AN INTRODUCTION TO L

EVY PROCESSES
WITH APPLICATIONS IN FINANCE
ANTONIS PAPAPANTOLEON
Abstract. These notes aim at introducing Levy processes in an infor-
mal and intuitive way. Several important results about Levy processes,
such as the Levy-Khintchine formula, the Levy-Ito decomposition and
Girsanovs transformations, are discussed. Applications of Levy pro-
cesses in nancial modeling are presented and some popular models in
nance are revisited from the point of view of Levy processes.
1. Introduction Motivation
Levy processes play a fundamental role in Mathematical Finance, as well
as in other elds of science, such as Physics (turbulence, laser cooling),
Engineering (telecommunications, queues, dams) and the Actuarial science
(insurance risk). A comprehensive overview of some applications of Levy
processes can be found in Prabhu (1998), in Barndor-Nielsen, Mikosch,
and Resnick (2001) and in Kyprianou (2005).
Levy processes have become increasingly popular in Mathematical Fi-
nance because they can describe the observed reality of nancial markets
in a more accurate way than models based on Brownian motion. In the
real world, we observe that asset price processes have jumps or spikes, see
Figure 1.1, and risk-managers have to take them into account. Moreover,
the empirical distribution of asset returns exhibits fat tails and skewness,
behavior that deviates from normality, see Figure 1.2. Hence, models that
accurately t return distributions are essential for the estimation of prot
and loss (P&L) distributions. In the risk-neutral world, we observe that
implied volatilities are constant neither across strike nor across maturities
as stipulated by the Black and Scholes (1973) (actually, Samuelson 1965)
model, see Figure 1.3. Therefore, traders need models that can capture the
behavior of the implied volatility smiles more accurately, in order to handle
the risk of trades. Levy processes provide us with the appropriate frame-
work to adequately describe all these observations, both in the real and
in the risk-neutral world.
Paul Levy. Processes with independent and stationary increments are
named Levy processes after the French mathematician Paul Levy (1886-
1971), who made the connection with innitely divisible laws, characterized
their distributions (Levy-Khintchine formula) and described their structure
These lecture notes were prepared for mini-courses taught at the University of Piraeus
in April 2005 and the University of Leipzig in November 2005. I am grateful for the
opportunity of lecturing on these topics to George Skiadopoulos and Thorsten Schmidt.
1
2 ANTONIS PAPAPANTOLEON
100
105
110
115
120
125
130
135
140
145
150
Oct 1997 Oct 1998 Oct 1999 Oct 2000 Oct 2001 Oct 2002 Oct 2003 Oct 2004
USD/JPY
Figure 1.1. USD/JPY foreign exchange rate, October
1997-October 2004.
0.02 0.01 0.0 0.01 0.02
0
2
0
4
0
6
0
8
0
Figure 1.2. Empirical distribution of daily log-returns for
the GBP/USD exchange rate and tted Normal distribution
(Levy-Ito decomposition). Paul Levy is one of the founding fathers of the
theory of stochastic processes and made major contributions to the eld
of probability theory. Among others, Paul Levy contributed to the study
of Gaussian variables and processes, the law of large numbers, the central
limit theorem, stable laws, innitely divisible laws and pioneered the study
of processes with independent and stationary increments.
More information about Paul Levy and his scientic work, can be found
at the websites
http://www.cmap.polytechnique.fr/~rama/levy.html
and
http://www.annales.org/archives/x/paullevy.html
(in French).
INTRODUCTION TO L

EVY PROCESSES 3
10
20
30
40
50
60
70
80
90 1
2
3
4
5
6
7
8
9
10
10
10.5
11
11.5
12
12.5
13
13.5
14
maturity
delta (%) or strike
im
p
lie
d

v
o
l
(
%
)
Figure 1.3. Implied volatilities of vanilla options on the
EUR/USD exchange rate on November 5, 2001.
2. Definition
Let (, T, F, P) be a ltered probability space, where T = T
T
and the l-
tration F = (T
t
)
t[0,T]
satises the usual conditions (cf. Jacod and Shiryaev
2003, I.1.3). Note that, by abuse of the standard notation, whenever we
write t 0 that means t [0, T].
Denition 2.1. A c`adl`ag, adapted, real valued stochastic process L =
(L
t
)
t0
with L
0
= 0 a.s. is called a Levy process if the following conditions
are satised:
(L1): L has independent increments, i.e. L
t
L
s
is independent of T
s
for any 0 s < t T.
(L2): L has stationary increments, i.e. for any s, t 0 the distribution
of L
t+s
L
t
does not depend on t.
(L3): L is stochastically continuous, i.e for every t 0 and > 0:
lim
st
P([L
t
L
s
[ > ) = 0.
The simplest Levy process is the linear drift, a deterministic process.
Brownian motion is the only (non-deterministic) Levy process with contin-
uous sample paths. Other examples of Levy processes are the Poisson and
compound Poisson processes. Notice that the sum of a Brownian motion
and a compound Poisson process is again a Levy process; it is often called
a jump-diusion process; we shall call it a Levy jump-diusion process,
because there exist jump-diusion processes which are not Levy processes.
3. First example: a L evy jump-diffusion
Assume that the process L = (L
t
)
t0
is a Levy jump-diusion, i.e. a
Brownian motion plus a compensated compound Poisson process. It can be
4 ANTONIS PAPAPANTOLEON
0.0 0.2 0.4 0.6 0.8 1.0
0
.0
0
.0
1
0
.0
2
0
.0
3
0
.0
4
0
.0
5
0.0 0.2 0.4 0.6 0.8 1.0

0
.1
0
.0
0
.1
0
.2
Figure 2.4. Examples of Levy processes: linear drift (left)
and Brownian motion
0.0 0.2 0.4 0.6 0.8 1.0

0
.4

0
.2
0
.0
0
.2
0
.4
0
.6
0
.8
0.0 0.2 0.4 0.6 0.8 1.0

0
.1
0
.0
0
.1
0
.2
0
.3
0
.4
Figure 2.5. Examples of Levy processes: compound Pois-
son process (left) and Levy jump-diusion
described by
L
t
= bt +W
t
+
_
N
t

k=1
J
k
t
_
(3.1)
where b R, R
+
, W = (W
t
)
t0
is a standard Brownian motion,
N = (N
t
)
t0
is a Poisson process with parameter (i.e. IE[N
t
] = t)
and J = (J
k
)
k1
is an i.i.d. sequence of random variables with probability
distribution F and IE[J] = < ; F describes the distribution of jump
size. All sources of randomness are mutually independent.
It is well known that Brownian motion is a martingale; moreover, the
compensated compound Poisson process is a martingale. Therefore, L =
(L
t
)
t0
is a martingale if and only if b = 0.
The characteristic function of L
t
is
IE
_
e
iuL
t

= IE
_
exp
_
iu
_
bt +W
t
+
N
t

k=1
J
k
t
__
_
= exp
_
iubt

IE
_
exp
_
iuW
t
_
exp
_
iu
_
N
t

k=1
J
k
t
__
_
;
INTRODUCTION TO L

EVY PROCESSES 5
since all the sources of randomness are independent, we get
= exp
_
iubt

IE
_
exp
_
iuW
t
_
_
IE
_
exp
_
iu
N
t

k=1
J
k
iut
_
_
;
taking into account that
IE[e
iuW
t
] = e

1
2

2
u
2
t
, W
t
Normal(0,
2
t)
IE[e
iu
P
N
t
k=1
J
k
] = e
t(IE[e
iuJ
1])
, N
t
Poisson(t)
(cf. also Appendix B) we get
= exp
_
iubt

exp
_

1
2
u
2

2
t
_
exp
_
t
_
IE[e
iuJ
1] iuIE[J]
_
_
= exp
_
iubt

exp
_

1
2
u
2

2
t
_
exp
_
t
_
IE[e
iuJ
1 iuJ]
_
_
;
and because the distribution of J is F we have
= exp
_
iubt

exp
_

1
2
u
2

2
t
_
exp
_
t
_
R
_
e
iux
1 iux
_
F(dx)
_
.
Now, since t is a common factor, we re-write the above equation as
IE
_
e
iuL
t

