PART

Gettin}tarted

CHA p TER 1
Intragwtwn to Managerial Economics

CHAPTER 2
Basic Training

INTEGRATING CASE STUDY

Stagy I: Olsen i Preowned CD Players
 CHAPTER

Introduction o Managerial
Eco mics:

• Preview
• The Circular Flow of Economic Activity
The Natere of the Firm
The Rationale for the Firm
The Objective of the Firm'
Maximizing versus Satisficing
The Principal-Agent Problem
Constrained Decision Making.
• The Concept of Economic Profit
• Profit in a Market. System
U Economics and Decision Making
I Summary
I Discussion Questions
U Problems
• Appendix: Present Value Analysis
Present Value of an Amount
Present Value of an Annuity
Problems

3
 4 PART I Getting Started

PREVIEW
For most purposes, economics can be divided into two broad categories: microeco-
nomics and macroeconomics. Macroeconomics is the study of the nconornic system
ajpl4lt includes techniques for analyzing chap ge s in totatotitpinitntah"
the_ er price index, the—unempl2yment rate, ^fia —
ployment, exports and im,
pts.
I
addresses questions about the effect of changes in invest-
ment, government spending, and tax policy on exports, output, employment, and
prices Only aggregate levels of these variables are considered. But concealed in the
aggregate data are countless changes in the output levels of individual firms, the
consumption decisions of individual consumers, and the prices of particular goods
and services.
Although macroeconomic issues and policies command m'uch of the attention in
newspapers and on television, the mkroditnensions of the economy are also important
and are often more relevant to the day-to-day problems facing the manager cr
nomics focuses on the behavir of the individu 1actors on the economic stage that is,
firms and individuals-and their interaction in markets.
Managerial economics should be thought of as applied microecondmics. It is an ap-
plication Of the part-of microeconomics that fOcuses on the topics that are of gërest-
ortance to managers. The iQpicslnciude_uenlanu, - -: production, cost,
---------------
interest and tmp
pricing, market structure and government tegulation.IA strong as the principles
that govern the- economic behavior of firms and indiiduals is an important managerial
talent. The rational application of these principles should result in better managerial
decisions, higher profits, and an increae in the value of the firm.
In general, managerial economics can be used by the goal-oriented manager in
two ways. First, given an existing economic environment, the principles of manager-
ial economics provide a framework for evaluating whether resources are being
allocated efficiently within a firm. For eample, economics can help the manager -de -
termine if profit could be increased by reallocating labor from a marketing activity to
the production line. Second, these principles help managers respond to various eco-
nomic signals. For example, given an increase in the price of output or the develop-
ment of a new lower-cost production technology, the appropriate managerial
response would be to increase output. Alternatively, an increase in the price of one in-
put, say labor, may be a signal to substitute other inputs, such as capital, for labor in
the production process.
The tools developed in the following chapters will increase the effectiveness of de-
cision making by expanding and sharpening the analytical framework used by man-
agers to make decisions. Thus, a working knowledge of the principles of managerial eco-
nomics can increase the value of both the firm and the manager.
This chapter sets the stage for the development of managerial economic skills. First,
the interrelationships among consumers, firms, and resource owners in a market econ-
omy are outlined. Next, the nature and objective of the firm and the importance of
profit as an incentive for firms to respond to consumer demands for output are dis-
cussed. Following that, the potential for owners and managers to have different objec-
tives (i.e., the principal-agent problem) is discussed. Finally, the role of economics in de-
cision making is considered.
 / CHAPTER 1 Introduction to Managerial Economics 5
°
($)
Product
Goods and services Goods and services
markeLs

Households
LFirmms

Economic Economic
resources i resources
Income (S) Factor
markets PaYmt5

THE CIRCULAR FLOW OF ECONOMIC ACTWITY

Individuals and firms are the funiamental participants in a market economy. Individu-
als own or control resources that have value to firms because they are necessary inputs
in the production process. These resources are broadly classified as labor, capital, and
natural resources. Of course, there are many types and grades of each resource. Labor
specialties vary from street sweepers to brain surgeons; capital goods range from
brooms to computers. Most people have labor resources to sell, and many own capital
and/or natural resources that are rented, loaned, or sold to firms to be used as inputs in
the production process. The money received by an individual from the sale of these re-
sources is called a factor payment. This income to individuals then is used to satisfy their
consumption demands for goods and services.
The interaction between individuals and firms occurs in two distinct arenas. First,
there is a product market where goods and services are bought and sold. Second, there
is a market for factors of production where labor, capital, and natural resources are
traded. These interactions are depicted in Figure 1. 1, which describes the circular flow
of income, output. meresources,
inand a
factor
ma a teconomy
In e uc mar et shown in the top part of the figure, individuals demand goods
and services in order ta atisfy thcir consumption dCsires The .y ruakethese demands
known by bidding in the product market for these goods and services. Firms, anxious to
earn profit& respond to these demands by supplying goods and services to that market. The
firm's production technology and input costs determine the supply conditions, while con-
sumer preferences and income (i.e., the ability to pay) determine the demand conditions.

'It is necessary to distinguish between wani and demand. Many individuals want goods and services but
cannot afford to buy them. The term demand implies that the consumer has both the desire to have the
good and the ability to bu y it.
 6 PART I Getting Started

The interaction of supply and demand determines the price and quantity sold
uct market, purchasing power, usually in the form of money, flovs from consumers to
Mfirms. At the same e ime, g o s an services ow m e O on— corn firms to
consumers.
.Ictorjnai1cejshown at the bottom of Figure 1.1. Here, the flows are the re-
verse of those in thëroduct market. Individuals are the suppliers in the factor market.
They supply labor services, capital, and natural resources to firms that demand them to
produce goods and services. Firms indicate the strength of their desire for these inputs
by bidding for them in the market. The flow of money is from firms to individuals, and
factors of production flow from individuals to firms. The prices of these prodnctive fac-
tors are set in this market.
Prices and profits serve as the signals for regulating the flows of money and re-
sources through the factor markets and the flows of money and goods through the
product market)For example, relatively high prices and profits in the personal om-
puter industry in the 1980s signaled producers to increase production and send more
units of output to the product market. To produce more computers, more labor and cap-
ital were required. Firms raised the prices they would pay for these resources in the fac-
tor market to signal resource owners that higher returns were now available. The result
was rapid growth in the personal computer industry as resources were bid away from
other industries. in the early 1990s, this market hecam very competitive and prices fell
substantially. Although total unit sales increased, the profit on each computer was
smaller, and some firms struggled to keep total profit at an acceptable level. Consumers
benefited greatly as better and more powerful computers became available at lower
and lower prices.
In the market economy depicted by this circular flow, individuals and firms are
highly interdependent; each participant needs the others. For example, an individual's
labor will have no value in the market unless there is firm that is willing to pay for it.
Alternatively, firms cannot justify production unless some consumers want to buy their
products. As a result, all participants have an incentive to provide what others want. All
participate willingly because they have something to gain by doing so. Firms earn prof-
its, the consumption demands of individuals are satisfied, and resource owners receive
wage, rent, and interest payments, if some do not benefit by buying and selling in these
markets, they are not required to do so. Thus, one can be sure that no individual is made
worse off by voluntary trade in these factor and product markets. Indeed, the gains that
accrue to the individual participants form the essence of a market economy.

THE NATURE OF THE FIRM -. .---•

i n order to earn profits, the firm organizes the factors of production to produce goods
and services that will meet the demands of individual consumers and other firms. The
concept of the firm plays a central role in the theory and practice of managerial eco-
nomics. Thus, a significant part of this text is focused on production, cost, and the orga-
nization of firms in the marketplace. These topics form the basis for what is known in
economics as the theory of the firm. An understanding of the reason for the existence
of firms, their specific role in the economy, and their objective provides a background
for that theory.
 CHAPTER 1 Introduction to Managerial Economics 7

The Rationale for the Firm

In a free-market economy, the organization and interaction of producers (i.e.. firms)
and conumers is accomplished through the price system. There is no need for an y cen-
tral direction by government, nor is such central control or planning thought to be de-
sirable. Within the firm, however, transactions and the organization of productive fac-
tors are generally accomplished by the central control of one or more managers. For
example, workers subject themselves almost completely to management during the
work period.Thus. there is an apparent dichotomy in the organization of production in
a market economy. Theprice system guides the decentralized irttcraction among con-
sumers and firms, whereas central planning and control tend to guide the interaction
within firms. This raises the question, Why is the production system not completely
guided by price signals? That is, why do firms exist in a market economy?
Essentially. firms exist as otgaxlizations because the total cost of producing an y rate
of output is lower than if the firm did not exist: There are several reasons wh y these
costs are lower. First, there is a cost of using the price s ystem to organize production.
The cost of obtaining information on prices and the cost of negotiating and concluding
separate contracts for each step of the production process would be very hurdens<me.
Firms often hire labor for long periods of time under agreements that specify onl y that
a wage rate per hour or day will be paid for the workers doing what ihey are asked.That
is, one general contract covers what usually will be a large number of transactions be-
tween the owners and workers. Te two parties do not have to negotiate a new contract
every time the worker is given a new assignment. The saving of the transactions costs
associated with such negotiations is advantageous to both parties, and thus both labor
and management voluntarily seek out such arrangements,
A secondary explanation for the existence of firms is that some government inter-
ference in the marketplace applies to transactions among firms rather than within firms.
For example, sales taxes usually apply only to transactions between one firm and an-
other. In some states, a construction company may have to pay sales tax on cabinets pur-
chased from an independent cabinetmaker. By hiring that person, this tax is avoided
and the cost of producing output is reduced. By internalizing some transactions within
the firm that would otherwise be subject to those interferences, production costs are re-
duced. Because this is a secondary factor, firms would exist in the absence of such in-
terference, but it probably contributes to the existence of more and larger firms.
Given that production costs are reduced by organizing production factors into
• firms, why won't this process continue until there is just one large firm, such as a giant
General Motors or Exxon that produces all goods and services for the entire economy?
There are at least two reasons. First, the cast of organizing transactions within the firm
tends to rise as the firm gets larger. Logic dictates that the firm will intenialize the
lower-cost transaction first, and then the higher-cost transactions. At some point, these
internal transactions costs will equal the costs of transacting in the market. At that
point, the firm will cease to grow. For example, all automobile producers in the world
buy UrCS from companies that specialize in the production of rubber products. Surely.

2This section draws on the classic article by R. H Coase, "The Nature of the Firm," Econo,nica (November
1937): 386-405.
 8 PART I Getting Started

Ford and General Motors must have considered building plants to produce their own
tires. It can be inferred that the cost of developing the new management skills required
for such a different type of production, and the difficulty of managing an even larger
and more complicated business, must have been greater than the cost of continuing to
buy tires from Goodyear, Michelin, or other producers.
Another example is legal services. Usually, attorneys are not an integral part of the
production process, but are needed periodically. It would be too costly for many firms
to have full-time attorneys whose services would not be needed on a continual basis. So,
rather than having a full-time lawyer employed by the firm, legal services are contracted
on a when-needed basis- The cost of such an arrangement for most flrms islower. In con-
trast, large firms that have a continual need for legal services generally have an in-house
legal staff, but even they make extensive use of outside law firms.
A second factor constraining firm size is the liniitatión of an entrepreneur's organi-
zational skill. Resources within the company may not be efficiently allocated if the firm's
size exceeds the manager's ability to control the operation. To overcome this problem,
many large firms are organized into groups of divisions referred to as profit centers. The
management of each of these seeks to maximize that division's profit. By having a num-
ber of smaller organizations, each being managed somewhat independently, the problem
of limited ability to control the larger firm is at least partially overcome.
Both of these reasons for a limit on the size of the firm fall under the heading of
what economists have termed diminishing returns to management. Stated another way,
production costs per unit of output will tend to rise as firms grow larger, because of lim-
ited managerial ability. It should be noted that many large firms recognize the problem
of excessive size and decentralize by establishing a number of separate divisions or
profit centers that act as individual firms.

Key Ct rncepts
• The ifiteractioLt of individuals and firip.s n a market ecannw can
as a circular flow of money, goods aid services. and resources.thrtt h rcduit
nd factor markets.
a •.
• Firms exist because the costs 01 production ar lower and re1gu.to ie owa
of labor and capital arc highe.t tta if.the firn*did w' evz........
• Limits are imposed on the size of flr because the cost ofrganizi tg rarc -
tions rises as the firm becomes larger and . because manaeril aIilit is linttd..

The Objective of the Firm

This book approaches microeconomics from the perspective of efficient management
of a business or other organization. To be able to discuss efficient or optimal decision
making requires that a goal or objective be established. That is,a management decision
can only he evaluated against the goal that the firm is attempting to achieve. Tradition-
ally, economists have assumed that the objective of the firm is to maximize profit. That
is, it is assumed that managers consistently make decisions in order to maximize profit.
But profit in which period? This year? The next five years? Often, managers are ob-
served making decisions that reduce current year profits in an effort to increase profits
 CHAPTER 1 Introduction to Managerial Economics
9
in future years, Expenditures for research and deve lopment; new capital equipment,
and major marketing programs are but a few examples of activities that reduce profits
initially but will significantly increase profits in later years.3
As'both current and future profits are important, it is assumed that the goal is to
maximize the present or discounted value of all future profits. 4 Formally stated, the goal
or objective function for the firm is to

maximize: PV()
1+r + (1+r) 2 • ry(1-1)
±
whe re 'It , is piofii in
time period r, and r is an appropriate discount rate used to reduce
future profits to their present value. Using the Greek letter Y, to indicate that each of
the terms on the right-hand side of the eq uation have been added together, the objec-
tive function can be written as

maximize: PV(rr) -
,
Thepresent value of all future profits also can be interpreted as t he value of the
firm, that is, what a willing buyer would pay for the business. Thus, to maximize the dis-
counted value of all future profits is equivalent to maximizing
terms profit maximization and value maximization the value of the firm, The
remainder of this book. will be used interchangeably in the
Most of the large firms in mrket economies are corporations where ownership is
spread among literally thousands of individuals, each of whom owns shares of stock that
represent that ownership,These owners elect a board of directors that, in turn, hires ex-
ecutives who will manage the firm. It is now common to hear these managers talk in
terms of making decisions that will maximize shareholder value. This is simply another
way of stating the goal of profit maximization,

Maximizing versus Satisficing

Considerable controversy has developed about the "realism" of this assumed manage-
ment goal. Advocates of alternative theories of the firm argue that the behavior of real-
world managers is not always consistent with the profit-maximization goal. Other ob-
jectives are seen as being at least as important. Among the alternatives that have been
identified are maximizing total revenue, maximizing employment tenure and depart-
mental budget, maximizing executive salaries, achieving "satisfactory" profit levels,
maximizing the manager's individual utility function, and maximizing market share
subject to a satisfactory profit constraint. Indeed, an entire literature has developed on
the subject of "satisficing" rather than "maximizing" management objecti*s.

3
Some managers who must report profit Performance monthly or quarterly claim that the pressure for
increased
long-term short-term
profit. profits may cause them to make decisions that increase these profits at the expense of
4 Many students will already studied the concept of present value. For those who have not, this
principle is developed more fully in the appendix to this chapter.
See H. A. Simon, "Theories of Decision Making in Economics," American Economic Review 49, no.3
(1959):253-283; and R. M. Cyert and J. A. March, A Behavioral Theory of the Firm (Upper Saddle River,
NJ: Prentice-Hall, 1963).
10 PART Getting Started

Some critics of the profit-maximization assumption argue that it is unrealistic be-

iause managers must function in an environment characterized by inadequate infor-
mation and uncertainty about the outcome of any strategy that might be adopted.
Therefore, as a practical matter, it is really impossible to maximize profit. Economist
Fritz Machlup responded to this criticism by comparing managing a firm to driving a
car.6 When deciding whether to pass another car on a two-lane highway, a driver must
consider a variety of conditions. A few of these include: the speed of the driver's car, the
car that is to be passed, and that of any oncoming vehicles; the road conditions; the ex-
istence of any curves or intersections that may be upcoming; and lighting conditions.
Obviously, to analyze this situation completely prior to passing the ear ahead might re-
quire the assistance of a computer and several working days.
Clearly, however, the typical driver makes an intuitive evaluation of these conditions
in a matter of a few seconds or less and makes a decision. Almost all of the time, it is the
correct decision. Machiup contends that managers are seeking to maximize profit and
simply react intuitively to a set of conditions characterized by incomplete information
and uncertainty in the same manner as the automobile driver. Thus, while there may be
additional information that would result in even greater profit, managers are striving to
generate the greatest possible profit given the limited information available.
Although some managers may have other goals, most of the criticism leveled at the
profit-maximization assumption may be irrelevant. Economics is less interested in how
some managers really act than in understanding the economic environment in which
managers must function and, more importantly, in developing a framework for predict-
ing managerial responses to important changes in that environment.
In this context, Machiup argues that it is fruitless to worry about how real-world
business managers really behave and what their goals really are. The essential question
is: If we assume that firms attempt to maximize profit, will the principles of economics
derived from that objective function explain the behavior of real-world firms? If the an-
swer is yes, it does not matter if the assumed goal is not entirely realistic. For example,
suppose that there are two firms, A and B, that are identical except in one respect: The
manager of A works 16 hours a day, seven days a week, whereas the manager of B works
only four days eachweek, and leaves the office shortly after lunch each day to play golf,
Because of manager A's commitment to work, it is probable that profits for firm A are
greater than at firm B, but this difference in commitment to work and its effect on profit
is not the issue. The fundamental question is: Will the two managers respond the same
way to changes in economic conditions? If the price of their product increases, do both
use the profit-maximizing strategy of increasing the rate of output? The manager of A
may take action first, because he is on the job most of the time, but the response should
be the same for both firms.
Most economists agree that the principles of managerial microeconomics do in-
deed allow accurate predictions of managerial decisions, and that profit maximization
provides a useful assumption in that context. Indeed, no general theory has yet been
proven to predict more accurately than the models based on profit maximization. Thus,
in this book, it will be assumed that the objective of the firm is to maximize profit or,
equivalently, to maximize the value of the firm.

6F, Machiup. "Theories of the Firm: Matginalist, Behavioral, Managerial," American Economic Review 57.
no. 1 (1967):1-33.
 CHAPTER I Introduction to Managerial Economics
11
The Principal-Agent Problem
In the modern corporation, the owners or stockholders (i.e., the principals) hire man-
agers(i.e., agents) to conduct the day-to-day operation of the firm, These managers are
paid a salary to represent the interest of the owners, ostensibly, to maximize the value
of the firm. A board of directors is elected by the owners to meet regularly with the
managers to oversee their activity and to tr y to ensure the managers are, in fact, acting
in the best interest of the owners.
Because of the difficulty of monitoring the managers on a continual basis, it is pos-
sible that goals other than profit-maximition may be pursued, i
n addition to iliose
mentioned earlier, the managers may seek to enhance their positions by spending cor-
porate funds on fancy offices, excessive and expensive travel, club. rnemberships, and so
forth. In recent y ears, many corporations have taken action to align the interests of own-
ers with the interests of the managers by tying a large share of managerial compensa-
tion to the financial performance of the firm.
For example, the manager may be given a basic salary plus potentially large
bonuses for meeting such goals as attaining a specified return on capital, growth in
earnings, and/or increase in the price of the firm's stock. With regard to the latter, the
use of stock options awarded to top managers is a most effective wa y to insure that
managers act in the interest of the shareholders. ipicaiiy, the art
al igement provides
that the manager is to receive an option to buy a specified number of shares of com-
mon stock at the current markt price for a specified number of years. The only way
the executives can benefit from such an arrangement is if the price of stock rises dur-
ing the specified term. The option is exercised by buying the shares at the specified
price, and the gain equals the increase in share price multiplied by the number of
shares purchased. Sometimes the agreement specifies that the stock must be held for/
several years following purchase.
Essentially, this option arrangement makes the manager a de facto owner, even if
the option has not been exercised. In almost every case of a report of unusually high ex-
ecutive compensation, the largest part of that compensation is associated with gains
from stock options.
Consider the following example. Smith is hired as the president of a firm at an an-
nual salary of $500,000 plus a five-year option to buy 100,000 shares of stock at the cur-
rent market price of $50 per share. Assume that within five years the price of the stock
has increased to $75 per share. Smith exercises the option by buying 100,000 shares for
$5,000,000, which have a market value of $7,500,000.
In the year the options are exer-
cised, Smith has a gain (i.e., additional compensation) of $2,500,000. However, if the
price of the stock had remained unchangd or had declined, this option would have no
value and Smith would have received no additional compensation.'

7A study
at the University of Wishirgton (Narayanan, 1996) found that managers who were paid only a
cash salary (no stock opfions) tended to make decisions that resulted in less investment that would increase
future profits. In contrast, those whose contracts were heavily weighted in common stock grants and
options tended to overinvest. It was shown that management contracts that provided for both cash and
stock tended to produce efficient long-term investment rates. See M. P. Narayanan, "Form of Compensation
and Managerial Decision Horizon," Journal of Ffnwcja1 and Quantitative Analysis
1966); 467-491. 31, no- 4 (December
12 PART I Getting Started

Executive Pay in the United States

If there were any questions about the potential rewards to the top executives
at America's largest corporations, consider the levels of compensation of the 10
highest-paid executives in 1996:
1.996 Salary and Long-term Total
Bonus Compensation Pay
Lawrence Coss
Green Tree Financial $102,499,000 none $102,499,000
Andrew (rove
Intel 3,003,000 $94,587,000 97.590.000
Sanford Weill
Travelers Group 6,330,000 87,82,000 - 94,158,000
Theodore Waitt
Gateway 2000 965,000 80,361,000 81,326,000
Anthony 0-Reilly
H. J. Heinz 2,736,000 61,500,000 64,236,000
Sterling Williams
Sterling Software 1,448,000 56,801,000 58,249,000
• L..T Deeu.
,,O,in
Citicorp 3,467,000 40,143,000 43,610,000
Stephen Hilbert
Conseco 13,962,000 23.450,000 37,412,000
Casey Cowell
U. S. Robotics 3,430,000 30,522,000 33,952,000
James Moffett
Freeport-McMoran C&G 6,956,000 26,776,000 33,732,000

Note that except for Lawrence Coss, by far, the largest part of compensation
comes in the form of long-term compensation, usually stock options.
Bill Gates, the CEO at Microsoft, was paid "only" about $350,000. Lest one
worry about his financial condition, the value of his 300 million shares of stock in-
creased by $11.4 billion during theyear!

Constrained Decision Making

The essence of the science of economics is determining optimum behavior where that
behavior is subject to constraints. Profit maximization is constrained by the limited in-
formation available to the manager. In genera!, constraints on managerial decisions in-
volve legal, moral, contractual, financial, and technological considerations. Legal con-
straints include the array of federal, state, and local laws that must be obeyed by all
citizens, both individual and corporate. Areas where managers seem to be having some
legal difficulty include environmental laws, especially those relating to pollution and
the disposal of hazardous wastes, and employment law, including wrongful termination
and sexual harassment matters. Moral constraints apply to actions that are not illegal
but are sufficiently inconsistent with generally accepted standards of behavior to be
considered improper. Contractual constraints bind the firm because of some prior
 CHAPTER 1 introduction to Managerial Economics 13
agreement such as a long-term lease on a building, or a contract with a labor union that
represents the firm's employees.
Finlly, there are financial and technological constraints. An exairiple of a financial
constraint occurs when a department of a firm is assigned a budget for the next year and
managers are given orders such as "maximize production subject to this budgeted
amount." Technological constraints set physical limits on the amount of output per unit
of time that can he generated by particular machines or workers. As will be shown in
subsequent chapters, making optimizing decisions under financial or technological con -
straints is a fiindamenial part of managerial economics

Case Study
Adam Smith and the Invisible Hand

The eighteenth-century philosopher Adam Smith is regarded as the father of modern

economics. He understood clearly how a decentralized market economy would function
with little, if any, outside regulation required. In particular, he saw that by seeking to
maximize their self-interest, the individual participants in this economy would achieve
socially desirable results. In his book An Inquiry into the Nature and Causes of the
Wealth of Nations* (originally published in 1776), Smith writes
It is not from the benevolence of the butcher, the brewer, or the baker
that we expect our dinner, but from their regard to their own self-interest. We
address ourselves, not to their humanity but to their self-love, never talk to
them of our own necessities, but of their advantages.
The coordinating mechanism in the economy, the price system, was described by
Smith as an "invisible hand" that guided private decisions in socially beneficial ways:
Every individual endeavors to employ his capital so that its produce may
be of greatest value. He generally neither intends to promote the public inter-
est, nor knows how much he is promoting it. He intends only his own security,
only his own gain. And he is in this led by an invisible hand to promote an end
which was no part of his intention. By pursuing his own interest he frequently
promotes that of society more effectually than when he really intends to pro-
mote it.
In addition to this rationale for a market system unfettered by outside (e.g., gov-
ernment) influence or control, Smith's example of the efficiencies associated with the
mass production of pins provides insight into the role of the firm in market economy
and the advantages of mass production.

*New York: Raiidom HousefModern Library. 1,985.

 14 PART I Getting Started

To take an example . - the trade of the pinmaker; a workman not edu-

cated to this business . . . nor acquainted with the use of the machinery em-
ployed in it .. . could scarce, perhaps, with his utmost industry, make one pin in
a day, and certainly could not make tventy. But in the way in which this busi-
ness is now carried on not only the whole work is a peculiar trade, but it is di-
vided into a number of branches, of which the greater part are likewise pecu-
liar trades. One man draws out the wire, another straightens it, a third cuts it, a
fourth points it, a fifth grinds it at the top for receiving the head; to make the
head requires two or three distinct o perations: . and the important business
of making a pin is, in this manner, divided into about eighteen distinct opera-
tions.. .. I have seen a small manufactory of this kind where ten men only were
employed making upwards of forty-eight thousand pins in a day. Each person,
therefore. making a tenth part of forty-eight thousand pins, might be consid-
ered as making four thousand eight hundred pins in a day. But if they had all
wrought separately and independently, and without any of them having been
educated to this peculiar business, they certainly could not each of them have
made twenty, perhaps not one pin in a day.
Even in 1776, Adam Smith could see the advantages of an economic system based
on the self-interest of the individuai participants and the development of firms to orga-
nize labor and capital for the purpose of low-cost mass production. U

Key concepts
• It is assumd that the objective of the firm is o maximize the present vaióe of
all fw:ure profits, subject to various legal, moral, contractual, financial, and tech-
nological constrainis.
• The principles of managerial economics allow accurate prcdiciion of decision
makin g in business and other organizations.
• The principal-agent problem refers to the posihility thai owners and their
managers may have different rthje.ctives.Thesc interests can he aligned through
the use ofmanagerial compensation arrungements that tie individual compen-
satin t0 :the overall performance of the firm.

THE CONCEPT OF ECONOMIC PROFIT

Havingdetermined that the goal of the firm is profit maximization, it is necessary to de-
fine the term profit. The conventional notion of profit is relatively straightforward: Profit
is defined as revenues minus costs. However the definition of cost is quite different for
the economist than for the accountant. Consider an individual who has an MBA degree
and is considering investing $200.000 in a retail store that she would manage. The pro-
jected income statement for the year as prepared by an accountant is as shown here.
 CHAPTER 1 Introduction to Managerial Economics 15
Sales $90,000
Less: Cost of goods sold 40,000
Gross profit $50,000
Less: Advertising $ 10.000
Depreciation 10,060
Utilities 3,000
Property tax 2,000
Miscellaneous expenses 5,000 30,000
Net accounting profit - $20,000

This accounting or business profit is what is reported in publications such as the Wall
Street Journal and in the quarterly and annual financial reports of businesses. It is a
meaningful concept asfar as it goes—it just does rjot go far enough. Furthermore, the
use of this concept may result in making the wrong decision,
The economist recognizes other costs, defined as implicit costs. These costs are not re-
flected in cash outlays by the firm, but are the costs associated with foregone opportuni-
ties. Such implicit costs are not included in the accounting statements, but must he included
in any rational decision-making framework. There are two major implicit costs in the pre-
ceding example. First, the owner has $200,000 invested in the business. Suppose the best al-
ternative use for this money is a bank account paying a 5 percent interest rate. Therefore,
this investment would return $19,000 annually. Thus, $10,000 should be considered as the
implicit or opportunity cost of having the $200,000 invested in the retail store.
The second implicit cost includes the manager's time and talent. The annual wage
return on an MBA degree from a reasonably good business school may be $60,000 per
year. This is the implicit cost of managing this business rather than working for some-
one else. Thus the income statement should be amended in the following way in order
to determine economic profit:
Sales $90.000
Less: Cost of goods sold 40,000
Gross profit $50,000
Less: Explicit Costs:
Advertising $10,000
Depreciation 10,000
Utilities 3,000
Property tax 2,000
Miscellaneous expenses 5,000 30,000
Accounting profit (i.e., profit $20,000
before implicit costs)
Less: Implicit costs:
Return on $200,000
of invested capital 10,000
Foregone wages 60,000 70,000
Net 'economic profit" -- - $-50,000

From this broader perspective, the business is projected to lose $50,000 in the first
year. The $20,000 accounting profit disappears when all "relevant" Costs arc
included. Obviously, with the financial information reported in this way, an entirely
 16 PART I Getting Started

different decision might be made on whether to start this business. Another way of look-
ing at the problem is to assume that $200,000 had to be borrowed at 5 percent interest and
an MBA graduate hired at $60,000 per year to run store. In this case, the implicit costs
become explicit, and the accounting profit is the same as the economic profit (i.e.,
$-50,000) because all costs, both explicit and implicit, have been considered.
Most dccision makers are aware of this concept. The carpenter who works as an in-
dependent contractor rather than as an employee of another firm knows that his op-
portunity cost is the market wage rate for carpenters. The contractor's lament, "I lost
money on that job" may mean that his accounting profit was $100 per day when he
could have made $150 per day working fo ouiCone else.
Me important point of this discussion is that an entirely different signal may be given
to management when the concept of economic profit is used.Sometimes the operation of
economically unprofitable businesses is continued because of a failure to understand and
properly include implicit cOsts. Rational deciibn making requires that all relevant costs,
both explicit/and implicit, be recognized. The concept of economic profit accounts for all
costs and tVerefore is a more useful
management tool than the more normally defined con-
cept of Icounting profit. In the following chapters, the economic definition of profit will
be useSY, and the-term cost is defined to include all relevant costs, both explicit and implicit.

Example itv Costs
Sharon Smith is a full-time homemaker and is also an excellent seamstress She has mate-
rial for which she paid $5 per yard several years ago. The material has increased in value
during that time and could be sold back to the local fabric shop for $15 per yard. Sharon
is considering using that material to make dresses, which she would sell to her friends
and neighbors. She estimates that each dress would reqire four yards of material and
four hours of her time, which she values at $10 per houri.f the dresses could be sold for $90
each, could Sharon earn a positive econQmic profit by. making and selling the dresses?
Solution The key to this decision is appropriately accounting for both Sharon's
time ($10 per hour) and the true opportunity cost of the material, $15 per yard (the
amount she could receive by selling it to the fabric Shop). The profit calculation per
dress would be as follows:
Revenue $90
Less: 4 hours of labor at $10/hr 40
4 yards of material at $451yd ,5Q
Economic profit
Clearly, making the dresses is not going to be profitable.
If Sharon had not included the value of her time and had used the historic price of
$5 per yard as the cost of the material, she would have estimated a "profit" of $70 per
dress, that is
Revenue $90
Less: 4 yards of material at $5/yd
"Profit" $20
This is not an accurate measure of profit because it fails to account for the true oppor-
tunity cost of Sharon's time and the opportunity cost of a yard of material. She could
sell both in the market and make more than she could by producing the dresses.
 CHAPTER 1 Introduction to Managerial Economics 17

PROFIT IN A MARKET SYSTEM

Except for those who espouse the view that chai it y is the only virtue, contributors to
output expect to he paid. Each workday,rnany
, people willingly spend about one-half of
their waking hours in employment for which: the y are paid wages. Those who own land
and structures will let others use those resources as long as rent is paid for their use.
Owners of financial resources make their assets available in return for the payment of
interest.
Of the total income generated in the U.S. economy each year. abOut 73 percent is
in the form of wages. Rent accounts for 2 percent and interest comprises about 10 per-
cent. Profit, although representing only 15 percent of national income, plays two criti-
cally important roles in the functioning of the economic system. First, profit acts as a sig-
nal to producers to change the rate of output, or to enter or leave an industr y, Second,
profit is a reward that encourages entrepreneurs to organize factors of production and
take risk.
High profits in an industry are usually a signal that buyers want more output from
that industry. Those profits provide the incentive for existing firms to increase output
and for new firms to enter the market. Conversely, low profits are a signal that less out-
put is being demanded by consumers and./ or that production methods are not efficient.
When demand for output decreases, firms reduce production of that product, and re-
sources are made available for thF production of other goods and services for which de-
mand has increased. If low profits are an indication of inefficienc y in production, rather
than inadequate demand, management is signaled to reorganize the production process
or otherwise reduce costs.
It seems clear that profit is the objective of most managers and entrepreneurs. Al-
though sometimes characterized by greed, avarice, misrepresentation, or outright law-
breaking, the desire for profit drives the market economy. This quest for profit is no
more selfish than the quest for higher wages. Profit is a residual income after other par-
ticipants in production have been paid.The sellers of labor inputs want their wages be-
fore the final goods are sold to consumers. Generally, they do not want to share the risks.
Thus, it is not surprising that entrepreneurs will not commit time and resources to risky
activities unless there is the prospect of earning a profit. One function of profit is to re-
ward entrepreneurs for accepting the risk associated with their business decisions. Vir-
tually all of these decisions carry the risk that money will be lost. Clearly, if there is a
chance of a loss, there must be the chance for a gain in the form of profit; otherwise, such
risks will not he taken.
Certain firms seem to be able to earb above-normal profits on a consistent basis.
In some cases, they are continually innovative in developing new products, in reducing
production costs on existing products, or in providing unusually good customer service.
Although other firms may follow their lead and copy any new product or duplicate a
production-cost advantage, over time the truly innovative firm will have developed
even more new products and production technolog y, thus maintaining its above-
normal profit performance. In such cases. profit plays a useful social function; it en-
courages firms to develop new products. to lower production costs, and to provide
better service.
18 PART I Getting Started

In other cases, firms may earn above-normal profit because they have monopoly
power in a market. That is, other firms are prevented from offering the same product or
service for sale, thus allowing that firm to maintain high prices and profits. For example.
Resorts International, Inc., was the first firm to be granted a gambling license in At-
lantic City, New Jersey, when that activity was legalized in 1978. For three years, the firm
made large profits. As soon as licenses were given to other firms, competition increased
and Resorts' profit fell more than 50 percent. In another case, Bausch & Lomb, a large
manufacturer of precision lenses, was given permission by the U.S. Food and Drug Ad-
ministration (FDA) to produce and sell soft contact lenses. For several years the firm
had _-Monopoly bccause no othei firm was able to obtain FDA permission. During that
period, prices of these lenses were very high, as were profits for Bausch & Lomb. As
soon as other firms were able to enter the market, both prices 'and profits declined.
In cases such as these, the above-normal profit is not socially useful. Indeed, a firm
with monopoly power does not have as much, incentive to hold down prices and pro-
duction costs, to develop new products, or to provide good customer service as does a
firm in a very competitive industry. However, in some industries, the lowest production
cost can be achieved by having only one firm in the market. The generation and deliv-
ery of electric power is an example of such an industry. In most areas of the country, one
firm is given a mono poly franchise but is subject to government regulation of price to
erisute that "reasonable" prices are charged and that only a "fair" profit is earned. In
this way, the advantage of low production cost is achieved without the above-normal
profits usually associated with monopoly power.
Some critics of the capitalist system argue that profit is little more than a windfall
gain that is randomly conferred on certain people. As such, they believe that it should
be taxed away because it is "unearned.' However, even if they were correct in claiming
that profit is no more than a random windfall, the free enterprise system cannot work
without profit. That is, no one will "play the game" (Le., make investments in business)
if there is no chance of winning a prize iic the form of profit.
In a market charcterized by many firms competing against one another, above-
normal profits provide important signals, but are not likely to be maintained over long
periods of time. That is, firms already in the market respond to higher profits by in-
creasing output, and new firms will have an incentive to enter the market as well. The
result will be an increased supply of the product, lower prices, and, ultimately, lower
profits. The result in competitive markets is that profits provide important signals, but
are somewhat transitory in nature.

ase Study
dliam Henry Gates Ill and the Microsoft Money Machine

Several years ago. when his fortune was a mere several hundred million dollars, a
weekly magazine labeled Bill Gates as "America's richest 'nerd.' " By 1992, at age 36,
he had passed Donald Trump, Ross Perot, and others to be listed as America's wealth-
 CHAPTER 1 Introduction to Managerial Economics
19
iest person by Forbes magazine; at that time the value of his holdings had grown to an
estimated $6.3 billion. More recently his wealth surpassed $40 billion. How did the free
enterprise system help him to attain such phenomenal wealth?
After graduating from high school in Seattle in 1973. Gates went to Harvard.
While there, he learned that the personal computer (PC) was in the development stage.
He dropped out of school and threw himself completely into designing an operating sys-
tem for the PC. The operating system is the program that coordinates the hardware and
software of the computer. His system. MS-DOS (Microsoft Disk Operating System),
was so good that IBM agreed to use it in their line of personal computers. With IBM
setting the industry standard, other computer manufacturers quickly adopted MS-DOS
as well. Later, MS-DOS was replaced by the Windows operating system, another of Mi-
crosoft's de v elopments. Today more than 80 percent of all personal computers in the
world use this system. Gates' firm, Microsoft, Inc., titakes money on every computer
sold with Windows as the operating system. In 1997, the firm recorded over $10 billion
in revenue and more than $2 billion in net profit. It ranks third in size in the industry,
behind IBM and Hewlett-Packard. Gates' personal holdings of some 300 million shares
of common stock represent about a 25 percent ownership share of the company.
Microsoft also produces programs for word processing, spreadsheets, and a van-
etv of other applications. One of Gates' ventures has been to puichase the ekciroiiie
reproduction rights to thousands of art and photographic works from museums and li-
braries around the world.Thee will be used as part of his plan for interactive home en-
tertainment systems.
With extremely hard work,a creative mind, and a willingness to take risks, Gates has
demonstrated how the market rewards the successful entrepreneur. He was able to pro-
duce what consumers wanted at a price the y were willing to pay; the result was that both
he and they are better off! This is the essence of the free-market economic system.

ECONOMICS AND DECISION MAKING

Where do the principles of microeconomics fit in the arena of managerial decision mak-
ing? Nobel prize-winning economist Herbert Simon identifies the primary activities in
decision making:8
1. Finding occasions for making decisions.
2. Identifying possible courses of action.
3. Evaluating the revenues and costs associated with each course of action.
4. Choosing that one course that best meets the goal or objective of the firm (i.e.,
that maximizes the value of the firm).
The primary role of managerial economics is evaluating the implications of alternative
courses of action and choosing the best or optimal course of action from among those
several alternatives.

H. A. Simon. The Decision-Makin c, Prseecs: in T/n' Exce!l(iI • C as

The New Science of
Management Decisions (New York: Harper & Row. 1960).
20 PART I Getting Started

Decision making in this context implies the need for optimizing behavior. The mar-
keting vice-president strives to maximize sales revenue, the production manager attempts
to minimize cost or maximize production, and the division president's goal is to maximize
profit. As discussed in an earlier section, these management targets are constrained by
other parameters relating to that decision. For example, production costs might be mini-
mized by producing nothing, but this would he inconsistent with the firm's goal of profit
maximization.A more typical goal for the production manager would be to minimize cost,
subject to producing a specified output rate, while the objective of the marketing vice-
president would be to maximize sales, subject to a given advertising budget.
The essence of efficient and iaiunai management is constrained optimization. Vir-
tually all choices and decisions are subject to limitations, and this is where the tools of
managerial economics are most useful. The manager who can achieve the most despite
those constraints will be rewarded with a high salary, stock options, and the other
perquisites usually associated with success.
Optimization principles of managerial economics are also important in the not-for-
profit sectors of the economy. For example, universities strive to maximize the value of
teaching and research outputs, subject to an annual budget constraint. Decisions about
the level of tuition, the right mix of faculty and secretaries, and the balance between
classrooms and laboratories will benefit from the correct application of the constrained
optimization principles that are basic to economics.

key Concepts -
• Economic profit refers to revenues minus all relevant costs, Ih explicit and
• implicit.
Profit plays two roles in a market economy: (1) Changes in profit signal pro-
ducers to change the rate of productip and (2) profit is a revard to entrepre-
neurs for taking risks, being especially innovative in developing new products,
and reducing production costs.
• Firms can earn economic profits because they have monopoly power in a mar-
ket. In general, such profits are not socially useful.
• The primary decision-making role of managerial economics is in determining the
optimal course of action where there are constraints imposed on the decision.

IVIMARY
Managerial economics can be viewed as an application of that part of microeconomics
that focuses on such topics as risk, demand, production, cost, pricing, and market struc-
ture. Understanding these principles will help to develop a rational decision-making
perspective and will sharpen the analytical framework that the executive must bring to
bear on managerial decisions.
Individuals and firms interact in both the product and the factor markets. Prices of
outputs and inputs are determined in these markets and guide the decisions of all mar-
ket participants. The firm is an entity that organizes factors of production in order to
produce goods and services to meet the demands of consumers and other firms. In a
 CHAPTER 1 Introduction to Managerial Economics 21
market system, the interplay of individuals and firms is not subject to central control.
The prices of both products and factors of production guide this interaction. Within
firms, however, activity is directed by managers. Central control within the firm is ad-
vantageous because transactions and information costs are reduced.The size of the firm
is limited because transaction costs within the firm will rise as the firm grows, and be-
cause management skill is limited.
It is assumed that the goal of the firm is to maximize the value of the firm or the
present value of all future profits, defined as revenue less all costs, explicit and implicit.
Opportunity costs such as the remuneration and nterest that owners and managers
have foregone on their labor and capital must be included as costs. Failure to account
for these implicit costs may result in an inefficient allocation of resources. The objective
of profit maximization is subject to legal, moral, contractual, financial, and technologi-
cal constraints. Some economists argue that the firm's objective is a 'satisfactor y" level
of profit rather than maximum profit. The principal-agent problem arises where the
owner of a firm and the manager of that firm have different objectives.The problem can
be solved by tying part of the manager's salar y to profits and/or changes in the price of
the firm's stock.
Profit plays two primary roles in the free-market system. First, it acts as a si g nal to
producers to increase or decrease the rate of output, or to enter or leave an indusi..
Second, profit is a reward for entrepreneurial activity, including risk taking and inno-
vation. in a competitive industry, economic profits tend to be transitory. The achiee-
ment of high profits by a firm usuall y results in other firms increasing their output of
that product, thus reducing price and profit. Firms that have monopoly power ma y be
able to earn above-normal profits over a longer period: such profit does not play a so-
cially useful role in the economy.
A primary role of economics in management is in making optimizing decisions
where constraints appl y. The application of the principles of managerial economics will
help managers ensure that resources are allocated efficiently within the firm, and that
the firm makes appropriate reactions to changes in the economic environment.

