Voluntary Submissions
Analytical Framework Teleconference Series
What Is Dominance?
Monday, 7 March 2016
Fair Trading Commission, Barbados
UCWG WORKBOOK: 2015-2016 SESSIONS
Session #1: What is Dominance /Substantial Market Power?
What knowledge or law should be drawn upon to understand the concept of
“dominance / substantial market power”?
Economic theory, precisely the law of supply and demand and the definition of a
market should be drawn upon to understand the concept of dominance and
substantial market power. The assumption that buyers and sellers constitute a
market establishes competition. Albeit not definitive, this principle introduces
competition which allows prices to change in response to variations in supply and
demand 1. Microeconomics introduces multiple forms of market structure such as
monopoly, oligopoly and perfect competition providing fundamentals on demand
and supply–side substitutability and pricing which aid in the defining of a market.
In addition to the definition, further knowledge of any specific industry or market
must be included.
It is also critical that the legislation in the respective jurisdiction be clearly
understood as well as any judicial precedent set.
Exactly what is meant when we say that a company is “dominant” or has
“substantial market power”?
The Fair Competition Act CAP. 326C Section 16 (2) defines “dominant” as
“ For the purposes of this Act, an enterprise holds a dominant position in a market if, by itself
or together with an affiliated company, it occupies such a position of economic strength as
will enable it to operate in the market without effective competition from its competitors or
potential competitors.”
1
Economics Basics: Monopolies, Oligopolies and Perfect Competition
In this regard, there is only a mention of “dominant position” and not of “substantial
market power”. However, one may infer from the definition or requirement to be
dominant that “substantial market power” is included as well. This is especially true
since there is the mention of “occupies a position of economic strength as will enable
it to operate in the market without effective competition….”. This is consistent with
the general understanding of dominance and substantial market power.
A company holds a dominant position when it maintains the ability to prevent
effective competition on the relevant market through elements such as profitable
price increases with a high degree of market power 2. A dominant company also has
significant freedom from competitive restraints imposed by rivals and consumers in
the market. A company has substantial market power when prices can be raised
above the benchmark price in a profitable manner.
What are the important considerations in calibrating the height of the dominance
bar?
The dominance bar constitutes a set of variables used to inform whether an
undertaking occupies a position of dominance. The bar generally considers whether
the defending firm has the ability to maintain dominance by analysing high market
share, financial depth, size of market and/or economy, policy implications, type of
industry and whether it operates in an environment with high barriers to entry.
These variables are necessary to guide the Competition Agency in determining if a
particular type of conduct warrants further inquiry but are nonetheless insufficient
to confirm dominance. Dominance is therefore confirmed by tests which investigate
whether the firm has the ability to raise its price without consideration to the
profitability resulting from the price increase and price responses from competitors.
Once it is shown that a firm has power to maintain profits following the price
increase then dominance is confirmed.
2
Competition Policy Theory and Practice, Massimo Motta
While it is important to use tests to verify dominance it is also critical to consider
beforehand the implications in choosing the level at which dominance is presumed.
Should the dominance bar be set high or low?
In deciding the height of the bar, the Agency will have to consider the realities of its
economy.
In developed nations where capital markets are more advanced and
where there exist opportunities to exploit economies of scale, it is prudent to assume
market entry to be more likely than in developing nations. For that reason, mature
jurisdictions may opt for a higher dominance bar. On the other hand, in lesser
developed jurisdictions where capital markets are less vibrant, it is more favourable
to use lower thresholds. Experience in small open economies, in particular island
states, proves that dominant enterprises area more likely to exist, because, in
addition to the imperfect capital markets, small landmass and population size make
entry less attractive.
If there is to be a market-share safe harbour, what should it be? Why?
As a general guide, the Commission 3 considers a firm that has had a sustained
market share of 50% or more is presumably in a position of dominance, whereas a
firm with a market share of less than 40% is less likely to hold a position of
dominance.
While market share is an important factor it is not, by itself,
determinative of dominance. Once again, this answer is not definitive but lends
some measure of benchmark for the authority and the business community.
A market-share safe harbour is seen as a threshold used to determine the level of
market share where dominance generally will not be found. Safe harbor is generally
used to provide some clarity and comfort, in this case for businesses. In such a
situation where a company has below a certain market-share threshold it is
presumed or certain that it will not be seen as dominant. It also reduces the time and
3
The Fair Trading Commission - Barbados
burden on agencies in determining if a company is dominant or for that matter if the
conduct is anticompetitive.
Is there a tradeoff between the height of the dominance bar and the standard of
proof for abuse?
If the standard of proof for abuse is consistent then there should be no trade-off with
respect to the height of the dominance bar as the latter is usually utilized as a mere
guide. However, if the bar is set inappropriately, what emerges is the issue of
enforcement errors4.
4
Type I and Type II errors
Competition Bureau of Canada
International Competition Network - Unilateral Conduct Working Group
UCWG Workbook - Session #1: What Is Dominance/Substantial Market Power?
Canada’s Competition Bureau (the "Bureau") is pleased to provide this submission to the
International Competition Network’s Unilateral Conduct Working Group’s 18 December 2015
dialogue on the analytical framework for evaluating unilateral conduct, addressing the question
of “What is Dominance/Substantial Market Power?” The Bureau, headed by the Commissioner
of Competition (the "Commissioner"), is an independent law enforcement agency responsible for
the administration and enforcement of the Competition Act1 (the "Act") and certain other statutes.
The Competition Tribunal (the “Tribunal”) has jurisdiction to hear and dispose of all applications
made by the Commissioner under certain sections of the Act, including the abuse of dominance
provision (section 79). In carrying out its mandate, the Bureau strives to ensure that Canadian
businesses and consumers have the opportunity to prosper in a competitive and innovative
marketplace. Answers to the questions posed by the UCWG follow below.
1. What knowledge or law should be drawn upon to understand the concept of
“dominance/substantial market power”?
In Canada, the concept of dominance is crucial in assessing whether an abuse of a dominant
position has occurred under subsection 79(1) of the Act. The elements of subsection 79(1) of the
Act are set out below:
79(1) Where, on application by the Commissioner, the Tribunal finds that
(a) one or more persons substantially or completely control, throughout
Canada or any area thereof, a class or species of business,
(b) that person or those persons have engaged in or are engaging in a practice of
anti-competitive acts, and
(c) the practice has had, is having or is likely to have the effect of preventing or
lessening competition substantially in a market,
the Tribunal may make an order prohibiting all or any of those persons from engaging in
that practice.
1
Competition Act, RSC 1985, c C-34.
Paragraph 79(1)(a) of the Act focuses on market power. The Bureau considers market power, in
the economic sense, to be the ability of a single firm or a group of firms to profitably maintain
prices above the competitive level (or similarly restrict non-price dimensions of competition) for
a significant period of time. Market power can also arise on the buying side when a single firm
or group of firms has the ability to profitably depress prices paid to sellers to a level that is below
the competitive price for a significant period of time. Market power has also been described as
the ability to set prices above competitive levels by restricting the output of one's rivals. This can
arise through the power of a firm to exclude new products from a market or through the
imposition of limits on competition using products of which it is a key provider. The assessment
of market power under paragraph 79(1)(a) accounts not only for a firm’s pre‑existing market
power, but also for market power derived from the alleged anti‑competitive conduct. Where a
firm does not presently appear to have and is not likely to acquire market power through the
alleged conduct within a reasonable period of time, the Bureau will generally not investigate
allegations of abuse of dominance relating to that conduct under paragraph 79 of the Act.
Further details about the Competition Bureau’s views regarding dominance within the
framework of unilateral conduct analysis can be found in the Bureau’s Abuse of Dominance
Guidelines. The Bureau also relies on current economic literature and jurisprudence from
Canadian courts, Canada’s Competition Tribunal, as well as guidance from foreign enforcement
agencies and courts for insights as to the proper approach in given situations of unilateral
conduct.
2. Exactly what is meant when we say that a company is “dominant” or has
“substantial market power”?
The Bureau considers market dominance to be synonymous with market power. In a general
sense, market power is the ability of a single firm or a group of firms to profitably maintain
prices above the competitive level, or other elements of competition such as quality, choice,
service, or innovation below the competitive level, for a significant period of time. Economic
theory also describes market power as the ability to set prices above competitive levels by
restricting the output of one’s rivals for a significant period of time. The Bureau normally
regards a “significant” period of time for the purposes of establishing market power to be one
year.
3. What are the important considerations in calibrating the height of the dominance
bar?
Market power can be measured directly and indirectly. Direct indicators of market power, such
as profitability or evidence of supra‑competitive pricing, are not always conclusive; practical
difficulties can arise in defining the “competitive” price level and the appropriate measure of cost
to which prices should be compared. Consequently, in addition to assessing any available
information in regard to direct measures of market power, the Bureau assesses a number of
indirect indicators of market power, including technological change, recent entry into or exit
from the market, industry supply capacity, and countervailing market power on the parts of
customers and distributors. Factors on which the Bureau places particular emphasis are market
shares and barriers to entry. The objective of this analysis is to determine the extent to which a
firm or group of firms is constrained from raising prices owing to the presence of effective
competition or the likelihood of competitive entry. In the case of entry barriers, including those
created by the firm’s conduct, the Bureau will assess not only whether entry is likely, but also the
time period required for an entrant to become a viable and effective competitor of sufficient scale
and scope such that an attempted price increase is not likely to be sustainable.
4. Should the dominance bar be set high or low? Why?
The element of dominance is an important threshold consideration in the abuse of dominance
analysis, but as mentioned above, it is but one part of a three-part test under the Canadian
legislation. In the context of paragraph 79(1)(a), the Bureau must establish that the a firm or
group of firms in question have “control” or market power in a relevant market. The standard for
“control” should not be set so high as to preclude anti-competitive conduct that has resulted in a
substantial lessening or prevention of competition merely because the firm in question lacked a
high market share (as an example).
5. If there was to be market-share safe harbor, what should it be? Why?
Though there are no hard and fast rules governing the relationship between market share and
market dominance, the Bureau is guided in its approach by the following general criteria when
examining market conditions:




