You are on page 1of 16

Chapter 4

Foundation of Portfolio Theory


Cheng-Few Lee, Alice C. Lee, and John Lee

Abstract In this chapter, we first define the basic concepts


of risk and risk measurement. Using the relationship of risk
and return, we introduce the efficient-portfolio concept and
its implementation. Then the concept of dominance principle
and performance measures are also discussed and illustrated.
Finally, the interest rate and market rate of return are used as
measurements to show how the commercial lending rate and
the market risk premium can be calculated.
Keywords Risk r Systematic risk r Unsystematic risk r
Return r Efficient frontier r Efficient portfolio r Performance
measure r Risk premium r Sharpe measure r Jensen measure r
Treynor measure

4.1 Introduction
This chapter focuses on the various types of risk. An
understanding of the sources of risk and their determination is very useful for doing meaningful security analysis
and portfolio management; thus, discussing risk prior to an
exploration of portfolio-selection models is essential. In this
chapter methods of measuring risk are defined by using basic statistical methods, and the concepts and applications of
the dominance principle and portfolio theory are discussed in
some detail.
A discussion of the different types of risk and their classification is followed by a review of the concepts and applications of portfolio analysis. The dominance as well as the
necessity of using performance measures as a means to comC.-F. Lee ()
Rutgers University, New Brunswick, NJ, USA
e-mail: lee@business.rutgers.edu
A.C. Lee
State Street Corp., Boston, MA, USA
e-mail: alice.finance@gmail.com
J. Lee
Center for PBBEF Research, Hackensack, NJ, USA
e-mail: johnleeexcelvba@gmail.com

pare performance among different portfolios are treated. The


determination of the commercial lending rate in accordance
with risk and return concepts is then analyzed, and the final
section of this chapter concerns calculations of the market
rate of return and the market risk premium.

4.2 Risk Classification and Measurement


Risk is defined as the probability of success or failure. To
be able to measure risk, it is necessary to define the range
of outcomes and the probability that these outcomes will occur. A probability distribution may be determined either subjectively or objectively; a subjective determination is made
by the individual about the likelihood of various future outcomes, while an objective determination involves measuring
past data that are associated with the frequency of certain
types of outcomes.
In either case, a subjective or objective determination
requires a quantitative measure of risk. The measure most
commonly used is the standard deviation or the variance of
the possible returns. In considering two securities, A and B,
it is possible to use a subjective approach, an objective approach, or a combination of the two approaches to indicate
the expected range of outcomes and the probability that each
outcome will occur. Figure 4.1 illustrates the distribution of
return possibilities for these two securities.
While the standard deviation is stated in rates of return,
the variance is stated in terms of the rate of return squared.
As it is more natural to discuss rates of return rather than rates
of return squared, risk is usually measured with the standard
deviation of returns. Nevertheless, for statistical purposes it
is usually more convenient to use the variance rather than
the standard deviation. Either risk measure is appropriate because the standard deviation is merely a simple mathematical
transformation of the variance.
It is the job of the security analyst to provide estimates
of a financial assets risk. Supplying estimates of a security
variance or standard deviation is one method of estimating
the total risk of a security.

C.-F. Lee et al. (eds.), Handbook of Quantitative Finance and Risk Management,
c Springer Science+Business Media, LLC 2010
DOI 10.1007/978-0-387-77117-5_4, 

53

54

C.-F. Lee et al.

Fig. 4.1 Probability distributions


between securities A and B

Why do securities have differing levels of total risk? What


factors cause the returns of securities to vary? Table 4.1
defines various types of risk and in so doing offers answers to these questions, and in analyzing the risks defined in Table 4.1 the following paragraphs provide further
illustration.
At any point a security or a portfolio could be subject to
a number of these risks. In general, the various risks are not
mutually exclusive, nor are they additive. By mutually exclusive is meant that the fact that a portfolio contains a certain

kind of risk does not mean that it also contains other types
of risk. By additive is meant that if a portfolio contains a
number of different types of risk, the total risk of the portfolio is simply the sum of the individual risks. Additionally,
some the risks defined in Table 4.1 are easily quantified or
measured for example using beta to measure systematic
risk while other types of risk may not have well-defined
quantitative risk measures such as management risk or political risk. Finally, it should be noted that the types of risk
listed on Table 4.1 are mutually exclusive.

4 Foundation of Portfolio Theory

55

Table 4.1 Types of risk


Risk type

Description

Call risk

The variability of return caused by the repurchase of the security before its stated maturity

Convertible risk

The variability of return caused when one type of security is converted into another type of security

Default risk

The probability of a return of zero when the issuer of the security is unable to make interest and principal
payments or for equities, the probability that the market price of the stock will go to zero when the
firm goes bankrupt

Interest-rate risk

The variability of return caused by the movement of interest rates

Management risk

The variability of return caused by bad management decisions; this is usually a part of the unsystematic
risk of a stock, although it can affect the amount of systematic risk

Marketability risk

The variability of return caused by the commissions and price concessions associated with selling an
illiquid asset

Political risk

The variability of return caused by changes in laws, taxes, or other government actions

Purchasing-power risk

The variability of return caused by inflation, which erodes the real value of the return

Systematic risk

The variability of a single securitys return caused by the general rise or fall of the entire market

Unsystematic risk

The variability of return caused by factors unique to the individual security

4.2.1 Call Risk

4.2.2 Convertible Risk

An investor who purchases a security, whether it is a debt


instrument or equity, usually has a pretty good idea of his or
her investment horizon that is, how long he or she intends
to hold the security. If the issuer calls the bond or repurchases
the stock at a point in time prior to the end of the investors
investment horizon, the return earned by the investor may
be less than expected. The investor is facing call risk (see
Table 4.1 for a detailed definition).
Generally all corporate bonds and preferred stocks are
callable at some point during the life of the security. Investors
do not like this call feature, because bonds are most often
called after interest rates have fallen, so that the issues can
replace them with a new issue of bonds at a lower interest
rate. When a bond is called, instead of receiving the expected
interest flows for the remainder of his or her investment horizon, the investor must reinvest the proceeds received from
the issuer at a lower rate of interest. It is this reinvestment at
a new lower rate of interest that causes the yield or return of
the bond to be different than what the investor expected.
For this reason, in periods of high interest rates, when
interest rates are expected to drop to lower levels in the future, bonds that are call protected for a certain number of
years usually sell at a higher price than callable bonds. Thus,
a deferred call feature can cause an interest spread or differential to develop vis--vis a similar issue with no call feature
or an issue that is immediately callable. The size of the interest differential varies with the investors expectations about
the future movement of interest rates.
Whether a bond is callable immediately after it is issued
or after a deferred period, the actual call of the bond involves
the payment of a premium over the par value of the bond by
the issuer.

