Free Essay

Lognormal Stock-Price Models

In: Science

Submitted By a6syed
Words 5362
Pages 22
Making sense of . . .

LogNormal stock-price models in Exams MFE/3 and C/4
James W. Daniel Austin Actuarial Seminars http://www.actuarialseminars.com June 26, 2008

c Copyright 2007 by James W. Daniel; reproduction in whole or in part without the express permission of the author is forbidden.

Foreword
This document briefly describes the ideas behind the use of LogNormal models for stock prices in some of the material for Exams MFE and C of the Society of Actuaries and Exams 3 and 4 of the Casualty Actuarial Society. Not a traditional exam-prep study manual, it concentrates on explaining key ideas so that you can then understand the details presented in the textbooks or study manuals. It can be especially useful to anyone taking Exam C/4 without having studied the material for Exam MFE/3.

2

Chapter 1

LogNormal stock-price models
1.1 Why LogNormal models?

Why learn about and use LogNormal models for stock prices? I could answer “Because it’s on the exam syllabi” or “Why not?”, but that wouldn’t be helpful. Instead, I’ll take a little space to motivate this. Suppose that the price of a stock or other asset at time 0 is known to be S(0) and we want to model its future price S(10) at time 10—note that some texts use the notation S0 and S10 instead. Let’s break the time interval from 0 to 10 into 10,000 pieces of length 0.001, and let’s let Sk stand for S(0.001k), the price at time 0.001k. I know the price S0 = S(0) and want to model the price S10000 = S(10). I can write (1.1) S(10) = S10000 = S2 S1 S10000 S9999 ··· S0 . S9999 S9998 S1 S0

Now suppose that the ratios Rk = SSk that appear in Equation 1.1 that represent the growth factors k−1 in price over each interval of length 0.001 are random variables, and—to get a simple model—are all independent of one another. Then Equation 1.1 writes S(10) as a product of a large number of independent random variables Rk . You know from probability that the sum of a large number of random variables Wk can, under reasonable hypotheses, be approximated well by a Normal random variable with the same mean and variance as the sum. Unfortunately Equation 1.1 involves a product, not a sum. But if we take the natural log of both sides, we get (1.2) ln S10000 = ln R10000 + ln R9999 + · · · + ln R2 + ln R1 + ln S0 .

This equation represents ln S(10) as ln S0 = ln S(0) plus the sum of a large number of independent random variables. Under reasonable hypotheses such as that all the ln Rk have the same probability distribution with positive variance, this implies that ln S(10) is well approximated by ln S(0) plus a Normal random variable with mean and variance each 10000 times that of a typical ln Rk . Note that if I took this same approach on the interval from time 0 to time 20, the mean and variance of the Normal would be 20000 times that of ln Rk —when I doubled the length of the time interval, I doubled the mean and variance of the Normal. That is, the mean and variance of the Normal grow linearly with the length of the time interval. So it seems that a plausible model for the price S(t) of a stock at time t is that ln S(t) should be ln S(0) plus a Normal random variable with mean and variance both proportional to t: (1.3) ln S(t) = ln S(0) + N (µt, σ 2 t), 3

where N (m, s2 ) denotes a Normal random variable with mean m and variance s2 . Exponentiating both sides of Equation 1.3 gives (1.4) S(t) = S(0)eN (µt,σ
2

t)

,

which models S(t) as S(0) times a LogNormal random variable with parameters µt and σ 2 t. This indicates that LogNormal models (with linearly growing parameters) are plausible models for the price of stocks and other assets. More precisely, the LogNormal models are examples of so-called Geometric Brownian Motion; the remaining sections of this note explain in more detail the precise nature of these models and how to calculate with them.

1.2

The first step—Standard Brownian motion B

Recall that for SoA Exam MLC or CAS Exam 3 you learned about Poisson Processes, a particular type of stochastic process—that is, a collection of random variables—that had independent increments. You don’t need to remember about Poisson Processes, just stochastic processes with independent increments. To be cautious, I’ll review that for you. A stochastic process X for t ≥ 0 is such that, for each value of t ≥ 0, X(t) is a random variable. For example, X(t) might represent the number of insurance claims filed with a particular insurance company by time t, or the value of all those claims, or the interest rate at time t, or, well, whatever entity you might want to model. To understand the Standard Brownian Motion stochastic process B, you need to understand the meaning and probability behavior of the increment B(t + h) − B(t) from time t to time t + h, where h > 0 and of course t ≥ 0. The increment is simply the change in the value of B from time t to time t + h. Note that—since you’ll soon see that B(0) = 0—the value B(t) = B(t) − B(0) itself can be viewed as an increment, namely from time 0 to time t. A fundamental property of Standard Brownian Motion processes is that increments on nonoverlapping time intervals are independent of one another as random variables—stated intuitively, knowing something about the change in B over one interval gives you no information about the change over a non-overlapping interval. But notice the important modifier “non-overlapping”. While the increments B(5.2) − B(3.1) and B(2.7) − B(1.6) are independent, B(5.2) − B(3.1) and B(4.3) − B(3.7) are not independent since the time intervals overlap. Similarly, the increments B(5.2)−B(3.1) and B(3.8) − B(1.5) are not independent since the time intervals overlap. Note, however, that B(5.2) − B(3.1) and B(3.1) − B(2.4) are independent since the intervals do not overlap; that is, touching at an endpoint is “OK”—time intervals that touch only at an endpoint do have independent increments. KEY ⇒ Definition 1.5 (Standard Brownian Motion) Standard Brownian Motion B has the following properties: 1. B is a stochastic process with B(0) = 0; 2. B has independent increments—any set of increments B(tj + hj ) − B(tj ) for j = 1, 2, . . . , n is independent, provided that the time intervals (tj , tj + hj ] are non-overlapping (touching only at an endpoint is OK); 3. for all t ≥ 0 and h > 0, the increment B(t + h) − B(t) is a Normal random variable with mean 0 and variance h. Let’s look at an example of how the properties of Standard Brownian Motion are used, especially that of independent increments. 4

