pricing options and equity-indexed annuities in...
TRANSCRIPT
Title Pricing options and equity-indexed annuities in regime-switching models by trinomial tree method
Author(s) Yuen, Fei-lung; 袁飛龍
Citation
Issue Date 2011
URL http://hdl.handle.net/10722/133208
Rights unrestricted
PRICING OPTIONS AND
EQUITY-INDEXED ANNUITIES IN
REGIME-SWITCHING MODELS BY
TRINOMIAL TREE METHOD
by
YUEN FEI LUNG
A thesis submitted in partial fulfillment of the requirements for
the Degree of Doctor of Philosophy
at The University of Hong Kong.
December 2010
Abstract of thesis entitled
PRICING OPTIONS AND
EQUITY-INDEXED ANNUITIES IN
REGIME-SWITCHING MODELS BY
TRINOMIAL TREE METHOD
Submitted by
YUEN FEI LUNG
for the degree of Doctor of Philosophy
at The University of Hong Kong
in December 2010
Starting from the well-known paper in pricing vanilla European call and put
options by Black and Scholes in 1973, there are many different research papers
on option valuation. The basic assumption of the Black-Scholes model is that the
price of the underlying asset, usually the stock, is a geometric Brownian motion.
The Markov regime-switching model (MRSM) introduced by Hamilton (1989)
improves the adaptability of the Black-Scholes model by allowing the parameters
of the stock price process change according to the financial situation. However,
due to the additional uncertainty brought by the Markov process, option pricing
in the MRSM is complicated and is usually done by simulation, or by solving a
system of partial differential equations.
Lattice model (or tree model) provides a simple way to price derivatives. In the
existing literature, additional branches are required in the lattice for derivative
pricing under the MRSM. In this case, the lattice is not recombining and its
efficiency is reduced significantly. Based on the trinomial tree model introduced
by Boyle (1988), a multi-state trinomial tree is introduced to price various options
in the MRSM. The key idea is to use the same lattice to accommodate all different
regimes by adjusting the probability measure. The method is simple and efficient.
The numerical results of the option prices obtained by this method are analyzed.
There are different MRSMs. The stock price in the MRSM of Elliott, Chan
and Siu (2005) is a continuous process while the price of derivatives jumps when
the regime switches. In order to study the jump risk of assets in details, the jump
diffusion model of Naik (1993) is studied and extended. The multi-state trinomial
tree is used to price options in this jump diffusion model and the nature of jump
risk is discussed.
Asian option is a strong path-dependent option. The payoff of Asian option
depends on the whole path of the asset price process and pricing Asian options
is not an easy task in the MRSM. Equity-indexed annuities (EIAs) are derivative
products linked to the performance of an equity index and is now a popular
product in the market. The payoff structure of EIAs can take different forms
according to the needs of the investors. The multi-state trinomial tree model is
modified using the idea of Hull and White (1993) and applied to Asian options.
The problem of quadratic approximation suggested by Hull and White (1993) is
identified and solved. Using the result of Asian option, the price of Asian-option-
related EIAs is obtained by an iterative equation.
Declaration
I declare that this thesis represents my own work, except where due acknowl-
edgements are made, and that it has not been previously included in a thesis,
dissertation or report submitted to this University or to any other institution for
a degree, diploma or other qualification.
Signed
YUEN FEI LUNG
i
Acknowledgements
I would like to express my most sincere gratitude to my supervisor, Prof.
Yang, Department of Statistics and Actuarial Science, the University of Hong
Kong, for his guidance. He always spends a large part of his valuable time to
teach us and monitor our progress. I would also like to express my thanks to
all of the department members for the help and support in these four years of
study.
ii
Table of Contents
Declaration i
Acknowledgements ii
Table of Contents iii
List of Tables v
1 Introduction 1
2 Multi-state Trinomial Tree 7
2.1 Introduction 7
2.2 Multi-state Trinomial Lattice 8
2.3 Numerical Results and Analysis 15
2.4 Alternative Models 27
2.5 Hedging Risk of Regime Switching 36
2.6 Conclusions 42
3 Pricing Regime-switching Risk 44
3.1 Introduction 44
3.2 Jump Diffusion Model 45
3.3 Arrival Rates of Jumps under Risk Neutral Measure 54
3.4 Trinomial Tree Pricing under Jump Diffusion Model 59
3.5 Numerical Results and Analysis 63
3.5.1 Jump Risks are Not Priced 63
3.5.2 Jump Risks are Priced 73
3.6 Conclusions 77
iii
4 Pricing Asian Option and Related EIAs 78
4.1 Introduction 78
4.2 A Modified Trinomial Lattice 79
4.3 Pricing Equity-Indexed Annuities 84
4.4 Numerical Results and Analysis 92
4.5 Conclusions 102
5 Concluding Remarks 107
References 110
iv
List of Tables
2.1 Comparison of different methods in pricing Euroean call option in
MRSM 16
2.2 Pricing European call option with trinomial tree 18
2.3 Pricing European put option with trinomial tree 18
2.4 Pricing American call option with trinomial tree 20
2.5 Pricing American put option with trinomial tree 21
2.6 Pricing down-and-out barrier call option with trinomial tree 22
2.7 Pricing double barrier call option with trinomial tree 23
2.8 Price of double barrier call options with different barrier levels 24
2.9 Pricing European call option with trinomial tree: great derivation
in volatilities 25
2.10 Pricing European call option under model with three regimes 27
2.11 Pricing European call option under model with three regimes using
trinomial tree 34
2.12 Pricing European call option under model with three regimes using
combined trinomial tree 35
3.1 Pricing European call option when jump risk is not priced 65
3.2 Pricing European put option when jump risk is not priced 65
3.3 Comparison of European call option prices in jump and non jump
models 67
3.4 Comparison of European put option prices in jump and non jump
models 67
3.5 Comparison of American call option prices in jump and non jump
models 70
v
3.6 Comparison of American put option prices in jump and non jump
models 71
3.7 Comparison of down-and-out barrier call option prices in jump and
non jump models 72
3.8 Comparison of European call option prices with priced and non-
priced jump risk 74
3.9 Comparison of European put option prices with priced and non-
priced jump risk 75
3.10 Comparison of American call option prices with priced and non-
priced jump risk 75
3.11 Comparison of American put option prices with priced and non-
priced jump risk 76
3.12 Comparison of down-and-out call option prices with priced and
non-priced jump risk 76
4.1 Comparison of the prices of (Eurpean type) average price call op-
tions in simple BS model (linear approximation of representative
value) 93
4.2 Comparison of the prices of (American type) average price call
options in simple BS model (linear approximation of representative
value) 94
4.3 Comparison of the prices of (Eurpean type) average price call op-
tions in simple BS model (quadratic and modified quadratic ap-
proximation, simple average asset price) 96
4.4 Comparison of the prices of (American Type) average price call
options in simple BS model (quadratic and modified quadratic ap-
proximation, simple average asset price) 97
vi
4.5 Comparison of the price of average price call options in MRSM
found by different methods I 100
4.6 Comparison of the price of average price call options in MRSM
obtained by different methods II 101
4.7 Price of average price call options with early exercise option in
MRSM 104
4.8 Price of one-year EIA in MRSM 105
4.9 Price of unit annual reset EIA in MRSM 106
vii
Chapter 1
Introduction
In the past decades, there are a lot of researches on option pricing and many
models have been proposed for the underlying asset. Markov regime-switching
model (MRSM), which allows the parameters of the market model controlled by
a Markov process, have become one of the popular models recently. It is found
to be consistent with the market data and has gained its popularity because it
can reflect the information of the market environment which cannot be modelled
by linear Gaussian process solely. Markov process can ensure the parameters
changing according to the market environment and preserve the simplicity of the
model. It is also consistent with the efficient market hypothesis, all the effects of
the information about the stock price are reflected on the current stock price.
Merton (1969) uses stochastic differential equation (SDE) to study the continuous-
time portfolio theory; since then geometric Brownian motion (GBM) becomes one
of the most commonly used stochastic process in financial mathematics because of
its highly random nature and simplicity. Black and Scholes (1973), based on the
work of Merton (1969), give the no-arbitrage price of an European option when
the price of the underlying assets is a GBM. The development of the formulae of
various derivatives is easy based on this model. However, when the parameters
1
of the SDE are not constant but controlled by a Markov process, the price of the
options cannot be found easily.
There are many papers about option pricing in multi-state models such as
MRSM. Some of them use lattice model. Boyle (1988) uses a pentanomial tree
lattice to find the price of derivatives with two states. Kamrad and Ritchken
(1991) suggest a (2k + 1)-branch model for k sources of uncertainty. Bollen
(1998) constructs a pentanomial tree which is excellent in finding a fair price of
European option and American option in two-regime situation. Aingworth, Das
and Motwani (2006) use a lattice with 2k branches to study the k-state model.
The increasing number of branches reduces the efficiency of the tree models and
so some other methods are used by different researchers to price derivatives.
Buffington and Elliott (2002) find the price of European option and American
option using partial differential equations (PDE). Mamon and Rodrigo (2005)
find an explicit solution to European options in regime-switching economy by
considering the solution of the associated PDE. Elliott, Chan and Siu (2005) use
Esscher transform to find the explicit price formula for European option. Boyle
and Draviam (2007) find the price of exotic options under regime switching using
PDE. PDE has become the focus of most researchers for option valuation in
MRSM as it is flexible.
Since the introduction of binomial tree model by Cox, Ross and Rubinstein
2
(1979), lattice model is one of the most popular methods to calculate the price of
simple options like European option and American option. Various lattice models
are suggested after that, see, for example, Jarrow and Rudd (1983) and Boyle
(1986). Trinomial tree model of Boyle (1986) is highly flexible. The extra middle
branch of it gives one degree of freedom to the lattice and that makes the lattice
very useful in regime-switching model. Boyle and Tian (1998) use this property
of the trinomial tree to price double barrier option and Bollen (1998) uses the
similar idea to construct an efficiently recombining tree. In Chapter 2, a new
method is introduced so that a trinomial tree can be used to find a fair price
of options under MRSM efficiently. The MRSM used by Buffington and Elliott
(2002) is a good and popular choice, and it is used to illustrate the idea of the
multi-state trinomial tree model.
Regime-switching market is not complete. There are many different ways
to find a fair price of the options. Miyahara (2001) uses the minimal entropy
martingale measure to find the price which maximizes the exponential utility.
Elliott, Chan and Siu (2005) use Esscher transform to obtain a fair price. Guo
(2001) introduces change-of-state (COS) contracts to complete the market. Naik
(1993) shows that the price of options can also be found by the given market
price of risks. In the MRSM of Buffington and Elliott (2002), stock price is a
continuous process and jump risk seems not a systematic risk in a certain sense.
3
In Chapter 3, the two-regime model of Naik (1993) is extended to k states and the
option prices under this market model are tested using the multi-state trinomial
tree.
Various modern insurance products are introduced into the market in these
years, including equity-indexed annuities (EIAs) and variable annuities (VAs).
Their payoffs are linked to the preformance of some assets and indices. Different
special features such as minimum yearly return, ceiling rate and participation
rate make the products more flexible and more complicated. Tiong (2000) gives a
details study on EIAs and prices the products using Esscher transform. Lee (2002,
2003) proposes several new designs of EIAs and finds their explicit formulae under
the Black-Scholes framework. Lin and Tan (2003) consider an Asian-option-
related EIA and price it under a stochastic interest rate model. Asian option is
a strong path-dependent option of which the value of payoff depends on the path
of the asset price process. Its valuation is complex under MRSM. In Chapter
4, the multi-state trinomial tree model is used to price Asian options in MRSM
using the idea of Hull and White (1993). The problem of quadratic approximation
suggested by Hull and White (1993) is identified and solved. Using the Markovian
property of the regime-switching process, the price of Asian-option-related EIA
can be obtained by an iterative equation with the price of Asian option.
The theoretical MRSM of Buffington and Elliott (2002) is presented here for
4
completeness.
We consider the real world probability space (Ω,F , P ). We let T be the
time interval [0, T ] that is being considered. W (t)t∈T is a standard Brownian
motion on (Ω,F , P ). X(t)t∈T is a continuous-time Markov process with finite
state space X := x1, x2, . . . , xk, which represents the economic condition and is
observable. A set of unit vector e1, e2, . . . , ek where xi = ei = (0, . . . , 1, . . . , 0) ∈
Rk is used to denote the current state of the Markov process. For simplicity, the
state ei is called the state i. We denote the set of states to be K := 1, 2, . . . , k.
Let A(t) = [aij(t)]i,j=1,...,k be the generator of the Markov process. By the
semi-martingale representation theorem,
X(t) = X(0) +
∫ t
0
X(s)A(s)ds+M(t), (1.1)
where M(t)t∈T is a Rk-valued martingale with respect to the P -augmentation
of the natural filtration generated by X(t)t∈T .
There are two basic investment tools in the model, one is bond and the other
is stock. The market interest rate is denoted by r(t,X(t))t∈T which depends
on the current state of economy only,
r(t) := r(t,X(t)) = 〈r,X(t)〉, (1.2)
where r := (r1, r2, . . . , rk); ri > 0 for all i ∈ K and 〈·, ·〉 denotes the inner product
in Rk.
5
The bond price process B(t)t∈T satisfies the equation
dB(t) = r(t)B(t)dt, B(0) = 1. (1.3)
The rate of return and the volatility of the stock price process are denoted
by µ(t,X(t))t∈T and σ(t,X(t))t∈T , respectively. Similar to the interest rate
process, they are only affected by the state of economy,
µ(t) := µ(t,X(t)) = 〈µ,X(t)〉, σ(t) := σ(t,X(t)) = 〈σ,X(t)〉, (1.4)
where µ := (µ1, µ2, . . . , µk) and σ := (σ1, σ2, . . . , σk) with σi > 0 for all i ∈ K.
The stock price process S(t)t∈T is a Markov-modulated geometric Brownian
motion. Z(t) is the cumulative rate of return of the stock over time interval [0, t],
that is, Z(t) = ln(S(t)/S(0)). Then, we have
S(t) = S(u) exp(Z(t)− Z(u)), (1.5)
dS(t) = µ(t)S(t)dt+ σ(t)S(t)dW (t), (1.6)
Z(t) =
∫ t
0
(µ(s)− 1
2σ2(s)
)ds+
∫ t
0
σ(s)dW (s). (1.7)
6
Chapter 2
Multi-state Trinomial Tree
2.1 Introduction
Since the binomial tree model was introduced by Cox, Ross and Rubinstein
(1979), the lattice model has become a popular way to calculate the price of
simple options like the European option and the American option. It is mainly
because the lattice method is simple and easy to implement. Various lattice mod-
els have been suggested after that, see, for example, Jarrow and Rudd (1983) and
Boyle (1986). The trinomial lattice of Boyle (1986) is highly flexible, and has
some important properties that the binomial lattice lacks. The extra branch of
the trinomial tree gives one degree of freedom to the lattice and makes it very
useful in the regime-switching model. Boyle and Tian (1998) use this property
of the trinomial tree to price double barrier options, and propose an interesting
method to eliminate the error in pricing barrier options. Bollen (1998) uses a
similar idea to construct an efficiently recombining tree. There are many other
researches using tree methods for derivative pricing in multi-state model. Boyle
(1988) uses a pentanomial tree model to calculate the price of derivatives with
two states. Kamrad and Ritchken (1991) suggest a 2k + 1-branch model for k
7
sources of uncertainty. Aingworth, Das and Motwani (2006) uses a lattice with
2k branches to study the k-state regime-switching model. However, when the
number of states is large, the tree models mentioned above are not efficient. We
propose a multi-state trinomial tree to price the options in a regime-switching
model. The trinomial tree that we propose is recombining. Instead of increasing
the number of branches in the tree for different regimes, we use different sets
of risk neutral probabilities for different regimes. Since it is a recombining tree,
option valuation is fast, simple and efficient using this method.
2.2 Multi-state Trinomial Lattice
In CRR binomial tree, when σ is the asset’s volatility and ∆ is the size of time
step, the ratios of change are given by eσ√
∆ and e−σ√
∆, the risk neutral proba-
bilities of getting up and down are specified so that the expected rate of return
of the stock matches the risk-free interest rate. In the trinomial model, with
constant risk-free interest rate and volatility, the stock price is allowed to remain
unchanged, go up or go down by a ratio. The upward ratio must be greater than
eσ√
∆ to ensure that a risk neutral probability measure exists. If πu, πm, πd are
the risk neutral probabilities of the stock price increases, remains unchanged and
decreases in the tree, r is the risk-free interest rate, then, for a constant λ, we
8
have
πueλσ√
∆ + πm + πde−λσ√
∆ = er∆, (2.1)
(πu + πd)λ2σ2∆ = σ2∆ (2.2)
λ should be greater than 1 so that the risk neutral probability measure exists.
In the literature, λ is usually taken to be√
3 (Figlewski and Gao (1999), Baule
and Wilkens (2004)) or√
1.5 (Boyle (1988), Kamrad and Ritchken (1991)). After
fixing the value of λ, the risk neutral probabilities can be found and the whole
lattice can be constructed.
However, in the Markov regime-switching model (MRSM), the risk-free inter-
est rate and the volatility are not constant. They change according to the Markov
process. More branches can be introduced into the lattice so that extra regimes
and information can be incorporated in the tree, for example, Boyle and Tian
(1988), Kamrad and Ritchken (1991) and Bollen (1998). The increasing number
of branches makes the lattice model more complex. Bollen (1998) suggests an
excellent recombining tree to solve the option prices in two-regime case, but the
multi-regime problem still cannot be solved effectively.
Here, we propose a different way to construct the tree. Instead of increasing
the number of branches in the tree, we change the risk neutral probability measure
under different regimes so that a recombining tree allows more regimes. The
9
method relies greatly on the flexibility of the trinomial tree model and the core
idea of the multi-state trinomial tree model here is to change the probability
measure to accommodate different regimes in the same recombining lattice.
