Advanced Financial Models Michael Tehranchi

Example sheet 1 - Michaelmas 2014

Problem 1. Consider a two-asset model with prices given by

(P 1 , P 2 ) (3, 9)
w;
1/4
www
ww
ww
(4, 6) / (6, 8)
GG 1/4
GG
GG
1/2 GG#
(6, 4)
Is there arbitrage in this market? If not, find all martingale deflators.
Solution 1. Let H ∈ R2 be a candidate absolute arbitrage, such that H · P0 ≤ 0 ≤ H · P1 .
We will show H · P0 = 0 = H · P1 and hence there is no arbitrage.
We will label the possible outcomes Ω = {A, B, C}, so that P1 (A) = (3, 9) and P{A} =
1/4, etc.
Let H = (h, k). The inequalities become
(0): 4h + 6k ≤ 0
(A): 3h + 9k ≥ 0
(B): 6h + 8k ≥ 0
(C): 6h + 4k ≥ 0
By subtracting appropriate multiples of inequality (0) from the others to eliminate h, we
have
(A0 ): (9 − 34 × 6)k = 9k/2 ≥ 0
(B 0 ): (8 − 46 × 6)k = −k ≥ 0
(C 0 ): (4 − 46 × 6)k = −5k ≥ 0
Inequalities (A0 ) and (B 0 ) together imply k = 0. Plugging this into (0) and (A) above imply
h = 0, as desired.
A martingale deflator in one period corresponds to a positive random variable Y such that
E(Y P1 ) = P0 . Letting Y (A) = a, Y (B) = b and Y (C) = c we have
1 1 1
(3a) + (6b) + (6c) = 4
4 4 2
1 1 1
(9a) + (8b) + (4c) = 6
4 4 2
All positive solutions of these two equations in three unknowns can be written as
   
40 6 8 12
(a, b, c) = u , 0, + (1 − u) , , 0 , 0 < u < 1.
21 7 15 5
Problem 2. Consider the one-asset model with price process Pt = 1{t<τ } where τ is a
random variable taking values in {1, 2, . . .}.
(a) Suppose that there is a non-random time T > 0 such that τ ≤ T almost surely. Show
that there is an arbitrage.
1
(b) Suppose P(τ = k) = (1 − p)k−1 p for a constant 0 < p < 1, so τ has the geometric
distribution. Show that there is no arbitrage. Can you find a martingale deflator?
Solution 2. (a) In words, an arbitrage is to immediately sell the asset short and consume
the revenue. Then at time T , buy the asset back. This can always be done since PT = 0
a.s. by assumption. In notation, let H0 = 0 = HT +1 , and Ht = −1 for 1 ≤ t ≤ T . The
corresponding consumption stream is c0 = 1 and ct = 0 for 1 ≤ t ≤ T .
Alternatively, let Y be a candidate martingale deflator. In particular, suppose that P Y is
a martingale. Note that P0 = 1 and PT = 0 a.s., so that
Y0 = Y0 P0 = E(YT PT ) = 0.
Hence the process Y is not strictly positive, and hence not a martingale deflator. By the
fundamental theorem of asset pricing, there must exist an arbitrage.
(b) Assuming that the filtration is generated by P , we have
E(Pt |Ft−1 ) = P(t < τ |t − 1 < τ )1{t−1<τ }
= (1 − p)Pt−1 .

Hence Yt = (1 − p)−t is a martingale deflator. By the 1FTAP, there is no arbitrage.

Problem 3. Consider a three-asset model with asset 1 cash so that P01 = P11 = 1 and assets
2 and 3 given by
(P 2 , P 3 ) (9, 8)
;
ww
ww
1/3
ww
ww
(6, 7)
GG
GG
GG
2/3 GG#
(3, 5)
Is there arbitrage in this market? If not, find all equivalent martingale measures with respect
to the cash numéraire.
Solution 3. A candidate arbitrage H = (g, h, k) would satisfy
g + 6h + 7k ≤ 0
g + 9h + 8k ≥ 0
g + 3h + 5k ≥ 0
The set of solutions to the above system of inequailities is given by
(g, h, k) = u(15, 1, −3) + v(9, 2, −3) + w(7, 1, −2), u, v, w ≥ 0.
In particular, yes, there are arbitrages. For instance taking H = (15, 1, −3) yields H · P0 =
0 ≤ H · P1 a.s. and P(H · P1 = 3 > 0) = 2/3 > 0.
By the 1FTAP, there are no equivalent martingale measures. But let’s see this directly in
this example: Let Ω = {A, B} and let Q be a candidate equivalent martingale measure with
2
Q{A} = a, Q{B} = b. Then we have three equations and two unknowns:
a+b=1
9a + 3b = 6
8a + 5b = 7.