If a bond or a preferred stock is convertible into a stated


number of shares of common stock of the corporation issuing
the original security, the rate of return of the investment may
vary because the value of the underlying common stock has
increased or decreased. This kind of investment is facing convertible risk, as described in Table 4.1. A convertible security
normally has a lower coupon rate, or stated dividend (in the
case of preferred stocks), because investors are willing to accept a lower contractual return from the company in order to
be able to share in any rise in the price of the firms common
stock. The size of the yield spread between straight and convertible securities is dependent on the future prospects of the
individual firm and the general level of interest rates.
The difference between the value paid for the conversion
right and the actual returns experienced over the life of the
security increase the variability of returns associated with the
investment.

4.2.3 Default Risk


The quality of the issue or the chances of bankruptcy for the
issuer have a significant impact on the rate of return of a security. A lower-quality issue will sell at a lower price and thus
offer a higher yield than a similar security issued by a higherquality issuer. This is true in all segments of the securities
markets. Hence, most investors hold diversified portfolios
in order to reduce their exposure to any one security going
into default or the issuer going bankrupt. The risk relating to
default or bankruptcy is called default risk, as described in
Table 4.1.

56

4.2.4 Interest-Rate Risk


As the general level of interest rates changes, the value of
an individual security also changes. As interest rates rise, the
value of existing securities falls, and vice versa. Additionally,
longer-term securities are more affected by a given change in
the general level of interest rates than are shorter-term securities. The fluctuation of security returns caused by the movement of interest rates is called interest-rate risk, as described
in Table 4.1.
For example, Bond A has a coupon of 7% and a maturity
of 5 years. If interest rates go from 7 to 8%, the price of A
will fall $6.75 for each $100 of par value, whereas the price
of B will fall only $4.00 for each $100 of par value.
Another factor that affects the amount of price change for
a given interest rate changes is the size of the coupon. The
lower the coupon rate of an issue, the more widely its price
will change for a given change in interest rates. For example,
Bond C is a 20-year bond with a percent coupon and a market
value of $68 when the general level of interest rates are 7%.
Bond D is a 20-year bond with 7% coupon, so its market
value is par $100. If interest rates go from 7 to 8%, the value
of Bond C will fall to $60.38, a decrease in value of 11.3%
based on its market price [(60.3868)/68], whereas Bond Ds
value will fall by 9.8% [(90.2100)/100].
In general, a shorter-maturity, higher-coupon security is
subjected to less interest-rate risk than a long-term, low
coupon security.

4.2.5 Management Risk


Management risk, as indicated in Table 4.1, is caused by
errors of a firms managers when they make business and
financing decisions. Business risk refers to the degree of fluctuation of net income associated with different types of business operations. This kind of risk is related to different types
of business and operating strategies. Financial risk refers to
the variability of returns associated with leverage decisions.
How much of the firm should be financed with equity and
how much should be financed with debt?
It should be noted that both business risk and financial
risk are not necessarily constant over time. Both risks can
be affected by the fluctuations of the business cycle or by
changes in government policy.
Jensen and Meckling have presented a theory, called
agency theory, that deals with the problem of management
errors and their impact on security owners. An agentprincipal relationship exists when decision-making authority is delegated. The modern corporation is a good example
of this phenomenon. The owners or shareholders of the

C.-F. Lee et al.

firm delegate the decision-making authority to the firms


managers, who are in reality employees or agents of the owners. Other things being equal, the managers make decisions
that satisfy the needs of the managers rather than the desires
of the owners unless the firms owners can protect themselves
from management decisions. Some have argued that it is the
function of financial arrangements using bond covenants, options, and so on, to narrow the divergence of goals between
a firms owners and its managers. In any case, it is the possibility of management error or divergent goals that is one of
the causes of variability of return for securities.

4.2.6 Marketability (Liquidity) Risk


The marketability risk (or liquidity risk) of a security (see the
description in Table 4.1) affects the rate of return received by
its owner. Marketability is made up of two components: (1)
the volume of securities that can bought or sold in a short
period of time without adversely affecting the price, and (2)
the amount of time necessary to complete the sale of a given
number of securities. Other things being equal, the less marketable a security, the lower its price or the higher its yield.
A highly liquid security for example, IBM shares traded on
the New York Stock Exchange (NYSE) can be purchased
or sold in large quantities in a very short time. Millions of
shares of IBM are traded every day on the NYSE. The stock
of a small firm traded over the counter (OTC) may trade only
a few hundred shares a week and is said to be illiquid.
One good measure of the marketability of a security is the
spread between the bid and ask prices. The bid price is the
current price at which the security can be sold and the ask
price is the current price at which the security can be bought.
The bid price is always lower than the ask price. The bid-ask
spread is the cost of selling the asset quickly. That is to say,
it is the amount of margin required by the market maker to
stand ready to buy or sell reasonable amounts of the security
quickly. The more illiquid the security the wider the bid-ask
spread.

4.2.7 Political Risk


International political risk (as described in Table 4.1) stems
from the political climate and conditions of a foreign country. For instance, if a government is unstable, as is the case
in some South American countries, there may be wild fluctuations in interest rates due to a lack of confidence by the
population and the business community. There may even be
some chance of sabotage of plant and equipment or other acts

4 Foundation of Portfolio Theory

of terrorism. Contracts may not be upheld or enforceable.


These are all factors that may cause the rate of return on certain securities to vary.
Domestic political risk takes the form of laws, taxes, and
government regulations. As the government changes the tax
law, the returns of securities can be greatly affected. For example, the Tax Reform Act of 1986 probably had a very important role in the bull market of January to August 1987,
when the Dow-Jones Industrial Average went from less than
1,900 to above 2,700.

4.2.8 Purchasing-Power Risk


Purchasing-power risk, as described in Table 4.1, is related
to the possible shrinkage in the real value of a security even
though its nominal value is increasing. For example, if the
nominal value of a security goes from $100 to $200, the
owner of this security is pleased because the investment has
doubled in value. But suppose that, concurrent with the value
increase of 100%, the rate of inflation is 200% that is, a basket of goods costing $100 when the security was purchased
now costs $300. The investor has a money illusion of being better off in nominal terms. The investment did increase
from $100 to $200; nevertheless, in real terms, whereas the
$100 at time zero could purchase a complete basket of goods,
after the inflation only two-thirds of a basket can now be purchased. Hence, the investor has suffered a loss of value.