Assuming that there are k regimes in the MRSM, the corresponding risk-free
interest rate and volatility of price of the underlying asset under these regimes
be r1, r2, . . . , rk and σ1, σ2, . . . , σk, respectively. The up-jump ratio of the lattice
is taken to be eσ√
∆. For a lattice which can be used by all regimes,
σ > max1≤i≤k
σi. (2.3)
For the regime i, let πiu, πim, π
id are the risk neutral probabilities of the stock
price increases, remains unchanged and decreases in the branch of the tree. Then,
similar to the simple trinomial tree model, the following set of equations can be
obtained, for all i ∈ K,
πiueσ√
∆ + πim + πide−σ√
∆ = eri∆, (2.4)
(πiu + πid)σ2∆ = σ2
i ∆. (2.5)
If λi is defined to be σ/σi for each i, then, λi > 1 and the values of πiu, πim, π
id
10
can be found, in terms of λi,
πim = 1− σ2i
σ2= 1− 1
λ2i
, (2.6)
πiu =eri∆ − e−σ
√∆ − (1− 1/λ2
i )(1− e−σ√
∆)
eσ√
∆ − e−σ√
∆, (2.7)
πid =eσ√
∆ − eri∆ − (1− 1/λ2i )(e
σ√
∆ − 1)
eσ√
∆ − e−σ√
∆. (2.8)
Therefore, the set of risk neutral probabilities depends on the value of σ. In
order to ensure that σ is greater than all σi, we might take
σ = max1≤i≤k
σi + (√
1.5− 1)σ, (2.9)
where σ is the arithmetic mean of σi. Root mean square is another suitable choice
of σ. The most efficient choice of σ is unknown. In this section, σi are assumed to
be not greatly different from each other; and the selection of σ is not important
as long as it is comparable with the volatilities of different regimes.
After the whole lattice is constructed, the main idea of the pricing method is
presented here. We let T be the expiration time of the option, N be the number
of time steps, then ∆ = T/N . At time step t, there are 2t+1 nodes in the lattice,
the node is counted from the lowest stock price level, and St,n denotes the stock
price of the nth node (it starts from the 0th node, for convenience) at time step
t. As all the regimes are sharing the same lattice and the regime state cannot be
reflected by the position of the nodes, each of the nodes has k possible derivative
11
prices corresponding to the regime state. Let Vt,n,j be the value of the derivative
at the nth node at time step t under the jth regime state.
The transition probability of the Markov process can be found by the generator
matrix. The generator matrix is assumed to be constant and taken to be A. We
define pij(∆) as the transition probability from regime state i to regime state
j for the time interval with length ∆; and for simplicity, it is denoted by pij.
The transition probability matrix, denoted by P , can be found by the following
equation,
P (∆) =
p11 · · · p1k
.... . .
...
pk1 · · · pkk
= eA∆ = I +∞∑l=1
(∆)lAl/l!. (2.10)
With the transition probability matrix, the price of a derivative at each node
can be found by iteration. We start from the expiration time, for example, for
an European call option with strike price K,
VN,n,i = (SN,n −K)+ for all states i, (2.11)
where SN,n = S0 exp[(n−N)σ√
∆].
We assume that the Markov process is independent of the Brownian motion
under the real market measure and the transition probabilities are not affected by
the use of risk neutral measure. With the derivative payoff at expiration, using
12
the following equation recursively,
Vt,n,i = e−ri∆
[k∑j=1
pij(πiuVt+1,n+2,j + πimVt+1,n+1,j + πidVt+1,n,j)
], (2.12)
the price of the option under all regimes can be obtained.
Regime switching is another source of risk because we do not know the time
of regime switching before it takes place. Moreover, due to regime switching,
the market is incomplete and the derivatives do not have a unique no-arbitrage
price. There are many ways to treat the additional risk from regime switching,
for example, not pricing the regime-switching risk (Bollen (1998)), or introducing
change-of-state (COS) contracts into the model (Guo (2001)). The first way is
used in the previous calculation. Some derivatives benefit while some are suffered
by the regime switching which depends on the initial regime, the transition prob-
abilities and the structure of the derivatives. It is also hard to make a compromise
in choosing appropriate transition probabilities if the market is not complete. It
is a reasonable choice of not pricing the regime-switching risk as long as there is
no arbitrage opportunity in the market and the Markov process is independent of
the Brownian motion. New securities COS can be introduced into the model to
complete the market. However, the regime is referring to the macroeconomic con-
dition, hedging or pooling regime-switching risk is complicated in the incomplete
market and insurance companies might not be willing to take the risk. We as-
sume that there are not suitable COS securities in the market. The risk premium
13
comes from the risk of Brownian motion only. We focus on the multi-state trino-
mial tree model in this chapter and the regime-switching risk will be discussed in
details in the other parts.
If we have to price American option, the value of the option at each node
under different regimes can be compared with the payoff of exercising the op-
tion immediately; and the larger value is used as the price for iteration. The
calculation is similar to the valuation of American option in simple lattice model.
For barrier option, the idea of Boyle and Tian (1998) can be applied. The
whole lattice is constructed from the lower barrier. As the initial price of the
underlying asset is not necessarily at the grid, a quadratic approximation is used
to calculate the price of the down-and-out option. The price of a down-and-in
option can be found using the idea that the sum of down-and-out option and
down-and-in option is a vanilla option. For a double barrier option, we have used
the flexibility of trinomial tree lattice but the value of σ is in fact not fixed. We
can set both of the upper and lower barriers on the node level by a fine adjustment
of the lattice parameter σ. The price of curved barrier option and discrete-time
barrier option can also be found, using a similar method suggested by Boyle and
Tian (1998).
The regime is observable, the payoff of the derivatives can depend on the
regime state, because the prices of the derivative under all regimes are found in
14
each node, the model is also applicable to price this kind of derivatives.
2.3 Numerical Results and Analysis
Based on the model introduced in the last section, we calculate the prices of
various options in different regimes. In this section we study the European option,
the American option, the down-and-out barrier option, the double barrier option,
and their prices are calculated by the multi-state trinomial tree. The results give
us some insights into the price of derivatives in the MRSM and the effects of
regime switching. First of all, the model is tested by comparing with the results
given by Boyle and Draviam (2007).
Table 2.1 shows that the option price obtained by using the trinomial lattice
is very close to the value obtained by using the analytical solutions derived in
Naik (1993), and also close to those obtained using partial differential equations in
Boyle and Draviam (2007). This verifies that the trinomial tree method proposed
in this chapter is applicable.
We now study the values of different types of options in the MRSM. The
underlying asset is assumed to be a stock with initial price of 100, following a
geometric Brownian motion of a two-regime model with no dividend. In Regime
1, the risk-free interest rate is 4% and the volatility of stock is 0.25; in Regime
15
Table 2.1: Comparison of different methods in pricing Euroean call option in
MRSM
European Call Option I
Regime 1 Regime 2
S0 Naik B&D Lattice Naik B&D Lattice
94 5.8620 5.8579 5.8615 8.2292 8.2193 8.2297
96 6.9235 6.9178 6.9229 9.3175 9.3056 9.3181
98 8.0844 8.0775 8.0827 10.4775 10.4647 10.4772
100 9.3401 9.3324 9.3369 11.7063 11.6929 11.7049
102 10.6850 10.6769 10.6828 13.0008 12.9870 13.0001
104 12.1127 12.1045 12.1108 14.3575 14.3436 14.3571
106 13.6161 13.6082 13.6143 15.7729 15.7591 15.7725
European Call Option II
Regime 1 Regime 2
S0 Naik B&D Lattice Naik B&D Lattice
94 6.2748 6.2705 6.2760 7.8905 7.8844 7.8943
96 7.3408 7.3352 7.3422 8.9747 8.9680 8.9789
98 8.5001 8.4938 8.5010 10.1335 10.1264 10.1374
100 9.7489 9.7423 9.7489 11.3641 11.3568 11.3673
102 11.0820 11.0755 11.0833 12.6631 12.6659 12.6674
104 12.4937 12.4877 12.4959 14.0267 14.0197 14.0317
106 13.9777 13.9726 13.9805 15.4510 15.4446 15.4565
†S0 is the initial stock price and the strike price is set to be 100. The volatilities of the stock
in Regime 1 and Regime 2 are 0.15 and 0.25 respectively. The option lasts for 1 year and the
lattice is set to have 1000 time steps. The generators of the regime-switching process are −0.5 0.5
0.5 −0.5
and
−1 1
1 −1
for the above two sets of data respectively.
16
2, the risk-free interest rate is 6% and the volatility of stock is 0.35. All options
expire in one year with strike price equal to 100. The generator for the regime-
switching process is taken to be −0.5 0.5
0.5 −0.5
.
The transition probabilities of the branch of state up, middle and down with 20
time steps are 0.177003, 0.641304 and 0.181693 in Regime 1; 0.351844, 0.296956
and 0.351200 in Regime 2, respectively. These values depend on the size of time
step, but the values with other sizes of time step are not much different from
these values because the time step is small in general. The values in 20-step case
can already give the idea of the size of the risk neutral probabilities. We study
the numerical results to see if there are any special characteristics of the prices
of these derivatives and the convergence properties of the model.
Tables 2.2 and 2.3 show that the convergence rate of the European call and
the European put options is fast. We know that the price of European call options
and European put options found by the CRR model converges with order 1, that
is, the error of the price is halved if the number of time steps is doubled (Baule
and Wilkens (2004), Omberg (1987)). We can see from the tables that most of
the ratios shown in the tables are close to 0.5. However, it is not the case for
the European call option when the number of iterations is large for Regime 2.
17
Table 2.2: Pricing European call option with trinomial tree
European Call Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 12.6282 0.0654 0.4954 15.7560 0.0043 0.5581
40 12.6936 0.0324 0.5000 15.7603 0.0024 0.5417
80 12.7260 0.0162 0.5000 15.7627 0.0013 0.4615
160 12.7422 0.0081 0.4938 15.7640 0.0006 0.6667
320 12.7503 0.0040 0.5000 15.7646 0.0004 0.2500
640 12.7543 0.0020 0.5000 15.7650 0.0001 1.0000
1280 12.7563 0.0010 0.5000 15.7651 0.0001 1.0000
2560 12.7573 0.0005 15.7652 0.0001
5120 12.7578 15.7653
†N is the number of time steps used in calculation. Diff is referring to the difference in price
calculated using various numbers of time steps and ratio is the ratio of the difference.
Table 2.3: Pricing European put option with trinomial tree
European Put Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 8.37107 0.05781 0.4959 10.2660 0.0119 0.5210
40 8.42888 0.02867 0.4977 10.2779 0.0062 0.5000
80 8.45755 0.01427 0.4989 10.2841 0.0031 0.5161
160 8.47182 0.00712 0.5000 10.2872 0.0016 0.5000
320 8.47894 0.00356 0.5000 10.2888 0.0008 0.5000
640 8.48250 0.00178 0.5000 10.2896 0.0004 0.5000
1280 8.48428 0.00089 0.4944 10.2900 0.0002 0.5000
2560 8.48517 0.00044 10.2902 0.0001
5120 8.48561 10.2903
18
This is because the approximation errors for the two regimes are different and
the round-off error. Boyle (1988) shows that using the trinomial tree model, the
approximation error is smaller if the three risk neutral probabilities of the tree are
almost equal with same number of time steps. In our case, we can see that the risk
neutral probabilities of Regime 1 are not as close as those of Regime 2. Therefore,
in Regime 2, the change in prices is smaller which implies a smaller approximation
error as shown in the numerical results in the tables. The differences between
the price changes for Regime 2 are less than one-tenth of that for Regime 1 most
of the time. However, the prices of the asset in both regimes affect one another.
The larger pricing error in Regime 1 affects the accuracy of the price in Regime 2.
The result is that the value in Regime 2 converges in a faster, but more unstable
way. On the other hand, the error in Regime 2 is small compared with that in
Regime 1; thus the convergence patterns in Regime 1 are more stable. Moreover,
the change of prices in Regime 2 is small when the number of time steps is large.
The round-off error then becomes significant.
When we apply the put-call parity to each of the regimes, the interest rate
found in the two regimes are 4.37% and 5.63% respectively using the result of
5120 time steps. It is reasonable because both of them are between 4% and 6%,
the interest rate found by Regime 1 data is close to Regime 1 rate while the same
is true for Regime 2. Interestingly, the deviations between the current interest
19
Table 2.4: Pricing American call option with trinomial tree
American Call Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 12.6282 0.0654 0.4954 15.7560 0.0043 0.5581
40 12.6936 0.0324 0.5000 15.7603 0.0024 0.5417
80 12.7260 0.0162 0.5000 15.7627 0.0013 0.4615
160 12.7422 0.0081 0.4938 15.7640 0.0006 0.6667
320 12.7503 0.0040 0.5000 15.7646 0.0004 0.2500
640 12.7543 0.0020 0.5000 15.7650 0.0001 1.0000
1280 12.7563 0.0010 0.5000 15.7651 0.0001 1.0000
2560 12.7573 0.0005 15.7652 0.0001
5120 12.7578 15.7653
rate and the interest rate found by put-call parity in both regimes are equal
to 0.37%. It is due to the symmetry of two regimes in terms of the transition
probabilities. The mechanism behind and the meaning of it will be discussed in
the next chapter.
The result of the American option is similar to that of the CRR model. The
prices of the American call option found by the trinomial tree is the same as
the European call option. It is consistent with the fact that it is not optimal to
exercise an American call option before expiration if the underlying asset pays
no dividend. We know that this result is also true for the MRSM. The prices
of the American put option in the table are larger than those of the European
option, meaning that early exercise of the option is preferred and there are some
20
Table 2.5: Pricing American put option with trinomial tree
American Put Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 8.80315 0.05236 0.5107 10.8942 0.0007 2.5714
40 8.85551 0.02674 0.4862 10.8949 0.0018 0.1111
80 8.88225 0.01300 0.4869 10.8967 0.0002 0.5000
160 8.89525 0.00633 0.4945 10.8969 0.0001 0.0000
320 8.90158 0.00313 0.4984 10.8970 0.0000 N/A
640 8.90471 0.00156 0.4936 10.8970 0.0000 N/A
1280 8.90627 0.00077 0.4935 10.8970 0.0000 N/A
2560 8.90704 0.00038 10.8970 0.0000
5120 8.90742 10.8970
situations in which we have to exercise the American put option before expiration.
The convergence pattern of the American put option is more complicated than
the European one. The rate of convergence for Regime 2 is very fast, even faster
than that of the European put option. The American put option is optimal
to be exercised somewhere before the maturity, so the approximation error is
smaller than that of the European option. The convergence pattern of Regime
2 is unstable, which is consistent with the results for the European option case;
larger initial pricing error in Regime 1 and round-off error affect the convergence
of the price in Regime 2.
For the down-and-out barrier call option, the prices found in both regimes
are smaller than those of the European call option due to the presence of the
21
Table 2.6: Pricing down-and-out barrier call option with trinomial tree
Down-and-out Barrier Call Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 8.97860 -0.01239 -0.6917 9.73967 -0.02790 0.0487
40 8.96621 0.00857 -0.4831 9.71177 -0.00136 4.9779
80 8.97478 -0.00414 0.1304 9.71041 -0.00677 0.3840
160 8.97064 -0.00054 -0.2778 9.70364 -0.00260 0.3269
320 8.97010 0.00015 -0.4667 9.70104 -0.00085 1.2588
640 8.97025 -0.00097 -0.3505 9.70019 -0.00107 0.0748
1280 8.96928 0.00034 -0.2059 9.69912 -0.00008 2.1250
2560 8.96962 -0.00007 9.69904 -0.00017
5120 8.96955 9.69887
The barrier level is set to be 90.
down-and-out barrier. The prices in the two regimes are closer to each other
compared with the case of European option. Although the volatility in Regime 2
is greater and has a higher chance to achieve a higher value at expiration, the high
volatility also increases the chance of hitting the down-and-out barrier and thus
eliminates its advantage. The convergence pattern of barrier option is complex.
It is difficult to get any conclusions from the data. However, we can see that
apart from converging uniformly in one direction, the values of the option found
in Regime 1 oscillate and the differences still have a decreasing trend in absolute
value. It is likely due to the effect of quadratic approximation.
The price of the double barrier option can also be obtained by the trinomial
22
Table 2.7: Pricing double barrier call option with trinomial tree
Double Barrier Call Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 6.15869 -0.15826 0.7097 4.54096 -0.13822 0.6130
40 6.00043 -0.11232 0.7314 4.40274 -0.08473 0.5189
80 5.88811 -0.04845 0.4111 4.31801 -0.04397 0.3834
160 5.83966 -0.01992 0.5954 4.27404 -0.01686 0.6109
320 5.81974 -0.01186 0.5320 4.25718 -0.01030 0.5029
640 5.80788 -0.00631 0.6133 4.24688 -0.00518 0.6120
1280 5.80157 -0.00387 0.1731 4.24170 -0.00317 0.2145
2560 5.79770 -0.00067 4.23853 -0.00068
5120 5.79703 4.23785
The barrier level is set to be 70 and 150.
model. The method suggested by Boyle and Tian (1998) is adopted here. The
lattice is built from the lower barrier and touches the upper barrier by controlling
the value of σ used in the lattice. Table 2.7 shows the price of the double barrier
option with different numbers of time steps being used. The lower barrier is 70 and
the upper barrier is 150. The values decrease progressively and converge. Table
2.8 summarizes the values of the double barrier options with different barrier
levels using 1000 time steps. When the difference between the upper and lower
barriers is smaller, the price of the options is reduced as there is a higher chance
of touching the barrier and becoming out of value. The effect of barriers is
more significant for Regime 2 because the stock has a higher volatility in Regime
23
Table 2.8: Price of double barrier call options with different barrier levels
Double Barrier Call Option in Regime 1
90 80 70 60 50
110 0.00063 0.0249 0.0498 0.0544 0.0546
120 0.10229 0.4310 0.5773 0.5952 0.5970
130 0.71002 1.6257 1.9120 1.9422 1.9451
140 1.88418 3.4101 3.8049 3.8446 3.8463
150 3.30481 5.3336 5.8019 5.8474 5.8490
200 7.87455 10.8888 11.4649 11.5163 11.5183
Double Barrier Call Option in Regime 2
90 80 70 60 50
110 0.00004 0.0049 0.0202 0.0285 0.0297
120 0.01567 0.1446 0.2909 0.3385 0.3440
130 0.01933 0.7381 1.1160 1.2117 1.2210
140 0.73257 1.8882 2.5051 2.6410 2.6515
150 1.62095 3.4224 4.2422 4.4065 4.4181
200 6.65432 10.5198 11.7835 11.9909 12.0042
The price of the double barrier options with lower barrier of 90, 80, 70, 60, 50 and upper
barrier of 110, 120, 130, 140, 150, 200 in the two regimes are calculated using 1000 time steps.
24
Table 2.9: Pricing European call option with trinomial tree: great derivation in
volatilities
European Call Option
Regime 1 Regime 2
N Price Diff Ratio Price Diff Ratio
20 9.07428 0.37247 0.5368 19.9973 -0.0409 0.4572
40 9.44675 0.19995 0.4475 19.9564 -0.0187 0.4706
80 9.64670 0.08948 0.4641 19.9377 -0.0088 0.5000
160 9.73618 0.04153 0.4869 19.9289 -0.0044 0.4773
320 9.77771 0.02022 0.4936 19.9245 -0.0021 0.5238
640 9.79793 0.00998 0.4971 19.9224 -0.0011 0.4545
1280 9.80791 0.00496 0.5000 19.9213 -0.0005 0.6000
2560 9.81287 0.00248 19.9208 -0.0003
5120 9.81535 19.9205
The volatilities of the two regimes are 0.10 and 0.50 respectively.
