Sumit Joshi

Microeconomic Theory

Price Discrimination

1 Introduction
Price discrimination can be broadly defined as the practice of selling
units of the same good at different prices to the same consumer or different
consumers. Price discrimination allows a seller to extract consumer surplus
on existing units as well as earn additional surplus through the sale of new
units. The practice is feasible when:

• Firms have monopoly/market power and therefore the ability to set

prices.

• Consumers cannot easily engage in arbitrage (e.g. one consumer buys

the product at a low price and resells it to another just below the price
of the seller).

In this chapter we will discuss various forms of price discrimination.

2 Classification
The classification of main types of discrimination follows Pigou1 . It is based
on the information that a seller has regarding consumer preferences.

• First degree price discrimination: This corresponds to the case

where the seller has complete information about the preferences of
each consumer. It is therefore the most favorable case from the seller’s
point of view. The seller can charge each consumer the maximum
price the consumer is willing to pay for each unit thereby extracting
all consumer surplus. This form of pricing is also called perfect price
discrimination or personalized pricing since prices differ across
consumers and units based on maximum willingness-to-pay.
1
A.C. Pigou (1920), The Economics of Welfare. Fourth Edition. London: Macmillan.

1
 • Second degree price discrimination: This corresponds to the
other extreme where the seller has only incomplete information about
consumer preferences. Given this limitation of information, the seller
cannot price discriminate across consumers. Rather, the seller at-
tempts to extract consumer surplus by putting forward different “pack-
ages” of a good at different prices and letting the consumers self-select
among these packages. Therefore this form of pricing is also sometimes
called menu pricing. We will consider the case where second degree
price discrimination takes the form of quantity-based price discrim-
ination, i.e. prices differ depending on the quantity package bought.
All consumers face a common non-linear price schedule involving dif-
ferent prices for different quantities. The seller can extract consumer
surplus by devising a suitable non-linear pricing schedule that targets
specific price-quantity package for each class of consumers. Incomplete
information about consumer preferences implies that a seller cannot
extract all the consumer surplus as in the case of first degree price
discrimination.

• Third degree price discrimination: This case lies in between the

other two forms of price discrimination in terms of informational con-
tent. The seller, while not having complete information about individ-
ual preferences, nevertheless observes some signal or indicator related
to consumer preferences (such as age, location, occupation etc). The
seller can then partition consumers into groups based on these signals
and charge each group a different price. This form of discrimination
is therefore also called group pricing: each member within a group
faces a common price schedule, but members from different groups
face different price schedules. Once again, since information on con-
sumer preferences is partial, the seller cannot extract all the consumer
surplus.

3 First Degree Price Discrimination

Since information about consumer preferences is complete, a monopoly firm
can price discriminate according to the maximum willingness-to-pay (ability-
to-pay or reservation price) of each consumer. As the name suggest, this
is simply the maximum price that a consumer is willing to pay for each unit
and is given by the consumer’s demand curve.

2
 P

A
P1
B
P2
Consumer
Demand

0 5 10 Q

Figure 1: Demand Curve Showing Consumer’s Reservation Prices

In Figure 1 for example, the consumer’s reservation price for the 5th unit is
P1 while the reservation price for the 10th unit is P2 . Under first degree price
discrimination the consumer pays her reservation price for each unit.2 Recall
that when the firm charges only one price, then marginal revenue is less than
price. However, when the firm practices first degree price discrimination,
then each unit is charged the reservation price. In other words, the firm’s
revenue from selling each additional unit is exactly the reservation price for
that unit. Therefore, under first degree price discrimination, the marginal
revenue curve coincides with the demand curve.

We know that a monopoly firm produces where M R = M C. Since marginal

revenue coincides with the demand curve, it follows that under first degree
price discrimination a firm produces the same output as a perfectly compet-
itive industry (see Figure 2). Each unit is sold at the consumer’s reservation
price for that unit, with the last unit sold at the perfectly competitive price.

Under this form of discrimination, the consumer pays the maximum price
she is willing to pay for each unit. Therefore the monopoly firm captures
the entire consumer surplus. Also note that since the firm produces the
same output as a perfectly competitive market structure, there is no social
deadweight loss. Therefore the monopoly firm captures the maximum total
surplus. This is shown in Figure 3 below.
2
In the real world information about consumer preferences is seldom complete. We can
think of the process where the seller does not set a fixed price but instead arrives at a
price with each buyer through negotiation and bargaining as approximating first degree
price discrimination.

