chapter

7
Portfolio
Theory

W ith $1 million to invest, you can have a nice portfolio of securities, but what is a portfolio exactly? You have
heard people say, don’t put all of your eggs in one basket, so how does that apply to investing? If you decide to put
the entire $1 million in two stocks, what will the trustee say? Popular press articles routinely discuss the importance
of diversification, or note that some stocks react negatively to economic developments such as the threat of rising infla-
tion, while others respond positively. Therefore, it seems like a good idea to learn about the basics of portfolio theory.
Then you will not be intimidated when someone starts talking about Markowitz (modern) portfolio theory, which is a
universal concept that is widely known and discussed by investors.
In this chapter, we outline the nature of risk and return as it applies to making investment decisions. Unlike
Chapter 6, we are talking about the future, which involves expected returns, and not the past, which involves realized
returns. Investors must estimate and manage the returns and risk of their investments. By building diversified port-
folios, they reduce risk to the maximum extent possible without negatively affecting returns. Therefore, we should
consider the investor’s total portfolio and analyze investment risk accordingly. As we shall see, diversification is the key
to effective risk management. We will consider the critically important principle of Markowitz diversification, focusing
primarily on the concepts of correlation and covariance as applied to security returns.

AFTER READING THIS CHAPTER YOU WILL BE ABLE TO:

▶ Understand the meaning and calculation of expected ▶ Calculate portfolio return and risk measures as
return and risk measures for an individual security. formulated by Markowitz.
▶ Recognize what it means to talk about modern ▶ Understand how diversification works.
portfolio theory.

Dealing With Uncertainty

In Chapter 6, we discussed average returns, both arithmetic and geometric, that investors
have experienced from investing in the major financial assets available to them. We also con-
sidered the risk of these asset returns as measured by the standard deviation or variance.
Realized returns are important for several reasons. First, investors need to know how
their portfolios have performed relative to relevant market indices. Second, realized returns
are important in helping investors form expectations about future returns by providing a
foundation upon which to make estimates of expected returns. For example, if over a long
period Treasury bills have averaged less than 4 percent on a geometric mean basis, it would be

c07.indd 169 1/23/2016 10:07:44 PM

 170 CHAPTER 7 PORTFOLIO THEORY

unrealistic to expect long‐run average compound returns of 6 or 7 percent in the future from
Treasury bills unless the investing environment has changed significantly.
How do we go about estimating returns, which is what investors must actually do in
managing their portfolios? First, note that we will use the return and risk measures developed
in Chapter 6. The return measure is applicable whether one is measuring realized returns or
estimating future (expected) returns because it includes everything the investor can expect to
receive over any specified future period.
Similarly, the variance, or its square root, the standard deviation, is an accepted measure
of variability for both realized returns and expected returns. We will calculate both the variance
and the standard deviation in the following text and use them interchangeably as the situation
dictates. Sometimes it is preferable to use one, and sometimes the other.
To estimate the returns from various securities, investors must estimate the cash flows
these securities are likely to provide. The basis for doing so for bonds and stocks will be cov-
ered in their respective chapters. For now, it is sufficient to remind ourselves of the uncertainty
of the future, a problem emphasized at the outset of Chapter 1.
Exhibit 7-1 presents an interesting discussion of risk and how best to understand it. In
this essay, Peter Bernstein, one of the most prominent observers of the investing environment,
argues that risky decisions are all about three elements. His second point, “expect the unex-
pected,” turned out to be particularly relevant given what happened to the financial system in
2008. The unexpected did occur, and very few were prepared to deal with the situation. The
resulting damage was enormous.

USING PROBABILITIES
The return an investor will earn from investing is not known; it must be estimated. Future
return is an expected return and may or may not actually be realized. An investor may expect
the return on a particular security to be 10 percent for the coming year, but in truth, this is only
a “point estimate.” Risk, or the chance that some unfavorable event will occur, is present when
investment decisions are made. Investors are often overly optimistic about expected returns.
We can use the term random variable to describe the one‐period rate of return from a stock
(or bond)—it has an uncertain value which fluctuates randomly.
To deal with the uncertainty of returns, investors need to think explicitly about a
security’s distribution of probable returns. In other words, while investors may expect a secu-
rity to return 10 percent, this is only a one‐point estimate of the entire range of possibilities.
Given that investors must deal with the uncertain future, a number of possible returns can,
and will, occur.
In the case of a Treasury bond paying a fixed rate of interest, the interest payment will
be made with 100 percent certainty barring a financial collapse of the country. The probability
of occurrence is 1.0, because no other outcome is possible.
With the possibility of multiple outcomes, which is the norm for common stocks, each
possible likely outcome must be considered and a probability of its occurrence assessed. The
probability for a particular outcome is simply the chance that the specified outcome will occur
and is typically expressed as a decimal or fraction.

PROBABILITY DISTRIBUTIONS
A probability distribution for a security brings together the likely outcomes that may occur
along with the probabilities associated with these likely outcomes. The set of probabilities in
a probability distribution must sum to 1.0, or 100 percent because they must completely
describe all the (perceived) occurrences.

c07.indd 170 1/23/2016 10:07:44 PM

 Dealing With Uncertainty 171

Exhibit 7-1
Risk: The Whole versus the Parts
Many years ago, in the middle of a staff meeting, a colleague of information about risk that assails us. Now we can break
passed me a scrap of paper on which he had written, “When all down the problem of risk into what appear to me to be its
is said and done, more things are said than done.” When I three primary constituent parts.
consider the plethora of books, articles, consultants, and First, what is the balance between the consequences of being
conferences on risk in today’s world, my friend’s aphorism has wrong and the probability of being wrong? Many mistakes do
never seemed more appropriate. Are we never going to nail risk not matter. Other mistakes can be fatal. No matter how small the
down and bring it under control? How much more can anyone probability you will be hit by a car when you cross against the
reveal to us beyond what we have already been told? lights, the consequences of being hit deserve the greater weight
In a very real sense, this flood of material about risk is inherently in the decision. This line of questioning is the beginning, and in
risky. Sorting out the pieces and searching for main themes has some ways the end, of risk management. All decisions must pass
become an escalating challenge. The root of the matter gets lost through this sieve. It is the end if you decide not to take the risk,
in the shuffle while we are analyzing all the elegant advances in but it is also the end in the sense that distinguishing between
risk measurement and the impressive broadening of the kinds of consequences and probabilities is what risk management is all
risks we seek to manage. More is said than is done, or what is about.
done loses touch with what has been said. Second, expect the unexpected. That sounds like an empty
If we go back to first principles for a moment, perhaps we can cliché, but it has profound meaning for risk management. It is
put the multifarious individual pieces into some kind of a larger easy to prepare for the risks you know—earnings fail to meet
framework and optimize the choices among the masses of expectations, clients depart, bonds go sour, and a valued
information we are attempting to master. associate goes to a competitor. Insurance and hedging strategies
Professor Elroy Dimson of the London Business School once cover other kinds of risks lying in wait out there, from price
said risk means more things can happen than will happen. volatility to premature death.
Dimson’s formulation is only a fancy way of saying that we do not But preparation for the unexpected is a matter of the
know what is going to happen—good or bad. Even the range of decision‐making structure, and nothing else. Who is in charge
possible outcomes remains indeterminate, much as we would like here? That is the critical question in any organization. And if it is
to nail it down. Remember always: Risk is not about uncertainty just you there when the unexpected strikes, then you should
but about the unknown, the inescapable darkness of the future. prepare in advance for where you will turn for help when
If more things can happen than will happen, and if we are matters seem to be running out of control.
denied precise knowledge of the range of possible outcomes, Finally, note that word “control.” With an exit strategy—when
some decisions we make are going to be wrong. How many, how decisions are easily reversible—control over outcomes can be a
often, how seriously? We have no way of knowing even that. secondary matter. But with decisions such as launching a new
Even the most elegant model, as Leibniz reminded Jacob product or getting married, the costs of reversibility are so high
Bernoulli in 1703, is going to work “only for the most part.” that you should not enter into them unless you have some
What lurks in the smaller part is hidden from us, but it could control over the outcome if things turn out differently from
turn into a load of dynamite. what you expect. Gambling is fun because your bet is irreversible
The beginning of wisdom in life is in accepting the inevitability and you have no control over the outcome. But real life is not a
of being wrong on occasion. Or, to turn that phrase around, the gambling casino.
greatest risks we take are those where we are certain of the These three elements are what risky decisions are all about—
outcome—as masses of people are at classic market bottoms consequences versus probabilities, preparation for dealing with
and tops. My investment philosophy has always been that victory unexpected outcomes, and the distinction between reversibility
in the long‐run accrues to the humble rather than to the bold. and control. These are where things get done, not said.
This emphasis on ignorance is the necessary first step toward
SOURCE: Peter Bernstein, “Risk: The Whole Versus the Parts,” CFA Magazine,
the larger framework we need if we hope to sort out the flood March/April 2004, p. 5. Reprinted by permission..

