Professional Documents
Culture Documents
Abstract
In this paper, we investigate recent developments in option pricing based on Black-Scholes processes, pure
jump processes, jump diffusion process, and stochastic volatility processes. Results on Black-Scholes model
with GARCH volatility (Gong, Thavaneswaran and Singh [1]) and Black-Scholes model with stochastic
volatility (Gong, Thavaneswaran and Singh [2]) are studied. Also, recent results on option pricing for jump
diffusion processes, partial differential equation (PDE) method together with FFT (fast Fourier transform)
approximations of Pillay and O’ Hara [3] and a recently proposed method based on moments of truncated
lognormal distribution (Thavaneswaran and Singh [4]) are also discussed in some detail.
Keywords: Stochastic Volatility, Black-Scholes Partial Differential Equations, Option Pricing, Monte Carlo
2, we study option pricing for pure jump processes, jump Then from our assumption that V t h t , X t for
diffusion models, stochastic volatility models, and jump any twice differentiable function f t , x and the Itô for-
diffusion models with stochastic volatility. Moreover, closed mula, we have (2.4).
form solutions are obtained by solving the two-dimensional When we equate the drift and diffusion coefficients
partial differential equations for stock price in some ex- from Equation (2.3) and Equation (2.4), we find a simple
amples. Section 3 closes the paper with conclusions. expression for the size of the stock portion of our repli-
cating portfolio:
2. Option Pricing and Partial Differential a t f x t, X t ,
Equations
and find
Consider the stock and bond model as in Steele [27] t, X t f x t, X t b t r t, X t t
dX t t , X t dt (t , X t )dW t ,
ft t , X t f xx t , X t 2 t , X t
1
and
d t r t , X t t dt ,
2
where all of the model coefficients f x t, X t t, X t .
t , X t , t , X t , and r t , X t are given by
The t , X t f x t , X t terms cancel, and b t
explicit functions of the current time and current stock is given by
price. First, we will use the coefficient matching method
to show that arbitrage price at time t of a European op- b t
tion with terminal time T and payout h t , X t is
ft t , X t f xx t , X t t , X t .
1 1 2
given by f t , X t where f t , x is the solution of r t, X t t 2
Because V t is equal to both f t , X t and
the partial differential equation (PDE)
1
ft t , x r t , x xf x t , x 2 t , x f xx t , x a t X t b t t , the values for a t and b t
give us a PDE for f t , X t :
2 (2.1)
r t, x f t, x
f t, X t f x t, X t X t
1
r t, X t
with terminal condition
f T , x h x .
If we let a t denote the number of units of stock
ft t , X t f xx t , X t 2 t , X t .
1
that we hold in the replicating portfolio at time t and let 2
b t denote the corresponding number of units of the
Now, when we replace X t by x, we arrive at the
bond, then the total value of the portfolio at time t is
general Black-Scholes PDE (2.1) and its terminal bound-
V t a t X t b(t ) t . ary condition (2.2).
The condition where the portfolio replicates the con- We can solve the Black-Scholes PDE to get the time-t
tingent claim at time T is simply call price. The link between the stochastic solution to
V T h X T . PDE and the martingale approcach is given by the fol-
lowing Feynman-Kac Theorem.
From the self-financing condition Theorem 2.1 A function f t , x defined by
d V t a t d X t b t d t (2.2)
f t , x e h X T X t x ,
t V X d
T
dV t f t t , X t d t f xx t , X t 2 t , X t dt f x t , X t dX t
1
2
(2.4)
ft t , X t f xx t , X t 2 t , X t f x t , X t t , X t dt f x t , X t t , X t dW t .
1
2
1 In Case (d),
ft t , x t , x f x t , x 2 t , x f xx t , x
2 f t, x
V t, x f t, x
r T t
Q e log X t e
2
2 T t W T t
with the terminal condition f T , x h x . X t x
Consider the Black-Scholes PDE
1
ft t , x rxf x t , x 2 x 2 f xx t , x rf t , x ,
e
r T t
log x 2 T t .
2
2
with the terminal condition: (a) f T , x x K ; (b)
2.1. Option Pricing for Pure Jump Processes
f T , x h x for some function h : 0, ; (c) (PDE Approach)
f T , x x ; (d) f T , x log x .
