Four Derivations of the Black-Scholes Formula

by Fabrice Douglas Rouah

www.FRouah.com
www.Volopta.com
In this note we derive in four separate ways the well-known result of Black
and Scholes that under certain assumptions the time-t price C(o
|
, 1, T) of a
European call option with strike price 1 and maturity t = T t on a non-
dividend stock with spot price o
|
and a constant volatility o when the rate of
interest is a constant r can be expressed as
C(o
|
, 1, T) = o
|
(d
1
) c
:r
1(d
2
) (1)
where
d
1
=
ln
St
1
+
_
r +
c
2
2
_
t
o
p
t
and d
2
= d
1
o
p
t, and where (j) =
1
p
2t
_

1
c

1
2
|
2
dt is the standard normal
cdf. We show four ways in which Equation (1) can be derived.
1. By straightforward integration.
2. By applying the Feynman-Kac theorem.
3. By transforming the Black Scholes PDE into the heat equation, for which
a solution is known. This is the original approach adopted by Black and
Scholes [1].
4. Through the Capital Asset Pricing Model (CAPM).
Free code for the Black-Scholes model can be found at www.Volopta.com.
1 Black-Scholes Economy
There are two assets: a risky stock o and riskless bond 1. These assets are
driven by the SDEs
do
|
= jo
|
dt +oo
|
d\
|
(2)
d1
|
= r
|
1
|
dt
The time zero value of the bond is 1
0
= 1 and that of the stock is o
0
. The
model is valid under certain market assumptions that are described in John
1
Hulls book [3]. By Itos Lemma the value \
|
of a derivative written on the
stock follows the diusion
d\
|
=
0\
0t
dt +
0\
0o
do +
1
2
0
2
\
0o
2
(do)
2
(3)
=
0\
0t
dt +
0\
0o
do +
1
2
0
2
\
0o
2
o
2
o
2
dt
=
_
0\
0t
+jo
|
0\
0o
+
1
2
o
2
o
2
|
0
2
\
0o
2
_
dt +
_
oo
|
0\
0o
_
d\
|
.
2 The Lognormal Distribution
2.1 The Lognormal PDF and CDF
In this Note we make extensive use of the fact that if a random variable 1 2 R
follows the normal distribution with mean j and variance o
2
, then A = c
Y
follows the lognormal distribution with mean
1 [A] = c
+
1
2
c
2
(4)
and variance
\ ar [A] =
_
c
c
2
1
_
c
2+c
2
. (5)
The pdf for A is
d1

(r) =
1
or
p
2
exp
_

1
2
_
lnr j
o
_
2
_
(6)
and the cdf is
1

(r) =
_
lnr j
o
_
(7)
where (j) =
1
p
2t
_

1
c

1
2
|
2
dt is the standard normal cdf.
2.2 The Lognormal Conditional Expected Value
The expected value of A conditional on A r is 1

(1) = 1 [AjA r]. For

the lognormal distribution this is, using Equation (6)
1

(1) =
_
1
1
1
o
p
2
c

1
2
(
ln x

)
2
dr.
Make the change of variable j = lnr so that r = c

, dr = c

dj and the
Jacobian is c

. Hence we have
1

(1) =
_
1
ln 1
c

o
p
2
c

1
2
(
y

)
2
dj. (8)
2
Combining terms and completing the square, the exponent is

1
2c
2
_
j
2
2jj +j
2
2o
2
j
_
=
1
2c
2
_
j
_
j +o
2
__
2
+j +
1
2
o
2
.
Equation (8) becomes
1

(1) = exp
_
j +
1
2
o
2
_
1
o
_
1
ln 1
1
p
2
exp
_
_

1
2
_
j
_
j +o
2
_
o
_
2
_
_
dj. (9)
Consider the random variable A with pdf )

(r) and cdf 1

(r), and the scale-

location transformation 1 = oA+j. It is easy to show that the Jacobian is
1
c
,
that the pdf for 1 is )
Y
(j)=
1
c
)

c
_
and that the cdf is 1
Y
(j) = 1

c
_
.
Hence, the integral in Equation (9) involves the scale-location transformation
of the standard normal cdf. Using the fact that (r) = 1 (r) this implies
that
1