2, hence it has a greater chance of reaching the barriers. When the difference
between the barriers increases, their effect on the barrier options is reduced and
the options in Regime 2 with a larger volatility have a higher price than that of
the same option in Regime 1. Their prices are lower than those of the vanilla call
option, which has prices of 12.7557 and 15.7651 in the two regimes, respectively,
found by trinomial tree with 1000 time steps.
We now consider a few more examples. We predict that the convergence rate
reduces if the volatilities of different regimes are largely different from each other.
We want to find if the prediction is true. All the other conditions are assumed
to be the same, but the volatilities of the asset under the two regimes become
25
0.10 and 0.50. The prices of the European call option are found. The risk neutral
probabilities of Regime 1 of 20 time steps case in the three branches are 0.0224138,
0.968941, 0.00864505, respectively. Most of the probabilities are distributed on
the middle branch.
The price of European option is positively related to the volatility and so
the value in Regime 1 decreases while the value in Regime 2 increases, when
compared with the results of previous example. The pricing error in Regime 1
is larger compared with the results in the previous example as a large σ is being
used to construct lattice. We can make use of the fact that the price of European
option converges with order 1 so that a better result can be obtained even with
a smaller number of time steps.
Next we consider a three-regime example. This example is used to examine
the efficiency of the trinomial tree under multi-state market. The interest rate and
the volatility in the three regimes are 4%, 5%, 6% and 0.20, 0.30, 0.40, respectively.
The initial price and strike price are both set as 100 and the generator matrix is
taken as −1 0.5 0.5
0.5 −1 0.5
0.5 0.5 −1
. (2.13)
The numerical results are shown in Table 2.10. They show that the conver-
26
Table 2.10: Pricing European call option under model with three regimes
European Call Option
Regime 1 Regime 2 Regime 3
N Price Diff Price Diff Price Diff
20 11.9484 0.1196 14.2232 0.0510 16.6246 -0.0143
40 12.0680 0.0582 14.2742 0.0255 16.6103 -0.0065
80 12.1262 0.0289 14.2997 0.0126 16.6038 -0.0031
160 12.1551 0.0143 14.3123 0.0064 16.6007 -0.0015
320 12.1694 0.0071 14.3187 0.0031 16.5992 -0.0008
640 12.1765 0.0036 14.3218 0.0016 16.5984 -0.0004
1280 12.1801 0.0018 14.3234 0.0008 16.5980 -0.0002
2560 12.1819 14.3242 16.5978
†N is the number of time steps used in calculation. Diff is referring to the difference in price
calculated using various numbers of time steps.
gence pattern is similar to that of the two-regime case. That is, the convergence
rate is still order 1 even for the three-regime case. The convergence property is
very useful as it can help us approximate the price of vanilla options even with a
small number of time steps.
2.4 Alternative Models
Several amendments can be made to improve the rate of convergence or adapt-
ability of the model under other situations. In the last section, it is assumed that
the generator of the Markov process is a constant matrix and the volatilities of
different regimes do not greatly deviate from the others. These two constraints
27
can be relaxed in some situations.
The generator process can be a function of time. If it is continuous, an
approximation can be used in the branches of each time point, for example, at the
branches at time t to t+∆, the transition probability matrix can be approximated
by the following equation,
P (t,∆) =
pt,11(∆) · · · pt,1k(∆)
.... . .
...
pt,k1(∆) · · · pt,kk(∆)
≈ eA(t)∆. (2.14)
The value of the options found by the lattice still converges to the value of the
options under a continuous-time model. Apart from using I +∑∞
l=1(∆)lA(t)l/l!
to approximate the value of transition probability matrix, another expression can
also be used,
P (t,∆) ≈ limn→∞
(I +A(t)∆
n)n = lim
n→∞(I +
A(t)∆
2n)2n
. (2.15)
This expression has also a good performance in approximating the value of P (t,∆)
using recursion in computer. It is important because the transition probability
matrix has to be calculated for each time step. A good approximation method
can greatly improve the efficiency of computation.
When the number of regime states is large, the volatilities of the asset in
different regimes might not be close to each other. The lattice in the last section
is constructed by a value, σ, which is larger than the asset’s volatilities in all
28
regimes, so that all regimes can be incorporated into this recombining lattice.
This simplifies calculations. However, when the volatilities in different regimes
largely deviate from one another, volatilities are relatively small in some regimes.
But since the model still has to accommodate the largest σi, the σ used in the
model is large. For those regimes with small volatilities, due to the up and down
ratios used in the tree are large, a high risk neutral probability has to be assigned
to the middle branch. The initial error of these regimes is relatively larger. A
recombining trinomial tree can be used to solve the problem.
When we confront a number of regimes corresponding to quite different volatil-
ities, we can divide the regimes into groups according to their size of volatility.
The regimes with large volatility are grouped together, and so are the regimes
with small volatility. The trinomial model can be applied to each group with
regimes whose volatilities are close to each other. The trinomial lattices are then
combined to form a multi-branch lattice, which is similar to the model suggested
by Kamrad and Ritchken (1991) in the (2k + 1)-branch model. More branches
can be introduced to handle more complex situations in the market. All of them
share the same middle branch. The problem is that the parameters σ in differ-
ent trinomial lattices do not necessarily match. When the lattices are combined,
the branches in each of the lattices need not meet each other, that is, the ratios
used in one lattice are not multiples of the other lattices and the simplicity of
29
the model disappears because the branches do not recombine in the whole lattice
efficiently and the number of nodes in the tree is very large.
In order to preserve the simplicity of the model and improve the rate of con-
vergence for the low-volatility regimes at the same time, a similar idea used in
the lattices by Bollen (1998) can be adopted. All regimes are divided into two
groups. In fact, they can be separated into more than two groups, but for pur-
poses of illustration, we only use two groups here. Again, the σ used in trinomial
lattice by the group with larger volatility is not necessarily a multiple of the σ
used by the other group. The problem can be solved by adjusting the value of
σ in either group or even both of the groups, depending on the situation. The
volatility of the group with large volatility should be at least double that of the
small volatility group; otherwise the multi-state trinomial tree in the previous
section should be good enough for pricing. If the ratio between the two values is
larger than 2, the values of lattice parameters σ in both groups are adjusted so
that their ratio is set to 2. In practice, the ratio should not be very large. This
model should be able to handle real data.
Similar to the model that we propose in the last section, assume that there are
k regimes and they are divided into two groups, k1 of them in the low volatility
group and k2 of them in the high volatility group. The states of economy are
30
arranged in ascending order of volatility and so
σ1 ≤ σ2 ≤ . . . ≤ σk1 ≤ . . . ≤ σk.
We now construct the combined trinomial tree in which the stock can increase
with factors e2σ√
∆ and eσ√
∆, remain unchanged, or decrease with factors e−σ√
∆
and e−2σ√
∆. At time step t, there are 4t + 1 nodes in the lattice, the node is
counted from the lowest stock price level, and St,n denotes the stock price of
the nth node at time step t. Each of the nodes has k possible derivative prices
corresponding to the regime states. Let Vt,n,j be the value of the derivative at the
nth node at time step t in the jth regime state. The regimes of group 1 use the
middle three branches with ratios eσ√
∆, 1, and e−σ√
∆. The regimes of group 2
use the branches with ratios e2σ√
∆, 1, and e−2σ√
∆.
We have to ensure that the combined trinomial tree can accommodate all
regimes so that the risk neutral probabilities of all regimes exist. That is
σ > max1≤i≤k1
σi and 2σ > maxk1+1≤i≤k
σi. (2.16)
For the regime i, πiu, πim, and πid are the risk neutral probabilities for up,
middle and bottom branches of the tree, respectively. Similar to the trinomial
tree model, the following set of equations can be obtained. For 1 ≤ i ≤ k1,
πiueσ√
∆ + πim + πide−σ√
∆ = eri∆, (2.17)
(πiu + πid)σ2∆ = σ2
i ∆; (2.18)
31
for k1 + 1 ≤ i ≤ k,
πiue2σ√
∆ + πim + πide−2σ√
∆ = eri∆, (2.19)
(πiu + πid)(2σ)2∆ = σ2i ∆. (2.20)
Solving the equations above, we have,
for 1 ≤ i ≤ k1,
πim = 1− σ2i
σ2, (2.21)
πiu =eri∆ − e−σ
√∆ − πim(1− e−σ
√∆)
eσ√
∆ − e−σ√
∆, (2.22)
πid =eσ√
∆ − eri∆ − πim(eσ√
∆ − 1)
eσ√
∆ − e−σ√
∆, (2.23)
and for k1 + 1 ≤ i ≤ k,
πim = 1− σ2i
4σ2, (2.24)
πiu =eri∆ − e−2σ
√∆ − πim(1− e−2σ
√∆)
e2σ√
∆ − e−2σ√
∆, (2.25)
πid =e2σ√
∆ − eri∆ − πim(e2σ√
∆ − 1)
e2σ√
∆ − e−2σ√
∆. (2.26)
With the payoff of a derivative in different regimes at expiration, the price of the
derivative at different regimes can be found using the following two equations
recursively.
For 1 ≤ i ≤ k1,
Vt,n,i = e−ri∆
[k∑j=1
pij(πiuVt+1,n+3,j + πimVt+1,n+2,j + πidVt+1,n+1,j)
], (2.27)
32
For k1 + 1 ≤ i ≤ k,
Vt,n,i = e−ri∆
[k∑j=1
pij(πiuVt+1,n+4,j + πimVt+1,n+2,j + πidVt+1,n,j)
]. (2.28)
A simple example is given here to illustrate the idea. We assume that there are
three regimes in the market. The corresponding volatilities and risk-free interest
rates in these regimes are 0.15, 0.40, 0.45 and 4%, 6%, 8%, respectively. The
generator matrix of the regime-switching process is−1 0.5 0.5
0.5 −1 0.5
0.5 0.5 −1
. (2.29)
Under the trinomial model in Section 2, the suggested value of σ is 0.524915 and
the risk neutral probabilities of Regime 1 under the up, middle and down state
with 20 time steps used are 0.0469448, 0.918341, 0.0347143, respectively. The
convergence rate of the price of derivatives in this regime is affected due to the
volatility difference. If the three regimes are divided into two groups, Regime 1
forms the low volatility group and Regimes 2 and 3 form the high volatility group.
By (2.9), the corresponding σ value in each of the trinomial trees is suggested to
be
σ(1) = 0.15 + (√
1.5− 1)0.15 = 0.183712,
σ(2) = 0.45 + (√
1.5− 1)(0.40 + 0.45)/2 = 0.545517.
33
Table 2.11: Pricing European call option under model with three regimes using
trinomial tree
European Call Option
Regime 1 Regime 2 Regime 3
N Price Diff Price Diff Price Diff
20 11.9872 0.2082 17.5029 0.0167 19.0695 -0.0226
40 12.1954 0.0966 17.5196 0.0088 19.0469 -0.0103
80 12.2920 0.0468 17.5284 0.0045 19.0366 -0.0050
160 12.3388 0.0232 17.5329 0.0022 19.0316 -0.0024
320 12.3620 0.0114 17.5351 0.0012 19.0292 -0.0012
640 12.3734 0.0058 17.5363 0.0006 19.0280 -0.0006
1280 12.3792 0.0028 17.5369 0.0003 19.0274 -0.0003
2560 12.3820 17.5372 19.0271
Note that σ(2) is about three times of σ(1). In order to make it adaptive to the
combined trinomial tree model, we must make adjustments to their values. For
example, we can take σ(1) to be 0.272758, which is half of σ(2). That is, the
value of σ used by group 1 is 0.272758. The risk neutral probabilities with 20
time steps for Regime 1 in the combined tree are 0.163008, 0.697569, 0.139423.
Tables 2.11 and 2.12 present the price of the European call option using the
trinomial tree and the combined trinomial tree. The pricing error in the combined
trinomial tree for Regime 1 in which the stock has a small volatility is smaller than
that in the trinomial tree. For the combined tree, the approximation errors of the
three regimes are closer to each other compared with those of the trinomial tree,
which is consistent with the result of Boyle (1998). However, we note that if N
34
Table 2.12: Pricing European call option under model with three regimes using
combined trinomial tree
European Call Option
Regime 1 Regime 2 Regime 3
N Price Diff Price Diff Price Diff
20 12.2024 0.0928 17.5325 0.0032 19.0964 -0.0346
40 12.2952 0.0452 17.5357 0.0010 19.0618 -0.0475
80 12.3404 0.0223 17.5367 0.0004 19.0443 -0.0087
160 12.3627 0.0111 17.5371 0.0002 19.0356 -0.0044
320 12.3738 0.0056 17.5373 0.0001 19.0312 -0.0022
640 12.3794 0.0027 17.5374 0.0000 19.0290 -0.0011
1280 12.3821 0.0014 17.5374 0.0000 19.0279 -0.0006
2560 12.3835 17.0574 19.0273
time steps are used, the number of nodes of the combined tree is (2N+1)(N+1);
it is about double of that ((N + 1)2) of a trinomial tree; and the pricing error
of the combined trinomial tree in Regime 3 is greater than that of the trinomial
tree. These suggest that the trinomial tree is more effective than the combined
trinomial tree, especially when the diffusion volatilities of different regimes are
comparable. Therefore, in most of the situations, the simple trinomial tree model
should be good enough and there is no need to use this combined trinomial
tree. These also suggests that trinomial tree is in some sense better than the
pentanomial tree of Bollen (1998) even in the two-regime case.
35
2.5 Hedging Risk of Regime Switching
We assume that there is only one risky underlying asset and one risk-free asset.
The market is not complete and the risk neutral probability is not unique. Our
model is different from the jump-diffusion model in which the price of the un-
derlying asset has jumps. There are some works regarding the pricing of options
in incomplete markets and the choice of risk neutral probability measure. For
example, Follmer and Sondermann (1986), Follmer and Schweizer (1991) and
Schweizer (1996) identify an equivalent martingale measure by minimizing the
variance of the hedging loss under the basic measure. In fact, the quadratic
loss of the hedge position can be related to the concept of a quadratic utility
(Boyle and Wang (2001)). Davis (1997) proposes the use of a traditional eco-
nomic approach of pricing, called the marginal rate of substitution, for pricing
options in incomplete markets. He determines a specific pricing measure, and
hence a fair price, of an option by solving a utility maximization problem. An-
other popular method in the literature is by minimizing entropy. Cherny and
Maslov (2003) justify the use of the Esscher transform for option valuation in
a general discrete-time financial model with multiple underlying risky assets by
maximizing exponential utility under the minimal entropy martingale measure.
They also highlight the duality between the exponential utility maximization and
the minimal entropy martingale measure (Frittelli (2000)). We assume that the
36
Markov process is independent of the Brownian motion under the real measure
and the pricing measure. In our model, when the regime changes, the volatility
of the underlying stock changes (and the risk-free rate also changes), the price of
the stock does not jump as the dynamic of the stock price is a continuous pro-
cess. The change in (expected) volatility changes the option price. For different
corresponding volatilities, the option prices are different. This means that the
option price jumps when the regime state changes. We think that the nature of
regime-switching risk is somehow different from that of the market risk. There-
fore, we can use the real transition probability for pricing. That is, we should not
price the regime-switching risk.
From the very basic concept of valuation, we know that in a complete market,
the risk neutral probability is the probability measure which determines the no-
arbitrage price of all assets in the market by taking discounted expectation using
the risk-free interest rate as the discounting rate. The ultimate tool that helps
us in finding the price of derivative assets is still the assumption of no arbitrage
in the market, which is useful in complete, and incomplete markets. As long
as there is no arbitrage, the price of a derivative asset can be anything. If we
want to price a derivative, it is rational to do it by comparing it with other
related securities in the market. As we know, the price of assets in the market is
determined by people, who have different views on the future and have different
37
risk preferences. Securities are traded in the market according to their investment
characteristics, and an equilibrium price is achieved in the market. In our model,
the real transition probability is assumed to be known, but it needs not be the
transition probability that is used by us in valuation. In practice, if the MRSM
is applied as the dynamic of risky assets in the market, the transition probability
matrix is not known and our estimation of this matrix is important. When a new
derivative is traded in the market with a price that the traders think suitable,
people trade this derivative in the market and an equilibrium price is achieved.
However, we can only observe the price of the derivatives in the current regime. If
we do not have the price information of the assets in all regimes, the no-arbitrage
price of the assets found is not unique.
In finance, when the price of a derivative is considered, the required return
of the derivative should be related to the risk involved. However, the measure of
risk and return, the exact relation between risk and return are still not clear. The
capital asset pricing model (CAPM) suggests that the risk premium of the asset is
proportional to its market risk measure β, which is useful and easy to understand
and therefore widely accepted. The price of the stocks is a continuous process in
the MRSM in this Chapter. Stocks of a company can be viewed as parts of its
business, where the business is something that can earn money by selling things
with a higher price than their costs. They generate values by transforming raw
38
materials into a more useful and valuable form. Derivatives are not present in
the market naturally, but introduced by some financial institutions. They are
just a form of betting; its outcome is related to the price of the underlying assets.
The trading of a derivative is a zero sum game. Therefore, when the regime-
switching risk of derivatives is considered, the issuers should not be rewarded
even it seems to bear the market risk. The regimes refer to the market situation,
but regime switching is a diversifiable risk. The price is unfair if either the issuers
or the buyers are rewarded by taking this jump risk. Hence, we suggest that the
original transition probability can be used in pricing in this model.
Under the continuous-time Markov regime-switching model, due to the regime-
switching risk, the market is incomplete. Guo (2001) uses the change-of-state
contracts to complete the market and price the options. In a model of k regimes,
there are k − 1 possible jumps for derivatives and thus there are k − 1 indepen-
dent derivatives at most in the market. Here, independent refers to the linear
independence of the jump sizes of the derivatives. Therefore, we can add k − 1
derivatives into the market and complete the market. The idea of having a risk
neutral transition probability emerges. There are k2 entries in the transition
probability matrix with k(k − 1) degrees of freedom. If all of the derivatives are
independent in terms of their jump sizes, each of the derivatives has k price infor-
mation for the k regimes, therefore k− 1 independent derivatives are required to
39
complete the market. There is a unique risk neutral transition probability that is
used by all the k−1 derivatives for pricing. In fact, all the other derivatives in the
market should also be priced using this unique risk neutral transition probability
to avoid arbitrage. Theoretically, when the price information of k − 1 additional
independent derivatives in all different regimes at each time point are known, the
market is completed and the unique risk neutral transition probability matrix
exists. The risk neutral transition probability matrix is the only matrix process
which is consistent with the price process of all assets. However, it is not easy
to construct the risk neutral transition probability matrix, especially when the
number of regimes is large.
