Journal of Financial Economics 10 (1982) 331-345.

North-Holland Publishing Company

COMMENTS ON THE VALUATION OF DERIVATIVE ASSETS

Avi BICK*
University of California, Berkeley, CA 94720, USA

Received June 1981, final version received April 1982

This paper presents an alternative approach to derive the Breeden-Litzenberger valuation

formula, which expresses the price of an arbitrary derivative security in terms of call options
prices. This valuation formula follows from the observation that a continuous derivative security
can be replicated by a portfolio including a bond and call options with all possible exercise
prices. Discrete terms are added to the original Breeden-Litzenberger formtila to reflect possible
discontinuities of the call option prices derivative with respect to the exercise price. These
discontinuities are subsequently shown to correspond to mass points of the probability
distribution of the stock price. Several applications of the Breeden-Litzenberger valuation
formula are demonstrated.

1. Introduction and summary

Breeden and Litzenberger (1978), henceforth BL, show that c,,rPc/ax,
where c(7;x) is the price of a European call option on a given underlying
security with exercise price x and maturity date 7; is a pricing density
function for the valuation of general derivative assets on the same underlying
security. Thus, the current price of a derivative security that pays q(M,) at
time 7: where M, is the value of the underlying security at time IT; and zero
at any other time, will be

W = 7qW,,(IT: y)dy. (1)

0

In the case of a derivative cash stream over time with density q(T, M) (i.e.,
q(7;M,)dT is paid in the time interval [7:T+dT]), (1) is generalized
(denoting the present by T=O) to

Wd=y
3d7;y)c,,(Ty)dyd7:
00
(1)
*I wish to thank Meir Schneller for introducing me to the topic and Victor Konrad and Mark
Rubinstein for helpful comments on a previous version of this paper. Valuable suggestions by
the referee, Douglas Breeden, have helped improve this paper. I remain solely responsible for all
errors. Partial support from the Dean Witter Grant to the School of Business, University of
California, Berkeley, is gratefully acknowledged.

0304405x/82/OOOMOOO /$02.75 0 1982 North-Holland

J.F.E.- D
332 A. Bick, Valuation o/derivative assets

In section 2,an alternative proof for the valuation formula (1) is presented.
The reason for this is twofold: first, while BL implicitly assume the existence
of c*, at each point, this paper will generalize the formula for the case in
which &/ax may not exist in some points. (The economic interpretation of
this case will be discussed in section 3.) For example, (1) is obviously not true
if c is piecewise linear in the exercise price [see, e.g., Cox, Ross and
Rubinstein (1979, eq. (6)) or Cox and Ross (1976, eq. (29)) for such cases].
Even if J2c/ax2 is continuous, (1) implies that the current price of a bond
which pays one dollar at time T is - &/iix [(T,O+,, which is not necessarily
true (see section 3).
Second, while BLs proof is intuitively appealing, it is not complete from a
formal point of view. They implicitly use an approximation theorem which
states that the given continuous-state economy (in which Mr. can obtain
any non-negative value) can be approximated by a discrete-state economy
(in which M, can obtain only integer non-negative multiples of some AM) in
the sense that price systems will converge to the price system in the given
economy as AM-O. While it is plausible that such a theorem can be proved
under certain conditions2 this problem can be circumvented altogether by
remaining in the given economy and approximating on portfolios. The
method employed here is to construct the derivative security explicitly from
the call options, thus generalizing the known piecewise-linear case [see
Garman (1976), Cox and Rubinstein (1982)], and then the implied valuation
formula will turn out to be mathematically equivalent to the generalized BLs
formula.
Section 3 deals with the relation between investors beliefs and the
structure of call options prices (as a function of the exercise price). In section
4, the valuation formula will be written for the BlackkScholes case, and it
will be shown that this is consistent with the continuous-trading-strategy
approach. The usefulness of this formula will be demonstrated in sections 5
and 6, where it will be employed to price compound options and call options
on a stock with constant dividend yield, respectively.

