Chapter 20

Brownian Motion and Its Lemma

Question 20.1.
If y = ln (S) then S = e
y
and dy =
_
(S,t )
S

(S,t )
2
2S
2
_
dt +
(S,t )
S
dZ
t
,
a) dy =
_

e
y


2
2e
2y
_
dt +

e
y
dZ
t
.
b) dy =
_
a
e
y


2
2e
2y
_
dt +

e
y
dZ
t
.
c) dy =
_


2
2
_
dt +dZ
t
.
Question 20.2.
If y = S
2
then S =

y and dy =
_
2S (S, t ) + (S, t )
2
_
dt +2S (S, t ) dZ
t
where (S, t ) is
the drift of S and (S, t ) is the volatility of S. For the three specications:
a) dy =
_
2

y +
2
_
dt +2

ydZ
t
.
b)
dy =
_
2

y
_
a

y
_
+
2
_
dt +2

ydZ
t
(1)
=
_
2a

y 2y +
2
_
dt +2

ydZ
t
. (2)
c) dy =
_
2 +
2
_
ydt +2ydZ
t
.
Question 20.3.
If y = 1/S then S = 1/y and dy =
_
S
2
(S, t ) +S
3
(S, t )
2
_
dt S
2
(S, t ) dZ
t
,
a) dy =
_
y
2
+
2
y
3
_
dt y
2
dZ
t
.
b) dy =
_

_
ay
2
y
_
+
2
y
3
_
dt y
2
dZ
t
.
c) dy =
_
+
2
_
ydt ydZ
t
.
257
Part 5 Advanced Pricing Theory
Question 20.4.
If y =

S then S = y
2
and
dy =
_
1
2
S
1/2
(S, t )
1
8
S
3/2
(S, t )
2
_
dt +
1
2
S
1/2
(S, t ) dZ
t
(3)
=
_
1
2y
(S, t )
1
8y
3
(S, t )
2
_
dt +
1
2y
(S, t ) dZ
t
(4)
a) dy =
_

2y


2
8y
3
_
dt +

2y
dZ
t
.
b) dy =
_
a
2y


2y
2


2
8y
3
_
dt +

2y
dZ
t
.
c) dy =
_

2


2
8
_
ydt +

2
ydZ
t
.
Question 20.5.
Let y = S
2
Q
0.5
, then
dy
y
=
_
2 (
S

S
) +

Q

Q
2
+
2
S


2
Q
8
+
S

Q
_
dt +2
S
dZ
S
+b
Q
dZ
Q
. (5)
Question 20.6.
If y = ln (SQ) = ln (S) +ln (Q) then
dy = d ln (S) +d ln (Q) (6)
=
_

2
S
/2 +
Q

2
Q
/2
_
dt +
S
dZ
S
+
Q
dZ
Q
. (7)
Question 20.7.
With = 0, the prepaid forward price for S
a
1
is
F
P
0,1
_
S
a
_
= S
a
0
exp
_
(a 1) r +
1
2
a (a 1)
2
_
. (8)
a) If a = 2, F
P
0,1
_
S
2
_
= 100
2
exp
_
.06 +.4
2
_
= 12461.
b) If a = .5, F
P
0,1
_
S
.5
_
= 10 exp
_
.03
.4
2
8
_
= 9.5123.
c) If a = 2, F
P
0,1
_
S
2
_
= 100
2
exp
_
.18 +3
_
.4
2
__
= 1.349 9 10
4
.
258
Chapter 20 Brownian Motion and Its Lemma
Question 20.8.
Since the process y = S
a
Q
b
follows geometric Brownian motion, i.e. dy =
y
ydt +
y
ydZ
y
the
price of the claims will be e
r
E

