IEOR E4707: Financial Engineering: Continuous-Time Models Fall 2013

⃝c 2013 by Martin Haugh

Local Volatility, Stochastic Volatility and

Jump-Diffusion Models
These notes provide a brief introduction to local and stochastic volatility models as well as jump-diffusion
models. These models extend the geometric Brownian motion model and are often used in practice to price
exotic derivative securities. It is worth emphasizing that the prices of exotics and other non-liquid securities are
generally not available in the market-place and so models are needed in order to both price them and calculate
their Greeks. This is in contrast to vanilla options where prices are available and easily seen in the market. For
these more liquid options, we only need a model, i.e. Black-Scholes, and the volatility surface to calculate the
Greeks and perform other risk-management tasks.
In addition to describing some of these models, we will also provide an introduction to a commonly used fourier
transform method for pricing vanilla options when analytic solutions are not available. This transform method is
quite general and can also be used in any model where the characteristic function of the log-stock price is
available. Transform methods now play a key role in the numerical pricing of derivative securities.

1 Local Volatility Models

The GBM model for stock prices states that
dSt = µSt dt + σSt dWt
where µ and σ are constants. Moreover, when pricing derivative securities with the cash account as numeraire,
we know that µ = r − q where r is the risk-free interest rate and q is the dividend yield. This means that we
have a single free parameter, σ, which we can fit to option prices or, equivalently, the volatility surface. It is not
all surprising then that this exercise fails. As we saw before, the volatility surface is never flat so that a constant
σ fails to re-produce market prices. This was particularly true after the crash of ′ 87 when market participants
began to correctly identify that lower strike options should be priced with a higher volatility, i.e. there should be
a skew.
After this crash, researchers developed alternative models in an attempt to model the skew. While not the first1
such model, the local volatility model is probability the simplest extension of Black-Scholes. It assumes that the
stock’s risk-neutral dynamics satisfy
dSt = (r − q)St dt + σl (t, St )St dWt (1)
so that the instantaneous volatility, σl (t, St ), is now a function of time and stock price. The key result2 of the
local volatility framework is the Dupire formula that links the local volatilities, σl (t, St ), to the implied volatility
surface.
Theorem 1 (The Dupire Formula) Let C = C(K, T ) be the price of a call option as a function of strike
and time-to-maturity. Then the local volatility function satisfies
∂C
+ (r − q)K ∂C
+ qC
σl2 (T, K) = ∂T
K2 ∂2C
∂K
. (2)
2 ∂K 2
1 Earlier models included Merton’s jump-diffusion model, the CEV model and Heston’s stochastic volatility model. Indeed

the first two of these models date from the 1970’s.

2 The local volatility framework was developed by Derman and Kani (1994) and in continuous time by Dupire (1994).
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 2

Proof: Recall first Kolmogorov’s forward equation for the PDF of the underlying stock. In particular, if p(y, t)
is the PDF of the stock price at time t and evaluated at St = y, then the forward equation is
1
−pt − (r − q)(yp)y + (σ 2 (t, y)y 2 p)yy = 0 for t > 0 (3)
2
with initial condition p(y, t) = δS0 (y) at t = 0. We can also write the call option price as
∫ ∞
−rT
[ ] −rT
C(K, T ) = e E0 (ST − K) +
= e (y − K)p(y, T ) dy (4)
K

and if we differentiate across (4) twice with respect to K we obtain

CKK (K, T ) = e−rT p(K, T ). (5)

Differentiating across (5) with respect to T yields

CKKT (K, T ) + rCKK (K, T ) = e−rT pT (K, T )

( )
1 2
= e−rT (σ (T, K)K 2 p)KK − (r − q)(Kp)K (6)
2
1 2
= (σ (T, K)K 2 CKK )KK − (r − q)(KCKK )K (7)
2
where (6) follows from (3) and (7) follows from substituting CKK for e−rT p(K, T ). Integrating across (7) with
respect to K yields
1 2
CKT (K, T ) + rCK (K, T ) − (σ (T, K)K 2 CKK )K + (r − q)(KCKK ) = h(T ) (8)
2
for some function, h(T ). Since KCKK = (KCK )K − CK , we can rewrite (8) as
1 2
CKT (K, T ) − (σ (T, K)K 2 CKK )K + r(KCK )K − q((KCK )K − CK ) = h(T ). (9)
2
Integrating across (9) again with respect to K yields
1 2
CT (K, T ) − σ (T, K)K 2 CKK + r(KCK ) − q(KCK − C) = h(T )K + g(T ). (10)
2
for some function, g(T ). Now note that as K → ∞ the call option price as well as its derivatives, CT , CK and
CKK all go to zero. But (10) then implies that h(T ) = g(T ) = 0 for all T after which (10) reduces to (2).

Given the implied volatility surface we can easily compute the corresponding call option price surface which is
the graph of C(K, T ) as a function of K and T . It is then clear from (2) that we need to take first and second
derivatives of this latter surface with respect to strike and first derivatives with respect to time-to-maturity in
order to compute the local volatilities. Calculating the local volatilities from (2) is therefore difficult and can be
unstable as computing derivatives numerically can itself be very unstable. As a result, it is necessary to use a
sufficiently smooth Black-Scholes implied volatility3 surface when calculating local volatilities using (2).

Remark 1 It is worth emphasizing that the local volatility model (1) with σl (·, ·) computed according to (2)
is, by construction, a self-consistent model that is capable of producing the implied volatility surface observed in
the market place.
3 It is also possible to write the Dupire formula in terms of the implied volatilities rather than the call option prices. One

can then work directly with the implied volatility surface to compute the local volatilities.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 3

Local volatility is known to suffer from several weaknesses. For example, it leads to unreasonable skew dynamics
and underestimates the volatility of volatility or “vol-of-vol”. Moreover the Greeks that are calculated from a
local volatility model are generally not consistent with what is observed empirically. Nevertheless, it is an
interesting model from a theoretical viewpoint and is often used in practice for pricing barrier4 options for
example.

Gyöngy’s Theorem
Gyöngy’s Theorem is an important theoretical result that links local volatility models to other diffusion models
that are also capable of generating the implied volatility surface. Consider a general n-dimensional Itô process,
Xt , satisfying
dXt = α(t, ω) dt + β(t, ω) dWt
where α(t, ω) and β(t, ω) are n × 1 and n × m adapted processes, respectively, and ω is a sample path of the
m-dimensional Brownian motion, Wt . Then Gyöngy’s Theorem states that there is a Markov process, Yt ,
satisfying
dYt = a(t, Yt ) dt + b(t, Yt ) dWt
where Xt and Yt have the same marginal distributions, i.e. Xt and Yt have the same distribution for each t.
Moreover, Yt can be constructed by setting

a(t, y) = E0 [α(t, ω)|Xt = y] and

[ ]
b(t, y)b(t, y)T = E0 β(t, ω)β T (t, ω)|Xt = y .

In a financial setting, Xt might represent the true risk-neutral dynamics of a particular security. Then b(t, y)/y
represents the local volatility function σl (t, ·) in (1). Because Xt and Yt have the same marginal distributions
then we know (why?) that European option prices can be priced correctly if we assume the price dynamics are
given by Yt . In particular Yt can produce the correct implied volatility surface. Moreover Gyöngy’s Theorem
therefore implies that the local volatility model of (1) is in some sense the simplest diffusion model capable of
doing this, i.e. reproducing the implied volatility surface. Gyöngy’s Theorem has been used recently to develop
stochastic-local volatility models as well as approximation techniques for pricing various types of basket options.