Discussion Questions
1-1. Explain how firms and individuals participate and interact in the product market
and in the factor market.
1-2. Describe the difference between the accounting and the economic concept of
profit. How might accounting practices be changed to make financial statements
and reports more useful for managerial decision making?
1-3. Explain how the principles of free-market economics that guide interrelations
among firms and individuals might guide pricing and resource allocation deci-
sions within the large. multiplant firm. -
1-4. Why is it important to state a managerial objective? Could the assumption that
the managers' objective is profit maximization be useful even if their real objec-
tive is maximizing market share or their salaries?
1-5. What might he the objective or objectives of each of the following nonprofit in-
stitutions?
a. The college of business at a major state university.
h. A municipal police department.
 22 PART 1 Getting Started

C. The emergency room of a hospital..

d. A museum.
1-6. Explain the role of profit in a free marketeconomic system How can such a small
share of national income (about 15 percent) be such an important determinant of
resource allocation?
1-7. Provide examples of managerial decisions that might reduce profits for the next
few years, but would increase the value of the firm. Explain.
1-8, Some argue that businesses need Lobe "socially responsible" (e.g., reduce pollu-
tion, employ more minority workers, cease buying from and selling to countries
that are out of political favor, sell at-lower pricesto low-income people, etc).
Evaluate the effects on the firm .of being socially responsible in a competitive en-
vironment. Can one firm in a market afford to be socià}y responsible if its com-
petitors are not? Explain. Would you invest in a company that devoted a signifi-
cant part of ILS resources to being socially responsible? Why?
1-9 Milton Friedman has argued that the realism of a model is not as important as its
Predictive ability That is, a model of economic behavior that appears to be unre-
alistic but predicts well is superior to a model that seems more realistic but does
not predict well. Do YOU agree? Why or why not?
1-10. In some large businesses, division managers must report profits monthl y and. in
some cases, more often. Some managers claim that the pressure to report favor-
able results continually on a month-to-month basis causes them to make deci-
sions. that will enhance short-term profit at the expense of profit in the long run.
What actions might a manager take that would increase profit in the short run but
that would reduce shareholder value?
1-11. A firm's profit has been at approximately $10 million per year in each of the last
five years.The company has two million shares of stock outstanding, and the mar-
ket price has been about $60 per share. The top management officials have been
receiving only a salary during thisVperiod. Describe how a combined bonus and
stock-option plan couldbe developed to provide these executives with greater in-
centive to increase profit.

roblems
1-1. A recent engineering graduate turns down ajob offer at $30,000 per year to start
his own business. He will invest $50,000 of his own money, which has been in a
bank account earning 7 percent interest per year. He also plans to use a building
he owns that has been rented for $1,500 per month. Revenue in the new business
during the first year was $107,000, while other expenses were
Advertising $ 5,000
Rent io,00o
Taxes 5,000
Employees' salaries, 40,000
Supplies 5,000

Prepare two income Statements, one using the traditional accounting approach
and one using the opportunity cost approach to determine profit.
 CHAPTER 1 Introduction to
Managerial Economics 23
1-2. Tempo Electronics,
Inc., has an inventory of 5,000 unique electronic chips origi-
nally purchased at $2.50 each; then- market value is now $5 each, The Production
department has proposed to use these by puffing each one together with $6 worth
of labor and other materials to produce a w
Should that proposal be implemented? Explain. ristwatch that would be sold for $10.
1-3. Smith, a college sophomore, generally spends his summers working on the uni-
versity maintenance crew at a wage rate of $6.00 per hour for a 40-hour week.
Overtime work is always available at an hourly rate of
1.5 times the regular wage
rate. For the Coming summer, he has been offered the pizza stand
the Student Union buildj, which would havc to be open 10 hours cper oncession
day, sixat
days a week. He estimates that he can sell 100 pizzas a week at $600 each. The
production cost of each pizza is $2.00 and the rent on the stand is $150 per week.
Should Smith take the pizza concession? Explain.
1-4. An executive's employment contract caii
for a salary of $400,00j per year, a
bonus equal to 2 percent of profits in excess of
$10,000,000,and an option to buy
5,000 shares of common stock at d price of $50 per share. The market price of the
stock is $70 per share, and the firm's profits for the current year are $12,000,000
Assuming the executive e
is the total compensation forxercises the stock option and then sells the stock, what
the year?
15. A manufacrer of personal co
mputers has an inventory of 10,000 back-up Stor-
age drives that sold for $100 per unit last year. The current market price of these
drives is now $70 per unit y
computers, the price of eachByc adding one of these drives to their stock of personal
omputer is increased by $80 per unit. Should the dri-
ver be added? What is the opportunity cost of these drivers? Explain
 24 PART I Getting Started

Appendix

Present- Value Aatysis

Many transactions involve making prreceiving cash payments at various future dates.
A person who takes-out a mortage loan trades a promise to make monthly payments
for thirty years for a large amount of cash now to pay for a home. A person injured in
an automobile accident accepts an insurance company's settlement of $1,000 per month
for life as compensation for the damage associated with that injury. }ftgh-priced pro-
fessional athletes offer their skills for multivear, no-cut contracts. in all these cases, con-
ceptsrelating to the time value of money are required to make sound decisions. The
time value of money refers to the fact that adollar to be received in the future is not
worth a dollar today. Therefore, it is necessary to have techniques for measuring the
value today (i.e., the present value) of dollars to be received or paid at different paints
in the future. This section outlines the approach to analyzing problems that involve pay-
ment and/or receipt of money at one or more points in tinte
Understanding the following terms is -essential to applying time value of money
principles: - -
- Annuity: a series of equal payments per period for a specified length of time. For
example, the repayment of a loan by making forty-eight monthly payments of $200 each
isaiorm of annuity.
Amount: a specific number of dollars to be paid or received on a specified date.
Present Value: the value today of an amount or an annuity. taking into considera-
tion that interest can be earned.

1 RESENT VALUE OF AN AMOUNT -

The basic equation for the present value (PV) of an amount S is

PV=S[(1 QA-1)
+f)!
The bracketed term
1
- (1+i)
is the present value of $1 in n-periods if. the interest rate is i percent. It is called the
present-value interest factor (PVIF).
As an example, what is the present value of $1,080 in 1 year if the interest rate is 8
percent per year? Substituting S = 1,080,1 = 0.08 (i.e.. the decimal equivalent of 8 per-
cent', and n I in equation (iA-i) yields
r ___
P1' = $LO8OL(lo8)I S1.000
 CHAPTER 1 IntrOductio i to Managerial Economics
25
Note that $1,000 would increase to $1,080 in one year at 8 percent interest. Thus, th
present-value concept explicitly takes account of the potential interest that could b
eried.
Consider another problem. What is the present value of $100,000 to be received a
the end of 10 years if the interest rate is 10 percent? The problem can be expressed as

PV = 100,004--1
(1.10)10
The present-value interest factors (PVIF . )
for a range of interest rates and periods
are-given in Table I at the end of the book. Part of that table is reproduced as
Table 1A.1.

pvij =
1, 1 -•-i)'
Interest Rate

------------
1 09259 0.9091 0.8929
2 08573 0.8264 0.7972
3 0.7938 0.7513 0.7118
4 0.7350 0.6830 0.6355
6 0.6302 0.5645 0.5066
8 0,5403 04665 0.4039
10 0.4632 0.3855 03220

By reading down the "10%" column in Table IA.1 to the row for n = 10 periods,
the factor 0.3855 is found. This is the present value of $1 in 10 years at 10 percent inter-
est. Multiplying this factor by $100,000, we obtain
PV = 100,000(0.3855) $38,550
If 10 percent is the appropriate interest rate, $100,000 in 10 years is equivalent to
$38,550 today.
The process of reducing a future amount to its present value is often referred to as
discounting
because the present value is always less than the future amount. In this con-
text, the interest rate used in present-value problems is generally referred to as a dis-
count rate.
Note that the present-value factors decrease as the number of periods increases
and as the interest rate increases. Because the interest rate is in the denominator of the
present-value equation, there is an inverse relationship between the present value and
the interest rate. Furthermore, the longer the period of time. before an amount is paid,
the lower the present value of any amount.
 26 PART I Getting Started

PRESENT VALUE OF AN ANNUITY

An annuity has been defined as a series of periodic equal payments. Although the term
is often thought of in terms of a retirement pension, there are many other examples of
annuities; The repayment schedule for a mortgage loan is an annuity. A father's agree-
ment to send his son $200 each month while he is in college is another example. Usu-
ally, the number of periods is specified, but not always. Sometimes retirement benefits
are paid monthly as long as a person is alive. In other cases, the annuity is paid forever
and is called a perpetuity.
It must be emphasized that the strict definition of an annuity implies equal pay-
ments. A contract to make 20 annual payments, which increase each year by, say, 10 per-
cent, would not be an annuity. As some financial arrançments provide for payments
with periodic increases, care must be taken not to apply an annuity formula if the flow
of payments is not a true annuity.
The present value of an annuity can be thought of as the sum of the present values
of each of several amounts. Consider at annuitv of three $100 payments at the end of
each of the next three years at 10 percnt interest. The present value of each payment is

PV,100
1.1()
PV2 100

PV3 =tooj-j
and the sum of these would be

PV= 100 — + 100 (1.10)2 + 100 (1.10)

1.10
or

PV = 1O() li—
.10+ (1.10)2 + (110)3]
Substituting the appropriate present-value interest factors fromTable 1A.1 and multi-
plying yields, the present value of this annuity is
PV = 100(0.9091 + 0.8264 + 0.7513) = 100(2.4868) = 248.68
Although this approach works, It clearly would be cumbersome for annuities of more
than a few periods. For example, consider using this method to find the present value of
a monthly payment for forty years if the monthly interest rate is 1 percent That would
require evaluating the present value of each of 480 amounts!
In general, the formula for the present value of an annuity of A dollars per period
for n periods and a discount rate of i is

This can be written as



CHAPTER 1 Introduction to Managerial Economics 27

PV = (IA-2)

Equatn (1A-2) is the general equation for the present value of an annuity. Recall that
means the sum of n separate components, the first where t = 1, the second where t = 2, and
so on, to I i.
The term
n 1

is called the present-value annuity factor (PVAF . ). It is the present value of an annuity
of $1 per period for n periods at a discount rate of i percent. Table H at the end of the
book provides these factors for a variety of interest rates and periods. Part of that table
is reproduced as Table 1A.2 here.

1
PVAF
I
+ ij
Interest Rate
Periods 1% 2% 3%
12 11.2551 10.5753 9.9540
24 21.2434 18,9139 16.9355
30 25.8077 22.3965 19.6004

Consider the present value of an annuity of $3,522 per month for thirty months with
an interest rate of 1 percent per month. Note that this problem differs from the first two
examples because it considers monthly (not annual) payments and a monthly discount
rate. This should not he confusing. The general problem refers to n periods and a dis-
count rate of 1 percent per period. As long as the length of period (i.e., month, year, etc.)
and the interest rate for that period correspond, the approach is straightforward. For
example, if the periods are years, the interest rate must be a yearly rate.
The equation for the present value'of this monthly annuity is

PV = 3,522
(1-01
The factor
30

is the present value of an annuity of $1. From Table 1A.2, that value is 25.8077. Substi-
tuting that value and multiplying gives PV = $3.522(25.8077) $90,895. This means
 28 PART I Getting Started

that an amount of $90,895 invested at an interest rate of 1 percent per month would be
just adequate to make thirty monthly payments of $3,522.
A similar problem might be stated in the following way. What is the present value of
a series of 30 monthly payments of $150 if the interest rate is 24 percent per year? Note
that the payIients are monthly, but the interest rate is stated as an annual rate. By divid-
ing the annual interest rate by 12, the appropriate monthly rate is found (i.e., 24/12 21).
Thus the solution would be

PV = 150[(-)1 = 150(22.3965) = $ 3,359

With the advent of financial calculators, there is l,ess need for actually using the
present-value equations and the tabular data, on presett-value interest and annuity
factors. However, it is essential that the principles underlying these financial calcula-
tions be understood.

Fninle Solomon Keith and the New York Lottery

3-

Solomon Keith, a 55-year-old bank custodian,.won the $5 million New York lottery in
1987, but unfortunately died man auto accident 15 months later after collecting only
the first two installments of $240,245 each. As the Wall Street Journal put it:" . . his num-
ber came up not once but twice." His estate collected three more payments, leaving 16
annual payments to be made.*
Primarily because of income taxes, estate taxes, and the administrative costs asso-
ciated with the estate, the remaining payments were put up for bid. The Prudential Life
Insurance Company made the winning bid of $2,075,000 (of which $700000 was used
to pay estate taxes, $550,000 went to attorneys and the administrator of the estate, and
the remaining $825,000 was divided among Keith's nine relatives).
What discount rate did Prudential use to come up with their winning bid?
Solution The discount rate is found by using the present value of an annuity formula

PV=j—

and making appropriate substitutions, that is,

2,075,000 = 240,245 16 111

solving for the present-value annuity factor
2,075,000
(PVAF,) = ---------- = 8.6370
240,2 45
and then using Table II (or a financial calculator) to determine that i = 0.0839 (i.e., a
discount rate of 8.39 percent was used.)

*See Wall Street Jour,wl. June. 30 1992 and the Salt Lake Tribune. June 3, 1992.
 CHAPTER 1 Introduction to Managerial Economics 29

AR - -
Pt
en the interest rate per period {the nu1 b-r-.-6f periods (it), and the amount
(S)or the annuity paYment ),theie are two basic, present-value problems
I resent Value oa'i 'n'OmL PTV
41 1J
is the presentis viue of an amount S to be received (or paid) in n periods if
the Interest rate t iercerit ner oenod The term f((1i4- ,FL ti
Interest factor (PV F,,)
Z PrLsent Value fan Aniway PV A ---LJ
1( 1 --•4t I_
were PV i the preseit talue onannuity of A perpetiod pdld at theeiid
of eac,h
fn nods if the nrest ritc is i percent per periotL'fle term
Ei 1(1 *
t1esent varue awwty tactor PVAF

Problams

IA-1. Robert Ryan, general manager of the Chicago Stars professional football team,
is currently negotiating a new contract with Ronnie Smith, the team's star run-
ning back. Under league rules, Smith is now a free agent, which means that he
is free to negotiate a contract with any other team in the league. Smith has pre-
sented Ryan with a final contract demand consisting of alternatives for a five-
year contract. If Ryan does not agree to one of these, Smith will sign with an-
other teain.The alternative contract demands are
a. A $2,000,000 bonus payment immediately, a payment of $500,000 at the end
of each of the next five years, and a deferred payment of $1,000,000 at the
end of the fifth year of the contract.
b. A $500,000 bonus payment now, payments of $300,000 at the end of each of
the next five years, and deferred payments of $200,000 each payable at the
end of years ii through 20
Ryan has determined that Smith's value to the team over the next five years is
about $3,000,000 (in terms of the present value of additional revenue from gate
receipts and television discounted at 12 percent per year). Should Ryan accept
one of Smith's contract demands, and if so, which one? Explain fully.
I A-2. A rich uncle gives you the choice of one of the following legacies
a. $15,000 each year for the next 12 years.
b. $13,000 each year for the next 18 years.
c. $11,000 each year for the next 12 years plus a lump-sum payment of $81,000
at the end of the 18th year.
30 PART I Gelling Started
Which would you take and why? Assume that the appropriate discount rate is
10 percent and all amounts would be received at the end of the year.
1A-3. Mity-Lite, Inc., a manufacturer of plastic tables for institutional use, is consid-
ering a capital spending program involving annual expenditures of $100,000 for
each of the next five years. The firm estimates that its annual profit of $100,000
would increase by 50 percent when the capital program was completed. As-
suming the firm has a 20-year life and the appropriate interest rate is 12 per-
cent, should the capital spending program be implemented?
1A-4. The Smith Construction Company borrows the entire cost of anew dump truck.
The luau has an annual interest rate of 12 percent and calls for monthly pay-
ments of $1,000 over a five-year period. What is the cost of the truck?
1A-5. A couple borrows $10,000 to buy a car.The loan agreement specifies that monthly
payments are to be made for four years. The annual interest rate is 12 percent.
Determine the monthly payment.
1A-6. A firm develops a new product that will add $50,00() to profit each year for five
years. If the discount rate is 10 percent per year, how mtch will this new prod-
uct add to shareholder value?
IA-7. It is estimated that the annual after-tax profits of the Micrwave Corporation
will he $500,000 per year for each of the next 30 years. Gvef a discount rate of
14 percent, what is the value of this firm?
IA-8. A loan agreement specifies that payments of $133.33 are to he made each
month for five years. The annual interest rate specified is 6 percenL What is the
amount of the loan?
1A-9. An executive's three-year employment contract calls for an annual salary of
$600,000 (paid in monthly installments) for the first year with a 10 percent in-
crease in the salary each year. What is the present value of this contract? The
appropriate discount rate is 8 percent per year.
lA-tO. The HAL Computer Corporation is considering an increase in its annual ad-
vertising expenditures from $10 milliOn to $15 million for a five-year period
(i.e., in years ito 5). The marketing department estimates that the increased ad-
vertising will increase profits by $4 million in years 3 to 7 and by $3 million in
years 8 to 10, after which profits will return to the level they were at prior to the
new program. If the firm uses a discount rate of 12 percent, will the proposed
advertising program increase shareholder value?
lA-il. Lightco Inc., has just replaced its production machinery with new, more effi-
cient equipment that will reduce annual production costs by $5 million per year.
The cost of the new machines is $28.8 million, If the firm uses a discount rate of
8 percent per year, how many years will it take for the firm to recoup its in-
vestment?
1A-12. A state department of transportation is considering replacement of a bridge at
a cost of $20 million. The life of the new bridge is 30 years, and it is estimated
that improved safety and reduced congestion would be valued by bridge users
at $3 million per year. The department uses an interest rate of 10 percent to
evaluate capital projects.
a. Should this proposal be implemented?
K. If not, what is the maximum cost of the new bridge that could he justified by
the bcncfits to users?.
 CHAPTER 1 Introduction to Managerial Economics
31
1A-13. On Monday, Smith borrows $100,000 at 8 percent annual interest with repay-
mnt to be made in monthly installments for 30 years. Overnight, the market in-
terest rate increases to 12 percent. The lender, anxious to get its money back
' from Smith so that it can be loaned out at the higher interest rate, makes Smith
the following offer: If she will double her monthly payment (thus significantly
reducing the term of the loan), the lender will reduce the interest rate on
Smith's loan to 6 percent per year Should Smith take this offer? Explain. (Hint:
Smith's opportunity cost is 12 percent per year. That is, she can invest at the
same rate that the bank can lend.)
1A-14. A corporate bond specifies that interest of $400 is to be paid every six months
for 20 years, and then the $1,000 principal or face value of the bond also will be
paid. The market interest rate for bonds of this quality is 10 percent per year.
What is the market value of this bond?
IA-15. A pedestrian is seriously injured when hit by an automobile and files a lawsuit
against the driver.The drivers insurance company offers the victim a Structured
settlement, which the company announces to be worth $2,000,000. The offer
consists of the following components:
1. $1,000 per month for 360 months;
ii.
one-time payments of $100,000, $200,000. and 5300,0(K) at the end of 35,
40, and 45 years, respectively; and
UL a final payment of $1,000,000 at the end of the 50th year.
a. Given that the market interest rate is 12 percent per year, evaluate this offer.
b. Develop a structured settlement that has a true present value of $2,000,000.
 CHAPTER

Basi

Preview
• Functional Relationships: Total, Average, and Marginal
M Economic Models
• Probability and Probability Distributions
Probability
Probability Distributions
Statistics of a Probability Distribution
• Summary
Discussion Questions
• Problems
• Appendix: Calculus and Managerial Economics
The Derivative of a Function
Higher-Order Derivatives
Calculus and Optimization
The Partial Derivative
Optimization and Mu1tivariae Functions
Problems

32


• CHAPTER 2 Basic Training 33

PREVIEW

Just as the skilled craftsman needs tools to build a new home, the manager needs tools
thsist in making decisions that will ultimately maximize the value of the firm. In-
creasingly, these tools are of the quantitative sort. Algebra, statistics linear program-
ming, and, to some extent, calculus are used in all the functional areas of business today.
This is especially true for decision makers in production and finance, but even those in
marketing and business law increasingly find these tools of value.
Many management decisions fall into the categor y
mization refers of optimization problems. Opti-
to finding the best way to allocate itsources given an objective ñinction.
For example, a production manager may seek to maximize production for given inputs
of capital and labor or to minimize the cost of producing a specified rate of output. In
Contrast, the president of the firm, who has a broader perspective, may want to organize
the firm's labor arid
capital resources in order to maximize profit and shareholder value.
Terms such as minimize and tnaIim ize
imply an Optimization problem. Sometimes man-
agers seek optimal decisions by trial and error or other informal methods. In many sit-
uations, however, data on the firm and its market can be combined with quantitative an-
alytical tools to scientifically determine the optimal management strategy.
The purpose of this chapter is to introduce the basic quantitative tools that arc com-
monly used in managerial decision making and to demonstrate their applications. In
the first section, functional relationships are discussed, and a special set of functions—
total, average, and marginal,_js developed and applied to problems in managerial eco-
nomics. Next, the nature and structure of an economic model is discussed. Typically,
such models include several equations and conditions that constrain the nature of these
functions, in the last section, basic principles of probability are developed, with empha-
sis on measures of central tendency and dispersion for probability distributions.
In addition, a review of the elements of calculus and their application to econom-
ics is included as an appendix. While all of the important concepts in managerial eco-
nomics can be understood without using calculus, its use helps to understand important
relationships and allows certain problems to be solved.

FUNCTIONAL RELATIONSHIPS: TOTAL, AVERAGE,

AND MARGINAL
In ma thematics,an equation of the form

yf(x) (24)
is read "y is a function of x."Thjs mens that the value of y depends on the value of x in
a systematic way and that there is a uni q ue value of y for each value of
X. Usually the
variable on the left-hand side of the equation. y in this case, is called the
dependent vari-
able. The variable on the right-hand side is called the independent variable,
Of course, y may depend on two or more independent variables, such as
Y - g(x,z) (2-2)
or

y–h(x1,x2 ... . . x) (2-3)



34 PART I Getting Started

The letters f g, and h have no specific meaning other than to indicate that a functional
relationship exists between y and the independent variable or variables. Equation (2-1)
is referred to as a univariate function because it has only one independent variable.
Equations (2-2) and (2-3) are multivariate functions.
In many economic models, a special set of functional relationships called total, av-
erage, and marginal functions is used. Such functions are involved in the theory of de-
mand, cost, production, and market structure. A basic command of these concepts is es-
sential to understanding the principles of managerial economics.
The following production example will help in understanding these relationships.
Suppose that there is a small building containing four machines and a stock of raw ma-
terials ready to be processed.Ten equally skilled and diligent workers are lined up out-
side ready to go to work in this factory. If there are no workers, output will be zero. As
workers are added, output increases. The total amount of output associated with a par-
ticular input of labor working with those four machines is called the total product of
example total product of one worker might be two units of output. As the
labor input changes, so does total output. An example of a total product schedule is
shown in the first two columns of Table 2.1.

L p I•4 .aI, 'tr2j. .nfl

Product f Labor SJiu3e'
Number of iWuI Average .%1ur,,i,u,I
Workers Product Product Product
(U) (Q) (AP) (MP)
U V —
1 2 2,0 2
2 5 2.5 3
3 9 3.0 4
4 14 3.5 5
5 22 4.4 8
6 40 6.7 18
7 57 8.1 17
8 63 7.9 6
9 64 7.1 1
10 63 6.3 –1

As just indicated, one person working alone in this factory produces two units of
output. Adding a second person and organizing the production system so that the work-
ers complement each other results in total product increasing to five units. The first
worker is associated with a two-unit increase in output; having two workers instead of
one will increase output by three units; three workers will increase output by four units;
and so on. Thc change in output associated with a one-unit change in workers is called
the marginal product of labor. Using the Greek capital letter delta (is) to indicate a
change, the marginal product function (MP) can be defined as
 CHAPTER 2 Basic Training 35

MP-2
AL (2-4)
wbtre Q represents output and L represents the input of labor.'
Note in Table 2.1 that for
the first six workers, the marginal product increases as
the rate of labor input increases. However, marginal product declines thereafter. That
is, the sixth worker adds 18 units to output, but the seventh adds only 17 units and the
eighth worker only 6. Finally, having 10 workers actually causes output to decline; the
marginal product associated with the tenth worker is negative. The point has been
reached where there are too many workers in this plant Perhaps they are getting in
each other's way; in any event, the presence of 10 workers has thwarted the achieve-
ment of efficient production.2
The average product of laborfrancrjon (AP)
labor used. Average product is found b y measures the average Output per Unit of
dividing total product by labor input.That is,

(2-5)
The total product function is plotted in Figure 2.1a, and the average and marginal
Product functions are shown in Figure 2.1b. There
There are important r elationships among
the three functions that are true fox all total, average, and marginal functions. First,
the value of the average function at any point along that curve is equal to the slope of
a ray drawn from the origin to the total function at the C
orresponding point. For ex-
ample, from Table 2.1 it is kiown that the average product of six workers is 6.7. Thus,
the slope of a line (QA) drawn from the origin to point
A on the total product func-
tion has a slope of 6.7. Similarly, the average product of seven workers is 8.1: this is
equal to the slope of a line (OB) drawn from the origin to point
uct function. B on the total prod-

'The marginal product function is the slope of the total product function. Slope; can be deened as the
change in the dependent variable divided by the change in the independent variable. Consider the function
Y = 10 + 3x. A one-unit change in z say from x = 5 to x =
from y = 25 toy = 28. Thus th e change in y (i.e., 6, is associated with a three-unit change in y,
= 3/1 3. y) is 3, and the change in x (i.e., &r) is 1,so the slope is
2 That marginal product must ultimately decline as
w orkers are added is clear in this example- The building
has a finite amount of floor space. The change in output should be positive if one or two workers are added.
But for some number of workers, outpt will be zero as workers f
ight for a place to stand rather than
produce output.Therefore beyond some rate of labor input, marginal product must decline.
3
A11 three of the functions are drawn with graphs having the same units on both axes (i.e., output on the
vertical axis and number of workers on the horizontal axis). Therefore, all three could have been included
in the same graph. However, the range of total product (0 to 64) is so much greater than the range of
average and marginal product that it is more descriptive to use two graphs. Note that thc height of each
interval on the vertical axis in Figure 2.1b is considerably greater than in Figure 2.la. This allows the
relationships among all three functions to be seen more clearly than if the
y were all drawn in the same
graph. Note also that the horizontal axis in both graphs is exactly the same. This allows comparison of
points on one graph that correspond to points on the other.


Total product (Q)

70
roduct (Q)

60

50

40

30

20

10
Rate of
labor impul
- 10
0
(a) Total product function

Average and
marginal product (Q)

20

15

10

age product
5

II I I I Rate of
5 10 labor input
(L)
(b) Average and marginal product functions

MIA

36
 CHAPTER 2 Basic Training 37
Another key relationship is that the value of the marginal function is equal to the
slope of the line drawn tangent to the total function at a corresponding point. For ex-
an$, the slope of the dashed line CD, which is drawn tangent to the total function at
E, is about 4. This means that the marginal product corresponding to this point (i.e., be-
tween 2 and 3 units of labor) is 4.
Point Eon the total function is called an inflection point. To the left of point F the
total function is increasing at an increasing rate; to the right of F, the total function is in
creasing but at a decreasing rate. Note that this inflection point, corresponding to about
six workers.occurs at the point where the marginal product function is at a maximur
(i.e., point H in Figure 2.1b).
If the marginal and average functions intersect, that point of intersection will be at
the minimum or maximum point on the average function. In Figure 2,lb j the intersec-
tion ccurs at point G, which is the maximum point of the average product function Jhe
logic of this relationship is quite straightforward. Suppose that the average product for
two workers is 2.5 and the marginal pioduct of three workers is more than that average,
say, 4. Thus the average product for three workers must increase to 3. If the additional
output associated with hiring another worker is above the previous average product,
the average product must be increasing.
Conversely; suppose that eight workers average 79 units of output per period and
that the marginal product of nine workers is only 1. This will cause average product to
fall to 7.1. If an additional worker adds less to total output than the average product
prior to that addition, average product must fall.
This logic leads to the following conclusion. For any set of average and marginal
functions, if the marginal function is greater than the average function, the average must
be rising. If marginal is below average, the average must be failing. This implies that the
intersection of the two functions must occur where the average function is at a maxi-
mum or minimum. As will be shown in later chapters, this result is important in some
managerial economics problems.
If the marginal function is positive, the total function must be rising. Note that it
does not matter whether the marginal function is increasing or decreasing, as long as it
is positive. Conversely, if the marginal function is negative, the total function must be
declining. Again, it does not matter if marginal is rising or falling. If marginal is nega-
tive, the total function will be declining. For the data in Table 2.1 and Figure 2.1, the mar-
ginal product is positive for the first nine workers. It declines after the sixth worker but
remains positive through the ninth. Note that total product increases until the tenth
worker is added.The marginal product of the tenth worker is —1, and this negative mar-
ginal product is associated with a decline in total product.
Because the total function increases as long as the marginal function is positive and
decreases when marginal is negative, it follows that total product is at a maximum when
the marginal function is zero. In Figure 2.1 the maximum of the total product function oc-
curs at nine workers. This point corresponds to the point where the marginal function in -
tersects the horizontal axis, that is, where marginal changes from being positive to negative.
An understanding of total, average, and marginal relations is an important founda-
tion for the effective study of managerial efii tics. erms such as average cost. mar-
ginal product, and total revenue are integral parts of the manager's vocabulary, and the
associated principles are some of the n ?thost powerful tools.
 38 PART I Getting Started

Case Study
Decision Making in the Public Sector:
Marginal Analysis and Automobile Safety

Suppose a specific automobile model has so many engineering defects that the risk of
having an accident is greatly increased. Although the car is no longer being manufac-
tured, vehicles already on the road arc expected to cause 10 deaths during the next year.
For simplicity, assume that the engineering defects are 'so severe that none of the cars
will be running after one year. The number of deaths can be reduced if the manufac-
turer recalls the cars and corrects some or all of the defects. Clearly, the ntorc defects
that are corrected, the greater will be the reduction in the number of deaths.
As shown in the table, the marginal cost of repairing and modifying these cars in
order to reduce the number of deaths from 10 to nine is $200.000, but marginal cost rises
as the number of deaths decreases. For example, the marginal cost of going from one
death to no deaths is $2,000,000. This cost rises because some of the defects are easily
corrected, whereas others require a substantial modification of the vehicle.
Suppose that the National Highway Traffic Safety Administration, a federal
agency responsible for regulating automobile safety, uses marginal principles to make
recall decisions and that the agency has determined that each life saved is worth
$800,000 to society. That is, $800,000 is the marginal benefit associated with saving a life.
The benefits and costs of the recall/repair program shown in the table can be com-
pared to make a decision. The marginal cost of saving the first life is $200,000 and the
marginal benefit is $800,000. Thus it is clear that the manufacturer should be required
to take some action. However, it is çqually clear that it would be inefficient to attempt
to eliminate all defects. For example, the marginal cost of going from one death to zero
deaths is $2,000,000 but the marginal benefit is only $800,000.

Marginal Cost of Marginal Benefit Total

Deaths Death Prevention of a Life Saved Net Benefits
0
9 200,000 $800,000 $ 600,000
8 400,000 800,000 1,000,000
7 600,000 800,000 1,200,000
6 800,000 800,000 1,200,000
5 1,000,000 800,000 1,000,000
4 -. 1,200,000 800,000 600,000
3 1,400,000 800,000 0
2 1,600,000 800,000 —800,000
1,800,000 800,000 —1,800,000
0 - 2,000,000 800.000 —3.000,000

The recall program should be designed to reduce those defects (and associated
deaths) to the point where marginal benefits equal marginal costs, In this example, that
 CHAPTER 2 Basic Training 39

equality occurs at six deaths. Note that total net benefits (a net benefit is defined as nir-
ginal benefit minus marginal cost) are maximized ($1,200,000) at six deaths. Total ben-
eliare $3,200,000 (i.e., four lives saved multiplied b y $800,000 per life) and total costs
are2,000.000. To require additional repairs to reduce the number of deaths below six
is not socially efficient because the costs exceed the benefits.

Key Concepts
• A functional relationship of the form
Y 7 ft . .k... L'.......

means th:.ij there is a systematic relationship between the dependent variable

and the independent varihles and that there is a uniquL value of
X ) , ...........,,

e for an'. of the ndependent variables.

• t-eir
ny totat functioa 'e.c.. total produ'1 , a. re'enue. ttc. there is an :,n'i-

ated marginal (unction and average. function.

Theleyrelat ionsbips among the total, aver: and marginal functions are
1. The value of the average (unction at any poirn is the slope 6f ,a ra y drawn
from the oniam to the total [unction at that p int.
2. The value of the marginal function at an y pont is the slope of aline drawn
1antent to the total function at that point.
3. 'ftc marginal luncion will intersect the average functIon at either, a mini-
mum or a maximum point of the average function.
If the marginal function is positive, the total function will he increasing, if
the marginal function is negative, the total function will he decreasing.
5. The total function reaches a maximum or minimum when the marginal func-
tion equals zero. :

ECONOMIC MODELS
In the aerospace industry, small model airplanes are flown in wind tunnels to test the
flig ht characteristics of full-sized planes having the same characteristics. In economics,
graphs and/or equations are used to explain economic relationships and phenomena
and to predict the effects of changes in such economic parameters as prices, wage rates.
and the price of capital. Although such models are abstractions from reality and may
seem unrealistic, they are useful in studying the way an economic s ystem works. Just as
it is not sound practice to build a new jet aircraft before a model is used to test its flight
characteristics, an economic decision should not be made without having first analyzed
its implications by using an economic model.
40 PART I Getting Started

Price per unit

($) S

LI
D Quantity per unit of
time (Q)

An economic model usually consists of several related functions, some restrictions

on one or more of the coefficients of these functions, and equilibrium conditions. Re-
call the concepts of supply and demand from introductory economics. As shown in Fig-
ure 2.2, the demand curve (DD) slopes downward from left to right and shows the quan-
tity of output that consumers are willing and able to buy at each price. The negative
slope implies that a larger quantity is demanded at lower prices than at higher prices.
The supply curve (SS) shows the amount that firms will produce and offer for sale at
each price. This curve has a-positive slope because firms will supply a larger quantity at
higher prices than at lower prices.4
Equilibrium in a market exists-when the cpiaptity demanded equals the gptity
ppiI&This is shown graphically as the intersection of the demand and supply func-
tions in Figure 2.2. In this example, P, and Q e.are the equilibrium price and quantity, re-
spectively. In equilibrium there is no incentive for buyers or sellers to change price or
the quantity. At point {P, Qe), buyers' demands are met exactly and suppliers are sell-
ing exactly the number of units they desire to sell at that price.
The preceding example is an economic model depicted graphically. A simple alge-
braic model can be used to describe exactly the same economic phenomenon. The quan-
tity demanded (Qd) and the quantity supplied (Q) are both functions of price. That is,
Qd = f(P) and Q3 = g(P) -
Suppose that the demand function is
Qd = a + bPwhere b <0 (2-6)
and the supply function is
Q5 = c + dP where d > 0 (2-7)

4Tlie principles of demand and supply arc discussed in greater detail in chapters 3 and 9.
 CHAPTER 2 Basic Training 41
The restrictions on the parameters (i.e.. b <0 and d > 0) simply mean that the demand
curve must slope downward (i.e.. have a negative slope) and that the supply curve must
slope upward (i.e., have a positive slope). By adding an equilibrium condition that the
quantity supplied equals the quantity demanded, that is,
Qd=Q (2-8)
the economic model, consisting of equations (2-6), (2-7). and (2-8), is complete.
By equating supply and demand, the equilibrium price (Pe) can be determined.
Substifuling equations (2-6) and (p7) fd? Q
d and Q--,
-in (2-8) yields
a + bP c + dP
Solving for P the equilibrium price is

4M_qq^kn^
tiis found bN substituting the equilibrium price into
either the
supply or demand function and solving for
quantity. Using the demand function, the
rium

Q,. = a +
If the values of the parameters a, b, c, and d are known, the actual values of
Pe and Q
can easily be calculated. In chapter 4, statistical methods used to estimate the numeri-
cal values of these parameters are reviewed.
Models of this type are used extensively in the study of economics. Indeed, one of
the strengths of the discipline is that important and powerful results can be derived
from models that are quite simple. To be sure, some economic models in advanced
books are very complex. However, a thorough grasp of many of the key principles of
economics can be obtained by applying rather basic concepts and models to business
and social problems. Clearly, an important reason for studying managerial economics is
to develop a set of economic models that can be used to analyze the many resource al-
location problems faced by managers.

Exampl,/ Determining Equilibrium Price and Quantity

Suppose parameters of the demand and supply equations have been estimated and that
the equations are

Qd = 14— 2P
and

2 + 4/'
Determine the equilibrium price and quantity.
Solution Note that the slopes of these functions meet the constraints specified in
the chapter. That is, the demand function slopes downward and the supply curve
slones upward. Setting the demand function equal to the supply function
 42 PART I Getting Started

14— 2P = 2 + 4P
and solving for the price yields
P=2
Substituting P = 2 into the demand function yields the equilibrium quantity
Qe = 14 - 2(2) = 10
Thus the Jprice, quantity} combination {P = 2, Qe = 101 results in equilibrium in this
market This combination correspors to (Pc' QJ in Figure 2.2.

Key Concepts
• An economic model typical!) consists of several functional relationships, con-
ditions, or constraints on one or all of these functions, and one or more equilib-
rium conditions.
• Generally, econolntc models are used to demonstrate an economic principlc to
explain an economic phenomenon, or to predict the economic implications of
some change affecting one or more of the functional relationships.

PROBABILITY AND PROBABILITY DISTRIBUTIONS

Managers are often faced with making decisions that have the potential for a variety of
outcomes. An outcome is a possible result of some action. For example, flipping a coin
will result in one of two outcomes—heads or tails. A management decision to introduce
a new product could result in a range of outcomes varying from wide consumer accep-
tance to no interest whatsoever. The quality of any decision will be enhanced by iden-
tifying the possible outcomes of the decision and then estimating the relative chances
of eachoccurrirLg. This listing of oufcomes and the chance of each occurring is a proba-
bility distribution that can be evaluated using quantitative techniques. Often the chance
of an outcome occurring will depend on the state of nature that prevails. The term state
Of nature refers to conditions in the business environment that will influence the out-
come of a decision but that are not subject to control by the decisiOn maker. For exam-
ple, the condition of the general economy is a state of nature that will affect the out-
comes of most business decisions.
Probability and probability distributions are an important part of the manager's tool
kit. In this section, methods are developed to setup the probability distribution and then
evaluate measures of central tendency and dispersion for that distribution. In chapter 14,
these prthciplesjre extended to analyze the concept of risk in decision making.

Probability
The probability of an event is the relative frequency of its occurrence in a large number
of repeated trials. For example, in repeated tossing of a coin, a hcad will appear about
one-half the time. That is, the relative frequency or probability of a head occurring is
0.50. In rolling a die, there are six possible outcomes, 1.2, 3. 4. 5.6.The relative frequency
of any one of these outcomes is 1/6, implying a probability of 0.167 for each outcome.
 CHAPTER 2 Basic Training 43
The probability of some events is known or can be computed with certainty. For
example, the probabilities of most outcomes associated with rolling dice, drawing
playing cards from a deck, and tossing coins are easily determined using standard
principles of probability. Other probabilities are not easily determined mathemati-
cally but can be determined by repeating the processmany times and observing how
many times particular outcomes occur. For instance, there may be no mathematical
way to estimate the probability of winning a particular game. But by playing the game
many times and noting the number of times it is won, the probability of winning can
be determined.
In other cases there is no accurate way to estimate probabilities except by using
judgment. The weather forecaster's statement that "the probability of measurable pre-
cipitation tomorrow is 0.40" is based on an evaluation of how prevailing breezes,radar
maps, and upper-air charts are associated with given weather conditions; Assessing the
probability of a recession next year or a significant change in consumer preferences falls
into the same category. Based on an analysis of current economic conditions, surve ys of
business capital-spending plans, and other information, a judgment can be made as to
that probability. This judgment is necessarily suhective and rry differ gfiuitly
among analysts, but predictions of this type are made daily by business manaeers, econ-
OTfiiSts, and other decision makers.