The above system has no solution.

Problem 4. Let (Ω, F, P) be a probability space, and let W be a random vector with the
d-dimensional normal Nd (0, I) distribution, where I is the d × d identity matrix. Fix a
constant vector α ∈ Rd and define an equivalent measure Q on (Ω, F) by the density
dQ 2
= eα·W −|α| /2 .
dP
Prove that the random variable Ŵ = W − α has the Nd (0, I) distribution under Q.
Solution 4. Let A ⊆ Rd be a (Borel) set. The following computation proves the claim

Q(Ŵ ∈ A) = EP [eα·W −|α| /2 1{Ŵ −α∈A} ]

2

Z −|w|2 /2
α·w−|α|2 /2 e
= e dw
A+α (2π)d/2
2
e−|w−α| /2
Z
= d/2
dw
A+α (2π)
Z −|u|2 /2
e
= d/2
du
A (2π)
= P(W ∈ A)
where the second to last line follows from the change of variables u = w − α.
Alternatively, since moment generating functions, when finite on an open set, determine
the distribution of a random variable, we could also argue as follows:
2 /2+θ·(W −α)
EQ [eθ·Ŵ ] = EP [eα·W −|α| ]
−θ·α−|α|2 /2
=e EP [e(α+θ)·W ]
2 /2+|α+θ|2 /2
= e−θ·α−|α|
2 /2
= e|θ|

and we recognise the moment generating function of the Nd (0, I) distribution.

Problem 5. Consider a d+1 asset model P = (1, S), where the first asset is cash while assets
2, . . . , d + 1 have time-0 price S0 ∈ Rd and time-1 prices S1 with the Nd (µ, V ) distribution,
where µ ∈ Rd and V is a positive semi-definite d × d matrix. Find necessary and sufficient
conditions on the data S0 , µ and V such that there is no arbitrage. When there is no
arbitrage, use the previous problem to find a martingale deflator.
3
Solution 5. Let H = (φ, π) be a candidate arbitrage. Since there is a numéraire assets, we
set φ + π · S0 = 0 and ask that φ + π · S1 ≥ 0 a.s. Note that
π · (S1 − S0 ) ∼ N (π · (µ − S0 ), π · V π).
The support of a normal random variable is all of R if its variance is positive. If we want
π · (S1 − S0 ) ≥ 0 a.s.
then we must have π · V π = |V 1/2 π|2 = 0. Now if V 1/2 π = 0 then
π · (S1 − S0 ) = π · (µ − S0 ),
a constant. Therefore, there is no arbitrage if and only if V 1/2 π = 0 ⇒ π · (µ − S0 ) = 0.
From linear algebra, an equivalent formulation is
no arbitrage ⇔ µ − S0 ∈ Range(V ).
When there is no-arbitrage, we can use the previous to exhibit an equivalent martingale
measure. Claim: if there is no arbitrage, there exists a vector α ∈ Rd such that
V 1/2 α = µ − S0 .
(The case when V is non-singular is easy, since α = V −1/2 (µ − S0 ). )
Now, by assumption S1 = V 1/2 W + µ, where W has the Nd (0, I) distribution. Let Ŵ =
W + α and
dQ 2
= e−α·W −|α| /2 .
dP
Note that Ŵ has the Nd (0, I) distribution under Q, so that
EQ (S1 ) = EQ [V 1/2 Ŵ + S0 ] = S0 ,
2 /2
and hence Q is an equivalent martingale measure, and hence the density Y = e−α·W −|α| is
a martingale deflator.
Problem 6. (Stiemke’s theorem) Let A be a m × n matrix. Prove that exactly one of the
following statements is true:
• There exists an x ∈ Rn with xi > 0 for all i = 1, . . . , n such that Ax = 0.
• There exists a y ∈ Rm with (AT y)i ≥ 0 for all i = 1, . . . , n such that AT y 6= 0.
What does this have to do with finance?
Solution 6. Let Ω = {1, . . . , n}, F the set of all subsets of Ω, and P({j}) = 1/n for all
j ∈ Ω. Let S0 = 0, and define a random variable by S1 : Ω → Rm by the S1i (j) = ai,j where
A = (ai,j )i,j . Consider a (1 + m)-asset market model with prices P = (1, S).
Since there is a numéraire, an arbitrage is a portfolio (φ, π) ∈ R1+m such that φ+φ·S0 = 0
and φ + π · S1 ≥ 0 almost surely, and P(φ + π · S1 > 0) > 0. Note that since S0 = 0 then
φ = 0 and that
m
X
π · S1 (j) = π i aij = (AT π)j .
i=1