(1) = exp
_
j +
o
2
2
_

_
ln1 +j +o
2
o
_
. (10)
See Hogg and Klugman [2].
3 Solving the SDEs
3.1 Stock Price
Apply Itos Lemma to the function lno
|
where o
|
is driven by the diusion in
Equation (2). Then lno
|
follows the SDE
d lno
|
=
_
j
1
2
o
2
_
dt +od\
|
. (11)
Integrating from 0 to t, we have
_
|
0
d lno
u
=
_
|
0
_
j
1
2
o
2
_
dn +o
_
|
0
d\
u
so that
lno
|
lno
0
=
_
j
1
2
o
2
_
t +o\
|
since \
0
= 0. Hence the solution to the SDE is
o
|
= o
0
exp
__
j
1
2
o
2
_
t +o\
|
_
. (12)
Since \
|
is distributed normal (0, t) with zero mean and variance t we have
that lno
|
follows the normal distribution with mean lno
0
+
_
j
c
2
2
_
t and
variance o
2
t. This implies by Equations (4) and (5) that o
|
follows the lognor-
mal distribution with mean o
0
c
|
and variance o
2
0
c
2|
_
c
c
2
|
1
_
. We can also
integrate Equation (11) from t to T so that, analogous to Equation (12)
o
T
= o
|
exp
__
j
1
2
o
2
_
t +o (\
T
\
|
)
_
(13)
and o
T
follows the lognormal distribution with mean o
|
c
r
and variance given
by o
2
|
c
2r
_
c
c
2
r
1
_
.
3
3.2 Bond Price
Apply Itos Lemma to the function ln1
|
. Then ln1
|
follows the SDE
d ln1
|
= r
|
dt.
Integrating from 0 to t we have
d ln1
|
d ln1
0
=
_
|
0
r
u
dn.
so the solution to the SDE is 1
|
= exp
_
_
|
0
r
u
dn
_
since 1
0
= 1. When in-
terest rates are constant then r
|
= r and 1
|
= c
:|
. Integrating from t to T
produces the solution 1
|,T
= exp
_
_
T
|
r
u
dn
_
or 1
|,T
= c
:r
when interest rates
are constant.
3.3 Discounted Stock Price is a Martingale
We want to nd a measure Q such that under Q the discounted stock price that
uses 1
|
is a martingale. Write
do
|
= r
|
o
|
dt +oo
|
d\
Q
|
(14)
where \
Q
|
= \
|
+
:t
c
t. We have that under Q, at time t = 0, the stock price o
|
follows the lognormal distribution with mean o
0
c
:t|
and variance o
2
0
c
2:t|
_
c
c
2
|
1
_
,
but that o
|
is not a martingale. Using 1
|
as the numeraire, the discounted
stock price is
~
o
|
=
St
1t
and
~
o
|
will be a martingale. Apply Itos Lemma to
~
o
|
,
which follows the SDE
d
~
o
|
=
0
~
o
01
d1
|
+
0
~
o
0o
do
|
(15)
since all terms involving the second-order derivatives are zero. Expand Equation
(15) to obtain
d
~
o
|
=
o
|
1
2
|
d1
|
+
1
1
|
do
|
(16)
=
o
|
1
2
|
(r
|
1
|
dt) +
1
1
|
_
r
|
o
|
dt +oo
|
d\
Q
|
_
= o
~
o
|
d\
Q
|
.
The solution to the SDE (16) is
~
o
|
=
~
o
0
exp
_

1
2
o
2
t +o\
Q
|
_
.
This implies that ln
~
o
|
follows the normal distribution with mean ln
~
o
0

c
2
2
t and
variance o
2
t. To show that
~
o
|
is a martingale under Q, consider the expectation
4
under Q for : < t
1
Q
_
~
o
|
j1
s
_
=
~
o
0
exp
_

1
2
o
2
t
_
1
Q
_
exp
_
o\
Q
|
_

F
s
_
=
~
o
0
exp
_

1
2
o
2
t +o\
Q
s
_
1
Q
_
exp
_
o
_
\
Q
|
\
Q
s
__

F
s
_
At time : we have that \
Q
|
\
Q
s
is distributed as (0, t :) which is identical
in distribution to \
Q
|s
at time zero. Hence we can write
1
Q
_
~
o
|
jF
s
_
=
~
o
0
exp
_

1
2
o
2
t +o\
Q
s
_
1
Q
_
exp
_
o\
Q
|s
_

F
0
_
.
Now, the moment generating function (mgf) of a random variable A with normal
distribution (j, o
2
) is 1
_
c

= exp
_
jc +
1
2
c
2
o
2
_
. Under Q we have that
\
Q
|s
is Q-Brownian motion and distributed as (0, t :). Hence the mgf of
\
Q
|s
is 1
Q
_
exp
_
o\
Q
|s
__
= exp
_
1
2
o
2
(t :)
_
where o takes the place of c,
and we can write
1
Q
_
~
o
|
jF
s
_
=
~
o
0
exp
_

