Lecture notes for course

Stochastic Processes for Finance

Contributed by
F. Boshuizen, A. van der Vaart, H. van Zanten, K. Banachewicz, P. Zareba and E. Belitser

Last updated: December 18, 2014

Contents
1 Pricing Derivatives
1.1 Note on Continuous Compounding . . . . . . . . . . . . . . . . .
1.2 Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Hedging a Forward . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Binomial Model
2.1 One Period Model
2.2 Two Period Model
2.3 Portfolios . . . . .
2.4 N Period Model .

1
1
2
3

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

6
6
8
9
10

3 Discrete Time Stochastic Processes

3.1 Stochastic Processes . . . . . . . .
3.2 Conditional Expectation . . . . . .
3.3 Filtration . . . . . . . . . . . . . .
3.4 Martingales . . . . . . . . . . . . .
3.5 Change of Measure . . . . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

13
13
14
15
16
16

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

4 Binomial Model Revisited

18
4.1 Martingales in the Binomial Model . . . . . . . . . . . . . . . . . 18
4.2 Pricing and Hedging . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.3 Towards Continuous Time . . . . . . . . . . . . . . . . . . . . . . 22
5 Continuous Time Stochastic Processes
5.1 Stochastic Processes . . . . . . . . . .
5.2 Brownian Motion . . . . . . . . . . . .
5.3 Filtrations . . . . . . . . . . . . . . . .
5.4 Martingales . . . . . . . . . . . . . . .
5.5 Generalized Brownian Motion . . . . .
5.6 Variation . . . . . . . . . . . . . . . .
5.7 Stochastic Integrals . . . . . . . . . . .
5.8 Geometric Brownian Motion . . . . . .
5.9 Stochastic Differential Equations . . .
5.10 Markov Processes . . . . . . . . . . . .
5.11 Quadratic variation revisited . . . .
5.12 It
o Formula . . . . . . . . . . . . . . .
5.13 Girsanovs Theorem . . . . . . . . . .
5.14 Brownian Representation . . . . . . .

ii

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

24
24
24
25
26
26
26
28
28
29
29
30
31
32
34

6 Black-Scholes Model
6.1 Portfolios . . . . . . . . . . . . . . .
6.2 The Fair Price of a Derivative . . . .
6.3 European Options . . . . . . . . . .
6.4 The Black-Scholes PDE and Hedging
6.5 The Greeks . . . . . . . . . . . . . .
6.6 General Claims . . . . . . . . . . . .
6.7 Exchange Rate Derivatives . . . . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

38
38
39
41
42
44
44
45

7 Extended Black-Scholes Models

7.1 Market Price of Risk . . . . . .
7.2 Fair Prices . . . . . . . . . . . .
7.3 Arbitrage . . . . . . . . . . . .
7.4 PDEs . . . . . . . . . . . . . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

46
47
47
48
49

8 Interest Rate Models

8.1 The Term Structure of Interest Rates
8.2 Short Rate Models . . . . . . . . . . .
8.3 The Hull-White Model . . . . . . . . .
8.4 Pricing Interest Rate Derivatives . . .
8.5 Examples of Interest Rate Derivatives

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

51
51
52
56
58
60

9 Risk Measurement
9.1 Value-At-Risk . . . . . . . . . . . .
9.2 Normal Returns . . . . . . . . . . .
9.3 Equity Portfolios . . . . . . . . . .
9.4 Portfolios with Stock Options . . .
9.5 Bond Portfolios . . . . . . . . . . .
9.6 Portfolios of Bonds and Swaptions
9.7 Diversified Portfolios . . . . . . . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

62
62
63
65
66
67
69
70

iii

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

Preface
These are lecture notes for the course Stochastic Processes for Finance. Parts
marked by * are either hard or regarded to be of secondary importance.
Knowledge of measure theory is not assumed, but some basic measure theoretic notions are required and therefore provided in the notes. The notes are
relatively succinct, a significant part of theory (e.g., proofs of certain assertions) is in fact contained in the exercises. Below is some relevant literature. It
is recommended that you read these books, next to the lecture notes.
1. Baxter, M. and Rennie, A., (1996). Financial calculus. Cambridge University Press.
2. Bj
ork, T., (2009). Arbitrage Theory in Continuous Time. Oxford Finance
Series.
3. Etheridge, A., (2002). A Course in Financial Calculus. Cambridge University Press.
4. Shreve, S.E., (2004). Stochastic Calculus for Finance I. The Binomial
Asset Pricing model. Springer.
5. Shreve, S.E., (2004). Stochastic Calculus for Finance II. Continuous-Time
Models. Springer.

iv

Chapter 1

Pricing Derivatives
A derivative, or contingent claim is a financial contract that promises a certain
payment to the owner, or delivery, depending on the value of some underlying
asset. Examples of the latter include stocks, bonds, interest rates, exchange
rates and commodities. Many different types of derivatives exists and they are
widely traded, for a variety of reasons. Some types are traded on exchanges
(basic call and put options on stocks for instance, see below), others are more
specialized (e.g. derivatives on interest rates) and are traded directly between
the parties involves, typically large financial institutions. Apart from giving
investment opportunities, derivatives are extremely useful for risk management:
financial vulnerability can be reduced by fixing a price for a future transaction.
In this chapter we introduce some basic concepts through examples; formal
definitions and theory follow in later chapters.

1.1

Note on Continuous Compounding

Our model for a (fixed interest) savings account is that a capital of size R0 placed
in such an account at some time t increases to the amount R0 erT at time t+t.
The capital in the account remains ours without restriction: we can withdraw it
at no cost at any time. The constant r is the continuously compounded interest
rate and is not quite an ordinary rate for a savings account, which is more
often a yearly or monthly rate. If interest is added to the account at the end
of a time period of one unit, then this would increase the capital from R0 to
R1 = (1 + r1 )R0 , the interest being r1 R0 , and r1 being the rate per time unit.
If instead we would obtain the interest in two installments, the first after half
a time unit, and the second after one time unit, then the initial capital would
first increase to R1/2 = (1 + r2 )R0 and next to R1 = (1 + r2 )R1/2 = (1 + r2 )2 R0 .
The second time we receive interest on interest. The rate r2 would be the
rate per half time unit, and hence 2r2 should be compared to r1 . However,
the comparison would not be exact, because (1 + r1 /2)2 > 1 + r1 , i.e., a more
frequent compounding rule used in this way yields a greater return. It would be
logical that r1 and r2 relate through the equation (1 + r2 )2 = 1 + r1 , apart from
possibly a correction for the benefit of early payment in the second scheme. The
interest on interest is making the difference.
We could continue this thought experiment and break the time unit in n

equal parts. The reasonable rate rn per (1/n)th time unit would satisfy (1 +
rn )n = 1 + r1 , or

nrn n
1+
= 1 + r1 .
n
Here nrn is the rate per time unit. Taking the limit as n , and assuming
that nrn tends to a limit r we obtain the equation er = 1 + r1 . This r is the
rate of interest per time unit earned in a savings account in which the interest
is added continuously in time. Using the continuous rate r is convenient, as
exponentials multiply.
From now on, if we say that there is a constant continuously compounded
interest rate r, this means that 1 euro in the bank at time t grows to ert euros
at time t. In the discrete setting, the discrete model of a savings account (where
a capital R0 at time zero grows to the capital R1 = (1 + r1 )R0 at time 1 for
some chosen fixed time unit and some interest rate r1 ) can be reduced to the
model with a constant continuously compounded interest rate r by relating r1
and r via the equation er = 1 + r1 .

1.2

Derivatives

The simplest example of a derivative is a so called forward. In a forward contract

one party agrees to deliver a certain asset to the other party on a fixed date
in the future, for a fixed price. There exist for instance forwards on foreign
currency, shares, but also on commodities such as oil or potatoes. Forwards can
be used as an insurance against future price fluctuations of the underlying asset.
The party, or person in a forward contract that promises to buy the asset at
the agreed future date is called the buyer, or holder of the forward and is said
to hold the long position. The other party is the writer, or seller of the contract
and is said to hold the short position. The agreed time in the future at which
the transaction in the underlying takes place is called the time of maturity, or
expiry time, or exercise time of the contract.
Consider a forward on an asset with value St at time t, strike price K and
maturity T . Then at time T the holder of the contract receives an asset which
is worth ST euros and has to pay K euros to the writer. Both T and K are
written in the contract, but ST will be known only at the expiry time. Hence,
the net effect is that the holder of the forward receives ST K euros. This
amount is called the payoff of the forward.
A forward as just described is an example of a European contract, where the
payoff is a fixed function of the value of the underlying asset at maturity. Other
basic examples are European call and put options. A European call option gives
the holder the right (but not the obligation) to buy an underlying asset for a
fixed strike price at a fixed maturity time in the future. Similarly, a European
put option gives the holder the right (but not the obligation) to sell an underlying
asset for a fixed strike price at a fixed maturity time in the future. Consider a
call on an asset with strike K and maturity T , let St be the value of the asset at
time t. Then at maturity T , two things can happen. First of all, the asset price
ST can be greater than the strike price K. Then it is advantageous for the holder
of the call to exercise his right and buy it for K euros. By immediately selling
it again, he makes a net profit of ST K euros. If ST is less than K however,
it makes no sense to exercise the right that the option gives and hence 0 euros

exchange hand. In other words, the payoff of the call is given by (ST K)+ ,
where x+ = x if x 0 and zero otherwise. The same reasoning shows that the
payoff of a put with strike K and maturity T is given by (K ST )+ .
Calls and puts can be used for risk management purposes. The owner of a
certain asset can for instance also buy an appropriate put option on the asset
the insure himself against a too large decrease of the price in a certain period
of time. However, calls and puts are also widely used by speculators, who use
them to gamble on the increase or decrease of asset prices.
The derivatives considered so far are all of European type, with a payoff
of the form C(ST ), a fixed function of the asset value ST at maturity. There
also exist different types of derivatives. First of all, there exist for instance call
and put options that can not only be exercised at maturity, but which give the
right to buy or sell and asset for a fixed price, at some time before maturity, to
be determined by the buyer. Such contracts are called American call and put
options.
Yet another type of options are options whose payoff depends on the whole
history of the asset price up to the time of maturity. Such derivatives are called
path-dependent, because their payoff depends on the whole path (St : t [0, T ]).
The payoff of an Asian option depends on the average of the asset price over a
certain time interval. For instance, an Asian call option on an asset with value
St at time t, with strike price K and maturity T , is a derivative which pays
RT
( T1 0 St dt K)+ to the holder at the time of maturity T .
There exists all kinds of other path dependent options, for instance barrier
options, whose payoff at maturity T depends on whether or not the asset price
has reached certain fixed levels before time T , or lookback options, whose payoff
depends on the minimum or maximum asset price before maturity.

1.3

Hedging a Forward

The owner of a derivative receives some uncertain payoff in the future, depending
on the future development of the value of the underlying asset. Since this might
result in financial gain for one of the parties in the contract, basic economic
reasoning says that this must come at a price. One of the key questions is: how
to determine the fair price of a derivative?
In fact, it is not at all immediately clear what we mean by the fair price
of a derivative. Consider for instance a forward on a stock with strike K and
maturity T , with payoff ST K for the buyer. What should the buyer of the
forward be willing to pay for entering into this contract?
Consider a market that consists of only one risky asset (the stock) and one
riskless asset (the savings account, or bond ). Suppose that we have, besides
buying the contract, two other options to invest our money:
(i) We can put our money in a savings account against a fixed, predetermined
interest rate r. One unit of money placed in the account grows to ert units
during a time interval [0, t] and is freely available. A negative balance in
our account is permitted, thus allowing us to borrow money at the same
interest rate r. If we borrow one unit at time 0, than we owe ert units at
time t, which is equivalent to having a capital of ert units.
(ii) We can invest in the stock. The stock price St at time t is a stochastic
variable, dependent on t. It may be assumed that we know the probability
3

distributions of these variables. For instance, a popular model is that the

variable ST /S0 is log normally distributed.
Next we discuss two answers to the pricing question.
A first idea (that will turn out to be wrong!) is to assume some probability
distribution for the future stock price ST , say ST is distributed according to a
probability density p on (0, ). Then the average, or expected payoff of the
contract at time T equals
Z
Z
E(ST K) =
(x K)p(x) dx =
xp(x) dx K,
0

provided of course that the integral is finite. If the interest rate is zero, then
the buyer should perhaps be willing to pay this amount of money at time 0 for
the forward contract. First, this does not really solve the pricing problem yet,
because in practice we would need to make some concrete choice for the density
p, perhaps on the basis of statistical analysis of stock price data. Much more
importantly however is that, as we argue below, pricing a forward in this way
leads to economic unreasonable and hence undesirable prices.
The key observation is that, by following a very simple strategy, it is possible
to replicate the payoff of the forward. Such strategies are called hedging. We
suppose that the interest rate is 0 for simplicity. At time 0 we borrow K euros
from the bank. We throw in another S0 K euros of our own money (with S0
the value of the stock at time 0) and use the resulting S0 euros to buy one stock.
Then we simply wait until time T . At that time, we pay back the K euros to
the bank and we sell the stock. The net effect of this strategy is easy to see: at
time 0 it costs us S0 K euros to be able to carry out the strategy. At time T
we end up with ST K euros in our hand.
In other words, the payoff that we get by following this simple trading strategy is exactly equal to the payoff of the forward. Basic economic reasoning then
implies that the cost of entering the forward at time 0 should also exactly equal
the cost of carrying out the trading strategy, i.e., the buyer of the forward should
pay S0 K euros to the seller (still under the assumption that the interest rate
is 0). Indeed, suppose the seller accepts a price of P < S0 K euros. Then the
buyer can apply the following strategy:
(i) At time 0, borrow P euros from the bank and enter into a the forward
contract. Also, borrow a stock from someone and sell it for S0 euros.
(ii) At time T , borrow K more euros from the bank and use the money to
buy one stock, as required by the forward contract. Return the stock to
the person it was borrowed from. Of the S0 euros that were obtained at
time 0, use K + P euros to settle the debt with the bank.
Observe that this strategy does not cost any money at time 0. At maturity T ,
it leads to a net payoff of S0 K P , which, by assumption, is strictly positive
in this case. If on the other hand the buyer accepts to pay a price P > S0 K
for the forward, the seller of the contract can follow a similar strategy in order
to earn some riskless money; see Exercise 1.1 below.
A trading strategy of this type, which costs no money at time 0, does not involve any injections or withdrawals of money between times 0 and T , and leads
to a nonnegative payoff at time T which is strictly positive with positive probability (in this case with probability 1 in fact), is called an arbitrage opportunity.
It is economically reasonable to assume the absence of arbitrage opportunities
4

in financial markets. The idea is that if they would exist, then immediately
there would be a huge demand for following the strategy, the price would then
immediately rise, and as a consequence the opportunity would disappear. We
simply have to set the value of the forward at time 0 to F = S0 K (for r = 0),
since any other value would introduce arbitrage opportunities in the market.
The forward example is somewhat special in the sense that the correct forward price F = S0 K does not depend on the distribution of ST . In fact,
it depends on the asset price only through its (observable) value at time 0. Is
there no role for probability theory in evaluating financial contracts?
There is. First we note that the expected gain of owning a contract is equal
to E(ST K), which does depend on the distribution of the asset price. For
2
example, E(ST K) = S0 eT + T /2 K if ST /S0 is log-normally distributed
with parameters T and 2 T , i.e., log(ST /S0 ) N (T, 2 T ).
Second, it turns out that the correct solution can be found from computing
an expectation, with respect to a special probability measure called a martingale
measure. To evaluate the price of a forward, this route would be overdone, as the
preceding hedging strategy is explicit and simple. However, the prices of other
contracts may not be so easy to evaluate, and it will be necessary to assume
some reasonable probabilistic models on the asset price.
The pricing and hedging of derivatives are important tasks in the financial
world. In the subsequent chapters we will treat two classical models that provide
a framework for carrying out these tasks. The first one is the binomial model,
which is a discrete-time tree model for asset prices. The second one is the
famous Black-Scholes model, which is a continuous-time model.
EXERCISE 1.1. Suppose the interest rate is zero. Show that if the buyer
accepts a price P > S0 K for the forward, there exists an arbitrage opportunity
for the seller. Give the explicit construction of the strategy.
EXERCISE 1.2. Suppose that there is a constant continuously compounded
interest rate r. Determine the fair price of the forward in this case.
EXERCISE 1.3. Show that if the price of a forward contract (with maturity
T ) is equal to 0, the fair strike price must be K = S0 erT .
EXERCISE 1.4. Consider European call and put options on a stock, with the
same maturity T and the same strike K. Let C0 , P0 and S0 be, respectively, the
price of the call, put and stock at time 0. Give an arbitrage argument for the
so-called put-call parity C0 P0 = S0 KerT .
2
EXERCISE 1.5. If log(ST /S0 ) N (T, 2 T ), show that EST = S0 eT + T /2 .
2
Hint: first establish EeZt = et /2 .

Chapter 2

Binomial Model
A main objective of financial engineering is to find a fair price of a derivative,
where by fair we mean a price acceptable both for the buyer and the seller.
Following the work by Black and Scholes in the 1970s the prices of derivatives are
found through the principle of no arbitrage introduced in the previous chapter.
In this chapter we introduce the notion of portfolio and briefly discuss the
pricing of derivatives in the so called Binomial Model, a finite security market
where trading takes place at discrete time instants t {0, 1, . . . , N }. Traded are:
one risky asset, a stock with price St at time t; and one riskless asset, a savings
account (or bond) with interest rate r 0 (theory works for r > 1). We
assume that an amount of one unit of money deposited in the savings account
at time 0 grows to a guaranteed amount of ert units at time t.
The goal is to determine the fair price of a composite contingent claim that
pays C(S0 , . . . , SN ) at maturity N . Sometimes we use terms claim and contract
for the payoff function C = C(S0 , . . . , SN ).

2.1

One Period Model

At time 0 we can invest in a stock with price S0 = s0 > 0, or put money in a

savings account (bond) with a fixed interest rate r. We model the stock price
at time 1 as a random variable S1 that can take only two values
(2.1)

P(S1 = us0 ) = p,
P(S1 = ds0 ) = 1 p.

Here u (for up) and d (for down) are two known constants, with u > d, and
p is a number in (0, 1) that may be unknown; cases p = 0, p = 1 are trivial.
The initial date is t = 0 and the terminal date t = 1, with trading possible
only at these two dates. We want to find the fair price at time 0 of a contract
that pays the amount C = C(S1 ) (called the payoff ) at time 1.
Example 2.2. A European call option corresponds to C = (S1 K)+ , for a
given strike price K. The payment on a forward contract is equal to C = S1 K.
Suppose that at time 0 we buy assets and put an amount of money
units in the savings account, , R. A negative value of corresponds to

borrowing money from the bank and a negative value of 0 to short selling of
the stock. Then we have a portfolio (, ) whose value at time 0 is given by
V0 = s0 + 1.

(2.3)

Then the (random) value of the portfolio at time 1 is given by

V1 = S1 + er .
From the perspective of today (time 0) this is a random variable, that we cannot
know with certainty. However, at time 1 the stock price S1 can only be either
us0 or ds0 . Depending on the outcome of S1 , the contract payoff is either C(us0 )
or C(ds0 ), and the value of the portfolio is either us0 + er or ds0 + er .
Suppose we have a portfolio (, ) such that its value at time 1 agrees exactly
with the payoff of the contract, for each of the two possibilities, i.e.,
(
us0 + er = C(us0 ),
(2.4)
ds0 + er = C(ds0 ).
This portfolio will cost us V0 at time 0, and is guaranteed to have the same
value at time 1 as the contract with payoff C(S1 ), whether the stock moves
up or down. We should therefore have no preference for the portfolio or the
contract, hence a fair price for the contract at time 0 is the price of the portfolio,
i.e., V0 corresponding to the portfolio (, ) satisfying the equations (2.4). The
portfolio that has this property is called hedging against the contingent claim.
The equations (2.4) form a system of two linear equations in the unknowns
and and can be readily solved: as s0 > 0 and u > d,
(2.5)

C(us0 ) C(ds0 )
,
s0 (u d)

uC(ds0 ) dC(us0 )
.
er (u d)

We say (, ) replicates C(us0 ) and C(ds0 ) with initial value s0 . Substituting

(2.5) in V0 = s0 + (2.3), we obtain the value of the portfolio at time zero:
(2.6)

V0 = er qC(us0 ) + (1 q)C(ds0 ) ,

where

q=

er d
.
ud

This is the fair price of the contract at time 0. Interestingly, the original probability p in (2.1) turns out to be unimportant for the fair price determination.
An essential feature of a financial market is that it is free of arbitrage, meaning that there is no riskless way of making money. Formally, an arbitrage opportunity is a trading strategy such that P(V0 = 0) = 1, P(V1 0) = 1 and
P(V1 > 0) > 0. A market is arbitrage free if no arbitrage opportunities exist.
Proposition 2.7. The one-period binomial market is free of arbitrage if and
only if d < er < u.
EXERCISE 2.8. Prove Proposition 2.7.
For the rest of the chapter, assume d < er < u. Then the number q from
(2.6) is contained in the interval (0, 1) and can be considered as an alternative probability for the upward move of the stock price S1 in (2.1). This new
probability distribution Q of the stock price (given by (2.1) with p replaced by
7

q) is called the risk-neutral measure or martingale measure, the meanings of

this terminilogy will become clear later. The price of the contract (2.6) can be
rewritten as
(2.9)

V0 = Eq (er C(S1 )),

where the subscript q in Eq means that the expectation is computed under the
measure Q. It can be seen that q is the unique probability such that
Eq (er S1 ) = s0 .
We can write the above equation also in the form Eq (er S1 |S0 ) = S0 (with
S0 the random variable that is equal to the constant s0 with probability one).
In more sophisticated terms, we say that the discounted asset price process
S0 , er S1 is a martingale under Q.
Example 2.10 (Forward). The value at time 0 of a forward is Eq er (S1 K) =
Eq er S1 er K = S0 er K; cf. Exercise 1.2. The strike price that makes
this value equal to zero is K = er S0 ; cf. Exercise 1.3.
In all the exercise below consider the one period model as described above.
EXERCISE 2.11. Let r = 0, s0 = 100, u = 1.2 and d = 0.8. Determine the
price of a European call option which has strike price K = 90. Compute the
corresponding replicating portfolio.
EXERCISE 2.12. Let r = 0, s0 = 100, u = 1.2 and d = 0.8. Determine
the price of a European put option which has strike price 100. Compute the
corresponding replicating portfolio.
EXERCISE 2.13. Prove the put-call parity stated in Exercise 1.4 by using the
martingale measure.