The above the Black-Scholes equation has a stochastic In order to price European options based on jump processes
solution of the form (Vecer [28]), we need to know the evolution of both
X Y t and YX t in order to determine both martin-
f X T X t x ,
r T t
f t , x e gale measures X and Y . It is possible to preserve the
where, under the neutral risk measure , symmetry of the evolution of the prices with the excep-
tion that the jump preserves the direction: when X Y t
dX t rX t dt X t dW t ,
jumps up, YX t jumps down, and vise versa. The jump
and X t is a geometric Brownian motion of the form N t belongs to the pair of and Y; it cannot be indivi-
X T X t exp r 2 2 T t W T t . dualized to one asset in contrast to the geometric Brow-
nian motion model, where the noise factor W X is asso-
For specific terminal conditions, the closed form time- ciated with the asset X, and the noise factor W Y is asso-
t price can be obtained by finding the stochastic solutions. ciated with the asset Y. In the case of Poisson evolution,
In Case (a), the time- t price is given by the Black-Scho- it is the intensity of the Poisson process that is asso-
les formula ciated with the particular asset. Under the X measure,
f t, x N t has intensity X t and the process N t X t is
a martingale, while under the Y measure, N t has in-
Q e X T K X t x
r T t
where
log x K r 2 2 T t X Y t X Y 0 exp N t e 1 Y t . (2.5)
d1 ,d 2 d1 T t .
T t The inverse price process YX t satisfy the SDE of
In Case (b), the time-t price can be represented as the form
f t, x
dYX t e 1 YX t d N t e X t ,
r T t
Q e f
2
2 T t W T t
X t e X t
and has a geometric Poisson process representation
x
YX t YX 0 exp N t e 1 X t . (2.6)
r T t
r 2 2 T t T t Z
e f xe . The values of Y and X are linked by the rela-
tionship X e Y , where is the size of the jump of
In Case (c), for some real v, log X Y t . Let V be a contract that pays off f Y X Y T
f t, x units of an asset Y at time T. The price of the contract V
with respect to the reference Y is given by
Q e X t e X t x
r T t ( 2 2T t W T t
VY t Yt f Y X Y T .
2
e x e
r T t r 2 T t W T t
X utX t , e x u X t , x e 1 xu xX t , x 0 where t is a deterministic function that represents the
equilibrium mean level of the asset against time, k is the
with the terminal condition mean reverting intensity of the asset, a t is a determi-
u X T , x f X
x . nistic function that describes the equilibrium mean level
of the volatility process against time and b is the mean
(b) The price function based on the geometric Poisson pro- reversion speed of the volatility process. The constant
cess (2.6), uY t, x Yt f Y X Y T XY t x satisfies is the volatility coefficient of the volatility process, and
W1 t and W2 t are correlated with correlation coef-
utY t , x Y utY t , e x uY t , x e 1 xuYx t , x 0 ficient .
with the terminal condition On the probability space , , , a Poisson process
N t is further defined for all 0 t T , with a constant
uY T , x f Y x . intensity parameter 0 . The process N t is assumed
Proof of this theorem is given in [28] (pp. 249-250). to be independent of both W1 t and W2 t . Furthermore,
For a geometric Poisson model (Vecer [28]) for two a sequence of random variables e Ji for 1 i N t is
no-arbitrage assets X and Y, where the price follows defined to represent the jump sizes of the Poisson proc-
dX Y t e 1 X Y t d N t Y t , ess. Each of the e Ji ’s are log-normally, identically and
independently distributed over time, where Ji ~ N , 2
and by letting N t m k , the price of a contract that and 0 . Then by defining the following process
pays off VT I N T k YT in terms of a reference X t ln S t and applying the Itô-Doebin formula to
asset Y is given by the two-factor mean reverting process with stochastic
volatility and jumps we have
VY t Yt VY T Yt N T k
v t m
tY N T k tY N T N t k m dX t k X t dt v t dW1 t
2 k k
T t
T t k m
e e J 1 X t dN t ,
Y
Y
.
k m ! (2.7)
Moreover, the price of VX t using the reference asset X dv t b(a v t )dt v t dW2 t , (2.8)
is given by
dW1 t dW2 t dt. (2.9)
T t
Y T t k m
Y
e
VX t VY t YX t YX t . The conditional characteristic function of the X t
k m ! process (2.7) is defined as
m
B t, T
k
iu e k T t 1 ab D s, T ds u T t , C t , T iu e k T t 1 ,
t
T
2ku a , b , ba a 1, b 1,
*
k T t
1 2 i * * * *
2 2
f v D f ; t , x, v B Ri aD .