1
2
o
2
t +o\
Q
s
_
exp
_
1
2
o
2
(t :)
_
=
~
o
0
exp
_

1
2
o
2
: +o\
Q
s
_
=
~
o
s
.
We thus have that 1
Q
_
~
o
|
jF
s
_
=
~
o
s
, which shows that
~
o
|
is a Q-martingale.
Pricing a European call option under Black-Scholes makes use of the fact that
under Q, at time t the terminal stock price at expiry, o
T
, follows the normal
distribution with mean o
|
c
:r
and variance o
2
|
c
2:r
_
c
c
2
r
1
_
when the interest
rate r
|
is a constant value, r. Finally, note that under the original measure the
process for
~
o
|
is
d
~
o
|
= (j r)
~
o
|
dt +o
~
o
|
d\
|
which is obviously not a martingale.
3.4 Summary
We start with the processes for the stock price and bond price
do
|
= jo
|
dt +oo
|
d\
|
d1
|
= r
|
1
|
dt.
We apply Itos Lemma to get the processes for lno
|
and ln1
|
d lno
|
=
_
j
1
2
o
2
_
dt +od\
|
d ln1
|
= r
|
dt,
which allows us to solve for o
|
and 1
|
o
|
= o
0
c
(
1
2
c
2
)|+cVt
1
|
= c
R
t
0
:sJs
.
5
We apply a change of measure to obtain the stock price under the risk neutral
measure Q
do
|
= ro
|
+oo
|
d\
Q
|
)
o
|
= o
0
c
(:
1
2
c
2
)|+cV
Q
t
Since o
|
is not a martingale under Q, we discount o
|
by 1
|
to obtain
~
o
|
=
St
1t
and
d
~
o
|
= o
~
o
|
d\
Q
|
)
~
o
|
=
~
o
0
c

1
2
c
2
|+cV
Q
t
so that
~
o
|
is a martingale under Q. The distributions of the processes described
in this section are summarized in the following table
Stochastic Lognormal distribution for o
T
j F
|
Process a
Process mean variance martingale
do = jodt +ood\ o
|
c
r
o
2
|
c
2r
_
c
c
2
r
1
_
No
do = rodt +ood\
Q
o
|
c
:r
o
2
|
c
2:r
_
c
c
2
r
1
_
No
d
~
o = o
~
od\
Q
with
~
o =
S
1
~
o
|
~
o
2
|
_
c
c
2
r
1
_
Yes
d
~
o =(j r)
~
odt +o
~
od\ o
|
c
(:)r
o
2
|
c
2(:)r
_
c
c
2
r
1
_
No
.
This also implies that the logarithm of the stock price is normally distributed.
4 The Black-Scholes Call Price
In the following sections we show four ways in which the Black-Scholes call price
can be obtained. Under a constant interest rate r the time-t price of a European
call option on a non-dividend paying stock when its spot price is o
|
and with
strike 1 and time to maturity t = T t is
C(o
|
, 1, T) = c
:r
1
Q
_
(o
T
1)
+

F
|
_
(17)
which can be evaluated to produce Equation (1), reproduced here for conve-
nience
C(o
|
, 1, T) = o
|
(d
1
) 1c
:r
(d
2
)
where
d
1
=
log
St
1
+
_
r +
c
2
2
_
t
o
p
t
and
d
2
= d
1
o
p
t
=
log
St
1
+
_
r
c
2
2
_
t
o
p
t
.
6
The rst derivation is by straightforward integration of Equation (17); the sec-
ond is by applying the Feynman-Kac theorem; the third is by transforming the
Black-Scholes PDE into the heat equation and solving the heat equation; the
fourth is by using the Capital Asset Pricing Model (CAPM).
5 Black-Scholes by Straightforward Integration
The European call price C(o
|
, 1, T) is the discounted time-t expected value of
(o
T
1)
+
under the EMM Q and when interest rates are constant. Hence
from Equation (17) we have
C(o
|
, 1, T) = c
:r
1
Q
_
(o
T
1)
+

F
|
_
(18)
= c
:r
_
1
1
(o
T
1)d1(o
T
)
= c
:r
_
1
1
o
T
d1(o
T
) c
:r
1
_
1
1
d1(o
T
).
To evaluate the two integrals, we make use of the result derived in Section (3.3)
that under Q and at time t the terminal stock price o
T
follows the lognormal
distribution with mean lno
|
+
_
r
c
2
2
_
t and variance o
2
t, where t = T t is
the time to maturity. The rst integral in the last line of Equation (18) uses
the conditional expectation of o
T
given that o
T
1
_
1
1
o
T
d1(o
T
) = 1
Q
[ o
T
j o
T
1]
= 1
S
T
(1).
This conditional expectation is, from Equation (10)
1
S
T
(1) = exp
_
lno
|
+
_
r
c
2
2
_
t +
c
2
r
2
_

