Lecture 4 Collusion

Competition and Regulation
Lecture 4
Collusion
Overview
• Definition
• Where does collusion arise? What
facilitates collusion?
• Detecting cartels; Policy
2
Definition
• Agreement to control prices, market share, or
a quality attribute of a good or service, but
also control of necessary inputs and essential
facilities to prevent further entry;
• The law should punish only explicit collusion
because of uncertain assessment and
potential instability of tacit collusion. However
studying implicit (tacit) collusion to make it
unstable is important as well.
3
Conditions for collusion
• A collusive
equilibrium may
be unstable due
to temptations to
deviate from
cooperation.
Collude Not
collude
Collude 10
0
10
Not
collude
20
20
2
0
2
4
Conditions for collusion
• However with repeated interaction,
collusion is possible when a punishment
for deviation is possible and credible;
• A) deviations should be detectable
immediately;
• B) Punishment should be credible and
sufficiently costly for deviating firm;
5
Repeated games
A game repeated
two or a known
number of times
can be solved with
backward induction
Collusion is never
a NE in this type of
games;
Collude Not
collude
Collude 10
0
10
Not
collude
20
20
2
0
2
6
Conditions for collusion
An infinitely repeated
Collude
game (or with unknown
end period) instead can
have a collusive outcome
if, after deviation is
Collude 10
detected, colluding again
becomes impossible;
10
This can be enforced by
the choice of trigger
Not
20
strategy (or tit for tat) by at
collude
least one player;
0
Not
collude
0
20
2
2
7
Coordination
• Under tacit collusion a further problem is
how to coordinate. Ex. How to choose
collusive price? How to adjust it when
circumstances/costs conditions change?
• Market sharing agreement (particularly
geografic) solve this problem and also the
problem of detection of deviation
8
Factors
• Concentration. Smaller numbers facilitate
collusion:
– Less incentive to deviation to gain mkt shares;
– Easier to coordinate;
– Easier to detect deviation.
But asymmetric mkt. shares make it more unlikely. More
temptation to deviate→ HHI not a good measure.
• Entry. The easier entry, the less useful collusion.
Entrants will destroy the equilibrium.
• Cross ownership or agreements (joint ventures).
Comunication and interest convergence
9
Factors
• Frequency of orders. Large one-off orders may
destroy collusive equilibria;
• Buyer power.
• Elasticity. Less to be gained by colluding if the
elasticity is high. Also incentives to deviation may
be larger;
• Demand stability. Easier to collude/more
transparency;
• Product Homogeneity. Easier to detect deviations
• Symmetry. Asymmetric firms are less likely to
collude. Difficult to reach agreements (esp tacit).
10
Large firms could be difficult to punish.
Factors
• Multi-market interaction. Punishment in
more than one market is more costly to
deviant. Not so clear however: also the
temptation is larger;
• Excess capacity. More difficult if large,
however as above result is ambiguous
• Observability of market shares and prices
is tantamount; policy should focus on tools
that facilitate transparency and monitoring
11
Exchange of info
• Collusion could be enforced through
exchange of info. Market studies, prices,
market shares. Even aggregate market
data could facilitate.
• In theory more transparency could
increase welfare, but exchange of
individual information is almost certainly a
sign of collusion
• Should be forbidden
12
Coordination signals
• Focal points. No need of info. exchange.
Usually status quo (especially market
division).
• Announcement of future prices. If private
or without commitment value then
collusive. Ex auctions;
• Public binding announcement instead may
be competitive in nature;
13
Coordination signals
• Meeting competition clauses. Use
consumer as enforcer of agreement, they
deliver the information and the gains for
the deviant are smaller
• Uniform delivery price increases
observability;
• A commitment not to sell at a discount
may facilitate collusion (MFN clause)
14
Conditions (para. 4.2.5.1)
• Suppose n firms.
Πc is the present profits from collusion
Πd is present profit from deviation
Vc is future value of colluding (eg the stream of future
profits continuining the collusive outcome)
Vp future value under punishment (eg the stream of
future profits after collusion collapses)
δ= discount factor=1/(1+r) where r is a discount rate
Then for firm i collusion is justified if present value of
collusion is larger than the present value of deviation
15
Conditions
16
Conditions
• Hence if δ is larger than a threshold (r
lower than a threshold) then collusion is
sustainable; More impatient firms are less
likely to collude;
• Depends on the other side on temptation
and punishment; the larger the gains from
deviating and the lower the future loss of
profits, the less likely is collusion;
Should we punish tacit collusion?
• No reliable data to calculate monopoly
price. How close should it be?
• Conviction on high price grounds is
dangerous;
• Price parallelism can be a result of
common shocks or fears of retaliation
rather than collusion
• Inferring collusion for mkt data is not
justifiable
18
Ex-ante Competition policies
• Penalties: fines, third-party compensation,
criminal penalties for management as a
deterrent are important; but so is the probability
of being caught;
• Auction design can prevent signalling practices;
ex: contemporaneous rather than sequential
auctioning; anonymous bidding;
• Black list of practices:
– information sharing, about individual firms data and
esp. future strategies
– cross shareholdings
19
Ex post Competition Policies
• Dawn raids to uncover evidence of explicit
collusion
• Leniency programmes: automatic
(voluntary, before investigation starts) and
discretionary;
• Both in the US and EU the ‘discount’ on
the fine may be very large (especially for
automatic case) up to automatic immunity;
20