Transcript 0-83 - CEA

Kilpailu- ja yhtiöoikeuden taloustiede
• Teacher: Markku Stenborg, PhD (Penn State)
– Research Fellow, ETLA, http://www.etla.fi/
• Innovation, regulation, and the changing terms of
competition in wireless telecommunications funded by
Nokia and Tekes, http://brie-etla.org/
• Economics of OSS
– Senior Research Scientist, HIIT, http://www.hiit.fi
• Managing Privacy and Trust in Mobile P2P
– Consultant at CEA
– Previously
• Assistant Prof with Turku Business School
• Senior Adviser at Finnish Competition Authority
• Senior Manager at KPMG Transaction Services
– email: [email protected]
– During this week: room M 12 (3rd floor, “Optimi”)
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
0
Joensuu 2004
Kilpailu- ja yhtiöoikeuden taloustiede
• Course homepage: http://www.cea.fi/joe.htm
• This course covers theoretical and empirical issues related to
Economics of strategic competition and competition policy:
– Price and non-price competition
– Strategies to affect competition
– Market delineation
– Dominance
– Mergers
– Cartels and coordination of market conduct
– Vertical restrictions
• Focus more on Economics, less on Law
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
1
Joensuu 2004
Kilpailu- ja yhtiöoikeuden taloustiede
• No textbook, but we will look at some papers and EU and Finnish
notices and cases
• Some useful reading:
– Aalto-Setälä et al. (2003) Kilpailulait ja laki julkisista
hankinnoista, 3rd ed. (or 2nd ed., 2001)
– Besanko, Dranove,& Shanley (2000) Economics of Strategy
– Cabral (2000) Introduction to Industrial Organization
– Carlton & Perloff (2000) Modern Industrial Organization
– Church & Ware (2000) Industrial Organization: Strategic
Approach
– Motta (2004) Competition Policy: Theory and Practice
– Whinston (2003) Lectures on Antitrust Economics, draft at
http://www.csio.econ.nwu.edu/
• I assume you have grasp of basic Economics concepts such as
demand, marginal benefit and cost, supply, efficiency, surplus, …
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
2
Joensuu 2004
Kilpailu- ja yhtiöoikeuden taloustiede
Luennot
• ma 13.9. klo 16-20, sali B6
• ti 14.9. klo 10-14, sali MA155
• ke 15.9. klo 14-18, sali MA155
• to 16.9. klo 10-14, sali B6
• pe 17.9. klo 10-14, sali MA155
Final exams
• Ke 29.9. klo 12-16, sali K4
• Ti 12.10. klo 14-18, sali K4
• 2+2 questions, 1+1 answers
• One case, one more technical question
• Two straightforward explanations
• Prize: best student receives Aalto-Setälä et al. Kilpailulait ja laki
julkisista hankinnoista, 3rd ed, Tietosanoma 2003.
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
3
Joensuu 2004
Course Outline
1. History and objectives of competition policy
• Introduction and objectives for competition policy
– Suggested reading: Motta, Ch 1, Kovacic and Shapiro (1999)
“Antitrust Policy: A Century of Economic and Legal Thinking”,
UC Berkeley, Working Paper No. CPC99-09
2. Market power and welfare
• Market power, allocative and productive efficiency
• Competition policy and innovation
• Market power and entry
– Aalto-Setälä et al. Ch 8.1-2
3. Market delineation and market power
• Product and geographic market definition
• How to measure market power
– Aalto-Setälä Ch 7; US Merger Guidelines,
http://www.usdoj.gov/atr/public/guidelines/hmg.htm, Section
1 "Market Definition"
– Case: Commission's Volvo/Scania decision
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
4
Joensuu 2004
Course Outline
4.
•
•
•
•
5.
•
•
•
Oligopoly, cartels and tacit collusion
How oligopolists compete
What is collusion
Factors that facilitate collusion
Ex-ante and ex-post measures to fight collusion
– Aalto-Setälä Ch 5.1-2; Stenborg (2004) “Forest for the Trees:
Economics of Joint Dominance”; Europe Economics (2001)
“Distinguishing between Competitive and Dominant
Oligopolies in Merger Control”
Horizontal mergers
Incentives to merge
Competitive and welfare effects of mergers
Which variables matter? How to deal with merger cases?
– Aalto-Setälä Ch 15; Epstein and Rubinfeld (2001), “Merger
Simulation: A Simplified Approach with New Applications”,
Antitrust Law Journal;
– Case: Commission's UPM/Haindl and Volvo/Scania decisions
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
5
Joensuu 2004
Course Outline
6.
•
•
•
•
Vertical restraints
Vertical externalities: double marginalization
Vertical restraints to internalize externalities
Welfare effects of vertical restraints
Foreclosure: exclusive dealing and other instruments
– Aalto-Setälä Ch 9.1-2; Dobson & Waterson (1996) “Vertical
Restraints and Competition Policy”, OFT Research Paper 12
– Case:
7. Predatory practices
• Predatory prices: long-purse, reputation, financial market effects
• Tests of anti-competitive behavior
– Aalto-Setälä Ch 8.1-2, 8.4.7;
– Grout (2001) “Recent Developments in the Definition of
Abusive Pricing in European Competition Policy”,
http://www.bris.ac.uk/Depts/CMPO/workingpapers/wp23.pdf
– Case:
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
6
Joensuu 2004
1. Intro: Competition Law and Policy
• Why do we need competition law and policy?
– Economic competition is self-steering process that guides
production, distribution, pricing, etc. decisions
– Competition is an efficient way to organize many activities in
society
• Efficient ≈ Pareto optimal: maximize well-being or surplus
generated by production and exchange, from assets
possessed in society
– Allocative efficiency
– Productive or X-efficiency
– Dynamic efficiency
– Market power and restraints on competition
• reduce efficiency and/or
• restrict and disturb self-guiding process on markets
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
7
Joensuu 2004
1. Intro: Competition Law and Policy
Goals of competition laws
• Promote efficiency?
– Obviously, …
– … but with nonprice competition simple formulas for efficiency
(consumer surplus + producer surplus) are deceptive and
misleading
• What is efficiency?
– Do not strive for perfect competition but promote ”workable
competition”
– With non-price competition, consumer welfare becomes multidimensional
• Customers have preferences over quality, speed and
security of supply, introduction of new products and
services, etc.