_
_
ln1 + lno
|
+
_
r
c
2
2
_
t +o
2
t
o
p
t
_
_
= o
|
c
:r
(d
1
),
so the rst integral in the last line of Equation (18) is
o
|
(d
1
). (19)
7
Using Equation (7), the second integral in the last line of (18) can be written
c
:r
1
_
1
1
d1(o
T
) = c
:r
1 [1 1(1)] (20)
= c
:r
1
_
_
1
_
_
ln1 lno
|

_
r
c
2
2
_
t
o
p
t
_
_
_
_
= c
:r
1 [1 (d
2
)]
= c
:r
1(d
2
).
Combining the terms in Equations (19) and (20) leads to the expression (1) for
the European call price.
5.1 Change of Numeraire
The principle behind pricing by arbitrage is that if the market is complete we
can nd a portfolio that replicates the derivative at all times, and we can nd an
equivalent martingale measure (EMM) N such that the discounted stock price
is a martingale. Moreover, the EMM N determines the unique numeraire
|
that discounts the stock price. The time-t value \ (o
|
, t) of the derivative with
payo \ (o
T
, T) at time T discounted by the numeraire
|
is
\ (o
|
, t) =
|
1
N
_
\ (o
T
, T)

F
|
_
. (21)
In the derivation of the previous section, the bond 1
|
= c
:|
serves as the nu-
meraire, and since r is deterministic we can take
T
= c
:T
out of the expectation
and with \ (o
T
, T) = (o
T
1)
+
we can write
\ (o
|
, t) = c
:(T|)
1
N
_
(o
T
1)
+

F
|
_
which is Equation (17) for the call price.
5.1.1 Black Scholes Under a Dierent Numeraire
In this section we show that we can use the stock price o
|
as the numeraire and
recover the Black-Scholes call price. We start with the stock price process in
Equation (14) under the measure Q and with a constant interest rate
do
|
= ro
|
dt +oo
|
d\
Q
|
. (22)
The relative bond price price is dened as
~
1 =
1
S
and by Itos Lemma follows
the process
d
~
1
|
= o
2
~
1
|
dt o
~
1
|
d\
Q
|
.
The measure Q turns
~
o =
S
1
into a martingale, but not
~
1. The measure P
that turns
~
1 into a martingale is
\
P
|
= \
Q
|
ot (23)
8
so that
d
~
1
|
= o
~
1
|
d\
P
|
is a martingale under P. The value of the European call is determined by using

|
= o
|
as the numeraire along with the payo function \ (o
T
, T) = (o
T
1)
+
in the valuation Equation (21)
\ (o
|
, t) = o
|
1
P
_
(o
T
1)
+
o
T

F
|
_
(24)
= o
|
1
P
[ (1 17
T
)j F
|
]
where 7
|
=
1
St
. To evaluate \ (o
|
, t) we need the distribution for 7
T
. The
process for 7 =
1
S
is obtained using Itos Lemma on o
|
in Equation (22) and
the change of measure in Equation (23)
d7
|
=
_
r +o
2
_
7
|
dt o7
|
d\
Q
|
= r7
|
dt o7
|
d\
P
|
.
To nd the solution for 7
|
we dene 1
|
= ln7
|
and apply Itos Lemma again,
to produce
d1
|
=
_
r +
c
2
2
_
dt od\
P
|
. (25)
We integrate Equation (25) to produce the solution
1
T
1
|
=
_
r +
c
2
2
_
(T t) o
_
\
P
T
\
P
|
_
.
so that 7
T
has the solution
7
T
= c
ln 2t

:+

2
2

(T|)c(V
P
T
V
P
t
)
. (26)
Now, since \
P
T
\
P
|
is identical in distribution to \
P
r
, where t = T t is the
time to maturity, and since \
P
r
follows the normal distribution with zero mean
and variance o
2
t, the exponent in Equation (26)
ln7
|

_
r +
c
2
2
_
(T t) o
_
\
P
T
\
P
|
_
,
follows the normal distribution with mean
n = ln7
|

_
r +
c
2
2
_
t = lno
|

_
r +
c
2
2
_
t
and variance = o
2
t. This implies that 7
T
follows the lognormal distribution
with mean c
u+u/2
and variance (c
u
1) c
2u+u
. Note that (1 17
T
)
+
in the
expectation of Equation (24) is non-zero when 7
T
<
1
1
. Hence we can write
this expectation as the two integrals
1
P
[ (1 17
T
)j F
|
] =
_ 1
K
1
d1
2
T
1
_ 1
K
1
7
T
d1
2
T
(27)
= 1
1
1
2
9
where 1
2
T
is the cdf of 7
T
dened in Equation (7). The rst integral in
Equation (27) is
1
1
= 1
2
T
_
1
1
_
=
_
ln
1
1
n

