Public Economics

General information
Each exercise set contains selected Discussion Questions from the textbook and Ad-
ditional Exercises.
Unless speci…ed otherwise, the discussion in the tutorials only covers the Additional
Exercises. You can do the questions from the textbook by yourself; solutions are
provided. If you have questions, you can ask your tutorial teacher.

Practicum 1

Discussion Questions Rosen:

8th edition 9th & 10th editions

Chapter 3 4, 9, 10 6, 12, 14
Chapter 4 1 1

Additional Exercises:

1.1 Two independent villages, Tinytown and Smallville, are located on opposite
sides of a river. Currently, there is no bridge connecting the two villages. The
villages decide independently on whether to (help) …nance a bridge. Building a
bridge would cost 100. If both villages help to …nance the bridge, they both pay
50; if only one village …nances the construction, it has to pay 100. The bene…t
of a bridge to Tinytown is 60, and to Smallville 75.

(a) Is it socially e¢ cient to build the bridge? Suppose the two villages decide
independently whether to build the bridge. Will the bridge be constructed
(determine the Nash equilibrium)?

Now suppose that the bene…t of the bridge to Smallville rises to 125.

(b) Suppose the two villages decide independently whether to build the bridge.
Will the bridge be constructed? Which village is best o¤?

1.2 Thelma and Louise are neighbours. During the winter, a snowplow can clear the
street in front of their houses. Thelma’s marginal utility (= marginal bene…t)
from snowplowing services is 12 Z euro, where Z is the number of times the
street is plowed. Louise’s marginal utility is 8 2Z euro. The market price per
hour equals 16 euro, which is also equal to the marginal cost of snowplowing.

1
 (a) Sketch the marginal utility curves and the aggregate willingness to pay.
Find the e¢ cient level of snowplowing.
(b) Suppose Thelma and Louise decide independently how many hours of snow-
plowing to pay for (of course taking the presence of the other into account).
What is the Nash-equilibrium outcome?
(c) How large is the di¤erence in welfare between the e¢ cient outcome and the
outcome under private provision (the outcome under b)?

Now suppose that the price of snowplowing drops to 8 euro.

(d) What would happen if the marginal cost of plowing suddenly drops below
8?

1.3 Adam and Britt derive utility from a public good. Adam’s total utility derived
from consuming Q units of the public good, measured in euro, is given by

UA = ln Q

Britt’s total utility derived from consuming Q units of the public good, measured
in euro, is given by
UB = ln Q
Total production cost of the public good are given by T = cQ. The market
price per unit of the public good is c. Both Adam and Britt maximise their
utility, and they decide independently from each other how many units of the
public good to buy. Assume that > > 0 and c > 0.

(a) Derive the socially e¢ cient level of public good provision.

(b) In the Nash-equilibrium, how many units of the public good does Britt
buy? How many units does she consume?

Suppose the government provides the public good, and covers the cost by taxing
Adam and Britt such that they both pay half of the total cost of providing the
public good. However, the government does not know the preferences of Adam
and Britt. The government makes the following proposal: Adam and Britt
(independently) state how many units of the public good they prefer. The
government then provides the average number of units (and adjusts the tax
accordingly). In other words, if Adams states to prefer QA units and Britt
states to prefer QB units, the government provides 21 (QA + QB ) units and both
pay a tax of 21 c 21 (QA + QB )

2
 (c) Determine the Nash equilibrium of statements QA and QB .
(d) Compare the outcome of c) to both the socially e¢ cient level (part a) and
the outcome in case of private provision (part b). What does this exercise
illustrate?

1.4 Consider a (very) small village, where only two families live: family Adams and
family Besley. The size of the available land around the village is 5 acres. This
land must be divided into space for agriculture, which creates sandwiches s, and
a park of size z acres. The park is a pure public good. Each acre agricultural
land yields 50 sandwiches.

(a) Find the production possibilities frontier. What is the Marginal Rate of
Transformation between sandwiches and park size?

The two families derive utility from the number of sandwiches and from the size
of the park. The utility of Adams and Besley is given by, respectively:

UA = sA + 20z
1 p
UB = sB + 30 z
2
(b) Derive the Marginal Rate of Substitution between sandwiches and park
size for both families. (Hint: MRS is the slope of an indi¤erence curve,
which is equal to the ratio of the marginal utilities of z and s.)
(c) What is the socially e¢ cient size of the park?
(d) How large would the park be if the two families decide independently about
their contribution to park size (assume that they both own half of the land
around the village)?

1.5 Consider a country with N individuals. Suppose the utility of individual i can
be expressed by p
Ui = y i + i G i
where Gi is the level of consumption of good G by individual i, yi is the level of
consumption of all other (private) goods by individual i, and i is a parameter
re‡ecting individual i’s relative preference for good G. The price per unit of
good y equals 1, the price per unit of good G equals p. Both prices are equal
to marginal production cost, which is assumed to be constant in production.1
1
This sentence implies that the supply curves are perfectly elastic (horizontal).

3
 All individuals have the same income, given by Y . Individuals di¤er in their
preferences: half of the individuals have i = 1, and the other half has i =
H > 1.
Suppose G is a private good.

(a) Determine individual i’s demand for good G, as a function of p.

(b) Determine total demand for good G, as a function of p.

Now suppose G is a public good.

(c) Determine the socially e¢ cient level of provision of G. How does it depend on
N , on p, and on H ? Explain.

Suppose all individuals have i = H . Furthermore, suppose each individual

can provide maximally 1 unit of the public good G.

(d) If individuals decide independently whether or not to provide a unit of the public
good, what is the Nash-equilibrium level of G?

4
Answers to exercises week 1

The solutions on the Discussion Questions (DQ) in Rosen & Gayer are numbered as
follows: Rosen [number DQ in 8th edition] / [number DQ in 9th & 10th editions].
So, Rosen 3.4 / 3.6 gives the solution to DQ 3.4 in the 8th edition and to DQ 3.6 in
the 9th & 10th editions of the textbook.

Rosen 3.4 / 3.6:

a) The social indi¤erence curves are straight lines with slope of –1. As far as
society is concerned, the “util”to Augustus is equivalent to the “util”to Livia.
b) Social indi¤erence curves are straight lines with slope of –2. This re‡ects the
fact that society values a “util”to Augustus twice as much as a “util”to Livia.
c) See the …gure.

5
 Figure 3.4

Rosen 3.9 / 3.12

Although Victoria’s marginal rate of substitution is equal to Albert’s, these are not
equal to the marginal rate of transformation and the allocation is, therefore, Pareto
ine¢ cient. Both people would give up 2 cups of tea for 1 crumpet but, according to
the production function, could actually get 6 crumpets by giving up 2 cups of tea. By
giving up tea and getting crumpets through the production function, both utilities
are raised.

Rosen 3.10 / 3.14

a) False. As shown on page 41 of the text, equality of the marginal rates of
substitution is a necessary, but not su¢ cient, condition. The MRS for each individual
must also equal the Marginal Rate of Transformation.
b) True. As long as the allocation is an interior solution in the Edgeworth box,
the marginal rates of substitution must be equal across individuals. This need not be
true, however, at the corners where one consumer has all the goods in the economy.
c) False. If a policy is a Pareto-improvement, it also increases social welfare, but
the other way around is not always true. Social welfare may increase by, for instance,
redistributing money from rich to poor. Clearly, this is not a pareto-improvement.
d, 8th edition) False. Smokers, especially those without children, are worse o¤ if
the tax on cigarettes is raised in order to fund education. Social welfare may improve
though, through better education and reduced externalities from smoking.
d, 10th edition) False. See Figure 3.10: a move from point q to point p3 increases
e¢ ciency, but reduces the utility of Eve. Hence, it is not a Pareto-improvement. Note
that given an allocation within the UPC, there exists an allocation on the UPC that
yields a Pareto-improvement.

Rosen 4.1 / 4.1:

a) Wilderness area is an impure public good – at some point, consumption
becomes rival; it is, however, nonexcludable.
b) These television signals are nonrival in consumption. Producers may be able
to choose to make the signal either excludable or non-excludable.
c) Medical school education is a private good.
d) Television programs are potentially excludable (Pay-tv etc.) and non-rival
(given that you have su¢ cient tv-sets).
e, 8th edition) An Internet site is nonrival in consumption (unless there is a lot
of tra¢ c), and is in principle excludable.
e, 9th & 10th edition) An ATM machine is excludable. It is non-rival (everyone
can use it), unless it is very busy (in which case you may have to wait).

6
Exercise 1.1
a) It is e¢ cient to build the bridge, as total bene…t is 135 > 100 (= total cost).
Both villages decide not to …nance the construction however. Consider this payo¤
matrix, where the …rst entry in a cell is the payo¤ to Tinytown, and the second entry
the payo¤ to Smallville:

S
build not build
T build 10; 25 40; 75
not build 60; 25 0; 0

For both villages, regardless of what the other village decides, it is always best to
choose not build. (Recall from Micro: this is a prisoners’dilemma game, ‘not build’
is here a dominant strategy and the equilibrium (not build, not build) is called a
Nash-equilibrium). Note that both villages would be better o¤ if some third party,
for instance the government, forced them to contribute both towards the construction
of the bridge.
b) Now, Smallville will build the bridge alone. Tinytown is best o¤. The payo¤
matrix changes to:
S
build not build
T build 10; 75 40; 125
not build 60; 25 0; 0
For Tinytown, not build is still the best option, regardless of Smallville’s choice (dom-
inant strategy). It follows that Smallville optimally chooses to build.

Exercise 1.2
a) Thelma’s marginal bene…t is M BT HELM A = 12 Z, and Louise’s is M BLOU ISE =
8 2Z. The marginal bene…t for society as a whole is the sum of the two marginal
bene…ts, or M B = 20 3Z (for Z 4), and is equal to Thelma’s marginal bene…t
schedule afterwards (for Z > 4). The marginal cost is constant at M C = 16. Setting
M B = M C along the …rst segment gives 20 3Z = 16, or Z = 34 , which is the
e¢ cient level of snowplowing.

7
b) Both Thelma and Louise would not pay for snowplowing: Z = 0. Marginal
cost is higher than marginal bene…t at Z = 0 for both.
c) With constant marginal cost, producer surplus is 0. Hence, we have to com-
pute the di¤erence in (total) Consumer Surplus. Under b), CS is 0. Under a),
total CS equals the area between the aggregate demand curve and the price. Hence,
CSa = 12 (20 16) 34 = 83 . Alternative way of calculating is to add the bene…ts
to Thelma and Louise of Z = 43 (area below MB-curves) and subtract the cost:
R4 R4 3 4 2
0
3
(M BT + M BL M C) = 03 (4 3Z) = 4 43 2 3
= 83 .
d) For M C 8, it is possible for Louise to credibly free ride, while Thelma
provides the e¢ cient allocation. This occurs because if Thelma believes that Louise
will free ride, Thelma chooses the amount of Z that maximises her own utility. With
M C 8, this gives Z 4. Yet, for Z 4, Louise’s marginal bene…t is zero, implying
that the social MB equals Thelma’s MB, see the …gure. Hence, if M C 8, Louise
free-rides and Thelma pays for Z 4, which is socially e¢ cient.
Beyond the question: Note that it depends on the speci…cs of the exercise that de-
spite free-riding, the outcome is socially e¢ cient. For instance, suppose Louise would

8
have marginal utility of snowplowing equal to 8 Z (rather than 8 2Z), and that
marginal cost would be equal to 6. Optimal provision would have Z = 7 (As for
Z 8, social marginal bene…t would be 20 2Z). However, if Thelma and Louise
decide independently, the only Nash equilibrium is where Thelma pays for Z = 6 and
Louise contributes nothing. Similar reasoning as above shows that this is indeed a
Nash equilibrium: Given that Thelma pays for Z = 6, marginal cost are higher than
marginal bene…t for Louise, so that she optimally contributes nothing. And given
that Louise does not contribute, it is optimal for Thelma to pay for Z = 6. Hence,
the outcome is not socially e¢ cient!
Note that the case where Louise pays for Z = 2 and Thelma does not contribute
is not a Nash equilibrium: If Thelma would not contribute, Louise optimally pays
for Z = 2. However, if Louise pays for Z = 2, marginal bene…ts to Thelma are
larger than marginal cost (10 > 6) so Thelma would contribute something. But then
Louise would want to reduce her contributions, etc. Hence, the only Nash equilibrium
is where Thelma pays for Z = 6 and Louise contributes nothing.

Exercise 1.3
a) M UA + M UB = M C gives Q + Q = c ! Q = +c . You can also maximise
UA + UB cQ with respect to Q.
b). The only Nash equilibrium is where Adam buys Q = c and Britt Q = 0. Hence,
both Adam and Britt consume Q = c (note: at Q = c , we have M UB = =c = c,
which is smaller than c by > . Hence, if Adam buys c units, Britt prefers to buy
0 units). start with person with highest marginal utility
c) The only Nash equilibrium is where QA = 4 c and QB = 0. As a result, the outcome
is Q = 2 c You can show this by (i) maximising utility of A and B and infering that
> implies you end up in a corner solution, or by (ii) intuitively arguing that for
any Q, the marginal cost are the same for A and B, but the marginal bene…ts are
always higher for A (as > ). Hence, if A is content with a particular Q, B prefers
a lower level and, hence, sets a lower QB . In turn, A sets a higher level.
d) As > , the outcome under c) is higher than the socially e¢ cient outcome. This
illustrates that where private provision of public goods tends to yield underprovision
(market failure), public provision tends to yield overprovision (government failure).
Which of these two is more e¢ cient depends on and .
Note 1: Observe that the di¤erence in stated preferences QA and QB is much larger
than the ‘true’di¤erence in preferred levels of Q.
Note 2: This exercise assumes people have no problem with lying about their prefer-
ences. Of course, if all people would be honest about their preferences (even if this
a¤ects their price or tax levels), outcomes would be more e¢ cient.

for c: MU(A) = MC(A) -> both partial derivatives wrt Qa -> Qa = 4a/c

for d: compare all utility - costs for plugging9in all values found for Q to find the ranking of situations
 1 acre gives park of size z = 1 or 50 sandwiches

Exercise 1.4
a) Each acre land can yield 50 s or 1 z. There are 5 acres land, so the maximum
amount of sandwiches is 250, and each acre park reduces s by 50. Hence, the PPF
equals s = 50(5 z). It follows that the MRT equals 50. (In general, the MRT is the
absolute value of the slope of the PPF, which can be written as M Cz
M Cs
)
@U=@z
b) MRS between s and z is given by . For family Adams, this gives
@U=@s
15
p
20 z 30
M RSA = 1 = 20; for family Besley, this gives M RSB = 1 = p .
2
z
c) Condition for optimal provision of public goods: M RSA + M RSB = M RT .
Substituting gives
30
20 + p = 50
z
Solving for z gives z = 1.
A note: in principle, both MRT and MRS could be de…ned in a reversed manner,
@U=@s @U=@z
so that MRS = rather than . However, this does not work in case of
@U=@z @U=@s p
1
deriving the e¢ cient level of public goods provision ( 20 + 30z 6= 50
1
for z = 1). This is
because we have to add up the marginal utilities of public good consumption across
all individuals, i.e. we want to know how many sandwiches the individuals together
@U=@z M Uz
want to give up for one additional unit of z. With MRS = = , the MRS
@U=@s M Us
is the marginal utility of public good z in terms of utility from consuming private
@U=@s
good s, which is what we want to know. With MRS = , we get how much of
@U=@z
the public good an individual is willing to give up for an additional private good s.
Adding this up over individuals is nonsense, because the additional private good can
only be consumed by one individual.
Note that if all goods are private goods, it does not matter whether MRS is written
@U=@x @U=@y
as or as , as long as the MRT is de…ned in the same way. It follows
@U=@y @U=@x
1
directly from the equilibirum condition: if M RSX = M RSY = M RT , then M RS X
=
1 1
M RSY
= M RT .
d) Regardless of the decision by family Besley, family Adams’marginal value of an
acre Park equals 20 sandwiches, while the marginal cost are equal to 50. Hence, family
Adams does not contribute to the park. Knowing this, family Besley maximises their
own family’s utility with respect to farm size. Hence, they optimally set M RSB =
30
M RT , which gives p = 50. Solving for z gives z = 0:36.
z
Clearly, this is less than optimal (as found under c)). The reason is that neither

10
family takes the bene…ts of the park to the other family into account.