1.1 The CEV Model

The constant elasticity of variance (CEV) model is a particular parametric local volatility model that was
introduced by Cox (1975). The risk-neutral dynamics are assumed to follow

dSt = (r − q)St dt + σStβ dWt (11)

where as usual q is the dividend yield, r is the risk-free rate and σ and β ∈ [0, 1] are the remaining model
parameters. Note that the CEV model generalizes GBM which is obtained when we set β = 1. The popularity of
the CEV model is due to its tractability. In particular, analytic5 expressions for options prices are available in
terms of the non-central χ2 distribution. This tractability also accounts for its use in term structure modeling,
often in conjunction with LIBOR market models.
By writing (11) as
dSt
= (r − q) dt + σStβ−1 dWt
St
we see that there is a negative relationship between price level and instantaneous volatility when β < 1. The
CEV model is therefore able to capture some of the skew that is observed empirically in practice. It is also worth
noting that when β < 1/2, there is a strictly positive probability that the CEV process will hit zero.
4 While the Black-Scholes GBM framework can be used to barrier options analytically, it is well known that the Black-

Scholes model is in fact a truly awful model for barrier options and that it should never be used in practice. As a result more
sophisticated models and numerical methods such as PDE or Monte-Carlo methods are used. We will return to barrier options
at a later stage to highlight the danger of using just one model to price exotic options.
5 See Cox (1975) but also Schroder (1989).
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 4

1.2 How Well Do Local Volatility Models Work in Practice

A local volatility model that is calibrated using the Dupire formula and he market’s implied volatility surface will
of course perfectly capture the observed skew at any maturity. A more interesting question is whether or not the
dynamics implied by (1) are an accurate representation of the true dynamics. Leaving aside the inability of any
diffusion model to capture jumps, the answer must be “no”. Gyöngy’s Theorem provides some justification for
this answer in that it suggests that the local volatility model is in some sense only a projection of a richer and
possibly more accurate diffusion model. This projection property also suggests that local volatility models
underestimate the volatility of volatility, also known as the vol-of-vol. Moreover this observation has been borne
out in practice where it has been found that local volatility models often significantly underprice structured
products. This is also related to a phenomenon whereby local volatility models generate forward skews that are
too flat. As such, they are unsuited for pricing securities such as cliquet options that are sensitive to the forward
skew.
Nonetheless, local volatility models are often used in practice to price various types of barrier options, for
example. In order to account for the possibility of jumps and the inability to hedge these jumps, various ad-hoc
adjustments are often made to the local volatility prices. There has also been some controversy6 regarding the
accuracy of deltas that are obtained from local volatility models. Indeed, users of local volatility models will
often persist in using Black-Scholes deltas to hedge rather than the deltas implied by their local volatility model.

2 Stochastic Volatility Models

The most well-known and important stochastic volatility model is due to Heston (1989) and in this section we
will concentrate exclusively on this model. It is a two-factor model and assumes separate dynamics for both the
stock price and instantaneous volatility. In particular, it assumes
√ (s)
dSt = (r − q)St dt + σt St dWt (12)
√ (vol)
dσt = κ (θ − σt ) dt + γ σt dWt (13)
(s) (vol)
where Wt and Wt are standard Q-Brownian motions with constant correlation coefficient, ρ. In practice, ρ
is generally (why?) negative.

Remark 2 The volatility process in (13) is commonly used in interest rate modeling where it is known as the
CIR7 model. It has the property that the process will remain non-negative with probability one. For certain
parameter combinations, it will always be strictly positive with probability one.

Whereas the local volatility model is a complete model, Heston’s stochastic volatility model is an incomplete8
(s)
model. This should not be too surprising as there are two sources of uncertainty in the Heston model, Wt and
(vol)
Wt , but only one risky security and so not every security is replicable. Put another way, while the drift in
(12) must be r − q under any EMM with the cash account as numeraire, we could use Girsanov’s Theorem to
change the drift in (13) in infinitely many different ways without changing the drift in (12).
To see this let us first suppose that the P -dynamics of St and σt satisfy
√ (1)
dSt = µt St dt + σt St dWt (14)
√ ( √ )
(1) (2)
dσt = νt dt + γ σt ρdWt + 1 − ρ2 dWt (15)
(1) (2)
where µt and νt are some Ft -adapted processes, and Wt = (Wt , Wt ) is a standard 2-dimensional
P -Brownian motion. Let us now define
( ∫ t ∫ )
1 t ′
Lt := exp − ηs′ dWs − ηs ηs ds
0 2 0
6 The SABR model was introduced to overcome the problems associated with the deltas of local volatility models.
7 AfterCox, Ingersoll and Ross (1985) who used this model for modeling the dynamics of the short interest rate.
8 Assuming as usual that the stock and the cash-account are the only traded securities.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 5
(1) (2)
for t ∈ [0, T ] and where ηt = (ηt , ηt ) is a 2-dimensional adapted process. Then9 Girsanov’s Theorem implies
∫
ct := Wt + t ηs ds is a standard 2-dimensional Qη -Brownian motion where dQη /dP = LT . In particular the
W 0
Qη -dynamics of St and σt satisfy
( √ (1) ) √
dSt = µt − σt ηt St dt + σt St dW ct(1) (16)
( √ √ √ ) √ ( √ )
dσt =
(1)
νt − γ σt ρηt − γ σt 1 − ρ2 ηt
(2)
dt + γ σt ρdW ct(1) + 1 − ρ2 dWct(2) . (17)

In order for Qη to be an EMM it is only necessary (why?) that

√ (1)
µt − σt ηt = r − q
(2)
and we are free to choose ηt subject only to the usual integrability constraints. This means that there are
(2)
infinitely many EMM’s and so10 the model is incomplete. In the Heston model we choose ηt so that
√ (1) √ √ (2)
νt − γ σt ρηt − γ σt 1 − ρ2 ηt = κ (θ − σt )

is satisfied. We therefore recover (12) and (13) once we identify W ct(1) with Wt(s) and Wt(vol) with (via Levy’s
√
Theorem) ρW ct(1) + 1 − ρ2 Wct(2) . Note that we still have several free parameters which in practice we would
determine by calibrating the model to the market prices of European options. This is the typical method of
choosing an EMM in incomplete market models.

2.1 The Pricing PDE

Because Heston’s model is incomplete it is not possible to price options using the replication arguments that
apply to complete market models. Instead, one assumes Q-dynamics as in (12) and (13) and then prices all
securities using this EMM. (The model parameters are chosen by calibrating model prices to observable market
prices.) The pricing PDE that the price, C(t, St , σt ), of any derivative security must satisfy in Heston’s model is
given by

∂C 1 ∂2C ∂2C 1 ∂2C ∂C ∂C

+ σS 2 2 + ρσγS + γ 2 σ 2 + (r − q)S + κ (θ − σ) = rC. (18)
∂t 2 ∂S ∂S∂σ 2 ∂σ ∂S ∂σ

This PDE follows of course via Feynman-Kac and the fact that
[ ]
M (t, St , σt ) := e−rt EQ
t e −r(T −t)
Option Payoff = e−rt C(t, St , σt ) (19)

must be (why?) a Q-martingale. Derivative prices can then be obtained by solving (18) subject to the relevant
boundary conditions or by using Monte-Carlo methods to estimate (19).
Heston succeeded in solving (18) in the case of European call options (and therefore put options via put-call
parity) by conjecturing a solution of the form
C(t, St , σt ) = St P1 (t, St , σt ) − Ke−r(T −t) P2 (t, St , σt ) (20)
where K is the option strike, T > t is the option maturity, and P1 and P2 are functions to be determined. Each
of the two terms on the right-hand-side of (20) must (why?) also satisfy (18). Substituting each of them in turn
into (18) leads to a corresponding PDE and terminal condition for Pj , for j = 1, 2. It is not possible to solve
these PDE’s for P1 and P2 in closed form but Heston was able to compute their Fourier transforms by guessing
their functional forms and then reducing each PDE to a series of two ODEs which could be solved analytically.
These transforms could then be inverted numerically to obtain the price of the call option via (20). Heston was
also able to interpret P1 and P2 as risk-neutral probabilities (with respect to different EMMs) of the option
expiring in the money. This observation should not11 be surprising and indeed it holds more generally.
9 We are assuming the necessary conditions. e.g. Novikov’s condition, to ensure that Lt is a martingale.
10 But we could make the model complete by introducing into the model another security whose price process depends on σt .
11 See also Exercise 16.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 6

Table 1: Call Option Price Estimates Using an Euler Scheme for Heston’s Stochastic Volatility Model.
The true option price is 13.085.

