On 22 February, the District Court of Amsterdam decided to refer preliminary questions to the Court of Justice of the European Union (CJEU) in order to receive guidance enabling it to assess the legality of the EU antitrust rules of price parity clauses (PPs) used by Booking.com, an online travel agent platform (OTA).* More specifically, the Amsterdam District Court hopes to receive clarity as to i) whether broad and narrow PPs may be considered ancillary restrictions in the context of Article 101(1) TFEU, and ii) how the relevant market should be defined where it concerns transactions which are being mediated by an OTA.
PPs under scrutiny
In the proceedings before the Amsterdam Court, Booking.com is requesting a declaration that it did not infringe Article 101(1) or Article 102 TFEU by including in its contracts with German hotels that feature(ed) on its platform ‘wide’ price parity clauses (WPPs) and ‘narrow’ price parity clauses (NPPs). Conversely, the German hotels are seeking a ruling that Booking.com has infringed said Treaty provisions by including the PPs.
Based on the WPPs, used by Booking.com until 2015, the German hotels were obliged to offer Booking.com the best price they made available for a hotel room. The German hotels were hence precluded from offering more favourable prices to other OTAs or via other sales channels, including direct online or offline sales by the individual hotels. In 2015, Booking.com changed the WPP to a NPP. The NPP obliged the German hotels to offer Booking.com no less favourable conditions – including price – which they offered elsewhere, except for other OTAs or via offline sales channels. Although the German hotels were, thus, able to offer more favourable prices to other OTAs and via offline sales channels, they were still obliged to offer Booking.com no less favourable conditions than those used on their own websites.
Both types of PPs have been scrutinized by various national competition authorities (NCAs). In 2013, the Bundeskartellamt (BKA) found that the use by HRS – another OTA – of WPPs infringed Article 101(1) TFEU. The Oberlandesgericht Düsseldorf confirmed this finding. In 2015, the French, Italian and Swedish NCAs accepted commitments from Booking.com that entailed changing its WPPs to NPPs, which would apply solely to conditions offered by hotels through their own direct online sales channels. In the same year, however, the BKA found that the use by Booking.com of NPPs also infringed Article 101(1) TFEU. Although a Swedish first instance court found that Booking.com’s NPPs infringed the EU antitrust rules, a second instance court overturned this ruling. Conversely, in Germany, the BKA’s Booking.com decision was quashed by a first instance court, only to be later confirmed by the Bundesgerichtshof (BGH).
Currently, Belgium, France, Italy and Austria have laws in place prohibiting the use of both WPPs and NPPs. Unsurprisingly, PPs became a subject to be discussed by the new vertical block exemption regulation (VBER). Article 5(1)(d) of the ‘new’ VBER excludes WPPs – more particularly, across-platform PPs – from the block exemption. Other types of PPs may benefit from the block exemption, provided that the VBER’s market share thresholds are not exceeded. What is more, Article 5(3) of the Digital Markets Act prohibits gatekeepers from using both WPPs and NPPs. One could therefore say that views on the legality of PPs, especially NPPs, seem to differ throughout the EU.
Nonetheless, the Amsterdam District Court limited its preliminary questions to the issues of ancilliarity and the delineation of the relevant market in view of the application of the ‘old’ VBER.
Application of the ancillary restraints doctrine
With its first question, the Amsterdam court inquires whether the PPs fall within the concept of ancillary restraints. This concept covers any restriction of competition which is: i) directly related and ii) necessary to the implementation of a ‘main operation’ which is neutral or even positive as regards its effects in terms of competition. If the PPs are directly related and necessary to achieve the main operation, the compatibility of the PPs must be examined with that of the main operation. Where the main operation does not fall foul of Article 101(1) TFEU the same will hold true for the PPs. Similarly, if the main operation is a restriction within the meaning of Article 101(1) TFEU but benefits from an exemption under the clause of Article 101(3) TFEU, the same is true for the PPs.
The first requirement ‘directly related’ demands that the PPs are subordinate to the implementation of the main operation and have an evident link with it (M6, para 105), whereas the second requirement regarding the necessity of the restrictions implies a two-fold test according to which: i) the PPs must be ‘objectively necessary’ to the implementation of the main operation and ii) ‘proportionate’ to the objectives of the main operation (Hoffmann-La Roche, para 69).
The EU Courts, especially since Mastercard, follow a narrow interpretation of the ancillary restraints doctrine and of the notion of ‘objectively necessary’, in particular. This strict approach has been justified by two issues. The first is that too broad a scope of ‘objective necessity’ would enable too many restrictions of competition to fall outside Article 101(1) TFEU, impairing the effectiveness of this prohibition. The second issue is that allowing restrictions that are essentially based on efficiency considerations to fall within the ancillary restraints doctrine would be contrary to the structure of Article 101 TFEU. According to this article, efficiency considerations are dealt with by application of 101(3) TFEU and not under Article 101(1) TFEU (EU competition law does not recognise a US-style rule of reason within Article 101(1) TFEU, see Generics para 104).