Exercise1.5)
a) Using budget constraint Y = yi +pGi to substitute for yi = Y pGi into individual
i’s utility function and maximising with respect to Gi gives …rst-order condition
2
i i
p + p = 0 ! Gi =
2 Gi 2p
M UG pG pG
Note: it is also possible to use M RS = M RT , yielding M Uy
= py
(note that py
=
@Ui pi
pG
= p1 , and M RSi = @Gi 2 Gi
M RT ). It is given that py @Ui = 1
. This yields the same
@yi
outcome as above.
b) We know individual demand. To get to total demand, we must add all individuals’
demand. Demand depends on i . Given how i is distributed, we get that total
demand is given by
!
2 2
1 1 1 H 1
D=N + = 2 N 1 + ( H )2
2 2p 2 2p 8p

c) Note: if G is a public good, we have that Gi = G for all i (all individual consume
the same level of G). To …nd the socially e¢ cient level of G, we must equate the sum
of all individuals’ marginal rate of substitution to marginal rate of transformation.
@Ui pi
@Gi 2 Gi
Note that M RSi = pi
2 Gi
and M RT = p. Individual i’s M RSi = @Ui = 1
= pi .
2 Gi
@yi
Hence, society’s marginal bene…t equals

1 1 1
N p + pH
2 2 G 2 2 G
which can be rewritten as
N (1 + H )
p
4 G
2 2
Hence, socially e¢ cient level of G is given by N (1+ )
p H = p ! G =
4 G
N (1+ H )
16p2
.
This increases in N : given that people have positive marginal utility, an increase in
the number of people consuming G increases societies marginal bene…t of G.
This increases in H : if marginal bene…t is higher, it is e¢ cient to provide more of
the public good.
This decreases in p: if marginal cost is higher, it is e¢ cient to provide less of the
public good.

11
d) An individual provide a unit whenever the private marginal bene…t is at least equal
to marginal cost p. This gives
2
pH = p ! G = H2
2 G 4p

Note that this is considerably smaller than the socially e¢ cient level (which is G =
N 2 ( H )2
4p2
if all individuals have i = H ).

12
 Public Economics
Practicum 2

Discussion Questions:

8th edition 9th edition 10th edition

Chapter 5 2, 3a, 3d, 8, 11 4, 6a, 6d, 11, 13 4, 7a, 7d, 12, 14

Additional Exercises:

2.1 Anne and Beatrice are neighbours. Beatrice loves pets, and considers adopting
either a cat or a dog. Anne is very scared of dogs, and would prefer to stay
inside her house permanently if her neighbour would adopt a dog. Consider the
following payo¤s for Anne and Beatrice.

Beatrice
cat dog
Anne stay inside 150; 200 150; 350
go outside 450; 200 0; 350

(a) If Anne and Beatrice decide independently, what is the Nash equilibrium? Is this
an e¢ cient outcome?

(b) Suppose Anne and Beatrice can make binding agreements, what will they do?

Now suppose that Anne’s payo¤ of staying inside after Beatrice adopts a dog is
350.

(c) If Anne and Beatrice decide independently, what is the Nash equilibrium? Sup-
pose Anne and Beatrice can make binding agreements, what will they do? Ex-
plain why this di¤ers from the di¤erence between (a) and (b).

2.2 A …rm produces Q liters of paint at costs 10 + Q2 . The …rm sells its paint
at a …xed price of 10 per liter. Emissions, like Volatile Organic Components,
are given by the function W = 41 Q. Environmental damage D due to these
emissions is estimated at D = 4W 2 .

(a) Show in one …gure the marginal private bene…t, the marginal private cost,
and marginal external damage curves.

1
 (b) Compare the privately optimal outcome and the socially e¢ cient outcome
in terms of output, pro…ts, and damage.
(c) What is the level of the Pigouvian tax?
(d) Now suppose that the …rm reduces its costs to 10+cQ2 , where c < 1. Show
that the Pigouvian tax increases when c decreases. Explain why.

2.3 A …rm located on a riverbank produces Q tons of chemicals, which it can sell
at a price of P per ton. Production costs equal 21 cQ2 . Production pollutes the
river. The level of pollution in the river is given by W = 21 Q. Downstream, a
farmer uses the water from the river to irrigate his land. To prevent damage
to his crops, he must have clean water. Removing pollution W from the river
costs D = 2dW 2 . Assume that P > 0, c > 0, and d 0.

(a) Show in one …gure the marginal private bene…t, the marginal private cost,
and marginal external cost curves.
(b) Suppose that the …rm has the property rights to the river and that bar-
gaining is costless. What is the outcome of bargaining between the …rm
and the farmer, in terms of production and payments from one party to
the other?
(c) How much does social e¢ ciency increase due to the agreement you found
under b), compared to the case without agreement? How does this increase
in social e¢ ciency depend on d?
(d) Is the outcome of the bargaining di¤erent if the farmer has the property
rights to the river?

2.4 Suppose that there are two …rms, …rm 1 and …rm 2. Currently, both …rms emit
1000 ton CO2. The …rms di¤er in their cost of reducing emissions, C(E), where
E is the reduction in tons CO2. These costs, which include both lower produc-
tion and measures to make production less polluting, are given by, respectively
1
C1 (E1 ) = (E1 )2
6
1
C2 (E2 ) = (E2 )2
4
The government has decided that the CO2 emissions of these …rms together
should be reduced to 1500 ton. Hence E1 + E2 = 500.

(a) Suppose the government uses a Cap-and Trade policy (with the cap at 1500
ton). What is the price that arises on the ‘market’for a permit to emit a
ton CO2? (hint: a picture a la Figure 5.10 may help)

2
 (b) What is the di¤erence in total cost of reducing emissions between a policy
that demands that both …rms reduce emissions by 250 ton and the Cap-
and-Trade policy?
(c) Does it matter how the permits are initially distributed among the two
…rms?

Now suppose that the cost of CO2 emissions to society increase. The govern-
ment considers to demand even lower total emissions of the two companies.
Denote the total reduction by R: R = E1 + E2 . Assume that the marginal
external cost of emissions are equal to d euro per ton CO2, where d > 100.

(d) Suppose the government uses a Cap-and Trade policy. What is the socially
e¢ cient Cap on total emissions? How does the price that arises on the
‘market’for a permit to emit a ton CO2 depend on d?
(e) Now suppose the government demand that both …rms reduce their emis-
sions by the same amount (E1 = E2 = 21 R). What is the socially e¢ cient
level of emissions? Explain the di¤erence with question d).

2.5 University education has positive externalities. Suppose that total demand for
university education in a country is given by P = 100 2q, where P is the
price of university education and q the total quantity of university education.
Furthermore, marginal external bene…ts of university education are given by
M EB = 25 21 q. Marginal cost of university education is constant and equal
to 20, which is also the price of university education when the government does
not intervene in this market.

(a) Derive the socially e¢ cient level of university education, and the level of
university education that arises without government intervention.
(b) What are the total expenditures on a Pigouvian subsidy on university ed-
ucation?

Now suppose that the government decides to provide university education for
free (= provides a subsidy of 20 per unit).

(c) Calculate the e¢ ciency gain or loss that arises if university education is
provided for free, compared to the case where there is no government
intervention at all. Explain why e¢ ciency increases or decreases. (Hint:
draw a …gure)

3
 Now suppose that the private returns to education increase, leading demand to
change to P = a 2q, where a > 100.

(d) Derive how the Pigouvian subsidy depends on a. Explain.

2.6 (this is an old exam exercise with propositions, which can be true or false)
A village has 10 identical rich inhabitants, who all have 1000 euro they do not
need to spend directly. Every inhabitant has two options: saving the money or
buying a sheep for 1000 euro. Saving yields an interest rate of 10 percent per
year. A sheep can be placed on the …eld next to the village, and can be sold
after a year to a butcher. Taking care of a sheep is costless. The price P that
an inhabitant gets for his sheep after a year depends on the total number of
sheep S on the …eld:
P = 1700 100S
The …eld is part of the village, but has no owner, and can only be used for
holding sheep. Inhabitants of the village decide independently between saving
and buying a sheep

(a) The Nash-equilibrium number of sheep is 7.

(b) The socially e¢ cient number of sheep is 3.

(c) The ‘tragedy of the commons’arises when resources are non-rival and exclud-
able.

Suppose there is an eleventh inhabitant of the village. This person is poor and does
not have 1000 euro to save. The marginal utility of income of the poor inhabitant is
higher than the marginal utility of income of the 10 rich inhabitants. The social wel-
fare function of the village is equal to the sum of utilities of all inhabitants (additive).
Suppose that the poor inhabitant …nds a document that proves that he actually owns
the …eld in the village. This means that he now can ask for a payment from rich
inhabitants who want to put a sheep on the …eld.

(d) The poor inhabitant can earn maximally 900 euro per year with exploiting the
…eld.

(e) Social welfare is higher after the poor inhabitant becomes owner of the …eld.

(f) If the social welfare function is additive, then a Pareto-improvement always

leads to higher social welfare.

4
Answers to exercises week 2
Rosen 5.2 / 5.4

a) The number of parties per month that would be provided privately is P .

b) See schedule MSBp.
c) P . Give a per-unit subsidy of $b per party.
d) The total subsidy=abcd. “Society” comes out ahead by ghc, assuming the
subsidy can be raised without any e¢ ciency costs (no ‘excess burden’or ‘deadweight
loss’). (Cassanova’s friends gain gchd; Cassanova loses chd but gains abcd, which is
a subsidy cost to government.)

Rosen 5.3 / 5.6 / 5.7

a) Given that there are only a few neighbors, the costs of negotiation should be
fairly small.
d) Too many internet users to negiotate optimal agreement (i.e. transaction cost are
too high).

Rosen 5.8 / 5.11 / 5.12

Privately optimal level of X is given by M P B = M P C, or 10 X = 5, which gives
X = 5. The socially e¢ cient level is given by M P B = M P C +M D, or 10 X = 5+2,
which gives X = 3.
The gain to society is the area of the triangle whose base is the distance between
the e¢ cient and actual output levels, and whose height is the di¤erence between
private and social marginal cost at the e¢ cient level. Hence, the e¢ ciency gain is
1
2
(5 3)(7 5) = 2. Alternatively: integrate Marginal net social bene…ts M P B
M P C M D = 3 X to get total e¢ ciency 3X 12 X 2 . Substituting for X = 5 and
X = 3 yields a gain of (9 4:5) (15 12:5) = 2.
The optimal (or Pigouvian) tax is equal to the Marginal Damage at the socially
e¢ cient level. Here, this is 2. Hence, the tax raises 2 3 = 6 in revenue.

5
Rosen 5.11 / 5.13 / 5.14
If marginal costs turn out to be lower than anticipated, cap-and-trade achieves too
little pollution reduction and an emissions fee achieves too much pollution reduction.
With an inelastic marginal social bene…t function, cap-and-trade is not too bad from
an e¢ ciency standpoint, while an emissions fee causes pollution reduction to be much
greater than the e¢ cient level when marginal cost is lower than anticipated. When
marginal social bene…ts are elastic, the opposite is true.

Exercise 2.1
a) Dog is a dominant strategy for Beatrice. Given that Beatrice will choose dog,
Anne will choose to stay inside. Hence, the Nash-equilibrium is (Anne stay inside;
Beatrice dog). This is not the e¢ cient outcome. Total payo¤ is 500, whereas (Anne
go outside; Beatrice cat) yields a total payo¤ of 650.
b) They will commit to (Anne go outside; Beatrice cat). Anne pays Beatrice 150
P 300.
c) The Nash-equilibrium is still (Anne stay inside; Beatrice dog). However, this is
now Pareto-e¢ cient. Hence, they will not sign any agreement (or: the agreement will
yield the same outcome, without any monetary transfers between Anne and Beatrice).
In other words, there is no Pareto-improving agreement possible.

Exercise 2.2
a) Marginal private bene…t (here marginal revenue) M P B = 10. Marginal pri-
vate cost M P C = 2Q. Total damage is given by D = 4( 14 Q)2 = 14 Q2 . Hence, marginal
damage is given by M D = dD dQ
= 12 Q. (You can also calculate marginal damage by
using the chain rule).
b) Privately optimal level of Q given by M P B = M P C (marginal revenue equals
marginal cost). This gives Qp = 5, which yields p = 15 and Dp = 6 41 . (note: this is
of course equal to maximising pro…t, which is given by = 10Q [10 + Q2 ]).
Socially e¢ cient level of Q given by M P B = M P C + M D, which gives Qs = 4,
leading to s = 14 and Ds = 4.
c) Pigouvian tax equals M D at social optimum Qs = 4. Hence, optimal tax
equals 2. Note: with a tax of 2 per unit, the …rm indeed optimally chooses to produce
Q = 4 (pro…t 10Q 10 Q2 2Q is maximal at Q = 4).
d) Now, M P C = 2cQ, so that M P B = M P C + M D, yields 10 = 2cQ + 12 Q !
10 20
Qs = 2c+ 1 = 4c+1 (note: if c goes down, Qs increases). Hence, Pigouvian tax
2
P t = 21 4c+1
20 10
= 4c+1 . This expression decreases when c increases: @P
@c
t 40
= (4c+1) 2 < 0.

Hence, if c decreases, the Pigouvian tax should go up. The reason is that here, dam-
age increases quadratically in Q, so that M D increases in Q. Hence, if the socially
e¢ cient Q increases (here: due to a reduction in c), the Pigouvian tax also increases.