Time Steps Sticky Zero Reflection

100 28.3 45.1
200 27.1 41.3
500 25.6 37.1
1000 24.8 34.6

2.2 Simulating the Heston Model

Suppose we need to simulate a multi-dimensional SDE of the form

dXt = µ(t, Xt ) dt + σ(t, Xt ) dWt (21)

where Wt is an n-dimensional Brownian motion. Then the simplest discretization scheme for doing so is the
so-called Euler scheme which satisfies
( ) ( ) √
bkh = X
X b(k−1)h + µ (k − 1)h, X
b(k−1)h h + σ (k − 1)h, X b(k−1)h h Zk (22)

where the Zk ’s are MVNn (0, In ). We can then simulate {Xbh , X

b2h , . . . , X
bmh }, where m is the number of time
steps, h is a constant and mh = T . The smaller the value of h, the closer the discretized path in (22) will be to
the continuous-time path of (21) that we wish to simulate. Of course this will be at the expense of greater
computational effort.
Special care must be taken, however, if we wish to simulate the dynamics in (12) and (13). In fact the Euler
scheme of (22) does not always converge well when the dynamics are as in (13), even when we take m very
large. For example, Andersen12 (2007) considers the problem of pricing an at-the-money 10-year call option
(s) (vol)
when r = q = 0. He takes κ = .5, σ0 = θ = .04, γ = 1, S0 = 100 and ρ := Corr(Wt , Wt ) = −0.9. The
true value of the option is 13.085. Using one million sample paths and a “sticky zero” or “reflection”
assumption13 , he obtains the estimates displayed in the table above for the option price as a function of m, the
number of discretization points. It is clear that neither scheme has converged even when taking as many as
1, 000 discretization points.
One therefore needs to be very careful when applying an Euler scheme to the Heston SDE and indeed to
jump-diffusion models where the diffusion component has the dynamics of (13). In fact when simulating Heston
dynamics it is much better to use an alternative scheme such as that proposed by Andersen (2007). More
generally, it is a good idea to price vanilla securities in the Monte-Carlo alongside the more exotic securities that
are of direct interest. If the estimated vanilla security prices are not comparable to their analytic prices, then we
know the scheme has not converged.

2.3 The Characteristic Function of the log-Stock Price

We will see in Section 4.1 that the characteristic function, i.e. the Fourier transform, of the log-stock price is
very useful for pricing options. For the Heston model of (12) and (13), the characteristic function satisfies

ϕT (u) = E[exp(iulog(ST ) | S0 , σ0 )]
= exp(iu(log(S0) + (r − q)T ))
× exp(θκγ −2 ((κ − ργui − d)T − 2 log((1 − g exp(−dT ))/(1 − g))))
× exp(σ02 γ −2 (κ − ργui − d)(1 − exp(−dT ))/(1 − g exp(−dT ))) (23)
12 “Simple, Efficient, and Robust Simulation of Stochastic Volatility Models, ICBI (2007) conference presentation.
13 The sticky zero assumption simply means that anytime the variance process, Vt , goes negative in the Monte-Carlo it is
replaced by 0. The reflection assumption replaces Vt with |Vt |. In the limit as m → ∞, the variance will stay non-negative with
probability 1 so both assumptions are unnecessary in the limit and the option prices should be identical.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 7

where i = 0 + 1i is imaginary and

√
d = (ργui − κ)2 + γ 2 (iu + u2 )
g = (κ − ργui − d)/(κ − ργui + d).

It is worth noting that there is another equivalent14 representation, ϕbT (u) say, of the characteristic function that
is very similar to ϕT (u) in (23). While this representation is correct, using the principal value of the log function
in ϕbT (u) causes ϕbT (u) to jump discontinuously when the imaginary component of the argument of the log
function crosses the negative real axis. As a result, when ϕbT (u) is used in a numerical integration scheme such
as (62) to price options, wildly inaccurate prices can be obtained. The representation in (23) avoids this problem.

2.4 How Well Does Heston’s Model Capture the Skew?

An interesting question that arises is whether or not Heston’s model accurately represents the dynamics of stock
prices. This question is often reduced in practice to the less demanding question of how well the Heston model
captures the volatility skew. By “capturing” the skew we have in mind the following: once the Heston model
has been calibrated, then European option prices can be computed using numerical techniques such as
Monte-Carlo, PDE or transform methods. The resulting option prices can then be used to determine the
corresponding Black-Scholes implied volatilities. These volatilities can then be graphed to create the model’s
implied volatility surface which can then be compared to the market’s implied volatility surface.

Figure 1: An Implied Volatility Surface under Heston’s Stochastic Volatility Model

Figure√1 displays the implied volatility surface for the following choice of parameters: r = .03, q = 0,
σ0 = .0654, γ = .2928, ρ = −.7571, κ = .6067 and θ = .0707. Perhaps the most noticeable feature of this
surface is the persistence of the skew for long-dated options. Indeed the Heston model generally captures
longer-dated skew quite well but it typically struggles to capture the near term skew, particularly when the latter
is very steep. The problem with a steep short-term skew is that any diffusion model will struggle to capture it as
there is not enough time available for the stock price to diffuse sufficiently far from its current level. In order to
solve this problem jumps are needed.

Note that some instruments can be priced analytically in Heston’s model. For example, the price of the
continuous-time version of a variance swap has a closed-form solution as the following example demonstrates.
14 Heston’s original paper as well as most papers in the literature report this alternative representation. See Section 4.6 of

Applications of Fourier Transform to Smile Modeling (2010), Springer, by J.Zhu for a discussion of this problem and other
possible remedies.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 8

Example 1 (Variance-swaps in Heston’s model)

Let Vc (0, T ) denote the total continuous realized variance in the Heston model from t = 0 to t = T . Then we
know from (13) that (see Exercise 3)
[∫ ]
T
EQ
0 [Vc (0, T )] = EQ
0 σt dt
0
∫ T
= EQ
0 [σt ] dt (24)
0
∫ T ( −κt )
= e (σ0 − θ) + θ dt (25)
0
1 − e−κT
= (σ0 − θ) + θT
κ
so that the annualized variance is given by

EQ
0 [Vc (0, T )] 1 − e−κT
= (σ0 − θ) + θ. (26)
T κT

Note that the fair price of a variance-swap in Heston’s model does not depend on γ or ρ. This should not be
too surprising.