2.2

Two Period Model

Suppose that at time 0 we have the same possibilities for investing as in the
preceding section, but we now consider a full trading horizon of three times:
0, 1, 2. Suppose we have a contract (contingent claim) that pays the amount
C = C(S2 ) at time 2, and we wish to find its fair price at time 0. General case
C = C(S0 , S1 , S2 ) can be considered as well (Exercise 2.15).
The prices of the stock at the three time instants are S0 , S1 , S2 , where we
assume that S0 = s0 is fixed, S1 is equal to either dS0 or uS0 , and S2 is equal
to either dS1 or uS1 . Thus the stock prices follow a path in a binary tree. We
assume that at each node of the tree the decision to move up or down is made
with probabilities p > 0 and 1 p, independently for the different nodes.
Again we may deposit money in a savings account (also a negative amount,
indicating that we borrow money) at a fixed interest rate r. One unit in the
savings account grows to er units at time 1, and to e2r units at time 2.
We can evaluate the claim recursively, backwards in time. Let Vn denote
the value of the contract at time n = 0, 1, 2. Clearly, at time n = 2 the claim
is worth V2 = V2 (S2 ) = C(S2 ). At n = 1, given S1 = s1 , S2 can be either us1
or ds1 , leading to two possible payoffs: C(us1 ) or C(ds1 ). Thus, we are in the

one period model, and by using (2.6) we can evaluate the value of the claim at
time n = 1 as a function of the stock price S1 at time n = 1:
(


er qC(u2 s0 ) + (1 q)C(dus0 ) , if S1 = us0 ,


V1 = V1 (S1 ) =
er qC(uds0 ) + (1 q)C(d2 s0 ) , if S1 = ds0 .
We should see V1 as the value of the contract at time 1 (or the fair price of the
contract at time 1). Now we treat the two possible values of V1 as the payoff on
our contract at time 1 in the one period model. Hence, by using (2.6) and the
above expression for V1 , the value of the contract at time 0 is




V0 = er qV1 (us0 ) + (1 q)V1 (ds0 ) = e2r q qC(u2 s0 ) + (1 q)C(dus0 )



+ (1 q) qC(uds0 ) + (1 q)C(d2 s0 )


= e2r q 2 C(u2 s0 ) + 2q(1 q)C(uds0 ) + (1 q)2 C(d2 s0 )


= Eq e2r C(S2 ) .
(2.14)
In this way, the fair price of the contract once again turns out to be the expectation of the discounted payoff e2r C(S2 ) under the martingale measure Q.
By using the one period model, we obtain that Eq (er S1 |S0 ) = S0 and
Eq (er S2 |S1 , S0 ) = S1 (or Eq (e2r S2 |S1 , S0 ) = er S1 ). This means that the
process (ern Sn , n = 0, 1, 2) is a martingale. We shall elaborate on this later.
EXERCISE 2.15. Generalize the above approach to a contract of the form
C = C(S0 , S1 , S2 ) and apply this to determine the price of an Asian call option

+
which has payoff (S0 + S1 + S2 )/3 100 , in a a two period model with r = 0,
s0 = 100, u = 1.2, d = 0.8. Determine also the hedging portfolio.
To summarize, in the two period model we treated three one period submodels by backward recursion in time. First we determined the values of the
contract at time 1, V1 (us0 ) and V1 (ds0 ) (depending on the realized value of S1 ),
then the value of the contract V0 at time 0. In each one period submodel, we
used the formula (2.6), but there are also the three underlying hedging portfolios defined by (2.5). Denote them by (2 (us0 ), 2 (us0 )), (2 (ds0 ), 2 (ds0 )) and
(1 , 1 ), respectively. Notice that the hedging portfolio in the two period model
consists of two couples (n , n ), n = 1, 2, where (2 , 2 ) is actually a function
of the stock price S1 at time 1. The portfolio is thus dynamic in time and at
each time moment it is based on the available (by that time) information such
as the stock prices.
EXERCISE 2.16. Show that V1 = 1 S1 + 1 er = 2 S1 + 2 er .

2.3

Portfolios

The property of the portfolio claimed in Exercise 2.16 is called self-financing.

At time 1 we get to see the value of the stock S1 and reallocate the positions
and without injection or withdrawal of money. In fact, we implicitly imposed
the self-financing property in the recursive construction, when interpreting the
obtained value V1 of the portfolio as the payoff of the contract at time n = 1.
Indeed, first we determined the two possible values of (2 , 2 ) that replicate
the payoff of the claim at n = 2, then constructed corresponding two possible
values of 2 S1 + 2 er = V1 . Finally we determined the unique pair (1 , 1 ) that
9

replicates these two values of V1 . At that point we used 1 S1 +1 er = V1 . Thus,

the fact 2 S1 +2 er = V1 = 1 S1 +1 er is simply contained in the construction.
The left hand side of the last identity is the market value of the portfolio just
after it has been selected at time n = 1, the right hand side represents the
market value of the portfolio at time n = 1 just before any changes are made in
the portfolio.
In the more general N period model, a portfolio is sequence of real pairs
(n , n ), with the interpretation that n is the amount of stock that an investor
holds at time n and n the number of bonds. Negative values of n and n are
possible. Furthermore, we will not allow n , n to depend on future values of
the stock, the investor is not clairvoyant. For all n, (n , n ) may only depend on
S0 , . . . , Sn1 . We say process (n , n ) is predictable, to be made precise later.
With a predictable portfolio (n , n ), we associate the value process Vn :
V0 = 1 S0 + 1 ,

Vn = n Sn + n enr ,

n = 1, . . . , N.

The idea is to use the results for the one period model recursively to solve
the pricing and replication problem for the N period model (as we did this for
the 2 period model). As we demonstrated in the 2 period model, to be able to
realize this idea, any reallocation (after an initial investment) of the portfolio in
time should be made without infusion or withdrawal of money, i.e., in a budget
neutral way. Precisely, a portfolio is called self-financing if
(2.17)

Vn = n Sn + n enr = n+1 Sn + n+1 enr ,

n = 1, . . . , N 1.

Note that this concept is not relevant to one period models.

A desirable property of a financial market is that it is free of arbitrage,
meaning that it is impossible to make a profit without being exposed to the risk
of incurring a loss. Formally, we call a portfolio with associated value process
Vt an arbitrage opportunity (over the discrete times n = 0, . . . , N ) if it is selffinancing, and P(V0 = 0) = 1, P(VN 0) = 1 and P(VN > 0) > 0. A market is
arbitrage free if no arbitrage opportunities exist.

2.4

N Period Model

In a binomial model with N periods and a constant interest rate r > 1, we

can price a composite claim C(S0 , . . . , SN ) with maturity N by extending the
backwards induction argument. The random variables Zn = Sn /Sn1 , n =
1, . . . , N , are independent, each taking the two values: d and u; P(Zn = u) = p,
n = 1, . . . , N . As before, assume that S0 = s0 > 0 is fixed and d < er < u.
In line with (2.9) and (2.14), the fair price of the claim C(S0 , S1 . . . , SN ) is


(2.18)
V0 = Eq eN r C(S0 , S1 . . . , SN ) ,
where the expectation is taken with respect to the measure Q which is described
r
d
. Recall
by the up-move probability Q(Zn = u) = q, n = 1, . . . , N , with q = eud
that the actual distribution of the stock price is determined by the probability
p of the up-move: P(Zn = u) = p, n = 1, . . . , N .
For example, the price of a European derivative C(SN ) with some payoff
that only depends on the stock price SN at the terminal date is given by


V0 = Eq eN r C(SN ) ,
10

where SN is equal to uXN dN XN S0 for XN the number of up-moves. Clearly,

the variable XN Bin(N, q), a binomial distribution with parameters N and
q. In this case, the fair price V0 can be expressed as a sum, the expectation of
a certain function of a binomially distributed random variable XN .
The induction argument to prove (2.18) is straightforward, but tedious. We
provide some details in the next section which you may want to skip at first
reading. The point is that we shall give a more elegant derivation after developing some martingale theory, which will also give us the intuition needed to
tackle the continuous time models later on.
Computing explicit expressions for the hedging portfolio (n , n )N
n=1 can
also be done in a recursive manner, by considering all one period submodels.
EXERCISE 2.19. How many one period submodels one needs to consider in
order to compute the price of a European option (the payoff depends only on
SN ) at time 0? And what if the option is path dependent?
An important feature of the N period binomial market is that every contingent claim can be generated (hedged) by some trading strategy. One then says
that, by definition, the N period binomial market is complete. This market is
also free of arbitrage, provided that d < er < u, as we demonstrated in the 1
period model. A precise elegant derivation of this fact for the N period model
is based on the martingale technique, and is presented in Chapter 4.
Note that in the computations of the fair price and the hedging portfolio we
need to know the constants u and d. In practice one might observe the prices of
some options on the market, and next calibrate the constants u and d so that
the prices given by the formula agree with the market prices. We would do this
when applying the binomial model. Most people consider the continuous time
models more believable.
One interesting property of the obtained fair price (2.18) of a derivative (and
the replicating portfolio) is that it is based on the martingale measure and does
not depend on the so called real dynamics of the stock. Suppose that two
parties (banks, financial institutions, persons, etc.) want to buy a European
call option with some maturity date and strike price, but have different views
on the stock price movement in the N period model with r = 0 and some
d < 1 < u: the optimistic party believes that the probability of an up-move
is 0.99, the pessimistic one believes that it is 0.01. Although at first glance it
seems reasonable to think that the former is willing to pay more for the option
than the latter, their perceptions of the market movements are immaterial: if
they both handle rationally, they should agree on the same price for this option!

*More on N Period Model

We construct a self-financing dynamic (predictable) portfolio such that at time
N its value VN = VN (S0 , . . . , SN ) is equal to the payoff of the claim. Thus,
VN = N SN + N eN r = C(S0 , . . . , SN ),
where (N , N ) is constructed at time N 1 on the basis of the observed values
S0 , . . . , SN 1 . At time N , the stock price SN takes only the two values SN 1 u

11

and SN 1 d. The above identity implies the following two equations:

N SN 1 u + N eN r = VN (S0 , . . . , SN 1 , SN 1 u),
N SN 1 d + N eN r = VN (S0 , . . . , SN 1 , SN 1 d).
These equations are like the equations (2.4) for the one period case, with the
difference that now we have eN r instead of er . Similarly to (2.5), we obtain
VN (S0 , . . . , SN 1 , SN 1 u) VN (S0 , . . . , SN 1 , SN 1 d)
,
SN 1 (u d)
uVN (S0 , . . . , SN 1 , SN 1 d) dVN (S0 , . . . , SN 1 , SN 1 u)
=
.
eN r (u d)

N =
N

To get the value VN 1 of the hedging portfolio at time N 1, we use the selffinancing property: VN 1 = N 1 SN 1 +N 1 e(N 1)r = N SN 1 +N e(N 1)r .
r
d
For q = eud
, inserting the expressions for N and N results in (check this!)


VN 1 = er qVN (S0 , . . . , SN 1 , SN 1 u) + (1 q)VN (S0 , . . . , SN 1 , SN 1 d) .
Iterating these arguments, for n = N 1, . . . , 0, backwards in time, we obtain


Vn = er qVn+1 (S0 , . . . , Sn , Sn u) + (1 q)Vn+1 (S0 , . . . , Sn , Sn d) ,
(2.20)

Vn (S0 , . . . , Sn1 , Sn1 u) Vn (S0 , . . . , Sn1 , Sn1 d)

,
Sn1 (u d)
uVn (S0 , . . . , Sn1 , Sn1 d) dVn (S0 , . . . , Sn1 , Sn1 u)
n =
.
enr (u d)
n =

Let sn be one of possible values of Sn and let Fn = {S0 = s0 , . . . , Sn = sn }, n =

0, . . . , N . Then we
for VN 1 as VN 1 (s0 , . . . , sN 1 ) =
 can rewrite
 the expression

er Eq VN |FN 1 = er Eq C|FN 1 = er Eq C(s0 , . . . , sN 1 , sN 1 ZN ). The
first identity in (2.20) can be rewritten as Vn (s0 , . . . , sn ) = er Eq [Vn+1 |Fn ],
which implies
Vn (s0 , . . . , sn ) = er Eq [Vn+1 |Fn ] = e2r Eq [Vn+2 |Fn ] = . . .
= e(N n)r Eq [VN |Fn ] = e(N n)r Eq [C(S0 , . . . , SN )|Fn ]
= e(N n)r Eq C s0 , . . . , sn , sn Zn+1 , . . . , sn

N
Y

Zm .

m=n+1

In particular, V0 = eN r Eq C(S0 , . . . , SN ).
Notice the only difference in formulas for the portfolio in multi-period and
one period cases: factor er in (2.5) becomes enr in (2.20). One can formally
have the same formulas if we change the definition on the portfolio and the
associated value process: V0 = 1 S0 + 1 , Vn = n Sn + n er n = 1, . . . N . The
property of self-financing is then Vn = n+1 Sn + n+1 , n = 0, . . . , N 1. The
new quantity n = n e(n1)r has then the interpretation of the total amount
of money at a savings account just before time n.

12

Chapter 3

Discrete Time Stochastic

Processes
A -field 1 F on a sample space is a collection of subsets of . A probability
measure P is a function P : FP
[0, 1], that satisfies the properties: (i) P() = 0,
P() = 1; (ii) P(n Fn ) = n P(Fn ) for all countable collections (Fn )nN of
pairwise disjoint sets from F. The triplet (, F, P) is a probability space.
A random variable (vector) X is a measurable function X : Rn if
{ : X() B} F for all B B, where B is the Borel -field on Rn . A -field
corresponding to observing a random vector X : Rn , is
(X) = {{X() B}, B B},

B is the Borel -field on Rn .

This (X) is called the -field generated by X. Notice that (X) F. For a
set A, the indicator function is 1A (x) = 1 if x A and 1A (x) = 0 if x 6 A.

3.1

Stochastic Processes

A stochastic process in discrete time is a (finite or infinite) sequence (Xn )n0 =

(X0 , X1 , . . .) of random variables or vectors, defined on a given probability space
(, F, P). Mathematically, random variables are measurable functions Xn :
R that map outcomes into numbers Xn (). A probability measure
P on subset of gives rise to a probability measure for (Xn )n0 . The stochastic
process (Xn )n0 maps every outcome into a sequence of numbers (Xn ())n0
called a sample path.
The best way to think of a stochastic process is to visualize the sample
paths as random functions. We generate an outcome according to some
probability measure on sample space and next consider a function n 7 Xn ()
on the domain N {0}. This domain is referred to as the set of discrete times.
Example 3.1 (Binomial model). The binomial model for the stock price is a
stochastic process (Sn )N
n=0 = (S0 , S1 , . . . , SN ), where each possible sample path
1 A rigorous mathematical definition of -field F includes the requirements (i) F ; (ii)
if F F , then F c F ; (iii) if F1 , F2 , . . . F , then i Fi F . The Borel -field B on Rd ,
d N, is the smallest -field of subsets of Rd that satisfies the above properties and contains
all intervals. In what follows, all considered sets from Rd will be Borel sets.

13

is given by a path in the binomial tree, and the probability of a sample path
is the product of the probabilities of the edges along the path. This gives an
intuitively clear description of the process, but for later use it is instructive to
define the stochastic process also formally as a map on a given sample space .
Take to be the set of all N -tuples = (1 , . . . , N ), where each i {0, 1}.
The stochastic process can be formally defined by setting S0 = s0 and
Sn (1 , . . . , N ) = s0 u

Pn

i=1

i n

Pn

i=1

n = 1, . . . , N.

If i = 1, the sample path goes up in the tree at time i, whereas i = 0 indicates

a down move. The value Sn is determined by the total number of moves up
and down in the tree up till time n. All probabilities concerning the stock price
process can be computed by using the probability mass function on :
PN
PN


p() = p (1 , . . . , N ) = p i=1 i (1 p)N i=1 i .

For example, the probability that the

P stock price at maturity N is bigger than
the strike price K is P(SN > K) = :SN ()>K p().

3.2

Conditional Expectation

For a discrete random variable (or vector) X and a discrete random vector Y ,
the conditional expectation of X given the event Y = y is given by
X
E(X|Y = y) =
xP(X = x|Y = y).
x

If we denote this function of y by f (y) = E(X|Y = y), then we write E(X|Y )

for f (Y ). This is a random variable, called the conditional expectation of X
given Y . Some important properties are given in the following lemma.
Lemma 3.2.
(i) EE(X|Y ) = EX.
(ii) E(E(X|Y, Z)|Z) = E(X|Z) (tower property).
(iii) E(X|Y ) = X if X = f (Y ) for some function f .
(iv) E(g(Y )X|Y ) = g(Y )EX if X, Y are independent, g is some function.
These properties can be proved from the definition, but are intuitively clear.
The first assertion says that the expectation of X can be computed in two steps,
first using the information on another vector Y , and next taking the expectation
of the result. Assertion (ii) gives exactly the same property, with the difference
that the expectations are computed conditionally on a vector Z. Property (iii)
says that we can predict X in terms of Y exactly if it is a function of a known Y ,
which is obvious. The fourth property follows from the fact that the conditional
(given Y ) distribution of X is unconditional if X and Y are independent.
EXERCISE 3.3. Suppose you generate N points in the interval [0, 1] as follows.
First you choose N from the Poisson distribution with mean 100. Next given
N = n you generate a random sample of n random variables from a given
distribution F on [0, 1]. What is the expected number of points in an interval
B [0, 1]?

14

We shall use the notation E(X|Y ) also if X or Y are continuous random

variables or vectors. Then the preceding definition does not make sense, because
the probabilities P(X = x|Y = y) are not defined if P(Y = y) = 0, which is the
case for continuous random variable Y . However, the conditional expectation
E(X|Y ) can still be defined as a function
of Y , namely
as the function such
that, for every function g, E E(X|Y )g(Y ) = E Xg(Y ) . The validity of this
equality in the case of discrete random variables can be easily checked. For
general random variables X and Y , we take this as a definition of conditional
expectation2 , where it is also understood that E(X|Y ) must be a function of Y .
The properties from the lemma continue to hold for this extended definition of
conditional expectation.
In most cases this abstract definition agrees perfectly with your intuition of
the expected value of X given Y . However, in some cases where there are many
sets {Y = y}, all with probability zero, your intuition could deceive you. The
problem is then usually that there are several equally good, but incompatible
intuitions.

3.3

Filtration

A filtration in discrete time is an increasing sequence F0 F1 of -fields.

The -field Fn may be thought of as the events whose occurrence is determined
up to time n, the known events at time n. Filtrations allow to model the
increasing flow of information as time goes by, especially when filtrations are
generated by stochastic processes.
The natural filtration of a stochastic process (Xn )n0 is defined by
Fn = (X0 , . . . , Xn ),

n = 0, 1, . . . .

In a way, Fn contains all the information available by observing the process till
time n.
We say that a process (Xn )n0 is adapted to the filtration (Fn )n0 if (Xn )
Fn for every n. Thus the events connected to an adapted process up to time n
are known at time n. The natural filtration corresponding to a process is the
smallest filtration to which it is adapted. If the process (Yn )n0 is adapted to the
natural filtration of a stochastic process (Xn )n0 , then Yn = fn (X0 , X1 , . . . , Xn )
for some function fn , for each n.
We say that a process (Yn )n0 is predictable relative to the filtration (Fn )n0
if (Yn ) Fn1 for each n. Thus the events connected to a predictable process
are known one time instant before they happen. If Fn = (X0 , . . . , Xn ), then
this is equivalent to Yn being a function Yn = fn (X0 , . . . , Xn1 ) of the history
of the process (Xn )n0 up to time n 1, for some fn .
If FY is the -field generated by Y , then we define E(X|FY ) = E(X|Y ). In a
way, E(X|F) can be interpreted as the expected value of X given the information
F. The trivial -field {, } is the -field containing no information.
Lemma 3.4.
2 A rigorous mathematical definition of E(X|Y ) is as follows: first define E(X|A) for a
-field A F as an A-measurable function that satisfies E[E(X|A)1A ] = E[X1A ] for each
A A; next define E(X|Y ) = E(X|(Y )), where (Y ) is the -field generated by Y .

15

(i) E(X|{, }) = EX.

(ii) For -fields F G, the tower property holds: E(E(X|G)|F) = E(X|F).
(iii) E(X|Y ) = X if (X) (Y ).
Example 3.5. In the N period binomial model, a portfolio (n , n ) is predictable with respect to the filtration (Fn )N
n=0 generated by the stock price
process (Sn )N
,
i.e.,
F
=
(S
,
.
.
.
,
S
).
The
process the corresponding value
n
0
n
k=0
process (Vn )N
is
adapted
to
the
filtration
(F
)N
n
n=0
n=0 .