6
Exercise 2.3
a) Marginal private bene…t (here: marginal revenue) is given by M P B = P .
Marginal private cost is M P C = cQ. Damage is given by D = 2d( 21 Q)2 = 12 dQ2 .
Marginal Damage equals M D = dQ.
b) Without bargaining, the …rm produces Q = Pc . At Q = Pc , M P B = M P C
(Again, this is the same as maximising pro…t P Q 21 cQ2 with respect to Q, which
yields …rst-order condition p cQ = 0). Hence, marginal pro…t is zero at Q = Pc , while
marginal damage at that level of Q is d Pc > 0. Hence, the farmer can compensate
the …rm for a reduction in Q. This is possible as long as marginal pro…t (which
equals M P B M P C = P cQ) is smaller than marginal damage . This holds until
P
P cQ = dQ, or Q = c+d , the socially e¢ cient level. (Note: in the absence of external
cost, d = 0, the socially optimal level is the same as the privately optimal level).
Comparing Q = Pc with Q = c+d P
, we see that the …rm’s pro…t ( = P Q 21 cQ2 )
1 P2 cP 2
drops from P = 2c P 2 to S = c+d 2(c+d)2
. The damage to the farmer decreases
2 2
from DP = 12 d Pc to DS = 12 d c+d
P
. Hence, both the farmer and the …rm agree
P
on a deal where the …rm produces Q = c+d and the farmer pays an amount between
1 P2 cP 2 P 2 P 2
2c
P2 c+d 2(c+d)2
and 12 d c
1
2
d c+d
to the …rm.
Using a numerical example: If P = 12, c = 1, and d = 21 , we get P = 72, S =
144 144 P
3 2 = 64, D = 36, and DS = 16. Hence, the outcome is that the …rm
2 2( 32 )
produces Q = 8 and the farmer pays an amount between 8 and 20 to the …rm
1 1 2
c) Without bargaining, social e¢ ciency is P DP = 2c P2 2
d Pc . With
P2 cP 2 1 P 2 P2
Coasian bargaining, this is S DS = c+d 2(c+d)2 2
d c+d = 2(c+d) . Hence,
P 21 2 2 2
bargaining increases social e¢ ciency by 2(c+d) 2c
P 2 12 d Pc = 12 c2d(c+d)
P
:
The e¤ect of d on the increase in social e¢ ciency can be found by di¤erentiating this
@ 1 d2 P 2 2 2c+d
term: @d 2 c2 (c+d)
= 12 dP
c2 (c+d)2
> 0. This is intuitive: an increase in d implies a
larger marginal external cost of production, implying an even larger di¤erence between
the private and social cost of production.
d) Now, without bargaining the …rm is not allowed to produce: Q = 0. Then,
marginal damage is zero, whereas marginal pro…ts at Q = 0 are P . Hence, the …rm
can (and is willing to!) compensate the farmer for allowing him to produce, as long
P
as marginal pro…t exceeds marginal damage. This holds until Q = c+d .
p P2
With Q = 0, pro…ts and damage are zero; with Q = c+d , pro…ts equal S = c+d
cP 2 2
and damage equals DS = 12 d c+d
2(c+d)2
P
. Hence, both parties agree with a deal
p
where Q = c+d and the …rm pays the farmer an amount between DS and S . In the
numerical example, this is an amount between DS = 16 and S = 64.
Comparing b) and d), we see that in both cases, the socially e¢ cient level arises.

7
However, property rights do a¤ect the distribution of rents.

Exercise 2.4
a) Total reduction in emissions, E1 + E2 , must be equal to 500. Marginal cost of
reducing CO2 for …rm 1 equals M C1 = 31 E1 , and for …rm 2 M C2 = 12 E2 . In equilib-
rium, it must be that M C1 = M C2 and the price of a permit to emit 1 ton CO2 P
must be equal to marginal cost (see Rosen page 89-90). Hence, we have 31 E1 = 12 E2
and E1 + E2 = 500. Solving gives E1 = 300 and E2 = 200. Hence, the price for a
permit equals 100.
b) Total cost of the Cap-and-Trade system is 25000. Total cost of demanding
both …rms to reduce emissions by 250 equals 26041 23 .
c) It does not matter how the permits are distributed for the reduction in emis-
sions of each …rm (unless the government sells permits at P > 100, in which case total
emissions < 1500). However, the initial distribution of permits does a¤ect transfers.
Clearly, permits will not be bought and sold if …rm1 gets 700 permits and …rm 2 gets
800. However, if , for instance, both …rms get 750 permits, then …rm 1 will sell 50
permits to …rm 2, and receive (maximally) 5000 in return.
Note that if the government sells the permits rather than giving them for free, the
government obtains some revenue (at the expense of the …rms).
d) Let R = E1 + E2 . Marginal bene…ts of reducing emissions are given by d. Using b)
marginal cost are given by 31 E1 = 12 E2 . As E2 = R E1 , we get 13 E1 = 12 (R E1 ) !
E1 = 53 R. Equating marginal bene…t to marginal cost, we get d = 31 E1 = 15 R, so
that R = 5d. Hence, the government sets a Cap of 2000 5d ton CO2. The price
that arisis for a permit equals d (if R = 5d, then E1 = 3d, so that marginal cost in
equilibrium equal d).
This corresponds perfectly to theory: marginal external cost of emissions are con-
stant, and equal to d. Hence, a Pigouvian tax (or emission fee) would be equal to d.
Under Cap-and-trade with the same total emissions as would arise under a (Pigou-
vian) emission fee, the price for a permit equals the (Pigouvian) emission fee.
Note: You can also …nd the optimal R by maximising social e¢ ciency dR 16 (E1 )2
1
4
(E2 )2 with respect to E1 and E2 , subject to R = E1 + E2 .
e) Now, the simplest approach is to maximise social e¢ ciency dR 16 (E1 )2 41 (E2 )2
with respect to R, subject to E1 = 21 R and E2 = 12 R. After substituting for E1 and
E2 , we di¤erentiate dR 16 ( 21 R)2 14 ( 21 R)2 to R, and get …rst-order condition

1 1 24
d R R=0!R= d
12 8 5
Hence, the optimal total reduction is smaller than under Cap-and-trade. As Cap-
and-trade yields a more e¢ cient way of reducing emissions than demanding equal
reductions, it is optimal to reduce emissions further under Cap-and-trade.

8
Exercise 2.5
a) Note that the demand function gives the private marginal bene…t of education:
For each level of q, we have M P B = P . Socially e¢ cient q: P + M EB = M C, which
gives 100 2q + 25 12 q = 20 ! q = 42.
Privately optimal q (no intervention): P = M C gives 100 2q = 20 ! q P = 40.
Note: this exercise is not about some …rm maximising pro…t. Hence, maximising
pro…t of some imaginary monopolist by maximising = P q cq = (100 2q) q 25q
with respect to q is a completely wrong approach to the exercise.
b) Pigouvian subsidy is equal to marginal damage bene…t at social optimum, so
equal to M EB at q . At q = 42, M EB = 4 = Pigouvian subsidy. As q = 42, total
expenditures on the subsidy are 42 4 = 168.
c) Total e¢ ciency (= totale value created by the maket) is the sum of consumer
surplus, producer surplus, (total!) external e¤ects, and subsidy cost.
Two ways of calculating: using integration or using graph. Make sure you understand
what you do and why!
Using integration: Without subsidy, q P = 40. With subsidy: price = 0, so q S = 50.
Net marginal bene…t of education as a function of q is given by M P B + M EB
M C = 100 2q + 25 21 q 20 = 105 2 12 q. Hence, e¤ect on e¢ ciency is given by
R 50
40
105 2 12 q dq = 105 (50 40) 45 (50)2 (40)2 = 75.
Using a graph: Without subsidy, q P = 40. Price equal to constant marginal cost
implies no producer surplus. Consumer surplus equals 21 (100 20)40 = 1600. Total
external bene…ts from education is the area below the MEB-curve. This equals 21 (25
5)40 + 5 40 = 600. (note: area below MEB-curve can be split into a triangle and
a square; you can also integrate the MEB-function). Hence, total e¢ ciency without
subsidy = 2200.
With subsidy: price = 0, so q S = 50. Consumer surplus equals 21 (100 0)50 = 2500.
Total external bene…ts are 12 (25 0)50 = 625. Subsidy cost are 20 50 = 1000. Hence,
total e¢ ciency is 2500 + 625 1000 = 2125.
Conclusion: Free univerisity education yields an e¢ ciency loss of 75 (or: implies that
the value created by the market reduces by 75).
Explanation: without any subsidy, consumption is too low from social perspective.
With free education, there is overconsumption (q S = 50 > 42 = q ). Given the
numbers in this exercise, the latter ine¢ ciency is larger than the former.
Beyond the question: maximal e¢ ciency arises in the social optimum. Pigouvian
subsidy of 4 yields q = 42, consumer surplus then equals 21 (100 16)42 = 1764. Total
external bene…ts = 21 (25 4)42 + 4 42 = 609. Subsidy cost equal 4 42 = 168. Hence
total welfare equals 1764 + 609 168 = 2205.
d) When P = a 2q, we get q p = a 220 and q = a+5 21
. Pigouvian subsidy equals
2
a+5
M EB at q : 25 5
. Hence, the Pigouvian subsidy decreases in a. The reason is

9
that here, the marginal external bene…ts of education decrease in q. Hence, when
higher private bene…ts imply that more people take more education, this reduces the
marginal positive externality in equilibrium, yielding an optimally smaller Pigouvian
subsidy.
Beyond the question: note that this implies that if the returns to education increase,
this does not immediately imply that education should be subsidized more!

Exercise 2.6
a) False. In equilibrium, 6 inhabitants will by a sheep. Saving yields after one year
1100 euro. Hence, an inhabitants will buy a sheep if and only if this yields at least
1100 euro. Hence, in the Nash equilibrium, the price of a sheep must be equal to 1100
euro. This happens when there are 6 sheep.
b) True. Total value of the investment of all 10 inhabitants is P (S) S +(10 S) 1100,
where P (S) is the price of a sheep as given and 10 S is the number of inhabitants
that saves. Substituting for P (S) and maximising to S gives S = 3:
c) False. Tragedy of the commons arises when resources are rival and non-excluded.
d) True. With a fee of F per sheep, there will be S sheep, where S follows from
F
1700 F 100S = 1100 ! S = 6 (the choice of independent inhabitants
100
F
between saving and a sheep). Total income of the poor inhabitant is F (6 ):
100
F
Maximising gives …rst-order condition 6 = 0 ! F = 300. This yields 3 sheep,
50
and total income is 900 euro. Alternative reasoning using Coase Theorem: Property
rights are now assigned, so the outcome will be socially e¢ cient: 3 sheep (question
b). Maximal fee such that there will be 3 sheep is 300 euro.
e) True. No-one of the rich inhabitants is worse o¤ (all still have 1100 euro after one
year, as under question a) and the poor inhabitant is better o¤. Hence, social welfare
increases, given the additive SW-function.
f) True. In case of a Pareto-improvement (as in case of questions d and e) no one
gets lower utility while at least one person gets higher utility. With an additive SW
function, this results in higher social welfare.

10
 Public Economics
Practicum 3

Discussion Questions Rosen:

8th edition 9th & 10th editions

Chapter 9 5, 6, 7 6, 7, 8
Chapter 10 4, 5 5, 6
Chapter 11 8 8

Additional Exercises:

3.1 Suppose a group of individuals di¤ers in ability a. Ability is private information:

individuals know their own ability, but companies that might employ these
individuals only know that ability is uniformly distributed between 0 and 1.
When self-employed, an individual with ability ai earns 5000 ai . When employed
by a company, the individual creates revenue equal to 5000 ai + 1000. Assume
that companies compete for employees.

(a) When all potential employees would be working for a company, what is the
maximum wage the company would be willing to pay?
(b) If all companies o¤er the wage derived under (a), would all employees
choose to work for a company? If not, which employees would choose to
be self-employed?
(c) In equilibrium, what wage is o¤ered by companies, and which individuals
work for a company?
(d) Is this outcome e¢ cient? Is there a way to improve upon this outcome (in
terms of e¢ ciency)?

3.2 Consider the market for second-hand cars. Suppose that second-hand cars are
either good or bad, and that the fraction of bad cars is 21 . A bad car will surely
break down within 6 months, a good car never breaks down within 6 months.
Sellers know the quality of their cars. A seller’s valuation of a good car is 3000
euro; sellers’valuation of a bad car is 1000 euro. Buyers do not know the quality
of a given car, but do know that cars can be either good or bad (and, hence,
that a seller’s valuation is either 1000 or 3000 euro). Suppose …rst that buyers’
valuation of a bad car is 3000 euro (despite breaking down) and of a good car
5000 euro.

(a) Which cars will be sold, and at which price?

1
 Now suppose that buyers’valuation of a bad car is 1600 euro and of a good car
4000 euro.

(b) Is it e¢ cient for both car types to be traded?

(c) Which cars will be sold, and at which price?

Now suppose that there is a third party: the inspectors. By inspecting a car,
which costs 336 euro, they can (credibly) determine whether the quality is good
or bad.

(d) If buyers always have the car inspected before they buy it, how does this
a¤ect your answer to (c)? Does this make the market fully e¢ cient?
(e) It is actually not necessary to have all cars inspected. Show that if sellers
have to bear the cost of inspecting, sellers of bad cars will honestly declare
that their car is bad, such that only good cars are inspected.
(f) It is possible to improve e¢ ciency (i.e. decrease inspection cost) even
further by randomly inspecting only a fraction of the cars o¤ered by sellers
who claim that their car is good. What is the minimum fraction of cars
that have to be inspected when the seller claims that the car is good to
make sure that sellers of bad cars optimally declare that their car is bad?
(g) How would you react to a non-economist arguing: "It is better to buy a
car from marktplaats or e-bay than from an o¢ cial dealer, as thereby you
do not have to pay for the pro…ts of the dealer. The bene…ts are obvious
if you compare the prices."

3.3 Consider a small village somewhere in East-Asia, where the only source of in-
come is bamboo, which can be sold after it has been cut. The utility of a
bamboo-cutter is given by p
U =8 Y
where Y is income. Bamboo-cutting is dangerous, due to the presence of poiso-
nous snakes. However, bamboo-cutters can take precautions, which reduce the
probability of being bitten. Without precautions, this probability is 12 . With
precautions, the probability of being bitten is lowered to 41 . Unfortunately, the
precautions are unpleasant: it reduces utility by 1. If a bamboo-cutter does not
get bitten, his income is YG , if he does get bitten by a snake, his income is YB .

(a) What is the expected utility of a bamboo-cutter who takes precautions?

And what is it if he does not take precautions?

2
Now suppose that YG = 4 and YB = 0.

(b) Show that a bamboo-cutter optimally takes precautions.

Now suppose that bamboo-cutters can insure themselves against the loss of
income after a snake-bite. An insured bamboo-cutter pays a premium p. If
he gets bitten, the bamboo-cutter receives a payment of 4 from the insurance
company.

(c) Show that an insured bamboo-cutter does not take precautions.

The insurance market is perfectly competitive, such that premiums are actu-
arially fair (i.e. the insurance companies make no pro…ts, in expected terms).
However, insurance companies cannot determine whether bamboo-cutters take
precautions or not.

(d) Calculate the premium. Is it optimal for bamboo-cutters to buy insurance?

(e) Show that if bamboo-cutters would be able to commit to taking precautions,
and the premium would adjust to account for the lower probability of
snake-bites, then bamboo-cutters would be better o¤. Discuss.

Now suppose that there is a second group of bamboo-cutters. These people are
identical to the …rst group, except that they are born with a special ability for
avoiding snakes. This implies that their probability of being bitten is only 0:1
(regardless of precautions). The two groups are equally large.

(f) If insurance companies would be able to distinguish between the two groups,
would the second group insure themselves against the income loss due to
snake-bites?
(g) If insurance companies would not be able to distinguish between the two
groups, would the second group insure themselves against the income loss
due to snake-bites?

3
Answers to exercises week 3

Rosen 9.5 / 9.6

a)  = 100 − 25, with  = 50 gives  = 2 visits per year. Total cost is 100.
b) With 50 percent coinsurance, the individual pays $25 per visit and the quantity
demanded is 3 visits per year. The individual’s out-of-pocket costs are $75 and the
insurance company pays $75 ($25 per visit, 3 visits per year).
c) The introduction of insurance caused the quantity demanded to increase from
2 to 3 because the individual’s effective price fell from $50 to $25, but the marginal
cost is still $50 per visit. The individual consumes medical services past the point
where the marginal benefit to the individual equals the marginal cost, leading to
inefficiency or deadweight loss. The deadweight loss (triangle) is equal to $12.50.
d) If the marginal external benefit of visiting the doctor is $50, the socially
optimal outcome is  = 4. As the individual still chooses  = 3, there is a suboptimal
outcome now (i.e. the subsidy is below the marginal external benefit). Compared to
the social optimum, this outcome is still $1250 less efficient.

