Columbia University

Graduate School of Business

Managerial Economics (B6006)

Classes 9-14

Industry Analysis
Imperfect Competition in the Short Run

I. Introduction.
In this section of the course we study imperfectly competitive industries. This term covers a wide
variety of markets that we describe later on. We start by listing the main features of such
markets:

1. The product is differentiated.

2. Firms have market power.

Points 1 and 2 are closely related. Product differentiation is the source of market power. What we
mean by “market power” is that the firm has some control over the price. In other words,
consumers DO NOT view competing products as perfect substitutes (for example, Coke and
Pepsi in the market for soft drinks, or different brands of cigarettes). However, it should be
noted that different products/brands within each of these market have to be close substitutes. If
there is a good/brand for which there are no substitutes, or all substitutes are poor substitutes, we
might want to view this market as a monopoly, i.e., a market with only one seller.
Market power implies that if the firm raised the price of its product, some customers will keep
buying the product at a higher price. This is in contrast with the perfectly competitive industry
(homogeneous product), where consumers will never pay a price above the market price, because
they do not differentiate between homogeneous products sold by different firms.

It should be pointed out that there are other sources of market power in addition to product
differentiation. For example, there are markets where, although the product is homogeneous
(e.g., oil), firms have market power. In such markets the source of power is frequently the
presence of large barriers to entry. This type of industry is called homogeneous oligopoly (few
sellers). We will study such industries in the next section of the course when we discuss strategic
interaction (game theory).

The number of firms operating in imperfect competition can vary dramatically and is likely,
therefore, to alter the nature of the analysis. It can vary from thousands (e.g., restaurants in Paris)
to less than twenty (e.g., the market for sport utility vehicles (SUV) in the US), or even to one
firm (e.g., De Beers in the world diamond market).
In order to keep the analysis simple and keep the number of competitors out of the discussion,

1
we will assume, for now, the following:
1. The number of competitors is given and,
2. Competitors do not change their choices (e.g., prices, quantities, marketing, etc.)

II. Consumers and the Demand Faced by the Individual Firm.

We will keep the buyers’ side of the story as simple as possible. We, therefore, limit the analysis
to industries where:
1. There are many small and price taking buyers, and
2. Search costs are negligible.

As we already mentioned we define imperfect competition as a market in which the individual

firm/brand is facing a downward sloping demand curve. This is a new concept in the course that
requires some analysis.
First, remember what we had in perfectly competitive markets. We only talked about market
demand, and didn’t discuss the demand faced by the individual firm. Why?
In perfect competition the individual firm has NO market power and is, therefore, a price taker.
In other words, it faces a perfectly horizontal demand curve at the market price, as shown in the
graph below:

Demand faced by individual firm in a competitive market

P* D

In contrast, in imperfect competition the firm is facing a downward sloping demand curve, as
shown in the graph below:

2
 P
Demand faced by individual firm imperfect competition

P1

P2

P3

Q
Q1 Q2 Q3
It is clear that one of the factors that will determine the shape of the demand curve is the
behavior of competitors. Assume, for example, that the above demand curve represents the
monthly demand for Coke in the US. It is very likely that what Pepsi does (change prices,
conduct marketing campaign, etc.) will affect the demand for Coke. However, as already
mentioned above, in our short run-analysis we exclude this possibility. We assume that the
firm/brand has a given, fixed, consumer base. For that reason our short run analysis will be
identical for all firms facing a downward sloping demand curve, regardless of the number of
competitors, including zero competitors. Thus, the analysis will be identical to that presented in
standard economic textbooks under the title “monopoly analysis”.

Notice that while in the competitive market the price was given and the firm had only to choose
the quantity that will maximize its profits, here the situation is different. As can be seen in the
graph above, the firm has to choose a point on the demand curve. Whether it chooses a quantity
or a price doesn’t matter, because one implies the other.

III. Calculating Revenues and Marginal Revenues.

Reviewing Perfect Competition

Before solving the profit maximization problem of the firm, it is worth spending some time on
understanding how revenues and marginal revenues are calculated in imperfect competition.

First, recall what we had in perfect competition. The firm faced the market price, P*. Its revenues
were simply P* Q (market price multiplied by the quantity). What was the marginal revenue
(MR)? In other words, what was the change in revenues associated with a change in sales of one
unit? It was…

3
Constant!!! Regardless of the level of output, if you sold one more unit, your change in revenues
would simply equal the market price. Therefore, in perfectly competitive markets, the total
revenue and marginal revenue graphs look like the figures below:

P P
Perfect Competition, Total Revenues

Perfect Competition, Marginal Revenue

TR

P* MR
+P*

+1

Q Q

In terms of equations:
TR = P*·Q
AR = (P*·Q)/Q = P*
MR = ∂(TR)/∂Q = ∂(P*·Q)/∂Q =P*
(here ‘∂’ means ‘partial derivative’, which is nothing but the derivative of a function with two or
more arguments with respect to only one of them, treating all other arguments as constants).

IV. Marginal Revenue and the Cost of increasing Sales in Imperfect Competition.
In the above example, of perfect competition, the firm could increase sales at no cost. By
definition, it could sell as many units as it wishes, at the market price. The marginal revenue was
constant and equal to the market price, regardless of the level of output.
This is not the case in imperfect competition, where the firm is facing a downward sloping
demand curve. Increasing sales has a cost. We will illustrate this point with a very simple
example:

You run a day-care center in Morningside Heights. You currently have 10 kids, each paying
$800 per month. Since you are below capacity, you consider increasing the number of kids by
reducing tuition. Following some research, you conclude that you face the following demand for
your day-care center:

4
 P

800

700

Q
10 11
Thus, in order to increase the demand from 10 to 11 you will have to reduce tuition by $100 per
month, from $800 to $700.
We can breakdown the change in revenues associated with increasing the number of kids by one
into two components.
1. Increase in revenues by having one more kid (1·700).
2. Decrease in revenues associated with the reduction in tuition for your existing customers
($100·10).
This change in revenue of $1000 represents the cost of generating this one additional sale. We
will give a more formal definition of this term in the next example.

Notice that we exclude the possibility of charging different prices for different parents (price
discrimination). We will discuss price discrimination separately and extensively later on.

More generally:

ΔR = P + QΔP = (price of extra unit) + (# of units previously sold*price change)

In the above example P = 700, QΔP = (10*(-100)) = -1000, and therefore, ΔR = -300. The cost of
increasing the number of kids from 10 to 11 is greater than the revenues generated by the
additional consumers, thus the net MR is negative.
The cost of sales is the difference between the price and the change in revenues. It is therefore
equal to -QΔP. Hence, the cost of sale is the reduction in revenues generated by the price break
needed in order to make an extra sale.

"Formal" Math (notice that while in a competitive market P is constant, here it is a function of

5
Q):

TR = Q ⋅ P(Q)
∂ TR ∂P
= P+ ⋅Q
∂Q ∂Q

Notice that the marginal revenue is lower than the price because the derivative of price with
respect to quantity is negative (quantity increases price decreases).

A Numerical Example (taken from the textbook).