3 An Introduction to Jump-Diffusion Models

We now give a brief introduction15 to jump-diffusion models.
Definition 1 We say that Jt is a pure jump process if it is constant between jumps and is adapted and
right-continuous.
Typical examples of pure jump processes are Poisson processes, compound Poisson processes as described in
Example 2 and point processes more generally.
Example 2 (A Compound Poisson Process)
Let Nt be a Poisson process with intensity λ. Then

∑
Nt
Xt := Yi
i=0

where the Yi ’s are IID random variables is a compound Poisson process. It is easy to check that E[Xt ] = λtµy
where µy := E[Y ] and that Var(Xt ) = λtE[Y 2 ]. It is also easy to check that Mt := Xt − λµy t is a martingale
and that the moment generating function (MGF) of Xt is given by

ϕXt (u) := eλt(ϕY (u)−1) (27)

where ϕY (u) is the MGF of Y .

We now state16 an important result that aids our understanding of compound processes and jump-diffusions
more generally. It effectively states that if we have a compound Poisson process with a finite number of possible
jump sizes, then we can view this process equivalently as a sum of independent Poisson processes in which the
size one jumps are replaced by jumps of a fixed size.
15 This introduction is taken from Chapter 11 of Stochastic Finance for Finance II: Continuous-Time Models by Steve Shreve.

This chapter is an excellent introduction and contains far more material than we can cover in these notes.
16 This is Corollary 11.3.4 in Shreve which also contains an outline proof of the result.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 9

Theorem 2 Let y1 , . . . , yM be a finite set of nonzero numbers and let p1 , . . . , pM be positive numbers that
sum to 1. Let N (t) be a Poisson process with intensity λ and define the compound Poisson process

∑
N (t)
Xt := Yi
i=1

where the Yi ’s are an IID sequence of random variables with P (Yi = yj ) = pj for all i, j. For m = 1, . . . , M , let
Nm (t) be the number of jumps of size ym up to and including time t. Then

∑
M ∑
M
N (t) = Nm (t) and Xt = ym Nm (t).
m=1 m=1

The processes N1 , . . . , NM are independent Poisson processes and each Nm has intensity λpm .

Definition 2 A jump-diffusion or jump process is a process of the form

∫ t ∫ t
Xt = X0 + γs dWs + θs ds + Jt (28)
0 0
=: Xtc + Jt

where Jt is a pure jump process and Xtc is the continuous part of Xt . Of course, γs and θs are adapted
processes and Ws is a standard Brownian motion.

Note that definition 1 implies that jump-diffusion processes as defined in (28) are right-continuous. This means
that Xt = lims↓t Xs . We therefore use Xt− to denote the left-continuous version of the process so that

Xt− := lim Xs .
s↑t

The difference between Xt and Xt− is then the jump-size, ∆Xt = ∆Jt , at time t. In the case of the
jump-diffusion of (28) we see that
∫ t ∫ t
Xt− = X0 + γs dWs + θs ds + Jt−
0 0

as the Riemann (or Lebesgue) and stochastic integral components of Xt are both continuous in t.

3.1 Stochastic Integrals with Respect to a Jump-Diffusion

We can now define the stochastic integral of an adapted process with respect to Xt .

Definition 3 The stochastic integral of the process Φt with respect to a jump-diffusion, Xt , is

∫ t ∫ t ∫ t ∑
Φs dXs = Φs γs dWs + Φs θs ds + Φs ∆Js . (29)
0 0 0 0<s≤t

In differential notation we write (29) as

Φt dXt = Φt γt dWt + Φt θt dt + Φt ∆Jt .

Local Volatility, Stochastic Volatility and Jump-Diffusion Models 10

Example 3 (Shreve E.G. 11.4.4)

Let Xt := Nt − λt where Nt is a Poisson process with intensity λ. Now let Φs = ∆Ns so that Φs is 1 if there is
jump at time s and 0 otherwise. Using Definition 3 we find
∫ t ∫ t ∫ t
Φs dXs = Φs dNs − λ Φs ds
0 0 0
= Nt − 0 = Nt . (30)

It is perhaps surprising that the stochastic integral in Example 3 fails to be a martingale despite the fact that
the integrator, Xt , is a martingale. This occurs because the integrand, Φs , is not left-continuous. Indeed we
have the following17 theorem.

Theorem 3 If the jump-diffusion process, Xt , of (28) is a martingale and Φt is left-continuous and adapted,
∫t ∫t
and E[ 0 γs2 Φ2s ds] < ∞ for all t ≥ 0, then the stochastic integral 0 Φs dXs is a martingale.

When we work with jump processes we will often want to insist that Φs be left-continuous or (almost
equivalently), predictable. This is particularly true for financial applications where Φs can then be interpreted as
a trading strategy. However, it is worth emphasizing that (29) is still defined if we only assume Φs is adapted.

Example 4 (Shreve E.G. 11.4.6)

Let Xt be as in Example 3 and define Φs := I[0,S1 ] (s) where S1 is the time of the first jump. Note that Φs is
left-continuous and we now obtain
∫ t {
−λt, 0 ≤ t < S1
Φs dXs =
0 1 − λS 1 t ≥ S1

= I[S1 ,∞) (t) − λ(t ∧ S1 ). (31)

where x ∧ y := min(x, y). We can verify by direct computation that I[S1 ,∞) (t) − λ(t ∧ S1 ) is a martingale as
implied by Theorem 3. It can also be checked that if we had taken Φs = I[0,S1 ) (s) which is right-continuous but
not left-continuous, then we would have found that
∫ t
Φs dXs = −λ(t ∧ S1 ).
0
∫t
Since E0 [−λ(t ∧ S1 )] = e−λt − 1 it follows that in this case 0
Φs dXs is not a martingale.

3.2 Quadratic Variation and Cross Covariation

Consider a partition of the time interval [0, T ] given by Π := 0 = t0 < t1 < t2 < . . . < tn := T . Let Xt be a
jump-process as in (28) and consider the sum of squared changes

∑
n
( )2
QΠ (X) := Xti − Xti−1 (32)
i=1

so that the quadratic variation of X on [0, T ] is defined as

[X, X](T ) := lim QΠ (X).

||Π||→0

17 This is Theorem 11.4.5 in Shreve. He points out that the result holds more generally when the integrand is predictable.

Given a given probability space (Ω, F, P ), we say that a process Yt is predictable if Y : Ω × [0, ∞) → R is measurable with
respect to the σ-algebra generated by the set of all left-continuous adapted processes.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 11
∫t ∫T
If for example, Xt = 0 γs dWs then we know from our results on diffusions that [X, X](T ) = 0 γs2 ds which
of course is stochastic in general as γs itself is stochastic (and adapted) in general.
Now let X1 and X2 be two jump processes so that for i = 1, 2 we have
∫ t ∫ t
Xi (t) = Xi (0) + γi (s) dWs + θi (s) ds + Ji (t) (33)
0 0

where Ji (t) is right-continuous. We then define the cross-variation process

[X1 , X2 ](T ) := lim CΠ (X1 , X2 )

||Π||→0

where
∑
n
CΠ (X1 , X2 ) := (X1 (ti ) − X1 (ti−1 )) (X2 (ti ) − X2 (ti−1 )) .
i=1

The following result18 is straightforward to prove given the corresponding results for diffusion processes.

Theorem 4 Let X1 and X2 be two jump processes as in (33). Then

∫ T ∑
[X1 , X1 ](T ) = γ12 (s) ds + (∆J1 (s))2
0 0<s≤T

and
∫ T ∑
[X1 , X2 ](T ) = γ1 (s)γ2 (s) ds + ∆J1 (s) ∆J2 (s)
0 0<s≤T
= [X1c , X2c ](T ) + [J1 , J2 ](T ). (34)

Note that in differential notation, the identity (34) states that

dX1 (t) dX2 (t) = dX1c (t) dX2c (t) + dJ1 (t) dJ2 (t)

so that in particular
dX1c (t) dJ2 (t) = dX2c (t) dJ1 (t) = 0
where Xic (t) is the continuous part of Xi (t). More generally, we can see from (34) that we need both processes
to have a diffusion component or both processes to have simultaneous jumps in order for the cross-variation
process to be non-zero.