Rosen 9.6 / 9.7

a) There is a 95 percent chance of no illness, in which case income is $30,000,
and a 5 percent chance of illness, in which case income is $10,000 because of the
$20,000 loss. Thus, expected income = 29,000. The utility of having a certain in-
come of $29,000 is 11.66, but the expected utility of this uncertain income is only
095 · (30 000) + 005 · (10 000) = 1164.
b) An actuarially fair premium would be $1,000 since there is a one in twenty
chance that the insurance company will have to cover losses of $20,000. If the indi-
vidual buys insurance for $1,000, then they have certain income of $29,000 and the
utility of $29,000 is 11.66.
c) Setting the expected utility equal to 11.64 and solving for income yields ap-
proximately $28,388, indicating that the individual is indifferent between bearing the
risk and having expected income of $29,000 or purchasing insurance with a certain
income of $28,388. Hence, if the insurance costs $30,000 - $28,388 = $1,612, the
individual is indifferent between having insurance and not having insurance.

Rosen 9.7 / 9.8

In both options, the expected fine is 20. Denoting the value of littering by  and
income by  , the first option gives an expected value of littering equal to 08(  )+
02 (  − 100), whereas the expected value of littering with the second option is
09 (  ) + 01 (  − 200). Assuming diminishing marginal utility of income
(i.e. risk-aversion with respect to income), expected utility is smaller in the second
case than in the first case. Hence, the second option has a stronger deterrent effect,

4
leading to a larger reduction in littering. In addition, setting higher fines is cheaper
than employing more police officers.

Rosen 10.4 / 10.5

The budget constraint initially has units of Medigap on the x-axis, and other goods
on the y-axis. Given initial prices of $1 per unit for each good, and $30,000 of income,
the budget constraint has a slope of -1, and the intercepts on both axes are at 30,000
units. It is assumed that the initial utility maximizing bundle consumes 5,000 units
of Medigap, hence the indifference curve is tangent at (5000,25000). All of this is
illustrated in the figure below.

Medigap choice without minimum

standards
Other Goods

30,000

25,000

U0

5,000 30,000 Medigap

efficiency
units

Rosen 10.4(a)

After the “minimum Medigap” mandate, the consumer can either choose 0 units of
Medigap or 8,000 or more units of Medigap. Thus, part of the budget constraint is
eliminated (though the overall shape remains the same as before). After the mandate,
the point (0,30000) is available, as well as all of the points to the southeast of the
point (8000,22000). Clearly, the person’s utility must fall since the preferred choice,
(5000,25000) is no longer available. If the person attains a higher level of utility as
(0,30000) compared with (8000,22000), the person chooses to not purchase Medigap.
In this case, the marginal rate of substitution is no longer equal to the price ratio.

5
This is illustrated below.
Medigap choice with minimum
standards; no Medigap is purchased
Other Goods

30,000

25,000

U0
U1
22,000

5,000 8,000 30,000 Medigap

efficiency
units

Rosen 10.4(b)

Rosen 10.5 / 10.6

If individuals are allowed to purchase supplemental private insurance, then the budget
constraint in Figure 10.5 of the textbook is modified by drawing a line starting at
point B that runs to the southeast and is parallel to AC.

6
 Individuals can purchase
Other Goods
supplemental private insurance
A
B

C
M
Health
Insurance

Rosen 10.5(b)

If individuals pay for health insurance (rather than perceiving it as being free), and the
insurance was paid for with a lump sum tax, then the budget constraint shifts in by an
amount that depends on the household’s share of the tax burden. If the household’s
tax burden exactly equals the cost of health insurance, the budget constraint is no
longer the line segment AD but rather the segment BCD, where the segment CD
runs along the original budget constraint, except that the minimum amount of health
insurance consumed is M.

7
 Other Goods
Government health insurance is financed by
A taxes

B C

D
M
Health
Insurance

Rosen 10.5(b)

Rosen 11.8 / 11.8

a) The problem does not provide information about the utility function, so the
optimal point is where the indifference curve is drawn tangent to the budget line,
which can occur at different values depending on how the curve is drawn. In the dia-
gram below, the optimal point involves saving $8,000 and future consumption consists
of period 2 income ($5,000) plus savings with interest ($8,800). (note that the present
value of consumption is $24,525, rather than the $25,545 specified in the figure

8
 $27,000

$13,800 . Optimal
P i

Endowment
$5,000 .Point

$12,000 $20,000 $25,545 Present

Consumption

Rosen 11.8

b) If Social Security takes $3,000 from the individual in the first period and
pays him this amount with interest in the second period, then private savings falls
from $8,000 to $5,000. There would be no change in optimal consumption values.

Exercise 3.1
Drawing the uniform distribution helps!
a) Average revenue equals 12 ([5000 · 0 + 1000] + [5000 · 1 + 1000]) = 3500 (uniform
distribution of ability). Hence, when all employees work for a company, on average
they create 3500 in revenue, so that the company is maximally willing to pay a wage
equal to 3500.
b) Given wage  = 3500, all individuals with ability  such that 5000· ≥ 3500
prefer to be self-employed. Rewriting yields  ≥ 07. Hence, the most productive
employees leave the company: adverse selection.
c) Companies anticipate upon the above: when they would pay a wage equal to 3500,
the average revenue of employees would only be 12 ([5000·0+1000]+[5000·07+1000]) =
2750, implying that companies would make a loss (on average). Hence, companies
pay lower wages, which in turn implies that more (relatively able) individuals choose
to be self-employed.
To find the equilibrium, denote the ability of individuals who are just indifferent
between being self-employed and being employee by ∗ . Hence, ∗ is given by ∗ =
5000 · ∗ , where ∗ is companies’ wage in equilibrium. The wage must be such that
the average productivity of all employees is equal to this wage (zero-profit). This

9
gives: ∗ = 12 ([5000 · 0 + 1000] + [5000 · ∗ + 1000]). Substituting for ∗ = 5000 ∗ ∗
gives
1
5000 · ∗ = ([5000 · 0 + 1000] + [5000 · ∗ + 1000])
2
which can be rewritten to
1 1 1000
5000 · ∗ = (1000 + 1000) → ∗ = = 04
2 2 2500
Which implies that equilibrium wage ∗ = 2000. Hence, the least able 40% of all
individuals works for companies and earns a wage of 2000, the rest is self-employed.
d) Efficiency requires that all individuals work for a company (revenue created is
1000 higher). Inefficiency arises because of private information. If information can
be (credibly) shared, outcome may be improved. Two ways to achieve this are:
— Pay-for-performance (or screening): if companies can make individuals’ wages de-
pended on the revenue they create, the adverse selection problem does not arise. See
Personnel Economics for more info.
— Signalling: If individuals can somehow credibly signal their ability (for instance by
doing studies which are more easy to finish with high ability), companies can pay
higher wages to high-ability individuals.
Note: both of these possible solutions (can) have negative effects: pay-for-performance
may not work perfectly (risk, distortions), which reduces efficiency, and when sig-
nalling is costly, there may be too much signalling (from an efficiency perspective).

Exercise 3.2
a) When price  ≥ 3000, all cars will be traded. Then, expected value for buyer
is 4000. Hence, maximum price they are willing to pay is 4000. When the price 
would be lower than 3000, high-quality cars would not be sold, and buyers would
anticipate this. Hence, all cars are sold, for a price between 3000 and 4000.
b) Yes, for both good and bad cars, buyers’ valuation is higher than sellers’
valuation.
c) If both types are offered, then the expected value for a buyer is 2800. Hence,
if all cars would be traded, buyers would be willing to pay maximally 2800. At that
price, sellers of good cars do not offer their car for sale. Hence, only low quality cars
are sold. Buyer anticipate this, which implies that they will never pay more than
1600. Hence, the price lies somewhere between 1000 and 1600.
d) Buyers can determine the quality of the car, at the cost of 336. Hence,
including the inspection cost, their valuation of a bad car is 1264 and of a good car
3664. Now, (before inspection) buyer and seller can agree to a price between 1000
and 1264 if the car turns out to be bad and a price between 3000 and 3664 if the car
turns out to be good. Even though all cars are being traded now, which is efficient,

10
the inspection cost are a waste. Hence, the inspection cost are a necessary bad in
order to avoid an even greater inefficiency: the withdrawal of good quality cars from
the market.
e) It is better to have the cars inspected only when the seller claims that the car is
good. Since sellers know the quality of their car, inspecting bad cars is not necessary.
Hence, a better solution is that sellers of good cars have their cars inspected, and that
inspected cars trade for a price between 3336 and 4000. Bad cars are not inspected,
and are traded for a price between 1000 and 1600.
f) As inspection is costly, less inspections is better (i.e. more efficient). However,
if no car is inspected and the price of a good car would be higher than the price of a
bad car, all sellers would claim that their car is good. Here, we look for the minimum
probability of inspection of cars claimed to be good by their sellers, denoted by , at
which sellers of bad cars still do not have an incentive to claim that their car is good.
Suppose the price of a good car is  and the price of a bad car is  . If a seller
claims the car is good, the probability that the car is inspected is . Cars are sold as
good cars if they are inspected and found good, as well as when the seller claims the
car is good but the car is not inspected. Clearly, a seller of a good car will always
claim that the car is good. Consider a seller of a bad car. If he claims the car is bad,
he gets price  . If he claims the car is good, either the car is not inspected (with
probability 1 − ), in which case he gets  , or the car is inspected (which happens
with probability ), yielding price  − 336 (inspection cost). Hence, the seller of a
bad car is honest when

 ≥ (1 − ) +  [ − 336] → (1 − ) ≥ (1 − ) − 336

To find the minimum  for which this inequality holds (so that inspections are min-
imised), we should substitute for the highest possible  and the lowest possible  .
Buyers maximally pay 1600 for a bad car, sellers are willing to sell a good car at
minimally 3000 + 336 (as in expectation, they have to pay 336 in inspection cost).
This gives

(1 − )1600 ≥ (1 − )[3000 + 336] − 336

0 ≥ 1400(1 − ) − 3362
5
which can be solved for  ≥ 6
(using a good calculator/math program).

Thus, in order to maximise efficiency in this market, it should be that  = 1600,

 = 3000 + 336 · 56 = 3280, and the probability of inspection in case the seller claims
the car is good = 56 .
g) Part of the higher price of the dealer is the inspection cost and the warranty.
The information dealers provide about quality is valuable.

11
Exercise 3.3 √ √
a) Expected utility with precautions:  =√6  +√2  − 1.
Expected utility without precautions:  = 4 √+ 4 √  .
b) Expected utility with precautions:  = √6 4√ + 2 0 − 1 = 11.
Expected utility without precautions:  = 4 4 + 4 0 = 8. √ √
c) √Expected utility with insurance and precautions:  = 6 4 − +2 4 − −
1 = 8 4 −  − 1. √ √

Expected
√ utility with insurance but without precautions:   = 4 4 − +4 4−=
8 4 − .
d) The premium  should be equal to the expected payout, given that bamboo-
cutters do not take precautions:  = 12 0 + 12 4 = 2. Hence, expected utility with

√
insurance (and without precautions) is  = 8 2 = 113. Expected utility with-
out insurance is  = 11 (note that under (b), we derived that without insurance,
they would take precautions). Hence, they prefer to buy insurance.
e) If bamboo-cutters could (credibly) commit √ to taking precautions, the premium
3 1 
would be  = 4 0+ 4 4 = 1. Hence,  = 8 3−1 = 129. Thus, although moral haz-
ard makes bamboo-cutters better off given that they have insurance, the fact that the
insurance companies anticipate this moral hazard and adjust their premium implies
that bamboo-cutters are actually worse off due to their own moral hazard behaviour.
They would prefer to be able to commit themselves to always taking precautions.
f) Premium for the second group would √ be 2 = √ 09 · 0 + 01 · 4 = 04. Expected
utility without insurance is  √ = 09 · 8 4 + 01
√ · 8 0 = 144. Expected utility with
insurance equals  = 09 · 8 4 − 04 + 01 · 8 4 − 04 = 152.
g) There
£ can be ¤ only one premium. If everyone is insured, expected payout is
given by 12 12 0 + 12 4 + 12 [09 · 0 + 01 · 4] = 12 (note: first part is expected payout to
a member of group 1, second part expected payout to member of group √ 2). Expected
utility
√ of a member of group two with this insurance is  = 09 · 8 4 − 12 + 01 ·
8 4 − 12 = 134. This is smaller than their utility without insurance. Hence, group
two will not insure themselves, implying that only group 1 insures, at a premium
 = 2.
Note that this is not the efficient outcome, as groups 2 does benefit from insurance.
This is an example of how asymmetric (or private) information can lead to adverse
selection, and, thereby, to reduced efficiency.

12
 Public Economics
Practicum 4

Discussion Questions Rosen:

8th edition 9th edition

Chapter 3 8 10
Chapter 8 5 5
Chapter 11 3 3
Chapter 12 7, 8 7, 8
Chapter 13 1, 4, 7 1, 4, 7

Rosen 8.5 and 13.7 will be discussed in the tutorial.

Additional Exercises:

4.1 Suppose there are two individuals, André and Bernard, who must split 400 euro.
André’s utility of income equals

UA = ln IA

and for Bernard, it equals

UB = ln IB
where IA and IB are the amounts of income André and Bernard receive, respec-
tively, and > 0 and > 0.

(a) What is the optimal distribution of income if the social welfare function is
additive (W = UA + UB )? Explain the e¤ects of and .
(b) Suppose = = 1. What is the optimal distribution of income if the
social welfare function is given by W = ! A UA + ! B UB ? Explain the e¤ects
of ! A and ! B .

4.2 Dunya and Nikki have identical utility functions:

p
Ui = Ii

where Ii is monthly income. Suppose that Dunya works, and earns 2000 euro
per month, whereas Nikki does not have a job and, hence, does not earn any
money. Suppose that the government can freely redistribute income, and that
the social welfare function is additive.

(a) What is the optimal distribution of income?

1
 Now suppose that the government incurs costs in redistributing income. In
particular, the cost of giving Y to Nikki is equal to 0:1Y .

(b) How much income is optimally collected from Dunya, and how much does
Nikki receive (round o¤ at whole euro’s)?

4.3 A country has the following welfare program for helping the poor. Each person
has the right to a bene…t of G euro per month, but for each euro someone earns,
the bene…t is reduced with t euro. Suppose that G = 600 and t = 0:4. Consider
Betty, who can work for an hourly wage of 10 euro. Assume that a month has
30 days (such that total number of hours T = 720).

(a) What is the break-even point of the welfare program, in terms of earnings
and hours worked?
(b) What is the mathematical expression of the budget constraint under the
welfare program? Sketch Betty’s budget constraint with and without the
existence of the welfare program.

Now suppose that Betty’s utility of income (in euro’s) Y and leisure (in hours)
L equals
1 5
U = Y 6 L6

(c) What is Betty’s optimal amount of leisure if there would be no welfare

program?
(d) What is the Betty’s optimal amount of leisure under the welfare program?
How large is the bene…t she receives? What is her utility?

Now suppose that the welfare program would be such that Betty receives a
bene…t equal to the exact amount she received under d), but now independent
of her own earnings.

(e) What would be Betty’s optimal amount of leisure? Why is it di¤erent from
c) and d)?
(f) Compare Betty’s utility under d) and e). How does this relate to the trade-
o¤ between e¢ ciency and distribution?

