Beruflich Dokumente
Kultur Dokumente
Nalan Gülpınar
Department of Computing
ng2@doc.ic.ac.uk
1
Contents
1 Introduction to Investment Theory 3
1.1 Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 The Basic Theory of Interest Rates . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Simple Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Compound Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Future Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Present Value (Discounting) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 Annuity Valuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1
7 Option Pricing Models 61
7.1 Binomial Lattice Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
7.1.1 Single-Period Option Pricing . . . . . . . . . . . . . . . . . . . . . . . . . 62
7.1.2 Multi-period Option Pricing . . . . . . . . . . . . . . . . . . . . . . . . . . 64
8 Black-Scholes Equation 65
8.1 Black-Scholes Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
8.2 Analysis of Black-Scholes Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2
1 Introduction to Investment Theory
Financial institutions continuously develop and routinely use very advanced analytical, statis-
tical and numerical techniques in order to create and price financial instruments. Advanced
quantitative techniques are used in the trading of financial instruments as well as risk manage-
ment. As a consequence there is a growing demand for knowledge in the area of quantitative
finance and the analysis of financial markets.
Computational finance is a research area which applies the most appropriate features of com-
puter science, statistics, mathematics to model and solve problems in finance. Sophisticated and
computationally intensive mathematical and statistical techniques for the analysis of real time
financial market data, portfolio management and risk management are needs in modern com-
puterised world. This opens up new opportunities as well as challenges in computational finance
for portfolio management, hedging, speculative international trading and financial decisions.
1.1 Terminology
Finance
Finance is defined as the commercial or government activity of managing money, debt, credit,
and investment.
Investment
In general term, investment is defined as the current commitment of resources in order to achieve
later benefits. If resources and benefits take the form of the money, investment is basically the
present commitment of money for the purpose of receiving more money later. If you invest in
some amount of money, you use your initial money in a way that, your capital will increase, by
paying it into a bank, or buying shares or property. An investor is a person or an organisation
that buys shares or pays money into a bank in order to receive a profit. Investment science is
the application of scientific tools, generally mathematical and computational, to investments.
Cash Flows
There is also a broader viewpoint of investment which is based on the idea of flows of expenditures
and receipts spanning a period of time. The objective of investment is to make the pattern of
flows over time as profitable as possible for the investor. If the expenditures and receipts are
denominated in cash, the receipts at any time period are termed cash flow. Therefore, an
investment is defined in terms of its resulting cash flow sequence; the amount of money that will
flow to and from an investor over time.
Cash flows are either positive or negative to occur at known specific time periods. This is the
case when the flows are known deterministically such as bank interests and mortgage payments.
For example, if the time period is one year, a stream defined by a series of numbers, (−1, 1.20),
corresponds to an initial payment of 1 pound at the beginning of investment period and the
receipt of 1.20 a year later. A cash flow of (−1500, −1000, +3000) describes £1500 and £1000
payments at the beginning of investment period and after a year and receipt of £3000 after two
years. A cash flow is denoted by (a0 , a1 , · · · , an ) at discrete time periods t = 0, 1, 2, · · · , n.
Basic investment problems can be classified as asset pricing, hedging and portfolio selection.
We will broadly describe them below.
Asset Pricing
Given an investment with known payoff characteristics (which may be random), asset pricing
models are concerned with what the reasonable price is or; equivalently what price is consistent
with the other securities that are available. For example, consider an investment opportunity
that will pay exactly £110 at the end of year. The question to be asked here is “How much is
this investment worth today?”. If the current interest rate for one year investment is 10%, then
the investment should have a price of exactly £100. In this case, £110 paid at the end of year
would correspond to a rate of return 10%.
3
Hedging
Hedging is the process of reducing the financial risks that either arise in the course of normal
business operations or are associated with investments. A form of hedging is the insurance where,
by paying a fixed amount (a premium), you can protect yourself against certain specified possible
losses. There are many examples of financial risks that can be reduced by hedging. Hedging can
be carried out in many ways through futures contracts, options, and other special arrangements.
4
1.2.2 Compound Interest
Most bank accounts and loans employ some form of compounding, producing compound interest.
Consider an account that pays interest at a rate of r per year. If interest is compounded yearly,
then after one year, the first year’s interest is added to the original principal to define larger
principal base for the second year. Thus, during second year the account earns interest on the
interest.
Yearly compounding
Let one year compound interest rate be r1 . Under yearly compounding, after a year the principal
A provides W1 which is computed as
W1 = A(1 + r1 )
If the subsequent one-period compound interest rate is r 2 , then the amount hold at t = 2 is
computed by reinvesting on W1 as
W2 = A(1 + r1 )(1 + r2 ) = W1 (1 + r2 )
In other words, after the second year, the money grows by another factor of (1 + r 2 ) to
(1 + r1 )(1 + r2 ). After n years, the initial capital becomes
Wn = A(1 + r1 )(1 + r2 ) · · · (1 + rn )
If the compound interest rates for n years are constant, r 1 = r2 = · · · = rn = r, then after n
years, the principal provides Wn
Wn = A(1 + r)n
This is the analytic expression of the account growth under compound interest. Notice that the
compound interest shows the geometric growth because of its nth-power form. When n increases,
the growth due to compounding can be substantial. The simple and compound interest cases are
shown in Figure 1 for an interest rate of 10%. Notice that simple interest leads to linear growth
over time, whereas compound interest leads to an accelerated increase defined by the geometric
growth.
Example 1.2
Suppose that you invest £1 in a bank account that pays 8% interest rate per year. What
will you have in your account after 5 years under compound and simple interests?
Year Compound Interest Rate Simple Interest Rate
1
t=1 1.08 = 1.(1 + 0.08) 1.08 = 1.(1+0.08)
2
t=2 1.1664 = 1.(1 + 0.08) 1.16 = 1.(1+0.08+0.08)
3
t=3 1.259712 = 1.(1 + 0.08) 1.24 = 1.(1+0.08+0.08+0.08)
4
t=4 1.36048896 = 1.(1 + 0.08) 1.32 = 1.(1+0.08+0.08+0.08+0.08)
5
t=5 1.4693280768 = 1.(1 + 0.08) 1.40 = 1.(1+0.08+0.08+0.08+0.08+0.08)
5
Example 1.3
Assume that initial amount to invest is A = £100 and annual interest rate is constant. What is
the compound interest rate and simple interest rate in order to have £150 after 5 years?
Compound Interest Rate Simple Interest Rate
W5 = A(1 + r)5 W5 = A(1 + 5r)
150 = 100(1 + r)5 150 = 100(1+5r)
150 51 150
1+r = 100 1+5r = 100
r = 1.084 -1 5r = 0.5
r = 8.4 % r = 10 %
1. a year is divided into a fixed number of equally spaced periods–say m periods. For example,
for monthly compounding m = 12 is set.
r
2. The interest rate for each of the m periods is thus m, where r is the nominal annual interest
rate.
r
r k
3. The account grows by (1 + m) during 1 period. After k periods, the growth is 1 + m .
r k
As a result, the initial amount A becomes W = A 1 + m after k periods.
Continuous Compounding
We now divide a year period into smaller periods and apply compounding monthly, weekly, daily
or even, minutes and seconds. This leads to continuous compounding where we consider the limit
of ordinary compounding as the number of periods m in a year goes to infinity;
h r im
lim 1 + = er
m→∞ m
We can also calculate how much an account will have grown after an arbitrary length of time.
Let t denote time measured in years. Select a time t and then divide the year into a number m
k
of small periods. Each period corresponds to a length of 1/m and for some k periods t = m .
Using the general formula for compounding, the growth factor for k periods is obtained as
h r ik h r imt
1+ = 1+
m m
h r ik nh r im o t
lim 1 + = lim 1+ = ert
m→∞ m m→∞ m
As a result, the investment of A (under the continuous compounding) becomes W = Ae rt after
t time periods.
6
Example 1.4
If the nominal interest rate is 8% per year, then with continuous compounding the growth
would be e0.08 = 1.0844 and hence the effective interest rate is 8.44%. Recall that quarterly
compounding produces an effective rate of 8.24%.
Example 1.5
See the following table for the effective interest rates for the different time periods in a year.
Example 1.6
Suppose you get two payments; £5000 today and £5000 exactly one year from now. Put these
payments into a saving account and earn interest at a rate of 5%. What is the balance in your
savings account exactly 5 years from now?
See the following table for interest and balance values.
Year Cash Inflow Interest Balance
(£) (£) (£)
0 5,000 0.00 5,000.00
1 5,000 250.00 10,250.00
2 0 512.50 10,762.50
3 0 538.13 11,300.63
4 0 565.03 11,865.66
5 0 593.28 12,458.94
7
The future value of the cash flow is also computed as follows;
A similar transformation applies to future obligations such as the repayment of debt. In order to
calculate the present value of this obligation, you determine how much money you would need
now to cover the future obligation. The process of evaluating future obligations as an equivalent
present value is also referred to discounting. The present value of the future monetary amount
is less than the face value of that amount, so the future value must be discounted to obtain the
present value. The factor by which the future value must be discounted is called discount factor.
Therefore, the discount factor for a year investment for the example above is
1
d1 =
(1 + r)
If the bank quotes rates with compounding, then such a compound interest rate must be used in
the calculation of the present value. Suppose that the annual interest rate r is compounded at
the end of each of m equal periods per year and W will be received at the end of the kth period.
The discount factor is computed as
1
dk =
r k
1+ m
The present value of a payment of W to be received kth periods in the future is P V k = dk W .
Notice that the present value of initial investment is itself since no discounting is necessary.
However, for the present value of ak , the flow must be discounted k periods.
8
Present Value of Frequent Compounding
Let r be nominal interest rate which is compounded at m equally spaced periods per year. For
n periods, consider the cash flow stream (a0 , a1 , · · · , an ) which is paid initially and at the end of
each period. Then the present value of the cash flow is calculated by the following formula
a1 a2 an
PV = a0 + + + ···+
r n
1+ mr r 2 1 +
1+ m m
n
X ak
=
r k
k=0 1+ m
Example 1.7
Assume that you have just bought a new computer for £3,000. The payment terms are two
years the same as the cash. If you can earn 8% on your money, how much money should you set
aside today in order to make the payment when due in two years?
3000
PV = = £2, 572.02
(1 + 0.08)2
Example 1.8
Consider the cash flow stream (−2, 1, 1, 1). Calculate the present and future values using an
interest rate of 10%.
1 1 1
PV = −2 + + 2
+ = 0.487
1.1 (1.1) (1.1)3
FV = −2 × (1.1)3 + 1 × (1.1)2 + 1 × (1.1) + 1 = 0.648
Example 1.9
FV
Show that the relationship P V = (1+r)n between present value and future value of a cash flow
holds.
In deciding whether or not to undertake an investment NPV is used in the following way: invest
in any project with a positive net present value. This is termed the net present value rule.
9
Example 1.10
At these days buying a two bedroom flat in London costs £150,000 on average. Experts predict
that a year from now it will cost £175,000. Thus you would be investing £150,000 now in the
expectation of the expected £175,000 a year. You have to make a decision on whether you should
buy a flat or invest £150,000 on safe government securities with 6% interest.
You should buy a flat if the present value of the expected £175,000 payoff is greater than the
investment of £150,000. Therefore, you need to ask, ”what is the value today of £175,000 to be
received a year from now, and is that present value greater than £150,000?”. The PV and NPV
are calculated as
175, 000
PV = = £165, 094
1 + 0.06
NPV = £165, 094 − £150, 000 = £15, 094 (2)
There is another way to evaluate investment projects. It involves calculating rates of return on
projects. This rule says that investment is worth under-taking if its rate of return exceeds the
cost of capital. This is termed the rate of return rule. The rate of return on investment in the
residential property is simply the profit as a proportion of the initial outlay.
P rof it
Rate of return =
Investment
175, 000 − 150, 000
= = 0.167 or 16.7%
150, 000
Since the 16.7% return on the residential building exceeds the 6% opportunity cost, you should
go ahead with the investment.
Example 1.11
The following cash flow is obtained from the construction and sale of an office building
Year 0 Year 1 Year 2
-150,000 -100,000 +300,000
Given 7% interest rate, create a present value worksheet and show the net present value, NPV.
t at dt P Vt
0 -150,000 1.0 -150,000
1
1 -100,000 1+0.07 = 0.935 -93,500
1
2 +300,000 (1+0.07)2 = 0.873 +261,900
NPV = 18,400
10
1
If we multiply both side of the last equation with (1+r) we obtain
1 a a a
P VA = + + ···+ (5)
(1 + r) (1 + r)2 (1 + r)3 (1 + r)n+1
By subtracting (5) from (4),
1 a a
1− P VA = − (6)
(1 + r) (1 + r) (1 + r)n+1
r a a
P VA = − (7)
(1 + r) (1 + r) (1 + r)n+1
a 1
P VA = 1− (8)
r (1 + r)n
1
which is equivalent to right side of the formula (3) for d = 1+r . When we consider m
r
periods per year, the interest rate corresponding to per period is i = m and the discount
1 −1
factor is d = 1+ r = (1 + i) . This simplifies Equation (3) as
m
a.(1 − (1 + i)−n )
P VA = (9)
i
Notice that Equations (3) and (9) are equivalent. Suppose that the kth payoff is a(1 + g) k
(instead of a constant cash flow) where g is the growth rate of the payoff per year. The
present value of growing annuity is calculated as
n
a 1+g
P VGA = 1−
r−g 1+r
Example 1.12
Suppose that you borrow £250,000 mortgage and repay over 15 years. The interest rate is 9% an-
nually and payments are made monthly. The effective periodic rate of interest is 9%/12 = 0.75%
per month. What is the monthly payment which is needed to pay off the mortgage?
Given n = 180, m = 12, T = 15, r = 9%, P VA = £250, 000, we can calculate the discount
factor
1
d = r
1+ m
1
= = 0.9925558
1 + 0.09
12
a.d(1 − dn )
P VA =
1−d
a 0.9925558(1 − 0.9925558180 )
250, 000 =
(1 − 0.9925558)
= a(98.59319)
250, 000
a = = £2, 535.67
98.59319
The present value of perpetuity: Perpetuities are assets that generate the same cash flow
forever. In other words, the annuity is called a perpetuity when the number of payments
becomes infinite. Perpetuity pays a coupon at the end of each year and never matures.
