On the Pricing of Contingent Claims under Constraints

I. KARATZAS

Departments of Mathematics
and Statistics
Columbia University
New York, NY 10027
ik@stat.columbia.edu
S. G. KOU
Department of Statistics
Hill Center, Busch Campus
Rutgers University
Piscataway, NJ 08855
kou@stat.rutgers.edu
November 1994; revised, November 1995
Abstract
We discuss the problem of pricing contingent claims, such as European call-options,
based on the fundamental principle of absence of arbitrage and in the presence of con-
straints on portfolio choice, e.g. incomplete markets and markets with short-selling con-
straints. Under such constraints, we show that there exists an arbitrage-free interval which
contains the celebrated Black-Scholes price (corresponding to the unconstrained case); no
price in the interior of this interval permits arbitrage, but every price outside the interval
does. In the case of convex constraints, the endpoints of this interval are characterized
in terms of auxiliary stochastic control problems, in the manner of Cvitanic & Karatzas
(1993). These characterizations lead to explicit computations, or bounds, in several inter-
esting cases. Furthermore, a unique fair price p is selected inside this interval, based on
utility maximization and marginal rate of substitution principles; again, characterizations
are provided for p, and these lead to very explicit computations. All these results are also
extended to treat the problem of pricing contingent claims in the presence of a higher in-
terest rate for borrowing. In the special case of a European call-option in a market with
constant coecients, the endpoints of the arbitrage-free interval are the Black-Scholes prices
corresponding to the two dierent interest rates; and the fair price coincides with that of
Barron & Jensen (1990).
AMS 1991 Subject Classication: Primary 90A09, 93E20, 60H30; Secondary 60G44, 90A10,
90A16, 49N15.
Key words and phrases: pricing of contingent claims, constrained portfolios, incomplete
markets, two dierent interest rates, Black-Scholes formula, utility maximization, stochas-
tic control, martingale representations, equivalent martingale measures, minimization of
relative entropy.

Research supported by the National Science Foundation, under Grant NSF-DMS-93-19816.

Part of this authors work was carried out while he was visiting the Courant Institute of Mathematical
Sciences, New York University, in Fall 1994; he extends his appreciation to his hosts at the Institute, for their
hospitality.
1
Contents
1 Introduction and summary 2
2 The nancial market model 4
3 Portfolio, consumption and wealth processes 5
4 Contingent claims and arbitrage in the unconstrained market 8
5 Upper and lower arbitrage prices 11
6 Representations for convex constraints 15
7 A fair price 24
7.1 Denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7.2 Connections with Arbitrage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
7.3 A representation for convex constraints . . . . . . . . . . . . . . . . . . . . . . . . 31
8 European call-option in a market with constant coecients 37
8.1 Lower and upper arbitrage prices . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
8.2 Computation of the fair price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.3 Counterexamples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
9 Market with higher interest rate for borrowing 45
10 A table 48
11 Discussion 50
1 Introduction and summary
The famous Black & Scholes (1973) formula provides the unique price of a European contingent
claim in an ideal, complete and unconstrained market, as laid out in sections 2 and 3 of the
present paper, based on the fundamental principle of absence of arbitrage opportunities. In
other words, this price is the unique one for which there are no arbitrage opportunities by
taking either a short or a long position in the claim, and investing wisely in the market. This
price coincides with the minimal initial capital, starting with which one can duplicate exactly
2
the claim at the terminal time, and also with the expectation of the claims discounted value
under the unique, risk-neutral equivalent probability measure (cf. Merton (1973), Cox &
Ross (1976), Cox & Rubinstein (1984), Harrison & Kreps (1979), Harrison & Pliska (1981),
Karatzas (1989); see also section 4 of this paper for a brief survey).
However, in the presence of constraints on portfolio choice (e.g., constraints on borrowing,
on short-selling of stocks, even on accessing certain stocks at all, as in the case of incomplete
markets), there ceases to exist a unique price for a contingent claim based solely on the principle
of absence of arbitrage. Instead, there appears an arbitrage -free interval [h
low
, h
up
] which
contains the Black-Scholes price u
0
; see the following gure. Here, h
up
represents the least
price the seller can accept without risk, and h
low
the greatest price the buyer can aord to pay
without risk.
- r r r r
0 h
low
u
0
h
up
This interval has the following properties:
(i) every price-level outside the interval leads to an arbitrage opportunity;
(ii) there are no arbitrage opportunities for price-levels in the interior of the interval.
These facts are demonstrated, to our knowledge for the rst time, in section 5 of this
paper. Furthermore, if the constraints on portfolio choice are convex, it turns out that the
endpoints of the arbitrage-free interval can be characterized as the values of certain suitable
stochastic control problems, as in Cvitanic & Karatzas (1993), or El Karoui & Quenez (1995)
for incomplete markets; see section 6 and, in particular, Theorem 6.1. Roughly speaking, the
upper (resp., lower) endpoint of the interval is equal to the supremum (resp., inmum) of
the Black-Scholes prices of the claim over a family of auxiliary, slightly more complicated in
structure but unconstrained, markets.
There remains the question of how to choose then a unique price for the claim, in the
presence of constraints on portfolio choice. There seems to be no denitive answer to this
question, though several approaches have been suggestedmost of them in the context of
incomplete markets (e.g. Follmer & Sondermann (1986), Foldes (1990), Follmer & Schweizer
(1991), Due & Skiadas (1991), Davis (1994), etc.), and some in dierent but related contexts
(dierent interest rates for borrowing and saving, Barron & Jensen (1990); transaction costs,
Hodges & Neuberger (1989)). We adopt in section 7 the approach of Davis (1994), which is
based on utility maximization and on the principle of zero marginal rate of substitution .
3
These considerations lead to the notion of a fair price p (Denition 7.3), which, under
certain mild conditions (cf. Assumptions 7.1, 7.2), is shown to lie within the arbitrage-free
interval (Theorem 7.1). Counterexamples for which the fair price lies outside the arbitrage-free
interval are also given in section 8.3. In the special case of convex constraints, we show that the
fair price admits a Black-Scholes representation under a certain minimal or least-favorable
equivalent probability measure (Theorem 7.4). In the derivation of this latter result, we draw
on the powerful results of Cvitanic & Karatzas (1992) for utility maximization under convex
portfolio constraints (cf. Karatzas, Lehoczky, Shreve & Xu (1991) for the special case of in-
complete markets). The representation of Theorem 7.4 leads to explicit computations of the
fair price p (Examples 7.1-7.4) for rather general portfolio constraints, including incomplete
markets, short-selling or borrowing constraints, etcetera. In particular, it is shown that p is
independent of both initial wealth and utility function, in a market with deterministic coe-
cients and in the presence of cone-constraints on portfolios; and in this case, the corresponding
equivalent martingale measure is also obtained by means of relative entropy minimization.
Section 8 oers a host of explicit computations for h
low
, h
up
and p in the special but
important case of a European call option, for a market with constant coecients and under
various kinds of constraints; these computations are tabulated in section 10, and constitute one
of the main results of this paper. Explicit computations are also possible for a path-dependent
(or look-back) option; see Example 7.4.
A most interesting result, from a practical point of view, is that the same ideas and tech-
niques can also treat the problem of pricing contingent claims in a market with higher interest
rate for borrowing than for saving. More precisely, it is shown in section 9 that in this case
there also exists an arbitrage-free interval, and a fair price p which always lies within that
interval. In the special case of European call option in a market with constant coecients, the
endpoints of the arbitrage-free interval are the two Black-Scholes prices corresponding to the
two dierent interest rates; and the fair price p coincides with the so-called minimax price in
Barron & Jensen (1990), if a power-type utility function is employed.
2 The nancial market model
We shall deal exclusively in this paper with a nancial market / in which d + 1 assets (or
securities) can be traded continuously. One of them is a non-risky asset, called the bond (also
4
frequently called savings account), with price P
0
(t) given by
dP
0
(t) = P
0
(t)r(t)dt, P
0
(0) = 1. (2.1)
The remaining d assets are risky; we shall refer to them as stocks, and assume that the price
P
i
(t) per share of the i
th
stock, is governed by the linear stochastic dierential equation
dP
i
(t) = P
i
(t)[b
i
(t)dt +
d

j=1

ij
(t)dW
j
(t)], P
i
(0) = p
i
, i = 1, 2, . . . d. (2.2)
In this model, W(t) = (W
1
(t), . . . , W
d
(t))

is a standard Brownian motion in

d
, whose
components represent the external, independent sources of uncertainty in the market /; with
this interpretation, the volatility coecient
ij
() in (2.2) models the instantaneous intensity
with which the j
th
source of uncertainty inuences the price of the i
th
stock.
As is standard in the literature, / is assumed to be an ideal market; in other words, we
have innitely divisible assets, no constraints on consumption, no transaction costs or taxes.
We shall allow, however, for constraints on portfolio choice, such as limitations on borrowing
(from the savings account ) or on short-selling (of stocks), and so on; see the Examples in
Section 6.
The probabilistic setting will be as follows: the Brownian motion W will be dened on a
complete probability space (, T, P), and we shall denote by T
t
the P-augmentation of the
natural ltration T
W
t
= (W(s); 0 s t). The coecients of /, that is, the interest rate
process r(t), the appreciation rate vector process b(t) = (b
1
(t), . . . , b
d
(t))

of the stocks, and the

volatility matrix-valued process (t) =
ij
(t)
1i,jd
, will all be assumed to be progressively
measurable with respect to T
t
and bounded uniformly in (t, ) [0, T] . We shall also
impose that the following strong non-degeneracy condition on the matrix a(t)

= (t)

(t),

a(t) | |
2
, (t, ) [0, T]
d
(2.3)
holds almost surely, for a given real constant > 0. All processes encountered throughout the
paper will be dened on the xed, nite horizon [0, T], and adapted to the ltration T
t
. We
shall introduce also the relative risk process
(t)

=
1
(t)[b(t) r(t)1

], (2.4)
where 1

= (1, 1, . . . , 1)

. The exponential martingale

Z
0
(t)

= exp
_
t
0

(s)dW(s)
1
2
_
t
0
| (s) |
2
ds, (2.5)
5
the discount process

0
(t)

= exp
_
t
0
r(s)ds (2.6)
and the Brownian motion with drift
W
0
(t)

= W(t) +
_
t
0
(s)ds, 0 t T (2.7)
will be employed quite frequently.
REMARK 2.1. It is a straightforward consequence of the strong non-degeneracy condition
(2.3), that the matrices (t),

(t) are invertible, and that the norms of ((t))

1
, (

(t))
1
are
bounded above and below by and 1/, respectively, for some (1, ); compare with
Karatzas & Shreve (1991) (hereafter abbreviated as [KS]), page 372. The boundedness of b(),
r() and (())
1
implies that of (); therefore, the process Z
0
() of (2.5) is indeed a martingale,
and not just a local martingale.
3 Portfolio, consumption and wealth processes
Consider now a small economic agent, whose actions cannot aect market prices, and who can
decide, at any time t [0, T],
(i) how many shares of bond
0
(t), and how many shares of stocks, (
1
(t),
2
(t), . . . ,
d
(t))

to hold, and
(ii) what amount of money C(t + h) C(t) 0 to withdraw for consumption during the
interval (t, t + h], h > 0. Of course, all these decisions can only be based on the current
information T
t
, without anticipation of the future. More precisely, we have the following.
DEFINITION 3.1. A trading strategy in the market / is a progressively measurable
vector process (
0
(t),
1
(t), . . . ,
d
(t)) such that
_
T
0

2
i
(t)dt < , 0 i d, almost surely.
The processes
0
and
i
represent the number of shares of the bond and the i
th
stock,
respectively, 1 i d, which are held or shorted at any given time t. A short position in the
bond (respectively, the i
th
stock), i.e.,
0
< 0 (resp.,
i
< 0 ), should be thought of as a loan.
DEFINITION 3.2. A cumulative consumption process is a non-negative progressively
measurable process C(t), 0 t T with increasing, RCLL paths on (0, T] (Right Continuous
with Left Limits), and with C(0) = 0, C(T) < a.s.
A basic assumption in the market /, is that trading and consumption strategies should
6
satisfy the so-called self-nancing condition
d

i=0

i
(t)P
i
(t) =
d

i=0

i
(0)P
i
(0) +
d

i=0
_
t
0

i
(u)dP
i
(u) C(t), 0 t T (3.1)
almost surely. The meaning of the equation is that, starting with an initial amount x =

0
(0) +

d
i=1

i
(0)p
i
of wealth, all changes in wealth are due to capital gains (appreciation of
stocks, and interest from the bond), minus the amount consumed.
For both economic and mathematical considerations, it is useful to introduce wealth and
portfolio processes.
DEFINITION 3.3. A portfolio process is a progressively measurable process () =
(
1
(), . . . ,
d
()): [0, T]
d
.
DEFINITION 3.4. For a given initial capital x, a portfolio process () as in Denition
3.3, and a cumulative consumption process C() as in Deniton 3.1, consider the wealth equation
dX(t) = X(t)[1
d

i=1

i
(t)]
dP
0
(t)
P
0
(t)
+
d

i=1
X(t)
i
(t)
dP
i
(t)
P
i
(t)
dC(t)
= X(t)[1
d

i=1

i
(t)]r(t)dt +
d

i=1
X(t)
i
(t)[b
i
(t)dt +
d

j=1

ij
(t)dW
j
(t)] dC(t), (3.2)
= X(t)r(t)dt +X(t)

(t)(t)dW
0
(t) dC(t), X(0) = x,
or equivalently

0
(t)X(t) = x
_
t
0

0
(s)dC(s) +
_
t
0

0
(s)X(s)

(s)(s)dW
0
(s), 0 t T, (3.3)
in the notation of (2.1), (2.2) and (2.5)-(2.7). If this equation has a unique solution X()
X
x,,C
(), this is then called the wealth process corresponding to the triple (x, , C). 2
The interpretation here is that () represent the propotions of the wealth X() which are
invested in the respective stocks i = 1, . . . , d.
REMARK 3.1. In the setup of Denition 3.4, notice that for the stochastic integral to be
well dened we must have
_
T
0
X
2
(t)[[(t)[[
2
dt < , a.s. Furthermore, if we dene

i
(t) =
_
X(t)
i
(t)/P
i
(t) ; i = 1, . . . , d
X(t)(1

d
j=1

j
(t))/P
0
(t) ; i = 0
_
, for 0 t T,
then () = (
0
(),
1
(), . . . ,
d
())

constitutes a trading strategy in the sense of Denition 3.1

and we have
X(t) =
d

i=0

i
(t)P
i
(t), 0 t T, (3.4)
7
as well as the self-nancing condition (3.1), which follows then from the wealth equation (3.2).
Notice that the wealth process X() can clearly take both positive and negative values.
The equation (3.3) leads us to consider the process
N
0
(t)

