Considering the discussion at the roundtable, the delegate submissions, and the background paper, the following key points emerge: • Mutually aware that their actions will produce reactions from rivals, oligopolists have a strong incentive to substitute anti-competitive co-operation for vigorous competition. Such behaviour, referred to in what follows as “co-ordinated interaction”, has negative welfare effects. This is especially clear in the case of collusion. The welfare losses associated with co-ordinated interaction are qualitatively the same as those linked to monopoly, i. e.
higher prices, X-inefficiency, slower innovation, and reduced product variety. Good examples of co-ordinated interaction other than collusion are price leadership, delivered pricing, and price stickiness around certain focal points. • There is a considerable gap between the desire to engage in co-ordinated interaction and the ability to do so successfully. It can be difficult to reach mutually acceptable terms of co-operation, and to ensure that firms do not deviate from them. Potentially co-operating oligopolists could have dissimilar costs and face different demand conditions as well.
While such differences may not pose a major problem for anti-competitive coordination of non-price rivalry, they could make it difficult for firms to settle on mutually beneficial price levels and price gaps among the various goods on offer. Even where these issues are resolved, successful co-operation regarding prices is not assured. Once price has been raised significantly above marginal cost, each oligopolist will have an incentive to shade its prices especially if it believes this will escape detection and retaliation.
Durable co-operation often requires that oligopolists work out ways to increase the probability of detection and the certainty of punishment for cheating, all the while avoiding prosecution for anti-competitive behaviour. [Despite the evident difficulties, large scale and long running cartels are known to exist as evidenced by the increasing number of such cartels currently being prosecuted in a number of OECD Members. ] Behavioural Remedies • The principal behavioural remedy for co-ordinated interaction is prohibition of collusion backed up by heavy fines and, in some countries, imprisonment.
Because of the heavy sanctions, colluders generally avoid creating direct evidence of anti-competitive communications. This makes it difficult to detect and prove collusion, and explains why competition agencies are often forced to rely on circumstantial evidence. However, evidence of parallel conduct is not and should not be considered sufficient proof of collusion. 7 DAFFE/CLP(99)25 The most common circumstantial evidence consists of closely parallel behaviour. The legal and economic problem with such evidence is that parallel behaviour could have causes other than collusion.
In oligopoly settings, parallel price movements for example could arise simply through independent rational behaviour. To convince courts that parallel behaviour has arisen through some kind of agreement rather than merely resulting from oligopolistic interdependence, competition authorities must usually demonstrate that something more has occurred, i. e. establish the existence of one or more “plus factors”. These are basically forms of conduct that would not be economically rational absent some agreement among competitors. Courts have also been willing to consider evidence that a particular market structure facilitates collusion.
This may be changing, however, since it is increasingly recognised that a market structure conducive to collusion would also facilitate forms of co-operation falling short of such behaviour. Some courts are inclined to presume that firms would not take the legal risks of colluding if they could readily achieve the same result in a more flexible, less risky manner. Some OECD countries have reacted to the problems associated with proving collusion by widening the net to prohibit “concerted practices”, and in a few countries, “facilitating practices” (see below) as well.
One country’s competition act goes even further by providing presumptions that certain acts constitute evidence of collusion even if there has been no explicit agreement. Attempts to widen what is caught by prohibitions of anti-competitive horizontal agreements and to lower burdens of proof could result in countries effectively outlawing the kind of parallel behaviour that oligopolistic interdependence strongly fosters. But parallel behaviour induced by interdependence cannot be eradicated with behavioural measures short of ongoing regulation of competitive conduct, a cure that is worse than the disease.
• In order to increase the chances for successful collusion, and in particular to improve abilities to detect and punish cheating, co-operating firms may employ what are known as “facilitating practices”. Some countries prohibit such practices either as abuses of dominance or as themselves constituting anti-competitive agreements. In others, they are counted along with other “plus factors” as circumstantial evidence of an agreement. Just about any behaviour making it easier for oligopolists to co-operate could be, under the right circumstances, a facilitating practice.
