Cristina Rossi Lamastra

Politecnico di Milano School of Management

[email protected]

E. Advanced competition models

Advanced competition models: 2
Table of contents
• E.1. Oligopolistic models and their extensions
• E.1.1. Introduction to (oligopolistic) competition
• E.1.2. Bertrand model and its extensions
• E.1.3. Cournot model and its extensions
• E.1.4. von Stackelberg model

• E.2. Concentration and collusion

• E.2.1. Definitions and key concepts
• E.2.2. Concentration indexes
• E.2.3. Collusion

• E.3. Entry and entry barriers

• E.3.1. Introduction to entry and entry barriers
• E.3.2. How entry barriers work
• E.3.3. The theory of contestable markets

• E.4. Competition in network and platform industries

• E.4.1. Competition in network industries
• E.4.2. Competition in platform industries
Cristina Rossi Lamastra
 3

E.1. OLIGOPOLISTIC
MODELS AND THEIR
EXTENSIONS

Cristina Rossi Lamastra

 4

E.1.1. Introduction to (oligopolistic)

competition

Cristina Rossi Lamastra

 The dimensions influencing competition in an 5
industry
The strength of competition in an industry depends on many dimensions; we
focus here on
• The so-called structural dimensions, describing how an industry looks like

Number of firms

Concentration
Structural
dimensions
Possibility of product
differentiation

Entry barriers

• Strategic competition → Results of competition depend on strategic

interactions among firms. There are no strategic interactions in
• Monopoly, as there is only one firm in the industry
• Perfect competition, as there are too many firms in the industry

Cristina Rossi Lamastra

 Why is competition important? (1) 6

Competition is an essential force in the economy

Urges firms to be

Improves the allocation of Fosters innovation and
efficient resources increases the quality of
products/services
Name a possible disadvantage of competition. Respond at:
https://pollev.com/cristinaross747
Cristina Rossi Lamastra
 Why is competition important? (2) 7

For consumers For industries

Increased efficiency and innovativeness

Low prices Wide access to For whole economy

products

High-quality Fostering economic development

Widen choices products
Cristina Rossi Lamastra
 Introduction to (oligopolistic) competition: 8
Oligopoly and game theory (1)
Oligopoly: a competitive regime in which there are few firms in the industry; it features
• Strategic interactions, the results of each firm’s actions affect (and are affected by) other
firms’ actions
• Entry barriers, which hamper the entry of new firms

How do firms compete in an oligopoly?

• Non-cooperatively → Competition on price vs. quantity
• Cooperatively → Collusion

Game theory is a tool to study oligopolies → It models strategic interactions among firms as
rational economic agents whose actions affect those of other agents

Main assumptions
1. Firms are rational economic agents and aim to maximize their profits
2. Rationality is common knowledge
3. Information is perfect
Key notions
• Dominant strategy: it makes firm A better off regardless of what other firms do
• Nash equilibrium: A’s strategy is the best reply to the strategies of the other firms and vice-
versa → Firms have no incentive to deviate from the equilibrium
Cristina Rossi Lamastra
Introduction to (oligopolistic) competition: 9
Oligopoly and game theory (2)
Two Nash
Z strategy is
equilibria
dominated
by the others

Three types of oligopolistic models based on

• Strategic variable: price vs. quantity
• Timing of the choices: simultaneous vs. sequential

Strategic variable Timing of the choices Oligopolistic model

Price Simultaneous Bertrand
Quantity Simultaneous Cournot
Quantity Sequential von Stackelberg

Cristina Rossi Lamastra

 10

E.1.2. Bertrand model and its

extensions

Cristina Rossi Lamastra

Basic Bertrand model: 11
Assumptions
The basic Bertrand model relies on the following assumptions

1. Two firms (i and j) compete in the market → Duopoly

2. They produce goods that are perfect substitutes → Homogenous goods

3. There are no switching costs, and consumers buy from the firm charging the
lower price

4. Price is the strategic variable, which firms set independently and

simultaneously
• When deciding its price, each firm has not yet observed the other firm’s price
→ No information advantage

5. Both firms have no capacity constraints and the same linear cost function
• → Total cost: C(q)=cq and thus MC(q)=c

Competition is modeled through a one-shot, non-cooperative, symmetric

game
Cristina Rossi Lamastra
Basic Bertrand model: 12
Demand functions and profit maximization
Demand function of firm i  Di ( pi ) if pi < p j
 1
Di ( pi , p j ) =  Di ( pi ) if pi = p j
2
 0 if pi > p j
As consumers buy from the firm charging the lowest price, if firm i sets a price
• Lower than the price of firm j: firm i captures the whole market demand
• Equal to the price of firm j: each firm captures half of the market demand
• Higher than the price of firm j: firm i has no demand (all consumers buy from j)
The same holds for firm j

Each firm chooses the price that maximizes its profit

Firm i:
max π i ( pi , p j ) = pi Di ( pi , p j ) − cDi ( pi , p j ) = ( pi − c) Di ( pi , p j )
pi

Firm j: max π j ( pi , p j ) = p j D j ( pi , p j ) − cD j ( pi , p j ) = ( p j − c) D j ( pi , p j )
pj

Cristina Rossi Lamastra

 Basic Bertrand model: 13
Nash equilibrium
Nash equilibrium → Set of strategies such that each firm has no incentive to change its
strategy, given the strategy of the other firm(s)
• Set of best replies: no firm can increase its profit by changing its strategy unilaterally
• The Nash equilibrium in a duopoly where firms compete on prices is the vector

* *  π i ( pi * , p j * ) ≥ π i ( pi , p j * ) ∀ pi ≠ pi *
( pi , p j )

π
 j i j
( p
*
, p
*
) ≥ π j ( pi
*
, p j ) ∀ p j ≠ p j
*

In the (Nash) equilibrium of a Bertrand duopoly, the two firms charge the same price,
which equals the marginal cost ∗ ∗
𝑝𝑝 𝑖𝑖 = 𝑝𝑝 𝑗𝑗 = 𝑐𝑐

Bertrand paradox: same results as in perfect competition with just two firms
Specifically, one can demonstrate that, for each firm, setting the price
• Lower than the marginal cost determines negative profits → This strategy is strictly
dominated by any other strategy, which brings positive profits or no profits
• Higher than the marginal cost is not an equilibrium

Cristina Rossi Lamastra

 Basic Bertrand model: 14
Proof of the Nash equilibrium
∗ ∗
The proof consists in showing that 𝒑𝒑 𝒊𝒊 = 𝒑𝒑 𝒋𝒋 = 𝒄𝒄 is the only possible equilibrium: in
all the other cases, firms have the incentive to deviate

pi > p j > c
• Firm j gets the entire demand and earns positive profit, while firm i earns no profit
• Firm i has the incentive to decrease its price at an intermediate level between pj
and c, thus earning positive profit → This is not an equilibrium

pi = p j > c
• The two firms share the market and earn positive profits
• Each firm has the incentive to decrease its price to serve the entire demand (for
a slight price reduction, profits will be higher) → This is not an equilibrium

pi > p j = c
• Firm j gets the entire demand but earns no profit, while firm i earns no profits as it
sells nothing
• Firm j has the incentive to increase its price, while firm i has the incentive to
decrease it → This is not an equilibrium
Cristina Rossi Lamastra
Bertrand model: 15
Extensions
In the Nash equilibrium of the (classical) Bertrand model, for each firm, it holds that
p=c, but this means no profit for both firms
Bertrand paradox: in a Bertrand competition, the price mechanism causes the
same result as in perfect competition

