EcoPub PDF Complet

Chapter 1
Introduction
I Public Economics and Economic Efficiency
1.1 Public Economics

• Public economics analyses government economic
intervention
– The use of taxes, expenditures, regulations
• It studies how decisions are made (descriptive)
– The processes through which government decisions are
reached
• It analyses what decisions should be made
(prescriptive)
– The decisions that would be in the best interest of
society
Presentation slides for J. Hindriks and G. D. Myles, Intermediate Public Economics, Second Edition
© Gareth D. Myles, August 2013
1.2 Methods
• Public economics uses models to investigate policy
– The possibilities for experimentation are limited
– Past experience cannot always be relied upon
• Models can take two forms
– Partial equilibrium models focus only on one or two
markets taking behavior elsewhere in the economy as
given
– General equilibrium models describe a complete economic
system with prices equilibrating supply and demand on all
markets simultaneously
1.2 Methods
• Actions of economic agents : Rationality paradigm
– Consumers maximize private welfare
– Firms maximize profits
• The government chooses policy instruments
• Reactions to a policy change
– The reactions of economic agents are predicted through
the solutions to the optimizations
– The independent decision-making of agents distinguishes
economic models
– Agents do not respond mechanically
Methods
Optimisation
Exo variables + Endo variables
Equilibrium
1.2 Methods
• Once a model is constructed its implications are
derived
– Logical reasoning is used to derive formally correct
conclusions
– These conclusions are interpreted in terms of the initial
policy question
• The institutional setting is invariably the mixed
economy
– Individual decisions are respected but the government
intervenes
– A range of objectives can be assigned to the government
1.2 Mixed economy

• Market
– Allocate ressources via (relative) price mechanism
– Market competition (market shares)
– Market failures
• Government
– Allocate ressources (and make policy choices) via
vote mechanism
– Political competition (vote shares)
– Political failures
1.3 Analyzing Policy

• The effect of a policy is determined by contrasting
the equilibrium with the policy to equilibrium
without
• Policy can be analyzed from a positive or a
normative perspective
• Positive analysis is about explaining why there is a
public sector, how government policies are chosen
and how these policies affect the economy
– An example is analyzing the effect of a corporate tax on
inward investment
Methods: Tax
TAX Price change

Change
Optimisation
Exo variables + Endo variables
Equilibrium
Tax reform
Source: Hindriks & Serse (JPubE2019)

Tax overshifting
VAT Electricity
Source: Hindriks & Serse (IJIO 2022).


• Normative analysis investigates what the best policy
is, and aims to provide a guide to good government
– An example would be an assessment of whether the level
of pensions should be indexed to average wages
• Normative analysis assumes the government has an
objective and chooses its actions to best achieve the
objective
• Positive and normative analysis are not distinct
– To evaluate a policy (normative) its effect must be
determined (positive): The credibility revolution with more
and better data (Angrist & Pischke, JEcoPersp 2010)

• The government’s objective is often taken to be the
aggregate level of welfare
• This raises questions about welfare measurement
• Any aggregate measure assumes some degree of
comparability of individual utility
• It is possible to proceed assuming utility is
comparable and to derive general principles that
apply for any degree of comparability
1.3 Pareto efficiency

• what do we mean by good policy
– Market incidence
– Welfare effect
• Identify limited instances of unequivocally
good policy
– Makes some people better off and no one
worse off

• Each individual's preferences can be
represented with a utility function
• The utility function, Ui, represents person
i's preferences:
• If person i prefers a policy x to another
policy y, then Ui(x) > Ui(y)
• If person i is indifferent between x and y,
then Ui(x) = Ui(y)

• A policy x Pareto dominates another policy
y if two conditions are satisfied:
1. No one strictly prefers y to x, that is, for all i
Ui(x) ≥ Ui(y)
2. At least one person strictly prefers x to y,
that is, for at least one i,
Ui(x) > Ui(y).
• Unanimity on x over y

• The move from a policy y to an alternative
policy x is a Pareto improvement if x
Pareto dominates y.

• A policy x is Pareto efficient if no other
policy Pareto dominates it (negative def)
• A policy x is Pareto inefficient if at least
one other policy Pareto dominates it.
• Pareto efficiency matters because:
– The set of policies from which there is no
policy change unanimously accepted
– We know a lot about how to achieve Pareto
efficiency

• We know how to achieve Pareto efficiency
• We care about achieving Pareto
improvements
• a move to efficiency need not be a Pareto
improvement (they may exist losers).
• This involves addressing distributional
concerns
1.4 Efficiency and distribution

x y z
person 1 5 2 4
person 2 1 3 7
person 3 4 1 1
• y is (Pareto) inefficient: status quo
• x and z are Pareto efficient
• But move from y to x is not unanimously
accepted (person 2 loses)
• Compensate welfare loss by transfer
1.5 An example of policy

• A policy has two components
– An action (efficiency)
– A transfer scheme (distribution)
• Think of the action as the efficiency
oriented policy lever to be pulled (Free
trade, carbon tax)
• The transfer scheme is a redistribution of
money

• Suppose you can separate person i's utility
from a policy (a; t) into two components
– Payoff from action a is vi(a)
– Payoff from transfer ti is simply ti
Ui(a; t) = vi(a) + ti
• Transferable utility (money = utility)
• Separability of policy (efficiency) and
transfer (distribution)

• Think about the sum of the vi's as the
amount of utility pie"
• Suppose one action yields a small pie the
other yields a large pie
• Can use transfers to create any division of
the pie you like
U2
Slope=-1
U1

U2
Slope=-1
U1
1.6 Teacher pay

• Two people: the School’s Principal (S) and
the Teacher's Union (T)
• Two actions:
– Using test scores to evaluate teacher
performance and determine pay, called Pay for
Performance (P)
– Paying teachers solely based on education and
seniority, called Seniority Pay (S)
1.6 Teacher pay

• Without transfer t=(0,0) and a=(P,S)
• School’s principal prefers pay for
performance
US(P; (0; 0)) = 10
US(S; (0; 0)) = 2
• Teacher's Union prefers seniority pay
UT (P; (0; 0)) = 1
UT (S; (0; 0)) = 6
1.6 Teacher pay

• It may be possible to get both to agree to a
move to performance pay by transferring
money to the teachers
– raising average salary
• Enough of a transfer might compensate the
teachers union for the utility loss from
adopting performance pay
1.7 Free Trade or not?

• Split policy choice in two
– Efficiency: choose action to maximize “pie”.
– Distribution: Use transfers to compensate any
losers.
• Trade Adjustment Assistance(TAA ) in the Trade
Expansion Act of 1962 to help workers and firms adjust to
disruption caused by trade liberalization. “It is justified now,
as it was then, on grounds that the government has an
obligation to help the 'losers‘ of policy-driven trade
opening.(Congressional Research Service).
1.7 Free Trade or not?

Two People: Capital (C) / Labor (L)
Two actions: Free Trade (F) / Protectionism (P)
vC(F) = 12 vC(P) = 4
vL(F) = 2 vL(P) = 9
• Free trade is efficient
• Compensate losers tl=-tc=+7
1.8 efficiency and utilitarism

• Under utilitarianism, prefer (x; t) to (y; t’) if
and only if
U1(x; t) + U1(x; t) > U1(y; t’) + U2(y; t’)
v1(x) + t1 + v2(x) + t2 > v1(y) + t1’+ v2(y) + t2’
• Balanced budgets: t1 + t2 = t1’ + t2’ = 0
v1(x) + v2(x) > v1(y) + v2(y)
• Transferable utility: transfers don't matter for
utilitarianism, only efficiency matters
1.8 efficiency and utilitarism

• Once at a utilitarian optimum, there is no
way to create more utility
• Any policy change either reduces the total
amount of utility or, at best, redistributes it
• This must make at least one person worse
off
1.9 Policy in practice

• Separation efficiency and distribution
– Choose policy to max total utility
– Compensate losers with budget balanced
transfers
• Relies on two critical assumptions
– Correct transfers will be chosen
– Transferable utility (utility=income)
1.9 Transfers Constraints

• Technological
– Collecting transfers is hard or expensive
• Informational
– Who are the winners and losers?
• Economic/behavioral
– Transfers may induce behavior distorsions
• Political
– losers may lack power (future generations)
1.9 Utility is not money

• Transferring money from one person to
another need not imply transferring an equal
amount of utility
• Suppose the action hurts those who value
money very little
• Have to transfer a lot of money to make up
for utility loss
• Take the money from people who value it a
lot Presentation slides for J. Hindriks and G. D. Myles, Intermediate Public Economics, Second Edition
1.9 Equity beyond efficiency

• We've concentrated on efficiency with
compensation transfers
• Redistribution policy: social security,
progressive taxation, poverty policy
• Redistribution limited by incentive,
information and mobility constraints.
• Optimal redistribution=Equity vs efficiency
• Information revolution (from Edgeworth
1919 to Vickrey 1942 and Mirrlees 1971)
1.9 values beyond utility

• We've slipped in a welfarist
consequentialism
• Pareto efficiency are unequivocally good if
the only thing we care about is utility
• There are perfectly sensible normative
frameworks that would reject some policies
that raise everyone's utility
• Values beyond welfare
1.9 Good policy?

• Which policies improve welfare the most
(prof Nathan Hendren, Harvard)
• The MVPF approach
Policyinsigths.org
Policyimpacts.org
1.10 Evidence based policy
Evidence-Based Medicine
Effectiveness
gap
Cost gap
Evidence-based medicine
Prostate cancer costs
– Prostatectomy $7,300
– Brachytherapy $19,000
– Radiation therapy $46,900
Prostate cancer effectiveness
Pelroth-Goldman-Garber (2011): same effect
=>Gains from deceptive practices
=>Find way to « nudge » patients away from
wasteful costs (Obama strategy)
“Following an acrimonious health care reform debate
involving charges of "death panels," in 2010, Congress
explicitly forbade the use of cost-effectiveness analysis in
government programs of the Patient Protection and
Affordable Care Act. In this context, comparative
effectiveness research emerged as an alternative strategy
to understand better what works in health care. Put simply,
comparative effectiveness research compares the efficacy
of two or more diagnostic tests, treatments, or health care
delivery methods without any explicit consideration of
costs.”
(Chandra, Jena & Skinner 2011)
Cost effectiveness
– Perception of rationing
– Cost easier to measure
Comparative effectiveness
– Perception of rationalizing
– effectiveness harder to measure
=> Widespread political/public resistance to the
idea of using prices (and costs) in health care
Ashton and Wray 2013

« we know how to generate high-quality evidence about the
benefits and harms of clinical interventions. But evidence can make
winners and losers of parties that have a stake in the decision that
will flow from the evidence. Consequently, though this flies in the
face of scientific objectivity, medical ethics, and the high ideals of
methods and ethics of human subjects research, parties will try to
manipulate the design and conduct of research so that the process
yields answers favorable to their interests. »
Types of public policy
• Market oriented: leveraging self- interest
(cap and trade carbon tax)
• Behavioral: leveraging behavioral
mistakes (nudging)
• Regulatory: constraining freedom of
choice via rules and regulations
(lockdown)
• Moral: prioritizes fairness and equity with
deterministic of human initiatives (The Veil
of Ignorance)
Regulation
• The US federal regulatory code lists
12,000 restrictions and rules for apple
orchards, regulations that have put many
producers out of business
• The US Preschool Development Grants
allocate over $750 million to children in
disadvantaged communities: applicants
will be required to comply with 2,400
“performance standards”
Bureaucracy
• Excessive regulation has created a
nightmare of meaningless paperwork,
oversight and requirements that have
further eroded what little humanity is left in
medicine
France saucepan with a roof
• France public decree No. 2014-797
requiring that restaurants display a
“saucepanwith-a-roof” symbol to denote
home-made food.
• Frozen pommes frites are certified as
house-made as long as they are baked in
an oven at the restaurant. Similarly, frozen
vegetables can be labelled as home-made
as long as a house sauce is added.
Uber regulation challenge
• We don't have to beg for forgiveness because
we are legal (Travis Kalanick 2014)
• The old way of creating services and
regulations around producers doesn't work
anymore. If you design systems around
producers it means more rules and laws and
those become quickly out of date, and
privilege the groups that were the best
political lobbyists when the laws were written.
(Neelie Kroes 2014)
Chapter 2
Equilibrium and Efficiency
Reading
• Essential reading
– Hindriks, J and G.D. Myles (2013) Intermediate Public Economics.
(Cambridge: MIT Press) Chapter 2.
• Further reading
– Duffie, D. and H. Sonnenschein (1989) ‘Arrow and General Equilibrium
Theory’, Journal of Economic Literature, 27, 565 - 598.
– Koopmans, T.C. (1957) Three Essays on the State of Economic Science.
(New York: McGraw-Hill).
– Ng, Y.-K. (2004) Welfare Economics. (Basingstoke: Palgrave
Macmillan).
• Advanced reading
– Debreu, G. (1959) Theory of Value, (New York: Wiley).
– Mirrlees, J.A. (1986) ‘The theory of optimal taxation’ in K.J. Arrow and
M.D. Intrilligator (eds) Handbook of Mathematical Economics
(Amsterdam: North-Holland).
2.1 Introduction
• Adam Smith’s “invisible hand” described the link
between competition and efficiency
– Decisions motivated by individual gain produce a socially
efficient outcome
• The competitive model captures this independent
decision-making
• Equilibrium is achieved by the adjustment of prices
• Prices act as the coordinating mechanism for
individual decisions
2.2 Economic Models

• A model is a simplification designed to capture
essential aspects of the problem under investigation
• Most models begin with a specification of objectives
for the economic agents
• Equilibrium is found and the effect of policy choice is
determined
• The “art” of successful modeling is the choice of the
level of detail
• There is typically a trade-off between generality and
clear predictions
2.3 Competitive Economies

• The fundamental competitive assumption is that
firms and consumers do not believe they can affect
prices
– This is justified when economic agents are negligible in
size
– It can always be imposed as an assumption on behavior
• Prices are central to the analysis
– Prices measure values
– Act as signals to consumers and firms
– Adjustment of prices equates supply and demand
2.3 Competitive Economies

• Information is central to the functioning of an
economy
– It is assumed in this chapter that all agents have symmetric
information
– The consequences of relaxing this are investigated in
Chapter 10
• Symmetric information permits uncertainty as long
as all agents are equally uninformed
– Future profits of firms may be uncertain
– But the directors must be no better informed than the
shareholders
2.4 The Exchange Economy

• The exchange economy models the simplest form of
economic trading
• Assume there are two consumers
• Each consumer has an initial endowment of the two
commodities
– The consumers exchange quantities of the two
commodities to achieve an improved consumption plan
• Market prices determine the rate of exchange
• Consumers believe their choices cannot affect prices

• The initial endowment of consumer h is
 h = 1h , 2h ( )
• When prices are p1, p2 a consumption plan
(
x h = x1h , x2h )
satisfies the budget of h if
p1x1h + p2 x2h = p11h + p22h
• The consumption plan maximizes the utility function
(
U h x1h , x2h )
subject to the budget constraint

• The Edgeworth box depicts the set of feasible
consumption allocations
– Feasible allocations are represented as points in a
rectangle
– The lengths of the sides are equal to total endowments
• Consumptions plans for the two consumers are
measured from the corners of the box
– The south-west corner is the zero consumption point for
consumer 1
– The north-east corner is the zero consumption point for
consumer 2

• A pair of consumption plans x12
2
is feasible if
x1i + xi2 = i1 + i2 , i = 1,2 x12 •
x
x22
 12 +  22
• Feasible plans are shown in •

the Edgeworth box in Fig.