We know that, although the jumps of derivative price correspond to the change
of regimes which indicates the change of market situation, the regime-switching
risk is different from the market risk in nature. If we really want to price the
jump risks of derivatives due to regime switching, regime-switching risk cannot
be eliminated by diversification in the market and hence it is a fundamental risk.
That is, stock prices should jump when the regime switches. Naik (1993) presents
a good and simple model under this framework. The prices of risk due to the
fluctuation of the Brownian motion and the risk of jump due to regime switching
are defined and used to find the risk neutral transition probability matrix. It will
be studied in details next chapter.
40
We consider a new situation. Apart from the bond, the risky asset and its
derivatives, there are some other assets in the market. If the total market value
has a significant change during regime switching, which implies that there ex-
ist some fundamental assets, rather than derivatives, having jump during regime
switching. Regime-switching risk is a market risk as it cannot be eliminated by
diversification and then it is appropriate to price this risk even if the risky under-
lying asset that we are now considering does not jump during regime switching.
Guo (2001) considers the jump risk premium in a continuous two-state regime-
switching asset price model and identifies a risk neutral measure by changing
the regime transition rates. Siu, Yang and Lau (2008) uses Esscher transform
to obtain the risk neutral measure in which the transition rates change. Their
results are consistent with pricing method used in jump-diffusion model in Naik
(1993) and Yuen and Yang (2009), where changing the transition rates is the
key to obtain the risk neutral measure. Again, our goal is to identify a pricing
transition probability.
Under the current asset price model, we suggest that if the real transition
probability is given, it can be used to price the first and all the other derivatives
of the asset; however, if the prices of the derivatives are already available in
the market, we should try to price the newly developed one using a transition
probability which is consistent with the current prices of all the assets. The real
41
transition probability would no longer be the one used in pricing but the risk
neutral transition probability would take its role and this is parallel to the idea
of risk neutral measure in the Black-Scholes-Merton model.
2.6 Conclusions
MRSM is gaining its popularity in the area of derivative valuation. However, the
difficulties in pricing and hedging derivatives in MRSM limit its development. In
this chapter, we introduced the multi-state trinomial tree lattice. Option pricing
under the MRSM using this lattice is now as simple as that in the Black-Scholes
model using binomial tree. The options which can be priced using the CRR model
under the simple Black-Scholes framework can also be priced using the trinomial
tree under the MRSM of Buffington and Elliott (2002).
The nature of jump risk has been discussed in details. The market is not
complete in the MRSM. As long as there is no arbitrage, jump risk (regime-
switching risk) is suggested not to be priced because the regime-switching process
is independent of the Brownian motion under the real and pricing measures and
jump risk during regime switching is not a kind of market risk in this asset price
model. If the real (market) transition probability is given, it can be used directly
to find an appropriate option price. If prices of derivatives in the market are
available, these prices can be used in order to determine a risk neutral transition
42
probability and the corresponding price of jump risk.
43
Chapter 3
Pricing Regime-switching Risk
3.1 Introduction
The regime-switching market is not complete. Therefore, many researches have
been done in order to find a fair price of the derivatives. Based on the result
of the last chapter, if all the underlying assets in the market have a continuous
price dynamic, regime-switching risk can be diversified and is not a systematic
risk. The real transition probability can be used in pricing derivatives. In general,
underlying assets and the market can have jump in value during regime switching.
A jump diffusion type MRSM is required to illustrate the dynamic of these assets’
price.
Naik (1993) considers this problem in depth and develops a two-state stock
price model. A specific risk neutral transition probability and hence a fair price
of each derivative can be found if the prices of jump risk under different states are
known. In this chapter, this idea is extended to a k-state market. The multi-state
trinomial tree is found to be useful in this jump diffusion model.
44
3.2 Jump Diffusion Model
The model in this section is based on the work of Naik (1993). We make use of
the same notation and similar assumptions as in Chapter 1, but the stock price
process jumps during regime switching.
We consider the real-world probability space (Ω,F , P ). Let T be the time
interval [0, T ] that is being considered. W (t)t∈T is a standard Brownian motion
on (Ω,F , P ). X(t)t∈T is a continuous-time Markov Chain with finite state
space e1, e2, . . . , ek, where ei = (0, . . . , 1, . . . , 0) ∈ Rk, which is a unit vector
representing the economic condition and is independent of the Brownian motion
under P . For simplicity, the state ei is called the state i.
Let A(t) = [aij(t)]i,j=1,...,k be the generator of the Markov chain. By the
semi-martingale representation theorem, we have
X(t) = X(0) +
∫ t
0
X(s)A(s)ds+M(t), (3.1)
where M(t)t∈T is a Rk-valued martingale with respect to the P -augmentation
of the natural filtration generated by X(t)t∈T . Point processes N(t; j)t∈T ,j∈K
count the number of state transitions to state j over time 0 to time t. Therefore,
if we assume that X(t−) = i, and i is not equal to j, the arrival rate of the point
process N(t; j) is just the corresponding entry aij(t) of the generator matrix.
Obviously, when X(t−) = j, the arrival rate of N(t; j) is zero. A new matrix is
45
set up so that the arrival rate of jump can be summarized in the following matrix,
A(t) =
0 a12(t) · · · a1k(t)
a21(t) 0 · · · a2k(t)
......
. . ....
ak1(t) ak2(t) · · · 0
= A(t)− diag(a11(t), a22(t), · · · , akk(t)), (3.2)
where diag(a11(t), a22(t), · · · , akk(t)) represents the diagonal matrix with elements
given by the vector (a11(t), a22(t), · · · , akk(t)). If the generator matrix A(t) is
assumed to be constant, the matrix of arrival rate A(t) would also be constant
and it is denoted by A = [aij]k×k.
We assume that the risk-free interest rate depends on the current state of the
economy only and therefore
r(t) := r(X(t)) = 〈r,X(t)〉 (3.3)
where r := (r1, r2, . . . , rk); ri > 0 for all i = 1, 2, . . . , k and 〈·, ·〉 denotes the inner
product in Rk.
Given the interest rate process, the bond price process B(t)t∈T satisfies the
equations,
dB(t) = r(t)B(t)dt, B(0) = 1. (3.4)
The rate of return and the volatility of the stock price process are denoted
by µ(t,X(t))t∈T and σ(t,X(t))t∈T , respectively. Similar to the interest rate
46
process, they are affected by the state of economy only,
µ(t) := µ(X(t)) = 〈µ,X(t)〉, σ(t) := σ(X(t)) = 〈σ,X(t)〉, (3.5)
where µ := (µ1, µ2, . . . , µk) and σ := (σ1, σ2, . . . , σk) with σi > 0 for all i =
1, 2, . . . , k.
The stock price process is assumed to jump during the transition of states
to ensure that the risk of state transition is a non-diversifiable risk. We assume
that the jump size depends on the state before and after the state change and
the current stock price only. If exp(yij)−1 denotes the ratio of jump of the stock
price during the state transition from i to j to the current stock price, then, the
stock price process S(t) is assumed to satisfy
dS(t)
S(t−)= µ(t−)dt+ σ(t−)dW (t)
+k∑j=1
(exp(yX(t−),j)− 1)(N(dt; j)− aX(t−),jdt). (3.6)
We denote N(t; j)−N(t−; j) by N(dt; j) and we find that yiji,j∈K has an impor-
tant property which makes the trinomial tree method applicable to price deriva-
tives in this model, that is,
yil + ylj = yij for all i, j, l ∈ K. (3.7)
The equation holds because of the Markovian property of X(t). For a Markov
chain, all information about the process in the past are represented by the current
47
information. We consider two situations; one is that the state of economy changes
from state i to state j, and the other is that the economic state changes from
state i to state l and goes to state j immediately. All of the other conditions in
these two situations are assumed to be the same. Due to the Markovian property,
the past information is not useful given the current state information. A person
cannot distinguish the difference in value of these two stock price processes under
these two situations. The prices of stocks in these two cases are the same and
equation (3.7) should hold. If j and l in equation (3.7) are taken to be i,
yii + yii = yii ⇒ yii = 0, (3.8)
for all i ∈ K. We should also have the following condition,
yij + yji = yii = 0 ⇒ yij = −yji, (3.9)
for all i, j ∈ K.
Here, we base on the works of Elliott, Chan and Siu (2005) on the double-
indexed σ-algebra. Let FXt t∈T and FZt t∈T be the the natural filtration of
X(t)t∈T and Z(t)t∈T , respectively. We define Gt to be the σ-algebra FXt ∨FZt
and Gt,s to be the double indexed σ-algebra FXt ∨FZs . A risk neutral probability
measure can be obtained so that the price of the derivatives can be calculated
easily. Let Q denote the risk neutral probability measure. We write ξ(T ) =
dQ/dP , ξ(t) = E(ξ(T )|Gt). For a derivative of the asset S(t) with a final payoff
48
of g(S(T )), its price at time t can be calculated by
V g(t) = EQ
[exp
(−∫ T
t
r(s)ds
)g(S(T ))
∣∣∣Gt] (3.10)
= E
[exp
(−∫ T
t
r(s)ds
)ξ(T )
ξ(t)g(S(T ))
∣∣∣Gt] , (3.11)
where exp(−∫ t
0r(s)ds
)ξ(t) is known as the state-price density. Because ξ(t) =
E(ξ(T )|Gt), ξ(t) is a martingale under the probability measure P . By the
martingale representation theorem, we have
dξ(t)
ξ(t−)= η0(t−)dW (t) +
k∑j=1
η(t−; j)(N(dt; j)− aX(t−),jdt) (3.12)
for some predictable processes η0(t−) and η(t−; j), where η(t−; j) > −1. Ac-
cording to Naik (1993), η0(t−) and η(t−; j) can be interpreted as the price of
continuous risk generated by the Brownian motion and the price of discontinuous
risk of the change in volatility, respectively. We assume that the risk characteris-
tic of the investors depends on the economic state only. Then, η0(t−) depends on
X(t−), that is, η0(t−) = η0(X(t−)); and the value of η(t−; j) depends on X(t−)
and j only, that is, η(t−; j) = η(X(t−), j). We write η(t−; j) as ηX(t−),j. The pro-
cess
exp(−∫ t
0r(s)ds
)ξ(t)S(t)
is a martingale under the probability measure
49
P . By Ito’s formula, we have
d[S(t)ξ(t)] = S(t−)dξ(t) + ξ(t−)dS(t) + 〈dS(t), dξ(t)〉
= S(t−)ξ(t−)
[µ(t−)dt+ σ(t−)dW (t) + η0(t−)dW (t)
+k∑j=1
(exp(yX(t−),j)− 1
)(N(dt; j)− aX(t−),jdt
)
+k∑j=1
η(t−; j)(N(dt; j)− aX(t−),jdt
)+ σ(t−)η0(t−)dt
+k∑j=1
(exp(yX(t−),j)− 1
)η(t−; j)N(dt; j)
]
= S(t−)ξ(t−)
[µ(t−)dt+ σ(t−)dW (t) + η0(t−)dW (t)
+k∑j=1
(exp(yX(t−),j)− 1
)(N(dt; j)− aX(t−),jdt
)
+k∑j=1
η(t−; j)(N(dt; j)− aX(t−),jdt
)+ σ(t−)η0(t−)dt
+k∑j=1
(exp(yX(t−),j)− 1
)η(t−; j)
(N(dt; j)− aX(t−),jdt
)
+k∑j=1
(exp(yX(t−),j)− 1
)η(t−; j)aX(t−),jdt
]. (3.13)
At the same time, r(t) is bounded and is continuous almost everywhere on T ,
thus we have
∫ t
0
r(s)ds =
∫ t
0
r(s−)ds, (3.14)
and therefore,
exp(−∫ t
0r(s−)ds
)ξ(t)S(t)
is also a martingale. By Ito’s for-
50
mula again,
d
[exp
(−∫ t
0
r(s−)ds)S(t)ξ(t)
]= exp
(−∫ t
0
r(s−)ds)S(t−)ξ(t−)
[µ(t−)dt+ σ(t−)dW (t)
+k∑j=1
(exp(yX(t−),j)− 1
)(N(dt; j)− aX(t−),jdt
)− r(t−)dt
+k∑j=1
η(t−; j)(N(dt; j)− aX(t−),jdt
)+ η0(t−)dW (t)
+k∑j=1
(exp(yX(t−),j)− 1
)η(t−; j)
(N(dt; j)− aX(t−),jdt
)
+k∑j=1
(exp(yX(t−),j)− 1
)η(t−; j)aX(t−),jdt+ σ(t−)η0(t−)dt
]. (3.15)
Apart from the time point of jump, we have
µ(t−)− r(t−) + σ(t−)η0(t−) +k∑j=1
(exp(yX(t−),j)− 1
)η(t−; j)aX(t−),j = 0. (3.16)
We rearrange the equation above and we have
µ(t−)− r(t−) = −σ(t−)η0(t−)−k∑j=1
(exp(yX(t−),j)− 1
)aX(t−),jη(t−; j). (3.17)
The risk premium of the stock has two components: the risk premium from the
Brownian motion and the risk premium from the jump of stock price. From the
capital asset pricing model under a pure diffusion formulation, the risk premium
is proportional to the volatility of the stock price. Here, the risk premium of a
jump due to a transition of the Markov chain is proportional to the size of jump
51
exp(yX(t−),j) − 1 and the arrival rate of jump aX(t−),j. As mentioned in Naik
(1993), −η0(t−) and −η(X(t−), j) represent the price of risk of diffusion volatility
and jump, respectively.
−η0(t−) =Risk Premium from Brownian Motion Risk
σ(t−)(3.18)
−η(X(t−), j) =Risk Premium from State j Jump Risk
(exp(yX(t−),j)− 1)aX(t−),j
. (3.19)
To ensure that the risk premiums of these two sources of risk are non-negative,
we have
η0(t−) ≤ 0 and (3.20)
η(X(t−), j)(exp(yX(t−),j)− 1) ≤ 0 for all X(t−), j ∈ K. (3.21)
Since exp(yX(t−),j)− 1 and yX(t−),j have the same sign, the second condition can
be simplified,
η(X(t−), j)yX(t−),j ≤ 0 for all X(t−), j ∈ K. (3.22)
From (3.6) and (3.12) and Ito’s formula, it is not difficult to see that the
52
expressions of S(T ) and ξ(T ) are given by
S(T ) = S(0)exp
[ ∫ T
0
µ(t−)dt+
∫ T
0
σ(t−)dW (t)− 1
2
∫ T
0
σ2(t−)dt
+k∑j=1
∫ T
0
yX(t−),jN(dt; j)−k∑j=1
∫ T
0
(exp(yX(t−),j)− 1
)aX(t−),jdt
],
(3.23)
ξ(T ) = exp
[ ∫ T
0
η0(t−)dW (t)− 1
2
∫ T
0
η20(t−)dt
+k∑j=1
∫ T
0
ln(1 + ηX(t−),j
)N(dt; j)−
k∑j=1
∫ T
0
aX(t−),jηX(t−),jdt
].(3.24)
Because there are jump processes N(t; j) in the expression, we apply Taylor
expansion to equation (3.23) and (3.24). We can verify that
dS(t) = S(t−)
[µ(t−)dt+ σ(t−)dW (t)− 1
2σ2(t−)dt+
k∑j=1
yX(t−),jN(dt; j)
−k∑j=1
(exp(yX(t−),j)− 1
)aX(t−),jdt+
1
2σ2(t−)dt
+1
2
k∑j=1
y2X(t−),jN(dt; j) +
∞∑l=3
1
l!
k∑j=1
ylX(t−),jN(dt; j)
]
= S(t−)
[µ(t−)dt+ σ(t−)dW (t)
+k∑j=1
(exp(yX(t−),j)− 1
)(N(dt; j)− aX(t−),jdt
)], (3.25)
53
dξ(t) = ξ(t−)
[η0(t−)dW (t)− 1
2η2
0(t−)dt+k∑j=1
ln(1 + ηX(t−),j
)N(dt; j)
−k∑j=1
aX(t−),jηX(t−),jdt+1
2η2
0(t−)dt
+∞∑l=2
1
l!
k∑j=1
[ln(1 + ηX(t−),j
)]lN(dt; j)
]
= ξ(t−)
[η0(t−)dW (t) +
k∑j=1
η(t−; j)(N(dt; j)− aX(t−),jdt
)]. (3.26)
Equations (3.23) and (3.24) satisfy the stochastic differential equations of S(t)
and ξ(t) and they are useful for us to understand the dynamic of stock price and
the change in probability measure.
3.3 Arrival Rates of Jumps under Risk Neutral
Measure
In the geometric Brownian motion model, Girsanov theorem is used to change
the real probability measure to the risk neutral probability measure. Under a
risk neutral probability, the expected rate of return of the risky asset becomes
the risk-free rate. The same idea is used in this section. A risk neutral probability
can be obtained such that the underlying security and all of its derivatives have
the risk-free interest rate as the expected return. The volatility of the diffusion
part of the risky asset under the risk neutral measure is the same as that under
54
the real-world measure, but the arrival rates of jumps change when the measure
changes. The arrival rates of jumps, which are the same as the rates of state
transitions, will be discussed in this section.
To understand the probabilistic properties of the standard Brownian motion
W (t)t∈T and the Markov chain X(t)t∈T under the risk neutral measure, the
stochastic differential equations governing the evolution of these two processes
over time can be considered. ξ(T ) is the Radon Nikodym derivative, then,
d(ξ(t)W (t)) = ξ(t−)dW (t) +W (t)dξ(t) + 〈dW (t), dξ(t)〉
= ξ(t−)dW (t) +W (t)dξ(t) + ξ(t−)η0(t−)dt. (3.27)
W (t) and ξ(t) are (G,P)-martingales, and G := Gtt∈T . Using the above equa-
tion, another process can be studied,
d
[ξ(t)(W (t)−
∫ t
0
η0(s−)ds)
]= ξ(t−)
[dW (t)− η0(t−)dt
]+(W (t)−
∫ t
0
η0(s−)ds)dξ(t) +[
dW (t)− η0(t−)dt]dξ(t)
= ξ(t−)dW (t) +(W (t)−
∫ t
0
η0(s−)ds)dξ(t). (3.28)
Therefore, we know thatξ(t)
(W (t)−
∫ t0η0(s−)ds
)is a (G,P)-martingale.
Obviously, the quadratic variation of (W (t)−∫ t
0η0(s−)ds) is t. By Levy’s theorem,
it is a standard Brownian motion under Q.