=
0
(t)X(t) +
_
t
0

0
(s)dC(s) = x +
_
t
0

0
(s)X(s)

(s)(s)dW
0
(s), 0 t T, (3.5)
which is seen to be a continuous local martingale under the so-called risk-neutral probability
measure (or equivalent martingale measure)
P
0
(A)

= E[Z
0
(T)1
A
], A T
T
, (3.6)
in the notation of (2.5).
DEFINITION 3.5. A portfolio/consumption process pair (, C) is called admissible for
the initial capital x , and we write (, C) /(x), if
(i) the pair (), C() obeys the conditions of Denitions 3.2-3.4;
(ii) the solution X
x,,C
() X() of equation (3.2) satises, almost surely:
X
x,,C
(t) , 0 t T. (3.7)
Here, is a nonnegative random variable with E
0
(
p
) < , for some p > 1. 2
The admissibility requirements in Denition 3.5 are imposed in order to prevent pathologies
like doubling strategies (c.f. Harrison & Pliska (1981), Karatzas & Shreve (1995)); such strategies
achieve arbitrarily large levels of wealth at t = T, but require X() to be unbounded from below
on [0, T].
If (, C) /(x), the P
0
-local martingale N
0
() of (3.5) is also bounded uniformly from
below, and is thus a P
0
-supermartingale. Consequently
E
0
[
0
(T)X
x,,C
(T) +
_
T
0

0
(t)dC(t)] x, (, C) /(x). (3.8)
Here E
0
denotes the expectation operator corresponding to the probability measure P
0
of (3.6);
under this measure the process W
0
() of (2.7) is standard Brownian motion, by the Girsanov
theorem (e.g. Karatzas & Shreve (1991), Section 3.5), and the discounted stock processes

0
()P
i
() are martingales, since
dP
i
(t) = P
i
(t)[r(t)dt +
d

i=1

ij
(t)dW
(j)
0
(t)], P
i
(0) = p
i
; i = 1, . . . , d, (3.9)
8
from (2.2) and (2.7), where W
(j)
0
is the j
th
component of W
0
.
REMARK 3.2. For any x and (, C) /(x), let F = X
x,,C
(T). Then for any
a ,= 0, we have X
ax,,aC
() = a X
x,,C
() from (3.2). In particular,
(i) if a > 0: (, aC) /(ax), X
ax,,aC
(T) = aF, a.s.
(ii) if a = 1, C() 0: X
x,,0
(T) = F.
4 Contingent claims and arbitrage in the unconstrained market
The dynamics of the market / become more interesting, once we introduce contingent claims
such as options. Suppose, in particular, that at time t = 0 we sign a contract which gives
us the right (but not the obligation, whence the term option) to buy, at the specied time T
(expiration date), one share of the stock i = 1 at a specied price q (exercise price). At
expiration t = T, if the price P
1
(T, ) of the share is below the exercise price, the contract is
worthless to us; on the other hand, if P
1
(T, ) > q, we can exercise our option at time t = T,
which means to buy one share of the stock at the exercise price q, and then sell the share
immediately in the market for P
1
(T, ). In other words, this contract entitles its holder to a
payment of B(T) B(T, ) = (P
1
(T, )q)
+
at time t = T; it is called a European call option,
in contradistinction with an American call option that can be exercised at any stopping time
(with values) in [0, T]. See Myneni (1992) for a survey on the pricing of American options with
unconstrained portfolios. In this paper we shall deal primarily with the pricing problem under
constraints on portfolio choice, and conne ourselves to European options; the similar problem
for American options will be treated elsewhere.
The following denition generalizes the concept of European call option.
DEFINITION 4.1. A European Contingent Claim (ECC) is a nancial instrument con-
sisting of a payment B(T) at maturity time T; here, B(T) is a nonnegative, T
T
-measurable
random variable with E[(B(T))
1+
] < for some > 0.
We shall denote the price at time t = 0 of the ECC by B(0). The main purpose of this paper
is to nd out what B(0) should be in the market /; in other words, how much an agent should
charge for selling such a contractual obligation, and how much another agent could aord to
pay for it.
It turns out that the answer depends on the structure of the market /. In this section, we
consider the simplest case: that of a complete, unconstrained market, i.e., one in which every
asset can be traded, and unlimited short-selling of both the bond and stocks is also permitted
9
(subject to the admissibility requirements of Denition 3.5). More precisely,
i
() takes values
in , for each 1 i d. In this case the answer to the pricing problem is well known. A
standard approach to this problem is to utilize the concept of arbitrage in the market / with
the ECC, denoted by (/, B) for short, with B standing for the pair (B(0), B(T)).
DEFINITION 4.2. There is an arbitrage opportunity in (/, B), if there exist an initial
wealth x 0 (respectively, x 0), an admissible pair (, C) /(x), and a constant a = 1
(respectively, a = 1), such that
x +a B(0) = X
x,,C
(0) +a B(0) < 0
at time t = 0, and
X
x,,C
(T) +a B(T) 0 a.s.
at time t = T. The values a = 1 indicate long or short positions in the ECC, respectively. 2
This denition of arbitrage is standard in the literature; see, for example, Chapter 6 in
Due (1992) and Myneni (1992). Such an arbitrage opportunity represents a riskless source of
generating prot, strictly bigger than the prot from the bond, by the combination of a trad-
ing/consumption strategy and the ECC. Furthermore, from the scaling properties in Remark
3.2, we know then that the prots from such a scheme are limitless. Such opportunities should
not exist in a well-behaved, rational market.
One of the most interesting classical results on option pricing is that by only excluding
such arbitrage opportunities, the price of the ECC can be uniquely determined, namely as
u
0

= E
0
[
0
(T)B(T)] = E[
0
(T)B(T)Z
0
(T)]. (4.1)
More precisely, if the ECC has a price B(0) > u
0
at time t = 0, then there is an arbitrage
opportunity involving a trading/consumption strategy (
0
, . . . ,
d
, 0)

and a short position in

the ECC; conversely, for any ECC having price B(0) < u
0
, there is also an arbitrage opportunity
using exactly (
1
, . . . ,
d
, 0)

and taking a long position in the ECC. Hence the price for the
ECC has to be u
0
, if no arbitrage is allowed in /. This price is called the arbitrage-free price,
also known as Black-Scholes price. Furthermore, corresponding to the Black-Scholes price u
0
,
there is a hedging portfolio process () (hence also a corresponding trading process ()) and
a consumption process C() 0, such that
X
u
0
,,0
(T) = B(T); (4.2)
10
and with the same portfolio () (hence the opposite trading strategy ()), we have
X
u
0
,0
(T) = B(T). (4.3)
REMARK 4.1. We have u
0
< in (4.1); indeed, with c < denoting a common upper
bound on [[()[[ and [r()[, and with p = 1 +, 1/p + 1/q = 1,
u
0
e
cT
_
E(B(T))
p
_
1/p
_
E(Z
0
(T))
q
_
1/q
e
cT+(q1)c
2
T/2
_
E(B(T))
p
_
1/p
< .
If /is a market with constant coecients b, r, in (2.1) and (2.2), then explicit calculations
are possible for u
0
of (4.1) in the following cases.
EXAMPLE 4.1. European call option, B(T) = (P
1
(T) q)
+
; then
u
0
= p (
+
(T, p)) qe
rT
(

(T, p)), p = P
1
(0) (4.4)
where

(t, p)

=
1

t
[log(p/q) + (r
2
/2)t] and (z) =
1

2
_
z

e
u
2
/2
du (4.5)
is the cumulative standard normal distribution function; we have set =
11
> 0. Furthermore,
the portfolio process in (4.2) and (4.3) satises

1
(t) > 1 and
i
(t) = 0, 2 i d, a.s. (4.6)
We refer the reader to Harrison & Pliska (1981), Cox & Rubinstein (1984), Karatzas (1989),
Karatzas & Shreve (1991) or Due (1992) for details.
EXAMPLE 4.2. Path-dependent (look-back) option
B(T) = max
0tT
P
1
(t).
of Goldman et al. (1979). Then the price of (4.1) is given by
u
0
= pe
rT
_

0
f(T, ; )e

d, p = P
1
(0)
where =
11
> 0,

=
r

2
and
f(t, ; )

= 1
_
t

t
_
+e
2
_
1
_
+t

t
__
. (4.7)
11
Further, the portfolio () of (4.2), (4.3) is given by
i
(t) 0, i = 2, . . . , d and

1
(t) =
e
(t)
f(T t, (t); ) +
_

(t)
f(T t, ; )e

d
e
(t)
+
_

(t)
f(T t, ; )e

d
(4.8)
for 0 t T, where
(t)

= max
0st
(W
0
(s) +s) (W
0
(t) +t) =
1

log
_
max
0st
P
1
(s)
P
1
(t)
_
.
We refer the reader to Karatzas & Shreve (1996), section 2.4 for the details. 2
A main drawback in the above classical argument is its dependence on the assumptions of
completeness and unconstrainedness for the market /. More to the point, as we have seen in
the above discussion, it is critical to be able to use as a trading strategy, if is permitted
in the market, and to trade in all (d+1) assets if necessary. However, if we are in a constrained
market, for instance, a market in which short-selling of stocks is prohibited (i.e., with
i
() 0,
for each i = 1, . . . , d), then the admissibility of the strategy (
0
, . . . ,
d
)

does not imply that

of (
0
, . . . ,
d
)

. Furthermore, in an incomplete market, not all the assets are accessible.

A general arbitrage argument is needed to cover these cases as well as the classical uncon-
strained case.
5 Upper and lower arbitrage prices
Let us introduce now further constraints on portfolio choice, in addition to those of Denition
3.5. Suppose that we are given two nonempty Borel subsets K
+
and K

of
d
; for any x ,
we shall consider portfolio/consumption pairs in the class
/

(x)

= (, C) /(x) : (t) K
+
if X
x,,C
(t) > 0, and (5.1)
(t) K

if X
x,,C
(t) < 0, t [0, T), a.s..
In other words, K
+
( respectively, K

) represents our constraint on portfolio choice when the

wealth is positive (resp., negative). We shall see examples in Section 6 where such dierent
constraints on portfolio, depending on the sign of the level of wealth, arise quite naturally.
DEFINITION 5.1. Given a European contingent claim B(T) as in Denition 4.1, intro-
duce the lower hedging class
L

= x 0 : ( ,

C) /

(x), such that X

x, ,

C
(T) B(T), a.s. (5.2)
12
and the upper hedging class
|

= x 0 : ( ,

C) /
+
(x), such that X
x, ,

C
(T) B(T), a.s. . (5.3)
Here we have set
/

(y)

= ( ,

C) /(y) : (t) K

and X
y, ,

C
(t) 0, 0 t < T, a.s. , for y 0 ,
/
+
(z)

= ( ,

C) /(z) : (t) K
+
and X
z, ,

C
(t) 0, 0 t < T, a.s. , for z 0.
The elements ( ,

C) (resp., ( ,

C)) in the denitions of the classes L and | are called lower
(resp., upper) hedging strategies for the ECC. 2
Clearly, the set L contains the origin. On the other hand, it is a straightforward consequence
of the Denition 5.1, that both sets L and | are (connected) intervals. More precisely, we have
the following result.
PROPOSITION 5.1 For any x
1
L, 0 y
1
x
1
implies y
1
L. Similarly, for any
x
2
|, y
2
x
2
implies y
2
|.
PROOF. Suppose (
2
, C
2
) /(x
2
) satises the conditions of (5.3). Then, with y
2
x
2
,
one just consumes immediately the amount y
2
x
2
; in other words, with

C
2
(t) = C
2
(t) +
(y
2
x
2
) 1
(0,T]
(t), we have X
y
2
,
2
,

C
2
(t) X
x
2
,
2
,C
2
(t) for all 0 < t T, and thus y
2
|. A
similar argument works for L. 2
The purpose of this section is to show that, in the presence of constraints as in (5.1), the
Black-Scholes price u
0
= E
0
[
0
(T)B(T)] is replaced by an interval [h
low
, h
up
] which contains
u
0
and is dened by (5.4) below, in the following sense: If B(0), the price of the ECC at time
t = 0,
(i) does not belong to [h
low
, h
up
], then there exists an arbitrage opportunity (Theorem 5.2);
(ii) belongs to the interior (h
low
, h
up
) of this interval, or if B(0) = h
low
= h
up
, then
arbitrage opportunities do not exist (Theorem 5.3 and Corollary 5.1).
DEFINITION 5.2. The lower arbitrage and the upper arbitrage prices are dened by
h
low

= supx : x L, h
up

= infx : x |, (5.4)
respectively.
Here we adopt the convention that inf = +. In Section 6 we shall provide characteri-
zations of the numbers h
low
, h
up
in terms of suitable stochastic control problems, which lead
to explicit computation in several interesting special cases (cf. Section 8).
13
REMARK 5.1. Heuristically, the upper arbitrage price may be viewed as the minimal
amount necessary for the seller of the ECC to set aside at time t = 0, in order to make sure
that he will be able to cover his obligation at time t = T. Similarly, the lower arbitrage price
can be viewed as the maximal amount that the buyer of the ECC is willing to pay at t = 0,
and still be sure that he will be able to cover, at time t = T, the debt he incurred at t = 0 by
purchasing the ECC.
This intuition suggests that the lower arbitrage price h
low
cannot be larger than the upper
arbitrage price h
up
. The following theorem shows in fact that for general constraint sets K
+
and K

, a stronger result holds.