c07.indd 171 1/23/2016 10:07:44 PM

 172 CHAPTER 7 PORTFOLIO THEORY

How are these probabilities and associated outcomes obtained? In the final analysis,
investing for some future period involves uncertainty, and therefore subjective estimates.
Although past occurrences (frequencies) may be relied on heavily to estimate the probabilities,
the past must be modified for any changes expected in the future.
Probability distributions can be either discrete or continuous. With a discrete probability
distribution, a probability is assigned to each possible outcome. In Figure 7-1a, five possible
returns are assumed for General Foods for next year. Each of these five possible outcomes—1,
7, 8, 10, and 15 percent—has an associated probability; these probabilities sum to 1.0, indi-
cating that the possible outcomes that an investor foresees for General Foods have been
accounted for.
With a continuous probability distribution, as shown in Figure 7-1b, an infinite number
of possible outcomes exist. Because probability is now measured as the area under the curve
in Figure 7-1b, the emphasis is on the probability that a particular outcome is within some
range of values.
The most familiar continuous distribution is the normal distribution depicted in
Figure 7-1b. This is the well‐known bell‐shaped curve often used in statistics. It is a
two‐parameter distribution in that the mean and the variance fully describe it.

CALCULATING EXPECTED RETURN FOR A SECURITY

To describe the single most likely outcome from a particular probability distribution, it is
necessary to calculate its expected value. The expected value of a probability distribution is the
weighted average of all possible outcomes, where each outcome is weighted by its respective
probability of occurrence. Since investors are interested in returns, we will call this expected

FIGURE 7-1 1
(a) A Discrete
Probability
Distribution
(b) A Continuous
Probability

Probability 0.5
Distribution

0
5 10 15
Rate of return (%)
(a)
Probability function

Rate of return (%)

(b)

c07.indd 172 1/23/2016 10:07:44 PM

 Dealing With Uncertainty 173

Expected Return The value the expected rate of return, or simply expected return, and for any security, it is
ex ante return expected
calculated as
by investors over some m
future holding period
E R Ri pri (7-1)
i 1

where
E R the expected rate of return on a security
Ri the ith po
ossible return
pri the probability of the ith return Ri
m the number of possible returns

CALCULATING RISK FOR A SECURITY

Investors must be able to quantify and measure risk. To calculate the total (stand‐alone) risk
associated with the expected return, the variance or standard deviation is used. As we know from
Chapter 6, the variance and its square root, standard deviation, are measures of the spread or
dispersion in the probability distribution; that is, they measure the dispersion of a random vari-
able around its mean. The larger this dispersion, the larger the variance or standard deviation.

✓ The tighter the probability distribution of expected returns, the smaller the standard
deviation, and the smaller the risk.

Based on your analysis, you think that next year’s return for General Foods will range from
Example 7-1 1 to 15 percent as described earlier. The expected value of the probability distribution for
General Foods returns is calculated in the first three columns of Table 7-1. We call this
expected value the expected rate of return, or simply expected return.

To calculate the variance or standard deviation from the probability distribution, first
calculate the expected return of the distribution using Equation 7-1. Essentially, the same
procedure used in Chapter 6 to measure risk applies here, but now the probabilities associated
with the outcomes must be included, as in Equation 7-2:
m
2 2
Variance of returns Ri E R pri (7-2)
i 1
and

2 1/ 2
Standard deviation of returns (7-3)

where all terms are as defined previously.

Note that the standard deviation is simply a weighted average of the deviations from the
expected value. As such, it provides some measure of how far the actual value may differ from
the expected value, either above or below. With a normal probability distribution, the actual
return on a security will be within ±1 standard deviation of the expected return approximately
68 percent of the time, and within ±2 standard deviations approximately 95 percent of
the time.

c07.indd 173 1/23/2016 10:07:45 PM

 174 CHAPTER 7 PORTFOLIO THEORY

The variance and standard deviation for General Foods, using the information reported above,
Example 7-2 is calculated in Table 7-1.

Calculating a standard deviation using probability distributions involves making subjec-

tive estimates of the probabilities and the likely returns. However, we cannot avoid such esti-
mates because future returns are uncertain. The prices of securities are based on investors’
expectations about the future. The relevant standard deviation in this situation is the ex ante
standard deviation and not the ex post standard deviation, which is based on realized returns.
Although standard deviations based on realized returns are often used as proxies for
ex ante standard deviations, investors should be careful to remember that the past cannot be
extrapolated into the future without modifications.

✓ Standard deviations calculated using historical data may be convenient, but they are
subject to errors when used as estimates of the future.

TABLE 7-1 Calculating the Standard Deviation Using Expected Data

(1) (2) (3) (4) (5) (6)

Possible Return (%) Probability (1) × (2) Ri − E(R) (Ri − E(R))2 (Ri − E(R))2pri

1 0.2 0.2 −7.0 49.0 9.8

7 0.2 1.4 −1.0 1.0 0.2
8 0.3 2.4 0.0 0.0 0.0
10 0.1 1.0 2.0 4.0 0.4
15 0.2 3.0 7.0 49.0 9.8

1.0 8.0 = E(R) 20.2

(20.2)1 / 2 4.49%

Checking Your Understanding

1. The expected return for a security is typically different from any of the possible outcomes
(returns) used to calculate it. How, then, can we say that it is the security’s expected
return?
2. Having calculated a security’s standard deviation using a probability distribution, how
confident can we be in this number?

Introduction to Modern Portfolio Theory

In the 1950s, Harry Markowitz, the father of Modern Portfolio Theory (MPT), developed the
basic portfolio principles that underlie MPT. His original contribution was published in 1952,
making portfolio theory over 60 years old. Over time, these principles have been widely
adopted by the financial community in a variety of ways, creating a broad legacy for MPT.1

1
See Frank J. Fabozzi, Francis Gupta, and Harry M. Markowitz, “The Legacy of Modern Portfolio Theory,” The Journal
of Investing (Fall 2002): 7–22.

c07.indd 174 1/23/2016 10:07:45 PM

 Portfolio Return and Risk 175

The primary impact of MPT is on portfolio management because it provides a frame-

work for the systematic selection of portfolios based on expected return and risk principles.
Most portfolio managers today are aware of, and to varying degrees apply, the basic principles
of MPT. Major mutual fund families employ the implications of MPT in managing their funds,
financial advisors use the principles of MPT in advising their clients, many financial commen-
tators use MPT terms in discussing the current investing environment, and so forth.
Before Markowitz, investors dealt loosely with the concepts of return and risk. Investors
have known intuitively for many years that it is smart to diversify, that is, not to “put all of your
eggs in one basket.” Markowitz, however, was the first to develop the concept of portfolio
diversification in a formal way—he quantified the concept of diversification. He showed
quantitatively why, and how, portfolio diversification works to reduce the risk of a portfolio to
an investor.
Markowitz sought to organize the existing thoughts and practices into a more formal
framework and to answer a basic question: Is the risk of a portfolio equal to the sum of the
risks of the individual securities comprising it? The answer is no! Markowitz was the first
to show that we must account for the interrelationships among security returns in order to
calculate portfolio risk and in order to reduce portfolio risk to its minimum level for any given
level of return.