Demand
P Q TR MR AR
6 0 0
5 1 5 5 5
4 2 8 3 4
3 3 9 1 3
2 4 8 -1 2
1 5 5 -3 1

From the above data we can plot the demand curve and the MR curve

Dollars per
Unit of 6
Output

Average Revenue
3 (Demand)

Marginal
2 Revenue

0
1 2 3 4 5 6 Output

Demand: Q = 6 - P or P = 6 - Q
TR = P*Q = 6Q - Q2
MR = 6 - 2Q

6
At each level of output the vertical distance between the price and MR represents the cost of
generating an additional sale: Cost of sale = (p - MR)

V. The General Case of a Linear Demand Curve

The general form of any linear demand curve is:

(1) Q = A -Bp; (demand curve)

where, both A and B are strictly positive numbers. A is the maximum potential volume of sales.
It is achieved when the price is equal to 0. Since sales cannot be negative, they will equal to zero
for any price greater or equal than (A/B).

Total revenues are equal to price times sales. In order to have an expression of the total revenues
that depends only on the level of sales, we solve (1) for p by getting:

(2) p = A/B - Q/B (inverse demand curve: what we generally draw)

Substitute (2) in the total revenue expression and get:

R(Q) = pQ = (A/B -Q/B)Q = (A/B)Q - Q2/B

The MR curve is the derivative with respect to Q of the total revenue curve. It is, therefore:

MR(Q) = A/B - 2Q/B.

The picture below illustrates graphically the construction of the MR-curve. The marginal
revenue curve is a downward sloping line. Both the demand and the MR curves intersect at the
price (or the “y” axis) at p = A/B. However, the demand curve intersects the quantity (or the x)
axis at Q = A, while the MR- curve intersects it half way, at Q = A/2. In the output range 0< Q<
A/2, the marginal revenue is positive, while in the range A /2 <Q<A, it is negative.

7
 Cost/Unit

A/B

Demand Curve

MR Curve

Q
A/2 A

VI. Profit Maximization

We can now move to the final step in our analysis, calculating the profit maximizing output and
price. The logic that we will follow is identical to the one we developed in the case of
competitive markets and is based on the fundamentals of cost-benefit analysis.
Facing a given level of prices and output, we consider the decision to increase production, for
example, by one unit.
The rules are very simple:
1. If change in revenues is greater than change in costs, then go ahead and expand. In other
words, if MR>MC increase output.
2. If change in revenues is lower than change in costs, then don’t expand and consider
reducing output. In other words, if MR<MC reduce output.

Given the above two rules, it can be seen that if either MR>MC or MR<MC the firm can
increase profits by changing output and prices. Therefore, the firm will maximize profit only
when:

MR = MC

We will work out the profit maximizing choices of a price setter firm in two different ways,
graphically and analytically with numerical examples.

8
Graphical Analysis
Consider the graph below:
Cost/Unit

A
P1 MC

P*
P2 B

Q
Q1 Q* Q2 D
MR

Let’s assume that the firm is operating at an output level of Q1, charging a price of P1 (point A).
Notice that at this level of operation the benefit of increasing output by one unit is greater than
the cost (MR>MC). Therefore, Q1 is not a profit maximizing choice. Higher level of output will
increase profits. The opposite is true for point B, where the firm charges a price of P1 and sells a
quantity of Q2. On the last unit sold the firm loses money because MR<MC. The difference
between MR and MC is the increase in profit that the firm could have by reducing its level of
production by one unit. Following this logic the only profit maximizing level of output in the
above graph is point C, where the firm is operating at an output level of Q*, charging a price of
P*. Notice that the profit-maximizing price is at the demand curve!! This point will become
clearer after you go over the examples below.
As an exercise at home, try to highlight the areas that represent the increased in profits associated
with moving from point A to point C and moving from point B to point C (if not already done in
class or read in the textbook).

Hence, the profit maximizing choice of sales, Q*, is such that MR(Q*) = MC(Q*)

Analytical Analysis
We start the analysis with two pieces of information: The demand curve faced by the firm and its
cost structure. From its cost structure we derive the marginal cost and average cost curves as
shown in the graph below (we didn’t introduce the AC curve so far because it is not required in
order to find the profit maximizing price and output). The next step is to derive the marginal
revenue curve and to find its intersection with the marginal cost. This intersection determines the

9
profit maximizing output Q*. Now, the problem is to find the right price, i.e., the price at which
the firm’s customers are willing to buy exactly Q* units of output. This is readily found by
substituting Q* in the demand curve, thereby obtaining P*.
Last, we would like to have a graphical idea of the profits realized by our firm.
Remember that profits are equal to sales times the difference between sale price and average cost
of production. They are, therefore, the area of the shaded rectangle in the picture below. The base
of the rectangle is Q*, while its height P*-AC(Q*).
Cost/Unit

MC
C
P*
AC

AC*

Q
Q* D
MR

Examples of the Monopoly Problem:

(As we mentioned earlier, in our short run analysis we treat all firms that face a downward
sloping demand curve identically, weather it’s a monopoly or any other market structure. We use
the term “monopoly problem” because it’s the common name used for this analysis).

A firm is facing a decreasing demand curve for its product. The problem of the firm is to select
the best point on its demand curve, i.e., a price-quantity pair that maximizes its economic profits.
It is evident that since each point on the demand curve specifies both quantity and price, we can
think about the profit maximization either in terms of selecting a profit maximizing output or in
terms of selecting a profit maximizing price. We find it easier (given what we have done in
perfect competition) to think in terms of quantity choice.
The firm has knowledge of the following data:
I) Demand curve for its own product,
II) Cost curve indicating the economic costs of producing any given level of output.

In order to find the profit maximizing output, QM, and the associated price, PM, we go through
the following steps:

10
Step 1: Compute the Marginal Revenue curve (MR) from the demand curve and the Marginal
Cost curve (MC) from the cost curve.

Step 2: Compute the profit maximizing output, QM, by solving the equation MR(QM) = MC(QM).

Step 3: Compute PM, the “right price” for QM (i.e., the price at which buyers are willing to buy
QM units of output) by substituting QM in the demand curve.

The next two examples illustrate this procedure. In the third example, the firm is facing a
production capacity constraint, i.e., it cannot produce more than a certain amount. The presence
of a capacity constraint slightly modifies step 3.

Example 1:
Data:
(1) Q = 100 - p; (demand curve for the firm’s output)
(2) C(Q) = 1,000 + 20Q; (cost curve)

Step 1: Compute the MR and MC curves

We start by computing the marginal revenue curve. The marginal revenue is the increase in
revenues generated by the sale (the production) of an extra unit of output. If we think about
infinitesimal units, the marginal revenue can be thought as the derivative with respect to Q of the
total revenue, TR.
Evidently, the total revenue is:

TR(Q) = pQ.

However, in order to compute the marginal revenue we need to express the total revenues just in
terms of output and take the derivative with respect to Q. Hence, in order to do that we go
through the following steps:

i) Express the price in terms of output.

Solve the demand curve, equation (1), in terms of price. From (1), we get

(3) p = 100 - Q.

Equation (3) is called the “inverse demand curve”, since it expresses prices in terms of quantity
(while the demand curve expresses quantities in terms of prices).

ii) Express the TR curve in terms of output

Substitute (3) in the total revenue expression and get

R(Q) = pQ = (100 - Q)*Q = 100Q - Q2

iii) Compute the Marginal Revenue

11
Just take the derivative of TR with respect to Q and get:

(4) MR(Q) = dR(Q)/dQ = 100 -2Q

MC - curve
As we already know the MC-curve is just the derivative with respect to Q of the cost curve,
equation (2). Hence:

(5) MC(Q) = dC(Q)/dQ = 20

Step 2: Compute the profit maximizing output QM

Equate the marginal revenue curve, equation (4), to the marginal cost curve, equation (5), and
get:

100 - 2Q = 20 or QM = 40

Step 3 :Compute pM

Substitute QM = 40 into the demand curve, equation (2), and get:

40 = 100 - p or pM = 60
At this point we can easily compute the optimal economic profits realized by the firm:

profits = pMQM - C(QM) = 40*60 - 1,000 - 20*40 = 600.

