradicts the fact that > 0.

Remark 7.6. The decomposition in part (b) of Theorem 7.5 is sometimes called the

optional decomposition of the P -supermartingale U . The existence of such a decomposition was first proved by El Karoui and Quenez [105] and D. Kramkov [182]

in a continuous-time framework where B is an optional process; this explains the

terminology.

}

7.3

sup E H t < 1 for all t ,

P 2P

which is equivalent to the condition that the upper bound of the arbitrage-free prices

of H is finite

sup .H / D sup sup E H < 1:

P 2P 2T

Our aim in this section is to construct self-financing trading strategies such that the

seller of H stays on the safe side in the sense that the corresponding portfolio value

is always above H .

Definition 7.7. Any self-financing trading strategy whose value process V satisfies

V t Ht

Sometimes, a superhedging strategy is also called a superreplication strategy. According to Definition 6.33, H is attainable if and only if there exist 2 T and a

superhedging strategy whose value process satisfies V D H P -almost surely.

Lemma 7.8. If H is not attainable, then the value process V of any superhedging

strategy satisfies

P V t > H t for all t > 0:

360

Chapter 7 Superhedging

WD inft 0 j H t D V t :

Then P D 1 D P V t > H t for all t . Suppose that P D 1 D 0. In

this case, V D H P -a.s so that we arrive at the contradiction that H must be an

attainable American claim.

Let us now turn to the question whether superhedging strategies exist. In Section 6.1, we have already seen how one can use the Doob decomposition of the Snell

envelope U P of H together with the martingale representation of Theorem 5.38 in

order to obtain a superhedging strategy for the price U0P , where P denotes the

unique equivalent martingale measure in a complete market model. We have also

seen that U0P is the minimal amount for which such a superhedging strategy is avail

able, and that U0P is the unique arbitrage-free price of H . The same is true of any

attainable American claim in an incomplete market model.

In the context of a non-attainable American claim H in an incomplete financial

market model, the P -Snell envelope will be replaced with the upper Snell envelope

U " of H . The uniform Doob decomposition will take over the roles played by the

usual Doob decomposition and the martingale representation theorem. Since U " is a

P -supermartingale by Theorem 7.2, the uniform Doob decomposition states that U "

takes the form

"

Ut

"

U0

t

X

s .Xs Xs1 / B t

(7.4)

sD1

Ht

for some predictable process and some increasing process B. Thus, the self-financing trading strategy D . 0 ; / defined by and the initial capital

"

1 X D D U0 D sup .H /

is a superhedging strategy for H . Moreover, if VQ is the value process of any superhedging strategy, then Lemma 7.8 implies that VQ0 > E H for all 2 T and each

P 2 P . In particular, VQ0 is larger than any arbitrage-free price for H , and it follows

that VQ0 sup .H /. Thus, we have proved:

Corollary 7.9. There exists a superhedging strategy with initial investment sup .H /,

and this is the minimal amount needed to implement a superhedging strategy.

We will call sup .H / the cost of superhedging of H . Sometimes, a superhedging

strategy is also called a superreplication strategy, and one says that sup .H / is the cost

of superreplication or the upper hedging price of H . Recall, however, that sup .H /

361

is typically not an arbitrage-free price for H . In particular, the seller cannot expect to

receive the amount sup .H / for selling H .

On the other hand, the process B in the decomposition (7.4) can be interpreted

as a refunding scheme: Using the superhedging strategy , the seller may withdraw

successively the amounts defined by the increments of B. With this capital flow, the

hedging portfolio at time t has the value U t" H t . Thus, the seller is on the safe side

at no matter when the buyer decides to exercise the option. As we are going to show

in Theorem 7.13 below, this procedure is optimal in the sense that, if started at any

time t , it requires a minimal amount of capital.

Remark 7.10. Suppose sup .H / belongs to the set .H / of arbitrage-free prices

for H . By Theorem 6.31, this holds if and only if sup .H / is the only element of

.H /. In this case, the definition of .H / yields a stopping time 2 T and some

P 2 P such that

sup .H / D E H :

Now let V be the value process of a superhedging strategy bought at V0 D sup .H /.

It follows that E V D sup .H /. Hence, V D H P -a.s., so that H is attainable in

the sense of Definition 6.33. This observation completes the proof of Theorem 6.34.

}

Remark 7.11. If the American claim H is not attainable, then sup .H / is not an

arbitrage-free price of H . Thus, one may expect the existence of arbitrage opportunities if H would be traded at the price sup .H /. Indeed, selling H for sup .H / and

buying a superhedging strategy creates such an arbitrage opportunity: The balance

at t D 0 is zero, but Lemma 7.8 implies that the value process V of cannot be

reached by any exercise strategy , i.e., we always have

V H

and

P V > H > 0:

(7.5)

Note that (7.5) is not limited to exercise strategies which are stopping times but holds

for arbitrary FT -measurable random times W ! 0; : : : ; T . In other words,

sup .H / is too expensive even if the buyer of H would have full information about

the future price evolution.

}

Remark 7.12. The argument of Remark 7.11 implies that an American claim H

is attainable if and only if there exists an FT -measurable random time W !

0; : : : ; T such that H D V , where V the value process of a superhedging strategy. In other words, the notion of attainability of American claims does not need the

restriction to stopping times.

}

We already know that sup .H / is the smallest amount for which one can buy a superhedging strategy at time 0. The following superhedging duality theorem extends

362

Chapter 7 Superhedging

"

this result to times t > 0. To this end, denote by U t .H / the set of all F t -measurable

random variables UQ t 0 for which there exists a d -dimensional predictable process

Q such that

UQ t C

u

X

(7.6)

kDtC1

Theorem 7.13. The upper Snell envelope U t" of H is the minimal element of U"t .H /.

More precisely

"

"

(a) U t 2 U t .H /,

"

"

(b) U t D ess inf U t .H /.

Proof. Assertion (a) follows immediately from the uniform Doob decomposition of

"

"

the P -supermartingale U " . As to part (b), we clearly get U t ess inf U t .H /

"

from (a). For the proof of the converse inequality, take UQ t 2 U t .H / and choose a

"

predictable process Q for which (7.6) holds. We must show that the set B WD U t

UQ t satisfies P B D 1. Let

"

"

UO t WD U t ^ UQ t D U t IB C UQ t IB c :

"

"

Then UO t U t , and our claim will follow if we can show that Ut UO t . Let

denote the predictable process obtained from the uniform Doob decomposition of the

P -supermartingale U " , and define

if s t ,

Os WD s

s IB C Qs IB c if s > t .

"

With this choice, UO t satisfies (7.6), i.e., UO t 2 U t .H /. Let

"

VOs WD U0 C

s

X

kD1

Us"

Then VOs

for all s t . In particular VOt UO t , and hence VOT HT , which

O

implies that V is a P -martingale; see Theorem 5.25. Hence, Doobs stopping theorem

implies

"

U t D ess sup ess sup E H j F t

P 2P

2T t

h

i

X

ess sup ess sup E UO t C

P 2P

2T

D UO t ;

which concludes the proof.

kDtC1

363

We now take the point of view of the buyer of the American claim H . The buyer

allocates an initial investment to purchase H , and then receives the amount H 0.

The objective is to find an exercise strategy and a self-financing trading strategy

with initial investment , such that the portfolio value is covered by the payoff of

the claim. In other words, find 2 T and a self-financing trading strategy with value

process V such that V0 D and V C H 0. As shown below, the maximal for

which this is possible is equal to

#

inf .H / D sup inf E H D inf sup E H D U0 ;

2T P 2P

P 2P 2T

where

#

U t D ess inf U tP

P 2P

P 2P

2T t

2T t

P 2P

is the lower Snell envelope of H with respect to the stable set P . More generally,

we will consider the buyers problem for arbitrary t 0. To this end, denote by

#

U t .H / the set of all F t -measurable random variables UQ t 0 for which there exists

a d -dimensional predictable process Q and a stopping time 2 Tt such that

UQ t

X

Q k .Xk Xk1 / H

P -a.s.

kDtC1

#

#

Theorem 7.14. U t is the maximal element of U t .H /. More precisely

#

(a) U t 2 U t .H /,

#

Proof. (a): Let be a superhedging strategy for H with initial investment sup .H /,

and denote by V the value process of . The main idea of the proof is to use that

V t H t 0 can be regarded as a new discounted American claim, to which we

can apply Theorem 7.13. However, we must take care of the basic asymmetry of

the hedging problem for American options: The seller of H must hedge against all

possible exercise strategies, while the buyer must find only one suitable stopping time.

#

It will turn out that a suitable stopping time is given by t WD infu t j Uu D Hu .

With this choice, let us define a modified discounted American claim HQ by

HQ u D .Vu Hu / IuD t ;

u D 0; : : : ; T:

364

Chapter 7 Superhedging

ess sup ess sup E HQ j F t D ess sup E HQ t j F t

P 2P

2T t

P 2P

D V t ess inf E H t j F t

P 2P

#

D Vt Ut ;

where we have used that V is a P -martingale in the second and Theorem 6.46 in

#

the third step. Thus, V t U t is equal to the upper Snell envelope UQ " of HQ at time

Q

t . Let be the d -dimensional predictable process obtained from the uniform Doob

decomposition of UQ " . Then, due to part (a) of Theorem 7.13,

V t U t# C

u

X

for all u t .

kDtC1

Thus, WD Q is as desired.

#

(b): Part (a) implies the inequality in (b). To prove its converse, take UQ t 2 U t , a

d -dimensional predictable process ,

Q and 2 T t such that

X

UQ t

Q k .Xk Xk1 / H

P -a.s.

kDtC1

E

h X

i

Q k .Xk Xk1 / F t D 0

for all P 2 P .

(7.7)

kDtC1

UQ t E H j F t ess sup E H j F t

2T t

#

for all P 2 P . Taking the essential infimum over P 2 P thus yields UQ t U t and

in turn (b).

To prove (7.7), let

GQ s WD Ist C1

s

X

s D 0; : : : ; T:

kDtC1

GQ is a P -martingale. Hence (7.7) follows.

365

We conclude this section by stating explicitly the corresponding results for European claims. Recall from Remark 6.6 that every discounted European claim H E

can be regarded as the discounted American claim. Therefore, the results we have

obtained so far include the corresponding European counterparts as special cases.

Corollary 7.15. For any discounted European claim H E such that

sup E H E < 1;

P 2P

there exist two d -dimensional predictable processes and such that P -a.s.

ess sup E H

j Ft C

P 2P

k .Xk Xk1 / H E ;

(7.8)

k .Xk Xk1 / H E :

(7.9)

kDtC1

ess inf E H

P 2P

T

X

T

X

j Ft

kDtC1

sup E H E C

P 2P

T

X

k .Xk Xk1 / H E

P -a.s.

kD1

Thus, the self-financing trading strategy arising from and the initial investment

1 X 0 D supP 2P E H E allows the seller to cover all possible obligations without any downside risk. Similarly, (7.9) yields an interpretation of the self-financing

trading strategy which arises from and the initial investment

1 X 0 D inf E H E :

P 2P

The latter quantity corresponds to the largest loan the buyer can take out and still be

sure that, by using the trading strategy , this debt will be covered by the payoff H E .

}

Remark 7.17. Let H be a discounted European claim such that

sup E H E < 1:

P 2P

O H D sup .H /. If D . 0 ; / is a superhedging

Suppose that PO 2 P is such that E

strategy for H , then

O H C

HO WD E

T

X

kD1

k .Xk Xk1 /

366

Chapter 7 Superhedging

from Theorem 5.25 that

O HO D E

O H :

E

This shows that HO and H are identical and that H is attainable. We have thus obtained

another proof of Theorem 5.32.

}

As the last result in this section, we formulate the following superhedging duality

theorem, which states that the bounds in (7.8) and (7.9) are optimal.

Corollary 7.18. Suppose that H E is a discounted European claim with

sup E H E < 1:

P 2P

"

Denote by U t .H E / the set of all F t measurable random variables UQ t for which there

exists a d -dimensional predictable process Q such that

UQ t C

T

X

P -a.s.

kDtC1

Then

"

P 2P

#

By U t .H E / we denote the set of all F t measurable random variables UQ t for which

there exists a d -dimensional predictable process Q such that

UQ t

T

X

Q k .Xk Xk1 / H E

P -a.s.

kDtC1

Then

P 2P

are acceptable in the sense that there exists a d -dimensional predictable process

such that

ZC

T

X

k .Xk Xk1 / 0

P -a.s.

kD1

.Z/ D infm 2 R j m C Z 2 A;

Z 2 L1 .; FT ; P /:

367

.Z/ D sup E Z :

P 2P

Remark 7.20. Often, the superhedging strategy in a given incomplete model can

be identified as the perfect hedge in an associated extremal model. As an example,

consider a one-period model with d discounted risky assets given by bounded random

variables X 1 ; : : : ; X d . Denote by

the distribution of X D .X 1 ; : : : ; X d / and by

.

/ the convex hull of the support of

. The closure K WD .

/ of .

/ is convex

and compact. We know from Section 1.5 that the model is arbitrage-free if and only if

the price system D . 1 ; : : : ; d / is contained in the relative interior of .

/, and

the equivalent martingale measures can be identified with the measures

with

barycenter . Consider a derivative H D h.X/ given by a convex function h on K.

The cost of superhedging is given by

Z

sup h d

D inf./ j affine on K, h

-a.s. ;

which is a special case of the duality result of Theorem 1.32. Since h is convex

and closed, the condition

. h/ D 1 implies h on K. Denote by M./ the

class of all probability measures on K with barycenter . For any affine function

with h on K, and for any

Q 2 M./ we have

Z

Z

h d

Q d

Q D ./:

Thus,

O

h./

D

Z

sup

h d

Q

(7.10)

2M./

Q

fO WD inf j affine on K, f

-a.s. :

The supremum in (7.10) is attained since M./ is weakly compact. More precisely,

it is attained by any measure

O 2 M./ on K which is maximal with respect to

the balayage order <bal defined for measures on K as in (2.24); see Thorme X. 41

in [87]. But such a maximal measure is supported by the set of extreme points of

the convex compact set K, i.e., by the Choquet boundary of K. This follows from a

general integral representation theorem of Choquet; see, e.g., Thorme X. 43 of [87].

In our finite-dimensional setting,

O can in fact be chosen to have a support consisting

of at most d C 1 points, due to a theorem of Carathodory and the representation of K

368

Chapter 7 Superhedging

as the convex hull of its extreme points; see [221], Theorems 17.1 and 18.5. But this

means that

O can be identified with a complete model, due to Proposition 1.41. Thus,

O

the cost of superhedging h./

can be identified with the canonical price

Z

O WD h d

O

of the derivative H , computed in the complete model

.

O Note that

O sits on the

Choquet boundary of K D .

/, but typically it will no longer be equivalent or

absolutely continuous with respect to the original measure

. As a simple illustration,

consider a one-period model with one risky asset X 1 . If X 1 is bounded, then the

distribution

of X 1 has bounded support, and .

/ is of the form a; b. In this case,

the cost of superhedging H D h.X 1 / for a convex function h is given by the price

p h.b/ C .1 p /h.a/;

computed in the binary model in which X 1 takes only the values a and b, and where

p 2 .0; 1/ is determined by

p b C .1 p /a D 1 :

The following example illustrates that a superhedging strategy is typically too expensive from a practical point of view. However, we will see in Chapter 8 how superhedging strategies can be used in order to construct other hedging strategies which

are efficient in terms of cost and shortfall risk.