3
 P
MC

A
Ppc

Demand = MR

Q
Q pc

Figure 2: Optimal Output Choice Under First Degree Discrimination

P
MC
Producer Surplus

A
Ppc

Demand = MR

Q
Q pc

Figure 3: Total Surplus Captured Through First Degree Discrimination

We now turn to a formal treatment. For any given price p, the represen-
tative consumer chooses the quantity that maximizes the quasilinear utility
function, u(Q) + y subject to the budget constraint pQ + y = m, where m
is income. The function u satisfies u(0) = 0, u0 > 0 and u00 < 0. Recall
that from the first order condition, we derive the demand function of the
representative consumer as:
p = u0 (Q) (1)
If each unit is sold at the maximum price that the consumer is willing to

4
pay, then total revenue from selling Q units is equal to:
Z Q
T R(Q) = u0 (q)dq (2)
0

It then follows that marginal revenue is equal to:3

Z Q
d
M R(Q) = u0 (q)dq
dQ 0
= u0 (Q)
= p (3)

In other words, marginal revenue under first degree price discrimination

coincides with the demand function. Let C(Q) denote the cost function of
the monopoly firm, satisfying C(0) = 0, C 0 > 0 and C 00 > 0. The profits of
the firm, and the first and second order conditions are respectively:

π(Q) = T R(Q) − C(Q)

π 0 (Q) = u0 (Q) − C 0 (Q) = 0
π 00 (Q) = u00 (Q) − C 00 (Q) < 0

The profit maximizing output Q∗ corresponds to:

u0 (Q∗ ) = C 0 (Q∗ ) (4)

i.e. where marginal revenue (equal to price) is equal to marginal cost. The
second order condition is satisfied because u00 < 0 and C 00 > 0. The result
follows by noting that the first order condition corresponds to that for total
3
In obtaining the marginal revenue, the following differentiation rule is useful. We will
need to differentiate an expression of the following kind:
Z θ
d
f (x)dx
dθ α
i.e. we are differentiating the integral with respect to a variable that occurs in one of
the limits of the integral. Suppose f (x) = F 0 (x). Then, using the Fundamental Theorem
of Calculus: Z Z
θ θ
f (x)dx = F 0 (x)dx = F (θ) − F (α)
α α
It now follows that:
Z θ
d d
f (x)dx = F (θ) = F 0 (θ) = f (θ)
dθ α dθ

5
surplus maximization. Also recall that (4) corresponds to the first order
condition for a perfect competitive industry (when the supply side is rep-
resented by a representative firm with cost function C(Q)). Therefore the
first degree price discriminating monopoly is efficient in the sense of max-
imizing total surplus. The move from a uniform price monopoly to a first
degree price discriminating monopoly increases total surplus and is therefore
welfare-enchaning. However this does not imply that perfect price discrimi-
nation is desirable. We also have to consider the issue of equity, because all
surplus is transferred from the consumer to the monopoly firm.

4 Second Degree Price Discrimination

Recall that this form of discrimination corresponds to the case where the
monopoly firm has incomplete information about consumer preferences. There-
fore the monopoly firm cannot distinguish among consumers and is forced
to offer them a common price schedule. However, the firm can vary the price
according to some characteristic of the good (quality, quantity etc). We will
consider the case where the price varies based on the quantity purchased.
This leads to a non-linear pricing scheme: The consumers pay a lower
price per unit if they purchase a larger number of units. Examples include
utilities who divide units of output into discrete blocks, and units in higher
blocks are sold at a lower price per unit (block pricing). For example,
Washington Gas Company prices consumers as follows:

Blocks of Units Price per unit

First 25 therms 46.21 cents per therm
Next 100 therms 29.68 cents per therm
Over 125 therms 24.93 cents per therm

Similarly, the rates for Virginia Electric and Power Company are:

Blocks of Units Price per Unit

First 800 kilowatt hours (kWh) 2.418 cents per kWh
Over 800 kWh 1.363 cents per kWh

This gives rise to a pricing schedule that is non-linear.

6
We will consider the case of two consumers, one of whom has a smaller
demand for the good than the other. The monopoly firm however cannot
distinguish between the two consumers. Suppose the monopoly firm offers
the two consumers the non-linear price schedule shown in Figure 4 below:

$10
1

$5
Q
0 5 10

Figure 4: A Non-Linear Price Schedule

The prices are chosen so that the low-demand consumer will buy 5 units and
the high-demand consumer will buy 10 units. The low-demand consumer
pays a total amount equal to 5×$2 = $10 while the high-demand consumer
pays a total amount equal to (5 × $2) + (5 × $1) = $15. We can therefore
equivalently imagine the monopoly firm offering the two consumers a choice
between the following (total amount-quantity) “packages”:
(r = $10, Q = 5); (r = $15, Q = 10)
and letting the low-demand consumer self-select the package (r = $10, Q =
5) and the high-demand consumer self-select the package (r = $15, Q =
10). We will now examine how a monopoly firm can design a self-selection
mechanism that induces each consumer type to buy the correct package for
their type and allow the monopolist to extract the maximum surplus.