3.3 Examples of Popular Jump-Diffusion Models

Merton (1975) was the first to propose a jump-diffusion model for pricing options. While the assumptions of
Merton’s model are not very realistic, it is easy to analyze and worth investigating in some detail.

Merton’s Jump-Diffusion Model

Merton’s jump diffusion model assumes that the time t stock price, St , satisfies

∏
Nt
(µ−σ 2 /2)t+σWt
St = S0 e Yi (35)
i=1

where Nt is a Poisson process with mean arrival rate λ, and the Yi ’s are IID log-normal random variables with
µy := E[Yi ] for all i. The Poisson process and Brownian motions are independent processes and between jumps
18 This is Theorem 11.4.7 in Shreve.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 12

the stock price behaves like a regular GBM. If the dynamics in (35) are under an EMM, Q, then the model
parameters are constrained in such a way that the Q-expected rate of return must equal r − q where q is the
stock’s dividend yield. To be specific, note that
[N ]
∏t
Q µt Q
E0 [St ] = S0 e E0 Yi
i=1
[ [N ]]
∏t
= S0 eµt EQ
0 EQ
0 Yi | Nt
i=1
[ Nt ]
= S0 eµt EQ
0 µy
∑∞
(λt)i i
= S0 eµt e−λt µ
i=0
i! y
= S0 eµt+λt(µy −1) . (36)
If Q is an EMM (with the cash account as numeraire) then the expected growth rate under Q must be r − q
and so (36) implies that we must have
µ + λ(µy − 1) = r − q. (37)
This is an equation in three unknowns and so it has infinitely many solutions. We can therefore conclude from
the Second Fundamental Theorem of Asset Pricing that Merton’s model is incomplete. Indeed this is true of
almost all jump-diffusion models.
We would like to be able to price European options in Merton’s model and there are several ways to do this
including Monte-Carlo simulation and Laplace or Fourier transform methods. We can also price these options,
however, by expressing them as an infinitely weighted sum of Black-Scholes options prices. To see this, note
that conditional on NT = n we can write
( )
∑n
ST = S0 exp (µ − σ /2)T + σWT +
2
Zi
i=1
( √ )
= dist
S0 exp (µ − σ /2)T + nµz +
2
σ 2 + nσz2 /T WT
(( ) )
2nµz + nσz2
= S0 exp µ+ −σ
bn /2 T + σ
2
bn WT (38)
2T
√
bn :=
where Zi := log(Yi ) ∼ N (µz , σz2 ) are IID, “=dist ” denotes “equal in distribution” and σ σ 2 + nσz2 /T .
Conditional on NT = n, we therefore see that the risk-neutral drift of St in (38) is given by
2nµz + nσz2 2nµz + nσz2
µ+ = r − q − λ(µy − 1) + (39)
2T 2T
µz +σz2 /2
= r − q − λ(e − 1) + n(µz + σz2 /2)/T (40)
= r − qbn (41)
2
where qbn := q + λ(eµz +σz /2 − 1) − n(µz + σz2 /2)/T . Note that (39) follows from (37) and (40) follows
because µy = exp(µz + σz2 /2). We are now in a position to derive an expression for European call options in
Merton’s jump-diffusion model. We obtain
∞
∑ (λT )n Q −rT
EQ
0 [e
−rT
(ST − K) ]
+
= e−λT E0 [e (ST − K)+ |NT = n]
n=0
n!
∑∞
(λT )n
= e−λT Cbs (S0 , K, r, qbn , σ
bn , T ) (42)
n=0
n!
where Cbs is the usual Black-Scholes call option price and (42) follows from (41) and the log-normality of ST
conditional on NT = n.
An interesting question to consider is how well Merton’s jump-diffusion model can replicate the implied volatility
surfaces that are typically observed in the market. Note that in contrast to the geometric Brownian motion
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 13

(GBM), we have five parameters19 , σ, λ, µ, µy and σy2 and just one equation to satisfy, namely (37). We
therefore have much more flexibility than GBM which can only achieve constant implied volatility surfaces.
Figure 2 displays the implied volatility surface under a Merton jump-diffusion model when σ = 20%, r = 2%,
√
q = 1%, λ = 10%, µz = −.05 and σz = 0.1. While other shapes are also possible by varying these

Figure 2: An Implied Volatility Surface under Merton’s Jump Diffusion Model

parameters, Figure 2 demonstrates one of the principal weaknesses of Merton’s jump-diffusion model, namely
the rapid flattening of the volatility surface as time-to-maturity increases. For very short time-to-maturities,
however, the model has no difficulty with producing a steep volatility skew. This is in contrast to stochastic
volatility models which do not allow jumps.

Kou’s Model
Kou (2002) developed the double-exponential jump-diffusion model where the jump-sizes have a double
exponential distribution. In particular, the stock price process, St , has Q-dynamics that satisfy
(N )
dSt ∑ t

= (r − λξ) dt + σ dWt + d (Vi − 1)

St− i=1

where Wt and Nt are a Q-Brownian motion and Poisson process with intensity, λ, respectively. A simple
application of Itô’s Lemma (see Section 3.4) implies that the log-stock price process, Xt := log(St /S0 ), then
satisfies
( ) ∑Nt
Xt = r − σ 2 /2 − λξ t + σWt + Yi , X0 = 0
i=1

where Yi = log(Vi ). The Yi ’s are an IID sequence of double-exponential random variables with density
fY (y) = pη1 e−η1 y 1{y≥0} + qη2 eη2 y 1{y<0}
where p, q ≥ 0 with p + q = 1, η1 > 1 and η2 > 0. In order to ensure that Q is indeed an EMM, it must be the
case that
pη1 qη2
ξ := E [V ] − 1 = + − 1. (43)
η1 − 1 η2 + 1
The double-exponential jump-diffusion model is quite tractable due to the memoryless property of exponential
random variables. This enables us, for example, to compute analytical expressions for expectations involving first
passage times.
19 We could use σz and σz2 in place of µy and σy2 .
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 14

The Bates Model

The Bates model assumes the stock price, St , has risk-neutral dynamics given by
√ (s)
dSt /St = (r − q − λ(ᾱ − 1)) dt + σt dWt + (αt − 1) dNt (44)
√ (vol)
dσt = κ (θ − σt ) dt + γ σt dWt (45)

where Nt is a Poisson process with intensity λ and αt is log-normally distributed with EQ

t− [αt ] = ᾱ. It is clear
that the Bates model is simply the Heston model with independent jumps added to the security price dynamics.
This independence implies that we can also easily compute the characteristic function of the log-stock price in
this model given that we already know how to do so for the Heston model.

3.4 Itô’s Lemma for Jump-Diffusions

The stochastic calculus we have studied for diffusion processes can be extended to jump-diffusions and indeed
other stochastic processes, including Levy processes and more generally, semimartingales. In this section we will
state Itô’s Lemma for jump-diffusions.