3.4

Martingales

A stochastic process (Xn )n0 is a martingale relative to a given filtration

(Fn )n0 if it is adapted to this filtration and E(Xn |Fm ) = Xm for every m < n.
By default, if we say that a certain process is a martingale without specifying
any filtration, we always mean the natural filtration of that process.
The martingale property is equivalent to E(Xn Xm |Fm ) = 0 for every
m < n, expressing that the increment Xn Xm given the past Fm has
expected value 0. A martingale is a stochastic process that, on the average,
given the past, does not grow or decrease.
Example 3.6 (Random walk). Let X1 , X2 , . . . be independent random variables
with mean zero. Define S0 = 0 and Sn = X1 + + Xn for n N. Then S is a
martingale relative to its natural filtration. Indeed, S is adapted by construction
and E(Sn+1 |Fn ) = E(Sn + Xn+1 |Fn ) = Sn + EXn+1 = Sn , since Sn Fn and
Xn+1 is independent of Fn .
Example 3.7 (Doob martingale). If Y is a random variable with E|Y | <
and Fn an arbitrary filtration, then Xn = E(Y |Fn ) defines a martingale. This
follows from the tower property of conditional expectations: E E(Y |Fn )|Fm ) =
E(Y |Fm ) for any m < n.
The martingale property can also be equivalently characterized as follows: a
process (Xn )n0 is a martingale if E(Xn+1 |Fn ) = Xn for every n.
EXERCISE 3.8. Prove the above claim by using the tower property.
EXERCISE 3.9. Let Z
, Z2 , . . . be independent
N (0, 1)-variables. Show that

 1P
n
n
forms
a
martingale.
the sequence Sn = exp
Z

i=1 i
2
EXERCISE 3.10. In a branching process we start with N0 = 1 individuals
at time 0, and at each time n each individual has a random number (chosen
from a fixed distribution) of offspring independent of the other individuals. The
new generation consists of the offspring only. Thus, given there are Nn = k
individuals at time n, the number of individuals Nn+1 in the (n+1)th generation
Pk
(n)
(n)
(n)
is distributed as i=1 Xi for iid random variables X1 , . . . , Xk . Show that
(n)
(Nn )n0 is a martingale if and only if EXi = 1.

3.5

Change of Measure

In mathematical finance, an important step in evaluating derivatives is the

change from a real world probability measure P to a risk-neutral measure Q,
making a discounted asset price process a martingale. If a process (Xn , n 0)

16

is a martingale relative to Q, then typically it is not a martingale relative to

P. This is because the martingale property involves the expected values, and
measures P and Q may assign different probabilities to the same events. The
two measures should agree, however, on which stock price paths are possible
(i.e., which are of positive probability). This leads to the notion of equivalent
measures.
On a finite sample space , consider two probability measures P and Q with
mass functions p() and q() respectively. If p() > 0 for all , then
L() =

q()
p()

is called the Radon-Nikodim derivative of Q with respect to P. Then we can

relate the expectations (and the probabilities) Eq , Ep with respect to measures
Q and P: for any random variable Y and any event A Eq Y = Ep [LY ]. In
particular, Q(A) = Ep (1A L). Besides, Ep L = 1.
Two measures P and Q on a finite sample space are said to be equivalent
if they agree which events have probability zero, i.e., P(A) = 0 iff Q(A) = 0.
Clearly, if P(L > 0) = 1, P and Q are equivalent.
Example 3.11 (N period binomial model). Let p, q (0, 1) be two probabilities of an up-move in the N period binomial model, leading to two resulting
measures P and Q on = {(1 , . . . , N ), i {0, 1}}. Then P and Q are
clearly equivalent, and the Radon-Nikodim derivative of Q with respect to P is
L() =

 1 q N PN
 q PN
i=1 i
i=1 i
p

1p

EXERCISE 3.12. In Example 3.11, let p (0, 1), but q = 0 (or q = 1). Show
that the resulting measures P and Q are not equivalent.
EXERCISE 3.13. Show that P and Q are equivalent if they agree which events
have probability one.

17

Chapter 4

Binomial Model Revisited

In this chapter we solely consider the multi-period binomial model introduced
in Chapter 2. Assume that the numbers u and d (d < u) are known, and we
can save or borrow money at a fixed interest rate r > 1. At each trading
time instant t, the stock price St moves up with a probability p (0, 1). We
exclude the cases p = 0, p = 1 as uninteresting, since the stock price evolves in
the trivial deterministic way in either of these cases. The up-move probability
p (0, 1) determines the underlying (real-world) probability measure P for the
stock price process (Sn )n0 described in Example 3.1.
In view of Proposition 4.6 below, we should assume d < er < u to allow no
arbitrage in the market. In this case, we can introduce an alternative (fictitious)
r
d
up-move probability q = eud
(0, 1), the corresponding (so called risk-neutral)
measure for the stock price process is denoted by Q.
Introduce the default filtration Fn = (S0 , . . . , Sn ), n 0. In this chapter,
when we say that a certain process is adapted, or predictable, or a martingale
without specifying the pertaining filtration, we always mean (Fn )n0 . All considered portfolios are predictable with respect to the default filtration. Sometimes we use a shorthand notation X for a process (Xn )n0 or (Xn )N
n=0 . A
nr

discounted version of X is denoted by X = (Xn )n0 = (e

Xn )n0 .

4.1

Martingales in the Binomial Model

Here we consider some martingales in the binomial model.

Example 4.1. In the binomial model, compute P(Sn+1 = uSn |Fn ) = 1
P(Sn+1 = dSn |Fn ) = p and


E(Sn+1 |Fn ) = uSn p + dSn (1 p) = Sn up + d(1 p) .
The last expression is equal to Sn if the identity up + d(1 p) = 1 holds. For
instance, if u = 2 and d = 1/2, then (Sn )n0 is a martingale for p = 1/3.
Example 4.2 (Discounted stock). In the binomial model, compute
E(e(n+1)r Sn+1 |Fn ) = pue(n+1)r Sn + (1 p)de(n+1)r Sn .
The discounted stock price process S = (Sn )n0 = (enr Sn )n0 is a martingale
only if the right side of the last display is equal to enr Sn , which in turn holds if
18

d
. But p (0, 1), hence d < er < u must hold. Finally, we conclude
p = q = eud
r
that if d < e < u, then (Sn )n0 is a martingale for p = q.

In Example 4.2 we have seen that the process S defined by Sn = enr Sn in

the binomial model is a martingale if the up-move occurs with the probability
r
d
q = eud
. In this section we shall show that all other martingales in this setting
can be derived from S in the sense that the increments Mn = Mn Mn1 of
any martingale M0 , M1 , . . . must be multiples n Sn for a predictable process
= (0 , 1 , . . .). In other words, the change Mn of an arbitrary martingale at
with the proportionality constant
time n 1 is proportional to the change in S,
At
n being a function of the preceding values S0 , . . . , Sn1 of the process S.
time n 1 the only randomness to extend M0 , . . . , Mn1 into Mn is in the
increment Sn .
Theorem 4.3. If M is a martingale in the binomial model under the measure
Q, then there exists a predictable process (n )n0 such that
Mn = n Sn ,

n N.

Proof. Because M is adapted to the filtration generated by S = (Sn )n0 , for

each n the variable Mn is a function of S0 , . . . , Sn . Given Fn1 the values
of S0 , . . . , Sn1 are fixed and hence Mn can assume only two possible values,
corresponding to a downward or upward move in the tree. By a similar argument we see that the variable Mn1 is fixed given Fn1 , and hence Mn
has two possible values given Fn1 . If we fix S0 , . . . , Sn1 , then we can write
Mn = gn (Sn ) for some function gn (which depends on the fixed values of
S0 , . . . , Sn1 ). Similarly we can write Sn = fn (Sn ). The martingale properties of the processes M (by assumption) and S (by Example 4.2) give that
Eq (Mn |Fn1 ) = 0 = Eq (Sn |Fn1 ), or
qgn (uSn1 ) + (1 q)gn (dSn1 ) = 0,
qfn (uSn1 ) + (1 q)fn (dSn1 ) = 0.
This implies that gn (uSn1 )/fn (uSn1 ) = gn (dSn1 )/fn (dSn1 ). We can define n as this common ratio.

4.2

Pricing and Hedging

In Example 4.2 we actually established that, provided d < er < u, the discounted stock price process (Sn )n0 forms a martingale under the measure Q,
r
d
the one with the up-move probability q = eud
. This implies the following.
Proposition 4.4. In the binomial model with d < er < u, let (n , n ) be a
self-financing portfolio with the associated value process Vn , n 0. Then the
discounted value process (Vn )n0 = (enr Vn )n0 is a martingale under Q.
That is way Q is called the risk-neutral (and martingale) measure.
EXERCISE 4.5. Prove Proposition 4.4.
The condition d < er < u from Proposition 4.4 has already occurred in the
one period model to prevent arbitrage. The same reason applies here. Indeed,
19

if er d, then the returns on the asset are with certainty bigger than the return
on the savings account, whereas if er u, then the returns are with certainty
smaller. Then the riskless savings account is never or always preferable over the
risky asset, respectively, and a reasonable portfolio will consist of only one type
of investment. We make this precise below.
Recall that an arbitrage opportunity is a self-financing trading strategy with
associated value process (Vn )N
n=0 , such that P(V0 = 0) = 1, P(VN 0) = 1 and
P(VN > 0) > 0. A market is arbitrage free if no arbitrage opportunities exist.
Proposition 4.6. The multi-period binomial market is free of arbitrage under
any probability measure P that is equivalent to Q iff d < er < u.
EXERCISE 4.7. Prove Proposition 4.6 (hint: use Propositions 2.7 and 4.4).
EXERCISE 4.8. Show that any P with an up move probability p (0, 1) is
equivalent to Q with an up move probability q (0, 1).
In view of Proposition 4.6 and Exercise 4.8, we require that d < er < u.
Now we are ready to solve the problem of pricing a claim C = C(S0 , . . . , SN ),
with the expiry time N , by using the martingale technique.
EXERCISE 4.9. Derive the solution (2.18) of the pricing problem. More generally, show that the fair price Vn of the contract C(S0 , . . . SN ) at time n is


(4.10)
Vn = e(N n)r Eq C(S0 , . . . , SN )|Fn , n = 0, 1, . . . , N.
Hint: use Proposition 4.4.
Clearly, Vn in (4.10) is a function of S0 , . . . Sn . It can be computed as
follows. Introduce the iid random variables Zn = Sn /Sn1 , each taking only
the values:
Qnd and u, Q(Zn = u) = q, P(Zn = u) = p, n = 1, . . . , N . Clearly,
Sn = S0 k=1 Zk . Given the (observed) values S0 = s0 , . . . , Sn = sn , the fair
price Vn = Vn (s0 , . . . , sn ) of the contract C at time n {0, 1, . . . , N } is
(4.11)

N


Y
Vn (s0 , . . . , sn ) = e(N n)r Eq C s0 , . . . , sn , sn Zn+1 , . . . , sn
Zm ,
m=n+1

EXERCISE 4.12. Prove (4.11).

Notice that Exercise 4.9 solves the pricing problem, but does not give the
corresponding replicating portfolio. Of course, this can be done by using many
one period submodels in the backward recursion approach, as is outlined in
Section 2.4. However, we look at the problem of constructing a hedging portfolio
yet from another perspective. For a claim C = C(S0 , . . . , SN ), define
Vn = Eq (eN r C|Fn ),

n = 0, 1, . . . N.

In view of Example 3.7, the stochastic process V is a martingale. Therefore, by

Theorem 4.3 there exists a predictable process = (n )N
n=1 such that
(4.13)

Vn = n Sn ,

n = 1, . . . , N.

Given this process, we define another process by

(4.14)

n = Vn1 n Sn1 ,

n = 1, . . . , N.

Since is predictable and V and S are adapted, the process is predictable.

We now interpret (n , n ) as a portfolio at time n:
20

(i) n is the number of assets held during the period (n 1, n],

(ii) n is the number of units in the saving account during the period (n1, n].
Because both processes are predictable, the portfolio (n , n ) can be created
at time n 1 based on information gathered up to time n 1, i.e., based on
observation of S0 , . . . , Sn1 . We shall think of the assets and savings changing
values (from Sn1 to Sn and from e(n1)r to enr ) exactly at time n, so that we
can adjust our portfolio right after time n. Then the value of the portfolio at
time n is
Vn = n Sn + n enr .
Just after time n we change the content of the portfolio; the value of the new
portfolio becomes
n+1 Sn + n+1 enr .
The following theorem shows that this amount is equal to Vn and hence the
new portfolio can be formed without additional money: the portfolio process
(, ) is self-financing. Furthermore, the theorem shows that the value VN of
the portfolio is exactly equal to the payoff of the claim C at the expiry time N .
As a consequence, we should be indifferent to owning the contract with claim
C or the portfolio (1 , 1 ) at time 0, and hence the fair price of the contract is
the value V0 of the portfolio.
Theorem 4.15. The portfolio process (, ) defined by (4.13)(4.14) is selffinancing. Furthermore, its value process V satisfies Vn = enr Vn , n = 0, . . . , N .
In particular, VN = C with probability one.
Proof. The equation (4.14) that defines n+1 can be rewritten in the form
n+1 Sn + n+1 ern = enr Vn , and Vn = Vn1 + Vn , where Vn = n Sn
by (4.13). Therefore,
n+1 Sn + n+1 enr Vn = enr Vn Vn
= enr Vn1 + enr n Sn (n Sn + n enr )
= enr Vn1 + enr n (Sn Sn ) n enr
= enr Vn1 enr n Sn1 n enr .
The right side is zero by the definition (4.14) of n . Thus the portfolio is selffinancing, as claimed. It also follows from these equations that enr Vn Vn = 0,
whence Vn = enr Vn is the discounted value of the portfolio at time n, for every
n. Furthermore, the value of the portfolio at time N is
VN = erN VN = erN Eq (erN C|FN ) = C,
since C is a function of S0 , S1 , . . . , SN , by assumption.
Because VN = C with certainty, the portfolio (, ) replicates the payoff of
the claim C (and is called hedging). The claim value C is a random variable
that depends on S0 , . . . , SN . However, no matter which path the asset prices
takes in the binomial tree, the portfolio always ends up having the same value
as the claim. This is achieved by reshuffling assets and savings at each time n,
based on the available information at that time.

21

The portfolio management can be implemented in practice. If we have sufficient funds to form the portfolio at time 0, then we never run into debt when
carrying out the hedging strategy. We interpret the value of the portfolio at
time 0, the amount of money needed to create the portfolio (1 , 1 ), as the fair
price of the claim at time 0. Since V is a (Doob) martingale under Q, this is
V0 = V0 = Eq (erN C|F0 ) = Eq (erN C).
Note that the formula expresses the price in the claim C without intervention
of the portfolio processes. Notice that we in passing solved the pricing problem
(2.18) (and also (4.10)) once again.
Example 4.16 (Forward). The claim of a forward with strike price K is C =
SN K. The value of the forward at time 0 is equal to V0 = Eq erN (SN K) =
Eq erN SN erN K. Since the process (ern Sn )n0 is a martingale under the
measure Q, V0 = Eq S0 erN K = S0 erN K. In Section 1.3 we obtained the
same result by describing an explicit hedging strategy.
Example 4.17 (European call option). The claim of a European call option
with strike price K is C = (SN K)+ . The fair price of the option at time 0
is V0 = Eq erN (SN K)+ . The variable SN is distributed as S0 uXN dN XN ,
where XN is the number of upward moves in the tree. Because XN Bin(N, q)
under the martingale measure Q, it follows that
V0 = erN

N
X

(S0 ux dN x K)+

x=0

 
N x
q (1 q)N x .
x

This expression is somewhat complicated, but easy to evaluate on a computer.

An alternative method of computation is backwards induction, as in Chapter 2.
An approximation formula for large N is given in the next section.
EXERCISE 4.18. Denote the values of the European call and put options at
time n {0, . . . , N } by Cn and Pn respectively. Assume that the both options
have the same strike price K. Derive the put-call parity
Cn Pn = Sn er(N n) K,

4.3

n = 0, . . . , N.

Towards Continuous Time

In the real world asset prices change almost continuously in time. The binomial
tree model can approximate this if the number of steps N is large. Mathematically we can even compute limits as N , in the hope that this gives a
realistic model.
A limit exists only if we make special choices for the relative up and down
moves u and d. Unless u and d tend to 1 as the number of moves N increases,
the asset price will explode and our model does not tend to a limit. We shall
think of the N moves in the binomial tree taking place in a fixed interval [0, T ],
at the times N , 2N , . . . , N N for N = T /N . Then it is reasonable to redefine
the interest rate in one time instant as rN , giving a total interest of r over the
interval [0, T ]. We also assume that, for given constants R and > 0,
dN = eN

uN = eN +

,
22

Then both d and u approach 1 as the length of a time interval tends to zero.
These choices are somewhat special, but can be motivated by the fact that the
resulting model tends to continuous time model considered later on.
Denoting by XN the number of times the stock price moves up in the time
span 1, . . . , N , the stock price SN at the terminal time N can be written as

(XN N/2) 
N N XN
p
.
SN = S0 uX
= S0 exp T + T
N dN
N/4
To price a European claim C(SN ), we need to determine the distribution of
SN under the martingale measure Q. Under the measure Q, XN Bin(N, qN ),
Using a Taylor expansion ex = 1 + x + x2 /2 + o(x2 ) as x 0, we evaluate

qN

1 1 p  + 2 r 
erN eN N

=
=
N
+ O(N ),
2 2

eN + N eN N

as N 0. Recalling that N =
log

T
N,

we obtain qN (1 qN )

1
4

as N and

2
 1 
 XN N qN  r 
SN
2
p
T
+O
= T + T
S0

N
N/4
(XN N qN )


2
2
p
+ T (r 2 )
N (r 2 )T, 2 T , N
= T
N/4

The last relation is the convergence in distribution which follows since

p
qN (1 qN ) XN N qN
XN N qN
p
p
=
N (0, 1), N ,
1/2
N/4
N qN (1 qn )
by the central limit theorem. Thus, under the martingale measure Q, in the
2
limit log SSN0 is normally distributed with mean (r 2 )T and variance 2 T .
Evaluating the (limiting) option price is now a matter of straightforward inte

2
gration. Let Z N (0, 1) and ST be such that log(ST /S0 ) N (r 2 )T, 2 T .
For a claim C = C(SN ) with expiry time T , the fair price of at time 0 is

2


V0 = erT EqN C(SN ) erT E[C(ST )] = erT E C(S0 e(r 2 )T + T Z ) .

Example 4.19 (European call option). The (limiting) fair price of a European
call option with expriry time T and strike price K is V0 = erT E(ST K)+ ,


2
where log(ST /S0 ) N (r 2 )T, 2 T . This can be computed to be
 log(S /K) + (r +
0

V0 = S0
T

2
2 )T

rT

Ke

 log(S /K) + (r
0

2
2 )T

where (z) = P(Z z) for Z N (0, 1). This is the famous Black-Scholes
formula found in 1973. We shall recover it later in a continuous time setup.
EXERCISE 4.20. Suppose that Z N (0, 1), S, K > 0, R, 6= 0, are some
constants. Show that, for z0 = 1 (log(K/S) ),
E(Se+ Z K)+ = Se+

2
2

( z0 ) K(z0 ).

Using this, derive the (asymptotic) Black-Scholes formula from Example 4.19.

23

Chapter 5

Continuous Time Stochastic

Processes
5.1

Stochastic Processes

A continuous-time stochastic process is an indexed collection of random variables

X = (Xt : t 0) = (Xt )t0 , defined on a given probability space. Thus every
Xt is a map Xt : R mapping outcomes into numbers Xt (). The
functions t 7 Xt () attached to the outcomes are called sample paths, and the
index t is referred to as time. The best way to think of a stochastic process is
to view it as a random function on the domain [0, ), with the sample paths
as its realizations.
For any finite set t1 < t2 < < tk of time points the vector (Xt1 , . . . , Xtk )
is an ordinary random vector in Rk , and we can describe a great deal of the
process by describing the distributions of all such vectors. On the other hand,
qualitative properties such as continuity or differentiability of a sample path
depend on infinitely many time points.