_
(28)
=
_
_
ln1 + lno
|
+
_
r +
c
2
2
_
t
o
p
t
_
_
= (d
1
) .
Using the denition of 1
2
T
(r) in Equation (10), the second integral in Equation
(27) is
1
2
= 1
_
_
1
1
7
T
d1
2
T

_
1
1
K
7
T
d1
2
T
_
(29)
= 1
_
1
P
[7
T
] 1
2
T
_
1
1
__
= 1
_
c
u+u/2
c
u+u/2

_
ln
1
1
+n +
p

__
= 1c
u+u/2
_
_
1
_
_
ln
St
1

_
r
c
2
2
_
t
o
p
t
_
_
_
_
=
1
o
|
c
:r
(d
2
)
since 1 (d
2
) = (d
2
). Substitute the expressions for 1
1
and 1
2
from
Equations (28) and (29) into the valuation Equation (24)
\ (o
|
, t) = o
|
1
P
[ (1 17
T
)j F
|
]
= o
|
[1
1
1
2
]
= o
|
(d
1
) 1c
:r
(d
2
)
which is the Black-Scholes call price in Equation (1).
6 Black-Scholes From the Feynman-Kac Theo-
rem
6.1 The Feynman-Kac Theorem
Suppose that r
|
follows the process
dr
|
= j(r
|
, t)dt +o (r
|
, t) d\
Q
|
(30)
10
and suppose the dierentiable function \ = \ (r
|
, t) follows the partial dier-
ential equation given by
0\
0t
+j(r
|
, t)
0\
0r
+
1
2
o(r
|
, t)
2
0
2
\
0r
2
r(t, r)\ (r
|
, t) = 0 (31)
with boundary condition \ (r
T
, T). The Feynman-Kac theorem stipulates that
\ (r
|
, t) has solution
\ (r
|
, t) = 1
Q
_
c

R
T
t
:(u,u)Ju
\ (A
T
, T)

F
|
_
. (32)
In Equation (32) the time-t expectation is with respect to the same measure Q
under which the stochastic portion of Equation (30) is Brownian motion. See
the Note on www.FRouah.com for illustrations of the Feynman-Kac theorem.
6.2 The Theorem Applied to Black-Scholes
To apply the Feynman-Kac theorem to the Black-Scholes call price, note that
the value \
|
= \ (o
|
, t) of a European call option written at time t with strike
price 1 when interest rates are a constant r follows the Black-Scholes PDE
0\
0t
+ro
|
0\
0o
+
1
2
o
2
o
2
|
0
2
\
0o
2
r\
|
= 0 (33)
with boundary condition \ (o
T
, T) = (o
T
1)
+
. The Note on www.FRouah.com
explains how the PDE (33) is derived. This PDE is the PDE in Equation (31)
with r
|
= o
|
, j(r
|
, t) = ro
|
, and o(r
|
, t) = oo
|
. Hence the Feynman-Kac
theorem applies and the value of the European call is
\ (o
|
, t) = 1
Q
_
c

R
T
t
:(u,u)Ju
\ (o
T
, T)

F
|
_
(34)
= c
:r
1
Q
_
(o
T
1)
+

F
|

which is exactly Equation (18). Hence, we can evaluate the expectation in

(34) by straightforward integration exactly in the same way as in Section 5 and
obtain the call price in Equation (1).
7 Black-Scholes From the Heat Equation
In this section we follow the derivation explained in Wilmott et al. [4]. We
rst present a denition of the Dirac delta function, and of the heat equation.
We then transform the Black-Scholes PDE into the heat equation, apply the
solution through integration, and convert back to the original (untransformed)
parameters. This will produce the Black-Scholes call price.
11
7.1 Dirac Delta Function
The Dirac delta function c(r) is dened as
c(r) =
_
0 if r 6= 0
1 if r = 0
and
_
1
1
c()d = 1.
For an integrable function )(r) we have that
)(0) =
_
1
1
)()c()d
and
)(r) =
_
1
1
)(r )c()d. (35)
7.2 The Heat Equation
The heat equation is the PDE for n = n(r, t) over the domain fr 2 1, t 0g
given by
0n
0t
=
0
2
n
0r
2
.
The heat equation has the fundamental solution
n(r, t) =
1
p
4t
exp
_
r
2
4t
_
(36)
which is the normal pdf with mean 0 and variance 2t. The initial value of the
heat equation is n(r, 0) = n
0
(r) which can be written in terms of the Dirac
delta function c as the limit
n
0
(r) = lim
r!0
n(r, t) = c(r).
Using the property (35) of the Dirac delta function, we can write the initial
value as
n
0
(r) =
_
1
1
c(r )n
0
()d (37)
We can also apply the property (35) to the fundamental solution in Equation
(36) and express the solution as
n(r, t) =
_
1
1
n(r )c()d (38)
=
_
1
1
n(r )n
0
()d
=
_
1
1
1
p
4t
c
(r)
2
/4r
n
0
()d,
with initial value
n(r, 0) =
_
1
1
c(r )n
0
()d = n
0
(r),
as before.
12
7.3 The Black-Scholes PDE as the Heat Equation
Through a series of transformations we convert the Black-Scholes PDE in Equa-
tion (33) into the heat equation. The rst set of transformations convert the
spot price to log-moneyness and the time to one-half the total variance. This
will get rid of the o and o
2
terms in the Black-Scholes PDE. The rst trans-
formations are
r = ln
o
1
so that o = 1c
r
(39)
t =
c
2
2
(T t) so that t = T
2t
o
2
l(r, t) =
1
1
\ (o, t) =
1
1
\
_
1c
r
, T 2t,o
2
_
.
Apply the chain rule to the partial derivatives in the Black-Scholes PDE. We
have
0\
0t
= 1
0l
0t
0t
0t
=
1o
2
2
0l
0t
,
0\
0o
= 1
0l
0r
0r
0o
=
1
o
0l
0r
= c
r
0l
0r
,
0
2
\
0o
2
=
1
o
2
0l
0r
+
1
o
0
0o
_
0l
0r
_
=
1
o
2
0l
0r
+
1
o
0
0r
_
0l
0r
_
0r
0o
=
1
o
2
0l
0r
+
1
o
2
0
2
l
0r
2
=
c
2r
1
_
0
2
l
0r
2