Theorem 5 (Itô’s Lemma for 1-Dimensional Jump-Diffusions)

Let Xt be a jump process and f (x) a function for which f ′ (x) and f ′′ (x) are defined and continuous. Then
∫ t ∫ t ∑
1
f (Xt ) = f (X0 ) + f ′ (Xs ) dXsc + f ′′ (Xs ) dXsc dXsc + [f (Xs ) − f (Xs− )] (46)
0 2 0 0<s≤t

where Xtc denotes the continuous, i.e. non-jump, component of Xt and the summation in (46) is over the jump
times of the process.

The statement20 of Theorem 5 should not be surprising given our knowledge of Itô’s Lemma for diffusions. In
particular, the jump-diffusion behaves as a diffusion in between jumps so it is only necessary to add the finite
number of changes in f (·) due to jumps to the usual terms that come from the diffusive component of the
process.
Example 5 (Merton’s Jump Diffusion Model)
∑Nt
Let Xt := log(St ) where St is given by (35). Then Xt = log(S0 ) + (µ − σ 2 /2)t + σWt + i=1 log(Yi ) and so

dXt = (µ − σ 2 /2) dt + σ dWt + log(Yi ) dNt .

Applying Itô’s Lemma to St = eXt we recover the dynamics for St and obtain21
1
dSt = St (µ − σ 2 /2) dt + σSt dWt + St σ 2 dt + St− (Yt − 1) dNt
2
= St (µ + (µy − 1)λ) dt + σSt dWt + [St− (Yt − 1) dNt − St (µy − 1)λ dt] . (47)

Referring to (47) note that the dWt term and the term in the square brackets are both martingales. Therefore if
(47) describes the risk-neutral dynamics of St is must be the case that the drift term equals St (r − q) dt and so
we obtain (37) once again.

20 This is Theorem 11.5.1 in Shreve where a proof of the result may also be found.
21 We could of course have written the dynamics for St directly using (35) but then we wouldn’t have been able to practice
using Itô’s Lemma.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 15

Example 6 (Shreve E.G. 11.5.2: Geometric Poisson Process)

Suppose St satisfies
St := S0 exp (Nt log(σ + 1) − λσt) = S0 e−λσt (σ + 1) t
N

where σ > −1 and Nt is again a Poisson process. Note that if σ > 0 then this process only jumps up and drifts
down between jumps. If −1 < σ < 0 then the process only jumps down and drifts up between jumps. Let’s
apply Itô’s Lemma to show that St is a martingale.
Define Xt = Nt log(σ + 1) − λσt so that St = S0 f (Xt ) where f (x) = exp(x). Itô’s Lemma now implies

St = S0 f (Xt )
∫ t ∑
= S0 f (X0 ) − λσS0 f ′ (Xu ) du + S0 [f (Xu ) − f (Xu− )]
0 0<u≤t
∫ t ∑
= S0 − λσ Su du + [Su − Su− ]
0 0<u≤t
∫ t ∑
= S0 − λσ Su du + σSu− (48)
0 0<u≤t
∫ t ∫ t
= S0 − λσ Su− du + σSu− dNu (49)
0 0
∫ t
= S0 + σ Su− (dNu − λ du).
0

where we have used the fact that Su = (σ + 1)Su− if a jump takes place at time u and where Su− is the value
of S immediately before that jump. Note that because we were able to write the jump in S at time u in terms
of Su− , we can write the SDE in differential form

dSt = −λσSt dt + σSt− dNt = σSt− dXt

where Xt := Nt − λt is clearly a martingale. Note also that Theorem 3 implies (why?) that St is a martingale.

∫t ∫t
Remark 3 Note that in going from (48) to (49) we replaced 0 Su du with 0 Su− du and this presents no
problem as the two expressions are equal whenever there are only countably many jumps. This is indeed the case
for all jump-diffusion processes.

Example 7 (Independence of Brownian Motion and a Poisson Process)

Let Wt and Nt be a Brownian motion and Poisson process with intensity λ, respectively, where both processes
are defined on the same probability space (Ω, F, P ) and relative to the same filtration, Ft , t ≥ 0. Using Itô’s
Lemma we will show that Wt and Nt are independent.
Towards this end, let Yt := exp(Xt ) where
1
Xt := u1 Wt + u2 Nt − u21 t − λ(eu2 − 1)t
2
where u1 and u2 are fixed real numbers. By applying Itô’s Lemma to the function f (x) = ex we obtain after
some calculations ∫ t ∫ t
Yt = 1 + u1 Ys dWs + (eu2 − 1) Ys− dMs (50)
0 0

where Ms := Ns − λs. From (50) we can easily conclude (why?) that Yt is a martingale from which it follows
that E[Yt ] = 1 for all t. But this implies that the joint MGF of Wt and Nt can be written as
[ ]
E0 eu1 Wt +u2 Nt = e 2 u1 t eλ(e −1)t .
1 2 u2
(51)
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 16

We can therefore conclude (why?) that Wt and Nt are independent. Given this key step, how would you prove
the full result?

Example 7 is the main step in proving a more general result. This result22 states that a Brownian motion and a
Poisson process, both defined on a common probability space and relative to the same filtration, must be
independent. We now state Itô’s Lemma for multi-dimensional jump-diffusions.

Theorem 6 (Itô’s Lemma for Multi-Dimensional Jump-Diffusions)

Let Xt := (X1 (t), . . . , Xn (t)) where Xi (t) is a jump process for i = 1, . . . , n and let f (·) be a C 2 function of n
variables. Then
∑ n ∫ t n ∫
1 ∑ t ∑
f (Xt ) = f (X0 ) + fi (Xs ) dXic (s) + fij (Xs ) dXic (s)dXjc (s) + [f (Xs ) − f (Xs− )] (52)
i=1 0
2 i,j=1 0
0<s≤t

where fi and fij denote the first and second partial derivatives, respectively, with respect to the appropriate
arguments of f .

The following special case of Theorem 6 arises frequently enough as to warrant special attention.
Theorem 7 (Itô’s Product Rule for Jump-Diffusions) Let X1 (t) and X2 (t) be jump processes. Then
∫ t ∫ t
X1 (t)X2 (t) = X1 (0)X2 (0) + X2 (s) dX1c (s) + X1 (s) dX2c (s) + [X1c , X2c ](t)
0 0
∑
+ [X1 (s)X2 (s) − X1 (s−)X2 (s−)] (53)
0<s≤t
∫ t ∫ t
= X1 (0)X2 (0) + X2 (s−) dX1 (s) + X1 (s−) dX2 (s) + [X1 , X2 ](t) (54)
0 0

While (53) follows almost immediately from an application of Theorem 6, proving (54) requires some additional
work. In particular, we can use (34) and the fact that jumps in Xi (t) are the same as jumps in
Ji (t) := Xi (t) − Xic (t) to go directly from (54) to (53).

Definition 4 (Doleans-Dade Exponential) Let Xt be a jump-diffusion. Then the Doleans-Dade

exponential of X is defined to be
( ) ∏
1 c
Zt := exp Xt − [X , X ](t)
X c c
(1 + ∆Xs ). (55)
2
0<s≤t

Theorem 8 The Doleans-Dade exponential, ZtX , is the solution to the stochastic differential equation
∫ t
X X
Zt = 1 + Zs− dXs . (56)
0
( )
Theorem 8∏can be proven by applying Itô’s product rule to ZtX = Yt Vt where Yt := exp Xtc − 12 [X c , X c ](t)
and Vt := 0<s≤t (1 + ∆Xs ). The Doleans-Dade exponential plays an important role in stochastic calculus,
particularly in the context of Girsanov’s Theorem. This follows because we can see from (55) and (56) that ZtX
will be23 a positive martingale with E[ZtX ] = 1 if Xt is a martingale. In that case it can be used to define a
change of probability measure.