Example 7.21. Consider a simple one-period model where S11 has under P a Poisson

distribution and where S 0 1. Let H WD .S11 K/C be a call option with strike

K > 0. We have seen in Example 1.38 that inf .H / and sup .H / coincide with the

universal arbitrage bounds of Remark 1.37

.S01 K/C D inf .H /

and

sup .H / D S01 :

Thus, the superhedging strategy for the seller consists in the trivial hedge of buying

the asset at time 0, while the corresponding strategy for the buyer is a short-sale of the

}

asset in case the option is in the money, i.e., if S01 > K.

7.4

In practice, some derivatives such as put or call options are traded so frequently that

their prices are quoted just like those of the primary assets. The prices of such liquid

options can be regarded as an additional source of information on the expectations of

the market as to the future evolution of asset prices. This information can be exploited

in various ways. First, it serves to single out those martingale measures P which are

369

compatible with the observed options prices, in the sense that the observed prices

coincide with the expectations of the discounted payoff under P . Second, liquid

options may be used as instruments for hedging more exotic options.

Our aim in this section is to illustrate these ideas in a simple setting. Assume that

there is only one risky asset S 1 such that S01 is a positive constant, and that S 0 is

a riskless bond with interest rate r D 0. Thus, the discounted price process of the

risky asset is given by X t D S t1 0 for t D 0; : : : ; T . As the underlying space of

scenarios, we use the product space

WD 0; 1/T :

We define X t .!/ D x t for ! D .x1 ; : : : ; xT / 2 , and denote by F t the -algebra

generated by X0 ; : : : ; X t ; note that F0 D ;; . No probability measure P is given

a priori. Let us now introduce a linear space X of FT -measurable functions as the

smallest linear space such that the following conditions are satisfied:

(a) 1 2 X.

(b) .X t Xs / IA 2 X for 0 s < t T and A 2 Fs .

(c) .X t K/C 2 X for K 0 and t D 1; : : : ; T .

The functions in the space X will be interpreted as (discounted) payoffs of liquid

derivatives. The constant 1 in (a) corresponds to a unit investment into the riskless

bond. The function X t Xs in (b) corresponds to the payoff of a forward contract on

the risky asset, issued at time s for the price Xs and expiring at time t . The decision

to buy such a forward contract at time s may depend on the market situation at time s;

this is taken into account by allowing for payoffs .X t Xs / IA with A 2 Fs . Linearity

of X together with conditions (a) and (b) implies that

Xt 2 X

for all t .

Finally, condition (c) states that call options with any possible strike and any maturity

up to time T can be used as liquid securities.

Suppose that a linear pricing rule is given on X. The value .Y / will be interpreted as the market price of the liquid security Y 2 X. The price of a liquid call

option with strike K and maturity t will be denoted by

C t .K/ WD ..X t K/C /:

Assumption 7.22. We assume that W X ! R is a linear functional which satisfies

the following conditions:

(a) .1/ D 1.

(b) .Y / 0 if Y 0.

370

Chapter 7 Superhedging

(d) C t .K/ D ..X t K/C / ! 0 as K " 1 for all t .

The first two conditions must clearly be satisfied if the pricing rule shall not create arbitrage opportunities. Condition (c) states that Xs is the fair price for a forward

contract issued at time s. This condition is quite natural in view of Theorem 5.29. In

our present setting, it can also be justified by the following simple replication argument. At time s, take out a loan Xs .!/ and use it for buying the asset. At time t ,

the asset is worth X t .!/ and the loan must be paid back, which results in a balance

X t .!/ Xs .!/. Since this investment strategy requires zero initial capital, the price

of the corresponding payoff should also be zero. The continuity condition (d) is also

quite natural.

Our first goal is to show that any such pricing rule is compatible with the paradigm that arbitrage-free prices can be identified as expectations with respect to some

martingale measure for X. More precisely, we are going to construct a martingale

measure P such that .Y / D E Y for all Y 2 X. On the one hand, this will

imply regularity properties of . On the other hand, this will yield an extension of our

pricing rule to a larger space of payoffs including path-dependent exotic options.

As a first step in this direction, we have the following result.

Lemma 7.23. For each t , there exists a unique probability measure

t on 0; 1/

such that for all K 0

Z

C t .K/ D ..X t K/C / D .x K/C

t .dx/:

In particular,

t has the mean

Z

x

t .dx/ D X0 :

Proof. Since K 7! .X t K/C is convex and decreasing, linearity and positivity of

imply that the function t .K/ WD ..X t K/C / is convex and decreasing as well.

Hence, there exists a decreasing right-continuous function f W 0; 1/ ! 0; 1/ such

that

Z K

f .x/ dx

C t .K/ D C t .0/

Z

D X0

0

K

f .x/ dx;

0

i.e., f .K/ is equal to the right-hand derivative of C t .K/ at K. Our fourth condition

on yields

Z

1

f .x/ dx D X0 ;

0

371

t on .0; 1/

such that

f .x/ D

t ..x; 1// for x > 0.

Fubinis theorem implies

Z

x

t .dx/ D

f .y/ dy D X0

.0;1/

and

Z

C t .K/ D X0

Z

D

.0;1/

Iy<x

t .dx/ dy

.x K/C

t .dx/:

.0;1/

t can be extended to a probability measure on 0; 1/, i.e.,

we must show that

t ..0; 1// 1. To this end, we will use the put-call parity

C t .K/ D X0 K C ..K X t /C /;

which follows from our assumptions on . Thus,

Z K

C

..K X t / / D

g.x/ dx;

0

where g.x/ D 1 f .x/. Since K 7! ..K X t /C / is increasing, g must be nonnegative, and we obtain 1 f .0/ D

t ..0; 1//.

The following lemma shows that the measures

t constructed in Lemma 7.23 are

related to each other by the balayage order <bal , defined by

Z

Z

<bal

f d

f d for all convex functions f

for probability measures with finite expectation; see Remark 2.63.

Lemma 7.24. The map t 7!

t is increasing with respect to the balayage order <bal :

tC1 <bal

t

for all t .

.X tC1 K/C .X tC1 X t / IX t >K .X t K/C

.X tC1 K/C .X tC1 K/C IX t >K

0:

372

Chapter 7 Superhedging

Since the price of the forward contract .X tC1 X t / IX t >K vanishes under our pricing

rule , we must have that for all K 0

Z

Z

.x K/C

tC1 .dx/ .x K/C

t .dx/

D C tC1 .K/ C t .K/

0:

An application of Corollary 2.61 concludes the proof.

Let us introduce the class

P D P 2 M1 .; F / j E Y D .Y / for all Y 2 X

of all probability measures P on .; F / which coincide with on X. Note that for

any P 2 P ,

E .X t Xs / IA D 0

so that P consists of martingale measures for X. Our first main result in this section

can be regarded as a version of the fundamental theorem of asset pricing without an

a priori measure P .

Theorem 7.25. Under Assumption 7:22, the class P is non-empty. Moreover, there

exists P 2 P with the Markov property: For 0 s t T and each bounded

measurable function f ,

E f .X t / j Fs D E f .X t / j Xs :

Proof. Since

tC1

R <bal

t , Corollary 2.61 yields the existence of stochastic kernel

Q tC1 such that y Q tC1 .x; dy/ D x and

tC1 D

t Q tC1 . Let us define

P WD

1 Q2 QT ;

i.e., for each measurable set A D 0; 1/T

Z

Z

Z

P A D

1 .dx1 / Q2 .x1 ; dx2 / : : : QT .xT 1 ; dxT / IA .x1 ; x2 ; : : : ; xT /:

Clearly,

t is the law of X t under P . In particular, all call options are priced correctly by calculating their expectation with respect to P . Then one checks that

P -a.s.

Z

E f .X tC1 / j F t D f .y/ Q tC1 .X t ; dy/ D E f .X tC1 / j X t :

(7.11)

The first identity above implies E X tC1 X t j F t D 0. In particular, P is

a martingale measure, and the expectation of .X t Xs / IA vanishes for s < t and

A 2 Fs . It follows that E Y D .Y / for all Y 2 X. Finally, an induction

argument applied to (7.11) yields the Markov property.

373

So far, we have assumed that our space X of liquidly traded derivatives contains

call options with all possible strike prices and maturities. From now on, we will

simplify our setting by assuming that only call options with maturity T are liquidly

traded. Thus, we replace X by the smaller space XT which is defined as the linear

hull of the constants, of all forward contracts

.X t Xs / IA ;

0 s < t T; A 2 Fs ;

.XT K/C ;

K 0,

modeled by a linear pricing rule

T W XT ! R:

We assume that T satisfies Assumption 7.22 in the sense that condition (d) is only

required for t D T

(d0 ) CT .K/ WD T ..XT K/C / ! 0 as K " 1.

By

PT D P 2 M1 .; F / j E Y D T .Y / for all Y 2 XT

we denote the class of all probability measures P on .; F / which coincide with

T on XT . As before, it follows from condition (c) of Assumption 7.22 that any

P 2 PT will be a martingale measure for the price process X. Obviously, any linear pricing rule which is defined on the full space X and which satisfies Assumption 7.22 can be restricted to XT , and this restriction satisfies the above assumptions.

Thus, we have PT P ;.

Proposition 7.26. Under the above assumptions, PT is non-empty.

Proof. Let

T be the measure constructed in Lemma 7.23 from the call prices with

maturity T . Now consider the measure PQ on .; F / defined as

PQ WD X0 X0

T ;

i.e., under PQ we have X t D X0 PQ -a.s. for t < T , and the law of XT is

T . Clearly,

we have PQ 2 PT .

A measure P 2 PT can be regarded as an extension of the pricing rule T

to the larger space L1 .P /, and the expectation E H of some European claim

H 0 can be regarded as an arbitrage-free price for H . Our aim is to obtain upper

and lower bounds for E H which hold simultaneously for all P 2 P . We will

derive such bounds for various exotic options; this will amount to the construction of

certain superhedging strategies in terms of liquid securities.

374

Chapter 7 Superhedging

H dig

1 if max0tT X t B

WD

0 otherwise,

which has a unit payoff if the price process reaches a given upper barrier B > X0 . If

we denote by

B WD inft 0 j X t B

the first hitting time of the barrier B, then the payoff of the digital option can also be

described as

H dig D IB T :

For simplicity, we will assume from now on that

CT .B/ > 0;

so that in particular

T ..B; 1// > 0.

Theorem 7.27. The following upper bound on the arbitrage-free prices of the digital

option holds:

CT .K/

:

(7.12)

max E H dig D min

0K<B B K

P 2PT

Proof. For 0 K < B, we have XB B and

H dig D IB T

.XT K/C

X XT

:

C B

I

B K

B K B T

E H dig

CT .K/

1

C

E .XB XT / IB T :

B K

B K

Since P is a martingale measure, the stopping theorem in the form of Proposition 6.36 implies

E XT IB T D E XB IB T :

This shows that

sup

P 2PT

E H dig

inf

0K<B

CT .K/

:

B K

375

WD 1

inf

0K<B

CT .K/

2 .0; 1/;

B K

then the infimum on the right-hand side is attained in K if and only if K belongs to

the set of
-quantiles for

T , i.e., if and only if

T .0; K//

T .0; K/:

In particular, it is attained in

K WD inf K j

T .0; K/
:

Proof. The convex function CT has left- and right-hand derivatives

.CT /0 .K/ D

T .K; 1//

and

T ..K; 1//I

see also Proposition A.4. Thus, the function g.K/ WD CT .K/=.B K/ has a minimum in K if and only if its left- and right-hand derivatives satisfy

0

.K/ 0 and

g

0

gC

.K/ 0:

0 and g 0 , one sees that these two conditions are equivalent to the

By computing g

C

requirement that K is a
-quantile for

T ; see Lemma A.15.

CT .K/

PO B T D min

:

0K<B B K

Moreover, PO can be taken such that

B D T 1

and

XB D B

PO -a.s. on B T

and

XT > K B T XT K modulo PO -nullsets,

where K is as in Lemma 7:28.

Proof. Let
be as in Lemma 7.28, and let

q.t / WD q

.t / D infK j

T .0; K/ t

; see A.3. We take an auxiliary probability space

Q FQ ; PQ / supporting a random variable U which is uniformly distributed on .0; 1/.

.;

By Lemma A.19, XQ T WD q.U / has distribution

T under PQ . Let be such that

X0 D C B.1 /:

376

Chapter 7 Superhedging

XQ T 1 WD IU C B IU > ;

and we let XQ t WD X0 for 0 t T 2.

We now prove that XQ is a martingale with respect to its natural filtration FQt WD

Q and hence

.XQ 0 ; : : : ; XQ t /. To this end, note first that FQT 2 D ;; ,

Q XQ T 1 j FQT 2 D E

Q XQT 1 D X0 D XQT 2 :

E

Furthermore, since K D q. /,

Q XQ T I XQ T 1 D B D E

Q XQT I U >

E

Q .XQ T K /C C K PQ U >

D E

D CT .K / C K .1 /

D .1 /.B K / C K .1 /

D B PQ XQ T 1 D B :

Hence,

Q XQT E

Q XQT I XQ T 1 D B

Q XQ T I XQ T 1 D D E

E

D X0 B PQ XQ T 1 D B

D PQ XQT 1 D :

It follows that

Q XQ T j FQT 1 D E

Q XQ T j XQT 1 D XQT 1 ;

E

and so XQ is indeed a martingale.

As the next step, we note that

XQT B XQT > K

U > D XQT 1 D B

U XQ T K ;

where we have used the fact that K D q. /. Hence, if we denote by

QB WD inft 0 j XQ t B

the first time at which XQ hits the barrier B, then

377

CT .K/

;

PQ QB T D PQ XQT 1 D B D 1 D min

0K<B B K

and the distribution PO of XQ under PQ is as desired.

Remark 7.30. The inequality

H dig

.XT K/C

X XT

C B

I

B K

B K B T

appearing in the proof of Theorem 7.27 can be interpreted in terms of a suitable superhedging strategy for the claim H dig by using call options and forward contracts:

At time t D 0, we buy .B K/1 call options with strike K, and at the first time

when the price process passes the barrier B, we sell forward .B K/1 shares of

the asset. This strategy will be optimal if the strike price K is such that it realizes the

minimum on the right-hand side of (7.12). By virtue of Lemma 7.28, such an optimal

strike price can be identified as the Value at Risk at level 1
of a short position XT

in the asset.

}

Let us now derive bounds on the arbitrage-free prices of barrier call options. More

precisely, we will consider an up-and-in call option

call

Hu&i WD

0

otherwise,

and the corresponding up-and-out call

.XT K/C

call

WD

Hu&o

0

if max0tT X t < B,

otherwise.

If the barrier B is below the strike price K, then the up-and-in call is identical to a

plain vanilla call .XT K/C , and the payoff of the up-and-out call is zero. Thus,

we assume from now on that

K < B:

Recall that K denotes the minimizer of the function c 7! CT .c/=.B c/ as constructed in Lemma 7.28.

Theorem 7.31. For an up-and-in call option,

if K K,

CT .K/

call

max E Hu&i D BK

P 2PT

BK CT .K / if K > K.

378

Chapter 7 Superhedging

call

Hu&i

B K

cK

.XT c/C C

.XB XT / IB T :

B c

B c

call

Indeed, on XT K or on B > T the payoff of Hu&i

is zero, and the right-hand

side is non-negative. On XT c; B T and on c > XT > K; B T we have

. The expectation of the right-hand side under a martingale measure P 2 PT is

equal to

B K

cK

B K

CT .c/ C

E .XB XT / IB T D

CT .c/;

B c

B c

B c

due to the stopping theorem. The minimum of this upper bound over all c 2 K; B/

is attained in c D K _ K , which shows in the assertion.

Finally, let PO be the martingale measure constructed in Lemma 7.29. If K K

then

.XT K/C IB T D .XT K/C PO -a.s.,

O H call D CT .K/. If K > K then PO -a.s.

and so E

u&i

B K

K K

C

call

.X

K

/

C

.B XT / IB T D Hu&i

:

T

B K

B K

Taking expectations with respect to PO concludes the proof.