In particular, an arbitrage can be identified with a vector π ∈ Rm with (AT π)j ≥ 0 for all
j = 1, . . . , n such that AT π 6= 0.
4
 An equivalent martingale measure (with respect to the cash numéraire) is a probability
measure Q on {1, . . . , n} such that Q{j} > 0 for all j and that EQ (Y S1 ) = 0. Letting
qj = Q{j} we have
n
X
EQ S1i = aij qj = (Aq)i
j=1
n
Pn martingale can be identified with a q ∈ R with qj > 0 for all
In this context an equivalent
j such that Aq = 0 and j=1 qj = 1.
The first fundamental theorem of asset pricing is that there exists an equivalent martingale
measure if and only if no arbitrage exists.

We have now seen one proof of the 1FTAP in lecture. For the sake of developing intuition,
here’s yet another one: As before, the easy direction is to show that the existence of an
equivalent martingale measure implies no arbitrage: Suppose that there exists an x ∈ Rn
with xi > 0 for all i = 1, . . . , n such that Ax = 0. If there exists a y ∈ Rm with (AT y)i ≥ 0
for all i = 1, . . . , n then x · (AT y) = (Ax) · y = 0. Hence AT y = 0.
Now we prove the harder direction using a version of the separating hyperplane theorem.
Suppose that there is no arbitrage: If there exists a y ∈ Rm with (AT y)i ≥ 0 for all
i = 1, . . . , n then AT y = 0. Let
S = {AT y : y ∈ Rm } ⊆ Rn
and let ( )
n
X
n
C= u ∈ R : ui ≥ 0 for all i = 1, . . . , n and ui = 1 ⊂ Rn .
i=1
By assumption, the subspace S and the convex compact set C are disjoint. Indeed, if v ∈ S,
= AT y for some y. If vi ≥ 0 for all i, then v = 0 by the no-arbitrage assumption.
then v P
Hence i vi = 0 6= 1 and v is not in C.
The situation is illustrated by the figure. A version of the separating hyperplane theorem,
says there exists a vector λ ∈ Rn such that
λ·v =0 for all v ∈ S
λ·u>0 for all u ∈ C.
(Try proving this!) We will be done once we show that λj > 0 for all j. So fix a j ∈ {1, . . . , n}
and let e ∈ Rn be given by ej = 1 and ei = 0 for all i 6= j. Then e ∈ C and λ · e = λj > 0 as
desired.
Problem 7. Consider a discrete time model with n asset with prices (Pt )t≥0 and dividends
(δt )t≥1 . Explain why an appropriate self-financing condition is
Ht · (Pt + δt ) = Ht+1 · Pt + ct .
Let Z be a positive process such that the process
t
!
X
Zt Pt + Zs δs
s=1 t≥0

is a martingale. Show that there is no arbitrage in this market.

5
Solution 7. At the end of the t-th trading period, the investor’s wealth held in a given asset
is the sum of the (ex dividend) price of the asset and the dividend payout. In total then,
the wealth is given by the formula
Ht · (Pt + δt ).
Now, the investor uses this money to consume and to rebalance his portfolio for the next
period so that
Ht · (Pt + δt ) = Ht+1 · Pt + ct+1
as required.
Let
Xt
Xt = Zt Pt + Zs δs
s=1
As in the lecture, let
t
X
Mt = Zt Ht+1 · Pt + Zs cs
s=0
t
X
= Z0 H0 · P0 + Hs (Xs − Xs−1 ).
s=1
Assuming X is a martingale implies that M is a local martingale. Now if HT +1 we have that
MT ≥ 0 a.s. so that M is true martingale. If H0 = 0 then E(MT ) = M0 = 0. Hence MT = 0
almost surely, and hence ct = 0 a.s. for all t. In particular, there is no arbitrage.
Problem 8. Consider a discrete time model with an asset with positive prices (Pt )t≥0 and
non-negative dividends (δt )t≥1 . Show that there is
 a self-financing trading strategy with cor-
Qt δs
responding wealth process Qt = Pt s=1 1 + Ps . What is the financial significance of this
6
 Pt