This result strongly depends on the model assumptions; it is possible to obtain

equilibrium prices higher than the marginal cost by changing these assumptions

1. Capacity constraints: each firm cannot serve the entire demand by itself; it has a
maximum capacity level of k
• Key notion: residual demand (RD): if pi<pj and D(pi)>ki → qi=ki and RDj = D(pj)-ki
• The equilibrium price is such that both firms produce at their capacity level →
qi=ki, qj=kj, and p=p(ki+kj)

2. Not homogeneous goods: goods are not perfect substitutes; thus, it is not true that the
demand of firm i(j) goes to zero if pi>pj (pj>pi)

3. Diverse cost functions: for instance, because firms use different technologies

4. Repeated games: firms compete on the market repeatedly and thus realize that a price
war hurts them both by driving the profits to zero → Collusion
Cristina Rossi Lamastra
 Bertrand model - Extensions: 16
Not homogeneous goods
If products are not homogenous, a price reduction does not imply that the rival gets no
demand → p=c is no longer the equilibrium, and firms earn positive profits
• E.g., the same good sold in different locations (→ Hoteling model)

Example → Try to solve it on your own (Prof. Silvestri will revise the exercise)
• Two firms with not homogenous goods competing on prices face the following demand
functions, respectively 1
q1 ( p1 , p2 ) = 300 − p1 + p2
2
1
q2 ( p1 , p2 ) = 300 − p2 + p1
2
• By assuming zero cost of production for both firms, find equilibrium prices, quantities,
and profits
Each firm, given the price of the other firm, chooses the price that maximizes its profit
1
• For firm 1, this means: max 𝜋𝜋1 = 𝑝𝑝1 ∗ 𝑞𝑞1 − 0 = 𝑝𝑝1 (300 − 𝑝𝑝1 + 𝑝𝑝2 )
𝑝𝑝1 2
𝜕𝜕𝜋𝜋1 1
• First-order condition implies: = 300 − 2𝑝𝑝1 + 𝑝𝑝2 = 0.
𝜕𝜕𝑝𝑝1 2
𝜕𝜕𝜋𝜋2 1
Given the symmetry of the problem, for firm 2 it is: = 300 − 2𝑝𝑝2 + 𝑝𝑝1 = 0
𝜕𝜕𝑝𝑝2 2

The optimal prices by solving the system with the two first-order conditions: 𝑝𝑝1∗ = 𝑝𝑝2∗ = 200.
Thus, the two firms produce 𝑞𝑞1∗ = 𝑞𝑞2∗ = 200 and gain profits 𝜋𝜋1∗ = 𝜋𝜋2∗ = 40,000
Cristina Rossi Lamastra
 Bertrand model - Extensions: 17
Diverse cost functions (1)

If firms have diverse cost functions, they produce the same good at
different costs

Let us assume that both firms have linear cost functions, with c1 < c2
• Firm 1 has a cost advantage
• The lowest price that firm 2 can set is c2

The optimal price for firm 1 depends on its monopoly price and on c2.
If
• 𝑝𝑝1𝑀𝑀 < 𝑐𝑐2 → 𝑝𝑝1∗ = 𝑝𝑝1𝑀𝑀 firm 1 can set its monopoly price as firm 2
cannot capture the entire demand by setting a price lower than 𝑝𝑝1𝑀𝑀

• 𝑝𝑝1𝑀𝑀 > 𝑐𝑐2 → 𝑝𝑝1∗ = 𝑐𝑐2 − 𝜀𝜀 firm 1 cannot set its monopoly price as firm 2
can capture the entire demand by setting a price lower than 𝑝𝑝1𝑀𝑀
• Firm 1 sets a price just a little lower than c2, the lowest price that
firm 2 can set
Cristina Rossi Lamastra
 Bertrand model - Extensions: 18
Diverse cost functions (2)

Example → Try to solve it on your own (Prof. Silvestri will revise the exercise)
Two firms compete on prices and face this demand function: q(p)=6,000–60p
• What is the equilibrium when c1=10 and c2=12?
• How does the equilibrium change if c2=60?
• Firm 1 has a cost advantage over firm 2, as c1<c2
• Let us compute the monopoly price for firm 1, pM. It results from MR = MC1
• The inverse demand curve is p=(6,000 - q)/60=100-1/60*q
• MR=100-1/30*q. Hence, 100-1/30*q = 10 → qM = 2,700 and pM= 55
As the monopoly price is pM=55
• When c2=12, in equilibrium, firm 1 cannot set the monopoly price and it sets p1=c2-
ε=12–ε
 Considering p1 ≈ 12, firm 1 produces q1 = 5,280 and gets a profit of π1=10,560
 Firm 2 sets p2=c2=12, does not produce (no demand), and gets zero profits
• When c2=60, in equilibrium, firm 1 can sets the monopoly price p1=pM=55
 Considering the demand function, firm 1 produces q1=2,700 and gets a profit of
π1=121,500
 Firm 2 sets the price p2=c2=60, does not produce (no demand), and gets zero
profits
Cristina Rossi Lamastra
 19

E.1.3. Cournot model and its

extensions

Cristina Rossi Lamastra

Basic Cournot model: 20
Assumptions
The basic Cournot model relies on the following assumptions
1. Two firms (1 and 2) compete in the market → Duopoly

2. They produce goods that are perfect substitutes → Homogenous goods

3. There are no switching costs, and consumers buy from the firm charging the lower price

4. Quantity is the strategic variable that firms set independently and simultaneously
• This is a mild form of competition as changes in quantities
• (Indirectly) cause changes in prices
• Are time-consuming (e.g., setting up new production plants)
• When deciding its quantity, each firm has not yet observed the other firm’s quantity →
No information advantage

5. Price results from total quantity offered on the market Y = y1 + y2. Specifically,
considering a linear demand function: p(Y) = 1-Y (with a=b=1)

6. Both firms have the same linear cost function and no capacity constraints
• Marginal cost is constant: MC(yi)=c (c<1)

A one-shot, non-cooperative, symmetric game models competition

Cristina Rossi Lamastra
Basic Cournot model: 21
Nash equilibrium
For each firm, the best reply function reports the quantity which maximizes profit,
whatever quantity the rival chooses
• Given y2, the profit maximization problem that firm 1 solves is
max𝜋𝜋1 𝑦𝑦1 , 𝑦𝑦2 = 1 − 𝑦𝑦1 − 𝑦𝑦2 𝑦𝑦1 − 𝑐𝑐𝑦𝑦1 = (1 − 𝑦𝑦1 − 𝑦𝑦2 − 𝒄𝒄)𝑦𝑦1
𝑦𝑦1

• The first-order condition is ∂π ( y1 , y2 )

= 1 − 2 y1 − y2 − c = 0
∂y1
• From which, we get firm 1’s best reply function 1 − y2 − c
y1 =
2
• As the problem is symmetric, firm 2’s best reply function is 1 − y1 − c
y2 =
2

The Nash equilibrium is at the intersection of the two best reply functions
• Equilibrium quantities 1− c
y1* = y2* =
3
1 + 2c
• Equilibrium price p* =
3 π =π
* *
=
(1 − c)
2