5.2

Brownian Motion

Brownian motion (often called also Wiener process) is a special stochastic process, which is of much interest by itself, but will also be used as a building block
to construct other processes. It can be thought of as the standard normal
process. A Brownian motion is often denoted by the letter W , after Wiener,
who was among the first to study Brownian motion in a mathematically rigorous
way. The distribution of Brownian motion is known as the Wiener measure.
A stochastic process W is a Brownian motion if
(i) the increment Wt Ws N (0, t s), for any 0 s < t;
(ii) the increment Wt Ws is independent of (Wu : u s), for any 0 s < t;
(iii) W0 = 0;
(iv) any sample path t 7 Wt () is a continuous function.
It is certainly not clear from the definition that Brownian motion exists, in
the sense that there exists a probability space with random variables Wt defined
on it that satisfy the requirements (i)(iv). However, it is a mathematical
24

theorem that Brownian motion exists, and there are several constructive ways
of exhibiting one. We shall take the existence for granted.
Properties (i) and (ii) can be understood in the sense that, given the sample
path (Wu : u s) up to some point s, Brownian motion continues from its
present value Ws by adding independent (normal) variables. In fact it can
be shown that given (Wu : u s) the process t 7 Ws+t Ws is again a
Brownian motion. Thus at every time instant Brownian motion starts anew
from its present location, independently of its past.
The properties of Brownian motion can be motivated by viewing Brownian
motion as the limit of the process in a binomial tree model, where starting from
S0 = 0 a process (Sn )n0 is constructed
Pnby moving up or down 1 in every step,
each with probability 1/2, i.e., Sn = k=1 Xk , for an i.i.d. sequence (Xk )kN
with P(Xk = 1) = P(Xk = 1) = 1/2. For a given N , we could set the values
of the process S0 , S1 , S2 , . . . at the time points 0, 1/N, 2/N, . . . and rescale the
vertical axis so that the resulting process remains stable. This leads to the
process W (N ) given by
1 X
(N )
Xi .
Wt =
N i:itN
By the Central Limit Theorem, with
(N )

Wt

1
Ws(N ) =
N

denoting convergence in distribution,

Xk

N (0, t s),

as N .

k:sN <ktN

Furthermore, the variable on the left side is independent of the variables Xk

(N )
with index k not contained in the sum, and hence of Ws . Thus in the limit
(N )
as N , the processes W
satisfy the properties (i)(iii). It can indeed
be shown that the sequence W (N ) converges, in a suitable sense, to a Brownian
motion process. The main challenge in proving existence of Brownian motion is
the required continuity (iv) of the sample paths.
t )t0 are indeEXERCISE 5.1. Suppose c > 0, [1, 1], (Wt )t0 and (W
pendent
Brownian motions. Which of the following are Brownian motions? (i)

( tW1 )t0 , (ii) (Wt )t0 , (iii) (W2t Wt )t0 , (iv) (c1/2 Wct )t0 , (v) (Wt +
t )t0 . Motivate your answers.
(1 2 )1/2 W

5.3

Filtrations

A filtration (Ft )t0 in continuous time is an increasing collection of -fields

indexed by [0, ). Thus Fs Ft for every s < t. A stochastic process X =
(Xt )t0 is adapted to a given filtration if (Xt ) Ft for every t. If we say that a
process is adapted, there is a filtration meant by default. All events concerning
the sample paths of an adapted process until time t are contained in Ft . The
natural filtration of a stochastic process X = (Xt )t0 is
Ft = (Xs : s t).
This filtration corresponds exactly to observing the sample paths of X up to
time t, and is the smallest filtration to which X is adapted.

25

5.4

Martingales

A martingale in continuous time relative to a given filtration (Ft )t0 is an

adapted process X such that
E(Xt |Fs ) = Xs ,

for every s < t.

This property is equivalent to the increments Xt Xs having expected value 0

given the past and present: E(Xt Xs |Fs ) = 0. If a process X is said to be a
martingale without specifying a filtration, we mean the natural filtration of X.
For example, a Brownian motion W is a martingale relative to its natural
filtration since E(Wt Ws |Fs ) = E(Wt Ws ) = 0 by the properties (i) and (ii).
EXERCISE 5.2. Let c > 0 and (Wt )t0 be a Brownian motion. Which of the
following are martingales? (i) (Wt2 t)t0 , (ii) (cWt/c2 )t0 , (iii) (ecWt )t0 , (iv)


Rt
(exp{cWt c2 t/2})t0 , (v) tWt 0 Ws ds t0 . Motivate your answers.

5.5

Generalized Brownian Motion

When working with a Brownian motion it is sometimes useful to include more

information into a filtration than given by observing the Brownian sample paths.
Given a filtration (Ft )t0 we replace property (ii) of a Brownian motion by the
alternative property
(ii) W is adapted to (Ft )t0 and the increment Wt Ws is stochastically
independent of Fs , for any 0 s < t.
By the requirement for W to be adapted, the filtration (Ft )t0 is necessarily
larger than the natural filtration of W . Therefore, the property (ii) requires
more than the corresponding (ii).
It can be checked that a generalized Brownian motion is still a martingale.

5.6

Variation

Brownian motion has strange sample paths. They are continuous by assumption,
but they are not differentiable. We can see this by studying the variation of the
sample paths.
Let 0 = tn0 < tn1 < < tnkn = t be a sequence of partitions of a given interval
[0, t] such that the mesh width maxi (tni tni1 ) tends to zero as n . Then
for a continuously differentiable function f : [0, t] R we have, as n ,
Z t
kn
kn
X
 n
 X
 0 n  n

f (ti ) f (tni1 )
f (ti1 )ti tni1 
|f 0 (s)| ds.
i=1

i=1

The left side of this equation is called the variation of f over the given partition.
The approximation can be shown to be correct in the sense that the variation
indeed converges to the integral on the right as the mesh width of the partition
tends to zero. We conclude that the variation of a continuously differentiable
function is bounded if the mesh width of the partition decreases to zero. As a
consequence the quadratic variation
kn
kn
X
 n


X
 n

f (ti ) f (tni1 )2 max f (tni ) f (tni1 )
f (ti ) f (tni1 )
i=1

i=1

26

tends to zero as n , since the maximum of the increments tends to zero by

the continuity of f and the variation is bounded.
The sample paths of Brownian motion do not possess this property. In
fact, the quadratic variation rather than the variation of the sample paths of
Brownian motion tends to a nontrivial limit.
This is true in a stochastic sense. It will be convenient to use the notation
L2

for convergence in second mean: a sequence of random variables Xn is

said to converge in second mean or converge in L2 to a random variable X,
L2
notation Xn
X as n , if
E(Xn X)2 0

as n .

Because the second moment of a random variable is the sum of its variance and
the square of its expectation, convergence in L2 is equivalent to EXn EX
and var(Xn X) 0 as n .
Lemma 5.3. For any sequence of partitions 0 = tn0 < . . . < tnkn = t of [0, t]
2 L 2
Pkn 
t as n .
Wtni Wtni1 
with maxi (tni tni1 ) 0, we have i=1
Proof. The increments Wtni Wtni1 of Brownian motion over the partition are
independent random variables with N (0, tni tni1 )-distributions. Therefore,
E

var

kn
X

|Wtni Wtni1 |2 =

kn
X
(tni tni1 ) = t,

i=1

i=1

kn
X

kn
X

|Wtni Wtni1 |2 =

i=1

var Wtni Wtni1

2

i=1

=2

kn
X
(tni tni1 )2 2t max |tni tni1 | 0,
i

i=1

as n . Here we used that var(Z 2 ) = 2 4 for Z N (0, 2 ).

Because the quadratic variation of a continuously differentiable function
tends to zero as the mesh width of the partition tends to zero, the sample
paths of Brownian motion cannot be continuously differentiable. Otherwise the
limit in the lemma would have been 0, rather than t. This expresses that the
sample paths of Brownian motion possess a certain roughness. This is nice if
we want to use Brownian motion as a model for irregular processes, such as
the Brownian motion of particles in a fluid or gas, or a financial process, but it
complicates the use of ordinary calculus in connection to Brownian motion.
For instance, if W were the price of an asset and we would have ti1 assets
in our portfolio during the time interval (ti1 , ti ], then our increase in wealth
due to changes in value of the asset during the full interval (0, t] would be
X
X
ti1 (Wti Wti1 ) =
ti1 Wti .
i

We would like to extend this to portfolios t that changeR continuously in time,

and thus would like to be able to compute somethingR like t dWt .R For a continuously differentiable function f we would interpret (t) df (t) as (t) f 0 (t) dt.
Because
R the sample paths of Brownian motion do not have derivatives, we cannot
define t dWt in this way. Stochastic integrals provide a precise mathematical
framework for this.
27

5.7

Stochastic Integrals

Let W be a Brownian motion relative to a given filtration (Ft )t0 . We define

Rt
an integral 0 Xs dWs for given stochastic processes X defined on the same
probability space as W in steps:
R
(a) If Xt = 1(u,v] (t)A for a random
variable A Fu , then Xs dWs is the


random variable Wv Wu A.
R
P (i)
P R (i)
(b) If Xs = i Xs , then Xs dWs = i Xs dWs .
R (n)
R
(c) If E (Xs Xs )2 ds 0 for some sequence X (n) , then Xs dWs is the
R (n)
L2 -limit of the sequence Xs dWs .
Rt
R
(d) 0 Xs dWs = (1(0,t] (s)Xs ) dWs .
Rt
The following theorem shows that the integral 0 Xs dWs can be defined by this
Rt
procedure for any adapted process X with E 0 Xs2 ds < . If this integral is
finite for all t > 0, Rthen we obtain a stochastic process denoted by X W and
t
given by X W = ( 0 Xs dWs : t 0).
Rt
Theorem 5.4. Let X be adapted and satisfy E 0 Xs2 ds < for every t 0.
Rt
Then 0 Xs dWs can be defined through steps (a)(d) and
Rt
(i) E 0 Xs dWs = 0,

2
Rt
Rt
(ii) E 0 Xs dWs = E 0 Xs2 ds,
Rt
(iii) the process ( 0 Xs dWs : t 0) is a martingale.
A sketch of the proof of the theorem is given in Section 5.14. The interested
reader is also referred to e.g. the book by Chung and Williams. TheR intuition
t
behind assertion (iii) is that the increments Xsi Wsi of the integral 0 Xs dWs
satisfy


E Xsi Wsi+1 |Fsi = Xsi E(Wsi+1 |Fsi ) = 0,
because the increments of Brownian motion have mean zero and are independent
of the past. This reasoning is insightful, perhaps more so than the formal proof.
As a mathematical justification it is wrong, because the integral is a much more
complicated object than a sum of (infinitesimal) increments.

5.8

Geometric Brownian Motion

The state Wt of a Brownian motion at time t is normally distributed with

mean zero and hence is negative with probability 1/2. This is an embarrassing
property for a model of an asset price. One way out of this difficulty would be
to model the asset prices as the sum f (t) + Wt of a deterministic function and
Brownian motion. Brownian motion with a linear drift, the process + t + Wt ,
is a special example. This type of model uses Brownian motion as a noisy
aberration of a deterministic asset price f (t). If the deterministic function
satisfies f (t)  0, then the probability that f (t) + Wt is negative is very small,
but still positive.
Another way out is to model the asset price as a geometric Brownian motion,
which is given by
St = eWt ++t = S0 eWt +t ,

t 0,

with S0 = e . Putting the process in the exponential certainly makes it positive.

28

5.9

Stochastic Differential Equations

A more general approach to modeling using Brownian motion is in terms of

differential equations. An asset price process could be postulated to satisfy, for
given stochastic processes and ,
Z t
Z t
(5.5)
St = S0 +
s ds +
s dWs .
0

This integral equation is usually written in differential form as

(5.6)

dSt = t dt + t dWt .

In the case t = 0, (5.6) reduces to the ordinary differential equation dSt = t dt.
Adding the term t dWt introduces a random perturbation of this differential
equation. The infinitesimal change dSt in St is equal to t dt plus a noise term.
Because the increments of Brownian motion are independent, we interpret the
elements dWt asRindependent noise variables.
t
The integral 0 s dWs in (5.5) must of course be interpreted as a stochastic
Rt
integral in the sense of Section 5.7, whereas the integral 0 s ds is an ordinary
integral, as in calculus. The stochastic differential equation (5.6), or SDE, is
merely another way of writing (5.5), the latter integral equation being its only
mathematical interpretation. The understanding of dWt as a random noise
variable is helpful for
R t intuition, but does not make mathematical sense.
For the integral 0 s dWs to be well defined, the process must be adapted.
In many examples the process t and t are defined in terms of the process
S. For instance, a diffusion equation takes the form
(5.7)

dSt = (t, St ) dt + (t, St ) dWt ,

for given functions and on R [0, ). Then the stochastic differential

equation is recursive and the process St is only implicitly defined, and in fact
there is no guarantee that it exists. Just as for ordinary differential equations,
existence of solutions for stochastic differential equations is an important subject
of study. There are several general theorems that guarantee the existence of
solutions under certain conditions, but we omit a discussion.

5.10

Markov Processes

A Markov Process X is a stochastic process with the property that for every
s < t the conditional distribution of Xt given (Xu : u s) is the same as the
conditional distribution of Xt given Xs . In other words, given the present Xs
the past (Xu : u s) gives no additional information about the future Xt .
Example 5.8 (Brownian motion). Because Wt = Wt Ws +Ws and Wt Ws is
normal N (0, t s) distributed and independent of (Wu : u s), the conditional
distribution of Wt given (Wu : u s) is normal N (Ws , t s) and hence depends
on Ws only. Therefore, Brownian motion is a Markov process.
Example 5.9 (Diffusions). A diffusion process S, as given by the SDE (5.7),
does not possess independent increments as Brownian motion. However, in an
29

infinitesimal sense the increments dSt depend only on St and the infinitesimal
increment dWt , which is independent of the past. This intuitive understanding
of the evolution suggests that a diffusion process may be Markovian. This is
indeed the case, under some technical conditions.

5.11

Quadratic variation revisited

In Section 5.6 we have seen that the quadratic variation of Brownian motion
converges to a limit as the mesh width of the partitions tends to zero. This is
true for general solutions to SDEs, except that in general the convergence must
be interpreted in probability. We say that a sequence of random variables Xn
converges in probability to a random variable X if, as n ,


P |Xn X| > 0, for every > 0.
P

This is denoted by Xn
X. Denote also Yn = op (1) if Yn
0.
Lemma 5.10. Let a stochastic process S satisfy the SDE (5.6) for adapted
processes and . Then for any sequence of partitions 0 = t0,n < . . . < tkn ,n = t
of the interval [0, t] with maxi (ti,n ti1,n ) 0 as n , we have that
Z t
kn
X
2 P



Sti,n Sti1,n

s2 ds as n .
Vn =
0

i=1

Sketch of the proof. We omit the Rindex n in the

R tinotation of the partition points.
ti
The increments are Sti Sti1 = ti1
s ds+ ti1
s dWs . The first term on the
right is an ordinary integral and gives no contribution
to the quadratic variation,
P
since the sum of its squares is of the order i (ti ti1 )2 0. The second term
Pkn 2
is approximately ti1 (Wti Wti1 ). Denote for brevity An = i=1
ti1 (Wti
Pkn 2
2
Wti1 ) and an = i=1 ti1 (ti ti1 ). We thus have Vn = An + op (1). Now
E(An an ) = 0 and
var(An an ) =

kn
X




var t2i1 (Wti Wti1 )2 (ti ti1 )

i=1

kn
X

kn
X

2
Et4i1 E (Wti Wti1 )2 (ti ti1 ) = 2
Et4i1 (ti ti1 )2 0.

i=1

i=1

Here we used the relevant properties of Brownian motion. Together these equaP
tions suggest that E(An an )2 0, which in turn implies An an
0.
Combined with the convergence
an =

kn
X

t2i1 (ti ti1 )

i=1

s2 ds,

P Rt
this would give the result: Vn = An + op (1) = an + An an + op (1)
0 s2 ds.
This proof can be made precise without much difficulty if the process is
bounded and left-continuous. For a complete proof we need to use a truncation
argument involving stopping times.

30

2
Pkn 
Sti,n Sti1,n  is called the quadratic
The limit of the sums of squares i=1
variation of the process S. It is denoted by [S]t , and also known as the square
Rt
bracket process. For a solution S to the SDE (5.6) we have [S]t = 0 s2 ds.
Besides the quadratic variation of a single process, there is also a cross
quadratic variation of a pair of processes R and S, defined as the limit (in
probability)
[R, S]t = lim

kn
X

Rti,n Rti1,n

Sti,n Sti1,n .

i=1

EXERCISE 5.11. Suppose that the processes R and S both satisfy an SDE
(5.6), but with different functions and . Guess [R, S] if
(i) R and S depend in (5.6) on the same Brownian motion.
(ii) the SDEs for R and S are driven by independent Brownian motions.

5.12

It
o Formula

The geometric Brownian motion is actually a special case of the SDE approach.
By a celebrated formula of It
o it can be shown that geometric Brownian motion
satisfies a SDE.
It
os formula is a chain rule for stochastic processes, but due to the special
nature of stochastic integrals it takes a surprising form. The version of Itos
formula we present here says that a transformation f (St ) of a process that
satisfies an SDE by a smooth function f again satisfies an SDE, and gives an
explicit expression for it.
Recall that for a stochastic process S as in (5.6), the quadratic variation is
the process [S] such that d[S]t = t2 dt.
Theorem 5.12 (It
os formula). If the stochastic process S satisfies the SDE
(5.6) and f : R R is twice continuously differentiable, then
df (St ) = f 0 (St ) dSt + 12 f 00 (St ) d[S]t .
Sketch of the proof. For a sequence of sufficiently fine partitions 0 = tn0 < tn1 <
< tnkn = t of the interval [0, t] with maxi (tni tni1 ) 0 we have
f (St ) f (S0 ) =

kn
X

f (Stni ) f (Stni1 )

i=1

kn
X

f 0 (Stni1 )(Stni ) Stni1 ) +

i=1

1
2

kn
X

f 00 (Stni1 )(Stni Stni1 )2 .

i=1

Rt
The first term tends to the stochastic integral 0 f 0 (Ss ) dSs . By the same arguRt
ments as used in Section 5.11, the second sum tends to 21 0 f 00 (Ss )s2 ds.
We have written It
os formula in differential form, but as usually it should
be mathematically interpreted as a statement about integrals.
The striking aspect of It
os formula is the second term 12 f 00 (St ) d[S]t , which
would not appear if the sample path t 7 St were a differentiable function. As
the proof shows it does appear, because the variation of the sample paths of S
is not finite, whereas the quadratic variation tends to a nontrivial limit.
31

Example 5.13. Brownian motion W itself certainly satisfies a stochastic differential equation: the trivial one dWt = dWt .
Applied with the function f (x) = x2 Itos formula gives dWt2 = 2Wt dWt +
Rt
1
conclude that Wt2 = 2 0 Ws dWs +t. Compare this
2 2 dt, because [W ]t = t. We
R
t
to the formula f 2 (t) = 2 0 f (s) df (s) for a continuously differentiable function
f with f (0) = 0.
Example 5.14 (Geometric Brownian motion). As a consequence of Itos formula, the geometric Brownian motion St = exp( + t + Wt ) = S0 exp(t +
Wt ), with S0 = e , satisfies the SDE
dSt = ( + 21 2 )St dt + St dWt .
To see this, apply It
os formula with the process Xt = t+Wt and the function
f (x) = S0 exp(x).
It
os theorem is also valid for functions of more than one process. For instance, consider a process f (t, St ) for a function f : [0, ) R R of two arguments. Write ft , fs and fss for the partial derivatives /tf (t, s), /sf (t, s)
and 2 /s2 f (t, s), respectively.
Theorem 5.15 (It
os formula). If the stochastic process S satisfies the SDE
(5.6) and f : [0, ) R R is twice continuouly differentiable, then
df (t, St ) = ft (t, St ) dt + fs (t, St ) dSt + 12 fss (t, St ) d[S]t .
As a second example consider a process f (Rt , St ) of two stochastic processes
R and S. If an index r or s denotes partial differentiation with respect to r or
s, then we obtain the following formula.
Theorem 5.16 (It
os formula). If the stochastic processes R and S satisfy the
SDE (5.6) and f : R2 R is twice continuously differentiable, then
df (Rt , St ) = fr (Rt , St ) dRt + fs (Rt , St ) dSt + 21 frr (Rt , St ) d[R]t
+ 12 fss (t, St ) d[S]t + frs (Rt , St ) d[R, S]t .

5.13

Girsanovs Theorem

Rt
The stochastic integral 0 Xs dWs of an adapted process relative to Brownian
motion is a (local) martingale. Thus the solution S to the SDE (5.6) is the sum
Rt
Rt
of a local martingale 0 s dWs and the process At = 0 s ds. The sample paths
of the process A are the primitive functions, in the sense of ordinary calculus,
of the sample paths of the process , and are therefore differentiable. They
are referred to as drift functions. The presence of a drift function destroys the
martingale property: a solution of an SDE can be a martingale only if the drift
is zero.
Rt
Rt
Lemma 5.17. The process S = (St )t0 defined by St = 0 s ds + 0 s dWs
is a local martingale if and only if = 0.

32

Sketch of the proof. If = 0, S is a martingale. Let us prove Rthe converse.

t
If S is a local martingale, then so is the process At = 0 s ds, because
Rt
the process 0 s dWs is a local martingale and the difference of two local martingales is a local martingale. Because the sample paths of A are differentiable, the rules of ordinary calculus apply, and yield that d(A2t ) = 2At dAt ,
Rt
or A2t = 0 2As dAs . The local martingale property of A carries over to every
Rt
process of the form 0 Xs dAs for an adapted process X. This can be proved by
first considering simple adapted processes, and next limits, along the same lines
as the martingale property of stochastic integrals was proved. In particular,
we may choose X = 2A, and we see that the process A2 is a local martingale.
If it is a martingale with finite second moments, then we can conclude that
EA2t = EA20 = 0, whence A = 0. The general case can be handled by a stopping
time argument.
EXERCISE 5.18. Let (Wt )t0 be a Brownian motion and (t), (t) be adapted
processes. Show that a generalized geometric Brownian motion (St )t0 , with

Rt
Rt
St = S0 exp 0 (s)dWs 0 ((s) + 21 2 (s))ds , is a martingale if (t) 0.
EXERCISE 5.19. Let (Wt )t0 be a Brownian motion and g(t) be a deterRt
Rt
ministic function such that 2 (t) = 0 g 2 (s)ds < . Show that 0 g(s)dWs
N (0, 2 (t)). (Hint: use the previous exercise and the fact that Z N (0, 2 ) iff
2 2
EeZ = e /2 for all R.)
The martingale property refers to the underlying probability distribution on
the outcome space. Therefore a process may well be a martingale relative to a
probability measure Q, whereas it is not a martingale if the outcome space is
equipped with another probability measure P. If the process is given by an SDE
under P, then this somehow means that the drift of the process can be made to
disappear by changing the probability distribution on the space of outcomes.
This observation turns out to be crucial in the pricing theory.
R t It will be sufficient to consider this for the case that = 1, i.e. St = Wt +
ds. If W is a Brownian motion under the probability measure P, then W
0 s
is a martingale under P and hence S cannot be a martingale under P, (unless
= 0). Girsanovs theorem shows that for most processes there exists
another probability measure Q such that S is a martingale, and even a Brownian
motion, under Q.
Theorem 5.20 (Girsanov). If (Wt : 0 t T ) is a Brownian motion under the
RT
probability measure P and is an adapted process with E exp( 21 0 2s ds) < ,
Rt
then there exists a probability measure Q such that the process (Wt + 0 s ds :
0 t T ) is a Brownian motion under Q.
There is even a constructive formula for finding the martingale measure Q
from P, given by
RT
1 RT 2 

Q(A) = E 1A e 0 s dWs 2 0 s ds ,
where the expectation is computed under the probability measure P. The conRT
dition E exp( 21 0 2s ds) < ensures that the formula in the preceding display
indeed defines a probability measure Q. If the process is bounded (e.g. constant), then the condition is clearly satisfied. In general, the condition says that
should not grow too big.
33

5.14

Brownian Representation

Let (Ft )t0 be the natural filtration of a given Brownian motion W . Stochastic
processes defined on the same outcome space that are martingales relative to
this Brownian filtration are referred to as Brownian martingales. Brownian
motion itself is an example, and so are all stochastic integrals X B for adapted
processes X.
The following theorem shows that these are the only Brownian martingales.
Theorem 5.21. Let {Ft } be the (completion of the) natural filtration of a
Brownian motion process W . If M is a (cadlag) local martingale relative to
Rt
{Ft }, then there exists a predictable process X with 0 Xs2 ds < almost
Rt
surely for every t 0 such that Mt = M0 + 0 Xs dWs .
This Brownian representation theorem remains true if the filtration is generated by multiple, independent Brownian motion processes W (1) , W (2) , . . . , W (d) .
Then an arbitrary (cadlag) local martingale can be written as Mt = M0 +
Pd R t (i)
(i)
i=1 0 Xs dWs .