55
Similarly, the Markov chain can be studied using stochastic differential equa-
tion, but instead of studying the Markov chain directly, the point processes
N(t; j)t∈T ,j∈K derived from the Markov chain are our focus. If the regime
is at state X(t−) just before time t, the point process N(t; j) has arrival rate
equal to aX(t−),j, then
d
(ξ(t)N(t; j)
)= ξ(t−)N(dt; j) +N(t−; j)dξ(t) +N(dt; j)dξ(t)
= ξ(t−)N(dt; j) +N(t−; j)dξ(t) + ξ(t−)η(t−; j)N(dt; j)
= ξ(t−)(1 + η(t−; j))(N(dt; j)− aX(t−),jdt) +N(t−; j)dξ(t)
+ξ(t−)(1 + η(t−; j))aX(t−),jdt. (3.29)
As X(t−) is a left continuous function of t, N(t−; j) represents the number of
state transitions to state j from time 0 to just before time t. We now consider
56
another process,
d
[ξ(t)
(N(t; j)−
∫ t
0
(1 + η(s−; j))aX(s−),jds)]
= ξ(t−)(N(dt; j)− (1 + η(t−; j))aX(t−),jdt)
+(N(t−; j)−
∫ t
0
(1 + η(s−; j))aX(s−),jds)dξ(t)
+dξ(t)(N(dt; j)− (1 + η(t−; j))aX(t−),jdt)
= ξ(t−)(N(dt; j)− aX(t−),jdt)
+(N(t−; j)−
∫ t
0
(1 + η(s−; j))aX(s−),jds)dξ(t)
−ξ(t−)η(t−; j)aX(t−),jdt+ ξ(t−)η(t−; j)N(dt; j)
= ξ(t−)(1 + η(t−; j))(N(dt; j)− aX(t−),jdt)
+(N(t−; j)−
∫ t
0
(1 + η(s−; j))aX(s−),jds)dξ(t). (3.30)
By the definition of the Radon Nikodym derivative ξ(T ), which is always positive,
the risk neutral probability measure Q is equivalent to the real probability mea-
sure P . N(t; j) is still a jump process under Q. Furthermore, if we divide a time
interval [s, t] into m pieces with mesh Π, and ∆N(tl; j) represents the change of
57
N(t; j) in the corresponding lth small time interval, then
EQ[
limΠ→0
m∑l=0
∆N(tl; j)∆N(tl; j)]
= E[ξ(T ) lim
Π→0
m∑l=0
∆N(tl; j)∆N(tl; j)]
= E[ξ(T ) lim
Π→0
m∑l=0
∆N(tl; j)]
= EQ[
limΠ→0
m∑l=0
∆N(tl; j)]. (3.31)
This shows that N(t; j) is a point process under Q. From stochastic differential
equation (3.30), the arrival rate of N(t; j) is (1 + η(t−; j))aX(t−),j under Q. That
is, under the risk neutral probability measure Q, the arrival rate matrix A∗ is
[a∗ij]k×k =
0 (1 + η12)a12 · · · (1 + η1k)a1k
(1 + η21)a21 0 · · · (1 + η2k)a2k
......
. . ....
(1 + ηk1)ak1 (1 + ηk2)ak2 · · · 0
. (3.32)
and the corresponding generator matrix A∗ is
[a∗ij]k×k =
a∗11 (1 + η12)a12 · · · (1 + η1k)a1k
(1 + η21)a21 a∗22 · · · (1 + η2k)a2k
......
. . ....
(1 + ηk1)ak1 (1 + ηk2)ak2 · · · a∗kk
(3.33)
where
a∗ii = −∑j 6=i
(1 + ηij)aij = −∑j 6=i
(1 + ηij)aij. (3.34)
58
3.4 Trinomial Tree Pricing under Jump Diffu-
sion Model
Based on the result of the last section, we are able to find the price of the deriva-
tives using trinomial tree that developed in Chapter 2. The life of the derivative
T can be divided into N time steps with length ∆. Some values of σ that sat-
isfied equation (2.3), for example, the one that suggested by (2.9), can be used.
eσ√
∆ is used as the ratio in the lattice. Each of the nodes accommodates price
information of k states. qij(∆) denotes the probability of state transition from
state i to state j in a time step ∆ under the risk neutral probability Q, which
can be found by the generator matrix A∗ using the matrix exponential suggested
by (2.10). That is,
[qij(∆)]k×k =
q11(∆) · · · q1k(∆)
.... . .
...
qk1(∆) · · · qkk(∆)
= eA∗∆. (3.35)
With no ambiguity, we write qij(∆) as qij.
For the regime i, πiu, πim and πid are the risk neutral probabilities corresponding
to stock price increases, remains the same and decreases, respectively. Due to the
presence of jumps in stock price, the values of these probabilities change. For
59
each i ∈ K, we have,
(πiue
σ√
∆ + πim + πide−σ√
∆)( k∑
r=1
qijeyij
)= eri∆, (3.36)
(πiu + πid)σ2∆ = σ2
i ∆. (3.37)
Equation (3.36) can be rearranged and expressed as
πiueσ√
∆ + πim + πide−σ√
∆ = exp(ri∆− ln
( k∑r=1
qijeyij
)). (3.38)
If λi is defined as σ/σi for each i, then, λi > 1 and the value of πiu, πim, π
id can
be calculated in terms of λi,
πim = 1− σ2i
σ2= 1− 1
λ2i
, (3.39)
πiu =eri∆−ln(
∑kr=1 qije
yij ) − e−σ√
∆ − (1− 1/λ2i )(1− e−σ
√∆)
eσ√
∆ − e−σ√
∆, (3.40)
πid =eσ√
∆ − eri∆−ln(∑k
r=1 qijeyij ) − (1− 1/λ2
i )(eσ√
∆ − 1)
eσ√
∆ − e−σ√
∆. (3.41)
As all the regimes share the same lattice and the regime state cannot be
reflected by the position of the nodes, each of the nodes has k possible prices
corresponding to the regime state at the node. At time step t, there are 2t + 1
nodes in the lattice, the node is counted from the lowest stock price level, and
St,n,j denotes the stock price of the nth node at time step t under regime j. The
stock prices in different regimes sharing the same node are different to the others
and the difference correspond to the ratio of jump that is given by yij. So, we
60
have
St,n,i = St,n,jeyji for all i, j ∈ K. (3.42)
Given the initial stock price and regime state, the stock prices at each node of the
lattice in different regimes can be found. Let Vt,n,j be the value of the derivative
at the nth node at time step t under the jth regime state. The trinomial model
can price European options, American options and barrier options. If we consider
a European call option at time T , we have
VN,n,i = (SN,n,i −K)+ for all i ∈ K, all n. (3.43)
Now, with the derivative prices in all regimes at expiration, by conditioning
on the economic state and the stock price level after one time step, we can apply
the following equation recursively,
Vt,n,i = e−ri∆
[k∑j=1
qij(πiuVt+1,n+2,j + πimVt+1,n+1,j + πidVt+1,n,j)
], (3.44)
and the price of the option can be obtained.
The price of American option can be calculated using the method suggested
in Chapter 2, by comparing the value of option at different regimes and its value
when it is exercised immediately at each node.
However, there is a problem when we price barrier options. Boyle and Tian
(1998) observes that the price of the barrier option obtained by the lattice model
61
is more accurate when the grid of the lattice touches the barrier level. Under our
jump diffusion model, the prices of stock at different regimes are not necessarily
the same even if they are in the same node. The stock price of all regimes might
not touch the barrier level at grids. We apply the method of Boyle and Tian
(1998) to start the lattice at the barrier level at a particular regime so that the
grid can touch the barrier level at least at one regime. In the two-regime case,
the value of σ can be adjusted so that the grid can touch the barrier level in
the lattice for both regimes. When more regimes are involved, it seems hard to
ensure that the grid touches the barrier level in the lattice at all regimes. It is
even harder for the stock prices at grid to touch both barriers for a double barrier
option in two or more regimes. More time steps in the lattice are required to
reduce the error.
We have to be careful about the difference between the model without jump
and the jump diffusion model under the risk neutral probability. Under the risk
neutral probability measure Q, the stock price process is given by
S(t) = S(0) exp
[ ∫ t
0
r(s−)ds+
∫ t
0
σ(s−)dW (s) +k∑j=1
∫ t
0
yX(s−),jN(ds; j)
−1
2
∫ t
0
σ2(s−)ds−k∑j=1
∫ t
0
(exp(yX(s−),j)− 1
)a∗X(s−),jds
]. (3.45)
The stock price jumps when there is regime switching. Given the same volatility
for the diffusion part, the jump process gives extra variability to the stock price
62
and can result in a higher option price than that in the model with no jump.
The stock prices of different regimes are different even if they share the same
node. When the option prices of different regimes are calculated, apart from
the effects of jumps on volatility, the risk-free interest rate, the intensity and
magnitude of jumps, the price of the underlying assets should also be considered
as they need not be the same at different regimes under this model.
3.5 Numerical Results and Analysis
In this section, we consider some examples and use them to study the properties
of our jump diffusion model. We study the convergence property and the effect
of the regime-switching risk. The value of the European options, the American
options and the down-and-out barrier options are considered. We first focus on
the case when the jump risk is not priced, then the option pricing when jump
risk is priced.
3.5.1 Jump Risks are Not Priced
Similar conditions are used so that the data obtained in this section can be
easily compared with those of the last chapter. We consider the market has two
regimes. The underlying asset is assumed to be a stock with initial price of 100
63
in Regime 1, following a geometric Brownian motion of the two-regime model
with no dividend. In Regime 1, the risk-free interest rate is 4% and the diffusion
volatility of stock is 0.25; in Regime 2, the risk-free interest rate is 6% and the
diffusion volatility of stock is 0.35. All options expire in one year with strike price
equal to 100. The generator for the regime-switching process under P is taken as −0.5 0.5
0.5 −0.5
.
The jump parameters yiji,j∈K are summarized in the following matrix,
Y =
0 y12
y21 0
=
0 .1
−.1 0
(3.46)
From this, we can obtain the price of the stock in all regimes at each node in the
lattice using (3.42). For example, initial stock price in Regime 2 is 100exp(0.1).
The price of jump risk is taken to be zero so we have
η =
0 η12
η21 0
=
0 0
0 0
. (3.47)
With the value of η, under the risk neutral probability measure Q, the arrival
rate matrix is
A∗ =
0 (1 + η12)a12
(1 + η21)a21 0
=
0 0.5
0.5 0
. (3.48)
With all the information given above, we can use the trinomial tree to find the
prices of options.
64
Table 3.1: Pricing European call option when jump risk is not priced
European Call Option
Regime 1, 100 Regime 2, 110.5171
N Price Diff Ratio Price Diff Ratio
20 12.9940 0.0862 0.2575 23.2553 0.0316 -0.6709
40 13.0802 0.0222 0.6532 23.2869 -0.0212 0.1981
80 13.1024 0.0145 0.7586 23.2657 -0.0041 -1.0714
160 13.1169 0.0110 0.2636 23.2615 0.0045 -0.5111
320 13.1279 0.0029 0.9310 23.2660 -0.0023 -0.4348
640 13.1308 0.0027 0.2593 23.2637 0.0010 -0.6000
1280 13.1335 0.0007 0.7143 23.2647 -0.0006 0.0000
2560 13.1342 0.0005 23.2641 0.0000
5120 13.1347 23.2641
†N is the number of time steps used in calculation. Diff is referring to the difference in price
calculated using various numbers of time steps and ratio is the ratio of the difference. The
value next to the regime number is the initial stock price at this regime.
Table 3.2: Pricing European put option when jump risk is not priced
European Put Option
Regime 1, 100 Regime 2, 110.5171
N Price Diff Ratio Price Diff Ratio
20 8.73688 0.07863 0.2336 7.24824 0.03917 -0.4447
40 8.81551 0.01837 0.6892 7.28741 -0.01742 0.1355
80 8.83388 0.01266 0.7915 7.26999 -0.00236 -2.3178
160 8.84654 0.01002 0.2445 7.26763 0.00547 -0.3400
320 8.85656 0.00245 1.0082 7.27310 -0.00186 -0.7043
640 8.85901 0.00247 0.2146 7.27124 0.00131 -0.4122
1280 8.86148 0.00053 0.9623 7.27255 -0.00054 -0.1296
2560 8.86201 0.00051 7.27201 0.00005
5120 8.86252 7.27208
65
In Table 3.1 and Table 3.2, we can see the convergence patterns, they are not
as smooth as those in the non-jump model shown in Table 2.2 and Table 2.3.
The jumped stock price during the state transition makes the situation more
complicated.
In the following, we will investigate the effect of the stock price jump by
comparing the numerical results for the models with and without jumps. In the
non-jump model, we assume that the risk-free rate and the diffusion volatility at
each regime are the same as that in the model with jump, but the dynamic of
the stock price process is a regime-switching geometric Browian motion without
jump. We also assume the same generator for both regime-switching processes.
Since we have different initial prices for the stock at different regimes in the
model with jumps, we use different initial prices at different regimes for the
model without jump to compare the prices of options in these two models. The
results are given in Tables 3.3 and 3.4. For the same initial stock price, diffusion
volatility, risk-free interest rate and keeping all the other assumptions, the jump
model gives a higher price. This is due to the jump. We can observe that the
convergence behavior in the jump diffusion model is more complex that that in
the continuous model. We can also observe that, at both regimes, the differences
between the call option price and put option price in the two models converge
to the same value. This is because of the put-call parity. This is shown in the
66
Table 3.3: Comparison of European call option prices in jump and non jump
models
European Call Options
Regime 1, 100 Regime 2, 110.5171
N Jump Non Jump Diff Jump Non Jump Diff
20 12.9940 12.6282 0.3658 23.2553 23.0144 0.2409
40 13.0802 12.6936 0.3688 23.2869 23.0464 0.2405
80 13.1024 12.7260 0.3764 23.2657 23.0033 0.2624
160 13.1169 12.7422 0.3747 23.2615 22.9917 0.2698
320 13.1279 12.7502 0.3776 23.2660 22.9965 0.2695
640 13.1308 12.7503 0.3765 23.2637 22.9915 0.2722
1280 13.1335 12.7563 0.3772 23.2647 22.9926 0.2721
2560 13.1342 12.7573 0.3769 23.2641 22.9913 0.2728
5120 13.1347 12.7578 0.3769 23.2641 22.9911 0.2730
†Jump is referring to the jump diffusion model and Non Jump is referring to the original
model. Diff is the difference of option prices between these two models.
Table 3.4: Comparison of European put option prices in jump and non jump
models
European Put Options
Regime 1, 100 Regime 2, 110.5171
N Jump Non Jump Diff Jump Non Jump Diff
20 8.73688 8.37107 0.36581 7.24824 7.00727 0.24097
40 8.81551 8.42888 0.38663 7.28741 7.04691 0.24050
80 8.83388 8.45755 0.37633 7.26999 7.00756 0.26243
160 8.84654 8.47182 0.37472 7.26763 6.99784 0.26979
320 8.85656 8.47894 0.37762 7.27310 7.00357 0.26953
640 8.85901 8.48250 0.37651 7.27124 6.99910 0.27214
1280 8.86148 8.48428 0.37720 7.27255 7.00044 0.27211
2560 8.86201 8.48517 0.37684 7.27201 6.99920 0.27281
5120 8.86252 8.48561 0.37691 7.27208 6.99912 0.27296
67
following. The prices of risk from Brownian motion and from Markov chain are
fixed. The risk neutral probability measure and thus the corresponding prices of
the derivatives can be found. We consider the situation that the stock pays no
dividend. Let c and p be the prices of the European call and the put option,
respectively. In the two models, by the put-call parity, for any given regimes, we
have,
c− p = S0 −KEQ[exp(−
∫ T
0
r(s)ds)]. (3.49)
The generator matrices and thus the arrival rate matrices are the same for both
models as the jump risk is not priced. The expected discounting factor for the
exercise price K shown in equation (3.49) is the same for the two models. The
result indicates that when the call option has a higher price due to the extra
variability from the jump of stock, the put option price increases by the same
amount if regime-switching risk is not priced.
If we substitute the values of the European call option and the European put
option of the two models obtained by iterations to equation (3.49), the value of
the expected discounted factor in the equation can also be obtained at the two
regimes. The value of the expected discounted factor can also be obtained by
the generator matrix. The probability of the regime changing from state i to
state j after a period of time t is given by the ij entry of etA∗. The generator
matrix under Q is the same as the one under the probability measure P because
68
the jump risk is not priced. The expected discounted factor can be obtained by
considering the expected amount of time that the Markov chain spends on each
regime. If A∗ was non-singular matrix, we have,∫ T
0
exA∗dx = (A∗)−1
(eTA
∗ − I)
if det A∗ 6= 0. (3.50)
However, by equation (3.34), let 1k and 0k be the column vector of size k with
all entries equal to 1 and 0, respectively. We know that,
A∗(1k) = 0(1k) = 0k. (3.51)
Therefore, 0 is an eigenvalue of A∗ and thus A∗ is a singular matrix. Equation
(3.50) cannot be used to find the expected amount of time that the Markov chain
spends on each regime. Given we are in state i, the expected amount of time that
the Markov chain is in state j over the period of length T is defined as τij. From
the values obtained using 5120 time steps, at each regime, we can calculate the
fixed interest rate that is equivalent to the stochastic interest rate being used.
Each fixed interest rate, in turn, reflects the expected amount of time that the
market spends on the corresponding regime, given the current state.
For state 1 : r1 =−1
1ln
(100− 13.1347 + 8.86252
100
)= 4.37%
⇒ Expected proportion of time in Regime 1 =6− 4.37
6− 4= 81.5%.
For state 2 : r2 =−1
1ln
(100e1.1 − 23.2641 + 7.27208
100
)= 5.63%
⇒ Expected proportion of time in Regime 2 =6− 5.63
6− 4= 18.5%.
69
Table 3.5: Comparison of American call option prices in jump and non jump
models
American Call Options
Regime 1, 100 Regime 2, 110.5171
N Jump Non Jump Diff Jump Non Jump Diff
20 12.9940 12.6282 0.3658 23.2553 23.0144 0.2409
40 13.0802 12.6936 0.3688 23.2869 23.0464 0.2405
80 13.1024 12.7260 0.3764 23.2657 23.0033 0.2624
160 13.1169 12.7422 0.3747 23.2615 22.9917 0.2698
320 13.1279 12.7502 0.3776 23.2660 22.9965 0.2695
640 13.1308 12.7503 0.3765 23.2637 22.9915 0.2722
1280 13.1335 12.7563 0.3772 23.2647 22.9926 0.2721
2560 13.1342 12.7573 0.3769 23.2641 22.9913 0.2728
5120 13.1347 12.7578 0.3769 23.2641 22.9911 0.2730
The expected proportion of time that the Markov chain stays in each state ob-
tained here is very close to the values obtained by numerical estimation, which is
given by
∫ 1
0
exp
−0.5x 0.5x
0.5x −0.5x
dx ≈
0.816060 0.183940
0.183940 0.816060
. (3.52)
The approximation errors of equation (3.52) should be much smaller than that by
using the information of option prices. For the continuous model, similar results
can be found in Elliott, Chan and Siu (2005).
