Introduction to Stochastic Calculus Applied to Finance Damien Lamberton L’Université arne la Vallée France and Bernard Lapeyre L'Ecole Nationale des Ponts et Chaussées , France Translated by Nicolas Rabeau
Centre for Quantitative Finance Imperial College, London and Merrill Lynch Int Ltd., London and ' Francois Mantion Centre for Quantitative Finance Imperial College London
CHAPMAN & HALL/CRC
Trang 34223 > [U2 math ^L9 O5 TM › K2 U G Bess L 56/3 | ‘FD, | | Contents Library of Congress Cataloging-in-Publication Data Catalog record is available from the Library of Congress Introduction vii Options vii
This book contains information obtained from authentic and highly regarded sources Reprinted material Arbitrage and put/call parity Vill ~ is quoted with permission, and sources are indicated A wide variety of references are listed Reasonable Black-Scholes model and its extensions ix efforts have been made to publish reliable data and information, but the author and the publisher cannot Contents of the book x assume responsibility for the validity of all materials or for the consequences of their use Acknowled gements
, X
Apart from any fair dealing for the purpose of research or private study, or criticism or review, as permitted ‘
under the UK Copyright Designs and Patents Act, 1988, this publication may not be reproduced, stored “ | 1 Discrete-time models 1 or transmitted, in any form or by any means, electronic or mechanical, including photocopying, micro- , : sa 1.1 Discrete-time formalism : - 1 filming, and recording, or by any information storage or retrieval system, without the prior permission : "
in writing of the publishers, or in the case of reprographic reproduction only in accordance with the “ ’ 1.2 Martingales and arbitrage opportunities 4
terms of the licenses issued by the Copyright Licensing Agency in the UK, or in accordance with the 1.3 Complete markets and option pricing 8
terms of the license issued by the appropriate Reproduction Rights Organization outside the UK 1.4 Problem: Cox, Ross and Rubinstein model 12 The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for - ⁄ ,
creating new works, or for resale Specific permission must be obtained in writing from CRC Press LLC 2 Optimal stopping problem and American options 17
for such copying 2.1 Stopping time 17
Direct all inquiries to CRC Press LLC, 2000 N.W Corporate Blvd., Boca Raton, Florida 33431 2.2 The Snell envelope 18
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are 2.3 Decomposition of supermar tingales 21
used only for identification and explanation, without intent to infringe 2.4 Snell envelope and Markov chains 22 2.5 Application to American options
Visit the CRC Press Web site at www.crepress.com 2.6 Exercises Ppa P " 23 25
First edition 1996 oo
First CRC reprint 2000 ik 3 Brownian motion and stochastic differential equations 29 ? © 1996 by Chapman & Hall 3.1 General comments on continuous-time processes 29
Ồ 3.2 Brownian motion 31
No claim to original U.S Government works : :
International Standard Book Number 0-412-71800-6 I 3.3 Continuous-time martingales 32
Printed in the United States of America 234567890 3.4 Stochastic integral and Ité calculus 35
3.5 Stochastic differential equations 49 Printed on acid-free paper
Trang 4vi
4 The Black-Scholes model
4.1 Description of the model :
4.2 Change of probability Representation of martingales 4.3 Pricing and hedging options in the Black-Scholes model 4.4 American options in the Black-Scholes model
4.5 Exercises
5 Option pricing and partial differential equations 5.1 European option pricing and diffusions 5.2 Solving parabolic equations numerically 5.3 American options 5.4 Exercises 6 Interest rate models 6.1 Modelling principles 6.2 Some classical models 6.3 Exercises 7 Asset models with jumps 7.1 Poisson process
7.2 Dynamics of the risky asset 7.3 Pricing and hedging options 7.4 Exercises
8 Simulation and algorithms for financial models 8.1 Simulation and financial models
8.2 Some useful algorithms 8.3 Exercises
Appendix
A.1 Normal random variables A.2 Conditional expectation A.3 Separation of convex sets References Index Contents 63 63 65 67 72 77 95 95 103 110 118 121 121 127 136 141 141 143 150 159 161 161 168 170 173 173, 174 178 179 183 Introduction
The objective of this book is to give an introduction to the probabilistic techniques
required to understand the most widely used financial models In the last few
years, financial quantitative analysts have used more sophisticated mathematical
concepts, such as martingales or stochastic integration, in order to describe the
behaviour of markets or to derive computing methods,
In fact, the appearance of probability theory in financial modelling is not recent At the beginning of this century, Bachelier (1900), in trying to build up a ‘Theory of Speculation’, discovered what is now called Brownian motion From 1973, the publications by Black and Scholes (1973) and Merton (1973) on option pricing and hedging gave a new dimension to the use of probability theory in finance Since then, as the option markets have evolved, Black-Scholes and Merton results have developed to become clearer, more general and mathematically more rigorous The theory seems to be advanced enough to attempt to make it accessible to students
Options
Our presentation concentrates on options, because they have been the main motiva- tion in the construction of the theory and still are the most spectacular example of the relevance of applying stochastic calculus to finance An option gives its holder the right, but not the obligation, to buy or sell a certain amount of a financial asset, by a certain date, for a certain strike price
The writer of the option needs to specify:
e the type of option: the option to buy is called a call while the option to sell is a
put, ,
e the underlying asset: typically, it can be a stock, a bond, a currency and so on;,
Trang 5vill Introduction e the amount of an underlying asset to be purchased or sold;
e the expiration date: if the option can be exercised at any time before maturity, it is called an American option but, if it can only be exercised at maturity, it is
called a European option;
e the exercise price which is the price at which the transaction is done if the option is exercised
The price of the option is the premium When the option is traded on an organised ‘market, the premium is quoted by the market Otherwise, the problem is to price the option Also, even if the option is traded on an.organised market, it can be interesting to detect some possible abnormalities in the market
Let us examine the case ofa European call option on a stock, whose price at time ¢ is denoted by S; Let us call T the expiration date and K the exercise
price Obviously, if K is greater than S, the holder of the option has no interest
whatsoever in exercising the option But, if Sy; > K, the holder makes a profit of Sr — K by exercising the option, i.e buying the stock for A’ and selling it back on the market at Sr Therefore, the value of the call at maturity is given by
(Sp — K)4 = max (Sp — K,0)
If the option is exercised, the writer must be able to deliver a stock at price K
It means that he or she must generate an amount (S7 — A), at maturity At the
time of writing the option, which will be considered as the origin of time, Sy is
unknown and therefore two questions have to be asked: , 1 How much should the buyer pay for the option? In other words, how should we
price at time t = 0 an asset worth (Sy — A); at time T’? That is the problem
of pricing the option
2 How should the writer, who earns the premium initially, generate an amount
(Sp — K)x at time T? That is the problem of hedging the option
Arbitrage and put/call parity
We can only answer the two previous questions if we make a few necessary assumptions The basic one, which is commonly accepted in every model, is the absence of arbitrage opportunity in liquid financial markets, i.e there is no riskless profit available in the market We will translate that.into mathematical terms in the first chapter At this point, we will only show how we can derive formulae relating European put and call prices Both the put and the call which have maturity T and exercise price K are contingent on the same underlying asset which is worth S; at
time ¢ We shall assume that it is possible to borrow or invest money at a constant rate r
Let us denote by C; and P, respectively the prices of the call and the put at time
t Because of the absence of arbitrage opportunity, the following equation called
Introduction ix
puvcall parity is true for all ¢ < T
C; — Đ = St - Kerrữ-9,
To understand the notion of arbitrage, let us show how we could make a riskless profit if, for instance,
CO.-P> Si- Ke t(T-*), ;
At time t, we purchase a share of stock and a put, and sell a call The net value of the operation is :
€ị — Pp - St
If this amount is positive, we invest it at rate r until time T’, whereas if it is negative
we borrow it at the same rate At time T, two outcomes are possible:
e Sr > K: the call is exercised, we deliver the stock, receive the amount K and
clear the cash account to end up with a wealth K + e"(7-*)(C, — P, — S;) > 0
e Sr < K: we exercise the put and clear our bank account as before to finish
with the wealth K + e'Œ~Ð9(Œ; ~ P, — S,) > 0
In both cases, we locked in a positive profit without making any initial endowment: this is an example of an arbitrage strategy
There are many similar examples in the book by Cox and Rubinstein (1985) We will not review all these formulae, but we shall characterise mathematically the notion of a financial market without arbitrage opportunity
Black-Scholes model and its extensions
Even though no-arbitrage arguments lead to many interesting equations, they are not sufficient in themselves for deriving pricing formulae To achieve this, we need to model stock prices more precisely Black and Scholes were the first to suggest a model whereby we can derive an explicit price for a European call on a’ stock that pays no dividend According to their model, the writer of the option can
hedge himself perfectly, and actually the call premium is the amount of money needed at time 0 to replicate exactly the payoff (Sr — K)4 by following their
dynamic hedging strategy until maturity Moreover, the formula depends on only one non-directly observable parameter, the so-called volatility
It is by expressing the profit and loss resulting from a certain trading strategy as a stochastic integral that we can use stochastic calculus and, particularly, Ité
formula, to obtain closed form results In the last few years, many extensions of
Trang 6x Introduction Contents of the book
The ‘first two chapters are devoted to the study of discrete time models The link between the mathematical concept of martingale and the economic notion of arbitrage is brought to light Also, the definition of complete markets and
the pricing of options in these markets are given We have decided to adopt the
formalism of Harrison and Pliska (1981) and most of their results are stated in the
first chapter, taking the Cox, Ross and Rubinstein model as an example
The second chapter deals with American options Thanks to the theory of optimal stopping in a discrete time set-up, which uses quite elementary methods, we introduce the reader to all the ideas that will be developed in continuous time
in subsequent chapters
Chapter 3 is an introduction to the main results in stochastic calculus that we will
use in Chapter 4 to study the Black-Scholes model As far as European options are
concerned, this model leads to explicit formulae But, in order to analyse American
options or to perform computations within more sophisticated models, we need
numerical methods based on the connection between option pricing and partial differential equations These questions are addressed in Chapter 5
Chapter 6 is a relatively quick introduction to the main interest rate models and Chapter 7 looks at the problems of option pricing and hedging when the price of the underlying asset follows a simple jump process
In these latter cases, perfect hedging is no longer possible and we must define a criterion to achieve optimal hedging: These models are rather less optimistic than the Black-Scholes model and seem to be closer to reality However, their mathematical treatment is still a matter of research, in the framework of so-called incomplete markets
Finally, in order to help the student to gain a practical understanding, we have included a chapter dealing with the simulation of financial models and the use of computers in the pricing and hedging of options Also, a few exercises and longer questions are listed at the end of each chapter
This book is only an introduction to a field that has already benefited from considerable research Bibliographical notes are given in some chapters to help the reader to find complementary information We would also like to warn the reader that some important questions in financial mathematics are not tackled Amongst them are the problems of optimisation and the questions of equilibrium for which the reader might like to consult the book by D Duffie (1988).-
A good level in probability theory is assumed to read this book: The reader is referred to Dudley (1989) and Williams (1991) for prerequisites However, some
basic results are also proved in the Appendix l
Acknowledgements
This book is based on the lecture notes taught at /’Ecole Nationale des Ponts
et Chaussées since 1988 The-organisation of this lecture series would not have
Introduction xi
been possible without the encouragement of N Bouleau Thanks to his dynamism, CERMA (Applied Mathematics Institute of ENPC) started working on financial modelling as early as 1987, sponsored by Banque Indosuez and subsequently by Banque Internationale de Placement ‘
Since then, we have benefited from many stimulating discussions with G Pages and other academics at CERMA, particularly O Chateau and G Caplain A few
people kindly read the earlier draft of our book and helped us with their remarks
Amongst them are S Cohen, O Faure, C Philoche, M Picqué and X Zhang Finally, we thank our colleagues at the university and at INRIA for their advice
Trang 7
Discrete-time models
The objective of this chapter is to present the main ideas related to option theory within the very simple mathematical framework of discrete-time models Essen- tially, we are exposing the first part of the paper by Harrison and Pliska (1981) Cox, Ross and Rubinstein’s model is detailed at the end of the chapter in the form of a problem with its solution
1.1 Discrete-time formalism 1.1.1 Assets
A discrete-time financial model is built on a finite probability space (0, F, P) equipped with a filtration, i.e an increasing sequence of o-algebras included in F: Fo,F1, , nN Fn can be seen as the information available at time n and is sometimes called the o-algebra of events up to time n The horizon N will often correspond to the maturity of the options From now on, we will assume
that Fo = {0,2}, Fy = F = P(Q) and Ww € N, P({w}) > 0 The market
consists in (d + 1) fifancial assets, whose prices at time n are given by the non- negative random variables S), S),, ,S4, measurable with respect to F, (investors know, past and present prices but ‘obviously not the future ones) The vector S, = (S°,S1, , S42) is the vector of prices at time n The asset indexed by 0 is the riskless asset and we have S° = 1 If the return of the riskless asset
over one period is constant and equal to r, we will obtain Sẽ = (1+r)” "The
coefficient 8, = 1/S° is interpreted as the discount factor (from time n to time 0): if an amount 8, is invested in_the riskless asset at time 0, then one dollar will