The British consol bond is an example of a perpetuity. When n goes to infinity, the present
value of annuity gives the value of a perpetuity. The present value of a stream of payments
of a, discounted at a rate of r, is equal to ar ;
a
P VP = (10)
r
For m periods per year, the formulae (10) becomes
a a
P VP = r = (11)
m
i
11
The present value of growing perpetuity: Similiarly, the present value of the growing per-
petuity can be found as
a
P VGP =
r−g
where g is the growth rate.
Corporate Bond is issued by a corporation, usually through the public securities markets,
which trades publicly just like stocks. They offer a higher yield because they carry a
higher default risk than government bonds. Three major types of corporate bonds are
mortgage bonds, debentures and convertible bonds. Mortgage bonds are secured by real
property such as real estate or buildings. Debentures are unsecured debt and backed only
by the name and goodwill of the corporation. Convertible bonds can be exchanged for the
stock in the corporation.
Municipal Bond is issued by a municipality or state. The advantage of “muni’s” is that the
returns are tax-free and no tax is paid on the interest you earn. Since it is tax-free, the
yield is usually lower than for a taxable bond because the “tax savings” are priced into the
bond. Muni’s are very popular with people in higher tax brackets.
Treasury Bond is issued by a national government. In most cases, they are considered safe
investments because the taxing authority of the government backs them. In the US, interest
on Treasury bonds is not subject to state income tax. T-bonds have maturity greater than
10 years, while notes and bills have lower maturity. They pay coupons twice per year, with
the principal paid in full at maturity.
Treasury Notes are similar to the treasury bonds except a treasury note is issued for a shorter
time period, typically 1 to 10 years. They pay coupons twice per year, with the principal
paid in full at maturity.
12
Treasury Bills are held for a shorter time period, maturity vary from 3, 6, or 9 months (some-
times up to 12 months). There is no coupon payment and return is paid at maturity.
Therefore, bills are priced at a discount. T-bills are zero coupon bonds, so the only cash
flow is the face value received at maturity.
Column 1: Issuer is the company, state (or province), or country that is issuing the bond.
Column 2: Coupon refers to the fixed interest rate that the issuer pays to the lender. The
coupon rate varies by bond.
Column 3: Maturity Date is the date when the borrower will pay the lenders (investors)
their principal back. Typically only the last two digits of the year are quoted, 25 means
2025, 04 is 2004, etc.
Column 4: Bid Price is the price in which someone is willing to pay for the bond. It is quoted
in relation to 100, no matter what the par value is. Think of the bond price as a percentage,
a bond with a bid of 93 means it is trading at 93% of its par value.
Column 5: Yield indicates annual return until the bond matures. Yield is calculated by the
amount of interest paid on a bond divided by the price. It is a measure of the income
generated by a bond.
13
Notation Explanation
P market price of the bond
F principal payment (face or par value)
M annual coupon rate of the bond: yearly coupon payment
m the number of coupon payments per year
c periodic coupon rate (c = M/m)
R APR (Annual Percentage Rate) for today’s cash flows
i effective periodic interest rate (i = R/m)
t the number of years to maturity
λ yield to maturity
n the total number of periods (n = mt)
The coupon amount is described as a percentage of the face value. For example, consider a 20%
coupon bond with par value of £100. The bond has three years from now to maturity and pays
the coupon semi-annually. The annual percentage rate is 13%. The bond will have a coupon of
£20 in a year and £10 after six months. The notation is, therefore,
F = £100, M = £20, c = £10, m = 2, t = 3, n = 6, R = 13%, i = 6.5%.
The price of the bond is simply the sum of the present values of all future payments.
Xn
M
P = P V [F ] + P Vk
m
k=1
X n M
F m
= n +
1+ mR R k
k=1 1 +
m
14
which gives the change of the price (in currency) of the bond in response to a change in the level
of the interest rate. Notice that the sign of the derivative is negative, which means that the
price of the zero coupon bond or the present value of the cash flow decreases with an increase
in the interest rate. Note that the first derivative of the price function is a first order (or linear)
approximation to the slope of the function. This approximation is accurate for small changes
in the interest rate. Hence, the price response ∆P to a change in the interest rates by ∆R is
approximately
∆P = −t(1 + i)−n−1 .F ∆R
In other words, it is an approximation for the absolute price change in pounds ∆P in response
to a shift in the interest rate by ∆R. The percentage response of the bond price is obtained by
dividing the derivative by the value of the bond as follows;
∂P 1 −t(1 + i)−n−1 .F
=
∂R P (1 + i)−n .F
−t
=
1+i
The percentage price change of the bond in response to a change by ∆R is
∆P −t
= ∆R
P 1+i
As it can be easily seen from this expression, the percentage price change of a zero coupon bond
is proportional to the maturity of the bond.
Example 2.2
Consider a zero coupon bond with a term to maturity of 5 years, a face value of £1000. The
interest rate is 8%. The change of price of the bond in response to change in the level of interest
rate is computed as
∂P 1
= −1000 × 5 × (1 + 0.08)−6 × = 31.5085
∂R 100
It means that when interest rate changes by one percentage point, then the bond price changes
by £31.51. See the following table for the comparison of this with the exact numbers obtained
from different interest rates of R = 8%, 9%, 7%.
Interest Rate Bond Value Change (£) Change (%)
8% 680.5832
9% 649.9314 -30.6518 -4.5038
7% 712.9862 +32.4030 +4.7611
Average 31.5274 4.6324
Hence, the error for the case of an interest rate movement of one percentage point is small.
It said that a coupon bond sell
• at face value if the coupon rate is equal to the market interest rate
Bonds Volatility
The absolute value of percentage price change is also called volatility. The factors affecting bond
volatility are level of yield, time to maturity, and coupon rate.
15
2.1.1 Yield to Maturity
Imagine you are interested in buying a bond, at a market price that’s different from the bond’s
par value. There are three numbers commonly used to measure the annual rate of return you are
getting on your investment: coupon rate (annual payout as a percentage of the bonds’ par value),
current yield and yield to maturity. Yield to maturity (YTM) is the percentage rate of return
measuring the total performance of a bond (coupon payments as well as capital gain or loss)
from the time of purchase until maturity. The yield-to-maturity is the best measure of the return
rate, since it includes all aspects of the investment. YTM includes all the interest payments you
will receive up to maturity and also assumes that you reinvest the interest payment at the same
rate as the current yield on the bond.
For bond valuation, the future payments are discounted at the same interest rate R. Now,
we reverse the present value procedure. Given the bond market price, we solve for the interest
rate that equates the present value of the cash flows to its price. The solution is the interest
rate at which the present value of the stream of payments (consisting of the coupon payments
and the final face value redemption payment) is exactly equal to the current price. This value is
termed the yield to maturity. Yield is the interest rate implied by the payment structure.
Suppose that a bond with a face value F makes m coupon payments of M m per period and
there are n periods remaining. The coupon payments sum to M within a year. The current price
of the bond is P . Then the yield to maturity is the value of λ such that the following equality
holds
Xn M
F m
P = +
λ n
1+ m λ k
k=1 1 + m
Note that the first term is the present value of the face value payment. The kth term in the
summation is the present value of the kth coupon payment M m . The sum of the present values is
set to equal to the bond price. As a result, the yield is determined at that particular discount
rate that makes the present value of all future payments to the bondholder equal to the current
market price. There is no easy way to calculate the yield to maturity for coupon-paying bond.
A computer program solves the equation numerically using iterative methods.
λ −n
For a zero coupon bond the bond price is P = F 1 + m . and the yield is found
" 1 #
F n
λ=m −1
P
With continuous compounding the bond price is computed as P = e −λt F . Hence, rear-
ranging and taking logarithms, the yield is obtained as λ = 1t ln F
P
Example 2.3
Consider a zero coupon bond with par value £100. It matures in six years from now and is
trading £55. What is the yield under annual, semiannual and continuous compoundings?
1
100
Annual compounding: λ= 55
6
− 1 = 10.48%
1
Semiannual compounding: λ = 2( 100
55
12
− 1) = 10.22%.
Continuous compounding: λ = 16 ln( 100
55 ) = 9.96%
Notice that the more periods per year, the lower the yield we obtain and the yield with
continuous compounding is always the lowest.
Price-Yield Curve
Price-yield curves show how yield and price are related. Consider bonds with 5%, 10%, 15%
coupon rates and 10 years to maturity. The bond 10% coupon rate means 10% of the face value
is paid each year. Price-yield curves of the bonds are plotted in Figure 3 where the price is
shown as a function of YTM expressed in percentage terms. Yield to maturity varies between 0
and 25%. It can be observed that
16
Figure 3: Price-yield curves
• the price-yield curve has a negative slope; that is, price and yield have an inverse relation.
If yield goes up, price goes down.
• the price of the bond must tend to zero as the yield increases - large yields imply heavy
discounting, so even the nearest coupon payment has little present value. Overall the shape
of the curve is convex.
• when the value of the bond is equal to the par value, it means that the yield is exactly
equal to the coupon rate. The bond is called a par bond.
If the yield is high, then it gives the low volatility. This also implies that volatility is asymmetric
respect to direction of yield change (for large change). For a given yield and maturity, the lower
the coupon rate implies the greater the price volatility.
Example 2.4
This example demonstrates some of problems in using the yield to maturity. Consider two types
of bonds A and B. They both cost £1000, have 3 years maturity, and compounded annually.
The cash flows and yields are presented in the following table.
Bond A Bond B
Year 1 145 430
Year 2 145 430
Year 3 1145 430
Yield (%) 14.5 13
From this table it might appear as if Bond A is better since it has a higher yield. However, the
yield to maturity rule might not lead us the higher returns since there is an assumption of equal
annual rates of return in computing the yield to maturity.
When we consider the different interest rates over periods 1, 2, and 3, this might not be the
case. Suppose i1 = 10%, i2 = 20%, i3 = 15% interest rates prevailing over periods 1, 2, and 3,
respectively. The present value of two cash flows for the bonds A and B are
145 145 1145
P (A) = + + = £996
1.1 1.1 × 1.2 1.1 × 1.2 × 1.15
430 430 350
P (B) = + + = £1000
1.1 1.1 × 1.2 1.1 × 1.2 × 1.15
From these calculations, the present value of bond B is greater than the present value of bond
A. As a result, the yield to maturity rule does not always work as a guide to higher returns, as
we vary the pattern of i1 , i2 , i3 .
17
Time to Maturity
Now let’s consider the influence of the time to maturity on price of the bond. Bonds with long
maturities have steeper price-yield curves than bonds with short maturities. Hence the prices
of long bonds are more sensitive to interest rate changes than those of short bonds. Figure 4
presents the price-yield curves for three different bonds. Each bond has 10% coupon rate with
300
3-Years
250
10-Years
15-Years
200
Prices
150
100
50
0
0 5 10 15 20 25
Yield to maturity
par value of £100, and 3 years, 10 years, and 15 years to maturity. The bonds are at par when
the yield is 10%; hence the curves all pass through the common par point. However, the curves
pivot upward around that point by various amounts, depending on the maturity. As maturity
is increased, the price-yield curve becomes steeper. This indicates that longer maturities imply
greater sensitivity of price to yield. As it will be discussed in the next section, duration is high
if maturity is long; so higher maturity implies higher volatility.
Bond holders are subject to yield risk as well as the interest rate risk. If yields change, bond
prices also change. This is immediate risk, affecting the near-term value of the bond. You may
continue to hold the bond and thereby continue to receive promised coupon payments and the
face value at maturity. The cash flow stream is not affected from interest rates; however, if you
plan to sell the bond before maturity, the price will be governed by the price-yield curve.
Example 2.5
Show the price-yield relation in tabular form for bonds with a 8% coupon rate, 1, 3, 5, and 10
years to maturity and 3%, 5%, 8%, and 10% yield values, respectively.
Time to Yield
Maturity 3% 5% 8% 10%
1 year 104.85 102.85 100 98.18
3 years 114.14 108.16 100 95.02
5 years 122.89 112.98 100 92.41
10 years 142.65 123.16 100 87.71
For example, the price of bound with yield to maturity of λ = 5%, and 3 years to maturity
is calculated as
100 8 8 8
P = + + +
(1 + 0.05)3 (1 + 0.05) (1 + 0.05)2 (1 + 0.05)3
P = £108.1697
Notice that the bond with 10 years to maturity is much more sensitive to yield changes than the
one with 1 year maturity. For the 8% coupon bond at yield λ = 8%, the price will be equal to
the par value of F = £100.
18
2.1.2 Duration
Maturity itself does not give a complete quantitative measure of interest rate sensitivity. Another
measure of time length termed duration gives a direct measure of interest rate sensitivity. It is
calculated as a weighted average of the times that cash flow payments are made. The weighting
coefficients are the present values of the individual cash flows.
Suppose that cash flows a0 , a1 , · · · , an are received at times t0 , t1 , t2 , · · · , tn , respectively. The
duration of the stream, which has units of time, can be formalised as
P V (a0 )t0 + P V (a1 )t1 + · · · + P V (an )tn
D =
PV
where P V (ak ) is the present value of cash flow ak at k and P V is the total present value of the
cash flow. The expression for D is indeed a weighted average of the cash flow times. Hence D
itself has units of time. When the cash flows are all nonnegative, then it is clear that t 0 ≤ D ≤ tn .
Duration is a time intermediate between the first and the last cash flows. It is a fact that a
zero-coupon bond, which makes only a final payment at maturity, has a duration equal to its
maturity date, D = T . Nonzero-coupon bonds have durations strictly less than their maturity
dates. This shows that duration can be viewed as a generalised maturity measure. If maturity
is long, then duration is high. So higher maturity implies higher volatility.
Macaulay Duration
The definition of duration given above is a bit vague about how the present value is calculated,
what interest rate to use. For a bond it is natural to base those calculations on the bond’s yield.
If the yields is used, the general duration formula (D) becomes the Macaulay duration, (D mac ).