1 2
• Both firms earn positive profits 9
Cristina Rossi Lamastra
Basic Cournot model: 22
Graphical representation

1 − y2 − c 1 − y1 − c
y1 = y2 =
2 2

1−𝑐𝑐
If 𝑦𝑦𝟏𝟏 = 𝟎𝟎 →𝑦𝑦𝟐𝟐 =
2
1−𝑐𝑐
If 𝑦𝑦𝟐𝟐 = 𝟎𝟎 →𝑦𝑦𝟏𝟏 =
2

Cristina Rossi Lamastra

 Cournot model: 23
Extensions
If firms have diverse cost functions, they produce the same good at different
costs; let us assume linear cost functions
• If c1 < c2 (firm 1 has a cost advantage), in equilibrium, we have
1 − 2𝑐𝑐𝟏𝟏 + 𝑐𝑐𝟐𝟐 1 − 2𝑐𝑐𝟏𝟏 + 𝑐𝑐𝟐𝟐 2
𝑦𝑦𝟏𝟏∗ = 𝜋𝜋𝟏𝟏∗ =
3 9

• The firm with the lower cost produces more and earns higher profits

Cournot with n symmetric firms

• For each firm i = 1,…n, in equilibrium, we have
1− c 1− c  1− c 
2
y = *
p =c+
*
π =
*

1+ n 1+ n
i
+
i
 1 n 
As the number of firms increases
• The price tends to marginal cost, c (as in perfect competition)
• Profits tend to be 0 (as in perfect competition)
• As the price decreases, the total quantity increases
• Social welfare increases
Cristina Rossi Lamastra
 24

E.1.4. von Stackelberg model

Cristina Rossi Lamastra

von Stackelberg model: 25
Definition and follower’s choice
In the von Stackelberg model, the assumptions of the Cournot model hold, but firm 1
is the first to choose its quantity, then firm 2 responds to this choice
• Quantity is the strategic variable
• Goods are homogeneous
• Firm 1 is the leader, firm 2 is the follower
• A sequential game (two periods) models the competition

Given the demand: p(Y)=1-Y, Y=y1+y2, we resort to backward induction

• We start from period 2 and derive the follower’s best reply, given the choice of
max π 2 ( y1 , y2 ) = (1 − y1 − y2 − c) y2
the leader
• Follower’s profit maximization y2
• Follower’s first-order condition and the best reply

∂π 2 ( y1 , y2 ) 1 − 𝑦𝑦1 − 𝑐𝑐
= 1 − 2 y2 − y1 − c = 0 𝑦𝑦∗2 =
2
∂y2
• We insert the follower’s best reply in the leader’s profit function and compute the
optimal quantity y1*

Cristina Rossi Lamastra

 von Stackelberg model: 26
Leader’s choice
1 − y1 − c
The leader knows how the follower will behave max π 1 ( y1 ) = (1 − y1 − − c) y1
y1 2
• Leader’s first order condition
∂π 1 ( y1 ) 1 1
= 1 − 2 y1 − + y1 − c = 0
∂y1 2 2
Hence, in equilibrium

Quantities y1* = 1 − c 1 − y1* − c 1 − c 1 − c 1 − c 1 + 3c

• y = *
2 = − = Price p* =
2 2 2 4 4 4

1 1− c  1 1− c 
2 2

π 1* =   π 2* =  
• Profits 2 2  4 2 

In equilibrium, the leader produces more and earns higher profits than the
follower
• First-mover advantage
Cristina Rossi Lamastra
 27

E.2. CONCENTRATION AND

COLLUSION

Cristina Rossi Lamastra

 28

E.2.1. Definition and key concepts

Cristina Rossi Lamastra

 Industry concentration 29

Concentration deals with the number and size (i.e., the dimensional distribution)
of firms in an industry

• Few firms of large size → High concentration, concentrated industry

• Many firms, but few of large size→ High concentration, a concentrated
industry with “a competitive fringe”
• Many firms, none of large size → Low concentration, “dispersed” industry

The firm’s size is computed based on the number of employees, output volumes,
revenues, added value…

Concentration is a vital feature of an industry and provides information about

competitive mechanisms

• High concentration → Oligopolistic structure, with a high risk of collusion.

Firms have market power → They can set prices above marginal costs

• Low concentration → (Monopolistic) competition

Cristina Rossi Lamastra

 Market shares and concentration vector 30

Market share (sij) of firm i in industry j, in which N firms operate, is the ratio of
the output of firm i in industry j (qij) over the total output of j (Qj , sum of
outputs of all firms in industry j)

Firm Output Market Share

1 q1 q1/Qj s1
𝐍𝐍
2 q2 q2/Qj s2 𝐪𝐪𝐢𝐢,𝐣𝐣
𝐰𝐰𝐰𝐰𝐰𝐰𝐰𝐰 𝐐𝐐𝐐𝐐 = � 𝐪𝐪𝐤𝐤,𝐣𝐣
… … … 𝐐𝐐𝐣𝐣
𝐊𝐊=𝟏𝟏
i qi qi/Qj si
… … …
N qN qN/Qj sN
𝐍𝐍

𝐐𝐐𝐣𝐣 = � 𝐪𝐪𝐤𝐤𝐤𝐤
𝐤𝐤=𝟏𝟏

• The sum of market shares in an industry is equal to 1 ∑𝐍𝐍

𝐤𝐤=𝟏𝟏 𝐬𝐬𝐤𝐤𝐤𝐤 = 𝟏𝟏
• The concentration vector contains the market shares in decreasing order
𝑺𝑺 = 𝒔𝒔𝟏𝟏𝒋𝒋 , 𝒔𝒔𝟐𝟐𝒋𝒋 , . . . 𝒔𝒔𝒏𝒏𝒋𝒋
Cristina Rossi Lamastra
Concentration curve and concentration 31
indexes
The concentration curve represents the concentration of an industry in a given
period on a Cartesian plane
• Horizontal axis: firms ordered according to their decreasing (non-increasing)
market shares (from the largest to the smallest one)
• Vertical axis: cumulative market shares (CMS), it is a monotonic curve
(increasing, at decreasing rates)
CMS

0,5

N Firms
Concentration indexes synthesize all the information contained in the
concentration vector
• They allow comparing concentration
𝑁𝑁 across periods and industries
• General formula 𝐼𝐼 = � ℎ𝑖𝑖 (𝑠𝑠𝑖𝑖 )𝑠𝑠𝑖𝑖
𝑖𝑖=1
• A concentration index is the weighted sum of market shares
• Depending on the weights, hi(si), we define diverse concentration indexes
Cristina Rossi Lamastra
 32

E.2.2. Concentration indexes

Cristina Rossi Lamastra

Concentration ratio 33

The sum of the market shares of the first k firms in the industry
• Firms are ordered in decreasing order according to their market shares
• Weight of the first k firms → hi(si) = 1 k
• Weight of the other N-k firms → hi(si) = 0 Ck = ∑ si
i =1
• Conventionally, k = 3, 4, 8, 20
• Given k, industry A is more concentrated than industry B if Ck(A) > Ck(B)
• 0 < Ck ≤ 1, equal to 1 when the first k firms cover the whole market, equal to
0 in perfect competition