Investments Intuition

Clearly, investors thought about diversifying a port- of portfolio theory are widely used today, by them-
folio before Markowitz’s landmark work. But they selves or in conjunction with other techniques,
did so in general terms. And it is true that not and by both institutional investors and individual
everyone uses his analysis today. However, the tenets investors.

Portfolio Return and Risk

When we analyze investment returns and risks, we must consider the total portfolio held by
an investor. Individual security returns and risks are important, but it is the return and risk to
the investor’s portfolio that ultimately matters. Optimal portfolios can be constructed if port-
folios are diversified correctly. As we learned in Chapter 1, an investor’s portfolio is his or her
combination of assets.
As we will see, portfolio risk is a unique characteristic and not simply the sum of indi-
vidual security risks. A security may have high risk if held by itself, but much less risk when
held in a portfolio of securities. Since investors are concerned primarily with the risk to their
total wealth, as represented by their portfolio, individual stocks are risky only to the extent
that they add risk to the portfolio.

PORTFOLIO EXPECTED RETURN

Portfolio Weights The percentages of a portfolio’s total value that are invested in each
Portfolio Weights portfolio asset are referred to as portfolio weights, which we will denote by w. The combined
Percentages of portfolio
portfolio weights sum to 100 percent of total investable funds, or 1.0, indicating that all port-
funds invested in each
security folio funds are invested. That is,

n
w1 w2 " wn wi 1.0 (7-4)
i 1

c07.indd 175 1/23/2016 10:07:45 PM

 176 CHAPTER 7 PORTFOLIO THEORY

With equal dollar amounts invested in three securities, the portfolio weights are 0.333, 0.333,
Example 7-3 and 0.333. For an equal‐weighted portfolio of five securities, each security would have a port-
folio weight of 0.20. Of course, weights do not have to be equal. A five‐stock portfolio could
have weights of 0.40, 0.10, 0.15, 0.25, and 0.10, or 0.18, 0.33, 0.11, 0.22, and 0.16.

Calculating the Expected Return on a Portfolio The expected return on

any portfolio p can be calculated as a weighted average of the individual security expected
returns:
n
E(R p ) wi E(Ri ) (7-5)
i 1
where
E(R p ) the expected return on the portfolio
wi the portfolio weight for the ith security
wi 1 .0
E(Ri ) the expected return on the ith security
n the number of different securities in the portfolio

Consider a three‐stock portfolio consisting of stocks G, H, and I with expected returns of 12,
Example 7-4 20, and 17 percent, respectively. Assume that 50 percent of investable funds are invested in
security G, 30 percent in H, and 20 percent in I. The expected return on the portfolio is

E(R p ) 0.5(12%) 0.3(20%) 0.2(17%) 15.4%



  
 

  
   


  


we 
  


  


 
 
  


  
 



  

✓ Regardless of the number of assets held in a portfolio, or the proportion of total

investable funds placed in each asset, the expected return on the portfolio is always
a weighted average of the expected returns for individual assets in the portfolio.

c07.indd 176 1/23/2016 10:07:46 PM

 Analyzing Portfolio Risk 177

Investments Intuition

The expected return for a portfolio must fall bet- where it falls is determined by the percentages of
ween the highest and lowest expected returns for the investable funds placed in each of the individual secu-
individual securities making up the portfolio. Exactly rities in the portfolio.

PORTFOLIO RISK
Return and risk are the basis of all investing decisions. Risk is measured by the variance (or
standard deviation) of the portfolio’s return, exactly as in the case of an individual security.
Typically, portfolio risk is stated in terms of standard deviation.
A major implication of MPT is as follows: Although the expected return of a portfolio is
a weighted average of the expected returns of the individual securities included in the portfo-
lio, portfolio risk (as measured by the variance or standard deviation) is not a weighted average
of the risk of the individual securities in the portfolio. Symbolically,

n
E(R p ) wi E(Ri ) (7-6)
i 1
n
2 2
p wi i (7-7)
i 1

Equation 7-7 is an inequality, and, in fact, investors can reduce the risk of a portfolio
beyond what it would be if risk were simply a weighted average of the individual securities’
risk. In order to see how this risk reduction is accomplished, we analyze portfolio risk in detail.

✓ Portfolio risk is always less than a weighted average of the risks of the securities in the
portfolio unless the securities have outcomes that vary together exactly, which is an
almost impossible occurrence.

Analyzing Portfolio Risk

RISK REDUCTION—THE INSURANCE PRINCIPLE
To begin our analysis of how forming a portfolio of assets can reduce risk, assume that all
security risk sources in a portfolio are independent. As we add securities to the portfolio, the
exposure to any particular source of risk becomes small. According to the Law of Large
Numbers, the larger the sample size, the more likely it is that the sample mean will be close to
the population expected value. Risk reduction in the case of independent risk sources can be
thought of as the insurance principle, named for the idea that an insurance company reduces
its risk by writing many policies against many independent sources of risk.
We start by assuming the most extreme case in which rates of return on individual secu-
rities are statistically independent such that any one security’s rate of return is unaffected by

c07.indd 177 1/23/2016 10:07:46 PM

 178 CHAPTER 7 PORTFOLIO THEORY

another’s rate of return. In this situation, and only in this situation, the standard deviation of
the portfolio is given by

i (7-8)
p
n1/ 2
Equation 7-8 describes the extreme case of independent risk sources, which implies that
all risk is firm specific. In this situation, the total risk continues to fall toward 0 as securities
are added. Unfortunately, when it comes to investing in financial assets, the assumption of
statistically independent returns is unrealistic.

The risk of a portfolio declines quickly as more securities are added. Using Equation 7-8 and
Example 7-5 assuming that each security’s standard deviation is 20 percent, the risk of a 100‐security port-
folio is reduced to 2.0 percent:
20
p
1001/ 2
2 .0 %

In practice, most stocks are positively correlated with each other; that is, the movements
in their returns are related, not independent. We call the variation in returns that is attributable
to general market moves, market risk. While total risk can be reduced, it cannot be eliminated
because market risk cannot be eliminated. Unlike firm‐specific risk, common sources of risk
(market risk) affect all firms and cannot be diversified away. For example, an increase in
interest rates affects most firms adversely because most firms borrow funds to finance part of
their operations.

DIVERSIFICATION
Another major implication of MPT is that there are two general sources of risk, firm‐specific
and market risk. Because the sources of risk are not entirely independent, adding securities
reduces the firm‐specific risk, but not the market risk. The process of adding securities to a
portfolio to reduce firm‐specific risk is referred to as diversification.

✓ Diversification is the key to the management of portfolio risk because it allows investors
to significantly lower portfolio risk without adversely affecting return.

We consider two forms of portfolio diversification, beginning with random diversification

and moving to efficient portfolio diversification, which is based on MPT principles.

Random Diversification Random or naive diversification refers to the act of randomly

diversifying without regard to how security returns are related to each other. An investor sim-
ply selects a relatively large number of securities randomly—the proverbial “throwing a dart
at The Wall Street Journal page of stock quotes.” As we add securities to a portfolio, the total
risk associated with the portfolio of stocks declines rapidly even without the use of any invest-
ment insight. The first few stocks cause a large decrease in portfolio risk.
The benefits of diversification kick in immediately—two stocks are better than one, three
stocks are better than two, and so on. However, diversification cannot completely eliminate the
risk in a portfolio because the market risk cannot be eliminated. As additional stocks are added,
risk is reduced, but the marginal risk reduction becomes smaller with each security added.

c07.indd 178 1/23/2016 10:07:46 PM

 The Components of Portfolio Risk 179

Although random diversification is clearly beneficial, it is generally not optimal. To take

full advantage of the benefits of diversification, we need to understand efficient diversification;
that is, we need to understand portfolio risk within an MPT context.

