The German Federal Supreme Court’s Judgment in Booking.com as a Case Study of the Limitations of Competition Law

The German Federal Supreme Court (BGH) held that the use of narrow price parity clauses (NPCs) between Booking.com and hotels infringed Art. 101 Treaty on the Functioning of the European Union (TFEU), confirming the Federal Cartel Office’s decision. The Supreme Court based its judgment mainly on the following grounds: (1) the use of these clauses restricts intra and inter-brand competition between hotels; (2) NPCs cannot be considered an ancillary restraint in the sense of Art. 101(1) TFEU; and (3) avoiding free-riding through these clauses cannot be considered as a justifying efficiency in the present case under Art. 101(3) TFEU. The present comment focuses on the analysis of the Court regarding (3) and argues that the reasoning in the judgment made conceptual mistakes and statements against settled EU law. The case shows the difficulties the Court had in handling a case where dynamic considerations are important. Therefore, even though the Court may have had valid concerns regarding NPCs, the judgment at hand is a step in the wrong direction regarding the judicial function of guidance of undertakings throughout the economy that wish to comply with competition law.

courts enforce its rules. The effects of signals that an administrative decision or court ruling send to firms in the economy are at least as important as the effects of correcting anticompetitive conduct in the specific market under investigation. 1 Under this lens of consistency, we will be discussing the German Federal Supreme Court's reasoning regarding efficiencies in the context of Art. 101(3) Treaty on the Functioning of the European Union (TFEU), which is one of the main aspects in the judgment, together with clarifications that the Court made regarding the ancillary restraints doctrine in the context of Art. 101(1) TFEU. Consistent results mean that courts apply the same legal and economic principles to similar circumstances to arrive at results that can be predictable to a reasonable degree. In the present case, this means that the restrictive effects and efficiencies attached to narrow price parity clauses (NPCs) should have been analyzed under the same economic principles that other comparable behavior is subjected to. This comment examines the reasons why the Federal Supreme Court failed at this task and how we can make sense of the judgment at hand. Section 2 provides a brief overview of the case while Sect. 3 offers contextual considerations. Section 4 takes a deep dive into the reasoning of the Supreme Court on efficiencies of NPCs. Section 5 concludes.

Case Summary
The case 2 concerns the use of ''most favored nation'' (MFN) clauses in the contracts between hotels and Booking.com. Broadly speaking there are two types of such clauses, broad and narrow ones. Broad price parity clauses restrict the ability of hotels to offer lower prices on other online platforms and their own online distribution channels. NPCs prohibit hotels from offering a lower price only on their own online distribution channels but are left free to do so on other online platforms. The ability to advertise lower prices on the hotels' online reservation portals is also restricted by both types of clauses. The hotels are free to advertise and offer lower prices through offline channels. The present case concerns the legality of NPCs.
The anticompetitive effect that the Federal Supreme Court associates to NPCs is the restriction of the ability to offer lower prices to consumers who book directly on hotel portals. According to the Court, this limits the ability of hotels to make lastminute offers, which are of particular importance in this market. 3 The clauses also reduce the incentives of hotels to lower the prices offered on competing platforms because that would mean a higher likelihood of diverting consumers away from direct sales -for which the hotel does not pay a commission -since the price on the hotel portal has to match the higher price on Booking. The main justification offered by Booking.com to include NPCs in the contracts with the hotels is to avoid free-riding. This could happen if, for example, a consumer uses Booking.com to search for accommodation options but then goes to the hotel portal to check whether a better price is offered and books through the hotel's website. In this scenario, Booking.com creates value mainly by reducing search costs and information asymmetries but does not earn a remuneration since it only makes money through a commission charged to hotels for bookings made on the Booking.com portal.
In succinct terms, the procedural history of the case is as follows: the Federal Cartel Office found NPCs to be in breach of Art. 101 TFEU. 5 The Düsseldorf Court of Appeal annulled the decision finding that NPCs were an ancillary restraint in an otherwise procompetitive agreement. 6 The Federal Supreme Court, on its part, reversed the lower court's judgment and agreed with the Federal Cartel Office regarding the overall anticompetitive character of the clauses in question.