= exp
_
t
_
iub
u
2

2
2
+
_
R
(e
iux
1 iux)F(dx)
_
_
. (3.2)
Since the characteristic function of a random variable determines its dis-
tribution, we have a characterization of the distribution of the random
variables underlying the Levy jump-diusion. We will soon see that this dis-
tribution belongs to the class of innitely divisible distributions. We will also
see that equation (3.2) is a special case of the celebrated Levy-Khintchine
formula.
4. Infinitely divisible distributions and the L evy-Khintchine
formula
There is strong interplay between Levy processes and innitely divisible
distributions. We rst dene innitely divisible distributions and give some
examples and then describe their relationship to Levy processes.
Let X be a real valued random variable, denote its characteristic function
by
X
and its law by P
X
, hence
X
(u) =
_
R
e
iux
P
X
(dx).
Denition 4.1. The law P
X
of a random variable X is innitely divisible,
if for all n N there exist i.i.d. random variables X
(1/n)
1
, . . . , X
(1/n)
n
such
that
X
d
= X
(1/n)
1
+. . . +X
(1/n)
n
. (4.1)
Equivalently, the law P
X
of a random variable X is innitely divisible if for
all n N there exists another law P
X
(1/n) of a random variable X
(1/n)
such
6 ANTONIS PAPAPANTOLEON
that
P
X
= P
X
(1/n) . . . P
X
(1/n)
. .
n times
. (4.2)
Alternatively, we can characterize an innitely divisible random variable
using its characteristic function.
Denition 4.2. The law of a random variable X is innitely divisible, if for
all n N, there exists a random variable X
(1/n)
, such that

X
(u) =
_

X
(1/n) (u)
_
n
. (4.3)
Example 4.3 (Normal distribution). Using the second denition, we can
easily see that the Normal distribution is innitely divisible. Let X
Normal(,
2
), then we have

X
(u) = exp
_
iu
1
2
u
2

2
_
= exp
_
iun

n

1
2
u
2
n

2
n
_
= exp
_
n(iu

n

1
2
u
2

2
n
)
_
=
_
exp
_
iu

n

1
2
u
2

2
n
_
_
n
=
_

X
(1/n) (u)
_
n
where X
(1/n)
Normal(

n
,

2
n
).
Example 4.4 (Poisson distribution). Following the same procedure, we
can easily deduce that the Poisson distribution is innitely divisible. Let
X Poisson(), then we have

X
(u) = exp
_
(e
iu
1)
_
= exp
_
n

n
(e
iu
1)
_
=
_
exp
_

n
(e
iu
1)
_
_
n
=
_

X
(1/n) (u)
_
n
where X
(1/n)
Poisson(

n
).
Remark 4.5. Other examples of innitely divisible distributions are the
compound Poisson distribution, the exponential, the -distribution, the geo-
metric, the negative binomial, the Cauchy distribution and the strictly stable
distribution. Counter-examples are the uniform and binomial distributions.
The next theorem provides a complete characterization of random vari-
ables with innitely divisible distributions via their characteristic functions;
this is the celebrated Levy-Khintchine formula.
INTRODUCTION TO L

EVY PROCESSES 7
Theorem 4.6. The law P
X
of a random variable X is innitely divisible if
and only if there exists a triplet (b, c, ), with b R, c R
+
and a measure
satisfying (0) = 0 and
_
R
(1 [x[
2
)(dx) < , such that
IE[e
iuX
] = exp
_
ibu
u
2
c
2
+
_
R
(e
iux
1 iux1l
{|x|<1}
)(dx)
_
. (4.4)
Proof. See Theorem 8.1 in Sato (1999).
The triplet (b, c, ) is called the Levy or characteristic triplet and the
exponent in (4.4)
(u) = iub
u
2
c
2
+
_
R
(e
iux
1 iux1l
{|x|<1}
)(dx) (4.5)
is called the Levy or characteristic exponent. Moreover, b R is called the
drift term, c R
+
the Gaussian or diusion coecient and the Levy
measure.
Now, consider a Levy process L = (L
t
)
t0
; using the fact that, for any
n N and any t > 0,
L
t
= L
t/n
+ (L
2t/n
L
t/n
) +. . . + (L
t
L
(n1)t/n
) (4.6)
together with the stationarity and independence of the increments, we con-
clude that the random variable L
t
is innitely divisible.
Moreover, for all u R and all t 0, dene

t
(u) = log IE[e
iuL
t
]; (4.7)
making use of (4.6) and the stationarity and independence of the increments,
we have for any m N
m
1
(u) =
m
(u) (4.8)
and hence, for any rational t > 0
t
1
(u) =
t
(u). (4.9)
For an irrational t, we can choose a sequence of rationals that decreases to
t and use right continuity of L to prove that (4.9) holds for all t 0.
Therefore, we have that for every Levy process, the following property
holds
IE[e
iuL
t
] = e
t(u)
(4.10)
= exp
_
t
_
ibu
u
2
c
2
+
_
R
(e
iux
1 iux1l
{|x|<1}
)(dx)
_
_
where (u) :=
1
(u) is the characteristic exponent of L
1
:= X (say), a
random variable with an innitely divisible distribution.
We have seen so far, that every Levy process can be associated with the
law of an innitely divisible distribution. The opposite, i.e. that given any
random variable X, whose law is innitely divisible, we can construct a Levy
process L = (L
t
)
t0
such that L
1
:= X, is also true. This will be the subject
of the next section.
8 ANTONIS PAPAPANTOLEON
5. The L evy-It o decomposition
Theorem 5.1. Consider a triplet (b, c, ) where b R, c R
+
and
is a measure satisfying (0) = 0 and
_
R
(1 [x[
2
)(dx) < . Then,
there exists a probability space (, T, P) on which four independent Levy
processes exist, L
(1)
, L
(2)
, L
(3)
and L
(4)
, where L
(1)
is a constant drift, L
(2)
is a Brownian motion, L
(3)
is a compound Poisson process and L
(4)
is a
square integrable (pure jump) martingale with an a.s. countable number of
jumps on each nite time interval of magnitude less that 1. Taking L =
L
(1)
+ L
(2)
+ L
(3)
+ L
(4)
, we have that there exists a probability space on
which a Levy process L = (L
t
)
t0
with characteristic exponent
(u) = iub
u
2
c
2
+
_
R
(e
iux
1 iux1l
{|x|<1}
)(dx) (5.1)
for all u R, is dened.
Proof. See chapter 4 in Sato (1999) or chapter 2 in Kyprianou (2005).
The Levy-Ito decomposition is a hard mathematical result to prove; here,
we go through some steps of the proof because it reveals much about the
structure of the paths of a Levy process. We split the Levy exponent (5.1)
into four parts
=
(1)
+
(2)
+
(3)
+
(4)
where

(1)
(u) = iub,
(2)
(u) =
u
2
c
2
,

(3)
(u) =
_
|x|1
(e
iux
1)(dx),

(4)
(u) =
_
|x|<1
(e
iux
1 iux)(dx).
The rst part corresponds to a deterministic linear process (drift) with pa-
rameter b, the second one to a Brownian motion with coecient c and
the third part to the characteristic function of a compound Poisson pro-
cess with arrival rate := (R (1, 1)) and jump magnitude F(dx) :=
(dx)
(R\(1,1))
1l
{|x|1}
.
The last part is the most dicult to handle; let L
(4)
denote the jumps
of the Levy process L
(4)
and
L
(4)
denote the random measure counting
the jumps of L
(4)
. Next, one constructs a compensated compound Poisson
process
L
(4,)
t
=

0st
L
(4)
s
1l
{1>|L
(4)
s
|>}
t
_
_
1>|x|>
x(dx)
_
=
t
_
0
_
1>|x|>
x
L
(4)
(dx, ds) t
_
_
1>|x|>
x(dx)
_
INTRODUCTION TO L

EVY PROCESSES 9
and shows that the jumps of L
(4)
form a Poisson point process; using results
for Poisson point processes, we get that the characteristic function of L
(4,)
is

(4,)
(u) =
_
<|x|<1
(e
iux
1 iux)(dx).
Finally, there exists a Levy process L
(4)
which is a square integrable mar-
tingale and L
(4,)
L
(4)
uniformly on [0, T] as 0+ with Levy exponent

(4)
.
Therefore, we can decompose any Levy process into four independent
Levy processes L = L
(1)
+L
(2)
+L
(3)
+L
(4)
, i.e.
L
t
= bt +

cW
t
+
t
_
0
_
|x|1
x
L
(ds, dx) (5.2)
+
_
t
_
0
_
|x|<1
x
L
(ds, dx) t
_
|x|<1
x(dx)
_
where L
(1)
is a constant drift, L
(2)
is a Brownian motion, L
(3)
a compound
Poisson process and L
(4)
is a pure jump martingale. This result is the
celebrated Levy-Ito decomposition of a Levy process.
6. The L evy measure and path properties
The Levy measure is a measure on R that satises
(0) = 0 and
_
R
(1 [x[
2
)(dx) < . (6.1)
That means, that a Levy measure has no mass at the origin, but singularities
(i.e. innitely many jumps) can occur around the origin (i.e. small jumps).
Intuitively speaking, the Levy measure describes the expected number of
jumps of a certain height in a time interval of length 1.
For example, the Levy measure of the Levy jump-diusion is (dx) =
F(dx); from that we can deduce that the expected number of jumps, in
a time interval of length 1, is and the jump size is distributed according
to F.
More generally, if is a nite measure, i.e. (R) =
_
R
(dx) = < ,
then F(dx) :=
(dx)

