Contract terms up and down the supply chain have included practices such as resale price maintenance; most-favoured customer terms; exclusive contracts and conditional discounting; and tying and bundling, all of which have caught the attention of antitrust regulators in the US and the EU for many years. While these vertical restraints have been of concern to antitrust regulators because of the potential to inhibit competition, that scrutiny has been tempered particularly in the US because these restraints may also result in measurable benefits for customers, which can significantly complicate and sometimes outweigh potential anticompetitive effects. However, the use of these tactics in e-commerce settings, and the increasing importance and dynamism of e-commerce in the economy, have attracted renewed interest from regulators.
In this chapter, we first define ‘e-commerce’ and how it may differ from traditional commercial transactions. We then review vertical relationships and restraints in e-commerce and the challenges in evaluating them by traditional competition models. Explicitly defining markets, evaluating marketing power, and identifying the pro- and anticompetitive effects embedded in these vertical relationships all require bespoke tools and approaches. These issues will be specifically explored in the context of two recent cases involving e-commerce: Apple e-Books and In re: Online Travel Company Hotel Booking Antitrust Litigation.
What is E-commerce?
E-commerce applies technologies including computing power and the internet to reshape how firms interact with their suppliers and customers. These tools are nearly universal in many parts of the world (the US, EU and much of Asia) and widespread almost everywhere. Consumers and businesses alike have unprecedented access to product and service information online, and can conduct transactions remotely. E-commerce business models fall into three categories, each of which has different market economics.
Tangible good models include products that require some form of physical delivery or presence for consumption. Many tangible products offered online are identical to those found on the shelves of brick-and-mortar stores – the only difference is that the product can be researched and purchased online. Amazon is the most prominent firm using this model, delivering tangible products, most of which are also available through traditional retailers across the globe (initially a business-to-consumer model or B2C). Notably, Amazon began as an online distributor of books but quickly expanded its business to offer a nearly infinite array of other tangible goods sold to customers through the internet. Subsequently, Amazon has also become a business-to-business (B2B) service provider offering cloud computing and redistribution services to businesses, as well as a traditional brick-and-mortar retailer with Whole Foods grocery stores and product showrooms.
Intermediary models provide services to customers (often facilitating the purchase of other goods) through an online interface that changes the role of the intermediary with respect to understanding and acquiring the good or service. The online interface in an intermediary model often replaces a human agent, with resulting benefits and challenges for the user experience. For example, travel and hospitality services can be procured by paying a travel agent, or by going online and using Orbitz. Ride-sharing services like Uber or Lyft organise the vehicles for hire differently and offer consumers a different way to call for a ride than a traditional taxi service. In both cases, the consumer’s goal in engaging with either service (travel agent vs online booking service, cab company vs Lyft app) is the same, but the service itself is fundamentally different.
Online products are wholly purchased and consumed online. For example, many forms of media content – video, music, books – can be accessed and consumed directly online. Because the cost of maintaining and distributing this content is often low, and because content creation may occur through separate third parties upstream, the value offered by these firms is often in aggregating content customers want in one location, pressuring providers to offer ‘must-have’ content and often generating revenue through advertising rather than subscription or pay per view, read or listen. For example, music streaming service Spotify connects consumers with content providers but also connects advertisers with consumers. It generates revenue from subscriptions for its ad-free premium services but it also generates revenue from advertisers on its free service. It then pays royalties to the content providers.
Altogether, the different forms of e-commerce make up a growing proportion of the US economy for both B2C and B2B markets. In B2C markets, some reports show that approximately 95 per cent of all Americans shop online.In the US, online sales comprise about 9 per cent of total retail sales, which lags behind many other countries; for example, online sales account for nearly a quarter of total retail value in China.
E-commerce also accounts for a significant portion of B2B markets. At the end of 2017, Forrester estimated that B2B e-commerce amounted to US$889 billion in the US, representing 11 per cent of total US B2B sales. Worldwide, the value of B2B e-commerce (US$7.6 trillion in 2017) is estimated to be more than three times that of B2C (US$2.4 trillion).
Economic characteristics of e-commerce
E-commerce takes many forms and is difficult to systematically distinguish from traditional commerce. Nonetheless, it has several key economic characteristics relevant to the antitrust analysis of vertical relationships.
First, e-commerce allows firms and customers to interact across long distances and borders. While this discussion focuses on matters in the US, the antitrust analysis may vary for different transactions depending on the locations of the parties to the vertical relationship (both seller and customer). For example, resale-price maintenance (RPM) is evaluated under the rule of reason in the US, but not in Europe.Thus, contractual arrangements that are acceptable in one jurisdiction may not be in another.