Suppose that the utility functions of two representative consumers for the
monopolist’s product are given by:
Ui (Qi , yi ) = ui (Qi ) + yi , i = 1, 2
where ui (0) = 0, u0i > 0, u00i < 0, and y denotes money. It is assumed that
consumer 1 is high-demand and consumer 2 is low-demand in the sense that:
u1 (Q) > u2 (Q)

7
i.e. consumer 1 has a greater total willingness-to-pay for any quantity Q than
consumer 2. In addition, consumer 1 also has a greater marginal willingness-
to-pay for any quantity Q in the sense that:4

u01 (Q) > u02 (Q)

which is equivalent to saying that if Q > Q, then:

u1 (Q) − u1 (Q) > u2 (Q) − u2 (Q)

Recalling that p = u0i corresponds to the demand function for consumer i,

this property states that the demand curve for consumer 1 lies above the
demand curve for consumer 2 at all levels of output. We will assume for
simplicity that the monopolist has zero costs of production.

4.1 The Case of Observable Demands

To fix ideas, let us suppose initially that the monopoly firm has complete
information about consumer preferences. In other words, the firm can dis-
tinguish between the low-demand and high-demand consumer. Since this
corresponds to first-degree price discrimination, we know that the firm can
then extract away all consumer surplus by charging each consumer their
maximum willingness-to-pay. We can show this formally as follows. The
monopolist’s problem is:

max r1 + r2 (5)
Q1 ,Q2 ,r1 ,r2

This is however not an unconstrained problem. The monopolist will have to

ensure that both consumers will indeed buy the quantity that is intended
for them. In other words, assuming that the utility from not consuming the
good is equal to zero, we will need the restrictions that:

u1 (Q1 ) − r1 ≥ 0 (IR1)
u2 (Q2 ) − r2 ≥ 0 (IR2)

Note that ui (Qi ) is the total willingness-to-pay of consumer i for quantity

Qi (see Figure 5 below):
4
This property is also sometimes referred to as the Spence-Mirrlees single crossing
property and plays a critical role in the economics of information.

8
 P
Ui(Qi)

P = ui’(Q)

Q
0 Qi

Figure 5: Total Willingness-to-Pay

Therefore the monopoly firm cannot charge a total amount ri that ex-
ceeds the total willingness-to-pay of consumer i. The constraints (IR1) and
(IR2) are referred to as the individual rationality or participation con-
straints. The monopolist has to maximize (5) subject to the two individual
rationality constraints. Let (Qo1 , Qo2 , r1o , r2o ) denote the solution to the mo-
nopolist’s problem, where the superscript “o” indicates that this solution
corresponds to the case of observable demands

Note that the two individual rationality constraints must hold as an equal-
ity (i.e. they must be binding) at the maximum. If they do not hold as
an equality, then the monopolist is not maximizing profits. For example,
suppose that at the solution (Qo1 , Qo2 , r1o , r2o ), the constraint (IR1) holds as a
strict inequality, i.e. u1 (Qo1 ) > r1o . Let ² > 0 be such that:
u1 (Qo1 ) = r1o + ²
Then consumer 1 will continue to buy Qo1 for the amount r1o + ². Moreover,
the profit of the monopolist has further increased by ². But this contradicts
the fact that (Qo1 , Qo2 , r1o , r2o ) was the profit-maximizing solution. Therefore
we can restrict ourselves to equality constraints:
u1 (Q1 ) − r1 = 0
u2 (Q2 ) − r2 = 0
Incorporating the constraints directly into the objective function, the mo-
nopolist’s problem can be written as the following unconstrained problem:
max u1 (Q1 ) + u2 (Q2 ) (6)
Q1 ,Q2

9
The first order conditions yield:

u01 (Qo1 ) = 0
u02 (Qo2 ) = 0

i.e. each type of consumer is sold the quantity at which their demand is
equal to the marginal cost of zero. It can be checked that the second order
conditions are satisfied. The amount they are charged is equal to their total
surplus from the individual rationality constraints:

r1o = u1 (Qo1 )
r2o = u2 (Qo2 )

It is important to note that in the case of observable demands, the monop-

olist sells to each consumer the efficient level of output5 (corresponding to
price equal to marginal cost) at which total surplus is maximum and there
is no social deadweight loss. Consumers however receive no surplus.

4.2 The Case of Unobservable Demands

Now consider the case where the monopolist cannot observe whether a given
consumer is low-demand or high-demand. In this case the monopolist cannot
charge the amounts r1o and r2o . This is because the high-demand consumer
will refuse to buy Qo1 . Consider Figure 6 below. By purchasing Qo1 for the
amount r1o = A + B + C, the high-demand consumer gets zero surplus. On
the other hand, from the figure on the left in Figure 6, the high-demand con-
sumer can get a strictly positive surplus (equal to the area B) by purchasing
Qo2 for the amount r2o = A. Therefore, from the figure on the right in Figure
6, in order to induce the high-demand consumer to buy Qo1 , the monopolist
can charge a maximum amount equal to the area A + C. In other words,
the monopolist will have to ensure that the high-demand consumer gets at
least a surplus equal to the area B from buying Qo1 . Therefore, relative
to the observable case, the monopolist has to concede some surplus to the
high-demand consumer in order to induce the consumer to buy the intended
larger quantity.
5
The efficient solution when demands are observable (i.e. there are no informational
or other constraints) is also referred to as the first best solution.