process? Let Z be a positive adapted process. Show that the process Zt Pt + s=1 Zs δs t≥0
is a martingale if and only if (Zt Qt )t≥0 is.
Solution 8. Let
Qt−1
Ht =
Pt−1
for t ≥ 1. Notice that H is predictable and self-financing, since
Ht (Pt + δt ) = Qt = Ht+1 Pt .
The strategy H consists of holding one share of the asset initially, and reinvesting the divi-
dend payments.
Note the identity
Zt+1 Qt+1 − Zt Qt = Ht+1 (Zt+1 Pt+1 + Zt+1 δt+1 − Zt Pt ).
Letting
t
X
Mt = Zt Pt + Zs δs
s=1
we see that
t
X
Zt Qt = Z0 P0 + Hs (Ms − Ms−1 ).
s=1
We recognise ZQ as the martingale transform of the predictable process H with respect to
M . If M is a martingale, then ZQ is a local martingale. But non-negative local martingales
in discrete time are true martingales.
Similarly, writing
t
X 1
Mt = Z0 P0 + (Zs Qs − Zs−1 Qs−1 )
s=1
H s

we see that if ZQ is a martingale, then M is martingale.

Problem 9. In discrete-time models, an asset is called risk-free iff its price process is pre-
dictable. Suppose that asset 1 and asset 2 are risk-free, that the market with prices (P 1 , P 2 )
has no arbitrage, and that P01 = P02 . Show that Pt1 = Pt2 a.s. for all t ≥ 0.
Solution 9. We will construct a candidate arbitrage. Let τ = inf{t ≥ 0 : Pt1 6= Pt2 }. Since
P 1 and P 2 are predictable, the event {τ = t} is Ft−1 measurable. Now let
Ht = sign(Pτ1 − Pτ2 )1{t=τ } (−1, 1)
and ct = |Pτ1 − Pτ2 |1{t=τ } Note that the process H is predictable and satisfy the self-financing
condition. Hence the H is an arbitrage if and only if P(τ < ∞) > 0.
Problem 10. A bond1 is an asset that pays a fixed amount (the principal) at a fixed future
time (the maturity date). Suppose that assets 1, . . . , n are bonds, each of principal amount
£1, where asset i matures at time i. Show that if there is no arbitrage in the market,
then each bond is a numéraire asset. Let H be the investment strategy of holding bond of
1Many real world bonds pay the bondholder a fixed amount on a regular basis, every three months for
instance, before the maturity date. These payments are called ‘coupons.’ The bonds we are now considering
here are actually zero-coupon bonds.
7
maturity t during the interval (t − 1, t]. Show that the corresponding wealth process (called
the money-market account) is risk-free.
Solution 10. Let PtT be the price at time t of a bond with maturity T , where 0 ≤ t ≤ T .
Note that by definition PTT = 1 a.s. for each T ≥ 1.
First we show that if there is no arbitrage then PtT > 0 a.s. for all 0 ≤ t ≤ T . The
primal approach is to construct an arbitrage if P(PtT ≤ 0) > 0 for some 0 ≤ t ≤ T . The
dual approach is to appeal to the 1FTAP: If there is no arbitrage there exists a state price
density process Y such that
E(YT |Ft )
PtT = .
Yt
Note that the right hand of the equation is strictly positive.
Let rt = P t1 − 1, and let
t−1
t
Y
Bt = (1 + rs ).
s=1

Note that B is predictable since rt is Ft−1 -measurable. Let Hti = 0 if i 6= t and Htt = Bt , i.e.
Ht = (0, . . . , 0, Bt , 0, . . . , 0).
(In the notation above, we are pretending that the matured bonds continue to exist. But
since we are not trading in them, their prices are irrelevant.) Then H is predictable and
self-financing since
Ht+1 · Pt = Bt+1 Ptt+1
= Bt
= Ht · Pt .
and its corresponding wealth process is B, the money market account.
Problem 11. Let X and Y be martingales (with respect to the same filtration). Show that
if XT = YT almost surely for some non-random T > 0, then Xt = Yt almost surely for all
0 ≤ t ≤ T.
Solution 11. If XT = YT almost surely, then
Xt = E(XT |Ft ) = E(YT |Ft ) = Yt
for all 0 ≤ t ≤ T .
Problem 12. (Bayes’s formula) Let P and Q be equivalent probability measures defined on
(Ω, F) with density Z = dQ
dP
. Let G ⊆ F be a sigma-field. Prove the identity:
EP (ZX|G)
EQ (X|G) =
EP (Z|G)
for each bounded random variable X.
P
Solution 12. Let Y = EEP(ZX|G)
(Z|G)
. Note that Y is G-measurable. Hence, we need only verify
the equation E (Y 1G ) = E (X 1G ) for all G ∈ G; equivalently, we must verify EP (ZY 1G ) =
Q Q