Second, because the cost of marketing and distributing products online requires lower up front investment, certain barriers to entry may be lower, particularly at the stage of distribution closest to the end customer. As any traditional brick-and-mortar retailer can attest, e-commerce has increased competition, facilitating and diversifying the number of competitors in a number of segments.At the same time, certain barriers to entry may be higher in e-commerce. Entry may be particularly challenging when significant investments in brand or infrastructure are required, or when customers have a strong preference for particular products, services or content and will not use a distributor that cannot offer them. Furthermore, some have argued that data about consumers and their preferences may form such a barrier, as long-standing firms with deeper data may utilise that data to better target consumer needs.
Third, the expansion of competing options for distribution, paired with the lower costs of reaching customers with both products and information, has also changed the way consumers collect information about products. The general expectation is that e-commerce reduces search costs for customers,making it more difficult to sustain supracompetitive pricing. However, overwhelming options or platform loyalty may reduce the benefits of lower search costs. E-commerce provides customers with direct access to products and services. These services may be accessed through the service originator’s digital channels, such as an airline’s or a hotel chain’s own web pages, or through third-party digital platforms that aggregate pricing and other information, provide payment mechanisms and allow for comparison shopping. Such platforms may co-exist with or substitute for the service originator’s own digital channels.
Fourth, e-commerce facilitates aggregation – competing firms may offer a wider and cheaper array of products and services and generate substantial consumer welfare gains.The tendency for e-commerce towards aggregation may also lead e-commerce firms to transform into more diversified platforms that bring together buyers and sellers, or content creators and content consumers. The two-sided or multisided nature of those platforms may create economic value through network effects across the platform. Specifically, the more buyers a platform recruits, the more attractive it may become to sellers (and vice versa). In some scenarios, the platform is truly multisided. For example, Spotify receives revenue from some users in the form of subscription fees and these users are attracted by its library of musical content. Spotify also sells time to advertisers, who are attracted by its community of subscribers. Spotify uses the revenue from both sources to compensate the providers of content. Balancing these dimensions is an aspect of some, though not all, e-commerce models.
Vertical restraints and rule of reason analysis in e-commerce
Contractual constraints between buyers and suppliers are common in a variety of industries and sectors, including those in which e-commerce has made significant inroads. Those vertical restraints can and do have effects on competition. However, unlike horizontal agreements among competitors, they are not per se violations of antitrust law in the US. Such distinctions reflect both the limited risk to competition from vertical relationships absent market power and the potential procompetitive benefits.
Under such circumstances, regulators and economists in the US generally apply a three-step rule of reason analysis in which plaintiffs first must show that the vertical restraint causes harm to competition. Defendants may then demonstrate an offsetting benefit to competition, which, if successful, then shifts the burden back to the plaintiffs, who must demonstrate that such benefits could be achieved using less restrictive tactics. As part of the analysis, regulators and courts examine market definition and market power so that consumers at risk can be identified and potential benefits and effects evaluated. Specifically, vertical agreements require an analysis of competition both upstream (in which manufacturers or suppliers compete with similar firms offering competing brands or solutions) and downstream (in which retailers selling similar products compete for end customers).
Types of vertical agreements
Vertical agreements may directly affect the prices charged or may limit or influence the nature and scope of product distribution. For example, such agreements may restrict the price charged or advertised to consumers or require that customers make a certain amount of their purchases from the seller. These agreements have the potential to exclude rivals, but may also protect legitimate incentives to investment.
The need to consider explicitly these offsetting effects was specifically articulated in the 2007 Leegin decision by the US Supreme Court. This decision specifically ruled that vertical restraints should be assessed under the rule of reason and precluded holding such restrictions as per se illegal under §1 of the Sherman Act. The court noted that ‘respected economic analysts . . . conclude that vertical price restraints can have procompetitive effects.’To understand the potential pro- and anticompetitive effects of such vertical agreements requires a careful review of the contractual restraints in the context of e-commerce.
For example, consider tying and bundling, which are arrangements whereby a supplier makes the purchase of one good or service (the tying good) conditional to the purchase of a second good (the tied good), or literally combines the goods into a single package (a bundle). When a new entrant seeks to introduce a new complement, bundling or tying arrangements may be procompetitive and even necessary to introduce a competitive alternative to existing solutions. However, such arrangements may be more likely to be anticompetitive when aggregation creates opportunities for firms to privilege their own options in the downstream market, and potentially exclude rivals and inhibit entry. Similarly, a platform provider that also offers its own products through the platform may leverage such arrangements and exclude rivals.A more subtle form of vertical restraint in the context of e-commerce that raises similar issues is preferential placement. For example, a search engine may preferentially feature search results associated with its own brand or a cable or wireless provider may prioritise bandwidth for its own content.