57

4.3 Portfolio Analysis and Application


Essential to adequate diversification is a knowledge of the
primary concepts of portfolio analysis and their application.
This section focuses on basic analysis and applications; more
advanced portfolio theory and methods will be discussed in
the next chapters.
A portfolio can be defined as any combination of assets or
investments. Portfolio analysis is used to determine the return and risk for these combinations of assets. Portfolio concepts and methods are employed here to formally develop the
idea of the dominance principle and some other measures of
portfolio performance.

4.3.1 Expected Return on a Portfolio


The rate of return on a portfolio is simply the weighted
average of the returns of individual securities in the portfolio. For example, 40% of the portfolio is invested in a security
with a 10% expected return (security A); 30% is invested in
security B with a 5% expected return; and 30% is invested in
security C with a 12% expected return. The expected rate of
return on this portfolio can be expected:
RN p D Wa RN a C Wb RN c C Wc RN c
D .0:4/.0:1/ C .0:3/.0:5/ C .0:3/.0:12/
D 0:091

4.2.9 Systematic and Unsystematic Risk


In Table 4.1, total risk was defined as the sum of systematic
and unsystematic risk. Total risk is also equal to the sum of
all the risk components just discussed. However, the importance and the contribution to total risk depends on the type of
security under consideration. The total risk of bonds contains
a much larger fraction of interest-rate risk than the total risk
of a stock. Each of the types of risk discussed may contain a
systematic component that cannot be diversified away and an
unsystematic component that can be reduced or eliminated,
if the securities are held in a portfolio.
It is assumed throughout this text that rational investors
are average to risk, whatever its source; hence a knowledge
of various risk-management techniques is essential. Toward
that end this chapter now focuses on the management of risk
through diversification.

in which Wa ; Wb , and Wc are the percentages of the portfolio


invested in securities A, B, and C, respectively. The summation of these weights is equal to one. Rp represents the
expected rate of return for the portfolio and is the weighted
average of the securities expected rate of return. In general, the expected return on an n-asset portfolio is defined
by Equation (4.1):
RN p D

n
X

RN i Wi

(4.1)

i D1

where:
n
P
i D1

Wi D 1
Wi D the proportion of the individuals investment
allocated to security i I and
RN i D the expected rate of return for security i:

58

C.-F. Lee et al.

4.3.2 Variance and Standard Deviation


of a Portfolio

Var.W1 R1t C W2 R2t / D

t D1

The riskiness of a portfolio is measured by the standard


deviation (or variance) of the portfolio. To calculate the standard deviation of a portfolio, we should first identify the
covariance among the securities within the portfolio. The covariance between two securities used to formulate a portfolio
can be defined as:

Cov.W1 R1 ; W2 R2 / D

N
X
.W1 R1t  W1 RN 1 /.W2 R2t  W2 RN 2 /
N 1
tD1

D W1 W2

N
X
.R1t  RN 1 /.R2t  RN 2 /
N 1
tD1

D W1 W2 Cov.R1 ; R2 /

N
P

(4.2)

D .0:4/.0:1/ C .0:6/.0:05/
.0:4/.0:15/ C .0:6/.0:1/ 2 =2
C.0:4/.0:15/ C .0:6/.0:1/
.0:4/.0:15/ C .0:6/.0:1/ 2 =2
C.0:4/.0:2/ C .0:6/.0:15/
.0:4/.0:15/ C .0:6/.0:1/ 2 =2
.0:12  0:12/2
.0:07  0:12/2
C
D
2
2
.0:17  0:12/2
C
2
Varportfolio D 0:0025

The riskiness of a portfolio can be measured by the standard


deviation of returns, as indicated in Equation (4.3):

where:
R1t D the rate of return for the first security
in period tI
R2t D the rate of return for the second security
in period tI
N
N
R1 and R2 D average rates of return for the first security
and the secondsecurity; respectivelyI and
Cov.R1 ; R2 / D the covariance between R1 and R2 :
The covariance as indicated in Equation (4.2) can be used to
measure the covariability between two securities (or assets)
when they are used to formulate a portfolio. With this measure the variance for a portfolio with two securities can be
derived:
Var.W1 R1t C W2 R2t /
2

N
P
.W1 R1t C W2 R2t /  .W1 RN 1 C W2 RN 2 /
D
N 1
tD1

.W1 R1t CW2 R2t /.W1 RN 1 CW2 RN 2 / 2


N 1

p D

v
u N
uP
N 2
u
t t D1 .Rpt  Rp /
N 1

where p is the standard deviation of the portfolios return


and RN p is the expected return of the n possible returns.
Figure 4.2 illustrates possible distributions for two portfolios, assuming the same expected return but different levels
of risk. Since portfolio Bs variability is greater than that of
portfolio A, investors regard portfolio B as riskier than portfolio A.

4.3.3 The Two-Asset Case


To explain the fundamental aspect of the risk-diversification
process in a portfolio, consider the two-asset case. Following
Equation (4.3):

Example 4.1 provides further illustration.


Example 4.1. If you have two securities, you can use
Equation (4.2) to calculate the portfolio variance as follows:
Security 1

Security 2

W1 D 40%
t D1
RN 1t D 10%
D2
15%
D3
20%
RN 1 D 15%

W2 D 60%
t D1
RN 2t D 5%
D2
10%
D3
15%
RN 1 D 10%

(4.3)

Fig. 4.2 Probability distributions of returns for two portfolios

4 Foundation of Portfolio Theory

s
p D
D

N
P
t D1

N
P
tD1

59

If 12 D 1, Equation (4.7) reduces from:

.Rpt RN p /2
N 1
W21 .R1t RN 1 /2 CW22 .R2t RN 2 /2 C2W1 W2 .R1t RN 1 /.R2t RN 2 /