, which is a probability measure. Thus, is the expected

number of jumps and F(dx) the distribution of the jump size x. If (R) = ,
then an innite number of (small) jumps is expected.
The Levy measure is responsible for the richness of the class of Levy
processes and carries useful information about the structure of the process.
Path properties of a Levy process can be read from the Levy measure: for
example, Figures 6.6 and 6.7 reveal that the compound Poisson process has
a nite number of jumps on every time interval, while the NIG and -stable
processes have an innite one; we then speak of an innite activity Levy
process.
10 ANTONIS PAPAPANTOLEON
1
2
3
4
5
1
2
3
4
5
Figure 6.6. The distribution function of the Levy measure
of the standard Poisson process (left) and the density of the
Levy measure of a compound Poisson process with double-
exponentially distributed jumps.
1
2
3
4
5
1
2
3
4
5
Figure 6.7. The density of the Levy measure of an NIG
(left) and an -stable process.
Proposition 6.1. Let L be a Levy process with triplet (b, c, ).
(1) If (R) < then almost all paths of L have a nite number of jumps
on every compact interval. In that case, the Levy process has nite
activity.
(2) If (R) = then almost all paths of L have an innite number of
jumps on every compact interval. In that case, the Levy process has
innite activity.
Proof. See Theorem 21.3 in Sato (1999).
Whether a Levy process has nite variation or not also depends on the
Levy measure (and on the presence or absence of a Brownian part).
Proposition 6.2. Let L be a Levy process with triplet (b, c, ).
(1) If c = 0 and
_
|x|1
[x[(dx) < then almost all paths of L have
nite variation.
(2) If c ,= 0 or
_
|x|1
[x[(dx) = then almost all paths of L have
innite variation.
Proof. See Theorem 21.9 in Sato (1999).
The dierent functions a Levy measure has to integrate in order to have
nite activity or variation, are graphically exhibited in Figure 6.8. The
INTRODUCTION TO L

EVY PROCESSES 11
compound Poisson process has nite measure, hence it has nite variation
as well; on the contrary, the NIG has an innite measure and has innite
variation.
0.2
0.4
0.6
0.8
1
Figure 6.8. The Levy measure must integrate [x[
2
1 (red
line); it has nite variation if it integrates [x[ 1 (blue line);
it is nite if it integrates 1 (orange line).
Moreover, the Levy measure carries information about the niteness of
the moments of a Levy process; this is extremely useful information in math-
ematical nance, for the existence of a martingale measure. The niteness
of the moments of a Levy process is related to the niteness of an integral
over the Levy measure (more precisely, the restriction of the Levy measure
to jumps larger than 1 in absolute value, i.e. big jumps).
Proposition 6.3. Let L be a Levy process with triplet (b, c, ). Then
(1) L
t
has nite p-th moment for p R
+
(IE[L
t
[
p
< ) if and only if
_
|x|1
[x[
p
(dx) < .
(2) L
t
has nite p-th exponential moment for p R (IE[e
pL
t
] < ) if
and only if
_
|x|1
e
px
(dx) < .
Proof. The proof of these results can be found in Theorem 25.3 in Sato
(1999). Actually, the conclusion of this theorem holds for the general class of
submultiplicative functions (cf. Denition 25.1 in Sato 1999), which contains
exp(px) and [x[
p
1 as special cases.
In order to gain some understanding of this result and because it blends
beautifully with the Levy-Ito decomposition, we will give a rough proof of
the suciency for the second part (inspired by Kyprianou 2005).
Recall from the Levy-Ito decomposition, that the characteristic exponent
of a Levy process was split into four independent parts, the third of which
is a compound Poisson process with arrival rate := (R (1, 1)) and
jump magnitude F(dx) :=
(dx)
(R\(1,1))
1l
{|x|1}
. Finiteness of IE[e
pL
t
] implies
12 ANTONIS PAPAPANTOLEON
niteness of IE[e
pL
(3)
t
], where
IE[e
pL
(3)
t
] = e
t

k0
(t)
k
k!
_
_
_
R
e
px
F(dx)
_
_
k
= e
t

k0
t
k
k!
_
_
_
R
e
px
1l
{|x|1}
(dx)
_
_
k
.
Since all the summands are nite, the one corresponding to k = 1 must also
be nite, therefore
e
t
_
R
e
px
1l
{|x|1}
(dx) < =
_
|x|1
e
px
(dx) < .
1
2
3
4
Figure 6.9. A Levy process has rst moment if the Levy
measure integrates [x[ for [x[ 1 (blue line) and second
moment if it integrates x
2
for [x[ 1 (orange line).
The graphical representation of the functions the Levy measure must
integrate so that a Levy process has nite moments is given in Figure 6.9.
The NIG process possesses moments of all order, while the -stable does
not; one can already see from Figure 6.7 that the tails of the Levy measure
of the -stable are heavier than that of the NIG.
7. Some cases of particular interest
We already know that a Brownian motion, a (compound) Poisson process
and a Levy jump-diusion are Levy processes, their Levy-Ito decomposition
and their characteristic functions. Here, we present some other special cases
of Levy processes that are of special interest.
7.1. Subordinator. A subordinator is an a.s. increasing (in t) Levy pro-
cess. Equivalently, for L to be a subordinator, the triplet must satisfy
(, 0) = 0, c = 0,
_
(0,1)
x(dx) < and = b +
_
(0,1)
x(dx) > 0.
INTRODUCTION TO L

EVY PROCESSES 13
The Levy-Ito decomposition of a subordinator is
(7.1) L
t
= t +
t
_
0
_
(0,)
x
L
(ds, dx)
and the Levy-Khintchine formula takes the form
(7.2) IE[e
iuL
t
] = exp
_
t
_
iu +
_
(0,)
(e
iux
1)(dx)
_
_
.
Two examples of subordinators are the Poisson and the inverse Gaussian
process, cf. Figures 7.10 and A.13.
7.2. Jumps of nite variation. A Levy process has jumps of nite varia-
tion if and only if
_
|x|1
[x[(dx) < . In this case, the Levy-Ito decompo-
sition of L resumes the form
(7.3) L
t
= t +cW
t
+
t
_
0
_
R
x
L
(ds, dx)
and the Levy-Khintchine formula takes the form
(7.4) IE[e
iuL
t
] = exp
_
t
_
iu
u
2
c
2
+
_
R
(e
iux
1)(dx)
_
_
,
where is dened as above.
Moreover, if ([1, 1]) < , which means that (R) < then L is a
compound Poisson process.
7.3. Spectrally one-sided. A Levy processes is called spectrally negative
if (0, ) = 0. Similarly, a Levy processes is called spectrally positive if L
is spectrally negative.
7.4. Finite rst moment. As we have seen already, a Levy process has
a nite rst moment if and only if
_
|x|1
[x[(dx) < . Therefore, the
Levy-Ito decomposition of L resumes the form
(7.5) L
t
= b

t +cW
t
+
_
t
_
0
_
R
x
L
(ds, dx) t
_
R
x(dx)
_
and the Levy-Khintchine formula takes the form
(7.6) IE[e
iuL
t
] = exp
_
t
_
iub

u
2
c
2
+
_
R
(e
iux
1 iux)(dx)
_
_
,
where b

= b +
_
|x|1
x(dx).
Remark 7.1 (Assumption (M)). For the remaining parts we will work only
with Levy process that have a nite rst moment. We will refer to them as
Levy processes that satisfy assumption (M). For the sake of simplicity, we
suppress the notation b

and write b instead.