Checking Your Understanding

3. What does it mean to an investor that the benefits of diversification kick in immediately
but are limited?

The Components of Portfolio Risk

In order to develop an equation for portfolio risk, as measured by variance or standard devia-
tion, we must account for two factors:

1. The weighted individual security risks (i.e., the variance of each individual security,
weighted by the percentage of funds invested in the security)
2. The weighted comovements between securities’ returns (i.e., the weighted covariance
between each security’s returns and the returns of all other securities in the portfolio)

The covariance and the correlation coefficient are prominent measures of comovement
between security returns and are both used commonly in deriving portfolio risk. These two
measures are discussed in detail in the following text.

COVARIANCE
Covariance An absolute Covariance is defined as the extent to which two random variables covary (move together)
measure of the extent over time. In financial markets, the variables in question are typically the returns on two secu-
to which two variables rities. Covariance can be:
tend to covary, or move
together
1. Positive, indicating that the returns on the two securities tend to move in the same
direction at the same time; when one increases (decreases), the other tends to do the
same.2
2. Negative, indicating that the returns on the two securities tend to move inversely; when
one increases (decreases), the other tends to decrease (increase).
3. Zero, indicating that the returns on the two securities are independent and have no
tendency to move together.

The formula for calculating the expected covariance for two securities is
m

AB RA,i E RA RB,i E RB pri (7-9)

i 1
where
AB the covariance between securities A and B3
RA,i one possible return on security A
E RA the expected return on security A
m the number of likely outcomes for a security for the period

2
Another way to say this is that higher‐than‐average values of one random variable tend to be paired with
higher‐than‐average values of the other random variable.
3
The order does not matter, because AB BA.

c07.indd 179 1/23/2016 10:07:46 PM

 180 CHAPTER 7 PORTFOLIO THEORY

Equation 7-9 indicates that covariance is the sum of the weighted products of the return
deviations from their expected values. In Equation 7-9, if the returns for both A and B are
above their expected value or below their expected value at the same time, the product will be
positive, leading to a positive covariance. If, on the other hand, A is above its expected value
when B is below its expected value, the product will be negative, and with enough counter
occurrences, the covariance will be negative.
The size of the covariance measure depends on the units of the variables involved and
changes when these units are changed. Therefore, the covariance primarily provides informa-
tion to investors about whether the association between asset returns is positive, negative, or
zero because simply observing the number itself, without any context with which to assess the
number, is not very useful.
We offer the following as an example of the limitation of covariance as a measure of
comovement. Assume you are told two securities, R and S, have a covariance of 325. What
does this indicate about the comovement between security R and security S? Since the value
is positive, you know the two have a positive association, that is, when one performs well
(poorly), the other tends to also perform well (poorly). But how strong is the association? This
is where covariance offers little guidance. We really do not know whether a covariance of 325
indicates a weak or a strong positive association. Fortunately, the correlation coefficient, which
we discuss next, overcomes this limitation of covariance.

THE CORRELATION COEFFICIENT

Correlation Coefficient In investments, the correlation coefficient (correlation) is a statistical measure of the relative
A statistical measure of comovements between security returns. Like covariance, correlation (ρij) measures the extent
the extent to which two to which the returns on two securities move together. Furthermore, both measures denote the
variables are associated
association between the two securities; however, they do not offer guidance regarding causa-
tion. For example, two oil companies may have security returns that have a high covariance,
that is, they are highly correlated. This does not mean the movements in one cause the move-
ments in the other; instead, both firms are reliant on a common underlying factor, the price
of oil.
Covariance and correlation are related in the following manner:
AB
AB (7-10)
A B

This equation shows that the correlation is simply a standardized covariance measure.
To derive correlation, covariance is standardized by dividing the value by the product of the
two standard deviations.
Or alternatively, by rearranging the formula, the covariance can be written as

AB AB A B (7-11)

Therefore, by knowing the covariance, we can calculate the correlation (and vice versa)
because the standard deviations of the assets’ returns are already known. From the formulas, we
can establish that when the covariance is positive, the correlation will also be positive. Likewise,
when the covariance is negative, the correlation will also be negative. This relationship is evi-
dent because standard deviation (as the square root of variance) is always a positive value.
While correlation and covariance are both measures of comovement, the primary
difference between the two is that correlation is a bounded measure of comovement.
In particular, correlation is a relative measure of association that is bounded by +1.0 and
−1.0 with

c07.indd 180 1/23/2016 10:07:46 PM

 The Components of Portfolio Risk 181

ij 1
perfect positive correlation
ij 1
perfect negative (inverse) correlation
ij 0.0
zero correlation

When analyzing how security returns move together, it is more convenient to talk about
the correlation because we can immediately assess the degree of association. Remember that
hypothetical covariance of 325 that we assumed between security R and security S. Depending
upon the security standard deviations, that covariance may translate to a correlation of 0.9,
which indicates a very strong positive correlation (the highest is 1.0). On the other hand, it
may translate to a correlation of 0.1, which indicates a weak positive correlation.
Overall, we can conclude that covariance and correlation are similar measures of
comovement; both provide information about the association of two securities. Correlation,
however, extends the information that covariance offers by indicating both direction of asso-
ciation and strength of association. For this reason, correlation is much more popular as a
measure of comovement.

Perfect Positive Correlation With perfect positive correlation, the returns have a
perfect direct linear relationship. Knowing what the return on one security will do allows an
investor to forecast perfectly what the other will do. In Figure 7-2, stocks A and B have identi-
cal return patterns over the six‐year period 2009–2014. When stock A’s return goes up, stock
B’s does also. When stock A’s return goes down, stock B’s does also.
Notice that a portfolio combining stocks A and B (portfolio AB), with 50 percent
invested in each, has exactly the same return as does either stock by itself, since the returns
are identical. The risk of the portfolio, as measured by the standard deviation, is identical to
the standard deviation of either stock by itself.

✓ When returns are perfectly positively correlated, portfolio risk is simply a weighted
average of the individual securities’ risks. This is the one case where diversification does
not lead to a reduction in risk.

Perfect Negative Correlation In contrast, with perfect negative correlation, the

securities’ returns have a perfect inverse linear relationship to each other. Therefore, knowing
the return on one security provides full knowledge about the return on the second security.
When one security’s return is high, the other is low.
In Figure 7-3, stocks A and C are perfectly negatively correlated with each other. Notice
that each stock has exactly the same return and standard deviation. When combined, how-
ever, the deviations in the stocks’ returns around their average return of 12 percent cancel out,
resulting in a portfolio return of 12 percent. This portfolio has no risk. It earns 12 percent each
year over the period measured, and the average return is 12 percent.
Notice carefully what perfect negative correlation does for an investor. By offsetting all
the variations around the expected return for the portfolio, the investor is assured of earning
the expected return. At first glance, it might appear that offsetting a negative return with an
exactly equal positive return produces a zero return but that is not the case. The expected
return for the portfolio is a positive number (otherwise, we would not invest). What is being
offset in this case are variations around the expected return.

c07.indd 181 1/23/2016 10:07:46 PM

 182 CHAPTER 7 PORTFOLIO THEORY

Stock A Stock B
40 40
30 30

Rate of return (%)

Rate of return (%)

20 20
10 10
0 0
–10 –10
–20 –20
–30 –30
–40 –40
2009 2010 2011 2012 2013 2014 2009 2010 2011 2012 2013 2014