Remark 7.32. The inequality

call

Hu&i

B K

cK

.XT c/C C

.XB XT / IB T

B c

B c

appearing in the preceding proof can be interpreted as a superhedging strategy for the

up-and-in call with liquid derivatives: At time t D 0, we purchase .B K/=.B c/

call options with strike c, and at the first time when the stock price passes the barrier

B, we sell forward .c K/=.B c/ shares of the asset. This strategy will be optimal

}

for c D K _ K.

We now turn to the analysis of the up-and-out call option

call

Hu&o

D .XT K/C IB >T :

call

D CT .K/ CT .B/ .B K/

T .B; 1//:

max E Hu&o

P 2PT

379

Proof. Clearly,

call

.XT K/C IXT <B

Hu&o

(7.13)

Taking expectations yields in the assertion.

Now consider the measure PQ on .; F / defined as

PQ WD X0 X0

T ;

i.e., under PQ we have X t D X0 PQ -a.s. for t < T , and the law of XT is

T . Clearly

PQ 2 PT , and (7.13) is PQ -a.s. an identity.

Using the identity

call

call

C Hu&i

;

.XT K/C D Hu&o

Corollary 7.34. We have

min

P 2PT

call

E Hu&i

D CT .B/ C .B K/

T .B; 1//;

and

min

P 2PT

call

E Hu&o

D

0

CT .K/

BK

BK

CT .K /

if K K,

if K > K.

Chapter 8

Efficient hedging

In an incomplete financial market model, a contingent claim typically will not admit

a perfect hedge. Superhedging provides a method for staying on the safe side, but the

required cost is usually too high both from a theoretical and from a practical point of

view. It is thus natural to relax the requirements.

As a first preliminary step, we consider strategies of quantile hedging which stay

on the safe side with high probability. In other words, we maximize the probability

for staying on the safe side under a given cost constraint. The main idea consists

in reducing the construction of such strategies for a given claim H to a problem of

superhedging for a modified claim HQ , which is the solution to a static optimization

problem of NeymanPearson type. Typically, HQ will have the form of a knock-out

option, that is, HQ D H IA . At this stage, we only focus on the probability that a

shortfall occurs; we do not take into account the size of the shortfall if it does occur.

In Sections 8.2 and 8.3 we take a more comprehensive view of the downside risk.

Our discussion of risk measures in Section 4.8 suggests to quantify the downside risk

in terms of an acceptance set for suitably hedged positions. If acceptability is defined

in terms of utility-based shortfall risk as in Section 4.9, we are led to the problem of

constructing efficient strategies which minimize the utility-based shortfall risk under

a given cost constraint. As in the case of quantile hedging, this problem can be decomposed into a static optimization problem and the construction of a superhedging

strategy for a modified payoff profile HQ . In Section 8.3 we go even one step further

and assess the shortfall risk in terms of a general convex risk measure. For a complete

market model and the case of AV@R, we discuss the structure of the modified payoff profile HQ and point out the relation to the problem of robust utility maximization

discussed in Section 3.5.

8.1

Quantile hedging

sup .H / D sup E H < 1:

P 2P

We saw in Corollary 7.15 that there exists a self-financing trading strategy whose

value process V " satisfies

"

VT H

P -a.s.

381

By using such a superhedging strategy, the seller of H can cover almost any possible

obligation which may arise from the sale of H and thus eliminate completely the

corresponding risk. The smallest amount for which such a superhedging strategy is

available is given by sup .H /. This cost will often be too high from a practical point

of view, as illustrated by Example 7.21. Furthermore, if H is not attainable then

sup .H /, viewed as a price for H , is too high from a theoretical point of view since

it would permit arbitrage. Even if H is attainable, a complete elimination of risk by

using a replicating strategy for H would consume the entire proceeds from the sale of

H , and any opportunity of making a profit would be lost along with the risk.

Let us therefore suppose that the seller is unwilling to put up the initial amount

of capital required by a superhedge and is ready to accept some risk. What is the

optimal partial hedge which can be achieved with a given smaller amount of capital?

In order to make this question precise, we need a criterion expressing the sellers

attitude towards risk. Several of such criteria will be studied in the following sections.

In this section, our aim is to construct a strategy which maximizes the probability of a

successful hedge given a constraint on the initial cost.

More precisely, let us fix an initial amount

v < sup .H /:

We are looking for a self-financing trading strategy whose value process maximizes

the probability

P VT H

among all those strategies whose initial investment V0 is bounded by v and which

respect the bounds V t 0 for t D 0; : : : ; T . In view of Theorem 5.25, the second

restriction amounts to admissibility in the following sense:

Definition 8.1. A self-financing trading strategy is called an admissible strategy if its

value process satisfies VT 0.

The problem of quantile hedging consists in constructing an admissible strategy

such that its value process V satisfies

P VT H D max P VT H

(8.1)

where the maximum is taken over all value processes V of admissible strategies subject to the constraint

(8.2)

V0 v:

Note that this problem would not be well posed if considered without the constraint

of admissibility.

Let us emphasize that the idea of quantile hedging corresponds to a Value at Risk

criterion, and that it invites the same criticism: Only the probability of a shortfall is

382

taken into account, not the size of the loss if a shortfall occurs. This exclusive focus

on the shortfall probability may be reasonable in cases where a loss is to be avoided

by any means. But for most applications, other optimality criteria as considered in

the next section will usually be more appropriate from an economic point of view. In

view of the mathematical techniques, however, some key ideas already appear quite

clearly in our present context.

Let us first consider the particularly transparent situation of a complete market

model before passing to the general incomplete case. The set

VT H

will be called the success set associated with the value process V of an admissible

strategy. As a first step, we reduce our problem to the construction of a success set of

maximal probability.

Proposition 8.2. Let P denote the unique equivalent martingale measure in a complete market model, and assume that A 2 FT maximizes the probability P A

among all sets A 2 FT satisfying the constraint

E H IA v:

(8.3)

H WD H IA

solves the optimization problem defined by (8.1) and (8.2), and A coincides up to

P -null sets with the success set of .

Proof. As a first step, let V be the value process of any admissible strategy such that

V0 v. We denote by A WD VT H the corresponding success set. Admissibility

yields that VT H IA . Moreover, the results of Section 5.3 imply that V is a

P -martingale. Hence, we obtain that

E H IA E VT D V0 v:

Therefore, A fulfills the constraint (8.3) and it follows that

P A P A :

As a second step, we consider the trading strategy and its value process V .

Clearly, is admissible, and its success set satisfies

VT H D H IA H A :

On the other hand, the first part of the proof yields that

P VT H P A :

It follows that the two sets A and VT H coincide up to P -null sets. In particular,

is an optimal strategy.

383

Our next goal is the construction of the optimal success set A , whose existence was

assumed in Proposition 8.2. This problem is solved by using the NeymanPearson

lemma. To this end, we introduce the measure Q given by

H

dQ

WD

:

dP

E H

(8.4)

Q A WD

v

E H

(8.5)

Thus, an optimal success set must maximize the probability P A under the constraint Q A . We denote by dP =dQ the generalized density of P with respect

to Q in the sense of the Lebesgue decomposition as constructed in Theorem A.13.

Thus, we may define the level

dP

c WD inf c 0 Q

>

c

E

H

;

(8.6)

dQ

and the set

dP

dP

A WD

> c E H D

>c H :

dQ

dP

(8.7)

Q A D ;

then A maximizes the probability P A over all A 2 FT satisfying the constraint

E H IA v:

Proof. The condition E H IA v is equivalent to Q A D Q A .

Thus, the particular form of the set A in (8.7) and the NeymanPearson lemma in

the form of Proposition A.29 imply that P A P A .

By combining the two Propositions 8.2 and 8.3, we obtain the following result.

Corollary 8.4. Denote by P the unique equivalent martingale measure in a complete market model, and assume that the set A of (8.7) satisfies

Q A D :

Then the optimal strategy solving (8.1) and (8.2) is given by the replicating strategy

of the knock-out option H D H IA .

384

Our solution to the optimization problem (8.1) and (8.2) still relies on the assumption that the set A of (8.7) satisfies Q A D . This condition is clearly satisfied

if

dP

P

D

c

H

D 0:

dP

However, it may not in general be possible to find any set A whose Q -probability

is exactly . In such a situation, the NeymanPearson theory suggests replacing the

indicator function IA of the critical region A by a randomized test, i.e., by an FT measurable 0; 1-valued function . Let R denote the class of all randomized tests,

and consider the following optimization problem:

E

D maxE

j

2 R and EQ

;

where Q is the measure defined in (8.4) and D v=E H as in (8.5). The generalized NeymanPearson lemma in the form of Theorem A.31 states that the solution

is given by

D I

dP >c H

dP

C I

dP Dc H

dP

D

dP

Q dP

> c H

dP

Q dP

D c H

(8.8)

D , i.e.,

dP

D c H 0:

in case P

dP

Definition 8.5. Let V be the value process of an admissible strategy . The success

ratio of is defined as the randomized test

V

D IVT H C

VT

:

I

H VT <H

Note that the set V D 1 coincides with the success set VT H of V . In the

extended version of our original problem, we are now looking for a strategy which

maximizes the expected success ratio E V under the measure P under the cost

constraint V0 v.

Theorem 8.6. Suppose that P is the unique equivalent martingale measure in a

complete market model. Let be given by (8.8), and denote by a replicating

strategy for the discounted claim H D H . Then the success ratio V of

maximizes the expected success ratio E V among all admissible strategies with

initial investment V0 v. Moreover, the optimal success ratio V is P -a.s. equal

to .

385

We do not prove this theorem here, as it is a special case of Theorem 8.7 below

and its proof is similar to the one of Corollary 8.4, once the optimal randomized test

has been determined by the generalized NeymanPearson lemma. Note that the

condition

dP

P

Dc H D0

dP

implies that D IA with A as in (8.7), so in this case the strategy reduces to

the one described in Corollary 8.4.

Now we turn to the general case of an arbitrage-free but possibly incomplete market

model, i.e., we no longer assume that the set P of equivalent martingale measures

consists of a single element, but we assume only that

P ;:

In this setting, our aim is to find an admissible strategy whose success ratio V

satisfies

(8.9)

E V D max E V ;

where the maximum on the right-hand side is taken over all admissible strategies

whose initial investment satisfies the constraint

V0 v:

(8.10)

sup E H

such that

D v;

(8.11)

P 2P

among all

E H

v

for all P 2 P .

(8.12)

H D H

with initial investment sup .H / solves the problem (8.9) and (8.10).

Proof. Denote by R0 the set of all

take a sequence n 2 R0 such that

E

n

! sup E

2R0

as n " 1.

386

converging P -a.s. to a function Q 2 R. Clearly, Q n 2 R0 for each n. Hence, Fatous

lemma yields that

E H Q lim inf E H Q n v

n"1

for all P 2 P ,

E Q D lim E Q n D lim E

n"1

n

D sup E

n"1

;

2R0

We must also show that (8.11) holds. To this end, note first that P D 1 D 1 is

impossible due to our assumption v < sup .H /. Hence, if supP 2P E H < v,

then we can find some " > 0 such that " WD "C.1"/ 2 R0 , and the expectation

E " must be strictly larger than E . This, however, contradicts the maximality

of E .

Now let be any admissible strategy whose value process V satisfies V0 v. If

V denotes the corresponding success ratio, then

so

H

D H ^ VT VT :

E H

Therefore,

E VT D V0 v:

(8.13)

E

E

:

(8.14)

process. Clearly, is an admissible strategy. Moreover,

V0 D sup .H / D sup E H

D v:

P 2P

V

satisfies

E

V

E

:

(8.15)

H

V

D H ^ VT H ^ H D H

equal to one on H D 0, and we obtain that V P -almost surely. According

to (8.15), this can only happen if the two randomized tests V and coincide P almost everywhere. This proves that solves the hedging problem (8.9) and (8.10).

387

8.2

Our starting point in this section is the same as in the previous one: At time T , an

investor must pay the discounted random amount H 0. A complete elimination of

the corresponding risk would involve the cost

sup .H / D sup E H

P 2P

v 2 .0; sup .H //:

This means that the investor is ready to take some risk: Any partial hedging strategy

whose value process V satisfies the capital constraint V0 v will generate a nontrivial shortfall

.H VT /C :

In the previous section, we constructed trading strategies which minimize the shortfall

probability

P VT < H

among the class of trading strategies whose initial investment is bounded by v, and

which are admissible in the sense of Definition 8.1, i.e., their terminal value VT is

non-negative. In this section, we assess the shortfall in terms of a loss function, i.e.,

an increasing function ` W R ! R which is not identically constant. We assume

furthermore that

`.x/ D 0

for x 0 and

E `.H / < 1:

A particular role will be played by convex loss functions, which correspond to risk

aversion in view of the shortfall; compare the discussion in Section 4.9.

Definition 8.8. Given a loss function ` satisfying the above assumptions, the shortfall

risk of an admissible strategy with value process V is defined as the expectation

E `.H VT / D E `. .H VT /C /

of the shortfall weighted by the loss function `.

Our aim is to minimize the shortfall risk among all admissible strategies satisfying

the capital constraint V0 v. Alternatively, we could minimize the cost under a

given bound on the shortfall risk. In other words, the problem consists in constructing

strategies which are efficient with respect to the trade-off between cost and shortfall

risk. This generalizes our discussion of quantile hedging in the previous Section 8.1,

which corresponds to a minimization of the shortfall risk with respect to the nonconvex loss function

`.x/ D I.0;1/ .x/:

388

Remark 8.9. Recall our discussion of risk measures in Chapter 4. From this point of

view, it is natural to quantify the downside risk in terms of an acceptance set A for

hedged positions. As in Section 4.8, we denote by AN the class of all positions X such

that there exists an admissible strategy with value process V such that

V0 D 0 and

X C VT A P -a.s.

for some A 2 A. Thus, the downside risk of the position H takes the form

N

.H / D infm 2 R j m H 2 A:

Suppose that the acceptance set A is defined in terms of shortfall risk, i.e.,

A WD X 2 L1 j E `.X / x0 ;

where ` is a convex loss function and x0 is a given threshold. Then .H / is the

smallest amount m such that there exists an admissible strategy whose value process

V satisfies V0 D m and

E `..H VT /C / x0 :

For a given m, we are thus led to the problem of finding a strategy which minimizes

the shortfall risk under the cost constraint V0 m. In this way, the problem of

quantifying the downside risk of a contingent claim is reduced to the construction of

efficient hedging strategies as discussed in this section.

}

As in the preceding section, the construction of the optimal hedging strategy is

carried out in two steps. The first one is to solve the static problem of minimizing

E `.H Y /

among all FT -measurable random variables Y 0 which satisfy the constraints

sup E Y v:

P 2P

that 0 Y H or, equivalently, that Y D H for some randomized test ,

which belongs to the set R of all FT -measurable random variables with values in

0; 1. Thus, the static problem can be reformulated as follows: Find a randomized

test 2 R which minimizes the shortfall risk

E `. H.1

among all

/ /

(8.16)

E H

v

for all P 2 P .

(8.17)

The next step is to fit the terminal value VT of an admissible strategy to the optimal

profile H . It turns out that this step can be carried out without any further assumptions on our loss function `. Thus, we assume at this point that the optimal of step

one is granted, and we construct the corresponding optimal strategy.

389

Theorem 8.10. Given a randomized test which minimizes (8.16) subject to (8.17),

a superhedging strategy for the modified discounted claim

H WD H

with initial investment sup .H / has minimal shortfall risk among all admissible

strategies which satisfy the capital constraint 1 X 0 v.

