Manag Econ
Manag Econ
D.J. Reyniers
MN3028, 2790028
2011
Undergraduate study in
Economics, Management,
Finance and the Social Sciences
This subject guide is for a 300 course offered as part of the University of London
International Programmes in Economics, Management, Finance and the Social Sciences.
This is equivalent to Level 6 within the Framework for Higher Education Qualifications in
England, Wales and Northern Ireland (FHEQ).
For more information about the University of London International Programmes
undergraduate study in Economics, Management, Finance and the Social Sciences, see:
www.londoninternational.ac.uk
This guide was prepared for the University of London International Programmes by:
Prof D.J. Reyniers, Director, Interdisciplinary Institute of Management, and Professor of
Management, London School of Economics and Political Science, University of London.
This is one of a series of subject guides published by the University. We regret that due to
pressure of work the author is unable to enter into any correspondence relating to, or arising
from, the guide. If you have any comments on this subject guide, favourable or unfavourable,
please use the form at the back of this guide.
Contents
Introduction ............................................................................................................ 1
Aims and objectives ....................................................................................................... 1
Subject guide breakdown .............................................................................................. 1
Learning outcomes ........................................................................................................ 2
Reading ........................................................................................................................ 2
Mathematics for managerial economics ......................................................................... 3
Online study resources ................................................................................................... 3
Examination advice........................................................................................................ 4
Some advice and ideas on how to study ......................................................................... 5
Glossary of abbreviations ............................................................................................... 5
Chapter 1: Decision analysis................................................................................... 7
Aims ............................................................................................................................. 7
Learning outcomes ........................................................................................................ 7
Essential reading ........................................................................................................... 7
Introduction .................................................................................................................. 7
Decision trees ................................................................................................................ 8
Attitude towards risk ................................................................................................... 10
Some applications ....................................................................................................... 12
The expected value of perfect information .................................................................... 14
A reminder of your learning outcomes.......................................................................... 17
Sample exercises ......................................................................................................... 17
Chapter 2: Game theory ....................................................................................... 19
Aims ........................................................................................................................... 19
Learning outcomes ...................................................................................................... 19
Essential readings ........................................................................................................ 19
Further reading............................................................................................................ 19
Introduction ................................................................................................................ 20
Extensive form games .................................................................................................. 21
Normal form games ..................................................................................................... 23
Nash equilibrium ......................................................................................................... 25
Prisoners’ dilemma ...................................................................................................... 29
Perfect equilibrium....................................................................................................... 30
Perfect Bayesian equilibrium ........................................................................................ 32
A reminder of your learning outcomes.......................................................................... 34
Sample exercises ......................................................................................................... 34
Chapter 3: Bargaining........................................................................................... 37
Aims ........................................................................................................................... 37
Learning outcomes ...................................................................................................... 37
Essential reading ......................................................................................................... 37
Further reading............................................................................................................ 37
Introduction ................................................................................................................ 37
The alternating offers bargaining game ........................................................................ 38
Incomplete information bargaining .............................................................................. 39
A reminder of your learning outcomes.......................................................................... 40
Sample exercise ........................................................................................................... 40
i
28 Managerial economics
ii
Contents
iii
28 Managerial economics
iv
Introduction
Introduction
Learning outcomes
At the end of the course, and having completed the Essential reading and
exercises, you should be able to:
• prepare for Marketing and Strategy courses by being able to analyse
and discuss consumer behaviour and markets in general
• analyse business practices with respect to pricing and competition
• define and apply key concepts in decision analysis and game theory.
Reading
Essential reading
This guide is intended for intermediate level courses on economics for
management. It is more self-contained than other subject guides might
be. Having said this, I do want to encourage you to read widely from the
recommended reading list. Seeing things explained in more than one way
should help your understanding. In addition to studying the material,
it is essential that you practise problem-solving. Each chapter contains
some sample questions and working through these and problems in the
recommended texts is excellent preparation for success in your studies.
You should also attempt past examination questions from recent years;
these are available online. Throughout this guide, I will recommend
appropriate chapters in the following books.
Tirole, J. The theory of Industrial Organisation. (Cambridge, Mass: The MIT
Press, 1988) [ISBN 9780262200714] this is excellent for the second part of
the guide dealing with pricing practices and strategic interactions.
Varian, H.R. Intermediate Microeconomics. (New York: W.W. Norton and Co.,
2006) seventh edition [ISBN 9780393928624] this can be useful mainly
for the review part of the course and those who prefer a less mathematical
treatment
Note on older editions: most of the relevant material from Varian
(2006) can also be found in the sixth edition of this book. Relevant
chapters for older edition are listed in Appendix 2.
Detailed reading references in this subject guide refer to the editions of the
set textbooks listed above. New editions of one or more of these textbooks
may have been published by the time you study this course. You can use
a more recent edition of any of the books; use the detailed chapter and
section headings and the index to identify relevant readings. Also check
the virtual learning environment (VLE) regularly for updated guidance on
readings.
Further reading
We list a number of journals throughout the subject guide which you may
find it interesting to read.
Please note that as long as you read the Essential reading you are then free
to read around the subject area in any text, paper or online resource. You
will need to support your learning by reading as widely as possible and by
thinking about how these principles apply in the real world. To help you
read extensively, you have free access to the VLE and University of London
Online Library (see below).
2
Introduction
The VLE
The VLE, which complements this subject guide, has been designed to
enhance your learning experience, providing additional support and a
sense of community. It forms an important part of your study experience
with the University of London and you should access it regularly.
The VLE provides a range of resources for EMFSS courses:
• Self-testing activities: Doing these allows you to test your own
understanding of subject material.
• Electronic study materials: The printed materials that you receive from
the University of London are available to download, including updated
reading lists and references.
• Past examination papers and Examiners’ commentaries: These provide
advice on how each examination question might best be answered.
• A student discussion forum: This is an open space for you to discuss
interests and experiences, seek support from your peers, work
collaboratively to solve problems and discuss subject material.
• Videos: There are recorded academic introductions to the subject,
interviews and debates and, for some courses, audio-visual tutorials
and conclusions.
3
28 Managerial economics
Examination advice
Important: the information and advice given here are based on the
examination structure used at the time this guide was written. Please
note that subject guides may be used for several years. Because of this
we strongly advise you to always check both the current Regulations for
relevant information about the examination, and the VLE where you
should be advised of any forthcoming changes. You should also carefully
check the rubric/instructions on the paper you actually sit and follow
those instructions.
I want to emphasise again that there is no substitute for practising
problem solving throughout the year. It is impossible to acquire a
reasonable level of problem solving skills while revising for the exam. The
examination lasts for three hours and you may use a calculator. Detailed
instructions are given on the examination paper. Read them carefully! The
questions within each part carry equal weight and the amount of time
you should spend on a question is proportional to the marks it carries.
Part A consists of compulsory, relatively short problems of the type that
accompanies each of the chapters in the guide. Part B is a mixture of some
essay type questions and longer analytical questions. In part B a wider
choice is usually available. Appendix 1 contains the exam I set in 1995 for
my second year undergraduates at LSE and indicates the level and type of
questions you should expect.
Remember, it is important to check the VLE for:
• up-to-date information on examination and assessment arrangements
for this course
• where available, past examination papers and Examiners’ commentaries
for the course which give advice on how each question might best be
4 answered.
Introduction
Glossary of abbreviations
Following is a list of abbreviations which are used throughout this subject
guide:
CE certainty equivalent
CEO Chief Executive Officer
CR concentration ratios
CRS computer reservation system
CV compensating variation
DGFT Director-General of Fair Trading
DM downstream monopolist
EU expected utility, European Union
EV equivalent variation, expected value
5
28 Managerial economics
6
Chapter 1: Decision analysis
Aims
The aim of this chaper is to consider:
• the concept of EVPI and how it can be used
• why we may want to use expected utility rather than expected value
maximisation
• the concept of certainty equivalent and how it relates to expected
value for a risk loving, risk neutral and risk hating decision-maker
• the application of decision analysis in insurance and finance.
Learning outcomes
By the end of this chapter, and having completed the Essential readings
and exercises, you should be able to:
• structure simple decision problems in decision tree format and
derive optimal decision
• calculate risk aversion coefficients
• calculate EVPI for risk neutral and non risk neutral decision-makers.
Essential reading
Varian, H.R. Intermediate Microeconomics. [ISBN 0393928024] Chapter 12.
Introduction
It seems appropriate to start a course on economics for management with
decision analysis. Managers make decisions daily regarding selection of
suppliers, budgets for research and development, whether to buy a certain
component or produce it in-house and so on. Economics is in some sense
the science of decision-making. It analyses consumers’ decisions on which
goods to consume, in which quantities and when, firms’ decisions on the
allocation of production over several plants, how much to produce and
how to select the best technology to produce a given product. The bulk
of economic analysis however considers these decision problems in an
environment of certainty. That is, all necessary information is available to
the agents making decisions. Although this is a justifiable simplification,
in reality of course most decisions are made in a climate of (sometimes
extreme) uncertainty. For example, a firm may know how many employees
to hire to produce a given quantity of output but the decision of whether
or how many employees to lay off during a recession involves some
estimate of the length of the recession. Oil companies take enormous
gambles when they decide to develop a new field. The cost of drilling for
oil, especially in deep water, can be over US$1 billion and the pay-offs in
terms of future oil price are very uncertain. Investment decisions would
definitely be very much easier if uncertainty could be eliminated. Imagine
what would happen if you could forecast interest rates and exchange rates
with 100 per cent accuracy.
Clearly we need to understand how decisions are made when at least some
of the important factors influencing the decision are not known for sure.
The field of decision analysis offers a framework for studying how these
7
28 Managerial economics
Decision trees
A decision tree is a convenient representation of a decision problem. It
contains all the ingredients of the problem:
• the decisions
• the sources of uncertainty
• the pay-offs which are the results, in terms of the decision-maker’s
objective, for each possible combination of probabilistic outcomes and
decisions.
Drawing a decision tree forces the decision-maker to think through
the structure of the problem s/he faces and often makes the process of
determining optimal decisions easier. A decision tree consists of two
kinds of nodes: decision or action nodes which are drawn as squares and
probability or chance nodes drawn as circles. The arcs leading from a
decision node represent the choices available to the decision-maker at
this point whereas the arcs leading from a probability node correspond
to the set of possible outcomes when some uncertainty is resolved. When
the structure of the decision problem is captured in a decision tree, the
pay-offs are written at the end of the final branches and (conditional)
probabilities are written next to each arc leading from a probability node.
The algorithm for finding the optimal decisions is not difficult. Starting
at the end of the tree, work backwards and label nodes as follows. At a
probability node calculate the expected value of the labels of its successor
nodes, using the probabilities given on the arcs leading from the node.
8
Chapter 1: Decision analysis
This expected value becomes the label for the probability node. At a
decision node x (assuming a maximisation problem), select the maximum
value of the labels of successor nodes. This maximum becomes the label
for the decision node. The decision which generates this maximum value
is the optimal decision at this node. Repeat this procedure until you reach
the starting node. The label you get at the starting node is the expected
pay-off obtained when the optimal decisions are taken. The construction
and solution of a decision tree is most easily explained through examples.
Example 1.1
0.617 0.25 2
0.41
0.7
private firm
0.34
-0.5
4
0.617
0.45
government 0.425
0.55
-2.5
Example 1.2
9
28 Managerial economics
should the executive take to maximise her expected profit? Try writing
down and solving the decision tree yourself without peeking! You should
get the following result.
600
168 oil
400
0.4 oil
drill
3000 ft -120
0.6
no oil drill
5000 ft no oil
168
-250
0
-160
10
Chapter 1: Decision analysis
is your certainty equivalent? Are you a risk lover?)? Note that it is possible to
be a risk lover for some lotteries and a risk hater for others.
40
200
p
1-p
0
Utility values can be obtained in a similar way for the other possible
outcomes of the decision problem. Replacing the monetary values by the
utility values and proceeding as before will lead to the expected utility
maximising decisions.
The definition of risk aversion can be rephrased in terms of the utility
function:
• a risk averse decision-maker has a concave utility function
• a risk lover has a convex utility function
• a risk neutral decision-maker has a linear utility function.
money
money
12
Chapter 1: Decision analysis
EU0
Example 1.3
13
28 Managerial economics
Maximising EU(A) and assuming an interior solution (i.e. 0 < A < W) leads
to the following (first order) condition:
EU'(A) = pU'(RW + (H – R)A)(H – R) + (1 – p)U'(RW + (L – R)A)(L – R) = 0
Note that, if the risky asset always yields a lower return than the riskless
asset (H, L < R), there can be no solution to this condition since U' > 0.
In this scenario the investor would not invest in the risky asset (A = 0).
Similarly, if the risky asset always yields a higher return than the riskless
asset (H, L > R) there can be no solution to this condition and under these
circumstances the investor would invest all of her wealth in the risky asset
(A = W). For the other scenarios (L < R < H) the first order condition above
allows us, for a specific utility function, to calculate the optimal portfolio.
14
Chapter 1: Decision analysis
600
drill
0.40
400
oil at
5000 ft drill
0.12
288 400 don’t 0
no oil
0.48
0 don’t drill
Why?
15
28 Managerial economics
1/3
16
O1
2/3
safe
O2
16
1/3 81
risky O1
2/3
O2 1
81
risky
O1 safe
1/3 16
O2 1
2/3 risky
16
16
Chapter 1: Decision analysis
Sample exercises
1. London Underground (LU) is facing a courtcase by legal firm Snook&Co,
representing the family of Mr Addams who was killed in the Kings Cross
fire. LU has estimated the damages it will have to pay if the case goes to
court as follows: £1,000,000, £600,000 or £0 with probabilities 0.2, 0.5 and
0.3 respectively. Its legal expenses are estimated at £100,000 in addition to
these awards. The alternative to allowing the case to go to court is for LU
to enter into out-of-court settlement negotiations. It is uncertain about the
amount of money Snook&Co. are prepared to settle for. They may only
wish to settle for a high amount (£900,000) or they may be willing to
settle for a reasonable amount (£400,000). Each scenario is equally likely. If
they are willing to settle for £400,000 they will of course accept an offer
of £900,000. On the other hand, if they will only settle for £900,000 they
will reject an offer of £400,000. LU, if it decides to enter into negotiations,
will offer £400,000 or £900,000 to Snook & Co. who will either accept (and
waive any future right to sue) or reject and take the case to court. The legal
cost of pursuing a settlement whether or not one is reached is £50,000.
Determine the strategy which minimises LU’s expected total cost.
2. Rickie is considering setting up a business in the field of entertainment
at children’s parties. He estimates that he would earn a gross revenue of
£9,000 or £4,000 with a 50–50 chance. His initial wealth is zero. What is
the largest value of the cost which would make him start this business:
17
28 Managerial economics
100
2/3
1/3
200
50
1/3
5/9
150
1/9
300
18
Chapter 2: Game theory
Aims
The aim of this chaper is to consider:
• the concept of information set and why it is not needed in decision
analysis
• why it is useful to have both extensive form and normal form
representations of a game
• the importance of the prisoners’ dilemma as a paradigm for many social
interactions
• the concept of dominated strategies and the rationale for eliminating
them in analysis of a game
• the concept of Nash equilibrium (this is absolutely essential!)
• the concept of non-credible threats and its application in entry
deterrence.
Learning outcomes
By the end of this chapter, and having completed the Essential readings
and exercises, you should be able to:
• represent a simple multi-person decision problem using a game tree
• translate from an extensive form representation to the normal form
representation
• find Nash equilibria in pure and mixed strategies
• explain why in a finitely repeated prisoners’ dilemma game
cheating is a Nash equilibrium
• explain the chainstore paradox.
Essential readings
Tirole, J. The Theory of Industrial Organization. Chapter 11.
Varian, H.R. Intermediate Microeconomics. Chapters 28 and 29.
Further reading
Axelrod, R. The evolution of cooperation. (New York, Basic Books 1984)
[ISBN 0465021212].
Kreps, D. and R. Wilson ‘Reputation and imperfect information’, Journal of
economic theory (1982) 27, pp.253–79.
Milgrom, P. and J. Roberts ‘Predation, reputation and entry deterrence’, Journal
of economic theory (1982) 27, pp.280–312.
‘Mumbo-jumbo, Super-jumbo’, The Economist, 12 June 1993, p.83
Nash, J. ‘Noncooperative games’, Annals of Mathematics 54 1951, pp.289–95.
‘Now for the really big one’, The Economist, 9 January 1993, p.83.
‘Plane wars’, The Economist, 11 June 1994, pp.61–62.
Selton, R. ‘ The chain store paradox’, Theory and Decision (9) 1978, pp.127–59.
‘The flying monopolists’, The Economist, 19 June 1993, p.18.
19
28 Managerial economics
Introduction
Game theory extends the theory of individual decision-making to
situations of strategic interdependence: that is, situations where players
(decision-makers) take other players’ behaviour into account when making
their decisions. The pay-offs resulting from any decision (and possibly
random events) are generally dependent on others’ actions.
A distinction is made between cooperative game theory and
noncooperative game theory. In cooperative games, coalitions or
groups of players are analysed. Players can communicate and make binding
agreements. The theory of noncooperative games assumes that no such
agreements are possible. Each player in choosing his or her actions, subject
to the rules of the game, is motivated by self-interest. Because of the larger
scope for application of noncooperative games to managerial economics, we
will limit our discussion to noncooperative games.
To model an economic situation as a game involves translating the essential
characteristics of the situation into rules of a game. The following must be
determined:
• the number of players
• their possible actions at every point in time
• the pay-offs for all possible combinations of moves by the players
• the information structure (what do players know when they have to
make their decisions?).
All this information can be presented in a game tree which is the game
theory equivalent of the decision tree. This way of describing the game is
called the extensive form representation.
It is often convenient to think of players’ behaviour in a game in terms of
strategies. A strategy tells you what the player will do each time s/he has
to make a decision. So, if you know the player’s strategy, you can predict
his behaviour in all possible scenarios with respect to the other players’
behaviour. When you list or describe the strategies available to each player
and attach pay-offs to all possible combinations of strategies by the players,
the resulting ‘summary’ of the game is called a normal form or strategic
form representation.
In games of complete information all players know the rules of the game.
In incomplete information games at least one player only has probabilistic
information about some elements of the game (e.g. the other players’ precise
characteristics). An example of the latter category is a game involving an
insurer – who only has probabilistic information about the carelessness of an
individual who insures his car against theft – and the insured individual who
knows how careless he is. A firm is also likely to know more about its own
costs than about its competitors’ costs. In games of perfect information all
players know the earlier moves made by themselves and by the other players.
In games of perfect recall players remember their own moves and do not
forget any information which they obtained in the course of game play. They
do not necessarily learn about other players’ moves.
Game theory, as decision theory, assumes rational decision-makers. This
means that players are assumed to make decisions or choose strategies which
will give them the highest possible expected pay-off (or utility). Each player
also knows that other players are rational and that they know that he knows
they are rational and so on. In a strategic situation the question arises whether
it could not be in an individual player’s interest to convince the other players
that he is irrational. (This is a complicated issue which we will consider in
20
Chapter 2: Game theory
the later sections of this chapter. All I want to say for now is that ultimately
the creation of an impression of irrationality may be a rational decision.)
Before we start our study of game theory, a ‘health warning’ may be
appropriate. It is not realistic to expect that you will be able to use game
theory as a technique for solving real problems. Most realistic situations are
too complex to analyse from a game theoretical perspective. Furthermore,
game theory does not offer any optimal solutions or solution procedures for
most practical problems. However, through a study of game theory, insights
can be obtained which would be difficult to obtain in another way and game
theoretic modelling helps decision-makers think through all aspects of the
strategic problems they are facing. As is true of mathematical models in
general it allows you to check intuitive answers for logical consistency.
Example 2.1
t 3,0
T 2
3,-2
b
1
t 1,-1
B
2
b 4,2
21
28 Managerial economics
endpoints of the tree. For example, when Player 1 chooses B and Player
2 chooses t, they receive pay-offs of 1 and –1 respectively. Games of
perfect information are easy to analyse. As in decision analysis, we can
just start at the end of the tree and work backwards (Kuhn’s algorithm).
When Player 2 is about to move and he is at the top node, he chooses t
since this gives him a pay-off of 0 rather than –2 corresponding to b. When
he is at the bottom node, he gets a pay-off of 2 by choosing b. Player 1
knows the game tree and can anticipate these choices of Player 2. He
therefore anticipates a pay-off of 3 if he chooses T and 4 if he chooses B.
We can conclude that Player 1 will take action B and Player 2 will take
action b.
Let us use this example to explain what is meant by a strategy. Player
1 has two strategies: T and B. (Remember that a strategy should state
what the player will do in each eventuality.) For Player 2 therefore, each
strategy consists of a pair of actions, one to take if he ends up at the top
node and one to take if he ends up at the bottom node. Player 2 has four
possible strategies, namely:
• (t if T, t if B)
• (t if T, b if B)
• (b if T, t if B)
• (b if T, b if B)
• or {(t, t), (t, b), (b, t),(b, b)} for short.
Example 2.2
The game tree below is almost the same as in Example 2.1 but here
Player 2 does not observe the action taken by Player 1. In other words,
it is as if the players have to decide on their actions simultaneously.
This can be seen on the game tree by the dashed line linking the two
decision nodes of Player 2: Player 2 has an information set consisting of
these two nodes. This game (of imperfect information) cannot be solved
backwards in the same way as the game of Example 2.1.
t 3,0
T 2
b 3,-2
1
1,-1
B t
2 4,2
b
22
Chapter 2: Game theory
Player 2
t b
Player 1 T 3,0 3,–2
B 1,–1 4,2
Normal form for example 2.2
Example 2.3
Firm B
Advertise Don’t
23
28 Managerial economics
Example 2.4
In the pay-off matrix below, only one pay-off, the pay-off to the row player,
is given for each pair of strategies. This is the convention for zero-sum
games (i.e. games for which the pay-offs to the players sum to zero for
all possible strategy combinations). Hence, the entry 10 in the (1, 1)
position is interpreted as a gain of 10 to the row player and a loss of 10 to
the column player. An application of this type of game is where duopolists
compete over market share. Then one firm’s gain (increase in market
share) is by definition the other’s loss (decrease in market share). In
zero sum games one player (the row player here) tries to maximise his
pay-off and the other player (the column player here) tries to minimise
the pay-off.
If we consider the row player first, we see that the middle row weakly
dominates the bottom row. For a strategy A to strictly dominate a strategy
B we need the pay-offs of A to be strictly larger than those of B against all
of the opponent’s strategies. For weak dominance it is sufficient that the
pay-offs are at least as large as those of the weakly dominated strategy.
In our case, the pay-off of M is not always strictly larger than that of B
(it is the same if Player 2 plays R). If we eliminate the dominated strategy
B, we are left with a 2 × 3 game in which Player I has no dominated
strategies. If we now consider Player 2, we see that C is weakly
dominated by R (remembering that Player 2’s pay-offs are the negative
of the values in the table!) and hence we can eliminate the second
column. In the resulting 2 × 2 game, T is dominated by M and hence we
can predict that (M, R) will be played.
Player 2
L C R
Player 1 T 10 20 –30
M 20 10 10
B 0 –20 10
Activity
Verify that, in the normal form of Example 2.1, this process leads to the outcome we
predicted earlier, namely (B,(t,b)) but that, in the normal form of Example 2.2, there are
no dominated strategies.
Sometimes, as in the example above, we will be left with one strategy pair,
which would be the predicted outcome of the game but the usual scenario
is that only a small fraction of the strategies can be eliminated.
24
Chapter 2: Game theory
Nash equilibrium
A Nash equilibrium is a combination of strategies, one for each player,
with the property that no player would unilaterally want to change his
strategy given that the other players play their Nash Equilibrium strategies.
So a Nash equilibrium strategy is the best response to the strategies that a
player assumes the other players are using.
Pure strategies are the strategies as they are listed in the normal form of
a game. We have to distinguish these from mixed strategies (which will
be referred to later). The game below has one Nash equilibrium in pure
strategies, namely (T, L). This can be seen as follows. If the row player plays
his strategy T, the best the column player can do is to play his strategy L
which gives him a pay-off of 6 (rather than 2 if he played R). Vice versa, if
the column player plays L, the best response of the row player is T. (T, L) is
the only pair of strategies which are best responses to each other.
L R
T 5, 6 1, 2
B 4, 3 0, 4
The story corresponding to this game is that of a husband and wife who
enjoy the pleasure of each other’s company but have different tastes in
leisure activities. The husband likes to watch football whereas the wife
prefers a night out on the town. On a given night the couple have to
decide whether they will stay in and watch the football match or go out.
The pay-off matrix could look like this.
Wife
In Out
In 10, 5 2, 4
Husband
Out 0, 1 4, 8
This game has two Nash equilibria: (in, in) and (out, out). Only when
both players choose the same strategy is it in neither’s interest to switch
strategies. The battle of the sexes game is a paradigm for bargaining
over common standards.
When electronics manufacturers choose incompatible technologies they
are generally worse off than when they can agree on a standard. For
example, Japanese, US and European firms were developing their own
versions of high definition television whereas they would have received
greater pay-offs if they had coordinated. The computer industry, in
particular in the area of operating system development, has had its
share of battles over standards. This type of game clearly has a first
mover advantage and, if firms succeed in making early announcements
which commit them to a strategy, they will do better.
25
28 Managerial economics
Example 2.6
A O
A –2, 2 20, 10
O 10, 20 –1, –1
There are two Nash equilibria in this game: (A, O) and (O, A). Note that
there is a ‘first mover advantage’ in this game. If Firm 1 can announce
that it will develop the drug for Alzheimer’s it can gain 20 if the
announcement is believed (and therefore Firm 2 chooses strategy 0).
Firms in this situation would find it in their interest to give up flexibility
strategically by, for example, signing a contract which commits them
to delivery of a certain product. In our scenario a firm could, with a
lot of publicity, hire the services of a university research lab famous for
research on Alzheimer’s disease.
A\B D DD
D –3, –3 10, –1
DD –1, 10 0, 0
26
Chapter 2: Game theory
There are two Nash equilibria: one in which Airbus builds the aircraft
(and Boeing doesn’t) and one in which Boeing builds it (and Airbus
doesn’t). In this game there is a significant ‘first-mover advantage’: if we
allow Boeing to make a decision before its rival has a chance to make
a decision it will develop the aircraft. (In reality the game is of course
more complicated and there are more strategies available to the players.
For example, Boeing could decide to make a larger version of its existing
450-seat 747, which would not be very big but could be developed
relatively quickly and at lower cost. Or it could decide to collaborate
with Airbus.)
The role of government regulation is not clear cut here. On the one
hand, governments may want to prevent the inefficient outcome of
both firms going ahead with development but, on the other hand, the
prospect of monopoly is not attractive either. Particularly if Boeing and
Airbus collaborate, the consequences of what would be an effective
cartel would be disastrous for struggling airlines. Not only would they
have to pay a high price for the super-jumbo but, if they want to buy a
smaller aircraft, they would have to rum to Boeing or Airbus who might
increase the prices of these smaller aircrafts to promote the super-
jumbo.
One way to avoid the problem of both companies starting development
is for the EU to announce that it will give a large subsidy to Airbus if
it develops the aircraft, regardless of whether Boeing also develops it.
Then ‘develop’ may become a dominant strategy for Airbus and one
Nash equilibrium would be eliminated. (To see this add 5 to A’s pay-off
if A chooses strategy D.)
The United States has persistently complained about Airbus subsidies and,
in 1992, an agreement limiting further financial support was reached.
As of July 1993 both firms had plans to go ahead independently with
development of a large aircraft despite discussing a partnership to build
a super-jumbo jet (the VLCT or very large civil transport project). In
June 1994 Airbus unveiled a design of the A3xx- 100, a double decker super
jumbo which would cost US$8 billion to develop.1 1
Case based on ‘Now
for the really big one’;
‘Mumbo jumbo, super
Example 2.7
jumbo’; ‘The flying
In the pay-off matrix below there is no Nash equilibrium ‘in pure monopolists’; ‘Plane
Wars’
strategies’ (i.e. none of the pairs (T, L), (T, R), (B, L) or (B, R) are stable
outcomes). Consider, for example, (B, L). If the row player picks
strategy B then the best response of the column player is L but, against
L, the best response of the row player is T, not B. A similar analysis
applies to the other strategy pairs.
