Free Essay

In: Science

Submitted By a6syed

Words 5362

Pages 22

Words 5362

Pages 22

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…...

Premium Essay

...ll Why Does a Stock's Price Rise or Fall? Introduction Watching the ups and downs of stock prices can be enough to make you seasick. If you own stocks, you've undoubtedly followed their prices with a feeling of either satisfaction or disappointment, depending on how your investments have done. Short-term swings can be bewildering, and sometimes it seems as though stock prices follow a logic all their own. Even so, it is possible to break down stock performance in ways that help distinguish solid growth from inflated expectations. Case Study in Rising Prices: Wal-Mart A stock's price is determined by its fundamentals (how the company has actually performed) combined with its valuation (how much the market is willing to pay for that performance and the promise of future growth). Similarly, when a stock's price goes up, that rise comes from either (1) growth in the underlying business or (2) an increase in the stock's valuation, as the market becomes more optimistic about the company's future. As an illustration of these two drivers of stock performance, consider the history of Wal-Mart WMT. In the 1970s and 1980s, Wal-Mart was the quintessential growth stock. It was profitable, with returns on equity consistently above 20%. Year after year, its revenue and earnings grew by more than 25%, often by more than 30%, as it opened new stores all over the country. As a result of this consistent growth, the annualized return of Wal-Mart's stock from 1970 to 1990......

Words: 1430 - Pages: 6

Premium Essay

...Regression Analysis for Determining the Price of Alcan Stock [pic] Jason Scott May 1, 2006 Introduction In this project, we have developed a model using stepwise regression to predict the price of Alcan’s stock, based on the impact of eight independent variables on the price of Alcan’s stock. The company’s stock is listed on the New York Stock Exchange (NYSE) under the ticker symbol AL. Of the eight variables we will be looking at, we are most interested in the impact that the price of aluminum has on Alcan’s stock. Alcan is the second largest producer of aluminum in the world, and may be one of the purest plays on aluminum, given that Alcan spent the mid 1990’s spinning off all operations not involved with the production of aluminum. If any company would reflect the impact of the change in aluminum prices, it would be Alcan. Additionally, we have a keen interest in this field as both of us are employees of a company recently spun-off from Alcan. The eight independent variables that we chose for our analysis, and our reasons for choosing them, are detailed below. Definition of Variables Dependant Variable Alcan (ALCAN) The dependent variable for our model is the monthly stock price of Alcan. Alcan is publicly traded on the New York Stock Exchange (NYSE) under the ticker symbol AL. The stock price is collected daily and these daily numbers are averaged together to give one monthly value. We have collected the......

Words: 12428 - Pages: 50

Premium Essay

...“IMPACT OF DIVIDEND ON STOCK PRICES” ABSTRACT The project aims to establish the impact of dividend on market price of a share. This has been done for individual companies in Steel sector. After studying the basic concepts of dividends and dividend policy I am able to get a proper perspective of the requirements of the project and also gain a better understanding of the results obtained. I have looked to find the relation between pre dividend price change and the dividend using regression analysis. Similarly, I have analyzed the relation between the post dividend price change and the dividend. It is a matter of fact that dividends are declared by a company primarily to generate capital and also at certain times to maintain the market sentiment. It is in the best interests of the company to maximize the market value of its share and companies use dividend as a tool to maintain their corporate image. However, the degree of correlation between the dividend and the market price is low which implies that several other internal and external factors affect the market value of shares. To gain a holistic picture, I also did a comparative study of two peers in a sector (steel)to better understand how the specific requirements of each sector also impact the dividend policy of a company. The conclusions derived from the analysis performed further consolidated theoretical knowledge and deviations were better understood. SYNOPSIS “Impact......

Words: 8285 - Pages: 34

Premium Essay

...CHAPTER 5. FLEXIBLE PRICES: THE MONETARY MODEL The monetary approach was the earliest approach to explain the related phenomena of exchange rate variations, on the other hand, and the BOP, in the other. While it is completely inadequate in explaining the wild day-to-day, month-to-month gyrations seen in recent years, it nonetheless provides some useful insights into the broader picture of long run trends. It consists of two basic building blocks: PPP and demand for money. 5.1 The Simple Monetary Model of a Floating Exchange Rate 5.1.1 The Setting: a. AS is vertical This does not imply that output is constant. It can only vary as the result of a change in the productivity of the economy: accumulation of capital, growth in the labor force or its educational level, etc.. And a vertical AS curve presupposes perfect price flexibility in all markets. b. The demand for real money balances is a stable function of only a few domestic macroeconomic variables. In fact, for the moment and till further notice, we shall work with the Cambridge quantity equation provided below: Md = kPy k > 0 Given a money stock, MS0, equilibrium means: MS0 = Md = kPy = kY The figure below shows that the fall in the price level is of the same proportion as the rise in income as we move from A to D. However, a change in MS would shift AD curve. AD curve will shift upward and to the right when there is an increase in MS. At point B, P1 must be greater than P0 by the same......