10
 P P

B B

A A u2'
u2' C u1' C u1'
Q Q
Q20 Q 10 Q 20 Q10
Surplus earned by high-demand by Amount charged to high-demand to induce
purchasing Q20 purchase of Q 10

Figure 6: Conceding Surplus to High-Demand Consumer

It turns out that the monopolist can increase profits further by continuing
to sell Qo1 (the efficient quantity where price equals marginal cost) to the
high-demand consumer but reducing the quantity sold to the low-demand
consumer. By selling an inefficient amount to the low-demand consumer
(one where u02 > 0, so that price is not equal to marginal cost), the monopo-
list loses some surplus. However, it more than makes up this loss through a
smaller concession to the high-demand consumer. This is shown in Figure 7
below. Let (Qu1 , Qu2 , r1u , r2u ) denote the solution to the monopolist’s problem
in the unobservable case (where the superscript “u” refers to the fact that
demands are unobservable).6 By selling Qu2 < Q02 , the monopolist is able to
realize a net gain in surplus.
6
The solution that maximizes total surplus in the presence of informational and other
constraints is called the second best solution.

11
 P P

u1' u1'

u2' u2'

Q Q
Q 2u Q20 Q10 Q 2u Q20 Q 10

Loss in surplus from low-demand Gain in surplus from high-demand

consumer consumer

Figure 7: Gains and Losses in Unobservable Case

Let us now go through the formal workings of this model. Since the monop-
olist cannot observe if a consumer is low or high demand, the monopolist has
to ensure that each type of consumer purchases the quantity intended for
their type. Therefore, in addition to the individual rationality constraints,
we also need the incentive compatibility (self-selection) constraints.
The monopolist’s problem is:

max r1 + r2 (7)
Q1 ,Q2 ,r1 ,r2

This is however not an unconstrained problem. The monopolist will have to

ensure that both consumers will indeed buy the quantity that is intended
for them. In other words, assuming that the utility from not consuming the
good is equal to zero, we will need the restrictions that:

u1 (Q1 ) − r1 ≥ 0 (IR1)
u2 (Q2 ) − r2 ≥ 0 (IR2)

u1 (Q1 ) − r1 ≥ u1 (Q2 ) − r2 (IC1)

u2 (Q2 ) − r2 ≥ u2 (Q1 ) − r1 (IC2)

12
(IC1) states that the high demand consumer should not buy the quantity
intended for the low demand consumer. Similarly, (IC2) requires that the
low demand consumer should not buy the quantity intended for the high
demand consumer. All four constraints are not independent. Moreover,
some of them will be binding at the profit-maximizing solution (i.e. hold as
an equality).

1. (IR2) and (IC1), along with u1 > u2 , imply (IR1).

Note that:

u1 (Q1 ) − r1 ≥ u1 (Q2 ) − r2 from (IC1)

> u2 (Q2 ) − r2 since u1 > u2
≥ 0 from (IR2)

and therefore (IR1) is satisfied (as a strict inequality). We can there-

fore remove (IR1) from the list of constraints.

2. (IR2) must be binding.

The argument is the same as in the observable case. Suppose (IR2)
holds as a strict inequality at the profit-maximizing solution, (Qu1 , Qu2 , r1u , r2u ),
i.e.
u2 (Qu2 ) > r2u

Increase r2u to r2u + ², ² > 0, so that:

u2 (Qu2 ) = r2u + ²

Correspondingly increase r1u to r1u +². Note that (Qu1 , Qu2 , r1u +², r2u +²)
satisfies (IR2) as an equality as well as the two incentive compatibility
constraints (because the LHS and RHS of (IC1) and (IC2) increase by
the same amount ²). The monopolist’s profits are now equal to:

(r1u + ²) + (r2u + ²) = r1u + r2u + 2²

Since profits have increased by 2², it follows that (Qu1 , Qu2 , r1u , r2u ) is
not profit-maximizing, a contradiction. Therefore we can write (IR2)
as an equality.

13
 3. Monotonicity of quantity: The high demand consumer is sold
a greater quantity than the low demand consumer, i.e. Q1 ≥
Q2 .
Adding (IC1) and (IC2) yields:

u1 (Q1 ) + u2 (Q2 ) ≥ u1 (Q2 ) + u2 (Q1 ) (8)

and rearranging we have:

u1 (Q1 ) − u1 (Q2 ) ≥ u2 (Q1 ) − u2 (Q2 ) (9)

Since u01 > u02 , (9) implies that Q1 ≥ Q2 .7

4. (IC1) must be binding.