Problems
1. Which k to choose?
2. Ck provides the same results for industries with different concentration
vectors
• Industry A: IA = (0.1, 0.1, 0.1, 0.1, …), C4A=0.4
• Industry B: IB = (0.37, 0.01, 0.01, 0.01, …), C4B=0.4
• C4 disregards that, in industry B, there is a firm totaling nearly 40% of the
industry output
Cristina Rossi Lamastra
 34
Herfindahl index
N

The weighted sum of the market shares of all firms in the industry HI = ∑ si
2

i =1

• The weight of each share is equal to the share itself; thus, HI is the sum of the
squared market shares of all firms in the industry
• Small firms contribute to the value of the index LESS than large firms

It holds that 1/N ≤ HI ≤ 1

• HI = 1/N when there are N firms of the same size
• Each firm account for 1/N of the total output
• In perfect competition: HI = 0 (N → ∞)
• HI = 1 in monopoly

A qualitative classification relates concentration, as measured by HI, and market

structure
Nature of competition Range of HI
(Monopolistic) competition < 0.2
Oligopoly 0.2 - 0.6
Monopolistic structure > 0.6
Cristina Rossi Lamastra
 Entropy index 35

The weighted sum of the market shares of all firms in the industry N
1 
EI = ∑ si ⋅ ln 
i =1  si 
• The weight of each share is equal to the (natural) logarithm of the inverse of
the market share itself
• Small firms contribute to the value of the index MORE than large firms (as
the index is based on the inverse of the market share)

Entropy is a measure of disorder/dispersion

• 1 firm → Minimum industry entropy, EI=0

N
1 N
EI = ∑1 ⋅ ln  = ∑1 ⋅ 0 = 0
i =1  1  i =1

• N firms of equal size → Entropy increases as N increases

1 1
𝐸𝐸𝐸𝐸 = ∑𝑁𝑁
𝑖𝑖=1 � 𝑙𝑙𝑙𝑙 1 = 𝑙𝑙𝑙𝑙(N)
𝑁𝑁
𝑁𝑁

Cristina Rossi Lamastra

 36

E.2.3. Collusion

Cristina Rossi Lamastra

 Collusion: 37
Definition and conditions
Collusion occurs when rival firms in an industry cooperate for mutual benefit
It often occurs in concentrated industries where
• Firm can easily find a collusive agreement
• The decision of a few firms to collude significantly impacts the whole industry

Two types of collusion exist

1. Explicit collusion: firms behave as an organized cartel (e.g., OPEC)
2. Tacit collusion: firms agree to maximize joint profits and on a rule for sharing them
without explicitly saying so (collusion is forbidden by anti-trust law)

Collusion allows firms to limit competition, increase prices, reduce quantities, and
exert market power
• Ultimately, it reduces social welfare as firms jointly behave as a monopolist

Conditions for collusion to occur

1. Firms must engage in repeated interactions → Repeated games with an infinite
(indefinite) time horizon
2. Firms can detect alleged deviations from the collusive agreement
3. A credible punishment for deviation exists, which discourages deviations
Cristina Rossi Lamastra
Collusion: 38
Interactions over a finite-time horizon
We can model collusion through the Prisoner’s Dilemma game

Firm 1

D: Deviate from the C: Stick on the collusive

collusive agreement agreement

D: Deviate from the

collusive (2;2) (12;1)
agreement
Firm 1
C: Stick on the
collusive (1;12) (10;10)
agreement

• The Nash equilibrium is the set of strategies (D, D)

• The set of strategies that maximizes the sum of payoffs is (C, C)

Collusion does not arise if firms interact

• Just one time → (C, C) is not the equilibrium of a one-shot game
• A finite number of times → Because of backward induction, in each interaction,
the result is the same as in the first one (D, D)
Cristina Rossi Lamastra
Collusion: 39
Interactions over an infinite time horizon
Assumptions

• n firms interact an infinite (indefinite) number of times → Infinite time

horizon game
• In each period, firm i can either collude (C) or deviate from the collusive
agreement (D)
• If firm i deviates, the other firms will punish it ( → punishment phase)

Definition

• πiC current profit of firm i if it colludes and all the other firms collude

• ViC future profits of firm i if it colludes and all the other firms collude

• πiD current profit of firm i if it deviates and all the other firms collude

• ViP future profits of firm i in the punishment phase

• δ discount factor (from 0 to 1) for future profits, which is the same for all
the firms and expresses their “degree of patience”
Cristina Rossi Lamastra
Conditions for collusion: 40
Critical discount factor
Collusion arises only if each firm prefers collusion to deviation; for each firm,
it should hold (→ incentive constraint)
π iC + δVi C ≥ π iD + δVi P i = 1,  , n
• Rearranging the terms of the incentive constraint, it holds that

• [1] 𝜋𝜋𝑖𝑖𝐷𝐷 − 𝜋𝜋𝑖𝑖𝐶𝐶

𝛿𝛿 ≥ 𝐶𝐶 ≡ 𝛿𝛿𝑖𝑖 𝑖𝑖 = 1, … , 𝑛𝑛
𝑉𝑉𝑖𝑖 − 𝑉𝑉𝑖𝑖𝑃𝑃

[1] does not depend on whether competition is on quantities or prices

[1] tells that collusion arises only if the discount factor is higher than the
“critical discount factor” 𝛿𝛿𝑖𝑖 , namely if firms are patient and attach importance
to future profits

Collusion is more likely if 𝛿𝛿𝑖𝑖 is low; this happens, for instance, when
• The deviation profit (πiD) is low
• The future profits in the punishment phase (ViP) are low
Cristina Rossi Lamastra
Collusion with price competition: 41
Main assumptions
There are n firms
• Interacting repeatedly over an infinite time horizon
• Producing homogeneous goods
• Facing a linear demand function
• Having the same constant marginal cost, c
• Having no capacity constraints
• Having the same discount factor, δ
• Aiming to maximize their total discounted profits
The equilibrium if they compete (Bertrand game with homogeneous goods) would be
pi = pj = … = pn= c

If firms are allowed to collude, they

• Set prices equal to the one that maximizes joint profits (i.e., the monopolistic
price): pi = pj = … = pn = pM
• Produce and sell the quantity: qi = qj = … = qn = QM/n
• Earn a profit: πi = πj = … = πn = πM/n

Does collusion arise? Is it sustainable?

Cristina Rossi Lamastra
 Collusion with price competition: 42
Trigger strategy
All firms play the intertemporal trigger strategy (→ Tit for tat: this for that, Axelrod, 1984)
• Firm i chooses pM in the first period
• In subsequent periods, the other firms keep choosing pM only if firm i keeps choosing pM
• Once firm i deviates, the other firms play the Bertrand strategy (which lasts forever): pi=c

With n firms, collusion arises if no firm has the incentive to deviate → If, for each firm, profits
of collusion are higher than profits of deviation
𝜋𝜋𝑀𝑀 𝜋𝜋𝑀𝑀 𝜋𝜋𝑀𝑀
+ 𝛿𝛿 + 𝛿𝛿 2 +. . . ≥ 𝜋𝜋 𝑀𝑀 𝑝𝑝𝑀𝑀 − 𝜀𝜀 = 𝜋𝜋 𝑀𝑀 𝑝𝑝𝑀𝑀 = 𝝅𝝅𝑴𝑴
𝑛𝑛 𝑛𝑛 𝑛𝑛

