EC201 Lecture Notes Michaelmas 2014

LondonSchoolofEconomics
EC201Microeconomic
PrinciplesI
MichaelmasTerm2014
LectureNotes
DrMargaretBray
EC201 Reading
Course textbook
Snyder, C. and Nicholson, W. "Microeconomic Theory: Basic Principles
and Extensions", 11th edition, South Western, 2012.
Alternative text for Michaelmas Term material,

much less good than Snyder and Nicholson for the
Lent Term
Perlo, J.M. Microeconomics with Calculus, 3rd edition Pearson 2013,
or Microeconomics with Calculus with MyEconLab, 3rd edition Pearson
2013
or previous editions
the rst edition is Microeconomics: Theory and Applications with Calculus, Pearson 2007
Perlo is also the author of Microeconomics, Do not use this book
Other books
These are not appropriate for EC201 but may be
helpful as an introduction
EC102 text Morgan W., M. Katz and H. Rosen, "Microeconomics", 2nd
edition, McGraw Hill 2009
No calculus so unsuitable for EC201.
Varian, H, R. Intermediate Microeconomics, Norton, (any edition)
calculus in the appendices to chapters.
Varian, H.R. "Intermediate Microeconomics with Calculus" Norton,
(2014) incorporates these appendices into the body of the text
Additional reading will be provided as links from the

website
1
EC201 Michaelmas Term Syllabus

1.
1.1
Consumer theory and its applications

Preferences and utility
Maximizing models
Marshalls model of consumer demand
General utility functions
Hicks on indierence curves
Modern assumptions on preferences
Completeness, transitivity, continuity, nonsatiation and convexity
Checking the nonsatiation assumption
Checking for convexity
Can indierence curves cross?
Finding the marginal rate of substitution
Reading
Snyder & Nicholson Chapter 3
or
Perlo Chapter 3
Introductory reading below the level of EC201
Morgan, Katz & Rosen, Chapters 2.1, 2.2
or
Varian Chapters 3, 4
1.2
Utility maximization and uncompensated demand

Budget line and budget set
Denition of uncompensated demand
Tangency solutions
Finding uncompensated demand with Cobb-Douglas utility
The eects of changes in prices and income on uncompensated demand
Demand curves
Elasticity
2
Normal and inferior goods

Corner solutions
Finding uncompensated demand with quasilinear utility
Substitutes and complements
Finding uncompensated demand with perfect complements utility
Finding uncompensated demand with perfect substitutes utility
Finding uncompensated demand with nonconvex utility
Reading
or
Perlo Chapters 2.5 (elasticity) and 3
Morgan, Katz & Rosen, Chapters 2.1, 2.2, 3
or
1.3
Expenditure minimization and compensated demand

Denition of compensated demand and the expenditure function
Income and substitution eects
Properties of compensated demand
Downward sloping compensated demand curves
Finding compensated demand with Cobb-Douglas, quasilinear, perfect complements and perfect substitutes utility functions
Properties of the expenditure function
Consumer surplus with quasilinear utility
The Slutsky Equation
Reading
or
Perlo Chapters 4, 5
Morgan, Katz & Rosen, Chapter 4
or
Varian Chapter 8
3
1.4
Price changes and welfare

Price indices
Substitution bias
Compensating variation (CV) and change in consumer surplus with
quasilinear utility
Compensating variation (CV) and change in consumer surplus with
income eects
Equivalent variation (EV)
Income eects, CV, EV and consumer surplus
Using EV to assess the eect of a tax
Using EV to assess the eect of a subsidy
Does compensating a consumer for a price increase imply that the
price increase has no eect on demand?
Benets in kind
Reading
Snyder & Nicholson Chapter 5, Extensions (following chapter
19) chapter 5
or
Perlo Chapters 4.4, 5,
or
Varian Chapter 14
1.5
Labour supply, taxes and benets

The budget constraint
Income and substitution eects of an increase in the real wage
Quick overview of the UK tax system
Income tax
Benets
Reading

or
Perlo Chapters 5,
Morgan, Katz & Rosen, Chapter 5.1
or
Varian Chapter 9
1.6
Saving and borrowing

Choices between income streams
The intertemporal budget line and present discounted value
Perfect capital markets
The simplest model of asset pricing
Modelling saving and borrowing decisions
Eects an interest rate cut
Dierent borrowing and lending rates
Reading
or
Perlo Chapter 15.4,
or
Varian Chapter 10
2
2.1
Firms, costs and prot maximization

Firms and costs
The objective of the rm
Why do some economists assume prot maximization?
Prot and present discounted value
Opportunity cost
Cost of capital
Production Function
5
Marginal product and marginal rate of technical substitution

Denition and properties of the cost function
Returns to scale
Economies of Scale
Reading
Snyder & Nicholson Chapters 9, 10
or
Perlo Chapters 6, 7,
or
Varian Chapters 18, 19, 20, 21
2.2
Prot maximization and costs for a price taking rm

Denition of price taking
Shutdown and output rules, average and marginal cost.
Cost curves, prot maximization and supply with constant returns
to scale
Cost curves, prot maximization and supply with decreasing returns
to scale
Cost curves with increasing returns to scale
Price taking is impossible for a rm with economies of scale at all
levels of output
Cost curves and supply with a u-shaped average cost curve
Long run and short run costs and supply
Derivation of long run and short run costs with Cobb-Douglas and
xed proportion production functions
Fixed and variable costs
Limitations of the long run short run model
Reading
or
Perlo Chapter 3,
6

or
3
3.1
Industrial Organization
Perfect competition and monopoly
Introduction to industrial organization
Perfect competition
Markets with a xed number of price taking rms
Entry and exit
Entry & exit in an industry with a xed proportion production function
Entry & exit, short run & long run costs, with a u-shaped average
cost curve
Increasing and decreasing cost industries
Firms with dierent costs & economic rent
Very simple model of economic rent in a price taking oil industry
Welfare economics of a tax with supply and demand
Producer Surplus
Welfare economics of a subsidy with supply and demand
Deadweight losses from taxes and subsidies
Monopoly
Deadweight loss from a monopoly
Cartels
Price discrimination
Reading
or
Perlo Chapters 2, 8, 9, 11,
or
Varian Chapters 23, 24, 25
3.2
Oligopoly and games

Introduction to game theory
Prisonersdilemma
Dominant strategy equilibrium
Cournot-Nash equilibrium
Nash equilibrium
Bertrand-Nash equilibrium
Pure and mixed strategies
Multiple equilibria
Simultaneous (normal form) and sequential move (extensive form)
games
Stackelberg equilibrium
A two stage entry game
Repeated games
Reading
Snyder & Nicholson Chapters 8, 15
or
Perlo Chapters 13. 14,
Morgan, Katz & Rosen, Chapters 15, 16
or
Microeconomic Principles I
Introduction to EC201
EC201
2014-15
Lecturers: Margaret Bray (Michaelmas)
Erik Eyster (Lent)
See the website and course information

documents for details of reading
Course textbook
Snyder, C. and Nicholson, W. "Microeconomic Theory:
Basic Principles and Extensions", 11th edition, South
Western, 2012.
Other books you may find helpful

Perloff, J.M. Microeconomics with Calculus, 3rd edition
Pearson 2013,
or previous editions
First edition is Microeconomics: Theory and Applications
with Calculus, Pearson 2007
2013-14 textbook, good for Michaelmas Term, not so good
for Lent Term.
Other books you may find helpful as an

introduction to topics
EC102 year text Morgan W., M. Katz and H. Rosen,
"Microeconomics", 2nd edition, McGraw Hill 2009
No calculus so not suitable for EC201.
The objective of EC201 is to develop

your knowledge of economic models,
the analytical, logical and mathematical skills needed to
understand the models,
Varian, H.R. "Intermediate Microeconomics: A Modern

Approach", any edition.
your ability to think critically about the models and their

relationship to the world.
Short clear chapters with a calculus treatment in the

appendices, but below the level of EC 201.
Skill development requires practice over a long period.
Varian, H.R. "Intermediate Microeconomics with Calculus"

Norton, (2014) incorporates these appendices into the
body of the text.
1.1 Preferences and utility
The exam tests these skills.

Employers value these skills.
Course Outline Michaelmas Term

2 Firms, costs and profit maximization

2.1 Firms and costs
2.2 Profit maximization and costs for a price taking firm
1.2 Utility maximization and uncompensated demand

1.3 Expenditure minimization and compensated demand
1.4 Price changes and welfare
3. Industrial organization
3.1 Perfect competition and monopoly
3.2
Oligopoly and games
1.5 Labour supply, taxes and benefits

1.6 Saving and borrowing
Models and maps
Economic Models
Each model has assumptions.
Models include definitions.
A logical argument derives conclusions from the

assumptions and definitions .
A map is a model
The reality of Londons geography is very complicated.
A map is a model and like all models

gives a simplified view of reality
leaving out many things & distorting others.
Maps are useful because they simplify greatly.
It is good for some questions.

It is very misleading for other questions.
Look at the London Underground map.

Which is the best route from St Pauls to Barbican
change at Bank & Moorgate
or change at Holborn & Kings Cross?
Beware of implicit assumptions, e.g. the London

Underground map works for underground journeys but
not for walking.
1. Preferences and utility

List of topics
1. Preferences and utility

List of topics
1
Example of a maximizing models:

internship applications
Marshalls model of consumer demand
Modern assumptions on preferences

Completeness, transitivity, continuity,
nonsatiation & convexity
1.1 Example of a
maximizing model:
1.1 Example of a
maximizing model:
Models in standard microeconomics assume that consumers
and producers either maximize or minimze something.
Critics of microeconomics often argue that this is a very bad
assumption.
The relevant question is whether a maximizing model is
appropriate for a particular application.
Probability of success
Assume
You make 5 internship applications.
The probability that an application is successful is 0.5.
Make a guess, which number is closest to the probability
that at least one application is successful?
Internship applications
What is the probability of getting at least one
offer?
Probability of at least 1 successful application

= 1 - (1 r)x
Probability of success in one application = r

r = 0.3
Probability of failure in one application = 1 r

Assume that applications are independent, i.e. the probability
that an application succeeds does not depend upon whether
other applications succeed.
r = 0.1
Then probability of failure in all x applications (1 r)x.

so probability of success in at least 1 application = 1 - (1 r)x
x
If r = 0.5 and x = 5 then = 0.96875.
The students objective
More assumptions: the students objective

Assume that a student chooses the number of applications
x that maximizes N px
(x has to be an integer).
= probability of at least one successful application
N is the value of an internship (measured in )
p is the cost of an application (measured in )
the student chooses the number of
applications x that maximizes
Total value total cost
Assume N = 5000, p =200

total value total cost
r = 0.3
= V(x) px
where V(x) = N = (1 - (1 r)x )N
px = cost of x applications
r = 0.1
Assume p =200, value of internship N = 5,000

x
Any optimization problem
Marginal conditions for the

internship problem
has a solution which depends on parameters of the

value function (here N and r)
and of the cost function (here p).
Aim here is to maximize
net value = total value total cost
If N = 5,000, r = 0.1 and p =200, solution x = 9, net
value is 1263
If N = 5,000, r = 0.3 and p =200, solution x = 6, net
value is 3212
N = 5000, r = 0.3
Marginal conditions are very often helpful in

solving optimization problems.
For this internship example
total value
Marginal value of xth application

= MV(x) = V(x) V(x-1)
marginal value
= gradient of tangent line
= N{(1 (1- r)x) (1 (1-r)x-1)} = Nr(1 r)x-1

This is decreasing in x because 0 < r < 1.
Marginal cost = c.
Assume N = 5000, r = 0.1

marginal value
Marginal value MV and total value V
V(0)
MV(1)
V(1) - V(0) = V(1)
MV(2)
V(2) V(1)
MV(3)
V(3) V(2)
Add these equations

MV(1) + MV(2) + MV(3) = V(3)
x
N = 5,000
p =200
r = 0.1
Assume N = 5000, r = 0.1

marginal value and total value
entire shaded area is the total

value of 3 applications.
total cost of 3 applications
total value total cost of
3 applications
total value of 3 applications

= area under marginal value graph
marginal cost
marginal value
marginal value
x
N = 5,000
p =200
r = 0.1
N = 5,000
p =200
r = 0.1
entire shaded area is the total

value of 4 applications.
total cost of 4 applications
MV(4) = 365 > 200 so increasing x from 3 to

4 increases
total value total cost
total value total cost of

4 applications
marginal cost
marginal cost
marginal value
marginal value
x
marginal conditions
If marginal value > marginal cost
increasing x increases total value total cost.
If r = 0.1 MV > MC = 200 for x 9,

MV < MC = 200 for x > 9
x = 9 is optimal.
If marginal value < marginal cost

decreasing x increases total value total cost.
If r = 0.3 MV > MC = 200 for x 6,

MV < MC for x > 6
x = 6 is optimal.
Assume N = 5000, r = 0.1

marginal value
Marginal conditions with integers

If marginal value MV is decreasing
If p =200 9
If p =300 6
If p =400 3
marginal cost c is constant
is optimal
is optimal
is optimal
then x is optimal if MV(x) c MV(x+1)

So in this model x is optimal if
Nr(1 - r)x-1 c Nr(1 - r)x-1
or equivalently r(1 - r)x-1 c/N r(1 - r)x.
x
TP
N = 5,000, p = 200, r = 0.1
Definition of surplus
If p = 200 x = 9 is optimal
The optimal value of x maximizes
Entire shaded area

is total value
total cost
net value = total value total cost

Surplus is total value total cost
when x is at the optimal level.
marginal cost
marginal value
x
Probability of at least 1 successful application

= 1 - (1 r)x
Marginal probability r(1-r)x-1

Low x, the main effect comes through r
r = 0.3
higher r gives higher marginal probability

r = 0.1
High x, main effect comes through (1-r)x-1

lower r gives higher marginal probability
Marginal probability r(1-r)x-1
Predictions
If the cost value ratio c/N is high enough applicants with
a low probability of success r make zero applications.
r = 0.3
high c/N
If the cost value ratio c/N is low applicants with a higher

probability of success r make more applications.
If the cost value ratio c/N is very low applicants with a
higher probability of success r make fewer applications.
low c/N
r = 0.1
very low c/N
1.3 Maximizing models:

are they useful?
Where high and low are depends on the parameters r

and c/N.
Maximizing models: are they useful?

Standard critique: microeconomic models assume that
people make lightning calculations which they dont.
Friedman response: as if model. Expert billiard players
do not know the maths describing the laws of physics but
act as if they did.
Milton Friedman Essays in positive

Economics (1953)
How do expert billiard players play?
. not at all unreasonable that excellent
predictions would be yielded by the
hypothesis that the billiard player made his
shots as if he knew the complicated
mathematical formulae that would give the
optimal directions of travel .
Response to Friedman: but .

Billiard players are taught the consequences of the laws
of physics even thought they cannot solve the
mathematical equations. (See YouTube)
D. Kahneman Thinking Fast and Slow Penguin, 2011
Expert intuition is good in
an environment that is sufficiently regular to be
predictable
with an opportunity to learn these regularities
through prolonged practice
immediate and unambiguous feedback
Billiards is like this. Many economic situations are not.
Mathematical models in economics and

biology: the brothers
Niko Tinbergen: 1907 88 Nobel Prize: physiology or
medicine 1973
Quantitative models of animal behaviour based on
very detailed observation
Jan Tinbergen: 1903 94 Nobel Prize: economics 1969
Quantitative models in economics (econometrics)
Behavioural Ecology:
optimizing models in biology
Applied to many species, e.g. birds, bees and lions.
Objective is reproductive success
Evolution thought natural selection + genes
Tradeoffs
Game theory
Experiments
Evolutionary psychology
Big data
Using evolutionary ideas to explain human behaviour,

e.g. food, sex, childcare
Now cheaper and easier to collect and analyze data than

ever before.
Argument
Easy to experiment, e.g. on look of website.
Our species emerged approximately 65,000 years

ago
Domestication of plants and animals approximately
12,000 years ago
We are not adapted to the world we live in
Biology, culture and environment all affect human behaviour.
The internship model again

We assumed that the objective is success in internship
applications this year.
The scarce resource is money.
The internship model again: suppose

you care about your academic work as well as your
internship because for example you care about
what you get out of the course
exam results for their own sake
exam results because they affect your chances of a
job after graduating.
The scarce resource is time
Cannot model this at this stage but will do so later.
1.4 Marshalls model of

consumer demand
Late 19th century orthodoxy Marshall.

Principles of Economics, 1890
Mid 20th century orthodoxy Hicks,
Value & Capital (1939)
builds on Edgeworth (1881), Pareto (1909), Slutsky (1915)
The modern textbook theory .
The purpose of consumer theory
Alfred Marshall (1842-1924)
to understand how household behaviour

buying, selling, working, saving and borrowing
is affected by changes in
prices, income, wages, interest rates, tax rates ..
to provide a way of
showing that there are situations in which everyone
gains from trade
measuring the effect on household welfare of
changes in prices, income etc..
1. Marshalls model of consumer demand

roots in 19th century utilitarian philosophy
achievements
the demand curve
consumer surplus
Cambridge
Principles of Economics (1890)
supply and demand curves
elasticity
consumer surplus
Marshalls theory of
consumer demand with
integer quantities,
1, 2, 3
gaps
lots of words, no modelling of
the effects of income & prices of other goods
10
Marshalls assumptions on consumer

behaviour
Marshalls assumptions on consumer

behaviour
Marginal value MV(x) = V(x) V(x-1).
The consumer chooses quantity x to maximize V(x) px.
maximum willing to pay for one more unit.
The price p does not depend on x

V(x) is measured in units of money, , V(0) = 0.
Marginal value is positive.
V(x) is the most a consumer would be willing to pay for x units

if the alternative were 0 units.
Marginal value decreases with the number of units.

Consumers maximize V(x) px (p is price).
Marshall calls V(x) utility
Maths is exactly like the internship model.
I call V(x) value to avoid confusion with the modern

interpretation of utility.
Total value:
marginal value = gradient of total value graph
Marginal conditions with integers

Objective is to max V(x) px
MV(x) = V(x+1) V(x)
Total value
If MV(x) > p increasing x to x + 1 increases V(x) px

Total value is close
to a smooth curve.
Marginal value is its
derivative.
If MV(x) < p increasing x to x + 1 decreases V(x) px

If MV(x) is decreasing then x is optimal if
MV(x) p MV(x+1)
number of teabags
Marshalls Theory applied to the Bray family

demand for teabags
per
tea
bag
The marginal value curve is

the demand curve.
Demand = 6 when p =0.5
number of teabags
Marshalls Theory applied to the Bray family

demand for teabags
per
tea
bag
Shaded area
total value of 6 teabags,
area under the
marginal value curve i.e.
the integral.
number of teabags
11
Consumer surplus in Marshalls model
Consumer surplus in
Marshalls model with
integer quantities
Marshall defines consumer surplus as

The most the consumer would be willing to pay
- amount the consumer pays
= V(x) px.
Consumer surplus in Marshalls model is exactly the same
as surplus in the internship model.
N = 5,000, p = 200, r = 0.1

Internship model
Consumer Surplus
Entire shaded area

is total value
Entire shaded area

is total value
total cost
per
tea
bag
total cost
Consumer
surplus
marginal cost
marginal value
x
number of teabags
Marshall and calculus
Calculus revision
If units of quantity are small enough

The total value and demand curves become smooth.
Calculus can be used to model the situation.
This is very convenient.
12
Going further needs calculus

Quick calculus test
This function has a

maximum
at 0.
But it has a
local max & local min.
At these points the derivative
is 0.
0
x
local min
Is the first derivative positive or negative?

0
Is the first derivative increasing or decreasing?
TP
y
y
This
function
has no
maximum.
0
x
Is this function concave or convex?
i.e. the line joining any two points on the graph is below the
graph.
Concave functions have decreasing first derivatives
so concave functions have negative second derivatives
If a function is concave a point where the

derivative is 0 maximizes the function.
y is a concave function of x
so
dy
is decreasing
dx
Marshalls Theory with

calculus and concavity
dy is 0 at x = 1 and is
dx
> 0 when x < 1 where the function is increasing

< 0 when x > 1 where the function is decreasing.
So the function has a max at x = 1
13
Marshalls theory with calculus and

concavity: assumptions
Total value is an increasing function of x so dV > 0
dx
Total value is a concave function of x so
dV
0
dx 2

concavity
Total value is a concave function of x implies that
total value total cost = V(x) px is concave.
Why?
as V(x) is concave the second derivative
Consumers maximize total value total cost = V(x) - px
The second derivative of V(x) px is also

so V(x) px is concave.

concavity
As V(x) px is concave the first order condition
dV
dx
d 2V
0
dx 2
2
dV
0
dx 2
Marshalls Theory with Calculus

xA = demand if p = pA
p
because at this point

p = MV
= p
pA
marginal value = price
Marginal value
= demand
Maximizes total value total cost

gives the maximum.
0
xA

xA = demand if p = pA
p
pA
Total
value
if
p = pA
Consumer surplus in
Marshalls model with
calculus and concavity
Marginal value
= demand
xA
14
Consumer surplus in Marshalls model

Marshall defines consumer surplus as
Marshall defines
consumer surplus in
his model as
total value
- total cost
The most the consumer would be willing to pay

- amount the consumer pays
total
value
total cost
= V(x) px.
pA
Marginal value
= demand
total cost
N = 5,000, p = 200, r = 0.1
Internship model
Measuring consumer
surplus is difficult
It requires knowing the
whole demand curve.
total cost
marginal cost
pA
Measuring the fall in

consumer surplus
when p rises from pA
to pB requires knowing
demand at pA and pB
and the elasticity of
demand.
pB
consumer
surplus
Marginal value
= demand
total cost
marginal value
Entire shaded area

is total value
xA
xA
Diamond water paradox

Water is essential, diamonds are not
But diamonds have a much lower price than
water.
p
pA
total cost
xA
High consumer
surplus low price,
e.g. water
low consumer
surplus high price,
e.g. diamonds
15
Water, high total value,

low price, low marginal value,
large CS.
Marshall and Hicks on

utility and indifference
curves
Diamonds, low total value,

high price, high marginal value,
small CS.
Marshalls theory explains why an

essential good (water) has a much lower
price than a luxury (diamonds).
What has Marshall achieved?

1. Explained the downward sloping demand curve.
2. Demonstrated that consumers gain from trade.
3. Giving a monetary measure of the gain from trade:
consumer surplus but this requires knowing the whole
demand curve.
What are the difficulties with Marshalls

model ?
It does not distinguish between
changes in demand due to changes in the price of
other goods & income
changes in demand due to changes in tastes, e.g.
fashion.
4. Giving a monetary measure of the loss to a consumer

caused by a price increase: loss of consumer surplus.
It does not model the effects of changes in other prices &

income.
5. Explained why a good may have a low price but high

total value, price = marginal value, not total value.
What is utility and can it be measured in money?
General utility functions
This is a big 19th century philosophical debate.
Implications of the utility function

1: completeness
Marshall has no model of how the utilities of two goods

interact.
Because utility is a number for any two bundles of goods (x1A,x2A)

and (x1B,x2B) one of these relationships must hold:
Move to utility depending on quantities of all goods,
u(x1A,x2A) > u(x1B,x2B) the consumer prefers (x1A,x2A) to (x1B,x2B)
Consuming x1, x2 .xn of goods 1,2n gives utility
u(x1A,x2A) = u(x1B,x2B) the consumer is indifferent between(x1A,x2A)

& (x1B,x2B).
u(x1, x2 .xn) which is a number. This term n = 2.

Note the implicit assumption, utility depends on the goods
you consume, can be extended to other peoples
consumption
u(x1A,x2A) < u(x1B,x2B) the consumer prefers (x1B,x2B) to (x1A,x2A)
There is no mention of human relationships in this theory.
16

2: transitivity

3: indifference curves
Because utility is a number for any three bundles of goods (x1A,x2A)

and (x1B,x2B) and (x1C,x2C)
All the points on an indifference

curve have the same utility
If u(x1A,x2A) > u(x1B,x2B) and u(x1B,x2B) > u(x1C,x2C)
so consumers are indifferent

between two points on the same
indifference curve.
then u(x1A,x2A) > u(x1C,x2C).
There is an indifference curve

through every point but we only
show some of them.
x2
u(x1, x2) = 3
u(x1, x2) = 2
u(x1, x2) = 1
x1
Without further assumptions

indifference curves can have a
strange shape.
Hicks: Value & Capital
3. Hicks on indifference curves:

Paretos theory
2nd edition, OUP 1939

Building on
In order to determine the quantities of goods which an

individual will buy at give prices,
Pareto (1909)
Edgeworth (1891)
Marshalls theory implies that we must know his utility

surface; (i.e. utility function);
Slutsky (1915)
Hicks worked with indifference curves
John Hicks,
1904 89
And that conveys less information than the utility surface.
LSE 1926 to 1935
It only tells us that that the individual prefers one particular

collection of goods to another ..; it does not tell us by
how much the first collection is preferred to the second.
Nobel prize 1972

Source picture LSE website
Indifference Curves & Ordinal Utility
Indifference curves
Does replacing 1, 2, 3 by 1, 4,
9 change preferences?
x2
u(x1, x2) = 3
u(x1, x2) = 2
Does replacing 1, 2, 3 by
1, 30, 14 change preferences?
u(x1, x2) = 1
Paretos theory only assumes that we must know his

indifference map.
x1
Indifference curves show whether or not the consumer

prefers one bundle of goods to another.
They do not tell us by how much one bundle is preferred
to another.
Different utility functions can generate the same
indifference curves.
Economists jargon, utility is ordinal not cardinal.
TP
17
Hicks made two assumptions on the shape

of indifference curve
Nonsatiation: more is better
Nonsatiation, more is better.
Consumers prefer more to less.
x2
A
B is preferred to A because
there is more good 1 and the
same amount of good 2.
Decreasing marginal rate of substitution

C
0
x1
To get to C on the same

indifference curve as A good 2
must be reduced.
Nonsatiation implies downward sloping

Hicks introduced the term Marginal Rate of

Substitution, MRS
x2
A
B
C
x1
BC
AB
= - slope of AC
Hicks assumed decreasing MRS
MRS = - gradient of indifference

curve
x2
= average amount of
good 2 needed to
compensate for loosing
1 unit of good 1.
0
Summary
Marshall started his analysis with a utility functions.
Hicks assumes that the MRS

decreases along an indifference
curve as x1 increases.
x1
With more x1 the consumer

requires less extra x2 to make up
for loosing a unit of x1.
1.5 Modern assumptions

on preferences:
Hicks dropped utility functions and started his analysis

with indifference curves.
The modern theory starts with preferences
18
4. Modern assumptions on preferences

1. Completeness
2. Transitivity
Modern assumptions on
preferences:
completeness
For each of these

assumption I give
3. Continuity
a definition
4. Nonsatiation
an intuitive explanation
5. Convexity
a critical discussion.
Modern Assumption 1: Completeness

If a consumer is choosing between two
bundles A and B one of the following
possibilities holds
x2
Choices over Chinese vegetables
.B
.A
she prefers A to B
si-gua
jay-lan
she prefers B to A
0
she is indifferent between A and B.
x1
mao-gua
This term we work with 2 goods so consumption bundles

can be shown in a diagram.
TP
Consumers may find it impossible to rank some options, or

do so in a way that is in some sense wrong for them.
Decision making takes time and effort, we are often
ignorant
uncertain
preferences:
transitivity
unable to understand what a product is and does

subject to systematic biases
This standard model is not based on psychological
research.
Behavioural economics is based on psychological
research.
19
Modern Assumption 2: Transitivity
World Cup 2014
Transitivity if A is preferred to B and B to C then A is

preferred to C.
England
Check that you understand what transitivity means by

clicking the correct answers.
Costa Rica draw
Costa Rica
beat
Italy
Italy
beat
England
1. The relationship A is taller than B is transitive.

Transitivity would imply that England beat Italy.
2. The relationship team A beat team B last time they

played is transitive.
The relationship beat in sport is not transitive.
TP
Group decision making, e.g. voting
preferences:
continuity
May not result in transitive decisions

Next term
Modern Assumption 3: Continuity
Summary: Modern theory starts with a

preference relation and assumes
x2
If A is preferred to B
& B is very close to C
1. Completeness, if A and B are two bundles of goods

then
.B
.C
.A
either A is preferred to B,
or A and B are indifferent
or B is preferred to A
then A is preferred to C.
x1
This can be stated much more precisely but not for EC201.
2. Transitivity if A is preferred to B and B to C then A is

preferred to C
3. Continuity if A is preferred to B and C is close to B then
A is preferred to C.
20
The completeness, transitivity and continuity

assumptions imply
Consumer theory:
ordinal utility
preferences can be represented by a utility function and

Take this on trust. Proving it requires maths beyond the
scope of EC201.
Indifference Curves & Ordinal Utility
Indifference curves
Replacing 1, 2, 3 by 1, 4, 9,
does not change preferences.
x2
u(x1, x2) = 3
u(x1, x2) = 2
Replacing 1, 2, 3 by
1, 30, 14 does change
u(x1, x2) = 1
x1
Indifference curves show whether or not the consumer

prefers one bundle of goods to another.
They do not tell us by how much one bundle is preferred
to another.
Different utility functions can generate the same
preferences
Economists jargon, utility is ordinal not cardinal.
Different utility functions representing the

same preferences

same preferences
Suppose utility u(x1,x2) = x13/4 x21/2
Suppose utility u(x1,x2) = x13/4 x21/2
If x1 > 0 and x2 > 0 u > 0
If x1 > 0 and x2 > 0 u > 0
Then as u2 is an increasing function of u (for u>0)
Then as ln u is an increasing function of u (for u>0)
u2
if v = then
v(x1,x2) =(u(x1,x2) )2 = x13/2 x2 represents the same
preferences as u(x1,x2)
This is because the order of numbers attached to
indifference curves does not change.
if w = ln u then
w(x1,x2) = ln(u(x1,x2) ) = (3/4) ln x1 + (1/2) ln x2
represents the same preferences as u(x1,x2)
21

same preferences

same preferences
Suppose utility u(x1,x2) = min [2x1, 3x2]
If f(u) is an increasing function of u
If x1 > 0 and x2 > 0 u > 0
Then f(u(x)) represents the same preferences as u(x).
Then as 2u is an increasing function of u
Utility is ordinal not cardinal.
if w = 2 u then
z(x1,x2) = 2 min [2x1, 3x2] = min[4x1, 6x2]
represents the same preferences as u(x1,x2)
preferences:
nonsatiation
Modern Assumption 4: Nonsatiation

(More is better)
General idea: more is better
A consumer prefers having more of each good to having less.
If A = (x1A,x2A), B = (x1B, x2B),
If x1A > x1B and x2A x2B then A is preferred to B.
If x1A x1B and x2A > x2B then A is preferred to B.
Nonsatiation in the indifference curve

diagram
x2
Nonsatiation means that any

point such as A inside or on the
boundary of the shaded area is
preferred to B.
.A
B
x1
Nonsatiation in Snyder & Nicholson

Section on economic goods
the variables are taken to be goods, i.e. whatever
quantities they represent we assume that more of any
particular good . is preferred to less.
Here starting from B increasing

x1 and/or increasing x2
increases utility.
Nonsatiation implies indifference curves slope

downwards
22
The Queen
Bill Gates
"It is quite clear that, far from being 'the richest
Bill Gates net worth in 1999: $100 billion.
woman in the world', the Queen is not,
Since1999 Gates gave more than 52% of his
in terms of disposable assets,

even the richest person in the UK.
wealth to charity.
Bill and Melinda Gates Foundation mostly funds
Her 'private wealth'

would form only a fraction of that
research on illnesses in developing
disclosed recently in the courts
countries. (AIDS, TB, malaria).
as the collective assets of the Beatles,"
$2.8billion dollars a year.

