FIN 651: PDEs and Stochastic Calculus Solutions

↑ Student name ↑
Final Exam
December 14, 2012
Instructor: Bjørn Kjos-Hanssen

Disclaimer: It is essential to write legibly and show your work. If your work is absent or
illegible, and at the same time your answer is not perfectly correct, then no partial credit can
be awarded. Completely correct answers which are given without justification may receive
little or no credit.
During this exam, you are not permitted to use calculators, notes, or books, nor to
collaborate with others.

0
 Score:
Problem Score/4
1
2
3
4
5
6
Total/20
×5/2 =Total/50

The top 5 of 6 questions count.

0
 Score:

RT
Problem 1. Calculate the variance of the Ito integral 0 Wt dWt .
Hint #1: you may use without proof the Ito isometry, which says that
Z 2 !
T Z T
E ∆t dWt =E ∆2t dt.
0 0

Hint #2 (alternative to Hint #1): You may start by using Ito’s formula to calculate d(W 2 ).
R R
Solution: ∆ = W and E(Ws2 )ds = sds = T 2 /2.
Alternatively,
1
d(W 2 ) = 2W dW + 2dt
2
so Z
2
WT = 2W dW + T
so Z
1
W dW = (WT2 − T )
2
and Z
1 1 1
Var W dW = Var (WT2 − T ) = Var(WT2 ) = (E(WT4 ) − (E(WT2 ))2 )
2 4 4
using the fourth moment of N (0, σ 2 ) is 3σ 4 ,
1
= (3T 2 − T 2 ) = T 2 /2
4

1
 Score:

Problem 2. (a) Solve the stochastic differential equation

dXt = t dt − dWt

(in other words, find Xt ) by integrating both sides from 0 to T .

(b) Consider the stochastic differential equation

dYt = t dt + dWt

What can be said about Xt and Yt , assuming X0 = Y0 and the same Brownian motion Wt is
used to define both Xt and Yt ? Justify.

(i) Xt = Yt almost surely?

(ii) Xt and Yt are not equal almost surely, but they have the same probability distribution?

(iii) Xt and Yt are not equal almost surely, and they also do not have the same probability
distribution?

(c) Draw (sketch) what a typical random path of the solution Yt may look like in a (t, Yt )
coordinate system.

2
 Score:

Space for work on Problem 2. (a) Xt = X0 + 21 t2 − (Wt − W0 ) and W0 = 0. (b) (ii)

not equal almost surely, but have the same distribution since Wt and −Wt have the same
distribution. (c) Sketch 21 t2 and some random motion around it.

2
 Score:

Problem 3. The Black-Scholes-Merton partial differential equation for the price c(t, x) of
a derivative security at time t when the current stock price is x = St is

∂c ∂c 1 2 2 ∂ 2 c
+ rx + σ x = rc
∂t ∂x 2 ∂x2
(a) Show that
c(t, x) = ax + bert
is a solution for any constants a and b.

(b) Describe (in words, in finance terms) the derivative security with price process given
by the function in (a), in other words c(t, St ) = aSt + bert .

Solution: The option pays the value of a shares of the stock, plus b invested in a money
market account with continuously compounded interest rate r.

3
 Score:

Problem 4. Evaluate the stochastic integral

Z T
1/2
Wt dWt .
0

• Explanation: Find an equivalent expression that does not involve stochastic (Ito,
“dW ”) integrals, but only regular (but possibly random) integrals and functions.
3/2
• Hint: use Ito’s formula to expand d(Wt ).

Solution:  
3/2 3 1/2 1 3 1 −1/2
d(Wt ) = Wt dWt + W dt
2 22 2
Z Z
3/2 3 1/2 3
Wt = W dWt + W −1/2 dt
2 t 8
Z Z
2 3/2 1/2 1
Wt = Wt dWt + W −1/2 dt
3 4
Z T
2 3/2 1 T −1/2
Z
1/2
Wt dWt = WT − W dt
0 3 4 0

4
 Score:

Problem 5. Recall that a stochastic process Xt is a Markov process if for each Borel
measurable function f , there is a function g such that if s < t then
Es (f (Xt )) = g(Xs ).
In other words, our estimate at time s of the distribution of Xt is determined by the value
Xs and does not depend on Xu , u < s. Here Es (·) is what the textbook calls E(· | Fs ).
(a) Simplify the expression:
E1 (W2 )
where Wt , t ≥ 0, is Brownian motion. E1 (W2 ) = W1 since Brownian motion is a
martingale and 1 < 2.
(b) Calculate Es (Xt ) and Es (Yt ) for the following processes.
Xt = Wt2 − t
Z t
Yt = Ws ds
0

(c) Can you rule out that one or more of them are Markov processes based on your answer
to (b)?
Es [Xt ] = Es [Wt2 ] − t = Es [((Wt − Ws ) + Ws )2 ] − t
= Es [(Wt − Ws )2 + 2(Wt − Ws )Ws + Ws2 ] − t
= t − s + 0 + Ws2 − t = Ws2 − s = Xs
so we cannot rule out that Xt is a Markov process (and by the way, we just showed that Xt
is a martingale). And Z t 
Es [Yt ] = Es Wu du
0
Z t Z s 
= Es Wu du + Wu du
s 0
Z t Z s
= (Es Wu )du + Wu du
s 0
Z t Z s
= Ws du + Wu du
s 0
Z s
= (t − s)Ws + Wu du = (t − s)Ws + Ys
0
which is not determined by Ys , so Yt is not a Markov process. Note: if you are unable
to do the formal calculation in part (b), you can try to guess or give an informal reasoning
for what the answer should be. Also in part (c) you can try to explain informally whether
the answers in (b) depend only on Xs , Ys or whether they depend also on Xu , u < s or Yu ,
u < s.

5
 Score:

Space for work on Problem 5.

5
 Score:

Problem 6. Let W be any normal random variable with mean 0 and variance 1. Calculate
the moment generating function

MW (u) = E(euW ).

Hint: This will involve “completing the square”.

Solution:
Z ∞   
uW ux 1 1 2
E(e ) = e √ exp − x dx
−∞ 2π 2
Z ∞  
1 1 2
=√ exp ux − x dx
2π −∞ 2
Z ∞  
1 1 2 
=√ exp − x − 2ux dx
2π −∞ 2
Z ∞  
1 1 2 2 2

=√ exp − x − 2ux + u − u dx
2π −∞ 2
Z ∞  
1 1 2 2

=√ exp − (x − u) − u dx
2π −∞ 2
Z ∞  
u2 /2 1 1 2

=e √ exp − (x − u) dx
2π −∞ 2
2 /2
= eu