• This is equivalent to
1 πM 1
≥ 1− δ ⇔ δ ≥ 1−
1
=δ*
≥πM
1− δ n n n

Collusion is sustainable if the discount factor is higher than the critical discount factor δ*
• If δ* is high, even patient firms (i.e., δ high) choose not to collude
• δ* threshold increases with the number of firms → Collusion is less sustainable in
competitive/poorly concentrated industry
• In the duopoly case (n = 2), the threshold is ½ (if δ ≥ ½ collusion arises, if δ < ½, it does
not)
Cristina Rossi Lamastra
Factors enabling collusion: 43
A taxonomy
Identifying factors enabling collusion to support anti-trust authorities in deciding
whether and how to intervene
Name a factor making collusion more likely (one short sentence). Respond at:
https://pollev.com/cristinaross747

We can group these factors into

A. Structural elements
A.1. Number of firms, concentration, symmetry (i.e., firms have similar market shares)
A.2. Frequency of transactions
A.3. Evolution of the demand
A.4. Other industry characteristics, e.g., diversification and consumers’ bargaining
power

B. Information availability
B1. Transparency
B2. Information sharing

C. Contract clauses
Cristina Rossi Lamastra
 Factors enabling collusion: 44
Structural elements (1)
A.1. Number of firms, concentration, symmetry: collusion is more likely
• The fewer firms in the industry
• The higher the concentration
• The higher the symmetry, which means there is a limited power imbalance

A.2. Frequency of transactions

• High frequency of transactions facilitates collusion due to timely punishment
 If little time passes between a transaction and the subsequent one, punishment
would be quick, and the firm would prefer not to deviate
• A substantial transaction, which grants immediate large profits, is a solid incentive to
deviate

A.3. Evolution of demand: the effect of collusion depends on

• Positive/negative (→ sudden increase/decrease) shocks in demand. Collusion is
• Less likely if a positive demand shock occurs → Immediate profits are large and
create the incentive to deviate for capturing them
• More likely, if a negative shock occurs → Immediate profits are low and do not
create an incentive to deviate for capturing them
• Forecasting of the future demand
• Collusion is less likely if the future demand is complex to forecast → Firms have the
incentive to deviate for capturing immediate profits
Cristina Rossi Lamastra
Factors enabling collusion: 45
Structural elements (2)
A.4.1. Entry barriers: the lower the entry barriers, the more challenging to
sustain collusion
• New entrants have incentives to deviate as they need immediate profits

A.4.2. Links among competitors: if a firm’s manager is in the BoD of a rival

firm, the two firms can easily coordinate and, thus, collusion is more likely

A.4.3. Consumers’ power: collusion is less likely in the presence of large

consumers, which can use their bargaining power to stimulate competition
among sellers

A.4.4. Diversification: it has an ambiguous effect on collusion. In diversified

firms
• On one side, collusion is less likely as punishment is more difficult
• Firms can compensate for losing collusive profits in one industry with
profits in another industry
• On the other side, collusion is more likely as the deviation is less
profitable
• Each industry accounts for a limited profit share
Cristina Rossi Lamastra
Factors enabling collusion: 46
Other factors
B. Transparency and information sharing
B1. Transparency (i.e., prices and quantities are observable) favors collusion, as firms
can easily come to a collusive agreement
B2. Information sharing may favor collusion. Specifically, information sharing on
• Past and current prices allow identifying deviators and targeting punishments
• Past and current quantities increase the predictability of the industry’s total supply
• Future prices and quantities may occur through
• Private announcements (directed only to competitors) favor coordination among
competitors
• Public announcements (directed to competitors and consumers)
• Favor coordination among competitors
• They also reduce information asymmetries for consumers, raising suspects
about uniformity in prices

C. Contractual clauses
• Some clauses in (long-term) contracts may favor collusion as they disclose relevant
price information
• The “meeting competition”: if the customer receives a better offer from another firm,
the focal firm commits to match the price → This discloses information about the price
Cristina Rossi Lamastra
 47

E.3.1. Introduction to entry and

entry barriers

Cristina Rossi Lamastra

 Introduction: 48
Entry and entrants

Firms enter (and exit) industries, and this affects the competition

(Potential) entrants: firms intending to enter an industry to commercialize goods that are
substitutes for those commercialized by firms already in the industry (→ Incumbents)
• De novo entrants → Newly created firms (→ startups), which suffer from liability of
newness and smallness but benefit from flexibility
• De alio entrants → Incumbents entering the industry through diversification
 De alio entrants are usually more dangerous for incumbents than de novo ones,
as they have more resources to compete

New firms enter an industry as profits, which incumbents earn, attract them. In general
• In deciding if to enter the industry, entrants compare expected profits and entry
costs
• Following entries, the industry supply increases and, thus, the price decreases
• The entry process ends when p=MC → No firm makes a profit

In short, entry increases competition and reduces incumbents’ profits → Incumbents

rely on entry barriers to shield against it

Cristina Rossi Lamastra

 Entry barriers: 49
Definitions and conditions

Entry barriers: obstacles that make it difficult to enter an industry → They

raise entry costs and make it unprofitable to enter the industry; thus, they
“protect” incumbents’ profits
We distinguish
1. Institutional entry barriers created by the policymaker’s intervention
2. Non-institutional entry barriers
• 2.1. Structural entry barriers related to incumbents’ cost advantages
• 2.2. Strategic entry barriers related to incumbents’ actions → Entry
deterrence

Based on entry barriers, we define three main entry conditions

1. Blockaded entry: institutional and structural entry barriers are so high that
incumbents do not need to resort to entry deterrence
2. Accommodated entry: institutional and structural entry barriers are low, but
entry deterrence does not pay off
3. Deterred entry: institutional and structural entry barriers are low, and entry
deterrence does pay off
Cristina Rossi Lamastra
 50

E.3.2. How entry barriers work

Cristina Rossi Lamastra

 Institutional and structural entry barriers 51

1. Policymakers create institutional entry barriers, as in the case of

• Requirements for licenses and permits to enter the industry (e.g., taxi licenses)
• Ownership of intellectual property rights by the incumbents (e.g., patents)

2. Structural entry barriers, (cost) advantages of the incumbents (mainly) related to

technology and demand conditions
• These barriers usually are a side-effect of incumbents’ (profit-maximizing)
decisions

Prominent examples of structural entry barriers

1. Control of crucial resources in the value chain (e.g., key inputs or distribution
channels) → Entrants must bear (significantly) high costs to access these
resources

2. Economies of scale → To be efficient, firms must be large. If large incumbents

already exist, entrants cannot achieve economies of scale and bear high costs
• Being efficient, incumbents can reduce their prices if new firms enter the industry

3. Incumbents’ marketing advantage → In terms, for instance, of customers’ loyalty

and umbrella branding → Entrants must bear the high cost of advertising
Cristina Rossi Lamastra
 Strategic entry barriers 52

When institutional and structural entry barriers are low, incumbents may engage in
actions to deter entry → Strategic entry barriers

Entry deterrence does pay off if

1. Incumbents earn higher profits if the entry does not occur (it is always the case)
2. Entry deterrence changes entrants’ expectations of their post-entry profits
• Entrants compare expected revenues with entry costs, expecting retaliation by
incumbents (e.g., price reductions), and thus low profits, they decide not to enter
3. It is based on a credible threat → If entry occurs despite entry deterrence, it is still
optimal for incumbents to engage in retaliation (e.g., reducing price)