THEOREM 5.1 We have for any nonempty constraint sets K
+
and K

in B(
d
),
0 h
low
u
0
h
up
,
where u
0
= E
0
[
0
(T)B(T)] is the Black-Scholes price of (4.1).
PROOF. By (3.8) and the denition of |, we get
x E
0
_

0
(T)X
x, ,

C
(T) +
_
T
0
(s)d

C(s)
_
E
0
[
0
(T)B(T)] = u
0
, x |.
Hence, h
up
u
0
. Similarly,
y E
0
_

0
(T)X
y, ,

C
(T) +
_
T
0

0
(s)d

C(s)
_
E
0
[
0
(T)(B(T))] = u
0
, y L,
whence y u
0
and h
low
u
0
. 2
One feature of the above theorem is that it holds for any constraint sets, therefore it is
applicable to many situations. For instance, in the case of a European call-option B(T) =
(P
1
(T) q)
+
on the rst stock, and assuming that this stock can be traded, we have P
1
(0) |,
and thus: 0 h
low
u
0
h
up
P
1
(0) < .
We dene the notion of arbitrage with portfolios constrained as in (5.1), by analogy with
Denition 4.2.
DEFINITION 5.3. We say that there exists in (/, B) an arbitrage opportunity with
constrained portfolios, if there exists an initial wealth x 0 (resp., x 0), an admissible
portfolio/consumption process pair (, C) in the class /
+
(x) (respectively, /

(x)) of Denition
5.1, and a constant a = 1 (resp., a = 1) such that
x +a B(0) = X
x,,C
(0) +a B(0) < 0 (5.5)
14
and
X
x,,C
(T) +a B(T) 0, a.s. (5.6)
Again, the values a = 1 represent long or short positions in the ECC, respectively. 2
THEOREM 5.2 For any ECC price B(0) > h
up
, there exists an arbitrage opportunity in
the sense of Denition 5.3; similarly for any ECC price B(0) < h
low
.
PROOF. Suppose that B(0) > h
up
; then for any x
1
(h
up
, B(0)) we know that x
1
|,
by the denition of h
up
. Thus there exists a ( ,

C) /
+
(x
1
) such that
X
x
1
, ,

C
(0) B(0) = x
1
B(0) < 0,
and
X
x
1
, ,

C
(T) B(T) B(T) B(T) = 0, a.s.
Hence (5.5) and (5.6) in Denition 5.3 are satised with a = 1.
For the case B(0) < h
low
, there is an arbitrage opportunity which satises (5.5) and (5.6)
with a = 1. The argument is similar to the rst one, and we omit the details. 2
THEOREM 5.3 For any ECC price B(0) , (|

L), there is no arbitrage in (/, B) with

constrained portfolios.
PROOF. We shall prove this by contradiction. Suppose B(0) , |, B(0) , L and that there
is an arbitrage opportunity in (/, B) with constrained portfolios. Two cases may arise.
Case 1: The arbitrage opportunity satises (5.5) and (5.6) with a = 1. In this case, there
exist an initial wealth x [0, ) and a pair (
1
, C
1
) /
+
(x), such that
x = X
x,
1
,C
1
(0) < B(0)
and
X
x,
1
,C
1
(T) B(T), a.s. (5.7)
From (5.7) and the denition of | we know that x = X
x,
1
,C
1
(0) |, whence B(0) |, thanks
to x < B(0) and Proposition 5.1; a contradiction.
Case 2: The arbitrage opportunity satises (5.5) and (5.6) with a = 1. The proof is similar
to that of Case 1, so we omit the details. 2
15
COROLLARY 5.1 If h
low
< h
up
, then for any price B(0) (h
low
, h
up
) of the ECC
there is no arbitrage opportunity in (/, B) with constrained portfolios.
In view of Theorems 5.2 and Corollary 5.1, the interval [h
low
, h
up
] is the best possible
interval for the ECC price that one can obtain by using only arbitrage arguments. We shall
call [h
low
, h
up
] arbitrage-free interval.
REMARK 5.2. In an unconstrained market, i.e., with K
+
= K

=
d
, we know from
the classical results that the Black-Scholes price u
0
belongs to both the lower hedging class L
and the upper hedging class | (see Chapter 6 in Due (1992)); thus we have h
low
= h
up
= u
0
according to Theorem 5.1.
REMARK 5.3. If the option price is equal to one of the two endpoints h
low
or h
up
, it may
well be that in some situations there is no arbitrage, while in others there may be an arbitrage
opportunity, depending on the consumption process. For example, in the unconstrained case,
if B(0) = h
up
= u
0
, there is no arbitrage, as it can be shown that the consumption process for
the hedging strategy is almost surely zero (see [KS], p.378). On the other hand, if B(0) = h
up
,
h
up
| and

C(T) > 0 a.s. (for instance, as in Remark 8.1), then this consumption can be
viewed as a kind of arbitrage opportunity.
Within the arbitrage-free interval, a unique fair price might be determined by considerations
based on utility maximization, or on a stochastic game between the buyer and the seller. An
approach using utility maximization, originally due to Davis (1994), is discussed in detail in
Section 7.
6 Representations for convex constraints
We shall concentrate in this section on the important special case where the constraint sets
K
+
, K

of (5.1) are nonempty closed, convex sets in

d
. For such sets, we shall obtain in this
section representations of h
low
, h
up
in terms of auxiliary stochastic control problems (cf. (6.7)
and (6.8)), which will lead in turn to explicit computations in Section 8.
We start by introducing the functions
(x)

= sup
K
+
(

x) :
d

_
+
and

(x)

= inf
K

x) :
d

_
.
16
In the terminology of convex analysis, () and

() are the support functions of the convex sets

K
+
and K

, respectively; they are closed, positively homogeneous, proper convex functions

on
d
(Rockafellar (1970), p.114). The support functions () and

() are nite on their

eective domains

K
+
and

K

, respectively, where,

K
+

= x
d
; s.t.

x , K
+
= x
d
; (x) < ,

= x
d
; s.t.

x , K

= x
d
;

(x) > .
Notice that both

K
+
and

K

are convex cones. The following two assumptions will be imposed

throughout this section.
ASSUMPTION 6.1. The functions () and

() are continuous on

K
+
and

K

, respec-
tively.
ASSUMPTION 6.2. The function () is bounded uniformly from below by some real
constant.
These two assumptions are satised by all of the examples below. In particular, Theorem
10.2, p.84 in Rockafellar (1970) guarantees that Assumption 1 is satised, if

K
+
,

K

are locally
simplicial; and Assumption 6.2 is satised if and only if K
+
contains the origin.
The convex constraints are perhaps among the most important constraints that arise in
practice. A few of them are listed below.
EXAMPLE 6.1. All of the following examples satisfy the Assumptions 6.1 and 6.2.
(i) Unconstrained case:
d+1
. In other words, K
+
= K

=
d
. Then

K
+
=

K

= 0
and =

0 on

K
+
and

K

, respectively.
(ii) Prohibition of short-shelling of stocks:
i
0, 1 i d. In other words, K
+
= [0, )
d
,
K

= (, 0]
d
. Then

K
+
=

K

= [0, )
d
, and 0 on

K
+
,

0 on

K

.
(iii) Constraints on the short-selling of stocks: A generalization of (ii) is K
+
= [k, )
d
for some k 0 and K

= (, l]
d
for some l . Then

K
+
=

K

= [0, )
d
, and (x) =
k

d
i=1
x
i
,

(x) = l

d
i=1
x
i
on

K
+
and

K

, respectively.
(iv) Incomplete market, in which only the rst m stocks can be traded:
i
= 0, i =
m + 1, . . . , d for some xed m 1, . . . , d 1, d 2. In other words, K
+
= K

d
;
i
= 0, i = m + 1, . . . , d. Then

K
+
=

K

= x
d
; x
i
= 0, i = 1, . . . , m and
=

0 on

K
+
and

K

.
17
(v) Incomplete market, with prohibition of investment in the rst m stocks:
i
= 0, 1 i
m for some 1 m d, d 2. In other words, K
+
= K

=
d
;
i
= 0, 1 i m. Then

K
+
=

K

= x
d
; x
m+1
= . . . = x
d
= 0, and =

0 on

K
+
and

K

.
(vi)Both K
+
and K

are closed, convex cones in

d
. Then

K
+
(

K

) = x
d
;

x
0, K
+
(K

) and (

) 0 on

K
+
(

K

). This clearly generalizes all the previous examples

except (iii).
(vii)Prohibition of borrowing:
0
0. In other words, K
+
=
d
:

d
i=1

i
1,
K

=
d
;

d
i=1

i
1. Then

K
+
=

K

= x
d
: x
1
= x
2
= = x
d
0, and
(x) = x
1
on

K
+
,

(x) = x
1
on

K

.
(viii) Constraints on borrowing: A generalization of (vii) is K
+
=
d
:

d
i=1

i
k,
for some k 1 and K

d
i=1

i
l for some l . Then

K
+
=

K

= x
d
: x
1
=
= x
d
0, (x) = kx
1
on

K
+
,

(x) = lx
1
on

K

.
Explicit formulae or bounds for h
low
and h
up
, for all these examples, in the case of a
European call option in a market with constant coecients, will be presented in detail in
Section 8. It is interesting to notice that, for all these examples,

K
+
is equal to

K

(in this
connection, see also Proposition 7.2). In general, this will not be the case; see Example 8.8.
The technique to handle such convex constraints is developed in Cvitanic & Karatzas (1993),
hereafter abbreviated as [CK2]. The basic idea is to introduce a family of auxiliary markets, in
which the unconstrained (hedging) problem is relatively easy to solve, and then try to come back
to the original market. This basic idea will help us here to give representations for the lower
arbitrage price h
low
and the upper arbitrage price h
up
, in terms of appropriate stochastic
control problems which involve optimization with respect to parameters of the auxiliary
markets.
In order to introduce these families of auxiliary markets, the notation of sections 5 and 6 in
[CK2] will be carried over here for K
+
; in addition, we shall consider the analogous notation
for K

. Dene the class H (resp.,

H) to be the set of progressively measurable process =
(t), 0 t T with values in

K
+
(resp.,

K

), which satisfy E
_
T
0
([[(t)[[
2
+ ((t)))dt <
(resp., E
_
T
0
([[(t)[[
2
+

((t)))dt < ); also introduce, for every H

H, the analogues

(t)

= (t) +
1
(t)(t),

(t)

= exp[
_
t
0
r(s) +((s)ds],

(t)

= exp[
_
t
0
r(s) +

((s))ds], (6.1)
18
Z

(t)

= exp[
_
t
0

(s)dW(s)
1
2
_
t
0
|

(s)|
2
ds], (6.2)
W

(t)

= W(t) +
_
t
0

(s)ds, (6.3)
of the processes in (2.4)-(2.7), as well as the measure
P

(A)

= E[Z

(T)1
A
] = E

[1
A
], A T
T
(6.4)
by analogy with (3.6). Finally, denote by T (resp.,

T) the subset consisting of the processes
H (resp.,

H) such that is bounded uniformly in (t, ) [0, T] :
sup
(t,)[0,T]
[[(t, )[[ < . (6.5)
Therefore, for every T

T, the exponential local martingale Z

() of (6.2) is actually a
martingale, from which we conclude that the measure P

of (6.4) is a probability measure and

the process W

() of (6.3) is a P

-Brownian motion, by the Girsanov theorem; in terms of this

new Brownian motion W

(), the stock price equations (2.2) can be re-written as

dP
i
(t) = P
i
(t)
_
(r(t)
i
(t))dt +
d

j=1

ij
(t)dW
(j)

(t)
_
, i = 1, . . . , d. (6.6)
In the special case of an incomplete market (Example 6.1 (iv)), this equation shows that the
discounted prices
0
()P
i
(), i = 1, . . . , d are martingales under every probability measure in the
class P

D
of (6.4).
THEOREM 6.1 With the above notation, we have:
(i) the lower arbitrage price is given by
h
low
= inf

D
E

(T)B(T)] =: g, (6.7)
provided that the function

() is bounded uniformly from below by some real constant;
(ii) the upper arbitrage price is given by
h
up
= sup
D
E

(T)B(T)], (6.8)
and if the right-hand side of (6.8) is nite, then h
up
|.
In particular, taking 0 in (6.7) and (6.8) we recover the result 0 h
low
u
0
h
up
of Theorem 5.1. For T (resp.

T), observe that the number E

(T)B(T)] (resp.
19
E

(T)B(T)]) is exactly the Black-Scholes price of the contingent claim in a new auxiliary
market with unconstrained portfolios.
Notice that

0 in all the cases of Example 6.1, except in (iii) when l > 0, and in (viii)
when l < 0. We shall treat these two cases separately (see Example 8.1 and Example 8.2) by
employing the denition of h
low
directly.
The representation (6.8) for h
up
is proved as in [CK2], although a set bigger than our T
is used there, so we only need to establish (6.7). As in [CK2], the proof uses the martingale
representation and Doob-Meyer decomposition theorems, and relies on the construction of a
submartingale with regular sample paths.
Let us denote by o the set of all T
t
-stopping times with values in [0, T], and by o
,
the subset of o consisting of stopping times such that () () (), , for any
two stopping times o and o such that a.s. For every o, consider also the
T

-measurable random variables

V ()

= essinf

D
E

_
B(T) exp
_
T

(r(s) +

((s)))ds

_
(6.9)
and

()

=

V ()e

0
(r(u)+

((u)))du
=

V ()

(),

T. (6.10)
LEMMA 6.1 For every

T, o, o
,T
we have the submartingale property

() E

[

Q

()[T

], a.s.
LEMMA 6.2 There exists a RCLL modication

V
+
() of the process

V (). Furthermore,
if we dene

Q
+

() by analogy with (6.10 ), then

Q
+

(t), T
t
, 0 t T is a P

-submartingale
with RCLL paths.
The proofs of Lemma 6.1 and Lemma 6.2 are carried out in a manner similar to that of the
Appendix in [CK2].
LEMMA 6.3 For the processes

V (),

Q

() of (6.9) and (6.10) we have

E
0
[ sup
0tT
(

V (t))
p
] < , p (1, 1 +), (6.11)
E

[ sup
0tT

(t)] < ,

D. (6.12)
In particular, E

(T)B(T)] = E

(

Q

(T)) < ,

T.
20
PROOF. From (6.9) it follows that
0

V (t) E
0
[B(T)e

_
T
t
r(s)ds
[T
t
] e
cT
B(t), 0 t T (6.13)
holds almost surely, in the notation of Remark 4.1 and with B(t)