Portfolio AB
40
30 Year Stock A Stock B Return on portfolio AB
Rate of return (%)

20 2009 36 36 36
2010 –12 –12 –12
10
2011 –10 –10 –10
0 2012 34 34 34
2013 –6 –6 –6
–10
2014 30 30 30
–20 Av. return 12 12 12
–30 Std. dev. 21.5 21.5 21.5

–40
2009 2010 2011 2012 2013 2014

Figure 7-2 Returns for the Years 2009–2014 on Two Stocks, A and B, and a Portfolio Consisting of 50 Percent A and
50 Percent B, When the Correction Coefficient Is +1.0

Stock A Stock C
40 40
30 30
20 20
Rate of return (%)
Rate of return (%)

10 10
0 0
–10 –10
–20 –20
–30 –30
–40 –40
2009 2010 2011 2012 2013 2014 2009 2010 2011 2012 2013 2014

Portfolio AC
40
30 Year Stock A Stock C Return on portfolio AC
2009 36 –12 12
20
2010 –12 36 12
Rate of return (%)

10 2011 –10 34 12
0 2012 34 –10 12
–10 2013 –6 30 12
2014 30 –6 12
–20
Av. return 12 12 12
–30
Std. dev. 21.5 21.5 0.0
–40
2009 2010 2011 2012 2013 2014

FIGURE 7-3 Returns for the Years 2009–2014 on Two Stocks, A and C, and a Portfolio Consisting of 50 Percent A and
50 Percent C, When the Correction Coefficient Is −1.0

c07.indd 182 1/23/2016 10:07:47 PM

 The Components of Portfolio Risk 183

Zero Correlation With zero correlation, there is no linear relationship between the
returns on the two securities. Combining two securities with zero correlation reduces the risk
of the portfolio. However, portfolio risk is not eliminated in the case of zero correlation. While
a zero correlation between two security returns is better than a positive correlation, it does not
produce the risk reduction benefits of a negative correlation coefficient.

Less Than Perfect Positive Correlation Figure 7-4 illustrates a case where
stocks A and D are positively correlated with each other at a level of ρ = +0.55. Investors may
encounter situations such as this and feel there is not much benefit to be gained from diversi-
fying. Note that the standard deviation of each security is 21.5 percent, with an average return
of 12 percent, but when combined with equal weights of 0.50 into the portfolio, the risk is
reduced, to a level of 18 percent. Any reduction in risk that does not adversely affect return is
considered beneficial.
With positive correlation, risk can be reduced, but it cannot be eliminated. Other things
being equal, investors wish to purchase securities with the least positive correlation possible.

✓ Ideally, investors would like securities with negative correlation or low positive correla-
tion, but they generally are faced with positively correlated security returns.

Over the decade ending in 2014, the average correlation between the stocks in the
S&P 500 and the Index itself was about 0.55, so our previous example reflects the actual situ-
ation. Of course, the correlation fluctuates. For example, in 2011 it rose as high as 0.86 before
dropping back.

Stock A Stock D
40 40
30 30
Rate of return (%)

Rate of return (%)

20 20
10 10
0 0
–10 –10
–20 –20
–30 –30
–40 –40
2009 2010 2011 2012 2013 2014 2009 2010 2011 2012 2013 2014

Portfolio AD
40
30 Year Stock A Stock D Return on portfolio AD
2009 36 25 30.5
Rate of return (%)

20
2010 –12 13 0.5
10 2011 –10 19 4.5
0 2012 34 28 31.0
–10 2013 –6 –35 –20.5
–20 2014 30 22 26.0
Av. return 12 12 12.0
–30
Std. dev. 21.5 21.5 18.0
–40
2009 2010 2011 2012 2013 2014

FIGURE 7-4 Returns for the Years 2009–2014 on Two Stocks, A and D, and a Portfolio Consisting of 50 Percent A and
50 Percent D, When the Correction Coefficient Is +0.55

c07.indd 183 1/23/2016 10:07:47 PM

 184 CHAPTER 7 PORTFOLIO THEORY

Checking Your Understanding

4. Why is negative correlation between two securities in a portfolio better than no (zero)
correlation?

Calculating Portfolio Risk

Correlations and covariances quantitatively account for comovements in security returns and
can be used to calculate portfolio risk. We first consider the simplest possible case, two securi-
ties, in order to see what happens in the portfolio risk equation. We then consider the case of
many securities, where the calculations become too large and complex to analyze with any
means other than a computer.

THE TWO‐SECURITY CASE

The risk of a portfolio, as measured by the standard deviation of returns, for the case of two
securities, 1 and 2,
1/ 2
P w12 2
1 w22 2
2 2 w1 w2 1,2 1 2 (7-12)

Equation 7-12 shows that the risk for a portfolio encompasses not only the individual security
risks but also the covariance between the two securities. Furthermore, three characteristics
determine portfolio risk:

◨ The variance of each security, as shown by 12 and 22 in Equation 7-12

◨ The covariance between securities, as shown by 1,2 1 2 in Equation 7-12
◨ The portfolio weights for each security, as shown by the wi’s in Equation 7-12

Note the following about Equation 7-12:

◨ The covariance term contains two covariances—the covariance between stock 1 and
stock 2 and between stock 2 and stock 1. Since each covariance is identical, we simply
multiply the first covariance by two. Otherwise, there would be four terms in
Equation 7-12, rather than three.
◨ We first solve for the variance of the portfolio and then take the square root to obtain the
standard deviation of the portfolio.

Consider the returns between Southeast Utilities and Precision Instruments for the period
Example 7-6 2005–2014. The summary statistics for these two stocks are as follows:

Southeast Precision

Return (%) 10.1 15.4

Standard deviation (%) 16.8 27.5
Correlation coefficient 0.29

c07.indd 184 1/23/2016 10:07:48 PM

 Calculating Portfolio Risk 185

Assume, for expositional purposes, we place equal amounts in each stock; therefore, the
weights are 0.5 and 0.5:
1/ 2
p w12 2
1 w22 2
2 2 w1 w2 1,2 1 2

2 2 2 2 1/ 2
0.5 16.8 0 .5 27.5 2 0.5 0.5 0.29 16.8 27.5
1/ 2
70.56 189.06 66.99
18.1%
Alternatively,
1/ 2
p w12 2
1 w22 2
2 w1 w2 1,2 1 2 w2 w1 2,1 2 1

[ .5 2 16.8
2
0.5
2
27.5
2
0.5 0.5 0.29 16.8 27.5
0.5 0.5 0.29 27.5 16.8 ] 1/ 2

1/ 2
70.56 189.06 33.5 33.5
18.1%

The Impact of the Correlation Coefficient The standard deviation of the port-
folio is directly affected by the correlation between the two stocks. Portfolio risk is reduced as
the correlation coefficient moves downward from +1.0.

Extending Example 7-6, the correlation between Southeast Utilities and Precision Instruments
Example 7-7 returns is +0.29. In order to focus on the effects of a changing correlation coefficient, we con-
tinue to assume weights of 0.5 each—50 percent is placed in each security. Summarizing the
data in this example,

SU 16.8
PI 27.5
wSU 0 .5
wPI 0 .5
With these data, the standard deviation (risk) for this portfolio,
2 2 2 2 1/ 2
P 0 .5 16.8 0 .5 27.5 2 0.5 0.5 16.8 27.5
1/ 2
70.56 189.06 229.32

since 2(0.5)(0.5)(16.8)(27.5) = 229.32.