Proof. The proof extends the last argument in the proof of Theorem 8.7. As a first

step, we take any admissible strategy such that the corresponding value process V

satisfies the capital constraint V0 v. Denote by

VT

I

H VT <H

the corresponding success ratio. It follows as in (8.13) that

V

D IVT H C

E H

V

for all P 2 P .

v

Thus, the optimality of implies the following lower bound on the shortfall risk

of :

E `.H VT / D E `. H.1 V / / E `. H.1 / /:

In the second step, we consider the admissible strategy and its value process V .

On the one hand,

V0 D sup .H / D sup E H

v;

P 2P

so satisfies the capital constraint. Hence, the first part of the proof yields

E `.H VT / D E `. H.1

On the other hand, VT H D H

,

V

V / /

E `. H.1

/ /:

(8.18)

and therefore

P -a.s.

Hence, the inequality in (8.18) is in fact an equality, and the assertion follows.

Let us now return to the static problem defined by (8.16) and (8.17). We start by

considering the special case of risk aversion in view of the shortfall.

Proposition 8.11. If the loss function ` is convex, then there exists a randomized test

2 R which minimizes the shortfall risk

E `. H.1

among all

/ /

E H

v

for all P 2 P .

is uniquely determined on H > 0.

(8.19)

390

Proof. The proof is similar to the one of Proposition 3.36. Let R0 denote the set of

all randomized tests which satisfy the constraints (8.19). Take n 2 R0 such that

E`. H.1 n / / converges to the infimum of the shortfall risk, and use Lemma 1.70

to select convex combinations Q n 2 conv n ; nC1 ; : : : which converge P -a.s. to

some Q 2 R. Since ` is continuous and increasing, Fatous lemma implies that

E `. H.1 Q / / lim inf E `. H.1 Q n / / D inf E `. H.1

2R0

n"1

/ /;

where we have used the convexity of ` to conclude that E`. H.1 Q n / / tends to

the same limit as E`. H.1 n / /.

Fatous lemma also yields that for all P 2 P

E H Q lim inf E H Q n v:

n"1

uniqueness part is obvious.

Remark 8.12. The proof shows that the analogous existence result holds if we use a

robust version of the shortfall risk defined as

sup EQ `. H.1

/ /;

Q2Q

where Q is a class of equivalent probability measures; see also Remark 3.37 and

Sections 8.2 and 8.3.

}

Combining Proposition 8:11 and Theorem 8.10 yields existence and uniqueness

of an optimal hedging strategy under risk aversion in a general arbitrage-free market

model.

Corollary 8.13. Assume that the loss function ` is strictly convex on 0; 1/. Then

there exists an admissible strategy which is optimal in the sense that it minimizes

the shortfall risk among all admissible strategies subject to the capital constraint

1 X 0 v. Moreover, any optimal strategy requires the exact initial investment v,

and its success ratio is P -a.s. equal to

where

IH >0 C IH D0 ;

and Theorem 8.10. Strict convexity of ` implies that is P -a.s. unique on H > 0.

Since ` is strictly increasing on 0; 1/, and the success ratio V of any optimal

391

equal to 1 by definition.

Since ` is strictly increasing on 0; 1/, we must have that

sup E H

V

is

D v;

P 2P

for otherwise we could find some " > 0 such that " WD " C .1 "/ would

also satisfy the constraints (8.17). Since we have assumed that v < sup .H /, the

constraints (8.17) imply that 1 and hence that

E `. H.1

" / /

< E `. H.1

/ /:

Since the value process V of an optimal strategy is a P -martingale, and since

VT H

V

DH

v V0 D sup E VT sup E H

P 2P

D v:

P 2P

Beyond the general existence statement of Proposition 8.11, it is possible to obtain

an explicit formula for the optimal solution of the static problem if the market model

is complete. Recall that we assume that the loss function `.x/ vanishes for x 0. In

addition, we will also assume that

` is strictly convex and continuously differentiable on .0; 1/.

Then the derivative `0 of ` is strictly increasing on .0; 1/. Let J denote the inverse function of `0 defined on the range of `0 , i.e., on the interval .a; b/ where

a WD limx#0 `0 .x/ and b WD limx"1 `0 .x/. We extend J to a function J C W 0; 1 !

0; 1 by setting

C1 for y b,

C

J .y/ WD

0

for y a.

From now on, we assume also that

P D P ;

i.e., P is the unique equivalent martingale measure in a complete market model. Its

density will be denoted by

dP

:

' WD

dP

392

Theorem 8.14. Under the above assumptions, the solution of the static optimization

problem of Proposition 8:11 is given by

J C .c ' /

^ 1 P -a.s. on H > 0;

H

where the constant c is determined by the condition E H D v.

D1

Proof. The problem is of the same type as those considered in Section 3.3. It can in

fact be reduced to Corollary 3.43 by considering the random utility function

u.x; !/ WD `.H.!/ x/;

0 x H.!/:

Just note that the shortfall risk E `.H Y / coincides with the negative expected

utility Eu.Y; / for any profile Y such that 0 Y H . Moreover, since our

market model is complete, it has a finite structure by Theorem 5.37, and so all integrability conditions are automatically satisfied. Thus, Corollary 3.43 states that the

optimal profile H WD Y which maximizes the expected utility E u.Y; / under

the constraints 0 Y H and E Y v is given by

H .!/ D I C .c ' .!/; !/ ^ H.!/ D .H.!/ J C .c ' .!///C :

Dividing by H yields the formula for the optimal randomized test

.

Corollary 8.15. In the situation of Theorem 8:14, suppose that the objective probability measure P is equal to the martingale measure P . Then the modified discounted

claim takes the simple form

H D H

D .H J C .c //C :

Example 8.16. Consider the discounted payoff H of a European call option .STi

K/C with strike K under the assumption that the numraire S 0 is a riskless bond, i.e.,

that S t0 D .1 C r/t for a certain constant r 0. If the assumptions of Corollary 8.15

hold, then the modified profile H is the discounted value of the European call option

struck at KQ WD K C J C .c / .1 C r/T , i.e.,

H D

Q C

.STi K/

:

.1 C r/T

Example 8.17. Consider an exponential loss function `.x/ D .e x 1/C for some

> 0. In this case,

1

y C

C

J .y/ D

; y 0;

log

c' C

1

log

^ H:

}

H DH

393

`.x/ D

xp

;

p

x 0;

for some p > 1, then the problem is to minimize a lower partial moment of the

difference VT H . Theorem 8:14 implies that it is optimal to hedge the modified

claim

H p D H .cp ' /1=.p1/ ^ H

(8.20)

}

where the constant cp is determined by E H p D v.

Let us now consider the limit p " 1 in (8.20), corresponding to ever increasing

risk aversion with respect to large losses.

Proposition 8.19. Let us consider the loss functions

`p .x/ D

xp

;

p

x 0;

L1 .P / to the discounted claim

p

.H c1 /C

where the constant c1 is determined by

E .H c1 /C D v:

(8.21)

.' / .p/ ! 1

P -a.s. as p " 1.

.pn /

lim H

n"1

pn

D H cQ ^ H D .H cQ /C :

Hence,

E H

pn

! E .H cQ /C :

E .H cQ /C D v;

which determines cQ uniquely as the constant c1 of (8.21).

394

Example 8.20. If the discounted claim H in Proposition 8.19 is the discounted payoff

of a call option with strike K, and the numraire is a riskless bond as in Example 8.16,

then the limiting profile limp"1 H p is equal to the discounted call with the higher

}

strike price K C c1 ST0 .

In the remainder of this section, we consider loss functions which are not convex

but which correspond to risk neutrality and to risk-seeking preferences. Let us first

consider the risk-neutral case.

Example 8.21. In the case of risk neutrality, the loss function is given by

`.x/ D x

for x 0.

E .H VT /C

under the capital constraint V0 v. Let P be the unique equivalent martingale measure in a complete market model. Then the static problem corresponding to Proposition 8:11 is to maximize the expectation

E H

under the constraint that

2 R satisfies

E H

v:

H

dQ

D

dP

E H

and

H

dQ

D

:

dP

E H

under the side condition

v

:

EQ WD

E H

Since the density dQ=dQ is proportional to the inverse of the density ' D dP =dP ,

Theorem A.31 implies that the optimal test takes the form

1

P -a.s. on H > 0,

c1 D supc 2 R j E H I ' < c v ;

and where the constant 2 0; 1 is chosen such that E H

1 can be arbitrary.

1

D v. On H D 0,

}

395

Assume now that the shortfall risk is assessed by an investor who, instead of being

risk-averse, is in fact inclined to take risk. In our context, this corresponds to a loss

function which is concave on 0; 1/ rather than convex. It is not difficult to generalize

Theorem 8.14 so that it covers this situation. Here we limit ourselves to the following

explicit case study.

Example 8.22. Consider the loss function

`.x/ D

xq

;

q

x 0;

for some q 2 .0; 1/. In order to solve our static optimization problem, one could apply

the results and techniques of Section 3.3. Here we will use an approach based on the

NeymanPearson lemma. Note first that for 2 R

`.H.1

// D .1

/q `.H / `.H /

`.H /:

E `.H.1

// E`.H / E

`.H / :

(8.22)

The problem of finding a minimizer of the right-hand side is equivalent to maximizing the expectation EQ under the constraint that EQ v=E H for the

measures Q and Q defined via

dQ

Hq

D

dP

E H q

and

dQ

H

;

D

dP

E H

optimal test must be of the form

I1>c ' H 1q C I1Dc ' H 1q

q

(8.23)

for certain constants cq and . Under the simplifying assumption that

P 1 D cq ' H 1q D 0;

(8.24)

D

0 otherwise.

(8.25)

q

By taking D

for E `.H.1

q

}

// under the constraint that E H v.

396

H I1>c H ' ;

0

which was obtained as the solution to the problem of quantile hedging by taking the

limit q # 0 in (8.25). Intuitively, decreasing q corresponds to an increasing appetite

for risk in view of the shortfall.

Proposition 8.23. Let us assume for simplicity that (8.24) holds for all q 2 .0; 1/,

that P H > 0 D 1, and that there exists a unique constant c0 such that

E H I1>c H ' D v:

(8.26)

q

0

D I1>c H '

0

Proof. Take any sequence qn # 0 such that .cqn /1=.1qn / converges to some cQ 2

0; 1. Then

:

lim qn D I1>cH

Q

'

n"1

Hence,

E H

qn

! E H I1>cH

:

Q

'

Since we assumed (8.24) for all q 2 .0; 1/, the left-hand terms are all equal to v, and

it follows from (8.26) that cQ D c0 . This establishes the desired convergence.

8.3

.H VT /C

arising from hedging the discounted claim H with a self-financing trading strategy

with initial capital

V0 D v 2 .0; sup .H //:

In this section, our aim is to minimize the shortfall risk

..H VT /C /;

where is a given convex risk measure as discussed in Chapter 4. Here we assume that

is defined on a suitable function space, such as Lp .; F ; P /, so that the shortfall

397

risk is well-defined and finite; cf. Remark 4.44. In particular, we assume that .Y / D

.YQ / whenever Y D YQ P -a.s.

As in the preceding two sections, the construction of the optimal hedging strategy can be carried out in two steps. The first step is to solve the static problem of

minimizing

..H Y /C /

over all FT -measurable random variables Y 0 that satisfy the constraint

sup E Y v:

P 2P

Y H , and we can reformulate the problem as

minimize .Y H / subject to 0 Y H and sup E Y v:

(8.27)

P 2P

The next step is to fit the terminal value VT of an admissible strategy to the optimal

profile Y . It turns out that this step can be carried out without any further assumptions on our risk measure . Thus, we assume at this point that the optimal Y of step

one is granted, and we construct the corresponding optimal strategy.

Proposition 8.24. A superhedging strategy for a solution Y of (8.27) with initial

investment sup .Y / has minimal shortfall risk among all admissible strategies whose

value process satisfies the capital constraint V0 v.

Proof. Let V be the value process of any admissible strategy such that V0 v. Due

to Doobs systems theorem in the form of Theorem 5.14, V is a martingale under any

P 2 P , and so

sup E VT D V0 v:

P 2P

..H VT /C / D .Y H / .Y H /:

Next let V be the value process of a superhedging strategy for Y with initial

investment

V0 D sup .Y / D sup E Y :

P 2P

Then we have

V0

v and

VT

This concludes the proof.

398

Proposition 8.25. If is lower semicontinuous with respect to P -a.s. convergence of

random variables in the class Y j 0 Y H and .Y / < 1 for one such Y ,

then there exists a solution of the static optimization problem (8.27). In particular,

there exists a solution if H is bounded and is continuous from above.

Proof. Take a sequence Yn with 0 Yn H and supP 2P E Yn v such that

.Yn H / converges to the infimum A of the shortfall risk. We can use Lemma 1.70

to select convex combinations Zn 2 convYn ; YnC1 ; : : : which converge P -a.s. to

some random variable Z. Then 0 Z H and Fatous lemma yields that

E Z lim inf E Zn v

n"1

.Z H / lim inf .Zn H /:

n"1

the desired minimizer.

Combining Proposition 8:25 and Proposition 8.24 yields the existence of a riskminimizing hedging strategy in a general arbitrage-free market model. So far, all

arguments were practically the same as in the preceding two sections.

Beyond the general existence statement of Proposition 8.25, it is sometimes possible to obtain an explicit formula for the optimal solution of the static problem if the

market model is complete and so P D P . In this case, the static optimization

problem (8.27) simplifies to

minimize .Y H / subject to 0 Y H and E Y v:

By substituting Z for H Y , this is equivalent to the problem

Q

minimize .Z/ subject to 0 Z H and E Z v;

(8.28)

.

Our approach relies on the general idea that a minimax problem can be transformed

into a standard minimization problem by using a duality result for the expression

involving the maximum. In the case of AV@R

, we can use the following representation of AV@R

from Lemma 4.51,

AV@R

.Z/ D

1

1

min.E .Z r/C C
r/ D min.E .Z r/C C
r/ (8.29)

r2R

r0

399

for Z 0. Our discussion of problem (8.28) will be valid also beyond the setting

of a complete discrete-time market model, whose underlying probability space has

necessarily a discrete structure by Theorem 5.37. In fact it applies whenever P and

P are two equivalent probability measures on a given measurable space .; F /.

This is important in view of the application of the next theorem in Example 3.50.

Theorem 8.26. Suppose that H 2 L1 .P / and denote by ' WD dP =dP the price

density of P with respect to P . Then the problem (8.28) admits a solution for D

AV@R

which is of the form

Z D H I'>c C .H ^ r /I'<c C .H C .1 /.H ^ r //I'Dc

(8.30)

Proof. Recall from Section 4.4 that

AV@R

.Y / D sup EQ Y ;

Q2Q

where Q

is the set of all probability measures Q

P whose density dQ=dP is

P -a.s. bounded by 1=
. Thus it follows from Fatous lemma that AV@R

is lower

semicontinuous with respect to P -a.s. convergence of random variables in the class

Y j 0 Y H (the same argument also gives upper semicontinuity and hence

continuity, but this fact is not needed here). Proposition 8.25 hence yields the existence of a solution Z of the minimization problem (8.28). By (8.29), Z must

solve

Q

minimize E .Z r /C subject to 0 Z H and E Z v;

(8.31)

AV@R

.Z / D

1

E .Z r /C C r :

(8.32)

Let us now solve (8.31). To this end, we consider first the case in which r D 0.

Then we are in the situation of Example 8.21 and obtain

Z D H I'>c C H I'Dc

for constants c > 0 and 2 0; 1. This is indeed a special case of (8.30).

Now we consider the case r > 0. Note first that we must have Z H ^ r .

Indeed, let us assume P Z < H ^ r > 0. Then we could obtain a strictly lower

risk AV@R

.Z / either by decreasing the level r in case P Z H ^ r D 1

or, in case P Z > H ^ r > 0, by shifting mass of Z from Z > H ^ r to

the set Z < H ^ r .