When time periods are given in terms of years as t = 1, 2, · · · , T and the corresponding cash flow
at , the Macaulay duration is defined as
a1 a2 aT
1× (1+λ) + 2 × (1+λ)2 + · · · + T × (1+λ)T
Dmac = a1 a2 aT
(1+λ) + (1+λ)2 + · · · + (1+λ)T
T
X T
X
at
t× t × P V (at )
t=1
(1 + λ)t t=1
= =
T
X at P
t=1
(1 + λ)t
Suppose that a financial instrument makes payments m times per year, with the payment in
period k being ak and there are n periods remaining. The Macaulay duration becomes
n
X k ak
λ k
m (1 + m
k=1
)
Dmac =
PV
n
X ak k
where λ is the yield to maturity and P V = . Note that the factor m in the
λ k
1+
k=1 m
numerator of the formula for Dmac is time, measured in years.
Example 2.6
A bond has a maturity of 5 years with a coupon payments of £50 and principal repayment of
£500 on redemption. The current price of the bond is £539.93. Find its yield to maturity and
duration.
The cash flow is (50, 50, 50, 50, 550) for t = 1, 2, 3, 4, 5 and the price P = £539.93. Recall
that yield to maturity is the rate of interest that equates the PV of future cash flows to the price
and computed as
50 50 50 50 50 + 500
539.93 = + + + +
(1 + λ) (1 + λ)2 (1 + λ)3 (1 + λ)4 (1 + λ)5
By solving the equation, λ = 8% is found.
19
t at P V (at ) t × P V (at )
1 50 46.30 46.30
2 50 42.87 85.73
3 50 39.69 119.08
4 50 36.75 147.01
5 550 374.32 1871.62
P 539.93
5
X
t × P V (at ) = 2269.73
t=1
2269.73
Then Macaulay duration is Dmac = 539.93 = 4.20 years. This is the average time to receipt of
the cash flows.
Modified Duration
Duration is used to measure directly the sensitivity of price to changes in yield. This also follows
a simple expression for the derivative of the present value of expression. In the case where
payments are made m times per year and the yield is based on those same periods, we have
P V (ak ) = akλ k . The first order derivative of P V (ak ) with respect to λ is
[1+ m ]
k k
dP V (ak ) m ak m
= − = − P V (ak )
dλ λ k+1 1+ λ
1 +m m
n
X
Similiarly, the first order derivative of bond price P = P V (ak ) with respect to λ is calculated
k=1
as
n
X n
X k
dP dP V (ak ) m
= =− λ
P V (ak )
dλ dλ 1+ m
k=1 k=1
1
= − λ
Dmac P
1+ m
≈ −DM P
where DM = D λ and is called the Modified duration. It is the usual duration modified by the
mac
1+ m
extra term in the denominator. This measures the relative change in a bond’s price directly as
λ changes. By using the approximation dP ∆P
dλ = ∆λ , the change in price due to a small change in
yield (or vice versa) can be estimated as
∆P ≈ −DM P ∆λ (12)
Explicit values for the impact of yield variations can be obtained by using ∆P = P N − P0 and
∆λ = λN − λ0 (where PN and P0 are the new and old prices and λN and λ0 are the new and
old values of the yields) as
PN ≈ P0 − DM P0 (λN − λ0 )
Example 2.7
Consider the bond given in Example 2.6. What is the new price of the bond if yield increases
from 8% to 8.1%? Use the Modified Duration.
The Modified duration is
4.20
DM = = 3.89 years
1 + 0.08
The change on price is obtained as
∆P ≈ −DM P ∆λ
≈ −3.89 × 539.93 × (0.081 − 0.08) ≈ −2.1003277
20
and the new price is calculated as
PN − P O ≈ −2.1003277
PN ≈ £537.83
As a result, price decreases from £539.93 to £537.83 when the yield increases from 8% to 8.1%.
2.1.3 Convexity
Modified duration measures the relative slope of the price-yield curve at a given point. This
leads to a linear approximation to price-yield curve. However, a better approximation can be
obtained by including a second order (quadratic) term which is based on convexity, C. Convexity
is defined as
1 ∂2P
C =
P ∂λ2
This is the relative curvature at a given point on the price-yield curve. Given a cash flow a t , for
t = 1, 2, · · · , n, the convexity is calculated as
n
1 X d2 P V (ak )
C =
P dλ2
k=1
n
X
1 k(k + 1)ak
= 2
(13)
P (1 + λ) (1 + λ)k
k=1
Xn
1
= k(k + 1)P V (ak )
P (1 + λ)2
k=1
Suppose that a financial instrument makes payments m times per year, with the payment in
period k being ak and there are n periods remaining. Then Equation (13) becomes
n
X
1 k(k + 1) ak
C =
λ 2 m2 λ k
P 1+ m k=1 1+ m
This reflects the sensitivity of modified duration to changes in yield. Assume that price P and
the corresponding yield λ are given and DM , C are calculated. The second order approximation
to the price-yield curve is
1
∆P ≈ −DM P0 ∆λ + P0 C(∆λ)2 (14)
2
Let ∆λ be change on yield from λ0 to λN and ∆P be the corresponding change on price of bond
from P0 to PN . The new price of bond is found as
1
PN ≈ P0 − DM P0 (λN − λO ) + P0 C(λN − λO )2
2
21
Expectation Theory (Fisher) Long-term rates should reflect expected future short-term rates.
It implies that the implied forward rate for a given period is equal to the expected futures
zero rate for that period. Theory starts with assuming that market participants which trade
on default free bonds are risk neutral, i.e. are indifferent to risk. They have no preference
for a maturity over another and seek bonds with highest expected return. Markets are
assumed to operate efficiently. Arbitrage across maturities forces interest rates on different
assets to conform to market expectations of the appropriate rate for each instrument.
Liquidity Preference Theory (Keynes and Hicks) This theory suggests that market in-
vestors prefer liquidity and they demand compensation in term of higher yield, for moving
to longer maturity. In this theory market participant as assumed risk-averse and in gen-
eral long term bonds are less liquid because higher bid-ask spread and are more volatile.
Borrowers on the other hand prefer to borrow at a fixed rate for a long period of time.
Financial intermediaries do not want to finance long-term fixed rate loans with short-term
deposits because of interest rate risk. To match depositors and borrowers than financial
intermediaries would rise long-term interest rates. So long-term interest rates are deter-
mined by market expectation plus a liquidity premium. The theory implies that implied
forward rates are upward-biased estimates of future zero rates.
Preferred Habitat Theory (Modigliani and Sutch) It says that different investor have
preferences for different maturities because of the nature of their liabilities or their risk
aversion. Participant are reluctant to switch across maturities. So securities of different
maturities are imperfect substitute for one another and intertermporal arbitrage does not
work. Thus the short-term rate is determined by demand and supply in the short-term
market, the medium-term rate is determined by demand and supply in the medium-term
market, and the long-term rate is determined by demand and supply in the long-term
market. The theory implies that forward rates are biased estimates of future zero rates;
this can be upward or downward biased.
Under yearly compounding convention: The spot rate st is defined such that (1 + st )t is
the factor by which a deposit held t years will grow. Then A becomes A(1 + s t )t
st mt
Under a convention of compounding m periods per year: The factor is (1 + m) and
st mt
the account grows to A(1 + m ) .
Under continuous compounding convention: The growth factor is est t and the capital pro-
vides Aest t after t years.
Assume that the 1-year spot rate s1 is known. In order to determine 2-year spot rate, we can
solve the following the equation for s2 ,
M M +F
P = +
1 + s1 (1 + s2 )2
where the bond has coupon payments of amount M at end of both years with a face value of F
and its current price P . This basically means that the price should equal to discounted value of
the cash flow stream. Carrying out forward in this way we can determine s 3 , s4 , · · · step by step.
Once the spot rates have been determined the corresponding discount factors d k at time period
k can be defined for various compounding conventions as follows;
1
For yearly compounding, dk = (1+sk )k
22
1
For compounding m periods per year, dk = sk mk
(1+ m )
(1 + sj )j = (1 + si )i (1 + fi,j )j−i
where si and sj are spot rates at i and j, respectively. The left side represents the factor by
which money grows if it is directly invested for j years with spot rate s j . The right side is the
factor by which money grows if it is invested first for i years with spot rate of s i and then in
a forward contract (arranged today) between years i and j (i < j) with rate of f ij . The term
(1 + fi,j ) is raised to the (j − i)th power because the forward rate is expressed in yearly terms.
Under various compounding conventions the forward rates can be determined as follows;
For yearly compounding, forward rate fi,j satisfies (1 + sj )j = (1 + si )i (1 + fi,j )j−i and is
calculated as
1
(1 + sj )j (j−i)
fi,j = −1
(1 + si )i
ssi i fi,j j−i
For compounding m periods per year, forward rate satisfies (1+ mj )j = (1+ m ) (1+ m )
and is obtained as
s 1
(1 + mj )j (j−i)
fi,j = m si i −m
(1 + m )
23
For continuous compounding, forward rate fi,j satisfies esj j = esi i efi,j (j−i) and is calculated
as
sj j − s i i
fi,j =
j−i
Example 2.9 Taken from Luenberger’s Book Chp 4 - Exercises 1
The spot rates for 1 and 2 years are s1 = 6.3% and s2 = 6.9%. What is the forward rate f1,2 ?
(1 + s2 )2
f1,2 = −1
1 + s1
(1.069)2
= − 1 = 7.5%
1.063
Common Stock is just that, “common”. The majority of stocks trading today are in this form.
Common stock represents ownership in a company and a portion of profits (dividends).
Investors also have voting rights (one vote per share) to elect the board members who
oversee the major decisions made by management. In the long term, common stocks, by
means of capital growth, yield higher rewards than other forms of investment securities.
This higher return comes at a cost as common stocks entail the most risk. Should a
company go bankrupt and liquidate, the common shareholders will not receive money
until the creditors, bond holders, and preferred shareholders are paid.
Preferred Stock represents some degree of ownership in a company but usually don’t have
voting rights (this may vary depending on the company). On preferred shares investors
are guaranteed a fixed (or sometimes variable) dividend forever, meanwhile common stocks
have variable dividends. However, one advantage is in the event of liquidation; they are
paid off before the common shareholder (but still after debt holders). Preferred stock may
also be callable, meaning that the company has the option to purchase the shares from
shareholders at anytime, and usually for a premium.
Investors purchase stocks for their returns which come in the form of capital gains (the appreci-
ation in value over time) and dividends (paid periodically by most companies). There are three
ways of transacting in stocks;
Buy: The stock will appreciate in value over time, or require the stock for its characteristics as
part of portfolio. It is also said that we are long in the stock.
Sell: The stock will depreciate in value over time or we require funds for another purpose.
24
Short sale: It is also said that we are short in the stock. Here, we do not own the stock; but
we borrow it from another investor, sell it to a third party and receive the proceeds. We
are obligated to pass on to the lender of the stock any dividends declared on the stock
and also to pay to the lender the market price of the stock if we decide to sell. When the
short sale is the case, it is expected that the stock will decline in value in order to enable
us to buy it back at a low price later on to make up out obligations to the lender. We are
expecting a bearish market for the stock.
Columns 1 and 2: 52-Week Hi and Low are the highest and lowest prices that a stock has
traded at over the previous 52-weeks (1 year). This typically does not include the previous
day’s trading.
Column 3: Company Name (and Type of Stock) lists the name of the company. If there
are no special symbols or letters following the name, it is common stock. Different symbols
imply different classes of shares. For example, “pf” means the shares are preferred stock.
Column 4: Ticker Symbol is the unique alphabetic name which identifies the stock on the
exchange’s ticker. The ticker tape will quote the latest prices alongside this symbol. If you
are looking for stock quotes online, you always search for a company by the ticker symbol.
Column 5: Dividend Per Share indicates the annual dividend payment per share. If this
space is blank, the company does not currently pay out dividends.
Column 6: Dividend Yield is the percentage return for the dividend, which is calculated as
annual dividends per share divided by price per share.
Column 7: Price/Earnings Ratio is calculated by dividing the current stock price by earn-
ings per share from the last four quarters.
Column 8: Trading Volume shows the total number of shares traded for the day, listed in
hundreds. To get the actual number traded, add ”00” to the end of the number listed.
Column 9 and 10: Day High - Low indicate the price range the stock has traded at through-
out the day’s trading. In other words, these are the maximum and the minimum price
people have paid for the stock.
Column 11: Close is the last trading price recorded when the market closed on the day. If the
closing price is up or down more than 5% than the previous day’s close, the entire listing
for that stock is bold-faced. Keep in mind you are not guaranteed to get this price if you
buy the stock the next day. Because a stock price is constantly changing (even after an
exchange is closed for the day) the close merely serves as an indicator of past performance.
Column 12: Net Change is the dollar value change in the stock price from the previous day’s
closing price. When you hear about a stock being: “up for the day” it means the net change
was positive.
25
3.2 Valuation of Stocks
Stock prices are changed everyday by “the market”. Buyers and sellers cause prices to change
as they decide how valuable each stock is. Basically, share prices change because of supply and
demand. If more people want to buy a stock than sell it, then the price moves up. Conversely,
if more people want to sell a stock, there would be more supply (sellers) than demand (buyers);
therefore, the price would start to fall.
Stock represents an ownership in a company. Therefore, the price of a stock shows what
investors feel the company is worth. Stock prices can change at any rate, some have dramatic
swings in one day while others stay the same for a long time. There are hundreds of variables
which drive stock prices, the most important of which is earnings. Think of earnings as the profit
of a company, the money left after all expenses have been paid, this is what shareholders desire.
Securities issued by the firm have no set maturity date or dividend rate; however, the par
value is set by the firm. Unlike bonds, there are many approaches to the valuation of stocks.
The reason is that cash flows associated with stocks are extremely difficult to estimate. In this
section we will consider ”discounted dividend model” where the stock price is the present value
of future cash flows to be received by the investor. The price the investor is willing to pay for a
share of stock depends upon magnitude, timing of expected future dividends and the risk of the
stock.
If the investor buys a stock, then he is entitled to receive all future dividends and can sell
the stock in the future. The cash flow to holders of common stock consists of dividends plus a
future sale price. Consider the following cash flow which investors expect to receive.