Example 1.6 Suppose that B is a Standard Brownian Motion process. Also suppose that you’ve observed B(1.21) = 3.4. [By the way, this value is surprisingly far from its mean, since B(1.21) = B(1.21) − B(0) is a Normal random variable with mean 0 and variance 1.21, so standard deviation 1.1, revealing that B(1.21) is over three standard deviations away from its mean.] Taking this observation of B(1.21) into account, you want to understand the probability behavior of B(1.85). You might (momentarily) wonder whether you should expect B to continue to be surprisingly far from its mean. But then you remember that Standard Brownian Motion has independent increments, so knowing what happened from time 0 to time 1.21 should have no impact on what happens from time 1.21 to time 1.85. More precisely, write B(1.85) as B(1.85) = [B(1.85) − B(1.21)] + [B(1.21) − B(0)]. Given the observation that B(1.21) = 3.4, you know that the second term in square brackets [ ] equals 3.4. Because of independent increments, the first term in square brackets (given the observation) is simply a Normal random variable—call it W —with mean 0 and variance 1.85 − 1.21 = 0.64. Thus, given the observation, the distribution of B(1.85) is given by [B(1.85) B(1.21) = 3.4] ∼ W + 3.4, with W ∼ N (0, 0.64) a Normal random variable with mean 0 and variance 0.64 (and with ∼ meaning “has the same probability distribution as”). This makes it easy to translate probability questions about B(1.85), given B(1.21) = 3.4, to questions about W . For example, for the mean E[B(1.85) B(1.21) = 3.4] = E[W + 3.4] = E[W ] + 3.4 = 0 + 3.4 = 3.4, and for the variance Var[B(1.85) B(1.21) = 3.4] = Var[W + 3.4] = Var[W ] = 0.64; the equality of Var[W + 3.4] and Var[W ] in the above followed from the fact that adding a constant to a random variable shifts all its random values by the same amount but not how those values vary among themselves. You compute probabilities just as easily; for example, Pr[B(1.85) ≤ 2.6 B(1.21) = 3.4] = = Pr[W + 3.4 ≤ 2.6] = Pr[W ≤ −0.8] = Pr[N (0, 0.64) ≤ −0.8]

Pr[N (0, 0.64)/0.8 ≤ −1] = Pr[N (0, 1) ≤ −1] ≈ 0.1587,

using a table of probabilities for the standard Normal distribution. ¶ On the next page you’ll find the graph of a simulated approximation to a Standard Brownian Motion for 0 ≤ t ≤ 1. 5

Standard Brownian Motion

(1.7)

1.3

The second step—Arithmetic Brownian Motion A

Standard Brownian Motion always has mean 0 but has linearly growing variance equal to t. Arithmetic Brownian Motion—also often called Brownian Motion with Drift—allows for linearly growing mean and for the variance to be proportional to t rather than simply equal to t. Definition 1.8 (Arithmetic Brownian Motion) Arithmetic Brownian Motion A with drift parameter µ, volatility parameter σ, and initial value A0 is a stochastic process A with A(t) = A0 + µt + σB(t), where B is Standard Brownian Motion. It’s easy to see how increments of A behave, since they are completely determined by corresponding increments of B. That is, A(t + h) − A(t) = [A0 + µ(t + h) + σB(t + h)] − [A0 + µt + σB(t)] = µh + σ[B(t + h) − B(t)]. Since B(t + h) − B(t) ∼ N (0, h), this makes A(t + h) − A(t) ∼ N (µh, σ 2 h). Relating increments in A to increments in B in this manner and using Key Definition 1.5 on Standard Brownian Motion leads to the following key facts—which you should compare with the correspondingly numbered facts about Standard Brownian Motion in Key Definition 1.5. KEY ⇒ Fact 1.9 (Arithmetic Brownian Motion properties) Suppose that A is an Arithmetic Brownian Motion with drift parameter µ, volatility parameter σ, and initial value A0 . Then: 1. A is a stochastic process with A(0) = A0 ; 2. A has independent increments—any set of increments A(tj + hj ) − A(tj ) for j = 1, 2, . . . , n is independent, provided that the time intervals (tj , tj + hj ] are non-overlapping (touching only at an endpoint is OK); 3. for all t ≥ 0 and h > 0, the increment A(t+ h)− A(t) is a Normal random variable N (µh, σ 2 h) with mean µh and variance σ 2 h. Note that Standard Brownian Motion is just Arithmetic Brownian Motion with drift parameter 0, volatility parameter 1, and initial value 0. 6

Example 1.10 Suppose that A is an Arithmetic Brownian Motion with drift parameter µ = −1, volatility parameter σ = 2, and initial value A0 = 3. Suppose that you’ve observed that A(1.21) = 5 and you wish to know the behavior of the random variable A(1.85)—that is, of A(1.85) A(1.21) = 5. Simply write A(1.85) = A(1.85) − A(1.21) + A(1.21) = A(1.85) − A(1.21) + 5 and then use Item 3 of Fact 1.9, namely that A(1.85) − A(1.21) is Normal with mean (−1)(1.85 − 1.21) = −0.64 and variance (22 )(1.85 − 1.21) = 2.56. Since A(1.85) is just this Normal plus 5, producing a Normal with the same variance but the mean increased by 5, [A(1.85) A(1.21) = 5] ∼ N (4.36, 2.56), a Normal with mean 4.36 and variance 2.56. ¶ Note that, in general, by arguments like those in Example 1.10, (1.11) A(t) = A(t) − A(0) + A(0) = A(t) − A(0) + A0 ∼ N (µt, σ 2 t) + A0 ∼ N (A0 + µt, σ 2 t).