be available at time n The assets indexed by 1 = 1 d are called risky assets
1.1.2 Strategies
Trang 82 Discrete-time models
shares of asset 7 held in the portfolio at time n @ is predictable, i.e độ 1s Zo-measurable
Vi € {0,1, , đ}
and, form > 1: ở; ¡s *a_¡-measurable
This assumption means that the positions in the portfolio at time ø (Ó2., ó5, , 64)
are decided with respect to the information available at time (n — 1) and kept until time n when new quotations are available
The value of the portfolio at time n is the scalar product
d Va(6) = dn.Sa = À dan : —
Its discounted value is : Valo) = Ba (¢n-Sn) = n-Sn, with-đ„ = 1/5 and Sy = (1,BnS}, ,BnS4) is the vector of discounted prices " A strategy is called self-financing if the following equation is satisfied for all neé {0,1, ,N -—1} on-Sn = On41-Sn -
The interpretation is the following: at time n, once the new prices S0, „S2, are
quoted, the investor readjusts his positions from ¢n to @n+1 without bringing or consuming any wealth
Remark 1.1.1 The equality on-Sn = $n41-Sn 1s obviously equivalent to bnti-(Sn41'— Sn) = bngi-Sn4i — bn-Sn,
or to
Vn41(¢) — Val) = n+i-(Sn+1 —.Ốn)
At time n + 1, the portfolio is worth dng1-Sn41 and Ony1-Sn41 — On41-Sn is
the net gain caused by the price changes between times n and n + 17 Hence, the profit or loss realised by following a self-financing strategy is only due to the price
_ moves
The following proposition makes this clear in terms of discounted prices Proposition 1.1.2 The following are equivalent
(i) The strategy ¢ is self-financing (ii) Foranyne {1, ,N}, Va($) = Vol?) + 52 9; - AS;, j=l ‘ where AS, is the vector S; — Sj-1 yee Discrete-time formalism 3 đi) For anyn € {1, ,N}, a Vn() = Vo(d) + 3 9; - AŠ;, j=l
where AS; is the vector S; ~ §;-1 = B;S; — Bj-1S;~1
Proof The equivalence between (i) and (ii) results from Remark 1.1.1 The equivalence between (i) and (iii) follows from the fact that @n.S, = dn41-Sn if and only if @n-Sp = On41-Sn Oo
This proposition shows that, if an investor follows a self-financing strategy, the
discounted value of his portfolio, hence its value, is completely defined by the
initial wealth and the strategy ($),, ., 64) pen< n (this is only justified because AS? = 0) More precisely, we can prove the following proposition
Proposition 1.1.3 For any predictable process ((bn -164) gene and for Tt? `
any Fo-measurable variable Vo, there exists a unique predictable process ($9) <„›<y such that the strategy o= (9, gi, , j2) is self-financing and its initial value is Vo ‘ ‘ : Proof The self-financing condition implies Vn (b) = Ort nSn to + bnSe = Wt > (GAS + + 4;A57) j=l
which defines $y, We just have to check that ớP is predictable, but this is obvious
if we consider’ the equation
d1 =W+S (9A8) + +-4/45/)+(sk (S§t j=1 ¡) 4-468 (-SL,))
QO
1.1.3 Admissible strategies and arbitrage
We did not make any assumption on the sign of the quantities ¢i, If 6° <0, we
have borrowed the amount |#9| in the riskless asset If 61, < 0 for ¿ > 1, we say
that we are short a number ¢}, of asset i Short-selling and borrowing is allowed but the value of our portfolio must be positive at all times
Definition 1.1.4 A'strategy ¢ is admissible if it is self-financing and if V,,(¢) > 0
Trang 94 Discrete-time models The investor-must be able to pay back his debts (in riskless or risky asset) at any time The notion of arbitrage (possibility of riskless profit) can be formalised as follows:
Definition 1.1.5 An arbitrage strategy is an admissible strategy with zero initial value and non-zero final value
Most models exclude any arbitrage opportunity and the objective of the next section is to characterise these models with the notion of martingale
1.2 Martingales and arbitrage opportunities
In order to analyse the connections between martingales and arbitrage, we must
first define a martingale on a finite probability space The conditional expectation plays a central role in this definition and the reader can refer to the Appendix for a quick review of its properties ‘
1.2.1 Martingales and martingale transforms
In this section, we consider a finite probability space (Q, F, P), with F = P(Q) and Ww € 2, P({w}) > 0, equipped with a filtration (F,)o<n<n (without necessarily assuming that Fy = F, nor Fy = {0, 2}) A sequence (X;,)o<n<n
of random variables is adapted to the filtration if for any n, Xp, is 7„-measurable
Definition 1.2.1 An adapted sequence (Mn)o<n< N Of real random variables is: © amartingale ifE (Mn+1|Fn) = Mn foralln < N -1;
© asupermartingale ifE(MnzilFn) < Mn forall n < N — 1; ° a submartingale ifE(Mysi|Fn) > Mn foralln < N — 1
These definitions can be extended to the multidimensional case: for instance, a sequence (M,,)o<n<wn of IR?-valued random variables is a martingale if each component is.a real-valued.martingale
In a financial context, saying that the price (S4)o<n<w of the asset 2 is a
martingale implies that, at each time n, the best estimate (in the least-square sense) of S7 „¡ is given by Sto
The following properties are easily derived from the previous definition and stand as a good exercise to get used to the concept of conditional expectation
1 (M,)o<n<n is a martingale if and only if
E(Ma+;|#a) = Mạ Vj >0
2 If (Mn noo i is a martingale, thus for any n: E (M,,) = E (Mo)
3 The sum of two martingales is a martingale
4 Obviously, similar properties can be shown for supermartingales and submartin- gales
wine
Martingales and arbitrage opportunities 5 Definition 1.2.2 An adapted sequence (Hn)o<n<n of random variables is pre- dictable if, for alln > 1, Hy is Fax 1 measurable
Proposition 1.2:3 Let (Mn)g<,<y be a martingale and (Hn)geney 4 pre- dictable sequence with respect to the filtration (Fn)yc,cy Denote AM, = Mr, — Mn-1 The sequence (Xn)ocn<n defined by ` -
Xo = HoMo „
Xn = HoMo+ MAM, + - + H,AM,, forn>1
is a martingale with respect to (#a)s<a< N-
(Xa) is sometimes called the martingale transform of (M„) by (Hạ): A conse-
quence of this proposition and Proposition 1.1.2 is-that if the discounted prices of
the assets are martingales, the expected value of the wealth generated by following
a self-financing strategy is equal to the initial wealth
Proof Clearly, (X,,) is an adapted sequence Moreover, for n > 0 E (Xn4i — Xa|Zn) = E (Hasi(Mnta — Ma)|Zn) = HyiitE(Ma+1 — Mn|Fn) since Hạ¿¡ is F,-measurable = 0 Hence oe A E(X s22 ) = E(XnlFn) = Xn
That shows that (X,,) isa martingale , có n
The following proposition is a very useful characterisation of martingales
Proposition 1.2.4 An adapted sequence of real random variables (Mạ) is amar-
_ tingale if and only if for any predictable sequence (Hn), we have
E (>: Hoan.) =0
Proof If (M,) is a martingale, the sequence (X,) defined by Xo = 0 and, forn > 1, X, = ye 1 HnAM,, for any predictable process (H,) is also a martingale, by Proposition 1.2.3 Hence, E(X yy) = E(Xo) = 0 Conversely, we notice that if 7 € {1, ,N}, we can associate the sequence (H,,) defined by Hạ = 0forn # 7 + 1 and Hj41 = 14; for any F;-measurable A Clearly, (Ha)
1s predictable and E (or, HẠAM, n) = = 0 becomes ˆ
E (14 (Mjsi1 — M;)) = 0
Trang 106 Discrete-time models
1.2.2 Viable financial markets
Let us get back to the discrete-time models introduced in the first section
Definition 1.2.5 The market is viable if there is no arbitrage opportunity eee
Lemma 1.2.6 If the market is viable, any predictable process Ce rests $°) satis-
fies , -
‹Gn(9) £ I
Proof Let us assume that Gv (¢) € I First, if Gr(¢) > Oforalln € {0, ,N}
the market is obviously not viable Second, if the G,,(@) are not all non-negative,
we define n = sup {kIP (¿,() < 0) > 0} It follows from the definition of n that _ ; n<N-1, P(Galg) <0) >Oand¥m>n Gmn(d) 20 We can now introduce a new process w 0 ifj<n v Ú; (0) = 1A(0)ó;(0) ifj>n
where A is the event {Gn(¢) < 0} Because ¢ is predictable and A is 7a-
measurable, w is also predictable Moreover
- 0 ifj<n
G;0) =| 1s (Gj ()-Ga()) iff >n
thus, G; () > 0 for all j € {0, ,N} and Gn (W) > 0 on A That contradicts
the assumption of market viability and completes the proof of the lemma o
Theorem 1.2.7 The market is viable if and only if there _exists-a-probability |
: *
measure P* equivalent | to P such that the discounted prices of assets are P*-
martingales
Proof (a) Let us assume that there exists a probability P* equivalent to P under which discounted prices are martingales Then, for any self-financing strategy
(bn), (1.1.2) implies
Ủa(ó) = Vo(d) + >, )-A5;
- j=l
Thus by Proposition 1.2.3, (Va (¢)) is a P*- martingale Therefore Vw (¢) and
Vo (#) have the same expectation under P*: -
E* (Vw (¢)) = E* (Yo(#))
ili i if ly if for any event † robability measures P; and P2 are equivalent if and only 1
ee (A) " é 2 Po (A) = 0 Here, P* equivalent to P means that, for any w € 9,
P* ({w}) > 0
Martingales and arbitrage opportunities 7 If the strategy is admissible and its initial value is zero, then E* (Yw(¢)) = 0, with Vy (¢) >-0 Hence Vy (¢) = 0 since P* ({w}) > 0, for allw € 2
(b) The proof of the converse implication is more tricky Let us call I’ the convex cone of strictly positive random variables The market is viable if and only if for
any admissible strategy ¢: Vo (¢) = 0 > Vy (¢) ¢ I
(b1) To any admissible process ($1,, , 62) we associate the process defined by
n
Ớ, (6) = 3) (0)A5} + + đ/A 8),
J=1
That is the cumulative discounted gain realised by following the self-financing strategy $,,, , $4 According to Proposition 1.1.3, there exists a (unique) process (25) such that the strategy ((Ø), #4, .,4)) is self-financing with zero initial
value Gn (¢) is the discounted value of this strategy at time n and because the market is viable, the fact that this value is positive at any time, i.e Gn(#) > 0 for
n=1, ,N, implies that Gy(¢) = 0 The following lemma shows that even if
we do not assume thạt Ớ„ (ó) are non-negative, we still have Gu(¢) ¢ I (b2) The set V of random variables Gn (¢), with ¢ predictable process in IR%, is
clearly a vector subspace of IR® (where IR® is the set of real random variables
defined on 92) According to Lemma 1.2.6, the subspace V does not intersect I Therefore it does not intersect the convex compact set K = {X €T| 3)„ X(œ) = 1} which is included in I’ As a result of the convex sets separation theorem (see
the Appendix), there exists (A (w)) cq sich that: WX EK, Ÿ`A(@)X(0) >0 2 For any predictable 6 DL rwW)Ew (4) (w) = 0 w From Property 1: we deduce that A(w) > 0 for all w € 9, so that the probability P*definedby ss, Aw) P* ({w}) = =—>_ >a ) wen A(w") is equivalent to P
Moreover, if we denote by E* the expectation under measure P*, Property 2
means that, for any predictable process (¢,) in R¢,
N
E* {5° 4,48; | =0
Trang 118 Discrete-time models It follows that for alli € {1, ,d} and any predictable sequence (¢?,) in IR, we have N E't |5 ø9jA5;] =0 j=l Therefore, according to Proposition 1.2.4, we conclude that the discounted prices ($1), ,(S4) are P* martingales n 1.3 Complete markets and option pricing 1.3.1 Complete markets’
We shall define a European option* of maturity N by giving its payoff h > 0, Fy-measurable For instance, a call on the underlying S! with strike price K will be defined by setting: h = (S} — K) , A put on the same underlying asset with the same strike price K will be defined by h = (K — Sy) ,- In those two examples, which are actually the two most important in practice, h is a function of Sy only There are some options dependent on the whole path of the underlying asset, i.e h is a function of So, S1» -, Sn That is the case of the so-called Asian options where the strike price is equal to the average of the stock prices observed during a certain period of time before maturity -
Definition 1.3.1 The contingent claim defined by h is attainable if there exists an
admissible strategy worth h at time N
Remark 1.3.2 In a viable financial market, we just need to find a self-financing strategy worth A at maturity to say that h is attainable Indeed, if ¢ is a self- financing strategy and if P* is a probability measure equivalent to P under which
discounted prices are martingales, then (V.(#)) is also a P*-martingale, being a martingale transform Hence, for n € {0, .,N} Va(¢) = E* (x@)1#:): Clearly, if (ở) > 0 ứn particular if Vw (ó) = A), the strategy ¢ is admissible
‘ Definition 1.3.3 The market is complete if every contingent claim is attainable ’ To assume that a financial market is complete is a rather restrictive assumption that
does not have such a clear economic justification as the no-arbitrage assumption The interest of complete markets is that it allows us to derive a simple theory of contingent claim pricing and hedging The Cox-Ross-Rubinstein model, that we shall study in the next section, is a very simple example of complete market modelling The following theorem gives a precise characterisation of complete, viable financial markets
* Or more generally a contingent claim
Complete markets and option pricing 9 Theorem 1.3.4 A viable market is complete if and only if there exists a unique probability measure P* equivalent to P under which discounted prices are mar- tingales
The probability P* will appear to be the computing tool whereby we can derive closed-form pricing formulae and hedging strategies
Proof (a) Let us assume that the market is viable and complete Then any
non-negative, Fjy-measurable random variable h can be written as h = Vy (¢)
where ¢ is an admissible strategy that replicates the contingent claim h Since ¢
is self-financing, we know that N 50 = VN (6) = Vo (¢) + 3 0/.A5; =1 ‘ Thus, if P; and P2 are two probability measures under which discounted prices arti hag arem ingales, (Vn ()) o<nen 84 martingale under both P, and Py It follows that,forÂ=lori=2 ~Đ 7 E; (w (6)) = E¿ (Wo (9)) = Wo (9), the last equality coming from the fact that Fo = {9, 0} Therefore A h E, (— )=E,{— (s ) óc & ) and, si hi i =
oaual nee | is arbitrary, P; = P»2 on the whole o-algebra Fy assumed to be
(b) Let us assume that the market is viable and incomplete Then, there exists a random variable h > 0 which is not attainable We call V
_ _ en call V the set of random N Ut >> bn-ASn, (1.1) n=1 where Up is Fo-measurable and ((¢), -,4%))gcncy is an IR”-valued pre- dictable process
It follows from Proposition 1.1.3 and Remark 1.3.2 that the variable h /S%, does
not belong to V Hence, V is a strict subset of the set of all random variables on
(2, F )- Therefore, if P* is a probability equivalent to P under which discounted
Prices are martingales and if we define the following scalar product on the set
of random variables (X,Y) + E* (XY), we notice that there exists a non-zero
random variable X orthogonal to V We also write
** Ww _ Xứ) *
Trang 1210 Discrete-time models
with ||_X |]oo = SUP eg |X (w)| Because E* (X) = 0, that defines a new proba-
bility measure equivalent to P and different from P* Moreover N
E** (> áo a6) =0
n=1
for any predictable process ((¢4, ,4%)) ycney- It follows from Proposition
1.2.4 that (Sn)o<new is a P**-martingale © Oo
1.3.2 Pricing and hedging contingent claims in complete markets
The market is assumed to be viable and complete and we denote by P* the unique
probability measure under which the discounted prices of financial assets are
martingales Let h be an Fyy-measurable, non-negative random variable and @ be an admissible strategy replicating the contingent claim hence defined, i.e Vn (¢) = A The sequence (¥.) is a P*-martingale, and consequently 0<n<N We(9) = E" (Pw(9)) › ứ that is Vo(¢) = E* (h/S3,) and more generally Va(ó) = SE" (= Fn) , n=0,1, ,N ˆ N