Time 0 1 2 ··· T ···
Cash Flow D1 D2 ··· D T + PT ···
where the current dividend is denoted by D0 . The stock’s discount rate re is the rate of return
that investors expect to earn on securities with similar risk. Shareholders require a rate of return
for buying a share. They buy for P0 and sell after one year for P1 and receive dividend D1 .
D1 + P1
P0 =
1 + re
The next buyer also sells after one year:
D2 + P2 D1 D2 + P2
P1 = =⇒ P0 = +
1 + re 1 + re (1 + re )2
This formula shows that the current value of stock is the present value of all future cash flows,
PT
i.e., dividends and expected selling price. Note that the expression (1+r e)
T can be neglected for a
T
large time horizon since (1+re ) becomes very large as T becomes very large. The disadvantages
of this approach are that the value of T is determined by the investor, the dividends are uncertain
and we need to estimate the expected selling price in order to calculate the current price.
Required Returns
D1 +P1
Consider P0 = 1+re . We can reexpress it in terms of returns as
D1 P1 − P 0
re = + (16)
P0 P0
The first part on the right hand side is a financial ratio widely used by practitioners and called
as the dividend yield. However, in practice, D1 is not known since it is an expected value about a
26
future dividend payment. Practitioners commonly refer to the dividend yield as D 0 /P0 . Because
of this difference we will refer to D0 /P0 as the historic or trailing dividend yield, and to D 1 /P0
as the prospective dividend yield. The second part on the right hand side of (16) is the capital
gain, which is the percentage of the current stock price. As a result, the return on equity is the
sum of prospective dividend yield and expected capital gain.
The Zero Dividend Growth Model: The simplest assumption about dividends is that they
stay constant over time with zero growth, so that D 1 = D2 = · · · = D̄. Then the formula
(15) becomes
D̄ D̄
P0 = =⇒ re = (17)
re P0
As a result, if the dividend is expected to stay constant over time (the cash flow is always
the same), then shares can be valued like perpetual bonds and the expected return on equity
is equal to the dividend yield. This model does not reflect reality because of constancy of
dividends. More general assumption is the constant dividend growth.
The Constant Dividend Growth Model: An amount that grows at a constant rate forever
is called a growing perpetuity. The stock with a constant dividend growth is actually a
growing perpetuity. Let g be a constant rate which dividends grow forever. Therefore,
dividends are computed as
D2 = D1 (1 + g)
D3 = D2 (1 + g) = D1 (1 + g)2
D4 = D3 (1 + g) = D1 (1 + g)3
···
DT = DT −1 (1 + g) = D1 (1 + g)T −1
When we substitute them in Equation (15), the current price of the stock is obtained as
D1 D1 (1 + g) D1 (1 + g)2 Dt (1 + g)T −1
P0 = + 2
+ 3
+ ···+ + ···
1 + re (1 + re ) (1 + re ) (1 + re )T
Assume that g is smaller than re . In this case the general formula (15) becomes
D1
P0 = (18)
re − g
Notice that it gives the equation (17) when g = 0. From (18), we can obtain the expected
return on equity as
D1
re = +g (19)
P0
which is the sum of prospective dividend yield and growth rate. Using (16) with (19), the
growth rate is obtained as g = P1P−P
0
0
and therefore, P1 = (1 + g)P0 . Hence, if we assume
that the company is in steady state where dividends are expected to grow at a constant
rate g, we also expect that the stock price grows at the same constant rate g.
Example 3.1
If the dividend at t = 1 is £2 and the expected growth rate is 5%, what is the dividiend
D5 at t = 5?
27
Example 3.2
Next year dividends per share for company X is expected to be £0.95. The dividends are
expected to grow at 14% per year in the future. What should be the current price if the
required rate of return is 16% per year?
Non-constant Dividend Growth Model: The model with a constant dividend growth rate
is not a suitable for the companies which are not in a steady state. For simplicity, we can
view g as a kind of average. In this case, we need to extend the constant dividend growth
model and define sub-periods with different growth rates. Then we can estimate the value
of the stock by considering each sub-period with individual discounting in two steps:
Dt+1
Pt = (21)
re − g
Then the present value of the stock is basically sum of present values of all dividends as
P0 = P V (N CD) + P V (Pt )
Example 3.3
The next three years dividends for Company Y are expected to be £0.50, 1.00, 1.50. Then
the dividends are expected to grow at a constant rate of 5% forever. If the required return
is 10%, what is the value of the stock?
For the non-constant dividend growth
3
X Dk (N C)
P V (N CD) =
(1 + re )k
k=1
0.50 1.00 1.50
= + +
(1 + 0.10) (1 + 0.10)2 (1 + 0.10)3
= 0.454 + 0.826 + 1.127 = 2.407
For the constant dividend growth based on the third year dividend D 3 ,
D3 (1 + g)
P3 =
re − g
1.5(1 + 0.05)
= = 31.50
0.10 − 0.05
P3 31.50
P V (P3 ) = 3
= = 23.67
(1 + re ) 1.331
Then P0 = P V (N CD) + P V (P3 ) = 2.407 + 23.67 = 26.07 is the present value of the stock.
28
Dividend Forecast
The general stock valuation formula is a function of all future dividends. A forecasting method
for dividends D1 , D2 , · · · , D∞ which extends into all periods is needed.
If we forecast the values of D1 , D2 , · · · , DT for a finite time horizon T , then the price of
stock becomes
T
X Di PT
P0 = i
+ (22)
i=1
(1 + re ) (1 + re )T
Then apply the constant growth formula for the horizon value P T by substituting for the
dividends after period T ,
DT +1 (1 + g)DT
PT = =
re − g re − g
then the general stock valuation formula given in Equation (22) becomes
T
X Di (1 + g)DT
P0 = +
i=1
(1 + re )i (1 + re )T (re − g)
If we forecast the dividends one period into the future, then a special version of (22)
for T = 1 is obtained. In this case, we obtain the standard equation for the dividend
growth model as
D1
P0 =
re − g
If we do not have a forecast for this year’s dividend, then the historic dividend value D 0
can be used so that D1 = D0 (1 + g). The stock price is obtained as
(1 + g)D0
P0 =
re − g
Example 3.4
Consider a company pays a dividend of £0.75 per share. Demand for this company’s product is
growing at 2% per year and inflation averages 2.5% per year. The company expects its profits
and dividends to grow at about 4.55% per year (1.02 × 1.025 = 1.0455). Stockholders require a
10% rate of return. What is the market price of this company’s stock?
The dividend next period is 0.75 × 1.0455 = 0.0784. Using the formula for a growing perpe-
tuity P = (reD−g)
1
, we obtain
0.75 × 1.0455
P = = £14.39
0.10 − 0.0455
E1 P0 p
re = × p + g =⇒ =
P0 E1 re − g
29
Therefore, companies should have the same P/E ratio if they have the same pay out ratio
(similar technology, efficiency), discount rate (business risk, financial risk), growth rate (business
prospects, market share development). Analysts often report the historical (or trailing) price-
earnings ratio, P0 /E0 where E0 is the current earnings. The relation between the historical and
forward-looking P/E can be seen from the following equation
P0 P0 1
=
E1 E0 (1 + g)
where E1 = E0 (1 + g). This shows that the historical P/E overstates the forward-looking P/E.
When it is assumed that dividends and earnings grow at the same constant rate g from now on,
(that is D1 = (1 + g)D0 , E1 = (1 + g)E0 ), the required return and P/E ratio are obtained as
D0
re = g + (1 + g) (23)
P0
P0 P0 p(1 + g)
= (1 + g) =
E0 E1 re − g
D
re − P 0 D0
Rearranging (23), we can find g = D
0
where P0 is referred as the historical dividend yield.
1+ P 0
0
Each pj can be thought of as the relative frequency with which y j would occur if an
experiment of observing yj were repeated infinitely often. The quantity y, characterised in
this way before its value known, is called a random variable.
Expected Value: The expected value (mean value or mean) of a random variable y is the
average value obtained by regarding the probabilities as frequencies. For the case of a
finite number of possibilities, it is defined as
m
X
E(y) = p j yj (24)
j=1
30
Variance: Variance is the measure of the degree of possible deviation from the mean. Given
the expected value, ȳ, of a random variable y, the variance of y is the expected value of
the squared variable (y − ȳ)2 , which is the measure of how much y tends to vary from its
expected value. It can be mathematically formulated as
var(y) = E (y − ȳ)2 = E(y 2 ) − 2E(y)ȳ + ȳ 2 (25)
= E(y 2 ) − ȳ 2
cov(y1 , y2 ) σ12
ρ12 = p = (27)
var(y1 )var(y2 ) σ1 σ2
The sign of the correlation coefficient defines the direction of the relationship, either positive
or negative. If ρ12 > 0, then two variables are said to be positively correlated. This means
that as the value of one variable increases, the value of the other variable increases; as one
decreases the other decreases. If If ρ12 < 0, the two variables are said to be negatively
correlated. This indicates that as one variable increases, the other decreases, and vice-
versa. When ρ12 = +1 (or −1), then it is called perfect positive (or negative) correlation.
If two random variables are independent, then they are called uncorrelated and σ 12 = 0.
Asset return: An investment instrument that can be bought and sold is called an asset. The
amount of money to be obtained when selling an asset is uncertain at the time of purchase.
In this case, the return is random and the uncertainty can be described in probabilistic
terms. Suppose that you purchase an asset at time zero and one year later you sell it. The
total return on this investment is defined as
amount received
total return = (28)
amount invested
Let x0 and x1 be amounts of money invested and received after one year, respectively. The
total return R is formalised as
x1
R= (29)
x0
The rate of return is defined as
amount received − amount invested
rate of return =
amount invested
and formulated
x1 − x 0
r= (30)
x0
Notice that total return and rate of return are related by R = 1 + r and the equation (30)
can be rewritten as
x1 = (1 + r)x0
This shows that the rate of return acts much like an interest rate.
31
Example 4.1
and plotted in Figure 6. Although both assets have the same mean return value of r¯1 =
r¯2 = 5, their standard deviations are σ1 2 = 3.535534 and σ2 2 = 6.284903.
n
X n
X
Equivalently, we have r = wi ri , since wi = 1.
i=1 i=1
Proposition 1 Both the total return and the rate of return of a portfolio of assets are
equal to the weighted sum of the corresponding individual asset returns, with the weight of
32
an asset being its relative weight in the portfolio; that is,
n
X n
X
R= wi R i , r= wi r i (32)
i=1 i=1
The mean return of a portfolio is calculated by taking the weighted sum of the individual expected
rates of return as
n
X
E(rp ) = E(r1 )w1 + · · · + E(rn )wn = wi E(ri )
i=1
Example 4.2
Consider a portfolio of a risk-free asset and two risky stocks. Suppose that there exists three
equally likely states presented in the scenario tree presented in Figure 7. The root node represents
today and the future uncertainty is discretised by three events (states). The following table
displays the asset returns for each state s. What is the expected return of the portfolio formed?
Returns (%)
States T-Bill Stock A Stock B
s i=1 i=2 i=3
s1 − boom 5 16 3
s2 − normal 5 10 9
s3 − recession 5 1 15
p p p
1 2 3
s s s
1 2 3
E[r] = 5 × w1 + (w2 + w3 ) × 9
1
For equally weighted portfolio, w1 = w2 = w3 = 3 and the expected return of the portfolio is
1
E[rew ] = [5 + 2 × 9] = 7.666
3
33
Risk
According to Investopedia dictionary, “risk” is defined as the chance that an investment’s actual
return will be different than expected. This includes the possibility of losing some or all of the
original investment. Those who work hard for every penny earned have a harder time parting
with money. These people are considered to be more risk averse. On the other end of the
spectrum, day traders feel if they aren’t making dozens of trades a day there is a problem, these
people are risk loving. When investing in stocks, bonds, or any investment instrument there is
a lot more risk than you’d think. Two basic types of risk are described here.
There are several types of risk that a smart investor should consider and pay careful attention
to. Deciding your potential return while respecting risk is the age old decision that investors
must make. More specific types of risk, particularly when we talk about stocks and bonds are
summarised as follows;
Credit Risk is a risk that a company or individual will be unable to pay the contractual interest
or principal on its debt obligations. It is sometimes called default risk.
Country Risk is a risk that a country won’t be able to honor its financial commitments. When
a country defaults it can harm the performance of all other financial instruments in that
country as well as other countries it has relations with.
Foreign Exchange Risk appears when investing in foreign countries. The fact that currency
exchange rates can change the price of the asset as well must be considered.
Interest Rate Risk is a risk that interest rates will rise during the term of your investment.
A rising interest rate hurts the performance of stocks and bonds.
Political Risk is a financial risk that a government in a country will suddenly change its poli-
cies. This is a major reason that second and third world countries lack foreign investment.
Market Risk is the one we are all familiar with. It’s the day to day fluctuations in a stocks
price and referred to volatility.
Portfolio Risk
There are alternative risk measures such as value at risk (VaR), downside risk and utility func-
tions. In the mean-variance framework, risk is measured as the variance of the portfolio return.
Let σi2 and σp2 denote the variances of the return of asset i and portfolio, respectively. The
covariance of return of asset i with asset j is σij . The variance of rate of return of the portfolio
is calculated as follows;
!2
Xn n
X
σp2 = E (r − r̄)2 = E wi r i − wi r¯i
i=1 i=1
! n
n
X X
= E wi (ri − r¯i ) wj (rj − r¯j ) (33)
i=1 j=1
n
X n
X
= E wi wj (ri − r¯i )(rj − r¯j ) = wi wj σij
i,j=1 i,j=1
34
Example 4.3
Consider the portfolio given in Example 3.2. What is variance of the portfolio?