@p
D W12 12 C .1  W1 /2 22 C 2W1 .1  W1 /12 1 2 1=2
@W1

2W1 12  2.1  W1 /22 C 2.1  2W1 /12 1 2


R
W1 12 22  2 12 1 2  12  12 1 2
D
W12 12 C .1  W1 /2 22 C 2W1 .1  W1 /12 1 2

N 1

q
D W21 Var.R1t / C W22 Var.R2t / C 2W1 W2 Cov.R1 ; R2 /
(4.4)

where W1 C W2 D 1. By the definitions of correlation


coefficients between R1 and R2 .12 /, the Cov R1 ; R2 can be
rewritten:
(4.5)
Cov.R1 ; R2 / D 12 1 2
where 1 and 2 are the standard deviations of the first and
second security, respectively. From Equations (4.4) and (4.5),
the standard deviation of a two-security portfolio can be
defined:
p
p D qVar.W1 R1t C W2 R2t /
D W12 12 C .1  W1 /2 22 C 2W1 .1  W1 /12 1 2
(4.6)
Example 4.2 provides further illustration.

to:

2 .2  1 /
.2  1 /.2  1 /
2
D
2  1

W1 D

Returning to the Examples for securities 1 and 2, since the


standard deviations are equal and the securities are perfectly
positively correlated, Equation (4.7a) indicates that there is
no value for W1 that will minimize the portfolio variance. The
weight from W1 that gives the minimum-variance portfolio is
0.05/(0.050.05) or 0.05/0, which is undefined.
If 12 D 1, Equation (4.7) reduces to:
W1 D

Example 4.2. For securities 1 and 2 used in the previous example, applying Equation (4.6) should give the same results
for portfolio variance.
Security 1

Security 2

12 D 0:0025
W1 D 0:4
12 D C1

22 D 0:0025
W2 D 0:6

(4.7a)

2 .2 C 1 /
.1 C 2 /.1 C 2 /
2
.2 C 1 /

(4.7b)

However, if the correlation coefficient between 1 and 2 is 1,


then the minimum-variance portfolio must be divided equally
between security 1 and security 2 that is:
0:05
0:05 C 0:05
D 0:5

W1 D

p
p D p.0:4/2 .0:0025/C.0:6/2 .0:0025/C2.0:4/.0:6/.1/.0:05/.0:05/
D 0:0025

p D 0:05 or Var portfolio D 0:0025, the same answer as for


Example 4.1.
If 12 D 1:0 Equation (4.6) can be simplified to the linear
expression:
p D W1 1 C W2 2

As an expanded form of Equation (4.6), a portfolio can be


written:
31=2
2
n
n
n1 X
X
X
p D 4
W12 12 C 2
Wi Wj ij i j 5
i D1

where W2 D .1  W1 /. Since Equation (4.6) is a quadratic


equation, some value of W1 minimizes p . To obtain this
value, differentiate Equation (4.6) with respect to W1 and set
this derivative equal to zero. Solving for W1 :
2 .2  12 1 /
W1 D 2
1 C 22  212 1 2

D4

n
n X
X

i D1 j Di C1

31=2
Wi Wj Cov.Ri t ; Rjt /j5

(4.8)

j D1 i D1

where:
(4.7)

A condition needed for Equation (4.7) is that W1  1; that


is, no more than 100% of the portfolio can be in any single security, and negative positions (short positions) can be
maintained on any security.

Wi and Wj D the investors investment allocated to security


i and securityj; respectivelyI
ij D the correlation coefficient between security i
and security j I and
n D the number of securities included in the
portfolio:

60

C.-F. Lee et al.

Since Equation (4.8) has n securities, there are n variance


terms (that is, W12 12 ) and .n2  n/ covariance terms (that
is, Wi Wj 12 i j ). If n D 200, Equation (4.8) will have 200
variance terms and 39,800 covariance terms, and any practical application will require a great amount of information
as well as many computations. This issue will be discussed
in more detail in Chap. 5; in the meanwhile, Example 4.3
provides further illustration.
Example 4.3. Consider two stocks, A and B: RN A D 10;
RN B D 15; A D 4, and B D 6. (1) If a riskless portfolio could
be formed from A and B, what would be the expected return
of Rp ? (2) What would the expected return be if AB D 0?
Solution
1=2

1. p D WA2 A2 C.1  WA /2 B2 C2WA .1  WA /AB A B
If we let AB D 1
p D WA A  .1  WA /B
p D 0 D 4WA  6.1  WA /RB
WA D 3=5
so
RN p D WA RN A C .1  WA /RB

2
3
.10/ C
.15/
D
5
5
D 12%
2. If we let AB D 0
i1=2
h
p D WA2 A2 C .1  WA /2 B2
i
@p
1h
D 2WA A2 C 2.1  WA /B2 .1/
@WA
2h
i1=2
D0

WA2 A2 C .1  WA /2 B2


WA A2  .1  WA /B2 D 0 

WA D B2 =A2 C B2
D B2 = 42 C 62
D 9=13
4
9
.10/ C .15/
RN P D
13
13
D 11:54%

4.4.1 The Efficient Portfolio


By definition, a portfolio is efficient, if there are no other
portfolios having the same expected return at a lower variance of returns. Moreover, a portfolio is efficient if no other
portfolio has a higher expected return as the same risk of
returns.
This suggests that given two investments, A and B, investment A will be preferred to B if:
E.A/ > E.B/ and Var.A/ D Var.B/
or
E.A/ D E.B/ and Var.A/ < Var.B/
where:
E.A/ and E.B/ D the expected returns of A and B: and
Var (A) and Var (B) D their respective variances or risk.
The mean returns and variance of every investment opportunity can be calculated and plotted as a single point on a
mean-standard deviation diagram, as shown in Fig. 4.3.
All points below curve EF represent portfolio combinations that are possible. Point D represents a portfolio of investments with a return RD and risk D . All points above EF
are combinations of risk and returns that do not exist. Point
B would therefore represent risk and return that cannot be
obtained with any combination of investments.
The EF curve is also called the efficient frontier because
all points below the curve are dominated by a point found on
the curve. For instance, suppose a firm is willing to assume a
maximum level of risk D . It can obtain a return of RD with
portfolio D or move to point C on the frontier and receive a
higher return RC with that portfolio. Therefore, C dominates
D because it would be preferred to D. For the same level of
risk, it has a higher return.
A similar argument could be made in terms of risk. If the
firm wants to achieve a return of RA , it will select portfolio

4.4 The Efficient Portfolio and Risk


Diversification
Utilizing the definitions of standard deviation of expected
return of a portfolio discussed previously, this section discussed the concepts of the efficient portfolio and risk diversification.
Fig. 4.3 The efficient frontier in portfolio analysis