The picture below illustrates graphically the determination of pM, QM.

12
 Cost/Unit

100

M
60

MC
20

Demand Curve

MR
Q
40 100

Example 2:
In Example 2, the firm is facing the same demand curve of example 1, but a more complex cost
structure.

Data:
(1) Q = 100 - p; (demand curve for firm output)
(2’) C(Q) = 100 + 20Q +Q2; (cost curve)

Step 1: Compute the MR and MC curves

Since the demand curve is identical to the one in example 1, the MR curve is expressed by
equation (4) of the previous example. In order to compute the MC-curve, take the derivative with
respect to Q of the cost curve, equation (2’). Hence:

(5’) MC(Q) = dC(Q)/dQ = 20 + 2Q

Step 2: Compute the profit maximizing output QM

Equate the MR, equation (4) to the MC, equation (5’), and get:

100 - 2Q = 20 + 2Q or QM = 20

Step 3: Compute pM
Substitute QM = 20 into the demand curve, equation (1), and get:

20 = 100 - p or pM = 80

13
At this point we can easily compute the economic profits realized by this firm

profits = pMQM - C(QM) = 80*20 - 100 - 20*20 - (20)2 = 700.

The picture below illustrates graphically the determination of pM, QM.

Cost/Unit

100

MC

M
80

20
Demand Curve

MR
20 50 Q
100

Example 3: Capacity Constraint

In this example we use the same cost and demand functions of example 1. However we introduce
a capacity constraint. Remember that in example 1, the profit maximizing output is QM = 40.

Will the profit maximizing output change if the firm has a productive capacity greater than 40?
The answer is obviously no. Without capacity constraints the firm, by producing QM, realizes
profits higher than profits associated to any other output level. Since, with a capacity constraint
higher than 40, the firm can still achieve maximum profits by producing QM = 40, it will keep
doing so.

Suppose now that the capacity constraint is less than QM = 40, for instance it is 30. What will the
profit maximizing output be in this case? Remember that it is worth producing each additional
unit for which MR > MC. Since both demand and cost curves are identical to example 1, the MR
and MC curves are the ones indicated in equations (4) and (5) of example 1. Hence, the
difference between MR and MC is given by:

MR(Q) - MC(Q) = 100 - 2Q - 20 = 80 - 2Q.

14
Observe that MR(Q) > MC(Q) for each output 0 < Q < 40. Hence, if the firm finds it worthwhile
to produce, it will produce at capacity, i.e., Q’ = 30. By substituting Q’ = 30, in the demand
curve, we get

30 = 100 - p or p’ = 70.

The profits associated to Q' = 30 and p' = 70 are:

Profits = p’Q’ - C(Q’) = 70*30 - 1,000 - 20*30 = 500.

Finally, observe that since profits are positive the firm will find it worthwhile to produce. How
would you have changed your answer if the profits were negative?

A Few Additional Points for Discussion

Below are a few topics that can be discussed and analyzed using our monopoly framework. It is
strongly suggested that if these issues were not already discussed in class you should not read the
answers/discussions before spending sometime on it by yourself. You can view these points
below as testing your depth of understanding the material

1. Maximizing revenues v. maximizing profits.

Managers are frequently motivated to maximize sales (revenues) rather than profits.
Under what circumstances maximizing profits and maximizing revenues will be the
same?

2. Negative MR?
In the graphs that represent the monopoly analysis the MR curve gets negative beyond a
certain level of output. What does negative MR mean?

3. Do you always fill the theater?

You are the general manager of a theater. You realize that almost always your theater is
half empty. Having studied some economics, you come up with the following line of
reasoning: “At the current prices that I charge the quantity demanded is less than the
supply. Given that there is excess supply it must be the case that I charge a price that is
above equilibrium. I should, therefore, reduce my prices.”
What’s wrong with this line of reasoning?

15
VII. Elasticity.
In many obvious circumstances, it is important for a price-setting firm to analyze the effect on
sales of a price change. However, if we are looking for a precise quantitative answer, the units
used to measure either output or price could blur the analysis. If we say that the price changed by
“one”, is it a change of $1, €1, or 1 yen? Is the change in output measured in gallons, liters tons,
cans? A major multinational operating in countries with different currencies and different unit of
measure needs an answer (that expresses the responsiveness of sales to price change)
independent of the units of measurement. Thus we are naturally led to the notion of price
elasticity.

The firm, facing a demand curve, is currently operating at a given point on the curve,
representing a certain level of prices, p, and sales, Q. We ask the following question:
What is the percentage change in quantity generated a 1% increase in price?
The answer to this question is the price elasticity of demand at Q, which we will denote by
Ep(Q). Since Ep(Q) is a ratio of percentage changes, its value is independent of the units used to
measure sales and prices. Writing down the definition of Ep(Q) in terms of equation gives:

%ΔQ
E p (Q) = (percentage change in quantity divided by percentage change in price).
%ΔP

Some Algebra:

ΔQ
⋅100
%ΔQ Q ΔQ P
E p (Q) = = = ⋅
%ΔP ΔP ⋅100 ΔP Q
P
Thus, Ep(Q) equal, for instance, to –2 means that if we increase the price by 1% sales would
contract by 2%.
There are many different ways used to compute the elasticity demand depending on how the
terms ΔQ and ΔP are measured. Here, we are mostly interested in the conceptual meaning of the
price elasticity and, therefore, we simply identify ΔQ/ΔP with the derivative of the sales with
respect to price. Thus, we use the following notion of price elasticity:
∂Q P
Ep = ⋅
∂P Q

Since by using calculus we consider infinitesimal changes, elasticity is now defined as the
percentage change in sales generated by an infinitesimal percentage change in prices.

∂Q
Notice that since the demand curve is decreasing, i.e., < 0 , price elasticity is always a
∂P
negative number.

16
The following is a conventional terminology used by all technical publications and textbooks. A
demand curve at a sale level Q is called:

i) elastic, if Ep(Q)<-1 (e.g., -2, -200 so on)

ii) inelastic if Ep(Q)>-1 (e.g., -(1/2), -.10)
iii) unit elastic if Ep(Q)=-1.

To better understand the meaning of i) and ii), we look at two extreme cases, a totally inelastic
(or 0-elastic) demand curve and a totally elastic (or infinitely elastic) demand curve.
A demand is 0-elastic if a price change does not induce any changes in sales. Thus a 0-elastic
demand is a vertical line (water in the desert). An infinitely elastic demand curve is a horizontal
line (at some price level, p). If prices are increased above p, the demand vanishes, while if they
are decreased, sales explode to infinity.

Thus, as a first approximation, we get the idea that if the demand is inelastic at Q, the demand
curve is very steep around Q. Conversely, if it is very elastic then it is fairly flat around Q. This
is just an intuitive, first approximation and by far not precise view of elasticity.

An Example
Suppose that we are facing, as in most problem sets, a linear demand curve
(1) Q=A-Bp.
Then, ∂Q/∂p=-B and thus
(2) Ep(Q)=-B(p/Q)
It is convenient to rewrite p by solving (1) in terms of prices, i.e.,
(3) P=(A/B)-(1/B)Q.
Substituting (3) into (2), we get
(4) Ep(Q)=- B[(A/B)-(1/B)Q]/Q=-(A/Q)+1.