L R
T 10, 5 2, 10
B 8, 4 4, 2
Nash (1951) showed that games in which each player has a finite number
of strategies always have an equilibrium. However, players may have to
use mixed strategies at the equilibrium. A mixed strategy is a rule
which attaches a probability to each pure strategy. To see why it makes
sense to use mixed strategies think of the game of poker. The strategies
are whether to bluff or not. Clearly, players who always bluff and players
who never bluff will do worse than a player who sometimes bluffs. Players
27
28 Managerial economics
using mixed strategies are less predictable and leaving your opponent
guessing may pay off. To see how to find a Nash equilibrium in mixed
strategies for a two-player game in which each of the players has two pure
strategies, consider the pay-off matrix of Example 2.7 again.
Example 2.7 (con’d)
Suppose the row player uses a mixed strategy (x, 1 – x) (i.e. he plays
strategy T with probability x and B with probability 1 – x) and the
column player uses a mixed strategy (y, 1 – y) (i.e. he plays strategy L
with probability y and R with probability 1 – y). Then the expected pay-
offs to the row and column player are respectively:
πr = 10xy + 2x(1 – y) + 8(1 – x)y + 4(1 – x)(1 – y)
and
πc = 5xy + 10x(1 – y) + 4(1 – x)y + 2(1 – x)(1 – y).
The row player chooses x so as to maximise her expected pay-off and the
column player chooses y so as to maximise his expected pay-off. Given y,
the expected pay-off to the row player is increasing in x as long as
y >1/2 and decreasing in x for y < 1/2 (check this by differentiating πr
with respect to x) and therefore the best response to y is x = 0 for y < 1/2,
x = l for y>1/2 and any x is optimal against y = 1/2. Following a similar
derivation for the column player we find that his best response to a
given x is to set y = 0 for x > 2/7, y = 1 for x < 2/7 and any y is optimal
against x > 2/7. These best response functions are pictured in bold in
Figure 2.3.
L
1/2
R
0 2/7 1
B X T
Prisoners’ dilemma
A class of two-person noncooperative games which has received much
attention not only in economics but in social science in general, is the
class of prisoners’ dilemma games. The story the game is meant to model
concerns two prisoners who are questioned separately, without the
possibility of communicating, about their involvement in a crime. They are
offered the following deal. If one prisoner confesses and the other does
not, the confessor goes free and the other prisoner serves 10 years; if both
confess, they each spend seven years in prison; if neither confesses, they
each serve a two-year term. This is summarised in the pay-off matrix below.
Confess Don’t
This game is very easy to analyse: for both players the strategy ‘confess’
dominates. (Remember that you want to minimise the pay-off here!)
There is one Nash equilibrium in which both prisoners serve seven-year
sentences. What is interesting about this game is that, if the prisoners
could set up a binding agreement, they would agree not to confess
and serve two years. (This type of model is used to explain difficulties
encountered in arms control for example.)
The typical application of the prisoners’ dilemma to managerial economics
translates the prisoners’ plight into the situation of duopolists deciding
on their pricing policy. If both set a high price they achieve high profits; if
both set a low price they achieve low profits; if one firm sets a low price
and its rival sets a high price the discounter captures the whole market
and makes very high profits whereas the expensive seller makes a loss. At
the Nash equilibrium both firms set low prices.
Of course in reality firms do not interact only once but they interact in
the market over many years and the question arises whether collusive
behaviour could be rationalised in a repeated prisoners’ dilemma game.
When the game is played over several periods rather than as a one-shot
game, players might be able to cooperate (set a high price) as long as their
rival is willing to cooperate and punish when the rival cheats (deviates
from cooperation). This possibility of punishment should give players
more of an incentive to cooperate in the ion- term. Axelrod (1984) ran
a contest in which he asked game theorists to submit a strategy or the
repeated version of the prisoners’ dilemma. He then paired the given
strategies (some of which were very complicated and required significant
computer programming) and ran a tournament. The ‘tit-for-tat’ strategy,
which consistently outperformed most of the others, is very simple. This
strategy prescribes cooperation as long as the other player cooperates
but deviation as soon as and as long as the other player deviates from
cooperation. It never initiates cheating and it is forgiving in that it only
punishes for one period. If two players use the tit-for-tat strategy they will
always cooperate.
Let’s think about what game theory can contribute to understanding
players’ behaviour in the repeated prisoners’ dilemma. If the game is
repeated a finite number of times then collusive behaviour cannot be
rationalised. To see this, remember that the only reason to cooperate is to
29
28 Managerial economics
avoid retaliation in the future. This means that, in the last period, there
is no incentive to cooperate. However, if both players are going to cheat
in the last period, the next-to-last period can be analysed as if it were the
last period and we can expect cheating then etc. so that we end up with
the paradox that, even in the repeated prisoners’ dilemma game, cheating
is the unique equilibrium. (Of course we are assuming, as always, that
players are intelligent, can analyse the game and come to this conclusion.
If a player is known to be irrational, an optimal response could be to
cooperate.) However, if the game is repeated over an infinite horizon or if
there is uncertainty about the horizon (i.e. there is a positive probability
(< 1) of reaching the horizon), then cooperation can be generated. What
is needed is that the strategies are such that the gain from cheating in one
period is less than the expected gain from cooperation. For example both
players could use trigger strategies (i.e. cooperate until the other player
cheats and then cheat until the horizon is reached). This will be a Nash
equilibrium if the gain from cheating for one period is smaller than the
expected loss from a switch to both players cheating from then onwards.
Perfect equilibrium
So far, with the exception of the section ‘Extensive form games’, we
have considered games in normal form. In this section we return to the
extensive form representation. Consider Example 2.1 and its normal form
representation at the beginning of the section on ‘Normal form games’.
From the pay-off matrix it is clear that there are three Nash equilibria
(T,(t, t)), (B,(t, b)) and (B,(b, b)). Two of these, the ones which have Player 2
playing b(t) regardless of what Player 1 plays, do not make much sense in
this dynamic game. For example, (B,(b, b)) implies that Player 2 – if Player
1 plays T – would rather play b and get a pay-off of –2 than t which gives
pay-off 0. The reason this strategy is a Nash equilibrium is that Player 1
will not play T.
The notion of perfect equilibrium was developed as a refinement of Nash
equilibrium to weed out this type of unreasonable equilibria. Basically, the
requirement for a perfect equilibrium is that the strategies of the players
have to form an equilibrium in any subgame. A subgame is a game starting
at any node (with the exception of nodes which belong to information
sets containing two or more nodes) in the game tree such that no node
which follows this starting node is in an information set with a node which
does not follow the starting node. In the game tree in Figure 2.4, a is the
starting node of a subgame but b is not since c, which follows b, is in an
information set with d which does not follow b.
1
b 2
1
c
1 2
1
d 1
3 1
a
30
Chapter 2: Game theory
t 10,7
T 2
b 3,0
1
7,10,-1
t
B
2
b 1,2
Player 2
(t,t) (t,b) (b,t) (b,b)
Player 1 T 10,7* 10,7* 3,0 3,0
B 7,10 1,2 7,10* 1,2
Normal form
31
28 Managerial economics
don’t
3,4
enter I
-2,0
E fight
don’t
0,10
32
Chapter 2: Game theory
T2 2,0,0
T1 2 0,2,0
T3
B2
1 M1 3 B3
2 0,2,2
T2
T3 2,8,0
B1 3
B2
Example 2.10
33
28 Managerial economics
0,10
don’t enter -2,0
E fight
‘sane’ x enter
I don’t fight
3,4
don’t enter 0,10
1-x -2,5
‘crazy’ E
fight
enter I
don’t fight 3,4
Sample exercises
1. Consider the following matrix game.
L R
T 2, 5 1, 4
B 5, –1 3, 1
Are there any dominated strategies? Draw the pay-off region. Find the
pure strategy Nash equilibrium and equilibrium pay-offs. Is the Nash
equilibrium Pareto efficient? Which strategies would be used if the
players could make binding agreements?
2. Find the Nash equilibrium for the following zero sum game. The
tabulated pay-offs are the pay-offs to Player I. Player II’s pay-offs are
the negative of Player I’s. How much would you be willing to pay to
play this game?
Player II
–1 2 –3 0
Player I 0 1 2 3
–2 –3 4 –1
34
Chapter 2: Game theory
to collude in the first year and both offer a most favoured customer
clause. What is the pay-off matrix for the second year if they colluded
the first year?
4. Find the pure and mixed strategy equilibria in the following pay-off
tables. How might the –100 pay-off affect the players’ actions?
L R L R
T 12, 10 4, 4 T 12, 10 4, 4
B 4, 4 9, 6 B 4, –100 9, 6
5. Students don’t enjoy doing homework and teachers don’t like grading
it. However, it is considered to be in the students’ long-term interest
that they do their homework. One way to encourage students to do their
homework is by continuous assessment (i.e. mark all homework), but
this is very costly in terms of the teachers’ time and the students do not
like it either. Suppose the utility levels of students and teachers are as
in the pay-off matrix below.
teacher
check don’t check
student work 0, -3 0, 0
no work -4, 4 1,- 2
a b c d
u 100, 3 2, 2 2, 1 0, –500
d 0, –500 3, –500 3, 2 1, 10
7. Two neighbours had their house broken into on the same night and
from each house an identical rare print went missing. The insurance
company with whom both neighbours had taken out home contents
insurance assures them of adequate compensation. However, the
insurance company does not know the value of the prints and offers
the following scheme. Each of the neighbours has to write down the
cost of the print, which can be any (integer) value between £10 and
£10,000. Denote the value written down by Individual 1 as x1 and the
one written down by Individual 2 as x2. If x1 = x2 then the insurance
company believes that it is likely that the individuals are telling the
truth and so each will be paid x1. If Individual 1 writes down a larger
number than Individual 2, it is assumed that 1 is lying and the lower
number is accepted to be the real cost. In this case Individual 1 gets
x2 – 2 (he is punished for lying) and Individual 2 gets x2 + 2 (he is
35
28 Managerial economics
rewarded for being honest). What outcome do you expect? What is the
Nash equilibrium?
8. Consider the extensive form game below. Find all Nash equilibria. Find
the subgame perfect equilibrium.
u
2 0,0
U
1 d
3,1
D
1,3
9. Consider the extensive form game below. What are the players’
information sets? Write this game in normal form and analyse it using
the normal form and the game tree.
t 3,2
2
T b 2,1
M t 1,2
1 2 b
1,4
B
t 2,1
2
2,4
b
10. A firm may decide to illegally pollute or not. Polluting gives it an extra
pay-off of g > 0. The Department of the Environment can decide to
check for pollution or not. The cost of this inspection is c > 0. If the
firm has polluted and it is inspected, it has to pay a penalty p > g; in
this case, the pay-off to the department of the environment is s – c > 0.
If the firm has not polluted and it is checked, no penalty is paid. If the
department of the environment does not inspect, its pay-off is 0.
a. Suppose that the Department of the Environment can observe
whether the firm has polluted or not before it formally decides
whether or not to check. Draw the game tree. What is the
equilibrium?
b. Now suppose the pollution cannot be observed before checking.
Draw the game tree. Is there a pure strategy equilibrium? Compute
the mixed strategy equilibrium, How does a change in the penalty
affect the equilibrium?
36
Chapter 3: Bargaining
Chapter 3: Bargaining
Aims
The aim of this chaper is to consider:
• the cooperative and conflictual aspects of bargaining
• the nature of the inefficiencies associated with incomplete information
bargaining.
Learning outcomes
By the end of this chapter, and having completed the Essential readings
and exercises, you should be able to:
• analyse alternating offer bargaining games with finite or infinite
number of rounds.
Essential reading
Tirole, J. The Theory of Industrial Organisation. Section 11.5.2.
Further reading
Fundenberg, D. and J. Tirole Game Theory. (Cambridge Mass: The MIT Press,
1991) [ISBN 0262061414].
Rubenstein, A. ‘Perfect equilibrium in a bargaining model’, Econometrica (1982).
Introduction
The literature on bargaining has developed dramatically in the last
decade or so due to advances in noncooperative game theory. Bargaining
is an interesting topic of study because it has both cooperative and
conflictual elements. For example, when a seller has a lot, reservation
price for an object and a buyer has a high reservation price then, clearly,
if the two parties can agree to trade, they will both be better off. On the
other hand, conflict exists regarding the divisions of the gains of trade. The
seller will naturally prefer a high price and the buyer will prefer a low price.
Game theory helps us to model bargaining situations carefully and allows
us to check our intuition regarding, for example, how the outcome of
the bargaining will depend on the parties’ bargaining power and so on.
Questions economists are interested in include:
• under which conditions will bargaining lead to an efficient outcome
• what are good bargaining strategies?
Bargaining problems arise whenever pay-offs have to be shared among
several players. When firms succeed in running a cartel,1 for example, 1
See Chapter 11,
agreeing on how to divide cartel profits is a major problem. Managers ‘Oligopoly’
are interested in bargaining models for their predictions in the context of
management (e.g. for labour (union) negotiations and strikes). However,
most game theoretic models of bargaining are either very simplistic (and
that is certainly true for the ones I discuss in this chapter) or extremely
complex and unrealistic in their assumptions about players’ ability to
reason and calculate.
In the hope that this does not discourage you too much, let us proceed.
37
28 Managerial economincs
38
Chapter 3: Bargaining
As the time between offers and counteroffers shrinks, the discount factor
approaches 1 and the asymmetry, caused by who moves first, disappears.
It is not very difficult to extend this analysis to allow for different discount
factors of the two bargaining parties. The conclusion of this modified
bargaining game is that the more patient bargaining partner will get a
larger slice of the cake.
Experimental work
Although the alternating offers bargaining game has a simple ‘solution’
and a stark prediction about the duration of the bargaining (one
offer only), the game theoretic findings are not always replicated
in experiments. About 10 years ago, when I was a student at LSE, I
(and many others) participated in a series of bargaining experiments.
Subjects were paired to an unknown bargaining partner and could only
communicate via a formal computer program with their partner who was
sitting in a different room. Experiments of this type generally show that
‘real’ players have a tendency to propose and accept what they consider
a fair offer while rejecting what they consider a mean offer even if this
rejection means they will be in a worse position. If you were offered one
penny in the last round of the game in this section, would you accept?
39
28 Managerial economincs
Sample exercise
Consider the alternating offers bargaining game over £100, with Bert
making the first offer and Ernie making a counter offer if he wants to.
Suppose Bert has an outside option of £50, that is, at any point during the
game, Bert can stop bargaining and get £50 (discounted if he gets it after
the first period). If Bert takes his outside option, Ernie gets zero. How does
this effect the equilibrium strategies and pay-offs?
40
Chapter 4: Asymmetric information
Aims
The aim of this chapter is to consider:
• why asymmetric information represents a problem if we are concerned
with efficiency the ‘lemon’ problem
• how using a ‘deductible’ can overcome moral hazard in the insurance
context
• the difference between signalling and screening
• the relevance of the participation and incentive compatibility
constraints in the principal-agent problem.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• give examples (preferably your own) of adverse selection and
moral hazard
• discuss how adverse selection in insurance can lead to low risk
individuals being priced out of the market and explain the use of a
partial cover insurance policy in this context
• explain Spence’s education as a signal model and work out an
example
• analyse a simple principal-agent problem.
Essential reading
Varian, H.R. Intermediation Microeconomics. Chapter 37.
Further reading
Akerlof, G.A. ‘The market for “lemons”: quality uncertainty and the market
mechanism’, Quarterly Journal of Economics (85) 1970, pp.488–500.
‘Flattened, sort of’, The Economist, 5 November 1994, p.100.
‘Generations X-onomics’, The Economist, 19 March 1994, pp.55–56.
‘Keep taking the tablets’, The Economist, 4 December 1993, p.81.
Milgrom, P. and J. Roberts ‘Price and advertising as signals of product quality’,
Journal of Political Economy 94(4) 1986, pp.796–821.
Rothschild, M. and J. Stiglitz, J. ‘Equilibrium in competitive insurance markets:
an essay on the economics of imperfect information’, Quarterly Journal of
Economics (91) 1976, pp.629–49.
Spence, M. Market signalling. (Cambridge: Harvard University Press, 1972).
‘Still money in that franchise’, International Banking Survey, The Economist, 30
April 1994, pp.43–46.
‘The consequences of kindness’, The Nordic Countries Survey, The Economist, 5
November 1994, pp.13–16.
41
28 Managerial economics
Introduction
Situations of asymmetric information arise when one party involved
in a transaction knows more than the other about relevant variables.
For example, a seller may have more information about the quality of
the product he is trying to sell than the buyer does. With asymmetric
information decisions are taken which would be inefficient under
symmetric information. In extreme cases, markets which would show
vigorous trade under perfect information may collapse completely. It
becomes very difficult under asymmetric information to draw inferences
from people’s behaviour, as they may be trying to fool you. For example,
when bargaining, a buyer may make a low bid in the hope that the seller
will believe he has a low reservation price. By bluffing in this way he
may increase his expected gain at the expense of an inefficient delay in
reaching an agreement.
Adverse selection
Asymmetry of information may lead to adverse selection where
less desirable agents are more likely to voluntarily participate or ‘self-
select’ in trade. In adverse selection problems, the type of agent is
not observable and has to be guessed. This is why adverse selection
problems are also known as hidden information problems. For
example, people who suffer from terminal diseases are more likely than
others to want to buy medical insurance or life insurance. The individuals
considering buying insurance have better information about their risks
than an insurance company does.
As a consequence of this adverse selection problem, the insurance
company cannot use risk estimates from the general population to set
its premiums. The premiums will have to be higher than those based on
the general population. Similarly, if a bank charges the same interest to
low risk and high risk borrowers, the high risk ones will be more likely to
borrow. The bank faces an adverse selection of borrowers. If you advertise
a job vacancy at a given wage, you only get people applying who are
willing to work at that wage. From a pool of potential applicants, only the
less desirable workers might apply. Adverse selection can be responsible
for market failure (i.e. there may be no market for a good whereas
profitable transactions between buyers and sellers would be possible if
everyone had full information).
42
Chapter 4: Asymmetric information
Example 4.1
Although Example 4.1 was phrased in terms of the used car market, the
‘lemons’ problem can occur in any market where buyers and sellers have
different information about the quality of the good being sold. In such
markets there is a tendency for low quality to crowd out high quality.
You might think that buyers and sellers should be able to write a contract
stipulating the quality sold, with appropriate penalties if the delivery is
later found to be of low quality. However, writing such contracts is costly
and only makes sense if they can be enforced. It is often very difficult for
a buyer, let alone a court, to assess quality accurately and to determine
whether any defect existed at the time of the sale or was caused by
negligent use. The adverse selection problem can be attenuated if
reputation is important, as is the case when a seller aims to repeatedly sell
to a given buyer, and by the use of warranties.1 1
Discussed in ‘Signalling
and screening’
Example 4.2
43
28 Managerial economics
R ≥ 400ph + 100p1
As long as there are many low claim individuals this premium does not
exceed the low claim individuals’ reservation price:
400ph + 100p1 = 400 (1 – p1) + 100p1 < 130 for p1 > 90%.
However, when there are too few low claim individuals, the premium
required by the insurance company to cover its cost would have to
increase to above £130. This implies that only high claim individuals
buy insurance at a high premium R > 400. Note that under perfect,
symmetric information everyone would buy insurance at the
appropriate rate for their risk category. However, because of the
asymmetry of information, if the insurance company continues to set
the premium based on the average claim, it will make losses since only
the high risk properties will be insured. So it must set the premium
based on the high risk clients and only these will be insured.
Example 4.3
The value of Target Ltd. under its current management is known only
to the current management team. EFM Ltd. is interested in taking
over Target and estimates that Target’s current value is 50, 60 or 75
(£ million) – all equally likely – whereas, after takeover and under
EFM’s management, the value would be 50, 70 or 85 respectively.
The procedure for the takeover bid is that EFM makes an offer which
Target can accept or reject. EFM only gets one chance to make an offer.
How much should they bid? Adverse selection occurs because, for any
bid EFM makes, Target will only be interested if it has a low value
(below the level of the bid). In fact we have an extreme case of adverse
selection here in that it prohibits trade altogether:
• if EFM offer 50 it cannot make a profit since Taret would only sell
if its value was 50
• if EFM offers 60, Target sells if its value is 50 or 60 which gives an
expected gain to EFM of (1/3)(–10) + (1/3)(10) = 0
• if EFM offers 75 then Target will accept the offer for sure but EFM
expects a loss: (1/3)(–25) + (1/3)(–5) + (1/3)(10) < 0.
44
Chapter 4: Asymmetric information
In Example 4.3, a takeover bid is only accepted if the target’s value under
its current management is below it. If information were symmetric (i.e. if
the bidder had equal access to information about the target’s value as the
target’s current management, the inefficiency associated with the absence
of a takeover could be avoided). A successful bid could be made in any
case, whatever the target’s value is, and trade would take place at some
price between the target’s value under its current management and its
value after the takeover under a new management team.
Moral hazard
‘Moral hazard’ is about incentive problems where one agent cannot
observe the actions of the other agent. This is why moral hazard
problems are also known as problems of hidden action. In the typical
example, being insured changes an individual’s behaviour. If someone has
taken out insurance for theft or collision he is likely to be more careless
than when he was not insured. Whereas the term ‘adverse selection’
generally applies to characteristics or qualities of one of the parties
before a contract is entered into (pre-contractual opportunism),
‘moral hazard’ describes situations in which one of the parties misbehaves
after a contract is signed (post-contractual opportunism). A
restaurant which offers an all-you-can eat deal gets an adverse selection of
big eaters as its clientele. If a group of people eat out and decide to share
the bill equally, moral hazard implies that they are likely to overindulge.
If a company rewards employees based on seniority rather than
performance, it may get an adverse selection of low-achieving applicants.
When compensation is based on team performance rather than individual
performance, moral hazard leads to reduced efforts.
In the 1980s the American government paid out US$300 billion when
hundreds of savings and loan associations failed. This debacle has
been explained in terms of moral hazard on the part of the banks who
(are required by law to) take out deposit insurance provided by the
government. With the deposit insurance, banks do not have the proper
incentive to avoid excessive risk taking.2 For life insurance policies in the 2
See ‘Still money in that
US, the insurance company pays the beneficiary after a suicide only if the franchise’
suicide took place a year or two years from the time the policy was issued.
Life insurance statistics show that the suicide rate is lowest in the 12th and
24th month and highest in the 13th and 25th month of the policy!
Of course, if the insurer can observe the behaviour of individuals (e.g.
where they park their cars or whether they lock them – in the case of
motor insurance; whether they smoke or take drugs in the case of life
insurance), it could group them in different risk classes and set a different
premium for each class. The problem is that it is often impossible or very
costly to monitor behaviour. To counteract the moral hazard problem
in insurance, companies often do not give complete insurance but use
deductible provisions.
Example 4.4
Elmo wants to insure his car, of value V, against theft. Elmo who has
wealth W (this includes the value of his car) is risk averse and has a
strictly concave utility of money function U. Elmo knows he should
really always lock his car and fix the anti-theft device he keeps in his
trunk. However, everything else equal, he prefers to be lazy and leave
the car unlocked. If he is careful his car gets stolen with probability pf;
45
28 Managerial economics
(1 – pi)U(W – R) + piU(W – V + D – R)
is decreasing in pi and Elmo therefore has an incentive to be careful. The
insurance company can now afford to base its premium on anticipated
careful behaviour (i.e. R = pf D). Note that the size of the deductible is
unimportant; in fact, for utility maximisation, we want D as close to V as
possible but not equal to it.
46
Chapter 4: Asymmetric information
47
28 Managerial economics
Education as a signal
The classic example of signalling, first analysed by Spence (1974), is
one in which high productivity individuals try to differentiate themselves
from low productivity individuals by investing in education. The firm
cannot distinguish between high and low quality workers but the workers
themselves know their abilities. As I will show in the following example,
good workers can use education as a signal of productivity since only
the most productive workers invest in education. In other words, the
signalling cost to the good workers is lower than to the low productivity
workers and therefore the two types of workers can be identified because
they make different choices.
Example 4.5
48
Chapter 4: Asymmetric information
You should be able to show that, if the expected tenure for the first
job is less than four years, signalling does not pay and no worker gets
a degree. In this case we find a pooling equilibrium in which the
two types of workers use the same strategy and are paid £22,500.
Similarly, if the expected tenure is long enough, say eight years or more,
signalling will not work because everyone would prefer to get a degree
and earn £30,000. Firms which want to break even cannot pay the high
salary to graduates in this case and as a result nobody gets a degree.
Note that the high productivity workers are negatively affected by the
existence of low productivity workers. Either they have to invest in
signalling (in a separating equilibrium) or they get a wage below their
productivity (in a pooling equilibrium). This type of externality is an
important aspect of the adverse selection problem.
In the US, where labour markets are less regulated than in Europe and the
minimum wage is only US$4.25 per hour, the wage differential between
college graduates and others is large and increasing. In 1994, college
graduates earned an average of 77 per cent more than high-school
graduates.4 Of course, people increase their productivity in college by 4
See ‘Generation
learning computer skills and other skills valued by employers. At the same X-onomics’
time, the sectors of the economy which employ unskilled workers have
seen fiercer foreign competition which drives down wages. In addition it
seems plausible that employers use college education as a proxy or a signal
of ability and high productivity and that this explains part of the wage gap.
Contrast this with Sweden’s labour market where workers with a
university education are paid 3.5 per cent more than workers who did not
get a university education (few Swedes go to university).5 5
See ‘The consequences
of kindness’
An insurance menu
Uninformed parties can screen by offering a menu of choices or possible
contracts to prospective (informed) trading partners who ‘self-select’ one
of these offerings. This type of screening was developed in an insurance
context by Rothschild and Stiglitz (1976). They show that, if the insurer
offers a menu of insurance policies with different premiums and amounts
of cover, the high risk clients self-select into a policy with high cover.
Example 4.6 illustrates how, in an insurance market, inefficiency due to
adverse selection can be ameliorated through the insurance company
offering clients two contracts: a low premium, partial cover policy and a
high premium full cover policy.
Example 4.6
49
28 Managerial economics
policies is to get clients to self select such that the low risk group
chooses the first policy and the high risk group the second. Assuming,
as before, that the insurance industry is competitive and therefore profit
on any policy is zero, the premiums are set according to Rl = plDl and
Rh = ph V.
Clearly, the low risk group prefers the fair (premium equal to expected
claim) insurance (Rl, Dl) to no insurance. Similarly, high risk individuals
prefer (Rh, V) to no insurance. What is most important though is that
the high risk group prefers (Rh, V) to (Rl, Dl). If it prefers the contract
designed for the low risk group then the insurance company cannot
break even. So we need:
ln(10,000 – 700) > (0.9) ln(10,000 – 0.01 Dl) + (0.1) ln(10,000 – 0.01 Dl + Dl – 7,000)
or
Dl < 1945.
The low risk group cannot be offered more than a 28 per cent
(1,945/7,000) partial cover.
Principal-agent problems
Principal-agent problems are an important class of moral hazard problems
in which one party, the principal, hires another party, the agent, to take
certain actions. The agent generally has more information than the
principal. In particular, the agent knows how much or how little effort
he made in pursuing the principal’s objectives. Since the principal and
the agent have different objectives the agent has to be given appropriate
incentives to act in the principal’s interest in order to avoid or limit
the moral hazard problem. In corporate governance, for example, the
principals are the shareholders and the agents are the managers.
Designing appropriate incentive systems is of great practical importance.
In the bank deposits insurance case, mentioned above, the government
50
Chapter 4: Asymmetric information
effort. A moral hazard problem arises: if the worker is not monitored or tablets’
Forcing contract
The employer could propose to pay the worker a lump sum L which gives
him his reservation utility if he makes effort e* i.e. L = u + C(e*) and zero
otherwise. Clearly, the participation and incentive compatibility constraints
are satisfied under this simple payment scheme. This arrangement is called
a forcing contract because the employee is forced to make effort e*. In
‘payment based on effort’ and ‘franchise’, the worker can choose his effort
level.
51
28 Managerial economics
Franchise
Suppose the worker keeps the profit of his efforts in return for a certain
payment to the principal. This can be interpreted as a franchise structure.
How should the ‘employer’ set the franchise fee F? The worker now
maximises P(e) – C(e) – F and therefore chooses the same optimal effort as
before: e* such that P'(e*) = C'(e*). The principal can charge a franchise fee
which leaves the worker with his reservation utility: F = P(e*) – C(e*) – u.