8
EP (ZX 1G ) for all G ∈ G.
EP (1G Y Z) = EP [E(1G Y Z|G)] tower property
= EP [1G Y E(Z|G)] taking out what’s known
= EP [1G EP (XZ|G)]
= EP [EP (1G XZ|G)] pulling in what’s known
= EP (1G XZ) tower property
Problem 13. Consider a single period market with two assets. The first asset is a riskless
bond with prices B0 = 1 and B1 = 1 + r for a constant r. The second asset is a stock with
prices (St )t∈{0,1} .
Let (φ∗ , π ∗ ) be the optimal solution to the problem
maximise E U (φB1 + πS1 ) subject to φB0 + πS0 = x
for a given concave increasing utility function U . Prove that the investor is holds a non-
negative number of shares of the stock if
E S1 > (1 + r)S0
Does this agree with your intuition?
Solution 13. We can solve for φ and rephrase the problem as
maximise E V (πξ)
where
V (y) = U [(1 + r)x + y]
and ξ = S1 − (1 + r)S0 .
Let π ∗ be the choice of π ∈ R that maximizes E V (πξ). On the one hand, we have
V (0) ≤ E V (π ∗ ξ)
by the optimality of π = π ∗ as compared to π = 0.
On the other hand, we have
E V (π ∗ ξ) ≤ V (π ∗ E ξ)
by Jensen’s inequality. Since U , and hence V , is increasing, we have
0 ≤ π ∗ Eξ.
If E ξ > 0 then π ∗ ≥ 0 as desired.

Another way of looking at this problem is to consider the function π 7→ E[V (πξ)]. Since
this function is maximized at π ∗ , the gradient E[V 0 (πξ)ξ] is non-negative if and only if
π ≤ π ∗ . Letting π = 0 above and noting that V 0 (0) = U 0 [(1 + r)x] is positive and non-
random completes the argument.
Problem 14. (Tower property of conditional expectation) Let X and Y be identically
distributed random variables taking values in the set {2n : n ≥ 0} such that X/Y ∈ {1/2, 2}
almost surely and
1
P(X = 2n , Y = 2n+1 ) = 2−n = P(X = 2n+1 , Y = 2n ) for n ≥ 0.
4
9
 1
(a) Show that P(X = 1) = 4
and

3
P(X = 2n ) = 2−n for n ≥ 1.
4
(b) Show that P(Y = 2|X = 1) = 1 and
1
P(Y = 2n+1 |X = 2n ) = = 1 − P(Y = 2n−1 |X = 2n ) for n ≥ 1.
3
(c) Let Z = Y − X. Show that E(Z|X = 1) = 1 and

E(Z|X = 2n ) = 0 for n ≥ 1.

(d) From part (c) we have E(Z|X) = 1{X=1} and hence

1
E(Z) = E[E(Z|X)] = > 0.
4
However, by symmetry we also have E(Z|Y ) = −1{Y =1} and

1
E(Z) = E[E(Z|Y )] = − < 0.
4
What has gone wrong?!

Solution 14. (a)

P(X = 2n ) = P(X = 2n , Y = 2n+1 ) + P(X = 2n , Y = 2n−1 )

1 1
= 2−n + 2−(n−1) 1{n≥1}
4 4
1
= 2−n (1 + 21{n≥1} )
4
(b) By Bayes’ formula

P(X = 2n , Y = 2n+1 )
P(Y = 2n+1 |X = 2n ) =
P(X = 2n )
1
=
1 + 21{n≥1}

by part (a).
(c) Using the formula from part (b) yields

E[Z|X = 2n ] = 2n+1 P(Y = 2n+1 |X = 2n ) + 2n−1 P(Y = 2n−1 |X = 2n ) − 2n

1 − 1{n≥1}
 
n
=2
1 + 21{n≥1}
= 1{n=0}
10
(d) The problem is that Z is not integrable. Indeed,
X∞
+
E(Z ) = (2n+1 − 2n )P(X = 2n , Y = 2n+1 )
n=0
∞
X 1
=
n=0
2
=∞
Notice that the conditional expectation (given an event)
E(Z 1{X=2n } )
E(Z|X = 2n ) =
P(X = 2n )
is defined, but the conditional expectation (given a sigma-field) E(Z|X) is not defined2.

2In fact, it is possible to define a generalised notion of conditional expectation so that E(Z|X) = 1{X=1} .
But this definition is fickle, and in particular, it would not obey the tower propery of conditional expectation.
11