One can solve these two ODEs under the conditions
f xx C i f ; t , x, v
2
B 0 0 and D 0 0 to obtain
(2.17)
f vv D 2 f ; t , x, v a 1 ged
B Ri i c d 2 ln ,
2 1 g
f xv C i D f ; t , x, v
i c d ed 1
f B C x D v f ; t , x, v . D d
,
2 1 ge
Substituting (2.17) into (2.15), it yields
where
( B C x D v) ( x)(C i ) i c d
g ,d i c 2 2ui 2 .
2
i c d
v D x v (C i ) D (2.18)
The FFT method can be used to obtain the call price
2 x C i 2 v D 2 0.
1 2 1 from the conditional characteristic function.
2 2
Equation (2.18) leads to a system of ODEs. We can 2.4. Option Pricing for Jump Diffusion Model
get the characteristic function by solving this ODE system. with Stochastic Volatility
As an example, for the Stochastic Volatility model
studied by Christoffersen, Heston and Jacobs [19] de- Thavaneswaran and Singh [4] considered the price proc-
fined by ess S t following a conditional jump diffusion model
dv t (a cv t )dt v t dW2 t , S t S t
y t log log t Z t ,
S t 1 S t 1
X t ln S t ,
where v r m t 2 , m eY 1 , and W t
2
where dW1 t dW2 t dt , and we know the equations is a standard Brownian motion, t is a stationary
that x R uv t , x v t , v a cv t process, and the number of jumps N t on 0,t fol-
lows a Poisson process with rate . Equivalently the
and v v t . Substitute them into (2.16), we have model can be written as
S t N t
1 m T
n
vt t W t Yi ,
log
S e 1 m T Ke rnT t g t ,
i 1 n!
n 1
where
S t S t S 2
y t log (t ) Z t ,
1
S t 1
log
S t 1 log K rnT 2 t
f t ,
t
where Yi ’s are identically distributed independent nor-
mal random variables having mean and variance 2 .
Let t t n 2 T . Then conditioning on t S 1
log K rnT 2 t
2 2 2
and taking the expectation as in Theorem 1 in [4], the g t t ,
call price is given by v
C S ,T
S T K
max S T K ,0.
1 m T
n
1 m T S is the initial value of S t , r is the risk-free interest
e e rnT t ( S T K )
n 1 n! rate, K is the strike price, and T is the expiry date. For
given N t n , we have rn r m n log 1 m T
and the t process has mean , variance 2 ,
1 m T
n
skewness , and kurtosis .Then the following
S t f t
1 m T
e
n 1 n! theorem is given in Thavaneswaran and Singh [4].
Theorem 2.3 For any twice continuously differentiable function f x and g x , the call price is given by
C S,T
1 m T 1 m T
n n
e
1 m T
n!
S t f t e
1 m T
n!
Ke
rn T
t g t
n 1 n 1
1 m T
n
t 2 t 4 2 t 1
1
e S f 2 t
1 m T
2
f 2 (t ) 2
n 1 n! 2
1 m T
n
E t 2 t
1
e g 2 t g 2 t 4 2 t 1)
1 m T
(
2
rn T 2
Ke
n 1 n! 2
1 m T
n
1 y
1
e S f 2 t f 2 t 1 2 2 t
1 m T
n 1 n! 2 3
1 m T
n
1 y 2 2
g t g t 1 t
1
e
1 m T rn T 2 2
Ke
n 1 n! 2 3
where S 1
log K rnT 2 t
2
t t
4
f 2 t ,
2 (t )
2 t [ t ]
2
S
is the kurtosis of volatility process t ,
1
log K rnT 2 t
2
y 4 t
g 2 t
,
y 2 t
2 y 2 t
is the kurtosis of the observed log-return y t , and
S 2 2 S
2
t 2 log rn T t 4 log rn T
2
f 2 t
1 K K
8 t
4 t t
2 2 2 2 2
S S
2
6 log r n
T
2
t
2 log rn
T
2
t
K exp K
,
8 2 2 t 2 t 8 t
2
S 2 2 S
2
t 2 log rn T (t ) 4 log rn T
2
g 2 t
1 K K
t
4 t t
2 2 2 2 2
S S
2
6 log r T
2
t
2 log rn T
2
t
exp
n
K K
.