1.4

The final step—Geometric Brownian Motion G

A Geometric Brownian Motion G is nothing other than what you get by exponentiating an Arithmetic Brownian Motion A—that is, G = eA . You know from Equation 1.11 above that A(t) is Normal with linearly growing mean A0 + µt and linearly growing variance σ 2 t—that is, A(t) ∼ N (A0 + µt, σ 2 t). Since G(t) is then just the exponential of a Normal random variable, it’s a LogNormal random variable. Note also that G0 = G(0) = eA(0) = eA0 ; equivalently, A0 = ln G0 . Definition 1.8 on Arithmetic Brownian Motion defined A via A(t) = A0 + µt + σB(t). This makes G(t) = eA(t) = eA0 +µt+σB(t) = eln G0 +µt+σB(t) = G0 eµt+σB(t) . This last is the form I’ll use in the definition. Definition 1.12 (Geometric Brownian Motion) Geometric Brownian Motion G with drift parameter µ, volatility parameter σ, and initial value G0 is a stochastic process G with G(t) = G0 eµt+σB(t) , where B is Standard Brownian Motion. Here’s the graph of a simulated approximation to a Geometric Brownian Motion for 0 ≤ t ≤ 1, with G(0) = 1; it results from exponentiating the simulated Standard (thus Arithmetic) Brownian Motion shown in Graph 1.7. ⇐ KEY

Geometric Brownian Motion

(1.13)

7

Increments A(t + h) − A(t) played a key role with Arithmetic (and Standard) Brownian Motion. For Geometric Brownian Motion, growth factors G(t+h) play a key role. Why? Because a growth G(t) factor (1.14)
G(t+h) G(t)

for G is an exponentiated increment of an Arithmetic Brownian Motion: G0 eµ(t+h)+σB(t+h) G(t + h) = = eµh+σ[B(t+h)−B(t)] . G(t) G0 eµt+σB(t)

This relationship produces the following three facts about growth factors for Geometric Brownian Motion from the corresponding facts in Key Definition 1.5 on Standard Brownian Motion or in Fact 1.9 about increments in Arithmetic Brownian Motion; the second part of the third fact follows from recalling—or looking in the tables for Exams 3 and C/4 to see—that the mean and variance of a 2 LogNormal W = eN (m,s ) are given by (1.15) E[W ] = em+ 2 s , and Var[W ] = {E[W ]}2 (es − 1).
1
2 2

Fact 1.16 (Geometric Brownian Motion properties) Suppose that G is a Geometric Brownian Motion with drift parameter µ, volatility parameter σ, and initial value G0 . Then: 1. G is a stochastic process with G(0) = G0 ; 2. G has independent growth factors—any set of growth factors G(tj +hj )/G(tj ) for j = 1, 2, . . . , n is independent, provided that the time intervals (tj , tj + hj ] are non-overlapping (touching only at an endpoint is OK); 3. for all t ≥ 0 and h > 0, the growth factor G(t + h)/G(t) is a LogNormal random variable eN (µh,σ
2

h)

with mean e(µ+ 2 σ
G(t) G(0) G0 ,

1

2

)h

and variance e(2µ+σ

2

)h

(eσ

2

h

− 1).

Since G(t) = (1.17)

item 3) in Fact 1.16 implies that E[G(t)] = G0 e(µ+ 2 σ
1
2

)t

.

When textbooks in financial economics (or financial mathematics or mathematical finance) such as the Derivatives Markets book for the actuarial exams MFE/3 and C/4 use a Geometric Brownian Motion G(t) at time t to model a stock price S(t) at time t in terms of S(0) = S0 , they sometimes 1 use α to denote µ + 2 σ 2 . Then Equation 1.17 becomes (1.18) E[S(t)] = S0 eαt .

The parameter α above can then be interpreted as the continuously compounded expected rate of return on the stock. The parameter σ still represents volatility, and the drift parameter µ is calculated as µ = α− 1 σ 2 . More generally, if the stock pays dividends and δ denotes the continuously 2 compounded rate of dividends on the stock, you’ll see in financial economics that the drift parameter 1 µ is calculated as µ = α − 2 σ 2 − δ.

1.5

Calculations on exams

Section 1.1 explained why LogNormal models are at least plausible for stock prices, and Section 1.4 finished the development of Geometric Brownian Motion as a more precise description of these LogNormal models. Throughout this final section I consider a Geometric Brownian Motion S as a model of a stock price; it has drift parameter µ, volatility σ, and initial value S(0) = S0 . Remember from the end of the preceding section that µ might be computed as µ = α − 1 σ 2 or µ = α − 2 1 2 2 σ − δ, where α is the continuously compounded expected rate of return on the stock and δ is the 8

continuously compounded rate of dividends on the stock, as in the Derivatives Markets textbook for Exams MFE/3 and C/4. Recall from Definition 1.12 on Geometric Brownian Motion that you can view S as S(t) = S0 eµt+σB(t) , where B is Standard Brownian Motion. Actuarial exam questions may well ask you to perform calculations involving S(t); in many cases you can decline and instead calculate with B(t), and I strongly recommend that you do so since B(t) is just a Normal random variable. I’ll show you what I mean in the following subsections. ⇐ Exams!

1.5.1

Probabilities

Suppose you are asked to compute Pr[S(t) ≤ z], say. Instead of directly dealing with the LogNormal random variables S(t), write (1.19) Pr[S(t) ≤ z] = Pr[S0 eµt+σB(t) ≤ z] = Pr B(t) ≤ ln(z/S0 ) − µt σ

and deal with the Normal random variable B(t) ∼ N (0, t). Example 1.20 Suppose that µ = −1.9, σ = 2, and S0 = 100, and that you need to compute the probability Pr[S(2) ≤ 105]. Repeating the development in Equation 1.19 with these values shows that the probability is given by Pr B(2) ≤ ln(105/100) − (−1.9)(2) 2 = = ln(1.05) + 3.8 = 1.9244 2 1.9244 Pr N (0, 1) ≤ √ = 1.3608 ≈ 0.9131. ¶ 2 Pr N (0, 2) ≤