2. Derivation of the pricing formula

Consider a future time T which will be held fixed throughout this section
(and hence omitted from the notation). The following securities, among other
securities, are assumed to be traded in a perfect financial market:

They assume (p. 622) that the underlying securitys value at time T has a continuous
probability distribution, but they do not establish the relation between these two properties.
For example, Cox, Ross and Rubinstein (1979) apply the central limit theorem in a special
case.
 A. Bick, Valunlion of derivativemsds 333

A given underlying stock whose price at time T is denoted M (a

random variable).
(ii) For each x10, a European call oDtion4 on the above stock with
maturity at T and exercise price x. (I; subsequent sections we shall call
it the (7: x)-call option.) Its price c(x) at t =0 is assumed to be
continuous in x.
(iii) A bond which yields one dollar at time 7Y Its price at t =0 will be
denoted B = B,.
The market is assumed to be open at times t =0 and t = 7; but no
assumptions are made regarding whether or not there is trade in between or
regarding the dividend pattern of the underlying stock.

Proposition 1. Let q:[O, co)+R be a function with a continuous second

derivative except in a finite set of points in which the left-hand and right-hand
derivatives q_, q+, ql , q; exist and are finite. Then:

(a) The cost of the derivative asset, which pays q(M) ut time T and zero at
any other time, is given by6

Wd=qW - c
=D(q+)
(q+(a)-qO)c+(a)-~~~q_)(q(a)-q-(a))c~(a)

+ + C (q+(a)- q-(a))44 + qWfy)dy~

q+(OMO)
at D(q)
$ (2)

where D(q) is the set of discontinuity points of q, D(q+) is the set of points a,
in which

q +(4 = lim 4(x)# q(a),

X-O+

and D(q_) is defined similarly.

(b) Assume that c has a continuous second derivative with respect to the
exercise price except in a finite set of points in which the left- and right-hand

31n BLs derivation the underlying asset is a portfolio, but here the analysis is restricted to the
case of a stock. In the case of a portfolio that can obtain only non-negative values (as is the case
for the market portfolio, which is used by BL) the analysis will be identical.
41n the sequel, call option will always mean European.
If this set is allowed to be discrete (possibly infinite). the analysis will be similar with the
addition of some further technical points. A similar remark would apply to D(c) in part (b).
Here y-(u) means lim,_,_q(x), etc.
6This derivative asset need not be traded in the market and its cost V(q) (which is not
necessarily linite) is to be understood as the limit of the costs of the approximating portfolios (in
a sense to be explained below). Of course, if it is traded, then by arbitrage considerations Y(q)
must be its current market price.
334 A. Bick, Valuation of derivative assets

derivatives c-, I?+, CY and c; exist and are jnite. Then,

1 (0)+ B)+aE&r,J
%I) = qWWy + q(OW+ qbW+(4 -c- (a)), (3)

where D(c) is the set of discontinuity points of c.

Proof (a) Suppose first that q is continuous. Consider a portfolio of all the
call options with weight function q. The portfolios cost will be

$dWcCy)dy~
and its payout at time T, given a realization of M, will be

It can be viewed as a limit of a sequence of portfolios, each one with a tinite

number of options, which corresponds to the Riemanns sums in the
definition of the integral. Suppose, for the sake of simplicity, that there is
only one positive point a in which q does not necessarily have a continuous
second derivative. Integration by parts gives
M
d q(y)(M-y)dy=q(M)-q(O)-q+(O)M if Msa,

and
M
/ q(y)(M-y)dy=/q(y)(M-y)dy+j(q(y)(M-y)dy
(I
=q(M)-q(O)-q+(O)M-(q+(a)-q_(a))(M-a) if a<M.

Thus q(M) can be produced by holding q(0) bonds, q+(O) call options with
zero exercise price (or q+(O) shares of the underlying stock, if it is certain that
the stock does not pay dividends in the time interval [0,7j), q+(a)-q-(a)
call options with exercise price a, and all call options weighted by q. The
cost of this portfolio is given by (2) [where in this case D(q) is one point and
the second and third terms are zero].
Next, if q is not continuous, it can be written in the form

Notice that the second term in (3) can be considered as a part of the summation (the third
term) as B = -c_(O) in the following sense: for x ~0, a portfolio of one call option with exercise
price zero and --x bonds can be considered as a call option with exercise price x, as it will pay
M - x = max (A4 - x, 0). Its price is c(x) = c(0) - rB, hence c_(0) = - B.
 A. Bick, Valuation of deriootiue assets 335

Xl..mdw={;
:t;;x,)=,_, lim n[max(M -x,0)-max(M -x- l/n,O)],

and

Xlr.m,(M)_r~
:t;e::se}=_, lim n[max (M -x + l/n, 0) -max (M -x, O)],

and Q is a continuous function that has the same derivatives as q and such
that Q(0) = q(0).It is clear that the cost of xCX,,,(M) is

lim n[c(x)-c(x+ l/n)] = -c+(x),

n-rm

and the cost of x,_,) (M) is -c_(x). It is left to the reader to conclude the
argument.