4 Foundation of Portfolio Theory

61

A over D, because A represents the same return at a smaller


level of risk or standard deviation: A < B . Therefore,
point D is not efficient but points A and C are. A decision
maker could, therefore, select any point on the frontier and
be secure in knowing that a better portfolio is not available.
Example 4.4 further illustrates this concept.
Example 4.4. To show how the portfolio concepts and methods discussed in this section can be used to do practical analysis, monthly rates of return for January 1980 to December
1984 for Pennzoil and Coca Cola are used as examples. The
basic statistical estimates for these two firms are average
monthly rates of return and the variance-covariance matrix.
The average monthly rates of return for Pennzoil (PZ) and
Coca Cola (CK) are 0.0093 and 0.0306, respectively. The
variances and covariances are listed in the following table.
Variance-covariance matrix
PZ
CK
Pennzoil
Coke

0.0107391
0.000290

0.000290
0.0495952

From Equation (4.7), we have:


0:0495952 C 0:000290
0:0107391 C 0:0495952 C 0:00058
0:0498852
D
0:069143
D 0:8189

W1 D

W2 D 1:0  0:8189 D 0:1811


Using the weight estimates and Equations (4.2) and (4.3):
E.RN P / D .0:8189/.0:0093/ C .0:1811/.0:0306/
D 0:01316
P2 D .0:8189/2.0:0107391/ C .0:1811/2.0:0495952/

4.4.2 Corporate Application of Diversification


The effect of diversification is not necessarily limited to securities but may have wider applications at the corporate level.
Frequently, managers will justify undertaking many product lines because of the effects of diversification. Instead of
putting all of the eggs in one basket, the investment risks
are spread out among many lines of services or products in
hope of reducing the overall risks involved and maximizing returns. To what degree this diversification takes place
in other types of corporate decisions depends on the decision
makers preference for risk and return. The overall goal is to
reduce business risk fluctuations of net income. However, it
should be noted that investors can do their own homemade
diversification, which generally reduces the need of corporate diversification.
This type of corporate diversification can be taken to the
multinational level. For example, General Motors has overseas divisions throughout the world. Although these divisions all produce the same product, autos and auto parts,
GMs status as a multinational corporation allows it to take
advantage of the diversifying effects of different exchange
rates and political and economic climates.

4.4.3 The Dominance Principle


The dominance principle has been developed as a means of
conceptually understanding the risk/return tradeoff. As with
the efficient-frontier analysis, we must assume an investor
prefers returns and dislikes risks. For example, as depicted
in Fig. 4.4, if an individual is prepared to experience risk associated with A , he or she can obtain a higher expected return with portfolio A.RN A / than with portfolio B.RN B /. Thus
A dominates B and would be preferred. Similarly, if an individual were satisfied with a return of RN B , he or she would
select portfolio B over C because the risks associated with B

C2.0:8189/.0:1811/.0:000290/
D 0:0088
P D 0:0940
when 12 is less than 1.00 it indicates that the combination
of the two securities will result in a total risk less than their
added respective risks. This is the diversification effect. If 12
were equal to 1.00, this would mean that the combination of
the two securities has no diversification effect at all.
The correlation coefficient of 12 D 0:0125659 indicates that a portfolio combining Pennzoil and Coca Cola
would show a diversification effect and a reduction in risk.

Fig. 4.4 The dominance principle in portfolio analysis

62

C.-F. Lee et al.

are less .B < C /. Therefore using the dominance principle reinforces the choice of an efficient portfolio and is also
a method in determining it.
Figure 4.4 makes clear that points A and B are directly
comparable because they have a common standard deviation,
A . Points B and C are directly comparable because of a
common return, RN B . Now consider portfolio D. How does
its risk versus return compare with the other portfolios shown
in Fig. 4.4? It is difficult to say because the risk and return
are not directly comparable using the dominance principle.
This is the basic limitation of the dominance principle that
portfolios without a common risk or return factor are not directly comparable.

4.4.4 Three Performance Measures


For such a situation it becomes necessary to use some
other performance measure. There are basically three important portfolio performance measures taught in investment
courses: (1) the Sharpe, (2) the Treynor, and (3) the Jensen
measures.
The Sharpe measure (SP) (Sharpe 1966) is of immediate concern. Given two of the portfolios depicted in Fig. 4.4,
portfolios Band D, their relative risk-return performance can
be compared using the equations:
SPD D

RN D  Rf
RN B  Rf
and SPB D
D
B

where:
SPD ; SPB
RN D ; RN B
Rf
D ; B

D Sharpe performance measuresI


D the average return of each portfolioI
D risk-free rateI and
D the respective standard deviation on risk
of each portfolio:

Because the numerator is the average return reduced by the


risk-free rate, it represents the average risk premium of each
portfolio. Dividing the risk premium by the total risk per
portfolio results in a measure of the return (premium) per
unit of risk for each portfolio. The Sharpe performance measure equation will therefore allow a direct comparison of any
portfolio, given its risk and returns.
Consider Fig. 4.5; portfolio A is being compared to portfolio B. If a riskless rate exists, then all investors would prefer A to B because combinations of A and the riskless asset
give higher returns for the same level of risk than combinations of the riskless asset and B. The preferred portfolio lies on the ray passing through Rf that is furthest in
the counterclockwise direction (the ray that has the greatest
slope). Example 4.5 further illustrates this concept.

Fig. 4.5 Combinations of portfolio and the risk-free investment

Example 4.5. An insurance firm is trying to decide between


two investment funds. From past performance it was able
to calculate the average returns and standard deviations for
these funds. The current T-bill rate is 9.5% and the firm will
use this as a measure of the risk-free rate.

Average return R (percent)


Standard deviation  (percent)

Smyth fund

Jones fund

18
20

16
15

Risk-free rate D Rf (percent) D 9:5


Using the Sharpe performance measure, the risk-return
measurements for these two firms are:
0:18  0:095
D 0:425
0:20
0:16  0:095
D 0:433
D
0:15

SPSmyth D
SPJones

It is clear that the Jones fund has a slightly better performance and would be the better alternative of the two.
The Sharpe measure looks at the risk-return decision from
the point of view of an investor choosing a portfolio to represent the majority of his or her investment. An investor choosing a portfolio to represent a large part of his or her wealth
would likely be concerned with the full risk of the portfolio,
and the standard deviation is a measure of that risk.
On the other hand, if the risk level of the portfolio is already determined by the investor and what is important is to
evaluate the performance of the portfolio over and above the
total market performance, perhaps the proper risk measure
would be the relationship between the return on the portfolio
and the return on the market, or beta. All combinations of a
riskless asset and a risky portfolio lie on a straight line con-

4 Foundation of Portfolio Theory

63

necting them. The slope of the line connecting the risky asset
A and the risk-free rate is .RN A  Rf /=A . Here, as in the
Sharpe measure, an investor would prefer the portfolio on the
most counterclockwise ray emanating from the riskless asset.
This measure, called the Treynor measure (TP), developed
by Treynor in 1965, examines differential return when beta
is the risk measure. Example 4.6 provides further illustration.