14 ANTONIS PAPAPANTOLEON
0.0 0.2 0.4 0.6 0.8 1.0

0
.5
0
.0
0
.5
0.0 0.2 0.4 0.6 0.8 1.0
0
1
0
2
0
3
0
4
0
Figure 7.10. Simulated path of a normal inverse Gaussian
(left) and an inverse Gaussian process
8. Elements from semimartingale theory
A semimartingale is a stochastic process X which admits the decomposi-
tion
X = X
0
+M +A
where X
0
is nite and T
0
-measurable, M is a local martingale with M
0
= 0
and A is a nite variation process with A
0
= 0. X is a special semimartingale
if A is predictable.
Any special semimartingale X has the following, so-called, canonical de-
composition
X
t
= X
0
+B
t
+X
c
t
+
t
_
0
_
R
x(
X

X
)(ds, dx)
where X
c
is the continuous martingale part of X and
_
t
0
_
R
x(
X

X
)(ds, dx)
is the purely discontinuous martingale part of X.
X
is called the random
measure of jumps of X; it counts the number of jumps of specic size that
occur in a time interval of specic length.
X
is called the compensator of

X
(for more details, see chapter II in Jacod and Shiryaev 2003).
Returning to the Levy-Ito decomposition (5.2), we can easily see that a
Levy process with triplet (b, c, ) which satises assumption (M), has the
following canonical decomposition
L
t
= bt +

cW
t
+
t
_
0
_
R
x(
L

L
)(ds, dx),
where
t
_
0
_
R
x
L
(ds, dx) =

0st
L
s
and
IE
_
t
_
0
_
R
x
L
(ds, dx)
_
=
t
_
0
_
R
x
L
(ds, dx) = t
_
R
x(dx);
INTRODUCTION TO L

EVY PROCESSES 15
one also writes
L
(ds, dx) = (dx)ds.
We denote the continuous martingale part of L by L
c
and the purely
discontinuous martingale part of L by L
d
, i.e.
L
c
t
=

cW
t
and L
d
t
=
t
_
0
_
R
x(
L

L
)(ds, dx). (8.1)
Remark 8.1. Every Levy process is also a semimartingale. Every Levy pro-
cess with nite rst moment (i.e. satises assumption (M)) is also a special
semimartingale; conversely, every Levy process that is a special semimartin-
gale, has a nite rst moment.
9. Girsanovs theorem
We will describe a special case of Girsanovs theorem for semimartingales,
where a Levy process remains a process with independent increments (PII)
under the new measure.
Let P and

P be probability measures dened on the ltered space (, T, F).
Two probability measures P and

P are equivalent measures, if P(A) = 0

P(A) = 0 and one writes P

P.
Whenever P

P, there exists a unique positive P-martingale = (
t
)
t0
such that IE[
d
e
P
dP

T
t
] =
t
, t 0 (the Radon-Nikodym derivative); is called
the density process of

P with respect to P.
Conversely, given a measure P and a positive P-martingale = (
t
)
t0
,
one can dene a measure

P on (, T), where P

P, via the Radon-Nikodym
derivative IE[
d
e
P
dP

T
T
] =
T
.
Theorem 9.1. Let L be a Levy process with triplet (b, c, ) under P, that
satises (M). It has the canonical decomposition
L
t
= bt +

cW
t
+
t
_
0
_
R
x(
L

L
)(ds, dx). (9.1)
(A1): Assume that P

P. Then, there exists a deterministic process
and a measurable non-negative deterministic process Y , satisfying
t
_
0
_
R
[x
_
Y (s, x) 1
_
[(dx)ds < ,
and
t
_
0
_
c
2
s
_
ds <
16 ANTONIS PAPAPANTOLEON
such that the density process = (
t
)
t0
has the form

t
= IE
_
d

P
dP

T
t
_
= exp
_
t
_
0

cdW
s

1
2
t
_
0

2
s
cds (9.2)
+
t
_
0
_
R
(Y (s, x) 1)(
L

L
)(ds, dx)

t
_
0
_
R
(Y (s, x) 1 ln(Y (s, x)))
L
(ds, dx)
_
.
(A2): Conversely, if is a positive martingale of the form (9.2), then it
denes a probability measure

P on (, T), such that P

P.
(A3): In both cases, we have that

W
t
= W
t

_
t
0

s
ds is a

P-Brownian
motion,
L
(ds, dx) = Y (s, x)
L
(ds, dx) is the

P-compensator of
L
and L has the following canonical decomposition under

P
L
t
=

bt +

W
t
+
t
_
0
_
R
x(
L

L
)(ds, dx), (9.3)
where

bt = bt +
t
_
0
c
s
ds +
t
_
0
_
R
x
_
Y (s, x) 1
_

L
(ds, dx). (9.4)
Proof. Theorems III.3.24, III.5.19 and III.5.35 in Jacod and Shiryaev (2003)
yield the result.
Remark 9.2. In the statement of Theorem 9.1, we have implicitly assumed
that L has nite rst moment under

P as well, i.e.

IE[L
t
[ < , for all t 0.
Remark 9.3. The process L is not necessarily a Levy process under the
new measure

P; it depends on the tuple (, Y ). We have the following cases
(G1): if (, Y ) are deterministic and independent of time, then L re-
mains a Levy process under

P; its triplet is (

b, c, Y ).
(G2): if (, Y ) are deterministic but depend on time, then L becomes
a process with independent (but not stationary) increments under

P, often called an additive process.

(G3): if (, Y ) are neither deterministic nor independent of time, then
L is a semimartingale under

P.
Remark 9.4. Notice that c, the diusion coecient, and
L
, the random
measure of jumps of L, did not change under the change of measure from P
to

P. That happens because c and
L
are path properties of the process and
do not change under an equivalent change of measure. Intuitively speaking,
the paths do not change, the probability of certain paths occurring changes.
INTRODUCTION TO L

EVY PROCESSES 17
Example 9.5. Assume that L is a Levy process with canonical decompo-
sition (9.1) under P. Assume that P

P and the density process is

t
= exp
_

cW
t
+
t
_
0
_
R
x(
L

L
)(ds, dx) (9.5)

_
c
2
2
+
_
R
(e
x
1 x)(dx)
_
t
_
,
where R
+
and R are constants.
Then, comparing (9.5) with (9.2), we have that the tuple of functions
that characterize the change of measure is (, Y ) = (, f), where f(x) =
e
x
. Because (, Y ) are deterministic and independent of time, L remains
a Levy process under

P, its Levy triplet is (

b, c, e
x
) and its canonical
decomposition is given by equations (9.3) and (9.4).
Remark 9.6. Actually, the change of measure of example 9.5 corresponds
to the so-called Esscher transformation or exponential tilting. In chapter 3 of
Kyprianou (2005), one can nd a signicantly easier proof of Girsanovs the-
orem for Levy processes for the special case of the Esscher transform. Here,
we reformulate the result of example 9.5 and give a short proof (inspired by
Eberlein and Papapantoleon 2005).
Proposition 9.7. Let L = (L
t
)
t0
be a Levy process with canonical decom-
position (9.1) under P and assume that IE[e
uL
t
] < for all u [p, p],
p > 0. Assume that P

P with Radon-Nikodym derivative
T
, where

t
=
e
L
c
t
e
L
d
t
IE[e
L
c
t
]IE[e
L
d
t
]
(9.6)
= exp
_

cW
t
+
t
_
0
_
R
x(
L

L
)(ds, dx)

_
c
2
2
+
_
R
(e
x
1 x)(dx)
_
t
_
,
for R, [[ < p. Then, L remains a Levy process under

P, its Levy triplet
is (

b, c, f), where f(x) = e

x
and its canonical decomposition is given by
equations
L
t
=

bt +

W
t
+
t
_
0
_
R
x(
L

L
)(ds, dx), (9.7)

b = b +c +
_
R
x
_
e
x
1
_
(dx). (9.8)
18 ANTONIS PAPAPANTOLEON
Proof. Applying Theorem 25.17 in Sato (1999), the moment generating func-
tion M
L
t
of L
t
exists for u C with 'u [p, p]. We get

IE[e
zL
t
] = IE[e
zL
t

t
]
= IE[e
zL
t
e
L
c
t
e
L
d
t
IE[e
L
c
t
]
1
IE[e
L
d
t
]
1
]
= IE[e
zbt
e
(z+)L
c
t
e
(z+)L
d
t
]IE[e
L
c
t
]
1
IE[e
L
d
t
]
1
]
= exp
_
_
t
_
zb +
(z +)
2
c
2
+
_
R
x
_
e
(z+)x
1 (z +)x
_
(dx)

2
c
2

_
R
x
_
e
x
1 x
_
(dx)

_
_
= exp
_
_
t
_
zb +
z
2
c
2
+zc +
_
R
e
x
_
e
zx
1 zx
_
(dx)
+
_
R
zx
_
e
x
1
_
(dx)

_
_
= exp
_
_
t
_
z
_
b +c +
_
R
x
_
e
x
1
_
(dx)
_
+
z
2
c
2
+
_
R
_
e
zx
1 zx
_
e
x
(dx)
_
_
_
.
The statement follows by proving that e
x
(dx) is a Levy measure, i.e.
_
R
(1 x
2
)e
x
(dx) < . It suces to note that
(9.9)
_
|x|1
x
2
e
x
(dx) C
_
|x|1
x
2
(dx) < ,
because is a Levy measure, where C is a constant, while the other part
follows from the assumptions.
10. Martingales and L evy processes
We give a condition for a Levy process to be a martingale and discuss
when the exponential of a Levy process is a martingale.
Proposition 10.1. Let L = (L
t
)
t0
be a Levy process with Levy triplet
(b, c, ) and assume that IE[L
t
[ < . L is a martingale if and only if b = 0.
Proof. Follows from Theorem 5.2.1 in Applebaum (2004).
Proposition 10.2. Let L = (L
t
)
t0
be a Levy process with Levy exponent
and assume that IE[e
uL
t
] < , u R. The process M = (M
t
)
t0
, dened
as
M
t
=
e
uL
t
e
t(u)
INTRODUCTION TO L