2.1 1
x11
11 + 12
• All feasible plans are
located in this box
• The common endowment
point is denoted  Figure 2.1
Typical Edgeworth Box
• Point x is a feasible
consumption allocation

• The consumers have a 2
common budget constraint

that passes through the x2
x1
endowment point 
• The gradient of the budget

1
constraint is
− p1 / p2
• Given this budget x1 and x2
are utility-maximising Figure 2.2
choices Preferences and demand

• In equilibrium supply equals 2
demand
• Prices adjust to achieve this x2
x1
• Plans x1 and x2 do not form 
an equilibrium
• Fig. 2.3 shows the effect of 1
an increase in p1 relative to
p2
• The budget constraint
pivots about the Figure 2.3
endowment point Relative price change

• An equilibrium is a position where supply is equated
to demand
• Prices adjust until the indifference curves have a
common point of tangency on the budget line
– When this is achieved demand is equal to supply
– These are called equilibrium prices
• Given consumer demand functions
the equilibrium prices solve xi = xi ( p1 , p2 )
h h
x1i ( p1 , p2 ) + xi2 ( p1 , p2 ) = i1 + i2 , i = 1,2

• An equilibrium is shown in 2
Fig. 2.3
• The budget constraint is at 1
x2
a point of common x

tangency with the
indifference curves 1
• The sum of demands from

the two consumers is equal
to the sum of endowments
• Does such an equilibrium Figure 2.4
always exist? Equilibrium

• Observe that only relative prices matter in
determining the level of demand
• The budget constraint can be written as
p  h
+ p  h
p1 h  p1 h h p1 h
x2 =
h 1 1 2 2
− x1 =  1 +  2  − x1
p2 p2  p2  p2
• This is determined by relative prices p1/p2
• As a consequence demand is homogenous of degree
zero
xih ( p1, p2 ) = xih (p1, p2 ), for any   0

• Consumers are concerned with real purchasing
power
– Nominal income is equal to the value of endowment
– A change in the level of prices raises nominal income and
the cost of purchases equally
• The implication is that a price normalization must be
adopted to fix the level of prices
• One normalization is to select a numeraire and fix its
price equal to 1
– The numeraire becomes the unit of account

• Define excess demand by
Z i = xi1 + xi2 − i1 − i2
   
• Then
p1Z1 + p2 Z 2 = p1 x11 + x12 − 11 − 12 + p2 x12 + x22 − 12 − 22 = 0
• This is Walras’ law: the value of excess demand is
zero
• If demand is equal to supply for good 1 then demand
must also equal supply for good 2
– Equilibrium in one market necessarily implies equilibrium
in other second market

• The existence of an equilibrium can be demonstrated
as follows:
– Choose good 1 as the numeraire (so p1 = 1 )
– Plot the excess demand for good 2 as a function of p2
– An equilibrium is occurs where the graph of excess
demand crosses the horizontal axis
– If the graph starts above the axis and finishes below there
must be a crossing point
– When the excess demand for good 2 is zero by Walras’ law
excess demand must also be zero for good 1

• Fig. 2.5 illustrates the
Z 2 (1, p2 )
argument that an
equilibrium exists
• Excess demand for good 2 is p2
positive when p2 is low

• Excess demand is negative
when p2 is high
• If the excess demand
function is continuous it
must cross the axis
Figure 2.5
• Generally there will be an Equilibrium and excess demand
odd number of equilibria
2.5 Production and Exchange

• Adding production to the exchange economy
provides a complete model of economic activity
• Production is undertaken by firms which aim to
maximize profit
• Some goods are initial endowments (e.g. labor),
some are intermediate inputs, and some are final
consumption goods
• The fully-developed model is called the Arrow-
Debreu economy

• Firms use inputs to produce outputs with the
objective of maximizing profit
• Each firm has a production technology
– The technology describes feasible input-output
combinations
• The profits of the firms are distributed as dividends
to consumers
– Consumers hold shares in firms
– The dividend received is proportional to shareholding

• A firm’s technology is
described by a production Good 2
set, denoted Yj
• A typical production set is Y j
3
shown in Fig. 2.6
• Inputs are negative (a -2 Good 1
subtraction from economy’s

stock) and outputs are
positive (an addition)
• The production plan (-2, 3) Figure 2.6
is feasible Typical production set

• The profit of firm j from
production plan (y1j , y2j ) is  =0
 0
Good 2
 j = p1 y1j + p2 y2j  0
y*
• Isoprofit curves show all
production plans giving a p2
constant level of profit Good 1

p1
• The curve for p = 0 always

goes through the origin
• Profit is maximized on the
highest attainable isoprofit
curve Figure 2.7
• This is at production plan y* Profit maximization
in Fig. 2.7

• Profit maximization by firm j implies a supply of good
i given by yij ( p ) and a level of profit
 j = in=1 pi yij ( p )
• With m firms the aggregate supply for good i is
Yi ( p ) =  mj=1 yij ( p )
• Shareholding of consumer h in firm j is  hj and hH=1 hj = 1
• Including dividends the budget constraint of
consumer h is
i =1 i i
n
p x h
= in=1 piih + mj=1 hj j

• Summing over consumers gives the level of
aggregate demand
X i ( p ) = hH=1 xih ( p )
• Excess demand including the supply from firms is
Z i ( p ) = X i ( p ) − Yi ( p ) −  hH=1ih
• Equilibrium occurs when Z i ( p ) = 0 for all goods i
• Equilibrium can be shown to exist under fairly weak
conditions
2.6 Efficiency of Competition

• Efficiency is about the best use of resources
– Efficiency is achieved when more cannot be obtained with
existing resources
• If there is a single decision-maker an efficient
allocation will maximize utility
– This is illustrated by the single-consumer economy
• Characterizing efficiency is more complex when
there are multiple decision makers
– Competing preferences must be resolved
– Pareto-efficiency is one solution to this difficulty
– Chapter 13 considers other solutions

• Fig. 2.8 shows utility Good 2
maximization in a single ~
consumer economy x*
1
• Good 1 is supplied by the
consumer to the firm 
• Good 2 is supplied by the p1 Good 1
firm to the consumer

• The consumer receives
dividend 
• Given the budget constraint Figure 2.8
utility is maximized at ~
x* Utility maximization

• The equilibrium of the Good 2
economy is shown in Fig.
2.9 ~
x* = y *
*
• The production plan y is 1
profit maximizing given the 

prices
• Consumption* ~x *is equal to
p1 Good 1
production y
• The equilibrium is efficient
– The consumption plan is on
the highest attainable
indifference curve given the Figure 2.9
production set Efficient equilibrium

• The gradient of the indifference curves is the
marginal rate of substitution MRS1,2
– The rate at which good 1 can be traded for good 2 holding
utility constant
• The gradient of the production frontier is the
marginal rate of transformation MRT1,2
– The rate at which good 1 can be transformed into good 2
• At the efficient equilibrium the two gradients are
equal so
MRS1, 2 = MRT1, 2

• Utility maximization equate the gradient of the
indifference curve and budget constraint
MRS1,2 = p1 p2
• Profit maximization equates the gradient of the
production frontier to an isoprofit curve
MRT1, 2 = p1 p2
• The consumer and the firm react to the same prices
so
MRS1,2 = p1 p2 = MRT1,2
• Prices succeed in coordinating decisions

• Constant returns to scale  =0 Good 2
provide some simplification
• In equilibrium the firm must ~
x* = y *
earn zero profit so the price 1
vector is orthogonal to the

production frontier p1 Good 1
• The budget constraint goes
through the origin and is
coincident with the
production frontier
• This is illustrated in Fig. 2.10 Figure 2.10
Constant returns to scale

• If there is more than consumer a definition of
efficiency must accommodate competing
preferences
• Pareto efficiency is a test of efficiency that can be
applied with many consumers
• Pareto efficiency searches for unexploited gains
– Is there a feasible reallocation of resources that can benefit
at least one consumer without harming any other?
– If no improving reallocation can be found the initial
position is Pareto efficient

• Pareto efficiency can be defined formally
• An allocation x is feasible if x = y + 
• Definition 2.1 A feasible consumption allocation x̂ is
Pareto-efficient if there is does not exist an
alternative feasible allocation x such that:
– Allocation x gives all consumers at least as much utility as
x̂
– Allocation x gives at least one consumer more utility than
x̂
• Note how Pareto-efficiency is defined by a negative
property and sidesteps distribution

• The Two Theorems of Welfare Economics are the
basis for claims concerning the desirability of the
competitive outcome
• The First Theorem states that a competitive
equilibrium is Pareto-efficient
• The Second Theorem states that any Pareto-efficient
allocation can be decentralised as a competitive
equilibrium
• These theorems are based on applying Pareto-
efficiency to the competitive economy

• In Fig. 2.11 allocation a is
not Pareto efficient 2
– The move to b raises the c

utility of both consumers d
b
• Allocation c is Pareto- a
efficient 1
– Any change in allocation must

make at least one consumer
worse-off
– At c MRS1, 2 = MRS1, 2
1 2
• There are many Pareto- Figure 2.11

Pareto-efficiency
efficient allocations
– d illustrates a second

• Pareto-efficient allocations
occur at the tangency of 2
two indifference curves

• The locus of tangencies is e
the contract curve 
• In Fig.2.12 the competitive 1

equilibrium is at point e Contract Curve
• The budget line is a

common tangent
• The competitive equilibrium
is Pareto-efficient Figure 2.12
The First Theorem

• Each consumer maximizes utility given the prices so
MRS11, 2 = p1 p 2 = MRS12, 2
• This equality of the MRSs is the Pareto-efficiency
condition
– Individual decisions are coordinated via prices
– This is achieved through individual optimization
• Theorem 2.2 (First Theorem of Welfare Economics)
The allocation of commodities at a competitive
equilibrium is Pareto-efficient

• The Second Theorem asks: can a Pareto efficient
allocation be made into a competitive equilibrium?
– This is the process of decentralisation
• Decentralisation is possible if the consumers’
indifference curves are convex.
– The common tangent to the indifference curves at a
Pareto-efficient allocation provides the budget constraint
– Given the budget line convexity ensures that the Pareto-
efficient allocation is the optimal choice of the consumers
– Decentralization is completed by choosing a point on this
budget line as the initial endowment point

• Decentralization is
2
illustrated in Fig. 2.13
• The Pareto-efficient e'
allocation e’ is made an  '
equilibrium by: 1
– Selecting ’ as the
endowment point
– Allowing trade to move the
allocation from ’ to e’
Figure 2.13
• If the initial endowment The Second Theorem
point is  a transfer of
endowment is necessary

• The formal description of this construction is the
Second Theorem
• Theorem 2.3 (Second Theorem of Welfare
Economics) With convex preferences, any Pareto-
efficient allocation can be made a competitive
equilibrium
• The important step is placing the economy at the
correct starting point
• This requires a process for redistributing initial
wealth

• The extension of the
Second Theorem to
production is shown in Fig. Z
2.14 W x̂
Price
• W is the set of feasible x̂1

Line
production plans x̂ 2
Feasible
• Z describes quantities of Set
the two goods that would

permit a Pareto-
improvement over xˆ1 , xˆ 2   Figure 2.14
• The price line separates W Proof of the Second Theorem
and Z
2.7 Lump-Sum Taxation

• It is implicit in the Second Theorem that wealth is
reallocated
• The practical value of the Second Theorem depends
on the possibility of redistribution
• The Theorem sees this as achieved through lump-
sum transfers of endowments between consumers
– A transfer is lump-sum if no change in a consumer’s
behaviour can affect the size of the transfer
– The transfer is optimal if the resulting equilibrium is the
policy maker’s most preferred outcome

• A lump-sum transfer is
2
shown in Fig. 2.15
• At the initial point the e
income level of h at the ' 
prices p̂ is p
h
ˆ ~
1 x11
• The value of the required

transfer to consumer h is
pˆ  h '− pˆ  h
• Achieved by transferring a Figure 2.15
quantity ~x11 of good 1 Lump-sum transfer

• Lump-sum transfers can be rephrased in terms of
lump-sum taxes
• Consumer 1 pays tax of amount T 1 = pˆ ~x11
• The tax revenue is given to consumer 2

• This pair of taxes moves the budget constraint in
exactly the same way as the lump-sum transfer
• The taxes are also lump-sum
– The value of the taxes cannot be affected by any change in
behaviour

• Lump-sum taxes have a central role in public
economics
• They achieve distributional objectives without
reducing the total endowment
– Redistribution is achieved with no efficiency cost
• If they can be employed in the manner described
they are the perfect taxes
– But in practice information limits the use of lump-sum
taxation
– This is explored in Chapter 13
2.8 Discussion of Assumptions

• The competitive economy is based on a set of strong
assumptions
• Competitive behavior rules out monopoly power but
does not explain who sets prices
• Asymmetry of information can develop through
product experience
• Market failure occurs when the assumptions are not
satisfied and efficiency is not achieved
• Chapters 6 – 10 analyze forms of market failure
Chapter 3
Behavioral Economics
3.1 Outline
Welfare
Policy
Market
Individual
3.1 Introduction
• Traditional public economics assumes that
people know what is best for themselves and
they can get themselves to act according to their
own best interest.
– rational individuals and (irrational?) markets
• People can make invalid choices or mistakes
• We consider public intervention motivated by the
fact that people make invalid choices or
mistakes
– Irrational individuals and (rational?) markets.
Bias and Noise
• Distribution of errors
bias
noise
x
x=0 E(x)
Mean Square Error (MSE)
𝑣𝑎𝑟 𝑥 = 𝐸 𝑥 2 − 𝐸(𝑥)2
𝐸 𝑥 2 =𝑣𝑎𝑟 𝑥 + 𝐸(𝑥)2
𝑀𝑆𝐸=𝑁𝑜𝑖𝑠𝑒 2 + 𝐵𝑖𝑎𝑠 2
3.1 Introduction
• Paternalism represents policies aimed at benefiting
individuals who cannot be relied upon to pursue self-
interest (authority)
• Welfarism represents policies intended to benefit
individuals when self-interest cannot be relied upon
because of the presence of market failures or the need
for redistribution. (externality)
• Behavioralism represents policies intended to benefit
individuals when self-interest cannot be relied upon
because of the presence of internal conflicts (internality)
3.1 Introduction
• People can make two types of mistakes that can
motivate public intervention
• Lack of knowledge:
– people cannot easily distinguish healthy and non-
healthy food
– policy enforcing effective “healthy” food labeling
• Lack of (self-)discipline
– People cannot get themselves to act according to
plan
– Policy helping them to act in the correct way.
Obesity and Info
Price of food Price of food fell
rose 1% during 14% during this
this period! period!
Estimated that reduction in food price accounts for 42% of increase in young
adults’ BMI 1981-1994
Helping Consumers Use Nutrition Information:
Effect of Format & Presentation
Downs, Wisdom & Loewenstein 2015
American Journal of Health Economics

https://doi.org/10.1162/AJHE_a_00020
11
Power of Info %
% Daily
Daily
Min. on
Snack
Item Calories
Treadmill*
Calories*
• Control (no Lays Classic Potato Chips 12%
23
230
115%
information)
• Numerical Values: Snickers candy bar 14%
28
280
140%
– Calorie Info Only York Peppermint Pattie 7%