Example 4.7
52
Chapter 4: Asymmetric information
he wants to prevent the agent quitting his job or, in other words, the
participation constraint should be satisfied:
(1/3)U(x, 8) + (2/3) U(y, 8) ≥ 8, or
(1/3)(√x – 8) + (2/3)(√y – 8) ≥ 8. (1)
Also, given the compensation scheme (x, y) the agent should prefer
working to shirking (i.e. the incentive compatibility constraint should be
satisfied):
(1/3) U(x, 8)(2/3) U(y, 8) ≥ (1/2) U(x, 0) + (1/2)U(y, 0), or
(1/3)(√x – 8) + (2/3)(√y – 8) ≥ (1/2) √x + (1/2) √y. (2)
The constraints (1) and (2) can be rewritten as:
√x + 2√y ≥ 48 and – √x + √y – 48 ≥ 0.
respectively. Therefore the area above the upwards sloping line in
Figure 4. 1 represents the set of feasible compensation schemes.
1/2
y
(2)
48
24
(1)
1/2
48 x
Figure 4.1: Feasible compensation schemes
The risk neutral principal will want to maximise his expected revenue
which, given that the agent works, equals:
(1/3)(0 – x) + (2/3)(10,000 – y).
Maximising this is equivalent to minimising 2y + x which is achieved
(over the region of feasible compensation schemes) at √x = 0 and
√y = 48. This indicates that the principal should pay the agent only when
he secures the £10,000 revenue and then the payment should be 482 =
2,304. The principal thus expects a net revenue of:
(2/3)(10,000 – 2,304) = 5,130.67
which exceeds his expected net revenue of inducing no effort:
(1/2)(0) + (1/2)(10,000) – 64 = 4,936.
The principal will therefore motivate the agent to work.
The agent receives expected utility equal to his reservation utility of
8 when his effort can be observed. When effort is unobservable, his
expected utility is (1/3)(0 – 8) + (2/3)(48 – 8) = 24. His expected wage is
also higher when effort cannot be observed: (1/3)(0) + (2/3)(2304) = 1536
> 256. This higher expected wage is necessary to compensate the agent
for carrying risk. Ideally the principal should carry all risk since he is
risk neutral but, if the agent is paid a fixed wage, inefficiency due to
moral hazard results.
53
28 Managerial economics
Sample exercises
1. The value of a four year old Honda Accord to the owner is uniformly
distributed between £3,800 and £5,800. Whatever the value to the
owner is, the car is worth £200 more to a potential buyer. There are
many potential buyers for a four year old Honda Accord. Find the
equilibrium price.8 8
Hint: At any price p,
which cars are offered
2. A travel agent sells holidays to the Cayman Islands. Consumers are for sale? What are they
willing to pay £1,400 if the holiday is excellent and £800 if the holiday worth to a buyer?
is mediocre. The travel agent knows the quality of the holidays he sells
but the consumer does not.
a. If a fraction x of travel agents sell excellent holidays, how much is a
risk neutral consumer willing to pay for a holiday?
b. Suppose it costs a travel agent £1,000 to book a mediocre holiday
and travel agents are perfect competitors. Is there a market
equilibrium in which all holidays are mediocre?
c. Suppose it costs a travel agent £1,000 to book a mediocre holiday
and £1,100 to book an excellent holiday. Is there an equilibrium in
which all holidays are excellent?
d. Suppose it costs the travel agent the same amount, £1,000, to book
a mediocre or an excellent holiday. Find the market equilibrium
(equilibria). How much consumer surplus is generated at the
equilibrium (equilibria)?
e. For the scenarios of (c) and (d), if it were possible, would it be good
competition policy to ban the sale of mediocre holidays?
f. How can the inefficiencies resulting from asymmetric information
be avoided in this context?
3. McClean & Co. and McDirty & Co. are in the entertainment business.
McClean is interested in taking over McDirty. It does not know
McDirty’s value v under its current management but believes it to be
between £3 million and £4 million, all values equally likely. The value
v is known only to McDirty’s management team who will decide on any
takeover bid. Whatever the value v is, the company is worth 1.5v – £1.5
million to McClean.
a. Draw the value to McClean as a function of v and conclude that
McDirty is always worth more in McClean’s hands than under its
current management.
54
Chapter 4: Asymmetric information
Revenue
R = 10 R = 30
e=1 2/3 1/3
e=2 1/3 2/3
a. Calculate the expected revenue if the agent works hard and if he
shirks.
b. If the principal could observe effort, how much would she have to
pay to get the agent to work hard? To get the agent to make a low
effort? What is her net revenue in either case? What is the optimal
forcing contract?
c. Suppose the principal can only observe revenue, not effort, and
so has to pay the agent more for the high revenue than for the
low revenue if she wants the agent to work hard. Write down the
55
28 Managerial economics
56
Chapter 5: Auction and bidding
Aims
The aim of this chapter is to consider:
• the difference between private and common value auctions
• the structure of the auctions discussed and the optimal bidding
strategies
• the IID assumption on buyers’ valuations
• the effect of the number of bidders on the auction revenue and the
bids
• how a bidders’ ring works
• the revenue equivalence result
• the winner’s curse
• the effect of buyer risk aversion in a first price sealed bid auction
• the effect of seller risk aversion on his preference between first and
second price sealed bids.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• explain why bidding your true valuation is a dominant strategy in
English and second price sealed bid auctions
• derive the optimal bidding strategy in first price sealed bid auctions
with uniformly distributed private values
• analyse simple auctions of the type discussed.
Essential reading
Varian, H.R. Intermediate Microeconomics. Chapter 17.
Further reading
Boyes, W.J. and S.K. Happel ‘Auctions as an allocation mechanism in academia:
the case of faculty offices’, Journal of Economic Perspectives 3(3) 1989,
pp.37–40.
Capen, E., R. Clapp and W. Campbell ‘Competitive bidding in high-risk
situations’, Journal of Petroleum Technology 23(1) 1971, pp.641–53.
Cassady, R., Jr. Auctions and Auctioneering. (Berkeley: University of California
Press, 1967).
‘Inside the stockade’, The Economist, 2 April 1994, pp.69–70.
McAfee, R. and J. McMillan ‘Auctions and bidding’, Journal of Economic
Literature 25(2) 1987, pp.699–754.
Milgrom, P ‘Auction theory’, in Bewley, T.(ed) Advances in Economic Theory,
Fifth World Congress. (Cambridge: Cambridge University Press, 1987).
[ISBN 0521340446].
Milgrom, P. and R. Weber ‘A theory of auctions and competitive bidding’,
Econometrica 50(5) 1982, pp.1089–1122.
‘Speeding allowed’, The Economist, 16 April 1994, p.69.
‘Teetering’, The Economist, 27 August 1994, 7.
57
28 Managerial economics
Introduction
While the model of perfect competition involves many sellers competing to
sell, auction theory analyses situations in which many buyers compete to
buy from one seller. Auctions are a widely used alternative to the market
mechanism for transferring goods from sellers to buyers as the following
list of examples illustrates.
• Around 500 BC Babylonian women were auctioned off annually to
prospective husbands. The most attractive ones were offered first and
fetched high prices, while others were auctioned with large dowries
made up from the sale of the attractive women.1 1
Cassady (1967)
58
Chapter 5: Auction and bidding
Given that auctions are so prevalent, we should consider the reasons for
their use and the circumstances under which they are more appropriate
than other methods of sale. Auctions are never used to sell standardised
products for which there are competitive markets. The time and expense
needed to organise an auction and gather all the interested parties in the
same place would be wasted since the outcome would be the same as in
the market. However, auctions are virtually the only mechanism used to
sell special, unique items such as antiques, real estate, art, rare wine, oil
drilling sites and mineral leases. It is very difficult to post a price for such
items because of the uncertainty about demand for them; there are no
historical data a seller can use to form an estimate about potential buyers’
willingness to pay. Auctions are therefore used to determine prices for
these special objects.
Standard (or first bids are submitted the highest bid bid below your
price) sealed bid secretly and valuation
simultaneously
Dutch auctioneer starts with the price at which bid below your
high price and reduces the auctioneer valuation
it gradually until a stopped
bidder stops him
Second price sealed bids are submitted the second highest bid your true
bid secretly and bid valuation
simultaneously
The optimal strategy in a private value English auction is to stay in the auction until
your valuation is reached.
61
28 Managerial economics
The optimal strategy in an IID private value first price sealed bid auction is to bid the
expected value of the second highest valuation among the bidders assuming your
valuation is the highest.
I will not give a proof of this result but I will show you how it works in the
special case of uniformly distributed valuations. Suppose there are two
risk-neutral bidders: you and an opponent. Each of you has a valuation
which is known only to you and which is drawn independently from a
uniform distribution on [0.1]. Your valuation is v1 and you bid b1. Suppose
your opponent’s strategy tells him to bid a fraction f of his valuation v2 (i.e.
b2 = f v2). Would you ever bid above f ? If you bid above f you are sure to
win since b2 < f for v2 < l. If you bid below f your probability of winning is
the probability that the other bid is below yours:
P (win) = P(b2 < b1) = P(f v2< b1) = P(v2 < b1 / f ) = b1 / f.
Hence your expected gain from a bid b1 equals:
(v1 – b1) P(win) = (v1 – b1) (b1 / f ).
Note that there is a tradeoff between the probability and the profitability
of winning: if the bid increases, the profit (v1 – b1) decreases but the
probability (b1 / f ) of winning increases. Optimising the expected gain
with respect to b1 gives b1 = v1/2. This optimal bid is independent off: an
optimal response to any linear bid is to bid half your valuation. It follows
that the strategy pair in which bidders bid half their valuations forms a
Nash equilibrium. How does this relate to the general rule above? Well, if
you assume your value v1 is the highest, then your expectation of the other
bidder’s value is equal to v1/2.
Let us extend the analysis above to a sealed bid auction with n bidders,
each with a valuation drawn from the uniform distribution on [0,1].
Assume you are Bidder 1 and all your opponents use the same linear
bidding strategy: bi = fvi, i = 2,..., n. You win the auction if all of your rivals’
bids are below yours, that is, when bi = fvi < b1 for all i:
P (win) = P(v2< b1 / f and v3 < b1 / f and ...vn < b1 / f ) = (b1 / f )n-1.
62
Chapter 5: Auction and bidding
The optimal strategy in a private value Dutch auction is same as for a first price sealed
bid auction.
In essence, the Dutch auction and the first price sealed bid auction
describe the same game. Each player’s strategy consists of a function of his
valuation. The only (immaterial) difference is that, in a sealed bid auction,
the bid is submitted in writing and a Dutch auction is an oral auction. In a
Dutch auction, the bidding strategy can be interpreted as the plan to claim
63
28 Managerial economics
the item if the bid level goes down to the number predetermined in the
bidding strategy. A bidder in a private value Dutch auction does not gain
anything from his presence; he could send an agent to do the bidding.
Although Dutch and English auctions are both oral auctions, they are
very different from the bidder’s perspective. As mentioned before, in an
English auction, as the price increases, the bidder can assess at any time
whether he can gain by increasing the bid level. In particular, when the
bid level reaches the bidder’s valuation it is clear that he cannot make a
gain from the auction. In a Dutch auction, the bidder is not certain at any
price (below his valuation) whether he can increase his gain by waiting or
bidding. He knows that he will win the auction if he bids but he could get
a larger surplus (valuation minus bid) if he waits.
The optimal strategy in a private value second price sealed bid auction is to bid your
true valuation.
Note that we found before that this is also the optimal strategy in an
English auction. Bidding your valuation is a dominant strategy, as it is for
the English auction. Why? Suppose you bid below your valuation. Then
you will gain zero if you don’t win. If you win, then you would have won
bidding your true valuation and paid the same amount (the second highest
bid). So, bidding below your valuation is dominated by bidding your
valuation since bidding lower only decreases your chances of winning and
obtaining a positive pay-off. Now suppose you bid above your valuation.
If you lose, you would certainly have lost bidding your valuation (your
bid would have been lower). Hence you did not do better by bidding
higher than your valuation in this case. If you win and the second bid is
above your valuation you will get a negative pay-off. You do worse than
by bidding your valuation. If you win and the second bid is below your
valuation, you would have won bidding your valuation as well. Bidding
above your value is dominated by bidding your value since bidding above
only increases your probability of winning the auction when you don’t
want to win it.
Note that the optimality or dominance of the strategy ‘bid your
valuation’ does not depend on knowing other bidders’ valuations or their
distributions. In other words, whatever the other bidders’ strategies are,
you cannot do better than bidding your valuation. As was mentioned
for the English auction, the dominance of the optimal bidding strategy
makes second price sealed bidding an attractive auction form. Bidders
do not have to gather any information; they do not in fact have to think
strategically. They just have to be rational enough to realise that they
should bid their valuation.
64
Chapter 5: Auction and bidding
Auction revenue
An important issue for anyone considering the sale of an object at an auction
is how much revenue can be expected from such a sale and which type of
auction generates the highest revenue. Let us first look at the revenue to
the seller from first and second price sealed bid auctions using the example
of independent, uniformly distributed valuations. We will use the following
property of the uniform distribution. For a uniform distribution on [0,1],
the expected value is 1/2; if you make two independent draws from the
distribution, the expected value of the largest is 2/3 and of the smallest
1/3; in general, if you make n independent draws, the expected value of the
largest is n/(n + l), the second largest has expected value (n – 1)/(n + l) and so
on; the minimum has expected value 1/(n + 1).
There is no dominant strategy for a common value English auction and the
equilibrium bidding strategy depends on the structure of the information
among the bidders (who knows what?) and the possibilities for obtaining
additional information about the value during the auction.
To avoid the winner’s curse you should bid below your estimate.
bids estimates
v-d v
Figure 5.1
In competitive procurements there are frequent occurrences of the
winner’s curse. If you get the consultancy job you tendered for, you are
likely to have underestimated the costs. In common value competitive
bidding for, say, an offshore oil lease, you have to correct for the possibility
of overestimation. Capen, Clapp & Campbell (1971) analyse bids for
offshore oil tracts, auctioned by the US government in the late 1960s. They
present evidence that the winners may have been cursed! For example,
the winning bid for offshore Texas Tract 506 was US$43.5 million, nearly
three times the second highest bid of US$15.5 million. Failure to properly
discount their estimates may partly explain why some independent 6
See ‘You say you want
television firms are in trouble, having paid too much for their franchises a revolution’
which were sold by the UK government in a bidding process in 1991.6
If there are many bidders, you have to shade your bid even more in order
to account for the winner’s curse phenomenon. To see this, suppose
66
Chapter 5: Auction and bidding
each bidder estimates the value as its real value plus some (positive or
negative) error and bids his estimate minus some discount. Then the
probability of the bid of the winner exceeding the value is larger when
there is a large number of bidders since the probability that the maximum
of n IID variables exceeds any number increases in n. This is an interesting
observation because there is a natural tendency to bid more aggressively
when there are many bidders.
Note that I haven’t told you how to bid in a common value first price
sealed bid auction other than that you should discount your valuation.
This is because no simple optimal bidding strategy exists. How you
should bid depends on the nature of uncertainty about the value and the
information available to the bidders. As for the private value versions, the
optimal bidding strategy in a common value Dutch auction is the same as
in a common value first price sealed bid auction.
Phantom bids
As mentioned above, bidders in common value English auctions obtain
information by observing who drops out and when. If the number of
active bidders decreases rapidly, bidders are likely to think they have
overestimated the value and revise their estimate downwards. This
explains the temptation by sellers to use agents who keep bidding against
the last bidder. The disadvantages of using agents in this way are that
(a) the seller incurs a risk of the high bidder dropping out and one of
his agents ‘winning’ the auction and (b) bidders may account for the
possibility that they are bidding against an agent and shade their bids
accordingly. Clearly, using phantom bids skillfully is an art!
Can the auctioneer use phantom bids profitably in a private value English
auction? Yes, he can. Recall that the winner pays the second highest
valuation. If the auctioneer takes a phantom bid after all but one ‘real’
bidder have dropped out, he increases the seller’s revenue. The larger the
difference between the highest and the second highest valuation, the more
profitable the use of phantom bids will be.
In a second price (private or common value) sealed bid auction, an
unscrupulous seller could ‘cheat’ by inserting a phantom bid just below the
winning bid, thus increasing the price paid by the winner. This may partly
explain the unpopularity of the second price sealed bid auction. Of course,
the seller could only profitably cheat in this way if the bidders are naive. If
67
28 Managerial economics
the bidders account for the possibility of a dishonest seller, they will shade
their bids and they will tend to bid as in a first price sealed bid auction.
Do you know why? Think about what the winner pays if the seller uses a high
phantom bid.
Collusion
In some auctions it is possible for bidders to collude. For example, a
‘ring’ of bidders in an English auction could agree not to bid against each
other (i.e. not to bid when the current bid was made by one of the ring
members). In this way the ring can buy the object for less than would be
possible without the ring. The object is then sold in the ring to the highest
bidder and the profit is divided among the ring members. Rings are more
likely when bidders know each other. In 1994, property developers in
Hong Kong were accused of collusion in government land auctions.7 7
See ‘Speeding allowed’
Although bidders could also collude in sealed bid auctions, the anonymity
of the bidding process makes cheating (i.e. breaking the collusive
agreement) more likely. This may be the reason why sealed bid auctions
are preferred to English auctions in procurement. Especially in auctions
which are not one-off events but are repeated very regularly with more or
less the same bidder population (for example Treasury auctions), collusion
is likely. To counteract collusion it is advisable for the seller to withhold
information such as the identity of the winner and the winning bid. This
eliminates the possibility of colluders punishing the cheaters which in turn
increases the probability of cheating.
Risk aversion
In deriving optimal strategies we have implicitly assumed that bidders are
risk-neutral. Would we find different results if bidders are assumed to be
risk averse? The answer to this question depends on the type of auction
under consideration. In a private value English or second price sealed bid
auction the dominance of bidding your true valuation still applies. The
argument given for this dominance does not assume anything about the
bidders’ attitude to risk.
However, risk averse and risk neutral bidders do bid differently in a
private value first price sealed bid auction and a Dutch auction. When the
clock in a Dutch auction is ticking away, indicating lower and lower prices,
a risk averse bidder will become increasingly nervous as the price moves
below his value. He is likely to stop the clock earlier than a risk neutral
person with the same value. A risk averse person bids higher because he
is willing to pay more to avoid the zero pay-off associated with losing. If
this does not make sense to you intuitively, think of the auction as a lottery
in which there are two possible pay-offs: a surplus equal to the value of
the item minus the bid if you win and zero if you lose. The probability
of winning depends on your bid. If you are risk averse you may prefer a
low positive pay-off with a high probability to a high positive pay-off with
a low probability. It follows that a (risk neutral) seller can make more
revenue in a first price sealed bid auction than in a second price sealed bid
auction if the bidders are risk averse, hence the qualification referring to
risk neutrality in the revenue equivalence result.
68
Chapter 5: Auction and bidding
Example 5.1
Consider a private value first price sealed bid auction with two risk
averse bidders who have identical utility functions U(x) = x1/2 and values
drawn independently from the uniform distribution on [0,1]. Recall that
the Nash equilibrium strategy for risk neutral bidders is to bid half their
valuation. As before, assume Bidder 2 uses a linear bidding strategy: b2
= fv2. The probability that Bidder 1 wins with a bid b1 , remains
b1 /f. Bidder 1 maximises his expected utility U = (b1 / f )(b1 – v1)1/2 which
results in b1 = 2/3v1. The best response against any linear bid function is
b1 = 2/3v1. We have therefore shown that, for risk averse bidders with a
square root utility of money function, bidding 2/3 of their valuation is a
Nash equilibrium strategy. The seller gets 2/3 of the expected maximum
valuation (2/3), that is 4/9, which is higher than the expected revenue of
1/3 in the risk neutral version.
A risk averse seller will also prefer first price over second price sealed
bid auctions. The reason is that the variance of the winning bid is larger
in the latter auction form. Example 5.2 (which can be skipped if you are
frightened of integrals!) illustrates this.
Example 5.2
69
28 Managerial economics
Sequential auctions
In most auction models it assumed that the sale of an item can be
considered in isolation. In reality however, many auctions are of a
sequential or repeated nature (e.g. the government may have an annual
round of procurement tenders for stationery; several art objects may be
auctioned on the same day). In some procurement auctions the quantity
auctioned is not fixed but can be determined by the buyer (i.e. the buyer
solicits bids and his quantity demanded is given by a demand function q(p)
where p is fixed by the auction). In some share auctions a bidder bids a
price and a quantity he wants to buy. The winner will be the bidder who
offered the highest price who will get the quantity he bid. If there are any
shares left, the bidder who offered the second highest price will get the
quantity he bid at the price he offered and so on. Bidders in share auctions
may be allowed to submit several price-quantity bids. Conceptually this is
equivalent to submitting a demand curve.
I have already mentioned that repeated auctions are more likely to lead
to bidder collusion but there are other complications. The value of a good
to a potential buyer may depend on whether he has been able to acquire
complementary goods in an earlier auction. Imagine a property developer
interested in purchasing a substantial chunk of land to build a new
entertainment complex. Suppose the land is sold in relatively small land to
build a new entertainment complex. Suppose the land is sold in relatively
small plots. If the bidder is not successful in the early bidding rounds, his
valuations for plots coming up for sale later on may decrease dramatically.
In particular, the value of any plot may depend on whether he is able to
acquire the adjoining plots. Similarly, a construction company, with limited
capacity, values winning a road maintenance contract according to its
available capacity which depends on the number of contracts it has won
recently.
Another consideration in sequential auctions is the effect of winning in
an early round on your ability to bid later on. If you have a fixed budget
or are not able to borrow easily, winning the auction for the modem
painting may leave you resourceless and hence without a chance to win
the silverware auction later on. Other bidders may realise this and may be
able to use this fact to their advantage. By bidding against you to ensure
you pay a large amount of money for the painting, they hope to ensure
themselves a good deal in the silverware auction.
Other factors
Procurement contracts are not always awarded on the basis of price alone.
In many instances a buyer will consider quality issues as well. He could
do this either when the bids are opened (if he has asked bidders to specify
quality variables) or by preselecting the suppliers who are allowed to
tender. For reasons of public accountability, governments and other public
institutions are more likely to use the latter option, especially if quality is
not easily assessed objectively.
Time may be an important consideration in building contracts. When the 8
See ‘Speeding allowed’
Santa Monica freeway was damaged in the 1994 Los Angeles earthquake,
the California Department of Transportation (Caltrans) scrapped its
normal bidding procedures and construction firms were required to bid on
both the cost and the time needed to get the repairs done.8
70
Chapter 5: Auction and bidding
Conclusion
I have only discussed a few types of auctions but many more exist. For
the more advanced and interested reader I recommend Wilson (1979),
Milgrom and Weber (1982), Milgrom (1987) and McAfee and McMillan
(1987) as further reading. Even for the auctions described in this chapter
there are many variants and the assumptions which are made to enable
us to analyse these auctions would have to be modified to deal with each
variant. For example, it is quite common for the seller to set a reserve price
below which he is not willing to sell. This reserve could be several millions
of pounds for an Impressionist painting for example. An astute auctioneer
or seller may be able to generate a higher expected revenue by carefully
selecting a reserve price. In the models considered here, the number of
buyers is fixed but in reality it is endogenous because potential suppliers
trade off the cost of preparing a bid (which may include an ‘entry fee’)
against their expected pay-off from participating in the tender. It is fair to
say that game theory has offered significant insights into the mechanisms
of auctions and strategies for rational bidders. At the same time, many
practical issues are ignored in current auction theory which limits its value
for real life bidders.
Sample exercises
1. A house which is currently rented is going to be sold. There are two
potential buyers: the current tenant and an outsider. The owner solicits
a price from the outside buyer and allows the tenant to buy the house
if he matches this offer. The buyers and the owner know that the two
buyers’ values for the house are independent of each other and that
for each buyer the value is drawn from a uniform distribution on
[£100,000; £160,000].
a. How should the tenant decide whether or not to match the outside
offer?
b. What offer should the outside buyer make if her value is £120,000?
What offer should she make as a function of her value v?
c. (Only attempt this part if you feel brave!) What is the seller’s
expected revenue? Assuming there are two bidders, what would the
seller’s expected revenue be under an English auction?
2. In a private value first price sealed bid auction there are two bidders
with values v1 and v2 drawn independently from a uniform distribution
on [0, 1]. Assume bidder 2’s strategy is to bid b2 = v21/2. Show that the
optimal response of bidder 1 is to bid b1 = 2/3v1. Do these strategies
form a Nash equilibrium?
71
28 Managerial economics
72
Chapter 6: Topics in consumer theory
Aims
The aim of this chapter is to consider:
• the decomposition of the price effect into income and substitution
effects (Hicks and Slutsky method)
• the concepts of EV and CV
• the relationship between price elasticity and the effect of a price change
on revenue
• the relationship between price elasticity and the optimal markup
• the role of the income effect in generating backward bending
labour supply
• the tradeoff between risk and return in investment
• the importance of correlation between asset returns for risk reduction
• why everyone chooses the same portfolio of risky assets when a risk-
free asset is available
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• derive demand functions and labour supply functions given a utility
function
• derive the Slutsky equation
• derive Slutsky and compensated demand functions
• calculate elasticities
• explain the state-contingent commodities model
• set up the intertemporal choice problem and analyse the effect of
changes in the interest rate
• set up labour supply model
• derive the shape of the feasible set of risk-return combinations for
portfolios of two risky assets.
Essential reading
Varian, H.R. Intermediate Microeconomics. Chapter 17.
Further reading
Battalio, R.C., L Green and J.H. Kagel ‘Income-leisure tradeoffs of animal
workers’, American Economics Review 71(4) 1981, pp.621–32.
Biddle, J.E. and D.S. Hamermesh ‘Sleep and the allocation of time’, Journal of
Political Economy 98(5) 1990, pp.922-43.
Gravelle, H. and R. Rees Microeconomics. (Harlow: FT Prentice Hall, 2004)
third edition [ISBN 0582404878] Chapter 3 and 19. You may find this
treatment too theoretical and I do not expect you to study the material there.
Solberg, E.J. Microeconomics for Business Decisions (Lexington: D.C. Heath and
Co., 1992) [ISBN 0669167053] Chapters 4, 5, 6 and 7.
73
28 Managerial economics
Introduction
In this chapter my aim is to guide you through some fundamental concepts
in the theory of the consumer. You have studied many of these concepts in
your introductory course so I will leave it up to you to review some of the
material there.
Consumer theory deals with the individual consumer’s choice of how to
spend his income. The standard model of consumer choice assumes that
consumers maximise a utility function subject to a budget constraint.
In addition to a brief review of the theory of the consumer, I will show
you some of its important applications to finance, welfare economics,
uncertainty, intertemporal decision-making and labour supply.
You should know how the budget constraint is affected by taxes, quantity
discounts and vouchers. A voucher is an income transfer which can be
spent on a particular good.2 2
I recommend Varian
(2006) pp.29–31 on the
You should remember how demand for a good can be derived from the
Food Stamp programme
utility maximisation problem subject to a budget constraint. For the as an example.
two goods case, at an interior solution, the optimality condition for
consumer choice is that the marginal rate of substitution equals the price
ratio. In a corner solution one of the goods is not consumed. If the
goods are perfect substitutes you will always find a corner solution unless
the price ratio equals the marginal rate of substitution. You should be able
to derive consumer demand algebraically. Study the derivation of demand
for Cobb-Douglas utility U(x1, x2) = x c1x d2 in Varian (2006) p.93. The
resulting demand functions
x1 = cm and x2 = dm where m is income.
(c + d )p1 (c + d )p2
Note that for Cobb-Douglas utility the expenditure shares of each good are
constant (i.e. p1x1/m = c/(c + d) and p2x2/m = d/(c + d ) do not vary with prices
or income). Empirically this implies that we may be justified in assuming
Cobb-Douglas utility if the fraction of their income which consumers spend
on various products is relatively stable. Once we know a consumer’s utility
function we can assess how he is affected by price and income changes;
this is useful in evaluating policy decisions regarding taxation for example.