= E
0
[B(T)[T(t)]. Now with
1 < p < 1 + , r = (1 + )/p and 1/r + 1/s = 1, we have from the Holder inequality and the
Doob maximal inequality:
E
0
[ sup
0tT
(B(t))
p
] const E
0
(B(T))
p
= const E[Z
0
(T)(B(T))
p
]
const (E[(B(T))
pr
])
1/r
(E[(Z
0
(T))
s
])
1/s
.
Therefore,
E
0
[ sup
0tT
(B(t))
p
] const (E[(B(T))
1+
])
1/r
exp(
s 1
2
c
2
T) < , (6.14)
which proves (6.11) in conjuction with (6.13).
On the other hand, from (6.13), (6.10), and the assumption that

() is uniformly bounded
from below by some real constant, we obtain that
0

Q

(t) =

V (t) exp[
_
t
0
(r(s) +

((s)))ds] const B(t), 0 t T (6.15)
also holds almost surely. With 1 < p < 1 +, 1/p + 1/q = 1 we get then, for any xed

D:
E

[ sup
0tT
B(t)] = E
0
_
Z

(T)
Z
0
(T)
sup
0tT
B(t)
_

_
E
0
[ sup
0tT
(B(t))
p
]
_
1/p

_
E
0
_
Z

(T)
Z
0
(T)
_
q
_
1/q
< .
We have used again the Holder and Doob inequalities, (6.14), as well as the uniform boundedness
of the process
1
()() in
Z

(t)
Z
0
(t)
= exp
_
t
0
(
1
(s)(s))

dW
0
(s)
1
2
_
t
0
[[
1
(s)(s)[[
2
ds, 0 t T.
In conjuction with (6.15), this leads then to (6.12). 2
PROOF OF THEOREM 6.1. The proof is similar to [CK2]. From now on we consider only
the RCLL modications of

V and

Q

, hence we can assume that these processes do have RCLL

paths.
Part I. We shall rst prove the inequality h
low
g. This is obvious if g = 0 so let us assume,
for the remainder of this part of the proof, that g > 0, and try to show that g L. From Lemma
6.2 and (6.12),

Q

() is a submartingale of class T[0, T] under P

, for every

D. Thus from
21
the martingale representation theorem (section 3.4 in [KS]) and the Doob-Meyer decomposition
for submartingales (section 1.4 in [KS]), we have for every

T :

(t) =

V (0) +M

(t) +A

(t) = g +
_
t
0

(s)dW

(s) +A

(t), 0 t T (6.16)
where M

(t) =
_
t
0

(s)dW

(s), 0 t T, is an (T
t
, P

)-martingale,

() is an
d
-valued,
T
t
-progressively measurable and almost surely square integrable process, and A

() is T
t
-
predictable with increasing, RCLL paths and A

(0) = 0, E

(T) < . Introduce the negative

process

X(t)

=

V (t) =

(t)

(t)
, 0 t T, for every

T. (6.17)
Then

X(0) =

V (0) = g, and

X(T) = B(T).
Hence, in order to show g L, it is enough to nd an admissible pair ( ,

C) /

(g) such
that

X() = X
g, ,

C
(); recall from (6.11) that

V () =

X() is dominated by the random
variable = sup
0tT

V (t) 0 with E
0
(
p
) < for some p > 1.
Let us start by observing that for any

T,

T we have from (6.10),

(t) =

Q

(t) exp
_
_
t
0

((u)du
_
t
0

((u))du
_
, 0 t T.
Thus, from the dierential form of (6.16) we get
d

Q

(t) = exp[
_
t
0

((s))ds
_
t
0

((s))ds] [

Q

(t)

((t))

((t))dt +

(t)dW

(t) +dA

(t)]
= exp[
_
t
0

((s))ds
_
t
0

((s))ds] [

X(t)

(t)

((t))

((t))dt (6.18)
+dA

(t) +

(t)
1
(t)((t) (t))dt +

(t)dW

(t)],
where the last equation comes from the denition of

X() and the connection between W

()
and W

() (cf. (6.3)). Comparing (6.18) with the Doob-Meyer decomposition

d

Q

(t) =

(t)dW

(t) +dA

(t), (6.19)
we conclude from the uniqueness of this decomposition that

(t) exp[
_
t
0

((s))ds] =

(t) exp[
_
t
0

((s))ds], 0 t T,
22
so that the process
h(t)

=

(t) exp[
_
t
0

((s))ds], 0 t T, does not depend on . (6.20)

We claim that we also have, almost surely,
_
T
0
1
{

X(t)=0}
[[h(t)[[
2
dt = 0. (6.21)
Indeed, consider the nonnegative P
0
-submartingale Q()

Q
0
() of (6.16). From the Tanaka-
Meyer formula (Meyer (1976), p.365, equations (12.1), (12.3)) we have
Q(t) = g +
_
t
0
1
{Q(s)>0}
dQ(s) + (t) +

0<st
1
{Q(s)=0}
Q(s),
where () is the local time of Q() at the origin: a continuous increasing process, at o the
set 0 t T; Q(t) = 0, a.s. Comparing this expression with (6.16), we obtain that
M(t)

=
_
t
0
1
{Q(s)=0}
dM
0
(s) = (t)+

0<st
1
{Q(s)=0}
Q(s)
_
t
0
1
{Q(s)=0}
dA
0
(s), 0 t T
is a continuous martingale of bounded variation. Thus, its quadratic variation
<M> (T) =
_
T
0
1
{Q(t)=0}
d <M
0
> (t) =
_
T
0
1
{Q(t)=0}
[[h(t)[[
2
dt
is almost surely equal to zero, and (6.21) follows (recall here that M
0
(t) =
_
t
0

0
(s)dW
0
(s) =
_
t
0
h

(s)dW
0
(s) from ((6.16) and (6.20)).
Therefore, if we x an arbitrary K

and dene
(t)

=
1

0
(t)

X(t)
(

(t))
1
h(t) 1
{

X(t)<0}
+ 1
{

X(t)=0}
, (6.22)
we obtain a portfolio process that satises almost surely

0
(t)

X(t)

(t)(t) = h

(t), a.e. on [0, T].

From this and from (6.18)-(6.20), we have
exp[
_
t
0

((s))ds
_
t
0

((s))ds]
[

X(t)

(t)

((t))

((t))dt +dA

(t) +

(t)
1
(t)((t) (t))dt] = dA

(t),
whence
exp[
_
t
0

((s))ds
_
t
0

((s))ds]
23
[

X(t)

(t)

((t)) +

(t)(t)

((t))

(t)(t)dt +dA

(t)] = dA

(t),
thanks to (6.22). Therefore, the process

C() dened as

C(t)

=
_
t
0

1

(s)dA

(s)
_
t
0

X(s)[

((s)) +

(s) (s)]ds, 0 t T, (6.23)

is independent of

T. In particular, taking 0, we see that

C(t) =
_
t
0

1
0
(s)dA
0
(s), 0 t T
is an increasing, adapted, RCLL process with

C(0) = 0 and

C(T) < almost surely. In other
words,

C() is a consumption process.
The same argument as on p. 664 of [CK2] shows then that

((s)) +

(s) (s) 0, 0 s T
holds almost surely, for every

T. Therefore, the proof in [CK1], p. 782-783, and Theorem
13.1 in Rockafellar (1970), p. 112, give us () K

, a.s. Notice that, for these arguments to

work, we need the continuity of

() as well as the condition that

() be bounded uniformly
from below by some real constant.
Now putting the various pieces together, we obtain
d(

X(t)

(t)) = d

Q

(t) =

(t)dW

(t) +dA

(t)
=

(t)[d

C(t) +

X[

((t)) +

(t) (t)]dt

X(t)

(t)(t)dW

(t)],
so that,
d(

X(t)

(t)) =

(t)d

C(t)

(x)

X(t)[

((t)) +

(t) (t)]dt (6.24)

+

(t)

X(t)

(t)(t)dW

(t).
Taking 0 in (6.24), we obtain the wealth equation (3.2) in the form
d(
0
(t)

X(t)) =
0
(t)d

C(t) +
0
(t)

X(t)

(t)(t)dW
0
(t),

X(0) = g,
whence

X() = X
g, ,

C
(). The proof of h
low
g is now complete.
Part II. Let us consider the proof of the reverse inequality h
low
g. This is obvious if
h
low
= 0, so we assume that h
low
= 0. Thus we have L ,= in (5.2), and for any x L there
exists (, C) /

(x) such that X

x,,C
(T) B(T) almost surely. It is easy to see from
(3.3) and (6.1) that the analogue of (6.24) holds, and thus

(t)X
x,,C
(t) +
_
t
0

(s)dC(s) +
_
t
0

(s)X
x,,C
(s)[

((s)) +

(s)(s)]ds, 0 t T
24
is actually a P

-local martingale, whence a supermartingale. This is because

()X
x,,C
() is
bounded from below by a P

-integrable random variable, thanks to (3.7), (6.5), and the Holder

inequality. Therefore,
x E

(T)X
x,,C
(T) +
_
T
0

(s)dC(s) +
_
T
0

(s)X
x,,C
(s)(

((s)) +

(s)(s))ds
_
E

(T)(B(T))]
for every x L,

T, or equivalently x E

(T)B(T)], from which h

low
g follows. 2
7 A fair price
We have seen that, if the upper arbitrage price h
up
is strictly bigger than the lower arbitrage
price h
low
, then the arbitrage argument alone is not enough to determine a unique price for
the contingent claim. Several approaches have been proposed to get around this problem in
the special case of incomplete markets (as in Example 6.1 (iv)); see, for example, Follmer &
Sondermann (1986), Follmer & Schweizer (1991), Due & Skiadas (1991), Foldes (1990) and
Davis (1994). There are also some approaches that have been suggested in dierent, but related,
contexts, such as pricing in the presence of transaction costs (Hodges & Neuberger (1989)) or
under dierent interest rates for borrowing and saving (Barron & Jensen (1990)). Although
perhaps none of them is totally satisfactory, we shall try in this section to generalize one of
them, the Davis (1994) approach, to the constrained setup of Section 5.
The purpose of this section is not to solve the problem completely (because it might turn out
that, from a practical point of view, the most convenient price to use is still the Black-Scholes
price u
0
; cf. Remark 11.4 in Section 11), but rather to see when the generalization works and
when it does not, and hopefully to focus attention on the study of possible connections between
arbitrage and utility maximization.
7.1 Denition
Daviss fair price is only dened for an agent with positive wealth and involves the concept
of utility function. Before presenting the denition of the fair price, we shall briey recall that
of utility function.
DEFINITION 7.1. A function U : (0, ) will be called a utility function, if it is
strictly increasing, strictly concave, of class C
1
and satises
U(0)

= U(0+), U

(0+)

= lim
x0
U

(x) = , U

()

= lim
x
U

(x) = 0.
25
We shall denote by I() the inverse of the function U

(). Notice that the function I() maps

(0, ) onto itself and satises
I(0+) = , I() = 0, U

(I(x)) = x.
Consider the following constrained portfolio optimization problem
V (x)

= sup
(,C)A
+
(x)
E
_
U
_
X
x,,C
(T)
__
, 0 < x < , (7.1)
where one tries to maximize expected terminal utility over portfolio/consumption pairs in the
class /
+
(x) of Denition 5.1. Clearly, we have
V (x) EU
_
xexp[
_
T
0
r(t)dt]
_
U(xe
r
0
T
) > ,
where r
0
is a lower-bound on r().
ASSUMPTION 7.1. For all x > 0, the value V (x) of (7.1) is attainable; in other words,
V (x) = E
_
U
_
X
x

(T)
__
, where X
x

(T)

= X
x,

,C

(T), (7.2)
for some (

, C

) /
+
(x), and we assume that the derivative of V () exists and is strictly
positive: V

() > 0, on (0, ). 2
This assumption is satised in many interesting cases, in particular with C

() = 0. Indeed,
it holds for all convex constraint sets K
+
, subject to the rather mild Assumptions 6.1 and 6.2;
see 7.3.
Suppose that at time t = 0, the price of the contingent claim is p = B(0) and one diverts
an amount , [[ < x, of money into the contingent claim B (i.e, buys /p shares of the
contingent claim). Then one chooses an optimal portfolio/consumption strategy to achieve
maximal expected utility from terminal wealth. Formally, one solves the stochastic control
problem
W(, p, x)

= sup
(,C)A

(x)
EU
_
X
x,,C
(T) +

p
B(T)
_
, [[ < x, (7.3)
where we set formally U(x) = for x < 0. Notice that W(0, p, ) coincides with the function
V () of (7.1) for every p > 0, and that we can actually take X
x,,C
(T) > 0 in (7.3) above. If
the contingent claim price p is set so that this small diversion of funds has a neutral eect on
W, in the sense
W

(0, p, x) = 0, (7.4)
26
then we tend to call this p the fair price of the contingent claim. Indeed, Davis (1994) uses
exactly (7.4) to dene the fair price. However, the dierentiability of W(, p, x) is often dicult
to check directly. Here we shall use a requirement weaker than dierentiability, and reminiscent
of the notion of viscosity solutions as in Crandall and Lions (1983).
DEFINITION 7.2 For a given x > 0, we call p a weak solution of (7.4) if, for ev-
ery dierentiable function (, p, x) satisfying (, p, x) W(, p, x) for all (x, x), and
(0, p, x) = W(0, p, x) V (x), we have

(0, p, x) = 0.
Notice the similarity of this notion with that of viscosity subsolution (see, for example
Denition 7.2 in Shreve and Soner (1994), or Fleming and Soner (1993) p. 66).
DEFINITION 7.3 Suppose that for any given x > 0, the weak solution p = p(x) >0 of
(7.4) is unique. Then we call this p(x) the fair price for the contingent claim at time t = 0,
corresponding to initial wealth x > 0.
In economic terms, the requirement (7.4) postulates a zero marginal rate of substitution
for W(, p(x), x) at = 0. Generally speaking, Daviss fair price depends on the utility function
U() and on the particular initial wealth x > 0. However, for convex constraint sets K
+
and
K

, we shall present in 7.3 conditions under which p(x) can be rendered independent of the
utility function U() and/or the initial wealth x > 0.
7.2 Connections with Arbitrage
An immediate question that we have to settle, is whether there exist any arbitrage opportunities
in (/, B) if the contingent claim price B(0) is set to be p(x). In other words, whether p(x)
belongs or not to the interval [h
low
, h
up
], for every initial wealth x > 0. In general, the answer
can be armative or negative, depending on the constraint sets K
+
and K

(indeed, several
counterexamples are given in Section 8.3); however, if we adopt the fairly general Assumption
7.2 below, then the answer is always armative.
ASSUMPTION 7.2. Suppose that (
(1)
, C
(1)
) /

(x) and (
(2)
, C
(2)
) /

(y), for
arbitrary but xed x , y . Then there exists a (, C) /

(x + y) such that the

corresponding terminal wealth X
x+y,,C
(T) is obtained by superposition:
X
x+y,,C
(T) = X
x,
(1)
,C
(1)
(T) +X
y,
(2)
,C
(2)
(T), a.s.
27
THEOREM 7.1 Suppose that the Assumptions 7.1 and 7.2 are satised, and that the fair
price p(x) exists for every x > 0; then
x > 0, h
low
p(x) h
up
. (7.5)
The meaning of Assumption 7.2 is that, whenever an agent chooses to invest in two dierent
accounts X
1
() X
x,
(1)
,C
(1)
() and X
2
() X
y,
(1)
,C
(2)
() separately, then this is equivalent, in
terms of terminal wealth, to investing and consuming according to some strategy (, C) which is
admissible for the initial wealth level x+y, for arbitrary real numbers x and y. This assumption
holds, in particular, if the pair
= (
(1)
X
1
+
(2)
X
2
)/(X
1
+X
2
), C = C
(1)
+C
(2)
is indeed in /

(x +y). A sucient condition, for Assumption 7.2 to hold in the case of convex
sets K

, is given along these lines in Proposition 7.1 below.

