Needed: A theory of harm for the Intel case

Giacomo Calzolari, Vincenzo Denicolò

20 January 2016

a

A

With a market share of over 80%, Intel holds a dominant position in the market for microchips, an essential component of computers and other electronic devices. As a result, Intel is subject to special antitrust scrutiny. In particular, its pricing strategies have been the object of a long antitrust controversy. The crux of the matter is the discounts given from 2001 to 2006 to computer makers for buying all, or most, of their chips from Intel.

In the US, antitrust cases brought by AMD (Intel’s main competitor), the Federal Trade Commission, and the New York State Attorney General were all settled in 2009. Intel agreed to stop the discounts and paid AMD $1.25 billion in damages. The same year, Intel was condemned by the European Commission. It appealed, but the European General Court upheld the Commission’s decision, confirming the record-breaking fine of €1.06 billionIntel appealed again, this time to the European Court of Justice, whose decision is still pending.

Although various antitrust authorities and courts have already dealt with the case, the exact reason for regarding Intel’s discounts as anticompetitive is still unclear. In the US, the out-of-court settlement made it unnecessary to articulate a theory of harm. In Europe, the Commission argued at length that the discounts were effectively predatory. However, the European General Court refused to review the Commission’s arguments, holding that discounts that are conditioned on exclusivity, or on meeting certain market-share requirements, are illegal by their very nature.  As the European General Court stated, “it is thus not necessary to show that they are capable of restricting competition on a case by case basis”.  The European General Court also argued that this approach is consistent with the European Court of Justice’s case law since Michelin.

However, some commentators believe that this time, the European Court of Justice may open the door to a more flexible approach based on the rule of reason. This would be a very sensible policy move. In our opinion, per se illegality should be reserved only for those practices (e.g. collusion) which economic analysis has shown to be anticompetitive in all but the most exceptional circumstances.

Clearly, this is not the case for exclusive contracts or market-share discounts. The theory here is largely unsettled. Several alternative models have been developed, with different predictions, and the empirical evidence is sparse and conflicting. In sum, there is no general economic consensus at present. By itself, this would speak for a rule of reason approach. Ideally, exclusive contracts should be prohibited only on the basis of a coherent theory of harm that clarifies the framework in which the dominant firm operates, the incentives that it has to offer exclusive contracts, and the reason why these contracts may harm the buyers and ultimately the final consumers.

Antitrust theory

The problem is that most existing theories do not fit the facts of antitrust cases such as Intel, and therefore can hardly be used as a basis for policy. For example, some theories (e.g. Aghion and Bolton 1987) maintain that exclusive contracts may be profitable only to the extent that they commit the buyer to pay penalties for breaching exclusivity. However, Intel’s contracts could be terminated at will, or at very short notice, without any contractual penalty being incurred.

Other theories (e.g. the so-called naked exclusion models pioneered by Rasmusen et al. 1991) hold that exclusive contracts may be a means to raising rivals’ cost, by depriving rivals of economies of scale. These models are difficult to apply to cases such as Intel for several reasons. First, they assume that rivals are potential entrants that cannot contract with the buyers, whereas AMD has been active in the microchips market for more than 30 years. Second, naked exclusion theories require that a substantial amount of the market be foreclosed. But Intel’s contracts could foreclose at most some 10-15% of the market, which would still leave plenty of room for a competitor to achieve economies of scale and prosper. Finally, the short-term nature of Intel’s contract is a problem for theories of naked exclusion, too. Entry can hardly be deterred if buyers are committed only for a short period of time.

Pro-competitive theories, which argue that exclusive dealing may foster relation-specific investments (see e.g. Marvel 1982), may also be difficult to apply. In particular, such explanations may be plausible when various active firms employ exclusive contracts, but may be viewed with a certain scepticism when exclusive contracts are used by the dominant firm only, as in Intel and many other cases.

Predation

Faced with these difficulties, the European Commission resorted to a predation story. While recognising that the average price charged by Intel was above its unit costs, the Commission argued that if the discounts were apportioned to the contestable share of the market only, then the effective price would fall below cost.