74
Chapter 6: Topics in consumer theory
When the price of a good changes relative to prices of other goods, there
are two effects. The first one is the substitution effect which represents
the fact that the rate at which you can trade off the good under study
with another good has changed. The second effect is the income effect
which represents the idea that your purchasing power increases after a
price decrease and decreases after a price increase. There are two versions
of the substitution effect. In the Slutsky version the budget line rotates
through the original consumption bundle until it has the same slope as the
new budgetline. The tangency of this rotated budgetline and the highest
possible indifference curve determines the change in consumption due to
the Slutsky substitution effect. Since the rotated budgetline passes through
the original consumption bundle, the latter is still affordable and in that
sense purchasing power has remained constant. In the Hicks version
the budgetline rolls under the original optimal indifference curve until it
has the same slope as the new budgetline and the new tangency (with the
same indifference curve so that utility is now held constant) determines
the change in consumption due to the Hicks substitution effect.
The Slutsky demand for a good depends on the initial consumption
bundle since we are holding the consumer’s purchasing power constant in
the sense that he can still (just) afford to buy his original bundle. Figure
6.1 illustrates the Slutsky substitution effect and demand.
x2
A
C B
x1
a c b
75
28 Managerial economics
original prices for which A is the optimum are p10 and p20 and the quantities
consumed at A are x10 and x20. From the figure it is easy to see that the
Slutsky demand (x1s) which is obtained by rotating the budgetline through
A as the price of good I decreases and drawing the relationship between
price and quantity, is in fact ordinary demand (x1) where income is held
such that A is just affordable or:
x1s ( p1, p2, x10, x20) = x1(p1, p2, m = p1x10 + p2x20) (1)
To derive the Slutsky equation, we take the derivative with respect to p1:
∂x1s ( p1, p2, x10, x20)/ ∂p1 = ∂x1 ( p1, p2, m = p1x10 + p2x20)/ ∂p1
+ (∂x1( p1, p2, m = p1x10 + p2x20)/ ∂m)x10
which can be rearranged to give:
∂x1 ( p1, p2, m = p1x10 + p2x20)/ ∂p1 = ∂x1s( p1, p2, x10, x20)/ ∂p1
– ∂x1( p1, p2, m = p1x10 + p2x20)/ ∂m)x10 (2)
This last equation decomposes the price effect into the Slutsky substitution
effect and the income effect.
Example 6.1
Assume Cobb-Douglas utility U(x1, x2) = (x1, x2)1/2 and original prices
p10 = p20 = 1 and income m = 100. From the section on ‘Utility, budget
constraint and demand’, we know that the ordinary demands are given
by x1 = m/(2p1) and x2 = m/(2p2) and hence x10 = x20 = 50. From (1) we know
that Slutsky demand is given by:
x1s( p1, p2 , x10 = 50, x20 = 50) = x1( p1, p2, m = p1x10 + p2x20 = 50( p1+p2)) =
25( p1 + p2)/p1
This implies that, when the price of good 1 decreases from 1 to 1/2,
ordinary demand for good 1 increases from 50 to 100 whereas Slutsky
demand increases from 50 to 75. The difference between 75 and 100 is
due to the income effect. You should check the Slutsky equation (2) for
this example:
–75/(2p12) = – 25p2 /p12 – 1/(2p1)50 = –150 for p1 = 1/2 and p2 = 1.
76
Chapter 6: Topics in consumer theory
x2
A C
B p1
1
x1
p 10
p0
1
p1
1
x1
Example 6.2
Suppose utility is given by U(x1, x2) = (x1, x2)1/2. What do the compensated
demands look like? The optimisation problem is:
min p1x1 + p2 x2 s.t. U(x1, x2) = u0.
From the tangency condition we know that the price ratio has to equal
the marginal rate of substitution or:
77
28 Managerial economics
Example 6.3
Using the same utility function U(x1, x2) = (x1, x2)1/2 and initial prices
and income p1 = p2 = 1, m =100, we can calculate the values of the three
measures of changes in consumer welfare when p1 decreases to 1/2. CV
is the amount of money we can take away from the consumer after the
price decrease to bring him back to his original utility level. To calculate
CV we need to know how optimal utility is affected by a change in price.
If we substitute the demand functions x1 = m/(2p1) and x2 = m/(2p2) into U
we find the indirect utility function:
78
Chapter 6: Topics in consumer theory
V = m/(2( p1 p2)1/2)
so that utility after the price change is:
V1= m/(2(1/2.1)1/2) = m/21/2
and utility before the price change is:
V0= m/(2(1.1)1/2) = m/2
CV is the amount of money we can take away from the consumer to
leave him as well off as initially. Hence (m – CV)/21/2 = m/2 or
(100 – CV)/21/2 = 50 which gives CV = 29.3. EV is the amount of money
you have to give the consumer to make him as well off as after the price
decrease. This means that EV is such that:
(m + EV)/2 = m/21/2 or (100 + EV)/2 = 100/21/2
which gives EV = 41.4. The change in consumer surplus due to the price
decrease is the area to the left of demand between the original and the
new price, hence:
1
m
CS = ∫ dp = 50(log(l) – log(1/2)) = 34.7
½
2p1 1
Elasticity6 6
Read Solberg (1992)
Chapter 4; Varian (2006)
Elasticity measures tell us how responsive demand is to changes in price, Chapter 15
income, prices of substitutes or complements, etc. Whether we look at
price elasticity or income elasticity or cross price elasticity, the measure is
always defined in proportional terms. It tells us what percentage change in
demand we can expect if the other variable changes by one per cent.
You should be aware that elasticities are time-dependent. For most
products it is unrealistic to assume that the price and income elasticities
will be unchanged over a long period of time. The simplicity, at least in
theory, of calculating values for, say, income elasticity is misleading. For
marketing purposes this figure has to be interpreted carefully. Let us take
the example of Malaysian car sales. A 40 per cent rise in incomes between
1987 and 1991 was accompanied by a 290 per cent rise in car sales, giving
a rough estimate of income elasticity of 7. Should the car industry attach
much importance to this number? A closer look at the car market and the
market for many goods in fast growing economies reveals that households
do not consume a good when their income is below a threshold level
but as soon as their income reaches the threshold level they do buy. This
means that, even with small increases in per capita income, large increases
in purchases could occur. At the same time, a large increase in per capita
income may not have a significant effect on sales if the households
whose income increased were already above the threshold level. Clearly,
it is more useful to have a good forecast of the growth in percentage of
households above the threshold than of growth in per capita income.7 7
See ‘Buyers and
builders’
Whether demand for a good is price elastic depends on several factors
including:
• the number and closeness of substitutes. The elasticity of
demand for a particular brand of cigarettes is large whereas the
demand for cigarettes as a whole is inelastic. The demand for petrol
and salt is inelastic
• the period of time over which the response to a price
change is assessed. For many goods if you study the response after
79
28 Managerial economics
∂R(p)
∂p
=p
∂q(p)
∂p
+ q(p) =
( p ∂q(p)
q(p) ∂p
+1
( q(p) = (1- η) q(p)
∂R
∂q
= p(q) + q
∂p
∂q ( q ∂p
= p 1+ p
∂q
( (
= c or p 1– 1η
( =c
p=c
( η
η–1
(
η
The optimal markup over cost η – 1 (η >1) is decreasing in the price
80
Chapter 6: Topics in consumer theory
81
28 Managerial economics
employed
450 line
m
m-p
unemployed
0 b-p m-p
income in state 2
certainty line
A income in state 1
82
Chapter 6: Topics in consumer theory
The iso-expected income line is the locus of income bundles (m1, m2)
which have the same expected value E(m). The equation for the iso-
expected income line for given probabilities (q1, q2) is q1m1 + q2m2 = E(m).
The slope of this line is also – q1 /q2 which implies that the tangency of an
indifference curve and an iso-expected income line occurs at the
intersection of the indifference curve with the certainty line. This is
illustrated in Figure 6.5.
m2
certainty line
B
E
e
c
C iso-expected income line
m1
b c e
83
28 Managerial economics
lend at interest rate r. If he does not consume in period 1 then the amount
available for consumption in period 2 equals m2 + (1 + r)m1. Similarly, if
he plans not to consume in period 2, he can borrow an amount m2 /(1 + r)
in period 1 and hence pay back m2 in period 2. In general, consumption in
period 2 will equal income in period 2 plus any savings from period 1 or
minus any repayments of loans from period 1:
c2 = m2 + (1 + r) (m1 – c1). (6)
Equation (6) is the equivalent of the budget constraint in the standard
consumer choice problem. Note that, as the interest rate r changes, the
budget line rotates through the endowment point E, becoming steeper
with an increase in r and flatter with a decrease in r.
If the individual can borrow but not lend, he has to choose a point on the
line segment to the right of E and, if he can lend but not borrow, he has to
choose a point on the segment to the left of E. If the borrowing and
lending interest rates are different, the budget constraint is piecewise
linear (and concave if the borrowing rate is higher than the lending rate)
with a kink at E.
c2
m2+ lending
(1+r)m1
E
m2
borrowing
c1
m1 m1 + m2/(1+r)
Why?
At the optimum the MRS equals the slope of the ‘budget line’ –(1 + r)
which implies that each individual saves or borrows such that the rate r at
which he can trade current for future consumption in the market coincides
with the rate at which he wishes to make these trades.
Using this model, we can derive the amount of saving or borrowing every
individual in the market plans to do, given the interest rate. The market
interest rate can be determined by plotting the total amount of saving and
the total amount of borrowing against the interest rate. You read off the
equilibrium interest rate where the two curves intersect.
84
Chapter 6: Topics in consumer theory
Example 6.4
Suppose the consumer has utility function U(c1, c2) = ln(c1) + a ln(c2),
where a (< 1) is a constant. The return on loans to and from the bank
is r and income is m1 in period 1 and m2 in period 2. The consumer’s
optimisation problem is:
Max ln(c1) + a ln(c2) s.t. c2 = m2 + (m1 – c1)(1 + r).
If we substitute the constraint into the objective function we obtain:
ln(c1) + a ln(m2 + (m1 – c1)(1 + r)).
Setting the derivative with respect to c1 equal to zero, we find:
1/c1 + (a/(m2 + (m1 – c1)(1 + r))(–1)(1 + r) = 0
which can be solved for c1:
c1* = (m1 + m2/(1 + r))/(1 + a)
If the consumer can lend but not borrow then the solution remains
unchanged as long as c1*< m1. If c1* > m1 i.e. m2 > a m1(1 + r) then the
consumer would like to borrow but, if he is prevented from doing so, he
will consume his endowment:
(c1, c2) = (m1, m2).
c2 c2
b c
b
c
a aE
E
c1 c1
(a) borrower (b) saver
85
28 Managerial economics
As an exercise, show that a saver can be made better or worse off by an interest rate
decrease. Note that a saver’s initial consumption plan is not feasible after the interest
rate decrease
Labour supply14 14
Read Solberg (1992)
Chapter 5; Varian (2006)
The standard consumer choice problem is easily rephrased in the context Chapter 9
of the labour supply decision. The ‘goods’ are leisure l and a composite
good c (with price = 1) which can be interpreted as ‘income available to
spend on consumer goods’. The budget constraint indicates how much
consumption is possible for each choice of leisure and its implied choice of
work time given a wage rate w. The worker may have non-labour income
M which is taken into account in the formulation of the budget constraint.
In addition to the monetary constraint, the consumer also faces a time
constraint: work time and leisure time have to add up to 24 hours per day
or, more generally, H hours per time period. The worker’s optimisation
problem, represented graphically in Figure 6.8, is: Max U(c,l) s.t. c =
M + w(H – l) and at the optimum the MRS equals the price ratio-w.
86
Chapter 6: Topics in consumer theory
C
M+w(H-1)
C*
1
1* H
87
28 Managerial economics
Example 6.5
∂U/∂c αcα–1l1–α 1 αl 1
= α = w or =
∂U/∂l c (1–α)l–α (1–α)c w
88
Chapter 6: Topics in consumer theory
(
l = (1-α) H +
M
w
(
and the labour supply curve is therefore:
H – l = αH – (1–α) M .
w
The number of hours worked H – l, is increasing in w and hence, for
this particular utility function, the labour supply curve is not backward
bending.
The simple model discussed above can be used to analyse many interesting
questions. Suppose the government pays a fixed amount in welfare benefit
for unemployed workers (i.e. if they work zero hours they get the benefit
but if they work at all, they do not).
You should be able to show that this provides an incentive to be idle especially if the
wage rate is low
90
Chapter 6: Topics in consumer theory
return (i.e. his indifference curves in the risk-return plane are upward
sloping: expected return is a ‘good’ whereas risk is a ‘bad’). The level of
risk associated with an investment is usually measured as the standard
deviation of the return. (This may be a good time to look for your statistics
notes!) Investments with larger standard deviations have to be rewarded
by greater expected returns. For example, average annual returns for
venture capital from 1945 to 1993 were 15.9 per cent and the standard
deviation in the returns was 25.5 per cent; stocks included in the S&P
500 averaged a return of 11.7 per cent over the same period and a
correspondingly lower standard deviation of 16.3 per cent. The exceptions
to the rule were gold and silver which had low average returns of about
five per cent and very high standard deviations of 25.8 per cent and 55.8
per cent respectively.21 21
See ‘Risk and return’
91
28 Managerial economics
to attain the combinations of expected return and risk given by (7) and
(8). Consider the (easier) case of ρ = 1 first. For ρ = 1,
Varp = (α σA + (1–α) σB)2 or σB = α σA + (1–α) σB
which combined with (7) implies that (p, rp) lies on a straight line
between (A, rA) and (B, rB) as is shown in Figure 6.10.
r
B
ρ <1
r ρ =1
A
σ σ
A B σ
92
Chapter 6: Topics in consumer theory
r
B
C
x
x
x
xx
x
xx
x x x
r x
f xx
xx
x
x x x
xx
x
Sample exercises
1. Goods x and y are perfect substitutes and consumers have utility
functions of the form U(x, y) = x + y. Consumer i has income mi.
Derive and draw his demand for good x. Derive and draw the market
demand for good x.
2. True/False. If the income elasticity is positive, the compensated
(Hicks) demand curve is steeper than the ordinary demand.
3. Joanna likes lipstick (x1) and earrings (x2) which give her utility
U(x1, x2) = x1x2. Her pocket money is M per week and lipsticks and
earrings cost p1 = 4 and p2= 1 respectively. Suppose p1 falls to 1. By
how much would her pocket money have had to increase to make
her as well off as after this price change? By how much could her
pocket money be reduced while keeping her as well off as before the
price decrease? Which is the compensating variation? Which is the
equivalent variation? Calculate her increase in consumer surplus due
to the price decrease.
4. The price elasticity of demand for wixies is 1.5. The wixies producers
form a cartel and agree wixies quotas which will have the effect of
reducing output such that price rises by 10%. What is the effect on
total revenue? How does your answer change if the price elasticity of
demand is 0.75?
5. Using the state-contingent commodities model, show graphically
that, for full insurance, a risk averse individual is willing to pay an
insurance premium which is higher than his expected loss.
6. A consumer lives for two periods and receives income 100 in each
period. In the first period he can invest some of his income in a
stock which has a return r of 10% or 20%, each equally likely. His
utility function U(c1, c2) is a function of his consumption levels in
both periods: U(c1, c2) = c12 + 0.7 c22 for c1, c2 ≥ 0. Find the optimal
consumption plan.22 22
Hint: maximise
expected utility
7. A worker can choose the fraction of a day, y, that he is not working
(so he works a fraction 1 – y of a day). He derives utility from
consumption goods c (assumed to have price = 1) and leisure y
according to U(c, y) = c4y2. He has a fixed nonlabour income of M per
day and the wage rate is w per day. Write his budget constraint and
derive his labour supply curve. Is it backward bending at high wages?
8. Empirical studies show that (a) for male employment small changes
in the wage rate do not affect the number of hours worked and (b)
the number of hours women work increases when taxes fall because
of their increased labour market participation. Explain these findings
graphically.
9. Portfolio theory assumes that investors care about expected return
and risk. Draw indifference curves for a risk neutral investor.
10. The return on stock A will be –10% or +20%, each with probability
0.5. The return on stock B will be –5% or +15%, with probabilities
0.3 and 0.7 respectively. The correlation coefficient between the
returns of the two stocks is 0.6. Calculate the expected return and
the variance of the return for each stock. Calculate the covariance
94
Chapter 6: Topics in consumer theory
between the returns of the stocks. Can you decrease your risk by
investing in a combination of these two stocks rather than in either of
the stocks?
11. Suppose the risk-free rate of return is 5% and you consider investing
in a risky asset with r = 10% and σ = 4%. Show all combinations of
risk and expected return which are feasible if you invest in a portfolio
consisting of the risk-free asset and the risky asset. If you are willing
to carry a risk of 3% but not more what is the best portfolio?
12. During the past six years the returns on Air Angleterre and French
Airways have moved as follows:
Year 1 2 3 4 5 6
AA 25 –16 18 5 13 2
FA 10 –3 15 3 20 8
Calculate all the parameters you need to derive the locus of feasible
expected return-risk combinations if you invest a share of your budget
in AA and the remainder in FA.
95
28 Managerial economics
Notes
96
Chapter 7: Production, factor demands and costs
Aims
The aim of this chapter is to consider:
• how the smooth isoquants continuous production model relates to the
linear production model
• the difference between conditional and unconditional input demands
• the problems involved in estimating production and cost functions
• the MRP = ME rule for determining unconditional input demands
• why imposition of a minimum wage may increase employment in a
monopsony
• why the industry demand for an input is not necessarily the horizontal
sum of firm demands
• why MR should equal MC in each producing plant of a multiplant firm.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• derive the cost function from a given production function i.e. derive
conditional input demands (input as a function of output) and write
the expenditure on these inputs as a function of output
• derive unconditional input demands for monopsony, monopoly
and price-taking firms
• solve the problem of which plants(s) to operate and at what level for
a two plant firm given the plants’ cost functions and the demand
function.
Essential reading
Varian, H.R. Intermediate Microeconomics. Chapters 18, 19 and 26.
Further reading
‘Doleful’, The Economist, 9 October 1993, p.17.
‘The trade unions scent a victory’, The Economist, 3 September 1994, pp.27–28.
‘Turning in to the future’, The Economist, 5 September 1992, pp.27–28.
Moroney, J. ‘Cobb-Douglas production functions and returns to scale in US
manufacturing industry’, Western Economic Journal 1967, pp.39–51.
Introduction
What I want to do in this chapter goes under the heading of ‘theory of the
firm’ in most microeconomics textbooks. However, over the last 15 years or
so, there has been a growing interest in what determines the boundaries
of the firm, how firms are organised internally, how their organisational
design can be explained in terms of efficiency and so on. All of these topics
are elements of a ‘theory of the firm’. The neoclassical view of the firm
97
28 Managerial economics
In Figure 7.1, the change in the input ratio corresponds to the difference in
slope of rays OA and OB and the change in the MRTS corresponds to the
difference in slope of the lines a and b tangent to the isoquant. Clearly, if
for the input ratios determined by A and B, the change in MRTS is larger,
98
Chapter 7: Production, factor demands and costs
then the isoquant is more convex and is smaller. The extreme values for
s are 0 for perfect complements (L shaped isoquants) and ∞ for perfect
substitutes (linear isoquants).
Activity
Why? Convince yourself by drawing a graph.
x2 B
a
0 x1
capital
PP1
50 A
PP2
B
10
man-hours
10 100
99
28 Managerial economics
100
Chapter 7: Production, factor demands and costs
x2
isocost line
x2*
0 x1
x1*
101
28 Managerial economics
Example 7.1
102
Chapter 7: Production, factor demands and costs
Example 7.2
Now let us consider the more complicated case of a firm which is not a
price taker. In the output market this means that the market price of the
firm’s output depends on how much it decides to sell. In the input market
similarly it means that the price the firm pays per unit of input depends
on the quantity it buys. As an example consider a university which is the
only employer of cleaners in a small town. If the university wants to hire
10 cleaners it can offer a wage of £7 per hour. However, if 11 cleaners are
needed, the wage has to rise to £7.50 per hour (the supply of cleaners
is upward sloping). If the university cannot discriminate between the
workers, the marginal cost of the 11th worker is not just his salary of £7.50
but also the increase in the other workers’ salary: £7.50 + £0.50*
10 = £12.50. Intuitively we can deduce that, if a firm is not a price taker, it
will hire fewer people because it takes into account the effect of its demand
on the wage level whereas a price taking firm ignores this externality.
A monopsony is a firm which is the only buyer in a market with many
sellers who take the market price as given. A monopsony could be a
103
28 Managerial economics
ME
w*
MRP
L
L*
104
Chapter 7: Production, factor demands and costs
Example 7.3
ME
w(L) MRP
supply
demand w(L)
A B
wmin
w* w*
L' L* L L* L' L
(a) Price takers (b) Monopsony
105
28 Managerial economics
106
Chapter 7: Production, factor demands and costs
w0
w1
∑ MRP (p0)
∑ MRP (p1)
Example 7.4
107
28 Managerial economics
Monopoly
Let us consider first what happens if the industry consists of one firm, a
monopoly – monopsony. Its total expenditure on labour equals
Lw = L (I + L) so that marginal expenditure equals ME = 1 + 2L.
Marginal revenue for the monopolist is MR = 2 – 2q = 2 – 2L1/2 and
MP = (1/2) L–1/2. Applying the rule MRP = ME, (and doing some
numerical work because this equation cannot be solved analytically),
gives an unconditional input demand of L* = 0.18. The corresponding
wage can be found from the labour supply function as w = 1.18.
Given its demand for labour, we know from the production function
how much output the monopolist produces: q = 0.42, and hence final
product price (from the demand function) is p = 1.58.
Perfect competition
Now let us see what happens if a number of (identical) firms in this
industry behave as price takers in input and output markets. Each firm
sets MRP = p MP equal to ME = w or p /(2L1/2) = w. To determine the
industry demand for labour we cannot just add these individual demand
curves as we have seen above. We need to take the output price effect
of increased input demand into account. The optimality condition for
the firm becomes (2 – nq)/(2L1/2) = w, where n is the number of firms
in the industry or, after substituting for q, (2 – nL1/2)/(2L1/2) = w. We can
solve this latter equation for L and find L = (w + n/2)–2 as the amount of
labour used by each firm. The total demand for labour in the industry
is n times this amount: L′ = n (w + n/2) –2 = 4n(2w + n)–2. The equilibrium
wage is such that demand and supply of labour are equal: 4n(2w + n)–2 =
w – 1. This equation cannot be solved analytically. The table below gives
the equilibrium wage (rounded to two decimal places) for some specific
values of n. It is not difficult to show that the limiting value of w, as the
number of firms n increases, equals 1.
108
Chapter 7: Production, factor demands and costs
w L* q Q p Π
monopsony 1.18 0.18 0.42 0.42 1.58 0.45
n = 10 1.26 0.26 0.16 1.6 0.4 0.03
n = 20 1.16 0.16 0.09 1.79 0.21 0.01
109
28 Managerial economics
110
Chapter 7: Production, factor demands and costs
MC1 MC2
∑ MC
MR
Example 7.5
111
28 Managerial economics
2 is largest when F1 > 0.83 and F2 < 21.67 + F1. The north-east corner
of Figure 7.8 corresponds to the scenario where fixed costs are too
large at both plants so that it is impossible for the firm to make positive
profits.
F2
22.5
0.83 20 F1
products and may hide important factors such as economies of scope (i.e.
the scenario in which it is cheaper to produce several products together
rather than separately).
Using time series data leads to additional problems. To get reliable results,
we need a reasonable sample size of, say, 40 observations. Ideally we
should use monthly observations; it is unreasonable to assume that firms
are using the same production technology (and hence operate according
to the same cost function) over a period of 40 years! In agriculture for
example mechanisation has dramatically changed the labour intensity of
many production activities over timespans of 10 or 20 years.
Remember that in the definition of the cost function it is assumed that
firms are using the most efficient combination of inputs. If the real life
firms in the sample are not operating efficiently, the estimated cost
function will be biased. In fact, we cannot expect firms to be operating
‘efficiently’ at all times; this would require perfect foresight and
immediate adaptation of scale of inputs to changed market environments.
An alternative method which is used to estimate the cost function is the
engineering approach. Engineers are asked to estimate the quantity of
inputs such as plant size and machinery required for given levels of output.
To estimate the cost function, these requirements are simply multiplied by
the price of the inputs. Because of the technical nature of this exercise, the
costs of controlling and managing are often overlooked.
Sample exercises
1. Draw an isoquant map representing increasing returns to scale up to
some output level and then decreasing returns to scale.
2. You can cook meals from ‘scratch’ using fresh ingredients or you can
buy frozen dinners. Draw an isoquant assuming the only inputs in meal
production are time and money.
3. For the linear version of the production model where every production
process is characterised by a ray through the origin in the isoquant
plane, show (graphically) the effect of technological progress which
enables the same production levels as before with less labour input.
4. In many LDCs, production is constrained by the availability of capital.
Show graphically (on an isoquant map) how the cost function reflects
this constraint on the amount of capital available. How would you find
the cost function analytically?
5. A firm’s production function is given by q(K, L) = K1/3 L2/3 and it faces
fixed input prices r (for capital) and w (for labour).
113
28 Managerial economics
114
Chapter 8: Topics in labour economics
Aims
The aim of this chapter is to consider:
• the role of unemployment both as a consequence and a necessary
condition in efficiency wage theory
• the dependence of the efficiency wage on the size of a potential
gain from cheating, the probability of being caught and unemployment
benefit
• why an individual firm’s demand for labour could increase when
it introduces efficiency wages
• why we observe long-term employment
• the role of promotions in the internal labour market.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• discuss the difficulties in attempting to apply standard neoclassical
models of labour supply and demand to internal labour markets
• analyse a simple efficiency wage model
• set up and solve the minimum cost implementation problem
• give explanations for increasing wage profiles
• discuss Frank’s theory of wage compression
• discuss managerial and executive compensation and the role of
performance related pay in this context
Further reading
‘A bit rich’, The Economist, 26 November 1994, p.89.
‘A racket in need of reform’, The Economist, 27 August 1994, pp.21–28.
‘Bosses on the run’, The Economist, 28 January 1995, p.26.
Frank, R.H. ‘Are workers paid their marginal products?’ American Economic
Review 74(4) 1984, pp.549–71.
Lazear, E.P. ‘Agency, earnings profiles, productivity, and hours restrictions’,
American Economic Review 71(4) 1981, pp.606–20.
‘Low risks, high rewards’, The Economist, 11 December 1993, pp.123–25.
Main, B.G., A. Bruce and T. Buck ‘Total hoard remuneration and company
performance’, (Discussion paper, Department of Economics, University of
Edinburgh, 1994).
Malcomson, J.M. ‘Work incentives, hierarchy, and internal labour markets’,
Journal of Political Economy 92(3) 1984, pp.486–507.
‘Nicely does it’, The Economist, 19 March 1994, p.94.
‘Ordinary deaths’, The Economist, 5 November 1994, p.74.
Ransom, M.R. ‘Seniority and monopsony in the academic labour market’,
American Economic Review 83(1) 1993, pp.221–33.
Shapiro, C. and J.E. Stiglitz, ‘Equilibrium unemployment as a worker discipline
device’, American Economic Review 74(3) 1984, pp.433–44.
‘The high price of freeing markets’, The Economist, 19 February 1994, pp.43–44.
‘The sour taste of gravy’, The Economist, 5 November 1994, p.50.
115
28 Managerial economics
Introduction
The previous chapters have provided all the ingredients of the traditional
neoclassical model of the labour market. The supply of labour is derived
from individual utility maximising decisions regarding consumption of
leisure and other goods. The demand for labour by a firm is derived from
the marginal revenue product rule which says that firms hire an amount
of labour such that the marginal return (the marginal revenue product
or MRP) equals the marginal expenditure. The total demand for labour
(as for other inputs) is obtained by summing the firms’ demands taking
into account any effects of this aggregation on the marginal revenue of
output (see the section on industry demand for inputs). The equilibrium
employment and wage levels are determined by the intersection of supply
and demand.
Although the traditional neoclassical framework summarised above is
undoubtedly an elegant construction which provides a wealth of testable
predictions, as it stands, it is neither realistic nor a great help to managers.