Tables 3.5 and 3.6 show the American call and put results. From the tables, we
can observe that the prices of the American call option found by the lattice model
70
Table 3.6: Comparison of American put option prices in jump and non jump
models
American Put Options
Regime 1, 100 Regime 2, 110.5171
N Jump Non Jump Diff Jump Non Jump Diff
20 9.12138 8.80315 0.31823 7.59267 7.36070 0.23197
40 9.19402 8.85551 0.33851 7.62430 7.39476 0.22954
80 9.21489 8.88225 0.33264 7.60882 7.36040 0.24842
160 9.22818 8.89525 0.33293 7.60721 7.35298 0.25423
320 9.23700 8.90125 0.33542 7.61088 7.35672 0.25416
640 9.23974 8.90471 0.33503 7.60932 7.35306 0.25626
1280 9.24192 8.90627 0.33565 7.61021 7.35395 0.25626
2560 9.24254 8.90704 0.33550 7.60970 7.35286 0.25684
5120 9.24298 8.90742 0.33556 7.60971 7.35275 0.25696
shown in Table 3.5 are exactly the same as the European call option prices. This
is consistent with the theoretical result that the American call option should not
be exercised before expiration. If we use C to denote the price of the American
call, for any given regimes, we would have the following inequality at time t before
expiration,
C ≥ c ≥ St −KEQ[
exp(−∫ T
t
r(s)ds)]≥ St −K. (3.53)
The American call is more valuable than the European call because of its early
exercise option. By equation (3.49), we use the fact that European put option
should have a non-negative value, we have the second inequality. The risk-free
interest rate under all regimes should be positive and thus the expected discounted
71
Table 3.7: Comparison of down-and-out barrier call option prices in jump and
non jump models
Down-and-out Call Options
Regime 1, 100 Regime 2, 110.5171
N Jump Non Jump Diff Jump Non Jump Diff
20 9.19243 8.97860 0.21383 19.8270 19.5065 0.3205
40 9.59712 8.96621 0.63091 20.3482 19.4800 0.8682
80 9.20153 8.97478 0.22675 19.7730 19.4623 0.3107
160 9.11647 8.97064 0.14583 19.6408 19.4521 0.1887
320 9.20268 8.97010 0.23258 19.7587 19.4489 0.3098
640 9.11665 8.97025 0.14640 19.6336 19.4468 0.1868
1280 9.20246 8.96928 0.23318 19.7546 19.4452 0.3094
2560 9.11626 8.96962 0.14664 19.6316 19.4451 0.1862
5120 9.11352 8.96955 0.14397 19.6272 19.4448 0.1824
factor should also be less than 1 unless t = T . Therefore, we obtain the third
inequality. The value St−K is just the value of the American call if it is exercised
immediately with a positive value. The price of the American call can be equal
to its immediately exercise price only if t = T . In all the other situations, the
price of the American call option is greater than its immediately exercise value.
Early exercise is always not optimal for American call option.
The price of the American put option is higher than that of the European
put option and the option price in the jump model is higher than that in the
continuous model. Both observations are consistent with our intuition and the
theoretical results in finance.
72
In Table 3.7, we can find that the convergence speed of the down-and-out
barrier call option is not stable, compared with the data of non-jump model,
even we have chosen an optimal σ value to ensure that the stock prices touch the
barrier level in both regimes, especially when N is small.
3.5.2 Jump Risks are Priced
We consider the option pricing when the jump risk is priced in this subsection.
We assume that all the set-up and conditions are the same as those in the last
subsection, except that the price of the jump risk is not 0 here. The price of the
jump risk is taken as
η =
0 η12
η21 0
=
0 −0.1
0.1 0
. (3.54)
With the value of η, under risk neutral probability measure Q, the arrival rate
matrix A∗ is
[a∗ij] =
0 (1 + η12)a12
(1 + η21)a21 0
=
0 0.45
0.55 0
. (3.55)
We can apply the trinomial tree method and find the price of the options. The
values are compared with the values of which the jump risk is not priced as shown
in the last subsection.
When jump risk is priced, we might expect that the prices of the options is
73
Table 3.8: Comparison of European call option prices with priced and non-priced
jump risk
European Call Options
Regime 1, 100 Regime 2, 110.5171
N Priced Non Priced Diff Priced Non Priced Diff
20 12.9940 12.8789 -0.1151 23.2553 23.1855 -0.0698
40 13.0802 12.9619 -0.1183 23.2869 23.2153 -0.0716
80 13.1024 12.9845 -0.1179 23.2657 23.1951 -0.0706
160 13.1169 12.9990 -0.1179 23.2615 22.1911 -0.0704
320 13.1279 13.0095 -0.1184 23.2660 22.1953 -0.0707
640 13.1308 13.0125 -0.1183 23.2637 22.1931 -0.0706
1280 13.1335 13.0151 -0.1184 23.2647 22.1941 -0.0706
2560 13.1342 13.0158 -0.1184 23.2641 22.1935 -0.0706
5120 13.1347 13.0163 -0.1184 23.2641 22.1935 -0.0706
†Priced is referring to the priced jump risk prices and Non Priced is referring to the prices
that jump risk is not priced. Diff is the option price of priced model minus non priced model.
greater. However, in Table 3.8 to Table 3.12, we calculate the prices of various
options for the two situations, the prices of all types of options are lower if jump
risk is being priced. The reason is that, under the risk neutral probability, the
only difference in pricing between the two models is the arrival rate matrix. When
the jump risk is priced, under the risk neutral probability measure, the market
spends more time on the first regime in which the underlying asset has a smaller
diffusion volatility and a smaller risk-free interest rate. This results in a lower
option price when the jump risk is priced.
74
Table 3.9: Comparison of European put option prices with priced and non-priced
jump risk
European Put Options
Regime 1, 100 Regime 2, 110.5171
N Priced Non Priced Diff Priced Non Priced Diff
20 8.73688 8.65535 -0.08153 7.24824 7.21182 -0.03642
40 8.81551 8.73153 -0.08398 7.28741 7.24995 -0.03746
80 8.83388 8.75078 -0.08310 7.26999 7.23391 -0.03608
160 8.84654 8.76353 -0.08301 7.26763 7.23193 -0.03570
320 8.85656 8.77321 -0.08335 7.27310 7.23725 -0.03585
640 8.85901 8.77575 -0.08326 7.27124 7.23555 -0.03569
1280 8.86148 8.77814 -0.08334 7.27255 7.23683 -0.03572
2560 8.86201 8.77874 -0.08327 7.27201 7.23633 -0.03568
5120 8.86252 8.77920 -0.08332 7.27208 7.23640 -0.03568
Table 3.10: Comparison of American call option prices with priced and non-priced
jump risk
American Call Options
Regime 1, 100 Regime 2, 110.5171
N Priced Non Priced Diff Priced Non Priced Diff
20 12.9940 12.8789 -0.1151 23.2553 23.1855 -0.0698
40 13.0802 12.9619 -0.1183 23.2869 23.2153 -0.0716
80 13.1024 12.9845 -0.1179 23.2657 23.1951 -0.0706
160 13.1169 12.9990 -0.1179 23.2615 22.1911 -0.0704
320 13.1279 13.0095 -0.1184 23.2660 22.1953 -0.0707
640 13.1308 13.0125 -0.1183 23.2637 22.1931 -0.0706
1280 13.1335 13.0151 -0.1184 23.2647 22.1941 -0.0706
2560 13.1342 13.0158 -0.1184 23.2641 22.1935 -0.0706
5120 13.1347 13.0163 -0.1184 23.2641 22.1935 -0.0706
75
Table 3.11: Comparison of American put option prices with priced and non-priced
jump risk
American Put Options
Regime 1, 100 Regime 2, 110.5171
N Priced Non Priced Diff Priced Non Priced Diff
20 9.12138 9.03967 -0.08171 7.59267 7.55410 -0.03857
40 9.19402 9.10973 -0.08429 7.62430 7.58468 -0.03962
80 9.21489 9.13110 -0.08379 7.60882 7.57031 -0.03851
160 9.22818 9.14429 -0.08389 7.60721 7.56895 -0.03826
320 9.23700 9.15286 -0.08414 7.61088 7.57252 -0.03836
640 9.23974 9.15563 -0.08411 7.60932 7.57108 -0.03824
1280 9.24192 9.15774 -0.08418 7.61021 7.57194 -0.03827
2560 9.24254 9.15838 -0.08416 7.60970 7.57147 -0.03823
5120 9.24298 9.15882 -0.08416 7.60971 7.57149 -0.03822
Table 3.12: Comparison of down-and-out call option prices with priced and
non-priced jump risk
Down-and-out Call Options
Regime 1, 100 Regime 2, 110.5171
N Priced Non Priced Diff Priced Non Priced Diff
20 9.19243 9.16036 -0.03207 19.8270 19.8061 -0.0209
40 9.59712 9.57579 -0.02133 20.3482 20.3321 -0.0161
80 9.20153 9.16939 -0.03214 19.7730 19.7497 -0.0233
160 9.11647 9.08205 -0.03441 19.6408 19.6162 -0.0246
320 9.20268 9.17050 -0.03218 19.7587 19.7347 -0.0240
640 9.11665 9.08220 -0.03445 19.6336 19.6086 -0.0250
1280 9.20246 9.17026 -0.03220 19.7546 19.7305 -0.0241
2560 9.11626 9.08178 -0.03448 19.6316 19.6063 -0.0250
5120 9.11352 9.07897 -0.03455 19.6272 19.6021 -0.0251
76
3.6 Conclusions
Naik’s (1993) Markov regime-switching model (MRSM) has been re-examined and
extended to k-state. The trinomial tree method developed in last chapter was
shown to be useful for finding the price of European options, American options
and barrier options by using the Markovian property of the regime-switching
process. Unlike the MRSM used in Elliott, Chan and Siu (2005), Naik’s model
allows the stock to jump when the regime switches. The jump of stock price
in Naik’s model theoretically gives ground to price regime-switching risk. If the
stock price dynamic of all underlying assets does not have a jump term, we think
that the regime-switching risk is not a fundamental risk, and need not be priced.
Therefore, Naik’s model is a good starting point if we want to price the regime-
switching risk.
We first considered the market where the jump risk during regime switching
was not priced. It is found that the possibility of stock price jump could result
in a greater option price as it increases the total volatility of the stock. For
the market with priced jump risk, the risk-free transition probability was found
by modifying the real transition probability and used to calculate the price of
derivatives. Therefore, the effect of priced jump risk can be understood more
easily by considering its effect on the pricing transition probability.
77
Chapter 4
Pricing Asian Option and Related EIAs
4.1 Introduction
In Chapter 2, we introduce the multi-state trinomial tree model to price the
European options, American options and barrier options in the Markov regime-
switching model (MRSM) of Buffington and Elliott (2002). The model is extended
to a MRSM with jump in Chapter 3. For strong path-dependent options, like
Asian option, the payoff of the derivatives depends on the path of the underlying
asset price process. We cannot apply the tree model directly to calculate their
price.
Equity-indexed annuities (EIA) is a popular insurance product in recent years.
The design of EIA is flexible and that makes the valuation of EIA a challenging
task. We use the method in Hull and White (1993) to price the Asian option in
our tree model and use the results to price some Asian-option-related EIAs by
iterative equations. We also identify the problem of quadratic approximation as
suggested by Hull and White (1993) and introduce a simple solution to it.
78
4.2 A Modified Trinomial Lattice
When we want to price Asian options, the average stock price plays a role which
makes the pricing of the Asian option more complex. The average asset price
depends on the path of the stock price process and can take a lot of different values
which cannot be reflected by the node position directly in the lattice model. This
causes the pricing of this strong path-dependent option a difficult problem in the
past. Hull and White (1993) presents an idea of representative sets of values, the
prices of the Asian options with the average stock price equal to representative
sets of values are calculated; when the average stock price level is not in the sets,
linear approximation is used to obtain the option price. The idea is a natural
extension of the lattice model and can be used in this chapter.
We assume T to be the expiration time of the option, N to be the number of
time steps, then ∆ = T/N . At time step t, there are 2t+ 1 nodes in the lattice,
the node is counted from the lowest stock price level, and St,n denotes the stock
price of the nth node at time step t. We let S0 = S0,0 be the initial stock price,
u = eσ√
∆ and d = u−1.
At a particular node, the highest average stock price can be achieved when the
stock price increases continuously and then decreases and reaches this node, and
similarly, the stock price decreases continuously and then increases can result in
79
the lowest average stock price value. Instead of calculating their values directly,
we use recursive method to find the highest average and lowest average of each
node in the lattice. We consider the node (t, n). If n equals to 0 or 2t, in all
time intervals, the stock price only drops or rises, the maximum average and
minimum average stock price level would be the same in these two situations.
We let ASmax(t, n) and ASmin(t, n) be the highest and lowest average stock price
at node (t, n). We have
ASmin(t, 0) = ASmax(t, 0) =S0
t
[1− dt+1
1− d− 1
2(1 + dt)
], (4.1)
ASmin(t, 2t) = ASmax(t, 2t) =S0
t
[ut+1 − 1
u− 1− 1
2(1 + ut)
]. (4.2)
The weight of the first and last node is half of the others and we base on this
method to find the average stock price in this section. We can also use the
approximation that each time step has the same weight. These approximations
are consistent with the discrete lattice model used in Hull and White (1993).
For an arbitrary node in the lattice, say (t, n), the path of stock price results in
lowest average comes from (t−1, n−2) just before reaching (t, n) and the highest
average comes from (t−1, n). This provides us with an easy way to calculate the
lowest and highest averages recursively at each node. However, we note that the
argument has a little problem, for the node (t, 1), there is no (t − 1,−1) for the
minimum path, and similarly, there is no (t − 1, 2t − 1) for the maximum path
of node (t, 2t − 1). Because we are now using a trinomial tree, the stock price
80
is allowed to remain unchange in any time interval. The lowest average for (t, 1)
is achieved when the stock price remains unchanged in the last step, so is the
highest average for (t, 2t− 1). Therefore, we have
ASmin(t, 1) = ASmin(t, 0) +1
2tS0d
t(u− 1), (4.3)
ASmax(t, 1) = ASmin(t, 1) +1
tS0(1− dt−1), (4.4)
ASmax(t, 2t− 1) = ASmax(t, 2t)−1
2tS0u
t(1− d), (4.5)
ASmin(t, 2t− 1) = ASmax(t, 2t− 1)− 1
tS0(ut−1 − 1). (4.6)
For the other nodes where n is not equal to 0, 1, 2t− 1, 2t, we have
ASmin(t, n) =1
t
[(t− 1)ASmin(t− 1, n− 2) +
1
2S0(un−2−t+1 + un−t)
],(4.7)
ASmax(t, n) =1
t
[(t− 1)ASmax(t− 1, n) +
1
2S0(un−t+1 + un−t)
]. (4.8)
We can calculate the number of representative values of each node based on
the idea in Hull and White (1993). The representative levels are taken as the form
S0emh where h is a constant and m is an integer (can be negative or zero) known
as the representative value. We first find the largest and smallest representative
values at each node. Let Mmin(t, n) and Mmax(t, n) represent the minimum and
the maximum value of m in node (t, n), we have
Mmin(t, n) = bln(ASmin(t, n))/hc , (4.9)
Mmax(t, n) = dln(ASmax(t, n))/he , (4.10)
81
where b. . .c and d. . .e represent the floor and the ceiling integral value, respec-
tively. Then, Mmin(t, n) and Mmax(t, n) are the possible range of representative
values of node (t, n) and we can calculate the price of the Asian option at each
node with the average stock price equals to all the representative levels of the
form S0emh. That is, the value m represents the average price level.
As all the regimes share the same lattice and the regime state cannot be re-
flected by the position of the nodes, each node has k possible derivative prices
corresponding to the current regime state at that node. The derivative price also
depends on the path of the stock which is now summarized using the represen-
tative value and each node can take one of the representative values between
Mmin(t, n) and Mmax(t, n) and obviously cannot be shown directly by the node
position. At the nth node of time step t under the jth regime state, let Vt,n,m,j
be the value of the derivative with representative value m, and V a(t, n, s, j) be
the value of the derivative with average stock price s. By definition, we have
Vt,n,m,j = V a(t, n, S0emh, j).
We assume that the Markov chain is independent of the Brownian motion
under the real market measure P , and the transition probabilities are not af-
fected by changing the probability measure from the physical probability to the
risk neutral probability. The transition probability does not change when the
probabilities assigned to the three branches change from the real measure to a
82
risk neutral measure and they can be considered separately during the calcula-
tion of the expectation. The price of the derivative at each node can be found by
iteration. We start from the expiration time, for example, for an average price
call option with strike price K,
VN,n,m,i = (S0emh −K)+ for all states i and all nodes n. (4.11)
Similar to the previous chapters, qij denotes the pricing transition probability
from regime i to regime j, πiu, πim, πid denote the risk neutral probabilities of the
stock price increases, remains unchanged and decreases, respectively, in the tree.
With the derivative payoff at expiration, using the following equation recursively
can help obtain the price of the Asian option,
Vt,n,m,i = V a(t, n, S0emh, i)
= e−ri∆k∑j=1
qij
[πiuV a(t+ 1, n+ 2,
tS0emh + St,n(1 + u)/2
t+ 1, j)
+πimV a(t+ 1, n+ 1,tS0e
mh + St,nt+ 1
, j)
+πidV a(t+ 1, n,tS0e
mh + St,n(1 + d)/2
t+ 1, j)].(4.12)
When the average price level is not at a representative level, we use a linear
approximation. If we have m = bln(AVE)/hc where AVE denotes the average
83
stock price value, then
V a(t, n, AVE, j) =AVE − S0e
mh
S0emh(eh − 1)V a(t, n, S0e
(m+1)h, j)
+S0e
(m+1)h − AVES0emh(eh − 1)
V a(t, n, S0emh, j)
=AVE − S0e
mh
S0emh(eh − 1)Vt,n,m+1,j +
S0e(m+1)h − AVE
S0emh(eh − 1)Vt,n,m,j
(4.13)
and the price of the option in all regimes can be obtained.
For an average strike option, the above method is still valid. Taking an average
strike call option as an example, we have
VN,n,m,i = (SN,n − S0emh)+ for all states i and all nodes n. (4.14)
Using the above recursive method, we can obtain the option price.
