Vous êtes sur la page 1sur 21

university-logo

Problem Formulation
Main Theorem
Numerical Results
Conclusions
Path-Dependent Dividends
and the American Put Option
M.H. Vellekoop
1
J.W. Nieuwenhuis
2
1
Department of Applied Mathematics
University of Twente
2
University of Groningen
Numerical Methods in Finance, Paris 2009
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Outline
1
Problem Formulation
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
2
Main Theorem
Calculating the Optimal Exercise Boundary
3
Numerical Results
Knock Out Dividend Model
Proportional Cash Dividend Model
Fixed Cash Dividend Model
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
Standard Equity Model
Consider the standard Black-Scholes-Merton model for stock and bond prices
dS
t
= rS
t
dt + S
t
dW
t
dB
t
= rB
t
dt
for time period t [0, T] with
S
0
, B
0
, r , given strictly positive constants and
W a one-dimensional Brownian Motion on the ltered probability space
(, F, (F
t
)
t [0,T]
, Q).
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
American and European Put Options
Let AP and EP denote the American and European Put price processes for
the maturity T and a strike K > 0:
EP
t
= E
Q
[
B
t
B
T
(K S
T
)
+
| F
t
]
def
= E(t , S
t
)
AP
t
= ess sup
T
[t ,T]
E
Q
[
B
t
B

(K S

)
+
| F
t
]
def
= A(t , S
t
)
where T
[t ,T]
is the set of all stopping times with values in [t , T].
Optimal stopping problem for American Put has the solution

= inf{u t : A(u, S
u
) = (K S
u
)
+
} = inf{u t : S
u
S

u
}
where S

denotes what is called the optimal exercise boundary.

Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
The difference between the American and European option price is known as
EE(t , s) = A(t , s) E(t , s)
and it can be characterized as follows:
Early Exercise Representation
(Carr et al 92), (Jacka 93), (Kim 90)
EE(t , s) = rK E
Q
[

T
t
e
r (ut )
1
{S
u
S

u
}
du | S
t
= s ]
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
Dividend Models
We would like to extend these results to the case where dividends are
included, i.e.
dS
t
= rS
t
dt + S
t
dW
t
dD
S
t
with, for example, continuous, proportional or xed cash dividends
dD
S
t
= qS
t
dt
dD
S
t
= (1 )S
t
d1
{t t
D
}
dD
S
t
= min{d, S
t
}d1
{t t
D
}
.
for a given t
D
]0, T[ and d > 0, q > 0, , ]0, 1[.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
Path-Dependent Dividend
Here, we focus on knock-out version of proportional dividends:
dD
S
t
= (1 )S
t
1
{ min
u[0,t
d
]
S
u
S
0
}
d1
{t t
d
}
This models the fact that the company which issued the stock will only pay
dividends if the stock price has not fallen below the level S
0
before the
dividend date.
This obviously makes the dividend path-dependent.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
Path-Dependent Dividend
European Put option with knockout dividends can be priced in closed form:
Lemma
For general ]0, 1] and m
t
S
0
we have that the European put option
price E(t , S
t
) with strike K and maturity T equals, for all t [0, T],
P
K,T
(t , S
t
) + 1
{t <t
d
}
Ke
r (Tt )

h
t
d
,T
(S
0
, K) L

t

h
t
d
,T
(
S
t
L
t
,
K
L
2
t
)

1
{t <t
d
}
e
r (Tt )

H
t
d
,T
(S
0
, K) L
2+
t

H
t
d
,T
(
S
t
L
t
,
K
L
2
t
)

with L
t
=
S
0
S
t
, = 2r
2
1 and

H
t
1
,t
2
(x
1
, x
2
) = E
Q
[

S
t
2
1
{

S
t
2

x
2

,

S
t
1
>x
1
}
| F
t
]
E
Q
[

S
t
2
1
{

S
t
2
x
2
,

S
t
1
>x
1
}
| F
t
]

h
t
1
,t
2
(x
1
, x
2
) = Q(

S
t
1
> x
1
, x
2
<

S
t
2

x
2

| F
t
)
where

S is the asset process with zero dividends.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
Dividend Model Assumptions
In more general setup, we assume
1
S is an adapted cdlg semimartingale, Markov, and distribution of S
t
has a density for all t
2
D
S
is an adapted, increasing cdlg semimartingale, continuous in all
but countable number of time points
3
S/B +

dD
S
/B is a Q-martingale, and the functions
x E
Q
[(S
u
) | S
t
= x], x E
Q
[

u
t
dD
S
v
/B
v
| S
t
= x]
are increasing for all 0 t < u T and for all increasing functions
: R R.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Standard Black-Scholes-Merton Model
Including Dividends
Assumptions
Dividend Model Assumptions
These assumptions include the dividend models mentioned earlier, but also
stock-dependent volatility models (Babilua et al. 07).
They guarantee that if we dene
S

t
= inf{s > 0 : A(t , s) > K s}
then
s > S

t
A(t , s) > K s
i.e. it excludes possibility that there
are several boundaries which
separate different continuation and
stopping regions.
dD
S
t
= min{S
t
D

, max{40 S
t
D

, 0}}d1
{t t
D
}
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Calculating the Optimal Exercise Boundary
For stock price models with dividends, the early exercise premium can be
characterized as follows:
Theorem (Early Exercise Representation including Dividends)
Under the assumptions stated above, and the additional assumption that the
optimal exercise boundary S

is continuous apart from at most a countable

number of points, the American and European Put price processes satisfy
AP
t
EP
t
= B
t
E
Q
[