For a start, the model cannot explain involuntary unemployment. If
there is unemployment (demand less than supply), wages simply adjust
downwards until an equilibrium is reached. This ignores, among other
things, the effect of minimum wage regulations, welfare payments and
union power. Furthermore, downward pressure on wages can only exist
if the pool of unemployed workers is available where they are needed; if
they are in a geographically different location they may face prohibitive
moving costs. Also, the theory predicts that wages should be continually
adjusted to reflect a worker’s productivity whereas in practice wages are
fairly stable and are almost never adjusted downwards. Similarly, firms are
assumed to adjust the size of their workforce continually in response to
changes in demand conditions whereas in practice labour turnover, at least
in Europe, is limited.
The MRP rule is hardly a practical guideline for managers making hiring
decisions and setting wage levels. At the time a worker is hired, there
is uncertainty about his productivity. This problem is likely to be worse
for skilled or professional workers for whom, even after they have been
hired, it is difficult to measure performance or productivity. Managers do
not treat workers like other inputs. In addition to making decisions about
how many workers to hire and how much to pay them, there are the
important problems of whom to hire, whether and how to train them and
how to motivate them. To explain certain features of the internal labour
market such as promotions, life-time employment with the same company,
increasing wages over a career path etc. different models or at least
variations on the conventional framework are needed.
Efficiency wages
Standard labour economics models assume that labour is of uniform
quality and that this quality can be observed by the employer. Indeed,
the employer is assumed to base the worker’s reward on his quality
or productivity. In reality of course employers do not have perfect
information about workers’ quality or (lack of) effort. If the employer
cannot detect ‘cheating’ such as entertaining friends instead of customers
on the business expense account, selling company secrets, taking bribes,
working less than contracted for, etc. then workers will cheat. Even when
116
Chapter 8: Topics in labour economics
an employer can detect these activities, the possibilities for punishment are
limited. According to recent reports in China’s Legal Daily, a worker in an
American joint venture company was forced to lick clean a glass product
which the supervisor found dirty and a woman who stole two pairs of
shoes from a Taiwanese factory was made to sit in a dog cage.1 These types 1
See ‘Ordinary deaths’
of punishments are not ordinarily available to employers.
In many cases, the only means of disciplining a worker is to fire him or
her. However, if the terminated worker can immediately find a new job
at the same wage (there is no unemployment), even termination is not
an effective punishment and hence threat of termination will not induce
good behaviour. Only when workers fear that they will not be able to find
a job or at least not as good a job, are they deterred from cheating by the
threat of termination. The question then arises whether by paying workers
a wage premium in the form of an efficiency wage, the employer can
induce good quality work. Note that ‘good quality work’ can have many
interpretations ranging from higher productivity and better attitudes
towards customers to reduced absenteeism and staff turnover rate. If
workers are paid a wage higher than the market wage, they incur a real
penalty when they are fired. This question was examined by Shapiro
and Stiglitz (1984) in their seminal paper on efficiency wages.2 Shapiro 2
Yellen (1984) provides
and Stiglitz focused on the consequences of payment of efficiency wages a good overview of
efficiency wage models
on unemployment. If firms find it profitable to pay higher wages., the
argument goes, then the demand for labour decreases which creates
unemployment. Furthermore, unemployment is not only a byproduct of
efficiency wages, it is a necessary ingredient in the efficiency wage model.
If there is full employment, as mentioned above, the threat of termination
is not effective. Unemployment is a worker discipline device.
p(w)
w
b
Clearly, C has to be larger than g; the cost to the firm due to the worker
shirking has to be larger than the gain to the worker for both condition
(3) and condition (4) to be satisfied. If these conditions are satisfied, the
move to an efficiency wage situation constitutes a Pareto improvement:
both the firm and the workers are better off. Efficiency is increased hence
the terminology ‘efficiency wages’! The efficiency wage is higher than the
original market wage w0 and, if the MRP is unaffected, the demand for
labour decreases and unemployment is generated. In this simple model, a
rise in unemployment benefit is likely to increase the efficiency wage and
hence increase unemployment.
Real-life firms and some enlightened management consultants know that
being nice to workers may pay off. In the efficiency wage model, it is
shown that it may be possible to increase the quality of work, reduce staff
turnover or eliminate unproductive behaviour by paying higher wages.
Nordstrom, an American department store chain, pays its sales staff twice
the industry average in the hope of attracting and retaining good workers.
Clearly similar arguments could be made for non-pecuniary rewards for
workers. It may be more profitable (less costly) for a firm to increase
job satisfaction than to increase wages. Employees may appreciate non-
monetary aspects of their jobs such as company cars, training programmes,
pride in their work, more responsibility and autonomy. Chaparral Steel, a
Texan steel producer, allows workers to travel around the world to select
their own equipment. Every employee of Motorola spends at least a week
every year in training. Levi Strauss ‘empowers’ workers by allowing them
to redesign parts of the production process. In half of America’s large
companies, workers are organised in self-managing teams.3 When firms 3
See ‘Nicely does it’
create a worker-friendly environment, the effect on monitoring costs is
equivalent to that of an increase in wage. If firms are nice to workers,
their tradeoff between wage and detection probability (see Figure 8.1)
can be shifted to the left so that, for the same amount of monitoring, firms
can pay a lower efficiency wage or, for the same wage, they can do less
monitoring. Within organisations, the latter possibility creates conflict as
the reliance on middle managers to do the monitoring is diminished.
119
28 Managerial economics
Are the workers better off when the government introduces minimum
efficiency wages? The answer is ambiguous. The workers who keep their
jobs are now receiving w' = g(1 – p)/p + b compared to w0 + g. Given b < w0,
workers are likely to be worse off unless p is low which it would be if firms
face high costs of monitoring. Of course the workers who lose their jobs are
worse off. The move to efficiency wages cannot be a Pareto improvement.
w' A
join &
don’t cheat
join &
cheat
don’t join
B
w + cH – c L
c
(cH – cL)/p + cL
don’t join
H
0 p′ 1 p
MR MRP
MRPH
w′ + M(p)
MRPL
xq
x H
qLL0 qHL0 L0 L
Q
(a) Marginal revenue (b) Marginal revenue product
To keep things simple, let us assume the MR function is linear and set
qL = 1 and qH > 1. Figure 8.3a represents the MR function. Figure 8.3b gives
the corresponding MRP functions before (MRPL) and after (MRPH) the
introduction of the efficiency wage. These curves are derived from the MR
curve as follows. Consider MRPL first. From (10) we know that:
MRPL (L) = MR(L) for qL = 1.
Hence for L = 0, MR(L) is given by the intercept in Figure 8.3a. For
L = L0, we can read MR(L0 ) on Figure 8.3a. These two points are sufficient
to draw MRPL in Figure 8.3b. Now consider MRPH. For L = 0, MRPH = MR(0)
qH which gives us the intercept for MRPH in Figure 8.3b. For L = L0, we see
that MR(qHL0) = x in Figure 8.3a which gives MRPH(L0) = qH x in Figure
8.3b.
Conditions (9) and (10) tell us that the firm demand for labour is
determined by the intersection of MRPL and w and the intersection of
MRPH and w' + M(p) respectively. If the initial wage w is relatively high as
on the figure and if the efficiency wage and monitoring costs are not too
high, the firm’s demand for labour could increase when it starts paying
efficiency wages!
123
28 Managerial economics
a recent study, Ransom (1993) found that for his samples of university
professors, after controlling for experience and some other variables, there
was actually a negative correlation between seniority and salary.
In some circumstances, human capital theory offers an explanation for
increases in wages. Firms investing heavily in training which increases
the workers’ value not only in their current job but elsewhere as well,
tend to pay new trainee employees a low wage. This wage is increased
after the training is completed. Examples of this are found in accountancy,
law and architecture. The reason for the jump in wage is that, while the
employee is developing marketable skills at the firm’s expense, he pays for
this training by working at a low wage. As soon as the training phase ends
and outside offers would be forthcoming, the firm is in a position to offer
a higher wage and retain the trained workers. What is harder to explain
is that older workers are paid more than younger and sometimes more
productive workers with the same training and qualifications.
Lazear (1981) uses an efficiency argument to explain such rising wage
profiles. Workers are paid less than their MRP early in their careers and
more than their MRP later in their careers so that over their lifetime
they are getting the value of their marginal product. Given this wage
profile it is obviously painful for a worker to lose his job after he has
worked with a firm for a few years: he loses his ‘investment’ in the form of
larger wages later on. Therefore workers with rising wages are more likely
to work hard and avoid being fired whereas workers whose wages do not
increase do not have the same incentive to put in high effort. By offering
increasing wages the employer can succeed in making both the worker
and himself better off because higher efforts by the worker lead to higher
productivity. The increased cost of shirking leads to a Pareto improvement.
Because workers are paid very high wages when they are older, mandatory
retirement is necessary otherwise workers would prefer to continue
working after they had been paid their lifetime value to the firm. Also,
there has to be some mechanism to prevent the firm from firing workers
(without cause) before it has paid them the higher wages to which they
are entitled. If firms care about their reputations, they will not renege on
the (implicit) agreement to pay the workers their lifetime productivity
value by firing them early because then they will not be able to attract
workers in the future.
The efficiency explanation above rests on the idea of using rising wages as
an incentive mechanism. In jobs where the employer can easily monitor
whether the worker is performing adequately or where alternative
incentive mechanisms such as piece-rates or commissions can be used,
we would not expect to see wages rise much with seniority. Conversely, in
large organisations where monitoring costs are high, rising wage profiles
are likely.
Efficiency wage theory can also explain why wages rise over a career
path up to a point where they exceed marginal revenue product. Initially
employees are usually offered low responsibility jobs which are easier
to monitor. In terms of the minimum cost implementation problem, this
implies that the firm can set a high detection probability p at low cost M(p)
which in turn leads to a relatively low wage. Later on, when the employee
has more responsibility, the monitoring cost increases so less monitoring
will be done and a higher wage results.
Firms also offer seniority-based pay for jobs where it is important to retain
people. Paying a low salary initially and compensating later screens for
loyalty in the sense that only workers who intend to stay with the firm
124
Chapter 8: Topics in labour economics
over a longer period find this an attractive proposition. Firms also use
‘golden handcuffs’ such as deferred compensation and unvested pensions
as screening devices. For example, Bell South, an American regional
telephone company, encourages employees to put up to 25 per cent of
their pay in a special account. The company augments this contribution
and, when they retire, the employees get more than twice the normal
market interest whereas, if they quit, they just get the market interest.
When an employer hires a new worker neither she nor the worker knows
for sure how productive the worker will be. Workers are typically risk
averse and would prefer a contract which offers them a secure job at a
wage which corresponds to average productivity to a contract offering
them a wage corresponding to their actual productivity. However, if a firm
pays a wage equal to the value of the average productivity it will make
a loss: the low productivity workers will stay and the highly productive
workers will leave for better outside offers. To guard against this, the firm
has to start everyone at a low wage and increase wages of workers who
turn out to be productive. The difference between the low initial wage and
MRP for the productive workers is like an insurance premium they pay at
the time when they do not know their productivity.
Finally, in cross section data, wages can show positive correlation with
seniority even if individual workers do not experience rising wage
profiles. There are several explanations for this. Firstly, workers who are
productive in their jobs are highly paid and are likely to stay whereas
workers who are not productive and get low pay tend to leave. As a
consequence, we find that the highly paid workers have higher seniority.
Efficiency wage theory offers an alternative explanation: firms which pay
efficiency wages have low turnover which means that workers in these
higher paying firms have higher seniority again generating the positive
correlation.
Wage compression
Another feature of internal labour markets is that within firms there seems
to be less variability in wages than in productivity. An explanation for this
fact could be that it is just too difficult to measure individual productivity
especially where team work is involved. A more interesting explanation is
given by Frank (1984) in terms of status. The starting point is the idea that
people care about their position in the income hierarchy of the firm and
they care more the more they interact with their co-workers. Employees
are willing to sacrifice some income if they can be the highest paid in the
126
Chapter 8: Topics in labour economics
firm and conversely they need to be compensated for being the being the
one with the lowest pay. This is illustrated in Figure 8.4. The three lines
crossing the 45 degree line represent wage schedules for three firms. An
individual with marginal productivity x could choose to work in Firm
A where he is near the top of the salary scale and enjoy relatively high
status. He would then pay a price ab in terms of the difference between his
wage and his marginal productivity. If he does not care at all about relative
status the individual with marginal productivity x could earn a wage
premium bc by working for Firm C near the bottom of its salary scale.
B
b
A
a
marginal product
127
28 Managerial economics
been public outrage in Britain over the ‘excessive’ salary increases top
executives of newly privatised industries award themselves. In January
1995, Cedric Brown, the CEO of British Gas was asked to justify to
Parliament his 75 per cent salary increase to £475,000.8 One of the
reasons for these huge remuneration packages seems to be the use of
incentive schemes for high-level executives. The fixed salary part of the
compensation package may be relatively modest but CEOs typically also
receive stock options which are rights to buy company shares at or above
the current share price. Clearly, there is no downside risk involved here.
If the share price is low during the specified time in which the options
can be exercised, the manager does not buy any shares. Nevertheless
share options can be a useful device to lengthen the CEO’s horizon and
give him an incentive to maximise the long-term value of the firm. The
remuneration package may also contain straightforward stock awards
which consist of shares given or sold at a discount. Sometimes there
are restrictions on the sale of these shares, for example, they may only
be sold after retirement. This is another example of the use of golden
handcuffs to align an employee’s interest to the long-term future of the
company.
Another reason for what seems like excessive compensation at the top of
large firms is that, once an employee reaches this highest level, promotion
can no longer be used as a reward. There is also a ‘political’ explanation
for high executive compensation. Nonexecutive directors who determine
executive pay clearly have an interest in exaggerating these pay awards
as they are often executives elsewhere and can only benefit if the ‘going
rate’ for top jobs rises. A similar story can be told to explain the inactivity
of institutional shareholders. Top executives of banks and pension funds
hardly have an interest in calling attention to high pay.9 The absolute 9
See ‘A racket in need of
size of CEO pay should not be a major concern of shareholders as long as reform’
there is a link with performance. The structure of pay and the provision of
incentives is what matters. It seems that too often executives are paid large
amounts of money when their companies are not doing so well. It is all the
more surprising therefore that large British institutional investors call for
ceilings on the amount of share options that can be issued to executives.
The Association of British Insurers (ABI) and the National Association of
Pension Funds (NAPF) recommend a limit of four times the fixed salary
component for stock options.10 10
Main et al. (1994)
128
Chapter 8: Topics in labour economics
Sample exercises
Review the minimum cost implementation problem for efficiency wages
and use this framework to argue whether efficiency wages will be higher
or lower than in the standard model if:
a. workers dislike monitoring
b. workers incur a large psychological cost when they are unemployed
c. workers are risk averse
d. monitoring is very costly
e. workers have high switching costs (moving expenses, loss of firm
specific human capital) when they lose their jobs.
129
28 Managerial economics
Notes
130
Chapter 9: Market structure
Aims
The aim of this chapter is to consider:
• the importance of entry barriers as determinants of market structure
and firm behaviour
• the concepts concentration curve, CRm, HHI and Lerner index
• why in a perfectly competitive industry the supply curve is not
necessarily the horizontal sum of individual firm supply curves
• why we may not observe MR = MC in a monopoly
• the Dorfman-Steiner model.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• describe the important factors in the determination of market
structure giving (preferably your own) examples
• calculate the market structure measures
• derive perfectly competitive firm and industry supply
• find profit maximising price and output for a monopoly
• explain short run and long run equilibrium in a monopolistically
competitive industry.
Essential reading
Varian, H.R. Intermediate Microeconomics Chapters 22, 23 and 24.
References cited
‘Ahead for now’, The computer industry survey, The Economist, 17 September
1994, pp.12–16.
Dorfman, R. and P.O. Steiner, ‘Optimal advertising and optimal quality’,
American Economic Review 44 1954, pp.826–36.
Gourvish, T.R. and R.G. Wilson, The British Brewing Industry, 1830-1980
(Cambridge: Cambridge University Press, 1994) [ISBN 0521452325]
‘Insurers get that sinking feeling’, The Economist, 20 August 1994, pp.65–66.
Martin, S. Advanced industrial economics, (Oxford: Blackwell, 1993)
[ISBN 063117852X]
‘National lottery’, The Economist, 28 May 1994, pp.28–33.
‘Squeezing into Hong Kong’, The Economist, 4 December 1993, p.88.
Sutton, J. Sunk costs and market structure. (Cambridge, Mass: The MIT Press,
1992) [ISBN 0262193051].
‘The Post Office’, The Economist, 8 January 1994, p.29.
‘Unsure about insurance’, The Economist, 31 October 1992, p.101.
131
28 Managerial economics
Introduction
Industries differ in the way they are structured. An important defining
characteristic of market structure is the number of firms serving the
market. However, it is certainly not the case that we can predict how firms
will behave just on the basis of how many firms there are in the industry.
The presence of barriers to entry also plays a crucial role in maintaining
market structure. It is generally believed that, when entry is easy, firms are
disciplined in their pricing policies even when the industry is concentrated.
Empirically, high entry barriers coincide with high profit margins.
Economists classify industries according to the degree of market power
firms have (i.e. do they behave as price takers or price makers?), whether
firms behave strategically and whether the product is homogenous
or differentiated. For example, in a competitive industry all firms are
assumed to be price takers; in an oligopoly there is strategic interaction
between firms.
Economies of scale
The world market for disposable plastic syringes is dominated by three
firms: Becton Dickinson, Sherwood/Brunswick and the Japanese firm
Turumo, with worldwide market shares in 1992 of 31 per cent, 16 per
cent and 18 per cent respectively. There are significant economies of
scale in this industry: the production technology is such that, as output
increases, average costs fall dramatically. The source of economies of
scale may be large setup or fixed costs. In extreme cases the presence of
significant economies of scale leads to a ‘natural monopoly’ which refers
to the situation where average costs decrease beyond the output level
corresponding to market demand. In such cases it is obviously cheaper
for one firm to satisfy market demand. The natural monopoly argument
has been used as an argument against privatisation of the Royal Mail in
Britain. Local newspapers are also characterised by significant economies
of scale so that often there will be just one local newspaper in a town.
The size of the market relative to the degree of economies of scale
seems a relevant predictor of market structure. A useful concept in this
context is the minimal efficient scale (MES) of an establishment in
the industry. It is usually defined as the output level at which long-run
average cost (LRAC) is minimised. An alternative definition of MES is
‘the output level beyond which further increases in output would lead to
a reduction in LRAC of no more than 10 per cent’. In practical terms this
means it is the output level where the LRAC curve begins to flatten out.
If we know the MES for a given production technology and we know the
total consumption in the market, we can calculate how many firms can be
accommodated in the industry. For example, if the MES for commercial
aircraft manufacturing corresponds to 10 per cent of the US market, then
there is room for 10 fans to serve this market. If the MES is large relative
to the size of the market, the industry is likely to be concentrated.
The importance of economies of scale as a determinant of market structure
is made evident by the changes in structure after technological innovation.
132
Chapter 9: Market structure
Strategy of incumbents
Market structure and the presence of entry barriers undoubtedly affect the
behaviour of established or incumbent firms in an industry. Conversely,
incumbents may engage in activities designed to maintain the market
structure or alter it in their favour. An example of the former is limit
pricing where a monopolist, rather than maximising short-run profits,
prices such that it is not possible for an entrant to make a profit at the
current price. An example of the latter is predatory pricing, where a firm
sets a low price (often below cost) in order to drive a rival out of business.
Firms have instruments other than prices to their disposal when trying to
drive competitors out of business. For example, after the deregulation of
buses in some British cities, some bus companies tried to organise their
timetables so that they would pick up passengers a few minutes before their
rivals. Note that limit pricing and predatory pricing are only possible if the
firm has a cost advantage. In some cases a monopolist does not have to set
a low price to deter entry. The threat to retaliate and flood the market when
a competitor enters may be sufficient. Such a threat can be made credible if
the monopolist has excess capacity so that increasing output (after entry) is
not very costly.
Incumbent firms manipulate dealer relationships to shut out
competition. An extreme case is that of exclusive dealership where a
manufacturer or wholesaler agrees to supply retailers only if they do not
sell any substitute products. Supermarkets typically carry a fairly limited
number of brands of consumer goods. Manufacturers negotiate deals which
involve the supermarkets allocating them a certain amount of shelf space.
Some PC makers sell their machines with a copy of MS-DOS or Windows
already installed. Some airlines have developed computer reservation
systems for use by travel agents. Of course flights by the airline which
developed the system will be more prominently displayed and easier to
book than others and, once travel agents have invested in learning how to
use such a system, they are ‘locked in’.
Incumbent firms often try to erect entry barriers by heavily advertising and
promoting their products. This is effective in industries where customer
loyalty is important or where buyers prefer established brands so that
it becomes very difficult for an entrant to convince consumers to try his
product. Large advertising budgets also increase an entrant’s capital
requirements.2 Companies can increase customer loyalty by offering loyalty 2
See next section
rebates. Airlines’ frequent flyer programmes are a good example of this.
Capital requirements
Large capital requirements can be a source of a cost disadvantage for a
potential entrant. Entrants are usually perceived as riskier prospects by
banks and other money-lenders. This is not without reason as in many
industries entrants are much more likely than established firms to go
out of business. They therefore pay a higher risk premium. Large sunk
costs (investments such as R&D which cannot be recuperated upon exit
from the market) make the entry decision particularly risky. They are
133
28 Managerial economics
Control of resource
An obvious instance of market structure determined by entry barriers is
when incumbents control a special non-replicable resource such as mineral
deposits. Some well-known examples are De Beers in diamonds, OPEC
and French Champagne. Take-off and landing slots at airports are also
very important resources without which entry into the airline industry
is impossible. In the same industry, the practice at most airports of long-
term leasing of gates represents an entry barrier. There are however limits
to the market power which derives from owning a special resource. The
Organisation of Petroleum Exporting Countries (OPEC) found that, when
they raised the price of oil too much, other countries started searching
for oil. The oil crisis of the early 1970s was largely responsible for the
exploration of oil fields in the North Sea.
134
Chapter 9: Market structure
135
28 Managerial economics
Perfect competition
In a perfectly competitive market there are many buyers and sellers who
act independently (non-strategically) and have perfect knowledge of the
market and in particular of the price charged by all sellers. The product is
homogenous so that firms do not engage in non-price competition. There
is no point in advertising for example if consumers are perfectly informed
and all products are identical. This also means that there is no price
dispersion (i.e. there is a unique market price at which everyone buys and
sells). No firm has control over this market price. It is often also assumed
that all firms have access to the same technology and this assumption,
combined with the fact that firms face the same input prices, implies
that the firms will have identical LRAC curves. Inputs are assumed to be
perfectly mobile and responsive to monetary incentives and there are no
barriers to entry in the long-run. Entry then occurs until economic profit
(of the marginal firm if they are not all identical) is zero. Zero economic
profit of course allows for a normal rate of return on equity.
As a result of these assumptions a firm behaves as if the demand for its
product is horizontal. Clearly the firm is in a sense making a mistake
here: when it decides to increase its supply the price of output is affected
because the industry demand curve is downward sloping. The same
comment applies to consumers. If any consumer increases his consumption
of the good he has a (very small) effect on price since industry supply is
upward sloping. The idea that competitive firms and consumers ignore
their effect on output price should be seen as a convenient simplification.
The assumptions of the perfect competition model may seem unrealistic
but there are several real world industries which conform relatively closely
to the competitive framework. Markets for some agricultural products
operate in a competitive fashion as do financial markets for savings and
136
Chapter 9: Market structure
stocks. The packaged tour business is also very competitive with operators
regularly going out of business. There are virtually no barriers to entry in
this business. Chartering aircraft or setting up a travel agency does not
require major capital investment.
137
28 Managerial economics
Example 9.1
As an exercise you should check how many firms could enter this
industry and still make a profit.6 6
Hint: Let the total
number of firms be
n and calculate the
The distinction between short and long-term is also important for industry
equilibrium market
supply. Long-run supply is more elastic than short-run supply because all output and price as a
inputs are variable and there is entry and exit. Sometimes the notion of function of n. Then find
intermediate run is used meaning the period of time in which existing the maximum n for
firms can adjust most or all of their inputs whereas in the long-run market which profit is positive.
entry and exit can take place. If all firms are identical the entry and exit
process ensures that each firm produces the output level at which its LRAC
are minimised (i.e. at its efficient scale) and price equals LRAC. Why is
this so? Clearly price cannot go below minimum LRAC or nothing would
be produced. If it is above LRAC then firms can, by choosing their output
level carefully, make positive profits which would induce entry. Note that,
if firms are not identical, long-run supply is no longer constant but upward
sloping. At the equilibrium in such an industry only the ‘marginal’ firms
make zero profits; the other firms make positive profits.
p
∑ MC ∑ MC
1 0
p
1
p
0
elastic than the horizontal sum of firm supply curves. This input price
effect of a change in output price is likely to be significant when the
industry is a large user of any of its inputs.
It is very difficult to assess the size or the significance of the input price
effect. In some cases it may be counteracted by new entry into the
industry as higher product prices lead to (temporary) larger profits. When
industry output increases there may be some positive externality effects
as well such as technological improvements through learning by doing for
example or improved support services. These factors would tend to make
industry supply more elastic than the horizontal sum of MC curves.
Monopoly
A monopoly is an industry consisting of one firm. Apart from state owned
or heavily regulated industries such as electricity there are not many
markets which conform to this extreme market structure. In practice
however, we assume that a firm will behave as a monopolist when it is
has a dominant market share of, say, 60–70 per cent and there is no major
rival. Microsoft and Intel for example have market shares of about 80 per
cent in operating software and microprocessor chips for PCs respectively.7 7
See ‘Ahead for now’
Monopolies are relatively common in emerging economies and former
communist countries because of protectionist measures and the lack of
antitrust legislation. For example, in Mexico, Vitra has a market share of
90 per cent in flat glass. Philip Morris took over Tabak (a Czech tobacco
company) which gives it control over 80 per cent of the market in the
Czech Republic. A few other examples of monopolies are:
Utilities
Whereas in Britain, New Zealand and the US you can choose your
telephone company, in many countries the telecom industry is a monopoly.
Hongkong Telecom will face competition from three new competitors from
1995 but is allowed to keep its monopoly on international calls (which 9
See ‘Squeezing into
generate most of its profits) until 2006.9 In Britain the Post Office keeps its Hong Kong’
monopoly on letter deliveries below £1.9.10 10
See ‘The Post Office’
The standard monopoly model is very simple. Industry demand equals
firm demand and, although we could make a distinction between short-
and long-term with respect to the monopolist’s cost function, there is no
139
28 Managerial economics
example is that of an airline route between two small cities. Entry is easy
for anyone who has planes available (on other less lucrative routes) and
exit is easy because planes can be sold or leased or switched to other routes.
In these circumstances, a monopolist cannot exercise his market power
because, if he did, he would be replaced by a willing and able entrant. In
fact monopolists may even behave as if they were in a perfectly competitive
industry if the market is contestable.
Monopolistic competition
Monopolistic competition is the market structure which emerges if we relax
the assumption of homogenous goods in the perfect competition model.
There are many sellers producing a differentiated product. For example, in
the beer industry, each brewery markets a unique product (although it has
many close substitutes). The main consequence of relaxing the assumption
of homogeneity of products is that some forms of non-price competition
such as advertising and providing different types of service can be profitable.
Monopolistic competition is associated with industries of firms with large
advertising budgets. Many ‘fast moving consumer goods’ (FMCG) sectors
fall into this category. Firms make enormous efforts to convince their
consumers that the washing powder they market, for example, is not
identical to the one marketed by their rivals. You could say that advertising
is necessary in monopolistic competition to ensure that buyers and sellers
have perfect information about the market which is what we continue to
assume. There are, apart from advertising, many methods firms use to create
differentiation. The following list contains a few examples:
• packaging: two chemically identical dishwasher detergents are
perceived as different by consumers if one is packaged in a plastic bottle
and the other in a cardboard box
• product design: shampoo and conditioner are sold separately and
combined
• type of service: in the PC industry it is becoming increasingly difficult
to produce a differentiated product. The same components are used in
the assembly of most PCs. Almost all PCs use Intel’s microprocessors. Dell
however has been very successful in this industry because it invented a
new way to sell PCs. Dell computers are sold by mail order and there is a
telephone hot-line which offers technical advice and after-sale service
• location: supermarkets and restaurants, even when they offer the same
range of products, are differentiated by their location. It is unrealistic to
assume that, if a supermarket charges a penny more for butter than the
supermarket at the other end of the high street, it will lose all its business,
or conversely that, when it undercuts its rival by a small amount, it
captures the whole market.