The most common and perhaps important are measures designed to increase the transparency of the market, i. e. improve participants’ knowledge of rivals’ competitive behaviour including the prices they actually charge. Good examples include: advance notice of nonbinding price changes; meeting competition and most favoured nation clauses; resale price maintenance; delivered pricing; sharing non-complementary intellectual property rights; and, in some cases, building a web of joint ventures, strategic alliances, shared ownership and cross-directorships.
Participating in industry associations could also be regarded as a facilitating practice especially where associations are used to promote or disguise the exchange of sensitive information, adoption of anti-competitive standards, or changes in government regulations which would facilitate co-ordinated interaction. This is not to say that all industry associations or all of their practices are anti-competitive. The point is merely that industry associations deserve close scrutiny in oligopolistic markets. Some competition laws prohibit
certain facilitating practices in oligopoly markets even when such practices are not being used in concert with any anti-competitive act. This should be done on a case by case basis since the same general type of practice could both enhance efficiency and facilitate coordinated interaction. 8 DAFFE/CLP(99)25 • Facilitating practices are sometimes instituted and even required by various regulatory bodies. Competition agencies should advocate their removal unless absolutely necessary for some legitimate regulatory purpose.
When telecommunication regulators, for example, require regulated companies to publish nonbinding future price intentions, there is a risk this will facilitate price collusion. The same could be said of public tendering rules requiring prompt publication of all rather than just the winning bid. It should also be noted that government or private sector purchase through a tender procedure is rendered less competitive and more prone to co-ordinated interaction if product rather than performance criteria are specified.
Competition agencies should point out that prices are likely to rise when facilitating practices are imposed, and be willing to work with regulators to develop satisfactory alternatives. • Prohibitions of horizontal exclusionary behaviour could help ensure that markets remain competitive rather than slide towards co-ordinated interaction enhancing oligopolistic structures.
The principal enforcement problem as regards horizontal exclusionary behaviour, predatory pricing for example, is to distinguish between vigorous, efficiency enhancing competition, which often harms competitors, and rivalry intended or having the effect of permanently excluding competitors from the market so as to reduce competition. Properly applied, however, prohibitions of horizontal exclusionary behaviour could play a valuable role in preventing the emergence or strengthening of co-ordinated interaction.
Structural Remedies • Because of difficulties inherent in proving collusion and problems in devising remedies for weaker forms of co-ordinated interaction, most countries’ competition laws supplement behavioural with structural measures. Chief among them is merger review. Merger review is undertaken in order to block or condition mergers that appear to have a significant potential to harm consumer interests. Some mergers create this risk by increasing an acquiring firm’s unilateral ability to act in an anti-competitive fashion.
Others threaten competition because they create market structures more conducive to collusion or other forms of co-ordinated interaction that might not themselves be illegal under a country’s horizontal agreement or abuse of dominance prohibitions. • There are a number of factors whose presence considerably increases the probability of co-ordinated interaction because they affect firms’ abilities and incentives to reach terms of co-operation and subsequently to detect and punish cheating on the arrangement.
The factors likely to be of particular importance include: seller concentration levels; height of barriers to entry/expansion; degree of symmetry in products, costs, firm sizes and objectives; significance of economies of scale and sunk costs; industry phase of development; variability of market demand and incidence of cost shocks; size and frequency of purchases; prevalence of buyer power; the incidence of multimarket (either product or geographic) contact among leading firms; and degree of transparency and ability to commit to certain behaviours including punishment for cheating.
• Determining a particular factor’s effect on the probability of co-ordinated interaction and assessing the impact of all factors taken together is very much a case by case exercise. Economic analysis can help identify relevant factors and provide a framework for analysing their influence, but economic 9 DAFFE/CLP(99)25 theory will seldom provide an off the shelf model yielding reliable predictions in a specific situation facing a competition agency.
While there is no broadly applicable framework that could be used to aggregate the effects of the many factors impinging on the probability of co-ordinated interaction, there appears to be a consensus that such behaviour is unlikely to occur if concentration and barriers to entry are sufficiently low. Some countries build safe harbours around those two factors, but it must be conceded that there is a certain degree of arbitrariness in their definition.
It is difficult to determine what constitute safe levels of concentration and barriers to entry, and the answer could well differ from market to market. • Some countries take action against mergers predicted to substantially lessen or prevent competition. Others block or seek to modify those believed to create or strengthen a dominant position. From the point of view of prohibiting mergers significantly increasing the risk of co-ordinated interaction, there may be little difference between the two types of tests.