PROOF OF THEOREM 7.1: We establish the upper bound rst. Suppose that p(x) > 0
is the fair price of Denition 7.3 for the initial wealth x > 0. For arbitrary y |, we want
to show that p(x) y. Now for any (x p(x)/y, 0) (x, 0), by Remark 3.2 and the
denition of the class | in (5.3), there exists an admissible pair (
(1)
, C
(1)
) /
+
() with

= (y/ p(x)) (0, x), such that
X
,
(1)
,C
(1)
(T) (/ p(x)) B(T) (7.6)
holds almost surely. On the other hand, by Assumption 7.1, there is an admissible pair
(
(2)
, C
(2)
) /
+
(w) which is optimal for the problem of (7.1) with the initial wealth
w = x +y/ p(x) = x > x > 0 (recall that < 0),
i.e., the resulting wealth process X
w,
(2)
,C
(2)
() 0 satises
V (w) = V (x ) = E
_
U
_
X
w,
(2)
,C
(2)
(T)
__
. (7.7)
Thus, from Assumption 7.2, we know that there is an admissible pair (
(3)
, C
(3)
) /

(x )
such that
X
x,
(3)
,C
(3)
(T) = X
,
(1)
,C
(1)
(T) +X
w,
(2)
,C
(2)
(T) X
w,
(2)
,C
(2)
(T)

p(x)
B(T), (7.8)
28
by (7.6). Hence, by the denition of W in (7.3),
W(, p(x), x) EU
_
X
x,
(3)
,C
(3)
(T)+(/ p(x))B(T)
_
EU
_
X
w,
(2)
,C
(2)
(T)
_
= V
_
x+y/ p(x)
_
,
thanks to (7.8) and (7.7). Therefore, for any function as in Denition 7.2, we have
(, p(x), x) (0, p(x), x)

W(, p(x), x) V (x)

V (x +y/ p(x)) V (x)

,
since < 0, and in the limit, as 0,
0 =

(0, p(x), x)
_
y
p(x)
1
_
V

(x).
Since V

(x) > 0 by Assumption 7.1, we obtain y p(x), from which the upper bound in (7.5)
follows.
Now consider the lower bound. For arbitrary z L, we want to show z p(x). Given
any (0, x), again by Remark 3.2 and the denition of L, we know that there exists a pair
(
(4)
, C
(4)
) /

() with

= z/ p(x) > 0 such that
X
,
(4)
,C
(4)
(T) (/ p(x))(B(T)), a.s. (7.9)
Also by Assumption 7.1, there exists a pair (
(5)
, C
(5)
) /
+
() where

= x + =
x +z/ p(x) > 0, with corresponding wealth process X
,
(5)
,C
(5)
() 0 which satises
V () = V (x +z/ p(x)) = E
_
U
_
X
,
(5)
,C
(5)
(T)
__
. (7.10)
From Assumption 7.2, we know that there exists a pair (
(6)
, C
(6)
) /

(x ) such that
X
x,
(6)
,C
(6)
(T) = X
,
(4)
,C
(4)
(T) +X
,
(5)
,C
(5)
(T) X
,
(5)
,C
(5)
(T)

p(x)
B(T), (7.11)
almost surely. Therefore,
W(, p(x), x) EU
_
X
x,
(6)
,C
(6)
(T)+(/ p(x))B(T)
_
EU
_
X
,
(5)
,C
(5)
(T)
_
= V (x+z/ p(x)),
via (7.9), (7.10) and (7.11). Thus, for any function as in Denition 7.2, we have
(, p(x), x) (0, p(x), x)

V (x +z/ p(x)) V (x)

, (0, x),
and in the limit, as 0,
0 =

(0, p(x), x)
_
z
p(x)
1
_
V

(x).
29
Again, V

(x) > 0 leads to the lower bound z p(x) of (7.5). 2

REMARK 7.1. It is readily seen that the rst part of the proof of Theorem 7.1 goes
through, and thus the upper bound p(x) h
up
of (7.5) is valid, even in the absence of As-
sumption 7.2, provided that the set K
+
is convex.
PROPOSITION 7.1 If the constraint sets K
+
and K

are convex, then a sucient con-

dition for the validity of Assumption 7.2 is

+
K
+
,

:
+
+ (1 )

_
K
+
, if 1
K

, if 0
_
. (7.12)
PROOF. For x
i
and (
(i)
, C
(i)
) /

(x
i
), let X
i
() X
x
i
,
(i)
,C
(i)
(), i = 1, 2 be the
corresponding wealth processes and dene C()

= C
(1)
() + C
(2)
(), x = x
1
+ x
2
, X()

=
X
1
() +X
2
(). Then it is not hard to see from the wealth equation (3.2) that X() = X
x,,C
(),
where the portfolio () is given by
(t)

= [(t)
(1)
(t) + (1 (t))
(2)
(t)]1
(X(t)=0)
, (t) = X
1
(t)/X(t). (7.13)
To show that (, C) /

(x), we have to check that

(t) K
+
on X(t) > 0, and (t) K

on X(t) < 0. (7.14)

Now on X
1
(t) > 0, X
2
(t) = 0 we have (t) =
(1)
(t) K
+
in (7.13); similarly, (t) =

(2)
(t) K
+
on X
1
(t) = 0, X
2
(t) > 0. By analogy, we have (t) K

on X(t) <
0, X
1
(t)X
2
(t) = 0. It remains to see what happens on X
1
(t)X
2
(t) ,= 0. We distinguish
several cases.
(i) X
1
(t) > 0, X
2
(t) > 0: On this event,
(i)
(t) K
+
(i = 1, 2) and 0 < (t) < 1, so
(t) K
+
by the convexity of K
+
.
(ii) X
1
(t) < 0, X
2
(t) < 0: By similar arguments, (t) K

.
(iii) X
1
(t) > 0 > X
2
(t), X(t) > 0: Then
1
(t) K
+
,
2
(t) K

, (t) > 1 and (t) K

+
,
by (7.12).
(iv) X
1
(t) > 0 > X
2
(t), X(t) < 0: Here (t) < 0, and (7.12) gives (t) K

.
(v) X
2
(t) > 0 > X
1
(t), X(t) > 0 and
(vi) X
2
(t) > 0 > X
1
(t), X(t) < 0 can be treated by analogy with (iii), (iv).
30
In all these cases, (7.14) holds. 2
The condition (7.12) is satised in the context of Examples 6.1, for the cases (i), (ii), (iii)
with l k, (iv), (v), (vi) with K

= K
+
, (vii), (viii) with l k. For a discussion of how
things can go wrong in (7.5) if the condition (7.12) fails, see the examples of 8.3.
PROPOSITION 7.2 For any two closed convex sets K
+
, K

that satisfy (7.12 ), we have

K
+
=

K

and ()

() on

K
+
(=

K

); futhermore, if K
+

,= , then () =

() on

K
+
(=

K

).
PROOF. Fix an arbitrary x

K
+
, so that (x) < . For > 1 and arbitrary
+
K
+
,

we have
x

(
+
) +x

((1 )

) = x

(
+
+ (1 )

) inf
K
+
(x

) = (x).
Therefore, taking inma, and recalling the positive homogeneity properties of () and

(),
we get
(x) + ( 1)

(x) (x) > .

It follows that

(x) > (thus

K
+


K

) and in fact (x)

(x).
Now x an arbitrary x

K

, so that

(x) < . For < 0 and arbitrary

+
K
+
,

we have
x

(
+
) +x

((1 )

) = x

(
+
+ (1 )

) sup
K

(x

) =

(x).
Therefore, again by taking suprema, and using the same homotheticity properties, we obtain
(x) (1 )

(x)

(x) < .
It follows that (x) < (whence

K
+


K

) and again (x)

(x).
The inequality

(x) (x) on
d
follows directly from K
+

,= . 2
REMARK 7.2. If the two closed convex sets K
+
, K

satisfy the conditions (7.12) and

K
+

,= , then the endpoints of the arbitrage-free interval [h

low
, h
up
] are characterized
solely in terms of the set K
+
(recall Theorem 6.1 and the notation of section 6).
31
7.3 A representation for convex constraints
The following result will be used to obtain the representation (7.25) for the fair price p(x). It
was estabilished by Davis (1994), but we provide here an alternative arguments, based on our
Denitions 7.3, 7.2 for the fair price.
THEOREM 7.2 Under the Assumption 7.1, the fair price p(x) is uniquely determined by
p(x) =
E
_
U

_
X
x,

,C

(T)
_
B(T)
_
V

(x)
, x > 0. (7.15)
PROOF. We shall use the inequalities
U(x) + (y x)U

(x) U(y) U(x) + (y x)U

(y), 0 < x < y < , (7.16)

which is a simple consequence of concavity. With the notation of (7.2), we have from the second
inequality in (7.16), for x > > 0, p > 0:
W(, p, x) E
_
U
_
X
x

(T) +

p
B(T)
__
E[U(X
x

(T))] +

p
E
_
U

_
X
x

(T) +

p
B(T)
_
B(T)
_
.
Since x X
x

(T) is nondecreasing, we get

W(, p, x) V (x ) +

p
E
_
U

_
X
x

(T) +

p
B(T)
_
B(T)
_
. (7.17)
Thus, from Fatous lemma,
liminf
0
W(, p, x) W(0, p, x)

lim
0
V (x ) V (x)

+
1
p
liminf
0
E[U

(X
x

(T) +

p
B(T)) B(T)]
V

(x) +
1
p
E[U

(X
x

(T)) B(T)].
On the other hand, with < 0, p > 0, we have, from the rst inequality in (7.16), that (7.17) is
valid again (with the interpretation U

(x) U

(0+) for x < 0), and thus by the monotone

convergence theorem,
limsup
0
W(, p, x) W(0, p, x)

lim
0
V (x ) V (x)

+
1
p
lim
0
E[U

(X
x

(T) +

p
B(T)) B(T)]
V

(x) +
1
p
E[U

(X
x

(T)) B(T)].
32
Therefore, for all x > 0 and p > 0,
limsup
0
W(, p, x) W(0, p, x)

(x) +
1
p
E[U

(X
x

(T)) B(T)] (7.18)

liminf
0
W(, p, x) W(0, p, x)

.
Let denote an arbitrary function as in Denition 7.2; then (7.18) yields

(0, p, x) = V

(x) +
1
p
E[U

(X
x

(T)) B(T)],
from which it is easy to check that p(x) dened in (7.15) is the unique weak solution of (7.4) in
the sense of Denition 7.2. 2
To give an explicit form of the fair price for convex constraints, we need a result from
[CK1] along with some additional notation and assumptions. For each T, introduce the
(continuous, strictly decreasing) function

(y)

= E

(T)I(y

(T)Z

(T))], 0 < y < ,

along with its inverse

(x)

=
1

(x), 0 < x < .