• These may not be measurable
• And even if they are measurable, value judgments are
necessary for efficiency analysis
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
8
Joensuu 2004
1. Intro: Competition Law and Policy
• Protect economic freedom and opportunity by promoting
competition, so that competition can create
– lower prices
– better quality
– greater choice
– more innovation
• Sometimes competition laws have also other goals:
– In EU, competition laws are used to promote single market
within EU
– Competition laws also protect SMEs in some cases
– These other goals can conflict with the main goal of protecting
economic freedom and opportunity
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
9
Joensuu 2004
1.1 Competition Laws
1. Restrictions on Competition
• Article 81(1) of EU Treaty states that
”agreements between undertakings, decisions by associations
of undertakings and concerted practices […] which have as
their object or effect the prevention, restriction or distortion
of competition within the Common Market” [shall be
prohibited]
– Article 81 covers much more than formal cartel arrangements
– Not only collusion, but also many beneficial forms of
horizontal and vertical cooperation are prohibited
• Browse (http://www.europa.eu.int/comm/competition/):
– Guidelines on the applicability of Article 81 of the EC Treaty to
horizontal cooperation agreements
– The Competition rules for supply and distribution agreements
– Guidelines on Vertical Restraints
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
10
Joensuu 2004
1.1 Competition Laws
2. Abuse of Market Power
• Article 82 states that
”Any abuse […] of a dominant position […] shall be prohibited
[…]. Such abuse may, in particular, consist in:
imposing unfair purchase or selling prices or other unfair
trading conditions;
limiting production, markets or technical development to the
prejudice of consumers;
applying dissimilar conditions to equivalent transactions with
other trading parties, thereby placing them at a competitive
disadvantage;
making the conclusion of contracts subject to acceptance by
the other parties of supplementary obligations which, by their
nature or according to commercial usage, have no connection
with the subject of such contracts.”
• Read “määräävän markkina-aseman väärinkäyttö” at
http://www.kilpailuvirasto.fi/
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
11
Joensuu 2004
1.1 Competition Laws
Per se and rule of reason
• Per se: conduct is prohibited if it fulfills the legal test regardless
of other issues
• Rule of reason: conduct is prohibited if its negative consequences
outweigh the positive
• Articles 81 and 82 of EU Treaty seem to be per se
• But to prove that firm has abused its dominant position,
authorities must
– show that the firm has dominant position
– conduct was abusive
– In practice, Article 82 has flavor of rule of reason analysis
• Article 81(1) does not apply to insignificant restrictions
• Article 81(3) and Commission Notices exempt various restrictions
– Article 81(1) also has flavor of rule of reason
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
12
Joensuu 2004
1.1 Competition Laws
• In some legal systems, many vertical restraints are dealt with
rule of reason
– Eg. previous Finnish law
• A priori, effects of vertical restraints to competition and efficiency
are ambiguous
– Many vertical restraints are solutions to problems, not
problems for competition
• Vertical restraints can align private incentives in supply
and distribution
• Double marginalization (two vertical monopolies)
– Hence the Block exemptions
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
13
Joensuu 2004
1.1 Competition Laws
3. Merger review
• Transactions that lead to increase in market power or to some
other competition problems may be prohibited
– Illegal to monopolize markets by M&As
• In EU
”A concentration which would significantly impede effective
competition […] in particular as a result of the creation or
strengthening of a dominant position, shall be declared
incompatible with the common market.”
– EU previously had pure dominance-test
• US: ”the effect of such acquisition may be substantially to lessen
competition, or to tend to create a monopoly”
– Usually, SLC-test poses lower threshold for intervention
– Monopoly is US legal jargon  dominant position  monopoly
in Economics textbooks
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
14
Joensuu 2004
1.2 Market Power in Case Law
• Read Motta, Ch 2 & 3, and Volvo/Scania decision, Market
Definition section pp 5-21
Assessment of market power in abuse and merger cases
1. Define relevant antitrust markets
2. Evaluate market power within the relevant markets
• Relevant markets are defined basically by demand substitution
– Only those goods that provide immediate and intense
competitive constraints to each other belong to the same
relevant market
– In some instances, also supply substitution and entry by
potential competitors are taken into account in market
delineation
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
15
Joensuu 2004
1.2 Market Power in Case Law
• Market power on relevant markets is analyzed:
– calculate market shares
– analyze competitive strengths of firms
– evaluating degree of actual competitive pressure firm faces
– entry barriers and other supply substitution
• In abuse cases, analyze whether conduct of dominant firm was
misuse of market power
– In EU, dominant firms have special obligations
– Dominant firms cannot use their market power to impair
conditions of competition
– Idea is to protect competition, not competitors
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
16
Joensuu 2004
2. Economics of Market Definition
• Read EU and US Guidelines on relevant markets
– http://www.europa.eu.int/comm/competition/antitrust/relev
ma_en.html
– http://www.usdoj.gov/atr/public/guidelines/hmg.htm
• Why we need to define relevant antitrust markets in case law?
– To calculate meaningful market shares
– Market shares tell us something about market power
• Market shares do not need to imply or correlate with
market power
– More on this in Oligopoly and Merger sections
– Identify main competitors and competitive constraints
– We are interested in market definition only to extent it helps
in analyzing market power
– Sometimes we can identify and measure market power w/o
defining markets
• More on this will follow (Merger section)
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
17
Joensuu 2004
2.1 How to define markets? — SSNIP
• On which set of goods market power can be exercised? Which
goods compete immediately with each other
– Market for cars? Separate markets for minivans, luxury
sedans, large family cars, compacts, subcompacts…?
• Relevant antitrust market is something that can be monopolized
– If it cannot be monopolized, it is too narrow
– Then important competitive pressures are left out of
candidate market
• Test: Small but Significant Non-temporary Increase in Price
– Take a small set of substitute goods and a geographic area
– Assume all goods produced by hypothetical monopoly
– Incentive to permanently increase prices by 5-10 %?
– Yes: candidate market = relevant market
• Proceed to analyze market power etc
– No: candidate market < relevant market
• Include more goods and repeat
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
18
Joensuu 2004
2.1 How to define markets? — SSNIP
• Logic: Goods on relevant market create intense competition to
each other
– Once this competition is removed, incentive to increase price
– If strong competition remains, price increase is not possible,
and goods do not constitute relevant market
• Leave out significant constraints on market power,
candidate market is too small
• Keep in firms and products that are not significant
constraints, market is too large
• Price increase leads to
– Consumers substitute away
– Outside producers increase output or enter
• SSNIP asks: how much demand shifts away for a price increase
and does this make price increase non-profitable?
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
19
Joensuu 2004
2.1 How to define markets? — SSNIP
• SSNIP in Economics jargon: What is demand elasticity for this set
of goods?
– Basically: elastic demand  market too narrow, inelastic
demand  relevant antitrust market
– But recall effects from costs and supply subsitituition
Digression on demand
• Individual demand is ultimately derived from customer
preferences
– Hold everything else constant and vary the price of the good
 customer’s demand curve
– Demand as function of own price vs shifts in demand function
– Sum up all customer demand’s  market demand
• Note: economics textbooks (sort of) assume relevant
market has been defined when discussing demand
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
20
Joensuu 2004
2.1 How to define markets? — SSNIP
• Market demand vs firm demand
– Demand for colas vs demand for Coca-Cola, say?