Probability Distributions
For a decision or an experiment of some type with several possible outcomes, the prob-
ability of the ith outcome occurring is indicated by P1. where
0P1 s1 i=1,2. . .,n (2-9)
That is. the probability must take on a value in the range 0 to 1. Negative probabili-
ties and those in excess of unit y have no meaning. For example, if an event had
a probability of 1.25, that would mean that it wouldoccur more than 100 percent of
the time!
Furthermore, the sum of the probabilities of all possible outcomes or events of an
experiment must equal 1. That is,

P, = 1 (2-10)
This means that when the experiment is conducted (i.e., the dice rolled, the coin tossed,
or the investment made), one of the outcomes must occur. It then follows that the prob-
ability that an event will not occur is I. minus the probability that it will occur.-For ex-
ample, if the probability of a 6 occurring when rolling a die is 116, this implies that the
probability of not rolling a six is 1 - 1/6 5/6This corollary is useful because it is some-
times easier to find the probability of an event not occurring than the probability that
the event will occur.
A listing of each outcome of an experiment and its probabilit y defines a probabil-
ity distribution. For example, the probabilities of tossing 0, 1,2. or 3 heads in three tosses
of a coin are shown in Table 2.2. In this example, X, is the number of heads observed in
each experiment of three tosses. Note that each probability is between 0 and 1 and that
the, sum of the probabilities is equal to 1
44 PART I Getting Started

111T
Number of Heads (X,) Probability (P1)
Q 0.125
1 0.375
2 0.375
3 0.125

Statisiics of a Probability Distribution

For any probability distribution, there is a set of statisticsor measures that describes or
provides summary information about the distribution.Thenost important of these are
measures of central tendenc y and dispersion.These have many applications in business,
science, and engineering. In managerial economics they can be used to evaluate and
compare the returns associated with alternative strategies and thus help to make sound
managerial decisions.
Expected Value: The first of these statistics is the expected value or mean of a pob-
ability distribution. EffectIvely, it is a weighted average of the outcomes using the preb
abilities of those outcomes as weights. The expected value is a measure of central ten-
dency because in repeated trials of most experiments, the values of the outcomes tend
to be concentrated around this statistic. For example, the expected value of the distrib-
ution for rolling two dice is 7. The range of outcomes is 2 through 12, but the outcomes
6 through 8 occur much more often than the outcomes 2, 3, 11, or 12. This is because
there are more combinations of the faces of the dice that yield 6, 7, or 8 than combina-
tions that yield 2,3, 11, or 12.There is only one combination (1,1) that yields 2, but there
are six combinations (1, 6), (2,5), (3,4), (4,3), (5,2), and (6, 1) that yield 7.
The expected value (ii.) of any probability distribution is computed by multiplying
each outcome by its respective probability and then summing the products. Thus

(2-11)

Recall the probability distribution for the number of heads observed in three tosses of
a coin.
Heads Probability
0 0.125
1 0.375
- 2 0.375
3 0.125

- The expected value of this distribution is computed to be

= 0.125(0) ± 0.375(1)± 0.375(2) + 0.i25(3)— 1.50
This means that if this experiment were repeated many times and the number of
heads observed for each trial recorded, thç average number of heads would be 1.50. Ob-
viously, it is not possible to toss 1.5 heads n any trial. The expected value is not a par-
 CHAPTER 2 Basic Training 45

ticular outcome but is merely an average of the outcomes for a large number of repeti-
tions of the experiment.
- Understanding this principle and applying it when making managerial plans and de-
cisions can be very profitable. The financial success of gambling casinos in Las Vegas and
Atlantic City is evidence of the truth of this statement. The managers of these firms use
the principles of probability to structure the gambling games played in their casinos. The
payoffs to the players are always set so that the house wins more than an equal share of
the amounts wagered.Theretum on a gambling game is often referred to as the expected
payoff It is computed by multiplying the prbability of winning by Uie amount to be won
and adding the product of the probability of losing and the amount to be lost. Consider
a slot machine that costs $1 per play. The machine is programmed so that on average it
returns $8 once in every 10 plays for a net payoff of $7 (i.e., the $8 gross return less the
$1 cost of playing the game). The probability distribution is as follows:
State of
Nature Outcome Probability
Win $7
Lnc' —1 0.90

The expected payoff to the player is $-0.20, that is,

= 0.110(7) ± 0.9(-1) = —0.20
A game is said to be fair if the expected return or payoff is zero. Clearly, the pre-
vious example is not a fair game; the player expects to lose $0.20 (equivalently, the
casino expects to win $0.20) every time the slot machine is played. Gambling games in
casinos are not fair in a statistical sense because the expected values are negative for
the player.
Standard Deviation: The second statistic of interest measures the dispersion of pos-
sible outcomes around the expected value. This measure is called the standard devia-
tion (a) and is given by the equation
C r = \/!P1 (X
-
(2-12)
As just indicated, the standard deviation measures the variability of outcomes around
the expected value of the distribution. If there is no variation in the outcomes, that is, if
all X = g., then a 0. As the amount of variation about the expected value increases,
the value of a also increases
The computation of or for the probability distribution for tossing three coins is as
follows:
a \/0.125(0 1.5) 2 + 0.375(1 - 1 . 5)2 + 0.375(2 1 .5)2 + 0.125(3_1.5)2
= 0.87
Because the standard deviation 'provides information about the dispersion of the
individual values or outcomes around the expected value, this measure of variation can
be of use in decision making. Suppose that a manager was considering two investment
alternatives, A and B. The probability distributions for each are shown in Figure 2.3.
Note that the expected value of profit (i.e., $100) is the same for both, but there is a
much grcatcr range of outcomes for B than A. This implies that tT B is greater than aA
46 PART I' Getting Started

Probability

Profit
$100
(L)

(a) Investment A

Probability

fit
(tL)
(b) Investment B

(i.e., B is a riskier investment than A).

Most decision makers will pick A, although there
may be others who would select B
because of differences in their preference for risk.
The nature of these preferences is discussed in more detail in chapter 14.
Coefficient of Varitjon:
The third statistic is the coefficient of variation, which relates
the variation of outcomes to the mean. It is defined as the ratio of the standard devia-
tion to the expected ahic

(213)


CHAPTER 2 Basic Training 47

This statistic provides a useful way to compare the variation to the expected value of a
probabilit y distribution. It measures variation per unit of expected value, The use of the
coefficient of variation can be illustrated b y considering two probability distributions,
C and D, having the following statistics:

P. c 100 ac50
D30
Both the mean and standard deviation for distribution C are greater than for distribu-
tion D, but the coefficient of variation is less than for C than for P. That
50 30
ii c = 0.5<u 1, = = 0.6
This means that distribution C has less dispersion or risk relative to its mean than does
distribution D. Thus, not only does the absolute amount of risk (o-) have to be estimated
and evaluated, but the level of that risk relative to the expected return on an investment
also should be considered.

Ky Cn p
• A probability distribution lists the possible outcomes of an experiment and the
probabiht asscciatcd with each outcome.
• 'The probabilit y of an y outcome must be in the range 0 P(X 1.0. and the
sum of the probabilities for all outcomes must equal-I.:,
• '1 he.,expected value of a probability distribution (p. P, X1 ) is a measure of
the average outcome from repeated...ials of an experiment
• The siandard deviati.ui [r .. ¶y - t)2 j no the dispersion of out-
conies around the expected value and is used to assess risk.
• 11e coefficient of variation (u oip..) is a measure of variation per unit of e x-
pected value.

SUMMARY
A functional relationship such as y = f(x 1 , x,,. . . , z,,) means that the value of the de-
pendent variable y depends on the values of the independent variables x , X 2,. .. , x,, in
a systematic way and that there is a unique value of y for any set of values for those in-
dependent variables. Total, average, and marginal relationships are fundamental to un-
derstanding and using economics. The marginal function is the slope of the total func-
tion. The intersection of the marginal and average functions occurs at either the
minimum or the maximum point of the average function.
An economic model generally consists of several related functions and certain con-
ditions related to these functions. The model can be used to explain economic phenom-
ena and to predict the effects of changes in one or more variables or functions. Ofter.,
 48 PART I Getting Started

simple models can be used to explain complex relationships and to derive powerful re-
sults. The effectiveness of managerial decision making can be enhanced significantly by
using economic models.
An outcome is a possible result o some action. A state of nature is a condition that
affects the outcome but cannot be controlled by the decision maker. The probability
of an event happening is its relative frequency of occurrence. The probability of an
outcome must be in the range 0 P (X i ) 10 and the sum of all probabilities must
equal one.
The combination of all possible outcomes and their probabilities comprises a prob-
ability distribution. For any such distribution there is a set of statistics. The expected
value or mean of a probability distribution measures central tendency.The standard de-
viation measures the dispersion of the outcomes, and thecoefficient of variation pro-
vides information on the amount of dispersion relative to the central tendenc y of the
distribution. The use of probability distributions and their associated statistics has wide
applicability in managerial decision making.

Discussion Questions

2-I. Explain the concept of an economic model. Why do economists and managers use
such models as part of the decision-making process?
2-2. Economists are not the only scientists to use models in their work. De-
scribe how other disciplines use models or similar abstractions from reality in
their work.
2-3. Explain the relationship among the total, average, and marginal functions. Intu-
itively explain why any intersection of the average and the marginal functions will
occur at a maximum or a minimum point on the average function.
2-4. The president of a major firm (who has had no training in managerial economics)
complains during a board meeting that quantitative techniques are of little value
because they are always subject to error and therefore should not be part of the
decision-making process. If the chairman asked for your opinion what would be
your response?
2-5. Explain how principles of probability are used to set the rules and payoffs for var-
ious gambling games in casinos.
2-6. Is the statement "y is a function of x" equivalent to sa ying that ' y is caused by x"?
In this context, critically evaluate the following statement: "The incidence of lung
cancer is significantly higher for heavy smokers than for nonsmokers. Thcreforc,
smoking causes cancer."
2-7. Explain how an insurance company could use its historical experience on deaths
and accidents to set its insurance rates.
2-8. The General Mills Company has several brands of breakfast cereal on the mar-
ket. Explain how you could use information on historic sales of these products to
develop a probability distribution for the sales of a new brand of cereal.
 CHAPTER 2 Basic Training 49

Problems
iven the following supply and demand equations
QD=lOO-5P
Qç=10+5P
where qand Qs represent the quantities demanded and supplied, respectively,
and P is the price.
)D etermine the equilibrium price and quantity.
/ J( If the government sets a minimum price of $10 per unit, how many units would
/ be supplied and how many would be demanded?, Flow c^o— u ld the government
maintain this minimum price?
the government sets a maximum price of $5 per unit, how many tin would
be supplied and how many would be demanded?
/ d. If demand increases to
Vo = 200 5P
determine the new equilibrium p ric e and quantity.
2-2 Given the following demand function
Q = 20 - 0.IOP
where P = price and = rate of output, complete the following table. (Note that
total revenue is equal to price times quantity)
Quantity Price Total Revenue Average Revenue Marginal Revenue

4
S

6
7

9
10

12

4'-Given the total cost function

TC = 150Q - 3Q 2 + 0.25Q3
complete the following table by computing the total, average, and marginal costs
associated with each quantity indicated.

\'
50 PART I Getting Started

Quantity Total Cost Average Cost Marginal Cost

2
3
4
5
6
7
8
0

10
11
12

2-4. Using the data developed in Problems 2-2 and 2-3,

a. Plot the total, average, and marginal functions for both revenue and cost.
[Note: Use a two-part graph similar to Figure 2.1 that places the total function
in the upper part and the average and marginal functions in the lower half.
Also, it is conventional to plot the marginal values at the midpoint between
the quantity values to which they relate. For example, the marginal revenue as-
sociated with going from two units of output to three units is $150. This value
should be plotted midway between the two- and three-unit marks on the quan-
tity (i.e., horizontal) axis.]
b. What is true of the marginal revenue function at that level of output where to-
tal revenue is at a maximum?
c. Determine total profit by subtracting total cost from total revenue at each rate
of output. Plot this function in the upper part of your diagram.
d. By comparing the marginal revenue and marginal cost curves and relating
them to the profit curve, can you think of a rule the firm might use to deter-
mine the output rate that would maximize profit?
S. Determine the average function for each of the following total functions:
' a) Total revenue = 100Q -
(j>Total cost 1,000 + 1OQ + 0.01Q2
/' c. Total profit = 50Q - 0.1Q2 - 1,000
"2 Given the following total revenue (TR) and total cost (TC)
equations, determine
k-' the output rate that would result in a breakeven (i.e., zero profit) situation for the
firm
TR5IQ—Q2
TC = 625 + Q
2-7. Do each of the following distributions meet the requirements for a probability
A
istribution? Why or why not?


CHAPTER 2 Basic Training 51

a. X1 P(X) b. Y, P0',) C. Z, P(Z,)

10 0.10 5 —0.20 —2 0.20

—20 0.20 10 0.40 —6 0.30
p —30 0.30 20 0.40 —8 0.40

—50 0.40 40 0.40 —10 0.20

2-8. Given the following probability distribution

P(X)
2 0.10
4 0.20
6 0.30
8 0.30
10 0.10

Compute the expected value, standard deviation, and the coefficient of variation.
2-9. A firm is contemplating building a new factory that will have three possible lev-
els of profit. depending on business conditions. The possible levels of profit and
the probability of each occuriing are
Profit Probability
—1,000 0.20
5,000 030
10,000 0.50

Determine the expected value, standard deviation, and the coefficient ofvaria-
tion for this probability distribution.
2-10. An advertising program has three possible outcomes: excellent success (i.e., 1,000
units sold); good success (i.e., 500 units sold); and little success (i.e., 100 units
sold). The probability of excellent success is 0.4 and the probability of little suc-
cess is 0.2. Find the mean, standard deviation, and coefficient of variation for this
probability distribution.

Problems Requiring Calculus

J4. A firm's total revenue (TR) and total cost (TC) functions are
TJ=110Q-5Q2
TC 10Q - Q 2 + 0.33Q3
=
Determine:
Equations for marginal revenue and average revenue.
t _V Equations for marginal cost and average cost.
,,,5/The output rate that maximizes total revenue:
&
52 PART I Getting Started

4/The output rate that minimizes average cost.

e. The output rate that maximizes profit.
2-12. Given the following function that relates total revenue (TR) to.-output (Q),
TR=20Q-2Q2
determine
.p{That rate of output that results in maximum total revenue.
JVThe marginal revenue function.
A1 'he rate of output for which marginal revenue is zero.
&lt there is any connection between your answers to parts (a) and (c)? Explain.
2-13. Given the total cost function
TC = IOOQ - Q2 + 0.3Q3
where Q = rate of output and TC = total cost, determine
a. The marginal and average cost functions.
b. The rate of output that results in minimum average cost.
2-14. Given the firm's demand function
Q = 55— 0.5P
(where P = price and Q = rate of output), and the total cost function
TC =20+ Q + 0.2Q2
where TC = total cost, determine
a. The total revenue function for the firm. (Hint: To find the total revenue func-
tion, solve the demand function for P and then multiply both sides of the equa-
tion by Q.)
b. The marginal revenue and, marginaJ cost functions and find the rate of output
for which marginal revenue equals marginal cost.
c. An equation for profit by subtracting the total coSt function from the total rev-
enue function. Find the level of output that maximizes total profit. Compare
your answer to that obtained in part (b). Is there any correspondence between
these answers?
2-15 The profit function for a firm selling two products is
i-50Q1 —Q +.1OOQ2-4Q
where Q1 and Q2 represent output rates of products 1 and 2, respectively. Deter-
mine the profit maximizing output rates for the two products and the profit as-
sociated with these output rates
 CHAPTER 2 Basic Training 53

Appendix

Calculus and Managerial Economics

Recall that many of the decisions facing manaers fall into the c.egory of optimization
problems. For example, decisions relating to maximizing profit or minimizing cost
clearly involve optimization. Often such problems can be solved graphically or by using
algebra. In other cases, however, the solution requires the use of calculus. More impor-
tantly, in many cases calculus can he u s ed to solve such problems more easily and with
greater insight into the economic principles underlying the solution.
Consider a firm whose total revenues from sales are given by the function
TR= 20Q - Q2
where Q represents the rate of output. Assume that the total cost of producing any rate
of output is give by the equation
TC = 50 + 4Q
Given the firm's ohjecti've of maximizing profits (i.e., total revenue minus total
cost), how much output should he produced? One approach would be to graph both the
total revenue and total cost functions. The vertical distance between the two functions
is profit. By identifying that point where this vertical distance is the largest, the profit-
maximizing output is found. This is shown in Figure 2A.1. which shows profit is maxi-
mized by producing eight units of output (Q 8).
An alternative approach is to develop a table showing revenue, cost, and profit at
each rate of output. Such data are provided in Table 2A.1. The tabular method has the
advantage of being somewhat more precise, That is, at an output rate of 8, total revenue.
is 96, total cost is 82, and profit is 14. From the graph in Figure 2A.1, these values can
only be approximated.
The problem is that neither method is very efficient. What if the profit-maximizing
output level is 8,000 or 8,000,000 instead of 8? The answer could be found using either
approach, but finding that answer could take considerable time. What if there had been
two outputs (Q1 and Q2) and three inputs (land, labor, and capital)? In this case, there
would be no practical way to determitie profit-maximizing output rates for Q1 and Q7
using these method.s.
A more powerful technique is needed so that the solution process can be both
precise and straightforward. Elementary calculus is easily adapted to optimization
problems in economics. Indeed, a few basic principles can be used in many different
kinds of problems. The profit-maximization problem just outlined could have been
solved quickly using the most elementar y calculus. In the following pages, some basic
principles of calculus are outlined and their application to economic problems is
demonstrated.
54 PART I Getting Started

Total revenue
Total cost
($)
120

100 Lal cost

nue
so

60

40

20

I I
Rate of
I I ,•. I. I I___..___
0 5 10
El MIMI alms

Profit
Q TR TC (TR—TC)..
0 0 .50 —50
1 19 54 —35
2 36 58 —22
3 51•.. 62 —11
4 (41. —2
5 75 70 5
6 84 74 10
7 91. 78 .13
8. %:. 82. 14
9 99 86 . .13
10 .109 90. 10
11 .99 .94 5
 CHAPTER 2 Basic Training aS
THE DERIVATIVE OF A FUNCTION
From, algebra recall that for the function y
€nge in y (denoted by A y) = f(x), the slope of that function is th
divided by a change in x (i.e., ix). The slope sometimes i
referred to as the rise
the run (the change in the variable measured on the vertical axis) ove
(the change in the variable measured on the horizontal axis). The slope isposi
tive if the curve slopes upward from left to right and
negative
downward from left to right. A horizontal line has a zero slope, if the function stope
to have an infinite slope. For a positive change in x (i.e., and a rtical line is said
Ax > 0), a positive slope implies
that X y is positive, and a ne gative.
slope impiie that y is negative.
The function y = 10 +
xis graphed in Figure 2A.2. To determine the average slope of
this function over the range x = I to x = 2, first find the corresponding y values. If x1
then Yi = 11, and if x 2 = 2, then = 1,
Y2 = 14. Then the slope is found by using the formula

slope =-Y = Y2- 7- Y1 14-11

=3
Lx x2 —x 1 2-1
In reality, this method determines the slope of a straight line through the points a
in Figure 2A.2. Thus, it Is only a rough a pproximation of the slope of the function and b
+ x,2 y = 10
which actually changes at every point on that function. By making the inter'a
smaller, a better estimate of the
slope is determined. For example, consider the slope over
the intervaix I to = 1.1. If x 1 = 1, then y1 = 11, and x2
Thus the slope is = 1.1 implies that y 2 11.21.

LIy 11.21 - 11 0.21

slope = == 2.1

10

10
56 PART I Getting Started

By using calculus, the exact slope at any point on the function can be determined.
The first derivative of a function (denoted as dy/dx) is simply the slope of a func-
tion when the interval along the horizontal axis is infinitesimally small. Technically, the
derivative is the limit of the ratio iXy/& as Ax approaches zero, that is,
dy
lim AY
dx Ax-4o Ax
The derivative of y = f(x) is sometimes written as f(x). The calculus term dy/dr is anal-
ogous to A y/x, but dy/dx is the precise slope at a point, whereasAy/Ax is the average
slope over an interval of the function. The derivative can be thought of as the slope of
h line drawn tangent to the function at that point. For example, the slope of the
a sti aigi

function y = 10 + x2 at point a in Fugure 2A.2 is the slope of th;e straight line drawn
tangent to that function at a.
What is the significance of this concept for managerial economics? Recall from the dis-
cussion of total and marginal relationships that the marginal function is simply the slope
of the total function. Calculus offers an easy way to find the marginal function by taking
the first derivative of the total function. Calculus also offers a set of rules for using these
derivatives to make optimizing decisions such as minimizing cost or maximizing profit.
Standard calculus texts present numerous formulas for the derivative of various
functions, In the hands of the skilled mathematician, these formulas allow the dcriva-
tive of virtually any function to be found. However, only a few of these rules are neces-
sary to solve most of the relevant problems in managerial economics. In this section
each of these basic rules is explained and its use demonstrated.

The Derivative of a Constant

The derivative of any constant is zero. When plotted, the equation of a constant (such
as y = 5) is a horizontal line. For any x, the change in y is always zero. Thus for any
equation y = a, where a is a constant,

(2A-1)
dx
The Derivative of a Constant Times a Function
The derivative of a constant times a function is that constant times the derivative of the
function. Thus the derivative of y = ajx), where a is constant, is

af'(x) (2A-2)
dx
For example, if y = 3x, the derivative is
dy
= 3f'(x) = 3(1) = 3

The Derivative of a Power Function

For the general power function of the form y = axb, the derivative is
dy b-
dx = bax(2A-3)
 Cl-IA PTER 2 Basic Training 57

The function y = x2 is a specific case of a power function where a = I and

b = 2. Hence
the derivative of this function is
= 2-1 =
dx
The interpretation of this derivative is that the slope of the function y = x 2 at any point
x is 2x. For example, the slope at x = 4 would be found by substituting x = 4 into the
derivative. That is,

4y—)
( dx
Thus when x 4, the change in y is 8 times a small change in
x.
The function y = x2
ally. The slope at x = 4 is is shown in Figurc 2A.3. Note that the slope changes continu-
8. As x increases, the slope becomes steeper. For negative val-
ues of x, the slope is negative. For example, if x = —3, the slope is —6.

The Derivative of a Sum or Difference

The derivative of a function that is a sum or difference of several terms is the sum (or
difference) of the derivatives ofeäch of the terms. That is, if y ftx) ± g(x), then
dy
= f'(x) + g' (X) (2A-4)
58 PART I Getting Started

For example, the derivative of the function

Y = 10 + 5x + 6x2
is equal to the sum of the derivatives of each of the three terms on the right-hand side.
Note that the rules for the derivative of a constant, a constant times a function, and a
power function must be used. Thus
dy
= 0 + (1)50 + (2)6x' = 5 + 12x
dx
Consider another function: y = 2.x - 6x 1 - 4x + 10
The derivative is = (3)2x2 - (- 2)6x 3 -4 - 0
dX
6x2 + 12x 3 -4

The Derivative of the Product of Two Functions

Given a function of the form y = f(x)g(x), the derivative dy/dx is given by

f'(x)g(x) + x)g'(x) (2A-5)

di
That is, the derivative of the product of two functions is the derivative of the first func-
tion times the second function plus the first function times the derivative of the second.
For example, the derivative of the function
y = (x2 - 4) (x 3 + 2x + 2)
is
dy
= (2x)(x 3 + Zx + 2) * - 4)(3XZ + 2) = 5x4 - 6 + 4x - 8
dx

The Derivative of a Quotient of Two Functions

For a function of the form y = f(x)/g(x) the derivative is
dy g(x)f'(x) - f(x)g'(x)
dx 2A-6)
[g(x)]2
Given the function
- 3x
Y=

the derivative would be

dy - x2 (2x - 3) - ()r2 - 3x)(2x) 3x2 - 3
dx (2)2 - x4 x

-The Derivative of a Function of a Function

The function
5)3
y (2x +
 CHAPTER 2 Basic Training 59
is really two functions combined. That is, by writing
U f(x)
whe7e
U = 2 -- 5
and

Y g)

y- U3

it is seen that v is a function of a funci00. That is.

Y g(u) = gff(x)]
This derivative is the derivative of y with respect to u and multiplied by the derivative
of u with respect to x, or

dv - dv du
dx - du dx
Now, using the rule for the derivative of a power function yields
dy
= 3u 3(2x + 5)2
du
and
du
dx
so
dy
= [3( + 5)2} '2 = 6(2 + 5)2
dx
Consider another example:

which can be rewritten as

Y = (x5 + 2x +
In this case,

Y=
and
U = x5 + 2x + 6
so the solution is
60 PART I Getting Started

= - = - 13/2
du 2 2
and
du
=5x +2
dx
Substituting x5 + 2x + 6 for u and multiplying the two derivatives just given yields

i±2A±G)7(5A4±2)
kx
dx du dx 2
These seven rules of differentiation are sufficient to determine the derivatives of all
the functions used in this book. However, sometimes two or more of the rules must be
used at the same time.

Finding the Marginal Function

Given a total revenue function
TR = 50Q - 0.5Q2
and a total cost function
TC = 2,000 + 200Q - 0.2Q2 + 0.001Q3
find the marginal revenue and marginal cost functions.
Solution Recall that a marginal function is simply the slope of the corresponding total
function. For example, marginal revenue is the slope of total revenue. Thus, by finding the
derivative of the total revenue function, the marginal revenue function will be obtained:

MR = 50— 2(0.5)Q2
dQ
=50— Q
Similarly, the marginal cost function will be found by taking the first derivative of
the total Cost function:

MC = d = 0 + 200 - 2(0.2Q 3WA0IQ1
dQ
- 200 - 0.4Q + 0.003Q'

Key Concepts

• Thctopeofa function" ftx -is the change

responding change in x (i.e.. &).
• Fora function v = jx!. the icnvauve, written as dy, rorf'xi, is tse of
the function at a particular point on Ibe function. Equivakntiv, th-i1c4vative is
the slope of a straight liiw c.lriwn tangent to the function at that point.
• By using one or more of the cven formulas outlined in this enhx, th de-
rivative of most fcnctions encountered in manarial economIcs cafl be foid..


CHAPTER 2 Basic Training 61

HIGHER-ORDER DERIVATIVES
The derivative of a function sometimes is called the first derivative to indicate that there
are' her-order derivatives. The second derivative of a function is simply the first de-
rivative of the first derivative; it is written d 2 v/dx 2 orf". In the context of economics, the
first derivative of a total function is the marginal function. The second derivative of the
total function is the slope of the first derivative or the rate at which the marginal func-
tion is changing.
Higher-order derivatives are easy to find. One simply keeps taking the first deriva-
tive again. Given the function
y=lOx 3 +jx-5x--6
-,,

the first derivative is

dy
=3Ox2+6x_5
dx
and the higher-order derivatives are
d2
second. -- = 6th 1 6

d3
third. = 60

fourth: =0

The second derivative has an important application in finding the maximum and/or
minimum of a function. 'Ibis concept is explained in the following section.

CALCULUS AND OPTIMIZATION

Recall from the discussion of total and marginal relationships that if the marginal func-
tion is positive, the total function must be increasing; if the marginal function is nega-
tive, the total function must be decreasing. It was shown that if the marginal function is
zero, then the total function must be at either a maximum or a minimum. In Figure 2A.4,
a total function and its associated marginal function are shown. At point a, which cor-
responds to x = x 1 , the total function is at a maximum and the marginal function is zero.
At point b, corresponding to x x, the total function is minimized and the marginal
function is again zero. Thus the marginal curve is zero at both x 1 and x2 . However, note
the difference in the slope of the marginal function at these points. At x 1 . the marginal
curve has a negative slope, whereas at x2 , the slope is positive.
Because the total function is at a maximum or a minimum (i.e., an extremum) when
its SiOpC is zero, one way to find the value of x that results in a maximum or a minimum
is to set the first derivative of the total function equal to zero and solve for x. This is a
better approach than the trial-and-error method used earlier. In that example, a total
revenue function,
TR = 20Q - Q2


62 PART I Getting Started

Total function

Marginal function

x.

I' g U* ZA.4 fT OLI Mi MRdm

and a total cost function,

TC = 50 + 40
were given. The problem was to find the rate of output, Q, that maximized profit.
The total profit function (it) is found by subtracting total cost from total revenue:
it = TR - TC
= 20Q - Q 2 - 50 - 4Q
= + 16Q - 50
The profit function will have a slope of zero where that function is at a maximum and also
at its minimum point. To find the profit-maximizing output, take the first derivative of the
profit function with respect to output, set that derivative equal to zero, and solve for out-
puLThat is, find the rate (or rates) of output where the slope of the profit function is zero.-5

"'Sometimes the result of this optimization process is a quadratic equation. that is,
ax2+bx+c=0

The solution (i.e., the value or values of x for which the equation is true) is given by the general equation
- - 4ac

For example, given the quadratic equation

- 60x + 500 = 0
the solution is
(-60)V(--60)7-4(l)(500)
X1, x2 -
2(1)
= 50
X, = 10
 CHAPTER 2 Basic Training 63
dir
dQ = —2Q + 16 = 0
2Q=16
Q=8
But will t
his output rate result in a profit maximum or a profit minimum?
Remember, setting the first derivative equal to zero and solving results in an ex-
tremum, but it could be a maximum or a minimum. But Figure 2A.4
shows that if the
total function is maximized, the marginal function has a negative 5l C
a fflifl jniun Conversely,
point on a total function is associated with a positive slope
­1­ of the mar-
ginal function.
Because the slope of the marginal function is the first derivative of that marginal
function, a simple test to determine if a point is a m
aximum or minimum is suggested.
Find the second derivative of the total function and evaluate it at the point where the
slope of the total function is zero, if the second derivative is negative (i.e., the marginal
function is decreasing), the total function is at a maximum. If the second derivative is
positive at that point (i.e., the marginal function is increasing), the total function is at a
minimum point.
In the profit-majn{j 0
problem just discussed, the second derivative of the
profit function is —2, which is negative. Therefote, profit is maximized at Q
8.
When finding the extremum of any function, setting the first derivative equal to
zero is called the first-order Sondition,
meaning that this condition is necessary for an
extremum but not sufficient
-to determine if the function is at a maximum or a mini-
mum. The test for a maximum or a minimum using the second derivative is called the
second-order condition.
The first- and second-order conditions together are said to be
sufficient to test for either a maximum or a minimum point. These conditions are sum-
marized as follows:

Maximum Minimum
First-order condition =0

Second-order condition

In some problems there will be two or possibly more points where the first de-
rivative is zero. Therefore, all these points will have to be evaluated using the second-
order condition to test for a minimum or a maximum. As shown in Figure 2A.5, a
function could have several points, such as a. b, and d ,
Points a and c are relative maxima and where the slope is zero.
b and d are relative minima. The term
tive rela-
means that the point is an extremum only for part of the function. Point a is a
maximum relative to other points on the function around it. It is not the maximum
point for the entire function because the value of y at point
e. c exceeds that at point
To find the absolute maximum, the value of the equation must be determined for
all relative maxima within the range that the function is defined and also at each of
the end points.
 PART' Getting Started

THE PARTIAL DERIVATIVE

Many economic phenomena are described by multivariate functions (i.e., equations
that have two or more independent variables). Given a general multivariate function
such as
Y = AX, z)
the first partial derivative of y with respect to x, denoted as b yl 6x or f, indicates the
slope relationship between y and x when z is held constant. This partial derivative is
found by considering z to be fixed and taking the derivative of y with respect to x in
the usual way. Similarly, the partial derivative of y with respect to z (i.e., Sy/z or f)
is found by considering x to be a constant and taking the first derivative of y with re-
spect to z.
For example, consider the function
yx2+3xz+z2
To find the partial derivative äy/Ox or f, consider z as a fixed and take the derivative.
Thus
f=2x1+(3z)+02X+3Z
This partial derivative means that a small change in x is associated with y changing at
the rate 2x + 3z when z is held constant at a specified level. For example, if z 2. the
 Cl-I APTER 2 Basic Training 65
slope associated with y and x is 2x + 3(2) or 2x + 6. If z = 5, the slope associated with
y and x would be 2x + 3(5), or 2x + 15.
Similarly, the partial derivative of y with respect to z would be

= 0 + 3x + 2z 21 = 3x -f 2z
This means that a small change in z is associated with y
changing at the rate 3x + 2z
when x is held constant.

OPTIMIZATION AND MULTIVARIATE FUNCTIONS

The approach to finding the maxin urn or minimum value of a multivariate function in-
volves three steps. First, find the partial derivative of the function with respect to each
independent variable. Second, set all'the partial derivatives equal to zero. Finally, solve
the system of equations determined in the second step for the values of each of the in-
dependent variables. That is, if

y=4— x2 -2 ±xz+2z2
then the partial derivatives are

ay —2x+
ax
ay
az
Setting these derivatives equal to zero

—2x + z =0
—2 + x + 4z = 0
and solving these two equations simultaneously forx and z yields

I 4

These values of x and z minimize the value of the function. The approach to test-
ing whether the optimizing solution results in a maximum or minimum for a multi-
variate function is complex and beyond the scope of this book. In this text, the
context of the problem will indicate whether a maximum or minimum has been
determined.
 66 PART I Getting Started

KeyConcepts.. : .
• Higher-order .ertvatwes are found by repeatedly ta1ungTh fustdenvative of
each resultant derivaUve
• The maximum or tUmImUm point of a fUnCtlOfl LV t)1 can be-found . by set.
twg the first derivative of tite fur on equal to zero and solviag for lb.Oialue
or values of x
• Whc the first dc.rwative of a fund on is zero the function s at a maximum fthe
second derivatiy is iiegativ,or at a 1 minimum if the second derivative is positive.
For a funcT ion navmg two or n1ore independent variables 1 6 -9-Yy . f(x, z)J the
partildenvative cy/3x is the slope rclationship between y and ss ing to
be-held,cmstaut
• Optimizing a mtiltwriatc function requires setting each p artial- &nvative
equal to zero and then solving the resulting s tern of equations simu1taneousl
(or the values of each independent variable

Problems
2A-i. Determine the first and second derivatives of each of the following functions.
a.Y=10 b.Y=3X2+4X+25
2 + 3X + 4
C. Y = (X2 — 4)(X2 + 2X + 5) d. Y= X
e. Q 100 - 0.2P2 f. R 500Q(1 5Q)
g. Y=aX2 +bX+c h. C=2,000-200X2+3X3
1. Y = (3X - 2)2 j. Y = 3X2 (' 2X3 — 2)
2A-2. Determine all the first-order partial derivatives for each of the following functions.
a. y = 3X22XZ + Z2
+ b. Q = 10K°-5L06
1. 0 d. C = 200 + 10 + 2X1X2 + 3X
C. Q lOOP1 - 2p Y
2A-3. Given the multivariate function
Y=50+18X+ 1OZ-5XZ--2X2
determine the values of X and Z that maximize the fiincticn.
2A4. The total revenue(TR) func ti on f1tiin igweuby
TR = 1,000Q — 10Q2
where Q is the rate of output per period. Determine the rate of output that re-
sults in maximum total revenue.. (Be sure that you have maximized, not mini-
mized, total revenue.)
2A-5 Smith and Wesson have written a newmanagerial economics book for which
they receive royalty payments of 15 per of total revenue from sales of the
book. Because their royalty income is tied to revenue, not profit, they want the
publisher to set the price so that total revenue is maximized. However, the pub-
lisher's objective is maximum profit. If the total revenue function is
 CHAPTER 2 Basic Training 67
TR 100,000Q - 10Q2
and the total cost function is
TC = 10,000 -1- 20Q +
Q2
determine
a. The output rate that will maximize total royalty revenue and the amount of
royalty income that Smith and Wesson would receive.
h. The
output rate that would maximize profit to the publisher. Based on this
rate of output, what ishe
t aiuuuiit
of royalty income that Smith and Wesson
would receive? Compare the royalty income of Smith and Wesson to that de-
termined in part (a). (Hint,'
First determine a function for total profit by sub-
tracting the cost function from the total revenue function.)
2A-6. A study indicated that the average cost function for a high school is
AC 10,3' OAQ + 0.00012Q2
Where Q is the number of students in the school
a. What size school (i.e., in terms of number of students) results in minimum av-
erage cost?
b. Find equations for rota] and marginal cost.
2A-7. A firm has determined that its annual profits depend on the number of sales-
people it employs and the amount spent on advertising. Specifically, the rela-
tionship between profits, ii (in millions), salespeople, S (in thousands), and ad-
vertising expenditures, A (in millions), is
1T — 10+60S+ iDA —2S2—A2
Determine the number of salespeople and the amount of advertising expendi-
tures that would maximize the firm's profits.
2A-8. A firm produces two products, milk and
cheese. Q1 and Q2 represent the output
rates for milk and cheese, respectively. The profit function is
Il — 100 + 20Q 1 + 60Q2 - 10Q 12
- 2Q 22 — 2Q1Q2
Determine the output rate for each product that will maximize profit.
 inteA I
Olsen's Pre-Owned CD Players

Sally Olsen, a recent business school graduate, is thinking about starting her own busi-
ness, and she has determined that she can procure and. market recycled compact disk
changers at $200 apiece. An expert in the field has offered the opinion-that unit sales in
any year will he 800, 1,000, or 1,200 with probabilities 0.2, 0.4, and 0.4, respectively.
Given the nature of this industry, investors expect to earn a rate of return of about 12
percent per year. As the market for this product has matured, no growth is expected.
This expert also thinks that realistically, Sally probably could maintain this business for
about 15 years. After that, she probably would have to develop a new business.
The direct cost of purchasing each unit and refurbishing it is $100. In addition,
Sally has the following annual fixec! costs:
Rent $20,000
Utilities 15,000
Property tax 5,000
Insurance 5,000

Sally would pay herself a salary of $500 per week, which is lower than the $40,000
per year offer that she is regularly offered by several large firms, including National
Electronic Sales, Inc. This salary offer is the same as that being offered to others with
Sally's training and experience.
Requirements:
1. Develop accounting and economic income statements for each possible level of
unit sales.
2. Based on expected profit, what is the value of this firm?
3. After Sally completed her business plan but before she actually started the opera-
tion, National Electronics increased its salary offer to Sally to $60,000 per year.
Again, this is now the salary being offered to others with skills similar to Sally's. Un-
der this condition, repeat requirements 1 and 2 as indicated above.0
 PART

I1I:;I'
CHAPTER 3
Demand Theory andAnalyeis

CHAPTER 4
Regression Techniaes and Demand Estimation

CHAPTER 5
Busuzess and Economic Forecasth

INTEGRATING CASE STUDY

Study Ii: Southern Turkey
 CHAPTER

Demand The
Zand Analysis
C
U Preview
I Individual Demand
• Market Demand
Determinants of Market Demand
The Markct Demand Equation
Market Demand versus Firm Demand
U Total and Marginal Revenue
S Price Elasticity
Point Versus Are Elasticity
Price Elasticity and Marginal Revenue
Determinants of Price Elasticity
Price Elasticity and Decision Making
S Income Elasticity
Inferior Goods, Necessities, and Luxuries
C Income Elasticity and Decision Making
• Cross Elasticity
Substitutes and Complements
Cross Elasticity and Decision Making
• Summary
I Discussion Questions
• Problems
• Appendix; Behind the Demand Crve
The Theory of Consumer Choke
Consumer Preferences
Indifference Curves
Budget Constraints
Utility Maximization
Consumer Choice and the Demand Curve
Consumer Choice and Changes in Demand

71
 72 PART!! Demand

PREVIEW
Demand theory and anal y sis can be a source of many useful insights for business deci-
sion nib-"qZ. Indeed, it is difficult to overstate the importance of understanding de-
S%1CCSS or failure of a business depends primarily on its at;'tn
mand. U L imatel
generate revenues by satisfn emarids of consumers. Firms that are unable to at-
tract the dollar votes of consumers are soowro 1 t of the market.
The fundamental objective of demand theory is to idem d analyze the basic
determinants of consumer needs and wants. An understanding of the forces 'ind de-
mand is n pnwrfl1l tool for managers Such knowledge provides the background
needed to make pricing decisions, forecast sales, and formulate marketing strategies.
This chapter begins with a discussion of individual demand, market demand, and
the demand faced by the firm. The focus then shifts to a basic concern of managers—
the total revenue earned by the firm. Finally, the concept of elasticity is introduced as a
tool for measuring the responsiveu:s of quantity demanded to changes in prices and
income. Three elasticity measures are discu:d: pric income elasticity, and
cross elasticity.

ThiDI\TIDUAL DEMAND
Consumer choice can be a difficult task in a modern economic system. in the United
States and in other industrialized nations, tens of thousands of goods and services are of-
fered for sale. But the purchases of most consumers are constrained by their income.
Many would like to drive a Ferrari, dine at the best restaurants, and vacation in Europe.
However, the relevant options often are a used Ford, a pizza, and a few days at the beach.
In determining what to purchase, individual consumers face a constrained opti-
mization problem. That is, given their income (the constraint), they select that combi-
nation of goods and services that maximizes their personal satisfaction. Implicitly, these
choices involve a comparison of the satisfaction associated with having a good or ser-
vice and its opportunity cost, that is, what must be given up in order to obtain it. In a
market economy, opportunity costs are reflected by prices. Thus prices act as signals to
guide consumer decisions. A high price denotes a significant opportunity cost, while a
lower price indicates that less must be given up.
One of the most basic concepts in demand theory is the law of demand. In its most
simple form, this law states that there is an inverse relationship between price and quan-
tity demanded—as price increases, quantity demanded will decrease. The law of de-
mand can be explained in terms of substitution and income effects resulting from price
changes.' The' substitution effect reflects changing opportunity costs. When the price of
a good increases, its opportunity cost in terms of other goods also increases. Conse-
quently, consumers may substitute other goods for the good that has become more ex-
pensive. The purchase of relatively more chicken and pork when beef prices increase is
an example of a substitution effect.
Next consider the income effect. When the price of a good increases, the consumer
Purchasing power is reduced.That is, at higher prices the individual cannot buy the same
bundle of goods as before. For example, $30 will buy six pounds of chocolates at $5 per

'See appendix. pp. 101-11 1.