A market share of less than 35 percent, held by one firm, will generally not prompt
further examination.
A market share between 35 and 50 percent, held by one firm, will generally only
prompt further examination if it appears the firm is likely to increase its market share
through the alleged anti-competitive conduct within a reasonable period of time.
A market share of 50 percent or more, held by one firm, will generally prompt further
examination.
In the case of a group of firms alleged to be jointly dominant, a combined market share
equal to or exceeding 65 percent will generally prompt further examination.
The Bureau considers that a market share of less than 35 percent will normally not give rise to
concerns of market power or dominance. In joint abuse cases, the Bureau considers that a
combined market share of the group of firms alleged to be jointly dominant of less than 65
percent will normally not give rise to concerns of market power or dominance. It is important to
note that the Tribunal has not considered a contested joint dominance case.
However, there may be situations where market power will be found despite a firm’s low market
share (i.e. below the 35% “threshold”). For example, the Tribunal in Visa/MasterCard found a
duopolist firm with less than 35% of market share in the relevant market to be dominant
(although not jointly dominant), taking into account the firm’s pricing discretion, its margins and
the very high barriers to entry. Therefore, an examination of market conditions can be important
in determining whether a firm with market share below 35% will be considered to possess
market power.
6. Is there a tradeoff between the height of the dominance bar and the standard of
proof for abuse?
The fact that a firm holds market power is not, in and of itself, sufficient to warrant intervention
under section 79. Rather, all elements of subsection 79(1) must be satisfied “on a balance of
probabilities” to constitute an abuse of dominance. Therefore, the degree of market power, the
nature and severity of the anti-competitive acts, and the degree to which competition in the
market has been impacted will all form part of the determination.
The Tribunal in Tele-Direct found that where a firm with a high degree of market power is found
to have engaged in anti-competitive conduct, smaller impacts on competition resulting from that
conduct will meet the test of being "substantial" than where the market situation was less
uncompetitive to begin with.
Directorate-General of Competition of the
European Commission
Submission by the Directorate-General for Competition of the
European Commission for the ICN Unilateral Conduct Working
Group's first round of dialogue in preparation of the drafting
of a chapter on the analytical framework for the ICN
Unilateral Conduct workbook
I.
SUBSTANTIAL MARKET POWER AND DOMINANCE DEFINED
1. Market power is the power to profitably influence market prices, output, innovation,
the variety or quality of goods and services, or other parameters of competition on the
market. An undertaking1 that is capable of substantially increasing prices above the
competitive level2 for a significant period of time holds substantial market power and
possesses the requisite ability to act to an appreciable extent independently of
competitors, customers and consumers. Unlike undertakings in a market characterised
by effective competition, an undertaking enjoying substantial market power is not
subject to effective competitive constraints
2. In the European Union an undertaking may be held liable for distorting competition by
way of unilateral conduct only if it has market power which confers on it a dominant
position in a relevant market. Holding substantial market power amounting to
dominance is a legal pre-condition for challenging anticompetitive unilateral actions.3
However, neither the holding of a dominant position, nor the acquisition of such a
position is reproachable under the EU rules on unilateral conduct. Only the abuse of
such a position constitutes an infringement of these rules.4
3. Conduct that may constitute an abuse when performed by a dominant firm, can be
procompetitive when performed by firms that are not dominant. Alternatively, firms
that are not dominant are not likely to succeed in harming the competitive process for
a significant period of time when using that conduct. The requirement for dominance
1
In EU law an undertaking is any entity engaged in an economic activity regardless of the legal status of the
entity or the way in which it is financed. See Case C-41/90 Höfner and Elser [1991] ECR I-1979, paragraph
21.
2
The expression 'increase prices' is used as a short-hand for the various ways in which the parameters of
competition can be influenced to the advantage of the dominant undertaking and to the detriment of
consumers.
3
Article 102 of Treaty on the Functioning of the European Union prohibits any abuse by one or more
undertakings of a dominant position within the internal market or in a substantial part of in so far as it may
affect trade between Member States.
4
According to settled case law the concept of abuse of a dominant position is an objective concept relating to
the behaviour of an undertaking in a dominant position which, on a market where the degree of competition
is already weakened precisely because of the presence of the undertaking concerned, through recourse to
methods different from those governing normal competition, has the effect of hindering the maintenance of
the degree of competition still existing in the market or the growth of that competition. See Case 85/76
Hoffmann-La Roche, paragraph 91; Case 322/81 Michelin I, paragraph 70; Case 62/86 Akzo, paragraph 69.
1
therefore serves as a “filter” allowing competition authorities to focus on conduct that
is most likely to harm competition.
4. Dominance has been defined under European Union law as a position of economic
strength enjoyed by an undertaking, which enables it to prevent effective competition
being maintained on a relevant market, by affording it the power to behave to an
appreciable extent independently of its competitors, its customers and ultimately of
consumers.5
5. This notion of independence is related to the degree of competitive constraint exerted
on the undertaking in question. Dominance entails that these competitive constraints
are not sufficiently effective and hence that the undertaking in question enjoys
substantial market power over a period of time.
6. The presence of dominance does not mean that all opportunity for competition in the
market is eliminated. Dominance may be found even if there is some degree of
residual competition provided that the investigated undertaking is able to act without
having to take account of such competition in its market strategy and without suffering
detrimental effects from such behaviour.6 Therefore, the fact that there may be
competition on the market is a relevant but not a decisive factor for determining
whether a dominant position exists. 7 On the other hand, the fact that an undertaking is
compelled by the pressure of its competitors' price reductions to lower its own prices
is in general incompatible with the independent conduct which is the hallmark of a
dominant position. In that case the undertaking concerned is likely to be subject to
effective competitive constraints, which is incompatible with the existence of
substantial market power.
7. Higher than normal profits may be an indication of a lack of competitive constraints
on an undertaking. More in general, the way in which a firm acts in a market may in
itself be indicative of substantial market power, for instance where an undertaking
increases its price while benefiting from falling costs. However, an undertaking's
economic strength cannot be measured by its profitability at any specific point in time;
even short-run losses are not incompatible with a dominant position.8
II.
ASSESSMENT OF DOMAINCE
8. The existence of a dominant position may derive from several factors which, taken
separately, are not necessarily determinative.9
5
See Case 27/76 United Brands Company and United Brands Continental v Commission [1978] ECR 207,
paragraph 65; Case 85/76 Hoffmann-La Roche & Co. v Commission [1979] ECR 461, paragraph 38.
6
Case 85/76 Hoffmann-La Roche, paragraph 70.
Judgment of the General Court of 30 January 2007, France Télécom v Commission, T-340/03,
ECLI:EU:T:2007:22, paragraph 101.
See Case 27/76 United Brands, cited in footnote 5, paragraph 126, and Case 322/81, NV Nederlandsche
Banden Industrie Michelin v Commission (Michelin I) [1983] ECR 3461, paragraph 59.
See Case 85/76 Hoffmann-La Roche, cited in footnote 5, paragraph 39.
7
8
9
2
9. The assessment of dominance will take into account the competitive structure of the
market, and in particular the following factors: (i) constraints imposed by the existing
supplies from, and the position on the market of, actual competitors (the market
position of the dominant undertaking and its competitors), (ii) constraints imposed by
the credible threat of future expansion by actual competitors or entry by potential
competitors (expansion and entry), (iii) constraints imposed by the bargaining strength
of the undertaking's customers (countervailing buyer power).
Market position of the dominant undertaking and its competitors
10. The analysis of the market position of the allegedly dominant undertaking and its
rivals provides insight into the degree of actual competition on the market. The
starting point for this analysis is the market shares of the various players. Market
shares provide useful first indications of the market structure and of the competitive
importance of various undertakings active on the market. If the undertaking concerned
has a high market share compared to other players on the market, it is an indication of
dominance, provided that this market share has been held for some time.10 If market
shares have fluctuated significantly over time, it is an indication of effective
competition. However, this is only true where fluctuations are caused by rivalry
between undertakings on the market. Fluctuations caused, for instance, by mergers are
not in themselves indicative of such rivalry.
11. Although current market shares are indicative, historic market shares may be used if
market shares have been volatile, for instance when the market is characterised by
large, lumpy orders. Changes in historic market shares may also provide useful
information about the competitive process and the likely future importance of the
various competitors, for instance, by indicating whether firms have been gaining or
losing market shares. In any event, market shares should be interpreted in the light of
likely market conditions, for instance, whether the market is highly dynamic in
character and whether the market structure is unstable due to innovation or growth.
12. The strength of any indication based on market share depends on the facts of each
individual case. Low market shares are generally a good proxy for the absence of
substantial market power. Experience suggests that dominance is not likely if the
undertaking's market share is below 40 % in the relevant market. However, there may
be specific cases below that threshold where competitors are not in a position to
constrain effectively the conduct of a dominant undertaking, for example where they
face serious capacity limitations. Such cases may also deserve attention.
13. Experience suggests that the higher the market share and the longer the period of time
over which it is held, the more likely it is that it constitutes an important preliminary
indication of the existence of a dominant position and, in certain circumstances, of
possible serious effects of abusive conduct, justifying an intervention. However, as a
matter of principle, it is not advisable to come to a final conclusion as to whether or
not a case should be pursued without examining all the factors which may be
sufficient to constrain the behaviour of the undertaking.
10
See Case 27/76 United Brands, cited in footnote 5, paragraph 111 and Case 85/76 Hoffmann-La Roche cited
in footnote 5, paragraph 41.
3
14. The relevance of market shares may be qualified by an analysis of the degree of
product differentiation in the market. Products are differentiated when they differ in
the eyes of consumers for instance due to brand image, product features, product
quality, level of service or the location of the seller. The level of advertising in a
market may be an indicator of the firms’ efforts to differentiate their products. When
products are differentiated the competitive constraint that they impose on each other is
likely to differ even where they form part of the same relevant market. Substitutability
is a question of degree. In assessing the competitive constraint imposed by rivals, it
must therefore be taken into account what is the degree of substitutability of their
products with those offered by the allegedly dominant undertaking. It may be that a
rival with 10% market share imposes a greater competitive constraint on an
undertaking with 50% market share than another rival supplying 20% of the market.
This may for instance be the case where the undertaking with the lower market share
and the allegedly dominant undertaking both sell premium branded products whereas
the rival with the larger market share sells a bargain brand.
Barriers to expansion and entry
15. Competition is a dynamic process and an assessment of the competitive constraints on
an undertaking cannot be based solely on the existing market situation. If the barriers
to expansion faced by rivals and to entry faced by potential rivals are low, the fact that
one undertaking has a high market share may not be indicative of dominance.
16. The potential impact of expansion by actual competitors or entry by potential
competitors, including the threat of such expansion or entry, is also relevant. An
undertaking can be deterred from increasing prices if expansion or entry is likely,
timely and sufficient.
17. Expansion or entry could be considered likely if it is sufficiently profitable for the
competitor or entrant. When assessing whether or not expansion or entry would be
profitable, the likely evolution of the market should also be taken into account.
Expansion or entry is more likely to be profitable in a market that is expected to
experience high growth in the future relative to a market that is expected to decline or
stagnate.
18. Whether expansion or entry is profitable depends in particular on the cost of (efficient)
expansion or entry and the likely prices post expansion or entry. The higher the cost of
expansion or entry and the lower the likely post expansion or entry prices, the greater
the risk that expansion or entry will be unprofitable and therefore not attempted. On
the other hand, entry may be particularly likely if suppliers in other markets already
possess production facilities that could be used to enter the market in question, thus
reducing the sunk costs of entry.
19. The prices post expansion or entry depend firstly on the impact on prices of the
additional output put on the market by the expansion or by the new entrant, but also on
the reaction of incumbents, in particular the allegedly dominant undertaking. Likely
strategic responses from the incumbents could be relevant for the assessment. An
aggressive competitive response from incumbents would be particularly likely if they
have committed to large excess capacity. The allegedly dominant undertaking may
also have built a reputation of responding aggressively to expansion or entry.
4
20. For expansion or entry to be considered timely, it must be sufficiently swift to deter or
defeat the exercise of substantial market power.
21. For expansion or entry to be considered sufficient, it cannot be simply small-scale
entry, for example into some market niche, but must be of such a magnitude as to be
able to deter any attempt to increase prices by the putatively dominant undertaking in
the relevant market.
22. The history of the industry can be relevant for the assessment of barriers to expansion
or entry. Significant barriers to expansion and entry are unlikely to be found in an
industry that has experienced frequent and successful examples of entry. On the other
hand if previous attempts to expand in or enter into the market have been
unsuccessful, perhaps due to deterring behaviour by incumbents, then expansion and
entry would seem less likely to have constituted an effective constraint.
23. When identifying possible barriers to expansion and entry it is important to focus on
whether rivals can reasonably replicate circumstances that give advantages to the
allegedly dominant undertaking. Barriers to expansion and entry can have a number of
origins relating to the legal or economic environment that pertains on the relevant
market:
-
Legal barriers: the legislative framework covering the relevant market can be
an important barrier. Such legislation may limit the number of market
participants, for example by granting special or exclusive rights in the shape of
concessions, licenses or intellectual property rights. Legislative measures that
grant a single undertaking the exclusive right to perform a certain activity
excludes rivals and may lead to such an undertaking having a legal monopoly
in a relevant market. Planning laws and licensing laws that impose limits on
the number of retail outlets limits expansion possibilities of existing and entry
possibilities for new retailers, which in turn may make it more difficult for
suppliers to gain access to efficient distribution. Intellectual property rights
may also prevent expansion and entry or make it more difficult. However,
intellectual property rights do not as such confer dominance on the holder. The
impact of intellectual property rights on expansion and entry depends on the
nature and actual strength of the intellectual property right held by the
allegedly dominant undertaking. Finally, also tariff and non-tariff barriers can
give advantages to incumbent firms.
-
Capacity constraints: competitors may have to commit large sunk investments
in order to expand capacity. An investment or cost is sunk when it cannot be
recovered if the undertaking exits the market. Moreover, even existing excess
capacity may be so expensive to employ that these costs constitute a barrier to
expansion; for instance, the costs of introducing another shift in a factory may
constitute a barrier to expansion.
-
Economies of scale and scope: large-scale production or distribution may give
the allegedly dominant undertaking an advantage over smaller competitors.
Scale and scope economies result from the spreading of fixed costs over larger
output or a broader set of products, leading to a reduction of average costs.
When economies of scale or scope are important and require a substantial
5
production capacity compared to the size of the market, efficient expansion or
entry is more costly and risky. Large fixed costs have to be committed and the
output produced will constitute a significant increase in output, which is likely
in itself to have a significant impact on price post expansion or entry. If
expansion or entry occurs at an inefficient scale, the competitive constraint
imposed on the incumbents will be less effective. In assessing barriers to
expansion and entry it is therefore useful to consider the minimum efficient
scale in the market concerned. The minimum efficient scale is the level of
output required to minimise average cost, exhausting economies of scale.11
11
-
Absolute cost advantages: these include preferential access to essential
facilities, natural resources, innovation and R&D, intellectual property rights
and capital conferring a competitive advantage on the allegedly dominant
undertaking, which makes it difficult for other undertakings to compete
effectively. In the large majority of cases financial strength is unlikely to be an
issue. However, in some cases it may be one of the factors that contribute to a
finding of a dominant position, in particular in those cases where (i) finance is
relevant to the competitive process in the industry under review; (ii) there are
significant asymmetries between competitors in terms of their internal
financing capabilities; and (iii) particular features of the industry make it
difficult for firms to attract external funds.
-
Privileged access to supply: the allegedly dominant undertaking may be
vertically integrated or may have established sufficient control or influence
over the supply of inputs that expansion or entry by smaller rival firms may be
difficult or costly.
-
A highly developed distribution and sales network: the allegedly dominant
undertaking may have its own dense outlet network, established distribution
logistics or wide geographical coverage that would be difficult for rivals to
replicate.
-
The established position of the incumbent firms on the market: it may be
difficult to enter an industry where experience or reputation is necessary to
compete effectively, both of which may be difficult to obtain as an entrant.
Factors such as consumer loyalty to a particular brand, the closeness of
relationships between suppliers and customers, the importance of promotion or
advertising, or other reputation advantages will be taken into account.
Advertising and other investments in reputation are often sunk costs which
cannot be recovered in the case of exit and which therefore make entry more
risky.
-
Other strategic barriers to expansion or entry: these encompass situations
where it is costly for customers to switch to a new supplier. This may for
example be the case where personnel have been trained to use the product of
the allegedly dominant undertaking or where due to network effects the value
Scale economies are normally exhausted at a certain point. Thereafter average costs will stabilise and
eventually rise due to diseconomies of scale and scope.
6
of rivals’ products are lower because they do not have a large installed base of
customers. For instance the value of a piece of software may not only depend
on the intrinsic qualities of the product but also on how many people use it and
thus with whom the new buyers can exchange files. Finally, the incumbent
firms may through the use of long-term contracts with customers have made it
difficult for rivals at a particular point in time to find a sufficient number of
customers able to switch supplier and therewith prevent that expansion or entry
would be profitable.
Market position of buyers
24. The market position of buyers provides an indication of the extent to which they are
likely to constrain the allegedly dominant undertaking. However, given the fact that
dominance is assessed in relation to a market, it is not sufficient that certain strong
buyers may be able to extract more favourable conditions from the allegedly dominant
undertaking than their weaker competitors. The presence of strong buyers can only
serve to counter a finding of dominance if it is likely that in response to prices being
increased above the competitive level, the buyers in question will pave the way for
effective new entry or lead existing suppliers in the market to significantly expand
their output so as to defeat the price increase.12 In other words, the strong buyers
should not only protect themselves, but effectively protect the market.
25. On the other hand, if one or more strong buyers are able to extract more favourable
conditions from the allegedly dominant undertaking than their weaker competitors, it
may be appropriate to define separate relevant markets for, respectively, strong and
weak buyers.13
CONCLUSION
26. Under European Union law the assessment of dominance is a crucial first step for the
assessment of anticompetitive unilateral conduct. In practice, different market factors,
amongst which market shares, play a role in answering the question whether market
power is significant and whether a dominant position exists in an individual case. It is
up to the competition authority to thoroughly examine all these factors in order to
come to a well-reasoned conclusion in its enforcement actions.
12
See in this respect Case T-228/97, Irish Sugar plc v Commission [1999] ECR II-2969, paragraph 101.
13
See Commission Notice on the definition of the relevant market cited in footnote 11, paragraph 43.
7
Korea Fair Trade Commission
UCWG WORKBOOK: 2015-16 SESSIONS(KFTC)
Session #1: What Is Dominance/Substantial Market Power?
•
What knowledge or law should be drawn upon to understand the concept of “dominance/substantial
market power”?
The Monopoly Regulation and Fair Trade Act (hereafter MRFTA) of Korea, Enforcement
Decree of The Monopoly Regulation and Fair Trade Act and Guidelines for Review of the
Abuse of Market Dominant Position stipulate the concept of “dominance”.
•
Exactly what is meant when we say that a company is “dominant” or has “substantial market
power”?
Under MRFTA Article Article 2 No. 7, the term “market dominant undertaking” means the
undertaking that holds market position with which it, as a supplier or purchaser,
individually or jointly with other undertaking, may determine, maintain, or change the
price, quantity, or quality of goods or services, or other trading condition in a certain line of
trade.
•
What are the important considerations in calibrating the height of the dominance bar?
Whether an enterpriser is a market-dominating enterpriser is determined by examining the
market share as well as whether there is a barrier to entering the market; ditto for its degree,
relative scale of competing business, possibility of competitors working jointly, similar
products, existence of adjacent market, and market freezing ability and resources.
Among the factors to be considered to determine market dominant position, market share is
the most important. Under MRFTA Article 4 (Presumption of Market Dominant
Undertaking), the undertaking that has one of the following market shares in a certain line
of trade (except an undertaking whose the amount of annual sales turnover or the amount of
annual purchases in a certain line of trade is less than 4 billion won) is presumed as the
market dominant undertaking in Article 2 No. 7.
1. The market share of one undertaking is 50/100 or more.
2. The combined market share of three or fewer undertakings is 75/100 or more. In this
case, however, undertakings that have market share below 10/100 shall be excluded.
1
•
Should the dominance bar be set high or low?
Why?
Not applicable
•
If there was to be market-share safe harbor, what should it be?
Why?
Not applicable
•
Is there a tradeoff between the height of the dominance bar and the standard of proof for
abuse?
Not applicable
2
Session #2: What Makes Conduct Exclusionary?
•
What does it mean to assess the competitive effects of conduct?
in different cases?
Can it mean different things
The competitive effects of conduct mean different things in different cases. More
specifically, Guidelines for Review of the Abuse of Market Dominant Position delineate the
review standards for examining anticompetitive effects as followings:
A. General Principle
(1) Whether the conduct in question causes anti-competition effects under MRFTA Article 3-2
shall be determined by taking account of the following factors which are not exhaustive.
Article 3-2 (Prohibition of the Abuse of Market Dominant Position)
(1) Any market dominant undertaking shall not engage in any of the following conducts (hereinafter “abusive
conducts”):
1. Unjustly determining, maintaining, or changing the price of goods or services (hereinafter “price”).
2. Unjustly controlling the sales of goods or the supply of services.
3. Unjustly hindering the business activity of other undertaking.
4. Unjustly impeding new competitors’ market entry.
5. Transacting with the purpose of unjustly excluding competitors or unjustly and substantially impairing
consumers’ interest.
(2) In determining anti-competition effects, the methods of comparing to similar market or
adjacent market, or comparing with the but-for market situation wherein the conduct in
question does not exist, can be used.
(3) The factors are not competing each other. Two or more anticompetitive effects may occur
at the same time. A factor can be the cause or result of other one. For instance, the decrease in
diversity of goods or service may lessen competition in a certain line of trade, and thereby
causing increase of price or decrease of output.
B. Increase of Price or Decrease of Output
(1) It should be taken into account whether the conduct in question is likely to, or actually,
cause the increase of price or the decrease of output of goods or services in a certain line of
trade or goods or services in the adjacent markets.
(2) It should be noted that it may take a considerable period of time to occur the actual
increase of price or decrease of output in the relevant market; such period depends on the
characteristic of industry or the conduct in question.
(3) Increase of price or decrease of output can be the result of other anticompetitive effects
such as foreclosure or raising rivals’ cost.
C. Limitation of diversity of goods or services
(1) It should be taken into account whether the opportunity of buying diverse goods or
services, the opportunity of buying cheaper goods or services that are complementary or
(actually or potentially) competitive with the products by a market dominant undertaking, is
limited.
(2) Decrease in the number of competitors due to the conduct in question may cause limitation
of diversity of goods or services that can be chosen by buyers.
(3) Limitation of diversity of goods or services can be the result of other anticompetitive
effects such as impediment of innovation
3
D. Impediment of Innovation
(1) It should be taken into account whether the conduct in question impedes the incentive to
innovate technology, research and development, services, or quality that are beneficial to
consumers.
(2) Impediment of Innovation can be the result of other anticompetitive effects such as
foreclosure or raising rivals’ cost.
E. Foreclosure Effect
(1) It should be taken into account whether the opportunity of market entrance or expansion of
competitors is, or likely to be, foreclosed. Foreclosure may decrease the number of effective
competitors, lessen competition, and thereby increase price, decrease output, or limit the
diversity of goods or services.
- In examining a degree of foreclosure effect, it should be taken account of dynamic
aspects, such as the extent of exclusion or difficulty of competitors’ access due to the conduct
in question; the impact of foreclosure on the growth of competitors (including potential
competitors) and new entrance, changes in market share, and refutation of the conductor in
question when it trade with other undertakings.
(2) It should be noted that it may take a considerable period of time to occur the actual
foreclosure effect in the relevant market; such period depends on the characteristic of industry
or the conduct in question.
F. Raising Rivals’ Costs
(1) It should be taken into account whether the conduct in question raise or is likely to raise
(actual or potential) rivals’ costs. As raising rivals’ costs lessens the competition pressure of a
market dominant undertaking, it may cause anticompetitive effects, such as increase of price,
decrease of output, limitation of diversity of goods or services, or impediment of innovation
in a certain line of trade or adjacent market.
(2) In determining the effect of raising rivals’ cost, it should be taken into account whether the
conduct in question forecloses competitors from efficient distribution or supply, hinders
competitors’ acquisition of resources necessary for production or distribution, or creates
artificial barriers to entry, and consequently competitors suffer a considerable cost and period
of time to overcome such obstacles.
•
Can the character of conduct ever be such that it necessarily constitutes lawful competition on
the merits? If so, what conduct? Why?
Not applicable
•
Can the character of conduct ever be such that it is necessarily abusive?
Why?
If so, what conduct?
Not applicable
•
Should a welfare standard be used in developing the rules or analytical framework for
particular practices? Should an error-cost analysis be used, and if so, how are error costs
evaluated?
4
The subparagraph 5 of the Article 3-2 (1) of the MRFTA defines conducts that would unjustly and
substantially impair consumers’ interest as one of the types of abuses of market dominant enterprisers.
However, the enforcement decree of the Act and guidelines do not stipulate specific welfare standards for the
rules or analytical framework for particular practices. Though, in its precedent rulings on some welfare-related
cases, the Supreme Court of Korea set forth standards below.
Whether a conduct of market dominant enterprisers would unjustly and substantially impair consumers’
interest should be determined after proving that a conduct with a risk of harming consumers’ interest exists;
that the harms consumers would suffer are substantial; and that the conduct at issue is unjust. Whether the
conduct at issue has a risk of ‘substantially’ impairing consumers should be determined in a specific and on a
case by case manner, after taking into comprehensive considerations of various factors. They include the
character of the goods or services at issue; the scope of consumers’ interest impaired; trade conditions of other
enterprisers in the similar markets; the degree of changes of dominant enterpriser’s cost after changes
including trade condition changes are applied; and the gap between the prices and values of goods and
services at issue. (The Supreme Court’s decision on CJ Hellovision case in March 2010)
Whether the conduct at issue is unjust should be determined after taking into account specific factors
including whether there is a reason to see that the intent or the purpose of the market dominant enterpriser’s
practice is getting excessive benefits of its monopoly power, or whether the conduct in question caused or has
a risk of causing impairing consumers’ interest in the market at issue based on the character of the product, the
character of the practice at issue, the period during which the practice took place, the structure and character of
the market. (The Supreme Court’s decision on TBroad II case in May 2010)
Should a welfare standard be used to assessing the facts of particular cases?
done only in some cases?
If so, in which cases?
Not applicable
5
Why?
Should this be
Comisión Federal de Competencia Económica
(COFECE)
COFECE’s submission, following to the UCWG’s call on papers related to the drafting of
Chapter 2 of the Unilateral Conduct Workbook: “The Analytical Framework for Evaluating
Unilateral Conduct”.
Exactly what is meant when we say that a company is “dominant” or has “substantial
market power”?
Market power, that is to say, a firm’s ability to set prices above competitive levels 1 and/or to
restrict output 2, is a matter of degree. Indeed, except for those firms whose interaction falls
within the framework of perfect competition, all firms hold a certain degree of market power 3.
Hence, the key word is the adjective “substantial”. Nonetheless, drawing a line between what
constitutes “normal” market power and substantial market power (SMP) is not an easy task, since
there is no one-size fits-it-all formula.
Intuitively, for a firm to enjoy SMP, it must be able to set prices above a competitive level during
a meaningful period, without being disciplined by other economic agents –for example, direct
competitors, providers or consumers –. This approach is close in spirit to the strategic
management framework known as “The five forces” 4.
It goes without saying that
SMP
is never possessed or exerted in a vacuum. Firms possess
SMP
within a relevant market 5. In this regard, market share analysis can be a first step, but it can fail to
capture the broad range of factors that determine
SMP
6
. This is so, because, as it was previously
highlighted, the competitive constraints that a firm faces are not only horizontal. Hence, an undue
emphasis on market shares can provide an unclear picture of the actual degree of SMP.
1
In its simplest form, the competitive price level is considered to be marginal cost. Cfr. Cabral, L. (2000)
Introduction to Industrial Organization. The MIT Press.
2
It must be highlighted that when prices are regulated, firms can only set quantities. In that scenario, it is more
convenient to think in terms of output.
3
Kaplow, L. & Shapiro, C. (2007) Antitrust. NBER Working Paper No. 12867
4
Porter, M. (2008). The five forces that shape strategy. Harvard Business Review.
5
We will not delve into what is a relevant market, as such an exposition is addressed in substantial bibliography.
Suffice to say that it “should be defined in a way such that the competitive constraints a firm faces, i.e. demand and
supply side substitution, are captured as accurately as possible”. Cfr. OECD Policy Roundtables (2012) Market
definition.
6
ICN’s Unilateral Conduct Working Group (2011) Unilateral Conduct Workbook Chapter 3: Assessment of
dominance.
1 de 2
It is useful to consider barriers 7 when assessing the existence of
albeit not a sufficient condition for the existence of
SMP.
SMP.
Barriers are a necessary,
In particular, there can be industries
where the ease of entry is low, but where, due to structural factors (e.g. high bargaining power of
consumers vis – a – vis firms), incumbents are unable to charge large markups. It is difficult, if
not impossible, to think of a case where a firm holds
SMP
and where there are no significant
barriers.
Operationally, SMP is relevant insofar as it works as a filter to establish harm in unilateral conduct
cases. Commercial strategies carried out by firms without
SMP
are considered unlikely to be
harmful. However, the mere possession of SMP is not unlawful; only the abuse of SMP. It is key to
understand the means by which a firm obtained
SMP
8
. Therefore, a priori, unlawfulness should
not be attached to the holding of SMP.
7
In this regard, we don’t endorse any particular definition. “[T]hese terms are intended simply to denote something
that is relevant to entry analysis because it tends to delay or prevent entry.” Cfr. OECD Policy Roundtables (2005)
Barriers to entry.
8
Cfr. OECD Policy Roundtables (2005) Competition on the merits.
2 de 2
Alden Abbott
Defining Market Dominance under Competition Law
By Alden F. Abbott, Heritage Foundation, U.S. NGA 1
In response to the call of the International Competition Network’s (ICN) Unilateral Conduct
Working Group (UCWG), I am pleased to present one United States NGA’s perspective on
defining market dominance for purposes of competition law. As a general matter, a finding of
dominance is a prerequisite to a determination of whether a single firm has abused a dominant
position in violation of competition law norms. My essay is informed by United States antitrust
jurisprudence, which employs the term “monopoly power” rather than market dominance. I
believe that the experience of the United States in determining whether a firm has monopoly
power under federal antitrust law, rooted in substantial experience and sound economics, serves
as a useful template for assessments of dominance. Accordingly, while this essay employs the
American terminology of monopoly power, I submit that it is entirely applicable to discussions
of “dominance” within the ICN. 2
I.
Overview
Market power is a seller’s ability to exercise some control over the price it charges. Virtually all
products that are differentiated from one another, if only because of consumer tastes, seller
reputation, or producer location, convey upon their sellers at least some degree of market power.
Thus, a small degree of market power is very common and understood not to warrant antitrust
intervention.
Market power and monopoly power are related but not the same. The U.S. Supreme Court has
defined market power as “the ability to raise prices above those that would be charged in a
competitive market,” and monopoly power as “the power to control prices or exclude
competition.” Precisely where market power becomes so great as to constitute what the law
deems to be monopoly power is largely a matter of degree rather than one of kind. Clearly,
1
The views set forth herein are solely attributable to the author.
This essay draws upon U.S. Department of Justice, COMPETITION AND MONOPOLY: SINGLE-FIRM CONDUCT
UNDER SECTION 2 OF THE SHERMAN ACT, ch. 2 (Sept. 2008), available at
http://www.justice.gov/sites/default/files/atr/legacy/2009/05/11/236681.pdf. Although the Justice Department
subsequently withdrew this report in 2009, that withdrawal centered on disagreement with certain report policy
recommendations for determining whether particular single firm conduct is illegal. The withdrawal was not
motivated by the report’s assessment of dominance itself, which reflected a great deal of consensus research. See
Remarks of Assistant Attorney General Christine Varney Before the U.S. Chamber of Commerce Regarding Vigorous
Antitrust Enforcement in this Challenging Era (May 12, 2009) (the withdrawn report “collect[ed] and evaluat[ed]
the opinions and expertise of antitrust enforcement officials from the United States and abroad, leading
economists and legal scholars, antitrust practitioners, and representatives of the business community” and
“provided a comprehensive evaluation of the history of single-firm enforcement and careful consideration of the
risks and benefits of particular enforcement strategies”), available at http://www.justice.gov/atr/speech/vigorousantitrust-enforcement-challenging-era.
2
however, monopoly power requires, at a minimum, a substantial degree of durable (non-fleeting)
market power.
Although monopoly power will generally result in the setting of prices above competitive levels,
the desire to obtain profits that derive from a monopoly position provides a critical incentive for
firms to invest and create the valuable products and processes that drive economic growth. For
this reason, U.S. antitrust law does not regard as illegal the mere possession of monopoly power
where it is the product of superior skill, foresight, or industry. Only anticompetitive conduct is
prohibited.
Monopoly power is conventionally demonstrated by showing that both (1) the firm has a high
share of a relevant market and (2) there are entry barriers—perhaps ones created by the firm’s
conduct itself—that permit the firm to exercise substantial market power for an appreciable
period. Unless these conditions are met, a firm is unlikely to have either the incentive or ability
to exclude competition.
II.
Market Share and Durability of Market Power
In determining whether a competitor possesses monopoly power in a relevant market, U.S. courts
typically have required a dominant market share before inferring the existence of monopoly
power. As a practical matter, a market share of greater than fifty percent has been necessary for
U.S. courts to find the existence of monopoly power. No U.S. court that has found that a
defendant possessed monopoly power when its market share was less than fifty percent.
Although a dominant market share is a useful starting point in determining monopoly power,
modern U.S. judicial decisions consistently hold, however, that proof of monopoly power
requires more than a dominant market share. Even a very high share does not guarantee
substantial power over price for a significant period: if smaller fringe firms in the market can
readily and substantially increase production at their existing plants in response to a small
increase in the large firm’s price (that is, if the fringe supply is highly elastic), a decision by the
large firm to restrict output may have no effect on market prices. However, even if fringe firms
cannot readily and substantially increase production, a firm with a very high market share is still
not guaranteed substantial power over price if the quantity demanded decreases significantly in
response to a small price increase—in other words, if market demand is highly elastic. That is,
when demand is elastic, a firm may be unable to raise price without losing so many sales that it
will prove to be an unprofitable strategy. Furthermore, in markets characterized by rapid
technological change, any power a firm may have may be both temporary and essential to the
competitive process. Indeed, in the extreme case, market structure may be a series of temporary
monopolies in a dynamically competitive market. Entry conditions are also important.
Specifically, if there are no significant barriers to entry, a very high market share may not allow
a firm to price significantly above cost, if it appears that new firms would quickly enter the
market and render such pricing unsustainable.
Notwithstanding that a high share of the relevant market does not always mean that monopoly
power exists, a high market share should be one of the most important factors in the examination
of whether a firm has monopoly power. A high share indicates that it is appropriate to examine
other relevant factors, such as barriers to entry. For example, if a firm has maintained a very
large market share (say, over 70 percent) for a significant time period, and market conditions (for
instance, high barriers to entry) are such that the firm’s market share is unlikely to be eroded in
the near future, an enforcement agency may wish to view such evidence as a rebuttable
presumption that the firm possesses monopoly power.
Durability of market power is reflected in the ability to engage in supracompetitive pricing for a
sustained period of time. In that regard, low barriers to entry, or the existence of highly elastic
demand, may effectively preclude a firm with a high market share from maintaining long-term
anticompetitive pricing levels.
III.
Relevant Market Definition
The market-definition requirement brings discipline and structure to the monopoly power
inquiry, thereby reducing the risks and costs of error. The relevant product market is defined
with regard to demand substitution, which focuses on buyers’ views of which products are
acceptable substitutes or alternatives. In examining proposed mergers, U.S. antitrust agencies
traditionally have looked at the profitability of a “hypothetical monopolist’s” small but
significant non-transitory price increase in setting market boundaries. In adapting this approach
for examining potential monopolists, however, enforcers should keep in mind that a firm may
already be charging a monopoly price, and, thus, may be exercising substantial market power
even if a price increase would be unprofitable (whether a firm is currently charging at a
monopoly level involves a factual inquiry). In general, although there may be no reliable
paradigm for defining the relevant market in every case, courts often are able to draw sound
conclusions about the relevant market based on the facts and circumstances of the industry.
Some commentators have argued that market definition could be avoided and monopoly power
demonstrated instead solely through direct evidence – for example, proof of high profits.
Relying exclusively on direct evidence of profits to establish monopoly power, however,
presents a number of difficult issues. High accounting profits do not necessarily reflect the
exercise of monopoly power. In particular, cost measures are normally available only from
reports prepared in conformity with accounting conventions, but economics and accounting have
significantly different notions of cost. In addition, determining if a firm is earning a profit
reflecting the exercise of monopoly power should take into account the opportunity cost of
employing those assets in their current use, which accounting fails to do. Moreover, available
estimates of a firm’s capital costs, an important input into calculating a firm’s profitability, are
generally based on accounting rules that do not account for the riskiness of the investment. More
generally, when all relevant economic costs are properly accounted for, what may at first seem to
be a supracompetitive return may be no more than a competitive one (or vice versa).
IV.
Direct Evidence of Firm Margins and Demand Elasticities
Using price-cost margins, rather than profits, as evidence of monopoly power is also
unsatisfactory. Monopoly power requires that the firm be able profitably to charge prices high
enough to earn a supranormal return on its investment. It is not clear how much price must
exceed short-run marginal cost before there is monopoly power. Depending on the size of the
firm’s fixed costs, even a significant margin between price and short-run marginal cost may be
insufficient to earn even a normal return. Indeed, a firm should not be found to possess
monopoly power simply because it prices in excess of short-run marginal cost and hence has a
high price-cost margin. In principle, a better measure of margin would be the ratio of price to the
firm’s long-run marginal cost, but such data is unlikely to be available. Furthermore, as an
indicator of monopoly power, demand elasticities suffer from the same fundamental problem that
margins do: neither tell us whether the firm is earning durable, supernormal profits. (Short-run
high demand elasticities may merely indicate significant product differentiation, not monopoly.)
In short, direct evidence of a firm’s profits, margins, or demand elasticities is not likely to
provide an accurate or reliable alternative to the traditional approach of first defining the relevant
market and then examining market shares and entry conditions when trying to determine whether
the firm possesses monopoly power.
V.
Direct Evidence of Anticompetitive Effects
If a dominant firm’s conduct has been demonstrated to cause competitive harm, one could rely
simply on that evidence and dispense with the market definition requirement entirely. However,
there are concerns with taking such an approach. One important concern is that effects evidence,
while very valuable, is generally imperfect, and sometimes subject to differing interpretations.
For this reason, also requiring a traditional market-definition exercise – incorporating, perhaps,
available evidence of alleged effects – likely adds value by strengthening inferences and thereby
avoiding potentially costly errors. In some circumstances, an inability to find any
anticompetitive effects may serve as a useful screen, enabling courts or enforcement officials to
conclude quickly that an abuse of a dominant position is implausible. In other cases, there may
be effects evidence strongly suggestive of harm and the existence of a relevant market that has
been subjected to abuse a dominant firm.
Ricardo Mederios de Castro
Concept of dominance:
dichotomy and continuity divide
Ricardo Medeiros de Castro
Simple dominance (and simple market power), collective dominance1 (and collective
market power), super dominance2 (and super market power): generally, Antitrust Authorities use
these terms to determine who has a certain market position [and may abuse this privileged status].
In the United States, finding of monopolization in violation of section 2 of the Sherman Act may
require an initial determination that the defendant has “monopoly power” or, in other words, a
high degree of market power.
Sometimes, market shares are used as a first approximation of market power. In an
attempt to economize decision costs, many Agencies around the world use such approximation:
Region
Presumption of market power
Legal basis
Singapore
60% market share
Singapore Guidelines on Section 47 Prohib. §3.8 - 2007
Israel
50% market share
Israel Restrictive Trade Practices § 26
European Union
50% market share
Case 62/86, AKZO v Commission, para 60
China
50% market share
China Antimonopoly Law – art.19
Russia
50% market share
Russia Competition Law – art. 5
Indonesia
50% market share
Indonesia Competition Law – art.17
South Korea
50% market share
South Korea Fair Trade Act – art. 4
Taiwan
50% market share
Taiwan Fair Trade Act – art. 5-1
South Africa
45% market share
South Africa Competition Act § 7
Saudi Arabia
40% market share
Saudi Arabia Executive Regulation art. 7(a)
Poland
40% market share
Act on competition and consumer protection art.4(10)
Serbia
40% market share
Article 15 of the Competition Act
Ukraine
35% market share
Ukraine Competition Law – art. 12.2
Egypt
25% market share
Egypt Competition Law art.4
Brazil
20% market share
Brazilian Competition Law (12.529/11 – art.36 §2 )
Table 1 – Some jurisdictions that use market share as a presumption of market power
Source: (ICC, 2008) (ELHAUGE & GERADIN, 2011) (CONCURRENCES.COM)3
Brazil, in this list, has the lowest [but rebuttable] threshold to presume market power.
The International Chamber of Commerce (ICC) understands that the presumption of
market power based on market shares is an unwise decision, especially if it is non-rebuttable. That
would occur because these presumptions would discourage “potentially procompetitive behavior
by companies wrongly presumed to have market power” (ICC, 2008, p. 1). The ICC argues, also,
1
See European Cases Alsatel v Novosam Case (247/86); Kali und Salzi Case (C-68/94 and C-30/95); Società Italiana
Vetro SpA v. Commission, Joined Cases T-68, T-77–78/89, [1992] E.C.R. II-1403. (SZYSZCZAK, 2011) Irish Sugar
case - Case T-228/97, [1999] ECR II-2969.
2 See European Case - Opinion of Advocate General Fennelly, Compagnie Maritime Belge, [2000] E.C.R. I-1365,–
UK Case - Napp Pharmaceutical Holdings Ltd Subsidiaries v. Director General of Fair Trading (SZYSZCZAK, 2011)
3
According
to
http://www.concurrences.com/Droit-de-la-concurrence/AntitrustEncyclopedia/?questions=576&lang=fr, verified in October, 26th , 2015
that the use of market share ascribes monopoly power or dominance to firms, when, in fact, they
do not possess such power.
Using similar kind of arguments, Kaplow argues that:
“market definition process should be abandoned. The central, conceptual argument is that there does not exist any coherent way to choose
a relevant market without first formulating one’s best assessment of market power, whereas the entire rationale for the market definition
process is to enable an inference about market power. Why ever define markets when the only sensible way to do so presumes an answer
to the very question that the method is designed to address? A market definition conclusion can never contain more or better information
about market power than that used to define the market in the first place. Even worse, the inferences drawn from market shares in relevant
markets generally contain less information and accordingly can generate erroneous legal conclusions — unless one adopts a purely
results-oriented market definition stratagem under which one first determines the right legal answer and then announces a market
definition that ratifies it. Additional, largely unavoidable difficulties are identified with the economic logic underlying market
redefinition. Because virtually all of the argument reveals inherent problems in the very conception of the market definition / market
share paradigm, it follows that the conclusions here do not depend on one’s assessment of the quality of various means of measuring
market power either in general or in particular cases and that they are independent of the legal application at hand. Prior criticism of the
market definition / market share paradigm is extensive”. (KAPLOW, 2010, p. 440)
If this criticism is right about the fact that market power cannot be inferred by market
shares, then, a lot of other questions arise, for example: what is the right way to define such
concept? How easy is it to find an alternative interpretation, without underestimating uncertainty?
How is it possible to define who has and who has not market power? Is it possible to make such
definition in abstract terms? What is the reason to define dominant position in first place?
Dominance necessarily is a mean of obtaining something or it can be the result of an act? In other
words: is dominance a precondition for an anticompetitive conduct or can it also be the outcome
of a specific undue practice that raises barriers of entry and modifies the conditions of a
competitive market?
By trying to find an alternative concept to the structural analysis, the European
jurisprudence attempted to use a different wording, defining dominant position as one that
“relates to a position of economic strength enjoyed by an undertaking, which enables it to prevent
effective competition being maintained on the relevant market by affording it the power to behave
to an appreciable extent independently of its competitors, its customers and ultimately of the
consumers” (see cases Hoffmann-La Roche case and United Brands case)4. Similar language is
found in several national legislations, such as the ones from Slovakia; Switzerland [Competition
Act Article 4(2)]; Turkey [Competition Act Section 3]; Czech Republic [Competition Act Art.10,
par.1] and among several others.
However, some authors understand that it is difficult to translate in economic terms how
a dominant player should be independent in regard of its own consumers and its own competitors
(MOTTA, 2004, p. 34). For O´Donoghue and Padilla “only a monopolist operating in a market
protected by insurmountable barriers to entry and facing a completely inelastic demand would
be able to behave independently of its competitors, competitive fringe and consumers”
(O´DONOGHUE & PADILLA, 2006, p. 108). The only problem is that the monopolist, in such
context, is not supposed to sell anything in the inelastic part of demand. So even this unlikely
scenario painted by O´Donoghue and Padilla may underestimate the implausibility of an
“independent dominant firm”. Also, if a dominant player is completely independent from its
competitors (with zero diversion ratio) in middlestream, downstream and upstream markets, then,
unilateral exclusionary practices would not render any benefit to this dominant player, depriving
him from the intent to execute anticompetitive actions.
Another way to express what is market power is to make some reference to a perfect and
abstract world (that resembles John Lennon´s song called Imagine): the model known as “perfect
Case 85/76 – Hoffman La Roche & Co. AG vs. Commission [1979] ECR 461, par.39. Case 27/76 United Brands v
Commission [1978] ECR 207, [1978] 1 CMLR 429.
4
competition” [where market power is absent, where there is an infinite amount of players, selling
and buying things, where perfect knowledge of everything is shared among everyone, where there
are no barriers of entry and exit whatsoever, the product negotiated is certainly homogeneous,
among other very narrow conditions]. Unfortunately, such model seems to exist only in the
imagination of those who study Economics. Moreover, Lerner Index (Lerner, 1934), as it will be
explained in this article, is the concept created to reflect the foundations of this perfect world (or
to segregate this unlikely scenario from other real ones). Such concept, however, seems to present
some loopholes when one tries to implement it.
In this debate, this paper will argue that it is truly difficult to present a closed and
uncontestable manner of screening dualistically what market power is or is not, since market
power means different things, in different situations, to different interpreters. This conclusion
seems to be not stressed enough when criticism of structural analysis appears.
In addition, it will be argued that the use of non-classical logics helps to understand why
this concept can change through time [dynamically] and depending on circumstantial features of
the case. Indeed, dominance and market power are (sometimes) simultaneously defined with
several other factors, which influence endogenously such definitions.
Therefore, this paper aims at discussing these concepts and how they can affect the
antitrust enforcement. In this context, criticism of traditional manner of market power definition
can render a better understanding of competitive pressures, and should be encouraged. However,
easy and overly simplistic criticism of market power definition based on some traditional
heuristics may underestimate the huge and inherent indetermination of the alternative approach.
That is why it is crucially important to discuss the degree of such indeterminacy.
1. Challenging classical logic
“To be or not to be? : that is the” [dichotomic] “question”, (SHAKESPEARE, 1623)
made by prince Hamlet, in hesitation to avenge his father's murder [King Hamlet]. Prince Hamlet
was wondering what would be worst: to live a long life knowing that his uncle, stepfather, and
new king Claudius murdered his father [and doing nothing against him] or to avenge his dead
father, even if such act would cost his own life. Maybe there could be a third option [not involving
violence], but it would also transform the Shakespearean classical tragedy, so appreciated in
literature, in something else.
Shakespeare knew about dichotomy of life [to be=1] (the appearance of a continuous
interval that we know) and death [not to be=0] (the end of everything familiar). Therefore, death
would mean the existence of the absolute, definite and final stage of our existence. There are
several crossroads or bifurcations in life and once someone has decided to follow, unfortunately,
it is not possible to have a second chance to reconsider. However, sometimes, it is also possible
– to avoid pain – not deciding anything at all. And that is, when Shakespeare questioned, through
the mouth of Hamlet, how courageous people are (i) to face the possibility of dichotomous bad
[but courageous] choice or (ii) to live a very long [and continuous] life comfortably without
making any bad decision, but with regret of not making it, without taking chances, and with a
little anger, bitterness or even anguish of not taking part of an important, but dualistic decision
that could have succeeded (or not).
If this existentialist approach were to be applied to Antitrust Law, one could see several
dummy (dualistic) specifications regarding market power, relevant market, conduct definitions,
burden of proof and several other subjects that would lead Antitrust enforcer to the final and
binary conclusion of what is legal or illegal, allowed or not, using something like the binary
Luhmanian code of decision (LUHMANN, Ökologische Kommunikation: Kann die moderne
Gesellschaft sich auf ökologische Gefährdungen einstellen?, 1989) (LUHMANN, 1995). And if
society makes good, conscious Antitrust decisions, it may raise the welfare of everyone, may
decrease prices and, sometimes, may allow poor people to buy medicines, to buy food, or some
other goods, that, otherwise, in the absence of such welfare, their own lives would be at stake.
And here the existentialist question appears once again.
Fighting abusive practice of concentrated structures in a concentrated world seems to
have important distributive aspects, especially if definitions of relevant markets are not just part
of an Idealistic-Hegelian debate, but may also be linked with real borders of international markets,
where immigrants try so hard to access and where there is so much inequality.
Figure 1 – Champagne-Glass Distribution (CONLEY, 2008)
It is surrounded in these concentrated international structures that practices as tying,
refusal to deal, retail price management, predatory practices, international cartels, and so many
others worldwide or regional anticompetitive conducts may be discovered and punished by only
some Authorities affected by them. Therefore, reparation of such undue practices may also be
concentrated in some areas of the world (generally rich ones), in the sense that evidences,
witnesses, and enterprise´s assets that allow the enforcement of certain antitrust decisions may,
sometimes, also be concentrated.
There may be Agencies with better screening tools of what is anticompetitive or not, or
may better deal with grey areas of Antitrust Theory. Rightful screening is important, given that
society, lawyers, judges and public servants need to know what is right (1) or wrong (0) in order
to prescribe what should be done and proscribe/punish what should not, even if it is hard to
explain, emulating a dummy variable decision.
However, dummy variables may be called dummies for simplifying the reality in a very
strong manner. Of course, simplification is a necessary rational tool to understand complex
environments, and in legal terms, clear definitions are important to regulate conducts. On the other
hand, oversimplifications could represent danger pitfalls in some situations.
In this context, substantial market power and dominance are terms that may lead an
unadvised interpreter to face some dichotomic, absolute and discrete decisions without a little bit
of restraint. And even if the careful decision maker thinks that the world is not simplistic divided
among firms that have or have not market power (believing that market power is a matter of
degree, varying continuously), it is important to face this further question: since Antitrust
Authorities should decide concrete cases, what is the specific point or frontier to draw a line
between substantial (and not substantial) market power? Or, again, what is a dominant (and not
dominant) position?
These concepts, based on negative (-) and positive (+) definitions, seem to follow a
Classical Logic of Aristotle, (LEITE, 2004) based on three main principles:

Identity principle: Everything is identical to itself.
∀ x, x ↔x

Non-contradiction principle: It is not possible to be and not to be at same time.
∀ x, ¬ ¬ ( x ) ↔ x
¬ (¬x ∧ x)

Principle of the exclusion of the middle: Everything is or is not. There is no third
option.
∀ x, ¬ x ∨ x
By using such principles, it is possible to claim, dualistically, that something is either
inside or outside the market; with or without market power; and having or having not the status
of dominant player, without considering the broader context of the conduct.
On the other hand, non-classical logic (such as dialethism, paraconsistent5 and/or
paracomplete logic) could accept the flexibility of some classical arguments (LEITE, 2004, p. 2),
allowing a reflection over these problems (and maybe rephrasing Shakespearean dilemma).
5

Dialethism – Dialethism argues that sometimes contradictions can exist and be true.
In the field of law, legal concepts are not found in physical world independently from
the point of view of the interpreter. Would it be totally illogical or incoherent that
some persons think that relevant market is well defined, using certain theories, while
others think it is not, using others [or having a different point of view]? Or would it
be impossible to disagree of what market power is? Even in Physics, the Theory of
Relativity could provide some examples of how contradictions views of reality can
exist simultaneously (such as in twin paradox (EINSTEIN, 1905), in which twins
exposed to different realities, can experience time and life itself in different manners;
or that light could assume the form of a wave without weight, in certain circumstances,
but in others be a particle with mass and weight). String Theory´s multiverse
(EVERETT, 1957) hypothesis also admits the possibility of having different
simultaneous realities. (CARR, 2007) By using these inputs from dialethism, it will
be argued on this paper, that it is possible to exist (i) valid different interpretations
about market power and (ii) situations that an enterprise can have and not have market
power, at same time, depending on a set of variables.

Paraconsistency - “A paraconsistent logic is a way to reason about inconsistent
information without lapsing into absurdity. In a non-paraconsistent logic,
inconsistency explodes in the sense that if a contradiction obtains, then everything
(everything!) else obtains, too. Someone reasoning with a paraconsistent logic can
begin with inconsistent premises—say, a moral dilemma, a Kantian antinomy, or a
semantic paradox—and still reach sensible conclusions, without completely exploding
Despite the debate about dialethism (if contradiction can or not be true), it is possible to think in terms of
paraconsistent logics, as a way to avoid “explosion” of the arguments. Explosion could occur if the
interpreter finds a contradiction, and based on that contradiction, everything else, in logical terms, falls
apart. Using this kind of thought, maybe it would be possible that two people, with different concepts about
how relevant market ought to be interpreted, may agree on what constitute market power.
into incoherence. Paraconsistency is a thesis about logical consequence: not every
contradiction entails arbitrary absurdities. Beyond that minimal claim, views and
mechanics of paraconsistent logic come in a broad spectrum, from weak to strong”6.
Using this kind of thought, maybe it would be possible that two people, with different
concepts about how relevant market ought to be interpreted, or different interpretation
of Lerner Index, or a dissonant view over any other concept, could agree on what
constitute an anticompetitive conduct in a concrete case. In other words, the
inconsistency of some terms, such as market definition, sometimes, does not explode
into absurdity the possibility of consensual inference over the existence of market
power or illegality of conducts, possibly being contained by this logical different
structure.

Other logics - There are a lot of other logics in a pluralist way of thinking, as fuzzy
logic, intuitionistic logic, modal logic, three valued logic, multiple valued logic, and
other kinds of logic, with different interpretation of how one should make inferences
about reality [and if the principle of the exclusion of the middle should prevail]. These
logics will be important to question how traditional Econometric inference, widely
used in Economics [and in Antitrust], may not be considering correctly the role of
critical values and their possible intervals (confident or credible ones). Trying to
exclude the middle, intervals of critical values are treated as points or lines, reinforcing
the dichotomic classical logic, without even considering the implications of such
choice (that ultimately deals with burden of proof), as it will be explained in some
other parts of this article.
In sum, maybe the use of other kinds of logic could help to understand some situations
that classical simple logic (and dualistic approach) tends to oversimplify about antitrust dilemmas.
The way to solve such problems is not easy or univocal, but, certainly, is not something that
should be hidden from the social, scientific and political debate.
2. Relativity
In respect specifically to unilateral anticompetitive conducts, it is not possible in abstract
terms to say that an enterprise has (or has not) market power.
This indeterminacy may occur because of (i) the possibility to perform, (ii) interests to
conclude and (iii) effects of some anticompetitive conducts may vary to some relational aspects,
and not to some absolute abstract ones. Depending on how these relations or scenarios are drawn,
it is possible to have substantial market power [in certain conditions or relations] and not to have
[on others].
So, knowing that Antitrust Law analyzes lot of conducts C={C1, C2, C3, … Cn}, regarding
several players P={P1, P2, P3, …., Pn} that occurs in specific conditions X={X1, X2, X3, … Xn}
maybe it would be important to determine the relational aspects of such variables {horizontally,
vertically and diagonally}, and not analyze – in vacuum – the concept of market power, as an
abstract precondition to every single antitrust punitive enforcement of all anticompetitive conduct
or structure [one size fits all].
Indeed, it is possible to have, inside a specific market, different companies with different
diversion ratios, with different price-cost margins, with different marketing strategies and with
different products, and so on. If an enterprise succeeds to exclude from the market its nearest
6
http://www.iep.utm.edu/para-log/
competitor through some exclusive contracts with retail sector [P1 exclude P2 performing conduct
C1], having P1 and P2 very high diversion ratios among themselves in a specific niche of the
market with high barriers of entry and low rivalry among all other players of this niche [X1 and
X2 Conditions], it is possible that this conduct could generate a great impact with a substantial
price increase, regardless of how relevant market is defined in abstract terms or the specific
amount of market share that both companies may have.
However, if the same conduct is directed to some other parts or niches of the same
relevant market [Let´s say that P1 tries to exclude P3 performing conduct C2, knowing that
diversion ration between P1 and P3 is almost zero], the same conduct may not interest P1 the same
way. So, the interest to perform an anticompetitive practice is a relational concept, since the
exclusion of the nearest competitor gives bigger payoffs (in other words, it gives a bigger reward
to the company that is able to succeed on this, and possibly will generate a greater impact in terms
of raising prices).
It is possible that P3 rests on different niche of the market with different conditions (entry
barriers, rivalry levels, capacity constraints, among others [Market conditions X3, X4, X5]) inside
the same relevant market. These conditions could interfere not only on the interest, but also on
the possibility of exclusion of some rival. If that is the case, P1, with a certain amount of market
share, can exclude P2, but not P3.
Sham litigation or fraud litigation are examples of how the capacity to exclude rival may
be linked with the conduct itself [and not necessarily with market share]. Indeed, even the smallest
player of the market can enter with a judicial or administrative claim, and legally exclude all other
players [and acquire market power] raising, subsequently, insurmountable legal market barriers
to rivals through a successfully implemented sham or fraud litigation strategy.
Furthermore, what is important and substantial market power to one conduct [C1] may
not be substantial or important to some other conduct [C2]. For example, contractual relationships
with several retailers on downstream market could be important to determine what is right or
wrong in terms of exclusive contracts. An enterprise could enforce contractual clauses to delay or
restraint the entry of a rival in a specific market through exclusive contracts with retailers [C1]
(because such enterprise has many contractual relationships). However, this same company may
have difficulties to perform a predatory pricing in the same market and targeting the same
competitors [C2], depending on how aggressive can be the sacrifice of its own profits vis-à-vis
how efficient are its own rivals in terms of production costs.
In some vertical conducts, it is not enough to measure ONE lerner index, ONE market
power, looking just to ONE relevant market. Indeed, for analyzing properly the rationality of
some vertical foreclosures, one should understand, simultaneously, the relationship between
lerner index from one layer of the market (downstream, for example) with another layer
(middlestream or upstream, for example). Therefore, analysis of profitability of both layers (upper
and lower layer) are interconnected.
For example, as mentioned in European Guidelines on the assessment of non-horizontal
mergers (2008/C 265/07), input foreclosure can depend on the measurement of lerner index in
both levels (upstream and downstream market):
The incentive to foreclose depends on the degree to which foreclosure would be profitable. The vertically integrated firm will take into
account how its supplies of inputs to competitors downstream will affect not only the profits of its upstream division, but also of its
downstream division. Essentially, the merged entity faces a trade-off between the profit lost in the upstream market due to a reduction
of input sales to (actual or potential) rivals and the profit gain, in the short or longer term, from expanding sales downstream or, as the
case may be, being able to raise prices to consumers. The trade-off is likely to depend on the level of profits the merged entity obtains
upstream and downstream. Other things constant, the lower the margins upstream, the lower the loss from restricting input sales.
Similarly, the higher the downstream margins, the higher the profit gain from increasing market share downstream at the expense of
foreclosed rivals 7
Such measure is not necessary [or even possible] in some kinds of unilateral conducts that
involves just ONE relevant market, such as some “predatory price” conducts; some sham
litigations; or some raising rival costs conducts. Therefore, there are different preconditions
regarding market power, depending on the analyzed conduct.
There are, also, certain models used when both buyers and sellers have market power, see,
for example, (HENDRICKS & MCAFEE, 2010).8 Hence, depending on the conduct or the
market, lerner index can mean different things (in terms of preconditions to perform a conduct).
3. Concepts
Given that, in the same relevant market “P1, with a certain amount of market share, can
exclude P2, but not P3”, one could question if this apparent contradiction could be avoided, if
relevant market were narrowed, defining relevant market as a sub segment (niche) that P1 and P2
belongs to, but P3 has no place on it.
Of course, market definition plays an important role in structural analysis, and there may
be some definitions that could capture better the competitive dynamics of a given market than
other. However, product heterogeneity is a reality in several markets, and there are many ways to
aggregate or to segregate markets. Also, it is difficult to consider every single possible set of
combinations of how market can be segregated (or aggregated) since the very beginning of
antitrust analysis.
Moreover, even if all the information is present (and all possible scenarios are
considered), there could be technological problems in market definitions. Let´s say that the
Antitrust Authority is using Hypothetical Monopolist test, that consider the exercise to see if
hypothetical Small but Significant and Non-transitory Increase in Price (SSNIP) is possible in
order to define a relevant market. If that is the case, maybe the specification of a default SSNIP
of 5% to niche where P1-P2 are present is not possible [but a very close SSNIP of 4,5% is]. In this
specific case, the standard procedure would be to enlarge the borders of relevant market, even
knowing that, by so doing, it is possible to include in the relevant market, several other
heterogeneous products with low relationship with P1 and P2.
Therefore, the problem may not always be the relevant market definition itself, and,
indeed, in some circumstances, P1 will be able to exclude P2, but not P3, even with the “correct”
market interpretation, according to predefined concepts. This example is very illustrative of how
(i) market power is a relative concept when applied in anticompetitive conduct analysis; and (ii)
SSNIP is a continuous concept behind the discrete definition of relevant market.
Another question that deserves attention is the following: regarding buyer power, could
there be a hypothetical monopsonist test? In other words, to measure “dominance” in buyer power
side, would it be correct to use the threshold of Small but Significant and Non-transitory Decrease
of Price (SSNDP) instead of SSNIP? If that is correct, would the test follow the same methodology
(looking only market power horizontally) or would it be necessary to measure impacts of a
hypothetical monopsonist in upstream, middlestream and downstream markets? Certainly, there
could be several possibilities to answer to these specific questions, and not right or wrong answers.
7
Official
Journal
of
the
European
Union,
18.10.2008,
C
265,
according
lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:C:2008:265:0006:0025:en:PDF, verified in 10th, March, 2015.
8 See http://vita.mcafee.cc/Bin/Vertical/mhi.html
to
http://eur-
Moreover, there are many other continuous concepts that could influence what relevant
market is. For example, even narrowing the analysis to seller power, there are several hermeneutic
options to choose to define relevant market, ranging from: qualitative aspects of the product 9;
verification of the level of prices10, level of diversion ratios, level of price-cost margins, level of
the available capacity; analysis of the movement of some variables, specially prices11 [simple
correlation, cointegration12, impulse response analysis, variance decomposition of vector errorcorrection model, granger causality test13 and several other exercises] (FORNI, 2004) (WERDEN
& FROEB, 1993); the use of critical loss analysis (HARRIS & SIMONS, Focusing Market
Definition: How Much Substitution Is Enough? , 1989) (SCHEFFMAN & SIMONS, 2003) 14; the
use of critical loss with aggregated diversion ratio (KATZ & SHAPIRO, 2003); generalized
critical loss analysis (COATE & WILLIAMS, 2005); and several other methodologies. Also there
are methods tailored to geographic specification such as Elzinga-Hogarty test (ELZINGA &
HOGARTY, 1973) (ELZINGA & HOGARTY, 1978)15, gravity model test (ANDRADE, et al.,
2010); just to name some of possible exercises.
It is important to mention that critical loss can vary between profit maximizing model or
breakeven model. It could be the case that using one of both models could mean the difference
between the approval of a merger or its blocking; the acquittal of an enterprise or its conviction,
especially if structural analysis is used.
Some authors, also, suggest that it is possible to use a Full Equilibrium Relevant Market
(FERM) (IVALDI & LÖRINCZ, 2005) simulating merger among all players in a specific market
to see if the price will raise x%, in an equilibrium situation and comparing such results with other
theoretical modalities of market definition. These authors (IVALDI & LÖRINCZ, 2005) claimed
that results observed in their experiment suggested that FERM (equilibrium model) differ from
critical loss analysis (out of equilibrium model, according to such article). Other authors pointed
to the possibility of use of other forms of equilibrium, such as Marshallian Profit-Maximizer test
(MPM) and Price-Leader, Profit-Maximizer test (PLPM) (FIUZA, 2008). Also, (BUCCIROSSI,
2000) made some models taking into account several Nash Equilibrium and game theory concepts
to define what relevant market is.
So, hermeneutic options (HE) is a set of several methodologies that may lead to same or
different outcomes:
HE = {HE1, HE2, HE3, …, HEN}
Equation 1 – Hermeneutic options
As mentioned before, depending on the preference of the interpreter, it is possible to
choose one or more hermeneutic options. However, if, in abstract terms, such choice (or
preference) may lead to different market definitions, in concrete cases, it is expected that, using a
paracomplete logic, for example, the trivalent logic of Jan Lukasiewicz (FRONHÖFER, 2011)
that relevant market could be defined as follows:
∀ HE _interpreter-j → RM ={ T ,  ,
9
} = {well defined, not well defined, neutral}
See Brown Shoe Co. v. United States, 370 U.S. 294
See European Case IV/M.582 – Orkla/Volvo
11
See Nestlé/Perrier
12
See Merger 08012.001885/2007-11 judged by CADE.
13
See Merger 04-07/75-18 judged by Turkish Authority
14
See U.S. cases FTC v. Swedish Match e United States v. Sungard Data Systems, Inc
15
See Merger involving Barloworld Coatings and Midas Paints judged by South African Antitrust
Authority
10
Equation 2 – Paracomplete trivalent logics applied to Market Definition for jth interpreter
The neutrality ( ) occurs because certain hermeneutic options are not applicable to some
situations. For example, when there are stationary prices, a cointegration test is not applicable.
So, this HECointegration, when applied to a case, within such context, will not be able to generate
neither negative, nor positive results in terms of market definition. The same thing can occur if
the Authority wants to use some parametric test without right specification (for example, it may
not be possible to find, all the times, a normal distribution of error terms).
Also, interpreter-1 (private agent); interpreter-2 (Antitrust Authority); interpreter-3
(judge of the case), and so on, may not reach a consensual decision of who had picked the right
hermeneutic choice.
So, it is important to stress that, in a conceptual level, relevant market (and
consequentially substantial market power defined structurally on the basis of market share) could
represent different valid concepts depending on the theoretical choice of the interpreter, that could
perfectly lead to different decisions of what represents dominance (due to divergent ideal nature
of these different abstract concepts).
4. Information
To complement this complex inference, in order to define market power or relevant
market properly, the interpreter should be able to have access to relevant amount of information
(∑ 𝐼𝑛𝑓) with good quality (Q). Each of these variables, when normalized, could range from 0 to
1. Zero would mean that the Antitrust Authority has no data at all (∑ 𝐼𝑛𝑓 = 0) or that the
information of the market is so confusing that it is not even possible to understand what market it
is (Q=0). The opposite situation is that when the Antitrust Authority has perfect information of
everything (then ∑ 𝐼𝑛𝑓=Q=1). In real terms, given asymmetric information, it is expected that in
merger and in conduct cases, both numbers should lie in the middle of such extremes.
Some countries, with electronic discovery or with the possibility of making dawn raids in
unilateral conduct cases, may have more access to information that helps to define more
accurately the competitive pressures of the market, in dominance cases, tending to be closer to 1
if compared to other countries that do not have such instruments. Thus, investigatory powers of
the Authority may interfere in this regard, allowing discovery of several evidences. And, on the
level of international cooperation, the Authorities (i) have incentives to cooperate, trying to
decrease its own asymmetric information, about theoretical questions, models, softwares and facts
of the market (and specific conducts) and of what other Authorities are doing to identify problems
and how to remedy them as well as (ii) have incentives not to cooperate, trying to minimize the
costs of cooperation itself [regarding different timetables, legal difficulties and interests to share
evidences, costs of bilaterally trying to communicate what are the main concerns of a specific
conduct and so on].
5. Robustness, convergence and burden of proof
As important as having information [obtained nationally or internationally] is the
processing of such information. In this regard, the strategy of identification (SI) of models or tests
could be used in order to comprehend the competitive pressures of the market in concrete cases.
For example, if an Antitrust Authority is trying to understand what amount of diversion ratio
between two enterprises is, it may use the following equation:
𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑜𝑛_𝑅𝑎𝑡𝑖𝑜 =
𝜀𝑖𝑗
𝜀𝑖𝑖
ii = Own demand price elasticity (which refers to several confidence/credible intervals, when estimated robustly)
ij = Cross demand price elasticity (which refers to several confidence/credible intervals, when estimated robustly)
Equation 3 – Diversion Ratio
Since own and cross price elasticities are intervals, diversion ratio, itself, is in interval of
intervals (although it is treated in some papers as just a point). In this sense, the reason for this
phenomenon may be the use of a robust exercise. If the Antitrust Authority hypothetically has
several algorithms to try, it is necessary to make a robustness check:
“A now common exercise in empirical studies is a "robustness check," where the researcher examines how certain "core" regression
coefficient estimates behave when the regression specification is modified in some way, typically by adding or removing regressors.
Leamer (1983) initially advocated investigations of this sort, arguing that "fragility" of regression coefficient estimates is indicative of
specification error, and that sensitivity analyses (i.e., robustness checks) should be routinely conducted to help diagnose misspecification.
Such exercises are now so popular that standard econometric software has modules designed to perform robustness checks automatically;
for example, one can use the STATA commands rcheck or checkrob. A finding that the coefficients don´t change much is taken to be
evidence that these coefficients are "robust." If the signs and magnitudes of the estimated regression coefficients are also plausible, this
is commonly taken as evidence that the estimated regression coefficients can be reliably interpreted as the true causal effects of the
associated regressors, with all that this may imply for policy analysis and economic insight”. (WHITE & LU, 2014)
However, it is not easy to know what “core variables” are, in a specific equation.
Therefore, it is important to know if by adding or removing regressors, the exercise is still valid,
and if problems regarding normal errors, endogeneity, among several others are controlled
(WHITE & LU, 2014). Hence, maybe only some of the results that are tested can present good
specification.
Even if only good models (means) are selected (with good specification), it is possible
that Antitrust Authorities select an interval of possible diversion ratios (means).
Figure 2 – Hypothetical example of possible diversion ratios derived from a robustness test
*For the purpose of the hypothetical exercise, the only valid results are the yellow ones, that specification test (such
as Sargan or Reset test) can reject at level of 10%.
However, in this example, diversion ratios (means) lie between 0,65 to 0,75. What if the
difference between 0,65 to 0,75 were a crucial difference to know if there is an anticompetitive
problem or not? It is not easy to reply to this question [and to choose among valid possible results].
Another problem is that all these valid model results (yellow points), actually are not
points, but lines or intervals (that were not plotted in graphic). Therefore, the interval of diversion
ratios (hypothetical yellow lines) seems to be underestimated, since the analysis focused only on
means but not on the entire interval. The maximum and minimum possible values were not
presented in this example.
If standard deviations were plotted (if yellow lines were plotted), maybe, it would be
possible to know if model results are converging to a point.
In terms of distributions, the convergence exists, if, as the number of observations (n)
tends to infinity, standard deviation () tends to 0 and the mean of random variables X of a given
sample (𝑋̅) tends to the population mean (𝜇𝑋 ). In mathematical notation, it is possible to write:
lim = 𝑋̅
𝑛→∞
Equation 4 – Convergence in a given distribution
Depending on consistency, unbiasedness and minimum variance of estimators  the
convergence process can be more or less efficient. Random vectors could also have the ability to
converge to a specific number, and maybe this concept could be applied, with some restraint, to
help how to deal with robustness tests [in other words, how to deal with valid but possible
contradictory answers of different algorithms or different model specifications, that are expressed
in terms of intervals]. Here the word “maybe” is underlined, because: (i) sometimes it is not
possible to replicate infinitely econometric model results in order to see if standard deviations
tends to 0, depending on the specification16; and (ii) there may be good algorithms and bad ones,
and if there are a lot of bad specifications and a few good ones, the convergence of econometric
model results may be misleading.
Nevertheless, before presenting how different models can converge (or not), it is
important to know why models can present different perspectives of reality.
Econometrics is used to understand scientific issues, and it unifies Statistical knowledge,
Economic Theory and Mathematics in an applied manner (GREENE, 2003, p. 1). However, these
three areas, also, comprehend a variety of schools and approaches, that may lead to contradictory
model responses (and plausibility views of reality) inherently attached to the use of different
hermeneutic choices (HE).
Regarding Economic Theory (in realm of demand estimation), the interpreter should
decide if the product is a Giffen good or a normal good; if product is homogeneous or
heterogeneous, since the very beginning of the analysis (and sometimes before any empirical test
is implemented). If the market is homogeneous than the Antitrust Authority is supposed to (at
least) consider the use of Cournot´s Model, otherwise if the market presents some heterogeneity,
then the Authority is supposed to consider Bertrand´s Model, or some other model (for example,
Stackelberg´s Model).
However, there are grades of homogeneity or heterogeneity, and some products may have
some sort of differentiation, even being labeled as similar ones. Oil (depending on the viscosity),
Iron ore (depending on the purity) 17 could be considered heterogeneous (even though for someone
that is not an expert in these markets may have different perception about the heterogeneity of
such products). Trademarks for cattle meat producer, in Brazil, in past, did not play an important
role for differentiation. Nowadays, it is possible to see a huge amount of publicity to differentiate
these products.18 Therefore, it is even harder to see, in a dynamic perspective, what is
16
There are standards deviations of a set of means of given parameter in different models, but models,
themselves, have their own standard deviations. It is interesting to know if both measures tend to zero.
17
CADE´s Merger analysis - 08012.002838/2001-08 (Companhia Vale do Rio Doce e Ferteco Mineração S.A.)
CADE´s Merger analysis - 08700.010688/2013-83 (Rodopa Indústria e Comércio de Alimentos Ltda. JBS S.A.;
Forte Empreendimentos e Participações Ltda.)
18
heterogeneous, what is becoming heterogeneous and what is not heterogeneous. Therefore, it may
be difficult to understand – at the very beginning of relevant market definition or market power
definition – what would be the best theoretical approach on this matter.
Independently of this debate, there are a lot of ways of how to implement a demand
model, for example: AIDS (Almost Ideal Demand System) (DEATON & MUELLBAUER, 1980),
o LAIDS (Linear AIDS) (FUJII & MARK, 1985)], QUAIDS (Quadratic Almost Ideal Demand
System) (BANKS, BLUNDELL, & LEWBEL, 1997), RAIDS (Rationed Almost Ideal Demand
System) (DEATON & MUELLBAUER, 1981), among others methodologies of these kind, as
well as Logit; Nested Logit demand; Mixed Logit demand; and assess Willingness To Pay of
consumers 1 (CAPPS, DRANOVE, & SATTERHWAITE, 2003); or it is possible to use some
calibration, as PCAIDS (Proportionally Calibrated Almost Ideal Demand ) (EPSTEIN &
RUBINFELD, 2003)]; ALM (Antirust Logit Model) (WERDEN & FROEB, The Effects of
Mergers in Differentiated products Industries: Logit Demand and Structural Merger Policy, 1994)
(WERDEN, FROEB, & TARDIFF, 1996); AMLM (Antitrust Mixed Logit Model) (DeSOUZA,
2009)].
Depending on the model used to estimate demand, it is possible to have different diversion
ratios or different own price elasticities, impacting in the decision of the Antitrust Authority.
There are other decisions that could affect the outcome of the analysis, for example: What
is the proper index to deflate prices? Is it correct to put some kind of filter that would smooth time
series that is being analyzed by the Authority? What are the rightful variables to analyze in the
Econometric model?
Regarding models, how should the Authority weight the result of different algorithms?
Some other decisions also are extremely important that could influence the outcome of the desired
tests, such as to use linear, isoelastic or some other form of dependent or independent variables;
how to solve problems such as the absence of normality of errors in parametric regressions? How
to solve heteroscedasticity? What would constitute good instruments for a concrete case?
Whenever two stage least square is used, what is the rightful approach regarding Sargan test, and
how strong should be the instruments – using Stock-Yogo methodology or some other? How
plausibility of results should be interpreted in first and second stages of the regressions? How to
measure if demand is kinked or not?
Sometimes, it is possible to have disagreement about these (or other) questions. Also,
there is the inferencist and Bayesian divide: Indeed, part of the academy is in favor of using
baeysian analysis (STRNAD, 2007) (DOWNEY, 2012), and even bayesian priors (POSNER R. ,
2008, p. 67) while other authors are more skeptical of using such methodologies without some
restraint (SPANOS, 2007) (MAYO, 1996) (GELMAN, 2008) (NIDA-RÜMELIN, 2008).
Regarding bayesianism, there are several strategies to deal with autocorrelation and to achieve a
convergent model.
In addition, there are some other problems, linked to the concept of some variables, such
as: critical loss or lerner index should be estimated using average variable costs or it should be
necessary to calculate marginal costs? According to Barry Harris:
“The Critical Loss is equal to Y ÷ (Y + CM) x (100%) where Y is equal to the Merger Guidelines' hypothesized price increase and CM
is equal to the contribution margin of the producers in the group. The contribution margin is defined as the difference between the
original price and average variable cost stated as a percentage of the original price. Variable cost is a proxy for the actual costs saved
because of the reduction in sales. The variable cost element should be consistent with the level of lost sales and the associated time
period.” (HARRIS B. C., 2015)
Certainly average variable costs (AVC) is a more direct and concrete measure. However,
if marginal cost (MC) ought to be empirically estimated, then, the complexities of the exercise
increase (and AVC model may lead to a different critical value when compared to MC model).
The concept of MC is currently being used in some other aspects, as measurement of GUPP
(Gross Upward Pricing Pressure) and UPP (Upward Pricing Pressure), and in several other
simulations.
Therefore, conflict of results is something expected, whenever someone is dealing with
so many different concepts or different possible paths to implement a model.
So, econometric exercises that an Antitrust Authority is supposed to present/deliver to
society in order to estimate some aspects of the competitive environment is very far from being a
dichotomic, clear and obvious task. This endeavor could encompass a series of strategies of
identification (SI = {SI1, SI2, SI3….SIn}).
Considering all of this, the question that appears worth to be made is not what are the
borders of a specific market (or even what is the market power regardless of market share), but:
(i)
(ii)
(iii)
(iv)
(v)
what are the definitions supporting concepts (HE) used in a concrete model?
how these definitions were measured (SI)?
what are the maximum and minimum values of each estimated value (given a
specific interval)19?
what are the critical values (or critical intervals with maximum and minimum
critical values)?
how different models [tailored for testing robustness] allow all possible estimated
intervals and critical intervals to interact among themselves?
Sometimes, outcomes of econometric measured intervals can converge to a number or a
direction. In this sense, it could be discovered a pattern among measured intervals and critical
intervals.
Then, instead of having a yes or no answer, true or false, or in terms of relevant market
the measure of in and out of a specific territory, maybe it would be more accurate to define a
probability of what relevant market is; and the probability of a given enterprise, in a given
circumstance, exercise market powers towards some specific player. Such exercise is not trivial.
“In Votorantim/Fischer/JV the notifying parties estimated own and cross price elasticities for orange juice and other dirinks using
monthly retail data for France, Germany, the Netherlands and the UK. The Commission criticized the parties´ analysis on a number
of technical grounds, noting in particular the small sample size and paucity control variables, and noted that the results were subject
to a wide margin of error (that is, the reported coefficients exhibited wide confidence intervals).” (GORE, LEWIS, LOFARO, &
DETHMERES, 2013)
For example, in a recent case (Braskem tried to acquire Solvay – Merger Case
08700.000436/2014-27), CADE estimated critical elasticity value as being - 2,57.
Claimants presented to CADE 48 models (with different specifications) in order to
present a robust estimate of elasticity level. The Economic Department of CADE estimated
several other models, but for illustrative purpose, it is interesting to focus just on models present
by claimants of this merger.
Some of the 48 models did not discarded the hypothesis that elasticity would be zero (with
95% of confidence interval above zero) [stressing that CADE disagreed that null hypothesis were
in favor of the claimants]. Some of the models that had this feature were considered not
informative of competitive pressures of the market. In this case, the parties argued that if their
19
Diversion ratio treated sometimes as a line or a point, indeed, involves an interval of cross elasticities divided by an
interval or own price demand elasticities
own model is not informative it means that they do not have market power. This is an incorrect
argument, in the sense that in the majority of markets it is not plausible that there is no elasticity
whatsoever, and that enterprises could raise their price in an infinite manner and that consumers
will still be willing to buy such products no matter what their own budget constraints are. If the
parties cannot discard zero elasticity, it is very likely that their model is not very precise.
Claimants even argued that in some specifications the confidence interval of demand elasticity
varied from + 50 to - 60. That is almost the same thing to say that temperature is between +1.000oC
or -1.000 oC.
Indeed, claimants may have interest to construe bad and inconclusive models, and to use
this inconclusiveness to argue that they do not have market power.
However, on the specific case, as it is possible to see on the bellow figures presented,
several other valid models presented own-elasticity of demand, of the entire market, with values
statistically negative (within 95% of confidence interval different from zero).
Figure 3 – All models presented by Claimants in Merger Case 08700.000436/2014-27
*Green points represent estimated means and white lines represent confidence intervals
**Elasticities are not presented in absolute values. These are real measured values.
Figure 4 – 35 models presented by Claimants in Merger Case 08700.000436/2014-27 (ordered by size of standard deviation)
*Green points represent estimated means and white lines represent confidence intervals
**Elasticities are not presented in absolute values. These are real measured values.
Comparing different valid results, it might be interesting to know if models converge to
a specific number. Sometimes it is difficult to increase model specifications towards infinite to
know the direction of convergence. So, one possible way to know if models are “converging”, is
to give more weight to models that have more precision (low standard deviation [or simply
“Std”]). For the purpose of this paper, “Std” were weighted observing this formula:
𝑊𝑒𝑖𝑔ℎ𝑡 = 𝑆𝑡𝑑−2
Equation 5 – Suggested weight
Therefore, models with low standard deviations will have higher weights.
Based on this concept, the following graphic was plotted, observing:



on horizontal axis - the inferior/lower limit of the interval of confidence of each
estimated elasticity
on vertical axis – the mean value of the estimated elasticities
plotted circles using the weight as the radius of such circles
It is possible to see that (using weights) the biggest area (convergence) is concentrated
where relevant market is well defined.
Figure 5 – Convergence of results – Claimant´s model - Merger 08700.000436/2014-27 (ordered by size of standard deviation)
**Elasticities are not presented in absolute values. These are real measured values.
There are other ways to represent graphically the convergence. It would be possible, for
example, to plot inferior limit, upper limit and elasticity mean of each test, in 3 dimensions.
Figure 6 – Convergence of results – Claimant´s model - Merger 08700.000436/2014-27 (ordered by size of standard deviation)
**Elasticities are not presented in absolute values. These are real measured values.
The problem of such representations is that when 𝑆𝑡𝑑−2 is used to represent radius of
circles or spheres, they enlarge the distributions that by concept are concentrated (with lower
standard deviation and maybe more efficient results). A possible solution to this problem would
be to represent graphically different distributions, in three dimensions, using cubes or in a smooth
manner, observing the correct standard deviation, but giving more observations (weights) to
accurate models. In other words, models will have (𝑆𝑡𝑑−2 ) observations (N).
Figure 7- Convergence of results – Claimant´s model - Merger 08700.000436/2014-27 (ordered by size of standard deviation)
**Elasticities are not presented in absolute values. These are real measured values.
Using such procedure, some minor observations had trespassed the limit of critical value.
On the other hand, the majority of observations (and even entire distributions) stayed inside the
area where market was considered to be well defined. In addition, it is very likely that the
observations that trespassed the boundaries of critical values may not be statistically relevant
(taking into account the overall distributions).
It is important to mention that the Economic Department of CADE observed some
deficiencies in models that were presented by parties and made its own exercises, that helped
CADE to decide this case in a more consistent and robust approach.
This graphics help the interpreter to understand that (even if there were only one
distribution and only one model to analyze) diversion ratios, own elasticities, cross elasticities,
marginal costs and several other concepts are not discrete concepts. Indeed, such definitions
belong to a very rich spectrum of possibilities and probabilities.
Another aspect to highlight is that generally statistical and econometrical tests compare
measured values with critical ones. Measured values, as explained before, are not points but
distributions. The same thing can be said about some important critical values. Indeed, critical
values themselves could be considered a wide range of possibilities.
For example, the price-cost margin (used in Lerner Index) is generally referred as a concept
that helps to understand what market power is. Moreover, the idea of Lerner Index could be used
in critical loss test and in several other parts of Antitrust Policy. Such concept could be referred
as:
𝐿𝑒𝑟𝑛𝑒𝑟 𝐼𝑛𝑑𝑒𝑥 𝒎𝒆𝒂𝒏 =
Equation 6 – Lerner Index
𝑃𝑟𝑖𝑐𝑒 − 𝑀𝑎𝑟𝑔𝑖𝑛𝑎𝑙_𝐶𝑜𝑠𝑡𝑠
𝑃𝑟𝑖𝑐𝑒
However, although it appears to be a point, indeed, Lerner represents a distribution, taking
in consideration the following arguments:

Regarding prices, this concept may vary a lot, depending on the client, on the features
of the products, on the period of time, or on some other aspects. Therefore, to simplify
reality, Lerner Index contemplates a mean of all these values. However, how precise
is Lerner Index? If prices have a huge standard deviation, maybe, it will not be a good
representation of what is important to set prices and to receive profits in a given
market, affecting concepts as critical loss.

About marginal costs, this value could be econometrically estimated. Therefore,
quantification of marginal cost is dependent of a series of other decisions that the
interpreter does and about the quantity and quality of information that interpreter has
regarding production costs. The problem is that (i) there may be some enterprises,
which do not even know how to measure, correctly, their own costs, how to keep
record of this information, or how to maximize its own profits (ii) the Antitrust
Authority may not understand correctly the structure of costs of a given sector; (iii)
and there could be different products and different production plants with different
marginal costs “inside” a given market definition that is being tested.

-
In addition, marginal costs could be econometrically estimated. So even with a
single estimation (and with full knowledge of the market), it is supposed that such
value lies within a confidence interval [inferencist approach] or a credible interval
[Bayesian approach]. And if the interpreter looks for a robust estimation of
marginal costs, then, the problem of convergence of model results may apply to
this concept.
-
Whenever perfect competition is mentioned in Economic textbooks (and the use
of lerner index), it is stressed that interpreters should not pay so much attention to
“accounting costs”, but to “economic costs” (that also encompass opportunity
costs). On the other hand, opportunity cost is a relative concept. In real world,
there are almost infinite possible alternatives to invest efforts to produce
something. The interpreter should limit its analysis if the intent is to draw a valid
inference of market power.
Since there is a distribution of values of prices and a distribution of several valid
probable marginal costs [inside confidence or credible intervals taking into account
specific opportunity costs], it seems plausible to refer to these concepts not as points
or lines, but a range of numbers.
𝐿𝑒𝑟𝑛𝑒𝑟 𝐼𝑛𝑑𝑒𝑥 𝑰𝒏𝒕𝒆𝒓𝒗𝒂𝒍 =
𝑃𝑟𝑖𝑐𝑒 ∗ −𝑀𝑖𝑛(𝑀𝑎𝑟𝑔𝑖𝑛𝑎𝑙_𝐶𝑜𝑠𝑡𝑠) 𝑃𝑟𝑖𝑐𝑒 ∗ −𝑀𝑎𝑥(𝑀𝑎𝑟𝑔𝑖𝑛𝑎𝑙_𝐶𝑜𝑠𝑡𝑠)
….
𝑃𝑟𝑖𝑐𝑒 ∗
𝑃𝑟𝑖𝑐𝑒 ∗
Equation 7 – Lerner interval changing marginal costs
* Since there are many combinations, price can mean average price, minimum price and maximum price, or any
other number between such values
If marginal costs and prices are used to settle what critical values are [critical elasticities or
GUPP for example], then, critical values themselves – also – cannot be considered just as points
or lines, but should be represented by some other kind of geometric form. In this situation, the
observed interval become as important as the estimated mean point.
If that is correct, then, classical statistical tests could also be difficult to implement without
some kind of adjustment. If classical statistical tests are used, then, there will be just two options:
-
First (Null) hypothesis (𝐻𝑂 ) : that lies on one side of critical value
Second (Alternative) hypothesis (𝐻1 ) : that lies on the other side of critical value
If critical value is not a point or a line, then, a third hypothesis may be needed to complete
(fairly) the exercise (since there will be some values between critical value distribution).
-
First (Alternative) hypothesis (𝐻1 ) : that lies on one side of critical interval
Second (Alternative) hypothesis (𝐻2 ) : that lies on the other side of critical interval
Third (Null) hypothesis (𝐻0 ) : that lies inside critical interval
That is similar to what happened when classical logic is compared to a three or multiple
valued logic. Classical logic is not tailored to deal with this problem [2 or more dimensions of
critical values], in the sense that, according to Classical thought, the reality is either A (Null
hypothesis) or not A (Alternative hypothesis).
Of course, Antitrust Authorities, in real world, should decide [at least in the majority of the
cases] in a dualistic framework: which means to define if someone is either guilty or not of some
conduct; if a merger should be blocked or not; and so on. However, to deny the existence of this
problem that underlies the critical value continuous concept may obscure the debate about the
alternative approaches on how to deal with these issues [especially if critical value is estimated
with great standard deviation using a non-efficient estimator].
Regardless of this discussion (and assuming a two valued logic), another interesting debate
is how burden of proof should be distributed? More specifically, who has the right to have null
hypothesis? Whoever possess the null hypothesis is in a good and comfortable position, in the
sense that the burden of proof lies on the other side (on the alternative hypothesis).
So, whenever the discussion of null hypothesis is raised, the question that is underneath the
surface is: who is responsible to prove dominance or any other thing? And in what extent?
Indeed, whenever the distribution of the measured value crosses the “line” of the critical
value, assuming it is a line, it is important to define, among other thing:



Who has null hypothesis?
What is the amount of statistical significance that Authority should demand in order
to minimize type I error (accepting a false hypothesis)? 90%, 95%, 99%, or other
value?
What is the amount of power (severity test) that Antitrust Authority should adopt in
order to minimize type II error (rejecting a true hypothesis)? 90%, 95%, 99%, or other
value?
Obvious decision - relevant market
Clear “victory” of one hypothesis
Not so obvious decision - relevant market
Possible tie: Victory should be arbitrated
Figure 8 – Obvious and not so obvious decisions about critical values
**Elasticities are not presented in absolute values
Whenever the whole distribution of the measured value lies in one side of the critical value,
then, it is possible to observe a “clear victory” of one hypothesis over the other one.
When the Antitrust Authority is faced with a “not so obvious” decision (in other words,
when the distribution of measured value crosses the critical value), it means that there is room for
doubts of what borders of the market are. If critical value is a line, it is possible (at least) to have
(i) a precautionary or (ii) a self-restraint approach (in other words, if there is doubt what side is
correct, it is possible to take another binary decision of who is right = Doubt={0,1}).
Precautionary approach prescribes that the null hypothesis is in favor of narrow definition
of the market. In other words, if there is doubt, then, the Antitrust Authority should prefer narrow
definitions. The self-restraint approach, on the opposite, prescribes that null hypothesis should be
established in favor of the enlargement of the market. So, in doubt, Antitrust Authorities, by this
approach, should decide for broader markets.
𝐻𝑜 = |𝐸| ≤ |𝐸𝐶𝑅 |; 𝐻1 = |𝐸| > |𝐸𝐶𝑅 |
Equation 8 – Victory of (generally) the precautionary approach = narrow market has the null hypothesis
𝐻𝑜 = |𝐸| ≥ |𝐸𝐶𝑅 |; 𝐻1 = |𝐸| < |𝐸𝐶𝑅 |
Equation 9 – Victory of (generally) the self-restraint approach = narrow market has the alternative. hypothesis
∗ |𝐸| = 𝑎𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑒𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦
∗∗ |𝐸𝐶𝑅 | = 𝑎𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑒𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦
Depending on how such approaches are preferred [about burden of proof], two different
persons could reach different, and contradictory solutions, looking to the same data, same results,
and same valid scientific models. Whenever a quantitative tie is established between two
competing hypothesis, the faith of relevant market definition would ultimately depend on the
subjective preference of the interpreter. That is a dilemma that the antitrust analyst has with just
one tie. In merger cases, CADE prefers the precautionary approach (see merger
08700.000436/2014-27).
Nevertheless, what if “clear victory of narrow market definition” occurs in some models
together with “arbitrated victory” (ties) in some other models? And what if “clear victory of
narrow markets” occurs in some models simultaneously with “clear victory of broader market” in
some other models? How to deal with these problems, taking into account that critical values may
not be lines?
The concept of convergence of different model responses [and to know the direction of
such convergence] may, sometimes, help to understand what is the most probable (i) measured
value and (ii) critical value. However, sometimes, it is possible that convergence is not achieved
and, even in this situation, it is interesting to make this kind of questions, regarding statistical
burden of proof.
In the following exercise, 300 random distributions were inferred (and standard deviations
were normally distributed to avoid convergence). Let´s imagine that they are real measures of
elasticities of a case. Also, let´s say that confidence interval of 95% of marginal costs can predict
price-cost margin of 31,72% to 40,25%, which will mean a isoelastic breakeven critical elasticity
between 3 and 2.4.
Distribution of model results and critical values
1
2
3
mean
4
5
Critical values - 3 to 2.4
0
100
200
300
model
Elasticities in absolute numbers
**Elasticities are presented in absolute values
Figure 9 – Different distributions of elasticities
Results of exercises - several models and different critical values
350
300
Results
250
200
Should be expanded
150
Undefined
100
Well defined
50
0
3
2,9
2,8
2,7
2,6
2,5
2,4
Critical Elasticities
**Elasticities are presented in absolute values
Figure 10 – Results of the application of critical elasticity test with several models
Decisions \Crit.Values
Undefined (𝑯𝒐 )
Well defined (𝑯𝟏 )
Should be expanded (𝑯𝟐 )
3,0
288
6
6
2,9
285
3
12
2,8
281
2
17
2,7
274
2
24
2,6
267
1
32
2,5
253
0
47
2,4
237
0
63
**Elasticities are presented in absolute values
Table 2 - Results of the application of critical elasticity test with several models
* ∀ HE, SI → RM ={ T ,  ,
} = {well defined, not well defined, neutral/undefined}
𝐻𝑜 = |𝐸| = |𝐸𝐶𝑅 | = Undefined
𝐻1 = |𝐸| < |𝐸𝐶𝑅 | = Well defined
𝐻2 = |𝐸| > |𝐸𝐶𝑅 | = Should be expanded
Equation 10 – Different hypothesis in a trivalent model to define Relevant Market
As it is possible to verify, if critical value is 3, then, 6 models will declare clear victory for
well-defined market, 6 other models will declare clear victory for the need to expand the market,
and all the other results (undefined) will be inconclusive or depend on the burden of the proof
decision. It is a very unclear scenario.
On the other hand, if the interpreter looks not only to 3 (as line of critical value) but to the
aforementioned distribution of critical values that goes from 3 to 2,4, will see that there are still
many undefined situations, but will also see that clear victory of the hypothesis regarding the need
for expansion of the market is increasing. In this sense, this exercise opens a debate on how the
interpreter should consider this phenomenon: (i) undefined results ought to be considered? (ii) If
the answer is yes, model results could be weighted somehow or it would be better to use, for
pragmatic purpose, two-valued logic?
Again, depending on the hermeneutic interpretation, it is possible to answer these questions
in different manners. Some other authors tried to use three valued logic in statistical tests,
(HARRIS R. , 2009) (HARRIS R. , 1997) (KAISER, 1960), but in a quite different manner (not
taking into account, for example, (i) different distributions of model´s responses, nor (ii) the
context in which a clear binary outcome is expected, to define relevant markets, to condemn a
certain practice, among other things.
Another important aspect to mention is that sometimes (and only sometimes) narrow
definition of the market does not represent a precautionary attitude.
For example, let´s suppose that in a specific market (market M1) there are 3 players, the first one
(incumbent) with 90% of market share and the other two with 5% each. The hypothetical
monopolization of this market will lead to a 10% increase in price. So this could represent the
“smallest hypothetical market”.
However, let´s suppose also that there is another market (market M2) with the same features, having
one player (incumbent) also with 90% of market shares and the other two with 5% each.
Let´s suppose that market M2 and market M1 have products with a certain degree of substitutability.
If the incumbent of market M1 exclude incumbent of market M2 from the market (or have a merger
with it), would it be possible to argue that such exclusion or merger has no impact at all, because,
according to the abstract definition of “smallest hypothetical market” both players belong to distinct
markets?
In fact, to search for the “smallest possible market” in which market power can be exercised does
not exhaust the possibility of having another (larger) market definition, in which market power
could, still, be exercised (sometimes even in a greater extent).
In this sense, what is relevant market for one situation (monopolization of market M1) may not be
a correct tool for some other situation (monopolization of market M1 and M2 by the same
enterprise). This argument is a little bit different from the traditional cellophane fallacy that tries to
see if – in conduct cases – it makes or not sense to define relevant market borders, given that the
conduct itself could have altered the conditions and sometimes even the borders of the market. The
problem here is not that conditions changed through time, but how different scenarios (and different
kinds of conducts) for the same market demand different market definitions.
Together with the discussion of the burden of proof of the definition of relevant market
itself or market conditions, there may be other important discussions about burden of proof.
Generally, analysts just segregate, dualistically, between illicit per se conducts [in which
the existence of the conduct is enough to condemn an enterprise without complementary
economic analysis] or conducts that should be analyzed by rule of reason, that may demand some
economic analysis.
However, there are several other questions behind this debate, such as:
(A) what should be proved?
(i) The existence of a conduct C? Is it necessary to prove intent?
(ii) What conduct means by an economical and/or juridical perspective? For
example, in abstract terms, predatory price is proven by what measure costs
(avoidable costs, average variable costs, marginal costs, others)?
(iii) Possibility of harm is proven by: Market share? Lerner index?
Upstream/downstream profits? Contracts? Sham judicial claims?
Coercion? How much foreclosure is admitted? Others?
(iv) Harm itself: is it necessary to quantify harm? Deadweight loss or just
overprice? In long term or short term? How to deal with potential harm
[elimination of potential players, for example]?
(v) Justifications: Is there any efficiency associated with the conduct (statically
or dynamically) or other justification?
(vi) Harm outweigh justifications? How to deal with allocative inefficiencies,
distributive inefficiencies, X inefficiencies, and so on?
(B) What are the types of evidences E = {E1, E2, E3,…EN} needed? Testimonies, telephone
wires, indirect evidences are accepted? Who should produce documents and information about
contracts, price, quantities sold, costs and other market information? How should that be legally
discovered by the government?
(C) In each of these categories mentioned before, if there is doubt, who wins [Doubt={0,1}]
or, statistically speaking, who has the null hypothesis?
(D) What are the different levels of the statistical significance and what is an accepted level
of power regarding statistical tests that may influence results of themes mentioned in A?
There are a lot of possible combination of A, B, C and D that could not be simply reduced
in two options: per se illegal conducts or conducts subject to rule of reason. So Wi is a specific
subset of combination of A,B,C, D and other variables of this kind, while W is a set of all possible
subsets regarding how burden of proof should be established or distributed.
W = {W1, W2, W3, …, WN}
Equation 11 – Interpreter´s choice on how burden of proof should be established/distributed
Although there are many possible combinations of these complex questions, the doctrine
and jurisprudence appears to begin to recognize a limited number of intermediate modalities
between rule of reason and per se illicit conducts.
Indeed, in the United States, there are some other categorizations, such as “inherently
suspect conduct” (hereinafter just “ISC”) that appear to be an intermediate solution to burden of
proof (between strict rule of reason and per se prohibition). The problem is how to define such
concept.
According to Richard Liebeskind and Joseph R. Tiffany (LIEBESKIND & TIFFANY,
2009), once market power and ISC itself is proved, the burden of proving the reasonability “shifts
from plaintiff to defendant” (In re Nine West C 3937 Polygram Holdings; Leegin Creative
Leather Products, Inc. v.. PSKs, Inc., Bbs Kay's Kloset). Liebeskind and Tiffany argued that
resale price maintenance (RPM) of Nine West case was not condemned because “Nine West did
not appear to have market power” (LIEBESKIND & TIFFANY, 2009, p. 4). Oliver Geoffrey,
on the other hand, argued that in “Realcomp, the FTC challenged practices by a Michigan
multiple listing service that limited user access to the listings of discount real estate brokers.
Labeling the practices “inherently suspect,” the Commission’s decision stated that, in the
absence of a procompetitive justification, the practices could be condemned without the need of
showing market power or actual anticompetitive effects.” (OLIVER, 2010, p. 40)
It appears, that part of the doctrine thinks that, in ISC, plaintiff should prove market power
of defendant (and other part of the doctrine disagrees).
About this debate, Spencer Weber Waller summarized Justice Stevens view of intermediate
burden of proofs using this graph, that divide conducts by a “rule of reason continuum” (and not
dichotomous rule of reason):
Figure 11 – Rule of reason - Adaptation of a Graph presented in (WALLER, 2009)
According to Spencer Weber Waller, there would be a certain tying conducts or certain
boycotts, that (together with ISC) seems to be in the middle of a “rule of reason continuum”. In
this graphical scheme, on the other hand, there is no econometric or statistical debate about burden
of proof (what is the statistical tolerable amount of significance?).
At CADE, the Brazilian Administrative Proceeding No 08012.001271/2001-44, involving
SKF do Brasil Ltda., took this intermediate view that the burden of proof of the justification of
the practice (to determine whether there are efficiencies or if efficiencies outweigh competitive
harm) should be on respondent side in cases involving RPM.
About burden of proof, one thing is also important to stress is that it is not just a procedural
issue. Indeed, in the conceptual level, to define (i) who wins null hypothesis; (ii) level of
significance of tests (iii) and the strategy of identification of concepts or models, also helps to
define what relevant market is, what lerner index (or interval) is, and, consequentially, what
market power (interval) is, which are material matters.
After choosing a specific strategy of identification of reality may affect the understanding
of what can produce harm on economic environment (or not); and what is (or should be) “allowed”
or “not”, materially. Therefore, this assessment helps to prescribe or proscribe conducts, in
general terms, and not just a procedural matter. Thus, it is not just a debate, made on a single case
on who is telling the true version of facts, where “right” and “wrong” [in abstract terms] are well
established and non-disputed concepts.
6. Dynamics
Sometimes, Market Power (or substantial market power) is estimated to know if certain
anticompetitive practices are “possible”. In Criminal Law, there is the concept of “impossible
crime” [impossibility derived from the mean or the object]. An impossible mean to commit a
crime is to administer sugar thinking that it is poison, so it is not possible to punish this crime
(unless the victim is diabetic or suffer a similar disorder). An impossible object to commit a crime
is someone that shoots a dead body, thinking it is alive and wanting to murder the victim, without
knowing that victim was already dead.
The problem of these two examples is that causality between illicit action and illicit result
is well established, making impossible the commission of the illicit practice. However, when this
issue is transported to social arena and specifically to Antitrust Law in order to know if enterprises
can exclude other rivals from the market, or have interest to do so, things become a little bit
fuzzier.
In this respect, the former Commissioner of CADE, Cesar Mattos, formulated this
hypothetical example: if two taxi drivers from the nearest point to CADE agree not give discounts
to customers, in his opinion, it could hardly be considered an anticompetitive illicit given the total
lack of market power of these two little players to impose price coordination to all other taxi
drivers (Administrative Proceeding No 08012.004484 / 2005-51).
This example seems to be not a very suitable one, since (i) price coordination is a per se
antitrust offense and may raise some questions if market power, market share, harm, justifications,
and the proper balance between harm/efficiencies should be discussed or not in antitrust defenses
for cartels; and (ii) the recent controversy around Uber demonstrate that the barriers of entry in
taxi industry are not negligible.
On the other hand, even market structures that at first glimpse do not have the optimal
conditions to sustain an anticompetitive conduct and they may, in a dynamic process, be modified.
Reflexivity phenomenon described by Anthony Giddens (GIDDENS, 1989) and Pierre Bourdieu
(BOURDIEU, 1998) claims that social [and market] structures can mold the conduct of agents,
as well as agents can change structure itself (that is the concept of structured and structuring
structure). Both processes occur simultaneously. Autopoietic logic of Luhmann also admits
changes in social structures (LUHMANN, 1989), but for endogenous factors, while Giddens and
Bordieu understand that exogenous factor could play this role.
So, what in first moment could be seen as an “impossible conduct to perform” (in markets
with low barriers of entry) could – in second moment – become a “possible conduct”, especially
if agents of a specific market increase barriers of entry, for example. It does not seem “impossible”
that players with low market power [but with high persuasion] in a price fixing scheme obtain,
somehow, the approval of several other players of the market for a given conduct, tacitly or
expressly.
Dynamic considerations can involve many other factors: taxes can be raised, antidumping
penalties can be imposed, logistics can be compromised, clients can enter in loyalty rebates,
among other conducts. Time matters.
Antitrust decision can be made on the first moment (t1) approving a merger for example
based on the argument that barriers of entry are low [and market has an international scope], but
in a second moment (t2) parties can enter with an antidumping claim, obtaining antidumping
protection for example, and altering market conditions and, maybe, changing the grounds that
allowed the favorable antitrust decision taken in first moment (t1), about mergers or market power.
So strategic use of antidumping and antitrust could interfere in market power perception (and in
relevant market definition).
Therefore, the timing (and dynamic aspect) of the analysis is important:
T = {t1, t2, t3, … tn}
Equation 12 – Timing (and dynamic aspects) of the analysis
Another aspect that should be highlighted is that several economic models try to measure
market power looking to profits, Lerner Index, market shares, demand features, and other
variables related to financial incentives to form an anticompetitive restraint (if it is profitable or
not to make a unilateral or coordinated action).
On the other hand, violent resources may be used to enforce unilateral or coordinated
anticompetitive practices [irrespectively from a logical or a traditional commercial point of view].
Posner mentioned several price-fixing schemes that were imposed by violence (POSNER R. A.,
1995, p. 88), during 1925 to 1939.
On the Brazilian Administrative Proceeding No 08012.002959/1998-11, one citizen that
declined to enter a price-fixing scheme, denouncing the anticompetitive practice to CADE, had
strafed his house three times, with eighteen shots on the front wall of his house. In this context,
the incentives, possibilities and damages of a violent anticompetitive practice may not be reduced
to lerner index, or to a mathematical evaluation of what overpricing is [and the traditional
tradeoffs involving this conduct].
Indeed, the decision to participate or not in an anticompetitive conduct could mean to put
at risk its own life and life of family members, in some situations. Let´s remember the
Shakespearean dilemma between life (the appearance of continuous interval that we know) and
death (the end of everything familiar). If life is the most precious asset that one has, and that is
what is at stake when comes the decision to participate or not in a cartel; to be or not part in
exclusive deal; to accept or not a RPM, then the discussion of “market power” or dominance may
not be limited to lerner index.
Enterprises, which are violent, may establish coercion to downstream market, upstream
market, middlestream market, and its own market, based on factual considerations. Even a small
player with violent resources can increase price threating entrants and can force other players to
adhere to the conduct (regardless of their profit maximization and their will to participate).
MC = {MC1, MC2, MC3, … , MCn}
Equation 13 – Means of coercion
Therefore, not everything that appears, statically, to be impossible, is, indeed, impossible
in dynamic terms. Such dynamics are hardly captured by a simple [and quick] answer to the
question of how one should define market power.
7. Teleology
What is [are] the goal[s] or value[s] that should be protected by law [and by Antitrust Law]?
In order to answer that question, it is possible that theories (and scientists) raise political
flags saying the law serves to maximize, minimize, divide, multiply or other mathematical
weighting about a set of pre-determined factors, such as values, principles or some other aspect,
depending on the theory. Once defined the function of law, it can be determined if the Law fulfills
its role effectively or efficiently, depending, also, on quantitative considerations.
One possible approach is the interpreter's neutrality with respect to social wealth. Through
this approach, the interpreter would say that Law in general (and consequentially Antitrust Law)
should not have any allocative or distributive preference. On the other hand, Julian Lamont and
Christi Favor understand that the defense of legal and teleological neutrality on this regard is
misguided because:
“reveals a confusion about the nature of the choices always facing each society. To claim that we should not pursue any changes to our
economic structures in light of a distributive justice argument is, by its very nature, to take a stand on the distributive justice of (or, if one
prefers, the ‘morality’ of) the current distribution and structures in the society compared to any of the possible alternative distributions and
structures practically available.”. (FAVOR & LAMONT, 2013)
For example, it is possible that interpreter defend a specific allocative distribution of
welfare, such as:








Strict egalitarianism (ROEMER, 1982) (ROEMER, 1985) (COHEN, 1988);
Difference principle of Rawls (RAWLS, 1971)
Luck egalitarianism of Ronald Dworkin, (DWORKIN, 2000);
Utilitarianism (BENTHAM, 1781) ;
Theories of equality of some authors, such as (RILEY, 1989) (MILLER, 1976)
(SARDUSKI, 1985) (DICK, 1975) (MILNE, 1986).
Libertarianism (NOZICK, 1974)
The criterion of Kaldor-Hicks efficiency, advocated by Richard Posner, in the first
phase of his work (POSNER R. , 2003).
Or some other mixed theory (BRUERS, 2010)
What seems interesting to highlight is that (i) it is possible to apply some specific
teleological criteria as something that should redefine the concentration of social wealth or values
already accumulated and or (ii) it is possible that such teleological criteria are applied for analysis
of concentrations that are yet to happen. It is also possible to think that the law can be framed as
to encourage actions of individuals in order to observe these criteria.
Bringing this debate to Antitrust, the interpreter can choose to find if acts are abusive or
not, depending on certain standards.
Therefore, [and besides what is per se illegal], several times when rule of reason is
established, critical values (and theories of what would represent harm derived from market
power) can change depending on the predetermined teleological standard of analysis.
Standard
Welfare standard
Explanation
Depending on how surplus is allocated or distributed, the use [or the
obtaining process of] market power can be harmful (and unlawful) or not.
As it will be explained in next table, there are a lot of ways to determine
what is the rightful welfare standard.
Range of options
Consumer choice
According to Lande, "the role of antitrust can best be understood in terms of a
fundamental standard-the standard of consumer choice. The antitrust laws are intended to
ensure that the marketplace remains competitive so that worthwhile options are produced
and made available to consumers, and this range of options is not to be significantly
impaired or distorted by anticompetitive practices. The antitrust laws thus ensure that the
economy responds to the aggregate signals of consumer demand, rather than to
government directives or the preferences of individual businesses. An optimal level of
consumer choice, which has elsewhere been termed "consumer sovereignty” is the state of
affairs where the consumer has the power to define his or her own wants and the ability to
satisfy these wants at competitive prices. The concept of consumer choice even embodies
some implicit notions about the rights of the individual in the broader society; it is implicitly
part of the Western world's response to Marxism and the other totalitarianisms of the
Twentieth Century." (LANDE, 2001).
Not predetermined When Dworking questioned if wealth is a value to be protected by law ,
Multiple goal
no matter what are the other consequences, he defended that there are other
values that should be protected by law, rather than simple efficiencies
concerns (DWORKING, 1980). According to Schwartz “The difficult question
is not whether non-economic considerations are a proper, indeed conventional, component
of the antitrust calculus, but how to take them into account” (SCHWARTZ, 1979, p.
1080).
Process itself
Competitive process
As explained by Jacobson "competitive process standard” [is] “articulated by
Gregory Werden and others. Under this approach, practices and transactions that interfere
with competition as a process would be prohibited, focusing only on economic effect, but
without focusing on any particular welfare standard. Practices that do not impair the
competitive process would not be prohibited, even if there is some negative impact on
consumer surplus”. (JACOBSON, 2015, p. 3)
Table 3 – Some standards about Antitrust goals
Standard
Price
Consumer Surplus
Producer Surplus
Hillsdown
Total Welfare
Important value
Surplus
Main question
Will price increase?
Will CS decrease?
Will PS decrease?
Will efficiencies generated exceed reduction in CS?
Efficiencies generated are lower than DWL?
Result
If yes, there is abusive MP
If yes, there is abusive MP
If yes, there is abusive MP
If no, there is abusive MP
If yes, there is abusive MP
CS = Consumer Surplus
PS = Producer Surplus
MP = Market Power
DWL = Deadweight loss
**Also it is possible to have some intermediate forms of balancing efficiencies, CS, PS or other variables, attributing, for example,
specific weights for each observable variable (such as efficiencies, DWL, PS, CS or some other variable).
Table 4 – Ranking of some Welfare Standards
One way to illustrate how such standards can interfere in antitrust reasoning is to look how,
in Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co., Inc., 127 S. Ct. 1069 (2007), [or
just Weyerhaeuser case], Rosch and Werden disagreed. Thomas Rosch believes that Antitrust
Law should not protect sellers and buyers equally. If consumers are not harmed, there would be
no reason to support conviction of unilateral practices20. That is why Rosch did not agree with the
20
“ In my view the antitrust laws protect consumers – and by “consumers” I mean consumers who buy the output in the relevant market. Having
practiced antitrust law for more than forty years, I yield to no one in my belief in the value and benefits of the Sherman Act. But I don’t think the Act
is supposed to cure all societal ills by preventing allocative inefficiencies. (…) In Weyerhaeuser, Ross-Simmons – a saw mill in the Pacific Northwest
– claimed that Weyerhaeuser engaged in a variety of anticompetitive conduct in the late 1990s in an effort to monopolize the relevant lumber market.
outcome of decision of U.S. Supreme Court on this case. Following this understanding, the former
Commissioner of CADE, Luiz Carlos Delorme Prado, in the Preliminary Inquiry No
08012.010713/2004-96, stated that “it is not role of Antitrust Authority to arbitrate profit margins
in a productive chain, if there is no harm to consumers whatsoever”.
On the other side, Gregory Werden claims that the decision of the Supreme Court was
correct in Weyerhaeuser case, even if the predatory practice did not impact directly consumers,
in the sense that anticompetitive process should be the rightful goal to pursue (and not protecting
this goal could even mean indirect harmful impact to consumers that the other position purports
to defend) (WERDEN G. , 2007). 21
There may be other theories that would impute or exclude other goals to Antitrust Law.
Depending on theoretical subjective preferences, interpreter can be more akin to accept
regulation or not, or more akin to accept the likelihood of a vertical restraint. In this regard,
Michael H. Riordan summarized the problem of vertical integration (Chicago and post-Chicago
divide) with these words:
Vertical integration is an enduring topic for economics. The structure-conduct-performance perspective of the 1950s and 1960s viewed
vertical integration suspiciously, worrying about exclusionary practices that foreclose competitors and leverage monopoly from one
market to another. The Chicago School of the 1960s and 1970s rebutted these concerns by pointing out the weak microeconomic
foundations of leverage theory, and explaining why vertical integration increases economic efficiency. Transaction Cost Economics of the
1970s and 1980s staked a middle ground, identifying new efficiency rationales for vertical integration, while cautioning that firms with
market power may have strategic goals poorly aligned with consumer welfare (Williamson, 1975; 1985). Most recently, a new literature
on vertical foreclosure (a.k.a. Post-Chicago Economics) applied game-theoretic tools to develop new theories of strategic vertical
integration and identify circumstances in which vertical integration alters industry conduct to the detriment of competitors and
consumers. The rich intellectual history of industrial organization economics thus reveals assorted approaches to the topic. Vertical
integration raises contentious issues for antitrust policy and industry regulation. Antitrust policy in the United States recognizes that a
vertical merger can create incentives for anticompetitive foreclosure or facilitate collusion, while remaining mindful that vertical
integration can achieve efficiencies (ABA, 2003). Vertical integration raises a similar conflict for the economic regulation of industries.
While foreclosure concerns offer a rationale to restrict the conduct of vertically integrated firms, faith in market efficiency and doubt
about the regulatory benevolence support a trend toward deregulation (Stigler, 1971). While Chicago School critiques of foreclosure
theory and cautions about the difficulties of collusion (Stigler, 1964) urge a permissive approach to vertical mergers and the regulation
One allegation was that Weyerhaeuser had purposely overpaid for inputs (alder sawlogs) and bought more than it needed in an effort to increase its
rivals’ costs and drive them out of business. The jury returned a verdict for the plaintiff despite finding that the plaintiff had failed to prove that alder
lumber was a distinct product market from all hardwood lumber. In the hardwood lumber market, Weyerhauser had less than a 10% market share
and the jury, in a special verdict, found that Weyerhaeuser lacked market power in that market. Nonetheless the jury awarded damages to the plaintiff
because it found, in accordance with the district court’s instructions, that Weyerhaeuser had purchased more alder sawlogs than “necessary,” paid a
higher price than “needed,” and prevented plaintiff from obtaining logs at a “fair price.” (…) If, but only if, the trier of fact finds that the defendant
enjoys market power in the output market, would it be necessary to determine whether the defendant also enjoyed monopsony (or
oligopsony) power vis-a-vis the input market and, if so, whether it exercised that power in a fashion that enabled it to exercise market power
in the output market. This test would dispose of the “predatory bidding” claims in Weyerhaeuser. As previously stated, the jury found that
Weyerhaeuser did not have market power in the downstream market (or output) – in my view that fact was dispositive. Given my conviction that the
antitrust laws are supposed to protect consumers in that market, I do not believe Section 2 liability should attach to predatory bidding allegations if
it does not create or maintain monopoly power in the downstream (or output) market – or create a dangerous probability of creating that monopoly
power”. (ROSCH, 2006)
21
According to Werden: “Congress responsible for the Sherman Act and the courts that have interpreted the Act were far from indifferent to
the plight of sellers exploited by buyer cartels or monopsonies (…)this essay argues that promoting consumer welfare is a goal of the Sherman Act,
but only a goal, and that making end-user welfare the touchstone under the Act could have extraordinarily undesirable consequences (…) Both enduser welfare and aggregate welfare are concerned with the well being of people, but people normally do not participate in the markets within which
Sherman Act violations occur.111 Determining the legality of business conduct on the basis of its effects on the welfare of people, thus, could force
antitrust analysis to look far beyond the relevant market.112 If end-user welfare were made the touchstone, it would become necessary to trace the
incidence of effects all the way down the distribution chain. This necessarily would impose an additional burden on plaintiffs and the courts;
moreover, in some cases, no end-user harm flows from conduct normally considered anticompetitive. Restricting output and raising price are the
usual effects of cartelization, monopolization, and other conduct addressed by the Sherman Act. Restricting output reduces the efficiency of resource
allocation and thus lessens aggregate welfare. Raising price transfers wealth from trading partners and normally also causes transfers from end users,
but the latter effect need not occur. (WERDEN G. , 2007).Werden was worried if: (i) price increase resulting from the elimination of
competition may affect only the fixed costs of the companies that are the direct purchasers. If so, direct purchaser companies typically
would not pass on the higher prices in the short term, but maybe in long term there could have some effect to consumers (ii) a firm
that engages in predation to become a monopolist in the sale of a consumer product sold to retailers through a two-part pricing scheme
with a fixed fee plus a per-unit charge. The creation of a monopoly at the manufacturing level would raise the fixed fee but not the
per-unit charge, which is optimally set at competitive level.
of vertically integrated industries, Post-Chicago theories of harmful vertical integration nevertheless featured prominently in some recent
merger reviews and regulatory proceedings. (RIORDAN, 2008, p. 145)
Depending on how one should assess the role (and the possibility) of the State in interfering
with vertical agreements, it is possible to see different teleological preferences of what Antitrust
should do, in concrete cases.
So, a merger or a specific conduct can be considered anticompetitive (or not) depending on
a specific teleological approach. It means that the use or abuse of market power is not clearly
harmful or unlawful (or even substantial), irrespectively of a teleological choice of the interpreter.
TL = {TL1, TL2, TL3, …, TLN}
Equation 14 – Interpreter´s teleological available choices
There is a subtle, but important question to address at this point. So relevant to determine
what should be protected or not by law, is to design:
(i)
(ii)
who, personally, has, nowadays, [or should ideally have in a perfect institutional
design] competence to deal with such questions and
what is, nowadays, [or should ideally be] the correct branch of law that should deal
with such matter.
Again, such determinations are made purely by subjective (or dogmatic) preference.
For example, let´s hypothetically say that a merger between all television channels or all
newspapers decreases prices and increases consumer and producer welfare (in terms of quantity
socially produced of newspapers). In such hypothetical example, probably it will have decrease
democratic participation of society, in the sense that different editorial views from the past will
be concentrated in a specific political group. In this sense, the spread of ideas could be harmed by
such merger. Moreover, in this scenario, one could argue if democracy is an issue that should be
protected by Antitrust enforcers? Using this hypothetical merger, one should question if it is a
case to block the merger or not.
One answer [hopefully not shared by many interpreters], is to understand that democracy
should not be protected at all (and such answer or choice is derived from interpreter´s own
subjective preference). Because of this totalitarian and antidemocratic preference, there would be
no room to adjudicate this case at all (inside Antitrust arena or outside this branch of law), having
no reason to block this hypothetical merger.
Another answer is to recognize that there is a problem with such different manners of
“market power” and that democracy should be protected. However, there may be disagreements
if the protector of democracy should be Antitrust Law or not.
The ones who defend that Antitrust Law have no competence to block the merger, they can
raise several arguments. For example, it is possible to claim that government should create other
instances to judge and maybe block this merger. Some may say that such option is better for
transparency purposes, to delimit the competence of Antitrust Authority as an organ who is
responsible to judge ONLY “welfare”, “competitive process” and/or “consumer choices”,
segregating it from other authorities vested with competence to deal with all other kinds of issues,
including justice, fairness, democracy or some other value. Some might even argue that there is
“a widespread agreement” against multiple goals considerations [and everything that is not
categorized as “welfare” or “economic value” is seen pejoratively as “populist goals” of
Antitrust]:
there is now widespread agreement that this evolution toward welfare and away from noneconomic considerations has benefitted
consumers and the economy more broadly. Welfare-based standards have led to greater predictability in judicial and agency decision
making. They also rule out theories of liability (e.g., a transaction will tend to reduce the number of small businesses in a market) and
defenses (e.g., the restraint upon trade is necessary to save consumers from the consequences of competition) that would significantly
harm consumers. Further, the focus upon economic welfare has led the Court to reject per se prohibitions of conduct once thought
anticompetitive but now, owing to advances in our economic knowledge, understood to be efficient. Untethered from an economic
welfare standard, it is difficult to imagine a rationale for eliminating those per se prohibitions. (WRIGHT & GINSBURG, 2013)
[T]here is no reasonable basis for presuming that courts must give priority or even weight to populist goals where the pursuit of such
goals might injure consumer welfare by interfering with competitive pricing, efficiency, or innovation. (…) The pursuit of these goals
would broaden antitrust's proscriptions to cover business conduct that has no significant anticompetitive effects, would increase
vagueness in the law, and would discourage conduct that promotes efficiencies not easily recognized or proved (TURNER, 1987)
("The biggest advantages conferred by the use of relatively traditional microeconomics as the guiding principle for antitrust are two:
coherence and welfare. . . . [P]opulist goals should be given little or no independent weight in formulating antitrust rules and
presumptions. As far as antitrust is concerned, they are substantially served by a procompetitive policy framed in economic terms....
[I]njection of populist goals, by broadening the proscriptions of business conduct, would multiply legal uncertainties and threaten
inefficiencies not easily recognized or proved… [Despite some inadequacies,] economics gives a focus to antitrust interpretation and is
critical to any formulation of rational rules." (AREEDA & HOVENKAMP, 2006, p. 110)
“Populist goal” is a very subjective term, since, for example, what Gregory Werden
(WERDEN G. , 2007) thinks about what populist goals are may be different from Thomas Rosch´s
view about the same thing (ROSCH, 2006).
Of course, when multiple tradeoffs are considered, unfortunately, there is room to defend
inefficient market structures (or to blindly belief in an inefficient structure).22 On the other hand,
as explained here, the lack of “predictability”, “vagueness of law” or “uncertainties” are necessary
effects of fairly considering non-classical logics. One thing is to endorse inefficient structures.
Another thing is to recognize that there are other values besides economic efficiencies
(DWORKING, 1980), such as democracy, for example, that deserves to be protected.
Also, the problem to define what is “economic” or “non-economic” considerations is a little
bit problematic, in the sense that in order to be a “non-economic” issue something should be
immune from scarcity problem and could not be modelled by rational way of thinking, which is
a quite rare phenomenon.
There are other problems with pushing competence to deal with “democratic values” to
other branches of law: specifically, what if there is no previous authority formed to deal with all
other “non-economic” issues derived from concentrated structures?
Besides that, the problem of common goods (and an undefined competence) is that
whatever is everyone´s responsibility to deal with also creates incentives that no one claims
responsibility for taking action in regard of that specific issue. Can Antitrust Agencies states, for
example, that if democracy is harmed by the approval of the hypothetical merger, it is someone
else´s fault? Yes, lawyers, judges and agencies might choose to act like Poncio Pilatos for the
sake of “predictability”. However, there may be other interpretations of what is the current
competence or the ideal competence (or institutional design) about how to deal with so many
multidimensional tradeoffs that Antitrust Law can face.
Another thing that must be taken in consideration is that allowing Agencies to balance “multiple goals”, sometimes,
there will be not so obvious values to fight for or even there may be contradictory goals. For example, the protection
of small business or “small dealers and worthy men”, were values mentioned in United States v. Trans-Missouri Freight
Ass’n, 166 U.S. 290, 323 (1897). Also, some other precedents tried to ascertain that the goal of Antitrust is to protect
“small, locally owned businesses” (Brown Shoe Co. v. United States, 370 U.S. 294, 344 (1962); accord, e.g., United
States v. Alcoa, 148 F.2d 416, 427 (2d Cir. 1945).
22
Here, also, it is important to stress that some authors may believe that structure (or
institutional governmental design) can influence conduct of Antitrust enforcers and its
performance to reach good or bad Antitrust decisions over these multidimensional tradeoffs [and
that is why some prefer to ascertain Antitrust competence as an absolute “should be”].
On the other hand, there may be the case that a specific governmental design may not
guarantee good decisions [or predictable decisions]. Before judging what is a good or a bad
antitrust decision, firstly, it is necessary to know what is the teleological choice of the interpreter,
that should be observed.
Sometimes it is difficult to say how teleological preferences could interfere in the analysis
if a merger or a conduct creates or endorses an “abusive” market power or not; and if such merger
or conduct should be accepted or not, in Antitrust grounds.
8. Non-structural analysis
Jonathan Baker23 and W. Blumenthal24 consider that if it is possible to show Market Power
or competitive harm directly, there is no need to delineate the borders of relevant market, in the
sense that the “thing” (effect) speaks by itself [“Res ipsa loquitur”]. Such Latin wording is derived
from the speech Pro Tito Annio Milone ad iudicem oratio (Pro Milone). Marcus Tullius Cicero,
52 b.C., made such speech in the benefit of his friend Titus Annius Milo, that was accused to kill
Publius Clodius Pulcher. Cícero argued that, considering circumstantial evidence, there was no
need to prove directly the innocence of his friend, because circumstance speaks by itself (in other
words, it was so clear that Titus was innocent that Cícero did not have to prove any other thing).
Although ironically Cícero lost such case25, this principle was adopted in Byrne v Boadle (2 Hurl.
& Colt. 722, 159 Eng. Rep. 299, 1863) by Chief Baron Pollock in a Tort Law claim, and rescued
by Baker in Antitrust Law26.
The problem, however, is that even if someone is not skeptical about res ipsa loquitor, it is
possible to assert that thing, sometimes, can speak by itself, but it needs a translator to be correctly
understood by the interpreter. In this sense, what is seen to be a more sophisticated (or even
superior) way to deal with market power (without the need to define market or lerner index) is
not exempt by subjective influences. It is also not possible to guarantee that the outcome of this
approach will be, necessarily better than some alternative procedure or even immune from
uncertainty.
It is important to stress that, although some extremist views of non-structuralist approach
claims that market definitions are not needed, databases used in Econometric exercises are
tailored to contain just a limited number of variables. Even a “non-struturalist” approach may be
dependent, in some degree, of heuristics simplification.
8.1.
With control groups
In Antitrust Law, one can try to measure the effects of some practices, in order to determine
if conduct cases generate some specific harm (using a specific standard). For example, it is
possible to look to what happened in prices before and after the observed conduct [and compare
such result with prices of markets where conduct did not happen]. Also, in merger cases it is
possible to use this analysis to see if the merger will create unduly market power. However, there
is no univocal way of making such “direct inference” on the effects of a specific conduct or
23
BAKER, Jonathan. Product Differentiation Through Space and Time: Some Antitrust Policy Issues. Disponível em: <
http://www.ftc.gov/public-statements/1996/02/product-differentiation-through-space-and-time-some-antitrust-policy >. Acesso em:
17 de agosto de 2009.
24
BLUMENTHAL, William. Why Bother?: On Market Definition under the Merger Guidelines. FTC/DOJ Merger Enforcement
Workshop. Disponível em: <http://www.justice.gov/atr/public/workshops/docs/202600.htm>. Acesso em: 13 de maio de 2010.
25
MURRAH, Morgan. Res ipsa loquitur De acordo com o site http://officialinformationact.blogspot.com.br/2012/10/the-thingspeaks-for-itself-usually-but.html verificado em 20 de outubro de 2014.
26
Baker, Jonathan B. and Reitman, David, Research Topics in Unilateral Effects Analysis (November 9, 2009). American University,
WCL Research Paper No. 09-37. Available at SSRN: http://ssrn.com/abstract=1504863 or http://dx.doi.org/10.2139/ssrn.1504863
merger. Indeed, there are many ways of making such measurement. For example, it is necessary
to choose an adequate algorithm to perform treatment effects analysis, and there is a wide menu
for that:
𝑦 = 𝛽0 + 𝛽1 𝑑𝑇 + 𝛿0 𝑑𝑡 + 𝛿1 𝑑𝑡𝑑𝑇 + 𝑢
Equation 15 - Difference-in-Difference analysis (DD)
̅̅̅̅̅̅̅̅̅̅𝑏𝑒𝑓𝑜𝑟𝑒 − 𝑐𝑜𝑛𝑡𝑟𝑜𝑙_
̅̅̅̅̅̅̅̅̅̅̅𝑎𝑓𝑡𝑒𝑟 ]
̅̅̅̅̅̅̅̅̅̅̅̅̅̅𝑏𝑒𝑓𝑜𝑟𝑒 − ̅̅̅̅̅̅̅̅̅̅̅̅̅̅
𝛿̂1 = [𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡
𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝑎𝑓𝑡𝑒𝑟 ] − [𝑐𝑜𝑛𝑡𝑟𝑜𝑙
Equation 16 – DD estimator
𝑦 = 𝛽0 + 𝛽1 𝑑𝑇 + 𝛽2 𝑑𝑂 + 𝛽3 𝑑𝑇𝑑𝑂 + 𝛿0 𝑑𝑡 + 𝛿1 𝑑𝑡𝑑𝑇 + 𝛿2 𝑑𝑡𝑑𝑂 + 𝛿3 𝑑𝑡𝑑𝑇𝑑𝑂 + 𝛽2 𝑋 + 𝑢
Equation 17 – Triple Difference or Difference-in-Difference-in-Difference analysis (DDD)
y = outcome of interest.
dT = dummy variable differentiating treatment and control groups
dt = dummy variable differentiating before and after treatment
dO = dummy variable differentiating other dimension
u = residuals
* It is possible to include other control variables in these equations
(GRUBER, 1994) (CARPENTER & STEHR, 2011)
If the implementation of these models is made through panels (and the effect is observed
only on the second period of time), the interaction term may not be needed.
𝑦𝑖𝑡 = 𝛽0 + 𝛿0 𝑑𝑡 + 𝛽1 𝑑𝑇𝑖𝑡 + 𝑎𝑖 + 𝑢𝑖𝑡
𝑦𝑖𝑡 = outcome of interest of individual i in time t
dt = dummy variable differentiating before and after treatment
dT = dummy variable differentiating treatment and control groups
𝑎𝑖 = fixed effects
𝑢𝑖𝑡 = residuals
Equation 18 – DD in panel analysis
𝑦𝑖𝑡 = 𝛽0 + 𝛿0 𝑑𝑡 + 𝛽1 𝑑𝑇𝑖𝑡 + 𝑎𝑖 + 𝑢𝑖𝑡
(𝑦𝑖2 − 𝑦𝑖1 ) = 𝛽0 + 𝛽1 (𝑑𝑇𝑖2 − 𝑑𝑇𝑖1 ) + (𝑢𝑖2 − 𝑢𝑖1 )
∆𝑦𝑖 = 𝛽0 + 𝛽1 ∆𝑑𝑇𝑖 + ∆𝑢𝑖
E (∆𝑑𝑇𝑖 ∆𝑢𝑖 = 0)
Equation 19 – First Difference in panel - individual level (WOOLDRIDGE, 2003, p. 427)
𝑦𝑖𝑠𝑡 = 𝐴𝑠 + 𝐵𝑡 + 𝑐𝑋𝑖𝑠𝑡 + 𝛽𝐼𝑠𝑡 + ∈𝑖𝑠𝑡
𝐴𝑠 = 𝑓𝑖𝑥𝑒𝑑 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 𝑠𝑡𝑎𝑡𝑒
𝐵𝑡 = 𝑓𝑖𝑥𝑒𝑑 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 𝑡𝑖𝑚𝑒
𝑋𝑖𝑠𝑡 = 𝑟𝑒𝑙𝑒𝑣𝑎𝑛𝑡 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠
∈𝑖𝑠𝑡 = 𝑒𝑟𝑟𝑜𝑟 𝑡𝑒𝑟𝑚
𝛽̂ = 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑛𝑡𝑒𝑟𝑣𝑒𝑛𝑡𝑖𝑜𝑛
Equation 20- Generalization of DD for multiple controls – panel (BERTRAND, DUFLO, & MULLAINATHAN, 2004, p. 2)
There are, also, some concerns about DD estimator:
“conventional DID estimator is based on strong assumptions. In particular, the conventional DID estimator requires that in absence of the
treatment, the average outcomes for treated and controls would have followed parallel paths over time. This assumption may be implausible
if pre-treatment characteristics that are thought to be associated with the dynamics of the outcome variable are unbalanced between the
treated and the untreated group. (ABADIE, 2005)
Also, (BERTRAND, DUFLO, & MULLAINATHAN, 2004) identified that DD can
present “severe serial correlation”. To administer such problem, authors made some suggestions,
as: (i) the use of nonparametric technique, block bootstrap (EFRON & TIBISHIRANI, 1994); or
(ii) to remove time dimension, aggregating it into only two periods (pre and post treatment); (iii)
among other possible solutions.
There are also who suggested, in realm of DD, the use of: feasible generalized least squares
estimators (HAUSMAN & KUERSTEINER, 2008); semiparametric estimators (ABADIE,
2005); and others. With multiple control groups, there are authors who suggest some
simplification of DD models. For example, (ABADIE, DIAMOND, & HAINMULLER,
Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s
Tobacco Control Program, 2010, p. 503) claim that “comparative case study research has broad
potential in the social sciences. However, the empirical implementation of comparative case
studies is plagued by inferential challenges and ambiguity about the choice of valid control
groups”. Because of that, these authors, in the presence of two (or more) control groups, prefer
the use of a synthetic control group [SCG] (a weighted average of the available control units).
The weights that are suggested by the authors depend basically on a vector of observed covariates
(not affected by the intervention) and a vector of unobserved common factors [see for example
command synth in stata and in R]. After determining SCG, authors perform a “placebo-like
technique”, by measuring intervention (treatment) mean squared prediction error (MSPE), as a
matter of gap between SCG and treated group, and compare such value with the gap between
SCG and other control groups individually.27
When some advocate the use of SCG (aggregate control group), others think that control
groups should be disaggregated. According to Stephen Donald and Kevin Lang, the lack of
control of “common groups” or different groups in a sample, in some DD analysis, can generate
bias:
“in the typical differences-in-differences model, we regress outcomes at the individual level (e.g. employment in a firm in state s in year t)
on a policy that applies to all individuals in the group (e.g. the minimum wage in state s in year t). Moulton (1990) shows that in regression
models with mixtures of individual and grouped data, the failure to account for the presence of common group errors can generate estimated
standard errors that are biased downwards dramatically. The differences-in-differences estimator is a special case of this model. Researchers
use a number of standard techniques to adjust for common group effects:
• random-effects feasible GLS estimation which under certain conditions is asymptotically efficient,
•correcting the standard errors using the error covariance matrix based on common group errors as in Moulton,
• correcting the standard errors using a robust covariance estimator according to a formula developed by Liang and Zeger (1986) and more
commonly known as the Stata cluster command
27
For example, it is important to see if an individual who/ receives treatment is subject to selection bias or not, even
in non-random assignment: 𝐸(𝑦𝑖𝑡 | 𝑑𝑇 = 0) ≠ 𝐸(𝑦𝑖𝑡 | 𝑑𝑇 = 1) (HECKMAN & HOTZ, 1989). Also, some authors use
placebo-like techniques: “The inferential techniques proposed in this article are related to Abadie and Gardeazabal (2003). In their study of
the economic effects of terrorism, Abadie and Gardeazabal (2003) use a synthetic control region to estimate the economic growth that the Basque
Country would have experienced in the absence of terrorism. To assess the ability of the synthetic control method to reproduce the evolution of a
counterfactual Basque Country without terrorism, Abadie and Gardeazabal (2003) introduce a placebo study, applying the same techniques to
Catalonia, a region similar to the Basque Country but with a much lower exposure to terrorism. In this paper, we extend the idea of a placebo study
to produce quantitative inference in comparative case studies. The idea of the placebo test proposed here is akin to the classic framework for
permutation inference, where the distribution of a test statistic is computed under random permutations of the sample units’ assignments to the
intervention and nonintervention groups. As in permutation tests, we apply the synthetic control method to every potential control in our sample.
This allows us to assess whether the effect estimated by the synthetic control for the region affected by the intervention is large relative to the effect
estimated for a region chosen at random. This inferential exercise is exact in the sense that, regardless of the number of available comparison regions,
time periods, and whether the data are individual or aggregate, it is always possible to calculate the exact distribution of the estimated effect of the
placebo interventions. Notice also that the inferential exercise proposed here produces classical randomization inference for the case where the
intervention is indeed randomized across regions, a rather restrictive condition. More generally, our inferential exercise examines whether or not the
estimated effect of the actual intervention is large relative to the distribution of the effects estimated for the regions not exposed to the intervention.
This is informative inference if under the hypothesis of no intervention effect the estimated effect of the intervention is not expected to be abnormal
relative to the distribution of the placebo effects. In this sense, our inferential procedure is related to that of DiNardo and Pischke (1997) and Auld
and Grootendorst (2004). DiNardo and Pischke (1997) compare the wage differential associated with computer skills (as reflected in the on-the-job
computer use) to the wage differentials associated with the use of other tools (pencils, telephones, calculators) that do not proxy for skills that are
scarce in the job market”. (ABADIE, DIAMOND, & HAINMULLER, Synthetic Control Methods for Comparative Case
Studies: Estimating the Effect of California’s Tobacco Control Program, 2010, p. 503)
(…) standard asymptotics cannot be applied when the number of groups is small as in the case where we compare two states in two years,
two cities over a small number of years, or self-employed workers and employees over a small number of years. In such cases, failing to
take account of the group-error structure will not only generate underestimates of the standard errors as in Moulton, but applying the normal
distribution to corrected t-statistics will dramatically overstate the significance of the statistics”. (DONALD & LANG, 2007, p. 221)
Independently how to treat different control groups, maybe it would be possible or even
preferable to use a multilevel mixed-effects linear regression, in some cases, incorporating random
and fixed effects.
Also, it is important to know if there are spillover effects from the treatment to control
group or not. This assessment is important in order to guarantee that control group is not
influenced by treatment [and that stable unit treatment value assumption – SUTVA, remains valid
after treatment] or, at least, if there is a manner to measure or to treat this problem:
“Communication between participants, band wagon effects, and other social psychological processes may violate what Rubin (1986) has
termed the "stable unit treatment value assumption" (SUTVA), which is routinely invoked when drawing causal inferences about
experimental effects. SUTVA holds that there is no interference between units; the experimental assignment of one subject has no effect
on other subjects' potential outcomes.1 SUTVA rules out "spillover effects" that occur, for example, when treated individuals transmit the
information contained in the treatment to the control group (Rosenbaum 2007). Other examples of SUTVA violations outside the realm of
elections include the displacement of crime from treatment areas that receive heightened police surveillance to control areas (Sherman and
Weisburd 1995), social comparisons that cause the control group assessments to be influenced by the intervention received by the treatment
group (Sobel 2006), strategic interaction between subjects such that the control group adjusts its behavior in light of prior treatments and
treatments received by others (Bednar et al. 2010), and strategic calculations that lead political actors in one jurisdiction to take cues from
neighboring jurisdictions that receive a treatment, such as financial audits or election monitoring (Hyde 2010; Silva 2010). Although
SUTVA is fundamental to causal analysis, experiments (as well as observational studies) have typically downplayed the possibility of
spillovers.” (SINCLAIR, McCONNELL, & GREEN, 2012)
Therefore, one should always analyze if there are spillover effects from treated group to
control group.
Another path that interpreter can opt is to use a quasimaximum likelihood (QML) technique
to infer equation estimators (EE) in the outcome model (OM) or the treatment model (TM) in
order to compare them, by a system of equations.
Through this procedure it is possible to run effect analysis through Regression adjustment
[there are who are in favor or who criticize such method, see (LIN, 2013), (FREEDMAN, 2008)
(RUBIN, 1974) ]; Endogenous treatment-effects estimation (CERULLI, ivtreatreg: A command
for fitting binary treatment models with heterogeneous response to treatment and unobservable
selection, 2014); Linear regression with endogenous treatment effects (VELLA & VERBEEK,
1999); Inverse-probability weighting (CERULLI, treatrew: A user-written command for
estimating average treatment effects by reweighting on the propensity score, 2014); Augmented
inverse-probability weighting; Inverse-probability-weighted regression adjustment; Multivalued
treatment effects; Nearest-neighbor matching; propensity-score matching, among several other
valid methodologies.
These estimators observe some assumptions, such as:



independent and identically distributed (i.i.d.) sampling - ensures that the outcome and
treatment status of each individual are unrelated to the outcome and treatment status of
all the other individuals in the population.
conditional-independence (CI) - once there is a control for all observable variables, the
potential outcomes are independent of treatment assignment;
overlap - each individual has a positive probability of receiving treatment.
Given that y1 are individuals who got treatment (t=1) and y0 are individuals of control
group (t=0) , it is possible to measure:
ATE = E(y1 - y0)
Equation 21 - Average treatment effect (ATE) in the population
POMt = E(yt)
Equation 22 - Potential-outcome mean (POM)
ATET = E(y1 - y0|t = 1)
Equation 23 - Average treatment effect on the treated (ATET)
The algorithm may change depending on the selection of functional form of model, that
could be linear; logit; probit; hetprobit; poisson; and other forms.
Susan Athey and Guido W. Imbens, also, developed another non-parametric and nonlinear
model, called Change-in-Changes (ATHEY & IMBENS, 2006). Other authors, also, tried to
combine approaches regarding the use of methods based on the exogeneity assumption (such as
matching methods) and methods based on change-score (difference-in-differences) (HO, IMAI,
KING, & SUART, 2007).
There is a wide menu to choose regarding what the proper algorithm is and how to best
design the experiment in order to draw a good counterfactual analysis.
In this sense, variables themselves can influence the result. It is very important to possess
a good and trustable database, in order to make this kind of analysis.
Taking into account all possible ways to perform a regression using control groups, it is
possible to say that the problem of robustness check, convergence of results, and the use of nonclassical logic, can arise in this kind of exercises as well. Of course, there could be discussions if
the control group defined for the exercise is indeed [or not] a valid control group for subjective
reasoning.
8.2.
Without control groups
Looking to level of variables can be informative (work as an evidence) to know how much
market power some enterprise has.
For example, it is important to see: (i) if consumers are discriminated; (ii) if there is stability
of market shares; (iii) if production or selling decrease with simultaneous profit increase; (iv) if
level of profitability or level efficiency28 is compared to other industries; (v) what are the levels
of demand price elasticity and supply price elasticity; (vi) if there are capacity constraints in the
industry; (vii) how many competitors are in the market; (viii) how many consumers are in the
market; (ix) how easy is to entry in the market; (x) if competitors should negotiate among
themselves and how often; (xi) if there is transparency in relation to price, production, capacity,
bids or other variables; (xii) among several other important factors.
Also, someone might be interested to know how a specific conduct changed price or
profitability of a given enterprise (and consequentially its own market power), and that may be
done through a chow test or any test for a structural break test, to see if that specific act or conduct
changed the observed parameter (to know the effects of an exclusive contract, a refusal to deal
and so on).
28
Jan Boone coined the concept of relative profit differences (RPD), that deals with profitability and efficiency. According to the
author it is defined as follows. “Let (n) denote the variable profit level of a firm with efficiency level n ∈ ℝ+ where higher n denotes
higher efficiency (more details follow below on how variable profits and efficiency are defined). Consider three firms with different
efficiency levels, n’’ > n’ > n, and calculate the following variable
[π(𝑛’’)− π (𝑛)]
RPD = [π(𝑛’)− π (𝑛)] (BOONE, 2008)
For example, if b is “break date” (when an anticompetitive conduct happened, or a merger
or any other event, such as entry or exit from the market of a given player), then, a very simple
model that would capture structural break could be the following one:
𝑦 = 𝑥𝑡 𝛽 + 𝜖𝑡
𝑖𝑓 𝑏 ≤ 𝑡
𝑦 = 𝑥𝑡 (𝛽 + 𝜕) + 𝜖𝑡 𝑖𝑓 𝑏 > 𝑡
𝐻𝑜 = 𝜕 = 0 = 𝑁𝑜 𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑎𝑙 𝑏𝑟𝑒𝑎𝑘
𝐻1 = 𝜕 > 0 = 𝑆𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑎𝑙 𝑏𝑟𝑒𝑎𝑘
Equation 24 – Structural break test
Of course, this is a very simplistic model, and there may be other ways to perform such
analysis. For example, it is possible to run an event study to know how a specific rate of return of
a specific activity [y] suffers the impact of given event “s” [D] (SHARPE, 1963) (LAFONTAINE
& SLADE, 2007):
𝑅𝑖𝑡 = 𝛼𝑖 + 𝛽𝑖𝑡 𝑅𝑚𝑡 + ∑ 𝛾𝑠𝑖 𝐷𝑠𝑡 + 𝑢𝑖𝑡
𝑠
𝑅𝑖𝑡 = 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑎𝑠𝑠𝑒𝑡 𝑖 𝑖𝑛 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡
𝑅𝑖𝑡 = 𝑚𝑎𝑟𝑘𝑒𝑡 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑏𝑟𝑜𝑎𝑑𝑙𝑦 𝑏𝑎𝑠𝑒𝑑 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑖𝑛 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡
𝐷𝑠𝑡 = 𝐷𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑒𝑣𝑒𝑛𝑡 "𝑠"
𝐷𝑠𝑡 = 1 𝑖𝑓 𝐸𝑣𝑒𝑛𝑡 "𝑠" 𝑜𝑐𝑐𝑢𝑟𝑟𝑒𝑑 𝑜𝑟 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Equation 25 – Event analysis
There are many events that can be measured: capacities constraints, exit or entry of
competitors, specific conducts, among other things.
One of the several kinds of possible events studies deals with the evaluation of the reaction
of agents who invest in the financial market when they receive the news about the existence of a
merger. Through this methodology, the price reaction of shares of merging parties and its
competitors are measured after the announcement of the operation, to assess whether the financial
market expects supranormal pricing after completion of the transaction. On the other hand, there
must be a high degree of care in handling this type of methodology, since the explanatory power
of this type of test can be contested [especially if the financial markets are not efficient] or if other
factors that may explain share movements, both of merging parties and of its competitors, are not
fully comprehended by the econometric model.. (ECKBO, 1983) (McAFEE & WILLIAMS, 1988).
Another path is to treat all time series “breakings” or events as unknown features, in order
to, in a second moment, try to verify if Econometric model find these breaks and if they coincide
with the event that is being analyzed. It is possible to run – for example - a Markov-switching
regression. Such regression exhibit different dynamics across unobserved states using statedependent parameters to accommodate structural breaks or other multiple-state phenomena. If the
regression presents evidence that the structural breaks coincides with the expected conduct, then,
it may be possible to infer some impact or effect derived from the conduct itself.
The problem that appears when control groups are not used is that sometimes some
variation of price or profitability may not be due to the observed event, but for something else,
that is not being properly controlled or observed by the interpreter.
For example, in Brazil, May 2007, SDE, together with SEAE, the Federal Police and the
State Prosecutors of the State of Paraíba launched a dawn raids in João Pessoa and Recife to
obtain evidence of a cartel in the fuel retailer sector. The operation involved 190 agents that
searched 26 different places and served 16 prison warrants. The dawn raids were called “274
Operation”, named because the investigated cartel allegedly agreed for the liter of gasoline to set
prices in R$ 2,74.
After dawn raids, the price decreased from R$ 2,74 to R$2,37.
Figure 12 – Pact 274
Is it possible to say that the dawn raids decreased the price? Maybe, but it is important,
also, to understand if such decrease were not due to simultaneous cost reductions, to simultaneous
demand decrease, or to some other simultaneous explanatory variable that is not controlled, by an
econometric exercise, in order to avoid a spurious correlation.
There are authors that tried to build models to infer market power, from certain parameters,
not using control groups. Indeed, “conduct parameter method (CPM), which employs an
empirical model based on the theory of conjectural variations to estimate a conduct parameter.
This parameter is purported to measure the competitiveness of a market in a very general way,
yielding an elasticity-adjusted price-cost margin and simultaneously nesting the perfectly
competitive, monopoly, and classical Cournot models”. (CORTS, 1999, p. 228)

Bresnahan-Lau Model - This model (BRESNAHAN, 1982) (LAU, 1982) suggests the
rotation of the demand curve in order to estimate the degree of power (). If the enterprise
has no market power whatsoever (), according to the model, a shift on the demand
slope (given that supply slope is not changed) should not affect the equilibrium price. The
reason for that is that the price, in perfect competition, is equal to marginal costs. Since
marginal costs did not change (only demand slope change) such change should not
influence equilibrium price. However, if there is some degree of market power (),
then, in this situation, if there is a rotation of demand curve (and supply curve is not
changed), the equilibrium price is expected to vary (and the parameter tries to identify
this situation).29
Kenneth Corts claimed that this strategy has some flaws. Corts seems to have made a
very simplified approach to Bresnahan-Lau Model. For example, the author in his model
used a constant marginal cost assumption and did not include in his analysis an
interaction term that would take in consideration only rotation of demand curve to see
how prices would react to it. Corts was worried that market power and supply slope could
vary over time affecting the result of conduct parameter estimate, that, in his view, “is
29
See also (PORTER, 1983)
valid only if the true process underlying the observed equilibrium generates behavior that
is identical on the margin, and not just on average, to a conjectural variations game”.
(CORTS, 1999, pp. 234-235).
After that, author used Nash´s game theory to test how N-firm symmetric oligopoly game
would behave in an environment where there were an efficient supergame equilibrium in
which deviations were punished by reversion to one-shot Cournot equilibrium strategies
forever. The author concluded that:
“ if observed equilibrium behavior results from efficient supergame collusion, the estimated conduct parameter underestimates the degree
of market power if demand shocks are not fully permanent, and may fail to detect any market power whatsoever when demand shocks
are completely transitory, even if average price-cost margins are near the monopoly level”. (CORTS, 1999)
Maybe such analysis could have a different outcome if Bresnahan-Lau Model with
interaction term were used. Nonetheless, it is an important warning on how this kind of
methodology can present some limitations.

Panzar-Rosse Model – According to this empirical methodology (PANZAR & ROSSE,
1987), the measure of marginal cost - price elasticity (called as H statistics) could be
informative of market power. The “key point is that a monopolist´s output and total
revenue decline when his marginal cost curve shifts upward. On the contrary, in a
perfectly competitive sector, an increase in marginal costs would be fully reflected in
price, thus increasing total revenues one-to-one for the sector as a whole. In between
these two extremes is the case of oligopolistic structure: as the marginal cost curve shifts
upward, total revenues increase by less than one-to-one with the increase in costs”
(BELAISCH, 2003).
Other authors, however, understand that:
“a Panzar-Rosse price function or scaled revenue equation – which have both been widely applied in the empirical competition literature
– cannot be used to infer the degree of competition. Only an unscaled revenue equation yields a valid measure for competitive conduct.
Our theoretical findings have been confirmed by an empirical analysis of competition in banking industry, based on a sample covering
more than 11,000 bank year observations on almost 18,000 banks in 67 countries during 1986-2004 period. (BIKKER, SHAFFER, &
SPIERDIJK, 2009)

Carman – Sexton Model - (CARMAN & SEXTON, 2005) – These authors tried to
measure the amount and the speed in which retail market pass through consumers
increases or decrease of costs, to know if they are non-reversible functions (HOUCK,
1977).
There can be many other manners to perform a non-structural analysis.
There are a lot of strengths and weaknesses that such techniques can present.
Sometimes, parties try to use this kind of methodology in order to invalidate direct
evidences of cartel. About this, Counsellor Fernando de Magalhães Furlan understood that:
“Claims based on projections of possible economic scenarios and informed on econometric studies derived from a predefined fact
have reduced evidentiary strength when confronted with direct evidences of the case. There is direct evidence of collusion. There are
rules regulating the cartel interactions. There is explicit division of customers. There are even current accounts at the headquarters of
each company to calculate gains and division of the illegal agreement. And in the face of all this myriad of evidence, there are opinions
saying that the market cannot afford to support a cartel and that there is no evidence of collusive behavior. There is no credible
interpretation authorizing such arguments in the face of the evidences in the file. Companies and executives do not spend years writing,
storing and supplying each other with documents proving an illegal practice that do not perform. I cannot conceive that several
companies have come together to set rules of a cartel only for sport. I do not admit the idea that documented and explicit mentions
of division of markets and customers are only acts of concealment that had no intention to occur. Such allegations are not only baseless,
but, with all due respect, absurd.”
(Processo Administrativo 08012.009888/2003-70, com os seguintes representados: AGA S.A., Linde Gases Ltda., Air Liquide Brasil Ltda.,
Air Products Brasil Ltda., Indústria Brasileira de Gases Ltda., S.A. White Martins, White Martins Ltda, White Martins Gases Industriais
Ltda , Carlos Alberto Cerezine, Gilberto Gallo, Hélio de Franceschi Junior, José Antônio Bortoleto de Campos, Moacyr de Almeida Netto,
Newton de Oliveira, Vitor de Andrade Perez e Walter Pilão).
Indeed, when economists try to ignore direct evidence of a cartel, and, based on this
dissociation from reality, try to affix the inexplicable conclusion that there would be no conditions
for a cartel in the sector (based on a Bresnahan-Lau method; Panzar-Rosse method; or other), it
seems that science is being used for tautological purposes (where reality serves to explain the
theory, and not the other way around).
Such tests should be used with a great amount of cautions and restraints, when used to
prove an anticompetitive conduct or to infer market power, in the sense that strategy of
identification may not reflect properly the level of competition of the sector or may not capture,
rightfully, how strategic performance of enterprises allocate gains of a cartel, in a repeated game
scenario.
There are other models that try to understand [or simulate] “collective” market power
derived from a specific merger, such as CPPI (MORESI, REITMAN, SALOP, & SARAFIDIS,
Gauging Parallel Accommodating Conduct Concerns with the CPPI, 2011); Model of
Compte/Jenny/Rey (COMPTE, JENNY, & REY, 2002); cGUPPI (MORESI, REITMAN,
SALOP, & SARAFIDIS, cGUPPI: Scoring incentives to engage in Parallel Acommodating
Conduct, 2015), among others [See also Merger cases n°08012.010195/2004-19 Suzano Bahia
Sul Papel - Celulose S.A., Votorantin Celulose e Papel - VCP - Ripasa S.A. Celulose e Papel and
Merger nº 08700.000658/2014-40 BRF S.A. - Minerva S.A.].
The important fact to stress is that all methodologies can present some limitations
inherently from the process of statistical inference and inherently to the difficult task that is to
choose the “rightful” (or most suitable) identification strategy.
9. International Relations and “market” power
Market power and some Antitrust Theories based on abstract models are issues generally
studied in abstract terms, looking to consumer or social welfares in a vacuum, protected from real
world and from International Relations Theories of how world politics works. Indeed, Neorealism
(MEARSHEIMER, The tragedy of Great Power Politics, 2001) (MEARSHEIMER, 1990)
(WALTZ, 1979) and Critical Theory (COX, 1981) (ASHLEY & WALKER, 1990) (ASHLEY R.
, 1984) (LINKLATER, 1996), and several others points of view seem to be apart from the classical
approach of Antitrust over a series of issues. These International Relation´s theories try to
understand “balance of power”, a term barely used in Antitrust.
There are many reasons why a country desires to protect its own markets (increasing market
power to domestic enterprises and decreasing market power of foreigners), based on “national
interests”.
Protectionism is several times linked with greed of specific sector driven by patrimonialism
[rent seeking] or corruption, or any other manner to obtain undue, illegitimate and personal
advantages over the entire society, protecting some firms against a healthy international
competitive environment. Therefore, protectionism is viewed as a form to benefit few in detriment
of society´s welfare. That, certainly, exists.
On the other hand, there may be other values that may justify, at least, some considerations
about it. Indeed, countries:






can try to protect internal jobs, in short term, when an internal acute macroeconomic crisis is able
to drive a series of national enterprises to bankruptcy;
can worry about internal security matters [trying to restrict trade against enemies, to maintain
balance of power, to achieve political purposes, to defeat terrorism or some other goal];
can consciously try to avoid a predatory subsidy from another area of the world [that will harm in
short term internal industry, but in long term will allow market power of this new subsidized
foreign enterprise, similarly to a national anticompetitive conduct of predatory price. In this case,
fostering national enterprise can represent a remedy against such undue international practices];
can boycott certain goods that are produced without a specific social standard (environmental
standard, for example {as discussed in Shrimp/Turtle WTO Panel or in Tuna/Dolphin WTO
Panel})
can try to protect the nation against unilateral anticompetitive acts of other nations that restrict
important technology, creating and favoring certain national enterprises, that are engaged with
R&D internally.
among several other goals, that are not limited to benefit few in detriment of society´s welfare.
The irony of Antitrust is that fiercely defense of competition with fancy theories that focus
on “welfare based on price” (and claim that that defense of other values is just a “populist”
decision) seems to deny autistically other non-price values/tradeoffs, contrasting with worldwide
dependence on some other topics. This antitrust silence, surrounded by diplomatic and strategic
interests, that, normally, Antitrust forums do not dare to deal with, hinders, instead of fostering,
the fight against undue international [and here called as “sovereign”] market power (that is
different from domestic and civil market power).
SM = {0,1}
Equation 26 – Nature of Market Power
* SM = 0 in case of a domestic use of market power or 1 in case of a “sovereign” market power
To show how market power is intertwined with international political pressures and interest
rather than just marginal costs, some industries will be analyzed, very briefly, in this article, as
examples of how tricky certain topics can appear in International Antitrust realm.
9.1.
Oil industry and “sovereign” cartels?
In this regard, for example, Christopher Lento made a very good picture of how Antitrust
Law conflicts with some rules of Organization of Petroleum Exporting Countries (OPEC), while,
at same time, diplomatic interests are at stake. His view is from a United States and European
perspective, but it seems that other countries or regions have not so different treatment to this
question:
“The United States’ dependence on foreign oil has historically led to an enormous reliance on the goodwill of an organization that is per
se illegal under U.S. antitrust law, but antitrust authorities are conspicuously quiet when it comes to confronting OPEC’s market controls
and manipulations. In the last few decades, there have been suggestions that the United States may have the means to assert subject-matter
jurisdiction over OPEC and should move towards pursuing claims against the organization despite the dismissal of an earlier attempt to
assert jurisdiction in 1981. There, the court held that OPEC was shielded from jurisdiction because it was acting in its governmental capacity
when it regulated production. However, in 1993, this earlier contention was cast into doubt by Hartford Fire Insurance Co. v. California ,
in which the Supreme Court ruled that the Sherman Act could be applied to the acts of foreign corporations committed in foreign countries
“that were meant to produce and did in fact produce some substantial effect in the United States,” which OPEC’s actions consistently do.
Increasing gasoline prices and higher prices for oil used by industry, such as airplane and jet fuel, have also led to legislative pressure from
both the Senate and House. In April 2000, and again in April 2001, Senator Arlen Specter sent letters to both President Clinton and President
Bush urging litigation against OPEC. Senator Specter also argued in front of Congress on June 22, 2005, urging the legislature to find that
OPEC’s immunity from jurisdiction be nullified. In the past, OPEC’s immunity from suit sprung from its classification as a governmental
entity, which would place it under the Foreign Sovereign Immunities Act of 1976, providing that a foreign defendant shall be immune from
suit in any federal or state court if the defendant qualifies as a “Foreign State” and unless a statutory exception to immunity applies.
However, Senator Specter argued against the courts, finding that OPEC Members’ cooperation to fix pricing was a “governmental activity”
as opposed to “commercial activity” and suggested that OPEC should be subject to suit in either U.S. federal court or the International
Court of Justice at the Hague. (…)
In 2011, the Fifth Circuit Court of Appeals affirmed the dismissal of a major challenge to OPEC’s antitrust violations. In Spectrum Stores,
et al. v. Citgo Petroleum Corp., et al., a group of gasoline retailers brought class actions against oil companies owned by OPEC Member
Nations, alleging that the national oil companies conspired with OPEC to fix crude oil prices in the United States through production limits.
Although the suit was brought against oil production companies rather than the OPEC Member Nations themselves, the Fifth Circuit
affirmed the dismissal by the district court under both the political question and act of state doctrines. (….) citing the act of state doctrine,
the Court of Appeals held that adjudication of the suit would call into question the acts of foreign governments concerning their natural
resources, which was outside the sphere of the Judicial Branch
(…)
While this is damaging in a concrete sense, it is mitigated by the fact that the price of oil is pinned to the dollar, and the normal standards
of economics are skewed as far as they pertain to the United States. Most oil importing countries are forced to reserve capital in the form
of U.S. dollars in order to maintain imports at the necessary levels, and oil exporting countries similarly hold, as their currency reserve,
billions in U.S. dollars, the currency in which they are paid. This reservation of reserve capital in U.S. dollars, in turn, creates a constant
upwards pressure on the dollar, independent of economic conditions within the United States; this upward pressure on the dollar allows the
United States to discount bond rates to other countries. Because of these discounted bond rates, oil exporters and producers are able to
invest profit made on oil straight back into the U.S. economy, with virtually zero currency risk. This allows the United States to run higher,
and virtually permanent, trade deficits at a more sustainable level than most other countries and also maintains relatively low prices on
imported goods.” (LENTO, 2014)
It is not by chance that some countries have more “ability” to finance their deficits while
other, as Brazil, that has to import overcharged oil, suffer the deadweight loss of some
anticompetitive “sovereign” practices.
The origin of OPEC market power is based on a predefined concept of what is immune of
adjudication or not. This issue (how to limit this undue “sovereign” market power), certainly, is
a complicated matter for a single country to deal with (and certainly is different from market
power measured by Lerner Index). What are the proper international institutions that could deal
with these cases? Or could a single country, isolated from all others, deal with this issue alone,
by itself? These questions are worth to be made in international forums. What seems clear is that
anticompetitive acts of this kind should not be shielded or imposed this way, in a democratic
international society, without a huge debate about it.
9.2.
Antidumping and predatory pricing?
Another area that deserves more debate in International Relations is Antidumping: this
policy, accepted by several countries around the world, aims (or is tailored) to avoid international
predatory pricing, using a thumb rule of “normal price”. If exports prices are lower than internal
prices, then, according to antidumping rule it represents an illicit act.
However, in Antitrust Law, to infer what is predatory price, generally, a thumb rule is not
enough to condemn the conduct. Export markets can have different rivalry levels or different
market structures that would allow a healthy international price discrimination, without
configuring a predatory international practice. In Antitrust jurisprudence, predatory pricing is
extremely rare (because it is hard to occur and because mistakes in this area are too costly for
society). So, it appears that the use of such antidumping “thumb rule” that forbids international
price discrimination, rather than just saving decision costs, can, sometimes, works as
patrimonialist-protectionist tool, if it does not discriminate properly certain specific situations
(such as the improper use of subsidies that distort measures of costs and are made with solely
predatory purposes). In this area, International Politics seems a little bit far from Antitrust
Principles and strict screening rules of what is considered anticompetitive practices, that wants to
limit only undue market power. And if there are so much antidumping measures around the world,
why there are so few predatory price cases condemned? It certainly is a good subject for reflection.
9.3.
Nuclear technology and refusal to deal?
Violence and coercion can sustain market power, internally and internationally. The very
framework in which the United Nations (UN) itself is established, ultimately, is a structure forged
by force and not by rational or Democratic principles. United Nations Security Council only
grants veto power to those nations that assumedly have force, nuclear weapons and dual
technology. All the other States that do not possess such weapons, according to UN rules, do not
deserve the same treatment and are subject to “nuclear non-proliferation” regimes. Some countries
(inferior ones) ought to disarm themselves while others (superior ones) claim to have special
powers derived from their nuclear power.
The problem with that (besides, of course, International Democracy and the right for selfprotection) is the fact that non-belligerent use of nuclear technology [and research on this field]
maybe subject to unilateral anticompetitive practices.
For example, Brazil bought Angra 1 (Brazilian Nuclear Power Plant) from a NorthAmerican enterprise called Westinghouse. This enterprise had a huge market and bargaining
power that was able to impose to Brazil a contract without stipulating any penal clause for
delaying the construction of Angra 1, refusing to transmit any nuclear technology to Brazil and
even refusing to supply Angra 1 with enriched uranium (that was necessary to make Agra I work,
as expected). This Power Plant was ironically named as a “nuclear firefly”30, because was not able
to produce nuclear energy at all. Although the initial price of the Westinghouse plant was 330
million dollars, the total cost of the project to Brazil was increased to US$ 2.2 billion (OLIVEIRA,
1999), what clearly shows the charge of an overprice.
When Brazil tried to negotiate with German [Kraftwerk Union A. G. (KWU)] and French
[Framatome] competitors of Westinghouse, the construction of another nuclear power plant (with
a better contract and with technology transfer), North American government itself intervened,
raising several commercial sanctions against Brazil and some other countries, and made extremely
difficult such negotiations. (BANDEIRA, Estado nacional e política internacional na América
Latina: o continente nas relações Argentina-Brasil (1932-1992), 1995) (BANDEIRA, O “Milagre
Alemão” e o Desenvolvimento do Brasil - As Relações da Alemanha com o Brasil e a América
Latina (1949-1994), 1995). New York Times, from the June 13, 1975 edition, called GermanBrazilian agreement as a “nuclear madness”. John Pastore, US Senator and Chairman of Joint
Committee on Atomic Energy insisted on the Agreement's nullification and suggested a
reconsideration of the U.S.'s NATO commitments to Germany to demonstrate the U.S.'s stance
on nonproliferation.31
“it was the largest and, at the time, the most expensive transfer of advanced technology to a developing country; it was the first breach
of the U.S. monopoly over the world export market for nuclear reactors by a non-American vendor” (KOLLMANN,
2012)
For protecting these monopolistic interest, in March 1977, Jimmy Carter took measures
against both Brazil and Germany: he pressed two American banks, Chase Manhattan Bank and
Eximbank to suspend all financing activities negotiated with Brazil, and halted the supply of
enriched uranium to Germany. (BANDEIRA, Estado nacional e política internacional na América
Latina: o continente nas relações Argentina-Brasil (1932-1992), 1995) (BANDEIRA, O “Milagre
Alemão” e o Desenvolvimento do Brasil - As Relações da Alemanha com o Brasil e a América
Latina (1949-1994), 1995).
Again, the case of Brazil-US relationship is mentioned here, but there are a lot of others
kinds of relations involving States and Antitrust-nuclear matters. In addition, the Department of
Justice and the Federal Trade Commission of US play an extremely important role, being truly
world beacons of how Antitrust is better applied. However, this other face of international trade
30
AGOSTINHO,Victor Angra 1 testa sistema de emergência. Após 20 meses, Furnas tenta colocar usina nuclear de Angra dos Reis
em funcionamento Folha de São Paulo. 24/11/94. Seção Cotidiano. p.3.
31
http://www.coldwar.hu/html/en/conferences/Third_2012/Papers/Conference_paper_-_Thomas_Kollmann.pdf
(and refusal to deals) should be understood from a wider and bigger perspective, especially if
governments with specific political interests endorse “market power”.
This is just one example from the past of what can happen in this area. Nonetheless,
nowadays, there is still a lot interference in this realm.
9.4.
International governmental boycotts
Without entering in a deep analysis of the controversial Webb-Pomerene Act, it is possible
to focus the analysis on some recent issues regarding international security matters. About this
issue, International Traffic in Arms Regulations – ITAR interfered in some commercial Brazilian
transactions. For example, in 2006, Embraer of Brazil was prevented from selling Super Tucano
aircraft to Venezuela by the U.S. Authorities (CORREA, 2012). Recent cooperation between
CBERS - China-Brazil Earth Resources Satellite also suffered certain difficulties imposed by US
Bureau of Industry and Security [hereinafter BIS] (GEIER & DERRIEN)32. How Brazilian
technology, Brazilian exports and market power of Brazilian industry are affected by those
decisions?
BIS implements export controls for the Department of Commerce through the Export
Administration Regulations (EAR). This legislation possesses boycott provisions, with specific
sanctions for those who disobey such boycotts [15 C.F.R. Part 760 (“Restrictive Trade Practices
or Boycotts”)]. These sanctions include requirements that the U.S. person provide information
about business relationships with a boycotted country or refuse to do business with persons on
certain boycott lists.
Cartel or boycotts, when organized by enterprises represent antitrust illicit conducts.
However, what if boycotts are implemented by sovereign states? According to records of BIS3334
, some boycotts are widely recognized and implemented observing EAR provisions. For
example:

Between November 2001 and July 2006, Buehler Limited of Lake Bluff, Illinois, made 80 exports of a
product called “Coolmet,” a mixture containing triethanolamine (TEA) that is used as a lubricant with cutting
tools, to various destinations including Brazil, without the required U.S. government authorization. Buehler
Limited agreed to pay a $200,000 civil penalty. Indeed, triethanolamine is listed on Schedule 3, part B of
Chemical Weapons Convention, because is raw material for mustard gas. However, it is also raw material for
several other commercial applications, such as shampoos, hygiene and cosmetic products, shaving foams and
several other commercial uses.

In 2004 and 2005, Cabela’s Incorporated, an outdoor equipment outfitter based in Sidney, Nebraska, exported
optical sighting devices for firearms to various destinations including Brazil, without the required U.S.
government license. On November 7, 2008, Cabela’s agreed to pay a $680,000 civil penalty.
Between 2000 and 2002, E.D. Bullard, of Kentucky, exported and re-exported thermal imaging cameras to
various destinations including Brazil, without the required U.S. government license. E.D. Bullard agreed to
pay a $330,000 administrative penalty in June 2005.


32See
Engineering Dynamics Inc., a Louisiana company that produced software to design offshore oil and gas
structures, exported and attempted to export software to Iran through a co-conspirator in Brazil without
having first obtained the required authorization from BIS and the U.S. Department of Treasury’s Office of
Foreign Assets Control. Nelson Galgoul, director of the Brazilian engineering company, Suporte, acted as an
agent for Engineering Dynamics Inc. in the marketing and support of this software and trained users of the
software in Iran. Each defendant was fined $250,000, and ordered to forfeit $218,583. On May 22, 2008,
Galgoul was sentenced to 13 months in prison, three years of supervised release, a $100,000 criminal fine,
and a $109,291 forfeiture for his part in the conspiracy. In April 2008, Engineering Dynamics, Inc. agreed to
pay a civil penalty of $132,791.39. In addition to the civil penalty paid to BIS, Engineering Dynamics Inc.
paid $132,791.39 to OFAC.
http://www.ieav.cta.br/peice/30_10/30_10_6.pdf
See https://www.bis.doc.gov/index.php/forms-documents/doc_view/535-don-t-let-this-happen-to-you-2010
34 See http://efoia.bis.doc.gov/index.php/component/docman/doc_view/436-e2044?Itemid=
33
In sum, enterprise cooperation and the exports of chemical materials or flow of technology
of optical sighting devices; thermal imaging cameras; and softwares to design offshore oil and
gas structures, in international scenario, can still be subject to discretionary understanding of some
governments of what is possible to trade or not.
This option can enhance market power of certain firms about such technologies, depriving
others enterprises and even, in some cases, other sovereign entities to access them (which is a
kind of refusal to deal, a typical unilateral Antitrust conduct, but with government blessing).
Certainly, non-proliferation policies can be made with good and peaceful purposes to maintain
balance of power, in order to avoid global war or to contain terrorism. Nevertheless, there is, at
least, a moderate risk that such interpretations and the implementation of such regimes may
interfere in purely commercial activities, reinforcing undue international market power.
Sometimes, the access to technology used by military forces can have important spillover
effects in several other areas. Internet itself (and Arpanet) are examples of how what begins as
dual technology at first sight can help to foster new researches in other social and purely
commercial areas. That is why boycotts (and even sovereign boycotts) related to some
technologies can have some harmful effects of preventing this kind of effects, depending on some
circumstances.
9.5.
Pharmaceuticals: “free” riding or “free” trade?
Another aspect that governments can interfere and protect market power is to force
certain agreements. For example, in pharmaceutical industry, there is a debate about how State´s
should deal with “data exclusivity” and “data protection” of “New Chemical Entities”.
According to Palmedo:
“ Data exclusivity is a form of intellectual property (IP) protection required by many trade agreements that protects the data submitted
by pharmaceutical and chemical firms to regulatory authorities for marketing approval. The pharmaceutical industry argues it is an
important form of IP protection needed to support further investment in new medicines. However, advocates for access to medicines
have warned that it delays generic competition and therefore raises prices of drugs. When a branded firm applies for marketing approval
from a regulator like the U.S. Food and Drug Administration, it must present clinical test data to regulators showing that the drug is safe
and effective. The process of obtaining the clinical trial data is very expensive, risky, and time consuming. Studies of the cost of drug
development find that the majority of the cost of bringing a new drug to market is conducting the clinical trials.
If generic firms needed to reproduce the same data, the high expenses would require them to charge prices nearly as high as the innovator.
Additionally, it would violate international ethical standards that prohibit doctors from conducting experiments on humans in situations
where the results are already known. Therefore, most nations’ regulatory authorities grant generic firms marketing approval based on
bioequivalence data, which shows that their products are chemically the same, and are absorbed into the body the same way, as the
originators’ products. Generic firms rely on the clinical trials conducted by the originator firms for proof their products’ safety and
efficacy.
Data exclusivity establishes a period of time (generally 5-10 years for small-molecule drugs) during which generic firms cannot win
marketing approval based on the clinical data submitted by brand name producers. It can keep generics off the market even in cases
where there is no patent in place, or when a compulsory license has been issued.” (PALMEDO, 2013)
However, data exclusivity is not the same thing as data protection. According to TRIPS
Article 39.3, countries [that signed TRIPS] should protect such data against disclosure, except
where necessary to protect the public or unless steps are taken to ensure that the data are protected
against unfair commercial use. There is nothing on TRIPS demanding countries to grant
exclusivity (and market power) to originators firms. Indeed, United States tried to include on
TRIPS data exclusivity obligation, but all other countries did not agree with such understanding.35
“It must be admitted that the following of Article 39.3 does not, from a prima facie reading, appear to impose data exclusivity
during a certain period of time. This lack of clarity is the obvious result of a difficult negotiation process where divergences of views
arose between developing and industrialized countries as to the necessity of EC/U.S. like type of data protection as well as among
industrialized countries on the length of the data exclusivity period” (EU, 2001, p. 3). Protection Of Data Submitted For The
35
The thesis to implement “data exclusivity”, forcing other firms that want to enter in
market (1) to repeat clinical trials or (2) to respect this market power can have serious social
consequences, in the sense that usually the efficacy of medicine is made by difference in
difference analysis, where the treated group will receive the medicine, while the control group
could receive a palliative treatment or even placebos. Deaths or worsening the health of control
group will prove the efficacy of something that is already known by enterprises. The worsening
the health of control group is something desired and needed to “prove” the efficacy of a drug. It
is a very dark way to protect “investments” for certain molecules that are not subject to patent law
protection: or the health of some group of people get worst or the investment of incumbent firm
(with first mover advantage) is not recouped “against free riders”.
This thesis was clearly refuted on TRIPS´s negotiation. However, when United States
tried to negotiate “free trade” agreement with some countries, this clause (about data exclusivity)
were mentioned as a precondition for signing the international treaty of “free trade”.
Millions of people lack access to affordable medicines. The intellectual property rules in the Central America Free Trade Agreement
(CAFTA) provide pharmaceutical companies with monopoly protections that allow them to market some drugs without competition by
less costly generics (…)
Our study suggests that CAFTA’s intellectual property rules on data exclusivity and patents are responsible for the removal of several
lower-cost generic drugs from the market in Guatemala and for the denial of entry to a number of others. (…). The U.S. Congress
removed the data-exclusivity provision of the trade agreement with Peru in May 2007, recognizing potentially negative consequences
for lower-income countries. As an initial step, the U.S. Trade Representative and the Department of Commerce should extend this
recognition to all CAFTA countries. It should assert its intention not to implement the data-exclusivity provisions of CAFTA and should
proactively cooperate with Central American governments that take action not to implement these provisions. A more limited reform
proposal, generated prior to the Peru policy, would require generic producers to pay a small sum for the use of pharmaceutical test data,
rather than the present system of establishing exclusive rights to protect investments in the data. (SHAFFER & BRENNER)
Nonetheless, according to Section 301 do US Trade Act of 1974, United States can impose
commercial sanctions to other States that “do not respect Intellectual Property”, putting them on
a black list or Priority Watch List. There were some movements to force Brazil to adopt “data
exclusivity protection”:
301 Report of 2012 - “The United States encourages Brazil to clarify and strengthen its system for protecting against unfair commercial
use, as well as unauthorized disclosure of test and other data generated to obtain marketing approval for pharmaceutical products. (…)
The United States looks forward to continuing to work with Brazil to address these and other matters”36
It is important to clarify that Brazil do not disclosure any test result that is registered in
Anvisa for approval of Generic version of medicines. Brazil, simply, do not recognize data
exclusivity (and market power) to medicines that do not possess patents or other intellectual
recognized Brazilian property right´s Law.
CADE also is analyzing a case to know if some enterprises had abused their “intellectual
property rights”, claiming, in Judiciary, that (i) their data was unduly disclosed, when, Anvisa
claim it did not disclose anything (ii), in order to recoup their investment, they need data
exclusivity. However, differently from Europe or United States, since Brazil did not recognize
such monopoly at all (data exclusivity), some enterprises claim that data exclusivity is not limited
on time. Therefore, they can simply wipe out all the generics from the market forever, since no
medicine could legally rely on their research without originator´s authorization to prove safety of
the medication, independently if it is a “New Chemical Entity” and independently from any other
variable. If successful this strategy, it would be possible to forestall Generics´s Brazilian policy.
CADE did not judge such case yet [Administrative Proceedings No 08012.006377/2010-25;
Registration Of Pharmaceuticals: Implementing The Standards
http://apps.who.int/medicinedocs/en/d/Jh3009ae/12.html
36 See http://www.ustr.gov/sites/default/files/2012%20Special%20301%20Report.pdf
Of
The
Trips
Agreement.
2002.
According
to
Plaintiff: Pró Genéricos; Defendants: H Lundbeck A/S, and Lundbeck Brasil Ltda; Conduct under
investigation: Abuse of data protection rights regarding Lexapro (antidepressant) to prevent
generic entry]. Independently of the outcome of this case, certainly, it is a manner to create market
power through a specific kind of arguments that were clearly not accepted in International realm.
9.6.
Telecommunications and mergers
Recently, Brazilian and German Presidents were spied by other countries.37 Such fragility
in national telecommunications systems can be fostered by mergers or anticompetitive conducts,
when foreign countries dominate important telecommunication structures.
In this regard, it is worth mentioning, for example, the case of Huawei Technologies, which
did not have their approved mergers in United States38 and it was barred from participating in
bidding for broadband provision in Australia39, due to national defense interests.
It is possible that there are other interests at stake, rather than simple “protectionism”, when
a governmental authority decides to block a merger. Blocking a merger can occur based on
nationality of the enterprise (and based non-economic reasons) in order to avoid foreign countries
to dominate important telecommunication structures.
9.7.
ICN and State-created dominant enterprises
In this scenario, it is possible to see that there are sovereign cartels; sovereign refusal to
deals; sovereign boycotts; sovereign espionages (and unfair international competition derived
from such acts) and the use of thumb rules to infer predatory pricing that unduly close
international markets.
Therefore, sometimes, nationality of enterprises matters in Antitrust. Sometimes, the use
of national enterprises could represent a strategy to escape from overpricing and international
wrongful acts that unduly create market power. For example, Brazil created Nuclebrás in order to
escape nuclear technology dependence over international monopolies. There may be reasons to
take in consideration nationality of enterprises when antitrust analysis is made.
However, on recommended practices of ICN in relation to “State-Created Monopolies”,
ICN assert that Antitrust Authorities should treat all enterprises as “private undertaking by using
standard antitrust analysis to assess dominance/substantial marker power, regardless of state
ownership or legal status of the firm”. Antitrust Authorities should “advocate for an expeditious
liberalization of barriers to entry in markets with state-created dominant enterprises”.
37
See http://www.dw.com/en/stung-by-the-nsas-reach-brazil-and-germany-prepare-for-closer-ties/a-18648528
Conforme site http://www.enterprisenetworkingplanet.com/netsysm/u.s.-blocks-huawei-from-government-networks.html,
verificado em 31 de julho de 2013. Conforme site http://www.washingtontimes.com/news/2007/oct/3/merger-opens-us-defense-tochina/?page=all, verificado em 31 de julho de 2013. “ American companies and its government should avoid doing business with China's two
38
leading technology firms, Huawei and ZTE, because they pose a national security threat to the US, the House of Representatives' intelligence
committee will warn in a report to be published on Monday. The Republican-controlled panel also says US regulators should block mergers and
acquisitions in the US by the two companies, which are among the world's leading suppliers of telecommunications gear and mobile phones. The
panel's report will cause transatlantic friction over the role of the Chinese companies. In the UK, Huawei is a major supplier to the telecoms provider
BT, and has supplied infrastructure being used in the new 4G superfast mobile network built by Everything Everywhere – the merged Orange/TMobile. Huawei provides access to its source code for GCHQ specialists who have reportedly examined it for threats and passed it as safe for use.
Huawei is a private company founded by a former Chinese military engineer, and has grown rapidly to become the world's second largest supplier,
behind Sweden's Ericsson, of telecommunications network gear, with operations in more than 140 countries. ZTE is the world's fourth largest mobile
phone manufacturer, with 90,000 employees worldwide, and fifth-largest maker of telecoms equipment. While both companies' sales of mobile
devices such as smartphones have grown in the US, espionage fears have proscribed any move into network infrastructure sales. ZTE has also enjoyed
growth in its sale of mobile devices, although in recent months it has faced allegations about banned sales of US-sanctioned computer equipment to
Iran. The FBI is probing reports that the company obstructed a US Commerce Department investigation into the sales. (…) Reflecting growing US
governmental and commercial concern over cyber-attacks traced to China, the report also recommends that US government computer systems not
include any components from the two firms because that could pose an espionage risk.” (http://www.theguardian.com/technology/2012/oct/08/chinahuawei-zte-security-threat)
39
Conforme site http://www.reuters.com/article/2012/03/26/us-australia-huawei-nbn-idUSBRE82P0GA20120326 verificado em 31
de julho de 2013.
Certainly, “in vacuum”, these principles can make sense. Nonetheless, since “sovereign”
market power and “sovereign” anticompetitive practices are different from traditional market
power (more difficult to control or adjudicate) and immersed in a multiple dimensional tradeoff
world [where nationality matters], maybe dogmatic ways to determine what is the rightful
antitrust strategy towards state ownership or enterprise´s nationality deserve a little bit flexibility,
rather than a traditional, straightforward and narrow guideline in this regard.
10.
Privatization of competence or minimalist approach?
As already mentioned, given that there are a lot of subjective roads that lead to market
power definition, defining competence [in other word, defining who has the legal authority to
choose what is the correct identification strategy of market power] is a huge issue in Antitrust
Law.
The definition of what constitutes anticompetitive practices can be made by public
Authorities or by private parties through arbitration. Two approaches are recognized when there
is an intersection between antitrust and arbitration. According to OECD:

“Under the maximalist approach, national courts are required to carry out an in depth review of arbitral awards when they
are challenged, or when enforcement is required. This involves a full review of the entire case and all the evidence associated
with it. The rationale for this approach is to avoid the risk that arbitration will be used to circumvent competition law. Courts
can therefore consider in detail whether competition law has been applied ‘correctly’.

Under the minimalist approach no special treatment is given for awards raising competition law issues, and emphasis
remains on taking the case outside of the courts, and settling it via arbitrators. The rationale for this approach is that if a full
review of the award is carried out, this arguably defeats the purpose of going into arbitration in the first place and
undermines the trust afforded to arbitrators and the institution of arbitration. Courts should therefore only overturn awards
where there is a fundamental breach of public policy.” (OECD, 2011, p. 14)
Both views can have excesses (and the mere choice of one or another theory, as a binary
decision can perhaps oversimplify the actual underlying social concerns of individual cases).
On one hand, in the maximalist theory, in which the state revises any and all acts and
arbitration awards, can improperly reward a party that deliberately includes a "competitive issue"
in middle of a contract together with an arbitration clause, so if the arbitration award is
unfavorable to such losing party, it would be possible to use “antitrust discussion” as an escape
valve to invalidate arbitration award.
On the other hand, the minimalist view can also pose social problems, especially when the
interpreter or judge is less likely to consider material matters, being more formalistic (seeking,
only unburden the judiciary from its backlog and preserve the decision of the Arbitration Court
no matter what, even if eventually this decision will affect the price of goods consumed by several
people who did not have the chance to participate in the arbitration agreement entered between
private parties).
This dualistic way of seeing arbitration may deserve a third theory, that is worried with
both kinds of excesses and some other institutional design (ID).
ID = { ID 1, ID 2, ID 3, … ID n}
Equation 27- Different institutional designs
For example, (i) Mitsubishi Motors v. Soler Chrysler-Plymouth, (1983, US First Circuit
and 1985 US Supreme Court) seems to be a very different approach from French cases such as
(ii) Thales v. Euromissile, although both decisions are all classified as belonging to the
“minimalist” approach.
In the first case (Mitsubishi Motors v. Soler Chrysler-Plymouth) US First Circuit of Appeals
dealt with a vertical restraint, which is not an obvious antitrust violation. This Court understood
that “the mere appearance of an antitrust dispute does not alone warrant invalidation of the
selected forum on the undemonstrated assumption that the arbitration clause is tainted”.
However, US First Circuit of Appeals allowed district court to analyze:
(i) “whether the matters are sufficiently separable to justify separate and
contemporaneous treatment;
(ii) Whether the permeation of antitrust issues halted or not arbitrability, in case of not
being possible to segregate private and antitrust issues.
As district court did not have the opportunity to decide such matters, the case was remanded
in these regards. It is not a simple delegation to private parties to estimate if public law was
breached or if there was public harm derived from such private contracts. There is a specific
methodology to observe in order to avoid minimalist excesses, while granting deference to the
choice of forum of private parties involved in a contract.
A very different approach seems to appear in Thales v. Euromissile. According to
Cuniberti:
In November 2004, the Paris Court of Appeal had ruled in Thales Air Defense v. GIE Euromissiles that there was such a procedural rule in
France. The French rule was that only violations of French public policy which were “obvious, actual and concrete” (flagrante, effective
et concrete) would be sanctioned. As a consequence, in Thalès, the Court had dismissed a challenge in a case where the parties had
arguably shared the relevant European market. The issue of the validity of the contract had not been raised during the arbitration.
(CUNIBERTI, 2008)
If that is correct, in such case, one party confessed that the contract, which included
arbitration clause was the result of a cartel agreement, with the relevant market division. Also,
according to Emmanuel Gallard, Paris Court of Appeals preferred to grant deference to arbitral
award without issuing an “economic analysis” of the practice, in order to decide if there was a
breach of an Antitrust obligation or not. (GAILLARD, 2007)
Therefore, the Paris Court of Appeals chose to adopt a highly formalistic understanding,
mentioning that argument of cartel agreement was not raised during arbitration. Ergo, there was
no "flagrant" violation of competition rules, even though one party clearly confessed that agreed
with a very long non-competition clause in certain regions of Europe (and that, in fact, such
agreement was a fruit of a cartel that divided European market). If such a claim is true and if Paris
Court of Appeals understands that it does not mean flagrant violation of competition rules, hardly
other conducts would fulfill such standard. (WALLE, 2013, p. 215).
Undue market power or a cartel can be maintained and even legally enforced if a dispute
involving such issue is settle by arbitration [and such issue is not even raised during arbitration].
11.
Conclusion
To be or not to be a dominant player?
Trying to fairly deal with this question, the answer (without the Shakespearean´s
eloquence), as expected, is “well, it depends on a lot of variables”, such as the ones on this list:
Variables that might influence if there is dominance
Notation
Type of conduct (or merger)
Players (who are involved in conducts or mergers)
What are the circumstances of the practice
Hermeneutic options regarding methodologies
Amount and quality of interpreter´s available information
C={C1, C2, C3, … Cn}
P={P1, P2, P3, …., Pn}
X={X1, X2, X3, … Xn}
HE={HE1, HE2, HE3, …, HEN}
∑ 𝐼𝑛𝑓 = {0, … ,1} & Q= {0, … ,1}
Strategy of identification
How burden of proof should be established or distributed
Timing (and dynamic aspects) of the analysis
Means of coercion
Interpreter´s teleological choice
Sovereign nature of market power
Who is the interpreter (what is the institutional jurisdictional design)
SI={SI1, SI2, SI3….SIn}
W={W1, W2, W3, …, WN}
t={t1, t2, t3, … tn}
MC={MC1, MC2, MC3,… , MCn}
TL={TL1, TL2, TL3, …, TLN}
SM={0,1}
ID={ ID 1, ID 2, ID 3, … ID n}
Certainly, these are just some of possible influences about what is important to measure
dominance (being possible that several other variables that were not properly mentioned in this
paper can also contribute for this debate).
Some risk would have to be taken, in order to define this issue properly. Maybe nonclassical logic can help to understand what can of risks are at stake. Indeed, to draw a definition
of market power, it is extremely difficult to make in abstract terms, without knowing, beforehand,
who is making an anticompetitive conduct, who is the victim and what are the conditions (or
circumstances in which the practice were perpetrated). This circumstantial information not only
is needed, but it is also needed with detailed precision. Also:

Several variables are intervals of intervals, including critical values (if robustness tests are
made). Therefore, to estimate the degree of uncertainty of every single empirical estimation
it is highly recommended that much attention is paid to convergence processes. Also,
standard deviations should be reported;

Traditional econometric tests are not enough to deal with some problems related with
robustness indeterminacy. Indeed, burden of proof should not be limited to classical
aristotelic questions to define who has the right to possess null hypothesis, or what is the
correct statistical significance or what is the correct power of the test to be used. When
critical values are understood as an interval of intervals, robustly speaking, one can clearly
see that a new methodology are needed to deal with situations when measured values are
found exactly in the middle of critical intervals. In this kind of situation, the recourse of
three or multiple valued logic approach or some other approach of this kind may be helpful.