Exclusive distribution is another form of vertical restraint found among e-commerce firms. While e-commerce firms may aggregate upstream firms’ offerings to consumers (downstream), upstream firms may prefer to limit distribution by rival providers to prevent competitors from developing a foothold. Given such incentives, upstream firms may seek exclusive distribution arrangements or volume commitments. Again, these restraints may be explicit, in the form of exclusive distribution agreements, or more subtle, such as offering discounts or preferential pricing to firms that commit to quantities or share of sales of a certain brand. In some circumstances, inter-brand competition (between competing supplier or manufacturer brands) may be sufficient to limit effects on customers. However, when consumer preferences or limited competitive options give upstream suppliers some market power, the weakening of intra-brand competition (i.e., competition between distributors offering the same brands) may lead to price increases or other anticompetitive effects.
Finally, vertical restraints may constrain prices offered, advertised or charged to consumers. Given the diversity and complexity of e-commerce competition, firms are often concerned with making investments in quality and marketing efforts, if those investments can be exploited by others who do not incur the investments and sell at lower prices. To preclude free-riding of this type, firms may use resale price maintenance (in which the upstream firm specifies the price at which a retailer can resell the product) and most favoured customer clauses (in which a firm specifies that it will receive a price no worse than that of any rival).
Again, if inter-brand competition is sufficient, such arrangements may on balance be procompetitive. However, they may weaken intra-brand price competition, resulting in offsetting effects that may be of concern to regulators.
Defining a market
Applying the rule of reason to evaluate vertical arrangements in e-commerce in the US requires an understanding of the relevant market. Whether products are available through traditional physical channels and online retailers or only available online, the competitive dynamics, evolving business strategies and solutions, and shifting consumer substitution patterns are all relevant to antitrust analysis. However, defining the relevant market is complicated in e-commerce, not only because substitution is less constrained by physical location or physical products, but also because supply-side substitution may come from unexpected places.
Online and offline distribution channels may serve the same market, even if the offerings are quite different (think Lyft and cabs). At the same time, for some consumers, online and offline options are not substitutable even when the end product is identical (a boutique on Rodeo Drive may not compete with a website that sells the same pair of shoes because the customers and their experience of the purchasing process are significantly different). In addition, rapid technological innovation may mean that product substitution options will be expanded through cross-platform competition facilitated by the low capital costs associated with incremental geographic or product expansions.
Finally, the agreements at issue may affect upstream inter-brand competition or downstream intra-brand competition. Assessing the competitive effects from foreclosure, collusion, or weakened price or quality competition requires some sense of their likelihood. Such questions are particularly challenging when firms are continually evolving; for example, Amazon’s shift from being merely an aggregator in a vertical chain of B2C commerce to also becoming a horizontal competitor through its private label products.
Assessing market power
Just as market definition methodology is challenged by the rapid evolution of e-commerce strategies and shifting consumer substitution patterns, the evaluation of market power in such a dynamic context is more complicated than in traditional industries. While traditional assessments of market power such as HHI,UPPI, and direct evidence of firms exercising pricing power can be helpful when the underlying market is properly defined, near-term shifts in strategy and preferences may significantly affect conclusions.
The twin issues of demand and supply side substitution, as well as the relevant horizon to evaluate competitive effects, are central to evaluating transactions affected by or within e-commerce. For example, a federal judge blocked the 2016 merger between Staples and Office Depot in part because he rejected the argument that Amazon’s recent foray into B2B office supplies and its expected expansion was sufficient to discipline competition post-merger. Some have asked whether such a position is reasonable, given the speed with which Netflix took over Blockbuster and Hollywood Video after their merger was blocked in 2005.While the judge ultimately blocked the transaction, lingering questions include whether a merger of these traditional retailers provides sufficient market power to affect e-commerce related competition in the near-term and, if so, for how long? Might they have provided a more viable long-term competitive option if the merger had been approved?
On the other hand, the evaluation of an acquisition of competing online retail platforms is often focused on demand side substitution. For example, traditional competition analysis in conjunction with a regulatory review of Amazon’s proposed acquisition of Zappos.com in 2009 may have focused on whether consumers buying shoes at Zappos were also shopping or gathering pricing information about shoes on Amazon. It may also have examined whether Zappos could move into the distribution of other products using its existing user interface, distribution experience and assets, and customer base (the acquisition was completed without a regulatory challenge in November 2009).