This conclusion was based on two broad lines of reasoning: (1) NPCs could not be considered an ancillary restraint to an otherwise procompetitive or competitive neutral agreement, and therefore do not escape Art. 101(1) TFEU; 7 and (2) avoiding free-riding through the use of NPCs in the specific case at hand was not an efficiency in the sense of Art. 101(3) and therefore cannot be exempted on that basis. 8 The case shows the difficulties in separating the analysis made under paragraphs 1 and 3 of Art. 101 TFEU. In brief terms, Art. 101 TFEU imposes the following analytical structure: first, it has to be determined whether a conduct is captured by paragraph 1 because of its anticompetitive object or effects. If the answer is affirmative, then the examiner must evaluate whether the conduct can be justified by efficiencies that fulfill all requirements of paragraph 3. According to the Federal Supreme Court, at the first stage of analysis -when determining whether a conduct is captured by Art. 101(1) -there can be no balancing of effects, which should be done only at the stage of justifying efficiencies under Art. 101(3). 9 In determining whether paragraph 1 is applicable to a clause, the analysis has to be done at a more abstract level. 10 This way of framing the analytical framework is unclear but can be interpreted in the following way: a specific behavior can be considered to fall outside the scope of Art. 101(1) if it can be safely assumed that in most circumstances it will lead to procompetitive or competitive neutral results.
Another way of making sense of the structure of Art. 101 is through a framework of shifting of the burden of proof. 11 The analysis made under paragraph 1 can be compared to the requirement in US law for the plaintiff to make a prima facie case of anticompetitive effects. 12 Simply put, a plaintiff has to convince the court to a sufficient degree that the conduct can be anticompetitive so that the burden of proof is shifted to the defendant to show the efficiencies caused by the conduct that can render it either neutral or positive in terms of consumer welfare effects. This shifting of the burden naturally occurs only during the first instance where the record is developed. Higher courts can review the application of this rule based on the evidence that is already available.
Given these considerations, it is relatively uncontroversial that the Federal Supreme Court concluded that NPCs fall under Art. 101(1) TFEU 13 and decided to analyze the free-riding justification under the requirements of paragraph 3 -which is the main focus of the present comment. It is difficult to argue that the effects of avoiding free-riding clearly compensate the restrictions in price that NPCs bring about. This is an empirical question of effects. In addition, free-riding has an impact on the quality of service of Booking.com, which is hard to balance in the abstract against possible price effects of bookings and the intermediation commission of platforms.

National and International Context of the Judgment
Before taking a deeper dive into the conclusions of the Federal Supreme Court regarding efficiencies, it is useful to consider the legal and policy context in which the decision was issued. There are two sets of contextual considerations that can shed light on the Court's decision in Booking.com. The first one is antitrust policy trends in digital markets in Germany. The second one is the different responses that NPC's elicited in the European Union.
Regarding antitrust enforcement in the digital economy in Germany, it is safe to affirm that both the Federal Cartel Office and the Supreme Court seem to have taken a proactive stance against dominant tech companies. One good recent example is the Facebook case regarding abuse of dominance through the terms and conditions imposed on its users. Labelling the outcome of the case as proactive is of course a subjective take. On the other hand, it can be stated that the theory of harm in the case was novel, at least in the Federal Cartel Office's decision, which relied heavily on an infringement of the General Data Protection Regulation (GDPR). 14 This was slightly modified in the Federal Supreme Court's decision of 23 June 2020, but the Federal Cartel Office's decision was still confirmed. 15 Experience with antitrust law enforcement in Germany -including the Facebook case -and in the EU served as an inspiration to the 10th amendment to the German Act against Restraints of Competition (ARC), which came into force in January 2021. The amendment introduced ex ante regulation for undertakings of paramount significance for competition across markets. The regime is sketched in Sec. 19a of the ARC and requires a two-step process: first, the Federal Cartel Office has to designate the undertaking as one of paramount significance and, second, the authority has to identify which specific conduct will be prohibited from an exhaustive list in para. 2 of Sec. 19a. This ex ante regime is itself a precursor of the Digital Markets Act (DMA) adopted on 18 July 2022 in the EU. The Federal Cartel Office has been active in this field already designating Alphabet/Google, 16 Amazon, 17 and Meta 18 (formerly Facebook) as firms of paramount significance for competition across markets.