The variances of assets i = 1, 2, 3 are computed as follows;
σ12 = V ar[r1 ] = 0
1 114
σ22 = V ar[r2 ] = (16 − 9)2 + (10 − 9)2 + (1 − 9)2 =
3 3
1 72
σ32 = V ar[r3 ] = 2 2
(3 − 9) + (9 − 9) + (15 − 9) =2
3 3
and covariances are
1 −90
σ23 = Cov[r2 , r3 ] = [(7)(−6) + (1)(0) + (−8)(6)] =
3 3
σ21 = Cov[r2 , r1 ] = 0
σ31 = Cov[r3 , r1 ] = 0
35
Figure 8: An optimal portfolio
risky portfolio” is usually determined to be somewhere in the middle of the curve, this is because
as you go higher up the curve you take on proportionately more risk for a lower incremental
return and so you’re assuming a larger amount of risk for a smaller increase in returns. On the
other end, low risk/low return portfolios are pointless because you can achieve a similar return
by investing in risk-free assets like government securities.
36
Maximum Expected Wealth Problem: This problem maximises the expected portfolio re-
turn. It is a risk-neutral approach which does not take risk attitudes into account. This
can be achieved by the following linear programming problem
n
X
max wi r i
w
i=1
subject to
Xn
wi = 1 (34)
i=1
wi ≥ 0 i = 1, · · · , n
Minimum Variance Portfolio: The minimum variance portfolio is obtained by solving the
following quadratic programming problem which only takes risk into account.
n X
X n
min wi wj σij
w
i=1 j=1
subject to
Xn
wi = 1 (35)
i=1
wi ≥ 0 i = 1, · · · , n
Mean Variance Model: The mean-variance problem describes basically the trade off be-
tween expected rate of return and variance of rate of return in a portfolio. It naturally
leads to diversification by taking risk attitudes into account. The mean-variance problem
can be stated as follows;
n X
X n
min wi wj σij
w
i=1 j=1
subject to
Xn
wi ri = r̄ (36)
i=1
Xn
wi = 1
i=1
wi ≥ 0 i = 1, · · · , n
The first constraint presents the expected portfolio return fixed at a certain value. The
other linear constraints describe the allocation of initial investment and bounds on decision
variables. Notice that the non-negativity of weights are to prevent short selling. Notice
that the mean-variance model becomes the minimum variance problem when the expected
wealth constraint is ignored.
Financial reality dictates that the highest performing portfolio strategy is also the most risky
efficient strategy available. In order to obtain other points on the Markowitz efficient frontier,
it is necessary to consider risk (variance) in conjunction with the mean return. In this case the
required expected return can be provided as constant r̄.
The risk-return tradeoff could easily be called the “sleep at night” test. Deciding what
amount of risk you can take on while still being able to get a comfortable rest at night without
worrying yourself to death is a most important decision. The risk/return tradeoff is the balance,
an investor must decide on between the desire for the lowest possible risk and the highest possible
returns. Remember to keep in mind that low levels of uncertainty (low risk) are associated with
low potential returns and high levels of uncertainty (high risk) are associated with high potential
returns. A higher standard deviation means a higher risk, which for the most part means a higher
potential return. How do you know what risk is most appropriate for you? This isn’t an easy
question to answer. The amount of risk you can comfortably undertake differs from person to
person. Generally, if you are having anxiety attacks every time when the market moves up or
down, then perhaps you should consider reducing your risk.
37
Markowitz Solution without Short-sale
Consider the mean-variance optimisation problem presented above. Introducing dual variables
¯ corresponding to the constraints of quadratic programming problem, respectively, the
µ, µ̄, µ̄
Lagrangian function can be stated as
n
n X
" n # " n # n
X X X X
¯) =
L(w, µ, µ̄, µ̄ wi wj σij + µ wi ri − r̄ + µ̄ wi − 1 − ¯ i wi
µ̄
i=1 j=1 i=1 i=1 i=1
The KKT conditions are presented below. The solution of linear equation system provides the
optimal investment strategy.
n
X
¯i
wj σij + µri + µ̄ − µ̄ = 0 i = 1, 2, · · · , n
j=1
n
X
wi r i = r̄
i=1
Xn
wi = 1
i=1
¯i
wi µ̄ = 0, i = 1, 2, · · · , n
¯
wi , µ̄i ≥ 0, i = 1, 2, · · · , n
1.34
1.32
1.3
1.28
Return
1.26
1.24
1.22
1.2
1.18
1.16
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Risk
volatility you are willing to bear in your portfolio is done by choosing any other point that falls
on the “Efficient Frontier”, since this will give you the maximum return for the amount of risk
you wish to take on. The set of points (that corresponds to portfolios) is called feasible set or
feasible region. Suppose that investor’s choice of portfolio is restricted to the feasible points on
a given horizontal line in mean-variance plane. All portfolios on this line have the same mean
rate of return, but different variances. Most investors prefer the portfolio corresponding to the
leftmost point on the line; that is the point with the smallest variance with the given mean. An
investor who agrees with this viewpoint is said to be risk averse. In this case, he or she seeks
to minimise risk as measured by variance. An investor who would select a point other than the
one of minimum standard deviation is said to be risk preferring.
Optimising your portfolio is not something you can calculate in your head. In fact, there
are computer programs dedicated to estimating 100’s (and sometimes 1000’s) of estimates for
expected returns in determining optimal portfolios for each given amount of risk.
38
5 Asset Pricing Models
Two main problem types dominate the investment science.
1. Determining the best course of action in an investment situation: problems of this type
include how to determine the best portfolio, how to devise the optimal strategy for man-
aging an investment, how to select from a group of potential investment projects, and so
forth. Mean-variance model is an example of this type presented in the previous section.
2. Determining the correct, arbitrage-free, fair, or equilibrium price of an asset: we also saw
some examples of this type such as the formula for the correct price of a bond in terms of
the term structure of interest rates.
We will carry on the pricing issue in this chapter, especially the correct price of risky asset
within the framework of the mean-variance setting. This is the Capital Asset Pricing Model
(CAPM) which was developed by Sharpe, Lintner and Mossin. An alternative asset pricing
model, we consider here is the Arbitrage Pricing Model (APT), is based on factor models. The
APT model was first developed by Stephen Ross, then have been created for applications in
most cash and derivative markets. Most of the criticism of the CAPM comes from two sources.
The first is the assumption that investors are mean-variance optimisers. The second comes from
single–period assumption.
39
asset’s total capital value to the total market capital value. These weights are also referred
as capitalisation weights.
In the idealised world, every investor is a mean-variance investor. In other words, every
investor is facing the same mean variance efficient frontier. This means that all investors
hold the same proportion of each asset, including the risk free asset and have the same
estimates, and everyone buys the same portfolio (which must be equal to the market
portfolio). Therefore, if we aggregate all investors’ portfolios we have a portfolio that is
on the frontier, but the aggregation of all portfolios is the entire market. Hence we have
shown that the market is mean-variance efficient. The implication is that investors do not
need to perform the optimisation. The efficient portfolios are the ones connecting the risk
free rate to the market portfolio.
Market Equilibrium The return on an asset depends on both the initial and final prices. The
investors solve the mean-variance portfolio problem using their common estimates, and
they place orders in the market to acquire their portfolios. If the orders placed do not
match what is available, the prices must change. The prices of assets under heavy demand
will increase; the prices of assets under light demand will decrease. Of course, the price
changes affect the estimates of asset returns directly, and hence investors will recalculate
their optimal portfolios. This process continues until demand exactly matches supply; that
is it carries on until there is an equilibrium.
The Capital Market Line The single efficient fund of risky assets is the market portfolio and
we can label this fund on the r̄ − σ diagram with M for market. The efficient set therefore
consists of the straight line, emanating from the risk free point and passing through the
market portfolio. This line is called the Capital Market Line (CML) and presented in
Figure 10. It is also called as a pricing line, since prices should adjust so that efficient
assets fall on this line. In mathematical terms, the capital market line states that
r¯M − rf
r̄ = rf + σ
σM
where r¯M and σM are the expected value and standard deviation of the market rate of
return and r̄ and σ are the expected value and standard deviation of the rate of return of an
arbitrary efficient asset. This equation describes all portfolios on capital market line which
r¯ −r
are combinations of riskless asset and portfolio M where r f is the intercept and MσM f is
the slope of the line. The slope is also called the price of the risk and describes how much
the expected rate of return of a portfolio must increase if its standard deviation increases
by one unit. As a consequence all portfolios that contain some proportion of portfolio M
and risk free asset will be perfectly correlated with it. An investor can attain positions
between rf and M by investing some of his/her money in portfolio M and the rest of it in
the risk free bond. For these positions he/she buys the risk free asset. On the other hand,
40
the positions on the line beyond point M can be obtained by selling the risk free bond and
using the proceeds of sale to buy portfolio M .
Proposition 2 If the market portfolio M is efficient, then the expected return E[r i ] of an asset
i satisfies
E[ri ] − rf = βi (E[rM ] − rf )
where
σiM
βi = 2
σM
Proof:
Consider the portfolio consists of a portion λ invested in asset i and a portion 1 − λ invested
on the market portfolio M . Here λ < 0 is allowed, which corresponds to borrowing at the
risk-free rate. In order to derive the CAPM we consider the portfolio with return
where ri is the return of asset i and rM is the return on the market. Taking expectation of two
sides of Equation (37) gives
r¯λ = λr¯i + (1 − λ)r¯M
and the standard deviation of the rate of return is
where E[r∗ ] = r¯∗ and σ∗2 are the expected return and the variance of r∗ . When λ = 0, it
corresponds to the market itself, so rλ is on CML; in other words, it is efficient. Observe that
for any λ the point (σλ , µλ ) must lie to the right of the CML; otherwise it would be a portfolio
with the same variance but higher expected return than an efficient portfolio, thus violating the
definition of efficiency. As λ varies, the values of r¯λ and σλ trace the curve.
The tangency condition can be translated into the condition that the slope of the curve is
equal to the slope of the capital market line at point M . In order to show this, we first need to
set up the following derivations;
dr¯λ
= r¯i − r¯M
dλ
dσλ λσi2 + (1 − 2λ)σiM + (λ − 1)σM
2
=
dλ σλ
dσλ
2
σiM −σM dr¯λ dr̄λ /dλ
Thus dλ λ=0 = σM . In addition, using the relation dσλ = dσλ /dλ we obtain the slope as
dr¯λ (r¯i − r¯M ) σM
=
dσλ λ=0 σiM − σM 2
which must be equal to the slope of the capital market line. Hence we find
41
The value r¯i − rf is called the expected excess rate of return of asset i. This is basically the
amount by which the rate of return is expected to exceed the risk-free rate. Likewise r¯M − rf is
the expected excess rates of the return of market portfolio. According to this, CAPM refers that
the expected excess rate of the return of an asset is proportional to the expected excess rates of
the return of market portfolio and proportionality factor is β.
The value βi is referred to the beta of an asset and measures the riskiness of each asset with
respect to the market portfolio M . High beta assets earn higher average return in equilibrium
because of the factor of βi (r¯M − rf ). The beta of the market portfolio is then
Cov(rM , rM )
βM = =1
V ar(rM )
This is used as a reference point which the risk of other assets can be measured. The average
risk of all assets is the beta of the market, which is one. Assets or portfolios that have a beta
greater than one have above average risk, tending to move more than market. For example, if
the riskless rate of interest (T-bill rate) is 5% per year and the market rises by 10%, assets with
a beta of 2 will tend to increase by 15%. If however, the market falls by 10%, assets with a beta
of 2 will tend to fall by 25% on average. Conversely, assets with betas less than one are of below
average risk and tend to move less than the market portfolio. Assets that have betas less than
zero tend to move in opposite direction to the market. These assets are known as hedge assets.
Beta of Portfolio
It is easy to calculate the overall beta of a portfolio β p in terms of the betas of the individual
assets in the portfolio. Suppose that the portfolio has n assets with the weights w 1 , · · · , wn . If
Xn Xn
the rate of return of the portfolio is r = wi ri and the cov(r, rM ) = wi cov(ri , rM ), then
i=1 i=1
n
X
βp = wi βi . The CAPM can be expressed in graphical form by regarding the formula as a
i=1
linear relationship which is called security market line. The CAPM in the covariance form and
2
in beta form is expressed in Figure 11. The market portfolio is at the point σ M in the first graph
and at β = 1 at the second graph. The security market line expresses the risk-reward structure
of assets according to CAPM. In other words, it describes the risk associated with each asset in
the portfolio as a function of its covariance with market or equivalently a function of its beta.
42
is affected by such general market moves, that asset entails systematic risk. The second term
σe2i is called non-systematic (idiosyncratic, or specific risk). This represents the component of
an asset’s volatility, which is uncorrelated with general market moves. Therefore, the specific
risk can be reduced by diversification. The systematic risk measured by beta is most important
since it directly combines with the other assets’ systematic risk.
Notice that this can also be seen as a discounting formula. In the deterministic case, it is
1
appropriate to discount the future payment at the interest rate r f , using a factor of 1+r f
. In
the random case, the appropriate interest rate is r f + β(r̄M − rf ), which can be regarded as a
risk-adjusted interest rate.
ri = ai + b i f + e i i = 1, · · · , n (39)
Notice that the rates of return and the single factor are linearly related. In Equation (39), a i
and bi are fixed constants. The quantity ei is random and represents the errors. We assume that
errors
Equation (39) also defines a linear fit to data. Imagine that several independent observations
are made of both the rate of return and factor. When the data is plotted the lines are likely to
be scattered. A straight line defined by the single factor equation is fitted through these points
in such a way that the average value of the error is zero. The error is measured by the vertical
distance from a point to the line. The fitting process is carried out for each asset separately to
obtain the parameters as follows. The mean return and the variance of assets i = 1, · · · , n are
calculated as
43
and
σij = bi bj σf2 i 6= j
cov(ri , f )
bi = (42)
σf2
Substituting (42) in (40), parameter ai can be obtained. The estimated parameters are valid
only for the current data set. Different values of b i and ai are obtained for the different sets
of historical observations of ri and f . Since ai is the intercept of the line for asset i with the
vertical axis, it is sometimes called intercepts. The parameter b i measures the sensitivity of the
return to the factor; therefore, it is named factor loading. If a portfolio is formed with weights
n
X
of wi for i = 1, · · · , n such that wi = 1, then
i=1
n
X n
X n
X
rp = wi a i + wi b i f + wi e i
i=1 i=1 i=1
= a + bf + e (43)
where
n
X n
X n
X
a= wi a i , b = wi b i , e = wi e i
i=1 i=1 i=1
1
If the variance of each asset is bounded by some constant, say σ, and if we put w i = n, then we
get
σ2
V ar[e] = → 0 as n → ∞
n
In a well diversified portfolio, the error term in the factor equation is small.
ri = ai + b1i f1 + b2i f2 + ei
The constant ai is called the intercept, and b1i , b2i are the factor loadings. The factors f1 and
f2 and the error ei are random variables. It is assumed that the expected value of error is zero
and the error is uncorrelated with two factors as well as with errors of other assets. However, it
is not assumed that two factors are uncorrelated with each other. These factors are observable
44
variables and their statistical properties can be studied independently of the asset returns. We
easily derive the following values for the expected rates of return and the covariances
Solution of this equation system gives the values of unknowns b 1i and b2i . When the factor
loadings are substituted in Equation (45) the intercepts a i can be found.
