On 19 January 2023, the European Court of Justice (ECJ) issued its preliminary ruling relating to Unilever’s fine from the Italian Competition and Markets Authority1 (AGCM) for unlawfully inducing ice-cream sellers not to sell rival brands. This ECJ judgment is the fourth recent judgment overturning findings that exclusivity provisions were anti-competitive.
On 14 September 20222 , the General Court (GC) annulled the European Commission’s (EC or the Commission) findings3 regarding payments made by Google to Android mobile device manufacturers and network operators for the pre-installation of Google services. It was the third recent judgment from the GC overturning findings that rebates paid or revenues shared were anti-competitive, following Intel and Qualcomm rulings.
In short, the ECJ's decision relied on previous GC’s judgments to conclude that competition authorities must ensure that exclusivity clauses are capable of excluding competitors that are as efficient as the dominant entity, and must assess all justifications and evidence put forward by the dominant entity.
The judgments have focused on the need to ensure that the analysis of alleged anti-competitive exclusivity clauses, exclusionary rebates and revenue sharing provisions is rigorous. They have not changed the underlying law, and do not identify new forms of exclusionary conduct. The overarching theme is essentially the difficulty experienced by the EC in applying the legal test, at least in part because of the information asymmetry between the EC and allegedly dominant entities.
Exclusivity provisions can take various forms beyond outright exclusivity clauses, including rebates and revenue sharing in connection with exclusive purchase/supply obligations. As the ECJ said in Hoffman-La Roche, “an undertaking which is in a dominant position on a market and ties purchasers – even if it does so at their request – by an obligation or promise on their part to obtain all or most of their requirements exclusively from the said undertaking abuses its dominant position within the meaning of Article 102 of the Treaty, whether the obligation in question is stipulated without further qualification or whether it is undertaken in consideration of the grant of a rebate”.4
On 26 January 2022, the GC partially overturned the EC’s Intel decision and annulled the €1.06 billion fine. The EC had concluded that the rebate scheme in question aimed to ensure that laptop manufacturers would use only Intel’s chips.
In 2009, the EC fined5 Intel for having abused its dominant position on the worldwide market for processors through the implementation of a strategy to exclude competitors from the market. Intel’s conduct included the payment of conditional rebates to four original equipment manufacturers (OEMs), i.e., Dell, Lenovo, HP, and NEC. The rebates were conditional on those OEMs purchasing all or almost of their CPUs from Intel.
In addition, Intel was found to have made payments to Medium-Saturn Holding (a European retailer of microelectronic devices) on the condition that it only sells computers incorporating Intel’s CPUs. These rebates were found to have reduced the ability of competitors to compete on the merits of their own processors.
The ECJ referred the case back to the GC to conduct an in-depth examination of the circumstances and assess the effects of the fidelity rebates on competition (the “as-efficient-competitor” (AEC) test). On referral,6 the GC concluded that the EC’s analysis was incomplete, and had not established that the rebates were capable of having, or were likely to have, anti-competitive effects by foreclosing a competitor as efficient as Intel.
The GC found errors in the EC’s application of the AEC test to the different OEMs. In relation to Dell7
, it founds that the data used to assess the contestable share, i.e., the share of demand which Intel’s customers were willing and able to switch to another supplier, was insufficient. In relation to HP8
, the EC did not demonstrate that the foreclosure lasted for the entire period of the infringement, in relation to Lenovo9
, the errors were in the assessment of the non-cash advantages, and for NEC10
, the EC had extrapolated the results for a single quarter to the entire infringement period.
The GC found that the EC did not properly take into account all of the criteria that determine whether pricing practices can have a foreclosure effect. Notably, it did not properly consider the criterion relating to the share of the market covered by the practices, and it did not correctly analyse the duration of the rebates. As EU law made clear, the Commission must analyse12
- The extent of the undertaking’s dominant position on the relevant market
- The share of the market covered by the contested practice
- The conditions and arrangements for the grant of rebates
- The duration and amount of these arrangements
- The possible existence of a strategy to exclude at least as-efficient competitors.13
On 15 June 2022, the GC annulled the fine of approximately €1 billion imposed by the EC on Qualcomm relating to a rebate structure that the EC found ensured that Apple did not use Qualcomm’s competitors’ chips in certain products.
In 201814, the EC fined Qualcomm for an abuse of dominance on the worldwide market for LTE chipsets. It found that Qualcomm abused its dominant position on the LTE chipsets market by imposing agreements providing incentives which required Apple to obtain its LTE chipsets exclusively from Qualcomm. The EC took the view that these exclusivity payments were capable of having anti-competitive effects, reducing Apple’s incentives to switch to competing LTE chipsets.