– What happens after ”market-wide” price increase?
• Marshallian demand based on ceteris paribus assumption and
measures effect of price change by keeping all other prices
constant
– Merger Guidelines assume that “the terms of sale of all other
products are held constant” = Marshallian demand
– Direct demands are hard to estimate
• Suppose condidate for relevant market has n goods and
their demand depend on each others’ prices
• Need to estimate at least n2 parameters to get any info
on Marshallian demand
• Price change in some goods do not leave all other prices
constant
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
21
Joensuu 2004
2.1 How to define markets? — SSNIP
• Residual demand curve is the demand curve faced by an
individual firm
– Residual demand = total market demand curve - supply of all
other firms in market
qi = Q(p) - qj
– Residual demand curve incorporates effects of changes in
prices of other products in response to changes in this
product’s price
– Residual demand is good tool for market definition to as it is
relatively easy to estimate from data available
• We do not observe demand curves, but price-output
pairs, determined jointly in equilibrium
• Can one identify demand and supply?
• Need econometrics to get from data to demand curve
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
22
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
• SSNIP-test should be applied by estimating own elasticity of
demand
• What value of elasticity is large enough for concluding that
given set of goods comprise relevant market?
• Notation:
P0 = Current observed price
P1 = P0 plus some specified price increase t
C = Marginal cost
L = (P – C)/P price-cost margin or Lerner-index:
T = Price increase deemed significant (eg. 0.05 or 0.1)
T = (P1 – P0)/P0
dQ P
dQ / Q

dP Q
dP / P
elasticity of demand
e(P ) 
• Assume C is constant, profits are then (P-C)Q - F
• For profitable price increase, profits with higher price must at
least equal profits from selling more at lower price
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
23
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
• Break-even condition is Q(P0)(P0 – C) = Q(P1)(P1 – C)
where P1 = break-even price
• Rearranging: Q(P1)/Q(P0) = (P0 – C)/(P1 – C)
• Using definitions of T and L:
(P 0 - C ) / P 0
(P1 - C ) / P1

L
LT
• For linear demand Q = (A - P)/B:
Q(P1 )
0
Q(P )

A  P1
AP
0
1
P1  P 0
P0
P0
A  P0
– Recall elasticity of demand here is e(P0) = P0/(A - P0), which
gives Q(P1)/Q(P0) = 1 - Le(P0)
• Break-even requires Q(P1)/Q(P0) = L/(L+T) so this gives us
L/(M+T) = 1 – Te(P0), and solving gives us critical elasticity
e(P0) = 1/(L+T)
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
24
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
•
•
•
•
•
– When demand is isoelastic, break-even elasticity is
e(P0) = [log(L+T) – log(L)]/log(1+T)
Critical sales loss for a price increase = proportionate decrease in
quantity sold as a result of the price increase large enough to
make price increase unprofitable
Sales loss resulting from price P0  P1 is
1 - Q(P1)/Q(P0)
For linear demand Q = (A–P)/B we can write this as
1 – Q(P1)/Q(P0) = 1 – (A – P1)/(A – P0)
= [(P1 – P0)/P0][P0/(A – P0)] = Te(P0)
Applying break-even value of e(P0) derived above gives value for
break even critical sales loss Y = T/(L+T)
If actual sales-loss after price increase is less than Y, it is
profitable to increase price
– The break-even value of the critical sales loss is the same for
both linear and isoelastic demand curves
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
25
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
• Relationship between market power index L, and critical e and Y
for 5 % price increase:
L%
e
Y%
50
1.82
9.1
40
2.22
11.1
30
2.86
14.3
20
4.00
20.0
10
6.67
33.3
Cross-Price Elasticity
• Sometimes in case law market definition is based on cross-price
elasticity of demand
• Cross price elasticity of demand eij = (dQi/dPj)/(Qi/Pj), where Qi
and Pi denote the quantity and price of products i and j
• Cross price elasticity = How demand for good i reacts to price
increase of good j?
• Sounds like nice idea to delineate markets: goods belong to
same relevant market if they are good-enough substitutes
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
26
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
• Rule used in some competition cases:
– Goods i and j are on same market if cross-price elasticity is
large enough and otherwise are on different markets
• Cross price elasticity is not a good measure for market
delineation
– Market delineation is not question of how much demand will
flow from i to j as Pi increases
• Cross-price elasticity is usually not symmetric, eij  eji
– eij: ”i and j on same mkt” and eij: ”i and j on different mkt” is
possible
• Even if a cross price elasticity is small, market need not be
narrow, as there may be many other goods that restrict the
market power of hypothetical monopolist
– If there are many substitutes, price increase will divert
demand to many goods  cross price elasticity must be small
– This indicates competition, not sepatate markets
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
27
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
• Price and cross price elasticities are connected:
– Own-price elasticity = 1 + weighted average of all cross-price
elasticities
• Weight: share of income in relation to share of income of
the good in question
”Cellophane Fallacy”
• US Supreme Court: high cross price elasticy between cellophane
and paper wrapping  relevant market wider than cellophane 
Du Pont not dominant
– Also SSNIP test ignores fact that firm may already have
market power
• Firm with market power wants to increase price to level where
competition constrains start to bite
– Demand usually turns more elastic as price increases
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
28
Joensuu 2004
2.2 Critical Elasticity of Demand and Loss
• Then goods actually outside of relevant market seem to be
substitutes
• High cross price elasticy indication of use of market power, not
an indication of wide market
• Cellophane fallacy means that different approach is required in
abuse of dominance cases
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
29
Joensuu 2004
3. Market Power
• Market power = ability to profitably charge P > MC
• Some sources of market power
– Only few firms active in the market
– Products are differentiated, and some customers prefer one
firm’s product to other
• Firm that attempts to ”steal” customers from its
competitor must reduce price a lot
• Then firms have less incentives to lower their prices
– Capacity constraints
• Firm have less incentive to win more customers, giving
other firms incventives to charge higher prices
– Customers are not informed of all firms’ prices
• Incentive to lower price is reduced
– Switching costs
• Each firm monopolizes its customer base
– Cartel or collusion (later in Oligopoly section)
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
30
Joensuu 2004
3. Market Power
• Market power is source of inefficiency
– Allocative inefficiency
• Harberger triangle
• Rent seeking
– X-inefficiency
• Less need to control costs or to concentrate on key
capabilities
• Less need to provide value to customers
– Dynamic inefficiency
• Less incentive to innovate
• Market power allows restrictions on competition
– Entry deterrence
– Predation
– Price squeeze
– Cartel or collusion
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
31
Joensuu 2004
3. Market Power
Monopoly’s profit maximization
• Assuming constant MC, monopoly’s profits are
 = [P(Q) - C] Q - F
• To maximize profits, set d/dQ = 0:
d/dQ = P(Q) + Q dP/dQ - C = 0 (MR - MC = 0) 
P*(Q) - C = -Q dP/dQ
• Divide both sides by P*
– (P* - C)/P* = -(Q/P*)(dP/dQ)
• Rewrite this as L = 1/e
– L = Lerner Index
– e = elasticity of market demand
• Under perfect competition or perfectly elastic demand:
P = C, hence L = 0
• L is usefull measure of market power: 0 ≤ L ≤ 1/e
• If we can estimate P, e and C, we can estimate market power L
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
32
Joensuu 2004
3.1 Measuring Market Power
• To measure market power, we need to estimate demand
consistently
• Basic idea: how well other goods substitute for goods produced
by firm i and constrain its market power?