141
28 Managerial economics
Each firm has some monopoly power. Indeed one of the reasons to engage
in non-price competition such as improving service is that a firm which
produces a differentiated product does not have to cut its price whenever a
rival cuts his price.
The short-run monopolistic competition model is in fact exactly the same
as the monopoly model with firms setting output levels such that MR =
MC. As in perfect competition there are no or low barriers to entry so that
in the long-run entry and exit ensure that profits are zero. The long run
equilibrium conditions for a monopolistically competitive industry are as
follows:
• at the optimal output level q* for each firm, price should equal average
cost (AC) so that profit is zero
• at its optimal output level each firm maximises profit and hence MR =
MC at q*.
From the long-run equilibrium conditions we know that, for the optimal
output level q*, MR intersects MC. The price corresponding to q* can be
read off the demand curve and has to equal AC as is illustrated in Figure
9.2. This leaves the possibility of demand intersecting AC rather than
being tangential to it at q*. However, this would mean that there are
output levels for which the firm could make a positive profit contradicting
the fact that q* is profit maximising. The figure shows that, at any output
level other than q*, the firm makes a loss. Because demand has to be
tangent to AC at the long-run equilibrium, firms operate at output levels
to the left of the efficient AC minimising level. If they could increase
their output, their AC would decrease. The cost of this ‘excess capacity’
is often referred to as the price of variety. Consumers are willing to pay a
premium, so the argument goes, to have a choice between differentiated
products. When, during your weekly food shopping, you have to find your
way through the enormous variety of breakfast cereals for sale on the
supermarket shelves you may be forgiven for being sceptical about this. In
the US in 1990 alone, 48 new cold or cough remedies were introduced. It
is hard to believe that consumers want this much variety.
p
MC
AC
p*
D
MR
q* q
142
Chapter 9: Market structure
143
28 Managerial economics
Sample exercises
1. The market demand for a product is as indicated in the table below.
The industry is perfectly competitive and the second table gives each
firm’s long-run total cost. Each firm can produce only integer numbers
of units of output. How many firms will be in the industry in the long-run?
price (£) quantity demanded
30 200
20 300
10 400
5 600
3 800
output total cost (£)
1 10
2 12
3 15
4 30
Price per candy bar (pence) Thousands of candy bars sold per game
20 10
25 9
30 8
35 7
40 6
45 5
50 4
5. Demand for a monopolist’s goods is given by q = S/p0 if p ≤ p0 and q = 0
if p > p0. S is a constant. What is the monopoly price?
145
28 Managerial economics
Notes
146
Chapter 10: Monopolistic pricing practices
Aims
The aim of this chapter is to consider:
• the differences between first, second and third degree price
discrimination
• why under first degree price discrimination all consumer surplus can be
extracted
• why under two part pricing, at least for identical consumers, unit price
should be set equal to marginal cost
• the relationship between price in a market and its price elasticity when
third degree price discrimination is practised
• intertemporal price discrimination and the Coase conjecture
• the peak-load pricing model
• the difference between pure and mixed commodity bundling
• the concepts of loss leaders, cannibalisation, joint products and
economies of scope
• the joint products model and why a monopolist might waste rather
than sell some of his output
• the three transfer pricing scenarios
• the relevance of taxation systems on the practice of transfer pricing.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• give (your own) examples of price discrimination and commodity
bundling
• derive the optimal take-it-or-leave-it-offer in first degree price
discrimination
• derive the optimal two-part pricing scheme
• derive optimal prices and quantities in third degree price discrimination
(analytically and graphically) and determine the optimal price and
quantity when the monopolist faces two demand curves and is not
allowed to price discriminate
• solve a simple two period skimming model with myopic consumers
• solve a simple peak-load pricing problem
• show graphically how consumers with different reservation prices
distribute themselves over the various options in commodity
bundling
• for a multiproduct firm derive prices and quantities for two
products when
a. each product is handled by a separate division and divisions act
noncooperatively and
147
28 Managerial economics
Essential reading
Tirole, J. The Theory of Industrial Organisation. Sections 1.1.2, 1.5.2, Chapter 3.
Varian, H.R. Intermediate Microeconomics. Chapter 25.
Further reading
Adams, W. and J. Yellen ‘Commodity bundling and the burden of monopoly’,
Quarterly Journal of Economics 90 1976, pp.475–98.
‘Ad nauseam’, The Economist, 5 February 1994, p.71.
‘A hell of an operating system’, The Economist, 18 January 1995, pp.17–18.
‘Better than price fixing?’, The Economist, 3 October 1992, p.100.
‘By the seat of their pants’, Airlines Survey, The Economist, 12 June 1993,
pp.26–28.
‘Europe’s car market. Carved up’, The Economist, 31 October 1992, p.94.
‘Death of the brand manager’, The Economist, 9 April 1994, pp.79–80.
Fisher, K. ‘Return to sender’, The Economist, 4 February 1995, p.8.
‘Hard sell’, The Economist, 4 March 1995, pp.89–92.
Hirshleifer, J. ‘On the economics of transfer pricing’, Journal of Business 29 1956,
pp.172–84.
‘IOU all over again?’ The Economist, 2 July 1994, pp.40–41.
‘Labouring in obscurity’, The Economist, 17 September 1994, p.94.
Lan, L. and A Kanafani ‘Economics of park-and-shop discounts: a case of bundled
pricing strategy’, Journal of Transport Economics and Policy 3/27, pp.291–303.
‘Learning to fly all over again’, Airlines survey, The Economist, 12 June 1993,
p.13.
‘Life after Lenin’, The Economist, 29 January 1994, p.74.
‘Managing the future’, The Economist, 19 December 1992, pp.70–75.
‘Movie mystery’, The Economist, 19 November 1994, p.36.
‘Not dead yet’, The Economist, 28 January 1995, pp.76–80.
Oi, W. ‘A Disneyland dilemma: two-part tariffs for a Mickey Mouse monopoly’,
Quarterly Journal of Economics 85 1971, pp.77–90.
Schmalensee, R. ‘Gaussian demand and commodity bundling’, Journal of Business
57(1,2) 1984, S211–S230.
‘Tax deficient’, The Economist, 22 May 1993, p.83.
Stigler, G.J. ‘United States v. Loew’s Inc: A note on block booking’, Supreme
Court Review (1963) pp.152–57. Reprinted Stigler, G.J. The organisation
industry. (Homewood, I11: Irwin, 1968).
‘Taxing questions’, The Economist, 22 May 1993 p.83.
‘The high-tech war’, The Economist, 26 December – 8 January 1993, pp.83–84.
‘The richest islands in the world, maybe’, The Economist, 6 November 1993, p.80.
‘Unhappy returns’, The Economist, 2 April 1994, p.88.
‘Unhappy returns for Barclays’, The Economist, 25 June 1994, p.97.
van Ackere, A. and Reyniers, D.J. ‘A rationale for trade-ins’, Journal of Economics
and Business 45 1993, pp.1–16.
van Ackere, A. and Reyniers, D.J. ‘Trade-ins and introductory offers in a
monopoly’, RAND Journal of Economics (1995) 26 pp.58–74.
Webster, T. ‘Rooms with a view to saving’, Evening Standard, 30 November 1993,
p.38.
‘What went wrong at IBM’, The Economist, 16 January 1993, pp.23–25.
148
Chapter 10: Monopolistic pricing practices
Introduction
In our discussion of monopoly pricing (Chapter 9) we found that the
monopolist sets price such that marginal revenue equals marginal cost.
We implicitly assumed that the monopolist had no option but to set a
single price, the same for all buyers, independent of the quantity they buy,
whether they buy any other products from the monopolist, etc. In reality,
sellers often have the opportunity to market their products at different
prices (price discrimination) or sell them as part of a package (commodity
bundling). So far, we have also assumed that the firm markets one product
or that we could analyse production and marketing of individual goods in
isolation. However, most firms produce a range of products and they find
it profitable to keep the pricing of the entire product line in mind when
making pricing decisions for each product.
Many firms are structured as multidivisional organisations with one
division buying goods from another division within the firm. The transfer
price at which such goods are sold between divisions is of crucial
importance for the profitability of the firm. If selling divisions are rewarded
on the basis of their divisional profit, they will tend to set the transfer price
too high which is detrimental to overall profitability. Clearly, transfer prices
have to be determined centrally in such situations. In this chapter we turn
our attention to these types of pricing problems.
Price discrimination
Of all pricing practices, monopolistic or otherwise, price discrimination
in its many forms is certainly the most visible. Price discrimination occurs
whenever the same good or service (i.e. a good or a service produced at
identical cost) is sold at different prices. More generally, we can say that,
when price differences between consumers are larger than is warranted
by cost differences, there is price discrimination. For example, British Gas
offers a reduction of your gas bill if you pay by direct debit. To the extent
that collecting revenue by direct debit is less costly than by alternative
methods, the discount is not an indication of price discrimination. It is
more difficult to argue that price differences between first and second class
on trains, or business and economy on planes, are justifiable on the basis of
cost differences. There is something else going on here.
For a firm in a perfectly competitive industry it is not possible to indulge
in price discrimination. If such a firm sets a price above the market price,
it loses all of its customers. Some degree of market power is necessary
to enable a firm to price discriminate. It is easiest to analyse price
discrimination by a monopolist although in reality it occurs mostly in
oligopolies. (Models of price discrimination by oligopolists are the subject
of recent and current research in industrial organisation.) It is easy to
understand the attraction of charging different prices for the same good.
Clearly, profits cannot decrease as a consequence of price discrimination
since one option is to set the same price in all markets. In general, profits
increase if price discrimination is possible.
Why is price discrimination more common in the sale of services than
manufactured goods? For services, resale is generally not possible. I
cannot get a haircut or a massage and sell it to you, for example. Utilities
such as telephone and electricity companies often successfully use price
discrimination. Customers cannot resell their services.
149
28 Managerial economics
m- P*
x
x*
150
Chapter 10: Monopolistic pricing practices
Example 10.1
p
a
D
b
c
d
x
0 x′ x*
152
Chapter 10: Monopolistic pricing practices
m-F
a
x
x* (m-F)/p
153
28 Managerial economics
154
Chapter 10: Monopolistic pricing practices
p2
p1
p1' MC'
D1 Σ MR
p1*
MC
MR1 p2* D2
D
MR2
q1*+ q2* = q0
q1* q 0 q2* q2 q
q1' q0 q1 2
1 q 1' + q2'
155
28 Managerial economics
Can you draw a MC curve on Figure 10.4 for which the monopolist
only sells in both markets when he is allowed to charge different
prices?
Example 10.2
156
Chapter 10: Monopolistic pricing practices
157
28 Managerial economics
the least elastic demand. Every day, you can see applications of this
principle. People are charged more when they want to have things done in
a hurry (their demands are likely to be inelastic): for example, one-hour
versus 24-hour photo processing. This type of price discrimination based
on urgency of service also applies to same day versus same week dry-
cleaning etc.
Japanese car manufacturers charge more for their cars in Japan than
abroad because the overseas car market is more competitive (and hence
its price elasticity higher) than the protected Japanese market. Dumping
is illegal for goods imported in the US. To offset the price differential,
extra duties are imposed on importers found dumping. The situation in
the European Union is similar. However, in an interesting case of double
standards, the EU persistently dumps surplus agricultural commodities,
arising from farm support programs, in poor countries.
Because the income effect of a price change is likely to be larger for poor
customers, their demands are generally more elastic. In many instances we
do observe that poorer customers are charged lower prices. Now you know
that this may have nothing to do with charity as it can be explained by
pure profit maximising behaviour on the part of the seller. It is rational for
doctors and dentists to charge poor people a lower fee. Often geographical
proxies are used in the sense that everyone in a generally poorer area
benefits from lower prices and conversely everyone in a relatively richer
area is asked to pay more. For example, Warner Brothers’ cinema in
Leicester Square in London charges £7 for a ticket about double of what
is charged (£3.80) in its cinema in Bury near Manchester.4 Apparently, 4
See ‘Movie mystery’
Russian newspapers have caught on to this trick. In January 1994,
Izvestia charged 20m roubles ($13,300) for a full page advertisement
if the customer was a Russian company and $30,000 if it was a foreign
company.5 In China, outdoor advertising rates for foreign products are up
5
See ‘Life after Lenin’
to five times those for domestic products.6 Keith Fisher (1995), the director 6
See ‘Not dead yet’
of Royal Mail International, points out in a letter to The Economist that the
Royal Mail charges a developing country one penny to deliver an incoming
letter whereas an advanced country pays 16 pence. Measures have to be
taken to avoid abuse by ‘remailers’ who essentially arbitrage by shipping
bulk mailings to developing countries.
Marketing managers use various devices to discriminate between segments
of the market. Although grocery coupons have other uses such as
gathering information on demand by experimenting with the price, it
seems that sellers may be trying to discriminate between people who
collect and redeem coupons and those who do not bother. The assumption
is that the redeemers have more elastic demand. Coupons are used for
many FMCG such as breakfast cereals. The importance of this type of
promotion is illustrated by the fact that grocery-coupon redemption
amounted to $6 billion in the US in 1992.7 Similar reasoning applies 7
See ‘Death of the
to price matching which refers to the practice of offering to match brand manager’
any lower price the consumer can find. Usually an advertisement of the
lower price has to be presented. Again there are other explanations for
this marketing practice but, since very few customers collect on price
guarantees, price matching is an easy way to discriminate between
customers with low search costs (for lower prices) and those with high
search costs. It can be argued that the latter have more inelastic demands.
158
Chapter 10: Monopolistic pricing practices
consumer buys at most once. A frequently observed pattern for these types
of goods is that they are initially sold at extremely high prices. After a
few years these innovative goods become more affordable. This pricing
pattern of decreasing prices over time is referred to as intertemporal price
discrimination. It is a form of third degree price discrimination in the sense
that consumers who buy later pay a lower price than those who buy early
on. At the same time you could argue that rational consumers can predict
that prices will decrease and can therefore ‘self-select’ when they will buy,
which makes intertemporal price discrimination more like second degree
price discrimination.
Price decreases over time are the norm in publishing. Publishers are in fact
monopolists since each book is sold by only one publisher. Although there
are some cost differences between hardbacks and paperbacks, it is difficult
to justify the price differences on a cost basis. Furthermore, paperbacks
are always sold when the demand for the hardback version has virtually
dried up. Between the hardback and the paperback stage, books are sold
through bookclubs as well. The Softback Preview bookclub is an English
company which specialises in high-quality paperback versions of original
hardback editions at around half the publisher’s hardback price. An
example of market skimming, where price cuts are not disguised by quality
differences is provided by Intel. Until Intel lost its monopoly, (when AMD
(Advanced Micro Devices) entered the market), its pricing strategy was
to keep new microprocessor prices very high and ease them down slowly
over time.
Sometimes intertemporal and third degree price discrimination are used
simultaneously. Tickets for the Kasparov-Short chess championship in
London, sponsored by the Times, were initially offered at up to £150.
When they did not prove popular and very few seats were sold, prices
were reduced down to £20 (intertemporal price discrimination)
and readers of the Sun (a London daily newspaper) were offered tickets
for £10 (third degree price discrimination). The frequently used
marketing practice of offering a ‘trade-in discount’ on goods such as
vacuum cleaners and other domestic appliances can be explained as an
attempt to price discriminate both intertemporally and between customers
who have bought before and those who have not.8 8
van Ackere and
Reyniers (1993, 1995)
If we assume that consumers are myopic, which means that they do not
think ahead and hence do not anticipate price changes, price skimming
is easy to model. The population of buyers and their willingness to pay
can be represented by a demand curve p(q), as in Figure 10.5. If the
monopolist can costlessly change the price he ends up with the entire
consumer surplus (above marginal cost c). He ‘skims’ the market by
selling at a high price to the most eager (or rich) consumers with very
high willingness to pay first and then gradually moves down the demand
curve selling to consumers with lower and lower reservation prices. In
reality of course durable goods prices do not change every hour. The
monopolist may have to keep the price constant for a given period of time,
for example, because the price has appeared in a printed advertisement or
a catalogue. In that case we would expect the price to jump from period
to period. As in Figure 10.5, during the first year, the price may be p1 and
q1 units are sold. At the beginning of the second year, since all consumers
with reservation price above p, have left the market, the monopolist faces
a new residual demand curve representing consumers still in the market
(line AB in Figure 10.5). He then picks a new (lower) price p2 and sells
q2 – q1 units at this price and so on. Clearly, a large fraction of the
consumer surplus can be captured in this way.
159
28 Managerial economics
A
p1 p(q)
p2
p3
c q
q1 q2 q3 B
Example 10.3
160
Chapter 10: Monopolistic pricing practices
161
28 Managerial economics
profitable) flights. Airlines now even have flexibility with respect to the
proportion of seats allocated to first, business and economy classes. The
new Boeing 777 airliner uses quick changing bulkheads so that these
proportions can be changed rapidly.10 10
See ‘Learning to fly
all over again’; ‘By the
Although yield management is mainly associated with the airline seat of their pants’; ‘The
industry, hotels face very similar problems and use similar revenue high-tech war’
maximising techniques. Discounts for guests booking in advance,
differential pricing between weekdays and weekend nights, loyalty
bonuses and quantity discounts (e.g. 50 per cent off for a two-night
stay) are just a few of the marketing ploys characterising this business.
One of the nicest examples of third degree price discrimination must
be the deal offered by The Royal Orchid Sheraton Hotel in Bangkok. To
celebrate its 10th birthday in 1993, it offered a percentage discount on
the room rate rising according to the guest’s age.11 11
Webster (1993)
Peak-load pricing
For many goods and services, demand fluctuates with the seasons or time
of day so that the seller faces several demand curves. The demand at these
different times is satisfied by a common facility but typically different
prices are charged for peak and off-peak use. The demand for electricity
peaks in the morning and the evening and there is seasonal variation
because of heating and air conditioning. Telephone companies have lower
rates for long-distance calls in the evenings and at weekends. Health clubs
sometimes offer off-peak membership at a lower fee. Hotels are cheaper
off-season when they have spare capacity.
Charging different prices at different times of day is not really price
discrimination if costs vary with output because of overtime or use of
less efficient production methods when demand is high. Often marginal
cost is constant until a critical capacity is reached. Electricity generating
companies use technologically efficient methods for lower levels of
output but when these are exhausted they have to use older technologies
which could be ten times as expensive per unit of output (Kilowatt hour).
The same is true for airlines which charge high fares in times of peak
demand when older and less fuel-efficient planes have to be brought into
operation. It is not always easy to establish whether a price difference
can be justified by a cost difference. In many instances, demand in the
peak period is less elastic and therefore, even if costs were the same, a
monopolist would charge a higher price.
In its most general form, peak-load pricing is not easy to model. We
should really start out with a model of consumers deciding whether they
want to consume in the peak or off-peak period depending on the relative
prices. However, we will skip this step here and take the demand functions
which would result from such a process as given. Let us look at a simple
model of peak-load pricing in which the peak (Dp( p)) and off-peak (D0( p))
demands have the same elasticity so that the standard third degree price
discrimination argument would give identical prices. An example of such
a situation is when, for every possible price, peak demand is a multiple of
off-peak demand: Dp( p)= mD0( p), where m is a constant. Assume further
that the cost of providing the service, ignoring capacity costs, is C(q), the
same for both periods. The same capacity K is used in peak and off-peak
periods. The opportunity or investment cost per unit of capacity is constant
at c per unit. The profit maximisation problem is:
max П = pp(qp ) qp + p0 (q0)q0 – C(qp ) – C(q0) – cK (6)
162
Chapter 10: Monopolistic pricing practices
subject to qp , q0 ≤ K, where qp and q0 are the units sold in peak and off-peak
periods respectively. There are two cases we need to consider. The first case
is when capacity is fully used in the peak period only (q0 < qp = K). Optimal
prices are derived by rewriting (6) as:
max П = pp(qp ) qp + p0 (q0 )q0 – C(qp ) – C(q0 ) – cqp
for which the first order conditions are:
MRp(qp) = MC(qp ) + c (7)
and
MR0(q0) = MC(q0).
The other case we need to consider is when capacity is fully used in both
periods (q0 = qp = q = K). This leads to rewriting (6) as:
max П = pp(q) q + p0(q) q – C(q) – C(q) – cq
which leads to the following result:
MRp (q) + MR0 (q) = 2MC(q) + c. (8)
In either case, the optimality conditions (7) and (8) determine quantity
sold in each period, which by substitution in the demand functions gives
the optimal prices. To determine whether it is best to set prices so that
capacity is only binding in the peak period or whether the seller should
attempt to equalise demand in both periods, profit has to be calculated for
both cases.
Commodity bundling
Sellers sometimes offer special deals when goods or services are bought
in packages. For example, it is often cheaper to buy a holiday package
which includes flight, accommodation and meals than to buy these items
separately. Some computer manufacturers (e.g. Gateway 2000 and Zeos)
sell bundled software as part of a package with a computer. PCs are often
sold in a package with peripherals. Many durable goods sellers offer
financing with their products. Luxury cars may be sold with leather seats,
sunroof, wooden dashboard, CD player and driver and passenger airbags
included at no extra cost. Again this marketing practice can be explained
in terms of price discrimination but, as it is rather special, it deserves a
special section.
Commodity bundling is sometimes used when the seller has to offer all
buyers the same deal because explicit price discrimination is impossible or
illegal. As for most forms of price discrimination, there may be cost-based
justifications for offering discounts when a bundle of products is bought.
For example, when a stereo system consisting of a specific tape deck, tuner
and CD player is sold as a bundle, the manufacturer and the retailer may
benefit from this standardisation and resulting economies of scale and
lower handling costs. Similarly, a restaurant which offers set meals reduces
waste of perishable items.
Commodity bundling, also called joint purchase discounts, was first
analysed by Adams and Yellen (1976) although the idea that bundling
can be used to extract consumer surplus first appeared in a paper by
Stigler (1963) who discusses the case of ‘block booking’ by the American
film industry. Monopoly producers used to (before it was ruled illegal)
bundle films when offering them to theaters. Theaters could not buy the
film they were interested in if they were only interested in one film; they
had to buy the bundle or nothing. This is an example of pure rather than
mixed bundling. Pure bundling refers to the situation in which the
163
28 Managerial economics
Figure 10.6 illustrates pure and mixed bundling for consumers with
independent demands for two goods. Each consumer is represented by a
point in the square OABC, marking the pair of reservation prices for both
goods. Consumers buy at most one unit of each product and a consumer’s
reservation price for the bundle equals the sum of his reservation prices for
the two goods. Consumers cannot resell goods. If goods are sold separately
(unbundled sales) then all consumers with reservation prices above the
set prices p1 and p2 buy Product 1 and Product 2 respectively (see Figure
10.6a).
C
r B
2
Buy 2 Buy
only both
p
2
Buy Buy
neither 1 only
r
0 1
p A
1
Figure 10.6a: Unbundled sales
If the seller only offers the goods as a bundle, then only consumers whose
reservation price for the bundle exceeds the bundle price (r1 + r2 ≥ pb ) will
buy (see Figure 10.6b).
r
2
Buy
bundle
p
b
Don’t
buy
r
p 1
b
p - p
b 2
r
2 G r +p - p
H 1 2 1
p
b
B
p A
2
C E p -p1
D E b
0 p r
p 1
1 b
166
Chapter 10: Monopolistic pricing practices
year. Total yearly VCR sales however are about 50 million. Arthur D. Little
Enterprises (ADLE), the consultancy marketing the new technology, not
surprisingly, tries to convince VCR manufacturers to sell their products
with Commercial Brake fitted.13 13
See ‘Ad nauseum’
The term tied sales is often used to describe a marketing practice which
is related to commodity bundling. Whereas in commodity bundling the
proportions of the items in the bundle are fixed (e.g. one bar of soap and
one towel), under tied sales, the consumer can decide on the quantities
in the bundle. Under tied sales, a manufacturer refuses to supply a
product for which he has a monopoly unless the consumer agrees to
also buy some (complementary) product from the manufacturer. When
this complementary product is available in a competitive market, tying
sales can be seen as an attempt to extend monopoly power. Usually there
is another reason to tie sales, namely to discriminate between types of
customers. As part of their standard leasing arrangement for photocopiers,
Xerox used to insist that customers bought Xerox paper. This allowed
it to price discriminate between high intensity and low intensity users
by charging a high paper price. If higher use of paper indicates a higher
willingness to pay for the photocopier lease then customers reveal their
type through the amount of paper they use. Tied sales is a mechanism to
extract more consumer surplus from the keener users. Of course Xerox
could have negotiated different rentals with low and high intensity users
but this leaves the problem of identifying which class any user belongs to.
Manufacturers who supply replacement complementary goods to the
durable good they sell may be tempted to design their products so
that only products in their range are compatible. This constitutes a
technological tie. For example, a printer manufacturer could design
his products so that only cartridges from the same make can be used.
Camera manufacturers design lenses and camera bodies so that they have
limited compatibility - usually only within the brand name - and often
there is incompatibility between successive generations of products. In
the US there have been several lawsuits concerning technological ties and
designed incompatibility.
Multiproduct firms
In mainstream microeconomic models the firm is usually modelled as a
single product manufacturer. In our discussion of production and pricing
we have focused on the case of a firm producing a single output. This is
done for convenience of course as we know that practically all firms are
multiproduct firms. In fact some of the most interesting pricing questions
arise when firms have to take into account that their pricing policy for
one of their products has significant implications on the demand and
profitability of another. Generally, when pricing and marketing decisions
for individual products are not coordinated the firm does not perform
as well as it could. Particularly, a company structure’ in which separate
divisions are given responsibility for the profitability of an individual
product can be undesirable.
167
28 Managerial economics
Example 10.4
168
Chapter 10: Monopolistic pricing practices
169
28 Managerial economics
the price net of transportation costs is not high enough. Gas which
is released when drilling for oil is often burnt because its transportation
cost from remote oil fields is too high. In the Falkland Islands where
sheep are farmed for wool, the meat goes to waste presumably because
of the high transportation costs. This state of affairs is not helped by
British regulations requiring that all meat for the British forces is imported
including ‘mutton granules’.15 15
See ‘The richest
islands in the world,
Example 10.5 maybe’
26
∑ MR
16 MC
10
8 MC ′
D D
j p
25 40 q
50 80
Figure 10.7: Joint products
Now suppose the cost function is C(q) = 25 + q2/35 and there is free
disposal (i.e. it is costless to discard excess pineapple juice). Note in
Figure 10.8 that marginal cost (MC' ) now intersects the sum of the
marginal revenue curves where marginal revenue of juice would be
negative if all of the juice produced is sold. The profit maximising
producer should set MRp(qp) = MC'(qp) or (80 – 2q)/5 = 2q/35 which
implies production of qA = 35 cans of pineapple. Rather than selling 35
cans of pineapple juice, the seller should restrict sale of juice to the
level where marginal revenue is zero (i.e. qj = 25 cans).
170
Chapter 10: Monopolistic pricing practices
Transfer pricing
A transfer price is the price which is charged when one division of
a vertically integrated firm (say the production division) sells an
intermediate good or service to another division (say the marketing
division). When such internal sales take place, maybe in addition to sales
of the intermediate good on the external market, the determination of
the transfer price is important for the firm’s profitability. At first sight
this statement may puzzle you: why should it matter what price GM’s
parts division charges its assembly plants? By definition, the payment
for parts is a transfer; it is like putting money from the left pocket into
the right pocket. The reason transfer prices are important is that they are
used to provide proper incentives for individual divisions to take actions
which maximise total firm profit. Many large-scale enterprises have a
decentralised structure consisting of semi-autonomous profit centres.
Such stand-alone divisions are often established to avoid excessive
communication and coordination costs in large firms. Since in this
scenario divisions are rewarded on the basis of their performance, the
selling and buying divisions have opposing interests in the determination
of the transfer price. The selling division has an interest in setting the
transfer price high, whereas the buying division prefers a low transfer
price. If transfer prices are not set properly or if the transfer price is left to
negotiation between divisions, total firm profit suffers.