Furthermore, let us impose, in addition to the requirements of Denition 7.1, the following
conditions on the utility function U:
U()

= lim
x
U(x) = , (7.19)
U(0) > , or U(x) = log x, (7.20)
x xU

(x) is nondecreasing on (0, ), (7.21)

and
for some (0, 1), (1, ) we have U

(x) U

(x), x (0, ). (7.22)

The following result of [CK1] describes the terminal wealth corresponding to the optimal
pair (

, 0) /
+
(x) for the constrained portfolio optimization problem of (7.1) under the
conditions (7.19)-(7.22), and shows that these guarantee the validity of Assumption 7.1.
THEOREM 7.3 Suppose that the constraint set K
+
is closed, convex, and satises As-
sumptions 6.1 and 6.2; assume also that the conditions (7.19)-(7.22) hold. Then, for every
x > 0, there exists a =
x
T and a pair (

, 0) /
+
(x) with corresponding terminal wealth
X
x,

,0
(T) = I(

(x)

(T)Z

(T)), a.s. (7.23)
33
This pair attains the supremum V (x) of (7.2), i.e., is optimal for the problem of (7.2). The
value function V () is continuously dierentiable, and its derivative can be represented as
V

(x) =

(x) > 0, x > 0. (7.24)
The process () T is optimal in a dual (minimization) stochastic control problem, whence
the adjectives minimal, dual-optimal or least-favorable for it. Now Theorem 7.3 leads
directly to a representation for p(x) in the market with convex constraints.
THEOREM 7.4 We have for all x > 0,
p(x) = E

[

(T)B(T)]. (7.25)
PROOF. We have for any given x > 0,
E[U

(X
x

(T))B(T)] = E[U

(X
x,

,0
(T))B(T)]
= E[U

(I(

(x)

(T)Z

(T)))B(T)] (by (7.23))
= E[

(x)

(T)Z

(T)B(T)] (using U

(I(x)) = x)
= V

(x) E[

(T)Z

(T)B(T)] (by (7.24))
= V

(x) E

[

(T)B(T)].
We can now apply Theorem 7.2, to get (7.25). 2
REMARK 7.3. Combining the representation for p(x) in the above Theorem 7.4, the
representations for h
low
and h
up
in Theorem 6.1, and Proposition 7.2, we recover (7.5): namely,
if the two closed convex sets K
+
, K

satisfy the condition (7.12), then p(x) [h

low
, h
up
] for
all x > 0.
It follows from Theorem 7.4 that p(x) is the Black-Scholes price E

[

(T)B(T)] of B(T) in
an auxiliary unconstrained market /

with interest rate r() +( ()), appreciation rate vector
b()+ ()+( ())1

and volatility matrix (), corresponding to the minimal (dual-optimal)

process () =
x
() of Theorems 7.3, 7.4. Here are some examples from [CK1], in which this
process can be computed explicitly.
34
EXAMPLE 7.1. Logarithmic utility function, general random adapted coecients. If
U(x) = log x, then it is shown in [CK1], p. 790 that () is given by
(t) = argmin
y

K
+
[2(y) +[[(t) +
1
(t)y[[
2
], 0 t T. (7.26)
Thus, () (as well as p) does not depend on the initial wealth x (0, ).
In particular, if K
+
is a cone (thus

() 0 on

K
+
), the expression of (7.26) becomes
(7.26)

(t) = argmin
y

K
+
[[(t) +
1
(t)y[[
2
, 0 t T;
this () also minimizes the relative entropy
H(P[P

)

= E(log
dP
dP

) = E(log Z

(T)) = E[
_
T
0

(t)dW(t) +
1
2
_
T
0
[[

(t)[[
2
dt]
=
1
2
E
_
T
0
[[(t) +
1
(t)(t)[[
2
dt
over T, answering a question of John van der Hoek.
Now consider the special case K
+
=
d
;
i
= 0, i = 1, . . . , m of an incomplete
market as in Example 6.1 (v) for some m = 1, . . . , d 1, d 2. Then (7.26)

becomes
(t) =
_
r(t)1

b(t)
0

n
_
, 0 t T
where

b(t) = (b
1
(t), . . . , b
m
(t))

and n = d m; see Karatzas, Lehoczky, Shreve & Xu (1991),

p. 721 and CK[1], pp. 797-798 (as well as Hofmann et al. (1992), who show that P

, the
least-favorable equivalent martingale measure of Karatzas et al. (1991), coincides in this case
with the minimal equivalent martingale measure in the sense of Follmer & Schweizer (1991)).
2
EXAMPLE 7.2. Deterministic coecients, utility function of power-type. Suppose that
the coecients r(), b(), () of the market / in (2.1), (2.2) are non-random (deterministic)
functions, and that the utility function U() is of the so-called power-type
U

(x)

=
_
x

/; 0 < < 1
log x = lim
0
x

/; = 0
_
, 0 < x < . (7.27)
Then it is shown in [CK1], p.802 that
(t) = argmin
y

K
+
[2(1 )(y) +[[(t) +
1
(t)y[[
2
], 0 t T (7.28)
is again independent of the initial wealth; the same is thus true of p. 2
35
EXAMPLE 7.3. Deterministic coecients, cone constraints. Suppose again that r(),
b(), () are deterministic, and that the constraint set K
+
is a (closed, convex) cone in
d
(as
in Examples 6.1 (i), (ii), (iv)-(vi)), so that

() 0 on

K
+
. Then it is shown in [CK1], p.801
that, under certain mild conditions on the utility function U(), the function
(t) = argmin
y

K
+
[[(t) +
1
(t)y[[
2
, 0 t T (7.29)
is independent , not only of the initial wealth x > 0 , but also of the utility function U(); these
same properties are inherited by p as well. Notice that () of (7.29) minimizes not only the
relative entropy H(P[P

) as in Example 7.1, but also the relative entropy

H(P

[P)

= E

(log
dP

dP
) = E

[
_
T
0

(t)dW(t)
1
2
_
T
0
[[

(t)[[
2
dt]
= E

[
_
T
0

(t)dW

(t) +
1
2
_
T
0
[[

(t)[[
2
dt]
=
1
2
E

[
_
T
0
[[(t) +
1
(t)(t)[[
2
dt]
over T. 2
In any of these Examples 7.1-7.3, and with deterministic market coecients (r(), b(),
()), the process of (7.26), (7.28) or (7.29) is again a non-random (deterministic) function
: [0, T]

K
+
. Suppose, furthermore, that
_

_
B(T) = (P(T)), where P() = (P
1
(), . . . , P
d
())

is the vector
of stock price processes and (p) : (0, )
d
[0, ) a continuous
function that satises polynomial growth conditions
in both [[p[[ and 1/[[p[[.
_

_
(7.30)
Then from (7.25), (7.30), (6.6) and the Feynman-Kac theorem (cf. [KS], p. 366), we see that
the fair price for B(T) is given by
p = e

_
T
0
(r(s)+( (s)))ds
Q(0, P(0)). (7.31)
Here Q(t, p) : [0, T] (0, )
d
[0, ) is the solution of the Cauchy problem for the linear
parabolic equation
Q
t
+
1
2
d

i=1
d

j=1
a
ij
(t)p
i
p
j

2
Q
p
i
q
j
+
d

i=1
(r(t)
i
(t))p
i
Q
p
i
= 0; 0 t < T, (7.32)
subject to the terminal condition
Q(T, p) = (p), p = (p
1
, . . . , p
d
) (0, )
d
, (7.33)
36
where we recall that the matrix a(t) = (a
ij
(t)) = (t)

(t) is as in (2.3). The Cauchy problem

of (7.32), (7.33) has a unique classical solution, subject to mild regularity conditions on the
coecients and on the terminal condition ; see Chapter 1 in Friedman (1964).
REMARK 7.4. In the case of constant coecients (r() = r, b() = b, () = ), the
formulae (7.31)-(7.33) become
p = e
(r+( ))T
Q(0, P(0)), (7.34)
Q(T t, p) =
_
(2t)
d/2
_
R
d
(h(t, p, z))e
||z||
2
/2t
dz ; 0 < t T, p (0, )
d
(p) ; t = 0, p (0, )
d
_
(7.35)
where = argmin
y

K
+
[2(1 )(y) +[[
1
(b r +y)[[
2
] of (7.28) is now a constant vector in

K
+
, and the function h : [0, T] (0, )
d

d
(0, )
d
is given by
h
i
(t, p, y)

= p
i
exp[(r
i

1
2
a
ii
)t +y
i
], i = 1, . . . , d. (7.36)
The Gaussian computation of (7.35) takes a very explicit form in the special case of a
European call option on the rst stock, where (p) = (p
1
q)
+
, 0 < p
1
< , for some exercise
price q > 0 in a market with constant coecients (r, b, ). Then (7.34) becomes
p = e
(
1
+(
1
))T
u
0
(r
1
, q; P
1
(0)), (7.37)
where _
u
0
(r, q; p) is the Black-Scholes price of (4.4), (4.5)
with interest rate r, exercise price q, and P
1
(0) = p
_
. (7.38)
EXAMPLE 7.4. Look-back option B(T) = max
0tT
P
1
(t) with constant coecients,
d = 1 and U() = U

() as in (7.27).
Again = argmin
y

K
+
[2(1 )(y) +[[
1
(b r +y)[[
2
] is constant, and (7.25) becomes
p = P
1
(0)e
(r+( ))T
_

0
f(T, ; )e

d
in the notation of (4.7) with

=
r

2
.
37
8 European call-option in a market with constant coecients
In this section, we use the general results of previous sections to study in detail the three prices
h
low
, h
up
and p for a European call option on the rst stock B(T) = (P
1
(T) q)
+
, in a market
with constant coecients, i.e., when the coecient b() b = (b
1
, b
2
, . . . , b
d
)

, r() r and
() = (
ij
) in (2.2) and (2.1) are all constants. All our examples involve closed, convex
sets K
+
, K

as in Section 6.
8.1 Lower and upper arbitrage prices
EXAMPLE 8.1. Constraints on Borrowing, Example 6.1 (viii) with K
+
= (, k],
K

= [l, ) for some k 1 and l 1.

It is easy to see from (4.6) that the Black-Scholes price u
0
belongs to L, thus
h
low
= u
0
, (8.1)
by Theorem 5.1. On the other hand, we claim that
h
up
E
0
[
0
(T)(
k 1
k
P
1
(T) q)
+
] +
1
k
P
1
(0) =: a
k
. (8.2)
PROOF OF (8.2). By the denition of h
up
it is enough to show that we can nd for
a
k
an admissible pair ( ,

C) /(a
k
), such that () k and X
a
k
, ,

C
() 0, X
a
k
, ,

C
(T)
(P
1
(T) q)
+
almost surely. Actually, we can take

C 0.
Dene for 0 t T,
X
(1)
(t)

=
1

0
(t)
E
0
_

0
(T)
_
k 1
k
P
1
(T) q
_
+

T
t
_
+
1
k
P
1
(t)
=

U
(1)
(T t, P
1
(t)) +
1
k
P
1
(t), 0 t T, (8.3)
where

U
(1)
(t, x)

= E
0
_
e
rt
_
k 1
k
P
1
(t) q
_
+

P
1
(0) = x
_
, 0 t T, 0 < x < . (8.4)
It is clear from (8.3) that
X
(1)
(0) = a
k
, X
(1)
(T) =
_
k 1
k
P
1
(T) q
_
+
+
1
k
P
1
(T) (P
1
(T) q)
+
= B(T).
38
Using the function

U
(1)
(t, x) of (8.4), we can dene

(1)
(t) =

U
(1)
(Tt,P
1
(t))
x
P
1
(t) +
1
k
P
1
(t)

U
(1)
(T t, P
1
(t)) +
1
k
P
1
(t)
.
We shall show that
X
(1)
() = X
a
k
,
(1)
,0
(), and
(1)
() k.
Notice, by the Feynman-Kac formula (cf. [KS], p. 366) and (8.4), that the function

U
(1)
(t, x)
satises the Cauchy problem
_

U
(1)
t
+r

U
(1)
=
1
2

2
x
2
2
U
(1)
x
2
+rx

U
(1)
x

U
(1)
(0, x) = (
k1
k
x q)
+
.
(8.5)
From (8.5), (3.9) in the form dP
1
(t) = rP
1
(t)dt +P
1
(t)dW
0
(t), and Itos rule, we obtain
d

U
(1)
(T t, P
1
(t))
=

U
(1)
t
dt +


U
(1)
x
dP
1
(t) +
1
2

2

U
(1)
x
2
d <P
1
(t)>
=
_
1
2

2
P
2
1
(t)

2

U
(1)
x
2
+rP
1
(t)

U
(1)
x
r

U
(1)
_
dt
+

U
(1)
x

_
rP
1
(t)dt +P
1
(t)dW
0
(t)
_
+
1
2

2

U
(1)
x
2

2
P
2
1
(t)dt
= r

U
(1)
(T t, P
1
(t))dt +


U
(1)
(T t, P
(
t))
x
P
1
(t)dW
0
(t).
Therefore,
dX
(1)
(t) = d

U
(1)
(T t, P
1
(t)) +
1
k
dP
1
(t)
= r

U
(1)
(T t, P
1
(t))dt +


U
(1)
(T t, P
1
(t))
x
P
1
(t)dW
0
(t)
+
1
k
rP
1
(t)dt +
1
k
P
1
(t)dW
0
(t)
= rX
(1)
(t)dt +
_

U
(1)
(T t, P
1
(t))
x
P
1
(t)dW
0
(t) +
1
k
P
1
(t)dW
0
(t)
_
= rX
(1)
(t)dt +X
(1)
(t) (t)dW
0
(t).
Thus X
(1)
() satises the equation (3.2) with C 0, whence the pair ( , 0) is indeed the one
we needed, except we have to verify () k. This can be checked easily from (8.4) and the
inequality
x(

(x) +
1
k
) k((x) +
1
k
x),
39
where (x) = (x(k 1)/k q)
+
. 2
REMARK 8.1. The case k = 1 corresponds to the so-called no-borrowing constraints
and is discussed in [CK2], where it is also shown that h
up
= a
1
= P
1
(0) (for k = 1). In addition,
these authors show that the consumption process C corresponding to the hedging strategy can
be taken as C(t) = 0, for 0 t < T, and C(T) = min(P
1
(T), q) > 0 at time t = T.
REMARK 8.2. If k > 1, then we can rewrite a
k
as
k 1
k
u
0
(r, qk/(k 1); P
1
(0)) +
1
k
P
1
(0).
Furthermore, if k increases (in other words, as the constraint becomes weaker) it is readily seen
that the upper bound a
k
converges to the Black-Scholes price u
0
:
h
up
k
u
0
= u
0
(r, q; P
1
(0)) = Black-Scholes price .
EXAMPLE 8.2. Constraints on short-selling, Example 6.1 (iii) with d = 1, K
+
=
[k, ), K

= (, l], for some k 0 and l > 1.