Regarding the international context of the Booking.com case, one has to take into consideration that the company was under investigation in a handful of EU countries. Therefore, there was an attempt at coordinating this effort within the framework of the European Competition Network. This resulted in the French, Swedish and Italian national competition authorities (NCAs) producing the same outcome: all three authorities ended their national investigations accepting the commitments proposed by Booking.com. Specifically, the company undertook to abolish broad parity clauses -which as mentioned earlier impede the ability to offer lower prices on competing platforms -but left NPCs intact. 19 Although the Federal 14 For a critical assessment of this theory of harm and the conflicts it causes with the enforcement of the GDPR see Colangelo and Maggiolino (2019). 15 Federal Supreme Court, 23 June 2020, Case KVR 69/19 -Facebook (interim decision). For a deeper analysis and comparison of the theory of harm used by the Federal Supreme Court and the Federal Cartel Office see Mackenrodt (2020). 16 Federal Cartel Office, press release 5 January 2022 ''Alphabet/Google subject to new abuse control applicable to large digital companies -Bundeskartellamt determines 'paramount significance across markets''' https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2022/05_01_ 2022_Google_19a.html?nn=3591568. Accessed 12 August 2022.
Cartel Office participated at some point in this coordination, it did not share the same opinion and as is known it held NPCs to be anticompetitive in the specific case at hand. Soon after, the issue was taken up by legislators in France, Austria and Italy, who basically agreed with the Federal Cartel Office. In France, all forms of price parity clauses in the hotel industry were outlawed by Art. L311-5-1 of the French Tourism Code, adopted on 8 August 2015 as part of a set of economic growth measures known as the 'Macron Law'. Austria followed suit in 2016 20 and Italy in 2017. 21 It is noteworthy that the French and Italian Parliaments took a stricter stance than that of their respective competition authorities. 22 Another important part of the international context story is that after the decisions of the mentioned competition authorities, there was a monitoring study conducted under the coordination of the ECN, which included 10 competition authorities, 23 to analyze the effects of changes in parity clauses. The study used data from a survey of hotels and prices gathered through web scraping tools, mainly to determine the extent to which price differentiation existed among different sales channels. 24 The study's results leave most issues open, especially after acknowledging the limitations of the survey and web scraping data. 25 As is known, the Federal Cartel Office conducted nonetheless its own study as it was ordered to do so by the Düsseldorf Court of Appeal. This study showed a greater extent of price differentiation in Germany after the abolition of NPCs in 2015 regarding offerings on booking platforms and on the hotels' own websites. 26 Finally, as has been noted elsewhere, 27 another important international development in this context is the treatment of NPCs in recent EU legislation passed after the Federal Supreme Court's judgment.  29 The DMA, on the other hand, does include a blanket prohibition of NPCs for gatekeepers (Art. 5(3)).
These two legislative outcomes can be explained from a practical perspective. The Commission -which appears to have a more lenient view towards NPCs as shown by the mentioned coordinated effort with the French, Italian and Swedish NCAs -issued the VBER and drafted the DMA proposal. The latter originally excluded NPCs from the catalogue of prohibited behavior. The final version adopted by the Council of Ministers, which did include the NPCs prohibition, was the product of a long legislative debate.
In the abstract, these two different treatments of NPCs can be nonetheless consistent with each other. It appears that concerns regarding NPCs grow along with the size of the undertaking. Under the VBER, for any undertaking with a market share of over 30%, NPCs will be subject to an effects analysis. The DMA's absolute ban, on its part, may be due to the belief that such conduct by gatekeepers (which will tend to be some of the large firms in the EU) can be presumed to have net harmful effects. Although the concept of gatekeeper is not exactly a variant of dominance or market power, there is some relationship. Gatekeepers, in general, can be considered as undertakings with enhanced power over market conditions, which is one of the rationales of the legislation. 30 The main takeaways from these contextual considerations can be summed up as follows: (1) given antitrust policy trends in digital markets in Germany the result achieved by the Federal Supreme Court does not come as a surprise; and (2) the development in the rest of the EU and at the supranational level has not been uniform, to say the least. Therefore, the big question going forward is what other digital intermediators can expect from the German authorities -and even beyond if the developments in this jurisdiction have a transnational influence. Part of the answer to this question lies in the signals that the efficiencies analysis of the Court has sent to undertakings operating in the German market.