*Proof of Theorem 5.4

In this section we provide details for the construction of the stochastic integral
in Section 5.7. Because this material is mathematically quite involved, we do
not give a full proof of Theorem 5.4, but we indicate the most essential steps.
A process of the type Xt = 1(u,v] (t)A for a random variable A Fu as in
(a) is adapted and hence so is process of the type
X
Xt =
1(ui ,vi ] (t)Aui ,
ui < vi , Aui Fui .
i

A process of this type is called simple adapted. By splitting up sets if necessary,

it is always possible to represent such a simple adapted process with disjoint
intervals (ui , vi ]. For X as in the preceding display we define
Z
X
(5.22)
Xs dWs =
Aui (Wvi Wui ).
i

Because the representation of X in terms of the intervals (ui , vi ] and Aui is not
unique (we could for instance split up the intervals further), it must be verified
that this definition is consistent, but we omit this part of the proof.
We next verify property (ii) for simple adapted processes X
Pand t = . 2If,
as we assume, the intervals (ui , vi ] are disjoint, then Xt2 =
i 1(ui ,vi ] (t)Aui .
Therefore, the right side of (ii) with t = is equal to
Z
Z X
X
E Xs2 ds = E
1(ui ,vi ] (t)A2ui ds =
EA2ui (vi ui ).
i

The left side of (ii) with t = is given by, in view of (5.22),

Z
XX

2
E
Xs dWs = E
Aui Auj (Wvi Wui )(Wvj Wuj ).
i

34

Because the intervals (ui , vi ] are disjoint, we have that E(Wvi Wui )(Wvj
Wuj ) = 0 for i 6= j, by the independence and the zero means of the increments
of Brownian motion. It follows that the diagonal terms in the double sum vanish,
whence the preceding display is equal to
X
X
X
E
A2ui (Wvi Wui )2 =
EA2ui E(Wvi Wui )2 =
A2ui (vi ui ),
i

where in the second step we use the independence of the increment Wvi Wui of
Fui and hence of Aui . Thus we have verified that for simple adapted processes
X
Z
Z

2
E
Xs dWs = E Xs2 ds.
In words we have shown that the integral is a linear isometry. A linear
isometry between normed spaces X and Y is a linear map I : X Y such that
kI(x)kY = kxkX for every x X. This isometry is the basis for the extension of
the integral to general adapted processes, by way of the following result from
analysis.
Any linear isometry I : X0 X Y from a linear subspace X0
of a normed space X into a complete normed space Y possesses
 a
0 = X
unique extension to an isometry defined on
the
closure
X

X : {Xn } X0 such that kXn Xk 0 of X0 in X.
In our situation
we take the space X equal to all adapted processes X with
Rt
kXk2X = E 0 Xs2 ds < , and X0 equal to the Rcollection of all simple adapted
processes. We have seen that the map I : X 7 Xs dWs is an isometry into the
set Y of random variables with finite second moments, with kY k2Y = EY 2 . Thus
the integral can be extended to the closure of the set of simple Radapted processes.
t
That this closure is the set of all adapted processes with E 0 Xs2 ds < can
be shown by approximation by step functions. We omit the details of this part
of the proof.
Thus the integral is defined. The verification of its properties (i)(iii) proceeds by first verifying that these assertions hold on the set X0 of simple processes and next showing that these properties are preserved under taking limits.
For a simple adapted process of the form Xt = 1(u,v] (t)A with (A) Fu
and s < t we have
(
Z t
Z s
A(Wtv Wsu ), if t v > s u,
Xr dWr
Xr dWr =
0,
otherwise.
0
0
Because A is known at time u s u and Brownian motion is a martingale we
have E(A(Wtv Wsu )|Fsu ) = AE(Wtv Wsu |Fsu ) = 0. Therefore, with
the help of the tower property of conditional expectation, it follows that
Z s
Z t

E
Xr dWr
Xr dWr |Fs = 0.
0

Rt

Thus the stochastic integral 0 Xs dWs is a martingale for X of this form. Because the sum of two martingales is again a martingale, this conclusion extends
to all simple adapted processes. Because the martingale property is preserved
under taking L2 -limits, it next extends to the stochastic integral in general.
35

*Stopping
Stopping times are intuitively meaningful objects that have interest on their
own, and are also essential for extension of the definition of stochastic integrals,
as given in the next section. However, we shall not need the material in this
and the following section in later chapters.
A stopping time relative to a filtration (Ft )t0 is a random variable T with
values in [0, ] such that {T t} Ft for every t 0. A stopping time
formalizes a strategy to play (or invest) on the market until a given time, which
need not be predetermined, but may be based on observing the market. The
requirement that the event {T t} is known at time t says that the decision to
stop trading must be made based on information collected in past and present.
If the filtration is generated by a process X, then this requirement implies that
the decision to stop at time t must be based on the sample path of X until t.
Example 5.23 (Hitting time). If X is an adapted process with continuous
sample paths, then T = inf{t 0 : Xt B} is a stopping time for every (Borel)
set B. This is known as the hitting time of B.
Stopping times are important tools in the theory of stochastic processes,
but are also crucial to evaluate American options. These are contracts that give
the holder the right to collect a certain payment at a time t in a given interval
[0, T ] of his own choosing. The amount of the payment depends on the history
of an asset price up to the time of payment. For instance, an American call
option on an asset with price process S gives the right to buy the asset at a
predetermined price K at any time t in an interval [0, T ]. This corresponds
to a payment of (St K)+ at the chosen time t. The financial problem is to
determine an optimal stopping time for the payment, and to evaluate the value
of the resulting contract.
Given a stopping time T and a stochastic process the stopped process X T is
defined as the stochastic process such that
(X T )t = XT t .
The sample paths of the stopped process are identical to the sample paths X
up to time T and take the constant value XT for t T .
Theorem 5.24. If X is a martingale, then so is X T .
More explicitly, the theorem says that, if X is a martingale, then
E(XT t |Fs ) = XT s ,

s < t.

In particular, we have EXT t = EXT s = EX0 , because we can choose s = 0.

If T is a bounded stopping time, then we may choose t T and we find
EXT = EX0 .
This says that stopping does not help if the pay-off process X is a martingale.
No matter how clever the stopping strategy T , the expected pay-off EXT is
EX0 .

36

Example 5.25. The process Wt2 t is a martingale. It can be shown that

T = inf{t 0 : |Wt | = a} is finite almost surely, whence WT2 = a2 . The identity
E(WT2 T ) = E(W02 0) = 0 reduces to ET = a2 .
However, it is not permitted to apply this identity directly, as T is not a
bounded stopping time. A way around this is to apply the identity with T n
for a given n and next take limits as n . Because T n is bounded we
find EWT2 n = E(T n). Because WT2 n a2 , we have EWT2 n EWT2 = a as
n by the dominated convergence theorem. Also we have ET n ET by
the monotone convergence theorem. Thus the formula ET = a2 is correct.
EXERCISE 5.26. For given a > 0, let T = inf{t 0 : Wt = a}.
1

(i) Show that Yt = eWt 2 t is a martingale, for every R.

(ii) Show that E exp(T ) = exp( 2a).

(iii) Show that ET = .

*Extended Stochastic Integrals

Using stopping times we can define a useful extension of the definition
R t of the
stochastic integral. We have already defined the stochastic integral 0 Xs dWs
Rt
for any adapted process X with E 0 Xs2 ds < . We shall now extend this to
all adapted processes X such that
Z
(5.27)
Xs2 ds < a.s.
This is a larger set of adapted processes, as finiteness of the expected value of a
positive random variable implies finiteness of the variable with probability one,
but not the other way around.
We truncate a given adapted process by stopping it appropriately. For a
given n we define the stopping time
Z t
n
o
Tn = inf t 0 :
Xs2 ds n .
0

Rt
The finiteness (5.27) of the (nondecreasing) process 0 Xs2 ds implies that Tn
Rt
as n . From the definition of Tn it follows immediately that 0 Xs2 ds n
Rt
R tT
if t Tn . Consequently E 0 (Xs 1sTn )2 ds = E 0 n Xs2 ds En = n < .
We can therefore define, for every n and t,
Z t
(Xs 1sTn ) dWs .
0

Rt

We define 0 Xs dWs as the limit of these variables, in the almost sure sense, as
n . It can be shown that this limit indeed exists.
Each of the Rprocesses in the preceding display is a martingale. The stochastic
t
integral Yt = 0 Xs Ws is the limit of these martingales, but need not be a
martingale itself. (The limit is only in an almost sure sense, and this is not strong
enough to preserve the martingale property.) However, the stopped process Y Tn
Rt
is exactly the the integral 0 (Xs 1sTn ) dWs and hence is a martingale. This has
Rt
gained the stochastic integral 0 Xs dWs the name of being a local martingale.
37

Chapter 6

Black-Scholes Model
In this chapter we assume that we can trade continuously in a (riskless) bond
and some risky asset, for instance a stock. We assume that the bond price B
evolves as
Bt = ert ,
where r is the riskless interest rate. The price process S of the risky asset is
assumed to be a geometric Brownian motion, i.e.
St = S0 et+Wt ,

t 0.

Here W = (Wt )t0 is a Brownian motion, R is called the drift of the process,
and the volatility. We denote by (Ft )t0 the filtration generated by the price
process S. Observe that (Ft )t0 is also the natural filtration of the Brownian
motion W , since both processes generate the same flow of information.
For some fixed T > 0, let C be a FT -measurable random variable, i.e., its
value is determined by the information up till time T . We think of C as the
pay-off at time T of some contingent claim. For technical reasons, we assume
that EC 2 < . We want to answer the same question as in the discrete-time
setup: What is the fair price of the claim C at time zero?
To answer this question we follow the same route as in Chapter 4. We first
use Girsanovs theorem to change the underlying probability measure in such
a way that the discounted asset price St = ert St becomes a martingale under
the new measure Q. Then we consider the Q-martingale Vt = EQ (erT C|Ft )
and use the representation theorem to construct a self-financing trading strategy
that replicates the pay-off C. By an arbitrage argument, the value of this trading
portfolio at time zero must be the fair price of the claim. As in the binomial
model, the fair price of the claim at time t = 0 turns out to be EQ erT C, the
expectation under the martingale measure of the discounted pay-off.

6.1

Portfolios

Before we can carry out the programme outlined in the preceding section we
have to give a mathematically precise definition of a self-financing portfolio in
the present continuous-time setting. A portfolio is just a pair of predictable
processes (t , t ). We interpret t as the number of risky assets held at time t,

38

and t as the number of bonds. Predictability roughly means that to determine

the positions t and t , only the information available before time t is used.
For technical reasons we assume that almost surely,
Z T
Z T
2
|t | dt +
|t | dt < .
0

With the portfolio (, ) we associate the value process V = (Vt )t0 defined by
Vt = t St + t Bt .
For hedging strategies we need the notion of a self-financing portfolio. Such
a portfolio is created using some starting capital at time zero, and after time
zero the portfolio is only changed by rebalancing, i.e., by replacing bonds by
the risky asset or vice versa. No additional injections or withdrawals of money
are allowed. Loosely speaking, such a portfolio has the property that in an
infinitesimally small time interval [t, t + dt], the changes in the portfolio value
are only caused by changes in the price processes S and B, and not by changes
in t and t which are due to injections or withdrawals of money. Therefore,
we call a portfolio (, ) self-financing if its price process V satisfies the SDE
dVt = t dSt + t dBt .
A replicating, or hedging portfolio for the claim C is a self-financing portfolio
(, ) with a value process V which satisfies VT = C. If such a portfolio exists,
then an arbitrage argument shows that the fair price of the claim at time
t [0, T ] equals the value Vt of the portfolio.
Of course, the arbitrage argument is an economic one, and not a mathematical argument. When we use the phrase fair price in mathematical theorems
below, the fair price or value will always be understood to be defined as
the value process of a replicating portfolio. (We shall be a bit careless about
the still open trap that there may be more than one replicating portfolios, with
different value processes.)

6.2

The Fair Price of a Derivative

Let us now derive the pricing formula for the derivative C announced in the
previous section.
The discounted asset price is St = Bt1 St = ert St = S0 exp{(r)t+Wt },
whence it is a geometric Brownian motion with drift r and volatility . By
Example 5.14, it satisfies the SDE
dSt = ( r + 12 2 )St dt + St dWt .
t = Wt + t( r + 1 2 )/, this simplifies to
If we define W
2
(6.1)

t.
dSt = St dW

By Girsanovs theorem, there exists a new underlying probability measure Q

is a Brownian motion under Q. Hence, the preceding SDE implies
such that W
that the process S = (St )t0 is a Q-martingale.
39

Now consider the process Vt = EQ (erT C|Ft ), t [0, T ]. By the tower property of conditional expectations, this is a Q-martingale relative to the filtration
(Ft )0tT . It is obvious that the natural filtration (Ft ) of W is also the natural
, the processes generate the same flow of information.
filtration of the process W
Hence, by the Brownian representation theorem, Theorem 5.21, there exists a
t . Defining t = t / St , we have
predictable process such that dVt = t dW

dVt = t St dWt and, by (6.1),

(6.2)

dVt = t dSt .

Next we introduce the process t = Vt t St . Then Vt = t St + t .

We claim that (, ) is a hedging portfolio for the derivative C. To prove
this, consider the value process V of the portfolio (, ): Vt = t St + t Bt . By
construction we have
Vt = t St + t Bt = Bt (t St + t ) = Bt Vt .
In particular, VT = B0 V0 = EQ (C|FT ) = C, and indeed the portfolio has the
value C at time T . To prove that it is self-financing, the result of the following
exercise (integration by parts formula) is useful.
EXERCISE 6.3. Use It
os formula to show that if X satisfies an SDE and F is
a differentiable (deterministic) function, then d(F (t)Xt ) = F (t) dXt + Xt dF (t).
Now we can compute dVt . By the result of the exercise we have
dVt = d(Bt Vt ) = Vt dBt + Bt dVt .
If we use the definition of t to rewrite the first term on the right-hand side,
use (6.2) to rewrite the second term and recall that ert = Bt , we find that
dVt = (t St + t )dBt + t Bt dSt = t (St dBt + Bt dSt ) + t dBt .
By the result of the exercise, dSt = d(Bt St ) = St dBt + Bt dSt . Hence
dVt = t dSt + t dBt ,
which shows that the constructed portfolio (, ) is indeed self-financing.
In view of the standard arbitrage argument, the fair price of the claim C
at time t is given by Vt = Bt Vt = ert Vt = EQ (er(T t) C|Ft ). Hence, we have
proved the following theorem.
Theorem 6.4. The value (fair price) of the FT -measurable claim C at time
t [0, T ] is given by Vt = EQ (er(T t) C|Ft ), where Q is the measure under
which the discounted stockprice process St = ert St , t [0, T ], is a martingale.
In particular, the price at time t = 0 is EQ erT C.
In the above approach, we introduce a Q-martingale V and used it to construct a self-financing portfolio that also replicates the claim C. Then we derived
the fair price of the claim as the value process of the constructed portfolio.
On the other hand, suppose now we have a self-financing portfolio that
replicates the claim. According to the Black-Scholes paradigm, the fair price of
the claim is the value process of the portfolio.

40

EXERCISE 6.5. Show that a portfolio process (, ) with the value process V
is self-financing iff (6.2) holds for the discounted value process Vt = ert Vt .
Since S is a martingale under Q, a self-financing portfolio makes the discounted value process V a martingale under Q, by Exercise 6.5. This establishes
again Theorem 6.4.
EXERCISE 6.6. Let V be the value process of a self-financing portfolio that
replicates a simple claim C = C(ST ). Show that there exists a process g such
t.
that V satisfies dVt = rVt dt + gt dW
Let us turn to arbitrage portfolios. By definition these are self-financing and
such that the corresponding value process V satisfies P(V0 = 0) = 1, P(VT
0) = 1 and P(VT > 0) > 0. Clearly, we can equivalently rephrase the latter
three conditions in terms of the discounted valued process V , P(V0 = 0) = 1,
P(VT 0) = 1 and P(VT > 0) > 0. As before, a market is arbitrage free, if no
arbitrage portfolios exist.
EXERCISE 6.7. Show that Black-Scholes market is arbitrage free (you may
use the fact that the martingale measure Q and the measure P are equivalent).

6.3

European Options

If the claim C is European, meaning that it is of the form C = f (ST ) for some
function f , then we can derive a more explicit formula for its fair price.
Recall that
2
St = S0 et+Wt = S0 e(r

2 )t+ Wt ,

t = Wt + t( r + 2 )/ is a Brownian motion under Q. So, under

where W
2
Q, the stockprice process S is also a geometric Brownian motion, with drift
r 2 /2 and volatility . In particular, we have under Q that
ST = St e(r

2
2 )(T t)+ T tZ ,

t [0, T ],

where Z N (0, 1), independent of St . It follows that the value process is

Vt = EQ (er(T t) C|Ft ) = EQ (er(T t) f (ST )|Ft )


2

 
= er(T t) EQ f St e(r 2 )(T t)+ T tZ Ft = F (t, St ),
where
(6.8)

er(T t)
F (t, x) =
2


 z2

2
f xe(r 2 )(T t)+z T t e 2 dz.

Thus, we have proved the following theorem.

Theorem 6.9. The fair price of a European claim C = f (ST ) at time t [0, T ]
is Vt = F (t, St ), where F (t, St ) is defined by (6.8).
For a given f , it is typically not possible to evaluate the above integral analytically and one has to resort to numerical integration. However, we can derive
explicit expressions for the prices of European calls and puts. The European

41

call option with strike K and maturity T corresponds to f (x) = (x K)+ . By

using Theorem 6.9 and Exercise 4.20, we compute the price of the call option:




Vt = St d1 (T t), St ) Ker(T t) d2 (T t, St ) , t [0, T ),
and VT = f (ST ), where (x) = P(Z x) for Z N (0, 1) and
d1 (, x) =

log(x/K) + (r + 2 /2)

d2 (, x) = d1 (, x) .

This is the celebrated Black-Scholes formula.

EXERCISE 6.10. Denote by Ct and Pt the prices at time t [0, T ] of European
call and put options respectively with the same maturity T and strike K. Derive
the put-call parity Ct Pt = St er(tT ) K. Use the call-put parity to derive
the Black-Scholes formula for the price of a European put option.

6.4

The Black-Scholes PDE and Hedging

The pricing function F (t, x) from Theorem 6.9 can also be obtained as the
solution of the so-called Black-Scholes partial differential equation (PDE). This
provides a second method for finding the price of a European claim. Usually
this PDE cannot be solved analytically and one has to use numerical methods.
Theorem 6.11. The value of a European claim C = f (ST ) at time t [0, T ]
is given by Vt = F (t, St ), where F is the solution of the partial differential
equation
2
Ft (t, x) + rxFx (t, x) + 2 x2 Fxx (t, x) rF (t, x) = 0,
subject to the boundary condition F (T, x) = f (x).
Proof. The function F is smooth in both arguments. Therefore, we can apply
It
os formula to the value process Vt = F (t, St ) to see that this satisfies the SDE
dVt = Ft (t, St ) dt + Fx (t, St ) dSt + 21 Fxx (t, St ) d[S]t .
t under Q, St = S0 e(r
In terms of a Browinian motion W
t,
dSt = rSt dt + St dW

2 )t+ Wt ,

so that

d[S]t = 2 St2 dt.

Substituting these identities in the preceding display for dVt gives

dVt = Ft (t, St ) + rSt Fx (t, St ) +

2 2
2 St Fxx (t, St )

t.
dt + Fx (t, St )St dW

t , so that
On the other hand, (6.1) and (6.2) imply that dVt = t St dW
Vt = ert Vt satisfies
t = rF (t, St ) dt + t St dW
t.
dVt = rert Vt dt + ert dVt = rVt dt + t ert St dW
Comparison of the dt-terms of the last two equations for dVt yields the PDE for
the function F , with the boundary condition F (T, ST ) = VT = C = f (ST ).