Source theage.com.au 2 January 2003
Quoting UK govt document, written 1972 disclosed 2003
5. Checking the nonsatiation assumption
The Beatles
If
Ill buy you a diamond ring my friend,
If it makes you feel alright
u
x 1
0,
u
x 2
then increasing one or both of of x1 and x2 increases utility.
Ill give you anything my friend,
The indifference curve slopes downwards.
If it makes you feel alright
The preferred set is above the indifference curve.
Cause I dont care too much for money
Nonsatiation is satisfied.
5. Checking the nonsatiation assumption

If
u
x 1
0,
u
x 2
Nonsatiation with perfect complements
then the nonsatiation assumption is satisfied
Perfect complements utility functions e.g. min[x1,2x2]
e.g. if u = x12/5x23/5
do not have partial derivatives points where x1 = 2x2
See Consumer Theory Worked

Example 1 on the website:
Checking the nonsatiation and
3 / 5
1
u
x 1
u
x 2
53 x 12 / 5 x 2 2 / 5 0
2
5
convexity assumptions.
3/ 5
2
0,
Cant use calculus to check for nonsatiation.
Discussion when deriving demand functions with perfect

complements utility.
so the nonsatiation assumption is satisfied.
23
Is nonsatiation satisfied here?

diagram
x2
Nonsatiation implies that points

above an indifference curve are
preferred to points on the
indifference curve.
preferred
set
x2
preferred set
These points are in the

preferred set.
0
x1
Nonsatiation implies downward sloping

preferences:
convexity
x1
TP
Modern Assumption 5: Convexity

The preferred set is convex.
Mathematically a set is convex if any straight line joining
two points in the set lies in the set.
Which of these sets are convex?
D
TP

Interpreting the convexity assumption

A and C are indifferent.
x2
x2
preferred
B set
preferred
set
C
0
B is the
x1
of A and C
Convexity implies that B is preferred to

0
x1
Convexity means preference
24
Do these preferences satisfy convexity

and nonsatiation?
Nonconvex preferences with nonsatiation

x2
preferred
Figure 1
set
x2
Figure 2
preferred
x2
x2
preferred
set
set
0
Figure 3
preferred
x1
set
set
x1
tea & coffee consumed simultaneously
convexity
convexity
convexity
nonsatiation
nonsatiation
nonsatiation
alcohol & drugs
x1
possible examples of nonconvex preferences
x1
Getty Images
TP
6. Checking for convexity
x2
convex functions
A function is convex if a
x2
Is this function convex?
A
B
straight line joining two

points on its graph lies
above the graph.

0

preferred
set
x1
Is this function convex?
x1
x2
x1
x2
TP
x1
Functions with a positive second

derivative are convex.
x2
Functions with a
increasing first derivative
are convex.
If nonsatiation
holds the
convexity
assumption is
equivalent to
assuming that the
indifference curve
is the graph of a
convex function.
Functions with a positive

second derivative have an
increasing first derivative
so functions with a
positive second derivative
are convex.
0
x1
25

Hicks assumed a decreasing

Marginal Rate of Substitution (MRS)
x2
MRS = - gradient of indifference

curve
x2
With 2 goods the

convexity and
decreasing
marginal rate of
substitution
assumptions are
the same.
preferred
set
Hicks assumes that the MRS

decreases along an indifference
curve as x1 increases
x1
so the derivative increases

implying that the graph of
the indifference curve is a convex
function.
How to check for convexity
x1
Example of checking for convexity

Use the formula for the utility function to x2 as a function

of x1 & u.
Find the second derivative of this function with respect to
x1.
If the second derivative is positive the convexity
assumption is satisfied.
If
u=
x12/5x23/5
then
Example 1 on the website:

Checking the nonsatiation and
x2 = u5/3 x1-2/3 so
x2 = - (2/3)
x1
u5/3 x1-5/3
convexity assumptions.
and
2x2 = (10/9) u5/3 x1-8/3 > 0

x12
so the indifference curve is convex.
Convexity with perfect complements
Now look at the implications of the

assumptions on:
Crossing of indifference curves
Do not have partial derivatives points where x1 = 2x2
Finding the Marginal Rate of Substitution (MRS)
Cant use calculus to check for convexity.
Using the MRS to see if two utility functions represent

the same preferences

Comparison of two peoples utility.
26
7. Can indifference curves cross?
Can indifference curves

cross?
A and B are on the same

indifference curve so are
x2
.B
.A u(x1,x2) = 6
B and C are on the same
indifference curve so are
Transitivity
implies
indifference
curves
Transitivity implies that A and C

are.
But A has
u(x1,x2) =4
utility than C.
x1
8. Finding the marginal rate of substitution
Marginal Rate of
Substitution (MRS)
Definition and formula

Explaining the formula
Example: Cobb-Douglas utility function
Using the MRS to find whether two utility functions
represent the same preferences
Convexity and MRS
Marginal Rate of Substitution: Definition
Marginal Rate of Substitution: Formula

u
x1
u
x2
Hicks and most textbooks defines

the MRS as
- gradient of indifference curve.
x2
A
Perloff defines the MRS as

gradient of indifference curve
See consumer theory

worked example 2,
marginal rate of
substitution & ordinal utility
x2
See Consumer Theory
Worked Example 2: MRS
x1
and ordinal utility
x1
27
Explaining the formula for MRS
If x2 stays at x2A
If x1 stays at x1B
x1 changes from x1A to x1B
x2 changes from x2B to x2A
x2
u changes from u0 to u1
Then
(x1B - x1A)
so
u0
x2A
(u1 u0) u
u1
x1
u1 u0 (x1B - x1A) u
Then
x2B
(u1 u0) u
x2A
(x2A - x2B)
so
0 x1B
x1
x2
u changes from u0 to u1
x1A
x1
u0
u1
x2
u1 u0 (x2A - x2B) u
0 x1B
x2
x1A
x1
From the previous slides

(u1 u0) (x1B - x1A) u
x2
x1
x2B
(x2A - x2B) u
u0
x2A
x2
x2B
B
u0
x2A
u1
u
0 x1B
x1
(x1B - x1A)
x2
u1
So slope of line AB =
(x2B - x2A)
When A and B are close the

slope of AB is close to the
derivative of the indifference
curve, - MRS
x1A
x1
0 x1B
x1A
x1
u
x2
MRS example: Cobb-Douglas utility

function
MRS perfect complements

utility u(x1,x2) = u = x12/5x23/5

MRS
u
x 1
u
x 2
do not have an MRS at points where x1 = 2x2

Cant use calculus to check for convexity.
2 3 / 5 3 / 5
x1 x 2
5
3 2 / 5 2 / 5
x1 x 2
5
2x 2

3x 1

28
Using the MRS to find whether two utility

functions represent the same preferences
If f is a strictly increasing function and (x1,x2) = f(u(x1,x2)) then
u and represent the same preferences.
Using the MRS to find whether two utility

functions represent the same preferences
MRS from utility function u(x1,x2)
u
x1
u
x2
What if you have two utility functions and cannot see whether
they represent the same preferences?
Same preferences imply same indifference curves imply same
MRS.
Check whether utility functions they have the same MRS to
see whether they represent the same preferences..
See Consumer Theory Worked Example 2 on the website: Marginal rate
of substitution and ordinal utility.
Using the chain rule MRS from utility function

(x1,x2) = f(u(x1,x2))
x1
x2
df u
du x1 =
df u
du x2
u
x1
u
x2
Can you compare different peoples utility?
Questions on preferences
and utility
TP
Regret
How far does our welfare depend on
Have you ever chosen to do or consume something and

then regretted it?
Choice under uncertainty?
Inability to commit yourself?
29
Questions on preferences
Where we have got to?
Are people rational?
What is the effect of other consumers on your choices?
We have thought critically about the assumptions

(completeness, transitivity, continuity, nonsatiation &
convexity)
What is the relationship between utility, welfare and

happiness?
we have got ordinal utility & indifference curves from the

assumptions
What is the effect of what we buy and sell on our quality

of life?
We can use calculus to
Who is the consumer?
Should we take psychology seriously?

Is the simple model useful?
check for nonsatiation & convexity

find the MRS
see if two different utility functions represent the same
preferences.
30
1.2 Utility maximization and uncompensated

demand
1.2 Utility maximization

and uncompensated
demand
1. Budget line and budget set

2.Definition of uncompensated demand
3.Tangency and corner solutions
4.Finding uncompensated demand with Cobb-Douglas
utility
5.The effects of changes in prices and income on
uncompensated demand
The budget set and

budget line
6. Demand curves
7. Substitutes and complements
7. Finding uncompensated demand with perfect
complements utility
8. Finding uncompensated demand with perfect
substitutes utility
9. Finding uncompensated demand with nonconvex
utility
1. The budget set and budget line

x2
What is the gradient of the budget line?
Assume that it is impossible to

consume negative quantities.
budget line
Budget set points with x1 0, x2 0

p1x1 + p2x2 m.
budget
set
0
Budget line p1x1 + p2x2 = m
x1
Rearranging gives p2x2 = -p1x1 + m
Notation
Varian Snyder & N Snyder & N

lectures
2 goods
n goods
quantities
x1
prices
p1 p2
px py
pi
p1 p2
income
x2
xi
budget line p1x1 + p2x2 = m
Perloff
so
q1
x2 = -(p1/p2) x1 + (m/p2)
q2
Definition of
uncompensated
demand
Where does the budget line meet the axes?

x2
budget line p1x1 + p2x = m
gradient p1/p2.
x1
2. Definition of uncompensated
demand functions
Definition:
To get uncompensated demand fix

income and prices which fixes the
budget line.
x2
The consumers demand functions
Get onto highest possible

indifference curve.
x1(p1,p2,m) and x2(p1,p2,m) maximize utility u(x1,x2)

subject to the budget constraint p1 x1 + p2 x2 m
x1
and non negativity constraints x1 0 x2 0.
Later we call this uncompensated demand.

Some books use the term Marshallian demand.
Examples of utility
maximization
(uncompensated demand)
Examples of utility maximization

1. Cobb-Douglas utility
2. Quasilinear utility
3. Perfect complements
4. Perfect substitutes
Examples of utility maximization

8 steps for finding

For each example we will look at
Indifference curve diagram
Effect of prices on demand, own price and cross price

elasticities
Effect of income on demand, normal and inferior goods,

income elasticity
Demand curve diagram
8 steps for finding uncompensated demand

5. Draw a diagram based on the tangency and budget line
conditions.
1. Write down the problem you are solving

2. What is the solution a function of?
3. Check for nonsatiation and convexity using calculus if
the utility function has partial derivatives
6. Remind yourself what you are finding and what it

depends on.
7. Write down the equations to be solved.
Explain their implications.

8. Solve the equations and write down the solution as a
function. If at this point x1 0 and x2 0 you have
found the utility maximizing point.
4. Use the tangency and budget line conditions.
Why check for nonsatiation and convexity?

If they are not satisfied there can be
a tangency point A where
MRS = price ratio
that does not solve the problem.
x2
A point like A that is not a tangency cannot maximize utility.

The point B with x1 > 0 and x2 > 0 maximises utility.
It must be a tangency point.
Here nonsatiation fails.

preferred
set
The tangency is at A
but B maximizes utility.
x2
The point C is a tangency

point but does not
maximize utility.
B.
AA
x1
B
x1
Here
convexity
fails

Here convexity fails.
Logic of first order conditions
Very important.
The tangency is A but B maximizes utility.

x2
x2
If the nonsatiation and convexity

conditions are satisfied then any
tangency point at which
preferred
set
preferred
set
MRS = price ratio

x1 0, x2 0
A
budget
solves the utility maximizing

problem.
set
0
x1
x1
Finding a tangency solution

The gradient of the indifference curve is MRS.
We have already found that
The gradient of the budget line is p1/p2.
u
x1
MRS
u
x2
If MRS = p1/p2 the point is tangent to some budget line

with gradient p1/p2.
x2
If in addition p1x1 + p2x2 = m the

point is on the budget line with
income m.
u
p
x1
1
u
p2
x2
x1
Another way to look at the tangency

condition
Get x1 more units of x1
u
x1
x1
increase in utility
so MRS = price ratio requires that
Spend 1 more on x1 gives 1 / p1 more units of good 1

increase in utility
1 u
p1 x1
Spend 1 less on x2
fall in utility
Spending 1 more on x1 and 1 less on good 2

increases utility if
1 u
1 u
p 1 x1 p2 x2
Spending 1 less on good 1 and 1 more on good 2

increases utility if
1 u
1 u
p 2 x2 p1 x1
Utility maximization requires
1 u
p2 x2
1 u
1 u
p 1 x1 p2 x2
Utility maximization requires 1 u 1 u
p 1 x1
or
p2 x2
u
p
x1
1
u
p2
x2
Finding uncompensated
demand with
Cobb-Douglas utility
i.e. MRS = price ratio

5. Draw a diagram based on the tangency and budget line
conditions.
1. Write down the problem you are solving

2. What is the solution a function of?
3. Check for nonsatiationand convexity using calculus if
the utility function has partial derivatives
6. Remind yourself what you are finding and what it

depends on.
7. Write down the equations to be solved.
Explain their implications.

4. Use the tangency and budget line conditions.
4. Finding uncompensated demand with

Step 1: What problem are you solving?
The problem is maximizing utility u(x1,x2) =
x12/5x23/5
8. Solve the equations and write down the solution as a

function. If at this point x1 0 and x2 0 you have
found the utility maximizing point.
Finding uncompensated demand with

Step 3: Check for nonsatiation and convexity
We have already done this, both are satisfied.
subject to non-negativity constraints x1 0 x2 0

and the budget constraint p1x1 + p2x2 m .
Step 2: What is the solution a function of?

Demand is a function of prices and income so is
x1(p1,p2,m)
x2(p1,p2,m)

Example 3

Step 4: Use the tangency and budget line
conditions
Easy to lose exam marks

Failing to say that
Because nonsatiation and convexity are satisfied any point
on the budget line at which
MRS = price ratio, x1 0 and x2 0
solves the solves the utility maximizing problem.
Because convexity and nonsatiation are satisfied any point with

p1x1 + p2x2 = m
so it is on the budget line
and
MRS =
here we have already found

MRS =

2x2 = p1
3x1
p2
2 3 / 5 3 / 5
x1 x 2
5
3 2 / 5 2 / 5
x1 x 2
5
2x 2
3x 1

You are finding demand x1 and x2 which is a

function of p1, p2 and m.
tangency condition
MRS =
u
x 1
u
x 2
solves the problem
Step 6: Remind yourself what you are

finding and what it depends on.
Step 5: Draw a diagram based on the

tangency and budget line conditions
x2
p1
Step 7: Write down the equations to be

solved.
p2
The equations are p1x1 + p2x2 = m and

2x2 = p1
p1x1 + p2x2 = m budget line

0
3x1
x1

Step 8 solve the equations and write down
the solution as a function.
(You will do some algebra here.)
Solving the equations simultaneously for x1 and x2 gives
(uncompensated) demand which is a function of p1, p2, m.
x1(p1,p2,m) = 2
x 2(p1,p2,m) = 3 m
p1
5 p2
because the conditions x1 0, x2 0 are satisfied.
p2
An alternative approach: using Lagrangians

You can also use Lagrangians to find demand.
See Consumer Theory Worked Example 3 and note the
explanation that is expected, e.g. checking for convexity.
With two goods, the Lagrangian is not essential. You can
base your analysis on graphs.
This term EC201 works with 2 goods, you do not need
Lagrangians.
With more than 2 goods you have to use Lagrangians.
Mathematical definition of homogeneous functions

A function f(z1,z2,z3..zn) is homogeneous of degree 0 if for all
numbers k > 0
Homogeneity of
f(kz1,kz2,kz3..kzn) = k0 f(z1,z2,z3..zn) = f(z1,z2,z3..zn).

Multiplying z1,z2,..zn by k > 0 does not change the value of f.
A function f(z1,z2,z3..zn) is homogeneous of degree one if

for all numbers k > 0
f(kz1,kz2,kz3..kzn) = k1 f(z1,z2,z3..zn) = kf(z1,z2,z3..zn)
Multiplying z1, z2 zn multiplies the value of f by k.
All prices and income are multiplied by k > 0.

x2
p x + p2x2 = m
m/p2 1 1
gradient
p1/p2
budget
How does the budget line change?

How does demand change?
set
All uncompensated demand functions are

homogeneous of degree 0 in prices and income
This means that if all prices and income are all
multiplied by k > 0 demand does not change.
With Cobb-Douglas utility u(x1,x2) = x12/5x23/5
m/p1
x1
The consumers demand functions x1(p1,p2,m) and x2(p1,p2,m)

are
in prices and income.
x 2(p1,p2,m) = 3 m
p1
5 p2
The values of these functions do not change when p1, p2

and m are all multiplied by k > 0.
That is if k > 0
x1(kp1,kp2,km) =
x1(p1,p2,m) = 2
x2(kp1,kp2,km) =
Easy to lose exam marks

EC201 always requires explanations
Asked what happens to uncompensated demand when
prices and income are all multiplied by 2
Saying nothing happens because uncompensated demand
is homogeneous of degree 0 in prices
Changes in demand and

demand curves
This is a statement of what happens it is not an

explanation.
TP
p1 increases, p2 and m do not change. What

happens to demand for goods 1 and 2?

x1(p1,p2,m) = 2
x 2(p1,p2,m) = 3 m
p1
5 p2

x1(p1,p2,m) = 2
x 2(p1,p2,m) = 3 m
p1
5 p2
Demand for good 2 is not

affected by the price of
good 1.
x2
price consumption curve
TP
x1
6. Demand curves
Demand for good 1

If p1 increases from p1A to p1B there is a
movement
the demand curve,
demand
Demand
p1
x1(p1,p2,m) = 2
p1
p1
To draw the demand curve with p1 on the vertical

axis rearrange this formula to get p1 as a
function of x1
p1 =
x1
p1B
p1A
x1
x1
Demand for good 1
Demand for good 1
If p2 changes there is
If income m increases
in demand for good 1.
x1B x1A
demand for good 1
p1
TP
The demand curve
p1
0
x1
x1
TP
x1
x1
TP
7. Elasticity
Measuring the impact of changes in prices & income
Elasticity
Own price elasticity is
% change in quantity
% change in own price
Elasticity captures intuition better than

numerical change in quantity
numerical change in price
Elasticity matters
for every decision on prices: e.g.
for a monopoly or oligopoly deciding on prices
A price increase from 1 to 2 is large.

A price increase from 10 000 to 10 0001 is small.
Elasticity does not depend on units ( $ or , kilos or
pounds) because % changes do not depend on units.
for a government deciding on taxes.
Elasticity and demand curves
Own price elasticity of demand

either
x1
=
x1 = x1 p1
x1 p1
p1
p1 x1
p1 x1
(negative, Snyder &

Nicholson,lectures)
price of
good 1
B
p1
p1
or
x1
x1 =
x1 p1
x1 p1
p1
p1 x1
p1 x1
Some authors
makes elasticity
positive
quantity of good
x1
Which demand curve is more elastic A or B?
p1
TP
Uncompensated demand for good 1 is

2m
(from Cobb
x1 ( p1 , p2 , m )
Douglas utility)
5 p1

2m
(from Cobb
x1 ( p1 , p2 , m )
Douglas utility)
5 p1
Find
Find
own price elasticity
x1 p1
2m 5 p
2 1 p1 1
p1 x1
5 p1 2m
own price elasticity
x1 p1
2m 5 p
2 1 p1 1
p1 x1
5 p1 2m
cross price elasticity
x1 p2
5p
0 1 p2 0
p2 x1
2m
x1 p2
5p
0 1 p2 0
p2 x1
2m
income elasticity
x1 m
2 5 p1
See Consumer
m 1Theory
m x1 5 pWorked
1 2m Example 10
income elasticity
x1 m
2 5 p1
See Consumer
m 1Theory
m x1 5 pWorked
1 2m Example 10

2m
(from Cobb
x1 ( p1 , p2 , m )
Douglas utility)
5 p1
Find

x1(p1,p2,m) = 2
x 2(p1,p2,m) = 3 m
p1
5 p2
See Consumer Theory

Worked Example 10
x p
2m 5 p
own price elasticity 1 1 2 1 p1 1
p1 x1
5 p1 2m
x1 p2
5p
0 1 p2 0
p2 x1
2m
income elasticity
x1 m
2 5 p1
m 1
m x1 5 p1 2m
MRS = price ratio implies

x2 =
x2
3p1 x1
2p2
income consumption curve
x1
8. Normal and inferior goods
Normal & inferior goods
A good is normal if consumption

increases.
when income
A good is inferior if consumption

increases.
when income
income
elasticiti ty
m x1
x 1 m
m x1
x1 m
positive if x 1 is a
negative if x 1 is an
TP
10
Estimates of elasticities
Ready to eat breakfast cereals
Own price,
Estimates of elasticities
Cars
Own price,
between - 2.277 (CapN Crunch) & - 4.252 (Shredded Wheat)

Cross price (with Corn Flakes)
between 1.199981 (Fiat Uno) & - 4.843053 (Citroen XM)

Cross price small
between 0.012 (Lucky Charms) & 0.212 (Frosted Flakes)
Aviv Nevo, Measuring Market Power in the Ready-to-Eat

Cereal Industry Econometrica, Vol. 69, No. 2. (March
2001), 307-342
Paul Davis & Pasquale Schiraldi The Flexible Coefficient

Multinomilal Logit (FC-MNL) Model of Demand for
Differentiated Products Rand Journal of Economics, Vol. 45,
No. 1, pages 3263, Spring 2014.
TP
Median Estimates of
elasticities of demand for alcohol
own price
elasticity
beer
wine
spirits
demand with
quasi-linear utility &
corner solutions
Wagenaar, A.C. et al Effects of beverage alcohol price and tax levels on

drinking: a meta-analysis of 1003 estimates from 112 studies Addiction,
104, p. 179 - 190 February 2009

quasilinear utility

quasilinear utility
Utility functions with the form u(x1,x2) = V(x1) + x2 are called
There are two goods x1 and x2, with prices p1 and p2.
quasilinear utility functions.
A consumer has income m and a utility function
Assume V(0) = 0 first derivative
u(x1,x2) = x11/2 + x2
Find her (uncompensated) demand for goods 1 and 2.
V(x1) > 0 so V(x1) is increasing

second derivative V(x1) < 0 so V(x1) is concave.
Example V(x1) = x11/2.
11
Logic of first order conditions

If the nonsatiation and convexity
conditions are satisfied then any
point on the budget line at which
You have already

.
seen this.
Logic of first order conditions: new

If convexity and nonsatiation are satisfied then:
preferred
set
MRS = price ratio
x2
solves the utility maximizing

problem.
budget
set
x1
But there may be no

tangency point with x1 0,
x2 0

Check for nonsatiation
u( x1 , x 2 ) V ( x1 ) x 2 .

x2(p1,p2,m)

conditions
0
x1
Any corner point on
the budget line with
MRS p1/p2
x1 0, x2 = 0
maximizes utility.
0
x1
Any corner point on
the budget line with
MRS p1/p2
x1 = 0, x2 0
maximizes utility.
The problem is maximizing utility u(x1,x2) = V(x1) + x2
x1(p1,p2,m)
u dV
0,
x1 dx1
u
1 0 so nonsatiation is satisfied.
x 2
Check for convexity

Getting x 2 as a function of u and x1 gives x 2 u
dV
x 2
dx1
x1
Indifference cures with quasilinear utility

x2
Quasilinear utility
x1 0, x2 0
u
dV
x1 dx1 dV
MRS
u
1
dx1
x 2
V ( x1 ) so
d 2V
2 x2
2 0 so convexity is satisfied.
dx1
x 21

p1x1 + p2x2 = m
preferred
set
x1 0, x2 0
preferred
set
x2
x2
u(x1,x2) = V(x1) + x2
implies parallel
indifference curves (note
arrows)
= p1/p2
MRS depends only on x1.
solves the problem.

Note that because V is a function of x1 only MRS depends
only on x1
x1
12
Step 5: Draw a diagram based on the tangency

and budget line conditions
x2

tangency condition
You are finding demand x1 and x2 which is a

function of p1, p2 and m.
MRS = dV = p1
dx1
If V(x1) =
p2
x11/2

solved.
this is x1-1/2 = p1
2
p2
With V(x1) = x11/2 the equations are p1x1 + p2x2 = m

x1-1/2 = p1
and
2
p2
p1x1 + p2x2 = m budget line

0
x1

If p1 p22
4m
x2
(You will do some algebra here.)
x1(p1,p2,m) = p22
Solving the equations simultaneously for x1 and x2 gives

x2(p1,p2,m) =
(uncompensated) demand which is a function of p1, p2, m.

x1(p1,p2,m) = 1
p22
p12
x 2(p1,p2,m) =
Properties of demand with quasilinear utility

x1
4m
>
x1
p1
x2(p1,p2,m) =
x1
If p1 < p22
x1(p1,p2,m) = m
4p1
The solution is a tangency.
4p1
These conditions satisfy x1 0, but it is possible that x2 < 0 in

which case this is not a solution.
x2
m - p2
p2
m - p2
p2
4p12
This is a corner solution.
p22
, x2
4 p12
m
,
p1
x2
m
p
2
p2 4 p1
if p1
if p1
p22
4m
p22
4m
Like all demand functions this is homogeneous of degree 0 in

prices and income.
If
p1
p22
4m
demand for good 1 does not depend on
income but does depend on the relative prices of goods 1

0
x1
and 2. There is no income effect on demand for good 1.
13
Demand curve with quasilinear utility

u(x1,x2) = x11/2 + x2
Income consumption curve with quasilinear

utility
x2
p1
x1 = p22
When p1 p22
tangency solution
4p12
m has no effect
x1 = p22
4m
tangency solution
4p12
p22
x1 = m
(4m)
p1
income m has
on demand for good 1
corner solution
(4m2)/p22
p2 has no effect
x1
x1
Compare income consumption curve with

x1(p1,p2,m) = 2
x 2(p1,p2,m) = 3 m
p1
5 p2
x2
Income has an effect

on demand for good 1.
Quasilinear utility and

Marshalls model of
consumer demand
Income elasticity of
demand for good 1 is
1.
0
x1
Another way of looking at the tangency

condition with quasilinear utility
Consumer maximizes V(x1) + x2
Another way of looking at the tangency

condition with quasilinear utility
The first order condition is
From the budget constraint p1x1 + p2x2 = m
dV
p1 = 0
dx1
p2
so x2 = (m - p1x1)/p2
and V(x1) + x2 = V(x1) (p1/p2)x1 + m
The consumer choses x1 to maximize this.
If good 2 is interpreted as spending on other goods so p2 = 1

the foc is the same as Marshalls
dx1
As V(x1) is concave V(x1) (p1/p2)x1 + m

is concave (check this by looking at the 2nd derivative)
so the first order condition gives a maximum.
dV = p1
You can think of Marshalls model as the

special case of the general model with
quasilinear utility.
14
10. Substitutes and complements

If demand for good 1 increases when the price of good
2 increases goods 1 and 2 are substitutes
Substitutes &
complements
If demand for good 1 decreases when the price of

good 2 increases goods 1 and 2 are complements
cross price elasticitity
p2 x1
x 1 p 2
p 2 x1
x 1 p 2
positive if x 1 and x 2 are

negative if x 1 and x 2 are
3TP
Cross price elasticity with quasilinear utility

cross price elasticity of demand
Price of
Shifts in demand curves
good
p1
p 2 x1
x 1 p 2
Quantity of good
p2
p2
With quasilinea r utility if m 2 then x 1 22 .
4 p1
4 p1
Shift A to C.
Find the cross price elasticity of demand.
This is an increase in demand.
x1
Causes? Increase or decrease in price of a complement?

Increase or decrease in price of a substitute?
Increase or decrease in income for a normal good.
TP
demand with
perfect complements utility
Increase or decrease in income for an inferior good.

In general u(x1,x2) = min(ax1,bx2)
here u(x1,x2) = min( x1, x2)
x1 bike wheels, x2 bicycle frames
u
u(x1,x2) = min( x1,100)
100
if x2 = 100.
0
200
x1 wheels
Getty Images
15

diagram
Perfect complements utility: indifference curves

Consumer Theory Worked
Example 6 Uncompensated
demand with perfect
complements utility
u(x1,x2) = min( x1,x2)
frames
x2
x1 bicycle wheels,
x2 bicycle frames
x2
x2 = x1
if x2 < x1 increasing x1
does not change utility
In EC201 nonsatiation means

that increasing both goods
increases utility.
preferred
set
If the utility function has

derivatives there is nonsatiation
if the partial derivatives are both
strictly positive.
x1
if x2 > x1 increasing x1
increases utility.
0
wheels
x1
frames
x2
Increasing both goods

increases utility.
The perfect
complements utility
function satisfies
nonsatiation.
A
0
wheels
x1
frames
x2
Calculus cannot be
used to check for
nonsatiation.
wheels
x1
Perfect complements: utility maximization

u(x1,x2) = min( x1,x2)
budget line
p1x1 + p2x2 = m
utility maximization
frames
implies that (x1,x2)
x2
x2 = x1
lies at the kink of the

indifference curves so
x2 = x1
and satisfies the budget
constraint so
u(x1,x2) = min( x1,x2)

frames
x1 bicycle wheels,
x2 bicycle frames
x2
x2 = x1
But the utility function

has no derivatives at
a point like A.
Perfect complements utility: indifference curves
wheels
x1
increases utility.
wheels
x1
Perfect complements: utility maximization

u(x1,x2) = min( x1,x2)
utility maximization implies that (x1,x2) lies at the kink of the
indifference curves so x2 = x1
and satisfies the budget constraint so p1x1 + p2x2 = m.
Solving simultaneously for x1 and x2 gives
x1 =
2m
(2p1 + p2)
x2 =
m
(2p1 + p2)
p1x1 + p2x2 = m.
16
Demand curves and changes in prices and

income with perfect complements utility
p1
p1 = m - p2
x1
demand curve diagram,

price on vertical axis
quantity on horizontal axis
x1
demand with
perfect substitutes utility:
corner solutions again
Increase in p1 results in
Increase in p2 results in
Increase in m results in
TP

perfect substitutes utility
Perfect substitutes utility

In general u(x1,x2) = ax1 + bx2
The problem is maximizing utility u(x1,x2) = 3x1 + 2x2
u(x1,x2) = 3x1 + 2x2.

indifference curves
x2
u = 3x1 + 2x2
gradient 3/2
0
x1
Consumer Theory Worked

Example 8 Uncompensated
demand with perfect
substitutes utility


x1(p1,p2,m)
x2(p1,p2,m)

conditions
u
u
2 0 so nonsatiation is satisfied.
3 0,
x1
x2
Getting x 2 as a function of u and x1 gives x2 (u
x 2
3/ 2
x1
3x1 )/ 2 so
p1x1 + p2x2 = m
so is on the budget line
and
MRS =
x2
0
x 21
2
convexity is satisfied.
MRS =
u
x 1
u
x 2
p1
Problem
p2
What if p1/p2 = 3/2 ?
2 3 / 5 3 / 5
x1 x 2
5 3
3 2 /25 2 / 5
x1 x 2
5
It is better to use a
diagram.
2x 2
3x 1
3TP
17
x2
A
x2
x2
p1
B
B
C x1
Demand curves and changes in prices

and income with perfect substitutes
x1
p1/p2 < 3/2
p1/p2 = 3/2
p1/p2 > 3/2
solution at
solution at
solution at
x1
(3/2)p2
(2m)/3p2
x1
Increase in p1,
movement along curve.
Demand curves and changes in prices

and income with perfect substitutes
demand if convexity is not
satisfied
p1
p1
(3/2)p2
(3/2)p2
(2m)/3p2
x1
(2m)/3p2
x1
Increase in m,
Increase in p1,
movement along curve.
Increase in demand.
Increase in p2, increase

in demand.
The demand curve with nonconvex

preferences

nonconvex utility
x2
x2
p1
See CT
Worked
Example 5.
x2
supply
p2
0
x1
p1 < p2
x1 = m/p1, x2 = 0
x1
p1 = p2
x1 = m/p1, x2 = 0
or x1 = 0, x2 = m/p2
x1
demand
m/p2
There is a jump in
the demand curve.
There may be no
price at which
supply = demand.
x1
p1 > p2
x1 = 0, x2 = m/p2
The consumer never consumes both goods. The

solution is not at a tangency point.
18
1.3 Expenditure minimization and

compensated demand
1.3 Expenditure
minimization and
compensated demand
1. Definitions of compensated & uncompensated demand

2. Definition of the expenditure function
3. Homogeneity of the compensated demand and
expenditure functions
4. Income & substitution effects
5. The slope of compensated demand curves
6. Compensated demand & the expenditure function with
Expenditure minimization and compensated

demand
7. Compensated demand & the expenditure function with
quasilinear utility
8. Consumer surplus with quasilinear utility
Expenditure minimization and compensated

demand
Why bother?
1. Income and substitution effects, essential for
9. Properties of the expenditure function
understanding the effects of changes in wages and
10. The Slutsky equation
taxes on labour supply and interest rates on savings.

