Numerical Methods For Pricing Basket Options
Numerical Methods For Pricing Basket Options
By
2004
find closed form solutions for their values. Finding numerical solutions for
the governing pricing equations becomes therefore an appealing approach to
pricing, especially since powerful desktop computers are now available.
We also take a look at one of the most competitive markets today, The
benefits and how they are related to the lookback put options.
ii
In the memory of my parents
iii
ACKNOWLEDGMENTS
A paper is almost always the product not only of its authors, but also
of the environment where the authors work, of the encouragements and crit-
ics gathered from colleagues and teachers, conversations after seminars and
and Misu Negritoiu from the University of Bucharest and Neil Falkner, Neil
Robertson and John Hsia from The Ohio State University.
iv
VITA
Bucharest
1998 . . . . . . . . . . . . . . . . . . . . . . B.S. International Transactions, Academy
of Economical Studies, Bucharest
FIELDS OF STUDY
v
TABLE OF CONTENTS
ABSTRACT ................................. ii
ACKNOWLEDGMENTS ........................ iv
VITA ...................................... v
1 INTRODUCTION ........................... 1
1.2 Hedging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Arbitrage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 A Simple Model for Asset Pricing–The Lognormal Random
Walk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
vi
1.6 Derivation of the Black-Scholes Formula for Two Assets . . . . 6
1.7 Options on Dividend Paying Assets . . . . . . . . . . . . . . . 9
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 The Monte-Carlo Simulation Algorithm . . . . . . . . . . . . . 31
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
vii
4.5 Guaranteed Minimum Death Benefits . . . . . . . . . . . . . . 45
4.6 Numerical Example . . . . . . . . . . . . . . . . . . . . . . . . 49
APPENDICES ................................ 54
BIBLIOGRAPHY ............................. 72
viii
LIST OF FIGURES
ix
CHAPTER 1
INTRODUCTION
possess shares (also known as equity certificates or stocks) and may or may
not receive dividends, depending on whether the company makes a profit and
investors about the likely dividend payments, future earnings and resources
that the company will control.
1
One of the simplest financial derivatives is a European call option. This is
a contract with the following conditions: at a prescribed time in the future,
known as the expiry date, the holder of the option may purchase a prescribed
asset, known as the underlying asset for a prescribed amount, known as the
The word may in the above description implies that, for the holder of
the option, the contract is a right and not an obligation. The other party
to the contract, who is known as the writer, does have a potential obligation:
The right to sell an asset is known as a put option and has payoff prop-
The options that may be exercised only at expiry are called European
options and those that may be exercised at any time prior to the expiry date
are called American options.
depend on the history of an asset price, not just on its value at expiry date.
1.2 Hedging
investors enter into large financial positions. These entities and individuals
2
wish to protect themselves from risk and uncertainty or wish to limit the risk
and uncertainty to tolerable levels.
struments.
1.3 Arbitrage
ential between instruments, generally with the belief that the return will be
It is often stated that asset prices must move randomly because of the effi-
cient market hypothesis. There are several different forms of this hypothesis
with different restrictive assumptions, but they are basically asserting two
things:
3
1. The past history is fully reflected in the present price, which does not
hold any further information, and
Thus the modelling of asset prices is really about modelling the arrival of
it is $100.
We can write:
dS
= µdt + σdW,
S
where:
µ = drift, a measure of the average rate of growth of the asset
price;
σ = volatility, a measure of the standard deviation of the return;
W = a Wiener process,
µ ¶
σ2
d(ln(St )) = µ− dt + σdWt .
2
4
The solution of the above equation is:
·µ ¶ ¸
σ2
St = S0 exp µ − t + σWt .
2
Let V (S1 (t), S2 (t), t) be the price of a derivative, with S1 (t) and S2 (t) the
we obtain:
∂V ∂V 1 ∂ 2V 2 1 ∂ 2V 2 1 ∂ 2V
dV = dt + dS1 + 2
(dt) + 2
(dS 1 ) + 2
(dS2 )2
∂t ∂S1 2 ∂t 2 ∂S1 2 ∂S2
2 2 2
∂ V ∂ V ∂ V
+ dS1 dt + dS2 dt + dS1 dS2 + higher order terms
∂S1 ∂t ∂S2 ∂t ∂S1 ∂S2
dS1 = S1 µ1 dt + S1 σ1 dW1
dS1 = S2 µ2 dt + S2 σ2 dW2 ,
with:
p
W2 = ρW1 + 1 − ρ 2 W3
5
Since (dt)2 ' 0, dtdW1 ' 0, dtdW2 ' 0, (dW1 )2 ' dt, (dW2 )2 ' dt and
dW1 dW2 ' ρdt, we get:
dtdS1 ' 0
dtdS2 ' 0
Then dV becomes:
∂V ∂V ∂V ∂V
dV = σ1 S1
∂S1
dW1 + σ2 S2
∂S2
dW2 +
∂t
[
+ µ1 S 1
∂S1
+
∂V 1 ∂ 2V 1 ∂ 2V ∂ 2V
+µ2 S2 + σ12 S12 2 + σ22 S22 2 + ρσ1 σ2 S1 S2
∂S2 2 ∂S1 2 ∂S2 ∂S1 ∂S2
dt ]
2. The risk-free interest rate r, the assets volatilities σ1 and σ2 and the
correlation coefficient ρ are known functions of time over the life of the
option;
6
3. There are no transactions costs associated with hedging a portfolio;
Construct a portfolio of one option and −∆1 shares of asset 1 and −∆2
π = V − ∆1 S1 − ∆2 S2 .