These questions are also relevant when examining market power in the context of vertical restraints. An exclusive distribution agreement or preferential product placement may be of little concern when a new e-commerce site is introduced. However, if that site becomes a leading aggregator or secures a large loyal or locked-in customer base, it may have sufficient market power upstream or downstream to stifle competition. Given the rapid pace of innovation and shifting strategies of many e-commerce companies, the evaluation of closeness of competition in the near-term, and the question of the appropriate horizon for competitive discipline from the supply side, are more complicated than in other industries. Thus, while traditional economic analyses may be sufficient to examine these industries, their application may be more complex when e-commerce is involved.
Analysing pro- and anticompetitive effects
As clearly articulated by the Supreme Court in Leegin (2007) and demonstrated in subsequent cases,alternative theories of competitive harm are relevant when evaluating market power in a clearly defined market, as well as the efficiency motives and potential welfare enhancements that arise from vertical agreements. When firms depend on up- and downstream parties to provide their services or reach their customers, coordination and information sharing become necessary. To correctly coordinate decision making; to account for competition and consumer preferences; and to ensure profitable investments, vertical agreements and restraints may be necessary. For example, when suppliers and retailers are making relationship-specific investments, hold-up risk (in which the bargaining position of the party making the investment decreases once the investment is made) may lead to exploitation and sub-optimal outcomes. Long-term contracts including volume commitments or exclusivity provisions may provide the necessary assurances to both parties. Such benefits must be then compared to the potential foreclosure of competing upstream suppliers and whether the contractual arrangement is overly restrictive given the procompetitive justifications.
Similar issues arise when competitors can free-ride. For example, the combination of physical showrooms with an online site by one retailer may provide necessary information for consumers to comparison shop for a luxury good. The retailer making such investments may seek an exclusivity agreement, at least for certain parts of the product line or in certain countries. Without such exclusivity, other retailers will free-ride on the information it provided to consumers and undercut its prices. In such circumstances, consumers may be better off with some vertical restraints, as long as inter-brand competition among luxury good suppliers is sufficient to preserve competition. In other circumstances, if the upstream supplier has market power, those agreements may weaken competition downstream.
As illustrated by the differing positions of EU and US agencies, vertical restraints on end-customer prices may be the most difficult to evaluate. The efficiency motives cited for resale price maintenance and other pricing restraints generally fall into two categories – coordination of decision-making and signalling. Suppliers and retailers often have distinctly different information: local preferences may or may not reflect market-wide preferences and the prices charged by one retailer may not be optimal from a market perspective (e.g., if too many units are sold at depressed prices in select markets, supplier profitability and long-term investments may be adversely affected). Furthermore, given that price is often a signal of quality, suppliers offering premium products may seek to control prices, retail discounting and the channels of distribution to maintain their reputation. However, when suppliers directly limit intra-brand price competition, upstream market power is highly relevant. Without sufficient competition among suppliers, such contractual arrangements may, on balance, be anticompetitive.
While these effects are well known, e-commerce challenges the analysis of some vertical restraints. For instance, territorial exclusivity clauses may be procompetitive when they constrain free-riding; however, e-commerce may jeopardise any procompetitive effects and result in heightened concerns for such clauses. Similarly, selective distribution – whereby a supplier applies criteria to admit retailers to its distribution network – is common and well understood. However, e-commerce has made such clauses more common as suppliers try to protect their brand image on the internet. An extreme version of such clauses is the outright ban of online sales. A supplier may prefer to forgo sales on the internet because of information asymmetry (consumers are not able to physically see and try the product) or the fear that retailers will have insufficient incentives to invest in quality.
Recent antitrust enforcement
In 2011, the US Department of Justice (DOJ) accused Apple and five of the largest US publishers of conspiring to increase the price for e-books above Amazon’s US$9.99 price in the run-up to the launch of the iPad in 2010.The DOJ claimed that Apple and the publishers had cost consumers millions of dollars by colluding to raise the price for e-books. The companies allegedly achieved this by simultaneously agreeing to switch to an agency sales model, which gave the power to publishers, rather than retailers, to set the price for the consumer. The publishers denied wrongdoing, but all eventually chose to settle. Apple, however, vigorously defended its actions, claiming that it was simply a new entrant breaking the hold of an existing monopolist (Amazon). Ultimately, after multiple appeals and the Supreme Court declining to hear the case, Apple was found guilty in 2016 and required to re-negotiate all e-book distribution deals and pay US$450 million in damages to consumers. The central issues of this case centred around two fundamental components of antitrust law and how to apply them in the case of a digital platform.