2. Expenditure function, essential for consumer
surplus and welfare economics.
Definitions of
compensated and
uncompensated
demand
1. Definitions of compensated and

What we have been calling demand up to now is
uncompensated (Marshallian ) demand which
maximizes utility u given prices p1 and p2 and income m,
so is a function of p1, p2, m,
notation x1(p1,p2,m), x2(p1,p2,m).
Compensated (Hicksian) demand minimizes the cost of

obtaining utility u at prices p1 and p2 and is a function of
utility u, p1, p2, notation h1(p1,p2,u), h2(p1,p2,u).
To get uncompensated demand fix

income and prices which fixes the
budget line.
x2
Get onto highest possible

indifference curve.
0
x1
Definition of the
expenditure function
To get compensated demand fix

utility and prices which fixes the
indifference curve and gradient of
budget line.
x2
x1
and intuition
Get onto lowest possible budget

line.
2. Definition of the expenditure function

The expenditure function E (p1, p2,u) is the minimum
amount of money you have to spend to get utility u with
prices p1 and p2. It is a function of p1, p2 and u.
The amount of goods which minimizes the cost of

getting utility u is compensated demand h1 (p1, p2,u),
Intuition for the expenditure function

A student buys 100 packs of sandwiches a year. The sandwich
price rises from 1 to 1.50. Could the student maintain the
same level of utility with
50 more?
60 more?
40 more?
20 more?
h2 (p1, p2,u)
so E(p1, p2,u)= p1h1 (p1, p2,u) + p2 h2 (p1, p2,u).
Getty Images
TP
The logic of choice
Prices change, income changes so that you

can still buy the same bundle of goods
You cannot have lower utility.
You have higher utility if there is any possibility of

substitution.
Given a choice between grapes, cherries and
apple you chose grapes.
Essential logic
Are you worse off if you are forced to chose

between grapes, apricots & blackberries?
No because you can still get grapes.
3. Homogeneity of the compensated

demand and expenditure functions
Compensated demand is homogeneous of degree 0 in
prices.
Homogeneity of the
compensated demand
and expenditure functions
If k > 0
h1(kp1,kp2,u) = h1(p1,p2,u).
Expenditure function is homogeneous of degree 1 in prices.

If k > 0
E(kp1,kp2,u) = k E(p1,p2,u).
The next slides explain why.
Mathematical definition of homogeneous functions
To get compensated demand fix

utility and prices which fixes the
indifference curve and gradient of
budget line.
x2
A function f(z1,z2,z3..zn) is homogeneous of degree 0 if for all

numbers k > 0
f(kz1,kz2,kz3..kzn) = k0 f(z1,z2,z3..zn) = f(z1,z2,z3..zn).
Multiplying z1,z2,..zn by k > 0 does not change the value of f.
x1
Get onto lowest possible budget

line.
Compensated demand depends on the indifference curve

and the slope p1/p2 of the budget line.
A function f(z1,z2,z3..zn) is homogeneous of degree one if

for all numbers k > 0
Multiplying p1 and p2 by k does not change the slope so

does not change compensated demand so
f(kz1,kz2,kz3..kzn) = k1 f(z1,z2,z3..zn) = kf(z1,z2,z3..zn)
h1(p1,p2,u) = h1(kp1,kp2,u)
Multiplying z1, z2 zn by k multiplies the value of f by k.
From the definition of the expenditure function

E(p1, p2,u)= p1h1 (p1, p2,u) + p2 h2 (p1, p2,u)
& E(kp1, kp2,u)= kp1h1 (kp1, kp2,u) + kp2 h2 (kp1, kp2,u)
As h1(p1,p2,u) = h1(kp1,kp2,u)
h2(p1,p2,u) = h2(kp1,kp2,u).
E(kp1, kp2,u)= kp1h1 (p1, p2,u) + kp2 h2 (p1, p2,u)

= k [p1h1 (p1, p2,u) + p2 h2 (p1, p2,u)}
h2(p1,p2,u) = h2(kp1,kp2,u).

prices.
Income and substitution

effects and compensated
and uncompensated
demand
so E(kp1, kp2,u)= k E(p1, p2,u).
The expenditure function is homogeneous of degree 1

in prices.
4. Income and substitution effects

The price of good 1
increases from p1 to p1.
x2
effect B to C,
change in
demand due
The logic of
compensated
demand and the
effect A to B,
change in
demand due to
effect A to C change
in
demand due to
m/p1
m/p1
x1
3TP
(h1,h2) is by definition
the cheapest way of
getting utility u at
prices (p1, p2).
By definition
compensated demand h1(p1,p2,u), h2(p1,p2,u)
x2
is the cheapest way of getting utility u at prices p1, p2

and costs p1 h1(p1,p2,u) + p2 h2(p1,p2,u) = E(p1,p2,u)
Therefore any other way of getting utility u costs the
same or more at prices p1, p2
Thus if u(x1,x2) = u then E(p1,p2,u) p1x1 + p2x2
(h1,h2) is by definition
the cheapest way of
getting utility u at
prices (p1, p2).
x2
(h1,h2)
So at prices (p1, p2)
slope
- p1/p2
0
x1
any other (x1, x2) with

utility u cannot lie
below the budget line
through (h1,h2) with
slope - p1/p2
(h1,h2)
So at prices (p1, p2)
slope
- p1/p2
0
x1
any other (x1, x2) with

utility u must lie on or
above the budget line
through (h1,h2) with
slope - p1/p2
Compensated demand
curves cannot slope
upwards:
geometric proof
Same information as in previous slide
Geometric proof that the compensated

demand curve cannot slope upwards
5. The slope of compensated demand

curves
The substitution effect of an increase in the price of a
good decreases or leaves unchanged the demand for
the good.
The compensated demand curve can never slope
upwards.
IMPORTANT
p1
RESULT
compensated
The next
demand curve
slides explain
0
h1A(p1A,p2,u) h2A(p1A,p2,u) compensated demand with

prices p1A,p2 and utility u.
Therefore u(h1A,h2A) = u.
h1B(p1B,p2,u) h2B(p1B,p2,u) compensated demand with
prices p1B,p2 and utility u.
Therefore u(h1B,h2B) = u.
x1
h1A(p1A,p2,u) h2A(p1A,p2,u) is the cheapest way of getting
x2
utility u at prices p1A,p2

u(h1B,h2B) = u so h1B,h2B is another way of getting
utility u
(h1A,h2A)
slope
- p1A/p2
therefore h1B,h2B cannot cost less than h1A,h2A at prices p1A,p2
x1
h1B,h2B cannot cost less h1A,h2A at prices p1A,p2

so (h1B,h2B) cannot lie in the shaded area.
h1B(p1B,p2,u) h2B (p1B,p2,u) is the cheapest way of getting
x2
utility u at prices p1B,p2

u(h1A,h2A) = u so h1A,h2A is another way of getting
utility u
therefore h1B,h2B cannot cost more than h1A,h2A at prices
p1B,p2
(h1A,h2A)
slope
- p1B/p2
0
x1
h1B,h2B cannot cost more than h1A,h2A at prices p1B,p2
so (h1B,h2B) cannot lie in the shaded area.
slope
Compensated demand
curves cannot slope
upwards:
algebraic proof
- p1B/p2
x2
(h1A,h2A)
slope
- p1A/p2
x1
Here p1B > p1A

(h1B,h2B) cannot lie in either shaded area
(h1B,h2B) must lies in the white triangle, so h1B h1A.
Proof that the compensated demand curve

cannot slope upwards
h1A(p1A,p2,u) h2A(p1A,p2,u) is the cheapest way of getting
utility u at prices p1A,p2
Algebraic proof that the compensated

h1B(p1B,p2,u) h2B (p1B,p2,u) is the cheapest way of getting
utility u at prices p1B,p2
u(h1B,h2B) = u so h1B,h2B is another way of getting

utility u
u(h1A,h2A) = u so h1A,h2A is another way of getting

utility u
Therefore at prices p1A,p2 quantities h1A,h2A cannot cost more

than h1B,h2B
Therefore at prices p1B,p2 quantities h1B,h2B cannot cost more

than h1A,h2A.
so in algebra p1Ah1A + p2h2A p1Ah1B + p2h2B
so in algebra p1Bh1B + p2h2B p1Bh1A + p2h2A
Remember the logic used here to derive the inequalities
Remember the logic used here to derive the inequalities
Algebraic proof that the compensated

Inequalities from the two previous slides
Simplify the inequality from the previous slides
p1Ah1A + p2h2A p1Ah1B + p2h2B
p1Ah1A + p2h2A + p1Bh1B + p2h2B

p1Ah1B + p2h2B + p1Bh1A + p2h2A
p1Bh1B + p2h2B p1Bh1A + p2h2A

Add the inequalities to get
Subtract p2h2A and p2h2B from both sides to get
p1Ah1A + p2h2A + p1Bh1B + p2h2B
p1Ah1A + p1Bh1B +
p1Ah1B + p1Bh1A
p1Ah1B + p2h2B + p1Bh1A + p2h2A

Remember the logic used here and derive the inequalities
Everything that follows comes from simplifying this inequality.
Remember the logic used here and derive the inequalities

Everything that follows comes from simplifying this inequality.
From the previous slide

p1Ah1A + p1Bh1B +
p1Ah1B + p1Bh1A
Rearrange to get 0 ( p1B - p1A ) ( h1A h1B)

so if the price rises from p1A to p1B so p1B - p1A > 0
either h1A h1B = 0 so compensated demand does not change
Compensated demand
and the expenditure
function with CobbDouglas utility
or h1A h1B > 0 so compensated demand falls.
6. Compensated demand and the expenditure

function with Cobb-Douglas utility
Finding compensated demand with CobbDouglas utility
The problem is minimising expenditure p1x1 + p2x2 subject to
non-negativity constraints x1 0 x2 0
and the utility constraint u(x1,x2) = x12/5x23/5 u.
Why does this matter?
See the next slide.
Worked example 4
Compensated demand
with convexity and a
differentiable utility
function.
Compensated demand is a function of prices & utility so is

h1(p1,p2,u)
h2(p1,p2,u)
Essential logic
With nonsatiation and convexity
utility u
x2
if there is a tangency
point such as A
Step 4: Use the tangency and utility

conditions
where MRS = p1/p2
Tangency requires that
MRS =
and utility is u
A
B
x1
this is compensated
demand
here we have already found
because any cheaper

point such as B gives
lower utility.
MRS =
u
x 1
u
x 2
2 3 / 5 3 / 5
x1 x 2
5
3 2 / 5 2 / 5
x1 x 2
5
p1
p2
2x 2
3x 1

tangency and utility conditions
Step 4: Use the tangency and utility conditions
utility u
x2
2x2 = p1
tangency condition
3x1
utility condition
p2
x12/5 x23/5 = u
because if x12/5 x23/5 > u there is a

cheaper way of getting utility u or more.
0
x1


You are finding compensated demand h1 and h2

which is are functions of p1, p2 and u.
2p
3p
This gives x1 2 u x 2 1 u
3 p1
2 p2
h1 ( p1 , p2 , u ) h2 ( p1 , p2 , u )
using notation
3/ 5

solved.
The equations are
x12/5
x23/5
=u
3/ 5
2x2 = p1
3x1
for compensated demand

2p
h1 ( p1 , p2 , u ) 2 u
3 p1
Note h1 ( p1 , p2 , u ) 0,
and
p2
Uncompensated and compensated demand

with Cobb-Douglas utility
2/5
3p
h2 ( p1 , p 2 , u ) 1
2 p2
h2 ( p1 , p2 , u ) 0.
2/5
u.
The expenditure function with Cobb-Douglas

utility
Compensated demand is
3/ 5
E ( p1 , p2 , u ) p1h1 ( p1 , p2 , u ) p2 h2 ( p1 , p2 , u )
Compensated demand
3/ 5
2p
h1 ( p1 , p2 , u ) 2 u
3 p1
2/5
2p
3p
h1 ( p1 , p2 , u ) 2 u
h2 ( p1 , p2 , u ) 1 u
p
3
1
2 p2
so the expenditure function is
Uncompensated demand
2m
3 m
x1 ( p1 , p2 , m)
x2 ( p1 , p2 , m)
5 p1
5 p2
2/5
3p
h2 ( p1 , p2 , u ) 1 u.
2 p2
3/ 5
2/5
2p
3p
p1 2 u p2 1 u
p
3
1
2 p2
3/ 5
2/5
2
3
p12 / 5 p23 / 5 u
3
2
Income and substitution effects with CobbDouglas utility
Income and
substitution effects with
x2
The price of good 1

Substitution effect
decreases
increases
Income effect
decreases
TP

with Cobb-Douglas utility
m/p1
m/p1
x1
Income and substitution effects with

Uncompensated demand
x1 ( p1 , p2 , m)
2 m
5 p1
x2 ( p1 , p2 , m)
income effect
3 m
5 p21
Compensated demand
2p
h1 ( p1 , p2 , u ) 2
3 p1
3/ 5
3p
h2 ( p1 , p2 , u ) 1
2 p2
substitution effect
2/5
u.
Compensated demand
and the expenditure
function with perfect
complements utility
7. Compensated demand and the

expenditure function with perfect
complements utility
u(x1,x2) = min( x1,x2)
frames
x1 bicycle wheels,
x2 bicycle frames
x2
x2 = x1
increases utility.
wheels
x1
Finding compensated demand with perfect

complements utility
u(x1,x2) = min( x1,x2) = u
min( x1,x2) = u
x2
x2 = x1
Expenditure minimization
implies that (x1,x2)
lies at the kink of the

indifference curves so
x1 ( p1 , p2 , m)
x2 = x1
h1 ( p1 , p2 , u ) 2u
and gives utility u so

0

with perfect complements utility
x1 = x2 = u.
x1
2m
,
2 p1 p2
x2 ( p1 , p2 , m)
m
2 p1 p2
h2 ( p1 , p2 , u ) u
substitution effect compensated demand
h1(p1,p2,u) = 2u, h2(p1,p2,u) = u.

Income and
substitution effects with
perfect complements
utility
income effect
substitution effect
TP
Income and substitution effects with perfect

complements utility
h1(p1,p2,u) = 2u, h2(p1,p2,u) = u.
The price of good 1
x2
x2 = x1
A
C
0
m/p1
There is
The income effect
reduces demand for x1
and x2.
The expenditure function with perfect

complements utility
Compensated demand is
h1 ( p1 , p2 , u ) 2u
h2 ( p1 , p2 , u ) u
E ( p1 , p2 , u ) p1h1 ( p1 , p2 , u ) p2 h2 ( p1 , p2 , u )
p1 2u p2u (2 p1 p2 )u
m/p1 x1
10
8. Compensated demand and the

expenditure function with quasilinear utility
Compensated demand
and the expenditure
function with
quasilinear utility
Utility functions with the form u(x1,x2) = V(x1) + x2 are called

quasilinear utility functions.
Assume V( 0 ) 0,
dV
0
dx1
Example V ( x1 ) x11 / 2 .
Finding compensated demand with

quasilinear utility
The problem is minimising expenditure p1x1 + p2x2 subject
to
d 2V
0.
dx12
Worked example 4
Compensated demand
with convexity and a
differentiable utility
function.

quasilinear utility
non-negativity constraints x1 0 x2 0
and the utility constraint u(x1,x2) = V(x1) + x2 u.
Why does this matter? See the next two slides.

Compensated demand is a function of utility & prices so is
h1(p1,p2,u)
h2(p1,p2,u)

Essential logic

Essential logic
If there is a tangency
point such as A
x2
x2
if there is a corner point

such as C where x2 = 0
where MRS = p1/p2
utility = u
and utility is u
this is compensated
demand
because any cheaper

point such as B gives
lower utility.
B
0
and MRS > p1/p2
this is compensated
demand
x1
x1
because any cheaper

point such as D gives
lower utility.
11

quasilinear utility
conditions
Tangency requires that
MRS =
p1

quasilinear utility
conditions
dV = p1
tangency condition
dx1
p2
x1-1/2 = p1
If V(x1) = x11/2 this is
u
dV
x1 dx1 dV
MRS
u
1
dx1
x 2
if V(x1) = x11/2 this is

tangency and utility conditions
p2
V(x1) + x2 = u
utility condition

quasilinear utility
p2
x11/2 + x2 = u

quasilinear utility
You are finding compensated demand h1 and h2
which is are functions of p1, p2 and u.
x2

solved.
A
x1-1/2 = p1
The equations are
p2
and
x11/2 + x2 = u.
x1

quasilinear utility
2
p
p
This gives x1 2
x2 u 2
2 p1
2 p1
using notation
h1 ( p1 , p2 , u ) h2 ( p1 , p2 , u )
for compensated demand
p
h1 ( p1 , p2 , u ) 2
2 p1
h2 ( p1 , p2 , u ) u
p2
2 p1
if p1 < p2/(2u) then this formula makes h2 < 0. This is

impossible, in this case there is a corner solution.

quasilinear utility
Tangency and corner solutions
If p1
p2
there a tangency solution
2u
2
p
p
h1 ( p1 , p2 , u ) 2
h2 ( p1 , p2 , u ) u 2
2 p1
2 p1
p
If p1 2 there is a corner solution,
2u
h1 ( p1 , p2 , u ) u 2
h2 ( p1 , p2 , u ) 0.
12
The expenditure function with quasilinear

utility
If p1
The expenditure function with quasilinear

utility
p2
compensated demand is
4u
p2
compensated demand is
4u
h1 ( p1 , p2 , u ) u 2 h2 ( p1 , p2 , u ) 0
If p1
p
p
h1 ( p1 , p2 , u ) 2 h2 ( p1 , p2 , u ) u 2
2
p
2
p1
1

E ( p1 , p2 , u) p1h1 ( p1 , p2 , u ) p2 h2 ( p1 , p2 , u )
E ( p1 , p2 , u) p1h1 ( p1 , p2 , u ) p2 h2 ( p1 , p2 , u )
p1u 2 p2 0 p1u 2
p
p1 2 p2 u 2
2 p1
2 p1
2
p
p2 u 2
4 p1

quasilinear utility
Tangency and corner solutions

with quasilinear utility
p2
solution is a tangency
2u
such as A.
If p1
x2
p
If p1 2 solution is a corner C.
2u
A
C
0
If p1
p22
uncompensated demand,
4m
2
p
x1 ( p1 , p2 , m) 2 x2 ( p1 , p2 , m)
2 p1
If p1
m
p
2
p2 4 p1
p2
compensated demand
2u
p
h1 ( p1 , p2 , u ) 2
2 p1
h2 ( p1 , p2 , u ) u
p2
2 p1
x1

quasilinear utility
Income and
substitution effects
quasilinear utility
The price of good 1

x2
Substitution effect A to B
B
decreases x1
A
increases x2.
Income effect B to C
x1
decreases x2.
There is no income effect
on demand for x1.
TP
m/p1
m/p1
x1
13
9. Consumer surplus with quasilinear utility

Marshall max V(x) px, First order condition V(x) = p
Consumer surplus with

quasilinear utility
Quasilinear u(x1,x2) = V(x1) +x2 , V(0) = 0

Interpret x2 as money, so p2 = 1.
Budget constraint with p2 = 1 implies that p1x1 + x2 = m so
utility is V(x1) + m p1x1
Foc at tangency V(x1) = p1, like Marshall,
Same interpretation as Marshall.
Quasilinear utility and consumer surplus

With quasilinear utility u(x1,x2) = V(x1) + x2
by assumption V(0) = 0, V(x1) > 0, V(x1) < 0.
If demand is not at a corner
Interpreting the quasilinear utility function

Assume that good 2 is money.
Assume that the consumer has a choice between
1. paying A and getting x1 units of good 1 and having

m A to spend on other goods. Utility V(x1) + m A.
there is no income effect

the compensated and uncompensated demand curves
are the same.
Interpreting the quasilinear utility function

If V(x1) + m A < m
2. Paying 0, getting 0 units of good 1 and having m to spend

on other goods. Utility V(0) + m = m (because V(0) = 0).
Quasilinear utility
Basic calculus, if V(0) = 0
then area under graph =
the consumer will not accept the offer.
If V(x1) + m A m the consumer will accept the offer.
x1
V ' ( x )dx V ( x ) V (0) V ( x )

1
The highest value of A at which the consumer will accept is

when V(x1) + m A = m
V(x1)
p1
V(x1)
or A = V(x1).
V(x1) is the most the consumer is willing to pay for x1 units if
the alternative is having 0 units.
x1
14
Quasilinear utility
With quasilinear utility
V(x1) = area under graph

= max consumer is willing to pay for x1 units
if good 2 is interpreted as expenditure on other goods so

p2 = 1 the MRS = p1/p2 condition becomes
V(x1) = p1
V(x1)
p1
V(x1) = uncompensated =
compensated demand
The curve V(x1) is

both the uncompensated and the compensated demand
curve.
x1
Quasilinear utility
Quasilinear utility
p1x1 = amount paid for good
Consumer surplus
= V(x1) - p1x1
= max willing to pay for x1
amount paid for x1.
p1
compensated demand
p1x1
x1
Properties of the
1, 2, 3
1. Increasing in utility
2. Increases or does not change when a price increases.
V(x1) - p1x1
p1
compensated demand
x1
10. Properties of the Expenditure

Function
1.
Increasing in utility
2.
The expenditure function increases or does not

change when a price increases.
3.
Homogeneous of degree 1 in prices.
4.
Shephards lemma
5.
Concave in prices
E(p 1 ,p 2 , u )
p 1
h 1 (p 1 ,p 2 ,u )
3. Homogeneous of degree 1 in prices.
15
Definition of the expenditure function

The expenditure function E (p1, p2,u) is the minimum
amount of money you have to spend to get utility u with
prices p1 and p2. It is a function of p1, p2 and u.
The amount of goods which minimizes the cost of

getting utility u is compensated demand
1. The expenditure function is increasing in

utility
x2
Utility increases
from u1 to u2.
u1
E(p1,p2,u2)
p2
The expenditure
function increases.
u2
E(p1,p2,u1)
p2
u2
h1 (p1, p2,u), h2 (p1, p2,u)

so E(p1, p2,u)= p1h1 (p1, p2,u) + p2 h2 (p1, p2,u).
u1
0
2. The expenditure function increases or

does not change when prices increase
The price of good 1
increases from p1A
to p1B,
compensated
demand moves
from A to B.
x2
E(p1B ,p2,u2)
p2
E(p1A ,p2,u1)
p2
The expenditure
function increases.
x1
3: The expenditure function is homogeneous

of degree 1 in prices.
prices.
h1(kp1,kp2,u) = h1(p1,p2,u).
If k > 0
Expenditure function is homogeneous of degree 1 in prices.

If k > 0
E(kp1,kp2,u) = k E(p1,p2,u).
The expenditure
function does not
change if demand
for good 1 is 0 at A.
0
Already explained.
x1
4. Shephards Lemma
Properties of the
expenditure function 4
Shephards Lemma
E(p 1 ,p 2 , u )
p 1
E(p 1 ,p 2 , u )
p 1
derivative of expenditure
The next
slides
explain
this.
h 1 (p 1 ,p 2 , u )
=
compensated
demand for good 1
function w.r.t. p1
gradient h1(p1A,p2,u)
h 1 (p 1 ,p 2 , u )
E(p1,p2,u)
u constant
p2 constant
p1 varies
p1A
p1
16
h1A(p1A,p2,u) h2A(p1A,p2,u) compensated demand with
E(p1,p2,u) is the cheapest way of getting utility u at prices

(p1,p2)
prices p1A,p2 and utility u.
(h1A,h2A) is another way of getting utility u, and
Therefore u(h1A,h2A) = u and
at prices (p1,p2) costs p1h1A + p2 h2A so
p1Ah1A + p2 h2A = E(p1A,p2,u)
E(p1,p2,u) p1h1A + p2 h2A
h1(p1,p2,u) h2(p1,p2,u) compensated demand with
for all p1
E(p1A,p2,u) = p1Ah1A + p2 h2A
prices p1, p2 and utility u.

Therefore u(h1,h2) = u
and p1h1A(p1,p2,u) + p2 h2A(p1,p2,u) = E(p1,p2,u)
New diagram
In this diagram (h1A,h2A) u and p2 do not vary, p1 varies.
E(p1,p2,u) p1h1A + p2 h2A
for all p1
E(p1A,p2,u) = p1Ah1A + p2 h2A.

Cost of buying (h1A,h2A) at
prices (p1, p2) is p1h1A + p2 h2A
amount
spent
p1h1A + p2 h2A
The graph of E(p1,p2,u) cannot be anywhere inside

the shaded area.
p1h1A + p2 h2A slope h1A
is a straight line
with gradient h1A
E(p1A,p2,u)
p1 price
p1A
p1 price good 1
p1A
The graph of E(p1,p2,u) meets the line with slope
Shepards Lemma
h1A at A and is never above the line.

so the line is tangent to the graph of E(p1,p2,u) at A
so the derivative of E(p1,p2,u) with respect to p1 at A is h1A
(compensated demand for good 1)
p1h1A + p2 h2A slope h1A
The derivative of the expenditure function

E(p1,p2,u) with respect to p1
is compensated demand for good 1
E(p 1 ,p 2 , u )
p 1
E(p1,p2,u)
h 1 (p 1 ,p 2 , u )
E(p1A,p2,u)
p1A
p1 price
17
5. The expenditure function is concave

in prices
Properties of the
expenditure function 5
concave in prices
because the graph of the
expenditure function lies on or below
all its tangent lines.
p1
11. The Slutsky equation

m E( p1, p2 , u)
When
The Slutsky equation
total effect of change in price

on demand at constant
income
x1 ( p1, p2 , m)
p1
h1 ( p1, p2 , u)
x1 ( p1, p2 , m)
x1 ( p1, p2 , m)
p1
m
substitution effect
of change in price
on demand, at
constant utility
Why bother with the Slutsky Equation?
h1(p1, p2,u)
x2
income effect of
change in
income on
demand
2TP
compensated demand for good 1.
x 1(p1, p2,m) uncompensated demand for good 1.

B
It is the only way of knowing how big income and

substitution effects are.
C
Combined with elasticity estimates it tells us that income

effects are too small to bother with except for goods that
are a large proportion of the budget.
gradient (p1 + p1)/p2

gradient p1/p2
x1
A to B, change in compensated demand. This is the

substitution effect.
B to C income effect. This is the shift in uncompensated
demand when income m changes.
18
Reminder: special cases
x2
Intuition for the Slutsky Equation
Quasilinear utility
Uncompensated demand for good
1 does not depend on income
except at low levels of income and
x1
x2
When p1 rises to p1 + p1 this is as if income falls by x1 p1

because this is how much more it costs to buy x1, so the
effective change in income is m = -x1 p1 .
The change in x1 through the income effect is approximately
x1
m
m
is equal to compensated demand.

No income effect.
Total change
depend on prices. No substitution
h1 ( p1 , p2 , u )
effect.
x1 ( p1 , p2 , m ) when m
E ( p1 , p2 , u )
But
p1
As h1 ( p1 , p2 , u )
h1 ( p1 , p2 , u )
p1
x1 ( p1 , p2 , m )
p1
E ( p1 , p2 , u )
x1 ( p1 , p2 , m ) E ( p1 , p2 , u )
m
p1
h1 ( p1 , p2 , u ) (Shephard' s lemma).
x1 ( p1 , p2 , m ) this implies that
x1 ( p1 , p2 , m )
p1
x1 ( p1 , p2 , m )
x1 ( p1 , p2 , m ) .
m
h1 ( p1 , p2 , u )
p1
x1 ( p1 , p2 , m )
x1 ( p1 , p2 , m ) .
m
p1 h1
h1 p1
Rearrangin g
x1 ( p1 , p2 , m )
p1
p1 x1
x1 p1
Own price elasticity
h 1
p1
p1
h1
p1
so
x1
p1
x1
x1 p
m
x1
x1
m
for EC201
so from the chain rule for partial derivative s

h1 ( p1 , p2 , u )
p1
x1
x1 p1
m
h1
p1
p1
x1
Compensated demand does not
Calculus proof of the Slutsky equation, not
The change in x1 through the substitution effect is approximately
Perfect complements utility
x1
m x1 p1 x1
x1 m m
Own price
Income
budget
of uncompensated
elasticity of
elasticity of
share
demand
compensated
uncompensate
demand
d demand
When m E ( p1 , p2 , u )
Slutsky equation
x1 ( p1 , p2 , m )
p1
h1 ( x1 , x 2 , u )
p1
x1 ( p1 , p2 , m )
x1 ( p1 , p2 , m )
m
leave out the arguments to make this easier to write down

x1
h1
x
x1
p1
p1
m
multiply by p1 / x1
p1 x1
x1 p1
p1 / h1 because x1
p1 h1
h1 p1
h1
m x1 p1 x1
x1 m m
Slutksy equation in elasticities
Income effects and the Slutksy Equation
p1 x1
x1 p1
p1 h1
h1 p1
Sign?
m x1 p1 x1
x1 m m
Sign for a normal
good
Sign for an inferior
good
19
Price of
Elasticity of demand curves
good
p1
Quantity of good
x1
Which demand curve is more elastic A or B?

If good 1 is a normal good which is more elastic,
compensated or uncompensated demand?
If good 1 is an inferior good which is more elastic,
compensated or uncompensated demand?
TP
20
1.4 Price change and welfare

1. Price indices
1.4 Price change and

welfare
2. Substitution bias
3. Compensating variation
4. Compensating variation and change in consumer
surplus with quasilinear utility
5. Compensating variation (CV) and change in consumer
surplus with income effects
6. Equivalent variation (EV)
7. Income effects, CV, EV and consumer surplus

8. Using EV to assess the effect of a tax
Base weighted price

indices
9.Using EV to assess the effect of a subsidy

10.Does compensating a consumer for a price increase
imply that the price increase has no effect on demand?
11.Benefits in kind
1. Price indices measure inflation

Public sector use
Inflation targeting
Adjusting levels of taxes, benefits, public pensions
Indexed government bonds (gilts)
Measurement of real wages
Private sector use

Pensions
Measurement of real wages
Price & wage setting
The design of price indices matters and is

controversial
Prices of different things change at different rates.
Price indices are weighted averages
What should be included in the index?
Weights
Formula
TP
1. Price Indices
Base weighted price index
Also called Laspeyres Price Index
Perloff uses CPI index to mean a base weighted index
p1Bx1A + p2Bx2A , .pnBxnA
Base weighted
price index
p1Ax1A + p2Ax2A , .pnAxnA
= w1 p1B + w2 p2B + + wn pnB

p1A
p2A
where w1 =
UK Price Indices: CPI and RPI
The base weighted price index is a weighted average of

proportionate price increases where weight for good i is
the proportion of expenditure spent on good i at date A.
Consumer Price Index (CPI)
CPI covers all private household
Retail Price Index (RPI)
RPI covers most private households but
CPI & RPI are essentially base weighted indices but the
weights change over time as expenditure patterns change.
etc.
x1A, x2A , .xnA consumption at date A, index is

p1Bx1A + p2Bx2A , .pnBxnA
pnA
p1A x1A
not pensioners,
not top 4% income
not foreign visitors
not foreign students

RPI
includes but CPI does not include

council tax, estate agent fees
CPI
includes but RPI does not include

university fees for foreign students
student accommodation.