14
140
70%
– Calorie Info + Daily Intake
Reference 2000 Apple Crisps 2%4
40
20%
– Calorie Info + Daily Snack

Choc. Chip Cookie 2 pack 17%
34
340
170%
Intake Reference 200
– % of Daily Calories Lays Baked Potato Chips 7%
13
130
65%
– % of Daily Snack Calories
– Minutes on a Treadmill Hostess apple pie 24%
47
470
235%
*Based onexperts
Nutrition
*To burn
*Based nutrition
the
experts experts’
recommend
calories
on nutrition therecommendation
200
inexperts’
recommend snack calories
that(this
men per of
assumes
should
recommendation day200
afor
you
eat calories
snacks
burn
about
2,000 per
or
102,400day
calories
calorie for
dessertpersnacks
(10% of aorday,
minute).
percalories
day per
diet.
and dessert
women(10%2,000 ofcalorie
should aeat
2,000percalorie
about day percalories
diet).
2,000 day diet).per day.
12
Power of Info (continued) Traffic

Nutrition
Light
Expected
Item Rating*
Grade*
Body Size*
Calories
• Heuristic Cues Lays

AppleClassic
Crisps Potato Chips C
40
– Nutrition Grade
Snickers candy
Lays Baked bar Chips
Potato C
130
– Expected Body Size
York Peppermint Pattie B
140
– Traffic Light Rating
• Ascending Order Apple Crisps Potato Chips
Lays Classic A
230
– No Info (Control) Choc. Chip

Snickers Cookie
candy bar 2 pack D
280
– Calorie Info Lays

Choc.Baked Potato 2
Chip Cookie Chips
pack B
340
Hostess apple pie F

470
*For someone
*Based who eats
on nutrition this recommendation
experts’ snack routinely over time,
of 200 based per
calories on nutrition
day for
experts’
snacksrecommendation
or dessert (10%ofof200 calories
a 2,000 per per
calorie dayday
for snacks
diet)
or dessert (10% of a 2,000 calorie per day diet)
13
Preliminary results…
Significant overall result of format, F(2, 591) = 3.46, p = .03
240
Calories of Chosen Snack
220
200
180
160
140
120
100
Control Numerical Heuristic Cues
Values
• Overweight vs. Normal Weight Participants
– No effect in normal weight sample, F(2,382) = 0.64, p = .53
– Significant effect in overweight sample, F(2,204) = 4.73, p = .01
280
Calories of Chosen
260
240
Snack
220
200
180
160
140
120
100
Control
Numerical Values
Heuristic Cues Control
Numerical Values
Heuristic Cues
Will Power
Will Power
– Without commitment device (control)
– With Commitment devices (deposit or Lottery)
3.2 Behavioral individuals
• Three forms of deviation from eco rationality
– Incorrect preferences
– Incorrect beliefs
– Social preferences
• Imperfect behavior and individual mistakes may

give the government new opportunities to step in
with policy interventions.
• The PBC Model of Choice
preferences
choices
Beliefs Constraints
INCORRECT PREFERENCES
INCORRECT CHOICES
PRESENT BIAS:
• people face a choice that is liable to change
their own preference in the future
• Addiction:
– The initial decision to start the addictive activity, leads
to a future decision of whether to continue the
addictive activity (Initial preference is to give it a try
and then to quit)
– Because of the addictive nature of the activity, future
preferences are changed and people are hooked to
the addiction.
• Preference reversal between today self and
future self (internal conflict).
• There is one individual self today and an entirely
different individual self in the future.
• Naïve people ignore this preference reversal
and (wrongly) believe they can stop later on.
• Sophisticated (rational) people anticipate this
preference reversal and choose not to start with
addiction.
The (β, δ) Model of Self-Control
U = u0 + β(δu1 + δ2u2 + . . . + δTuT ),

now later
where δ < 1 is the standard discount rate and β ≤ 1

is self-control discounting (now/later)
• The discount rate between any two periods in
the future is δ
• The discount rate between the present and the
immediate future is βδ
• consumption decision involving
– utility u1 at t = 1 and
– delayed utility u2 at t = 2.
• Investment good must trade off

the cost u1 < 0 against a future benefit u2 > 0.
• Temptation good must trade off
the benefit u1 > 0 against a future cost u2 < 0
• From ex-ante perspective, If the consumer could
commit to a choice in advance, say at time t = 0,
she would consume if
β(δu1 + δ2u2 ) >0
or
u1 + δu2 >0
• However, the consumer actually consumes at
time t = 1 if
u1 + βδu2 ≥ 0.
• Naïve consumer with self-control problem (β<1)
will underweight period 2 since βδ<δ
– underconsumes investment goods (with delayed

benefit u2 > 0) and
– overconsumes temptation goods (with delayed cost u2
< 0),
• Sophisticated agent without self-control problem
(β=1) will consume according to the plan.
• Going to the Gym
• Assume that β= 1/2 and δ= 1.
• Suppose that current effort is 8 and generates
delayed benefits of health improvement 10.
• Will you go?
• Gym Now: -8 + 1/2 [10] = -3
• Gym Later: 0 + 1/2 [-8 + 10] = +1
Agent would prefer going later, so agent won’t
exercise without commitment.
• Empirical evidence (Della Vigna & Malmendier
2006)
– Average cost of gym membership: $75 per month
– Average number of visits: 4
– Average cost per visit: $19
– Cost of “pay per visit”: $10
• Membership is costly commitment devise but

help to solve the self-control problem
Immediate Reward
Source: Tali Sharot (2017) the influential mind

REFERENCE-DEPENDENCE
• Standard model: people only value income
levels.
• Reference-dependent model: People value
income change () as well as income levels ($)
• preferences over income, x relative to reference
point r :
v(x; r) = x − r for x ≥ r,
= λ(x − r) otherwise.
• Loss aversion: For λ>1, the value function is
steeper for losses below the reference point (x
<r) than for gains over the reference point (x>r).
• Loss aversion can also explain endowment
effect:
– homeowner values the sale according to the extent of
deviation from the purchase price.
– If there is loss aversion, the homeowner will
overweight a price loss compared to a price gain.
• Stock market sell “winners” and hold back
“losers”
INCORRECT BELIEFS
INCORRECT CHOICES
Rational ignorance?
Optimism bias
Source :Chowdhury et al (2014 )Psychological Medicine, 44, 2003–2012

Incorrect beliefs
• Beliefs >< preferences
• Incorrect preference=> choice mistakes
• Incorrect beliefs => choice mistakes
• Mistakes in the formation of beliefs
– Gambler’s fallacy
– Confirmation bias/ hindsight bias
– Overconfidence bias
Gambler’s fallacy
Let {h, t} represents a lottery that pays $h if the
next flip of a coin comes up heads (H) and $t if the
next flip comes up tails (T )
– If the person has observed no flips before, she
chooses a lottery such that h = t , considering that
heads and tails are equally likely
– If the person has observed HHH, she chooses a
lottery such that h<t, considering that heads are less
likely than tails…which is wrong!
Gambler’s fallacy in our classroom
In our test (version 1): players are endowed with 10 tokens;

the roulette pockets are half Red, half Black. Each token
bet on the winning color will be converted in 1 €. How many
tokens should be bet on Black?
– Players that have not observed anything before (C control) have

bet on average 6,05 tokens on Black
– Players that have observed Red for three times in a row (D
Treatment) have bet on average 5.625 tokens on Black
– Players in group D were only slightly influenced by the
observation and actually behaved more “rationally” than players
in group C
Gambler’s fallacy in our classroom – version 1
70%
60% 50% Red - 50% Black

Group D Group C (control)
50%
40%
30%
20%
10%
0%
1 2 3 4 5 6 7 8 9 10
Black Tokens (out of 10)

Gambler’s fallacy in our classroom
In our test (version 2): players are endowed with 10 tokens; the
roulette pockets are 40% Red, 60% Black. Each token bet on the
winning color will be converted in 1 €. How many tokens should
be bet on Black?
– The strategy that maximizes expected gain is to play all tokens on

Black ( 0 x 40% + 10 x 60% = 6 €), but a prudent player might prefer
to bet fifty-fifty as she gets 5 € for sure.
– Players who have not observed anything (B Control) have bet on
average 6,64 tokens on Black.
– Players that have observed Red for three times in a row (A
Treatment) have bet on average 6,9 tokens on Black.
– Players in Group A were influenced by the observation
Gambler’s fallacy in our classroom – version 2
40%
35%
40% Red - 60% Black Group A Group B (control)
30%
25%
20%
15%
10%
5%
0%
1 2 3 4 5 6 7 8 9 10
Black Tokens (out of 10)
Confirmation bias is inferring less than is justified
from the observation of recent events
Biased updating of information by
– overweighting information that confirms initial opinion
– Underweighting information that contradicts initial
opinion : wishful thinking (not drinking)
Hindsight bias is inferring more than is justified
from the observation of a recent event
– rapidly rewrite our memory of the past to fit what we
have just learned (e.g. sport, politics, epidemic
outbreak)

Overconfidence bias
Overconfidence bias is the tendency to
overestimate one’s own (relative) abilities and
expect the resulting outcomes to be better than
they will be
– people overestimate the output they can generate or
the productivity of their effort.
– Reckless drivers: Young and male drivers
overestimate their driving skill (less so for women and
mature drivers!)
– Women are less overconfident which may explain
why they work harder at school?
Framing bias occurs when different but equivalent
descriptions of a decision problem lead to
systematically different decisions
Evidence for incoherence in individual choices.
– Risk framing
– Attribute framing
– Goal framing
– Visual framing (Titchener illusion)
Risk framing: The tropical disease problem
Preparing for outbreak of an unusual tropical
disease, which is expected to kill 600 people.
Make a choice among sure vs risky options
Framing 1 “gain perspective”:
A: If this program is adopted, 200 people will be saved.
B: If this program is adopted, there is a one-third
probability that 600 people will be saved and a two-
thirds probability that no people will be saved.
Framing 2: “loss perspective”
C: If this program is adopted, 400 people will die.
D: If this program is adopted, there is a one-third
probability that nobody will die and a two-thirds
probability that 600 people will die
A=C : 200 people saved =
400 people will die (sure thing)
B= D: one-third probability to save everybody =
one-third probability nobody will die.
Participants tend to prefer
– the “sure option” A when given options A and B,
– but tend to prefer the risky option D when offered
options C and D
Participants appear to be
– risk-averse for gains (A > B),
– risk-seeking for losses (C< D)
Tropical disease in our classroom
Group A and C % Reponses (93 ppl) VS % Reponses (95 ppl) Group B and D
Programme A 77% 45 % Programme C
Programme B 23% 55 % Programme D

Tropical disease in our classroom
Group A and C % Reponses (125 ppl) VS % Reponses (84 ppl) Group B and D
Programme A 81% 48 % Programme C
Programme B 19% 52 % Programme D

Attribute framing
objects described in terms of a positive attribute
are generally evaluated more favorably than
equivalent objects described in terms of
negative attributes
Goal framing
More likely to participate in activity when
disadvantage of not participating are stressed
rather than advantage of participating
Perception bias
– Titchener illusion,
– Choices are sensitive to contrast

– Regret : comparison with foregone choices
– Disappointment: comparison with unrealized outcomes.
– More choices is less choices
Economist Subscription
You are very passionate about Economics and Business and you decided to buy
a subscription to The Economist. What type of subscription would you buy?
Group B and C
Group A and D
(Control)
a) Economist.com subscription, 59€:
One-year subscription to Economist.com.
77% 80%
Includes online access to all articles from
The Economist since 1997.
b) Print subscription, 125€:
One-year subscription to the print edition of 2% (not available)
The Economist
c) Print & web subscription, 125€:
One-year subscription to the print edition of
The Economist and online access to all articles
21% > 20%
from The Economist since 1997.
Economist Subscription
You are very passionate about Economics and Business and you decided to buy
a subscription to The Economist. What type of subscription would you buy?
Group B and C
Group A and D
(Control)
a) Economist.com subscription, 59€:
One-year subscription to Economist.com.
77% 80%
Includes online access to all articles from
The Economist since 1997.
b) Print subscription, 125€:
One-year subscription to the print edition of 2% (not available)
The Economist
c) Print & web subscription, 125€:
One-year subscription to the print edition of
The Economist and online access to all articles
21% > 20%
from The Economist since 1997.
A weekend in Paris or in Rome?
Control (A-D): Group B Group C

« Paris or Rome Additional option: Additional option:
all-included » « Rome, no breakfast » « Paris, no breakfast** »
(106 observations) (39 observations) (44 observations)
Rome 81% 82% 79%
Paris 19% 18% 21%

Rome 81% 82% 79%
Paris 19% 18% 21%
* Two people have chosen the option Rome with the coffee not included.

Rome 81% 82% 79%
Paris 19% 18% 21%
- The addition of « Paris, no breakfast » in Group C increases the

peference for Paris (21% vs 19 % in the control group).
** The option Paris with coffee not included has not been selected by any respondent

all-included » « Rome, no breakfast* » « Paris, no breakfast* »
Rome 68% 63% 59%
Paris 32% 35% 41%
- The addition of « Rome, no breakfast » in Group B decreases the

preference for Rome (63% vs 68% in Control group) .
- The addition of « Paris, no breakfast » in Group C increases the

peference for Paris (41% vs 32 % in the control group).
* In Group B one person chose Rome - breakfast not-included.

Conformism bias
– Individuals compare to each others
– Conformism in opinions (beliefs) / in choices (actions),
– Lack of autonomy: agents may renounce their choices
(or opinions) and conform to an erroneous choice or
opinion of the others (e.g. Asch’s experiment)
U=u(xi)-p(xi,m)
where p(xi,m) represent deviation cost between individual

choice xi and normal (median) choice m
3.3 Behavioral Markets

Questions: Psychology and the market
– Does individual mistakes matter at the group level?