4.4 (this is an old exam exercise with propositions)

2
Aïsha and Nora are sisters. Nora is relatively rich. Her income YN = 2500 euro.
Aïsha is relatively poor, her income is only YA = 400 euro. Aïsha’s utility can be
given by p
U A = CA
where CA is her level of consumption in euro’s. Nora´s utility function can be given
by p
U N = CN + U A
where CN is Nora’s level of consumption in euro’s and is a parameter. We assume
that 0 < 1. Nora considers giving Aïsha some amount of money G 0. After
this gift, both sisters spend their disposable income on consumption, implying that
CN = YN G and CA = YA + G.

(a) If > 0:4, then G > 0.

(b) If = 0:5, then a gift of G = 500 euro leads to a pareto-improvement compared

to G = 0.

Now suppose that the government introduces a public poverty alleviation programme.
Through this programme, Aïsha receives a bene…t of B = 200 euro, implying that
CA = YA + B + G.

(c) If = 0:5, then G = 20 euro.

(d) If > 0:4, the introduction of the public poverty alleviation programme leads
to crowding-out of private poverty alleviation.

3
Answers exercises week 4
Rosen 3.8 / 3.10 / 3.10
Social welfare is maximized when Mark’s marginal utility of income is equal to Judy’s
marginal utility of income. Taking the derivative of Mark’s utility function to …nd
his marginal utility function yields M UM = p50
IM
and taking the derivative of Judy’s
100
utility function yields M UJ = IJ . If we set M UM equal to M UJ , the condition
p

for maximization becomes IJ = 4IM and, since the …xed amount of income is $300,
this means that Mark should have $60 and Judy should have $240 if the goal is to
maximize social welfare = UM + UJ .

Rosen 8.5
a) Bill is willing to pay 25 cents to save 5 minutes, so he values time at 5 cents
per minute. The subway saves him 10 minutes per trip, or 50 cents. The value of 10
trips per year is $5. The cost of each trip is 40 cents, or $4 per year. The annual net
bene…t to Bill is therefore $1. The present value of the bene…ts = $5/0.25 = $20; the
present value of the costs is $4/0.25 = $16.
Note: the present value of an income stream of x per year that starts next year and
lasts forever equals xr , where r is the interest rate.
b) Total bene…ts = $20 55,000=$1,100,000. Total costs = $16 55,000 = $880,000.
Net bene…ts = $220,000.
c) Costs = $1.25 55,000 = $68,750. Bene…ts =($62,500/1.25) + ($62,500/1.252)
= $90,000. Net bene…t = $21,250.
d) The subway project has a higher present value. If a dollar to the “poor” is
valued the same as a dollar to the “middle class,”choose the subway project.
e) Let be the distributional weight of a poor person relative to a rich person.
The two projects are equivalent in terms of social welfare if
62; 500 62; 500
220; 000 = 68; 750 + +
1:25 [1:25]2
which can be solved for = 3:21. In other words, this distribution weight means that
$1 of income to a poor person must be viewed as more important than $3.21 to the
middle class for the legal services to be implemented rather than the subway project.

Rosen 12.7 / 12.7 / 12.8

Initially the price of food was $2 and the price of other goods was $1. The black
market for food stamps changes the price of food sold to $1. In Figure 12.3 of the
textbook, as one moves to the “northwest”from point F, the segment will now have
a slope (in absolute value) of 1 rather than 2. The black market may make the
individual better o¤ if the best point on her budget constraint AFD was initially at

4
the corner solution of point F, and the black market certainly does not make her
worse o¤. It is important to note that the black market does not always make the
recipient better o¤. If the (absolute value) of the marginal rate of substitution (MRS)
were between 1 and 2, the indi¤erence curve would not “cut” into the new part of
the budget constraint with the black market.

Black market where food stamps

are sold for fifty cents on the dollar,
Other Goods no better off

Sell food stamps for other

goods on black market

A F
U0

D
Food
Stamp Food
Allotment

Rosen 12.7A

If the MRS were less than (or equal to) 1 in absolute value, the person would be
made better o¤ and would reduce food consumption by selling the food stamps on
the black market.

5
 Black market where food stamps
are sold for fifty cents on the dollar,
Other Goods higher utility

Sell food stamps for other

goods on black market

A F U1
U0

D
Food
Stamp Food
Guarantee

Rosen 12.7B

Rosen 12.8 / 12.8 / 12.9

Pareto e¢ cient redistribution is a reallocation of income that makes some (or all)
people better o¤ while no-one is made worse o¤. This is impossible if people only
care about their own income. However, with these two consumers, Marsha’s utility
increases as Sherry’s utility increases. Thus, it may be possible to reallocate income
from Marsha to Sherrypand raise both of their utility. With Sherry’s initial utility
function of US = 100 YS , her utility withpYS = 100 equals US = 1; 000. With
Marsha’s initial utility function of UM = 100 YM + 0:8US , her utility with YM = 100
of income is UM = 1000 + 0:8 1000 = 1800. If the social welfare function is additive,
then initial welfare is W = US + UM = 2; 800.
If $36 is reallocated from Marsha to Sherry, then Sherry’s utility becomes US =
1; 166:19. Marsha’s utility becomes UM = 800 + 932:952 = 1; 732:95. In this case,
Sherry’s utility increases from 1,000 to 1,166.19, while Marsha’s utility falls from
1,800 to 1,732.95. Social welfare increases with this redistribution, going from 2,800
to 2,899.14. Thus, this redistribution increases social welfare, but it is not Pareto
e¢ cient redistribution.

Rosen 13.1
Note that the …gure below shows the correct shape of the budget constraint, but the
numbers themselves are outdated.
a) With a wage rate of $10 per hour, Elizabeth earns $100. Because the deduction

6
in California is $225, none of her earnings are counted against the $645 welfare bene…t.
Thus, her total income is $745 (=$100+$645).

Rosen 13.1

b) The actual welfare bene…ts collected by a person equals B = G t(Earnings

D), where B = actual bene…ts, G = welfare grant, t = tax rate on earned income,
and D = standard deduction. Thus, (Earnings D) is the net earnings that are
taxed away in the form of reduced bene…ts. When bene…ts equal zero (B = 0), the
expression becomes 0 = G t(Earnings D), which can be rweritten to: Earnings =
G=t + D. This is known as the “breakeven formula.” In the California context here,
the expression becomes Earnings = $645=0:5 + 225, or Earnings = $1; 515. With a
wage rate of $10 per hour, this corresponds to 151.5 hours of work per month.
c) See Figure. The number 577 should be replaced with 645 and 9 hours should
be replaced with 22.5 hours.
d) The diagram above shows one possibility. In this case, Elizabeth is both
working and on welfare, but she collects a reduced welfare bene…t.

Rosen 13.4
a) He participates in the public housing program as long as P1 P2 ca > cef .

7
 Rosen 13.4

Rosen 13.7
In all cases, the demand curve for housing slopes downward.
a) If the price of low income housing gets bid up but there is no increase in the
stock of housing, then the supply curve is perfectly inelastic (vertical).

FIGURE 8.7a –Demand curve shifts outward, perfectly inelastic supply

PHOUSING
S

P1

P0

D1

D0

Q0
QHOUSING

Rosen 13.7A

b) If there is no increase in the price of housing, but there is an increase in the

stock of housing, then the supply curve is perfectly elastic (horizontal).

8
 FIGURE 8.7b –Demand curve shifts
outward, perfectly elastic supply
PHOUSING

P0 S

D1

D0

Q0 Q1
QHOUSING

Rosen 13.7B

c) If there is an increase in both the price and quantity of housing, then the
supply curve slopes upward.
FIGURE 8.7c –Demand curve shifts
outward, upward sloping supply curve
PHOUSING

P1

P0

D1

D0

Q0 Q1
QHOUSING

Rosen 13.7C

9
Rosen 11.3
Use the basic formula for balance in a pay-as-you-go social security system:
Nb B
t=
Nw w
Call 1990 year 1 and 2050 year 2. Then
B1
t1 = 0:267( )
w1
B2
t2 = 0:458( )
w2

It follows that to keep B1

w1
=B 2
w2
we require tt21 = 0:458
0:267
= 1:72. That is, tax rates would
have to increase by 72 percent. Similarly, to keep the initial tax rate constant, we
would require B 2 B1
= = :267
w2 w1 :458
= 0:58. Relative to wages, bene…ts would have to fall
almost by half.

Rosen 8.5
a) Bill is willing to pay 25 cents to save 5 minutes, so he values time at 5 cents
per minute. The subway saves him 10 minutes per trip, or 50 cents. The value of 10
trips per year is $5. The cost of each trip is 40 cents, or $4 per year. The annual net
bene…t to Bill is therefore $1. The present value of the bene…ts = $5/0.25 = $20; the
present value of the costs is $4/0.25 = $16.
Note: the present value of an income stream of x per year that starts next year and
lasts forever equals xr , where r is the interest rate.
b) Total bene…ts = $20 55,000=$1,100,000. Total costs = $16 55,000 = $880,000.
Net bene…ts = $220,000.
c) Costs = $1.25 55,000 = $68,750. Bene…ts =($62,500/1.25) + ($62,500/1.252)
= $90,000. Net bene…t = $21,250.
d) The subway project has a higher present value. If a dollar to the “poor” is
valued the same as a dollar to the “middle class,”choose the subway project.
e) Let be the distributional weight of a poor person relative to a rich person.
The two projects are equivalent in terms of social welfare if

62; 500 62; 500

220; 000 = 68; 750 + +
1:25 [1:25]2

which can be solved for = 3:21. In other words, this distribution weight means that
$1 of income to a poor person must be viewed as more important than $3.21 to the
middle class for the legal services to be implemented rather than the subway project.

10
Exercise 4.1
a) SW is additive and, hence, is maximal when M UA = M UB , which gives
IA
= IB
, or IB = IA . As IA + IB = 400, the optimal distribution is IA = 400 +
and IB = 400 + . (Alternatively, you use Lagrange to maximise UA + UB under
constraint IA + IB = 400). As the SW-function gives the same weight to utility of oth
individuals, the person who derives higher (marginal) utility from income, as given
by and , optimally receives higher income.
b) Now, SW is maximal if ! A M UA = ! B M UB . This gives ! A I1A = ! B I1B , or
IB = !!BA IA . As IA + IB = 400, the optimal distribution is IA = 400 !A!+! A
B
and
!B
IB = 400 !A +!B . (Alternatively, you use Lagrange to maximise ! A UA + ! B UB under
constraint IA + IB = 400). Now, utility functions are identical, but the SW function
weighs utility di¤erently (! A versus ! B ). The person who has a higher weight in the
SW-function optimally receives more income.

Exercise 4.2 p p
a) Social Welfare equals W = ID + IN , where ID + IN = 2000. Marginal
utility of income equals dU
dI
= 2p1 I for both. Given the additive SW function, welfare
is maximised when marginal utilities are equal:
1 1
p = p
2 ID 2 IN
which yields ID = IN . As ID + IN = 2000, it must be that ID = p
IN = 1000 (Alterna-
p
tively, you can substitute for ID = 2000 IN and maximise W = 2000 IN + IN ).
b) Now we have that ID = 2000 1:1Y and IN = Y . Substituting this into SW-
function and maximising wrt Y gives
dW 1:1 1
= p + p =0
dY 2 2000 1:1Y 2 Y
p p
! 2000 1:1Y = 1:1 Y

which can be solved for Y 865:80. Hence, Nikki receives 865:80 and Dunya pays
in taxes 1:1 865:80 952:38. Note that reduced e¢ ciency of redistribution leads
to less redistribution (and, hence, more inequality): trade-o¤ between e¢ ciency and
distribution.

Exercise 4.3
a) Break-even point is the lowest amount of earnings E at which the individ-
ual does not receive any bene…t. Bene…t B is given by B = G tE. Solving for
B = 0 gives E = G=t ! E = 1500. Hence, after working 150 hours (or, equivalently,
L = 570) she receives no bene…t anymore.

11
b) Maximum earnings is 7200. Without welfare program, the budget constraint
is Y = 7200 10L. Each hour of leisure has opportunity cost of 10 euro. With welfare
program, the budget constraint consists of two parts: i) for L 570: Y = 7200 10L;
ii) for L 570: Y = 4920 6L. The latter part we can derive from the fact that
Y = 600 if L = 720 and Y = 1500 if L = 570.
c) This is maximisation with a constraint. Hence: either use Lagrange, or sub-
stitute the budget constraint into the utility function. The …rst method is probably
best here. Lagrange equation is
1 5
= Y 6 L 6 + (7200 10L Y)

Partial di¤erentiation wrt Y , L, and gives:

@ 1 5 5
= Y 6 L6 =0
@Y 6
@ 5 1 1
= Y 6 L 6 10 = 0
@L 6
@
= 7200 10L Y = 0
@
5 5 1 1
Combining the …rst and second gives 10 6
Y 6 L 6 = 56 Y 6 L 6 , or 2L = Y . The last
one gives Y = 7200 10L. Combining these two expressions gives 2L = 7200 10L.
Solving gives L = 600. Earnings are 10(720 600) = 1200
d) As the optimal L without welfare is larger than 570, with welfare the second
part of the budget constraint as derived under b) is the one Betty takes into account.
Hence, the lagrange equation now equals
1 5
= Y 6 L 6 + (4920 6L Y)

Partial di¤erentiation wrt Y , L, and gives:

@ 1 5 5
= Y 6 L6 =0
@Y 6
@ 5 1 1
= Y 6L 6 6 = 0
@L 6
@
= 4920 6L Y = 0
@
5 5 1 1
Combining the …rst and second gives Y 6 L 6 = 56 Y 6 L 6 , or L = 56 Y . The last one
gives Y = 4920 6L. Combining these two expressions gives L = 65 [4920 6L].
Solving gives L = 683 13 . Earnings equal 10(720 683 13 ) = 366 23 . Bene…t equals 600 -
0.4*366 23 = 453 13 . Total income is 820. Utility equals 704.42. Hence, welfare makes

12
Betty better o¤ (see e.g. Figure 13.4 in the book), but (in this example) her income
is lower.
Some additional explanation on solving 4.3 c) and d):
10 5 5 1 1
6
Y 6 L 6 = 65 Y 6 L 6 can be written as
5 1
10 L 6 5Y 6
=
6 Y 56 6 L 16
5 1
L6 Y6
2 5 = 1
Y6 L6
1 5 1
a
as x = xa
. Next step is to multiply both sides with Y 6 L 6 . This gives
5 5 1 1 5 1
L6 Y 6 L6 Y 6 Y 6 L6
2 5 = 1
Y6 L6
5 1 1 5
2L 6 L 6 = Y 6 Y 6
2L = Y

as xa xb = xa+b .
e and f) Now, Y = 453 13 + 7200 10L. Maximising Betty’s utility yields the same
relation between Y and L as under c): 2L = Y . Hence, we get 12L = 7653 31 ! L =
1 5
637:78 and Y = 1275:56. Utility equals U = (1275:56) 6 (637:78) 6 = 715:88. Betty
works less than under c, as (given her utility function) the additional income implies
that at the margin, leisure becomes relatively more attractive than the additional
consumption generated by working (in other words: leisure is a normal good). She
works more than under d), as the incentive to work is stronger (by working the same
number of hours as under d, she would get the exact same utility. But now she gets
an additional 10 euro income for working one more hour, rather than only 6 euro).
She gets a higher utility than under d), implying that the welfare program under d)
reduces e¢ ciency (relative to a system that does not distort the choice between work
and leisure as in e).