From April 2011, benefits, tax credits, public sector
pensions
indexed by CPI, not RPI
From April 2012 tax thresholds indexed by CPI not RPI
Basic state pension, increases by highest of
CPI inflation, increase in average earnings, 2.5%
Indexed gilts (government securities) used by pension funds
still indexed by RPI.
Percentagechange
0.4
3.6
0.2
3.2
1.1
2.6
1.3
0.7
1.5
10.3
2.8
0.6
1.6
CPI Annual Inflation, 1.6%, July 2014
Source, Office of National Statistics, Consumer Price

Inflation July 2014 (published 19.08.2014)
Food&nonalcoholicbeverages
Alcohol&tobacco
Clothing&footwear
Housing&householdservices
Furniture&householdgoods
Health
Transport
Communication
Recreation&culture
Education
Restaurants&hotels
Miscellaneousgoods&services
CPIAllItems
UnitedKingdom
CPI12monthratetoJuly2014
CPI 12-month rate to July 2014
United Kingdom
CPI Annual Inflation, 1.6%, July 2014
CPI is internationally comparable RPI is not

CPI was originally developed for decisions on
macroeconomic convergence for Euro adoption
CPI does not include owner occupied housing
Housing markets differ across counties
Impossible to get international agreement
The ONS (Office for National Statistics) is working on

CPIH which includes owner occupied housing: hard to
do.
Food & non-alcoholic beverages

Alcohol & tobacco
Clothing & footwear
Housing & household services
Furniture & household goods
Health
Transport
Communication
Recreation & culture
Education
Restaurants & hotels
Miscellaneous goods & services
CPI All Items
Percentage change
-0.4
3.6
-0.2
3.2
1.1
2.6
1.3
0.7
1.5
10.3
2.8
-0.6
1.6
Source, Office of National Statistics, Consumer Price

Inflation July 2014 (published 19.08.2014)
CPI inflation
Oil
CPI services
excluding VAT
Agriculture
CPI goods excluding

energy & VAT
Industrial metals
Sources: Bloomberg, S&P indices and Thomson Reuters Datastream.
Bank of England Inflation Report August 2013
Inflation by age group July 2014
Source: Alliance Trust, Inflation Alert Aug 14
(a) Brent forward prices for delivery in 1021 days time in US dollars.
(b) Calculated using S&P (US dollar) commodity price indices.
Source Bank of England Inflation Report August 2014
Spending Weights for previous graph
How do spending patterns vary with age?

Inflation by Age Group Jul 14
Source:http://www.ons.gov.uk/ons/rel/family-spending/family-spending/family-spending-2011-edition/family-spending-2011-pdf.pdf
National price indices are averages

So may be poor measures of inflation for some people
Expenditure weights are different for different people
Price changes are different in different places.
Price indices and

substitution bias
x2
2. Substitution bias
p1Bx1A + p2Bx2A
p1Ax1A + p2Ax2A
The base weighted price index

measures the proportional increase
in the cost of (x1A,x2A)
E(p1B,p2B,u)
E(p1A,p2A,u)
The expenditure function price index

measures the proportional increase
in the cost of getting utility u
Fact
(x1A, x2A) maximizes

utility subject to the
budget constraint
p1Ax1 + p2Ax2 = m1
u1
m1/p2A
(x1A, x2A) is the

cheapest way of getting
utility u1 at prices
(p1A,p2A) so the
base weighted index expenditure function price index.
The next slides explain why.
m3/p2B
m2/p2B
( x1A,x2A)
gradient
p1A/p2A
0
x1
E(p1A,p2A,u1)
= p1Ax1A + p2Ax2A = m1
x2
m3 = p1Bx1A + p2Bx2A
x2
gradient p1B/p2B
m3/p2B
m2/p2B
is the cost of ( x1A,x2A)

at prices (p1B,p2B)
(x1B, x2B) is the

cheapest way of
getting utility u1 at
prices p1B, p2B
u1
m1/p2A
( x1A,x2A)
is the cost of ( x1A,x2A)

at prices (p1A,p2A).
0
x1
proportional increase in income
needed to continue to buy (x1A,x2A)
after the price change
= base weighted price index.
m1
p1Ax1A + p2Ax2A
x2
u2
u1
m1/p2A
so the expenditure
function
m1 = p1Ax1A + p2Ax2A
gradient p1B/p2B
m3/p2B
m2/p2B
( x1A,x2A)
( x1B,x2B)
E(p1B,p2B,u1)
= p1Bx1B + p2Bx2B = m2
x1
(x1A, x2A) also gives utility u1 so costs the same or more than
than (x1B, x2B) at prices p1B, p2B
so m2 = p1Bx1B + p2Bx2B p1Bx1A + p2Bx2A = m3.
x2
gradient p1B/p2B
m3/p2B
m2/p2B
m3/p2B
m2/p2B
u2
u1
m1/p2A
gradient p1B/p2B
u2
u1
m1/p2A
( x1A,x2A)
( x1A,x2A)
gradient
( x1B,x2B)
( x1B,x2B)
x1
Here m2 < m3 so a consumer with income m3 has higher

utility than a consumer with income m2.
Base weighted price index

m1
p1Ax1A + p2Ax2A

needed to continue to buy ( x1A,x2A)
after the price change.
is greater than
m2 = p1Bx1B + p2Bx2B
m1
p1Ax1A + p2Ax2A
=
E(p1B,p2B,u1)
E(p1A,p2A,u1)
p1A/p2A
0
If prices change from (p1A,p2A) to (p1B,p2B) and income
changes from m1 to m2 utility does not change.
If prices change from (p1A,p2A) to (p1B,p2B) and income
changes from m1 to m3 utility increases.
x1
Substitution bias
If income grows at the same rate as a base weighted price
index utility either increases or stays the same.
If there is any possibility of substitution utility increases.

needed to continue to have utility u1
after the price change.
Thus base weighted price indices overstate the rate of

inflation.
expenditure function price index
If different prices increase at very different rates so relative prices

change the substitution bias is larger. US calculations, due to
inflation bias the CPI overstates inflation by 0.5 % see Perloff
page 127.
Reminder: the expenditure function is

homogeneous of degree 1 in prices
that is if k 0 then
E (kp1 , kp2 , u ) kE ( p1 , p2 , u )
so if p1B kp1 A & p2 B kp2 A , so both prices grow at the same rate
expenditure function price index
base weighted price index
E ( p1B , p2 B , u A ) kE ( p1 A , p2 A , u A )
k
E ( p1 A , p2 A , u A ) E ( p1 A , p2 A , u A )
The UK CPI formula is different from the RPI formula, in part to

allow for substitution bias.
There have been big recent changes in relative prices.
p1B x1 A p2 B x 2 A kp1 A x1 A kp 2 A x 2 A
k.
p1 A x1 A p2 A x 2 A
p1 A x1 A p2 A x 2 A
Both price indices increase at the same rate.
TP
Over the long term the introduction of new

goods is a big issue for price indices.
Who today would buy a film stored on
this?
Problems with the expenditure function price

index
Calculating the expenditure function price index requires
knowledge of the expenditure function.
The expenditure function is derived from the utility
function.
www.freephoto1.com
The utility function is unobservable.

Who in 1990 downloaded a film from
the internet?
Getty Images
There is a neat way round this for situations where only

one price changes.
This is the compensating and equivalent variation
developed by Hicks which turn out to be closely related
to consumer surplus.
3. Compensating variation (CV)

Definition
Compensating variation
The compensating variation for a price increase from p1A to

p1B is the amount of extra money the consumer needs to
get back to the same level of utility as before the price
change.
Compensating variation and the expenditure

function
x2
At prices (p1A,p2) the consumer buys (x1A,x2A) giving utility

u(x1A,x2A) = uA.
gradient p1B/p2
E(p1B,p2,uA)
p2
After the price change & compensation the consumer gets the
same level of utility by buying (x1B,x2B) so u(x1B,x2B) = uA.
CV
p2
The consumer minimises the cost of getting utility so
uA
( x1A,x2A)
E(p1A,p2,uA)
the amount spent at prices (p1A,p2) is E(p1A,p2,uA),
gradient
p2
the amount spent at prices (p1B,p2) after compensation is

E(p1B,p2,uA).
( x1B,x2B)
p1A/p2
Compensating variation = E(p1B,p2,uA) - E(p1A,p2,uA)
x1
The relationship between differentiation and

integration
CV and Shephards Lemma

The problem remains, CV depends on the expenditure
function so on utility so is unobservable.
If y f ( x )
Remember Shephards lemma
E(p 1 ,p 2 , u )
p 1
f(x)
dy
f ' ( x)
dx
h 1 (p 1 ,p 2 , u )
To work with this you need to remember the relationship

between differentiation and integration.
then f(b) f ( a ) f ' ( x )dx

a
p1 B
CV and Shephards Lemma

Shephards lemma
E(p 1 ,p 2 ,u )
h 1 (p 1,p 2 , u )
p 1
p1
h1 ( p1 , p2 , u A )dp1
p1 A
= compensating variation
when p1 rises from p1A to
p1B.
p1B
p1A
implies that
compensated demand
curve h1(p1,p2,uA)
0
x1B x1A
x1
Price increases from p1A to p1B

compensated demand falls from x1A to x1B
Compensating variation
and change in consumer
surplus with quasilinear
utility
4. Compensating variation and change in

consumer surplus with quasilinear utility
u(x1,x2) = V(x1) + x2
V(0) = 0, V(x1) > 0, V(x1) < 0.
There are no income effects.
The compensated and uncompensated demand curves are
the same.
The compensating variation is the fall in consumer surplus.
The next slides show why.
Quasilinear utility
Consumer surplus
= V(x1) - p1x1
With quasilinear utility

the fall in consumer surplus
p1
= compensating variation
= max willing to pay for x1

amount paid for x1.
V(x1) - p1x1
p1
compensated demand
p1B
p1A
x1
Compensating variation and Marshall
compensated demand
x1
5. Compensating variation and change

consumer surplus in with income effects
Marshalls change in consumer surplus looks the same as

compensating variation and is defined in the same way
In effect Marshall assumes quasilinear utility.
Compensating variation is the loss of consumer surplus
associated with the compensated demand function.
Compensating
Variation ACDF
p1B
compensated demand
h1(p1,p2 ,uA)
p1A
When there are income effects compensating variation and

change in consumer surplus are different.
x1
Fall in consumer
surplus ABDF
In the modern approach

consumer surplus is
defined as this area
p1B
p1B
x1(p1,p2 ,m)
p1A
F
x1(p1,p2 ,m)
p1A
x1
x1
uA is the level of utility that the consumer gets with

prices (p1A,p2) and income m.
compensated demand
Change in Consumer Surplus is the area bounded by the
uncompensated demand curve.
Compensating variation is the area bounded by the

compensated demand curve with utility uA.
h1(p1,p2 ,uA)
p1B
x1(p1,p2 ,m)
p1A
x1
Compensated demand is less elastic than

uncompensated demand.
Income and substitution effects work in the
same direction. This is a normal good
Fall in consumer
surplus ABDF
compensated demand
p1B
h1(p1,p2 ,uA)
x1(p1,p2 ,m)
p1A
F
compensated demand
p1B
x1
h1(p1,p2 ,uA)
x1(p1,p2 ,m)
p1A
F
x1
Compensating
Variation ACDF
Compensating
Variation ACDF
Fall in consumer
surplus ABDF
compensated demand
p1B
h1(p1,p2 ,uA)
p1B
x1(p1,p2 ,m)
p1A
F
With a price rise for a

normal good
fall in CS < CV.
p1A
F
x1
x1
6. Equivalent Variation (EV)

Definition
Compensating variation &

equivalent variation
EV is the amount of money that taken away from the

consumer without changing prices has the same effect
on utility as the price change.
EV = E(p1B,p2,uB) - E(p1A,p2,uB)
The next
slide shows
why.
compensated demand
x2
EV = E(p1B,p2,uB) - E(p1A,p2,uB)
E(p1B,p2,uB)
h1(p1,p2 ,uB)
CV = E(p1B,p2,uA) - E(p1A,p2,uA)
p2
uB is the level of
utility that the
consumer gets with
prices (p1B,p2) and
income m.
compensated demand
EV
p2
p1B
h1(p1,p2 ,uA)
x1(p1,p2 ,m)
p1A
E(p1A,p2,uB)
p2
uB
gradient p1B/p2
uA
gradient p1A/p2
0
Indifference curve diagram
x1
x1
10
compensated demand
compensated demand
EV Equivalent Variation ABEF
h1(p1,p2 ,uB)
Normal good
EV
h1(p1,p2 ,uB)
< change CS <
ABEF
compensated demand
p1B
h1(p1,p2 ,uA)
x1(p1,p2 ,m)
p1A
F
Normal goods, CV & EV

Uncompensated demand is more elastic than compensated
demand because income and substitution effects work in
the same direction.
For a price rise
EV < change in consumer surplus < CV
h1(p1,p2 ,uA)
x1(p1,p2 ,m)
p1A
F
x1
ACDF
compensated demand
p1B
ABDF
CV
x1
Compensating variation,
equivalent variation
consumer surplus and
income effects
because EV is measured at a lower level of utility.

As the good is normal less is consumed at lower utility.
7. Income effects, CV, EV and consumer

surplus
The Slutsky equation in elasticities

shows the size of the income effect
When there are no income effects, e.g. with quasilinear

utility, uncompensated and compensated demand are
the same so
the loss in consumer surplus due to an increase in p1 is
the same as the CV & EV.
p1 x1
x1 p1
p1 h1
h1 p1
m x1 p1 x1
x1 m m
The difference between CV, EV and the change in

consumer surplus is due to income effect.
11
substitution
effect
p1 x1
x1 p1
budget
income effect
p1 h1
h1 p1
Which measure to use?
share
m x1 p1 x1
x1 m m
Income elasticity of
of uncompensated
of compensated
uncompensated
demand
demand
demand
If income effects are small, for example because the budget

share is small, CV, EV and change in CS are close. Use
change in CS because it is easy to measure.
If income effects are large, it depends on the question you
want to answer.
CV if it is how much is needed to compensate.
Income effects are small when either or both of the income

elasticity of uncompensated demand and the budget share are
small.
If income effects are small the change in consumer surplus is a
good approximation to the compensating variation.
EV if it is what is the monetary equivalent of the price

change.
Examples
A government wants to reduce CO2 generation to combat
global warming.
Fuel is a large share of expenditure so the income effect is
significant.
When does an increase in price from p1A to p1B

have a big adverse impact on the consumer?
It does this through taxation.

To make this politically acceptable it needs to compensate
people for the increase in tax, the compensating
variation is relevant.
It wants to know how different groups, e.g. pensioners,
families with young children are affected.
This is measured by equivalent variation.
uncompensated
demand
p1
p1B
p1A
TP
TP
Fall in consumer surplus

due to p1 increasing from
p1A to p1B .
demand
p1
p1B
p1A
Diagram A
0
x1B x1A
p1
x1
uncompensated
demand.
p1B
p1A
Diagram B
x1B
x1A
x1
In which diagram is
demand more elastic
A or B?
In which diagram is
the fall in consumer
surplus bigger, A or
B?
demand curve
diagram
x1B x1A
x1
The fall in consumer surplus due to a price increase

from p1A to p1B is less than (p1B p1A)x1A the extra
money needed to buy x1A.
The fall in consumer surplus is less when
demand is elastic, usually because there is a good
substitute available.
demand curve diagram
12
demand
p1
Formula for fall into

consumer surplus.
p1B
p1A
Fall in Consumer Surplus
Let p1 = p1B p1A > 0

Let x1 = x1B x1A < 0
x1B x1A
1
p1 x1 A p1x1
2
x1
fall in consumer surplus =
1
( p1B p1 A ) x1 A ( p1B p1 A )( x1 A x1B )
2
1
p1 x1 A p1x1
2
The fall in consumer surplus is less when

demand is more elastic.
1 x p p
1 x1
p1 x1 A 1 1 1 1
p1 x1 A 1
2 p x
x
2
1A
1 1 A p1
1 p
x p
p1 x1 A 1 e 1 where e 1 1 0 elasticity
p
2
p1 x1 A
Why is the fall in consumer surplus less

when demand is more elastic?
1 p
Fall in consumer surplus p1 x1 A 1 e 1
2 p1 A
Demand is elastic when it there is a good
1 p
p
1 ( p1 x1 A )1 e 1
p
1A
2 p1 A
p
where 1 proportionate price increase
p1 A
e
substitute available,
x1 p1 A
elasticity
p1 x1 A
p1 x1 A expenditure on good 1 before the price increase.
Adding up demand curves and

consumer surplus
Each person in group 1 earns 8,000 per year.

Each person in group 2 earns 100,000 per year,
Add the demand curves

There are 100 people in each group.
Situation 1, everyone in group 1 losses CS 250, everyone in
group 2 looses CS 50. Total loss CS
price
= 100 x 250 + 100 x 50= 30000.

D1
Situation 2, everyone in group 1 losses CS 50, everyone in

group 2 looses CS 250. Total loss CS
D2
quantity
TP
= 100 x 50 + 100 x 250= 30000.
TP
13
Adding up consumer surplus geometrically implies a value

judgement that giving 1 to one consumer has the same
social benefit as giving 1 to any other consumer.
If you disagree with this judgement you would want to

evaluate the losses to each group, and then consider how
to use them as input into a decision.
Using equivalent variation to

assess the effect of a tax
This can be modelled mathematically.
8. Using EV to assess the effect of a tax

Suppose that a tax causes the price of good 1 to rise from
p1A to p1A + t, where t is tax. (This is called an excise
tax.)
(We will see later that this is a special case, it happens
when supply is perfectly elastic.)
Demand for good 1 falls from x1A to x1B. Demand for good 2
rises from x2A to x2B.
Tax revenue t x1B

Good 2 is expenditure on other goods so p2 = 1.
From the budget constraint without tax
p1A x1A
x2A
=m
budget constraint with tax

(p1A + t) x1B +
x2B
= m
Subtract these equations to get

(p1A x1A + x2A ) (p1Ax1B + x2B) - t x1B = 0.
How much revenue does the tax raise?
Tax revenue
x2
EV
(p1A x1A + x2A ) (p1Ax1B + x2B) - t x1B = 0.
So tax revenue = t x1B = (p1Ax1A + x2A) (p1A x1B + x2B)
Excess burden =
A
EV tax revenue
(x1B,x2B)
(x1A,x2A)
C
Cost of original bundle
(x1A,x2A) at prices
p1A , p2 = 1.
Cost of new bundle

(x1B,x2B) at prices p1A,
p2 = 1.
budget lines without tax

slope p1A
0
IC diagram, p2 = 1
x1
budget line with tax

slope (p1A + t)
2TP
14
Definition: The excess burden
EV tax revenue
The lump sum tax raises the same

revenue as the excise tax but gives
higher utility.
x2
of an excise tax is
IC utility uA
= loss to consumer tax revenue.
With an excise tax there is an excess burden.
Budget constraint
with lump sum tax
R
(x1A,x2A)
Suppose that instead of imposing an excise tax that increased

the price of good 1 from
EV
>R
IC utility uC
p1A to p1A + t and raised revenue R
IC utility uB
the government imposed a lump sum tax that took away

R from the consumer so the budget constraint is
x1
Budget constraint
with excise tax
p1Ax1 + p2 x2 = m R.
IC diagram, p2 = 1
This is a general argument.
9. Using EV to assess the effect of a subsidy.
Suppose the government wants to raise revenue R.

A lump sum tax that reduces income by R that does not
depend on anything the consumer does reduces utility
by less than a tax raises R where the revenue could be
changed by changing consumption, work or saving.
Suppose that a subsidy causes the price of good 1 to fall

from p1A to p1A - s , where s is the subsidy per unit.
(e.g. excise tax, VAT, income tax)

Demand for good 1 changes from x1A to x1C. Demand for
good 2 changes from x2A to x2C.
The only feasible lump sum tax is a poll tax where

everyone pays the same amount.
Is a poll tax ethically desirable?
How much does the subsidy cost?
Is a poll tax politically feasible?

2TP
Subsidy costs s x1C
x2
Good 2 is expenditure on other goods so p2 = 1.

From the budget constraint without subsidy
m
budget line
(x1C,x2C)
with subsidy
slope (p1A - s)
(x1A,x2A)
budget line without

subsidy slope p1A
0
x1
IC diagram, p2 = 1
p1A x1A
x2A
=m
budget constraint with subsidy

(p1A - s) x1C +
x2C
= m

(p1A x1A + x2A ) (p1Ax1C + x2C) + sx1C = 0.
TP
15

(p1A x1A + x2A ) (p1Ax1C + x2C) + sx1C = 0.
x2
Cost of subsidy?
Equivalent variation?
B
C
Rearrange cost of subsidy

= s x1C =
(p1A x1C + x2C)
(p1A x1A + x2A)
(x1C,x2C)
budget line with subsidy

gradient (p1A s)
(x1A,x2A)
Cost of new bundle
(x1C,x2C) at prices
Cost of original bundle

(x1A,x2A) at prices
p1A, p2 = 1.
p1A, p2 = 1.
budget line without

subsidy gradient p1A
0
TP
x1
IC diagram, p2 = 1
2TP

A lump sum subsidy that increases income by a fixed
amount that does not depend on anything the consumer
The equivalent variation (EV) of a subsidy is the amount of
extra income the consumer needs to get to have the
same effect on utility as the subsidy.
does increases utility more than a subsidy costing the

same amount where the cost of the subsidy can be
changed by changing consumption, work or saving.
Does compensating a
consumer for a price
increase imply that the
price increase has no
effect on demand?
10. Does compensating a consumer for a

price increase imply that the price increase
has no effect on demand?
Suppose a price (e.g. heating fuel) rises from p1A to p1B .

Demand for the good falls.
Suppose the consumer is compensated by being given the

compensating variation (CV). Does the consumer go
back to consuming the same amount of heating fuel?
16
x2
The price increase moves the

consumer from
IC utility uA
A price increase reduces demand even if the consumer is

compensated.
Following compensation the

consumer is at
B
budget
line after
price
change
is the substitution effect of

the price rise. Does it increase or
decrease x1?
CV?
This is an economists insight.
IC utility uB
budget line before price

change gradient p1A
It comes from knowing about income and substitution

effects.
gradient
p1B
x1
If income effects are small compensation has little effect

on the demand for the good.
TP
TP
IC diagram, p2 = 1
11.
Benefits in Kind
budget line
x2
with benefit in kind
Benefits in kind
The consumer gets a

benefit in kind, getting
x1D units of good 1
costing p1x1D for free.
budget line without

benefit in kind
0
In this figure would the

consumer be better off
just getting the money
as extra income?
x2
x1
budget line with

extra income
x2
budget line with

extra income
x1D
IC
In this figure would the

consumer be better off
just getting the money
as extra income?
IC
budget line without

benefit in kind
budget line without
benefit
0
x1D
TP
x1
IC diagram, p2 = 1
x1D
x1
TP
17
Why Benefits in Kind?
Why Benefits in Kind?
The last slides suggest that it is sometimes better and never worse
for a consumer to get a sum of money rather than a benefit in
kind costing the same amount.
So why are benefits in kind common?
The star at this month's Grammy ceremony was the $34,000 gift
bag (actually more of a suitcase), which contained an iPod, a
stereo system, clothing, holidays, a voucher for eye surgery and
a bottle of Trump fragrance for men.
http://film.guardian.co.uk/oscars2005 February 26, 2005
Screen Actors Guild Gift Bags Getty Images
Why do we give gifts

not money?
Why do
governments
provide health and
education free at
the point of service?
Mother and daughter (18-21 months) giving
food basket to senior woman Getty Images
Students Across The UK Return To School For Start Of The Autumn Term
Clinical Trials Begin For New Vaccine Against Avian Influenza Getty Images
18
1.5 Labour supply, taxes

and benefits
1. The budget constraint

2. Income and substitution effects of an increase in the
real wage
3. Quick overview of the UK tax system
4. Income tax
5. Benefits
Essential additional reading: chapters from

the Mirrlees Review
Chapter 3: Labour Supply and Taxes, Costas Meghir
and David Philipps, of
Dimensions of Tax Design: J. Mirrlees et.al. OUP April
2010.
Link from the website reading list, also at the Institute for
Fiscal Studies website
http://www.ifs.org.uk/mirrleesreview/dimensions/ch3.pdf
The labour economics

budget constraint and
utlity
Chapter 5: Integrating Personal Taxes & Benefits of

Tax by Design: the Mirrlees Review, J. Mirrlees et.al.
OUP September 2011
Link from the website reading list, also at the Institute for
Fiscal Studies website
http://www.ifs.org.uk/mirrleesreview/design/ch5.pdf
Notation
Assume that:
Utility u(c,n) depends only on consumption c and time outside
paid employment leisure n
endowment of time
e.g. one year = 365 x 24 = 8760 hours
n hours not in paid employment
(leisure)
T-n hours in paid employment
(work)
c consumption of composite good
P price of composite good
W hourly wage rate (that is earnings per hour)
w = W/P real wage rate
Non satiation, utility is increasing in both consumption and

leisure.
A standard but bad assumption
if the gain from working is not only the money earned, but
also other benefits such as status, meaning and purpose,
on the job consumption and social contact.
The other assumptions of consumer theory are satisfied
completeness, transitivity, continuity and convexity

1
1. The budget constraint
The Budget Constraint

Standard consumer theory budget
total consumption total earnings

Pc W(T - n)
or
p1x1 + p2x2 m term m on RHS given along with

prices p1, p2
Pc + Wn WT
or, dividing by P and recalling that
w = W/P
Labour economics budget constraint
budget constraint can be rewritten as:
c + wn wT
price of c is 1, price of n is w
c + wn wT
term wT on RHS depends on the real wage w.
Consumption + w leisure value of total time available

Non satiation implies that the budget constraint is satisfied as
an equality c + wn = wT
Whose behaviour is being modelled?
Does consumption = earnings each week in this

model?
An individual?
Does consumption = earnings each week in
A family?
reality?
When does consumption = earnings?
A group?
TP
Budget constraint: c + wn = wT
Where does the budget constraint meet
the axes?
A
c
wT
What is the gradient of the budget line?

wT
Preferences are represented by indifference curves.

A corner solution ( 0 consumption or 0 leisure is most
unlikely).
At a tangency solution MRS =
What happens to the budget line when w increases?
An increase in the real wage w is like a decrease in the

price of good 2 in standard consumer theory.
When w increases the budget line meets the horizontal axis
at the same point T, but becomes steeper.
TP

2
2. Income and substitution

effects of an increase in the
real wage w.
wT
Income and substitution

effects on labour supply
wT
In this diagram is n a normal

good?
C
A
Substitution effect A to B
Income effect B to C
Income and substitution effects on labour supply work in
directions.
real
real
I argued using the Slutsky equation that the size of the
income effect on demand for good 1 is small when the
budget share p1x1/m is small.
Here the budget share of n (leisure) is
wn/wT = n/T leisure/total time.
Is this budget share small?
TP
wage
rate w
labour
supply
wage
labour
rate w
supply
labour
labour
Labour supply
increases when
the wage rate
rises
Labour supply
decreases when
the wage rate
rises
which is bigger
substitution effect or
income effect?
which is bigger
substitution effect or
income effect?
2TP
Estimating the elasticity of labour supply

is hard due to
real wage
complicated budget constraints, depending on family

circumstances
w*
labour
supply
The labour supply curve is backward-bending
linked decisions
whether to get paid employment
how many hours, child care?
people who are not on their labour supply schedule,
unemployment, conventional hours.
when the substitution effect dominates the income

effect for wages below a certain wage w*, and the
income effect dominates the substitution effect
There is very little evidence for workers whose pay does

not depend on current hours worked, e.g. professionals.
above w*.

3
Husbands Wives
Mean uncompensated labour supply
elasticity
0.207
Mean compensated labour supply

elasticity
0.169
Effects of Unemployment
0.844
The labour supply model tells us that the cost of
unemployment to a worker is lost consumption.
0.941
In fact, the costs of unemployment in reality extend far

beyond just the monetary:
Unemployment is a disaster similar to marriage break-
Womens labour supply is generally more elastic than

mens
source Ashenfelter, The Labor Supply Response of Wage
Earners, in Palmer & Pechman, Welfare in Rural Areas, The
North Carolina-Iowa Income Maintenance Experiment
Brookings,1978 Data collected in1970 - 72
upin each case you cease to be neededthere is a

huge psychic coston top of whatever income an
unemployed person loses.
(Richard Layard, Happiness Has social science a clue?
Robbins Lectures 2002/3, LSE)
UK Taxes on income
Income Tax is paid on all income including income from
employment and income from investments.
Income & consumption taxes:

the simplest model
National Insurance Contributions (NICs) are paid on

income from employment only.
More details follow.
4. Modelling the effects of an income tax

Budget constraint without tax
Assume at 25% tax rate on consumption and so the price

increases to 1.25P. Tax revenue 0.25 Pc.
Pc = W(T n) or Pc + Wn = WT
or c + wn = wT
Budget constraint with 20% proportional income tax

total tax paid = 0.2 W (T-n)
Budget constraint Pc = W (T n) 0.2 W(T-n)
or Pc = 0.8 W (T n) or
Modelling the Effects of a Tax on Consumption
c + 0.8 wn = 0.8wT
Budget constraint with consumption tax

(1.25) Pc = W (T n) or Pc = 0.8 W (T n)
or c + 0.8 wn = 0.8wT same as with 20% income tax.
Tax revenue = 0.25Pc = 0.25(0.8 W (T n)) = 0.2 W(T-n) same
as 20% income tax.
In general a proportional income tax at rate tm and a proportional
consumption tax at rate tc raise the same revenue and have
the same effect on the budget constraint if
(1 - tm ) (1+ tc ) = 1.

4
Tax revenue given T- n* and c* as labour and consumption

= 0.20 W (T n*)
This diagram is distorted to make it easier

to follow. The gradient of the budget line
with tax is much too small.
wT
c* + wn*
tax revenue = WT (Pc* + Wn*)
= P (wT (c* + wn*) )
=
P
(0.2 = tax rate, W wage, T n* labour)

= W T 0.8 WT - 0.2 W n*
0.8wT
= WT (Pc* + 0.8 W n* ) 0.2 W n*

(because from the budget constraint 0.8 WT = Pc* + 0.8Wn* )

gradient 0.8 w
= WT (Pc* + Wn*)
gradient - w
= WT cost of (c*, n*) at pre tax prices P and W

0
Definition: Equivalent Variation for a price

From the
change
Price
The price of good 1 starts at p1A giving utility uA. Changes
The price of good 1 rises to p1B
and
Welfare
p2 does not change.
Taking away the equivalent variation, EV, without Slides
changing p1 from p1A has the same effect on utility as
increasing p1 from p1A to p1B without changing income.
wT
0.8wT
Definition: Equivalent Variation of a tax

gradient 0.8 w
The tax changes the price of good 1 leisure from W to

0.8W.
Taking away the equivalent variation, EV, without
changing the price of leisure has the same effect on
utility as imposing the tax.
wT
budget constraint no
tax gradient - w
n
gradient - w
0
tax gradient - w
n
wT
tax revenue =
c* + wn*
P
equivalent variation = P
0.8wT
0.8wT

gradient 0.8 w
gradient - w
0
tax gradient - w
n

gradient 0.8 w
gradient - w
0
tax gradient - w
n

5
wT
tax revenue =
c* + wn*
equivalent variation = P
excess burden =
equivalent variation - tax revenue
= P
0.8wT
wT
Taking
from the consumer as a
lump sum gives the
budget constraint
which gives the same tax revenue
as the income tax but is better for
the consumer.
c* + wn*
0.8wT

gradient 0.8 w
gradient - w
0
tax gradient - w
n

gradient 0.8 w
gradient - w
0
tax gradient - w
n

A lump sum tax that reduces income by a fixed amount
that does not depend on anything the consumer does
reduces utility by less than a tax raising the same
amount of revenue where the revenue can be changed
by changing consumption, work or saving.
(e.g. excise tax, VAT, income tax)
The only feasible lump sum tax is a poll tax where
everyone pays the same amount.
Income tax:
a more realistic model
Is a poll tax ethically desirable?