EVY PROCESSES 19
is a martingale.
Proof. We have that IE[e
uL
t
] = e
t(u)
< , for all t 0.
For 0 s t, we can re-write M as
M
t
=
e
uL
s
e
s(u)
e
u(L
t
L
s
)
e
(ts)(u)
= M
s
e
u(L
t
L
s
)
e
(ts)(u)
.
Using the fact that a Levy process has stationary and independent incre-
ments, we get
IE
_
M
t

T
s
_
= M
s
IE
_
e
u(L
t
L
s
)
e
(ts)(u)

T
s
_
= M
s
e
(ts)(u)
e
(ts)(u)
= M
s

The stochastic exponential c(L) of a Levy process L = (L

t
)
t0
is the
solution Z of the stochastic dierential equation
dZ
t
= Z
t
dL
t
, Z
0
= 1
dened as
c(L)
t
= exp
_
L
t

ct
2
_

0st
_
1 + L
s
_
e
L
s
. (10.1)
The stochastic exponential of a Levy process that is a martingale is a local
martingale (cf. Jacod and Shiryaev 2003, Theorem I.4.61) and indeed a
martingale when working of a nite time horizon (cf. Kallsen 2000, Lemma
4.4).
11. Construction of L evy processes
Three popular methods to construct a Levy process are described below.
(C1): Specifying a Levy triplet; more specically, whether there ex-
ists a Brownian component or not and what is the Levy measure.
Examples of Levy process constructed this way include the stan-
dard Brownian motion, which has Levy triplet (0, 1, 0) and the Levy
jump-diusion, which has Levy triplet (b,
2
, F).
(C2): Specifying an innitely divisible random variable as the density
of the increments at time scale 1 (i.e. L
1
). Examples of Levy process
constructed this way include the standard Brownian motion, where
L
1
Normal(0, 1) and the normal inverse Gaussian process, where
L
1
NIG(, , , ).
(C3): Time-changing Brownian motion with an independent increas-
ing Levy process. Let W denote the standard Brownian motion; we
can construct a Levy process by replacing the (calendar) time t
by an independent increasing Levy process , therefore L
t
:= W
(t)
,
t 0. The process has the useful in Finance interpretation
as business time. Models constructed this way include the stan-
dard Brownian motion, the normal inverse Gaussian process, where
Brownian motion is time-changed with the inverse Gaussian process
(hence its name) and the variance gamma process, where Brownian
motion is time-changed with the gamma process.
20 ANTONIS PAPAPANTOLEON
Naturally, some processes can be constructed using more than one meth-
ods. Nevertheless, each method has some distinctive advantages which are
very useful in applications. The advantages of specifying a triplet (C1) are
that the characteristic function and the pathwise properties are known and
allows the construction of a rich variety of models; the drawbacks are that
parameter estimation and simulation (in the innite activity case) can be
quite involved. The second method (C2) allows the easy estimation and
simulation of the process; on the contrary the structure of the paths might
be unknown. The method of time-changes (C3) allows for easy simulation,
yet estimation might be quite dicult.
12. Simulation of L evy processes
We shall briey describe simulation methods for Levy processes. We con-
centrate on nite activity Levy processes (i.e. Levy jump-diusions) and
some special cases of innite activity Levy processes, namely the normal
inverse Gaussian and the variance gamma process. Here we do not discuss
methods for simulation of random variables with known density; various
simulation algorithms can be found in Devroye (1986) (also available online
at http://jeff.cs.mcgill.ca/~luc/rnbookindex.html). Moreover, sev-
eral speed-up methods for the Monte Carlo simulation of Levy processes are
discussed in Webber (2005).
12.1. Finite activity. Assume we want to simulate the Levy jump-diusion
L
t
= bt +W
t
+
N
t

k=1
J
k
where N
t
Poisson(t) and J F(dx).
We can simulate a discretized trajectory of the Levy jump-diusion L at
xed time points t
1
, . . . , t
n
as follows:
simulate a standard normal variate
transform it into a normal variate with variance t, where t =
t
i
t
i1
(denoted G
i
)
simulate a Poisson random variate with parameter t
simulate the law of jump sizes J, i.e. simulate F(dx)
if the Poisson variate is larger than zero, add the value of the jump.
The discretized trajectory is
L
t
i
= bt
i
+
i

j=1
G
i
+
N
t
i

k=1
J
k
.
12.2. Innite activity. We will describe simulation methods for two very
popular models in mathematical nance, the variance gamma and the nor-
mal inverse Gaussian process. These two processes can be easily simu-
lated because they are time-changed Brownian motions; we follow Cont
and Tankov (2003) closely. A general treatment of simulation methods for
innite activity Levy processes can be found in Cont and Tankov (2003) and
Schoutens (2003).
INTRODUCTION TO L

EVY PROCESSES 21
Assume we want to simulate a normal inverse Gaussian (NIG) process
with parameters , , ; we can simulate a discretized trajectory at xed
time points t
1
, . . . , t
n
as follows:
simulate n independent inverse Gaussian variables I
i
with parame-
ters
i
=
(t)
2

and
i
= t, where t = t
i
t
i1
, i = 1, . . . , n
simulate n standard normal variables G
i
set L
i
= I
i
+

I
i
G
i
The discretized trajectory is
L
t
i
=
i

j=1
L
j
.
Assume we want to simulate a variance gamma (VG) process with pa-
rameters , , ; we can simulate a discretized trajectory at xed time points
t
1
, . . . , t
n
as follows:
simulate n independent gamma variables
i
with parameter
t

where
t = t
i
t
i1
, i = 1, . . . , n
set
i
=
i
simulate n standard normal variables G
i
set L
i
=
i
+

i
G
i
The discretized trajectory is
L
t
i
=
i

j=1
L
j
.
13. Asset price model
We describe an asset price model driven by a Levy process, both under
the real and under the risk-neutral measure.
13.1. Real-world measure. Under the real-world measure, we model the
asset price process as the exponential of a Levy process, that is
S
t
= S
0
exp L
t
; (13.1)
here, L is the Levy process whose innitely divisible distribution has been
estimated from the data set available for the particular asset. Therefore,
the log-returns of the model have independent and stationary increments,
which are distributed along time intervals of specic length, for example
1 according to an innitely divisible distribution X, i.e. L
1
d
= X.
Naturally, the path properties of the process L carry over to S; if, for
example, L is a pure-jump Levy process, then S is also a pure-jump process.
This fact allows us to capture, up to a certain extend, the microstructure of
price uctuations, even on an intraday time scale.
The fact that the price process is driven by a Levy process, makes the
market, in general, incomplete; the only exceptions are the markets driven
by the Normal (Black-Scholes model) and Poisson distributions. Therefore,
there exists a large set of equivalent martingale measures, i.e. candidate
measures for risk-neutral valuation.
22 ANTONIS PAPAPANTOLEON
Eberlein and Jacod (1997) provide a thorough analysis and characteriza-
tion of the set of equivalent martingale measures; moreover, they prove that
the range of, e.g. call, option prices under all possible equivalent martingale
measures spans the whole no-arbitrage interval (i.e [(S
0
Ke
rT
)
+
, S
0
] for a
European call option with strike K). In addition, Selivanov (2005) discusses
the existence and uniqueness of martingale measures for exponential Levy
models in nite and innite time horizon and for various specications of
the no-arbitrage condition.
The Levy market can be completed using particular assets, such as mo-
ment derivatives (e.g. variance swaps), and then there exists a unique equiv-
alent martingale measure; see Corcuera, Nualart, and Schoutens (2005a,
2005b). For example, if an asset is driven by a Levy jump-diusion
L
t
= bt +W
t
+
N
t

k=1
J
k
(13.2)
where F(dx) 1, then the market can be completed using only variance
swaps on this asset.
The process (S
t
)
t0
is the solution of the stochastic dierential equation
dS
t
= S
t
(dL
t
+
1
2
ct +e
L
t
1 L
t
),
where L
t
= L
t
L
t
is the jump of L at time t. We could also specify S
as the solution to the following SDE
dS
t
= S
t
dL
t
,
whose solution is the stochastic exponential S
t
= S
0
c(L
t
). The second
approach is unfavorable for nancial applications, because (a) the asset price
can take negative values, unless jumps are restricted to be larger than 1
(i.e. supp() [1, )) and (b) the distribution of log-returns is not known.
In the special case of the Black-Scholes model, the two approaches coincide.
Remark 13.1. For the connection between the natural and stochastic ex-
ponential for Levy processes, we refer to Lemma A.8 in Goll and Kallsen
(2000).
13.2. Risk-neutral measure. Under the risk neutral measure, denoted by
P, we will model the asset price process as an exponential Levy process
S
t
= S
0
exp L
t
(13.3)
where the Levy process L must satisfy Assumptions (M) and (EM).
Assumption (EM). We assume that the Levy process L has a nite rst
exponential moment, i.e.
IE[e
L
t
] < .
There are various ways to choose the martingale measure such that it
is equivalent to the real-world measure. We refer to Goll and R uschendorf
(2001) for a unied exposition in terms of f-divergences of the dierent
methods for selecting an equivalent martingale measure (EMM). Note that,
some of the proposed methods to choose an EMM preserve the Levy property
INTRODUCTION TO L