Exercise 4.4 p p
a) True. Nora maximises 2500 G + 400 + G with respect to G, yielding …rst-
1 1
order condition 2p2500 G + 2p400+G = 0 ! 2 (2500 G) = 400 + G ! G =
2500 2 400
1+ 2
Hence, G > 0 when 2500 2 400 > 0 ! 2 > 0:16 ! > 0:4.
.
b)
p False. Aishap is better o¤
p with anypgift G > 0. Nora is not worse o¤ when
2500 G + 400p+ G 2500 + 400. Solving and substituting for = 0:5
and G = 500 yields 2000 + 0:5 30 60 ! 59:72 60 which is false.

13
 p p
c) True. Maximising 2500 G + 0:5 600 + G with respect to G yields …rst-order
1 1
condition 2p2500 G
+ 0:5 2p600+G = 0. Solving gives 0:25 (2500 G) = 600 + G !
G = 20.
d) True. For > 0:4, Nora would give a positive amount of money to Aisha (see a).
This gift is reduced when Aisha receives more money (as Aisha’s marginal utility of
income reduces). See also lecture 4.

14
 Public Economics
Practicum 5

Discussion Questions Rosen:

8th edition 9th & 10th editions

Chapter 14 3, 5, 6, 8 3, 5, 6, 8
Chapter 15 5, 7, 8 9, 11, 12

Rosen 14.3 will be discussed in the tutorial.

Additional Exercises:
2
5.1 The demand for duck-tape is given by P = 10 3
Q, the supply by P = 1 + 31 Q.

(a) Find equilibrium price, quantity, and elasticities of demand and supply in
equilibrium.

An unit tax t is imposed on suppliers. Suppose t = 3.

(b) Find the new equilibrium. What is the incidence of the tax? How is this
related to the elasticities of demand and supply you found under (a)? Why
is tax incidence related to elasticities?
(c) Find the excess burden of the tax. How do you think the excess burden
depends on elasticities of demand and supply?

5.2 Betty (see exercise 4.3) has moved to another country. In this country, she can
earn a hourly wage of 20 euro. The total number of hours available is 720, and
she has to decide how to divide these between work and leisure. Of course, her
preferences have not changed: Betty’s utility of net income (in euro’s) Y and
leisure (in hours) L equals
1 5
U = Y 6 L6

(a) How many hours does Betty work?

Now suppose that the government introduces a proportional income tax: T =

0:2I, where I is gross income, so that Y = I T .

(b) How many hours does Betty work? What is the tax revenue?

1
 (c) Determine the income and the substitution e¤ect of the income tax on the
consumption of leisure. (Hint: compute utility under (b). Then …nd out
which combination of L and Y would yield this level of utility if relative
prices of Y and L would not have changed, by computing the level of a
lump-sum tax that yields the utility under b).
(d) What is the excess burden of the tax? (Note: You need to compute the
equivalent variation. Figure 15.3 may help).

5.3 Harry has a gross hourly wage of 20 euro and can divide 500 hours each month
between leisure and work. He has diminishing marginal utility of both leisure
and income. Without redistribution, Harry would work 200 hours per month.
Now suppose that the country where Harry works, has the following system of
redistribution: Everyone gets 600 euro from the government. This is paid by a
15% tax on gross income from work.

Argue whether Harry works more, less, or the same number of hours due to the system
of redistribution, as compared to the 200 hours he would work without redistribution.

5.4 Suppose there are 100 equivalent individuals in a village. The mayor of the
village is considers building a park in the centre of the village. Construction
of the park costs 800. The utility each individual derives from the park will
be equivalent to 10. To pay for the construction of the park, the mayor would
introduce a unit tax t on gasoline. Individuals’ demand for gasoline can be
given by qi = 16 2p. Supply of gasoline is perfectly elastic: p = 4.

(a) Suppose construction cost could be paid for by a lump-sum tax. Is it

e¢ cient to construct the park?
(b) Show that to raise 800, the tax t needs to be 2.
(c) What would be the excess burden of this tax?
(d) Should the park be constructed?

Now suppose that it is possible to vary the size of the park, which also a¤ects
1
the cost. Suppose the utility each individual derives from the park equals 80 C,
where C is the total amount of money invested in the park.

(e) Determine the optimal size of investment in the park C.

2
 Figure 1: Rosen 14.3

Answers to exercises week 5

Rosen 14.3: See …gure
Rosen 14.5
a) Before-tax equilibrium: P = $10 and Q = 300,000. After-tax equilibrium: P
= $10.60 and Q = 288,000. Consumers pay $10.60 and producers receive $9.60.
b) Revenue = $288,000. Consumers bear 60 percent of the tax burden and
producers bear 40 percent. So, $172,800 comes from consumers and $115,200 from
producers.
c) With a more elastic demand curve, quantity consumed will decrease even more
as a result of the tax, so the liquor tax will be more e¤ective at reducing consumption
among young drinkers.

Rosen 14.6
You can view this exercise as proving (for the case of linear demand and supply
curves) that tax incidence does not depend on whether the tax is levied on consumers

3
or producers.
Without tax, the equilibrium price can be calculated by equating demand and supply
a c
a bP = c + dP ! P =
b+d
The equilibrium output can be determined by substituting the equilibrium price into
either the supply or demand equation, which gives
ad + bc
Q=
b+d
Now suppose that a unit tax of u dollars is imposed. We will show that it doesn’t
matter which party it is imposed upon (the consumer or producer); the new equilib-
rium will be the same in either case. First, suppose the unit tax is imposed upon the
consumer. Then, the e¤ective demand curve changes into QDT = a b(P + u). Note,
the price P is here de…ned as the price consumers pay to producers. On top of this
price, consumers pay tax u. Equating demand QDT and supply QS gives
a c bu
a b (P + u) = c + dP ! P =
b+d
This is the price producers receive. Consumers pay P + u, or
a c bu a c + du
+u=
b+d b+d
Equilibrium quantity follows from substituting P into QDT or QS
a c + du cb + da dbu
QDT = a b =
b+d b+d
Note: Compared to the case without tax, producers receive a lower price, consumers
pay a higher price, and quantity is reduced.
Now suppose that the tax is imposed on producers. Then, the e¤ective supply curve
changes to QST = c + d(P u). Again, note that the price P is here de…ned as the
price consumers pay to producers. Producers have to pass on u to the government.
Equating demand QD and supply QST gives
a c + du
a bP = c + d(P u) ! P =
b+d
This is what consumers pay to producers. The producers keep P u, or
a c + du a c bu
u=
b+d b+d

4
Equilibrium quantity is given by
a c bu bc + ad dbu
QST = c + d( )=
b+d b+d
Hence, we …nd the same prices and quantity.

Rosen 14.8
Progressivity depends on the average tax rate. In the tax scheme T = 4000 + 0:2I,
the percentage of income I paid in taxes is AT R = 4000 I
+ 0:2. The best way to
determine whether the tax scheme is progressive or regressive is to see what happens
to AT R if I increases. Taking the derivative of AT R with respect to I gives
dAT R 4000
= 2 >0
dI I
Hence, the tax scheme is progressive.
a
A general scheme of this sort is T = a + tI. Here, average tax rate is AT R = I
+ t.
The e¤ect of I on AT R equals
dAT R a
=
dI I2
Hence, this tax scheme is progressive of a < 0 and regressive if a > 0.

Rosen 15.5 / 15.9

It is likely that the elasticity of demand for television is quite inelastic. It follows that
the excess burden from a $160 per year television tax is small relative to the revenues
that are collected.

Rosen 15.7 / 15.11

a) The value of the marginal product of capital in the corporate sector is given
by V M Pc = 100 Kc , and the value of the marginal product of capital in the
noncorporate sector is given by V M Pn = 80 2Kn . Without taxation, capital will
be allocated such that the marginal product is equal in both sectors (otherwise an
owner of a unit of capital in the sector where the marginal product is lowest can
gain by moving the unit to the other sector). Thus, setting V M Pc = V M Pn gives
100 Kc = 80 2Kn . Substituting for Kc + Kn = 50 gives 100 50 + Kn = 80 2Kn ,
implying that Kn = 10 and Kc = 40.

5
 Allocation of capital to the
corporate and non-corporate sectors
VMPCORPORATE VMPNONCORPORATE

100

80

VMPN=80-2KN VMPC=100-KC

QCORPORATE KC=40 QNONCORPORATE

KN=10

Rosen 15.7a

b) If a unit tax of $6 is leveled on capital employed in the corporate sector, the

after-tax value of the marginal product in the corporate sector falls. It is now given by
V M Pc0 = 100 Kc 6 = 94 Kc . Setting V M Pc0 = V M Pn gives 94 Kc = 80 2Kn .
Substituting for Kc + Kn = 50 gives 94 50 + Kn = 80 2Kn , implying that Kn = 12
and Kc = 38.
As can be seen in the …gure, the excess burden of this tax is a triangle with height
6 (the tax) and base 2 (the amount of capital shifted from the corporate to the
noncorporate sector due to the tax). Hence, the tax burden is 21 6 2 = 6.

6
 FIGURE 13.7b –Reallocation after
per-unit tax on corporate capital
VMPCORPORATE VMPNONCORPORATE

100

94

80

Excess burden from

$6 per unit tax

VMPN=80-2KN
VMPC=94-KC

QCORPORATE KC=38 QNONCORPORATE

KN=12

Rosen 15.7b

Rosen 15.8 / 15.12 (relates to 14.5)

a) Before-tax equilibrium: P = $10 and Q = 300,000. After-tax equilibrium:
P = $10.60 and Q = 288,000. Consumers pay $10.60 and producers receive $9.60.
Excess Burden = 12 (12,000)($0.60) + 12 (12,000)($0.40) = $6,000.

P
tax revenue

S
$10.6 excess burden
0
$10
unit tax

$9.6
0

D0

D1

288,00 300,00 Q
0 0

7
 Rosen 15.8
If the negative external cost were equal to $1 per gallon, then a $1 tax would achieve
an e¢ cient allocation and would create no excess burden. With a negative external
cost of $0.50 per gallon, there is still an excess burden associated with a $1 per gallon
tax, but it is smaller since the e¢ cient level of output in this market would be between
288,000 and 300,000.
Exercise 5.1
a) Equilibrium Q = 9, P = 4. Demand elasticity:
dQd P
"d =
dP Qd
d
To …nd dQdP
, either rewrite demand function to Qd = 15 1:5P and take the derivative
to P , or simply take the derivative of P = 10 32 Qd to Qd (Note: this gives dQ dP
d)
d d
and use dQdP
1
= dP=dQ dQ
d . Both give dP = 1:5. Hence, in equilibrium, "d = 2
3
= 23 .
s
Similarly, "s = dQ P
dP Qs
= 43 .
b) E¤ective supply function is now P 3 = 1 + 31 Qs or P = 4 + 13 Qs . New
equilibrium: Q = 6, P = 6. So consumers pay 6, producers keep 3. Tax revenue
equals 3 6 = 18. For consumers, price has increased by 2, for producers, the price has
decreased by 1. Hence, consumers e¤ectively pay 12 in taxes, producers 6. Consumers
thus bear two-thirds of the tax burden, producers one-third. It is no coincidence that
the elasticity of supply is twice as large as the elasticity of demand in the pretax
equilibrium and that consumers pay twice as much in taxes as producers. In general,
the distribution of the tax burden over consumers and producers depends on the
relative elasticities of demand and supply. Equal elasticities (ignoring the di¤erence
in sign) imply an equal split, and the more elastic one curve is compared to the
other, the smaller the share of the tax burden this group has to bear. This re‡ects
that relative quality of alternatives for consumption and production: if consumers
can easily switch to consuming another good, but producers have few alternative
production options, producers will end up bearing most of the tax burden.
c) Excess burden is 21 t Q = 12 3(9 6) = 4 12 . Note that this implies that to raise 18
euro, consumers lose 15 and producers lose 7.5. As discussed in week 6: if elasticities
are larger, the reduction in equilibrium quantity from a given tax is larger, which
implies that the exces burden is larger.
Exercise 5.2
a) Maximal earning are 20 720 = 14400. This gives budget constraint Y =
14400 20L. Hence, Lagrange equation is
1 5
= Y 6 L 6 + (14400 20L Y)

8
Di¤erentiating to Y , L, and gives
@ 1 5 5
= Y 6 L6 =0
@Y 6
@ 5 1 1
= Y 6 L 6 20 = 0
@L 6
@
= 14400 20L Y = 0
@
5 5 1 1
Combining the …rst two gives 16 Y 6 L 6 = 120 5
Y 6 L 6 , which can be rewritten to L =
1
4
Y . Substituting this into the third derivative gives 14400 6Y = 0. Hence, Y = 2400
and L = 600.
b) Now, maximal net income is (1 0:2) 20 720 = 11520. Budget constraint is
now Y = 11520 16L. Lagrange equation is
1 5
= Y 6 L 6 + (11520 16L Y)

Di¤erentiating to Y , L, and gives

@ 1 5 5
= Y 6 L6 =0
@Y 6
@ 5 1 1
= Y 6 L 6 16 = 0
@L 6
@
= 11520 16L Y = 0
@
5 5 1 1
5
Combing the …rst two gives 16 Y 6 L 6 = 96 Y 6 L 6 . This can be rewritten to L = 30
96
Y.
Substituting into the third derivation gives 11520 6Y = 0. Hence, Y = 1920 and
L = 600. Tax revenue is 0:2 20(720 600) = 480.
c) ith Y = 1920 and L = 600, her utility equals
1 5
U = (1920) 6 (600) 6 728:35

In case of a lump-sum tax T , Betty’s budget constraint is Y = 14400 T 20L. How-

ever, relative prices of income and leisure are identical to those under (a). Hence, only
the last derivative of the Lagrange under (a) is a¤ected by this lump-sum tax. This
implies that from the …rst two, we have again that L = 14 Y or 4L = Y . Substituting
that into the utility function, we get
1 5 1
U = (4L) 6 (L) 6 = 4 6 L

If we set this equal to 728:35, we are on the same indi¤erence curve as under (b).
1
4 6 L = 728:35 gives L 578:1. Hence, Y = 4 578:1 2312:37. Hence, the income

9
e¤ect of the tax is to reduce leisure by about 21.9 hours. The substitution e¤ect of
the tax (which makes leisure relatively less expensive) increases leisure by the same
amount.
d) Even though Betty keeps working the same hours, there is an excess burden, as the
excess burden stems from the substitution e¤ect. The easiest way to see that there
is a excess burden is by imposing a lump-sum tax of 480 euro on Betty (rather than
the 20% income tax). Then, budget constraint is Y = 14400 480 20L. Hence,
compared to (a), the only di¤erence is that 14400 is changed into 13920. Hence, with
a lump-sum tax of 480, Betty’s optimal Y and L are Y = 2320 and L = 580. The fact
the we …nd something di¤erent under (b) indicates that there is an excess burden to
this tax, even though there is no change in L.
To …nd the excess burden of the income tax, we can use the outcomes derived under
c). Substituting for L 578:1 and Y = 2312:37 into the budget constraint with
lump-sum tax T (recall: Y = 14400 T 20L), we …nd T 525:63. This means
that if we could levy a lump-sum tax rather than the 20% income tax, a lump-sum
tax of 525.63 would make Betty as well o¤ as the 20% income tax. However, the
latter only raises 480 in revenue. Hence, the excess burden of the income tax is
525:63 480 45:63: This is almost 10% of tax revenue.