Words: 2404 - Pages: 10

Premium Essay

...Macro-Economic Variables on the Stock Market Impact of Macro-Economic Variables on the Stock Market Arunabha Dhar (Roll No. 008) Gaurav Bhatt (Roll No. 017) Amartya Ray (Roll No. 067) Bodhisatva Basu (Roll No. 075) Rahul Jain (Roll No. 094) Arunabha Dhar (Roll No. 008) Gaurav Bhatt (Roll No. 017) Amartya Ray (Roll No. 067) Bodhisatva Basu (Roll No. 075) Rahul Jain (Roll No. 094) Contents UNDERSTANDING ON RESEARCH PROBLEM IDENTIFICATION & DEFINITION 3 ABSTRACT 3 INTRODUCTION 3 LITERATURE REVIEW 6 GAP in Research 8 MOTIVATION 8 DATA COLLECTION/SAMPLE SELECTION 9 HYPOTHESIS 10 Research Methodology 10 FINDINGS 11 CONTRIBUTION TO LITERATURE 23 CONCLUSIONS 23 References 25 UNDERSTANDING ON RESEARCH PROBLEM IDENTIFICATION & DEFINITION Relationship between macroeconomic variables and broad market index: A causal relationship between Nifty CNX and macroeconomic variables in India ABSTRACT The relationship between macroeconomic variables and broad market index by now are well documented in the literature. However a void in the literature relates to examining the causal relationship between Nifty CNX and macroeconomic variables such as FDI, FPI, weighted average lending rate (WALR), GDP and oil import in India and correlation among the macro variables. INTRODUCTION Globalization of Indian economy post liberalization has been spurred by capital and stock investment in terms of FDI & FPI. Indian stock market both securities......

Words: 6806 - Pages: 28

Premium Essay

...[pic] Michael G. Foster School of Business Using a Valuation Model to Estimate a Firm’s Stock Price* In the ongoing search for bargains in the stock market, analysts and investors rely on models to estimate the intrinsic value of a firm’s equity. By comparing the valuation suggested by their model to the actual value in the marketplace, they form opinions as to whether a given stock is under or over valued. Valuation models are also used by investment bankers as an aid to pricing initial public offerings, and to inform parties involved in assorted private transactions such as selling a business or division, dividing property among owners, and settling estates. In this note, we introduce a relatively simple but powerful model of equity (stock) valuation.[1] 1. The basic idea behind valuation Valuation models in finance are typically based on discounted future cash flows or discounted future dividends. Keep in mind that, holding underlying assumptions constant, all valuation models should yield the same result. A model for valuing equity based on accounting data may be preferable in some cases, in that: o Benchmarks for performance are almost always given in earnings per share (EPS) – not cash flows or dividends. o Since real world dividend payout policies tend to be stable for long periods, valuation models based on dividends are less useful for modeling changes in value. o Earnings generally receive far more attention from the......

Words: 3809 - Pages: 16

Premium Essay

...Stock Price Analysis For each organization, it is very important to consider and review the stock price analysis. Here, the stock price analysis of the organizations Amazon and Ebay shall be carried out. In relation to the stock price analysis, the analysis of the stock prices shall be reviewed from January 1, 2010, 2011 and 2012 and in addition to this, the stock price analysis for the organization shall also be considered regarding the stock price of the organization since the date of its incorporation. The graph showing the changes in the stock prices of the organizations is provided below (Yahoo Finance, 2014). The chart given above shows the changes in the stock prices for Ebayfrom January 1,2010, January 1,2011 and January 1,2012. These charts show the information that, the organization has been in a position to enhance its stock prices since the date of incorportation. So although the stock price has increased since being incorporated, it hasn’t been a substantial increase amount. Based on this chart, it can be said that, the stock price of the Ebay has definitely grown since the date of its incorporation till the present date, but the growth has been small yet. Though, the increase has not been a significant change, the organization has made sure that, it carries out its activities in the most appropriate manner without any kind of problem or issue for it. In relation to the Amazon, the stock prices shall be analyzed as well.. At the time of incorporation......