1.5.2

Simulation

The syllabus for Exam C/4 contains material on simulation by Inversion. To Inversion simulate a continuous-type random variable W with strictly increasing cumulative distribution function, you obtain a value U from a Uniform random variable on the interval from 0 to 1, solve the equation Pr[W ≤ z] = U for z, and set the Inversion-simulated value of W to be z; I’ll denote this Inversionsimulated value by ISim W . Example 1.21 Suppose you need to simulate a standard Normal random variable Z ∼ N (0, 1) by Inversion. Given a Uniform(0,1) value U , you need to find the number z so that Pr[N (0, 1) ≤ z] = U ; such a number z can be found from tables for the standard Normal distribution, and let’s call this value zU . Then ISim N (0, 1) = zU , where Pr[N (0, 1) ≤ zU ] = U. For example, if U = 0.975, then zU = 1.96. ¶ A useful fact about simulation by Inversion is that, if random variables X and Y are related by Y = g(X) for some non-decreasing function g, then for a given value U of a Uniform(0,1) random variable it turns out that ISim Y and g(ISimX) are identical. This is the monotonic transformation property of (simulation by) Inversion. 9

Example 1.22 Suppose that you need to Inversion simulate a Normal random variable W ∼ N (m, s2 ) with mean m and variance s2 . We can think of W as W = g(Z) = m + sZ for a standard Normal random variable Z ∼ N (0, 1). According to the monotonic transformation property of Inversion in the preceding paragraph, ISim W = m + sISim Z. But by Example 1.21, ISim Z = zU . Thus ISim W = m + szU . That is, ISim N (m, s2 ) = m + szU . ¶ You may be asked to Inversion simulate values of S(t). Remember that S(t) = S0 eµt+σB(t) . By the monotonic transformation property of Inversion you’ll get precisely the same result by Inversion simulating B(t) and computing S(t) from it as by Inversion simulating S(t) directly. And B(t) is easy to Inversion simulate as in Example 1.22 since B(t) ∼ N (m, s2 ) with m = 0 and s2 = t. You simply get a Uniform(0,1) value U and compute √ √ ISim B(t) = 0 + t zU = t zU . Thus the efficient way to Inversion simulate S(t) is to Inversion simulate B(t) as above and then compute ISim S(t) = S0 eµt+σISim B(t) . In practice—and on exams—you seldom want to simulate just a single value S(t); rather you want to simulate a sequence of values S(t1 ), S(t2 ), . . . , S(tn ), with 0 < t1 < t2 < · · · < tn . Since each S(tj ) = S0 eµtj +σB(tj ) , all you need do is simulate each B(tj ) and then compute S(tj ) from that value. Suppose that you simulate B(t1 ), possibly by Inversion. Since the intervals (0, t1 ] and (0, t2 ] overlap, B(t2 ) is dependent on B(t1 ) and this dependence must be taken into account in simulating B(t2 ). Recall from Key Definition 1.5 on Standard Brownian Motion that B has independent increments on non-overlapping time intervals. So B(t2 )−B(t1 ) can be simulated independently from B(t1 ) = B(t1 ) − B(0), and similarly for each increment B(tj+1 ) − B(tj ). And don’t forget that— letting hj denote tj+1 −tj —you know how to Inversion simulate B(tj+1 )−B(tj ) = B(tj +hj )−B(tj ) since by Key Definition 1.5 on Standard Brownian Motion this increment is just a Normal random variable N (0, hj ) and you can Inversion simulate Normal random variables as in Example 1.22. This process is sometimes described as “Inversion in steps”. An example should clarify how you do this. Example 1.23 Suppose that µ = 0.06, σ = 2, and S0 = 100, and that you need to simulate S(t) at t1 = 0.5 and t2 = 1 in steps of h = 0.5 using the two independent Uniform(0,1) values U1 = 0.4602 and U2 = 0.6554, respectively. I’ll do so as described in the preceding paragraph by simulating B(t) at t1 = 0.5 and t2 = 1 using Inversion on the increments in steps of h = 0.5 and then computing S(t) as S(t) = S0 eµt+σB(t) . Watch. For convenience I’ll denote the first increment B(0.5) = B(0.5)−B(0) by I1 and the second B(1)− B(0.5) by I2 . Since B(0) = 0, you also can say that B(0.5) = I1 and B(1) = I1 + I2 . As explained in the preceding paragraph, both I1 and I2 are Normal random variables N (0, 0.5) with mean 0 and √ variance 0.5. As in Example 1.22, each is therefore Inversion simulated as ISim N (0, 0.5) = 0 + 0.5 zU ≈ 0.70711zU . For our two values of U , tables of the standard Normal distribution give z0.4602 = −0.1 and z0.6554 = 0.4. This gives ISim I1 ≈ 0.70711(−0.1) = 0.070711 and ISimI2 ≈ 0.70711(0.4) = 0.28284. Thus, the simulated values B(0.5) of B(0.5) = I1 and B(1) of B(1) = I1 +I2 are B(0.5) = −0.070711 and B(1) = −0.070711 + 0.28284 ≈ 0.21213, respectively. So my simulated value S(0.5) of S(0.5) is S(0.5) = S0 eµ×0.5+σB(0.5) = 100e(0.06)(0.5)+(2)(−0.070711) ≈ 89.456. Similarly, my simulated value S(1) of S(1) is S(1) = S0 eµ×1+σB(1) = 100e(0.06)(1)+(2)(0.21213) ≈ 162.30. ¶ 10