EVY PROCESSES 23
of log-returns; examples are the Esscher transformation and the minimal
entropy martingale measure (cf. Esche and Schweizer 2005).
The market practice is to consider the choice of the martingale measure
as the result of a calibration to market data of vanilla options. Hakala
and Wystup (2002) describe the calibration procedure in detail; moreover,
Cont and Tankov (2004, 2005) and Belomestny and Rei (2005) present
numerically stable calibration methods for Levy driven models.
Because we have assumed that P is a risk neutral measure, the asset price
has mean rate of return r and the discounted and re-invested process
(e
(r)t
S
t
)
t0
, is a martingale under P. Here r 0 is the (domestic) risk-
free interest rate, 0 the continuous dividend yield (or foreign interest
rate) of the asset.
The process L has the canonical decomposition
L
t
= bt +W
t
+
t
_
0
_
R
x(
L

L
)(ds, dx) (13.4)
and hence, the drift term b equals the expectation of L
1
and can be written
as
b = r

2
2

_
R
(e
x
1 x)(dx). (13.5)
where 0 the diusion coecient and W a standard Brownian mo-
tion under P.
L
is the random measure of jumps of the process L and

L
(dt, dx) = (dx)dt is the compensator of the jump measure
L
, where
is the Levy measure of L
1
.
14. Popular models
14.1. Black-Scholes. The most famous asset price model based on a Levy
process is that of Samuelson (1965), Black and Scholes (1973) and Merton
(1973). The log-returns are normally distributed with mean and variance

2
, i.e. L
1
Normal(,
2
) and the density is
f
L
1
(x) =
1

2
exp
_

(x )
2
2
2
_
.
The characteristic function is

L
1
(u) = exp
_
iu

2
u
2
2
_
,
the rst and second moments are
E[L
1
] = , Var[L
1
] =
2
,
while the skewness and kurtosis are
skew[L
1
] = 0, kurt[L
1
] = 3.
The canonical decomposition of L is
L
t
= t +W
t
and the Levy triplet is (,
2
, 0).
24 ANTONIS PAPAPANTOLEON
14.2. Merton. Merton (1976) was the one of the rst to use a discontinuous
price process to model asset returns. The canonical decomposition of the
driving process is
L
t
= t +W
t
+
N
t

k=1
J
k
where J
k
Normal(
J
,
2
J
), k = 1, ..., hence the distribution of the jump
size has the density
f
J
(x) =
1

2
exp
_

(x
J
)
2
2
2
J
_
.
The characteristic function of L
1
is

L
1
(u) = exp
_
iu

2
u
2
2
+
_
e
i
J
u
2
J
u
2
/2
1
_
_
,
and the Levy triplet is (,
2
, f
J
).
The density of L
1
is not known in closed form, while the rst two moments
are
E[L
1
] = +
J
and Var[L
1
] =
2
+
2
J
+
2
J
14.3. Kou. Kou (2002) proposed a jump-diusion model similar to Mer-
tons, where the jump size is double-exponentially distributed. Therefore,
the canonical decomposition of the driving process is
L
t
= t +W
t
+
N
t

k=1
J
k
where J
k
DbExpo(p,
1
,
2
), k = 1, ..., hence the distribution of the jump
size has the density
f
J
(x) = p
1
e

1
x
1l
{x0}
+ (1 p)
2
e

2
x
1l
{x>0}
.
The characteristic function of L
1
is

L
1
(u) = exp
_
iu

2
u
2
2
+
_
p
1

1
iu

(1 p)
2

2
+iu
1
__
,
and the Levy triplet is (,
2
, f
J
).
The density of L
1
is not known in closed form, while the rst two moments
are
E[L
1
] = +
p

(1 p)

2
and Var[L
1
] =
2
+
p

2
1
+
(1 p)

2
2
.
14.4. Generalized Hyperbolic. The generalized hyperbolic model was in-
troduced by Eberlein and Prause (2002) following the works on the hyper-
bolic model by Eberlein and Keller (1995); the class of hyperbolic distribu-
tions was invented by O E. Barndor-Nielsen in relation the so-called sand
project (cf. Barndor-Nielsen 1977). The increments of time length 1 fol-
low a generalized hyperbolic distribution with parameters , , , , , i.e.
INTRODUCTION TO L

EVY PROCESSES 25
L
1
GH(, , , , ) and the density is
f
GH
(x) = c(, , , )
_

2
+ (x )
2
_
(
1
2
)/2
K

1
2
_

2
+ (x )
2
_
exp
_
(x )
_
,
where
c(, , , ) =
(
2

2
)
/2

1
2
K

2
_
and K

denotes the Bessel function of the third kind with index (cf.
Abramowitz and Stegun 1968). Parameter > 0 determines the shape,
0 [[ < determines the skewness, R the location and > 0 is a
scaling parameter. The last parameter, R aects the heaviness of the
tails and allows us to navigate through dierent subclasses. For example,
for = 1 we get the hyperbolic distribution and for =
1
2
we get the
normal inverse Gaussian (NIG).
The characteristic function of the GH distribution is

GH
(u) = e
iu
_

2

2
( +iu)
2
_
2
K

2
( +iu)
2
_
K

2
_ ,
while the rst and second moments are
E[L
1
] = +

2

K
+1
()
K

()
and
Var[L
1
] =

2

K
+1
()
K

()
+

2

2
_
K
+2
()
K

()

K
2
+1
()
K
2

()
_
,
where =
_

2
.
The canonical decomposition of a Levy process driven by a generalized
hyperbolic distribution (i.e. L
1
GH) is
L
t
= tE[GH] +
t
_
0
_
R
x(
L

GH
)(ds, dx)
and the Levy triplet is (E[GH], 0,
GH
); the Levy measure of the GH distri-
bution is known, but has a quite complicated expression, see Raible (2000,
section 2.4.1)
The GH distribution contains as special or limiting cases several known
distributions, including the normal, exponential, gamma, variance gamma,
hyperbolic and normal inverse Gaussian distributions; see Eberlein and
v. Hammerstein (2004).
14.5. Normal Inverse Gaussian. The normal inverse Gaussian distribu-
tion is a special case of the GH for =
1
2
; it was introduced to nance in
Barndor-Nielsen (1997). The density is
f
NIG
(x) =

exp
_

2
+(x )
_
K
1
_

_
1 + (
x

)
2
_
_
1 + (
x

)
2
,
26 ANTONIS PAPAPANTOLEON
while the characteristic function has the simplied form

NIG
(u) = e
iu
exp(
_

2
)
exp(
_

2
( +iu)
2
)
.
The rst and second moments of the NIG distribution are
E[L
1
] = +

_

2
and Var[L
1
] =

_

2
+

2

(
_

2
)
3
,
and similarly to the GH, the canonical decomposition is
L
t
= tE[NIG] +
t
_
0
_
R
x(
L

NIG
)(ds, dx),
where now the Levy measure has the form

NIG
(dx) = e
x

[x[
K
1
([x[)dx.
The NIG is the only subclass of the GH that is closed under convolution,
i.e. if X NIG(, ,
1
,
1
) and Y NIG(, ,
2
,
2
) and X is independent
of Y , then
X +Y NIG(, ,
1
+
2
,
1
+
2
).
Therefore, if we estimate the returns distribution at some time scale, then
we know it in closed form for all time scales.
15. Pricing European options
We review the idea of Sebastian Raible for the valuation of European
options using Laplace transforms; see chapter 3 and appendix B.1 in Raible
(2000). We assume that the following conditions are in force.
15.1. Summary of assumptions. The necessary assumptions regarding
the return distribution of the asset are:
(D1): Assume that
L
T
(z), the extended characteristic function of L
T
,
exists for all z C with z I
1
[0, 1].
(D2): Assume that P
L
T
, the distribution of L
T
, is absolutely contin-
uous w.r.t. the Lebesgue measure
\
with density .
The necessary assumptions regarding the payo function are:
(P1): Consider a European-style payo function f(S
T
) that is inte-
grable.
(P2): Assume that x e
Rx
[f(e
x
)[ is bounded and integrable for
all R I
2
R.
Finally, we need the following assumption involving both the payo func-
tion and the return distribution of the asset:
(B1): Assume that I
1
I
2
,= .
INTRODUCTION TO L