At any time, the value of an admissible strategy replicating his completely deter- mined by h It seems quite natural to call V,,(¢) the price of the option: that is the wealth needed at time ? to replicate A at time N by following the strategy ¢ If, at time 0, an investor sells the option for ' ‘
h » B(x): E* | =>
he can follow a replicating strategy ¢ in order to generate an amount hat time N In other words, the investor is perfectly hedged
Remark 1.3.5 It is important to notice that the computation of the option price only requires the knowledge of P* and not P We could have just considered a
measurable space (2, F) equipped with the filtration (F,,) In other words, we
would only define the set of all possible states and the evolution of the information over time As soon as the probability space and the filtration are specified, we do not need to find the true probability of the possible events (say, by statistical means) in order to price the option The analysis of the Cox-Ross-Rubinstein
Complete markets and option pricing 11 model will show how we can compute the option price and the hedging strategy in practice
1.3.3 Introduction to American options
Since an American option can be exercised at any time between 0 and N, we shall
define it as a positive sequence (Z,,) adapted to (F,), where Z,, is the immediate
profit made by exercising the option at time n In the case of an American option
on the stock S! with strike price K, Z, = (S} — K) ,) in the case of the put,
Zn = (K -S}) +: In order to define the price of the option associated with
(Zn)o<n<w, we shall think in terms of a backward induction starting at time
Indeed, the value of the option at maturity is obviously equal to Uy = Zy At what price should we sell the option at time N ~ 1? If the holder exercises straight away he will earn Zy_, or he might exercise at time N in which case the writer must be ready to pay the amount Zy Therefore, at time N — 1, the writer has
to earn the maximum between Zy_1 and the amount necessary at time N — 1 to
generate Zy at time N In other words, the writer wants the maximum between
Zn-1 and the value at time N — 1 of an admissible strategy paying off Zy at time
0 « {> : 5 `
N, i Sy_,E (ZxIZu¬) with Zwy = ZN/SÂ As we see, it makes sense to
price the option at time N — 1 as
Un_-1 = max (Zv-1,5%_1E" (Zn Fy-1))
By induction, we define the American option price forn = 1, ,N by #I-)) Un-1 = max — - So If we assume that the interest rate over one period is constant and equal to r So =(1 +r)" and = 1 * |
Un—1 = max (20-1, ler e (Un |Fa-1 ) ’ let U, =U, /S° be the discounted price of the American option
Proposition 1.3.6 The sequence (0.) -isa P*-supermartingale lt is the
0<n<N smallest P*-supermartingale that dominates the Sequence (Z.) , 0<n<N sy
We should note that, as opposed to the European case, the discounted price of the American option is generally not a martingale under P* _
Proof From the equality °
Ủ„_¡ = max (Z.-1,E" (0 IZa~)) ›
Trang 1312 " Discrete-time models now consider a supermartingale (Tn)o<n<N that dominates (Za)o<n<N- Then Ty > Un andif T, > U, we have f,-1 >E* (f \Fn1) > E* (0: \Fn1) whence
T,-1 > max (Z.-1,E* (Gn \Fn—1 )) = Unt
A backward induction proves the assertion that (T;,) dominates (Úa) oO
1.4 Problem: Cox, Ross and Rubinstein model
The Cox-Ross-Rubinstein model is a discrete-time version of the Black-Scholes
model It considers only one risky asset whose price is S, attimen,O<n<QN,
and a riskless asset whose return is r over one period of time To be consistent with the previous sections, we denote So =(1+r)"
The risky asset is modelled as follows: between two consecutive periods the relative price change is either a or 6, with -1 <a < b:
- Jf Sn(1 +a)
Sata = { Sa +0)
The initial stock price So is given The set of possible states is then N= {1+
a,1+}% Each N-tuple represents the successive values of the ratio Sp4i/Sn, n = 0,1, ,N — 1 We also assume that Fp = {0,9} and F = P(Q) For n = 1, ,N, the o-algebra F,, is equal to o(S1, , Sn) generated by the
random variables S,, ,S, The assumption that each singleton in 2 has a strictly positive probability implies that P is defined uniquely up to equivalence We now
introduce the variables T, = Sn/Sn-1, forn = 1, ,N If (11, ,2N) is
one element of 2, P{(11, ,2N)} = P(T,; = 11, ,Tn = xn) Asa result,
knowing P is equivalent to knowing the law of the N-tuple (T,,T2; ,Tv) We’ also remark that forn > 1, ¥, = 0(T1, -,Tn)-
1 Show that the discounted price (S,) is a martingale under P if and only if
E(Tn4i|Fa) = Ptr, Wn € {0,1, ,N — 1}
The equality B(§n41|Fn) = §, is equivalent to E(§n41/Sn|Fn) = 1, since Sn is #,-measurable and this last equality is actually equivalent to E(Tn+1 |Fn) = 1+r 2 Deduce that r must belong to Ja, b[ for the market to be arbitrage-free
If the market is viable, there exists a probability P* equivalent to P, under which (Sn)
is a martingale Thus, according to Question 1 `
E'(TeailZa)=1+r 7
and therefore B*(Tn41) = 1+ Since Tn+1 is either equal to 1 + a or 1 + 6 with non-zero probability, we necessarily have (1 +r) €]1 + a,1+ 6]
Problem: Cox, Ross and Rubinstein model 13 3 Give examples of arbitrage strategies if the no-arbitrage condition derived in
Question (2.) is not satisfied
Assume for instance that r < a By borrowing an amount 5» at time 0, we can purchase one share of the risky asset At time N, we pay the loan back and sell the risky asset We realised a profit equal to Sw — %a(1 + r)Ÿ which is always positive, since Sw > So(1 +a)” Moreover, it is strictly positive with non-zero probability There is arbitrage opportunity If r > 6 we can make a riskless profit by short-selling the risky
asset ‘
4 From now on, we assume that r € Ja, b[ and we write p = (b — r)/(b — a)
Show that (Sn) isa P-martingale if and only if the random variables 7}, To, tes Tw are independent, identically distributed (ID) and their distribution is
given by: P(T; =1+a) =p=1-P(T, = 1+5) Conclude that the market is arbitrage-free and complete
If T; are independent and satisfy P(T; = 1+a)=p=1- P(T; = 1+6), we have - Efn+i|7fa) = B(n+i) = p( +4) +(1— p)(1 +6) =1+z
and thus, (S,,) is a P-martingale, according to Question 1
Conversely, if forn = 0,1, ,N — 1, E(Tn41|Fn) = 1417, wecan write
ạ + a)E (1¢7, 4,=14+0}!Fn) + lại + b)E (17, 4:=148)|Fa) =l+r
Then, the following equality
E (10,,i=i+a)l#f2) + B (,¿y=i+e 7n) = 1,
implies that E (1¢7, : ° n41=l+e}|“n =14a}|F; ) = p and E (1‡, ~ d Fr) = 1-
induction, we prove that for any 2; € {1+a,1+ (i wnt) |Fo) — P By
P(N =Z1, ,Ía = Zn) =|»:
i=1
where p; = pif x; =1+aandp; =1- if2; = i i= pifx; = 1+ That shows that th ( i
T; are IID under measure P and that P(T; = 1+) =p _— We have shown that the very fact that (S,,) is a P-martingale uniquely determines the distribution of the N-tuple (Ti, T2, , Ty ) under P, hence the measure P itself
Therefore, the market is arbitrage-free and complete
We denote by C,, (resp P,) the value at time n, of a European call (resp put) ona share of stock, with strike price K and maturity N
(a) Derive the putcall parity equation 7 Ca — Py =S,~—K(14+r)7 8-9),
knowing the put/call prices in their conditional expectation form
If we denote E” the expectation with respect to the probability measure P* under
which (S,.) is a martingale, we have ‘
ICn— Pa = (1+) EY (Sy — K)4 — (K - Sw) 4|Fa)
= (L+r) O° ET (Sw — K|Fa)
Trang 1414 Discrete-time models
the last equality comes from the fact that (52) is a P*-martingale
(b) Show that we can write C,, = c(n, S,) where c is a function of K, a; br and p
When we write Sv = Sp Tas T;, we get
Cn = (1+ 7) OME (s ÏI=- “K) Fn i=n+l
Since under the probability P*, the random variable Tin 41 T; is independent of
Fp and since S, is Fn-measurable, Proposition A.2.5 in the Appendix allows us to write: C, = c(n, Sn), where c is the function defined by
c{n, z)
(t+r)~N-")
nhám 7
= treat Goren enna),
6 Show that the replicating strategy of a call is characterised by a quantity H, = A(n, S,—1) at time n, where A will bé expressed in terms of function c
We denote H® the number of riskless assets in the replicating portfolio We have
HẠ(1L +r)” + AnSn = c(n, Sn)
Since Hp and H„ are Fn—1-measurable, they are functions of 5i, „0„—1 Only and,
since S,, is equal to S,-1(1 + @) or Sn-1(1 + 6), the previous equality implies Ho(i +r)" + AnSn-1(1 +a) = cín, S2—1(1 + đ))
and
Ho(itr)” + AnSn-i(1 + 6) = e(n, Sn—i(1 +9)
Subtracting one from the other, it turns out that :
c{n, z(1 + b)) — c{n, z(1+ 3)
A(n,z) = z(b — a)
7, We can now use the model to price a call or a put with maturity T on a single stock In order to do that, we study the asymptotic case when N converges
to infinity, and r = R7/N, log((1 +a)/(1+r)) = —ơ/VN and log((1 + b)/(1+r)) = o/VN The real number R is interpreted as the instantaneous rate at all times between 0 and T, because e®? = limy_, (1 + r)% 0? can
be seen as the limit variance, under measure P*, of the variable log(Sj), when
N converges to infinity %5 (a) Let (Yw)w>¡ be a sequence of random variables equal to
Yy= XÈ +X? +-::+XN
Problem: Cox, Ross and Rubinstein model 15
where, for each N, the random variables X N are HD, belong to {-ø/VN, o/VN},
and their mean is equal to wy, with limny_,.o(Nun) = p Show that the sequence (Yn) converges in law towards a Gaussian variable with mean u and variance ø? We just need to study the convergence of the characteristic function gyy of Yn We obtain ‘ @yx (u) =E (exp(tuYn)) = H E (exp (iux}’)) (E (exp (ux?)))”
= (1 + tUpin —07u?/2N + o(1/N)) ”
Hence, limyoo $¥y(u) = exp (iup — o?u?/ 2), which proves the convergence
in law
(b) Give explicitly the asymptotic prices of the put and the call at time 0
For a certain N, the put price at time 0 is given by 2 + N on Pi®) (1+ RT/N)-‘E* (x-s I) sử + ẹ ‘n=l E* (a + RTJN)"YK — So exp(Yw)) | ._ K¬N with Ywy'= 3n log(Tz/(1 + r)) According to the assumptions, the variables N Xj = log(T;/(1 + r)) are valued in {—z/VN,øơ/VN}, and are IID under probability P* Moreover : 4 _ o/VN _ -o//N ) = (1- 2p) = FE ee VN e2Z/VN _ c-z/VN VN"
Therefore, the sequence (Yw) satisfies the conditions of Question 7.(a), with u=
—ø”/2 If we write (w) = (Ke~*T — Soe")+, we are able to write lý”) — Et ((¥w))| - E* ((a + RT/N)-" K — Soexp(¥w)) , E*(x;’ — (Ke? _ gy exp(¥w)),,} K|[(+ RT/N)"" —e7*7| IA
Since w is a bounded l continuous function and because the sequence (Yn) con- verges in law, we conclude that
im im B* (Fn)
_ Noo
Trang 1516 | Discrete-time models +00 = 1 / (Ke "7 - SoeT— 2/2199) a— 92/2 gu, V2mrJ_—« The integral can be expressed easily in terms of the cumulative normal distribution F, so that lim PO”) = Ke®? F(—d2) - SoF(-41), N¬œ where dị = (log(z/K) + RT + ø?/2)/ø, dạ = dị — ơ and F4) = , wo N The price of the call follows easily from put/call parity lim yoo CÁ ) = SoF(di)— Ke~*TF(da)
Remark 1.4.1 We note that the only non-directly observable parameter 1s ø Íts interpretation as a variance suggests that it should be estimated by statistical methods However, we shall tackle this question in Chapter 4
2
e? (dz
Notes: We have assumed throughout this chapter that the risky assets were not offering any dividend Actually, Huang and Litzenberger (1988) apply the same ideas to answer the same questions when the stock is carrying dividends The theorem of characterisation of complete markets can also be proved with infinite probability spaces (cf Dalang, Morton and Willinger (1990) and Morton (1989)) In continuous time, the problem is much more tricky (cf Harrison and Kreps (1979), Stricker (1990) and Delbaen and Schachermayer (1994)) The theory of complete markets in continuous-time was developed by Harrison and Pliska (1981 , 1983) An elementary presentation of the Cox-Ross-Rubinstein model Is given in the book by J.C Cox and M Rubinstein (1985) -
2
Optimal stopping problem and American options
The purpose of this chapter is to address the pricing and hedging of American options and to establish the link between these questions and the optimal stopping problem To do so, we will need to define the notion of optimal stopping time, which will enable us to model exercise strategies for American options We will also define the Snell envelope, which is the fundamental concept used to solve.the optimal stopping problem The application of these concepts to American options will be described in Section 2.5
2.1 Stopping time -
The buyer of an American option can exercise its right at any time until maturity The decision to exercise or not at time n will be made according to the informa- tion available at time n In a discrete-time model built on a finite filtered space (2, F, (Fn)ocn<n > P), the exercise date is described by arandom variable called stopping time
Definition 2.1.1 A random variable v taking values in {0,1,2, , N} isa Stop-
ping time if, for any n€ {0,1, -, N},
{v=n}e Fy
Remark 2.1.2 As in the previous chapter, we assume that F = P() and P({w}) > 0, Wwe ©ˆ This hypothesis is nonetheless not essential: if it does
not hold, the results presented in this chapter remain true almost surely However, we will not assume Fo = {0,2} and Fy = F, except in Section 2.5, dedicated
to finance
Remark 2.1.3 The reader can verify, as an éxercise, that v is a stopping time if
and only if, for any n € {0,1, ,N},
{U < n} € 7a
Trang 1618 Optimal stopping problem and American options Let us introduce now the concept of a ‘sequence stopped at a stopping time’ Let (Xa)s<„< be a sequence adapted to the filtration (Fn)ocncy and let v be a stopping time The sequence stopped at time v is defined as
Xã (2) = Xy(w)an (w) )
i.e., on the set {⁄ = 7} we have
„_( X; Wj<n x ={ n ifj>n Note that X¥, (w) = Xiu (w) (= X; on {v = j})
Proposition 2.1.4 Let (Xn) be an adapted sequence and v be a stopping time The stopped sequence (Xx )ocnen #5 adapted Moreover, if (X,,) is a martingale
(resp a supermartingale), then (Xy) is a martingale (resp a supermartingale) Proof We see that, form > 1, we have
Xvan = Xo + » $; (X; — Xj-1) ’
j=l "
where $; = Ì{;<u}: Since {j < v} is the complement of the set {y<j} =
{ < 7 — 1}, the process (Ón)o<n< 38 predictable
It is clear then that (X„An)o<a<w 1S adapted to the filtration (Ta)o<n<N-
Furthermore, if (X,,) is a martingale, (X_,n) is also a martingale with respect to (Fp), since it is the martingale transform of (Xn) Similarly; we can show that
if the sequence (X,,) is a supermartingale (resp a submartingale), the stopped sequence is still'a supermartingale (resp a submartingale) using the predictability
and the non-negativity of (¢;)o<j<n- đ
2.2 The Snell:envelope `
In this section, we considér an adapted sequence (Zn)o<n<n> and define the sequence (U,,)o<n<wn as follows:
Un = Zn
U, = max(Zn,E(Unt|Fn)) Vn<N~—1
The study of this sequence is motivated by our first approach of American options (Section 1.3.3 of Chapter 1) We already know, by Proposition 1.3.6 of Chapter 1, that (Un)o<n<wn is the smallest supermartingale that dominates the sequence (Za)o<n<N We call ít the Snell envelope of the sequence (Zn)o<n<n-
By definition, U,, is greater than Z,, (with equality forn = -N) and in the case of
a strict inequality, Un = E(Un4i|Fn) It suggests that, by stopping adequately the
sequence (U,), it is possible to obtain a martingale, as the following proposition shows ,
The Snell envelope 19
Proposition 2.2.1 The random variable defined by
vp = inf {n > 0|Un = Zn} (2.1)
se sopping time and the stopped sequence (Unarrg)o <n<wn #8 4 martingale
ros - Since Uy = Zn, UY is a well-defined element of {0,1, , N} and we
{Uo = 0} = {Uo = Zo} € Fo,
and fork > 1
{v0 =k} = {Uo > Zo} 1 - 1 {Ue-1 > Ze-1} 1 {Uk = Ze} € Fe
To demonstrate that (U*%°) i i ;
214: (Ux) is amartingale, we write as in the proof of Proposition
Un? = Unave = Uo,+ >> b;AU;,
j=l
where $; = 11,,>,} So that, forn € {0,1, ,N—1}, Una ~ U,° nti (On+1 — Un)