1.5.3

Options and option pricing

The syllabi for Exams MFE/3 and C/4 contain material on European call options and European put options with strike price K at time T on a stock having price modeled by S(t). As a quick reminder: 1. A European call option with strike price K at time T gives the option owner the right to buy one share of the stock at time T for the price K if the owner wishes to do so. If the stock price S(T ) then is above K, the owner will exercise the option, buy the share for K and immediately sell it for S(T ), making a net amount N = S(T ) − K. If the price is at or below K, the owner won’t exercise the option and makes a net amount N = 0. More compactly, I’ll write the option owner’s net amount N at time T as N = (S(T ) − K)+ , where the expression (Y )+ denotes the non-negative part of Y : (Y )+ equals Y if Y ≥ 0 and equals 0 if Y < 0. 2. A European put option with strike price K at time T gives the option owner the right to sell one share of the stock at time T for the price K if the owner wishes to do so. If the stock price S(T ) then is below K, the owner will buy the share for S(T ) and immediately exercise the option by selling the share for K, making a net amount N = K − S(T ). If the price is at or above K, the owner won’t exercise the option and makes a net amount N = 0. More compactly, I’ll write the option owner’s net amount N at time T as N = (K − S(T ))+ . If you are asked to simulate the net amount N on an option, you merely simulate the stock price S(T ) at time T as discussed in the preceding subsection and then compute the net amount. The famous Black-Scholes formulas compute the price to purchase one of the above options at time 0 as the expected value of the present value—called the actuarial present value in the life contingencies material on Exam MLC/3—of the net amount N . In the Derivatives Markets textbook, the force of interest used in discounting to obtain present values is denoted by r and called the risk-free rate; moreover, the continuously compounded expected rate of return α on the stock is assumed to equal the risk-free rate r, so α = r. For consistency with that text, I now assume that the stock price S is Geometric Brownian Motion with drift parameter µ = r − 1 σ 2 − δ, 2 volatility σ, and initial value S0 , where r is the risk-free force of interest used in discounting and the continuously compounded expected rate of return on the stock. The Black-Scholes price C for a European call option with strike price K at time T is then computed as (1.24) C = E[e−rT N ] = e−rT E[(S(T ) − K)+ ], while the price P for a European put option with strike price K at time T is computed as (1.25) P = E[e−rT N ] = e−rT E[(K − S(T ))+ ].

So the question is: on exams, how do you compute C or P above? For exam MFE, unfortunately you should memorize the formulas from the textbook or your study manual. But for Exams 3 and C/4, you can use information in the distribution tables provided and needn’t memorize any formulas. To use the tables, you need to remember a relationship also useful in the Exam C/4 syllabus material on severity. Recall that the notation A ∧ B stands for the smaller of A and B. The useful relationship is that (1.26) (A − B)+ + (A ∧ B) = A ⇐ Exams!

for all numbers A and B. [You should check this by considering the two cases A ≥ B and A < B.] To use this for the call price C in Equation 1.24, for example, write (S(T ) − K)+ + (S(T ) ∧ K) = S(T ), from which it follows that (S(T ) − K)+ = S(T ) − (S(T ) ∧ K). Therefore the expected value required for C in Equation 1.24 can be computed as (1.27) E[(S(T ) − K)+ ] = E[S(T )] − E[S(T ) ∧ K]. 11

So calculating the price C of the European call option reduces to calculating the two expected values on the right-hand side of Equation 1.27. How do you do that? By looking them up in the tables for Exams 3 and C/4! Remember that S(T ) is a LogNormal random variable, and the tables contain formulas for the moments E[W k ] and limited moments E[(W ∧ x)k ] of a LogNormal random variable W . Example 1.28 Suppose that the initial stock price is S0 = 41, the strike price on a European call option at time T = 0.25 is K = 40, the volatility is σ = 0.3, the risk-free force of interest (and continuously compounded expected rate of return on the stock) is r = 0.08, and that there are no dividends—that is, that δ = 0. This yields a drift parameter µ = r − 1 σ 2 − δ = 0.035. I want to 2 compute the Black-Scholes call price C in Equation 1.24 via C = e−(0.08)(0.25) E[(S(0.25) − 40)+ ]. Now S(0.25) = = S0 eµ×0.25+σB(0.25) = 41e(0.035)(0.25)+(0.3)B(0.25) eln 41+0.00875+0.3N (0,0.25) = eN (3.7223, 0.0225) ,
2

which shows that S(0.25) is a LogNormal random variable W = eN (m,s ) with m = 3.7223 and s2 = 0.0225 (so s = 0.15). To use Equation 1.27, I need both E[W ] and E[W ∧ 40]. The Exams 3 and C/4 tables for the LogNormal give E[W ] = em+ 2 σ ≈ e3.73355 ≈ 41.827; similarly, they give E[W ∧ 40] = em+ 2 s Φ
1
2

1

2

ln 40 − m − s2 s

+ 40 1 − Φ

ln 40 − m s

,

where Φ(y) = Pr[N (0, 1)] ≤ y] gives the cumulative distribution function for the standard Normal distribution. Calculating produces E[W ∧ 40] ≈ 38.364. So, from Equation 1.27, E[S(0.25) − 40)+ ] = E[(W − 40)+ ] = E[W ] − E[W ∧ 40] ≈ 41.827 − 38.364 = 3.463. Finally, from Equation 1.28 we get the price of the European call as C ≈ e−0.02 (3.463) ≈ 3.395. Tedious, but no memorized formulas! ¶

12

Similar Documents

Premium Essay

Mathematical / Statistical Background for Option Pricing

...expectation of the function g  X  to be the integral E  g  X      g  x  f  x  dx  Note that g  X  is also a random variable The Moment Generating Function (MGF) The MGF of a random variable X is a function of t denoted by M X  t   E  e xt  which is an expectation MGF of normal If X ~ N   ,  2  1  x     1 Xt xt Then M X  t   E  e     e e 2     2   Lognormal Distribution:  2 1   t   2t 2  dx  e 2   Y has the lognormal distribution with parameters   , 2  if:   its logarithm is normally distributed X  log e Y  ~ N  , 2 . This in turn means that Y  e X 2 The cumulative density function of Y is  log e  y     FY  y   Pr Y  y   N      x 12 1  2t where N  x    e dt the cdf of the standard normal distribution 2  The probability density function of Y is  1  log  y    2  1 e fY  y   exp      2  y 2    for y > 0 The mean and variance of the Lognormal distribution:  The mean of Y is E Y   e 1 2   2 e 1 E  X...