EVY PROCESSES 27
15.2. Raibles method.
Denition 15.1. Let L
h
(z) denote the bilateral Laplace transform of a
function h at z C, i.e. let
L
h
(z) :=
_
R
e
zx
h(x)dx.
The next Theorem gives an explicit expression for the value of an option
with payo function f() and driving process L.
Theorem 15.2. Assume that the above mentioned conditions are in force
and let g(x) := f(e
x
) denote the modied payo function of an option with
payo f(x) at time T. Choose an R I
1
I
2
. Letting V () denote the price
of this option, as a function of := log S
0
, we have
V () =
e
RrT
2
_
R
e
iu
L
g
(R +iu)
L
T
(iR u)du, (15.1)
whenever the integral on the r.h.s. exists (at least as a Cauchy principal
value).
Proof. According to arbitrage pricing, the value of an option is equal to its
discounted expected payo under the risk-neutral measure P. We get
V = e
rT
IE[f(S
T
)]
= e
rT
_

f(S
T
)dP
= e
rT
_
R
f(S
0
e
x
)dP
L
T
(x)
= e
rT
_
R
f(S
0
e
x
)(x)dx
because P
L
T
is absolutely continuous with respect to the Lebesgue measure
(by D2). Dene the function g(x) = f(e
x
) and let = log S
0
, then
V = e
rT
_
R
g( x)(x)dx = e
rT
(g )() (15.2)
which is a convolution of g with at the point , multiplied by the discount
factor.
The idea now is to apply a Laplace transform on both sides of (15.2) and
take advantage of the fact that the Laplace transform of a convolution equals
the product of the Laplace transforms of the factors. The resulting Laplace
transforms are easier to calculate analytically. Finally, we can invert the
Laplace transforms to recover the option value.
For the functions g and we have that x e
Rx
[g(x)[ is bounded,
_
R
e
Rx
[g(x)[dx <
28 ANTONIS PAPAPANTOLEON
for R I
2
(by P2) and
_
R
e
Rx
(x)dx =
L
T
(iR) <
for R I
1
(by D1). Therefore, for R I
1
I
2
(use B1) the prerequisites of
Theorem B.2 in Raible (2000) are met and we get that, the convolution in
(15.2) is continuous and absolutely convergent. Applying Theorem B.2, we
get, for z = R +iu, u R
L
V
(z) = e
rT
_
R
e
zx
(g )(x)dx
= e
rT
_
R
e
zx
g(x)dx
_
R
e
zx
(x)dx
= e
rT
L
g
(z)L

(z).
Moreover, from the same Theorem we have that V () is continuous and
_
R
e
R
[V ()[d < .
Hence, the prerequisites of Theorem B.3 in Raible (2000) are satised and
we can invert this Laplace transform to recover the option value.
V () =
1
2i
R+i
_
Ri
e
z
L
V
(z)dz
=
1
2
_
R
e
(R+iu)
L
V
(R +iu)du
=
e
R
2
_
R
e
iu
e
rT
L
g
(R +iu)L

(R +iu)du
=
e
rT+R
2
_
R
e
iu

L
T
(iR u)L

(R +iu)du.

15.3. Examples of options.

Example 15.3 (Plain vanilla option). A European plain vanilla call option
pays o f(S
T
) = (S
T
K)
+
, for some strike price K. The Laplace transform
of its modied payo function g (normalized for strike K = 1) is
L
g
(z) =
1
z(z + 1)
(15.3)
for z C with 'z = R I
2
= (, 1).
Similarly, for a European plain vanilla put option that pays o f(S
T
) =
(KS
T
)
+
, the Laplace transform of its modied payo function g (normal-
ized for strike K = 1) is given by (15.3) for z C with 'z = R I
2
= (0, ).
INTRODUCTION TO L

EVY PROCESSES 29
0.1
0.2
0.3
0.4
0.5
0.6
0.1
0.2
0.3
0.4
0.5
0.6
Figure 16.11. Densities of hyperbolic (red), NIG (blue) and
hyperboloid (green) distributions (left). Comparison of the
GH (red) and Normal (blue) distributions.
Example 15.4 (Digital option). A European digital call option pays o
f(S
T
) = 1l
{S
T
>K}
. The Laplace transform of its modied payo function g
is
L
g
(z) =
1
z
_
K
S
0
_
z
(15.4)
for z C with 'z = R I
2
= (, 0).
Similarly, for a European digital put option that pays o f(S
T
) = 1l
{S
T
<K}
,
the Laplace transform of its modied payo function g is
L
g
(z) =
1
z
_
K
S
0
_
z
(15.5)
for z C with 'z = R I
2
= (0, ).
15.4. Examples of distributions.
Example 15.5 (Normal). The Normal distribution has a (known) Lebesgue
density and its moment generating function M
N
(u) exists for all u R,
therefore we have that I
1
= R.
Example 15.6 (Generalized Hyperbolic). The Generalized Hyperbolic dis-
tribution has a (known) Lebesgue density and its moment generating func-
tion M
GH
(u) exists for all u ( , ), therefore we have that
I
1
= ( , ).
16. Empirical evidence
Levy processes provide a framework that can easily capture the empirical
observations both under the real world and the risk-neutral measure.
We provide here some indicative examples.
Under the real world measure, Levy processes can be generated by dis-
tributions that are exible enough to capture the observed fat-tailed and
skewed (leptokurtic) behavior of asset returns. One such class of distribu-
tions is the class of generalized hyperbolic distributions (cf. section 14.4). In
Figure 16.11, various densities of generalized hyperbolic distributions and a
comparison of a generalized hyperbolic and normal density are plotted.
A typical example of the behavior of asset returns can be seen in Figures
1.2 and 16.12. The tted normal distribution has lower peak, fatter anks
30 ANTONIS PAPAPANTOLEON
0.02 0.0 0.02
0
2
0
4
0
6
0
0.02 0.01 0.0 0.01 0.02

0
.0
2

0
.0
1
0
.0
0
.0
1
0
.0
2
quantiles of GH
e
m
p
iric
a
l q
u
a
n
tile
s
Figure 16.12. Empirical distribution of EUR/USD daily
log-returns with tted GH (red) and the corresponding Q-Q
plot.
and lighter tails than the empirical distribution; this means that, in real-
ity, tiny and large price movements occur more frequently, and small and
medium size movements occur less frequently, than predicted by the normal
distribution. On the other hand, the generalized hyperbolic distribution
gives a very good statistical t of the empirical distribution; this is further
veried by the corresponding Q-Q plot.
Under the risk-neutral measure, the exibility of the generating distri-
butions allows to accurately capture the shape of the implied volatility smile
and to a lesser extend the shape of the whole volatility surface.
17. Related literature
The following articles are recommended for further reading.
Eberlein and Keller (1995), Cont (2001), Eberlein (2001), Barndor-
Nielsen and Prause (2001), Carr et al. (2002), Eberlein and

Ozkan
(2003).
The following books are recommended for further reading.
Levy processes: Bertoin (1996), Sato (1999), Applebaum (2004),
Kyprianou (2005)
Levy processes with applications in nance: Cont and Tankov (2003),
Schoutens (2003)
Semimartingale theory: Protter (2004), Jacod and Shiryaev (2003)
Semimartingale theory with applications in nance: Shiryaev (1999).
Appendix A. Poisson random variables and processes
Denition A.1. Let X be a Poisson distributed random variable with pa-
rameter R
+
. Then, for n N the probability distribution is
P(X = n) = e

n
n!
and the rst two centered moments are
E[X] = and Var[X] = .
INTRODUCTION TO L

EVY PROCESSES 31
Denition A.2. An adapted c`adl`ag process N
t
: R
+
N0 is called
a Poisson process if
(1) N
0
= 0,
(2) N
t
N
s
is independent of T
s
for any 0 s < t < T,
(3) N
t
N
s
is Poisson distributed with parameter (t s) for any 0
s < t < T.
Then, 0 is called the intensity of the Poisson process.
Denition A.3. Let N be a Poisson process with parameter . We shall
call the process N
t
: R
+
R where
N
t
:= N
t
t (A.1)
a compensated Poisson process.
A simulated path of a Poisson and a compensated Poisson process can be
seen in Figure A.13.
0.0 0.2 0.4 0.6 0.8 1.0
0
.0
0
.5
1
.0
1
.5
2
.0
2
.5
0.0 0.2 0.4 0.6 0.8 1.0
0
.0
0
.1
0
.2
0
.3
Figure A.13. Plots of the Poisson (left) and compensated
Poisson process.
Proposition A.4. The compensated Poisson process dened by (A.1) is a
martingale.
Proof. We have
(1) The process N is adapted to the ltration because N is adapted (by
denition),
(2) IE[[N
t
[] < because IE[[N
t
[] < ,
(3) Let 0 s < t < T, then
IE[N
t
[T
s
] = IE[N
t
t[T
s
]
= IE[N
s
(N
t
N
s
)[T
s
] (s (t s))
= N
s
IE[(N
t
N
s
)[T
s
] (s (t s))
= N
s
s
= N
s
.