Suppose (IC1) holds as a strict inequality at the profit-maximizing
solution, (Qu1 , Qu2 , r1u , r2u ), i.e.

u1 (Qu1 ) − r1u > u1 (Qu2 ) − r2u

Increase r1u to r1u + ², ² > 0, such that:

u1 (Qu1 ) − (r1u + ²) = u1 (Qu2 ) − r2u

Leave r2u unchanged. Then (Qu1 , Qu2 , r1u + ², r2u ) satisfies (IR2), (IC1)
and (IC2). The monopolist’s profits from (Qu1 , Qu2 , r1u + ², r2u ) are:

(r1u + ²) + r2u = r1u + r2u + ²

Since profits have increased by ², it follows that (Qu1 , Qu2 , r1u , r2u ) is not
profit-maximizing, a contradiction. Therefore we can write (IC1) as
an equality.
7
Suppose not and let Q1 < Q2 . Then u01 > u02 implies that:

u1 (Q2 ) − u1 (Q1 ) > u2 (Q2 ) − u2 (Q1 )

and rearranging we get:

u1 (Q1 ) + u2 (Q2 ) < u1 (Q2 ) + u2 (Q1 )

which contradicts (8). This in turn implies that at least one of the incentive compatibility
conditions is violated.

14
 5. (IC1), along with u01 > u02 and the monotonicity of quantity,
implies (IC2).
Note that:

r1 − r2 = u1 (Q1 ) − u1 (Q2 ) from (IC1)

≥ u2 (Q1 ) − u2 (Q2 ) from u01 > u02 and Q1 ≥ Q2

and rearranging yields (IC2). We can therefore remove (IC2) from the
list of constraints and substitute it with the monotonicity restriction
Q1 ≥ Q2 .

The monopolist’s problem now becomes:

max r1 + r2 (10)
Q1 ,Q2 ,r1 ,r2

subject to:

u2 (Q2 ) − r2 = 0 (IR2)
u1 (Q1 ) − r1 = u1 (Q2 ) − r2 (IC1)
Q1 ≥ Q2 (Monotonicity)

Note that from (IR2) and (IC1):

r2 = u2 (Q2 )
r1 = u1 (Q1 ) − u1 (Q2 ) + r2
= u1 (Q1 ) − u1 (Q2 ) + u2 (Q2 )

We now proceed by ignoring the monotonicity constraint and focusing only

on (IR2) and (IC1). This relaxed monopolist’s problem can be rewritten
as the following unconstrained problem:

max [u1 (Q1 ) − u1 (Q2 ) + u2 (Q2 )] + u2 (Q2 )

Q1 ,Q2

The standard procedure is to solve this relaxed problem and then ensure
that the solution indeed satisfies the monotonicity restriction.

The first order conditions for the relaxed problem are:

u01 (Q1 ) = 0 (11)

−u01 (Q2 ) + u02 (Q2 ) + u02 (Q2 ) = 0 (12)

15
Letting (Qu1 , Qu2 ) denote the profit-maximizing outputs in the unobservable
case, from (11):
u01 (Qu1 ) = 0 (13)
Therefore the high-demand consumer is sold the efficient (or first best) level
of output similar to the observable case. The marginal willingness-to-pay, or
price, is equal to marginal cost for the high demand consumer. From (12):

u02 (Qu2 ) = u01 (Qu2 ) − u02 (Qu2 )

> 0 (14)

The marginal willingness-to-pay, or price, is greater than marginal cost for

the low-demand consumer. Therefore the low demand consumer is sold an
inefficient amount. To see this, note that in the observable case:

u02 (Qo2 ) = 0

Therefore:
u02 (Qu2 ) > u02 (Qo2 )
and since u002 < 0 (i.e. u02 is decreasing):

Qu2 < Qo2

The low demand consumer is sold less than the efficient amount.

From (IR2):
r2u = u2 (Qu2 )
The low-demand consumer pays an amount equal to the total surplus re-
ceived and therefore gets zero consumer surplus. Now consider the high-
demand consumer. Recall that (IR1) is satisfied as a strict inequality, i.e.
the high demand consumer gets a strictly positive consumer surplus. Note
that (IC1) is satisfied as an equality. Therefore the high demand consumer
is conceded a surplus exactly equal to the surplus that the high demand
consumer will get by defecting to the low quantity. From (IC1):

r1u = u1 (Qu1 ) − u1 (Qu2 ) + r2u

= u1 (Qu1 ) − u1 (Qu2 ) + u2 (Qu2 )
= u1 (Qu1 ) − [u1 (Qu2 ) − u2 (Qu2 )]
< u1 (Qu1 )

16
The high demand consumer gets positive consumer surplus, but just enough
to meet the incentive compatibility constraint.