The GC annulled the Commission’s decision,15 based on a number of procedural irregularities and errors in the analysis of the anti-competitive effects of the incentive payments. The GC found that the Commission had failed to consider all of the relevant factual circumstances in its analysis of the effects of the incentives. In particular, the EC concluded that the incentives reduced Apple’s incentives to switch suppliers in circumstances where Apple had no alternatives to Qualcomm’s chipsets for the majority of its needs during the period concerned. The GC concluded that the EC did not conduct an investigation of the existence of competing LTE chipsets that Apple could have used instead. Given the EC’s theory of harm, the EC could not ignore the fact that Apple was unable to switch.
The Commission argued that, rather than incentivising Apple in relation to the chips for which there were no technical alternatives, the incentives had foreclosed competitors from the contestable share. However, the GD held that this did not reflect the theory of harm in the decision. The theory of harm referred to “the capability of the payments concerned to have anti-competitive effects on the whole of the relevant market for LTE chipsets on the ground that Apple was dissuaded from sourcing from the applicant’s competitors for the whole of its LTE chipset requirements for iPhones and iPads”.16
The GC held that concluding that the payment reduced Apple’s incentives to switch to competitors for certain IPad models was insufficient, to support a finding that the payments were anti-competitive in relation to all of Apple’s requirements. It was necessary to examine all of the circumstances to assess whether the payments could have anti-competitive effects by foreclosing hypothetical at least as efficient competitors in the period concerned.
In September 2022, the GC annulled the EC’s findings regarding payments made by Google to Android mobile device manufacturers and network operators for the pre-installation of certain Google services.17 The GC annulled the decision in so far as it found the portfolio-based revenue sharing agreements in themselves to constitute an abuse.
The GC began by finding that the AEC test “can be useful” to assess whether the practices excluded competitors that were at least as efficient as the dominant company.18 It did not conclude that the test must necessarily be applied. However, where it is applied, it must be done rigorously.
In examining the EC’s application of that test, the GC found that only 5% of the market defined by the Commission had been covered by the conduct, while previous decisions had entailed coverage of between 39 and 85%.19 This level of coverage did not support the Commission’s finding that the agreements covered a “significant part” of the relevant markets. The GC also identified several errors of reasoning in applying the test, including the estimate of the costs attributable to a hypothetically at least efficient competitor, the revenues that could be shared by such a competitor, the contestable share, the assessment of the competitor’s ability to obtain pre-installation of its app, and the estimate of the likely revenues on the basis of the age of mobile devices in use:
- Costs attributable to a hypothetically at least as efficient competitor:20 Having noted that a hypothetically at least as efficient competitor is a competitor which, at least, has the same ability to generate revenue and is faced with the same costs as those of the undertaking in a dominant position, the GC found that the EC should have only taken incremental costs (i.e., the percentage of revenues shared with the other contracting party) into account, rather than operational costs (around 10-20%). Further, the costs were sourced from a document generated by a third party, not from an answer generated by Google, and was contested by Google.
- Revenues that could be shared by a hypothetically at least as efficient competitor:21 The GC confirmed the Commission’s assessment of revenues to be taken into account (i.e., the exclusion of revenues generated through search queries carried out on the webpage of its search engine, since Google did not share such revenues, and those revenues were generated independently of revenue sharing agreements with OEMs and MNOs).
- Share of search queries that could be contested by a hypothetically at least as efficient competitor:22 The GC concluded that the EC had appropriately relied on the share of search queries actually contested by competing general search services on PCs, instead of relying on the share that might be contested by a hypothetically at least as efficient competitor on mobile devices.
- Extent of pre-installation of an app of a hypothetically at least as efficient competitor:23 The EC concluded that OEMs and MNOs would only receive limited additional third-party revenue, in view of Google’s market share and its ubiquity at the points of access to general search services, and that third parties would face higher transaction costs and technical problems, e.g., regarding storage capacity. The GC found that the Commission had not demonstrated that a competitor would not be able to have its app pre-installed. The EC had noted that the Mobile Application Distribution Agreements (MADAs) prevented OEMs and MNOs from pre-installing exclusively a competing general search service app. However, the EC’s decision concerned pre-installation in addition to Google Search, not in its absence. As a result, the GC found that the EC’s findings were of no relevance to the theory of harm. In addition, the GC ruled that the EC cannot qualify the ability of a hypothetically at least as efficient competitor to respond to Google’s portfolio-based revenue sharing agreements by concluding that competitors could have their apps pre-installed on a limited number of OEM’s or MNO’s mobile devices.
- Applying the ratione temporis of the AEC test:24 The GC concluded that the EC committed an error by not quantifying the effect of devices already sold on the ability of a hypothetically at least as efficient competitor to offset the portfolio-based revenue sharing. The EC effectively assumed, without further analysis, the capacity of new and old mobile devices to generate the same general search revenues.