Then:
dπ = dV − ∆1 dS1 − ∆2 dS2 .
dπ = dt [ ∂V∂t + µ S ∂S
1 1
1
∂V ∂V
− ∆ 1 µ1 S 1 + µ2 S 2
∂S2
− ∆2 µ2 S2 +
1 ∂ 2V 1 ∂2V
+ σ12 S12 2 + σ22 S22 2 + ρσ1 σ2 S1 S2
2 ∂S1 2 ∂S2
∂ 2V
∂S1 ∂S2
+]
∂V ∂V
+dW1 (σ1 S1 − ∆1 σ1 S1 ) + dW2 (σ2 S2 − ∆2 σ2 S2 ).
∂S1 ∂S2
∂V
∆1 =
∂S1
∂V
∆2 = .
∂S2
7
The result is a portfolio whose increment is wholly deterministic:
riskless assets would see a growth of rπdt in a time dt. So if the right-hand
side of the above equation were greater than this amount, an arbitrager could
portfolio. Conversely, if the right-hand side of the equation were less then
rπdt, then the arbitrager would short the portfolio and invest π in the bank.
Thus we have:
But:
∂V ∂V
π = V − ∆1 S1 − ∆2 S2 = V − S1 − S2 ,
∂S1 ∂S2
so we obtain:
∂V 1 ∂ 2V 1 ∂ 2V ∂ 2V
+ σ12 S12 2 + σ22 S22 2 + ρσ1 σ2 S1 S2 +
∂t 2 ∂S1 2 ∂S2 ∂S1 ∂S2
∂V ∂V
+rS1 + rS2 − rV = 0.
∂S1 ∂S2
8
1.7 Options on Dividend Paying Assets
The price of an option on assets that pay out dividends is affected by the
payments, so we must modify the Black-Scholes formula.
Consider a constant dividend yield, so in a time dt the asset pays out a
Arbitrage considerations show that in each time-step dt, the asset price
must fall by the amount of the dividend payment in addition to the usual
fluctuations. It follows that the random walk for the asset price is modified
to:
dS = σSdW + (µ − q)Sdt.
Since we receive qi · Si · dt for every asset of type i held and since we hold
−∆i shares of asset i (i = 1, 1), our portfolio changes to:
∂V 1 ∂ 2V 1 ∂ 2V ∂ 2V
+ σ12 S12 2 + σ22 S22 2 + ρσ1 σ2 S1 S2 +
∂t 2 ∂S1 2 ∂S2 ∂S1 ∂S2
∂V ∂V
+(r − q1 )S1 + (r − q2 )S2 − rV = 0.
∂S1 ∂S2
9
1.8 The Explicit Finite Difference Method
Rarely can we find closed form solutions for the values of options. Unless the
problem is very simple, we are going to have to solve a partial differential
equation numerically.
∂u ∂u
= .
∂τ ∂x2
∂u
The partial derivative ∂τ may be defined as a limit:
∂u u(x, τ + dτ ) − u(x, τ )
(x, τ ) = lim ,
∂τ dτ →0 dτ
∂ 2u
and the second partial derivative can be defined as:
∂x2
Next, we divide the x-axis into equally spaced nodes at distance dx apart,
and the τ -axis into equally spaced nodes at distance dτ apart. This di-
10
vides the (x, τ )-plane into a mesh, where the mesh points have the form
(ndx, mdτ ). We then concern ourselves only with the values of u(x, τ ) at the
mesh points(ndx, mdτ ).
We write U (n, m) = u(ndx, mdτ ). Using the above differences, the diffu-
dτ
where α = .
(dx)2
If we choose a constant x step dx, we cannot solve the problem for all
−∞ < x < ∞ without taking an infinite number of x-steps. We get around
−N dx ≤ x ≤ N dx,
11
When using numerical schemes for solving partial differential equations,
one needs to address three fundamental issues:
are trying to solve as the space and time steps tend to zero;
the numerical solution and the exact solution at a fixed point in the
domain of interest tends to zero uniformly as the space and time dis-
cretizations tend to zero.
one underlying asset, it can be shown that the system is stable if 0 < α ≤ 12 .
The stability condition puts severe constraints on the size of the time-
When dealing with options on two assets, the stability problem becomes
much more complicated. In one special, but very important case, when we
12
have a pure diffusion problem:
∂V ∂ 2V ∂ 2V
=a 2 +b 2,
∂τ ∂x1 ∂x2
dt dt 1
0≤a 2
+b 2
≤ .
(dx1 ) (dx2 ) 2
In the last chapter we will attempt to give some directions for overcoming
The basis of Monte-Carlo simulation is the strong law of large numbers, stat-
To simulate the final payoff, we first need to simulate a path for S(t), the
stock price process. We can use the following:
Algorithm:
13
2. Generate N independent random numbers Yi which are standard nor-
mally distributed;
W (0) = 0
µ ¶ µ ¶ r
T T T
W j = W (j − 1) + · Yj , j = 1, · · · , N
N N N
µ ¶ µ ¶
T T N
W (t) = W (j − 1) + t − (j − 1) · ·
N N T
· µ ¶ µ ¶¸ · ¶
T T T T
· W j − W (j − 1) , for t ∈ (j − 1) , j .
N N N N
1 2 )t
S(t) = S0 (t) · e(r− 2 σ · eσW (t) , t ∈ [0, T ];
14
Disadvantages: Even given high speed computers, the method is time-
consuming, as both n and N have to be very large to yield good esti-
mates for the option price.