The EU courts have limited the notion of ‘objectively necessary’ in two ways. Firstly, a restriction of competition will only fall within this notion if it is not possible to dissociate the restriction from the main operation without jeopardising the existence of that operation. Solely restrictions absent of which it would be ‘impossible’ to carry out the main operation can be considered objectively necessary. Restrictions must be ‘strictly indispensable’ (Hoffmann-La Roche, paras 71-72). Put differently, the restriction of competition must constitute a conditio sine qua non to the main operation (AG Cosmas’s Opinion in Masterfoods, para 88 and Mastercard, para 94). Importantly, the fact that the main operation is more difficult to implement or even less profitable absent the restriction (i.c. the PPs) is not enough (Mastercard, para 91).
The requirement of ‘objectively necessary’ implies that there is a particular issue that could endanger the existence of the main operation. The issue can be dealt with by restricting the commercial autonomy of a party to the agreement in question which may involve a restriction of competition. This restriction is, however, essential to address the issue that endangers the main operation (similarly AG Rantos’ Opinion in EDP, para 104). The relevant question that needs to be answered in the proceedings before the Amsterdam Court appears to be whether Booking.com would or could no longer engage in the main operation if it cannot use the WPPs and/or the NPPs.
Secondly, restrictions of competition that are necessary for efficiencies to arise cannot be considered objectively necessary in the context of Article 101(1) TFEU. Instead, these restrictions must be assessed based on Article 101(3) TFEU. AG Mengozzi noted (para 85) that “it is in the context of the latter provision and not in the context of Article 101(1) that restrictions that make it possible to implement the main operation, improve its efficiency or ensure its commercial success and, in general, those that are ‘indispensable’ in view of the competitive situation on the market, must be taken into account”. The notion of ‘objectively necessary’ therefore does not involve an analysis of whether absent the restriction of competition, the main operation is commercially effective but whether that operation is possible in the first place.
In the proceedings before the Amsterdam Court, the greatest uncertainty as to the application of the ancillary restraints doctrine seems to reside in the assessment of the second requirement. Booking.com argues that the PPs are necessary because they are required to combat the free riding-problem. This problem may occur where hotels while making use of the Booking.com platform, offer their hotel rooms against more favourable conditions on other OTAs or their own website. As a consequence, consumers divert to websites other than Booking.com and the OTA misses out on the commission it receives when consumers book a room via its platform.
As noted, the assessment of the objective necessity of the PPs requires answering whether Booking.com will or can no longer engage in its main operation if it cannot use the WPPs and/or the NPPs. Since Booking.com argues that the PPs are necessary to combat the issue of free riding, this will include analysing whether indeed such a problem exists. It must be established whether consumers first visit Booking.com and, subsequently, book the hotel room of choice on another OTA or on the hotel’s own website. In addition, it must be established that free-riding poses a significant problem to Booking.com (e.g. given the sheer number of customers leaving Booking.com for other OTAs or the hotels’ own website). Only if the free riding-problem makes it ‘impossible’ for Booking.com to carry on its main operation, the PPs, provided that they address this free riding-problem, may be considered objectively necessary. Since only in such a case will the PPs constitute a condition sine qua non to the main operation. In that respect, the BKA and the BGH have doubted the existence or the size of the free riding-problem. Moreover, it will be interesting to see in this regard whether Booking.com’s main operation has become impossible in the Member States where PPs have been prohibited.
If indeed a free-riding problem exists, it must be assessed whether the PPs address this problem so that they can be considered objectively necessary. Furthermore, it must be analysed if the PPs are proportionate or whether there are less restrictive alternatives that may address the free riding-problem equally well. Although one may argue that WPPs are not objectively necessary – this may, for example, be inferred from the fact that Booking.com has agreed with several NCAs to limit its WPPs in the future to NPPs indicating that WPPs are not strictly indispensable to the main operation – to argue the same for NPPs seems slightly more difficult. Nonetheless, the BGH concluded that NPPs cannot be considered objectively necessary. Furthermore, it follows from the new Guidelines to the new VBER that the Commission does not consider that the ancillary restraints doctrine is applicable to PPs as they constitute efficiency justifications. Paragraph 372 notes that “where retail parity obligations produce appreciable restrictive effects, possible efficiency justifications need to be assessed under Article 101(3) of the Treaty” and that such efficiency justifications include “the use of retail parity obligations by providers of online intermediation services […] to address a free-rider problem”.
Looking for the relevant (product) market
It follows from its second question that the Amsterdam Court is uncertain as to the definition of the relevant (product) market. More particularly, it is uncertain whether this market includes solely OTAs or whether it must define according to a broader perspective so as to include the hotels’ own sales channels, both online and offline (i.e. hotels’ websites as well as, for example, sales channels such as telephone). In short, the Amsterdam Court is uncertain about whether and which sales channels must be included in the definition of the relevant market. Booking.com argues that the relevant market is the market for booking and distribution of travel accommodation. The different sales channels (online and offline) form one relevant market as they are substitutable for both hotels and travellers (equally). The German hotels argue that the relevant market is for OTAs since solely OTAs offer the combination of searching, comparing and booking.