(b) Transforming eq. (2) into the form (3) is only a matter of calculus.
Again, suppose for the sake of simplicity that there is only one positive point
a in which c or q do not necessarily have a continuous second derivative and
in which q is continuous on the left but not necessarily on the right. The
origin may possibly be another discontinuity point of q. The reader can
verify that integration by parts twice gives

% qW&Wy= -q+tO)ctO)+q+(O)c+tO)-(q+(a)-q~ta))c(a)

+ q +6+?+(a)- qtw (4 + 7
0
dWty)dy

+ Liz Cd(b)44 - qt&(b)l. (4)

It is assumed that the last term is zero9 (a more rigorous approach is

presented in appendix 1). Now substitute (4) in (2) (modified to our simple

Integrate Iz and I.separately and then add and take the limit as b-co.
This limit is zero for a large set of functions q. as (a) lim,,, c(x)=0 is a very plausible
assumption because lim,,, max(M-x,0)=0 for each M, and (b) lim,,,xc(x)=O ~ to prove
this, recall that c is convex [see Smith (19?6)] and see Taylor (1955, problem 10, p. 642), or
Polya and Szegij (1972, problem 113).
336 A. Bick, Valuation of dmivative assets

case) and note, in particular, that if c is continuous at a, all the discrete

terms which correspond to a are cancelled. This will give the desired
result. Q.E.D.

The general valuation formula (3) is a combination of equation (2) in BLs

paper, which is correct when c,, is continuous and c,(7:0+)+BT=0, and
their discrete distribution formula [their equation (l)] which correctly values
payoffs of mass points. The above proof follows the economic interpretation
of eq. (2) of this paper, and (3) follows from it mathematically. It is also
interesting to prove Proposition 1 by applying BLs method, which provides
insights for eq. (3). This will be done in appendix 2.
The reader can verify that (3) holds in particular for the case of a bond, a
call option, or the stock itself. lo Applying the formula for the case q(M,)
= max (0, x - MT), one obtains

V(q) = c(x) - c(0)+ XB,

i.e., the put-call parity.

Note that the valuation functional V( .) is linear and satisfies V(y) 2 0 for
42.0 [as ~20, c+(O)+BzO - see Smith (1976), and c+(a)-c_(u)20 - see
Rockafellar (1970, theorem 25.3)], in accordance with the works of
Rubinstein (1976), Garman (1978), Ross (1978), and others.

3. Investors beliefs and the structure of call options prices

The price of a derivative asset is expressed in the previous section in terms

of the bonds and the call options prices. This relies on the spanning
property of these securities and the lack of arbitrage opportunities, and it is
true regardless of the way in which the price system is determined. For
example, the analysis is valid in a two-date, multi-period or continuous-time
framework and nothing is assumed on investors preferences or beliefs. While
c(7x) may be determined in equilibrium by all these characteristics of the
economy, the functional relation between V(q) and c(7,x) is independent of
them.
The discrete terms which were added to the original BLs formula
correspond to mass points in the probability distribution of the underlying
stock M,. If M, has a continuous probability distribution with no mass
points, (2) and (3) reduce to the BL (continuous) formula (1). As c,(?;u+)
-c,(T,(t-) is the implicit price of a contingent claim that pays one dollar
at time T if and only if MT=u, it is clear that this is positive if and only if

OThe latter case requires that the stock does not pay dividends, so that tt can be considered
as a security which pays M, at time T and zero at any other time. Again, this condition is not
required for the formula to be true for other derivative assets which pay only at time T
 A. Bick, Valuation of derivative assets 337

all investors attribute positive probability to this event. Similarly c,(7:0+)