42

It should be noted that the PDE for the value process is the same for every
type of European option. The type of option is only important for the boundary
condition.
In the proof of the preceding theorem we only compared the dt-terms of
the two SDEs that we obtained for the value process V of the claim. By
t -terms, we obtain the following explicit formulas for the
comparing the dW
hedging portfolio of the claim.
Theorem 6.12. A European claim C = f (ST ) with value process Vt = F (t, St )
can be hedged by a self-financing portfolio consisting at time t of t risky assets
and t bonds, where


t = Fx (t, St ),
t = ert F (t, St ) Fx (t, St )St .
t -terms of the
Proof. The formula for t follows from the comparison of the dW
two SDEs for V that we obtained in the proof of the preceding theorem. Recall
that t = Vt t St . Substituting Vt = F (t, St ) and t = Fx (t, St ) yields the
formula for t .
The hedging strategy exhibited in the preceding theorem is called the delta
hedge for the claim. Note that in general the numbers of stocks and bonds in the
hedging portfolio change continuously. In practice it is of course not possible to
trade continuously. Moreover, very frequent trading will not always be sensible
in view of transaction costs. However, the delta hedge can be derived to indicate
what a hedging portfolio should look like.
In all exercises below we consider the standard Black-Scholes model.
EXERCISE 6.13. For a < b, consider the claim C(ST ) = K1(a,b) (ST ) (called
a binary spread ). Determine the fair price of this claim at any time t [0, T ].
EXERCISE 6.14. Consider a straddle claim with pay-off at maturity T equal
to C = |ST K|. Determine the fair price of the straddle at any time t [0, T ].
This claim can also be hedged with a constant portfolio that not only consists
of shares and bonds, but contains as a third component European call options
as well. Find this portfolio.
EXERCISE 6.15. Consider a bull spread claim with the following pay-off C =
min{max{ST , A}, B}, where B > A > 0. This claim can be hedged with a
constant portfolio consisting of stocks, bonds and European call options. Find
this portfolio and the price process.
EXERCISE 6.16 (Feynman-Kac theorem). Consider the following SDE: dXt =
(t, Xt )dt + (t, Xt )dWt , t [0, T ], for a fixed T > 0, X0 is independent of W .
Let h(x) be a Borel measurable function such that E|h(XT )| < . Then there
exists a function g(t, x) : R2 R, such that g(t, Xt ) = E(h(XT )|Ft ), t [0, T ],
Ft = (Ws , s t). Assume that g is twice continuously differentiable. Show
that g(t, x) satisfies the PDE gt (t, x)+(t, x)gx (t, x)+ 21 2 (t, x)gxx (t, x) = 0 with
the boundary condition g(T, x) = h(x). Formulate a version of Feynman-Kac
theorem also for the discounted function f (t, Xt ) = E(er(T t) h(XT )|Ft ).
EXERCISE 6.17. For a fixed T > 0, consider the PDE ft (t, x) + 2tfx (t, x) +
t4
2
2 fxx (t, x) = 0, with the boundary condition f (T, x) = x . Find an explicit
solution of this problem by using the Feynman-Kac theorem (see Exercise 6.16)
and verify this solution by straightforward computations.

43

EXERCISE 6.18. Generalize Theorems

R t 6.11 and 6.12 for composite claims of
the form C = f (ST , ZT ), with Zt = 0 g(u, Su ) du for some g : R2 R.
RT
EXERCISE 6.19. Consider the claim C = 0 Su du. Determine the fair price
Vt of this claim for each t [0, T ]. Find a hedging strategy for this claim. For
Rt
Zt = 0 Su du, write Vt = F (t, St , Zt ) and determine a PDE for F (t, x, z).

6.5

The Greeks

Parties which are trading a claim with associated value function Vt = F (t, St )
are often interested in the sensitivity of the price of the claim with respect to
the price of the underlying asses, and also with respect to time t, interest rate r
and volatility . Reasonable measures for these sensitivities are the derivatives
of the function F (t, x). These derivatives have special names. The quantities
= Fx ,

= Fxx ,

= Fr ,

= Ft ,

V = F

are called the delta, gamma, rho, theta and vega of the claim, respectively.
Together they are called the Greeks.
For instance, the delta of a claim measures the first order dependence of the
price of the claim relative to the price of the underlying asset. A very high delta
means that small changes in the asset price cause relatively large changes in the
value of the claim. Observe that the delta is precisely the number of stocks in
the hedging portfolio of Theorem 6.12.
EXERCISE 6.20. Calculate the Greeks for the European call option and give
the delta hedging strategy for the claim.

6.6

General Claims

If the claim C FT is not of European type, we typically have no nice closed

form expression for its price. In that case, one can use simulation to find the
approximate price. The price of the claim is given by EQ erT C and in the
preceding section we saw that under the martingale measure Q
St = S0 e(r

2 )t+ Wt ,

is a Brownian motion. To approximate the price, the following procewhere W

dure can be followed:
1) Simulate a large number, say n, of realizations of the process S under Q.
2) For each realization, compute the corresponding payoff of P
the claim, yieldn
ing n numbers C1 , . . . , Cn . Compute the average cn = n1 i=1 Ci .
3) Then by the law of large numbers, the discounted average erT cn is a
good approximation for the price EQ erT C if n is large enough.
We can quantify the quality of the approximation by obtaining aP
confidence
n
1
interval. The sample standard deviation sn is defined by s2n = n1
i=1 (Ci
cn )2 . By the central limit theorem and the law of large numbers, we have the

44


convergence in distribution n(
cn EQ C)/sn
N (0, 1) as n . Hence, for
large n, we have the approximation

  c E C 
 n
Q 
Q n
 > 1.96 0.05.
sn
It follows that
 rT


e
(
cn 1.96sn / n), erT (
cn + 1.96sn / n)
is an approximate 95%-confidence interval for the price of the claim C. The
length of the interval tends to zero as n , which means that our simulation
scheme can achieve arbitrary accuracy if we simulate long enough. In practice
we shall be limited by computation time.

6.7

Exchange Rate Derivatives

Companies that do business in a country with a different currency are often

interested in reducing the risk due to uncertainty in the exchange rate. One
possibility to reduce this risk is to buy a suitable exchange rate derivative. For
instance, a Dutch company that will place a large order in the US one month
from now may want to have an option to buy a large number of dollars for a
specified price (in euros) at that time. In this section we use the developed
Black-Scholes theory to derive the fair price of such a derivative.
We assume that there exist dollar bonds in the US and we can trade in
euro bonds in The Netherlands. The prices of these bonds (in their respective
currencies) are supposed to be given by
Dt = eqt ,

Bt = ert ,

respectively, where r is the European interest rate and q is the US interest rate.
The exchange rate Et , the euro value of one dollar, is modelled as a geometric
Brownian motion for certain parameters , and a Brownian motion W :
Et = E0 et+Wt .
From the Dutch perspective, we can now trade in two assets: the riskless
euro bond B and the risky US bond S, which have (euro) price processes
Bt = ert ,

St = Et Dt = S0 e(q+)t+Wt ,

respectively. In other words, S is a geometric Brownian motion with drift q +

and volatility . From the point of view of a Dutch trader, this is just a standard
Black-Scholes market and we know how to price derivatives.
Consider for instance a contract giving a Dutch trader the right to buy one
US dollar for K euros at time T > 0. The payoff of this contract at time T in
euros is (ET K)+ . By the standard theory, the fair euro price of the contract
is erT EQ (ET K)+ , where Q is the martingale measure, under which the
discounted price process St = ert St is a martingale. Note that
erT EQ (ET K)+ = eqT EQ erT (ST KeqT )+ .
This is eqT times the standard Black-Scholes price of a European call option
with maturity T and strike KeqT . For this claim we have an explicit formula.
45

Chapter 7

Extended Black-Scholes
Models
The classical Black-Scholes model that we considered in the preceding chapter
can be extended in several directions. So far we only considered markets in
which a single bond and one risky asset are traded. We can also study the more
complex situation when there are several risky assets with price processes that
do not evolve independently. This allows the pricing of derivatives which depend
on the behaviour of several assets. The assumption of a constant drift and
volatility can also be relaxed. They can be replaced by arbitrary, predictable
stochastic processes.
In general we can consider a market in which a bond is traded with price
process B and n risky assets with price processes
S 1 , . . . , S n . We assume that
Rt
the bond price is of the form Bt = exp( 0 rs ds) for rt the interest rate at
time t, so that it satisfies the ordinary diffential equation
dBt = Bt rt dt,

B0 = 1.

The interest rate r may be an arbitrary predictable process and hence depend on
all information before time t. We assume that the asset price processes satisfy
the system of stochastic differential equations
(7.1)

dSti = it Sti dt + Sti

d
X

tij dWtj ,

i = 1, . . . , n,

j=1

where W 1 , . . . , W d are d independent Brownian motions and the i and ij are

predictable processes. Then the processes i model the drift, and the ij model
both the volatility and the dependence structure of the price processes.
Under certain conditions such general market models are also free of arbitrage and have the property that each claim that is a function of the asset prices
can be hedged by a self-financing trading strategy. Explicit pricing formulas are
usually not available in such general models. However, if the model is free of arbitrage and complete, the general fact that price is expectation of discounted
payoff under a martingale measure is still true. The SDEs satisfied by the
price processes under the martingale measure are typically easily obtained, so
the simulation method can be used to approximate claim prices. This requires
the simulation of solutions of multi-dimensional SDEs.
46

7.1

Market Price of Risk

The key structural condition needed to push through the theory is the existence
of a predictable, vector-valued process = (1 , . . . , d ), called the market price
of risk, such that
(7.2)

d
X

tij tj = rt it ,

i = 1, 2, . . . , n.

j=1

We can write this system of equations in vector form as t t = rt 1 t , and

hence the existence of the market price of risk process requires that the vector
rt 1t is contained in the range space of the (nd)-matrix t . This is immediate
if the rank of t is equal to the number n of stocks in the economy, as the range
of t is all of Rn in that case. If the rank of t is smaller than the number of
stocks, then existence of the market price of risk process requires a relationship
between the three parameters , r and . This situation is certain to arise
if the number of components of the driving Brownian motion is smaller than
the number of risky assets, i.e., d < n. Hence we can interpret the condition of
existence of a process as in the preceding display as implying that the random
inputs W (i) to the market should be at least as numerous as the (independent)
risky assets. We shall see a somewhat different interpretation when discussing
models for the term structure of interest rates, where the market price of risk
assumption will come back in the natural, intuitive form that a market cannot
have two different interest rates.

7.2

Fair Prices

In the present extended Black-Scholes model, a portfolio is still a pair (, )

of a predictable process t , giving the number of bonds, and a vector-valued
predictable process t = (1t , . . . , nt ), giving the numbers of assets of the various
Pn R T
types. To make the integrals well defined, we assume that i=1 0 |it |2 dt +
RT
|t | dt
0
Pn< . The portfolio is called self-financing if its value process Vt =
t Bt + i=1 it Sti satisfies
(7.3)

dVt =

n
X

it dSti + t dBt .

i=1

By definition, the fair price of a claim C at time t is the value Vt of a replicating

strategy at time t, where a replicating strategy is exactly as before: a selffinancing strategy whose value at T is equal to VT = C.
In the present situation we discount using the process B rather than the
exponential factors ert . Thus the discounted stock processes are


R t ij
Rt
Rt
P
1 Pd
j
(sij )2 ds+ d
1 i
i
i 0 rs ds+ 0 is 2

j=1
j=1 0 s dWs .
St = Bt St = S0 e
The second equality follows from the definition of B, the SDE (7.1) for the asset
prices and It
os formula, applied as in Example 5.14. By Itos formula,
dSti = Sti (it rt ) dt + Sti

d
X
j=1

47

tij dWtj .

If there exists a market price of risk process , this can be rewritten in the form
d
X

dSti = Sti

tj ,
tij dW

j=1

Rt

t = Wt s ds. Unless = 0, the process W

will not be a Pwhere W
0
Brownian motion in view of Lemma 5.17. However, according to Girsanovs
is a Brownian
theorem, there exists a probability measure Q under which W
motion (if is appropriately integrable). Under this martingale measure Q,
the discounted stock prices are local martingales.
We can now follow roughly the reasoning in Section 6.2 to construct a
replicating portfolio for a claim that pays an amount C at time T . A key
element in this construction is to find a process t such that the martingale
Vt = EQ BT1 C|Ft ) is representable as dVt = t dSt . In the present vectorvalued situation this is to be understood as
n
X
dVt =
it dSti .
i=1

If the matrices t are square and invertible, then this representation can be easily
obtained from the vector-valued version of the Brownian representation theorem,
Theorem 5.21, by the same arguments as in Section 6.2. More generally, the
desired representation is typically possible if the filtration (Ft ) is generated by
the asset processes Sti . In the following theorem we refer to this assumption by
assuming that the stock price processes possess the representation property.
Theorem 7.4. Assume that there exists a predictable process satisfying (7.2),
and that the stock price processes possess the representation property. Further
 RT

 RT
more, assume that E exp 21 0 ks k2 ds < and E exp 21 0 ks k2 ds <
. Then the value of the claim C FT at time t [0, T ] is given by
Bt EQ (BT1 C|Ft ), where Q is the measure under which the discounted price
processes Bt1 Sti are local martingales.

7.3

Arbitrage

In the preceding it was seen that existence of the market-price-of-risk process is

essential for the construction of a martingale measure Q under which the discounted stock price processes are local martingales. To underline the necessity
of the existence of the market-price-of-risk we shall now show that without it,
the market allows arbitrage.
By (7.3), the value process of a self-financing strategy (, ) is
Z t
Z t
n Z t
n

X
X
i
i
Vt V0 =
s Bs rs ds +
s dSs =
Vs
it Sti rs ds
0

i=1

n Z t
X
i=1

is dSsi =

Vs rs ds +
0

i=1

n Z t
X
i=1

is (dSsi Ssi rs ds).

The last display and the fact that dBt = rt Bt dt imply that
d(Bt1 Vt ) = Vt Bt2 dBt + Bt1 dVt = Bt1

n
X
i=1

48

it (dSti Sti rt dt).

Hence, in view of (7.1), the discounted value process takes the form
(7.5)

Vt = Bt1 Vt = V0 +

n Z
X
i=1

Bs1 Ssi is

d
X


sij dWsj (rs is ) ds .

j=1

This formula does not make explicit reference to the amount invested in
the bond, which has been eliminated. A partial strategy defines a value
process P
through (7.5), and given we can define from the equation Vt =
n
t Bt + i=1 it Sti . By retracing the calculations, the resulting strategy (, )
can be seen to be self-financing and to possess value process Vt . Thus, to see
which value processes are possible, it suffices to construct the stock portfolio .
Nonexistence of a market price of risk process implies that the vector rt 1t
is not contained in the range of t , for a set of times t. Then there exists a vector
t such that the vector (St1 1t , . . . , Stn nt ) is orthogonal to this range and its inner
product with the vector rt 1 t is strictly negative for a set of times t:
n
X

Sti it tij = 0,

j = 1, . . . , d,

i=1

n
X

Sti it (rt it ) < 0.

i=1

We can arrange it so that the latter inner product is never positive and hence,
by (7.5), the corresponding discounted gain process will be zero at time 0 and
strictly positive at time T . This is an example of arbitrage.
On the other hand, if the market price of risk process exists, then the
discounted gains process in (7.5) can be written as a stochastic integral relative
R
, for W
t = Wt t s ds. Under the martingale measure Q
to the process W
0
is a Brownian motion, and hence the discounted gains process
the process W
will be a local martingale. Under the integrability assumptions of Theorem 7.4
it is a Q-martingale, and hence cannot become strictly positive as its mean must
remain zero. Thus existence of the market price of risk is intimately connected
to the nonexistence of arbitrage.

7.4

PDEs

= (W
1, . . . , W
d ) defined
Under the conditions of Theorem 7.4, the process W
Rt
t = Wt s ds is a Brownian motion under the martingale measure Q.
by W
0
Because option prices can be written as expectations under Q, it is useful to
. If
rewrite the stochastic differential equation (7.1) in terms of the process W
we also assume that the processes r and take the forms rt = r(t, Bt , St ) and
t = (t, Bt , St ), then the equations describing the asset prices become
dBt = Bt r(t, Bt , St ) dt,
(7.6)

dSti = Sti r(t, Bt , St ) dt + Sti

d
X

tj ,
ij (t, Bt , St ) dW

i = 1, . . . , n.

j=1

As usual we assume that (B, S) is adapted to the augmented natural filtration

. Then, under regularity conditions on r and , the process (B, S)

FtW of W
will be Markovian relative to this filtration. If we assume in addition that is
can be expressed in (B, S) by inverting the second equation,
invertible, then W
49


are the same.
and hence the filtrations Ft and FtW generated by (B, S) and W
The process (B, S) is then Markovian relative to its own filtration Ft . In that
case a conditional expectation of the type EQ (X|Ft ) of a random variable X
that is a function of (Bs , Ss )st can be written as F (t, Bt , St ) for a function F .
This observation can be used to characterize the value processes of certain
options through a partial differential equation. The value process of a claim
that is a function C = g(ST ) of the final value ST takes the form

Vt = Bt EQ

 g(S )  
RT


T 
Ft = EQ e t r(s,Bs ,Ss ) ds g(ST )|Ft .
BT

If the process (B, S) is Markovian as in the preceding paragraph, then we can

write Vt = F (t, Bt , St ) for a function F . We assume that this function possesses
continuous partial derivatives up to the second order. For simplicity of notation
we also assume that S is one-dimensional. Then, by Itos formula,
dVt = Ft dt + Fb dBt + Fs dSt + 21 Fss d[S]t .
Here Ft , Fb , Fs are the first order partial derivatives of F relative to its three
arguments, Fss is the second order partial derivative relative to its third argument, and for brevity we have left off the argument (t, Bt , St ) of these functions.
A second application of It
os formula and substitution of (7.6) for (B, S) yield
dVt = d(Bt1 Vt ) = Bt2 Vt dBt + Bt1 dVt = Bt1 F r dt + Bt1 dVt
t.
= Bt1 (F r + Ft + Fb Bt r + Fs St r + 21 Fss St2 2 ) dt + Bt1 Fs St dW
The process Vt was seen previously to be a Q-local martingale. Because the
is a Brownian motion, this can only be true if the drift term on the
process W
right side of the preceding display is zero, i.e.,
(F r)(t, b, s) + Ft (t, b, s) + (rFb )(t, b, s)b
+ (rFs )(t, b, s)s + 12 ( 2 Fss )(t, b, s)s2 = 0.
This is an extension of the Black-Scholes partial differential equation.
This partial differential equation is useful for the numerical computation of
option prices. Even though the equation is rarely explicitly solvable, a variety of numerical methods permit to approximate the solution F . The equation depends only on the functions r and defining the stochastic differential equation (7.6). Hence it is the same for every option with a claim of
the type C = g(ST ), the form of the claim only coming in to determine the
boundary condition. Because C = g(ST ) = F (T, BT , ST ), this takes the form
F (T, b, s) = g(s). For instance, for a European call option on the stock S, this
becomes F (T, b, s) = (s K)+ .

50

Chapter 8

Interest Rate Models

8.1

The Term Structure of Interest Rates

In the classical Black-Scholes model the interest rate is a deterministic constant.

In reality the situation is much more complicated of course. In general, it is
not even possible to talk about the interest rate, since short term and long
term rates are usually different. Moreover, the time evolution of interest rates
typically has a random component.
In this chapter we introduce interest rate models that capture these properties of the time value of money. Such models are necessary for the pricing
of so-called interest rate derivatives. These are financial contracts that are designed to trade and manage the risk that is caused by the uncertainty about the
time value of money.

8.1.1

Discount Bonds

Pure discount bonds are simple financial contracts that capture the time value
of money. A discount bond which matures at time T > 0, also called a T -bond,
is a contract which guarantees a pay-off of 1 euro at time T . The price of a
T -bond at time t T is denoted by P (t, T ). It is the amount we are willing to
pay at time t to receive 1 euro at time T . The collection {P (0, T ) : T > 0} of
all bond prices at time t = 0 completely determines the time-value of money at
time 0. It is called the term structure of interest rates.
For fixed t, the function T 7 P (t, T ) is typically smooth, since, for instance,
the price of a bond that matures 9 years from now will be close to the price of
a bond that matures 10 years from now. For a fixed maturity T > 0 however,
the function t 7 P (t, T ) will appear to fluctuate randomly. By construction it
holds that P (T, T ) = 1.

8.1.2

Yields

If we have 1 euro at time t, we can use it to buy 1/P (t, T ) T -bonds. At time
T we then receive 1/P (t, T ) euros. Hence, a euro at time t grows to 1/P (t, T )
euros at time T . If the interest rate over the interval [t, T ] had been constant,
say r, a euro at time t would have grown to er(T t) at time T . If we compare

51

these (i.e., 1/P (t, T ) = er(T t) ), we see that buying the T -bonds at time t leads
to a constant interest rate over the time interval [t, T ] of
Y (t, T ) =

(8.1)

log P (t, T )
.
T t

We call this the yield over [t, T ]. The collection of all yields of course contains
exactly the same information as the collection of all bond prices. However, the
yields have a somewhat easier interpretation in terms of interest rates.

8.1.3

Short Rate

Although the interest rate does not exist, we can construct an object that can
be interpreted in this way. We just saw that the yield Y (t, T ) can be interpreted
as the constant interest rate valid in the time interval [t, T ]. The number



log P (t, T )
rt = lim Y (t, T ) =
T t
T
T =t
can therefore be viewed as the interest rate at time t (or in the infinitesimal
interval [t, t + dt]). We call rt the short rate at time t. From its definition it is
clear that in general, the short rate does not contain all information about the
time value of money.