This assumes, however, that dominance is defined to include joint dominance, and that structural ties are not necessary in order to establish joint dominance. With regard to increased risk of co-ordinated interaction, there appears to be a question whether the substantial lessening test could be used to prevent a series of small mergers from eventually yielding an anti-competitive market structure. Such a set of mergers could at some point be halted in jurisdictions applying dominance based tests.
• Required access to an essential facility could, in some instances, amount to a structural remedy when it is applied to expand the number of suppliers in an oligopolistic industry. As with all remedies for co-ordinated interaction, application of the essential facilities doctrine presents some important difficulties. Care must be taken not to discourage necessary network investments including, where feasible, the development of facilities based competition. Moreover, access terms must be set so as to permit efficient competition without subsidising entry by inefficient providers.
• An optimal oligopoly policy may involve some difficult tradeoffs. A number of studies have shown that a modicum of concentration may be required in order to increase innovation activity, but that too much concentration could stifle it. The exact thresholds are of course difficult to estimate and could vary from market to market. There may also be a trade-off between the benefits of allowing firms to meet in order to adopt various pro-competitive standards and the costs associated with raising the risk of co-ordinated interaction. 10
DAFFE/CLP(99)25 NOTE DE SYNTHESE par le Secretariat Au vu des debats qui se sont deroules lors de la table ronde, des contributions des delegues et de la note de synthese, on peut degager les principaux points suivants : • Mutuellement conscients que leurs actions entraineront une reaction de la part de leurs rivaux, les oligopoles ont tout interet a substituer une cooperation anticoncurrentielle a une vive concurrence. Une telle attitude, designee dans la suite de ce texte comme une « interaction coordonnee » est prejudiciable au bien-etre.
C’est particulierement clair dans le cas de collusion. Les pertes de bien-etre associees a l’interaction coordonnee sont qualitativement les memes que celles qu’induisent les situations de monopole, a savoir la hausse des prix, l’X-inefficience, le ralentissement de l’innovation et la moins grande variete de produits. Parmi les exemples d’interaction coordonnee autres que la collusion, on peut citer les prix directeurs, la fixation de prix livraison comprise et la viscosite des prix autour de certains pivots.
• Il y a une difference considerable entre le desir de s’engager dans une interaction coordonnee et la possibilite de le faire avec succes. Il peut etre difficile de s’entendre sur des conditions mutuellement acceptables de cooperation et de s’assurer que les societes ne s’ecartent pas de ces conditions. Des oligopoles susceptibles de pratiquer la cooperation pourraient avoir des couts differents et etre confrontes a des conditions egalement differentes de la demande.
Meme si ces differences peuvent ne pas poser de veritable probleme s’opposant a une coordination anticoncurrentielle dans le cadre d’une rivalite ne portant pas sur les prix, elles peuvent poser aux firmes des difficultes pour s’entendre sur des niveaux de prix mutuellement avantageux ainsi que sur des ecarts de prix entre les differents biens proposes. Meme lorsque les firmes parviennent a resoudre ces questions, le succes de la cooperation sur les prix n’est pas assure.
Une fois que le prix a ete rehausse de facon importante au-dessus du cout marginal, chaque firme composant l’oligopole aura interet a baisser ses prix surtout si elle est convaincue que cela passera inapercu et qu’elle ne subira pas de represailles. Une cooperation durable exige souvent que l’oligopole se donne toutes les chances de detecter ce type de comportement et fasse comprendre aux tricheurs qu’ils seront punis a coup sur, le tout en evitant d’etre poursuivi pour comportement anticoncurrentiel.
[En depit des difficultes evidentes, on sait qu’il existe des ententes de grande envergure et etablies de longue date, comme le prouve la multiplication des poursuites a l’encontre de telles ententes dans un nombre de pays Membres de l’OCDE]. Les recours comportementaux • Le principal recours comportemental face a une interaction coordonnee reside dans l’interdiction des collusions assortie d’amendes importantes, voire, dans certains pays, de peines d’emprisonnement. En raison de la lourdeur des sanctions, ceux qui se livrent a des collusions evitent en general de donner