External factors It is very common that factors are chosen to be variables that are external
to the securities being explicitly considered in the model, such as gross national product
(GNP), consumer price index (CPI), unemployment rate.
Extracted factors It is possible to extract factors from the known information about security
returns. For example, the rate of return on the market portfolio is the most frequently
used factor. It is constructed directly from the returns of the individual securities. The
rate of return of one security can also be used as a factor for others. The method of
principal components is used to extract factors. The method uses the covariance matrix
of the returns to find the combinations of securities that have a large variances.
ri = a i + b i rM + e i
ri − r f = ai − (1 − bi )rf + bi (rM − rf ) + ei
= αi + bi (rM − rf ) + ei
E[ri ] = rf + αi + bi (E[rM ] − rf )
and the intercept is obtained as in the CAPM. Notice that b i = βi . An example of CAPM as a
factor model is presented in Figure 12. While y-axis represents the returns on the asset Lloyds,
the market returns are on the x-axis.
45
Figure 12: An Example of Capital Asset Pricing Model as a Factor Model
46
5.3.1 The Simple Version of APT
Consider a single–factor model. Assume that the factor model holds exactly (that is there is no
ei term);
ri = a i + b i f
The uncertainty comes from the factor f . The APT says that a i and bi are related if there is
no arbitrage. Choose another asset j, which is different from the asset i, such that b i 6= bj and
form a portfolio with w of asset i and (1 − w) of asset j.
rp = wai + (1 − w)aj + [wbi + (1 − w)bj ]f
bj
If we choose w = bj −bi so that wbi + (1 − w)bj = 0, then
ai b j aj b i
rp = + (49)
bj − b i bi − b j
and since rp has no exposure to f , it is risk-less. Therefore, rp must be equal to the risk-less rate
rf . The special portfolio is risk free because the equation for r i contains no random. If there
is a separate risk-free asset with a rate of return r f , it is clear that the portfolio constructed in
(49) must have the same rate; otherwise there would be an arbitrage opportunity. Even if there
is no explicit risk-free asset, all portfolios constructed this way, with no dependence on f must
have the same rate of return. We denote this rate by α 0 , (α0 = rf , if there is a risk free asset).
Setting the right-hand side of (49) equal to α0 , we find
ai b j aj b i
α0 = +
bj − b i bi − b j
α0 (bj − bi ) = a i bj − a j bi
aj − α 0 ai − α 0
= (50)
bj bi
The relationship must hold for all assets, therefore ai −α
bi
0
must be a constant c which is not
depending on i. Hence we obtain the following
ai − α 0
=c
bi
for all i and a constant c. This shows explicitly that the values of a i and bi are not independent;
that is indeed
ai = α 0 + b i c
We can use this information to write a formula for the expected rate of return of asset i.
E[ri ] = ai + bi E[f ]
= α0 + bi c + bi E[f ]
= α0 + bi E[f + c]
If E[f ] + c = α1 , then we can get
E[ri ] = α 0 + b i α1 (51)
Notice that once the constants α0 and α1 are known, the expected return of an asset i is
determined entirely by the factor loading bi since ai must follow bi .
The pricing formula given in (51) looks similar to the CAPM. If the factor f is chosen to be
the rate of return on the market rM , then we can set α0 = rf and α1 = E[rM ] − rf . Therefore,
the APT is identical to the CAPM when bi = βi . Therefore the CAPM is a consequence of the
APT if an exact single factor model holds. This result can also be extended for multi factor
models as follows:
Proposition 3 Suppose that there are n assets whose rates of return are governed by m < n
m
X
factors according to the equation ri = ai + bij fj for i = 1, · · · , n. Then there are constants
j=1
m
X
α0 , α1 , · · · , αm such that E[ri ] = α0 + bij αj for i = 1, · · · , n.
j=1
47
5.3.2 The APT Model for Well-Diversified Portfolios
Consider more realistic factor models which have error terms as well as factor terms. Suppose
that there are n assets and the rate of return on asset i can be represented by
m
X
ri = a i + bij fj + ei
j=1
where E[ei ] = 0 and E[e2i ] = σe2i . Also assume that ei is uncorrelated with the factors and
with the error terms of other assets. Let us form a portfolio using the weights w 1 , · · · , wn with
Xn
wi = 1. The rate of the return of the portfolio is
i=1
m
X
r =a+ b j fj + e i
j=1
where
n
X n
X n
X
a= wi a i , b j = wi bij , σe2 = wi2 σe2i
i=1 i=1 i=1
If n → ∞ and σe2i ≤ S 2 hold, then σe2 → 0. In other words, the error term associated with
well-diversified portfolio of an infinite number of assets has a variance of zero. If we apply the
simple APT, we obtain the following;
Proposition 4 Suppose that there are n assets whose rates of return are governed by m < n
m
X
factors according to the equation ri = ai + bij fj for i = 1, · · · , n. Then there are constants
j=1
α0 , α1 , · · · , αm such that for any well-diversified portfolio having an expected rate of return is
Xm
E[ri ] = α0 + bij αj for i = 1, · · · , n
j=1
This is again basically a relation describing that a i is not independent of bij ’s. The risk
free term must be related to the factor loadings. This is true even when there are error terms,
provided there is a large number of assets so that error terms can be effectively diversified away.
ri = ai + bi1 f1 + bi2 f2 + ei
If the market represents a well-diversified portfolio, the term Cov(r M , ei ) is ignored since there
is no error term in this portfolio. The beta of the asset i is
48
where
σM,f1 σM,f
βf 1 = 2 , β f2 = 2 2
σM σM
Hence the overall beta of asset can be obtained from underlying factor betas that do not depend
on the particular asset. The weights of these factor betas in the overall asset beta is equal to
the factor loadings. Notice that different assets have different betas since they have different
loadings.
2. Assume that a contract gives one the right, but not the obligation to purchase 100 shares
of GM stock for $60 per share in exactly 3 months. This is an option to buy GM. The
payoff of this option will be determined in 3 months by the price of GM stock at that time.
If GM is selling then for $70, the option will be worth $1000 because the owner of option
could at that time purchase 100 shares of GM for $60 per share according to the option
contract, and immediately sell those shares for $70 each.
49
the commodity at specific time in the future at a specific price agreed upon today. The contract
is between two parties, buyer and seller. The buyer is said to be long and seller is said to be
short. Being long or short, a given amount is the position of the party. Most forward contracts
specify that all claims are settled at the defined future date; both parties must carry out their
side of the aggrement at that time. The buyer (long position) of a forward contract is obligated
to take delivery of the asset and pay agreed-upon price at the maturity date. The seller (short
position) of a forward contract is obligated to deliver the asset and accept the agreed-upon price
at the maturity date.
Forward contracts are traded over-the-counter rather than on exchanges. The forward price
is the price that applies at delivery, which is negotiated so that the initial payment is zero.
The open market for immediate delivery of underlying asset is called the spot market. This is
distinguished from the forward market, which trades contracts for future delivery. During the
course of a forward contract, the spot market price may fluctuate as a function of the spot price
of the underlying asset.
where d(0, T ) is the discount factor between 0 and T . Buying the commodity at price S 0 is
exactly the same as lending an amount S0 of cash for which we will receive an amount F at time
T since storage is costless. We can assert the relation (52) using present value analysis (which is
based on an assumption of perfect markets, no transaction cost and arbitrage-free possibilities)
as the present value of the cash flow stream (−S0 , F ) must be zero. In addition, the relation (52)
can be proved using the arbitrage argument.
S0
Suppose that F > d(0,T ) . The portfolio can be constructed as follows; at the present time
borrow S0 amount of cash, buy one unit of the underlying asset on the spot market at
price S0 , and take one-unit short position in the forward market. The total cost of the
portfolio is zero. At time T , we deliver the asset, receiving a cash amount F , and repay
S0 S0
our loan in the amount d(0,T ) . As a result, we obtain positive profit of F − d(0,T ) for zero
net investment. This is an arbitrage opportunity.
50
S0
Suppose that F < d(0,T ) . We can construct the reverse portfolio. This requires that we short
one unit of the underlying asset. The shorting is executed by borrowing the asset from
someone who plans to store it during this period, then selling the borrowed asset at the spot
price, and replacing the borrowed asset at time T . The arbitrage portfolio is constructed
by shorting one unit, and taking a one-unit long position in the forward market. The net
S0
cash flow of the portfolio at time zero is zero. At time T , we receive d(0,T ) from our loan
and pay F one unit of asset and we return this unit to the lender who made the short
S0
possible. Our profit is d(0,T ) − F.
As a result, the relationship (52) must hold, since we obtain an arbitrage opportunity in either
case.
1. Portfolio A: buy a share for S0 and sell the share forward at T for the forward price F ,
Following two strategies, we form portfolios A and B whose payoff values are summarised in the
following table.
Payoff now Payoff at maturity
Strategy t=0 t=T
Buy stock −S0 ST + D T
Sell the stock forward 0 F − ST
Invest at risk-free rate −S0 S0 (1 + r)
Portfolio A −S0 F + DT
Portfolio B −S0 S0 (1 + r)
Since both portfolios have the same payoff values, we can find the forward price as
F = S0 (1 + r) − DT = (S0 − D0 )(1 + r)
Notice that when the stock pays no dividend, the forward price is F = S 0 (1 + r).
Example 6.1
Consider the simplest case of a stock trading at £145 today that pays no dividend during the
next 3 months and risk free rate is of 8%. What is the forward price?
The first portfolio is formed by buying one share for £145 and selling the share forward in
3 months for the forward price F while the second portfolio is constructed by the investment of
£145 in a bank account at the interest rate of 8%.
Today (t=0) 3 months from now (t=T)
Strategy (£) (£)
Buy share -145 +ST
Sell forward 0 F − ST
Portfolio A -145 F
Portfolio B -145 147.9193
The payoff of the portfolio A is certain and equal to F although we do not know the price of the
stock after 3 months. In the second strategy, we invest £145 today in a riskless bank account
and receive 145 × (1 + 0.08 3
12 ) = 147.9193 in three months time. By considering no arbitrage
principle, two portfolios must have the same payoff. Hence, the forward price is F = 147.9193.
51
Forward Arbitrage
The no-arbitrage principle is that prices must always adjust so that no market participant can
make a riskless profit.
Case 1: Suppose that the forward contract is overpriced as F = 149.00. How could you profit
from this? In this case, we can buy portfolio A and sell portfolio B. The output of the
investment strategies is summarised in the following table.
As a result we can profit £1.08 without taking any risk which is independent of the future
stock price.
Case 2: Suppose that the forward contract is underpriced as F = 143. The opposite strategy,
which is to sell the stock, buy it forward and invest the proceeds from the sale in the risk
free asset, leads to an arbitrage profit as £4.91.
When we price forwards using the principle of no-arbitrage, we assume that risk-less profit
opportunities without investment as in the above example cannot exist. Therefore, we conclude
that the only price that is consistent in the arbitrage free market is F = 147.9193.
where the dividend yield d is expressed as an annual percentage rate, similiar to interest rate.
52
Example 6.2
Consider a six-month forward contract on a stock that is currently trading at £95 and has a
dividend yield of 2%. The risk free rate is 7%. Show that the price of six-month forward contract
is £97.40.
If you buy e−dT = e0.02(−0.5) = 0.99 units of the stock, it means that you invest 0.99 × 95 =
£94.05. You also reinvest the all dividends, so in six months you own 1 unit of the stock. You
sell this unit forward so the return on your portfolio is riskless. You could also invest your £94.05
at the risk free rate, and obtain a payoff 94.05e0.5×0.07 = £97.40. Therefore, the price of forward
is found as
F = S0 e(r−d)T
= 95.00e(0.07−0.02)0.5 = 95.00 × 1.02532 = £97.40
If F < 97.40, then an arbitrage profit can be obtained by selling the stock and buying it back
forward, investing the proceeds in bonds. If F > 97.40, then we can make an arbitrage profit by
buying stock and selling it forward, where we would borrow the money to purchasing the stock.
In case of commodity forwards, the owner of commodities has to maintain their value, which
requires storage (wheat, gold), feeding (live hogs), or security (gold). The cost is called cost
of carry and expressed as an annual percentage rate q. It is treated as a negative dividend.
Therefore, the valuation formula for commodity forwards become
F = S0 e(r+q)T (54)
53
price to fall so that he or she can sell the asset for more than it is worth. Therefore, the put
option is exercised if the stock falls below the strike price.
Column 1 – Strike price (or exercise price) is the stated price per share for which under-
lying stock may be purchased (for a call) or sold (for a put) by the option holder upon
exercise of the option contract. When you exercise a call option, this is the value you
purchase the shares for. Option strike price goes up and down by typically increment 2.50
or 5.00. (In this example, it moves in 2 increments. )
Column 2 – Expiry Date shows the end of the life of an options contract. Options expire on
the third Friday of the expiry month.
Column 4 – Volume indicates the total number of options contracts traded for the day. The
total volume of all contracts is listed at the bottom of each table.
Column 5 – Bid is the price which someone is willing to pay for the option contract.
Column 6 – Ask is the price which someone is willing to sell an options contract for.
Column 7 – Open Interest is the number of options contracts that are open. These are
contracts which have not expired or have not been exercised. Total open interest is given
at the bottom of the table.