15
CHAPTER 2
EUROPEAN LOOKBACK
PUT BASKET OPTION
WITH CONSTANT NUMBER
OF SHARES
2.1 Introduction
mum or minimum realized asset price over the life of an option. For example,
a lookback put has a payoff at expiry that is the difference between the max-
imum realized price and the spot price at expiry. This may be written as:
max(J − S, 0),
16
where J is the maximum realized price of the asset:
J = max S(τ ).
0≤τ ≤t
which the underlying assets have traded and also because by decreasing the
frequency at which the maximum and the minimum are measured, some
contracts become cheaper and therefore more appealing.
In this chapter, we analyze the lookback put option on a basket of two
and maximum realized basket price, but similarity reductions are going to
be used to obtain more efficient numerical solutions.
Consider two assets with S1 (t) and S2 (t) the prices at time t and α1 , α2 the
17
The lookback option is a path dependent option. Its value V is going to
be a function of S1 , S2 , time and the maximum realized basket price.
Therefore, the only random variables are S1 and S2 and the option price
must satisfy the Black-Scholes equation:
∂V 1 2 2 ∂2V ∂V
+ σ1 S1 2 + (r − q1 )S1 − rV +
∂t 2 ∂S1 ∂S1
1 ∂ 2V ∂ 2V ∂V
+ σ22 S22 2 + ρσ1 σ2 S1 S2 + (r − q2 )S2 =0
2 ∂S2 ∂S1 ∂S2 ∂S2
For arbitrage reasons, the realized option price must be continuous across
sampling dates, so if M is the maximum and t0 is a sampling time, we have:
V (S1 , S2 , M, t− +
0 ) = V (S1 , S2 , Max(M, α1 S1 + α2 S2 ), t0 ).
18
Then:
∂V ∂W
=M
∂t ∂t
∂V ∂V ∂s1 ∂W
= · =
∂S1 ∂s1 ∂S1 ∂s1
2 2
∂ V ∂ W 1
= ·
∂S1 ∂S2 ∂s1 ∂s2 M
2
∂ V ∂ 2W 1
= ·
∂S12 ∂s21 M
∂W 1 2 2 ∂ 2W ∂W
+ σ1 s1 2 + (r − q1 )s1 − rW +
∂t 2 ∂s1 ∂s1
1 ∂2W ∂2W ∂W
+ σ22 s22 2 + ρσ1 σ2 s1 s2 + (r − q2 )s2 =0
2 ∂s2 ∂s1 ∂s2 ∂s2
becomes:
V (S1 , S2 , M, t− +
0 ) = V (S1 , S2 , Max(M, α1 S1 + α2 S2 ), t0 ).
19
becomes:
W (s1 , s2 , t−
0 ) = max(1, α1 s1 + α2 s2 )·
s1 s2
·V ( , , t+ ).
max(1, α1 s1 + α2 s2 ) max(1, α1 s1 + α2 s2 ) 0
Boundary conditions:
∂W 1 2 2 ∂ 2W ∂W
+ σ2 s2 2 + (r − q2 )s2 − rW = 0.
∂t 2 ∂s2 ∂s2
∂W 1 ∂ 2W ∂W
+ σ12 s21 2 + (r − q1 )s1 − rW = 0.
∂t 2 ∂s1 ∂s1
∂W s1 ∂W s2
· + · ∼ 1 as s1 → ∞ and s2 → ∞.
∂s1 W ∂s2 W
20
• S1 → ∞ and S2 is finite. Then:
∂W s1
· ∼ 1.
∂s1 W
∂W s2
· ∼ 1.
∂s2 W
Thus, the strategy for valuing the lookback put option is as follows:
2. Apply the appropriate jump condition across the current sampling date
Method
21
spaced nodes at distance ds2 apart and the t-axis into equally spaced nodes
at distance dt apart.
s2 (j) = (j − 1)ds2 , 1 ≤ j ≤ J + 1
t(k) = (k − 1)dt, 1≤k ≤K +1
for example four times the value of the maximum realized basket price. So
4
0 ≤ s1 ≤ 4 and 0 ≤ s2 ≤ 4, which would imply ds1 = I
and ds2 = J4 .
For 1 ≤ i ≤ I + 1, 1 ≤ j ≤ J + 1, let:
A(i) = 1 2
σ
2 1
[(i − 1)ds1 ]2
B(i) = (r − q1 )(i − 1)ds1
C = −r
D(j) = 1 2
σ
2 2
[(j − 1)ds2 ]2
22
The explicit difference scheme for the PDE is:
Boundary conditions:
• s1 → 0 and s2 → 0. Then:
U (1, 1, k) = e−r(k−1)dt .
23
• s1 → 0. Then:
U (1, j, k) − U (1, j, k + 1)
+
· dt ¸
U (1, j + 1, k) − 2U (1, j, k) + U (1, j − 1, k)
+D(j) +
(ds2 )2
· ¸
U (1, j + 1, k) − U (1, j − 1, k)
+F (j) + CU (1, j, k) = 0 becomes:
2ds2
U (1, j, k + 1) U (1, j, k)
= +
· dt dt ¸
U (1, j + 1, k) − 2U (1, j, k) + U (1, j − 1, k)
+D(j) +
(ds2 )2
· ¸
U (1, j + 1, k) − U (1, j − 1, k)
+F (j) + CU (1, j, k).
2ds2
• s2 → 0. Then:
U (i, 1, k + 1) U (i, 1, k)
= +
· dt dt ¸
U (i + 1, 1, k) − 2U (i, 1, k) + U (i − 1, 1, k)
+A(i) +
(ds1 )2
· ¸
U (i + 1, 1, k) − U (i − 1, 1, k)
+B(i) + CU (i, 1, k).