32 ANTONIS PAPAPANTOLEON
Remark A.5. The characteristic functions of the Poisson and compensated
Poisson processes are respectively
IE[e
iuN
t
] = exp
_
t(e
iu
1)
_
and
IE[e
iuN
t
] = exp
_
t(e
iu
1 iu)
_
.
Appendix B. Compound Poisson random variables
Let N be a Poisson distributed random variable with parameter > 0
and J = (J
k
)
k1
an i.i.d. sequence of random variable with law F. Then, by
conditioning on the number of jumps and and using independence, we have
that the characteristic function of a compound Poisson distributed random
variable is
IE[e
iu
P
N
k=1
J
k
] =

n0
IE
_
e
iu
P
N
k=1
J
k

N = n
_
P(N = n)
=

n0
IE
_
e
iu
P
n
k=1
J
k
_
e

n
n!
=

n0
_
_
_
R
e
iux
F(dx)
_
_
n
e

n
n!
= exp
_
_

_
R
(e
iux
1)F(dx)
_
_
.
Appendix C. Notation
d
= equality in law
a b = mina, b
a b = maxa, b
Let C z = x +iy, x, y R. Then 'z = x and z = y.
1l
A
(x) =
_
1, x A,
0, x / A.
Appendix D. Datasets
The EUR/USD implied volatility data are from 5 November 2001. The
spot price was 0.93, the domestic rate (USD) 5% and the foreign rate (EUR)
4%. The data are available at
http://www.mathfinance.de/FF/sampleinputdata.txt.
The USD/JPY, EUR/USD and GBP/USD foreign exchange time se-
ries correspond to noon buying rates (dates: 22/10/1997 22/10/2004,
4/1/993/3/2005 and 1/5/20023/3/2005 respectively). The data can be
downloaded from
http://www.newyorkfed.org/markets/foreignex.html.
INTRODUCTION TO L

EVY PROCESSES 33
References
Abramowitz, M. and I. Stegun (Eds.) (1968). Handbook of mathematical
functions (5th ed.). Dover.
Applebaum, D. (2004). Levy processes and stochastic calculus. Cambridge
University Press.
Barndor-Nielsen, O. E. (1977). Exponentially decreasing distributions
for the logarithm of particle size. Proc. Roy. Soc. London A 353, 401
419.
Barndor-Nielsen, O. E. (1997). Normal inverse Gaussian distributions
and stochastic volatility modelling. Scand. J. Statist. 24, 113.
Barndor-Nielsen, O. E., T. Mikosch, and S. Resnick (Eds.) (2001). Levy
processes: theory and applications. Birkhauser.
Barndor-Nielsen, O. E. and K. Prause (2001). Apparent scaling. Finance
Stoch. 5, 103 113.
Belomestny, D. and M. Rei (2005). Optimal calibration for exponential
Levy models. WIAS Preprint No. 1017.
Bertoin, J. (1996). Levy processes. Cambridge University Press.
Black, F. and M. Scholes (1973). The pricing of options and corporate
liabilities. J. Polit. Econ. 81, 637654.
Carr, P., H. Geman, D. B. Madan, and M. Yor (2002). The ne structure
of asset returns: an empirical investigation. J. Business 75, 305332.
Cont, R. (2001). Empirical properties of asset returns: stylized facts and
statistical issues. Quant. Finance 1, 223236.
Cont, R. and P. Tankov (2003). Financial modelling with jump processes.
Chapman and Hall/CRC Press.
Cont, R. and P. Tankov (2004). Nonparametric calibration of jump-
diusion option pricing models. J. Comput. Finance 7(3), 149.
Cont, R. and P. Tankov (2005). Retrieving Levy processes from option
prices: regularization of an ill-posed inverse problem. SIAM J. Control
Optim.. (forthcoming).
Corcuera, J. M., D. Nualart, and W. Schoutens (2005a). Completion of a
Levy market by power-jump assets. Finance Stoch. 9, 109127.
Corcuera, J. M., D. Nualart, and W. Schoutens (2005b). Moment
derivatives and Levy-type market completion. In A. Kyprianou,
W. Schoutens, and P. Wilmott (Eds.), Exotic option pricing and ad-
vanced Levy models, pp. 169193. Wiley.
Devroye, L. (1986). Non-uniform random variate generation. Springer.
Eberlein, E. (2001). Application of generalized hyperbolic Levy mo-
tions to nance. In O. E. Barndor-Nielsen, T. Mikosch, and S. I.
Resnick (Eds.), Levy Processes: theory and applications, pp. 319336.
Birkhauser.
Eberlein, E. and J. Jacod (1997). On the range of options prices. Finance
Stoch. 1, 131140.
Eberlein, E. and U. Keller (1995). Hyperbolic distributions in nance.
Bernoulli 1, 281299.
Eberlein, E. and F.

Ozkan (2003). Time consistency of Levy models.
Quant. Finance 3, 4050.
34 ANTONIS PAPAPANTOLEON
Eberlein, E. and A. Papapantoleon (2005). Symmetries and pricing of
exotic options in Levy models. In A. Kyprianou, W. Schoutens, and
P. Wilmott (Eds.), Exotic option pricing and advanced Levy models,
pp. 99128. Wiley.
Eberlein, E. and K. Prause (2002). The generalized hyperbolic model:
nancial derivatives and risk measures. In H. Geman, D. Madan,
S. Pliska, and T. Vorst (Eds.), Mathematical Finance-Bachelier Con-
gress 2000, pp. 245267. Springer Verlag.
Eberlein, E. and E. A. v. Hammerstein (2004). Generalized hyperbolic
and inverse Gaussian distributions: limiting cases and approximation
of processes. In R. Dalang, M. Dozzi, and F. Russo (Eds.), Seminar
on Stochastic Analysis, Random Fields and Applications IV, Progress
in Probability 58, pp. 221264. Birkhauser.
Esche, F. and M. Schweizer (2005). Minimal entropy preserves the Levy
property: how and why. Stochastic Process. Appl. 115, 299327.
Goll, T. and J. Kallsen (2000). Optimal portfolios for logarithmic utility.
Stochastic Process. Appl. 89, 3148.
Goll, T. and L. R uschendorf (2001). Minimax and minimal distance mar-
tingale measures and their relationship to portfolio optimization. Fi-
nance Stoch. 5, 557581.
Hakala, J. and U. Wystup (2002). Hestons stochastic volatility model ap-
plied to foreign exchange options. In J. Hakala and U. Wystup (Eds.),
Foreign Exchange Risk. Risk Publications.
Jacod, J. and A. N. Shiryaev (2003). Limit theorems for stochastic pro-
cesses (2nd ed.). Springer.
Kallsen, J. (2000). Optimal portfolios for exponential Levy processes.
Math. Meth. Oper. Res. 51, 357374.
Kou, S. G. (2002). A jump diusion model for option pricing. Manag.
Sci. 48, 10861101.
Kyprianou, A. E. (2005). Introductory lectures on uctuations of Levy
processes with applications. Lect. Notes Math. (forthcoming). Springer.
Merton, R. C. (1973). Theory of rational option pricing. Bell J. Econ.
Manag. Sci. 4, 141183.
Merton, R. C. (1976). Option pricing with discontinuous returns. Bell J.
Financ. Econ. 3, 145166.
Prabhu, N. U. (1998). Stochastic storage processes (2nd ed.). Springer.
Protter, P. (2004). Stochastic integration and dierential equations (3rd
ed.). Springer.
Raible, S. (2000). Levy processes in nance: theory, numerics, and em-
pirical facts. Ph. D. thesis, University of Freiburg.
Samuelson, P. A. (1965). Rational theory of warrant pricing. Indust.
Manag. Rev. 6, 1331.
Sato, K.-I. (1999). Levy processes and innitely divisible distributions.
Cambridge University Press.
Schoutens, W. (2003). Levy processes in nance: pricing nancial deriva-
tives. Wiley.
Selivanov, A. V. (2005). On the martingale measures in exponential Levy
models. Theory Probab. Appl. 49, 261274.
INTRODUCTION TO L

EVY PROCESSES 35
Shiryaev, A. N. (1999). Essentials of stochastic nance: facts, models,
theory. World Scientic.
Webber, N. (2005). Simulation methods with Levy processes. In A. Kypri-
anou, W. Schoutens, and P. Wilmott (Eds.), Exotic option pricing and
advanced Levy models, pp. 2949. Wiley.
Department of Mathematical Stochastics, University of Freiburg, D79104
Freiburg, Germany
E-mail address: papapan@stochastik.uni-freiburg.de
URL: http://www.stochastik.uni-freiburg.de/~papapan