 CHAPTER 3 Demand Theory and Analysis 73
pound, but only five pounds at $6 per pound.trhe change in purchasing power is called
an income effect because the price increase is equivalent to a reduction in the con-
sumer's income./

Case Study
The Law of Demand in Perkasie, Pennsylvania

Among the many environmental problems facing the United States is how to dis-
pose of the vast amounts of garbage generated each day by households and businesses.
In 1960, Americans discarded an average of 2.6 pounds of trash per person per, day. hut
today the number is 3.8 pounds. As the volume of garbage grows, existing disposal sites
are filling up, and it is becoming increasingly difficult to find new locations near urban
areas for landfills.
One small community used the law of demand to ease its garbage collection prob-
lem. Residents of Perkasie, Pennsylvania, were paying an annual fixed fee of 120 per
resident for garbage collection and discarding a daily average of 2.2 pounds of trash per
person. Because the collection fee was fixed, the additional expense to residents of trash
disposal was zero, and they had no financial incentive to conserve on the amount of
trash they produced.
Perkasie began charging by the bag for garbage collection. The city required that
all trash be placed in special bags sold by the city. For example, a large bag had a Ca-
pacity of about 40 pounds and sold for $1.50. Thus, the marginal cost to residents of gen-
erating additional trash increased from zero to about four cents per pound. Garbage
that was not in an approved bag was not picked up. In addition, the city introduced a re-
cycling program. Each household was given buckets to be filled with cans and bottles
that were picked up every week. The city also arranged for newspapers to be collected
once a month.
The result was predictable—people began to dump less trash. During the first year
the program was in effect, trash collections per person declined to less than one pound
per day. Perkasie citizens benefited because they paid 30 percent less than before, and
the city reduced its garbage collection costs by 40 percent. U

MARKET DEMAND
Although choices by individuals are the basis of the theory of demand, it is total or mar-
ket demand that is of primary interest to managers. The market demand for a good or
service is the sum of all individual demands. For example, consider a market that consists
of only two buyers. The demand curves for these two consumers are depicted in Figure
3.1. These demand curves show the relationship between price and quantity demanded.
Consumer l's demand curie is shown in the first panel (D I D,) and that of consumer 2
74 PART II Demand

Price per Price per Price per

unit ($) unit ($) unit ($)
Market
demand
D2

15h\ +

0 45 02 8 0 6 13
Quantity Quantity Quantity
per period per period per period
H ':. g'w1

in the second panel (D2D2). At a price of $10, the individual quantities demanded are 5
and 8 units, respectively. Hence the total market demand (DMD I, as shown in the third
panel) is 13 units, The market demand at any price is the sum of the individual quantities
demanded at that price.
Graphically, the market demand curve is the horizontal summation of the individ-
ual demand curves. That is, for any given price, the market demand curve is the sum of
the horizontal distances from the vertical axis to each individual demand curve.

Determinants of Market Demand

The effect of a change in price is depicted as a movement from one point to another
along a particular demand curve. For example, in Figure 3.1, as price increases from $10
to $15, moving along the market demand curve DMDM depicts a decrease in quantity
demanded from 13 to 6 units. A movement along the demand curve in response to a
change in the price of a good or service is referred to as a change in quantity demanded.
The price of a good or service is not the only determinant of demand. However, in
plotting a demand curve, it is assumed that other factors that affect demand are held
constant. When these factors are allowed to vary; the demand curve will shift. Such shifts
are referred fo as changes in demand A shift to the right is called an increase in demand,
meaning that consumers demand more of the good or service at each price than they
did before. A leftward shift indicates a decrease in demand. That is, less is demanded at
each price than before.
In a market economy, firms must be responsive to consumer demands.Thus, it is im-
portant that managers understand the determinants of demand. Some of the most im-
portant are consumer preferences, income levels, and prices of other goods.2
Consumer Preferences Obviously, an important determinant of demand is the pref-1
erences of consumers. These preferences can change rapidly in response to advertising.

2 0ther factors. such as population. expectations, and government policies. can also affect demand. Although
they may be important, for case of exposition they are not considered here.
 CHAPTER 3 Demand Theor, andAnalysis
75
Price of Sprite increases
Price per Price of tennis rackets
unit ($) Price per increases
D'
unit (5)

L Quantity of 7-up
Quantity of tennis
up
per period
balls per period
(a) Substitutes
(b) Complements

MMM
fads, and cus
t011is. mere was a time when gloves were considered a must for the well-
dressed woman. Today, gloves are usually worn only on special occasions. This change
in preferences caused a deciase in the demand for gloves, meaning that fewer gloves
are demanded at any given price. In contrast, pants for women have become more pop-
ular in recent years. In this case, the shift in preferences resulted in an increase in de-
mand, wit ore women's pants now demanded at each price.
c e Demand is also affected by the amount of income that consumers
available to spend. For most goods, an increase in have
c onsinner Income would cause the de-
mand curve for the product to shift to the right. For example, a few years ago,
from oil and gas leases left the state of Alaska with surplus revenues. The state revenues
responded
by providing a cash grant to every resident of the state. Alaskans used much of this ex-
tra income to buy additional goods and services. That is, the increased income
resulted
in an increase in demand for many goods and services. Later, lower oil prices reduced in-
comes' Alaska. The result was a decrease in demand for many goods and services.
Other Goods
The demand for a good is often influenced by changes in the
prices oforother goods. The nature of the impact depends on whether the goods are sub-
stitutes complements. Substitutes
arc goods that have essentially the same use. When
the price of a good goes up, the demand for its substitutes is likely to increase. For ex-
ample, 7-Up and Sprite are similar lemon-flavored soft drinks. An increase in the price
of Sprite would cause people to purchase less of that beverage and consume more 7-
Up. Thus the demand curve for 7-Up would shift to the right from DD
to D'D', as shown
in Figure 3,2a. Note that more 7-Up is demanded at each price than before the shift.
Goods that are often used together are called complements.
An increase in the price
of one such good will cause the demand for its complement to decrease. Consider ten-
nis rackets and balls. If t
he price of rackets increases, fewer people will play tennis. With
fewer people involved in
is illustrated by the leftwardthe sport, fewer tennis balls will be purchased. This outcome
shift of the demand curve from DD to DD ' in Figure 32b.


76 PART II Demand

The Market Demand Equation

The market demand function can also be expressed mathematically. If the primary de-
terminants of demand are the price of the product, income, consumer preferences, and
the prices of other goods and services, the demand equation can be written as
QD = f(P,I,P0, 1) (3-1)
where P is the price of the good or service, I is income, P0 represents the prices of other
goods, and T is a measure of consumer tastes and preferences. Equation (3-1) suggests
that there is a correspondence between the quantity demanded and the variables on the
right-hand side. Hever, the equation implies only that there are general relation-
ships. It says nothing about their nature and magnitude. For example, equation (3-1)
provides no information about how quantity demanded would be affected by an in-
crease in income. Quantifying this information requires that a functional form be cho-
sen to represent the equation for market demand. The linear form is shown below.
QD=B+aPP+azI+aOPO+aTT (3-2)

The coefficients a,,, a1, a0, and a T indicate the change in quantity demanded of one-
unit changes in the associated variables. For example, a,, is the coefficient of price. Its
interpretation is that, holding the other three variables constant, quantity demanded
changes by a,, units for each one-unit change in price. In most cases, a,, will be. negative.
To illustrate, if a',, = —2 and price is measured in dollars, a $1 increase in price would be
associated with a two-unit decrease in quantity demanded.
For many purposes it is useful to focus on the relationship between quantity de-
manded and the price of the good or service while holding the other variables constant.
If 1, P0 and T are not allowed to vary, then demand is a function only of P Hence the
linear form of the demand equation can be written as
QD = B + aP (3-3)

where B represents the combined influence of all the other determinants of demand
and a,, 0. This simple demand equation is the basis for much of the analysis in the re-
mainder of the chapter.

Example The Market Demand for Tests

Max, a graduating senior, has accumulated an impressive file of testhdi.ir ng his college
/ career. But now he needs to sell his test collection to obtain money for his impending
marriage. Three wealthy friends express interest in buying some of the tests. Max de-
termines that their individual demand equations are as follows:
Q1=30.00—LOOP
Q2 = 22.50 - 0.75P
= 37.50 - 1.25P
where the quantity subscripts denote each of the three friends and price is measured in
dollars per test.
What is the market demand equation for Max's tests, and how many more tests can
he sell for each one-dollar decrease in price? If he has a file of 60 tests, what price should
he charge to sell his en, ire collection?
 CHAPTER 3 Demand Theory and Analysis 77
Solution Market demand, Qm, is the sum of the individual demands. Thus
Qrn = Q1 + Q2 + Q3 = ( 30.00 1.0oP) + (22.50 0.75P)
+ (37.50 1.25p)
Simplifying yields

Qm 9000 - 3.00P
Because P is measured in dollars, a one-dollar decrease in price will increase quan-
tity demanded by three tests. To sell the entire 60-test collection , the price must he
set such that Q. = 60. That is,

60 90 3.00P
Solvin g thise qu
a tion gives P = $i000. Substituting this price back into the individual
demand equations gives Q1 20, Q2 = 15, and Q=25.

Market Demand versus Firm Demand

Thus far, the discussion has focused on market demand curves. But from the perspec-
tive of mana g ers, it is the demand curie facing the individual fhm that is most relevant
in pricing and output decisions.
Where a firm is the only seller in a market, the relevant demand curve is the mar-
ket demand curve. Consequently, the firm will bear the entire impact of changes in in-
comes, consumer preferences, and prices of other goods. Similarly, the pricing policies
of the firm will have a significant impact on purchasers of the firm's product. But few
businesses are the only sellers of a good or service. In the vast majority of cases, a firm
supplies only part of the total market. Consequently, the demand curve faced by the in-
dividual firm is not the same as the market demand curve.
One major difference between firm and market demand is that additional factors
affect demand at the firm level. Perhaps the most important factor is decisions made by
competitors. For example, a price cut by one firm probably would decrease sales of ri-
val firms unless those firms also reduced price. Similarly, an effective advertising cam-
paign can increase a firm's sales at the expense of its competitors.
Another difference between firm and market demand is the quantitative impact of
changes in tastes, income, and prices of other goods. Consider beef and pork as exam-
ples of substitute goods. Suppose the market demand equation indicates that a 1-cent
per pound increase in the price of beef would increase the market demand for pork by
1 million pounds per y ear. Now consider the demand equation for a small meatpacker
with a I percent market share of pork sales. Because the firm's share is a small fraction
of the total market, the impact on the firm's sales resulting from the change in beef
prices will be much less than the total market change. That is, if the demand equation
for the small firm was estimated, the coefficient showing the effect of changes in the
price of beef would be much smaller than the coefficient for the market demand equa-
tion. Similarly, the coefficients showing the effects of changes in tastes and incomes
would also be less than for the market demand equation.
For the r emainder of this chapter, the discussion
will focus on the demand curve
faced by the individual firm. However, most of the concepts discussed are equally ap-
plicable to market demand curves.


78 PART 11 Demand

Key Cone• "...... . . . . ...

A change in demanded refers to a movement along a demand curve
causcdby achange m: of the.goód or service. •.
• A change in demand ls reprcsentd by ashiftofthe demand cürve' resulting '
;
from 'i change in consumer preL eices income or prices of other goods
• If two goods are substitutes, when the price of one good increases, demand for
the other also increases. . . .•.. ... .
o If two goods are compteiient a 'ugher price f0 tone will aus a dereasc in
thc tiem,mu foi tue oiner

TOTAL AND MARGINAL REVENUE

One indication of a firm's success is the total revenue generated by the sale of its prod-
ucts. Rankings of firm size are usually made on the basis of total revenue. Similarly,
growth is often expressed in terms of increases in total revenue. In that it reflects the
ability of the firm to satisfy consumer demands, the use of total revenue as a measuie
of success has some merit.
The first two columns of Table 3.1 provide information about the demand faced by
a firm. By multiplying price times quantity, the total revenue associated with each
price—quantity pair is determined. These data are also shown in Table 3.1. Note that to-
tal revenue increases as price goes from $1 to $5 and then decreases for prices greater
than $6. This suggests that sound pricing decisions require information about demand.
In some cases higher prices may increase total revenue, whereas in other circumstances
a price increase can have the opposite effect.
Table 3.1 also shows marginal revenue. Marginal revenue is defined as the change
in total revenue associated with the sale of one more unit of the product. For example,
as quantity goes from four to five, total revenue increases from $28 to $30. Hence, the

Price Quantity Total Revenue Marguwi Revenue

$10 ' 1 ••' $10
9 2' 18 $8
8 3'' 24 6
7 4 28 4
6 5 30 2
6 30 0
4 7 28 –2
3 8 24 -4
2 9 18" -6
1 10 10 –8
 CHAPTER 3 Demand Theory and Analysis 79
Total
revenue (5)

Quantity per
period

Price per
unit (S)

Quantity per
period

MR

marginal or extra revenue associated with the fifth unit is $2. Note that marginal rev-
enue declines as quantity increases. Beyond six units, marginal revenue is negative. The
explanation stems from the inverse relationship between price and quantity. To sell ex-
tra units, the firm must reduce the price of all the units sold. Negative marginal revenue
means that the dollars received from selling the extra Unit are not sufficient to com-
pensate for the dollars lost as a result of selling all other units at a lower price. Clearly.
a firm should not incrense output beyond the point where marginal rcvcnue is zero.
The total and marginal revenue data of Table 3.1 can be plotted on a graph. If frac-
tional units are allowed, the line has the appearance of a smooth curve, as shown in Fig-
ure 3.3. Note the relationship between the total and marginal revenue curves. As long

OU I'ARL 11 Demand

as total revenue is increasing, marginal revenue is positive. At the maximum point on

the total revenue curve, marginal revenue is zero. But beyond that point, marginal rev-
enue is negative.
Figure 3.3 also shows the relationship between the marginal revenue curve and
the demand curve. Note that the two curves intercept the price axis at the same point.
Using calculus, it can easily be shown that for linear demand equations, the value of
the slope of the marginal revenue curve is twice the value of the slope of the associ-
ated demand curve. Because the two curves have the same price intercept, this implies
that the quantity intercept of the marginal revenue curve is exactly half that of the de-
mand curve.
The marginal revenue equation can be derived from the demand equation. Suppos
that the demand equation is given b y Q = B -t- aF whee a !:-^ 0. Solving for price, the
demand equation becomes

(3-4)
ap a

Multiplying by quantity gives

TR = PQ = -+ Q2
(3-5)

Marginal revenue is the derivative of TR with respect to Q. Thus

MR=d°= (3-6)
dQ a,, a,,

Note that the marginal revenue equation has the same intercept (—B/a,,) as the demand
equation and that the slope of marginal revenue, 21a r. is twice the slope of the demand
equation, 1/a r . Also note that MR = Oat Q = B12, This point corresponds to the maxi-
mum value of the total revenue func'tion. For Q <B/2, marginal revenue is positive and
iota] revenue is increasing. For Q > B12, marginal revenue is negative and total revenue
is decreasing.

Key Concepts
• Marginal revenue is the change in revenue associated with a one-unit change in•
output.
• Marginal revenue is zero at the quantit y that generates maximum total revenue
and is negative beyond that point.
• For a linear demand curve, the absolute value of the slope ol the marginal rev-
enue curve is twice that of the demand curve.

ELASTICITY
In the preceding section it was demonstrated that higher prices do not always result in
greater total revenue. A price change can either increase or decrease total revenue, de-
pending on the nature of the demand function. The uncertainty involved in pricing de-


CHAPTER 3 Demand Theory and Analysis

8I
cisions could be reduced if managers had a method of measuring the probable effect of
price changes on total revenue. One such measure is price elasticit
y of demand, which is
defined as the percentage change in quantity demanded, divided
change in price. That is, b y the percentage

-.

where the symbol . is used to denote change. Thus, price elasticit y

indicates the per-
centage change in quantity demanded for a 1 percent change in price.
For example, suppose that a firm increases the price of its product b y
quantity demanded subsequentl y decreases b y 2 percent and
3 percent. The price elasticit y would he
—3%
E =------ = -
:'tice that E
wa y s less than orisequal
negative. Except in rare and unimportant cases, price elasticity k :1!-
to zero. 3 The ex p
lanation is the law of demand. which states that
price and quantity demanded are inversely related. Thus, when the price change is pos-
dive. the change in quantity demanded is negative, and
v ice versa.

Point versus Arc Elasticity

There are two approaches to computing price elasticities. The choice between the
tw
depends on the available data and the intended use Arc elasticities are appropriate for
anal y zin g
the effect of discrete (i.e.. measurable)
measurable) changes in price. For example,
pric
increase from $1 to £2 could he evaluated by computing the arc elasticit y
. In actual prac-
tice. most elasticity computations involve the are method. The other choice is point elas-
ticit.Thjs approach can be used to evaluate the effect of ver y
small price chan g es or to
compute the price elasticity at a particular price. Point elasticities are important in the-
tic economics.
c Elasticity
The percentage change in price is the change in price divided be price
- i.e.. PIP). Similarl y, the percentage change in quantity demande
d is the change in
quantity divided by quantity Ii-e., QiQ).
Thus price elasticity can he expressed as

EP ----
B y rearranaifl g terms, equation (3-8) can be written as

EP J ) Q r3u)
-. -

Note that P and Q specify

a particular point on the demand curve, while QP i tac
reciprocal of the slo p
e of the demand curve. Thus the price elasticity will vary depend
ing on where the measurement is taken on the demand curve and also wit
( he demand curve. h the slope of

'In Some CeMs. Price

elast icities are multiplied by — 1
her tu" 4utie it obscures the to make them positive This cointjon is not adopted
i n v erse relationship between price and quantit
y demanded


82 PART 11 Demand

The price and quantity information presented earlier in the chapter can be used to
demonstrate the calculation of the arc price elasticiy ,Tablc 3.1 on page 78 indicates thai
at a price of $6, quantity demanded is 5 units. The table also shows that if the price fin-
creases to $10, quantity demanded will be I unit, Thus for a $4 increase in price, the
change in quantity demanded is -4 units. That is,
L'IQ -4
1
AP 4
Determining the value of P1Q poses a problem. Which should be used, the price - quail-
tity data be lure the price change or the price-uantiY values after the price increase'
The choice makes a significant difference in the computed price elasticity. For exampk.
if the initial data are used, then
-ix= -L20

However, it the price-quantity data after the price change are selected, then
LQP - 10 - 1000
£ -
APQ 1
In the first case, the price elasticity estimate indicates that a 1 percent increase in pr'
results in a quantity decline of just over 1 percent. But the second estimate iinplie; tn
the percentage impact on quantity demanded is 10 times that of the percentage
change.
The conventional approach used to calculate arc elasticities is to use averagL
ues for price and quantity. Thus arc price elasticity is defined as
E =—-'-
P (Q2 + Q 1 )/21 (P2
+ P1)f2
2 are.thePrice - q -Li:
where P1 andQ 1 are the initial price-quantity pair and Pi-and Q
tity values after the price change. Simplifying and rearranging terms, equation (3-
can be written as 7
21 P2 Pi
p - p,_p1A

Now using the data from Tabl'S.1, the arc elasticity is computed to be
1-5 10+6
- ------- x ------- = -2.67

Thus as price increases from $6 toSlO, the average change in quantity demanded p
percent change in price is -267 percent.

Exam ple The Price Elasticity of Demand for Playing Cai$ -

Suppose the market demand for playing cards is given 1w ih equation
Q = 6,000,000 - 1,000,000Y
where Q is the number of decks of cards demanded each year and P is the price in
lar. For a price increase from $2 to $3 per deck, what is the arc price eiasticit'v?


CHAPTER 3 Demand Theory and Analysis

83
Solution
Using the market demand equation, the quantity demanded is 4,000,000
decks of cards at a price of $2. Similarly, it is determined that the quantity demanded
is 3,000,000 decks at a price of $3 per deck. Note that in equation (3-1 1),
P 1 ) is just XQ/P The market demand equation can be used to determine (Q2 - Q1)I(F2
AQIA,,p
In that each $I increase in price causes a 1.000,000 decrease in quantity demanded, it
is known that Q/P = —1,000,000. Thus
the arc price elasticity is
Ep 1,000,000 x 2+3
= -0.71

This mcai-is iliat a 1 percent increase in price will reduce the quantity demanded by
0.71 percent.

Point Elasticity Now let's consider extremely small changes in price. For
proaching zero, the term AQIP can be written as dQJdP, where 41P ap-
dQ!d.p is the deriva-
tive of Q with respect to R B asicall y,dQ/dp expresses the rate at which
for very small changes in P For a linear demand equation, Q is changing
dQIdP is Constant. For ex-
ample, in the demand equation Q = .8 + aP the derivative dQ/dP is
ap. Thus the rate
of change for a small price change is the same as for large changes. Hence for linear de-
mand eq uation, dQ!dP =
Fur small price changes, P1 and P2 are approximately equal. Thus either price can
be used in the calculation with no significant effect on the computed price elasticity.
Hence the equation for point elasticity can be written as

(3-12)
Equation (3-12) is used to calculate price elasticities at a particular point on the de-
mand curve For example, consider the price—quantity data used earlier in this section.
It has already been determined that dQ/dP —1. If price is $6 and
manded is 5 units, the price elasticity is the quantity de-

dQP 6
Ep_ a .-= —1X= 1,20

That is, for very small price changes above or below $6, the percentage change in quan-
tity demanded is —1.20 times the percentage change in price. Again, the minus sign de-
notes that there is an inverse relationship between price and quantity.
Point elasticities can also be computed from a demand equation. Suppose that the
demand equation is as follows:

Q=100-4P
Because the relationship is linear, dQIdP is constant and equal to the rate of change in
Q0 for each one-unit change in P Note that quantity demanded changes by
— 4 units
for each Unit increase in P Thus dQ!dP = —4.
Suppose that P = $10. Substituting this value into the demand equation yields
QD = 60. Thus the point elasticity at P = $10 is

E= —4x 1-2 = —0.67



84 PART H Demand

The interpretation is that 1 percent increase in price causes a 0.67 percent reduction in
the quantity demanded.
Consider a second example using the same demand equation. Suppose that P =
$20, which implies that Q = 20. The equation is linear, so dQ/dP = -4, as before. Thus

E= -4X= -4.0

That is, at P = $20, a small change in price generates a percentage change in quantit'
demanded that is four times the percentage change in price.
Price Elasticity and Marginal Revenue
Table 3.2 reproduces the price, quantity, total revenue, and
' marginal revenue inform
tion from Table 3.1. It also shows the point price elasticity at each price. Note that tl
absolute value of the elasticities becomes smaller s prices decrease. At F = $10, ti
elasticity is -10.00, while at P = $1. it is -0.10.
Often, it is useful to classify demand relationships on the basis of price elasticiti
The following classification scheme is frequently used:
if.- Then Demand Is Said to Be
E, < -I Elastic
Ep = -1 Unitary elastic
-I 0 inelastic
Thus, based on the data in Table 3.2, at some price between $5 and $6, demand is ui
tary elastic because the price elasticity is equal to -1. For prices below this price, d
mand is inelastic. For higher prices, demand is elastic.
Figure 3.4 shows the demand and marginal revenue curves for the price-quanti.
data of Table 12. The figure has been labeled to show where demand is elastic, unita
elastic, and inelastic. The figure indicates that demand becomes less elastic at low
prices. This is a characteristic of linear demand curves. Because the curve is linea
dQidP is a constant. Thus price elasticity is determined by the value of P/Q. But as pric
decreases, PIQ also decreases. Consequently, the absolute value of E becomes smalle
and demand becomes less elastic.

Price Quantivy Price Total

-J
Marginal
Elasticity Revenue Revenue
$10 1 -10,00 $10 -
9 2 -4.50 18 $8
8 3 -2.67 24 6
7 4 -1.75 28 4
6 5 -1.20 30 2
5 6 -0.83 30 0
4 7 -0,57 28 -2
3 8 -0.38 24 -4
2 9 -0.22 18 -6
1 10 -0.10 10 -8
 CHAPTER 3 Demand Theory and Analysis 85
Price per
Unit (5)

Quantity per
period

Figure 3.4 can be used to interpret the three elasticity categories. The figure shows
that the point of unitary elasticity corresponds to the point where the marginal revenue
crosses the quantity axis. That is, marginal revenue is zero where demand is unitary elas-
tic. The explanation is not difficult. Unitary elasticity means that a 1 percent increase in
price causes quantity demanded to decrease by 1 percent. But total revenue is com-
puted by multiplying price times quantity. Thus, if demand is unitary elastic, the increase
in price is exactly offset by the decrease in quantity demanded. As a result, there is no
change in total revenue and marginal revenue is zero.
Figure 3.4 also shows that marginal revenue is positive where demand is elastic, and
negative when demand is inelastic. Although the demand curve in Figure 3.4 is linear,
these relationships are also true for nonlinear demand curves. The point where marginal
revenue is zero always divides the elastic and inelastic regions of the demand curve.4

4 However, it should be noted that there are some demand relationships for which marginal revenue is
never zero. For example, consider a demand equation of the following form:

Taking the derivative with respect to P yields

dQ
dP
Thus the price elasticity is

E (aFt)
p
But P = Q. Thus E a. Because a is a constant, it follows that E,, is constant for all values of price and
quantity. If a < —1, demand is elastic for all points on the curve. Thus marginal revenue is always positive.
Conversely, if a> —1, demand is inelastic and marginal revenue negative. If a = - I, marginal revenue is
zero and demand is unitary elastic for all price —quantity combinations.
86 PART II Demand

The choice of the terms inelastic and elastic is appropriate when viewed in light of
the relationship between price elasticity and marginal revenue. The word inelastic car-
ries the connotation of something that is not flexible or responsive. Consider a vertical
demand curve. For such a curve, quantity demanded is not affected by changes in price.
That is, dQ/dP = 0 and thus E = 0. Such curves are sometimes referred to as com-
pletely inelastic.
In contrast, for a horizontal demand curve, quantity demanded is highly responsive
to changes in prices. In fact, an arbitrarily small change in price causes an infinitely large
change in quantity demanded. That is. dQidP = -x and hence EP = -x Horizontal de-
mand curves are said to be infinitely elastic. Although there are probably no goods or
services for which 111al k demand is infinitely elastic or completely inelastic, an under-
standing of these cases is useful in economic analysis. Also, , individual producers may
face demand curves that approach these extremes.

Key Concepts
Price elasticity is defined as the percentage change in quantity demanded that
results from a 1 per.cent change in price,
' Point elasticity is used where the. price change is Very small, Arc elasticity, is ap-
propri tor a iarer change in price.
Demand is elastic if F,, <. - . 1 1 inelastic if -1 F,, 0, and unitay elastic i f:
—1.
Demand is elastic where marginal revenue is positive, unitary elastic when mar-
gina! revenue is zero, and inelastic if marginal revenue is negative.

Determinants of Price Elasticity,

Price elasticities have been estimated for many goods and services; Table 3.3 provides
some examples. The short-run elasticities reflect periods of time that are not long
enough for the consumer to adjust completely to changes in prices. The long-run values
refer to situations where consumers have had more time to adjust.
Note the variation in elasticities in Table 3.3. The long-run demand for foreign
travel by U.S. residents is elastic (i.e., P2,, = —4.10). In contrast, the long-run demand for
water is highly inelastic (i.e., P2,, = —0.14). Electricity demand is inelastic in the short
run, but elastic in the long. run. In general, three factors determine the price elasticity of
demand. They are (I) availability of substitutes, (2) proportion of income spent on the
good or service, and (3) length of time.
Availability of Substitutes Products for which there are good substitutes tend to
have higher price elasticities than products for which there are few adequate substi-
tutes. Motion pictures are a good example. Movies are a form of recreation , but there
are many alternative recreational activities. When ticket prices at the movie theater in-
(.Tease, these substitute activities replace movies Thus, the demand for motion pictures
is relatively elastic, as shown in Table 3.3.
 CHAPTER 3 Demand Theory and Anal
ysjv 87

Good or Service
Estimated Price Elasticity
Electricity -013 Short run
Electricity -1.89 Lone run
Water
-0.14 Long run
Motion pictures --3.69 Long run
Gasoline -0.15 Short run
Gasoline
-0.78 Long run
Foreign travel
-4.i 0 Lone run
Sources: H. S. Houthakker and L. D. Taylor,
Consumer
Dema.d in the United Stares; .4nafsjs and Projections
(Cambridge, Mass.: Harvard University Press, 1970).
PP, 166-167 and.! L. Sweentv.' ­ "Abe Demand
for Gasoline
A Vintage Capita! Modem. Department 01 Engineering
Economics, Stanford Unmvarsiiy, 1975

At the other extreme, consider the short-run demand for

electricit y. When the lo-
cal utility increases prices, consumers have few options. The owner of an electrically
heated home can turn down the thermostat. but the savings are limited b
y the desire to
keep warm. The Owner of an electric clothes dryer may try drying clothes on an outside
line, but this option will not he very successful during the winter or for the residents of
some apartment buildings. Similarly, there are not many short-run alternatives to using
electricity for r efrigeration and li g
relatively inelastic. hting. Hence the short-run demand for electricity is

Proportion of Income Spent

Demand tends to be inelastic for goods and services
that account for only a small proportion of total expenditures. Consider the demand for
salt. A one-pound container of salt will meet the needs of the typical household for
months and costs only a few cents. If the price of salt were to double, this change would
not have a significant impact on the family's purchasing power. As a result, price
changes have little effect on the household demand for salt. In contrast, demand will
tend to be more elastic for goods and services that require a substantial portion of to-
tal expenditures.
Time Period
Demand is usually more elastic in the long run than in the
short run.
The explanation is that, given more time, the consumer has more opportunities to ad-
just to changes in prices. Table 3.3 indicates that the long-run elasticity for electricity is
more than ten times the short-run value. As indicated previously, in the short run, peo-
ple living in an electrically heated home have few options to reduce eicctrieitv con-
sumption, But over a lon g
er time period, they may switch to gas or improve the energy
efficiency of the home. Similarly, higher electricity prices ma y
ultimatel y cause con-
sumers to find other energy sources for cooking and clothes drying.


88 PART 11 Demand

Case Study
The Short-Run Versus Long-Run Demand for Gasoline

As shown in the following table, gasoline prices increased dramatically from 1973
to 1981. At first, consumers had little choice but to use about the same amount of gaso-
line and pay the higher prices. Some vacation trips were canceled and many commuters
started going to work in buses or car pools. but the options for relief were liniied. From
1973 to 19Th, average fuel consumption per vehicle declined from 736 to 685 gallons per
year, a decrease of 7 percent. However, given more time to adjust, consumers were able
to reduce the impact of higher gas prices. Smaller, fuel-efficient cars became popular,
and the average miles per gallon of gasoline for passenger cars increased from 13.3 in
1973 to 15.7 in 1981. People also changed jobs or moved closer to their places of work.
These and other changes in driving habits reduced the average number of miles driven
per car from 9,800 to 8,700 over the same period. The net effect of these changes was
that fuel consumption per vehicle in the United States declined from 736 to 555 gallons
per year between 1973 and 1981, a reduction of nearly 25 percent. Clearly, the long-run
demand for gasoline was more elastic than the short-run demand.

Average Miles Average Fuel

Average Price Average Miles Driven per Consumption
Year of Gasoline per Gallon Vehicle per Year (gallons)
1973 $0.40 13.3 9,800 736
1975 0.57 13.7 9.400 685
1977 0.62 14.1 9,600 680
1979 0.86 14.5 9,300 638
1981 1.31 15.7 8,700 555
Source: U.S. Department of Commerce, Bureau of the Census. StazisticalAbstract of
the United States (Washington. D.C.: U.S. Government Printing Office, selected years).

Price Elasticity and Decision Making

Information about price elasticities can be extremely useful to niallagers as they con-
template pricing decisions. If demand is inelastic at the current price, a price decrease
will result in a decrease in total revenue. Alternatively, reducing the price of a product
with elastic demand would cause revenue to increase. 5 The effect on total revenue
would be.the reverse for a price increase. However, if demand is unitary elastic, price
changes will not change total revenues.

t However, a price reduction is not always the correct strategy when demand is elastic. The decision must
also take into account the impact on the firm's costs and profits. More will be said about pricing strategy in
later chapters.


CH APTER 3 Demand Theory and Analysis

89
The re
lationship between elasticity and total revenue can be shown using calculus.
Total revenue is price times quantity. Taking the derivative of total revenue with respect
to quantity yields marginal revenue:
MR= d(TR) d(PQ) .p+Q dP
(3-13)
Equation (3-13) States that the additional revenue resulting from the sale of one more
Unit of a good or service is equal to the selling price of the last unit (P), adjusted for the
reduced revenue from all other units sold at a lower price
be 0 LLttI__jj (QdP/do).m1 e quation can

\
PdQ )
MR—p(1 QdP (3-14)
But note that (Q1P) dP/dQ 1/Er
Thus

MR= P(i
+ EP (345)
Equation (3-15)
indicates that marginal revenue is a function of the elasticity of de-
mand. For example, if demand is unitar,' elastic, E,,
I then
MR - P(i =
+
Because marginal revenue is
zero, a price change would have no effect on total revenue.
In contrast if demand is elastic, E < —1 and (1 + 1/Er) >
0. Hence, marginal revenue
is positive, which means that, by increasing quantity demanded, a price reduction would
increase total revenue. Equation (345) also implies that if demand is inelastic,
revenue is negative, indicating that a price reduction would decrease total marginal
revenue.
Some analysts question the usefulness of elasticity estimates. They argue that elastici-
ties are redundant, in that the data necessary for their determination could be used to de-
termine total revenues directly.Th managers could assess the effects of a change in price
without knowledge of price elasticity. Although this is true, elasticity estimates are valu-
able, in that they provide a quick way of evaluating pricing policies. For example, if demand
is known to be elastic, it is also known that a price increase will reduce total revenues.
fhLpts -. ..
-
U)j,J iti...

• - ;

Cc not VL J1%d time t ) cFJJt1't ft

tyw tiang4
AV flw&hae cn1Qf'thc 100
• .. .•,-. ,,I;,
ft
---'?.'•'• %IVt,J ne wul1 wirpase if Jae inand
-, is elstic ( I
Th*aI rcei
nc i. d)astjc (E,
Total reven (li will be
'U14L . -.1


90 PART II Demand

Example How Much Tuition for College Students?

The board of trustees of a leading state university is faced with a critical financial prob-
lem. At present tuition rates, the university is losing $7.5 million per year. The presider'
of the university, a well-known biologist, urges that tuition be raised $750 over the pre
sent $3,000 rate—a 25 percent increase. Based on the 10,000 students now attending th
school, he projects that this increase would cover the $7.5 million shortfall in revenue
Student leaders protest that they cannot afford a tuition hike, but the president r:
sponds that the only alternative is to cut back significantly on programs and faculty. 11
faculty supports the tuition increase as a means of preserving their jobs.
The students quickly realize that any appeal that involves compassion for th'
plight is likely to fall on deaf ears.Their only hope is to demonstrate that the tuition h
is not in the best interest of the university. What can theydo?
Solution The university administration sees a tuition increase as a means of
creasing total revenue. The board of trustees is likely to reject the administrations ç C)-
posal only if the students can demonstrate that it would not achieve this purpose.
As part of a term paper, an economics major discovers a journal article that discus s
the price elasticity of demand for a college education.The author of the article estim .s
that the elasticity for enrollment at state universities is –1.3 with respect to tuiion
changes. That is, a 1 percent increase in tuition would decrease enrollments by 1.3
cent. The data are current and the paper was written by a highly respected scholar.
Based on the elasticity estimate from the paper, the students calculate that the
posed tuitionhike of 25 percent would decrease enrollmentby32.S percent,or nearly 3,
students. This would result in a decrease in total revenue from $30,000,000 at present
$3,000X 10,000)to about $25,000,000 after the tuition increase (i.e., $3,750 X 6,700).
Faced with this startling information, the board of trustees asks the university p:
ident if the cost savings from fewer students would compensate for the revenue di
The president replies that most o(the university's costs are independent of enrollni
and, hence, there would not be a significant cost saving. By a unanimous vote, the bo.
;)f trustees rejects the tuition increase and orders the president to find some other
to meet the revenue deficiency.

Bø)ME ELASTICITY
Income elasticities are used to measure the responsiveness of demand to changes in -r
a good or service
come When other factors are held constant, the income elasticity of
- the percentage change in demand associated with a 1 percent change in income. Spec
the income elasticity of a good or service is defined as
acaJly,
%iQ Qi (3-1
E,– %M Qil

where I denotes income. Rearranging terms, equation (3-16) can be written

.1 MQ
(3-17
 CHAPTER 3 Demand Theory andAnalysis 91
As with price elasticity, income elasticity can be expressed in either arc or point
terms. Arc income elasticity is used when relatively large changes in income are being
considered and is defined as

E, ,
2 1
,_ (- )
2 I
where Q1 and I represent the initial levels of demand and income, and
the values after a change in income. Q2 and 12 are
For example, suppose that the demand for automobiles as a function of income per
capita is given by the equation
Q 50,000 ± 5 (1)
What is the income elasticity as per capita income increases from $10.000 to 11,000?
Substituting J = $10,000 into the equation, quantity demanded is 100,000 cars. Simi-
larly, at 12 = $ 11,000, quantity demanded is 105,000 automobiles. Thus
- 105,000 - 100,000 11,000 + 10,000
- 11,000 >< 0.512
-
The in terpretation of this result is that over the income range $10,000 io $11.000. each
1 percent increase in income causes about a five-tenths of 1 percent increase in quan-
tity demanded.
If the change in income is small or if income elasticity at a particular income level
needs to be determined, a point elasticity is appropriate. In this case, Q/Jis expressed
as dQ/d!. Thus

E1 =
( 3-19)
To illustrate, if Q 50,000 + 51 as before, each one-unit increase in income is associ-
ated with a five-unit increase in demand. Thus dQfdJ 5. For 1
$10,500, demand is
102,500 units and the income elasticity is

E,.=5x 0.512
102,500 -
Note that this value is identical to the arc elasticity between $10,000 and $11,000 of
income, They are equal because demand is a linear function of income.Thus, the rate of
change in quantity does not vary as income increases. Hence the average rate of change
between $10,000 and $11,000 is the same as the instantaneous rate of change at $10,500.
However, for nonlinear demand equations, the arc and point elasticities at the midpoint
of the arc are not necessarily equal.

Inferior Goods, Necessities, and Luxuries

Income elasticities can be either negative or positive. When they are negative, an in-
crease in income is associated with a decrease in the quantity demanded of the good or
service-Chi hot dogs might be an example. Those living on tight budgets may be un
able to afford any other kind of meat. But as their incomes increase, they give up hot
dogs and switch to other types of meat, such as roast beef and steak. Thus the increase
 92 PART II Demand

in income causes a decrease in demand for hot dogs. Goods with negative income elas-
ticities are defined as inferior goods.
Normal goods and services have positive income elasticities. They can be further
classified by the magnitude of E1. If 0 <E1 1, the percentage change in demand is
positive but less than or equal to the percentage change in income. Such goods and ser-
vices are referred to as necessities. That is, demand is relatively unaffected by changes
irl income. A good example is bread, a basic food item eaten by even the poorest fami-
lies. As a family becomes more affluent, it will consume more bread, but the increase is
usually not proportionate to the increase in income.
F;nally, luxurics are goods and services foi which E1 > i.Tnis means that the change
in demand is proportionately greater than the change in income. For example, if E1 =
4, a 1 percent increase in income would cause a 4 percent increase in demand. Jewelry
is an example of a luxury good. As individuals become wealthier, they have more dis-
posable income. Thus, purchases of necklaces, rings, and fine watches tend to represent
a larger share of their incomes.
Income Elasticity and Decision Making
The income elasticity for afinn's product is an important determinant of the firm's suc-
cess at different stages of the business cycle. During periods of expansion, incomes are
rising, and firms setting luxury items such as gourmet foods and exotic vacations will
find that the demand for their products will increase at a rate that is faster than the rate
of income growth. However, during a recession, demand may decrease rapidly. Con-
versely, sellers of necessities such as fuel and basic food items will not benefit as much
during periods of economic prosperity, but will also find that their markets are some-
what recession-proof That is, the change in demand will be less than that in the econ-
omy in general.
Knowledge of income elasticities can be useful in targeting marketing efforts. Con-
sider a firm specializing in expensive men's colognes. Because such goods are luxurie-
those in high-income groups would be expected to be the prime customers. Thus th
firm should concentrate its marketing efforts on media that reach the wealthier se
ments of the population. For example, advertising dollars should be spent on space i'
Esquire and the New Yorker rather than in the National Enquirer and Wrestling Today

Case Study
J3ngel's Law and the Plight of the Farmer

In the nineteenth century, a German statistician, Ernst Engel, etched his name in
economic history by proposing what has become known as Engel's law. Engel studied
the consumption patterns of a large number of households and concluded that the per-
centage of income spent on food decreases as incomes increase. That is, he determined
that food is a necessity. His finding has repeatedly been confirmed by later researchers.
Examples of estimated income elasticities are shOwn in the following table. Note that
only beef has an estimated elasticity greater than unity.


CHAPTER 3 Demand Theory and Analysis 93

Food I:snniuted Income Elasticity

Beef 1.05
Chicken 0.28
Pork 0.14
Tomatoes 0.24
Potatoes 0.15
S. D. B. Suits , Agrlculiure" in Structure of
American Industry, ed. W Adams, 8th ed. (New York:
Macmillan, 1990); and D. M. Shuffeu, The Demand
and Price Structure for Selected Vegetables
(Washington. D.C.: U.S. Department of Agriculture -
19"", TeChffiliCaj Buiirin 1105),

One of the i mplications of En Sel 's

law is that farmers may not prosper as much as
those in other occupations during periods of economic prosperity. The reason is that if
food expenditures do not keep pace with increases in gross domestic product, farm in-
comes may not increase as rapidly as incomes in general. However, this tendency has
been partially oflset by the rapid increase in farm pioductivity in recent years.
In 1940,
each U.S. farmer grew enough food to feed about 11 other people. Today, the typical
farm, worker produces enough to feed 80 people.

KeyConcepb :
.- ', .•'•, '•
lniv.. Lli. tit', is tht percentage nanc,. ui dnhm
ri pir etc't c. 7i . ng in
inc'.rnie
• Ne ?atjvi: ir.cme elasticjtji denote in.ieriot- oo&
thpo m al goods Are those
It t- , I ruisdLfm s inc
c.ssty.Fo,,uxurjE,> I.

