Investment Anal 1

Investment Analysis
Gareth D. Myles
May 2003
ii
Contents
Introduction
xiii
Investment Fundamentals
1 Securities and Analysis

1.1 Introduction . . . . . . . . . . . .
1.2 Financial Investment . . . . . . .
1.3 Investment Analysis . . . . . . .
1.4 Securities . . . . . . . . . . . . .
1.5 Non-Marketable Securities . . . .
1.6 Marketable Securities . . . . . .
1.6.1 Money Market Securities
1.6.2 Capital Market Securities
1.6.3 Derivatives . . . . . . . .
1.6.4 Indirect Investments . . .
1.7 Securities and Risk . . . . . . . .
1.8 The Investment Process . . . . .
1.9 Summary . . . . . . . . . . . . .
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
3
4
6
7
8
9
9
11
13
14
15
16
18
2 Buying and Selling

2.1 Introduction . . . . . . . . . . . . . . .
2.2 Markets . . . . . . . . . . . . . . . . .
2.2.1 Primary and Secondary . . . .
2.2.2 Call and Continuous . . . . . .
2.2.3 Auction and Over-the-Counter
2.2.4 Money and Capital . . . . . . .
2.3 Brokers . . . . . . . . . . . . . . . . .
2.4 Trading Stocks . . . . . . . . . . . . .
2.4.1 Time Limit . . . . . . . . . . .
2.4.2 Type of Order . . . . . . . . .
2.5 Accounts . . . . . . . . . . . . . . . .
2.5.1 Account Types . . . . . . . . .
2.5.2 Margin Requirement . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
21
21
21
22
23
23
24
24
25
25
26
26
27
27
iii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
iv
CONTENTS
2.6
2.7
II
2.5.3 Margin and Return . . . . . . . . . . . . . . . . . . . . . .

Short Sales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Portfolio Theory
33
3 Risk and Return

3.1 Introduction . . . . . . . . . . . .
3.2 Return . . . . . . . . . . . . . . .
3.2.1 Stock Returns . . . . . . .
3.2.2 Portfolio Return . . . . .
3.2.3 Portfolio Proportions . . .
3.2.4 Mean Return . . . . . . .
3.3 Variance and Covariance . . . . .
3.3.1 Sample Variance . . . . .
3.3.2 Sample Covariance . . . .
3.4 Population Return and Variance
3.4.1 Expectations . . . . . . .
3.4.2 Expected Return . . . . .
3.4.3 Population Variance . . .
3.4.4 Population Covariance . .
3.5 Portfolio Variance . . . . . . . .
3.5.1 Two Assets . . . . . . . .
3.5.2 Correlation Coefficient . .
3.5.3 General Formula . . . . .
3.5.4 Effect of Diversification .
3.6 Summary . . . . . . . . . . . . .
4 The
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
28
29
31
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
35
35
36
37
38
40
42
42
43
45
48
49
50
52
54
55
55
57
58
59
60
Efficient Frontier
Introduction . . . . . . . . . . . . . . . .
Two-Asset Portfolios . . . . . . . . . . .
Short Sales . . . . . . . . . . . . . . . .
Efficient Frontier . . . . . . . . . . . . .
Extension to Many Assets . . . . . . . .
Risk-free Asset . . . . . . . . . . . . . .
Different Borrowing and Lending Rates
Conclusions . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
65
65
66
75
75
79
82
86
90
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
93
. 93
. 94
. 98
. 102
. 104
. 105
5 Portfolio Selection
5.1 Introduction . . . . . . . . .
5.2 Expected Utility . . . . . .
5.3 Risk Aversion . . . . . . . .
5.4 Mean-Variance Preferences
5.5 Indifference . . . . . . . . .
5.6 Markovitz Model . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
CONTENTS
5.6.1 No Risk-Free . . . . . .
5.6.2 Risk-Free Asset . . . . .
5.6.3 Borrowing and Lending
Implications . . . . . . . . . . .
Conclusions . . . . . . . . . . .
5.7
5.8
III
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Modelling Returns
6 The
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
6.10
6.11
6.12
6.13
Single Index Model

Introduction . . . . . . . . . . .
Dimensionality . . . . . . . . .
Single Index . . . . . . . . . . .
Estimation . . . . . . . . . . .
Shortcomings . . . . . . . . . .
Asset Return and Variance . .
Portfolios Return and Variance
Diversified portfolio . . . . . .
Market Model . . . . . . . . . .
Beta and Risk . . . . . . . . . .
Adjusting Beta . . . . . . . . .
Fundamental Beta . . . . . . .
Conclusion . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
115
115
115
117
118
121
124
125
126
127
129
129
132
133
.
.
.
.
.
.
.
.
.
135
135
135
136
137
138
138
139
139
139
Equilibrium Theory
8 The
8.1
8.2
8.3
8.4
8.5
8.6
8.7
Capital Asset Pricing

Introduction . . . . . . .
Assumptions . . . . . .
Equilibrium . . . . . . .
Capital Market Line . .
Security Market Line . .
CAPM and Single-Index
Pricing and Discounting
106
107
108
109
110
113
7 Factor Models
7.1 Introduction . . . . . . . . . . . .
7.2 Single-Factor Model . . . . . . .
7.3 Two Factors . . . . . . . . . . . .
7.4 Uncorrelated factors . . . . . . .
7.5 Many Factors . . . . . . . . . . .
7.6 Constructing uncorrelated factors
7.7 Factor models . . . . . . . . . . .
7.7.1 Industry factors . . . . . .
7.7.2 Fundamental factors . . .
IV
.
.
.
.
.
143
Model
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
145
145
146
147
149
150
151
152
vi
CONTENTS
8.8
8.9
8.8 Market Portfolio . . . . . . . . . . . . . . . . . . . . . . . . . 154

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
9 Arbitrage Pricing Theory

9.1 Introduction . . . . . . .
9.2 Returns Process . . . . .
9.3 Arbitrage . . . . . . . .
9.4 Portfolio Plane . . . . .
9.5 General Case . . . . . .
9.6 Equilibrium . . . . . . .
9.7 Price of Risk . . . . . .
9.8 APT and CAPM . . . .
9.9 Conclusions . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
159
159
159
160
161
161
162
163
163
163
10 Empirical Testing
10.1 Introduction . .
10.2 CAPM . . . . .
10.3 APT . . . . . .
10.4 Conclusions . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
165
165
165
165
165
11 Efficient Markets and Behavioral Finance

11.1 Introduction . . . . . . . . . . . . . . . . .
11.2 Efficient Markets . . . . . . . . . . . . . .
11.3 Tests of Market Efficiency . . . . . . . . .
11.3.1 Event Studies . . . . . . . . . . . .
11.3.2 Looking for Patterns . . . . . . . .
11.3.3 Examine Performance . . . . . . .
11.4 Market Anomolies . . . . . . . . . . . . .
11.5 Excess Volatility . . . . . . . . . . . . . .
11.6 Behvioral Finance . . . . . . . . . . . . .
11.7 Conclusion . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
167
167
167
168
168
168
168
168
168
168
169
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Fixed Income Securities
12 Interest Rates and Yields

12.1 Introduction . . . . . . . . . . . . . .
12.2 Types of Bond . . . . . . . . . . . .
12.3 Ratings and default . . . . . . . . .
12.4 Yield-to-Maturity . . . . . . . . . . .
12.5 Semi-Annual and Monthly Coupons
12.6 CONTINUOUS INTEREST . . . . .
12.7 Interest Rates and Discounting . . .
12.7.1 Spot Rates . . . . . . . . . .
12.7.2 Discount Factors . . . . . . .
12.7.3 Forward Rates . . . . . . . .
171
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
173
173
174
175
175
179
180
180
180
182
183
vii
CONTENTS
12.8 Duration . . . . . . . . .
12.9 Price/Yield Relationship
12.10Bond Portfolios . . . . .
12.11Conclusions . . . . . . .
13 The
13.1
13.2
13.3
13.4
13.5
13.6
13.7
13.8
13.9
VI
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
185
186
188
188
Term Structure
Introduction . . . . . . . . . . . . . . . . .
Yield and Time . . . . . . . . . . . . . . .
Term Structure . . . . . . . . . . . . . . .
Unbiased Expectations Theory . . . . . .
Liquidity Preference Theory . . . . . . . .
Market Segmentation (Preferred Habitat)
Empirical Evidence . . . . . . . . . . . . .
Implications for Bond Management . . . .
Conclusion . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
189
189
189
189
190
191
192
192
192
192
Derivatives
193
14 Options
14.1 Introduction . . . . . . . . . . . . .
14.2 Options . . . . . . . . . . . . . . .
14.2.1 Call Option . . . . . . . . .
14.2.2 Put Options . . . . . . . . .
14.2.3 Trading Options . . . . . .
14.3 Valuation at Expiry . . . . . . . .
14.4 Put-Call Parity . . . . . . . . . . .
14.5 Valuing European Options . . . . .
14.5.1 The Basic Binomial Model
14.5.2 The Two-Period Binomial .
14.5.3 The General Binomial . . .
14.5.4 Matching to Data . . . . .
14.6 Black-Scholes Formula . . . . . . .
14.7 American Options . . . . . . . . .
14.7.1 Call Options . . . . . . . .
14.7.2 Put Option . . . . . . . . .
14.8 Summary . . . . . . . . . . . . . .
14.9 Exercises . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
195
195
196
196
198
199
199
205
206
207
212
215
217
218
220
220
222
224
225
15 Forwards and Futures

15.1 Introduction . . . . . . . . .
15.2 Forwards and Futures . . .
15.3 Futures . . . . . . . . . . .
15.3.1 Commodity Futures
15.3.2 Financial Futures . .
15.4 Motives for trading . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
227
227
227
229
229
230
232
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
viii
CONTENTS
15.4.1 Hedging . . . . . . . . . . . . . . . . .
15.4.2 Speculation . . . . . . . . . . . . . . .
15.5 Forward Prices . . . . . . . . . . . . . . . . .
15.5.1 Investment Asset with No Income . .
15.5.2 Investment Asset with Known Income
15.5.3 Continuous Dividend Yield . . . . . .
15.5.4 Storage costs . . . . . . . . . . . . . .
15.6 Value of Contract . . . . . . . . . . . . . . . .
15.7 Commodities . . . . . . . . . . . . . . . . . .
15.8 Futures Compared to Forwards . . . . . . . .
15.9 Backwardation and Contango . . . . . . . . .
15.10Using Futures . . . . . . . . . . . . . . . . . .
15.11Conclusions . . . . . . . . . . . . . . . . . . .
16 Swaps
16.1 Introduction . . . . . . . . . . . . . . .
16.2 Plain Vanilla Swaps . . . . . . . . . .
16.2.1 Interest Rate Swap . . . . . . .
16.2.2 Currency Swaps . . . . . . . .
16.3 Why Use Swaps? . . . . . . . . . . . .
16.3.1 Market Inefficiency . . . . . . .
16.3.2 Management of Financial Risk
16.3.3 Speculation . . . . . . . . . . .
16.4 The Swap Market . . . . . . . . . . . .
16.4.1 Features . . . . . . . . . . . . .
16.4.2 Dealers and Brokers . . . . . .
16.5 The Valuation of Swaps . . . . . . . .
16.5.1 Replication . . . . . . . . . . .
16.5.2 Implications . . . . . . . . . . .
16.6 Interest Rate Swap Pricing . . . . . .
16.7 Currency Swap . . . . . . . . . . . . .
16.7.1 Interest Rate Parity . . . . . .
16.7.2 Fixed-for-Fixed . . . . . . . . .
16.7.3 Pricing Summary . . . . . . . .
16.8 Conclusions . . . . . . . . . . . . . . .
VII
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
232
234
234
235
238
239
239
239
239
240
240
241
241
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
243
243
243
244
245
248
248
248
250
250
250
252
253
253
255
256
257
257
259
263
263
Application
17 Portfolio Evaluation
17.1 Introduction . . . . . .
17.2 Portfolio Consturction
17.3 Revision . . . . . . . .
17.4 Longer Run . . . . . .
17.5 Conclusion . . . . . .
265
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
267
267
267
267
267
267
ix
CONTENTS
VIII
Appendix
18 Using Yahoo!
18.1 Introduction
18.2 Symbols . .
18.3 Research . .
18.4 Stock Prices
18.5 Options . .
.
.
.
.
.
269
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
271
271
271
271
271
271
CONTENTS
Preface
This book has developed from the lectures for a final-year undergraduate course
and a first-level graduate course in finance that I have taught at the University of
Exeter for a number of years. They present the essential elements of investment
analysis as a practical tool with a firm theoretical foundation. This should make
them useful for those who wish to learn investment techniques for practical use
and those wishing to progress further into the theory of finance. The book
avoids making unnecessary mathematical demands upon the reader but it does
treat finance as an analytical tool. The material in the book should be accessible
to anyone with undergraduate courses in principles of economics, mathematics
and statistics.
xi
xii
PREFACE
Introduction
Finance, and the theory of finance, are important. Why? Because of the growth
of financial markets around the world, the volume of trade and the opportunities for profit. Finance theory is about the construction and management of
portfolios. This is helped by understanding theories of finance including the
pricing of derivatives.
The notes have an emphasis on calculation - of returns, variances etc. They
treat finance as an analytical subject but recognize the role and limitation of
theory.
xiii
xiv
INTRODUCTION
Part I
Investment Fundamentals
Chapter 1
Securities and Analysis

Learning investment analysis is a journey into a wealth of knowledge that is an exciting mix of the practical and the analytical. It
looks to technique to evaluate and to theory to explain. It is natural
to feel a degree of trepidation at the start of such a journey. To help
offset this we need to familiarize ourselves with the landscape and
landmarks, to develop an overview of our route. Some of these landmarks may be familiar others may be new or be seen from a different
perspective. Armed with this we can map out our route.
1.1
Introduction
This book is about the investment of wealth in financial securities. It provides

an introduction to the tools of investment analysis that can be used to guide
informed investment decisions. These tools range from the knowledge of the
securities that are available and how they are traded, through the techniques
for evaluating investments, to theories of market functioning.
Some investments can be very successful. An investor placing $10,000 in
August 1998 in the stock of Cephalon, a biopharmaceutical company traded
on Nasdaq, would have stock worth $107,096 in September 2003. Similarly, a
purchase of 10,000 in September 2001 of Lastminute.com stock, an internet
retailer traded on the London Stock Exchange, would be worth 134,143 in
August 2003. Cephalon and Lastminute.com are far from being alone in offering
these levels of gain. Many high technology companies match and can even
outstrip their performance. On the down side, losses in value can be even more
spectacular. Anyone investing $10,000 in September 2000 in Palm Inc., the
makers of handheld computers also traded on Nasdaq, would see that reduced
to $91 in April 2003. Such falls are not restricted to manufacturers. A holding
in July 2000 of 15 million in Exeter Equity Growth Fund would be worth
72,463 in August 2003 due to a fall in share price from 103.50 to 0.50.
What can be learnt from this book that would help choose investments like
3
CHAPTER 1. SECURITIES AND ANALYSIS
Cephalon and avoid Palm Inc.? The honest answer is that in September 2000
none of the evidence and none of the tools of investment analysis could have
forewarned that the stock of Palm Inc. would collapse in the way it did. Far
from being a condemnation of the methods, this observation shows precisely
why they are so valuable. How can this be so? Because it emphasizes that the
world of investments is plagued by uncertainty and unpredictability. No matter
how sophisticated are the tools we develop, or how rigorously we do our research
into potential investments, it is not possible for an investor to predict the future.
And that, in a nutshell, is why we need to learn investment analysis.
Investment analysis encompasses a methodology for accommodating the fundamental uncertainty of the financial world. It provides the tools that an investor can employ to evaluate the implications of their portfolio decisions and
gives guidance on the factors that should be taken into account when choosing
a portfolio. Investment analysis cannot eliminate the uncertainty, but it can
show how to reduce it. Moreover, although it cannot guarantee to guide you to
winners like Cephalon, it can help stop you being the investor that places all
their wealth in Palm Inc.
The starting point for investment analysis is the market data on the values
of securities which describes how they have performed in the past. In some parts
of the book, this market data is taken as given and we study how we should
invest on the basis of that data. This generates a set of tools which, even if an
investor does not apply them literally, provide a powerful framework in which
to think rationally about investment. This framework continually emphasizes
why many regretful investors have found to their cost that the maxim there is
no such thing as a free lunch is especially true in financial markets.
A serious investor will want to go beyond just accepting market data and
progress to an understanding of the forces that shape the data. This is the role of
financial theories that investigate explanations for what is observed. The deeper
understanding of the market encouraged by theory can benefit an investor by,
at the very least, preventing costly mistakes. The latter is especially true in the
world of derivative securities we meet later. But a theory remains just that until
it has been shown to unequivocally fit the data, and the wise investor should
never forget the limitations of theoretical explanations.
The book will provide information on how to choose which securities to invest
in, how they are traded, and the issues involved in constructing and evaluating
a portfolio. Throughout the text examples draw on the freely-available and
extensive data from Yahoo and show how the methods described can be applied
to this data.
1.2
Financial Investment
It is helpful to begin the analysis with a number of definitions that make precise
the subject matter that we will be studying. A standard definition is that
investment is the sacrifice of current consumption in order to obtain increased
consumption at a later date. From this perspective, an investment is undertaken
1.2. FINANCIAL INVESTMENT
with the expectation that it will lead, ultimately, to a preferred pattern of

consumption for the investor.
This definition makes consumption the major motivation for investment. In
contrast, many investors would argue that their motivation for investment is to
increase their wealth. This observation can be related back to the definition
by noting that wealth permits consumption or, in more formal language, an
increase in the stock of wealth permits an increase in the flow of consumption.
Wealth and consumption are, therefore, two sides of the same coin.
Looking more closely, two different forms of investment can be identified.
Real investment is the purchase of physical capital such as land and machinery to
employ in a production process and earn increased profit. In contrast, financial
investment is the purchase of paper securities such as stocks and bonds.
We do not explicitly discuss real investments in this book. Firms undertake
real investment to generate the maximum profit given the market conditions
that they face. There are many interesting issues raised by the real investment
activities of firms including issues of research and development, capacity expansion, and marketing. But consideration of these matters falls strictly outside
the scope of a text whose focus is upon financial investment. It should be noted,
though, that a real investment by an individual, such as the purchase of a house
or a painting, must be considered as part of the overall portfolio of assets held
by that investor.
There are, however, links between the two forms of investment. For example,
the purchase of a firms shares is a financial investment for those who buy them
but the motive for the issue of the shares is invariably that the firm wishes to
raise funds for real investment. Similarly, the commitment of a householder to a
mortgage, which is a financial investment, generates funds for a real investment
in property.
As a brief preview, the issues concerning financial investment that are addressed in the following chapters include:
The forms of security available: where and how they are bought and sold;
The investment process: the decision about which securities to purchase,
and how much of each;
Financial theory: the factors that determine the rewards from investment
and the risks.
The strategy employed to address these issues has the following structure.
The first step is to introduce the most important forms of securities that are
available to the investor and the ways in which they can be traded. The next
step is to analyze the general issues that are involved determining the preferred
choice of investment. This is undertaken abstracting from the particular features
of different securities. Next, we consider financial theories that try to explain
what is observed in the financial markets and which provide further insight into
the investment decision. Finally, we return to detailed analysis of some special
types of securities that raise especially interesting analytical questions.
1.3
Investment Analysis
The purpose of this book is to teach the principles of investment analysis. So,
what is investment analysis? One definition that moves us a little way forward
is that:
Investment analysis is the study of financial securities for the
purpose of successful investing.
This definition contains within it a number of important points. Firstly,
there are the institutional facts about financial securities: how to trade and
what assets there are to trade. Secondly, there are analytical issues involved
in studying these securities: the calculation of risks and returns, and the relationship between the two. Then there is the question of what success means
for an investor, and the investment strategies that ensure the choices made are
successful. Finally, there are the financial theories that are necessary to try to
understand how the markets work and how the prices of assets are determined.
It is clear that the more an investor understands, the less likely they are to
make an expensive mistake. Note carefully that this is not saying that the more
you know, the more you will earn. An explanation for this observation will be
found in some of the theories that follow. These comments partly address the
question Can you beat the market? Whether you can depends on the view you
may hold about the functioning of financial markets. One of the interpretations
of investment analysis is that this is just not possible on a repeated basis. An
alternative interpretation is that knowing the theory reveals where we should
look for ways of beating the market.
Example 1 The website for GinsGlobal Index Funds puts it this way Very
few professional fund managers can beat the market. Since there is no reliable way to identify the fund managers who will outperform the market, investors are best served by buying a broad spectrum of stocks at lower cost
(www.ginsglobal.co.za/company_profile.htm).
A knowledge of investment analysis can be valuable in two different ways. It
can be beneficial from a personal level. The modern economy is characterized
by ever increasing financial complexity and extension of the range of available
securities. Moreover, personal wealth is increasing, leading to more funds that
private individuals must invest. There is also a continuing trend towards greater
reliance on individual provision for retirement. The wealth required for retirement must be accumulated whilst working and be efficiently invested.
The study of investment analysis can also provide an entry into a rewarding
professional career. There are many different roles for which investment analysis is useful and the material covered in this book will be useful for many of
them. The training to become a financial analyst requires knowledge of much of
this analysis. Further, there are positions for brokers, bankers and investment
advisors for whom knowledge of investment analysis is a distinct advantage.
1.4. SECURITIES
Example 2 The Association for Investment Management and Research (AIMR)

is an international organization of over 50,000 investment practitioners and educators in more than 100 countries. It was founded in 1990 from the merger
of the Financial Analysts Federation and the Institute of Chartered Financial
R
) Program which
Analysts. It oversees the Chartered Financial Analyst (CFA
is a globally-recognized standard for measuring the competence and integrity of
financial analysts. CFA exams are administered annually in more than 70 countries. (For more information, see www.aimr.org)
1.4
Securities
A security can be defined as:

A legal contract representing the right to receive future benefits
under a stated set of conditions.
The piece of paper (e.g. the share certificate or the bond) defining the
property rights is the physical form of the security. The terms security or asset
can be used interchangeably. If a distinction is sought between them, it is that
the term assets can be applied to both financial and real investments whereas
a security is simply a financial asset. For much of the analysis it is asset that is
used as the generic term.
From an investors perspective, the two most crucial characteristics of a
security are the return it promises and the risk inherent in the return. An
informal description of return is that it is the gain made from an investment
and of risk that it is the variability in the return. More precise definitions of
these terms and the methods for calculating them are discussed in Chapter 3.
For the present purpose, the return can be defined as the percentage increase
in the value of the investment, so
Return =
final value of investment initial value of investment

100. (1.1)
initial value of investment
Example 3 At the start of 2003 an investor purchased securities worth $20000.

These securities were worth $25000 at the end of the year. The return on this
investment is
25000 - 20000
Return =
100 = 25%.
20000
The return on a security is the fundamental reason for wishing to hold it. The
return is determined by the payments made during the lifetime of the security
plus the increase in the securitys value. The importance of risk comes from the
fact that the return on most securities (if not all) is not known with certainty
when the security is purchased. This is because the future value of security is
unknown and its flow of payments may not be certain. The risk of a security is
a measure of the size of the variability or uncertainty of its return.
It is a fundamental assumption of investment analysis that investors wish

to have more return but do not like risk. Therefore to be encouraged to invest in assets with higher risks they must be compensated with greater return.
This fact, that increased return and increased risk go together, is one of the
fundamental features of assets.
A further important feature of a security is its liquidity. This is the ease with
which it can be traded and turned into cash. For some assets there are highly
developed markets with considerable volumes of trade. These assets will be
highly liquid. Other assets are more specialized and may require some effort to
be made to match buyers and sellers. All other things being equal, an investor
will always prefer greater liquidity in their assets.
The major forms of security are now described. Some of these are analyzed
in considerably more detail in later chapters because they raise interesting questions in investment analysis.
1.5
Non-Marketable Securities
The first form of security to introduce are those which are non-marketable,
meaning that they cannot be traded once purchased. Despite not being tradeable, they are important because they can compose significant parts of many
investors portfolios.
The important characteristics of these securities are that they are personal the investor needs to reveal personal details in order to obtain them so that the
parties on both sides know who is involved. They tend to be safe because they
are usually held at institutions that are insured and are also liquid although
sometimes at a cost.
The first such security is the savings account. This is the standard form of
deposit account which pays interest and can be held at a range of institutions
from commercial banks through to credit unions. The interest rate is typically
variable over time. In addition, higher interest will be paid as the size of deposit
increases and as the notice required for withdrawal increases. Withdrawals can
sometimes be made within the notice period but will be subject to penalties.
A second significant class are government savings bonds. These are the nontraded debt of governments. In the US these are purchased from the Treasury
indirectly through a bank or savings institution. The bonds receive interest
only when they are redeemed. Redemption is anytime from six months after
the issue date. National Savings in the UK deal directly with the public and
offers a variety of bonds with different returns, including bonds with returns
linked to a stock exchange index.
Two other securities are non-negotiable certificates of deposit (CDs). These
are certificates issued by a bank, savings and loan association, credit union,
or similar financial organization that confirm that a sum of money has been
received by the issuer with an implied agreement that the issuer will repay the
sum of money and that they are not a negotiable (or tradeable) instrument.
CDs can have a variety of maturities and penalties for withdrawal. They are
1.6. MARKETABLE SECURITIES
essentially a loan from an investor to a bank with interest paid as the reward. A
money market deposit account (MMDA) is an interest-earning savings account
offered by an insured financial institution with a minimum balance requirement.
The special feature of the account is that is has limited transaction privileges:
the investor is limited to six transfers or withdrawals per month with no more
than three transactions as checks written against the account. The interest rate
paid on a MMDA is usually higher than the rate for standard savings account.
1.6
Marketable Securities
Marketable securities are those that can be traded between investors. Some are
traded on highly developed and regulated markets while others can be traded
between individual investors with brokers acting as middle-men.
This class of securities will be described under four headings. They are
classified into money market securities which have short maturities and capital
market securities which have long maturities. The third group are derivatives
whose values are determined by the values of other assets. The final group are
classified as indirect investments and represent the purchase of assets via an
investment company.
1.6.1
Money Market Securities
Money market securities are short-term debt instruments sold by governments,

financial institutions and corporations. The important characteristic of these
securities is that they have maturities when issued of one year or less. The
minimum size of transactions is typically large, usually exceeding $100,000.
Money market securities tend to be highly liquid and safe assets. Because
of the minimum size of transactions, the market is dominated by financial institutions rather than private investors. One route for investors to access this
market is via money market mutual funds.
Treasury Bills
Short-term treasury bills are sold by most governments as a way of obtaining
revenues and for influencing the market. As later chapters will show, all interest
rates are related so increasing the supply of treasury bills will raise interest rates
(investors have to be given a better reward to be induced to increase demand)
while reducing the supply will lower them.
Treasury bills issued by the US federal government are considered to be the
least risky and the most marketable of all money markets instruments. They
represent a short-term loan to the US federal government. The US federal
government has no record of default on such loans and, since it can always print
money to repay the loans, is unlikely to default. Treasury bills with 3-month
and 6-month maturities are sold in weekly auctions. Those with a maturity
of 1 year are sold monthly. Treasury Bills have a face value of $1000 which is
10
the amount paid to the holder at the maturity date. They sell at a discount
(meaning a price different to, and usually less, than face value) and pay no
explicit interest payments. The benefit to the investor of holding the bill is the
difference between the price paid and the face value received at maturity.
An important component in some of the analysis in the later chapters is
the risk-free asset. This is defined as an asset which has a known return and
no risk. Because US Treasury Bills (and those of other governments with a
similar default-free record) are considered to have no risk of default and a known
return, they are the closest approximations that exist to the concept of a riskfree investment. For that reason, the return on Treasury Bills is taken as an
approximation of the risk-free rate of return.
Commercial Paper
Commercial paper is a short term promissory note issued by a corporation,
typically for financing accounts for which payment is due to be received and for
financing inventories. The value is usually at least $100,000 and the maturity
270 days or less. They are usually sold at a discount. These notes are rated by
ratings agencies who report on the likelihood of default.
Eurodollars
Eurodollars are dollar-denominated deposits held in non-US banks or in branches
of US banks located outside the US. Because they are located outside the US,
Eurodollars avoid regulation by the Federal Reserve Board. Eurodollars originated in Europe but the term also encompasses deposits in the Caribbean and
Asia. Both time deposits and CDs can fall under the heading of Eurodollars.
The maturities are mostly short term and the market is mainly between financial institutions. The freedom from regulation allows banks in the Eurodollar
market to operate on narrower margins than banks in the US. The market
has expanded as a way of avoiding the regulatory costs of dollar-denominated
financial intermediation.
Negotiable Certificates of Deposit

As for non-negotiable CDs, these are promissory notes on a bank issued in
exchange for a deposit held in a bank until maturity. They entitle the bearer
to receive interest. A CD bears a maturity date (mostly 14 days to 1 year), a
specified interest rate, and can be issued in any denomination. CDs are generally
issued by commercial banks. These CDs are tradeable with dealers making a
market (meaning they buy and sell to give the market liquidity). CDs under
$100,000 are called small CDs, CDs for more than $100,000 are called large
CDs or Jumbo CDs.
11
Bankers Acceptance
A bankers acceptance is a short-term credit investment created by a non-financial
firm but which is guaranteed by a bank. The acceptances can be traded at discounts from face value. Maturities range from 30 - 180 days and the minimum
denomination is $100,000. Bankers Acceptance are very similar to treasury
bills and are often used in money market funds.
Repurchase Agreements
A repurchase agreement involves a dealer selling government securities to an
investor with a commitment to buy them back at an agreed time. The maturity
is often very short with many repurchase agreement being overnight. They
constitute a form of short term borrowing for dealers in government securities.
The interest rate on the transaction is the difference between the selling and
repurchase prices. They permit the dealer to attain a short position (a negative
holding) in bonds.
1.6.2
Capital Market Securities
Capital market securities include instruments having maturities greater than

one year and those having no designated maturity at all. In the latter category
can be included common stock and, in the UK, consuls which pay a coupon in
perpetuity. The discussion of capital market securities divides them into fixed
income securities and equities.
Fixed income securities promise a payment schedule with specific dates for the
payment of interest and the repayment of principal. Any failure to conform to
the payment schedule puts the security into default with all remaining payments.
The holders of the securities can put the defaulter into bankruptcy.
Fixed income securities differ in their promised returns because of differences
involving the maturity of the bonds, the callability, the creditworthiness of the
issuer and the taxable status of the bond. Callability refers to the possibility
that the issuer of the security can call it in, that is pay off the principal prior to
maturity. If a security is callable, it will have a lower price since the issuer will
only call when it is in their advantage to do so (and hence against the interests
of the holder). Creditworthiness refers to the predicted ability of the issuer to
meet the payments. Income and capital gains are taxed differently in many
countries, and securities are designed to exploit these differences. Also, some
securities may be exempt from tax.
Bonds Bonds are fixed income securities. Payments will be made at specified
time intervals unless the issuer defaults. However, if an investor sells a bond
before maturity the price that will be received is uncertain.
12
The par or face value is usually $1000 in the US and 100 in the UK. Almost
all bonds have a term - the maturity date at which they will be redeemed.
Coupon bonds pay periodic interest. The standard situation is for payment
every 6 months. Zero coupon or discount bonds pay no coupon but receive the
par value at maturity. The return on a discount bond is determined by the
difference between the purchase price and the face value. When the return is
positive, the purchase price must be below the face value. Hence, these bonds
are said to sell at a discount.
Bonds sell on accrued interest basis so the purchaser pays the price plus the
interest accrued up until the date of purchase. If this was not done, sales would
either take place only directly after coupon payments or else prices would be
subject to downward jumps as payment dates were passed.
Treasury Notes and Bonds The US government issues fixed income
securities over a broad range of the maturity spectrum through the Treasury.
These are considered safe with no practical risk of default. Treasury notes have
a term of more than one year, but no more than 10 years. Treasury bonds have
maturities that generally lie in the range of 10 - 30 years.
Notes and bonds are sold at competitive auctions. They sell at face value
with bids based on returns. Both notes and bonds pay interest twice a year and
repay principal on the maturity date.
Similar notes and bonds are issued by most governments. In the UK, government bonds are also known as gilts since the original issues were gilt-edged.
They are sold both by tender and by auction.
Federal Agency Securities Some federal agencies are permitted to issue
debt in order to raise funds. The funds are then used to provide loans to assist
specified sectors of the economy. The are two types of such agencies: federal
agencies and federally-sponsored agencies.
Federal agencies are legally part of the federal government and the securities are guaranteed by the Treasury. One significant example is the National
Mortgage Association.
Federally-sponsored agencies are privately owned. They can draw upon the
Treasury up to an agreed amount but the securities are not guaranteed. Examples are the Farm Credit System and the Student Loan Marketing Association.
Municipal Bonds A variety of political entities such as states, counties,
cities, airport authorities and school districts raise funds to finance projects
through the issue of debt. The credit ratings of this debt vary from very good
to very poor. Two types of bonds are provided. General obligation bonds are
backed by the full faith and credit whereas revenue bonds are financed through
the revenue from a project.
A distinguishing feature of these bonds is that they are exempt from federal
taxes and usually exempt from the taxes of the state issuing the bond.
13
Corporate Bonds Corporate bonds are similar to treasury bonds in their

payment patterns so they usually pay interest at twice yearly intervals. The
major difference form government bonds is that corporate bonds are issued by
business entities and thus have a higher risk of default. This leads them to be
rated by rating agencies.
Corporate bonds are senior securities which means that they have priority
over stocks in the event of bankruptcy. Secured bonds are backed by claims on
specific collateral but unsecured are backed only by the financial soundness of
the corporation. Convertible bonds can be converted to shares when the holder
chooses.
Common Stock (Equity)
Common stock represents an ownership claim on the earnings and assets of
a corporation. After holders of debt claims are paid, the management of the
company can either pay out the remaining earnings to stockholder in the form
of dividends or reinvest part or all of the earnings. The holder of a common
stock has limited liability. That is, they are not responsible for any of the debts
of a failed firm.
There are two main types of stocks: common stock and preferred stock.
The majority of stock issued is common stock which represent a share of the
ownership of a company and a claim on a portion of profits. This claim is paid
in the form of dividends. Stockholders receive one vote per share owned in
elections to the company board. If a company goes into liquidation, common
stockholders do not receive any payment until the creditors, bondholders, and
preferred shareholders are paid.
Preferred Stock
Preferred stock also represents a degree of ownership but usually doesnt carry
the same voting rights. The distinction to common stock is that preferred stock
has a fixed dividend and, in the event of liquidation, preferred shareholders are
paid before the common shareholder. However, they are still secondary to debt
holders. Preferred stock can also be callable, so that a company has the option
of purchasing the shares from shareholders at anytime. In many ways, preferred
stock fall between common stock and bonds.
1.6.3
Derivatives
Derivatives are securities whose value derives from the value of an underlying
security or a basket of securities. They are also known as contingent claims,
since their values are contingent on the performance of the underlying assets.
Options
An option is a security that gives the holder the right to either buy (a call option)
or sell (a put option) a particular asset at a future date or during a particular
14
period of time for a specified price - if they wish to conduct the transactions.
If the option is not exercised within the time period then it expires.
Futures
A future is the obligation to buy or sell a particular security or bundle of securities at a particular time for a stated price. A future is simply a delayed
purchase or sale of a security. Futures were originally traded for commodities
but now cover a range of financial instruments.
Rights and Warrants
Contingent claims can also be issued by corporations. Corporate-issued contingent claims include rights and warrants, which allow the holder to purchase
common stocks from the corporation at a set price for a particular period of
time.
Rights are securities that give stockholders the entitlement to purchase new
shares issued by the corporation at a predetermined price, which is normally less
than the current market price, in proportion to the number of shares already
owned. Rights can be exercised only within a short time interval, after which
they expire.
A warrant gives the holder the right to purchase securities (usually equity)
from the issuer at a specific price within a certain time interval. The main
distinction between a warrant and a call options is that warrants are issued and
guaranteed by the corporation, whereas options are exchange instruments. In
addition, the lifetime of a warrant can be much longer than that of an option.
1.6.4
Indirect Investments
Indirect investing can be undertaken by purchasing the shares of an investment

company. An investment company sells shares in itself to raise funds to purchase a portfolio of securities. The motivation for doing this is that the pooling
of funds allows advantage to be taken of diversification and of savings in transaction costs. Many investment companies operate in line with a stated policy
objective, for example on the types of securities that will be purchased and the
nature of the fund management.
Unit Trusts
A unit trust is a registered trust in which investors purchase units. A portfolio
of assets is chosen, often fixed-income securities, and passively managed by a
professional manager. The size is determined by inflow of funds. Unit trusts are
designed to be held for long periods with the retention of capital value a major
objective.
1.7. SECURITIES AND RISK
15
Investment Trusts
The closed-end investment trust issue a certain fixed sum of stock to raise capital. After the initial offering no additional shares are sold. This fixed capital
is then managed by the trust. The initial investors purchase shares, which are
then traded on the stock market.
An open-end investment company (or mutual fund) continues to sell shares
after the initial public offering. As investors enter and leave the company, its
capitalization will continually change. Money-market funds hold money-market
instrument while stock and bond and income funds hold longer-maturity assets.
Hedge Funds
A hedge fund is an aggressively managed portfolio which takes positions on both
safe and speculative opportunities. Most hedge funds are limited to a maximum
of 100 investors with deposits usually in excess of $100,000. They trade in all
financial markets, including the derivatives market.
1.7
Securities and Risk
The risk inherent in holding a security has been described as a measure of the
size of the variability, or the uncertainty, of its return. Several factors can be
isolated as affecting the riskiness of a security and these are now related to the
securities introduced above. The comments made are generally true, but there
will always be exceptions to the relationships described.
Maturity The longer the period until the maturity of a security the more
risky it is. This is because underlying factors have more chance to change
over a longer horizon. The maturity value of the security may be eroded
by inflation or, if it is denominated in a foreign currency, by currency
fluctuations. There is also an increased chance of the issuer defaulting the
longer is the time horizon.
Creditworthiness The governments of the US, UK and other developed
countries are all judged as safe since they have no history of default in
the payment of their liabilities. Therefore they have the highest levels of
creditworthiness being judged as certain to meet their payments schedules.
Some other countries have not had such good credit histories. Both Russia
and several South American countries have defaulted in the recent past.
Corporations vary even more in their creditworthiness. Some are so lacking
in creditworthiness that an active junk bond market exists for high
return, high risk corporate bonds that are judged very likely to default.
Priority Bond holders have the first claim on the assets of a liquidated
firm. Only after bond holders and other creditors have been paid will stock
16

holders receive any residual. Bond holders are also able to put the corporation into bankruptcy if it defaults on payment. This priority reduces
the risk on bonds but raises it for common stock.
Liquidity Liquidity relates to how easy it is to sell an asset. The existence
of a highly developed and active secondary market raises liquidity. A
securitys risk is raised if it is lacking liquidity.
Underlying Activities The economic activities of the issuer of the security
can affect its riskiness. For example, stock in small firms and in firms
operating in high-technology sectors are on average more risky than those
of large firms in traditional sectors.
These factors can now be used to provide a general categorization of securities into different risk classes.
Treasury bills have little risk since they represent a short-term loan to the
government. The return is fixed and there is little chance of change in other
prices. There is also an active secondary market. Long-term government bonds
have a greater degree of risk than short-term bonds. Although with US and
UK government bonds there is no risk of default and the percentage payoff is
fixed, there still remains some risk. This risk is due to inflation which causes
uncertainty in the real value of the payments from the bond even though the
nominal payments are certain.
The bonds of some other countries bonds may have a risk of default. Indeed,
there are countries for which this can be quite significant. As well as an inflation
risk, holding bonds denominated in the currency of another country leads to an
exchange rate risk. The payments are fixed in the foreign currency but this does
not guarantee their value in the domestic currency. Corporate bonds suffer from
inflation risk as well as an enhanced default risk relative to government bonds.
Common stocks generally have a higher degree of risk than bonds. A stock
is a commitment to pay periodically a dividend, the level of which is chosen by
the firms board. Consequently, there is no guarantee of the level of dividends.
The risk in holding stock comes from the variability of the dividend and from
the variability of price.
Generally, the greater the risk of a security, the higher is expected return.
This occurs because return is the compensation that has to be paid to induce
investors to accept risks. Success in investing is about balancing risk and return
to achieve an optimal combination.
1.8
The Investment Process
The investment process is description of the steps that an investor should take
to construct and manage their portfolio. These proceed from the initial task
of identifying investment objectives through to the continuing revision of the
portfolio in order to best attain those objectives.
The steps in this process are:
1.8. THE INVESTMENT PROCESS
17
1. Determine Objectives. Investment policy has to be guided by a set of

objectives. Before investment can be undertaken, a clear idea of the purpose
of the investment must be obtained. The purpose will vary between investors.
Some may be concerned only with preserving their current wealth. Others may
see investment as a means of enhancing wealth. What primarily drives objectives
is the attitude towards taking on risk. Some investors may wish to eliminate
risk as much as is possible, while others may be focussed almost entirely on
return and be willing to accept significant risks.
2. Choose Value The second decision concerns the amount to be invested.
This decision can be considered a separate one or it can be subsumed in the
allocation decision between assets (what is not invested must either be held in
some other form which, by definition, is an investment in its own right or else
it must be consumed).
3. Conduct Security Analysis. Security analysis is the study of the returns
and risks of securities. This is undertaken to determine in which classes of assets
investments will be placed and to determine which particular securities should
be purchased within a class. Many investors find it simpler to remain with
the more basic assets such as stocks and fixed income securities rather than
venture into complex instruments such as derivatives. Once the class of assets
has been determined, the next step is to analyze the chosen set of securities to
identify relevant characteristics of the assets such as their expected returns and
risks. This information will be required for any informed attempt at portfolio
construction.
Another reason for analyzing securities is to attempt to find those that are
currently mispriced. For example, a security that is under-priced for the returns
it seems to offer is an attractive asset to purchase. Similarly, one that is overpriced should be sold. Whether there are any assets are underpriced depends
on the degree of efficiency of the market. More is said on this issue later.
Such analysis can be undertaken using two alternative approaches:
Technical analysis This is the examination of past prices for predictable
trends. Technical analysis employs a variety of methods in an attempt to
find patterns of price behavior that repeat through time. If there is such
repetition (and this is a disputed issue), then the most beneficial times to
buy or sell can be identified.
Fundamental analysis The basis of fundamental analysis is that the true
value of a security has to be based on the future returns it will yield.
The analysis allows for temporary movements away from this relationship
but requires it to hold in the long-rum. Fundamental analysts study the
details of company activities to makes predictions of future profitability
since this determines dividends and hence returns.
18
4. Portfolio Construction. Portfolio construction follows from security

analysis. It is the determination of the precise quantity to purchase of each
of the chosen securities. A factor that is important to consider is the extent of
diversification. Diversifying a portfolio across many assets may reduce risk but
it involves increased transactions costs and increases the effort required to manage the portfolio. The issues in portfolio construction are extensively discussed
in Chapters 4 and 5.
5. Evaluation. Portfolio evaluation involves the assessment of the performance of the chosen portfolio. To do this it is necessary to have some yardstick
for comparison since a meaningful comparison is only achieved by comparing
the return on the portfolio with that on other portfolios with similar risk characteristics. Portfolio evaluation is discussed in Chapter 17.
6. Revision. Portfolio revision involves the application of all the previous
steps. Objectives may change, as may the level of funds available for investment.
Further analysis of assets may alter the assessment of risks and returns and new
assets may become available. Portfolio revision is therefore the continuing reapplication of the steps in the investment process.
1.9
Summary
This chapter has introduced investment analysis and defined the concept of a
security. It has looked at the securities that are traded and where they are
traded. In addition, it has begun the development of the concepts of risk and
return that characterize securities. The fact that these are related - an investor
cannot have more of one without more of another - has been stressed. This
theme will recur throughout the book. The chapter has also emphasized the
role of uncertainty in investment analysis. This, too, is a continuing theme.
It is hoped that this discussion has provided a convincing argument for the
study of investment analysis. Very few subjects combine the practical value of
investment analysis with its intellectual and analytical content. It can provide
a gateway to a rewarding career and to personal financial success.
Exercise 1 Use the monthly data on historical prices in Yahoo to confirm the
information given on the four stocks in the Introduction. Can you find a stock
that has grown even faster than Cephalon?
Exercise 2 There are many stocks which have performed even worse than Exeter Equity Growth Fund. Why will many of these be absent from the Yahoo
data?
Exercise 3 If a method was developed to predict future stock prices perfectly,
what effect would it have upon the market?
1.9. SUMMARY
19
Exercise 4 At the start of January 1999 one investor makes a real investment
by purchasing a house for $300000 while a second investor purchases a portfolio
of securities for $300000. The first investor lives in the house for the next two
years. At the start of January 2001 the house is worth $350000 and the portfolio
of securities is worth $375000. Which investor has fared better?
Exercise 5 Is a theory which tells us that we cannot beat the market useless?
Exercise 6 You are working as a financial advisor. A couple close to retirement
seek your advice. Should you recommend a portfolio focused on high-technology
stock or one focused on corporate bonds? Would your answer be different if you
were advising a young newly-wed couple?
Exercise 7 Obtain a share certificate and describe the information written upon
it.
Exercise 8 By consulting the financial press, obtain data on the interest rates
on savings accounts. How are these rates related to liquidity?
Exercise 9 Taking data on dividends from Yahoo, assess whether the prices of
stocks are related to their past dividend payments. What does your answer say
about fundamental analysis?
Exercise 10 If all investors employed technical analysis, would technical analysis work?
Exercise 11 Are US treasury bills a safe asset for an investor who lives in
Argentina?
Exercise 12 Corporations usually try to keep dividend payments relatively constant even in periods when profits are fluctuating. Why should they wish to do
this?
20
Chapter 2
Buying and Selling

In chess, after learning the names of the pieces, the next step
is to understand the moves that the pieces may make. The ability of each piece to move in several ways provides the complexity of
the game that has generated centuries of fascination. By combining these moves, chess manuals describe the standard openings, the
philosophies of the middle game and the killer finishes. Similar rules
apply to trading securities. Much more is involved than simply buying and selling. Getting to know the rules of the game and the trades
that can be made will help the investor just as much as it helps the
chess player.
2.1
Introduction
A fundamental step in the investment process is the purchase and sale of securities. There is more to this than is apparent at first sight. An order to buy or sell
can take several forms, with characteristics that need to be determined by the
investor. A variety of brokers with different levels of service, and corresponding
fees, compete to act on the investors behalf. Some brokers are even prepared
to loan funds for the investor to purchase assets.
The chapter begins with a discussion of the markets on which securities are
traded. The role and characteristics of brokers are then described. Following
this, the focus turns to the purchase of common stock since it is here that there
is the greatest variety of purchasing methods. The choice of method can affect
the return on a portfolio just as significantly as can the choice of asset so the
implications for returns are considered.
2.2
Markets
Securities are traded on markets. A market is a place where buyers and sellers
of securities meet or any organized system for connecting buyers and sellers.
21
22
CHAPTER 2. BUYING AND SELLING
Markets are fundamental for the trading of securities.