Free-Riding as an Efficiency Justification in the Eyes of the Federal Supreme Court
The Federal Supreme Court correctly enumerates the four requirements of efficiencies defenses under Art. 101(3) TFEU concluding that the first one is already lacking: 31 that the agreement contributes to improving the production or distribution of goods or to promoting technical or economic progress. This is problematic in view of the nature of free-riding. As a procedural rule, the Federal Supreme Court analyzed this defense with a favorable view towards Booking.com regarding the evidence on the portion of consumers that use the platform to search for accommodation options and then go to the website of their hotel of preference and book there if the price is lower. 32 At this general level of reasoning, it is undeniable that a clause that reduces this behavior improves the production of a good. If a firm creates value but a part of it is appropriated by another economic agent, the firm has fewer incentives to produce such value. This will result in a sub-optimal amount/quality in the market. 33 In other words, Booking.com creates a positive externality for hotels and travelers for which it does not get compensated. It is fair to say that these are well-settled and basic aspects of the economic doctrine underlying competition law analysis. 34 The Federal Supreme Court framed the issue of whether Booking.com obtains commission revenue under the free-riding scenario as one of contractual fairness. 35 As such it considers it to fall outside the scope of competition law analysis. However, as explained above, this conclusion goes against well-settled economic principles. This makes the Court's reasoning harmful in terms of consistent law enforcement and the signals sent to economic agents throughout the economy. If firms cannot rely on courts applying principles around which a consensus exists, then the perception of uncertainty regarding judicial interpretation of rules increases.
At this point, it is important to clarify that this opinion does not argue that Booking.com should have been acquitted because avoiding free-riding reduces, in the abstract, inefficiencies caused by firms not being compensated for positive externalities that they produce. 36 What is argued here is that framing the issue as one of contractual fairness was wrong. After accepting that free-riding leads to inefficiencies, the inquiry should have proceeded to establish the extent of its occurrence based on concrete evidence. 37 This should have been balanced against the quantified anticompetitive effects of NPCs 38 to determine whether the first requirement of Art. 101(3) TFEU is met.
The Federal Supreme Court was concerned with the abstract nature of the freeriding defense, 39 and with reason. As the argument goes, Booking.com alleged that the reduced commission revenue caused by free-riding customers and hotels reduces 32 Id., para. 72. 33 Mankiw (2015), p. 199. 34 The analysis of externalities is covered by foundational textbooks in microeconomics. See Mankiw (2015) and Dixit (2014) as good examples. 35 Id., paras. 75-76. 36 Kathuria and Mackenrodt (2021), p. 6 (''a free-riding possibility does not provide a carte blanche to the parties''). 37 As explained below, this comment argues that the survey conducted by the Federal Cartel Office may have not been the best empirical strategy to gauge this phenomenon. 38 The Federal Supreme Court was content to establish harm by the possibility of price restriction and the importance of last-minute bookings for hotels. There was no direct quantification of the intensity of these two phenomena. Therefore, the question arises: How can a possibility, with no estimated amount, be balanced against potential or quantified efficiencies? A balancing exercise, properly speaking, can only exist if one has two amounts on the same units, say revenue lost by the platform (€) against over-price paid by consumers (€). 39 Id., paras. 60 and 87. The Federal Supreme Court takes issue with the specific mechanism at work: that lower revenues lead to less investment capital that may or may not be used to improve the service. the investments made on improving the service or expanding the platform's capacity. Court's should by no means be persuaded by abstract arguments, which may be true to a greater or lesser extent in practice. The inquiry should always be supported by evidence.