0l
0r
_
.
Substitute for the partials in the Black-Scholes PDE (33) to obtain
1o
2
2
0l
0t
+r1c
r
c
r
0l
0r
+
1
2
o
2
1
2
c
2r
c
2r
1
_
0
2
l
0r
2

0l
0r
_
rl = 0
which simplies to

0l
0t
+ (/ 1)
0l
0r
+
0
2
l
0r
2
/l = 0 (40)
where / =
2:
c
2
. The coecients of this PDE does not involve r or t. The
boundary condition for \ is \ (o
T
, T) = (o
T
1)
+
. From Equation (39),
when t = T and o
|
= o
T
we have that r = ln
S
T
1
which we write as r
T
, and
that t = 0. Hence the boundary condition for l is
l
0
(r
T
) = l(r
T
, 0) =
1
1
\ (o
T
1)
+
=
1
1
(1c
r
T
1)
+
= (c
r
T
1)
+
.
13
We make the additional transformation
\ (r, t) = c
or+o
2
r
l(r, t) (41)
where c =
1
2
(/ 1) and , =
1
2
(/ +1). This will convert Equation (40) into the
heat equation. The partial derivatives of l in terms of \ are
0l
0t
= c
oro
2
r
_
0\
0t
\(r, t),
2
_
0l
0r
= c
oro
2
r
_
0\
0r
c\(r, t)
_
0
2
l
0r
2
= c
oro
2
r
_
c
2
\(r, t) 2c
0\
0r
+
0
2
\
0r
2
_
.
Substitute these derivatives into Equation (40) to obtain
,
2
\(r, t)
0\
0t
+ (/ 1)
_
c\(r, t) +
0\
0r
_
+c\(r, t) 2c
0\
0r
+
0
2
\
0r
2
/\(r, t) = 0
which simplies to the heat equation
0\
0t
=
0\
2
0r
2
. (42)
From Equation (41) the boundary condition for \(r, t) is
\
0
(r
T
) = \(r
T
, 0) = c
or
T
l(r
T
, 0) (43)
=
_
c
(o+1)r
T
c
or
T
_
+
=
_
c
or
T
c
or
T
_
+
.
since , = c + 1. The transformation from \ to \ is therefore
\ (o, t) =
1
1
c
oro
2
r
\(r, t). (44)
7.4 Obtaining the Black-Scholes Call Price
Since \(r, t) follows the heat equation, it has the solution given by Equation
(38), with boundary condition given by (43). Hence the solution is
\ (r, t) =
1
p
4t
_
1
1
c
(r)
2
/4r
\
0
()d
=
1
p
4t
_
1
1
c
(r)
2
/4r
_
c
o
c
o
_
+
d.
14
Make the change of variable . =
r
p
2r
so that =
p
2t. +r and d =
p
2td..
\ (r, t) =
1
p
2
_
1
1
exp
_

1
2
.
2
_
(45)
exp
_
,
_
p
2t. +r
_
c
_
p
2t. +r
__
+
d.
Note that the integral is non-zero only when the second exponent is greater than
zero, that is, when ,
_p
2t. +r

c
_p
2t. +r

which is identical to .
r
p
2t
.
We can now break up the integral into two pieces
\ (r, t) =
1
p
2
_
1
r/
p
2r
exp
_

1
2
.
2
_
exp
_
,
_
p
2t. +r
__
d.