From the outset, the (potential) expectations of the Amsterdam Court with regard to receiving a ready answer from the CJEU to its second question should be tempered. Delineating the relevant market requires a fact-intensive investigation, one for which the CJEU is ill-equipped. The CJEU will presumably not spell out a relevant market definition that the Amsterdam Court can readily use. The preliminary reference procedure is meant for questions on the interpretation (or legality) of EU law. Consequently, fact-heavy questions are often left for national courts to decide. This does not mean that the CJEU will not address the question. It may reiterate and clarify the analysis that should be deployed when defining the relevant market in a situation such as the one in this case.
The Amsterdam Court’s doubts can be traced back to the intricacies of online platforms. Notwithstanding that already quite some ink has been dedicated to it, the issue of how to define the relevant market when dealing with online platforms is still not fully settled. What makes this a problematic issue is, inter alia, the often two- or multisided nature of online platforms. For example, OTAs provide, on the one hand, particular services to consumers (i.e. the platform users that are looking to book a hotel room). These include price comparison, search service and product review. On the other hand, OTAs provide hotels with an attractive showcase and ensure that the hotels can transact with a large number of consumers. OTAs have a two-fold function. The value of these platforms lies in the facilitation of transactions among their different customer groups. The existence of these different functions complicates establishing the relevant market.
In view of these difficulties, it might be advisable to first establish whether one is dealing with one or multiple potentially relevant markets. In order to do so, one can assess the substitutability of the OTA concerned – which constitutes the relevant point of reference for assessing substitutability – from the perspective of every customer group separately (i.e. the consumer, on the one hand, and the hotels, on the other). If substitutability for the customer groups differs, it might be that there are multiple potentially relevant markets. In such a case, there is a market catering to the particular demand of the hotels which is distinct from the market catering to the demand of consumers. Whereas the agreement between Booking.com and the hotels containing the PPs must then be placed in the context of the former market, the competitive relationship between Booking.com and the hotels will be found in the latter. In such a scenario it is not excluded that the (potentially) anticompetitive behaviour (i.c. the PPs) takes place in a different market from where the anticompetitive effects occur. Consequently, it is possible that Booking.com and the hotels are competitors in a market different from the market in which they concluded the agreement containing the PPs. Hence, Booking.com and the hotels may not be in the same upstream market despite the fact that there is horizontal competition between their sales channels in a downstream market. Booking.com seems to argue that, since substitution is similar for hotels and consumers, there is only one market. It, however, implicitly follows from their analyses that the BKA and the German courts take the view that there are multiple markets, including a distinct (upstream) market that caters for the hotels’ demand. This is because when establishing the relevant market, the BKA and the German courts predominantly (or even solely) considered the substitutability of different sales channels from the point of view of the hotels and their particular needs.
If multiple potentially relevant markets are found, one is forced by the VBER to pick one of these markets for its application. Especially as regards its market share threshold. Article 3(1) VBER notes that the exemption “shall apply on condition that the market share held by the supplier does not exceed 30 % of the relevant market on which it sells the contract goods or services and the market share held by the buyer does not exceed 30 % of the relevant market on which it purchases the contract goods or services”. The VBER requires an investigation of the market on which the undertakings involved act as buyer and seller. In case one considers that there is a market that caters for the demand of hotels which is distinct from the market that caters for the demand of consumers, that former market must be analysed for the application of the VBER’s market share thresholds. It will be on this market that the hotels act as buyers and the OTAs as suppliers (indeed, according to article 1(d) of the new VBER ‘supplier’ will include the provider of intermediation services).
Article 101(3) TFEU?
What is interesting furthermore, aside from the preliminary questions, is what is absent from the preliminary reference. The Amsterdam Court did not refer to a question on whether WPPs and/or NPPs can benefit from an exemption based on Article 101(3) TFEU. This will become relevant if it is concluded that WPPs and NPPs are restrictions within the meaning of Article 101(1) TFEU and that they are not ancillary nor that they can benefit from the VBER. The German courts have found that neither WPPs nor NPPs used by an OTA could benefit from an individual exemption.
The preliminary questions posed by the Amsterdam District Court to the CJEU will allow the Luxemburg court to settle, to a certain extent, the discussion with regard to PPs. Additionally, it will allow the CJEU to provide relevant guidance, both with regard to the application of the ancillary restraints doctrine as well as to how the analysis of the relevant market in case of online platforms must be conducted. The CJEU’s answer with respect to the ancillary restraints doctrine could be particularly salient since a finding by the CJEU that WPPs or NPPs can be considered ancillary restraints would potentially contradict the German judgements on the topic as well as affect the (application of the) new provisions introduced by the VBER.
* After checking (17/04/2023) the curia database, the request for preliminary questions seems not to have been registered or processed yet at the CJEU.