Observe, that although the demand curve has constant slope, Ep(Q) is close to zero when Q is
close to the maximum number of sales A, while it is a very large negative number when sales are
very low and close to zero. Finally, the elasticity is equal to –1 when sales are A/2. Remember
that in correspondence of those sales the marginal revenue is equal to zero. If it sounds
confusing, don’t worry. We will come back to this point later on.

More intuition:
Why is it the case that although we are dealing with a linear demand curve the price elasticity is
different at different points on the line?
Consider again the definition of price elasticity:
∂Q P
Ep = ⋅
∂P Q
Now, notice that the first term, ∂Q/∂P, is equals to 1/slope of the demand curve and is constant,
because the demand curve is linear (in the above example the value of this constant is –B).
However, the term P/Q is different at different points on the curve. It’s very large when Q is
small and gets smaller when Q increases. When output is relatively small, a one unit increase is a

17
relatively large in percentage, while the same one unit increase in output is relatively minor when
the level of output is large).

Elasticity, Revenue Changes (MR), and the Cost of Sale.

This section contains some technical parts. If you find these parts confusing, don’t worry; just
skip them.

There is a close analytical relationship between the concepts of marginal revenues, price
elasticity, and cost of sales. We will first build the intuition and then provide the more technical
analysis.

Let’s start with a knife-edge case by supposing that the price elasticity is –1. At this point, what
is the sign of the marginal revenue? In other words, if we increased output, what would happen
to revenues? Increase? Decrease?
The answer is pretty simple. We will experience no increase at all in revenues or, equivalently,
the marginal revenue is zero. If we want to increase sales, we have to decrease prices. However,
by definition, since the price elasticity is –1, the price will have to drop exactly by the same
percentage that sales will increase. Thus, the sales increase exactly compensates for the price
decrease and total revenues stay constant.

If, however, the firm operates at a point where the demand is elastic, i.e., Ep(Q)<-1., the
marginal revenue is positive, while the opposite is true if the demand is inelastic. Suppose that
we are told that Ep(Q)=–2 (or any number below –1). An elasticity equal to –2 implies that a 1%
decrease in price is accompanied by 2% increase in sales. If we want to increase sales, we have
to decrease prices. However, prices have to drop in percentage terms less (half if Ep(Q)=-2) than
the percentage increase in demand. Thus the increase in demand (quantity) more than
compensate for the decrease in prices and total revenues increase, i.e., MR>0. Therefore, the
more elastic the demand, the bigger is the marginal revenue. Remember that in a competitive
market, when the firm is facing a perfectly horizontal demand curve at the market price (and
therefore price elasticity is -∞), marginal revenue equals the price!!!
Similar, but opposite argument applies for inelastic demand curves.

Formal Analysis.
We start with the definition of MR (marginal revenue is the derivative of total revenue with
respect to output):
∂ (TR) ∂ ( P ⋅ Q)
MR = =
∂Q ∂Q
Now, in order to understand the next steps, you have to remember the following: P is a function
of Q, and, therefore, in order to take the above derivative you have to use the product rule for
differentiation which is: (f*g)’=f’g+g’f.

∂ ( P ⋅ Q) ∂P ∂Q ∂P ⎡Q ∂ P ⎤
MR = =Q +P =Q + P = P⎢ ⋅ + 1⎥
∂Q ∂Q ∂Q ∂Q ⎣P ∂Q ⎦

18
Notice that the first term in the squared brackets is the inverse of price elasticity. Therefore, the
marginal revenue expression can be rewritten as:

⎡ 1 ⎤ P
MR = P ⎢ + 1⎥ = P +
⎣ EP ⎦ EP

Again, remember that since price elasticity is negative, the marginal revenue is lower than the
price. Also, since we defined the cost of sale as p-MR, we can rewrite the cost of sale as:
P
Cost of sale= −
EP
The examples below illustrate the link between price elasticity and marginal revenues. As you
can see, the more elastic is the demand, the closer is the MR to the price.

Examples:
Ep =-10 MR=p+p/(-10)=.90p MR=90% of the price
Ep =-2 MR=p+p/(-2)=.50p MR=50% of the price
Ep =-1 MR=p-p MR=0
⎛ 1 ⎞
Ep =-1/2 MR= P⎜ 1 − ⎟ = −P
⎝ 1 / 2⎠
Exercise:
It has been suggested that managers should always follow the following rule:
If elasticity is less than 1 Raise your Price!!!
First, show why this rule is correct. Second, is the opposite true?
The answer is NO. Try to figure out why.
Third, research has shown that the demand for cigarettes is inelastic. If this is correct, doesn’t it
imply that cigarette companies violate the above rule and, therefore, do not maximize profits?
Fourth, try to show how this rule is perfectly consistent with our analysis of monopoly profit
maximizing price and output.

Additional points for discussion:

Ë What makes the elasticity of demand for a product more/less elastic?
Ë What is the difference between the elasticity of demand faced by the market and that
faced by the firm?
Ë There are other elasticities to consider: Income elasticity, Cross price elasticity, etc.

19
 Industry Analysis:
Pricing

I. Price Discrimination

The basic idea of price discrimination is to charge different prices to different customers for the
same (or very similar) product or service. Other variations of price discrimination include
charging different prices at different time periods, or charging different price per unit, depending
on the number of units bought.
In order to motivate our discussion of price discrimination, we first illustrate the potential for
increasing profits beyond those available by charging a single price.

Consider the graph below:

The firm is currently maximizing profits by charging a single price P*. Notice, however, that
there are two sources of additional profits that are not exploited by the firm.
1. There are current customers that buy the product at P* but are willing to pay a higher
price for it. The green segment of the demand curve represents these customers.
2. There are potential customers that currently do not buy the product because the
current price, P*, is too high for them. The price they are willing to pay, although
lower than P* is still above the marginal cost of selling this additional unit. The
yellow segment on the demand curve represents these customers.
Cost/Unit

MC

P*

Q
Q* D
MR

“Price Discrimination” covers a wide variety of pricing strategies aimed at increasing the firm’s
profit beyond and above those generated by charging a single price. Due to time constraint we
will cover only a limited set of strategies.

20
Price Discrimination and the Cost of Sale
In the example above, Q* is the profit maximizing level of sales. Why doesn’t the firm increase
its sales beyond Q* if there are additional customers who are willing to pay a price the marginal
cost of production?
The answer is that the cost associated with generating this additional sale was too high. Price
discrimination can be viewed as a way to reduce the cost of sale, thus allowing the firm to
increase sales. Remember that the cost of sale was generated by the price-break given to the
“old” customers. If, however, the firm could charge a lower, new price to the additional
customers while, at the same time, maintaining the old price for the old customers, than it avoid
the “bite” of the cost of sale.
Another obvious way to reduce the cost of increasing sales is through advertising. Advertising
campaigns attract new customers without requiring price cuts. However, advertising campaigns
are also costly. We will not investigate this topic. It suffices to say that if the cost of an extra
euro of advertising generates an increase in sales above $1, it is “cost-effective” to increase
advertising. Using the now familiar marginal analysis, we are led once again to the conclusion
that the optimal level of advertising is such that the marginal cost of advertising is equal to
marginal revenue it generates.

II. Conditions for Price Discrimination

The problem is to find a way to charge different prices to different people without knowing their
personal identities. Clearly, the latter implies that the first step in price discrimination is to
“separate” customers into groups differentiated by their willingness to pay. In order to have a
successful separation, two features have to hold true:
i) The cost for consumers to move from the high paying group to the low paying group
must be higher than the gain generated by the price change. Otherwise all customers will
move to the group facing a lower price Think of examples (there are numerous).
ii) Resale must be very difficult and extremely limited. Otherwise, the market will open to
arbitrage opportunities and those who can buy at a lower price will re-sell to those who
are charged a higher price. Again, think of examples.