8 2 2 t 2 t
8
2
t
Proof of the theorem follows from Gong, Thavanes- Volatility Models with Application in Option Pricing,”
waran and Singh [1] and Thavaneswaran and Singh [4]. Journal of Statistical Theory and Practice, Vol. 4, No. 4,
2010.
Superiority of the approach has been demonstrated for
Black-Scholes model with GARCH volatility and Black- [3] E. Pillay and J. G. O’ Hara, “FFT Based Option Pricing
under a Mean Reverting Process with Stochastic Volatil-
Scholes model with stochastic volatility in Gong, Tha-
ity and Jumps,” Journal of Computational and Applied
vaneswaran and Singh [1,2]. More details of related re- Mathematics, Vol. 235, No. 12, 2011, pp. 3378-3384.
cent empirical results are also given in Cao and Guo doi:10.1016/j.cam.2010.10.024
[31,32]. [4] A. Thavaneswaran and J. Singh, “Option Pricing for Jump
Diffusion Model with Random Volatility,” The Journal
3. Conclusions of Risk Finance, Vol. 11, No. 5, 2010, pp. 496-507.
doi:10.1108/15265941011092077
Recently Gong, Thavaneswaran and Singh [1,2] have de- [5] F. Black and M. Scholes, “The Valuation of Option Con-
monstrated the superiority of the truncated lognormal dis- tracts and a Test of Market Efficiency,” Journal of Fi-
tribution method for option pricing by carrying out ex- nance, Vol. 27, No. 2, 1972, pp. 99-417.
doi:10.2307/2978484
tensive empirical analysis of the European call option
valuation for S & P 100 index and showing that the pro- [6] F. Black and M. Scholes, “The Pricing of Options and
Corporate Liabilities,” Journal of Political Economy, Vol.
posed method outperforms other compelling stochastic
81, No. 3, 1973, pp. 637-659. doi:10.1086/260062
volatility pricing models. In this paper, option pricing is
[7] R. C. Merton, “Theory of Rational Option Pricing,” Jour-
discussed for Black-Scholes models, stochastic volatility
nal of Economics and Management Science, Vol. 4, No. 1,
models, pure jump process models and jump diffusion 1973, pp. 141-183.
process models. Option pricing using PDE method to-
[8] S. L. Heston, “A Closed-Form Solution for Options with
gether with FFT and the method based on a truncated Stochastic Volatility with Applications to Bond and Cur-
lognormal distribution for Black-Scholes process and jump rency Options,” The Review of Financial Studies, Vol. 6,
diffusion process are also discussed in some detail. No. 2, 1992, pp. 327-343. doi:10.1093/rfs/6.2.327
[9] S. L. Heston, “Option Pricing with Infinitely Divisible
4. References Distributions,” Quantitative Finance, Vol. 4, No. 1, 2004,
pp. 515-524. doi:10.1080/14697680400000035
[1] H. Gong, A. Thavaneswaran and J. Singh, “A Black- [10] J. C. Cox, J. E. Ingersoll and S. A. Ross, “A Theory of
Scholes Model with GARCH Volatility,” The Mathe- the Term Structure of Interest Rates,” Econometrica, Vol.
matical Scientist, Vol. 35, No. 1, 2010, pp. 37-42. 53, No. 2, 1985, pp. 385-407. doi:10.2307/1911242
[2] H. Gong, A. Thavaneswaran and J. Singh, “Stochastic [11] G. Bakshi, C. Cao and Z. Chen, “Empirical Performance
of Altemative Option Pricing Models,” Journal of Finance, [22] G. Barone-Adesi, H. Rasmussen and C. Ravanelli, “An
Vol. 52, No. 5, 1997, pp. 2003-2049. Option Pricing Formula for the GARCH Diffusion Model,”
doi:10.2307/2329472 Computational Statistics and Data Analysis, Vol. 49, No.