Markets can have a physical location such as the New York Stock Exchange
or the London International Financial Futures Exchange. Both of these have
a trading floor where trade is conducted. It is not necessary for there to be
a physical location. The London Stock Exchange once possessed a physical
location, but now trade is conducted through a computer network that links
dealers. The Nasdaq Stock Market also has no location but relies on a network
to link dealers. Recent innovations such as internet-based markets also have no
physical location.
Example 4 The New York Stock Exchange was founded in 1792 and registered as a national securities exchange with the U.S. Securities and Exchange
Commission on October 1, 1934. It was incorporated as a not-for-profit corporation in 1971. The Exchange building at 18 Broad Street was opened in 1903
and a number of additional buildings are now also in use. At the end of 2002,
2,959 stocks were listed with a combined value of $9,603.3 billion. In July 2003,
31,924.5 million shares were traded with a combined value of $896.0 billion and
an average share price of $28.07. Only members of the Exchange can trade and
to become a member a seat must be purchased. The highest price paid for an
NYSE seat was $2,650,000 on August 23, 1999. (www.nyse.com)
Example 5 Nasdaq opened in 1971 as the first electronic market and is currently the largest. It lists just under 4000 companies primarily in the technology,
retail, communication, financial services and biotechnology sectors. Information
on market activity is transmitted to 1.3 million users in 83 countries. There is
an average of 19 market makers for each listed company with Dell Computer
Corporation having 95 market makers. Annual share volume in 2002 was 441
billion shares with a value of $7.3 trillion. (www.nasdaq.com)
Markets can be classified in a number of different ways. Each classification
draws out some important aspects of the role and functioning of markets.
2.2.1
Primary and Secondary
Primary markets are security markets where new issues of securities are traded.
When a company first offers shares to the market it is called an initial public
offering. If additional shares are introduced later, they are also traded on the
primary market. The price of shares is normally determined through trade but
with new shares there is no existing price to observe. The price for initial public
offerings has either to be set as part of the offer, or determined through selling
the shares by tender or auction.
Secondary markets are markets where existing securities are resold. The
London and New York stock exchanges are both primarily secondary markets.
The role of the primary market in helping to attain economic efficiency is
clear: the primary market channels funds to those needing finance to undertake
real investment. In contrast, the role of the secondary market, and the reason
2.2. MARKETS
23
why so much attention is paid to it, is probably less clear. Two important roles
for the secondary market that can be identified:
Liquidity One of the aspects that will be important for the purchaser of a
new security is their ability to sell it at a later date. If it cannot be sold,
then the purchaser is making a commitment for the lifetime of the asset.
Clearly, given two otherwise identical assets an investor would prefer to
own the one which can most easily be traded. Thus new securities would
have a lower value if they could not be subsequently traded. The existence
of a secondary market allows such trading and increases the liquidity and
value of an asset.
Value Trading in assets reveals information and provides a valuation of
those assets. The assignment of values guides investment decisions by
showing the most valuable uses for resources and helps in the attainment
of economic efficiency. Without the secondary market this information
would not be transmitted.
2.2.2
Call and Continuous
A second way to classify markets is by the nature of trading and the time periods
at which trading can take place.
In a call market trading takes place at specified times. Those who wish to
trade are called together at a specific time and trade for a limited period. A
single price is set that ensures the market clears. This can cause significant
movements in price from one trading time to the next, so call markets can have
provisions to limit movement from the initial price.
Example 6 The main Austrian exchange, Wiener Brse, operates a call system to auction shares. The auction price is set to ensure that the largest volume of orders can be executed leaving as few as possible unfilled. An auction
schedule is published to announce the times when specific securities are called.
(www.wienerboerse.at)
In a continuous market there is trading at all times the market is open.
Requests to buy and sell are made continuously. Trade is often facilitated by
market makers who set prices and hold inventories.
Example 7 The London Stock Exchange operates as a continuous market and
is the largest equity market in Europe. On the London Stock Exchange trading
is performed via computer and telephone using dealing rooms that are physically
separated from the exchange. Almost 300 firms worldwide trade as members of
the Exchange. (www.londonstockexchange.com)
2.2.3
Auction and Over-the-Counter
In an auction market buyers and sellers enter a bidding process to determine

the trading price of securities. This typically takes place at a specified location.
The New York Stock Exchange is the primary example of an auction market.
24
An over-the-counter market involves direct negotiation between broker and

dealers over a computer network or by telephone. The market will have a
network of dealers who make a market and are willing to buy and sell at specified
prices. They earn profit through the spread: the difference between the price
at which they will buy and the price at which they will sell (the latter being
higher). Nasdaq is considered to be an over-the-counter market.
2.2.4
Money and Capital
The money market is the market for assets with a life of less than 1 year. This
includes money itself and near-money assets such as short term bonds.
The capital market is the market for assets with a life greater than 1 year
such as equity and long-term bonds.
2.3
Brokers
On most markets, such as the New York and London Stock Exchanges, an
individual investor cannot trade on the market directly. Instead they must
employ the services of a broker who will conduct the trade on their behalf. This
section discusses brokers and the services offered by brokerages.
A broker is a representative appointed by an individual investor to make
transactions on their behalf. The reward for a broker is generated through
commission charged on the transactions conducted. This can lead to incentive problems since it encourages the broker to recommend excessive portfolio
revision or churning. The accounts of individual investors at a brokerage are
dealt with by an account executive. Institutional investors deal through special
sections of retail brokerage firms
Brokerage firms can be classified according to the services offered and the
resulting level of fee charged. Traditional brokerages, now called full-service
brokers, offer a range of services including information, investment advice and
investment publications. They conduct the trading business of the clients and
aim to guide them with their investment decisions. In addition to earning income
from commissions, full-service brokers also generate revenue from a range of
other activities. Amongst these are trading on their own account, commission
from the selling of investment instruments such as mutual funds and payment
for participation in initial public offerings.
Example 8 In 2002, the assets of the retail customers of Morgan Stanley
amounted to $517 billion and they employed 12,500 financial advisors. Their
retail brokerage business now focuses on fee-based accounts rather than commission and has changed the incentive structure for financial advisors so that
the interests of the investor and the financial advisor coincide. The financial
advisors also take a more consultative approach with investors and emphasize
financial planning, asset allocation and diversification. Managed investment
products such as mutual funds, managed accounts and variable annuities have
become a major focus. (www.morganstanley.com)
2.4. TRADING STOCKS
25
Discount brokers offer fewer services and charge lower fees than full-service
brokers. Effectively, they do not provide advice or guidance or produce publications. Their major concentration is upon the execution of trading orders. Many
discount brokers operate primarily internet-based services.
Example 9 Quick & Reilly charge a minimum commission rate of $19.95 for
orders placed online for stocks priced over $2.00. A higher rate applies to stock
priced under $2.00 and for trades executed over the telephone or through financial consultants. A full schedule of fees can be found at www.quickandreilly.com.
2.4
Trading Stocks
To trade stocks through a broker it is necessary to provide a range of information. Some of this information is obvious, others parts require explanation. The
details of the transaction that need to be given to the broker are:
The name of the firm whose stock is to be traded;
Whether it is a buy or a sell order;
The size of the order;
The time limit until the order is cancelled;
The type of order.
Of these five items, the first three are self-explanatory. The final two are
now explored in more detail.
2.4.1
Time Limit
The time limit is the time within which the broker should attempt to fill the
order. Most orders can be filled immediately but for some stocks, such as those
for small firms, there may not be a very active market. Also, at times when the
market is falling very quickly it may not be possible to sell. In the latter case
a time limit is especially important since the price achieved when the order is
filled may be very different to when the order was placed.
A day order is the standard order that a broker will assume unless it is
specified otherwise. When a day order is placed the broker will attempt to
fill it during the day that it is entered. If it is not filled on that day, which
is very unlikely for an order concerning a sale or purchase of stock in a large
corporation, the order is cancelled.
An open-ended time horizon can be achieved by placing an open order, also
known as a good-till-cancelled order. Such an order remains in effect until it
is either filled or cancelled. In contrast, a fill-or-kill order is either executed
immediately or, if this cannot be done, cancelled. Finally, a discriminatory
order leaves it to the brokers discretion to decide when to execute or cancel.
26
2.4.2
Type of Order
The alternative types of order are designed to reduce the uncertainty associated
with variations in price.
A market order is the simplest transaction. It is a request for the broker
to either buy or sell, with the broker making their best effort to complete the
transaction and obtain a beneficial price. With a market order the price at
which the trade takes place is uncertain but, unless it is for a very illiquid asset,
it is usually certain that the broker will complete the transaction.
In a limit order a limit price is specified. For a stock purchase, the limit price
is the maximum price at which the investor is willing to buy. For a stock sale,
the limit price is the minimum they are willing to accept. Execution of a limit
order is uncertain since the limit price may be unobtainable. If the transactions
does proceed then the upper limit on price (if buying) or the lower limit on price
(if selling) is certain.
With a stop order, a stop price has to be specified. This stop price acts
a trigger for the broker to initiate the trade. For a sale, the stop price is set
below the market price and the broker is instructed to sell if the price falls
below the stop price. A stop-loss strategy of this form is used to lock-in profits.
Alternatively, for a buy order, the broker is instructed to buy if the price rises
above the stop price (which is set above the current market price). This strategy
could be employed by an investor waiting for the best moment to purchase a
stock. When its price shows upward movement they then purchase.
The execution of a stop order is certain if the stop price is passed. However the price obtained is uncertain, especially so if there are rapid upward or
downward movements in prices.
A stop-limit order combines the limit order and the stop order. A minimum
price is placed below the stop price for a sell and a maximum price is placed
above the stop-price for a buy. This has the effect of restricting price to be
certain within a range but execution is uncertain since no transaction may be
possible within the specified range.
2.5
Accounts
Before common stock can be through a broker it is first necessary to open an

account with a brokerage. This can be done by either physically visiting the
brokerage, by telephone or directly by the internet. It is necessary that some
personal details are given to the broker.
Example 10 The online account application form at Quick and Reilly requires
answers to five categories of question. These are: (i) personal details including
citizenship and social security number; (ii) financial details including income,
source of funds and investment objectives; (iii) details of current broker; (iv)
employment status; and (v) links with company directors and stock exchange
members.
27
2.5. ACCOUNTS
2.5.1
Account Types
When opening an account at a brokerage, an investor has a choice between the

two types of account. A cash account requires that the investor provides the
entire funds for any stock purchase. In contrast, a margin account with a broker
allows the investor to borrow from the broker to finance the purchase of assets.
This allows a portfolio to be partly financed by using borrowed funds. The
implications of this will be analyzed after first considering some further details
of margin accounts.
To open a margin account a hypothecation agreement is required. Under
such an agreement the investor has to agree that the brokerage can:
Pledge securities purchased using the margin account as collateral;
Lend the purchased securities to others.
To make this possible, the shares are held in street name by the brokerage.
This means that they are owned legally by brokerage but dividends, voting
rights and annual reports of the companies whose stock are purchased go to the
investor. In consequence, the investor receives all the privileges of owning the
stock even if they do not legally own it.
The reason that the shares can be pledged as collateral is because the brokerage requires some security for the loan it has advanced the investor. There
is always a possibility that an investor may default on the loan, so the brokerage retains the stock as security. Allowing the shares to be lent to other
investors may seem a strange requirement. However, this is necessary to permit
the process of short-selling to function. This is discussed in Section 2.6.
A margin purchase involves the investor borrowing money from the broker
to invest. The broker charges the investor interest on the money borrowed plus
an additional service charge.
2.5.2
Margin Requirement
A margin purchase involves an element of risk for the broker. The shares they
hold in street name form the collateral for the loan. If the value of the shares
falls, then the collateral is reduced and the broker faces the risk that the borrower may default. To protect themselves against this, the broker insists that
only a fraction of the investment be funded by borrowing. This is fraction
termed the initial margin requirement.
The initial margin requirement, expressed as a percentage, is calculated by
the formula
Initial Margin Requirement =
value financed by investor

100.
total value of investment
(2.1)
This can be expressed alternatively by saying that the initial margin requirement
is the minimum percentage of the investment that has to be financed by the
28
investor. In the US, the Board of Governors of the Federal Reserve system has
authorized that the initial margin must be at least 50%. Exchanges can impose
a higher requirement than this, and this can be raised even further by brokers.
Example 11 If the initial margin requirement is 60%, an investor must provide
at least $6,000 of a $10,000 investment and the brokerage no more than $4,000.
In the period following a margin purchase the value of the investment made
will change. If the value falls far enough, then the collateral the brokerage is
holding may no longer be sufficient to cover the loan. To guard against this, the
brokerage calculates the value of the securities each day. This is called marking
to market. From this is calculated the actual margin which is defined by
Actual Margin =
market value of assets - loan

100.
market values of assets
(2.2)
The actual margin can rise or fall as the asset prices change.
Example 12 Assume that a margin purchase of $10,000 has been made using
$7,000 of the investors own funds and $3,000 borrowed from the broker. If the
value of the investment rises to $12,000 the actual margin is 12,0003,000
100 =
12,000
75%. If instead the value of the investment falls to $6,000 the actual margin is
6,0003,000
100 = 50%.
6,000
A brokerage will require that the actual margin should not fall too far. If
it did, there would be a risk that the investor may default and not pay off the
loan. The maintenance margin requirement is the minimum value of the actual
margin that is acceptable to the brokerage. The New York Stock Exchange
imposes a maintenance margin of 25% and most brokers require 30% or more.
If the actual margin falls below the maintenance margin, then a margin call
is issued. A margin call requires that the investor must add further funds to
the margin account or deposit additional assets. Either of these will raise the
market value of assets in the account. Alternatively, part of the loan could be
repaid. In any case, the action must be significant enough to raise the actual
margin back above the maintenance margin.
Example 13 Assume that a margin purchase of $12,000 has been made with a
loan of $4,000. With a maintenance margin of 30%, the investor will receive a
margin call when
market value of assets - 4,000
100 < 30.
market values of assets
This is satisfied when the market value of assets is less than $5714.
2.5.3
Margin and Return
Buying on the margin has a both a benefit and a cost. Recall that the formula
(1.1) defined the return as the increase in value of the investor as a percentage
29
2.6. SHORT SALES
of the initial value. What changes when this formula is applied to a margin
purchase is that the initial value of the investment is measured by the funds
coming from the investors own resources. With a margin purchase the quantity
of the investors funds is reduced for any given size of investment by the value
of the funds borrowed from the brokerage. As the following example shows, this
reduction magnifies the return obtained from the investment.
Example 14 Consider an investment of $5,000 made using a cash account. If
the value of the investment rises to $6,500 the cash return is
Cash Return =
6, 500 5, 000
100 = 30%.
5, 000
(2.3)
Now consider the same investment using a margin account. Assume the initial
margin is 60% so the investor provides $3000 and borrows $2000. With an
interest rate of 10% charged on the loan the return is
Margin Return =
6, 500 3, 000 0.1 2, 000

100 = 110%.
3, 000
(2.4)
Example 14 reveals the general property of a margin purchase which is that

it raises the return above that of a cash purchase if the return is positive. This
is because the return is calculated relative to the contribution of the investor
which, due to the loan component, is less that for a cash purchase.
Margin purchases do have a downside though. As the following example
shows, a margin purchase also magnifies negative returns.
Example 15 Assume the value falls to $4,000. The the return from a cash
purchase is
4, 000 5, 000
100 = 20%,
(2.5)
Cash Return =
5, 000
and the return on the margin purchase is
Margin Return =
4, 000 3, 000 1.1 2, 000

= 40%.
3, 000
(2.6)
The conclusion from this analysis is that purchasing on the margin magnifies
gains and losses. Because of this, it increases the risk of a portfolio. Informally,
this suggests that a margin purchase should only really be considered when
there is a strong belief that a positive return will be earned. Obviously, this
conclusion can only be formally addressed using the techniques of portfolio
analysis developed later.
2.6
Short Sales
A short sale is the sale of a security that an investor does not own. This can be
achieved by borrowing shares from another investor. It is part of the role of a
broker to organize such transactions and to ensure that the investor from whom
the shares are borrowed does not suffer from any loss.
To provide the shares for a short sale, the broker either:
30
Broker
Provides
margin
Lends
stock
Pays
price
Receives
stock
Short
Initial
Buyer
Seller A
Owner B
Told C
owns stock
Firm
Figure 2.1: A Short Sale
Broker
Cash for
dividends
Dividends, reports,
voting rights
Reports
Short
Initial
Buyer
Seller A
Owner B
Firm
Dividends, reports,
voting rights
Figure 2.2: After the Short Sale

Uses shares held in street name;
or
Borrows the shares from another broker.

Figures 2.1 and 2.2 illustrate a short sale. Investor A is the short-seller.
The shares are borrowed from B and legally transferred to the buyer C. This
is shown in Figure 2.1. To ensure that B does not lose from this short sale, A
must pay any dividends that are due to B and the broker provides an annual
report and voting rights. The report can come from the firm and the voting
rights can be borrowed from elsewhere - either from other shares owned by the
broker or from other brokers. Figure 2.2 illustrates this.
To close the transaction, the investor A must eventually purchase the shares
and return them to B. A profit can only be made from the transaction if the
shares can be purchased for less than they were sold. Short-selling is only used
if prices are expected to fall.
31
2.7. SUMMARY
There is a risk involved for the broker in organizing a short sale. If the
investor defaults, the broker will have to replace the shares that have been
borrowed. The short-seller must make an initial margin advance to the broker
to cover them against this risk. This initial margin is calculated as a percentage
of the value of the assets short-sold. The broker holds this in the investors
account until the short-sale is completed and the investor finally restores the
shares to the initial owner.
Example 16 Let 100 shares be short-sold at $20 per share. The total value of
the transaction is $2000. If the initial margin requirement is 50%, the investor
must deposit a margin of $1000 with the brokerage.
To guard the brokerage against any losses through changes in the price of
the stock, a maintenance margin is enforced. Thus a margin call is made if the
actual margin falls below the maintenance margin. The actual margin is defined
by
short sale proceeds + initial margin - value of stock
100,
value of stock
(2.7)
where the value of stock is the market value of the stock that has been short-sold.
Actual Margin =
Example 17 If the value of the shares in Example 16 rises to $2,500 the actual
margin is
Actual Margin =
2, 000 + 1, 000 2, 500

100 = 20%,
2, 500
(2.8)
If instead they fall to $1,500 the actual margin becomes

Actual Margin =
2, 000 + 1, 000 1, 500

100 = 100%.
1, 500
(2.9)
With a short sale actual margin rises as the value of the stock sold-short falls.
2.7
Summary
Trading is a necessary act in portfolio construction and management. Securities

can be traded in a number of ways through brokers offering a range of service
levels. These trading methods have been described, especially the process of
short-selling which has important implications in the following chapters. The
process of buying using a margin account has been shown to raise return, but
also to increase potential losses. With the practical background of these introductory chapters it is now possible to begin the formalities of investment
analysis.
Exercise 13 A margin account is used to buy 200 shares on margin at $35 per
share. $2000 is borrowed from the broker to complete the purchase. Determine
the actual margin:
32

a. When the purchase is made;
b. If the price of the stock rises to $45 per share;
c. If the price of the stock falls to $30 per share.
Exercise 14 An investor buys 2000 shares at $30 each. The initial margin
requirement is 50% and the maintenance margin is 30%. Show that if the stock
price falls to $25, the investor will not receive a margin call. At what price will
a margin call be received?
Exercise 15 600 shares are purchased on the margin at the beginning of the
year for $40 per share. The initial margin requirement was 55%. Interest of
10% was paid on the margin loan and no margin call was ever faced. A dividend
of $2 per share is received. Calculate the annual return if:
a. The stock are sold for $45 per share at the end of the year;
b. If the stock are sold for $25 per share at the end of the year.
c. Calculate the return for (a) and (b) if the purchase had been made using
cash instead of on the margin.
Exercise 16 Using a margin account, 300 shares are short sold for $30 per
share. The initial margin requirement is 45%.
a. If the price of the stock rises to $45 per share, what is the actual margin
in the account?
b. If the price of the stock falls to $15 per share, what is the actual margin
in the account?
Exercise 17 Is it true that the potential loss on a short sale is infinite? What
is the maximum return?
Part II
Portfolio Theory
33
Chapter 3
Risk and Return

The first steps in investment analysis are to calculate the gains
from an investment strategy and the risk involved in that strategy.
Investment analysts choose to measure gains by using the concept of
a return. This chapter will show how returns can be calculated in a
variety of circumstances, both for individual assets and for portfolios.
Looking back over the past performance of an investment the calculation of risk is just an exercise in computation. Given the data, the
formulas will provide the answer. Where the process is interesting is
when we look forward to what the return may be in the future. The
challenge of investment analysis is that future returns can never be
predicted exactly. The investor may have beliefs about what the return will be, but the market never fails to deliver surprises. Looking
at future returns it is necessary to accommodate their unpredictability by determining the range of possible values for the return and the
likelihood of each. This provides a value for the expected return from
the investment. What remains is to determine just how uncertain
the return is. The measure that is used to do this, the variance of
return, is the analysts measure of risk. Together the expected return
and variance of alternative portfolios provide the information needed
to compare investment strategies.
3.1
Introduction
At the heart of investment analysis is the observation that the market rewards
those willing to bear risk. An investor purchasing an asset faces two potential
sources of risk. The future price at which the asset can be sold may be unknown,
as may the payments received from ownership of the asset. For a stock, both of
these features are immediately apparent. The trading price of stocks changes
almost continually on the exchanges. The payment from stocks comes in the
form of a dividend. Although companies attempt to maintain some degree of
35
36
CHAPTER 3. RISK AND RETURN
constancy in dividends, they are only a discretionary payment rather than a

commitment and their levels are subject to change.
These arguments may not seem to apply to bonds whose maturity value and
payments seem certain. But bond prices do fluctuate so, although the maturity
value is known, the value at any time before maturity is not. Furthermore, the
maturity value is given in nominal terms whereas the real value is uncertain as
inflation must be taken into account. The same argument also applies to the
real value of the coupon payments. Finally, there is the risk of default or early
redemption. Only the shortest term bonds issued by major governments can
ever be regarded as having approximately certain payoffs.
In order to guide investment choice, an investor must be able to quantify
both the reward for holding an asset and the risk inherent in that reward. They
must also be aware of how the rewards and risks of individual assets interact
when the assets are combined into a portfolio. This chapter shows how this is
done.
3.2
Return
The measure of reward that is used in investment analysis is called the return.
Although we focus on financial assets, the return can be calculated for any
investment provided we know its initial value and its final value.
The return is defined as the increase in value over a given time period as
a proportion of the initial value. The time over which the return is computed
is often called the holding period. Returns can be written in the raw form just
defined or, equally well, converted to percentages. All that matters in the choice
between the two is that consistency is used throughout a set of calculations. If
you start using percentages, they must used everywhere. The calculations here
will typically give both.
The formula for calculating the return can now be introduced. Letting V0
be the initial value of the investment and V1 the final value at the end of the
holding period, the return, r, is defined by
r=
V1 V0
.
V0
(3.1)
To express the return as a percentage the formula is modified to

r=
V1 V0
100.
V0
(3.2)
Example 18 An initial investment is made of $10,000. One year later, the

value of the investment has risen to $12,500. The return on the investment is
= 0.25. Expressed as a percentage, r = 1250010000
100 = 25%.
r = 1250010000
10000
10000
It should be emphasized that the return is always measured relative to the
holding period. The example used a year as the holding period, which is the conventional period over which most returns are expressed. For instance, interest
37
3.2. RETURN
rates on bonds and deposit accounts are usually quoted as an annual rate. The
precise description of the return in the example is consequently that the return
on the investment was 25% per year. Other time periods may be encountered
such as a month, a week, or even a day. Detailed analysis of stock prices often
employs daily returns.
Example 19 An investment initially costs $5,000. Three months later, the
investment is sold for $6,000. The return on the investment is r = 60005000

5000
100 = 20% per three months.
3.2.1
Stock Returns
The process for the calculation of a return can also be applied to stocks. When
doing this it is necessary to take care with the payment of dividends since these
must be included as part of the return. We first show how to calculate the
return for a stock that does not pay a dividend and then extend the calculation
to include dividends.
Consider a stock that pays no dividends for the holding period over which
the return is to be calculated Assume that this period is one year. In the formula
for the return, we take the initial value, V0 , to be the purchase price of the stock
and the final value, V1 , to be its trading price one year later. If the initial price
of the stock is p (0) and the final price p (1) then the return on the stock is
r=
p (1) p (0)
.
p (0)
(3.3)
Example 20 The price of Lastminute.com stock trading in London on May 29

2002 was 0.77. The price at close of trading on May 28 2003 was 1.39. No
dividends were paid. The return for the year of this stock is given by
r=
1.39 0.77
= 0.805 (80.5%).
0.77
The method for calculating the return can now be extended to include the
payment of dividends. To understand the calculation it needs to be recalled
that the return is capturing the rate of increase of an investors wealth. Since
dividend payments are an addition to wealth, they need to be included in the
calculation of the return. In fact, the total increase in wealth from holding the
stock is the sum of its price increase plus the dividend received. So, in the
formula for the return, the dividend is added to the final stock price.
Letting d denote the dividend paid by a stock over the holding period, this
gives the formula for the return
r=
p (1) + d p (0)
.
p (0)
(3.4)
Stocks in the US pay dividends four time per year and stock in the UK pay
dividends twice per year. What there are multiple dividend payments during
the holding period the value of d is the sum of these dividend payments.
38
Example 21 The price of IBM stock trading in New York on May 29 2002 was
$80.96. The price on May 28 2003 was $87.57. A total of $0.61 was paid in
dividends over the year in four payments of $0.15, $0.15, $0.15 and $0.16. The
return over the year on IBM stock was
r=
3.2.2
87.57 + 0.61 80.96

= 0.089 (8.9%).
80.96
Portfolio Return
It was noted in the introduction that the definition of a return could be applied
to any form of investment. So far it has only been applied to individual assets.
We now show how the method of calculation can be applied to a portfolios
of assets. The purchase of a portfolio is an example of an investment and
consequently a return can be calculated.
The calculation of the return on a portfolio can be accomplished in two ways.
Firstly, the initial and final values of the portfolio can be determined, dividends
added to the final value, and the return computed. Alternatively, the prices and
payments of the individual assets, and the holding of those assets, can be used
directly.
Focussing first on the total value of the portfolio, if the initial value is V0 ,
the final value V1 , and dividends received are d, then the return is given by
r=
V1 + d V0
.
V0
(3.5)
Example 22 A portfolio of 200 General Motors stock and 100 IBM stock is
purchased for $20,696 on May 29 2002. The value of the portfolio on May 28
2003 was $15,697. A total of $461 in dividends was received. The return over
= 0.219 (-21.9%).
the year on the portfolio is r = 15697+46120696
20696
The return on a portfolio can also be calculated by using the prices of the
assets in the portfolio and the quantity of each asset that is held. Assume that
an investor has constructed a portfolio composed of N different assets. The
quantity held of asset i is ai . If the initial price of asset i is pi (0) and the final
price pi (1) , then the initial value of the portfolio is
V0 =
N
ai pi (0) ,
(3.6)
N
ai pi (1) .
(3.7)
i=1
and the final value

V1 =
i=1
If there are no dividends, then these can be used to calculate the return as
N
N
V1 V0
i=1 ai pi (1)
i=1 ai pi (0)
r=
.
(3.8)
=
N
V0
i=1 ai pi (0)
39
3.2. RETURN
Example 23 Consider the portfolio of three stocks described in the table.
Stock
A
B
C
Holding
100
200
150
Initial Price
2
3
1
Final Price
3
2
2
Example 24 The return on the portfolio is

r
(100 3 + 200 2 + 150 2) (100 2 + 200 3 + 150 1)

100 2 + 200 3 + 150 1
= 0.052 (5.2%).
This calculation can be easily extended to include dividends. If the dividend

payment per share from stock i is denoted by di , the formula for the calculation
of the return from a portfolio becomes
N
N
ai [pi (1) + di ] i=1 ai pi (0)
r = i=1
(3.9)
N
i=1 ai pi (0)
Example 25 Consider the portfolio of three stocks described in the table.

Stock
A
B
C
Holding
50
100
300
Initial Price
10
3
22
Final Price
15
6
20
Dividend per Share

1
0
3
Example 26 The return on the portfolio is

r
(50 [15 + 1] + 100 [6] + 300 [20 + 3]) (50 [10] + 100 [3] + 300 [22])
50 [10] + 100 [3] + 300 [22]
= 0.122 (12.2%).
The calculation of the return can also be extended to incorporate shortselling of stock. Remember that short-selling refers to the act of selling an asset
you do not own by borrowing the asset from another investor. In the notation
used here, short-selling means you are indebted to the investor from whom the
stock has been borrowed so that you effectively hold a negative quantity of the
stock. For example, if you have gone short 200 shares of Ford stock, then the
holding for Ford is given by 200. The return on a short sale can only be positive
if the price of Ford stock falls. In addition, during the period of the short sale
the short-seller is responsible for paying the dividend on the stock that they
have borrowed. The dividends therefore count against the return since they are
a payment made.
Example 27 On June 3 2002 a portfolio is constructed of 200 Dell stocks and
a short sale of 100 Ford stocks. The prices on these stocks on June 2 2003, and
the dividends paid are given in the table.
40

Stock
Dell
Ford
Initial Price ($)

26.18
17.31
Dividend ($)
0
0.40
Final Price ($)

30.83
11.07
Example 28 The return over the year on this portfolio is

r
3.2.3
(200 30.83 + [100 11.47]) (200 26.18 + [100 17.31])

200 26.18 + [100 17.31]
= 0.43 (43%).
Portfolio Proportions
The calculations of portfolio return so far have used the quantity held of each
asset to determine the initial and final portfolio values. What proves more
convenient in later calculations is to use the proportion of the portfolio invested
in each asset rather then the total holding. The two give the same answer
but using proportions helps emphasize that the returns (and the risks discussed
later) depend on the mix of assets held, not on the size of the total portfolio.
The first step is to determine the proportion of the portfolio in each asset.
If the value of the investment in asset i at the start of the holding period is V0i ,
then the proportion invested in asset i is defined by
Xi =
V0i
,
V0
(3.10)
where V0 is the initial value of the portfolio. By definition, these proportions

must sum to 1. For a portfolio with N assets this can be seen from writing
N
N

Vi
V0
Xi = i=1 0 =
= 1.
(3.11)
V
V
0
0
i=1
Furthermore, if an asset i is short-sold then its proportion is negative, so Xi < 0.

This again reflects the fact that short-selling is treated as a negative shareholding.
Example 29 Consider the portfolio in Example 23. The initial value of the
portfolio is 950 and the proportional holdings are
XA =
200
600
150
, XB =
, XC =
.
950
950
950
Example 30 A portfolio consists of a purchase of 100 of stock A at $5 each,

200 of stock B at $3 each and a short-sale of 150 of stock C at $2 each. The
total value of the portfolio is
V0 = 100 5 + 200 3 150 2 = 800.
The portfolio proportions are
5
6
3
XA = , XB = , XC = .
8
8
8
41
3.2. RETURN
Once the proportions have been calculated it is possible to evaluate the

return on the portfolio. Using the proportions, the return is the weighted average
of the returns on the individual assets. The return can be calculated using
r=
N
(3.12)
Xi ri .
i=1
Example 31 From the figures in Example 23, the returns on the stocks are
rA =
1
1
32
23
21
= , rB =
= , rC =
= 1,
2
2
3
3
1
and from Example 29 the initial proportions in the portfolio are

XA =
200
600
150
, XB =
, XC =
.
950
950
950
The return on the portfolio is therefore

200
1
1
600
150
r=

(1) = 0.052(52%).
+
+
950
2
950
3
950
It is important to note that the portfolio proportions are calculated at the
start of the holding period. If a series of returns is to be calculated over a
number of holding periods, the proportions must be recomputed at the start of
each of the holding periods. This is necessary to take into account variations in
the relative values of the assets. Those that have relatively larger increases in
value will gradually form a greater proportion of the portfolio.
Example 32 A portfolio consists of two stocks, neither of which pays any dividends. The prices of the stock over a three year period and the holding of each
is given in the table.
Stock
A
B
Holding
100
200
p (0)
10
8
p (1)
15
9
p (2)
12
11
p (3)
16
12
Example 33 The initial value of the portfolio is V0 = 10010+2008 = 2600,

so the portfolio proportions are
XA (0) =
1000
1600
5
8
= , XB (0) =
= .
2600
13
2600
13
The portfolio return over the first year is then

r=
15 10
8
98
5
= 0.269 (26.9%)
13
10
13
8
At the start of the second year, the value of the portfolio is V1 = 100 15 +
200 9 = 3300. This gives the new portfolio proportions as
XA (1) =
5
6
1500
1800
= , XB (0) =
= ,
3300
11
3300
11
42
and return
r=
11
12 15
15
11
11 9
9
= 0.03 (3%).
Finally, the proportions at the start of the third holding period are
XA (2) =
1200
2200
6
11
= , XB (2) =
= ,
3400
17
3400
17
and the return is

r=
3.2.4
16 12 11 12 11
6
= 0.176 (17.6%).
17
12
17
11
Mean Return
The examples have illustrated that over time the return on a stock or a portfolio
may vary. The prices of the individual stocks will rise and fall, and this will
cause the value of the portfolio to fluctuate. Once the return has been observed
for a number of periods it becomes possible to determine the average, or mean,
return. For the moment the mean return is taken just as an average of past
returns. We discuss later how it can be interpreted as a predictor of what may
be expected in the future.
If a return, on an asset or portfolio, is observed in periods 1, 2, 3, ..., T, the
mean return is defined as
T

rt
r=
,
(3.13)
T
t=1
where rt is the return in period t.
Example 34 Consider the following returns observed over 10 years.

Year
Return (%)
1
4
2
6
3
2
4
8
5
10
6
6
7
1
8
4
9
3
10
.
6
Example 35 The mean return is

r=
4 + 6 + 2 + 8 + 10 + 6 + 1 + 4 + 3 + 6
= 5%.
10
It should be emphasized that this is the mean return over a given period of
time. For instance, the example computes the mean return per year over the
previous ten years.
3.3
Variance and Covariance
The essential feature of investing is that the returns on the vast majority of
financial assets are not guaranteed. The price of stocks can fall just as easily as
43
3.3. VARIANCE AND COVARIANCE
they can rise, so a positive return in one holding period may become a negative
in the next. For example, an investment in the shares of Yahoo! Inc. would
have earned a return of 137% between October 2002 and September 2003. Three
years later the return from October 2005 through to September 2006 was 31%.
The following year the stock had a return of 2%. Changes of this magnitude in
the returns in different holding periods are not exceptional.
It has already been stressed that as well as caring about the return on an
asset or a portfolio and investor has to be equally concerned with the risk. What
risk means in this context is the variability of the return across different holding
periods. Two portfolios may have an identical mean return but can have very
different amounts of risk. There are few (if any) investors who would knowingly
choose to hold the riskier of the two portfolios.
A measure of risk must capture the variability. The standard measure of
risk used in investment analysis is the variance of return (or, equivalently, its
square root which is called the standard deviation). An asset with a return that
never changes has no risk. For this asset the variance of return is 0. Any asset
with a return that does vary will have a variance of return that is positive. The
more risk is the return on an asset the larger is the variance of return.
When constructing a portfolio it is not just the risk on individual assets that
matters but also the way in which this risk combines across assets to determine
the portfolio variance. Two assets may be individually risky, but if these risks
cancel when the assets are combined then a portfolio composed of the two assets
may have very little risk. The risks on the two assets will cancel if a higher than
average return on one of the assets always accompanies a lower than average
return on the other. The measure of the way returns are related across assets
is called the covariance of return. The covariance will be seen to be central to
understanding portfolio construction.
The portfolio variance and covariance are now developed by first introducing
the variance of return as a measure of the risk and then developing the concept
of covariance between assets.
3.3.1
Sample Variance
The data in Table 3.1 detail the annual return on General Motors stock traded
in New York over a 10 year period. Figure 3.1 provides a plot of this data.
The variability of the return, from a maximum of 36% to a minimum of - 41%,
can be clearly seen. The issue is how to provide a quantitative measure of this
variability.
Year
Return %
Year
Return %
93-94
36.0
98-99
-1.2
94-95
-9.2
99-00
25.3
95-96
17.6
00-01
-16.6
96-97
7.2
01-02
12.7
97-98
34.1
02-03
-40.9
Table 3.1: Return on General Motors Stock 1993-2003

The sample variance is a single number that summarizes the extent of the
variation in return. The process is to take the mean return as a measure of
44
40
30
20
10
0
-10
-20
93- 94- 95- 96- 97- 98- 99- 00- 01- 0294 95 96 97 98 99 00 01 02 03
-30
-40
-50
Figure 3.1: Graph of Return

the normal outcome. The difference between the mean and each observed
return is then computed - this is termed the deviation from the mean. Some
of these deviations from the mean are positive (in periods when the observed
return is above the mean) and some are negative (when the observed return is
below the mean). The deviations from the mean are then squared and these
squares are summed. The average is then obtained by dividing by the number
of observations.
With T observations, the sample variance just described is defined by the
formula
T
1
2
(rt r) .
(3.14)
2 =
T t=1
The sample standard deviation is the square root of the sample variance so

T
1

(rt r)2 .
(3.15)
=
T t=1
It should be noted that the sample variance and the sample standard deviation
are always non-negative, so 2 0 and 0. Only if every observation of the
return is identical is the sample variance zero.
There is one additional statistical complication with the calculation of the
variance. We can view the sample variance as being an estimate of the population variance of the return (meaning the true underlying value). The formula
given in (3.14) for the sample variance produces an estimate of the population
variance which is too low for small samples, that is when we have a small number
45
of observations. (Although it does converge to the true value for large samples.)
Because of this, we say that it is a biased estimator. There is an alternative
definition of the population variance which is unbiased. This is now described.
The unbiased estimator of the population variance is defined by
T
2T 1 =
1
(rt r)2 ,
T 1 t=1
(3.16)
with the unbiased estimator of the population standard deviation being

T
1
(rt r)2 .
(3.17)
T 1 =
T 1 t=1
Comparing the formulas (3.14) and (3.16) it can be seen that the distinction
between the two is simply whether the average value is found by dividing by T
or T 1.
Either of these formulas is perfectly acceptable for a calculation of the sample variance. All that matters is that the same formula is used consistently.
However, from this point onwards we will use division by T . It should be observed that as the number of observations increases, so T becomes large, the
difference between dividing by T and by T 1 becomes ever less important. For
very large values of T the two formulas provide approximately the same answer.
The next example calculates the sample variance of the return on General
Motors stock using the data in Table 3.1.
Example 36 For the returns on the General Motors stock, the mean return is
r = 6.5.
Using this value, the deviations from the mean and their squares are given by
Year
rt r
(rt r)2
Year
rt r
(rt r)2
93-94
29.5
94-95
-15.7
95-96
11.1
96-97
0.7
97-98
27.6
870.25
98-99
-7.7
246.49
99-00
18.8
123.21
00-01
-23.1
0.49
01-02
6.2
761.76
02-03
-47.4
59.29
353.44
533.61
38.44
2246.76
Example 37 After summing and averaging, the variance is

2 = 523.4.
3.3.2
Sample Covariance
Every sports fan knows that a team can be much more (or less) than the sum
of its parts. It is not just the ability of the individual players that matters but
46
how they combine. The same is true for assets when they are combined into
portfolios.
For the assets in a portfolio it is not just the variability of the return on
each asset that matters but also the way returns vary across assets. A set of
assets that are individually high performers need not combine well in a portfolio.
Just like a sports team the performance of a portfolio is subtly related to the
interaction of the component assets.
To see this point very clearly consider the example in Table 3.2. The table
shows the returns on two stocks for the holding periods 2006 and 2007. Over the
two years of data the mean return on each stock is 6 and the sample variances
of the returns are 2A = 2B = 16. Both stocks have a positive sample variance
so are individually risky investments.
Stock
A
B
Return in 2006 Return in 2007

10
2
2
10
Table 3.2
The outcome with respect to risk changes considerably when these stocks
are combined into a portfolio. Consider a portfolio that has proportion 12 of
stock A and 12 of stock B. With these proportions the return on the portfolio
in 2006 was
1
1
(3.18)
rp = 10 + 2 = 6,
2
2
and in 2007 the return was
1
1
rp = 10 + 2 = 6.
2
2
(3.19)
This gives the sample mean return on the portfolio as

rp =
6+6
= 6.
2
(3.20)
This value is the same as for the individual stocks. The key point is the sample
variance of the portfolio. Calculation of the sample variance gives
2p =
[6 6]2 + [6 6]2
= 0,
2
(3.21)
so the portfolio has no risk. What the example shows is that assets that are
individually risky can be combined into a portfolio in such a way that their
variability cancels and the portfolio has a constant return.
The feature of the example that gives rise to this result is that across the two
years a high return on one asset is accompanied by a low return on the other
asset. Put another way, as we move between years an increase in return on one
of the assets is met with an equal reduction in the return on the other. These
changes exactly cancel when the assets are placed into a portfolio. This example
teaches a fundamental lesson for portfolio theory: it is not just the variability
47
of asset returns that matters but how the returns on the assets move relative
to each other. In our example the moves are always in opposite directions and
this was exploited in the design of the portfolio to eliminate variability in the
return on the portfolio. The complete elimination of risk in the portfolio is an
extreme feature of the example. The general property of portfolio construction
is to obtain a reduction in risk by careful combination of assets.
In the same way that the variance is used to measure the variability of
return of an asset or portfolio, we can also provide a measure of the extent to
which the returns on different assets move relative to each other. To do this we
need to define the covariance between the returns on two assets, which is the
commonly-used measure of whether the returns move together or in opposite
directions.
The covariance takes the deviations from the mean return for the two assets
at time t, multiplies these together, sums over time and then averages. Hence,
when both assets have returns above the mean, or both below the mean, a
positive amount is contributed to the sum. Conversely, when one is below the
mean and the other above, a negative amount is contributed to the sum. It is
therefore possible for the covariance to be negative, zero or positive. A negative
value implies the returns on the two assets tend to move in opposite directions
(when one goes up, the other goes down) and a positive value that they tend to
move in the same direction. A value of zero shows that, on average, there is no
pattern of coordination in their returns.
To provide the formula for the covariance, let the return on asset A at time
t be rAt and the mean return on asset A be rA . Similarly, the return on asset
B at time t and the mean return are rBt and rB . The covariance of the return
between these assets, denoted AB , is
AB =
T
1
[rAt rA ] [rBt rB ] .
T t=1
(3.22)
By definition, for any asset i it follows from comparison of formula (3.14) for
the variance and (3.22) for the covariance that ii = 2i , so the covariance of
the return between an asset and itself is its variance. Also, in the formula for
the covariance it does not matter in which order we take asset A and asset B.
This implies that the covariance of A with B is the same as the covariance of B
with A or AB = BA .
Example 38 The table provides the returns on three assets over a three-year
period.
Asset
A
B
C
Year 1
10
10
12
Year 2
12
14
6
Year 3
11
12
9
48
Example 39 The mean returns are rA = 11, rB = 12, rC = 9. The covariance

between A and B is
AB =
1
[[10 11] [10 12] + [12 11] [14 12] + [11 11] [12 12]] = 1.333,
3
while the covariance between A and C is

AC =
1
[[10 11] [12 9] + [12 11] [6 9] + [11 11] [9 9]] = 2,
3
and that between B and C

BC =
1
[[10 12] [12 9] + [14 12] [6 9] + [12 12] [9 9]] = 4.
3
For a set of assets the variances and covariances between the returns are
often presented using a variance-covariance matrix. In a variance-covariance
matrix the entries on the main diagonal are the variances while those off the
diagonal are the covariances. Since ij = ji , only half the covariances need to
be presented. Usually it is those below the main diagonal. For three assets A, B
and C the variance-covariance matrix would be of the form
A
B
C
A
2A
AB
AC
B
2B
BC
C
.
2C
Example 40 For the data in Example 38, the variance-covariance matrix is

A
B
C
3.4
A
0.666
1.333
2
B
2.666
4
C
.
6
Population Return and Variance
The concept of sample mean return that we have developed so far looks back
over historical data to form an average of observed returns. The same is true
of the formulation of the sample variance and sample covariance. The sample
values are helpful to some degree to summarize the past behavior of returns but
what is really needed for investment analysis are predictions about what may
happen in the future. An investor needs this information to guide their current
investment decisions. We now discuss the extent to which the sample returns
and sample variances calculated on historical data can become predictions of
future outcomes.
A conceptual framework for analyzing future returns can be constructed
as follows: take an asset and determine the possible levels of return it may
achieve, and the probability with which each level of return may occur. For
3.4. POPULATION RETURN AND VARIANCE
49
instance, after studying its current business model we may feel that over the
next year IBM stock can achieve a return of 2% with probability 14 , of 4%
with probability 12 , and 6% with probability 14 . The possible payoffs, and the
associated probabilities, capture both the essence of randomness in the return
and the best view we can form on what might happen. It will be shown in this
section how forming predictions in this way can be used to construct measures
of risk and return.
Before proceeding to do this, it is worth reflecting on the link between this
approach and the calculations of sample means and sample variances using
historical data. At first sight, it would seem that the two are distinctly different
processes. However, there is a clear link between the two. This link follows
from adopting the perspective that the past data reflect the outcomes of earlier
random events. The observed data then constitute random draws from the set
of possible outcomes, with the rate of occurrence governed by a probability
distribution.
Adopting the usual approach of statistical analysis, the historical data on
observed returns are a sample from which we can obtain estimates of the true
values. The mean return we have calculated from the sample of observed returns
is a best estimate of the mean for the entire population of possible returns. The
mean return for the population is often called the expected return. The name
of mean is correctly used for the value calculated from the outcome of observation, while the name of expected is reserved for the statistical expectation.
However, since the mean return is the best estimate of the expected return, the
terms are commonly used interchangeably.
The same comments also apply to the sample variance and the sample covariance developed previously. They, too, are sample estimates of the population
variance and covariance. This was the point behind the discussion of the population variance being a measure of the true variance. The issue of unbiasedness
arose as a desirable property of the sample variance as an estimator of the
population variance.
3.4.1
Expectations
The first step in developing this new perspective is to consider the formation of
expectations. Although not essential for using the formulas developed below, it
is important for understanding their conceptual basis.
Consider rolling a dice and observing the number that comes up. This
is a simple random experiment that can yield any integer between 1 and 6
with probability 16 . The entire set of possible outcomes and their associated
probabilities is then

1
1
1
1
1
1
, 2,
, 3,
, 4,
, 5,
, 6,
.
(3.23)
1,
6
6
6
6
6
6
The expected value from this experiment can be thought of as the mean of
the outcome observed if the experiment was repeated very many times. Let x
50
denote the number obtained by observing a roll of the dice. This is one observation of the random variable X. The expected value of the random variable
is denoted E (X) and is given by the sum of possible outcomes, x, weighted by
their probabilities. For the dice experiment the expected value is
E [X] =
1
1
1
1
1
1
1 + 2 + 3 + 4 + 5 + 6 = 3.5.
6
6
6
6
6
6
(3.24)
Notice the interesting feature that the expected value of 3.5 is not an outcome
which will ever be observed - only the integers 1 to 6 ever appear. But this does
not prevent 3.5 being the expected value.
Expressed in formal terms, assume we have an random event in which there
are M possible outcomes. If outcome j is a value xj , and occurs with probability
j , then the expected value of the random variable X is
E [X] =
M
j xj .
(3.25)
j=1
The idea of taking an expectation is not restricted to just the observed

values of the random experiment. Return to the dice rolling example. For this
experiment we may also be interested in the expected value of X 2 . This can be
computed as
1
1
1
1
1
1
E X 2 = 1 + 4 + 9 + 16 + 25 + 36 = 15.167. (3.26)
6
6
6
6
6
6
This expression is just the value of each possible outcome squared, multiplied
by the probability and summed.
Observing this use of the expectation, we can recall that the variance is
defined as the average value of the square of the deviation from the mean. This,
too, is easily expressed as an expectation. For the dice experiment the expected
value was 3.5 (which we can use as the value of the mean), so the expected value
of the square of the deviation from the mean is

1
1
1
2
2
2
2
E (X E [X]) = [1 3.5] + [2 3.5] + [3 3.5]
6
6
6
1
1
1
+ [4 3.5]2 + [5 3.5]2 + [6 3.5]2 = 2.9167.
(3.27)
6
6
6
This is the population variance of the observed value of the dice rolling experiment.
3.4.2
Expected Return
The expectation can now be employed to evaluate the expected return on an

asset and a portfolio. This is achieved by introducing the idea of states of the
world. A state of the world summarizes all the information that is relevant
for the future return of an asset, so the set of states describes all the possible
51
different future financial environments that may arise. Of course, only one of
these states will actually be realized but when looking forward we do not know
which one. These states of the world are the analysts way of thinking about,
and modelling, what generates the randomness in asset returns.
Let there be M states of the world. If the return on an asset in state j is rj ,
and the probability of state j occurring is , then the expected return on asset
i is
E [r] = 1 r1 + ... + M rM ,
(3.28)
or, using the same notation as for the mean,
r =
M
(3.29)
j rj .
j=1
Example 41 The temperature next year may be hot, warm or cold. The returns
to stock in a food production company in each of these states are given in the
table.
State
Return
Hot
10
Warm
12
Cold
18
Example 42 If each states is expected to occur with probability 13 , the expected

return on the stock is
1
1
1
E [r] = 10 + 12 + 18 = 13.333.
3
3
3
This method of calculating the expected return can be generalized to determine the expected return on a portfolio. This is done by observing that the
expected return on a portfolio is the weighted sum of the expected returns on
each of the assets in the portfolio.
To see this, assume we have N assets and M states of the world. The return
on asset i in state j is rij and the probability of state j occurring is j. Let Xi
be the proportion of the portfolio invested in asset i. The return on the portfolio
in state j is found by weighting the return on each asset by its proportion in
the portfolio then summing
rP j =
N
(3.30)
Xi rij.
i=1
The expected return on the portfolio is found from the returns in the separate
states and the probabilities so
E [rP ] = 1 rP 1 + ... + M rP M .
(3.31)
The return on the portfolio in each state can now be replaced by its definition
in terms of the returns on the individual assets to give
E [rP ] =
N

i=1
1 Xi ri1 + ... +
N

i=1
M Xi riM ,
(3.32)
52
Collecting the terms for each asset

E [rP ] =
N
Xi [1 ri1 + ... + M riM ] ,
(3.33)
i=1
which can be written in brief as

rP =
N
Xi ri .
(3.34)
i=1
As we wanted to show, the expected return on the portfolio is the sum of the
expected returns on the assets multiplied by the proportion of each asset in the
portfolio.
Example 43 Consider a portfolio composed of two assets A and B. Asset A
constitutes 20% of the portfolio and asset B 80%. The returns on the assets in
the 5 possible states of the world and the probabilities of those states are given
in the table.
State
Probability
Return on A
Return on B
1
0.1
2
5
2
0.2
6
1
3
0.4
1
0
4
0.1
9
4
5
0.2
2
3
Example 44 The expected return on asset A is

rA = 0.1 2 + 0.2 6 + 0.4 1 + 0.1 9 + 0.2 2 = 3.1,
and that on asset B is
rB = 0.1 5 + 0.2 1 + 0.4 0 + 0.1 4 + 0.2 3 = 1.7.
The expected portfolio return is
rP = 0.2 3.1 + 0.8 1.7 = 1.98.
Notice that the same result is obtained by writing
rP = 0.1(0.2 2 + 0.8 5)+0.2(0.2 6 + 0.8 1)+0.4(0.2 1 + 0.8 0)
+ 0.1 (0.2 9 + 0.8 4) + 0.2 (0.2 2 + 0.8 3) = 1.98.
3.4.3
Population Variance
The population variance mirrors the interpretation of the sample variance as

being the average of the square of the deviation from the mean. But where the
sample variance found the average by dividing by the number of observations
(or one less than the number of observations), the population variance averages
53
by weighting each squared deviation from the mean by the probability of its
occurrence.
In making this calculation we follow the procedure introduced for the population mean of:
(i) Identifying the different states of the world;
(ii) Determining the return in each state;
(ii) Setting the probability of each state being realized.
We begin with the definition of the population variance of return for a single
asset. The population variance is expressed in terms of expectations by

2
(3.35)
2 = E (r E [r]) .
In this formula E [r] is the population mean return. This is the most general
expression for the variance which we refine into a form for calculation by making
explicit how the expectation is calculated.
To permit calculation using this formula the number of states of the world,
their returns and the probability distribution of states, must be specified. Let
there be M states, and denote the return on the asset in state j by rj . If the
probability of state j occurring is j , the population variance of the return on
the asset can be written as
2 =
M

j=1

j [rj r]2 .
(3.36)
Since it is a positively weighted sum of squares the population variance is always

non-negative. It can be zero, but only if the return on the asset is the same in
every state.
The population standard deviation is given by the square root of the variance, so

M

j [rj r]2 .
(3.37)
=
j=1
Example 45 The table provides data on the returns on a stock in the five possible states of the world and the probabilities of those states.
State
Return
Probability
1
5
.1
2
2
.2
3
-1
.4
4
6
.1
5
3
.2
Example 46 For this data, the population variance is

2
= .1 [5 3] + .2 [2 3] + .4 [1 3] + .1 [6 3] + .2 [3 3]
= 7.9.
54
3.4.4
Population Covariance
The sample covariance was introduced as a measure of the relative movement

of the returns on two assets. It was positive if the returns on the assets tended
to move in the same direction, and negative if they had a tendency to move
in opposite directions. The population covariance extends this concept to the
underlying model of randomness in asset returns.
For two assets A and B, the population covariance, AB , is defined by
AB = E [(rA E [rA ]) (rB E [rB ])] .
(3.38)
The expression of the covariance using the expectation provides the most general
definition. This form is useful for theoretical derivations but needs to be given
a more concrete form for calculations.
Assume there are M possible states of the world with state j having probability j . Denote the return to asset A in state j by rAj and the return to
asset B in state j by rBj . The population covariance between the returns on
two assets A and B can be written as
AB =
M

j=1
j [rAj rA ] [rBj rB ] ,
(3.39)
where rA and rB are the expected returns on the two assets.