= l{n+i<ze} (Ua+n — Un) *
By definition, U, = max (Z,, E (Un41|Fn)) and on the set {n+ 1 < v9}, U, > - Zn Consequently U, = E (Un41|Fn) and we deduce
Uniti — Un? = Mnticin} (Unt1 — E (Un41|Fn)) and taking the conditional expectation on both sides of the equality
E((U73, ~ U29) [Za) = lexa<a)E ((U»¿i — B Until Fa))| Fa)
{n+ 1 < uọ} € Fy (since the complement of {n +1 < vo} is {vp <
n}) 7 7
Hence
| B (Uns ~ Us") Fa) =0,
which proves that U” is a martingale , n In the remainder, we shall note Jn, the set of stopping times taking values in
in, n+ 1, .,N} Notice that 7;, xv is a finite set since © is assumed to be finite
e martingale property of the sequence U”° gives the following result which relates the concept of Snell envelope to the optimal stopping problem
Corollary 2.2.2 The stopping time vo satisfies
Uo = E(Z,,|Fo) = sup, E(Z,|Fo)
" vVETO.N
If we think of Z,, as the total winnings of a gambler after n games, we see that stopping at time vp) maximises the expected gain given Fo
Proof Since U”° is a martingale, we have
Trang 1720 Optimal stopping problem and American options On the other hand, if v € 7o,n, the stopped sequence U” is a supermartingale So that Uo E (Ux|Fo) = E(U,|7o) E (Z,|Fo) ; which yields the result n > > Remark 2.2.3 An immediate generalisation of Corollary 2.2.2 gives U, = sup E(Z,|F,) U€Tn,N , = E (Zv, |Za) ,
where v,, = inf {j > n|U; = Z;}
Definition 2.2.4 A stopping time v is called optimal for the sequence (Zn)gcn<n
if
E(Z,|Fo) = sup E(Z,|Fo)
Ton
We can see that vo is optimal, The following result gives a characterisation of optimal stopping times that shows that vo is the smallest optimal stopping time Theorem 2.2.5 A stopping time v is optimal if and only if
{ fy = Uy and (Uvan)oen<n is 4 martingale (2.2)
Proof If the stopped sequence U” is a martingale, Up = E(U,|Fo) and con- sequently, if (2.2) holds, Up = E(Z,|Fo) Optimality of v is then ensured by Corollary 2.2.2 : Conversely, if v is optimal, we have Up = E(Z.|Fo) < E(UL|Fo) - But, since U” is a supermartingale, E (U.|Fo) < Uo Therefore and since U, > Z,, U, = Zp Since E (U,|Fo) = Uo and from the following inequalities Uo > E (Uvan|Fo) > E (U,|Zo)
(based on the supermartingale property of (UY )) we get
E (Uvan|Fo) = E (U,|Fo) = E(E (Us| Fn)| Fo)
But we have Uyan > E(U.|F,), therefore „A„ = E(U_| Fn), which proves that (U27) is a martingale n
Decomposition of supermartingales 21 2.3 Decomposition of supermartingales
The following decomposition (commonly called ‘Doob decomposition’) is used in viable complete market models to associate any supermartingale with a trading strategy for which consumption is allowed (see Exercise 5 for that matter) Proposition 2.3.1 Every supermartingale (Un)o<n<n has the unique following decomposition:
Uz = Ma — An,
where (M,,) is a martingale and (A,,) is a non-decreasing, predictable process,
null at 0
Proof It is clearly seen that the only solution for n = 0 is Mp = Up and Ap = 0 Then we must have
Unt — Un = Mnsi — Mn — (Anti — An)
So that, conditioning both sides with respect to F,, and using the properties of M
and A `
_ (Anti — An) =E (Un+1|Frn) -U, and
3 Mansi — Mn = Una — E(Da+i|Za)
(M,) and (A,) are entirely determined using the previous equations and we see
that (M,,) is a martingale and that (A,.) is predictable and non-decreasing (because (U,,) is a supermartingale) ¬ Oo Suppose then that (U,,) is the Snell envelope of an adapted sequence (Z,,) We
can then give a characterisation of the largest optimal stopping time for (Z,,) using
the non-decreasing process (A,,.) of the Doob decomposition of (U,):
Proposition 2.3.2 The largest optimal stopping time for (Za) is given by y -{m if An =0
me inf {n, An41 40} ifAn #0
Proof It is straightforward to see that v,,, is a stopping time using the fact that (An)o<n<w is predictable From U, = M, — A, and because A; = 0, for
j < Max, we deduce that U* = M= and conclude that U= is a martingale To
Trang 1822 - Optimal stopping problem and American options
We have E (Uj41|F;) = M; — 4;+¡ anđ, on the set {„ = 7}, 4; = 0 and 4;+i >0,so Ủy = M; and E( Uj41|F;) = Mj - Ajai < Uj It follows that
U;= max (Z;,E (UyailF;)) = Z; So that finally
BI = Zeina
It remains to show that it is the greatest optimal stopping time If v is a stopping
time such that vy > y,,, and P(v > „) > 0, then
E(U,) = E(M,) — B(A,) = E(Uo) — B(A,) < E(Uo)
and U” cannot be a martingale, which establishes the claim n
2.4 Snell envelope and Markov chains
The aim of this section is to compute Snell envelopes in a Markovian setting A
sequence (Xn)n>o of random variables taking their values in a finite set Eis
called a Markov chain if, for any integer n > 1 and any elements Zo, 2j, ,
In-1,2,y of EB, we have
P(Xn4i = y|Xo = Tọ; »Xn-1 = In-1,Xn = z)=P(X„¿¡ = y|Xn = #) `
The chain is said to be homogeneous if the value P(z, y) = P (Xn+i = yl|Xn = 2) does not depend on n The matrix P = (P(z,))(„, ye EXE indexed by Ex E,
is then called the transition matrix of the chain The matrix P has non-negative entries and satisfies: yeE P(z,y) = 1 forall x €'E; it is said to be a stochastic matrix On a filtered probability space (9, Fy (Fridoenen > P) , we can define the notion of a Markov chain with respect to the filtration: ,
Definition 2.4.1 A sequence (Xn)o<n<n of random variables taking values in
E is a homogeneous Markov chain with respect to the filtration (Fn) ycn<y» with
transition matrix P, if (Xn) is adapted and if for any real-valued function f on E, we have `
B( (Xa¿i) [#a) = Pƒ (Xa),
where P f represents the function which mapsx € Eto P f(x) = yee P(z,y)f(y)
Note that, if one interprets real-valued functions on E as matrices with a single
column indexed by E, then Pf is indeed the product of the two matrices P and
f It can also be easily seen that a Markov chain, as defined at the beginning of the section, is a Markov chain with respect to.its natural filtration, defined by Fn = 0(Xo, -,Xn) ` ˆ
The following proposition is an immediate consequence of the latter definition and the definition of a Snell envelope ˆ
Proposition 2.4.2 Let (Z,,) be an adapted sequence defined by Zn = b(n, Xn),
where (Xp) is a homogeneous Markov chain with transition matrix -P, taking
values in E, andy is a function from N x E to R Then, the Snell envelope (Un)
Application to American options 23
of the sequence (Z,,) is given by U, = u(n, Xn), where the function u is defined
by
u(N,z) =u(N,z) VieE
and, form < N - 1,
u{n, :) = max (#(n,-), Pu(n + 1,-)) 2.5 Application to American options
From now on, we will work in a viable complete market The modelling will be based on the filtered space (9, Z,(Ta)o<n<ụ P) ‘and, as in Sections 1.3.1 and
1.3.3 of Chapter 1, we will denote by P* the unique probability under which the
discounted asset prices are martingales
2.5.1 Hedging American options
In Section 1.3.3 of Chapter 1, we defined the value process (U,,) of an American
option described by the sequence (Z,,), by the system
Un = Zn
U, = max (Zn, S2E* (Unai/S Suil#a)) Vn<N-—I1
Thus, the sequence (U,,) defined by U,, = U,,/S® (discounted price of the option)
is the Snell envelope, under “ Of the sequence (Za) We deduce from the above Section 2.2 that Un = sup Et (Zuz.) - vETaN and consequently e Zz U, = S09 2 sup B* (Sir) E* (| (7, ) From Section 2.3, we can write - Un = Mạ — Ấn,
where (M,,) is a P*-martingale and (A,) is an increasing predictable process,
Trang 19-2A4 -~ Optimal stopping problem and American options and consequently
Un = Va(d) - A
Therefore
U, = Vn(¢) _
where A, = S° A,, From the previous equality, it is obvious that the writer of the option can hedge himself perfectly: once he receives the premium Up = Vo(¢), he can generate a wealth equal to V,,(@) at time n which is bigger than U,, and a fortiori 2a
What is the optimal date to exercise the option? The date of exercise is to be chosen among all the stopping times For the buyer of the option, there is no point
in exercising at time n when U,, > Z,,, because he would trade an asset worth U,,
(the option) for an amount Zy (by exercising the option) Thus an optimal date r of exercise is such that U; = Z, On the other hand, there is no point in exercising after the time
= inf {j, Ajzi AO}
(which is equal to inf {3 3 Aj41 z o}) because, at that time, selling the option provides the holder with a wealth Uj = V„„(Ó) and, following the strategy ¢ from that time, he creates a portfolio whose value is strictly bigger than the ‘option’s at times „„ + 1, Y%ax + 2, ,N Therefore we set; as a second condition, T < v,,, which allows us to say that Oi isa martingale As aresult, optimal dates of exercise are optimal stopping times for the sequence (Zp ), under probability P* To make this point clear, let us consider the writer’s point of view If he hedges himself using the strategy ¢ as defined above and if the buyer exercises at time 7
which is not optimal, then U; > Z, or A; > 0 In both cases, the writer makes a
profit V,(¢) — Z, = U; + Ar — Z,, which is positive 2.5.2 American options and European options |"
Proposition 2.5.1 Let Cy be the value at time n of an American option described
by an adapted sequence (Zn)gcncn and let Cn be the value at time n of the
European option defined by the Fy-measurable random variable h = Zy Then, we have Cy > Cn
Moreover, if Cn > Zn for any n, then
n=Cn Vn€({0,1, ,N}
The inequality C,, > c, makes sense since the American option entitles the holder to more rights than its European counterpart ©
Proof For the discounted value (Cn) i is a supermartingale under P*, we have
é, > E* (đ„IZ-) =E* (ẽw|#a) = ẽ
Hence Ca > cạ
Exercises 25
If cn, > Z, for any n then the sequence (é,), which is a martingale under P*,
appears to be a supermartingale (under P*) and an upper bound of the sequence
(Zn) and consequently
Cn SG, Wn {0,1, , N}
O
Remark 2.5.2 One checks readily that if the relationships of Proposition 2.5.1
did not hold, there would be some arbitrage opportunities by trading the options To illustrate the last proposition, let us consider the case of a market with a single risky asset, with price S, at time n and a constant riskless interest rate,
equal to r > 0 on each period, so that S? = (1 +1)” Then, with notations of Proposition 2.5.1, if we take Z, = (S, — K)4, cy is the price at time n of a
European call with maturity N and strike price K on one unit of the risky asset and C,, is the price of the corresponding American call We have
En (1+ r)-“E* ((Sw - K)+|Za)
E* (Sn —~K(1+ ry" |F a) |
= S,-—K(1+r)-%,
using the martingale property of (S,,) Hence: cn > Sy — K(1+1r)7(N-™) > S, — K, for r > 0.As cy’> 0, we also have c, > (S, — K)4 and by Proposition
2.5.1, Cy = Cn There is equality between the price of the European call and the price of the corresponding American call
-This property does not hold for the put, nor in the case of calls on currencies or dividend paying stocks
Notes: For further discussions on the Snell envelope and optimal stopping, one ~ may consult Neveu (1972), Chapter VI and Dacunha-Castelle and Duflo (1986), Chapter 5, Section 1 For the theory of optimal stopping in the continuous case, see El Karoui (1981) and Shiryayev (1978) :
IV
2.6 Exercises
Exercise 1 Let v be a stopping time with respect to a filtration (F,)o<n<n-
We denote by F, the set of events A such that AN {v = n} € F, for any neé {0, ,N}
1 Show that F, is a sub-o-algebra of Fy F, is often called ‘o-algebra of events
determined prior to the stopping time v’
2 Show that the random variable v is ¥,-measurable
3 Let X bea real-valued random variable Prove the equality
N
Yo l= E(X1F))
j=0
Trang 2026 Optimal stopping problem and American options 4 Let r be a stopping time such that r > v Show that F, C F;
5 Under the same hypothesis, show that if (/,,) is a martingale, we have
M, = B(M.|7.)
(Hint: first consider the case r = N.)
Exercise 2 Let (U,) be the Snell envelope of an adapted sequence (Z,) Without
assuming that Fo is trivial, show that
E(Ua)= sup E(Z,), U€To,N
and more generally
E (Un) = sup E (Z.) ` 1U€Ta,N
Exercise 3 Show that v is optimal according to Definition 2.2.4 if and only if E(Z,)= sup E(Z;) T€To,N
Exercise 4 The purpose of this exercise is to study the American put in the model of Cox-Ross-Rubinstein Notations are those of Chapter 1
1 Show that the price P,,, at time n, of an American put on a share with maturity N and strike price K can be written as Tn= Pam(n, Sn) : where Pam (n, x) is defined by Pam(N, 2) = (K ~x)4 and, forn < N—1, f(n+1,2) 1+r , with f(n + 1,2) = pPam(n +1,x2(1 + a)) + (1 — p)Pam(n + 1,2(1 + ð)) and p = (b.— r)/(b — a) , 2 Show that the function P, ,(0,.) can be expressed as Pam(0,2) = sup E* ((1+7r)7"’(K —2V,)4), vETO,N Pam(n, +) = max (œ ~ #)+,
where the sequence of random variables (Vạ )o<n<ụ is defned by: Vọ = l
and, forn > 1, V, = [iL Ui, where the U;’s are’ some random variables
Give their joint law under P*
3 From the last formula, show that the function 2 ++ Pam(0,2) is convex and
non-increasing ,
4 We assume a < 0 Show that there is a real number x* € (0, K] such that, for + <S #*, Pzm(0,+) = (K — +)+ and, for z c]z ,K/(+a)%[, Pam(0,2) >
(K — 2)+
5 An agent holds the American put at time 0 For which values of the spot So would he rather exercise his option immediately?
Exercises 27
6 Show that the hedging strategy of the American put is determined by a quantity Hạ = A(n, S,-1) of the risky asset to be held at time n, where A can be
written as a function of Pam
Exercise 5 Consumption strategies The self-financing strategies defined in Chapter 1 ruled out any consumption Consumption strategies can be introduced
in the following way: at time n, once the new prices So sẻ st are quoted, the
investor readjusts his positions from ¢, to ¢n4, and selects the wealth yn, to be consumed at time n + 1 Any endowment being excluded and the new positions being decided given prices at time n, we deduce
on41-5 = on-Sn — Yn4+1- (2.3)
So a trading strategy with consumption will be defined as a pair (¢,y), where @ is a predictable process taking values in IR, representing the numbers of
assets held in the portfolio and y = (Ya)i<n< N 1S a predictable process taking
values in IRt, representing the wealth consumed at any time Equation (2.3) gives the relationship between the processes ¢ and and replaces the self-financing
condition of Chapter 1
1 Let @ bea predictable process taking values in IR*? and let y be a predictable
process taking values in IR* We set Va(¢) = @n-Sn and Va(d) = da-Šn
Show the equivalence between the following conditions:
(a) The pair (4, 7) defines a trading strategy with consumption (b) For any n € {1, , N}, Va($) = Vo(d) + 3 9/.A6; — Ð 2a, j=l j=l (c) Foranyn € {1, ,N}, Va(d) = Vo(d) + 3 2ó;.AŠ; — Nxj/S9 j=l j=l /
2 In the remainder, we assume that the market is viable and complete and we
denote by P* the unique probability under which the assets discounted prices
are martingales Show that if the pair (¢,y) defines a trading strategy with consumption, then (V,,(¢)) is a supermartingale under P*
3 Let (U,) be an adapted sequence such that (U,,) is a supermartingale under
P* Using the Doob decomposition, show that there is a trading strategy with
consumption (¢, y) such that V,(¢) = U, for any n € {0, , N}
4 Let (Z,,) be an adapted sequence We say that a trading strategy with consump-
tion (¢, y) hedges the American option defined by (Z,) if Va(¢) > Zn for
any n € {0,1, ,.N} Show that there is at least one trading strategy with
consumption that hedges (Z,,), whose value is precisely the value (U,) of the
American option Also, prove that any trading strategy with consumption (4, +) `
Trang 21Hl