– Does individual mistakes affect the market outcomes?
“It doesn’t matter what you or I do

It’s how the whole group behaves
—Gary Becker
Questions:
• Does competitive pressure erase choice
mistakes and induce agents to make rational
choices?
• Will the market help people to make rational
choices, or will the market exacerbate choice
mistakes by exploiting irrational consumers ?
(e.g., eating disorders and the food industry)
Money pumps
• This argument establishes that an irrational
consumer (with intransitive preferences) is
doomed to bankruptcy when operating in the
market
• the market will “pump” an indefinite amount
of money out of the consumer always willing to
trade one consumption bundle for another.
Complementary mistakes
• Guessing game (illustration market bubble)
• Participants pick a number between 1-100
• The winner is the participant closest to 2/3
average pick
• No dominant strategy
• What is Nash equilibrium (when all participants
are rational)?
Nash equilibrium: by successive deletion of
dominated guesses
– Any guess above 66,6 is irrational for every player
since it cannot represent 2/3 of average guess
– But then since no guess is expected above 66,6, it is
irrational for every player to guess above 44,4 (i.e 2/3
of 66,6)
– But with no guess expected above 44,4, it is again
irrational to guess above 2/3 of 44,4 and so on until
we reach the Nash equilibrium guess of zero.
Experimental result abroad
– Some students guessed close to 100
– Many students guessed close to 33,3 (2/3 of 50)
random guess)
– Another smaller group guessed close to 22,2 (2/3 of
33,3)
– Average guess close to 33 and winning guess close
to 22.
– deviation from rationality: participants don’t play
Nash
Guessing game in our classroom
Updating 1 Updating 2
41,86 Average
34.57
27.90 Target 25.88
23.04
17.25
Cognitive hierarchy
– Level 0 rationality : guessing 50 (random guess)
– Level 1 rationality : guessing 33 in response to 50
– Level 2 rationality : guessing 22 in response to 33
– And so on
Keynes (1936) believed that similar behavior was
at work in the stock market and could explain a
market bubble:
– the price of shares is based on what people think
other people think their value is.
Coronavirus Impact on Stock
Market
The company said it may have the “potential to
treat and prevent Wuhan coronavirus”.
Rationality tug-of-war
Markets will correct irrationality if:

-consumers are receptive to advice, and
-there is more profit in protecting consumers than
taking advantage of them.
Coronavirus and misinformation
Lies spread faster than truth
• Concern over false news
• Vosoughi et al. (Science 9 March 2018) used a
data set of rumor cascades on Twitter from 2006
to 2017.
• False news reached more people than the truth;
• The fear, surprise and emotional reactions may
be responsible for the differences observed.
• “We need a vaccine against misinformation” (Dr
Mike Ryna, Head of WHO)
3.4 Behavioral policy

Question: Psychology and Public policy
– Does behavioral individual influence policy?
Welfare policy
– To correct information problems
– To correct externalities
– To regulate competition
– To redistribute resources
– To enforce law and order
Behavioral policy
– To correct/prevent individual mistakes

Externalities
– With welfare public policy, prices are wrong or misaligned
because of externalities.
– Externalities are costs that people impose on others but
do not internalize: prices do not reflect true costs to others.
Internalities
– With behavioral public policy, prices are wrong because of
internalities
– Internalities are costs that people impose on themselves
but do not internalize: prices do not reflect true costs to
themselves

Standard public policy can change behavior by
changing :
– relative prices, constraint
– budgets,
– Information pref & beliefs
Behavioral public policy can also change behavior
by
– Using people’s biases to help them make the correct
choices (e.g. default option)
– providing immediate rewards for healthy choices

Automatic enrollment
• The choice of the default option matters
• for organ donation schemes and
retirement saving plans the default option
affects participation rates,
• even though such a default neither affects
opportunities (low cost to opt out) nor
provides new information.

Organ donation:
– 85.9 % participation in Sweden (with
automatic enrollment) and
– 4.3% participation in Denmark (without
automatic enrollment)
Retirement saving plans 401(K)
– 88% participation with automatic enrollment
and contributions cluster around default 2%
– 40% participation without automatic enrollment
and dispersion of contributions.

SMarT Pension Plan (Save More Tomorrow)
• Smart pension plan was implemented in the
workplace to help workers save more
– Present-bias: willingness to save but only tomorrow
– Opportunity-cost bias: underweight (or neglect)
opportunity-cost relative to out-of-pocket cost.
– Status-quo bias: tendency not to act and so to adhere
to default option
• Workers could elect in advance a portion of their
current income, and also a possibly different
portion of future income, to be saved.

• Workers chose to save little out of current
income, they committed to save a large portion
of future income:
• Saving today out of tomorrow income solves
both the present-bias and the opportunity-cost
bias (saving out of future pay increase)
• Automatic enrollment solves the status-quo bias
• the average saving rate increased from 4.4 to
8.8 percent

• TED lecture on SMarT pension plan by its
designer Prof Benartzi (with prof Thaler)

Complementarity
• Psychology is interested in individual behavior
• Economy is interested in collective behavior
• it is the social interaction of people that will
eventually determine the final outcome of any
policy intervention
• Behavioral public policy should complement, not
replace, more substantive public policies.
• The focus of economists on people who are
“rational” remains a useful policy benchmark
3.5 Behavioral welfare

Standard welfare analysis:
Policy change=>rational response=>market
outcome=>welfare evaluation based on “revealed”
preferences : choices  preferences
Behavioral welfare analysis:
Policy change=>behavioral response =>market
oucome=>welfare evaluation based on “true”
preferences
Behavioral agents=> choices <≠> preferences

Choice-based welfare analysis
In behavioral models, agents choose from
generalized choice sets
G = (X, d)
where
X is a set of possible choices and
d is an ancillary condition that affects choice but by
assumption does not affect (true) preferences
(e.g., salience, framing, default option).

• Let C(X, d) denote choice made in a given
generalized choice sets G
• Choice mistakes and inconsistent choices imply
that different ancillary conditions d ≠ d’ lead to
different choices even if the choice set X is
unchanged
C(X, d) ≠ C(X, d’) for d ≠ d’

Welfare bounds
• How can choices be relied upon to infer partly
preference (by setting welfare bounds)?
• When the revealed preference relation P can be
defined such as x P x’ where x is always chosen
over x’ for any ancillary conditions d.
• With sufficiently many choices it is possible to
infer welfare bounds.
• Illustration with pension plan.

• Consider three different pension plans
– High contributions and benefits (H)
– Medium contributions and benefits (M),
– Low contributions and benefits (L)
• Consider the following revealed preferences
– H>M>L for framing condition d
– M>H>L for framing condition d’
• We can infer that L cannot be optimal given the
observed choices, no matter the framing
conditions.

• Optimal policy can be bounded between H and
M
• Welfare bounds and the set of optimal policies
are tight when choice is less sensitive to
ancillary conditions (less behavioral problems)
• Welfare bounds and the set of optimal policies
are large when choices are more sensitive to
ancillary conditions (more behavioral choices)
– Impossible to rule out plan L if is revealed preferred
to H and M for some framing condition d’’

Preference refinement
• Drop ancillary condition d’’ because considered
as contaminated and so rule out plan L from the
set of optimal policy
• Example: reference-dependence bias and
optimal redistributive policy:
– the aversion to loss would favor the rich facing
income loss against the poor facing income gain.
– Bias for less redistribution

Structural modeling
• Preferences can be identified from the observed
choices on the basis of a structural decision-
making model
• By building a behavioral model that can explain
how ancillary conditions affect the choices, and
then to use this model to predict which choices
reveal true preferences.
• The behavioral model seeks to explain the
deviation from rationality and “true” preferences

• Example
– Use a present-bias (β, δ) model that explains
observed choices, and then calculate the optimal
policy within such a model using normative
discounting with β = 1.
• Problem:
– many competing behavioral models to explain the
same deviation from rationality.

Application: global warming
– Use a present-bias (β=1/2, δ=1) model :
– agent can make investment at a cost of C=100 (i.e.,
pollution abatement costs) to gain delayed benefits
B=180
– Investing tomorrow is desirable
-βδ C+βδ2 B=-(1/2)100+(1/2)180=+40
– Investing today is not desirable
- C+βδ B =-100+(1/2)180=-10

Self-control problem: willingness to invest but only
tomorrow
– Agent will never act according to plan which is to
invest tomorrow
– Revealed choices of not investing to curb global
warming cannot be used to infer the true (inter-
temporal) preference of the agent, and cannot be
relied upon to make normative judgments
– Optimal policy based on normative criterion β =1
rather than behavioral parameter β =1/2
3.6 Social preferences
http://www.preferencesfortruthtelling.com/

Ultimatum game
• Two players bargain (anonymously) to divide a
fixed amount between them.
• The proposer chooses to give a share s ∈ [0, 1]
of the “pie” to the responder.
• The responder decides either accepts or rejects
• If accepted both player gets their agreed upon
shares, p(s)=1-s and r(s)=s
• If rejected players receive nothing p=r=0.

• Nash equilibrium: the rational solution predicts
that proposer should offer the smallest possible
share s*≈0 and responder would accept it.
• Güth, Schmittberger, Schwarze (1983)
– The first experiment on this game.
– The mean offer was 37% of the “pie”
• Further experiments confirm that players deviate
from rational prediction for some notion of fairness.
– The average offers are in the region of 40-50% of the pie
– About half of the responders reject offers below 30%

Ultimatum game in my classroom :
“Pie”: 10 €
Offer
The average offer was 4.82 €
90% willing to offer >3 €
Response
The minimum average amount willing to accept was 3.54 €
32% willing to reject offers < 3 €

• What is the fair sex?
• The minimum amount players were willing to accept

was on average 3.5 € for boys and 3.61 € for girls
• The average offer was 4.99 € for boys and 4,60 €

for girls

• Competitive environment: the rational solution
becomes consistent with experimental outcome
• Consider n > 1 proposers making offers
s = (s1, . . . , si, . . . , sn)
to a unique responder.
• If offers si accepted, the responder earn si and the
proposer earns 1-si. The other proposers earn 0
• Nash equilibrium s*≈ 1 is confirmed by experiments
– In competitive environment, agents act as if they are selfish.

Selfish, Altruist, Fair, Envy (SAFE)
Consider utility function on income levels
vi(yi, yj ) where i ≠ j .
• Selfish: vi is independent of yj
• Altruist: vi increases in yj
• Envy: vi decreases in yj
• Fair: vi decreases in the difference yi- yj

Market impact
• Separable preferences, V(Ci, C-i)=u(Ci)+ w(C-i)
– my consumption choices Ci are independent of
others’ consumption choices C-i ,
• Competitive markets,
– an agent’s decision does not influence the market
price nor the volume of trade,
– then he has no opportunity to change the material
consumptions of others in the economy.
– He behaves as if selfish, even though he is non-
selfish.

Theorem 3.2 (Dufwenberg et al. RES2011)
If all agents have separable preferences that
are strictly monotone in own consumption, any
Walrasian equilibrium of an economy with social
preferences is a Walrasian equilibrium of the
standard economy with selfish preferences.
=> Social preferences do not change market outcome

• Market equilibrium may not be affected by social
preferences
• But the market’s outcome will in general not be
efficient (contrarily to the standard competitive
economy). Ci=- C-i can be Pareto improving
• Also market outcome can be affected by social
preferences under market failures (e.g. public
good=> non separable pref)
Conclusions
• Behavioral economics has made clear headway
into the public policy mainstream
• Most economists recognize that some of the
people are not fully rational some of the time,
and some of the time that matters
• What most economists are not (yet?) willing to
accept is that bounded rationality carries much
weight in a market setting (mindless economy);
• they believe rational actors cancel out the
irrational ones and that markets can only fool
some people some of the time but not all the
people all the time.
Conclusions
• “The economist may attempt to ignore
psychology, but it is sheer impossibility for him to
ignore human nature,” J.M Clark Journal of
Political Economy 1918)
• Public policy is about externalities
• Behavioral policy is about internalities (e.g self-
control problems)
• Public policy is about changing prices and
income
• Behavioral policy is more about design, framing
and commitment device.
Social dilemmas:
Externalities,
Coordination, Commitment
Social dilemmas
III Departures from Efficiency
Outline
• Social dilemmas motivate public policy
– Individual rationality = collective irrationality
=>standard public policy
– Individual irrationality= collective irrationality

=> behavioural pub policy
• Three social dilemmas
– Externality
– Coordination trap
– Commitment problem
Externality
I would make the correct choice if I could
internalize the external effect of my choice on
others (e.g. pollution)
• Public policy to internalize the externality (e.g.
Pollution tax)
• We will proceed with variety of examples
Coordination
I would make the correct choice if I knew your choice in
advance. (what side of the street to drive?)
• Complementarity in actions => Multiple equilibria
• One equilibrium Pareto dominates another
• Agents can become “trapped" in an inefficient
equilibrium
• How? | reinforcing expectations (bank run)
• Different from externality because multiple equilibria:
good equilibrium is also an equilibrium
Commitment
I would make the correct choice if I could trust you.
• I can't trust you to follow through on your promise
• I take a different action as a result of the lack of trust
that makes both of us worse off
• If just one of us could (credibly) commit, that would
be beneficial for both of us
• The externalities assume everyone is able to commit
EXTERNALITY
8.1 Externality
• An externality is a link between economic agents
that lies outside the price system
– Pollution from a factory
– Covid/flue infection
• Externalities are not under the control of the
affected (infected) agent
– Efficiency theorems do not apply
– Competitive equilibrium (laissez-faire) is not efficient
• Externalities are of practical importance
– Climate change
– Pandemic
8.2 Externality Defined

Definition 8.1 (Externality) An externality is present
whenever some economic agent’s welfare is “directly”
affected by the action of another agent in the economy
• Production externality: the externality affects profit

• Consumption externality: the externality affects
utility
• Positive externality: raises utility or profit
• Negative externality: reduces utility or profit
8.2 Externality Defined

Pecuniary externality: an external effect that works
through prices
– An oil price rise affects the profitability of a fishery
• Pecuniary externalities do not create an inefficiency
• With externalities the actions of agents are not
independent nor determined solely by prices
• Strategic interdependence arises
• This is the source of inefficiency
8.3 Market Inefficiency

• Consider two consumers with utility functions
( ) ( )
U 1 = x1 + u1 z1 + 1 z 2
z is the externality
U 2 = x2 + u (z ) +  (z )
2
2
2
1
• The externality arises from consumption of good z

• The competitive equilibrium satisfies
( )
uh ' z h = 1
• The efficient allocation is described by
( ) ( )
u1 ' z1 +  2 ' z1 = 1
u ' (z )+  ' (z ) = 1
2
2
1
2

• The externality leads to a gap between private value
( u h ') and social value (u h '+ h~ ' )
• If the externality is positive  h ' 0
– Marginal utility of each consumer is lower at efficient
allocation than at market equilibrium
– Too little good z is consumed in equilibrium
• If the externality is negative  h ' 0
– Marginal utility of each consumer is higher at efficient
allocation than at market equilibrium
– Too much good z is consumed in equilibrium

• In Fig. 8.1 the market
outcome has private Marginal
benefit
marginal benefit (PMB) and cost SMB (vh~ '  0)
equal to marginal cost (MC) PMB
• The optimum allocation has MC

social marginal benefit
SMB (vh~ '  0)
(SMB) equal to MC zh
• Location of SMB relative to

PMB depends on external Figure 7.1
effect Deviation of private from
social benefits
Examples
• Environment Pollution
• Traffic Jam
• Study Choice
• Rat Race
• Natural Ressources Overuse
8.4 Externality Examples