The population covariance may be positive or negative. A negative covariance arises when the returns on the two assets tend to move in opposite
directions, so that if asset A has a return above its mean (rAj rA > 0) then
asset B has a return below its mean (rBj rB < 0) and vice versa. A positive
covariance arises if the returns on the assets tend to move in the same direction,
so both are either above the mean or both are below the mean.
Example 47 Consider the returns on three stocks in the following table. Assume the probability of the states occurring are: 1 = 12 , 2 = 14 , 3 = 14 .
State
Stock A
Stock B
Stock C
1
7
8
3
2
2
1
7
3
6
6
2
Example 48 The mean returns on the stocks can be calculated as rA = 5,

rB = 5 and rC = 4. The variance of return for the three stocks can be found as
2A
2B
2C
1
(7 5)2 +
2
1
(8 5)2 +
2
1
2
(3 4) +
2
1
(2 5)2 +
4
1
(1 5)2 +
4
1
2
(7 4) +
4
1
(6 5)2 = 4.5,
4
1
(6 5)2 = 8.75,
4
1
2
(2 4) = 3.75.
4
3.5. PORTFOLIO VARIANCE
55
The covariances between the returns are

1
1
1
(7 5) (8 5) + (2 5) (1 5) + (6 5) (6 5) = 6.25,
AB =
2
4
4
1
1
1
(7 5) (3 4) + (2 5) (7 4) + (6 5) (2 4) = 3.75,
AC =
2
4
4
1
1
1
BC =
(8 5) (3 4) + (1 5) (7 4) + (6 5) (2 4) = 5.0.
2
4
4
These can be summarized in the variance-covariance matrix
4.5
6.25
.
8.75
3.75 5.0 3.75
3.5
Portfolio Variance
The calculations of the variance of the return on an asset and of the covariance
of returns between two assets are essential ingredients to the determination of
the variance of a portfolio. It has already been shown how a portfolio may have
a very different variance from that of the assets from which it is composed. Why
this occurs is one of the central lessons of investment analysis. The fact that it
does has very significant implications for investment analysis
The variance of the return on a portfolio can be expressed in the same way
as the variance on an individual asset. If the return on the portfolio is denoted
by rP and the mean return by rP , the portfolio variance, 2P , is

2
2P = E (rP rP ) .
(3.40)
The aim now is to present a version of this formula from which the variance
can be calculated. Achieving this aim should also lead to an understanding of
how the variance of the return on the portfolio is related to the variances of the
returns on the individual assets and the covariances between the returns on the
assets.
The analysis begins by studying the variance of a portfolio with just two
assets. The result obtained is then extended to portfolios with any number of
assets.
3.5.1
Two Assets
Consider a portfolio composed of two assets, A and B, in proportions XA and

XB . Using the definition of the population variance, the variance of the return
on the portfolio is given by the expected value of the deviation of the return
from the mean return squared.
The analysis of portfolio return has shown that rP = XA rA + XB rB and
rP = XA rA + XB rB . These expressions can be substituted into the definition
of the variance of the return on the portfolio to write

2
2P = E ([XA rA + XB rB ] [XA rA + XB rB ]) .
(3.41)
56
Collecting together the terms relating to asset A and the terms relating to asset
B gives

2P = E (XA [rA rA ] + XB [rB rB ])2 .
(3.42)
Squaring the term inside the expectation

2
2
[rA rA ]2 + XB
[rB rB ]2 + 2XA XB [rA rA ] [rB rB ] .
2P = E XA
(3.43)
The expectation of a sum of terms is equal to the sum of the expectations of
the individual terms. This allows that variance to be broken down into separate
expectations

2
2
2
2
2P = E XA
[rA rA ] + E XB
[rB rB ]
+E [2XA XB [rA rA ] [rB rB ]] .
(3.44)
The portfolio proportions can then be extracted from the expectations because
they are constants. This gives

2
2
E [rA rA ]2 + XB
E [rB rB ]2
2P = XA
+2XA XB E [[rA rA ] [rB rB ]] .
(3.45)
The first expectation in this expression is the variance of return on asset A, the
second expectation is the variance of return on asset B, and the third expectation is the covariance of the returns of A and B. Employing these observation
allows the variance of the return on a portfolio of two assets, A and B, to be
written succinctly as
2 2
2 2
A + XB
B + 2XA XB AB .
2P = XA
(3.46)
The expression in (3.46) can be used to calculate the variance of the return
on the portfolio given the shares of the two assets in the portfolio, the variance
of returns of the two assets, and the covariance. The result has been derived
for the population variance (so the values entering would be population values)
but can be used equally well to calculate the sample variance of the return on
the portfolio using sample variances and sample covariance.
Example 49 Consider two assets A and B described by the variance-covariance
matrix

4
.
2 8
The variance of a portfolio consisting of
2P =
1
4
asset A and
3
4
12
32
13
4 + 8 + 2 2 = 8.125.
4
4
44
asset B is given by
57
Example 50 Consider two assets C and D described by the variance-covariance

matrix

6
.
3 9
The variance of a portfolio consisting of
2P =
2
3
asset C and
1
3
asset D is given by
22
12
21
(3) = 1.0.
6+ 9+2
3
3
33
It can be seen from formula (3.46) for the variance of return on a portfolio
that if the covariance between the two assets is negative, the portfolio variance
is reduced. This observation is emphasized in the example by the variance of
the portfolio of assets C and D being much lower that the portfolio of assets A
and B. The variance-reducing effect of combining assets whose returns have a
negative covariance is a fundamental result for investment analysis. It provides
a clear insight into how the process for constructing portfolios can reduce the
risk involved in investment.
3.5.2
Correlation Coefficient
The variance of the return on a portfolio can be expressed in an alternative way

that is helpful in the analysis of the next chapter. The covariance has already
been described as an indicator of the tendency of the returns on two assets
to move in the same direction (either up or down) or in opposite directions.
Although the sign of the covariance (whether it is positive or negative) indicates
this tendency, the value of the covariance does not in itself reveal how strong the
relationship is. For instance, a given value of covariance could be generated by
two assets that each experience large deviations from the mean but only have a
weak relationship between their movements or by two assets whose returns are
very closely related but individually do not vary much from their means.
In order to determine the strength of the relationship it is necessary to
measure the covariance relative to the deviation from the mean experienced
by the individual assets. This is achieved by using the correlation coefficient
which relates the standard deviations and covariance. The correlation coefficient
between the return on asset A and the return on asset B is defined by
AB =
AB
.
A B
(3.47)
The value of the correlation coefficient satisfies 1 AB 1.

A value of AB = 1 indicates perfect positive correlation: the returns on the
two assets always move in unison. Interpreted in terms of returns in different
states of the world, perfect positive correlation says that if the return on one
asset is higher in state j than it is in state k, then so is the return on the other
asset. Conversely, AB = 1 indicates perfect negative correlation: the returns
on the two assets always move in opposing directions, so if the return on one
58
asset is higher in state j than it is in state k, then the return on the other asset
is lower in state j than in state k.
Using the correlation coefficient, the variance of the return of a portfolio can
be written as
2 2
2 2
A + XB
B + 2XA XB AB A B .
(3.48)
2P = XA
It can be seen from this formula that a negative correlation coefficient reduces
the overall variance of the portfolio.
Example 51 A portfolio is composed of 12 of asset A and 12 of asset B. Asset
A has a variance of 25 and asset B a variance of 16. The covariance between
the returns on the two assets is 10. The correlation coefficient is
AB =
10
= 0.5,
54
and the variance of return on the portfolio is

2P =
3.5.3
2
2

1
1
1
1
25 +
16 + 2
0.5 25 16 = 110.25.
2
2
2
2
General Formula
The formula to calculate the variance of the return on a portfolio can now be
extended to accommodate any number of assets. This extension is accomplished
by noting that the formula for the variance of the return on a portfolio involves
the variance of each asset plus its covariance with every other asset.
For N assets in proportions Xi , i = 1, ..., N, the variance is therefore given
by
N
N

2 2
(3.49)
2P =
Xi Xk ik .
Xi i +
i=1
k=1
k=i
It should be confirmed that when N = 2 this reduces to (3.46). The presentation

of the formula can be simplified by using the fact that ii is identical to 2i to
write
N
N

2
Xi Xk ik .
(3.50)
P =
i=1 k=1
This formula can also be expressed in terms of the correlation coefficients.

The significance of this formula is that it provides a measure of the risk of
any portfolio, no matter how many assets are included. Conceptually, it can
be applied even to very large (meaning thousands of assets) portfolios. All the
information that is necessary to do this are the proportionate holdings of the
assets and the variance-covariance matrix. Later chapters consider how this
informational requirement can be reduced even further.
59
Example 52 A portfolio consists of three assets, A, B, and C. The portfolio

proportions are XA = 16 , XB = 12 , and XC = 13 . The variance-covariance matrix
is
3
4 12
.
2 1 9
The formula for the variance of the portfolio is
2 2
2 2
A + XB
B + XC2 2C + 2XA XB AB + 2XA XC AC + 2XB XC BC .
2P = XA
Using the data describing the portfolio

2
2
2
1
1
1
11
11
11
2
2
2
A +
B +
2C + 2 AB + 2 AC + 2 BC
P =
6
2
3
62
63
23
1 2
1
1
1
1
1
=
+ 2 + 2 + AB + AC + BC .
36 A 4 B 9 C 6
9
3
Substituting in from the variance-covariance matrix
2P
1
1
1
1
1
1
3 + 12 + 9 + 4 + 2 + (1)
36
4
9
6
9
3
= 4.6389.
=
Example 53 A portfolio consists of three assets, A, B, and C. The portfolio

proportions are XA = 14 , XB = 14 , and XC = 12 . The variances of the returns
on the individual assets are 2A = 16, 2B = 25, and 2C = 36. The correlation
coefficients between the returns are AB = 0.5, BC = 0.25, and AC = 0.75.
The formula for the variance of the portfolio is
2 2
2 2
2 2
A +XB
B +XC
C +2XA XB A B AB +2XA XC A C AC +2XB XC B C BC .
2P = XA
For the data describing the assets and the portfolio

2P
11
11
12
12
12
11
4 5 0.5 + 2
4 6 (0.75) + 2
5 6 0.25
16 + 25 + 36 + 2
4
4
2
44
42
42
= 5.375.
=
3.5.4
Effect of Diversification
As an application of the formula for the variance of the return of a portfolio this
section considers the effect of diversification. Diversification means purchasing
a larger number of different assets. It is natural to view diversification as a
means of reducing risk because in a large portfolio the random fluctuations of
individual assets will have a tendency to cancel out.
To formalize the effect of diversification, consider holding N assets in equal
proportions. This implies that the portfolio proportions satisfy Xi = N1 for all
assets i = 1, ..., N . From (3.49), the variance of this portfolio is
2
2
N
N

1
1
(3.51)
2P =
2i +
ik .
N
N
i=1
k=1,k=i
60
Observe that there are N terms in the first summation and N [N 1] in the
second. This suggests extracting a term from each summation to write the
variance as

N

N
N
1
N 1
1
1
2i +
ik .
(3.52)
2P =
N i=1 N
N
[N 1] N
i=1
k=1,k=i
Now define the mean of the variances of the N assets in the portfolio by
2a

N

1
=
2i ,
N
i=1
(3.53)
and the mean covariance between all pairs of assets in the portfolio by
ab =

N
N

i=1 k=1,k=i

1
ik .
[N 1] N
(3.54)
Using these definitions, the variance of the return on the portfolio becomes

1
N 1
2a +
ab .
2P =
(3.55)
N
N
This formula applies whatever the number of assets (but the mean variance and
mean covariance change in value as N changes).
Diversification means purchasing a broader range of assets which in the
present context is reflected in an increase in N. The extreme of diversification occurs as the number of assets in the portfolio is increased without limit.
Formally, this can be modelled by letting N and determining the effect
on the variance of the return on the portfolio.
It can be seen from (3.55) that as N the first term will converge to
zero (we are dividing the mean value by an ever increasing value of N ) and
the second term will converge to
ab (because as N increases N1
N tends to 1).
Therefore, at the limit of diversification
ab .
2P
(3.56)
This result shows that in a well-diversified portfolio only the covariance

between assets counts for portfolio variance. In other words, the variance of
the individual assets can be eliminated by diversification - which confirms the
initial perspective on the consequence of diversification.
3.6
Summary
The most basic information about assets is captured in their mean and variance
which are used by analysts as the measures of return and risk. This chapter
has shown how the sample return, sample variance and sample covariance can
61
3.6. SUMMARY
be calculated from data on individual assets. It has also shown how these can
be combined into measures of risk and return for portfolios, including portfolios
with short-selling of one or more assets.
These ideas were then extended to the calculation of population mean, variance and covariance. The calculation of population values was based upon the
idea that the sample data was a random draw from an underlying population.
Following this approach lead to the concept an expected value. The concepts
involved in calculating population values capture the very essence of unpredictability in financial data.
Finally, the chapter applied the concept of the population variance as an expectation to calculate the variance of return on a portfolio. The importance of
the covariance between the returns on the assets for this variance was stressed.
This was emphasized further by presenting the variance in terms of the correlation coefficient and by demonstrating how diversification reduced the portfolio
variance to the average of the covariances between assets in the portfolio.
Exercise 18 A 1969 Jaguar E-type is purchased at the beginning of January
2002 for $25000. At the end of December 2002 it is sold for $30000.
a. Given these figures, what was the return to the investment in the Jaguar?
b. Now assume the car was entered in a show and won a $500 prize. What
does the return now become?
c. If in addition, it cost $300 to insure the car and $200 to service it, what
is the return?
Exercise 19 The following prices are observed for the stock of Fox Entertainment Group Inc.
Date
Price
June 00
26.38
June 01
28.05
June 02
25.15
June 03
.
28.60
Exercise 20 No dividend was paid. Calculate the mean return and variance of
Fox stock.
Exercise 21 The returns on a stock over the previous ten years are as given in
the table.
Year
Return (%)
1
1
2
-6
3
4
4
12
5
2
6
-1
7
3
8
8
9
2
10
.
12
Exercise 22 Determine the mean return on the stock over this period and its
variance.
Exercise 23 The prices of three stocks are reported in the table.
Brunswick Corporation
Harley-Davidson Inc.
Polaris Industries Partners
June 00
16.56
8.503
31.98
June 01
24.03
47.08
45.80
June 02
28.00
51.27
65.00
June 03
23.00
43.96
63.04
62
Exercise 24 During these years, the following dividends were paid

Brunswick Corporation
Harley-Davidson Inc.
Polaris Industries Partners
00-01
0.52
0.12
0.94
01-02
0.26
0.09
0.53
02-03
0.50
0.12
1.18
Exercise 25 a. For each stock, calculate the return for each year and the mean
return.
b. Compute the return to a portfolio consisting of 100 Brunswick Corporation
stock and 200 Harley-Davidson Inc. stock for each year.
c. For a portfolio of 100 of each of the stock, calculate the portfolio proportions at the start of each holding period. Hence compute the return to the
portfolio.
Exercise 26 For the data in Exercise 23 calculate the variance of return for
each stock and the covariances between the stock. Discuss the resulting covariances paying particular attention to the market served by the companies. (If you
do not know these companies, descriptions of their activities can be found on
finance.yahoo.com.)
Exercise 27 Assume that there are 2 stocks and 5 states of the world. Each
state can occur with equal probability. Given the returns in the following table,
calculate the expected return and variance of each stock and the covariance between the returns. Hence find the expected return and variance of a portfolio
with equal proportions of both stock. Explain the contrast between the variance
of each stock and the portfolio variance.
Stock A
Stock B
State 1
5
9
State 2
7
6
State 3
1
5
State 4
8
4
State 5
3
8
Exercise 28 Given the following variance-covariance matrix for three securities, calculate the standard deviation of a portfolio with proportional investments
in the assets XA = 0.2, XB = 0.5 and XC = 0.3.
Security A
Security B
Security C
Security A
24
12
10
Security B
Security C
32
-8
48
Exercise 29 Consider the following standard deviations and correlation coefficients for three stocks.
Stock
A
B
C
9
6
10
Correlation
A
1
0.75
0.5
with
B
0.75
1
0.2
stock
C
0.5
0.2
1
3.6. SUMMARY
63
Exercise 30 a. Calculate the standard deviation of a portfolio composed of

50% of stock A and 50% of stock C.
b. Calculate the standard deviation of a portfolio composed of 20% of stock
A, 60% of stock B and 20% of stock C.
c. Calculate the standard deviation of a portfolio composed of 70% of stock
A, 60% of stock B and a short sale of C.
Exercise 31 From finance.yahoo.com, find the historical price data on IBM
stock over the previous ten years. Calculate the return each year, the mean
return and the variance. Repeat for the stock of General Motors and Boeing.
Hence find the expected return and variance of a portfolio consisting off 20%
IBM, 30% General Motors and 50% Boeing.
64
Chapter 4
The Efficient Frontier

To make a good choice we must first know the full range of alternatives. Once these are known it may be found that some can
dismissed as poor, simply giving less of what we want and more of
what we dont want. These alternative should be discarded. From
what is left, the choice should be made. In finance terms, no investor
wishes to bear unnecessary risk for the return the return that they
are achieving. This implies being efficient and maximizing return
for given risk. Given this, what remains is to choose the investment
strategy that makes the best trade-off between risk and return. An
investor needs to know more than just the fact that there is trade
off between the two. What it is necessary to find is the relationship
between risk and return as portfolio composition is changed. We already know that this relationship must depend on the variances of
the asset returns and the covariance between them. The relationship
that we ultimately construct is the efficient frontier. This is the set
of efficient portfolios from which a choice is made.
4.1
Introduction
The investment decision involves the comparison of the returns and risks of
different potential portfolios. The calculations of the previous chapter have
shown how to determine the expected return on a portfolio and the variance
of return. To make an informed choice of portfolio an investor needs to know
the possible combinations of risk and return that can be achieved by alternative
portfolios. Only with this knowledge is it possible to make an informed choice
of portfolio.
The starting point for investigating the relationship between risk and return
is a study of portfolios composed of just two risky assets with no short-selling.
The relationship between risk and return that is constructed is termed the portfolio frontier and the shape of the frontier is shown to depend primarily upon
65
66
CHAPTER 4. THE EFFICIENT FRONTIER
the coefficient of correlation between the returns on the two assets. The concept of a minimum variance portfolio is introduced and the efficient frontier
the set of assets with maximum return for a given level of risk is identified.
The minimum variance portfolio is later shown to place a central role in the
identification of efficient portfolios.
The restrictions on the number of assets and on short-selling are then relaxed in order to move the analysis closer to practical application. Permitting
short-selling is shown to extend the portfolio frontier but not to alter its shape.
Introducing additional risky assets generalizes the portfolio frontier into the
portfolio set, but the idea of an efficient frontier is retained. The extensions
are completed by allowing a risk-free asset, both with a single interest rate and
differing interest rates for borrowing and lending.
The outcome of this analysis is the identification of the set of portfolios from
which an investor should choose, and the set of portfolios that should not be
chosen. This information is carried into the next chapter where the efficient set
is confronted with preferences.
4.2
Two-Asset Portfolios
The analysis begins by considering the risk and return combinations offered
by portfolios composed of two risky assets. We start by assuming that there
is no risk-free asset and short sales are not possible. This simple case is the
basic building block for the analysis of more general situations that relax the
assumptions.
The two risky assets are labelled A and B. It is assumed that the expected
return on asset A is less than that of asset B, so rA < rB . For any investor
to choose asset A it must offer a lower variance of return than asset B. It is
therefore assumed that 2A < 2B . If these conditions were not met, either one
asset would never be chosen or, if the return and variance of both were the same,
the two assets would be identical and no issue of choice would arise.
A portfolio is described by proportional holdings XA and XB of the assets
with the property that XA + XB = 1. Ruling out short sales implies that the
holdings of both assets must be positive, so XA 0 and XB 0. The focus
of attention is the relation between the standard deviation of the return on
the portfolio, p , and the expected return of the portfolio, rp , as the portfolio
proportions XA and XB are varied. The reason for this interest is that this
relationship reveals the manner in which an investor can trade risk for return
by varying the composition of the portfolio.
Recall from (3.48) that the standard deviation of the return on a two-asset
portfolio is given by
2 2
1/2
2 2
A + XB
B + 2XA XB AB A B
.
p = XA
(4.1)
Now consider the variances of the two assets and the proportional holdings to
be given. The standard deviation of the return on the portfolio then depends
4.2. TWO-ASSET PORTFOLIOS
67
only upon the value of the correlation coefficient, AB . This observation motivates the strategy of considering how the standard deviation/expected return
relationship depends on the value of the correlation coefficient.
The analysis now considers the two limiting cases of perfect positive correlation and perfect negative correlation, followed by the intermediate case.
Case 1: AB = +1 (Perfect Positive Correlation)
The first case to consider is that of perfect positive correlation where AB = +1.
As discussed in Chapter 3, this can be interpreted as the returns on the assets
always rising or falling in unison.
Setting AB = +1, the standard deviation of the return on the portfolio
becomes
1/2
2 2
2 2
A + XB
B + 2XA XB A B
.
(4.2)
p = XA
The term within the brackets is a perfect square so its square root can be written
explicitly. Taking the square root gives the solution for the standard deviation
as
(4.3)
p = XA A + XB B .
Equation (4.3) shows that the standard deviation of the return on the portfolio
is a obtained as a weighted sum of the standard deviations of the returns on the
individual assets, where the weights are the portfolio proportions. This result
can be complemented by employing (3.12) to observe that the expected return
on the portfolio is
rp = XA rA + XB rB ,
(4.4)
so the expected return on the portfolio is also a weighted sum of the expected
returns on the individual assets.
Example 54 provides an illustration of the risk/return relationship that is
described by equations (4.3) and (4.4).
Example 54 Let asset A have expected return rA = 1 and standard deviation
A = 2 and asset B have expected return rB = 10 and standard deviation
B = 8. Table 4.1 gives the expected return and standard deviation for various
portfolios of the two assets when the returns are perfectly positively correlated.
These values are graphed in Figure 4.1.
XA 0
0.25 0.5 0.75 1
XB 1
0.75 0.5 0.25 0
rp
10 7.75 5.5 3.25 1
p
8
6.5
5
3.5
2
Table 4.1: Perfect Positive Correlation
As Example 54 illustrates, because the equations for portfolio expected return and standard deviation are both linear the relationship between p and
rp is also linear. This produces a straight line graph when expected return is
plotted against standard deviation. The equation of this graph can be derived
68
rp
12
10
10
Figure 4.1: Risk and Return

as follows. The portfolio weights must sum to 1 so XB = 1 XA . Substituting
for XB in (4.3) and (4.4), and then eliminating XA between the equations gives

rB A rA B
rA rB
rp =
(4.5)
+
p .
A B
A B
This result makes precise the details of the linear relationship between expected
return and standard deviation. It can be easily checked that the data in Table
4.1 satisfy equation (4.5).
The investment implication of the fact that the frontier is a straight line is
that the investor can trade risk for return at a constant rate. Therefore, when
the returns on the assets are perfectly positively correlated, each extra unit of
standard deviation that the investor accepts has the same reward in terms of
additional expected return.
The relationship that we have derived between the standard deviation and
the expected return is called the portfolio frontier. It displays the trade-off
that an investor faces between risk and return as they change the proportions
of assets A and B in their portfolio. Figure 4.2 displays the location on this
frontier of some alternative portfolio proportions of the two assets. It can be
seen in Figure 4.2 that as the proportion of asset B (the asset with the higher
standard deviation) is increased the location moves up along the frontier. It is
important to be able to locate different portfolio compositions on the frontier as
this is the basis for understanding the consequences of changing the structure
of the portfolio.
69
rp
X A = 0, X B = 1
rB
X A = 0.5, X B = 0.5
X A = 1, X B = 0
rA
Figure 4.2: Asset Proportions on the Frontier

Case 2: AB = 1 (Perfect Negative Correlation)
The second case to consider is that of perfect negative correlation with AB =
1. Perfect negative correlation occurs when an increase in the return on one
asset is met with by a reduction in the return on the other asset.
With AB = 1 the standard deviation of the portfolio becomes
2 2
1/2
2 2
A + XB
B 2XA XB A B
.
p = XA
(4.6)
The term expression within the brackets is again a perfect square but this time
the square root has two equally valid solutions. The first solution is given by
p = XA A XB B ,
(4.7)
p = XA A + XB B .
(4.8)
and the second is

It is easily checked that these are both solutions by squaring them and recovering
the term in brackets.
The fact that there are two potential solutions makes it necessary to determine which is applicable. This question is resolved by utilizing the fact that
a standard deviation can never be negative. The condition that p must be
non-negative determines which solution applies for particular values of XA and
XB , since when one gives a negative value for the standard deviation, the other
will give a positive value. For instance, if B > A , then (4.7) will hold when
XA is large relative to XB and (4.8) will hold when XA is small relative to XB .
70
rp
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
Figure 4.3: Perfect Negative Correlation

A = 2 and asset B have expected return rB = 4 and standard deviation B = 6.
Table 4.2 gives the expected return and standard deviation predicted by (4.7)
and (4.8) for various portfolios of the two assets when the returns are perfectly
negatively correlated. The positive values are graphed in Figure 4.3.
XA
0 0.25 0.5 0.75 1
XB
1 0.75 0.5 0.25 0
rp
4
3.5
3
2.5
2
p (4.7) -6
-4
-2
0
2
p (4.8) 6
4
2
0
-2
Table 4.2: Perfect Negative Correlation
The important fact about the portfolio frontier for this example is that the
portfolio XA = 34 , XB = 14 has a standard deviation of return, p , that is zero.
This shows that the two risky assets have combined into a portfolio with no risk
(we have already observed this possibility in Section 3.3.2). That a portfolio
with standard deviation of zero can be constructed from two risky assets is a
general property when there is perfect negative correlation.
To find the portfolio with a standard deviation of zero, substitute XB =
1 XA into either (4.7) or (4.8) and set p = 0. Then both (4.7) and (4.8)
71

provide the expression
XA A [1 XA ] B = 0.
(4.9)
Solving this equation for the proportion of asset A in the portfolio gives
XA =
B
,
A + B
(4.10)
which, using the fact that the proportions must sum to 1, implies the proportion
of asset B is
A
.
(4.11)
XB =
A + B
A portfolio with the two assets held in these proportions will have a standard
deviation of p = 0. The values in Example 55 can be confirmed using these
solutions.
Example 56 Let asset A have standard deviation A = 4 and asset B have
6
4
= 35 and XB = 4+6
= 25 . Hence
and standard deviation B = 6. The XA = 4+6
the standard deviation is

1/2
9
4
32
= 0.
16 +
36 2
46
p =
25
25
55
The general form of the portfolio frontier for AB = 1 is graphed in Figure
4.4 where the positive parts of the equations are plotted. This again illustrates
the existence of a portfolio with a standard deviation of zero. The second
important observation to be made about the figure is that for each portfolio on
the downward sloping section there is a portfolio on the upward sloping section
with the same standard deviation but a higher return. Those on the upward
sloping section therefore dominate in terms of offering a higher return for a given
amount of risk. This point will be investigated in detail later.
Case 3: 1 < AB < +1
For intermediate values of the correlation coefficient the frontier must lie between that for the two extremes of AB = 1 and AB = 1. It will have a
curved shape that links the positions of the two assets.
A = 2 and asset B have expected return rB = 8 and standard deviation B = 6.
Table 4.3 gives the expected return and standard deviation for various portfolios
of the two assets when AB = 12 . These values are graphed in Figure 4.5.
XA 0 0.125 0.25 0.375 0.5 0.625 0.75 0.875
XB 1 0.875 0.75 0.625 0.5 0.375 0.25 0.125
rp
8 7.25
6.5
5.75
5
4.25
3.5
2.75
p 6 5.13 4.27 3.44 2.65 1.95 1.50 1.52
Table 4.3: Return and Standard Deviation with AB = 12
1
0
2
2
72
rp
X A = 0, X B = 1
rB
X A = 0.5, X B = 0.5
rA
X A = 1, X B = 0
Figure 4.4: Portfolio Frontier with Perfect Negative Correlation
rp
9
8
7
6
5
4
3
2
1
0
0
Figure 4.5: Portfolio Frontier with Negative Correlation
73
It can be seen in Figure 4.5 that there is no portfolio with a standard deviation of zero, but there is a portfolio that minimizes the standard deviation.
This is termed the minimum variance portfolio and is the portfolio located at
the point furthest to the left on the portfolio frontier. The composition of the
minimum variance portfolio is implicitly defined by its location on the frontier.
Referring back to Table 4.3 it can be seen that for the data in Example 4.5 this
portfolio has a value of XA somewhere between 0.625 and 0.875. We will see
later how to calculate exactly the composition of this portfolio.
The observation that there is a minimum variance portfolio is an important
one for investment analysis. It can be seen in Figure 4.5 that portfolios with a
lower expected return than the minimum variance portfolio are all located on the
downward-sloping section of the portfolio frontier. As was the case for perfect
negative correlation, for each portfolio on the downward sloping section there
is a portfolio on the upward-sloping section with a higher excepted return but
the same standard deviation. Conversely, all portfolios with a higher expected
return than the minimum variance portfolio are located on the upward sloping
section of the frontier. This leads to the simple rule that every efficient portfolio
has an expected return at least as large as the minimum variance portfolio.
Example 58 Over the period September 1998 to September 2003, the annual
returns on the stock of African Gold (traded in the UK) and Walmart (traded
in the US) had a covariance of 0.053 (ignoring currency variations). The
variance of the return on African Gold stock was 0.047 and that on Walmart
was 0.081. These imply that the correlation coefficient is = 0.858. The portfolio
frontier for these stocks is graphed in Figure 4.6 where point A corresponds to
a portfolio composed only of African Gold stock and point B a portfolio entirely
of Walmart stock.
The analysis of the different values of the correlation coefficient in Cases 1 to
3 can now be summarized. With perfect positive correlation the portfolio frontier is upward sloping and describes a linear trade-off of risk for return. At the
opposite extreme of perfect negative correlation, the frontier has a downwardsloping section and an upward-sloping section which meet at a portfolio with
minimum variance. For any portfolio on the downward-sloping section there
is a portfolio on the upward-sloping section with the same standard deviation
but a higher return. Intermediate values of the correlation coefficient produce
a frontier that lies between these extremes. For all the intermediate values, the
frontier has a smoothly-rounded concave shape. The minimum variance portfolio separates inefficient portfolios from efficient portfolios. This information is
summarized in Figure ??.
The following sections are devoted to generalizing the assumptions under
which the portfolio frontier has been constructed. The first step is to permit
short selling of the assets but to retain all the other assumptions. The number
of assets that can be held in the portfolio is then increased. Finally, a risk-free
asset is introduced.
74
rp
B
Figure 4.6: African Gold and Walmart
rp
AB = 1
rB
AB = 1
rA
1 < AB < 1
Figure 4.7: Correlation and Portfolio Frontier
75
4.3. SHORT SALES
4.3
Short Sales
Permitting short sales removes the non-negativity restriction on the proportions

of the two assets in the portfolio. With short-selling the proportion of an asset
held can be negative but the proportions must still sum to unity. This allows
both positive and negative values of the portfolio proportions XA and XB . The
only restrictions is that XA + XB = 1. For example, if asset A is sold short, so
XA < 0, then there must be a correspondingly long position in asset B with
XB > 1.
The effect of allowing short sales is to extend the frontier beyond the limits
defined by the portfolios {XA = 0, XB = 1} and {XA = 1, XB = 0}. The consequences of this change can be easily illustrated for the case of perfect positive
correlation. Using (4.4), and the substitution XB = 1XA , the expected return
is given by
rp = XA rA + [1 XA ] rB .
(4.12)
Similarly, from (4.3) the standard deviation is
p = XA A + [1 XA ] B .
(4.13)
Without short sales, equations (4.4) and (4.3) hold only for values of XA that
satisfy 0 XA 1. But with short selling they are defined for all values of XA
that ensure p 0, which is the requirement that the standard deviation must
remain positive. This restriction provides a range of allowable proportions XA
that is determined by A and B
Asset A has expected return rA = 2 and standard deviation A = 4. Asset
B has expected return rB = 4 and standard deviation B = 10. Then p 0 if
XA 53 and hence XB 23 . The portfolio frontier is graphed in Figure 4.8.
Note that the choice of XA = 53 , XB = 23 produces a portfolio with rp = 0.5
and p = 0. Therefore, short selling can produce a safe portfolio when asset
returns are perfectly positively correlated.
The effect of short-selling in the general case of 1 < < 1 is to extend the
frontier as illustrated in Figure 4.9. The interpretation of points on the portfolio
frontier in terms of the assets proportions needs to be emphasized. Extending
the frontier beyond the portfolio composed solely of asset A is possible by going
long in asset A and short-selling B. Moving beyond the location of asset B
is possible by short-selling A and going long in B. The importance of these
observations will be become apparent when the choice of a portfolio by an
investor is considered in Chapter 5.
4.4
Efficient Frontier
The important role of the minimum variance portfolio has already been described. Every point on the portfolio frontier with a lower expected return than
the minimum variance portfolio is dominated by others which has the same standard deviation but a higher return. It is from among those assets with a higher
76
rp 7
X A = 1, X B = 2
5
4
X A = 0, X B = 1
3
2
X A = 1, X B = 0
1
X A = 5 / 3, X B = 2 / 3
0
0
10
15
p
20
Figure 4.8: Short Selling with Perfect Positive Correlation
77
4.4. EFFICIENT FRONTIER
rp
X A < 0, X B > 1
rB
X A > 0, X B > 0
rA
X A > 1, X B < 0
Figure 4.9: The Effect of Short Selling
return than the minimum variance portfolio that an investor will ultimately
make a choice. The minimum variance portfolio separates efficient portfolios
that may potentially be purchased from inefficient ones that should never be
purchased.
The set of portfolios with returns equal to, or higher than, the minimum
variance portfolio is termed the efficient frontier. The efficient frontier is the
upward section of the portfolio frontier and is the set from which a portfolio
will actually be selected. The typical form of the efficient frontier is shown in
Figure 4.10.
For every value of AB there is a portfolio with minimum variance. The
calculation of the proportional holdings of the two assets that constitute the
minimum variance portfolio is an important component of the next step in the
analysis. The proportions of the two assets are found by minimizing the variance
of return. The variance can in be expressed terms of the proportion of asset A
alone by using the substitution XB = 1 XA . The minimum variance portfolio
then solves
2 2
A + [1 XA ]2 2B + 2XA [1 XA ] AB A B .
min 2p XA
{XA }
(4.14)
Differentiating with respect to XA , the first-order condition for the minimization
78
rp
Efficient Frontier
r mvp
mvp
Figure 4.10: The Efficient Frontier

problem is
2p
XA 2A [1 XA ] 2B + [1 XA ] AB A B XA AB A B = 0. (4.15)
XA
Solving the necessary condition for XA gives the portfolio proportion
XA =
2A
2B A B AB
.
+ 2B 2A B AB
(4.16)
For a two-asset portfolio, this portfolio proportion for asset A (and the implied
proportion in asset B) characterizes the minimum variance portfolio for given
values of A , B and AB .
Example 59 With perfect positive correlation,
XA =
2A
2B A B
B
=
,
+ 2B 2 A B
B A
and with perfect negative correlation

XA =
2A
2B + A B
B
=
.
+ 2B + 2 A B
A + B
79
4.5. EXTENSION TO MANY ASSETS

When the assets are uncorrelated
XA =
2A
2B
.
+ 2B
Example 60 Using the data for Example 58, the minimum variance portfolio
of African Gold stock and Walmart stock is given by
1
XA
XB
0.081 + 0.047 2 0.081 2 0.858

= 0.57,
=
1
1
0.047 + 0.081 + 2 0.047 2 0.081 2 0.858
= 0.43,
where asset A is African Gold stock and asset B is Walmart stock. Given
an expected return on African Gold stock of 0.1 and an expected return on
Walmart stock of 0.2, the expected return on this portfolio is
rp = 0.1 0.57 + 0.2 0.43 = 0.029,
and the standard deviation is

12
1
1
p = 0.572 0.047 + 0.432 0.081 2 0.57 0.43 0.047 2 0.081 2 0.858 = 0.06.
Refer back to Figure 4.6. In the figure point A corresponds to a portfolio composed entirely of African Gold stock and point B to a portfolio entirely composed
of Walmart stock. It can be seen that the efficient frontier consists of all portfolios with a Walmart holding of at least 43% and an African Gold holding of
at most 57%.
4.5
Extension to Many Assets
The next step in the analysis is to introduce additional risky assets. The first
consequence of the introduction of additional assets is that it allows the formation of many more portfolios. The definition of the efficient frontier remains
that of the set of portfolios with the highest return for a given standard deviation. But, rather than being found just by varying the proportions of two
assets, it is now constructed by considering all possible combinations of assets
and combinations of portfolios.
The process of studying these combinations of assets and portfolios is eased
by making use of the following observation: a portfolio can always be treated
as if it were a single asset with an expected return and standard deviation.
Constructing a portfolio by combining two other portfolios is therefore not analytically different from combining two assets. So, when portfolios are combined,
the relationship between the expected return and the standard deviation as the
proportions are varied generates a curve with the form discussed above. The
shape of this curve will again be dependent upon the coefficient of correlation
between the returns on the portfolios.
80
rp
D
4
2
B
1
3
C
Figure 4.11: Construction of Portfolio Set
This is illustrated in Figure 4.11 for three assets. Combining assets A and B
produces the first solid curve. Combining assets C and D produces the second
solid curve. Then combining portfolio 1 on first curve with portfolio 2 on second
curve produces the first dashed curve. Then combining portfolio 3 on first curve
with portfolio 4 on second curve produces the second dashed curve. This process
can be continued by choosing a portfolio on one curve and combining it with a
portfolio from another curve.
This process of forming combinations can be continued until all possible
portfolios of the underlying assets have been constructed. As already described,
every combination of portfolios generates a curve with the shape of a portfolio
frontier. The portfolio frontier itself is the upper envelope of the curves found
by combining portfolios. Graphically, it is the curve that lies outside all other
frontiers and inherits the general shape of the individual curves. Hence, the
portfolio frontier is always concave. The efficient frontier is still defined as the
set of portfolios that have the highest return for any given standard deviation.
It is that part of the portfolio frontier that begins with the minimum variance
portfolio and includes all those on the portfolio frontier with return greater
than or equal to that of the minimum variance portfolio. These features are
illustrated in Figure 4.12.
As well as those portfolios on the frontier, there are also portfolios with
return and standard deviation combinations inside the frontier. In total, the
81
4.5. EXTENSION TO MANY ASSETS
rp
Efficient
Frontier
Minimum
Variance
Portfolio
Figure 4.12: The Portfolio Frontier as an Envelope
82
rp
Figure 4.13: The Portfolio Set

portfolio frontier and the portfolios located in the interior are called the portfolio
set. This set is shown in Figure 4.13.
In general, the portfolio frontier is found by minimizing the standard deviation (or the variance) for a given level of return. This is analyzed in detail in
the Appendix.
4.6
Risk-free Asset
The previous sections have considered only risky assets. A risk-free asset is now
introduced and it is shown that this has a significant effect upon the structure
of the efficient frontier.
The interpretation of the risk-free asset is important for understanding the
implications of the following analysis. It is usual to assume that the risk-free
asset is a treasury bill issued, for instance, by the US or UK government. Investment, or going long, in the risk-free asset is then a purchase of treasury bills.
The government issues treasury bills in order to borrow money, so purchasing a
treasury bills is equivalent to making a short-term loan to the government. Conversely, going short in the risk-free asset means that the investor is undertaking
borrowing to invest in risky assets. Given this interpretation of the risk-free
asset as lending or borrowing, we can think of its return as being an interest
rate.
83
4.6. RISK-FREE ASSET
With these interpretations, the assumption that the consumer can go long
or short in a risk-free asset at a single rate of return means that the interest rate
for lending is the same as that for borrowing. This is a very strong assumption
that is typically at variance with the observation that the rate of interest for
borrowing is greater than that for lending. We accept the assumption of the
single rate in this section and relax it in the next.
An idea that we have already employed is that a portfolio of risky assets
can be treated as if it were a single (compound) risky asset with a return and
a variance. This holds as long as the proportions of the assets in the portfolio
remain constant. Then combining such a portfolio with the risk-free asset is like
forming a portfolio of two assets. Using this approach, it is possible to discuss
the effect of combining portfolios of risky assets with the risk-free asset without
needing to specify in detail the composition of the portfolio of risky assets.
Consider a given portfolio of risky assets. Denote the return on this portfolio
by rp and its variance by 2p . Now consider combining this portfolio with the
risk-free asset. Denote the return on the risk-free asset by rf . Let the proportion
of investment in the risky portfolio be X and the proportion in the risk-free asset
be 1 X.
This gives an expected return on the combined portfolio of
rP = [1 X] rf + X rp ,
and a standard deviation of

1/2
2
P = [1 X] 2f + X 2 2p + 2X [1 X] p f pf
.
(4.17)
(4.18)
By definition the variance of the risk-free asset is zero, so 2f = 0 and pf = 0.

The standard deviation of the portfolio then reduces to
P = Xp .
Rearranging this expression
X=
P
.
p
(4.19)
(4.20)
Substituting into (4.17), the return on the portfolio can be expressed as

P
P
rP = 1
rf +
rp ,
(4.21)
p
p
which can be solved for rP to give
rP = rf +

rp rf
P .
p
(4.22)
What the result in (4.22) shows is that when a risk-free asset is combined
with a portfolio of risky assets it is possible to trade risk for return along a
r r
straight line that has intercept rf and gradient pp f . In terms of the risk/return
84
rP
rp
rf
Figure 4.14: Introducing a Risk-Free Asset
diagram, this line passes through the locations of the risk-free asset and the
portfolio of risky assets. This is illustrated in Figure 4.14 where the portfolio p
is combined with the risk-free asset.
Repeating this process for other points on the frontier gives a series of lines,
one for each portfolio of risky assets. These lines have the same intercept on
the vertical axis, but different gradients. This is shown in Figure 4.15 for three
different portfolios 1, 2, and 3.
The final step in the analysis is to find the efficient frontier. Observe in Figure
4.15 that the portfolios on the line through point 3 provide a higher return for
any standard deviation than those through 1 or 2. The set of efficient portfolios
will then lie on the line that provides the highest return for any variance. This
must be the portfolio of risky assets that generates the steepest line. Expressed
r r
differently, the efficient frontier is the line which makes the gradient pp f as
great as possible. Graphically, this line is tangential to the portfolio frontier for
the risky assets. This is shown in Figure 4.16 where portfolio T is the tangency
portfolio.
Consequently when there is a risk-free asset the efficient frontier is linear
and all portfolios on this frontier combine the risk-free asset with alternative
proportions of the tangency portfolio of risky assets. To the left of the tangency
point, the investor holds a combination of the risky portfolio and the risk-free
asset. To the right of the tangency point, the investor is long in the risky
85
4.6. RISK-FREE ASSET
rP
rf
P
Figure 4.15: Different Portfolios of Risky Assets
rp
X p > 1, X f < 0
T
X p < 1, X f > 0
rf
Figure 4.16: Efficient Frontier with a Risk-Free Asset
86
portfolio and short in the risk-free asset. The risky assets are always purchased
in the proportions implied by the structure of the tangency portfolio. The
gradient of the efficient frontier (the slope of the line) is the price of risk in
terms of the extra return that has to be offered to the investor in order for them
to take on additional unit of standard deviation.
Example 61 Assume that two risky assets, A and B, are available and that
their returns are uncorrelated. Letting X denote the proportion of asset A in
the portfolio of risky assets, the tangency portfolio is defined by
max
{X}
X rA + [1 X] rB rf
rp rf
=
1 .
p
2 2 2
2
2
X A + [1 X] B
Differentiating with respect to X, the first-order condition is

rA + [1 X] rB rf ] 2X 2A 2 [1 X] 2B
rA rB
1 [
= 0.

12 2

32
2
2
X 2 2A + [1 X] 2B
X 2 2A + [1 X] 2B
Solving the first-order condition gives

X=
2A [
rB
rA rf ]
2B [
.
rf ] + 2B [
rA rf ]
This analysis can be extended to consider the effect of changes in the rate
of return on the risk-free asset. Assume that there are two risky assets with
asset B having the higher return and standard deviation. Then as the risk-free
return increases, the gradient of the efficient frontier is reduced. Moreover, the
location of the tangency portfolio moves further to the right on the portfolio
frontier. This increases the proportion of asset B in the risky portfolio and
reduces the proportion of asset A. Through this mechanism, the rate of return
on the risk-free asset affects the composition of the portfolio of risky assets.
Example 62 Using the data for African Gold and Walmart stock in Example
58 the proportion of African Gold stock in the tangency portfolio is plotted in
Figure 4.17. This graph is constructed by choosing the proportion of African
r r
Gold stock to maximize the gradient pp f for each value of rf . It can be seen
that as the return on the risk-free asset increases, the proportion of African
Gold, which has the lower return of the two assets, decreases.
4.7
Different Borrowing and Lending Rates
It has already been noted that in practice the interest rate for lending is lower
than the rate for borrowing whereas the construction of the efficient frontier in
the previous section assumed that they were the same. This does not render the
87
4.7. DIFFERENT BORROWING AND LENDING RATES
XA
rf
Figure 4.17: Composition of Tangency Portfolio
88
previous analysis redundant but rather makes it a step towards incorporating

the more general situation.
Before proceeding to the analysis it is worth considering why the interest
rates should be different. Fundamentally, the reason has to be the existence of
some form of market inefficiency. If there were no inefficiency then all investors
would be able to borrow at same rate at which they could lend. The explanation
for such inefficiency can be found in the theories of information and the way in
which they affect market operation. In brief, lenders are imperfectly informed
about the attributes of borrowers. Some borrowers, such as the US and UK
government, have established strong reputations for honoring their debts and
not defaulting. Consequently, they can borrow at the lowest possible rates. In
contrast, private borrowers have limited reputations and lenders will be uncertain about their use of funds and consequent ability to repay. Furthermore, the
borrowers are usually more informed than the lenders. These factors result in
less reputable borrowers having to pay a premium on the interest rate for loans
in order to compensate the lender for the increased risk.
There are two effects of there being different rates of return for lending and
borrowing. Firstly, the efficient set cannot be a single line of tangency. Secondly,
each investor will face an efficient set determined by the rate at which they can
borrow (assuming that the lending rate corresponds to the return on treasury
bills which can be purchased by any investor).
Denote the borrowing rate facing an investor by rb and the lending rate by r .
The discussion above provides the motivation for the assumption that rb > r .
Denote the proportion of the investors portfolio that is in the safe asset by Xf .
If Xf > 0 the investor is long in the safe asset (so is lending) and earns a return
r . If Xf < 0 the investor is short in the safe asset (so is borrowing) and earns
a return rb . It is never rational for the investor to borrow and lend at the same
time.
The structure of the efficient frontier can be developed in three steps. Firstly,
if the investor is going long in the safe asset the highest return they can achieve
for a given standard deviation is found as before: the trade-off is linear and
the tangency portfolio with the highest gradient is found. This gives the line
in Figure 4.18 which is tangent to the portfolio frontier for the risky assets at
point T1 . The difference now is that this line cannot be extended to the right of
T1 : doing so would imply the ability to borrow at rate r which we have ruled
out. Secondly, if the investor borrows the efficient frontier is again a tangent
line; this time with the tangency at T2 . This part of the frontier cannot be
extended to the left of T2 since this would imply the ability to lend at rate rb .
This, too, has been ruled out. Thirdly, between the tangency points T1 and T2 ,
the investor is purchasing only risky assets so is neither borrowing or lending.
These three sections then complete the efficient frontier.
Example 63 If there are just two risky assets, A and B, whose returns are
uncorrelated, the result in Example 61 shows that the proportion of asset A in
89
4.7. DIFFERENT BORROWING AND LENDING RATES
rp
Xf <0
T2
T1
rb
Xf >0
rl
Figure 4.18: Different Returns for Borrowing and Lending

the tangency portfolio T1 is given by
XA1 =
rA r ]
2B [
,
2A [
rB r ] + 2B [
rA r ]
and in the tangency portfolio T2 by

XA2 =
2A [
rB
2B [
rA rb ]
.
rb ] + 2B [
rA rb ]
It can be shown that if rA < rB then XA1 > XA2 so at the second tangency the
proportion of the lower asset with the lower return is smaller.
In summary, when there are differing returns for borrowing and lending the
efficient frontier is composed of two straight sections and one curved section.
Along the first straight section the investor is long in the risk-free asset and
combines this with tangency portfolio T1 . At T1 all investment is placed in the
tangency portfolio. Between T1 and T2 the investor purchases only risky assets
with the portfolio composition changing as the move is made around this section
of the portfolio frontier. Beyond T2 the investor goes short in the risk-free asset
and combines this short position with a purchase of the risky assets described
by the portfolio at T2 .
90
4.8
Conclusions
The chapter has investigated the risk/return relationship as portfolio composition is varied. For portfolios consisting of only risky assets, a portfolio frontier is
obtained whose shape depends on the correlation of asset returns. The minimum
variance portfolio was defined and its role its separating efficient from inefficient
portfolios was identified. From this followed the determination of the efficient
frontier - the set of portfolios with return at least as great as the minimum variance portfolio. A risk-free asset was then introduced and the efficient frontier
was constructed as the tangent to the portfolio set. Finally, the consequence of
having different returns for borrowing and lending was considered.
The central message of this chapter is the fact that an investor is able to
distinguish some portfolios which are efficient from others which are not. It
is from the efficient set that a selection will ultimately be made. The second
important observation is the role of the risk-free asset, and whether lending and
borrowing rates are the same, in determining the structure of the efficient set.
Given this characterization of the efficient set, it is now possible to move to the
issue of portfolio choice.
Exercise 32 The table provides data on the return and standard deviation for
different compositions of a two-asset portfolio. Plot the data to obtain the portfolio frontier. Where is the minimum variance portfolio located?
X
rp
p
0
.08
.5
.1
.076
.44
.2
.072
.38
.3
.068
.33
.4
.064
.29
.5
.060
.26
.6
.056
.24
.7
.052
.25
.8
.048
.27
.9
.044
.30
1
.04
.35
Exercise 33 Assuming that the returns are uncorrelated, plot the portfolio frontier without short sales when the two available assets have expected returns 2 and
5 and variances 9 and 25.
Exercise 34 Using 10 years of data from Yahoo, construct the portfolio frontier
without short selling for Intel and Dell stock.
Exercise 35 Confirm that (4.7) and (4.8) are both solutions for the standard
deviation when AB = 1.
Exercise 36 Given the standard deviations of two assets, what is smallest value
of the correlation coefficient for which the portfolio frontier bends backward?
(Hint: assuming asset A has the lower return, find the gradient of the frontier
at XA = 1.)
Exercise 37 Discuss the consequence of taking into account the fact that the
two stocks in Example 58 are traded in different currencies. Furthermore, what
role may the short data series play in this example?
Exercise 38 Allowing short selling, show that the minimum variance portfolios
for AB = +1 and AB = 1 have a standard deviation of zero. For the case of
a zero correlation coefficient, show that it must have a strictly positive variance.
4.8. CONCLUSIONS
91
Exercise 39 Using the data in Exercise 33, extend the portfolio frontier to
incorporate short selling.
Exercise 40 Calculate the minimum variance portfolio for the data in Example
57. Which asset will never be sold short by an efficient investor?
Exercise 41 Using (4.16), explain how the composition of the minimum variance portfolio changes as the variance of the individual assets is changed and
the covariance between the returns is changed.
Exercise 42 Calculate the minimum variance portfolio for Intel and Disney
stock.
Exercise 43 For a two-asset portfolio, use (4.22) to express the risk and return
in terms of the portfolio proportions. Assuming that the assets have expected
returns of 4 and 7, variances of 9 and 25 and a covariance of 12, graph the
gradient of the risk/return trade-off as a function of the proportion held of the
asset with lower return. Hence identify the tangency portfolio and the efficient
frontier.
Exercise 44 Taking the result in Example 61, show the effect on the tangency
portfolio of (a) an increase in the return on the risk-free asset and (b) an increase
in the riskiness of asset A. Explain your findings.
Exercise 45 What is the outcome if a risk-free asset is combined with (a) two
assets whose returns are perfectly negatively correlated and (b) two assets whose
returns are perfectly positively correlated?
Exercise 46 Prove the assertion in Example 63 that if rA < rB then XA1 >
XA2 .
92
Chapter 5
Portfolio Selection
Choice is everything! But even when we have determined the
available options, it is necessary to know exactly what we want in
order to make the best use of our choices. It is most likely that
we have only a vague notion of what our preferences are and how
we should respond to risk. Dont immediately know this but must
work from a basic feeling to clearer ideas. Consequently, want to
summarize and construct preferences. We end up suggesting how
people should behave. Even though some may not act this way it
would be in their interests to do so.
5.1
Introduction
The process of choice involves two steps. The first step is the identification of
the set of alternatives from which a choice can be made. The second step is
to use preferences to select the best choice. The application of this process to
investments leads to the famous Markovitz model of portfolio selection.
The first step of the process has already been undertaken. The efficient
frontier of Chapter 4 identifies the set of portfolios from which a choice will
be made. Any portfolio not on this frontier is inefficient and should not be
chosen. Confronting the efficient frontier with the investors preferences then
determines which portfolio is chosen. This combination of the efficient frontier
and preferences defined over portfolios on this frontier is the Markovitz model.
This model is at the heart of investment theory.
The study of choice requires the introduction of preferences. The form that
an investors preferences taken when confronted with the inherent risk involved
in portfolio choice is developed from a formalized description of the decision
problem. This study of preferences when the outcome of choice is risky leads
to the expected utility theorem that describes how a rational investor should approach the decision problem. Once preferences have been constructed they can
be combined with the efficient frontier to solve the investors portfolio selection
93
94
CHAPTER 5. PORTFOLIO SELECTION
problem.
There are parts of this chapter that are abstract in nature and may seem
far removed from practical investment decisions. The best way to view these
is as formalizing a method of thinking about preferences and decision making.
Both these a slightly tenuous concepts and difficult to give a concrete form
without proceeding through the abstraction. The result of the analysis is an
understanding of the choice process that fits well with intuitive expectations
of investor behaviour. Indeed, it would be a poor representation of choice if it
did otherwise. But, ultimately, the Markovitz model very neatly clarifies how
an investors attitudes to risk and return affect the composition of the chosen
portfolio.
A reader that is not deeply concerned with formalities can take most of the
chapter on trust and go immediately to Section 5.5. This skips the justification
for how we represent preferences but will show how those preferences determine
choice.
5.2
Expected Utility
When a risky asset is purchased the return it will deliver over the next holding
period is unknown. What is known, or can at least be assessed by an investor,
are the possible values that the return can take and their chances of occurrence.
This observation can be related back to the construction of expected returns
in Chapter 3. The underlying risk was represented by the future states of the
world and the probability assigned to the occurrence of each state. The question
then arises as to what guides portfolio selection when the investment decision
is made in this environment of risk.
The first step that must be taken is to provide a precise description of the
decision problem in order to clarify the relevant issues. The description that we
give reduces the decision problem to its simplest form by stripping it of all but
the bare essentials.
Consider an investor with a given level of initial wealth. The initial wealth
must be invested in a portfolio for a holding period of one unit of time. At
the time the portfolio is chosen the returns on the assets over the next holding
period are not known. The investor identifies the future states of the world, the
return on each asset in each state of the world, and assigns a probability to the
occurrence of each state. At the end of the holding period the returns of the
assets are realized and the portfolio is liquidated. This determines the final level
of wealth. The investor cares only about the success of the investment over the
holding period, as measured by their final level of wealth, and does not look any
further into the future.
This decision problem can be given the following formal statement:
At time 0 an investment plan is chosen;
There are n possible states of the world, i = 1, ..., n, at time 1;
5.2. EXPECTED UTILITY
95
At time 0 the probability of state i occurring at time 1 is i ;

