Collusion
Competition Policy Seminar (B.Sc.)
WiSo Faculty, University of Cologne
12/16/2014
Fabion Abazaj
fabio.ab94@gmail.com
Contents
1.
Definition ................................................................................................................................1
2.
Types of Collusion ...................................................................................................................1
2.1.
2.1.1.
Simultaneous cartel formation ...............................................................................2
2.1.2.
Sequential cartel formation ....................................................................................3
2.1.3.
Network of market-sharing agreements ................................................................3
2.1.4.
Famous examples of cartels....................................................................................5
2.2.
3.
Overt Collusion ...............................................................................................................1
Tacit Collusion .................................................................................................................5
2.2.1.
Bertrand price competition ....................................................................................7
2.2.2.
Cournot quantity competition ................................................................................8
2.2.3.
Famous Examples ...................................................................................................9
Conclusion............................................................................................................................ 10
1. Definition
In economics collusion is defined as the oligopolistic market situation where firms
may refrain from undercutting each others prices or from selling in each others market
areas, regions or other divisions, and thus prices are higher than in a competitive benchmark.1In his book Competition Policy: Theory & Practice, Motta argues that in order
for collusion to arise there must be two elements that characterize the involved parties.
First, all the participants firms in collusion must be able to detect in time that a colluding
firm is deviating through undercutting the agreed price, and second, the should be able
to imply a punishment, i.e. either, selling higher quantities in the post-deviation period,
or setting even lower prices, in order to impede the deviant firm from gaining profits. 2 If
these two conditions exist, a collusive outcome will arise among the firms in a market.
2. Types of Collusion
As far as the coordination is concerned, collusion can be achieved in two ways: it
can either be overt (explicit) or tacit (implicit). Each type is analyzed below.
2.1.
Overt Collusion
The term explicit or overt collusion refers to formal agreements among firms, with
firms meeting often to discuss about prices and possible punishment in a case a colluding firm deviates. These agreements are officially called cartels. Cartel agreements may
provide for setting the least prices, limiting the output or capacity, restricting non-price
competition or posing measures to restrict entry to the market.3 Cartels are usually
banned under antitrust laws in modern economies. In the US they are banned under the
Sherman Act (Section 8.1) and in the EU under Article 81 & 82 of the EC Treaty.4 They
theoretically split into: a) the direct profit-maximization ones (which are formed simultaneously or subsequently) and b) the market-sharing agreements. In order to express
better a formalized analysis, we use the Cournot oligopolistic model.
Oxford Dictionary of Economics, J. Black, N. Hashimzade & G. Myles, Oxford University Press (2012),
lemma: Collusion
2
Competition Policy: Theory and Practice, M. Motta, Cambridge University Press 2004, Ch. 4, p. 140
3
Oxford Dictionary of Economics, J. Black, N. Hashimzade & G. Myles, Oxford University Press (2012),
lemma: Cartels
4
The Political Economy of Antitrust, Ghosal, V., Stennek, J. (Eds.), Elsevier B.V., Ch. 8, pp. 217
2.1.1. Simultaneous cartel formation
In a Cournot game of
symmetric firms (same constant marginal cost c) and sell an
homogeneous product (same product demand
, where
is the total
quantity produced by all firms), it can be proved that a firm is best response is to produce
and the profits for firm are
. (1)5.
] firms is formed. The Cournot game is now
We suppose now that a cartel of
played by
independent firms that do not participate in the cartel. To define safer
our math, we suppose that not all firms are going to participate in the cartel, so the independent firms are
Thus, transforming equation (1) (we substitute
with
), the profits for each firm outside the cartel in Cournot equilibrium will be
. Since the cartel may be considered as one firm, the whole cartel in
Cournot equilibrium will gain
(the same as a firm outside the cartel)
and because of the symmetry, profits are equally divided to
the cartel will gain
firms and each firm within
.For the sustainability of this cartel, no cartel
member should have an incentive to leave the cartel agreement, which is equivalent to
(2), or meaning that for a firm the profits should be bigger inside
the cartel than outside. Solving inequality (2) will lead to
(3).