When control groups are used, it is important that Authorities pay attention to several
issues, especially if stable unit treatment value assumption is violated or not (or if there is
some other source of bias). If the control groups are not used, then, it is highly
recommended to think if strategy of identification of market power is correctly delineated
and if control of several issues (including endogeneity and inclusion of all “important”
variables) is observed in order to create a valid counterfactual analysis.
So, when critics are made in regard of use of market shares as a first approximation of
market power, normally, these topics are not raised, showing some limitations of the alternative
approaches. Certainly, these alternative approaches can be very sophisticated. Nonetheless,
uncertainty is not eliminated by their use, especially when interpreter is conscious that there are
a lot of subjective options in a continuing array of choice, and his responsibility (his responsibility
alone) to find the exact point where such continuous interval is divided in only two parts. There
are many ways to make such choice.
Further, it is important to think in alternatives, for the sake of international distributive
justice concerns, how “sovereign undue market power” can be limited, in order to protect those
who suffer effects of sovereign international deadweight loss .
12.
Bibliography
ABADIE, A. (January de 2005). Semiparametric Difference-in-Difference Estimator. The Review of Economic Studies,
72(1), 1-19.
ABADIE, A., DIAMOND, A., & HAINMULLER, J. (2010). Synthetic Control Methods for Comparative Case Studies:
Estimating the Effect of California’s Tobacco Control Program. Journal of American Statistical Association, 493505. Acesso em 23 de November de 2015, disponível em http://www.hks.harvard.edu/fs/aabadie/ccsp.pdf
ANDRADE, M. V., GAMA, M. M., RUIZ, R. M., MAIA, A. C., MODENESI, B., & TIBURCIO, D. M. (2010).
Estrutura do mercado de saúde suplementar no Brasil. Texto de discussão 400, 21, 1-66. Acesso em 15 de October
de 2015, disponível em http://www.cedeplar.face.ufmg.br/pesquisas/td/TD%20400.pdf
AREEDA, P. E., & HOVENKAMP, H. (2006). Antitrust law : an analysis of antitrust principles and their application.
New York: Aspen Publishers.
ASHLEY, R. (Spring de 1984). The poverty of neorealism. International Organization, 38(2), 225-286.
ASHLEY, R. K., & WALKER, R. B. (September de 1990). Introduction: Speaking the Language of Exile: Dissident
Thought in International Studies. International Studies Quarterly, 34(3), 259-268.
ATHEY, S., & IMBENS, G. W. (March de 2006). Identification and inference in nonlinear difference-in-differences
models. Econometrica, 74 (2), 431–497.
BANDEIRA, L. A. (1995). Estado nacional e política internacional na América Latina: o continente nas relações
Argentina-Brasil (1932-1992). São Paulo: Ensaio.
BANDEIRA, L. A. (1995). O “Milagre Alemão” e o Desenvolvimento do Brasil - As Relações da Alemanha com o
Brasil e a América Latina (1949-1994). São Paulo: Ensaio.
BANKS, J., BLUNDELL, R., & LEWBEL, A. (1997). Quadratic Engel Curves and Consumer Demand . The Review
of Economics and Statistics, Vol. 79, No. 4., 527-539.
BELAISCH, A. (2003). Do brazilian banks compete? IMF Working Paper - WP 03/113. Acesso em 11 de October de
2015,
disponível
em
http://lnweb90.worldbank.org/CAW/Cawdoclib.nsf/0/92ABEA5695F7322D85256E1D00658004/$file/wp03113.
pdf
BENTHAM, J. (1781). Principles of Morals and Legislation. Kitchener: Batoche Books. Acesso em 20 de Setembro
de 2015, disponível em http://socserv.mcmaster.ca/econ/ugcm/3ll3/bentham/morals.pdf
BERTRAND, M., DUFLO, E., & MULLAINATHAN, S. (February de 2004). How Much Should We Trust
Differences-in-Differences Estimates? The Quarterly Journal of Economics, 119(1), 249-275.
BIKKER, J., SHAFFER, S., & SPIERDIJK, L. (2009). Assessing Competition with Panzar-Rosse Model: The role of
scale, costs and equilibrium. DNB Working Paper, 1-40.
BOONE, J. (August de 2008). The Economic Journal. 118 (531), 1245-1261.
BOURDIEU, P. (1998). O poder simbólico. Rio de Janeiro: Bertrand Brasil.
BRESNAHAN, T. F. (1982). The oligopoly solution concept is defined. Economic Letter, 10, 87-92.
BRUERS, S. (Outubro de 2010). A quantitative model for a theory of justice. Acesso em 20 de Setembro de 2015,
disponível em Stijn Bruers, the rational ethicist: https://stijnbruers.files.wordpress.com/2010/10/a-quantitativemodel-for-a-theory-of-justice-part-i2.pdf
BUCCIROSSI, P. (April de 2000). Scelte di policy e definizione del mercato rilevante: un modelo strategico. Temi e
problemi, 10, 1-54. Acesso em 14 de October de 2015, disponível em http://www.agcm.it/traspstatistiche/doc_download/2642-tp10.html
CAPPS, C., DRANOVE, D., & SATTERHWAITE, M. (2003). Competition and market power in option demand
markets. RAND Journal of Economics, Vol. 34, No. 4, 737–763.
CARMAN, H. F., & SEXTON, R. J. (2005). Supermarket Fluid Milk pricing practices in Western United States.
Agribusiness, 509-530.
CARPENTER, C. S., & STEHR, M. (May de 2011). Intended and Unintended Consequences of Youth Bicycle Helmet
Laws. Journal of Law and Economics, Vol. 54, No. 2, 305-324.
CARR, B. (2007). Universe or multiverse? Cambridge: Cambridge University Press.
CERULLI, G. (2014). ivtreatreg: A command for fitting binary treatment models with heterogeneous response to
treatment and unobservable selection. Stata Journal, 14, 453–480.
CERULLI, G. (2014). treatrew: A user-written command for estimating average treatment effects by reweighting on
the propensity score. Stata Journal, 14, 541–561.
COATE, M. B., & WILLIAMS, M. D. (2005). Generalized Critical Loss for Market Definition. Potomac Law and
Economics Working Paper No. 05-01, 1-23. doi:http://dx.doi.org/10.2139/ssrn.669146
COHEN, G. A. (1988). History, Labour, and Freedom: Themes from Marx. Oxford: Oxford University Press.
COMPTE, O., JENNY, F., & REY, P. (2002). Capacity Constraints, Mergers and Collusion. European Economic
Review, 1-29.
CONLEY, D. (2008). You may ask yourself: An introduction to thinking like a sociologist. Nova Iorque: W.W. Norton
and Company.
CORREA, F. (2012). Caças de Combate Aéreo para o Brasil: o que a história tem a dizer. Defesanet. Acesso em 3 de
December de 2015, disponível em http://www.defesanet.com.br/fx2/noticia/8952/DEFESA-EM-DEBATE--Cacas-de-Combate-Aereo-para-o-Brasil---o-que-a-Historia-tem-a-dizer/
CORTS, K. S. (1999). Conduct parameters and the measurement of market power. Journal of Econometrics, 88, 227250.
COX, R. (1981). Social Forces, States and World Orders: Beyond International Relations Theory. Millennium, 10(2),
126-155.
CUNIBERTI, G. (2008). Arbitral Awards Violating European Antitrust Laws: French Courts Cannot Help.
conflictoflaws.net. Fonte: http://conflictoflaws.net/2008/arbitral-awards-violating-european-antitrust-laws-frenchcourts-cannot-help/
DEATON, A., & MUELLBAUER, J. (1980). An Almost Ideal Demand System. American Economic Review, 70 , 312326.
DEATON, A., & MUELLBAUER, J. (1981). Functional forms for labor supply and commodity demands wíth and
without quantity restrictions . Econometrica, 49, 1521-1532.
DeSOUZA, S. (Maio de 2009). Análise Empírica de Modelos Agregados de Demanda. Working Paper, CAEN, UFC.
Acesso em 30 de Abril de 2015, disponível em CAEN - UFC: http://www.caen.ufc.br/attachments/article/37/seet05.pdf
DICK, J. (1975). How to Justify a Distribution of Earnings. Philosophy and Public Affairs.
DONALD, S., & LANG, K. (2007). Inference with difference-in-difference and other panel data. The Review of
Economics and Statistics, 89(2), 221–233.
DOWNEY, A. B. (2012). Think Bayes. Needham: Green Tea Press.
DWORKIN, R. (1980). Is Wealth a Value? Journal of Legal Studies, 9.
DWORKIN, R. (2000). Sovereign Virtue: Equality in Theory and Practice. Cambridge: Harvard University Press.
DWORKING, R. (March de 1980). Is Wealth a Value? The Journal of Legal Studies, 9 (2), 191-226. Acesso em 8 de
December de 2015, disponível em https://blogs.law.harvard.edu/hltf/files/2010/10/Dworkin.pdf
ECKBO, B. (1983). Horizontal mergers, collusion and stockholder wealth. Journal of Financial Economics 11, 241273.
EFRON, B., & TIBISHIRANI, R. (1994). An Introduction to the Bootstrap, Monograph in Applied Statistics. New
York: Chapman and Hall. Acesso em 23 de November de 2015, disponível em
http://www.hms.harvard.edu/bss/neuro/bornlab/nb204/statistics/bootstrap.pdf
EINSTEIN, A. (1905). Zur Elektrodynamik bewegter Körper. Annalen der Physik 17, 891–921.
ELHAUGE, E., & GERADIN, D. (2011). Global Competition Law and Economics. Oxford: Hart Publishing.
ELZINGA, K. G., & HOGARTY, T. F. (1973). The Problem of Geographical Market Delineation in Antimerger Suits.
Antitrust Bull. 45, 18.
ELZINGA, K. L., & HOGARTY, T. (1978). The Problem of Geographical Market Delineation Revisited: The Case of
Coal. Antitrust, 1.
EPSTEIN, R. J., & RUBINFELD, D. L. (20 de Agosto de 2003). Merger Simulation with Brand-Level Margin:
Extending PCAIDS with Nests" UC Berkeley Competition Policy Center Working Paper No. CPC03-40.
http://ssrn.com/abstract=495568. Acesso em 30 de Abril de 2015, disponível em SSRN:
http://ssrn.com/abstract=495568
EVERETT, H. (1957). "Relative State" Formulation of Quantum Mechanics. Rev.Mod.Phys., 454.
FAVOR, C., & LAMONT, J. (2013). Distributive Justice. Stanford Encyclopedia of Philosophy. Acesso em 23 de
August de 2015, disponível em http://plato.stanford.edu/entries/justice-distributive/
FIUZA, E. (2008). Relevant market delineation and horizontal merger simulation: an unnified approach. IPEA Working
Paper,
1-35.
Acesso
em
14
de
October
de
2015,
disponível
em
http://www.fea.usp.br/feaecon/incs/download.php?i=25&file=../media/livros/file_25.pdf
FORNI, M. (2004). Using stationarity tests in antitrust market definition. American Law and Economics Review 6(2),
2004, p. 441-464. GENESOVE, D.. “Comment on Forni’s ‘Using stationarity tests in antitrust market definition’”,
American Law and Economics Review, 6(2), 441-464.
FREEDMAN, D. (2008). On regression adjustments in experiments with several treatments. Annals of applied
statistics, 2, 176-196.
FRONHÖFER, B. (2011). Introduction to Many-Valued Logics. Dresden: Technische Universität Dresden. Acesso em
15
de
October
de
2015,
disponível
em
http://www.wv.inf.tu-dresden.de/Teaching/SS2011/mvl/mval.HANDOUT2.pdf
FUJII, E. M., & MARK, J. (1985). An Almost Ideal Demand System for Visitor Expenditures. Journal of Transport
Economics and Policy, 19, 161-171.
GAILLARD, E. (5 de April de 2007). Extent of Court Review of Public Policy. New York Law Journal, 237(65), 1-4.
Fonte: http://www.shearman.com/~/media/Files/NewsInsights/Publications/2007/04/Extent-of-Court-Review-ofPublic-Policy/Files/IA_NYLJ-Extent-of-Court-Review_040308_16/FileAttachment/IA_NYLJ-Extent-of-CourtReview_040308_16.pdf
GEIER, M., & DERRIEN, Y. (s.d.). Dificuldade de fornecimento de componentes - Dificuldade de desenvolvimento
de processo de blindagem. Acesso em 3 de December de 2015, disponível em Departamento de Ciência e
Tecnologia Aeroespacial - Instituto de Estudos Avançados: http://www.ieav.cta.br/peice/30_10/30_10_6.pdf
GELMAN, A. (2008). Objections to Bayesian statistics. Bayesian Analysis. Volume 3, Número 3, 445-449. Acesso em
01 de Julho de 2015, disponível em http://www.stat.columbia.edu/~gelman/research/published/badbayesmain.pdf
GIDDENS, A. (1989). A constituição da sociedade. São Paulo: Martins Fontes.
GORE, D., LEWIS, S., LOFARO, A., & DETHMERES, F. (2013). The Economic Assessment of Mergers under
European Competition Law. Cambridge - UK: Cambrigde University Press.
GREENE, W. H. (2003). Econometric Analysis. Nova Iorque: Prentice Hall.
GRUBER, J. (1994). The Incidence of Mandated Maternity Benefits. American Economic Review, 84(3), 622-41.
Acesso
em
23
de
November
de
2015,
disponível
em
https://www3.nd.edu/~wevans1/class_papers/gruber_maternity_benefits.pdf
HARRIS, B. C. (25 de June de 2015). Recent Observations About Critical Loss Analysis. Acesso em 16 de October de
2015, disponível em Department of Justice (DOJ) http://www.justice.gov/: http://www.justice.gov/atr/recentobservations-about-critical-loss-analysis
HARRIS, B. C., & SIMONS, J. J. (1989). Focusing Market Definition: How Much Substitution Is Enough? . Research
International Law and Economy 207.
HARRIS, R. (1997). Significance Tests have their place. Psychological Science, 8, n.1, 8-11. Acesso em 20 de October
de 2015, disponível em http://psychology.okstate.edu/faculty/jgrice/psyc5314/harris_1997.pdf
HARRIS, R. (2009). Reforming significance test via three-valued logic. Em L. L. HARLOW, S. A. MULAIK, & J. H.
STEIGER, What If There Were No Significance Tests? (pp. 145-174). New York: Psychology Press. Fonte:
https://books.google.com.br/books?id=5cLeAQAAQBAJ&pg=PA155&lpg=PA155&dq=hypothesis+test+three+
valued+logic&source=bl&ots=oLdqEEtQnN&sig=Kr3OLNqU7Z6Z63yJ39g7Rm_Tirw&hl=ptBR&sa=X&ved=0CCEQ6AEwAGoVChMIhf-WbHRyAIVhxiQCh1NLQI2#v=onepage&q=hypothesis%20test%2
HAUSMAN, J., & KUERSTEINER, G. (2008). Difference in Difference meets Generalized Least Squares: High Order
Properties of Hypothesis Test. Journal of Econometrics, 371-391.
HECKMAN, J. J., & HOTZ, V. J. (1989). Choosing Among Alternatives Nonexperimental Methods for Estimating the
Impact of Social Programs. Journal of the American Statistical Association, 862-874.
HENDRICKS, K., & MCAFEE, R. P. (2010). A Theory of Bilateral Oligopoly. Economic Inquiry, 48(2), 391-414.
HO, D. E., IMAI, K., KING, G., & SUART, E. A. (2007). Matching as nonparametric preprocessing for reducing
model dependence in parametric causal inference. Political Analysis, 15(3), 199-236.
HOUCK, J. P. (1977). An approach to specifying and estimating non reversible functions. American Journal of
Agricultural Economics, 59, 570-572.
ICC. (27 de March de 2008). Potential anti-competitive effects of presuming dominance or substantial market power
based on market shares. Acesso em 25 de October de 2015, disponível em International Chamber of Commerce:
http://www.iccwbo.org/advocacy-codes-and-rules/areas-of-work/competition/international-competition-network(icn)/
IVALDI, M., & LÖRINCZ, S. (2005). A Full Equilibrium Relevant Market Test: Application to Computer Servers ,
August 2005. IDEI Working Paper, n. 341, 1-51.
JACOBSON, J. M. (August de 2015). Another take on the relevant welfare standard for antitrust. theantitrustsource,
1-8. Acesso em 8 de December de 2015, disponível em https://www.wsgr.com/PDFSearch/jacobson-0815.pdf
KAISER, H. (1960). Directional Statistical Decisions. Psychological Review, 67, 160-167.
KAPLOW, L. (2010). Why (ever) define markets? Harvard Law Review, 124, 437-517.
KATZ, M. L., & SHAPIRO, C. (2003). Critical Loss: Let’s Tell the Whole Story. Antitrust , 49-56.
KOLLMANN, T. (10 de July de 2012). Nuclear Madness: What was Special About the Brazil-Germany Nuclear Accord
of
1975?
Acesso
em
3
de
December
de
2015,
disponível
em
Cold
War:
http://www.coldwar.hu/html/en/conferences/Third_2012/Papers/Conference_paper_-_Thomas_Kollmann.pdf
LAFONTAINE, F., & SLADE, M. (2007). Vertical Integration and Firm Boundaries: The Evidence. Journal of
Economic Literature, 45 (3), 629-685.
LANDE, R. H. (Spring de 2001). Consumer Choice as the Ultimate Goal of Antitrust. Research Paper Series. University
of Pittsburgh Law Review , 62 (3), 503-525. Acesso em 8 de December de 2015, disponível em
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1478680
LAU, L. (1982). On Identifying the Degree of Competitiveness from Industry Price and Output Data. Economics
Letters, 10, 93–99.
LEITE, A. d. (Janeiro de 2004). Capítulo 7 da Apostila Tópicos de Álgebra Moderna Elementar, 1.0. (UNESP, Editor)
Acesso em 6 de Outubro de 2015, disponível em Faculdade de Engenharia de Guaratinguetá:
http://www.feg.unesp.br/~anachiaradia/Material/Apostila%20do%20Prof%20Aury%20-%20Cap%207.pdf
LENTO, C. (2014). Shifting Sands: An Analysis of OPEC Under U.S. Antitrust and EU Competition Law an How U.S.
Oil Boom Might Change It All. LSU Journal of Energy Law and Resources, 281-307. Acesso em 2 de December
de 2015, disponível em http://digitalcommons.law.lsu.edu/cgi/viewcontent.cgi?article=1036&context=jelr
Lerner, A. (1934). The concept of monopoly and the measurement of monopoly power. Review of Economic Studies,
157-175.
LIEBESKIND, R., & TIFFANY, J. R. (31 de July de 2009). Two years after Leegin, questions remain on lawfulness
of resale price maintenance. Acesso em 21 de October de 2015, disponível em Pillsburylaw:
http://www.pillsburylaw.com/sitefiles/publications/053986e13bce687e8e0ca54e89372048.pdf
LIN, W. (2013). Agnostic notes on regression adjustments to experimental data: reexaminig Freedman´s critique. The
annals of applied statistics, 7(1), 295-318. Acesso em 7 de December de 2015, disponível em
http://www.columbia.edu/~wl2513/agnostic.pdf
LINKLATER, A. (1996). The Achievements of Critical Theory. Em S. SMITH, & K. B. BOOTH, International
Theory: Positivism and Beyond. Cambridge: Cambridge University Press.
LUHMANN, N. (1989). Law as a social system. Northwestern University Law Review, 83, 136-150.
LUHMANN, N. (1989). Ökologische Kommunikation: Kann die moderne Gesellschaft sich auf ökologische
Gefährdungen einstellen? Cambridge: Polity Press.
LUHMANN, N. (1995). Die Kunst der Gesellschaft. Frankfurt: Suhrkamp.
MAYO, D. G. (1996). Error and the Growth of Experimental Knowledge. Chicago: The University of Chicago Press.
McAFEE, R. P., & WILLIAMS, M. A. (1988). Can event studies detect anticompetitive mergers? Economics Letters
28 , 199-203.
MEARSHEIMER, J. J. (1990). Back to the future. Instability in Europe after the Cold War. International Security, 15.
MEARSHEIMER, J. J. (2001). The tragedy of Great Power Politics. New York: W.W.Norton.
MILLER, D. (1976). Social Justice. Oxford: Clarendon Press.
MILNE, H. (1986). Desert, effort and equality. Journal of Applied Philosophy, 3, 235-243.
MORESI, S., REITMAN, D., SALOP, S., & SARAFIDIS, Y. (2011). Gauging Parallel Accommodating Conduct
Concerns with the CPPI. SSRN. Acesso em 20 de Julho de 2015, disponível em
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1924516
MORESI, S., REITMAN, D., SALOP, S., & SARAFIDIS, Y. (2015). cGUPPI: Scoring incentives to engage in Parallel
Acommodating Conduct. ssrn, 1-32. Fonte: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2640962
MOTTA, M. (2004). Competition Policy: Theory and Practice. New York: Cambridge University Press.
NIDA-RÜMELIN, J. (2008). Non bayesian decision theory. Cambridge: Springer.
NOZICK, R. (1974). Anarchy, State, and Utopia. Oxford : Blackwell.
O´DONOGHUE, R., & PADILLA, J. A. (2006). The Law and Economics of article 82 EC. Oxford, UK: Hart
Publishing.
OECD, O. f.-o. (13 de Dezembro de 2011). OECD Hearings. Arbitration and Competition. DAF/COMP(2010)40.
Acesso em 30 de Junho de 2015, disponível em http://www.oecd.org/competition/abuse/49294392.pdf
OLIVEIRA, O. M. (1999). Os descaminhos do Brasil nuclear. Ijuí: Unijuí.
OLIVER, G. (2010). Tenors, Real Estate Brokers And Golf Clubs: A Quick Look at Truncated Rule of Reason
Analysis. . Antitrust, 40-46.
PALMEDO, M. (Summer de 2013). Do Pharmaceutical Firms Invest More Heavily in Countries with Data Exclusivity?
Currents International Trade Law Journal, 1-21. Acesso em 3 de December de 2015, disponível em
http://ssrn.com/abstract=2259797
PANZAR, J., & ROSSE, J. R. (1987). Testing for "Monopoly" Equilibrium. Journal of Industrial Economics, 35, 443456.
PORTER, R. (1983). A study of Cartel stability: the joint executive committe. Bell Journal of Economics, 301-314.
POSNER, R. (2003). Economic Analysis of Law. Nova Iorque: Aspen.
POSNER, R. (2008). How judges think. Cambridge: Harvard University Press.
POSNER, R. A. (1995). A Statistical Study of Antitrust Enforcement. Em F. S. McCHESNEY, & W. F. SHUGHART
II, The causes and consequences of Antitrust: public choice perspective (pp. 73-94). Chicago: The University of
Chicago Press.
RAWLS, J. (1971). A Theory of Justice. Cambridge: Belknap Press.
RILEY, J. (1989). Justice Under Capitalism, Markets and Justice. New York: J.W. Chapman.
RIORDAN, M. (2008). Competitive Effects of vertical integration. Em P. BUCCIROSSI, Handbook of Antitrust
Economics (pp. 145-182). London: The MIT Press.
ROEMER, J. (1982). A general theory of exploitation and class. Cambride, MA: Harvard Univ. Press.
ROEMER, J. (1985). Equality of talent. Economics and Philosophy, 155-188.
ROSCH, J. T. (7 de December de 2006). Monopsony and The Meaning of “Consumer Welfare”: a closer look at
Weyerhaeuser. Acesso em 8 de December de 2015, disponível em U.S. Federal Trade Commission:
https://www.ftc.gov/sites/default/files/documents/public_statements/monopsony-and-meaning-consumer-welfarecloser-look-weyerhaeuser/061207miltonhandlerremarks_0.pdf
RUBIN, D. (1974). Estimating Causal Effects of Treatments in Randomized and Nonrandomized Studies. Journal of
Education Psychology, 66, 688-701.
SARDUSKI, W. (1985). Giving Desert Its Due. Dordrecht: D. Reidel.
SCHEFFMAN, D. T., & SIMONS, J. J. (2003). The State of Critical Loss Analysis: Let’s Make Sure We Understand
the Whole Story. American Bar Association, 1-9. Acesso em 14 de October de 2015, disponível em
http://www.americanbar.org/content/dam/aba/publishing/antitrust_source/03/11/scheffman.authcheckdam.pdf
SCHWARTZ, L. B. (1979). "Justice" and other non-economic goals of antitrust. University of Pennsylvania Law
Review,
1076-1081.
Acesso
em
8
de
December
de
2015,
disponível
em
http://scholarship.law.upenn.edu/cgi/viewcontent.cgi?article=4868&context=penn_law_review
SHAFFER, E. R., & BRENNER, J. (s.d.). A Trade Agreement’s Impact On Access To Generic Drugs. Health Affairs.
Acesso em 3 de December de 2015, disponível em http://content.healthaffairs.org/content/28/5/w957.full
SHAKESPEARE, W. (1623). The Tragedy of Hamlet, Prince of Denmark. Acesso em 14 de October de 2015,
disponível em http://shakespeare.mit.edu/hamlet/hamlet.3.1.html
SHARPE, W. (1963). A Simplified Model for Portfolio Analysis. Management Science, 277-293.
SINCLAIR, B., McCONNELL, M., & GREEN, D. (October de 2012). Detecting spillover effects: design and analysis
of multilevel experiments. American Journal of Political Science, 56(4), 1055-1069.
SPANOS, A. (Dezembro de 2007). Philosophy of Econometrics. Acesso em 1 de Julho de 2015, disponível em
citeseerx: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.126.8707&rep=rep1&type=pdf
STRNAD, J. (2007). Should Legal Empiricists Go Bayesian? Stanford Law and Economics Olin Working Paper, No.
342. Acesso em 01 de Julho de 2015, disponível em http://ssrn.com/abstract=991335
SZYSZCZAK, E. (2011). Controlling Dominance in European Markets. Fordham International Law Journal, 33, Issue
6, 1738-1775.
TURNER, D. F. (1987). The Durability, Relevance, and Future of American Antitrust Policy,. California Law Review,
75, 797-798.
ULEN, T., & COOTER, R. (2004). Law and Economics. Nova Iorque: Pearson.
VELLA, F., & VERBEEK, M. (1999). Estimating and Interpreting Models with Endogenous Treatment Effects.
Journal of Business and Economic Statistics, 17, 473–478.
WALLE, S. V. (7 de Fevereiro de 2013). Private Antitrust Litigation in the European Union and Japan – A
Comparative Perspective. Portland: Maklu Publisher. Acesso em 01 de Julho de 2015, disponível em SSRN:
https://books.google.com.br/books?id=UdcLqbQgkWwC&pg=PA215&lpg=PA215&dq=SA+Thales+Air+D%C3
%A9fense+v.+Euromissile&source=bl&ots=cHD7J4VNRN&sig=7kFSL6ywbgRkQidInNPbWBnBv04&hl=ptBR&sa=X&ei=_y2UVaWsMsX2-QG27J-ACQ&ved=0CFEQ6AEwBg#v=onepage&q=SA%20Thales%20Ai
WALLER, S. W. (2009). Justice Stevens and the Rule of Reason . Southern Methodist University Law Review, 61.
Acesso em 21 de October de 2015, disponível em http://ssrn.com/abstract=1290770
WALTZ, K. (1979). Theory of International Politics. New York: McGraw-Hill.
WERDEN, G. (23 de Março de 2007). Monopsony and the Sherman Act: Consumer Welfare in a New Light. Acesso
em 01 de Julho de 2015, disponível em SSRN: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=975992
WERDEN, G., & FROEB, L. (1994). The Effects of Mergers in Differentiated products Industries: Logit Demand and
Structural Merger Policy. Journal of Law, Economics, & Organization Vol. 10, No. 2, 407-426.
WERDEN, G., & FROEB, L. M. (1993). Correlation, Causality, and All that Jazz: The Inherent Shortcomings of Price
Tests for Antitrust Market Delineation. Review of Industrial Organization, 8, 329-353. Acesso em 14 de October
de 2015, disponível em http://www.cas.hse.ru/data/999/860/1224/Froeb,_Werden_(1993).pdf
WERDEN, G., FROEB, L., & TARDIFF, T. (1996). The Use of the Logit Model in Applied Industrial Organization.
International Journal of the Economics of Business, vol. 3, issue 1, 83-105.
WHITE, H., & LU, X. (2014). Robustness check and robustness test in applied Economics. Journal of Econometrics,
178,
194-206.
Acesso
em
11
de
November
de
2015,
disponível
em
http://econ.ucsb.edu/~doug/245a/Papers/Robustness%20Checks.pdf
WILLIAMSON, O. (1968). Economies of Scale as an Antitrust Defense: The Welfare Trade-offs. American Economic
Review, 58, 18-36.
WOOLDRIDGE, J. (2003). Introductory econometrics: a modern approach. South-Western College Pub.
WRIGHT, J. D., & GINSBURG, D. H. (2013). The goals of antitrust: welfare trumps choice. Fordham L. Rev., 24052423.
YANG, H., WANG, Q., ZHANG, Q., & ZHANG, A. (2013). Market power and its determinants of the Chinese airline
industry. Acesso em 3 de November de 2015, disponível em International Centre for Maritime Studies:
http://www.icms.polyu.edu.hk/ifspa2013/Presentations/A06.pdf