There are two fundamental ways to create separation and market discrimination. The first and
simplest is to separate on the basis of observable characteristics (e.g., sex, geography, age and so
on) correlated with the willingness to pay. If, for example, the willingness to pay for a theater
show is lower among students (because they are poorer), then offering a student discount will be
a form of price discrimination where the student ID is the observed characteristic.

The second is to create “mechanisms” inducing a self-selection of people into groups with a
different willingness to pay. Classical examples of such mechanisms include: coupons,
discounted outlets in impossible locations, books sold in hard cover and paperback, discounted
airline tickets for Saturday stay over, shopping on Priceline.com, MBA vs. EMBA, volume
discount, and bundling.

As an exercise you should think of each of the above examples and try to figure out, in each
case, which are the specific groups that are being separated by these mechanisms, why one is
willing to pay more than the other, and what makes individuals choose to belong to the group

21
paying a higher price (Ignore the case of bundling. We will discuss this topic separately).

III. First Degree Price Discrimination

Our first model is the benchmark model of price discrimination. While it’s very unrealistic in its
pure form, it is a very useful model to start with.
Let’s think about a firm facing 10 customers. Each customer is willing to buy for the right price
one unit of the product supplied by our firm. Order the customer according to their reservation
prices so that customer number 1 is the customer willing to pay the highest reservation price,
customer number 2 the second highest and so on. In the picture below, the prices P1,…, P10 are
the reservation prices of the customers. So, for instance, the third reservation price P3 is the
maximum price that this customer is willing to pay in order to get the unit of the good. If the
price is even one cent above P3, this customer will not buy, while, quite obviously, if it is below
P3 the customer will buy the unit.

P Demand by 10 consumers
P1

P2
P3
P4

P5
P6
P7
P8

P9
P10
Q
1 2 3 4 5 6 7 8 9 10
Suppose now that our firm has already 5 units available for sale. In a world with uniform pricing,
in order to maximize the revenues generated by the sale of 5 units the firm charges P5. The total
revenues generated by this strategy are 5P5, i.e., the area of the (green) shaded rectangle in the
picture below. In some sense, customers 1 to 4 receive a “gift” from the market. They would be
willing to pay more, but in virtue of the price uniformity they just pay P5. The aggregate gift is
called consumer surplus. It is equal to the sum of the individual gifts, i.e., (P1- P5)+( P2- P5)+( P3-
P5)+( P4- P5), and is represented by the light shaded areas in the graph below. This is the amount
of dollars saved by the buyers over their reservation prices or, equivalently, the difference
between the maximum amount of money that buyers would be willing to spend to get the 5 units
(i.e., P1+P2+ P3+P4+ P5) minus what they actually pay (i.e., 5P5).

22
 P Demand by 10 consumers
P1 Single Price Revenues

P2
P3
P4

P5
P6
P7
P8

P9
P10
Q
1 2 3 4 5 6 7 8 9 10
If, however, the firm can perfectly price discriminate the situation is quite different. Each
customer will be charged her own reservation price. The total revenues generated by the sales of
the five units is equal to the maximum amount of money that buyers would be willing to spend to
get the 5 units, i.e., P1+P2+ P3+P4+ P5. In a sense, by perfectly price discriminate, the firm
transforms the entire consumer surplus into revenues. Indeed, with perfect price discrimination
the consumer surplus is zero.

In order to generalize the analysis, let us think now about a firm facing a smooth and downward
sloping demand curve. Although not necessary, it is helpful to think that each point on the
demand curve represents the reservation price of an individual for the (infinitesimal) unit of the
good. In the graph below we add only the marginal cost curve.

The most important point in this figure is that now the demand curve is also the MR curve. Why?
As the firm increases sales, “old” customers keep paying the same price. “New” customers pay
their reservation price, so the price that they pay is equal to the marginal revenue. The marginal
revenue from each additional sale is decreasing, following the demand curve.

Profits are maximized when the level of output is Q*. Beyond this level of output the cost of
selling an additional unit (MC) exceeds the price (MR) charged from the additional customer. At
Q*, the last unit sold at a point where MR=MC.

23
 P
First Degree Price Discrimination

MC

D=MR

Q
Q*
IV. Second Degree Price Discrimination
Consider the demand curve of an individual consumer for a given good or service, for example,
electricity consumption. It is reasonable to assume that the demand curve will be downward
sloping, as we discussed in an earlier class. In the case of electricity the consumer is willing to
pay a relatively higher price for the electricity that support that most essential needs, such as the
TV, Playstation, etc. Having covered the necessities, her willingness to pay for more electricity
(e.g., for refrigerator, lights, etc.) will be lowera.

Second degree price discrimination is built on the idea that individuals’ demand curves are
downward sloping, i.e., willingness to pay is decreasing with the number of units already
purchased. The firm is charging a different price depending on whether it’s the first, second, or
third unit purchased. While there are many markets in which such a practice is unfeasible, there
are other markets in which it is easy to do. Electric utility sometimes practice it by charging
decreasing prices for the first, second, etc. blocks of electricity.

Volume discount is a form of second degree price discrimination. Why? Think about the clearer
example of “buy a second pair for half price”: You pay a high price for the first pair, but charged
a lower price for the second. But this is equivalent to giving a volume discount, you pay less per
unit when you by two pairs than when you buy one.

a
These examples do not necessarily reflect the preferences of most adults….

24
V. Third Degree Price Discrimination
We talk about third degree price discrimination to describe situations where the firm has
successfully separated customers in different groups and is facing, therefore, distinct demand
curves one for each group.
To fix ideas, say that a firm has successfully separated its customers into two groups. Customers
cannot resale the product or switch from one group to the other. The firm knows its own cost
structure and the demand curves of both groups. Total costs depend on the total production
(sales) and not on how total sales are divided among groups.
While the firm can price discriminate between the two groups, the price within each group is
unique. The firm has to decide on the level of output to be sold to each group and the prices to
be charged. The goal is, as usual, profit maximization.

In order to understand the logic behind third degree price discrimination, we divide the argument
in two parts by asking the following questions.
1. Assume that the firm has somehow decided on total level of sales, say, 10,000 units.
Since total output has already been decided we can ignore the production costs. How
should we allocate the total output between the two groups in order to maximize
revenues?
2. Given the answer to 1), i.e., once understood how to optimally allocate output among
groups, how should we decide the total level of production in order to maximize profits?

Hence, let us start with an arbitrary division of sales, by allocating for instance 5,000 units to
each group, and analyze the consequences to total revenues of switching one unit from the
second to the first group. Since we are making an extra sale to the first group, the change in
revenues is equal to the marginal revenue of the first group, MRI. Since we are decreasing sales
to the second group by one unit, the change in revenues is minus the marginal revenue of the
second group, - MRII. Thus, the change in total revenues generated by this maneuver is equal to
the difference in marginal revenues of the groups, i.e., MRI - MRII. Thus, we can derive the
following simple algorithm:
If MRI > MRII, increase sales to the first group (and decrease them to the second);
If MRI < MRII, increase sales to the second group (and increase them to the second);
Stop when MRI = MRII.
Hence, the answer to the first question is simple:
Given any arbitrary level of output, Q*, it is optimal to allocate Q* among the two groups so that
MRI(Q*I)= MRII(Q*II), with Q*I and Q*II denoting the sales to the first and the second group and
Q*I + Q*II =Q*.