River Pollution
• Two firms located on a river
• The upstream firm, u, pollutes the river
• The pollution reduces the output of the downstream
firm, d
• Both produce the same output and sell at a price of 1
• Technologies are F u (Lu ), F d Ld , Lu where Li is labor ( )
use of firm i
• The profit level of firm i is  i = F i () − wLi

• Fig. 8.2 depicts the Revenue
allocation of labor Upstream
Cost
• Production function of  u
downstream shifts down

when Lu increases w
w u* d*
• Profit maximization leads 0 u 0 d
L ,L
to Lu*and Ld *  d
Downstream
• A small reallocation of labor Cost
from u to d does not affect  u Revenue
but raises  d
• The equilibrium is Figure 8.2
inefficient Equilibrium with river pollution

Traffic Jam
• This example considers the externalities that car
drivers impose on each other
• Assume there are N commuters with a choice of
commuting by train or car
• Travel by train takes 40 minutes
• The travel time by car increases as the number of car
users increases
• Commuters make the choice which minimizes their
personal travel time

• In Fig. 8.3 the equilibrium
choice of commuting mode Minutes
equalizes travel time by commuting
Maximum
train and car time saving Car
• 40% of commuters choose 40

travel by car Train
20
• The optimum choice
minimizes total travel time 0 20 40 % of car users
• This occurs when 20% travel

by car Figure 8.3
• The equilibrium has too Choice of commuting mode
many commuters choosing
to use cars

Pecuniary Externality
• A group of students must choose to be either
economists or lawyers
• Income declines when more students make the same
choice
• This is a pecuniary externality since an additional
student choosing to be an economist lowers income
for all economists
• Each individual ignores this externality when
choosing occupation

• Fig. 8.4 shows income in
each occupation Income of Income of
lawyers
• In equilibrium the income economists
levels are equal Lawyer
• A percentage E choose to
be economists Economist
• This equilibrium is efficient

0 E 100
• A reduction in income is a Percentage of economists
cost for an employee but a

benefit for an employer Figure 8.4
Job choice
• Pecuniary externalities do
not cause inefficiency

Rat Race Problem
• A rat race is a contest for relative position
• Assume performance is judged in relative terms and
not absolute terms
• An advantage is gained over rivals only by competing
harder than they do
• If all competitors compete hard the extra efforts
cancels out
• All competitors could gain by making an agreement
to reduce effort

• The Rat Race is shown in
Fig. 8.5
• Making high effort has a
cost of c Player 2
low high
• High effort is a dominant 1/2 1-c
low
strategy so the Nash Player 1
1/2 0
0 1/2 - c
equilibrium is {high, high} high
1-c 1/2 - c
• The Pareto-efficient
outcome is {low, low}
• Committing to low effort is Figure 8.5
a Pareto-improvement Rat Race

Tragedy of the Commons
• This example arises from the common right of access
to a resource
• Consider a lake that is used without restriction by
local fisherman
• The fishermen rent boats on a daily basis at cost c
• If B boats are hired each fisherman catches F(B) fish
• Each fisherman can earn wage w if they do not fish

• In Fig. 7.6 the equilibrium
number of boats solves Revenue
F(B*) – c = w
per boat
• The optimum solves

F(Bo) – c + Bo F’(Bo ) = w c+w
tax
Total cost
per boat
• F’(Bo ) < 0 implies Bo < B*

MR AR
• In equilibrium there are too B o
B* Number
many boats of boats
• Each fisherman ignores the

Figure 8.6
negative externality Tragedy of the Commons
• A tax can restore efficiency
Solutions
Public solutions
• Pigouvian taxes
• Licenses
Private solutions
• Integration
• Property rights and negotiation
8.5 Pigouvian Taxation

• Externalities cause inefficiency because of the
divergence between social and private benefits (or
costs)
• A tax can be used to raise the private marginal cost
– This assists efficiency with a negative externality
• A subsidy (a negative tax) can be used to reduce the
private marginal cost
– This assists efficiency with a positive externality
• Taxes that are used to combat externalities are
called Pigouvian taxes

• The use of a tax to correct
for a negative externality is Value PMC '
shown in Fig. 8.8 t PMC
• Social marginal benefit
(SMB) is below Private
marginal benefit (PMB) PMB
• The tax, t, shifts Private SMB
marginal cost from PMC to xo xm Quantity
PMC’
• The quantity consumed falls Figure 8.8
from xm to xo Pigouvian taxation
• xo is efficient with SMB =
PMC

• Pigouvian taxation seems a simple solution
– A tax is paid equal to the marginal damage
– A subsidy is received equal to marginal benefit
• There are limitations to the argument:
– The tax may need to be differentiated between
consumers, firms, and goods
– Without sufficient differentiation the externality is only
partially corrected
• Taxation should be seen as putting a price on the
externality (morally acceptable ? See cheat neutral
video)
8.6 Licenses
• Licenses can control externalities directly
– Supported by legislation that externalities can only be
generated if a license is held
• The optimal quantity of externality is calculated
• Licenses totalling to this quantity are issued
• Trading of licenses ensures they are used by those
who obtain the greatest benefit
• Administratively this is simpler than taxation
• But licenses have no informational advantage
8.6 Licenses
• Licenses and taxes are equivalent with certainty
– They can achieve the same efficient equilibrium
• This equivalence does not hold when there is
uncertainty
• Licenses achieve an outcome that is certain
• The outcome with taxation depends upon the
realized outcome of the (uncertain) abatement cost
• The precise circumstances determine which is best
– There is no general rule
8.6 Licenses
• In Fig. 8.9 PMC can be high
(PMCH) or low (PMCL)
Value PMCH
• Equating expected cost (PMCE) PMCE
to SMB gives abatement z* PMCL
• Licenses force abatement of z* t*
• A tax t* achieves either zH or zL SMB
• Choice between instruments zH z* zL Quantity

depends on slopes of PMC and
SMB
Figure 8.9
Uncertain costs
8.7 Integration
• Consider two producers who each causes a positive
externality for the other
– For example, a beekeeper and an orchard
• With no intervention each will ignore the externality
and produce too little
• If combined into a single firm they will internalize the
externality and produce at the efficient level
• But this may cause monopoly
• It may require unwilling partners to cooperate
8.8 The Coase Theorem

• The Coase Theorem proposes that economic agents
will solve externality problems without intervention
• Theorem 8.2 (Coase Theorem) In a competitive
economy with complete information and zero
transaction costs, the allocation of resources will be
efficient and invariant with respect to legal rules of
entitlement
• Legal rules of entitlement (or property rights)
determine ownership in the economy
• The theorem implies that policy should do no more
than establish and enforce property rights

• Coase sees externalities as arising through the
absence of property rights
– Pollution occurs when there is no right to clean air or clean
water
• If there was a property right anyone suffering an
externality would be paid compensation
• The compensation is the price paid by the party
causing the externality
• Competitive trading will ensure the correct price
emerges and efficiency is achieved

• The theorem also asserts that the equilibrium is
invariant to the assignment of property rights
• Will a firm pollute the atmosphere of a neighboring
house?
– Only if the benefit from doing so exceeds the
compensation required by the householder
– This applies whether the firm has the right to pollute or
the householder has the right to clean air
• The final distribution of income will be different
• Equilibrium will be unaffected by the allocation of
property rights if there are no income effects

• The Coase theorem has a number of practical
limitations
– The lack of clear property rights
– Transaction costs in reaching compensation agreements
– The potential thinness of the market implying bilateral
bargaining and potential inefficiency with incomplete
information
– Potential monopoly power
• The Coase theorem suggests a resolution to the
externality problem but there are reasons why the
market may not function
COORDINATION
Based on Ethan Bueno de Mesquita (2016) : Political Economy for Public Policy, Princeton University press
Coordination
I would make the correct choice if I knew your choice in
advance.
Complementarity in actions
=> Multiple equilibria (≠externality)
Agents can become “trapped" in bad equilibrium

Examples of coordination trap
• Social Conventions
– Foot binding
– Honor Killing
– Private vs public schools
• Economic
– Underdevelopment (1)
– Agglomeration
– Technology (2)
• Political
– Corruption (3)
To comply or not?
Benefit of compliance
Benefit of non compliance
Compliance rate (%) 100

0
Technology
• The bandwagon effect studies how standards are
adopted and wrong standards may be sustained
• The standard example is the typewriter keyboard
• The Qwerty keyboard was designed to prevent
jamming of mechanical keys
• It is claimed that faster keyboards can be designed
with alternative key locations
• Typists are reluctant to switch given their skills
Standard adoption
• A bandwagon effect is
illustrated in Fig. 8.7 1
• If less than p* typists use

Qwerty the proportion falls Percentage of
QWERTY
p*
to 0 over time users at

time t + 1
• Inefficiency of Qwerty
reflected in p* > 0.5
• There are equilibria at 0, p*, 0 p*
Percentage of QWERTY
1
and 1 users at time t
• The inefficient technology Figure 8.7

dominates if starting point Equilibrium keyboard choice
is to the right of p*
Under-development
• Pervasive complementarities may lead to situation
where the economy is locked in a “poverty trap”
while there exists a high equilibrium that requires all
agents to coordinate their action to reach it
• Underdevelopment= coordination trap
• Investments complementarities
• No investment because complementary investments
are not made
• Those complementary investments are not made
because the former are missing.
Policy interventions
• Reinforcing expectations can lead to bad
outcomes
• Policy interventions are quite different than
in externalities situations because the goal
is simply to move people to a different
equilibrium
– Short-run interventions (e.g. crack down)
– Communication and leadership
– Public investment (big push)
COMMITMENT
Based on Ethan Bueno de Mesquita (2016) : Political Economy for Public Policy, Princeton University press
Commitment problems
I would make the correct choice if could trust
you
• I take a different action by lack of trust
• That makes both of us worse off
• If you could commit, then I would take a
good action making both of us better off.
• Externality and coordination ≠ trust.
Costly conflict
• 2 parties are having a dispute
– War
– Law Suit (Brexit deal)
– Strikes (pension reform)
• Bargaining over an externality
• Costly conflict occurs if bargaining fails
• Because conflict is costly, a bargain exists
that makes both sides better off
Costly conflict
• Elites inside firm or organization blocking
technological change
• Social partners blocking (pension) reforms
• Labor/Management dispute
• Negotiating with terrorists
Bargaining surplus
Sharing Bargaining surplus

p1>p2
=
improve bargaining power
Preemptive conflict is avoided
Take away
The school of opportunity
JEAN HINDRIKS
MAT TÉO GODIN
Outline
Comparing schools : the school of opportunity:
Comparing school systems: the land of opportunity:

• High cross-country variation: beyond pure genetics and student sorting
• Equity-efficiency trade off
• The Great Gatsby curves
THE SCHOOL OF OPPORTUNITY: HINDRIKS & GODIN 2

Social mobility in education
• Relative versus absolute mobility
• Absolute mobility: rescaling
• Relative mobility: reshuffling
•Ex-post versus ex-ante equality opportunity
• Ex-ante approach Van de gaar (1993)
• Ex-post approach Roemer (1998)
• Ferreira &Gignoux (2011) =>ex-ante approach in education

Social mobility and social gradient
• Slope of social gradient y=h’(x)
• Linear relation (resilience= non linear)
•Intensity of social gradient
•Distribution dependant
Intensity= slope x (Sdx /Sdy)

More dispersed SES=> higher R2 (social heterogeneity)
More dispersed y => lower R2 (academic heterogeneity)

Mobility and social gradient
same social gradient (slope)
no mobility (left) – mobility (right)
12 12
y=x y=x
10 10
8 8
Student
Test rank 6 6
4 4
2 2
0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Student social rank Student social rank

Mobility and inequality
Perfect mobility (equal chance for all) Perfect inequality

Rank to Rank mobility
•For each student i within the same country
•test score in math => (national) test rank 𝑖 (in decile group)
•Social PISA index => (national) social rank 𝑖 (in decile group)
•Ranking by decile=> ranking less sensitive to sample selection
•Ordinal outcome => less sensitive to standardisation and measurement errors

THE SCHOOL OF OPPORTUNITY

The school of opportunity
Definition
School of opportunity =
+ School whose students perform beyond national-rank expectation (efficiency)
+ School with upward mobility of students (equity)
Methodology
 Measuring efficiency & equity
 Finding Correlates

Data
• PISA 2012 with focus on Mathematics
•Number of countries: OECD 32
•Number of schools: 4179
• Exclusion 10% most segregated schools (less relevant for social
mobility).
•Number of students: 113 000

SCHOOL EFFICIENCY

The rank-rank line for two countries
Different school
France
systems
Norway
Different efficiency
benchmark

School efficiency
The efficiency indicator
◦ School efficiency index is:
◦ is the estimated test rank of the student i in country c

◦ is the observed test rank of the student i in country c
◦ is the number of student in the school s
◦ In the US, the most efficient school (top efficiency decile) are 37% above the
expectation. Less efficient schools (bottom efficiency decile) are 42% below
expectation.

The efficiency of two schools in the US
distribution of their students around the national rank-rank line

Correlates of school efficiency
Regressing school efficiency on school-level variables
Exploiting both cross-national and intra-national variation in school efficiency and school
variables
Set of covariates from PISA 2012 at the school level
◦ Composition: average and variance ESCS
◦ Admittance policy :
◦ teaching methods
◦ Resources and Class size
◦ Autonomy &accountability
◦ Control and Monitoring
Using OLS Fixed effects model on efficiency index.
OLS FIXED EFFECTS MODEL ON EFFICIENCY COEF. Z
Mean SES 0.36*** (22.29)

Repeaters (%) -0.39*** (-16.03)
Urbanisation -0.01*** (-5.57)
ClassSkip -0.03*** (-10.26)
Selection BehavTransf -0.02*** (-5.63)
AcaTransf 0.03*** (6.71)
Admittance Rule-Perform 0.02*** (6.42)
Admittance Rule-Religion -0.01*** (-2.7)
Admittance Rule-Family -0.01*** (-3.22)
TeachingAdequacy (ability X pedagogy) 0.01*** (3.58)
Creative activity – intra muros 0.01*** (3.53)
Teaching
Teacher attitude-Socio Dev -0.01*** (-3.18)
methods
Teacher attitude-Academic Standard adjustement -0.01*** (-5.3)
Teach2Test 0.02*** (6.1)
Teacher competence- QualMath (%) 0.02*** (3.13)
Ressources Class size 0.01*** (5.07)
Class size^2 -0.002*** (-4.75)
MaterialShortage (instructional) -0.01*** (-3.44)
Accountability- student achievTracking 0.01* (1.83)
N 4179
R² 0,33

School Equity I
o The school Upward Mobility Rate (UMR) is the headcount ratio in school s (Chetti et al 2014)
1
UMRs 
ns
 is
ui
◦ with for (upward mobility) and

◦ In the US, the UMR varies between 76% in top decile schools and 21% in bottom decile
schools.