8.1.4

Forward Rates

Let t < S < T and consider the following strategy. At time t, we sell one S-bond,
giving us P (t, S) euros. We immediately use this money to buy P (t, S)/P (t, T )
T -bonds. At time S the S-bond matures, and we have to pay one euro to its
holder. At time T the T -bond matures, and we receive P (t, S)/P (t, T ) euros.
If we follow this strategy, the net effect is that one euro at time S grows
to P (t, S)/P (t, T ) euros at time T . If the interest rate were a constant r over
the time interval [S, T ], one euro at time S would grow to er(T S) at time T .
Hence, the constant interest rate over [S, T ] determined at time t is

log P (t, T ) log P (t, S)

.
T S

This quantity is called the forward rate for [S, T ], contracted at time t. If we
let S T , we get

log P (t, T ),
T
which is the forward rate at time T , contracted at time t. Note that the short
rate is a particularRforward rate, namely f (t, t) = rt . Moreover, it is easy to see
T
that P (t, T ) = e t f (t,s) ds , so the collection of all forward rates contains all
information about the term structure of interest rates.
(8.2)

8.2

f (t, T ) =

Short Rate Models

The classical approach to interest rate models is to specify a stochastic model

for the short rate rt and to assume that the bond price P (t, T ) is some smooth
function of rt . A model of this type is called a short rate model.
52

So let us suppose that under the real world probability measure P, the
short rate satisfies the SDE
(8.3)

drt = (t, rt ) dt + (t, rt ) dWt ,

where W is a P-Brownian motion, and and are certain functions on [0, )

R. Let (Ft ) be the filtration generated by the process r. We assume that we
can put money in a bank which pays the interest rate rt , in the sense
R t that one
euro at time zero grows to Bt euros at time t, where Bt = exp( 0 rs ds). In
differential notation, the process B satisfies
dBt = rt Bt dt.
For the bond prices we assume that P (t, T ) = F T (t, rt ), where F T is some
smooth function on [0, ) R which may depend on the time to maturity
T . Clearly, the functions should satisfy F T (T, r) = 1 for all T and r. In the
preceding section we noted that the short rate does not determine the whole
term structure of interest rates, so we can expect that we have some freedom in
choosing the functions F T . On the other hand we do not want to allow arbitrage
opportunities in the bond market. It is intuitively clear that this implies certain
restrictions on the relation between the prices of the T -bonds for various T ,
leading to restrictions on the functions F T . In the remainder of this section we
explain how we can construct arbitrage free short rate models.
The first step is the observation that by the absence of arbitrage there cannot
be banks with different rates of interest.
Lemma 8.4. Suppose there exists a self-financing portfolio with value process
V which satisfies dVt = qt Vt dt for some adapted process q. Then qt = rt for all
t 0.
Proof. We sketch the proof. Suppose for simplicity that q and r are constant
and that q > r. Then we can borrow 1 euro at rate r and invest it in the
portfolio which pays interest q. At time T , say, we sell the portfolio, giving
us exp(qT ) euros. We pay back our loan, which is now exp(rT ), and are left
with a risk-free profit of exp(qT ) exp(rT ) euros. This is clearly an arbitrage,
which is not allowed.
The general case of random, nonconstant processes q and r can be handled
similarly.
By assumption the price P (t, T ) of a T -bond is given by P (t, T ) = F T (t, rt ).
This is a smooth function of t and a process which satisfies an SDE. Hence, by
It
os formula, we have that
T
(t, rt ) d[r]t .
dP (t, T ) = FtT (t, rt ) dt + FrT (t, rt ) drt + 21 Frr

If we combine this with the SDE (8.3) for the short rate rt we obtain
(8.5)

dP (t, T ) = T (t, rt )P (t, T ) dt + T (t, rt )P (t, T ) dWt ,

where the functions T and T are given by

(8.6)

T =

T
FtT + FrT + 12 2 Frr
,
FT

53

T =

FrT
.
FT

Below we write tT and tT instead of T (t, rt ) and T (t, rt ).

To study the relation between the prices of bonds with different maturities
we are now going to consider a self-financing portfolio consisting of S-bonds and
T -bonds, for S < T . Suppose we are given such a portfolio, consisting at time
t < S of Tt T -bonds and St S-bonds, and let V denote its value process. Since
the portfolio is self-financing we have
dVt = Tt dP (t, T ) + St dP (t, S) = uTt Vt

dP (t, T )
dP (t, S)
+ uSt Vt
,
P (t, T )
P (t, S)

where uT and uS are the fractions of the portfolio consisting respectively of

T -bonds and S-bonds, defined by
uTt =

Tt P (t, T )
,
Vt

uSt =

St P (t, S)
.
Vt

If we combine this with the SDE (8.5) for P (t, T ) we get





dVt = uTt tT + uSt tS Vt dt + uTt tT + uSt tS Vt dWt .
This SDE holds for every self-financing portfolio consisting of S-bonds and
T -bonds. Conversely, we can construct a particular portfolio by specifying fractS
tT
S
tions uT and uS satisfying uTt + uSt = 1. The choice uTt = T
S , ut = T S
t
t
t
t
leads to a self-financing portfolio with value process V satisfying
 S T

t t tT tS
dVt =
Vt dt.
tT tS
The dWt -term has disappeared, so by Lemma 8.4 it must hold that for all t 0
tS tT tT tS
= rt ,
tT tS

equivalently

tS rt
T rt
= t T .
S
t
t

In other words, the ratio (tT rt )/tT must be independent of T . Thus, we

have proved the following lemma.
Lemma 8.7. There exists a function on [0, ) R, independent of T , such
that, for all t, T ,
T (t, rt ) rt
(t, rt ) =
.
T (t, rt )
Recall that tT and tT are the local rate of return and volatility of the T bond, respectively (cf. (8.5)). Hence, the difference tT rt can viewed as a risk
premium. It is the excess return that we get if we invest in the risky T -bond
instead of putting our money in the bank. The quantity (tT rt )/tT , i.e. the
risk premium per unit of volatility, is called the market price of risk of the T bond. In this terminology the preceding lemma states that in an arbitrage free
bond market all bonds have the same market price of risk.
If we combine the result of Lemma 8.7 with the definitions (8.6) of the
processes T and T , we arrive at a PDE for the pricing functions of T -bonds,
called the term structure equation.

54

Theorem 8.8. Let (t, rt ) denote the market price of risk. Then for every
T > 0 the function F T satisfies the PDE
T
FtT + ( )FrT + 21 2 Frr
rF T = 0,

subject to the boundary condition F T (T, r) = 1.

EXERCISE 8.9. Prove Theorem 8.8.
Under certain regularity conditions the term structure equation has a unique
solution for every T > 0, so the bond prices P (t, T ) = F T (t, rt ) are completely
determined by the functions , and .
It is now clear how we can construct a short rate model leading to an arbitrage free bond market:
1) Specify the drift and volatility for the short rate rt (under P) and
assume that rt satisfies the SDE (8.3), with a P-Brownian motion W .
2) Choose a function on [0, ) R, and for T > 0 let F T be the solution
of the term structure equation corresponding to , and .
3) Finally, define the price of a T -bond as P (t, T ) = F T (t, rt ).
Observe that the term structure equation for the price of a T -bond is very
similar to the Black-Scholes PDE for the pricing function of a European claim,
cf. Theorem 6.11. In the preceding chapter we saw that the price of a European
claim also equals the expectation of the discounted pay-off under a new measure
Q. We have the following analogous theorem for the price of a T -bond in a short
rate model.
Theorem 8.10. If and are the drift and volatility of the short rate rt (see
(8.3)) under P and is the market price of risk, the price of a T -bond at time
t is given by
 RT



P (t, T ) = Bt EQ BT1 |Ft = EQ e t rs ds |Ft ,
where Q is the measure under which the short rate satisfies the SDE


t,
drt = (t, rt ) (t, rt )(t, rt ) dt + (t, rt ) dW
is a Q-Brownian motion (W
t = Wt +
and W

Rt
0

(s, rs )ds).

Note that for every T > 0 the discounted price P (t, T ) = Bt1 P (t, T ) of a
T -bound satisfies


Bt1 P (t, T ) = EQ BT1 |Ft ,
so that for every T > 0 the process (P (t, T ))tT is a martingale under Q.
Therefore the measure Q appearing in the statement of the theorem is called
the martingale measure of the model. Observe that the formula
P (0, T ) = EQ BT1
for the current price of a T -bond is a statement of the usual form price of a
claim is the expectation of the discounted payoff under the martingale measure,
since a T -bond can be viewed as a claim which pays off 1 euro at time T .
Theorem 8.10 gives us a second method for the construction of a model for
an arbitrage free bond market:
55

1) Specify an SDE for the short rate rt under the martingale measure Q and
let (Ft ) be the natural filtration of the process r.
RT


2) Define the price P (t, T ) of a T -bond by P (t, T ) = EQ e t rs ds |Ft .
This second procedure for the construction of short rate models is known
as martingale modeling and has the obvious advantage that we do not have to
specify the market price of risk (and function ) explicitly. Instead, we only
need to specify the functions and , the volatility and the drift term of
the short rate SDE under the martingale measure Q.

8.3

The Hull-White Model

The Hull-White model for the term structure of interest rates assumes that
under the martingale measure Q the short rate rt satisfies the SDE
(8.11)

drt = ((t) art ) dt + dWt ,

where a 6= 0 and are certain numbers, is a deterministic function and W is

a Q-Brownian motion. The natural filtration of r (and W ) is denoted by (Ft )
and the price P (t, T ) of a T -bond at time t is defined by
RT


P (t, T ) = EQ e t rs ds |Ft .
By the preceding section this defines an arbitrage free model for the bond market.
It is possible to obtain concrete formulas for the bond prices in this model.
The main reason is that we have an explicit expression for the solution of the
SDE (8.11). This allows us to determine the conditional distribution of the
RT
integral t rs ds given Ft , which we need to calculate P (t, T ).
RT
Lemma 8.12. Given Ft , the integral t rs ds possesses a Gaussian distribution
RT
RT
with mean B(t, T )rt + t B(u, T )(u) du and variance 2 t B 2 (u, T ) du, where
B(t, T ) = (1 ea(T t) )/a.

(8.13)

Proof. We apply the It

o formula to calculate d(eat rt ) and use (8.11) to get that
Z s
Z s
(8.14)
rs = eas r0 + eas
(u)eau du + eas
eau dWu .
0

Integrating rs from t to T and repeatedly using integration by parts gives

Z T
Z t
Z T
at
at
au
rs ds = r0 e B(t, T ) + e B(t, T )
e (u) du +
B(u, T )(u) du
t

+ B(t, T )eat

Z
= B(t, T )rt +

eau dWu +

0
T

B(u, T ) dWu
t

Z
B(u, T )(u) du +

B(u, T ) dWu .
t

Clearly, the first two terms on the right-hand side are Ft -measurable. The
third one is independent of Ft and is Gaussian with zero mean and variance
RT
2 t B 2 (u, T ) du. This completes the proof.
56

We can now derive the bond price formula for the Hull-White model.
Theorem 8.15. In the Hull-White model the price of a T -bond is given by
P (t, T ) = eA(t,T )B(t,T )rt ,

(8.16)

where B is defined by (8.13) and A(t, T ) =

RT
t

2 2
2 B (u, T )

(u)B(u, T ) du.

RT
Proof. We have to calculate the expectation of exp( t rs ds) given Ft . By
the preceding lemma this boils down to computing the expectation of the exponential of a Gaussian random variable. If Z N (m, s2 ), it holds that
E exp(Z) = exp(m + s2 /2). Together with the lemma this yields the statement
of the theorem.
Short rate models in which the bond price is of the form (8.16) are called
affine models. The reason for this name is that the yields and forward rates are
affine in rt in that case. Indeed, by (8.16) and (8.1)(8.2), the yield Y (t, T ) and
the forward rate f (t, T ) in the Hull-White model are as follows:
Y (t, T ) =

A(t, T )
B(t, T )
rt
,
T t
T t

f (t, T ) = BT (t, T )rt AT (t, T ).

Now consider a specific bond market in which bonds of all maturities are
traded. Then at time zero, we can observe the bond prices and forward rates
with all maturities. We denote the observed prices and rates in the market by
P (0, T ) and f (0, T ), respectively. On the other hand, the Hull-White model
gives the formula for the forward rates:
f (0, T ) = BT (0, T )r0 AT (0, T ).

(8.17)

Obviously, we would like to match the theoretical rates f (0, T ) with the observed
rates f (0, T ). We will now show that this is possible by choosing an appropriate
function in (8.11). This procedure is called fitting the model to the term
structure of interest rates.
Theorem 8.18. Let the parameters a, in (8.11) be given. Then with the
choice


(T ) = af (0, T ) + fT (0, T ) + 2 B(0, T ) eaT + a2 B(0, T )
(8.19)
the theoretical Hull-White forward rates coincide with the observed rates. The
corresponding price of a T -bond is then given by
P (t, T ) =



2 2
P (0, T )
exp B(t, T )f (0, t)
B (t, T )(1 e2at ) B(t, T )rt .
P (0, t)
4a

Proof. If we insert the expressions for A and B into (8.17) we see that we have
to solve the equation

f (0, T ) = e

aT

Z
r0 +

ea(T u) (u) du

57

2
(1 eaT )2 = g(T ) h(T ),
2a2

where g is the solution of the differential equation g 0 + ag = , g(0) = r0

and h(t) = 2 B 2 (0, t)/2. Then g(T ) = f (0, T ) + h(T ), which shows that the
solution of the equation for is given by
(T ) = g 0 (T ) + ag(T ) = fT (0, T ) + hT (T ) + a(f (0, T ) + h(T )).
This proves the first part of the theorem.
The second statement can be obtained by inserting the expression for in
formula (8.16).
So after fitting the term structure of interest rates there are only two free
parameters left in the Hull-White model, a and . In practice these are determined by matching the theoretical prices for certain interest rate derivatives
with observed prices. This procedure is called the calibration of the model.

8.4

Pricing Interest Rate Derivatives

The result of Theorem 8.10 can be viewed as a pricing formula for the simplest
possible claim which has a payoff of one euro at time T . Using the same arguments as above it can be extended to a general claim which pays some random
amount C FT at time T .
Theorem 8.20.
Let C FT be a claim. Its value at time t T is given by
RT


Vt = EQ e t rs ds C|Ft , where Q is the martingale measure.
Note that for C 1 we indeed recover the formula for the bond price P (t, T ).
Many interest rate derivatives do not only have a payment at the time T of
maturity, but also at certain fixed intermediate times. Holding such a product
(called coupon) is equivalent to holding a portfolio of derivatives with different
maturities. Hence, Theorem 8.20 implies the following.
Theorem 8.21. Let 0 < T1 < < Tn = T and Ci FTi for i = 1, . . . , n.
Consider a derivative with a payment of Ci at time Ti for i = 1, . . . , n. Its value
at time 0 is given by
n
R Ti
X
V0 = EQ
e 0 rs ds Ci ,
i=1

where Q is the martingale measure.

The result of Theorem 8.21 can now be used to determine the price of a
given derivative by simulation methods, just as we discussed in Section 6.6 for
the Black-Scholes model. The procedure is as follows:
1) Simulate a large number of realizations, say n, of the short rate process r
under the martingale measure Q.
2) For realization number j, compute the payoff of the claim and determine


RT
P
an approximation cj for the discounted payoff i exp 0 i rs ds Ci .
Pn
3) Then by the law of large numbers, cn = n1 j=1 cj is a good approximation
of the claim price, provided that n is large enough.

4) By the central limit theorem the interval [

cn 1.96sn / n, cn +1.96sn / n]
is an approximate
interval for the price of the derivative,
P95%-confidence
n
1
2
where s2n = n1
(c

)
.
j
n
j=1
58

The usual approach to simulating realizations of the short rate is to discretize the SDE for rt under Q. Suppose that under Q the short rate satisfies
(8.3), where and are given functions. Then for small > 0 we have the
approximation
r(k+1) rk (k, rk ) + (k, rk )(W(k+1) Wk ),

k = 0, 1, 2, . . . .

The increments W(k+1) Wk are independent N (0, ) random variables. So the

approximation can be written as r(k+1) rk (k, rk ) + (k, rk )Zk+1 ,

where Z1 , Z2 , . . . are independent, standard Gaussian random variables. This
so called Euler approximation of the SDE (8.3) can be used to simulate sample
paths and to determine the corresponding approximation of the discounted payoff. A single realization is constructed as follows.
1) Partition the interval [0, T ] into m intervals of length = T /m and simulate m i.i.d. standard Gaussian random variables Z1 , . . . , Zm .
2) The short rate r0 at time 0 is given. The future rates r , r2 , . . . , rm are
computed recursively by the formula

r(k+1) = rk + (k, rk ) + (k, rk )Zk+1 , k = 0, . . . , m 1.

3) For this realization the pay-off is calculated and the discount factors
RT
P
exp( 0 rs ds) are approximated by exp( k rk ).
The outlined procedure works for any short rate model. In the special case
of the Hull-White model we have in fact an exact recurrence formula for the
discretized short rate process. This means that realizations can be simulated
without introducing approximation errors due to discretization of the SDE. If
d
we write X = Y , this means that X and Y have the same distribution.
Theorem 8.22. Suppose that the short rate rt satisfies (8.11) under the martingale measure Q, where is given by (8.19). Then rt = t + yt , where y0 = 0,
q
2
2
d
t = f (0, t) + 2 B 2 (0, t), y(k+1) ea yk = 2 B(0, 2)Zk+1 , k = 0, 1, . . . ,
for any > 0, with Z1 , Z2 , . . . i.i.d. standard Gaussian and B given by (8.13).
Proof. It follows from (8.14) that rt = t + yt , where
Z t
Z t


t = eat r0 +
(u)eau du
and yt = eat
eau dWu .
0

The expression for t in the statement of the theorem now follows after inserting
(8.19) and some straightforward calculations. To prove the recurrence formula
for the process y, observe that the random variables
Z (k+1)
ea(k+1) y(k+1) eak yk =
eau dWu , k = 0, 1, . . . ,
k

R (k+1) 2au
are independent, Gaussian, with zero means and variances 2 k
e
du =
2 2ak 2a
e
(e 1)/(2a). Hence, with i.i.d. standard Gaussian Z1 , Z2 , . . .,
r
2 (e2a 1)
d ak
a(k+1)
ak
e
y(k+1) e yk = e
Zk+1 , k = 0, 1, . . . .
2a
The proof is completed by dividing this by ea(k+1) .
59

8.5

Examples of Interest Rate Derivatives

Below we discuss some common interest rate products and their valuation.

8.5.1

Bonds with Coupons

In practice pure discount bonds are not often traded. Instead, bonds typically
do not only have a payoff at maturity, the so called principal value, but also make
smaller regular payments before maturity. Such a bond is called a coupon bond.
A 10-year, 5% coupon bond with a principal value of 100 euros for instance,
pays 5 euros every year until maturity and 100 euros at maturity (after ten
year).
More generally, suppose that a bond makes a payment of k euros at times
T1 < . . . < Tn = T , and pays off its principal value of 1 euro at time T . Then
holding this coupon bond is equivalent to holding k pure discount bonds with
maturity Ti for i = 1, . . . , n, and one T -bond. Hence, the value of the coupon
bond is
n
X
P (0, Ti ).
P (0, T ) + k
i=1

We remark that conversely, the prices of pure discount bonds may be expressed in terms of the prices of coupon bonds. In practice this is the usual way
in which the prices of discount bonds are inferred from market data.

8.5.2

Floating Rate Bonds

There also exist bonds with intermediate payments that depend on the interest
rates at the time of the payment. The LIBOR rate for the time interval [S, T ],
set at time S is defined as

1 
1
P (S, T ) 1
=
1 .
L(S, T ) =
(T S)P (S, T )
T S P (S, T )
This is simply the return per time unit on an investment at time S in a T -bond.
A floating rate bond is a bond with additional payments at times T1 < . . . <
Tn = T . The payment Ci at time Ti is
(Ti Ti1 )L(Ti1 , Ti ) =

1
P (Ti1 , Ti )

1.

This is precisely the gain we would have had at time Ti if we had bought one
euro worth of Ti -bonds at time Ti1 . The principal value of one euro is payed
at time T .
By Theorem 8.21, the price of this asset at time 0 is given by
X
P (0, T ) +
EQ BT1
(P (Ti1 , Ti )1 1).
i
i

By the tower property of conditional expectation and Theorem 8.10 the ith term
in the sum equals


EQ EQ BT1
(P (Ti1 , Ti )1 1)|FTi1 = EQ (P (Ti1 , Ti )1 1)BT1
P (Ti1 , Ti )
i
i1
= EQ BT1
EQ BT1
P (Ti1 , Ti ) = P (0, Ti1 ) P (0, Ti ).
i1
i1
60

Hence, the value of the bond is

P (0, T ) +

n
X

P (0, Ti1 ) P (0, Ti ) = P (0, 0) = 1

i=1

So the price of a floating rate bond equals its principal value.

EXERCISE 8.23. Derive the pricing formula for the floating rate bond by
showing that there exists a self-financing portfolio with initial value of one euro
which has the same payoff as the bond.

8.5.3

Swaps

A swap is contract that exchanges a stream of varying, interest rate dependent

payments for a stream of fixed amount payments, or vice versa. Consider for
example time points 0 < T1 < . . . < Tn . At time Ti we make a payment of
(Ti Ti1 )L(Ti1 , Ti ) euros and we receive k euros. In other words, we swap
the gain obtained from a one euro investment in Ti -bonds at time Ti1 for the
constant gain k. Buying this contract is equivalent to selling a floating rate
bond and buying a coupon bond which pays k euros at each time Ti . Hence,
the price of the swap at time 0 is
P (0, Tn ) + k

n
X

P (0, Ti ) 1.

i=1

8.5.4

Swaptions

A swaption is a contract giving the holder the right to enter into a swap at a
future date. Suppose for instance that the contract gives the right to enter at
time T0 > 0 into a swap described in the preceding section. Then the payoff at
time T0 of the option is


P (T0 , Tn ) + k

n
X

+
P (T0 , Ti ) 1 .

i=1

By Theorem 8.20, the price at time 0 of the swaption is therefore given by

EQ e

R T0
0

rs ds

P (T0 , Tn ) + k

n
X

+
P (T0 , Ti ) 1 .

i=1

In general this expectation can not be evaluated analytically and one has to
resort to numerical methods. Note by the way that the latter formula shows
that a swaption can also be viewed as a call option on a coupon bond.