400

to

0 ZO H H ^ r

and

E ZO vO WD vQ E H ^ r :

problem is equivalent to

O

minimize E ZO subject to 0 ZO H H ^ r and E ZO v.

(8.33)

Hence, it is solved by

ZO D .H H ^ r /I'>c C .H H ^ r /I'Dc

for some constants c > 0 and 2 0; 1. It follows that

Z D ZO CH ^r D H I'>c C.H ^r /I'<c C.H C.1/.H ^r //I'Dc :

We now solve (8.28) in the more specific situation in which H D 1. In this case,

Q The next result shows that these

one sees that r and c are functions of the capital v.

functions behave as follows. As long as vQ is below some critical threshold v , we

Q is determined by the requirement E Z D v.

Q Above

have r D 0, and c D c.v/

the critical threshold v , the value of c is always equal to c.v /, and now r > 0 is

Q

determined by the requirement E Z D v.

Theorem 8.27. Consider the setting of Theorem 8:26. Assume in addition that H D 1

and that ' has a continuous and strictly increasing quantile function q' with respect

to P and satisfies k'k1 > 1 . Then the solution Y of problem (8.28) is P -a.s.

unique. Moreover, there exists a critical capital level v such that

Z D I'>q' .t0 /

P -a.s. for vQ v ;

Q and

Z D r C .1 r /I'>q' .t /

where r D 1

equation

1vQ

.t /

> 0, .t / WD

Rt

0

is the unique solution of the

q' .t

/.t

1 C
/ D .t

/:

Finally, the critical capital level v is equal to 1 .t

/.

Proof. We fix vQ 2 .0; 1/. For a constant c we then let

Zr D r C .1 r/I'>c ;

(8.34)

401

Q i.e.,

r.c/ D

:

E 'I ' c

This makes sense as long as c c0 , where c0 is defined via vQ D E 'I ' > c0 .

Theorem 8.26 states that a solution of our problem can be found within the class

Zr.c/ j c c0 . Thus, we have to minimize

Z 1

Z 1

qZr.c/ .s/ ds D
r.c/ C .1 r.c//

Iq' .s/>c ds

AV@R

.Zr.c/ / D

1

1

over c c0 . Here we have used Lemma A.23 in the second identity. This minimization problem can be simplified further be using the reparameterization c D q' .t /,

which is one-to-one according to our assumptions. Indeed, by letting

%.t / WD r.q' .t // D 1

1 vQ

;

.t /

Z

R.t / WD AV@R

.Z%.t/ / D %.t / C .1 %.t //

1

I.t;1 .s/ ds

D .1 v/

Q

.t 1 C
/C

.t /

We show next that the function

.t / WD

t 1C

.t /

2 .1 ; 1, which will define the solution as soon as

t

t0 and as long as t D t0 does not give a better result. To this end, we note first

that

.t / .t 1 C /q' .t /

:

(8.35)

0 .t / D

.t /2

The numerator of this expression is strictly larger than zero for t 1 and equal

to .1 / > 0 at t D 1 . Moreover, for t > 1 ,

Z 1

Z t

q' .s/ ds C

q' .s/ q' .t / ds;

.t / .t 1 C
/q' .t / D

0

1

which is easily seen to be strictly decreasing in t . For t " 1 this expression converges

to 1
k'k1 , which is strictly negative due to our assumption k'k1 >
1 . Hence,

402

2 .1 ; 1/, which is the unique solution

of the equation

q' .t

/.t

1 C / D .t

/;

and this solution t

is consequently the unique maximizer of .

If t

t0 , then R has no minimizer on .t0 ; 1, and it follows that t D t0 is its

minimizer. Let us compare R.t

/ with R.t0 / in case t

> t0 . We have

Q

R.t

/ D .1 v/

.t

1 C
/C

.t

1 C
/C

D
.t0 /

.t

/

.t

/

and

R.t0 / D
.t0 C
1/C D
.t0 /

.t0 1 C
/C

:

.t0 /

Since t

is the unique maximizer of the function t 7! .t 1 C
/C =.t /, we thus see

that R.t

/ is strictly smaller than R.t0 /. Hence the solution is defined by

t WD t0 _ t

:

Clearly, t

is independent of v,

Q while t0 decreases from 1 to 0 as vQ increases from 0

to 1. Thus, by taking v as the capital level for which t

D t0 , we see that the optimal

solution has the form

I

for vQ v ,

' .t0 /

Z D '>q

r C .1 r /I'>q' .t / for vQ > v ,

1vQ

> 0.

where r D 1 .t

/

Finally, when vQ is equal to the critical capital level v , we must have that t

D t0 .

Q and so v D 1 .t

/.

But t0 was defined to be the solution of 1 .t / D v,

Let us now point out the connections of the preceding theorem with robust statistical

test theory as explained in Section 3.5 and in particular in Remark 3.54. To this end, let

R denote the set of all measurable functions W ! 0; 1, which will be interpreted

as randomized statistical tests; see Remark A.32. Problem (8.28) for H D 1 can then

be rewritten as

minimize sup EQ

subject to

2 R and E

v,

Q

Q2Q

where Q

is the set of all probability measures Q

P whose density dQ=dP is

P -a.s. bounded by 1=
. The solution of this problem is described in Theorem

8.27. It is easy to see that must also solve the following dual problem:

maximize E

subject to

2 R and sup EQ

Q2Q

;

403

where D supQ2Q EQ . Thus, is an optimal randomized test for testing the hypothesis P against the composite null hypothesis Q

; see Remark 3.54.

When Q

admits a least-favorable measure Q0 with respect to P in the sense of

Definition 3.48, then must also be a standard NeymanPearson test for testing the

hypothesis P against the null hypothesis Q0 . By Theorem A.31, it must hence be of

the form

D I

Dc C I

>c ;

where c > 0 and 2 0; 1 are constants and

D

dP

:

dQ0

Under the assumptions of Theorem 8.27, this is the case when D c.' _ q' .t

//,

where c is a suitable constant. It follows that

dQ0

1

'

D

:

dP

c ' _ q' .t

/

We have by (8.34),

E

i

1

'

D P ' > q' .t

/ C

E 'I ' q' .t

/

' _ q' .t

/

q' .t

/

D 1 t

C

1

.t

/ D
:

q' .t

/

dQ0

1

'

D

dP

' _ q' .t

/

(8.36)

with

D

dP

D
.' _ q' .t

//:

dQ0

Corollary 8.28. For a measure P P satisfying the assumptions of Theorem 8:27,

the measure Q0 in (8.36) is a least-favorable measure for

dQ

Q

D Q 2 M1 .P /

P -a.s.

dP

in the sense of Definition 3:48.

Chapter 9

So far, we have focused on frictionless market models, where asset transactions can

be carried out with no limitation. In this chapter, we study the impact of market imperfections generated by convex trading constraints. Thus, we develop the theory of

dynamic hedging under the condition that only trading strategies from a given class S

may be used. In Section 9.1 we characterize those market models for which S does not

contain arbitrage opportunities. Then we take a direct approach to the superhedging

duality for American options. To this end, we first derive a uniform Doob decomposition under constraints in Section 9.2. The appropriate upper Snell envelopes are

analyzed in Section 9.3. In Section 9.4 we derive a superhedging duality under constraints, and we explain its role in the analysis of convex risk measures in a financial

market model.

9.1

In practice, it may be reasonable to restrict the class of trading strategies which are

admissible for hedging purposes. As discussed in Section 4.8, there may be upper

bounds on the capital invested into risky assets, or upper and lower bounds on the

number of shares of an asset. Here we model such portfolio constraints by a set S of

d -dimensional predictable processes, viewed as admissible investment strategies into

risky assets. Throughout this chapter, we will assume that S satisfies the following

conditions:

(a) 0 2 S.

(b) S is predictably convex: If ; 2 S and h is a predictable process with 0

h 1, then the process

h t t C .1 h t / t ;

t D 1; : : : ; T;

belongs to S.

(c) For each t 2 1; : : : ; T , the set

S t WD t j 2 S

is closed in L0 .; F t1 ; P I Rd /.

(d) For all t , t 2 S t implies t? 2 S t .

405

In order to explain condition (d), let us recall from Lemma 1.66 that each t 2

L0 .; F t1 ; P I Rd / can be uniquely decomposed as

t D t C t? ;

where t 2 N t and t? 2 N t? ,

and where

N t D t 2 L0 .; F t1 ; P I Rd / j t .X t X t1 / D 0 P -a.s. ;

N t? D t 2 L0 .; F t1 ; P I Rd / j t t D 0 P -a.s. for all t 2 N t :

Remark 9.1. Under condition (d), we may replace t .X t X t1 / by t? .X t X t1 /,

and t? .X t X t1 / D 0 P -a.s. implies t? D 0. Note that condition (d) holds

if the price increments satisfy the following non-redundance condition: For all t 2

1; : : : ; T and t 2 L0 .; F t1 ; P I Rd /,

t .X t X t1 / D 0

P -a.s.

H)

t D 0

P -a.s.

(9.1)

}

Example 9.2. For each t let C t be a closed convex subset of Rd such that 0 2 C t .

Take S as the class of all d -dimensional predictable processes such that t 2 C t

P -a.s. for all t . If the non-redundance condition (9.1) holds, then S satisfies conditions (a) through (d). This case includes short sales constraints and restrictions on the

size of a long position.

}

Example 9.3. Let a; b be two constants such that 1 a < 0 < b 1, and take

S as the set of all d -dimensional predictable processes such that

a t X t1 b

P -a.s. for t D 1; : : : ; T .

This class S corresponds to constraints on the capital invested into risky assets. If

we assume that the non-redundance condition (9.1) holds, then S satisfies conditions

(a) through (d). More generally, instead of the two constants a and b, one can take

}

dynamic margins defined via two predictable processes .a t / and .b t /.

Let S denote the set of all self-financing trading strategies D . 0 ; / which arise

from an investment strategy 2 S, i.e.,

S D D . 0 ; / j is self-financing and 2 S :

In this section, our goal is to characterize the absence of arbitrage opportunities in S.

The existence of an equivalent martingale measure P 2 P is clearly sufficient.

Under an additional technical assumption, a condition which is both necessary and

sufficient will involve a larger class PS P . In order to introduce these conditions,

we need some preparation.

406

Q-martingale if there exists a sequence of stopping times .n /n2N T such that

n % T Q-a.s., and such that the stopped processes Z n are Q-martingales. The

sequence .n /n2N is called a localizing sequence for Z. In the same way, we define

local supermartingales and local submartingales.

Remark 9.5. If Q is a martingale measure for the discounted price process X , then

the value process V of each self-financing trading strategy D . 0 ; / is a local

Q-martingale. To prove this, one can take the sequence

n WD inft 0 j j tC1 j > n ^ T

as a localizing sequence. With this choice, j t j n on n t , and the increments

n

V tn V t1

D In t t .X t X t1 /;

t D 1; : : : ; T;

n

j F t1 D In t t EQ X t X t1 j F t1 D 0:

EQ V tn V t1

used in the proof of Theorem 5.25. Throughout this chapter, we will assume that

F0 D ;; and FT D F .

Proposition 9.6. A local Q-supermartingale Z whose negative part Z t is integrable

for each t 2 1; : : : ; T is a Q-supermartingale.

Proof. Let .n / be a localizing sequence. Then

Z tn

T

X

Zs 2 L1 .Q/:

sD0

In view of limn Z tn D Z t , Fatous lemma for conditional expectations implies that

Q-a.s.

n

D Z t1 :

EQ Z t j F t1 lim inf EQ Z tn j F t1 lim inf Z t1

n"1

n"1

follows.

Exercise 9.1.1. Let Z be a local Q-martingale with ZT 0. Show that in our

situation, where F0 D ;; , Z is a Q-martingale.

407

that

(9.2)

X t 2 L1 .PQ / for all t ,

and such that the value process of any trading strategy in S is a local PQ -supermartingale.

Remark 9.8. If S contains all self-financing trading strategies D . 0 ; / with

bounded , then PS coincides with the class P of all equivalent martingale measures. To prove this, let PQ 2 PS , and note that the value process V of any such is a

PQ -supermartingale by (9.2) and by Proposition 9.6. The same applies to the strategy

, so V is in fact a PQ -martingale, and Theorem 5.14 shows that PQ is a martingale

measure for X.

}

Our first goal is to extend the fundamental theorem of asset pricing to our present

setting; see Theorem 5.16. Let us introduce the positive cone

R WD j 2 S; 0

generated by S. Accordingly, we define the cones R and R t . Clearly, R contains no

arbitrage opportunities if and only if S is arbitrage-free. We will need the following

condition on the L0 -closure RO t of R t :

for each t , RO t \ L1 .; F t ; P I Rd / R t .

(9.3)

for all t .

Theorem 9.9. Under condition (9.3), there are no arbitrage opportunities in S if and

only if PS is non-empty. In this case, there exists a measure PQ 2 PS which has a

bounded density d PQ =dP .

Example 9.10. In the situation of Example 9.2, condition (9.3) will be satisfied as

soon as the cones generated by the convex sets C t are closed in Rd . This case includes short sales constraints and constraints on the size of a long position, which are

modeled by taking C t D a1t ; b t1 adt ; b td for certain numbers akt ; b tk such that

1 akt 0 b tk 1.

Example 9.11. Consider now the situation of Example 9.3. We claim that S does not

contain arbitrage opportunities if and only if the unconstrained market is arbitragefree, so that we have PS D P . To prove this, note that the existence of an arbitrage

opportunity in the unconstrained market is equivalent to the existence of some t and

some F t1 -measurable t such that t .X t X t1 / 0 P -a.s. and P t .X t

X t1 / > 0 > 0 (see Proposition 5.11). Next, there exists a constant c > 0 such that

these properties are shared by Qt WD t Ij t X t 1 jc and in turn by "Qt , where " > 0.

But "Qt 2 S t if " is small enough.

408

As to the proof of Theorem 9.9, we will first show that the condition PS ;

implies the absence of arbitrage opportunities in S.

Proof of sufficiency in Theorem 9:9. Suppose PQ is a measure in PS , and V is the

value process of a trading strategy in S such that VT 0 P -almost surely. Combining Lemma 9.12 below with Proposition 9.6 shows that V is a PQ -supermartingale.

Q VT , so V cannot be the value process of an arbitrage opportunity.

Hence V0 E

Lemma 9.12. Suppose that PS ; and that V is the value process of a trading

strategy in S such that VT 0 P -almost surely. Then Vt 0 P -a.s. for all t .

Proof. The assertion will be proved by backward induction on t . We have VT 0

by assumption, so let us assume that V t 0 P -a.s. for some t . For D . 0 ; / 2 S

with value process V , we let s.c/ WD s Ijs jc for c > 0 and for all s. Then the value

process V .c/ of .c/ is a PQ -supermartingale for any fixed PQ 2 PS . Furthermore,

.c/

V t1 Ij t jc D V t Ij t jc t .X t X t1 /

.c/

t

.X t X t1 /

.c/

.c/

D V t1 V t :

The last term on the right belongs to L1 .PQ /, so we may take the conditional expectaQ j F t1 on both sides of the inequality. We get

tion E

Q V .c/ V t.c/ j F t1 0

V t1 Ij t jc E

t1

PQ -a.s.

Let us now prepare for the proof that the condition PS ; is necessary. First we

argue that the absence of arbitrage opportunities in S is equivalent to the absence of

arbitrage opportunities in each of the embedded one-period models, i.e., to the nonexistence of t 2 S t such that t .X t X t1 / amounts to a non-trivial positive gain.

This observation will allow us to apply the techniques of Section 1.6. Let us denote

S 1 WD 2 S j is bounded:

Similarly, we define

S t1 WD t j 2 S 1 D S t \ L1 .; F t1 ; P I Rd /:

409

(a) There exists an arbitrage opportunity in S.