SPA D 0:84

The Treynor performance measure uses the beta coefficient


(systematic risk) instead of total risk for the j th portfolio
.j / as a risk measure. Applications of the TP are similar to
the SP as discussed previously.1
Jensen (1968, 1969) has proposed a measure referred to
as the Jensen differential performance index (Jensens measure). The differential return can be viewed as the difference
in return earned by the portfolio compared to the return that
the capital asset pricing line implies should be earned.
Consider the line connecting the riskless rate and the market portfolio. A manager could obtain any point along this
line by investing in the market portfolio and mixing this with
the riskless asset to obtain the desired risk level. If the constructed portfolio is actively managed, then one measure of
performance is the difference in return earned by actively
managing the portfolio, versus what would have been earned
if the portfolio had been passively constructed of the market
portfolio and the riskless asset to achieve the same risk level.
The slope of the line connecting the riskless asset and the
market portfolio is .RN M  Rf /=M , and the intercept must
be the riskless rate. The beta on the market portfolio is one;
therefore, the CAPM equation results in:

SPB D 0:73

RN P D Rf C .RN M  Rf /P :

Example 4.6. Rank the portfolios shown in the table based


on the Sharpe measure. Assume Rf D 8%. If Rf D 5%,
how does the order change?
Portfolio

Return (percent)

Risk (percent)

A
B
C
D
E

50
19
12
9
8:5

50
15
9
5
1

Solution Sharpe measure SPM D

RN M  Rf


SPC D 0:44
SPD D 0:20
SPE D 0:50
Ranked by the Sharpe measure, A > B > E > C > D.
The Sharpe measure indicates that portfolio A is the most
desirable: it has the highest return per unit of risk.
For Rf D 5 percent
SPA D 0:90
SPB D 0:933
SPC D 0:77
SPD D 0:80
SPE D 0:35

Jensens measure (JM) is the differential return of the managed portfolios actual return less the return on the portfolio
of identical beta that lies on the line connecting the riskless
asset and the market portfolio. Algebraically, Jensens measure is expressed:
JM D RN P  Rf C .RN M  Rf /p
Figure 4.6 depicts portfolio rankings for the Jensen measure,
and Example 4.7 provides further illustration.
Example 4.7. Rank the portfolio in the table according to
Jensens measure:

The order changes to E > B > A > D > C. E is now the


best portfolio as it has the highest return per unit of risk.
The Treynor measure can be expressed by the following:
TP D

RN j  Rf
j

where:
RN j D average return of j th portfolioI
Rf D ris  free rateI and
j D beta coefficient for j th portfolio:

Fig. 4.6 Jensens measure for portfolio rankings

Discussion of the Treynor measure adapted from Treynor (1965).


Adapted by permission.

64

1.
2.
3.
4.

C.-F. Lee et al.

Assuming RM D 10 percent and Rf D 8 percent


Assuming RM D 12 percent and Rf D 8 percent
Assuming RM D Rf D 8 percent
Assuming RM D 12 percent and Rf D 4 percent

Portfolio

Ri (percent)

(percent)

A
B
C
D
E

50
19
12
9
8:5

50
15
9
5
1

2.5
2.0
1.5
1.0
0.25

Solution

2.

3.

4.

RN P  Rf RN M  Rf

P
m

D SPP  SPm .commom constant/l


If the Jensen measure (JM) is divided by P , it is equivalent
TM
to the Treynor measure plus some constant common to all
portfolios:
RN P  Rf RN M  Rf P
JM
D

P
P
P
D TMP  RN M  Rf

Ri D Rf Ci .RM Rf /

1.

RN P  Rf RN M  Rf .pm /
JM
D

p
P
P
m
m

JM D .Ri Rf /i .RM Rf /

JMA D 37 percent
JMB D 7 percent
JMC D 1 percent
JMD D 2 percent
JME D 0 percent
Ranked by the Jensen measure A > B > C > E > D.
JMA D 32 percent
JMB D 3 percent
JMC D  2 percent
JMD D 3 percent
JME D 0:5 percent
The rank change to A > B > E > C > D.
JMA D 42 percent
JMB D 11 percent
JMC D 4 percent
JMD D 1 percent
JME D 0:5 percent
The rank is A > B > C > D > E.
JMA D 26 percent
JMB D 1 percent
JMC D 4 percent
JMD D 3 percent
JME D 2:5 percent
The rank now is A > E > B > D > C.

Interrelationship among Three Performance Measure. It


should be noted that all three performance measures are interrelated. For instance, if pm D pm =p m D 1, then the
Jensen measure divided by p becomes equivalent to the
Sharpe measure. Since
p D pm =m2 and pm D pm =p m
the Jensen measure (JM) must be multiplied by 1=p to derive the equivalent Sharpe measure:

D TMP  commom constant


Example 4.8 provides further illustration.

Example 4.8. Continuing with the example used for the


Sharpe performance measure in Example 4.5, assume that in
addition to the information already provided, the market return is 10%, the beta of the Smyth Fund is 0.8, and the Jones
Fund beta is 1.1. Then, according to the capital asset pricing
line, the implied return earned should be:
RN Smyth D 0:095 C .0:10  0:095/.0:8/ D 0:099
RN Jones D 0:095 C .0:10  0:095/.1:1/ D 0:1005
Using the Jensen measure, the risk-return measurements for
these two firms are:
JMSmyth D 0:18  0:099/ D 0:081
JMJones D 0:16  0:1005 D 0:0595
From these calculations, it is clear that the Smyth Fund has a
better performance and would be the better alternative of the
two. Note that this is the opposite of the results determined
from the Sharpe performance measure in Example 4.5. Computing the Treynor measure would reinforce the Jensen results. More analysis of these performance measures will be
undertaken in detail in next chapters.