SCHOOL EQUITY

School Equity
Individual mobility in two schools

Equity versus efficiency
Equity line
Efficiency line
Student test rank
(decile)
Student Social rank

(decile)

Equity versus efficiency

School Equity II
The school equity index is :
◦ 𝑖 / x𝑖 is the individual mobility ratio : test rank /social rank (>1 upward mobility)
◦ 𝑠 is the number of student in school s
◦ r is a relative mobility parameter (with 0≤r ≤1):
◦ Higher r put more weight on mobility at the bottom
◦ Higher r put more weight on the size of the jump

Equity index II
r=1
Equity
Higher r
r<1
1
Individual mobility
1 yi/xi
Downward Upward
mobility mobility

School Equity
Schools ESCS rank and equity index
Note: The bottom and top ESCS school are ignored (most segregated schools). They represent 10 % of all schools.

Equity-efficiency Correlation
Efficiency UMR Equity
r=1/4 r=1/2 r=1

Efficiency 1
UMR 0.7028 1
r=1/4 0.5787 0.8238 1
Equity r=1/2 0.4187 0.7323 0.9327 1
r=1 0.1675 0.4853 0.7289 0.8011 1

Correlates of school equity
Regressing school equity on school-level variables
Exploiting cross-national and intra-national variation in school equity
Set of correlates from PISA 2012 at the school level
◦ Composition: ESCS mean and variance, selection rule
◦ teaching methods
◦ Resources and Class size
◦ Autonomy&Accountability
◦ Monitoring
Using OLS fixed effects model on UMR and Equity index.
OLS fixed effects model on UMR z
Coef
Mean SES -0.30*** (-13.63)
UMR model Repeater(%) -0.48*** (-14.18)
Urbanization -0.01*** (-5.14)
ClassSkip -0.03*** (-8.66)
Selection BehavTransf -0.03*** (-6.17)
AcademicTransf 0.03*** (6.03)
TeachingAdequacy (ability X pedagogy) 0.01*** (3.72)
School Newspaper 0.02*** (4.15)
Teaching Teacher attitude-Socio Dev -0.01** (-2.06)
methods Teacher attitude-Academic Standard adjustement -0.02*** (-4.49)
Teach2Test
0.02*** (4.79)
Teacher competence- QualMath (%) 0.02** (2.51)
Class size 0.01*** (4.4)
Ressources Class size^2 -0.002*** (-4.24)
Material Shortage (instructional) -0.01** (-2.21)
Accountability- student AchievTracking 0.01** (2.19)
R² (overall) 0,11
N 4179

OLS fixed effects model on Equity Coef. z
Equity Model Mean SES -1.49*** (-28.73)

Variance ESCS
(r=1) Urbanization
0.36*** (14.27)
-0.04*** (-6.37)
Repeater(%) -1.16*** (-14.89)
Selection ClassSkip -0.08*** (-9.43)
BehavTransf -0.05*** (-4.28)
AcademicTransf 0.07*** (5.7)
TeachingAdequacy (ability X pedagogy) 0.01** (2.58)
School Newspaper 0.05*** (4.09)
Teaching Teacher attitude-Socio Dev -0.02** (-2)
methods Teacher attitude-Academic Standard adjustement
-0.04*** (-4.41)
Teach2Test 0.05*** (5.33)
Teacher competence- QualMath (%) 0.05*** (2.9)
Ressources Class size 0.02*** (4.65)
Class size^2 -0.002*** (-4.24)
Autonomy- CourseContent
0.02** (2.02)
R² (overall) 0,24
N 4179

The school of opportunity (r=1)

Belgium: Pooling FR & NL schools
Efficient schools
2/3 schools in NL
1/3 schools in FR
Equitable schools
2/3 schools in NL
1/3 schools in FR
Equitable & efficient schools
47% schools in NL
11% schools in FR

Pooling Official & Private schools
Efficient schools
26% official schools
59% private schools
Equitable schools
53% private schools

38% private schools

Pooling Official & Private schools
Efficient schools
56% private schools
Equitable schools
54% private schools

40% private schools

Correlates of school of opportunity
Regressing school equity & efficiency on school
composition and policy
Exploiting both cross-national and intra-national variation
in school equity & efficiency and school-level variables
Same set of covariates from PISA 2012 at the school level
Using Logit model with equity based either on UMR or
Equity index, with FE
ALL SAMPLE FE MODEL 1 MODEL 2 MODEL 3
MEAN SES -0.26 -0.70 -0.89**
SD SES 2.09*** 2.14*** 2.11***
REPEATERS (%) -10.12*** -10.13*** -9.86***
URBANISATION -0.22*** -0.23*** -0.28***
Selection CLASSSKIP -0.46*** -0.44*** -0.44***
BEHAVTRANSF -0.42*** -0.42*** -0.42***
ACATRANSF 0.42*** 0.40*** 0.39***
ADMITTANCE RULE-PERFORM 0.27*** 0.26*** 0.24***
TEACHINGADEQUACY (ABILITY X PEDAGOGY) 0.10*** 0.09***
CREATIVE ACTIVITY – INTRA MUROS 0.29*** 0.27***
Teaching TEACHER ATTITUDE-SOCIO DEV -0.14** -0.13*
methods TEACHER ATTITUDE-ACADEMIC STANDARD -0.23*** -0.23***
TEACH2TEST 0.33*** 0.29***
TEACHER COMPETENCE- QUALMATH (%) 0.24*
CLASS SIZE 0.13***
Ressources CLASS SIZE^2 -0.002***
MATERIALSHORTAGE (INSTRUCTIONAL) -0.13***
ICTRATE 0.05
AUTONOMY- BUDGFORMATION -0.01
AUTONOMY- COURSECONTENT 0.10*
Governance AUTONOMY-ASSESSMENTPOLICY 0.08
ACCOUNTABILITY- STUDENT ACHIEVTRACKING 0.13
MONITORING- EXTERNALEVALUATION -0.01
N (NUMBER OF SCHOOLS) 2.866 2.866 2.866

Marginal Effect Significant variables
(model 3, FE)
dy/dx
Mean SES -0.121
SD SES 0.281***
Repeaters (%) -1.318***
Urbanisation -0.037***
ClassSkip -0.059***
BehavTransf -0.055***
AcaTransf 0.051***
Admittance Rule-Perform 0.034***
TeachingAdequacy (ability X pedagogy) 0.013**
Creative activity – intra muros 0.036**
Teacher attitude-Socio Dev -0.017
Teacher attitude-Academic Standard adjustement -0.031**
Teach2Test 0.039***
Teacher competence- QualMath (%) 0.032*
Class size 0.017***
Class size^2 -0.0002***
MaterialShortage (instructional) -0.017**
N (number of schools) 2.866

OFFICIAL SCHOOLS FE MODEL 1 MODEL 2 MODEL 3
MEAN SES 0.08 -0.26 -0.47
SD SES 2.17*** 2.24*** 2.23***
REPEATERS (%) -9.30*** -9.26*** -8.93***
URBANISATION -0.24*** -0.25*** -0.31***
CLASSSKIP -0.50*** -0.47*** -0.48***
BEHAVTRANSF -0.28*** -0.28*** -0.28***
ACATRANSF 0.42*** 0.40*** 0.39***
CREATIVE ACTIVITY – INTRA MUROS 0.25** 0.23**
TEACHER ATTITUDE-SOCIO DEV -0.06 -0.05
TEACHER ATTITUDE-ACADEMIC STANDARD -0.22*** -0.21***
TEACH2TEST 0.23*** 0.18**
TEACHER COMPETENCE- QUALMATH (%) 0.22
CLASS SIZE 0.15***
CLASS SIZE^2 -0.002***
MATERIALSHORTAGE (INSTRUCTIONAL) -0.14***
ICTRATE 0.01
AUTONOMY- COURSECONTENT 0.06
AUTONOMY-ASSESSMENTPOLICY 0.18**
MONITORING- EXTERNALEVALUATION 0.08
GENERAL TRACK FE MODEL MODEL 2 MODEL 3
MEAN SES -2.89*** -3.33*** -3.46***
SD SES 4.98*** 5.03*** 4.98***
REPEATERS (%) -9.25*** -9.23*** -9.00***
URBANISATION -0.16*** -0.17*** -0.20***
CLASSSKIP -0.44*** -0.43*** -0.43***
BEHAVTRANSF -0.29*** -0.29*** -0.28***
ACATRANSF 0.43*** 0.43*** 0.42***
CREATIVE ACTIVITY – INTRA MUROS 0.28*** 0.27**
TEACHER ATTITUDE-SOCIO DEV -0.09 -0.08
TEACHER ATTITUDE-ACADEMIC STANDARD -0.25*** -0.23***
TEACH2TEST 0.31*** 0.27***
TEACHER COMPETENCE- QUALMATH (%) 0.44***
CLASS SIZE 0.08**
CLASS SIZE^2 -0.001**
MATERIALSHORTAGE (INSTRUCTIONAL) -0.12**
ICTRATE 0.12
AUTONOMY- BUDGFORMATION 0.06
AUTONOMY- COURSECONTENT 0.10
AUTONOMY-ASSESSMENTPOLICY 0.06
MONITORING- EXTERNALEVALUATION 0.18

RANDOM SPLIT (50%) MODEL 1 MODEL 2 MODEL 3
MEAN SES 0.16 -0.27 -0.61
SD SES 1.65*** 1.66*** 1.62***
REPEATERS (%) -10.12*** -10.24*** -10.07***
URBANISATION -0.27*** -0.29*** -0.36***
CLASSSKIP -0.52*** -0.51*** -0.52***
BEHAVTRANSF -0.38*** -0.39*** -0.40***
ACATRANSF 0.37*** 0.35*** 0.34**
TEACHINGADEQUACY (ABILITY X PEDAGOGY) 0.11** 0.10*
CREATIVE ACTIVITY – INTRA MUROS 0.41*** 0.39***
TEACHER ATTITUDE-SOCIO DEV -0.20** -0.19*
TEACHER ATTITUDE-ACADEMIC STANDARD -0.21** -0.22***
TEACH2TEST 0.27** 0.21*
TEACHER COMPETENCE- QUALMATH (%) 0.38***
CLASS SIZE 0.17***
CLASS SIZE^2 -0.002***
MATERIALSHORTAGE (INSTRUCTIONAL) -0.11
ICTRATE -0.11
AUTONOMY- COURSECONTENT 0.25***
AUTONOMY-ASSESSMENTPOLICY 0.11
MONITORING- EXTERNALEVALUATION -0.11
Robustness
Standard regressors robustly correlated to equity/efficiency
Sample split model gives same regressors
Logit and Probit with FE give same regressors
 Reading model (using test scores in reading) give same regressors
 Except that two correlates become non-significant:
 Teaching Attitude sociodev
 Accountability-student achievtracking

Sorting and Policy Endogeneity issues
Substantial Variation in School efficiency and equity
Two effects: the school effect and the student effect (sorting effect)
Policy endogeneity issues: school policy is unlikely to be random but endogenously determined.
Answer: looking for correlates (not identifying causality)
 Our regressions control for the school composition
 Outliers are removed (most segregated school).
 Policy variables based on the perception of staff are mostly exogenous (see Hindriks et al,2015)
 Equity/efficiency are country-specific using only ordinal variables

LAND OF OPPORTUNITY

Data
• PISA 2003-2015 (5 waves)
• Focus on Mathematics
•Countries: 27/ 32 OECD countries (participants to 5 waves)
•Students at 15 year: 1.031.451
•Use student weights (sample errors)
•Use 5-10 plausible values (test errors)

Normalized inter-decile mobility
•For each student i within the same country
•test score in math => (national) test rank 𝑖 (in decile group)
•Social PISA index => (national) social rank 𝑖 (in decile group)
•Nomalized relative to perfect mobility :
1  yi 
Mobilityc 
nc
ic  x  / 1.61
  1
 i 
•Where perfect mobility =

Normalized Social Mobility
SVK
HUN
DNK
FRA
CFL
ISR
CZE
DEU
CFR
CHL
NZL
LUX
ESP
AUT
IRL
OCDE
AUS
POL
GBR
ITA
GRC
TUR
PRT
USA
JPN
SWE
NLD
KOR
SVN
CHE
ISL
NOR
CGE
FIN
EST
CAN
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

Social mobility vs Intensity Social
gradient

MOBILITY – EFFICIENCY TRADE OFF?

Social mobility & performance
The opportunity of students

to move up the edu ladder
versus
the level of edu attainment
that can be reached

MOBILITY – INEQUALITY
The Great Gatsby Curve?
« So we beat on, boats against the current, borne back ceaselessly into the past »
(F.S Fitzgerald: The Great Gatsby)

Great Gatsby curve I: student inequality
.8
FIN CAN
.7
ISL
Inter-decile mobility
NOR
JPN KOR
SWE CHE
ESP POL
IRL AUS
.6
NLD
USA PRT
AUT
GBR NZL
ITA
LUX SVK
DNK DEU
FRA BEL
.5
CZE
HUN
Correlation = - 0.58
.4
80 85 90 95 100
Standard deviation of test score

Great Gatsby curve II: school inequality
.8
FIN CAN
.7
ISL
NOR
KOR JPN
SWE AUS CHE
ESP POL
IRL
.6
NLD
USA PRT
AUT
NZL GBR
ITA
LUX SVK
DNK DEU
FRA
BEL
.5
CZE
HUN
.4
.2 .3 .4 .5 .6 .7
Between school inequality

Great Gatsby curve III: social segregation
.8
FIN CAN
.7
ISL
NOR
KOR
JPN
SWE CHE GRC AUS
POL ESP
.6
IRL
NLD
USAPRT
AUT
GBR
NZL ITA
SVK LUX
DNK DEU
BEL FRA
.5
CZE
HUN
.4
.25 .3 .35 .4 .45

Dissimilarity index

Great Gatsby curve III: social segregation
Source : PISA 2003-2015 and own calculations

Available on http://www.itinerainstitute.org/fr/

ANNEXES ON EXPLANATORY VARIABLES

Mean SES
.2 .4 .6 .8
AUS
AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
THE SCHOOL OF OPPORTUNITY: HINDRIKS & GODIN

LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
Variation in school social composition
55
Variance SES
0 .5 1 1.5 2
AUS
AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
THE SCHOOL OF OPPORTUNITY: HINDRIKS & GODIN

NLD
NOR
NZL
POL
Variation in school social mix
PRT
SVK
SVN
SWE
TUR
USA
56
Variation in school behavioral transfer
Student Transfer - Behavioural problems
AUS
AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
Not likely Likely

Very likely

Variation in school low-achiever transfer
Student Transfer - Low achievement
AUS
AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
Not likely Likely

Very likely

Variation in school academic admission
AUS
AUT
CAN
Admission - Academic record CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
Never Sometimes
Always

Variation in school teacher practice
AUS
Teacher intentions - Adapt standards

AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
-4 -3
-2 -1

Variation in school teaching practice
AUS
AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
TeachAdequacy
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
-3 -2
-1 0
1 2

Variation in school class size
AUS
AUT
CAN
CFL
CFR
CGE
CHE
CHL
CZE
DEU
DNK
Class size
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
13 18
23 28
33 38
43 48
53

Variation in school autonomy-budget
AUS
AUT
CAN
CFL
Autonomy: Budget Formation
CFR
CGE
CHE
CHL
CZE
DEU
DNK
ESP
EST
FIN
FRA
GBR
GRC
HUN
IRL
ISL
ISR
ITA
KOR
LUX
NLD
NOR
NZL
POL
PRT
SVK
SVN
SWE
TUR
USA
0 20 40 60 80 100
percent
0 1
2

Description of the variables
Composition
ESCS Mean of the ESCS at school level. (Mean of the student ESCS levels expressed in decile. The decile
(meanESCS) sub-categorization is constructed at the country level. Continuous variable.
Variance (sdESCS2 ) Variance of the ESCS (decile) at school level. Continuous variable.
Truancy Q : In your school, to what extent is the learning of students hindered by the following
(Truancy) phenomena? => Student truancy. 1=not at all, 2=very little, 3= to some extent and 4=a lot. Ordinal
variable
Skipping classes Q : In your school, to what extent is the learning of students hindered by the following
(ClassSkip) phenomena? => Students skipping classes. 1=not at all, 2=very little, 3= to some extent and 4=a
lot. Ordinal variable

Repeaters rate Repeaters rate at school level. Continuous variable.
(rdoub)
Location Location of the school:1=village, 2=Small Town, 3=Town, 4=City,5 =Large City
(location)

Selection
Transf. behaviour Q : In your school, how likely is it that a student in 10thgrade would be transferred to another
(BehavTransf) school for the following reasons? =>Behavioral problems. 1=Not likely, 2=likely and 3=very likely.
Ordinal variable
Transf. achievement Q : In your school, how likely is it that a student in 10thgrade would be transferred to another
(AcaTransf) school for the following reasons? =>Low achievement. 1=Not likely, 2=likely and 3=very likely.
Ordinal variable
Admission rule: Q:How often are the following factors considered when students are admitted to your school?
Student Performance Student’s record of academic performance (including placement tests) 1=Never, 2=Sometimes,
(AdmRulePerf) 3=Always.
Religious philosophy Parents’ endorsement of the instructional or religious philosophy of the school
(AdmRuleRel) 1=Never, 2=Sometimes, 3=Always.
Family member Preference given to family members of current or former students
(AdmRuleFamil) 1=Never, 2=Sometimes, 3=Always.