(b) There exist t 2 1; : : : ; T and t 2 S t such that

t .X t X t1 / 0 P -a.s.,

(9.4)

Proof. The proof is essentially the same as the one of Proposition 5.11.

In order to apply the results of Section 1.6, we introduce the convex sets

K tS WD t .X t X t1 / j t 2 S t ;

for t 2 1; : : : ; T . Lemma 9.13 shows that S contains no arbitrage opportunities if

and only if the condition

K tS \ L0C D 0

(9.5)

holds for all t 2 1; : : : ; T .

Lemma 9.14. Condition (9.5) implies that K tS L0C .; F t ; P / is a closed convex

subset of L0 .; F t ; P /.

Proof. The proof is essentially the same as the one of Lemma 1.68. Only the following additional observation is required: If . n / is sequence in S t , and if and are

two F t1 -measurable random variables such that 0 1 and is integer-valued,

then WD 2 S t . Indeed, predictable convexity of S and our assumption that

0 2 S imply that

n

X

I Dk k 2 S t

kD1

D

1

X

I Dk k 2 S t :

kD1

E jXs j < 1

for all s.

(9.6)

For the purpose of proving Theorem 9.9, this can be assumed without loss of generality: If (9.6) does not hold, then we replace P by an equivalent measure P 0 which

410

has a bounded density dP 0 =dP and for which the price process X is integrable. For

instance, we can take

T

h X

i

jXs j dP;

dP 0 D c exp

sD1

where c denotes the normalizing constant. If there exist a measure PQ P 0 such that

each value process for a strategy in S is a local PQ -supermartingale and such that the

density d PQ =dP 0 is bounded, then PQ 2 PS , and the density d PQ =dP is bounded as

well.

Lemma 9.15. If S contains no arbitrage opportunities and condition (9.3) holds,

then for each t 2 1; : : : ; T there exists some Zt 2 L1 .; F t ; P / such that Z t > 0

P -a.s. and such that

E Z t t .X t X t1 / 0

for all 2 S 1 .

(9.7)

Proof. Recall that R does not contain arbitrage opportunities if and only if S is

arbitrage-free. Hence, for each t , K tR \ L0C .; F t ; P / D 0 by Lemma 9.13.

By the equivalence of conditions (a) and (c) of same lemma and condition (9.3), we

even get

O

(9.8)

K tR \ L0C .; F t ; P / D 0

where RO t denotes again the L0 -closure of R t . The cone RO t satisfies all conditions

required from S t , and hence Lemma 9.14 implies that each

O

C tR WD .K tR L0C .; F t ; P // \ L1

is a closed convex cone in L1 which contains L1

C .; F t ; P /. Furthermore, it

follows from (9.8) and the argument in the proof of (a) , (b) of Theorem 1.55

O

O

that C tR \ L0C D 0, so C tR satisfies the assumptions of the KrepsYan theorem,

which is stated in Theorem 1.62. We conclude that there exist Z t 2 L1 .; F t ; P /

O

such that P Z t > 0 D 1, and such that E Z t W 0 for each W 2 C tR . As

O

t .X t X t1 / 2 C tR for each 2 S 1 , Z t has property (9.7).

Now we can complete the proof of Theorem 9.9 by showing that the absence of

arbitrage opportunities in S implies the existence of a measure PQ that belongs to the

class PS and has a bounded density d PQ =dP .

Proof of necessity in Theorem 9:9. Suppose that S does not contain arbitrage opportunities. We are going to construct the desired measure PQ via backward recursion.

First we consider the case t D T . Take a bounded random variable ZT > 0 as

constructed in Lemma 9.15, and define a probability measure PQT by

ZT

d PQT

D

:

dP

E ZT

411

EQ T T .XT XT 1 / j FT 1 0

for all 2 S 1 .

(9.9)

WD EQ T T .XT XT 1 / j FT 1 j 2 S 1 :

For ; Q 2 S 1 , let

A WD EQ T T .XT XT 1 / j FT 1 > EQ T QT .XT XT 1 / j FT 1 ;

and define 0 by t0 D 0 for t < T and

T0 WD T IA C QT IAc :

The predictable convexity of S implies that 0 2 S 1 . Furthermore, we have

EQ T T0 .XT XT 1 / j FT 1

D EQ T T .XT XT 1 / j FT 1 _ EQ T QT .XT XT 1 / j FT 1 :

Hence, the family is directed upwards in the sense of Theorem A.33. By virtue

of that theorem, ess sup is the increasing limit of a sequence in . By monotone

convergence, we get

EQ T ess sup EQ T T .XT XT 1 / j FT 1

2S 1

D sup EQ T EQ T T .XT XT 1 / j FT 1

2S 1

1

D

sup E T .XT XT 1 / ZT

E ZT 2S 1

(9.10)

0;

where we have used (9.7) in the last step. Since S contains 0, it follows that

ess sup EQ T T .XT XT 1 / j FT 1 D 0

PQT -a.s.,

2S 1

Now we apply the previous argument inductively: Suppose we already have a probability measure PQtC1 P with a bounded density d PQtC1 =dP such that

EQ tC1 jXs j < 1

for all s,

412

EQ tC1 k .Xk Xk1 / j Fk1 0 P -a.s. for k t C 1 and 2 S 1 . (9.11)

Then we may apply Lemma 9.15 with P replaced by PQ tC1 , and we get some strictly

positive ZQ t 2 L1 .; F t ; PQtC1 / satisfying (9.7) with PQtC1 in place of P . We now

proceed as in the first step by defining a probability measure PQt PQtC1 P as

ZQ t

d PQt

D

:

EQ tC1 ZQ t

d PQtC1

Then PQt has bounded densities with respect to both PQtC1 and P . In particular,

EQ t jXs j < 1 for all s. Moreover, the F t -measurability of d PQt =d PQtC1 implies

that (9.11) is satisfied for PQt replacing PQtC1 . Repeating the arguments that led to

(9.9) yields

EQ t t .X t X t1 / j F t1 0 for all 2 S 1 .

After T steps, we arrive at the desired measure PQ WD PQ1 2 PS .

9.2

which can be decomposed as

U t D U0 C

t

X

k .Xk Xk1 / B t ;

(9.12)

kD1

adapted and increasing process such that B0 D 0. In the unconstrained case where

S consists of all strategies, we have seen in Section 7.2 that such a decomposition

exists if and only if U is a supermartingale under each equivalent martingale measure

P 2 P . In our present context, a first guess might be that the role of P is now played

by PS . Since each value process of a strategy in S is a local PQ -supermartingale for

each PQ 2 PS , any process U which has a decomposition (9.12) is also a local PQ supermartingale for PQ 2 PS . Thus, one might suspect that the latter property would

also be sufficient for the existence of a decomposition (9.12). This, however, is not

the case, as is illustrated by the following simple example.

Example 9.16. Consider a one-period market model with the riskless bond S00

S10 1 and with one risky asset S 1 . We assume that S01 1 and that S11 takes the

values S11 .! / D 12 and S11 .! C / D 32 on WD ! ; ! C . We choose any measure

P on which assigns positive mass to both ! C and ! . If we let S D 0; 1, then

a measure PQ belongs to PS if and only if PQ ! C 2 .0; 12 . Thus, for any positive

413

initial value U0 , the process defined by U1 .! / WD 0 and U1 .! C / WD 2U0 is a PS supermartingale. If U can be decomposed according to (9.12), then we must be able

to write

2U0 D U1 .! C / D U0 C .S11 .! C / S01 .! C // B1 .! C /

for some B1 .! C / 0. This requirement is equivalent to U0 =2. Hence the

decomposition (9.12) fails for U0 > 1=2.

}

The reason for the failure of the decomposition (9.12) for certain PS -supermartingales is that PS does not reflect the full structure of S; the definition of PS depends

only on the cone

j
> 0; 2 S

generated by S. In the approach we are going to present here, the structure of S will

be reflected by a stochastic process which we associate to any measure Q

P .

Definition 9.17. For a measure Q

P , the upper variation process for S is the

increasing process AQ defined by

Q

A0 WD 0

Q

AQ

tC1 A t WD ess sup tC1 .EQ X tC1 j F t X t /

and

2S

Q

EQ AT < 1

EQ jX tC1 X t j j F t < 1 P -a.s. for all t .

Clearly, the upper variation process of any measure Q P satisfies

Q

AQ

tC1 A t D ess sup tC1 .EQ X tC1 j F t X t /;

2S 1

condition EQ jX tC1 X t j j F t < 1 guarantees that

tC1 .EQ X tC1 j F t X t / D EQ tC1 .X tC1 X t / j F t ;

and it follows that

Q

Q

A tC1 A t D ess sup EQ tC1 .X tC1 X t / j F t for Q 2 QS .

(9.13)

2S 1

In particular, we have

Q

Q

AP

T D 0 for P 2 PS ,

(9.14)

P PS QS :

(9.15)

414

then V AQ is a local Q-supermartingale.

Proof. Let V be the value process of D . 0 ; / 2 S. Denote by n .!/ the first time

t at which

j tC1 .!/j > n or EQ jX tC1 X t j j F t .!/ > n:

If such a t does not exist, let n .!/ WD T . Then n is a stopping time. Since

n

jV tC1

V tn j In t C1 j tC1 j jX tC1 X t j 2 L1 .Q/;

n

EQ V tC1

V tn j F t D In t C1 tC1 .EQ X tC1 j F t X t /

n

.AQ /tC1

.AQ /t n :

Let us identify the class QS in some special cases.

Remark 9.19. If S 1 consists of all bounded predictable processes with non-negative components, then QS D PS . To prove this, take Q 2 QS , and note first that

AQ 0, due to (9.13) and the fact that S is a cone. Thus, value processes of strategies

in S are local Q-supermartingales by Proposition 9.18. By taking 2 S such that

j

ti 1 and t 0 for j i , we get that X i is a local Q-supermartingale, and Proposition 9.6 implies that X i is a Q-supermartingale. In particular, X ti is Q-integrable,

}

and we conclude Q 2 PS .

Remark 9.20. If S 1 consists of all bounded predictable processes , then QS D P .

This follows by combining Remarks 9.8 and 9.19.

}

Example 9.21. Suppose our market model contains just one risky asset, and S consists of all predictable processes such that

at t bt

1 < a t 0 b t < 1

P -a.s.

If we assume in addition that E jX tC1 X t j j F t > 0 P -a.s., then the nonredundance condition (9.1) holds, and S satisfies the assumptions (a) through (d)

stated at the beginning of this chapter. If Q P is any probability measure such that

EQ jX tC1 X t j j F t < 1 P -a.s. for all t ,

(9.16)

415

then

ess sup EQ tC1 .X tC1 X t / j F t

2S 1

<1

P -a.s.

We now state the uniform Doob decomposition under constraints, which is the main

result of this section.

Theorem 9.22. Suppose that PS is non-empty. Then for any adapted process U with

UT 0 P -a.s., the following conditions are equivalent:

(a) U AQ is a Q-supermartingale for every Q 2 QS .

(b) There exists 2 S and an adapted increasing process B such that B0 D 0 and

U t D U0 C

t

X

k .Xk Xk1 / B t

kD1

Q

Mt

WD U0 C

t

X

k .Xk Xk1 / AQ

t

kD1

Q

Q

Q

Lemma 9.12 that the negative part of M t is bounded from below by A t 2 L1 .Q/.

Q

So M is a Q-supermartingale by Proposition 9.6. Since

Q

Q

Q

Q

MT MT BT D UT AT AT

P -a.s.,

Q-supermartingale.

(a) ) (b): We must show that for any given t 2 1; : : : ; T there exist some

2 S and a non-negative random variable R t playing the role of B t B t1 such that

U t U t1 D t .X t X t1 / R t , i.e.,

U t U t1 2 K tS L0C .; F t ; P /;

where

K tS D t .X t X t1 / j 2 S:

416

The formulation of this problem does not change if we switch from P to any equivalent probability measure, so we can assume without loss of generality that P 2 PS .

In this case AP 0, and U is a P -supermartingale. In particular, Us 2 L1 .P / for

all s.

We assume by way of contradiction that

U t U t1 C tS D .K tS L0C .; F t ; P // \ L1 .P /:

(9.17)

Recall that we have proved in Lemma 9.14 that C tS is a closed convex subset of

L1 .; F t ; P /. The HahnBanach separation theorem, Theorem A.57, now implies

the existence of a random variable Z 2 L1 .; F t ; P / such that

WD sup E Z W < E Z .U t U t1 / DW < 1:

(9.18)

W 2C tS

0 . / E Z IZ<0

for every 0, and it follows that Z 0 P -almost surely.

In fact, we can always modify Z such that it is bounded from below by some " > 0

and still satisfies (9.18). To see this, note first that every W 2 C tS is dominated by

some t .X t X t1 / 2 K tS with integrable negative part. Therefore

E W E t .X t X t1 / lim inf E t .X t X t1 /Ij t jc 0;

c"1

where we have used our assumption that P 2 PS . If we let Z " WD " 1 C .1 "/Z,

then Z " still satisfies E Z " W 0 for all W 2 C tS , and for " small enough, the

expectation E Z " .U t U t1 / is still larger than . So Z " also satisfies (9.18).

Therefore, we may assume from now on that our Z with (9.18) is bounded from

below by some constant " > 0.

For the next step, let Z t1 WD E Z j F t1 and

Z

dQ

WD

:

dP

Z t1

Since this density is bounded and since P 2 PS , we get

EQ jXs Xs1 j j Fs1 < 1

(9.19)

EQ s .Xs Xs1 / j Fs1 D E s .Xs Xs1 / j Fs1 0

for s t ; (9.20)

417

see the proof of Theorem 7.5 for details. We now consider the case s D t . As we have

seen in the proof of Theorem 9.9, the family

EQ t .X t X t1 / j F t1 j 2 S

is directed upwards. Therefore, we may conclude as in (9.10) that

EQ Z t1 ess supEQ t .X t X t1 / j F t1

2S

D sup E t .X t X t1 / Z

(9.21)

2S

:

Since Z t1 ", (9.21) implies that

Q

Q

EQ A t A t1 D EQ ess sup EQ t .X t X t1 / j F t1 :

"

2S

Q

By (9.20) we may conclude that EQ AT =", and so (9.19) yields Q 2 QS .

As a final step, we show that U AQ cannot be a Q-supermartingale, thus leading

our assumption (9.17) to a contradiction with our hypothesis (a). To this end, we use

again (9.21)

D EQ Z t1 .U t U t1 /

D E Z .U t U t1 / D

>

EQ Z t1 ess sup EQ t .X t X t1 / j F t1

2S

Q

Q

D EQ Z t1 .A t A t1 / :

Q

EQ U t U t1 j F t1 AQ

t A t1

P -a.s.,

9.3

claim. Our goal is to construct a superhedging strategy for H that belongs to our class

S of admissible strategies. The uniform Doob decomposition suggests that we should

418

Q 2 QS . If we consider only one such Q, then the minimal process U which satisfies

these requirements is given by UQ Q C AQ , where

UQ tQ WD ess sup EQ H AQ

j F t t D 0; : : : ; T;

(9.22)

2T t

is the Snell envelope of H AQ with respect to Q. Thus, one may guess that

Q

ess sup.UQ t C A t /;

t D 0; : : : ; T;

Q2QS

is the minimal process U which dominates H and for which U AQ is a Q-supermartingale for each Q 2 QS . Let us assume that

Q

sup UQ 0 D sup sup EQ H AQ

< 1:

Q2QS

Q2QS 2T

Definition 9.23. The process

"

Q

Q

UQ t WD ess sup.A t C UQ t /

Q2QS

Q

D ess sup A t C ess sup EQ H AQ

j Ft ;

Q2QS

t D 0; : : : ; T;

2T t

The main result of this section confirms our guess that UQ " is the process we are

looking for.