4.5 Determination of Commercial


Lending Rate
This section concerns a process for estimating the lending
rate a financial institution would extend to a firm or the borrowing rate a firm would think is reasonable based on economic, industry, and firm-specific factors.

4 Foundation of Portfolio Theory

65

As shown previously, part of the rate of return is based


on the risk-free rate. The risk-free rate Rf must first be
forecasted for three types of economic conditions boom,
normal, and poor.
The second component of the lending rate is the risk
premium .RP /. This can be calculated individually for each
firm by examining the change in earnings before interest and
taxes (EBIT) under the three types of economic conditions.
The EBIT is used by the lender as an indicator of the ability
of the potential borrower to repay borrowed funds.
Table 4.2 has been constructed based on the methods discussed previously. In total there are nine possible lending
rates under the three different economic conditions. The construction of these lending rates is shown in Table 4.3. This
table shows that during a boom the risk-free rate is set at
12%, but the risk premium can taken on different values.
These is a 40% chance that it will be 3.0%, a 30% chance it
will be 5.0%, and a 30% chance it will be 8.0%. The products
of the RP probabilities and the Rf probability are the joint
probabilities of occurrence for the lending rates computed
from these parameters. Therefore, there is a 10% chance that
a firm will be faced with a 15% lending rate during a boom,
a 7.5% chance of a 17% rate, and a 7.5% chance of an 18%

Table 4.2 Possible lending rates


Economic
conditions
Rf (percent)

Probability

Boom

12.0

0.25

Normal

10.0

0.50

Poor

8.0

0.25

rate. This process applies for the other conditions, normal


and poor, as well.
Based upon the mean and variance Equations (4.1)
and (4.2) it is possible to calculate the expected lending rate
and its variance. Using the information provided in Table 4.3,
the weighted average can be calculated:
R D.0:100/.15%/ C .0:075/.17%/ C .0:075/.20%/
C .0:200/.13%/
C .0:150/.15%/ C .0:150/.18%/ C .0:100/.11%/
C .0:075/.13%/
C .0:075/.16%/
D15:1%
With a standard deviation of:
h
 D .0:100/.15  15:1/2 C .0:075/.17  15:1/2
C .0:075/.20  15:1/2
C .0:200/.13  15:1/2 C .0:150/.5  15:1/2

EBIT
($ millions)

Probability

Rp (percent)

2.5
1.5
0.5
2.5
1.5
0.5
2.5
1.5
0.5

0.40
0.30
0.30
0.40
0.30
0.30
0.40
0.30
0.30

3
5
8
3
5
8
3
5
8

Table 4.3 Construction of actual lending rates


Economic
conditions

(A)
Rf (percent)

(B)
Probability

(C)
Rp (percent)

(D)
Probability

.B  D/
Joint probability
of occurrence

.A C C/
Lending
rate (percent)

Boom

12

0.25

Normal

10

0.50

Poor

0.25

3.0
5.0
8.0
3.0
5.0
8.0
3.0
5.0
8.0

0.40
0.30
0.30
0.40
0.30
0.30
0.40
0.30
0.30

0.100
0.075
0.075
0.200
0.150
0.150
0.100
0.075
0.075

15
17
20
13
15
18
11
13
16

66

C.-F. Lee et al.

Fig. 4.7 Probability of Xi in the


intervals 1; 2; 3

C .0:150/.18  15:1/2
C .0:100/.11  15:1/2 C .0:075/.13  15:1/2
i1=2
C .0:075/.16  15:1/2
D.0:001 C 0:271 C 1:801 C 0:882 C 0:0015
C 1:2615 C 1:681

away. The remaining risk is systematic risk that which influences all risky assets.
The market rate of return can be calculated using one of
several types of market indicator series, such as the DowJones Industrial Average, the Standard and Poor (S&P) 500,
or the New York Stock Exchange Index, using the following
equation:
It  It 1
D Rmt
It 1

C 0:331 C 0:061/
D2:51%
If this distribution is indeed approximately normal, the mean
and standard deviation can be employed to make some
statistical inferences. Figure 4.7 makes it clear that 68.3% of
the observations of a standard normal distribution are within
one standard deviation of the mean, 95.4% are within three.
Since the average lending rate is assumed to be normally
distributed with a mean of 15.1% and a standard deviation of
2.51%, it is clear that almost all (99.7%) of the lending rates
will lie in the range of 7.5722.63%, because 7.57% is three
standard deviations below 15.1 and 22.63% is three standard
deviations above the mean. It is also clear that 68.3% of the
rates will lie in the range of 12.5917.61%.

4.6 The Market Rate of Return and Market


Risk Premium
The market rate of return is the return that can be expected
from the market portfolio. This portfolio is of all risky assets that is, stocks, bonds, real estate, coins, and so on. Because all risky assets are included, the market portfolio is a
completely diversified portfolio. All unsystematic risks related to each individual asset would, therefore, be diversified

(4.9)

where:
Rmt D market rate of return at time tI
It D market index at tI and
It 1 D market index at t  1:
This equation calculates the percent change in the market
index during period t and the previous period t  1. This
change is the rate of return an investor would expected to
receive in t had he or she invested in t  1.
A risk-free investment is one in which the investor is sure
about the timing and amount of income streams arising from
that investment. However, for most types of investments, investors are uncertain about the timing and amount of income
of their investments. The types of risks involved in investments can be quite broad, from the relatively riskless T-bills
to highly risky speculative stocks.
The reasonable investor dislikes risks and uncertainty and
would, therefore, require an additional return on his investment to compensate for this uncertainty. This return, called
the risk premium, is added to the nominal risk-free rate. The
risk premium is derived from several major sources of uncertainty or risk, as was discussed at the beginning of this
chapter.
Table 4.4 illustrates this concept. In this table the market rate of return using the S&P 500 was calculated using

4 Foundation of Portfolio Theory

67

Table 4.4 Market returns and


T-bill by quarters
Year

Month

S&P 500

2005
2006

Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec

1; 248:29
1; 280:08
1; 280:66
1; 294:87
1; 310:61
1; 270:09
1; 270:20
1; 276:66
1; 303:82
1; 335:85
1; 377:94
1; 400:63
1; 418:30
1; 438:24
1; 406:82
1; 420:86
1; 482:37
1; 530:62
1; 503:35
1; 455:27
1; 473:99
1; 526:75
1; 549:38
1; 481:14
1; 468:36
1; 378:55
1; 330:63
1; 322:70
1; 385:59
1; 400:38
1; 280:00
1; 267:38
1; 282:83
1; 164:74
968:75
896:24
887:88