It is easy to see why antitrust authorities like this predatory theory of harm – it brings the case to familiar territory, where the anticompetitive mechanism is well understood, and well-established tests are available for determining whether the dominant firm’s behaviour is anticompetitive or not. However, exclusive dealing is not predation. Its key feature is that the price is conditioned on rivals’ volumes, which may make these contracts anticompetitive even if prices do not fall short of costs. Applying predatory tests to exclusive dealing cases may thus result in many false negatives.

But in practice the Commission’s approach may result paradoxically in too many false positives rather than false negatives. The problem is that nobody knows what exactly the ‘contestable’ portion of the market is, and by narrowing it down and down, any discount may be argued to be predatory. The European General Court may thus have been right in dismissing the Commission’s approach after all, and it would not be a good idea for the European Court of Justice to revive it.

Dealing with cases like Intel

So, what insights can economists offer to antitrust authorities and courts for dealing with cases such as Intel? Fortunately, recent theoretical developments have enlarged our tools box. For lack of space, we shall focus here on our own recent work on the matter.

In two papers (Calzolari and Denicolò 2013, 2015), we have developed a new theory of exclusive contracts where the dominant firm’s market power is based on a competitive advantage, in terms of lower costs or better product quality, that it enjoys over its rivals. The theory assumes that the dominant firm cannot fully extract the buyers’ surplus by means of fixed fees and thus must resort to distortionary prices (i.e. prices that exceed marginal costs). Under this realistic assumption, we show that the dominant firm has an incentive to use exclusive contracts so as to better exploit its competitive advantage.

If the competitive advantage is large, the strategy is profitable. Buyers are harmed in terms of both higher prices and reduced variety. Competitors are harmed as they are excluded from at least a portion of the market.

If the competitive advantage is small, on the other hand, the strategy backfires. Exclusive contracts reduce prices and profits and benefit the buyers. However, competing firms are caught in a prisoners’ dilemma, and thus have an incentive to keep using exclusive contracts even if the profits are reduced.

In this theory, contracts have no commitment value and thus need not be long-term to be effective. They do not serve as a means to raising rivals’ cost, and thus the fraction of the market foreclosed need not be large. Marginal prices never fall short of marginal costs and thus there is no sacrifice of profits, nor any need of recoupment. The theory does not rely on competitors’ inability to contract with the buyers and thus can be applied also to cases in which competitors are active firms rather than potential entrants.

For all of these reasons, our theory seems more broadly applicable than alternative ones. It also confirms, however, that per se illegality is not a sound policy, and a rule of reason approach may be more appropriate.

In practice, the key issue that antitrust authorities and the courts should address is whether the dominant firm’s rivals can compete for exclusives affectively (in which case, exclusive contracts are pro-competitive) or not (in which case they are anticompetitive). That in turn depends on the size of the dominant firm’s competitive advantage. Admittedly, this is not easy to measure, but fortunately it is correlated with variables, such as the firms’ market shares, that are readily observable. Therefore, we believe that our approach might be the basis for a structured antitrust test that could be applicable in practice.

References

Aghion, P, and P Bolton (1987), “Contracts as a Barrier to Entry”, American Economic Review 77 (3): 388–401.

Calzolari, G and V Denicolò (2013), “Competition with exclusive contracts and market-share discounts”, American Economic Review 103(6): 2384-2411.

Calzolari, G and V Denicolò (2015), “Exclusive contracts and market dominance”, American Economic Review 105(11): 3321–3351.

Marvel, H P (1982), "Exclusive Dealing", Journal of Law and Economics 25: 1-26.

Rasmusen, E B, J M Ramseyer, and J S Wiley Jr (1991), “Naked Exclusion”, American Economic Review 81(5): 1137–45.

Footnotes

1 Another problem for the applicability of the predation theory is that AMD was never under any real threat of shutdown. This raises the issue of how could Intel recoup the profits lost in the short run. Luckily for the Commission, the European case law does not require proof of recoupment, so this was not actually perceived as a problem.

a

A

Topics:  Competition policy

Tags:  antitrust, Intel, ECJ, EGC, monopoly, EU

Professor of Economics, Bologna University ; Research Fellow, CEPR

Professor of Economics, University of Bologna and University of Leicester (currently on leave); Research Fellow, CEPR

Events