Theorem 9.24. The upper QS -Snell envelope of H is the smallest process U H

such that U AQ is a Q-supermartingale for each Q 2 QS .

For a European claim, we have the following additional result.

Proposition 9.25. For a discounted European claim H E with

sup EQ H E AQ

T < 1;

Q2QS

"

Q

Q

UQ t D ess sup.EQ H E AT j F t C A t /;

Q2QS

t D 0; : : : ; T:

419

Proposition 9.25 will follow from Lemma 9.30 below. The next result provides

a scheme for the recursive calculation of UQ " . It will be used in the proof of Theorem 9.24.

Proposition 9.26. For fixed Q0 2 QS let Q t .Q0 / denote the set of all Q 2 QS

which coincide with Q0 on F t . Then UQ " satisfies the following recursion formula:

Q

0

0

Q"

UQ t" AQ

D .H t AQ

t

t / _ ess sup EQ U tC1 A tC1 j F t ; t D 0; : : : ; T 1:

Q2Q t .Q0 /

The proofs of this proposition and of Theorem 9.24 will be given at the end of this

section. Let us recall the following concepts from Section 6.4. The pasting of two

probability measures Q1 Q2 in a stopping time 2 T D j is a stopping time

T is the probability measure

Q A D EQ1 Q2 A j F ;

Q

A2F:

It was shown in Lemma 6.40 that, for all stopping times and FT -measurable Y 0,

EQQ Y j F D EQ1 EQ2 Y j F_ j F :

(9.23)

Recall also that a set Q of equivalent probability measures on .; F / is called stable

if for any pair Q1 ; Q2 2 Q and all 2 T the corresponding pasting also belongs

to Q. A technical inconvenience arises from the fact that our set QS may not be

stable. We must introduce a further condition on which guarantees that the pasting

of Q1 ; Q2 2 QS in also belongs to QS .

Lemma 9.27. For 2 T , the pasting QQ of Q1 ; Q2 2 QS in satisfies EQQ jX tC1

X t j j F t < 1 P -a.s., and its upper variation process is given by

Q

Q

Q1

Q2

2

C .AQ

A t D A t^

t A / I<t :

Moreover, we have QQ 2 QS under the condition that there exists " > 0 such that

dQ2

"

dQ1 F

a.s. on < T .

EQQ jX tC1 X t j j F t D EQ1 jX tC1 X t j j F t I>t

C EQ2 jX tC1 X t j j F t It ;

and each of the two conditional expectations is finite almost surely.

(9.24)

420

Q

Q As above, (9.23) yields

Now we will compute the upper variation process AQ of Q.

C EQ2 t .X tC1 X t / j F t It :

Taking the essential supremum over t 2 S t1 gives

Q

Q

1

1

2

2

AQ

tC1 A t D .A tC1 A t /I>t C .A tC1 A t /It ;

Q

Q

In a final step, we show that QQ belongs to QS under condition (9.24). We must

Q

Q

show that EQQ AT < 1. Let Z t denote the density of Q2 with respect to Q1 on F t .

Q

Q

Q

Q

Q2

Q2

1

EQQ AT D EQQ AQ

C AT A

Q

Q2

2

1

D EQ1 AQ

C EQ2 AT A j F

h 1

i

Q

Q

2

EQ1 .AT 2 AQ

/

Z

j

F

EQ1 AT 1 C EQ1

T

Z

1

Q

Q

EQ1 AT 1 C EQ2 AT 2 ;

"

Lemma 9.28. Suppose we are given Q1 ; Q2 2 QS , a stopping time 2 T , and a set

B 2 F such that dQ2 =dQ1 jF " a.s. on B. Let QQ be the pasting of Q1 and Q2

in the stopping time

WD IB C T IB c :

Then QQ 2 QS , and the Snell envelopes associated with these three measures by (9.22)

are related as follows:

Q

Q

1

2

Q Q1 C AQ

Q Q2 C AQ

UQ Q C AQ

/ IB c C .U

/ IB

D .U

P -a.s.

(9.25)

Proof. We have dQ2 =dQ1 jF
", hence QQ 2 QS follows from Lemma 9.27. Let

Q

be a stopping time in the set T of all stopping times . The formula for AQ in

Lemma 9.27 yields

Q

Q1

Q1

Q1

Q2

Q2

AQ

D A C .A A /IB c C .A A /IB :

EQQ Y j F D EQ1 Y j F IB c C EQ2 Y j F IB

421

for all random variables Y such that all conditional expectations make sense. Hence,

Q

Q

EQQ H AQ

j F C A

Q1

Q2

Q2

1

D .EQ1 H AQ

j F C A / IB c C .EQ2 H A j F C A / IB :

time in T . Conversely, every 2 T can be written in that way for stopping times

1 and 2 . Thus, taking the essential supremum over all 1 and 2 and applying

Proposition 6.36 yields (9.25).

Q

Q

Qk

1

0

In fact, we have A D AQ

D AQ

in (9.25), as we will have A

in the following

lemma.

Lemma 9.29. For any Q0 2 QS , 2 T , and > 0, there exist a set 2 F such

that Q0 1 and measures Qk 2 QS such that Qk D Q0 on F and

k

Q"

UQ Qk C AQ

% ess sup.UQ Q C AQ

/ D U

P -a.s. on :

Q2QS

Proof. By Theorem A.33 and its proof, there exists a sequence .Qn0 / QS such that

0

Q

Q

lim max.UQ n C A n / D UQ "

k"1 nk

P -a.s.

, Q0 k 1 .1 2k /, and

k1

Q

Q

k

D max.UQ n C A n /

UQ Qk C AQ

nk

P -a.s. on k .

T

By letting WD k k , this will imply the first part of the assertion. We start this

recursion in k D 0 by taking Q0 and 0 WD .

0

implies that there exists some

For Qk given, the equivalence of Qk and QkC1

" > 0 such that the set

dQkC1

" 2 F

D WD

dQk F

satisfies Q0 D 1 2.kC1/ . Thus, kC1

WD k \ D satisfies Q0 kC1

.kC1/

/. We now define a set

1 .1 2

0

QkC1

B WD UQ

0

QkC1

C A

k

> UQ Qk C AQ

\ D;

422

0

and consider the pasting QkC1 of Qk and QkC1

in the stopping time WD IB C

T IB c . By Lemma 9.28, QkC1 2 QS and

0

Q

Q

kC1

k

Q QkC1 CAQ

/ IB

UQ kC1 CA kC1 D .UQ Qk CAQ

/ IB c C.U

0

kC1

k

Q QkC1 CAQ

/

D .UQ Qk CAQ

/ _ .U

0

P -a.s.

P -a.s. on D

Q

Q

D max .UQ n CA n /

P -a.s. on kC1

.

nkC1

Proof of Proposition 9:26. For Q0 2 QS and t 2 t; : : : ; T , Q t .Q0 / denotes the set

of all Q 2 QS which coincide with Q0 on F t . By Lemma 9.29 and by the definition

of UQ Q as the Snell envelope of H AQ ,

"

Q

Q

UQ t A t 0 D ess sup UQ t

Q2Q t .Q0 /

Q

Q

D ess sup ..H t A t / _ EQ UQ tC1 j F t /

Q2Q t .Q0 /

Q

Q

D .H t A t 0 / _ ess sup EQ UQ tC1 j F t :

Q2Q t .Q0 /

"

"

Q

Q

"

Q

UQ t A t 0 .H t A t 0 / _ ess sup EQ UQ tC1 A tC1 j F t :

(9.26)

Q2Q t .Q0 /

For the proof of the converse inequality, let us fix an arbitrary Q 2 Qt .Q0 /. For

any > 0, Lemma 9.29 yields a set 2 F tC1 with measure Q 1 and

Qk

"

% UQ tC1

AQ

Qk 2 Q tC1 .Q/ such that UQ tC1

tC1 P -a.s. on . Since Qk coincides

with Q on F tC1 , we have P -a.s. on

"

Q Qk

Q Qk

EQ UQ tC1

AQ

tC1 j F t D lim EQ U tC1 j F t D lim EQk U tC1 j F t

k"1

k"1

k"1

ess sup

Q

Q2Q

t C1 .Q/

Q

Q

UQ t

Q

Q

ess sup UQ t D UQ t" AQ

t

Q

Q2Q

t .Q/

0

D UQ t" AQ

t :

(9.26).

423

Proof of Theorem 9:24. Since Q0 2 QS is obviously contained in Q t .Q0 /, the recursion formula of Proposition 9.26 yields

"

Q

Q

"

Q0

"

0

j F t EQ0 UQ tC1

AQ

UQ t A t 0 .H t A t 0 / _ EQ0 UQ tC1 A tC1

tC1 j F t ;

"

that UQ " dominates H .

Let U be any process which dominates H and for which U AQ is a Q-supermartingale for each Q 2 QS . For fixed Q, the Q-supermartingale U AQ dominates

H AQ and hence also UQ Q , since UQ Q is the smallest Q-supermartingale dominating

H AQ by Proposition 6.10. It follows that

Q

Q

"

U t ess sup.UQ t C A t / D UQ t

Q2QS

Lemma 9.30. Let H be a discounted American claim whose payoff is zero if it is not

exercised at a given stopping time 2 T , i.e., H t .!/ D 0 if t .!/. Then its

upper QS -Snell envelope is given by

"

Q

Q

UQ t D I t ess sup.EQ H A j F t C A t /;

t D 0; : : : ; T:

Q2QS

Proof. By definition,

"

Q

UQ t D ess sup ess sup.EQ H AQ

j F t C A t /:

2T t

Q2QS

Hence,

0

on < t ,

"

UQ t D

H t on D t .

So we have to show that choosing is optimal on > t . If 2 T t is a

stopping time with P > > 0, then WD ^ is as least as good as , since

each process AQ is increasing. So it remains to exclude the case that there exists a

stopping time 2 T t with on > t and P < > 0, such that

yields a strictly better result than . In this case, there exists some Q1 2 QS such

that

Q1

1

EQ1 H AQ

j Ft C At

Q2QS

(9.27)

424

with strictly positive probability on > t . Take any PQ 2 PS and " > 0, and

define

Q

d P

" :

B" WD

dQ1 F

Now let Q" be the pasting Q1 and PQ in the stopping time IB" C T IB c . According

"

"

Q

to Lemmas 9.27 and 9.28, Q" 2 QS , and its upper variation process satisfies A t D

"

Q

Q

Q

A t 1 and A D A 1 as well as

"

Q

Q

1

A D A 1 IB c C AQ

IB"

"

P -a.s. on > t .

"

Q1

Q

1

EQ1 H AQ

E Q " H A Q

j Ft C At

j Ft C At

"

P -a.s. on B" .

By letting " # 0, the P -measure of B" becomes arbitrarily close to 1, and we arrive at

a contradiction to (9.27).

9.4

"

Q

UQ 0 D sup UQ 0 D sup sup EQ H AQ

< 1:

Q2QS

Q2QS 2T

Our aim in this section is to construct superhedging strategies for H which belong to

our set S of admissible trading strategies. Recall that a superhedging strategy for H

is any self-financing trading strategy whose value process dominates H . If applied

"

with t D 0, the following theorem shows that UQ 0 is the minimal amount for which a

superhedging strategy is available.

Q "t .H / the set of all F t -measurable random variables Ut 0 for which

Denote by U

there exists some 2 S such that

Ut C

u

X

k .Xk Xk1 / Hu

(9.28)

kDtC1

"

Theorem 9.31. The upper QS -Snell envelope UQ t of H is the minimal element of

"

Q .H /. More precisely

U

t

Q " .H /,

(a) UQ t" 2 U

t

"

Q " .H /.

(b) UQ t D ess inf U

t

425

Proof. The uniform Doob decomposition in Section 9.2 combined with Theorem 9.24

yields an increasing adapted process B and some 2 S such that

"

UQ u" D UQ t C

u

X

k .Xk Xk1 / C B t Bu

P -a.s. for u t .

kDtC1

"

Q "t .H / from (a). For the proof of the

As to part (b), we first get UQ t ess inf U

Q "t .H / and choose a predictable process 2 S for

converse inequality, take Ut 2 U

"

which (9.28) holds. We must show that the set B WD UQ t U t satisfies P B D 1.

Let

"

"

UO t WD UQ t ^ U t D UQ t IB C U t IB c :

Then UO t UQ t" , and our claim will follow if we can show that Ut" UO t . Let

denote the predictable process obtained from the uniform Doob decomposition of the

P -supermartingale UQ " , and define

s

if s t ,

Os WD

s IB C s IB c if s > t .

With this choice, O 2 S by predictable convexity, and UO t satisfies (9.28), i.e., UO t 2

Q " .H /. Let

U

t

s

X

"

Ok .Xk Xk1 /:

VOs WD UQ 0 C

kD1

by Propositions 9.18 and 9.6. Hence,

"

Q

UQ t D ess sup ess sup EQ H AQ

C At j F t

Q2QS

2T t

h

i

X

Q

Ok .Xk Xk1 / AQ

C

A

ess sup EQ UO t C

F

t

t

Q;

kDtC1

UO t :

"

Q " .H /.

This proves UQ t ess inf U

t

For European claims, the upper QS -Snell envelope takes the form

"

Q

Q

UQ t D ess sup.EQ H E AT j F t C A t /;

Q2QS

t D 0; : : : ; T:

426

"

Q

UQ 0 D sup .EQ H E EQ AT /

(9.29)

Q2QS

is the smallest initial investment which suffices for superhedging the claim H E . In

fact, the formula above can be regarded as a special case of the representation theorem

for convex risk measures in our financial market model. This will be explained next.

Let us take L1 WD L1 .; F ; P / as the space of all financial positions. A position

Y 2 L1 will be regarded as acceptable if it can be hedged with a strategy in S at no

additional cost. Thus, we introduce the acceptance set

T

t .X t X t1 / 0 P -a.s. :

AS WD Y 2 L1 9 2 S W Y C

tD1

Due to the convexity of S, this set AS is convex, and under the mild condition

infm 2 R j m 2 AS > 1;

(9.30)

S .Y / WD infm 2 R j m C Y 2 AS I

see Section 4.1. Note that condition (9.30) holds in particular if S does not contain

arbitrage opportunities. In this case, we have in fact

S .0/ D infm 2 R j m 2 AS D 0;

i.e., S is normalized. The main results of this chapter can be restated in terms of S :

Corollary 9.32. Under condition (9.3), the following conditions are equivalent:

(a) S is sensitive.

(b) S contains no arbitrage opportunities.

(c) PS ;.

If these equivalent conditions hold, then

Q

Y 2 L1 :

Q2QS

Q

EQ AT if Q 2 QS ,

.Q/ D

C1

otherwise.

(9.31)

427

Proof. That (a) implies (b) is obvious. The equivalence between (b) and (c) was

shown in Theorem 9.9. Since both sides of (9.31) are cash invariant, it suffices to prove

(9.31) for Y 0. But then the representation for S is just a special case of the superQ X ,

hedging duality (9.29). Finally, (9.31) and (c) imply that S .X / supPQ 2PS E

S

and the sensitivity of follows.

Chapter 10

incomplete market model. Instead of controlling the downside risk, we simply aim at

minimizing the quadratic hedging error. We begin with a local version of the minimization problem, which may be viewed as a sequential regression procedure. Its solution involves an orthogonal decomposition of a given contingent claim; this extends

a classical decomposition theorem for martingales known as the KunitaWatanabe decomposition. Often, the value process generated by a locally risk-minimizing strategy

can be described as the martingale of conditional expectations of the given contingent claim for a special choice of an equivalent martingale measure. Such minimal

martingale measures will be studied in Section 10.2. In Section 10.3, we investigate

the connection between local risk minimization and the problem of variance-optimal

hedging where one tries to minimize the global quadratic hedging error. The local and

the global versions coincide if the underlying measure is itself a martingale measure.