Exercise 5.3
Without redistribution, Harry has 300 hours of leisure and 4000 euro of income. In
that optimal point, Harry is indi¤erent between an additional hour of leisure and an
additional hour of work (indi¤erence-curve touches budget constraint). An hour of
work yields 20 euro, so that the marginal value of an hour of leisure in that point is
also 20 euro. With redistribution, the same point can be attained: If Harry works
200 hours, his income is 600 + 0:85(20 200) = 4:000 euro. Hence, he could choose
exactly the same point. However, with redistribution, an hour of work only yields net
17 euro. From the above argumentation, his value of an hour of leisure, given that
his income is 4000 euro and he has 300 hours of leisure, is 20 euro. Hence, Harry will
optimaly work less.
This is most easily seen when drawing the old and the new budget constraints: the new
budget constraint rotates exactly at the optimal point on the old budget constraint.
Hence, it must be that the indi¤erence curve that exactly touches the old budget
constraint at 300 hours of leisure lies under the new budget constraint for a range of
values L > 300. Hence, the optimal new point lies on the part of the new budget
constraint where L > 300.

Exercise 5.4
a) Yes. If it would be possible to levy a lump-sum tax, there is no excess burden.
Hence, total cost of the park equals 800, whereas total bene…t equals 1000.

10
b) Perfectly elastic supply at p = 4 implies that the e¤ective price including
unit tax t consumers pay equals pc = 4 + t. At this price, individual demand equals
qi = 16 2(4 + t). Total tax revenue should be 800. Two ways to approach. (1)
individuals are identical and, hence, will pay the same amount in taxes. Hence, each
individual has to pay 8 in taxes, or tqi = 8. This gives

tqi = 16t 8t 2t2 = 8 ! 4t t2 4=0

which can be solved for t = 2.

(2) Total demand equals 100qi . Hence, tax revenue is t 100qi , which must be equal
to 800. This gives

t 100qi = 1600t 800t 200t2 = 800 ! 4t t2 4=0

which can be solved for t = 2.

c) Without tax, each individual buys qi = 8. With t = 2, we get qi = 4. Hence,
for each individual, excess burden is 21 t qi = 12 2 4 = 4. Total excess burden is
100 4 = 400.
d) No. Now, total economic cost of building the park is 800 + 400 = 1200. Total
bene…ts are only 1000. Hence, welfare decreases if the park is constructed. Note: you
can also determine the total cost of the tax (tax revenue + EB) as the reduction in
consumer surplus in the market for gasoline. Before tax, CS = 100 21 (8 4) 8 = 1600.
After tax, CS = 100 12 (8 6) 4 = 400. Hence, building the park would yield 1000
in wellbeing related to the park, but costs 1200 in wellbeing related to consuming
gasoline.
e) Using b), tax t yields total revenues equal to t 100qi = 1600t 800t 200t2 .
This is also total investment in the park. Excess burden equals 100 21 t qi = 100
1
2
t (16 2(4) (16 2(4 + t))) = 100t2 . Hence, total net bene…t of investing in the
1
park equals Bene…t - Cost - EB = 100 80 (1600t 800t 200t2 ) (1600t 800t 200t2 )
2
100t = 50t (3t 4). Maximising this wrt t yields …rst-order condition 200 300t =
2
0 ! t = 32 . Hence, total investment in the park is optimally 800 32 200 32 =
4000
9
= 444 49 :

11
 Public Economics
Practicum 6

Discussion Questions Rosen:

all editions
Chapter 16 1, 7, 8

Additional Exercises:

6.1 The supply of both good X and good Y is perfectly elastic at P = 10. Suppose
the demand for good X can be represented by:
1
QX = 22 PX
4
and the demand for good Y by:

QY = 50 PY

(a) Find the elasticity of demand for both good X and good Y in equilibrium.
(b) Suppose that an ad valorem tax is placed on both goods. The government
needs to collect a given amount of tax revenue on the markets for good
X and good Y, and aims to minimise the total loss in e¢ ciency due to
taxation. Good Y is taxed at a rate of 5%. What must be the rate at
which good X is taxed?
(c) Now suppose the poor people consume relatively many units of good X
while rich people consume relatively many units of good Y. Also, suppose
that the government maximises social welfare, which is given by the sum
of the utilities of all citizens. Poor people have a higher marginal utility of
income than rich people. How would this e¤ect the optimal tax rates?

6.2 Consider two goods (say, soap and shampoo) with identical demand functions:
P = 10 Q. Suppose supply of both soap and shampoo is perfectly elastic at
the same price level: P = 2. Suppose one of the two goods is currently taxed
at t = 2 per unit.

(a) What is the revenue of this tax? What is the excess burden?

Now suppose both goods are taxed at t = 1 per unit.

1
 (b) What is the total revenue of these taxes? What is the total excess burden?
(c) From an economic e¢ ciency point of view, which of these two tax schemes
is preferable?

6.3 The Erasmus University considers placing baskets with ‡owers on the Woud-
estein campus. There is place for at most 100 baskets. After careful investiga-
tion, the EUR has learned that students’(total) bene…t of these ‡ower baskets
is equal to 2F , where F is the number of ‡ower baskets. The cost of each basket
is 1. The EUR wants to pay for the ‡owers by charging a unit tax tc on each
cup of co¤ee sold. Demand for co¤ee can be represented by Qc = 300 100Pc .
The supply of co¤ee is perfectly elastic at Pc = 1.

(a) If we ignore the excess burden of the tax, how many ‡ower baskets are
placed?
(b) What level of unit tax on co¤ee is necessary to cover the cost of planting
the ‡owers? What is the excess burden of this tax? How large is the
increase in campus welfare?
(c) Show that for any number of F , the marginal cost of a ‡ower basket, taking
tc
account of the excess burden of the co¤ee tax, equals M C = 1 + .
2 2tc
(Hint: use the chain rule: @EB
@F
C
= @EB
@tc
c @tc
@F
. Also, remember that @t
@F
c
=
1
@F=@tc
)
(d) What is the optimal number of ‡ower baskets and the optimal level of the
unit tax? What is the excess burden of this tax? How large is the increase
in campus welfare?

6.4 (this is an old exam exercise with propositions, apart from e and f)
A small city has 50 identical poor citizens and 50 identical rich citizens. On a plot
of land in the city, a park can be constructed. The cost of the park are 10:000 euro.
Each rich inhabitant values the park at 150 euro, and each poor inhabitant values the
park at 50 euro. The cost of constructing the park will be covered through two new
taxes. First, a part of the cost of constructing the park will be …nanced through a
unit tax on opera tickets equal to 5 euro per ticket, levied on the producers. Second,
the remaining cost will be covered through a head tax, such that each citizen pays
the same amount in this head tax.
Only the rich citizens go to the opera. The demand for opera tickets by all rich
inhabitants together is given by:
1
P = 100 Q
20
2
Marginal cost of opera tickets is constant. In the absence of a tax on opera tickets,
supply of opera tickets is given by P = M C = 20.
The social welfare function of the city is additive:

SW = 50UR + 50UP

where UR is the utility of a rich person and UP is the utility of a poor person. The
marginal utility of income of a rich person equals 1. The marginal utility of income
of a poor person equals .

(a) The Excess Burden of the tax on opera tickets equals 250 euro.

(b) If the demand for opera tickets would be more elastic, tax revenue would be
smaller.

(c) Each rich individual would pay in total 175 euro in taxes if the park would be
constructed.

(d) If = 1:1, building the park increases social welfare.

(e) Explain how the outcome at d) relates to the trade-o¤ between e¢ ciency and
distribution.

(f) Suppose that only the poor would go to the opera (or that another good would
be taxed, only consumed by poor, with demand as stated above). For which
level of would it be optimal to build the park?

6.5 Consider a village with 50 identical rich inhabitants and 50 identical poor in-
habitants. The village considers investing in a public good, which costs 3000
euro. Each inhabitant of the village values the public good at V euro. The
public good is …nanced by a tax on good X. Demand for good X can be given
by P = 100 52 Q, where P is the price per unit in euro and Q is quantity.
Supply can be given by P = 10 + 15 Q. The government implements a unit tax
of 30 euro, to be paid by producers of good X.
Only poor inhabitants consume good X. Only rich inhabitants produce good
X. The social welfare function of the village is additive:

SW = 50UR + 50UP

where UR is the utility of a rich person and UP is the utility of a poor person. The
marginal utility of income of a rich person equals 1. The marginal utility of income
of a poor person equals > 1.

3
 1. (a) Derive the share of the tax burden that is borne by the consumers of good
X.

(b) Explain why the rich would prefer a tax on good X over a head tax.

1. (c) Suppose V = 40. Derive for which levels of it is socially optimal to invest
in the public good given that the investment is …nanced by the tax on good
X. Explain
(d) How would your answer to (c) change if the poor had a larger weight in
the social welfare function than the rich?

4
Answers to exercises week 6
Rosen 16.1
The commodities should optimally be taxed according to the inverse elasticity rule.
Using the numbers in exercise, we get
tcable satellite 7:4
= = = 14:5
tsatellite cable 0:51
Thus, tax rates on basic cable should be 14.5 times higher than tax rates on satellite
television because basic cable is inelastically demanded, while demand for satellite
television is highly elastic.
Among the assumptions that go into the inverse elasticity rule are that goods are
neither complements nor substitutes. In this case, it is likely that the …rst of these
assumptions is false –basic cable and satellite television are likely substitutes for each
other.

Rosen 16.7, 8th and 9th edition

The problem with a one-time tax on pro…ts is that the government has an incentive
to renege on its promise. The time inconsistency of optimal policy occurs when
the government cannot implement an optimal tax policy because the stated policy
is inconsistent with the government’s incentives over time. Since pharmaceutical
companies know that government has an incentive to impose a tax on pro…ts more
than once, their behavior will be a¤ected by the tax, creating an excess burden.

Rosen 16.7, 10th edition

The problem is that the R&D cost are incurred before the drugs and vaccins can be
sold. After development, international organizations may put pressure on the phar-
maceuticals to sell the drugs at relatively low cost (a bit above marginal cost), which
would make it impossible to cover the (sunk) R&D cost. The optimal policy would
be a promise to pay a price su¢ ciently high to cover the R&D cost, but this promise
is not credible: there is a time-inconsistency problem in the sense that after the de-
velopment, the international organizations are better o¤ by reneging on their promise
and demand lower prices.
There are no easy solutions: Fixing prices ex ante (before development) is di¢ cult,
as true cost are hard to forecast before development. A …xed mark-up (bove mar-
ginal cost) is also problematic, is it removes incentives to make e¢ cient cost-reducing
investments by the pharmaceuticals. In practice, international organizations spon-
sor R&D. Another solution is patents, which the pharmaceuticals can use to reduce
competition by copy-cats.

Rosen 16.8
a) True. To illustrate, consider the simplest example where there are only two

5
goods: consumption goods and leisure. The budget constraint is equal to: pC C+wL =
I, where pC and w are the prices of consumption goods and leisure, C and L are the
quantities of consumption and leisure, and I is income. Then a proportional tax
on all goods changes the budget constraint to: (1 + )pC C + (1 + )wL = I, or
rearranging, pC C + wL = I 0 , where I 0 = (1+I ) < I. Thus, a proportional tax on
all goods does not change relative prices and is equivalent to taking away time (or
earnings capacity). So it is equivalent to a lump sum tax.
b) False. If leisure is not taxed, generally e¢ ciency requires di¤erent tax rate
for di¤erent goods (Ramsey’s inverse elasticity rule). Imagine a more complicated
budget constraint: pC C + pF F + wL = I. If leisure cannot be taxed, then a tax on
commodities leads to a budget constraint of (1 + )pC C + (1 + )pF F + wL = I,
which does change the relative price of leisure compared with food or consumption
goods. Thus, it is not a lump sum tax. Instead, the inverse elasticity rule given in
equation (16.9) would suggest that the ratio of the tax rates are inversely related
to the ratio of the compensated demand elasticities for all commodities that can be
taxed. That is, C = F = F = C .
c) True. See Figure 16.3 in the book.
d) One notion of horizontal equity is that people in equal positions should be
treated equally by the tax system. Under this traditional notion of horizontal equity,
the fact that Tom’s workplace provides free access to a …tness room suggests this kind
of compensation should be taxed; Jerry pays “full taxes” on his compensation while
Tom does not. Another notion of horizontal equity relies on the utility de…nition of
horizontal equity. This concept says that if two individuals have the same utility
without taxes, they should have the same utility with taxes, and the taxes should
not a¤ect the utility ordering. One implication of the utility de…nition is that any
existing tax structure does not violate the notion of horizontal equity if individuals
are free to choose their activities and expenditures. If Tom and Jerry have free choice
between the two di¤erent jobs (and identical preferences), then the net after-tax
rewards (including amenities) must be the same at both jobs; otherwise there would
be migration. In this case, the before-tax wage on Tom’s job adjusts for the fact that
there is a fringe bene…t.

Exercise 6.1
@QX PX 1 10 10
a) "x = @PX QX
= 4 19 12
= 0:128. Similarly, "Y = 1 40 = 0:25.
b) This implies the govenrment should use the inverse elasticity rule: ttXY = ""XY .
tX 0:25
Substituting gives 0:05 = 0:128 . Hence, tX = 0:0975 (9.75%).
c) This implies that a euro tax revenue collected on the market for good X reduces
social welfare more than a euro tax revenue collected on the markt for good Y.
The inverse elasticity rule only considers e¢ ciency, and does not incorporate any

6
redistributional preferences. Hence, if the government wants to redistribute from rich
to poor, it should adjust the tax rates (decrease in tx , increase in ty ), yielding ttXY < ""XY .

Exercise 6.2
a) Without tax, Q = 8. With t = 2, Q = 6. Tax revenue 2 6 = 12. Excess
burden = 21 t Q = 21 2 2 = 2.
b) With t = 1, Q = 7. Hence, tax revenue on each good is 7, which makes total
tax revenue 14. Excess burden on each good is 21 1 1 = 12 . Hence, total excess burden
is 1.
c) The second scheme yields both more revenue and less excess burden, so is
preferable (from an e¢ ciency point of view) to the …rst scheme.

Exercise 6.3
a) Marginal bene…t of a basket M BF = 2. Marginal cost without excess burden
is 1. So all 100 places will be …lled with ‡ower baskets.
b) Tax revenue is given by tc Qc = tc [300 100(Pc + tc )] = 200tc 100t2c . (note
that Pc = 1). Total cost of 100 baskets is 100, so it must be that 200tc 100t2c = 100,
or t2c 2tc + 1 = 0. This gives tc = 1.
Excess burden is given by 12 t Qc . Without tax, Qc = 200. With tc = 1, Qc = 100.
1
So, EB = 2 1 [200 100] = 50. Bene…t of ‡owers 2F = 200. Direct cost is 100. Excess
burden is 50. Hence, welfare has increased by 200 100 50 = 50.
c) The cost of placing F baskets is F . Hence, to place F baskets, tc must be such
that tax revenue tc Qc = F . Derived above, tc Qc = 200tc 100t2c . This gives F =
200tc 100t2c . Using @t@F
c 1
= @F=@tc
, we can derive:

@tc 1
=
@F 200 200tc
This gives how tc changes with F .
The excess burden of tc equals 21 tc Qc . Without tax, Qc = 200. With tax tc ,
Qc = 200 100tc . Hence, Qc = 200 [200 100tc ] = 100tc . Substituting this gives
EB = 50t2c . So F has e¤ect on EBc through tc (a change in F a¤ects tc , which in turn
a¤ects EBc ). In math, the marginal e¤ect of F on EBc is given by: @EB @F
C
= @EB
@tc
c @tc
@F
.
This gives:
@EBC 1 tc
= 100tc =
@F 200 200tc 2 2tc
So, the direct cost of an additional basket is 1, and the associated increase in excess
burden is 2 tc2tc . Hence, total marginal cost (the marginal cost of public funds) is
1 + 2 tc2tc . (Note: in the context of government cost-bene…t analysis, this expression
gives the marginal cost of public funds).