Caveat: There is an exception, somewhat pathological, to the previous analysis. Remember that
the marginal revenue curve is decreasing. It could be the case that the MRII(0)<MRI(Q*), i.e.,
that the customers in the second group are willing to pay much less than the customer of the first
group for any conceivable allocations of the total sales among the two groups. Quite obviously,
in such a situation we may as well ignore the customers of the second group by selling them
nothing.

25
Now we move to the second question. Since the optimal strategy is to keep equal the marginal
revenues among groups, the change in revenues generated by the production of an extra unit is
unambiguously equal to the common value of the marginal revenues. It is irrelevant if the extra
(infinitesimal) unit is sold to the first or to the second group, their marginal revenues are equal.
Thus, in a third degree price discrimination scenario, the firm, in order to maximize profits,
selects an optimal aggregate level of output Q*, the sum of sales to the first group Q*I and to the
second group Q*II, that satisfy:

Q*I + Q*II =Q*and MRI(Q*I)= MRII =MC(Q*).

Once Q*I, Q*II and Q* have been chosen, prices to the different groups are easily found through
the group demand curves.
We present now two numerical examples of third degree price discrimination.

Example 1

Data
Consider a firm facing two segmented market with the following demand curves for its product:
1. Q1 = 100 - p1 (demand of customers in group 1)
2. Q2 = 120 - .5p2 (demand of customers in group 2)

And the following cost function:

3. TC = 2,000 + 20( Q1 + Q2) (total cost of production)

Step1: Find the marginal revenues curves.

First, solve the demand equations for prices and get, from 1:
4) p1 = 100- Q1 ,

and from 2
5) p2 = 240 - 2Q2 .

Second, by exploiting equation 4), compute total revenue curve for group 1 (TR1(Q1)), i.e., the
total revenue to the firm when Q1 units are sold to group 1:

TR1(Q1) = p1(Q1) = (100- Q1)Q1=100Q1 - (Q1)2

and, by exploiting equation 5), total revenue curve for group 2 (TR2(Q2)):

TR2(Q2) = p2(Q2) = (240 -2Q2)Q2 = 240Q2 -2(Q2)2 .

Marginal revenues are obtained by taking the derivative of the total revenue curve with respect to
output (or sales). Therefore, we get:

MR1(Q1) = 100 - 2Q1,

MR2(Q2) = 240 - 4Q2.

26
Step 2: Find the optimal sales to the first, Q1*, and to the second, Q2*, group.
To achieve profit maximization, the firm has to find sales to the first group, Q1*, and to the
second group, Q2*, that satisfy the following conditions:

MR1(Q1*) = MR2(Q2*) = MC(Q1* + Q2*),

in words, the marginal revenues from the first group computed at optimal sales Q1* is equal to
the marginal revenues from the second group computed at optimal sales Q2* and both marginal
revenues are equal to the marginal cost of total production computed at (Q1* + Q2*).

From the total cost expression (equation 3)), remembering that marginal cost is the derivative of
total cost with respect to total output Q1 + Q2, we get:

MC(Q1 + Q2) = 20.

Therefore, MR1(Q1*) = MC(Q1* + Q2*) translates into:

100 - 2Q1* = 20

which implies Q1* = 40.

while, MR2(Q2*) = MC(Q1* + Q2*), becomes

240 - 4Q2* = 20

which implies Q2* = 55.

Step 3: Find prices to be charged to the first, p1*, and to the second, p2*, group.
Optimal prices are easily found by substituting the volume of optimal sales into the demand
curves. So, for group 1, substitute, Q1* = 40 into equation 1) (or, equivalently, equation 4)) and
get

Q1* = 100 - p1* or 40 = 100 - p1*

which implies p1* = 60.

For group 2, substitute, Q2* = 55 into equation 2) (or, equivalently, equation 5)) and get:

Q2* =120 - .5p2* or 55 = 120 - p2*

which implies p2* = 130.

Total profits are easily computed by using optimal values of sales and prices:

27
A* = p1* Q1* + p2* Q2* - TC(Q1* + Q2* ) = 60*40 + 130*55- 2,000 - 20*95 = 5650.

Remark 1: Use this example to understand the ideas behind price discrimination: in order to
maximize profits, the firm is charging $60 to first group and $130 to the second for each unit of
the product. Although the price to the second group is more than twice the price to the first it
would be a mistake to conclude that the firm should increase sales to the second group. Why?
The key indicators are the marginal revenues. In order to maximize profits, the firm equates the
marginal revenues, not the prices, of the different groups. We can look at the problem from a
different point of view. We have defined “cost of sale” as the difference between price and
marginal revenue. In the previous example the marginal revenue of the first group computed at
Q1* = 40 is MR(Q1* )=100 - 2Q1* =20. Since p1* = 60, the cost of sale of the first group is p1* -
MR1(Q1* )= 60 - 20 = 40. For the second group, the marginal revenue computed at Q2* = 55 is
MR(Q2* ) = 240 - 4Q2* = 240 - 220 = 20, which makes the cost of sale of the second group equal
to p2* - MR2(Q2* ) = 130 - 20 = 110. Hence, the firm charges a higher price to the second group
because its cost of sale is higher.

Example 2

This example has not been presented in class. The only complication introduced with respect to
the first example is in the total cost:

TC = 2,000 + (Q1 +Q2 )2.

The marginal cost is the derivative of the total cost with respect to total output Q1+Q2 and it is
therefore equal to

MC = 2(Q1+Q2).

Therefore, contrary to the first example where MC=20, in this case MC depends on total sales.
This creates some (mild) complications in solving the problem.

Demands by both groups are identical to example 1. Therefore, we can skip the first step and go
directly to the second.

Step 2: Find the optimal sales to the first, Q1*, and to the second, Q2*, group.

To achieve profit maximization, the firm has to find sales to the first group, Q*1, and to the
second group, Q2*, that satisfy the following conditions:

MR1(Q1*) = MR2(Q2*) = MC(Q1* + Q2*),

Therefore, MR1(Q1*) = MC(Q1* + Q2*), translates into

28
100 - 2Q1* = 2(Q1* + Q2*),

or, equivalently,

6) Q1* = 25 - Q2*/2

while, MR2(Q2*) = MC(Q1* + Q2*), becomes

240 - 4Q2* = 2(Q1* + Q2*),

or, equivalently,

7) 6Q2* = 240 - 2Q1*.

Substitute equation 6 into equation 7 and get:

6Q2* =240 - (50 - Q2*)

or

Q2* = 190/5 = 38.

Now substitute Q2*=38 into equation 6 and get

Q1* =25-38/2 = 6.

Step 3: Find prices to be charged to the first, p1*, and to the second, p2*, group.
Optimal prices are easily found by substituting the volume of optimal sales into the demand
curves. So, for group 1, substitute, Q1* = 6 into equation 1) (or, equivalently, equation 4)) and get
Q1* = 100 - p1* or 4 = 100 - p1*

which implies p1* = 94.

For group 2, substitute, Q2* = 38 into equation 2) (or, equivalently, equation 5)) and get
Q2* = 120 - .5p2* or 38 = 120 - .5p2*

which implies p2* = 164.

Total profits are easily computed by using optimal values of sales and prices:

A* = p1* Q1* + p2* Q2* - TC(Q1* + Q2* ) = 94*6 + 164*38 - 2,000 - (38+6)2 = 6732.