2ds1
• s1 → ∞ and s2 → ∞. Then:
∂W ∂W
W = s1 + s2
∂s1 ∂s2
U (I + 1, J + 1, k) = I [U (I + 1, J + 1, k) − U (I, J + 1, k)] +
+J [(U (I + 1, J + 1, k) − U (I + 1, J, k)]
I J
U (I + 1, J + 1, k) = U (I, J + 1, k) + U (I + 1, J, k).
I +J −1 I +J −1
24
• s1 → ∞. Then:
∂W
W = s1
∂s1
U (I + 1, j, k) = I [U (I + 1, j, k) − U (I, j, k)]
I
U (I + 1, j, k) = U (I, j, k).
I −1
• s2 → ∞. Then:
J
U (i, J + 1, k) = U (i, J, k).
J −1
U (i − 1, j − 1, k) U (i, j − 1, k) U (i + 1, j − 1, k)
• • •
U (i − 1, j, k) U (i, j, k) U (i + 1, j, k)
• • •
U (i − 1, j + 1, k) U (i, j + 1, k) U (i + 1, j + 1, k)
• • •
U (1, 1, k).
25
4. Boundary condition at s1 = 0: for 2 ≤ j ≤ J, knowing U (1, j − 1, k),
U (1, j, k) and U (1, j + 1, k), we can find U (1, j, k + 1).
U (I + 1, j, k) for 1 ≤ j ≤ J.
U (i, J + 1, k) for 1 ≤ i ≤ I.
i (1, 1, 1)
26
2.4 Numerical Examples
Table 2.2 shows the value of the option using different sets of sampling times.
In each case the values shown are at 10 years before expiry with zero dividend
yield for both assets and r = .1, σ1 = σ2 = .2, ρ = .1, α1 = .3, α2 = .7 and
Basketprice
M ax
Case 1 Case 2 Case 3 Case 4
For example, if the current basket price is 90 and the current maximum
is 100, then we must search along the row starting with .9. The value of the
27
option will be, under sampling strategy 2, 100 × .031 = 3.1. Observe that the
option price decreases as the number of samples decreases (from Case 1 to
Case 3). This is financially obvious, since the lower the number of samples
is, the lower the final payoff is expected to be. Decreasing the frequency of
since one of the main commercial criticisms of lookback options is that they
are expensive.
Note also that the option price reaches a minimum close to the point
where the basket price is equal to the current maximum. The option delta
can become positive, since it is beneficial for the holder of the option if the
basket price rises just before a sampling date and then falls.
28
CHAPTER 3
EUROPEAN LOOKBACK
PUT BASKET OPTION
WITH REBALANCING
3.1 Introduction
pling date, such that the percentages of the money invested in each stock are
constant.
29
We start with an initial amount D, and a constant vector:
with
φ(1) + φ(2) + · · · + φ(n) = 1,
At time zero, the number of shares of each stock in the basket is:
µ ¶
D · φ(1) D · φ(2) D · φ(n)
a0 = , ,··· , .
S1 (0) S2 (0) Sn (0)
then we compute the new vector containing the number of shares of each
stock in the basket:
µ ¶
S · φ(1) S · φ(2) S · φ(n)
a1 = , ,··· , ,
S1 (t1 ) S2 (t1 ) Sn (t1 )
30
The finite-difference method is much more complicated to implement,
therefore we choose to use the Monte-Carlo simulation method to compute
the value of this option.
The Monte Carlo simulation method uses the risk neutral valuation result.
The expected payoff in a risk-neutral world is calculated using a sampling
The payoff B, for a lookback put basket option, is a function of the price pro-
cesses S1 (t), S2 (t), · · · , Sn (t). Thus, to simulate B, we first have to simulate
Suppose that the process followed by ln(Si (t)) in a risk-neutral world is:
µ ¶
σ2
d(ln(Si (t)) = r− i dt + σi dWi .
2
µ µ ¶¶ µ µ ¶¶ µ ¶ r
T T σ2 T T
ln Si k − ln Si (k − 1) = r− i + σi xi .
N N 2 N N
31
or, equivalently:
µ ¶ µ ¶ õ ¶ r !
T T σ2 T T
Si k = Si (k − 1) exp r− i + σi xi .
N N 2 N N
corr (X(i, k), X(j, k)) = corr (Si , Sj ) for all 1 ≤ i, j ≤ n and 1 ≤ k ≤ N.
32
with:
ε(1) − 1 = 0
(ε(2))2 + (ε(3))2 − 1 = 0
..
.
· µ ¶¸2 · µ ¶¸2 · µ ¶¸2
n(n − 1) n(n − 1) n(n − 1)
ε +1 + ε +2 + ··· + ε +n − 1 = 0.
2 2 2
µ ¶ µ ¶ µ ¶ µ ¶
i(i − 1) j(j − 1) i(i − 1) j(j − 1)
ε +1 ε + 1 + ··· + ε +i ε +i −
2 2 2 2
−corr (Si , Sj ) = 0 for i = 1, n − 1, j = i + 1, n.
This system can be solved by using the MATLAB function fsolve, but
first we have to transform the correlation matrix:
33
n(n−1)
with 2 elements.
µ ¶ µ ¶ µ ¶ µ ¶
i(i − 1) j(j − 1) i(i − 1) j(j − 1)
ε +1 ε + 1 + ··· + ε +i ε +i −
2 2 2 2
µ ¶
(i − 1)i
−corrvec n(i − 1) − + i + j − 1 = 0 for i = 1, n − 1, j = i + 1, n.