Is a poll tax politically possible?
A more realistic model of income tax

Divide income into tax brackets.
e.g. 0 - 5 000, 5 000 - 20 000, > 20 000
An income tax system gives a marginal tax rate for each
bracket, with higher brackets having higher marginal tax
rates.
If the tax rates are
0 % in bracket 1
20% in bracket 2
40 % in bracket 3
Usual description of this tax scheme

total annual
< 5000
5000 - 20000
> 20000
marginal tax income rate

0 %
20 %
40 %
Total income tax =

0.00 (income in bracket 1) + 0.20 (income in bracket 2)
+ 0.40 (income in bracket 3).

6
Income 15 000
Definition: Marginal and Average Income Tax Rates

Marginal income tax rate is the number of extra pennies
tax you pay on 1 extra earnings.
0 in bracket 3
20 000
10 000 in bracket 2
With this income tax scheme the marginal income tax rate
for someone earning 15000 is 20%.
Average income tax rate = total income tax
total income
If someone earning 15000 pays 2000 tax
the average income tax rate = 2000 = 13%
15 000
5 000
5000 in bracket 1
With an income of 15 000 you have 5000 in bracket 1

15 000 5000 = 10 000 in bracket 2. 0 in bracket 3.
Total tax = (0.00 x 5000) + (0.20 x 10 000) + (0.40 x 0 ) = 2000
Marginal tax rate = 20%
Average tax rate =
tax
= 2000 = 13%
income
15000
Income 30 000
Budget Constraint
15 000 in bracket 2
5,000
0
5000 in bracket 1
With an income of 30 000 you have 5000 in bracket 1

15 000 in bracket 2, 10 000 in bracket 3
annual consum ption ()
10 000 in bracket 3
20,000
slope - 2.4
wage 4 per hour

17000
slope - 3.2
5000
3760
Total tax = (0.00 x 5000) + (0.20 x 15 000) + (0.40 x 10 000 ) =

7000
Marginal tax rate = 40%
Average tax rate = 100 tax
= 7000 = 23%
income
15000
Diagrams for seeing

budget constraints:
the economists
diagram
consumption
c
The economists
diagram has the
advantage that
indifference
curves have
their usual
shape.
But non
economists
dont
understand it.
8760
See consumer theory worked example 10 on website

for more explanation and details of how to calculate the
budget constraint.
Flip the economists diagram horizontally.

consumption
indifference curves
gradient the other
way.
budget set
7510
time outside employment (annual hours)
leisure n
hours worked
This diagram makes much more sense to

non economists.

7
You often see diagrams with

earnings before tax rather than
hours worked on the horizontal
axis.
income
after tax
Simplified diagram of tax rates

before abolition of personal
allowance for high earners.
income
(Assumes only two different

marginal tax rates.)
after tax
With this diagram the shape of the

graph does not depend on the
wage rate.
0
earnings
earnings
Contrast with previous

slides.
effect on those
earning above e1
Effect of the abolition of the personal

allowance for high earners increases
marginal tax rate at earnings between
e0 and e1
income
after tax
income
after tax
C
decrease in utility,
does this stop people
working in the UK?
e0
e1
earnings
0
no substitution effect
e0
e1
earnings
A to C, income effect, increases labour supply
effect on those earning

between e0 e1 and e1
sign of overall effect on
labour supply unclear.
income
after tax
A
B
C
decrease in utility,
The UK income tax

system
does this stop people

working in the UK?
e0
e1
earnings
A to B, substitution effect decreases labour supply

B to C, income effect, increases labour supply

8
Taxation has always

been politically
David Cameron
controversial.
Conservative Party Leader
"You don't have to persuade me that high
marginal tax rates are a bad idea. The
sort of tax system I believe in is one that's
effective in raising revenue
(Spectator 6 October 2009)
Government receipts 2014 -15
Getty Images
Ed Miliband
Labour party leader
We should look to do more from taxation.
We can take more from the banks and in
tackling tax avoidance.
Guardian
(The Telegraph 30 Sep 2010)
Budget Document, HM Treasury: March 2014
UK: Taxes on consumption

VAT (value added tax) is the main tax on consumption.
Before January 09
17.5 %
January 09 December 09
15 %
January 10 December 10
17.5 %
January 11
20 %
Some goods are exempt from VAT.
(e.g. childrens clothes, public transport, books and newspapers)
Some goods have reduced tax rate.

(e.g domestic fuel, installing energy-saving materials, children's car seats)
Some goods are subject to additional taxes

(e.g. alcohol, tobacco, petrol)
UK income tax brackets

marginal
tax rate 2011 12
0%
2012 13 2013 14 2014 15
0 7,475
0 - 8,105 0 - 9,440
0 - 10,000
7,475 42,475
8,105 42,475
9,440 41,540
10,000 41,865
Higher 42,475 rate: 150,000

40%
42,475 150,000
41,540 150,000
41,865 150,000
Additional
> 150,000
rate: 50%
> 150,000
N/A
N/A
Additional
N/A
rate: 45%
N/A
> 150,000
> 150,000
Basic
rate:
20%
Source HMRC This ignores the treatment of the personal allowance &
child benefit for high earners which are discussed later.
For historical reasons the UK has 2 taxes on

income: Income tax and National Insurance
Income tax: based on all income including income from
employment and income from investments.
Total amount depends on annual earnings.
PAYE (pay as you earn) paid monthly or weekly based on
estimated earnings.
Amount adjusted by refund or extra charge at year end.
National Insurance Contributions (NICs): based on income from
employment only.
base on weekly income
Total depends on weekly earnings.
Additional employers contribution
National Insurance Contributions (NIC) 2014-15

weekly income
employees NIC
marginal rate
employers NIC
marginal rate
< 111
0%
111 - 805
12%
13.8%
> 805
2%
13.8%
From http://www.hmrc.gov.uk/rates/nic.htm
4TP

9
UK percentiles and median income

(annual, full time employment)
Median April 2013 (source ONS)
Men
29,000
Women
25,900
UK policy on high earners
50,000 post tax income 95th percentile

100,000 post tax income 99th percentile
(source, Institute for Fiscal Studies: Where do you fit in? )
Policy on high earners

Alistair Darling, Labour 2009
The next 3 diagrams are based on

Withdrawal symptoms: the new 'High Income Child
Benefit charge
Removal of personal allowance at 100,000

Robert Joyce
George Osborne, Conservative, 2013

Removal of child benefit at 50,000
Marginal income tax rates 2013 - 14 if these policies

No children
1 child
2 children
3 children
4 children
were not in place
80%
Institute for Fiscal Studies, January 2013

http://www.ifs.org.uk/publications/6527
Budget Statement: April 2009 removal of

personal allowance for incomes > 100,000
70%
I believe that it is fair that those who have gained the most
should contribute more ..
Marginal
income 60%
tax rate
50%
fully withdraw the benefit of that allowance for

those with incomes over 100,000 from next April
40%
30%
Only those with incomes over 100,000 a year or 2 per

cent of the population will be affected.
20%
10%
0%
0
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Annual taxable income (1000's)
Alistair Darling
Labour

10
Proposed policy
If income 100,000 get personal allowance
Income > 100, 000 no personal allowance
UK income tax brackets for a family with no children

with total removal of all personal allowance @ 100,000
marginal tax rate
Marginal tax rate 40% at 100,000.
0%
2014 15
For income <
100,000
2014 15
For income >
100,000
0 - 10,000
Eliminating 10,000 personal allowance brings the start

of the 40% tax rate down from 41,865 to 31,865
Basic rate:
20%
10,000 - 41,865
0 - 31,865
If income increases from 100,000 to 100,001 income

tax increases tax by 0.40 * 10,000 +0.40 = 4000.04
unacceptable
Higher rate:
40%
41,865 - 100,000
31,865 - 150,000
N/A
> 150,000
Loss of personal allowance in effect changes tax

brackets
Additional rate:
45%
Marginal income tax rates 2013 14 for a household

No children
1 child
2 children
3 children
4 children
with no children
Partial solution
80%
Personal allowance tapered at 50% on incomes >

100,000
i.e. earning 1 more reduces allowance by 50p.
70%
Margina
60%
l
income
tax rate 50%
40%
In effect this introduces marginal tax rates (2013 14)

40%
41,865 - 100,000
60%
100,000 - 120,000
40%
120,000 - 150,000
45%
> 150,000
Withdrawal of
personal
allowance @
100,000
30%
20%
10%
0%
0
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Removal of child benefit for high earners

Before April 2013,
1000 per year for one child family
regardless of income
More for a larger family.
Partial solution
Withdraw benefit gradually by increasing marginal tax rates
between 50,000 and 60,000
George
Osborne
Conservative
Changes marginal tax rates.
Proposed policy after April 2013

no benefit for a household with one income > 50,000
Increase income for single earner family from 50,000 50,001 loses 1,000
unacceptable

11
Marginal income tax rates 2013 - 14 with withdrawal of

No children
1 child
2 children
3 children
4 children
personal allowance for a household with children
Additional issues on child benefit
80%
50%
No children
4
3
2
1
40%
70%
Marginal
income 60%
tax rate
Withdrawal of
child benefit @
50,000
30%
20%
Tax is paid on basis of individual incomes

Benefits are paid on basis of household incomes
Child benefit is withdrawn if one partner earns > 50,000
Withdrawal of
personal
allowance @
100,000
A household with two earners each getting 49,999

keeps child benefit.
10%
0%
0
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
A household with one earner getting 50,000 loses child

benefit.
What can economists usefully say on this?
Additional issues on child benefit

Who is a partner?
HMRC
Partner means someone youre not permanently
separated from who youre married to, in a civil
partnership with or living with as if you were.
Many people in the UK do not have to complete tax
forms. Child benefit requires people to complete a tax
form who would not otherwise do so.
Benefits
5. Benefits
In the UK there are many different benefits including
Do not try to remember

Income Support,
this list,
Jobseekers Allowance,
Employment and Support Allowance
These are paid
Housing Benefit,
largely to people with
Child Tax Credit
low or zero income.
Working Tax Credit.
Child Benefit
Amount depends on
Council Tax Benefit
earnings and other
circumstances in
Disability Living Allowance
complicated ways.
State Pension
Winter Fuel Allowance

12
Benefits that depend on earnings

The inevitable conflict
Suppose the objective is that every family should have at
least 200 per week.
This can be done by giving every family a cash benefit
200 so they have 200 + y
where y is earned income.
Suppose the objective is that every family should have at

least 200 per week.
This can be done by giving every family earning income
y less than 200 a benefit of 200 - y
where y is earned income.
This is targeted on the poorest families.
But this is badly targeted, the rich get as much as the

poor.
It is much less expensive than giving 200 to all families

so there is less need for taxes or government borrowing.
It is expensive. The money has to come from

somewhere, taxes or government borrowing.
Benefits and Budgets Constraints
Definition: The Benefit Withdrawal Rate
Assume for simplicity that this is a household that pays no

taxes.
Suppose that the benefits system is designed to give this
family at least 200 per week.
If the family earns above 200 its gets no benefit.
This is the amount by which the benefit is withdrawn if

someone earns 1 more.
If benefit = 200 y the benefit withdrawal rate is 100%
because benefit is withdrawn by 100% of 1 when
income increases by 1.
benefit
200
If the family earns less than 200 it gets benefit 200 y.
200 - y
0
income after
taxes and
benefits
Would anyone with

this budget
constraint take a job
paying less than
200 per week?
200
income
income after
taxes and
benefits
Would this person

take a job paying
more than 200 per
week?
200
45
0
200
45
200
y income before benefits
0
TP
200

13
Making Work Pay

income after
taxes and
benefits
Would this person

take a job paying
more than 200 per
week?
200
The policy response to this disincentive to work caused by

the benefit system was to introduce a form of benefit
called a tax credit.
Tax credits are paid to people in work.
Earned Income Tax Credit (USA)
Working Families Tax Credit (UK 1999 2003)
Child Tax Credit and Working Tax Credit
UK 2003 onwards.
45
0
200
With a 100% benefit withdrawal rate there is no incentive to work

for less than 200.
Universal Credit 2014 partial introduction. 2017 ?
Benefit as a
function of income
benefit
What happens with a lower withdrawal rate?

200 0.50 y
Suppose the withdrawal rate is 50%.
200
benefit with 50%

withdrawal rate
Benefit is 200 for someone earning 0.

Benefit is 200 0.50 y for someone earning y < 400
Benefit is 0 for someone earning y 400.
200 y
benefit with
100%
withdrawal
rate
200
400
D
C
income after
taxes and
benefits
Budget line with

100% benefit
withdrawal rate ABD
200
Budget line with

50% benefit
withdrawal rate ACD
45
0
200
400
IC
income after
taxes and
benefits
200
This person
moves from A
to B when the
withdrawal
rate falls.
Labour supply
increases.
200
400

14
The tax credit trade off

income after
taxes and
benefits
E
F
200
This person
moves from E to
G when the
withdrawal rate
falls. Subst effect
EF and income
0
effect FG both
decrease labour
supply.
Reducing the withdrawal rate from 100% to 50%

improves work incentive for people earning less than
200 but worsens work incentives for people earning
between 200 and 400.
More generally lower withdrawal rates improve work
incentives for low earners and worsen work incentives
for moderate earners.
200
400
This is an inevitable trade off.
Lower withdrawal rates result in more benefits being paid

so are more expensive.
The money has to come from somewhere (taxes or
government borrowing.)
Lower withdrawal rates result in more people receiving
benefits and make the benefit system more difficult to
administer.
Up to now I have been looking at someone who gets

benefits but does not pay taxes.
In fact in the UK many people both get benefits and pay
taxes.
When a UK family on 300 per week earns 1 extra it
pays extra income tax
pays extra NIC*
0.20
0.11
losses benefit (tax credit) 0.37

so looses in total
0.68
and thus takes home an additional 0.32.

* NIC National Insurance Contribution
Definition: Effective Marginal Tax Rate EMTR

If this family on 300 per week earns 1 more it pays extra
tax
(income tax + NIC) of 0.325.
Its marginal tax rate t is 32.5 %.
It losses benefit 0.37.
Its benefit withdrawal rate b is 37%.
When a UK family on 3000 per week earns 1 extra it

pays extra income tax
pays extra NIC*
0.50
0.01
losses benefit (tax credit) 0.00

so looses in total (EMTR) 0.51
and thus takes home an additional 0.49.
Its effective marginal tax rate is m = t + b = 69.5 %.

This is how much it looses from extra taxes and lower
benefits when it earns 1 more.
* NIC National Insurance Contribution

15
Work incentives for single-adult families , 2008/9

(simulated)
100%
90%
80%
Single no children, METR
Lone parent, METR
70%
Universal Credit
60%
50%
40%
30%
20%
10%
0%
-10%
200
400
600
800
1,000
1,200
Pre-tax weekly earnings
Source: from Brewer et al (2008) and based on analysis of FRS

using TAXBEN.
METR = marginal effective tax rate. Includes employers NI but
not consumption taxes.
5. Benefits
Universal Credit
In the UK there are many different benefits including

The governments main welfare reform programme
Integrates parts of the tax and benefit system
Intended to
make the system easier to understand and use
improve work incentives
move system online
all in a time of austerity reduction of government
spending
Cannot escape the tax credit tradeoff
Universal Credit
Income Support,
Jobseekers Allowance,
Employment and Support Allowance
Housing Benefit,
Child Tax Credit
Working Tax Credit.
Child Benefit
Council Tax Benefit
Disability Living Allowance
State Pension
Winter Fuel Allowance
Universal
credit will
replace
these
benefits.
Universal Credit: National Audit Office

Report September 2013
Nationwide introduction originally planned for all new

claimants from October 1913
Rushed timetable
Currently (September 2014) being implemented in small

number of places
Problematic simultaneous detailed design of policy

and IT systems.
expected full implementation end of 2017.

If you live with your partner and both claim Universal Credit
youll receive a single payment that covers you both.
Project management and leadership difficulties.

Department of Work and Pensions response
substantial progress since April 2013
new leadership

16
Howard Shiplee
Director General, Universal Credit Programme
Accountable for implementing Universal Credit This
includes:
owning and communicating the vision of the
programme
Universal Credit: National Audit Office

Report September 2013
Follow up study
Scheduled Late Autumn 2014
http://www.nao.org.uk/work-in-progress/universal-creditfollow-study/
ensuring the implementation of Universal Credit is

completed safely and securely
providing clear leadership to the programme team
https://www.gov.uk/government/people/howard-shiplee

17
1. Choices between income streams

2. The intertemporal budget line and present discounted
value
3. Perfect capital markets
4. The simplest model of asset pricing
1. Choices between
income streams
5. Saving and borrowing decisions

6. Effects an interest rate cut
7. Different borrowing and lending rates
You can borrow and lend at 10%. Choose between A, B and C.

2014
2014
2015
A 10,000
22,000
B 18,000
11,000
C 20,000
11,000
You can borrow and lend at 10%. Choose between A, B and C.

2015
A 10,000
22,000
B 18,000
11,000
C 20,000
11,000
C is better than B.
If you start with A, borrow 10,000 in 2014, repay 11,000 in
2015 you get C.
With borrowing and lending at 10% A and C are
2TP
2TP

1
2. The intertemporal budget line and

present discounted value
c1
Assume you know for sure

2014
2015
y0
y1
income
(y0, y1 )
c0
You can borrow and lend at interest rate r.

Note with 10% interest r = 0.10.
c0 2014 consumption c1 2015 consumption
Saving at date 0 s0 = y0 c0 (if s0 < 0 you are borrowing s0)
Consumption at date 1 c1 = y1 + (1 + r)s0 = y1 + (1 +r) (y0 c0)
Budget constraint c1 = y1 + (1 +r) (y0 c0)
c1
Which budget line would you choose, A or B?

gradient (1+r)
Which income stream would you choose (y0, y1) or ( y *0 , y *1 ) ?
(y0, y1 )
c1
c0
Budget constraint c1 = y1 + (1 +r) (y0 c0) is a straight line with
gradient (1+r).
( y *0 , y *1 )
It can be rearranged to get

c0 +
c1
(1+r)
B (y0, y1 )
y0 + y1
(1+r)
This straight line meets the horizontal axis at y0 + y1
c0
y *1
(1 r )
y *0
(1+r)
c0
y0
y1
(1 r )
c0
TP
The present discounted value of the income stream (y0,y1) is

y0 + y1
(1+r)
Equations of 2 budget lines
Present discounted value
c0
c0
c1
(1 r )
c1
(1 r )
y0
*
0
y1
(1 r )
y *1
(1 r )
In this model non satiation implies that any consumer would

choose the income stream with the highest present value.

2
gradient - (1 + r)
Definition
c1
( y *0 , y *1 )
Which income
stream is
optimal at
interest rate r ?
The present discounted value (pdv) at date 0 of

an income stream y0, y1, y2,....yt,....yT, paying yt at
date t, discounted at an interest rate r is
gradient - (1 + r)
Which income
stream is
optimal at
interest rate r ?
pdv
(y0, y1 )
y0
c0
....
of
y 0 , y 1 , y 2 ... y t ,... y T ,
y1
(1 r )
y2
(1 r )2
is
.....
yt
yT
.....
(1 r )T
(1 r )t
TP
The pdv of y0, y1, y2,....yt,....yT

is the maximum amount of debt at date 0 which you could repay
using the entire income stream y0, y1, y2,....yt,....yT.
If you get the income stream y0, y1, y2,....yt,....yT,
at interest rate r, start with no savings and no debt at date 0, you
can follow any consumption path
and leave no debt or savings at date T whose pdv
c0
....
c1
(1 r )
c2
(1 r )2
3. Perfect capital markets

a strong assumption
1.
no uncertainty (can be relaxed)
2.
You can borrow and lend at the same interest rate r.
3.
.....
ct
cT
.....
(1 r )t
(1 r )T
is equal to the pdv of the income stream.
The only constraint on borrowing is that you must have

enough income to repay your debt.
4.
There is a perfect and costless mechanism which

ensures that no one takes on loans that they cannot
repay and that all debts are repaid.
Credit Crunch Questions

Why did households, firms, banks and governments
borrow money they are now not able to repay?
The simplest model of asset

pricing
Why were banks willing to lend to them?

3
The simplest model of asset pricing

Suppose you own the chess set which you do not like so
keep it in the bank, you are only interested in it as an
Renowned French
artist Bernard
Maquin created the
Royal Diamond
Chess set
Valued at over 5 million, it contains diamonds, emeralds, rubies,

pearls and sapphires. The king piece alone weighs 165.2 grams of 18
carat yellow gold, 73 rubies and 146 diamonds.
Source: http://www.charleshollandercollection.com/chess.html
If the alternatives are sell the chess set now or sell in 10 years
sell now if
5 million > 15 million
(1+r)10
Sell the chess set in 10 years if
5 million < 15 million
(1+r)10
You are indifferent between selling the chess set now
and selling it in 10 years if
5 million = 15 million
(1+r)10
asset.
How do you decide whether to sell it now or in 5 years?
Suppose everyone knows that the price now is 5 million
and the price in 10 years will be 15 million.
Selling the chess set now adds 5 million to
of your monetary wealth.
Selling the chess set in 10 years adds
million
of your monetary wealth.
TP
The chess set is unique, there is no other identical chess

sets.
Now suppose that this is a financial asset that many
people (or organisations) own.
If 5 million > 15 million everyone wants to sell now.
(1+r)10
If 5 million < 15 million everyone wants to buy now.
(1+r)10
The market can only clear if the current price is
15 million
this is a no arbitrage argument.
(1+r)10
More generally no arbitrage arguments in a model with

certainty imply that if an asset pays dividends d1, d2, dT
at dates 1,2 T and is sold at price pT at date T then the price
p0 of the asset at date 0 must be the present discounted value
of the dividends + price at date T, that is
p0
d1
d2
d3
......
2
(1 r ) (1 r )
(1 r )3
dT
pT
T
(1 r )
(1 r )T
No arbitrage conditions are very important in standard

finance theory.
I have assumed no uncertainty and perfect knowledge of
what dividends and asset prices will be in the future.
No arbitrage arguments can be extended to allow for
uncertainty and are the foundation of the Black Scholes
model of options pricing.
Some financial economists argue that there are
considerable limits to arbitrage and that this has important
consequences for asset pricing.

4
The simple asset pricing models assumes that people know

what prices will be in the future.
The model can be extended to allow for uncertainty under
strong assumptions.
If an increase in current asset prices leads people to expect
that future asset prices will be even higher, an increase in
asset prices can increase demand and further increase
prices.
5. Saving and borrowing

decisions
If a fall in current asset prices leads to people to expect that

future asset prices will be even lower, a decrease in asset
prices can decrease demand and further decrease prices.
Assume preferences
satisfy the standard
assumptions of
completeness,
transitivity, continuity,
nonsatiation and
convexity so can be
represented by a
utility function
u(c0,c1).
The consumer
maximizes u(c0,c1)
subject to the budget
constraint
c0
c1
(1 r )
y0
5. Saving and
borrowing decisions
c1
consumption at A
gradient - (1 + r )
c1
> current income

At a tangency solution
MRS = 1 + r
(y0, y1 )
The household borrows.

A
(y0, y1 )
gradient - (1 + r)
c0
c0
y1
(1 r )
This households current

consumption at A
This households current
gradient - (1 + r )
c1
< current income

A
The household saves.
Effects of an
interest rate cut
(y0, y1 )
0
c0

5
This household borrows
6. Effects an interest rate cut
before the rate cut.
gradient - (1 + r )
c1
(y0, y1 )
Interest
rate cut
from r to r.
gradient
The substitution effect A to

In a simple macro model, cutting the interest rate
increases growth
increases inflation
by
increasing investment by firms
increasing current consumption by households.
The question here is:
does the simple consumer theory model imply interest
rate cuts increase current consumption?
B of the cut increases

consumption at date 0.
For a borrower the rate cut

always increases utility;
B
the rate cut is like an

increase in income.
If c0 is normal the
income effect B to C
A rate cut makes current consumption relatively cheaper.
c0
Income and substitution effects on
c0 work in the same direction.
increases c0.
Intuition for the effects of an interest cut on a

borrower
- (1 + r)
This household saves

both before and after the
gradient - (1 + r )
c1
interest rate cut.
The substitution effect A to

The substitution effect increases current and decreases
future consumption.
B of the cut increases
A borrower is made better off by the rate cut; this has the
same effect as an increase in income.
The household has
If current consumption is a normal good the rate cut

increases current consumption.
after the rate cut

the income effect B to C
Income and substitution effects work in the same

direction.
decreases c0
Income and substitution effects on c0

work in the opposite direction.
Someone who saves
The substitution effect is the same as for a borrower; it

borrows after the rate cut
before the rate cut and
- (1 + r)
c0
gradient - (1 + r )
c1
may have higher or lower

A
utility after the rate cut.

Here she has higher utility
(y0, y1 )
after the rate cut.
C
B
If current consumption is a normal good the rate cut

decreases current consumption.
Income and substitution effects work in opposite
directions.
gradient
(y0, y1 )
lower utility

saver
If the household continues to save after the rate cut it is

worse off; this has the same effect as a decrease in
income in which case
B
C
gradient
- (1 + r)
c0
Income and substitution effects on
c0 work in the same direction.

6
Someone who saves

before the rate cut and
gradient - (1 + r )
c1
A
borrows after the rate cut
The substitution effect is the same as for a borrower; it

may have higher or lower

utility after the rate cut.
If the household saves before the rate cut and borrows

after it may be better off or worse off after the rate cut.
Here she has lower utility
(y0, y1 )
at the low rate.
gradient
- (1 + r)

saver
c0
Income and substitution effects on c0

work in the opposite direction.
If the household is better off the income effect increases

current consumption, income and substitution effects
work in the same direction.
If the household is worse off the income effect decreases
current consumption, income and substitution effects
work in opposite directions.
7. Different borrowing and lending rates

Banks make a profit on the difference between the rate at
which they borrow and the rate at which they lend or invest
their money.
7. Different borrowing
and lending rates
In the standard simple model competition in banking leads

to 0 bank profits so borrowing and lending rates are the
same.
Where borrowers from banks may default the interest rate at
which banks lend has to be higher than the rate at which
they borrow if banks are not to make losses.
In 2008 banks were unwilling to lend to each other, due to
worries about default due to bank failure.
Sometimes it is possible to borrow at a low

rate and invest at a higher rate
at
date 1
If this can be done on an unlimited scale it can make

unlimited profits.
save
at date 0
at
date 1
borrow
(y0,y1 )
(y0,y1 )
Example, UK student loans, you can only borrow a

limited amount.
Carry trade. Borrow in Japanese yen, sell yen, buy
another currency to invest at a higher rate. Buy yen,
repay loan. But
How risky is your investment?
What about exchange rate risk?
at
date 1
neither borrow
nor save
(y0,y1 )
different
borrowing
and
lending
rates
at date 0
borrow at rate r,
steeper budget
line gradient 1 + r
save at rate r < r
at date 0
flatter budget line

gradient 1 + r

7
at date 1
at date 1
(y0,y1 )
at date 0
Budget constraint with different borrowing and
saving rates and a credit limit.
In the simple model

borrowing and saving
interest rates are the
same and there is no
credit limit. It is optimal
to choose the income
stream with the highest
pdv regardless of
preferences.
If there are different
interest rates for
borrowing and lending or
credit limits the choice
between income streams
depends on preferences,
(y0,y1 )
(y0,y1 )
at date 0
Have you got both debt and savings?

In this model if the rate at which you borrow is higher
than the rate you get on saving it is optimal to
use your savings to pay off your debt.
Possible exception you can continue to borrow now but
might not be able to borrow in the future.

8
2.1 Firms and Costs

1. The objective of the firm
2. Firms and Markets
2. Opportunity cost
3. Production function
4. Cost functions
2.1 Firms and costs
5. Economies of scale
6. Returns to scale
1. What is the objective of the firm?
1. What is the objective of

the firm?
Some firms have mission statements
They also have annual reports in which they set out objectives.
Profit maximization: the standard assumption

The objective of the firm is to maximize profits.
Profits = revenue opportunity cost.
This is the objective of the firms shareholders who control the
Reasons for for assuming profit

maximization
Competition in the product market
(if a firm can at best make zero profit it has to maximize
profits to stay in business.)
firm.
The simple model of the firm starts with either a production
function or a cost function.
It ignores the fact that a firm is an organisation run by people
who have individual objectives and agendas.
It ignores conflicts of interest between investors and senior
Incentives generated by the financial sector and the market for

managers.
(reputation, reward packages, takeover threat).
As a reasonable first approximation, particularly
in areas where conflicts of interest seem unimportant.
managers.
2.1 Firms and costs

1
Risk taking
Risk management
You have made a a trade that has made a big loss.

It has not yet been observed.
You will be sacked when the loss is observed.
You could make a second trade with a small probability of
making a very high return and a large probability of a big
Managing the risks of long term decisions
Some incentive schemes, e.g. bonuses and stock options

may
loss.
The situation is greatly complicated by asymmetric
If you make the high return you keep your job.
information and incomplete contracts.
If you make the big loss you loose your job.
This is the focus of much research in corporate finance.
Which economists assume profit maximization?
Research in Corporate Finance has focused

on relations between
Teachers of standard intermediate microeconomics courses.

Usually industrial economists at the research level.
Usually not corporate finance economists at the research level.
Providers of finance and senior managers

providers of different types of finance:
equity (shares), debt
In this course we want to understand how the nature of
different groups of equity holders:
competition in the product market affects prices and quantities.
family and others,
Assuming profit maximization we get good insights from this.
financial institutions, other firms

different types of debt, banks, bonds
Introducing the complications I have just discussed would make

this analysis much more complicated.
Firms get profits at many dates.
Measuring revenue can be problematic
Thestandardassumptionisthatfirmsmaximize
Tesco estimates of profits March August 2014
the
1,100 million
presentdiscountedvalueV ofprofits,
850 million
t profitsatdatet
Profits = revenue - costs
1
(1 r )
2
(1 r )
2014,
September
2014
principally due to the accelerated recognition of
3
(1 r )
August
......
commercial income and delayed accrual of costs
2.1 Firms and costs

2
Profits = revenue - costs

Costs are complicated if the firm has durable equipment,
intellectual property or long run contracts
economic cost (opportunity cost) can be different from cash
2. Opportunity Cost
flows to providers of inputs and from accounting costs.
2. Opportunity Cost
Defined as the value of an input in its best alternative use.
Opportunity cost to you of studying

The cost to you of studying at LSE is
fees (if you pay them yourself)
Sometimes called economic cost.

+ the opportunity cost of your time, that is the
For something a person or firm is buying now the
opportunity cost is the current price.
the value of your time in the best alternative use.

If the best alternative use of your time is employment
the salary you would earn if you were not at LSE is an
opportunity cost.
Opportunity cost to you of running your own

business
The opportunity cost to an entrepreneur of running the
business is
Opportunity cost to a firm

What is the opportunity cost of using an input from inventory
e.g. oil?
at least as big as the salary she could earn elsewhere.
If there is a more valuable alternative to employment the

opportunity cost is higher.
What is the opportunity cost of land, buildings and equipment?

in the short term?
in the long term?
2.1 Firms and costs

3
Opportunity cost
can be impossible to measure accurately,
is always useful to think about when making decisions.
The Cost of Capital
Accountants need to produce precise numbers.
The Cost of Capital
If you save p at interest rate r for 1 year you have cash

(1+r)p next year.
The simplest model.