VI. Bundling

29
Bundling refers to the practice of offering consumers several distinct products packaged
together, as a bundle. Pure bundling occurs when you cannot buy any of the packaged items
separately (to date, the New York Times does not sell separate sections of the newspaper; so,
even if all you wanted was to buy the sport section, you must also buy the obituaries). Mixed
bundling refers to the practice of offering consumers the choice between buying a bundle and
buying items separately (as an example, restaurants that offer dishes in a fixed menu and a la
carte).

Examples of Bundling
Stereo Components
Computers
Service Warranty
Restaurant Menus
Cable TV
Microsoft Office
Newspapers
Music CD
Magazine subscription (why?)
Vacation package

Our objective is to provide an intuitive explanation of when and why bundling and mixed
bundling enhance profits. A general analysis is technically very demanding and we will just
provide three simple examples. In all the examples, for the sake of simplicity, there is no cost of
production. Thus, profits coincide with revenues.

We start with two very simple examples (the first is taken from the best seller "Information
Rules"b). In the first, bundling increases profits; while in the second the opposite is true. In the
third example, we illustrate mixed bundling.

In the next two examples, we consider the two following strategies:

1. Sell the two products separately.

2. Sell them as a bundle.

Example 1

Word Spreadsheet
Processor

Mark 100 30
Noah 30 100
The numbers in the table are reservation prices. Thus 100 is the maximum price that Mark is
willing to pay in order to buy a WP. Evidently, the bundle contains one WP and one S. The

b
“Information Rules” by Carl Shapiro and Hal Varian, HBS Press, Boston 1999.

30
reservation price for the bundle is equal to the sum of the reservation prices of the two products.
Hence, Mark’s reservation price for the bundle is equal to 100+30 = 130

Strategy 1: We have to find the profit maximizing choice of the prices for the two products.
Given the reservation prices of Noah and Mark we just have to consider two choices: selling the
products at 30 or selling the products at 100. With the first choice, we sell two units of them
making a profit of 30*4 = 120. With the second, we just sell one unit of both products and we
make a profit of 100*2 = 200. Thus, the profit maximizing choice is to set the price of both
products equal to 100, thereby realizing a profit of 200.

Strategy 2: Both Mark and Noah are willing to pay 130 for the bundle. Hence, bundling
generates a profit of 130*2 = 260.

Hence, here bundling dominates selling the products separately. Why is bundling working?
Because Mark is willing to pay a lot for the Word processor (100) and much less for the
Spreadsheet (30), while for Noah it is exactly the other way around. Thus, Mark and Noah are
buying the bundle for opposite reasons. The first for the WP, the second for the S. If the
reservation prices are significantly negatively correlated (perfectly so, in the example) bundling
generates additional profits. Consumers, although paying the same price for the bundle, are price
discriminated on its components. If you ask Mark, after he bought the bundle, how much he paid
for WP he is going to indicate a price of 100, while Noah would answer to the same question
with a price of 30.

Thus bundling is the profit maximizing strategy.

Is it always the case? No. Let’s see.

Example 2

Word Spreadsheet
Processor

Mark 100 30
Noah 30 100
John 150 150

Strategy 1: Given the reservation prices of Noah, John and Mark we just have to consider three
choices: selling the products at 30, or at 100, or at 150. With the first choice, we sell three units
of each product making a profit of 30*6 = 180. With the second choice, we sell two units of each
product making a profit of 100*4 = 400. With the third choice we sell just one unit of each
product thereby realizing profits for 150*2 = 300. Hence, the profit maximizing choice with
strategy 1 is to set the prices of both products equal to 100. This strategy yields profits equal to
400.

31
Strategy 2: Both Mark and Noah are willing to pay (at most) 130 for the bundle, while John is
willing to pay (at most) 300. Evidently, the best bundling choice is to price the bundle at 130. At
this price everybody is willing to buy the bundle. Hence, bundling generates profits equal to 390.

Thus, in this second example pricing the products separately dominates bundling them. Why?
The difference is the presence of John. John is willing to pay considerably more for both
products. Thus, John breaks the negative correlation between the reservation prices, thereby
making bundling suboptimal.

We provide a last example where the best strategy is “mixed bundling”. Using mixed bundling
means that the products are sold both separately and also bundled together. Thus with mixed
bundling we have to provide three prices: one for the bundle and one for each commodity.

Example 3

Word Spreadsheet
Processor

Mark 100 50
Noah 50 100
John 110 10

We just analyze the profits generated by bundling and mixed bundling: it is trivial to verify that
only selling separately generates much lower profits.

Bundling: John is willing to pay (at most) 120 for the bundle, while the other two are willing to
pay (at most) 150. Hence, we just have to consider two prices: 120 and 150. At 120, we sell three
bundles realizing profits of 360. At 150 we sell two bundles realizing profits of 300. Thus the
profit maximizing bundling strategy is to price at 120. This strategy yield profits of 360.

Mixed Bundling: The problem is that John is an outlier. He is willing to pay more than anybody
else for WP and less than anybody else for S. With mixed bundling we can turn this feature to
our advantage. The idea is simple. We want Mark and Noah to buy the bundle, while we want
John to buy (and pay a lot for) S. Price the bundle at 150 and price WP at 110 and S at 110. The
prices of each product are above Mark’s and Noah’s reservation prices, while the price of the
bundle is equal to their reservation price (i.e., to the sum of their reservation prices for the two
products). Hence, with these prices, Mark and Noah will buy the bundle. However, the price of
the bundle and the price of S are above John’s reservation price, while the price of WP is equal
to John’s reservation price. Thus, John will buy one unit of WP. Summarizing, with the chosen
prices, we sell two bundles at 150 and one unit of WP at 110. Therefore, the profits generated by
mixed bundling are equal to 150*2+110 = 410.

32
Thus, in this case mixed bundling is the profit maximizing choice.

Of course, the bundling scheme works as for other price discrimination methods only if bundles
cannot be split and units resold to the single product markets. By now, you should be able to
explain why (think arbitrage).

33
 Imperfect Competition in the Long Run

I. Monopolistic Competition

The objective of this class is to conduct a long-run analysis of an imperfectly competitive

industry. Since the objective of a long-run analysis is to examine the implication of entry and
exit of firms, we must allow for such entry and exit to take place. We will, therefore, consider
only markets in which, in spite of the imperfectly competitive nature of the market, there is free
entry and exit.
We call such a market: “Monopolistic Competition”. “Monopolistic” indicates that each
firm/brand is facing a downward sloping demand curve, while “competition” indicates that there
is free entry and exit.

What kinds of markets have such features, i.e., imperfect competition on one hand and free entry
and exit on the other hand?
Commonly cited examples include the cereal market, the toothpaste market, high-end
motorcycles, snowboards, restaurants in New York City, etc. These are markets in which there is
a variety of brands, where each brand is differentiated enough to face a downwards sloping
demand, BUT, while there is an obvious barrier to entry to the brand itself (you cannot start
selling a soft drink under the name “Coca Cola”), nothing prevents other companies from
entering the market with their own brand.

We will start by describing such a market in a short-run equilibrium and then move to the long-
run equilibrium.

Short-run Equilibrium
Consider, for example, the market for sport utility vehicles (SUV) in the US. Assume that there
are about 10 existing brands right now. Although not strictly necessary, it will be helpful to
identify a brand with a firm. Let’s also assume that the typical brand faces the demand curve and
has the cost structure as described in the graph below (To make the analysis simpler, assume that
the cost structures of all brands are identical):

34
 $/unit

MC
PSR

AC

MR
Q
QSR
Each brand is maximizing its profits in the short run by charging the profit maximizing price
(PSR) and selling QSR. In the graph, since the price is above the average costs, the firm has
positive economic profits. The shaded area in the graph represents these profits.