The risk of this portfolio clearly depends heavily on the value of the third term, which
in turn depends on the correlation coefficient between the returns for SU and PI. To assess the
potential impact of the correlation, consider the following cases: a of +1, +0.5, +0.29, 0,

c07.indd 185 1/23/2016 10:07:48 PM

 186 CHAPTER 7 PORTFOLIO THEORY

−0.5, and −1.0. Calculating portfolio risk under each of these scenarios produces the follow-
ing portfolio risks:
If 1 .0 : p 22.2%
If 0 .5 : p 19.4%
If 0.29 : p 18.1%
If 0.0 : p 16.1%
If 0 .5 : p 12.0%
If 1 .0 : p 5 .4 %

These calculations clearly show the impact that combining securities with less than
perfect positive correlation has on portfolio risk. The risk of the portfolio steadily decreases
from 22.2 percent to 5.4 percent as the correlation coefficient declines from +1.0 to −1.0.
Note, however, that risk declines from 22.2 percent to only 16.1 percent as the correlation
coefficient drops from +1 to 0 and it is only cut in half (approximately) when is −0.5.

Investments Intuition

Correlations are a key variable when considering As Equation 7-12 shows, the benefits also depend on
how diversification reduces risk. However, a little the standard deviations of the asset returns and the
reflection indicates they are not the complete story. portfolio weights.

The Impact of Portfolio Weights We saw earlier (Figure 7-3) that with a two‐
stock portfolio and perfect negative correlation, the risk can be reduced to zero. Notice that
this did not happen in Example 7-7 (the risk when = −1.0 is 5.4 percent). The reason for
this is that the weights for each stock were selected to be 0.50 for illustration purposes.
To reduce the risk to zero in the two‐security case, and to minimize risk in general, it is neces-
sary to select optimal weights.
Let’s consider the impact of portfolio weights on portfolio risk. The size of the weights assigned
to each security has an effect on portfolio risk, holding the correlation coefficient constant.

Using the same data as Example 7-7, let’s consider portfolio risk. Recall that the correlation
Example 7-8 coefficient between Southeast Utilities and Precision Instruments is +0.29. For illustration
purposes, we examine five different sets of weights, each of which must sum to 1.0.

Southeast Precision σp (%)

0.1 0.9 25.3

0.3 0.7 21.3
0.5 0.5 18.1
0.7 0.3 16.2
0.9 0.1 16.1

c07.indd 186 1/23/2016 10:07:48 PM

 Checking Your Understanding 187

In this two‐stock portfolio example, holding the correlation coefficient constant at +0.29, the
risk of the portfolio varies as the weight for each of the assets changes. Because Southeast has
a substantially lower standard deviation than does Precision, portfolio risk decreases as the
weight assigned to Southeast increases. However, with a positive correlation, portfolio risk
reduction is somewhat limited.

✓ Portfolio risk is affected by both the correlation between assets and by the percentage of
funds invested in each asset.

THE n‐SECURITY CASE

The two‐security case can be generalized to the n‐security case. Portfolio risk can be reduced
by combining assets with less than perfect positive correlation. Furthermore, the smaller the
positive correlation, the better.
Portfolio risk is a function of each individual security’s risk and the covariances between
the returns on the individual securities. Stated in terms of variance, portfolio risk is
n n n
2
p wi2 2
i wi w j ij (7-13)
i 1 i 1 j 1
i j
2
p the variance of the portfolio s return
2
i the variance of security i s return
ij the covariance between the returns for securities i and j
wi the weight invested in security i
n n
a double summation sign, since i j, n (n 1) comovement terms are added
i 1 j 1 together ( i.e., all possible pairs of values for i and j, excluding i j)
Although Equation 7-13 appears formidable, it states exactly the same message as
Equation 7-12 for the two‐stock portfolio:
Portfolio risk is a function of:

◨ The weighted risk of each individual security (as measured by its variance)
◨ The weighted covariances among all pairs of securities
✓ As noted previously, three variables determine portfolio risk: variances, covariances, and weights.
Since the covariance of security i and security j (σij) equals the correlation of returns for
security i and j times the standard deviation of each security (Equation 7-11), we can easily
report Equation 7-13 using covariance or correlation. Therefore, in general discussions of
portfolio risk, these two comovement terms can be, and are, used interchangeably.
Because of its importance, we emphasize the components of portfolio risk in Figure 7-5.

Checking Your Understanding

5. Correlation and covariance are both measures of comovement. How are the two meas-
ures used and how are they related to one another?
6. Suppose we add a very risky stock to a well‐diversified portfolio. Could such an action
lower the portfolio’s risk?

c07.indd 187 1/23/2016 10:07:48 PM

 188 CHAPTER 7 PORTFOLIO THEORY

Figure 7-5 The sum of

The sum of
The Components of the
the
weighted
Portfolio Risk weighted Portfolio
covariances
risk of each risk
among all
individual
pairs of
security
securities

The Importance of Covariance One of Markowitz’s important contributions to

portfolio theory is his insight about the relative importance of variances and covariances.
When we add a new security to a large portfolio of securities, there are two impacts.

1. The asset’s own risk, as measured by its weighted variance, is added to the portfolio’s
risk.
2. Weighted covariances between the new security and every other security already in the
portfolio are also added.

✓ As the number of securities held in a portfolio increases, the importance of each

individual security’s risk (variance) decreases, while the importance of the covariance
relationships increases.

For example, in a portfolio of 150 securities, the contribution of each security’s own risk
to the total portfolio risk is extremely small. When a new security is added to a large portfolio
of securities, what matters most is its covariance with the other securities in the portfolio.
Portfolio risk consists almost entirely of the covariance risk between securities. Individual
security risks are diversified away in a large portfolio, but the covariance terms are not diversi-
fied away and therefore contribute to the risk of the portfolio.

Obtaining the Data

To calculate portfolio risk using Equation 7-13, we need estimates of the variance for each
security and estimates of the correlation coefficients or covariances. Both variances and covari-
ances can be (and are) calculated using either ex post or ex ante data. If an analyst uses ex post
data to calculate the variances and covariances and then uses these estimates in the Markowitz
model, the implicit assumption is that the relationship that existed in the past will continue
into the future. If historical data is thought to offer the best estimate of the expected variances
and covariances, it should be used. However, it must be remembered that measures of vari-
ance and covariance change over time as does the expected return.

SIMPLIFYING THE MARKOWITZ CALCULATIONS

Equation 7-13 illustrates the problem associated with the calculation of portfolio risk using
the Markowitz mean‐variance analysis. In the case of two securities, there are two covariances,
and we multiply the weighted covariance term in Equation 7-12 by two since the covariance
of A with B is the same as the covariance of B with A. Therefore, in the case of three securities,
there are six covariances, with four securities, 12 covariances, and so forth. The total number
of covariances in the Markowitz model is calculated as n(n − 1), where n is the number of
securities.
Table 7-2 shows the variance–covariance matrix associated with these calculations. For
the case of two securities, there are n2, or four, total terms in the matrix—two variances and

c07.indd 188 1/23/2016 10:07:49 PM

 Summary 189

Table 7-2 The Variance–Covariance Matrix Involved in

Calculating the Standard Deviation of a Portfolio

Two securities:
σ1,1 σ1,2
σ2,1 σ2,2
Four securities:
σ1,1 σ1,2 σ1,3 σ1,4
σ2,1 σ2,2 σ2,3 σ2,4
σ3,1 σ3,2 σ3,3 σ3,4
σ4,1 σ4,2 σ4,3 σ4,4

two covariances. For the case of four securities, there are n2, or 16 total terms in the matrix—
4 variances and 12 covariances. The variance terms are on the diagonal of the matrix and, in
effect, represent the covariance of a security with itself. Note that the covariance terms above
the diagonal are a mirror image of the covariance terms below the diagonal—that is, each
covariance is repeated twice since σAB is the same as σBA.

✓ The number of covariances in the Markowitz model is based on the calculation n(n − 1),
where n is the number of securities involved. Because the covariance of A with B is the
same as the covariance of B with A, there are [n(n − 1)]/2 unique covariances.