• Answer: elasticity of residual demand
– Residual demand does not tell who or what constrains market
power
• Straightforward approach is to specify system of demand
equations q = D(p;r), where q is vector of quantities demanded,
p is a vector of prices, and r is a vector of exogenous variables
that shift demand
– Need to define D(.) in a way that is both flexible and
consistent with economic theory
• Problem 1: Number of parameters estimated increases with
square number of products
– 10 firms, each with 20 brands  40 000 elasticities
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
33
Joensuu 2004
3.1 Measuring Market Power
• Problem 2: Simultaneity
– Equilibrium price and quantity determined jointly by demand
and supply schedule
– Price increase by i is followed by rivals  need to take into
account how rivals react  residual, not market demand
• Problem 3: Simple approach ignores consumer heterogeneity
• Solutions to problems include
– 1) assume problems away
– 2) assume symmetric representative consumer
– 3) assume multi-stage budgeting
– 4) use discrete choice/address models
1. Avoid problem
• Focus on aggregate demand
– All eastbound rail traffic, not differentiated across cities
• Focus on narrowly defined product
– Self service 87 octane
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
34
Joensuu 2004
3.1 Measuring Market Power
• Focus only on sub-markets
– Particular segment in beer industry
• This is enough in some cases
2. Symmetric representative consumer
• Use Constant Elasticity of Substitution (CES) utility function
– Dimensionality problem is solved by imposing symmetry
between products
– Estimation involves a single parameter, regardless of number
of products, and can be achieved using simple econometrics
• Cross-price elasticities are restricted to be equal, regardless of
how “close” the products are
– Are MB and Opel equally good substitues to BMW?
– This restriction can have important implications and in many
cases would lead to the “wrong” conclusions
• For some industries or cases this model of differentiation is
adequate, for most markets this is not
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
35
Joensuu 2004
3.1 Measuring Market Power
3. Separability and multi-stage budgeting
• Divide products into smaller groups and allow for flexible demand
within each group
• Multi-stage budgeting
– Consumer allocates expenditure in stages
• at highest stage expenditure is allocated to broad groups
(food, housing, clothing, transportation…)
• at lower stages group expenditure is allocated to subgroups (cereal, bread, cheese, ..)
• until expenditures are allocated to individual products
– At each stage, allocation decision is function of only that
group’s total expenditure and prices of commodities in that
group (or price indexes for the sub-groupings)
– Then we have cross elasticities between Opel sedan and VW
sedan, between Opel van and VW van, and between sedan
and van categories, but not between Opel sedan and VW van
– Reduces number of parameters to be estimated
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
36
Joensuu 2004
3.1 Measuring Market Power
• Three stage system
– Top level: overall demand for the product (cars or ready-toeat cereal)
– Middle level: demand for different segments (sedan, suv,
stw, minivan; or family, kids, adults cereal)
– Bottom level: brand demand corresponding to competition
between different brands within each segment
4. Discrete Choice Models
• Model products as bundles of characteristics
– sweetness, fiber content, …
– alcohol content, bitterness, ...
– horsepower, length, ...
• Preferences are defined over characteristics space
• Each consumers chooses the product with best characteristics for
her
– use bus if U(bus) > U(car), U(train), U(walk), ...
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
37
Joensuu 2004
3.1 Measuring Market Power
• Discrete choice models yield Logit demands (under some
assumptions)
– prob that n chooses i has logistic distribution
• Dimension of characteristics relevant dimension for empirical
work
• Heterogeneity is modeled and estimated explicitly
• Can be estimated using individual or aggregate market data
Comparison
• Symmetric average consumer models least adequate for
modeling demand for differentiated products
– Problem: all goods are assumed to be equally close, equally
good substitutes
• Logit models widely used because they are simple
• Multi-level model requires a priori segmentation of market into
relatively small groups, which might be hard to define
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
38
Joensuu 2004
3.1 Measuring Market Power
• Typically, multi-stage budgeting models assume all consumers
consume all products
– For broad categories like food and shelter reasonable
– For differentiated products, it is unlikely that all consumers
consume all varieties
• Multi-stage budgeting model is closer to classical estimation
methods and neo-classical theory, and more intuitive to
understand
• Discrete choice models require characteristics of products, are
more technical to use, and rely on distributional assumptions and
functional forms
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
39
Joensuu 2004
3.2 Application
• Nevo (2001) uses discrete choice models succesfully
– Panel of quantities and prices for 25 brands of cereal in 65
U.S. cities over 20 quarters, using scanner data
– Estimate own price and cross-price demand elasticities
– Compute price-cost margins implied by three industry
structures:
• each brand on its own
• actual structure of few multi-product firms
• monopoly or collusion
– Markups implied by current industry structure and imperfect
competition match observed price-cost margins
– High margins due to consumers' willingness to pay for
favorite brand, and to pricing decisions that take into account
substitution between own brands
– Market power entirely due to the firms' ability to maintain
portfolio of differentiated products and influence perceived
product quality through advertising
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
40
Joensuu 2004
4 Oligopoly
• Topics
– Basic models of oligopolistic competition
– How can firms change rules of game to their advantage?
– How can firms avoid intensive rivalry?
– When cartel or implicit collusion is stable?
• Read or review
– Chapter on Oligopoly in any modern Micro or Industrial
Organization textbook, and/or
– Europe Economics report available at
http://europa.eu.int/comm/enterprise/library/libcompetition/libr-competition.html
• I will use game theoretic reasoning and Nash equilibrium, so you
should soon get comfortable with these ideas
– Note: topics following oligopoly (Collusion and Mergers) will
be based on oligopoly theory
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
41
Joensuu 2004
4.1 Cournot or Quantity Competition
• Assumptions
– Market demand: price function of total quantity produced,
p = p(q), eg. p = a – bq
– Assume 2 firms on relevant market denoted by i and j
– Firms produce quantities qi and qj
– Firms have constant marginal costs ci
– No threat of entry
• Profits for firm i = Total Revenue - Total Costs
i = pi(qi,qj) qi - c(qi,qj) = p[(qi+qj)-ci]qi
– Note: i's profit depends on what rival j does, unlike in
monopoly or perfect competition
– Firm faces a problem of strategic interaction or plays a game
• How much will i want to produce?