+B, is positive if and only if all investors attribute positive probability to
the case of bankruptcy before or at time 7: In the pure discrete case where
M, is atomic (investors are assumed to agree on the mass points but not
necessarily on their weights) it follows from arbitrage considerations that
c(T,:x) must be piecewise linear in x, and thus only the discrete terms in (3)
remain.
This can be demonstrated more explicitly in the context of a multi-period
(finite or infinite horizon) exchange economy where investors beliefs are not
necessarily homogeneous. In addition to the given underlying stock and its
call options (with notation as before), other securities and their derivative
assets may be traded. Suppose at time t=O the ith individual maximizes
(with respect to all feasible dynamic strategies) c, Ei[U,,(di,)], where Ui, is
his/her utility of time t consumption, (dio,d,,.di,. .) is the consumption
stream in dollars (random variables for tz I), and Ei is the expectation
operator. Suppose the ith investors subjective probability distribution of M,
is determined, except in a finite number of mass points, by a continuous
density function fi,. It follows that in equilibrium, for each investor i and
time Tzl,

=K,,E,[max (MT-s,O)E,[U~,(d,,)/ MT]]

=Kir 7 ~fir(m)(m-X)Ei[~j:.T(diT) ( Mr=m]dm

X

(5)

where the summation is on the mass points of MT (as viewed by the ith

More specilically, c(T..~)=[,(7;ui)+(u,~x)Ai for a,. , <xzu,, where U, , and ~1, are two
adjacent mass points of M, and A, is the price of the derivative asset x,.,, , ,(Mr) (see section 2).
Eq. (5) is similar to eq. (5) in Brennan (1979) which is derived for a two-date model with a
finite number of securities. This can be generalized to our case by applying a variational method:
let di,,d,,,d,,,. ,. be the cash stream, as viewed by investor i at time I =O, resulting from his
optimal strategy (which is assumed to exist). Construct a new feasible strategy by combining this
optimal strategy with c times the following zero-investment strategy: Invest $1 in purchasing
I/c(T,x) units of the (7;x)-call option and -.$I in selling short l/B, units of the T-bond. The
zero-investment strategy pays D,(M,, x) = max (M, - x, O)/r( 7; x) - l/B, at time 7: Hence, because
of optimality,

02 ds r=l
,Fr Wi,(&)l +EiCUi,(dir+ED,(M,,x))l
I 1 I
= E~CW~AM~X)I.
Substituting D,(Mr,x) from above and rearranging will yield eq. (5)
338 A. Bick, Valuation of derivative assets

individual) and where

Differentiation yields

for each a>O, and

IMT=xI,
C,,(7;x)=1;.T(X)KiTEiCUfT(diT) (54

for each x which is not a mass point.

To further see the role of c,, as a density function and to demonstrate an

application of the valuation formula, it will now be shown that c,, satisfies
the Chapman-Kolmogorov equation [see, e.g., Schuss (1980, p. lOl)]. This is
true under the assumption that there is a deterministic function c(t, M, TX)
which gives the (7;x)-call options price at each time t as a function of the
underlying stocks price M = M,. This is the case, for example, in the Black-
Scholes model (see section 4) and related models, but here continuous
trading or an explicit distribution is not postulated. It is assumed, however,
that c is suitably smooth such that (3) is reduced to (1). Suppose, now, that
the market is open at the three times t<r < 7: Then at time t one knows that
the (7;x)-call option can be sold at time 7 for q(M,)=c(z, M,, 7:x). This
payoff can be constructed at time t from call options that expire at time 7,
and (I) gives

c(t, M, 7; 4 = [47, Y, 7: N~cl~x) Itr.,w.z.y&

Hence

a2@x2
lk M. T.x) = %Wax2
Itr.y.T.x,a2CW
I~r.M,r.yh4

which is the analog to the Chapman-Kolmogorov equation in probability

theory. Garman (1978), referring to general pricing operators, termed this
property intertemporal consistency.

4. The Black-Scholes case

It is interesting to write eq. (3) for this important case. To recapitulate, if

continuous trading is allowed and the underlying stocks price follows a
 A. Bick, Valuation ofderivativeassets 339

geometric Brownian motion, and, for the sake of simplicity, it does not pay
dividends in the relevant time interval, then the Black-&holes (1973) formula
implies that the price at time t of a European call option with maturity at
time T 2_t and exercise price x is

c(t,S, Tx)=SN(d,)-e-T-xN(d,), (6)

where

d,=d,(t,S,7:x)=[lnS/x+(r-a*/2)(T-t)]/afi,

d,=d2+aJT-t.