Example 6.3
This example describes how options work. Consider Microsoft (MS) stock with current price of
$67. The premium is $3.15 for a July 70 Call, which indicates that the expiration is 3rd of July
and the strike price is $70.
A stock option contract is the option to buy 100 shares so you must multiply the contract
by 100 to get the total price. Therefore, 1 call option contract with 2 months to expiration and
a strike price of $70 will cost: 3.15 × 100 (for the underlying shares) = $315. The strike price
of $70 means that the MS stock price must rise above $70 before the option is worth anything.
Furthermore, because the contract is $3.15 per share, the break-even price would be $73.15.
Keep in mind that the stock price is $67, less than the strike price of $70 so we actually just
54
paid $315 for an option that is theoretically worthless. But you might not lose the entire $315
because you are allowed to trade the options contract like a stock as long as it hasn’t expired.
Three weeks later, the stock price is $78. The options contract has increased along with the
stock price and is now worth 8.25 × 100 = $825. If you subtract what you paid for the contract,
then your profit is ($8.25 − $3.15) × 100 = $510. You more than doubled your money in just
three weeks. If you wanted, you could sell your options (called “closing your position”) at this
time and take your profits. If you think the stock price will continue to rise, you can let it ride.
On the expiration date, the MS stock price tanks, and is now $62. Since this is less than the
strike price and there is no time left, the option contract is worthless. We are now down the
original investment of $315. The process of our option investment is summarised in the following
table. Notice that on May 1st and at expiry date the stock price is less than the strike price
of $70 so we would lose our original investment of $315. The price swing for the length of this
contract from high to low was $825, which gives over double the original investment.
Options are considered very risky. On the other hand, options can be used in a way to reduce
risk. Suppose you have 100 shares of Microsoft and want to “lock-in” at the price the stock was
trading. You might want to lock-in to control your risk if you believed the markets might go
down substantially in the near future. In this case, buying 1 put option would protect you. If
the stock price goes down, you won’t lose (much) money because the loss from owning the stock
will be offset by the gains on your put option. In this way, options can be actually lower your
risk and act as an insurance. This is called “hedging”.
Types of Options
Call and Put options are known as Vanilla options since they are the simplest options. European
options are permitted to exercise only at expiry. American options allow the holder to exercise
at any time before expiry. Asian options are those whose payoff depend on the average price of
the underlying asset over a certain period of time. Note that the terms European, American or
Asian do not in any way refer to the continents on which the contracts are traded. Bermudan
options are half way between European and American options since the exercise is allowed on the
specific days or periods. Exotic options are a generic name given to derivative securities which
have more complex cash-flow structures than vanilla options (standard puts and calls). There
are literally hundreds of different exotic options and strategies using various positions in stocks,
options, and futures. Barrier, Digital, Lookback, Averagerate options (Asian options), Options
on baskets, Forwardstart, and Compound (options on options) are the most common exotic type
options. The principal motivation for trading exotic options is that they permit a much more
precise articulation of views on future market behavior than those offered by “vanilla” options.
Exotics can also be used as part of a risk management strategy or for speculative purposes.
55
you rights in the future? You may have no idea what the stock price will be between now and
expiry date. However, the contract has a value. At the very least you know that there is no
downside to owning the option. The contract gives you specific rights but no obligations.
The value of the contract depends on two things before expiry; i) how high the asset price
is today, ii) how long there is before expiry. The higher the underlying asset today, the higher
we might expect the asset to be at expiry of the option. Therefore, the more valuable we might
expect a call option to be. On the other hand, a put option might be cheaper by the same
reasoning. The dependence on time to expiry is more subtle. The longer the time to expiry, the
more time there is for the asset to rise or fall. Furthermore, the longer we have to wait until we
get any payoff, the less valuable the payoff is (simply because of the time value of money).
Option valuation is more complicated prior to an option’s expiration. Various option pricing
methodologies were proposed, but the problem wasn’t solved until the emergence of Black-Scholes
theory in 1973. Pricing formulae are to determine the theoretical fair value of an option. The
binomial opton pricing and Black-Scholes equation are the most common techniques. The Cox
Ross Rubinstein binomial model is widely accepted as the industry standard for equity options.
However, you should realise that pricing models calculate fair value based only on estimates
of volatility, interest rates and other factors. The price at which an option trades may not
necessarily be the same as its fair value.
−C = − max(ST − E, 0)
Example 6.4
Consider a call option on 200 ounces of gold struck with the current price of $400 and the strike
price of $375. The call option value at expiration will be: 200 max(0, 400 − 375) = 5000 This
reflects the fact that the option holder can exercise the option to purchase 200 ounces of gold
for $375 per ounce, and then immediately sell the 200 ounces at the market price of gold, which
is $400 per ounce.
Suppose instead that the price of gold were $360 when the option expired. In this case, it
would make no sense for the option holder to pay $375 for gold when the market price is only
$360. The option holder would not exercise the option, and it would expire worthless. Its market
value at expiration would be: 200 max(0, 360 − 375) = 0
56
and the current value of the underlier, whichever is greater. Therefore, the holder of a put option
prefer not to exercise the option when ST > E, since he/she can sell a share for more in the
market. If ST < E, the holder sells one share to the writer of the put option at the price of E.
Since this share is only worth ST , the put holder has made a profit of E − ST . The value of a
put option P is defined by payoff function as
P = max(E − ST , 0) (56)
When you sell a put option has the payoff −P = − max(E − S T , 0).
Example 6.5
Consider a put option on 200 ounces of gold struck at $400. If the market price of gold is $380
when the option expires, then the put’s expiration value is 200 max (0, 400 − 380) = 4000. This
reflects the fact that the option holder can purchase gold at the market price of $380 and then
exercise the put, selling the same gold for $400.
Payoff Diagram
We can plot the value of an option at expiry as a function of the underlying stock price in what
is known as a payoff diagrams. The payoff diagrams tell us what happens at expiry, how much
money the option contract is worth at that time. It makes no allowance for how much premium
you had to pay for the option. The payoff diagrams of buying and selling a call and a put options
are presented in Figure 14 and 15. C and P denote the call and put option values, respectively.
Payoff Payoff
P
C
E S E S
Stock Price Stock Price
PUT OPTION
CALL OPTION
Figure 14: Payoff diagrams of buying a call and a put option at expiry
Notice that a call option has value if ST > E and a put option has value if ST < E. In Figure
14, the first graph illustrates how the expiration value of a call option varies with the value of
the underlier at expiration. If, at expiration, the underlier value is below the strike price, the
option expires worthless. Otherwise, its expiration value is the value of the underlier less the
strike price.
Example 6.6
What are the payoffs of a call option and a put option at expiry if the exercise price is E = £50
and the stock prices are £20, 40, 60, 80.
Buy Call Write Call Buy Put Write Put
Stock Price max(ST − E, 0) − max(ST − E, 0) max(E − ST , 0) − max(E − ST , 0)
20 0 0 30 -30
40 0 0 10 -10
60 10 -10 0 0
80 30 -30 0 0
57
Payoff
Payoff
E S E S
Figure 15: Payoff diagrams of selling a call and a put option at expiry
Example 6.7
Suppose that the price of one IBM stock is $666 now. The cost of a 680 call option with expiry
in 3 months is $39. You expect the stock to rise significantly between now and expiry. How can
you profit if your prediction is right?
Suppose that you buy the stock for $666. Assume that, just before expiry, the stock has
risen to $730. I will have made a profit $64 per stock and my investment will have risen
by
730 − 666
× 100 = 9.6%
666
Suppose that you buy the call option for $39. At expiry, you can exercise the call, pay
$680 to receive something worth $730. You have paid $39 and get $50. This is the profit
of $11 per option. In percentage terms, it is
value of asset at expiry − strike − cost of call
Profit = × 100
cost of call
730 − 680 − 39
= × 100 = 28%
39
58
The payoff of the portfolio of options is formulated as
where ST is the value of the underlying asset at time T , C and P are the value of call and put
options values. The position of the portfolio constructed by a put and a call option at maturity
is summarised in the following table.
Type Option Value ST < E ST > E
Call Option C = max(ST − E, 0) 0 ST − E
Put Option −P = − max(E − ST , 0) −(E − ST ) 0
Portfolio C − P = max(ST − E, 0) − max(E − ST , 0) ST − E ST − E
The put-call parity can also be obtained at t for 0 ≤ t ≤ T as follows. If you buy the asset
today, then it will cost you St and be worth ST at expiry. The value of ST is uncertain, but
that amount can be guaranteed by buying the asset. Therefore, we can get S T if we decide to
buy that asset. On the other hand, the payment E at expiry T brings a cash flow of Ee −r(T −t)
at time t if we put in a bank with the interest rate of r. To lock in a payment of E at time T
involves a cash flow of Ee−r(T −t) at time t. In addition, a portfolio of a long call and a short put
gives the same payoff as a long asset, short cash position. This is summarised in the following
table.
Worth Today Worth at Expiry
Holding (t : 0 < t < T ) (t = T )
Buy call C max(ST − E, 0)
Sell put −P − max(E − ST , 0)
Buy stock −St −ST
Borrow cash Ee−r(T −t) E
Total C − P − St + Ee−r(T −t) 0
The equality of these cash flows is independent of the future behaviour of the stock and is model
independent. Therefore, the put-call parity at t becomes
C − P = St − Ee−r(T −t)
where the present value of E under continuous compounding is P V (E) = Ee −r(T −t) . In general,
the put-call parity at t can be formulated as C − P = S t − P V (E).
The call-put parity holds at any time up to expiry. Otherwise, there would be riskless
arbitrage opportunities. Here cash flows are set up to have a guaranteed value of zero at expiry.
Example 6.8
Suppose that European call and put options on stock A with the same exercise price of £40 and
six months to maturity are selling for £5 and £3, respectively. The current stock price is £40
and the risk-free interest rate is 8% p.a. Show whether put-call parity is satisfied under annual
compunding?
Here, C = 5, P = 3, E = 40, T = 0.5, St = 40, r = 8%. According to put-call parity the
relationship C = P + St − P V (E) must be satisfied if the arbitrage does not exist. Therefore,
1
C = P + St − E.
(1 + r)T
40
5 > 3 + 40 − = 4.51
(1 + 0.08)0.5
Hence, the put-call parity is not satisfied. The violation might be because of any one of three
possibilities:
1. the call is overpriced,
2. the put is underpriced, or
59
3. the stock is underpriced.
The arbitrage portfolio can be constructed at view of these possibilities as presented in the
following table.
Position Initial Value ST ≤ 40 ST > 40
Sell call 5 0 −(ST − 40)
Buy put -3 40 − ST 0
Buy stock -40 ST ST
Borrow 40(1 + 0.08)−0.5 38.49 -40 -40
Net portfolio Value 0.49 0 0
The arbitrage profit 0.49 is earned in forming such a portfolio.
Example 6.9
Consider a stock, a European put option, a European call option and T-bill. The stock is
currently selling for £100. Both put and call options have maturity of 3 months and the same
exercise price of £90. A call option has a price of £12 and a put £2. The annual interest rate is
5%. Is there an arbitrage opportunity available at these prices under continuous compounding?
From put-call parity the price of the call option should be C = P + S t − P V (E) where
P V (E) = E.e−rT .
C = P + St − P V (E)
12 < 2 + 100 − 90e−0.05×0.25 = 13.12
Therefore the call option is underpriced since the market price of the call is £12. Therefore, we
can realise an arbitrage profit of 1.12 by constructing a portfolio as follows: buy the call, sell
the put and stock, and invest P V (90) for 3 months. This process is summarised in the following
table.
Position Initial Value ST ≤ 90 ST > 90
Buy call -12 0 ST − 90
Sell put 2 ST − 90 0
Sell stock 100 −ST −ST
Buy T-bill −90e−(0.05)0.25 90 90
Net portfolio Value 1.12 0 0
where the function H(.) is called Heaviside function which takes the value one when its argument
is positive and zero otherwise.
A binary put option pays 1 pound at expiry if the asset price is smaller than the exercise
price. Therefore, a digital put has the payoff
1 if ST − E ≤ 0
PB = H(E − ST ) = (60)
0 if ST − E ≥ 0
The digital (or binary) option contracts have a payoff at expiry that is discontinuous in the
underlying asset price. An example of the payoff diagram for a binary call and a binary put is
shown in Figure 17.
If you think that the asset price will rise by expiry, to finish above the strike price, then you
might choose to buy either a vanilla call or a binary call. The vanilla call has the best upside
potential, growing linearly with S beyond the strike. However, the binary call can never pay
off more than 1 pound. If you expect the underlying to rise dramatically, then it may be the
60
Figure 17: Payoff diagrams of binary call and binary put options at expiry.
best to buy the vanilla call. If you believe that the asset rise will be less dramatic, then buy the
binary call. The gearing of the vanilla call is greater than that for a binary call if the move in the
underlying is large. On the other hand, the binary put would be bought by someone expecting
a modest fall in the asset price.
Like standard options, digitals can be classified as European or American style. The European
digital provides a payoff $1 if the asset ends above the strike price at the option’s maturity date
and zero otherwise. The American digital has a payoff of $1 if the underlying asset reaches the
value E before or at the expiration date T .
Sn+1 = Sn Hn+1 , 0 ≤ n ≤ N − 1
and p + q = 1.
If the stock price is known at the beginning of a period, the price at the beginning of the
next period is one of only two possible values. These two possibilities are usually defined to be
multiples of the price at the previous period – a multiple u (for up) and a multiple d (for down);
where u and d are both positive. Hence if the price at the beginning of a period is S, at the next
period, the stock price is either Su or Sd with probabilities of p and q = 1 − p, for 0 < p < 1.
At each step, the stock price either increases with a factor of u or decreases with a factor of d
based on the current stock price. At t = 1, if the current stock price is Su, then at t = 2 there
are two possibilities of up or down movements. Therefore, Su either increases to (Su)u = Su 2
61
with the probability of p or decreases to (Su)d = Sud with the probability of 1 − p. On the
other hand, the current price Sd goes up to either (Sd)u or goes down to (Sd)d = Sd 2 with
the corresponding probabilities. This carries out until the last time period. The general form of
such a lattice is presented in Figure 18.