Finally we ensure that the solution to the relaxed problem satisfies monotonic-
ity in quantities. Note that:

u01 (Qu1 ) = 0 from (13)

< u02 (Qu2 ) from (14)
< u01 (Qu2 ) from single-crossing

Therefore, u01 (Qu1 ) < u01 (Qu2 ). Since u001 < 0, it follows that Qu1 > Qu2 . There-
fore the solution meets the monotonicity constraint.

4.3 Welfare Comparison

We will now try to compare total surplus under second degree price dis-
crimination with that under a uniform price monopoly. Consider Figure 8
below where Ps is the price charged by a uniform price (simple) monopo-
list, and Qsi is the output produced by consumer i. (Note that this output
under a simple monopoly corresponds to the point where marginal revenue
is equal to zero marginal cost. We have omitted the marginal revenue curve
for simplicity).

P P

u1' u1'

u2' u2'

Ps Ps
L
G
Q Q
Q2u Q2s Q1s Q1u
Loss in surplus from low-demand Gain in surplus from high-demand
consumer due to price discrimination consumer due to price discrimination

Figure 8: Welfare Comparison

17
 Consider the figure on the left. The area under the demand curve u02 up
to output Qs2 is the total surplus (consumer surplus plus producer surplus)
under a simple monopoly. Under second degree price discrimination, the
total surplus is extracted by the monopolist and corresponds to the area
under the demand curve u02 up to output Qu2 . Therefore the loss in surplus
when moving from a uniform price to second degree price discrimination is
equal to area L. Now consider the figure on the right. The area under the
demand curve u01 up to output Qs1 is the sum of consumer and producer
surplus corresponding to a simple monopoly. The area under the demand
curve u01 up to output Qu1 is the total surplus (surplus conceded to high-
demand consumer plus surplus accruing to the monopolist) under second
degree price discrimination. Therefore the gain in surplus when moving
from a uniform price to second degree price discrimination is given by area
G. The change in welfare is thus ∆W = G − L.

Note from Figure 8 that:

L ≥ P s × (Qs2 − Qu2 )
G ≤ P s × (Qu1 − Qs1 )

Let us define:

∆Q1 = Qu1 − Qs1

∆Q2 = Qu2 − Qs2

Then ∆Q1 > 0 and ∆Q2 < 0. It follows that:

∆W = G−L
≤ [P s × (Qu1 − Qs1 )] − [P s × (Qs2 − Qu2 )]
= [P s × (Qu1 − Qs1 )] + [P s × (Qu2 − Qs2 )]
= PS × [∆Q1 + ∆Q2 ]

Therefore a necessary condition for ∆W ≥ 0 is that ∆Q1 + ∆Q2 ≥ 0. In

other words, ∆W ≥ 0 implies ∆Q1 + ∆Q2 ≥ 0. The (logically equivalent)
contrapositive of this result is that ∆Q1 + ∆Q2 < 0 implies ∆W < 0.
Therefore second degree price discrimination is welfare enhancing relative
to a uniform price only if aggregate output increases; if aggregate output
falls, then welfare under second degree price discrimination is lower.

We saw that first degree price discrimination was unambiguously welfare

enhancing relative to a uniform price monopoly. Under second degree price

18
discrimination, since ∆Q1 > 0 and ∆Q2 < 0, the welfare comparison is
ambiguous.

5 Third Degree Price Discrimination

Suppose consumers can be divided into two groups based on some observ-
able signal (e.g. age, location, occupation). Then the monopoly firm can
price discriminate between the groups based on the elasticity of demand.
The monopoly firm can divide the total market into groups based on their
elasticity of demand (or sensitivity to price). Consumers who belong to
a relatively more price inelastic group pay a higher price per unit. This
form of price discrimination is observed in the real world when firms divide
their market into private (elastic) versus business (inelastic), or senior citi-
zens/children (elastic) versus adults (inelastic). The objective of the seller
is to once again extract higher consumer surplus from the relatively more
price inelastic groups.

In Figure 9, the monopoly firm divides its total demand into two groups
based on price elasticity. It then determines the price and quantity for each
group as follows:

P P P
MC

P1

P2 D2

MC of
Qm { MR D1 + D2

Q1
MR1
D1 MR2

Qm Q2

Total Market Inelastic Group Elastic Group

Figure 9: Third Degree Price Discrimination

19
 • The monopoly first determines its total output Qm by equating the
marginal revenue corresponding to market demand (i.e. total demand
D1 + D2 ) to its marginal cost.
• In order to allocate this output across the two groups, it then equates
the marginal cost of Qm to the marginal revenue for each group. It
therefore sells Q1 to the inelastic group and Q2 to the elastic group.
• Once the allocation of quantities is fixed, the firm then determines the
price to be charged for each group by going to the demand curve for
that group. Thus the inelastic group pays P1 per unit while the elastic
group pays P2 per unit. Note that the inelastic group pays a higher
price.