S= S, is the underlying stocks price at time t, r is the continuously

compounded interest rate (a constant), r~ is the instantaneous variance of the
stocks rate of return (a constant), and N( .) is the cumulative distribution
function of the standard normal variable [the density function will be
denoted n( .)].I3 The reader can verify that in this case eq. (3) reduces to the
BL (continuous) formula (1). Differentiation of (6) yields

a2c/ax2=
(e- TmL)/xoF)n(d2).

Hence, as BL observe, eq. (1) implies that the value at time t of a derivative
asset, which pays q(S,) at time T and zero at any other time, is

V,(t,S)=(e- rT -%Jr-t, 1 (q(y)lyMd,(t, S, 7; My. (7)

Recall that in this model a call option with any exercise price can be
produced by continuously trading the stock and a bond [see Merton (1977)],
and thus assumption (ii) (in section 2) is satisfied. It is interesting to note that
(7) can also be proved directly without using BLs valuation formula: Merton
(1977)14 shows that one can construct a continuous trading strategy of the
stock and a bond that pays (in our notation) q(S,) at time T and zero at any
other time. Merton shows that the portfolios value, at any time t and for
any S=S,, coincides with the unique twice-continuously differentiable
function F(t, S) that solves the differential equation

aFlat++02s2(a2F/as2)+rs(aF/as)=
rF, (8)

with the boundary condition F( 7: S) = q(S).

Thus, it is left to prove
)The more conventional notation S for the stocks price is now used, instead of M which
was used before in order to keep notation similar to BLs. Note that S, and e-(-) correspond
to M and B, respectively, in Proposition 1, and the time of evaluation is t instead of 0.
%ze also Bergman (I 98 1).
340 A. Bick, Valuation of derivative assets

Proposition 2. 5 s(t,S) as in (7) is a solution to the differential equation (8)

with the above boundary condition.

Proof (outline). It is known that c(t, S, I,x) as in (6) satisfies (8), and this
implies that

1 q(y)(a/ay)c(t, S, 7: y)dy satisfies (8).

To see that the boundary condition is satisfied, take fixed t, S and T and
substitute w= fid,(t,S, T,y) in (7) to obtain
,-wZ/2(T-I)
ow+(r-o~/2)(T-I)
Vq(trS)=e-T-) _[ q(Se (9)
JgG)dW.

As t-+T, the second factor in the integrand tends to the Dirac function at
w=O [see, e.g., Schuss (1980, p. 22)], i.e., the integral will tend to q(S). Q.E.D.

This demonstrates that the BL approach, which is based on the

construction at one time t of a replicating portfolio from the bond and all
the call options, gives the same result as Mertons approach, which is to
construct a replicating continuous-trading strategy from the bond and the
stock. Of course, this method cannot offer an alternative to Mertons
approach when the derivative asset itself is a call option.

5. Example: The valuation of compound options

While Geske (1979) used stochastic calculus methods to price options on

options, the same result can be obtained merely by a mechanical substitution
in the BL valuation formula for the Black-Scholes case. The framework will
be as in section 4. In order to keep the notation similar to Geskes, let u, be
the value at time t of the underlying security which is assumed to follow a
geometric Brownian motion with instantaneous variance of return 02. Let S,
be the value of a (7;m)-European call option on the above security, and let c,
be the value of a (t*,K)-European call option on the above European call
option, where t* < 7: [See Geske (1979) for the economic interpretation.] The
payoff of the latter security at time t* is

q(u,.) = max (S,. - K, 0)

=max (v,,N(d,(t*, u,., Tm))- me-rcT-f*)N(d2(t*, v,+ 7; m))- K, 0)

15After having written this paper it was brought to the authors attention that this proposition
was also proved in an unpublished paper by Black (1974). who also recognized (7) as the value
of a general derivative asset in the Black-Scholes case.
 A. Bick, Valuation of deriuative assets 341

where S,.=S(c*, u,., Tm) is substituted from (6). Let V be the value of 0,. for
which S,, - K = 0. Then [see Smith (1976, eq. (45))]

S,. - K > 0 if and only if II,. > V.

Eq. (7) now gives, for t <t*,

c,= q4 4

=(e- r(r*mf)/afi) $S[yN(d,(t,

y,7;m)
-me-rcT-*)N(d2(t*, y, 7;m))- K] n(dz(4u,t*,y))dy,

where u = v,. Breaking this into three integrals, then substituting x

=dl(t, u, t*, y) in the first one and x=d2(t, u, t*, y) in the other two, results
in16

d,(t,u, Tm)-(JSIIZr)x
n(x)dx
JFw2 >

- Ke - N(d,(t, II, t*, I?)),

where r1 = t* - t, 52 = T-t. By using the last equation in Geske (1979, p. 80)

it follows without difficulty that eq. (10) is identical to eq. (4) in Geske (1979).