The level of wealth at time 1 in state i with investment plan is Wi ();
At time 1 the state of the world is realized and final wealth determined.
Example 64 An investor allocates their initial wealth between a safe asset and
a risky asset. Each unit of the risky asset costs $10 and each unit of the safe
asset $1. If state 1 occurs the value of the risky asset will be $15. If state 2
occurs the value of the risky asset will be $5. The value of a unit of the safe
asset is $1 in both states. Letting 1 be the number of units of the risky asset
purchased and 2 the number of units of the safe asset, the final wealth levels in
the two states are
W1 () = 1 15 + 2 ,
and
W2 () = 1 5 + 2 .
The example shows how an investor can compute the wealth level in each
state of the world. The investor also assigns a probability to each state, which
becomes translated into a probability for the wealth levels. Hence, it state 1
occurs with probability 1 then wealth level W1 () occurs with probability 1 .
We can safely assume that an investor prefers to have more wealth than less.
But with the risk involved in the portfolio choice problem this is not enough
to guide portfolio choice. It can be seen in the example that every choice of
portfolio leads to an allocation of wealth across the two states. For example, a
portfolio with a high value of 1 relative to 2 gives more wealth in state 1 and
less in state 2 compared to a portfolio with a relatively low value of 1 . The key
step in the argument is to show how the wish for more wealth within a state
translates into a set of preferences over allocations of wealth across states.
The first step is to formalize the assumption that the investor prefers more
wealth to less. This formalization is achieved by assuming that the preferences
of the investor over wealth levels, W , when these are known with certainty, can
be represented by a utility function U = U (W ) and that this utility function
has the property that U (W ) > 0. Hence, the higher is the level of wealth the
higher is utility. We will consider the consequences of additional properties of
the utility function in Section 5.3.
The utility function measures the payoff to the investor of having wealth
W. Such a utility function can be interpreted in three different ways. First,
the investor may actually operate with a utility function. For example, an
investment fund may set very clear objectives that can be summarized in the
form of a utility function. Second, the investor may act as if they were guided
by a utility function. The utility function is then an abbreviated description
of the principles that guide behavior and make them act as if guided by a
utility function. Third, the utility function can be an analysts summary of the
preferences of the investor.
96
Time 0
Time 1
W1 , U (W1 )
W2 , U (W2 )
W3 , U (W3 )
M
n
Wn , U (Wn )
Figure 5.1: The Decision Problem

Example 65 (i) The quadratic utility function is given by U = a + bW cW 2 .
This utility function has the property that U (W ) > 0 only if b 2cW > 0. It
has the desired property only if the wealth level is not too high.
(ii) The logarithmic utility function by U = log W . This utility function
always has U (W ) > 0 provided that W > 0. This utility function is not defined
for negative wealth levels (an investor in debt).
The second step in the analysis is to impose the assumption that the investor
can assess the probability of each state occurring. For state i, this probability is
denoted
n by i . Because they are probabilities, it follows that i 0, i = 1, ..., n,
and i=1 i = 1. This formulation leads to the structure shown in Figure 5.5.
The interpretation of the figure is that the investor is located at time 0 looking
forward to time 1. The branches emanating from time 0 are the alternative
states of the world that may arise at time 1. A probability is assigned to each
state. A choice of a portfolio determines the wealth level in each state. The
wealth levels determine the utilities.
The difficulty facing the investor is that the choice of portfolio must be made
before the state at time 1 is known. In order to analyze such ex ante choice
in this framework a set of preferences must be constructed that incorporate
the risk faced by the investor. To do this it is necessary to determine an ex
ante evaluation of the potential income levels {W1 , ..., Wn } that occur with
probabilities {1 , ..., n } given the ex post preferences U (W ) .
The preferences over wealth levels, represented by the utility function U (W ) ,
can be extended to ex ante preferences over the random wealth levels by assuming that the investor acts consistently in such risky situations. Rationality
means that the investor judges outcomes on the basis of their probabilities and
payoffs, and combines multiple risky events into compound events without any
inconsistencies. If the investor behaves in this way, then there preferences must
satisfy the following theorem.
97
5.2. EXPECTED UTILITY
Theorem 1 If a rational investor has utility of wealth U (W ) , their preferences

over risky outcomes are described by the expected utility function
EU = E[U (Wi )] =
n
i U(Wi ).
(5.1)
i=1
The theorem shows that the random consequences are evaluated by the
mathematical expectation of the utility levels. This theorem has played a very
important role in decision-making in risky situations because of the simplicity
and precision of its conclusion. It provides the link between the evaluation of
wealth when it is known with certainty and the evaluation of uncertain future
wealth levels.
Example 66 Consider an investor whose utility of wealth is represented by the
1
utility function U = W 2 . If there are three possible states of the world, the
expected utility function of the investor is given by
1
EU = 1 W12 + 2 W22 + 3 W32 .

The decision of an investor is to choose a portfolio . The chosen portfolio
determines a wealth level Wi () in each state i. What the expected utility
theorem states is that the investor should choose the portfolio to maximize
expected utility subject to the cost of the portfolio being equal to the initial
wealth they are investing. Let the cost of a portfolio be given by C () , so
the investor faces the constraint be given by W0 = C (). The decision problem
facing the investor is then described by
max E[U (Wi ())] subject to W0 = C () .
{a}
(5.2)
Example 67 Assume an investor with an initial wealth of $1,000 has a logarithmic utility function. Let the probability of state 1 by 23 . Assume that these
is a risky asset that costs $2 to purchase but will be worth $3 if state 1 occurs.
If state 2 occurs the risky asset will be worth $1. Assume that there is also a
risk-free asset that costs $1 and is worth $1 in both states. The decision problem
for the investor is
max
{1 ,2 }
2
1
ln (31 + 2 ) + ln (1 + 2 ) ,
3
3
subject to the budget constraint

1000 = 21 + 2 .
Eliminating 2 between these equation gives
max
{1 }
1
2
ln (1 + 1000) + ln (1000 1 ) .
3
3
98
differentiating with respect to 1 the necessary condition for the maximization

is
1
2
= 0.
3 (1 + 1000) 3 (1000 1 )
Solving the necessary condition gives
1 =
1000
.
3
This is the optimal purchase of the risky asset. The optimal purchase of the safe
asset is
1000
.
2 = 1000 21 =
3
This completes the general analysis of the choice of portfolio when returns
are risky. The expected utility theorem provides the preferences that should
guide the choice of a rational investor. The optimal portfolio then emerges as
the outcome of expected utility maximization. This is a very general theory with
wide applicability that can be developed much further. The following sections
refine the theory to introduce more detail on attitudes to risk and how such
attitudes determine the choice of portfolio.
5.3
Risk Aversion
One fundamental feature of financial markets is that investors require increased

return to compensate for holding increased risk. This point has already featured
prominently in the discussion. The explanation of why this is so can be found
in the concept of risk aversion. This concept is now introduced and its relation
to the utility function is derived.
An investor is described as risk averse if they prefer to avoid risk when there
is no cost to doing so. A precise characterization of the wish to avoid risk can
be introduced by using the idea of an actuarially fair gamble. An actuarially
fair gamble is one with an expected monetary gain of zero. Consider entering
a gamble with two outcomes. The first outcome involves winning an amount
h1 > 0 with probability p and the second outcome involves losing h2 < 0 with
probability 1 p. This gamble is actuarially fair if
ph1 + (1 p)h2 = 0.
(5.3)
Example 68 A gamble involves a probability 14 of winning $120 and a probability 34 of losing $40. The expected payoff of the gamble is
3
1
120 40 = 0.
4
4
If an investor is risk averse they will be either indifferent to or strictly opposed to accepting an actuarially fair gamble. If an investor is strictly risk averse
then they will definitely not accept an actuarially fair gamble. Put another way,
99
5.3. RISK AVERSION
a strictly risk averse investor will never accept a gambles that does not have a
strictly positive expected payoff.
Risk aversion can also be defined in terms of an investors utility function.
Let W0 be the investors initial wealth. The investor is risk averse if the utility
of this level of wealth is higher than the expected utility arising from entering a
fair gamble. Assume that ph1 + (1 p)h2 = 0, so the gamble with probabilities
{p, 1 p} and prizes {h1 , h2 } is fair. An investor with utility function U(W ) is
risk averse if
U (W0 ) pU (W0 + h1 ) + (1 p)U (W0 + h2 ).
(5.4)
The fact the gamble is fair allows the left-hand side of (5.4) to be written as
U (p(W0 + h1 ) + (1 p)(W0 + h2 )) pU (W0 + h1 ) + (1 p)U (W0 + h2 ). (5.5)
The statement in (5.5) is just the requirement that utility function is concave.
Strict risk aversion would imply a strict inequality in these expressions, and a
strictly concave utility function.
A strictly concave function is one for which the gradient of the utility function
falls as wealth increases. The gradient of the utility function, U (W ) , is called
the marginal utility of wealth. As shown in Figure 5.2, strict concavity means
that the marginal utility of wealth falls as wealth increases.
These statements can be summarized by the following:
Risk Aversion U (W ) concave,
(5.6)
Strict Risk Aversion U (W ) strictly concave.
(5.7)
and
Example 69 Consider an investment for which $10 can be gained with probability 12 or lost with probability 12 and an investor with initial wealth of $100. If
the investor has a logarithmic utility function then
ln (100) = 4.6052 >
1
1
ln (100 + 10) + ln (100 10) = 4.6001.
2
2
This inequality shows that the logarithmic function is strictly concave so the
investor is strictly risk averse.
Risk aversion is a useful concept for understanding the an investors choice
of portfolio from the efficient set. The value of the concept makes it worthwhile
to review methods of measuring the degree of an investors risk aversion. There
are two alternative approaches to obtaining a measure. One methods is via
the concept of a risk premium and the other is by defining a coefficient of risk
aversion.
An investors risk premium is defined as the amount that they are willing
to pay to avoid a specified risk. An alternative way to express this is that the
100
Figure 5.2: Strict Risk Aversion

risk premium is the maximum price the investor would pay for an insurance
policy that completely insured the risk. The risk premium is defined relative to
a particular gambles, so will vary for different gambles. But for a given gamble
it can be compared across different investors to judge who will pay the lowest
price to avoid risk.
Consider a gamble with two outcomes h1 > 0 and h2 < 0 which occur with
probabilities p and 1 p. Assume that
ph1 + [1 p] h2 0,
(5.8)
so that a risk-averse investor would prefer not to accept the gamble. The risk
premium is defined as the amount the investor is willing to pay to avoid the risk.
Formally, it is the amount that can be taken from initial wealth to leave the
investor indifferent between the reduced level of wealth for sure and accepting
the risk of the gamble. The risk premium, , satisfies the identity
U (W0 ) = pU (W0 + h1 ) + (1 p)U (W0 + h2 ).
(5.9)
The higher is the value of for a given gamble, the more risk-averse is the
investor. One way to think about this is that measures the maximum price
the investor is willing to pay to purchase an investment policy that ensures the
gamble will be avoided.
The risk premium is illustrated in Figure 5.3. The expected utility of the
gamble is pU (W0 + h1 ) + (1 p)U (W0 + h2 ), and this determines the certain
income level W0 that generates the same utility. From the figure it can be
seen that the more curved is the utility function, the higher is the risk premium
for a given gamble. In contrast, if the utility function were linear the risk
premium would be zero.
101
5.3. RISK AVERSION
Utility
U (W0 + h2 )
U(W0 ) = EU
U (W0 + h1 )
W0 + h1
W0
W0 + h2
Wealth
Figure 5.3: The Risk Premium

The observation that the size of the risk premium is related to the curvature
of the utility function suggests the second way of measuring risk aversion. The
curvature can be measured by employing the second derivative of utility. The
two measures of risk aversion that are defined in this way are:
Absolute Risk Aversion:
RA =
Relative Risk Aversion:
RR =
U
U ;
W U
U .
Absolute and relative risk aversion are equally valid as measures of risk
aversion. A higher value of either measure implies a higher risk premium for any
gamble. The meaning of the two measures can be investigated by considering the
size of a gamble that an investor is willing to take relative to their income level.
For instance, evidence indicates that investors are more willing to take a gamble
of monetary value when their wealth is higher. This behavior is equivalent to
absolute risk aversion being lower for investors with higher incomes. In contrast,
a lower value of relative risk aversion would mean that investors with higher
wealth were more likely to accept a gamble with monetary value equal to a
given proportion of their wealth. There is no evidence to support this behavior.
Example 70 For the negative exponential utility function, U (W ) = ebW ,
absolute risk aversion is constant with RA = b. If an investor with this utility
function and wealth W0 is willing to accept a gamble with probabilities {p, 1 p}
and prizes {h1 , h2 }, the investor will accept the gamble at any wealth level.
B
B
W B1 , W > 0, B >
Example 71 For the power utility function U (W ) = B1
1
0, relative risk aversion is constant with RR = B . If an investor with this
102
utility function and wealth W0 is willing to accept a gamble with probabilities {p, 1 p} and prizes {W0 h1 , W0 h2 }, the investor will accept the gamble
{p, 1 p},{W h1 , W h2 } at any wealth level W .
5.4
Mean-Variance Preferences
The preceding sections have detailed the construction of an expected utility

function that describes preferences over risky wealth levels. The key ingredients
of the analysis are the set of possible wealth levels and the probabilities with
which they may occur. In contrast, we have chosen to describe assets and
portfolios by their returns and risks. As a consequence the preferences we are
using do not sit comfortably with the characterization of portfolios. The purpose
of this section is to describe the resolution of this difference.
A very important specification of expected utility for finance theory is that
in which utility depends only upon the mean return and the variance of the
return on a portfolio. This is important since these two characteristics of the
portfolio are what underlie the concept of the efficient frontier. Preferences that
depend only on the mean and variance of return can be displayed in the same
diagram as the efficient frontier, and can be directly confronted with the set of
efficient portfolios to investigate the selection of portfolio. The conditions under
which expected utility depends on the mean and variance are now derived.
To undertake the investigation it is necessary to employ Taylors Theorem to
approximate a function. For any function the value at x2 can be approximated
by taking the value f (x1 ) at a different point, x1 , and adding the difference
between x1 and x2 multiplied by the derivative of the function at x1 , so
f (x2 ) f (x1 ) + f (x1 ) [x2 x1 ] .
(5.10)
The approximation can be improved further by adding half the second derivative
times the gradient squared. This process is the basis of Taylors Theorem which
states that for any function
1
f (x2 ) = f (x1 ) + f (x1 ) [x2 x1 ] + f (x1 ) [x2 x1 ]2 + R3 ,
2
(5.11)
where R3 is the remainder that needs to be added to make the approximation

exact.
Taylors Theorem can be applied to the utility function to determine the
situations in which only the mean and variance matter. Assume that wealth
random and may
any value in the range [W0 , W1 ]. Let the expected value
take
. For any value of wealth W
in the range [W0 , W1 ] Taylors
of wealth be E W
Theorem, (5.11), can be used to write

E W
+ U E W
= U E W
U W
W
2

1
E W
+ R3 .
(5.12)
W
+ U E W
2
103
5.4. MEAN-VARIANCE PREFERENCES

, is also random. This means
Wealth is random so the utility of wealth, U W
that the expectation of (5.12) can be taken. Two facts simplify

the expectation.
E W
= 0.
First, the expected deviation from the mean must satisfy E W
2

E W
= 2W
Second, by definition E W
. Using these facts the expected
value is
1

+ U E W
=U E W
E U W
(5.13)
2W
+ R3 .
2
It can be seen from (5.13) that there are two sets of conditions under which
only the mean and the variance of the wealth is relevant. These are either that
the remainder, R3 , is exactly zero or else the remainder depends only on the
mean and variance of wealth. In detail, the remainder can be written exactly
as
n

1 (n)
,
(5.14)
W E W
E W
R3 =
U
n!
n=3

. The remainder is comprised of the
where U (n) is the nth derivative of U W
additional terms that would be obtained if the approximation were continued
by adding derivatives of ever higher order.

,
These observations are important because the mean level of wealth, E W
is determined by the mean return on the portfolio held by the investor. This
follows since

= W0 (1 + rp ) .
(5.15)
E W
Similarly, the variance of wealth is determined by the variance of the portfolio.
Observe that
2

E W
2W
= E W
= E [W0 (1 + rp ) W0 (1 + rp )]
= W02 2p .
(5.16)
An expected utility function that depends on the mean and variance of wealth
is therefore dependent on the mean and variance of the return on the portfolio.
The first situation under which only the mean and variance enter expected
utility can be read directly from (5.14).
Condition 1 If the utility function is either linear or quadratic only the mean
and variance matter.
This condition applies because if the utility function is linear or quadratic
then U (n) = 0 for any n 3. The
R3 in (5.14) is then equal to 0
remainder
n
.
whatever the values of W E W
104
If utility is quadratic, expected utility can be written as

= E W
E U W

= E W
b 2
E W
2
2

b
2
E W + W
.
2
(5.17)
The second situation in which only the mean

enter expected
n
variance
and the
utility is obtained by focusing on the terms W E W

in the remainder.
n

E W
is the nth central moment of the disIn statistical language, W
2

E W
, is
tribution of wealth. Using this terminology, the variance, W
the second central moment. It is a property of the normal distribution that
the central moments, for any value of n, are determined by the value of the
mean
of the
distribution

and
the variance.
In short, for the normal distribution
n
n
2
W E W
= f E W , W
so knowing the mean and variance determines all other central moments. Therefore, for any utility function only the
mean and variance matter is wealth is normally distributed.
Condition 2 For all utility function only the mean and variance matter if
wealth is distributed normally.
If either of the conditions applies then the investor will have preferences that
depend only on the mean and variance of wealth. What this means for portfolio
choice is that these are the only two features of the final wealth distribution
that the investor considers. The fact that the mean and variance of final wealth
depend on the mean and variance of the portfolio return allows the preferences to
be translated to depend only on the portfolio characteristics. Therefore, if either
condition 1 or condition 2 applies, the investor has mean-variance preference
that can be written as

(5.18)
U = U rP , 2P ,
where rP is the mean (or expected) portfolio return and 2P is its variance.
5.5
Indifference
The utility function has been introduced as a way of representing the investors
preferences over different wealth levels. Using the arguments of the previous
section this can be reduced to a function that is dependent only upon the mean
and variance of portfolio returns. The implications of mean-variance preferences
are now developed further.
The basic concept of preference is that an investor can make a rational and
consistent choice between different portfolios. An investor with mean-variance
preferences makes the choice solely on the basis of the expected return and
variance. This means when offered any two different portfolios the investor can
5.6. MARKOVITZ MODEL
105
provide a ranking of them using only information on the mean return and the
variance of return. That is, the investor can determine that one of the two
portfolios is strictly preferred to the other or that both are equally good.
The discussion of the reaction of investors to different combinations of return
and risk makes it natural to assume that preferences must satisfy:
Non-Satiation For a constant level of risk, more return is always strictly
preferred;
Risk Aversion A portfolio with higher risk can only be preferable to one
with less risk if it offers a higher return.
Information about preferences can be conveniently summarized in a set of
indifference curves. An indifference curve describes a set of portfolios which
the investor feels are equally good so none of the set is preferred to any other.
An indifference curve can be constructed by picking an initial portfolio. Risk
is then increased slightly and the question asked of how much extra return is
needed to produce a portfolio that is just as good, but no better, than the
original portfolio. Conducting this test for all levels of risk then traces out a
curve of risk and return combinations that is as equally good as, or indifferent
to, the original portfolio. This curve is one indifference curve. Now consider a
portfolio that has a higher return but the same risk as the original portfolio.
From non-satiation, this new portfolio must be strictly better. In this case, it
is said to lie on a higher indifference curve. A portfolio which is worse lies on a
lower indifference curve.
The interpretation of risk aversion in terms of indifference curves is shown in
Figure 5.4. Risk aversion implies that the indifference curves have to be upward
sloping because more return is needed to compensate for risk. If one investor is
more risk averse than another then they will require relatively more additional
return as compensation for taking on an additional unit of risk. This implies
that the indifference curve of the more risk averse investor through any risk and
return combination is steeper than that of the less risk averse investor.
5.6
Markovitz Model
The point has now been reached at which the mean-variance preferences can be
confronted with the efficient frontier. This combination is the Markovitz model
of portfolio choice and is fundamental in portfolio theory. The model permits
portfolio choice to be analyzed and the composition of the chosen portfolio to
be related to risk aversion.
The Markovitz model makes a number of assumptions that have been implicit in the previous description but now need to be made explicit. These
assumptions are:
There are no transaction cost;
All assets are divisible;
106
rp
Less risk
averse
More risk
averse
Figure 5.4: Risk Aversion and Indifference Curves

Short selling is permitted.
The first assumption allows investors to trade costlessly so there is no disincentive to diversify or to change portfolio when new information arrives. The
second assumption permits the investor to obtain an optimal portfolio no matter
how awkward are the portfolio proportions. Some assets, such as government
bonds, are in large denominations and indivisible. The assumption of the model
can be sustained if investors can undertake indirect investments that allow the
purchase of fractions of the indivisible assets. The role of short selling in extending the portfolio frontier was made clear in the previous chapter. The strong
assumption is that short selling can be undertaken without incurring transaction
costs.
5.6.1
No Risk-Free
Portfolio choice is first studied under the assumption that there is no risk-free
asset. In this case the efficient frontier will be a smooth curve.
The optimal portfolio is the one that maximizes the mean-variance preferences given the portfolio frontier. Maximization of utility is equivalent to
choosing the portfolio that lies on the highest possible indifference curve given
the constraint on risk and return combinations imposed by the efficient frontier.
The point on the highest indifference curve will occur at a tangency between
the indifference curve and the portfolio set. Since the investor is risk averse the
indifference curves are upward sloping so the tangency point must be on the
efficient frontier. This means that the portfolio chosen must have a return at
least as great as the minimum variance portfolio.
107
5.6. MARKOVITZ MODEL
rp
p2
p1
Figure 5.5: Choice and Risk Aversion

Figure 5.5 shows choice of two investors with different degrees of risk aversion
when there are just two risky assets available. A and B denote the locations
of the two available risky assets. The more risk averse investor chooses the
portfolio at p1 which combines both risky assets in positive proportions. The
less risk averse investor locates at portfolio p2 . This portfolio involves going
short in asset A. Since no investor chooses a portfolio with a lower return than
the minimum variance portfolio, asset A will never be short-sold. In addition,
any portfolio chosen must have a proportion of asset B at least as great as
the proportion in the minimum variance portfolio. As risk aversion falls, the
proportion of asset B increase and that of asset A falls.
The same logic applies when there are many risky assets. The investor is
faced with the portfolio set and chooses a point on the upward sloping part of
the frontier. The less risk-averse is the investor, the further along the upwardsloping part of the frontier is the chosen portfolio.
5.6.2
Risk-Free Asset
The introduction of a risk-free asset has been shown to have a significant impact
upon efficient frontier. With the risk-free this becomes a straight line tangent
to the portfolio set for the risky assets. The availability of a risk-free asset has
equally strong implications for portfolio choice and leads into a mutual fund
theorem.
The portfolio frontier with a risk-free asset is illustrated in Figure 5.6 with
the tangency portfolio denoted by point T . The more risk-averse of the two
investors illustrated chooses the portfolio p1 . This combines positive proportions
of the risk-free asset and the tangency portfolio. In contrast, the less risk-averse
108
rp
p2
p1
rf
Figure 5.6: Risk-Free Asset and Choice

investor chooses portfolio p2 which involves going short in the risk-free to finance
purchases of the tangency portfolio.
The important point to note is that only one portfolio of risky assets is purchased regardless of the degree of risk aversion. What changes as risk aversion
changes are the relative proportions of this risky portfolio and the risk-free asset in the overall portfolio. Consequently, investors face a simple choice in this
setting. They just calculate the tangency portfolio and then have to determine
the mix of this with the risk-free. To do the latter, an investor just needs to
evaluate their degree of risk aversion.
This observation form the basis of the mutual fund theorem. If there is a riskfree asset, the only risky asset that needs to be made available is a mutual fund
with composition given by that of the tangency portfolio. An investor then only
needs to determine what proportion of wealth should be in this mutual fund.
As a prelude to later analysis, notice that if all investors calculated the
same efficient frontier then all would be buying the same tangency portfolio.
As a result, this would be the only portfolio of risky assets ever observed to be
purchased. There would then be no need for rigorous investment analysis since
observation of other investor would reveal the optimal mix of risky assets. The
assumptions necessary for this to hold and the strong implications that it has
will be discussed in detail in Chapter 8.
5.6.3
Borrowing and Lending
The outcome when borrowing and lending rates are not the same is an extension
of that for a single risk-free rate.
Figure 5.7 shows the outcome for three investors with different degrees of
109
5.7. IMPLICATIONS
rp
p3
p2
rb
p1
T2
T1
rl
Figure 5.7: Different Interest Rates

risk aversion. The most risk averse mixes the risk-free asset with the tangency
portfolio at T1 . The less risk-averse investor purchases risky assets only, with
the choice located at p2 . Finally, the investor with even less risk aversion locates
at p3 which combines the tangency portfolio T2 with borrowing, so the investor
is going short in the risk free asset.
In this case the structure of the portfolio of risky assets held does vary as
the degree of risk aversion changes. But the range of risky portfolios that will
be chosen is bounded by the two endpoints T1 and T2 . Also, the degree of risk
aversion determines whether the investor is borrowing or lending.
5.7
Implications
The analysis of this chapter has several general implications for portfolio choice.
Firstly, there is no simple relationship between the composition of a portfolio
and risk aversion. It is always the case that an increase in risk aversion will
move portfolio choice closer to the minimum variance portfolio. However, even
the minimum variance portfolio may involve short-selling which is usually seen
as a risky activity. This may be surprising since it is not natural to associate
short-selling with what could be extreme risk aversion. Furthermore, risk-averse
investors will generally bear some risk and can even bear considerable risk. The
only implication of risk aversion is that an investor will not bear unnecessary
risk.
To apply these methods the value of risk aversion needs to be determined.
This can be done either precisely or in general terms. It can be done precisely
by using experimental type approaches to test the reaction of the investor to
different risky scenarios. It can be done in general terms just by discussion
110
with the investor about their reaction to risk. Once risk aversion is known,
preferences can be confronted with the efficient frontier to determine choices.
5.8
Conclusions
This chapter has introduced a formalization of the portfolio decision problem

when there is uncertainty. It was shown how to model the randomness of returns
via the introduction of states of nature. This model brought in preferences over
wealth and lead to the expected utility theorem. The concept of risk aversion,
which is a measure of reaction to risk, was then considered. The final step was
to studied when utility could be reduced to mean/variance preferences.
These mean-variance preferences were then confronted with the efficient set
to analyze at portfolio choice. Three different situations were considered and
for each it was traced how the portfolio changed as the degree of risk aversion
changed. An important observation is that when there is a single risk-free rate
the investor will mix the tangency portfolio with the risk-free asset. So all that
is needed is this single risky portfolio which has the form of a mutual fund.
Exercise 47 If there are three possible future wealth levels, which occur with
equal probability, and utility is given by the square root of wealth, what is the
expected utility function?
Exercise 48 Assume there is one risky asset and one safe asset (with a return
of 0) and 2 states of the world (with returns r1 and r2 for the risky asset) which
occur with probabilities p and 1 p. Find the optimal portfolio for an investor
b
b
with the utility function U = Wb .U = Wb .
Exercise 49 Consider an investor with the utility function U = a + bW. Show
1
that they will be indifferent to taking on a fair gamble. Show that if U = a+bW 2
they will not take on the fair gamble, but will if U = a + bW 2 . Calculate the
marginal utility of wealth and the degree of absolute risk aversion for each case.
Comment upon the differences.
Exercise 50 An investor with utility function U = ln W and total wealth of
W = $2 is willing to enter a gamble in which $1 can be won or lost. What must
be the minimum chance of winning for the investor to participate in the gamble?
Exercise 51 The table provides information on portfolio returns and variances,
and the satisfaction derived from several portfolios. Use this information to
graph the indifference curves of the investor. Do they satisfy risk aversion?
What effect does doubling the utility number attached to these curves have?
Portfolio
rp (%)
p (%)
Utility
1
1
2
1
2
2
4
1
3
4
6
1
4
2
2
2
5
4
4
2
6
8
6
2
7
3
2
3
8
6
4
3
9
12
6
3
10
4
2
4
11
8
4
4
12
16
6
4
5.8. CONCLUSIONS
111
Exercise 52 Consider the quadratic utility function U = a + bW cW 2 . Find

the marginal utility of wealth. What happens to this as wealth increases? Does
this utility function provide a good model of preferences?
Exercise 53 Assume there are two risky assets whose returns are uncorrelated.
The expected returns of the assets are 2 and 3, and the standard deviations 5 and
6. There is also a risk-free asset with return
of 1. Find the efficient frontier.

When the utility function is U = 10r 0.25 r2 + 2p , find the optimal portfolio.
112
Part III
Modelling Returns
113
Chapter 6
The Single Index Model

If we want to make progress it is necessary to strip away some of
the details and to focus on issues of core importance. A deft application of Occams razor will simplify the task but retain the essence.
Deeper insights can be provided without losing the essentials.
6.1
Introduction
Using the matrix of variances and covariances for the returns on a set of assets,
the techniques of the previous chapter can be employed to calculate the efficient
frontier. What this simple statement obscures is the quantity of information
that is needed to put this into practice. The methods discussed in this chapter,
and the next chapter, present a method of reducing the information requirement.
The chapter first quantifies the extent of the information requirement by
determining the number of variances and covariances that must be calculated.
A model for reducing the information that is required is the introduced. It is
also described how this can be implemented. The implications of the model are
then explored. Finally, the practical interpretation of the model is discussed.
6.2
Dimensionality
Computing the variance of the return on a portfolio requires the input of information on the variance of return for each of the assets in the portfolio and
the covariance of the returns on each pair of assets. Although the computation
of the variance of the return on the portfolio is straightforward given the variances and covariances, obtaining these imposes considerable demands upon the
investor.
The extent of the information requirement can be appreciated by returning
to formula (3.49) that determines the variance of the return on a portfolio. For
a portfolio composed of N assets, the variance is given by
115
116
CHAPTER 6. THE SINGLE INDEX MODEL
N
N

2p =
Xi Xj ij .
(6.1)
1
1
N + N [N 1] = N [1 + N] .
2
2
(6.2)
i=1 j=1
For each term i = 1, ..., N in the first summation, there are N corresponding
terms in the second. The sum therefore involves a total of N 2 terms, composed
of the variances for each of the N assets and the N [N 1] covariances.
The number of pieces of information required is less than this because the
matrix of variances and covariances is symmetric. Since ij = ji for all i
and j, this implies that there are only 12 N [N 1] independent covariances.
Adding this number of covariances to the number of variances, the total number
of variances and covariances that an investor needs to know to compute the
variance of the return on a portfolio of N assets is
To see the consequences of the formula in (6.2), consider the following example.
Example 72 If a portfolio is composed of the shares of a 100 firms, then
1
1
2 N [1 + N] = 2 [100 101] = 5050.
In assessing the message of this example, it should be noted that a portfolio
with 100 assets is not an especially large portfolio. Many private investors
manage portfolios of this size and financial institutions are very likely to run
portfolios with many more assets than this. In fact, because (6.2) shows that
the number of variances and covariances essentially increases with the square
of the number of assets, the number rapidly becomes very large as the number
of assets in the portfolio increases. The effect that this has can be appreciated
from the next example.
Example 73 If an institution invests in all the stocks in the S+P 500 index,
125250 variances and covariances need to be known to calculate the variance of
the return on the portfolio.
The implications of these observations can be understood by considering how
information on variances and covariances is obtained. There are two standard
sources for the information:
Data on asset returns;
Analysts whose job it is to follow assets.
If data is collected it can be employed to calculate variances and covariances
in the way that was described in Chapter 3. The shortcoming with this approach is the demands that it places upon the data. To accurately estimate
what could be several thousand variances and covariances with any degree of
117
6.3. SINGLE INDEX
accuracy requires very extensive data. This can only work if the data reflect
the current situation regarding the interactions between assets. Unfortunately,
if the necessary quantity of data is obtained by using information on returns
stretching back into the past, then the early observations may not be representative of the current situation. The values calculated will then be poor estimates
of the actual values.
The role of analysts is to follow a range of stocks. They attempt to develop an
understanding of the firms whose stock they follow and the industries in which
the firms operate. Using this knowledge, analysts produce predictions of future
returns for the stocks and an assessment of the risks. Although analysts can
be employed to provide information to evaluate the variances of the returns of
the stocks they follow, it is unlikely that their knowledge can contribute much
to the calculation of covariances. This is partly a consequence of the typical
structure of a brokerage firm which divides analysts into sectoral specialists.
This structure is suited to inform about variances but not covariances since the
links across sectors which are needed to evaluate covariances is missing.
The conclusion from this discussion is that the large numbers of variances
and covariances required to evaluate the variance of a well-diversified portfolio
cannot be computed with any reasonable degree of accuracy. This leads to a
clear problem in implementing the methods of constructing the efficient frontier.
6.3
Single Index
Faced with the kinds of difficulties described above, the natural response is
to find a means of simplifying the problem that retains its essence but loses
some of the unnecessary detail. This is a standard modelling technique in all
sciences. A model is now provided that much reduces the information needed
to calculate the variance of the return on a portfolio and provides the investor
with an appealingly direct way of thinking about the riskiness of assets.
The basis of the model is the specification of a process for generating asset
returns. This process relates the returns on all the assets that are available to a
single underlying variable. This ties together the returns on different assets and
by doing so simplifies the calculation of covariances. The single variable can be
thought of for now as a summary of financial conditions.
Let there be N assets, indexed by i = 1, ..., N . The single index model
assumes that the return any asset i can be written as
ri = iI + iI rI + (iI ,
(6.3)
where ri is the return on asset i and rI is the return on an index. iI and iI

are constants and (iI is a random error term. What this model is saying is that
the returns on all assets can be linearly related to a single common influence and
that this influence is summarized by the return on an index. Furthermore, the
return on the asset is not completely determined by the index so that there is
some residual variation unexplained by the index - the random error. As will be
118
shown, if this process for the generation of returns applies, then the calculation
of portfolio variance is much simplified.
Before proceeding to describe the further assumptions that are made, some
discussion of what is meant by the index will be helpful. The index can be an
aggregate of assets such as a portfolio of stocks for all the firms in an industry
or sector. Frequently the index is taken to be the market as a whole. When
it is, the single index model is usually called the market model and rI is the
return on the market portfolio. As will be shown later the market model has
additional implications (concerning the average value of iI across the assets)
beyond those of the general single-index model. For the moment attention will
focussed on the single-index model in general with the market model analyzed
in Section 6.9.
The single-index model is completed by adding to the specification in (6.3)
three assumptions on the structure of the errors, (iI :
1. The expected error is zero: E [(iI ] = 0, i = 1, ..., N;
2. The error and the return on the index are uncorrelated: E [(iI (rI rI )] =
0, i = 1, ..., N ;
3. The errors are uncorrelated between assets: E [(iI (jI ] = 0, i = 1, ..., N,
j = 1, ..., N, i = j.
The first assumption ensures that there is no general tendency for the model
to over- or under-predict the return on the asset. The second ensures that the
errors random and unexplained by the return on the index The third assumption
requires that there is no other influence that systematically affects the assets.
It is possible in an implementation of the model for some of these assumptions
to be true and others false.
6.4
Estimation
Before proceeding to discuss the value of imposing the single-index model, a

method for estimating the constants iI and iI will be described. This also
provides further insight into the interpretation of these constants.
The standard process for estimating the model is to observe historical data
on the return on asset i and the return on the index I. A linear regression is
then conducted of the return on the asset and the return on the index.
The method of linear regression finds the line which is best fit to the data.
The best fit is defined as the line that minimizes the sum of the errors squared,
where the error is the difference between the observed value and the value predicted by the model. This minimization is undertaken by the choice of the values
of and . Figure 6.1 shows four data points and the associated errors which
are given by the vertical distances from the line. The line is adjusted until the
sum of this is minimized.
119
6.4. ESTIMATION
Figure 6.1: Linear Regression

In period t the observed return on asset i is ri,t and the return on index is
rI,t . Given values for iI and iI , the error in this observation is
ei,t = ri,t (iI + iI rI,t ) .
(6.4)
If data is collected for T periods, iI and iI are chosen to

min
{iI , iI }
T

t=1
[ri,t (iI + iI rI,t )]2 .
(6.5)
The choice of iI produces the first-order condition

2
T
[ri,t (iI + iI rI,t )] = 0,
(6.6)
rI,t [ri,t (iI + iI rI,t )] = 0.
(6.7)
t=1
and the choice of iI gives

2
T

t=1
Solving this pair of conditions, the estimated value of iI is
iI =
T
t=1
[ri,t ri ] [rI,t rI ]
T
t=1
(6.8)
[rI,t rI ]
where the denotes that this is an estimated value. It should be noted that the
can also be written as
formula for
iI
120
iI =
iI
covariance of i and I
=
,
2I
variance of I
(6.9)
so that is is determined by the covariance of the return on the asset with the
return on the index and the variance of the return on the index.
The first-order condition (6.6) can be re-arranged and divided by T to give
iI rI .
iI = ri
(6.10)
rI = 0.
iI
E [(iI ] = ri
iI
(6.11)
Now take the expectation of (6.4) to obtain
This equation shows that the process of linear regression always ensures that
E [(iI ] = 0. The linear regression model therefore satisfies, by construction, the
first assumption of the single-index model.
Returning to the choice of beta, the first-order condition can be written as
T

t=1
Now consider
rI,t [ri,t (iI + iI rI,t )] =
cov (ei,t , rI,t ) (rI,t rI ) =

=
T
rI,t ei,t = 0.
(6.12)
t=1
T
1
(ei,t ei ) (rI,t rI )
T t=1
T
1
ei,t rI,t
T t=1
= 0,
(6.13)
(6.14)
(6.15)
so that linear regression also ensure that the second of the assumptions of the
single-index model is satisfied by construction.
However linear regression cannot ensure that for two assets k and j the third
assumption, E [eiI , ekI ] = 0, is satisfied.
Example 74 The table provides data on the return of an asset and of an index
over a five year period.
ri,t 4 6 5 8 7
rI,t 3 5 4 6 7
Using this data, it can be calculated that ri = 6 and rI = 5. Then
T

t=1
and
[ri,t ri ] [rI,t rI ] = (2)(2) + (0)(0) + (1)(1) + (2)(1) + (1)(2) = 9,

T

t=1
[rI,t rI ] = (2)2 + (0)2 + (1)2 + (1)2 + (2)2 = 10.
121
6.5. SHORTCOMINGS
ri
9
8
7
6
5
4
3
2
1
0
0
rI
Figure 6.2: Regression Line

9
9
and i = 6 10
5 = 32 . The data points and the regression
These give i = 10
line are shown in Figure 6.2. The estimated errors for each period are
ei,t -0.2 0 -0.1 1.1 -0.8
It can be seen that these errors sum to zero and are uncorrelated with rI .
6.5
Shortcomings
Assume now that the investor has collected data on the returns for an index
and for a set of assets. The important point to note is that the single index
model can always be imposed upon those observations. By this it is meant that
the relation (6.3) can be always be used as a model of the process generating
returns. But this does not imply that it will be the correct model: if it is not
correct the assumptions upon the errors will not be satisfied. Even if they are
satisfied, it is not necessarily true that the model is a good one to use. As well
as satisfying the assumption it is also important to consider how much of the
variation in the returns on the assets is explained by the variation in the return
on the index. If it is very little, then the model is providing a poor explanation
of the observations. These two points are now discussed in turn.
The estimation of the single-index model by linear regression guarantees,
by construction, that the expected errors are zero and that the correlation of
error and index return is also zero. Hence, for all possible observations of data,
the first and second assumptions can be made to hold by suitable choice of the
values of iI and iI . However, even if they hold this does not guarantee that
the errors are small or that much of the variation in the return is explained.
This points are illustrated in the following example.
Example 75 The data on the returns on asset i and on the returns on two
122
12
10
8
6
4
2
0
0
0.5
1.5
2.5
3.5
4.5
Figure 6.3:
indices I1 and I2 are given in the table.
ri rI1 rI2
5
1
4
6
3
3.75
10 4
4
Using data on index I1 produces the single-index model
ri = 3 + 1.5rI1 + (iI1 ,
which is graphed in Figure 6.3. The errors from this relationship for the three
observations are 12 , 1 12 , 1, so their mean is 0. It can also be calculated that
they are uncorrelated with the index. The index for this model explains 75%
of the variation in the return on the asset. Using the data on index I2 the
single-index model is
ri = 16.5 + 6rI2 + (iI2 ,
which is graphed in Figure 6.4. The errors from this relationship are 2 12 , 0,
2 12 , so their mean is 0 and they are uncorrelated with the index. The index
for this model explains 10% of the variation in the return on the asset.Both of
these indices produce single index models which satisfy the assumptions on the
correlation of error terms but index I1 provides a much more informative model
than index I2 .
In contrast, the third assumption that there is no correlation in the errors
across assets need not hold for observed data data. This is just a reflection of
the fact that the single index model is an assumption about how returns are
generated and need not necessarily be true. Its failure to hold is evidence that
there are other factors beyond the index that are causing asset returns to vary.
In such a case the model will need to be extended to incorporate these additional
correlating factors. Such extensions are the subject matter of Chapter 7.
123
6.5. SHORTCOMINGS
12
10
8
6
4
2
0
3.7
3.75
3.8
3.85
3.9
3.95
4.05
Figure 6.4:
Example 76 Assume that the true model generating the observed returns on
two assets is
r1
r2
= 2 + 2rI1 + rI2 + 1 ,
= 3 + 3rI1 + 2rI2 + 2,
where I1 and I2 are the two indices that jointly determine the asset returns.
Over three periods of observation the returns and errors are
rI1 rI2 1
r1 2, r2
1
6
0
10 1
19
1
2
4
1
11 2 16 12
3
1
1 8
12 14 12
These values satisfy the requirement that E [i ] = 0 and E [1 2 ] = 0, so the
true errors are uncorrelated. If a single index model is imposed upon this data
using I1 as the index, the result would be the estimates
r1
r1
2
= 11 rI1 + e1 ,
3
1
1
= 21 2 rI1 + e2 ,
6
4
so the estimated errors are

e1 23 1 13
23
1
1
1
e2 12
6 12
These estimated errors satisfy E [ei ] = 0 and E [ei (rI1 rI1 )] = 0 but
E [e1 e2 ] = 19 . The non-zero covariance of the errors is the result of imposing an incorrect model. The second index has a role to play in generating the
observed returns and this is captured in the correlation.
124
6.6
Asset Return and Variance
The single-index model was introduced as a method for reducing the information
required to calculate the variance of a portfolio. It is now shown how this is
achieved.
The first step is to determine the implications of the return generation
process for an individual asset. Since the single-index model assumes
ri = iI + iI rI + (iI ,
(6.16)
then taking the expectation gives

ri = iI + iI rI .
(6.17)
Hence the expected return on the asset is determined by the expected return
on the index.
Example 77 If the expected return on an index is rI = 5, an asset described
by iI = 2 and iI = 1.2 has expected return ri = 2 + 1.2 5 = 8.
The variance of the return on the asset is defined by 2i = E [ri ri ]2 . Using
the single-index model
2i
= E [iI + iI rI + (iI iI iI rI ]
= E [ iI [rI rI ] + (iI ]2

2
= E 2iI [rI rI ] + 2(iI iI [rI rI ] + (2iI .
(6.18)
The next step is to use the linearity of the expectations operator to write

2
(6.19)
2i = 2iI E [rI rI ] + 2 iI E [(iI [rI rI ]] + E (2iI .
Using
2 of the single-index model and noting that E [rI rI ]2 = 2I ,
2 assumption
2
E (iI = i , the variance can be simplified to
2i = 2iI 2I + 2i .
(6.20)
From (6.20) it can be seen that the variance of the return on the asset is
composed of two parts:
Market (or systematic or syncratic) risk, 2iI 2I ;
Unique (or unsystematic or idiosyncratic) risk, 2i .
The market risk is the risk that can be predicted through knowledge of the
variance of the market index. The unique risk of the asset is related to the
asset-specific random variation.
Example 78 If 2I = 16, then for an asset with iI = 0.8 and 2i = 2, 2i =
0.82 16 + 2 = 12.24.
125
6.7. PORTFOLIOS RETURN AND VARIANCE
It should be noted that a low beta does not necessarily imply a low risk
because of the idiosyncratic error. A low beta asset has low systematic risk but
this is only one component of total risk.
Example 79 Assume 2I = 9. Then for asset A with AI = 0.9 and 2A = 8,
2A = 0.92 9 + 8 = 15.29. Similarly, for asset B with AI = 1.05 and 2B = 2,
2B = 1.052 9 + 2 = 11.923. Asset B has a lower total variance despite having
a higher value of .
6.7
Portfolios Return and Variance
Similar calculations can be used to derive the expected return and variance on
a portfolio.
Consider a portfolio of N assets with portfolio weights X1 , ..., Xn . Then the
expected return on the portfolio is given by
rp
N
Xi ri
i=1
N
Xi [iI + iI rI + (iI ]
i=1
N
Xi iI +
i=1
N
Xi iI rI +
i=1
= pI + pI rI + (pI .
N
Xi (iI
i=1
(6.21)
Hence taking the expectation

rp
where pI =
N
i=1

= E pI + pI rI + (pI
= pI + pI rI ,
Xi iI and pI =
N
i=1
(6.22)
Xi iI .
Example 80 Consider a portfolio comprised of two assets A and B with AI =

2, AI = 0.8 and BI = 3, BI = 1.2. Then
rp = (3XA + 2XB ) + (0.8XA + 1.2XB ) rI .
If XA = XB =
1
2
and rI = 5, then rp =
15
2 .
But the important calculation is how the use of the single-index model simplifies the calculation of the variance. The variance is defined
2p
= E [rp rp ]2
2
= E pI + pI rI + (pI pI pI rI

= E 2pI [rI rI ]2 + 2 pI [rI rI ] (pI + (2pI
= 2pI 2I + 2p ,
(6.23)
126
or, using the earlier definitions,

2

N
N

2
2
2 2
p =
Xi iI I +
Xi i .
i=1
(6.24)
i=1
In agreement with the definitions for an individual asset, the first term of this
expression is the systematic variance and the second term the non-systematic
variance.
Example 81 Let AI = 0.75, BI = 1.5, 2I = 25, 2A = 2, 2B = 4, then
2

2
+ 4XB
.
2p = [0.75XA + 1.5XB ]2 25 + 2XA
When XA = 13 , then
2

2
2
1
1
22
2p = 0.75 + 1.5
= 41.063.
25 + 2 + 4
3
3
3
3
It is from this expression for the portfolio variance that the effect of the
restricted return generation process can be seen. To calculate this it is only
necessary to know iI , 2I and 2i , for assets i = 1, ..., N . Hence only 2N + 1
pieces of information are needed rather than the N (N 1) for the unrestricted
model.
Example 82 Returning to the portfolio of FT 100 shares, it is only necessary to
know 101 variances and 100 betas, a significant reduction from the 5050 without
the restricted returns process.
This simple observation shows the value of the single-index model in reducing
the information required to calculate the variance of the return on a portfolio.
6.8
Diversified portfolio
Further insight into the implications of applying the single-index model can be
obtained by considering the variance of a well-diversified portfolio. We know
already that in the general case the variance reduces to the mean covariance. It
is interesting to see the analogue in this case.
Consider a large portfolio that is evenly held, so Xi = N1 for each of the
N assets. Using the single-index model, the portfolio variance is the sum of
systematic and non-systematic risk. Consider first the non-systematic risk on
the portfolio. This is given by
N

2
2 2
Xi i
p =
i=1
N 2
1
=
2i .
N
i=1
(6.25)
127
6.9. MARKET MODEL
In can be seen directly that 2p tends to 0 as N tends to infinity. Consequently,

for a diversified portfolio the non-systematic risk can be diversified away.
The market risk is given by
2pI 2I