2007

2008

Equation (4.9) to devise average monthly returns. Monthly


T-bill rates are listed in the column B. The T-bill rate was
deducted from the market return rate .Rm  Rf / to devise
the risk premium. In the last 2 months of 2007 and for most
months of 2007 the market was in decline, with low returns
resulting in each month. This allowed the T-bill investors to
obtain a higher than the market return and resulted in negative risk premiums.
The second half of 2006 demonstrated an increasing
market level and higher market returns. In October 1982 the
return was 3.15%, the highest in the past 10 months. This al-

(A)
Market return
(percent)
2:55
0:05
1:11
1:22
3:09
0:01
0:51
2:13
2:46
3:15
1:65
1:26
1:41
2:18
1:00
4:33
3:25
1:78
3:20
1:29
3:58
1:48
4:40
0:86
6:12
3:48
0:60
4:75
1:07
8:60
0:99
1:22
9:21
16:83
7:48
0:93

(B)
T-bill rate
(percent)

(AB)
Risk
premium
(percent)

0.34
0.36
0.37
0.38
0.39
0.39
0.41
0.42
0.41
0.40
0.41
0.41
0.41
0.42
0.42
0.41
0.40
0.39
0.40
0.40
0.36
0.32
0.31
0.25
0.27
0.17
0.14
0.12
0.12
0.15
0.15
0.14
0.14
0.07
0.04
0.01

2:21
0:32
0:74
0:84
3:48
0:38
0:10
1:71
2:05
2:75
1:23
0:85
0:99
2:60
0:58
3:92
2:86
2:17
3:60
0:89
3:22
1:16
4:71
1:11
6:38
3:65
0:74
4:64
0:95
8:75
1:14
1:08
9:35
16:90
7:53
0:94

lowed market rates to leap beyond the riskless T-bill rate, and
the result was a positive risk premium. During the period of
fluctuation in the level of stock market prices, the first half of
2007 revealed a fluctuated risk premium.
Theoretically it is not possible for a risk premium required
by investors to be negative. Taking on risk involves some
positive cost. Nevertheless, using short-run estimators as in
Table 4.4 may result in negative figures because they reflect
the fluctuations of the market. The basic problem with using actual market data to assess risk premiums is the difference between expected returns (which are always positive)

68

and realized returns (which may be positive or negative). It


becomes evident that investors expectations will not always
be realized.

4.7 Conclusion
This chapter has defined the basic concepts of risk and risk
measurement. The efficient-portfolio concept and its implementation was demonstrated using the relationships of risk
and return. The dominance principle and performance measures were also discussed and illustrated. Finally, the interest
rate and market rate of return were used as measurements to
show how the commercial lending rate and the market risk
premium can be calculated.
Overall, this chapter has introduced uncertainty analysis
assuming previous exposure to certainty concepts. Further
application of the concepts discussed in this chapter as related to security analysis and portfolio management are explored in later chapters.

References
Ben-Horin, M. and H. Levy. 1980. Total risk, diversifiable risk and
non-diversifiable risk: a pedagogic note. Journal of Financial and
Quantitative Analysis 15, 289295.
Bodie, Z., A. Kane, and A. Marcus. 2006. Investments, 7th Edition,
McGraw-Hill, New York.
Bowman, R. G. 1979. The theoretical relationship between systematic
risk and financial (accounting) variables. Journal of Finance 34,
617630.
Elton, E. J., M. J. Gruber, S. J. Brown, and W. N. Goetzmann.
2006. Modern portfolio theory and investment analysis, 7th Edition,
Wiley, New York.
Evans, J. L. and S. H. Archer. 1968. Diversification and the reduction of dispersion: an empirical analysis. Journal of Finance 23,
761767.

C.-F. Lee et al.


Francis, J. C. and S. H. Archer. 1979. Portfolio analysis, Prentice-Hall,
Englewood Cliffs, NJ.
Ibbotson, R. G. and R. A. Sinquefield. 1976. Stocks, bonds, bills, and
inflation: simulations of the future (19762000). Journal of Business 49, 313338.
Jensen, M. C. 1968. The performance of mutual funds in the period
19451964. Journal of Finance 23, 389416.
Jensen, M. C. 1969. Risk, the pricing of capital assets, and the
evaluation of investment portfolios. Journal of Business 42,
167185.
Lee, C. F. and S. N. Chen. 1981. The sampling relationship between
Sharpes performance measure and its risk proxy: sample size,
investment horizon and market conditions. Management Science
27(6), 607618.
Lee, C. F. and S. N. Chen. 1984. On the measurement errors and ranking of composite performance measures. Quarterly Review of Economics and Business 24, 617.
Lee, C. F. and S. N. Chen. 1986. The effects of the sample size, the
investment horizon and market conditions on the validity of composite performance measures: a generalization, Management Science 32(11), 14101421.
Lee, C. F. and A. C. Lee. 2006. Encyclopedia of finance, Springer,
New York.
Lee, C. F., J. C. Lee, and A. C. Lee. 2000. Statistics for business and
financial economics, World Scientific Publishing Co, Singapore.
Maginn, J. L., D. L. Tuttle, J. E. Pinto, and D. W. McLeavey. 2007.
Managing investment portfolios: a dynamic process, CFA Institute
Investment Series, 3rd Edition, Wiley, New Jersey.
Markowitz, H. M. 1959. Portfolio selection: efficient diversification of
investments, Wiley, New York, NY.
Modigliani, F. and G. A. Pogue. 1974. An introduction to risk and return. Financial Analysis Journal 30, 6986.
Robicheck, A. A. and R. A. Cohn. 1974. The economic determinants
of systematic risk. Journal of Finance 29, 439447.
Schall, L. D. 1972. Asset valuation, firm investment, and firm diversification. Journal of Business 45, 1128.
Sharpe, W. F. 1966. Mutual fund performance. Journal of Business
39, 119138.
Thompson, D. J. 1976. Sources of systematic risk in common stocks.
Journal of Business 49, 173188.
Tobin, J. 1958. Liquidity preference as behavior toward risk. Review
of Economic Studies 25, 6586.
Treynor, J. 1965. How to rate management of investment funds.
Harvard Business Review 43, 6375.
Wackerly, D., W. Mendenhall, and R. L. Scheaffer. 2007. Mathematical
statistics with applications, 7th Edition, Duxbury Press, California.