6. Example: The Merton proportional dividend model

Merton (1973, p. 171) [see also Smith (1976, p. 26)] modified the Black-
Scholes model for the case of a constant dividend yield. It will now be
demonstrated that this result can also be obtained from Black-Scholes
formula (with no dividends) and BLs valuation formula.
The continuous dividend stream is assumed to have a density of 6S,, where
6 is a positive constant and S, is the stocks price. This can be used to

16The following identities, which are readily verified from the detinitions, are used in the
transformation:

d,(t*,~~exp((r+a2/2)r,-xcr~),~m)=(d,(t,v,T,m)-~x)/~~~,
and the same formula is true with r-u/2 on the LHS and d, instead of d, in both sides.
342 A. Bick, Valuation of derivative assets

purchase 6 new shares per each share held. Fix a current time t, and a future
time 7: Then reinvestment of dividends will transform one share at time t, to
a(t)=e(-) shares at time t> t, (as da/a=ddt), regardless of the path of S.
S~=ed(-ro)SI, the value of a continuously reinvested position, is assumed to
follow a geometric Brownian motion with instantaneous variance of return

Now, the Black-Scholes model can be applied to 9; and call options on S^,
(which can be produced from ,$ and a bond). The value of such a (7, x)-call
option, t(t,,S, TX), is given by (6) with the obvious modification of notation.
The unprotected (I;x)-call option on S, can be viewed as a derivative asset of
St with payoff at time T max(S,-x,O)=max(e~sSr-xX,O), where z = T-t,.
It follows from (7) that its current price is

~GW,>S>77y))dy.

Breaking this into two terms, substituting w=d,(t,, S, 7; y) in the first and
w = d2(t0, S, 7: y) in the second, results in

c(t,, S, 7; x) =e-SN(d,(t,, S, 7: xed))-e-xN(d,(t,, S, 7;xe)),

which is identical to eq. (53) in Smith (1976). In the case where s=_/(t)
depends on the time (but not on the stocks price) one obtains the same
formula only with ftf(t)dt instead of 6~.

Appendix 1

In the proof of part (b) of Proposition 1 it was assumed, for simplicity,

that
L = lim [q(b)c(b) - q(b)c(b)] = 0.
b-+m

In fact, this is necessarily the case if V(q) is finite. More specifically,

f: q(y)c(y)dy converges [i.e., V(q) is defined and finite] if and only if
J-2 q(y)c(y)dy converges and in this case

lim [q(b)c(b) - q(b)c(b)] = 0.

b-m

Proof: Applying the fact that

lim c(x)=0 and lim xc(x) =O,

x-+m X-X
See footnote 9.
 A. Bick, Valuation of derivativeassets 343

integration by parts gives

1 (z - x)c(z)dz = c(x) for x 2 a.

Hence

7 q(y)c(y)dy = % [ (z - y)c(z)q(y)dzdy = s,s (z - y)c(z)q(y)dzdy

*

= -Tc"(z)[q'+(a)(z-a)+q+(a)]dz+ jc(z)q(z)dz
LI

= c+(a)q+(4 - c(a)q+(4 + ckMzkk 7

0
(11)

where

A={(y,z)eR2; azLy<cn, ySz<co}

={(y,z)~R; aszzm, asJsz},

and the technique of changing the order of integration in a double integral is

used. In the fifth and seventh equalities, integration by parts and the above
limits are applied.
This implies the if and only if part. Integration by part twice of the LHS
in (11) gives (as in section 2) the same expression as on the RHS, only with
the addition of J!,. This implies that L=O.

Appendix 2

An alternative proof I* for Proposition 1 is given, using BLs framework;

the notation will be as in section 2 of this paper. Consider the discrete case
as described by BL, and let P(M) be the current price of one dollar
contingent upon level M of the underlying stock at time 7: By arbitrage

This proof was provided by Douglas T. Breeden

344 A. Bick, Valuation of derivative assets

consideration, the value of the payoff q(M) is

Substituting BLs formula [their equation (I)] for P(M) gives

11.(12)
c(M+AM)-c(M) C(M)-c(M-AM)
w =& 4(M) AM

To prove (b), let AM+0 in (12). The discontinuity points of c will give

J& da)(c+(4 -c-(a)).