N
Xi iI
i=1
N

1
N iI
i=1
2
2
= I 2I ,
2I
2I
(6.26)
where I is the mean value of iI .

Putting these observations together,
2p = 2pI 2I + 2p ,
(6.27)
tends to
2
2p = I 2I ,
(6.28)
as N tends to infinity. For a well-diversified portfolio, only the systematic risk

remains. This can be interpreted as the basic risk that underlies the variation
of all assets. From this perspective, 2I can be termed undiversifiable market
risk and 2p diversifiable risk.
6.9
Market Model
The discussion to this point has been phrased in terms of a general index. The
most important special case is when the index is the return on the entire set of
assets that can be traded on the market. The single-index model then becomes
the market model.
To denote this special case, the expected return on the index is denoted rM ,
the variance of this return by 2M and the beta of asset i by iM .
With the market model
T
(ri,t ri ) (rM,t rM )
iM = t=1T
.
(6.29)
2
t=1 (rM,t rM )
The average value of beta across the assets, with the average taken using the
weights of each asset in the market portfolio, is
M =
N

i=1
Xi iM .
(6.30)
128
This value can be obtained from using linear regression of the market return on
itself, so
T
(rM,t rM ) (rM,t rM )
M = t=1 T
= 1.
(6.31)
2
t=1 (rM,t r M )
Therefore in the market model the weighted-average value of iM , with the

weights given by the shares in the market portfolio, is equal to 1. This is one of
the special features of the market model.
The second feature follows from noting that
rM
N
Xi ri
i=1
N
Xi (iM + iM rM )
i=1
N
Xi iM + rM ,
(6.32)
i=1

so N
i=1 Xi iM = 0. Hence the weighted-average value of iM is 0 for the market
model.
Furthermore, since the value of beta on the market portfolio is M = 1, an
asset that has iM < 1, has lower systematic risk than the market. If iM > 1,
then it has more systematic risk than the market.
Example 83 There are two risky assets available and two potential future states
of the world. Both states are equally likely. There are 100 units of asset A and
it has an initial price, pA (0) , of 10. There are 200 units of asset B and it has
an initial price, pB (0) , of 15. The final price of the assets in the two states of
the world are given in the table.
State 1 State 2
pA (1) 12
11
pB (1) 20
16
Given this data, it follows that rA = 0.15 and rB = 0.2. The proportions of
the two assets in the market portfolio are XA = 0.25 and XB = 0.75. Hence the
return on the market in state 1 is 0.3 and in state 2 is 0.075 and rM = 0.1875.
Using this data it is possible to calculate AM and BM using the population
covariance and variance to evaluate (6.29). Doing this gives
AM =
1
2
(0.2 0.15) (0.3 0.1875) +

1
2
(0.3 0.1875) +
1
2
1
2
(0.1 0.15) (0.075 0.1875)

(0.075 0.1875)2
= 0.444,
and
BM =
1
2
(0.333 0.2) (0.3 0.1875) +

1
2
(0.3 0.1875) +
1
2
1
2
(0.067 0.2) (0.075 0.1875)

(0.075 0.1875)2
= 1.185
6.10. BETA AND RISK
129
Using these values

M = 0.25 0.444 + 0.75 1.185 = 1.
The final step is to compute AM and BM . Using the fact that ri = iM +
iM rM ,
AM = 0.15 0.444 0.1875 = 0.0667,
BM = 0.2 1.185 0.1875 = 0.0222.
Hence XA AM + XB BM = 0.25 0.0667 + 0.75 (0.0222) = 0.
6.10
Beta and Risk
The beta of an asset plays a very important role in the practical application of
investment analysis techniques. The next sections consider it in some detail and
develop a practical interpretation of the theory.
Beta is seen as a measure of the systematic riskiness of an asset. This is
clear from (6.20) in which beta can be seen to act as a multiplying factor on the
variance of the index. It is also evident that this is not a complete description
of risk since the non-systematic risk has also to be taken into account. These
statements are also clearly true of a portfolio and the portfolio beta. Even so,
this perspective on beta is still helpful.
The observation that beta is related to risk leads to the following interpretations which are given for market model (they can be written equally for the
general single-index model):
If iM > 1 then the asset is more volatile (or risky) than the market. In
this case it is termed aggressive. An increase (or decrease) in the return
on the market is magnified in the increase (or decrease) in the return on
the asset.
If iM < 1 then the asset is less volatile than the market. In this case it is
termed defensive. An increase (or decrease) in the return on the market
is diminished in the increase (or decrease) in the return on the asset.
With these definitions, it is also possible to think in terms of the construction
of a defensive portfolio of low beta assets or an aggressive portfolio of high
beta. Although these are useful descriptions, it should not be forgotten that
the total risk must also include the idiosyncratic risk. Only in a well-diversified
portfolio can latter be set aside. In a small portfolio it can even dominate.
6.11
Adjusting Beta
It has already been noted that beta can be calculated by obtaining historical
data on the returns on an asset and on the index. A linear regression is then
130
ri
iM > 1
rM
Figure 6.5: Aggressive Asset
ri
iM < 1
rM
Figure 6.6: Defensive Asset
131
6.11. ADJUSTING BETA
conducted of asset returns on index returns. The intercept obtained is iI and

the slope coefficient iI . There are also several sources for ready-estimated
values of beta. Such publications generally provide information on estimates of
beta and the non-systematic errors.
The method of computing beta raises some questions about the accuracy
of the values obtained and suggests that it may be necessary to adjust the
estimated value.
Because an estimate is based on historical data it is implicitly made by
assuming that beta is constant over time. If beta changes, the estimates will be
imprecise. This suggests there may be value in adjusting an estimated beta to
give more emphasis to more recent data.
There is also a second issue involved in the estimation. If the index used is
the market return, then the average value of beta must equal 1. This follows
since the average value is the beta of the market and the market has a beta value
of 1. Now recall that the estimated beta should be viewed is a random variable
which is expected (if unbiased) to be equal to the true value. The randomness is
consequence of it being based on the observed data set which itself is a random
draw from the set of possible data sets.
Therefore if the estimated beta deviates from the expected value of 1 there
can be two reasons for this. Firstly, the true beta may be different to 1 or,
secondly, there is a random error in the estimation. The further the value is
from 1 the more likely it is that there is a large random error in the estimation.
This suggests that betas that deviate far from 1 may involve large random errors.
This isolates two reasons for considering adjusting estimated betas. Firstly,
the value of beta for the stock may change during the course of the data period. Secondly, there is a statistical tendency for large deviations from 1 to be
associated with large random errors. The following sections consider methods
of adjustment that can be used to correct estimated values of beta.
The two methods of adjustment that are now discussed are purely statistical
methods. They employ mechanical procedures to make the necessary adjustments to beta.
The analysis of Blume involved estimating beta for a set of stocks for two
sample periods, with one period pre-dating the other. The second set of estimates were then regressed on the first set in order to find the average relationship
between the betas estimated for the two periods. This process is intended to
capture the tendency for mean-reversion in the estimates.
Letting i1 denote the value of beta for stock i in the period 1948-1954 and
i2 the value for 1955-1961, the relationship between the two was found to be
i2 = 0.343 + 0.677 i1 .
(6.33)
This result shows clearly the mean-reverting tendency of beta. It also suggests a case for correcting downwards any observed value of beta greater than
1 and adjusting upwards any less than 1.
The correction suggested by Blume is a linear one. It does not put any
special emphasis on the sampling error (the extent of deviation from 1) of the
observed beta. The Vasilek method is an attempt to do this.
132
Let 2 denote the variance of the distribution of the historical estimates

1
of beta over a sample of stock and 2i 1 be the square of the standard error of
the estimate of beta for security i measured in time period 1. Vasilek suggested
that an estimate of i2 should be obtained as a weighted average of i1 and 1 ,
where 1 is the mean estimate of beta in period 1. The weighting suggested was
i2 =
2 + 2i 1
1 +
2 + 2 i 1
i1 .
(6.34)
This weighting procedure adjusts observations with large standard errors

further towards the mean than it adjusts observations with small standard errors. It also ensures that the more uncertain is an estimate, the less weight is
placed upon it.
6.12
Fundamental Beta
The previous section looked at mechanical methods of adjusting beta. In contrast, fundamental betas regard beta as a measure of risk that can be related to
the firm-level variables. The basic view is that small, new and indebted firms
are more risky.
Particular variables that can be considered are:
Dividend payout. Often measured by dividends divided by earnings. Since
management is more reluctant to cut dividends than to raise them, a high
dividend payout is indicative of confidence on the part of management concerning the level of future earnings. Also, dividend payments are less risky
than capital gains. Hence, the company that pays more of its earnings as
dividends is less risky.
Asset growth. Often measured by annual change in total assets. Growth
is usually though of as positively associated with beta. High-growth firms
are thought of as more risky than low-growth firms.
Leverage. Often measured as senior securities divided by total assets.
Leverage tends to increase the volatility of the earnings stream and hence
increases risk and beta.
Liquidity. Often measured as senior securities divided by current liabilities. A firm with high liquidity is thought to be less risky than one with
low liquidity and hence liquidity should be negatively related to beta.
Asset size. Measured by total assets. Large firms are often thought to
be less risky than small firms, if for no other reason than that they have
better access to the capital markets. Hence they should have lower betas.
Earning variability. Measured as thee standard deviation of the earning/price ratio. The more variable a companys earning stream and the
133
6.13. CONCLUSION
more highly correlated it is with the market, the higher its beta should
be.
Given these factors for each firm, the role of the analyst is to subjectively
judge how they can be compounded into a value of beta. A standard process
would be to start with an estimated beta and then adjust it if it appears to be
far out of line on any of these fundamental factors.
6.13
Conclusion
The calculation of the variance of the return for even a medium-sized portfolio can be informationally demanding. The single-index model is a means of
reducing the information required. It assumes that a single variable is responsible for generating the returns on all assets. The most important implication
of this assumption is that it greatly simplifies the calculation of the variance
of the return on a portfolio. Furthermore, it follows that the variance can be
decomposed into systematic and non-systematic componentes.
The beta values generated by the single-index model can also be used to
categorise assets as aggressive or defensive and provide a simple way of thinking
about portfolio construction. Since the betas are estimates, justifications were
given for adjusting the estimated value. This lead into a discussion of adjustment
methods and fundamental betas.
Exercise 54 You manage a portfolio of 50 assets and wish to calculate the
efficient frontier. If you decide that a sample of 30 observations is required to
calculate each variance and covariance, how many data points do you need in
total?
Exercise 55 One response to the data requirements may be to group stocks into
industries and assume that all firms in an industry have the same covariance
with all firms from another industry. A variance can then be calculated for each
stock and a single covariance. By considering the Ford, General Motors and
Dell stock, assess the success of this approach.
Exercise 56 Given the following
estimate the value of and .
Period 1
2 3 4 5
Asset
12 8 5 9 7
Index
8
7 9 8 12
observed returns on an asset and an index,

6
15
16
7
16
15
8
4
7
9
3
6
10
9
8
Exercise 57 The following table provides data on the returns on two assets and
an index. Assess whether the single-index model is appropriate for these assets.
Period
1 2 3 4 5 6 7 8 9 10
Asset 1 6 3 6 8 4 4 2 9 4 5
Asset 2 7 8 4 3 6 8 9 4 8 1
Index
2 4 3 9 5 2 8 4 7 1
134
Exercise 58 Assume returns are generated by a model where the market is the
single factor. The details of the model for three stocks are:
Stock Alpha Beta iI Portfolio Proportion
A
.1
1.1
4
0.6
B
-.2
0.9
3
0.2
C
.05
0.8
5
0.2
The expected return on the market is 12% with a standard deviation of 18%.
(i) What is the portfolios expected rate of return?
(ii) What is the standard deviation of the return on the portfolio?
Exercise 59 Calculate beta for IBM stock using the return on the Standard
and Poor 500 over the last 10 years as the index. (Simplify the calculation by
ignoring dividends paid on the index).
Exercise 60 Assume there are two stocks, A and B, with A = 1.4 and B =
0.8.
(i) If the mean return on the market portfolio is 10% and the risk-free rate
of return is 5%, calculate the mean return of the portfolios consisting of:
a. 75% of stock A and 25% of stock B,
b. 50% of stock A and 50% of stock B,
c. 25% of stock A and 75% of stock B.
(ii) If the idiosyncratic variations of the stocks are A = 4, B = 2 and
the variance of the market portfolio is 2M = 12, calculate the variance of the
portfolios in (a), (b), (c).
(iii) What are the mean return and variance of the portfolios in (ii) if they
are 50% financed by borrowing?
Exercise 61 Assume that two assets constitute the entire market. The possible
returns in the three future states of the world, which occur with equal probability,
and the initial market proportions are given in the table.
Asset Proportion State 1 State 2 State 3
A
0.4
3
2
5
B
0.6
4
4
6
(i) Determine and for both assets.
(ii) Determine the idiosyncratic errors.
(iii) Plot the portfolio frontier.
Exercise 62 If an investors risk aversion increases, can the average beta value
of their portfolio rise?
Chapter 7
Factor Models
7.1
Introduction
In a factor model, the return on a security is modelled as being determined

by one or more underlying factors. The single-index, or market model of the
previous chapter is an example of a single-factor model. In fact, the terminology
factor and index are used interchangeably.
There is no reason to use only a single factor. For instance, firms in the same
industry may have returns that rise and fall together due to some correlating
factor unique to that industry. If this is the case, the assumption of the single
factor model, that the random errors for any two firms are uncorrelated, is not
valid.
In general, additional factors may improve the statistical properties of the
model and will reduce the unexplained error. Two issues are explored here.
First, the returns and variance of a portfolio are derived for models with multiple
factors. Second, the set of relevant factors is considered.
7.2
Single-Factor Model
Repeating the definition of the previous chapter, but with a new notation for
factors models in general, the returns process for the single-factor model is
ri = ai + bi f + ei
(7.1)
where f is the single factor.

Repeating the derivations for the market model gives an expected return for
asset i
ri = ai + bi f,
(7.2)
2i = b2i 2f + 2ei ,
(7.3)
and variance
135
136
CHAPTER 7. FACTOR MODELS
where
bi =
if
.
2f
(7.4)
The covariance between two assets i and j is ij = bi bj 2f .

For a portfolio the return is
rp = ap + bp f + ep ,
(7.5)
and the variance

2p = b2p 2f + 2ep ,
n
n
n
where ap = i=1 wi ai , bp = i=1 wi bi and ep = i=1 wi ei .
7.3
(7.6)
Two Factors
The extension to many factors is now considered, beginning with the case of
two factors.
If it is assumed that the returns on asset i are determined by two factors
and a random error, the return process becomes
ri = ai + b1i f1 + b2i f2 + ei ,
(7.7)
where f1 and f2 are the values of factors 1 and 2. It is assumed that

cov (ei , fk ) = 0, k = 1, 2, all i,
(7.8)
cov (es , ej ) = 0, all i, j.
(7.9)
and
With this returns process the expected return on asset i becomes

ri = ai + b1i f 1 + b2i f 2 ,
(7.10)
and the variance of the return

var (ri ) = E (ri ri )2

2
= E (ai + b1i f1 + b2i f2 + ei ri )

2
= E b1i f1 f 1 + b2i f2 f 2 + ei
=
2
b2ki E
k=1

2

+ 2b1i b2i E f1 f 1 f2 f 2
fk f k
+E [ei ]
= b21i 2f1 + b22i 2f2 + 2b1i b2i f1 f2 + 2ei .
(7.11)
137
7.4. UNCORRELATED FACTORS

For two assets, i and j the covariance is
cov (ri , rj ) = E [(ri ri ) (rj rj )]

2
2

bki fk + ei ri
bkj fk + ej rj
aj +
= E
ai +
= E

k=1
2
k=1
2
k=1

bki fk f k + ei

2
bki bkj E fk f k
+ [b1i b2j + b2i b1j ] E

k=1
2
k=1

bkj fk f k + ej

f1 f 1 f2 f 2
= b1i b1j 2f1 + b2i b2j 2f2 + [b1i b2j + b2i b1j ] f1 f2 .

(7.12)
The bs can be calculated by a multiple regression of the return on asset i on

the values of the factors. This process guarantees that cov (ei , fk ) = 0, k = 1, 2,
all i, and cov (es , ej ) = 0, all i, j.
It can also be noted that
cov (ri , f1 ) = b1i 2f1 + b2i f1 f2 ,
(7.13)
cov (ri , f2 ) = b1i f1 f2 + b2i 2f2 .
(7.14)
and
The values of b1i and b2i can then be solved directly from these equations.
7.4
Uncorrelated factors
An important special case arises when the factors are uncorrelated. If they are
then
(7.15)
cov (f1 , f2 ) = 0.
Employing this assumption gives
var (ri ) = b21i 2f1 + b22i 2f2 + 2ei ,
(7.16)
cov (ri , rj ) = b1i b1j 2f1 + b2i b2j 2f2 .
(7.17)
and
The values of b1i and b2i follow even more immediately when f1 f2 = 0. In
this case
cov (ri , f1 ) = b1i 2f1 ,
(7.18)
138
and
cov (ri , f2 ) = b2i 2f2 ,
(7.19)
so the bs can be found directly. Section 11.6 shows how to construct uncorrelated
factors.
7.5
Many Factors
These calculations can be extended directly to any number of factors.

With n factors, the returns process is
ri = ai +
n
bki fk + ei ,
(7.20)
k=1
where cov (fk , ei ) = 0 and cov (ei , ej ) = 0.

The expected return becomes
ri = ai + +
n
bki f k ,
(7.21)
k=1
and the variance is

var (ri ) =
n
b2ki 2fk +
k=1
n
n
bki bli fk fl + 2ei .
(7.22)
k=1 l=1
For two assets, i and j the covariance is

cov (ri , rj ) =
n
n
bki blj fk fl .
(7.23)
k=1 l=1
7.6
Constructing uncorrelated factors
The calculations in Section 11.4 show the simplification that is achieved when
the factors are uncorrelated. It is always possible to construct uncorrelated
factors.
Consider a model with n factors f1 , ..., fn which are potentially correlated.
The aim is to create factors f1 , ..., fn which are uncorrelated. To do this, take
the first factor, f1 , (it does not matter which this is) and define f1 f1 . Then
conduct the regression
f2 = a + b1 f1 + e.
(7.24)
From this define

f2 = f2 a + b1 f1 = e.
(7.25)
By definition of the least squares estimator, the error, e, must be uncorrelated

with f1 . It captures that part of f2 that is unexplained by f1 .
139
7.7. FACTOR MODELS

To obtain fi then regress
fi = a +
i1

j=1
and define
fi = fi a
bj fj + e,
i1

j=1
(7.26)
bj fj = e.
(7.27)
The factors f1 , ..., fn obtained in this way are uncorrelated as required.
Using these uncorrelated factors, the covariance between two assets i and j
is
ij = bi1 bj1 2f + ... + bin bjn 2f .
(7.28)
1
7.7
Factor models
There are a number of alternative factor models which vary in the motivation
for the choice of factors. Two of the most significant are now discussed.
7.7.1
Industry factors
These models begin with the single-index model and add factors that capture
industry effects.
If the correlation between securities is caused by a market effect and additional industry effects, then the return generating process becomes
ri = ai + bim fm + bi1 f1 + ... + biL fL + ei ,
(7.29)
where fm is the market index and f1 , ..., fL are (uncorrelated) factors relating
to the L industries in which company i operates.
7.7.2
Fundamental factors
A broader range of factors can be introduced. A way of doing this is based

on the efficient market argument that current beliefs about future events are
already incorporated in asset prices, so it is only unexpected changes that can
affect return. Hence the additional factors should capture these unexpected
changes.
An example of an index created on the basis of this reasoning includes as
factors:
Default risk : the unexpected difference in return between 20-year government bonds and 20-year corporate bonds. Measured as the return
on long-term government bonds minus the return on long-term corporate
bonds plus half a per cent.
140
The term structure: the return on long-term government bonds minus the
return on a one-month Treasury bill one month in the future.
Unexpected deflation: the rate of inflation expected at the beginning of
month minus the actual rate of inflation realized at the end of the month.
Growth: unexpected change in the growth rate in real final sales.
Residual market: the difference between the excess return on the S&P
index and the expected excess return.
f1
f2
f3
f4
f5
Default
Term
Deflation
Growth
Market
R2
Sector
Cyclical
Growth
Stable
Oil
Utility
Transportation
Financial
-1.53
0.55
2.84
-1.04
-2.08
0.58
3.16
-0.92
-1.40
0.68
2.31
-0.22*
-0.63*
0.31
2.19*
-0.83*
-1.06
0.72
1.54
0.23*
-2.07
0.58
4.45
-1.13
-2.48
1.00
3.20
-0.56*
* Not significant at 5% level.
1.14
1.28
0.74
1.14
0.62
1.37
0.99
0.77
0.84
0.73
0.50
0.67
0.66
0.72
Exercise 63 Assume that returns of individual securities are generated by the

following two-factor model:
rit = ai + bi f1t + ci f2t + eit .
The following three portfolios are observed:
Expected Return bi
ci
A 25
1.0 1.0
B 10
2.0 0.5
C 20
0.5 1.5
(i) Find the relationship between expected returns and factor sensitivities.
(ii) Suppose you can find a portfolio, D, with expected return = 26, bD = 3.0,
cD = 1.4
(iii) Explain how you could construct a profitable arbitrage portfolio from
securities A, B and C and portfolio D.
Exercise 64 Assume that stock returns are generated by a two-factor model
rit = ai + bi f1t + ci f2t + eit .
Consider the following
Stock bi
ci
A
0.2 1.1
B
0.1 1.0
C
0.3 0.9
portfolio:
ei
0.6
0.5
0.4
7.7. FACTOR MODELS
141
Calculate the variance of an equally-weighted portfolio under the following

alternative assumptions:
(i) f1 , f2 uncorrelated and ei , ej uncorrelated (i = j).
(ii) f1 f2 = 0.5 and ei , ej uncorrelated (i = j).
142
Part IV
Equilibrium Theory
143
Chapter 8
The Capital Asset Pricing

Model
There are demands and supplies. There is a balance of forces that gives an
equilibrium. When balanced the returns have to be in line. Add some
assumptions and generate a clear outcome.
8.1
Introduction
The analysis to this point has considered the expected returns and variances
as given data and used these to determine investment policy. The intellectual
step now is to move to considering the explanation for the observed data. Equilibrium models explain the process of investor choice and market clearing that
lies behind the observed pattern of asset returns. That higher expected return
means higher risk is already clear. An equilibrium model predicts exactly how
much more expected return is required to compensate for additional risk.
The value of an equilibrium model, and of the Capital Asset Pricing Model
(CAPM) in particular, is that it allows the evaluation of portfolio performance.
The model generates an equilibrium relationship between expected return and
risk. If a portfolio delivers a lower level of expected return than predicted by
this relationship for its degree of risk then it is a poor portfolio. The CAPM
model also carries implications in the area of corporate finance. It can be used
as a tool in capital budgeting and project analysis.
The CAPM provides an explanation of asset returns uses the concept of a
financial market equilibrium. A position of equilibrium is reached when the
supply of assets is equal to the demand. This position is achieved by the adjustment of asset prices and hence the returns on assets. This adjustment occurs
through trading behavior. If the expected return on an asset is viewed as high
relative to its risk then demand for the asset will exceed supply. The price of
the asset will rise, and the expected return will fall until equilibrium is achieved.
The particular assumptions about investors preferences and information made
145
146
CHAPTER 8. THE CAPITAL ASSET PRICING MODEL
by a model then determines additional features of the equilibrium.

The CAPM determines very precise equilibrium relationships between the
returns on different assets. The basic assumption of the model is that all investors behave as described in the chapters above. That is, they construct
the efficient set and choose the portfolio that makes the value of their meanvariance expected utility as high as possible. Some additional assumptions are
then added and the implications are then traced.
It is shown that this model leads to especially strong conclusions concerning
the pricing of assets in equilibrium. If the model is correct, these can be very
useful in guiding investment and evaluating investment decisions.
8.2
Assumptions
The set of assumptions upon which the CAPM is based upon are now described.
The interpretation of each assumption is also discussed.
The first set of assumptions describe properties that all assets possess.
All assets are marketable This is the basic idea that all assets can be traded
so that all investors can buy anything that is available. For the vast majority
of assets this an acceptable assumption. How easily an asset can be traded
depends upon the extent to which an organised market exists. There are some
assets cannot be easily traded. An example is human capital. It can be rented
as a labor service but cannot be transferred from one party to another.g
All assets are infinitely divisible The consequence of this assumption is
that it is possible to hold any portfolio no matter what are the portfolio proportions. In practice assets are sold in discrete units. It is possible to move close to
this assumption by buying a fraction of a mutual funds. For instance, treasury
bills may have denominations of $100,000 but a fraction of one cna be bought
if it is shared between several investors.
The second set of assumptions characterize the trading environment.
No transaction costs Transactions costs are the costs of trading. Brokers
charege commission for trade and there is a spread between the buying and
selling prices. The role of the assumption is to allow portfolios to be adjusted
costlessly to continually ensure optimality.
Short sales are allowed The role of short sales has already been described
in the extension of the efficient frontier. They are permitted in actual financial
markets. Where the CAPM diverges from practice is that it is assumed there
are no charges for short selling. In practice margin must be deposited with the
broker which is costly to the investor since it earns less tha the market return.
8.3. EQUILIBRIUM
147
No taxes Taxes affect the returns on assets and tax rules can alter the benefit
of capital gains relative to dividends and coupons. The assumption that there
are no taxes removes this distortion from the system.
The next pair of assumptions imply that the market is perfect.
Lending and borrowing can be undertaken at the risk-less rate Investors face a single rate of interest. This is the assumption of a perfect capital
market. There are no asymmetries of information that prevent lending and
borrowing at a fair rate of interest.
No individual can affect an asset price This is idea of a competitive
market where each trader is too small to affect price. It takes away any market
power and rules out attempts to distort the market.
The next set of assumptions describe the trading behaviour of investors.
All investors have mean/variance preferences This allows us to set the
model in mean variance space and analyse choice through the efficient frontier.
All investors have a one period horizon This simplifies the investment
decision.
Final assumption ties together all the individual investors.
All investors hold same expectations This makes the investors identical
in some sense.
Note that the investors are not assume identical because they can differ in
their risk aversion. Some may be very risk averse some may be less risk averse.
Example 84 Give example of same information and different preferences.
This set of assumptions combines the Markowitz model of portfolio choice
developed in earlier chapters with the assumption that investors have the same
information and reach the same assessment of the expected return and variance
of return for every asset. It is the information and assessment assumptions that
permit the aggregation of individual choices into a market equilibrium with
specific properties.
8.3
Equilibrium
The general properties of equilibrium are now determined by tracing through

the implications of the CAPM assumptions.
The investors all have the same information and expectations. They use this
information to construct the portfolio frontier. Having the same expectations
it follows that the investors perform the same calculations. Hence all investors
construct the same portfolio frontier for risky assets and assess there to be the
148
rp
Figure 8.1: Portfolio frontier

same trade off between expected return and risk. The general form of portfolio
frontier for the risky assets constructed by all investors is shown in Figure 8.1.
Given this portfolio frontier, all the investors must face the same efficient
frontier. The risk-free rate is the same for all and the tangency profile must be
the same.
Given that they face the same efficient frontier, all investors must combine
the risky tangent portfolio M and the risk-free asset. However, the proportions
in which they are combined will differ according to the degree of risk aversion
of each investor. Some may borrow at the risk-free rate while others may lend.
Since all consumers are purchasing portfolio M , this must be the market
portfolio of risky assets. By market portfolio, it is meant a portfolio with the
risky assets in the same proportions as they are found in the market as a whole.
This is the separation principle that states an investor only needs to purchase
two assets. All investors combine the risk free asset and the market portfolio.
What differs between investors is the proportion of these in the portfolio. The
more risk averse is an ivestor the higher will be the proportion of the risk free
asset in the portfolio. Less risk averse investors will hold a larger proportion
of the market portfolio. Those with a low enough level of risk aversion will go
short in the risk free asset to invest in the market portfolio.
The market portfolio is assumed to be well-diversified. The consequence of
this is that non-systematic risk is diversified away by all investors since they
hold the market portfolio.
Finally, if all investors are purchasing the same risky portfolio, there can be
no short selling in equilibrium. If any investor were short-selling a risky asset,
all would be short-selling. This cannot be an equilibrium since the aggregate
demand for the asset would be negative.
149
8.4. CAPITAL MARKET LINE
rp
M
rf
rMVP
MVP
Figure 8.2: Efficient frontier and market portfolio
8.4
Capital Market Line
The capital market line is the name assigned to the efficient frontier in the
CAPM. All efficient portfolios must lie on this line which implies that there is
a linear relationship between risk and return for all portfolios chosen in equilibrium. Any portfolio above the line will be demanded by all investors. Its price
will rise, and so return fall, until in equilibrium it lies on the line. The opposite
applies to any portfolio below the line. Its price will fall and return rise until it
lies on the line.
Since the points (0, rf ) and ( M , rM ) are both on the capital market line, it
r r
gradient can be calculated to be MM f . From this it follows that any portfolio,
p, located on the Capital Market Line must satisfy the equation

rM rf
rp = rf +
(8.1)
p.
M
The interpretation of (8.1) is that rf is the reward for time. This is the
return earned when no risk is involved (p = 0) but consumption is postponed.
Holding the risk-free asset delays consumption for one period and an investor
requires compensating for this. The compensation received is the risk-free rate
of return.
r r
The gradient of the line MM f is the reward for risk or the market price
of risk. To hold risk an investor requires compensation beyond that given by
the risk-free rate. Each unit of standard deviation is rewarded by an extra
rM rf
r r
units of return. The term MM f is the Sharpe ratio which is used later
M
in portfolio evaluation.
Example 85 Assume rf , rM and M . The construct capital market line. Then
150
rp
rM
rf
Borrowing
Lending
Figure 8.3: Capital market line

take an asset i and find the implied ri .
Now is the return higher or lower? The buy or sell. Talk about the equilibrium and disequilibrium.
Example 86 Now talk of trading strategies and evaluation of portfolio performance.
Before proceeding the fact that rp is random must be recalled. The consequence is that in any particular period the realized portfolio return may be
above or below the value predicted by the capital market line. Only in expected
terms are they always upon the line. This is just to stress that randomness
distinguishes rp from rp . Although non-systematic risk may be diversified away
there is still the systematic risk.
8.5
Security Market Line
The CAPM also has implications for the returns on individual assets.
Consider plotting the covariance of an asset with the market against the
assets expected return. Combining M and risk free allows movement along a
line through the two points these assets determine.
The covariance of the risk-free asset with the market is zero and the assets
2
return is rf . The covariance
of the market with the market is M . Hence the
points (0, rf ) and 2M , rM can be linearly combined to determine the Security
Market Line. In equilibrium, all assets must offer return and risk combinations
that lie on this line. If there was an asset (or portfolios) located above this line,
all investors would buy it. Equally, if there was an asset that lay below the line,
151
8.6. CAPM AND SINGLE-INDEX
rp
rf
iM
Figure 8.4: Security Market Line
no investor would hold it. Trade in these assets must ensure that in equilibrium
they will lie on the line.
Using the two identified point, the equation of the Security Market Line is

rM rf
ri = rf +
(8.2)
iM ,
2M
or, defining iM = iM
2 ,
M
rM rf ] iM .
ri = rf + [
(8.3)
Hence there is a linear trade-off between risk measured by iM and return ri .

Example 87 Give some examples. Best to use S+P as the market and do some
asset calculation.
If there are any assets that lie above the line then they are underpriced and
should be purchased. Any below are over priced and should be sold. In this
way the CAPM can be used to identify assets to purchase and sell.
Example 88 Of the buying and selling process leading to equilibrium.
8.6
CAPM and Single-Index
The CAPM and the single-index model both generate a parameter which
determines the return on the asset. Consequently, it is important to make clear
the interpretation of iI and iM .
152
The basic difference is that iI is derived from an assumption about the

determination of returns. In particular, it is derived from a statistical model
of the return process. The index on which returns are based is chosen, not
specified by any underlying analysis.
In contrast, iM is derived from an equilibrium theory. It emerges from the
assumptions of that theory rather than being imposed upon it. The assumptions
also generate a precisely defined value for iM .
Also, in the single-index model, the index I is usually assumed to be the
market index, but in principal could be any index. In the CAPM model, M is
always the market portfolio
Finally, the CAPM provides a sufficient set off assumptions for the singleindex model to be the true representation of the return-generating process rather
than just an approximation. Under its assumptions, returns are generated by a
linear relationship.
QUESTION TO ANSWER AT THIS POINT - SHOULD APHAS BE DIFFERENT OR THE SAME?
Example 89 But of what? Show that the two will be equal. I.e. used the two
methods of caculation to show the same. That is use the basic relationship that
i =
cov (ri , rm )
cov (ri rf , rm rf )
=
var (rm )
var (rm rf )
Example 90 Use the general relationship to demonstrate that cov (ri , rm ) =

cov (ri rf , rm rf ) and var (rm ) = var (rm rf ).
8.7
Pricing and Discounting
The CAPM also has implications for asset prices. Since the returns of assets
are related by the Security Market Line in equilibrium, the prices must also be
related.
To derive the relationship for asset prices, note that the return on an asset
can be written as
qi pi
ri =
,
(8.4)
pi
where pi is the purchase price and qi the (random) sale price. If dividends are
paid, they can be incorporated within qi . From the security market line
So
ri = rf + iM [rM rf ] .
(8.5)
pi (1) pi (0)
= rf + iM [rM rf ] ,
pi (0)
(8.6)
or
pi (0) =
pi (1)
.
1 + rf + iM [rM rf ]
(8.7)
8.7. PRICING AND DISCOUNTING
153
This should be the equilibrium market price of the asset.

Note role here: work out expected price and dividend in period 1 and discount back to period 0. The role of iM is to adjust the risk free rate of return to
give the correct rate of discounting for the degree of risk of the asset. This illustrates a general principle for discounting to find the present value of a project.
Note that pi (1) can just be seen as the expected value of a future random payoff from any kind of investment project. Then pi (0), the value today, is just
the discounted value of the that set of payments. The discounting includes the
return on risk-free to represent the time element and the beta term to reflect
correctuion for risk. Notice that the higher is beta the greater is the discounting.
So more risky projects (more risky in terms of beta with market) are dicounted
more heavily.
To see this as a general process observe that theproblem
at the heart of
!
valuation is to take a sequence of random cash flows Ct , t = 0, ..., T, and to
construct a present value at time 0. If preferences are risk neutral, the present
value is found easily by discounted the expected cash flow at t and discounting
at the risk-free rate. This would give

E C1
E C2
+
+...,
(8.8)
P V0 = C0 +
1 + rf
[1 + rf ]2
where C0 is taken as known at time 0. The difficulties begin when there is risk
aversion. Several methods are now considered for achieving the valuation with
risk aversion.
Discount at a rate capturing the risk in the cash flow. The present value
then becomes

E C1
E C2
+
+...,
(8.9)
P V0 = C0 +
1 + rc
[1 + rc ]2
with rc = rf + rp . Here rp can be interpreted as the risk premium that the the
risky cash flow must pay in excess of the risk-free rate. The difficulty in using
this approach is the determination of rp . It should reflect the premium applied
to other assets with similar risk.
Use the Certainty Equivalent. For each random cash flow there is a certainty
equivalent that satisfies

U (Cte ) = EU Ct ,
(8.10)
so that the utility of the certainty equivalent is equal to the expected utility of
the random cash flow. The present value then becomes
P V0 = C0 +
C1e
C2e
+
+....
1 + rf
[1 + rf ]2
(8.11)
This method is limited by the need to employ the utility function to determine
the certainty equivalent.
154
Each of these methods will work but has its own drawbacks. A further
method is now proposed and then explored in detail. Apply CAPM. The risk
premium rp can be determined very easily if the CAPM model is appropriate.
If CAPM applies then the security market line gives the relationship
rc = rf + c [rM rf ] .
(8.12)
The drawback with using CAPM is that it relies on restrictive assumptions.

Example 91 Add a simple example of how this can be used. Three states,
market return, covarinace, payment on project.
Example 92 Next example give market variance, project covariance, expected
value. Find beta and value project.
8.8
8.8 Market Portfolio
The CAPM model relies on the use of a market portfolio in order to be operative.
This market portfolio is meant to the the entire set of risky assets that are
available. It is not clear how this is obtained.
The major difficulty is the breadth of the market portfolio. It is meant to
include all risy assets not just financial securities. For example, it includes real
assets such as art and property and other assets such as human capital. This is
obviously not easy to define.
There are three situations in which this problem of defining the market
portfolio arises. The first is in the calculation of the beta values for assets.
Recall that these are obtained by covariance of the return on an asset with
the market divided by the variance of the return on the market. If the market
portfoli is incorrectly defined both of these values will also be wrong and the
estimated beta will not be correct.
The next problem is the construction of the capital market line and the
security market line. If an incorrect market portfolio is chosen and the beta
values estimated on the basis of this are wrong then the two lines will not
provide the correct predictions on returns.
The final problem is that the problem of the market portfolio makes it difficult to test whether the CAPM model is correct or not. If the prediction of the
security market line is used as a test of th emodel then a rejection can show that
either the model does not apply or the wrong market portfolio is used. More is
said about this in Chapter 10.
8.9
Conclusions
The CAPM moves us from fact (the acceptance of returns and variances as
data and the analysis of choice) to modelling of where this data comes from.
The CAPM determines the returns in equilibrium by assuming that they are
determined by adjustment of returns to equate the demand and supply of assets.
155
8.9. CONCLUSIONS
CAPM gives very clear conclusions. It explains the returns on assets through
the relationship with the market portfolio. It also gives a guide to investment
behavior through the combination of the market portfolio and the risk free asset.
The model also formalies why betas are of interest in investment analysis. But
all of these properties must be confronted with evidence since the assumptions
are equally strong.
Exercise 65 Assume there are two stocks, A and B, with A = 1.4 and B =
0.8. Assume also that the CAPM model applies.
(i) If the mean return on the market portfolio is 10% and the risk-free rate
of return is 5%, calculate the mean return of the portfolios consisting of:
a. 75% of stock A and 25% of stock B,
b. 50% of stock A and 50% of stock B,
c. 25% of stock A and 75% of stock B.
(ii) If the idiosyncratic variations of the stocks are A = 4, B = 2 and
the variance of the market portfolio is 2M = 12, calculate the variance of the
portfolios in (a), (b), (c).
(iii) What are the mean return and variance of the portfolios if they are 50%
financed by borrowing?
Exercise 66 Assume there are just two risky securities in the market portfolio.
Security A, which constitutes 40% of this portfolio, has an expected return of
10% and a standard deviation of 20%. Security B has an expected return of
15% and a standard deviation of 28%. If the correlation between the assets is
0.3 and the risk free rate 5%, calculate the capital market line.
Exercise 67 The market portfolio is composed of four securities. Given the
following data, calculate the market portfolios standard deviation.
Security Covariance with market Proportion
A
242
0.2
B
360
0.3
C
155
0.2
D
210
0.3
Exercise 68 Given the following data, calculate the security market line and
the betas of the two securities.
Security 1
Security 2
Market portfolio
Risk free asset
Expected return
Correlation with market portfolio
Standard deviation
15.5
9.2
12
5
0.9
0.8
1
0
2
9
12
0
Exercise 69 Consider an economy

are given below.
Number of Shares Price
A 100
1.5
B 150
2
with just two assets. The details of these

Expected Return
15
12
Standard Deviation
15
9
156
The correlation coefficient between the returns on the two assets if 1/3 and
there is also a risk free asset. Assume the CAPM model is satisfied.
(i) What is the expected rate of return on the market portfolio?
(ii) What is the standard deviation of the market portfolio?
(iii) What is the beta of stock A?
(iv) What is the risk free rate of return?
(vi) Construct the capital market line and the security market line.
Exercise 70 Consider an economy with three risky assets. The details of these
are given below.
No. of Shares Price Expected Return Standard Deviation
A 100
4
8
10
B 300
6
12
14
C 100
5
10
12
The correlation coefficient between the returns on any pair of assets is 1/2
and there is also a risk free asset. Assume the CAPM model is satisfied.
(i) Calculate the expected rate of return and standard deviation of the market
portfolio.
(ii) Calculate the betas of the three assets.
(iii) Use solution to (ii) to find the beta of the market portfolio.
(iv) What is the risk-free rate of return implied by these returns?
(v) Describe how this model could be used to price a new asset, D.
Exercise 71 Exercise to show that in regression of excess returns the value of
the intercept must be zero. Describe why this is a test of CAPM.
Exercise 72 Let return on market and asset be observed.
State
Asset
Market
(i) Find the for the asset.
(ii) Given in which state is it above and below the security market line?
(iii) Show that in expected terms it is on the SML.
Exercise 73 Take two assets with betas A and B held in proportions XA and
XB which are the market portfolio of risky assets. If the return of A is ?? and
rf =?, rM =? and M =?. What must be rB =? If rB =? what would you do?
If rB =? what would you do?
Exercise 74 Use the CAPM2 example for two risky assets and a simplified
utility to get some cancellation.
Exercise 75 (i) Consider an asset with expected future price of 10 and a beta
of 1.2. If rf = 0.05 and rM = 0.1, what is the fair market price of the asset
today?
(ii) If the equilibrium price today is ?, what is the expected price next year?
8.9. CONCLUSIONS
157
Exercise 76 A project costs $1000 to undetake and its payoff is related to the
market as in the table.
State
1 2 3 4
Project
Market
(i) Find the return on the project in each state.
(ii) Calculate the beta of the project.
(iii) Is the PDV of the project positive or negative?
(iv) If ?? were changed to ??, would decision on project alter?
158
Chapter 9
Arbitrage Pricing Theory

9.1
Introduction
Arbitrage Pricing Theory (APT) is an alternative to CAPM as a theory of

equilibrium in the capital market. It works under much weaker assumptions.
Basically, all that is required is that the returns on assets are linearly related to
a set of indices and that investors succeed in finding all profitable opportunities.
Thus, the multi-factor model of Chapter 11 is assumed to apply exactly.
The equilibrium is then obtained by asserting that there can be no unrealized
returns. This results from investors arbitraging away all possible excess profits.
9.2
Returns Process
The foundation of the APT is the assumption that the return on asset i is
generated by an underlying set of factors. To introduce the model in its simplest
form, it is assumed initially that there are only two factors. The extension of
the argument to many factors will be given later.
With two factors, the return on asset i is given by
ri = ai + b1i f1 + b2i f2 + ei ,
(9.1)
E [ei ej ] = 0, i = j.
(9.2)
with
Condition (9.2) implies that the non-systematic errors are uncorrelated between
any two assets so the errors not explained by the factors are unique to each asset.
It is now assumed that the portfolio of each investor is well-diversified so
that non-systematic risk can be ignored (see Section ??). Only the systematic
risk caused by the variation of the factors is then relevant. With the return
process in (9.1), the expected return on a portfolio is
rp = ap + b1p f 1 + b2p f 2 ,
159
(9.3)
160
CHAPTER 9. ARBITRAGE PRICING THEORY
where
ap =
n
wi ai ,
(9.4)
wi bki , k = 1, 2.
(9.5)
i=1
and
bkp =
n

i=1
9.3
Arbitrage
The implication of arbitrage is that there are no risk-free profits to be earned.

To see the role of this, consider the following example.
Let there be three portfolios A, B and C with returns and factor sensitivities
as given in the following table.
Portfolio
A
B
C
Expected return %
8
10
6
b1i
0.7
0.6
1.0
b2i
1.1
1.4
0.7
Now let there be a further portfolio, D, with an expected return of rD = 9%

and factor sensitivities of b1i = 0.8 and b2i = 1.0. A portfolio, E, formed as
a combination of portfolios A, B and C will match the factor sensitivities of
portfolio D if weights wA , wB , wC can be found such that
wA b1A + wB b1B + wC b1C
wA b2A + wB b2B + wC b2C
= 0.8,
= 1.0,
(9.6)
(9.7)
and
wA + wB + wC = 1.
(9.8)
wA = 0.4, wB = 0.2, wC = 0.4.
(9.9)
Solving these gives

These weights imply that the expected return on portfolio E is
rE = 0.4(10) + 0.2(8) + 0.4(6) = 7.6.
(9.10)
Arbitrage can then be conducted between portfolio E and portfolio D to

realize a return with no risk. An arbitrage portfolio involves selling one portfolio and purchasing an equal value of another portfolio. This involves no net
investment but, if successful, will generate a positive return.
To how this works, let the portfolio weights be wD and wE with wD > 0
and wE = wD . The latter condition is the one defining an arbitrage portfolio.
The expected return on the arbitrage portfolio is
rap = wD rD + wE rE = wD [rD rE ] = wD .
(9.11)
161
9.4. PORTFOLIO PLANE
In principal, the return on this arbitrage portfolio can be increased without limit
as wD is raised. Therefore a positive expected return is realized without any
net investment on the part of the investor.
Its systematic risk is given by the factor sensitivities
bkap = wD bkD + wE bkE = 0, k = 1, 2,
(9.12)
since the portfolios have the same sensitivities. Hence the systematic risk of the
portfolio is zero and the positive expected return on the arbitrage portfolio is
achieved with no risk.
This situation cannot exist in equilibrium. In fact, as investors buy the
arbitrage portfolio the return on portfolio D will be driven down and that on E
driven up. The consequences of this will be considered below.
9.4
Portfolio Plane
An understanding of the reason why such an arbitrage portfolio can be constructed is obtained by looking at the relationship between the returns on the
assets A, B and C and their factor sensitivities.
The three portfolios can be viewed as determining a plane in a three dimensional space where the expected return is graphed on the vertical axis and the
factor sensitivity on the horizontal axes. The general equation of the plane is
given by
ri = 0 + 1 b1i + 2 b2i .
(9.13)
Using the data in the example, the coefficients 0 , 1 and 2 can be calculated
to give the resulting equation
ri = 3.6 + 4b1i + 8b2i .
(9.14)
Now look again at portfolio D. Since

rD > 3.6 + 4b1D + 8b2D = 7.6,
(9.15)
it can be seen that portfolio D lies above the plane determined by A, B and C.
In contrast portfolio E lies on the plane since
rE = 3.6 + 4b1E + 8b2E = 7.6
(9.16)
directly below D. This can be interpreted as saying the D gives a return in

excess of that implied by its sensitivities.
9.5
General Case
The general construction is to take n factors with the return process for the N
(N = n) portfolios given by
ri = ai +
n

k=1
bki fk + ei ,
(9.17)
162
with
E [ei ej ] = 0, i = j.
(9.18)
These portfolios determine a plane in n-dimensional space which has equation

ri = 0 +
n
(9.19)
k bki .
k=1
The coefficients 0 , ..., n can

0
..
. =
n
be found from solving

b1A
1 b1N
1
..
.
1
bnA
..
.
bnN
rA
..
.
rN
(9.20)
The arbitrage argument is that if there is a portfolio that is not on this

plane, then an arbitrage portfolio can be constructed. If the factor sensitivities
of the portfolio not on the plane are bkD , k = 1, ..., n then the arbitrage portfolio
is defined by
N
wi bki
= bkD , k = 1, ..., n,
(9.21)
= 0.
(9.22)
i=1
N
wi
i=1
9.6
Equilibrium
It is now possible to derive the fundamental conclusion of APT. The argument

has already been made that if an arbitrage portfolio can be found, then investors
will find it. Investment in the arbitrage portfolio will ensure that any portfolios
off the plane determined by the returns of the other portfolios will be driven
onto it as prices (and hence returns) change.
This argument can be formalized as follows. Take a set of portfolios equal
in number to the number of factors. These will define a plane in ndimensional
space. Any distinct set of n portfolios will do for this purpose. For example,
the argument above could have used a plane constructed from portfolios A, B
and D and then found an arbitrage portfolio against C. All that matters for
the argument was that the set of four portfolios - A, B, C, D - did not lie on
the same plane. Then search to see if there is any portfolio not on this plane.
If there is, then construct an arbitrage portfolio and realize the expected gains.
This arbitrage activity will ensure that in equilibrium there can be no portfolios either below or above the plane. Thus all portfolio returns must be related
by the equation of the plane relating factor sensitivities to expected return.
The contribution of APT is to conclude that this equilibrium plane exists and
to characterize its structure.
163
9.7. PRICE OF RISK
9.7
Price of Risk
There is one final point to be made about the APT. The coefficient i is the price
of risk associated with factor i. That is, an extra unit of bki will be rewarded
with an increase in expected return equal to i . This is just a reflection again
of the fact that an investor will only accept greater variability (measured by a
higher value of bki ) if more return is gained. In equilibrium, the i s determine
just how much greater this risk has to be.
The final term to consider is 0 . The asset with bki = 0, k = 1, .., n is the
risk-free asset. Hence 0 is return on the risk-free asset.
9.8
APT and CAPM
APT, and the multi-factor model, are not necessarily inconsistent with CAPM.
In the simplest case with one factor, the two are clearly identical. With more
than factor further conditions must be met.
To obtain an insight into these, assume that returns are generated from two
factors so
ri = ai + b1i f1 + b2i f2 + ei .
(9.23)
The equilibrium from the APT model is then determined by the equilibrium
equation
ri = rf + 1 b1i + 2 b2i ,
(9.24)
where the condition 0 = rf has been used. The interpretation of k is that
this is the return above the risk-free rate earned by an asset with bki = 1 and all
other values of bji = 0. From the CAPM the value of this excess return should
be
i = i [rM rf ] .
(9.25)
Substituting this into (9.24) gives

ri
= rf + b1i 1 [rM rf ] + b2i 2 [rM rf ]

= rf + b1i 1 + b2i 2 [rM rf ] .
(9.26)
(9.27)
This is exactly the CAPM model where i = b1i 1 + b2i 2 . The two remain
consistent provided this identity holds.
9.9
Conclusions
The APT is helpful.

Exercise 77 find something
164
Exercise 78 add something

Exercise 79 and something
Chapter 10
Empirical Testing
10.1
Introduction
The tests of these equilibrium models are important since they influence how
the market is viewed. If either is correct, then that gives a direct influence upon
how investment decisions are made and evaluated. For instance, it CAPM is
true then it is unnecessary to purchase anything but the market portfolio.
10.2
CAPM
Tests of the CAPM are based on prediction that the market portfolio is efficient.
But this efficiency has to be judged in the light of information that was available
at the points at which investments were made. In other words, it is efficient in
an expected sense. Definition of the market portfolio.
Can it really be tested?
10.3
APT
What factors should be included?

Joint test of factor choice and model.
10.4
Conclusions
What are these? Models dont fit?

Exercise 80 add
Exercise 81 add
Exercise 82 add
165
166
CHAPTER 10. EMPIRICAL TESTING
Chapter 11
Efficient Markets and

Behavioral Finance
Add some chat
11.1
Introduction
This chapter now tests a very basic feature of model.
11.2
Efficient Markets
Intorudcue the idea of an efficient market.

Efficient Market
A (perfectly) efficient market is one in which every securitys price equals
its investment value at all times.
- if efficient, information is freely and accurately revealed
Types of Efficiency
Form of efficiency embodied in prices
Weak
prices of securities
Semistrong
publicly available information
Strong
public and private information
Interpretation: cannot make excess profits using the form of information
embodied in prices
Evidence: markets are at least weak-form efficient, strong is very doubtful
This finding is not surprising given the number of professional and amateur
investors attempting to find profitable opportunities.
167
168CHAPTER 11. EFFICIENT MARKETS AND BEHAVIORAL FINANCE
11.3
Tests of Market Efficiency
11.3.1
Event Studies
Look at the reaction of security prices after new information is released

- markets appear to perform well
11.3.2
Looking for Patterns
The return on an asset is composed of

- the risk free rate
- a premium for risk
But latter can only be predicted via a model so finding patterns is a joint
test of model and efficiency.
- one finding here is the January effect (returns abnormally high)
11.3.3
Examine Performance
Do professional investors do better?