– Depends on how much i expects j to produce, qje
• How much will j want to produce?
– Depends on how much j expects i to produce, qie
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
42
Joensuu 2004
4.1 Cournot or Quantity Competition
• Note, for each qje, there is an optimal output
qi*(qje) = argmax i(qi,qje)
• qi*(qje) is called i’s reaction function
– Compare with monopoly profit max
Problem
• i needs to put himself on j’s position and try to predict how j will
behave
• j needs to put himself on i’s position and try to predict how i will
behave
• i needs to to put himself on j’s position and try to predict how j
will think how i will behave
• j needs to ... predict how i will think how j will behave
• etc. ad inf.
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
43
Joensuu 2004
4.1 Cournot or Quantity Competition
Solution
• Suppose both i and j know p(q), ci and cj, and also expect that
rival will produce profit-maximizing quantity
qi*(q-ie) = argmax j(qi,q-ie)
• Then i should choose qi* = argmax i(qi,qj*) and j
qj* = argmax j(qj,qi*)
– Each firm chooses its strategy taking rivals equilibrium
strategy as given
– Firm i needs to predict j’s equilibrium production
• Simultaneously but individually
max i = p(qi+qj)qi - cqi
max j = p(qi+qj)qi - cqj
• At a maximum, small change in output should not increase
profits; differentiating each max problem yields
di/dqi = qi(dp/dqi) + p(qi+qj) - ci = 0
dj/dqj = qj(dp/dqj) + p(qi+qj) - cj = 0
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
44
Joensuu 2004
4.1 Cournot or Quantity Competition

These are familiar 1st order conditions MR - MC = 0
 Compare with monopoly profit max
 Plug in p(qi+qj) = a - b(qi+qj) and solve for qi*(qje) and qj*(qie),
you get reaction fns:
(1)
qi*(qj) = (a - ci)/2b - qj/2
(2)
qj*(qi) = (a - cj)/2b - qi/2
Solve simultaneously [eg, insert qj*(qi) from (2) into (1) to
replace qj] to get Cournot-Nash equilibrium quantities
(3)
qi* = (a + cj - 2ci)/3b
(4)
qj* = (a + ci - 2cj)/3b


Note: Each firms is on her reaction function
In equilibrium, no firm has incentive to alter her strategy
choice unilaterally
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
45
Joensuu 2004
4.1 Cournot or Quantity Competition
• Insert then qi* and qj* to demand function to get equilibrium
price p*, and then plug these to profit function to get equilibrium
profits
 Reaction functions (1) and (2) are downward-sloping:
dqi*(qi)/dqj = -1/2 < 0
– This also applies to more general Cournot games
– If j increases her production (eg, due to reduction in marginal
cost cj), i will want to reduce his output
– Lower action by one firm induces higher reaction from rivals
– Note, these are equilibrium reactions
• Strategies qi are here strategic substitutes
• Downward-sloping reaction functions  strategic substitutes
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
46
Joensuu 2004
4.1 Cournot or Quantity Competition
Properties of Cournot-Nash Equil
• Go back to reaction functions (1) and (2), and rewrite as
(5)
p(q) - ci = -qi dp/dqi |:p 
(6)
Li = si/e, where
si = qi/q is i’s market share, q = iqi,
Li = (p - ci)/p is firm i’s mark-up or Lerner Index
e = -p(q)/qp’(q) is elasticity of market demand
(6) is basic Cournot pricing formula
• Compare to monopoly’s profit max condition
• In Cournot-Nash equilibrium, market share determined by
– firm’s relative cost efficiency
• Each firm has limited mkt power:
– i’s marginal revenue MRi is p + qip’, so
– p - MRi = qip’(q) > 0  MR > MC
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
47
Joensuu 2004
4.1 Cournot or Quantity Competition
• Smaller market shares s (or more rivals)  smaller mark-up,
more vigorous competion reduces mark-up
• Greater demand elasticity  larger mark-up, less competitive
equilibrium
• Mark-up is proportional to firm market share
• Market shares are directly related to firms cost-efficiency ci
– Market shares determined by cost-efficiency
• Less efficient firms are able to survive
– sj > 0 even if cj >> min c
• Average industry-wide mark-up i si (p - ci)/p = MU
• In Cournot-Nash equil, MU = i si2/e = HHI/e, where HHI is the
Herfindahl-Hirschman Index
– Market performance negatively related to HHI
• These properties give some basis for competition policy opposing
mergers on oligopolistic markets
– What if competition is not Cournot-type?
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
48
Joensuu 2004
4.2 Bertrand or Price Competition
• In reality, firms choose and compete with prices, not quantities
• Often prices are easier to adjust than quantities
• Who chooses prices in Cournot game?
– Cournot unrealistic model?
• Naive thought: firms select prices as in (6) above: pi* st.
(pi* - ci)/pi* = si/e?
• Bertrand paradox: No
• Model: identical product, mkt demand q = q(p), eg. q = a – bp
• Demand for firm i:
pi > pj  i cannot sell at all, qi = 0
pi = pj  i and j split demand, qi = q(p)/2
pi < pj  i sells total mkt demand, qi = q(p)
– Note: small change in rival’s price causes huge change in
firm’s demand
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
49
Joensuu 2004
4.2 Bertrand or Price Competition
• Suppose cj = ci = c
• If i charges pi > c, j can increase her profits by undercutting i
slightly
• If i charges pi < c, i is making losses but j can guarantee j = 0
by staying out of mkt
 Only equil price can be pi = pj = c
– Duopoly enough for perfect competition!
• Result depends crucially on
– firms able and willing to serve all customers at announced
price
– identical products
– customers have complete information eg on prices  firms
have no bargaining power wrt customers
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
50
Joensuu 2004
4.2 Bertrand or Price Competition
Product Differentation and Price Competition
• Simple example only
• Products are imperfect substitutes, demands are symmetric
qi = a - fpi + gpj
• Assume constant marginal costs ci
• Product differentation is assumed fact, not designed by firms
– g/f measures degree of product differentation (how?)
• Profit for i here
i = (pi - ci)(a - fpi + gpj)
• Bertrand-Nash equilibrium found similarly as above:
– Firm i maximizes profits wrt to strategy variable pi
– Solve for reaction functions
– Find where reaction functions intersect
– Then solve for equilibrium prices, quantities, and profits
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
51
Joensuu 2004
4.2 Bertrand or Price Competition
• Reaction functions slope up
– The higher the price i charges, the higher the price rival j
wants to charge
• Prices are strategic complements
– Higher strategy draws a higher reaction from rivals
• Upward-sloping reaction functions  strategic complements
Capacity Constraints and Price Competition
• Firms first choose capacities q and then select prices p?