Words: 7933 - Pages: 32

Premium Essay

Geometric Brownian Motion - Karl Sigman

...Copyright c 2006 by Karl Sigman 1 Geometric Brownian motion Note that since BM can take on negative values, using it directly for modeling stock prices is questionable. There are other reasons too why BM is not appropriate for modeling stock prices. Instead, we introduce here a non-negative variation of BM called geometric Brownian motion, S(t), which is defined by S(t) = S0 eX(t) , (1) where X(t) = σB(t) + µt is BM with drift and S(0) = S0 > 0 is the intial value. Taking logarithms yields back the BM; X(t) = ln(S(t)/S0 ) = ln(S(t))−ln(S0 ). ln(S(t)) = ln(S0 )+X(t) is normal with mean µt + ln(S0 ), and variance σ 2 t; thus, for each t, S(t) has a lognormal distribution. 2 As we will see in Section 1.4: letting r = µ + σ , 2 E(S(t)) = ert S0 the expected price grows like a fixed-income security with continuously compounded interest rate r. In practice, r >> r, the real fixed-income interest rate, that is why one invests in stocks. But unlike a fixed-income investment, the stock price has variability due to the randomness of the underlying Brownian motion and could drop in value causing you to lose money; there is risk involved here. (2) 1.1 Lognormal distributions If Y ∼ N (µ, σ 2 ), then X = eY is a non-negative r.v. having the lognormal distribution; called so because its natural logarithm Y = ln(X) yields a normal r.v. X has density f (x) = This is derived via computing xσ 1 √ e 2π −(ln(x)−µ)2 2σ 2 0, d dx F (x) , if x ≥ 0; if x < 0. for ...

Words: 4386 - Pages: 18

Premium Essay

Black Scholes Model

...Valuing Stock Options: The Black-Scholes-Merton Model Chapter 13 Fundamentals of Futures and Options Markets, 8th Ed, Ch 13, Copyright © John C. Hull 2013 1 The Black-Scholes-Merton Random Walk Assumption  Consider a stock whose price is S  In a short period of time of length Dt, the return on the stock (DS/S) is assumed to be normal with:  mean m Dt  standard deviation s Dt  m is the annualized expected return and s is the annualized volatility. Fundamentals of Futures and Options Markets, 8th Ed, Ch 13, Copyright © John C. Hull 2013 2 Why can we say that?  Assume that the Normal(m,s2) annual return is made up of the sum of n returns of shorter horizons (eg. monthly, weekly): m  E ( ri )  E (ri ) nE (ri ) i 1 i 1 n n 2 n n thus E (ri )  m / n thus Var (ri )  s 2 / n s  Var ( ri )  Var (ri ) nVar (ri ) i 1 i 1  We have n=1/Dt intervals of length Dt in a year (eg. for monthly n=1/(1/12) = 12 intervals of length 1/12 of a year), therefore: E (ri )  m / n  m / (1/ Dt )  mDt Var (ri )  s 2 / n  s 2 / (1/ Dt )  s 2 Dt Sigma(ri )  s Dt Fundamentals of Futures and Options Markets, 8th Ed, Ch 13, Copyright © John C. Hull 2013 3 The Lognormal Property  These assumptions imply that ln ST is normally (Gaussian) distributed with mean: ln S 0  (m  s 2 / 2)T and standard deviation: s T  Because the logarithm of ST is normal, the future value or price (at time T) of the stock ST is...

Words: 1783 - Pages: 8

Free Essay

Luenberger

...I NVESTMENT SCI ENCE I NVESTMENT SCI ENCE DA YID G. LUENBERGER STANFORD UNIVERSITY New York Oxford OXFORD UNIVERSITY PRESS 1998 OXFORD UNIVERSITY PRESS Oxford New York Auckland Bangkok Bogota Bombay Buenos Aires Cnlcutta Cape Town Dar es Salaam Delhi Florence Hong Kong Istanbul Karachi Athens Kuala Lumpur Mexico City Madras Nairobi Mndrid Paris Melbourne Singapore Taipei Tokyo Toronto F \--1& ljS1S,'L (Jml aHociated compallies ill Berlin Jbndon ' LE 4 /3 en where that last expression is valid in the limit as In goes to infinity, cOllesponding to continuous compounding Hence continuous compounding leads to the familiar expo~ nential growth CUlve Such a curve is shown in FigUle 2 2 for a 10% nominal interest late Debt We have examined how a single investment (say a bank deposit) glows over time due to intelest compounding It should be clem that exactly the same thing happens 10 debt It I bonoll' money from the biwk at an intelest rate 1 and make no payments to the bank, then my debt increases accOJding to the same formulas Specifically, if my debt is compounded monthly, then after k months my debt will have grown by a factor of [I + (I /12) l' 21 14 12 10 PRINCIPAL AND INTEREST 17 FIGURE 2.2 Expollential growth curve; cOllfinuous compoUlld growth, Under conl;nuotls compounding at 1D'X" the value of $1 doubles in abotll 7 yems In 20 yems it grows by a factor of ilbotll B !5 ~ 4 0...

Words: 24917 - Pages: 100

Premium Essay

Final Exam

...material first appears in your textbook. Good luck! / /1) d6nominated interest rate is 5%. The price of a 2-year U.S.-@Ddenominated T=2 pyloqtion on Canpdial dollars, with a strike price of S1, is S.tO. Find the price of (,ooo )-v"^, u.s.@Jlddenominated calloptions on canadifr-doilars. .,/r. r Suppose the current exchange rate is L.l-0 Canadian dollars per 1. U.S. dollar. Also, the Canadian-dollar-denominated interest rate is 4%, wbilqthe U.S.-dollar- rA* I : l'l cS/frf or. r Ycs =.0 v, f a =.oJ -l v 2)7You are considering entering into a box spread, whereby you buy a 45-strike call z/ option for 58.50, sell a 45-strike put option for 53.50, sell a So-strike call option for 56.50, and buy a S0-strike put option for 56.00. Assume that all options can only be exercised 1 year from now. Also, assume that the continuously compounded risk-free interest rate is 3%. Construct a profit table (at time T) to demonstrate than an arbitrage opportunity exists (based on the given option ' prices), and state the amount of the guaranteed profit that is realized (at t=T). Pur( .4oqo4(, t t'z) =,1 '/ You use a 2-period binomial tree model to price two call options on a futures contract. The initial futures price is 100, and subsequent prices (in the tree) are determined, assuming u*d = 1 and u/d = 1.5. Both call options have time-tomaturity of 6 months and strike price 100, but one option is European-style and the other...

Words: 860 - Pages: 4

Premium Essay

Game Shop Inc.