- problem of determining what is normal, again a joint test
- problem of random selection
Results
Those with inside information can always do well so strong-form is usually
rejected
- e.g. trading of company directors
Tests of semi-strong often isolate strategies that earn abnormal returns but
usually not enough to offset transactions costs.
Weak-form - some possibility that investors overreact to some types of information.
11.4
Market Anomolies
Is it worht listing anomolies or are these part of the section above?
11.5
Excess Volatility
Does this fit in the previous section?
11.6
Behvioral Finance
Look at this as an explanation of some of the failure of market efficiency.
11.7. CONCLUSION
11.7
Conclusion
Put some here: things that need to be explained.

Exercise 83 put in exercise
Exercise 84 put in the next exercise
Exercise 85 add another.
169
170CHAPTER 11. EFFICIENT MARKETS AND BEHAVIORAL FINANCE
Part V
171
Chapter 12
Interest Rates and Yields

12.1
Introduction
Bonds are securities that promise to pay a fixed income and so are known as
fixed income securities. They are important investment instruments in their own
right. The returns on bonds are also important in determining the structure of
interest rates on different types of loans.
The income from a bond takes the form of a regular coupon payment and
the payment of principal on maturity. One central issue is to find a method of
comparison of bonds that can have very different structures of payments and
lengths of maturity. Although the promised payments are known at the time
the bond is purchased, there is some risk of default. This provides a role for
ratings agencies to assess the risk of bonds.
One special case of a bond is the risk-free security that has played such a
prominent role in the theoretical analysis. In practice, the risk-free security is
typically taken to be a United States or a United Kingdom short-term bond.
These have little risk of default so that their payments are virtually guaranteed.
Even these bonds are not entirely risk-free since there is always some risk due
to inflation being unpredicatble.
The chapter first discusses different types of bond. Then it moves towards
making comparisons between bonds. The first comparison is based on the assessment of risk characteristics as measured by rating agencies. Then bonds
are compared using the concept of a yield to maturity. Following this, the focus is placed upon interest rates. Spot rates and forward rates are related to
the payments made by bonds and it is shown how these interest rates are used
in discounting. Finally, the chapters looks at the concept of duration, which
measures a further property of a bond, and this is related to the price/yield
relationship.
173
174
12.2
CHAPTER 12. INTEREST RATES AND YIELDS
Types of Bond
A bond is a promise to make certain payments. All bonds are issued with a
maturity date which is the date at which the final payment is received. (There
are some exceptions to this: UK consuls issued to finance Napoleonic War are
undated) On the maturity date the bond repays the principal. The principal is
also called the face value.
As well as the payment of principal, bonds can also make periodic coupon
payments. Coupons are typically made semi-annually or annually. The final
payment on a bond at the maturity date is the sum of the last coupon and the
principal.
There are two distinct categories of bond which differ in whether they make
coupon payments or not.
(a) Pure discount bonds.
These are bonds which provide one final payment equal to the face value (or
par value) of the bond. The return on the bond arises from the fact that they
typically sell for less than the face value or at a discount.
These are the simplest kind of bond and there analysis underlies all other
bonds. As noted in the discussion of the efficient frontier, a pure discount bond
is basically a simple loan from the bond purchaser to the bond seller with the
length of the loan equal to the maturity of the bond. For example, a one-year
bond is a one-year loan. This interpretation will be employed frequently in this
chapter.
(b)Coupon bonds.
A coupon bond provides a series of payments throughout the life of the bond.
These payments are the coupons on the bond. It is possible to regard the coupon
as an interim interest payment on a loan. This perspective will be found helpful
at numerous points below.
So, with a pure discount bond only the final repayment of the loan is made.
With a coupon bond, regular interest payments are made then the principal is
repaid.
A bond is callable if the final payment may be made earlier than maturity.
This may sometimes be at a premium meaning the issuer of the bond has to
make an additional payment to the holder in order to call. The bond will be
called if its issuer finds it advantageous to do so. If it is advantageous for the
issuer, it is usually not so for the holder. Hence callable bonds must offer a
better return than non-callable to compensate for the risk of calling.
A bond is convertible if it includes an option to convert it to different assets.
A sinking fund is a bond issue which requires that a fraction of the bonds are
redeemed each period. This has the advantage of avoiding the necessity for a
large payment on the maturity date.
12.3. RATINGS AND DEFAULT
175
If a Treasury note or Bond is non-callable, it is effectively a portfolio of pure

discount bonds. For example, if the bond has a maturity of two, can regard
the coupon payment in year 1 as a pure discount and the coupon plus principal
in year 2 as a second pure discount. Coupon stripping is the process of selling
each coupon as an individual asset. This can have the advantage of allowing
investors to purchase assets whose timing of payments best matches their needs.
Because of this, stripping can create additional value.
12.3
Ratings and default
The first way of comparing bonds is to look at ratings. Bonds have some chance
of default. This varies across bonds. Government bonds tend to be the safest,
while some corporate bonds can be very risky. There are agencies who produce
ratings of the riskiness of bonds.
Bonds are rated according to the likelihood of default.
The two most famous rating agencies are:
(1)Standard and Poors;
(2)Moodys.
The categories used in these ratings systems are:
(1)Investment Grade Aaa - Baa
(2)Speculative Grade Ba - below
Informally, the lowest category of bonds are known as junk bonds. These
have a very high probability of default.
For corporate bonds, better ratings are associated with:
Lower financial leverage;
Smaller intertemporal variation in earnings;
Larger asset base;
Profitability;
Lack of subordination.
The possibility of default implies that a premium must be offered above the
risk-free rate of return in order to encourage investors to hold the bonds. This
premium is known as the risk premium.
12.4
Yield-to-Maturity
The basic issue when making comparisons of bonds is how to compare bonds
with different structures of payoffs. One way to do this is to consider the
(promised) yield-to-maturity. The word promised is used since the bond may
be called or go into default. In either case the full set of promised payments
will not be made.
176
The yield-to-maturity is the rate of return that equates present discounted

value of payments to the price of the bond. It is the most common measure
of a bonds return and allows for comparisons between bonds with different
structures of payments.
Definition 1 The yield to maturity is the interest rate (with interest compounded at a specified interval) that if paid on the amount invested would allow
the investor to receive all the payments of the security.
This definition is applied to a series of increasingly complex bonds and then
to the general case.
Let the principal, or face value, of the bond be M . This is paid at the
maturity date T. The coupon payment in year t is denoted by Ct and the
purchase price of the bond by p. The yield-to-maturity (or just yield from this
point) is denoted y.
For a pure discount bond with principal M and maturity of 1 year, the
yield is found by considering the investor to have two choices. Either they can
purchase the bond for p and one year later receive principal M. Or they can
invest p at a fixed rate of interest and at the end of the year have p [1 + y]. The
rate of interest that ensures these two choices lead to the same final wealth is
the yield. Therefore the yield, y, satisfies
p [1 + y] = M.
(12.1)
Example 93 A bond matures in 1 year, with principal of $1000,If the present

price $934.58, the yield-to-maturity satisfies 934.58 [1 + y] = 1000, so y = 0.07
(7%).
Now consider a pure discount bond with a two year maturity. The choices
confronting the investor are again either to purchase the bond or invest at a
fixed rate of interest. Following the latter course of action, they will receive
interest at the end of the first year. This will give them at total of p [1 + y] .
Retaining this investment, interest will again be earned at the end of the second
year. The yield then has to satisfy
[p [1 + y]] [1 + y] = p [1 + y]2 = M.
(12.2)
Example 94 A pure discount bond matures in 2 years, with principal $1000,

2
If the present price is $857.34, the yield-to-maturity satisfies 875.34 [1 + y] =
1000, so y = 0.08 (8%).
Now consider a coupon bond with maturity of two years, principal M,
coupon C and purchase price p. The way to match this is as follows. The
amount p is invested at interest rate y. At the end of the first year after payment
of interest this has become p [1 + y]. The payment of the coupon is equivalent
to withdrawing C from this sum. At the end of the second year, interest is paid
on the remaining sum p [1 + y] C. The yield must then satisfy
[p [1 + y] C] [1 + y] = p [1 + y]2 C [1 + y] = M + C.
(12.3)
177
12.4. YIELD-TO-MATURITY
Example 95 A coupon bond with principal of $1000 pays a coupon of $50

each year and matures in 2 years. If the present price is $946.93, the yieldto-maturity satisfies [1 + y] [[1 + y] 946.50 50] = 1050, so y = 0.08 (8%).
Reviewing these formula, it can be seen that the last two are special cases
of the expression
M
C
C
+
(12.4)
p=
+
2
2,
[1 + y] [1 + y]
[1 + y]
where for the pure discount bond C = 0. Observing this, for a bond with
maturity of T , the general expression defining the yield is
p=
T

t=1
M
C
.
t +
[1 + y]
[1 + y]T
(12.5)
Example 96 A bond has a maturity of 10 years, pays a coupon of $30 and has
a face value of $1000. If the market price is $845.57, the yield satisfies
845.57 =
10

t=1
1000
30
+
,
[1 + y]t [1 + y]10
so y = 0.05 (5%).
It is helpful to add a short explanation of how the yield, y, is actually calculated for these more complex examples. Mathematically, there is no formula
when T > 3. The basic, but time-consuming, approach is to use trial and error.
An initial guess of 10% is usually worth trying. A more sophisticated
approach
T
C
M
is to employ a suitable package to graph the value of p t=1 [1+y]
t
[1+y]T
as a function of y. The value that makes it equal to zero is the yield, y.
Example 97 Consider a bond with principal of $1000 that pays an annual
coupon of $30. The bond has a maturity of 5 years and the current price is
$800.
Using trial and error, produces the following table
y
0.05
0.06
0.07
0.08
0.081
0.0801
5
30
1000
+
913.41
873.63
835.99
800.36
796.91
800.02
t
5
t=1 [1+y]
[1+y]
5
30
1000
A graph of 800 t=1 [1+y]t [1+y]5 is given in Figure 12.1.
The yield-to-maturity can be used to determine whether a bond is good

value. This can be done by comparing the yield-to-maturity with an estimated
appropriate return. In this approach, the investor determines what they feel
should be the yield a bond offers and then compares it to the actual yield.
The estimation of an appropriate yield should be based on factors related to
the structure of payments and the riskiness of the bonds. The following will be
relevant:
178
Figure 12.1: Finding the Yield

Time to maturity This will determine the time until the principal is received. If the case is required earlier, the bond will have to be sold.
Coupon payments Coupon payments relate to the timing of the payment
flow compared to preferred flow
Call provisions To see the effect of these, consider why a bond would be
called. As an example, assume that on issue the bond had a coupon of
$100 but now similar bonds have coupon of $50. It would pay the issuer
to call the bond and replace it with the lower coupon bond. So, bonds
are generally called when yields fall. This benefits the issuer but not the
purchaser. Hence bonds with call provision should have higher yields to
compensate.
Tax status Bonds which are tax exempt will have a lower yield-to-maturity
that reflects the tax advantage.
Marketability (Liquidity) Bonds that are not very marketable need to have
a higher yield-to-maturity to induce investors to purchase them.
Likelihood of default As already described, bonds that may default have
a risk premium so the yield-to-maturity is higher.
Taken together, these factors determine an overall view of the bond and from
this the appropriate return can be inferred.
12.5. SEMI-ANNUAL AND MONTHLY COUPONS
12.5
179
Semi-Annual and Monthly Coupons
Many government and corporate bonds make coupon payments on a semi-annual

basis. This section shows how these bonds, and indeed bonds that pay coupons
at any regular interval, can be incorporated in the framework above.
The assumption of the previous section was that coupons were paid annually.
In fact, there is no place in the analysis where this actually matters. All that
does matter is that the coupons are paid at regular intervals, and this interval
can be 3 months, 6 months or any other alternative. The formula determining
the yield, equation (12.5), then determines the yield as a rate of interest for that
period.
Example 98 A bond with a maturity of two years pays a coupon of $30 semiannually and has a maturity value of $1000. If its price is $929.08, the semiannual yield is defined by
929.08 =
30
30
1000
30
30
+
+
+
,
+
1 + y [1 + y]2 [1 + y]3 [1 + y]4 [1 + y]4
so y = 0.05 (5%). This equation can be understood by noting that there are
4 six-months periods in 2 years. The interpretation of the result is that the
semi-annual interest rate is 5%.
Example 99 A bond with a maturity of 1 year pays a coupon of $10 monthly
and has a maturity value of $1000. With a current price of $894.25, the monthly
yield is found from
894.25 =
12

t=1
10
1000
,
t +
(1 + y)
(1 + y)12
which gives y = 0.02.

The issue that remains is to convert the semi-annual or monthly interest
rates into annual equivalents. There are two ways that this can be done which
lead to slightly different answers.
To motivate the first of these, consider investing $1 for a year with interest
paid semi-annually at rate y. After 6 months, the $1 becomes $1 + y after the
payment of interest. Now assume that the interest is not reinvested, but the
capital sum of $1 is. At the end of the year the $1 that has been invested has
become $1 + y. Adding the interest of $y that was withdrawn after 6 months,
gives the investor at total of $1 + 2y. The annual interest rate can then be
interpreted as 2y. Under this first approach the semi-annual interest rate is
converted to an annual rate by multiplying by 2. More generally, if interest is
paid n times per year at rate y, the annual interest rate is ny.
The second method of converting to an annual rate is to assume that the
interest earned after 6 months is reinvested. After 6 months, the $1 investment
is worth $1 + y, and with reinvestment is worth $ (1 + y) (1 + y) after 1 year.
This corresponds to an annual interest rate of (1 + y)2 1.
180
Example 100 Assume the semi-annual interest rate is 5%. Without reinvestment, the annual interest rate is 2 5 = 10%. With reinvestment, the annual
2
interest rate is (1 + 0.05) 1 = 0.1025 (10.25%).
Example 101 An investment pays interest of 2% each month. Without rein12
vestment, the annual interest rate is 24%. With reinvestment it is (1 + 0.02)
1 = 0.26824 (26.824%).
As the examples illustrate, the annual interest rate with reinvestment is
higher than without. For semi-annual interest, the difference is given by
(1 + y)2 1 2y = y 2 .
(12.6)
When y = 0.05, the difference y2 = 0.0025, as found in the example. In general,

if interest is paid n times per year, the difference between the annual interest
rate with reinvestment and that without reinvestment is
(1 + y)n 1 ny.
(12.7)
Example 102 If interest of 1% is paid monthly, the difference between the two
annual interest rates is
(1 + 0.01)12 1 12 0.01 = 0.006825.
There is no right or wrong in which of these interest rates to use. Both are
derived from legitimate, though different, experiments. In the range of interest
rates usually encountered in practice, the difference is small but significant.
When such conversions are necessary in later parts of the text, the reinvestment
method will be used for simplicity.
12.6
CONTINUOUS INTEREST
Will make much se of continuous compounding below so introcude it here as an

extension of this analysis - the section in Hull is simple.
12.7
Interest Rates and Discounting
There are a series of interest rates in the market place. These must be related
to prevent arbitrage. Such arbitrage would involve constructing an arbitrage
portfolio of loans. This section now relates these. It also ties in with the idea
of discounting.
12.7.1
Spot Rates
The spot rate is the interest rate associated with a spot loan: a loan that
is granted immediately (on the spot) with capital and interest repaid at a
181
12.7. INTEREST RATES AND DISCOUNTING
specified date. The discussion of the efficient frontier in Chapter 4 has already
made the interpretation of a bond as a loan. So the spot rates must be related
to the yields on bonds.
For pure discount bonds the relationship is very straightforward. A pure
discount bond is simply a loan from the purchaser to the issuer with the length
of the loan equal to the maturity of the bond. The yield on the bond must
therefore be equal to the rate of interest on a spot loan of this length. This
gives the identity
spot rate = yield-to-maturity.
(12.8)
This identity is true of any pure discount bond. Therefore the price of a discount
bond with maturity T is related to the spot rate St by
p=
M
(1 + ST )T
(12.9)
Given a set of pure discount bonds of maturities T = 1, 2, ... the application of

this formula provides the spot rates S1 , S2 , ...
Example 103 Three pure discount bonds of with principal of $1000 and with
maturities of 1, 2 and 3 years have prices $934.58, $857.34, $772.18 respectively.
1000
1000
, 857.34 = (1+S
and
The corresponding spot rates found from 934.58 = 1+S
)2
1
2
1000
772.18 = (1+S
3 . Hence S1 = 0.07 (7%), S2 = 0.08 (8%) and S3 = 0.09 (9%).
3)
Therefore an interest rate of 7% is paid on an immediate loan to be re-paid in
1 year and an interest rate of 9% applies to an immediate loan which has to be
re-paid in 3 years.
The analysis has to be adjusted to apply to a coupon bond. In the discussion

of coupon stripping it was noted how each coupon payment could be treated
as a separate loan. Hence the coupon paid at the end of the first year can be
treated as repayment of principal on a 1-year loan. This should attract interest
at rate S1 . Similarly, the coupon payment at the end of the second year can be
treated as the repayment of principal on a 2-year loan. It attracts interest at
rate S2 . The same logic can be applied to all later payments. The price paid
for the bond must represent the sum of the values of the repayments on these
individual loans. Hence the relationship of the price, coupon payments and
principal to the spot rates for a coupon bond of maturity T is given by
p=
T

t=1
C
M
.
t +
T
(1 + St )
(1 + ST )
(12.10)
Example 104 A bond with maturity of 3 years has a principal of $1000 and
makes a coupon payment of $50. If the price is $900 then the spot rates satisfy
900 =
50
50
50
1000
+
+
+
.
1 + S1 [1 + S2 ]2 [1 + S3 ]3 [1 + S3 ]3
182
It is clear that the spot rates cannot be calculated using information on a

single coupon bond. Instead they must be constructed by an iterative process.
This works by first taking either a pure discount bond or a coupon bond with
a maturity of 1 year. With the pure discount bond, S1 is determined by p =
M
C
M
1+S1 and with the coupon bond by p = 1+S1 + 1+S1 . The spot rate S2 can
then be found using coupon bond with a maturity of 2 years by observing that
C
C
M
p = 1+S
+ [1+S
can be solved for S2 once S1 is known. Next, using
2 +
[1+S2 ]2
1
2]
S1 and S2 it is possible to use a coupon bond with a maturity of 3 years to find
S3 . This process can be continued to consecutively construct a full set of spot
rates using the prices of a series of bonds of different maturity.
Example 105 Three bonds have face values of $1000. The first is a pure discount bond with price $909.09, the second is a coupon bond with coupon payment
$40, a maturity of 2 years and a price of $880.45 and the third bond is a coupon
bond with coupon of $60, maturity of 3 years and price of $857.73.
The spot rate S1 is given by
909.09 =
1000
,
1 + S1
so S1 = 0.1 (10%). Using the fact that S1 = 0.1, S2 is determined by

880.45 =
40
1040
,
+
1.1 [1 + S2 ]2
so S2 = 0.11 (11%). Finally, using S1 and S2 , S3 solves

857.73 =
60
1060
60
+
+
.
2
1.1 [1.11]
[1 + S3 ]3
This gives S3 = 0.12 (12%).
12.7.2
Discount Factors
If a future flow of payments are to be received, the standard process is to convert

these to a present value by using discounting. The reason for doing this is that
it allows different flows to be directly compared by using their present values.
Discount factors are used to discount payments to find the present value of
future payments.
These discount factors can be used to find the present value of any security.
Let the payment in period t be Vt and assume the last payment is received in
period T . With discount factor dt the present value of the flow of payments is
PV =
T

t=1
dt Vt .
(12.11)
12.7. INTEREST RATES AND DISCOUNTING
183
This can also be expressed in terms of a discount rate. If the discount rate, ,
1
is constant then dt = (1+)
t and
PV =
T

t=1
1
t Vt .
(1 + )
(12.12)
This method of discounting can be used whether the payments are certain
or risky. When they are risky it is necessary to take explicit account of the
risk. One way to do this was seen in Chapter 8 where the expected value of the
payment in period t was used and the discount rate adjusted for risk using the
beta. An alternative way of incorporating risk in the discounting will be see in
Chapter 14.
If the payments are certain, then the there is no need to adjust for risk and
the discount factors can be related directly to the returns on bonds and the spot
rates. In fact, if dt is defined as the present value of $1 in t years, then
dt =
1
,
(1 + St )t
(12.13)
where St is the spot rate on a loan that must be repaid in t years. If the value
of $1 were above or below this value, then an arbitrage possibility would arise.
Using these discount factors, the present value of a flow of payments is
PV =
T

t=1
1
Vt .
(1 + St )t
(12.14)
Example 106 If d1 = 0.9346 and d2 = 0.8573 and a security pays $70 in 1

years time and $1070 in 2 years time then
P = 0.9346 70 + 0.8573 1070 = 982.73.
Example 107 If S1 = 0.09, S2 = 0.1 and S3 = 0.11, the present value of the
flow V1 = 50, S2 = 50 and S3 = 1050 is
PV =
12.7.3
50
1050
50
+
+
= 854.94.
2
1.09 (1.1)
(1.11)3
Forward Rates
The spot rates of interest relate to immediate loans. It is also possible to consider
agreeing today for a loan to be granted at some future date with repayment at
some even later data. For instance, an investor could agree to receive a loan
in one years time to be paid back in two years. Such loans are called forward
loans.
The rate of interest on a forward loan is called the forward rate. The interest
rate on the loan made in one years time to be paid back in two years is denoted
184
by f1,2 . It should be stressed that this is an interest rate agreed today for a loan
in the future. If the loan contract is accepted by the lender and borrower this is
the interest rate that will be paid on that loan. The important point is that it
need not be the same as the rate of interest that applies to one-year spot loans
in a years time.
Forward rates have to be related to current spot rates to prevent arbitrage,
so they link the spot rates for different years. To see how this link emerges,
consider two alternative strategies:
Invest for one year at spot rate S1 and agree today to invest for a second
year at forward rate f1,2 ;
Invest for two years at spot rate S2 .
To avoid any possibility of arbitrage, the returns on these two strategies
must be equal. If they were not, it would be possible to borrow at the lower
rate of interest and invest at the higher, yielding a risk-free return for no net
investment. A dollar invested in strategy 1 is worth (1 + S1 ) after one year and,
reinvested at interest rate f1,2 , becomes (1 + S1 ) (1 + f1,2 ) after two years. A
2
dollar invested in strategy 2 is worth (1 + S2 ) after two years. The equality
between the returns requires that
2
(1 + S1 ) (1 + f1,2 ) = (1 + S2 ) .
(12.15)
Hence
(1 + S2 )2
.
(12.16)
(1 + S1 )
The spot rates therefore determine the interest rate on a forward loan.
1 + f1,2 =
Example 108 Let S1 = 0.08 and S2 = 0.09. Then

2
1 + f1,2 =
(1 + 0.09)
,
(1 + 0.08)
so f1,2 = 0.1.
The same argument can be applied between to link the spot rates in any
periods t and t 1 to the forward rate ft,t1 . Doing so gives the general formula
for the forward rate between years t 1 and t as
t
1 + ft,t1 =
(1 + St )
t1 .
(1 + St1 )
(12.17)
Example 109 The spot rate on a loan for 10 years is 12% and the spot rate
on a loan for 11 years is 13%. To prevent arbitrage, the forward rate f10,11 has
to satisfy
(1 + 0.13)11
,
1 + f10,11 =
(1 + 0.12)10
so f10,11 = 23.5%.
185
12.8. DURATION
Forward rates are linked to spot rates and spot rates determine discount
factors. Therefore there is a link between forward rates and discount factors.
This is given by the relation
dt =
12.8
1
t1
(1 + St1 )
(1 + ft1,t )
(12.18)
Duration
Duration is a measure of the length of time until the average payment is made
on a bond. This can be used to compate different bonds. Duration can also be
used to capture the sensitivity of price to the interest rate. This section shows
hwo to calculate duration for a single bond and then for a portfolio of bonds.
If cash flows are received at times 1, ..., T then the duration, D, is given by
D=
P V (1) + 2 P V (2) + ... + T P V (T )

PV
(12.19)
where P V (t) is the present value of the cash flow at time t and is defined by
P V (t) =
Ct
t,
(1 + y)
(12.20)
and P V is the total present value of the cash flow. When this formula is applied
to a bond, the pricing ensures that P V is also the market price of the bond.
For a zero-coupon bond no payments are made prior to the final value. Hence
P V (T ) = P V so
D = T,
(12.21)
and the duration is equal to the time to maturity. For a coupon bond, the
intermediate payments ensure that the duration has to be less than the maturity,
giving
D < T.
(12.22)
Example 110 Consider a bond that pays an annual coupon of $40, has a face
value of $1000 and a maturity of 6 years. With a discount rate of 3%, the
following table computes the values required to calculate the duration
Time Cash Flow Discount Factor
PV of Cash Flow
t
1
1
40
=
0.97087
40
0.97087
=
38.835
38.835
1.03
1
2
40
40
0.94260
=
37.704
2 37.704 = 75.408
2 = 0.94260
[1.03]
1
3
40
=
0.91514
40
0.91514
=
36.606
3 36.606 = 109.82
[1.03]3
1
4
40
= 0.88849 40 0.88849 = 35.540
4 35.540 = 142.16
[1.03]4
1
5
40
= 0.86261 40 0.86261 = 34.504
5 34.504 = 172.52
[1.03]5
1
6
1040
=
0.83748
1040
0.83748
=
870.98
6 870.98 = 5225.9
[1.03]6
Using these values the duration is

PV t
38.835 + 75.408 + 109.82 + 142.16 + 172.52 + 5225.9

= 5.4684.
=
38.835 + 37.704 + 36.606 + 35.540 + 34.504 + 870.98
PV
186
The calculation of the duration can be extended to portfolios of bonds. Consider two bonds A and B with durations
T
tP V A (t)
A
,
(12.23)
D = t=1 A
PV
T
where P V A = t=1 P V A (t) , and
B
D =

with P V B = Tt=1 P V B (t) .
These facts imply that
P V A DA + P V B DB =
T
T

t=1
t=1 tP V
PV B
(t)
tP V A (t) +
(12.24)
T
tP V B (t) .
(12.25)
t=1
The duration of the portfolio is defined by

T
T
tP V A (t) + t=1 tP V B (t)
,
D = t=1
PV
(12.26)
where P V = P V A + P V B . But (12.25) implies that

D=
PV A A PV B B
D +
D .
PV
PV
(12.27)
This result establishes that the duration of a portfolio is a weighted sum of

durations of the individual bonds.
12.9
Price/Yield Relationship
From the fact that the price of the bond is determined by

P =
C
C +M
C
+
+ ... +
,
1 + y (1 + y)2
(1 + y)T
(12.28)
it can be observed that:

1. P and y are inversely related.
This follows from seeing that
dP
C
2C
T (C + M )
=
...
< 0.
2
3
T +1
dy
(1 + y)
(1 + y)
(1 + y)
(12.29)
2. The relationship is convex.

Calculation gives
2C
3C
(T + 1) (C + M )
d2 P
=
+
+ ... +
> 0.
3
4
T +2
2
dy
(1 + y)
(1 + y)
(1 + y)
(12.30)
187
12.9. PRICE/YIELD RELATIONSHIP
Price
True
Relation
P
Approximation
Using Duration
Yield-toMaturity
Figure 12.2: Price/Yield Relationship

The duration can also be used to link changes in the yield on a bond to
changes in its price. From (12.20)
t
dP V (t)
=
P V (t) .
dy
1+y
Hence using the fact that P V = P
d
dP
=
dy
or
T
T
t=1 P V
(12.31)
(t) ,
T

P V (t)
1
t
=
P V (t) =
DP,
dy
1
+
y
1
+
y
t=1
t=1
1
dP
= Dm P, Dm =
D,
dy
1+y
(12.32)
(12.33)
where Dm is called the modified duration.

This is an exact result that holds for a differential (meaning very small
change) in the yield. If can also be used as an approximation relating the price
to the modified duration and yield changes. So
P Dm P y.
(12.34)
This shows the approximation, but comparison with (12.30) also shows that the
use of duration overstates the effects of a yield increase.
This result is returned to later after the yield curve has been considered.
188
12.10
Bond Portfolios
Some stuff here on the management of bond protfolis.

Immunisation
???
12.11
Conclusions
The chapter has considered methods for comparing bonds with different structures of payments and different maturities. Bond ratings were analyzed as was
the yield as a measure of the return. Bonds represent one form of lending, so the
interest rates on bonds are related to the interest rates on loans. This analysis
tied together spot rates, forward rates and discount factors. The duration as
another measure of a bond was also considered and price/yield relationships
were investigated.
Exercise 86 What is coupon stripping? What are the benefits of this for
investors?
Exercise 87 Three pure discount bonds, all with face values of $1000, and maturities of 1, 2 and 3 years are priced at $940, $920 and $910 respectively.
Calculate their yields. What are their yields if they are coupon bonds with an
annual coupon of $40?
Exercise 88 An investor looks for a yield to maturity of 8% on fixed income
securities. What is the maximum price the investor would offer for a coupon
bond with a $1000 face value maturing in 3 years paying a coupon of $10 annually with the first payment due one year from now? What is the maximum price
if it is a pure discount bond?
Exercise 89 Three pure discount bonds, all with face values of $1000, and maturities of 1, 2 and 3 years are priced at $950.89, $942.79 and $929.54 respectively. Calculate:
a. The 1-year, 2-year and 3-year spot rates;
b. The forward rates from year 1 to year 2 and from year 2 to year 3.
Exercise 90 Calculate the duration of a bond with a coupon of $50 and maturity value of $1000 if it matures in six years and the discount rate is 4%.
Chapter 13
The Term Structure

13.1
Introduction
This chapter looks at the variation of yields with respect to time and reviews
theories designed to explain this.
13.2
Yield and Time
The yield curve shows the yield-to-maturity for treasury securities of various
maturities at a particular date. In practice, securities do not lie exactly on this
line because of differences in tax treatment and in callability.
It should be noted that the yield-to-maturity is a derived concept from the
flow of payments and it would equally informative to have used duration on
horizontal axis rather than maturity.
13.3
Term Structure
A similar graph can be constructed using spot rates on the vertical axis. This
is called the term structure of interest rates. Spot rates are more fundamental
than the yield-to-maturity.
The following question are raised by the term structure:
i. Why do rates vary with time?
ii. Should the term structure slope up or down?
Although the term structure can slope either way, periods in which it slope
upwards are more common than periods in which it slopes down.
The following theories have been advanced to answer these questions.
189
190
CHAPTER 13. THE TERM STRUCTURE
Yield-toMaturity
Time to
Maturity
Figure 13.1: Yield Curve
13.4
Unbiased Expectations Theory
This theory is based on the view that forward rates represent an average opinion
about expected rates in the future. So,
- if yield curve upward sloping, rates are expected to rise,
- if yield curve downward sloping, rates are expected to fall.
Example 111 Consider the investment of 1. Let the 1-year spot rate be 7%,
the two year spot rate be 8%.
Consider the following two strategies
a. invest now for two years.
2
Final return = 1 [1.08] = 1.664
b. invest for one year, then again for a further year
Final return = 1 [1.07] [1 + es1,2 ]
where es1,2 is the expected one year spot rate in year 2.
This strategies must yield the same return which implies es1,2 = 0.0901 .
Reversing this argument
- one year spot rate today is 7%
- one year spot rate expected next year is 9.01%
- so two year spot rate must be 9%
Hence the yield curve slopes upwards under the assumptions.
191
13.5. LIQUIDITY PREFERENCE THEORY

In equilibrium: it must be the case that
es1,2 = f1,2 ,
(13.1)
so that the expected future spot rate is equal to the forward rate. This would
be true for all time periods.
13.5
Liquidity Preference Theory
This theory is based on the idea that investors prefer, all things equal, shortterm securities to long-term securities. This can be justified by assuming that
investors place an intrinsic value on liquidity.
For example, consider making an investment for a two-year period. This can
be done using two different strategies.
i. Maturity Strategy
- hold a two-year asset
ii. Rollover Strategy
- hold two one-year assets
An investor who values liquidity would prefer the rollover strategy. They
might need cash at end of period 1 and with maturity strategy, price of asset
at end of year 1 is not known. Using the rollover strategy eliminates this price
risk. Consequently, in order to make them attractive, longer term securities
must have a risk premium
To see this
- expected return on 1 with rollover strategy is
1 [1 + s1 ] [1 + es1,2 ]
(13.2)
- expected return on 1 with maturity strategy is

1 [1 + s2 ]2
(13.3)
The maturity strategy must have higher return to compensate for loss of
liquidity so
2
[1 + s1 ] [1 + es1,2 ] < [1 + s2 ]
(13.4)
[1 + s1 ] [1 + f1,2 ] = [1 + s2 ]2
(13.5)
f1,2 > e1,2
(13.6)
f1,2 = e1,2 + L1,2
(13.7)
Since by definition
it follows that
or
192
CHAPTER 13. THE TERM STRUCTURE
where L1,2 is the liquidity premium.

Under the Liquidity Preference Theory the term structure again depends on
the expected spot rate but with the addition of the liquidity premium.
Note that if all spot rates are equal, the liquidity premium ensures the term
structure slopes upwards. For it to slope downwards, spot rates must be falling.
Thus the liquidity premium ensures that the term structure slopes upwards
more often than it slopes downwards.
13.6
Market Segmentation (Preferred Habitat)
The basic hypothesis of this theory is that the market is segmented by maturity
date of the assets. It motivates this by assuming that investors have different
needs for maturity.
The consequence is that supply and demand for each maturity date are
independent and have their return determined primarily by the equilibrium in
that section. Points on the term structure are related only by substitution of
marginal investors between maturities.
13.7
Empirical Evidence
Strict market segmentation - little empirical support

- observe continuity of term structure
There is some evidence that term structure conveys expectations of future
rates
- but with inclusion of liquidity preferences
- but these premiums change over time
13.8
Implications for Bond Management
Look at how bonds can be managed to protect against effects of interest changes.
Link the duration, and interest rates and term structure
Add immunization methods.
13.9
Conclusion
Complete explanation has not been found but term structure can be used to
provide information on expected level of future rates.
Exercise 91 Derive a term structure.
Exercise 92 Solve an immunization example.
Exercise 93 Do price/yield and duration example.
Exercise 94 Example on risk minimization.
Part VI
Derivatives
193
Chapter 14
Options
In the practise of investment analysis, as in life generally, commitment can be costly. Commitment can force a damaging course of
action to be seen through to the end long after it is clear that it is
wrong. Options, though, are valuable. They allow us to pick what
is right when it is right or to choose not to select anything at all.
Simple though it may be, the act of keeping our options open is
good investment advice. Financial markets have long realized these
facts and have developed financial instruments that allow options to
be kept open. Since having an option is valuable, it can command a
price and be traded on a market. The purpose of investment analysis
is to determine the value to place upon an option. This may seem an
imprecise question, but in no other area of finance has investment
analysis been more successful in providing both a very clear answer
and revolutionizing the functioning of the market.
14.1
Introduction
An option is a contract that gives the holder the right to undertake a transaction if they wish to do so. It also gives them the choice to not undertake the
transaction. Possessing this freedom of choice is beneficial to the holder of the
option since they can avoid being forced to make an undesirable trade. Options
therefore have value and the rights to them are marketable.
The issue that the investment analyst must confront when faced with options is to determine their value. It is not possible to trade successfully without
knowing the value of what is being traded. This applies equally to the financial
options traded on established markets and to more general instruments, such
as employment contracts, which have option-like features built in. This chapter
will describe the standard forms of option contract and then gradually build towards a general formula for their valuation. The individual steps of the building
process have independent worth since they provide a methodology for tackling
195
196
CHAPTER 14. OPTIONS
a range of valuation issues.
14.2
Options
There are two basic types of options. A call option gives the right to buy an
asset at a specific price within a specific time period. A put option gives the
right to sell an asset at a specific price within a specific time period. The price
at which the trade can take place is called the exercise or strike price. The
asset for which there is an option to buy or sell is often called the underlying
asset. If the underlying asset is a common stock, then the standard call and
put options are called plain vanilla options. This distinguishes them from other
more complex options which, for example, can provide the option to buy another
option. If the option is used, for example the holder of a call option chooses to
buy the underlying stock, the option is said to have been exercised.
14.2.1
Call Option
A plain vanilla call option is the right to buy specific shares for a given price
within a specified period. The premium on an option is the price paid by the
investor to purchase the option contract.
The contract for a plain vanilla call option specifies:
The company whose shares are to be bought;
The number of shares that can be bought;
The purchase (or exercise) price at which the shares can be bought;
The date when the right to buy expires (expiration date).
A European call can only be exercised at the time of expiration. This means
that the purchaser of the option must hold it until the expiration date is reached
and only then can choose whether or not to exercise the option. In contrast an
American call can be exercised at any time up to the point of expiration.
If an investor purchases a call option, they must have some expectation that
they will wish to exercise the option. Whether they will wish to do so depends
critically upon the relationship between the exercise price in the contract and
the price of the underlying asset. Clearly they will never exercise the right to
buy if the price of the underlying asset is below the exercise price: in such a case
they could purchase the underlying asset more cheaply on the standard market.
For a European call, the option will always be exercised if the price of the
underlying asset is above the exercise price at the date of expiration. Doing
so allows the investor to purchase an asset for less than its trading price and
so must be beneficial. With an American call, the issue of exercise is more
complex since there is also the question of when to exercise which does not arise
with European options. Putting a detailed analysis of this aside until later, it
remains correct that an American option will only be exercised if the price of
14.2. OPTIONS
197
the underlying is above the exercise price and will certainly be exercised if this
is true at the expiration date.
Example 112 On July 11 2003 Walt Disney Co. stock were trading at $20.56.
Call options with a strike or exercise price of $22.50 traded with a premium of
$0.05. These call options will only be exercised if the price of Walt Disney Co.
stock rises above $22.50.
In order for a profit to be made on the purchase of a call option it is necessary
that the underlying stock prices rises sufficiently above the exercise price to offset
the premium.
Example 113 Call options on Boeing stock with a strike price of $30.00 were
trading at $5.20 on June 23, 2003. If a contract for 100 stock were purchased
this would cost $520. In order to make a profit form this, the price on the
exercise date must be above $35.20.
The next example describes the financial transfers between the two parties
to an options contract.
Example 114 Consider A selling to B the right to buy 100 shares for $40 per
share at any time in the next six months. If the price rises above $40, B will
exercise the option and obtain assets with a value above $40. For example, if
the price goes to $50, B will have assets of $5000 for a cost of $4000. If the
price falls below $40, B will not exercise the option. The income for A from this
transaction is the premium paid by B to purchase the option. If this is $3 per
share, B pays A $300 for the contract. If the price of the share at the exercise
date is $50, the profit of B is $5000 - $4000 - $300 = 700 and the loss of A
is 300 - $1000 = $700. If the final price $30, the profit of A is $300 and the
loss of B is - $300.
Two things should be noted from this example. Firstly, the profit of one
party to the contract is equal to the loss of the other party. Options contracts
just result in a direct transfer form one party to the other. Secondly, the loss to
A (the party selling the contract) is potentially unlimited. As the price of the
underlying stock rises, so does their loss. In principle, there is no limit to how
high this may go. Conversely, the maximum profit that A can earn is limited
to the size of the premium.
The final example illustrates the general rule that call options with lower
exercise prices are always preferable and therefore trade at a higher price. Having a lower exercise price raises the possibility of earning a profit and leads to a
greater profit for any given price of the underlying.
Example 115 On June 23, 2003 IBM stock were trading at $83.18. Call options with expiry after the 18 July and a strike price of $80 traded at $4.70.
Those with a strike price of $85 at $1.75.
198
CHAPTER 14. OPTIONS
A final point needs to be noted. Exercise of the option does not imply that
the asset is actually sold by one party to the other. Because of transactions
costs, it is better for both parties to just transfer cash equal in value to what
would happen if the asset were traded.
14.2.2
Put Options
A plain vanilla put option is the right to sell specific shares for a given price
within a specified period. The contract for a plain vanilla put option specifies:
The company whose shares are to be sold;
The number of shares that can be sold;
The selling (or exercise) price at which the shares can be sold;
The date when the right to sell expires (expiration date).
As with calls, a European put can only be exercised at the expiration date
whereas an American put can be exercised at any date up to the expiration
date. The difference in value between American puts and European puts will
be explored later. But it can be noted immediately that since the American put
is more flexible than the European put, it value must be at least as high.
Example 116 On July 11 2003 Walt Disney Co. stock were trading at $20.56.
Put options with a strike or exercise price of $17.50 traded with a premium of
$0.10. These put options will only be exercised if the price of Walt Disney Co.
stock falls below $17.50.
It is only possible to profit from purchasing a put option if the price of the
underlying asset falls far enough below the exercise price to offset the premium.
Example 117 Put options on Intel stock with a strike price of $25.00 were
trading at $4.80 on June 23, 2003. If a contract for 100 stock were purchased
this would cost $480. In order to make a profit form this, the price on the
exercise date must be below $20.20.
In contrast to the position with a call option, it can be seen from the next
example that the loss to the seller of a put contract is limited, as is the potential
profit for the purchaser. In fact, the loss to A (or profit to B) is limited to the
exercise price and the loss of B (profit to A) is limited to the premium.
Example 118 A sells B the right to sell 300 shares for $30 per share at any
time in the next six months. If the price falls below $30, B will exercise the
option and obtain a payment in excess of the value of the assets. For example,
if the price goes to $20, B will receive $9000 for assets worth $6000. If the
price stays above $30, B will not exercise the option. The income for A is the
premium paid by B for the option. If this is $2 per share, B pays A $600 for the
14.3. VALUATION AT EXPIRY
199
contract. If the price of the stock at expiry of the contract is $20, the profit of
B is $9000 - $6000 - $600 = $2400 and the loss of A is $6000 + $600 - $9000
= - $2400. If the final price is $40, the loss of B is - $600 and the profit of A
is $600.
The next example illustrates that the higher is the strike price, the more
desirable is the put option. This is because a greater profit will be made upon
exercise.
Example 119 On June 23, 2003 General Dynamics stock were trading at $73.83.
Put options with expiry after the 18 July and a strike price of $70 traded at
$1.05. Those with a strike price of $75 traded at $2.95.
14.2.3
Trading Options
Options are traded on a wide range of exchanges. Most prominent amongst

those in the US are the Chicago Board Options Exchange, the Philadelphia
Stock Exchange, the American Stock Exchange and the Pacific Stock Exchange.
Important exchanges outside the US include the Eurex in Germany and Switzerland and the London International Financial Futures and Options Exchange.
Options contracts are for a fixed number of stock. For example, an options
contract in the US is for 100 stock. The exercise or strike prices are set at
discrete intervals (a $2.50 interval for stock with low prices, up to a $10 interval
for stock with high prices). At the introduction of an option two contracts
are written, one with an exercise prices above the stock price and one with an
exercise price below. If the stock price goes outside this range, new contracts
can be introduced. As each contract reaches its date of expiry, new contracts
are introduced for trade.
Quotes of trading prices for options contracts can be found in both The Wall
Street Journal and the Financial Times. These newspapers provide quotes for
the call and put contracts whose exercise prices are just above and just below
the closing stock price of the previous day. The price quoted is for a single
share, so to find the purchase price of a contract this must be multiplied by the
number of shares in each contract. More detailed price information can also be
found on Yahoo which lists the prices for a range of exercise values, the volume
of trade and the number of open contracts.
Market makers can be found on each exchange to ensure that there is a
market for the options. The risk inherent in trading options requires that margin
payments must be must in order to trade.
14.3
Valuation at Expiry
The value of an option is related to the value of the underlying asset. This is
true throughout the life of an option. What is special about the value of the
option at the expiration date is that the value can be computed very directly.
200
CHAPTER 14. OPTIONS
Vc
Figure 14.1: Value of a Call Option at Expiry

Prior to expiration the computation of value requires additional analysis to be
undertaken, with the value at expiration an essential component of this analysis.
If a call option is exercised the holder receives a sum equal to the difference
between the price of the underlying stock at expiry and the exercise price. An
option with an exercise price of $10 on an underlying stock with price $15 is
worth $5. If the option is not exercised, the exercise price must be above the
price of the underlying and the value of the option is $0. These observations
can be summarized by saying that value, or fair price, at expiration is given
by
(14.1)
V c = max {S E, 0} ,
where the max operator means that whichever is the larger (or the maximum)
of 0 and S E is selected. Hence if S E = 5 then max {5, 0} = 5 and if
S E = 2 then max {2, 0} = 0. The formula for the value of a call option at
expiry is graphed in Figure 14.1. The value is initially 0 until the point at which
S = E. After this point, each additional dollar increase in stock price leads to
a dollar increase in value.
Example 120 On June 26 2003 GlaxoSmithKline stock was trading at $41.
The exercise prices for the option contracts directly above and below this price
were $40 and $42.50. The table displays the value at expiry for these contracts
for a selection of prices of GlaxoSmithKline stock at the expiration date.
S
max {S 40, 0}
max {S 42.5, 0}
37.50
0
0
40
0
0
41
1
0
42.50
2.50
0
45
5
2.50
47.50
7.50
5
Setting aside the issue of timing of payments (formally, assuming that no

discounting is applied) the profit, c , from holding a call option is given by
201
V0c
Figure 14.2: Profit from a Call Option

its value less the premium paid. If the premium is denoted V0c , profit can be
written as
c
= V c V0c = max {S E, 0} V0c

= max {S E V0c , V0c } .
(14.2)
The relationship between profit and the price of the underlying asset is graphed
in Figure 14.2. The figure shows how the profit from purchasing a call option
is potentially unlimited.
If a put option is exercised the holder receives a sum equal to the difference
between the exercise price and the price of the underlying stock at expiry. An
option with an exercise price of $10 on an underlying stock with price $5 is
worth $5. If the option is not exercised, the exercise price must be below the
price of the underlying and the value of the option is $0. These observations
can be summarized by saying that value, or fair price, at expiration is given
by
(14.3)
V p = max {E S, 0} ,
so that the value is whichever is larger of 0 and E S. The formula for the value
of a put option is graphed in Figure 14.3. When the underlying stock price is
0, the option has value equal to the exercise price. The value then declines as
the underlying price rises, until it reaches 0 at S = E. It remains zero beyond
this point.
Example 121 Shares in Fox Entertainment Group Inc. traded at $29.72 on 7
July 2003. The expiry value of put options with exercise prices of $27.50 and
$30.00 are given in the table for a range of prices for Fox Entertainment Group
Inc. stock.
202
CHAPTER 14. OPTIONS
Vp
Figure 14.3: Value of a Put Option

S
max {27.50 S, 0}
max {30 S, 0}
20
7.50
10
22.50
5
7.50
25.00
2.50
5
27.50
0
2.50
30
0
0
32.50
0
0
The profit from purchasing a put option (again assuming no discounting so

the timing of payments can be ignored) is given by the difference between the
premium paid, V0p , and its value at expiry. Hence
p
= V p V0p = max {E S, 0} V0p

= max {V0p , E S V0p } .
(14.4)
This profit is graphed in Figure 14.4 as a function of the price of the underlying
stock at expiry. The figure shows how the maximum profit from a put is limited
to E V0p .
These results can be extended to portfolios involving options. Consider a
portfolio consisting of as units of the underlying stock, ac call options and ap
put options. A short position in any one of the three securities is represented
by a negative holding. At the expiry date, the value of the portfolio is given by
P = as S + ac max {S E, 0} + ap max {E S, 0} .
(14.5)
The profit from the portfolio is its final value less the purchase cost.
Example 122 Consider buying two call options and selling one put option,
with all options having an exercise price of $50. If calls trade for $5 and puts
for $10, the profit from this portfolio is
= 2 max {S 50, 0} max {50 S, 0} 2 5 + 10.
203
E V0p
V0p
Figure 14.4: Profit from a Put Option

For S < 50, the put will be exercised but not the call, leading to a profit of
= 50 + S.
For S > 50, the two calls are exercised, but not the put. The profit becomes
= 2S 100.
Profit is 0 when S = 50.
Portfolios of puts, calls and the underlying asset can be used to engineer
different structures of payoffs. Several of these have been given colorful names
representing their appearance. The first example is the straddle which involves
buying a put and a call on the same stock. If these have the same exercise price,
the profit obtained is
P = max {E S, 0} + max {S E, 0} V0p V0c
(14.6)
The level of profit as a function of the underlying stock price is graphed in Figure
14.5. This strategy is profitable provided the stock price deviates sufficiently
above or below the exercise price.
The strangle is a generalization of the straddle in which a put and call are
purchased that have different exercise prices. Denoting the exercise price of
the put by E p and the that of the call by E c , the profit of the strategy when
E p < E c is shown in Figure 14.6.
Finally, a butterfly spread is a portfolio constructed by purchasing a call with
exercise price E1c and a call with exercise price E3c . In addition, two calls with
204
CHAPTER 14. OPTIONS
E
p
c
0
V0 + V
Figure 14.5: Profit from a Straddle
Ep
V0p + V0c
Ec
S
Figure 14.6: Profit from a Strangle
205
14.4. PUT-CALL PARITY
E3c
E1c
V1c + 2V2c V3c
E2c
Figure 14.7: Profit from a Butterfly Spread

exercise price E2c halfway between E1c and E3c are sold. The profit level is
P = max {S E1c , 0} 2 max {S E2c , 0} + max {S E3c , 0} V1c + 2V2c V3c .
(14.7)
When the underlying stock price is below E1c none of the options is exercise and
a profit is made equal to V1c + 2V2c V3c . For S between E1c and E2c profit
rises. Once E2c has been reached, further increases in S reduce profit until E3c .
Beyond this point, profit is again equal to V1c + 2V2c V3c . This is graphed in
Figure 14.7.
14.4
Put-Call Parity
There is a relationship between the value of a call option and the value of a put
option. In fact, if one value is known, the other can be derived directly. This
relationship is determined by analyzing a particular portfolio of call, put and
the underlying asset.
Consider a portfolio that consists of holding one unit of the underlying asset,
one put option on that asset, and the sale of one call option, with the put and
call having the same exercise price. If V p is the value of the put option and V c
the value of the call, the value of the portfolio, P , is
P = S + V p V c.
(14.8)
P = S + max {E S, 0} max {S E, 0} .
(14.9)
At the expiration date, the final values for the two options can be used to write
the portfolio value as
206
CHAPTER 14. OPTIONS
If S < E at the expiration date, then the put option is exercised but not the
call. The value of the portfolio is
P = S + E S = E.
(14.10)
Conversely, if S > E the call options is exercised but not the put. This gives
the value of the portfolio as
P = S S + E = E.
(14.11)
Hence, whatever the price of the underlying asset at the expiration date, the
value of the portfolio is
P = E,
(14.12)
so the portfolio has the same value whatever happens to the stock price.
Since the value of the portfolio is constant for all S, the portfolio is a safe
asset and must pay the rate of return earned on the risk-free asset. If this return
is r, with continuous compounding the initial value of the portfolio if there are
t units of time until the date of expiry is equal to the discounted value of the
exercise price, so
S + V p V c = Eert .
(14.13)
Therefore, at any time up to the expiration date, if either V p or V c is known,

the other can be derived directly. This relationship is known as put-call parity.
Example 123 A call option on a stock has 9 months to expiry. It currently
trades for $5. If the exercise price is $45 and the current price of the underlying
stock is $40, the value of a put option on the stock with exercise price $45 and
9 months to expiry when the risk-free rate is 5% is
V p = V c S + Eer[T t] = 5 40 + 45e0.050.75 = 8.44.
14.5
Valuing European Options
The problem faced in pricing an option before the expiration date is that we do
not know what the price of the underlying asset will be on the date the option
expires. In order to value an option before expiry it is necessary to add some
additional information. The additional information that we use takes the form
of a model of asset price movements. The model that is chosen will affect the
calculated price of the option so it is necessary work towards a model that is
consistent with the observed behavior of asset prices.
The initial model that is considered makes very specific assumptions upon
how the price of the underlying asset may move. These assumptions may seem
to be too artificial to make the model useful. Ultimately though, they form the
foundation for a very general and widely applied formula for option pricing.
14.5. VALUING EUROPEAN OPTIONS
207
The method of valuation is based on arbitrage arguments. The analysis of

Arbitrage Pricing Theory emphasized the force of applying the idea that two
assets with the same return must trade at the same price to eliminate arbitrage
opportunities. To apply this to the valuation problem the process is to construct
a portfolio, with the option to be valued as one of the assets in the portfolio,
in such a way that the portfolio has the same return as an asset with known
price. In essence, the returns on the portfolio are matched to the returns on
another asset. The portfolio must then trade at the same price as the asset
whose returns it matches. Knowing the prices of all the components of the
portfolio except for the option then implies we can infer the value of the option.
This simple methodology provides exceptionally powerful for valuing options
and will be used repeatedly in what follows.
The analysis given in this section is for European options on an underlying
stock that does not pay any dividends. Dividends can be incorporated using the
same methodology but space limitations prevent this extension being undertaken
here. The valuation of American options requires a development of the analysis
for European options and is analyzed in Section 14.7.
14.5.1
The Basic Binomial Model
To begin the study of option pricing we first consider the very simplest model
for which the valuation problem has any substance. Although simple, solving
this teaches us all we need to know to progress to a very general solution.
Assume that when the option is purchased there is a single period to the
expiration date. No restriction needs to be placed on the length of this period,
as long as the rates of returns are defined appropriately for that period. When
the contract is purchased, the current price of the underlying stock is known.
What we do not know is the price of the underlying at the expiration date. If
we did, we could calculate the profit from the option, discount it back to the
date at which the contract is purchased and determine a precise value. It is this
missing piece of information about the future price of the underlying stock that
we must model. The modelling consists of providing a statistical distribution
for the possible prices at the expiration date.
The fundamental assumption of the basic binomial model is that the price of
the stock may take one of two values at the expiration date. Letting the initial
price of the underlying stock be S, then the binomial assumption is that the
price at the expiration date will either be:
Equal to uS, an outcome which occurs with probability p;
or
Equal to dS, an outcome which occurs with probability 1 p.
The labelling of these two events is chosen so that u > d 0, meaning that
the final price uS is greater than the price dS. Consequently, the occurrence of
the price uS can be called the good or up state and price dS the bad or
208
CHAPTER 14. OPTIONS

Stock Price uS
Option Value Vu
Probability p
Stock Price S
Option Value V0
Risk-free (gross) return R
Probability 1 - p
Stock Price dS
Option Value Vd
Figure 14.8: Binomial Tree for Option Pricing

down state. It can be seen how this model captures the idea that the price of
the underlying stock at expiration is unknown when the option is purchased.
The final component of the model is to assume that a risk-free asset with
return r is also available. Defining the gross return, R, on the risk-free asset by
R 1 + r, it must be case that the return on the risk-free asset satisfies
u > R > d.
(14.14)
This must hold since if R > u the risk-free asset would always provide a higher
return that the underlying stock. Since the stock is risky, this implies that noone would hold the stock. Similarly, if d > R no-one would hold the risk-free
asset. In either of these cases, arbitrage possibilities would arise.
Section 14.3 has already shown how to value options at the expiration
date. For example, if the stock price raises to uS, the value of a call option is
max {uS E, 0} and that of a put option is max {E uS, 0} . For the present,
it is enough to observe that we can calculate the value of the option at the
expiration date given the price of the underlying. The value of the option at
expiration is denoted Vu when the underlying stock price is uS and Vd when it
is dS.
The information that has been described can be summarized in binomial tree
diagram. Consider Figure 14.8. At the left of the diagram is the date the option
is purchased denoted time 0. At this time the underlying stock price is S and
the option has value V0 . It is this value V0 that is to be calculated. The upper
branch of the tree represents the outcome when the underlying price is uS at
expiration and the lower branch when it is dS. We also note the risk-free return
on the tree.
To use this model for valuation, note that there are three assets available: (1)
209
the underlying stock; (2) the option; and (3) the risk-free asset. Constructing
a portfolio of any two of these assets which has the same return as the third
allows the application of the arbitrage argument.
Consequently, consider a portfolio that consists of one option and units
of the underlying stock. The number of units of the underlying stock is chosen
so that this portfolio has the same value when the underlying has price uS at
the expiration date as it does when it has price dS. This then allows us to apply
the arbitrage argument since the portfolio has a fixed value and so must pay
the same return as the risk-free asset. The portfolio constructed in this way is
often referred to as the delta hedge for the option.
The cost of this portfolio at the date the option contract is purchased is
P0 = V0 S,
(14.15)
where V0 is the unknown which is to be determined. At the expiration date the

value of the portfolio is either
Pu = Vu uS,
(14.16)
Pd = Vd dS.
(14.17)
or
The value of is chosen to ensure a constant value for the portfolio at the
expiration date. Hence must satisfy
Vu uS = Vd dS,
(14.18)
giving
=
Vu Vd
.
S [u d]
(14.19)
Substituting this value of back into (14.16) and (14.17),

Pu = Pd =
uVd dVu
,
ud
(14.20)
so it does give a constant value as required.