2
After solving the system, we find X(i, k) and then simulate the paths of
µ ¶ µ ¶ ·µ ¶ ¸
T T 1 2 T
Si k = Si (k − 1) · exp r − σ (i) · ·
N N 2 N
" r #
T
· exp σ(i) · · X(i, k) ,
N
for 1 ≤ k ≤ N .
Next, we compute the matrix of the number of shares of each stock in the
basket at each sampling time with the algorithm described at the beginning
of this chapter.
Max = max [a(1, t)S1 (t) + a(2, t)S2 (t) + · · · + a(n, t)Sn (t)] ;
t∈{t1 ,t2 ,··· ,tfinal }
34
where {t1 , t2 , · · · , tfinal } is the set of sampling times.
The payoff becomes:
V = exp(−rT ) · B.
The number of simulation runs carried out depends on the accuracy re-
quired. If M is the number of runs and ω is the standard deviation of the
values of the option calculated from the simulation runs, the standard error
ω
of the estimate of V is √ .
M
The advantage of the Monte-Carlo simulation method is that is efficient
when there are several variables involved. This is true since the time taken
out to carry out a Monte-Carlo simulation increases approximately linearly
with the number of variables, whereas the time required for most other pro-
cedures increases exponentially with the number of variables.
35
3.3 Numerical Example
Consider 4 assets with the standard deviation vector σ and the percent-
age vector φ given by:
.005 .1
.052 .2
σ= , φ= .
.006 .1
.22 .6
Time to maturity: T = 1.
36
Using the MATLAB function lkb.m, with source code in Appendix B,
we obtain the following values for the lookback put basket option with re-
balancing:
No sampling 3.8
37
CHAPTER 4
GUARANTEED MINIMUM
DEATH BENEFITS
4.1 Introduction
The multiple decrement model provides a framework for studying many fi-
nancial security systems. For example, life insurance policies frequently pro-
vide for special benefits if death occurs by accidental means or if the insured
becomes disabled. Another major application is in pension plans. A plan,
38
In general, actuarial applications of multiple decrement models arise when
the amount of benefit payment depends on the mode of exit from the group
of active lives.
The Individual Variable Annuity marketplace is highly competitive. One
Return of premium (ROP): the death benefit is the greatest of the ac-
Reset: the death benefit is the greatest of the account value, the net pre-
mium paid and the highest account value on the last five anniversaries;
Ratchet: the death benefit is the greatest of the account value, the net
premium paid and the highest account value on all past anniversaries.
In this chapter, we will analyze these death benefits using the results
obtained in Chapter 2 and Chapter 3.
39
and set:
s(x) = 1 − FX (x) = Pr(X > x), for x ≥ 0.
s(x) is called the survival function. For any positive x, s(x) is the probability
that a newborn will attain age x.
The symbol (x) is used to denote a life-age-x. The future lifetime of (x),
The symbol t qx is interpreted as the probability that (x) will die within t
years and t px is the probability that (x) will attain age x + t.
fX (x)
1 − FX (x)
40
has a conditional probability density interpretation. For each age x, the
expression gives the value of the conditional probability density function of
X at exact age x, given the survival to that age. It will be denoted by µ(x)
and referred to as the force of mortality. We can write:
A widely used assumption for fractional ages is linear interpolation on log s(x+
t):
log s(x + t) = (1 − t) log s(x) + t log s(x + 1), for 0 ≤ t < 1.
t px = (px )t ,
and:
µ(x + t) = µ(x) = − log px , for 0 ≤ t < 1.
41
the marginal probability density function of J by fJ (j) and the marginal
probability density function of T by fT (t).
m
X
fJ (j) = 1
j=1
Z ∞
fT (t)dt = 1.
0
Z t
Pr [(0 < T ≤ t) ∩ (J = j)] = fT,J (s, j)ds =t qx(j) ,
0
and the probability of decrement due to all causes between a and b is:
m Z
X b
Pr [a < T ≤ b] = fT,J (t, j)dt.
j=1 a
The probability of decrement due to cause j at any time t in the future is:
Z ∞
fJ (j) = fT,J (s, j)ds =∞ qx(j) , j = 1, 2, · · · , m.
0
For t ≥ 0, the marginal probability density function for T , fT (t), and the
m
X
fT (t) = fT,J (t, j),
j=1
Z t
FT (t) = fT (s)ds.
0
42
Using the superscript (τ ) to indicate that a function refers to all causes, or
total force of decrement, we obtain:
(τ )
t qx = Pr(T ≤ t) = FT (t),
(τ )
t px = Pr(T > t) = 1 −t qx(τ ) ,
fT (t) d
µ(τ
x =
)
= − logt p(τ
x ,
)
1 − FT (t) dt
fT,J (t, j)dt = Pr[T > t] · Pr[(t < T ≤ t + dt) ∩ (J = j)|T > t].
This suggests the following definition for the force of decrement due to cause
j:
d (j)
fT,J (t, j) fT,J (t, j) q
dt t x
µ(j)
x (t) = = (τ )
= (τ )
.
1 − FT (t) t px t px
So:
Restated, the last equality implies that the probability of decrement between
(τ )
t and t + dt due to cause j is equal to the probability t px that (x) remains
(j)
in the given status until time t times the conditional probability µx (t) that
decrement occurs between t and t+dt due to cause j, given that the decrement
has not occurred before time t.