It is seen from (3) that for
which implies
, the inequality is impossible,
which means that every member has incentive to leave the agreement. Yet, the inequality shows that for
the cartel is sustainable. So, a duopoly may be led to a car-
tel agreement, but no oligopoly with 3 or more firms can attain such an agreement.6
Firms outside the cartel perceive the higher cartel price as a public good, so they have a
strong free-riding incentive on it. But we should recall that this analysis is based on the
5
6
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.3, p. 54-59
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 337-338
prerequisite that firms are cost-symmetric and produce a homogeneous product, which
both are simpler hypotheses. (for differentiated products in a market, see 7)
2.1.2. Sequential cartel formation
We now suppose an exogenously determined order of firms which have to respond
to a first firms proposal on the formation of a cartel. If all firms accept the proposal,
they thus form a cartel and exit from the Cournot game. Among the remaining independent firms, there will be again a first one to make another proposal, and so forth. If
one of the ordered firms rejects the proposal, the proposed cartel is not formed, but there
comes the proposal for another cartel by this rejecting firm. The cartel will be formed if
all the next questioned firms accept it and so forth, as previously.
The critical cartel size is given comparing the profits a firm gains inside a cartel of
size k, that is
, with the profits each firms gains when all the firms are independ(Its 1, and not 0, according to the previous model we had used with
ent,
independent firms). Thus, it is preferable for firms to create a cartel, if
(
) If
is the first
integer larger than the RHS of the inequality, denoting the firms forming the cartel, then
the first
firms are perceived as the ones to remain independent. Belleflamme &
Peitz, show that
, which
roughly means that the least lucrative size of a cartel is larger than 80% of the firms in
the industry.8
2.1.3. Network of market-sharing agreements
Market-sharing agreements are bilaterally signed collusive agreements between two
firms that are active in geographically or consumers different consumer markets. In an
industry with totally n firms, there may be such collusive pairwise links among many
firms
These links make up a set of total links or collusive network
Using
again the same hypotheses ( firms, Cournot competition on homogeneous product, demand
7
8
, symmetric marginal cost ) and the
firms initially being present on
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 338-339
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 340-341
distinct geographical markets, we can see that the profit of a firm (principally on market , but profiting in market j, too) are:
where in a collusive network
spectively,
is the number of firms active in market , and re-
the firms active in market . Reasonably, the first term denotes the
profit within the domestic market, and the second term in within the foreign market,
having signed no agreement yet. If an agreement is signed between a firm and a firm ,
that is a link
(being added to the network
while
the
), the profits of firm
incentive
to
sign
]
an
are restricted to:
agreement
is
where the first term in carets denotes the less competition on s market, and the second
term subtracted is the former access to s market, which is now lost. (2 opposite effects)
The difference above refers to the incentive of a firm to sign an agreement with another firm . But the incentive in case the network is empty or non-empty, in the
sense that no firms in the market may have signed any sharing agreement, or not.
In an empty network, symbolized by
, there is incentive that two firms sign an
agreement if, and only if,
only for
which is true
Hence, the network is stable for
, meaning that no pair of firms
would have incentive to form a market-sharing agreement.But, in a non-empty network
, two firms will have a benefit from signing an agreement if both
and
isfied as a system only by
, which because of symmetry, is always sat, which roughly means that if a non-empty net-
work is formed, it must be of complete collusion with every firm being a monopolist on
its own market.9
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 341-343
2.1.4. Famous examples of cartels
2.1.4.1. Vitamin cartels
In November 2001, the European Commission deterred multiple cartels of 8 vitamin
companies cooperating for a total of 10 years. The EC imposed a fine of 855.22 million in total on all the companies. There were later found out overall 13 companies involved in secret agreements of market-sharing and price fixing policies, and famous
ones leading, such as Hoffman-La Roche (CH), BASF (DE) or Takeda Chemical Industries (JP.) This has been the most damaging series of cartels that the EC had ever investigated because of the range of products involved.10
2.1.4.2.
OPEC
The Organization of Petroleum Exporting Countries (OPEC) has been attempting
since 1973 to function as a world oil-exporting cartel, setting negotiated prices and petroleum export quotas with the oil companies.11 Although it is not an agreed cartel between companies, but a negotiating organization, OPEC is considered as the most famous world cartel. Its members are Iraq, Saudi Arabia, Iran, Kuwait, Venezuela, Libya,
the United Arab Emirates, Qatar, Indonesia, Algeria, Nigeria, Ecuador, Angola and Gabon. Since its creation, OPEC has had a very big influence on oil prices, emerging even
on political crises. (1973 Yom Kippur War12) OPEC is not normally subject to competition law, since it is protected and legally recognized by the US foreign trade laws.13
2.2.