The arbitrage argument can now be applied. The portfolio of one option
and units of the underlying stock provides a constant return. Therefore
it is equivalent to holding a risk-free asset. Given this, it must pay the same
return as the risk-free or else one could be arbitraged against the other. Hence
the gross return on the portfolio must be R which implies
Pu = Pd = RP0 .
(14.21)
Now substituting for P0 and Pu gives

uVd dVu
= R [V0 S] .
ud
(14.22)
210
CHAPTER 14. OPTIONS
Using the solution for and then solving for V0

1 Rd
uR
V0 =
Vu +
Vd .
R ud
ud
(14.23)
This result gives the fair value for the option that eliminates arbitrage opportunities. In an efficient market, this would be the premium charged for the
option.
The valuation formula is defined for general values of Vu and Vd . What
distinguishes calls and puts are the specific forms that these values take. These
can be called the boundary values. These boundary values were calculated in
Section 14.3.
Example 124 For a call option with exercise price be E, the value of the option
at the expiration date is either Vu = max {uS E, 0} or Vd = max {dS E, 0} .
The initial value of the call option is then

1 Rd
uR
max {uS E, 0} +
max {dS E, 0} .
V0 =
R ud
ud
Example 125 Consider a call option with exercise price $50 written on a stock
with initial price $40. The price of the underlying stock may rise to $60 or to
$45 and the gross return on the risk-free asset is 115%. These values imply
u = 1.5, d = 1.125 and R = 1.15. The value of the option at the expiration date is
either Vu = max {uS E, 0} = max {60 50, 0} = 10 or Vd = max {dS E, 0}
= max {45 50, 0} = 0. The initial value of the call option is then

0.35
0.025
1
10 +
0 = $0.58.
V0 =
1.15 0.375
0.375
Example 126 For a put option with exercise price E, the value of the option
at the expiration date is either Vu = max {E uS, 0} or Vd = max {E dS, 0} .
The initial value of the put option is then

1 Rd
uR
max {E uS, 0} +
max {E dS, 0} .
V0 =
R ud
ud
Example 127 Consider a put option with exercise price $50 written on a stock
with initial price $40. The price of the underlying stock may rise to $60 or
to $45 and the gross return on the risk-free asset is 115%. These values imply
u = 1.5, d = 1.125 and R = 1.15. The value of the option at the expiration date is
either Vu = max {E uS, 0} = max {50 60, 0} = 0 or Vd = max {E dS, 0}
= max {50 45, 0} = 5. The initial value of the put option is then
V0 =

0.025
0.35
1
0+
5 = $4.06.
1.15 0.375
0.375
211
The valuation formula we have constructed can be taken in two ways. On

one level, it is possible to just accept it, and the more general variants that
follow, as a means of calculating the value of an option. Without developing
any further understanding they can be used to provide fair values for options
that can then be applied in investment analysis. At a second level, the structure
of the formula can be investigated to understand why it comes out the way it
does and what the individual terms mean. Doing so provides a general method
of valuation that can be applied to all valuation problems.
To proceed with the second approach, observe that the weights applied to
Vu and Vd in (14.23) satisfy
uR
Rd
> 0,
> 0,
ud
ud
(14.24)
and
uR Rd
+
= 1.
(14.25)
ud
ud
Since both weights are positive and their sum is equal to 1, they
have the
Rd
basic features of probabilities. To emphasize this, define q ud . Then the
valuation formula can be written as
V0 =
1
[qVu + [1 q] Vd ] .
R
(14.26)
In this expression, the value V0 is found by calculating the expected value at

expiration and discounting back to the initial date using the risk-free rate of
return. This shows that the value can be written in short form as
V0 =
1
Eq (V ) ,
R
(14.27)
where the subscript on the expectation operator indicates that the expectation
is taken with respect to the probabilities {q, 1 q}.
The idea that we value something by finding its expected value in the future
and then discount this back to the present is immediately appealing. This is
exactly how we would operate if we were risk-neutral. However, the assumption
in models of finance is that the market is on average risk-averse so that we
cannot find values this simply. How this is captured in the valuation formula
(14.27) is that the expectation is formed with the probabilities {q, 1 q} which
we have constructed not the true probabilities {p, 1 p} . In fact, the deviation
of {q, 1 q} from {p, 1 p} captures the average risk aversion in the market. For
this reason, the probabilities {q, 1 q} are known as a risk-neutral probabilities
they modify the probabilities so that we can value as if we were risk-neutral.
This leaves open two questions. Firstly, where do the true probabilities
{p, 1 p} feature in the analysis? So far it does not appear that they do. The
answer to this question is that the true probabilities are responsible for determining the price of the underlying stock. Observe that the price of the underlying
stock when the option is purchased must be determined by its expected future
212
CHAPTER 14. OPTIONS
payoffs. Hence, S is determined from uS and dS by a combination of the probabilities of the outcomes occurring, {p, 1 p}, discounting, R, and the attitude
to risk of the market. The true probability may be hidden, but it is there.
Secondly, are these risk-neutral probabilities unique to the option to be valued? The answer to this question is a resounding no. When risk-neutral probabilities can be found they can be used to value all assets. In this analysis there
are only three assets but all can be valued by using the risk neutral probabilities. Consider the underlying stock. For this asset, Vu = uS and Vd = dS.
Using these in the valuation formula

1
1 Rd
uR
uS +
dS = S.
(14.28)
V0 = [qVu + [1 q] Vd ] =
R
R ud
ud
Hence the risk neutral probabilities also value the underlying stock correctly.
For the risk-free asset

uR
1
1 Rd
R+
R = 1.
(14.29)
V0 = [qVu + [1 q] Vd ] =
R
R ud
ud
This process of calculating he expected value of returns using the risk-neutral
probabilities and then discounting back to the present using the risk-free rate of
return is therefore a general valuation method that can be applied to all assets.
Example 128 Consider a call option with exercise price $50 written on a stock
with initial price $50. The price of the underlying stock may rise to $60 or fall
to $45 and the gross return on the risk-free asset is 110%. The risk-neutral
probabilities are given by
q=
Rd
1.1 0.9
2
=
= ,
ud
1.2 0.9
3
and
1q =
1.2 1.1
1
uR
=
= .
ud
1.2 0.9
3
(14.30)
(14.31)
The initial value of the call option is then

1
1
1 2
V0 = Eq (V ) =
10 + 0 = $6.06.
R
1.1 3
3
In addition, the price of the underlying stock must satisfy

1 2
1
V0 =
60 + 45 = $50.
1.1 3
3
14.5.2
The Two-Period Binomial
The single-period binomial model introduced a methodology for valuing options

but does not represent a very credible scenario. Where it is lacking is that the
213
Stock price uS
Option value Vu
Probability q
Stock price S
Option value V0
Stock price uuS

Option value Vuu
Probability q2
Stock price udS

Option value Vud
Probability 2q[1 q]
Stock price dS
Option value Vd
Probability 1 - q
Stock price ddS

Option value Vdd
Probability [1 q]2
Figure 14.9: Binomial with Two Sub-Intervals

underlying stock will have more than two possible final prices. Having introduced the method of risk-neutral valuation, the task of relaxing this restriction
and moving to a more convincing environment is not at all difficult.
A wider range of final prices can be obtained by breaking the time period
between purchase of the option and the expiration date into smaller sub-intervals
and allowing the stock price to undergo a change over each sub-interval. As long
as the rate of return for the risk-free asset and the proportional changes in the
stock price are defined relative to the length of each sub-interval, the use of
risk-neutral valuation can be directly extended to this setting.
Consider Figure 14.9 which shows the period between purchase and expiration broken into two sub-intervals. Starting with an underlying stock price
of S, at the end of the first sub-interval the price will either be uS or dS. In
terms of the risk-neutral probabilities, these will occur with probabilities q and
1 q respectively. Starting from the price uS, it is possible to reach a final price
at the end of the second interval of either uuS or udS. Since the probability
of another u is q and of a d is 1 q, these final prices must have probabilities
q 2 and q [1 q] respectively. Similarly, starting from dS, another d occurs with
probability 1 q and a u with probability q. Hence the final price ddS is reached
2
with probability [1 q] and duS with probability [1 q] q. But udS = duS, so
the central price at the expiration date can be reached in two different ways
with total probability of arrival given by 2q [1 q] . The risk-free (gross) return,
R, is defined as the return over each sub-interval. Hence the return over the
period is R2 . This completes the construction of the figure.
The value of the option V0 can be obtained in two different ways. The first
way is to use a two-step procedure which employs risk-neutral valuation to find
214
CHAPTER 14. OPTIONS
Vu and Vd using the expiration values, and then uses these to find V0 . Although
not strictly necessary for a European option, it is worth working through these
two steps since this method is necessary when American options are valued.
The second way to value the option is to apply risk-neutral valuation directly
to the expiration values using the compound probabilities. Both give the same
answer.
To apply the two-step procedure, assume we are at the end of the first subinterval. The price of the underlying stock is either uS or dS. If it is uS, then
applying (14.26) the value of the option must be
Vu =
1
[qVuu + [1 q] Vud ] .
R
(14.32)
Similarly, if the price of the underlying stock at the end of the first sub-interval
is dS, the value of the option is
Vd =
1
[qVud + [1 q] Vdd ] .
R
(14.33)
Now move to the very beginning of the tree. At the end of the first sub-interval
the option is worth either Vu or Vd . Applying risk-neutral valuation, the value
of the option at the purchase date must be
Vd =
1
[qVu + [1 q] Vd ] .
R
(14.34)
Substituting into this expression using (14.32) and (14.33) gives

V0 =

1 2
2
V
+
2q
[1
q]
V
+
[1
q]
V
q
.
uu
ud
dd
R2
(14.35)
This is the fair value of the option at the purchase date. It should also be clear
that this is the result that would have been obtained by applying risk-neutral
valuation directly to the values at the expiration date using the risk-neutral
probabilities given in the binomial tree.
Example 129 For a call option with exercise price E,
"
#
Vuu = max u2 S E, 0 ,
Vud = max {udS E, 0} ,
"
#
Vdd = max d2 S E, 0 .
The value of the call is
V0
=
=

1 2
2
q Vuu + 2q [1 q] Vud + [1 q] Vdd
2
R
"
#

1
[1 q] max {udS
E,
0}
q 2 max u2 S E, 0 + 2q "
#
.
+ [1 q]2 max d2 S E, 0
R2
215
Example 130 Consider a put option with a year to expiry on a stock with
initial price of $50. Over a six month interval the stock can rise by 15% or by
5% and the risk-free rate of return is 107.5%. If the put option has an exercise
price of $65 the value of the contract is
2

32
1
1
13
4.625
+
0
+
9.875
= $5.557.
V0 =
2
44
4
[1.075] 4
14.5.3
The General Binomial
The process of either working back through the tree or applying risk-neutral
valuation directly to the expiration values can be applied to a binomial tree
with any number of sub-intervals. A general variant of the binomial formula
is now obtained that applies whatever number of sub-intervals the period is
divided into.
To derive this, note that in (14.35) the occurrence of a q in the expression
matches the occurrence of a u (and a 1 q matches a d). Furthermore, the
coefficients on the values at expiration are 1, 1 for the one-interval case and 1,
2, 1 for the two-interval case. These are the terms in the standard binomial
expansion. Using these observations, the valuation formula for a period divided
into n sub-intervals can be immediately derived as

n
1
n!
(14.36)
q j [1 q]nj Vuj dnj .
V0 = n
R
j!
[n
j]!
j=0
It is easy to check that for n = 1 and n = 2 this gives the results already derived
directly.
Example 131 When n = 4 the valuation formula is
4

1
q Vu4 + 4q 3 [1 q] Vu3 d + 6q 2 [1 q]2 Vu2 d2
V0 = 4
.
R
+4q [1 q]3 Vud3 + [1 q]4 Vd4
Although the result as given is time consuming to use when computing

results manually, it is easy to write a program that will compute it automatically.
Even when n is large, it will only take seconds to obtain an answer. The
valuation formula is therefore perfectly usable and the next sub-section shows
that it can reflect the actual data on stock price movements. However, it can
be improved further by specifying the details of the final valuations.
Consider a call option. In this case
"
#
(14.37)
Vuj dnj = max uj dnj S E, 0 .
Now define a as the smallest non-negative integer for which

ua dna S E > 0.
(14.38)
216
CHAPTER 14. OPTIONS
Hence it requires a minimum of a up moves to ensure that

" the option will
# be
in the money at expiry."Consequently, if#j < a then max uj dnj S E, 0 = 0
and if j > a then max uj dnj S E, 0 = uj dnj S E. With this definition
of a it is only necessary to include the summation in (14.36) the terms for j a,
since all those for lower values of a are zero.
Example 132 Let u = 1.05, d = 1.025, S = 20, E = 24 and n = 5. The values
of uj dnj S are given in the table.
u0 d5 S
22.63
u1 d4 S
23.18
u2 d3 S
23.75
u3 d2 S
24.32
u4 d1 S
24.92
u5 d0 S
25.53
Example 133 It can be seen that uj dnj S exceeds the exercise price E of 24
only when j 3. Hence a = 3.
Using the definition of a to remove from the summation those outcomes for
which the option is worthless at expiry, the value of the call becomes

n

1
n!
(14.39)
V0 = n
q j [1 q]nj uj dnj S E .
R j=a j! [n j]!
Separating this expression into terms in S and terms in E,
j nj

n

n!
u d
V0 = S
q j [1 q]nj
j!
[n
j]!
Rn
j=a

n

n!
ERn
q j [1 q]nj .
j!
[n
j]!
j=a
Now define
u
d
q, 1 q = [1 q] ,
R
R

n

n!
(a; n, q)
q j [1 q]nj ,
j!
[n
j]!
j=a
q =
and let
and
n

(a; n, q )
j=a

nj
n!
uq j d [1 q]
.
j! [n j]!
R
R
(14.40)
(14.41)
(14.42)
(14.43)
(14.44)
(a; n, q) (and equivalently (a; n, q )) is the complementary binomial distribution function which gives the probability that the sum of n random variables,
each with value 0 with probability q and value 1 with probability 1 q will be
greater than or equal to a. Because they are probabilities, both (a; n, q) and
(a; n, q ) must lie in the range 0 to 1.
217
Using this notation, the valuation formula can be written in the compact
form
(14.45)
V0 = (a; n, q ) S ERn (a; n, q) .
The value of the option is therefore a combination of the underlying stock price
and the discounted value of the exercise price with each weighted by a probability. This is an exceptionally simple formula.
14.5.4
Matching to Data
The next question to be addressed is how to make the formula in (14.45) into
a result that can be applied in a practical context. To evaluate the formula we
need to supply values for S, E, R, n and q. The underlying stock price S and the
risk-free return R can be obtained directly from market data. The exercise price
E is written into the option contract. The number of intervals, n, is chosen to
trade-off accuracy against ease of computation. All that is unknown is q, the
probability in the binomial tree.
To motivate the approach taken to providing a value for q, recollect that
the basic idea of the binomial tree is that the price of the underlying stock is
random. Given a value of R, the value of q is determined by u and d. The values
of u and d must be chosen to result in behavior of the underlying stock price
that mirrors that observed in the market place. This leads to the idea of fixing
u and d to provide a return and variance of the underlying stock price in the
binomial model that equals the observed variance of the stock in market data.
Let the observed expected return on the stock be r and its variance be 2 .
Each of these is defined over the standard period of time. If the time length of
each interval in the binomial tree is t, the expected return and variance on
the stock over an interval are rt and 2 t. If at the start of an interval the
stock price is S, the expected price at the end of the interval using the observed
return is Sert . Matching this to the expected price in the binomial model gives
puS + [1 p] dS = Sert ,
(14.46)
where it should be noted that these are the probabilities of the movements in
the statistical model, not the risk-neutral probabilities. Solving this shows that
to match the data
ert d
p=
.
(14.47)
ud
Over an interval in the binomial tree, the return on the underlying stock is u 1
with probability p and d 1 with probability 1 p. The expected return is
therefore pu + [1 p] d 1. The variance in the binomial model, sb , can then
2
be calculated as sb = pu2 + [1 p] d2 [pu + [1 p] d] . Equating this variance
is to the observed market variance gives
pu2 + [1 p] d2 [pu + [1 p] d]2 = 2 t.
(14.48)
218
CHAPTER 14. OPTIONS
Substituting for p from(14.47), ignoring terms involving powers of t2 and

higher, a solution of the resulting equation is
u = e
d=e
(14.49)
(14.50)
These values can then be used to parameterize the binomial model to match
observed market data.
Example 134 The data in Example 36 generate an annual variance of 523.4%
for General Motors stock. If the year is broken into 365 intervals of 1 day each,
1
then t = 365
= 0.00274 and = 22.88. Hence
u = e
and
d = e
These imply
q=
14.6
0.00274
= e22.88
= 3.3,
0.00274
= e22.88
= 0.3.
ert d
e6.50.00274 0.3
=
= 0.239.
ud
3.3 0.3
(14.51)
(14.52)
(14.53)
Black-Scholes Formula
In moving from the single-interval binomial to the general binomial the process
used was to reduce the interval between successive price changes. Continuing to
shorten the interval eventually leads to a situation where one price change follows
another without any time seeming to have passed. In the limit, we can then
think of price changes occurring continuously, rather than at discrete intervals
as in the binomial. Such continuity comes close to capturing the observation
that for most significant stocks a very large number of trades take place so the
actual process is almost continuous.
Taking the limit of the binomial model as the interval between trades shrinks
to zero leads to the Black-Scholes equation. The Black-Scholes equation is one
of the most fundamental results in investment analysis. Its value comes from
the fact that it provides an easily applied practical solution to the problem
of pricing options that can be evaluated using observable market data. The
construction of the equation revolutionized the way option markets functioned
since it provided an exact and easily computable fair value for an option.
The move from discrete intervals in the binomial model to continuous time
for Black-Scholes leads to two changes to the valuation formula (14.45). The first
is very simple: the discrete compounding captured in the term Rn becomes
the continuous analog erT where T is the time until the option expires and r
is the risk-free interest rate for a compatible time period. For example, if the
option has 9 months to expiry and r is the annual risk-free rate then T is defined
as written as a fraction of a year, in this case T = 0.75.
14.6. BLACK-SCHOLES FORMULA
219
The second change relates to the probabilities. In the general binomial

formula S and E are weighted by values from complementary binomial distributions. In the limit as the length of the time intervals shrink to zero, these
distributions converge to the normal distribution and the weights become values
from the cumulative function for the normal distribution. Being the cumulative
of the normal distribution, both weights are again between 0 and 1.
Collecting these points together, the Black-Scholes equation for the value of
a call option is given by
V c = N (d1 ) S EerT N (d2 ) ,
(14.54)
where N (d1 ) and N (d2 ) are values from the cumulative normal distribution and

ln (S/E) + r + 0.52 T
d1 =
,
(14.55)
T

ln (S/E) + r 0.52 T
d2 =
.
(14.56)
T
Recalling the discussion of applying the general binomial formula, S, E, r,
T can be directly observed and calculated from observed market data. Given
these values, the formula is applied by computing d1 and d2 then determining
N (d1 ) and N (d2 ) from statistical tables for the cumulative normal a table is
contained in the appendix. The formula is then evaluated.
Example 135 A call option with an exercise price of $40 has three months to
expiry. The risk-free interest rate is 5% per year and the stock price is currently
$36. If the standard deviation of the asset price is 0.5, then T = 0.25, E = 40,
S = 36, = 0.5 and r = 0.05. The formulas for the call option give

ln (36/40) + 0.05 + 0.5(0.5)2 0.25
d1 =
= 0.25,
0.5 0.25
and
ln (36/40) + 0.05 0.5(0.5)2 0.25
= 0.5.
d2 =
0.5 0.25
From the tables for the cumulative normal distribution
N (d1 ) = 0.4013, N (d2 ) = 0.3083.
Substituting into the Black-Scholes formula

40
c
V = [0.4013 36] 0.050.25 0.3085 = $2.26.
e
The Black-Scholes formula for the value for a put of option is
V p = N (d2 ) EerT N (d1 ) S,
where the definitions of d1 and d2 are as for a call option.
(14.57)
220
CHAPTER 14. OPTIONS
Example 136 If T = 0.25, E = 40, S = 36, = 0.5 and r = 0.05 then

d1 = 0.25 and d2 = 0.5. From the cumulative normal tables
N (d1 ) = N (0.25) = 0.5987,
and
N (d2 ) = N (0.5) = 0.695.
This gives the value of the put as

40
p
V = 0.695 0.050.25 [0.5987 36] = $5.90.
e
14.7
American Options
The analysis of European options is much simplified by the fact that they can
only be exercised at the expiration date. The fact that American options can be
exercised at any time up until the date of expiry adds an additional dimension to
the analysis. It now becomes necessary to determine the best time to exercise.
The best way to analyze this is to return to the two-interval binomial tree
displayed in Figure 14.9. The two-interval model provides a time after the
first price change at which the issue of early exercise can be addressed. With
American options it is also necessary to treat calls and puts separately.
14.7.1
Call Options
Assume that a call option is being analyzed and that the first price change has
lead to a price of uS for the underlying stock. The holder of the option then
has three choices open to them:
Exercise the option and obtain max {uS E, 0} ;
c
c
Hold the option and receive either Vuu
or Vud
depending on the next price
change;
Sell the option for its value Vuc .

Whether the option should be exercised depends on which of these three
alternatives leads to the highest return. First consider holding the option. The
payoff of this strategy can be evaluated by employing risk-neutral valuation.
c
c
or Vud
is
Hence the value of receiving either Vuu
Vuc =
1
c
c
[qVuu
+ [1 q] Vud
],
R
(14.58)
but this is precisely the fair market value of the option. The value of holding
the option is therefore the same as that of selling (though there is risk involved
in the former). Now compare exercising to selling. If the option is sold, Vuc is
221
14.7. AMERICAN OPTIONS
realized. If it is exercised, uS E is realized - there is no point exercising the

option if uS E < 0. By definition
"
#
c
Vuu
= max u2 S E, 0 u2 S E,
(14.59)
and
c
= max {udS E, 0} udS E.
Vud
(14.60)
Using risk-neutral valuation and the inequalities in (14.59) and (14.60)

Vuc
But
1
c
c
+ [1 q] Vud
]
[qVuu
R

1
2
qu S + [1 q] udS E .
R

1
2
E
u [qu + [1 q] d] S
qu S + [1 q] udS E =
R
R
R
> uS E,
(14.61)
(14.62)
where the last inequality follows from the fact that qu+[1 q] d = R and R 1.
Combining these statements
Vuc > uS E,
(14.63)
which shows that the option should never be exercised early. It is always better
to hold or to sell than to exercise.
Similarly, if after the first interval the price is dS, the choice of strategies is:
Exercise and obtain max {dS E, 0} ;
c
c
or Vdd
depending on the next price
Hold the option and receive either Vud
change;
Sell the option for its value Vdc .

Applying risk-neutral valuation shows that the second and third provide the
payoff Vdc . Noting that
"
#
c
= max d2 S E, 0 d2 S E,
(14.64)
Vdd
then
Vdc
1
c
c
+ [1 q] Vdd
]
[qVdu
R

1
qudS + [1 q] d2 S E ,
R
d [qu + [1 q] d] S E
=
R
R
> dS E.
(14.65)
222
CHAPTER 14. OPTIONS
Stock price uS
Option value Vup
Stock price S
p
Option value V0
Stock price uuS

Option value Vuup
Stock price udS

p
Option valueVud
Stock price dS
p
Option value Vd
Stock price ddS
p
Option valueVdd
Figure 14.10: American Put Option

Hence the conclusion obtained is that
Vdc > dS E,
(14.66)
so that it is better to hold or sell than to exercise.

These calculations illustrate the maxim that an option is Better alive than
dead, revealing that an American call option will never be exercised early. It
is always better to hold or to sell than exercise. Even though the options have
the feature of early exercise, if they are priced correctly this should never be
done.
14.7.2
Put Option
The same conclusion cannot be obtained for a put option. In this case it may
be better to exercise.
The two-interval binomial tree for an American put option is illustrated in
Figure 14.10. Consider being at the end of the first interval and observing a stock
price of dS. The value of the put option at this point is Vdp and early exercise
would obtain the amount E dS. The option should therefore be exercised early
if E dS > Vdp . Where an American put differs from an American call is that
this can hold in some circumstances and early exercise becomes worthwhile.
This can be seen by using the expiration values and the risk-neutral probabilities to obtain
1
p
p
+ [1 q] Vdd
].
(14.67)
Vdp = [qVdu
R
p
p
Numerous possibilities now arise depending upon whether Vdu and Vdd are positive or zero. That early exercise can be optimal is most easily demonstrated if
223
14.7. AMERICAN OPTIONS
Vuup
max E uS ,Vup
V0p
Vudp
max E dS , Vdp
}
V ddp
Figure 14.11: Value of American Put

p
p
= E udS and Vdd
= E ddS.
both are taken to be positive. In this case, Vdu
Then early exercise will be optimal if
E dS >
1
[q [E udS] + [1 q] [E ddS]] .
R
(14.68)
Substituting for q and 1 q then solving shows that the inequality in (14.68)
holds if
R > 1.
(14.69)
Therefore, the put option will be exercised early if the return on the risk-free
asset is positive.
A similar analysis can be undertaken to investigate the numerous other possibilities. But the important conclusion is that it is sometimes optimal to exercise
American puts early. So their value must be higher than for a European put.
The actual method of valuation of an American put option is to construct the
binomial tree and to assign the value of the option at each node as the maximum
of the early exercise value and the fair value of the option. This is shown in
Figure 14.11 which indicates the value at each node incorporating the option
for early exercise.
Example 137 Consider a two-interval binomial tree with R = 1.05, u = 1.1, d =
1 and an initial stock price of $10.
224
CHAPTER 14. OPTIONS
A European put option contract with exercise price $12 is worth

1 2 p
2 p
p
V
+
2q
[1
q]
V
+
[1
q]
V
q
V0p =
uu
ud
dd
R2
1
[0.25 0 + 0.5 1 + 0.25 2]
=
1.052
= $0.907.
An American put on the same stock has value
V0p =
1
[q max {E uS, Vup } + [1 q] max {E dS, Vdp }] .
R
Working back from the end of the binomial tree,

Vup
1
p
p
+ [1 q] Vud
]
[qVuu
R
1
=
[0.5 0 + 0.5 1]
1.05
= $0.476,
Vdp
1
p
p
+ [1 q] Vdd
]
[qVud
R
1
[0.5 1 + 0.5 2]
=
1.05
= $1.429.
and
=
Therefore max {E uS, Vup } = max {12 11, 0.476} = 1 (so the option is exercised early) and max {E dS, Vdp } = max {12 10, 1.429} = 2 (so the option
is exercised early). The initial value of the option is then
V0p
1
[0.5 1 + 0.5 2]
1.05
= $1.429.
=
As claimed, if it is optimal to exercise early, the American option has a higher

value than the European option.
14.8
Summary
The chapter has described call and put options, distinguishing between European and American contracts. Information on where these options can be traded
and where price information can be found has also been given.
The process of valuing these options began with a determination of the
value of the options at the expiration date. From these results the profit from
portfolios of options was determined. In particular, this process was used to
derive put-call parity.
14.9. EXERCISES
225
It was then noted that to provide a value before the expiration date it was
necessary to model the statistical distribution of future prices of the underlying
stock. European options were valued using the single-period binomial model.
The model was then gradually generalized, eventually resulting in the BlackScholes formula.
American options were then considered. It was shown that an American call
would never be exercised early but a put may be. American calls therefore have
the same value as European calls. American puts will be at least as valuable as
European puts and may be strictly more valuable.
14.9
Exercises
Exercise 95 Consider two call options on the same underlying stock. Option
1 has an exercise price of $60 and sells for $5 while option 2 has an exercise
price of $55 and sells for $6. Assuming they have the same expiration date,
calculate the profit from the strategy of issuing two $60 calls and purchasing one
$55. Sketch the level of profit versus the share price at the expiration date.
Exercise 96 If a call option on a stock trading at $40 has an exercise price
of $45 and a premium of $2, determine the premium on a put option with the
same exercise price if the annual risk-free rate of return is 5% and there is 6
months to expiration.
Exercise 97 Using the binomial pricing model calculate the value of a call option on a stock that currently sells for $100 but may rise to $115 or fall to $80
when there is 1 year to expiry, the risk free rate of return is 5% and the exercise price is $105. Repeat this exercise breaking the year in (i) two six month
intervals and (ii) three four month intervals but retaining $115 and $80 as the
maximum and minimum prices that can be reached.
Exercise 98 Prove that (14.49) and (14.50) are a solution to the equation relating observed market variance to the variance in the binomial model.
Exercise 99 Taking the prices from Yahoo, find the sample variance for Ford
stock (ten years of data) and hence compute u and d for a daily sub-interval.
Exercise 100 Determine the value of a call option with 9 months to go before
expiration when the stock currently sells for $95, has an instantaneous standard
deviation of 0.8, the exercise price is $100 and the continuously compounded
risk-free rate of return in 6%.
Exercise 101 Consider a stock that currently trades for $75. A put and call on
this stock both have an exercise price of $70 and expire in 150 days. If the riskfree rate is 9 percent and the standard deviation for the stock is 0.35, compute
the price of the options using Black-Scholes.
Exercise 102 Show that the values given for put and call options satisfy putcall parity.
226
CHAPTER 14. OPTIONS
Exercise 103 Consider a two-interval binomial tree with S = 20, E = 22, u =

1.05, d = 1.025 and R = 1.05. By applying the two-step procedure to work back
through the tree, show that an American call option on the stock will never be
exercised early.
Exercise 104 For the data in Exercise 103, determine at which points in the
tree the put will be exercised early.
Chapter 15
Forwards and Futures

Please excuse the facts the notes are not perfect but all the material you need is here.
15.1
Introduction
Forwards and futures are both contracts which involve the delivery of a specific
asset at an agreed date in the future at a fixed price. They differ from options
contracts in the fact that there is no choice involved as to whether the contract
is exercised. With both forwards and futures the agreed price must be paid
and delivery undertaken. Despite this, the underlying approach to valuation
remains the same.
Forward contracts, which are no more than commitments to a future trade,
have been in use for a very long time. One piece of evidence to this effect is that
the agreement to purchase dates whilst the dates were still unripe on the tree (a
forward contract) was prohibited in the early Islamic period. Commodity futures
also have a fairly long history. They were first introduced onto an exchange by
the Chicago Board of Trade in the 1860s to assist with the reduction in trading
risk for the agriculatural industry. Financial futures (which differ in significant
ways from commodity futures) are a much more recent innovation.
This chapter will introduce the main features of forward and future contracts
and describe where they can be traded. The motives for trading and potential
trading strategies will be analyzed. Finally, the valuation of the contracts will
be considered.
15.2
Forwards and Futures
Forwards and futures are two variants of the same basic transaction but there
are some important operational differences between them. These differences are
reflected in the valuations of the contracts. The forward contract is the simpler
form and this is described first.
227
228
CHAPTER 15. FORWARDS AND FUTURES
As an example of a forward contract consider a farmer growing wheat and

a baker who requires wheat as an ingredient. Assume that wheat is harvested
in September. A forward contract would be written if the farmer and the baker
committed in May to the baker purchasing 2 tons of wheat at $1000 per ton
when the wheat is harvested in September. The essential elements here are
the commitment to trade at a future date for a fixed price and quantity. No
money is exchanged when the forward contract is agreed. Money only changes
hands when the commodity is delivered. The financial question that arises is
the determination of the price (in this example $1000) written into the contract.
A futures contract has the almost all the features of forward contract. In
a futures contract there is also a commitment to trade and agreed quantity at
a fixed price at a future date. Where differences arise between forwards and
futures is in the timing of institutioanl arrangement and the timing of payment.
A forward is an over-the-counter agreement between two individuals. In
contrast, a future is a trade organized by an exchange.
A forward is settled on the delivery date. That is, there is a single payment
made when the contract is delivered. The profit or loss on a future is
settled on a daily basis.
To understand the process of daily settlement, assume a futures contract is
agreed for delivery of a commodity in three months. Label the day the contract
is agreed as day 1 and the day of delivery as day 90. Let the delivery price
written into the contract on day 1 be $30. Now assume that on day 2 new
contracts for delivery on day 90 have a delivery price of $28 written into them.
Those who are holding contracts with an agreement to pay $30 are in a worse
position than than those holding $28 contracts. The daily settlement process
requires them to pay $2 (the value by which their position has deteriorated) to
those who have sold the contract. The delivery price of $28 on day 2 is then
taken as the starting point for day 3. If the delivery price in new contracts rises
to $29 on day 3 then the holder of the futures contract from day 2 receives $1.
This process is repeated every day until day 90. Effectively, daily settlement
involves the futures contract being re-written each day with a new contract
price.
From this brief description, it can be seen that a futures contract involves a
continuous flow of payments over the life of the contract. In contrast, a forward
contract has a single payment at the end of the contract. This difference in the
timing of payments implies that the contracts need not have the same financial
valuation.
With a futures contract the exchange acts as an intermediary between the
two parties on different sides of the contract. The process of daily settlement
is designed to avoid the development of excessive negative positions and the
possibility of default. To further reduce the chance of default exchanges insist
upon the maintenance of margin. Margin must be maintained by both parties
to a sufficient level to cover daily price changes.
15.3. FUTURES
229
The next section of the chapter will focus upon the trading details of futures
contracts because these are the contracts that can be most readily traded. The
focus will then shift to forward contracts when valuation is considered. The
reason for focussing on forward contracts is that the single payment involved
makes valuation a very much simpler process. Finally a contrast will be drawn
between the valuation of a forward contract and the valuation of a futures
contract.
15.3
Futures
There are two basic types of futures contracts. These are commodity futures
and financial futures.
15.3.1
Commodity Futures
Commodity futures are trades in actual commodities. Many significant agricultural products are covered by futures contracts including wheat, pork and
orange juice plus other commodities such as timber. Futures contracts originated in an organized way with the Chicago Board of Trade and have since
been offered by numerous other exchanges.
Example 138 The Chicago Board of Trade was established in 1848. It has
more than 3,600 members who trade 50 different futures and options products
through open auction and/or electronically. Volume at the exchange in 2003
was 454 million contracts. Initially, only agricultural commodities such as corn,
wheat, oats and soybeans were traded. Futures contracts have developed to include non-storable agricultural commodities and non-agricultural products such
as gold and silver. The first financial futures contract was launched in October
1975 based on Government National Mortgage Association mortgage-backed certificates. Since then further futures, including U.S. Treasury bonds and notes,
stock indexes have been introduced. Options on futures were introduced in 1982.
(http://www.cbot.com/cbot/pub/page/0,3181,1215,00.html)
A contract with the Board of Trade, which is similar in structure to contracts
on other exchanges, specifies:
The quality of the product. The quality has to be very carefully defined
so that the parties to the contract know exactly what will be traded. This
is important when there are many different varieties and qualities of the
same product.
The quantity of a trade. The quantity that is traded is specified in the
contract. This is usually large in order to make delivery an economically
viable exercise. However, it does mean that these contracts are lumpy
so that the assumption of divisibility is not easily applied.
230
The place to which delivery is made. The importance of this is through

the high transport costs that can be involved in shipping the commodities
around.
The date of delivery (or interval in which delivery is to be made). This is
essential for the contract to function.
The price. This is the basic feature of the contract upon which profit and
loss is determined. The price is what will be paid at the delivery time.
These specifications have to be very precise and complete in order to ensure
that there can be no dispute about whether the correct product is ultimately
delivered.
Example 139 Soybeans Futures. 1. Contract Size 5,000 bu. 2. Deliverable
Grades No. 2 Yellow at par, No. 1 yellow at 6 cents per bushel over contract
price and No. 3 yellow at 6 cents per bushel under contract price *No. 3 Yellow
Soybeans are only deliverable when all factors equal U.S. No. 2 or better except
foreign material. See Chapter 10s - Soybean Futures in the Rules & Regulations
section. 3. Tick Size 1/4 cent/bu ($12.50/contract) 4. Price Quote Cents
and quarter-cents/bu 5. Contract Months Sep, Nov, Jan, Mar, May, Jul, Aug
6. Last Trading Day The business day prior to the 15th calendar day of the
contract month. 7. Last Delivery Day Second business day following the last
trading day of the delivery month. 8. Trading Hours Open Auction: 9:30 a.m.
- 1:15 p.m. Central Time, Mon-Fri. Electronic: 7:31 p.m. - 6:00 a.m. Central
Time, Sun.-Fri. Trading in expiring contracts closes at noon on the last trading
day. 9. Ticker Symbols Open Auction: S, Electronic: ZS 10. Daily Price Limit:
50 cents/bu ($2,500/contract) above or below the previous days settlement price.
No limit in the spot month (limits are lifted two business days before the spot
month begins).
(http://www.cbot.com/cbot/pub/cont_detail/0,3206,959+14397,00.html)
Although the contracts specify delivery of a commodity, most contracts are
closed before the delivery date. Less then 1% are delivered or settled in cash.
15.3.2
Financial Futures
Financial futures are contracts drawn up on the basis of some future price or
index such as the interest rate or a stock index. Generally, no good is delivered
at the completion of the contract and only a financial exchange takes place.
Generally is used because there are exceptions involving bond contracts.
Financial futures become possible when it is observed that the actual commodity need not be delivered at the end of the contract only the profit over
the current spot price is paid. For example, assume the futures contract price
is $3 and the spot price is $2. Then the buyer of futures contract pays $1 to the
seller and no transfer of asset needs to take place.
A financial future can also be formed by converting an index into a monetary
equivalent. For instance, a stock index future can be constructed by valuing each
15.3. FUTURES
231
10 points at $1. Thus an index of 6100 would trade at a price of $610. If the
index fell to 6000, the futures price would become $60. Using such a mechanism,
it becomes possible to construct such contracts on any future price.
Example of exchanges in the US where financial futures are traded are the
Chicago Board of Trade, Mid-America Commodity Exchange and New York
Board of Trade.
R
Futures
Example 140 NYSE Composite Index
R
Futures Small Contract Size
Contract REVISED NYSE Composite Index
$5 NYSE Composite Index (e.g., $5 5000.00 = $25,000) Symbol Value of
Minimum Move MU $2.50
R
Futures Reg. Contract Size
Contract REVISED NYSE Composite Index
R

(e.g., $50 5000.00 = $250,000) Symbol
$50 NYSE Composite Index
Value of Minimum Move YU $25.00
Price Quotation: Index Points where 0.01 equals $0.50
Daily Price Limits: Please contact the Exchange for information on daily
price limits for these contracts.
Position Limits: NYSE Regular (on a 10:1 basis) are converted into NYSE
Small positions for limit calculation purposes. Any One Month Limit 20,000
All Months Combined Limit 20,000
Cash Settlement: Final settlement is based upon a special calculation of the
third Fridays opening prices of all the stocks listed in the NYSE Composite
R
.
Index
(http://www.nybot.com/specs/yxrevised.htm)
In the UK, futures contracts are traded on LIFFE the London International
Financial Futures Exchange which was opened in 1982.
Example 141 LIFFE offers a range of futures and options, and provides an
arena for them to be traded. The Exchange brings together different parties
such as financial institutions, corporate treasury departments and commercial
investors, as well as private individuals some of whom want to offset risk,
hedgers, and others who are prepared to take on risk in the search for profit.
Following mergers with the London Traded Options Market (LTOM) in 1992
and with the London Commodity Exchange (LCE) in 1996, LIFFE added equity
options and a range of soft and agricultural commodity products to its existing
financial portfolio. Trading on LIFFE was originally conducted by whats known
as open outcry. Traders would physically meet in the Exchange building to
transact their business. Each product was traded in a designated area called a
pit, where traders would stand and shout the price at which they were willing to
buy or sell.
In 1998, LIFFE embarked on a programme to transfer all its contracts from
this traditional method of trading, to an electronic platform. This transition is
now complete. The distribution of LIFFE CONNECTTM stands at around 450
sites, more than any other trading system in the world, and covers all major
time zones. This distribution continues to grow.
(http://liffe.npsl.co.uk/liffe/site/learning.acds?instanceid=101765&context=100190)
232
There are three major types of futures traded on LIFFE.

Contracts on short term interest rates These are based on the three-month
money market rate and are priced as 100 - interest rate. Consequently,
when the interest rate goes up it implies the price of the futures contract
goes down.
Bond futures Bond future represent long-term interest rate futures. They
are settled by delivery of bonds, with adjustment factors to take account
of the range of different bonds that may be delivered. This is a financial
future which is settled by actual delivery of the commodity.
Equity index futures Equity index futures are cash settled and are priced
per index point.
15.4
Motives for trading
Two motives can be identified for trading forwards and futures. These are
hedging and speculation. These motives are now discussed in turn.
15.4.1
Hedging
Hedging is the use of the contracts to reduce risk. Risk can arise from either
taking demanding or supplying a commodity at some time in the future. The
current price is known but the price at the time of demand or supply will not
be known. A strategy of hedging can be used to guard against unfavorable
movements in the product price.
Two examples of the way in which hedging can be employed are now given.
Example 142 Consider a bakery which needs wheat in three months. It can:
i. wait to buy on the spot market;
or
ii. buy a future now.
If the baker followed (ii) they would be a long hedger this is the investor
who has committed to accept delivery.
Example 143 Consider a company in the UK who will be paid in three months
time in Euros. It can:
i. sell a future on the Euros now;
or
ii. wait to receive the Euros and sell them on the spot market.
If the firm followed (i) they would be a short hedger the investor who
commits to supply the commodity.
The advantage of a futures contract is that it fixes the price and guards
against price changes. For someone who has to buy in the future it can be used
233
15.4. MOTIVES FOR TRADING
to insure against price increases while for someone who has to sell in the future
it can insure against price falls.
A company that is due to sell an asset at a particular time in the future
can hedge by taking a short futures position. They then hold a short hedge. A
company that is due to buy an asset at a particular time in the future can hedge
by taking a long futures position a long hedge.
Hedging through the use of futures contracts reduces risk by fixing a delivery or purchase price. This insures against adverse price movements but also
means that profit is lost from favorable price movements. The optimal degree of
hedging determines the best trade-off between these. In effect, it is usually best
to cover some exposure by hedging but leave some uncovered in order to profit
from favorable price movements. The hedge ratio is the size of the position in
futures relative to size of exposure
One way of analyzing the optimal degree of hedging is to consider the strategy that minimizes the variance in a position. The optimal hedge ratio can be
determined by considering the variation in the spot price and the futures price.
Let S be change in spot price S over length of hedge and F be change in
futures price F over length of hedge. The standard deviation of S is denoted by
S and the standard deviation of F by F . Let be coefficient of correlation
between S and F and let the hedge ratio be denoted by h.
Consider a position which is long in the asset but short in future. With h
denoting the hedge ratio, the change in the value of the position over the life of
the hedge is
P = S hF.
(15.1)
Conversely, when long in the future but short in the asset the change in value
of position is
P = hF S.
(15.2)
For both of these positions, the variance of change in the value of hedged position
is
var(P ) = E (P E (P ))2
= E (S hF E (S hF )) .
(15.3)
Computing the expectation gives

var(P ) = 2S + h2 2F 2h S F .
(15.4)
One definition of an optimal policy is to choose the hedge ratio to minimize

this variance. The necessary condition for the hedge ratio is
dvar(P )
= 2h2F 2 S F = 0.
dh
(15.5)
Solving this condition, the hedge ratio that minimizes the variance is
h=
S
.
F
(15.6)
234
Given data on these standard deviations and the correlation, this optimal hedge
ratio is simple to compute.
Example 144 A company must buy 1m gallons of aircraft oil in 3 months.
The standard deviation of the oil price is 0.032. The company hedges by buying futures contracts on heating oil. The standard deviation is 0.04 and the
correlation coefficient is 0.8. The optimal hedge ratio is
0.8
0.032
= 0.64.
0.040
One heating oil futures contract is for 42000 gallons. The company should buy
0.64
1000000
= 15.2,
42000
contracts, which is 15 when rounded.