It is easy to see from (4.6) that u
0
|, whence
h
up
= u
0
. (8.6)
We claim that in this case,
h
low
E
0
[
0
(T)(P
1
(T) q)1
[P
1
(T)ql/(l1)]
] =:
l
. (8.7)
PROOF OF (8.7). Clearly, it is enough to show that
l
L. Dene the process,
X
(2)
(t)

=
1

0
(t)
E
0
[
0
(T)(P
1
(T) q)1
[P
1
(T)ql/(l1)]
[T
t
]
=

U
(2)
(T t, P
1
(t)), 0 t T, (8.8)
where

U
(2)
(t, x)

= E
0
[e
rt
(P
1
(t) q)1
[P
1
(t)ql/(l1)]
[P
1
(0) = x], 0 t T, 0 < x < . (8.9)
Then we have at time t = 0,
X
(2)
(0) = E
0
[
0
(T)(P
1
(T) q)1
[P
1
(T)lq/(l1)]
] =
l
< 0,
40
and at time t = T,
X
(2)
(T) = (P
1
(T) q)1
[P
1
(T)ql/(l1)]
(P
1
(T) q)
+
.
On the other hand, (8.9) gives 0

U
(2)
(t, x) E
0
[e
rt
P
1
(t)[P
1
(0)=x] = x, so that, from (8.8),
the positive process X
(2)
() is dominated by the P-integrable random variable max
0tT
P
1
(t).
Hence, it is enough to nd a pair (
(2)
, C
(2)
) /(
l
) with X
(2)
() = X

l
,
(2)
,C
(2)
(). Introduce

(2)
(t) =

U
(2)
(Tt,P
1
(t))
x
P
1
(t)

U
(2)
(T t, P
1
(t))
, 0 t T.
Again by the Feynman-Kac formula, the function

U
(2)
(t, x) of (8.9) satises the Cauchy problem
_

U
(2)
t
+r

U
(2)
=
1
2

2
x
2
2
U
(2)
x
2
+rx

U
(2)
x

U
(2)
(0, x) = (x q)1
[xql/(l1)]
,
and from Itos rule,
dX
(2)
(t) =
_

U
(2)
t
dt +


U
(2)
x
dP
1
(t) +
1
2

2

U
(2)
x
2
d <P
1
(t)>
_
=
_
1
2

2
P
2
1
(t)

2

U
(2)
x
2
+rP
1
(t)

U
(2)
x
r

U
(2)
_
dt

U
(2)
x
_
rP
1
(t)dt +P
1
(t)dW
0
(t)
_

1
2


2

U
(2)
x
2

2
P
2
1
(t)dt
= r

U
(2)
(T t, P
1
(t))dt


U
(2)
x
P
1
(t)dW
0
(t)
= rX
(2)
(t)dt +X
(2)
(t)
(2)
(t)dW
0
(t).
Hence the wealth equation (3.2) is satised with C = C
(2)
0. To check that
(2)
() l, we
need only verify that

U
(2)
(Tt,P
1
(t))
x
P
1
(t)

U
(2)
(T t, P
1
(t))
l.
This bound is not hard to derive, from (8.9) and the inequality

(x) x l(x), where (x) = (x q)1

[xql/(l1)]
.
The proof is now complete. 2
REMARK 8.3. Notice that we have from (2.2), P
1
(t) = e
(r
2
/2)t+N(t)
p, where
p = P
1
(0) > 0 and N() is standard Brownian motion under the probability measure P
0
.
41
Therefore,
l
can be rewritten as

l
= u
0
_
r, ql/(l 1); P
1
(0)
_
+E
0
_

0
(T)(P
1
(T) ql/(l 1))1
[P
1
(T)ql/(l1)]
_
= u
0
_
r, ql/(l 1); P
1
(0)
_
+q
0
(T)/(l 1) P
0
(P
1
(T) ql/(l 1))
= u
0
_
r, ql/(l 1); P
1
(0)
_
+
qe
rT
l 1
P
0
_
(r
2
/2)T +N(T) log(ql/(P
1
(0)(l 1)))
_
,
in the notation of (7.38). Invoking the normal distribution, we arrive after a bit of algebra at

l
= u
0
_
r, lq/(l 1); P
1
(0)
_
+
qe
rT
l 1
_
1
_
1

T
log
_
ql/(P
1
(0)(l 1)) (r
2
/2)

T
___
.
As before, if the constraints become weaker and weaker (i.e., l ), then
l
converges to the
Black-Scholes price u
0
:
h
low
u
0
, as l .
REMARK 8.4. If we consider the no short-selling case of Example 6.1 (ii) (or equivalently,
Example 8.2 with k = l = 0) then instead of the inequality (8.7), we can actually prove
h
low
= 0.
Indeed, we have from (6.6) that
e
_
t
0

1
(s)ds

0
(t)P
1
(t) = P
1
(0) exp[
_
t
0
(s)dW

(s)
1
2
_
t
0
(s)
2
ds],
which is a P

-martingale. Thus, if we denote

T
d
the subset of all non-random functions
: [0, T]

K

in the set

T, we have
E

(T)P
1
(T)] = P
1
(0)e

_
T
0
(

((s))+(s))ds
,

T
d
. (8.10)
By Theorem 6.1, we get the inequalities,
0 h
low
= inf

D
E

(T)(P
1
(T) q)
+
]
inf

D
E

(T)P
1
(T)]
P
1
(0) inf

D
d
e

_
T
0
(

((s))+(s))ds
= P
1
(0) inf

D
d
e

_
T
0
(s)ds
= 0,
as we can let tend to . Thus we conclude that h
low
= 0 in the no short-selling case.
42
EXAMPLE 8.3. Constraints on Borrowing, Example 6.1 (viii) with d = 1 and K
+
=
(, k], K

= [l, ) for some k 1, l k, l > 1.

Here again the upper bound h
up
a
k
on the upper arbitrage price holds, as in (8.2) of
Example 8.1. Now, however, h
low
= 0, so that the complete picture is
0 = h
low
< u
0
< h
up
a
k
< .
Indeed, we have here

K

= (, 0] and (x) = kx,

(x) = lx on

K

, so that for
deterministic (),
E

(T)P
1
(T)] = P
1
(0)e

_
T
0
(

((s))+(s))ds
= P
1
(0)e
(l1)
_
T
0
(s)ds
as in (8.10), and we obtain h
low
= 0 much like in Remark 8.4, except now letting () tend to
.
EXAMPLE 8.4. Constraints on short-selling, Example 6.1 (iii) with d = 1 and K
+
=
[k, ), K

= (, k] for some k 0. In this case,

0 = h
low
< u
0
= h
up
< .
Indeed, h
up
= u
0
follows as in (8.6) of Example 8.2. As for h
low
= 0, observe that we have
now

K

= [0, ), (x) =

(x) = kx on

K

, and (8.10) become

E

(T)P
1
(T)] = P
1
(0)e
(1+k)
_
T
0
(s)ds
for deterministic (); we conclude h
low
= 0, by letting () becomes very large as in Remark
8.4.
EXAMPLE 8.5. Incomplete market cases.
(a). Only the rst m stocks can be traded, with 1 m d 1, d 2 as in Example 6.1
(iv). Then by the explicit formula in (4.6), we have that h
up
= h
low
= u
0
.
(b). The rst m stocks cannot be traded, 1 m d1 as in Example 6.1.(v). In this case,
it can be shown that h
up
= , as in [CK2]. We can show that h
low
= 0. In fact, observe, once
again from (6.6), that
e
_
t
0

1
(s)ds

0
(t)P
1
(t) = P
1
(0) exp[
_
t
0

1
(s)dW

(s)
1
2
_
t
0
[[
1
(s)[[
2
ds],
where
1
= (
11
,
12
, . . . ,
1d
)

. Then the same argument as for the no short-selling case will

lead to the desired result.
43
8.2 Computation of the fair price
Let us compute in this subsection the fair price p of (7.25) in a few examples, with closed and
convex sets K

that satisfy the condition (7.12)so that p is in the interval [h

low
, h
up
] for all
these examples.
EXAMPLE 8.6. Cone constraints. Let K
+
be a (closed, convex) cone in
d
, and K

=
K
+
. Then from (7.29), (7.37) and the fact that 0 on

K
+
, we have
p = e

1
T
u
0
(r
1
, q; P
1
(0)) (8.11)
in the notation of (7.38), where
= argmin
y

K
+
[[
1
(b r +y)[[
2
. (8.12)
In particular, p does not depend on either the utility function or the initial level of wealth.
Here are some particular cases.
(a) Incomplete market with only the rst m stocks available, Example 6.1 (iv). Then (8.12)
gives
1
= 0, and (8.11) becomes
p = u
0
(r, q; P
1
(0)) = Black-Scholes price.
(b) Incomplete market with the rst m stocks unavailable, Example 6.1 (v). We again have
from (8.12) that
1
= r b
1
(see also Example 7.1), and (8.11) takes the form:
p = e
(rb
1
)T
u
0
(b
1
, q; P
1
(0)).
(c)Prohibition of short-selling, Example 6.1 (ii) with d = 1. Then it can be seen by simple
algebra that in this case = (r b)
+
in (8.12), thus (8.11) becomes
p =
_
u
0
(r, q; P
1
(0)) ; if r b
e
(rb)T
u
0
(b, q; P
1
(0)) ; if r > b
_
. (8.13)
EXAMPLE 8.7. Utility function of the power type (7.27). In this case, (7.28) or (7.26)
give
= argmin
y

K
+
[[[
1
(b +r y)[[
2
+ 2(1 )(y)] (8.14)
and p is then as in (7.37); for a set K
+
that is not a cone, this p depends in general on the
utility function through the constant [0, 1). Here are some concrete examples.
44
(a) Prohibition of borrowing, Example 6.1 (vii) with d = 1. Then (8.14) gives = (r b +
(1 )
2
)

, and thus (7.37) becomes

p =
_
u
0
((b + ( 1)
2
), q; P
1
(0)) ; if r b + ( 1)
2
u
0
(r, q; P
1
(0)) ; otherwise
_
.
(b) Constraints on borrowing, Example 8.3. Then (x) = kx on

K
+
= (, 0], for some
k 1, so (8.14) and (7.37) give = (r b + (1 )
2
)

and
p =
_
u
0
(r, q; P
1
(0)) ; if b +k( 1)
2
r
e
(k1)(b+k(1)
2
r)
u
0
_
b +k( 1)
2
, q; P
1
(0)
_
; otherwise
_
. (8.15)
(c) Constraints on short-selling, Example 8.4. Then (x) = kx on

K
+
= [0, ) for some
k 0, and (8.14), (7.37) lead respectively to = (r b +k( 1)
2
)
+
and
p =
_
u
0
(r, q; P
1
(0)) ; if r b +k(1 )
2
e
(1+k)(rb+k(1)
2
)
u
0
_
b +k(1 )
2
, q; P
1
(0)
_
; otherwise
_
. (8.16)
8.3 Counterexamples
Finally, let us demonstrate by some examples that the lower bound of (7.5) may fail, in the
absence of condition (7.12) on the sets K

. In all these examples the set K

+
is convex, so the
upper bound of (7.5) must hold; see Remark 7.1.
EXAMPLE 8.1 (contd) with K
+
= (, k], K

= [l, ) and k > 1, l 1. Here it is

easy to check that condition (7.12) fails; and with utility function U

(), 0 < 1 as in (7.27),

the fair price p is given by (8.15) and satises
p 0 as b ,
for xed (r, k, , q,
2
, l), since u
0
(x, q; P
1
(0)) P
1
(0) as x (See Cox & Rubinstein (1984),
p.216). However, we know from (8.1) that h
low
u
0
(r, q; P
1
(0)) > 0, whence h
low
> p > 0 for
all suciently large appreciation rates b.
EXAMPLE 8.2 (contd). Here K
+
= [k, ), K

= (, l] for some k 0, l > 1.

Again, it is veried that condition (7.12) fails; and with utility function of the type (7.27), the
fair price p is given by (8.16) and satises
p 0, as r
45
for (b, k, , q,
2
) xed. On the other hand, we have from Remark 8.3:
h
low

l
= u
0
(r, lq/(l 1); p) +
qe
rT
l 1

_
1
_
1

T
log
_
lq/(p(l 1)) (r
2
/2)

T
___
r
p P
1
(0) > 0.
Consequently, for all suciently large interest rates r, h
low
> p > 0.
EXAMPLE 8.8. Take d = 1, r > b, K
+
= [0, ), K

= [1, ) (a combination of
Examples 6.1 (ii), (vii)), so that (7.12) fails again. Now

K
+
= [0, ) and

K

= (, 0],
h
low
= u
0
(r, q; P
1
(0)) as in (8.1), and from (8.13):
p = e
(rb)T
u
0
(b, q; P
1
(0)) < u
0
(r, q; P
1
(0)) = h
low
.
9 Market with higher interest rate for borrowing
We have studied so far the pricing problem for contingent claims in a nancial market with
the same interest rate for borrowing and for saving. However, the techniques developed in the
previous sections can be adapted to a market /

with interest rate R() for borrowing higher

than the bond rate r() (saving rate).
We consider in this section an unconstrained market /

with two dierent (bounded, T

t
-
progressively measurable) interest rate processes R() r() for borrowing and saving, re-
spectively. In this market /

, it is not reasonable to borrow money and to invest money in

the bond, at the same time. Therefore, the relative amount borrowed at time t is equal to
_
1

d
i=1

i
(t)
_

. As shown in [CK1], the wealth process X() = X

x,,C
() corresponding to
initial wealth x and a portfolio/consumption pair (, C) as in Denition 3.5, satises now the
analogue of the wealth equation (3.2)
dX(t) = r(t)X(t)dt dC(t) +X(t)
_

(t)(t)dW
0
(t) (R(t) r(t))
_
1
d

i=1

i
(t)
_

dt
_
, (9.1)
whence
N(t)

=
0
(t)X(t) +
_
t
0

0
(t)dC(t) +
_
t
0

0
(t)X(t)[R(t) r(t)]
_
1
d

i=1

i
(t)
_

dt, 0 t T
is a P
0
-local martingale by Itos rule, in the notation of (2.4)-(2.7) and (3.6).
46
All the arguments in Section 5 go through under slight modications. For example, the
lower and upper hedging classes are now dened to be
L

= x 0 : ( ,

C) /(x), such that X
x, ,

C
() 0 and X
x, ,

C
(T) B(T), almost surely,
|

= x 0 : ( ,

C) /(x), such that X
x, ,

C
() 0 and X
x, ,

C
(T) B(T), almost surely.
The statements of Denition 5.2, Theorem 5.1, 5.2 and 5.3 hold without change.
We set ((t)) =
1
(t), 0 t T, for T, where T is the class of progressively
measurable processes : [0, T]
d
with r R
1
= =
d
0, l P a.e. Then
with this notation the theory of Section 6 also goes through with only minor changes, such as
replacing

by and

T by T, etc. In particular, Theorem 6.1 now states that
h
low
= inf
D
E

(T)B(T)] (9.2)
and that
h
up
= sup
D
E

(T)B(T)]. (9.3)
The proofs of (9.2) and (9.3) follow the same lines of Theorem 6.1 and [CK2], respectively. We
sketch the proof of (9.2) here.
SKETCH OF PROOF FOR (9.2). We rst repeat the proof of Theorem 6.1, right up to
(6.23). There, we change the denition of the consumption process