Suppose you must buy capital goods 1 year in advance at
which time you pay a price p.
If you buy the capital good, used it and sold it, you would
have cash p next year
After using the capital good you can if you wish sell it for
price p. You then have p in cash.
The opportunity cost is rp where r is the interest rate.
The difference in the amount of cash is the opportunity cost

of capital rp.
Why?
If p = 1 the cost of capital is r.
This is why textbooks use notation r for the cost of capital
More realistically the cost of capital depends

on
What matters for opportunity cost is where you are now.
rate of physical or technical deterioration
People know this
How you got there is irrelevant.
continuation or not of technical support
Look to the future, you cant change the past.
technical obsolescence
Let bygones be bygones.
changes in the price of the capital good

installation and transaction costs
taxes
Dont pour good money after bad.

But we find it hard to do.
What you think and feel about the past is part of the
present.
2.1 Firms and costs

4
3. Production functions
3. Production functions
Marginal Products
In this part of the course we assume the firm produces one
output from two inputs, capital K and labour L.
With more maths this can be generalized to many inputs
and outputs.
Output q = f(K,L), f(K,L) is the production function.
Standard assumptions on marginal products

output
gradient = f = MPL
L
f(K*,L)
f
K
marginal product of capital MPK
f
L
marginal product of labour MPL
Law of Diminishing Marginal Returns

A general statement of the previous slide.
If one input increases while the others are held constant the
fixed
0
labour L
f
With fixed K output increases as L increases MPL =
> 0.
L
MPL decreases as L increases,
.
marginal product of that input falls as output expands.
Example: labour in agriculture

with a fixed amount of land, seed, tractors etc beyond a certain
with fixed K output is a concave function of L,
point the extra output from increasing labour starts falling.
Similarly MPK > 0, MPK decreases as K increases.
Isoquants
K
Consumer Theory: Marginal Rate of

Substitution (MRS)
q=3
x2
MRS = - gradient of indifference curve
q=1
q=2
L
Isoquant q = 1 shows all the combinations of inputs for
which f(K,L) = 1.
u
x1
MRS
u
x 2
x1
marginal utility of good 1

marginal utility of good 2
Changing the numbers on the isoquants changes output

obtainable from inputs.
This matters. Different from indifference curves.
2.1 Firms and costs

5
Producer Theory: Marginal Rate of

Technical Substitution (MRTS)
4. Cost functions
K
MRTS = - gradient of isoquant.
Using the same maths as consumer theory
f
MRTS L
f
K
marginal product of labour

marginal product of capital
Standard assumption, decreasing MRTS,

moving along an isoquant by increasing L the MRTS
decreases.
4. Cost Functions
The cost function c(v,w,q) is the minimum cost of producing
output q using capital K and labour L with prices v and w.
This definition assumes that all inputs can be varied, later
we will call this a long run (as opposed to short run) cost
function.
Worked Examples on the website

1. Notes on Cost Functions
2. Conditional factor demand and cost functions with a
Cobb-Douglas production function.
3. Conditional factor demand and cost functions with a
fixed proportion production function.
Detailed discussion of the
technical material.
Sometimes this is called a total cost function (as opposed

to a marginal or average cost function).
Worked Examples on the website

4.
Graph sketching for cost functions
Parallels with Consumer Theory

Given a production function f(K,L) the cost function c(v,w,q) is the
minimum cost of producing output q using capital K and labour
A level material
Essential for EC201
L with prices v and w.
Given a utility function u(x1,x2) the expenditure function E(p1,p2,u)

is the minimum cost of obtaining utility u by consumption of x1
and x2 with prices p1 and p2.
2.1 Firms and costs

6
You find the expenditure function by
You find the cost function by
finding the levels of x1 and x2 that minimize
finding the levels of K and L that minimize
p1x1 + p2x2 subject to the constraints u(x1,x2) u and
vK + wL subject to the constraint f(K,L) q and non
non negativity constraints x1 0, x2 0.
negativity constraints K 0, L 0.
Write these as
Write these as
x1 = h1(p1,p2,u) and x2 = h2(p1,p2,u). This is
K(v,w,q) and L(v,w,q). Note they depend on (v,w,q).
compensated demand. Note compensated demand

depends on (p1,p2,u).
They are sometimes called conditional factor demand.

The cost function is
The expenditure function is

E(p1,p2,u) = p1h1(p1,p2,u) + p2h2(p1,p2,u)
c(v,w,q) = vK(v,w,q) + wL(v,w,q).

It depends on (v,w,q).
It depends on (p1,p2,u).
Checks
The Mathematics of Finding Cost Functions

This is exactly the same as the mathematics of finding
expenditure functions.
There are worked examples on website for a Cobb-Douglas
Check that increasing inputs increases output.

This is the equivalent of nonsatiation in consumer
theory.
Production Function q = K3/5 L2/5 and a fixed proportions

Production Function q = min( 3K,L).
For the exam and written homeworks answers need to be
Check for convexity.

This is exactly the same as in consumer theory.
written out in detail, as in the worked examples.
Parallels with Consumer Theory

For the Cobb-Douglas production function f(K,L) = K 3/5 L 2/5
K=(3w/2v)2/5 q
L= (2v/3w)3/5 q
minimize the cost of producing output q when price of

capital = v and price of labour = w.
The cost function is
c(v,w,q) = v (3w/2v)(2/5) q + w (2v/3w)(3/5) q
which is equal to
Given a production function f(K,L) the cost function c(v,w,q) is

the minimum cost of producing output q using capital K and
labour L with prices v and w.
Given a utility function u(x1,x2) the expenditure function
E(p1,p2,u) is the minimum cost of obtaining utility u by
consumption of x1 and x2 with prices p1 and p2.
Thus the cost function has exactly the same mathematical
properties as the expenditure function.
[(3/2)2/5 + (2/3)3/5] w2/5 v3/5 q.
2.1 Firms and costs

7
Properties of the Expenditure Function

E(p1,p2,u)
1.
Properties of the Cost Function c(w,v,q)
Increasing in utility
1. Increasing in output q
2.
Homogeneous of degree 1 in prices.
2. Homogeneous of degree 1 in
3.
Non-decreasing in prices.
4.
Concave in prices.
5.
Shephards lemma
From the
consumer
theory slides
input prices v,w

3. Non-decreasing in input prices.
4. Concave in input prices.
5. Shephards lemma for cost functions
E(p 1 ,p 2 , u )
p 1
h 1 (p 1 ,p 2 , u )
compensated demand
isocost line gradient -w/v
Cost
minimization at
a tangency
At the cost minimising

point A1 MRTS = w/v
c(v,w,q1)/v
c(v,w,q)
v
K(v,w,q)
c(v,w,q)
w
L(v,w,q)
The maths of the

cost function is
exactly the same
as the maths of
the expenditure
function so the
cost function has
the same
properties as the
expenditure
function.
If q2 > q2
K
c(v,w,q2)/v
producing q2 costs
more than producing q1
so the cost function
c(v,w,q2) > c(v,w,q1)
c(v,w,q1)/v
the cost function is

increasing in output.
A2
A1
A1
f(K,L) = q2
isoquants
c(v,w,q1)/w
f(K,L) = q1
L
If all input prices are multiplied by k

> 0 the isocost line doesnt change.
The input combination that
minimizes the cost of producing q
does not change, but the cost of
the inputs is multiplied by k.
c(v,w,q1)/v
The cost function is homogeneous

of degree 1 in input prices.
A1
0
c(v,w,q1)/w
If the price of K rises

from vA to vB, the
gradient of the isocost
line changes.
Cost minimizing inputs

for output q1 move
from A to B with less K
and more L.
c(vA,w,q1)/vA
c(vB,w,q1)/vB
c(v,w,q2)/w
f(K,L) = q1
L
The cost of producing

q1 increases.
B
isoquants
c(v,w,q1)/w
f(K,L) = q1
L
f(K,L) = q1
0
c(vA,w,q1)/w
c(vB,w,q1)/w
2.1 Firms and costs

8
5. Returns to scale
Returns to scale are a property of the production function.
Returns to scale
& economies of scale
They have implications for the cost function which

are important for understanding the competitive structure of an
industry.
6TP
Returns to scale with

Cobb Douglas production function
f(K,L) = KaLb
m>1
If f(mK,mL) = mf(K,L) there are
returns to scale.
a > 0, b> 0, m > 0, implies f(mK,mL) = (mK)a(mL)b = maKambLb
If f(mK,mL) < mf(K,L) there are
= ma+bKaLb = ma+b f(K,L)
returns to scale.
Assume m > 1. If a +b = 1, ma+b = m, so f(mK,mL) = m f(K,L)

Returns to scale?
If f(mK,mL) > mf(K,L) there are
If a +b > 1, ma+b > m, so f(mK,mL) > m f(K,L)
returns to scale.
Returns to scale?
If a +b < 1, ma+b < m, so f(mK,mL) < m f(K,L)
If this seems strange think about m = 2.
Returns to scale?
Division of Labour
6. Economies of Scale
Adam Smith, Wealth of Nations, 1776
Returns to scale are a feature of a production function.
Description of a pin factory
Economies and diseconomies of scale are a feature of a cost
One man draws out the wire, another straightens it, a
function.
third cuts it, a fourth points it, a fifth grinds it at the
Average cost AC = total cost/output

There are economies of scale if AC
There are diseconomies of scale if AC
top for receiving the head; to make the head

with output.
with output.
Increasing returns to scale in the production function imply

of scale, more on this later.
There other sources of economies of scale.
requires two or three distinct operations; to put it on

is a peculiar business, to whiten the pins is another;
it is even a trade by itself to put then into the
paper...... ten persons can make upwards of 48 000
pins in a day, one person certainly cannot make
4 800 in a day.
2.1 Firms and costs

9
Volume Effects
Division of labour on a
production line 1977
For ships, pipe lines, planes, chemical works, blast

furnaces, etc.
capacity depends on volume.
A more automated
production line 2009.
cost depends on surface area,

increasing size reduces cost per unit of output.
Note the change in

technology and the
substitution of capital for
labour.
EMMA MAERSK
Length: 379 meters
Capacity: 11,000 containers
each weighing 14 tons
Source: Wikipedia
Getty Images
Argument that there cant be decreasing returns to

scale
Suppose that a factory produces 1 million cars.
If the firm wants to produce 2 million cars it could
Where might this argument be wrong?

There might be some fixed input that cannot be doubled,
example: a manager with experience and expertise in running
the factory.
set up an identical factory that produces another 1 million cars

in exactly the same way.
So doubling inputs either doubles output or more than doubles
output.
Thus there must be constant or increasing returns to scale.
There may be managerial diseconomies of scale.

Management difficulties grow faster than the number of people.
A very small organisation may be able to coordinate during a
coffee break.
Bigger organisations need formal structures.
The transition from a small to a large organisation can be
difficult.
Economies of Scope
Other Influences on Costs

One firm producing a range of related products has cost
The production function model assumes a single
unchanging product and that the amount of output from a
given input is fixed by technology.
This may not be so owing to
economies of scope
learning by doing
Costs before production starts,
e.g. research and development, website
more on this later.
advantages over firms producing single products.

Are there economies of scope in running supermarkets and
home deliveries of groceries?
Ocado does not have supermarkets
and and originally delivered Waitrose
products. Runs from dedicated
warehouses.
Tesco , goods for delivery were
handpicked in stores.
Supermarkets In the UK Getty Images
2.1 Firms and costs

10
Tesco and Ocado, September 2012

Tesco plans to open more dark stores effectively stores
without customers that fulfil grocery orders placed online.
Tesco has four dark stores serving London and the southeast.
Ocado shrugged off the threat from Tescos expansion.

This is actually an opportunity for us.
Learning by Doing
As firms build experience in
producing a product the average
cost of producing the product
falls.
Examples, silicon chips
aeroplanes
Concord Production Getty Images
Financial Times, September 24 2012
2.1 Firms and costs

11
2.2 Profit maximization and costs for a price

taking firm
2.2 Profit maximization

and costs for a price
taking firm
1. Definition of price taking
2. The shutdown and output rules
3. Profit maximization by a price taking CRS firm
4. Cost curves, profit maximization and supply with

decreasing returns to scale
5. Cost curves with increasing returns to scale
6. Impossibility of price taking with economies of scale at all
1. Definition of price taking

A firm is a price taker if nothing it can do affects the prices it
levels of output
7. Cost curves and supply with a u-shaped average cost
pays for inputs and outputs.
In particular its output does not affect its output price.
curve
Price taking does not mean prices do not change.
8. Long run and short run costs and supply
Price taking is plausible if the firm has a small market
share.
2.
:
The shutdown & output rules
The shutdown rule and average cost.
2. The shutdown & output

rules
c(v, w, q) is total cost
so
c(v, w, q)
is average cost, AC
q
revenue pq, total cost c(v, w, q ) profits pq c(v, w, q)

Why we need to know
about average costs.
The shutdown rule
The shutdown rule
If the cost of producing 0 is 0, i.e. c(v, w,0) 0

the firm can make 0 profits by
The firm shuts down (does not produce)

if price AC at all levels of output. .
producing 0 so if the firm maximizes profits at q 0

profits pq c(v, w, q ) 0 so
c(v, w, q )
q
The shut down rule implies that

if a profit maximising firm
produces q 0 then price AC.
The output rule and marginal cost

Profit revenue - cost R(q) - c(v, w, q)
R c
marginal cost
If marginal revenue
q q
then increasing output q increases profits.
R c
marginal cost
q q
then increasing output q decreases profits.
If marginal revenue
Profit maximization at q 0 implies
Why we
need to know
about
marginal
costs.
R c
marginal cost.
q q
But marginal revenue marginal cost
does not necessarily imply profit maximization.
marginal revenue
You will see an example of this shortly.
Getting this wrong is a common

exam mistake.
What is marginal revenue?

c( v, w, q )
is marginal cost, MC
q
By definition for a price taking firm

marginal revenue = price and does not vary with output
The first order condition for profit maximizati on

implies that if the firm
maximizes profits at q 0
R
c( v, w, q)
q
q
output rule, marginal revenue MC
Why we
need to know
about
marginal
costs.
(note perfect competition implies price taking)

For a monopoly marginal revenue depends on the firms
output
For a firm in an oligopoly marginal revenue depends on

the firms output & the output of other firms.
The relationship between

marginal cost &
average cost
Producer theory worked example 4 on the

website
A level material
Graph sketching for cost functions
Essential for EC201
When you have formulae for MC & ACyou must use the
formula when sketching a graph of MC & AC.
Also use what you know about the relationship between

MC, AC and increasing and decreasing AC.
AC curves may not be u-shaped
AC
c(v, w.q)
q
Total marginal and average costs
Using the quotient rule the derivative of AC is
1 c(v, w, q )
(
q c(v, w, q ))
q
q2
1
(MC AC)
q
average cost at q1
So average cost increases when

MC
AC
average cost decreases when

MC
AC
When MC
AC the AC function has a critical point
(maximum, minimum or point of inflection.)
Total cost c(v,w,q).
AC
gradient of
marginal cost
c1
MC
gradient
0
q1
TP
3. Profit maximization by a price taking CRS

firm
3. Profit maximization by
a price taking CRS firm
Cost curves with CRS

The production function
K
q=3
q = f(K,L) has constant returns

to scale if
q=2
for all positive numbers m

mf(L,K) = f(mL,mK).
q=1
0
L
isoquants with CRS
Under constant returns to scale CRS the optimal ratio of

inputs (e.g. the capital labour ratio)
is the same at all levels of output.
Total cost function from a CRS

production function
Multiplying inputs by 2 multiplies output by 2.

c(v,w,q),
So it costs twice as much to produce 2 units of output as it
gradient = MC = AC
costs to produce 1 unit of output.

More generally with constant returns to scale, given input
prices v and w
output q
c(v,w,mq) = mc(v,w,q).
Marginal cost = average cost
vary with input prices v and w but not with output
Marginal and average cost from

a CRS production function
A firm is a price taker if nothing it can do changes the
price p at which it sells.
/unit
Profits = pq cq = (p c)q
c = MC = AC
c
If p > c, so (p c) > 0 increasing q always increases

profits, there is no profit maximizing output.
If p < c so (p c) < 0 the firm makes losses at all q > 0 so
produces 0.
output q
Marginal cost = average cost =c
p = c is the only price at which a price taking firm with
varies with input prices v and w but not with output
Supply by a price taking CRS firm

AC is not u-shaped
If p = c the firm makes 0 profit at any q.

constant returns to scale has a profit maximum at q > 0.
Supply by a price taking CRS firm

p
horizontal supply curve

c
horizontal supply curve

c
output q
output q
4. Cost curves, profit maximization and supply

with decreasing returns to scale
4.Cost curves profit

maximization & supply
with decreasing returns
to scale
Under decreasing returns to scale DRS multiplying inputs

by 2 multiplies output by
So it costs
than twice as much to produce 2 units of
output as it costs to produce 1 unit of output, so

c(v,w,2q)
Cost curves with DRS
c(v,w,mq)
m c(v,w,q)
so AC at output mq = c(v,w,mq)
c(v,w,q) = AC output q
mq
AC
2c(v,w,q).
Total cost function from a DRS

production function
More generally with decreasing returns to scale, given input

prices if m > 1
than 2.
with output.
0
output q
MC = gradient of tangent AC = gradient of chord.

MC
as q increases.
AC
as q increases.
MC
AC and MC from a DRS production function.
AC.
If MC is increasing p = MC gives a profit

maximum: intuition
/unit
MC
MC AC
everywhere
/unit
MC
p*
AC
MC
as q increases
q
Average cost is not u-shaped
AC
as
q increases
q1
output q
When q < q1 MC < p*, increasing output by 1 unit increases

costs by MC and revenue by p*. As p* > MC this increases
profits.
When q > q1 MC > p*, increasing output by 1 unit increases
costs by MC and revenue by p. As p < MC this decreases
profits.
Profit maximization with price taking and a

DRS production function
If MC is increasing p = MC gives a profit

maximum: calculus
(q) = pq c(q) ( c(q) = total cost )
First order condition for profit maximization
/unit
MC =
supply
(q) = p c(q) = price marginal cost = 0.

p
AC
As c(q) is the derivative of c(q) = MC increasing marginal
cost implies that c(q) > 0
The second derivative (q) = c(q) < 0 so
(q) = pq c(q) is a concave function of q.
The first order conditions give a maximum.
0
q*
q
With DRS MC is increasing and MC > AC
If p = MC then p > AC so the shutdown rule is satisfied.
The MC curve is the supply curve.
5. Cost curves with IRS

Under increasing returns to scale IRS multiplying inputs by 2
multiplies output by more than 2.
Price taking with

economies of scale at
all levels of output is
impossible
So it costs less than twice as much to produce 2 units of

output as it costs to produce 1 unit of output, so
c(v,w,2q) < 2c(v,w,q).
More generally with increasing returns to scale, given input
prices if m > 1
c(v,w,mq) < m c(v,w,q)
so AC at output mq = c(v,w,mq) < c(v,w,q) = AC at output q
mq
AC decreases with output.
Total cost function from an IRS

production function
Total cost function from an IRS

production function
output q
MC = gradient of tangent < AC = gradient of chord
output q
MC = gradient of tangent < AC = gradient of chord

MC decreases as q increases
AC decreases as q increases,
there are economies of scale
AC and MC from an IRS production function
6. Price taking with economies of scale at

all levels of output is impossible
MC < AC
everywhere
/unit
/unit
Because AC is
decreasing either
MC (decreases)
as q
AC
AC (increases)
as q
MC
p < AC for all q or

AC
p
0
p AC for large q.
MC
output q
Here p < AC for all q, a price taking firm cannot make a profit.
output q
The shut down condition is never satisfied.

AC is not u-shaped
Price taking with economies of scale at all

levels of output is impossible
/unit
Because AC is
decreasing either
p < AC for all q or

AC
p AC for large q.
Costs in R&D intensive industries

For computer software, pharmaceuticals, cars ... there are major
research and development (R & D) costs.
These are fixed (they do not vary with output).
MC
0
output q
Here p AC > MC for large q, there is no profit maximum.

A price taking firm can increase its profits indefinitely.
The firm will expand until it has a large market share and is not
a price taker.
Before R&D starts its costs are opportunity costs. If they are too
high the product is not profitable.
Once the R&D is done it is a sunk cost and not an opportunity

cost. It is irrelevant to production decisions.
Price taking with economies of scale is
A simple model of costs in an R&D intensive

industry
Total cost = F + cq
impossible
/unit
AC
F = R & D cost, opportunity cost before development not an

opportunity cost after development
very
important
c = constant marginal cost

MC
AC = F/q + c decreases at all levels of output.
This firm has economies of scale at all levels of output, it
cannot be a price taker.
q
R & D intensive industries with a fixed development
cost & constant marginal cost have economies of
scale at all levels of output.
7. Cost curves and supply with a u-shaped

average cost curve
7.Cost curves and supply

with a u-shaped average
cost curve
Often assumed for perfect competition.

For small q there are economies of scale AC falls.
For large q there are diseconomies of scale, AC rises.
total cost
function
total cost
function
output
AC falls when
MC falls when
AC rises when
MC rises when
AC has a minimum at
MC has a minimum when
At the minimum AC = MC.
The chord is
TP
Marginal and average cost with a u shaped

average cost curve
/unit
output
TP
Profit maximization by a price taking firm with a

u average shaped cost curve
/unit
MC
MC
AC
AC
min AC
p
output q
output q
MC deceases when q < a and increases when q > a.
Here p < min AC
AC decreases when q < b and decreases when q > b.
the firm makes losses at all q > 0,
At b AC has a minimum so its derivative is 0 implying that

MC = AC.
the shut down condition cannot be satisfied at q > 0 so q = 0.

/unit
p
q2 maximizes profits.
MC
AC
At q2 p = MC.
At q1 p = MC, this is a
local minimum of
profits.
min AC
0 q1
b q2

/unit
p
q2 maximizes profits.
MC
AC
min AC
output q
At q2 p = MC.
At q1 p = MC, this is a
local minimum of
profits.
0 q1
b q2
output q
Here p > min AC. The firm can make profits.

When q < q1 p < MC so increasing q decreases profits.
When q1 < q < q2 p > MC increasing q increases profits.
Price = MC at q1 but q1 does not maximize profits.
When q2 < q p < MC increasing q decreases profits.

/unit

supply curve
MC
AC
AC
MC
p = min AC
output q
The supply curve

is the upward
sloping part of the
MC curve where
MC AC.
output q
Here p = min AC. The firm can make at most 0 profits.
When p < min AC the firm produces 0.
It does this by producing b or producing 0.
When p = min AC the firm produces either 0 or b.
At the output b that minimizes AC p = MC = AC.
When p > min AC the firm produces on the point where

p = MC MC is increasing and MC > AC.
Both the shut down and output conditions are satisfied.
8. Long run and short run costs and supply

See the producer theory worked examples 1, 2 & 3.
Up to now, all inputs are chosen at same time.
Now there are two periods, the planning period and the
8. Long run and short run

costs and supply
production period.
Capital is fixed in the planning period & paid for in the
production period.
Labour is chosen and paid for in the production period.
If the firm knows output and input prices in the planning
period this makes no difference.
K and L are chosen to minimize total cost c(v,w,q).
isoquant
q2
K
c2/v
long run
expansion
path
c1/v
ci = LRTC of
output qi
LRTC = long run
total cost
(L2,K2)
(L1,K1)
q1
0
isocost line
gradient w/v
L
If in the planning period the firm knows output q and input

prices w & v in the production period it chooses the cost
minimizing point on the long run expansion path.
Short Run Costs

The firm installs K in the planning period, when it is
uncertain what input prices will be and how much it
will produce in the production period.
The cost s(v,w,K,q) of production depends on input
prices v and w, capital K and output q. It is called
short run total cost (SRTC).
Contrast LRTC c(v,w,q) does not depend on K.
Total inputs (Li,Ki) total cost LRTC = ci when output is qi.
Flexibility can never be bad in standard

economics models
If given the choice between fruit & cake you chose cake
the more restricted menu is worse.
isoquant
q2
K
c2/v
long run
expansion
path
c1/v
K1
If given the choice between fruit & cake you chose fruit
the more restricted menu is no better.
isocost line
gradient w/v
q1
0
L1
If in production period K = K1, input prices are w, v and

output is q1 the minimum cost of producing q1 is c1.
K
s2/v
c2/v
With capital K1
isoquant
q2
c2 = LRTC of
output q2
isocost line
gradient w/v
at output q1
s2 = SRTC of
output q2
SRTC = short run
total cost
K1
s2 > c2
K1
q1
q1
0
L2
If in production period K = K1, input prices are w, v and

output is q2 the firm has to use L2 labour.
Total cost s2/v > c2/v.
q0
q2
LRTC = SRTC.
long run
expansion At other outputs
path
LRTC < SRTC.
short run
expansion path
isoquant
L
With capital fixed at K1 in the planning period the firm is on the

short run expansion path in the production period.
At output q1 the firm is on the long run expansion path.
At other outputs the firm is not on the long run expansion path.
10
LRTC SRTC always

With a minimization problem you can often do better and
never do worse with more flexibility.
Short run and long run problems of getting to LSE

Short run problem, you have no bike, find the quickest way
of getting to LSE.
Long run problem, find the quickest way to LSE and get a
Long run (choose K and L)
bicycle if that is quickest.
more flexible than
If the bus is quickest the SR and LR problems give the
short run (choose L, K fixed).
same minimum time.
So either LRTC < SRTC or LRTC = SRTC
If the bike is quickest the LR problem gives a shorter

minimum time
So either long run time < short run time
or long run time = short run time.
Finding SRTC with a Cobb-Douglas

Production Function
Long run and short run

costs and supply with a
Cobb-Douglas production
function
Rearrange the production function equation

q = K3/5 L2/5 to get L = q5/2 K-3/2
So to produce q units of output with K = K* requires
q5/2 K*-3/2 units of labour and at prices v and w costs
SRTC = w q5/2 K*-3/2 + vK*.
I have already found that when K and L are chosen
freely
LRTC depend on v,w,q
LRTC = [(3/2)2/5 + (2/3)3/5] w2/5 v3/5 q
Total average and marginal long and

short run costs
Long run total cost LRTC
Long run average cost LRAC

Long run marginal cost LRMC
Short run total cost SRTC
Short run average cost SRAC

Short run marginal cost SRMC
c( v, w, q )
c( v, w, q )
q
c( v, w, q )
q
s( v, w, K *, q )
s ( v, w, K *, q )
q
s ( v, w, K *, q )
SRTC depend on
v,w,K*,q
With the Cobb - Douglas production function q K 3 / 5 L2 / 5

LRTC [(3 / 2) 2 / 5 (2 / 3) 3 / 5 ]q
so
LRAC LRMC [(3 / 2) 2 / 5 (2 / 3) 3 / 5 ]

The production function has constant returns to scale,
long run average and marginal costs are equal and do
not depend on output.
SRTC w q 5 / 2 K* -3 / 2 vK* so
Producer theory worked

examples 1 & 2
SRAC w q 3 / 2 K* - 3 / 2 vK*q 1
SRMC w(5 / 2) q 3 / 2 K* - 3 / 2
Short run average and marginal costs depend on output.
11
SRTC
LRTC
Marginal cost is the gradient
of the total cost curve.
At q* SRTC and LRTC are
tangent so have the same
gradient.
q*
At q* the fixed level of capital K* is at the cost minimizing level

and LRTC = SRTC so LRAC = SRAC.
As SRTC LRTC for all values of q, SRAC LRAC for all q.

LRTC [(3 / 2) 2 / 5 (2 / 3) 3 / 5 ]q
LRAC LRMC [(3 / 2)
2/5
so
(2 / 3) 3 / 5 ]
The production function has constant returns to scale,

long run average and marginal costs are equal and do
not depend on output.
SRTC w q 5 / 2 K* -3 / 2 vK* so
SRAC w q 3 / 2 K* - 3 / 2 vK*q 1
SRMC w(5 / 2) q 3 / 2 K* - 3 / 2
Short run average and marginal costs depend on output.
LRTC and SRTC are tangent at q* where LRMC = SRMC.
With a good grasp of A level calculus and

indices and the economics, you can establish
that
SRMC
/unit
LRMC = LRAC is a horizontal straight line.

SRAC
SRMC is 0 at q = 0 and increasing
LR & SR MC
& AC with this
CobbDouglas
production
function.
SRAC is u-shaped
(do this by seeing where SRMC > SRAC)
LRAC = LRMC
SRAC tends to infinity as q tends to 0

SRAC = SRMC= LRMC = SRAC at the point that
q*
At q = q* LRAC = LRMC = SRAC = SRMC
minimizes SRAC
See producer theory

worked example 4
At all values of q
SRAC LRAC
SRMC can be > or < LRMC
Fixed and variable costs

SRTC w q
5/ 2
K*
-3 / 2
vK*
short run variable cost w q 5 / 2 K* - 3 / 2 varies with q

short run fixed cost vK* does not vary with q
In the production period K * cannot be varied so is not
an opportunity cost.
The opportunity cost in the production period is
SRVC w q 5 / 2 K* -3 / 2
Short run average variable cost
SRAVC SRVC/q w q 3 / 2 K* -3 / 2 0
so SRAVC is increasing in q so SRMC SRAVC
short run variable cost SRVC w q 5 / 2 K* -3 / 2

As SRVC does not include K which is fixed in the short run
SRTC SRVC
12
Short run profit maximization

Output rule p = SRMC
Short run profit

maximization
Shutdown rule p SRAVC
SRMC is the derivative of SRTC and also of SRVC with

respect to q.
SRMC
/unit
SRMC
This firm has constant

returns to scale
SR supply
so LRMC = LRAC.
SRAC
Long run supply is a

horizontal line.
SRAVC
LRAC = LRMC
q*
Here SRMC > SRAVC for all q so short run supply is SRMC.
LRMC LR supply
p0
0
q*
This is a firm with CRS. If the price is p0 the firm is on

both its long run and short run supply curves at q* where
LRMC = SRMC.
The firm would plan to produce 0 if it knew in the

planning period that p < p0.
If the firm finds in the production period that
p >0 it produces q > 0 even if p < p0.
It makes an economic profit
whether it makes an accounting loss depends on the
The General Relationship

Between SR & LR
AC & MC
accounting treatment of capital.

whether it loses cash depends on debt servicing and
other obligations.
13
SRTC
The General Relationship Between SR & LR

AC & MC
gradient
SRAC LRAC for all q
= SRMC
= LRMC
SRAC = LRAC and SRMC = LRMC at q* where the capital
LRTC
stock K* would be chosen in the long run.
0
q*
q
At q* the fixed level of capital K* is at the cost minimizing level
and LRTC = SRTC so LRAC = SRAC.
In the next slides the firm plans for q*.
As SRTC LRTC for all values of q, SRAC LRAC for all q.

AT q* LRTC and SRTC curves are tangent so the gradient MC is
the same.
SRMC
SRAC
/unit
Firm would not plan

to produce if it knew
p < p0, but if it finds
0 < p at production
period it produces
even p < p0.
p
p*
LRAC
LRS
It makes economic
profits because K is
not an opportunity
cost in the short run.
LRAC
LRMC
SRAVC
p0
SRMC = SR supply
q*
q* output q
At q* LRAC = SRAC and LRMC = SRMC.

Here SRAVC < SRMC for all q, so SRMC is short run supply.
LRS SRS with U shaped AC
/unit
SRAC
SRMC
LRTC
SRTC
LRAC
gradient = LRMC = LRAC
p0
=SRMC =SRAC
0
q0
SRAVC
LRMC
Suppose the firm plans to produce at the level q0 of q that

minimizes LRAC. At this level LRAC = LRMC.
It installs the cost minimizing level of K for q0 so at q0
SRAC = LRAC and SRMC = LRMC.
q0
At q0 LRAC = LRMC = SRAC = SRMC

Here SRAVC < SRMC for all q, SRMC is SRS
LRS is LRMC curve above LRAC.
14
The firm would plan to produce 0 if it knew in the
planning period that p < p0.