Also we have:
Z t Z tX
m
(τ )
t qx = fT (s)ds = fT,J (s, j)ds =
0 0 j=1
m Z
X t m
X
(j)
= fT,J (s, j)ds = t qx .
j=1 0 j=1
43
and it follows that:
m
X
µ(τ )
x (t) = µ(j)
x (t),
j=1
that is, the total force of decrement is the sum of the forces of decrement due
to all causes.
it is possible to define a single decrement model that depends only on the par-
· Z t ¸
0(j) (j)
t px = exp − µx (s)ds ,
0
0(j)
t qx = 1 −t p0(j)
x (absolute rate of decrement).
Note that:
m
Y
(τ ) 0(j)
t px = t px and
j=1
0(j)
t px ≥t p(τ
x .
)
0(j)
The magnitude of other forces of decrement can cause t px to be considerably
(τ )
greater than t px .
44
For single decrement models, constant-force assumption implies:
µ(j) (j)
x (t) = µx (0),
µ(τ ) (τ )
x (t) = µx (0),
0(j)
¡ 0(j) ¢t
t px (t) = px ,
for 0 ≤ t < 1.
But this is exactly the payoff for a vanilla put option with the exercise price
equal to the premium. The present value of the death benefit paid is equal
greatest of the account value, the net premium paid and the highest account
45
The amount that the insurance company has to cover is:
This is the same as the payoff for a lookback put option with discretely
measured maximum, at the beginning of the period and every year for the
past five years.
The third basic GMDB is Ratchet, for which the death benefit is the
greatest of the account value, the net premium paid and the highest account
value on all past anniversaries.
This is the same as the payoff for a lookback put option with maximum
When the amount of benefit payment depends on the mode of exit from
46
Let B(j, t) be the present value of a benefit at age (x + t) incurred by a
decrement at that age by cause j. Then the actuarial present value of the
benefits that occur between times t1 and t2 is given by:
Pm R t2 (τ ) (j)
A= j=1 B(j, t) t px µx dt.
t1 | {z } | {z }
present value
function of the fT,J (t, j)
benefit payment
For example, in the case of death benefits, three causes of decrement are
usually considered:
(1) Mortality;
(2) Lapse;
(3) Partial withdrawal.
A benefit is paid only at death. The present value function of the benefit
payment,B(1, t), is equal to the price of an option (vanilla put for ROP and
Below is a computation of the present value of the claims over the evalu-
ation period [t1 , t2 ], for 1 ≤ j ≤ m causes of decrement.
We have, for t2 ≥ t1 + 1:
m Z
X t2
A= B(t, j)t p(τ ) (j)
x µx (t)dt =
j=1 t1
x µx (t)dt).
B(t, j)t p(τ ) (j)
+
[t2 ]
47
Define:
Z [t1 ]+k+1
A(k, j) = B(t, j)t p(τ ) (j)
x µx (t)dt, for 0 ≤ k ≤ [t2 ] − [t1 ] − 1,
[t1 ]+k
Z t1
A(t1 , j) = B(t, j)t p(τ ) (j)
x µx (t)dt,
[t1 ]
Z t2
A(t2 , j) = B(t, j)t p(τ ) (j)
x µx (t)dt.
[t2 ]
We have:
Z 1
(τ ) (τ ) (j)
A(k, j) = [t1 ]+k px B([t1 ] + k + t, j) t px+[t1 ]+k µx+[t1 ]+k (t)dt,
0
Z t1 −[t1 ]
(τ ) (τ ) (j)
A(t1 , j) = [t1 ] px B([t1 ] + t, j) t px+[t1 ] µx+[t1 ] (t)dt,
0
Z t2 −[t2 ]
(τ ) (τ ) (j)
A(t2 , j) = [t2 ] px B([t2 ] + t, j) t px+[t2 ] µx+[t2 ] (t)dt.
0
¡ ¢
µx(j) (t) = − ln p0(j)x
0(j)
¡ ¢t
t px = p0(j)
x ,
for 0 ≤ t < 1.
48
Then A(k, j), A(t1 , j) and A(t2 , j) become:
m ³
Y ´
0(j) 0(j)
A(k, j) = p0(j)
x px+1 · · · px+[t1 ]+k−1 ·
j=1
Z 1 m ³
Y ´t h ³ ´i
0(j) 0(j)
· B([t1 ] + k + t, j) · px+[t1 ]+k · − ln px+[t1 ]+k dt,
0 j=1
Y³
m
0(j) 0(j)
´
A(t1 , j) = p0(j)
x px+1 · · · px+[t1 ]−1 ·
j=1
Z 1 Ym ³ ´t h ³ ´i
0(j) 0(j)
· B([t1 ] + t, j) · px+[t1 ] · − ln px+[t1 ] dt,
0 j=1
Y³
m
0(j) 0(j)
´
A(t2 , j) = p0(j)
x px+1 · · · px+[t2 ]−1 ·
j=1
Z 1 Ym ³ ´t h ³ ´i
0(j) 0(j)
· B([t2 ] + t, j) · px+[t2 ] · − ln px+[t2 ] dt.
0 j=1
We obtain:
m [t2 ]−[t1 ]−1
X X
A= A(k, j) − A(t1 , k) + A(t2 , j) .
j=1 k=0
49
Standard deviation vector: σ1 = σ2 = .22;
No sampling 10.2379
Therefore, for $1000 invested today in the basket of two assets described
above, the expected value of the claims covered by the insurance company
for the next 10 years is about $27 in the case of Ratchet and $10 for ROP.
50
CHAPTER 5
CONCLUSION AND
FURTHER RESEARCH
1
1. Introduce an “intermediate” value V (i, j, k+ 2 );
1
2. Solve from time-step k to the intermediate time-step k+ 2 using explicit
1
3. Having found the intermediate value V (i, j, k+ 2 ), step forward to time-
step k + 1 using implicit differences in S1 and explicit differences in S2 .