• We have a 2-stage game (more on this later)
– In equilibrium, higher price than without capacity constraints
• Intuition
– Limited capacity  business stealing not attractive option
 want to price less aggressively  rivals price less
aggressively
 higher profits
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
52
Joensuu 2004
4.2 Sum Up
• Cournot outcome possible with price competition
– Interpret: Cournot = capacity competition followed by price
competition
• Under some assumptions
• Cournot
– Markets where production desicions in advance, price is
flexible, and storage costs arehigh
– Consistent with empirical evidence
• Bertrand
– More realistic assumptions, less realistic outcome?
– Generalized price competition models more consistent
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
53
Joensuu 2004
4.3 Dynamic Competition
• Simplest way to model dynamic rivalry is to introduce two stages
– Model j’s reactions to strategic moves by firm i
– Firms try change game they are playing
• Idea: Choose a strategy now that affects game you play
tomorrow so that your expected profits increase
• Capacity-Price -model above an example: Smaller capacity now
 reduce ability to compete aggressively in future  draw less
aggressive reactions from rivals  higher profit
Stackelberg Oligopoly
• Stackelberg-Cournot game: Firm i chooses its output first, and j
after i’s choice
• Precommitment by i is relevant, not physical timing of moves
• Solve by backward induction:
– First look at last possible moves of the game
– Then work backward to beginning of game, as in dynamic
programming
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
54
Joensuu 2004
4.3 Dynamic Competition
Last move
• Firm i chooses his capacity first
• When j chooses her capacity, she knows i’s capacity qiS
• j's optimal capacity determined by her reaction function (2)
qj*(qi) = a/2b - qi/2
Penultimate move
• To design good strategy, i must put himself on j’s shoes and try
to think how he would behave were he the last to move
• i chooses qiS to maximize profits, taking as given i’s reaction
function, not equilibrium output as in Cournot game
• i chooses best point from rival’s reaction function
• Plug (2) into i’s profit function (a - b(qi+qj))qi and solve for qiS
• Plug qiS back to (2) and solve for qj*, and then solve for prices
and profits
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
55
Joensuu 2004
4.3 Dynamic Competition
• In Stackelberg game, i’s profits higher and j’s lower than in
Cournot game
• First-mover advantage
• Intuition: Commit to flood the market  induce rival to lower
output  increases your profit
– Equilibrium above “subgame perfect Nash equilibrium”
– Also other Nash equil possible:
• i announces to produce qi s.t. p(qi) < cj if j enters
• This is not be credible: i will not want to undertake
threat should j enter (more on this later)
• Crucial reasons: 1) commitment, 2) strategies substitutes
– In Stackelberg-Bertrand duopoly, there is second mover
advantage
– Once rival has committed to a price, firm has strong
incentives to undercut
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
56
Joensuu 2004
4.4 Modeling Dynamic Competition
• 2 time periods, denoted by 1 and 2
• Firm i can take strategic action k on period 1
– k: Advertising, R&D, product design, …
– Strategic action measured by its cost
– Strategy k is sunk on 2nd period, i cannot revoke it
– k is investment, precommitment
• On period 2, i and j compete
– To concentrate on strategic effects, assume k does not affect
j’s demand or costs directly
• i’s 2nd period profits are i(qi,qj,k)
• i’s 1st period profits are i (qi,qj,k) - k
• k shifts i’s 2nd period profit fn
– Strategic move k alters i’s own incentives to choose later 2nd
period tactics
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
57
Joensuu 2004
4.4 Modeling Dynamic Competition
• To find equilibrium, solve for by backward induction starting from
2nd period game
2nd period
• For any given k, equilibrium again given by
di/dqi = 0
di/dqj = 0
• 2nd period reaction functions qi(qj,k) and qj(qi,k), and optimal
tactics qi*(k) are now functions of k
• Equilibrium profits are i(qi*(k),qj*(k),k)
1st period
• How to choose k?
• Profits are i(qi*(k),qj*(k),k) - k
• To find max profit, differentiate i wrt k to get MR – MC = 0; this
gives
d i dqi + d i dqj + d i 1 = 0
dqi dk dq j dk
dk
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
58
Joensuu 2004
4.4 Modeling Dynamic Competition
• First term is zero because i will choose second stage tactic qi st
di/dqi = 0; we have:
d i ( q*i ,q*j ,k ) dq*j d i

=1
dqj
dk dk
•
•
•
•
LHS 2nd term: direct effect
LHS 1st term: strategic effect
RHS: Direct cost of commitment
How can k alter j’s 2nd period tactics since k does not directly
affect j’s profits?
• Strategic move k alters i’s own incentives to choose  alters j’s
incentives to react  changes i’s profits
• Sign of strategic effect is equal to sign of
d 2 i d 2 i d i
dqi dk dqi q j dq j
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
59
Joensuu 2004
4.4 Modeling Dynamic Competition
Three effects
• How commitment k changes i’s own optimal tactics
• How j reacts to changes in i’s incentives
• How i’s profits are affected by changes in j’s tactics
Strategic effect > 0  overinvest in k
Strategic effect < 0  underinvest in k
• Example: Cost reduction in Cournot and Bertrand games
– How reaction functions shift as marginal costs of j are
decreased?
• Example: Increased marketing in Cournot and Bertrand games
– How reaction functions shift as j increases her marketing
expenses?
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
60
Joensuu 2004
4.4 Modeling Dynamic Competition
Taxonomy for Strategies
• Strategic substitutes vs complements
– Cournot game = strategic substitutes
– Bertrand game = strategic complements
• Commitment makes firm tough vs soft
• Investment k makes i tough
– i will produce more or price below
– k shifts i’s rf right and up in Cournot game
– k shifts i’s rf right and down in Bertrand game
• Investment k makes i soft
– i will produce less or price above
– k shifts i’s rf left and down in Cournot
– k shifts i’s rf left and up in Bertrand
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
61
Joensuu 2004
4.4 Modeling Dynamic Competition
Commitment makes firm
Stage 2 variables
are
Strategic
Complements
(eg, prices)
Strategic
Substitutes
(eg, capacities)
Tough
Soft
Puppy Dog Ploy
Strategic effect < 0
Commitment cause
rivals behave more
aggressively
Fat Cat Effect
Strategic effect > 0
Commitment cause
rivals behave less
aggressively
Top-Dog Strategy
Strategic effect > 0
Commitment cause
rivals behave less
aggressively
Lean and Hungry Look
Strategic effect < 0
Commitment cause
rival behave more
aggressively
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
62
Joensuu 2004
4.4 Modeling Dynamic Competition
Strategic Incentives to Commit in Cournot
• Commitment makes firm tough
– Reaction function shifts outward
– Firm will produce more for all given rivals’ output
– Example: Marginal cost reducing innovation
– Strategic effect might outweigh direct effect  Invest even if
NPV < 0!