ILUI[Y

Demand is also influenced b y

prices 8f other goods and services. The responsiveness of
quantity demanded to changes in price of other goods is measured by r,
Which is defined asfthe percentage change in quantity demanded of oneg
a 1 percent change—in-- caused by
the of some other good. That is,
price
%v ()
EC=
°/LP (3-20) -
where x and represent the goods or services being considered.
For large
computed as changes in the price of y, arc cross elasticity is used. The arc elasticity is
reici ii IJemana

E = Q2 - Q1 P y,2 + Pyi
C (3-21)
PY2 PYJ Q +
-
where the subscript 1 refers to the initial prices and quantities and 2 to the final values.
Suppose that demand for x in terms of the price of y is given by
100 ± 0.5P
If P, increases from $50 to $100, then, using the equation, it is determined that Q in-
creases from 125 to 150 units. Thus the cross price elasticity is
iSfi -125 100 ± 501
E 1O+125 0.47
100-50
The interpretation is that a 1 percent increase in the price of y causes a 0.27 percent in-
crease in the quantity demanded of x.
Point cross elasticities are analogous to the point elasticities already discussed. For
small changes in P,
dQP
EC ( 3-22)
=
Based on the demand equation Q = 100 ± 0.5P , the derivative, dQdP, / = 0 .5. At P,
$20. quantity demanded is 110 units. Hence the point cross elasticity is

Ec = 0.5 )< = 0.09

Substitutes and. Complements

Cross elasticities are used to classify the relationship between goods. If E> 0, an in-
crease in the price of y causes an increase in the quantity demanded of x, and the twc
products are said to be substitute& That is, one product can be used in place of (substi-
tuted for) the other. Suppose that the price of y increases. This means that the opportu-
nity cost of y in terms of x has increased The result is that Consumers purchase less y
and more of the relatively cheaper good x. Beef and pork are examples of substitutes.
An increase in the price of beef usually increases the demand for pork, and vice versa.
When E <0, the goods or services involved are classified as complements (goods
that are used together). Increases in the pricefyiduce the quantity demanded of that
product The diminishedIniiid for y causes a reduced demand for x Bread and but-
ter, cars and tires ad computers and computer software are examples of pairs of — goods
that are coEn,pkments. -
Cross ehisutcities are not always symnietncaI That is a change in demand for good x
caused by a change m the price of goody may not equal the change in demand for gen
erated by a chan_gemthepnce of x Cdiisider the case of margarine and buttrOne sfudy
detrmuiedthata1 p_eent jncreisiTffi the price of butter-caused a 081 percnt increase
in demand for mar rarine Howeve a 1 percent increasëii'i the price of mafganne in
&eased demand for butter By only 0.67 percent." Although the two elasticities are differ-
ent, note that they are both positive, indicating that butter and margarine are substitutes.

1-i. Wold, Demand Analysis (New York: Wile). 1953).

 CHAPTER 3 Demand Theory and Analysis 95
CrossElasticity and Decision Making
Many large corporations produce several related products. Gillette makes both razors
and razor blades. Ford sells several competing makes of automobiles. Where a com-
pany's products are related, the pricing of one good can influence the demand for other
products. Gillette will probably sell more razor blades if it lowers the price of its razors.
In contrast, if the price of Fords is reduced, sales of Mercurys may decline. Information
regarding cross elasticities can aid decision makers in assessing such impacts.
Cross elasticities are also useful in establishing boundaries between industries.
Sometimes it is difficult to determine which products should be included in an industry.
For example, should the manufacturing of cars and trucks be considered one industry or
two? One way to answer such questions is to specify industries based on cross elastici-
ties. This approach defines an industry as including firms whose products exhibit a high
positive cross elasticity Goods and services with negative or small cross elasticities are
considered to belong to different industries. The definition of an industry might seem to
be an unimportant matter, but the choice can have important implications. For example,
the outcomes of antitrust cases alleging illegal monopolies are sometimes determined
primarily by the industry definition accepted by the judges assigned to the case.

Example Using Elasticities in Decision Makixi

The R. J. Smith Corporation is . a. publisher of romance novels—nothing exotic or
erotic—just stories of common people falling in and out of love. The corporation hires
an economist to determine the demand. for.,its product. After months of hard work and
submission of an exorbitant bill the'
he analyst tells the company that demand for the
firm's novels (Q) is given by the following equation:
Q=12,0005,000p±51+500p
where P,, is the price charged for the R. J. Smith novels, I is income per capita, and P. is
the price of books from competing publishers.
Using this information, the company's managers want to
1. Determine what effect a price increase would have on total revenues
2. Evaluate how sale of the novels would change during a period of rising incomes
3. Assess the probable impact if competing publishers raise their prices
Assume that the initial values of P, 1, and P are $5, $10,000, and $6, respectively.
Solution
1. The effect of a price increase can be assessed by computing the point price elas-
ticity of demand. Substituting the initial values of I and P, yields
Q = 12,000 + 5(10,000) + 500(6) - 5,000 P
which is equivalent to
Q = 65,000 - 5,000P
Note that d QJdP = –5,000. At P = $5, quantity demanded is 40,000 books.
Using these data, the point price elasticity is computed to be

= –5,000 X = – 0.625
 96 PART H Demand

Because demand is inelastic, raising the price of the novels would increase total
revenue.
2. The income elasticity determines whether a product is necessity or a luxury. It
has already been determined that the initial quantity demanded at the given val-
ues of the price and income variables is 40,000. From the demand equation the
derivative, d QJdI = 5 Thus the income elasticity is
- - 10,000
1.25
40,000 -
Because E1 > 1, the novels are a luxury good. Th us as incomes increase, sales
should increase more than proportionately.
3. The demand equation implies that dQ/dP = 50().Thus it is known that Ec is
positive, meaning that Smith's romance novels and books from competing pub-
lishors arc viewed by consumers as substitutes Computhg Ec yields

E - 500 x 6.00= 0.075

Hence, a 1 percent increase in the price of other books results in a 0.075 percent
increase in demand for R. 1 Smith's romance novels.

SUMMARY
Demand refers to the number of units of a good or service that consumers are wilhin
and able to buy at each price during a specified interval of time. Market demand is th
sum of all individual demands. A movement along the demand curve is caused b:
changes in price and is referred to as a change in quantity demanded. A change in de
mand is represented by a shift of the demand curve. Changes in deand can be caused
by changes in tastes and preferences, income, and prices of other gods and services.
Marginal revenue is the change in total revenue per one-UP change in demanc
Total revenue is increasing when marginal revenue is positive. Marginal revenue is zer
at the point of maximum total revenue and total revenue is declining when margina
revenue is negative. For linear demand curves, the absôlutè value of the slope of th
marginal revenue curve is twice that of the demand curve.
Elasticities measure the responsiveness of demand to various factors. Price elastic
ity is defined as the percentage change in quantity demanded per 1 percent change ii
price. Arc price elasticity is used to assess the impact of discrete changes in price. Poin
price elasticity is appropriate for very small price changes.
Demand is said to be elastic where .E < —1 and inelastic for —1 <E 0. Deman
is unitary elastic if E = —1. Marginal revenue is positive where demand is elastic, zer
for unitary elasticity, and negative for inelastic demand. Demand tends to be more ela
 CHA pTER 3 Demand Theory and Analysis
97
tic when (1) there are many substitutes for a good or service, (2) a substantial
tion of total income is spent on the product, and (3) the time period is longer. propor-
Income elasticity is the percentage change in quantity demanded per 1 percent
change in income. Inferior goods have negative income elasticities, while normal goods
are those with positive income elasticities. For necessities, o <
goods and services with income elasticities greater than 1. 1. Luxuries are
-
Cross elasticity is defined as the percentage change in quantity demanded of one
good per 1 percent change in the price of some other good. Cross elasticities are posi-
tive for substitutes and negative for complements Goods with high positive cross elas-
ticitjs are considered to be in the same market.

Dic580 Questions
3-1. Suppose that the possession of narijuana was legalized in all states. What would
happen to the demand for marijuana? Explain.
3-2. Why is the market demand curve usually less elastic than demand curves faced by
the individual firms in that market?
3-3. The income elasticity for soft drinks is e
stimated to be 1.00. How could an exec-
utive of a bottling firm use this information to forecast sales?
3-4. The p rofitmaxi zing pricc Will
false? Explain. never be set where demand is inelastic. True or
3-5. If demand is unitary elastic, what action could
revenue? Explain. .a manager take to increase total
3-6. Why would demand for natural gas he more inelastic in the short run than in the
long run?
3-7. Which of the following pairs of goods, are Substitutes and which are comple-
ments? Explain.
a. Ins ulation and heating oil
b. Hot dogs and mustard
c. Television sets and vide ocassette recorders
d. Rice and potatoes
3-8.
In a world with just two goods where all income is spent on the two goods, both
goods cannot be inferior. True or false? Explain.
3-9. In 1988, tuition at Commonwealth College was $4,000 per year and enrollment
was 5,000 students, By 1998, tuition had increased to $8,000. but enrollment had
increased to 5,500 students. Does this change imply
curve? Explain. an upward-sloping demand

Problems

3-1. It is known that quantity demanded decreases by two units for each $1 increase
p price. At a price of $5,
quantity demanded is ten units.
\4"—. What will be the quantity de ma n ded if price is zero?
AWrite an equation for quantity demanded as a function of price.
Write an equation that expresses price as a function of quantity.
Write an equation for total revenue.
s of three people, A, B, and C whose individual demand equa-
"v' iions are as follows:
98 PART II Demand

A: P35-0.5d
B: P=50-025Q8
C: P=40-20QT
11)e industry supply equation is given by Q = 40 + 3.5P..
Determine the .equiiibriuni price and i2ijy
Determine theóthtaie purchased by each injividual.
A market consists of two individuals. Their demand equations are Qi = 16 - 4?
f and Q2 = 20 - 21 respectively.
What is thc market demand equatin?
'f b At a puce of $2, what is the point price elasticity for each person and for the

C3-Z. The demand equation faced by DuMont Electronics for its personal computers,is
,iven by P 10,000 - 4Q.
a. Write the marginal revenue equation.
b. At what price and quantity will marginal revenue be zero?
c. At what price and quantity will total revenue be maximized?
d. If price is increased from $6,000 to $7,000, what will the effect be on total rev-
enue? What does this imply about price elasticity?
& 3-5 ne demand for shirts produced by a Canadian manufacturer has been estimated
tobeP3O—Q/200.
a. Compute the point elasticity at P = $10; at P = $15.
b. How does the point elasticity vary with the price?
-6. A manager believes that the demand for her product is given by the equation P
=50—Q/100.
• a. What is the rc elasticity of demand as price dereas from $12 to $10?
b. What is the *c elasticity of demand as price ieases from $10 to $12?
k 1. Fcr each of th following equations, determine whether demand is elastic, inelas-
tic, or unitary elastic at the given price.
a. Q=100-4P and P=$20.
b.Q=l500-2OP and P=$5.
c, P =50 - 0.1Q and P = $ 20.
-8. Sailright Inc. manufactures and sells sailboards. Management believes that the
price elasticity of demand is —3.0. Currently, boards are priced at $500 and the
quantity demanded is 10,000 per year.
a. If the price is increased to $600, how many sailboards will the company be able
to sell each year?
b. How much will total revenue change as a result of the. price increase?
-9. Demand for a softback managerial econcimics-text..is given by Q =20,000 - 300?
The book is initially priced at $30..
a. Compute the point price elasticity of demand at P $30.
b. If the objective is to increase total revenue, should the pric&be increased or
decreased? Explain.
c. Compute the arc price elasticity for a price decrease from $30 to $20.
d. Compute the are price elasticity for a price decrease from $20 to $15.
3-10. Write a demand equation for which the price elasticity of demand is zero for all
Prices.
 CHAPTER 3 Demand Theory and Anal ysis 99

3-11. A consultant estimates the price—quantity relationship for New World Pizza to
be P = 50 - SQ.
a. At what output rate is demand unitary elastic?
b. Over what range of output is demand elastic?
c. At the current price, eight units are demanded eachperiod. It
objective is
to increase total revenue, should the price be increased or decreased? Explain.
The price elasticity ror rice is CStlldL . be — 0.4 and the income elasticity is 0.8.
t-a price of $0.40per pound and a per capita income of $20,000, the dèmand?r
rice is 50 million tons - per year.
rice an inferior good, a necessity, or a luxury? Explain,
per capita income increases to $20,500, what will be the quantity demanded
rice?
"if the price of rice increases to $0.41 per pound and income per capita remains
at LuO, what will be the quariity demanded?
7-Acme Tobacco is currently selling Qpounds of pipe tobacco per year. Due to
competitive pressur, :pric ppe declines from $15 to $12. As a
result, the demand for Acme pipe tobacco increases to QUO poms year.
3/What is the cross elasticity of demand for pipes and pipe tobacco?
h. Assuming that the cross elasticity does not change, at what
isice of pipes
wouki the demand for pipe tobacco be 3,000 pounds per year? Use $15 as the
initial price of a pipe.
;-14. The McNight company/is a major producer of steel. Management estimates that
the demand for the company's steel i5 given by the equation
Q 5,000 - 1,000P + 0.11 + lOOPa
where Q is steel demand in thousands of tons per year, P is the price of steel in
dollars per pound, I is income per capita, and Pa is the price of aluminum in dol-
lars per pound. Initially, the price of steel is $1 per pound, income per capita is
$20,000, and the price of aluminum is $0.80 per pound.
a. How much steel will be demanded at the initial prices and income?
b. What is the point income elasticity at the initial values?
c. What is the point cross elasticity between steel and aluminum? Are steel and
aluminum substitutes or complements?
d. If the objective is to maintain the quantity of steel demanded as computed in
part (a), what reduction in steel prices will be necessary to compensate for a
$0.20 reduction in the price of aluminum?
-15. The price of oil is $30 per barrel and the price elasticit y i, z,ant and equal to
—0.5. An oil embargo reduces the
available by 20 percenr.iisc-the arc
eiasticity formula to calci" -
e the percentage increase in the price of oil.
The arc adverti —
to $12 ng elasticity is 1.5 as advertising expenditure increases from $10
-- iflillion. If demand is 50 at an advertising expenditure of $12 million, what
will demand h at an advertising expenditure of $10 million?
1-17. ftc 'nand equation is estimated to be 50 - 3P * 2Pr,. where P0 is the.price of
some other good. Assume the average value of P is $3 and the average value of
P0 is $6.
a. What is the price elasticity at the average values of p and P? How should the
price of the good be changed to increase total revenues?
*
100 PART II Demand

b. What is the cross elasticity at the average values of P and F0 ? What is the re-
lationship between the two goods?
c. if the equation is correctly estimated, is the good inferior, a necessity, or a lux-
ury? Explain.

- Problems Requiring Calculus

3-18. The Inquiry Club at Jefferson Utversity has compiled a book that exposes the
private lives of many of the professors on campus. Economics majors in the club
estimate that total revenue from sales of the book is given by the equation
TR = 120Q - 0.1Q3
a. Over what output range is demand elastic?
b. Initially, the price is set at $71.60. To maximize total revenue, should the price
be increased or decreased? Explain..
3-19. The demand equation for a product is given by
- P=30-0.1Q2
a. Write an equation for the point elasticity as a function of quantity.
h At what price is demand unitar y elastic?
3-20. The demand equation for aproduct is given by
201
Q P

where I is income and P is price.

a. Write an equation for the point price elasticity. For what values of I and P is
demand unitary elastic? Explain.
b. Write an equation for the point income elasticity. For what values of I and P is
the good a necessity? Explain.
3-21. Given the demand equation, Q = 12,000 - 1OP2.
a. For this equation, write the expression for the point price elasticity of demand
as a function of P
b. Over what range of prices is the demand inelastic?
3-22. Consider the demand equation, Q = 150 - PP, where the subscripts x and o re-
fer to two different goods.
a. For this equation, write the expression for the point price and cross elasticities
of demand as a function of P0 and P.r.
. 1ht is the relationship between the price and cross elasticities?
 CHAPTER 3 Demand Theory andAnalysjs 101

Appendix

Be/thzd the Demand Curve:

The Theory of Condwner Cho&ie
The discussion of demand in chapter 3 began by assuming that the relationship between
price and quantity demanded was known,The purpose of thi s appendix is to explain th
e
derivation of the demand function and to develop a theory of consumer choice. This
theory provides an understanding of the consumer decision-making process. It also
serves as a guide for predicting consumer behavior under various conditions.

CONSUMER PREFERENCES

Individuals make choices based on their personal tastes and preferentes. Those who en-
joy skiing may spend a large percentage of their income on ski equipment, whereas
those who are inclined to less strenuous activities may use their income to buy books or
CDs. Tastes and preferences are shaped by many factors. Some of the determinants are
family environment, physical condition, age, sex, education, religion, and location. In the
analysis that follows, tastes and preferences will be viewed as a given, and the discus-
sion will focus on how those tastes and preferences are transformed into consumption
decisions.
In the well-established tradition of economics, four basic assumptions are made in
developing the model of consumer choice. First, it is assumed that individuals can rank
their preferences for alternative bundles of goods and services. Consider a world in
which only two goods are available: deodorant and mouthwash. Suppose that a con-
sumer is confronted with the following combinations of those two goods:
B
5 cans deodorant 2 cans deodorant 4 cans deodorant
2 bottles mouthwash 6 bottles mouthwash 4 bottles mouthwash
The ability to rank means that the individual can assess the relative amount of satisfac-
tion that would result from each bundle of goods. For example, suppose that B is con-
sidered the most desirable bundle, and that C and A are viewed as providing equal but
lesser amounts of satisfaction than B. Using the terminology of the theory of consumer
choice, it is said that B is preferred to both C and A and that the consumer is indiffer-
ent between C and A.
The second assumption is nonsatiation. This means that individuals consider them-
selves better off if they have more of a good or service than if they have less. Consider
a bundle D that consists of
102 PART II Demand

up
3 cans deodorant
7 bottles mouthwash

The nonsatiation assumption implies that this bundle would be preferred to bundle B
because it includes more deodorant and more mouthwash.
Transitivity is the third assumption. It can be thought of as requiring that prefer-
ences be consistent. Transitivity states that if bundle D is preferred to bundle B and if
B is preferred to both A and C, bundle D must be preferred to bundles A
and C.
Finally, it is assumed that in order to get additional units of one good, consumers
are willing to give up successively fewer units of other goods. For example, a consumer
may be willing to forgo the purchase of five cans of deodèrant to obtain the first bottle
of mouthwash. However, if the person already has three bottles of mouthwash, the
value of another bottle in terms of deodorant is likely to be less than five cans.

Key Concepts
. Four assumptions form.the basis io.r he.theory of co psu êrihoio...TheY 'ac
indvtduals can ranktheir prefrrences.
1 Noncat'ahoi—peopleprc for more to less
- rdnsuvjty--ra}kmgs are consistent
4 Indi duals aren1hng to gi ve lip successiv e h' smalk.r anioxnts of one good
of
in ordcr to get other goods
adthtjon1m1fls

DDIFFERENCE CURVES

Recall that bundles A and C were viewed as being equally desirable. That is, an indi-
vidual would be indifferent to having A rather than C. If asked to select one over the
other, there would be no basis for choice. Now suppose that other bundles, designated
as E, I G, and H, are also considered equivalent to A and C. If these bandies are plot-
ted on a graph as shown in Figure 3A.1, the points can be joined to form an indifference
curve that represents all bundles of goods that provide an individual with equal levels
of satisfaction.
Note that Figure 3A.1 shows several indifference curves. All points on the curve
through B arc considered equivalent to that bundle. Because B is prelei ted to A, the
assumption of transitivity guarantees that all the points on the indifference curve pass-
ing through B are preferred to all the points on the curve associated with A. Similarly,
are
since D provides more satisfaction than B, all points on the curve passing through D
preferred to those on the curves passing through B and C Because of nonsatiation,
higher indifference curves denote increased levels of satisfaction.
The four assumptions stated earlier determine the basic characteristics of indiffer-
ence curves.The assumption that individuals are capable of ranking their preferences im-
plies that indifference curves exist. The assumption of nonsatiation assures that the
curves will have a negative slope. This is easily shown by considering a curve with posi-
 CHAPTER 3 Demand Theory andAnalysis 103
Cans of
deodorant

Bottles of
Mouthwash

LL
Cans of
deodorant

L_ Bottles of
mouthwash
—I

tive slope, such as the curve passing through point 1 in Figure 3A.2. Pick any other point
on the curve, such as 2. Note that . point 1 denotes a bundle with more of both mouthwash
and deodorant than point 2. But because of the nonsatiation assumption, having more of
both goods implies that I is preferred to 2. Thus, the two points cannot be on the same
indifference curve. Hence, indifference curves must be downward sloping.
 104 PART II Demand

Transitivity and nonsatiation guarantee that two indifference curves will not inter,
sect. This can also be seen from Figure 3A.2, which shows two indifference curves cross-
ing at point 3. Consider points 4 and 5. The bundle denoted by 4 has more of both goods
than 5 and hence must be preferred to 5. Because 3 and 5
are on the same indifference
curve, transitivity requires that 4 be preferred to 3. But this is not true; 3 and 4 are on
the same indifference curve. Thus the assumption of transitivity has been violated be-
cause preferences are not Consistent. The assumption of transitivity is always violated
when indifference curves intersect.
The assumption that consumers will be willin g to give up successively fewer units
of one good in order to get additional units of anothei good assures that indifference
curves will have the convex shape shown in Figure 3A.1. The slope at any point on an
indifference curve is the tangent to the line at that point.This slope represents the num-
ber of cans of deodorant that will be given up to get one' more bottle of mouthwash,
white still remaining on the same indifference curve. That is, it is the rate at which the
individual is willing to trade deodorant for mouthwash. This trade-off is referred to as
the marginal rate of substitution (MRS). For
example, in Figure 3A.1, if the slope of the
tangent to the indifference curve at point A is —8, individual will be willing to sub-
stitute eight cans of deodorant for one bottle ofmouthwash.
outhwash.
To satisfy the fourth assumption, the absolute value of the MRS must decrease
when moving down anindifference curve. Indifference curves that are convex to the
origin have this properly. Note that the absolute value of the slope of the curve in Fig-
ure 3Ai decreases from point A to point H. If the slope of the tangent at point
H is
—1/8, the individual is now willing to give up eight bottles of mouthwash to get a single
can of deodorant. An indifference curve must be convex in order to depict this declin-
ing marginal rate of substitution.

Key Concepts
Indifference curves are downward slopmg. convex, and do not intersect. Highei
indifference curves refleci greater levels of sa I isfac!ion. .
• The slope of anindifference curve Ls called the ma,r 2inj! rate of suhstitutici
MRS ). It d enotes the rate at which an individual is willing to trade two
gxs
or services.
• Convex indifference Curves depict a dcc iui ing marginal rate of substitWion.

BUDGETCONSTRAjs
Wants rcfl&ct individuJ tastes-' preferences. But actual purchases are strongly in-
fluenced by income and prices. Let / be income and D and P. be the prices of de-
odorant and mouthwash, respectively. In a two-good world, possible purchases are de-
fined by the following expression:

1PDQD +PMQM (3A-1)

 CHAPTER 3 Demand Theory and Analysis 105
Carts at
deodorant

-a

0 Bottles of
mouthwash
I

where QD and QM are the quantities of deodorant and mouthwash purchased. 'Ihe in-
terpretation of equation (3X-1) isstraightforward. It S t ates that the sum of money spent
on deodorant plus the sum spent on mouthwash must be less than or equal to the total
income available.
If equation (3A-l .') is treated as ar, equality, it denotes the possible bundles of de-
odorant and mouthwash that can be purchased if all the income is spent. Solving this
equation for QD yields

QD = - - QM (3A-2)

which expresses possible purchases of deodorant in terms of income, prices, and the
quantity of mouthwash purchased, Equation (3A-2) can be graphed as shown in
Figure 3A.3,
The line in Figure 3A.3 is called a budget constraint. All the points to the left of
the curve, such as point A, represent quantities of deodorant and mouthwash that can
be purchased using less than all the available income. Points to the right of the bud-
get constraint, such as point B, are bundles that cannot be purchased with the avail-
able income.
Note that the vertical intercept of the budget constraints is "D• It represents the
number of cans of deodorant that could be purchased if all income was spent on that
commodity. The horizontal intercept is 1/Pu and has a similar interpretation. The slope
of the budget Constraint is It represents the rate at which one good can be sub-
stituted for another in the marketplace. For example, if PAIPD = 1/2, two bottles of
mouthwash must be given up to get one more can of deodorant.
 106 PART H Demand

UTILITY MAXIMIZATION
It is assumed that individuals strive to achieve the most satisfaction ppssib1e from their
purchase choices. This objective is often referred to as utility maximization. It involves
consideration of both indifference curves and the budget constraint. Suppose that a -
consumer is contemplating the purchase of bundle X as shown in Figure 3A.4. Point X
lies on the budget constraint, indicating that it is an affordable bundle. But Xis not the
bundle that will maximize the individual's utility. Note that Y is also on the budget con-
straint but is on a higher indifference curve. Thus Y is preferred to X.
Point Y lies at the point of tangency between the budget constraint and the indif-
ference curve.That is, point Y is the only point on the curve that touches the budget con-
straint. Note that a point such as Z on any higher indifference curve is above the bud-
get constraint and hence is not affordable. Thus the bundle, represented by point Y is
the utility-maximizing point.
Given the individual's tastes, preferences, and income, and the prices of the two
goods, there is no other point that will provide the same level of utility or satisfaction
as point Y Because Y is the point of tangency between the indifference curve and the
budget constraint, the slope of the two lines is equal at that point. Remember that the
slope of an indifference curve is the marginal rate of substitution, and the slope of the
budget constraint is the negative of the price ratio. PM'PD. Hence the utility maxi-
mizing bundle is the point where
- = MRS (3A-3)
PD
Equation (3A-3) is easily interpreted. The price ratio represents the rate at which the
market requires consumers to substitute the two goods. The MRS is the rate at which
the individual desires to substitute the goods. Utility maximization occurs where the
rate at which the consumer wants tosubstitute is just equal to the rate at which he or
she must substitute. if these two rates are not equal, purchases can be rearranged in
such a way as to increase satisfaction or utility.

Cans of
deodorant

Bottles of
mouthwash
 CHAPTER 3 Demand Theory and Analysis 107

Key Concepts
• The utility-maximizing Point occurs where the highest indifference curve is tan-
gent to the budget constraint,
• At the utility-maximizing point, the rate that products must be traded in the
market is just equal to the rate at which the m
dividual s willing to substitute
one good for another.

CONSUMER CHOICE AND THE DEMAND CURVE

The price—quantity combinations shown on an individual demand curve are the result of
utility-maximizing decisions at various prices. This can easily be seen b y considering the
denand for mouthwash. Suppose that tastes and preferences, income, and the prices of
deodorant and mouthwash generate indifference curves and budget constraints as shown
in Figure 3A.5. If mouthwash costs $1 per bottle, the utility-maximizing bundle will be
point } which denotes five bottles of this good. That is, at P = $1, quantity demanded is
five bottles. Thus Ycorresponds to one point on the demand curve for mouthwash.
If the price of mouthwash changes, Y will no longer be the utilit y -maximizing hun-
die. Suppose that M increases to $2. This change causes the budget constraint to rotate
in a clockwise manner, as shown in Figure 3A.5. Note that the vertical intercept is the
same as before ( 1/Pa) . But the s1ope of the budget constraint is PM/P D . Thus the in-
crease in P 1 causes the line to rotate clockwise.
With the new budget constraint, the point Y is no longer affordable. Hence the con-
sumer must select a new utilit y-maximizing. bundle. As before, this point is where the
highest indifference curve is tangent to the budget constraint. In Figure 3A.5 the
utility-maximizing point is Z. Note that Z is a bundle with less mouthwash (3 bottles
compared to 5) than point Y The higher price of mouthwash reduced purchasing power
and increased the opportunity cost of mouthwash relative to deodorant.
Cans of
deodorant

/
D

Bottles of
mouthwash
FrGURE 3A.5 Pric &. Chaige and Utility Maxjm izat-jo!,
 108 PART H Demand

Nice per
unit ($)

$2

$t

EA

Bottles of mouthwash
per year

Point Z represents another price –quantity pointon the demand curve. It is the quan-
tity demanded when 1M $2. As shown by Figure 3A.6, the individual demand curve is
generated by plotting these quantities as a function of price. Additional points are
obtained by changing the price of mouthwash and determining the utility-maximizing
quantity of the product.

CONSUMER CHOICE AND CHANGES N DEMAND

In the preceding section, the theory of consumer choice was used to show how changes
in the quantity demanded (movements along the demand curve) occur. The theory can
also be used to explain changes in demand (i.e., shifts of the demand curve). Such shifts
result from a change in tastes and preferences, a change in income, or a change in the
price of other goods. Consider the impact of an increase in income. When income in-
creases, the budget constraint shifts outward, as shown in Figure 3A.7. Because the
slope of the constraint is determined by prices and not by income, the new budget con-
straint is parallel to the initial budget constraint.
Initially, the utility-maximizing point was at Y where 11 is tangent to the budget
constraint, but the extra income allows movement to a hi gher indifference curve. The
utility-maximizing point is now 2, where 12 is tangent to the new budget constraint.
Note that Z involves more mouthwash than did Y At the same price of mouthwash,
there is a greater demand for mouthwash than before. The result of this income change
would appear as a rightward shift of the demand curve—more mouthwash would be de-
manded at each price.
The effect on demand of changes in tastes and preferences is illustrated by Figure
3A.8. Initially, let the utility-maximizing point be Y, as shown in the "before" panel.
Now, suppose that concern about environmental impacts of a deodorant ingredient
causes the consumer to view deodorant less favorablThis is indicated by a shift in the
 CHAPTER 3 Demand Theory and Analysis , 109
Cans of
deodorant

Bottles of
mouthwash

Cans of Cans of
deodorant deodorant

Bottles of Bottles of
mouthwash mouthwash

indifference curve from 1 to 12 and a new utility-maximizing point, Z, as shown in the

"after" panel. Note that at the new utility-maximizing point, less deodorant is selected
than before the change in preferences. This would be represented on thedemand curve
as a leftward shift. That is, less deodorant would be demanded at any price.
In a similar way, the theory of consumer choice can be used to show how changes
in the prices of other goods cause the dcmand curve to shift. Point Yin Figure 3A.9 is
the u tility-maximizing bundle for given income, tastes and preferences, and prices of
deodorant and mouthwash
Now, suppose that the price of mouthwash decreases. This is depicted in Figure 3A.9
by a counterclockwise rotation of the budget constraint. The new utility-maximizing bun-
dle is point Z,
where 12 is tangent to the new budget constraint. Note that this point spec-
110 PART 11 Demand

Cans of
deodorant

Bottles of mouthwash

ifies more mouthwash than was selected before the price reduction. This is because the
opportunity cost of buying mouthwash has declined. The price change would be depicted
as a movement from one point to another along the demand curve for mouthwash.
Note also that point Z involves less deodorant than did point Y. That is, the decrease
in the price of mouthwash resulted in a decrease in the demand for deodorant, This
change causes a leftward shift in the demand curve for deodorant—less will be de-
manded at each price.
The impact of a price change of 6ne good on the demand for the other good de-
pends on the relationship between the two products. Apparently, mouthwash and de-
odorant were considered by the consumer to be substitutes, Thus a decrease in the price
of mouthwash decreased the demand for deodorant. If the goods had been comple-
ments, analysis using the theory of consumer choice would have predicted an increase
in the demand for deodorant. In this case, a decrease in the price of mouthwash would
have resulted in a utility-maximizing point such as X, which specifies more deodorant
than before the price change.

Ke 1onupt
Change in income are shown bf japaraild shift of the budget constrainL A
charc in the pr of onc good caukathe l.'uJgct cUu.[iaiut ft wtate.
-'-At different price'. the quautitk demanded shown by the demandcurve cor-
resn -md tc ih1' int rftin;wv etwii hi.!d2e! strintc ant! i!j!fre.9ce
curves.
•.. Shifts of the demand curve reflect changes in the budget constraint or indiffer-
ence curves caused b y changes in ncome, prices of other goods, or tastes and
preferences. - -
 CHAPTER 3 Demand Theory and Analysis
111
Discuss ion Questions
3A-1. "The price ratio is the rate at which individuals must substitute goods in the mar-
ketplace." Explain this statement.
3A-2. How do changes in income affect the slope of the budget constraint? Explain.
3A-3. Suppose that bundle A is preferred to B, C is preferred to
D, and D is preferred
to B. What can be said about C in relation to B?
What about C in relation to A?
Are any of the assumptions of the theory of consumer choice violated by these
preferences?
Problems
3A-1. Consider a world in which there are only two goods. An individual has an income
of $30,000, the price of deodorant is $3 pecan, and the price of mouthwash is $4
per bottle.
a. Expressing deodorant as the dependent variable, write the equation for the
budget constraint.
b. What is the slope of the budget constraint?
c. if the price of mouthwash increases to $5,
write the new equation for the bud-
get Constraint.
3A-2. Assume that the budget constraint is given by the equation Q1 = 1,000 - 5Q2,
where Q1 and Q2 represent quantities of two goods. Normally, indifference
curves are convex to the origin, but assume in this case that they are linear with
a constant slope of —2.
a. Graph the budget constraint (with Q1 on the vertical axis).
b. Draw in a set of indifference curves and label the l
uti ity-maximizing point.
c. Where would the utility-maximizing point have been if the indifference
curves had aconstant slope of —6?
 CHAPTER

Regression l4iniques and

Deman-Itiniatioii

• Preview
• Regression Techniques
Estimating Coefficients
Testing Regression Estimates
Prediction Using Regression Equations
Multiple Regression
I Demand Estimation
Development of a Theoretical Model
Data Collection
Choice of Functional Form
Estimation and Interpretation of Results
I Probleniswith Regression Analysis
Omitted Variables
Identification
Multicollinearity
I Surunary
• Discussion Questions
• Problems

112
 CHAPTER 4 Regressioi Techniques and Demand Estimation 113

PREVIEW -: S

In the previous chapter, demand analysis was introduced as a tool for managerial deci-
sion making. For example, it was shown that a knowledge of price and cross elasticities
can assist managers in pricing and that income elasticities provide useful insights into
how demand for a product will respond to different macroeconomic conditions.
In chapter 3. it was assumed that these elasticities were known or that the data were
already available to allow them to be easily computed. Unfortunately, this is not usually
the case. For many business applications, the manager who desires information about
elaslicities must develop a data set and use statistical methods to estimate a demand
equation from which the elasticities can then be calculated. Ibis estimated equation can
then also be used to predict demand for the product, based on assumptions about prices,
income, and other factors.
In this chapter, the basic techniques of demand estimation are introduced. The first
major section considers regression analysis, which is a statistical method for fitting an
equation to a data set. Regression analysis is used for demand estimation in this chap-
ter and is also the technique used to estimate production and cost equations in later
chapters. The second section describes the four basic steps involved in estimating a de-
mand e q uation. Finally, three potential difficulties associated with regression analys
is
are discussed—omitted variables, the identification problem, and multicollinearity.

REGRESSION TECHNIQUES
Consider the simple demand equation Qd = B + aP The law of demand implies that
the coefficient a should be negative, indicating that less of the product is demanded at
higher prices. However, in making pricing decisions, it may not be sufficient to know
that quantity demanded and price are inversely related. An estimate of the numerical
value of a and also of the coefficient B may be required for decision making. Similarly,
to fully understand the correspondence between inputs and outputs in production func-
tions and between costs and output in cost functions, it is often necessary to quantify re-
lationships between the variables. The most widely used technique in economics and
many other sciences for estimating these relationships is the least-squares regression
method. The basic elements of this technique are developed in this section.
Although the focus of this chapter is on estimation of demand, the example used in
this section to introduce regression analysis is based on cost and output data. The rea-
son for selecting a cost function is that it is somewhat easier to explain and to under-
stand in a regression context than are psice–quantity relationships. After the basics of
least-squares analysis have been presented, thc remainder of the chapter considers how
the method is used to estimate demand functions.

Estimating Coefficients
Consider a firm with a fixed capital stock that has been vented under a long-term lease
for $100 per production period. The other input in the firm's production process is la-
bor, which can be increased or decreased quickly depending on the firm's needs. In this
case, the cost of the capital input ($100) is fixed and the cost of labor is variable. The
114 PART II Demand

manager of the firm wants to know the relationship between output and cost, that is, the
firm's total cost function. This would allow the manager to predict the cost of any spec-
ified rate of output for the next production period.
Specifically, the manager is interested in estimating the coefficients a and b of the
function
Y = a + bX
whete the dependent variable Yis total cost and the independent variable Xis total out-
put. If this function is plotted on a graph, the parameter a would be the vertical inter-
cept (i.e., the point where the function intersects the vertical axis) and b would be the
SiOpc of the function. Recall that the slope of a total function is the marginal function.
As Y = a + bX is the total cost function, the slope, b, is marginal cost or the change in
total cost per unit change in output.
Assume that data on cost and output have been collected for each of seven produc-
tion periods and are reported in Table 4.1. Note that there is a cost of $100 associated
with an output rate of zero. This represents the fixed cost of the capital input, which must
be paid regardless of the rate of output. These data are shown as points in Figure 4.1.
They suggest a definite upward trend, but they do not trace out a straight line. The prob-
lem is to determine the line that best represents the overall relationship between V atid
X. One approach would simply be to "eyeball" a line through these data in a way that
the data points were about equally spaced on both sides of the line. The coefficient a
would be found by extending that line to the vertical axis and reading the Y-coordinate
at that point. The slope, b, would be found by taking any two points on the line, {X1 , Y1)
and (X2, Y2 ,1 and using the slope formula
= Y2 -
b
x2-xi

Although this approach could be used, the method is quite imprecise and can be
employed only when there is just one independent variable. What if production cost de-
pends on both the rate of output and the size of the plant? To plot the data for these
three variables (total cost, output, and plant size) would require a three-dimensional di-
agram; it would be nearly impossible to eyeball the relationship in this case. The addi-
tion of another independent variable, say average skill levels of the employees, would
place the data set in the fourth dimension, where any graphic approach is hopeless.

Tata! Car: Total Output

Production Period (Y1) (X,)
I $100 0
2 150 5
3 160 8
4 240 10
5 230 15
6 370 23
7 410 25
 CHAPTER 4 Regression Techniques and T.)'#nand Estimation 115
Iota] cost (Y)

500

400
]2.21x

300

200

Iuqj

Output
y
C I

9.

There is a better way. Statisticians have demonstrated that the best estimate of the
coefficients of a linear function is to fit the line through the data points so that the sum
of squared vertical distances from each point to the line is minimid.This technique is
called least-squares regression estimatiOn.
Based on the output and cost data in Table 4.1, the least-squares regression equa-
tion will be shown to be
Y 87.08 + 12.21X
This equation is plotted in Figure 4.1. Note that the data points fall about equally on
both sides of the line.
Consider an output rate of 5. As shown in Table 4.1, the actual cost associated with
this output level is 150.1he value predicted by the regression equation, referred to as
is 148.13.That is,Y = 87.08 + 12.21(5) = 148.1 3.The deviation of the actual Yvalue from
the predicted value (i.e., the vertical distance of the point from the line), Y1 - } is re-
ferred to as the residual or the prediction error.
There are many values that might be selected as estimators of a and b, but only one
of those sets defines a line that minimizes the sum of squared deviations [i.e.. that mini-
mizes (Y - Y)2] . The equations for computing the least-squares estimators a and are

- X)2
(4-1)
and

(4-2)
where Y and X are the means of the Y and X variables.
Using the basic cost and output data from the example, the necessary calculations
are shown in Table 4.2. Substituting the appro p riate values into equations (4-1) and
(4-2), the estimates of b and 2 are computed to be
6,245.71 -
1221
- (X - X)2 - 511.40 -
a = V — bX -' 237.14— 12.21(12.29) = 87.08
116 PART II Demand

Cost Output
(Ye) (X,) - - ' (X, - ( t - - 19
100 —137.14 —12.29 151.04 1,685.45
150 —87.14 —7.29 53.14 635.25
160 —77,14 —4.29 18.40 330.93
240 10 —002.86 —2.29 524 —6.55
230 15 --7.14 2.71 7.34 —19.35
370 23 132.86 10.71 114.70 1,422.93
4• n "C
- U 172.86 12. 71 161.54
Y = 237.14 = 12.29
= 511.40 = 6,245.71

Thus the estimated equation for the total cost function is

Y = 87.08 + 12.21X
The estimate of the coefficient a is 87.08. This is the vertical intercept of the ré-
gression line. In the context of this example, a = 87.08 is an estimate of fixed Cost. Note
that this estimate is subject to error because it is known that the actual fixed cost is $100.
The value of b is an estimate of the change in total cost for a one-unit change in output
(i.e.. marginal cost), The value of 1, $12.21, means that, on average, a one-unit change in
output results in a $12.21 change in total cost. Thus b is an estimate of marginal cost.

Key Concepts
• The least-squares regression techniue s used to estimate the oeffic]ents of a
function by fitting a line through the data so that the sum of s.quared deviations
e , - is minimized
• Estimates of the coefficients of the function V = a + bX are gtvn by the equa-
tions
X1 -
an d ê- - - bX
-
• The valuA,of a estimates the vertical intercept or the estimated value of Ywhen
9
/fe value of b estimates the change in Yfor a one unit chauge tnX

4imating the Demand for Lobster Dinners

The basic regression toolsjust discussed can also be used to estimate demand relation-
ships. Consider a small restaurant chain specializing in fresh lobster dinners. The busi-
ness has collected information on prices, and the average number of meals served per day
for a random sample of eight restaurants in the chain. These data are shown below. Use
regression analysis to estimate the coefficients of the demand function Qd a + bP.
Based on the estimated equation, calculate the point price elasticity of demand at the
mean values of the variables.