51
This method is stable for all time-steps and the error is O(dt2 , dS12 , dS22 ).
∂2V
The only problem is that the cross derivative term ∂S ∂S causes lots of
1 2
problems in the basic implementation of the ADI, and the Black-Scholes
equation cannot be changed into the equation:
∂V ∂ 2V ∂ 2V
+ a(s1 , s2 , t) 2 + b(s1 , s2 , t) 2 = 0.
∂t ∂s1 ∂s2
s1 = x1 S1 + x2 S2
s2 = y1 S1 + y2 S2
We are still looking for an implicit finite differences method that can be
easy implemented in the case of a basket option.
The finite differences methods are suitable for solving financial problems
with two or three random factors. For more than that number, the Monte
Carlo simulation becomes a better method, which also works in the case of
complex payoffs.
The discrete lookback options are more difficult to hedge than regular
options because the delta of the option is discontinuous at all sampling times.
52
is: if two portfolios are worth the same on a certain boundary, they are worth
the same at all interior points of the boundary.
53
APPENDIX A
THE SOURCE CODE OF
LOOKBASKET.M
function
f=lookbackbasket(S1,S2,M,sigma1,sigma2,r,q1,q2,ro,T,alpha1,
alpha2,tbeg,tinc)
54
% q2 Divided rate of the second asset
% ro Correlation coefficient between the two assets
% T Time to maturity
% alpha1 Number of shares of the first asset in the basket
n=floor((T-tbeg)/tinc)+2;
a=zeros(I+1);
for i=1:(I+1)
a(i)=0.5*(sigma1^2)*((i-1)*ds1)^2;
end;
b=zeros(I+1);
for i=1:(I+1)
b(i)=(r-q1)*((i-1)*ds1);
end;
c=-r;
55
d=zeros(J+1);
for j=1:(J+1)
d(j)=0.5*(sigma2^2)*(((j-1)*ds2)^2);
end;
e=zeros(I+1,J+1);
for i=1:(I+1)
for j=1:(J+1)
e(i,j)=ro*sigma1*sigma2*((i-1)*ds1)*((j-1)*ds2);
end;
end;
f=zeros(J+1);
for j=1:(J+1)
f(j)=(r-q2)*(j-1)*ds2;
end;
v=zeros(I+1,J+1,K+1);
for i=1:(n+1)
w(i)=(i-1)*p;
end;
56
slin1=linspace(0,4,I+1);
slin2=linspace(0,4,J+1);
for k=2:(K+1)
v(1,1,k)=exp(-r*(k-1)*dT);
end;
z=v(1,1,K+1);
%end rule 3
for j=1:(J+1)
if (l==1)
v(i,j,1)=max(1-(alpha1*(i-1)*ds1+alpha2*(j-1)*ds2),0);
end;
if (l>1)
mx=max(1,alpha1*(i-1)*ds1+alpha2*(j-1)*ds2);
v(i,j,w(l)+1)=mx*interp2(slin1,slin2,u(:,:,w(l)+1),
(i-1)*ds1/mx,(j-1)*ds2/mx);
57
end;
end;
end;
%end rule 2
for k=(w(l)+1):w(l+1)
for i=2:I
for j=2:J
sum1=a(i)*(v(i+1,j,k)-2*v(i,j,k)+v(i-1,j,k))/(ds1^2);
sum2=b(i)*(v(i+1,j,k)-v(i-1,j,k))/(2*ds1);
sum3=c*v(i,j,k);
sum4=d(j)*(v(i,j+1,k)-2*v(i,j,k)+v(i,j-1,k))/(ds2^2);
sum5=e(i,j)*(v(i+1,j+1,k)-v(i+1,j-1,k)-v(i-1,j+1,k)+
v(i-1,j-1,k))/(4*ds1*ds2);
sum6=f(j)*(v(i,j+1,k)-v(i,j-1,k))/(2*ds2);
v(i,j,k+1)=dT*(v(i,j,k)/dT+sum1+sum2+sum3+sum4+sum5+sum6);
end;
end;
%end rule 1
58
for j=2:J
sum1=v(1,j,k)/dT;
sum2=d(j)*(v(1,j+1,k)-2*v(1,j,k)+v(1,j-1,k))/(ds2^2);
sum3=f(j)*(v(1,j+1,k)-v(1,j-1,k))/(2*ds2);
sum4=c*v(1,j,k);
v(1,j,k+1)=dT*(sum1+sum2+sum3+sum4);
end;
%end rule 4
for i=2:I
sum1=v(i,1,k)/dT;
sum2=a(i)*(v(i+1,1,k)-2*v(i,1,k)+v(i-1,1,k))/(ds1^2);
sum3=b(i)*(v(i+1,1,k)-v(i-1,1,k))/(2*ds1);
sum4=c*v(i,1,k);
v(i,1,k+1)=dT*(sum1+sum2+sum3+sum4);
end;
%end rule 5
for j=1:J
v(I+1,j,k+1)=I/(I-1)*v(I,j,k+1);
end;
%end rule 6
59
%rule 7: Boundary condition at s2=max
for i=1:I
v(i,J+1,k+1)=J/(J-1)*v(i,J,k+1);
end;
%end rule 7
%end rule 8
end;
u=zeros(I+1,J+1,K+1);
u=v;
end;
z=v(:,:,K+1);
f=interp2(slin1,slin2,z,S1/M,S2/M)*M;
60
APPENDIX B
THE SOURCE CODE OF
LKB.M
function f=lkb(stocks, sigma, T,inam,fi,t1,dt,r,corr)