28 Optimal stopping problem and American options 5 Let x be a non-negative number representing the investor’s endowment and let = (Yn)i<n<n be a predictable strategy taking values in IR* The consump- tion process (Yn) is said to be budget-feasible from endowment z if there is a
‘predictable process ¢ taking values in TR#?!, such that the pair (ở, y) defines a trading strategy with consumption satisfying: Vo(¢) = 2 and V,(@) 2 0, for any n € {0, , N} Show that (ya) is budget-feasible from endowment z if and only if E* (oa 74/591) <z ` 3 Brownian motion and stochastic differential equations
The first two chapters of this book were dealing with discrete-time models We had the opportunity to see the importance of the concepts of martingales, self- financing strategy and Snell envelope We are going to elaborate on these ideas in
a continuous-time framework In particular, we shall introduce the mathematical tools needed to model financial assets and to price options In continuous-time,
the technical aspects are more advanced and more difficult to handle than in discrete-time, but the main ideas are fundamentally the same
Why do we consider continuous-time models? The primary motivation comes from the nature of the processes that we want to model In practice, the price changes in the market are actually so frequent that a discrete-time model can
barely follow the moves On the other hand, continuous-time models lead to more explicit computations, even if numerical methods are sometimes required Indeed,
the most widely used model is the continuous-time Black-Scholes model which leads to an extremely simple formula As we mentioned in the Introduction, the connections between stochastic processes and finance are not recent Bachelier (1900), in his dissertation called Théorie de la spéculation, is not only among the first to look at the properties of Brownian motion, but he also derived option pricing formulae ;
We will be giving a few mathematical definitions in order to understand
continuous-time models In particular, we shall define the Brownian motion since
it is the core concept ofthe Black-Scholes model and appears in most financial
asset models, Then we shall state the concept of martingale in a continuous-time set-up and, finally, we shall construct the stochastic integral and introduce the
differential calculus associated with it, namely the 1t6 calculus ,
It is advisable that, upon first reading, the reader passes over the proofs in small print, as they-are very technical
Trang 2230 - Brownian motion and stochastic differential equations Definition 3.1.1 A continuous-time stochastic process in a space E endowed with
ao- algebra E is a family (Xt)em+ of random variables defined on a probability space (2, A, P) with values in a measurable space (E, €)
Remark 3.1.2
e In practice, the index ¢ stands for the time
e A process can also be considered as arandom map: for each w in 2 we associate the map from IRt to E: t > X;(w), called a path of the process
e A process can be considered as a map from IR* x 2 into E We shall always consider that this map is measurable when we endow the product set Rt x
with the product o-algebra B(IR*) x A and when the set E is endowed with €
e We will only work with processes that are indexed on a finite time interval
(0, T]
As in discrete-time, we introduce the concept of filtration
Definition 3.1.3 Consider the probability space (Q, A, P), a filtration (Fideso is
an increasing family of o-algebras included in A
The o-algebra Z7; represents the information available at time ¢ We say that a
process (X¢}z>0 is adapted to (Fz)t>0, if for any t, Xz is F-measurable
Remark 3.1.4 From now on, we will be working with filtrations which have the following property
If A € Aand if P(A) =0, then for any t, A€ F; -
In other words F; contains all the P-null sets of A The importance of this technical
assumption is that if X = Y Pas and Y is F,-measurable then we can show that X is also #;-measurable "
We can build a ñltration generated by a process (Ã;);>o and we write F, =
ơ(X.,s < t) In general, this filtration does not satisfy the previous condition
However, if we replace F; by #; which is the o-algebra generated by both F, and WV (the g-algebra generated by all the P-null sets of A), we obtain a proper filtration satisfying the desired condition We call it the natural filtration of the
process (X;):>0 When we talk about a filtration without mentioning anything, it is assumed that we are dealing with the natural filtration of the process that we are
considering Obviously, a process is adapted to its natural filtration
As in discrete-time, the concept of stopping time will be useful A stopping time ‘is a random time that depends on the underlying process in a non- anticipative way In other words, at a given time t, we know if the stopping time is smaller than t Formally, the definition is the following: ,
Definition 3.1.5 7 is a stopping time with respect to the filtration (Fe)e>o if TIS a random variable in R* U {+00}, such that for any t s0
{r<t)}€7
Brownian motion 31
The o-algebra associated with 7 is defined as
Fr, = {AE A, foranyt >0,AN{r<t}e€ Fi}
This o-algebra represents the information available before the random time r One
can prove that (refer to Exercises 8, 9, 10, 11 and 14):
Proposition 3.1.6
e ffSisa stopping time, S is Fs measurable
e If S is a stopping time, finite almost surely, and (Xt)e>0 is a continuous,
adapted process, then Xs is Fs measurable
e If S andT are two stopping times such that S<T Pas then Fs C Fr
e If S andT are two stopping times, then S AT = inf(S, T) is a stopping time
In particular, if S is a stopping time and t is a deterministic time S At is a
stopping time
3.2 Brownian motion
A particularly important example of stochastic process is the Brownian motion It will be the core of most financial models, whether we consider stocks, currencies
or interest rates
Definition 3.2.1 A Brownian motion is a real-valued, continuous stochastic pro-
cess (Xt)s>0, with independent and stationary increments In other words:
e continuity: Pas the map s+ X,(w) is continuous
e independent increments: If s < t, X_— X is independent of Fz =o(Xu,us
8)
e stationary increments: if:s < t, Xt —- Xs and X;_„ — Xo have the same
probability law sả
This definition induces the distribution of the process X¢, but the result is difficult
to prove and the reader ought to consult the book by Gihman and Skorohod (1980)
for a proof of the following theorem “
Theorem 3.2.2 If (Xt)t>0 is @ Brownian motion, then X, — Xo is a normal random variable with mean rt and variance o*t, where r and o are constant real numbers
Remark 3.2.3 A Brownian motion is standard if
Xo=0 Pas E(X:)=0, E(X?)=t.-
From now on, a Brownian motion is assumed to be standard if nothing else is
mentioned In that case, the distribution of X; is the following:
! e ( =) dx
V2zt P \ 2
Trang 2332 Brownian motion and stochastic differential equations The following theorem emphasises the Gaussian property of the Brownian motion We have just seen that for any t, X; is a normal random variable A stronger result is the following: Theorem 3.2.4 If (Xt)t>0 is a Brownian motion and if 0 < tị < - < tạ then (X2,, ., Xt,) is a Gaussian vector The reader ought to consult the Appendix, page 173, to recall some properties of Gaussian vectors ,
Proof Consider 0 < í¡ < - < tn, then the random vector (Xt, Xt -
Xt,, -,Xt, — Xt,_,) is composed of normal, independent random variables
(by Theorem 3.2.2 and by definition of the Brownian motion) Therefore, this vector is Gaussian and so is (Xt,, -,Xe,) D
We shall also need a definition of a Brownian motion with respect to a filtration
(Ft)
Definition 3.2.5 A real-valued continuous stochastic process is an (F;)-Brownian
motion if it satisfies:
e Foranyt > 0, X¢ is 7?¡-measurable ,
® ifs <t, X, — X,.is independent of the o-algebra F,
© ifs <t, X, — X, and Xt_, — Xo have the same law
Remark 3.2.6 The first point of this definition shows that o(X,,u < t) C Fy
Moreover, it is easy to check that an 7;-Brownian motion is also a Brownian
motion with respect to its natural filtration.-
3.3 Continuous-time martingales
As in discrete-time models, the concept of martingale is a crucial tool to explain the notion of arbitrage The following definition is an extension of the one in discrete-time
Definition 3.3.1 Let us consider a probability space (Q,A,P) and a filtration
(Ft)t>0 on this space An adapted family (M;z)1>0 of integrable random variables,
Le E(|Mt|) < +00 for any t is:
© amartingale if foranys <t, E (M.|Z,) = M,;
® asupermartingale if for anys < t, E(Mi|F;) < M,, ® a submartingale # for any s < t, Ð(M;|Z;) > M,
Remark 3.3.2 It follows from this definition that, If (Mf;);¿>o is a martingale, then
ĐB(M,) = E(Mo) for any t
Here are some examples of martingales
Proposition 3.3.3 If (Xz)t>0 is a standard F,-Brownian motion: 1 X is an 7ị-martingale
2 X}? —t is an F,-martingale C
Continuous-time martingales 33 3 exp (aX: — (a? /2)t) is an F,-martingale
Proof If s < ¢ then X; — X, is independent of the o-algebra F, Thus E(X;—
X,|Fs) = E(Xt — Xs) Since a standard Brownian motion has an expectation equal to zero, we have E(X; — X,) = 0 Hence the first assertion is proved To
show the second one, we remark that E (X? - X2|F,) E (Xt — Xs)? + 2X5(Xt — Xs)|Fs) E (Xt — X;)?|Z.) + 2X,E (Xt — X.|Z.) , and since (Ã;);¿>o is a martingale E (X:— X;|Z;) = 0, whence E (X}? — X2|F,) = E ((Xt — Xs)"|Fs) - Because the Brownian motion has independent and stationary increments, it fol- lows that E(Œœ%:- X,)?|JZ) = E (Xệ ) = t-s
The last equality is due to the fact that X, has a normal distribution with mean zero and variance t That yields E (X? —t|F,) = X? -s, ifs <t
Finally, let us recall that if g is a standard normal random variable, we know that + 4 *° + E (e*?) =f ee = er /2 On the other hand, if s < t¢ E (2“*:-”*“,) = — eos -o “2B (e zŒ=X2|7,) because X, is ,-measurable Since X; — X, is independent of F,, it turns out that E 2) = E (zœ-*») = B(eX-s) ì = E (v3) | = exp (50% _ ›)
That completes the proof | D
Tf (M:):>o Ìs a inaningale, the property E (M;|Zs) = Ms, is also true if ¢ and s are bounded stopping times This result is actually an adaptation of Exercise 1 in
Trang 24KK
34 Brownian motion and stochastic differential equations
Theorem 3.3.4 (optional sampling theorem) /f (4;):>0 is @ continuous mar-
tingale with respect to the filtration (F,)t>0, and if 7 and T2 are two stopping
times such that T) < To < K, where K is a finite real number, then M,, is
integrable and
E(M,,|F,,) = Mr, Pas
Remark 3.3.5
e This result implies that if 7 is a bounded stopping time then E(M,) = E(Mo)
(apply the theorem with 7, = 0,72 = 7 and take the expectation on both sides) e If M; is a submartingale, the same theorem is true if we replace the previous
equality by
E(M,,|F,,) > M,, Pas
We shall now apply that result to study the properties of the hitting time of a point by a Brownian motion
Proposition 3.3.6 Once again, we consider (X+z):>0 an F,.-Brownian motion If
ais a real number, we callT, = inf {s > 0,X, = a} or +00 if that set is empty
Then, T, is a stopping time, finite almost surely, and its distribution is charac- terised by its Laplace transform
E (eT) _ 7 V2Alal,
Proof We will assume that a > 0 First, we show that T, isa | stopping time
Indeed, since X, is continuous
{To < t} = Meeqt« {sup X, >a- eS = Meeqt: Useqt,s<t {X, >a- e}
ss
That last set belongs to ¥;, and therefore the result is proved In the following, we
write x A = inf(œ, 9)
Let us apply the sampling theorem to the martingale M; = exp (0X; — (0? / 2)t) We cannot apply the theorem to T, which is not necessarily bounded; however, if n is a positive integer, T, An is a bounded Stopping t time (see Proposition 3.1.6), and from the optional sampling theorem
E (MT, An) = ]1
On the one hand M,A„ = exp (ØXT,An — Ø "ức An)/2) < exp(ơa) On the other hand, if 7¿ < +oo, lima-;+ee ẤMr,An = Mr, and if Ty = +00,
X,< < aat any , therefore lima_;+eo Ấm, An = 0 Lebesgue theorem implies that
E(ltr,<+eeyMr,) = 1, i.e since Xr„, = a when Tz’ < +00
ơ2 `
E (1(2,<+001 exp (-‡z)) =e7%, \
Stochastic integral and Ité calculus 35
By letting o converge to 0, we show that P(T, < +00) = 1 (which means that
the Brownian motion reaches the level a almost surely) Also
2
E (exp (-$*)) =e" The case a < 0 is easily solved if we notice that
T, = inf {s >0,-X, = —a},
where (— Xz)t>0 is an F,-Brownian motion because it is a continuous stochastic
process with zero mean and variance t and with stationary, independent increments oO
The optional sampling theorem is also very useful to compute expectations involving the running maximum of a martingale If M;, is a square integrable
martingale, we can show that the second-order moment of supg<t<r |Ä;| can be
bounded This is known as the Doob inequality
Theorem 3.3.7 (Doob inequality) [f (M:)o<i<r is a continuous martingale, we have 2 E ( sup Imi") < 4E(|Mr|’) O<t<T The proof of this theorem is the purpose of Exercise 13 D
3.4 Stochastic integral and Ito calculus
In a discrete-time model, if we follow a self-financing strategy ¢ = (Hn)ocncn> the discounted value of the portfolio with initial wealth Vo is ~~
ˆ n
Vo + SAS; — 5-1)
j=l
That wealth appears to be a martingale transform under a certain probability
measure such that the discounted price of the stock is a martingale As far as continuous-time models are concerned, integrals of the form f H, dS, will help
us to describe the same idea
However, the processes modelling stock prices are normally functions of one or several Brownian motions But one of the most important properties of a Brownian motion is that, almost surely, its paths are not differentiable at any point In other
words, if (X,) is a Brownian motion, it can be proved that for almost every w € Q, there is not any time ¢ in IR* such that dX; /dt exists As a result, we are not able
to define the integral above as
[ toax.= [ 10954
Trang 2536 Brownian motion and stochastic differential equations motion, and we shall call them stochastic integrals That is the whole purpose of this section
3.4.1 Construction of the stochastic integral
Suppose that (W;)¢>0 is a standard F,-Brownian motion defined on a filtered
probability space (2, A, (F:):>0,P) We are about to give a meaning to the expression J f(s, w)dW, for a certain class of processes f(s,w) adapted to the filtration (F;):>0 To start with, we shall construct this stochastic integral for a
set of processes called simple processes Throughout the text, T will be a strictly positive, finite real number
Definition 3.4.1 (Htosesr is called a simple process if it can be written as
=3 al œ)1I¿,_, ¿,])
where Ö = tọ < tị < - < tp = T and ộ¡ is 7ị,_, -measurable and bounded Then, by definition, the stochastic integral of a simple process H is the continuous
process (I(H):)o<t<r defined for any t € Ite, te41] as
He = So oi(Wi, — We 1) + bi+1 (Wt — We, )
1<i<k
Note that ƒ(JJ); can be written as
I(H),= D> ®(WùAr— Wu, cay),
1<t<p
which proves the continuity of t » I(H), We shall write J H,dW, for I(H): The following proposition is fundamental
Proposition 3.4.2 If (H:)o<t<r is a simple process: 0 (fp HedW.) - siết ne) B( [ Has), +2 (sip| [x J)<® [ "Hits Proof In order to prove this proposition, we are going to use discrete-time is a continuous F,-martingale,
processes Indeed, to show that ( J H,aW,) is a martingale, we just need to
check that, for any t > s, ` t E (/ aw,IZ.) =Í H.dW, 0 0
Stochastic integral and It6 calculus 37
If we include s and ¢ to the subdivision tj = 0 < t) < - < t, = T, and if we call M, = f,” H.dW, and Gn, = 7¡„ for 0 < n < p, we want to show that My
is a G,-martingale To prove it, we notice that +
Mạ = |" Haw, = » We, — Wei.)