CHAPTER 4 Regression Techniques and Demand Estimation

[BIN

Meals per
Cm Day (Q) Price (P)
1 100 $15
2 90 18
3 85 19
4 110 14
5 120 13
6 90 19
7 105 16
8 100 14

Solution The mean values of the variables are O = 100 and fl= $16, The other data
needed to calcu!ate the coefficients of thc deniwid equation are shown below,
Cia' Q P, -'F (P, .-- (P P)(Q. -
1 0 —1 1 0
2 —10 2 4 —20
3 —15 3 9 —45
4 10 —2 4 —20
5 20 —3 9 —60
6 —10 3 9
7 —30
5 0 / 0 0
8 0 —2 4 0
- P)2 = 40 Z(P1 - - Q) = —175
As shown, the sum of the (P1 - P)2 is 40 and the sum of the
(P1 - P)(Q1 - Q) is
— l?5.Thus, using equations (4-1) and (4-2),
b = — 175/40 = —4.375
and a = IOU - (-4.375)(16) = 170.
Hence, the estimated demand equation is Qd = 170 - 4.375R Recall from chapter
3 that the formula for point price elasticity of demand is E (dQ/dP)(P/Q). Based on
the estimated demand function, dQ/dP = —4.375. Thus, using
the mean values for the
price and quantity variables, E (-4.375(16/100) —0.7.

Testing Regression Estimates

Once the parameters have been estimated, the strength of the r elationship between the
dependent variable and the independent variables can be measured in two ways The
first uses a measure called the coefficient of determination denoted as
R2. to measure
how well the overall equation explains changes in the dependent variable. The second
measure uses the -statistic to test the strength of the reationship between arindepen-
dent variable and the dependent variable.
Testing Overall Explanatory Power Define the squared deviation of any
the mean of Y [i.e., (Y, - Y) 2 } as the variation in Y Y, from
The total variation is found by sum-
ming these deviations for all values of the dependent variable. That is.
total variation -
(4-3)
118 PART II Demand

=j + cx

Unexplained Variation
Total Variation
(Y, - Y1)
Explained Variation
F)

.ff • ai' a v... ._.

I \itri.. J&,h in a l'tr.ioi
X Mokt

Total variation can be separated into two components: explained variation and unex-
plained variation. These concepts are explained below. For each X j value, compute the
predicted value of Y1 (denoted as Y) by substituting X1 in the estimated regression
equation:
fi =d+
The squared difference between the predicted value 1 and the mean value Y [i.e., (Y - Y)2]
is defined as ep/aif2ed variation. The word explained means that the deviation of Y from its
average value Y is the result of (i.e., is explained by) changes in X For example, in the data on
total output and cost used previously, one important reason the cost values are higher or lower
than Y is because output rates (Xi) are higher or lower than the average output rate.
Total explained variation is found by summing these squared deviations, that is,
total explained variation - Y)2 (44)
i.
Unexplained variation is the difference between Y, and S',. That is, part of the deviation
of Y, from the average value (Y) is "explained" by the independent variable, X The re-
maining deviation, V1 - Y, is said to be unexplained. Summing the squares of these dif -
ferences yields
total unexplained variation = - j7)2 (4.5)
The three sources of variation are shown in Figure 4.2.
The coefficient of determination (R2) measures the proportion of total variation in
the dependent variable that is "explained" by the regression equation. That is,
= total explained variation - ( ki -
R2
total variation - (Y1 - Y)2 (4-6)
The value of R2 ranges from zero to 1. If the regression equation explains none of the
variation in Y (i.e., there is no relationship between the independent variables and the
dependent variable), R2 will be zero. If the equation explains all the variation (i.e.. total
explained variation = total variation), the coefficient of determination will be 1. In gen-
eral, the higher the value of R2, the "better" the regression equation. The term fit is of-
ten used to describe the explanatory power of the estimated equation. When R2 is high.
the equation is said to fit the data well. A low R2 would be indicative of a rather poor fit.
 CHAPTER 4 Regression Techniques and Demand Estimation 119

Total Explained Unexplained

Variation Variation Variation
(Y _2
(Y )2 f?
-
100 18,807.38 87.08 22,518.00 166.93
150 7,593.38 148.13 7,922.78 3.50
160 5,950.58 184.76 2,743.66 613.06
240 8.18 209.18 781.76 949.87
230 50.98 270.2.3 1,094.95 1,618.45
11 17 , 65 1.78 .)I.91 17,100.79 4.37
410 29,880.58 392.33 24,083.94 312.23
Y= 237.14 Y)2 - 79,942.86 Y)2 = 76,245.88 - = 3,668,41

How high must the coefficient of determination be in order that a regression equa-
tion be said to fit well? There is no precise answer to this question. For some relation-
ships, such as that between consumption and income over time, one miaht expect to
be at least 0.95. in other cases, such as estimating the relationship between output and
average cost for fifty different producers during one production period, an R2 of 0.40 or
0.50 might be regarded as quite good.
Based on the estimated regression equation for total cost and output, that is,
',= 87.08 + 12.21X1
the coefficient of determination can be computed using the data on sources of variation
shown in Table 4.3.
= explained variation = 76,245.88
R2 = 0.954
total variation 79,942.86
The value of R2 is 0.954, which means that more than 95 percent of the variation in to-
tal cost is explained by changes in output levels. Thus the equation would appear to fit
the data quite well.
Evaluating the Explanatory Power of Individual Independent Variables The I-test
is used to determine if there is a significant relationship between the dependent vari-
able and each independent variable. This test requires that the standard deviation (or
standard error) of the estimated regressibn coefficient be computed. The relationship
between a dependent variable and an independent variable is not fixed because the es-
timate of b will vary for different data samples. The standard error of b from one of
these regression equations provides an estimate of the amount of variability in b. The
equation for this standard error is
2)
S&—-
where n is the number of observations. For the production-cost example used in this sec-
tion, n = 7 and the standard error of h is

120 PART II Demand

^'^7 W S4
5P cs ec.

Degree of Freedom t-value

1 12.706
3 3.182
5 2.571
7 2.365
10 2.228
20 2.086
30 2.043
60 2.000
120 1.980

• - C6-
8 4f/5
.

= 1.19
- 511.40
The least-squares estimate of is said to be an estimate of the parameter b. But it
is known that b is subject to error and thus will differ from the true value of the para-
meter b. That is why 1 is called an estimate.
Because of the variability in b, it sometimes is useful to determine a range or inter-
val for the estimate of the true parameter b. Using principles of statistics, a 95 percent
confidence interval estimate for b is given by the equation
± tfl_k_lS;
where tn_k_i represents the value of a articular probability distribution known as Stu-
dent's t distribution. The subscript (n - k - 1) refers to the number of degrees of free-
dom, where n is the number of observations or data points and k is the number of inde-
pendent variables in the equation. An abbreviated list oft-values for use in estimating 95
percent confidence intervals is shown in Table 4•41 In the example discussed here, n = 7
and k = 1, so there are five (i.e.,? 1 - 1) degrees of freedom, and the value of in the
table is 2.571. Thus, in repeated estimations of the output—cost relationship, it is ex-
pected that about 95 percent of the time the true value of parameter b will lie in the in-
terval defined by the estimated value of b plus or minus 2.571 times the standard error
of b. For the output-cost data, the 95 percent confidence interval estimale would be
12.21 ± 2.571(1.19)
or from 9.15 to 15.27. This means that the probability that the true marginal relation-
ship between cost and output (i.e., the value of b) within this range is 0.95.
If there is no relationship between the dependent and an independent variable, the
parameter b would be zero. A standard statistical test for the strength of the relation-
ship between Y and X is to check whether the 95 percent confidence interval includes

'A more complete list of r-valus is found in Table HI on page 668.

 CHAPTER 4 Regression Techniques and Demand Estimation 121
the value zero. If it does not, the relationship between X and Y as measured by I is said
to be statistically significant. If that interval does include zero, then b is said to be non-
significant, meaning that there does not appear to be a strong relationship between
the to variables. The confidence interval for in the output-cost example did not in-
clude zero, and thus it is said that b, an estimate of marginal cost, is statistically signifi-
cant or that there is a strong relationship between cost and rate of output.
Another way to make the same test is to divide the estimated coefficient (b) by its
standard error. The probability distribution of this ratio is the same as Student's t dis-
tribution; thus this ratio is called a t-value. If the absolute value of this ratio is equal to
or grcatcr thau the tabled value of t tor n - k - 1 degrees of freedom, b
is said to be
statistically significant. Using the output-cost data, the t-value is computed to be
= 12.21 = 10.26
=
1.19
Because the ratio is greater than 2.571,'the value of the t-statistic from Table 4.4, it is
concluded that there is a statisticall y significant relationship between cost and output.
In general, if the absolute value of the ratio b/Si is greater than the value from the table
for n - k - 1 degrees of freedom, the coefficient b is said to be statistically significant.

K.y (.ncep:
'-•:•
• ni Lflh'ieT 01 Jurination R, i a mcasi eof th of total var-
atiu,i in ffi ief'3z'L-u I1aW that icxDlau h iess'ion cquatior.
• A percufl CO fidcuce nktval eSTitnte of njeter b;s giyen b y h t
C,, :.S•.. if ttts nL'rval d c' no 1UUiUd zero, to hestaticallv signif-
• i.icniug ii1hre is atr' ,n. r;ionshi n the iident and:in-
deptncleii vail bleL
• U... ....:. ..t..... ...i --
u d er*e i-statitic)
variable.

Prediction Using Regression Equations

The regression equation can be used to predict or estimate the value of the dependent
variable given the value of the independent variable.
The estimated total cost function, *
Y 87.08 + 1221X
can be used to make a point estimate of the cost of a particular rate of output, say 20,
by substituting X = 20 and solving for Y Thus
Y =- 87.08 I 12.21(20) = 331.28
This means that the predicted cost of producing 20 units of output is $331.28.
Recall that the actual output-cost data points do not lie on the regression line
but are dispersed above and below that line. This means that a value predicted by the
regression equation will be subject to error. That is, one would not expect the
122 PART II Demand

predicted values to be 100 percent accurate. The standard error of the estimate (Se)
is a measure of the probable error in the predicted values. The formula for this stan-
dard error is
- -b (X - g)(Y—Y)
S—\i n—k—i
The predicted value Y is called a point estimate of the value of the dependent vari-
able to distinguish that estimate from a confidence interval estimate. The latter is a
range of values that is expected to include the actual Y value 95 percent of the timeThe
standard error of the estimate is a fundamental part of the confidence interval estimate.
For example, for a given predicted value of the dependent variable, Y, the 95 percent
confidence interval estimate is given by
± tn_kiSe
This means that 95 percent of the time the actual value of V will fall within that range.
Suppose that management is considering a production run of 22 units of output anc
wants to knOw the estimated cost. Substituting 22 for Xin the regression equation yield
the point estimate:
1' = 87.0 + 12.21(22) = 355.70
Using equation (4 . 7) and data from Tables 4.2 and 43, the standard error of the esti
mate is computed as
f,942.86 - (12.21)(6,245.71)
Sg
= ______- = 27.14
5
Thus, a 95 percent confidence interval estimate for the cost of producing 22 units of out
put is
1' ± 2.571 (S,.) or 355.70 ± 2.571(27.14)
Based on this statistical analysis, it is expected that in repeated production runs of 22
units, 95 percent of the time the cost will be in the range $285.92 to $425.48. Equivalently,
only 5 percent of the time will the actual cost be higher than $425.48 or lower than $285.92.

Case Study
The Value of a Poor Administrator

Many employees complain that their bosses do not deserve the large raises they receive
each year, and college professors are no exception. If questioned, most facult'i members
e
could identify a departm nt head, dean, vice-president, or president they believe is re-
ceiving salary adjustments that do not correspond to that person's contribution to the
institution.
Are undeserved raises the exception or the rule at colleges and universities? Is
there a relationship between raises received by administrators and their job perfor-
 CHAPTER 4 Regression Techniques and Demand Estimation 123

mance? Faculty members at the University of South Florida used regression analysis to
evaluate this question.
Each year, administrators at the university are rated by the faculty using a five-
point scale, where 1 equals very poor and 5 equals very good. About 20 percent of the
faculty respond to the survey. To investigate the relationship between performance and
pay, ratings for 14 administrators were compared with the raises those same administra-
tors received the same year. The equation used to estimate this relationship was Y = B
+ aX where Y is the administrator's raise and X is the average rating for the individual.
Using ordinary least-squares regression, the estimated equation was Y - 19,680
- 3,887X The sign of the coefficient of X implies that job performance and raises were
inversely related at the University of South Florida. The coefficient of X was statisti-
cally significant and the R2 was 0.30. The equation can be used to predict an adminis-
trator's raise based on the rating. For example, for a rating of 2, Y 19,680 - 3,887(2)
= $11.906. The other predictions for raises are shown below.
Rating Raise
Very poor (1.00) $15,793
Poor (2.00) 11,906
Av erage (3.00) 8,019
Good (4.00) 4,132
Very ,good (5.00) 245

One interpretation of these findings is that the university's system for awarding
raises is deeply flawed. How might the low response rate to the survey affect the valid-
ity of the conclusions?

Multiple Regression
Estimation of the parameters of an equation with more than one, independent variable
is called multiple regression. In, principle, the concept of estimation with multiple re-
gression is the same as with simple linear regression, but the necessary computations
can he much more complicated. For an equation with three or more independent vari-
ables, the time required to calculate the values and the likelihood of an arithmetic er-
ror make computation by hand impratical. Consequently, virtually all regression
analysis involving multivariate equations uses computers.There are dozens of programs
available that can perform this task.
Because most economic relationships involve more than a simple relationship be-
tween a dependent and a single independent variable, multiple regression techniques are
widely used in CC0iion1ic. For example, the demand for a product usually depends on
more than just the price of the good. Other variables, such as income and prices of other
goods, can also have an influence. Thus, a simple regression equation involving only
quantity and price would be incomplete and probably would result in an incorrect esti-
mation of the relationship between quantity and price.This is because the effects of other
variables omitted from the equation are not taken into account. Similarly, a regression
 124 PART II Demand

equation that included only the rate of output as the determinant of costs could gener-
ate inaccurate results because other factors, such as input prices, also affect costs. The
Problem of omitted variables is discussed later in this chapter.
With multiple regression, it is important that the user understand how to interpret
the estimated coefficients of the equation. Earlier in the chapter it was assumed that
costs were a function of output. Now assume that costs are also determined by the price
of labor. Thus, the multiple regression equation can be written as
Y = A +bX+cZ
where V is total cost, Xis out p ut. 7 is
the price of labor, and A, b, and c are the coeffi-
cients to be estimated. The coefficients of X and 7 indicate the effect on total cost of a
one-unit change in each variable, holding the influence of the other variable constant,
For example, b
shows the change in total costs for a one-unit hange in outpu
ing that the price of labor stays the same. t, assum-
The coefficient of Z estimates the effect of a
unit change in labor price, assuming that the rate of output is unchanged.

Key Concepts
• A regression equation can he used to
pr'di t h .... 'he of the dep ei, deit j
able for gwen viuc of the iiidepenknt vari-
Vfl:thlL'.
• The standaj d error of the dstimate is a measure ofthe error in prediction.
• With multiple regression, each estimated coefiicien measures
one vai-jahle on the depndcnt variable. hoJdiji cL nstrj t the impact of
variables. the influe nce of other

DEMAND ESTIMATION

Although the process can be very complex, regression analysis really involves just four
steps: (1) d
evelopment of a theoretical model, (2) data collection, (3) choice of a func-
tional form, and (4) estimation and i
nterpretation of results. In this section, these steps
are described in the context of estimating a demand equation.

Development of a Theoretical Model

In using regression, the analyst must formulate a theoretical model of the economic re-
lationships involved. This model should be based on Sound economic theory and be ex-
pressed in m athematical terms. F
undamentally, the m odel-building process involves de-
termining which variables should h e
included in the analysis and if there is a theoretical
rationale for predicting the nature and magnitude of the rel
ationships between vari-
ables. For example, if the objective is to estimate a demand equation, it would be rea-
sonable to assume that quantity demanded is a function of the
price of the good, in-
come, prices of other goods, and tastes and preferenecs, Thus, the general form of
demand equation could be written as

2 1n mathematical terms, b is the partial derivative of

derivative with respect to Z. Y with respect to X. Similarly, c
is the partial
 CHAPTER 4 Regression Techniques and Demand Estimation
125

Qd = f(1 1, F0, 7)
where P is the price of the good, I is income, P0
is the price of some other good, and T
is a measure of tastes and preferences.
Economic theory can also be used to analyze the expected relationships between
the dependent variable, Qd, and the independent variables, P I, P0.
and T The law of de-
mand states that price and quantity demanded are inversely related. For normal goods,
income and quantity demanded would be positively related. With respect to the price
of the other good, the relationship should be inverse if the two goods are complements
and positive if they are substitutes. Without some additional 1nformatiu1j on the vari-
able used to measure tastes and preferences, the nature of that association cannot be
predicted.
Prior knowledge, based on economic theory, about the relationships between vari-
ables can be used to assess the empirical results of regression analysis. If the signs of the
estimated coefficients are inconsistent with the predictions of economic theory, then the
process should be reexamined. For exam,le, if the estimated coefficient of price is not
negative, it is possible that the demand equation was incorrectly formulated or that
there were errors in data collection and entr y. Other possibilities are that relevant vari-
ables were omitted from the regression equation or that there is something unique
about the good or service. At the very least, a coefficient with an unexpected sign is a
signal that the analysis should be carefully reexamined.

Data Collection
To estimate demand, data for each of the variables that influence demand first must be
obtained. These data may be collected from surveys, market experiments, or existing
sources such as historical records of the firm or government publications. Either time-
series or cross-section data may be used. Time-series data consists of period-by-period
observations in a specific market for each of the variables that affect demand. Twenty
months of data on quantity demanded, price, income, and prices of other goods from a
single city would be an example of time-series data that could be used to estimate a de-
mand function.
In contrast, cross-section data are based on a number of markets at a single point
in time. For example, cross-section data might consist of quantity demanded, income,
price, and tastes and preferences for 100 different markets during a recent year. If the
cross-section data exhibit variation from market to market, they can be used to estimate
relationships between quantity demanded and the other variables.

i.ey oncepti
-i The first step in regression anal y sis is to formuie
bsecL on econemiC.
thetr - - .,.
•• '.Economjetbeonrrn he d tovh1atethe sizdasof
coefficients.
• Time-series daii consist of obervaioiis fron. a ingle inarkef over a period of
time. Cros-secth)fl rta re badjt info from a npniber of markets at
a sin gk pn
126 PART II Demand

Choice of Functional Form

Equation (4-8) indicates a general relationship between quantity and the factors ex-
pected to influence demand. That is,
Qd = f(P, I, P0 ,fl (4-8)
However, estimation using regression analysis requires the choice of a specific func-
tional form for the equation. A linear equation is the simplest possible form, The linear
equation corresponding to the general relationships of equation (4-8) is
Qd =B4- apP -f-alI±a 0 P0 +a T T (4-9)
The linear form has several advantages. First, it can be estimated without modifi-
cation—no transformations of the data are necessary. Second, the coefficients of the
variables have a simple interpretation. 11 values of the other independent variables re-
main unchanged, each coefficient represents the change in quantity per unit change in
the associated independent variable. Furthermore, the estimated changes are constant
for each independent variable and unaffected by values of the other variables. These
properties make computations much easier. For example, it is a simple matter to fore-
cast the impact on quantity demanded of changes in per capita income. If per capita in-
come increases by $1,000 and a 1 = 0.05, the increase in uuantitv demanded will he 0.05
X 1,000, or 50 units per period.

It is also possible to calculate elasticities based on the estimated coefficients. The

formula for point elasticity of demand is

E . - -
dP Qd
But dQ d/dP = a. Thus, by selecting a value for P and using the estimated equation to
compute Qd, the elasticity for that price—quantity combination can be computed. In-
come and cross elasticities can be determined using the same general approach.
Various functional forms can be used for regression analysis.. Other than the linear
equation, probably the most common is the multiplicative functional form. For esti-
mating demand, the multiplicative equation is
Qd
= BPWaZPTaT (4-10)
In its present form, this equation cannot be estimated using ordinary least squares be-
cause it is not linear. However, there is a simple transformation of the equation that al-
lows it to be estimated using least squares. First, take the logarithm of both sides of
equation (4-10). The result is
tog(Q) = 109(BPapPiP,T4r)
But the logarithm of a product is just the sum of the logarithms. Thus,
log(Q) = log(B) + log(Pp) + log(Pi)+ log(P,) + log(T°T)
The equation can be further simplified by noting that the logarithm of a number raised
to a power is equal to that power times the logarithm of the number. Hence,
logQ = 10gB + alogP + a1 log l + a,, log+ alog T (411)
 CHAPTER 4 Regression Techniques and Demand Estimation
127
Because this equation is linear in terms of the logarithms of the original variables, the
coefficients can be estimated using the ordinary least-squares method.
The coefficients a, a1 , a, and a7 in equations (4-10) and (4-11) have an important
interpretation. Using calculus, it can easily be shown that they are the elasticities for the
respective variables. Consider the price elasticity of demand. Taking the derivative of
equation (4-10) with respect to price yields

(a,,) BP I arP T0r

dP
Multiplying both sides of the equation by PlQd gives

° - dPQ =
£
Qd
But

Qd =BPpia1PTaT
Hence
t, - -
- (4,,

Using an analogous approach, it can be shown that a1 is the income elasticity, a0 the
cross elasticity, and ar is an elasticity measure for tastes and preferences. Thus, anisad-
vantage of the multiplicative form is that it yields estimates of elasticities. Note that in
contrast to the elasticities from the linear equation, these elasticities are constant—they
are unaffected by changes in the independent variables.
Another feature of the multiplicative form is that the change in quantity demanded
per unit change in an independent Variable is not constant as it is with the linear form.
Rather, it is determined not only by the associated variable, but also by the values of the
other independent variables. This means computations using multiplicative equations
are somewhat more difficult but they may be more realistic in depicting the relation-
ship between the variables.
The choice of an appropriate functional form depends on the underlying theoreti-
cal model and the intended use of the results. If quantity demanded is thought to be a
linear function of the independent variables, a linear form would be appropriate, In con-
trast, the multiplicative form may be a better choice if the objective is to estimate elas-
ticities or to allow for nonlinear relationships between the variables.

Case Study
The Pope and the Price of Fish

For over a thousand years, the Catholic Church required that members abstain from eat-
ing meat on Fridays as an act of penance. Many Catholics responded by including fish as
part of their Friday meals. The effect of this practice was to increase the demand for fish.
128 PART H Demand

But in February 1966, Pope Paul VI authorized local bishops to end meatless Fri-
days. In December 1966, church leaders in the United States stipulated that members
could eat meat on non-Lent Fridays (usually 46 Fridays during the year). With absti-
nence no longer required on these days, it was expected that Catholics would consume
less fish and more meat. As a result, the price of fish, at least in the short run, should
have decreased.
This hypothesis was examined by economist F.W. Bell. Noting that New England is
45 percent Catholic, he collected data on fish prices for seven different species consumed
in that region before and after the change. He also collected data on other factors that
affect fish prices, such as personal income
and prices of meat and pouitiy. Using multiple-
regression techniques to hold the effects of these other factors constant. Bell was able to
assess the impact of the action by the church. His estimated equations are not reported
here because they involved techniques beyond the scope of this book. However, the find-
ings with respect to the Pope's decree are rcported in the following table.
Estimated Percent
Decrease in Price
Resulting from the
Pope's Authorization
Species to Eat Meat on Fridays
Sea scallo ps —17
Yellowtail flounder —14
Large haddock —zi
Small haddock (scrod) —2
Cod —10
Ocean perch —io
Whiting —20

Note that the percentage decrease in prices ranged from 2 percent for small haddock to
21 percent for large haddock. The average price decline for the seven species was 12.5
percent. Clearly, many Catholics enjoyed being able to have a steak on Fridays.
SOURCE: F. W. Bell, "The Pope and the Price of Fish,"
American Economic Review (December 1968),
pp. 1346-1350.

Estimation and Interpretation of Results

It has already been noted that many software packages are available for regression analy-
ses. Most require little, more than that the analyst specify the form of the equation to be
estimated and input the data. The computer then performs the necessary calculations and
provides results in a format that is usually relatively easy to interpre. Consider the esti-
mation of the demand equation for a hypothetical good. The equation to be estimated is
Qd B+aP+ all + a0P0
and the price and income variables are measured in dollars.
 CHAPTER 4 Regression Techniques and Demand Estirnailon
129

- T'ariahje
Price of
Price Income Other Good
Constant (P) (I) (Pt,)
-
Estimated coefficient 50.7836 —4.9892 0.0034 —1.2801
Standard error 10.2189 1.3458 0.0045" 0.5890
t-statistic (4.97) (-3.71) (0.76) (-2.17)
Nitniher of observations - 182 R2 - 0.6837

Table 4.5 shows the output from the demand estimation problein:rhe estimated co-
efficient of the constant term suggests the quantity demanded if values of me other vari-
ables are zero. However, this coefficient has little economic meaning in most demand
estimation problems. The other coefficients estimate the change in quantity demanded
per dollar change in the associated independent variable. For example, the price coeffi-
cient is — 4.9892. This estimate implies that if the other independent variables are held
constant, a $1 change in the price of the good will result in a change in quantity de-
manded of almost five units.
In evaluating regression results, the estimated relationships between the variables
should be considered in terms of economic theory. Note that the coefficient of price is
negative. This implies an inverse relationship between price and quantity demanded.
which is consistent with economic theory. Note also the coefficients of income and price
of the other good. The signs of these coefficients could not he predicted using economic
theory, but theory does provide information regarding their interpretation. The coeffi -
cient of income is positive, indicating that the item is a normal good. The "price of other
good" coefficient is negative. Hence the two goods must be complements.
The standard errors reported in Table 4.5 indicate the precision of the estimates. Di-
viding the estimated coefficients by their standard errors gives the t-statistics. These can
be used for hypothesis testing. A common hypothesis is that a coefficient is not signifi-
cantly different from zero. As discussed earlier in the chapter, this hypothesis can be
tested by comparing the computed 1-statistics to the values shown for the t-distribution
in Table 4.4 or in Table III on page 668.
For samples with more than about 120 observations (such as in Table 4.5), 95 per-
cent of the f-distribution lies between + 1.960 and —1.960. Thus f-statistics greater
than + 1.960 or less than — 1.960 impl y that the hypothesis that an estimated coeffi-
cient is equal to zero can be rejected withonly a 5 percent probability of error. That
B, the p
i obability of erroneously concluding that a coefficient is not equal to zero is
5 percent or less.
In Table 4.5, the I-statistics for the constant term, price, and the price of the other
good all have absolute values greater than 1.960. Thus the traditional interpretation is
that these coefficiets are significant (i.e., the probability of erroneously rejecting the
hypothesis that they are equal to zero is less than or equal to 5 percent). In contrast, the
1-statistic for the income variable t '76. meaning that there is more than a 5 percent
probability of erroneously rejecting the hypothesis that the coefficient is equal to zero.
Consequently, the income coefficient is referred to as nonsignificant.


130 PART 11 Demand

Finally, the value (if the coefficient of determination or R2 indicates the overall ex-
planatory power of the model. This number represents the proportion of total variation
in the dependent variable explained by changes in the independent variables. The R2
value from Table 4.5 is 0.6837. Thus about two-thirds of the total variation in quantity
demanded is explained by price, income, and the price of the other good.

Key Concepts
• The functional form of an equation to he estimated should be selected based on
the underlying economic theory and the intendcd use of the estimates
• The estimated parameters of a linear equation indicate the impacts of a change
• io each independent variable. -
• S m
Withoat additional coputation }w cstr i paranieter of mutipLictive
.equatiouS can he interpreted as elasiicties.
ated paraeters of an equation are evaluated by examining their
• The estimated
signs and magnj ttmd r fhe associated -v aiues are used otest hypotheses about
statistical significance.

Example Interpreting an Estimated Demand Equation

A multiplicative demand function of the form
Qd = BP°pI4rP
is estimated using cross-section data. The results are as follows:
Variable

Constant Price Income Price of Other Good

Estimated coefficient 0.02248 —0.2243 1.3458 0.1034

Standard error 0.01885 0.0563 0.5012 0.8145
(-statistic (1.19) (-3.98) (2.69) (0.13)
Number of observations = 224 R2 = 0.2515

1. How should the coefficients and the R2 value be interpreted?

2. What will the quantity demanded be if the values of the independent variables are
a. Price = $10.
b. Income per capita = $9,000.
c. Price of the other good =4 15.
3. How much would quantity demanded change if price were decreased to $8 and the
values of the other variables held constant?
4. Is demand elastic or inelastic? What effect would a price increase have on iota] revenue?
5. Are the two goods substitutes or complements?

Solution
1. Because the functional form is multiplicative, the estimated coefficients represent
elasticities. The price elasticity is —0.2243, indicating that demand is inelastic, and
the income elasticity is 1.3458, suggesting that the good is a luxui-y. Both of these
 CHAPTER 4 Regression
Thchniques and Demand Estimation
131
elasticity estimates are si gnificantly differe
of the " nt from zero. In contrast, the coefficient
other good" price is positive but not
is a weak r elationship between the two goods. significant. The implication is that there
The R2
cr reported for this problem is 0.2515. This is a rather low value, even for
oss-section data. It means that only about one-fourth of the variation in demand
can he explained by the model. Thus forecasts generated by the estimated equation
are unlikely to be very .
accurate. A possible explanation for the low R2 is that im-
portant explanatory variables have been omitted from the model.
2. In m ultiplicative form, the estimated equation is

Qd - 0.02248P'0224311.3458po.1034
A forecast of demand can be obtained by-
substituting in the given values for the in-
dependent variables. Making these substitutions, the quantit y
mated to be 3,722 units. dejiianded is esti-
3. The impact of reducing price can be"estimated by
for P recomputing quantity demanded
= $8, while holding the other variables constant. The resulting Q is 3,913. Hence
the $2 price decrease is estimated to increase quantity demanded by 191 uni ts
4. The coefficient of P
is the estimated price elasticity. Because a,,, = -1 0.2243, demand
is i nelast ic and a price IflC1'C'ISe would increase total revenue.
& "1 he coefficient of P0
is positive, so the two goods are substitutes. However, the co-
efficient of P,, is not significant.

Although regression analysis is a valuable technique for estimating demand functions and
other economic r
elationships, serious problems can occur if the analyst is not careful in the
formulation of the model and interpretation of results. Three potential pitfalls are dis-
cussed in this section_(najtted variables, the i
dentification problem, and multicollinearity
Omitted Variables
It has already been observed that economic theory can be used to specify those vari-
ables that should be included in a regression equation. However, if there are
variables
that are Omitted, the results of regression analysis can be misleading. The following ex-
ample will help to explain the i
mportance of including all the relevant variables.
It is hypothesized that the salary (5)
the number of times the player strikes outof(K) a major league baseball player depends on
during the season. It is assumed that the
more times a player strikes out, the lower his salary would be. Hence the sign of the es-
timated regression coefficient is expected to he negative.
Based on data for 150 players, the following regression equation is estimated for an-
nual salary (measured in thousands) and strikeouts in a season with the following result:
S = 484.42 + 15.54K R2 = 0.44
(-5.32) (2.51)
where the [-sta
tistics are in parentheses. Note that the coefficient on strikeouts is positive
the I distribution significant
and statistically because the t-stalistjc of 2.51 is greater than 1.96, the value of
in Table 4.4, which c
orresponds to 148 degrees of freedom. But this result
PART 11 Demand

suggests the nonsensical conclusion that players are paid more if they strike out more of-
ten! Because salary is measured in thousands of dollars, the coefficient 15.54 means that
each additional strikeout is associated with an additional $15,540 in annual salary.
The problem here is that a misspecified equation has resulted in a meaningless and
r specifi-
misleading estimate of the relationship between strikeouts and salary. A bette .
cation would have salary dependent on strikeouts and home runs (H). It is hypothesized
that strikeouts are negatively related with salaries and that the coefficient of home runs
should be positive. Using the same data on salaries and strikeouts, but adding data on
home runs, resulLs in the following multiple-regression equation:
R2 = 0.92
S 462.8 - 1.28K + 17.14H
(3.71) (-0.33) (6.44)
By including information on the number of home runs, the estiMated relationship between
strikeouts and salary changes dramatically. That relationship now is negative (as hypothe-
2
sized), although the coefficient is not significantly different from zero. However, there is
very strong relationship between salary and home runs. The estimated coefficient (17.14)
suggests that each home run is associated with an additional $17,140 in salary Further
moie, this coefficient is statistically significant because the t-statistic is greater than 1.96.
Changing the specification of the e uation resulted in an entirely different set o
conclusions about how salary is related to performance. By including data on horw
runs, it is seen that the number of strikeouts is inversely related to salary. More impor
tantly, the multiple-regression equation shows that salary is determined primarily b'
was only 0.44. mean
the number of home runs hit. Furthermore, in the first equation, R2
ing that less than one-half the variation in the salary variable is explained. By includin
increased to 0.92. The multiple-regressioi
data on both strikeouts and home runs, R2
equation that includes both strikeouts and home runs fits the data much better.
The first equation failed to consider the number of home rims. By omitting this im
portant information, the strange resãlt was obtained that suggested strikeouts result ii
higher salaries. This erroneous conclusion resulted because many home run hitters ten
to strike out quite often. Obviously, their high salaries are tied to home-run production
not strikeouts. But by estimating the relationship between salary and strikeouts, it ap
peared that baseball players receive higher salaries for striking out more often.
The more complete analysis indicates that the number of strikeouts is relatively
unimportant when the effect of home runs is considered This suggests one of the advan-
tages; Of the multipleregresSiOn approach over simple regression. In multiple regression,
the regression coefficient measures the net or partial effect of each independent variable,
while holding the effect of the other independent variables constant. In the example, the
coefficient on home runs is 17.14, which means that while holding the effect of strikeouts
each additional home run is associated with an additional $17,140 in salary.The
constant
coefficient on strikeouts is —1.28.This means that after adjusting for the number of home
runs, each additional strikeout is associated with $1,280 fewer dollars of salary.
When regression results are inconsistent with economic theory, the omission of im-
portant variables could be the cause. Consider a demand function estimated from 15
B
quarters of time series data from a particular market. The estimating equation is Qd
Assume that the estimated coefficient for price is positive arid statistically signifi-
+ aP
cant. One explanation for this anomalous result could be that as price increased over time,
so did income and the number of people in the market. Because price is positively cone-


CHAPTER 4 Regressiun Techniques and Demand Estimation

133
Price
Pnce

5,
/

I()

6H
/ //D
I / /
/
Quantity
Ouantiiy
4.2ia SrnrL^
7 14

Jated with income and with population, the coefficient of the price variable is reflecting
increases in demand caused by the population and income changes. To separately identify
these influences, additional variables should be included in the regression equation.
Identification
Related to the idea of omitted variables is the identification problem. Suppose that the
following price and quantity information has been collected from a particular market.
These data result from the interaction of supply and demand in the market. Specifically.
they are the equilibrium prices and quantities observed for each year.
Year Price Quantity
1 10 100
2 8 120
3 6 140

In Figure 4.3a, the data have been plottednd a line fitted through the three points. Note
that this line has a negative slope. If other factors that affect demand did not change over
the three-year period and the supply curve shifted to the right from S, to
shown in Figure 4.3a. then DD can he legitimately identified as a demand curve. S2 to S. as
Alternatively, Suppose that income, prices of other goods, or changes in tastes and pref-
erences caused the demand curve to shift over time. In this case. the three points represent
equilibrium prices and quantities determined by the intersection of different demand and
supply curves, as shown in Figure 4.3b. Without more information, it is impossible to choose
between the two possibilities: the separate demand curves cannot he identified.
The basic cause of this identification problem is that there is simultaneit
y between
the supply and demand equations. That is, both the quantity demanded and the quantity


134 PART 11 Demand

supplied are affected by a change in the price of the good. Thus, the effects of price on
quantity demanded and quantity supplied cannot be separately determined.
The key to identifying the true supply-and-demand relationships is to add one or
more additional variables to each equation that are not included in the other equation.
For example, assume that the problem involves the supply and demand for gasoline. The
basic demand-and-supply model is
Qd = B + diPg
Q = C + sjPg
and Q, = Qd
where P. is the price of gasoline and the equation Q = Q, is included because the ob-
served data are equilibrium quantities and prices. The quantity demanded always
equals the quantity supplied: thus there is no way to sort out the values of the parame-
ters B, d1 , C, and s 1 . Hence, neither function can be uniquely identified.
What is needed is another variable that affects only demand and one that affect3
only supply. Let income,!, be added to the demand equation and the price of crude oil.
P ' be added to the supply function. Now the three-equation model is
Qd = B + diPg + d21
Qs = C + S j Pg ± 2PC
and Q5=Qd
Note that the dependent variables in this model are quantity and price of gasoline (even
though price appears on the right-hand side of the equations). The two independent
variables are income and the price of crude oil. Together, the three equations are re-
ferred to as the structural form of the model.
Because Qd = the demand and supply equations can be combined to form
B + di Pg + d21 = C + siPg + S2Pc

Solving this equation for Pg gives

s2 -
Pg = -± ( 4-12)
d1 - s1 ,
d1 — s + { _ d1 -. S1
Substituting the expression for P. back into the original demand equation and solving
for quantity gives

Q = B + d1{ z
B}
+ t d 12 - ( 4-13)

Equations (4-12) and (443) are called reduced form equations. Note that they are
linear and that only the independent variables PC and I appear on the righthand side.
Thus, these equations can be estimated using least squares because the simultaneity
problem has been eliminated. The two equations look complicated but really are not.
When estimated, they each have a constant term and a coefficient for each variable. if
the following substitutions are made:
A=B+dj B } , andg,


CHAPTER 4 Regression Techniques and Demand Estimation

135
then equation (4-13) can be written as
Q= A + 91P + 921
The estimated parameters A, g1 , and g2
structural equations. For example, shows combine the effects of the coefficients from the
91 the net effect of a change in the price of
crude oil on the e
preted in a similar quilibrium
manner, quantity demanded of gasoline, Equation (4-12 is inter-
When one variable on the right side of a structural equation is
sim uItaneousj y de-
termined with the variable on the left side, one approach is to determine the reduced
form equations of the model and use those equations for
rcgrcssj oj-i audvsis. This
method may be appropriate for estimating demand functions,
Flowever, in actual practice, the sim
ultaneity between quantity and price often is ig-
nored, and it is assumed that the price of the good is an independent variable. In this
case, an eqwution of the general form
Qd f(TI,P,)
is used to estimate the demand function. Any shifts in demand are captured b
y includ-
ing the income and other goods price variables. This approach is easier than working
with a simultaneous equation model, but may yield less accurate results,
Multicollinear;ty
The omitted variables and the identification problem discussed thus far have both in-
volved the need to add additional variables to the regression equation. But there are
situations where the problem is that there are too many variables. An important exam-
ple is when two or more i
referred to as ndependent variables are highly correlated. This problem is
mulsjco/jjneam,
Consider a student who has just completed a difficult course in American
literature.
So many books were assigned in the class that no one was able to finish all of the read-
ing. For an assignment in a statistics course, the student collected data on the grade per-
formance and study habits of a random sample of 40 Students who took the
class. His hypothesis was that Course grades should be Positively related to theliterature
number
of hours spent studying for the course and the amount of reading each person com-
pleted. Applying regression analysis to the data, the equation was estimated to be
G = 50.00 ± 0.4011 + 0.02P R2 0.80
(2.80) (0.80) (1.35)
where G is the percentage grade, H
read. The t-stat istics are in p ar is hours of study time, and P is total number of pages
entheses Note that the coefficients of both H and P are
Positive, but neither is St atistically Significant, However, the R2 is very high for cross-
sectional data, suggesting the equation should be a good predictor of grades.
The problem with the analysis is that hours spent studying for the literature class
and pages read are likely to be highly corr
course, t elated. The more time students devote to the
be nearlyhel more pages they will be able to read. In fact, the relationship probabl y will
inear,That is, twice as many hours will probably result in about twice as many
pages completed
Multic ollinearity causes problems in regression anal
y sis. When two variables are
highly correlated it is difficult to identify the 'unique influence that each variable has


136 PART II Demand

the dependent variable. For the example, more hours spent studying mean a better
grade, but more study hours also allow more pages to be read, and this also should re-
suit in a better grade.
When there is multicollinearity, the standard errors of the coefficients tend to be
large and, hence, the t-statistics will be small. Consequently, the coefficients are less
Likely to be statistically significant. One solution to the problem of multicollinearity is
to remove one of the highly correlated variables from the equation.
In the example, suppose that hours studied is dropped from the model. When the
new equation is estimated, the results are
en fin .
ou.'iu
n2
-r 0.03P 1 - U.!)
,, -'C.

(2.70) (3.00)
Note that the revised equation still has a high R2 and the coefficient of P is positive and
statistically significant.
Multicollinearity is not always easy to detect. If two variables are almost perfectly
correlated, most regression programs will indicate that they are unable to perform the
analysis. But in other cases, there may be no error message, and multicollinearity may
go unnoticed. One approach is to check for pairs of independent variables that are
highly correlated.

Key. Concepts ..
A misspecified re.gression equation can lead to erroneous tin
cclusons about the
• relationship between variahks.
• There may be an identification problem if price and quantity iTC simultane-
ously deiermined by supply and demand. The problem can be solved by
cluding additional variables and using reduced-form estimation.
• Multicoilinearity results jEt large stdard errors and small r-statistics for the p47
timated coefficients.