In 2017, the AGCM found that Unilever had abused a dominant position on the Italian market for the sale of individually packaged ice cream, intended for consumption outside the home, distributed through various sales outlets. It found that independent distributors of Unilever’s products had imposed exclusivity clauses on sales outlets25. Unilever appealed that decision to the Tribunale Administrativo Regionale per il Lazio, which dismissed Unilever’s action in its entirety.
It then appealed that judgment to the Consiglio di Stato, which referred questions to the ECJ for preliminary ruling. The first question concerned the circumstances in which the acts of formally autonomous and independent economic operators, namely distributors, may be imputed to the manufacturer of the products distributed26.
The ECJ held that the actions of distributors forming part of the distribution network for the goods and services of a producer in a dominant position may be imputed to that producer if it is established that those actions were not taken independently by those distributors but formed part of a policy unilaterally decided by the manufacturer that was implemented through the distributors. Unilever had expressly instructed the distributors to implement a particular commercial policy and to sign standard contracts with the sales outlets, which could not be modified (except with the express consent of Unilever). This was found to be sufficient to impute the distributors' conduct to Unilever.
The national court then asked whether Article 102 TFEU requires the authority to assess whether exclusivity clauses have the effect of foreclosing equally efficient competitors from the market and, in particular, to examine the economic analyses submitted by a party concerning the actual ability of the alleged conduct to foreclose equally efficient competitors from the market.27
The ECJ first recalled that exclusivity clauses are not necessarily capable of producing exclusionary effects. The ECJ stated that, where exclusivity clauses are included in distribution agreements, a competition authority must establish whether, in light of all the relevant circumstances and the economic analyses and justifications provided by the undertaking concerned regarding the ability of the clauses to exclude competitors which are as efficient as the dominant undertaking, those clauses are capable of restricting competition.28
It went on to confirm that its findings on the exclusionary effects of rebate systems, set out in Intel,29
also apply to exclusivity clauses. Exclusionary effects may be outweighed by efficiency gains which also benefit the consumer. However, the balancing of the positive and negative effects of such a practice can only be done after the authority has analysed the ability of the practice to foreclose at least equally efficient competitors.
The ECJ noted its earlier findings that the "as efficient competitor" test is optional30 but went on to state that such a test enables the effects of a practice to be quantified. In the case of exclusivity clauses, the test can be used to determine whether a hypothetical competitor with a cost structure similar to that of the dominant undertaking would be able to offer its products or services only at a loss or with insufficient margin if it had to compensate the sales outlets’ operators for their costs associated with switching suppliers or for the loss of the discounts.
Further, if such an analysis is submitted, the authority must assess its probative value. Finally, the ECJ stated that the authority cannot exclude the relevance of such evidence without 31:
- Stating the reasons why it considers that the evidence does not contribute to establishing the potential of the practices to produce exclusionary effects; and
- Giving the undertaking the opportunity to identify additional evidence.
While Intel addressed conditional/loyalty rebates, Qualcomm and Google considered exclusive purchasing obligations linked to incentive payments and revenue sharing agreements, and Unilever related to exclusivity clauses. The practices that all four entities were alleged to have engaged in required a detailed assessment of their alleged anti-competitive effects:
- In Intel, the GC concluded that the loyalty rebates are not per se infringements of Article 102 and the EC must analyse, in all cases, whether there were anti-competitive effects through these rebates. Furthermore, through recent case law, five conditions were identified to determine whether loyalty rebates violate Article 102 TFEU32. The EC analysis was found to be incomplete and insufficient to establish that the rebates were capable of having anti-competitive effects.
- In Qualcomm, the GC found that the EC did not take into account all of the factual circumstances, (e.g., that there were no technical alternatives to Qualcomm’s LTE chipsets for the majority of Apple’s needs during the infringement period).
- In Google, the GC concluded that the EC miscalculated the extent of the market covered by the contested practice, and that there were several errors of reasoning regarding the costs attributable to a hypothetically at least as efficient competitor, the assessment of the competitor’s ability to obtain pre-installation of its app, etc.
- In Unilever, the ECJ concluded that competition authorities must analyse whether, in light of all the relevant circumstances and the economic analyses produced regarding the ability of the clauses to exclude competitors which are as efficient as the dominant undertaking, those clauses are capable of restricting competition.
In short, the European court’s judgments have focused on the need to ensure that the analysis of alleged anti-competitive exclusionary rebates and revenue sharing in connection with exclusive purchase/supply obligations is rigorous. They don’t change the underlying law or require the identification of new forms of exclusionary conduct. The overarching theme is essentially the difficulty experienced by the EC in applying the AEC test (when it has chosen to do so), at least in part because of the information asymmetry between the EC and allegedly dominant entities.