C(), to read

C(t)

=
_
t
0

(s)dA

(s)
_
t
0

X(s)
_
((s)) +

(s)(s) + (R(s) r(s))

_
1
d

i=1

i
(s)
_

_
ds.
Taking (t) = (t)
1
(t)1

, where
1
(t)

= [r(t) R(t)]1
(

d
i=1

i
(t)>1)
, we get

C(t) =
_
t
0

(s)dA

(s)
as required. Skip the lines in which (t) K

is shown, and observe that we have now

d(

X(t)

(t)) = dQ

(t) =

(t)dW

(t) +dA

(t)
=

(t)
_
d

C(t)

X(t)
_
((t))+(R(t)r(t))
_
1
d

i=1

i
(t)
_

(t)(t)
_
dt+

X(t)

(t)(t)dW

(t)
_
=

(t)[d

C(t)

X(t)
,
(t)dt +

X(t)

(t)(t)dW

(t)],
47
where

,
(t)

= [R(t) r(t) +
1
(t)]
_
1
d

i=1

i
(t)
_

1
(t)
_
1
d

i=1

i
(t)
_
+
, 0 t T
is a nonnegative process. Now taking 0, we get

X() = X
g, ,

C
() by comparing with the
new wealth equation (9.1), where g is now the right-hand side of (9.2). The rest of the proof
proceeds in a clearly analogous way. 2
With the adoption of the new () and T, we can dene the fair price by analogy with (7.4),
and proceed in the same way as we did before, to obtain all theorems in Section 7. In particular,
an encouraging phenomenon is that the fair price always lies within the arbitrage interval, as
the Assumption 7.2 is always satised in this case.
The argument in [CK2] for computing h
up
in a market /

with d =1 and constant coef-

cients also works for h
low
, after slight adjustments. For example, change sup and max
in (9.8) and (9.9) of [CK2] to inf and min respectively; then from the Hamilton-Jacobi-
Bellman (HJB) equation we can also get
h
low
= u
0
(r, q; P
1
(0))
for the European call option B(T) = (P
1
(T) q)
+
. In other words, the lower arbitrage price
is exactly the Black-Scholes price with interest rate r, while, as it has be shown in [CK2], the
upper arbitrage price is the Black-Scholes price with interest rate R:
h
up
= u
0
(R, q; P
1
(0)).
For the fair price within the interval [h
low
, h
up
], we still use Theorem 7.4 and Remark 7.1 to
get the explicit fair price p for the constant coecient market /

. More precisely, with utility

function U

() as in (7.27), it is shown in [CK1], p.816 that the in Theorem 7.4 is given by

=
_

_
0 ; if r b
1
+
2
( 1)
r b
1

2
( 1) ; if r b
1
+
2
( 1) R
r R ; if b
1
+
2
( 1) R.
_

_
.
Hence, (7.37) gives the fair price
p =
_

_
u
0
(r, q; P
1
(0)) : if r b
1
+
2
( 1)
u
0
(b
1
+
2
( 1), q; P
1
(0)) ; if r b
1
+
2
( 1) R
u(R, q; P
1
(0)) ; otherwise
_

_
. (9.4)
48
REMARK 9.1. The expression (9.4) coincides with the so-called minimax price in
Barron & Jensen (1990), dened to be the number p = p(x) for which the function
W(, p(x), x) of (7.3) is minimized at = 0. (Clearly, with p(x) as in our Denition 7.3,
W(, p(x), x) is minimized at = 0, so we can take p(x) = p(x), justifying this coincidence.)
REMARK 9.2. More generally, with d 1, utility function U

() of the type (7.27) and

deterministic coecients (resp., =0 in (7.27) and general random coecients), the function
(resp., process) (t) =
1
(t)1

is given as

1
(t) = argmin
r(t)R(t)y0
_
[[
1
(t)(b(t) r(t) +y1

[[
2
2y
_
=
_

_
0 ;

(t) 0
r(t) R(t) ;

(t) R(t) r(t)

(t) ; 0

(t) R(t) r(t)

_

_
by analogy with (7.26) and (7.28), where

(t) = ( 1 +

(t)
1
(t)1

)/tr[(
1
(t))

(
1
(t))].
In the special case B(T) = (P(T)) of (7.30) with deterministic coecients, the computations
of (7.31)-(7.36) for the fair price p are all still valid.
10 A table
The results of previous discussions and examples, concerning the pricing of a European call
option B(T) = (P
1
(T) q)
+
in a market with constant coecients, can be summarized on a
table as follows.
49
h
low
h
up
p
Unconstrained market u
0
(r) u
0
(r) u
0
(r)
Incomplete market, with
rst m stocks available u
0
(r) u
0
(r) u
0
(r) *
Incomplete market,
rst m stocks unavailable 0 e
(rb
1
)T
u
0
(b
1
) *
No short-selling of stocks
(K
+
= [0, ), K

= (, 0])
0 u
0
(r)
_
u
0
(r); if r b
1
e
(rb
1
)T
u
0
(b
1
) ; if r > b
1
_

No borrowing
(K
+
= (, 1], K

= [1, ))
u
0
(r) P
1
(0)
_
u
0
(r); if r f
u
0
(f); otherwise
_

Constraints on short-selling
(K
+
= [k, ),
K

= (, k], k 0)
0 u
0
(r)
_
u
0
(r) ; if r f
c
k
; otherwise
_

Constraints on borrowing
(K
+
= (, k],
K

= [l, ), l k > 1)
0 a
k
_
u
0
(r) ; if r b
1
+k( 1)
2
d
k
; otherwise
_

Constraints on short-selling
(K
+
= [k, ),
K

= (, l], k 0, l > 1)

l
u
0
(r) not appropriate ( p < h
low
)
Constraints on borrowing
(K
+
= (, k],
K

= [l, ), k > 1, l 1)
u
0
(r) a
k
not appropriate ( p < h
low
)
Market with higher
interest rate R > r
for borrowing
u
0
(r) u
0
(R)
_

_
u
0
(r) ; if r f
u
0
(f) ; if r f R
u
0
(R) ; if f R
_

() For arbitrary utility function.

() For utility function U

() of the form (7.27) with 0 < 1.

In the above table, r is the interest rate of the bond (savings account); b
1
is the appreciation
rate of the rst stock, on which the option is written;
2
is the stock volatility; u
0
(x)
u
0
(x, q; P
1
(0)) is the Black-Scholes price for interest rate x and exercise price q; and P
1
(0) is
the price for the rst stock at time t = 0. Finally,
a
k
=
k 1
k
u
0
_
r, qk/(k 1); P
1
(0)
_
+
1
k
P
1
(0)
k
u
0
(r),

l
= u
0
_
r, lq/(l 1); P
1
(0)
_
+
qe
rT
l 1

_
1
_
1

T
log(ql/(P
1
(0)(l 1)) (r
2
/2)

T
___
l
u
0
(r)
c
k
= e
(1+k)(rb
1
+(1)k
2
)
u
o
(b
1
+ (1 )k
2
)
d
k
= e
(k1)(b
1
+(1)k
2
r)
u
0
(b
1
+ ( 1)k
2
)
50
f = b
1
+ ( 1)
2
.
REMARK 10.1. It should be observed that all the exact values, as well as the bounds, for
h
low
and h
up
are independent of the appreciation rate b of the stock, which is often dicult to
estimate. This makes the lower and upper arbitrage prices relatively easy to use. In contrast,
a main drawback of the fair price is that it does depend on b. Heuristically, it may well be that
hedging, as it is based on the arbitrage arguments, is a sort of global property. On the other
hand, the Denition 7.3 of the fair price looks like a local property, as it involves a derivative;
this makes the fair price p more likely to depend on the local drift b (appreciation rate) of
the price process.
11 Discussion
1. With a little additional care, the method also works for the European option with dividend
rate g(t). For example, the analogue for Theorem 6.1 will be
h
low
= inf

D
E

(T)B(T) +
_
T
0

(s)g(s)ds]
and
h
up
= sup
D
E

(T)B(T) +
_
T
0

(s)g(s)ds].
2. A similar lower price h
low
(for buyers, as opposed to the h
up
which refers to sellers)
was mentioned by El Karoui & Quenez (1992) in the incomplete market case, but without
justication based on considerations of arbitrage.
3. Suppose we want to consider constraints on the number of shares , or on the total amount
of money invested in every asset, instead of on the vector , of the proportions of wealth
invested in assets. Then the general arbitrage arguments in Section 5 still hold. However,
we no longer have an easy way to get all the representations of Section 6. For instance, the
nice equation (6.24) is changed, as the very helpful term ((s))+

(s)(s) disappears.
4. For practical purposes, one may recommend the use of the Black-Scholes price u
0
as a
rough-and-ready unique price, for a constrained market with the same interest rate for
borrowing and saving, when the fair price p is dicult to compute. The reasons are:
51
(a) The Black-Scholes price u
0
always lies within the arbitrage-free interval [h
low
, h
up
];
(b) As we saw in the case of constraints on borrowing and short-selling, the arbitrage-free
interval will shrink to u
0
as the constraints become weaker and weaker;
(c) The Black-Scholes price u
0
does not involve the stock appreciation rate b;
(d) Many numerical procedures, including software, have been developed to calculate
u
0
.
Acknowledgements: We wish to thank Professor Jaksa Cvitanic for his careful reading of
the manuscript and his helpful comments, and Professors N. El Karoui and Mark Davis for their
encouragement. We are indebted to Dr. Frank Oertel for a suggestion that led to our Example
7.4, to Prof. Steven Shreve for prompting us to tighten the argument in Theorem 7.2, and to
Dr. John van der Hoek for asking us to consider the connection of the fair price with entropy
minimization. Thanks also due to the anonymous referee for his very conscientious reading of
the paper, and his many corrections and suggestions.
References
[1] Barron, E., and Jensen, R. (1990). A stochastic control approach to the pricing of options.
Math. Oper. Res. 15 49-79.
[2] Black, F. and Scholes, M. (1973). The pricing of options and corporate liabilities. J. Polit.
Econ. 81 637-659.
[3] Cox, J. and Ross, S. (1976). The valuation of options for alternative stochastic processes.
J. Financial Econ. 3 145-166.
[4] Cox, J. and Rubinstein, M. (1984). Option Markets. Prentice-Hall, Englewood Clis, N.J.
[5] Crandall, M.G. and P.-L. Lions (1983). Viscosity Solutions of Hamilton-Jacobi equations.
Trans. A.M.S. 277 1-42.
[6] [CK1] Cvitanic, J. and Karatzas, I. (1992). Convex duality in constrained portfolio opti-
mization. Ann. Appl. Probab. 2 767-818.
[7] [CK2] Cvitanic, J. and Karatzas, I. (1993). Hedging contingent claims with constrained
portfolios. Ann. Appl. Probab. 3 652-681.
52
[8] Davis, M. (1994). A general option pricing formula. Preprint, Imperial College, London.
[9] Due, D. (1992). Dynamic Assets Pricing Theory. Princeton Univ. Press, N.J.
[10] Due, D. and Skiadas, C. (1991). Continuous-time security pricing: a utility gradient
approach. Preprint, Stanford University.
[11] El Karoui, N. and Quenez, M.C. (1995). Dynamic programming and the pricing of contin-
gent claims in an incomplete market. SIAM J. Control and Optim. 33 29-66.
[12] Fleming, W.H. and Soner, H.M. (1993). Controlled Markov Processes and Viscosity Solu-
tions. Springer-Verlag, New York.
[13] Foldes, L.P. (1990). Conditions for optimality in the innite-horizon portfolio-cum-savings
problem with semimartingale investments. Stochastics and Stochastics Reports 29 133-171.
[14] Follmer, H. and Sondermann, D. (1986). Hedging of nonredundant contingent claims.
Contributions to Math. Econ. (Hildenbrand, W. and MasCollel, A., eds.) 205-223.
[15] Follmer, H. and Schweizer, M. (1991). Hedging of contingent claims under incomplete
information. Applied Stochastic Analysis (Davis, M. and Elliott, R., eds.) Gordon and
Breach, New York & London, 389-414.
[16] Friedman, A, (1964) Partial Dierential Equations of Parabolic Type. Prentice Hall, En-
glewood Clis, N.J.
[17] Goldman, M.B., Sosin, H. B. and Gatto, M. A. (1979). Path-dependent options.
J. of Finance 34 1111-1127.
[18] Harrison, J.M. and Kreps, D.M. (1979). Martingales and arbitrage in multi-period security
markets. J. Econom. Theory 20 381-408.
[19] Harrison, M. and Pliska, S.R. (1981). Martingales and stochastic integrals in the theory of
continuous trading, Stochastic Process. Appl. 11 215-260.
[20] Hodges, S.D. and Neuberger, A. (1989). Optimal replication of contingent claims under
transaction costs. Rev. Futures Markets 8 222-239.
[21] Hofmann, N., Platen, E. and Schweizer, M. (1992). Option pricing under incompleteness
and stochastic volatility. Mathematical Finance 2 153-187.
53
[22] Karatzas, I. (1989). Optimization problems in the theory of continuous trading. SIAM J.
Control Optim. 27 1221-1259.
[23] Karatzas, I., Lehoczky, J.P., Shreve, S.E. and Xu, G.L. (1991). Martingale and duality
methods for utility maximization in an incomplete market. SIAM J. Control and Opti-
mization 29 702-730.
[24] [KS] Karatzas, I. and Shreve, S. (1991). Brownian Motion and Stochastic Calculus. 2nd
printing, Springer-Verlag, New York.
[25] Karatzas, I. and Shreve, S. (1996). Monograph on Mathematical Finance. In preparation.
[26] Meyer, P.A. (1976). Un cours sur les integrales stochastiques. Lecture Notes in Mathematics
511 245-398. Springer-Verlag, Berlin.
[27] Myneni, R. (1992). American Option. Ann. Appl. Probab. 2 1-23.
[28] Rockafellar, R.T. (1970). Convex Analysis. Princeton Univ. Press, New Jersey.
[29] Shreve, S.E. and Soner, H.M. (1994). Optimal Investment and Consumption with Trans-
action Costs. Ann. Appl. Probab. 4 609-693.
54