Pedagogy
Instructional Schools sometimes organize instruction differently for students with different abilities and
organization interests in mathematics. Which of the following options describe what your school does for
(TeachAdequacy) 10th-grade students in mathematics classes? Value from -3 to +2 :-3=similar content & different
level for all classes,+2=different content & different level for all classes. Ordinal variable
Creative extra- Index of creative extra-curricular activities at school. Total number of the following activities that
curricular activities occurred at school: i) band, orchestra or choir; ii) school play or school musical; and iii) art club or
(CREACTIV) art activities. Ordinal variable
Teacher intention : Q : How much do you agree with these statements about teachers in your school?=> There is
social development consensus among mathematics teachers that the social and emotional development of the
(TeachPrefSocdev) students is as important as their acquisition of mathematical skills and knowledge in
mathematics classes. -4= strongly disagree, -3=disagree, -2=agree, -1=strongly agree. Ordinal
variable
standard adaptation consensus among mathematics teachers that it is best to adapt academic standards to the
(TeachPrefAcaStand) students’ levels and needs. -4= strongly disagree, -3=disagree, -2=agree, -1=strongly agree.
Ordinal variable

achievement level consensus among mathematics teachers that academic achievement must be kept as high as
(TeachPref2Test) possible. -4= strongly disagree, -3=disagree, -2=agree, -1=strongly agree. Ordinal variable
Internet Use Q: In all subjects taken together, for how much of their work does the school expect 10th-grade
students to access the Internet? Work during class, Homework, Assignments

Ressource
Percentage of high Which percentage of the mathematics staff has a bachelor’s or master’s degree with a major in
qualified math mathematics, statistics, physics, or engineering
teachers
(PercQualMath)
Class size (CLSIZE) Average size of classes
Class size² Squared average size of classes

(CLSIZE2)
Material shortage Q:Is your school’s capacity to provide instruction hindered by any of the following issues?
(MatShortage) Shortage or inadequacy of instructional materials (e.g., textbooks) 1=not at all, 2=very little, 3= to
some extent and 4=a lot. Ordinal variable
Computers Approximative ratio computers/student in the tenth grade.
available
(PercComp)

Autonomy
Responsability: Q:Regarding your school, who has a considerable responsibility for the following tasks?
Budget formation Formulating the school budget. 0=State or national dept. of education, 1=School-level foverning
(RespBudgFormation) board, 2=Teachers or Principal.
Course content Determining course content. 0=State or national dept. of education, 1=School-level foverning
(RespCourseContent) board, 2=Teachers or Principal.
Responsability: Q:Regarding your school, who has a considerable responsibility for the following tasks? Budget
Budget allocation Allocation. 0=State or national dept. of education, 1=School-level foverning board, 2=Teachers
(RespBudgAlloc) or Principal.
Student assessment Establishing student assessment policies. 0=State or national dept. of education, 1=School-level
policies foverning board, 2=Teachers or Principal.
(RespAssessPol)
Selecting teachers for Selecting teachers for hire. 0=State or national dept. of education, 1=School-level foverning
hires board, 2=Teachers or Principal.
(RespHireTeach)
Accountability
Data: Achievement tracked Q:In your school, are achievement data used in any of the following accountability procedures? Achievement data are
by the authority tracked over time by an administrative authority, such as a district, state, or national education agency.-2=no, -1=yes
(AchievTracking)
Quality Assurance: Data Q:Which of the following measures aimed at quality assurance and improvement do you have in your school? Systematic
recording (DataRecording) recording of data including teacher and student attendance and graduation rates, test results and professional
development of teachers. -2=no, -1=yes
Quality Assurance: Internal Q:Which of the following measures aimed at quality assurance and improvement do you have in your school? Internal
evaluation (InternEval)
evaluation/self-evaluation . -2=no, -1=yes
Quality Assurance: Student Q:Which of the following measures aimed at quality assurance and improvement do you have in your school? Seeking
feedback
(StudentFeedback)
written feedback from students (e.g., regarding classes, teachers or resources) . -2=no, -1=yes
Data: Assessments to make Q:In your school, are assessments of students in the 10th grade used for any of the following purposes? To make decisions
decision about retention about students’ retention or promotion:-2=no,-1=yes
(Assess2Retention)
Data: Assessments to Q:In your school, are assessments of students in the 10th grade used for any of the following purposes? To monitor the
monitor school progress school’s progress from year + cxto year
(Assess2School) :-2=no,-1=yes

Accountability
Data: Assessments to Q:In your school, are assessments of students in the 10th grade used for any of the following purposes? To group students
instruction purposes for instructional purposes
(Assess2Instruction) :-2=no,-1=yes
Data: Assessments to Q:In your school, are assessments of students in the 10th grade used for any of the following purposes? To make judgments
monitor teachers about teachers’ effectiveness
(Assess2teach) :-2=no,-1=yes

Monitoring
Monitoring: During the last year, have any of the following methods been used to monitor the practice of
Teacher peer review
(MonitPeers)
mathematics teachers at your school? Teacher peer review (of lesson plans, assessment
instruments, lessons) No=-2, Yes=-1
Monitoring: During the last year, have any of the following methods been used to monitor the practice of mathematics teachers at your
Principal or Staff school? Principal or senior staff observations of lessons
observations No=-2, Yes=-1
(MonitStaff)
Monitoring: During the last year, have any of the following methods been used to monitor the practice of mathematics teachers at your
Inspector or external school? Observation of classes by inspectors or other persons external to the school
(MonitExternal) No=-2, Yes=-1
Teachers’ rewards: To what extent have appraisals of and/or feedback to teachers directly led to the following?
Change in work
responsabilities
Changes in work responsibilities that make the job more attractive
(RewardResponsab) . 1=No change, 2=small change, 3=moderate change, 4=large change
Teachers’ rewards: To what extent have appraisals of and/or feedback to teachers directly led to the following? A
Salary
(RewardSalary)
change in salary. 1=No change, 2=small change, 3=moderate change, 4=large change

Monitoring
Teachers’ rewards: To what extent have appraisals of and/or feedback to teachers directly led to the following? A financial bonus or another
Financial bonus kind of monetary reward. 1=No change, 2=small change, 3=moderate change, 4=large change
(RewardBonus)
Teachers’ rewards: To what extent have appraisals of and/or feedback to teachers directly led to the following? Public recognition from you .
Public recognition 1=No change, 2=small change, 3=moderate change, 4=large change
(RewardPubRecognition)

Annexes
USA
1.2
1
Variance
.8
.6
.4
.2 .4 .6 .8
meanESCS
Equity under the mean Equity above the mean

Chapter 6
Public Goods
Reading
• Essential reading
– Hindriks, J and G.D. Myles (2013) Intermediate Public Economics.
(Cambridge: MIT Press) Chapter 6.
• Further reading
– Andreoni, J. (1990) ‘Impure altruism and donations to public goods: a
theory of warm-glow giving’, Economic Journal, 100, 464 - 477.
– Abrams, B.A. and M.A Schmitz (1984) ‘The crowding out effect of
government transfers on private charitable contributions: cross
sectional evidence’, National Tax Journal, 37, 563 - 568.
– Bergstrom, T.C., L. Blume, L. and H. Varian (1986) ‘On the private
provision of public goods’, Journal of Public Economics, 29, 25 - 49.
– Bohm, P. (1972) ‘Estimating demand for public goods: an experiment’,
European Economic Review, 3, 55 - 66.
Reading
– Cornes, R.C. and T. Sandler (1996) The Theory of Externalities, Public
Goods and Club Goods. (Cambridge: Cambridge University Press)
Chapters 6 - 10.
– Cullis, J. and P. Jones (1998) Public Finance and Public Choice, (Oxford:
Oxford University Press) Chapter 3.
– Isaac, R.M., K.F. McCue and C.R Plott (1985) ‘Public goods in an
experimental environment’, Journal of Public Economics, 26, 51 - 74.
– Itaya, J.-I., D. de Meza and G.D. Myles (1997) ‘In praise of inequality:
public good provision and income distribution’, Economics Letters, 57,
289 - 296.
– Oakland, W.H. (1987) ‘Theory of public goods’ in A.J. Auerbach and M.
Feldstein (eds.), Handbook of Public Economics (Amsterdam: North-
Holland).
– Samuelson, P.A. (1954) ‘The pure theory of public expenditure’,
Review of Economics and Statistics, 36, 387 - 389.
Reading
– Warr, P.G. (1983) ‘The private provision of a pure public good is

independent of the distribution of income’, Economics Letters, 13, 207
- 211.
• Challenging reading
– Groves, T. and J. Ledyard (1977) ‘Optimal allocation of public goods: a solution
to the ‘free rider’ problem’, Econometrica , 45, 783 - 809.
– Foley, D.K. (1970) ‘Lindahl’s solution and the core of an economy with public
goods’, Econometrica, 38, 66 - 72.
– Laffont, J.-J. (1987) ‘Incentives and the allocation of public goods’, in A.J.
Auerbach and M. Feldstein (eds.), Handbook of Public Economics.
(Amsterdam: North-Holland).
– Milleron, J.-C. (1972) ‘Theory of value with public goods: a survey article’,
Journal of Economic Theory, 5, 419 - 477.
6.1 Introduction
• National defense: all inhabitants are simultaneously
protected
• Radio broadcast: received simultaneously by all
listeners in range of the transmitter
• These are both public goods
• If many consumers benefit from a single unit of
provision the efficiency theorems do not apply
• Public goods cause market failure so government
intervention may be beneficial
6.2 Definitions
• The pure public good has been subject to most
analysis
• A pure public good satisfies:
– Nonexcludability If the public good is supplied, no
consumer can be excluded from consuming it
– Nonrivalry Consumption of the public good by one
consumer does not reduce the quantity available for
consumption by any other
• A private good is excludable at no cost and is
perfectly rivalrous
6.2 Definitions
• Goods can possess different
combinations of rivalry and
excludability Rivalrous
Non-
rivalrous
• Club goods are studied in Private Club

chapter 7 Excludable
good good
Common
• Common property Non-
Excludable
property
resource
Public
good
resources are studied in

chapter 8
• These are both examples of Figure 6.1
impure public goods Typology of goods
6.3 Private Provision

• The characteristics of a public good lead to the
wrong incentives for consumers
• Each consumer has an incentive to rely on others to
provide the public good
• The reliance on others is called free-riding
• This leads to inefficiency since too little public good
is provided
• All consumers will benefit from providing more
public good

• Consider two consumers who allocate their incomes
between a private good and a public good
• The consumers take prices as fixed
• Each consumer derives a benefit from the public
good provision of the other
• This introduces strategic interaction into the decision
processes
• The Nash equilibrium in provision levels has to be
characterized

• This interaction is captured in the preferences of
consumer h with utility function
U h ( xh , g1 + g 2 )
• The choice must satisfy the budget constraint
M h = xh + g h
• Consider consumer 1, using the budget

constraint, his utility function is
U 1 (M 1 − g1, g1 + g 2 )

2
• Indifference curves obtain by tracing the rate at which g
1
can be traded for g keeping utility constant
dg 2 U 1x − U G1
1
= 1
dg U1
UG
1 2
• Let ĝ be the best response of consumer 1 to g
U U 1
x
1
G
g 1  ĝ 1
U =U 1
x
1
G
g 1 = ĝ 1
U U 1
x
1
G
g 1  ĝ 1

• Let g h be the provision of
consumer h g2
Budget
constraint
• Fig. 6.2 shows the
preferences of consumer 1
• Assume consumer 2 g2
provides g 2
• The utility of consumer 1 is
maximized at ĝ 1 ĝ 1 g1
2
• Varying g traces out the
locus of choices for Figure 6.2
consumer 1 Preferences and choice

• Fig. 6.3 constructs the locus
of choices for consumer 2 g2
Budget
constraint
• If consumer 1 chooses to
provide g 1consumer 2
chooses ĝ 2 ĝ 2
• The locus of choices is given

by the solid line through the
tangency points g1 g1
• This is the best-response

function of consumer 2 Figure 6.3
Best reaction for
consumer 2

• The Nash equilibrium is
where the choices of the g2
two consumers are the best
reactions to each other
• At this point neither has an ĝ 2
incentive to change their

choice
• This occurs where the best- ĝ 1 g1
response functions cross

• The equilibrium choices are ĝ 1 Figure 6.4
and ĝ 2 Nash equilibrium

• The private provision
equilibrium is inefficient g2 Locus of Pareto
efficient allocations
• But it is privately rational
• A simultaneous increase in
provision by both ĝ 2
Set of Pareto-
consumers leads to a Pareto improvements
improvement
g1
• Pareto-efficient allocations ĝ 1
are the points of tangency

between indifference Figure 6.5
curves Inefficiency of equilibrium
6.4 Efficient Provision

• At a Pareto-efficient allocation the indifference
curves are tangential
• This does not imply equality of the marginal rates of
substitution because the indifference curves are
defined over the quantities of the public good
purchased by the two consumers
• Instead the efficiency condition involves the sum of
marginal rates of substitution and is termed the
Samuelson rule

Horizontal summation in the private good market

Vertical summation in the public good market

• The tangency condition is
dg 2 dg 2
|
1 U 1const.
= |
1 U 2const.
dg dg
• Calculating the derivatives
U 1x − U G
1
U G2
=
1
UG U x2 − U G2
• The marginal rate of substitution is
h
UG
MRS Gh , x 
U xh

• The tangency condition then becomes
1
MRS G ,x + MRS G, x = 1
2
• This is the Samuelson rule

– The sum of marginal rates of substitution is equated to
the marginal rate of transformation between public and
private goods
– The marginal rate of substitution measures the marginal
benefit to a consumer of another unit of public good
– The marginal rate of transformation is the marginal cost
of another unit

• For two private goods the efficiency condition is
MRS i , j = MRS i , j
1 2
• Why the difference?