As a quick side note, however, the abstractness of such a defense is inherent to any dynamic consideration in competition law analysis, on which courts arguably do a worse job compared to the assessment of short-term price effects. 40 The possible efficiency that is discussed in the Booking.com case is prospective. The increased compensation to the platform will be converted to product improvement. This is not much different than assessing innovation effects in markets with intensive R&D investments, which is not an easy task to undertake. 41 However, it is an important point on which further academic and practice development is needed since innovation and other dynamic considerations may give antitrust law more relevance in digital markets.
In the eyes of the Federal Supreme Court, the price restriction of NPCs appeared clear and the potential benefits more abstract, which is a justifiable point to make. However, instead of reversing the judgment, a better option would have been to remand the case to further develop the record on the alleged competitive benefits of NPCs. This could be justified because the intensity of the efficiencies of the clauses in question was not established and this is an essential fact in order to conduct the balancing that Art. 101(3) TFEU requires. The Court did not consider this option explicitly. However, it seems like the performance of the market in the period after the Federal Cartel Office invalidated NPCs weighed heavy in the Court's mind. Given that there was evidence that the revenue of Booking.com grew in this period, as well as its base of hotels and market share, 42 the Court considered that NPCs were not necessary to maintain or improve the platform's services.
The Federal Supreme Court made a conceptual mistake in the conclusions it drew from the mentioned market performance trends. Because Booking.com was enjoying a growth period, there was no indication that the abolition of NPCs and the free-riding problem were considerable. The Court admitted that this growth might have been in part due to strong market demand. However, it was categorical in stating that for the purposes of Art 101(3) TFEU it was irrelevant whether Booking.com would have grown more in a counterfactual scenario where it would have been able to keep the NPCs in the contracts with hotels. 43 The use of counterfactuals to analyze the net effect of a conduct is a settled principle in Art. 101 TFEU case law, 44 which has a sound basis in economic analysis. In view of this, the reasoning of the Court can also be potentially harmful in terms of the signals it sends to the rest of the economy and in terms of consistent application of competition law.
40 McGowan (2001); Kokkoris and Valletti (2020);and Spulber (2022). 41 Kokkoris and Valletti (2020 One could justify the assertion of the Federal Supreme Court on the relevance of counterfactuals if one takes it out of context and in a literal meaning. It is true that the growth of the revenue of the undertaking itself is not important for the analysis of efficiencies. The focal point should be the actual benefits that can be transferred to consumers. However, the argument in the case was never that NPCs were needed for Booking.com to grow but that by avoiding free-riding they increased the incentive to invest in maintaining and improving the added value to consumers.
According to the Court's reasoning on market performance, one could conclude, for example, that NPCs were pro-competitive if there would have been evidence that accommodations demand grew during the period of validity of the clauses. That would have also been a mistake. The important fact to determine would have been the performance of market demand in an alternative scenario. The relevant evidence for this is not overall market performance but the isolation of the effect of NPCs.
The question is how one estimates the ceteris paribus effects of NPCs. This was another flaw in the case, which is attributable more to the Federal Cartel Office and the Düsseldorf Court of Appeal than to the Federal Supreme Court. To estimate the effects of the abolition of NPCs, the Federal Cartel Office decided to conduct a survey on hotels and booking platforms. This is somewhat puzzling since it seems like there was a perfect opportunity to conduct a statistical exercise using more objective measures. Specifically, the Federal Cartel Office could have instead required information from booking platforms and a sample of hotels on revenue per sold room or other information that captures price behavior more objectively. There could have also been an effort to measure free-riding more directly -for example, by measuring the effect of price differentials on bookings in the hotel's direct online channels. This and other variables could have been used in more robust estimation methodologies using econometric models. In fact, after the Booking.com saga throughout the EU, there was no lack of attempts at such empirical estimation exercises, which was done mostly using web scraping methods to obtain information on prices. 45 The Federal Cartel Office could have obtained much more quality data had it used its powers to request information in this context. The German competition authority did not state in its study whether it considered methodologies other than the survey or the reasons why the latter was better suited than other empirical methods.