61

Chapter 9

Risk Measurement
Financial institutions deal in risk of various types. Market risk is the exposure to
the changing prices of assets on the market, and can be limited by using appropriate portfolios that include instruments such as options and swaps. Managing
risk is important for:
- Internal management, e.g. optimizing profit subject to restrictions on risk.
- To fulfill the requirements of regulatory authorities, such as national banks.
- Credit ratings.
The management of risk requires that risk be measured. In this chapter we
discuss the most popular measure of risk: Value-at-Risk, abbreviated VaR.

9.1

Value-At-Risk

Let Vt be the value of a portfolio at time t, and Ft the information available at

time t. We fix a given time t + t in the future and confidence level 1 . The
variable Vt Vt+t is the loss that we shall incur in the period [t, t + t]. VaR
is defined as a number such that the loss is with high probability smaller than
this number.
Definition 9.1. VaR is the upper -quantile of the conditional distribution of
Vt Vt+t given Ft , i.e.,
P(Vt Vt+t < VaR|Ft ) 1 P(Vt Vt+t VaR|Ft ).
The Value-at-Risk depends of course on both t and . For regulators a
period of 10 days and = 1% is usual, whereas for other purposes periods from
one day to a year or = 0.05% may be considered appropriate. VaR as defined
here refers to the absolute value of the portfolio, but for some purposes it can
be useful to consider the Value-at-Risk per capital VaR/Vt instead.
The conditioning on the past information Ft in our definition appears natural, but is not always made explicit. Although we shall consider VaR in the
following only at one fixed time t, Value-at-Risk is in our definition a stochastic
process in time. For larger lags t the dependence on t will typically be small.
One criticism to the use of VaR for risk measurement is that it may say little
about expected losses. In particular, it says nothing about the sizes of the losses
bigger than VaR, which occur with frequency , and which could be arbitrarily

62

big. Related to this is that VaR is in general not subconvex under combination
of portfolios. If two subunits of a financial institution both control their VaR,
then there is no guarantee that the VaR of the financial institution as a whole
is also under control. This is illustrated in Example 9.2. Other measures of
risk, which are not open to these criticisms are the conditional variance (or
volatility) and the expected excess loss, given by
var(Vt Vt+t |Ft ),

and

E (Vt Vt+t c)+ |Ft ).

Here c is some given threshold. The volatility is a classical measure, dating back
to Markowitz, who first proposed to optimize profit under the side condition of
bounded volatility. It is somewhat unstable for heavy-tailed distributions and
perhaps can be criticized for being symmetric.
Example 9.2 (Nonconvexity). Suppose that the portfolio consists of two parts,
with values Xt and Yt at time t, so that the total value is Vt = Xt + Yt . Then
Vt Vt+t = Xt + Yt for Xt and Yt the losses on the two subportfolios in
the interval [t, t + t], and the relative contributions of the two subportfolios in
the total are wX = Xt /Vt and wY = Yt /Vt . If VaR(V ), VaR(X), VaR(Y ) are
the Value-at-Risks of the three portfolios, then it may happen that the VaR(V )
is bigger than the convex combination wX VaR(X) + wY VaR(Y ).
For example, choose some fixed and let the vector (Xt , Yt ) be distributed (conditionally on Ft ) on the set of points {(0, 0), (c, 0), (0, c)} according
to the probabilities 12, , , for a given c > 0. Then VaR(X) = VaR(Y ) = 0,
but VaR(V ) = c. We can make the discrepancy c between the total Value-atRisk and the convex combination arbitrarily large.
To determine VaR we need a model for the conditional distribution of Vt
Vt+t given Ft . There are many possibilities, such as the ARMA and GARCH
models from time series, or the Black-Scholes or Hull-White models from derivative pricing. It is important to note that we need the distribution of the value
process under the real-world measure, not the martingale measure. Thus given
a model the parameters are estimated from time series giving the actual prices
of the assets over time, so-called historical analysis. Because some of the parameters may be common to the real-world measure and martingale measure,
some parameters could also be calibrated using observed derivative prices.
In rare cases, depending on the model, it is possible to determine an analytic
expression for the conditional distribution of Vt+t Vt given Ft . More often
the VaR must be calculated numerically, for instance by stochastic simulation.
If we can generate a large number N (e.g. at least N = 10000 if = 1%) of
realizations from the conditional distribution of Vt Vt+t given Ft , then VaR
is approximately the (1 )N largest value among the realizations. Remember
that, unlike when using simulation to determine a derivative price, this time we
must simulate from the real-world measure.

9.2

Normal Returns

In practice it is often assumed that the return Vt+t /Vt 1 is conditionally

normally distributed given Ft . (It may even be assumed that the returns are
independent of the past and form an i.i.d. sequence if restricted to a discrete time

63

grid, but that is even more unrealistic and unimportant for the following.) If the
conditional mean and variance are t and t2 , then the conditional distribution
of Vt Vt+t is normal with mean Vt t and standard deviation Vt t , and the
Value-at-Risk is given by


(9.3)
VaR = Vt t 1 (1 ) t .
Note that it is proportional to the current capital Vt and linearly increasing in
the volatility t . A positive drift t decreases VaR.
If a portfolio consists of n assets or subportfolios, with value processes
1
V 1 , . . . , V n , then it is often assumed that the vector of returns (Vt+t
/Vt1
n
n
1, . . . , Vt+t /Vt 1) is conditionally multivariate-normally
distributed given Ft .
Pn
The value of the whole portfolio Vt = i=1 Vti can be written as
Vt+t Vt = Vt

n
X

wti

V i

i=1

wti

t+t
Vti


1 ,

Vti /Vt

where
=
is the relative contribution of asset i to the whole portfolio.
The sum is a linear combination of a Gaussian vector and hence is normally
distributed. If the return vector possesses conditional mean vector (1t , . . . , nt )
and covariance matrix (ti,j ), then the conditional distribution of Vt Vt+t
given Ft is normal with mean Vt t and standard deviation Vt t , for
(9.4)

t =

n
X

wti it ,

i=1

(9.5)

t2 =

n X
n
X

wti wtj ti,j .

i=1 j=1

The Value-at-Risk again takes the same form (9.3), but with the new values of
t and t substituted.
The Cauchy-Schwarz inequality says that the covariances satisfy |ti,j |
i j
t t , where ti = ti,i are the variances of the components. This shows that
t

n
X

wti ti .

i=1

Because the VaR is linear in the standard deviation, this shows that the combined portfolio has a smaller VaR than a similar single portfolio of the same
volatility. This expresses the well-known guideline that diversification of a portfolio helps to control the risk.
As Example 9.2 shows, diversification is not always useful to control VaR,
but the preceding shows that for portfolios with normal returns it is.
Empirical studies have now well established that economic time series are
not Gaussian random walks, an assumption of the past that still lives on in
many VaR-methods. Returns are not i.i.d. and their marginal distributions
deviate from normal distributions in that they are typically heavier tailed and
sometimes skewed. Conditional normality of the returns given the past, as is
assumed in this section, is also debatable, but not always rejected by statistical
tests on empirical data. In particular, GARCH models are often combined
with normal conditional distribution, which automatically leads to heavier-tailed
unconditional distributions.
64

9.3

Equity Portfolios

The value process of a portfolio of one stock with price St is equal to Vt = St . If

we adopt the Black-Scholes model, then St = S0 exp(t + Wt ) for a Brownian
motion process W , and hence the log returns satisfy
Rt := log

St+t
= t + (Wt+t Wt ).
St

Because the increments of Brownian motion are independent of the past, it

follows that the log returns are conditionally normally distributed with mean t
and variance 2 t. The loss can be expressed in the log returns as Vt Vt+t =
St (1 eRt ). Solving the equation P(Vt (1 eR ) v) = 1 for a N (t, 2 t)
distributed variable R and a fixed value Vt , yields the Value-at-Risk



1
(9.6)
VaR = Vt 1 e t ()+t .
This has similar features as the equation (9.3): the risk is proportional to the
current value Vt , increasing in the volatility and decreasing in the drift .
Because log x x 1 for x 1, the log return Rt is close to the ordinary
return St+t /St 1, if is small. If we make this approximation and still assume
the normal model forthe return, then we are in the situation of Section 9.2, with
t = t and t = t. The resulting formula (9.3) is identical to the formula
obtained by replacing the exponential in (9.6) by its linear approximation.
The value process of a combined portfolio consisting of it assets of price Sti
(i = 1, 2) is given by Vt = 1t St1 + 2t St2 . If we assume that the numbers ti do
not change in the interval [t, t + t], then the gain in this interval is given by
1
2
Vt+t Vt = 1t (St+t
St1 ) + 2t (St+t
St2 ).

To determine VaR we need a model for the conditional distribution of the vector
2
1
St2 ). There are many possibilities.
(St+t
St1 , St+t
A natural generalization of the Black-Scholes model would be to assume that
both asset price processes follow Black-Scholes models Sti = S0i exp(i t + i Wti ).
Here W 1 and W 2 are Brownian motions, of which it would be natural to assume
that they are also jointly Gaussian with some correlation. Then the joint returns
i
/Sti , will be bivariate Gaussian, and we can compute
(Rt1 , Rt2 ), for Rti = log St+t
the VaR in terms of the parameters i , i and the correlation of Rt1 and Rt2 , at
least by computer simulation.
If, as before, we simplify by assuming that the returns, and not the log
returns, are bivariate Gaussian, then we shall be in the situation of Section 9.2.
The VaR is then given by (9.3) with t and t given by (9.4), where wti =
it Sti /Vt .
Alternatively, we may apply more realistic, but more complicated models.
The conditional distribution of the loss will then typically be non-Gaussian, and
not analytically tractable, but the Value-at-Risk can often be obtained easily
by simulation.
Deriving VaR of portfolios of more than two stocks does not cause conceptual difficulties. However, making realistic models for the joint distribution of
many equities is difficult. Gaussian models are a possibility, but unrealistic.
Other standard models may include many parameters, that may be difficult to
estimate.
65

9.4

Portfolios with Stock Options

In the Black-Scholes model a European call option with strike K and expiry
time T has value
 log(S /K) + (r + 2 /2)(T t) 
t

Ct = St
T t
 log(S /K) + (r 2 /2)(T t) 
t

.
Ker(T t)
T t
The distribution of this random variable, or, more appropriately, the conditional
distribution of Ct+t given Ft , is not easily obtainable by analytic methods, but
it is easy to simulate, being an explicit function of the stock price St (or St+t )
and a number of nonrandom quantities. Thus the VaR of a portfolio consisting
of a European call option can be easily obtained numerically.
The value of some other European options on a given stock with price process St can also be written as an explicit function Ct = F (t, St ) of the stock
price. The Value-at-Risk can then also be obtained by one round of computer simulation. Given Ft the stock price St is known and hence the gain
Ct+t Ct = F (t + t, St+t ) F (t, St ) is stochastic only through St+t . We can
simulate the conditional distribution of the gain by simulating a sample from
the conditional distribution of St+t given Ft . In the Black-Scholes model we
have that
St+t = St et+(Wt+t Wt ) ,
and hence an appropriate simulation scheme
is to simulate a standard normal

t+ tZ
.
variable Z and next compute St e
Even though this seems easy enough, in practice one often uses approximations of the form
Ct+t Ct = F (t + t, St+t ) F (t, St )
Ft (t, St )t + Fs (t, St )(St+t St ) + 12 Fss (t, St )(St+t St )2 .
The three partial derivatives on the right side are exactly the Greeks , ,
and , already encountered in
Section 6.5. For small t, the increment St+t St
is typically of the order O( t) in probability, and hence the middle term on
the right dominates. If we neglect the other two terms, then we arrive in the
pleasant situation that the gain Ct+t Ct is a linear transformation St Rt of
the return Rt = St+t /St 1. If we also assume that this return is conditionally
normally distributed, then we are back in the situation of Section 9.2, and VaR
takes a familiar form.
Options for which there is no explicit pricing formula are more difficult to
incorporate. According to the general pricing theory, the value of an option
with payment C at time T in the Black-Scholes model is equal to


Ct = EQ er(T t) C|Ft .
For complicated claims C this could be determined numerically by simulation,
this time under the martingale measure Q. Combined with the preceding this
leads to a double (nested) simulation scheme, sometimes referred to as the full
monte. We shall consider the most complicated case, that of a claim that
depends on the full path (St : 0 t T ) of the stock price.
66

Given the information Ft , the beginning (Ss : 0 s t) of the sample path

is known and hence given Ft the claim can be written as a function C = ht (Ss :
t < s T ) of the future of the path. Therefore we can simulate the value of the
option given Ft by simulating many times the future path (Ss : t < s T ) from
its conditional distribution given Ft , evaluating each time ht (Ss : t < s T ),
and finally taking the average of these values. In this round of simulations we
simulate from the martingale measure. If the stock price process S is Markovian,
then the conditional distribution of (Ss : t < s T ) given Ft is the same as the
conditional distribution of this process given St . For instance, for the BlackScholes model we have
Ss = St e(r

s W
t)
/2)(st)+(W

. The conditional distribution of the right side is

for a Brownian motion W
s W
t given Ft , which is
determined from the conditional distribution of W
simply that of a Brownian motion. Note that we have taken the drift equal to
r 2 /2, because we must simulate under the martingale measure.
We can now perform a nested sequence of simulations to determine the
Value-at-Risk of a portfolio consisting of options and stocks as follows. We
denote the true-world and martingale measures by P and Q, respectively, and
abbreviate the value of the claim C given the initial path (Ss : 0 s t) by
ht (Ss : t < s T ).
FOR (i in 1:MANY)
{
SIMULATE (Ssi : 0 s t + t) ACCORDING TO P
FOR (j in 1:MANY)
{
GIVEN Sti SIMULATE (Ssj : t < s T ) ACCORDING TO Q
i
GIVEN St+t
SIMULATE (Ssj : t + t < s T ) ACCORDING TO Q
}
COMPUTE Cti AS AVERAGE ht (Ssj : t < s T ) OVER j
i
COMPUTE Ct+t
AS AVERAGE ht+t (Ssj : t + t < s T ) OVER j
}
i
.
VaR IS 1 LARGEST OF THE VALUES Cti Ct+t
This scheme is sufficiently involved that it will pay to use special computational techniques to make the simulations more efficient and more accurate.

9.5

Bond Portfolios

The value of a bond portfolio, consisting of bonds

different maturity, with or
Pof
n
without coupons, is a linear combination Vt = i=1 it Pt,Ti of discount bond
prices Pt,Ti . A term structure model gives exactly the joint distribution of the
zero-coupon bonds. Thus in principle every term structure model allows to
calculate the VaR of the portfolio, if necessary by simulation. In the present
situation we need the term structure model under the true world measure P!
As a particular example, consider the Hull-White model. In this model the
bond prices are given by an explicit formula of the form
Pt,T = eA(t,T )B(t,T )rt .
67

Thus we can simulate the bond price at time t by simulating the short rate rt .
Under the martingale measure Q the short rate is the sum of a deterministic
function and an Ornstein-Uhlenbeck process. Unfortunately, to compute VaR
we need to simulate under the true-world measure P, which may add a random
drift to the Ornstein-Uhlenbeck process. This may destroy the Gaussianity
and other nice properties, which the short rate process possesses under the
martingale measure.
More specifically, in the Hull-White model the short rate satisfies formula
(8.14), which can be written in the form
at

rt = (t) + e

eas dWs ,

Rt
for the deterministic function (t) = eat r0 + eat 0 (s)eas ds. The function
in this expression can be found by calibration on option prices observed in
the market, and so can the parameters a and . However, the process W in
the preceding display is a Brownian motion under the martingale measure Q,
and not under P. In agreement with GirsanovsR theorem, under P the process
t
W is a Brownian motion with drift, and Wt 0 (s, rs ) ds for the market
price of risk is a P-Brownian motion. There is no way the market price of
risk can be calibrated from derivative prices alone, but it can be determined
by historical analysis. If (t, rt ) does not depend on rt , then the change of
drift only changes the deterministic function in the preceding display, and
we can use the simulation scheme for Ornstein-Uhlenbeck processes discussed
in Section 8.4 and Theorem 8.22 to generate rt , the bond prices, and hence the
VaR. If (t, rt ) is random, then the drift of the short rate process under P is
random, and we must fall back on the more complicated simulation schemes for
diffusion processes, such as the Euler scheme discussed in Section 8.4. Rather
than calibrate a and from observed derivative prices on the market, we may
then also choose to fit the diffusion model
drt = (t, rt ) dt + dWt ,
with W a P-Brownian motion, directly to historical data. Note that the volatility parameter is common to both the P and Q models and hence can both be
calibrated and estimated.
This discussion of the Hull-White model extends to any model where the
bond prices are simple functions Pt,T = F T (t, rt ) of the short rate, and more
generally to multi-factor models in which the bond prices Pt,T = F T (t, Xt ) can
be written as a simple function of multivariate diffusion process X.
In practice one often uses simpler approaches based on approximations and
the assumption that theP
yields are multivariate normal. Given a bond portfolio
n
with value process Vt = i=1 it Pt,Ti , where it is assumed constant in [t, t + t],
we first approximate
Vt+t Vt =

n
X
i=1

n
X

it (Pt+t,Ti Pt,Ti ) =

n
X



it e(Ti tt)Yt+t,Ti e(Ti t)Yt,Ti

i=1

h
it Pt,Ti (Ti t)(Yt+t,Ti Yt,Ti ) + tYt+t,Ti +

i=1

68

(Ti t)2
(Yt+t,Ti
2

i
Yt,Ti )2 .

In the last step we use the approximation ey ex ex (y x + 12 (y x)2 )

for y x. The derivatives in the linear and quadratic parts are known as the
duration and the convexity, respectively. The conditional distribution of the
right side given Ft can be evaluated once we have a model for the conditional
joint distribution of the vector of yield increments (Yt+t,Ti Yt,Ti ) given Ft . In
practice this is often modeled by a mean zero multivariate normal distribution.
If we neglect the quadratic term (use only duration), then we fall back in the
situation of Section 9.2. VaR then takes the form as given in (9.3):
v
uX
n
u n X
1
VaR = (1 )t
it jt Pt,Ti Pt,Tj (Ti t)(Tj t)ti,j .
i=1 j=1

VaR based on using both duration and convexity in the approximation can be
determined by simulation.

9.6

Portfolios of Bonds and Swaptions

A swaption relative to the swap times T1 < T2 < < Tn pays the amount

+
Pn
PT0 ,Tn +K i=1 PT0 ,Ti 1 at time T0 < T1 . Depending on the term structure
model used, there may or may not be an explicit formula for the value of a
swaption at time t < t0 . If there is, then determining the VaR does not present
great difficulties.
In general, the value at time t < T0 can be evaluated as
n
 R T0
X

+  
Ct = EQ e t rs ds PT0 ,Tn + K
PT0 ,Ti 1 Ft .
i=1

In short rate models, such as the Hull-White model, the variable inside the
expectation can be written as a function of the short rate process (rs : 0
t Tn ). Given Ft the initial path (rs : 0 s t) is known and hence we
can evaluate the price Ct by computing an expectation under Q of a function
ht (rs : t s T ) of the future sample path, given its present state rt . For
instance, in the Hull-White model Ct is given by
n
h R T0
X

+  i
EQ e t rs ds eA(T0 ,Tn )B(T0 ,Tn )rT0 + K
eA(T0 ,Ti )B(T0 ,Ti )rT0 1 r0 .
i=1

In this particular case it is possible to evaluate the expectation analytically (at

least for n = 1), using the approach of Lemma 8.12. In general, we need to use
simulation or approximations.
To compute the VaR of a swaption portfolio, we need a double, nested round
of simulations, one under the true-world measure and one under the martingale
measure. If the bond price Pt,Ti can be expressed in rTi and the value of the
swaption at time t can be written as


EQ ht (rs : t < s T0 )|rt ,
then a simple (but computationally inefficient) scheme is as follows.
FOR (i in 1:MANY)
69

{
SIMULATE (rs : 0 s t + t) ACCORDING TO P
FOR (j in 1:MANY)
{
GIVEN rti SIMULATE (rsj : t < s T0 ) ACCORDING TO Q
i
GIVEN rt+t
SIMULATE (rsj : t + t < s Tn ) ACCORDING TO Q
}
COMPUTE Cti AS AVERAGE ht (rsj : t < s T0 ) OVER j
i
COMPUTE Ct+t
AS AVERAGE ht+t (rsj : t + t < s T0 ) OVER j
}
i
VaR IS 1 LARGEST OF THE VALUES Cti Ct+t
.
Rather than using this double simulation scheme we may here also apply
approximations. However, in general it is not easy to compute partial derivatives
of the value process relative to e.g. the yield (duration and convexity), and
we may need to use numerical (i.e. discretized) derivatives instead.

9.7

Diversified Portfolios

In the preceding section we have considered a variety of portfolios. The balance

sheet of a large financial institution will typically include a combination of the
assets considered so far. To compute the Value-at-Risk we can use the same
arguments, with the important complication that we need models for the joint
distribution of the various assets. For instance, the joint distribution of stocks
on Philips and IBM, and bonds of various maturities. There is no standard
approach to this.

70