Examples of entry deterrence strategies by the incumbents

• Fixing the limit price
• Other entry deterrence strategies
• Product strategies: bundling and product proliferation
• Signing long-term contracts
• Increasing production capacity
If entry occurs despite entry deterrence, incumbents can resort to predatory pricing
Cristina Rossi Lamastra
 Limit price: 53
The model of Bain, Modigliani, and Sylos-Labini
The incumbent, which is a monopolist, does not set the monopoly price but
the so-called limit price that drives the potential entrant’s profits to zero

• Limit price deters a potential entrant from entering the industry

• The incumbent sets the limit price indirectly by setting the limit quantity
• Limit price is determined via the demand curve

Assumptions of the model of the limit price

1. There are an incumbent, m, and a potential entrant, e
2. Consumers have no switching costs
3. The model unfolds over two periods, pre-entry (t=0) and entry period (t=1)
• t=0, m commits to producing a given quantity and, thus, charging a given
price
• t=1, e decides whether to enter or not. In making this decision, e assumes
that m will keep the same quantity and, thus, the same price it set at t=0
• This assumption is a somewhat unrealistic
4. There is perfect information
Cristina Rossi Lamastra
 Limit price: 54
Blockaded entry
Blockaded entry: given its costs, it is never profitable for e to enter the industry
→ The incumbent does not need to deter entry by setting the limit price

t = 1: the (potential) entrant evaluates its

t = 0: the incumbent chooses the residual demand and costs and decides
monopoly quantity and price, which not to enter the industry.
maximize its profits, given the demand As the average cost is higher than the price,
curve post-entry profits are negative

ACe

Cristina Rossi Lamastra

 Limit price: 55
Not blockaded entry (1)
Not blockaded entry: e’s costs allow for profitable entry, depending on the price set by i

t = 0: the incumbent chooses the monopoly t = 1: the (potential) entrant evaluates its
residual demand and costs and decides to
quantity and price, which maximize its profits,
enter the industry
given the demand curve As the average cost is lower than the price,
its post-entry profit is positive

The incumbent should deter entry by choosing a different combination of quantity and price
Cristina Rossi Lamastra
 Limit pricing: 56
Not blockaded entry (2)
Q: Which is the maximum price – limit price - which m can set, by choosing a
limit quantity, to deter entry?
A: Limit price (pL) and limit quantity (qL) are such that they make e indifferent
between entering and not entering the industry
• qL, pL are such that the residual demand of e is tangent to its AC → In this
case, e’s post-entry profit goes to zero

• Any price higher than pL causes positive profits for e and triggers its entry
Cristina Rossi Lamastra
 Critics to the Bain, Modigliani, Sylos-Labini 57
Model
i threats e to charge pL by choosing qL, but if e decides to enter despite the threat,
enacting this threat might not be optimal for i
• In case of entry, the best reply may be to choose the Cournot quantity, which
maximizes i’s profit in a duopoly → The incumbent’s commitment to the limit
quantity is not credible An exemple: Kodak
vs. Polaroid

• K has 2
strategies: stay
Out or going In
• Then P has 2
strategies:
Fighting or
Accomodate
In the one shot
game, two Nash
equilibria exist

(Out, F) is eliminated
by backward
induction

Cristina Rossi Lamastra

Other entry deterrence strategies: 58
Bundling and product proliferation

Bundling occurs when a firm sells a combination of goods at a price that is

lower than the price of buying the same goods separately

Typically, firms resort to bundling to reduce costs or for marketing reasons, but
bundling also allows to deter entry
• This happens when an incumbent, operating in industries i and j, is the
market leader in industry i and faces entry threats in industry j
• Bundling good i and good j induces consumers to buy the bundle from the
incumbent rather than buying good i from the incumbent and good j from the
(potential) entrant
 The incumbent can charge a lower price for the bundle

Product proliferation occurs when the incumbent enlarges the scope of its
offering to meet all customers’ preferences
• No room for profitable entry exists → A potential entrant should bear very
high costs for differentiating its goods from the incumbent’s ones
• Example: the case of ready-to-eat breakfast cereals in the US

Cristina Rossi Lamastra

Other entry deterrence strategies: 59
Long term contracts and increasing production capacity

Long-term contracts bond (large) consumer(s) to incumbents

• Potential entrants have limited room for maneuvering as incumbents have
already secured a sizable portion of the demand
• Why do consumers accept signing long-term contracts instead of waiting
for new entrants and (likely) lower prices?
• Name a factor inducing customers to accept a long-term contract
(one short sentence). Respond at: https://pollev.com/cristinaross747
• Assurance of quality on crucial inputs
• Reduction of uncertainty on crucial inputs
• Example: The case of Monsanto, Coca-Cola, and Pepsi

Increasing production capacity through such a move, incumbents

• Put in place a credible threat by bearing sunk costs
• Signal that they can reduce price, capture all the demand, and serve it
Potential entrants anticipate that no room for profitable entry exists and do not
enter the market
• Example: The case of DuPont and the patent on titanium dioxide
Cristina Rossi Lamastra
 Predatory pricing: 60
Definition and rationales
Predatory pricing: in the short run, the incumbent sets the price below its
marginal cost to get rid of new entrants and earn higher profits in the long run
• Low prices are generally associated with high efficiency
• In case of predatory pricing, the incumbent sets low prices for anti-competitive
reasons → Forcing new entrants out of the industry or deterring potential
entrants

With predatory pricing

• Potential entrants are forced to “think twice” about entering the industry as they
envisage that it is hard to gain post-entry profits
• New entrants are severely challenged in recovering their entry costs
• Predatory incumbent expects to recover the losses incurred while charging
predatory prices through future (monopolistic) profits
• In these cases, social welfare
• Increases in the short run because of low prices
• Decrease in the long run because monopoly prices
Key issue: it is hard to distinguish predatory prices from low prices due to
efficiency
Cristina Rossi Lamastra
 Predatory pricing: 61
Explanation and challenges
The main explanation for predatory pricing is the so-called “deep pocket” issue
• An incumbent deters potential entrants/drives new entrants out of the market by
engaging in a price war that generates losses for all the players
• Incumbents have a “deep pocket” (many resources): they can face these losses
• Potential/new entrants have a “small pocket” (they have limited resources) and
can hardly survive such losses for a long time

Predation is a challenging strategy

1. Larger market shares mean more significant losses
2. New entrants can resort to external financing to recover from losses
• The minority view of the Chicago School: predatory prices do not exist
 New entrants can always resist predation by resorting to external financing
 Incumbents anticipate this and do not resort to predation

3. Usually, predation is easier (and thus more common) in the presence of information
asymmetries and uncertainty
• The incumbent can leverage these drawbacks to convince (potential) entrants that
there is no room for gaining (high) profits

Cristina Rossi Lamastra

 62

E.3.3. The theory of contestable

markets

Cristina Rossi Lamastra

The theory of contestable markets: 63
How it works
Contestable markets are industries in which
“An entrant has access to all production techniques available to the incumbents, is
not prohibited from wooing the incumbent’s customers, and entry decisions can be
reversed without cost” (Baumol, Panzar, and Willig, 1982)

Features of a contestable market are

1. No entry barriers for firms → Free entry
2. No switching costs for customers → Elastic demand
3. No sunk costs, also because of well-functioning secondary markets for physical
capital → Free exit