Figure 18: An example of Binomial Lattice stock model for 4 time periods.
In order to specify the model completely, we must specify values of u, d and probabilities
p, q. They are chosen in such a way that the true stochastic nature of the stock is captured as
much as possible. Since the model is multiplicative in nature (the new value being uS or dS
with u > 0, d > 0), the price of the stock never becomes negative. It is therefore possible to
consider the logarithm of price as a fundamental variable. Let v be the expected yearly growth
rate. Consider v = E[ln( SST0 )] where S0 is the initial stock price and ST is the price at the end
of 1 year. In case of the deterministic process, the exponential growth rate is v = ln( SST0 ) and
the stock price at T is ST = S0 evT . The standard deviation is σ 2 = var[ln( SST0 )]. If a period
of length of ∆t is chosen, which is small compared to 1, the parameters of the binomial lattice
model are as follows;
1 1 v√
p = + ∆t (62)
2√ 2 σ
u = eσ ∆t (63)
√
d = e−σ ∆t
(64)
With this choice the binomial model will closely match the values of v and σ; that is the
expected growth rate of ln(S) in the binomial lattice model will be nearly v and the variance of
that rate will be nearly σ 2 . The closeness of the match improves if ∆t is made smaller, becoming
exact as it goes to zero. In addition, we will need to have a large number of time steps and hence
use a large number of nodes in order to implement the binomail lattice model on realistic time
scales In this section, we will first explain how to find the price of option based on the single
period binomial lattice, and then extend it to multi-period.
62
lattices in Figure 19. All three lattices have common arcs, in the sense that all move together
along the same arcs. If the stock price moves along the upward (downward) arc, then risk-free
asset and the call option both move along their upward (downward) arcs as well. Notice that
the risk free value is deterministic.
If the initial price S of the stock is known, then all values on the one-step three lattices are
known except the value of the call C. Let Cu and Cd be the values of a call option after a single
time period. They are obtained as
In order to duplicate these outcomes, purchase w s dollars worth of stock and wa dollars worth
of the risk free asset. This portfolio at the next time period will give payoff either uw s + Rwa
or dws + Rwa , depending on which path is taken. Since no-arbitrage is assumed, the option
outcomes match. Therefore, the following equations hold.
Solving this equation system, we find the weights of stock and asset in the portfolio as
Cu − C d Cu − uws uCd − dCu
ws = and wa = =
u−d R R(u − d)
Now we use the comparison principle (no-arbitrage principle) to assert that the value w s + wa
must be the value of the the call option C. The reason is that the portfolio we constructed
produces exactly the same outcomes as the call option. If the cost of the portfolio were less than
the price of the call option, we would never purchase the call. In fact we could make risk–free
arbitrage profit by buying this portfolio and and selling the call for an immediate risk free gain.
The opposite also were possible; in this case we would reverse the argument on our investment
strategy.
As a result we obtain the price of the call as
1 R−d u−R
C = Cu + Cd (69)
R u−d u−d
The portfolio made up of the stock and the risk–free asset that duplicates the outcome of the
option is referred to a replicating portfolio. It basically replicates the option. This option pricing
formula can also be derived by taking expected value of the option using the probability q, and
then discounting this value according to the risk-free rate. Therefore, the value of a one-period
call option on a stock governed by a binomial lattice is
1
C = [qCu + (1 − q)Cd ] (70)
R
63
where
R−d
q= (71)
u−d
is called risk-neutral probability and 0 < q < 1.
This procedure of the valuation works for all securities. In fact, q can be calculated by making
sure that the risk–neutral formula holds for the underlying stock itself; that is
1
S= [quS + (1 − q)dS]
R
Solving this equation again gives (71).
Equation (70) can also be written as
1
CT −1 = E[CT ]
R
where CT −1 and CT are the call values at T and T − 1, respectively, and E denote expectation
with respect to the risk neutral probabilities q and (q − 1).
Figure 20: Two-stage lattice representing two period call option and stock price.
stock is S and this price is modified by the up and down factors u and d while moving through
the lattice. The values of call option with strike E and the expiration time corresponding to
the final point in the lattice are shown in Figure 20. The value of the option is known at the
final nodes of the lattice. Starting from the final period and working backward toward the initial
time, the single-period risk free discounting is applied at every node of the lattice, At time period
two, the option value at each node of lattice is known as
64
Example 7.1
Consider a stock with a volatility of σ = 0.20. The current price of the stock is £62 and stock
pays no dividends. A call option on this stock has an expiration date 3 months from now and a
strike price is £60. The current interest rate is 10% and compounded monthly. Determine the
price of the call option by using the binomial option approach.
Let’s determine parameters for the binomial model of the stock price fluctuations. The time
1
period length is 1 month, which means ∆t = 12 . The other parameters are found as
√
u = eσ ∆t
√
= 1.05943
−σ ∆t
d = e = 0.94390
0.10
R = 1+ = 1.00833
12
The risk neutral probabilities are computed as
R−d
q= = 0.55770 and p = 1 − 0.55770 = 0.4423
u−d
The stock price evaluation on Binomial lattice is presented in Figure 21. In order to calculate
the call option price, we start from the final time period and enter the expiration values of the
call. This is the maximum of 0 and ST − E. For example, the entry for the top node is computed
as
1
[0.5577 × 13.72 + 0.4423 × 5.68] = 10.079
1.00833
We work backward until the initial value is reached. In this way, the price of the option is found
as £3.455. The option values on the binomial lattice is presented in Figure 22
8 Black-Scholes Equation
The price to buy or sell an asset at some future date depends on the expected future behaviour
of the price of the asset. This is often modelled by a stochastic process such as an Ito process,
based on Brownion motion. From this stochastic model, a partial differential equation, the
Black-Scholes equation can be derived for the dependence of value on asset price and time. In
this section, we first introduce Ito process and derive Black and Scholes equation — a linear
partial differential equation for the value of options.
Geometric Brownian Motion: Stock price defined as a function of time t, S(t) is a random
variable. Time evoluation of a stock price is often idealized as a diffusion process of
p
dS = µSdt + σS (dt)
where µ is the drift term (or the expected rate of return on the stock), σ is the volatility
of the stock price, and is a random variable following the normal distribution with
65
13.72
10.079
6.9525
5.68
3.1415
4.6075
0
1.7375
unit variance. Notice that the first term in the right hand side of the differential equation
describes the stock-price growth at a compounded rate of µ per unit time, when the second,
the stochastic term is zero. In this case,
S(t) = S0 exp(µt)
dS
which is the solution of differential equation dt = µS
Ito Formula: The Ito formula defines the stochastic differential equation for a stochastic process
that is defined as function of an Ito diffusion. Suppose that the random process X is defined
by the Ito process
dX(t) = a(X, t)dt + b(X, t)dW
where W is a standard Wiener process. If the process Y (t) is defined by Y (t) = F (X, t),
then Y (t) satisfies the Ito Equation has the following SDE
∂Y ∂Y 1 ∂2Y
dY = dt + dX + 2
dX 2
∂t
∂X 2 ∂X
∂Y ∂Y 1 2 ∂2Y ∂Y
= + a(t, X) + b (t, X) dt + b(t, X) dW
∂t ∂X 2 ∂X 2 ∂X
∂F ∂F 1 ∂2F ∂F
= + a(t, X) + b2 (t, X) 2
dt + b(t, X) dW
∂t ∂X 2 ∂X ∂X
The proof of this formula relies on the Taylor series extension and can be found in details
in Hull’s book.
Consider the function f (t, S) and the change in the function because of the value of S can
be described by f (S + ∆S). If we think of dt and dW as small quantities of ∆t and ∆W
then
66
The value of an option depends on t independently as well as through S. Let V = V (t, S)
be the value of the option for 0 < t ≤ T . When t changes to t + dt, then S changes to
S + dS and the value of option from V to V + dV . Therefore, we obtain
Here µ and σ are the drift rate (the expected return or the growth rate) and the volatility
of the asset and are defined as follows;
M M
1 X 1 X
µ= Ri and σ 2 = (Ri − R̄)2
M ∆t i=1 (M − 1)∆t i=1
where
Si+1 − Si √
Ri = = µ∆t + σ ∆t
Si
Example 8.1
Use Ito’s Lemma to show that it is possible to find a function f (S) which itself satisfies a
stochastic differential equation but with zero drift.
By Ito’s Lemma
2
∂f ∂f 1 2∂ f
df (S) = B(S) dWt + A(S) + B(S) dt
∂S ∂S 2 ∂S 2
For zero drift, we must have
∂f 1 ∂2f
A(S) + B(S)2 2 = 0
∂S 2 ∂S
This is a second order ordinary differential equation which can be solved for a particular speci-
fications of A and B.
67
Let the price S of an underlying security (which we shall refer to as a stock) be governed by a
geometric Brownian motion over a time interval [0, T ] described by the SDE:
Proposition 5 Suppose that the price of a security is governed by (81) and the interest rate is
r. A derivative of this security has a price f (S, t), which satisfies the partial differential equation
∂f ∂f 1 ∂2f
+ rS + σ 2 S 2 2 = rf. (82)
∂t ∂S 2 ∂S
Proof
In order to derive the Black-Scholes equation, we form a portfolio with portions of the stock
and the bond so that this portfolio exactly matches the return characteristics of the derivative
security. The value of this portfolio must equal the value of the derivative security, assuming
that there is no–arbitrage.
Assume the value of an option on S is a twice differentiable function f of S and time t. Then
by the Ito formula the value of the option f follows the SDE:
∂f ∂f 1 ∂2f ∂f
df = + µS + σ 2 S 2 2 dt + σSdW (83)
∂t ∂S 2 ∂S ∂S
This is an Ito processes for the price of the derivative security. This price fluctuates randomly
along with the stock price S and the Brownian motion W .
We form a portfolio of S and B that replicates the behavior of the derivative security. In
particular, at each time t we select an amount x t of the stock and an amount yt of the bond,
giving a total portfolio value of G(t) = xt S(t) + yt B(t). We wish to select xt and yt so that G(t)
replicates the derivative security value f (S, t). The gain in value of this portfolio due to changes
in security prices is
dG = xt dS + yt dB
If we expand the equation by substituting the dS value, we obtain
dG = xt dS + yt dB
= xt {µSdt + σSdW } + yt rBdt
= (xt µS + yt rB)dt + xt σSdW (84)
Since we want the portfolio gain of G(t) to behave just like the gain of f , we match the coefficients
of dt and dW in (84) to those of (83). In order to do this, we first match the dW coefficient by
∂f
setting xt = ∂S . Requiring G = xt S + yt B and G = f , gives
1 ∂f
yt = f (S, t) − S
B ∂S
Substituting these expressions in (84), and then matching the coefficient of dt in (83), the
following equality is obtained.
∂f 1 ∂f ∂f ∂f 1 ∂2f
µS + f (S, t) − S rB = + µS + σ 2 S 2 2
∂S B ∂S ∂t ∂S 2 ∂S
or
∂f ∂f 1 ∂2f 2 2
+ rS + σ S = rf
∂t ∂S 2 ∂S 2
68
which is the Black-Scholes equation.
This equation was first written down in 1969 but Fisher Black and Myron Scholes justified
the model and published in 1973. The Black-Scholes equation is a parabolic partial differential
equation (it is related to the heat or diffusion equation of mechanics). It is linear which means
that if there are two solutions of the equation then the sum of these is itself also a solution.
The advantage of this equation is that it is relatively easy to solve numerically. There are some
analytical and numerical techniques to solve this PDE. For that, the reader is referred to the
Wilmott’s Derivatives book, especially see the Chapter 6.
This PDE describes the evolution through time of any option that can be written as a
function of the current stock price S and the current time t. It is a “backwards” equation in
that it requires a final boundary condition at the option expiry time t = T rather than an initial
condition at the initial time t = 0. The final condition which is given by the option payoff, (and
possibly spatial boundary conditions) specify the specific type of option the PDE is to be solved
for.
Example 8.2
∂f ∂f ∂2f
= 0, = 1, =0
∂t ∂S ∂S 2
Hence (82) becomes rS = rS, which shows that it is a solution.
∂f ∂f ∂2f
= rert , = 0, =0
∂t ∂S ∂S 2
Hence (82) becomes rert = rert , which shows that f is a solution.
Π = f (S, t) − ∆S (85)
The first term on the right is the option and the second term is the short asset position. Notice
the minus sign in front of the second term. The quantity ∆ will for the moment be some constant
quantity of our choosing. We will assume that the underlying follows a log-normal random walk
dS = µSdt + σSdX. Now the natural question is “How does the value of the portfolio change
69
from time t to t + dt?” The change in the portfolio value is due partly to the change in the
option value and partly to the change in the underlying:
dΠ = df − ∆dS (86)
∂f 1 ∂2f ∂f
dΠ = dt + σ 2 S 2 2 dt + dS − ∆dS (88)
∂t 2 ∂S ∂S
No Arbitrage
No-arbitrage arguments state that a risk free portfolio must earn the risk free interest rate r.
If the portfolio had a return of q, higher than r, a trader would buy the portfolio, borrowing
the amount required to buy the portfolio at the risk free interest rate. They would thus make
a profit of q − r times the value of the portfolio, with no risk and with no investment: which is
known as an arbitrage. Similar arguments apply for when the portfolio return is lower than r.
After choosing the quantity ∆ as suggested above, we hold a portfolio whose value changes
by the amount
∂f 1 2 2 ∂2f
dΠ = + σ S dt (90)
∂t 2 ∂S 2
This change is completely riskless. If we have a completely risk-free change dΠ in the portfolio
value Π, then it must be the same as the growth we would get if we put the equivalent amount
of cash in a risk-free interest bearing account:
dΠ = rΠdt
∂f
= r(f − S)dt. (91)
∂S
This is an example of the no arbitrage principle. By equating Equations (89) and (91) for
the return on the portfolio and cancelling the dt factor, we obtain the Black-Scholes partial
differential equation:
∂f ∂f 1 ∂2f
+ rS + σ 2 S 2 2 = rf. (92)
∂t ∂S 2 ∂S
70