(y
1
) = y
0
e
(
y
r)
. We use Itos lemma, as in equation (20.38),
with = 0 and
S
=
Q
= r to arrive at the drift

y
= ar +br +
1
2
a (a 1)
2
S
+
1
2
b (b 1)
2
Q
+ab
S

Q
(9)
= .06 (a +b) +
.4
2
2
a (a 1) +
.2
2
2
b (b 1) .3 (.4) (.2) ab. (10)
a) Since a = b = 1, y
0
= 10000 and
y
= .12 .024 = .096 hence the claim is worth
10000e
.096.06
= 10366.56.
b) Since a = 1 and b = 1, y
0
= 1 and
y
= .2
2
+.024 = .064 hence the claim is worth
e
.064.06
= 1.004.
c) Since a = 1/2andb = 1/2, y
0
= 100and
y
= .029hence the claimis worth100e
.029.06
=
96.948.
d) Since a = 1 and b = 1, y
0
= 1/10000 and
y
= .056 hence the claim is worth
_
e
.056.06
_
/10000 = 9.960 1 10
5
.
e) Since a = 2 and b = 1, y
0
= 1000000 and
y
= .292 hence the claim is worth
1000000e
.292.06
= 1.2612 million.
Question 20.9.
It is obvious if t = 0 the proposed solution will be equal to X
0
. It is helpful to rewrite the solution
as
X
t
= e
t
_
X
0
+a
_
e
t
1
_
+
_
t
0
e
s
dZ
s
_
= e
t
Y
t
(11)
where dY
t
= ae
t
dt +e
t
dZ
t
. Since e
t
is deterministic,
dX
t
=
_
e
t
dt
_
Y
t
+e
t
dY
t
= (a X
t
) dt +dZ
t
. (12)
Question 20.10.
Note that if V (S) satises the given equation, then
E

(dV) =
_
(r ) SV
S
+
1
2

2
S
2
V
SS
_
dt = rVdt. (13)
259
Part 5 Advanced Pricing Theory
Since V (S) = kS
h
1
where a is constant, showing y = S
a
satises E

(dy) = rydt when a = h

1
is
sufcient (i.e. the constant term is irrelevant). Using Itos lemma,
E

(dy) = aS
a1
(r ) S +
1
2
a (a 1) S
a2

2
S
2
(14)
=
_
a (r ) y +a (a 1)

2
2
y
_
dt. (15)
If E

(dy) = rydt then a must satisfy

a (r ) +a (a 1)

2
2
= r. (16)
The two solutions are h
1
and h
2
as given (12.11) and (12.12) which one can verify directly.
Question 20.11.
As discussed in the hint, consider a strategy of 1 unit in Q, Q
i
/ (S
i

i
) for both i = 1 and 2. Let
I
t
be the amount of money in the risk free asset. The value of the portfolio is
V
t
= Q
t
(1
1
/
1

2
/
2
) +I
t
. (17)
The expected change in the value is
dV
t
= dQ
t

Q
t

1
S
1t

1
dS
1t

Q
t

2
S
2t

2
dS
2t
+rI
t
dt (18)
=
__

2
_
Q+rI
_
dt (19)
+(
1
Q
1
Q) dZ
1
+(
2
Q
2
Q) dZ
2
(20)
=
__

2
_
Q+rI
_
dt. (21)
Since this portfolio requires zero investment and there is no risk, the drift and V
t
must be zero.
Hence
_

2
_
Q+r
_
1

1

2
_
Q = 0. (22)
Rearranging

Q
r =

1

1
(
1
r) +

2

2
(
2
r) . (23)
260
Chapter 20 Brownian Motion and Its Lemma
Question 20.12.
We must try to nd a position in S and Q that eliminates risk. Let us buy one unit of S and let be
the position in Q. Let I
t
be our bond investment. We have V
t
= S
t
+
t
Q
t
+I
t
with V
0
= 0. Since
this strategy must be self nancing,
dV =
_

S
S +
Q
Q+rI
_
dt +(
S
S Q) dZ (24)
hence we will set =
S
S/ (Q). This will make our zero cost, self nancing strategy riskless.
Hence the drift and the value must be zero. Mathematically, if V
t
= 0 then I = S