Tacit Collusion
Since cartels are banned by competition law in most of the countries of the developed or quickly developing economies, a typical conversation about collusive coordination should focus on tacit or implicit collusion, which occurs without explicit communication between the colluding parties. Since it is risky for modern firms to sign explicit agreements on setting higher than competitive prices, which would give antitrust
authorities proof for penalties14, they do not cooperate but they agree on prices or
quantities of product observing each others moves in the market. By game theory, it can
10
http://europa.eu/rapid/press-release_IP-01-1625_en.htm?locale=en
Oxford Dictionary of Economics, J. Black, N. Hashimzade & G. Myles, Oxford University Press (2012),
lemma: Organization of Petroleum Exporting Countries
12
http://history.state.gov/milestones/1969-1976/OPEC
13
Foreign Commerce and the Antitrust Laws, W.L.Fugate, Aspen Publishers, 2002, p. 192
14
Competition Policy: Theory and Practice, M. Motta, Cambridge University Press 2004, Ch. 4, p. 138
11
be verified that tacit collusion is not possible when the horizon of the competition in the
game is finite ( periods), since they would know ex-ante their profits in their last period
of the game. (for the analysis, see15)
But in an infinite horizon (
) (not of competition, but better of coexistence
game in the same market) tacit collusion may emerge at the subgame perfect equilibrium. For simplicity, we consider two firms,
and , in an industry. We first define the
grim trigger strategy, according to which firm initially chooses the action that maximizes both firms profits, acting in a cooperative behaviour. (We will later see that the action may be setting either the price or the quantity of the product). Firm can follow that
action, and then firm
will choose again this action, as firm
tacitly agreed. Firm
will keep choosing the same action as long as both of them have done so in the previous
periods, marking thus a cooperation phase. Nevertheless, in case one of the two firms
chooses another action and thus deviates, that new action will trigger the beginning of
the so-called punishment phase. From that period on and infinitely both firms choose the
action that leads to the Nash equilibrium of the game, thus competing again.
The profits in each different phase are defined as follows: when the firms are in the
cooperation (or collusion) phase, each one gains a profit
, where
is the mo-
nopolistic profit, since both firms collude, thus acting as the sole firm in the market.
When one firm acts cooperative and the other deviates, the latter gains
mally is equal or to
, which nor-
(or a little lower), while the cooperative firm normally gains ze-
16
ro profits . During the punishment phase or, equally, at the Nash equilibrium of the static competitive game, both firms gain
, and naturally we have that
.A
firm always maximizing its profit will compare what it immediately gains from deviation with future losses because of the punishment phase that will follow, which depends
both on the magnitude of each type of profit and on the firms discount factor.
If we start in a punishment (or non-cooperative) phase, firm
will choose an action
according to the Nash equilibrium of the static game for the whole infinite horizon, and
15
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 344
The Economic Assessment of Mergers under European Competition Law, D. Gore, S. Lewis, A. Lofaro,
F. Dethmers, Cambridge University Press 2013, p. 334
16
firm s best reaction is to do so, too, at every period, that it is following the trigger
strategy. But if we start in a cooperative phase, and firm chooses the grim trigger strategy, it will gain
in each period, as the punishment phase will never be triggered by
firm . Taking into consideration a discount factor
will gain
, in present values, firm
(1). The same applies to firm . But
if one firm deviates, in the first period it will gain
(now punishment phase) it will gain
and for each following period
Thus in present discounted values, it will totally
gain
(2). Thus, a firm prefers to fol-
low the grim trigger strategy if and only if
, which eventually leads to
(3), with the LHS depicting the long-term losses as a result of
the punishment (in discount) and the RHS depicting short-term gain as a result of deviation. Solving (3) for , we have that
, which roughly means
that tacit collusion will be sustainable, if the discount factor of the firms is above a certain level. Hence, firms will mostly collude if they fear that deviation is paying less now
and that they will lose more in the future through punishment than now through sharing
profits, provided that deviation is observable by the firms so that punishment can start.17
But, the actions a firm can undertake may be either price-setting or quantity-setting.
And so is the way that the theoretical model above can be applied as.