In particular, the term which corresponds to M=O, if it is a mass point, will

bei

q(O)(d+(0) -cr. (0)) = q(O)(c+(0) + B).

Wherever c exists, the limit as AM-+0 of the bracketed expression in (12)

may be represented as c(M)dM. Since there are only a finite number of
discontinuities, the limit as AM+0 of the remaining infinite summation
equals

Combined with the above discrete terms, it gives eq. (3).

To prove (a), it is assumed that q is continuous (the discontinuous case can
then be obtained as in section 2). Let M take on discrete values {nA} for
n=O, 1,2... It follows from (12) that

v(q)= f dnAV(W
n=O

= dW + q(Ok(A)/A- dW(WA

+; f q(nA)c((n- l)A)-2 $ q(nA)c(nA)+ f q(nA)c((n+ l)A)

II=1 n=l =,

See footnote 7 for the interpretation of negative exercise price. It can also be verified directly
from arbitrage considerations that P(0) = [c(dM) - c(O)]/dM + B.
See also footnote 19.
 A. flick, Valuation of derivative assets 345

+; z q(nLl)c((n- l)A)-2 f q(nfl)c(nd)+ f q(n&((n+ l)d)

n-2 It=1 n=O 1
i

=q(o)~+q(A)-q(o)~c(o)
A

q((k+W-qNk+ l)A)_d(k+W)-q(kA)
+kzoc((k+
l)A) A A

assuming convergence of the infinite summations. Taking limits as A+0

where possible (recognizing only a finite number of discontinuities) gives the
valuation formula (2) for the case where q is continuous.

References
Bergman, Y.Z., 1981, A characterization of self-tinancing portcolio strategies, Research Program
in Finance working paper no. I 13 (University of Califorma, Berkeley, CA).
Black, F., 1974, The pricing of complex options and corporate liabilities, Mimeo. (University of
Chicago, Chicago, IL.).
Black, F. and M. Scholes, 1973, The pricing of options and corporate liabilities, Journal of
Political Economy 8 I, 637-659.
Breeden, D.T. and R.H. Litzenberger. 1978, Prices of state-contingent claims implicit in option
prices, Journal of Business 51, 621-651.
Brennan, M.J., 1979, The pricing of contingent claims in discrete time models, Journal of
Finance 34, 53-68.
Cox, J.C. and S.A. Ross, 1976, The valuation of options for alternative stochastic processes,
Journal of Financial Economics 3, 1455166.
Cox, J.C. and M. Rubinstein, 1982, Options markets (Prentice-Hall, Englewood Cliffs. NJ)
forthcoming.
Cox, J.C., S.A. Ross and M. Rubinstein, 1979, Option pricing: A simplified approach. Journal of
Financial Economics 7, 2299263.
Garman, M.B., 1976, An algebra for evaluating hedge portfolios, Journal of Financial Economtcs
3, 4033427.
Barman, M.B., 1978, A synthesis of the pure theory of arbitrage, Research Program in Finance
working paper no. 98 (University of California, Berkeley, CA),
Geske, R., 1979, The valuation of compound options, Journal of Financial Economics 7, 63-81.
Merton, R.C., 1973, Theory of rational option pricing, Bell Journal of Economics and
Management Science 4, 141-183.
Merton, R.C., 1977, On the pricing of contingent claims and the Modigliani~Miller theorem,
Journal of Financial Economics 5, 241-249.
Polya, G. and G. Szego, 1972, Problems and theorems in analysis. Vol. 1 (Springer-Verlag,
Berlin).
Rockafellar, R.T., 1970, Convex analysis (Princeton University Press, Princeton, NJ).
Ross, S.A., 1978, A simple approach to the valuation of risky streams, Journal of Business 51,
453-415.
Rubinstein, M., 1976, The valuation of uncertain income streams and the pricing of options, Bell
Journal of Economics 7,407-425.
Schuss, Z., 1980, Theory and applications of stochastic dilTerential equations (Wiley, New York).
Smith, C.W., 1976, Option pricing: A review, Journal of Financial Economics 3, 3 52
Taylor, A.E., 1955, Advanced calculus (Ginn and Co., Boston, MA).