1
p
2
_
1
r/
p
2r
exp
_

1
2
.
2
_
exp
_
c
_
p
2t. +r
__
d.
= 1
1
1
2
.
Complete the square in the rst integral 1
1
. The exponent in the integrand is

1
2
.
2
+,
p
2t. +,r =
1
2
_
. ,
p
2t
_
2
+,r +,
2
t.
The rst integral becomes
1
1
= c
or+o
2
r
1
p
2
_
1
r/
p
2r
c

1
2
(:o
p
2r)
2
d..
Make the transformation j = . ,
p
2t so that the integral becomes
1
1
= c
or+o
2
r
1
p
2
_
1
r/
p
2ro
p
2r
c

1
2

2
dj
= c
or+o
2
r
_
1
_

r
p
2t
,
p
2t
__
= c
or+o
2
r

_
r
p
2t
+,
p
2t
_
.
The second integral is identical, except that , is replaced with c. Hence
1
2
= c
or+o
2
r

_
r
p
2t
+c
p
2t
_
.
Recall that r = ln
S
1
, / =
2:
c
2
, c =
1
2
(/ 1) =
:c
2
/2
c
2
, , =
1
2
(/ + 1) =
:+c
2
/2
c
2
,
and t =
1
2
o
2
(T t). Consequently, we have that
r
p
2t
+,
p
2t =
ln
S
1
+
_
r +
c
2
2
_
(T t)
o
p
T t
= d
1
15
and that
r
p
2t
+c
p
2t = d
1
o
p
T t = d
2
.
Hence the rst integral is
1
1
= exp
_
,r +,
2
t
_
(d
1
) .
The second integral is identical except that , is replaced by c and involves d
2
instead of d
1
1
2
= exp
_
cr +c
2
t
_
(d
2
) .
The solution is therefore
\(r, t) = 1
1
1
2
(46)
= c
or+o
2
r
(d
1
) c
or+o
2
r
(d
2
) .
The solution in Equation (46), expressed in terms of 1
1
and 1
2
, is the solution for
\(r, t). To obtain the solution for the call price \ (o
|
, t) we must use Equation
(44) and transform the solution in (46) back to \ . From (44) and (46)
\ (o, t) = 1c
oro
2
r
\(r, t) (47)
= 1c
oro
2
r
[1
1
1
2
] .
The rst integral in Equation (47) is
1c
oro
2
r
c
or+o
2
r
(d
1
) = 1c
(oo)r
(d
1
) (48)
= o(d
1
)
since , c = 1. The second integral in Equation (47) is
1c
oro
2
r
c
or+o
2
r
(d
2
) = 1c
(o
2
o
2
)r
(d
2
) (49)
= 1c
:(T|)
(d
2
)
since c
2
,
2
=
2:
c
2
. Combining the terms in Equations (48) and (49) produces
the Black-Scholes call price in Equation (1).
8 Black-Scholes From CAPM
8.1 The CAPM
The Capital Asset Pricing Model (CAPM) stipulates that the expected return
of a security i in excess of the risk-free rate is
1 [r
I
] r = ,
I
(1 [r
1
] r)
where r
I
is the return on the asset, r is the risk-free rate, r
1
is the return on
the market, and
,
I
=
Co [r
I
, r
1
]
\ ar [r
1
]
is the securitys beta.
16
8.2 The CAPM for the Assets
In the time increment dt the expected stock price return, 1 [r
S
dt] is 1
_
JSt
St
_
,
where o
|
follows the diusion in Equation (2). The expected return is therefore
1
_
do
|
o
|
_
= rdt +,
S
(1 [r
1
] r) dt. (50)
Similarly, the expected return on the derivative, 1 [r
\
dt] is 1
_
J\t
\t
_
, where \
|
follows the diusion in (3), is
1
_
d\
|
\
|
_
= rdt +,
\
(1 [r
1
] r) dt. (51)
8.3 The Black-Scholes PDE from the CAPM
Divide by \
|
on both sides of the second line of Equation (3) to obtain
d\
|
\
|
=
1
\
|
_
0\
0t
+
1
2
o
2
o
2
|
0
2
\
0o
2
_
dt +
0\
0o
do
|
o
|
o
|
\
|
,
which is
r
\
dt =
1
\
|
_
0\
0t
+
1
2
o
2
o
2
|
0
2
\
0o
2
_
dt +
0\
0o
o
|
\
|
r
S
dt. (52)
Drop dt from both sides and take the covariance of r
\
and r
1
, noting that only
the second term on the right-hand side of Equation (52) is stochastic
Co [r
\
, r
1
] =
0\
0o
o
|
\
|
Co [r
S
, r
1
] .
This implies the following relationship between the beta of the derivative, ,
\
,
and the beta of the stock, ,
S
,
\
=
_
0\
0o
o
|
\
|
_
,
S
.
This is Equation (15) of Black and Scholes [1]. Multiply Equation (51) by \
|
to obtain
1 [d\
|
] = r\
|
dt +\
|
,
\
(1 [r
1
] r) dt (53)
= r\
|
dt +
0\
0o
o
|
,
S
(1 [r
1
] r) dt.
This is Equation (18) of Black and Scholes [1]. Take expectations of the second
line of Equation (3), and substitute for 1 [do
|
] from Equation (50)
1 [d\
|
] =
0\
0t
dt +
0\
0o
[ro
|
dt +o
|
,
S
(1 [r
1
r]) dt] +
1
2
0
2
\
0o
2
o
2
o
2
dt. (54)
17
Equate Equations (53) and (54), and drop dt from both sides. The term
involving ,
S
cancels and we are left with
0\
0t
+ro
|
0\
0o
+
1
2
o
2
o
2
|
0
2
\
0o
2
r\
|
= 0. (55)
We recognize that Equation (55) is the Black-Scholes PDE in Equation (33).
Hence, we can obtain the Black-Scholes call price by appealing to the Feynman-
Kac theorem exactly as was done in Section (6.2) and solving the integral as in
Section (5).
9 Incorporating Dividends
The Black-Scholes call price in Equation (1) is for a call written on a non
dividend-paying stock. There are two ways to incorporate dividends into the
call price. The rst is by assuming the stock pays a continuous dividend yield
. The second is by assuming the stock pays dividends in lump sumps, "lumpy"
dividends.
9.1 Continuous Dividends
We assume that the dividend yield is constant so that the holder of the stock
receives an amount o
|
dt of dividend in the time increment dt. After the
dividend is paid out, the value of the stock drops by the dividend amount.
In other words, without the dividend yield, the value of the stock increases
by ro
|
dt, but with the dividend yield the stock increases by ro
|
dt o
|
dt =
(r ) o
|
dt. Hence, the expected return becomes r instead of r, which
implies that he risk-neutral process for o
|
follows Equation (14) but with drift
r instead of r
do
|
= (r ) o
|
dt +oo
|
d\
Q
|
. (56)
Following the same derivation in Section (3), Equation (56) has solution
o
T
= o
|
exp
__
r
1
2
o
2
_
t +o\
Q
r
_
where t = T t. Hence, o
T
follows the lognormal distribution with mean
o
|
c
(:j)r
and variance o
2
|
c
2(:j)r
_
c
c
2
r
1
_
. Proceeding exactly as in Equa-
tion (18), the call price is
C (o
|
, 1, T) = c
:r
1
S
T
(1) c
:r
[1 1(1)] . (57)
18
The conditional expectation 1
S
T
(1) from Equation (10) becomes
1
S
T
(1) = exp
_
lno
|
+
_
r
o
2
2
_
t +
o
2
t
2
_
(58)