T
t
1
{S
u
S

u
S
u
=0}
(d(
K
B
u
) +
dD
S
u
B
u
) | F
t
].
Continuity of optimal exercise boundary is subject of ongoing research, in
collaboration with CERMICS.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Calculating the Optimal Exercise Boundary
Part I
First prove that s A(t , s) is Lipschitz continuous, uniformly over t
Apply Meyer-Ito formula for convex mappings to the process
X = (AP K + S)/B.
Use the fact that AP/B, as Snell envelope of upper semicontinuous
process, is a continuous positive supermartingale, and
that S

t
is continuous apart from a countable number of points,
to show that the local time of X in zero equals zero.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Calculating the Optimal Exercise Boundary
Part II
Since S/B +

dD
S
/B is a Q-martingale, it turns out that the remaining
technical point is to prove that

t
0
1
{S
u
>S

u
}
d(
AP
u
B
u
)
is also a martingale under Q.
We do this by extending earlier dened methods (El Karoui 79, Karatzas
& Shreve 88, Jacka 93) for optimal stopping problems, exploiting the
fact that we may show that we should never stop in points where S is
discontinuous.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Calculating the Optimal Exercise Boundary
Integral equation for Optimal Exercise Boundary
Inserting s = S

t
as initial condition at time t gives
Corollary
Under the assumptions of the previous theorem, the optimal exercise
boundary satises
K S

t
=
E(t , S

t
) B
t
E
Q
[

T
t
1
{S
u
S

u
S
u
=0}
(d(
K
B
u
) +
dD
S
u
B
u
) | S
t
= S

t
].
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Knock Out Dividend Model
Proportional Cash Dividend Model
Fixed Cash Dividend Model
Knock Out Dividend Model
Equity model
dS
t
= rS
t
dt + S
t
dW
t
dD
S
t
dD
S
t
= (1 )S
t
1
{ min
u[0,t
d
]
S
u
S
0
}
d1
{t t
d
}
.
Parameter values
S
0
= K = 100, = 0.50, r = 0.125, T = 0.50, t
D
= 0.30, = .99, = .20.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
0
10
20
30
40
50
60
70
80
90
100
Time
O
p
t
i
m
a
l

E
x
e
r
c
i
s
e

B
o
u
n
d
a
r
y
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Knock Out Dividend Model
Proportional Cash Dividend Model
Fixed Cash Dividend Model
Proportional Cash Dividend Model
Equity model
dS
t
= rS
t
dt + S
t
dW
t
dD
S
t
dD
S
t
= (1 )S
t
d1
{t t
D
}
.
Parameter values
S
0
= 100, = 0.30, r = 0.04, K = 100, T = 2.00, t
D
= 1.50, = 0.98
0 0.5 1 1.5 2
0
10
20
30
40
50
60
70
80
90
100
Time
O
p
t
i
m
a
l

E
x
e
r
c
i
s
e

B
o
u
n
d
a
r
y
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Knock Out Dividend Model
Proportional Cash Dividend Model
Fixed Cash Dividend Model
Fixed Cash Dividend Model
Equity model
dS
t
= rS
t
dt + S
t
dW
t
dD
S
t
dD
S
t
= min{D, S
t
}d1
{t t
D
}
.
results in following integral equation
EEP(t , s) = rK

t
d
t
e
r (ut )
N(
ln(S

u
/s)r (ut )

ut
)du
+ rKe
r (t
d
t )

T
t
d

z
N(
ln(se
r (t
d
t )+x

t
d
t
D)ln S

u
+r (ut
d
)

ut
d
)dN(x)du
where
z =
ln(D/s) r (t
d
t )

t
d
t
.
and where N is the cumulative standard normal distribution function.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Knock Out Dividend Model
Proportional Cash Dividend Model
Fixed Cash Dividend Model
Fixed Cash Dividend Model
Equity model
dS
t
= rS
t
dt + S
t
dW
t
dD
S
t
dD
S
t
= min{d, S
t
}d1
{t t
D
}
.
Parameter values
S
0
= 100, = 0.30, r = 0.04, K = 100, T = 2.00, t
D
= 1.50, d = 5.
0 0.5 1 1.5 2
0
10
20
30
40
50
60
70
80
90
100
Time
O
p
t
i
m
a
l

E
x
e
r
c
i
s
e

B
o
u
n
d
a
r
y
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Problem Formulation
Main Theorem
Numerical Results
Conclusions
Conclusions
We extended the early exercise representation formula to a rather
general model for stocks with knock-out dividends.
However, we need to assume that the optimal exercise boundary is
continuous apart from a countable number of points in time.
Establishing how to prove this a priori is known to be hard, and is the
subject of ongoing research.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Appendix References
References I
P. Babilua, I. Bokuchava, B. Dochviri, and M. Shashiashvili.
The American put option in a one-dimensional diffusion model with
level-dependent volatility.
Stochastics, 79:525, 2007.
P. Carr, R. Jarrow, and R. Myneni.
Alternative characterizations of American puts.
Mathematical Finance, 2:87106, 1992.
N. El Karoui.
Les aspects probabilistes du contrle stochastique, Lecture Notes in
Mathematics 876.
Springer-Verlag, Berlin, 1981.
S.D. Jacka.
Local times, optimal stopping and semimartingales.
Annals of Probability, 21:329339, 1993.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option
university-logo
Appendix References
References II
S. Kallast and A. Kivinukk.
Pricing and hedging American options using approximations by Kim
integral equations.
European Finance Review, 7:361383, 2003.
I.J. Kim.
The analytical valuation of American options.
Review of Financial Studies, 3:547472, 1990.
G. Peskir.
On the American option problem.
Mathematical Finance, 15:169181, 2005.
Vellekoop & Nieuwenhuis Path-Dependent Dividends and the American Put Option