10.1

In this section, we no longer restrict our discussion to strategies which are selffinancing. Instead, we admit the possibility that the value of a position is readjusted

at the end of each period by an additional investment in the numraire asset. This

means that, in addition to the initial investment at time t D 0, we allow for a cash

flow throughout the trading periods up to the final time T . In particular, it will now

be possible to replicate any given European claim, simply by matching the difference

between the payoff of the claim and the value generated by the preceding strategy

with a final transfer at time T .

. 0 ; / such that 0 D . t0 / tD0;:::;T is adapted, and such that D . t / tD1;:::;T is a

d -dimensional predictable process. The (discounted) value process V of . 0 ; / is

defined as

V0 WD 00

and

V t WD t0 C t X t

for t 1.

For such a generalized trading strategy . 0 ; /, the gains and losses accumulated up

429

to time t by investing into the risky assets are given by the sum

t

X

k .Xk Xk1 /:

kD1

V0 D

10

C 1 X0

V t D V0 C

and

t

X

k .Xk Xk1 /;

t D 1; : : : ; T;

kD1

investment V0 D 00 D 10 C 1 X0 . In this case, . t0 / tD1;:::;T is a predictable

process. In general, however, the difference

Vt

t

X

k .Xk Xk1 /

kD1

is now non-trivial, and it can be interpreted as the cumulative cost up to time t . This

motivates the following definition.

Definition 10.2. The gains process G of a generalized trading strategy . 0 ; / is given

by

t

X

k .Xk Xk1 /; t D 1; : : : ; T:

G0 WD 0 and G t WD

kD1

. 0 ; /

C t WD V t G t ;

t D 0; : : : ; T;

In this and in the following sections, we will measure the risk of a strategy in terms

of quadratic criteria for the hedging error, based on the objective measure P . Our

aim will be to minimize such criteria within the class of those generalized strategies

. 0 ; / which replicate a given discounted European claim H in the sense that their

value process satisfies

VT D H P -a.s.

The claim H will be fixed for the remainder of this section. As usual we assume

that the -field F0 is trivial, i.e., F0 D ;; . In contrast to the previous sections

of Part II, however, our approach does not exclude a priori the existence of arbitrage

opportunities, even though the interesting cases will be those in which there exist

equivalent martingale measures. Since our approach is based on L2 -techniques, another set of hypotheses is needed.

430

Assumption 10.3. Throughout this section, we assume that the discounted claim H

and the discounted price process X of the risky assets are both square-integrable with

respect to the objective measure P

(a) H 2 L2 .; FT ; P / DW L2 .P /.

(b) X t 2 L2 .; F t ; P I Rd / for all t .

In addition to these assumptions, the quadratic optimality criteria we have in mind

require the following integrability conditions for strategies.

Definition 10.4. An L2 -admissible strategy for H is a generalized trading strategy

. 0 ; / whose value process V satisfies

VT D H P -a.s.

G t 2 L2 .P /

for each t .

We can now introduce the local version of a quadratic criterion for the hedging

error of an L2 -admissible strategy.

Definition 10.5. The local risk process of an L2 -admissible strategy . 0 ; / is the

process

0

2

Rloc

t . ; / WD E .C tC1 C t / j F t ;

t D 0; : : : ; T 1.

An L2 -admissible strategy .O 0 ; /

all t ,

loc 0

O0 O

Rloc

t . ; / R t . ; / P -a.s.

0

for each L2 -admissible strategy . 0 ; / whose value process satisfies V tC1 D OtC1

C

O tC1 X tC1 D VOtC1 .

Remark 10.6. The reason for fixing the value V tC1 D VOtC1 in the preceding definition becomes clear when we try to construct a locally risk-minimizing strategy

O backwards in time. At time T , we want to construct O 0 ; O 0 ; OT 1 ; OT as

.O 0 ; /

T 1 T

a minimizer for the local risk RTloc1 . 0 ; /. Since the terminal value of every L2 admissible strategy must be equal to H , this minimization requires the side condition

T0 C T XT D H D VOT . As we will see in the proof of Theorem 10.9 below, miniO completely determines O 0 and OT and VOT 1 , but one is still free

mality of RTloc1 .O 0 ; /

T

0

to choose OT 1 and OT 1 among all T0 1 ; T 1 with T0 1 C T 1 XT 1 D VOT 1 .

In the next step, it is therefore natural to minimize RTloc2 . 0 ; / under the condition

that VT 1 is equal to the value VOT 1 obtained from the preceding step. Moreover, the

problem will now be of the same type as the previous one.

}

431

turn out that they are self-financing on average in the following sense:

Definition 10.7. An L2 -admissible strategy is called mean self-financing if its cost

process C is a P -martingale, i.e., if

E C tC1 C t j F t D 0

In order to formulate conditions for the existence of a locally risk-minimizing strategy, let us first introduce some notation. The conditional covariance of two random

variables W and Z with respect to P is defined as

cov.W; Z j F t / WD E W Z j F t E W j F t E Z j F t

provided that the conditional expectations and their difference make sense. Similarly,

we define the conditional variance of W under P is

var. W j F t / D E W 2 j F t E W j F t 2

D cov.W; W j F t /I

see also Exercise 5.2.6.

Definition 10.8. Two adapted processes U and Y are called strongly orthogonal with

respect to P if the conditional covariances

cov. U tC1 U t ; Y tC1 Y t j F t /;

t D 0; : : : ; T 1,

When we consider the strong orthogonality of two processes U and Y in the sequel, then usually one of them will be a P -martingale. In this case, their conditional

covariance reduces to

cov. U tC1 U t ; Y tC1 Y t j F t / D E .U tC1 U t /.Y tC1 Y t / j F t :

After these preparations, we are now ready to state our first result, namely the following characterization of locally risk-minimizing strategies.

Theorem 10.9. An L2 -admissible strategy is locally risk-minimizing if and only if it

is mean self-financing and its cost process is strongly orthogonal to X.

Proof. The local risk process of any L2 -admissible strategy . 0 ; / can be expressed

as a sum of two non-negative terms

0

2

R loc

t . ; / D var. C tC1 C t j F t / C E C tC1 C t j F t :

432

Since the conditional variance does not change if we add F t -measurable random variables to its argument, the first term on the right-hand side takes the form

var. C tC1 C t j F t / D var. V tC1 tC1 .X tC1 X t / j F t /:

(10.1)

E C tC1 C t j F t 2 D .E V tC1 j F t tC1 E X tC1 X t j F t V t /2 : (10.2)

In a second step, we fix t and V tC1 , and we consider tC1 and V t as parameters.

0

Our purpose is to derive necessary conditions for the minimality of Rloc

t . ; / with

respect to variations of tC1 and V t . To this end, note first that it is possible to change

the parameters t0 and t in such a way that V t takes any given value, that the modified

strategy is still an L2 -admissible strategy for H , and that the values of tC1 and

V tC1 remain unchanged. In particular, the value in (10.1) is not affected by such a

0

modification, and so it is necessary for the optimality of Rloc

t . ; / that V t minimizes

(10.2). This is the case if and only if

V t D E V tC1 j F t tC1 E X tC1 X t j F t :

(10.3)

The value of (10.1) is independent of V t and a quadratic form in terms of the F t measurable random vector tC1 . Thus, (10.1) is minimal if and only if tC1 solves

the linear equation

0 D cov.V tC1 tC1 . X tC1 X t /; X tC1 X t j F t /:

(10.4)

E C tC1 C t j F t D E V tC1 tC1 .X tC1 X t / j F t V t D 0:

Moreover, given (10.3), the condition (10.4) holds if and only if

E .C tC1 C t /.X tC1 X t / j F t D 0

where we have used the fact that the conditional covariance in (10.4) is not changed by

subtracting the F t -measurable random variable V t from the first argument. Backward

induction on t concludes the proof.

The previous proof provides a recipe for a recursive construction of a locally riskminimizing strategy: If V tC1 is already given, minimize

E .C tC1 C t /2 j F t D E .V tC1 .V t C tC1 .X tC1 X t ///2 j F t

with respect to V t and tC1 . This is just a conditional version of the standard problem

of determining the linear regression of V tC1 on the increment X tC1 X t . Let us now

consider the case

d D 1;

433

where our market model contains just one risky asset. Then the following recursive

scheme yields formally an explicit solution

VOT WD H;

cov. VOtC1 ; X tC1 X t j F t /

OtC1 WD

I
t C1 0 ;

2

tC1

(10.5)

2

Here tC1

is a shorthand notation for the conditional variance

2

WD var. X tC1 X t j F t /:

tC1

O whose

Defining Ot0 WD VOt Ot X t , we obtain a generalized trading strategy .O 0 ; /

O

terminal portfolio value VT coincides with H . However, an extra condition is needed

to conclude that this strategy is indeed L2 -admissible.

Proposition 10.10. Consider a market model with a single risky asset and assume

that there exists a constant C such that

.E X t X t1 j F t1 /2 C t2

(10.6)

O Moreover,

Then the recursion (10.5) defines a locally risk-minimizing strategy .O 0 ; /.

0 ; /

O

O

up to modifications

any other locally risk-minimizing strategy coincides with .

of Ot on the set t2 D 0.

O is L2 -admissible. To this end, observe that the

Proof. We have to show that .O 0 ; /

recursion (10.5) and the condition (10.6) imply that

E .Ot .X t X t1 //2

cov. VOt ; X t X t1 j F t1 /2

2

E .X t X t1 / j F t1 I
2 0

DE

t

t4

cov. VOt ; X t X t1 j F t1 /2

.1 C C / E

t2

.1 C C / E var. VOt j F t1 / :

The last expectation is finite if VOt is square-integrable. In this case, Ot .X t X t1 / 2

O follows by

L2 .P / and in turn VOt1 2 L2 .P /. Hence, L2 -admissibility of .O 0 ; /

0

O

O

backward induction. The claim that . ; / is locally risk-minimizing as well as the

uniqueness assertion follow immediately from the construction.

434

t

X

.E Xs Xs1 j Fs1 /2

;

var. Xs Xs1 j Fs1 /

t D 1; : : : ; T;

sD1

as the assumption of bounded mean-variance trade-off. Intuitively, it states that the

forecast E X t X t1 j F t1 of the price increment X t X t1 is of the same order

}

as the corresponding standard deviation t .

Remark 10.12. The assumption of bounded mean-variance trade-off is equivalent to

the existence of some < 1 such that

.E X t X t1 j F t1 /2 E .X t X t1 /2 j F t1

trade-off is equivalent to

2t C.E .X t X t1 /2 j F t1 2t /;

which is seen to be equivalent to (10.7) by choosing D C =.1 C C /.

Example 10.13. Let us consider a market model consisting of a single risky asset S 1

and a riskless bond

S t0 D .1 C r/t ; t D 0; : : : ; T;

with constant return r > 1. We assume that S01 D 1, and that the returns

R t WD

1

S t1 S t1

;

1

S t1

t D 1; : : : ; T;

of the risky asset are independent and identically distributed random variables in

L2 .P /. Under these assumptions, the discounted price process X , defined by

Xt D

t

Y

1 C Rs

;

1Cr

t D 0; : : : ; T;

sD1

is square-integrable. Denoting by m

Q the mean of R t and by Q 2 its variance, we get

m

Q r

;

1Cr

Q 2

2

j F t1 / D X t1

:

.1 C r/2

var. X t X t1

Thus, the condition of bounded mean-variance trade-off holds without any further assumptions, and a locally risk-minimizing strategy exists. Moreover, P is a martingale

measure if and only if m

Q D r.

}

435

Let us return to our general market model with an arbitrary number of risky assets

X D .X 1 ; : : : ; X d /:

The following result characterizes the existence of locally risk-minimizing strategies

in terms of a decomposition of the claim H .

Corollary 10.14. There exists a locally risk-minimizing strategy if and only if H admits a decomposition

H DcC

T

X

t .X t X t1 / C LT

P -a.s.,

(10.8)

tD1

t .X t X t1 / 2 L2 .P /

for all t ,

O is given

and satisfies L0 D 0. In this case, the locally risk-minimizing strategy .O 0 ; /

by O D and by the adapted process O 0 defined via O 0 D c and

0

Ot0 D c C

t

X

s .Xs Xs1 / C L t t X t ;

t D 1; : : : ; T:

sD1

Moreover, the decomposition (10.8) is unique in the sense that the constant c and the

martingale L are uniquely determined.

O is a given locally risk-minimizing strategy with cost process CO , then

Proof. If .O 0 ; /

O

O

L t WD C t C0 is a square-integrable P -martingale which is strongly orthogonal to

X by Theorem 10.9. Hence, we obtain a decomposition (10.8). Conversely, if such a

O has the cost process CO D c C L, and

decomposition exists, then the strategy .O 0 ; /

O is locally risk-minimizing.

Theorem 10.9 implies that .O 0 ; /

To show that L is uniquely determined, suppose that there exists another decompoQ and L.

Q Then

sition of H in terms of c,

Q ,

Qt D

N t WD c cQ C L t L

t

X

sD1

can be represented as a stochastic integral with respect to X . Strong orthogonality

means that

0 D E .Qt t / .X t X t1 /.X t X t1 / j F t1 :

436

0 D E ..Qt t / .X t X t1 //2 j F t1 ;

and so N t N t1 D .Qt t / .X t X t1 / D 0 P -almost surely. In view of

Q D L and in turn cQ D c.

Q 0 D L0 D 0, we thus get L

L

A decomposition of the form (10.8) will be called the orthogonal decomposition of

the contingent claim H with respect to the process X. If X is itself a P -martingale,

then the orthogonal decomposition reduces to the KunitaWatanabe decomposition,

which we will explain next. To this end, we will need some preparation.

Lemma 10.15. For two square-integrable martingales M and N , the following two

conditions are equivalent:

(a) M and N are strongly orthogonal.

(b) The product MN is a martingale.

Proof. The martingale property of M and N gives

E .M tC1 M t /.N tC1 N t / j F t D E M tC1 N tC1 j F t M t N t ;

and this expression vanishes if and only if MN is a martingale.

Let H 2 denote the space of all square-integrable P -martingales. Via the identity

M t D E MT j F t , each M 2 H 2 can be identified with its terminal value MT 2

L2 .P /. With the standard identification of random variables which coincide P -a.s.,

H 2 becomes a Hilbert space isomorphic to L2 .P /, if endowed with the inner product

.M; N /H 2 WD E MT NT ;

M; N 2 H 2 :

Recall from Definition 6.14 that, for a stopping time , the stopped process M is

defined as

M t WD M^t ; t D 0; : : : ; T:

Definition 10.16. A subspace S of H 2 is called stable if M 2 S for each M 2 S

and every stopping time .

Proposition 10.17. For a stable subspace S of H 2 and for L 2 H 2 with L0 D 0,

the following conditions are equivalent:

(a) L is orthogonal to S, i.e.,

.L; M /H 2 D 0

for all M 2 S.

437

E .L tC1 L t /.M tC1 M t / j F t D 0 P -a.s. for all t .

(c) The product LM is a martingale for each M 2 S.

Proof. The equivalence of (b) and (c) follows from Lemma 10.15. To prove (a) , (c),

we will show that LM is a martingale for fixed M 2 S if and only if .L; M /H 2 D 0

for all stopping times T . By the stopping theorem in the form of Proposition 6.36,

.L; M /H 2 D E LT M D EL M :

Using the fact that L0 M0 D 0 and applying the stopping theorem in the form of

Theorem 6.15, we conclude that .L; M /H 2 D 0 for all stopping times T if and

only if LM is a martingale.

After these preparations, we can now state the existence