22 The steps of Example 7 are presented in Corollary 11.5.3 of Shreve which states the result concerning the independence of
a Brownian motion and Poisson process. See also Exercises 11.4 to 11.6 in Shreve for related results on (i) the independence of
multiple Poisson processes that do not have simultaneous jumps and (ii) the independence of a compound Poisson process and
a Brownian motion.
23 Subject to technical restrictions on X . Otherwise Z X will be a local martingale which we will not define here.
t t
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 17

4 Fourier Transform Methods

Fourier and Laplace transform methods have proved particularly useful for pricing derivative securities when
closed-form solutions are not available. As mentioned earlier, they were first used in a financial engineering
context by Heston (1993). Transform methods can be used to solve the pricing PDE or to compute the
risk-neutral expected value directly. In section 4.1 below we briefly describe the Carr-Madan approach24 which
uses Fourier transforms to compute option prices directly.
A popular alternative approach is the Lewis method25 which computes the generalized Fourier transform of the
derivative price with respect to the log-stock price. The Lewis approach is more general than the Carr-Madan
approach below as it does not require the concept of a “strike” but it is also somewhat trickier to implement.

4.1 The Carr-Madan Approach

The Carr-Madan approach requires the characteristic function or the Fourier transform of the log-stock price.
This transform can be computed for a wide array of models including, for example, the Heston model. In fact
the characteristic function of the log-stock price in the Heston model is given in (23). The Carr-Madan
approach can be used to price European call options as we will now show.
First note that we can write the call option price as
∫ ∞
CT (k) = e−rT (es − ek )qT (s) ds (57)
k

where qT is the risk-neutral density of the log-stock price, s := log(ST ), T is the time-to-maturity and
k = log(K) is the log-strike. Because CT (k) → S0 as k → −∞, CT (k) will not be square-integrable. We
overcome this problem by defining
cT (k) := exp(αk)CT (k) (58)
for some α > 0. (Values of α = .75 have been recommended in the literature but depending on the application
at hand, a different value may be required. Note that (62) is valid for any positive α so any difficulties that
might arise with α are due to the difficulties that arise with the numerical inversion of the right-hand-side of
(62).) Consider now the Fourier transform of cT (k) which is defined as
∫ ∞
ψT (v) := eivk cT (k) dk. (59)
−∞

In Exercise 17 you are asked to show that

[ ]
e−rT E0 ei(v−(α+1)i) log(ST )
ψT (v) =
α2 + α − v 2 + i(2α + 1)v
e−rT ϕT (v − (α + 1)i)
= (60)
α2 + α − v 2 + i(2α + 1)v
where i = 0 + i is imaginary and ϕT (·) is the characteristic function of the log-stock price, log(ST ). The Fourier
inversion formula then implies that the option price is given by
∫ ∞
e−αk
CT (k) = e−ivk ψT (v) dv (61)
2π −∞
∫ ∞
e−αk ( )
= Re e−ivk ψT (v) dv (62)
π 0
24 This approach was developed in “Option Valuation Using the Fast Fourier Transform” by Carr and Madan in the Journal

of Computational Finance (1998). Other popular transform methods include the Lewis method
25 “A Simple Option Formula for General Jump-Diffusion and Other Exponential Levy Processes” (2001), available from

http:www.optioncity.net.
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 18

where Exercise 17 again asks you to justify going from (61) to (62). The option price in (62) can be found using
standard Fourier inversion techniques. If many options with different strikes but the same time-to-maturity need
to be priced, then the Fast Fourier transform can be used. Indeed this was the approach proposed originally by
Carr and Madan and for this reason we review the FFT in Appendix A. Alternatively, if just a single option price
is required a standard26 numerical integration of the right-hand-side of (62) can be performed. In fact, even if
we are pricing multiple options with the same time-to-maturity it is not clear27 that using the FFT is the most
efficient method for doing so.
Finally, we mention here the implied volatility surface in Figure 1 was constructed by using (62) to compute call
option prices in the Heston model.

Remark 4 As pointed out by Carr and Madan, for very short maturities the option price approaches its
intrinsic value which is non-analytic. This causes the integrand in (62) to be very oscillatory and therefore very
difficult to integrate numerically. They circumvent this problem by developing an expression for zT (k) where
zT (k) is the time value of the option. This expression is again in terms of the characteristic function of the
log-stock terminal price and is obtained in a similar manner to our derivation of (62).

Appendix A: The Fast Fourier Transform

Let X be a random variable with PDF, f (·), and let
∫ ∞
[ ]
fbx (s) := E e−isx = e−isx f (x) dx
−∞

be the Fourier transform of X, i.e. the Fourier transform of f (·). Then the standard Fourier inversion formula
states that ∫ ∞
1
f (x) = fbx (s)eisx ds. (63)
2π −∞
Now suppose we know fbx (·) and we wish to compute f (·) numerically using the inverse FFT. Then using (63)
we can approximate f (·) as

1 ∑b
N
f (x) ≈ fx (sj )eisj x η (64)
2π j=1

where η = “ds” and

sj := −T + η(j − 1) for j = 1, . . . , N (65)
so that the lower and upper limits of integrations are −T and −T + η(N − 1), respectively. Let ηλ = 2π/N .
The FFT returns then f (·) at the values
xn := −b + λ(n − 1) for n = 1, . . . , N. (66)
In particular substituting (66) and then (65) into (64) we obtain

1 ∑b
N
f (xn ) ≈ fx (sj )eisj (−b+λ(n−1)) η
2π j=1

η ∑b
N
= fx (sj )e−ibsj eiηλ(n−1)(j−1)−iλT (n−1)
2π j=1

ηe−iλT (n−1) ∑ b
N
fx (sj )e−ibsj e N (n−1)(j−1)
2πi
= (67)
2π j=1

26 Note that some software packages / languages such as Matlab can handle complex calculations whereas others such as VBA

require additional functions.

27 There is an ongoing debate as to what are the most efficient methods for evaluating (62).
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 19

since ηλ = 2π/N . Rearranging (67) we obtain

1 ∑b
N
2π
fx (sj )e−ibsj e N (n−1)(j−1)
2πi
f (xn )eiλT (n−1) =
ηN N j=1

or equivalently,
1 ∑
N
2πi
vn = uj e N (n−1)(j−1) (68)
N j=1

where
2π
vn := f (xn )eiλT (n−1) and (69)
ηN
uj := fbx (sj )e−ibsj .

(68) is now in the form required by most software packages. That is we first compute the uj ’s and then pass
them through the inverse FFT to obtain the vn ’s. Finally, we compute the f (xn )’s using (69). Note that for a
fixed N , there is a tradeoff between the accuracy of the numerical integration and the fineness of the grid where
we compute f (x). This is because we must have ηλ = 2π/N .

Exercises
1. This question refers to the exercise from the Black-Scholes and the Volatility Surface where an implied
volatility surface was fitted to European call and put prices and then (in a later exercise) used to compute
the price of a digital option. We will fit an implied volatility surface to the option data and then use this
surface to compute a local volatility surface using Dupire’s formula. (Note that there are better ways to
do the various steps below but if done properly they should produce a reasonable local volatility surface.
Feel free to try your own way if you prefer.)
(a) Use a spline function to fit the implied volatility skew for each option maturity. (As there are four
maturities you should have four fitted splines. Note also that in practice you should know (and be satisfied
with) how the splines extrapolate beyond the observed implied volatilities.)