The Federal Supreme Court states that the Düsseldorf Court of Appeal did not discuss Booking.com's objections to the Federal Cartel Office study, 46 which is not surprising given the fact that the first instance court absolved the undertaking based on its interpretation of the ancillary restraints doctrine. Therefore, there is very little that the Supreme Court could have done given the applicable standard of review at that instance. However, the point is noteworthy going forward. It is important that in future cases there is a more explicit analysis on the best way to estimate effects that are relevant to an antitrust inquiry.
The last point of the analysis of efficiencies on which this opinion focuses is the distinction that the Federal Supreme Court draws between the effects of NPCs and 45 See Hunold et al. (2018) and Mantovani et al. (2021). 46 Booking.com, para. 72. those of selective distribution systems. 47 According to the Court, the latter may restrict competition between dealers in order to reduce the risk that special services are no longer provided by dealers whose prices are undercut. The Court labels such services as benefits that go beyond the free-riding problem. This is also a mistaken application of the analytical framework.
It is clear that, to the extent that there is free-riding in the accommodation booking market, the continuation or improvement of the special services of Booking.com are at risk -although evidence is required to establish the extent of such a risk. From a competition law point of view, there is no reason to treat a potential decrease in quality differently from the discontinuation of a given service since both reduce consumer welfare in the counterfactual scenario. Therefore, the risk described by the Federal Supreme Court in the case of selective distribution systems applies as well to the free-riding problem that booking platforms face. In fact, the Court lists a number of special features/services of Booking.com for which it does not perceive a remuneration unless it generates an intermediation commission. However, the Court interprets these services as the foundations of the undertaking's network effects and market power. 48 The Court does not consider that the presence of network effects does not exclude the possibility of free-riding. A platform can still attract users because of the value that it creates but does not appropriate.
What would have been a more valid concern is whether avoiding free-riding and the associated risk of losing or deteriorating Booking.com's special services effects compensate the price restriction of NPCs, which is briefly analyzed by the court in the last part of the judgment. 49 This is why the preservation of special services seldom justifies resale price maintenance (RPM), which under EU law is a restriction by object, presumed to be particularly harmful to competition, even if one acknowledges that RPM does indeed avoid free-riding in a distribution system.
The Federal Supreme Court was more lenient in its evaluation of evidence on anticompetitive effects than on efficiencies. This can be affirmed based on the following: the Court was satisfied with the existence of potential anticompetitive effects of the clauses -as explained, there was no estimation on the actual effect on prices of the abolition of NPCs -but in the case of efficiencies required more concrete evidence. It is true that NPCs have the potential to restrict price differentiation between platforms because of the mechanism identified by the Court: 50 if a hotel charges a lower price on another platform, it does not necessarily divert a customer from Booking.com but possibly from its direct online sales channel since the latter will have to be priced at the same level as the price offered on Booking.com. However, since on a theoretical level NPCs can have both restrictive and pro-competitive effects, the question is an empirical/evidentiary one. This is an additional reason why the judgment under analysis sends the wrong guidance for future compliance and litigation. 47 Id., para. 79. 48 Id., para 82. 49 Id., paras 80-87. 50 And elsewhere in the literature. See Kathuria and Mackenrodt (2021).

Conclusions
The Booking.com judgment of the Federal Supreme Court is a part of the story of antitrust enforcement in the digital economy in Germany. It shows a proactive and strict stance from the Federal Cartel Office and the Court, which has been echoed in recent legislative changes, culminating in the 10th amendment of the German ARC. On the other hand, it is a case study of coordination efforts at the EU level within the European Competition Network and the effects of inconsistent approaches between and within countries. Judicial guidance in the digital economy is of great importance given that principles of analysis are not as well settled as is the case in more traditional markets, such as production and distribution of commodities. The Court appears to signal that when the market and the investigated undertaking are growing, there is little room to argue for compensating efficiencies of an anticompetitive practice. Although this result is not per se bad policy in the concrete case, it can be dangerous if it is reached through a flawed reasoning regarding the nature of an economic phenomenon -in this case, free-riding -and the evidentiary standard to prove harm and justifying efficiencies. A correct reasoning is as important as the outcome given the guiding function of court decisions. The Court did not fulfill this task in the Booking.com judgment.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by/4.0/.