2.2.1. Bertrand price competition
For simplicity, we have to recall the Bertrand model of competition with constant
for two firms. If a same price is set by firms, thus the market will split equally
for both firms. If firms set the monopoly price
one will gain
, which is the highest possible, each
(Its on their interests to charge the higher price while colluding,
even if they could charge a range of lower prices up to the competitive one). By slightly
undercutting the rivals price, a firm will mostly gain a deviation short-term profit almost
17
, since the customers have no incentive to buy anymore the product from
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 345-346
the firm with the higher price, and thus, the betrayed firm gets 0 profits. After the deviation, the punishment period begins, when each firm will eventually set the Nash equilibrium price
and they will both have
, and by substituting into the lowest dis(
count factor ratio we have that
So, in this price-setting duopoly of infinite horizon, any collusive profit level between
and
is sustainable, provided that the discount factor is
alize the logic above for
firms, with
remaining equal to
. We can gener-
, but
. So, using
again (1) and (2), we have that
, which
is increasing for n, that is, when there are more firms in the market, the discount factor
has to be greater and thus collusion is more difficult to sustain.18
2.2.2. Cournot quantity competition
But firms can also adjust their quantity to their competitors choices, rather than setting prices. We suppose a homogeneous product market of
game, where
for all the firms, and the inverse demand function is
for simplicity. Equalizing
stituting
this
firms with a linear Cournot
in
the
gives us the monopolistic quantity
profit
function
will
give
us
the
. Sub-
monopoly
, while the collusive profit is
profit:
. (1)
However, we already know that at the Nash equilibrium of this symmetric game a firms
best response is to produce
. (2) Since also
, which gives an equilibrium profit of
(
))
the firm may deviate if it considers that all the other firms are producing q, maximizing
the previous profit for
), thus having deviation profit
, and since the other firms choose the collusive
18
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 346-347
quantity
, the deviant firm will gain
(1), (2) and (3) we find that
This measure is 0.53 for
(3). From
.
and is increasing with
when
, which as in the
Bertrand price competition means that collusion will be sustainable for more firms if the
discount factor is even greater, which means difficult sustainability of collusion. We can
notice that for
, that roughly means collusion is easier in a duopoly
in price competition. But for
, quantity competition is easier for the sustainability
of collusion since
. This can be explained as follows: when
, there is
a stronger punishment which makes collusion easier in price competition. When
the deviation is stronger which makes collusion more difficult to sustain in price competition that in quantity competition.19
2.2.3. Famous Examples
2.2.3.1.
Wood Pulp
In 1984, 40 wood pulp producers, which had their offices outside the EU were found
out by the European Commission to have infringed Article 81 (it was then named A.85)
through price collusion. The EC based its claims on suspicious parallel behavior among
firms, consisting of quarterly price announcements, simultaneous price announcements
and eventually same prices. The EC had then imposed large fines on the producers. 20 In
1993, the European Court of Justice was called by the same producers to reinvestigate
the case, since they claimed that they were established outside the EU and thus could not
be considered to infringe the EU competition law. The ECJ had difficulty in establishing
a clear decision, since there was indeed missing evident documentation or agreements
that would leave no doubts for the formation of a cartel. In the light of the experts reports that conducted the new investigation, the ECJ reached the conclusion that cooperation is not the only possible reason for parallel behavior, but still supported the ECs decision, claiming that it was of high importance that these suspicious practices were undertaken within the territories of the EU, no matter where the firms were established.21
19
Industrial Organization, Market & Strategies, P. Belleflamme & M. Peitz, Ch.14, p. 346-347
Competition Policy in the EU, X. Vives, Oxford University Press 2009, Ch. 5 (M. Motta), p.111
21
Competition Policy in the EU, X. Vives, Oxford University Press 2009, Ch. 5 (M. Motta), p.112-113
20
2.2.3.2.
Samsung & LG
In 2012, the KFTC (regulatory authority for competition in S. Korea) fined Samsung
Electronics and LG Electronics for coordination on fixing prices of some electronic appliances during 2008-2009, holding secret meetings. Samsung was fined 18.86 million,
while LG was fined 13.75 million.22
3. Conclusion
Either overt or tacit, collusion damages consumers by leading the prices of products
to rise more than normally. Below, the graph shows how consumers are damaged from
i.e. the formation of a cartel in a duopoly comparing to the competitive benchmark:
If firms agree on the
intersecting with the
firm, where
price of monopoly, their joint
curve
curve of the market, gives out quantities
and
for each
. Provided that each firm has different costs, the profit of firm 1
maximized under this cartel is depicted by rectangle (
respectively for firm 2 by rectangle
due to higher costs and
due to lower costs. The total producers
surplus in the market is given by the area above
and to the border of the adja-
cent triangles. The consumers surplus is obviously reduced from
while we now encounter a deadweight loss of
to (
due to imperfect competition. So,
the effects of this example of collusion are the same as those of a normal monopoly.23
Collusion is obviously leading to negative effects for consumers. Especially tacit collusion, which is more difficult to be detected, can lead to absorbing much of the consumers surplus for long periods, since no intervention of competition authorities in time
may be possible.
22
23
http://www.bbc.co.uk/news/business-16540678
http://www.amosweb.com/cgi-bin/awb_nav.pl?s=wpd&c=dsp&k=collusion
10