Following Hirshleifer’s (1956) seminal paper on transfer pricing we will
develop the analysis for three scenarios distinguished by whether there
is an outside or external market for the intermediate good and if there
is, whether this market is competitive or not. To simplify the analysis
we assume that units are defined so that one unit of the final good
requires one unit of intermediate good. We also assume that there are
no technological synergies between the production and the marketing
division (i.e. the cost function for either division is not affected by the
output of the other).
171
28 Managerial economics
p, MC
∑ MC
p
f MCi
MCf
p
t
D
MR
q
f q
172
Chapter 10: Monopolistic pricing practices
p, MC
MCf + p
p e
f MCi
p =p
t e
MCf
D
MR
q q
f i q
of MCfc + pe and MR. The external market price represents the opportunity
cost of using a unit of the intermediate good internally. Of course the same
analysis applies when the final good is sold in a competitive market so that
its MR curve is horizontal at the market price.
If the firm is a net buyer of the intermediate good, its optimisation
problem becomes:
max pf (qf )qf – pe qe – Ci (qf – qe ) – Cf (qf )
which results in
MRf (qf) = MCi(qf – qe) + MCf (qf )
and
pe = MCi (qf – qe ).
This situation could be illustrated graphically as in Figure 10.9 but with
MCi and pe intersecting to the left of the optimal output level for the final
good.
When there is an external market, profit maximisation generally calls
for the vertically integrated firm to use it. The optimal quantities of
intermediate and final good produced are not equal. As a consequence,
when the divisions are prevented from using the external market,
profitability suffers. For example, a cement producer cannot increase
profitability when he buys a ready-mix-concrete firm with the sole
intention of using it as a captive customer.
The proper transfer price when there is a competitive external market
is the external market price. This will induce both divisions to produce
the optimal quantities if they are instructed to maximise profit. This
transfer price rule is also intuitively appealing to managers of the separate
divisions. The manufacturing division would not want to sell below the
market price and forego a more profitable outside sale and the marketing
division does not want to pay more than the market price. Headquarters,
when setting the transfer price in this scenario, does not need any cost
information. Because of its simplicity the rule of equating transfer price to
external market price is used almost without exception in practice.
Note that the marketing division does not benefit from being vertically
integrated with the manufacturing division. It could buy the intermediate
good under the same conditions on the external market. There is no
rationale for the vertically integrated structure under this scenario. We
obtain this rather surprising result because of the assumption of demand
independence. In the more realistic setting, when an increase in the
quantity of the intermediate good sold on the outside market has a
depressing effect on the demand for the final good, we would expect that
the firm sets the transfer price below the external market price. On the
other hand, when demand dependence and technological dependence
are negligible, there is a case to be made for ‘outsourcing’ or spinning off
a manufacturing division which operates in a competitive market. This
process has been taking place in the computer industry which used to
have a multi-layered vertical structure. Most PC makers in fact restrict
themselves to little more than contracting with electronics suppliers,
assembling and marketing their machines. Several of these electronics
manufacturers arose from restructuring in computer companies which
spun off some of their production capacity.18 18
See ‘Labouring in
obscurity’
The issues of make-or-buy decisions and demerger are however very
complex and we cannot hope that a simple transfer price model like
the one outlined in this section will give us all the answers. Arguments
174
Chapter 10: Monopolistic pricing practices
p
e
∑ MR
De
p* Df
e
p* MCi
f
MCf
q* q* q q
e q f f q* i
e i
Example 10.6
In all three scenarios we have considered, the transfer price should be set
equal to marginal cost. In scenario 1, marginal cost corresponding to the
optimal output level of the final good is relevant; in scenario 2, the market
price is used but this is equal to marginal cost when the manufacturing
division maximises profits; and in scenario 3, marginal cost corresponding
to total production (for internal and external use) is the valid price. In
practice, because it is convenient to set transfer price equal to market
price, mistakes are made when market conditions are as in scenario 3.
More complicated scenarios can be modelled for firms in different market
structures and where the assumptions of demand independence and
technological independence do not hold. This is however beyond the scope
of this subject guide.
176
Chapter 10: Monopolistic pricing practices
177
28 Managerial economics
Sample exercises
1. This is an extract from ‘Serving your needs. A guide to telephone
services in your home 1993–94.’ published by British Telecom:
‘We work out the cost of each call individually, and measure the call in whole
units. Each unit buys a period of time. (The basic unit currently costs 4.935 p
including VAT.) The length of time given for each unit depends on when and
where you are phoning from, and where your call is to. (The unit rate may
vary depending on how many calls you make... We have introduced three
`Customer Options’ which give discounts on the basic unit rate depending on
the number of calls you make:
Standard Personal: A 5% discount is automatically given on all direct
dialled calls you make over £58.75 in one quarter. This discount increases to
8% on calls made over £293.75 in a quarter. (A quarter is an average of 91
days.)
Option 15: For a £4 quarterly charge you can apply for our new high-value
scheme which offers a 10% discount on all direct-dialled calls at the basic unit
rate. You will benefit from this if your call charges are consistently more than
£40 per quarter including VAT.
Supportline: If you make very few calls (less than 125 units per quarter) you
can get: half-price standard rental and the first 30 units free, On the next 120
units per quarter, you pay a rate which is higher than the standard unit rate.
After you have used 150 units the price falls to the standard unit rate.’
a. Draw budgetlines corresponding to each of the three ‘Customer
Options’. The x-axis is the number of units per quarter and the
y-axis is remaining income. Remember the rental.
b. If my current phone bill is less than £40 per quarter, it does not
make sense for me to go for Option 15. True or false?
c. Why does BT offer these options?
2. In 1979 the Belgian government forced BMW to charge a low price for
cars sold in Belgium (price ceiling). BMW attempted to ban its dealers
from exporting the cheaper cars, offered for sale in Belgium, to other
countries. The European Commission condemned these bans. Show
the possible effect of the Commission’s decision by graphing BMW’s
price-discriminating strategy (selling at price ceiling in Belgium and at
higher price elsewhere) and graphing the non-discriminating scenario.
Show that BMW might decide to quit the Belgian market altogether or
alternatively, charge the Belgium controlled price throughout the EEC.
3. Off-peak demand for a service is given by q1 = 100 – 2p1 and peak
demand is q2 = 300 – p2. Marginal cost per unit is c = 10 and capacity
cost g = 90 per unit. Find the optimal peak and off-peak price.
4. Suppose Gillette knows it has two types of customers: about 50% are
of Type 1 and 50% are of Type 2. A Type 1 customer is willing to pay
up to £1 for a razor and up to £1.50 for a package of 10 razorblades. A
178
Chapter 10: Monopolistic pricing practices
179
28 Managerial economics
b. Suppose now that Hummer & Co has market power in the external
market for cotton. It cannot buy cotton from the external market,
but it can sell cotton according to an external demand function
pe = 358 – 0.09qe. Determine the optimal quantity and price for
bandages, and how much cotton (if any) is sold on the external
market and, if any is sold, the transfer price.
180
Chapter 11: Oligopoly
Aims
The aim of this chapter is to consider:
• the differences in assumptions with respect to decision variables and
strategic (a)symmetry between the various oligopoly models discussed
in this chapter
• why there is a temptation to cheat in a cartel
• the name of the Nash equilibrium in the Bertrand model
• the concept of a profit possibility frontier
• why the division of profit can be problematic in a cartel
• how MFC clauses can facilitate collusion
• the nature of regime-switching models, the role of uncertainty in
these models and their predictions about price wars.
Learning outcomes
By the end of this chapter, and having completed the Essential reading and
activities, you should be able to:
• solve Stackelberg, Cournot, Betrand and collusion models
• solve a dominant firm model analytically and graphically
• demonstrate why in a repeated Bertrand model cheating is more
likely when there are many firms are impatient.
Essential reading
Tirole, J. The Theory of Industrial Organisation. Chapters 5 and 6.
Varian, H.R. Intermediate Microeconomics. Chapter 27.
Further reading
‘A bonus for Saddam’, The Economist, 11 March 1995, pp.105–06.
Abreu, D., D. Pearce and E. Stacchetti ‘Optimal cartel equilibria with imperfect
monitoring’, Journal of Economic Theory 39(1) 1986, pp.191–225.
Allen, B. and J.-F. Thisse ‘Price equilibria in pure strategies for homogenous
oligopoly’, Journal of Economics and Management Strategy 1(1) 1992,
pp.63–82.
Anderson, S.P. and M. Engers ‘Stackelberg versus Cournot oligopoly
equilibrium’, International Journal of Industrial Organisation 10 1992,
pp.127–35.
Bertrand, J. Book review of ‘Theorie Mathematique de la Richesse Sociale’ and
of ‘Recherches sur les Principes Mathematiques de la Thoerie des Richesses’,
Journal des Savants 68 1883, pp.499–508.
Boyer, M. and M. Moreaux ‘Being a leader or a follower’, International Journal
of Industrial Organisation 5 1987, pp.175–92.
Brander, J.A. and A. Zhang ‘Dynamic oligopoly behaviour in the airline
industry’, International Journal of Industrial Organisation 11 1993,
pp.407–35.
181
28 Managerial economics
Brannman, L.E. and J.D. Klein ‘The effectiveness and stability of highway
bid-rigging’ in Audretsch, D.B. and Siegfried, J.J. (eds) Empirical Studies in
Industrial Organisation: Essays in Honour of Leonard W. Weiss. (Dordrecht:
Kluwer, 1992) [ISBN 0792318064] pp.61–75.
‘Business and Finance’, The Economist, 3 December 1994, p.7.
‘Clean streets’, The Economist, 12 March 1994, pp.53–56.
Cooper, T.E. ‘Most-favoured-customer pricing and tacit collusion’, RAND
Journal of Economics 17(3) 1986, pp.377–88.
Cournot, A.A. Recherches sur les Principes Mathematiques de la Theorie
des Richesses. (1838) English edition, Bacon, N. (ed) Researches into the
Mathematical Principles of the Theory of Wealth. (New York: Macmillan,
1897).
‘Disputes are forever’, 1986, 17 September 1994, pp.93–94.
Edgeworth, F. ‘La Teoria Pura dle Monopolio’, Giornale degli Economisti 40
1897, pp.13–31.
Encaouna, D. and A. Jacquemin ‘Degree of monopoly, indices of concentration
and threat of entry’, International Economics Review 21(1) 1980, pp.87–105.
Geroski, P.A., L. Philips and A. Ulph Oligopoly, competition and welfare. (Oxford:
Blackwell, 1985) [ISBN 063114479X].
Gravelle, H.S.E. and R. Rees Microeconomics. (Harlow: FT Prentice Hall, 2004)
third edition [ISBN 0582404878] Chapter 16.
Green, E.J. and R.H. Porter ‘Non-cooperative collusion under imperfect price
information’, Econometrica 52(1) 1984, pp.87–100.
Hall, R.L. and C.J. Hitich ‘Price theory and business behaviour’. Oxford
Economic Papers 2 1939, pp.12–45.
Kreps, D.M. and J.A. Scheinkman ‘Quantity precommitment and Bertrand
competition yield Cournot outcomes’, Bell Journal of Economics 14(2) 1983,
pp.326–37.
Levinstein, M. ‘Price wars and the stability of collusion: a study of the pre-
World War I bromine industry’, NBER Working Paper Series on Historical
Factors in Long-Run Growth (August 1993) p.50.
Neilson, W.S. and H. Winter ‘Bilateral most-favoured-customer pricing and
collusion’ RAND Journal of Economics 24(1) 1993, pp.147–55.
‘Oil rolls over’, The Economist, 2 April 1994, p.75.
Ono, Y. ‘The equilibrium of duopoly in a market of homogenous goods’,
Economica (1978) 45, pp.287–95.
‘Scored’, The Economist, 20 November 1993, p.5.
Spulber, D.F. ‘Bertrand competition when rivals’ costs are unknown’, The
Journal of Industrial Economics XLIII(1) 1995, pp.1–11.
‘Smelt a rat’, The Economist, 23 July 1994, pp.72–73.
‘Steel woes’, The Economist, 19 February 1994, p.18.
Stigler, G. ‘A theory of oligopoly’, Journal of Political Economy 72(1) 1964,
pp.44–61.
‘Still smokin’, The Economist, 11 March 1995, pp.93–94.
Sweezy, P.M. ‘Demand under conditions of oligopoly’, Journal of Political
Economy 47(4) 1939, pp.568–73.
‘Then there were seven’, The Economist, 5 February 1994, pp.19–24.
von Stackelberg, H. Marktform und Gleichgewicht. (Berlin: Springer, 1994).
Introduction
In terms of number of firms and consumer price, oligopoly is an
intermediate market structure between monopoly and perfect competition.
Firms may price at marginal cost as in perfect competition, they may
set a monopoly price or a price between these two extremes. An
oligopolistic industry consists of a few firms who recognise their strategic
interdependence. This strategic interaction, rather than the number of
182
Chapter 11: Oligopoly
184
Chapter 11: Oligopoly
Strategic asymmetry
In this section we study models in which firms for whatever reason do not
make their decisions simultaneously. Simultaneity does not necessarily
have a chronological meaning. In game theory ‘simultaneous decisions’
refers to the situations in which players make decisions ignorant of the
decisions taken by the other players. In sequential models there is a
‘leader’ who makes a first move, after which the ‘followers’ make their
decisions. This type of analysis is obviously only appropriate when there is
an industry leader and it is clear which firm assumes the leadership role.
Stackelberg
In the Stackelberg model, the ‘leader’ (Firm 1) makes a quantity decision,
which can be observed by the ‘follower’ (Firm 2) who in turn decides on
quantity. Price depends on total output and the exact relationship between
output and price is given by the demand curve p(q1 + q2). Let’s consider
the follower’s problem first. The follower has to determine an optimal
output level q2 given that the leader has chosen q1. Hence, his optimisation
problem is:
max p(q1 + q2) q2 – c2(q2 ).
q2
The result (i.e. the optimal choice of q2) will generally depend on q1: q2 =
q2 (q1). The leader can anticipate the follower’s choice of quantity and its
dependence on his own quantity decision and therefore his problem is:
max p(q1+q2(q1 ))q1 – c1(q1 ).
q1
Example 11.1
185
28 Managerial economics
Dominant firm
The dominant firm model applies to industries in which there is one large
firm (or a cartel) acting as a price leader and several small firms. The
small firms in the industry (the fringe) act as competitive firms in that
they take the price set by the dominant firm as given and they supply all
they want at this price. In other words, they maximise their profits by
determining their supply quantity as the quantity for which marginal cost
equals the given price. In contrast to the perfectly competitive model there
is no condition of zero profits. The dominant firm supplies the residual
demand. As in the Stackelberg model, the dominant firm as the leader has
the advantage of influencing the fringe firms’ or followers’ quantities, in
this case by imposing the price they have to sell at.
Figure 11.1 illustrates how the price decision is made. For any price, the
dominant firm can determine how much will be sold by the fringe firms
by horizontally adding their MC curves. This is indicated by the SMC (sum
of marginal cost) curve. The horizontal difference between the industry
demand curve D and the SMC is the residual demand (RD) faced by the
dominant firm. Profit maximisation by the price leader requires setting
marginal revenue (MR) corresponding to the residual demand equal to the
fern’s marginal cost (MC). This leads to an output of q* by the dominant
firm. The optimal price can be read from the residual demand curve as p*.
Fringe firms supply q′ at price p*.
SMC
p*
MC
RD D
MR
MC
q
q* q′
the fringe firms’ marginal cost SMC. The fringe firms therefore do not
produce and the dominant firm is the monopoly producer. In practice, a
dominant firm may decide not to undercut the fringe firms even when
that would deliver the highest profit. The reason is that refusing to
accommodate the fringe firms or taking over some of them may provoke
legal action.
If there are no barriers to entry and the small firms are making positive
profits, entry will occur, shifting the SMC curve to the right and lowering
the dominant firm’s market share. For example, US Steel’s market share
decreased from 75 per cent in 1903 to less than 25 per cent in the 1960s.
American Can which used to supply 90 per cent of the tin market in 1901,
saw its market share decrease to 40 per cent by 1960. Such market entry
may alter the way in which the industry operates. The dominant firm may
not be willing to tolerate a very large fringe sector.
Theoretically, for the dominant firm model to apply, the dominant firm
should be powerful so that it can credibly threaten to punish fringe firms
if they do not accept its price leadership. This punishment would take
the form of driving the fringe firms out of business through a price war.
Therefore the dominant firm should have low costs, substantial market
share and large production capacity to force the other firms in the industry
to set the same price. In the UK white salt market, however, there is
evidence of British Salt following ICI in its pricing whereas British Salt is
the low-cost producer. Examples of industries for which the dominant firm
model may be plausible include (dominant firms in brackets):
• the UK chemicals industry (ICI)
• the European soft drinks industry (Coca-Cola)
• the US aluminium industry (ALCOA)
• the US airline industry (American Airlines)
• the European steel industry (British Steel)
• photocopiers (XEROX)
• cars (GM)
• cameras and film (KODAK).
The reasons why particular firms get to be leaders may be related to size,
management style or they may be historical. For example, Coca-Cola was
exempt from sugar rationing during the war in return for supplying cheap
Coke to American troops in Europe. This gave Coca-Cola a significant
advantage over Pepsi Co at least in the European market. Federal Express,
probably because of first mover advantages, is seen as the price leader in
overnight delivery.
The leadership role can switch between large players in an industry. In the
US candy bar market Hershey was dominant in the 1960s. This dominance
was later challenged by Mars in the early 1970s. The term barometric
price leadership refers to firms deliberately alternating the dominant
firm role to avoid attracting the authorities’ attention. The firm which
takes the leader role is the first to announce a price change when a change
in cost or demand conditions warrants it.
The dominant firm model can be modified to allow for several large firms
dominating the market.2 These market leaders can be modeled as joint 2
See, for example,
profit maximisers (a cartel) in which case our analysis above is basically Encaoua and Jacquemin
unaltered or they may be assumed to behave in Cournot fashion with (1980)
187
28 Managerial economics
example, three giant companies (BAT Industries, Philip Morris and RJR
Nabisco) jointly have a 90 per cent market share.4 4
See ‘Still smokin
Example 11.2
Symmetric models
The theoretical difficulty of justifying asymmetric models lies in the fact
that the asymmetry is not explained within the model. There have been
some recent attempts5 to endogenise the strategic (a)symmetry in oligopoly 5
See, for example,
models. Unless we know the particular industry we are studying well, and Ono (1978); Boyer
and Moreaux (1987);
can appeal to, say, historical reasons for the asymmetry, it is perhaps more
Anderson and Engers
sensible to assume that no firm has a strategic advantage. This assumption (1992)
is the starting point for simultaneous moves oligopoly models.
Cournot
In the Cournot (1883) model firms make simultaneous quantity decisions.
In game theory terminology, the duopolists are engaged in a non-cooperative
game and we are interested in the Nash equilibrium of this game (i.e. a pair
of quantity choices such that it is in neither firm’s interest to alter its choice
unilaterally). These strategies are consistent in the sense that no firm has
ex post regret when it observes its competitor’s quantity decision. The good
produced by the Cournot duopolists is homogenous and therefore the price
is determined by the total quantity supplied: p(q1 + q2). To find the Nash
equilibrium, we find best response functions (i.e. we find the best quantity
choice of Firm 1(2) as a function of Firm 2 (1 )’s quantity choice). The
intersection of these best response functions gives the Nash equilibrium.
Mathematically this means that for Firm 1 we solve:
max p(q1 + q2 ) q1 – c1(q1 )
q1
and find q1 = f1 (q2) and for Firm 2 we find q2 = f2(q1) similarly. The Nash
equilibrium is then found by solving q1 = f1( f2(q1)).
188
Chapter 11: Oligopoly
Example 11.3
q
2
16
10
8
f
2
q
f 1
1
4 8 20
Example 11.4
Since all firms are identical we look for a symmetric Nash equilibrium
in which all output decisions are identical i.e. qi = Q/n so that the
condition above can be rewritten as:
()
Q ∂p
p+ n
∂Q
= c,
189
28 Managerial economics
or
p–c 1
.
p = nη
Bertrand
Bertrand (1883) wrote a book review of Cournot’s work and disagreed
with Cournot about his assumptions on the behaviour of oligopolists.
Bertrand thought that oligopolists would collude rather than compete and
to demonstrate how unrealistic Cournot’s description was, he developed
a model in which the decision variable is price. He showed how Cournot’s
way of thinking leads to the implausible result that in a duopoly the
perfect competition price is charged.
We can think of the Bertrand model as a non-cooperative game in which
each firm decides on price, taking the other prices as given. We then
look for a Nash equilibrium as in the Cournot model. The assumptions
of the Bertrand model are that the product is homogenous, consumers
have complete information and there are no search or transport costs
so that they buy from the firm charging the lowest price. If firms charge
the same price they are assumed to share the market and have identical
market shares. Assume firms have identical and constant marginal costs
and unlimited capacity. Under these assumptions we find the very striking
result that there is a unique Nash equilibrium in which both firms set
price equal to marginal cost. Why is this the Nash equilibrium? Clearly,
price could not be below marginal cost since then firms would be making
losses. Price cannot be above marginal cost since then one of the firms
could reduce its price by a very small amount, capture the whole market
and make a positive profit. If firms differ in (constant) marginal costs,
then the low cost firm prices just below the second lowest marginal cost at
equilibrium unless that price is above the monopoly price.
Example 11.5
190
Chapter 11: Oligopoly
The assumption of consumers buying from the firm charging the lowest
price, even if the price difference is very small, is quite restrictive and
justifiable only when the goods are very close substitutes. In fact,
oligopolists strategically differentiate their products to reduce tough price
competition. When products are differentiated, a firm can charge a higher
price than its rival without losing all its business. Even when the good
is homogenous, it is not realistic to assume that all demand goes to the
lower price firm. Allen and Thisse (1992) have developed a pure oligopoly
model in which some customers do not care about small price differences.
This implies that firms do not lose all of their sales if they are undercut
slightly by a rival. The equilibrium of their pricing game allows for some
market power by the firms. Spulber (1995) shows that the assumption of
firms knowing rivals’ costs is a crucial ingredient in the Bertrand model. If
firms know only their own costs and have probabilistic information about
rivals’ costs, they set price above marginal cost at the equilibrium of the
pricing game.
It is difficult to think of situations in which the Bertrand model is a
reasonable approximation of reality. The scenario which fits the original
Bertrand model best is maybe that of firms submitting sealed bids for a
procurement contract where the contract is awarded to the firm quoting
the lowest price. However, even in these circumstances, firms find ways
to avoid tough price competition. In the 1950s, General Electric and three
other firms rotated sealed-bid business for circuit breakers. They held
secret meetings to decide market shares in advance. More recently, two
of America’s biggest bakeries, Continental Baking and Campbell Taggert,
have been investigated for alleged bidrigging in sales of bread to schools
and hospitals. In Japan in 1993, the Fair Trade Commission conducted
raids on some big electronics companies, including Sony and Toshiba, who
allegedly colluded on bids for electronic billboards in sports stadiums.6 6
See ‘Scored’
191
28 Managerial economics
game, the capacities chosen in the first stage are the Cournot equilibrium
quantities and in the second stage firms choose (identical) prices such that
demand equals joint capacity.
Collusion
Chamberlin was one of the first economists (after Adam Smith) to suggest
that oligopolistic sellers would form a cartel and cooperate to set joint
profit maximising price and output levels. Cartels were common before
the antitrust laws were passed. Levinstein (1993) vividly describes how
the US and Europe bromine producers colluded in the 30 years before
World War I. Currently, in most of the western world (Switzerland is a
notable exception), the norm is that collusion is illegal although some
cartels are sanctioned by governments. For example, the International Air
Transport Association, consisting of American and European airlines flying
transatlantic routes, used to set uniform fares for transatlantic flights. The
US allows agricultural committees to determine prices and production
quotas for some products. Sometimes collusion between firms in the
domestic market is explicitly forbidden but governments allow firms to
participate in international cartels. The European steel industry operated
as a price fixing cartel until 1988 and was fined in 1994 by the European
Commission.7 In the same year, the Commission fined 42 cement firms and 7
See ‘Steel woes’
trade groups for running a cartel.8 8
See ‘Business and
Finance’
Because price fixing is illegal, there are not many open cartels but firms
may nevertheless find ways to cooperate at the expense of customers.
Powerful firms use the media to announce price changes which are followed
by the other firms in the industry. All firms in the industry use the leader’s
price as a focus point and the risk of ‘misunderstandings’ is minimal. This
is a form of implicit or tacit collusion. Trade associations or professional
associations are often vehicles for collusion. This certainly seems to be the
case for the Associations of Trade Waste Removers and their central council
which hold a tight grip on the rubbish-collection business for commercial
customers in New York City. The rate charged by New York carters is fixed
at three or four times the rate in Los Angeles or Chicago. The business is
notorious for its connections with the Mafia.9 9
See ‘Clean streets’
192
Chapter 11: Oligopoly
profit 2
NE o max
profit 1
which gives the result mentioned above: marginal revenue equals each
firm’s marginal cost at its optimal output level.
193
28 Managerial economics
We can use this analysis to show that, for cartel members, since they are
playing a prisoners’ dilemma game, there is always a temptation to cheat
by overproducing. Given that firms, even when they have formally agreed
to collude, have no recourse to a court of law (because collusion is illegal),
this temptation is very real. Consider firm l’s profit given that firm 2
produces the joint profit maximising quantity q2*:
π1 (q1, q2*) = p(q1 + q2*) q1 – c1 (q1 ).
If firm 1 assumes that firm 2 is going to stick to q2* it finds its optimal
output level q1* by differentiating this function with respect to q1 which
gives:
∂p ∂c1
p(q1+ q2* ) + q1 – (2)
∂Q ∂q1
Comparing this to the first order condition for q1 above in (1) reveals that
(2) is equal to:
∂p
– p(q2)
∂Q
which is positive. This indicates that Firm 1 can increase its profits by
increasing output if Firm 2 keeps output constant at the ‘collusive’ amount.
The intuition for this is that marginal revenue for the cheater is close to
the cartel price whereas his marginal cost equals marginal revenue and is
thus lower than the cartel price.
Even when cartels can monitor and enforce prices effectively, cartel
members may engage in non-price competition by offering free extras such
as after-sale service or free delivery and installation. Airlines offer better
meals, free in-flight entertainment etc., in order to lure customers away
from other cartel members. This type of non-price competition can be
interpreted as cheating and it is not as easily detected as price cutting.
Case: OPEC
The standard example of a cartel is OPEC which restricts oil output
by its members so as not to spoil the market. OPEC’s history has
repeatedly illustrated the difficulties of maintaining a cartel and
avoiding cheating by cartel members. Within OPEC there is tension
between the rich countries with large oil reserves and a small
population (e.g. Kuwait and Saudi-Arabia, which alone accounts for a
third of OPEC’s output) and poor countries with small oil reserves and
a large population (e.g. Libya and Indonesia). The countries with large
reserves want a low price since they are concerned about the long-
term effect of high oil prices: customers may start switching to other
energy sources and there may be market entry in the form of non-
OPEC exploration and production such as in the North Sea. Countries
with low reserves do not worry about the long-term effects and, since
short-term demand elasticity for oil is low, they argue for a high price.
The poor OPEC countries have consistently overproduced their quotas 10
See ‘Oil rolls over’; ‘A
and even the United Arab Emirates decided to increase its production bonus for Saddam’
to 1.5 million barrels a day from its allocation of one million barrels in
1988. The lifting of sanctions against Iraq, imposed by the UN after the
invasion of Kuwait, may put more pressure on the cartel.10
194
Chapter 11: Oligopoly
Case: Diamonds
When industries succeed in filtering sales through a common
distributor, collusion is not an unlikely mode of behaviour. The
Central Selling Organisation (CSO), a subsidiary of De Beers, aims to
control the world wholesale market in rough diamonds. In 1993, De
Beers itself accounted for 50 per cent of CSO sales and Russian sales
amounted to 26 per cent. Even this seemingly cosy cartel arrangement
has had problems. Russia agreed in 1990 to sell 95 per cent of its
output through the CSO for a period of five years but it is unlikely
that this agreement was honoured. In particular, Russia has exploited
loopholes such as classifying diamonds as ‘technical’ and hence not
covered by the agreement. De Beers has had to buy this leaked output
to keep control of the flow of diamonds on to the market and hence
their price.11 11
See ‘Disputes are
forever’
Given the incentive to overproduce, cartels use several mechanisms to
detect and prevent cheating e.g. forcing firms to publicise prices. Firms
may adopt ‘facilitating practices’ which eliminate the incentive to cheat.