The theory of contestable markets generalizes perfect competition theory →

Competition depends on entry threats, not only on actual competitors
Essential notion → “hit-and-run” strategy: a firm
1. Envisaging positive profits (as p>AC) decides to enter the industry
2. Undercuts its price
3. Makes short-term positive profits
4. Exits the industry at no cost before incumbents’ reaction
Provide an example of a contestable market (one short sentence). Respond at:
https://pollev.com/cristinaross747
Cristina Rossi Lamastra
Contestable markets and efficiency 64

Engaging in a “hit-and-run” competition is a credible threat if

• Consumers can easily switch to the new entrant
• Incumbents react slowly, and new entrants can exit quickly

Conditions for contestable markets push industries towards efficiency → To

deter entry; the incumbent should set price at the minimum average costs
• If “hit-and-run” competition is possible, in equilibrium, no firm makes
positive profits → Perfect competition outcome
• Firms are forced to produce at the minimum average cost
• Inefficient firms are driven out of the industry

The theory of contestable markets significantly departs from mainstream

competition theory
• Mainstream theory: positive relationship between the number of firms in the
industry and social welfare → Social welfare increases as the number of firms
increases
• Contestable market theory: monopolistic and oligopolistic industries may
assure high social welfare → Potential competition ensures productive and
allocative efficiency
Cristina Rossi Lamastra
 65

E.4. COMPETITION IN
NETWORK AND PLATFORM
INDUSTRIES

Cristina Rossi Lamastra

 66

E.4.1. Competition in network

industries

Cristina Rossi Lamastra

Definition: 67
Network goods
Firms offering network goods and thus operating in network industries face peculiar
competitive dynamics
Network goods are goods whose value increases with the number of their current (and
future) users; namely, they have both
• An intrinsic value, i.e., the value of the good per se
• A synchronization value, i.e., the value depending on the number of users

Network goods are subject to network externalities → Every user increases the value of
the good, thus causing a positive externality on the others
1. Direct network externalities: the value of the network good increases automatically as
the number of users increases
•E.g.: the email has no value per se: being the only one to have an email is useless; its
value increases with the number of email users
2. Indirect network externalities: when the number of users increases, the value of the
network good increases as the offering of complementary goods increases
•The hardware-software paradigm, e.g., the value of a game console increases with the
number of compatible games
1. The more the consoles’ users
2. The more the developers who want to produce compatible games
3. The more those who prefer to buy that console →The more the console’s users

Cristina Rossi Lamastra

 A telling example: 68
A dog and a garbage bin (1)
A funny story to explain competition in network industries

Six guys want to buy video game consoles, and three firms (A, B, C) produce three
console models. The choices of the six guys depend on
1. The characteristics of the console (intrinsic value)
2. How many other guys own the same console (synchronization value) so they can play
with friends, exchange videogames, and so on

The utility function describing the value of a console is Ui = Xk + w*Nk

•Xk is the intrinsic value of console k (with k=A, B, C)
•w*Nk is the synchronization value of console k
• Nk = number of users owning the console k
• w = parameter that accounts for the value of other users (e.g., w=0.2); it measures
the strength of network externalities

•During the weekend, the six guys browse the catalog, reporting consoles’
characteristics and, based on them, decide which console they prefer
•On Sunday evening, their preferences depend only on the intrinsic value of each model;
there is no synchronization value (and no network externality) as none has yet bought
the first console
Cristina Rossi Lamastra
 A telling example: 69
A dog and a garbage bin (2)
Suppose that the intrinsic values, which define preferences, are in table 1
Table 1 A B C Preferred With no network
console externalities,
Alan 0.4 0.6 0.5 B firms producing
Bud 0.6 0.5 0.4 A videogame
consoles share
Charlie 0.5 0.4 0.6 C
the market
David 0.4 0.6 0.5 B
Eliah 0.6 0.5 0.4 A
Frank 0.2 0.3 0.4 C

•On Monday at 6 a.m., Charlie’s dog was starving and turned the garbage bin
upside down
•Charlie wakes up suddenly, while the other guys are still sleeping, and thinks,
“Since I am awake, I will go early to the videogame shop and buy my console”
•At 8 a.m., he goes to the shop and buys console C, according to his
preferences
•As Charlie makes his choice, network externalities emerge, and the
preferences of other guys, whom the shop assistant informs, change
Cristina Rossi Lamastra
 A telling example: 70
A dog and a garbage bin (3)
After Charlie’s choice, other guys’ preferences change as the value of console C for
them is now Uc=Xc+ 0.2 (w*1). The preferences are those in table 2
Table 2 A B C Preferred consoles Alan, David, and Frank
after Charlie’s choice
now prefer C and will buy
Alan 0.4 0.6 0.5+0.2*1=0.7 C it
Bud 0.6 0.5 0.4+02*1=0.6 A/C
Charlie C
David 0.4 0.6 0.5+0.2*1=0.7 C
Eliah 0.6 0.5 0.4+0.2*1=0.6 A/C
Table 3 reports the
Frank 0.5 0.4 0.4+0.2*1=0.6 C preferences of the other
Table 3 A B C Preferred consoles guys after the choices by
after choice by Alan, David, Alan, David, and Frank
and Frank Console C has won the
Alan C market
Bud 0.6 0.5 0.4+0.2*4=1.2 C
Charlie C
David C
Eliah 0.6 0.5 0.4+0.2*4=1.2 C
Franl C
Cristina Rossi Lamastra
 Competition in network industries: 71
What the story of the dog and a garbage bin tells us
This short story tells us that in the case of network industries
1. Small initial events, which happen by chance, can determine the fate of the
competition
2. The first who benefits from network externalities can win the market
3. Competition can assume the form of “the-winner-takes-it-all”
• Once the number of users of a network good has reached the critical
mass, any new user wants to buy that good
• The good conquers the market

Critical mass
• In nuclear engineering: the minimum amount of uranium needed to cause
a self-sustaining nuclear reaction
• In the case of network goods, for any price, the number of users (N) such
that
• Number of users > N: the network succeeds as each user attracts
further users
• Number of users < N: the network fails as users progressively leave
the network
Cristina Rossi Lamastra
 Competition in network industries: 72
How critical mass works

Suppose that a potential user U is willing to pay 1€ for a network good only if
there are other 100 users in the network

•If, when the price is 1€, there are other 100 users, U buys the network good: the
network now consists of 101 users

•A potential user willing to pay 1€ only if there are other 101 users will join the
network, followed by a user willing to pay 1€ only if there are other 102 users,
followed by a user willing to pay 1€ only if there are other 103 users, and so on

The network grows larger and larger, and the good wins the market

•If,
when the price is 1€, there are 99 users, U does not buy the network good, and
the network still consists of 99 users

•If (for whatever reason) a user leaves the network, users become 98

•One of these users may be willing to pay 1€ only if there are other 99 users, this
user leaves the network, users will be 97, and so on

The network becomes smaller and smaller, and the good fails
Cristina Rossi Lamastra
 Competition in network industries: 73
Growth of the number of users
U = number of users
Maximum number of users
UN
Uh
Critical mass

Ul

Time
tl th tN
•Phase 1: start-up phase, before the critical mass is reached (0 → Ul)
•Phase 2: rapid growth after the critical mass is reached (Ul → Uh)
•Phase 3: maturity, growth slows down (Uh → UN)
Cristina Rossi Lamastra
 Competition in network industries: 74
The critical mass and the start-up problem