4.3 Pricing Equity-Indexed Annuities
Equity-indexed Annuities (EIAs) can protect the investors from a drop in equity
index and at the same time allow them to have a profit when the index appreci-
ates, which is an attractive feature for the investors. For the people who want to
do some investments in order to support their living after retirement, they do not
want to expose to the high risk of stocks but at the same time fear that the high
grade bonds cannot offer them a significant real return. The guarantee return of
84
EIA allow them to prepare for their living after retirement more easily as they
know at least how much they can obtain from EIA, and the EIA has a potential
to gain upside profits when the price of the reference asset appreciates.
For a long term investment, it is obvious that the expected return rate, the
volatility of stock prices and equity indices are not constant and it is more rea-
sonable to apply some stochastic models for the return rate and volatility when
we price an EIA as we know a small derivation in volatility can result in a large
error in EIA pricing.
Now we treat the security process S(t) as the equity index process and A(t)
is the average index level over time 0 to time t, that is
A(t) =1
t
∫ t
0
S(u)du. (4.15)
Then, we consider a general expression of a point-to-point Asian EIA which is
similar to the one used in Lin and Tan (2003) and has a cumulative return equal
to
C(t) = max[min[1 + αRt, (1 + ζ)t], (1 + g)t], (4.16)
where Rt = A(t)/S0 − 1. Rt is the average return of the equity index over the
period from time 0 to t; α is the participation rate that shows the extra return
received by the investors per unit of the average return of the equity index; ζ is
the cap rate which is the maximum annual return that can be enjoyed by the
85
investors and g is the guarantee rate which is the minimum annual return of the
EIA contract. Apart from the guarantee rate, the payment of this EIA contact has
some favorite features for the investors. Rt is the average return over the period
which can provide the investors a more stable return and can better reflecting the
performance of the equity index over the entire period. Various values of α can
be chosen by the investors according to their own risk preference and they can
expose to different levels of equity index risk. Due to these protection features,
the participation rate which is the potential extra return that can be obtained by
the investors might be small, if the investors require a high guarantee return. ζ
is a upper bound or the maximum return rate. By introducing this ceiling rate,
the investors can choose a higher participation rate while keeping the level of
guarantee return and the price of EIA.
The cumulative return of the EIA is very similar to a collar,
max[min[1 + αRt, (1 + ζ)t], (1 + g)t]
= 1 + αRt − [1 + αRt − (1 + ζ)t]+
+[(1 + g)t − (1 + αRt) + [1 + αRt − (1 + ζ)t]+]+
= 1 + αRt − [1 + αRt − (1 + ζ)t]+ + [(1 + g)t − (1 + αRt)]+. (4.17)
The last equality holds because when 1 +αRt > (1 + ζ)t, the value of the second
option is zero when ζ > g. Therefore, we can delete the redundant term [1 +
αRt − (1 + ζ)t]+ in the last expression. Furthermore, we can write the return of
86
EIA in terms of standard Asian options,
[1 + αRt − (1 + ζ)t]+ =α
S0
[S0
α+ A(t)− S0 −
S0
α(1 + ζ)t
]+
=α
S0
[A(t)− S0(1 +
(1 + ζ)t − 1
α)
]+
(4.18)
and
[(1 + g)t − (1 + αRt)]+ =
α
S0
[S0(1 +
(1 + g)t − 1
α)− A(t)
]+
. (4.19)
From the characteristics of EIA, we know that it consists of two options. The
minimum return guarantee results in a long position of a put option and the cap
return results in a short position of a call option. Because we are now considering
the return of EIA and thus a factor S−10 is used to obtain the proportional return
of the options. α is the participation rate. We notice that the participation
rate might not be equal to one, that means the EIA being considered does not
necessarily change by the same amount as what the average equity index does,
even the average return of the equity index is within the guarantee level and
the cap level. The strikes of both options are also adjusted according to the
participation rate.
Investors can demand a guarantee return every year rather than just a guaran-
tee return for the whole contract period so that they can lock the return annually
to avoid the fluctuation of the equity index in the future. In this way, they can
have a better forecast about what they get at the expiration of contract. This
87
kind of EIA is called the annual reset EIA or ratchet EIA and its cumulative
return is given by,
C(t) =t∏
k=1
max[min(1 + αR′k, 1 + ζ), 1 + g], (4.20)
where R′k =∫ kk−1
S(u)du/S(k−1)−1, which is the average index return of the kth
year. If the equity index follows the Black-Scholes framework, the appreciation
rate of the index in a time interval is independent of the return rate in the previous
intervals due to the independent increment property of Brownian motion and
so the expected return of the whole contract period is equal to the product of
expected return in each year. However, in our MRSM, the future return rate
and volatility of the index are affected by the current data due to the presence of
regime switching. For example, if the return at this time point is low, there is a
higher probability that we are now in a regime with low expected return and thus
the expected return in the next time period will be low because it is likely that
we are still in this low return regime state. Fortunately, the regime-switching
process is a Markov process, we are able to determine the expected return in the
year with the regime information at the very beginning of the year. Therefore,
we are able to solve this problem by considering a conditional expectation.
Similar to the geometric Brownian motion model, the risk neutral probability
can be obtained so that the value of the derivatives can be calculated as the risk-
free discounted expectation of the payoff of the derivatives. Let Q denote the risk
88
neutral probability measure. The expected risk neutral discounted value of the
unit ratchet EIA for t years with initial regime i is denoted by Vr(t, i) which is
equal to
EQ
[exp(−
∫ t
0
r(u)du)t∏
k=1
max[min(1 + αR′k, 1 + ζ), 1 + g]
∣∣∣∣∣X(0) = xi
]
= EQ
[exp(−
∫ 1
0
r(u)du) max[min(1 + αR′1, 1 + ζ), 1 + g] EQ[
exp(−∫ t
1
r(u)du)
×t∏
k=2
max[min(1 + αR′k, 1 + ζ), 1 + g]∣∣∣G1, X(0) = xi
]∣∣∣∣∣X(0) = xi
]
= Vr(1, i)k∑
j1=1
pij1(1)Vr(t− 1, j1)
= Vr(1, i)
[k∑
j1=1
pij1(1)Vr(1, j1)
×[ k∑j2=1
· · ·Vr(1, jt−2)[ k∑jt−1=1
pjt−2jt−1(1)Vr(1, jt−1)]· · ·]]. (4.21)
Therefore, with the values of 1-year Asian option of different initial regimes,
together with the 1-year transition probability, we are able to find the value of
the ratchet EIA recursively using the above equation.
We can include the mortality component into the EIA contact. We assume
that the ratchet EIA is payable at the end of the year that the investor dies or
the EIA contract expires. There is no selection effect and the future lifetime
random variable is independent of the Brownian motion and regime-switching
process under the real and pricing measures. Furthermore, we assume that the
89
numbers of participants of this EIA contract in different ages are large and so
the insurance companies can diversify their mortality risk and are risk neutral
towards this risk. We denote the future lifetime, the real probability of death,
survival in one year of a x-year-old person by T (x), qx and px respectively. Let
FMt t∈T be the natural filtration of the future lifetime random variable. We
define Ht to be the σ-algebra FXt ∨ FZt ∨ FMt , using the same notations as in
Chapter 2. The expected risk neutral discounted value of the unit life dependent
ratchet EIA for t years with initial regime i for a x-year-old investor is denoted
90
as Vm,x(t, i) which is equal to
EQ
[exp
[−∫ min(t,dT (x)e)
0
r(u)du]
×min(t,dT (x)e)∏
k=1
max[min(1 + αR′k, 1 + ζ), 1 + g]
∣∣∣∣∣X(0) = xi
]
= EQ
[exp
[−∫ 1
0
r(u)du]
max[min(1 + αR′1, 1 + ζ), 1 + g]
×[I(T (x) ≤ 1) + I(T (x) > 1)EQ
[exp[−
∫ min(t,dT (x+1)+1e)
1
r(u)du]
×min(t,dT (x+1)+1e)∏
k=2
max[min(1 + αR′k, 1 + ζ), 1 + g]∣∣∣H1, X(0) = xi
]]∣∣∣∣∣X(0) = xi
]
= Vm,x(1, i)
[qx + px
k∑j1=1
pij1(1)Vm,x+1(t− 1, j1)
]
= Vm,x(1, i)
[qx + px
k∑j1=1
pij1(1)Vm,x+1(1, j1)
[qx+1 + px+1
k∑j2=1
· · ·
×Vm,x+t−2(1, jt−2)[qx+t−2 + px+t−2
k∑jt−1=1
pjt−2jt−1(1)Vm,x+t−1(1, jt−1)]· · ·]].
(4.22)
If we have the values of ratchet EIAs with different expirations, together with the
mortality information, we can find the life dependent ratchet EIA easily,
Vm,x(1, i) = Vr(1, i) for all i and all x, (4.23)
Vm,x(t, i) =t−1∑n=0
npx qx+n Vr(n+ 1, i) + tpx Vr(t, i), (4.24)
where npx is the probability of a x-year-old person survives for n years. Because
the EIA contracts can be expressed in terms of Asian options, their values can be
91
calculated using the trinomial tree method or with the above iterative equations.
4.4 Numerical Results and Analysis
The trinomial tree introduced in this chapter is both efficient and easy to use for
pricing Asian options and the other related derivatives like EIA. In the beginning
of this section, we compare our results with those of Hull and White (1993) to
show the similarity between our trinomial tree and the binomial tree without
regime switching. We study the properties of the option prices found by the tree
model. We also compare our results with those of Boyle and Draviam (2007) in
order to show the validity of this model under regime-switching model. Other
numerical results are presented so that we can have a better understanding of this
pricing method and the characteristics of Asian options and related derivatives.
In Table 4.1 and Table 4.2, we use different methods to approximate the price
of European type and American type average price call option. The adjusted
average takes the weights of the first and the last nodes in the lattice be half
of the others while the simple average takes all nodes with the same weight. It
shows that the option price obtained by using the trinomial lattice is very close to
the value obtained by Hull and White (1993), and that means trinomial tree has
a similar performance in pricing Asian options as binomial tree of Hull and White
(1993). The results of using simple average are closer to those of adjusted average
92
Table 4.1: Comparison of the prices of (Eurpean type) average price call options
in simple BS model (linear approximation of representative value)
Average Price Call Option (Eurpean type)
Number of Time Steps (N)
h Results 20 40 60 80
0.100 Adjusted Average 4.6756 4.6870 4.6904 4.6919
Simple Average 4.6419 4.6705 4.6792 4.6839
Hull and White 4.663 4.679 4.685 4.687
0.050 Adjusted Average 4.6110 4.6174 4.6194 4.6204
Simple Average 4.5794 4.6011 4.6083 4.6124
Hull and White 4.588 4.605 4.612 4.614
0.010 Adjusted Average 4.5451 4.5443 4.5454 4.5462
Simple Average 4.5119 4.5274 4.5242 4.5377
Hull and White 4.517 4.530 4.536 4.539
0.005 Adjusted Average 4.5413 4.5365 4.5358 4.5360
Simple Average 4.5079 4.5194 4.5243 4.5274
Hull and White 4.513 4.522 4.526 4.529
0.003 Adjusted Average 4.5404 4.5347 4.5331 4.5326
Simple Average 4.5073 4.5177 4.5217 4.5240
Hull and White 4.512 4.520 4.523 4.525
†The initial stock price and strike price are 50, the risk free interest rate is 10% and the
volatility is 0.30.
93
Table 4.2: Comparison of the prices of (American type) average price call options
in simple BS model (linear approximation of representative value)
Average Price Call Option (American Type)
Number of Time Steps (N)
h Results 20 40 60 80
0.100 Adjusted Average 5.0210 5.3027 5.3468 5.3692
Simple Average 5.0583 5.2179 5.2826 5.3207
Hull and White 5.197 5.311 5.360 5.377
0.050 Adjusted Average 5.0721 5.1305 5.1586 5.1716
Simple Average 4.9405 5.0524 5.1006 5.1286
Hull and White 4.971 5.080 5.124 5.145
0.010 Adjusted Average 4.9540 4.9819 4.9958 5.0038
Simple Average 4.8202 4.9033 4.9402 4.9608
Hull and White 4.823 4.906 4.941 4.962
0.005 Adjusted Average 4.9479 4.9693 4.9790 4.9848
Simple Average 4.8138 4.8904 4.9231 4.9414
Hull and White 4.815 4.892 4.924 4.942
0.003 Adjusted Average 4.9466 4.9665 4.9749 4.9794
Simple Average 4.8127 4.8875 4.9189 4.9357
Hull and White 4.814 4.890 4.920 4.936
†The initial stock price and strike price are 50, the risk free interest rate is 10% and the
volatility is 0.30.
94
when N is larger, which is obvious as more time steps being used, the effect of
individual time step is smaller. Increasing number of time steps increases the
numbers of nodes, and at the same time the possible range of average asset price
for nodes. Because using the adjusted average to do the calculation can improve
the consistency of approximating, we use this adjusted average approximation.
Since the calculation in Hull and White (1993) is based on simple averages, both
approximations are used to find the option prices so that the comparisons between
different methods shown in the tables are more meaningful.
The option prices are calculated for the set of different representative values
and that imposes further approximation error on option valuation. Hull and
White (1993) suggest that quadratic interpolation can be used to improve the
accuracy of the approximation, but our numerical results are different from what
we predicted. In Table 4.3 and Table 4.4, the prices of the options calculated
by quadratic approximation are significantly lower than the results found by
linear approximation, especially when h and N are large and the options have
early exercise option. Our goal of using quadratic interpolation is to reduce the
approximation error when h is large so that the method can be of high efficiency.
However, the quadratic approximation tends to underestimate the price of the
Asian options rather than provides a more accurate result when h is large.
The payoff of the Asian option is linear to the average asset price to a certain
95
Table 4.3: Comparison of the prices of (Eurpean type) average price call options
in simple BS model (quadratic and modified quadratic approximation, simple
average asset price)
Average Price Call Option (Eurpean type)
Number of Time Steps (N)
h Method 20 40 60 80
0.100 Modified Quad. 4.5949 4.6021 4.6059 4.6083
Quadratic Appro. 4.3660 4.3720 4.3736 4.3754
Hull and White 4.663 4.679 4.685 4.687
0.050 Modified Quad. 4.5353 4.5456 4.5493 4.5512
Quadratic Appro. 4.4779 4.4869 4.4903 4.4920
Hull and White 4.588 4.605 4.612 4.614
0.010 Modified Quad. 4.5075 4.5185 4.5219 4.5237
Quadratic Appro. 4.5058 4.5158 4.5192 4.5209
Hull and White 4.517 4.530 4.536 4.539
0.005 Modified Quad. 4.5068 4.5170 4.5206 4.5224
Quadratic Appro. 4.5066 4.5165 4.5200 4.5217
Hull and White 4.513 4.522 4.526 4.529
0.003 Modified Quad. 4.5068 4.5168 4.5203 4.5221
Quadratic Appro. 4.5067 4.5166 4.5201 4.5219
Hull and White 4.512 4.520 4.523 4.525
†The initial stock price and strike price are 50, the risk free interest rate is 10% and the
volatility is 0.30.
96
Table 4.4: Comparison of the prices of (American Type) average price call options
in simple BS model (quadratic and modified quadratic approximation, simple
average asset price)
Average Price Call Option (American Type)
Number of Time Steps (N)
h Method 20 40 60 80
0.100 Modified Quad. 4.9529 5.0615 5.1114 5.1352
Quadratic Appro. 4.5826 4.5321 4.5076 4.5053
Hull and White 5.197 5.311 5.360 5.377
0.050 Modified Quad. 4.8644 4.9268 4.9655 4.9733
Quadratic Appro. 4.7590 4.7932 4.8162 4.8238
Hull and White 4.971 5.080 5.124 5.145
0.010 Modified Quad. 4.8128 4.8875 4.9178 4.9336
Quadratic Appro. 4.8112 4.8848 4.9149 4.9307
Hull and White 4.823 4.906 4.941 4.962
0.005 Modified Quad. 4.8119 4.8861 4.9169 4.9331
Quadratic Appro. 4.8117 4.8857 4.9164 4.9325
Hull and White 4.815 4.892 4.924 4.942
0.003 Modified Quad. 4.8120 4.8859 4.9166 4.9329
Quadratic Appro. 4.8120 4.8858 4.9165 4.9327
Hull and White 4.814 4.890 4.920 4.936
†The initial stock price and strike price are 50, the risk free interest rate is 10% and the
volatility is 0.30.
97
extent. When the average price is below the strike, the payoff is zero. Once the
average price is greater than the strike, the payoff increases linearly with respect
to the average price. Therefore, the value of the Asian option at expiration is two
joined linear equations with a kink at strike. Normally quadratic interpolation
is a better approximation than linear approximation because the linear equation
is just a special case of a quadratic equation. However, when the average asset
price is close to the strike, as the payoff is in fact two joined linear equations,
it is obvious that linear interpolation provides an exact solution while quadratic
interpolation which takes three points to do the calculation underestimates the
payoff. Underestimation of the payoff takes effect at various nodes and makes
the final result lower than the true value. We can use linear approximation when
the average asset value is close to the strike and use quadratic approximation
elsewhere. This modification provides a better result. However, we know that
linear approximation tends to overestimate the price while quadratic approxima-
tion tends to underestimate the price, this kind of mixture effects might cause a
chaotic convergent pattern.
We then compare our method with Boyle and Draviam (2007). Assume that
the underlying asset with initial price of S0, and there are two regimes. At Regime
1, the diffusion volatility of the asset price is 15%; and at Regime 2, the diffusion
volatility of the asset price is 25%. The risk-free interest rate at both regimes
98
is 5%. The average price call option lasts for a year and the generator matrix
of the Markov chain regime switching process is
a11 a12
a21 a22
=
−1 1
1 −1
.
We compare the values of the options with different initial asset prices S0 and
different strikes K.
We can see that the results using simple average asset price are very close
to those of Boyle and Draviam (2007) in Table 4.5. Boyle and Draviam (2007)
use partial differential equations to solve the problem in which simple average
and linear approximation are used during discretization. Because the price of the
Asian option is a convex function of the average asset price, linear approximation
overestimates the price of options, and therefore, the modified quadratic approx-
imation tends to be smaller compared to the results of the other two methods.
The same observation is found when we are considering the price of options under
different sets of transition probabilities in Table 4.6.
Comparing with other valuation methods, lattice model has an inborn advan-
tage of obtaining the price of derivative with early exercise option easily. We
consider the same situation as we did in Table 4.5 but for the price of a average
price call option with early exercise option in Table 4.7.