An analyst considering 100 securities must estimate [100(99)]/2 = 4,950 unique covariances.
Example 7-9 For 250 securities, the number is [250(249)]/2 = 31,125 unique covariances.

Obviously, estimating large numbers of covariances quickly becomes a major problem

for model users. Since many institutional investors follow as many as 250 or 300 securities,
the number of inputs required may become an impossibility. In fact, until the basic Markowitz
model was simplified in terms of the covariance inputs, it remained primarily of academic
interest.
On a practical basis, analysts are unlikely to be able to directly estimate the large number
of correlations necessary for a complete Markowitz analysis. In his original work, Markowitz
suggested using the relation of each security to an index as a means of generating covariances.

Summary
▶ The expected return from a security must be esti- therefore, it also incorporates the probabilities used in
mated. Since this is done under conditions of uncer- calculating the expected return.
tainty, it may not be realized. Risk (or uncertainty) is ▶ The expected return for a portfolio is a weighted aver-
always present in the estimation of expected returns age of the individual security expected returns.
for risky assets.
▶ Portfolio weights, designated wi, are the percentages of
▶ Probability distributions are used to calculate a secu- a portfolio’s total funds that are invested in each secu-
rity’s expected return. rity, where the weights sum to 1.0.
▶ The standard deviation or variance of a security’s ▶ Portfolio risk is not a weighted average of the indi-
expected return is a measure of the security’s risk; vidual security risks. To calculate portfolio risk, we

c07.indd 189 1/23/2016 10:07:49 PM

 190 CHAPTER 7 PORTFOLIO THEORY

must take account of the comovements between the ▶ The risk of a well‐diversified portfolio is largely attrib-
securities’ returns. utable to the impact of the covariances. When a new
security is added to a large portfolio of securities, what
▶ The correlation coefficient is a relative measure of the matters most is its covariance with the other securities
association between security returns. It is bounded by in the portfolio.
+1.0 and −1.0, with 0 representing no association.
▶ As the number of securities held in a portfolio
▶ Correlation and covariance are both quantitative increases, the importance of each individual security’s
measures of comovement, and either can be used in risk (variance) decreases, while the importance of the
deriving portfolio risk (portfolio variance). covariance relationships increases.
▶ The major problem with the Markowitz model is that
▶ Portfolio risk is a function of security variances, covar-
it requires a full set of security return covariances in
iances, and portfolio weights.
order to calculate portfolio variance.
▶ Covariance and correlation between security returns ▶ The number of covariances in the Markowitz model
determine how much portfolio risk can be reduced is n(n − 1); the number of unique covariances is
through diversification. [n(n − 1)]/2.

Questions
7‐1 Distinguish between historical return and expected 7‐12 How many total terms (variances and covariances)
return. would exist in the variance–covariance matrix for
7‐2 How is expected return for one security deter- a portfolio of 30 securities? How many of these are
mined? For a portfolio? variances, and how many covariances?
7‐3 The Markowitz approach is often referred to as a 7‐13 When, if ever, would a stock with a large risk
mean‐variance approach. Why? (standard deviation) be desirable in building a
portfolio?
7‐4 How would the expected return for a portfolio of
500 securities be calculated? 7‐14 Evaluate the following statement: As the number of
7‐5 What does it mean to say that portfolio weights securities held in a portfolio increases, the impor-
sum to 1.0 or 100 percent? tance of each individual security’s risk decreases.
7‐6 What are the boundaries for the expected return of 7‐15 Should investors generally expect positive correla-
a portfolio? tions between stocks and bonds? Bonds and bills?
Stocks and real estate? Stocks and gold?
7‐7 Many investors have known for years that they
should not “put all of their eggs in one basket.” How 7‐16 What are the inputs for a set of securities using the
does the Markowitz analysis shed light on this old Markowitz model?
principle? 7‐17 Evaluate this statement: For any two‐stock portfo-
7‐8 Evaluate this statement: With regard to portfolio lio, a correlation coefficient of −1.0 guarantees a
risk, the whole is not equal to the sum of the portfolio risk of zero.
parts. 7‐18 Agree or disagree with this statement: The variance of
7‐9 How many, and which, factors determine portfolio a portfolio is the expected value of the squared devia-
risk? tions of the portfolio’s returns from its mean return.
7‐10 What is the relationship between the correlation 7‐19 Evaluate this statement: Portfolio risk is the key
coefficient and the covariance, both qualitatively issue in portfolio theory. It is not a weighted aver-
and quantitatively? age of individual security risks.
7‐11 How many covariance terms would exist for a port- 7‐20 Agree or disagree with these statements: There are
folio of 10 securities using the Markowitz analysis? n2 terms in the variance–covariance matrix, where
How many unique covariances? n is the number of securities. There are n(n − 1)

c07.indd 190 1/23/2016 10:07:49 PM

 Questions 191

total covariances for any set of n securities. Divide c. Having established the portfolio weights, the
by two to obtain the number of unique calculation of the expected return on the
covariances. portfolio is independent of the calculation of
7‐21 Holding a large number of stocks ensures an opti- portfolio risk.
mal portfolio. Agree or disagree and explain your d. When adding a security to a portfolio, the
reasoning. average covariance between it and the other
securities in the portfolio is less important
CFA than the security’s own risk.
7‐23 Given the large‐cap stock index and the govern- 7‐27 Select the correct statement from among the
ment bond index data in the following table, cal- following:
culate the expected mean return and standard a. The risk of a portfolio of two securities, as
deviation of return for a portfolio 75 percent measured by the standard deviation, would
invested in the stock index and 25 percent invested consist of two terms.
in the bond index. b. The expected return on a portfolio is usually a
weighted average of the expected returns of
Assumed Returns,Variances, and Correlation
the individual assets in the portfolio.
Large‐Cap Government c. The risk of a portfolio of four securities, as
Stock Index Bond Index
measured by the standard deviation, would
Expected return 15% 5% consist of 16 covariances and 4 variances.
Variance 225 100 d. Combining two securities with perfect
Standard deviation 15% 10% negative correlation could eliminate risk
Correlation 0.5 altogether.
7‐28 Select the incorrect statement from among the
following:
CFA a. Under the Markowitz formulation, a portfolio
7‐24 Suppose a risk‐free asset has a 5 percent return of 30 securities would have 870 covariances.
and a second asset has an expected return of b. Under the Markowitz formulation, a portfolio
13 percent with a standard deviation of 23 per- of 30 securities would have 30 variances in
cent. Calculate the expected return and standard the variance–covariance matrix.
deviation of a portfolio consisting of 10 percent of c. Under the Markowitz formulation, a portfolio
the risk‐free asset and 90 percent of the second of 30 securities would have 870 terms in the
asset. variance–covariance matrix.
7‐25 Consider the following information for Exxon and d. Under the Markowitz formulation, a portfolio
Merck: of 30 securities would require 435 unique
t
Expected return for each stock is 15 percent. covariances to calculate portfolio risk.
t
Standard deviation for each stock is 7‐29 Concerning the riskiness of a portfolio of two
22 percent. securities using the Markowitz model, select the
t
Covariances with other securities vary. correct statements from among the following set:
Everything else being equal, would the prices of a. The riskiness depends on the variability of the
these two stocks be expected to be the same? securities in the portfolio.
Why or why not? b. The riskiness depends on the percentage of
7‐26 Select the correct statement from among the portfolio assets invested in each security.
following: c. The riskiness depends on the expected return
a. The risk for a portfolio is a weighted average of each security.
of individual security risks. d. The riskiness depends on the amount of
b. Two factors determine portfolio risk. correlation among the security returns.

c07.indd 191 1/23/2016 10:07:49 PM

 192 CHAPTER 7 PORTFOLIO THEORY

7‐30 Select the correct statement from the following statements regarding the Markowitz model:
a. As the number of securities held in a portfolio increases, the importance of each individual security’s risk
also increases.
b. In a large portfolio, portfolio risk will consist almost entirely of each security’s own risk contribution to the
total portfolio risk.
c. In a large portfolio, the covariance term can be driven almost to zero.
d. As the number of securities held in a portfolio increases, the importance of the covariance relationships
increases.

