We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
Arithmetic Asian Options with Continuous Sampling
Jin B. Zhang!
Version: May 1999
Key words: Asian options, Arithmetic mean, Partial Differential Equation
JEL Classification Code: G13
"Department of Economics and Finance, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon,
Hong Kong. Tel: (852) 2788-7315, Fax: (852) 2788 8806, email: efjzhang@[Link]. This project was
supported by City University of Hong Kong. The author acknowledges Peter Carr, Ernie Ou, Xueping
‘Wa, Jie-An Yan for helpful comments end suggestions. Arithmetic Asian Options with Continuous Sampling
Abstract
This paper studies the pricing of Arithmetic Asian options with continuous sampling
We derive a new analytical approximate formula to price and hedge the Arithmetic Asian
options. The correction to the analytical approximate formula can be evaluated by solving
2 Partial Differential Equation (PDE) numerically. Numerical experiments show that the
error of our semi-analytical approach, ie., analytical approximation with the correction, is
at least of the order of 10-%, for the options with wide range of parameters tested in this,
paper. The accuracy can be easily pushed even higher by decreasing the grid size for the
computation of the correction term. The CPU time taken for the numerical computation
is one to two seconds. Our method is more accurate than any existing methods in the
literature, and faster than other PDE methods. With the error well-controlled, our results
can be used as a benchmark to justify the error computed by other approximation methods,
including Monte Carlo simulation. 1 Introduction
Asian option is a kind of path-dependent option in which payoff function depends on the
historical average of the underlying asset. It is called average rate option if the average
price is in the position of asset price in the payoff function. It is called average strike option
if the average price is in the position of strike price in the payoff function. The Asian option
is separated into two types: Arithmetic Asian option and Geometric Asian option based
on the way of taking averages. Geometric Asian option is easy to price and hedge since
a closed-form solution is available, while Arithmetic Asian option is difficult to price and
hedge.
Asian option is very useful in financial industry. It can be used to hedge a thinly traded
asset over @ certain period of time. The hedge is cheaper than a portfolio of plain vanilla
options. In addition to the cost effectiveness, the Asian option has another advantage over
the plain vanilla option that it can be used to protect. price manipulation by either party
of the contract on the maturity date.
Pricing and hedging Arithmetic Asian option is very difficult, because of the intractabil-
ity of the arithmetic average of lognormal process. closed-form solution has not been found,
and it seems unlikely that one will be found any time soon. Huge amount of efforts have
been concentrated on approximate method. They fall into three categories.
The first is numerical approach. Kemna & Vorst (1990) first derive the closed-form
solution of Geometric Asian option, and evaluate Arithmetic Asian option by using Monte
Carlo simulation with variance reduction technique, in which the geometric average is used
as a control variate. But the variance can only be used to estimate the error with certain
confidence interval, say 95%, can not be used to judge the maximum error. Carverhill &
Clewlow (1990) calculate the density of the sum of random variable using the fast Fourier
transformation approach, they can not justify the maximum error either. Roger & Shi
(1995) and Aliziary, Decamps & Koehl (1997) compute the price of Asian option by solving
the PDE numerically with finite difference method.
3 The second is analytical approximation, Turnbull & Wakeman (1991) find an approxi-
mate formula by matching the first several moments of arithmetic average with a lognormal
process, they have to use Monte Carlo simulation as a benchmark to justify the error. Levy
(1992) derives the approximate formula for the discrete sampling case by matching only
the first two moments, he also use the results of Monte Carlo simulation as the true value
for the error estimation. Geman & Yor (1993) find a quasi-explicit solution of the Arith-
metic Asian option in term of inverse Laplace transformation of a confluent hypergeometric
function, they have to use numerical technique to invert the Laplace transform (see Geman
& Eydeland 1995), and finally they are lacking an error bound. Milevsky & Posner (1998)
approximate the sum of lognormals by using Reciprocal Gamma distribution, and find
a new approximate formula for the Arithmetic Asian option, they suffer the same prob-
lem of lacking an error bound justification. Posner & Milevsky (1998) approximate the
state-price density with first four moments in Johnson’s family, derive another closed-form
approximate formula, but not yet solve the problem of the error estimation.
The third is the lower and upper bound estimation. Roger & Shi (1995) provide a
lower and upper bounds of the Asian option by computing the expectation based on some
zero-mean Gaussian variable. But the difference between their upper bound and lower
bound is quite big, can be as large as 6% of the option value. Chalasani, Jha & Varikooty
(1998) use trinomial tree model to estimate the lower bound and upper bound for the Asian
option with discrete sampling. The difference between their upper bound and lower bound
is smaller, 0.2% of the option price. But their algorithm is not so straight forward that one
may not easily reproduce.
In this paper, we present a semi-analytical approach to price Arithmetic Asian option
with continuous sampling. The method is more accurate than any existing methods in the
literature, and faster than other PDE methods. We derive a new analytical approximation
formula by using a singularity-removing technique. And the correction to this approximate
formula is shown to be governed by a nice PDE with well-behaved coefficients and therefore can be solved fastly and accurately with numerical method.
This paper is organized as follow. In §2, we present our major results on the analytical
approximation and error estimation. The proofs of these results are attached in appendices.
We then numerically estimate the correction terms in §3 and compare our results with other
results available in the literature, Finally we conclude the paper and discuss the reason
why our approach is faster and more accurate than other PDE methods.
2 Pricing and hedging formulas
There are two kinds of Asian options: average rate option and average strike option. Aver-
age rate option is considered in this paper. The result can be extended to deal with average
strike option.
For an average rate option, the strike is fixed, denoted as EB. Its payoff at maturity date
T is given by following formula
max(3 — £,0) (ly
where 3 is the average price of the underlying over a certain period of time, eg., (Zo.
We only consider the pricing formula within the averaging period, ie., Ty)