[12] L. O. Scott, “Option Pricing When the Variance Changes 2, 2005, pp. 287-310. doi:10.1016/j.csda.2004.05.014
Randomly: Theory, Estimation, and an Application,” Jour- [23] G. Barone-Adesi, R. F. Engle and L. Mancini, “A
nal of Financial and Quantitative Analysis, Vol. 22, No. GARCH Option Pricing Model with Filtered Historical
4, 1987, pp. 419-438. doi:10.2307/2330793 Simulation,” Review of Financial Studies, Vol. 21, No. 3,
[13] J. Hull and A. White, “The Pricing of Options on Assets 2008, pp. 1223-1258. doi:10.1093/rfs/hhn031
with Stochastic Volatilities,” Journal of Finance, Vol. 42, [24] A. Thavaneswaran, S. Peiris and J. Singh, “Derivation of
No. 2, 1987, pp. 281-300. doi:10.2307/2328253 Kurtosis and Option Pricing Formulas for Popular Vola-
[14] J. Hull and A. White, “An Analysis of the Bias in Option tility Models with Applications in Finance,” Communica-
Pricing Caused by a Stochastic Volatility,” Journal of In- tions in Statistics.Theory and Methods, Vol. 37, 2008, pp.
ternational Economics, Vol. 24, No. 4, 1988, pp. 129- 1799-1814. doi:10.1080/03610920701826435
145. [25] A. Thavaneswaran, J. Singh and S. S. Appadoo, “Option
[15] P. Ritchken and R. Trevor, “Pricing Options under Gen- Pricing for Some Stochastic Volatility Models,” Journal
eralized GARCH and Stochastic Volatility Processes,” of Risk Finance, Vol. 7, No. 4, 2006, pp. 425-445.
Journal of Finance, Vol. 59, No. 1, 1999, pp. 377-402. doi:10.1108/15265940610688982
doi:10.1111/0022-1082.00109 [26] S. J. Taylor, “Asset Price Dynamics, Volatility, and Pre-
[16] J. B. Wiggins, “Option Values under Stochastic Volatil- diction,” Princeton University Press, Princeton, 2005.
ities,” Journal of Financial Economics, Vol. 19, 1987, pp. [27] J. M. Steele, “Stochastic Calculus and Financial Applica-
351-372. doi:10.1016/0304-405X(87)90009-2 tions,” Springer, New York, 2001.
[17] S. L. Heston and S. Nandi, “A Closed-Form GARCH [28] J. Vecer, “Stochastic Finance: A Numeraire Approach,”
Option Valuation Model,” The Review of Financial Stud- Chapman & Hall/CRC Press, London, 2011.
ies, Vol. 13, No. 1, 2000, pp. 585-625. [29] H. Y. Wong and Y. W. Lo, “Option Pricing with Mean
doi:10.1093/rfs/13.3.585 Reversion and Stochastic Volatility,” European Journal
[18] R. J. Elliot, T. K. Siu and L. Chan, “Option Pricing for of Operational Research, Vol. 197, No. 1, 2009, pp. 179-
GARCH Models with Markov Switching,” International 187. doi:10.1016/j.ejor.2008.05.014
Journal of Theoretical and Applied Finance, Vol. 9, No. [30] D. Duffie, J. Pan and K. Singleton, “Transform Analysis
6, 2006, pp. 825-841. doi:10.1142/S0219024906003846 and Asset Pricing for Affine Jump-Diffusion,” Econo-
[19] P. Christoffersen, S. Heston and K. Jacobs, “Option Val- metrica, Vol. 68, No. 6, 2000, pp. 1343-1376.
uation with Conditional Skewness,” Journal of Eco- no- doi:10.1111/1468-0262.00164
mtrics, Vol. 131, No. 1-2, 2006, pp. 253-284. [31] L. Cao and Z. F. Guo, “Applying Gradient Estimation
[20] L. Mercuri, “Option Pricing in a GARCH Model with Technique to Estimate Gradients of European Call fol-
Tempered Stable Innovations,” Finance Research Letter, lowing Variance-Gamma,” Proceedings of Global Con-
Vol. 5, No. 3, 2008, pp. 172-182. ference on Business and Finance, Vol. 6, No. 2, 2011, pp.
doi:10.1016/j.frl.2008.05.003 12-18.
[21] A. M. Badescu and R. J. Kulperger, “GARCH Option [32] L. Cao and Z. F. Guo, “Applying Variance Gamma Cor-
Oricing: a Semiparametric Approach,” Insurance Mathe- related to Estimate Optimal Portfolio of Variance Swap,”
matics and Economics, Vol. 43, No. 1, 2008, pp. 69-84. Proceedings of International Conference of Financial
doi:10.1016/j.insmatheco.2007.09.011 Engineering, London, 6-8 July 2011.