There are some important facts to note about third degree price discrimi-
nation. First, the optimal allocation of output across groups requires that
marginal revenues should be equated across the groups, i.e.
M R1 = M R2
If the marginal revenues are unequal, then the monopoly firm can increase
its profits further by redistributing output between the groups. For example,
suppose that:
M R1 = 20 > M R2 = 15
Then the firm can increase profits by transferring one unit from Group 2
to Group 1. By reducing the output for Group 2 by one unit, the loss in
revenue is $15. By selling this extra unit to Group 1, the gain in revenue is
$20. Therefore this reallocation of output increases the firm’s profits by $5.
As long as M R1 6= M R2 , there will always be scope for increasing profits
through redistribution of output. It is only when M R1 = M R2 (= M C)
that profits are maximized.

We can now examine this pricing scheme formally. Suppose consumers can
be divided into two groups based on some observable signal (e.g. age, loca-
tion, occupation).
u1 (Q1 ) + y1 utility of group 1; u01 (Q1 ) = P1 demand of group 1
u2 (Q2 ) + y2 utility of group 2; u02 (Q2 ) = P2 demand of group 2
Let C(Q) be the firm’s cost function. The maximization problem is:
max P1 (Q1 )Q1 + P2 (Q2 )Q2 − C(Q1 + Q2 )
Q1 ,Q2

20
The first order conditions can be written as:

M R1 (Q1 ) − C 0 (Q1 + Q2 ) = 0
M R2 (Q2 ) − C 0 (Q1 + Q2 ) = 0

In particular:
M R1 (Q1 ) = M R2 (Q2 )
Letting E1 and E2 denote the price elasticities of demand for Group 1 and
Group 2, we know from the relationship between price, marginal and price
elasticities that:
∙ ¸
1
M R1 = P1 1 −
E1
∙ ¸
1
M R2 = P2 1 −
E2

When the outputs are allocated correctly between the two groups, then
M R1 = M R2 . Therefore:
∙ ¸ ∙ ¸
1 1
P1 1 − = P2 1 −
E1 E2

and the prices charged to the two groups should depend on the price elas-
ticities for the two groups as follows:
h i
1
P1 1 − E2
=h i
P2 1− 1 E1

In particular, if E1 > E2 , then P1 < P2 . For example, if E1 = 4 > E2 = 2,

then: £ ¤
P1 1 − 12 2
= £ ¤
1 = 3
P2 1− 4
P1
Since P2 < 1, it follows that P1 < P2 .

Note that third degree price discrimination can only be effective if the
monopoly firm can prevent resale among the groups. If the monopoly firm
cannot prevent resale, then the elastic group can purchase the product from
the monopolist at the price P2 and sell it to the inelastic group at a price
just below P1 thereby undercutting the monopolist.

21
5.1 Welfare Comparison
We can compare how welfare changes when we move from a uniform price
monopoly to third degree price discrimination by using the same technique
as in the case of second degree price discrimination. Consider Figure 10 indi-
cating the price and output under third degree price discrimination (Pi , Qi ,
i = 1, 2) and for a simple uniform price monopoly (PS , Qsi , i = 1, 2). The
figure on the left shows the surplus gained and the figure on the right shows
the surplus lost.

Group 1 Group 2
P P

Area, G = gain Area, L = loss

P2
PS
P1 L
G
MC

Q1s Q1 Q2 Q2s
Q Q

Figure 10: Gains and Losses

We can bound these areas as shown below in Figure 11. Note that G ≤ A
and L ≥ B and therefore:

∆W = G−L
≤ A−B
= [PS − M C] [Q1 − Qs1 ] − [PS − M C] [Qs2 − Q2 ]
= [PS − M C] [Q1 − Qs1 ] + [PS − M C] [Q2 − Qs2 ]
= [PS − M C]∆Q1 + [PS − M C]∆Q2
= [PS − M C] [∆Q1 + ∆Q2 ]

Therefore a necessary condition for ∆W ≥ 0 is that ∆Q1 + ∆Q2 ≥ 0.

Therefore third degree price discrimination leads to an increase in welfare
if aggregate output increases. With linear demand curves and constant

22
marginal costs, it can be checked that aggregate output is unchanged, i.e.
∆Q1 + ∆Q2 = 0. Therefore ∆W ≤ 0.

However, it is quite possible that ∆Q1 + ∆Q2 > 0. This would be the case
if the simple monopoly did not sell to one group because its demand was
below marginal cost. Price discrimination allows a monopolist to sell to both
groups and thus enhance welfare.

Group 1 Group 2
P P

P2
PS
P1 B
A
MC

Q1s Q1 Q2 Q2s
Q Q

Figure 11: Bounding Gains and Losses

6 Two-Part Tariffs
We now consider another mechanism for extracting surplus called a two-
part tariff. This scheme requires a consumer to first pay an entry fee
for the right to consume a product, and then a constant price for each
unit consumed. An example would be paying a fixed monthly amount for
a cellular calling plan (entry fee) and then a price per minute for the calls
made. We will now examine how a monopolist should set the per unit price
and the entry fee in order to maximize profits.