_
_
ln1 + lno
|
+
_
r
c
2
2
_
t +o
2
t
o
p
t
_
_
= o
|
c
(:j)r
(d
1
)
with d
1
redened as
d
1
=
ln
St
1
+
_
r +
c
2
2
_
t
o
p
t
.
Using Equation (7), the second term in Equation (18) becomes
c
:r
1 [1 1(1)] = c
:r
1
_
_
1
_
_
ln1 lno
|

_
r
c
2
2
_
t
o
p
t
_
_
_
_
= c
:r
1(d
2
) (59)
with d
2
= d
1
o
p
t as before. Substituting Equations (58) and (59) into
Equation (57) produces the Black-Scholes price of a European call written on a
stock that pays continuous dividends
C (o
|
, 1, T) = o
|
c
jr
(d
1
) c
:r
1(d
2
).
Hence, the only modication is that the current value of the stock price is
decreased by c
jr
, and the return on the stock is decreased from r to r . All
other computations are identical.
9.2 Lumpy Dividends
To come. Same idea: the current value of the stock price is decreased by the
dividends, except not continuously.
References
[1] Black, F., and M. Scholes (1973). "The Pricing of Options and Corporate
Liabilities." Journal of Political Economy, Vol 81, No. 3, pp. 637-654.
[2] Hogg, R.V., and S. Klugman (1984). Loss Distributions. New York, NY:
John Wiley & Sons.
[3] Hull, J. (2008). Options, Futures, and Other Derivatives, Seventh Edition.
New York, NY: Prentice-Hall.
[4] Wilmott, P., Howison, S., and J. Dewynne (1995). The Mathematics of
Financial Derivatives: A Student Introduction. Cambridge, UK: Cambridge
University Press.
19