(b) Now write a function that takes each of the fitted splines from part (a) and uses them to compute an
implied volatility for any strike-maturity pair, (K, T ). A good way to do this is to interpolate using total
variance. That is, suppose we want σbs (K, T ) where K = 75 and T = .8. We then use the fitted splines
at T = .5 and T = 1 to compute σbs (75, .5) and σbs (75, 1). We could then estimate σbs (75, .8) as
√
5 × .5 σbs (75, .5) + 5 × σbs (75, 1)
2 2 3 2
σbs (75, .8) = . (70)
.8

Note that interpolating total variance as we do in (70) is much better than interpolating total volatility.
Note also that you need to make some assumption regarding how to extrapolate to maturities less than
.25 years and greater than 1.5 years. For the purpose of this question you can assume that the term
structure of implied volatility is constant up until t = .25 and constant beyond t = 1.5.

(c) Now plot your fitted implied volatility surface using a grid of strike-maturity pairs that is larger than
the original grid.

(d) Use the Dupire formula to write a function that computes the local volatility at a given strike-maturity
pair, (K, T ). You should use your function from part (b) to estimate the various partial derivatives
numerically. In anticipation of simulating many paths of the local volatility process simultaneously, it
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 20

would actually be a good idea if your function was vectorized along the strike dimension so that one call
to the function could return σl (K1 , T ), . . . , σl (Kn , T ) where n is an arbitrary number of strikes.

(e) Now plot your local volatility surface. What do you notice?

(f ) Simulate your local volatility model (using an Euler scheme) to price the original options that you used
to construct the volatility surface. Note that you can use the same Monte-Carlo to price all of the options.
(If necessary the Brownian bridge construction together with stratified sampling can be used to speed up
your Monte-Carlo.) How do your Monte-Carlo prices compare to the original prices? (Hint: they should be
almost identical with any differences being due to statistical error (from the Monte-Carlo) and numerical
error (from the Euler scheme and estimation of the derivatives).)

2. Consider the CEV model with Q-dynamics given by (11). Find the Q̃-dynamics of the deflated cash
account when we deflate by the risky asset and Q̃ is the corresponding EMM. (You may assume that
β > 1/2 so that there is no possibility of the risky asset price reaching zero.)
3. Justify the step where we went from (24) to (25) in Example 1. (Hint: Take expectations in (13) and then
use the martingale property of stochastic integrals to eliminate the last term. You can then obtain a
simple ODE for EQ 0 [σt ].)

4. Confirm that (18) is indeed the pricing PDE that corresponds to the Heston model of (12) and (13).
5. (Volatility and Variance Swaps in Heston’s Model)
A standard result states that the square-root function can be expressed as
∫ ∞
√ 1 1 − e−sx
x = √ ds. (71)
2 π 0 s3/2

(a) Use (71) to obtain an expression for the fair value of a √ volatility swap under the Heston model. You
∗ ∗
may assume that the payoff of a volatility swap is given by Vc (0, T ) − Kvol where Kvol is the strike
(typically chosen at inception so that the initial value of the swap is 0) and
∫
1 T
Vc (0, T ) = σt dt
T 0
is the continuous realized variance with σt having the dynamics of (13).

Hint: The Laplace transform of Vc (0, T ) can be computed analytically. In particular,

[ ]
E0 e−sVc (0,T ) | σ0 = eA(T,s)−B(T,s)σ0 (72)

where A(T, s) and B(T, s) are known deterministic functions of T . (See Cairns (2000) for details.)

(b) Use (72) to compute the fair value of a variance swap in Heston’s model. Your answer should of
course agree with the expression we found in (26).

6. Consider a two-dimensional Markov process for the foreign price of a stock, St , and the exchange rate, Xt ,
where both processes follow local volatility models:

dSt = (rf − q)St dt + σs (St , t)St dWs (t) (73)

dXt = (r − rf )Xt dt + σs (Xt , t)Xt dWx (t) (74)

where dWs dWx = ρ(St , Xt , t) dt, rf is the foreign risk-free rate, r is the domestic risk-free rate and q is
the dividend yield. In addition, Ws and Wx are Q̃- and Q-Brownian motions, respectively, where Q̃ is the
EMM for a foreign investor with the foreign cash account as numeraire, and Q is the EMM of a domestic
investor with the domestic cash account as numeraire. Find an expression for the Q-dynamics in a local
volatility model of St .
Local Volatility, Stochastic Volatility and Jump-Diffusion Models 21

7. Show that the MGF of a compound Poisson process, Xt , is given by ϕXt (u) = eλt(ϕY (u)−1) as in equation
(27) of Example 2.
8. Write a program to construct the implied volatility surface of Figure 2. To do this you need to write a
program to compute option prices in Merton’s model. This can be done, for example, by either evaluating
the expression in (42) or by using Monte-Carlo or by computing the characteristic function of the
log-stock price and then using Fourier transform techniques.

9. Referring to Example 4, show that the process Mt := I[S1 ,∞) (t) − λ(t ∧ S1 ) in (31) is indeed a martingale.
10. Use Itô’s Lemma to confirm that (43) must hold if the dynamics specified for St in Kou’s jump-diffusion
model are risk-neutral or Q-dynamics.
11. (Exercise 11.3 in Shreve) Let Nt be a Poisson process with intensity λ > 0, and let S0 and σ > −1
be given. Use the stationary and independent increments property of a Poisson process rather than Itô’s
Lemma to show directly that
St = exp (Nt log(σ + 1) − λσt) = (σ + 1)Nt e−λσt
is a martingale.

12. (Exercise 11.4 in Shreve) Suppose N1 (t) and N2 (t) are independent Poisson processes with
intensities λ1 and λ2 , respectively, both defined on the same probability space (Ω, F, P ) and relative to
the same filtration Ft , t ≥ 0. Show that almost surely N1 (t) and N2 (t) can have no simultaneous jump.
(Hint: Define the compensated Poisson processes M1 (t) = N1 (t) − λ1 t and M2 (t) = N2 (t) − λ2 t, which
like N1 and N2 , are independent. Use Itô’s product rule for jump processes to compute M1 (t)M2 (t) and
take expectations.)
13. Referring to Example 7, use Itô’s Lemma to derive (50).
14. Prove Itô’s Product Rule for jump-diffusions. That is, prove Theorem 7.
15. Use Itô’s product rule to prove Theorem 8.
16. (A Standard Fourier Transform Approach to Option Pricing)
−rT +
Let C0 = EQ 0 [e (ST − K) be the time t = 0 price of a call option with strike K, maturity T ,
risk-free-rate r and underlying security price process St . Note that Q is the EMM corresponding to the
cash account as numeraire.
(a) Show that the option price may be expressed as
C0 = S0 Q1 (XT > ln K) − e−rT KQ2 (XT > ln K) (75)
where XT := ln(ST ) and Q1 and Q2 are EMM’s corresponding to specific changes of numeraire. Be sure
to identify these numeraires. Does the expression in (75) agree with the interpretation of P1 and P2 that
we gave at the end of the paragraph immediately following (20)?
(b) Let f1 and f2 be the characteristic functions under Q1 and Q2 , respectively, of XT so that
Q [ ]
fj (u) := E0 j eiuXT , j = 1, 2.
Give an expression for C0 in terms of f1 and f2 . Be sure to simplify it as much as possible.

17. In this exercise we justify the Carr-Madan call option-pricing formula of (62).

(a) By substituting for cT (k) in (59) using (57) and (58), show that ψT (v) satisfies (60).
(b) Justify going from (61) to (62).

18. Using the Carr-Madan approach, write a program to recreate the implied volatility surface in Figure 1
corresponding to the Heston model with parameters as in Section 2.4. (Note that characteristic function
of the log-stock price is given in (23).)