If the firm finds in the production period that
LRS
p < p0 it produces q > 0
LRAC
p0
By assumption capital is not an economic cost, so the

firm makes economic profits.
SRMC = SR supply
0
Whether the firm makes an accounting loss depends

q0
output q
upon the accounting treatment of capital.

Whether it loses cash depends on its debt servicing
costs and other demands on cash flow.
Here SRAVC < SRMC for all q, SRMC is SRS

LRS is LRMC curve above LRAC.
LRS SRS with a fixed proportion production

function
Long run & short run

supply with a fixed
proportion production
function
See Producer Theory Example 3 on the website for details.

Production function f(K,L) = min( 3K,L).
I have already shown that at the long run cost minimizing point
L = q and K = q/3 so
LRTC = (w + v/3)q.
LRAC = LRMC = w + v/3.
Fix K = K* so q = min(3K*, L)
isoquant
q
3K*
L 3K* q = 3K*
Producer theory worked

examples 1 & 3
Fixed proportion
production function
output
q = min(3K,L)
long run
expansion path
K = L/3
isocost line
L < 3K*
q=L
0
0
Output as a function of L with K =
function q = min(3K, L).
3K*
K*
with the production
In the LR the firm produces where q = 3K = L, so K = q/3, K = L/3

LRTC = ( w+ v/3 )q.
15
isoquant q *= 3K*
K
Fixed proportion
production function
output
q = min(3K,L)
SRTC = wq + vK*
gradient w
long run
expansion
path
short run
expansion path
K*
LRTC = (w + v/3)q
gradient (w + v/3)
vK*
0
Long run and short

run total cost with a
fixed proportion
production function.
q*
In the SR with K fixed at K* the firm cannot produce more than 3K*.
SRTC and LRTC with q = min(3K,L), in SR capital fixed at K*,

q* = 3K*.
It can produce q K* by using labour q, SR total cost vK* + vq
It is impossible to produce more than q* in the SR.
Long run and short

run marginal &
average cost with a
fixed proportion
production function.
SRAC = w + vK*/q
/unit
Short run profit maximizing with fixed

proportions
In the short run it is impossible to produce more than q*.
If q q* the cost of producing q is wq + vK*.
LRAC = LRMC
w + v/3
w
= w + v/3
SRMC = w = SRAVC
q*
But K* is not an opportunity cost, so the firm maximizes

economic profits are (p w)q.
If p < w profits are maximized by producing 0.
SRAC, SRMC, LRAC, LRMC with q = min(3K,L), capital fixed

at K*, q* = 3K*.
If p = w any output between 0 and q* maximizes profits.

If p > w the firm maximizes output by producing the
maximum q*.
It is impossible to produce more than q* in the SR.
SRS
Short run and long run profit maximization

in the planning period
production period model
SRMC = w
LRS
w + v/3
w
If the firm can anticipate the future perfectly there is no

conflict between short and long run profit maximization.
q*
In the LR the firm has CRS, LRMC = SRMC = w + v/3, horizontal

long run supply.
When p = w + v/3 the firm is on its long and short run supply
curves at q*.
A price taking firm produces at a point where

LRMC = SRMC = p and LRAC = p = SRAC > SRAVC.
If the firm cant anticipate perfectly it produces where
SRMC = p and p SRAVC.
16
What does long run and short run mean?

Economics textbooks:
Is the short run long run

analysis useful?
in the long run all inputs are variable

in the short run some inputs are fixed.
Everyday sense
short run: a temporary state that wont persist
long run: the state things tend to revert to.
The meanings are similar if the economy tends to come
back to a steady state. Does it?
Heathrow:
Is the short run long run analysis useful?
Main London Airport since 1946

Yes, it captures the idea that unexpected things happen, and if
a firm has to decide on some inputs in advance it may
maximize economic profits by producing even if it makes an
economic loss.
15 miles (24 km) west of central London.

Aircraft approach over central London
Not on a major railway line.
Foggy
Limited space two runways, with a possible third.
But it is a simple model, and can be very misleading.

Would a major airport be built at Heathrow if there were not
one there already?
Will there be a major airport at Heathrow in 50 years?
17
3.1 Perfect competition and monopoly
3.1 Perfect competition

and monopoly
1. Introduction to industrial organization

2. Markets with a fixed number of price taking firms
3. Entry and exit
4. Entry & exit in an industry with a fixed proportion
production function
5. Entry & exit, short run & long run costs, with a u-shaped
average cost curve
6. Firms with different costs & economic rent
1. Introduction to industrial organization
7. Welfare economics of a tax with supply and demand
In EC201 we work with simple models
8. Welfare economics of a subsidy with supply and demand
One homogeneous good of known quality, no product

differentiation
9. Monopoly
One price for all buyers and sellers

(except when modelling price discrimination).
10. Cartels
No asymmetric information, buyers know the

characteristics of the good and the price charged by
each firm
11. Price discrimination
No transactions costs, e.g. search, negotiation, legal fees,

bid ask spread .
largely static models

partial equilibrium
Apart from the discussion of economic rent we model
one market at a time,
general equilibrium next term
Market structures
1. perfect competition price taking firms
2. monopoly one firm that affects price
3. oligopoly a small number of firms that affect
price
3.1 Perfect competition & monopoly
2. Markets with a fixed

number of price taking
firms
2. Markets with a fixed number of price

taking firms
Standard models of perfect competition

Withoutentry
andexit
key assumption is price taking

Short runsupply
curves
nothing a firm does affects the prices it gets for output and
pays for inputs.
Withentry
andexit
Shortrun
Long runsupplycurves
Longrun
nothing a purchaser does affects the price it pays for

Implicit assumption:
output.
See Industrial
Organization Worked
Example 1, Perfect
Competition
Firms renew their capital stock in the time span over

which entry & exit are possible.
EC201 models of perfect competition

Withoutentry Withentryand
andexit
exit
No distinction
betweenlongand
shortrunsupply
Marketswitha
fixednumberof
pricetaking
firms
Distinction between
longandshortrun
supply
When is price taking plausible?

A homogeneous good, all firms produce an identical product.
Large number of buyers and sellers each with a small market
share.
Everyone can observe prices.
Averysimple
modelofentry
andexit
More sophisticated model

Existing firms investment decisions depend on their
current capital stock.
New entrants are free to choose their capital stock.
Note
Price taking does not imply that prices do not change.
Price taking does imply that an individual buyer or seller
cannot do anything to change prices.
Entry and Exit?

We have already looked at profit maximizing by a price
taking firm.
We now look at an industry with a fixed number of price
taking firms assuming entry and exit is impossible.
p = MC follows from
p = AC, follows from
Profit maximization implies firms minimize
Then we look at an industry where entry and exit is

possible.
Profit maximization does not imply that firms minimize
How to solve this model
Markets with price taking firms

Assume for now a fixed set of firms price taking profit
maximizing firms.
p1
p1
In an equilibrium with a fixed set of price taking firms

0
each firm supplies a profit maximizing level of output

given input and output prices
q1
firm
The market clears, that is

supply = demand.
Q1
industry
industry supply = horizontal sum of firm supply
The next slides show how to solve the model.
Find industry price p1 and quantity Q1 by intersection of supply

and demand on industry diagram.
Supply and demand are shown as straight lines for

simplicity only.
Given the industry price find firm output q1 on firm diagram.
Shift in supply
Increase in demand
S1
s
p1
D2
p1
Demand curve shifts
0
(
, Q from Q1 to
Firm diagram firm quantity from q1 to
CRS
q1
0
industry
firm
Increase in input prices
Q1
MC.
Supply curves s1 & S1 move
demand at each price)
Industry diagram price from p1 to
Special
Case 1
0
Q2
Q1
s1
D1
0 q1
p1
p1
from p1 to
Industry diagram price
Firm diagram firm quantity
, Q
from q1 to
from Q1 to .
.
Equilibrium in a market with a fixed set of price

taking firms and CRS
Usually AC & MC depend on q and input

prices.
With constant returns to scale AC = MC and
does not vary with q. AC & MC do depend on
input prices.
industry
supply
demand
MC = AC = c
q
the firms
supply
firm
industry supply = horizontal sum of firms supply
0
When p = c the firm makes the same profits
(0) at any q. The firms output is not
determined.
Q
industry
industry quantity Q comes from intersection supply and demand

on the industry diagram.
Price = c.
Firm supply is not determined.
3. Entry and exit in perfectly

competitive markets
A very simple model of entry and exit

we work with the simplest model, a price change is
We now drop the assumption that the number of firms is
either temporary, there is no entry and exit, firms are
fixed. Now we allow for entry and exit.
on short run supply curves,

or permanent, in which case firms are on long run
In reality prices in the future are uncertain.
supply curves, firms enter if they can make economic
Price taking firms make decisions on entry, exit and capital
profits at the price, and exit if they make economic
investment depending on their beliefs about prices in the
losses at the price.
future.
There is no simple formula.
Market with a fixed number of price taking

firms
Equilibrium in a perfectly competitive market

with entry and exit
Each firm supplies a profit maximizing level of output given

input and output prices
Up to now
Each firm supplies a profit maximizing level of output given

input and output prices
The market clears, that is supply = demand.
The market clears, that is supply = demand.
For firms in the industry p min AC so the firms dont make

losses.
For firms in the industry p min AC so the firms dont make

losses.
New
For potential entrants p min AC so firms cannot enter and

make positive profits.
For potential entrants p min AC so firms cannot enter and

make positive profits.
If all firms have the same costs including potential entrants ( a

strong assumption) profits = 0 & p = min AC.
If all firms have the same costs including potential entrants

( a strong assumption) profits = 0 & p = min AC.
Modelling entry and exit with supply and

demand diagrams
c
All firms are identical including potential entrants
See Industrial
have constant marginal cost, no fixed cost
so MC = AC = c at all levels of output.
Industry price = c, industry quantity Q given by
supply = demand at price c.
Organization
Worked
Example 1,
Perfect
Competition
Firm quantity q is undetermined because firms make 0

profits at all levels of output.
A very special case
D
0
q
firm
Q1
industry
Industry price p = c = MC = AC, gives industry output Q1.

All firms make 0 profits, so the industry is in entry and exit
equilibrium.
The model does not give firm output q or number of firms n.
4. Entry & exit in an industry with fixed

proportion production functions
4. Entry & exit in an

industry with fixed
proportion production
functions
SRATC = w + vK*/q
SRS
Costs and
supply for a
firm
LRMC c =
LRS
LRAC = w + v/3
SRMC s = w
q*
All firm have the production function q = min(3K,L).

SR K fixed at K*, q* = 3K* = maximum possible output.
firm SRS
firm SRS
Industry SRS
Industry SRS
p2
p2
Firm
LRS
c
s
Industry
LRS
c
s
Firm
LRS
c
s
Industry
LRS
c
s
D1
0
q*
firm
D2
Q*
industry
q*
firm
With demand D1 price = c = LRMC = LRAC
Q* Q2
industry
In the SR, price rises to p2, industry quantity Q* & firm quantity
q* do not change.
Both LRS and SRS = demand, industry quantity Q* firm

quantity q*. Industry is in entry and exit equilibrium.
Firms make profits.
SRS1 SRS2
firm SRS
Industry SRS
p2
p2
Firm
LRS
c
s
Industry
LRS
c
s
c
s
Firm
LRS
p3
Industry
LRS
c
s
D2
0
Demand rises to D2.
firms makes 0 profits.
firm SRS
TP
q*
firm
Q* Q2
D3
Q
industry
q*
firm
Q*
industry
Demand rises to D2.
Demand falls to D3.
If demand is expected to continue at this level in LR either

firms expand capacity or there is entry.
In the SR, price falls to p3, industry quantity Q* & firm quantity
q* do not change.
Industry SRS moves from SRS1 to SRS2, quantity rises to Q2

and price returns to c.
Firms make losses.
firm SRS
Industry
SRS3 SRS
SRS1
Firm
LRS
c
s
firm SRS
Industry
LRS
c
s
Industry SRS
Firm
LRS
c
s
Industry
LRS
c
s
D3
0
q*
firm
Q3 Q*
D4
0
industry
q*
firm
Demand falls to D3.
Demand falls to D4.
If demand is expected to continue at this level in LR either

firms reduce capacity or there is exit.
SR, price s, Q = Q4
Industry SRS
firm SRS
Long run short run with uncertainty
But it may be impossible to predict the price, due to

unpredictable fluctuations in supply and demand.
In this example the firm is never on its LRS supply curve,
price fluctuates around the long run equilibrium price
LR price = c, Q = Q5
Industry SRS moves from SRS1 to SRS3, quantity falls to Q3

and price returns to c.
K is chosen in the planning period, based on expected

price in the production period
Q5 Q4 Q*
industry
pB
Firm
LRS
c
s
Industry
LRS
pA
DB
DA
0
q*
firm
0
industry
Quantity is chosen in the planning period
Q*
Q
Uncertain
demand
Demand in the production period fluctuates unpredictably

between DA and DB
The price varies unpredictably between pA and pB.
firm SRS
Industry SRS
Firm
LRS
c
s
Uncertain
supply
pB
Industry
LRS
pG
D
0
q*
firm
QB Q*QG
industry
If quantity is targeted at Q* in the planning period, but

varies in the production period between QG (good weather)
and QB (bad weather) the price varies between pG and pB
5. Entry & exit, short run

& long run costs, with a
u-shaped average cost
curve
5. Entry & exit, short run & long run costs,

with a u-shaped average cost curve
/unit
SRAC
SRMC
Assume all firms including potential entrants have the same

costs.
LRAC
Profit maximization implies p = MC.

p0
Assume identical firms, including potential entrants so entry

and exit equilibrium implies p = AC.
SRAVC
LRMC
p = MC = AC implies firm output q0 is at minimum AC and

p = p0 in the next slide.
q0
From the
firms, &
costs slides
Industry output Q given by supply = demand at price p.

Number of firms = Q/q.
Long run supply with u shaped average cost

curves
Assume for now that all firms are identical and input prices
do not vary with the size of the industry.
Firms enter if p > min AC = p0 and exit if p < min AC = p0.
Short run supply is the horizontal sum of firm short run
LRAC
srs
SRS1
p1
p1
LRS
D1
q1
Q1
supply.
industry
firm
Long run supply comes from entry.
With demand D1 price is p1, firms produce q1 and make 0

profits.
In this case LRS is a horizontal line at p0.
Both LRS and SRS = demand.
Long run supply is perfectly elastic.
The industry is in entry and exit equilibrium.

Industry output is Q1.
srs
LRAC
p2
p2
p1
p1
D2
LRS
q1 q2
firm
LRAC
p1
p1
D1
srs
SRS1
Q1
Q2 Q
industry
Demand shifts to D2. In the SR there is no entry.
D2
SRS1
SRS2
LRS
D1
q1 q2
firm
Q1
If demand D2 is expected to persist there is entry.
p2 and Q2 determined by SRS1 and D2 .
Industry SRS shifts from SRS1 to SRS2.
Given p2 firms produce q2 and make profits > 0.
Industry output rises from Q1 to Q3, price returns to p1.
The industry is not in entry and exit equilibrium.
Firm output returns to q1.
Q3
industry
Fracking
Is this a good story?

It is plausible that if the industry is in entry and exit equilibrium
it stays there.
But changes in demand or in input prices with a fixed number
of firms and no entry and exit move the industry away from
entry and exit equilibrium.
The model says where the new entry and exit equilibrium is.
It does not say how the transition to the new equilibrium works.
Why do exactly the right number of firms enter? There is no
good simple story for this.
Increasing cost industries

Up to now we have assumed input prices dont change as
the industry expands.
Extracts gas
Profitable at current gas
prices
Starting fracking increases supply
Gas prices drop. How fast and how far?
Decision: will a well be profitable when the price falls?
Decreasing cost industries

Standard example of decreasing cost industry, silicon chips
for computers.
If the industry input is not much used elsewhere this is

unlikely,
e.g. agriculture & agricultural land, steel and iron ore
But this is an industry with economies of scale so is not

perfectly competitive.
If input prices rise as the industry expands long run supply

is upward sloping, but firms make 0 profits.
Thinking about this properly requires modelling this

imperfectly competitive industry.
A fall in demand results in a fall in input prices.
6. Firms with different costs & economic rent

Up to now I have assumed that all firms including potential
entrants have the same costs.
6. Firms with different

costs & economic rent
Different firms may have different costs, due to different

technologies or access to different quality inputs.
I will look at the effects of access to different quality inputs.
Case 2: some firms have higher quality

inputs which cant be traded
Case 1: Entry Costs

There is a cost to entering the industry. For firms already
in the industry this is a sunk cost and not an opportunity
cost.
Some inputs come in different qualities.

If either the difference cant be observed so high and low
quality inputs trade at the same price
For potential entrants this is an opportunity cost.

or it is not possible to trade the inputs
If at the industry price
0 < profits for firms in the industry < entry costs
firms with higher quality inputs have lower costs and higher
profits.
firms in the industry make positive profits in entry and exit

equilibrium.
Case 3: some firms have higher quality

inputs which can be traded.
Economic Rent
Some inputs come in different qualities.
Originally this story about different quality inputs was

about farming with varying land quality.
Now suppose these inputs can be traded.

With no market for land farmers with high quality land
make economic profits.
Higher quality inputs have higher prices.
With a market for land all the profits go to the landowner in

the form of rent.
In the simplest case all the extra profits that firms have with
high quality inputs that cant be traded go into the price of
these inputs.
Farmers that do not own the land make 0 profits.

Now firms with high quality inputs make 0 profits.
Rent is sometimes used as a term for profits.
A very simple model of rent in a price

taking oil industry
There are two types of input.
High quality onshore wells, MC of production c1, total
capacity Q1.
Low quality offshore wells, MC of production c2 > c1,

total capacity Q2.
Onshore
supply
Offshore
supply
Industry supply
c2
c2
c1
c1
Q1 Q2
Q1
Q1 + Q2
add S curves horizontally
Industry supply & demand with different demand curves

$
c2
Solve for S =D
c1
accurate diagram
c2
p
c1
Solve for S =D
accurate diagram
or find Q at p = c1 & p = c2
0
Q1
or find Q at p = c1 & p = c2
then use a diagram
Q1 + Q2
Supply = demand
Q1
then use a diagram
Q1 + Q2
Supply = demand
0 < Q < Q1 p = c 1
Q = Q1 c 1 < p < c 2
Onshore wells make 0 profits & produce less than

maximum Q1
Onshore wells produce maximum Q1

Offshore wells do not produce and make 0 profits
Offshore wells do not produce and make 0 profits

$
c2
c1
Q1
Economic rent from a

quota
Q1 + Q2
0 < Q < Q1 p = c1.

Onshore produce Q1, profits,
Offshore produce < Q2 profits 0.
Economic rent from a quota

p
The situation with no quota
p1
Economic rent from a quota

p
The quota limits output

to Q1 < Q0.
p1
supply
supply
demand
0
Q1
Q0
demand
0
Q1
Q0
With no quota p = c, Q = Q0,
With quota p = p1 > c Q = Q0, CS
CS (consumer surplus)
Producers makes profit (rent)
With horizontal supply profits (rent) = 0.
Deadweight loss (loss CS profits)
10
Quotas are usually fixed by allocating

licences
Increasing the number of medallions
Example, New York taxi medallions
Increases the number of taxis, possibly lower fares.
Given for free to taxi drivers in 1930s
Decreases the price of current medallions, so is a loss to

medallion owners.
Aim to limit number of taxis, so increasing drivers earnings.

Now there is a market for medallions
(price August 2014 $1,045,000 (individual))
Medallion owners have a strong interest in not increasing

the number of medallions.
Drivers usually do not own medallions & get approximately

the minimum wage.
Other examples of quotas

Imports
European Union milk production
Fishing quotas (externalities are important here)
Carbon Dioxide emissions (externalities are important
here)
Quotas can be hard to enforce and have unintended
consequences.
Allocation mechanisms
Administrative
Sale with a price determined by the government
Auction
Rent seeking
Activities aimed at capturing rents
Public relations
Lobbying politicians
Corruption
Crime
7. Welfare economics of a tax with supply

and demand
Assume that firms costs = social costs
7. Welfare economics of
a tax with supply and
demand
A strong assumption requiring no externalities (next term)

and perfectly competitive input markets
(otherwise price inputs > marginal cost of inputs)
This assumes that distribution is not an issue, the social

gain of giving 1 to someone does not depend on
wealth, income .
See Perloff chapter 9 for many examples
11
Producer surplus for a firm

MC = v'(q)
p1
Industry consumer and producer surplus
Total cost = F + v(q)

v'(q) = derivative of variable cost
= MC.
q1
= fixed cost + variable cost.
Integration, v(q1) is area under MC

curve.
p1
p1
0
q1
firm
Total revenue = p1q1 =
industry consumer surplus

= total revenue variable cost
Producer surplus
Q1
industry
If there is no fixed cost producer surplus = profit.
Welfare economics with supply and demand

example 1 tax
sum the firms supply curve horizontally to get industry supply

and industry producer surplus
Effects of a tax in an industry with CRS

If the tax is t and price paid by consumers is p
Assume for simplicity the tax t is per unit sold, not % of

price
firms get p t.
Example excise tax on cigarettes, alcohol, petrol.
If all firms have CRS so constant AC & MC

profits from supplying q are (p c t)q
Perloff calls this a specific tax
profits are 0 when p = c + t,

supply with tax is perfectly elastic at c + t
The tax t per unit increases marginal cost by t
Assuming that input prices dont change when the industry

expands industry supply with the tax is horizontal at c + t.
Effect of an excise tax in an industry with CRS
c+t
c
S with tax
c+t
s no tax
price p
c+t
s with tax
Effects of an excise tax with CRS
supply
S no tax
demand
demand
q
firm
Q1
Q2
industry
Q2
Q1
With CRS there is no producer surplus.
with no tax price is c (= AC = MC), industry produces Q1
With no tax price = c, industry output Q1
with the tax price is c + t, industry produces Q2,
Consumer surplus = entire shaded area.
tax revenue = tQ2
With tax t price = c + t, Q falls to Q2, tax revenue =

loss consumer surplus =
Firms make
0 profits with
and without
tax.
> tax revenue
12

Definition of deadweight loss due to tax
price p
deadweight loss =
loss of producer & consumer surplus tax revenue
c+t
supply
In this case loss of producer surplus = 0
demand
loss of consumer surplus equivalent variation

Q2
deadweight loss equivalent variation tax revenue
Q1
Here demand is more elastic,

= excess burden of the tax to the consumer
= t (Q1 Q2)
deadweight loss area
is a larger fraction of tax revenue area
1
Deadweight loss tQ
2
(the is because Q Q2
Q1 0 )
Q p pQ
pQ
tQ
e
e
Q
p
Q
p
p
p
Q p
where e elasticity
so with downward sloping
p Q
demand curves elasticity is negative.
t
1 tQ 1
1
Deadweight loss tQ te (e)(tQ)
2
2 p 2
p
t
1 tQ 1
Deadweight loss te (e)(tQ)
2 p 2
p
tQ tax revenue,
t / p proportion ate increase in price
deadweight loss
tax revenue
1
- elasticity proportion ate increase in price
Profit maximization with an excise tax &

increasing MC
Implication
It is better to raise tax revenue by taxing goods whose
demand is inelastic.
But this ignores distribution.
If the good is a large proportion of the consumption of
poor people and you want to take this into account this
argument is too simple.
The analysis can be extended to take explicit account of
distributional value judgements, but has to be done with
maths not diagrams.
Not in this course.
= t Q2
MC + t = supply curve with tax t

MC = supply curve with no tax
p
p-t
If there is a tax t & consumers pay p the firm gets p t on

each unit.
Profit maximization implies p t = MC, or p = MC + t.
The tax shifts the supply curve upwards by t.
Producer surplus with tax =
13
S with tax
Supply and demand with tax
S with tax
p2
S no tax
p1
D
p2 - t
0
q2
q1
firm
Q2 Q1
industry
industry
diagram
S no tax
p2
p1
p2 - t
loss of consumer surplus
loss of producer surplus

0
Q2
Q1
with no tax price p1, industry Q1, firm q1
tax revenue
With tax price p2 industry Q2 at intersection D and S with tax.
deadweight loss = loss of surplus tax revenue
consumer surplus =
Any form of tax except lump sum taxes involves a

deadweight loss.
Politically acceptable lump sum taxes are not feasible.
producer surplus =
tax revenue tQ2 =
supply with tax
Tax incidence: who pays the tax?

p
Statutory incidence
supply with no tax
p2
p1
who sends the tax revenue to the tax authorities?
p2 - t
demand more elastic

than supply
Economic incidence
loss PS > loss CS
Who suffers a fall in price (producers) or an increase in

price (consumers)?
Q2 Q1
Economic incidence depends on the elasticity of supply and

demand.
S with tax
supply with tax
p
p2
p2
supply with no tax
S no tax
p1
p2 - t
p1
p2 - t
D
supply and
demand are
elastic,
deadweight loss
is a large fraction
of tax revenue
demand less elastic

than supply
loss PS < loss CS
Q2 Q1
Q2 Q1
14
S with tax
p
S no tax
p2
supply and demand

are inelastic,
deadweight loss
is a small fraction of
tax revenue
p1
p2 - t
D
Q2 Q1
The UK debate on
alcohol:
tax or minimum price
Policies being considered to reduce alcohol consumption:
Liam Donaldson:
UK Chief Medical Officer (1998 2010)
arguing for a minimum price per unit alcohol
1) Minimum price of 45p per unit

2) Tax reform
3) Ban of quantity based offers
Tax increase may not increase consumer prices

Supermarkets & brewers may absorb the tax
UK debate on alcohol
Discussion and following figures from

R. Griffith, A. Leicester and M. O'Connell, Price-based
measures to reduce alcohol consumption
Institute for Fiscal Studies Briefing Note 138 (March
2013) http://www.ifs.org.uk/publications/6644
Current taxes not based on amount of alcohol so dont

target heavy drinkers
Liam & Rutter, Addiction, p 737 738, April 2011
Griffith, Leicester & OConnell
Annual average alcohol consumption 1970 - 2012
A minimum price results in large transfers to

supermarkets & brewers.
Taxes raise government revenue
How far supermarkets & brewers absorb the tax is an
empirical question.
The structure of taxation needs reform but is
constrained by EU Directives.
Institute for Fiscal Studies Briefing Note 138 2013
R. Griffith, A. Leicester and M. O'Connell, Price-based measures to reduce

alcohol consumption http://www.ifs.org.uk/publications/6644
Based on data from the OECD Health Statistics 2012
15
Alcohol prices relative to allitems RPI inflation 1990 - 2012
Median Estimates of
elasticities
own price
elasticity
On trade: bars, pubs

Off trade: supermarket off licences
R. Griffith, A. Leicester and M. O'Connell, Price-based measures to reduce

alcohol consumption http://www.ifs.org.uk/publications/6644
Based on ONS Retail Price Index data
5TP
beer
- 0.46
wine
- 0.69
spirits
- 0.80
Wagenaar, A.C. et al Effects of beverage alcohol price and tax levels on

drinking: a meta-analysis of 1003 estimates from 112 studies Addiction,
104, p. 179 - 190 February 2009
Effects of a minimum price
price p
price p
p1
p1
supply
supply
demand
Q2
Q1
demand
Q2
Q1
Tax = p1 c fall in consumption Q1 Q2
Minimum price p1 fall in consumption Q1 Q2
tax tax revenue
tax profits of suppliers
Taxes on Alcohol 2014
The minimum price & tax have the same effect on

consumers.
The minimum price generates no revenue for the
government and increases profits of retailers &
manufacturers
The tax produces revenue for the government and in this
model with CRS does not change profits of retailers &
manufacturers.
UK taxation of alcohol is complicated and constrained by
EU directives.
(in addition to VAT)

Product and basis of duty
Duty
Rate per litre of pure alcohol

Spirits
28.22
per hectolitre (100 litres) per cent of

alcohol in the beer
Beer
High-strength beer (>7.5%)
18.74
24.03
per hectolitre (100 litres) of product

Wine
exceeding 5.5% - not exceeding 15%
alcohol per volume.
273.31
Source: http://www.hmrc.gov.uk/
16
Excise tax per

unit of alcohol
by strength and
type, 2012/13
Is this a
sensible tax
regime?
R. Griffith, A. Leicester and M. O'Connell, Price-based

measures to reduce alcohol consumption
Institute for Fiscal Studies Briefing Note 138 (March
2013) http://www.ifs.org.uk/publications/6644
compare
minimum unit price 45p / unit

simple tax regime estimated to have the
same effect on total alcohol consumption as
the minimum price
20p per unit tax on wine and spirit
7.1 per unit tax on other drinks
07
Average
purchase
change
%
Assumes elasticity of demand - 0.5
Units per adult per week

7 14 15 21 21 35 > 35
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
Tax is better targeted on heavy drinkers who buy stronger

cheaper alcohol
Tax reform: additional tax revenue to government
Minimum price: extra revenue to retailers & manufacturers
Minimum
unit price
Tax reform
Modelling issues
Imperfect competition: supermarkets are not price
takers.
Take model to data.
8. Welfare economics of a
subsidy with supply and
demand
17
Supply by a firm with a subsidy

8. Welfare economics of a subsidy with
supply and demand
Assume for simplicity the subsidy s is per unit sold, not
% of price,
MC = supply curve no subsidy

p+s
subsidies are particularly common on food, housing

and agricultural products.
With a subsidy s consumers pay p , producers get
p + s so supply at the point where MC = p + s or
p = MC - s
MC - s = supply curve with

subsidy
If there is a subsidy s & consumers pay p the firm gets p + s

on each unit.
Profit maximization implies p + s = MC, or p = MC - s.
The tax shifts the supply curve downwards by s.
Producer surplus with subsidy =
Supply &
demand
with a
subsidy
With no subsidy Q = Q1 p = p1,

with subsidy price consumers pay falls to p2
price producers get rises to p2 + s
Q expands to Q2
price
simple
welfare
economics
of a subsidy
gain in producer surplus

gain in consumer surplus
entire shaded area = cost of subsidy = sQ2
deadweight loss
= cost of subsidy
gain from subsidy
price
supply no
subsidy
p2 + s
p1
supply with
subsidy
p2
p2 + s
p1
supply with
subsidy
p2
demand from
consumers
Q1
Q2
demand from
consumers
Q1
The gain in consumer and producer surplus from a subsidy is

less than the cost of the subsidy.
Who gains from the subsidy on some agricultural products from
the Common Agricultural Policy of the European Union?
simple
welfare
economics
of a subsidy
Q2

gain in consumer surplus
entire shaded area = cost of subsidy = sQ2
deadweight loss
Effects on farmers outside the EU are also important.
The subsidies increases the value of agricultural land.

Land owners gain when the subsidy is introduced, and would
lose if the subsidy were removed.
New entrants do not gain.
= cost of subsidy
loss of surplus
price
Simple argument: the cost of subsidy exceeds the benefit to
farmers and consumers.
s
p2 + s
p1
With less elastic

supply most of the
gain in surplus
goes to producers
p2
demand
Q1Q2
18
simple
welfare
economics
of a subsidy
There is a large stock of housing.
With completely
inelastic supply all
the gain in surplus
goes to producers.
price
There is no
increase in quantity
or consumer
surplus There is no
deadweight loss
p1 + s
p1
demand
Q1
In the short term supply is very inelastic.

The main effect of a subsidy is to push up the price that
sellers get with very little reduction in the price buyers
pay.
If supply is completely inelastic all the benefit goes to the
sellers.
Subsidising buyers chiefly helps sellers.
UK Housing Benefit debate
UK Housing Benefit debate
Housing benefit is paid to low income households, both in

and out of work.
If the simple model is correct this is a question about relative

elasticities.
It is being reduced.
If supply is completely inelastic landlords lose, tenants do not.
Both landlords and tenants will lose.
Perfect competition: homogeneous goods, buyers can

costlessly switch to a lower price seller.
Question: how much do tenants lose?