Words: 419 - Pages: 2

Free Essay

...Another method of business forecasting would be qualitative models. Unlike a time series model, this type of model utilizes more opinions and judgments. There are four different approaches of qualitative methods. The first is the Delphi method. The Delphi method which uses questionnaires and modifies them until they receive the results necessary. Another approach is the jury of executive opinion which takes small amounts of managers to obtain the demand aspect. Sometimes statistic models are used to assist as well. Some use the consumer market survey which takes the opinions of consumers to forecast future purchasing. Lastly, sales force composite is when every member of a sales teams predicts a realistic sales number for their area. (Hanna, Render, & Stair, 2012) The Department of Energy’s Energy Information Administration (EIA) focusing a great deal on sales from main suppliers. The following chart beginning in 1990 to the most current data, it highlights when Americans were paying the most for gasoline as well as when it was the least expensive. It’s obvious to see that shortly after the recession in 2008 there a major increase in the price of gasoline from suppliers which lead to reduction of sales in the US (Short, 2014). Ever since that point the price of gasoline has slowly been decreasing to around three dollars where it is currently. Not only has the price per gallon lessened but sales have slowly been falling as well. Some people blame the fact there are other...

Words: 516 - Pages: 3

Premium Essay

...predictability CHAPTER ONE INTRODUCTION 1.1 BACKGROUND TO THE STUDY The performance of an economy is dependent largely on the efficient performance of its financial markets, since they facilitate the financing of productive activity and hence national output and economic growth. In this research report, the key roles and function of the financial markets are highlighted with the thrust of the discussion on the core issue of how the market works; directly and indirectly. One of the most important factors for rapid economic development is the effective mobilization and allocation of scarce resources within an economy. These resources can be real or financial, but they are scarce and command a price. The establishment of effective and efficient channel for the mobilization and allocation of scarce financial resources is therefore essential. The financial market, comprising of the money and capital markets, occupies an important place in most economies of the world. The primary function of a financial market is to enable funds to be sufficiently allocated from the surplus units of the economy to the deficit units for productive investment. The greater the transmission efficiency is, the higher the rate of growth of the economy (Olowe, 1997). The money market trades only in securities or debt instruments maturing in less than twelve months, while in the capital market, longer term debts as well as equity instruments are traded. The complementarity between money......

Words: 11803 - Pages: 48

Premium Essay

...The relationship between stock prices and exchange rates in China Mengyuan Chen Illinois Wesleyan University Dec 10, 2012 Abstract This paper uses the data of RMB exchange rates and stock market prices in China from 1994 to 2011 to estimate the relationship between stock prices and exchange rates. There are two major theories concerning the relationship. According to the portfolio balance effect, these two variables should be negatively related; in addition, according to the international trading effect theory, these two variables should be positively related. The linear regression model is adopted to observe the various relationships between stock and foreign exchange markets. The results conﬁrmed my hypothesis, which indicates that the international trading effect is more dominant, thus the net effect is a positive causal relationship from exchange rates to stock prices. I. Introduction Within the emerging Chinese market, China now has more open policies and advanced financial market instruments to promote globalization. For example, China started to allow the RMB to float within a larger daily range in 2005 and brought derivative options into the stock market. These significant steps all suggest that China is beginning to face a new economic condition. For instance, the challenging policy making of RMB exchange rate is one. Exchange rates and stock prices are both key indicators of the economy and financial markets. So the relationship between those two becomes......

Words: 2999 - Pages: 12

Premium Essay

...Does Voluntary Disclosure Improve Stock Price Informativeness? Author(s): K. Stephen Haggard, Xiumin Martin and Raynolde Pereira Reviewed work(s): Source: Financial Management, Vol. 37, No. 4 (Winter, 2008), pp. 747-768 Published by: Blackwell Publishing on behalf of the Financial Management Association International Stable URL: http://www.jstor.org/stable/20486678 . Accessed: 25/07/2012 04:27 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. . Blackwell Publishing and Financial Management Association International are collaborating with JSTOR to digitize, preserve and extend access to Financial Management. http://www.jstor.org Does Voluntary Stock Price Disclosure Improve Informativeness? K. Stephen Haggard, Xiumin Martin, and Raynolde Pereira* According to theory, comovement in stock prices reflects comovement in thefundamentalfactors underlying the values ofstocks. Recent theory contends that stockprice comovement can be driven by information markets or the informational opacity of the...

Words: 10226 - Pages: 41

Free Essay

...estacionario, debido al comportamiento funcional. Esto se debe b´sicamente a que en el caso de a costos marginales decrecientes se provocan rendimientos a escala crecientes, por lo que la ﬁrma querr´ producir cada vez m´s entonces su stock ser´ enorme y sus costos a a a tender´n a cero para poder ser m´s competitiva y eﬁcientes respecto al contricante. a a En el caso de los costos nulos e id´nticos, se puede notar que por la rigidez en estos, e los contratos impiden que la competencia en precios de el resultado deseado, ya que al ser manejado por contratos con distintos vencimientos, en ocasiones bajar tu precio no incrementar´ tu demanda y en otra puede incrementarla mucho, depende de cu´ntos a a contratos est´n por vencerse. e Este trabajo concluye que con este tipo de costos el equilibrio no es sostenible en el tiempo, genera incosistencias pero s´ permite la competencia. Se concluye que optimizar ı 12 din´micamente, puede llevar a desviaciones a lo largo del tiempo pero s´ mejora los a ı beneﬁcios en el caso que tal vez nunca lleguen a precio igual a costo marginal, lo que s´ pasar´ en un modelo est´tico. ı ıa a Referencias [1] Fujiwara, K. (2006) A Stackelberg Model of Dynamic Duopolistic Competition with Sticky Prices, ”Economics Bulletin”, 12, 12, pp. 1-9. [2] Mancero, X. Saavedra, E. (2006) Un Modelo de Entrada y Competencia en Telecomunicaciones, Revista de an´lisis econ´mico”, 21, 1, pp. 29 - 57. a o [3] Dockner, E., S.......