SUMMARY
The least-squares regression technique can be used to quantify the relationship be-
tween a dependent variable and one or more independent variables. In simple regres-
sion, there is one independent variable, whereas with multiple regression, there are two
or more independent variables. The estimated coefficients in a regression equation
measure the change in the value of the dependent variable for each one-unit change in
the independent variable, holding the other independent variables constant.
The coefficient of determination (R2) is used to test the explanatory power of the
entire regression equation. This statistic measures the proportion of the total varia-
tion in the dependent variabic that is explained by variations in the independent vari-
ables. Hypotheses regarding the coefficients of individual independent variables are
tested using the t-statistic, which is computed by dividing the estimated coefficient by
its standard error. If the absolute value of this ratio is greater than the value taken
from a table of the student's distribution, the coefficient is said to be statistically sig-
 CHAPTER 4 Regression Techniques and Demand Estimation
137
nificant. Regression equations can be used to make both point and interval estimates
of the predicted value of the dependent variable for given values of the independent
variables.
The first step in using regression analysis is to develop a theoretical model based on
economic theory. This model can specify the data that should be collected and can also be
used to interpret results. Regression analysis can be applied to either cross-section or time-
series data. Linear estimating equations have the advantage of simplicity, but if a multi -
plicative equation is used to estimate the demand function, its coefficients are elasticities.
If relevant variables are excluded from a regression equation, the equation is said
to be miss pecified. The use of a misspecified equation may lead to incor
rect conclusions
about the relationships between the dependent and independent variables. Where price
and quantity are simultaneously determined by supply and demand, it may be difficult
to identify the demand function. This problem can be resolved by adding explanatory
variables and using a reduced-form equation. When two or more independent variables
are highly correlated, this multicollinearity may result in large standard errors and,
hence, small t-statistics for individual coefficients.
Discussion Questions
4-1, It is sometimes argued that quantitative techniques, such as multiple regression
analysis, are of little value because they are always subject to error. Is this a valid
argument?
4. 2. A demand equation is estiMated, but the coefficient of income is not significant.
In using this equation to predict quantity demanded, should the income variable
be omitted? Explain.
4-3. Why would coefficients of determination tend to be higher for time-series than
for cross-section data?
44. What is the relationship between the sign of the estimated coefficient and the sign
of the associated i-statistic? Explain.
4-5. In a demand function, what is the economic interpretation of a statistically sig-
nificant coefficient for the price variable?
4-6. In transforming a multiplicative equation to a linear form, what would be the ef-
fect on the estimated coefficient of the price variable of using natural logarithms
rather than logarithms to the base 10?
4-7. Data on average price and quantity over the last 12 months are used to plot a
graph that is claimed to represent a demand function. Is this a valid interpreta-
tion? Explain.
4-8. Adding more independent variables to an estimating equation tends to increase
the R2. What problems could result from adding variables?
4-9. Is there any basis for choosing between a linear and a multiplicative equation for
estimating a demand function? Explain.
Problems

TT1T
4-1. Consider the following five data points:


138 PART H Demand

a. Use regression analysis to calculate by hand the estimated coefficients of the

equation Y B + aX.
b. Compute the standard error and the i-statistic for the coefficient of X.
4-2. Consider the following five data points:

X —1.0 0,0 1.0 2.0 3.0

Y —1.0 1.0 1.0 2.5 3.5

a. Use regression analysis to calculate by hand the estimated coefficients of the

equation Y = B + aX
D. Compute the coefficient of determination.
c. What is the predicted value of Y for X = 1.0? For X = 15?
4-3. The following regression equation estimates the relationship between the num-
ber of cups of hot chocolate sold (H) and number of s'iimmers (N) at a beach:
H=252.8-2.05N R2=0.45
(2.06) (-3.05)
(t-values are shown in parentheses.)
a. Explain or interpret the regression coefficient of N, the t-value, and the coef-
ficient of determination of this equation.
b. How is it possible that more hot chocolate is sold when there are fewer peo-
ple at the beach? Does this relationship suggest anything about the specifica-
tion of the equation?
'The vice-president of United Feeds, Inc., has provided the following quarterly
price —quantity data for UF Superb, a horse feed additive. He has asked that the
demand equation for this product be estimated.
Quarter
3 6 7 89 10
2
53 43 40 47 57 41 53 37 51
Price ($) 60
100 108 97 80 105 86 110 90
Quantity 83 93
/ a. Use the ordinary leastsquareSregreSSi011 method to estimate this function.
. The firm is considering a price increase to $66. Make a point and 95 percent
confidence interval estimate of sales volume if this price increase is carried out.
' 4-5. Annual prices and beef consumption per capita in six cities are as follows:
Price Consumption
City per Pound per Capita
1 $2.00 55
2 1.90 60
3 2.10 50
4 1.80 70
5 2.30 45
6 2.20 48


CHAPTER 4 Regression Techniques and Demand Estimation

139
a.
If a demand equation is to be estimated using these data, would the linear
form (i.e., Q D + a?) or the m
ultiplicative form (i.e., Q BP) be more ap-
propriate? Explain
b.
Could the resulting equation be properly interpreted as a demand function?
Explain.
4-6. The MacWend Drive-In has determined that demand for hamburgers is giveii by
the following equation:
Q 205.2 + 210A 200. OPM
+ 100.OP + 0.51
(1,85) (2.64) (—.5 61) (2.0)) (4.25)
where Q is the number of hamburgers sold per month (in 1,000s), A
vertising expenditures during the previous month (in $1,000s) , is the ad-
p,
Mac Wend burgers (dollars), P- is the price of hamburgers of the company'sis the price of
m a-
jot' Competitor (dollars), and I is income per capita in the surrounding
nity (in $ 1 ,0 commu-
(ft)The t-statistic fo each coefficient is shown in parentheses below
each coefficient
Are'the-signs of the individual coefficients consistent with predictions from
economic theory? Explain.
b. If A $5.000. PM $!, P
= $1.20, and I $20,000. how many hamburgers
will be demanded?
c. What is the advertising elasticity at
4-7. M A $5,000?
otorland Recreational Vehicles estimates the monthly demand (Q) for its prod-
uct is given by the equation
log Q = 1.00 - 1,50 log P + 3.00 log
1 R2 = 0.21
(1.20) (--2.50) (0.02)
where P is price and I
is income per capita in thousands. The t-statistics are shown
in parentheses and logarithms to the base 10 were used to transform the equa-
tion. Assume that estimates are generated by a sample of 400 observations.
a. Rewrite the expression as a m
ultiplicative demand equation.
b.
Based on the equation, is the product an inferior good, a necessity, or a luxury
good? How much confidence do you have in your answer? Explain.
c.
Is the equation likely to be useful in predicting demand for Motorland's prod-
uct? Why or why not?
4-8. Data from 20 cities were used to estimate the demand for face-lifts 1
regression equation was Fhe resulting

50.000 O.00IP + 0.0021 R2 0.55

(5.42) (-2.34) (2.00)
where Qd is face-lifts per 1,000 population per year,
P is the price in dollars, I is
income in dollars, and the i-statistics are shown in parentheses.
a.
In determining statistical significance of the coefficients, what number should
be used for degrees of freedom?
b.
Which of the coefficients are statistically significant at the 5 percent level?
140 PART II Demand

c. The mean values of P and I are $5,000 and $20,000, respectively. Compute the
point price elasticity of demand.
d. What is the predicted demand at the mean values of the independent variables?
4-9. A demand equation of the form Q = BP lb is estimated, and the coefficient of
price is —2 with a standard error of 0.8.The coefficient of income is 3 with a stan-
dard error of 2.0. The estimate is based on a sample size of 43,
a. How many degrees of freedom should be used in determining thc critical value
of the i-statistic? What is the table value for the 95 percent confidence level?
b. Which of the coefficients are statistically significant at the 95 percent confi-
dence level? Explain.
4-10. For a hypothesis test of the statistical significance of coefficients, what would be
the degrees of freedom for each of the following equations? What is the table
value of the i-statistic for the 95 percent confidence level?
a. Q BPa1P c and 11 observations?
b. Y = A + bX + cZ and 63 observations?
c. Y = A + bX + cZ and 200 observations?

Problems Requiring Calculus

4.11. The demand function for a product is estimated to be
Qd = l0 '°j20p°0
and the mean values for P and I are $150 and $18,000, respectively.
a. Predict the value for the dependent variables at the mean values of the inde-
pendent variables.
b. At the mean values, how much will a dollar change in price change the quan-
tity demanded? What about a dollar change in income?
4-12. Economists on a skiing vacation in Aspen, Colorado, decide to study the demand
for lift tickets in the area. Usingmultiple regression analysis, the demand equa-
tion is estimated to be
log Q1 3.00 - 2.0 log P ± 0.5 log S
where Qt is quantity demanded per pay, P is the price of a full-day lift ticket, and
S is accumulated snow depth in inches at the lodges. Using calculus, show that for
small changes, a 1 percent increase in accumulated snow depth will increase daily
lift ticket demand by 0.5 percent.
443. Starting with the general demand equation,

Q bPPaPPaI
demonstrate that a, is the income elasticity and a 0 is the cross elasticity.

Computer Problems
The following problems can be solved by using the TOOLS program (downloadable
from www.prenhall.com/petersen) or by using other computer software.
4-14. Data on income (in thousands of dollars). education (years), experience (years),
and age (years) for twenty people are shown here.
 CHAPTER 4 Regression Techniques and Demand Estimation
141
Person Job.
Income Age Education Experience
1 5.0 29 2 9
2 9.7 36 4 18
3 28.4 41 8 21
4 8.8 30 8 12
5 21.0 34 8 14
6 26.6 36 10 i6
7 25.4 61 12
8 23.1 29 12 9
9 22.5 54 12 18
10 19.5 30 12 5
11 21.7 28 12 7
12 24.8 29 13 9
13 30.1 35 14 12
14 24.8 59 - 14 17
15 28.5 65 15 19
16 26.0 30 15 6
17 38.9 40 16 17
18 22,1 23 16 1
19 33.1 58 17 10
20 48.3 / 60 21 17
a. Use multiple regression analysis to estimate income as a linear function of age.
Write the equation, i-statistics, and the coefficient of determination, Provide
an explanation for the sign of the age coefficient.
b. Use regression analysis to estimate income as a linear function of education,
job experience, and age. Write the equation, (-statistics, and the coefficient of
d etermination. How do the results of part (b) explain the results from part (a)?
c. Use the results from part (b) to estimate income of a typical person who has
14 years of education, has 10 years of job experience, and is 45 years of age.
4-15. Data on grade point average and IQ were obtained for 12 high school students.
Grade Point Average - !Q Grade Point Average IQ
2.1 116 2.9 126
2.2 129 2.7 122
3.1 123 ' 2.1 114
23. i21 1.7 109
3.4 131 3.3 132
2,9 134 3.5 140
a. tise regression analysis to estimate the effect of 1Q on grade point average.
Write the equation, t-statistics, and the coefficient of determination. Is the re-
suit consistent with your prior expectations? Explain.
b. Forecast the grade point average for a student with an 10 of 120 and for
a student with an 10 of 150. Which forecast do you have more confidence
in? Why?


142 PART II Demand

4-16. Data on electric power consumption (in billions of kilowatt-hours), GNP (in bil-
lions of dollars), and electricity prices (in cents per kilowatt-hour) for the period
1969-1983 are shown here.
Year Consumption GNP Price

1969 407.9 $ 9440 2.09

1970 447.8 992.7 2.10
1971 479.1 1,077.6 2.19
1972 511.4 1,185.9 2.29
iY/i JJ'+.h i,.)hO.'+ L..)O

1974 555.0 1,434.2 2.83

1975 586.1 . 1,594.2 3.21
1976 613.1 1,718.0 .45
1977 652.3 1,918.3 3.78
1978 679.2 2,163.9 4.03
1979 696.0 2,417.8 4.43
1980 734.4 2,631.7 5.12
1981 730.5 2,957.8 5.80
1982. 732.7 3069. 6.44
1983 750.9 3,304.8

a. Using regression analysis, estimate consumption as a linear function of GNP, price,

and the previous year's electricity consumption. (NcrrE: assume 1968 consump-
tion was $367.7 billion.) Write the equation, t-statistics, and the coefficient of de-
termination. Are the signs of the estimated coefficients consistent with economic
theory? Which of the coefficients are statistically significant at the 0.05 level?
b. In 1984, GNP was $3661.3 billion and the price of electricity was 7.16 cents per
kilowatt-hour. Use the estimating equation from part (a) to predict electricity
consumption for 1984.
4-17. Consumption of hamburgers (thousands of burgers per week) in 12 different
cities is shown here. Prices of hamburgers, income per capita (in $1,000s), and
prices of hot dogs for the cities are also shown.
Hamburger Hamburger income Hot Dog
City Consumption Price ($1,0004 Price
1 50 $1.50 12.0 $1.80
2 80 1.35 14.2 1.55
3 95 1.25 15.0 1.45
4 105 1.20 16.0 1.35
5 70 1.40 13.8 1.60
6 85 130 14.3 1.50
7 55 1.50 13.3 1.70
8 60 1.45 13.3 1.70
9 75 1.35 13.7 1.60
10 90 1.25 14.5 1.50
11 100 1.20 15.2 1.35
12 65 1.45 13.6 1.65
 CHAPTER 4 Regression Techniques and Demand Estimation
143
a. Use regression analysis to estimate hamburger consumption as a multiplica-
tive function of the price of hamburgers, income, and hot dog price. Write the
equation, t-statistics, and the coefficient of determination. Which coefficients
are significant at the 0.05 level?
b. Based on the estimates from part (a), what are the price, income, and cross
elasticities? Is the cross elasticity consistent with economic theory? Explain.
 CHAPTER

Busine riomic
F

U Preview
• Sources of Data
Expert Opinion
Surveys
Market Experiments
• Tune-Series Analysis
Trend Projection
Exponential Smoothing
U Barometric Forecasting
Leading Indicators
Composite and Diffusion Indices
• Input/Output Analysis
Transactions Matrix
Direct Requirements Matrix
Direct and Indirect Requirements Matrix
Forecasting with an InpuJOutput Model
U Summary
U Discussion Questions
• Problems

144
 CHAPTER 5 Business and Economic Forecasting 145

PREVIEW
The vast majority of business decisions involve some degree of uncertainty—managers
seldom know exactly what the outcomes of their choices will be. One approach to re-
ducing the uncertainty associated with decision making is to devote resources to fore-
casting Forecasting involves predicting future economic conditions and assessing their
effect on the operations of the firm.
Frequently, the objective-of forecasting is to predict demand. In some cases, man-
agers are intreted in the total demand for a product. For examp!e, the decision by an
office products firm to enter the home computer market may be determined by esti -
mates of industry sales growth. In other circumstances, the projection may focus on the
firm's probable market share. If a forecast suggests that sales growth by existing firms
will makc successful entry unlikely, the company may decide Lu look for other areas in
which to expand.
Forecasts can also provide informatibn on the proper product mix. For an automo-
bile manufacturer such as General Motors, managers must determine the number of
full-sized versus compact cars to be produced. In the short run, this decision is largely
constrained by the firm's existing production facilities for producing each kind of car.
However, over a Jongcr period, managers can build or modify podiictioii facilities. But
such choices must be made long before the vehicles begin coming off the assembly line.
Accurate forecasts can reduce the uncertainty caused by this long lead time. For exam-
ple, if the price of gasoline is expected to increase, the relative demand for compact cars
is also likely to increase. Conversely, a projection of stable or falling gasoline prices
might stimulate demand for larger cars.
Forecasting is an important management activity. Major decisions in large busi-
nesses are almost always based on forecasts of some type. In some cases, the forecast
may be little more than an intuitive assessment of the future by those involved in the
decision. In other circumstances, the forecast may have required thousands of work
hours and tens of thousands of dollars. It may have been generated by the firm's own
economists, provided by consultants specializing in forecasting, or be based on infor-
mation provided by government agencies.
This chapter focuses on some basic techniques of forecasting. The first section con-
siders various methods for collecting the necessary data. Next is a discussion of time-
series analysis. The third section considers barometric forecasting. The basic principles
of input/output analysis are presented in the last section of the chapter.

SOURCES OF DATA
Forecasting requires the development of a good set of data on which to base the analy-
sis. A forecast cannot be better than the data from which it is derived. Three important
sources of data used in forecasting are expert opinion, surveys, and market experiments.

Expert Opinion
The collective judgment of knowledgeable persons can be an important source of in-
formation. In fact, some forecasts are made almost entirely on the basis of the per-
sonal insights of key decision makers. This process may involve managers conferring
146 PART II Demand

to develop projections based on their assessment of the economic conditions facing

the firm. In other circumstances, the company's sales personnel may be asked to eval-
uate future prospects. In still other cases, consultants may be employed to develop
forecasts based on their knowledge of the industry. Although predictions by experts
are not always the product of "hard data," their usefulness should not be underesti-
mated Indeed, the insights of those closely connected with an industry can be of great
value in forecasting.
Methods exist for enhancing the value of information elicited from experts. One of
the most useful is the Delphi technique. Its use can be illustrated by a simple example.
Suppose that a panel of six outside experts is asked to forecast a firm's sales for the next
year. Working independently, two panel members forecast an 8 percent increase, three
members predict a 5 percent increase, and one person predicts no increase in sales.
Based on the responses of the other individuals, each expert is then asked to make a re-
vised sales forecast. Some of those expecting rapid sales growth may, based on the judg-
ments of their peers, present less optimistic forecasts in the second iteration. Con-
versely, some of those predicting slow growth may adjust their responses upward.
However, there may also be some panel members who decide that no adjustment of
their initial forecast is warranted.
Assume that a second set of predictions by the panel incii.ides one estimate of a 2
percent sales increase, one of 5 percent, two of 6 percent, and , two of 7 percent. The ex-
perts again are shown each other's responses and asked to consider their forecasts fur-
ther. This process continues until a consensus is reached or until further iterations gen-
erate little or no change in sales estimates.
The value of the Delphi technique is that it aids individual panel members in as-
sessing their forecasts. Implicitly, they are forced to consider why their judgment differs
from that of other experts. Ideally, this evaluation process should generate more precise
forecasts with each iteration.
• One problem with the Delphi method can be its expense. The usefulness of expert
opinion depends on the skill and insight-of the experts employed to make predictions.
Frequently, the most knowledgeable people in an industry are in a position to command
large fees for their work as consultants;Or they may be employed by the firm, but have
other important responsibilities, which means that there can be a significant opportu-
nity cost in involving them in the planning process. Another potential problem is that
those who consider themselves experts may be unwilling to be influenced by the pre-
dictions of others on the panel. As a result, there may be few changes in subsequent
rounds of forecasts. -

Surveys
Surveys of managerial plans can be an important source of data for forecasting. The ra-
tionale for conducting such surveys is that plans generally form the basis for future ac-
tions. For example, capital expenditure budgets for large corporations are usually
planned well in advance. Thus, a survey of investment plans by such corporations should
provide a reasonably accurate forecast of future demand for capital goods.
Several private and government organizations conduct periodic surveys of plant
and equipment expenditure plans. One of the most widely used is sponsored by Mc-
Graw-Hill, Inc. This survey is conducted twice yearly and includes corporations ac-
 CHAPTER 5 Business and Economic Forecasting
147
counting for more than 50
percent of total investment in the U.S. economy. An even
more comprehensive survey of capital expenditure plans is undertaken quarterly by the
U.S. Department of Commerce. The results of this survey are reported in the depart-
ment's Survey of Current Business.
Useful data for forecasting can also be obtained from surveys of consumer plans.
For example, the Survey Research Center at the University of Michigan polls con-
sumers about their intentions to purchase specific products such as household appli-
ances, housing, and automobiles. The results are used to project consumer demand and
also to measure the level of consumer confidence in the economy. The US. Burcau of
the Census also conducts surveys of consumer intentions.
If data from existing sources do not meet its specific needs, a firm may conduct its
own survey. Perhaps the most common example involves companies that are consider-
ing a new product or making a substantial change in an existing product. But with new
or modified products, there are no data on which to base a forecast. One possibility is
to survey households regarding their anticipated demand for the product. Typically,
such surveys attempt to ascertain the demographic characteristics (e.g., age, education,
and income) of those who are most likely to buy the product and find how their deci-
sions would be affected by different pricing policies.
Although surveys of consumer demand can providc useful data for forecasting, their
value is highly dependent on the skills of their originators. Meaningful surveys require
careful attention to each phase 91 the process. Questions must be precisely worded to
avoid ambiguity. The survey sample must be properly selected so that responses will be
representative of all customers. Finally, the methods of survey administration should
produce a high response rate and avoid biasing the answers of those surveyed. Poorly
phrased questions or a nonrandom sample may result in data that are of little value.
Even the most carefully designed surveys do not always predict consumer demand
with great accuracy. In some cases, respondents do not have enough information to de-
termine if they would purchase a product. In other situations, those surveyed may be
pressed for time and be unwilling to devote much thought to their answers. Sometimes
the response may reflect a desire (either conscious or Unconscious) to put oneself in a
favorable light or to gain approval from those conducting the survey. Because of these
limitations, forecasts seldom rely entirely on results of consumer surveys. Rather, these
data are considered supplemental sources of information for decision making.

Gade Study
The Use and Abuse of Survey Data

For years, firms have used consumer surveys to collect data on the demand for
their products. A more recent trend has been the use of survey data to promote those
products to potential customers. But where advocacy and persuasion are the goals,
there may be a tendency Co
stretch the limits of good survey design and interpretation.
In some cases,'Ehe problem with a survey is that the questions have the effect of
biasing the answers in a way that favors the products of the firm. A study sponsored by
148 PART II Demana

Levi Strauss asked college students to select which clothes they thought would be most
popular during the coming year.The company announced that Levi's 501s were chosen
as the most popular jeans. What the public was not told was that 501s were the only
jeans on the list.
A Black Flag survey stated that, "A roach disk . . . poisons a roach slowly. The dy-
ing roach returns to the nest and after it dies is eaten by other roaches. In turn, these
roaches become poisoned and die." Survey respondents were then asked, "How effec-
tive do you think this type of product would be in killing roaches?" Provided with this
"helpful" information, 79 percent said that the disk would be effective.
A sample that is not random can be used to generate survey results that are suspect.
A Chrysler study showed that its can were preferred to those of Toyota. However, none of
the people in the sample owned a foreign car, suggesting that they may have been predis-
posed to buy U.S. automobiles. A survey sponsored by American F'cpress and the French
government concluded that it was untrue that the French are unfriendly. But the sample
consisted of Americans who had visited France for pleasure more than once during the last
two years—presumably people who have positive feelings about vacationing in France.
Sample bias was the cause of what is generally considered to be the most inaccurate
political poll in U.S. history. in 1936, the Literary Digest predicted that Republican Alf Lan-
don would be a big winner over Franklin D. Roosevelt in the presidential election of that
year. But the sample consisted of those who had telephones, cars, or were subscribers to
the magazine—all characteristics of high-income voters at that time. History records that
Roosevelt was reelected to a second term by a lapdslide vote of the general population.
The selective use of data is not confined to Western society. In China, a popula-
tion census determined that the population of one province was 28 million. Five years
later, the same province was found to have 105 million people. An astounding birth rate
was not the cause. Rather, the first census was used for military conscription (hence, a
small number was better) and the secord was the basis for government aid to the vic-
tims of a famine in the regions (now a large estimated population was beneficial). m

Market Experiments
A potential problem with survey data is that survey responses may not translate into
actual consumer behavior. That is, consumers do not necessarily do what they say they
are going to do. This weakness can be partially overcome by the use of market experi-
ments designed to generate data prior to the full-scale introduction of a product or im-
plementation of a policy.
To set up a market experiment, the firm first selects a test market. This market may
consist of several cities, a region of the country, or a sample of consumers taken from a
mailing list. Once the market has been selected, the experiment may incorporate a num-
ber of features. It may involve evaluating consumer perceptions of a new product in the
test market. In other cases, different prices for an existing product might be set in vari-
ous cities inorder to determine demand elasticity. A third possibility would he a test of
consumer reaction to a new advertising campaign.

CHAFFER 5 J3usine.ss and Economic Forecasting 149

There are several factors that managers should consider in selecting a test market.
First, the location should be of manageable size. If the area is too large, it may be ex-
pensive and difficult to conduct the experiment and to analyze the data. Second, the res-
idents cf the test market should resemble the overall population of the United States
in age, education, and income. If not, the results may not be applicable to other areas.
Finally, it should be possible to purchase advertising that is directed only to those who
are being tested.
Market experiments have an advantage over surveys in that they reflect actual con-
sumer behavior, but they still have limitations. One problem is the risk involved. In test
markets where prices are incrcascd, consumers way switch to products of competitors.
Once the experiment has ended and the price reduced to its original level, it may be dif-
ficult to regain those customers. Another problem is that the firm cannot control all the
factors that affect demand. The results of some market experiments can- be influenced
V y bad weathei, changing economic conditions, or the tactics of competitors. Finally, be-

cause most experiments are of relatively short duration, consumers may not be com-
pletely aware of pricing or advertising changes.Thus their responses may understate the
probable impact of those changes.

Key cern . .-:

- :-.............. .
. -.............................
• The Delphi technique is an iterative teihniqie thai can be us IC; ih..tnc
value of expert opinion.
H,
Survvsofco nsunIcrintentions re careful
reful attention to wordii ,sarnpk se---
- lection, and methods of administratkm. .
• Market ex erirnentpro r ide actual thita on cor.suwjo.
njer heha HoeveT,w -
the y are expensive, rky. and ityb nf!uenccd by factors beyoni the firm's
control
...........p. - .......... ................-

TIME-SERIES ANALYSIS
Regression analysis, as described in chapter 4, can be used to quantify relationships be-
tween variables. However, data collection can be a problem if the regression model in-
cludes a large number of independent variables. When changes in a variable show dis-
cernable patterns over time, time-series analysis is an alternative method for forecasting
future values.
The focus of time-series analysis is to identify the components of change in the data.
Traditiona Fly, these components are divided into four categoi iCs:
L Trend
2. Seasonality
3. Cyclical patterns
4. Random fiuctuationg
A trend is a long-term increase or decrease in the variable. For example, the time series
of human population in the United States exhibits an upward trend, while the trend for
endangered species, such as the tiger, is downward. The seasonal component represents


150 PART TI Demand

changes that occur at regular intervals. A large increase in sales of skis in the fall and
early winter would be an example of seasonality.
Analysis of a time series may suggest that there are cyclical patterns, defined as sus-
tained periods of high values followed by low values. Business cycles fit this category.
Finally, the remaining variation in a variable that does not follow any discernable pat-
tern is due to random fluctuations. Various methods can be used to determine trends,
seasonality, and any cyclical patterns in time-series data. However, by definition,
changes in the variable due to random factors are not predictable. The larger the ran-
dom component of a time series, the less accurate the forecasts based on those data.

Trend Projection
One of the most commonly used forecasting techniques is trend projection. As the
name suggests, this approach is based on the assumption that' there is an identifiable
trend in a time series of data. Trend projection can also be used as the starting point for
identifying seasonal and cyclical variations.
Table 5.1 is a time series of a firm's quarterly sales over a three-year time span.
These data will be used to illustrate graphical and statistical trend projection and also
to describe a method for making seasonal adjustments to a forecast.
Grapie Curve Fitting Note that the data show generally increasing sales quarter
by quarter. But a useful forecast usually requires greater precision than is implied by
the statement "generally increasing sales." To be of value in forecasting, a numerical es-
timateof the increase in sales per quarter must be made. One way to make this estimate
is to fit a line to the data graphically. In Figure 5.1, a straight line is drawn through the
data points in such a way as to reflect the trend of the data as accurately as possible.
The upward slope of the line reflects sales increases rover time. By computing the
slope of this trend line, it is possible to determine the average rite of increase per quar-
ter. This value can be used to calculate Vies in future periods. Alternatively, by extend-
ing the trend line beyond the last data point (i.e., the fourth quarter of 1998), the esti-

.' i .I•.ti Salei

Period .\ uwliet Quarter Sales (inililuas)

1 1996:1 $300
2 1996:11 305
3 1996:111 315
4 1996:IV 340
5 1997:1 346
6 1997:11 352
7 1997:111 364
8 1997:IV 390
9 1998:1 397
10 1998:11 404
11 998:111 41S
12 199:1V 445
 CHAPTER 5 Business and Economic Forecasting 151
Saks per
quarter (Sr)
(million $)

50C

48( •0'
-

ME

300

Quarter (t)
1 11 1 1 1
1996 1 1997 1 1998 I 1999

mated sales can be read directly from the graph. Suppose that sales are to be forecast
for the third quarter of 1999. Based on an extrapolation of the data to that period, the
graph indicates that sales are projected to be about $480 million.
Statistical Curve Fitting One limitation of graphical curve fitting should be obvious—
the accuracy of forecast depends on the analyst's ability to fit a curve to the data. A more
so phisticated approach is to use statistical methods to fit the data to an equation of spec-
ified functional form. Basically, this involves using the ordinary least-squares concept
developed in chapter 4 to estimate the parameters of the equation.
CONSTANT RATE OF CHANGE Suppose that an analyst determines that a fore-
cast will be made assuming that there will be a constant rate of change in sales from one


152 PART II Demand

period to the next. That is, the firm's sales will change by the same amount between two
periods. The time-series data of Table 5.1 are to be used to estimate that rate of change.
Statistically, this involves estimating the parameters of the equation
S = S0 + In (5.1
where S denotes sales and t indicates the time period. The two parameters to be esti-
mated are S0 and b. The value of S0 corresponds to the sales (vertical) axis intercept of
the line in Figure 5.l.The parameter b is the constant rate of change and corresponds
to the slope of the line in Figure 5.1.
Many hand calculators can estimate the parameters of equation (5-1). Specific pro-
cedures vary from model to model, but usually the only requirement is that the user in-
put the data and push one or two designated keys.The machine then returns the estimated
parameters. For the data of Table 5. 1, the quarters would have t6 be inputted as sequen-
tial numbers starting with 1. That is, 1996:1 would be entered as 1.1996:11 would he. en-
tered as 2, and so forth. Based on the data from the table, equation (5-1) is estimated as
S, = 281.394 + 12.8111
The interpretation of the equation is that the estimated constant rate of increase in sales
per quarter is $12.811 million. A forecast of sales for any future quarter, S, can ho '-
tamed by substituting in the appropriate value for z. For example, the third quarter of
1999 is the 15th observation of the time series. Thus, the estimated sales for that quar-
ter would be 281.394 + 12.811(15), or $473.56 million.
CONSTANT PERCENTAGE RATE OF CHANGE Now suppose that the individual
responsible for the forecast wants to estimate a percentage rate of change in sales. That
is, it is assumed that sales will increase by a constant percent each period. This relation-
ship can be expressed mathematically as
S=$ 0 ± g)
Similarly,
S1 = S_ 2(1 + g)
where g is the constant percentage rate of change, or the growth rate. These two equa-
tions imply that
S1 = Sr_ 2 (1 + g)2
and, in general,
S1 = S0(1 + 9)1 (5-2)
As shown, the pat ameers of equation (5-2) cannot be estimated using ordinary
least squares. The problem is that the equation is not linear. However, there is a simple
transformation of the equation that allows it to be estimated using ordinary least
squares.
First, take the natural logarithm of equation (5-2).' The result is

'Either natural logarithms or logarithms to the base 10 can be used. In this section, natural logarithms
(designated as In) are used for all computations.
 CHAPTER 5 Business and Economic Forecasting 153

In S. = ln[S,(1 +
But the logarithm of a product is just the sum of the logarithms. Thus
In S 1 = In S,, + ]n[(l + g)1]
The right -hand side of the equation can be further simplified by noting that
In[(l + g] = t[ln( 1 + g))
Hence
In S = In S ) + / l In ( i + g)J (5-3)
Equation (5-3) is linear in form. This can be seen by making the following
substitutions:
In S
= to
b Jn(1 ± g)
Thus the new equation is
= + bt
k'-,--, /
which is linear.
The parameters of equation ('54) can easily he estimated using a hand calculator.
The key is to recognize that the sales data have been translated into logarithms. Thus,
instead of S,, it is In S that must be entered as data. However, note that the (values have
not been transformed. Hence for the first quarter of 1996, the data to be entered are In
300 = 5.704 and I: for the second quarter, in 305 = 5.720 and 2; and so forth. The trans-
formed data are provided in Table 5.2.

TABLE S.2 Nat Lo .hhms odj'

Period Natural Logarithm of

Number Sales (in millions)
(r) Quarter (5,)
1996:1 5.704
2 1996:11 5.720
3 1996:111 5.753
4 1996:IV 5.829
5 1997:1 :1847
6 1997:11 5.864
7 1997:111 5.897
8 i997.j\•
9 1998:1
10 1998:11
11 1998:111
12 1998-IV 6.098
154 PART II Demand

Using the ordinary least-squares method, the estimated parameters of equation (5-4),
based on the data from Table 5.2 are
= 5.6623 + 0.0353r
But these parameters are generated from the logarithms of the data.Thus, for interpreta-
tion in terms of the original data, they must be converted based on the relationships
In S0 V0 = 5.6623 and ln(1 + g) = b = 0.0353. Taking the antilogs yields S 0 = 287.810
and I + g = 1.0359. Substituting these values for S0 and 1 + g back into equation (5-2) gives
S = 287.810(l.0359)r
where 287.810 is sales (in millions of dollars) in period 0 and the estimated growth rate,
g, is 0.0359 or 3.59 percent.
To forecast sales in a future quarter, the appropriate value of t is substituted into
the equation. For example, predicted sales in the third quarter of 1999 (i.e., the fifteenth
quarter) would be 287.810(l.0359)'-5 , or $488.51 million.
Seasonal Variation in lime-Series Data Seasonal fluctuations in time-series data
are not uncommon. In particular, a large increase in sales for the fourth quarter is a
characteristic of certain industries. Indeed, some retailing firms make nearly half of
their total sales during the Christmas holida y period. Other business activities have
their own seasonal sales patterns. Electric utilities serving hot, humid areas have distinct
peak sales periods during the summer months because of the extensive use of air con-
ditioning, whereas those in colder regions may have peaks in winter. Similarly, housing
sales drop off during the winter, but the demand for accountants' services increases in
the first quarter as income tax deadlines approach.
A close examination of the data in Table 5.1 on page 150 indicates that the quar-
terly sales increases are not uniformly distributed over the year. The increases from the
first quarter to the second, and from the fourth quarter to the first, tend to be small,
while the fourth-quarter increase is consistently larger than. that of other quarters. That
is, the data exhibit seasonal fluctuations, as shown by the dashed line in Figure 5.2.
Pronounced seasonal variation can cause serious errors in forecasts based on time-
series data. For example, Table 5.1 indicates that actual sales for the fourth quarter of
1998 were $445 million. But if the estimated equation is used to predict sales for that
period (using the constant rate of change model), the predicted total is 281.394 +
12.811(12), or $435.13 million. The large difference between actual and predicted sales
occurs because the equation does not take into account the fourth-quarter sales jump.
Rather, the predicted value from the equation represents an averaging of individual
quarters. Thus, sales will be underestimated for the strong fourth quarter, conversely,
the predicting equation may overestimate sales for the other quarters.
The accuracy of the forecast can be improved by seasonally adjusting the data.
Probably the most common method of adjustment is the ratio-to-trend approach. Its use
can be illustrated using the data from Table 5.1. Based on the predicting equation,
= 281.394 4- 12.811t
actual and calculated fourth-quarter sales are shown in Table 5.3 on page 156. The final
column of the table is the ratio of actual to predicted sales for the fourth quarter. This
ratio is a measure of the seasonal error in the forecast.
 CHAPTERS
Business and Economic Forecasting 153
Sales per
quarter
ni III ion S)

51

40

30(

1996 1997 1998 1999

ON

As shown, for the three- y ear period, a v

erage actual sales for the fourth quarter
were 102 percent of the average forecastet sales for that quarter. The factor 1.02 can be
used to adjust future fourth-quarter sales estimates. For example, if the objective is to
predict sales for the fourth quarter of 1998. the predicting equation generates an esti-
mate of $435.13 million. Multiplying this number by the 1.020 adjustment factor, the
forecast is increased to $443.8 million, which is close to the actual sales of $445 million
for that quarter. A similar technique could be used to make a downward adjustment for
predicted sales in other quarters.
Seasonal adjustment can improve forecasts based on trend projection. However, trend
projection still has some shortcomings. One is that it is primarily limited to short-term
predictions. If the trend is e xtrapolated much be
y ond the last data point, the accuracy of
156 PART 11 Demand

Forecasted Actual
Fourth-Quarter Fourth-Quarter Actual/Predicted
Year -- Sales Sales Fourth-Quarer Sales

1996 332.64 $340 1.022
997 383.88 390 1.016
1998 435.13 445 1.023
Average = 1,020

the forecast diminishes rapidly. Another limitation is that factos such as changes in rela-
tve prices and fluctuations in the rate of economic growth are not considered. Rather, the
trend projection approach assumes that historical relationships will not change.

Key Concepts
• Forecasts'- based on Lhe assumptic'. f a c6hst— a, re cf change can be made h
fitting time-series data to an equation of the general form
= -+- hr
• Forecasts assuming a constant percentage rate of change involve stimating the
parameters of the equation
= so u t-4)
• Forecast errors due to seasonal variation can he reduced using- the ratio-to-
trend adjustment method.

Gase Study
Forecasting Winning Performances in the Olympic Games

The Summer Olympics brings together the world's best athletes. In most events, there
has been a general trend over the years for winning times, distances, weights, and scores
to improve as training methods and equipment are perfected and more athletes become
involved in the sport. The following table indicates the winning distance for the women's
shot put and the winning time for the men's 400-meter race for the 12 Olympics held be
tween 1948 and 1992.


CHAPTER 5 Business and Economic Forecasting 157

Winning Distance or Time
Women's Shot Put Men's 400 Meters
Year (meters) (seconds)
1948 13.75 46.2
1952 15.28 45.9
195i5 16.59 46.7
1960 17.32 44.9
1964 18.14 45.1
1968 19.61 43.8
1972 21.03 44.7
1976 21.16 44,3
1980 22.40 44.6
1984 20.48 44.3
1988 22.24 43.9
1992 21.06 43.5

Trend projection can be used to estimate the rate of improvement and forecast
gold-medal performances for future Olympic games. If the constant rate and percent-
age rate of change models are used to analyze the data, the
rsu1ts rc as follows;
Women's Shot put
= 14.403 + 0.721t / Forecast for 1996:
23.77 meters
= 14.585(1.041)' Forecast for 19'
24.44 meters
Men's 400 Meters
= 46.359 - 0.236t Forecast for 1996:
43.29 seconds
1', = 46.369(0.995)' Forecast for 1996:
43.31 seconds
The analysis implies that the winning distance for the women's shot put has in-
creased about seven-tenths of a meter, or 4 percent per Ol ympics. The gold medal 400-
meter time has decreased by about two-tenths of a second, or approximately one-half
of one percent during each four-year inter-val.
The actual winning times for the 1996 Olympics in Atlanta. Georgia, were 20.55
meters for the shot put and 43.5 seconds for the 400 meters. The trend projection equa-
tions for the 400 meters predict the 1996 winning time within two-tenths of a second,
but the forecasts for the women's shotput are much too high. The problem is that the
rapid increases in distance in the earlier years (e.g., 1948 to 1972) have not been
matched in more recent Olympics. In fact, the winning shotput distance for 1996 was
about the same as in 1972. One possible explanation is that the gender and drug tests
that are now required of athletes have reduced cheating in the field events. The failure
of the shot put equations to accurately predict 1996 results indicates the need for the
analyst to be aware of any conditions or events that could influence time-series data. 0
158 PART If Demand

Exponential Smoothing
Trend projection is actually just regression analysis where the only independent vari-
able is time. One characteristic of this method is that each observation has the same
weight. That is, the effect of the initial data point on the estimated coefficients is just as
great as the last data point. If there has been little or no change in the pattern over the
entire time series, this is not a problem. However, in some cases, more recent observa-
tions will contain more accurate information about the future than those at the begin-
ning of the series. For example, the sales history of the last three months may be more
relevant in forecasting future sales than data for sales 10 years in the past.
Exponential smoothing is a technique of time-series forecasting that gives greater
weight to more recent observations. The first step is to choose . a smoothing constant, a,
where 0 < a < 1.0. If there are n observations in a time series,'the forecast for the next
period (i.e., n + 1) is calculated as a weighted average of the observed value of the se-
ries at period n and the forecasted value for that same period.That is,
F 1 riX, + (12— a)F (5-5)
where F,. 1 is the forecast value for the next period,X,, is the observed value for the last
observation, and F is a forecast of the value for the last period in the time series. The
Corecasted values for F
and all the earlier periods are calculated in the same maLInr.
Specifically,
F1cX1+(l —cx)F 1 (5-6)
starting with the second observation (i.e.. I = 2) and going to the last (i.e., n). Note
that equation (5-6) cannot be used to forecast F1 because there is no X0 or F0. This prob-
lem is usually solved by assuming that the forecast for the first period is equal to the ob-
served value for that period.That is, F1 = X1 . Using equation (5-6), it can be seen that
this implies that the second-period forast is just the observed value for the first pe-
riod. or F2 = X1.
The exponential smoothing constant chosen determines the weight that is given to
different observations in the time series. As a approaches 1.0, more recent observations
are given greater weight. For example, if c.x = 1.0, then (1 - a) = 0 and equations (5-5)
and (5-6) indicate that the forecast is determined only by the actual observation for the
last period. In contrast . values or a give greator weight to observations from pre-
vious periods.
Assume that a firm's sales over the last 10 weeks are as shown in Table 5.4. By as-
sumption, F, = F1 X1. If a = 0.20. then

F = 0. 20(4.30) ± 080(400) = 406.0

and
F4 .. 0.20(420) + 0.80(406) = 408.8
The forecasted values for four different values of a are provided in Table 5.4. the tahie
also shows forecasted sales for the next period after the end of the time-series data. 'r
week 11. Using a = 0.20, the forecasted sales value for the 11th week is computed to he
0.20(420) ft90(435.7) = 432.56
 CHAPTER 5 Business and Economic Forecasting 159

Week Sales a = 0.20 a 0.40 a 0.60 a = 0.80

I F, F, F, F, _____

400 400.00 400.00 400.00 400.00

430 400.00 400.00 400.00 400.00

420 406.00 412.00 418.00 424.00

4 440 408.80 415.20 419.20 420.80
c 460

415.04

425.12

431.68

436.18

A'),l CI'
F; ,IAII
439.07 1.10

470 427.23 439.44 443.47 443.05

430 435.78 451.67 459.39 464.61

440 434.62 443.00 441.76 436.92

j.'j 420 435.70 441.80 440.70 439.38

432.56 433.08 428.28 423.88

'I able 5.4 suggests why this method is referred to as smoothing technique. Consider the
forecasts based on a = 0.20. Note that the smoothed data show much less fluctuation than
the original sales data. Note also that as .x increases, the fluetuations in the F increase, be-
cause the forecasts give more weight to the last observed value in the time series.
Choice of a Smoothing Constant' Any value of a could be used as the smoothing
constant. One criterion for selecting this value might be the analyst's intuitive judgment
regarding the weight that should be given to more recent data points. But there is also
an empirical basis for selecting the value of a. Remember that the coefficients of a re-
gression equation are chosen to minimize the sum of squared deviations between ob-
served and predicted values. This same method can be used to determine the smooth-
ing constant.
fle term (X, - F,)2 is the square of the deviation between the actual time-series
data and the.forecast for the same period. Thus, by adding these values for each obser-
vation, the sum of the squared deviations can be computed as

(X, -

One approach to choosing a is to select the value that minimizes this sum. For the data
and values of a shown in Table 5.4. these sums are
Smoothing Constant Sum of Squared !)eviaiions
0.20 6484.23
0.40 4683.87
0.60 4213.08
0.80 439.52

These results suggest that, of the four values of the smoothing constant. a 0.60 pro-
vides the best forecasts using these data. However, it should be noted that there may he
values of a between 0.60 and 0.80 or between 0.40 and 0.60 that yield even better results.
160 PART 11 Demand

Evaluation of Exponential Smoothing One advantage of exponential smoothing is

that it allows more recent data to be given greater weight in analyzing time-series data.
Another is that, as additional observations become available, it is easy to update the
forecasts. There is no need to reestimate the equations, as would be required with trend
projection-
The primary disadvantage of exponential smoothing is that it does not provide very
accurate forecasts if there is a significant trend in the data. If the time trend is positive,
forecasts based on exponential smoothing will he likely to be too low, while a negative
time trend NA,ill result in estimates that are too high. Simple exponential smoothing
works best when there is no discernahie time trend in the clnta There are. however,
more sophisticated forms of exponential smoothing that allow both trends and season-
ality to be accounted for in making forecasts.2

Case Study
Forecasting Economic Growth and the Federal Deficit

Among the hundreds 01 time series collected and published by the tederal government,
wo of the mostimportant are the rate of real economic growth and the federal deficit.
The numbers for these two series during the 1980s are shown here.
Growth Rate in Deficit
Year Real GOP (%)(billions)

1980 —0.2 61.3

1981 1.9 63.8
1982 —23- 145.9
1983 36 176.0
1984 6.8 169.6
1985 3.4 196.6
1986 2.7 206.9
198 3.4 1582
1988 4.5 1417
1989 2.5 134.3

Based on these data. the growth rates and deficits for future years can he forecast
using either trend projection or exponential smoothing.The estimated equations for the
constant rate-of-change trend projection model are
Growth, = 0.413 + 0.399t
Deficit, = 93.913 + 8.3851
Using exponentiai smoothing, a smoothing constant is selected by examining the
sum of squared devialions between the observed and forecasted values. For the 1980s

, Se e , example. S. Makrdkis and S. Wheelwright. Forecasting Methods for Manamen: ('kw York:
Wiley, [989). pp. 11-.94
 CHAPTER 5 Business and Economic Forecasting 161

data, it was found that the sum of squared deviations is minimized using a = 0.4 for the
growth rate and a = 1.0 for the deficit. The choice of a = 1.0 for the deficit data implies
that, with' exponential smoothing, the best forecast of the next period is just the ob-
served value for the previous period. In contrast, using growth data, the forecast for the
last period is weighted more heavily than the actual value for the previous period.
The actual and forecasted values for 1990 are provided below. Note that none of
the forecasts are close to the observed values for 1990. One explanation is that growth
rates and deficits don't follow a regular pattern over time. Rather, they are determined
a vat ut iaiuis, suii as tiai and iituiiiaiy puiay uL1u wuttu tAiiiuinie wuui-
tions. As such, regression models that incorporate other variables should generate more
accurate forecasts. However, even using very sophisticated econometric models, the
rack record for forecasting growth rates and deficits is not very impressive.0
Growth Rate Deficit
Trend projection forecast 4.8 191.6
Exponential smoothing 2.9 1343
forecast

Actual value for 1990 0.9 161.3

(ey nep
• )ntttiai sniot 'thin g are
d-- ic es for thc prevtus peii&J:
a.smoothing'èo im iZC-5 the
. sum of rei the.. for jsid tt: tes.