i=1
with @; G;-1-measurable Moreover, X, = W;,, isaG,-martingale (since (We) ido
is a Brownian motion) (Mn),,¢/o,p] turns out to be a martingale transform of (Xn) ne{o,pj- The Proposition 1.2.3 of Chapter 1 allows us to conclude that (Mn) ne{o,p) 18 a martingale The second assertion comes from the fact that n 2 (x: $¡(X¡ — x] E(Mz) #=1 S3 EB(;4;(XiT— Xi a)(XjT—Xjj)) — @D i=1 j=1 ‘ | ; Also, if i < j, we have
BAGG” Meal 3 Kea) nà
=_ E(E(2¡i2;(X:¡ — X;—i)(X; — X;~—¡)| đy~1))
2= E(đ9,(X;¡- — Kini JE (X5 ~ X5-119;- 1))-
Since X; is a martingale, E(X; - X3-1|G;-1) = 0 Therefore, ifi < j, E (¢:6;(X: — Xi-1)(X; — Xj-1)) = 0 If j > i we get the same thing Fi-
nally, if = 7,
E (ở (X:— X:_ï)”) E (E (47 (X; ~ Xi-1)?| Gi-1))
= E (G7 E (Xi — Xi-1)?| Gi-1)) ;
as a result
B((X¡ > Xi-¡)?|[đ—¡) = B (0, — Wyj_,)2) =0 T—b.a — 2)
From (3.1) and (3.2) we conclude that cu ca
E In _ x) =E (x: Bt =) =E ([ Has)
i=1 i=1
The continuity, of t > J H,dW, is clear if we look at the definition The’ third assertion is just a consequence of Doob inequality (3.3.7) applied to the continuous
martingale (5 H,aW,) So n
Trang 2638 Brownian motion and stochastic differential equations Remark 3.4.3 We write by definition
T T t : hi
| Haw, = | Haw, [ H,dM, Ỷ
t 0 0
Ift < 7T, and If A € 7¡, then s 1A1(¿<;yH; is sull a simple process and it is
easy to check from the definition of the integral that
T T
Ỉ 1AR,1t¿<,ydW; = 14 | H,dwW, (3.3) ie
0 t ,
Now that we have defined the stochastic integral for simple processes and stated some of its properties, we are going to extend the concept to a larger class of adapted processes H pe A EEE ` ` T ?í = (t60sser (Ft)t>0 — adapted process, E (/ sta) < te] 4 0 :
Proposition 3.4.4 Consider (W:):>o an 7,-Brownian motion There exists a unique linear mapping J from H to the space of continuous F,-martingales
defined on [0,T), such that:
1 If (Ht)t<r isa simple process, Pas foranyO<t<T, J(H), =I(H)t t
2 Ift<T,E(J(H)?) =E (| a
This linear mapping is unique in the following sense, if both J and J' satisfy the
previous properties then
Pas W<t<T, (UH) =J'(H) We denote, for H € H, [ H,dW;, = J(H);
On top of that, the stochastic integral satisfies the following properties:
Proposition 3.4.5 If (H:)o<t<r belongs to H then 1 We have ®(mp|[ ) <4E ự sta) (3.4) t<T 2 Ifr is an F,-stopping time T T Pas [ H,dW, = [ l(s<:+H.dW (3.5) 0 0 —
Proof We shall use the fact that if (H,)s<r is in H, there exists a sequence (H2).<r of simple processes such that :
T
lim (/ |H, — H®|’ is) =0,
n—+cœo 0
Stochastic integral and Ité calculus 39 A proof of this result can be found in Karatzas and Shreve (1988) (page 134,
problem 2.5)
If H € Hand (H™)n>o is a sequence of simple processes converging to H in the previous sense, we have t<T T E (sup |I(H*†P), — 1H") < 4E (/ |H?t? — HP? i) (3.6) , 0 Therefore, there exists a subsequence /7#() such that B (supine) _ 1(H?0®)), P) < <i t<T 2"
thus, the series whose general term is J(H#(®+1)) — J(H#(*)) is uniformly con- vergent, almost surely Consequently I(H 9(n)), converges towards a continuous
function which will be, by definition, the map t + J(H), Taking the limit in
(3.6), we obtain
T :
E (sup |J(H); — rựr,Ƒ) <4E ( [ |H, - H®|’ i) + BA)
t<T 0 :
That implies that (J(H)t)o<e<r does not depend on the approximating sequence
On the other hand, (J(H)t Joxesr i is a martingale, indeed
E(I (H"),|Z,) = I(H"),
Moreover, for any t, liMn++oo1(H"), = J(H), in L*(9;P) norm and, because
the conditional expectation is continuous in L?(, P), we can conclude
From (3.7) and from the fact that E([(H")?) = E (1P ds), it fol-
lows that E(J(H)?) =E bì \H.|° ds) In the same way, from (3.7) and since
EGup,<~ J(H")ÿ) < 4E (5 IH?i ds), we prove (3.4)
The uniqueness of the extension results from the fact that the set of simple processes is dense in H °
We now prove (3.5) First of all, we notice that (3.3) is still true if H € H This is justified by the fact that the simple processes are dense in H and by (3.7) We first consider stopping times of the form 7 = = Vicicn t;14,;, where
0<t <-+ <t, = T and the A;’s are disjoint and F;, measurable, and we
prove (3 5) i in that case First
c
T T
[ Ion Haw, = [ - » 1A,Í{s;>:,} H,dW,,
Trang 2740 Brownian motion and stochastic differential equations also, each 1,,5+;}14;Hs is adapted because this process is zero if s < t¢, and is
equal to 14,H, otherwise, therefore it song to H It follows that 1<:i<n T So Usrr}HsdWs = TP H,dW, = Lf nan, 1<i<n and then ƒ“ 1(;<„)H,dW, = jy H,dW, |
In order to prove this result for an arbitrary stopping time 7, we must notice that 7 can be approximated by a decreasing sequence of stopping times of the previous form If (k+ Nr,
™m™ = » on — Mage, <<“)
o<i<2"
T, converges almost surely to 7 By continuity of the map t 4 J H,dW, we can affirm that, almost surely, So” H,dW, converges to So H,dW,, On the other hand 2 E T ` , T Ỉ l¿<r+H,dW, — Ỉ 1{,<:„¡H.dW, 0 0
converges to i lys<r} HsdW, in L?(Q,P) (a subsequence converges almost surely) That completes the proof of (3.5) for an arbitrary stopping time oO In the modelling, we shall need processes which only satisfy a weaker integrability condition than the processes in H, that is why we define , T H = { Uinoses (F:)t>0 — adapted process, / H?ds < +oo Pas 0 The following proposition defines an extension of the stochastic integral from H to H ,
Proposition 3.4.6 There exists a unique linear mapping J from H into the vector space of continuous processes defined on [0,7], such that:
1 Extension property: Jƒ (H:)o<:<7T is a simple process then
Pas WO<t<T, J(H): ~ 1(Hì
2 Continuity property: if (H ")a>o is a sequence of processes in H such that fe ( (H™)? ds converges to 0 in probability then SUP¿< |J(H");| converges to 0 m probability Consistently, we write J H, dW, for J(H) T = E (/ — ` 0 This last term converges to 0 by dominated convergence, therefore f lr¿<;„)H;ởW,
Stochastic tmiegral and ltô calculus 4I
Remark 3.4.7 It is crucial to notice that in this case ( J H,dW,) is not
0<t<T necessarily a martingale
Proof It is easy to deduce from the extension property and from the continuity
property that if H € ?( then P as V < T, J(H); = J(H):
Let H € H, and define T, = inf {0 <s <T, J) H2du > n} (+00 if that set
is empty), and H? = H, l{s<T„}-
Firstly, we show that T,, is a stopping time Since {7 < t} = { J HỆdu > n}, we just need to prove that J H?du is F,-measurable This result is true if H is a simple process and, by density, it is true if H € H Finally, if H € H, J H?du ~ is also F;-measurable because it is the limit of J( (Hụ„ A K)?du almost surely as K tends to infinity Then, we see immediately that the processes H? are adapted and bounded, hence they belong to H Moreover t t [ Hraw, = | 1:;<r„yH?!đ4W,, 0 0 and relation (3.5) implies that t tATn / H?dW, = | H?*!4W, D 0 0
Thus, on the set {fo HỆdu < nj, for anyt < T, J(H"), = J(H"*?), Since
Ua>o{f H?du < n} = {fo H?du < +00}, we can define almost surely a process J(H), by: on the set Uy HỆdu < n},
Wte<T J(H): = J(H"):
The process t + J(H); is almost surely continuous, by definition The extension property is satisfied by construction We just need to prove the continuity property
of J To do so, > we first notice that
P (super |I(H).| > ) < P (Jj H2ds > 1/N)
>2
If we call ty = inf {s <T, f> H2du > 1/N} (+00 if this set is empty), then
on the set {Jo H2du < 1/N}, it follows from (3.5) that, for any t < T,
| { f° H2du<i/N} SUP <T 7H)
to t
[ H,dW, = Ï(H); = J(H)), = Ỉ H}1t(,<„„)dW, = Ỉ H,1(s<„„)dW,,
Trang 2842 Brownian motion and stochastic differential equations whence, by applying (3.4) to the process s ++ Hs lts<ry} We get
> ) P( #4: > 5)
+4/c?E (5 H?1(„<-„)45)
T
P(Í HỆ > À )* le Ne
As a result, if Jo ( (HT?) ds converges to 0 in probability, then Sup;< l7 (H"):|
converges to 0 in (ot sly
In order to prove the linearity of J, let us consider two processes belonging to H, called H and K, and the two sequences H? and K?* defined at the beginning of the proof such that Se (H? — H,)*ds and fe (KP - K,)?ds converge to 0 in probability By continuity of J we can take the limit in the equality J(AH” +
pK"), = AJ(A"): + pJ(K")t, to prove the continuity of J
Finally, the fact thatif H € 71 then ƒ\( (H,— H?)?dt converge to 0in probability and the continuity property yields the uniqueness of the extension
IA
P (sup J(H): t<T
IA
We are about to summarise the conditions needed to define the stochastic integral with respect to a Brownian motion and we want to specify the assumptions that make it a martingale
Summary:
Let ‘us consider an F;- Brownian motion (W:):>o and an Z7¡-adapted process
(Aiosesr- We are able to define the stochastic integral (fo H,dW,)o<t<T as
soon as fe H?ds < +oo Pas By construction, the process (fo H,dW,)o<t<t is a martingale if E ( Se H ?ds) < +00 This condition is not necessary Indeed,
the inequality E ( fe H Ras) < +00 is satisfied if and only if
2 (sp (ƒ H,aW, ) < +00 0<t<7T ,
This is proved in Exercise 15
3.4.2 It6 calculus : TA
It is now time to introduce a differential calculus based on this stochastic integral Tt will be called the /t6 calculus and the main ingredient is the famous /t6 formula In particular, the It6 formula allows us to differentiate such a function as t + f (W,) if f is twice continuously differentiable The following example will simply show that a naive extension of the classical differential calculus is bound to fail Let us try to differentiate the function t > W? in terms of ‘dW,’ Typically, for a
Stochastic integral and Ité calculus 43 or itaft function f(t) null at the origin, we have f(t)? = 2 So F( s)ds = 2 So £( ) In the Brownian case, it is impossible to have a nae vernal W? =2 ed m dW; Indeed, from the previous section we know that J W,dW,
1s a martingale (because E ( J W?ds) < +œ), null at zero If it were equal to
W7? /2 it would be non-negative, and a non-negative martingale vanishing at zero can only be identically zero
We shall define precisely the class of processes for which the Itô formula is
applicable ,
Definition 3.4.8 Let(Q, F, (Ft):>0, P) bea filtered probability space and (W:)1>0 an Fy-Brownian motion (X+)o<t<r is an IR-valued It6 process if it can be written as t t Pas.Wt<T X= Xo+ | Kuds + | H,dW,, - 0 0 where e Xo is Fo- measurable ® (K:)o<:<r and (H:)o<¡<r a are F,-adapted processes ° fe |K,|ds < +oo Pas ° fe |H,|2ds < too Pas
We can prove the following proposition (see Exercise 16) which underlines the uniqueness of the previous decomposition
Proposition 3.4.9 If (M:)o<t<t is a continuous martingale such that t : T , : MM =Í[ K,ds, with Pas [ |K,|ds < +00, 0 ` ; 0 then , Pas.Ví<7, Mị =0 This implies that:
~ An It6 process decomposition is unique That means that if
t t t t
X, = Xo +/ K,ds +/ H,dW, = Xọ + [ ‘Kids +/ HidW,
0“ 0 0 0
then
X= Xi dPas H,=H' dsxdPae K,=K' dsxdPae
— If (Xt)o<esr is a martingale of the form Xo+ J K,ds + J H,dW,, then
K, = 0 dt x dP ae
We shall state It6 formula for continuous martingales The interested reader should refer to Bouleau (1988) for an elementary proof in the Brownian case, i.e when (W,) is a standard Brownian motion, or to Karatzas and Shreve (1988) for a
Trang 2944 Brownian motion and stochastic differential equations Theorem 3.4.10 Let (X)o<:<r be an Ité process t t Xi=Xe+ | K,ds-+ [ H.dW., 0 0 and f be a twice continuously differentiable function, then f(X1) = f(Xo) +f £%0 where, by definition ,)dX, + sf rx d(X,X) t_ (X,Xỳ,= [ H}ds, A [res dX, = [ 10%) Likewise, if (t,z) — ƒ(, z) is a function which is twice differentiable with respect and Keds +f FO) :)H.dW
to x and once with respect to t, and if these partial derivatives are continuous with » respect to (t, x) (i.e f is a function of class C}:?), It6 formula becomes t ` ƒŒ,X⁄¿) = £(0,Xo0) + [ fi(s, Xs)ds 0 t 1 t * + [ f(s,XodX,+g [ f2(s,X4(X,X), 0 0 ,
3.4.3 Examples: Ité formula in practice
Let us start by giving an elementary example If f(z) = x? and X; = W:, we
identify K, = 0 and H, = 1, thus t : ‘1 rt W? =2 [ W,dW, + = Ỉ Qds 0 2 0 t W?-t= 2 | W;dW, 0
It turns out that
Since E ( Ss W2ds) < +00, it confirms the fact that W2 — ¢ is a martingale We ‘now want to.tackle the problem of finding the solutions (5;):>0 of + ` t ¬ › ; S,= 20+ | Ss (uds + odW,) _ 4.8) 0 This is often written in the symbolic form dS, = S; (udt+odW,), So = Zo (3.9) We are actually looking for an adapted process (S:)¢>0 such that the integrals
Stochastic integral and It6 calculus 45 i S,ds and Ss S,dW, exist and at any time ¢
t t
Pas Sy = Zo + / pS,ds + [ oS,dw,
0 0
To put it in a simple way, let us do a formal calculation We write Y¿ = log(S;)
where 5S; is a solution of (3.8) S; is an It6 process with K, = 5; and H; = aSs
Assuming that S; is non-negative, we apply Ité formula to f(z) = log(z) (at least formally, because f(x) is not a C? function!), and we obtain ‘dS, 1 ///-1 1 og(S;) = log(So) S,) =1 + m +5/ ak S2ds _—5 — 2œ2 Using (3.9), we get t t Y, = Yo +f (u — ø?/2) at + [ odW,, 0 0 and finally Y; = log(S¿) = log(So) +
Taking that into account, it seems that
St = Zo exp ((u — ơ?/2)t+ ơW,)
1s a solution of equation (3.8) We must check that conjecture rigorously We have
St = fit, W,) with
(u — ø2/2)t+ ơWi
^ ƒŒ,#) = zo exp ((uT— ø?/2)t + ơz)
Itô formula is now applicable and yields Se = ƒ0,W.) f(0, Wo) + | fi(s, W.)ds “3 eo x3 [ gam d(W,W), Furthermore, since (W, W), = t, we can write t 1 t Si = 20 + [ Ss:(u _ ơ?/2) ds+ [ S,adW, + 7 S,ơ?ds - 0 0 - 0 In conclusion Si = to + " Seuds +f S,odw,
Remark 3.4.11 We could have obtained the same result (exercise) by applying
Ité formula to S; = j(Z,), with Z; = (u—o? /2)t-+-oW; (which is an Ité process) and j(z) —zo exp(z)
Trang 3046 Brownian motion and stochastic differential equations Proposition 3.4.12 (integration by parts formula) Let X and Yt be two lô pro-
cesses, Xt = Xot fy Ks ds +f, H,dW, andY, = Yo +f Kids +f HidwW, Then t t X% = Xo¥o+ | X.aY, + [ Y,dX, + (X,Y), 0 0 with the following convention t (X,Y): = / H,H'ds 0 Proof By Itô formula
(Xi + Y,)? = (Xot+ Yo)? +2 (X, +Y,)d(X, + Y,) + J(H, + HỊ)?ds X2 = X2+2|2X.dX,+ [) H?ds Y2 = Yệ+2ƒV,dV,+ ft Hi'ds By subtracting equalities 2 and 3 from the first one, it turns out that t - ¿ t Xe, = Xo¥o+ | x.dy, + [ vax, + | H,H;ds 0 O° 0 D We now have the tools to show that equation (3.8) has a unique solution Recall that
St = x0 exp ((u — 07/2) t+ o0W)
is a solution of (3.8) and assume that (Ã;);>o is another one We attempt to compute the stochastic differential of the quantity X, 15, ) Define
L= 2 = exp ((—u +ø?/2) t — ơW;),
t
p' = —u + Ø2 and ơ' = —-o Then 2 = exp(w _ a” /2)t + ơ'W,) and the
verification that we have just đone shows that e t t Z¿=1+ / Z,(p'ds + o'dW,) =1+ / Z,((=u+ a”) ds — odW,) 0 0 From the integration by parts formula, we can compute the differential of X12 d(X;:Z:) = X,dzZt + Z.dX%: + d(X, 2)
Stochastic integral and Ité calculus 47
In this case, we have * t (X,Z):= (f x,oaw,,~ [ Z,0dW,): = -Ƒ ơ?X,Z.ds 0 0 0 Therefore d(X¿Z) =_ X:Z:((Tu+ø?) dị —- odW;)
+_ X;Z¡ (udt + ơdW/,) — X;Z.ơ?dt =
Hence, X;Z; is equal to XoZa, which implies that
VWt>0, Pas X,;=202Z,' =S
The processes X; and Z, being continuous, this proves that
“Pas Wt>0, Xp=20Z,! = S
We have just proved the following theorem:
Theorem 3.4.13 if we consider two real numbers a, 2 and a Brownian motion
(W:)t>0 and a strictly positive constant T, there exists a unique It process
(St)o<e<r which satisfies, for anyt < T, t St = Xo +[ S, (uds + odW,) 0 This process is given by ^J Sy = ao exp ((u — ø2/2) t + ơW,).- Remark 3.4.14
e The process S; that we just studied will model the evolution of a stock price in
the Black-Scholes model
e When pu = 0, S; is actually a martingale (see Proposition 3.3.3) called the exponential martingale of Brownian motion
Remark 3.4.15 Let © be an open set in IR and (Xt)o<e<7 an It6 process which stays in © at all times If we consider a function f from © to R which is twice continuously differentiable, we can derive an extension of It6 formula in that case
;œ%: f(Xo) + [res s)dX; + ; J2 X,)Hds
This result allows us to apply Itô formula to log(X;) for instance, if Ã; 1s a strictly
positive process
3.4.4 Multidimensional Ité formula
We apply a multidimensional version.of It6 formula when f is a function of'several
Trang 3148 Brownian motion and stochastic differential equations version will prove to be very useful when we model complex interest rate structures for instance
Definition 3.4.16 We call standard p-dimensional F;-Brownian motion an RRP -
valued process (W, = (W}, ,W?))>0 adapted to F,, where all the (W, 2)t>o0 are independent standard 7ị- Brownian motions
Along these lines, we can define a multidimensional Ité process
Definition 3.4.17 (Xt)o<t<r is an Ité process if
t Pp t
Xi = Xe + K,ds+S” | HịdW)
0 i=1 49
where:
e K, and all the processes (H}) are adapted to (Ft)
ef |Ke|ds < +00 Pas e fe (Hi) ds < +00 Pas Ité formula becomes: Proposition 3.4.18 Let (X}, , oo t - P t R x} =xi+ / Kids+ > [ Hi dw) 0 jai 10
then, if f is twice differentiable with respect to x and once differentiable with
respect to t, with continuous partial derivatives in (t, r) X?') be n Ité processes t ƒt,XỊ, ,X?)= 5.(5,X1,: , XP)ds : a 0 t Lá 5a! (s,X}, ,X2)dXi 2 suf a (s,X; ,Xz)d(X',X?), LiL; with: ` % 0 dX} = Kids + 57-¡ Hệ? dWj, 0 d(X', XI), = SP _, Hi" Hb ds
Remark 3.4.19 If (X, Jocrer and (Ÿ;)o<¿<7 are two Itô processes, we can de
fine formally the cross-variation of X and Y (denoted by (X,Y);) through the following properties:
e (X,Y) is bilinear and symmetric
© ({j Keds, X.)t = Oif Mosesr | is an Ité process |
Stochastic differential equations 49
o (fj HedWij, [> HidW?): = Oifi 43
o (f; HedWij, fo HidW)): = Jo HeHids
This definition leads to the cross-variation stated in the previous proposition 3.5 Stochastic differential equations
In Section 3.4.2, we studied in detail the solutions to the equation ‘ot X;=z+ [ X,(puds + odW,) 0 We can now consider some more general equations of the type t t X,=Z +/ b(s, X,)ds +t[ ơ(s, X,)dW, 0 0
These equations are called stochastic differential equations and a solution of (3.10) is called a diffusion These equations are useful to model most financial assets, whether we are speaking about stocks or interest rate processes Let us first study some properties of the solutions to these equations :
(3.10)
3.5.1 Hô theorem
What do-we mean by a solution of (3.10)?