S
S
Q
Q. The
drift being zero implies

S
S +

S
S
Q

Q
Qr
_
S +

S
S

_
= 0. (25)
Dividing both sides by S and simplifying leads to

Q
= r

S
r

S
. (26)
Since Q is negatively related to Z, if S has a positive risk premium then Q will negative risk
premium.
Question 20.13.
In the following we dene y
t
= S
a
t
Q
b
t
.
a) From equation (20.38), the (real world) expected value of y
T
is E (y
T
) = y
0
e
mT
where
m = a (
S

S
) +b
_

Q
_
+
a (a 1)
2
S
2
+
b (b 1)
2
Q
2
+ab
S

Q
(27)
is the real world capital gain. Given a (real world) expected return , the value of the claim is
e
T
E (y
T
) = y
0
e
(m)T
. Using Itos lemma and problem 20.11,
1
= a
S
and
2
= b
Q
. We
then have
= r +a (
S
r) +b
_

Q
r
_
(28)
and the value of the claim being
y
0
e
(m)T
= S
0
Q
0
e
rT
e
hT
(29)
where h = a (r
S
) +b
_
r
Q
_
+
1
2
a (a 1)
2
S
+
1
2
b (b 1)
2
Q
+ab
S

Q
. Note this agrees
with e
rT
E

(y
T
).
261
Part 5 Advanced Pricing Theory
b) The expected return of y is and the actual expected capital gain is m. The lease rate of y
would have to be the difference

= m which equals

= r(1 a b) +a
S
+b
Q

1
2
a(a 1)
2
S

1
2
b(b 1)
2
Q
ab
S

Q
.
The prepaid forward price must be y
0
e

T
= y
0
e
(m)T
which agrees with our previous answer.
We can rewrite it in an informative way. The forward price for a security paying S
a
is
F
0,T
(S
a
) = S
a
e
_
a(r
S
)+
1
2
a(a1)
2
S
_
T
.
The forward price for Q
b
is
F
0,T
(Q
b
) = Q
b
e
_
b(r
Q
)+
1
2
b(b1)
2
Q
_
T
.
Thus, we can rewrite the prepaid forward price as
F
P
0,T
(S
a
Q
b
) = e
rT
F
0,T
(S
a
)F
0,T
(Q
b
)e
ab
S

Q
T
. (31)
The expression on the right is the product of the forward prices times a factor that accounts for the
covariance between the two assets. The discount factor converts it into a prepaid forward price.
Question 20.14.
As mentioned in the problem, dZ appears in both dS and dQ. One can think of dQas an alternative
model for the stock (with dS being the standard geometric Brownian motion).
a) If there were no jumps, dQwould also be geometric Brownian motion. Since it has the same
risk component, dZ,
Q
must equal . If we thought of Q as another traded asset, this naturally
follows from no arbitrage.
b) If Y
1
> 1 then there are only positive jumps. We would therefore expect
Q
< to com-
pensate for this. Mathematically, dQ/QdS/S =
_

_
dt +dq
1
. If a jump occurs, dq
1
=
Y
1
1 > 0; if
Q
we could buy Q and short S. The only risk we have is jump risk but this will
always be good news for our portfolio. In order to avoid this arbitrage
Q
must be less than .
If we use a weaker assumption k
1
= E (Y
1
1) > 0 and we assume the returns to S and Q should
be the same (this makes sense if we are looking at Q as an alternative model instead of another
stock) then we arrive at a similar result. The expected return to Q is
Q
+
1
k
1
; setting this equal
to implies
Q
=
1
k
1
> 0.
262
Chapter 20 Brownian Motion and Its Lemma
c) Let

be the expected return of Q. Note that

Q
is not the expected return, it is the expected
return conditional on no jumps occurring. We have the following relationship,

= E
_
dQ
Q
_
/dt =
Q
+k
1

1
+k
2

2
(32)
where k
i
= E (Y
i
1). Hence
Q
=

k
1

1
k
2

2
. If

= (i.e. Q and S have the same

expected return) then
Q
= k
1

1
+k
2

2
. The sign of which could be positive or negative if
there are no restriction on k
1
and k
2
.
263