Neither of these assumptions hold for the rental housing
market.
How important is this?
With supply and demand analysis

Any policy that puts a gap between what consumers pay
and producers get moves quantity from the level that
maximizes surplus.
Remember the limits of supply and demand analysis
9. Monopoly
Assumes price taking.

Consumers know the prices and quality, no asymmetric
information
Input prices reflect social costs, so no externalities,
perfect competition in the input markets.
Distribution is not taken into account.
19
9. Monopoly
Up to now I have assumed that firms are price takers.
They are small relative to the market so their output
decisions have no effect on prices.
Now look at industries with one firm (monopoly)
Later look at industries with a small number of firms
(oligopoly).
See Industrial
Organization Worked
Example 2, Monopoly
Why are there industries with a small

number of firms?
Economies of scale
to be discussed later.
Legal barriers to entry
e.g. licenses, quotas, state monopolies
Access to a critical input or technology that is not
available to other firms,
patents
secrecy making imitation impossible
organizational culture
If q > 0 maximizes profits the first order conditions imply

(q) = R(q) c'(q) = 0 or
For a price taking firm marginal revenue MR = p price
MR marginal revenue = R(q) = c'(q) = marginal cost.
For a firm whose output affects the price it gets using

the product rule R = pq where p is a function of q
This is the marginal output rule.

If c(0) = 0 then if q = 0 profits = 0.
Profit maximization then implies (q) = R(q) c(q) 0
so if q > 0 maximizes profits
average revenue = R(q)/q = p c(q)/q = AC
That is p AC, the shutdown rule.
R
p
q p
p q
)
p (1
q
p q
q
1
p (1 )
e
p q
where e
own price elasticity of demand.
q p
e 0 because demand is downward sloping
1
so marginal revenue MR p (1 ) p price
e
MR
A simple model of demand

Profit maximization implies MC = MR < p so MC < p
Throughout this section
For a monopoly elasticity, MR and output depend only on

demand from buyers.
Q = industry supply = demand when = Q = - p

where p = price so p = a bQ
where a = / and b = 1/
For a firm in a oligopoly elasticity, MR and output depend

on demand from buyers and the output by other firms in
the industry.
Assume that > 0 & > 0 so a > 0 and b > 0.

For a monopoly total revenue pQ = (a bQ)Q = aQ bQ2.
Marginal revenue =
derivative of total revenue with respect to Q = a 2bQ.
20
Example monopoly with constant returns to

scale (CRS)
If a > c profits (a c)Q bQ2 are a quadratic in Q with a

negative coefficient of Q2 .
the algebra
A firm is a monopoly when there is no other firm producing
the same good.
With CRS the firms total cost is cQ where c = MC = AC.
If price = a bQ profits pQ cQ = (p c)Q
= (a bQ c) Q = (a c)Q bQ2.
If a c profits are negative for all Q > 0, the firm produces 0.
The foc (first order conditions) give a maximum where the

derivative is 0, that is a c 2bQ = 0 .
The same condition can be got from
MR = marginal revenue = marginal cost = c.
Total revenue = pQ = (a bQ)Q = a bQ2
differentiating with respect to Q gives MR = a 2bQ
so MR = MC gives a 2bQ = c
or a c 2bQ = 0.
Exam mistake
doing this analysis for
a perfectly
competitive industry
Monopoly with CRS, the diagram

From the first order condition MR = MC
a c 2bQ = 0 so profits are maximized when
Q = Qm = (a c)/b.
price pm = a b Qm = a - (a c) = a + c > c because
a > c.
profits = total revenue total cost = pmQm cQm
MR = a 2bQ = c = MC when Q = Qm = (a c)/b

If Q < Qm MR > MC profits increase when Q increases
If Q > Qm MR < MC profits decrease when Q increases
Profit max at Qm where MR = c = MC.
/unit a
pm= (a + c)
demand a bQ
c
= (pm c)Qm = (a c)Qm = (a c)2 /b.
MR = a 2bQ
0
Monopoly with the cost function c(q) = F + cq
Qm
a/2b
a/b Q
Monopoly with the cost function c(q) = F + cq
Demand gives p = a bQ, profits = total revenue total cost

= pQ cQ F = (a bQ)Q cQ - F = (a - c)Q bQ2 - F
If Q > 0 and profits are maximized the foc is a c 2bQ = 0

so Q = (a c)/b > 0
If a c profits < 0 for all Q > 0 the firm does not produce.
Price p = a bQ = a - (a c) = (a + c) = c + (a c) > c
If a > c profits are a quadratic function of Q with a negative

coefficient on Q2.
Profits (p c)Q F = (a c)2 /(4b) - F.

If F > (a c)2 /(4b) the firm cannot make a profit so shuts down.
21
/unit
Case 1 (a c)2 /(4b) - F 0
Case 2 (a c)2 /(4b) - F < 0
At Qm = (a c)/b MR = MC and
price pm = a + c > c + F/q = AC.
Price and quantity are the same as
in the CRS case but profits are
/unit
At Qa MR = MC and profits are

higher than at any other Q > 0
But price pa < c + F/q = AC.
The monopoly makes losses at all
(a c)2 /(4b) - F 0.
pm
Q > 0 so does not produce.
pa
demand
MR
AC = c + F/q
MC = c
Qm a/2b
c
MR
QC
a/b
Comparing monopoly
and perfect competition
there is a deadweight
loss due to monopoly.
But if there are fixed
costs perfect
competition is
impossible.
/unit
pm
Qm
MC = c
AC = c + F/q
Qa a/2b
a/b
10. Cartels
Assume there are no fixed costs, total cost = cQ.

Monopoly profits = (pm c) Qm
Consumer surplus
Compare perfect competition, p = c, Q = QC

Consumer surplus = entire shaded area.
Loss of consumer surplus due to monopoly monopoly profits =
deadweight loss due to monopoly
10. Cartels
Suppose that there are n firms in an industry that have a cartel
agreement to maximize industry profits.
Suppose firm i has costs cqi + F.
Assume F > 0 so there are economies of scale.
Cartels
Industry profits = (a bQ - c)Q nF
= (a c) Q bQ2 nF.
If a c the industry cannot make profits.
Total industry cost

If a > c and the industry produces Q > 0
= c(q1 + q2 ..qn) + nF = cQ + nF where Q = q1 + q2 ..qn.
Assume as before p = a bQ.
Industry profits pQ c Q nF = (a bQ - c)Q nF
the formula for profits is the same as with monopoly except the
fixed cost is nF not F.
Same as monopoly profits except last term is nF not F.
22
Cartels
If Q > 0 maximizes profits Q = (a c) /b
price p = (a + c),
p and Q are the same as with a monopoly but
cartel profits (a c)2 /(4b) - nF
If n > (a c)2 /(4bF) the cartel makes losses,

so at least one firm in the industry makes losses.
As the cartel maximizes industry profits if n > (a c)2 /(4bF)
there is no way the industry can be profitable at any level of
output.
An industry with n > (a c)2 /(4bF) is impossible when there is
entry and exit.
< monopoly profits (a c)2 /(4b) - F.

Dropping the one price assumption
A large fixed cost F is a barrier to entry
as is low demand, small a or large b.
Up to now p = price per unit same for all buyers, doesnt

depend on who the buyer is and how much purchased.
If F < (a c)2 /(4b) < 2F a monopoly can make profits but

if there are 2 or more firms the industry makes losses.
The industry must be a monopoly.
Now suppose p = p1 for lucky people, p = p2 > p1 for others.

If the lucky people can sell on to the others at p with
p1 < p < p2 they make a profit and no one buys at p2.
Dropping the One Price Assumption
If it is impossible, costly or difficult to resell people may end up

paying different unit prices for the same good.
This is called price discrimination.
Examples, student discounts, buy 2 get 1 free,
airline pricing, mobile phone pricing.
For EC201 we only look at price discrimination by a monopoly.
23
First Degree (Perfect) Price Discrimination

Suppose v(q) (value) is the most a consumer will pay for q
units.
The monopoly makes a take it or leave it offer, either get q for a
total payment of just less than v(q) or get nothing.
First Degree (Perfect) Price Discrimination

In perfect competition the consumer chooses q to maximize
v(q) pq,
foc v'(q) = p so v'(q) is the demand curve
v(q) is the area under the demand curve
Consumer accepts offer because payment < value.

v(q) pq is consumer surplus,
Monopoly profits v(q) c(q) maximized when
v'(q) = c'(q).
q is the same with perfect competition & perfect price

discrimination.
/ unit
/ unit
v(q) = total shaded area

monopoly
profits
area = v(q)
v'(q) demand curve
c'(q)
v'(q)
demand curve
total
cost
q
Compared to perfect competition there is no loss of
surplus, but all the consumer surplus now goes to the
monopolist.
Quantity (Second Degree) Price Discrimination
Multimarket (Third Degree) Price Discrimination
Make the price per unit vary with the amount purchased.
e.g, Quantity discounts, 2 for 1 offers
connection charge + price per unit (electricity)
Suppose customers can be sorted in some way, e.g.

where they buy, when they buy, age, student status
Choice of mobile phone plan.

The customers choice of quantity is informative about demand
and consumer surplus although it does not reveal everything.
Sellers can take advantage of this.
Then the firm can charge different prices to different

groups and make more profit than if it had to charge
everyone the same price.
This is too difficult to model for EC201.
24
Multimarket (Third Degree) Price Discrimination
ECONOMISTMAGAZINE
Suppose customers can be sorted in some way, e.g where

annual
cost
coverprice
student
subscription
1year(direct
debit)
regular
subscription
1year(direct
debit)
weekly
cost
5.00
they buy, when they buy, age student status
%ofcover
price
100
Then the firm can charge different prices to different groups

and make more profits than if it had to charge everyone
the same price.
126
2.42
48
n markets, market i has revenue Ri(qi) = piqi

total revenue = R1(q1) + R2(q2) + Ri(qi) + Rn(qn)
155
3.00
60
Total cost c(q1 + q2 . + qi + qn)
MRi
Multimarket price discrimination maximizes
total revenue total cost
Ri
p
q pi
pi qi i pi (1 i
)
qi
qi
pi qi
pi (1
1
) ( ei elasiticty in market so i so ei 0) so
ei
if - ei 1 MRi 0 so the firm does not sell in market i.

R1(q1) + R2(q2) + Ri(qi) + Rn(qn) - c(q1 + q2 . + qi + qn)
First order conditions for maximization set derivative with
respect to qi at 0 so
Ri(qi) = c(q1 + q2 . + qi + qn) for all i
Marginal revenue is the same in all markets and equal to
marginal cost.
If - ei 1 profit maximization implies MRi pi (1
1
) MC
ei
1
pi
is increasing in ei .
MC (1 1 )
ei
The mark up over marginal cost is higher when - ei is close to 1.
Intuition when demand is inelastic increasing price reduces
demand by less.
25
3.2 Oligopoly and Games
3.2 Oligopoly and Games
1. Introduction to Game Theory

2. Prisoners' dilemma
3. Dominant strategy & Cournot-Nash equilibria
4. Nash equilibrium
5. Bertrand-Nash equilibrium
1. Introduction to game theory

6. Pure and mixed strategies
This is the way modern economists model oligopoly

(industries with a small number of firms who take into account
each others actions)
7. Multiple equilibria
8. Simultaneous (normal form) and sequential move
(extensive form) games
It is also used to model many other situations.

Language for game theory
9. Stackelberg equilibrium
Games have players.
10. A two stage entry game
Each player has a strategy.
11. Repeated games
Payoffs depend on strategies and are illustrated in the payoff

matrix.
Firm 2
large
Firm
1
small
large
16, 16
20, 15
small
15,
18, 18
20
20, 15 in the top

right box means
player 1 gets 20
player 2 gets 15
when 1 plays
large & 2 plays
small.
Economics Lesson on Cartels

Think of this as a model of a cartel.
Limiting production increases profits for all firms.
But each firm has an incentive to increase output.
The players are firm 1 and firm 2.

The players strategies are large output & small output.
Cartels are difficult to sustain.
The payoff for each player depends on the choice of strategy

by all players.
The table is the payoff matrix
3.2 Oligopoly and games
2. Prisoners' dilemma
Eyster
confess
Prisoners' dilemma
Bray
not
confess
confess
16, 16
20, 15
not
confess
15,
18, 18
20
Bray and Eyster are prisoners. They are being interrogated

and are offered a reduction in prison sentence to anyone
who confesses.
Both have an incentive to confess, but they both do worse if
they both confess than they would do if neither confessed.
The prisoners' dilemma game was formulated in the Cold War

as a model of the nuclear arms race,
the players were USA and the Soviet Union.
Sustaining a cartel,
sustaining a fishery
Lessons from Game Theory

1. In the standard competitive model people acting
individually in their own self interest achieve a Pareto
efficient outcome.
In the prisoners' dilemma model the opposite happens,
the outcome when both act in their individual interest is
worse for both of them than if they act cooperatively.
international trade negotiations

international action on climate change
can all be modelled as prisoners' dilemma.
2. In sport games are zero sum, one teams win is another

teams loss.
Prisoners' dilemma is not zero sum.
Life is not zero sum.
3. Dominant strategy & Cournot-Nash equilibria

A strategy is dominant if it maximizes a players payoff
whatever the other player does.
Cournot-Nash
equilibrium
In prisoners dilemma both players have a dominant strategy,

confess.
The prisoners' dilemma has a dominant strategy
equilibrium, i.e. a situation in which each player has, and
plays, a dominant strategy.
Many games do not have a dominant strategy equilibrium, as
in the following model.
Definition of a Cournot-Nash equilibrium in a

duopoly model
In the Cournot model of a duopoly (industry with 2 firms) each
firms strategy is its output.
In the Cournot-Nash equilibrium the outputs q1 and q2 have the
property that
If demand is given by p = a bQ, one firm produces q & the other

produces s industry quantity Q = s + q.
The firm producing q has total revenue
pq = (a b (q + s))q .
Profits pq cq = (a - c b (q + s))q.
If a c the firm cant make profits at any s or q.
given q2 firm 1 maximizes its own profits by choosing q1.
If a c bs < 0 the firm cannot make profits with q > 0
given q1 firm 2 maximizes its own profits by choosing q2.
If a c bs 0 profits are maximized by setting
Industrial Organization
Worked Example 3
Applying this to firm 1 q1 is its own output,

s = q2 is the output of the other firm, taken as given by firm 1.
q = (a c - bs )/2b.
Price, quantity and profits in the Cournot duopoly

model
Profit max at q1 > 0 if a - bq2 c 0 & a 2bq1 bq2 = c.
Firm output q1 = (a c)/3b q2 = (a c)/3b
Similarly for firm 2 taking q1 as given profits are maximized at

q2 > 0 if a - bq1 c 0 & a 2bq2 bq1 = c.
Industry output Q = q1 + q2 = 2/3 (a c)/b
Solving simultaneously
a 2bq1 bq2 = c
a 2bq2 bq1 = c
gives q1 = (a c)/3b q2 = (a c)/3b
in which case a bq2 c = a bq1 c = 2/3 (a c) > 0 if a > c.
Reaction functions for the Cournot Model

The easiest way to solve the equations
a 2bq1 bq2 = c & a 2bq2 q1 is by elimination.
Another approach rearrange a 2bq1 bq2 = c to get
get q1 as a function of q2 getting q1 = (a - c bq2)/2b.
This is firm 1s reaction function, giving its profit maximizing
level of output given q2.
Similarly firm 2s reaction function is to set
q2 = (a c bq1)/2b.
in Cournot Nash equilibrium both firms are on their reaction
functions.
.
Industry price p = a b Q = 1/3 a + 2/3 c = c + 1/3 ( a c)

Profits for firm 1 = (p c)q1 = (a c)2/9b = profits for firm 2.
Profits are higher if costs are lower (c smaller), or if demand is
higher (a bigger or b smaller)
Starting from the reaction functions

(a c bq2 )
q1
2b
(a c bq1 )
q2
2b
you can find q1 and q2 by substitution
firm 1s
reaction
function
starting with
a bq2
q2
a c b 2b
q2
.
2b
This gives the same result as solving 0
for q1 and q2 from the first order conditions
but the algebra is harder.
Cournot
equilibrium
q1
firm 2s
reaction
function
Reaction functions and cartels

In a 2 firm cartel with the same cost and demand functions
industry output = (a c)/2b,
firm output (assuming equal shares) = (a c)/4b
price c + (a c), firm profits 1/8 (a c)2/b
In firm 2 sticks to the cartel output

firm 1 can increase profits by
increasing output.
firm 1s reaction
function
If firm 1 sticks to the cartel output

firm 2 can increase profits by
increasing output.
Firm 1s reaction function is q1 = (a - c bq2)/2b.
If both firms increase output both

make lower profits than with the
cartel.
If firm 2 produces the cartel output (a c)/4b firm 1s profit

maximizing output is
(a c b[(a c)/4b]/2b = 3/8 (a c)/b > (a c)/4b = cartel
output.
Cartel agreements are hard to

sustain, particularly if output is
unobservable or the cartel is
illegal.
Cournot-Nash
equilibrium
firm 1s
reaction to
cartel output
q2
0
cartel
output
q1
firm 2s
reaction
function
4. Nash Equilibrium
Nash equilibrium and

dominant stratagy
equilibrium
Cournot Nash equilibrium is a special case of a Nash

equilibrium.
In a Nash equilibrium each player's strategy maximizes his
payoff, given the strategies pursued by the other players.
In a Cournot model the strategy is output.
Nash equilibrium and dominant stratagy

equilibrium
5. Bertrand Nash Equilibrium with 2 firms and

identical goods
In a Bertrand duopoly game price is the strategic variable.
In prisoners' dilemma confess is a dominant strategy,

.
because
it is the best thing to do whatever the other
player do.
In a dominant strategy equilibrium each player has and
plays a dominant strategy.
A dominant strategy equilibrium is always a Nash
equilibrium.
But a Nash equilibrium is generally not a dominant
strategy equilibrium.
Suppose 2 firms produce an identical good and both have

total cost cq so AC = MC = c.
If both firms charge the same price p they share the market
equally.
If one firm charges a lower price it takes the entire market.
Industrial Organization Worked
Example 4
Bertrand Nash Equilibrium
Bertrand Nash Equilibrium with 2 firms producing

goods that are imperfect substitutes
If firm 1 charges p1 > c, firm 2s best response is to charge

p2 > c where p2 is just less than p1.
Two firms produce goods which are substitutes but not

perfect substitutes
Firm 1 sells nothing and makes 0 profits,
Demand for firm 1s output
But firm 1 could do better by charging just less than p2, so

p1, p2 is not a Nash equilibrium.
q1 = 2 - 3p1 + 3p2,
Class 9
Demand for firm 2s output

If either firm charges less than c it makes losses and could
do better by charging c.
q2 = 6 - 2p2 + p1.
The Bertrand Nash equilibrium has p1 = p2 = c.
Comparing Bertrand and Cournot

Assume there are 2 identical firms
both have cost total cost cq, MC = AC = c.
In Bertand Nash equilibrium p1 = p2 = c.
This is the same outcome as in perfect competition.
Each firm make 0 profits.
In Cournot Nash equilibrium if a > c p = 1/3 a + 2/3 c > c,
Each firm make profits (a c)2/9b > 0.
Bertrand or Cournot?
Which model is appropriate depends on the real world
situation you are trying to understand.
The result that price setting gives the same result as
perfect competition does not hold if
the goods the firms produce are not perfect
substitutes
or firms commit to quantity (capacity) before they
set prices.
n firm Cournot Model

Assume demand p = a bQ, there are n firms each firm has
total cost cq so MC = AC = c
firm output q
(a c)
(n 1 )b
n(a c)
(n 1 )b
n(a c)
(a c )
a bQ a
c
(n 1 )
( n 1)
industry output Q
nq
Using the same argument as with duopoly if the other firms

produce a total of s a firm producing q makes profits
(a bs c)q bq2.
price p
If a bs c > 0 profits are maximized at q = (a bs c)/b.
(a c)2
(n 1 )2 b
I will come back to this when I have a model of entry.
If all firms produce the same output q so s = (n-1)q these

conditions are satisfied if a > c and
(a c)
q
(n 1 )b
each firm makes profits ( p c ) q
6. Pure and mixed strategies

An enforcement game find the Nash equilibrium
car driver

traffic
warden
park
legally
park
illegally
patrol
- 5,
-10
15, - 100
not patrol
0,
- 10
0,
if the warden patrols the driver parks

if the driver parks legally the warden
if the warden does not patrol the driver parks
if the driver parks illegally the warden

A player plays a pure strategy if she does not randomize,
e.g. she always patrols, or always doesnt patrol.
TP
The enforcement game has an equilibrium in

mixed strategies
If the driver believes that the warden patrols with probability w
the drivers expected payoff from legal parking
A player plays a mixed strategy if she randomizes, e.g.

she always patrols with probability 1/3 and doesnt patrol
with probability 2/3.
= -10w - 10(1 w)
The drivers expected payoff from illegal parking
The enforcement game has no equilibrium in pure

strategies.

mixed strategies
The driver is indifferent between parking legally and parking
illegally and and is willing to randomize if
-10w - 10(1 w) = = -100 w + 0(1 w)
that is if w = 0.1
= -100 w + 0(1 w).

mixed strategies
If the warden believes driver parks legally with probability d
Expected payoff from patrolling -5 d + 15(1 d)
Expected payoff not patrolling 0 d + 0(1 d)
The warden is indifferent between patrolling
and not patrolling parking and is willing to randomize if
-5 d + 15(1 d) = 0 d + 0(1 d)
that is d = 0.75

mixed strategies
This game has an equilibrium in mixed strategies where the

warden patrols with probability 0.1 & the driver parks
legally with probability 0.75.
Existence Question (Not for this course)

Do all games have an equilibrium in either pure or mixed
strategies?
Yes, if the game has simultaneous moves and there are a
finite number of players who each have a finite number of
pure strategies.
Note: in the equilibrium in an mixed strategies

Result proved by Nash (1950).
the probability that the warden patrols is determined by the
drivers indifference condition
the probability that the driver parks legally is determined by
the wardens indifference condition
Nobel Prize for Economics 1994

Biography
Sylvia Nasar, A Beautiful Mind
Faber and Faber 2002
Lessons from Game Theory 3

There are games in which there is no equilibrium in pure
strategies, so in the model players must randomize.
Examples, enforcement (e.g. speed cameras)
Sports, hit right or left randomly to keep opponent
guessing.
Multiple equilibria
7. Multiple equilibria
Lessons from Game Theory 4
The computer choice game

economist
Mac
Games can have multiple equilibria.

So a game theoretic model may not give a prediction of the
outcome.
PC
Mac
2,
0,
PC
0,
1,
biologist
Does this game have a Nash equilibrium in pure strategies?
This is especially true if the same players play many times.

You wont see it in EC201, but the outcomes of game
theoretic model are very sensitive to the assumptions of
the model, so similar models may give very different
predictions.
Modelling entry to an industry as a game

incumbent = firm already in industry
fight if there
is entry
potential
entrant
= firm
deciding
whether
to enter
not fight if
there is entry
not enter
0,
0,
enter
-1 ,
1,
Simultaneous and
sequential move games
Nash equilibrium?
8. Simultaneous and sequential move

games
So far we have assumed that players choose their
strategies simultaneously and used a payoff matrix to
illustrate the game.
Games like this are called simultaneous move games
(also normal form games).
Entry is better modelled as sequential move game
(sometimes called an extensive form game).
This simple game has 2 stages.
stage 1 potential entrant chooses whether to enter
stage 2 incumbent chooses whether to fight.
Extensive form games are analysed using a game tree.
In the entry game the incumbent would like to commit to

fighting if there is entry so as to deter entry.
But the commitment is not credible because once there is
entry it is not profitable to fight it.
Commitment can be strategically useful.
Commitment strategies,
in the entry game investing in capacity to reduce
marginal cost.
when invading, burning boats
somehow reducing the payoff to not fighting.
The entry game looked at as a simultaneous move game
has two Nash equilibria (not enter, fight)
& (enter, not fight).
in (a,b) a = p. entrants profit,

b = incumbents profit
not
(a, b)
(0, 9)
enter
stage 1
potential
entrant
chooses
game tree
enter
stage 2
incumbent
chooses
fight
(-1, 0)
not
(1, 1)
fight
If the potential entrant does not enter the incumbent has no choice
to make. Potential entrant gets 0, incumbent 9.
If the potential entrant enters at stage 1 what does the incumbent
do at stage 2?
What does the potential entrant do at stage 1?
Always analyse sequential move games by backward

induction.
What does last player to move do, given what other players
have already done?
What does the next to last player to move, given what other
players have already done, and knowing how last player
to move will respond?
What does the first player to move do, knowing how

players will respond in all future moves?
Looking at this as a sequential move game it has one

equilibrium, (enter not fight).
9. Stackelberg equilibrium
Stackelberg equilibrium
There are two firms 1(leader) and 2 (follower) with costs

cq1, cq2
p = a b(q1 + q2).
Cournot assumes q1 and q2 are chosen simultaneously.
Stackelberg assumes 2 stages.
Stage 1 leader chooses q1.
Stage 2 follower chooses q2.
At stage 2 the follower choses q2 taking q1 as given.
(a c bq1 )
a c
and q1
2b
2b
(a c)
a 3c
price p a b( q1 q2 )
4b
4 4
The follower' s profits are maximized by setting.
As q2
(a c bq1 )
2b
At stage 1 the leader chooses q1 taking into
q2
account how the follower will respond when choosing q2 .
The leader' s profits are ( p c ) q1 ( a c ) 2 / 8b .
The leader' s profits are
The follower' s profits are ( p c) q2 ( a c ) 2 / 16b .
q2
( p c ) q1 ( a c b( q1 q2 ) ) q1
a c bq1
a c b q1
2b
1
1
q1 ( a c ) bq1 q1
2
2
a c
which are maximized by setting q1
at stage 1.
2b
Firm cost = cq (q firm output) price p = a bQ (Q industry

output)
price
firm output industry firm
industry
output profits profits
c
perfect
competition
(a c)
undetermined
b
n firm
CournotNash
c + (a c) (a c)
(n+1) (n+1)b
n(a c) (a c)2 n(a c)2

(n+1)b (n+1)2b (n+1)2b
2 firm
CournotNash
c + (a c) (a c)
3
3b
2(a c) (a c)2 2(a c)2

3b
9b
9b
Firm cost = cq (q firm output) price p = a bQ (Q industry

output)
price
firm
industry firm
industry
output
output
profits
profits
Stackel
- berg
c + (a c) leader
4
(a c)
2b
follower
(a c)
4b
3(a c)
4b
leader
(a c)2
8b
follower
(a c)2
16b
3(a c)2
16b
monopoly
c + (a c) (a c)
2
2b
(a c)
2b
(a c)2
4b
(a c)2
4b
Comparisons
Price and industry profits are highest in monopoly and lowest
with perfect competition.
A two stage entry game
As n, the number of firms in a Cournot-Nash model, gets

larger, price falls, industry output increases, industry profits
fall.
When n is very large price, industry output and industry
profits are close to their perfect competition levels.
Comparing Stackelberg and 2 firm Cournot-Nash,
in Stackelberg price is lower, industry profits are lower,
leaders profits are higher, followers lower than in C N.
10. A two stage entry game
I have already analysed the 2nd stage game

(a c)2
in stage 2.
(n 1 )2 b
Taking the cost F at stage 1into account
Each firm makes profits

At stage 1 firms decide whether to enter the market or not,
if they enter they pay F.
At stage 2 they have MC = AC = c and they play a
Cournot Nash Oligopoly game with p = a bQ, where
Q = q1 + q2 + +qn.
each firm' s profits from the 2 stages are

(a c)2
F.
(n 1 )2 b
n firms can make 0 or positive profits but n 1 firms can' t if
ac
n 1
bF
the number of firms is small if there is a large fixed
n2
or marginal cost, or demand is small (a small or b large).
Why is this 2 stage game important?

This is a simple model of a product cycle.
Stage 1 decide whether to start on product development
(computer, games machine, drug )
Stage 2 bring the product to market.
Decisions at stage 1 depend on what is expected to
happen at stage 2.
Market Share %
Operating System
Android
iOS (Apple)
BlackBerry
Symbian
Think of the fixed cost as R & D (research and

development).
High R & D costs result in fewer firms in the industry.
Firms that commit early may have a first mover advantage.
Firms that commit late may learn from others mistakes.
Windows
Linux
Others
Q2 2011
Q2 2012
Q2 2013
46.9
68.1
79.3
18.8
16.9
13.2
11.5
4.8
2.9
16.9
4.4
0.2
2.3
3.5
3.7
2.3
0.8
0.5
0.1
10
Eyster
confess
Bray
Repeated games
not
confess
11. Repeated
games
confess
16, 16
20, 15
prisoners'
not
confess
15,
18, 18
dilemma
20
If the game is played once confess is a dominant strategy for both

players.
If the game is repeated 10 times does it change things?
What will happen in round 10?
Infinitely repeated games: the trigger

strategy:
The situation is different when game is repeated an infinite
number of times, and the aim is to maximize present
discounted value of pay-offs, 0 + 1 +22 +..
t can be interpreted as the probability the game is stil
going at date t.
In the trigger strategy start by not confessing
If both players have not confessed in rounds 1,2..n, do not
confess in rounds n+1,
If either player has confessed at some point in rounds
1,2...n, confess in round n+1.
Both players playing the trigger strategy is a Nash

equilibrium of the repeated game if the discount factor
is not too small.
In the equilibrium players have no incentive to deviate from
the trigger strategy at any point, i.e. their commitment to
the strategy is credible.
(jargon, equilibrium is sub-game perfect)
Note trigger strategy depends on being able to observe
what other players do.
It is hard to sustain co-operation where cheating is
unobservable.
The trigger strategy can be a disaster when misperception
is possible:
11

EC201 Lecture Notes Michaelmas 2014

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

EC201 Lecture Notes Michaelmas 2014

Încărcat de

Drepturi de autor:

Formate disponibile

LondonSchoolofEconomics

Alternative text for Michaelmas Term material,

Additional reading will be provided as links from the

EC201 Michaelmas Term Syllabus

Consumer theory and its applications

Utility maximization and uncompensated demand

Normal and inferior goods

Expenditure minimization and compensated demand

Price changes and welfare

Labour supply, taxes and benets

Snyder & Nicholson Chapter 16

Saving and borrowing

Firms, costs and prot maximization

Marginal product and marginal rate of technical substitution

Prot maximization and costs for a price taking rm

Morgan, Katz & Rosen, Chapter 7

Oligopoly and games

See the website and course information

Other books you may find helpful

Other books you may find helpful as an

The objective of EC201 is to develop

Varian, H.R. "Intermediate Microeconomics: A Modern

your ability to think critically about the models and their

Short clear chapters with a calculus treatment in the

Skill development requires practice over a long period.

Varian, H.R. "Intermediate Microeconomics with Calculus"

1.1 Preferences and utility

The exam tests these skills.

Course Outline Michaelmas Term

Consumer theory and its applications

2 Firms, costs and profit maximization

1.2 Utility maximization and uncompensated demand

Oligopoly and games

1.5 Labour supply, taxes and benefits

Models and maps

Models include definitions.

A logical argument derives conclusions from the

A map is a model and like all models

Maps are useful because they simplify greatly.

It is good for some questions.

Look at the London Underground map.

1.1 Preferences and utility

Beware of implicit assumptions, e.g. the London

1. Preferences and utility

1. Preferences and utility

Example of a maximizing models:

Marshalls model of consumer demand

Modern assumptions on preferences

1.1 Preferences and utility

Probability of at least 1 successful application

Probability of success in one application = r

Probability of failure in one application = 1 r

Then probability of failure in all x applications (1 r)x.

If r = 0.5 and x = 5 then = 0.96875.

The students objective

More assumptions: the students objective

Assume N = 5000, p =200

Assume p =200, value of internship N = 5,000

1.1 Preferences and utility

Any optimization problem

Marginal conditions for the

has a solution which depends on parameters of the

Marginal conditions are very often helpful in

Marginal value of xth application