Definition 3.5.1 We consider a probability space (Q, A,.P) equipped with a fil-
tration (F:)t>0 We also have functions b : IRỶxIR> Ro: RtxR>Ra Fo-measurable random-variable Z and finally an F,- Brownian motion (W:):>o A solution to equation (3.10) is an 7-adapted stochastic process (Ä:):>o that Salisfies:
e Foranyt > 0, the integrals [ÿ b(s, X,)ds and So als, X,)dW, exist [Ws xolas < eoana f° Iz(s,X,)Ÿ ds < +00 Pas © (X1t)t>0 Satisfies (3.10), i.e + t t w29 Pas Xi=Z+ | b(s,X,)đs+ Í ơ(s,X,)dW l ` 0 0, Remark 3.5.2 Formally, we often write equation (3.10) as dX, = b(t, Xt) dt +a (t, Xt) dw, Xo = 2
The following theorem gives sufficient conditions on b and o to guarantee the
existence and uniqueness of a solition of equation (3.10)
Theorem 3.5.3 Jf b and o are continuous functions and if there exists a constant
Trang 3250 Brownian motion and stochastic differential equations 1 |b(t, x) — b(t, y)| + lo(t, x) — ø(,w)| < K|z — vị
2 |b(,z)| + |ơ(Œ,z)| < K( + |z|)
3 E(Z2}< +oo
then, for any T > 0, (3.10) admits a unique solution in the interval [0, T] Moreover, this solution (Xs)o<s<T satisfies
B( sup Ix.f) < +00 0<s<T
The uniqueness of the solution means that ƒ (X:)o<t<r and (Y:)o<t<T đre two
solutions oƒ (3.10), then Ð a.s VŨ St ST, Xp = tị
Proof We define the set
£ = {doses ¥,-adapted continuous processes, such that E (sup XP) < +00}
s<T
Together with the norm ||X||= (E (supe<:<z |X:|?)) 1⁄2 £ 1s acomplete normed vector space In order to show the existence of a solution, we are going to use
the theorem of existence of a fixed point for a contracting mapping Let ® be the
function that maps a process (X;)o<s<T into a process (®(X)s)o<s<7 defined
by t : t : :
6(X), = z+ Ws, Xe)ds+ | a(s,X,)dW, 0 0
If X belongs to €, (X) is well defined and furthermore if X and are both in
€ we can use the fact that {a +b)? < 2(a? + 6?) to write that
lð(X),— 8(V)# < 2 (supeceer [Jo (0(s, Xs) ~ bs, Y,))ds[
1)
+ §UPo<;<7 li, (s, X, )— ø(s, Y;))
and therefore by inequality (3.4)
B (sap |#(X) - #24) s<T
< an (sp, (ƒ |b(s, X;) — Đ(s, —we,Y2Ie) +8E ([ (a(s, Xs) — o(s, Ys)’ 24)
2(K?T2? + #er)B ( sup |X¿ — vi)
0<t<T
IA
Stochastic differential equations — 51
whenee ||&(X) #®(Y)|| < (2(K?T? + 4K?7))!/? ||X — Y|| On the other
hand, if we denote by 0 the process that is identically equal to zero and if we notice that (a + b +c)? < 3(a? + b? + c?) t 2 [ o(s,0)dW, 0 t 2 Ỉ b(s, 0)ds 0 E ( sup I#(0).P) < 3(E(Z’) + K?T? +4K°T) < +00 0<t<T + sup |#(0);|? <3 (z + sup 0<:t<T 0<:<T Therefore
We deduce that ® is a mapping from € to € with a Lipschitz norm bounded from above by k(T) = (2(K?T? + 4K?T))'/2 If we assume that 7' is small enough
_ so that k(T) < 1, ít turns out that ® is a contraction from € into € Thus it
has a fixed point in € Moreover, if X is a fixed point of ®, it is a solution of
(3.10) That completes the proof of the existence for T small enough On the other
hand, a solution of (3.10) which belongs to € is a fixed point of 6 That proves
the uniqueness of a solution of equation (3.10) in the class € In order to prove the uniqueness in the whole class of It6 processes, we just need to show that a solution of (3.10) always belongs to € Let X be a solution of (3.10), and define = inf{s > 0, |X,| > n} and f"(t) = E (supg<,ciar, |X|”) It is easy to
check that f”(t) is finite and continuous Using the same comparison arguments
as before, we can say that So ,; :
E (supocuctat, Xul’) S$, 9(Be +E ( fy" K(1+ |X Dds) +4B (f° K7(1 + |X.1)?ds) ) 3 (B(Z?) + 2(K?T + 4K?) x fo (1 +E (suPo<ucear, IXxl?)) ds) This yields the following inequality IA - mm ƒ*{s
In order to complete t the proof, let us recall the Gronwall lemma
Lemma 3.5.4 (Gronwall F0 If f is a continuous function such that for any
0<t<Tf, (ose s)ds, then f(T) < a(1 +e"), Proof Let us write u(t) = e~** ‘So £¢ s)ds Then,
a ul (t) = e~ (f(s) - of f(s)ds) < ae~™
`
By first-order integration we obtain u(T) < a/b and f(T) < a(1 + e°?) Oo
Trang 3352 Brownian motion and stochastic differential equations In our case, we have f"(T') < K < +00, where K is a function of T independent
of n It follows from Fatou lemma that, for any T’,
B ( sup Ix.f) < K < +00
0<s<T
Therefore X belongs to £ and that completes the proof for small 7' For an arbitrary T, we consider a large enough integer and think successively on the intervals
[0, T'/n], [T/n, 2T/n] |[[n — 1)T/n, TỊ Oo
3.5.2 The Ornstein-Ulhenbeck process
Ornstein-Ulhenbeck process is the unique solution of the following equation:
dX = —cX,dt + odW, Xo = @£
It can be written explicitly Indeed, if we consider Y, = X,e and integrate by parts, it yields
dY, = dX;e* + X;d(e°) + d(X, e°)y Furthermore (X,e%); = 0 because đ(e°?) = ce°td¿ It follows that dY, = ơe°tđjWW; and thus : : t X:=ze t+ ae [ e°°dW, 0 ` This enables us to compute the mean and variance of X;: t E(X;) = cet 4+ ge~“E (Í caw.) — tet , 0
(since E (Jo (e*)?as) < +00, J e°°dW, is a martingale null at time 0 and therefore its expectation is zero) Similarly
Var(X:) = E (Ge — E(X.))ˆ) = ơ?E Ga (ƒ caw.) ) " ñ ~2ct = ơ? lize" 1 ` 2c
We can also prove that X; is a normal random variable, since X; can be written as
J ƒ(s)dW, where ƒ(.) is a deterministic function of time and J f?(s)ds < +oo (see Exercise 12) More precisely, the process (Xt)>0 is Gaussian This means
Stochastic differential equations 53 that if A,, , An are real numbers and if 0 < t; < - < t,, the random variable Ar Xt, +++: +AnXt, is normal To convince ourselves, we just notice that
+00 t
X,, = ce + oe [ 1{;<¿,y€°°dW, =m; + [ fi(s)dW,
0 0
Then Xt, +++ + AnXe, = Dy Ams + fo (OL, Acfi(s)) dW, which is
indeed a normal random variable (since it is a stochastic integral of a deterministic function of time)
3.5.3 Multidimensional stochastic differential equations
The analysis of stochastic differential equations can be extended to the case when
’ processes evolve in IR” This generalisation proves to be useful in finance when
we want to model baskets of stocks or currencies We consider
e W = (W!, , WP) an IRP-valued Z;-Brownian motion e b:IRT xIR” OIR”,b(s,z) = (b1(s,z), -, b*{(s, z))
e z:TRỶ x]IR*” — IR^XP (which is the set of n x p matrices),
Ø(3,) = (Ø,7(s, Z))1<i<n ,1<j<p-
e Z=(Z', ,Z") an Fy-measurable random variable in IR”
We are also interested in the following stochastic differential equation:
t t
Xi= 2+ b(3,X,)ds+ f ơ (s, X,) dW; (3.11)
0 0
In other words, we are looking for a process (X;)o<¿<r with values in IR”,
adapted to the filtration (Z;);>o and such that P as , for any ¢ and for any ¿ < n
t Prt
xXp=H Zt [ bi(s, X,)ds + >| Ø:,;(s, X,;)dWỷ 0 jan J0
The theorem of existence and uniqueness of a solution of (3.11) can be stated as:
Theorem 3.5.5 If x € IR", we denote by |x| the Euclidean norm of x and if
Trang 3454 Brownian motion and stochastic differential equations 3.5.4 The Markov property of the solution of a stochastic differential equation The intuitive meaning of the Markov property is that the future behaviour of the process (X;)¢>o after t depends only on the value X; and is not influenced by the history of the process before t This is a crucial property of the Markovian model and it will have great consequences in the pricing of options For instance, it will allow us to show that the price of an option on an underlying asset whose price is Markovian depends only on the price of this underlying asset at time t
Mathematically speaking, an 7;-adapted process (X:):>o satisfies the Markov
property if, for any bounded Borel function f and for any s and £ such that s < , we have
E(f (Xt) |Fs) = E (f (Xe) |Xs)-
We are going to state this property for a solution of equation (3.10) We shall
denote by (X!*, s > t) the solution of equation (3.10) starting from = at time t
and by X* = X°* the solution starting from z at time 0 For s > t, X* satisfies Xt =24 / _b(u, XÉ®) du + / o (u, Xb") aW,, t t
A priori, X** is defined for any (t, x) almost surely However, under the assump-
tions of Theorem 3.5.3, we can build a process depending on (t,x, s) which is
almost surely continuous with respect to these variables and is a solution of the
previous equation This result is difficult to prove and the interested reader should refer to Rogers and Williams (1987) for the proof ,
The Markov property is a consequence of the flow property of a solution of a
stochastic differential equation which is itself an extension of the flow property of
solutions of ordinary differential equations
Lemma 3.5.6 Under the assumptions of Theorem 3.5.3, ifs >t tXƑ XoF =X, ‘ Pas Proof We are only going to sketch the proof of this lemma For any «x, we have 8 Pa.s X” =z+ / b (u, Xt") du + / ơ (u, X;”) dWa t t It follows that, Pas for any y € R, s§ s§ X‡Y=ụ+ / b(u, XS") du+ / ơ (u, X$*) dWu, St t and also
xi =Xƒ+ [i (u, x.) du + Ƒs (u, xu") dW |
These results are intuitive, but they can be proved rigorously by using the continuity
of y +» X*¥, We can also notice that X° is also a solution of the previous equation no Stochastic differential equations 55 Indeed, if t < s XP? = z+ So b(u, XZ) du + So o (u, XZ) dW = Xƒ+ ƒ, b(u,X?) du + [` ø (u,X£) dW, The uniqueness of the solution to this equation implies that XP“ = X?*' for t<s 0
In this case, the Markov property can be stated as follows:
Theorem 3.5.7 Let (X1)t>0 be a solution of (3.10) It is a Markov process with
respect to the Brownian filtration (F;)t>0 Furthermore, for any bounded Borel function f we have P as E(f (Xt) |Fs) = (Xs), with $(z) = B (ƒ(X?”)) Remark 3.5.8 The previous equality is often written as E (f (%X;) \Fs) =E ((X?”))|,~x.,
Proof Yet again, we shall only sketch the proof of.this theorem For a full proof,
the reader ought to refer to Friedman (1975)
The-flow property shows that, ifs < t, Xf = xX; Ke On the other hand, we
can prove that X;°* is a measurable function of x and the Brownian increments (W.4u — Ws, u > 0) (this result is natural but it is quite tricky to justify (see
Friedman (1975)) If we use this result for fixed s and t we obtain X;/°* = ®(z,W;+„ —W;; u > 0) and thus
Xf = 0(X2,Weiu — We; u > 0),
where X} is ¥,-measurable and (W.4 — W;)„>o is independent of F, If we apply the result of Proposition A.2.5 in the Appendix to X,, (Ws4u — W;)„>o, ®.and F,, it turns out that
E(f (B(XE, Wein — Wes w > 0))|Z,) = EB(f (Ole, Wau — We; U> O)leaxs
.= E (f (XP) )eaxe
mu
The previous result can be extended to the case when we consider a function of the whole path of a diffusion after time s In particular, the following theorem is useful when we do computations involving interest rate models
Theorem 3.5.9 Le (X;);¿>o be a solution of (3.10) and r(s, x) be a non-negative
Trang 3556 ˆ Brownian motion and stochastic differential equations measurable function Fort > s Pas B (eS s(x) |F,) = HX) with ~ fir uX a" )du $,z #2) = B (c Í n2 (Xe) ) , It is also written as
E (J r(uXu)du (Xt) Fe) -E (=! ror (xe)
Remark 3.5.10 Actually, one can prove a more general result Without getting into the technicalities, let us just mention that if ¢ is a function of the whole path of X; after time s, the following stronger result is still true:
Pas E (2X, t > 8) |Zs) = E(¿(X?”, t 2 8))Ì„—x, `
Remark 3.5.11 When ö and ø are independent of z (the diffusion is said to be
homogeneous), we can show that the law of X:%, is the same as the one of X}*,
which implies that if f is a bounded measurable function, then E (F(X) = E (f(x?)