...Portfolio Insurance, edited by Don Luskin (John Wiley and Sons 1988)] [reprinted in The Handbook of Financial Engineering, edited by Cliff Smith and Charles Smithson (Harper and Row 1990)] [reprinted in Readings in Futures Markets published by the Chicago Board of Trade, Vol. VI (1991)] [reprinted in Vasicek and Beyond: Approaches to Building and Applying Interest Rate Models, edited by Risk Publications, Alan Brace (1996)] [reprinted in The Debt Market, edited by Stephen Ross and Franco Modigliani (Edward Lear Publishing 2000)] [reprinted in The International Library of Critical Writings in Financial Economics: Options Markets edited by G.M. Constantinides and A..G. Malliaris (Edward Lear Publishing 2000)] Abstract This paper presents a simple discrete-time model for valuing options. The fundamental economic principles of option pricing by arbitrage methods are particularly clear in this setting. Its development requires only elementary mathematics, yet it contains as a special limiting case the celebrated Black-Scholes model, which has previously been derived only by much more difficult methods. The basic model readily lends itself to generalization in many ways. Moreover, by its very construction, it gives rise to a simple and efficient numerical procedure for valuing options for which premature exercise may be optimal. ____________________ † Our best thanks go to William Sharpe, who first suggested to us the advantages of the discrete-time...

Words: 13937 - Pages: 56

Premium Essay

Finance

...Financial Institutions Center Derivatives and Corporate Risk Management: Participation and Volume Decisions in the Insurance Industry by J. David Cummins Richard D. Phillips Stephen D. Smith 98-19 THE WHARTON FINANCIAL INSTITUTIONS CENTER The Wharton Financial Institutions Center provides a multi-disciplinary research approach to the problems and opportunities facing the financial services industry in its search for competitive excellence. The Center's research focuses on the issues related to managing risk at the firm level as well as ways to improve productivity and performance. The Center fosters the development of a community of faculty, visiting scholars and Ph.D. candidates whose research interests complement and support the mission of the Center. The Center works closely with industry executives and practitioners to ensure that its research is informed by the operating realities and competitive demands facing industry participants as they pursue competitive excellence. Copies of the working papers summarized here are available from the Center. If you would like to learn more about the Center or become a member of our research community, please let us know of your interest. Anthony M. Santomero Director The Working Paper Series is made possible by a generous grant from the Alfred P. Sloan Foundation Derivatives and Corporate Risk Management: Participation and Volume Decisions in the Insurance Industry By J. David Cummins Wharton School, University...

Words: 15024 - Pages: 61

Free Essay

Flaws with Black Scholes and Exotic Greeks

...the experts. In Richard Meyers’ estimation, risk managers or traders do not socialize enough. “It’s all about visibility,” he said. Meyers, chairman and CEO of Richard Meyers & Associates, a talent acquisition and management firm in New Jersey, relates the story of a firm that decided to adopt an Enterprise Risk Management (ERM) strategy. Instead of appointing its risk manager to head ERM, the company brought in someone else. Why? Time has come when organizations across the world have to do deep amendments in their Enterprise Risk Management (ERM) policies covering foreign exchange hedging programs, diversification in derivatives portfolio, Enterprise risk management policies and deeper and deeper understanding towards financial models. With this background paper would...

Words: 9364 - Pages: 38

Free Essay

Black-Scholes

...financial decision. Derivative financial instruments such as options, futures and others have been introduced and more commonly used to manage financial risk for improving decision making in this dynamic competitive environment. Options are defined as securities which one party has the right (no obligation) to buy or sell underlying assets with certain price within a certain/specific period of time (Hull, 2012). The option can be either call (right to buy) or put (right to sell) in the form of American options (exercised any time until expiry date) or European options (exercised on expiry date) as either traded options (standard option contracts) or overt-the-counter options (tailor made options). Due to various choices of options, different option pricing models such as Put-Call Parity, Black-Scholes, Cox-Rubenstein Binominal, Risk-Neutral valuation, the Greeks and others has been developed and applied in current financial market. Black-Scholes Option Pricing Model (BS) BS is designed and introduced by Fisher Black and Myron Scholes in 1973 with the assumptions of the market is efficient, returns are lognormal distributed, no commission or transaction cost is charged, no dividend is paid, no penalties to short selling, terms of European option is used, interest rate is remained constant and known rate (Black & Scholes, 1973). Thereafter, the assumption of no dividends has been relaxed by Robert Merton in the same year....

Words: 1927 - Pages: 8

Premium Essay

Finance Quantities

...APPENDICES A B C Quantitative Review References to CFA Questions Glossary A P P E N D I X A QUANTITATIVE REVIEW Students in management and investment courses typically come from a variety of backgrounds. Some, who have had strong quantitative training, may feel perfectly comfortable with formal mathematical presentation of material. Others, who have had less technical training, may easily be overwhelmed by mathematical formalism. Most students, however, will benefit from some coaching to make the study of investment easier and more efficient. If you had a good introductory quantitative methods course, and like the text that was used, you may want to refer to it whenever you feel in need of a refresher. If you feel uncomfortable with standard quantitative texts, this reference is for you. Our aim is to present the essential quantitative concepts and methods in a self-contained, nontechnical, and intuitive way. Our approach is structured in line with requirements for the CFA program. The material included is relevant to investment management by the ICFA, the Institute of Chartered Financial Analysts. We hope you find this appendix helpful. Use it to make your venture into investments more enjoyable. 1006 Appendix A 1007 A.1 PROBABILITY DISTRIBUTIONS Statisticians talk about “experiments,” or “trials,” and refer to possible outcomes as “events.” In a roll of a die, for example, the “elementary events” are the numbers 1 through 6...