For example, cartel members may agree to offer a ‘meet or release clause’.
Such a clause applies when a customer finds a lower price offered by
another supplier, in which case the low price is met by the original seller
or the customer is released from the obligation to buy. Thus way cartel
members are informed if any undercutting takes place. Another ‘trick’
used by colluders is the ‘most favoured customer clause’ (MFC). Under
MFC, firms promise their customers that if they ever lower the price, they
will offer a rebate, equal to the difference between the price customers
pay now and the new price. In this scenario, if firms can agree to collude
for a small number of periods while offering the MFC, then their pay-offs
are such that deviating from the collusive outcome by lowering price is
no longer profitable because of the penalties which would have to be
paid to previous customers. Both General Electric and Westinghouse,
manufacturers of turbine generators, used MFC, effective for six months
after a sale, in the 1960s and 1970s until they agreed to end the practice
to avoid antitrust prosecution.12 12
See Cooper (1986)
and Neilson and Winter
(1993)
Dynamic interaction
So far we have discussed models which describe oligopolistic behaviour
when the firms play a one-shot game. More realistic models assume a
dynamic setting and in particular the repeated prisoner’s dilemma
provides a natural framework. As we have seen before, collusion can be
an equilibrium if the time horizon is infinite or if there is uncertainty
about the length of the horizon. Such a collusive equilibrium is sustained
by players’ threats to retaliate (i.e. to revert to the non-cooperative Nash
outcome as soon as one player deviates). To illustrate this idea consider a
simple infinitely repeated Bertrand game. All n firms are identical and the
price which maximises joint profit is pm, the monopoly price. If all firms
charge pm each makes a profit equal to a share 1/n of the monopoly profit
Пm. If one of the firms undercuts however, it captures the whole market.
Although repetition of the one-period Bertrand solution with all firms
pricing at marginal cost in each period is an equilibrium in this game,
there may be a Nash equilibrium in which firms collude. Suppose firms
use trigger strategies of the following type. collude (i.e. set price pm)
until at least one firm deviates; when one or more firms deviate in a given
195
28 Managerial economics
period, set price equal to marginal cost from the next period onwards. For
this combination of trigger strategies to form an equilibrium it should be
the case that it does not pay for any firm to do something else assuming all
the other firms are sticking to their trigger strategies. If a firm decides to
deviate, it gains:
Пm – Пm/n = Пm (n – 1)/n
immediately as it becomes a monopoly. Its loss, compared to the collusive
outcome, due to retaliation from the next period onwards is:
(Пm/n)(δ + δ2 + ...) = (Пm/n) (δ/(1 – δ))
where δ is a discount factor. The firm cheats if the gain exceeds the loss or
when:
δ < 1 – 1/n.
This proves that, at least in this simple pricing model, cheating is more
likely when the number of firms in a cartel is large. When n = 50, a
discount factor above 0.98 is needed to sustain collusion whereas for an
industry with n = 2 firms, a discount factor above 0.5 is sufficient. The
dependence on the discount factor is clear: when firms are patient
(high δ), they evaluate their future losses more highly and this deters them
from cheating.
We have analysed repetition of the Bertrand pricing game here but a
similar story can be told about repeating the static Cournot game. The
trigger strategies there involve playing Cournot equilibrium output forever
as soon as cheating occurs. In contrast to the Bertrand situation where you
don’t sell at all when someone cheats, firms in a Cournot industry may not
be able to observe cheating immediately but maybe two or three periods
after it has taken place. Cheaters could get away with earning profit
above their share of the collusive profit for several periods. Detection lags
therefore make cheating more tempting.
Stigler (1964) has pointed out that, when demand fluctuates and there
is a lot of uncertainty, there is larger scope for misunderstanding moves
which merely reflect changed demand conditions as attempts to cheat.
This observation has formed the basis for recent game theoretic work on
dynamic oligopoly.13 Recall that the repeated Bertrand and Cournot models 13
See for example Green
predict that, when collusion is sustained, prices and outputs are stable. and Porter (1984); Abreu
et al. (1986)
Firms choose the joint profit maximising price or output and there are no
price wars. In the recent models which allow for uncertainty, price wars
do occur. To illustrate the main ideas here, let’s assume a homogenous
oligopoly with identical firms. The time horizon is infinite or there is
a chance, in each period, that the game ends. Firms decide on output
levels and they observe only their own output level and the market price
(which, as always, depends on total industry output). Each firm uses these
observables to deduce whether a rival has defected from the (tacitly)
collusive arrangement (i.e. when the market price is low, overproduction,
compared to the joint profit maximising output, has taken place). So
far, we are describing a repeated Cournot game. The ingredient which
has to be added to generate equilibrium price wars is uncertainty. The
relationship between industry output and price is stochastic so that a low
price can be caused by a rival flooding the market or an exogenous fall in
demand.
Green and Porter (1984) show that, when there is uncertainty, there is an
equilibrium in which firms periodically revert from the collusive solution
to static Cournot output levels for a few periods, after which they return
to the collusive output levels. At this equilibrium, firms are using trigger
196
Chapter 11: Oligopoly
Case: Aluminium
Aluminium producers certainly seem to be aware of the dangers of
excess capacity. They agreed to cut back capacity in 1994 and prices
and profits have increased since this agreement. However, Russia -
which is a large producer and exporter - is believed to be cheating
by cutting down its smelting capacity by much less than the agreed
500,000 tonnes. Furthermore, it is quite possible that, without the
agreement, Russia would have had to close more capacity than it has
done now. Russian costs have increased and inefficient Russian smelters
are kept in business only because prices have climbed. Also, as firms
become more profitable, the temptation to cheat and bring capacity
back into production may be hard to resist even for ‘honest’ western
producers.15 15
See ‘Smelt a rat’
There are models which reach exactly the opposite conclusion to Green- 16
See, for example,
Porter, namely: that cartels are less stable during expansionary periods
Brannman and Klein
because firms then have less capacity or inventories to punish the (1992)
cheaters. Also the gain from cheating is larger when there is a boom in
demand. Studies of construction industry auctions concur with this latter
prediction.16
197
28 Managerial economics
198
Chapter 11: Oligopoly
Sample exercises
1. Let the inverse demand curve for a homogenous product be
p = 70 – Q. Suppose there are two firms in the industry, each with
constant marginal costs of 10. Assuming they behave as Cournot
duopolists, what will be the price and output? What price and output
levels does the Stackelberg model predict? What price and output
levels does the Bertrand model predict? What is the collusive price
and output level? Now assume the firms interact each period and
the time horizon is infinite. For which values of the discount factor is
there a collusive equilibrium sustained by trigger strategies if firms
play Cournot forever after cheating has occurred? What if they play
Bertrand forever after cheating?
2. An industry consists of two firms, each of which produce output at a
constant unit cost of 10 per unit. The demand function for the industry
is Q = 1,000,000/p. Find the Cournot and Stackelberg equilibrium price
and quantity.
3. PowerGen and National Power are the only two electricity generating
companies in the UK. The demand for electricity is p = 580 – 3q. The
total cost function of PowerGen is TCP = 410qP and the total cost
function of National Power is TCN = 460qN.
a. If these two firms collude to maximise their combined profits, how
much will each firm produce?
b. How will they divide profits?
4. An industry consists of one dominant firm and five small (fringe) firms
which behave competitively, taking the price set by the dominant firm
as given. Industry demand is Q = 10 – p. The fringe firms have identical
cost functions C(q) = q2 and the dominant firm has constant marginal
cost of 1 per unit. Derive the dominant firm’s residual demand and
corresponding marginal revenue. Illustrate all your derivations on a
graph. What price will the dominant firm set? At this price, how much
does it supply and how much is supplied by the fringe firms?
199
28 Managerial economics
200
Appendix 1: Sample examination paper
Section A
Answer all four questions from this section (12.5 marks each)
1. Each of two firms (A and B) must decide, independently and without
knowing the other’s decision, whether to use its proprietary technology
to manufacture a product or to use its competitor’s technology. Each
firm would prefer that they both use the same technology rather than
different technologies but naturally each prefers that its technology be
the industry standard. The pay-offs are given in the matrix below.
B
use A’s use B’s
technology technology
A
use A’s
2,1 0,0
technology
use B’s
0,0 1,2
technology
a. Find all (including mixed strategy) equilibria.
b. Draw the pay-off region and mark the equilibrium pay-off pairs
corresponding to your answer to (a).
c. Which (if any) of the equilibria are Pareto efficient?
2. ‘It is foolish for a seller to use second price sealed bid auctions because
the winner who bids £10,000 may end up paying only £l which is much
less than his bid.’ Comment.
3. Low quality workers have a value (constant marginal revenue product)
of £10,000 per year whereas the value of high quality workers is
£15,000 per year. Firms cannot distinguish between the two types
of workers but they know that there are as many low quality as high
quality workers.
a. How much are the workers paid assuming the labour market is
competitive?
201
28 Managerial economics
5. Abel & Co. is the price leader in the market for MM-disorder drugs.
The industry demand is given by Q = 90 – 8p. There are four ‘follower’
or ‘fringe’ firms who accept the price set by the dominant firm Abel &
Co. Each of these followers has long run cost function C(qf) = 10qf +
qf2. The leader has a constant returns to scale technology with long run
cost function C(q1) = 10q1.
a. Derive the followers’ supply curve.
b. Derive the net demand facing Abel & Co. and determine the optimal
price and output for the dominant firm.
c. How much is supplied by the followers?
202
Appendix 1: Sample examination paper
Section B
Answer two questions from this section (25 marks each).
6. Lisa Listens supplies services for which the demand function is p(q) =
1– Q/2. The technology exhibits constant returns to scale. Labour is
the only variable input and the firm employs L workers. An individual
worker’s productivity cannot be costlessly observed but it is known
that the expected output of a shirking worker is 1 unit per hour (Q = L)
whereas that of a hard working employee is q (q > 1) units per hour
(Q = qL). The workers and the firm are risk neutral.
a. Draw the ‘marginal revenue product of labour’ curve for the case of
shirking employees and for the case of hard working employees. Do
these curves intersect? Give an interpretation of your findings.
b. At wage w = 1, how much labour is hired if Lisa Listens knows that
its workers will shirk?
c. Suppose Lisa Listens decides to induce its workers to work hard
by paying an efficiency wage and monitoring the workers. As a
consequence the cost per worker equals we+M(p) where we is
the efficiency wage and M(p) is the monitoring cost per worker
incurred to ensure a probability p of catching a shirking worker. A
worker suffers a disutility c from working. If she works hard, this
disutility is c = 0.75 whereas if she shirks it is c = 0. Lisa Listens
pays its workers we unless they are caught shirking in which case
they get nothing. A worker who is caught cheating cannot get a job
elsewhere. Workers have a choice between a job with Lisa Listens or
an outside low effort job (giving disutility c = 0) paying w = 1.
i. Given a detection probability p, what is the minimum wage
we Lisa Listens can pay and still attract workers if the workers
anticipate that they will have to work hard
ii. How large should we be to induce the workers to work hard?
d. Show that the solution to Lisa Listens’ minimum cost
implementation problem is p =1/7 and we = 21/4 for monitoring
cost M(p) =147 p/4. How much labour is hired in this scenario for
q=12?
7. Many charities offer differential membership fees. Depending on
whether the member wishes to receive the charity’s magazine or
whether only membership is desired, a different fee is charged. It is
usually not possible for non-members to purchase a subscription to the
magazine only. Assume that among potential members, willingness to
pay for membership ve is uniformly distributed on [0,E] and willingness
to pay for the magazine va is uniformly (and independently of va)
distributed on [0,A].
a. OUT, a charity for overworked university teachers, charges a
‘package’ price p for membership and a subscription to its magazine
Termtime Blues. Indicate on the [0, E] × [0, A] rectangle which
potential members are interested in this package. Determine
the revenue maximising price p and OUT’s optimal revenue per
potential member. Assume that 2A/3 < E < 3A/2 so that the optimal
price does not exceed A or E.
b. Now suppose OUT changes its membership policy and allows
potential members to buy membership only at a fee f or to ‘opt in’
and buy the magazine as well for a total cost of f + g. Indicate on
the [0, E] × [0, A] rectangle the potential members who opt in, those
203
28 Managerial economics
who are interested in membership only and those who will not
become members.
c. Interpret your findings.
8. An individual with wealth W and strictly concave von Neumann-
Morgenstern utility function U can face either of two states of nature.
In state 1 which occurs with probability 1-p he suffers no loss and
in state 2, which occurs with probability p, he suffers a loss V. The
individual has no control over probability p (there is no moral hazard).
a. An insurance firm offers the individual a policy (R, D), where R is
the premium and D is the payment the individual receives from the
insurance company in state 2. Write down the individual’s expected
utility and the (risk neutral) firm’s expected profit from this policy.
b. Show that if the insurance company maximises the individual’s
utility subject to its expected profit being nonnegative, full
insurance results.
c. Assume now there are two classes of individuals, low and high
risk, with probabilities of loss V of p1 and ph respectively. Although
individuals know their own risk, an insurer cannot distinguish
between high and low risk individuals. Explain why this may result
in only the high risk individuals being insured.
d. Given the scenario in (c) an insurance company decides to offer
a ‘menu’ of two policies: a low premium policy with a deductible
(R1,D1) where D1< V, designed for the low risk group and a high
premium full insurance policy (Rh,V) for the high risk group.
i. Assuming a perfectly competitive insurance industry, write down
the constraints on contract design.
ii. In what way does a low risk individual suffer from the existence
of high risk individuals?
9. Steve derives utility from Ben-and-Jerry’s ice cream (x1) and CDs (x2)
according to utility function U(x1, x2) – √x1+x2.
a. Given p1 and p2, the prices for tubs of Ben-and-Jerry’s ice cream
and CDs respectively, and Steve’s income m, solve his utility
maximisation problem.
b. The price of Ben-and-Jerry’s ice cream has recently gone up from
£2 per tub to £3 per tub. Calculate the equivalent variation and
the compensating variation of this price increase assuming the
price of CDs is £10 and m = 100.
c. Calculate the change in consumer surplus due to the price
change in (b).
d. Due to a MMC investigation into pricing practices in the music
business, the price of CDs decreases from £10 to £7. Calculate
the compensating variation of the combined price changes i.e the
price increase of ice cream and the price decrease of CDs
10. a. Discuss, using graphs where necessary, how the demand for an
input by a competitive industry differs from the horizontal sum
of its firms’ individual input demands.
b. Under World Bank pressure, the Zambian state marketing
monopoly for maize was recently dismantled. Its role was
subsequently taken over by a handful of private companies
which, according to Oxfam reports, pay farmers less than half
the old state price. The World Bank argues that farmer prices will
204
Appendix 1: Sample examination paper
END OF PAPER
205
28 Managerial economics
Notes
206
Appendix 2: References for older versions of textbooks
Main texts
Varian, H.R. Intermediate Microeconomics. (New York: W. W. Norton & Co.,
2003) sixth edition [ISBN 9780393978308].
Varian, H.R. Intermediate Microeconomics. (New York: W. W. Norton & Co.,
1999) fifth edition [ISBN 9780393973709].
Other books
Gravelle, H.S.E and R. Rees Microeconomics. (London: Longman, 1992) second
edition [ISBN 0582023866].
Chapter 2
Page 21: Main texts:
• Varian, sixth edition, Chapters 28 and 29
• Varian, fifth edition, Chapter 28
Chapter 4
Page 43: Main texts:
• Varian, sixth edition, Chapter 36
• Varian, fifth edition, Chapter 36
Chapter 5
Page 57: Main texts:
• Varian, sixth edition, Chapter 17
• Varian, fifth edition, Chapter 17
Chapter 6
Page 73: Main texts:
• Varian, sixth edition, Chapters 2, 3, 4, 5, 6, 8, 9, 10, 12, 13, 14, 15
• Varian, fifth edition, Chapters 2, 3, 4, 5, 6, 8, 9, 10, 12, 13, 14, 15
Page 73: References cited:
• Gravelle and Rees, second edition, Chapters 4, 20
Page 73: Margin note 1:
• Varian, sixth edition, Chapters 2, 3, 4, 5, 6;
207
28 Managerial economics
Chapter 7
Page 95: Main texts:
• Varian, sixth edition, Chapters 18, 19, 26
• Varian, fifth edition, Chapters 18, 19, 26 Chapter 9
Page 127: Main texts:
• Varian, sixth edition, Chapters 22, 23, 24
• Varian, sixth edition, Chapters 22, 23, 24 Chapter 10
Page 141: Main texts:
• Varian, sixth edition, Chapter 25
• Varian, fifth edition, Chapter 25
208
Appendix 2: References for older versions of textbooks
Chapter 11
Page 173: Main texts:
• Varian, sixth edition, Chapter 27
• Varian, fifth edition, Chapter 27
Page 189: Margin note 17:
• Gravelle and Rees, second edition, Chapter 12
209
28 Managerial economics
Notes
210
Appendix 3: Maths checkpoints
Inverse:
Any function can be expressed in its inverse form by changing the
explained and the explaining (endogenous) variable. Mathematically this
means solving the equation for the other variable. Formally the inverse of
a function f is indicated as f –1.
For the above demand function we get
4p = 100 – q
p(q) = 25 – (¼)q
We can see that for a market demand q = 60, the resulting market price
p = 25 . (1/4 × 60) = 10. Of course the relation between the two variables is
still the same, but the logic is now reversed. This function now answers the
question: What is the price p that could be charged (i.e., by a company)
for a specific market demand q?
Very often it is necessary to find the inverse of the demand function
in order to calculate a company’s revenues. Similarly to calculate the
marginal revenue product function (refer to section 5 of this appendix)
the inverse of the Production function needs to be derived (subject guide
reference: Chapter 7, Production, p.100, firm demand for inputs).
211
28 Managerial economics
2. Differentiation
Students are expected to be able to differentiate functions of several
variables. Typically this is done to find the maximum value a function can
reach and to identify the corresponding value for its explaining variable(s).
For example, a company’s profit function can be maximised to find the
profit optimising output quantity (see section 9). Below is a summary
of the main rules and the derivatives of some specific functions. Partial
derivatives are found by treating the other variables as a constant.
Main rules:
Constant factor y = a.f (x)y'=a. f '(x)
Example: y = 5x 5
a = 5 and f(x) = x5
y' = 5x 5x4 = 25x 4
Formal notations:
Differentiations can be expressed in several ways, all meaning exactly the
same. For example, for a cost function C(q):
C'(q) Typically used for functions with one variable
∂C(q)/∂q Often used for partial differentiations of functions with two
variables
MC(q) MC = Marginal Cost, implies the differentiation of a total
cost function C(q)
213
28 Managerial economics
4. Integration
To calculate the area under a function, which can for instance represent
consumer surplus, students need to understand integration as the
‘opposite’ of differentiation. Integration is used to calculate the area below
a function, for instance, to find the consumer surplus. In general, integrals
between the borders a and b can be calculated as follows
b
∫
a
f(x) dx = F(b) - F(a)
Example:
Demand is given by a function D: q (p) = 100 – 2p. The consumer surplus
equals the area between the price and the demand curve. For price 40,
consumer surplus is represented by the area below the demand curve and
above p = 40. At the price of 40, the market demands a quantity q = 20.
50
214
Appendix 3: Maths checkpoints
Thus, the gain in consumer surplus equals the difference between 400 –
100 = 300. Alternatively it can directly be calculated as
40
Example:
If a firm’s marginal cost function is given as MC (q) = 10, it is clear that
marginal cost (or the incremental production cost for the next unit)
is a constant 10 for any given production unit. Thus, total cost can be
calculated as the product of 10 times the level of production, or 10q, plus
any fixed cost F which occurs regardless of the level of production. The
total cost function equals C (q) = 10q + F, which is the integrated marginal
cost function or ∫MC. And marginal cost is the differentiated total cost
function. (See for instance Chapter 11, Oligopoly, p.177).
Similarly, if MC (q) = 10 + 4q + 6q2, then ∫MC =∫(10 + 4q + 6q2) dq = C(q) =
10q + 2q2 + 2q3 + F
215
28 Managerial economics
6. Uniform distribution
The concept of uniform distribution is used in Chapter 5, Auctions and
Bidding. It describes the way in which a sample of data is distributed
within a certain interval.
Example:
What are the expected valuations of bidders in an auction, if we know the
number of bidders as well as the interval from which their valuations are
drawn?
If the valuation of n bidders is uniformly distributed on [0,1] we can make
the following predictions:
n = 1, expected valuation: ½ (middle of the interval)
n = 2, expected minimum and maximum valuations: 1/3 and 2/3
n = 3, expected minimum, middle and maximum valuations: ¼, ½ and ¾
Thus, when n > I the expectation of the valuation of the highest bidder can
be calculated as (n)/(n + 1), of the second highest as (n – 1)/(n + l), and so
on.
If the distribution is not normalised, for instance for an interval [0, 5]
the expected valuations of 4 bidders would be at equal distances on the
interval [0....v1....v2.... v3….v4….5], or 1, 2, 3, and 4 respectively. The highest
valuation can be calculated as
(n)/(n + 1) × 5 = (4)/(4 + 1) × 5 = 4
For uniformly distributed valuations between [10, 15], the expected
valuations of 4 bidders are expected to be at equal distances on the
interval
[10.... v1….v2.... v3.... v4.... 15] or 11, 12, 13, 14 respectively.
Thus, the expected highest valuation can be calculated as
10 + (n)/(n + l) × 5 = 14
In general for n bidders with uniformly distributed preferences on [L,U]
the expected highest valuation equals
L + (n)/(n + 1) . (U – L)
7. Probabilities
Probabilities are often denoted by the variable ‘p’ which represents a
number between 0 and 1. If a decision maker faces two (or more) possible
scenarios, which he cannot influence, assumptions could be made about
the likelihood of each outcome. The probabilities of all outcomes must add
up to 1 (representing 100%). Suppose one in ten bikes gets stolen each
year. Outcome 1, bike gets stolen, occurs with probability p = 0.1 (equal to
10%) and outcome 2, bike does not get stolen, occurs with probability
(1 – p) or (1 – 0.1) = 0.9 (equal to 90%).
216
Appendix 3: Maths checkpoints
Example:
One bidder wants to buy a rare painting which he values at $400,000.
He does not know what the value is to the owner, but he knows that it is
uniformly distributed between $100,000 and $500,000. The bidder can
only make one offer, which the owner can either accept or reject. What is
the optimal offer that the bidder should make?
The owner’s valuation is said to be uniformly distributed, meaning that
any point in the interval [U, L] = [100,000; 500,000] is equally likely to be
the valuation. The painting will only be sold if the submitted bid exceeds
the owner’s valuation, otherwise the owner would consider the bid to be
too low. If the bidder offers $500,000 he would definitely win the auction,
but his ‘gain’ would be a negative $100,000 as the bid would be above
his own valuation. Thus, the bidder faces a basic trade-off between the
potential gain and the likelihood of the painting being sold. He can lower
his bid, but at the same time the probability of his bid being larger than
the owner’s valuation decreases and he becomes less likely to win the
auction. The bidder’s expected gain can be calculated as the gain from
buying the painting multiplied with the probability of winning.
(v – b) . P(win)
(v – b). [(b – L) /(U – L)]
(400,000 – b) . [(b –100,000) / (500,000 – 100,000)]
To understand the transformation of the owner’s distribution to the
probability values, consider a few examples:
Probability of
Bid Gain Expected gain
winning
500,000 -100,000 1 -100,000
400,000 0 0.75 0
300,000 100,000 0.5 50,000
200,000 200,000 0.25 50,000
100,000 300,000 0 0
By interpreting the above equation as a function of the variable b, we can
calculate the optimal bid by differentiating the function with respect to b.
Expected gain (b) = (400,000 – b) . [(b – 100,000) / 400,000]
∂Expected gain / ∂b = –1[(b – 100,000)/400,000] + (400,000 – b).(1/400,000)
(product rule!)
Set equal to 0 and solve for b to find optimal bid
– 1/400,000. b +1/4 + 1 – (1/400,000 . b) = 0
5/4 = (1/200,000) . b
217
28 Managerial economics
b = 250,000
For the optimal bid of 250,000 the expected gain equals 56,250.
218
Appendix 3: Maths checkpoints
Example:
If demand function is given as D: q = 100 – 4p
Marginal cost = 12q
(i) starting with the firm’s profit function П(q) = Revenue(q) – costs(q)
Revenues: Total Revenues TR (q) = price x number of units sold, or p.q, or (25-
1/4q) .q or 25q – ¼q2
Costs: Marginal Cost MC = 12q (given), Total Cost C(q) = ∫MC = 6p2 + F
П(q) = Revenues (q) – Costs (q)
П(q) = 25q – ¼q2 (6q2+F)
П(q) = 25q – 6.25q2 – F
∂П/∂q = 0
25 – 12.5q = 0
q=2
Thus, the firm should produce 2 units, which gives it a maximum profit of
П(q) = 25q – 6.25q2 – F
П(2) = (25 × 2) – 6.25(2)2 – F
П = 50 – 25 – F
П = 25 – F
219
28 Managerial economics
The market price can be found by substituting q back into the (inverse)
demand function
p = 25 – (¼)q
p = 25 – (¼ × 2)
p = 24.5
(ii) applying the optimality condition ‘MR = MC’
MR: Derive Marginal Revenue from the Demand function
- Find the inverse demand curve as p = 25 – (¼)q
- Total Revenue (TR) = price x number of units sold, or p.q, or (25 – (¼)q) . q or
25q – (¼)q2
- Marginal Revenue (MR) = ∂TR/∂q (25-q (¼)q2) = 25 – (½)q
MC: Given as 12q
MR = MC
25 – (½)q = 12q
q=2
220
Appendix 3: Maths checkpoints
Example
Tom, Dick and Harry manage a company which markets balloons. The balloon market is
perfectly competitive and the price of balloons is $5 per unit. The amount of capital is
fixed and the investment in capital can be considered a sunk cost. The cost of material
inputs is negligible but labour is expensive. Tom has asked Dick to gather some data on
the productivity of labour. Dick presents him with the figures in the middle of the table
below. Meanwhile Tom has asked Harry to get some information about labour costs.
Harry told him that these costs would depend on how many workers were employed, as
indicated in the table below on the right. Tom now has to decide how many workers to
hire. What should he do and what will the wage rate be?
Number of Total number of
Wage rate
workers balloons per hour
0 0 10
1 10 11
2 15 12
3 19 15
4 21 17
5 22 20
6 20 25
Step 1
Read the question carefully!
Step 2
Write down the most important information.
221
28 Managerial economics
Step 3
Relate it to the studied material.
• Company is a Monopsonist facing perfect competition in its output market.
• Chapter 7, Production, pages 98-103.
Step 4
Solve the question
• Plan ahead and be certain about what you are looking for. Here we need to find the
optimal number of workers, variable ‘L’ (explicitly asked in the question)!
• Remind yourself of the analytical steps required to solve the question.
Solution Alternative 1:
Solve in Total Values, finding maximum profit
L Total Revenue (p.q) Total Expenditure (L.w) Profits II
p q Rev. L w Exp. (Rev – Exp.)
0 5 0 0 0 10 0 0
1 5 10 50 1 11 11 39
2 5 15 75 2 12 24 51
3 5 19 95 3 15 45 50
4 5 21 105 4 17 68 37
5 5 22 110 5 20 100 10
6 5 20 100 6 25 150 –50
Answer: Hiring two workers and paying them a wage of 12 gives the maximum profit
of 51.
222
Appendix 3: Maths checkpoints
Solution Alternative 2:
Solve in Marginal Values, applying MRP = ME rule (company should continue to expand
the workforce as long as the additional worker’s revenue contribution is larger than the
incremental expenditure).
0 5 0 0 0 0 0 0
Answer: Workers are hired as long as MRP is larger than ME. Thus, two workers should
be hired at wage 12. (If the third worker would be hired, MRP would be very close to ME.
But since ME exceeds MRP by exactly 1, total profits would be driven down by 1 if the
company hires worker number 3.)
Solution Alternative 3:
It is also possible to derive the actual functions from the data in the table and to proceed
analytically. Clearly this would be more complicated than the two ways suggested
above. If a table lists a limited number of possibilities, it is usually quicker to check each
alternative to find the optimal solution.
223
Notes
224
Notes
Notes
225
Notes
226