Roadblocks – when should the essential facilities doctrine apply in EU competition law?

Historically, EU competition law has allowed dominant firms to refuse to deal with competitors except in very specific circumstances concerning “indispensable” infrastructure. This respect for freedom of contract and property rights reflects the essential facilities doctrine which was established in EU law in Bronner. Limited exceptions have been carved out over the years, but important developments in the EU courts threaten to significantly limit its application to digital platforms.

The most notable development was the well-publicized and clear judgment in Google Shopping. The CJEU (the highest EU court) found that the doctrine did not apply to “self-preferencing” – where access to infrastructure is granted but subject to discriminatory conditions. 

Another development last week flew under the radar but would go a few steps further. AG Medina’s opinion last week in Alphabet (Android Auto) suggests a more restrictive reading of Bronner.

In summary, AG Medina argues that: 

• The test set out in Bronner should not apply to “open” digital platforms;
• a refusal or delay by an open platform to provide access to a potentially competing developer is capable of anticompetitive effects; and
• a refusal or delay cannot be objectively justified by the time and cost required to develop a new template for developers. Android Auto operates based on “templates” created to enable developers to create their applications. Similar to an application programming interface (API), templates require investment to ensure the interoperability of the application with the platform technology.

What determines the applicable legal test in access related cases?

Citing Slovak Telekom and Baltic Rail, Medina concludes that Bronner is only relevant for cases where infrastructure has been developed and reserved for the dominant undertaking’s own use. She contrasts that with Android Auto, which is characterized as “deliberately open” to attract “as many ‘in-car apps’ as possible”. AG Medina argues that the logic underpinning the essential facilities doctrine does not apply as requiring access does not undermine the incentives to invest of an open platform (which benefits from more apps). 

This goes further than the Google Shopping judgment. There, the CJEU distinguishes from Bronner based on the conduct – i.e., a refusal of access is distinct from subjecting access to discriminatory conditions. There is no need to force access to infrastructure, as that access has already been granted. In that context, the CJEU found that there is less harm to freedom of contract. 

AG Medina instead distinguishes from Bronner essentially based on the nature of the platform – i.e., a refusal of access should be subject to a different legal test where a platform is otherwise “open”. But platforms may have varying degrees of openness. In this context, more applications do not invariably mean a better platform. Forcing a platform operator to facilitate a particular service on their platform still impinges on freedom of contract and impacts their ability to increase platform value. AG Medina flirts with the idea that the absence of an existing interface for EV charging apps might have reflected a wish to reserve those services but notes that Google did not raise the argument. 

Such a binary approach risks neglecting the long-term incentives and fundamental rights at play here. There are good reasons why the “open” or “closed” nature of a platform should not determine the applicable legal test. 

Capability and incentive to foreclose

AG Medina finds that anticompetitive effects can arise from the delayed launch of an app which is at least “potentially in competition with a product or service capable of being provided by the dominant undertaking”. 

This is a very wide net. What is a dominant company “capable” of providing – must there at least be concrete plans to develop a similar service? Or is it sufficient that they might plausibly enter and, if so, over what timeframe?

The ambiguity as it stands risks blurring the line between self-preferencing and more general refusals to supply. If a dominant undertaking is not active, nor has plans to enter, the downstream service, it is much less likely to have an incentive to foreclose downstream rivals. We should accordingly be cautious in assessing the two dissimilar situations in the same way. 

The reliance on objective justification

Having determined that the Bronner conditions for refusal of access should not apply, the AG argues that the “actual focus” of the analysis should be on whether there is an objective justification for the delayed access. 

She finds that the lack of an existing template for the specific needs of the developer would not objectively justify the refusal of an access request. A refusal would only be objectively justified where designing such a template is (a) impossible, or where access could (b) affect the performance of the platform or (c) run counter to the platform’s economic model or purpose. 

She considers that time or resource constraints also cannot justify an outright refusal to provide access. Before refusing access, the dominant company should indicate development time and can require the requestor to contribute to the costs of development.

This would be a high bar for dominant firms to prove that they need not mandate access. It risks entangling competition law in negotiations on acceptable development costs and their allocation. 

Is this a workable framework?

The AG’s opinion is potentially far-reaching. While she recognises that the reasoning behind Bronner is that forced access is “particularly detrimental for freedom of contract”, the AG suggests a system of access which would require platform operators to implement tailored requests to develop interfaces from (a wide range of potentially competing) developers in consideration for a monetary contribution. 

Much like…a contract?