Using equation (4.17), we are able to find the fair value of the EIA with the
Asian option prices obtained from the trinomial tree. In order to consider the
99
Table 4.5: Comparison of the price of average price call options in MRSM found
by different methods I
Average Price Call Option (European Type)
Regime 1 (σ1 = 0.15, r1 = 5%) Regime 2 (σ2 = 0.25, r2 = 5%)
S0 Method K = 90 K = 100 K = 110 K = 90 K = 100 K = 110
90 Simple 4.6104 1.1134 0.1969 5.8747 2.1804 0.6685
B&D 4.6204 1.1172 0.1966 5.8747 2.1808 0.6694
Modified 4.5964 1.0970 0.1899 5.8655 2.1688 0.6600
95 Simple 8.1076 2.6203 0.5802 9.1482 3.9854 1.4278
B&D 8.1132 2.6288 0.5809 9.1475 3.9850 1.4281
Modified 8.0937 2.6014 0.5668 9.1380 3.9731 1.4165
100 Simple 12.3354 5.1227 1.4520 13.0385 6.5274 2.7007
B&D 12.3374 5.1338 1.4574 13.0381 6.5274 2.7010
Modified 12.3253 5.1071 1.4331 13.0294 6.5172 2.6876
105 Simple 16.9513 8.5765 3.0859 17.3581 9.7668 4.6045
B&D 16.9523 8.5831 3.0956 17.3580 9.7659 4.6041
Modified 16.9453 8.5608 3.0651 17.3506 9.7554 4.5911
110 Simple 21.7336 12.7213 5.6350 21.9427 13.5779 7.1801
B&D 21.7353 12.7242 5.5472 21.9435 13.5774 7.1802
Modified 21.7306 12.7091 5.6179 21.9372 13.5675 7.1689
†B&D refers to the results of Boyle and Draviam (2007). For the lattice methods, N and h
are set to be 200 and 0.005 respectively. Simple refers to results using equal weight on each
node and linear approximation. Modified refers to results using adjusted weight on each node
and modified quadratic approximation.
100
Table 4.6: Comparison of the price of average price call options in MRSM ob-
tained by different methods II
Average Price Call Option (European Type)
Regime 1 (σ1 = 0.15, r1 = 5%) Regime 2 (σ2 = 0.25, r2 = 5%)
S0 Method K = 90 K = 100 K = 110 K = 90 K = 100 K = 110
0 Simple 12.1595 4.7007 1.0821 13.2292 6.8674 3.0211
B&D 12.1617 4.7135 1.0864 13.2286 6.8684 3.0228
Modified 12.1515 4.6864 1.0638 13.2187 6.8561 3.0073
0.5 Simple 12.2631 4.9493 1.3001 13.1168 6.6663 2.8315
B&D 12.2651 4.9609 1.3053 13.1165 6.6669 2.8325
Modified 12.2538 4.9340 1.2817 13.1073 6.6559 2.8183
1 Simple 12.3354 5.1227 1.4520 13.0385 6.5274 2.7007
B&D 12.3374 5.1338 1.4574 13.0381 6.5274 2.7010
Modified 12.3253 5.1071 1.4331 13.0294 6.5172 2.6876
1.5 Simple 12.3877 5.2480 1.5609 12.9819 6.4279 2.6072
B&D 12.3899 5.2589 1.5667 12.9812 6.4273 2.6069
Modified 12.3772 5.2325 1.5421 12.9729 6.4177 2.5940
2 Simple 12.4267 5.3414 1.6417 12.9396 6.3542 2.5380
B&D 12.4291 5.3523 1.6480 12.9387 6.3531 2.5372
Modified 12.4160 5.3261 1.6231 12.9306 6.3438 2.5247
†The generator of the Markov chain is taken to be
−λ λ
λ −λ
.
101
effect of the interest rate in regime-switching model, we change the interest rate
in the Regime 2, r2, from 5% to 7% in the following analysis. Table 4.8 shows the
value of Vr(1, i) as defined in equation (4.21) with different guarantee and ceiling
rates; the participation rate α is equal to 1. We find that two approximation
methods are very close to each other. With the value of Vr(1, i), we can obtain
the price of ratchet EIAs easily using recursive equation. We make use of the
previous results and consider the case when ceiling and guarantee rates are equal
to 20% and 3%, respectively. The prices of unit annual reset EIA with different
expirations are shown in Table 4.9.
4.5 Conclusions
In this chapter, we have applied a modified trinomial tree method to obtain the
price of Asian option under the Markov regime-switching model (MRSM). We
have extended the trinomial tree model in Chapter 2 using the idea of represen-
tative values of Hull and White (1993) and obtained the fair value of various
Asian options in the MRSM. From the numerical results we found that quadratic
approximation as suggested by Hull and White (1993) significantly underesti-
mate values of the Asian options when average price level is close to the strike
price. We suggested to use linear approximation when average price level is close
to the strike price and use quadratic approximation elsewhere so that the un-
102
derestimation problem can be solved, and at the same time we can take the
advantage of quadratic approximation. The weights of the asset price at the
start and expiration can be taken as half of the other time points to improve the
approximation of average asset price. However, these improvements are not sig-
nificant when the number of time steps is large and the representative constant is
small. The structure of the lattice model can help find the price of Asian options
with early exercise option. The prices of Asian options can also be used to price
point-to-point EIA and ratchet EIA by using a simple iterative equation with the
Markovian property of the regime-switching process.
103
Table 4.7: Price of average price call options with early exercise option in MRSM
Average Price Call Option (American Type)
Regime 1 (σ1 = 0.15, r1 = 5%) Regime 2 (σ2 = 0.25, r2 = 7%)
S0 Method K = 90 K = 100 K = 110 K = 90 K = 100 K = 110
90 Simple 5.0528 1.1510 0.1993 6.5088 2.2931 0.6845
Modified 5.0370 1.1333 0.1921 6.5067 2.2815 0.6757
95 Simple 9.2242 2.7757 0.5928 10.4691 4.2914 1.4812
Modified 9.2197 2.7548 0.5787 10.4824 4.2813 1.4694
100 Simple 14.1866 5.6142 1.5066 15.2530 7.2320 2.8513
Modified 14.1980 5.5967 1.4861 15.2867 7.2297 2.8385
105 Simple 19.2650 9.7290 3.2779 20.3650 11.1434 4.9701
Modified 19.2790 9.7214 3.2542 20.4078 11.1561 4.9590
110 Simple 24.3479 14.6381 6.1757 25.5137 15.8374 7.9552
Modified 24.3641 14.6491 6.1563 25.5604 15.8711 7.9526
104
Table 4.8: Price of one-year EIA in MRSM
Price of One-year EIA at Regime 1
Guarantee Ceiling Return
Return Method 5% 10% 15% 20% 25%
0% Simple 0.96881 0.98304 0.99085 0.99476 0.99659
Modified 0.96885 0.98307 0.99085 0.99470 0.99650
1% Simple 0.97292 0.98715 0.99496 0.99887 1.00070
Modified 0.97294 0.98716 0.99494 0.99879 1.00059
2% Simple 0.97741 0.99164 0.99946 1.00336 1.00520
Modified 0.97743 0.99165 0.99943 1.00328 1.00507
3% Simple 0.98229 0.99652 1.00434 1.00824 1.01008
Modified 0.98230 0.99652 1.00430 1.00815 1.00995
Price of One-year EIA at Regime 2
Guarantee Ceiling Return
Return Method 5% 10% 15% 20% 25%
0% Simple 0.96076 0.97743 0.98878 0.99610 1.00062
Modified 0.96079 0.97747 0.98880 0.99611 1.00061
1% Simple 0.96498 0.98166 0.99300 1.00032 1.00484
Modified 0.96500 0.98168 0.99301 1.00032 1.00482
2% Simple 0.96947 0.98615 0.99749 1.00481 1.00934
Modified 0.96949 0.98617 0.99750 1.00481 1.00931
3% Simple 0.97424 0.99092 1.00226 1.00958 1.01411
Modified 0.97425 0.99093 1.00227 1.00957 1.01407
105
Table 4.9: Price of unit annual reset EIA in MRSM
Regime 1
Length of Ratchet EIA (years)
Method 1 2 3 4 5 10 15 20
Simple 1.00824 1.01714 1.02619 1.03534 1.04457 1.09195 1.14149 1.19328
Modified 1.00815 1.01699 1.02599 1.03508 1.04425 1.09135 1.14057 1.19201
Regime 2
Length of Ratchet EIA (years)
Method 1 2 3 4 5 10 15 20
Simple 1.00958 1.01867 1.02776 1.03693 1.04617 1.09363 1.14325 1.19511
Modified 1.00957 1.01861 1.02765 1.03676 1.04595 1.09312 1.14243 1.19395
106
Chapter 5
Concluding Remarks
Option, which is expected to be a tool for hedging, is sometimes also used for
speculative purposes. It is very popular among some investors because of its high
leverage. Selling options can be a profitable business for the investment banks
and other financial institutions. However, option pricing is difficult and requires
profound knowledge in financial economics and long-time experience of trading
in markets. A simple option pricing method not only promotes the efficiency in
trading options but also reduces the cost for those who really wants to use options
to hedge their investment risks as options can be sold at a more reasonable price.
Stock price models become more complicated and various complex derivatives are
introduced into the market. Derivative pricing is challenging and is mainly done
by PDE and simulations. Simple methods like lattice model lose their glory.
Lattice models are simple and easy to understand. It is a good news if lattice
models can be used to find the price of options even in more advanced models.
Multi-state trinomial tree model was introduced to find a fair price of various
derivatives in MRSMs.
In Chapter 2, we focused on Buffington and Elliott’s (2002) model. We intro-
duced the multi-state trinomial tree model and used it to price different options.
107
The nature of regime-switching risk was discussed in details and we suggested
that this risk need not be priced under this model. Although pricing jump risk
need not result in arbitrage, jump risk can be eliminated in the market if the
fundamental asset price processes do not jump during regime switching.
To justify pricing the regime-switching risk, the model of Naik (1993) has
been studied and extended in Chapter 3 in which stock price jumps when regime
switches. By using the Markovian property of the regime-switching process, tri-
nomial tree is still efficient for option pricing under this model. The numerical
results were analyzed and it was found that some properties that are true in one-
state market are also true for the MRSM, for example, early exercise is also not
optimal for American call option in the MRSM if the underlying asset pays no
dividend.
Asian option is strongly path-dependent and we cannot directly use the ap-
proach in Chapter 2 and Chapter 3 to price Asian option. In Chapter 4, we
applied the method of Hull and White (1993) to price Asian option in Buffington
and Elliott’s (2002) MRSM. The problem of quadratic approximation suggested
in Hull and White (1993) was identified and a simple method was introduced to
solve this problem. The fair price of Asian-option-related EIAs can then be found
using iterative equations.
With the method to calculate the price of derivatives in MRSM, a more com-
108
plicated problem emerges, it is how to find an appropriate risk neutral probability
(and transition probability) from the market price of the options. One of the ma-
jor difficulties is that we can only have the price information of the current regime.
Further researches can focus in this area.
109
References
Aingworth, D. D., Das, S. R. and Motwani, R. (2006). A simple approach for
pricing equity options with Markov switching state variables, Quantitative
Finance, 6(2), 95–105.
Baule, R. and Wilkens, M. (2004). Lean trees – a general approach for improv-
ing performance of lattice models for option pricing, Review of Derivatives
Research, 7, 53–72.
Black, F. and Scholes, M. (1973). The pricing of options and corporate liabilities,
Journal of Political Economy, 81, 637–654.
Bollen, N. P. B. (1998). Valuing options in regime-switching models, Journal of
Derivatives, 6, 8–49.
Boyle, P. P. (1986). Option valuation using a three-jump process, International
Options Journal, 3, 7–12.
Boyle, P. P. (1988). A lattice framework for option pricing with two state
variables, Journal of Financial and Quantitative Analysis, 23(1), 1–12.
Boyle, P. P. and Draviam, T. (2007). Pricing exotic options under regime switch-
ing, Insurance: Mathematics and Economics, 40, 267–282.
110
Boyle, P. P., Evine, J. and Gibbs, S. (1989). Numerical evalution of multivariate
contingent claims, The Review of Financial Studies, 2(2), 241–250.
Boyle, P. P. and Tian, Y. (1998). An explicit finite difference approach to the
pricing of barrier options, Applied Mathematical Finance, 5, 17–43.
Boyle, P. P. and Wang, T. (2001). Pricing of new securities in an incomplete
market: The catch 22 of no-arbitrage pricing, Mathematical Finance, 11(3),
267–284.
Bremaud, P. (1981). Point processes and queues, Springer-Verlay, New York.
Buffington, J. and Elliott, R. J. (2002). American options with regime switching,
International Journal of Theoretical and Applied Finance, 5(5), 497–514.
Cherny, A. S. and Maslov, V. P. (2003). On minimization and maximization of
entropy in various disciplines, Theory of Probability and Its Applications,
48(3), 466–486.
Cox, J. C., Ross, S. A. and Rubinstein, M. (1979). Option pricing: a simplified
approach, Journal of Financial Economics, 7, 229–263.
Davis, M. H. A. (1997). Option pricing in incomplete markets, in Dempster,
M. A H. and Pliska, S. R. (eds.), Mathematics of Derivative Securities,
Cambridge University Press, Cambridge, 216–226.
111
Elliott, R. J., Aggoun, L. and Moore, J. B. (1994). Hidden Markov models:
estimation and control, Springer-Verlag, New York.
Elliott. R. J., Chan, L. L. and Siu, T. K. (2005). Option pricing and Esscher
transform under regime switching, Annals of Finance, 1, 423–432.
Elliott, R. J., Siu, T. K. and Chan, L. L. (2008). A PDE approach for risk
measures for derivatives with regime switching. Annals of Finance, 4, 55–
74.
Figlewski, S. and Gao, B. (1999). The adaptive mesh model: a new approach
to efficient option pricing, Journal of Financial Economics, 53, 313–351.
Follmer, H. and Sondermann, D. (1986). Hedging of contingent claims un-
der incomplete information, in Hildenbrand, W. and Mas-Colell, A. (eds.),
Contributions to Mathematical Economics, 205–223.
Follmer, H. and Schweizer, M. (1991). Hedging of contingent claims under
incomplete information, in Davis, M.H.A. and Elliot, R.J. (eds.), Applied
Stochastic Analysis, 389–414.
Frittelli, M. (2000). The minimal entropy martingale measure and the valuation
problem in incomplete markets, Mathematical Finance, 10(1), 39–52.
Fujiwara, T. and Miyahara, Y. (2003). The minimal entropy martingale mea-
112
sures for geometric Levy processes, Finance and Stochastics, 7, 509–531.
Gerber H. U. and Shiu, E. S. W. (1994). Option pricing by Esscher transforms,
Transactions of the Society of Actuaries 1994, 46, 99–191.
Guo, X. (2001). Information and option pricings, Quantitative Finance, 1, 37–
57.
Hamilton, J. D. (1989). A new approach to the economic analysis of nonsta-
tionary time series and the business cycle, Ecomometrica, 57(2), 357–384.
Hull, H. C. and White, A. (1993). Efficient procedures for valuing European
and American path-dependent options. The Journal of Derivatives, 1(1),
21–31.
Hull, H. C. (2006). Options, futures, and other derivatives, Addison Wesley,
New Jersey, 6th Edition.
Jarrow, R. and Rudd, A. (1983). Option Pricing, Dow Jones-Irwin, Homewood,
Ill.
Kamrad, B. and Ritchken, P. (1991). Multinomial approximating models for
options with k state variables, Management Science, 37(12), 1640–1652.
Landen, C. (2000). Bond pricing in a hidden Markov model of the short rate,
Finance and Stochastics, 4, 371–389.
113
Lee, H. (2002). Pricing exotic options with applications to equity-indexed annu-
ities, Ph.D thesis, Department of Statistics and Actuarial Science, Univer-
sity of Iowa, USA.
Lee, H. (2003). Pricing equity-indexed annuities with path-dependent options,
Insurance: Mathematics and Economics, 33, 677–690.
Lin, X. S. and Tan, K. S. (2003). Valuation of equity-indexed annuities under
stochastic interest rates. North American Actuarial Journal, 7(4), 72–91.
Mamon, R. S. and Rodrigo, M. R. (2005). Explicit solutions to European options
in a regime-switching economy, Operations Research Letters, 33, 581–586.
Mao, X. and Yuan, C. (2006). Stochastic differential equations with Markovian
switching, Imperial College Press, London.
Markovitz, H. (1952). Portfolio selection, The Journal of Finance, 7(1), 77–91.
McDonald, R. L. (2006). Derivatives Markets, Addison Wesley, Boston, MA,
2nd Edition.
Merton, R. C. (1969). Lifetime portfolio selection under uncertainty: the continuous-
time case, The Review of Economics and Statistics, 51(3), 247–257.
Merton, R. C. (1976). Option pricing when underlying stock returns are discon-
tinuous, Journal of Financial Economics, 3(1), 125–144.
114
Miyahara, Y. (2001). [Geometric Levy process & MEMM] Pricing model and
related estimation problems, Asia-Pacific Financial Markets, 8, 45–60,
Naik, V. (1993). Option valuation and hedging strategies with jumps in volatility
of asset returns, The Journal of Finance, 48(5), 1969–1984.
Øksendal, B. (1998). Stochastic differential equations: an introduction with
applications, Springer-Verlag, Berlin Heidelberg, 5th Edition.
Omberg, E. (1987). A note on the convergence of binomial-pricing and compound-
option models, The Journal of Finance, 42(2), 463–469.
Pedler, P. J. (1971). Occupation times for two state Markov chains, Journal of
Applied Probability, 8, 381–390.
Penaud, A., Wilmott, P. and Ahn, H. (1999). Exotic passport options, Asia-
Pacific Financial Markets, 6, 171–182.
Schweizer, M. (1996). Approximation Pricing and the variance-optimal martin-
gale measure, Annals of Probability, 24, 206–236.
Siu, T. K., Yang, H. L. and Lau, J. W. (2008). Pricing currency options un-
der two-factor Markov-modulated stochastic volatility models, Insurance:
Mathematics and Economics, 43(3), 295–302.
115
Steele, J. M. (2004). Stochastic Calculus and Financial Applications, Springer-
Verlag, New York.
Tiong, S. (2000). Valuing equity-indexed annuities, North American Actuarial
Journal, 4(4), 149–170.
Wilmott, P. (2000). Paul Wilmott on Quantitative Finance, Wiley, Chichester.
Zvan, R., Vetzal, K. R. and Forsyth, P. A. (2000). PDE methods for pricing
barrier options, Journal of Economic Dynamics and Control, 24, 1563–1590.
Yuen, F. L. and Yang, H. (2009). Option pricing in a jump-diffusion model with
regime-switching. ASTIN Bulletin, 39(2), 515–539.
Yuen, F. L. and Yang, H. (2010). Option pricing with regime switching by
trinomial tree method, Journal of Computational and Applied Mathematics,
233(8), 1821–1833.
Yuen, F. L. and Yang, H. (2010). Pricing Asian options and equity-indexed
annuities with regime-switching by trinomial tree method, North American
Actuarial Journal, 14(2), 256–272.
116