– Beneficial strategic side-effect
– Top-Dog: Big or strong to become aggressive
• Commitment makes firm soft
– Firm will produce less for all given rivals’ output
– Reaction function shifts inward
– Example: Marginal cost increasing entry into other mkt
– Negative strategic side-effect
– Lean and Hungry Look: Refrain from expanding to avoid
weakness
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
63
Joensuu 2004
4.4 Modeling Dynamic Competition
Strategic Incentives to Commit in Bertrand
• Commitment makes firm tough
– Firm will underprice
– Reaction function shifts inward
– Example: MC-reducing innovation
– Negative side-effect
– Puppy-Dog Ploy: stay small or weak to avoid agressive
competition  Do not lower costs!
• Commitment makes firm soft
– Firm will overprice
– Reaction function shifts outward
– Beneficial side-effect
– Example: Target small niche, Product differentation
– Fat-Cat Effect: Become soft to attract only weak competition
 Sumo-strategy
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
64
Joensuu 2004
4.4 Modeling Dynamic Competition
• Need to look more than just direct effects of irreversible
decisions
• Nature of future competition affects incentives to make
investments or commitments now
• Need to look at how equilibrium changes, not just first effetcs
Examples of 1st Stage Commitments
• Build excess capacity  deter entry
• Enter and underinvest  avoid attracting tough competition
• R&D: reduce costs  price aggressively / gain mkt share
• Build large customer base, costly to switch  less competition in
future
• Underinvest in marketing  less loyal customers  become
aggressive in 2nd stage
• Overinvest in marketing  loyal customers  become soft in 2nd
stage
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
65
Joensuu 2004
5 Cartels and Collusion
• Competition leads to less than jointly maximal profit  firms
have incentives to avoid competition
• These incentives are basis for competition policy
• Explicit cartels, implicit tacit collusion
• How would these show up in reaction fn picture?
How Can We Detect Cartels and Collusion?
• Hard without ”smoking gun”
• Lerner Index L = (p - ci)/p = si/e?
– If p, si and e known, make inference on p - ci
– Often not practical: p, ci and e not known accurately enough
– But with good enough data this can be done
• Identical prices?
– Not evidence for cartel
– Perfect competition  identical prices
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
66
Joensuu 2004
5.1 Explicit Cartel
• Intuition:
– ”Few” competitors  easy to form cartel/collude
– ”Many” competitors  hard to form cartel/collude
• Selten (1973): 4 is few, 6 is many
– Intuition: with 6 firms, staying outside cartel gives more than
joining cartel with 5 other firms
• Result from 2-stage model:
– 1. Decide to join/stay out
– 2. Choose output
– If n > 5, best strategy in stage 1 is to stay out
– If n < 5, best strategy in stage 1 is join cartel
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
67
Joensuu 2004
5.2 Implicit Collusion
• Implicit agreement or understanding not to compete
– Eg. firms ”agree” on monopoly price and output
• But this is unstable
– Cheating and undercutting gives even higher profits than
collusion, if rivals adher to agreement
• Need mechanism to remove incentives for cheating
• "Stick-and-Carrot" Theory:
– Cheating draws punishment and low profits in future
– Collusion draws rewards (high profits)
– Deters from cheating on promise to fix prices
• Future reward  Collude now
– Requires that future matter
• How to punish? Price war an example
– Punishment will also hurt the punisher
– Need incentives to punish
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
68
Joensuu 2004
5.2 Implicit Collusion
Collusion in Bertrand Competition
• Model: firms interact repeatedly
• Assume c = 0, mkt demand q = a - bp
• Per period profits now it = pit qit(pit, pjt)
• Bertrand equilibrium price for one-shot game = 0
• On each period t each firm chooses price pit knowing all previous
prices pit-s, s = 1,2,3,…
• No end-game problem: repeat per-period game infinitely many
times
– Or: Prob(next period is last) < 1
• Future matters but less than today: firms discount future profits
with discount factor 0 <  < 1
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
69
Joensuu 2004
5.2 Implicit Collusion
• Owners of firms value monetary stream m such that mt+1 = mt,

1
1 P

1 r 1 k
• where r is discount (or interest) rate, P probability that game ends
after this period and k firm's marginal cost of capital
• Firm goal: maximize present value of per-period profit stream
Vi = t tit
• Strategy?
– Plan ahead how to play entire game
– What per-period moves to choose after any history
– Think: players desing strategy before game starts and then
leave computers to execute strategy
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
70
Joensuu 2004
5.2 Implicit Collusion
• Examples of simple strategies:
– One-shot Bertand price always
– Tit-for-Tat: do today what rival did yesterday
– pi1= pM; pit= pM if pjt-1= pM, else pit= 0
• Equilibrium: No incentive to change strategy
• Is "always one-shot Bertrand equil behavior” still an equil strategy?
– Yes: if i always chooses pit = 0, best j can do is to choose pjt =
0  it = 0
• Both always charge monopoly price and earn it = iM/2 > 0
equilibrium?
– If j always charges pjt= pM, what should i do?
– Look at reaction function: i should choose pit= pM- 
– If i deviates from pM, it earns higher profits every period iD =
pM-  > pM/2 (D: deviate or defect), hence
ViD = t t it(piD,pjM) > ViM = t t it(piM,pjM)
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
71
Joensuu 2004
5.2 Implicit Collusion
•  Strategy ”always monopoly price” is not in equilibrium
• ”Grim Strategy” (GS):
– Choose pi1= pM
– Choose pit= pM if pjt-1= pM
– Else always choose pit= 0
• Suppose j knows i plays GS; what is best for j?