The two-part tariff calls for a uniform price per unit. Suppose the monopolist
sets this price at the monopoly price, Pm as in Figure 12. By charging one
price, Pm , the monopoly concedes the consumer surplus shown below. Under
a two-part tariff scheme, the monopolist can capture this entire consumer
surplus in the form of an entry fee. Thus the monopolist in principle can

23
extract away all the surplus from the consumer. Note, however, in Figure
12 that the monopoly firm is not maximizing the surplus that it can get.
By charging the monopoly price, and producing the monopoly output, it
is creating social deadweight losses. By reducing its per unit price, and
increasing its output, to the perfectly competitive level, the monopoly firm
can capture this potential surplus as well.

P Consumer Surplus = Entry Fee

MC

Pm

Ppc Deadweight
Loss

Producer
Surplus P
MR
Q
Qm Q pc

Figure 12: Surpluses Under Uniform Pricing

P
Consumer Surplus MC
= Entry Fee

A
P*=Ppc

Producer P
Surplus
Q
Q pc

Figure 13: Two-Part Tariff

Therefore, as shown in Figure 13, a profit-maximizing monopoly firm us-

24
ing a two-part tariff will set the price at the perfectly competitive level.
This ensures that total surplus is at its maximum because deadweight losses
are equal to zero. The monopoly then captures this maximum surplus by
extracting away the (now larger) consumer surplus as an entry fee.

We will now derive the above results formally. The set up is a little different
because it is convenient to work with the indirect utility function rather than
the direct utility function. The indirect utility function of the representative
consumer is given by:
V (P, m) = v(P ) + m
where m denotes income. The function v is assumed to satisfy standard
assumptions:

v(P ) = 0 for P ≥ P
v 0 (P ) < 0
v 00 (P ) > 0

where P is some sufficiently high “choke price” above which there is no

demand for this good by the representative consumer.

From Roy’s Identity, the direct demand function is:

∂V
∂P
Q(P ) ≡ − ∂V = −v 0 (P )
∂m

Since v 00 > 0, the direct demand function is downward-sloping:

Q0 (P ) = −v 00 (P ) < 0

Then consumer surplus from a given price P < P is given by:

Z P Z P
Q(p)dp = − v0 (p)dp
P P
Z P
= v 0 (p)dp interchanging the limits of the integral
P
= v(P ) − v(P ) from the Fundamental Theorem of Calculus
= v(P ) since v(P ) = 0

Therefore consumer surplus is the area under the direct demand function
between prices P and P .

25
Let us start with the individual rationality (or participation) constraint.
Suppose the monopolist sets an entry fee e and a per unit price P . The
representative consumer will buy the good at a price only if:

v(P ) + m − e ≥ m
|{z}
| {z }
Utility from buying good and paying (e,P ) Utility from not buying and retaining income m

from which we get:

e ≤ v(P )
Therefore, if the monopolist sets a price P , then the entry fee cannot exceed
the consumer surplus from P , v(P ). Also note that this individual rational-
ity constraint will be binding at the profit-maximizing solution. If e < v(P ),
then the solution cannot be profit-maximizing because the monopolist can
then increase the entry fee, and thereby profits, while ensuring that the
consumer will continue to buy the good.

The monopolist’s problem can now be written as:

max e + (P − c)Q(P )
e,P

subject to the individual rationality constraint e = v(P ). We can rewrite

this problem as the following unconstrained maximization problem:

max v(P ) + (P − c)Q(P )

P

Differentiating with respect to price, the first order condition is:

v0 (P ) + Q(P ) + (P − c)Q0 (P ) = 0 (15)

and since Q(P ) = −v 0 (P ), the first order condition becomes:

(P − c)Q0 (P ) = 0 (16)

It is assumed that the second order condition is satisfied, i.e. Q0 (P ) + (P −

c)Q00 (P ) < 0.

Note that Q0 (P ) = −v 00 (P ) < 0. Therefore from (16):

P∗ = c (17)

The monopoly firm prices at marginal cost. Substituting this price in the
individual rationality constraint:

e∗ = v(P ∗ ) = v(c) (18)

26
The entry fee is equal to the consumer surplus corresponding to the price
P ∗ = c. Since there are no deadweight losses, the move from a uniform price
monopoly to a two-part tariff (with identical consumers) is unambiguously
welfare-enhancing.

References
[1] L. Phlips (1983). The Economics of Price Discrimination. Cambridge:
Cambridge University Press.

[2] J. Tirole (1988). Theory of Industrial Organization. Cambridge: MIT

Press. (Chapter 3).

[3] H. Varian (1989). Price Discrimination in Handbook of Industrial Orga-

nization, Volume 1, Edited by Schmalansee and Willig. Elsevier Science
Publications. (Chapter 10).

27