Words: 16892 - Pages: 68

Premium Essay

Scenario Analysis

...1 PROBABILISTIC APPROACHES: SCENARIO ANALYSIS, DECISION TREES AND SIMULATIONS In the last chapter, we examined ways in which we can adjust the value of a risky asset for its risk. Notwithstanding their popularity, all of the approaches share a common theme. The riskiness of an asset is encapsulated in one number – a higher discount rate, lower cash flows or a discount to the value – and the computation almost always requires us to make assumptions (often unrealistic) about the nature of risk. In this chapter, we consider a different and potentially more informative way of assessing and presenting the risk in an investment. Rather than compute an expected value for an asset that that tries to reflect the different possible outcomes, we could provide information on what the value of the asset will be under each outcome or at least a subset of outcomes. We will begin this section by looking at the simplest version which is an analysis of an asset’s value under three scenarios – a best case, most likely case and worse case – and then extend the discussion to look at scenario analysis more generally. We will move on to examine the use of decision trees, a more complete approach to dealing with discrete risk. We will close the chapter by evaluating Monte Carlo simulations, the most complete approach of assessing risk across the spectrum. Scenario Analysis The expected cash flows that we use to value risky assets can be estimated in one or two ways. They can represent a probability-weighted...

Words: 17404 - Pages: 70

Premium Essay

America Online

...You are expecting IBM stock price to go up in next 8 months, however you are not completely sure. So you decide to use just one option, either European call or European put on IBM stock maturing in 8 months to bet on your view about IBM’s stock price prospects. Suppose that the current stock price and the strike price for both call and put on the IBM stock is $50. (a) What option will you invest in? Explain.  Call. Call price will go up if the stock price goes up. The losses are limited by the option premium paid. (b) At what price will you breakeven if both put and call options are sold for the same premium of $5 Breakeven stock price $50+$5 = $55 (c) Assume that the risk free rate is 3% per annum. Also assume that the standard deviation of IBM’s stock return is 30% per year. What is the Black-Scholes value of the option you have identified in part a? Step 1: find d1 and d2 d_1=(ln⁡(50/50)+(0.03+〖0.30〗^2/2)×8/12)/(0.30×√(8/12))=0.2041 d_2=0.2041-0.30×√(8/12)=-0.0408 Step 2: find N(d1) and N(d2) Using the cumulative normal table obtain N(d1) = N(0.20) = 0.5793 and N(d2) = N(-0.04) = 0.4841 Step 3: calculate the call option value c=$50×0.5793-$50×e^(-0.03×(8/12) )×0.4841=$5.2393 (d) What is the time value of the option you have identified in part a? Because the stock price equals the strike price ($50) the total value of the option would consist of time value only, therefore the time value of this option is $5.2393 Problem 2 You anticipate that the volatility...

Words: 2634 - Pages: 11

Free Essay

Stochastic Volatility

...with stochastic volatility (the Heston Model) Aarhus School of Business and Social Science 2011 2 Acknowledgements My gratitude and appreciation goes to my supervisor Peter Lø chte Jø rgensen, for his kind and insightful discussion and guide through my process of writing. I was always impressed by his wisdom, openness and patience whenever I wrote an email or came by to his office with some confusion and difficulty. Especially on access to the information on certain Danish structured products, I have gained great help and support from him. 3 Abstract My interest came after the reading of the thesis proposal on strucured products written by Henrik, as is pointed out and suggested at the last part of this proposal, one of the main limitations of this thesis may be the choice of model. This intrigues my curiosity on pricing Asian options under assumption of stochstic volatility. At first, after the general introduction of strucutred products, the Black Scholes Model and risk neutral pricing has been explained. The following comes the disadvanges of BS model and then moves to the stochastic volatility model, among which the Heston model is highlighted and elaborated. The next part of this thesis is an emricical studying of two structured products embbeded with Asian options in Danish market and follows with a conclusion. Key words: structured products, Asian options, Black-Scholes model, stochastic volatilty, Heston model, calibration 4 Table of Contents ...

Words: 17332 - Pages: 70

Premium Essay

Financial Modelling

...Advanced Modelling in Finance using Excel and VBA Mary Jackson and Mike Staunton JOHN WILEY & SONS, LTD Chichester ž New York ž Weinheim ž Brisbane ž Singapore ž Toronto Copyright  2001 by John Wiley & Sons, Ltd, Baffins Lane, Chichester, West Sussex PO19 1UD, England National 01243 779777 International (C44) 1243 779777 e-mail (for orders and customer service enquiries): cs-books@wiley.co.uk Visit our Home Page on http://www.wiley.co.uk or http://www.wiley.com All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 9HE, UK, without the permission in writing of the publisher. Other Wiley Editorial Offices John Wiley & Sons, Inc., 605 Third Avenue, New York, NY 10158-0012, USA Wiley-VCH Verlag GmbH, Pappelallee 3, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 42 McDougall Street, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 6045 Freemont Blvd, Mississauga, ONT, L5R 4J3, Canada British Library Cataloguing in Publication Data A catalogue record for this book is available from the British...

Words: 57326 - Pages: 230

Premium Essay

Sdas

...LECTURE 7: BLACK–SCHOLES THEORY 1. Introduction: The Black–Scholes Model In 1973 Fisher Black and Myron Scholes ushered in the modern era of derivative securities with a seminal paper1 on the pricing and hedging of (European) call and put options. In this paper the famous Black-Scholes formula made its debut, and the Itˆ calculus was unleashed upon the world o 2 of finance. In this lecture we shall explain the Black-Scholes argument in its original setting, the pricing and hedging of European contingent claims. In subsequent lectures, we will see how to use the Black–Scholes model in conjunction with the Itˆ calculus to price and hedge all manner of o exotic derivative securities. In its simplest form, the Black–Scholes(–Merton) model involves only two underlying assets, a riskless asset Cash Bond and a risky asset Stock.3 The asset Cash Bond appreciates at the short rate, or riskless rate of return rt , which (at least for now) is assumed to be nonrandom, although possibly time–varying. Thus, the price Bt of the Cash Bond at time t is assumed to satisfy the differential equation dBt (1) = rt Bt , dt whose unique solution for the value B0 = 1 is (as the reader will now check) t (2) rs ds . Bt = exp 0 The share price St of the risky asset Stock at time t is assumed to follow a stochastic differential equation (SDE) of the form (3) dSt = µt St dt + σSt dWt , where {Wt }t≥0 is a standard Brownian motion, µt is a nonrandom (but not necessarily...

Words: 3260 - Pages: 14