Long-run Equilibrium
The graph above describes the average, typical SUV brand as well as the situation that a
potential entrant is likely to face. If this is the situation, what is likely to happen in the long run?
The answer is identical to what we had in our analysis of perfect competition: Positive economic
profits attract entry. In our example, given that economic profits are positive, new brands of SUV
are likely to show up in the market.

What is likely to be the effect of entry on existing brands?

Be careful here. In the perfect competition case the effect of entry was a price reduction. You
cannot give such an answer here. Why? Simply because the firm decides what price to charge, so
such an answer will have no meaning.
What will happen here is that the demand curve faced by each brand will shift to the left. In other
words there will be a cut in the market share of each brand.
Notice that it is not clear what will happen to the shape of the demand curve as it “moves” to the
left. It can get steeper or flatter. It is, therefore, not clear whether, following a market share cut
the new profit maximizing price is going to be higher or lower. This is a non-intuitive point that
non-economists frequently fail to understand.

When will entry stop? Here the answer is identical to what we had in perfect competition. Entry
will stop as soon as the (expected) profits from entry are not higher than those expected in other
similar projects. In other words, entry will stoop when (expected) economic profits are zero.

35
If economic profits are zero, it must be the case that the market price is equal to the AC. If the
AC is exactly equal to the price, then it cannot be higher or lower than the demand. On the other
hand, to maximize profits the firm has to choose a quantity that equalizes marginal cost to
marginal revenue. The graph below describes this equilibrium:

$/unit

MC

PLR AC

D
MR

Q
QLR
Looking at the graph above you surely have noticed that the AC curve is tangent to the demand
curve. You might want to ask the following question:
Is it a coincidence that the AC is tangent to the demand curve EXACTLY at the same level of
output where the MR=MC?
The answer is: no. By definition AC will be tangent to AR (the demand) at the same point (QLR)
where MR = MC. The mathematical proof is trivial. Here it is without use of calculus:

1. At the point of tangency:

Slope of AC(QLR) = Slope of AR(QLR), and
AR(QLR) = AC(QLR)
MC − AC
2. Slope = Δ AC =
Q
MR − AR
3. Slope = Δ AR =
Q
4. Therefore, at tangency: MR(QLR) = MC(QLR)

Summary
Result is similar to LR in Perfect Competition.
Profits are driven down by entry.
In Perfect Competition it is the price drop that reduces profits.

36
Here the main force is market cut up. The result is similar.

Life is even more difficult:

You have to work on Demand (efficiency in marketing and distribution)
You have to beat the most efficient potential entrant in both Cost and Marketing!!
Niche Marketing or Differentiation is not enough
It must be a Protected Niche!
The key for success, therefore, is: Barriers to Entry.

II. Barriers to Entry

Barriers to entry are “obstacles” that make it costly for outsiders to enter an industry. Barriers,
when effective, translate into positive economic profits. It is almost a tautology in economics to
view positive economic profit as the product of barriers to entry.
There are three different forms of barriers to entry: customer barriers, cost barriers, and political
barriers. We discuss them in turn.

Consumer Barriers.
Consumer barriers translate into price advantage for the insiders. New entrants, in order to attract
customers, must sell at a considerable lower price. Obviously, the price differentials create
profits. For instance, suppose that a firm is operating in a market niche protected by a consumer
barrier. The insider and the outsiders have the same technology. However, outsiders are aware
that, if they had to enter the market, they could attract customers by offering a price cut of $5.
Entry will happen only if the difference between the market price and the average cost is $5.
Thus, in a long-run equilibrium, this market will show profits equal to $5 per unit.

There are different forms of consumer barriers that translate in consumer inertia or, to use
common words, brand image and consumer loyalty. One important form is related to situations
where customers have a “poor understanding” of the product. For instance, universities, law
firms, hospitals, and doctors often enjoy this type of barriers: customers are “unsophisticated”,
i.e., they have a relatively poor understanding (and are at times not really interested in
understanding) of the product. In all these markets, customers also exhibit some form of “risk
aversion.” To switch doctors or to change MBA program is a risky choice. The product is
complex and the consequences of a wrong choice can be devastating. Thus, competing with
established firms requires a substantial price cut.
Somewhat different, but with equivalent economic consequences, is the situation where
consumers that change brand (i.e., acquire the product from a different firm) bear a cost. There
are two different forms of the this phenomenon: a) “Switching cost” and b) “industry
standards” or “network externalities.”
a) Suppose that a corporation uses Microsoft Word as its word processing program.
Adopting new software would require retraining the entire staff, i.e., the corporation
would incur a substantial switching cost.
b) Some products are (more) valuable just because everybody else is using them, i.e.,
because they have become industry standards. Adobe’s acrobat PDF format is a good
example. Virtually everybody (individuals, universities and corporations) can send and

37
 read documents in PDF format. Competing with Adobe in this market is very hard. To
come up with a better product is not really the hard part of the problem. A successful
competing product must instantaneously acquire an important market share that makes
the competing product a viable alternative. This is clearly very difficult.

When firms are protected by market barriers profits are unusually high and market shares are
stable. When you look at actual data of actual markets, market share stability is a sign of little
entry activity. When combined with high profits, the latter is a sign of barriers. When only
customer barriers are present, profitability is often independent of size. Do not look for big
market shares.

Cost Barriers.
Whenever a firm can produce at lower cost than its (potential and/or effective) competitors, we
say that a cost barrier protects the firm. There are many sources that translate into cost barriers.
Some are natural, e.g., availability of the product (e.g., diamonds, oil), favorable location, better
inputs. Other sources are technological. Below we discuss briefly two classical forms of cost
barriers, economies of scale and economies of scope.
a) A technology displays “economies of scale” when the average cost of production is
declining over the relevant economic range. In industries with “economies of scale,“
firms with larger market shares have lower average costs. Sometimes, although the
average cost is declining, it becomes virtually flat at some market share level. That value
is called “Minimum efficiency scale” (MES). The term indicates the minimum market
share below which your average cost is prohibitively high. When we talk about industries
with economies of scale we think about industries with high MES (at least 6-7%). Entry
in an industry with economies of scale requires that entrants achieve very fast the MES.
Operating below MES produces losses that cannot be sustained for a long time. Of course
there are exceptions. For instance, AirBus, before becoming an effective competitor of
Boeing, experienced substantial losses —of course, it helped a lot to be backed by the
European tax-payers. A similar story was reproduced in the automotive industry when, in
the 80s, quotas on imports limited competition from Japanese car makers in Europe and
the U.S. .
b) Economies of scope occur when firms achieve cost savings by increasing the variety of
goods and services that they produce (joint production). Such effects arise when it is
possible to share components and to use the same facilities and personnel to produce
several products. For example, a bank may sell retail insurance products in its local
branches in order to spread the fixed costs (like the office rent) over a larger number of
products.

When you look at actual industries, where economies of scale are at work, you will notice three
things: a) share stability, b) unusually high profitability indexes, c) large market shares. In other
words, look for a positive correlation between size and profitability.

Other Barriers
We just mention here that legislation (governments) very often introduce barriers to protect

38
regional industries from international competition (e.g., quotas, tariffs and so on) or to encourage
firms and individuals in risky research activities (patents).

39