BAROMETRIC FORECASTING
Trend projection and exponential smoothiiig use time-series data to predict the future
based on past relationships. But if there is no clear pattern in a time series, the data are of
little value for forecasting. An alternative approach is to find a second series of data that
is correlated with the first. Hence, by observing changes in the second series, it may be pos-
sible to predict changes in the first. For example, suppose that a lumber company's sales
exhibited large yearly fluctuations over the last decade—so large that any forecast based
on a trend projection of sales is useless. But over the same period, it is also determined
that the firm's sales were highly correlated with the number of housing starts. Thus, if
housing starts can be predicted, this information can be used to forecast lumber sales.
A time series that is correlated with another time series is sometimes called an in -
dicator of the second series. Substantial time and effort have been expended searching


162 PART II Demand

4P#1
/ 4% / 4%

#2^

4 5 6
(a) Coincident indicator

#2 4% 4••

4 5
(b) Leading indicator
____ -.------:.•, •-.-*f-.;-- •. .., ..

for good indicators of economic trends. Business Conditions Digest, a monthly pub lica-.
tion of the Department of Commerce, reports information on over 300 tirn series.
These data are closely followed by economists, managers, and financial ana s.

Leading Indicators
If two series of data frequently increase or decrease at the same tithe, one- series may be
regarded as a coincident indicator of the other. For example. in Figure series i is a
coincident indicator of series 2 because the two series have their peaks anti rr-1ghs in
the same periods.
If changes in one series consistently occur prior to changes in another series,a lead-
ing indicator has been identified. In Figure 5.3b, series 1 can be considered a leading in-
dicator of series 2 because the peaks and troughs of series 1 consistently occur before
the corresponding peaks and troughs of series 2.
For purposes of forecasting, leading indicators are of primary interest. Much as a
meteorologist uses changes in barometric pressure to predict the weather, leading indi-
cators. can be used to forecast changes in general economic conditions. Consequently,
the use of such indicators is commonly referred to as barometric forecasting.
The value of a leading indicator depends on several factors. First, the indicator must
be accurate. That is, its fluctuations must correlate closely with fluctuations in the series
that itis intended to predict. Second, the indicator must provide adequaie lead time. Even
it two.senes are highly correlated, an indicator will be of little use if the lead time is too
short. For example, because of the time needed to put idle steel furnaces into operation,

Department of Commerce. Business ondiiio,is Digest (Washington. D.C.: U.S. Government Printing
Office, monthly).
 CHAPTER 5 Business and Economic Forecasting 163
a series that predicts changes in steel demand by, say, one week would not be a useful fore-
casting tool for managers in the steel industry. A third requirement is that the lead time
be relatively constant. If a series leads another by six months on one occasion and by two
years the next, the indicator will be of little use because it cannot provide useful forecasts.
Fourth, there should be a logical explanation as to wh y one series predicts another.
If enough time series are studied, it is likely that a correlation can be found between
pairs of series. However, unless there is a causal relationship between the two series, the
historical pattern may not be very useful in forecasting future events because there is
no reason to expect the pattern to be repeated. For example, it has been su22ested that
skirt lengths are a leading indicator of stock prices. Specifically, shorter skirts predict
higher prices as measured by the Dow Jones Industrial Average. Some attempts have
been made to explain this relationship, but they are not very convincing. Finally:an in-
dicator's value is affected by the cost and time necessary for data collection. A time se
ries that can be maintained onl y at a very high cost may not be worth the expense. Sim-
ilarly, if there is a long delay before the data are available, the effective lead time of the
indicator may be too short to be useful.
Table 5.5 lists selected leading indicators and an economic variable that each is used
to predict. For some of the indicators, there is an obvious link between the two series.
One example is new buildin g permits and private tiostng starts. When a permit is is
sued, this indicates a strong commitment to build a new house-Thus changes in permits
should be closely correlated to changes in housing starts. Clearly, housing permits rep-
resent a leading indicator because builders are legally required to have a permit before
beginning construction. Similar arguments can be made to explain new orders as lead-
ing indicators of sales in the durable goods and capital equipment industries.
For some of the indiators, it is more difficult to explain the correlation between the
two series. Consider stock prices as a predictor of general economic conditions. Histor-
ically, indices of common stock prices have been a relatively accurate predictor of cy-
cles in business activity. One possible explanation is that stock prices reflect expecta-
tions and plans of managers and consumers, which are implemented in future months.

Composite and Diffusion Indices

Although a time series showing the changes in stock prices may be somewhat useful in pre-
dicting general economic conditions, no single leading indicator has yet been identified

Economic Variable
-_
Lead.ii-1 irdjcator
I. Average workweek
Predicted b y the Indicator
Manufacturing output
2. Average weekly initial unemployment claims State unemployment insurance payments
3. New orders for durable goods Sales of durable goods
4. New orders for capital goods Sales of capital goods
5. New building permits Private housing starts
6. Change in manufacturing and trade inventories General economic conditions
7. Industrial material prices Consumer prices
8. Common stock prices Genera] economic conditions
164 PART II Demand

that comes close to having a perfect forecasting record. In addition, even when an indica-
tor correctly predicts variations in economic activity the lead time often shows consider-
able fluctuation. The basic problem is that any-time series contains random fluctuations
that do not conform to the general pattern of the data.
One way to improve barometric forecasting is by construction of an index. Such in-
dices represent a single time series made up of a number of individual leading indicators.
The purpose of combining the data is to smooth out the random fluctuations in each indi-
vidual series. Ideally, the resulting index should provide more accurate forecasts. The most
commonly used barometric forecasting indices are composite indices and diffusion indices.
Composite Indices A composite index is a weighted average of individual indica-
tors. The weights are based on the predictive ability of eaéh series. That is, a series that
does a better job of predicting would be given greater weight than a less accurate series.
The index is interpreted in terms of percentage changes from peridd to period. For ex-
ample, the most widely followed composite index is maintained by the U.S. Department
of Commerce and is based on 11 leading indicators.
Each month the department reports the change in the index. Monthly swings have
little significance, but sustained increases for several months are interpreted as a sign
that general economic conditions are likely to improve. Conversely, successive declines
in the index suggest that the economy is weakening.
Diffusion Indices The same 11 leading indicators also are used by the Department
of Commerce to construct a diffusion index. This index is ? measure of the proportion
of the individual time series that increase from one mc-: the iiet. For example, if
eight of the indicators increased from June to July, the diffusion index for July would he
8/11, or 72.7 percent. When the index is over 50 percent for several months, the forecast
is for improved economic conditions. As the index approaches 100 percent, the likeli-
hood of improvement increases. Conversely, if less than 50 percent of the indicators ex-
hibit an increase, a downturn is indicaied.
The use of indices-improves the accuracy of barometric forecasting. However, the
prediction record of this technique is far from perfect. On several occasions the De-
partment of Commerce indices have forecast recessions that failed to occur. Variability
in lead time is another weakness.A third problem is that, while the barometric approach
signals the likely direction of changes in economic conditions, it says little about the
magnitude of such changes. As such, it provides only a qualitative forecast. Finally, al-
though there has been much study on the leading indicators of general economic con-
ditions, the managers of individual fu-rns may find it difficult to identify leading indica-
tors that provide accurate forecasts for their specific needs.

Key Concepts :
• The value of a leading in&ator c a heaéeuray o prctiosth
• -and itabilitv of the lead time. and m.d availabiiit q*r1at
• Composite indices consist of n nfgc oa
• A diffusion index is eaLcttIed agtI tioi of Thcatots tharmcrae
from one pLTI&)44 to the nt
 CHAPTER 5 Business and Economic Forecasting
165
Example Calculating Composite and Diffusion Indices
Following are data for three leading indicators for a three-month period. The first
month represents the base period, and the three series are to be given equal weight.
Construct a composite and a diffusion index from the data.
Leading Leading Leading
Month Indicator I Indicator!! Indicator III
400 30 100
25 29 110
460 33 135
Solution The diffusion index is generated by determining whether each series in-
creased or decreased from month to month. For the second month, series I and III in-
creased, but series II decreased. Hence the index is 66.7 for that month. During the third
month, all of the series increased in comparison to month 2. Thus the index for that
month is 100.
The composite index can he computed by first calculating the percentage changes
(relative to the base month) for each series, Percentage changes during the second
month were 6.25 percent for the first series, —3.33
percent for the second, and 10 percent
for the third. Giving each series equal weight, the average percentage change was 4.31
percent. The value of the compose index for the first month is arbitrarily set to 100, so
the index for the second month is 104.31. For the third month, the changes from the base
period are 60/400 = 15 percent, 3/30 10 percent, and 35/100 = 35
percent, respectively.
Thus, the average change is 20.0 percent. Hence the index for that month is 120.0.
Values for the two indices for each month are shown In
the following table. Both in-
dices suggest that economic conditions should improve in future months.
Month Diffusion Index Composite index
1 •- 100.00
2 66.70 104.31
3 100.00 120Xi0

Econometric models can be used to forecast changes in demand in one sector, but such
models seldom have the detail necessary tossess the impacts of those changes on other
sectors. However, modern econOmic systems are highly interrelated, Changes in de-
mand in one sector of the economy can have significant impacts on demand in other
sectors. Some of these impacts are immediate and obvious. For example, steel, rubber,
glass, and plastics are all important inputs in the production of motor vehicles. Thus, an
increase in the demand for automobiles would cause an increase in the demand -for
those four products.
The direct impact that automobile sales have on the demand for steel, rubber, glass,
and plastics is a
ugmented by secondary effects generated in still other sectors. Consider the
resulting increase in demand for steel. As steel output increases to meet the requirements
for producing auto
mobiles,steel industry managers will find it necessary to purchase
166 PART II Demand

additional inputs. These may include iron ore, coal, and electricity If the increase in de-
mand is expected to be permanent, management may also decide to expand the capacity
of their production facility by acquiring additional capital goods such as blast furnaces.
In turn, these secondary impacts will affect other parts of the economy. Over time,
the increase in automobile demand may be the cause of change in hundreds of differ-
ent industries. To the casual observer, many of these changes would seem unrelated to
the production of automobile&
Input/output analysis is a useful technique for sorting out and quantifying sector-
by-sector impacts. This approach captures not only the direct effects but also related im-
pacts in other parts of the economy. For example, an input/output matrix developed by
the U.S. Department of Labor can estimate the impact of increased automobile sales on
nearly 500 individual sectors in the U.S. economy.

Transactions Matrix
Input/output analysis is based on a table that indicates historic patterns of purchases and
sales between industries. Data for this table are usually generated from surveying a sam-
ple of firms. A simple two-sector version of such a table is shown in Table 5.6. The two
sectors identified in the table are manufacturing and agriculture. Within each sector are
firms producing specific products. For example, the manufacturing sector might include
firms making electronic components and also firms that produce television sets and com-
puters. Similarly, the agriculture sector could include firms producing cotton and wool.
The rows of Table 5.6 show the disposition of each sector's output. Consider first
the manufacturing sector that is depicted in the first row. If the numbers represent bil-
lions of dollars, the first element in the first row indicates that firms engaged in manu-
facturing sold $8 billion of their product to other manufacturing firms. An example of
these "within-sector" transactions might be the sale of electronic components to a tele-
vision manufacturer. The second element of the first row shows that $10 billion of man-
ufactured goods were sold to firms in the agricultural sector. Together, these first two
elements indicate total sales by manufactumg to other firms. Such sales are sometimes
called intermediate sales because they represent output that is used as inputs for mak--
ing products sold to consumers.The first two elements of the second row are interpreted
in a similar manner. They show intermediate sales by the agriculture sector.
The third element of each row is designated as final demand and shows sales to ul-
timate consumers. Finally, the last element is total sales. This number is the sum of the

Sales to Sales to Final Total

Manufacturing + Agriculture + Demand Sales
Purchases from manufacturing 8 + 10 + 2 20

+ 12 = 30
Purchases from agriculture 6 + 12
+ +
+
Value added 6 8
20 30
Total sales
 CHAPTER 5 Business and Economic Forecasting 167

Sales to Sales to
Manufacturing Agriculture
Purchases from manufacturing 8 10
Purchases from agriculture 6 12

intermediate sales and final demand and represents the total sales by the sector. Thus
the $20 billion of total sales in manufacturing were made up of $18 billion in interme-
diate sales to other firms and $2 billion in sales to final consumers.
The columns of Table 5.6 indicate the use of revenues by each of the sectors. The
elements in the first column show that firms in the manufacturing sector spent $8 billion
to purchase intermediate goods from other manufacturing firms and $6 billion on goods
from the agriculture sector. Subtracting that $14 billion amount from total sector sales of
$20 billion leaves a residual of $6 billion, This amount is referred to as value added because
it represents the difference between the value of inputs purchased and the product finally
sold. Value added consists of payments to workers, owners of capital, and the government.
The elements of the table indicating intermediate sales and purchases are of par-
ticular interest for input/output analysis. Together, they make up the transactions ma-
trix or the model Based on the num hers in Thhle S h the transactions matrix is as shown
in Table 5.7. In matrix notation, let this matrix be designated as X, where
8 10
- 6 12
Direct Requirements Matrix
If each element in a column of the transactions matrix is divided by the total sales of
that sector's product, the result is the direct requirements matrix of an input/output
model. The elements of each column of the direct requirements matrix can be inter-
preted as indicating the immediate or direct change in revenue for a sector generated
by a $1 change in demand for the product of the sector represented by the column.The
direct requirements matrix for the sample problem is shown in Table 5.8.
Consider the elements of the manufacturing column. They were computed by di-
viding the first column of the transactions matrix by total sales in manufacturing—$20
billion. The first element indicates that for each dollar of manufacturing sales, $0.40 of
intermediate sales to other firms in the manufacturing sector are initially generated.
The second element shows that $030 in sales b y agricultural firms will be the direct re-
suit of each dollar of sales in the manufacturing Sector. The elements of the second col-
umn have a similar interpretation with respect to each dollar of sales by the agricultural
sector. Let the direct requirements matrix be designated as A. Thus
0.33
- [0'40
A - 0.30 0.40

Manufacturing Agriculture
Manufacturing sales per dollar 0.40 0.33
Agricultural sales per dollar 0.30 040
168 PART II Demand

Manufacturing Agriculture
Change in total 230 1.26
demand in manufacturing
Change in total 1.15 2.30
demand in agriculture

Direct and Indirect Requirements Matrix

The direct requirements matrix shnws the immediate or direct impact of changes in c
mand in a sector. But the elements of this matrix do not incorporate secondary and other
impacts. For example, Table 5.8 indicates that the direct result of a $1 increase in agri-
cultural sales would be a $0.33 increase in the demand for manufactured goods. How-
ever, such an increase would rcquire that manufacturing firms purchase additional goods
from other firms in the manufacturing .mnd agricultural sectors. These purchases would
stimulate still other purchases by firms in the two sectors. Thus the ultimate impact
would be greater than predicted by the coefficient of the direct requirerents matrix.
To take into account secondary impacts of changes in demand, inputiutput analy-
sis requires computation of a direct and indirect requirements matrix. The derivation
and calculation of this matrix
are beyond the scope of this book, but conve, tionally it
is designated by the notation (I - A)', where A is the direct requirements matrix, .I is
the identity matrix, and the exponent denotes the inverse of a matrix. 4 For the example
being considered, the direct and indirect requirements matrix is shown in Table 5.9.
The direct and indirect requirements matrix has a straightforward interpretation.
The elements of each column indicate the change in total demand in each sector that
would result from a $1 change in final demand for the sector designated by the column.
For example, consider the first column. The first element shows that after all direct and
secondary effects have been taken inta'account, a $1 change in final demand for manu-
factured goods will cause a $2.30 change in total demand in the manufacturing sector.
The second element of the first column estimates that change in total demand in agri-
culture resulting from a $1 change in final demand for manufactured goods. In this case,
the number is $1.15. Similarly, the elements of the second column show changes in total
demand in each sector caused by a $1 change in final demand in the agricultural sector.

Forecasting with an Input/Output Model

The direct and indirect requirements matrix can be used to forecast sector-by-sector im-
pacts of changes in final, demand. For example, suppose that an increase in exports
causes a $5 billion increase in final demand in the agricultural sector. After taking into
-account all the direct and secondary effects, what impact would this change have on to-
tal demand for manufactured and agricultural goods?
With respect to manufacturing, Table 5.9 shows that each $1 change in agricultural
final demand causes a $1.26 change in the total demand for manufactured goods. Thus

"An identity matrix has is on its main diagonal and Os for its other elements. It is analogous to the number
I in arithmetic. The inverse of a matrix is a matrix that if multiplied by the original yields the identity
matrix. The concept is analogous to the reciprocal in arithmetic.


CHAPTER 5 Business and Economic Forecasting

169
a $5 billion increase would increase total demand in manufacturing by
$5 billion X 1.26,
or $6.30 billion. Similarly, the increase in total demand in the agriculture sector would
be $5 billion x 2.30, or $11.50 billion.
Input/output analysis can also be used to forecast employment impacts. This is done
by assuming a constant ratio of employment to total demand. To illustrate, assume that
total demand in the manufacturing sector is $500 billion and the total number of work-
ers in that sector is 5,000.000. Hence the emplo yment ratio is Ito $100,000. That is, one
job exists for each $100,000 in total demand.
If this ratio is constant, one additional job in thc II1CLLILIIaL JUJJJJ g secLor Will
be cre-
ated for each $100,000 increase in total demand. Thus if a $5 billion increase in final de-
mand for agricultural products causes a $6.3 billion increase in total demand in the man-
ufacturing sector, the employment impact would he $6.3 billion divided by $100,000, or
63,000 new jobs. A similar computation could be made to estimate the employment in
crease in the agricultural sector.

Example Automobile Sales and the Demand for Glass

Assume that by 2002, automobile sales to consumers are forecast to bc $10 bifliori
greater than at present. A manufacturer of glass is contemplating a plant expansion and
wants to know the probable effect of the increase in automobile demand on the total
demand for glass. Union leaders in the industry want to know the employment impact.
The forecasts are to be based on a five-sector input/output model that includes sec-
tors for glass and automobiles. The direct and indirect requirements matrix is as shown.
The employment ratios are also shown. What will be the total demand and employment
impacts in the glass industr y of the projected increase in final demand for automobiles?
Direct and Indirect Requirements MarrLr
Steel Automobiles Computers Glass - - Electricity
Steel 1.034 0.334 0.008 0.010 0.010
Automobiles 0.008 1.010 0.009 0.002 0.007
Computers 0.003 0.004 1.110 0.001 0.100
Glass 0.009 0.090 0.048 1.004 01)05
Electricity 0.142 0.045 0.010 0.086 1.009
Employment Ratios
steel Aurpmob tiesComputersGloss Electricity
Jobs per million dollars 2.00 10.00 8.00 4.50
of total sales 1.00

Solution The elements in the second column of the direct and indirect requirements
matrix indicate the change in total demand resulting from a $1 change in final demand
for automobiles. For the glass industry, the coefficient is 0.090. Thus a $10 billion in-
crease in final demand for automobiles is estimated to increase total sales of glass by
0.090 X $10 billion, or $900 million. For the glass industry, the employment ratio is
4.5
jobs per $1 million in total sales. Thus the
$900 million increase in total sales of glass is
estimated to create 4,050 new jobs.
170 PART II Demand

The construction of input/output models is a very expensive and time-consuming

task. It would rarely be feasible for a firm to develop its own tables. Fortunately, several
large models have been developed by various governmental agencies. Perhaps the most
important is a 460-sector model of the U.S. economy maintained by the Department of
Labor. The department also publishes employment ratios. This information is used by
government analysts to evaluate impacts of public policy. It is also available for use by
firms in forecasting. In addition, there are state and regional input/output muels that
can be used for forecasting on a more localized basis.
The major value of input/output analysis is that it takes into account the interrcla-
tionships between sectrs. However, the approach has limitations in addition to its high
cost. A fundamental problem is that input/output forecasts are based on ratios that are
assumed to be fixed. This assumption may not be very realistic in situations where rapid
technological change is occurring.

Key Concepts
• Input/output analysis ecast th&dfectand indirect impactsof
changes in one sector on other sectors of the economy
• The transactions matrix shows the. pattern of sales and purchases between sac-
tors.The direct requirements matrix indicates the proportion of revenues used
to purchase goods and services from each sector.
The elements of the direct and indirect requirements mati ix show the impact
that a $1 change in final demand would have on total demand in each of the sec
tars of the model
.. input/output analysis can be used,tq pçdisectos .by-sectorsale and employ-
ment impacts

SUMMARY
Data for use in forecasting can be obtained from expert opinion, surveys, and market
experiments. Forecasts of experts can be refined by procedures such as the Delphi tech-
nique, which uses an iterative process to aid individuals in assessing their judgments.
Surveys of plans and attitudes by consumers and managers can be useful in forecasting.
However, considerable care must be exercised to assure that the responses reflect sub-
sequent consumer choices. Market experiments have an advantage over surveys be-
cause they provide information on actual consumer behavior. Their limitations include
the risk of losing customers, high cost, probkms of making consumers aware of changes,
and inability to control all factors.
Variations in time-series data are the result of trends, seasonal fluctuations. cyclical
patterns, and random factors. One method of analyzing a time series is trend projection.
Basically, the technique involves estimating the coefficients that provide the best fit of
data to a specified equation. The two most common equation forms estimate a constant
rate of change and a constant percentage rate of change. Fitting the constant percent-
age rate-of-change equation requires that the data be inputted in logarithmic form.
Forecasts based on trend projection of time-series data can be improved by adjusting
for seasonal Variations.
 CHAPTER 5 Business and Economic Forecasting 171
Another time-series technique for forecasting is exponential smoothing. Forecasts
based on this approach are a weighted average of observed and predicted values for the
last period. The weights are determined by choosing a smoothing constant that mini-
mizes the sum of squared deviations between observed and forecasted values. Expo-
nential smoothing works best when the time series does not have a strong time trend.
Barometric forecasting is based on time eries that are correlated. If the changes in
one series consistently occur prior to changes in another, a leading indicator has been
identified. Useful leading indicators must be accurate, provide adequate and constant
lead times, involve a logical explanation as to wh y one series predicts the other, and M.
available at reasonable cost. A composite index is a weighted average of several lead-
ing indicators. Diffusion indices measure the proportion of individual time series that
increase from one period to the next.
Inputioutput analysis can he used to forecast the direct and indirect impacts ot final
demand changes in one sector on total demand in other sectors of the economy. The
technique relies on data that show patterns of sales and purchases between Sectors.
These patterns are summarized ma transactions matrix. The direct requirements matrix
is computed by dividing the columns of the transactions matrix by sector total sa!es. It
shows the proportions of total revenues used to make purchases from the other -,COOTS.
The direct and indirect requirements matrix is the basic tool used for forecasting
with input/output analysis. Each number in a column shows the impact that a $1 change
in final demand in one sector would have on total demand in each of the other sectors
of the model. By assuming constant employment/total sales ratios, input/output analy-
sN can also be used to estimate sector-by-sector employment impacts.

Discussion Questions
5-1. The manager of Harrison Toys assembles a panel of five experts to assist in for-
mulating a sales forecast. The intent is to use the Delphi technique, but after ex-
amining the projections of the others, none of the experts are persuaded to make
any changes in their initial forecasts. What conclusion might the manager draw
from the inability of the experts to agree on a projection?
5-2. A business surveys its customers to generate information about the demand for
a proposed new product. Under what circumstances might the customers under-
state their true demand for the product? When might they overstate their true de-
mand? Explain.
5-3. To test consumer response, on January 1, a soft-drink manufacturer introduces a
new brand in the Chicago market area. Over a three-month period, sales are slow.
and the company concludes that the product is unlikely to he profitable in na-
tionwide marketing. Should the firm base its decision entirely on this experi-
ment? Why or why not?
5-4. Using trend projection, how could a manager decide whether to use the constant-
rate-of-change model rather than the constant-percentage rate of change?
5-5. Quarterly data from the beginning of) 988 to the end of 1998 are used to estimate
the coefficients of the equation S = S 0 + br, where S, is sales in the tth quarter
No seasonal adjustments are made. How accurate are each of the following fore-
casts likely to be? Explain.
a. Annual sales of a tire manufacturer for the year 2005.
 172 PART 11 Demand

b. Fourth-quarter sales for 1999 for a retail gift store.

c. Fourth-quarter sales for 1999 for a newspaper publisher.
5-6. A particular composite and a diffusion index are constructed based on the same
three times series of data. In a given month, could the composite index increase
while the diffusion index decreased? Use a simple example to explain your answer.
5-7. Could the lead time necessary for an indicator to be useful in forecasting he
longer in one industry than in another? Give an example.
5-8. The average workweek is considered a leading indicator of general economic ac-
tivity. Would longer workweeks predict an upturn or a downturn --economic ac-
tivity? Explain.
5-9. A district manager proposes that her division develop its own state-level in-
put/output model for use in forecasting. Is this a good idea? Why or why not?
c-rn A 20-sector inputfoutput model gives value added as a percent of SaIcs. Can this
estimate be used to rank relative profitability of the sectors? Wh y or ''fr not?
5-11. Would input/output analysis be useful in forecasting sates in a rapidl y chauging
industry such as personal computers? Explain.

Problems
5-1. The management of Romano's Foods decides to test-market a new frozen pizza.
As the person in charge of the market experiment, what considerations s ...iid af-
fect your decisions with respect to
a. The number and locations of cities in which the new pizza would be test-
marketed?
b. The duration of the experiment?
c Pricing of the product?
5-2. The trend line for sales of postcards (in millions) is estimated using annual data
from 1968 to 1998. The result is s= 10(1.05)'.
a. What is the interpretation of the "10" and the term in parentheses?
b. Forecast sales of postcards in 2000.
5-3 The trend for sates of beef (in millions of pounds) is estimated using data from
the first quarter of 1972 to the fourth quarter of 1998. The result is as shown be-
low with the i-value in parentheses:
Sr 12,5+ 0.14t
(3.28)
a. What is the interpretation of the coefficient oft? Is the coefficient statistically
significant at the 95 percent level? Explain,
b. Forecast the demand for beef in the third quarter of 1999.
5-4. Quarterly sales for Swarthmore Cycles are as follows:
Period Sales Period Sales
1997:1 100 1998:1 130
1997:11 110 1998:11 135
1997:111 115 1998:111 145
1997:IV 125 1998:IV 150
a. Plot the sales data and graphically fit a straight line to the points.
 CHAPTER 5 Business and Economic Forecasting
173
b. In using the data for trend prediction, would the constant rate of change or the
constant percentage change model be more appropriate? Explain.
c. Using a calculator or the computation methods described on pages 115-116 of
- chapter 4, use the data to estimate the coefficients of the equation S = S +
hi, where S is sales in the tth quarter.
d. Use the data to estimate the coefficients of the equation S
S0 (l + g.
5-5. Based on quarterly data from 1995:1 to 1998:IV, Highland Foods estimates that
potato chip sales can be projected using the equation
St = 5.000,000 + lOO,000t
where 1995:1 is period 1. Actual fourth-quarter sales were as follows:
1995 5.450000
1996 5860,00()
197 61270,000
1998 6,680,000

a. Project sales for the first three quarters of 1999.

b. Without using a seasonal adjustment, project sales for 1999:IV.
c. Project seasonally adjusted sales for 1999:W
5-6. The
winning heights (meters) for the Olympic pole vault for 1948-1996 were an -
alyzed using the two trend-projection models. The estimated equations were
PV, = 4.295 + 0.142t
PV = 4.289(1.033)t
where PV, is the forecasted winning height.
a On average, how much has the winning height increased from one Olympics
to the next? What has been the percentage increase?
b. Forecast the winning heights for 1976 and 1996.
c. The winning heights for 1976 and 1996 were 5.45 and 5.90 meters, respectively.
Is there an explanation for the large differences between the forecasted and
the actual values for 1996?
Over a six-year period, a firm's sales (millions) were $200, $220,$180,$200,$190,
and $210.
a. Using a smoothing constant of 0.5, forecast sales for the next period.
b. Graph the original data. Based on the results from part (a), graph the forecasted
sales. Do the smoothed results have less variability than the original data?
c. Forecast next period sales using smoothing constants of a = 1.0 and a = 0.0.
What are the implicit assumptions associated with these smoothing constants?
5-8. For three periods, the average workweek is 40,44, and 46 hours. An index of ma-
terials prices is 100,110, and 105 for the same three periods. Consider workweek
and materials prices as leading indicators of general economic activity.
A. Use these data to construct a conposite index for the three periods.

b. What does the index predict regarding general economic activity for the
fourth period? Explain.
5-9.
The demand for newsprint over a 10-year period is shown in the following table.
Also shown are three time series for the same time period.


PART IT Demand

Year

2 3 4 5 6 7 8 9 10
1
100
Quantity of 110 115 115 130 145 140 135 135 150
newsprint
Series A 25 30 25 35 30 40 45 50 45 45
Series B 200 220 225 225 240 260 255 250 255 240
Series C 20 20 23 26 25 24 24 28 28 26

a. Which of the time series is a roughly coincident indicator of newsprint demand?

n:_ L series :__1
U. VVI11411I
:.....I
4 14ULIL u1L1ILfl Ui
........-U) t'.-... f1,..
UULG1U i

dicator, what is the lead time?

c. Based on the leading indicator, is newsprint demand likely to increase, de-
crease, or stay the same in period 11? Explain.
5-10. Each year, sector A purchases $5 million of goods from sector B and $10 million
from sector C Sector B makes annual purchases of $2 million from sector A and
$8 million from sector C. Annual purchases of sector C are $3 million from sec-
tor A and $6 million from sector B. Withinsector purchases are $1 million for sec-
tor A, $4 million for sector B, and $5 'million for sector C. Total sales in the sec-
tors are $20 million, $20 million, and $30 million, respectively.
a. Write the transactions matrix for the three sectors.
b. How. much is value added for sector A?
c. How much is final demand in sector A?
d. Write the direct requirements matrix for the model.
5-11. The Ohio Planning Commission has developed a four-sector regional input/out-
put model with a direct and indirect requirements matrix as follows.
Sector
1 2 3 4__
11 1.20 0.45 0.90 0.60
I 2 0.20 1.50 0,80 0.75
Sector 0.63 1.10 0.10
0.89
0.24 0.80 0.43 1.24

a. What is the interpretation of the element in the first row of the second column?
b. If final demand decreases by $5 million in sector 3, what will be the effect on
total sales in sector 1?
c. The commission estimates that 1 ton of sulfur dioxide particles enters the at-
mosphere for each $10 million increase in total sales in sector 2. How much
pollution will result from a $5 million increase in final demand in sector 1?

Computer Problems

The following problems can be solved by using the TOOLS program (downloadable
from www.prenhall.comJpetersefl) or by using other computer software.
5-12. Electric power consumption data from 1969 to 1983 were provided in problem 4-14
on page 141.Those data are to be used for this problem.
a. Using trend projection, estimate the constant rate of change in electricity con-
g umption between 1969 and 1983.
 CHAPTER 5 Business and Economic Forecasting 175
b. Based on the result from part (a), forecast electricity consumption for 1984.
Actual U.S. consumption was 777.4 bi l lion kilowatt-hours. What might account
for the large prediction error?
5-13. Monthly sales for a firm over an 11-month period are as shown.
Month
Sales 4ionrh Sales
January 82.000 i
July $1,550
February 1,350
August 1,300
March 1,950
September 1200
April 1.975 Ocohcr I /
May 3.100
November 2.350
June 1.750

a. For smoothing consants of 0.2,0.4,0.6, and O.S. compute predicted sales. What
are the sums of squared deviations? What is the optimal smoothing constant?
b. Based on the result from part ('a), forecast sales for December.
5-14. The gold-medal-winning heights (meters) for the men's high jump for 1948 to
1992 are shown here.
Year Hei

1948 1.98 1972 2.23

1952 2.04 1976 2.25

1956 '2.12 1980 2.36

1960 2.16 1984 2.35

1964 2.18 1988 2.38

1968 2.24 1992 2.34
a. Use trend projection to estimate the constant rate of change and the constant
percentage rate of change for the period 1948 to 1992.
b. Use the estimated trend projection equations to forecast the winning height
for 1996.
c. Use exponential smoothing with a smoothing constant of a = 0.2 and a = 0.8
to forecast the winning height for 1996.
d. Which approach (trend projection or exponential smoothing) provides the
more accurate forecast? Explain.
.-15. Standard and Poor's provides a price index for stocks listed on the New York
Stock Exchange. The index for 1972 to 1982 is shown here.

Year Index Year Index

1972 109,2 1978 96.0

1973 107.4 1979 103.0

1974 82.9 1980 118.8

1975 86.2 1981 128.1

1976 102.0 1982 119.7

1977 98.2

a. What was the estimated constant percentage rate of change per period?
h. Using the data, project the Standard and Poor's Index for 1983 and 1984.


176 PART 11 Demand

c. The actual index for 1983 and 1984 was

Year index
1983 160.4
1984 160.5
Is trend projection a good tool for estimating future stock prices? Why or why not?
5-16. Quarterly profits data (in millions of dollars) for 1996:1 to 1998:IV are shown here.
Period Profit Period Profit
1996:1 -1.0 1 19971T1 -,-i 8
1996:11 -0.5 19971V +2.4
1996:111 -0.0 1998:1 4-23
19961V +0.6 1998:11 +2.9
1997:1 +1.1 1998:111 +3.3
1997:11 +13 1 1998-TV +3.6
a. Using trend projection, what is the estimated constant rate of change per period?
b. What happens when the trend projection program is used to estimate the con-
stant percentage rate of change? Explain.
5-17. Each year sector A purchases $5 million of goods from sectoi B and $10 million
from sector C. Sector B makes annual purchases of $2 million from sector A and
$8 million from sector C. Annual purchases of sector C are $3 million from Sector
A and $6 million from sector B. Within-sector purchases are $1 million for sector
A, $4 million for sector B, and $5 million for sector C. Total sales in each sector are
$20 million, $20 million, and $30 million, respectively. Final demands are $14 mil-
lion for sector A, $5 million for sector B, and $7 million for sector C.
a. Using input/output analysis what are the elements of the direct and indirect
requirements matrix? What is the interpretation of the element in the second
row and third column? -
b. Final demand in sector A increases by $2 million. What is the change in total
demand for each sector?
c. Starting from the original values, final demand decreased by $1 million each
in sectors A and B and increases by $2 million in sector C. What is the change
in total demand for each sector?
5-18. Consider the following four-sector transactions matrix:
Sales Sales Sales Saks Final Total
to A + to B + to C + to D + Demand Sales

Purchases 8 10 5 7 10 40
ftoni A

Purchases 6 9 10 8 2 35
from B

Purchases to 5 2 3 15 35
from C

Purchases 5 8 7 5 5 30
from D

Value added 11 3 11 7
Total sales 40 35 35 30
 CHAPTER 5 Business and Economic Forecasting 177
a. Write the direct and indirect requirements matrix. What is the interpretation
of the element in the fourth column of the fourth row?
b. If final demand in sector D increases by 5, what will be the changes in total de-
mand for each sector?
c. It is estimated that 50 jobs are created for each one-unit increase in total de-
mand in sector A and 20 jobs for each one-unit increase in total demand in sec-
tor 12 If final demand increased by 5 in sector D, how many additional jobs
would be created in each of these two sectors?
 Int€ASEy II
Southern Turkey

Southern Turkey's production division operates 10 turkey-producing facilities in Ten-

nessee, Arkansas, and Kentucky. The firm's marketing division sells frozen turkey and
turkey products in the southeastern United States. It has been in business since 1955 and
currently has an SO percent market share in the region. There are numerous other sellers.
hut they have little impact on Southern Turkey's operations. Quarterly sales data for the
company (in thousands of pounds) for four years are shown in the following table.
T? ;I

II III Iv
1995 10.000 10,500 10,5000 14,000
196 10.500 10.500 11.000 14,500
1997 11,000 11.500 11.500 15,000
1998 11500 11,500 12,000 16,000

A concern for managers at Southern Turkey is that they lack information about the
factors that affect demand for their product. In the past, some sales forecasts have been in-
accurate because of fluctuations in the p rices of substitute goods or changes in general eco-
nomic conditions. To obtain needed data, management approved a marketing experiment
in 100 cities in the firm's market area. This experiment was conducted during the first six
months of 1997 and involved charging different prices for turkey and then measuring the
per capita consumption in each city. In addition, data were collected on income per capita
and prices of chicken, beef and pork for each of the 100 cities. Then multiple-regression
techniques were used to estimate the following per capita demand function for turkey:
Q = 7.00 - 10.0 'T + 2.0 P + 1.0 P,3 0.50 P + .00031
(4.29) (-5.642) (3.333) (2.876) (0.503) (2.000)
= 0.75

178
 Integrating Case Sudy Ii 179
where: the f-statistics are in parentheses and
Q = annual turkey consumption per capita (pounds)
PT = price of turkey
Pc = price of chicken
P R = price of beef
Pp = price of pork
I = per capita income
At the. present time, the average wholesale selling price of Southein's i.urke y is
$0.5() per pound. In the southeastern United States, the average prices of chicken, beef,
and pork are $0.50,$1,50, and $1.50 per pound, respectively. Income per capita in that
region is $18,000.
Requirements
1. How much turkey will the firm sell in 1999 and in 000? How much will be sold in
the fourth quarter of each of those years? (Recall that the fourth quarter includes
the Thanksgiving and Christmas holidays.) Describe and justify your choice of a
forecasting technique. Identify possible sources of error in your forecast.
2. At this time, thc firm's production facilities are capable of producing 17,500,000
pounds of turkey every three months. However, management has tentative plans to
expand capacit y by constructing two new production facilities.These additions will
be started in mid-2000 and be operational by June 2001.1'he expansion will in-
crease the firm's production capacity to 19,000,000 pounds per quarter. Should
management proceed with its timetable for expanding capacity? Why or why not?
3. Historically, prices have been determined by using a markup over production cost.
Is there reason to believe that the current price of $0.50 per pound is not the profit
maximizing price? Explain.
4. In the data provided, does management have any information that would suggest
how the demand for turkey is affected by general economic conditions? Explain.
Be specific.
5. Government forecasts predict beef and chicken prices will be 10 percent higher in
1999, while pork prices will decline by 20 percent. The price of turkey is expected
to he unchanged. Estimate the individual impacts of each of these changes on the
per capita demand for turkey. How confident are you about your estimates? How
would this change your forecast of sales made in part 1? Explain.