– GS is best reply (among others)
•  GS is a best reply against itself
•  Both firms using GS is an equilibrium
• Punishment needs to be credible, otherwise it is only empty threat
– There must be incentives to start punishment
– Punishment must be part of equilibrium path from that
moment onward, so that no firm will want to deviate from
punishment
• One-shot Nash equil behavior always credible punishment
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
72
Joensuu 2004
5.2 Implicit Collusion
• GS punishes defection forever
• Punishment "too hard", lesser punishment suffices
• Optimal punishment: shortest number of periods T such that extra
profits gained by defection are vanished
– Stay on intended equil path: earn M/2 each period
– Temptation: gain M - M/2 -  = M/2 -  during defection
– Punishment: earn zero profits long enough so that profits
(defect + punishment) < profits (collusion)
• Minimum length of sufficient punishment depends on discount
factor 
• Often optimal punishment is minimax strategy of per period game,
ie tougher than one-shot equil behavior
• GS easy to use
• Point here collusive outcome, not details how one supports
outcome
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
73
Joensuu 2004
5.2 Implicit Collusion
• ”Folk Therorem": Any outcome that leaves each player more than
one-shot minmax outcome is sustainable as an equilibrium
outcome in infinitely repeated game
– There are many equilibrium strategies
– ”Anything” is in equil
– No predictive power without more assumptions
• Generally collusion is sustainable if temptation to defect is low
enough and punisment following the deviation strong enough
• Firm wants to keep colluding if present value of devi-ating is
smaller than present value of adhering to collusive agreement
• PV of collusion here
ViC = ttit(piC,pjC) = piC/(1-)
as t dt = 1/(1-d) if |d| < 1
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
74
Joensuu 2004
5.2 Implicit Collusion
• PV of deviation = profits reaped during deviation + present value
of profits earned during punishment:
ViD = D + ttit(piP,pjP) = D +  piP/(1-)
– Note: here punishment assumed to be infinitely long
• Collusion is sustainable if
P
D
C
 iC




  iD  i    iD iP
1
1
i  i
• Incentive to deviate depends on discount factor
• If discount factor is too low to support collusion, either toughen up
punishment or try to lower degree of collusion
– Longer or harder price war
– Reduce collusive prices from monopoly price
• Note: punisments are never observed
– None used since threat is enough
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
75
Joensuu 2004
5.2 Implicit Collusion
Collusion with Imperfect Information
• What if firms cannot observe rivals' exact prices nor outputs?
– Don't know if rival defected  know when to start price war
• No threat of price war  collusion not sustainable?
• Use other info: Sales were less than expected
– Think Bertrand oligopoly with identical goods and with
stochastic demand
– Firm has 0 demand today: somebody deviated and stole
customers or shift in demand?
– Start price war when price or demand drops "enough"
– Start price war even if you know nobody deviated, as nobody
has incentives to deviate
– Intuition: no punishment  no firm has incentives to collude 
per period equilibrium only possibility
– Better off with some price war instead of permanent rivalry
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
76
Joensuu 2004
5.2 Implicit Collusion
Factors that Help Collusion
• General idea: stronger, earlier and more certain punishment
increases possibilities to collusion
”Topsy-Turvy” principle: the more firms have
opportunities for aggressive competition, the less
competition there is
• Public prices and other market transparency
– Easy to observe deviation
• Size of cartel
– N equally sized firms
– Each firm receives 1/Nth share of total monopoly profits
– Collusion sustainable if one shot defection followed by
punishment leaves less profits that staying on collusive path:
1 p M Q( p M )  N  1
p Q( p ) 
1 
N 1 
M
M
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
77
Joensuu 2004
5.2 Implicit Collusion
• Product differentation works two ways
– More products are differentiated, the larger price decrease
needed to
• steal mkt share
• punish deviator
– More products are differentiated, less incentive to cheat and try
to steal mkt share
– More products are differentiated, less price war by rivals affects
profits
– Introduces non-price competition: more variables to monitor
and more ways to cheat
• Cost conditions and capacity utilization
– Capacity constraint or steeply rising MC reduce profit margin
for extra output
• Reduce incentive to cheat
• Reduces possibilities and incentives to punish
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
78
Joensuu 2004
5.2 Implicit Collusion
• Free capacity
– Increases temptation to cheat
– Allows harsher punishment  increases possibilities and
incentives to punish
• Elasticity of firm demand
– Inelastic firm demand  more mkt share means significant
reduction in price  less incentive to cheat
– More elastic demand is, the harder it is to sustain collusion
• Multimarket contact
– Firms produce several competing goods or operate on several
geographic mkts
– More opportunities to cheat
– Price war on all mkts  allows more severe punishments
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
79
Joensuu 2004
5.2 Implicit Collusion
• Firm symmetry
– Firms have different shares of a specific asset (capital) which
affects marginal costs
– Joint profit maximization: output is shifted away from small
(inefficient) firms towards large (efficient) firms
– Smallest firm has highest potential to steal business of its
rivals and, has highest incentives to disrupt collusive
agreement
– Incentives to deviate are reversed when equilibrium calls for
punishments
– Largest firm loses most at punishment phase, it will have
highest incentives to deviate from punishment
• Capacity constraints
– Incentives to stay in collusive equilibrium are very different for
large and small firms
– Small firm will have some incentive to cheat in short run, as it
can only increase its sales marginally up to capacity level
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
80
Joensuu 2004
5.2 Implicit Collusion
– Large firm has a lot more capacity available and can gain more
customers with same price deviation from collusive norm
• Large firms tend to have a greater incentive to deviate
from collusive price
– Asymmetry in capacities will also have an important effect on
effective punishments
• Worst punishment firm can impose on its competitors is to
produce up to full capacity
• Small firm that is already producing at almost full capacity
has low possibilities to punish rivals that do not follow
collusive norm
• Large firm competing with small firm will have large
incentives to deviate from any collusive norm without this
being disciplined threat of low prices in future
– Increases in asymmetries in capacities make collusion more
difficult
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
81
Joensuu 2004
5.2 Implicit Collusion
Collusion and Antitrust
• Read Motta Ch 4, Europe Economics report, UPM/Haindl decision p.
18-, and browse my ”forest” paper
– Joint dominance, coordinated effects are legal jargon ~
collusion in economics jargon
• How to identify collusion or separate collusion from competition?
• Authorities/customers argue that collusive equilibrium is played
• Suspected firms want to argue that behavior is as if noncooperative
Cournot or Bertrand equil is being played
• Possible to detect collusion from behavior alone?
– Firms have more mkt power than one shot equil?
– Estimate cost, demands and reaction fns and compare actual
behavior to non-cooperative and collusive equil
– Doable, but technical (eg. Nevo, Slade), see my ”forest” paper
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
82
Joensuu 2004
5.2 Implicit Collusion
• Core of policy problem: Collusion arises as equilibrium behavior
– Hard to prohibit or deal with ex post
• Solution: try to prevent collusion, ban business practices and
mergers that help facilitate collusion
• Analyses of asymmetry in assets and capacity constraints suggest
merger guidelines that differ from traditional wisdom
• For a given number of firms, Herfindahl and other concentration
tests tend to predict that a more symmetric industry is more likely
to be more competitive
• Asymmetry may be pro-competitive
– Asymmetry in industry may even more than compensate for
reduction in number of firms in merger involving large firm
– Increased asymmetry hurts collusion and may benefit
competition
Markku Stenborg: Kilpailu- ja yhtiöoikeuden taloustiede
83
Joensuu 2004