7
d) Marginal bene…t of F is 2. Marginal direct cost is 1. Marginal excess burden
is derived above. Hence, the optimal level of F is given by:
tc
2=1+
2 2tc

This can be solved for tc = 23 .

Substituting tc = 23 into F = 200tc 100t2c gives F = 800
9
. Excess burden is
12
23
200 133 31 = 22 29 .
Welfare improvement is now 2 800 9
800
9
22 29 = 66 23 .

6.4
a) True. Before the tax, the equilibrium is given by 100 20 1
Q = 20 ! Q = 1600._
With tax, this becomes 100 1
Q = 20 + 5 ! Q = 1500. _ Hence, the Excess
20
1
Burden equals 2 5 (1600 1500) = 250. Beyond the question: this is 5 euro per rich
inhabitant.
b) True. Equilibrium quantity would go down more, as demand is more price-sensitive.
This reduces tax revenue.
c) True. Total opera tax revenue equals 5*1500 = 7500, which is 150 per rich inhabi-
tant. The remaining 2500 euro is collected through a lumpsum tax, which is 2500/100
= 25 euro per person. Hence, rich inhabitants pay 175 euro in taxes for the park.
d) False. The park a¤ects each rich inhabitant’s net utility by 150 175 5 = 30
euro. Note that the 5 comes from the Excess burden of the opera tax. It a¤ects
each poor inhabitant’s net income by 50 25 = 25 euro, which corresponds to 25
extra utility. Hence, constructing the park leads to higher SW when

50 30 + 50 25 0! 1:2

e) The economic cost of the park are 10250 euro. The park is valued at 7500 + 2500
euro (ignoring the tax, each rich citizen’s willingness to pay for the park equals 150
and each poor citizen’s willingness to pay is 50). Hence, building the park is not
e¢ cient. However, given the tax on opera, the park also redistributes from rich to
poor. If distributional motives are su¢ ciently strong (here: the di¤erence in marginal
utility between rich and poor su¢ ciently high), social welfare can increase despite the
reduction in e¢ ciency.
Beyond the question: a Pareto-improvement is theoretically possible: if, rather than
building the park, each rich citizen would give an amount between 25 and 30 euro to
a (di¤erent) poor person, everyone would be better o¤.
f) Now, the park a¤ects each rich inhabitant’s net utility by 150 25 = 125 euro.
It a¤ects each poor inhabitant’s net utility by 50 175 5 = 130 euro. Hence,

8
constructing the park leads to higher SW when
6250
50 125 50 130 0! <
6500
As marginal utility of poor is generally larger than that of rich, will be larger than
one. Hence, it is not optimal to build the park.
Note that this exercise shows that: (i) you need to take the excess burden into account
when determining the e¢ ciency / social welfare e¤ects of investments, and (ii) that
for social welfare, it matters who is e¤ectively bearing the cost of taxation.

Exercise 6.5
a) Equilibrium without tax: 100 25 Q = 10+ 51 Q ! Q = 150 and P = 40. Equilibrium
with tax 100 52 Q = 10 + 30 + 15 Q ! Q = 100, consumers pay 60 and producers keep
30. Hence, consumers bear 2=3 of the tax burden. Note: tax revenue is 3.000 euro.
b) In case of the tax on X, there is an excess burden. The excess burden is 12 30 (150 100) =
750 euro. Hence, total cost are 3750 euro. The rich are able to shift 2/3 of these total
cost to the poor. Hence, the cost to each rich person equals 1250 / 50 = 25 euro. In
case of a head tax, there is no excess burden and rich and poor would share the tax
burden equally: all inhabitants would pay 30 euro. Hence, the rich prefer the tax on
X, despite the excess burden.

c) The rich bear 1/3 of the total cost of the tax, the poor bear 2/3. Hence, it is socially
optimal to invest if 50 40 + 50 40 1250 + 2500 ! 750 500 ! 1:5. The
investment bene…ts the rich, but hurts the poor. Total bene…ts are larger than total
cost. However, if society values a euro to the poor su¢ ciently more than a euro to
the rich (here: because the marginal utility of income of a poor person is su¢ ciently
higher), it is socially optimal to refrain from making the investment.
d) Intuitively: if the poor have a larger weight in the SW-function, society values
a euro to the poor even more than before. This makes it socially less attractive
to invest, such that the investment is done only for su¢ ciently low values of . In
math: Suppose the SW-weight of the poor is w > 1. it is socially optimal to invest if
1:5
50 40 + w 50 40 1250 + w 2500 ! 750 500 w ! w
, which decreases
in w.

9
 Public Economics
Practicum 7

Discussion Questions Rosen:

All questions will be discussed in the tutorial.

8th and 9th editions 10th edition

Chapter 6 1 b), 3, 11 1 b), 3, 10

In addition to 3:
Is there anything C can do to improve upon the log-rolling outcome?

In addition to 11:

Suppose John can set the agenda. What is the outcome if people vote sincerely? What
is the outcome if they vote strategically? Is having agenda-setting power bene…cial
or not?

Additional Exercises:

7.1 Consider the preferences as in Rosen’s DQ 11b. Suppose that, rather than using
pairwise majority voting, the three people use plurality voting: they vote over
the three alternatives simultaneously, and each person can cast one vote. The
alternative that gets the highest number of votes wins. Also, suppose that in
case multiple alternatives receive the highest number of votes, John has the
power to decide among these alternatives (he is the ‘chairman’).

(a) If they vote sincerely, what is the outcome?

(b) If they vote strategically, what is the outcome?
(c) Is being the chairman bene…cial or not?

7.2 5 people vote about how to spent the afternoon together. They can choose
between the Swimming Pool (S), the Library (L), and the Bar (B). The prefer-
ences of John and Jim are such that S > L > B. Tom, Claire, and Carla prefer
B > S > L.

(a) If they use the majority rule and vote sincerely, what is the outcome?

1
 Now suppose they use rank-order voting. This is a voting procedure where each
individual ranks the alternatives. The highest ranked alternative gets 1 point,
the second-highest 2 points, etc. The alternative which receives the lowest total
number of points wins.

(b) Compared to majority rule, what could be an advantage of using rank-order

voting?
(c) If they vote sincerely, what is the outcome?
(d) What would be the outcome if the only alternatives were S and B?
(e) Relate the answers to (c) and (d) to Arrow’s Impossibility Theorem.

7.3 In Utopia, life is almost perfect. The only issue on which citizens have di¤erent
opinions is on the size of the beach s. The utility of citizen i from a beach of
size s is
Ui = js si j
where si is citizen i’s most preferred size of the beach. Among citizens, si is
uniformly distributed between 0 and 1 mile.

(a) Are these preferences single-peaked?

(b) What is the si of the median citizen?

Utopia is a representative democracy and holds presidential elections. The

elected president will decide about the size of the beach. Their are two can-
didates, mr. Red and mrs. Brown. They have similar preferences as citizens.
The ideal point of mr. Red is sR < 21 , the ideal point of mrs. Brown is sB > 12 .

Suppose that before the elections, a candidate can credibly promise to choose
a certain beach size when elected.

(c) If mr. Red commits to sR and mrs. Brown to sB , what is the probability
that mr. Red wins the election?
(d) To which beach size will the candidates commit? Who wins the election?
(e) If the candidates would be unsure about the preferences of the median
voter, what would be the trade-o¤ they face in setting their promise about
s?

Now suppose that promises are not credible. In fact, when elected, candidates
can choose any beach size, regardless of their campaign promises.

2
 (f) Who wins the election? What will be the size of the beach?

7.4 Angela, David, and Emmanuel share an apartment. They have agreed to hire a
cleaning company to clean their apartment. However, they still need to decide
for how many hours H the apartment should be cleaned. The cleaning company
charges 12 euro per hour of cleaning. Angela’s valuation of cleaning is given by

UA = 40H H 2 euro.

David’s valuation is given by

3 2
UD = 60H H euro.
2
Emmanuel’s valuation is given by
1 2
UF = 20H H euro.
2

a) Show that the socially e¢ cient level of H is 18 hours.

Suppose that they agree on sharing the cost of cleaning equally, such that each of
them pays 4 euro per hour of cleaning. Furthermore, suppose that they decide on the
level of H by pairwise majority voting.

b) Show that (given the cost sharing), each individual’s preferences regarding H
are single-peaked.

c) Which level of H wins when they use the pairwise majority voting procedure?

David is unhappy about the outcome of the vote on the number of hours of cleaning
H. He threatens to stop contributing to the cleaning cost. After a heated debate,
Angela, David, and Francois cancel their agreements regarding cleaning. As a result,
they must decide independently how many hours of cleaning to buy individually from
the cleaning company.

d) In the resulting Nash-equilibrium, David pays for H = 16 hours of cleaning.

3
Answers to exercises week 7
Rosen 6.1
b) C wins in every pairwise vote. Thus, there is a stable majority outcome, de-
spite the fact that persons 1, 2, and 3 have double-peaked preferences. This demon-
strates that although multi-peaked preferences may lead to voting inconsistencies,
this is not necessarily the case.

Rosen 6.3
Without vote-trading, neither bill would pass. If there is vote-trading, then voter B
would agree to support issue X provided voter A supports issue Y, allowing both bills
to pass. The change in net bene…ts is +3 for Issue X and -2 for Issue Y, so logrolling
results in a gain of +1.
Addition: Logrolling is very bad for C. Hence, to prevent this, C may propose to A
to vote in favour of X if A votes against Y. Compared to having both bills passed,
this bene…ts both A and C, and increases welfare by +3.

Rosen 6.11 / 6.10

a) The outcome of the …rst election (M vs. H) is M. The outcome of the second
election (H vs. L) is L. The outcome of the third election (L vs. M) is M. Majority
rule leads to a stable outcome since M defeats both H and L. Giving one person the
ability to set the agenda would not a¤ect the outcome in this case.
b) With the change in Eleanor’s preference ordering, majority rule no longer
generates a stable outcome. In a vote between M and H, the outcome is H. In a
vote between H and L, the outcome is L. In a vote between L and M, the outcome
is M. Hence, the voting paradox arises: even though individual preferences are all
transitive, the ‘collective’preference ordening is not.
This implies that it matters who sets the agenda. For example, Abigail prefers H,
so she might pit L against M …rst in order to eliminate L and avoid having L defeat
H. Thus, by having a …rst-round vote between L and M and a second-round vote
between the winner of round 1 and H will result in H being chosen.
Addition: With sincere voting, if John …rst sets L against H to eliminate H. Then
follows L against M, such that his favorite, M, wins.
If they vote strategically, this does not work. In the …rst round, Eleanor understands
that if she follows her preferences and votes for L, then she ends up with M, her least
favorite outcome. Hence, she optimally votes for H in the …rst round, such that in
the second round H wins (from M).
However, John can still get his favorite, M. To do so, he must make sure that in the
second round, M is up against L (see b)). Starting with H versus L does not work
(see above). Starting with M versus L, then the second round has either M versus H
(H wins) or H versus L (L wins). As two prefer L over H, L will win. Hence, M does

4
not win. Lastly, by starting with M versus H, the last round has either M versus L
(M wins) or H versus L (L wins). As two people prefer M over L, M will win. Hence,
John optimally starts with M versus H.
It follows that agenda-setting power is bene…cial, both in case of sincere and in case
of strategic voting.

Additional exercise 7.1

a) They all vote for their most-favoured option, implying that all options receive
the same number of votes. Hence, as chairman, John can decide. He optimally decides
for M.
b) Abigail and Eleanor realise that if they vote for their …rst-best option, they
end up with M. Hence, they can do better by voting both for H. Then, the outcome is
H, as it receives the maximal number of votes. (Note that it cannot be an equilibrium
for both to vote for L, as Abigail prefers to get M over L, and by voting for H, she
can make sure M is the outcome)
c) Being the chairman is good if people vote sincerely, but bad if they vote
strategically.

Additional exercise 7.2

a) Tom, Claire, and Carla prefer B over L and B over S, so B wins any pairwise
majority voting.
b) It uses the whole ranking of options, rather than only two options. Hence,
it takes into account that even though a majority may favour a certain option, this
option may be so bad for a minority that another option is preferable for ‘society’.
c) S gets 2 + 6 = 8 points. L gets 4 + 9 = 13 points. B gets 6 + 3 = 9 points.
Hence, S wins.
d) Without L, S gets 2 + 6 = 8 points. B gets 4 + 3 = 7 points. Hence, B wins.
e) Apparently, this voting procedure does not ful…ll the criterion that in any
situation the ranking of options should be independent of irrelevant alternatives.
Note that L is the least-preferred option under (c). Yet, if we add option L, the
outcome using rank-order voting changes from B to S.

Additional exercise 7.3

a) Yes, the peak is at si .
b) The median of a uniform distribution is equal to the mean, which is sm = 21 .
c) If the median prefers Red’s position over Brown’s position, then a majority
of citizens prefer Red’s position, leading to a victory for Red. The same holds for
Brown. The median citizen prefers Red over Brown when jsR sM j < jsB sM j and
prefers Brown over Red otherwise.
d) Drawing is very helpful here. Suppose that the optimum of Brown is closer
to the median then the optimum of Red: jsR sM j > jsB sM j. Hence, if they

5
both stick to their optimum, Brown wins. This is very bad for Red, as sB sR .
Hence, Red can improve upon this by choosing an sR < sM such that jsR sM j <
jsB sM j leading to a victory for Red. Now Brown is worse o¤, leading Brown to set
a policy closer to the median than Red, which makes Red worse o¤, etc. They will
end up setting their policy equal to the optimum of the median voter: sR = sB = sM .
Hence, both have equal chance to win.
e) Under (d), getting closer to the median policy was necessary to win the elec-
tions. However, getting closer to the median hurts the candidates, as their own
optimal policy is not the median’s. Hence, if there is uncertainty about the real me-
dian, then the candidates will trade-o¤ the probability of winning the election (which
increases by getting closer to the expected position of the median) and the policy that
they have to implement when elected (which yields higher bene…t when the remain
close to their own optimum). Thus, when the candidates do not perfectly know the
median’s position, the probability of winning becomes a continuous variable of their
position. As a result, in equilibrium, the optimal policy proposals will not be equal
to the median’s optimal position, but rather somewhere in between the median’s op-
timal policy and the candidate’s optimal policy.
f) When campaign promises are not credible, (rational) voters will foresee that
candidates will simply implement their most-preferred policy when elected. Hence,
the candidate with the optimal policy closest to the median’s position will win the
election.

Additional exercise 7.4

a) Note that cleaning is a public good here (all individuals’utilities depend on H).
Hence, M UA + M UD + M UF = 12 ! 120 6H = 12 ! H = 18.
b) Each individual has (net) value of H that can be written like U = aH bH 2 4H.
This is a inverted parabola: Maximising U gives …rst-order condition a 2bH 4 = 0,
so that the optimum lies at H = a2b4 . As M U (= a 2bH 4) < 0 for H > H and
M U > 0 for H < H , utility is decreasing in the distance between H and H : the
graph of utility as function of H has only one peak.
c) It follows that Angela prefers HA = 18, David prefers HD = 18 32 , and Emmanuel
HE = 16. Hence, Angela is the median voter, so that her optimum wins against any
alternative in case of pairwise majority voting. Note: that the outcome happens to be
equal to the social optimum is a coincidence. Generally speaking, pairwise majority
voting can lead to outcomes that can be lower or higher than the socially e¢ cient /
social-welfare maximising level.
d) The only Nash equilibrium is where Angela and Francois contribute zero, and
David responds by setting H = 16. Note that this is Nash: If David buys H = 16, for
both M UA and M UF are smaller than M C = 12, implying that it is indeed optimal
to buy zero.