– An additional unit of a private good goes to either
consumer 1 or consumer 2
– Efficiency is achieved when both place the same marginal
value upon it
– An additional unit of public good benefits both
consumers
– The marginal benefits are therefore summed
6.5 Voting
• The level of public good provision is frequently
determined by voting
• Political parties promise different levels of provision
and compete for office
• The party with the majority wins the election and
chooses the level of public good
• Need to assess whether the process of voting attains
efficiency
6.5 Voting
• There is a population of H voters
• The cost of the public good is shared equally
• Consumer h has income Mh
• The utility function is
h  h G 
U x ,G = U  M − ,G
 H
(
)
• Each consumer votes for the value of G that
maximizes utility
6.5 Voting
Private
good
• Rank the consumers by

income so
M 1  M 2  ....  M H
• Fig. 6.6 shows that the G1 Gm GH
Public
good
preferred levels of public Utility
good satisfy
G1  G 2  ....  G H
• Assume an odd number of
voters Public
G1 Gm GH
• The median voter will be good
decisive and their choice Gm Figure 6.6

will win the vote Allocation through voting
6.5 Voting
• The choice of the median voter satisfies
m G 
max U  M m
− ,G
G  H 
• The necessary condition can be written as
1
MRS = m
H
• So voting achieves efficiency only if
MRS h
H
MRS m = 
h =1 H
6.5 Voting
• Can any predictions be made about voting and
efficiency?
– Observed income distributions have a long right tail
– If MRS falls with income the median MRS is greater than
the mean
– This implies voting results in the median choice, Gm,
exceeding the efficient level
• There is no guarantee that voting will achieve
efficiency
6.6 Personalized Prices

• With private goods
consumption is adjusted to
equate marginal valuation Private Public
with market price good good
• With public goods it is not Price Same Different
possible for consumers to
adjust consumption Quantity Different Same
• This suggests adjusting
prices to match the
valuations of the fixed Table 6.1
quantity Prices and quantities
• This is the basis of
personalized pricing
6.6 Personalized Pricing

• Personalized pricing can be achieved by making each
consumer pay for their specific share of the cost of
the public good
• The Lindahl mechanism asks each consumer to
announce their demand for the public good as a
function of their cost share
• The shares are adjusted until all consumers demand
the same quantity
• If the demands honestly reflect preferences the
equilibrium is efficient

• The tax (cost) shares for the
two consumers are t1 and t2 G1 G2
• The shares satisfy t1 + t2 =

1
G* G*
• The budget constraint of h
is xh + thGh = Mh Reaction
Reaction
• Gh is chosen to maximize of 2 of 1
utility t1 t2
• Equilibrium shares ensure

G1 = G2= G* Figure 6.7
• Efficiency is achieved Lindahl equilibrium

• The choice problem is
(
max G h U h M h − t h G h , G h )
• This has necessary condition
U Gh
=t h
U xh
• Summing over consumers
1
UG U G2
+ = t 1 +t 2 = 1
U 1x U x2
• The allocation satisfies the Samuelson rule

• Personalized pricing achieves efficiency but suffers
from significant drawbacks
• With many consumers there are practical difficulties
of implementation
• The Lindahl mechanism is not incentive compatible:
the consumers have no incentive to announce their
true demand functions
• The announcement of a false demand is always
optimal

• Assume consumer 1 is
honest G1 G2
• If consumer 2 were also

honest the equilibrium eL
would be eL
• The equilibrium can be eM
moved to eM if consumer 2
announces a false demand t1 t2
function
• Allocation eM maximizes the Figure 6.8
utility of consumer 2 given Gaining by false
the demand function of announcement
consumer 1
6.7 Mechanism Design

• If consumers are rational they will make false
announcements whenever this is advantageous
• The false announcements will distort the outcome
and result in the loss of efficiency
• Mechanism design is the search for allocation
mechanisms that cannot be manipulated
• A preference revelation mechanism ensures true
preferences are revealed
• Two examples of mechanisms that do not work are
now demonstrated

• Understatement
– The benefit of the public
good is vh = 1
– The cost of 1 is met by Announcement
of player 2
those reporting rh = 1 0
0
1
0
0
– Announcements either Announcement
of player 1
0
1
1
1
1 2
rh = 0 or rh = 1 0 1
2
– Provided if r1 + r2 ≥ 1
– Nash equilibrium is rh =
Figure 6.9
0, h = 1, 2
Announcements and payoffs
– The public good is not
provided but should be

• Overstatement
– Benefit of the public
good is v1 = 0, v2 = ¾
– Cost of 1 is shared Announcement
of player 2
equally 3
4 1
0 1
4
– Reports are r1 = 0 or 1, Announcement
0
0
1
−1
2
1
of player 1
r2 = ¾ or 1 1 −1
2
4
−1
2
4
– Provide if r1 + r2 ≥ 1
– Equilibrium r1 = 0, r2 = 1
Figure 6.10
– The public good is
Payoffs and overstatement
provided when it should
not be

• The Clarke-Groves mechanism ensures
– Consumers reveal their true valuation of the public good
– The public good is provided only when it should be
• Assume that the allocation of the cost of the public
good (if provided) among consumer is taken as given
• Consumers report their net benefits from provision
of the public good (benefit – cost)
• The public good is provided if the sum of net benefits
is positive

• If the public good is provided then side payments
are made
• These side payments reflect the fact that extracting
the true report is costly
• The side payments internalize the net benefit of the
public good to other players
• This allows efficiency to be achieved

• Net benefits are vh = -1 or
vh = 1 (the mechanism must
work for both)
• Reports are rh = -1 or rh = 1 Announcement of
player 2
• The public good is provided -1 +1
v 2 −1
if r1 + r2 ≥ 0 Announcement
of player 1
-1
0
0
v1 + 1
v +1
2
v2 +1
• If provided the payoffs are +1
v1 − 1 v1 + 1
v1 + r2 for player 1 and

v2 + r1 for player 2
• The rh in the payoffs are the Figure 6.11
Clarke-Groves mechanism
side payments
Mechanism Design
Announcement
of player 2
• When v1 = -1 a truthful -1 +1
report is weakly dominant Announcement

-1 0 0
of player 1
for player 1 +1 −2 0
v1 = −1
• When v1 = 1 player 1 is
indifferent between truth Announcement
of player 2
-1 +1
and false statement
-1 0 2
Announcement
• The mechanism ensures of player 1
+1 0 2
there is no incentive not to v1 = +1

be truthful Figure 6.12
Payoffs for Player 1

• The Clarke-Groves mechanism ensures truthful
reports and efficient provision of the public good
• The limitation of the mechanism is the cost of the
side payments need to be made
• If v1 = v2 =1 the total cost of the side payments is 2
• The side payments must be financed from outside
the mechanism

• The cost of the side payments can be reduced but
not eliminated
• One way to reduce cost is for each player to make a
side payment only if their choice changes the social
decision
• The Clarke tax requires a pivotal agent (one who
changes the social decision) to pay a tax equal to the
cost imposed on other agents by the changed
decision
• Clarke taxes ensure efficient provision and that true
values are revealed
6.8 More on Private Provision

• The analysis of private provision of a public good
demonstrated that the equilibrium will be inefficient
• This is not the only result that the model delivers
• The private provision model makes additional
predictions that can be contrasted to evidence
• These predictions have implications for government
policy

• Consider a transfer of D
from consumer 1 to
g2
consumer 2
• The transfer shifts the
ĝ 2 + D
indifference curves from
ĝ 2
the solid to the dashed
• The choices gˆ 1 − D, gˆ 2 + D 
delivers the same utilities ĝ 1 − D ĝ 1 g1
after the transfer as the


choices gˆ 1 , gˆ 2 did before  Figure 6.13
• The best response function Effect of a transfer
also shifts

• Consumer 1 reduces
contribution to the public g2
good by D
• Consumer 2 raises ĝ 2 + D
contribution by D ĝ 2
• Total public good provision

remains at
G = gˆ 1 + gˆ 2 ĝ 1 − D ĝ 1 g1
• Consumption of private
good does not change Figure 6.14
New equilibrium
• The equilibrium is invariant
to the transfer

• Assume H identical
consumers G H =3
H =2
• Let G be total provision of

all consumers but one
• Symmetry of equilibrium Pareto-
efficient
implies G
g=
Reaction
function
H −1 g
• As H increases the
equilibrium moves up the
reaction function Figure 6.15
Additional consumers
• The contribution of each
individual tends to zero

• The predictions of the model have been tested using
laboratory experiments
• In a typical experiment participants are divided into
groups of ten and are given a fixed income to spend
• The income can be divided between a public good
and a private good
• The private good has higher private benefit and the
public good a higher social benefit

• Fig. 6.16 shows the payoffs in a
typical experiment (Isaac, McCue,
and Plott, 1985)
Private good Public good
• It is individually rational to spend Private benefit 5 1
all income on the private good Social benefit 5 10
• It is socially optimal to spend all

income on the public good
• The Nash equilibrium of the one-
shot game has no investment in
the public good Figure 6.16
Public good experiment

• The experiments find that an average contribution to
the public good is 30 to 90 percent of income
• Most observations fall in the 40 to 50 percent range
• Among students the contribution level to the public
good is lowest for economists and falls with the
number of years of economics studied
• Repeating the game results in lower contributions in
later rounds

• A number of explanations have been proposed for
these results
• The players may use non-Nash conjectures (i.e. they
may expect reciprocity from other players)
– This response has the problem of being arbitrary
• Consumers may derive a warm glow from making a
contribution
– The warm glow raises the benefit from the public good
and so raises contribution
• Social interaction can be admitted
– This could be a social custom or a consideration of fairness
6.8 Experiment on voting



• Voting turn out rates
• Control group 30%

• Neighbors 38%
Source: Gerber, Alan S., Donald P. Green, and Christopher W. Larimer.(2008) “Social pressure and
vote turnout: Evidence from a large-scale field experiment.” American Political Science Review 102
6.9 Fund-Raising Campaigns

• The model of private provision used so far has a
single round of voluntary contributions
• If consumers observe the equilibrium is inefficient
they may wish to have a second round of
contribution
• The same logic can be applied repeatedly suggesting
efficiency may be approached after numerous
rounds of contribution
• This logic is now assessed using a fund-raising game

• In the fund-raising game a target level of funds must
be achieved before a public good can be provided
• The game has an infinite horizon
• The target for funds is C
• There are two identical players (X and Y) who derive
benefit B from the public good
• The good is socially desirable with B < C < 2B
• Both players have discount rate d for delaying
completion for one period
• The players alternate in making contributions

• In a contribution campaign the contributions are
paid at the time they are made
– This form of campaign is used if no credible commitment
to future payment can be made
• Assume it is now the turn of X to make a
contribution at time T
• The maximal contribution X will make to finish the
campaign is xT = [1 – d]B
• At T – 1 the maximum contribution Y will make
(knowing X will finish the campaign at T) is yT–1 =
d[1 – d2]B
– This reasoning can be continued back in time

• The path of contributions is
Total
shown in Fig. 6.17 contributions
• Summing the contributions C

xT
yT −1
gives xT −2
[1 – d]B + d[1 – d2]B

yT −3
xT −4
...
+ d3[1 – d2]B + …= B
• The total contributions T Time
never exceed the individual
valuation
• The contribution campaign Figure 6.17
Two-player contribution
will never finance the public campaign
good

• In the subscription campaign donation pledges are
made
• Contributions are only made when and if total
pledges are sufficient to finance the public good
• This alters the strategic structure and the amount
raised is equal to the total valuation of the
contributors
• Start at time T with X about to make a pledge
• The maximum X will pledge to finish the campaign is
xT = [1 – d]B

• The maximum that Y will pledge at T – 1 knowing
that X will complete the campaign is yT–1 = [1 – d2]B
• Summing these pledges gives
[1 – d]B + [1 – d2]B + d [1 – d2]B + …= 2B
• It is therefore possible to fund the project because C
< 2B
• This result shows that allowing contributions to be
repeated may remove the inefficiency but requires
the ability to make binding commitments
6.10 Conclusions
• Public goods have different characteristics to private
goods
• The condition for efficiency reflects these
characteristics
• Private provision will not secure efficiency
• The government is restricted by limited information
on the value of the public good to each individual
• Mechanisms can be designed that deliver efficiency
but these are costly to operate

EcoPub PDF Complet

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EcoPub PDF Complet

Uploaded by

Copyright:

Available Formats

Chapter 1

1.1 Public Economics

1.2 Mixed economy

1.3 Analyzing Policy

TAX Price change

Source: Hindriks & Serse (JPubE2019)

Source: Hindriks & Serse (IJIO 2022).

1.3 Analyzing Policy

1.3 Analyzing Policy

1.3 Pareto efficiency

1.3 Pareto efficiency

1.3 Pareto efficiency

1.3 Pareto efficiency

1.3 Pareto efficiency

1.3 Pareto efficiency

1.4 Efficiency and distribution

1.5 An example of policy

1.5 An example of policy

1.5 An example of policy

1.5 An example of policy

1.5 An example of policy

1.5 An example of policy

1.6 Teacher pay

1.6 Teacher pay

1.6 Teacher pay

1.7 Free Trade or not?

1.7 Free Trade or not?

1.8 efficiency and utilitarism

1.8 efficiency and utilitarism

1.9 Policy in practice

1.9 Transfers Constraints

1.9 Utility is not money

1.9 Equity beyond efficiency

1.9 values beyond utility

1.9 Good policy?

1.10 Evidence based policy

Ashton and Wray 2013

2.2 Economic Models

2.3 Competitive Economies

2.3 Competitive Economies

2.4 The Exchange Economy

2.4 The Exchange Economy

2.4 The Exchange Economy

2.4 The Exchange Economy

the Edgeworth box in Fig.

2.4 The Exchange Economy

common budget constraint

• The gradient of the budget

2.4 The Exchange Economy

2.4 The Exchange Economy

x1i ( p1 , p2 ) + xi2 ( p1 , p2 ) = i1 + i2 , i = 1,2

2.4 The Exchange Economy

• The sum of demands from

2.4 The Exchange Economy

2.4 The Exchange Economy

2.4 The Exchange Economy

2.4 The Exchange Economy

2.4 The Exchange Economy

positive when p2 is low

2.5 Production and Exchange

2.5 Production and Exchange

2.5 Production and Exchange

subtraction from economy’s

2.5 Production and Exchange