We can extend this result and show that if r is a function of x only then
s+t su t = l
E (J r( x3 (X38) =E (= + r(x? * (x2*))
In that case, the Theorem 3.5.9 becomes ,
E (=! r(Xu)du (X;) Fe) =E Gus rome (x08) r=X, z=X, 3.6 Exercises Exercise 6 Let (M:)t>0 be a martingale such that for any t, E(M?) < +00 Prove that if s <t , E (Me — Ms)’|F.) = E(M? — M3|F,)
Exercise 7 Let Xz be a process with independent stationary increments and zero initial value such that for any t, E (X; 2) < +00 We shall also assume that the
mapt > E (X2) 1s continuous Prove that E (X;) = c‡ and that Var(X;) = c’t, where c and c’ are two constants
Exercise 8 Prove that, if r is a stopping time,
F, ={AEA, forallt>0,AN{r <t} Ee Fi}
is a o-algebra
Exercises 57
Exercise 9 Let S be a stopping time, prove that S is Fs-measurable
Exercise 10 Let S and T be two stopping times such that S < T Pas Prove that Fs C Fr
Exercise 11 Let S be a stopping time almost surely finite, and (Xt)t>0 be an adapted process almost surely continuous
1 Prove that, P a.s., for any s
X,= lim S2 Ta (6+a)/n((9)XXk/a (6) r:i>-+co k>0 2 Prove that the mapping ({0, t] x 2, B([0, t]) x Fk) — GR, B(R)) (s,w) + : is measurable 3 Conclude that if SŠ < ý, Xs is F,-measurable, and thus that Xg is Fs- measurable
Exercise 12 This exercise is an introduction to the ak, of stochastic integra- tion We want to build an integral of the form f° f(s)dX,, where (X;);>o 1s an
F,-Brownian‘motion and if là is a measurable function from (IR*, B(IRT)) into (IR, BUIR)) such that f° 2(s)ds < +00 This type of integral is called Wiener
integral and it isa Da nulse case of Itô integral which is studied in Section 3.4 We'recall that the set 1 of functions of the form Do<icn—1 Fe; ,t241]> With
a; € R, and tp = 0 < ty < - < ty is dense in the space 120RẺ, đz) endowed 1/2 °/20)"2 1 Consider a; € IR, and 0 = tp < ty < + + < ty, and call f = > Oj] je, ¢544]- O<i<N-1 with the norm || fi] ,2 = (, We define I.(f) = > ai(Xt,,, — X¢;) O<i<N-1
Prove that I.(f) is a normal random variable and compute its mean and vari-
ance In particular, show that
E(Ie(f)’) = lI/lla-
2 From this, show that there exists a unique linear mapping I from L?(IR*, dz) into L?(0, F, P), such that I(f) = IIfllz2, fot any f in L?(R*) I,(f), (f), when f belongs to H and E(1(ƒ)2) when ƒ belongs to 1 and E(I
3 Prove that if (Xa)n>o is a sequence of normal random variables with zero-
Trang 3658 Brownian motion and stochastic differential equations
variable with zero-mean Deduce that if f € L?(IR*, dz) then I(f) is a normal
random variable with zero mean and a variance equal to i ƒ?(s)ds
4 We consider f € L?(IR*, dx), and we define
Z,= [ ƒ(s)dX, = J 1o,g(s)f(9)dÄX,,
prove that Z; is adapted to Fz, and that Z, — Z, is independent of + (hint: -
begin with the case ƒ € H)
5 Prove that the processes Z,, Z? — J f?(s)ds, exp(2: — ÿ " ƒ?(3)ds) are
Z;-martingales
Exercise 13 Let T be a positive real number and (Mz)o<t<r be a continuous
F,-martingale We assume that E(M/?) is finite 1 Prove that (|M:z|)o<e<7 is a submartingale
2 Show that, if M* = supy<per |Mil,
AP (M* >) SE (|Mr|lar->a}) -
(Hint: apply the optional sampling theorem to the submartingale | M;| between
+ AT and T where 7 = inf{t < T,|M,| > A} (if this set is empty 7 is equal
to +00).) :
3 From the previous result, deduce that for positive A
E((M A 4)?) < 2B((M* A A)|Mrl)
(Use the fact that (Mƒ* A A)? = An
4 Prove that E(M*) is finite and pzP~ ldz for p = 1,2.) B( sup is) < 4E(|Mr|’) O<t<T Exercise 14 s
1 Prove that if S and S’ are two Z-stopping times then Š A Š° = inf(S, 5”) and SV S' =sup(S, S’) are also two F;-stopping times
2 By applying the sampling theorem to the stopping time S V s prove that E (Ms l{s>s}|Fs) = M.1ts>s} 3 Deduce thatfors <S¿- _ E (Msatlts>s}|Fs) = Ms1¢s>5}- 4 Remembering that Mga, is F,-measurable, show that t 3 Mgaz is an F;- martingale Exercises 59 Exercise 15 1 Let (Ht)o<t<7 be an adapted measurable process such that L HỆdt < œ, t a.s Let Mp = J H,dW, Show that if E (suPo<¿<r M2) < oo, then T tra wo TK
E ( to A; dt) < oo Hint: introduce the sequence of stopping times 7, = inf{t >0| fj H2ds = n} and show that E(M?,,,) =E (0 Hds) 2 Let p(t,2) = 1//1 - texp(—2x?/2(1 — t)), forO < t < 1 and z € IR, and
p(1, 2) = 0 Define M; = p(t, W,), where (W;)o<¢<1 is standard Brownian motion (a) Prove that : t Op M=M — t a+ [ Bq (8 W,)dW (b) Let 8 H, — Arts W,)
Prove that Lo H2dt < 00, a.s and E (6 Hat) = +,
Exercise 16 Let (M.)o<:<r be a continuous Z7;-martingale equal to J K,ds,
where (;)o<¿<7 is an 7¿-adapted process such that fe |K;|ds < +œ Pas
we T '
1 Moreover, we assume that Pais to |K,|ds < C < +00 Prove that if we write t? = Ti/n forO0 <i < n, then
lim E (>> (w; - Me.) "= +1 ,
2 Under the same assumptions, prove that
E (= (Muy - Me.,)') = B (M3 - M3) i=l
Conclude that M7; =0 Pas.,andthus Pas Vt < T, M;, = 0
T i 4
3 J, |Ks|ds is now assumed to be finite‘almost surely as opposed to bounded n y as Opp | We shall accept the fact that the random variable f |&,|ds is F,-measurable
Show that T,, defined by
‹
t
T„ =inf{0 < s< r, [ |Ks{ds > n} - 0
(or T if this set is empty) is a stopping time Provethat P as limy.400 Tn = T Considering the sequence of martingales (M7, )e>0, prove that
Trang 3760 Brownian motion and stochastic differential equations
4 Let M; be a martingale of the form Ss H,dW, + Ss K,ds with Ss Hids <
.+00 Pas and Ss |K,|ds < +oo Pas Define the sequence of stopping times T, = inf{t < T, fs H?ds > n}, in order to prove that K; = 0 dt x Pas
Exercise 17 Let us call X; the solution of the following stochastic differential
equation
0
We write S; = exp ((u — ơ?/2)‡ + ơW/,)
1 Derive the stochastic differential equation satisfied by S,- 1 2 Prove that
{ dX, = (pXitp')dt+ (0X, + 0')dW,
d(X,S71) = S71 ((u! — o0')dt + o'dW)
3 Obtain the explicit representation of X;
Exercise 18 Let (W2)1>0 be an F;-Brownian motion The purpose of this exercise is to compute the law of (W:, sup, <; Ws)
1 Consider S a bounded stopping time Apply the optional sampling theorem to
the martingale M, = exp(izW; + 27/2), where z is a real number to prove
that if 0 <u < v then
E (exp (iz(Wu4s5 — Wuts)) |Futs) = exp (—z?( — u)/2)
2 Deduce that WS = Wu4s — Ws is an -F54u-Brownian motion independent
of the ơ-algebra Fs :
3 Let(Y¿);>o be a continuous stochastic process independent of the ø-algebra
such that E(supo<.<x |¥s|) < +00 Let T be a non-negative B-measurable
random variable bounded from above by K: Show that
E(¥r|B) = E(Y)hier-
We shall start by assuming that J can be written as 3°, <;<, tila,, where
0< < - < tạ = K,and the A4; are disjoint B-measurable sets
4 We denote by 7? the inf{s > 0, W, > A}, prove that if f is a bounded Borel
function we have *
E (F(W)1 rcey) =E (przez olt _ r^))
where ó(u) = E(f(W,, + A)) Notice that E(f(W + À)) = B(ƒ(—Wv +À))
and prove that =
B (FW) acy) = B(F2- Wel prcy)
5, Show that if we write W? = sup,c, W and if A > 0
Trang 384 The Black-Scholes model
Black and Scholes (1973) tackled the problem of pricing and hedging a European option (call or put) on a non-dividend paying stock Their method, which is based on similar ideas to those developed in discrete-time in Chapter 1 of this book, leads to some formulae frequently used by practitioners, despite the simplifying character of the model In this chapter, we give an up-to-date presentation of their work The case of the American option is investigated and some extensions of the model are exposed
4.1 Description of the model 4.1.1 The behaviour of prices -
The model suggested by Black and Scholes to describe the behaviour of prices is a continuous-time model with one risky asset (a share with price 5; at time £) and a riskless asset (with price S? at time t) We suppose the behaviour of S? to be encapsulated by the following (ordinary) differential equation:
dS? =rSedt, | (4.1)
where r is a non-negative constant Note that r is an instantaneous interest rate and should fot be confused with the one-period rate in discrete-time models We
set S? = 1, so that SP? = e”* fort > 0
We assume that the behaviour of the stock price is determined by the following stochastic differential equation:
dS, = S; (udt + odB,), (4.2)
where yz and o are two constants and (B;) isa standard Brownian motion
The model is valid on the interval [0,7] where T is the maturity of the option
As we saw (Chapter 3, Section 3.4.3), equation (4.2) has a closed-form solution
2
Trang 3964 The Black-Scholes model where So is the spot price observed at time 0 One particular result from this model
is that the law of S; is lognormal (i.e its logarithm follows a normal law) More precisely, we see that the process (S;) is a solution of an equation of type
(4.2) if and only if the process (log(5;)) is a Brownian motion (not necessarily standard) According to Definition 3.2.1 of Chapter 3, the process (St) has the
following properties:
e continuity of the sample paths; |
e independence of the relative increments: if u < t, St /S or (equivalently), the
relative increment (S; — S,)/S., is independent of the o-algebrao(S.,v < u); e stationarity of the relative increments: if u < t, the law of (S; — Su)/Su is
identical to the law of (S:-u — So)/So
These three properties express in concrete terms the hypotheses of Black and Scholes on the behaviour of the share price
4.1.2 Self-financing Strategies
A strategy will be defined as a process ¢ = (Ó:)o<:e<r=((H?, H,)) with values in IR?, adapted to the natural filtration (F;) of the Brownian motion; the components
H? and H, are the quantities of riskless asset and risky asset respectively, held in the portfolio at time t The value of the portfolio at time ¢ is then given by
Vi (¢) = Hp S? + HeSt
In the discrete-time models, we have characterise self financing strategies by the
equality: Visi(¢) — Va(¢) = on41-(Sn41 — Sn) (see Chapter 1, Remark 1.1.1)
This equality is extended to give the self-financing condition in the continuous- time case > dv, (¢) = H?ds? + H,dS; To give a meaning to this equality, we-set the condition T T | |H?|dt < too as | H?dt < +oœo a5 0 0 T Lo T Ỉ HỆdS? = [ H?re"dt 0 0 is well-defined, as is the stochastic integral T „ T °T [ HdS; = [ (Hi Sip) dt +/ oH, S.dB, 0 0 0
since the map t ++ S; is continuous, thus bounded on [0, T] almost surely
Definition 4.1.1 A self-financing strategy is defined by a pair of adapted pro- cesses (HP) c, cp and (Hi)o<t<T satisfying:
Then the integral
Change of probability Representation of martingales 65
T T
W IPla+ [ Hệ dt < +œ a.s
2 HPS? + HS; = HOS° + HoSo + [ Hodse + +f HudS, a.s
for allt € [0, TỊ
We denote by 5; = e~*S; the discounted price of the risky asset The following
proposition is the counterpart of Proposition 1.1.2 of Chapter 1
Proposition 4.1.2 Let ¢ = (CHP ,H:))o<tcr be an adapted process with values
in IR’, satisfying fy |HP|dt+ fr H?dt < +ooa.s We set: Vi(¢) = H°S9+H,S,
and V,() = e-**V,(¢) Then, ¢ defines a self-financing strategy if and only if - t Ủ,(9) = Wa(đ) + [ Hyd, as (4.3) for allt € [0,7] Proof Let us consider the self-financing strategy ¢ From equality _ dvi (¢) = —rV,(d)dt + e~"*dVi (4)
which results from the differentiation of the product of the processes (e~"') and (Vi(#)) (the cross-variation term d(e~", V.(#))4 is null), we deduce
dV,(¢) = ore" (HPe" + H,S;) dé + e""' HP d(e rty 4 e”tH,4S, = H, s re~”!S,dt + e~”!4S,)
= H,.d&,,
which yields equality (4.3) The converse is justified similarly O Remark 4.1.3 We have not imposed any condition of predictability on strategies unlike in Chapter 1 Actually, it is still possible to define a predictable process in
continuous-time but, in the case of the filtration of a Brownian motion, it does not
restrict the class of adapted processes significantly (because of the continuity of sample paths of Brownian motion)
In our study of complete discrete models, we had to consider at some stage a prob- ability measure equivalent to the initial probability and under which discounted prices of assets are martingales We were then able to design self-financing strate- gies replicating the option.‘The following section provides the tools which allow us to apply these methods in continuous time
4.2 Change of probability Representation of martingales 4.2.1 Equivalent probabilities
Let (2,.A,P) be a probability space A probability measure Q on (2, A) is
absolutely continuous relative to P if
Trang 4066 , The Black-Scholes model
Theorem 4.2.1 Q is absolutely continuous relative to P if and only if there exists
a non-negative random variable Z on (1, A) such that
VAEA ata) = | Z(w)dP(w)
A
Z is called density of Q relative to P and sometimes denoted dQ/dP
The sufficiency of the proposition is obvious, the converse is a version of the Radon-Nikodym theorem (cf for example Dacunha-Castelle and Duflo (1986), Volume 1, or Williams (1991) Section 5.14)
The probabilities P and Q are equivalent if each one is absolutely continuous
relative to the other Note that if Q is absolutely continuous relative to P, with
density Z, then P and Q are equivalent if and only ifP(Z>0)=1
4.2.2 The Girsanov theorem
Let (9, +, đo <:<T ,P) be a probability space equipped with the natural fil- tration of a standard Brownian motion (B:)o<,<7» indexed on the time interval
[0, 7'] The following theorem, which we admit, is known as the Girsanov theorem (cf Karatzas and Shreve (1988), Dacunha-Castelle and Duflo (1986), Chapter 8) Theorem 4.2.2 Let (6:)o<e<r be an adapted process satisfying fe 62ds < 00
a.s and such that the process (Lt)o<t<T defined by
t 1 t It = exp (- [ 6,dB, — >| 02s)
0 0
is a martingale Then, under the probability P() with density Lr relative to P, the process (Wi )o<t<r defined by W, = Bet+ J 6,ds, is a standard Brownian motion : Remark 4.2.3 A sufficient condition for (Lt)o<e<r to be a martingale is: 1 /Ÿ - E | exp 5 [ 62 dt < 00, : 0
(see Karatzas and Shreve (1988), Dacunha-Castelle and Duflo (1986)) The proof of Girsanov theorem when (6;) is constant is the purpose of Exercise 19 _ 4.2.3 Representation of Brownian martingales
Let (Bi)o<e<r be a standard Brownian motion built on a probability space (0, F,P) and let (F:)o<e<r be its natural filtration Let us recall (see Chap-
ter 3, Proposition 3.4.4) that if (Hz)o<e<7 is an adapted process such that E ( fe Hat) < oo, the process ( J H,dB,) is a square-integrable martin- gale, null at 0 The following theorem shows that any Brownian martingale can be
represented in terms of a stochastic integral
Pricing and hedging options in the Black-Scholes model 67
Theorem 4.2.4 Let (M:)o<:<r be a square-integrable martingale, with respect
to the filtration (Fy)o<t<r There exists an adapted process (H,)o<t<r such that E (A H?ds) < +00 and
t
Vt € [0,7] Mp= Mot Ỉ H,dB, as (4.4)
0
Note that this representation only applies to martingales relative to the natural filtration of the Brownian motion (cf Exercise 26)
From this theorem, it follows that if U is an Zr-measurable, square-integrable random variable, it can be written as
T
U=E(U) +f H,dB, as., 0
where (H;) is an adapted process such that E ( fe H}ds) < +00 To prove it,
consider the martingale M, = E (U|F,) It can also be shown (see, for example,
Karatzas and Shreve (1988)) that if (M:)o<:<r is a martingale (not necessarily square-integrable) there is arepresentation similar to (4.4) with a proce$s satisfying
only f, H?ds < oo, a.s We will use this result in Chapter 6
4.3 Pricing and hedging options in the Black-Scholes model 4.3.1 A probability under which (5) is a martingale
We now consider the model introduced in Section 4.1 We will prove that there exists a probability equivalent to P, under which the discounted share price S; = e—*'S, is a martingale From the stochastic differential equation satisfied by (S;), we have dS, = —re"'!S,dt+e "tdS, = St ((u — r)dt + odB;) % oe Consequently, if we set W, = B, + (u—r)t/o, dS, = SiodW, (4.5)
From Theorem 4.2.2, with 6, = (—r)/o, there exists a probability P* equivalent
to P under which (W;)o<t<7 is a standard Brownian‘ motion We will admit that the definition of the stochastic integral is invariant by change of equivalent probability (cf Exercise 25) Then, under the probability P*, we deduce from
equality (4.5) that (S;) is a martingale and that