global S;
global T;
61
global N;
global n;
global corr;
e0=ones(floor(n*(n+1)/2),1);
e=fsolve(@randomk,e0);
N=10; n=length(stocks);
end;
end;
for k=2:(N+1)
S(i,k)=S(i,k-1)*exp((r-(sigma(i))^2/2)*T/N+sigma(i)*
sqrt(T/N)*X(i,k-1));
end;
end;
maux=(T-t1)/dt;
62
if (floor(maux)==maux) M=floor(maux)-1;
end;
if (floor(maux)~=maux) M=floor(maux);
end;
tau(1)=t1;
for j=2:(M+1)
tau(j)=t1+(j-1)*dt;
end;
for i=1:n
a(i,1)=inam*fi(i)/S(i,1);
for k=2:(M+2)
s=0;
for j=1:n
s=s+a(j,k-1)*SS(j,tau(k-1));
end; a(i,k)=fi(i)*s/SS(i,tau(k-1));
end;
end;
for j=1:(M+1)
s=0;
for i=1:n
s=s+a(i,j+1)*SS(i,tau(j));
63
end;
v(j)=s;
end;
MAX=max(v);
s=0;
for i=1:n
s=s+a(i,M+2)*SS(i,T);
end;
payoff=max(MAX-s,0);
optionvalue=exp(-r*T)*payoff; f=optionvalue;
function f1=SS(i,t)
global S;
global T;
global N;
m=floor(t*N/T);
f1=interp1(linspace(0,T,N+1),S(i,:),t);
function f2=randomk(e) g
global n;
global corr;
64
g=[];
for i=1:n
s=0;
for j=1:i
s=s+(e(floor(i*(i-1)/2+j)))^2;
end;
gaux=s-1;
g=[g;gaux];
end;
for i=1:(n-1)
for j=(i+1):n
s=0;
for k=1:i
ii=floor(i*(i-1)/2);
ij=floor(j*(j-1)/2);
s=s+e(ii+k)*e(ij+k);
end;
gaux=s-corr(n*(i-1)-floor((i-1)*i/2)+j);
g=[g;gaux];
end;
end;
f2=g;
65
APPENDIX C
THE SOURCE CODE OF PV.M
function res=pv(t1,t2)
%m Mortality vector
%l Lapse vector
%pw Partial withdrawal vector
global k;
global t1;
global t2;
global px1;
global px2;
global px3;
66
N=10;
m=[0.000001 0.000001 0.000001 0.000001 0.000324 0.000301
0.000286 0.000328 0.000362 0.00039];
for i=1:length(m)
px1(i)=1-m(i);
end;
for i=1:length(l)
px2(i)=1-l(i);
end;
for i=1:length(pw)
px3(i)=1-pw(i);
end;
A=zeros(floor(t2)-floor(t1));
for k=0:(floor(t2)-floor(t1)-1)
P1=1;
67
for i=1:(floor(t1)+k)
P1=P1*px1(i);
end;
P2=1;
for i=1:(floor(t1)+k)
P2=P2*px2(i);
end;
P3=1;
for i=1:(floor(t1)+k)
P3=P3*px3(i);
end;
A(k+1)=P1*P2*P3*quad(@intfun,0,1);
end;
P1=1;
for i=1:floor(t1)
P1=P1*px1(i);
end;
P2=1;
for i=1:floor(t1)
P2=P2*px2(i);
end;
68
P3=1;
for i=1:floor(t1)
P3=P3*px3(i);
end;
k=0;
At1=P1*P2*P3*quad(@intfun,0,t1-floor(t1));
P1=1;
for i=1:floor(t2)
P1=P1*px1(i);
end;
P2=1;
for i=1:floor(t2)
P2=P2*px2(i);
end;
P3=1;
for i=1:floor(t2)
P3=P3*px3(i);
69
end;
k=floor(t2)-floor(t1);
At2=P1*P2*P3*quad(@intfun,0,t2-floor(t2));
s=0;
for i=1:(floor(t2)-floor(t1))
s=s+A(i);
end;
res=s-At1+At2;
function fp=intfun(t)
global k;
global t1;
global t2;
global px1;
global px2;
global px3;
fp=[];
for i=1:length(t)
70
c1=lookbackbasket(S1,S2,M,sigma1,sigma2,r,q1,q2,ro,floor(t1)
+k+t(i),alpha1,alpha2,tbeg,tinc);
c2=(px1(floor(t1)+k+1)*px2(floor(t1)+k+1)*px3(floor(t1)+k+1))
^t(i); c3=-log(px1(floor(t1)+k+1));
c=c1*c2*c3;
fp=[fp c];
end;
71
BIBLIOGRAPHY
[1] Newton Bowers, Hans Gerber, James Hickman, Donald Jones, and Cecil
Nesbitt. Actuarial Mathematics. The Society of Actuaries, 1997.
[2] Damiano Brigo and Fabio Mercurio. Interest Rate Models. Springer-
Verlag, 2001.
[3] John C. Hull. Options, Futures and Other Derivatives. Prentice Hall,
2003.
[4] Ralf Korn and Elke Korn. Option Pricing and Portfolio Optimization.
Prentice Hall, 2001.
[7] Nassim Taleb. Dynamic Hedging. John Wiley and Sons, Inc., 1997.
[8] Domingo Tavella and Curt Randall. Pricing Financial Instruments. John
Wiley and Sons, Inc., 2000.
[9] Paul Wilmott, Jeff Dewynne, and Sam Howison. Option Pricing. Oxford
Financial Press, 1993.
72