The example illustrates that the hedge does not have to be in the same
commodity but only in a similar commodity whose price is highly correlated
with the one being hedged. In addition, it also show that optimal hedging does
not necessarily imply that all of the exposure has to be covered. In the example
the company has an exposure of 1m gallons but buys futures contracts of 630000
gallons.
15.4.2
Speculation
The second reason for trading in futures is speculation. If the spot price is
expected to change, a trader can engage in speculation through futures.
A speculator has no interest in taking delivery of the commodity or of supplying it, but is simply interested in obtaining profit through trade. Consequently,
any trade they make must ultimately be matched by a reversing trade to ensure
that they do not need to receive or deliver.
For an expected price rise a speculator will:
i. Buy futures now;
ii. Enter a reversing trade to sell later after the price has risen.
Conversely, for an expected price fall, the speculator will:
i. Sell futures now;
ii. Enter a reversing trade to buy later after the price has fallen.
Clearly, even though the quantity of commodity to be traded is limited to
the amount produced, any number of speculative trades can be supported if
there are speculators on both sides of the market.
15.5
Forward Prices
The valuation issue involved with forward contracts is to determine the delivery
price, or forward price, that is written into the contract at its outset. At the
15.5. FORWARD PRICES
235
time the two parties on either side of a contract agree the trade, no payment
is made. Instead the forward price is set so that the contract is fair for both
parties. To be fair the contract must have a value of zero at the time it is agreed.
It is this fact that allows the delivery price to be determined.
As we will see, the forward price in the contract and the spot price of the
underlying asset at the time the contract is agreed are related. This relationship
is now developed as the basis for determining the forward price.
This section develops the valuation of forward contracts. Forwards are considered since the daily settlement involved in futures contracts makes their
analysis more complex. A later section explores the extent of the differences
between the values of the two contracts.
The focus of this section is upon investment assets. The important feature of
these is that it is possible to go short in these assets or reduce a positive holding
if it is advantageous to do so. This allows us the flexibility to apply an arbitrage
argument to obtain the forward price. A number of cases are considered which
differ in whether or not the asset pays an income.
15.5.1
Investment Asset with No Income
The process of valuation using arbitrage involves searching for profitable opportunities by combining the assets that are available. To determine the fair
futures price it is assumed that the assets available consist of a risk-free asset,
the asset underlying the forward contract and the forward contract. If the forward price is not correctly set, it becomes possible to produce arbitrage profits
by combining these assets.
The construction of an arbitrage portfolio is illustrated by the following
example.
Example 145 Consider a stock with a current spot price of $40, which will pay
no dividends over the next year, and a one-year risk free rate of 5%. Suppose
that the forward price for delivery in one year is $45 and a contract is for 100
shares. Given these numbers, it is possible to earn an arbitrage profit.
To achieve the profit, the following investment strategy is used:
1. Borrow $4000 for 1 year at the interest rate of 5%;
2. Buy 100 shares of the stock for $4000;
3. Enter into a forward contract to sell 100 shares for $4500 in 1 year.
On the delivery date of the forward contract at the end of 1 year, the loan
requires $4000e0.05 = $4205.1 to repay. The stock is sold for $4500. Hence
a profit of $294.9 is earned. Note that this profit is entirely certain since all
agreements are made at the outset of the forward contract. In particular, it does
not depend on the price of the underlying stock at the delivery date. Since a
risk-free profit can be earned, the forward price of 45 cannot be an equilibrium.
Now consider the formulation of an investment strategy for a lower forward
price.
236
Example 146 Consider a stock with a current spot price of $40, which will pay
no dividends over the next year, and a one-year risk free rate of 5%. Suppose
that the forward price for delivery in one year is $40 and a contract is for 100
shares. Given these numbers, it is possible to earn an arbitrage profit.
To achieve the profit, the following investment strategy is used:
1. Sell short 100 shares of the stock for $4000;
2. Lend $4000 for 1 year at the interest rate of 5%;
3. Enter into a forward contract to buy 100 shares for $4000 in 1 year.
On the delivery date of the forward contract at the end of 1 year, the loan is
repaid and provides an income of $4000e0.05 = $4205.1. The stock is purchased
for $4000. Hence a profit of $205.1 is earned. This profit is entirely certain so
the forward price of 40 cannot be an equilibrium.
In the first example, the loan requires $4205.1 to repay, so no profit will
be earned if the sale at the forward price earns precisely this same amount.
Similarly, in the second example, no profit is earned if the purchase of the shares
costs $4205.1. Putting these observations together, the only forward price that
eliminates arbitrage profits has to be $42.05. This price satisfies the relation
that
42.05 = 40e0.05 .
(15.7)
That is, the forward price is the current spot price compounded at the risk-free
rate up to the delivery date.
To express this for a general forward contract on an investment asset with
no dividend, let the forward price at the outset of the contract be F0 , the spot
price be S0 , the continuously compounded risk-free interest rate be r and the
time to the delivery date be T . The forward price agreed at the outset of the
contract must then be
F0 = S0 erT .
(15.8)
The construction of an arbitrage portfolio is only one method of obtaining

the forward price. Recall that a similar process lead to the valuation of an
option in the binomial model. Approaching forward contracts from this second
direction emphasizes the generality of the method of valuation and shows that
futures are not distinct from options.
Consequently, assume that spot price of the underlying asset at the outset
of the contract is S0 . Adopting the binomial assumption, the price of the underlying stock can change to either uS0 or dS0 at the delivery date in the forward
contract. For the investor who is short in the contract, the value of the forward
contract at the delivery date is either F0 uS0 when the asset price is uS0 or
F0 dS0 when the price is dS0 . These prices and values produce the binomial
tree in Figure 15.1.
Risk-neutral valuation can now be applied to the binomial tree. Let the riskneutral probability associated with a move to uS0 be q and that with a move
to dS0 be 1 q. With an option contract, a premium is paid for the contract
237
15.5. FORWARD PRICES
Price of
Underlying
Value of
Contract
uS0
F0 uS 0
S0
dS 0
F0 dS 0
Figure 15.1: Binomial Tree for a Forward Contract

and it is the fair value of this that is determined by risk-neutral valuation. In
contrast, with a forward contract no payment is made or received at the start
of the contract. Instead, the price in the contract F0 is chosen to make the
contract fair, or to give it zero initial value. Letting V0f be the initial value
of a futures contract, then F0 must satisfy
V0f =
1
[q [F0 uS0 ] + (1 q) [F0 dS0 ]] = 0.
R
(15.9)
Solving this equation for F0

F0 = quS0 + (1 q) dS0 .
Using the fact that q =
Rd
ud ,
(15.10)
this can be simplified to

F0 = RS0 ,
(15.11)
which is precisely the same price as in (15.8) when expressed in terms of discrete
compounding.
Furthermore, for a binomial tree with n sub-periods, the initial forward price
can be shown to satisfy
F0 = Rn S0 ,
(15.12)
so it converges to the result with continuous discounting as n . Hence, riskneutral valuation in the binomial tree can be used to value forward contracts in
exactly the same way as for options.
238
Price of
Underlying
Value of
Contract
uS 0 IR
F0 uS 0 + IR
S0
dS0 IR F0 dS 0 + IR
Figure 15.2: An Asset with Income
15.5.2
Investment Asset with Known Income
Many financial assets provide an income to the holder. The holder of a forward
on the asset does not receive this income, but the price of the underlying asset
decreases to reflect the payment of the income. This observation allows the
payment of income to be incorporated into the binomial tree.
If the asset pays an income with present value of I just prior to the delivery
date in the forward contract, the value of the asset will be reduced to uS0 IR
on the upper branch of the tree and dS0 IR on the lower branch. The modified
binomial tree is in Figure 15.2.
The application of risk-neutral valuation gives
V0f =
1
[q [F0 uS0 + IR] + (1 q) [F0 dS0 + IR]] = 0.
R
(15.13)
Solving this using the definitions of the risk-neutral probabilities provides the
forward price
F0 = [S0 I] R.
(15.14)
As before, this can be extended naturally to the continuous case as

F0 = [S0 I] erT .
(15.15)
Therefore, if the asset pays an income this reduces the forward price because
the person who is long in the forward contract does not receive this income but
is affected by the fall in the assets price immediately after the income is paid.
15.6. VALUE OF CONTRACT
15.5.3
239
Continuous Dividend Yield
Rather than making a single payment of income, an asset may have a continuous
flow of dividends. Let the rate of flow of dividends be q. Then the previous
result can be modified to
F0 = S0 e(rq)T .
(15.16)
A continuous flow of dividends has the effect of continually reducing the asset
price so reduces the forward price.
15.5.4
Storage costs
Storage costs are the opposite of income. They can be added into the expressions
directly.
Let U be present value of storage costs then
F0 = [S0 + U ] erT .
15.6
(15.17)
Value of Contract
It has already been noted that at the outset of the contract the forward price
is chosen to ensure that the value of the contract is zero. As time progresses,
the spot price of the underlying asset will change as will the forward price in
new contracts. The contract can then either have a positive value if the price
change moves in its favor and negative if it moves against.
To determine this value, let Ft be forward price at time t, and F0 the forward
price in a contract agreed at time 0. With time T t to the delivery date, the
value, f , of the forward contract is then given by
Vtf = [Ft F0 ] er[T t] .
(15.18)
As already noted, at the time the contract is written its value is zero. Now since
Ft = St er[T t] it follows that the value of the contract at time t is
Vtf = St F0 er[T t] .
(15.19)
With an income from the asset, this value becomes

Vtf = St I F0 er[T t] ,
(15.20)
f = S0 eqT F0 er[T t] .
(15.21)
and with a dividend
15.7
Commodities
Considering forward contracts on commodities does make a difference to these

results. The features of commodities are that there may be no chance to sell
240
short, and storage is sometimes not possible if the commodity is perishable.

This means the pricing relations have to be revised.
Returning to the basic strategies, it is possible to borrow money, buy the
underlying asset, go short in a forward, hold the asset until the delivery date
and then deliver and repay the loan. This must not be profitable.
Let U be present value of storage costs the strategy is not profitable if
F0 [S0 + U ] erT .
(15.22)
This relation puts an upper bound on the forward price. A lower bound cannot
be applied without the possibility of short sales or of sales from stocks. If
the good cannot be stored, then U can be thought of as the cost of actually
producing the commodity.
15.8
Futures Compared to Forwards
In general, futures and forwards will not have the same price because of the
daily settlement. This leads the two assets have different flows of payments.
When the risk-free interest rate is constant, then
forward price = future price.
(15.23)
This identity arises because with the constant interest rate the timing of the
payments does not matter since they have the same present value.
Prices need not be the same when interest rates vary because of daily settlement. Consider a situation where the spot price, S, is positively correlated
with the interest rate. With a long position, an increase in S earns a daily
profit. Positive correlation ensures this is invested when r is high. Conversely,
a decrease in S earns a loss which is covered when interest rates are low. This
implies the future is more profitable than the forward.
Despite the observations, the difference in price may be small in practice.
15.9
Backwardation and Contango
The final issue to address is the relationship between the futures price and the
expected spot price.
There are three possibilities that may hold.
1. Unbiased predictor.
In this case, the futures price is equal to the expected spot price at the
delivery date of the contract. Hence
F0 = E [ST ] .
(15.24)
2. Normal backwardation.
The argument for normal backwardation follows from assuming that
a. Hedgers will want to be short in futures,
241
15.10. USING FUTURES
Price of
Future
Normal
Contango
Price of
Stock
Normal
Backwardation
Delivery
Time
Figure 15.3: Backwardation and Contango

b. Will have to offer a good deal to speculators,
Together these imply that
F0 < E [ST ] .
(15.25)
3. Normal contango.
The argument for normal backwardation follows from assuming that
a. Hedgers will want to be long on average,
b. Must encourage speculators to be short,
Together these imply that
F0 > E [ST ] .
(15.26)
The empirical evidence on this issue seems to suggest that generally F0 <
E [ST ], so that normal backwardation holds.
15.10
Using Futures
Investment strategies with futures.
15.11
Conclusions
This chapter has introduced futures and forwards. The nature of the contracts
has been described and the methods of valuation analyzed. A fair price has
242
been determined by using both arbitrage arguments and the binomial model.
Chapter 16
Swaps
16.1
Introduction
In 1981 IBM and the World Bank undertook an exchange of fixed rate debt
for floating rate debt. This exchange was the start of the interest rate swap
industry. It is now estimated that the market is worth over $50 trillion per
year. But it is difficult to provide a precise valuation of the size of the market
because the market is not regulated and swaps are arranged between individual
parties and not through exchanges.
The financial swaps we will consider are agreements to exchange one sequence
of cash flows over a fixed period for another sequence of cash flows over the same
period. This is precisely what IBM and the World Bank did.
The two sequences of cash flows are tied to either to a debt instrument or
to a currency. This gives the two main types of swaps:
Interest rate swaps
Currency rate swaps
Why did swaps emerge? The first swaps were conducted in the late 1970s
to avoid currency UK currency controls. These controls limited the value of
currency that could be exchanged but this could easily be avoided by swapping
rather than exchanging. These were followed by the IBM and World Bank swap
in 1981. By 2001 it was estimated that $57 trillion in underlying value was
outstanding in swap agreements.
The next section describes interest rate swaps and currency swaps. The
use of swaps and the market for swaps are then described. The chapter then
proceeds to the valuation of swaps.
16.2
Plain Vanilla Swaps
The basic form of interest rate swap, the plain vanilla, is now described.
243
244
CHAPTER 16. SWAPS
The first step to do this is to introduce the LIBOR. This is the London Interbank Offered Rate the rate of interest at which banks lend to each other. This
rate is fundamental to valuing swaps since it acts as the basic floating rate of
interest.
Definition 2 British Bankers Association (BBA) LIBOR is the BBA fixing
of the London Inter-Bank Offered Rate. It is based on offered inter-bank deposit
rates contributed in accordance with the Instructions to BBA LIBOR Contributor Banks. The BBA will fix BBA LIBOR and its decision shall be final. The
BBA consults on the BBA LIBOR rate fixing process with the BBA LIBOR
Steering Group. The BBA LIBOR Steering Group comprises leading market
practitioners active in the inter-bank money markets in London. BBA LIBOR
is fixed on behalf of the BBA by the Designated Distributor and the rates made
available simultaneously via a number of different information providers. Contributor Panels shall comprise at least 8 Contributor Banks. Contributor Panels will broadly reflect the balance of activity in the inter-bank deposit market.
Individual Contributor Banks are selected by the BBAs FX & Money Markets Advisory Panel after private nomination and discussions with the Steering
Group, on the basis of reputation, scale of activity in the London market and
perceived expertise in the currency concerned, and giving due consideration to
credit standing. (http://www.bba.org.uk/bba/jsp/polopoly.jsp?d=225&a=1413)
16.2.1
Interest Rate Swap
A swap requires two parties to participate. For the purpose of the discussion,
call these party A and party B.
On one side of the swap, party A agrees to pay a sequence of fixed rate
interest payments and to receive a sequence of floating rate payments. A is
called the pay-fixed party.
On the other side of the swap, party B agrees to pay a sequence of floating
rate payments and to receive a sequence of fixed rate payments. B is called the
receive-fixed party
The tenor is the length of time the agreement lasts and the notional principal
is the amount on which the interest payments are based. With a plain vanilla
swap, interest is determined in advance and paid in arrears.
Example 147 Consider a swap with a tenor of five years and two loans on
which annual interest payments must be made. Let the notional principal for
each loan be $1m. Party A agrees to pay a fixed rate of interest of 9% on the
$1m. Party B receives this fixed rate, and pays the floating LIBOR to A.
In principal, the swap involves loans of $1m being exchanged between the
parties. That is, A has a floating interest rate commitment which is transfers
to B and B has a fixed-rate commitment that it transfers to A. But in practice
there is no need for these loans to exist and the principal can be purely nominal.
In fact only the net payments, meaning the difference in interest payments, are
made.
245
16.2. PLAIN VANILLA SWAPS
Table 16.1 illustrates the cash flows resulting from this swap agreement for
a given path of the LIBOR. It must be emphasized that this path is not known
when the swap agreement is made. The direction the LIBOR takes determines
which party gains, and which party loses, from the swap. The parties will
enter such an agreement if they find the cash flows suit their needs given the
expectations of the path of the LIBOR.
Year, t
0
1
2
3
4
5
16.2.2
LIBORt
8
10
8
6
11
Table 16.1:
Floating Rate(B A)
Fixed Rate(A B)
80,000
90,000
100,000
90,000
80,000
90,000
60,000
90,000
110,000
90,000
Cash Flows for a Plain Vanilla Swap
Currency Swaps
A currency swap involves two parties exchanging currencies. It will occur when
two parties each hold one currency but desire another. This could be for reasons
of trade or because they aim to profit out of the swap based on expectations of
exchange rate movements. The parties swap principal denominated in different
currencies but which is of equivalent value given the initial exchange rate.
The interest rate on either principal sum may be fixed or floating. As an
example, consider two parties C and D. Assume that C holds Euros but wants
to have dollars. For instance, C may have to settle an account in dollars. In
contrast, D holds dollars but wants to have Euros instead. The two parties can
engage in a swap and trade the dollars for Euros. Unlike an interest rate swap,
the principal is actually exchanged at the start of the swap. It is also exchanged
again at the end of the swap to restore the currency to the original holder.
The fact that the interest rates can be fixed or floating on either currency
means that there are four possible interest schemes:
C pays a fixed rate on dollars received, D pays a fixed rate on Euros
received
C pays a floating rate on dollars received, D pays a fixed rate on Euros
received
C pays a fixed rate on dollars received, D pays a floating rate on Euros
received
C pays a floating rate on dollars received, D pays a floating rate on Euros
received
The predominant form of contract is the second. If party D is a US firm,
then with a plain vanilla currency swap the US firm will pay a fixed rate on the
currency it receives.
246
CHAPTER 16. SWAPS
$10m
French
firm
US
firm
10m
Figure 16.1: Currency Swap
To show how a currency swap functions, consider a swap of type 1 which
involves exchanging fixed-for-fixed. The cash flows that occur with this swap
are:
The initial swap of currency at initiation
The periodic interest payments
The swap of principal at termination
A currency swap involves interest payments which are made in the currency
received. Consequently, since the two payments are in different currencies, there
is no netting of the interest payments.
Example 148 Consider a US firm that holds dollars but wants euros and a
French firm that holds Euros but wants dollars. Both parties agree to pay fixed
interest. Assume that:
a. The spot exchange rate is $1 = 1. The spot rate is the rate for immediate
exchange of currency.
b. The US interest rate is 10%
c. The French interest rate is 8%
d. The tenor of the swap is 6 years
e. Interest is paid annually
f. The principal swapped is $10m for 10m.
It should be noted that given the spot exchange rate, the principal exchanged
is of equal value. This implies the fact that the swap must always be of equal
value at the initial spot exchange rate. Figure 16.1 displays the exchange of
principal at the start of the swap.
The cash flows during the period of the swap are illustrated in Table 16.2.
This shows that the interest payments are made in the currency received. Since
the swap is fixed-for-fixed, the interest payments remain constant.
247
16.2. PLAIN VANILLA SWAPS
0
1
2
3
4
5
6
To US From US To French From French

10m $10m
$10m
10m
$1m
0.8m
0.8m
$1m
$1m
0.8m
0.8m
$1m
$1m
0.8m
0.8m
$1m
$1m
0.8m
0.8m
$1m
$1m
0.8m
0.8m
$1m
$1m
0.8m
0.8m
$1m
Table 16.2: Cash Flow for Fixed-for-Fixed
Given these payments, it is natural to ask which flow is best. The answer to
this question depends on (i) the needs of the two firms for currency, and (ii) the
course of exchange rates over the lifetime of the swap. Because the interest and
principal have to be repaid in a currency different to the one that was initially
held, entering a swap agreement opens the parties up to exchange rate risk.
Example 149 Consider a swap between a US firm and a Japanese firm. The
Japanese firm pays a floating rate on dollars received and the US firm pays a
fixed rate on the Yen received. Assume that:
a. The spot exchange rate be $1 = Y120.
b. The principal is $10m when denominated in dollars and Y1200m when
denominated in Yen.
c. The tenor of the swap is 4 years.
d. The Japanese 4-year fixed interest rate is 7%. This is the interest rate
paid on the Yen received by the US firm.
e. The rate on the dollar is the LIBOR, which is 5% at the initiation of the
swap.
The cash flows during the swap are determined by the path of the LIBOR.
Table 16.3 displays the flows for one particular path of the LIBOR. In this
table, the LIBOR rises over time so the interest payments received by the US
firm increase over time. If the exchange rate were constant, this would be
advantageous for the US firm. However, as will be seen later, the exchange rate
is related to the interest rate and this needs to be taken into account before this
claim can be established.
Time
0
1
2
3
4
LIBOR
5%
6%
7%
10%
Japanese in Japanese out US in

$10m
Y1200m
Y1200m
Y84m
$0.5m
$0.5m
Y84m
$0.6m
$0.6m
Y84m
$0.7m
$0.7m
Y1284m
$1.1m
$1.1m
Table 16.3: Cash Flow on Fixed-for-Floating
US out
$10m
Y84m
Y84m
Y84m
Y1284m
248
16.3
CHAPTER 16. SWAPS
Why Use Swaps?
There are three major reasons why swaps may be used. These are now considered in turn.
16.3.1
Market Inefficiency
A first reason for using swaps is to overcome market inefficiency. For example,
it could be the case that firms located in a country are able to borrow at a lower
rate in that country than firms located abroad. This creates a position in which
firms have a comparative advantage in borrowing in their countrys currency.
Given such a position of comparative advantage, it is possible for two parties to
find a mutually advantageous trade.
Such a trade is illustrated in Table 16.4 where the US firm can borrow dollars
at 9% but the UK firm must pay 10% to borrow dollars. The opposite position
holds for borrowing in the UK.
US $ rate UK rate
US firm 9%
8%
UK firm 10%
7%
Table 16.4: Interest Rates
Assume that the UK firm wants dollars and the US firm wants Sterling. If
they were to borrow directly at the rates in the table, the US firm would pay
a rate of interest of 8% on its sterling and the UK firm a rate of 10% on its
dollars.
If the firms were to borrow in their own currency and then swap, this would
reduce the rate faced by the US firm to 7% and that faced by the UK to 9%.
This swap is illustrated in Figure 16.2. The exploitation of the comparative
advantage is beneficial to both parties.
The existence of the comparative advantage depends on there being a market inefficiency that gives each firm an advantage when borrowing in its home
market. If the market were efficient, there would be a single ranking of the
riskiness of the firms and this would be reflected in the rates of interest they
pay in both countries. The internalization of financial markets makes it unlikely
that there will be significant inefficiencies to be exploited in this way.
16.3.2
Management of Financial Risk
Swaps can be used to manage financial risk. This is clearest when assets and
liabilities are mismatched.
The US Savings and Loans provide a good example of the possibility of risk
management using swaps. These institutions receive deposits from savers and
use the funds to provide loans for property.
The Savings and Loans pay floating rate interest on deposits but they receive
fixed rate interest on the loans they grant. Since the loans are for property they
are generally very long term.
249
16.3. WHY USE SWAPS?
$15m
US
firm
UK
firm
10m
$15m
Lender
US
10m
Lender
UK
Figure 16.2: Exploiting Comparative Advantage

This places the Savings and Loans in a position where they are exposed to
risk if the floating interest rate rises. Such a rise would create an increase in
their payments to depositors but would not be accompanied by any increase
from the long-term loans. Precisely this position was responsible, at least in
part, for the collapse of a number of these institutions in the 1980s.
Example 150 The Savings and Loan crisis of the 1980s was a wave of savings
and loan failures in the USA, caused by mismanagement, rising interest rates,
failed speculation and, in some cases, fraud. U.S. taxpayers took the brunt
of the ultimate cost, which totaled around US$600 billion. Many banks, but
particularly savings and loan institutions, were experiencing an outflow of low
rate deposits, as depositors moved their money to the new high interest money
market funds. At the same time, the institutions had much of their money
tied up in long term mortgages which, with interest rates rising, were worth
far less than face value. Early in the Reagan administration, savings and loan
institutions (S&Ls) were deregulated (see the Garn - St Germain Depository
Institutions Act of 1982), putting them on an equal footing with commercial
banks. S&Ls (thrifts) could now pay higher market rates for deposits, borrow
money from the Federal Reserve, make commercial loans, and issue credit cards.
(http://en.wikipedia.org/wiki/Savings_and_Loan_scandal)
A solution to the risk problem faced by the Savings and Loans would have
been to swap the fixed interest rate loans for floating interest rate loans. By
doing this, they could have ensured that any increase in interest rates is met by
an increase in expenditure and receipts.
It is clear that there are other possible responses for the Savings and Loans
to secure their position. For instance they could issue bonds with a fixed coupon
and buy floating rate notes. This would then balance their portfolio as a whole.
250
CHAPTER 16. SWAPS
The reason why such a trade may not be used is because of legislation which
limits the financial activities that can be undertaken.
16.3.3
Speculation
Expressed in the simplest terms, a swap is a no more than a bet on the direction
of interest rate and/or exchange rate movements. If the movement is in the
right direction, a profit can be earned. Swaps can therefore be used for purely
speculative reasons.
16.4
The Swap Market
This section discusses the major features of the swap market and the participants
in that market.
16.4.1
Features
The major features of the swap market are the following:

a. There is no publicly observable marketplace. Swaps are transactions
that take place either between individuals directly, between individuals with
the intermediation of a broker, or with a swap dealer. Brokers and dealers are
discussed further below.
b. There is limited government regulation. Since there is no marketplace
it is difficult for any government to provide regulation. There has been some
recent discussion of regulation.
Example 151 Americas continued financial leadership in the new economy is
at stake as Congress sets out to modernize the Commodity Exchange Act, the
law that covers futures and derivatives trading. The revised law is supposed to
liberalize the derivatives market, setting important legal terms that distinguish
traditional commodity futures from over-the-counter derivatives, or swaps. The
future of the U.S. swaps market depends on whether Congress can keep it free
of entangling regulations and legal uncertainty. Derivatives are an essential tool
of risk management for American businesses. They are the lubricants that let
financial markets allocate capital more efficiently. Foreign exchange swaps, for
example, diminish the risks associated with fluctuating currencies. Rate swaps
smooth out the effects of interest-rate fluctuations by converting long-term, fixedrate debt into short-term, variable-rate debt. OTC derivatives make businesses
more competitive by lowering their cost of capital. To be effective, the enforceability and legal status of swaps must be firmly established. Banks and other financial institutions have worried for years that the Commodity Futures Trading
Commission might begin applying futures regulations to swaps. That would be
disastrous, since futures contracts are legally enforceable only if they are traded
on a listed exchange, such as the Chicago Mercantile Exchange. Off-exchange
swaps are privately negotiated, custom-tailored contracts. Trillions of dollars
in interest-rate and currency-swap contracts would be undermined if they were
16.4. THE SWAP MARKET
251
suddenly regulated like futures. Banks furnishing swaps to large institutional

and corporate clients are poised to extend the benefits of these risk management
tools to their small business and retail customers. But they are wary of the
CFTC, which in 1998 considered regulating swaps. The legal uncertainty this
created was unsettling to the financial markets, which dont consider the CFTC
technically competent to regulate complex swap transactions. Unwarranted bureaucratic restrictions would reduce the technical precision of swaps and increase
their cost. House bill H.R. 4541, the Commodity Futures Modernization Act, is
supposed to rationalize the regulatory environment and provide legal certainty.
But this effort is fragmented because of the competing jurisdictions of regulatory agencies and congressional committees. An amendment recently offered by
House Banking Committee Chairman Jim Leach, R-Iowa, goes the furthest in
liberalizing OTC swaps, but still leaves room for regulatory meddling. Though
the CFTC couldnt regulate them, the Treasury Department or Federal Reserve
could. Other versions of H.R. 4541 set up a convoluted series of exemptions to
insulate most swaps from CFTC regulation, but dont exempt the entire universe
of swaps. Individual investors worth less than $5 million to $10 million in assets will likely face regulatory hurdles. Ostensibly these restrictions are meant to
protect retail investors from fraud. However, as Harvard University law professor Hal Scott testified to the House Banking Committee, the true purpose might
be to fence off exchange-traded derivatives markets from competition with OTC
derivatives markets for retail investors. Swap contracts completed over electronic trading facilities are potentially vulnerable under the bill. Specifically,
derivative transactions resulting from automated trade matching algorithms
are exposed to additional regulation. This language could inhibit the new economy innovators that match trades electronically using highly specialized software.
The big commodity exchanges would benefit from rules that hinder off-exchange
innovators. But the added red tape will only delay the inevitable. If regulatory
barriers are set up to protect the futures industry from electronic competition,
the innovators will simply move offshore. If Congress wishes to liberalize swaps,
it should do so by defining commodity futures narrowly and prohibiting any regulation of OTC derivatives outside the definition. Over-the-counter swaps should
be completely exempt from antiquated exchange rules that were designed for the
old economy. Rather than leaving any OTC derivatives in regulatory limbo,
Congress should confer ironclad legal certainty upon all kinds of swaps. (Swap
New For Old: Congress Shouldnt Impose Tired Rules On OTC Derivatives by
James M. Sheehan, August 9, 2000, Investors Business Daily)
c. Contracts cannot be terminated early. The nature of a swap deal is that
it is a commitment that must be seen through to the end. Once it is agreed it
is not possible to withdraw from the deal.
d. No guarantees of credit worthiness. With futures there is an exchange
which manages the contracts to avoid any possibility of default by ensuring
margin is held and limiting daily movements of prices. The fact that there is no
marketplace for swaps implies that there is no similar institution in the swap
market.
252
CHAPTER 16. SWAPS
Example 152 London Borough of Hammersmith and Fulham: A local government in the United Kingdom that was extremely active in sterling swaps between
1986 and 1989. Swap volume was very large relative to underlying debt, suggesting large scale speculation by the borough council. The speculation was unsuccessful and a local auditor ruled that the transactions were ultra vires-beyond the
powers of the council. The House of Lords sitting as the High Court ultimately
upheld the auditors ruling. The legal risk of some risk management contracts was established at considerable cost to the London financial community.
(http://riskinstitute.ch/00011654.htm)
16.4.2
Dealers and Brokers
For anyone wishing to conduct a swap there is the problem of finding a counterparty. For other derivatives, such as options and futures, this is less of a
problem since there are organized exchanges to assist with transactions.
In the early days of the swap market counterparties to a swap were originally
found via a broker. The market has developed so that swaps are now generally
conducted through dealers. This has increased the efficiency of the swap market.
Swap Broker
A swap broker acts as an intermediary in the market. Their role is to match
swap parties who have complementary needs.
A broker maintains a list of clients who are interested in entering into swap
deals and tries to match the needs of the clients.
But because it is necessary for a broker to find matching clients before any
trade can take place, the organization of a market through brokers does not
make for a very efficient market.
Swap Dealer
A swap dealer acts as a counter-party to a swap. They can be on either side of
the deal. The profit of a swap dealer is obtained by charging a spread between
the two sides of the deal.
The dealer accumulates a swap book. The book is constructed with the aim:
of balancing trades to limit risk.
The risks facing a swap dealer are the following:
1. Default risk
The party on the other side of a swap may default.
2. Basis risk
The basis risk arises from movements in interest rates.
3. Mismatch risk
Mismatch risk arises from the two sides of the dealers swap book not being
balanced.
16.5. THE VALUATION OF SWAPS
16.5
253
The Valuation of Swaps
The process of valuation relates to answering two related questions. How is a

swap correctly priced? How can the deal be fair for both parties?
As an example, consider a plain vanilla interest rate swap. The party on one
side of this swap will pay the floating LIBOR rate, while the party on the other
side pays a fixed rate of interest. The only variable in this transaction that can
be adjusted to make the deal fair for both parties is the fixed rate. By making
this higher, the receive-fixed party benefits. Make it lower and the pay-fixed
party benefits.
The fundamental issue is to determine what fixed rate should be used to
make the deal fair. Here fair means that both parties see the swap as equally
advantageous at the time at which it is agreed.
Before proceeding to determine the fixed rate, it is worth looking at how
swaps are related to bond portfolios. The reasoning is the same as that applied
to options and forwards: the swap is constructed so that there are no arbitrage
opportunities. Both of the earlier derivatives were priced by constructing a
replicating portfolio that gave the same payoffs as the derivative. Applying the
arbitrage argument then means the price of the derivative must be the same as
the cost of the replicating portfolio.
The same basic logic can be applied to swaps where bonds can be used to
replicate the position of a party who has entered a swap deal.
16.5.1
Replication
Definition: a floating rate note is a bond that pays a floating rate of interest
(LIBOR for this analysis)
1. Interest rate swaps
a. Plain vanilla receive-fixed
This is equivalent to
-a long position in a bond
-a short position in a floating rate note
Example 1. A 6% corporate bond with annual coupon maturity 4 years,
market value of $40m trading at par
2. A floating rate note, $40m principal, pays LIBOR annually, 4 year maturity
The cash flows are shown in Figure 16.3.
These flows match those for a swap with notional principal of $40m and a
fixed rate of 6%.
b. Plain Vanilla Pay-Fixed
The swap is equivalent to:
-issue bond (go short) a fixed-coupon bond
-but (go long) a floating rate note
254
CHAPTER 16. SWAPS
Buy Bond
Sell FRN
40
40
2.4
LIBOR 40
2.4
LIBOR 40
2.4
LIBOR 40
2.4
LIBOR 40
Figure 16.3: Cash Flows
255
16.5. THE VALUATION OF SWAPS
Buy Euro 7%
Issue US 6%
50
$50
3.5
$3
3.5
$3
3.5
$3
3.5
$3
Figure 16.4:
2. Currency Swaps
a. Fixed-for-Fixed Currency Swap
-buy a bond in one currency
-issue bond denominated in another
b. Plain Vanilla Currency Swap
-one bond fixed coupon
-one floating rate note
16.5.2
Implications
1. Motive for swaps?

Economize on cost of these bond portfolios
2. Pricing of swaps?
Since they can be replicated by bonds, must be related to interest rates on
bonds
256
16.6
CHAPTER 16. SWAPS
Interest Rate Swap Pricing
The essential item to be determined in pricing an interest rate swap is to set

the fixed interest rate so, given that the other party pays LIBOR, the swap is
fair.
To see how the argument functions, consider a plain vanilla interest rate
swap. The receive-fixed party pays LIBOR. The fixed rate has to be set so that
there are no arbitrage opportunities. Define the SFR as the Swap Fixed Rate.
This is the fixed rate that will be constructed to make the swap fair.
For there to be no arbitrage, the two flows of payments over the life of the
swap must have the same present value. This present value has to be computed
using the rates of interest observed in the market. Fundamental to this process
is the term structure and the implied forward rates. The term structure is the
set of spot interest rates for spot loans of different lengths. These spot rates
imply the forward rates. This was covered in Chapter 12.
Consider a swap with notional principal of $1m and a tenor of 4 years. The
floating interest rate in each year is predicted by the forward rate. Note that
these rates are all observed at the time the swap is organized and contracts
can be made to borrow and lend at these rates of interest. They need not, and
almost certainly will not, be the rates that actually hold when the future periods
are reached but they are the best predictor at the start of the swap.
Year Floating Rate Fixed Rate
1
f0,1
SF R
2
f1,2
SF R
3
f2,3
SF R
4
f3,4
SF R
Table 16.5: Interest Rates
Using the interest rates in Table 16.5, the present value of the cash flows
must be equal. Given that the value of the notional principal is $1m, the present
value of the series of floating interest payments is
P V (f loating) =
f1,2
f2,3
f3,4
f0,1
+
+
+
.
1 + s1 [1 + s2 ]2 [1 + s3 ]3 [1 + s4 ]4
(16.1)
The present value of the fixed interest payments is

P V (fixed) =
SF R
SF R
SF R
SF R
+
+
+
.
1 + s1 [1 + s2 ]2 [1 + s3 ]3 [1 + s4 ]4
(16.2)
Equating these two present values and solving, the SF R can be found to be
3
fn,n+1
n=0 [1+sn+1 ]n+1
1
m=0 [1+sm ]m
SF R = 4
(16.3)
This is the swap fixed rate that leads to no arbitrage being possible since it
equates the present values.
257
16.7. CURRENCY SWAP
Note further that the relation between spot rates and forward rates makes
it possible to translate between the two. In particular,
1 + s1 = 1 + f0,1 ,
(16.4)
[1 + s2 ] = [1 + f0,1 ] [1 + f1,2 ] ,
(16.5)
[1 + s3 ]3 = [1 + f0,1 ] [1 + f1,2 ] [1 + f2,3 ] ,
(16.6)
[1 + s4 ]4 = [1 + f0,1 ] [1 + f1,2 ] [1 + f2,3 ] [1 + f3,4 ] ,
(16.7)
Using these relations, SF R can be expressed either:

1. In terms of spot rates
or
2. In terms of forward rates.
Example 153 Let the spot rates be s1 = 4%, s2 = 5%, s3 = 6%, s4 = 7%. Then
f0,1 = 4%, f1,2 = 6%, f2,3 = 8%, f3,4 = 10%. So
SF R =
0.04
1.04
1
1.04
+
+
0.06
[1.05]2
1
[1.05]2
+
+
0.08
[1.06]3
1
[1.06]3
+
+
0.1
[1.07]4
1
[1.07]4
= 0.068 (6.8%)
In general, if interest is paid at intervals of length and the tenor of the

swap is N , then the formula for the swap fixed rate can be generalized to
SF R =
N
f[n1] ,n
n=1
z0,n
N
1
m=1 z0,m
(16.8)
where z0,n is the discount factor between time 0 and time n .

These results determine what the fixed rate should be in the swap to match
the floating LIBOR.
16.7
Currency Swap
With a currency swap there is the additional feature of changes in the exchange
rate. This requires an extension to the analysis. The extension has to relate
the swap fixed rates in the two countries to the term structure in both countries
and the exchange rates.
16.7.1
Interest Rate Parity
Consider two countries A and B. The information that is available at the

initiation of the swap consists of:
1. The term structure in A
2. The term structure in B
3. The rates for foreign exchange between the currencies of the two countries.
258
CHAPTER 16. SWAPS
(1 + s )
A t
t
Conversion to B
AB 0 , 0
1 + stB
BA e0,t
Interest in B
Conversion to A
Figure 16.5: Interest Rate Parity

Under (3) we observe both the spot exchange rates and the forward exchange
rates. Forward exchange rates give the rate now for an agreed currency exchange
at a fixed date in the future.
The notation is to use AB e0,0 to denote the value at time 0 for currency A
in terms of currency B for delivery at 0. This is the spot exchange rate. For
instance, if 1(currency A) = $1.5 (currency B) then AB e0,0 = 1.5.
Similarly, the notation AB e0,t denotes the value contracts made at time 0 for
currency A in terms of currency B for delivery of the currency at time t. This
is a forward exchange rate.
These exchange rates do not stand alone but are related via the spot rates of
interest. This is a consequence of the fact that transactions can be undertaken
to trade the currencies at spot and forward rates.
Consider the following two investment strategies:
Invest 1m in country A for t years
Convert 1m to currency of country B and invest for t years and enter
forward to convert back
The basis of this strategy is that all the interest rates and exchange rates
are known at time 0 so the cash flows are certain. The fact that everything
is certain implies that the payoffs of the two strategies must be the same. If
they were not, then arbitrage would take place. The two strategies are shown
in Figure 16.5.
To eliminate the possibility of arbitrage it must be the case that
t

t

=AB e0,0 1 + sB
(16.9)
1 + sA
t
t BA e0,t ,
so that given the spot rates it is possible to calculate the currency forward rates.
These currency forward rates can then be used these to obtain the present value
of a swap deal at the initiation of the swap.
The claim made here is that interest rate parity connects SF RA to SF RB .
If it did not then there would be arbitrage between the currencies of the two
259
16.7. CURRENCY SWAP
Country A
Term
Structure
Country B
Exchange
Rate
SFR A
Term
Structure
SFR B
Interest Rate
Parity
Figure 16.6: Interest Rates and Exchange Rates
countries. Therefore it is possible to use the SF R in each country as the fixed

rate in a currency swap.
16.7.2
Fixed-for-Fixed
Consider a fixed-for-fixed swap involving an exchange of dollars for a foreign

currency.
Let A be the party that receives dollars and pays a fixed rate on these dollars.
Let B be the party that receives the foreign currency and also pays a fixed
rate on this foreign currency.
To determine the fair value of the swap, the issue is to determine what fixed
rates should be used.
The answer that will be demonstrated is that:
Party A: pays dollar SFR - the SFR on a corresponding dollar plain vanilla
interest rate swap
Party B: pays foreign SFR - the SFR on a corresponding foreign plain
vanilla interest rate swap
Doing this ensures the present value of expected cash flows for A and B are
zero.
Two demonstrations of this are given. The first is taken from the text by
Kolb and involves adopting a set of numbers and evaluating an example. The
second demonstration shows the result algebraically for a swap with a very short
tenor.
260
CHAPTER 16. SWAPS
Demonstration A
Consider a $ for Dm swap. Assume that the spot rate for current exchange is $1
= DM2.5. Let the principal on the swap be $100m. This is equal to DM250m
at the initial spot rate. The tenor of the swap is 5 years.
The first step in constructing the correct values of the SFRs is to use the
term structure in each country to generate the implied path of the exchange
rate. The interest rate parity argument in (16.9) gives the relationship between
the spot rates and exchange rates as
t

1 + sDM
t
(16.10)

.
$DM e0,t =
$ t
DM$ e0,0 1 + st
This formula determines the forward exchange rates for the two currencies. It
should also be noted that by definition, the two exchanges rates are related by
DM$ e0,0
1
.
$DM e0,0
(16.11)
To allow a numerical demonstration, Table 16.6 assumes values for the $ and
DM term structures. Combining these with interest rate parity, the implied path
of the exchange rate can be derived. This is given in the final column.

t

t
1 + s$t
1 + sDM
Year s$t
sDM
$DM e0,t
t
t
0
2.50
1
0.08
1.08
0.05
1.05
2.430
2
0.085 1.177
0.052 1.106
2.349
3
0.088 1.289
0.054 1.171
2.71
4
0.091 1.421
0.055 1.240
2.181
5
0.093 1.567
0.056 1.315
2.097
Table 16.6: Term Structures and Exchange Rate
The second step is to use the term structure to calculate the implied set of
forward rates. These are shown in Table 16.7.
$
0.08
0.089
DM
0.5
0.053
0.095
0.058
0.102
0.059
f0,1 = s1
2
2)
f1,2 = (1+s
(1+s1 ) 1
f2,3 =
f3,4 =
(1+s3 )3
(1+s2 )2
(1+s4 )4
(1+s3 )3
(1+s5 )5
(1+s4 )4
f4,5 =
1 0.103 0.0605
Table 16.7: Forward Rates
The third step is to use these forward rates to generate the swap fixed rates
through the formula
5
ft1,t
t=1 (1+st )t
1
t=1 (1+st )t
SF R = 5
(16.12)
261
16.7. CURRENCY SWAP

Using the values in Table 16.7, the two swap fixed rates are
SF R$
0.074 + 0.076 + 0.074 + 0.072 + 0.066

0.926 + 0.850 + 0.776 + 0.704 + 0.638
= 0.929,
=
(16.13)
and
SF RDM = 0.056.
(16.14)
The SF R$ in (16.13) is the value that would be used in a $ interest rate swap
and the SF RDM in (16.14) is the rate that would be used in a DM interest rate
swap. These are the values that are consistent with the elimination of arbitrage
possibilities and make the swap fair for both parties.
The fact that these are the correct SFRs can be shown by using these values
to determine the expected cash flows during the life of the swap It should be
noted that these are the flows expected given the observed term structures. If
future interest rates are not as implied by the term structure, then the actual
cash flows will be different.
TABLE OF APPLICATION
The conclusion derived from observing the figures in this table is that these
SFR values do give a fair price so the swap is of fair value for both parties.
The initial present value of the swap, evaluated using interest rate parity to
determine the exchange rates, is zero for both parties.
Demonstration B
The second demonstration that the SF R is the correct rate to use undertakes
the calculations using the general definitions of the variables.
Consider a swap of DM for $ with a two-year tenor. Table 16.8 states the
cash flows for the two parties involved with the swap per $ of principal.
Year
0
1
2
DM cash flow
$ cash flow
DM value of $
$DM e0,0
1
$DM e0,0
DM
$
$
R
SF
R
$DM e0,0 SF

$DM e0,1 SF
R

DM
$
1 + SF R
$DM e0,2 1 + SF R$
$DM e0,0 1 + SF R
Table 16.8: Cash Flows
The next table presents the net DM cash flows.

Year
0
1
2
Net DM cash flow

$DM e0,0 $DM e0,0 = 0
DM
$DM e0,1 SF R$
$DM e0,0 SF R

DM
$DM e0,2 1 + SF R$
$DM e0,0 1 + SF R
Table 16.9: Net Cash Flows
Discount on DM
1
1
(1+sDM
)
1
1
(1+sDM
)2
2
262
CHAPTER 16. SWAPS
The present value of the DM cash flow is

1
DM
$

e
SF
R
e
SF
R
PV =
0,0
0,1
$DM
$DM
1 + sDM
1

1
DM
$
1
+
SF
R
+
e
e
1
+
SF
R
. (16.15)

0,0
0,2
$DM
$DM
2
1 + sDM
2
By definition, the forward exchange rates are

$DM e0,1
$DM e0,2
and

1 + sDM
1

,
=
$
DM$ e0,0 1 + s1
(16.16)
2

1 + sDM
2
=

,
$ 2
DM$ e0,0 1 + s2
DM$ e0,0
(16.17)
1
.
$DM e0,0
(16.18)
Using these exchange rates, the present value is

DM
1
+
s
1
DM
SF R$
$DM e0,0
$DM e0,0 SF R
1 + s$1

2

1 + sDM
1
2
DM
$
DM$ e0,0
,
+
2 $DM e0,0 1 + SF R
2 1 + SF R
1 + sDM
1 + s$2
2
(16.19)
1

PV =
1 + sDM
1
or, simplifying this expression,

1 + SF RDM
1 + SF R$
SF RDM
SF R$

+
+
P V =$DM e0,0

.
2
2
1 + sDM
1 + s$1
1 + sDM
1 + s$2
1
2
(16.20)
The swap fixed rate is defined by
SF R =
f0,1
1+s1
1
1+s1
+
+
f1,2
(1+s2 )2
1
(1+s2 )2
(1+s )2
=
=
2
1+s1 1
s1
1+s1 + (1+s2 )2
1
1
1+s1 + (1+s2 )2
(1 + s2 )2 (1 + s1 )
(1 + s1 )
(1 + s1 ) + (1 + s2 )
(16.21)
263
16.8. CONCLUSIONS
The SF R can be substituted into the definition for present value (16.20) to give
PV
=
=
=
2
(1+sDM
)+(1+sDM
) +(1+sDM
)
1
2
1
(1+sDM
)(1+sDM
) 2
1
2
$DM e0,0
$
$
$ 2
$
SF R (1+s1 )+(1+s2 ) +(1+s1 )
$
$ 2
(1+s1 )(1+s2 )
DM 2
)(1+sDM
)+(1+sDM
)
(1+s2 ) (1+sDM
1
1
1
(1+sDM
)(1+sDM
)2
1
2
$DM e0,0
$ 2
$
$
1+s
1+s
1+s
( 2 ) ( 1 ) ( 1 )+(1+s$1 )
2
(1+s$1 )(1+s$2 )
$DM e0,0 [1 1]
0.
(16.22)
SF RDM
This completes the demonstration that the present value of the swap is zero.
16.7.3
Pricing Summary
This use of the SF R in a fixed-for-fixed swap provides the insight necessary to

understand the interest rates used in other swaps.
A convenient summary of the results is the following:
1. Fixed-for-Fixed
Both parties pay the SFR for the currency received.
2. Floating-for-Fixed
The pay-floating party pays LIBOR, and the pay-fixed pays SFR (The LIBOR rate is that on the currency received).
3. Fixed-for-Floating
The pay-fixed party pays SFR, and the pay-floating pays LIBOR (The LIBOR rate is that on the currency received).
4. Floating-for-Floating
Both parties pay the LIBOR on the currency received.
16.8
Conclusions
This chapter has introduced swaps and the swap markets. It has also been
shown how these swaps can be priced by setting the swap fixed rate to give the
swap the same present value for the two parties on either side of the swap.
Exercise 105 Assume that a US and UK firm engage in a currency swap. Let
the spot exchange rate at the time of the swap be 1 = $1.60, the LIBOR rate
be 5% and the fixed UK rate be 6%. If the principal is 10m, chart the cash
flows for the two parties when the tenor is 5 years.
Exercise 106 Consider a swap dealer with the following swap book.
264
CHAPTER 16. SWAPS
Notional Principal
T enor
F ixed Rate
Dealers Position
( million)
(Y ears)
(%)
A
10
4
7
Receive-Fixed
B
35
3
6.5
Pay-Fixed
C
20
5
7.25
Pay-Fixed
D
40
4
7.5
Receive-Fixed
E
15
1
6.75
Receive-Fixed
If the applicable LIBOR rate is currently 5% but rises 1% per year, determine
the yearly cash flow of the dealer if no further deals are made.
What should the dealer do to reduce their risk? [6 marks]
Swap
Exercise 107 Consider the following term structures:

Year
0
1
2
3
US
5% 6% 7% 8%
UK 3% 4% 5% 4%
(i) If the current exchange rate is 1 = $1.5, find the fixed interest rate that
would be paid on a plain-vanilla currency swap.
(ii) Determine the cash flows for the currency swap above if the principal is
100m, and show that the present value of the net flow is 0 for the firm receiving
.
Part VII
Application
265
Chapter 17
Portfolio Evaluation
17.1
Introduction
This must tie together some of the various components.

The basic issue will be to go through the investment process of selection,
construction, investment and evalaution.
17.2
Portfolio Consturction
Could do this in a retrospective form

i.e. look at data in year 2000 to select a couple of different protfolios using
the techniques descibed
one low risk, one high risk.
Can be related to two different people with different requirements
such as young and old.
17.3
Revision
Then a year later inspect these

Possibly revise
Then check again.
17.4
Longer Run
Bring up to the year 2005 to see how they perform.
17.5
Conclusion
Look at the issues that have been learnt.

267
268
CHAPTER 17. PORTFOLIO EVALUATION
Exercise 108 Must do something similar as an exercise.
Part VIII
Appendix
269
Chapter 18
Using Yahoo!
18.1
Introduction
Examples throughout use Yahoo to get data.
18.2
Symbols
How to find symbol.
18.3
Research
The background information on the company
18.4
Stock Prices
How to quickly find stock prices
18.5
Options
How to find option prices and interpretation
271

Investment Anal 1

Uploaded by

Copyright:

Available Formats

Investment Anal 1

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Investment Anal 1

Uploaded by

Copyright:

Available Formats

Investment Analysis

1 Securities and Analysis

2 Buying and Selling

2.5.3 Margin and Return . . . . . . . . . . . . . . . . . . . . . .

3 Risk and Return

Single Index Model

Capital Asset Pricing

8.8 Market Portfolio . . . . . . . . . . . . . . . . . . . . . . . . . 154

9 Arbitrage Pricing Theory

11 Efficient Markets and Behavioral Finance

Fixed Income Securities

12 Interest Rates and Yields

15 Forwards and Futures

Securities and Analysis

This book is about the investment of wealth in financial securities. It provides

CHAPTER 1. SECURITIES AND ANALYSIS

1.2. FINANCIAL INVESTMENT

with the expectation that it will lead, ultimately, to a preferred pattern of

CHAPTER 1. SECURITIES AND ANALYSIS

Example 2 The Association for Investment Management and Research (AIMR)

A security can be defined as:

final value of investment initial value of investment

Example 3 At the start of 2003 an investor purchased securities worth $20000.

CHAPTER 1. SECURITIES AND ANALYSIS

It is a fundamental assumption of investment analysis that investors wish

1.6. MARKETABLE SECURITIES

Money Market Securities

Money market securities are short-term debt instruments sold by governments,

CHAPTER 1. SECURITIES AND ANALYSIS

Negotiable Certificates of Deposit

1.6. MARKETABLE SECURITIES

Capital Market Securities

Capital market securities include instruments having maturities greater than

CHAPTER 1. SECURITIES AND ANALYSIS

1.6. MARKETABLE SECURITIES

Corporate Bonds Corporate bonds are similar to treasury bonds in their

CHAPTER 1. SECURITIES AND ANALYSIS

Indirect investing can be undertaken by purchasing the shares of an investment

1.7. SECURITIES AND RISK

Securities and Risk

CHAPTER 1. SECURITIES AND ANALYSIS

The Investment Process

1.8. THE INVESTMENT PROCESS

1. Determine Objectives. Investment policy has to be guided by a set of

CHAPTER 1. SECURITIES AND ANALYSIS

4. Portfolio Construction. Portfolio construction follows from security

CHAPTER 1. SECURITIES AND ANALYSIS

Buying and Selling

CHAPTER 2. BUYING AND SELLING

Markets are fundamental for the trading of securities.

Primary and Secondary

Call and Continuous

Auction and Over-the-Counter

In an auction market buyers and sellers enter a bidding process to determine

CHAPTER 2. BUYING AND SELLING

An over-the-counter market involves direct negotiation between broker and

Money and Capital

2.4. TRADING STOCKS

CHAPTER 2. BUYING AND SELLING