Words: 4546 - Pages: 19

Premium Essay

...– Banking sector & Stock Exchange and Conceptual framework – Macroeconomic variable |3.1 An overview of Banking sector and Stock market of Bangladesh | We are interested in investigating the relationship between stock prices and macroeconomic variables because individual investors can earn abnormal profits by exploiting this relationship and the existence of this utilizable opportunity would then dangerously distort the market’s ability to proficiently allocate scarce resources. In other word, the stock market will lose its informational efficiency. Informational efficiency is defined as at any given time, stock prices fully reflect all available information of the market. Thus, no investor has an advantage in predicting a return on a stock price because no one has access to information not already available to everyone else. Identifying the relationship or informational efficiency thus can be used to correct the current economic stabilization policies. Therefore, the issue of whether stock prices and macroeconomic variables are related or not have received considerable attention. This paper provides empirical evidence of the relationship between stock prices with each of the macroeconomic variables: exchange rate, inflation rate, money supply variables and so on. The knowledge of the prevailing relationship between stock prices one the one hand, and micro variables like market price/ earnings, growth......

Words: 3958 - Pages: 16

Free Essay

...Empirically Consistent Model for Stock Price and Option Pricing HUADONG(HENRY) PANG∗ Quantitative Research, J.P. Morgan Chase & Co. 277 Park Ave., New York, NY, 10017 Third draft, May 16, 2009 Abstract In this paper, we propose a novel simple but empirically very consistent stochastic model for stock price dynamics and option pricing, which not only has the same analyticity as log-normal and Black-Scholes model, but can also capture and explain all the main puzzles and phenomenons arising from empirical stock and option markets which log-normal and Black-Scholes model fail to explain. In addition, this model and its parameters have clear economic interpretations. Large sample empirical calibration and tests are performed and show strong empirical consistency with our model’s assumption and implication. Immediate applications on risk management, equity and option evaluation and trading, etc are also presented. Keywords: Nonlinear model, Random walk, Stock price, Option pricing, Default risk, Realized volatility, Local volatility, Volatility skew, EGARCH. This paper is self-funded and self-motivated. The author is currently working as a quantitative analyst at J.P. Morgan Chase & Co. All errors belong to the author. Email: henry.na.pang@jpmchase.com or hdpang@gmail.com. ∗ 1 Electronic copy available at: http://ssrn.com/abstract=1374688 2 Huadong(Henry) Pang/J.P. Morgan Chase & Co. 1. Introduction The well-known log-normal model for stock price was......

Words: 7582 - Pages: 31

Premium Essay

...Jaypee Institute of Information Technology, Noida A-10, SECTOR 62, NOIDA, INDIA (12) December, 2009 Sujata Kapoor, JBS, JIIT,Dec’ 2009 TABLE OF CONTENTS S. no. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. Contents Introduction Review of the literature Research objectives Research Methodology Lintner Model: Analysis and findings Factor Analysis and Regression results on Extracted Factors Quadratic Polynomial Regression Analysis & Findings Event study: Analysis & Findings Conclusion Chapter plan Selected References Annexure Page no. 4-9 9-14 14-15 15-23 23-24 24-28 28-30 30-32 32-35 35-36 36-38 (III-XX) 2 Sujata Kapoor, JBS, JIIT,Dec’ 2009 KEY TERMS DIVIDEND PAYOUT RATIO: The percentage of earnings paid to shareholders in dividends. Calculated as: DIVIDEND POLICY: The policy a company uses to decide how much it will pay out to shareholders in dividends. SHAREHOLDERS’ VALUE: The value delivered to shareholders because of management's ability to grow earnings, dividends and share price. In other words, shareholder value is the sum of all strategic decisions that affect the firm's ability to efficiently increase the amount of free cash flow over time. LINTNER MODEL: A model stating that dividend policy has two parameters: (1) the target payout ratio and (2) the speed at which current dividends adjust to the target. AGENCY COST: A type of internal cost that arises from, or must be paid to, an agent acting on behalf of a principal. Agency costs arise because of core......

Words: 16016 - Pages: 65