When Facebook and Twitter, which are network goods, reached the critical mass

The start-up problem

•At the offset of the market, none is using network goods; there is no network (and no
network externality)
•The focal firm must convince users to buy its good
•By joining its network, users generate network externalities and attract further users
since the critical mass is reached
•If, instead, users buy the network good of another firm, this good may reach the critical mass,
while the focal good fails irrespective of its price (and quality)
Cristina Rossi Lamastra
 Competition in network industries: 75
How to solve the start-up problem
Firms can solve the start-up problem and reach the critical mass by
1. Receiving support from the government
•E.g., Internet in the US (1960), Minitel in France (1980), broadband in Italy (2000)

2. Charging meager prices to attract users from the very start; a firm can even set the
price to zero
•E.g., free access to social networks (e.g., Facebook)
•Why do software firms (e.g., Microsoft) sometimes tolerate piracy (one short sentence)?
Respond at: https://pollev.com/cristinaross747

3. Breeding users’ expectations that the network will grow large

•E.g., bearing sunk costs to show commitment to winning the market

4. Offering complementary products/services

•E.g., Apple is investing in Apps to promote the use of iOS

5. Establishing collaborations to reach the critical mass

•E.g., by making two network goods compatible
Cristina Rossi Lamastra
 Competition in network industries:
76
Switching costs and lock-in (1)

Once a network good A has reached the critical mass, the costs of choosing
another network good B and joining its network (i.e., switching costs)
become very high
• The market is locked in A irrespectively of its intrinsic value

Switching costs increase, and thus the probability of lock-in is higher,

depending on
1. Strength of the direct network externalities, i.e., the size of the current
network and the expected size of the future network
2. Strength of the indirect network externalities, in terms of availability of
complementary goods
3. The (relative) quality of the network good
4. The (relative) price of the network good
5. The learning costs: how much time and effort users spend to learn how to
use a good
•E.g., the time spent in learning to use an operating system

Cristina Rossi Lamastra

Competition in network industries:
77
Switching costs and lock-in (2)
Network externalities and switching costs explain market shares in the industry of operating
systems (desktop segment and mobile segment)

OSS advocates
claim that Linux
has better
performances

The case of the QWERTY keyboard

•Keyboards need trained typists, which generate indirect network externalities
•Today, QWERTY is the most diffused keyboard; it won the competition with the DVORAK
keyboard, which was claimed to perform better in terms of typing speed
•Event triggering QWERTY use: in early 1900, a typist won a competition on QWERTY
•Typists started to train on QWERTY, and QWERTY won the market due to indirect network
externalities that QWERTY-trained typists generate
Cristina Rossi Lamastra
 78

E.4.2. Competition in platform

industries

Cristina Rossi Lamastra

Definition: 79
Platform businesses
Platform businesses are firms that create value by operating platforms
• Platforms create value by enabling transactions between two (or more) groups of
participants who, in some sense, need each other
• Uber creates value by connecting drivers and passengers (share value: $57.93)
• Airbnb creates value by connecting tenants and people needing rooms (share value: $
174.25)

Platforms shift value creation from the transformation of inputs into outputs to the
coordination of participants from diverse groups

1. Platforms’ value for (prospective) users depends on the size of these groups
• Cross-side network externalities: the value of a platform for a participant of one group
increases as participants of the other group increase
• E.g., Airbnb’s value for renters rises as tenants increase
• Same-side network externalities: the value of the platform for a participant of one
group increases as participants of the same group increase
• E.g., Airbnb’s value for renters increases as renters increase due to an increase in
feedback
• They may also be harmful, e.g., competition between buyers and sellers on eBay
Provide an example of negative same-side network externalities (one short sentence).
Respond at: https://pollev.com/cristinaross747
Cristina Rossi Lamastra
The challenges of platform businesses: 80
Balanced growth and the chicken-egg problem
2. Platforms are “asset-light”: they need few assets to create value → They
benefit from economies of scale
• What is crucial for platform businesses is the interface, which allows
connecting participants from diverse groups
• Fixed costs are mainly related to the design of the interface
• Marginal costs per transaction are (very) low

Because of 1 and 2 platforms

• Face the so-called chicken-egg problem: at the offset of a platform, there
are no participants on one side, and thus it is not easy to attract
participants on the other side and vice-versa
• The challenge is to bring “both sides on board”
• Succeed in the market if they achieve a “balanced growth,” namely,
participants on both sides grow steadily
• Too few participants on both sides are problematic
• Too many participants on just one side are problematic alike
• E.g., Uber creates limited value if there are few drivers and few
passengers, but also if there are many drivers and few passengers and
vice-versa
Cristina Rossi Lamastra
Platform businesses: 81
Pricing strategies
To solve the chicken-egg problem and achieve balance growth, platforms
adopt both pricing and non-pricing strategies

PRICING STRATEGIES
Platform businesses usually charge a price menu, which consists of

1. Access (or subscription or entry) fees: the price paid for accessing the
platform
• Usually, EU crowdfunding platforms (CFPs) charge a subscription fee to
entrepreneurs who want to post their entrepreneurial projects on the platform
• Access fees for project funders are 0 in the large majority of the cases

2. Transaction (or usage) fees: the price paid for each transaction on the
platform
• Usually, CFPs retain a % of the capital raised by project proponents
• In the case of CFPs, usage fees for project proponents are positive, while
usage fees for project funders are zero in the large majority of the cases
• In the case of CFPs adopting the all-or-nothing approach, transaction fees
assume the form of success fees

Cristina Rossi Lamastra

Pricing strategies: 82
How they can solve platforms’ challenges
Pricing strategies allow for solving the chicken-egg problem and achieving a
balanced growth

To this end, platforms usually adopt a “divide-and-conquer strategy”

• They charge low prices to participants of the side, which has the more
elastic demand, to bring them on board
• The presence of participants on this side attracts participants on the other
side, which are thus willing to pay a higher price
• Platforms recover the (alleged) losses on the one side with the gains on
the other side

For instance, the divide-and-conquer strategy is used by

• CFPs, which in most cases charge no access and usage fees to funders
• Funders are more sensitive to price than proponents as funders can fund
projects on diverse platforms (→ Multi-homing)
• Dating platforms, which charge no access (and lower transaction) fees to
women, who are more reluctant than men to use dating platforms
• Social networks (e.g., Facebook), which charge no fee to private users and
charge favorable fees to business users

Cristina Rossi Lamastra

Platform businesses: 83
Non-pricing strategies
Platforms also resort to a wide array of non-pricing strategies
It is possible to group these strategies into the following classes
1. Accessibility. It deals with rules regulating access to the platforms, e.g.,
• Using one language vs. multiple languages in the case of CFPs
• Requiring to use of specific development tools in case of videogame platforms

2. Specialization. It deals with the span of the offering of the platform, e.g.,
• Hosting just music projects or any project in case of CFPs
• Focusing just on wealthy people or any people in the case of dating platform

3. Bundling. It deals with expanding the platform’s functionalities to incorporate

functionalities of other platforms, e.g., Amazon incorporating video streaming like
Netflix

4. Quality. It deals with the quality of the platform itself, e.g.

• A selling platform can be visually appealing like eBay or have a text-only
interface like Craiglist

Pricing and non-pricing strategies interact to determine the balanced growth

Cristina Rossi Lamastra