Problems
7‐1 Calculate the expected return and risk (standard deviation) for General Foods for
2014, given the following information:

Probabilities 0.15 0.20 0.40 0.10 0.15

Expected returns (%) 20 16 12 5 −5

7‐2 Four securities have the following expected returns: A = 15 percent, B = 12 percent,
C = 30 percent, and D = 22 percent.
a. Calculate the expected returns for a portfolio consisting of all four securities under
the following conditions:
b. The portfolio weights are 25 percent each.
c. The portfolio weights are 10 percent in A, with the remainder equally divided
among the other three stocks.
d. The portfolio weights are 10 percent each in A and B and 40 percent each in C
and D.
7‐3 Assume the additional information provided below for the four stocks in Problem 7‐2.

Correlations with

σ (%) A B C D

A 10 1.0
B 8 0.6 1.0
C 20 0.2 −1.0 1.0
D 16 0.5 0.3 0.8 1.0

a. Assuming equal weights for each stock, what are the standard deviations for the
following portfolios?
A, B, and C
B and C
B and D
C and D

c07.indd 192 1/23/2016 10:07:49 PM

 Computational Problems 193

b. Calculate the standard deviation for a portfolio consisting of stocks B and C,

assuming the following weights: (1) 40 percent in B and 60 percent in C and
(2) 40 percent in C and 60 percent in B.
c. In part a. which portfolio(s) would an investor prefer?

Computational Problems
The following data apply to Computational Problems 7‐1 through 7‐4. Assume expected returns
and standard deviations as follows:

EG&G GF

Return (%) 25 23
Standard deviation (%) 30 25
Covariance 112.5

The correlation coefficient, ρ, is 0.15.

Proportion in
(1) (3)
EG&G GF Portfolio Expected (2) Standard Deviation
wi wj = (1 − wi) Returns (%) Variance (%)

1.0 0.0 25.0 900 30.0

0.8 0.2 24.6 637 25.2
0.6 0.4 24.2 478 21.9
0.2 0.8 23.4 472 21.7
0.0 1.0 23.0 625 25.0

7‐1 Confirm the expected portfolio returns in column 1.

7‐2 Confirm the expected portfolio variances in column 2.
7‐3 Confirm the expected standard deviations in column 3.
7‐4 On the basis of these data, determine the lowest risk portfolio.
7‐5 Assume that the risk‐free rate is 7 percent, the estimated return on the market is
12 percent, and the standard deviation of the market’s expected return is 21 percent.
Calculate the expected return and risk (standard deviation) for the following portfolios:
a. 60 percent of investable wealth in riskless assets, 40 percent in the market
portfolio
b. 150 percent of investable wealth in the market portfolio
c. 100 percent of investable wealth in the market portfolio

c07.indd 193 1/23/2016 10:07:49 PM

 194 CHAPTER 7 PORTFOLIO THEORY

Spreadsheet Exercises
7‐1 Given two stocks and returns for five or six periods, construct combinations of returns
in Excel for these two stocks that will produce the following four different correlation
coefficients: −1.0, 0.0, +0.8, and −0.8. Use the CORREL function to show that your
returns achieve the indicated correlation coefficient. The following example shows
returns for two stocks, A and B, which produce a correlation coefficient of 1.0. You can
use either five periods or six periods. Note that numerous combinations are possible
in each case, so there is no one correct answer.

A B

3 3
9 9
6 6
10 10
2 2
19 19

CORREL 1 .0

7‐2 The data below are annual returns for General Foods (GF) and Sigma Technology (ST)
for the period 2000–2014. Sigma Technology is highly regarded by many investors for
its innovative products. It had returns more than twice as large as that of General
Foods. Assume an investor placed half her funds in General Foods and half in Sigma
Technology during this 15‐year period. Her objective was to earn a larger return than
that available in General Foods alone. Assess the performance of the portfolio relative
to the performance of each individual security.
a. Calculate the arithmetic mean return for each stock.
b. Calculate the standard deviation for each stock using the STDEV function in the
spreadsheet.
c. Calculate the correlation coefficient using the CORREL function in the
spreadsheet.
d. Calculate the covariance using the COVAR function in the spreadsheet.
e. Calculate the portfolio return assuming equal weights for each stock.
f. Set up a calculation for the standard deviation of the portfolio that will allow you
to substitute different values for the correlation coefficient or the standard devia-
tions of the stocks. Using equal weights for the two stocks, calculate the standard
deviation of the portfolio.
g. How does the portfolio return compare to the return on General Foods alone?
How does the risk of the portfolio compare to the risk of having held General
Foods alone?

c07.indd 194 1/23/2016 10:07:50 PM

 Spreadsheet Exercises 195

h. Assume that the correlation between the two stocks had been 0.10. How much
would portfolio risk have changed relative to the result calculated in f?

GF ST

2014 −0.141 0.222

2013 0.203 0.079
2012 −0.036 −0.220
2011 −0.204 0.527
2010 0.073 −0.628
2009 −0.111 0.684
2008 0.023 1.146
2007 0.291 0.564
2006 0.448 0.885
2005 0.482 0.433
2004 0.196 0.516
2003 0.103 −0.056
2002 0.075 0.153
2001 0.780 1.207
2000 0.254 0.736

7‐3 Fill in the spreadsheet below to calculate the portfolio return and risk between Zenon
and Dynamics, given the 10 years of annual returns for each stock and portfolio
weights of 50/50.
a. How would your answer change if the weights were 40 percent for Zenon and
60 percent for Dynamics?
b. How would your answer change if the weights were 30 percent for Zenon and
70 percent for Dynamics?

Zenon Dynamics

Expected return
Variance
Standard deviation

Covariance
Weight for Zenon 50%
Weight for Dynamics 50%

Expected portfolio return

Portfolio variance
Portfolio standard deviation

c07.indd 195 1/23/2016 10:07:50 PM

 196 CHAPTER 7 PORTFOLIO THEORY

Zenon Dynamics

Zenon return (%) Dynam return (%)

9.89 –47.67 2014
–12.34 30.79 2013
13.56 24.78 2012
34.56 7.89 2011
–15.23 24.42 2010
20.09 34.56 2009
7.56 67.56 2008
16.47 44.67 2007
18.34 78.56 2006
15.56 51.00 2005

Checking Your Understanding

7‐1 The expected return for a security is a weighted average of the possible outcomes that
could occur. It is the best one‐point estimate of the return. If this opportunity were to
be repeated for a large number of trials, the average return realized would be the
expected return.
7‐2 Assuming a normal probability distribution, we can be quite confident.
7‐3 The benefits of diversification kick in immediately. Therefore, two securities provide
better risk reduction than one, three are better than two, and so forth. However, at
some point, there is very little benefit to be gained by adding securities (the gains are so
small as to be insignificant), and therefore, the benefits of diversification are limited.
7‐4 Negative correlation means that security returns move inversely to each other. This
provides better risk reduction because the negative movement of one security is offset
by the positive movement of another security.
7‐5 Correlation is a more informative comovement measure because it provides insight
regarding both the direction and the strength or the relationship between security
returns, whereas covariance offers guidance only about the direction of the relation-
ship. Both are quantitative measures of comovement and either measure can be used
to calculate portfolio risk. The two are related as follows: AB= AB A B.
7‐6 When adding a security to a well‐diversified portfolio, what matters is its relationship
to the other securities and not its own individual risk. If the security is negatively cor-
related with the other securities in the portfolio, having a large risk will work to reduce
the overall risk of the portfolio.

c07.indd 196 1/23/2016 10:07:50 PM