First, the courts needed to decide whether market conditions in the early days of the e-book industry were sufficiently unique to warrant a review of Apple’s actions under the rule of reason standard. In cases where the economic impact of the defendant’s actions is not immediately obvious, the plaintiffs are required to prove a concerted action and harm to competition. Alternatively, if Apple’s alleged actions were per se anticompetitive, then the plaintiffs would only have to prove that an agreement existed. This more lenient standard is meant to be applied ‘only after courts have had considerable experience with the type of restraint at issue.’
The DOJ argued that the court did have experience with cases like Apple’s, and that the economics of bookselling through traditional physical platforms, such as brick-and-mortar bookstores, was well understood. Apple contended that digitisation of both book content and the platform had fundamentally changed the market. Both publishers and retailers were struggling with how to price books that cost substantially less to produce but delivered the same content as the physical copies. Further, Amazon’s market power in the e-book market was allowing it to engage in potentially anticompetitive behaviour by pricing below marginal cost. While the 2nd Circuit judge did say she was ‘mindful of Apple’s argument that the nascent e-book industry has some new and unusual features,’ she ruled that the DOJ could apply the per se standard. This ruling sent a warning to other digital platforms that new digital markets may not be considered sufficiently different from physical markets to warrant differential treatment. To consider Apple’s actions per se anticompetitive would also be tied to whether Apple’s agreements should be characterised as vertical or whether Apple was effectively the facilitator of horizontal agreements. As noted, horizontal agreements among competitors are ‘prohibited despite the reasonableness of the particular prices agreed upon,’ while the legality of vertical price restraints depends on the companies’ motives and the outcome of their actions.Apple argued that it had engaged in ‘inherently vertical, novel conduct’ with the procompetitive objective of launching the iBooks Store to compete with Amazon, which owned 90 per cent of the e-book market, while the DOJ claimed that Apple’s behaviour was targeted at horizontal coordination. However, the courts sided with the DOJ in accusing Apple of facilitating a horizontal scheme, reasoning that Apple had informed publishers of one another’s intentions and had made it clear that multi-publisher sign-on was essential. This ruling brings to light the challenge of evaluating the nature of coordination efforts when a platform is established.
In re: Online Travel Company Hotel Booking Antitrust Litigation
In 2012, a consolidated class action lawsuit was filed in the US District Court for the Northern District of Texas against 12 major hotel chains and nine online travel agencies (OTAs) for allegedly engaging in an industry-wide conspiracy to uniformly enact vertical agreements between hotels and OTAs to eliminate intra-brand competition between online hotel booking sites.In the vertical agreements, the hotels would publish the best available rate or lowest rate of their hotel rooms for the OTA to use on their website (another example of an RPM agreement). The agreements also contained most favoured nation (MFN) clauses to ensure that the published rate for each OTA was not lower than that for any other OTA.
Critically, the plaintiffs did not allege that any individual vertical agreement was anticompetitive; rather, the ‘Plaintiffs’ antitrust claims rested entirely on the circumstantial facts purportedly showing that Defendants entered into an “express or tacit” industry-wide conspiracy not to compete.’As such, the plaintiffs claimed that the matter ‘must be considered holistically, for without either [horizontal or vertical] component, the scheme would fail’.
The plaintiffs alleged both a per se violation and a violation under the rule of reason. Using the rule of reason framework, the parties disagreed on the correct market definition. The plaintiffs defined the relevant product market as ‘the direct online sale of hotel room reservations’ in the US. The defendants argued that the product market should be wider, including other methods of booking hotel rooms such as brick-and-mortar travel agencies. The defendants also argued that the market should be considered at a local level, as consumers substitute among hotels at their desired location, not among all hotels in the US.
The plaintiffs and defendants also used different methods to evaluate market power. The plaintiffs alleged that the defendants collectively had significant market power, citing the OTA defendants’ 94 per cent market share. The defendants argued that market power should be evaluated separately for each defendant, rather than collectively.
The defendants argued that the RPM agreements had procompetitive benefits. The defendants relied on the argument set forth in the Leegin decision that RPMs between the hotels and OTAs (which reduce intra-brand competition) allow for an increase in inter-brand competition among hotel companies. They also noted that the agreements allow hotels to communicate a consistent message to their customers and more efficiently deliver their products to the public.
In 2014, the court dismissed the lawsuit, primarily on the grounds that the plaintiffs did not provide sufficient facts to demonstrate an industry-wide conspiracy. The court concluded that the agreements could be explained by economic forces: a hotel could reasonably want to control the price of its rooms, for example, to protect the hotel’s brand, and an OTA would want an MFN clause in exchange for giving up the right to discount hotel rooms. Additionally, the court found that the plaintiffs’ ‘factual enhancements’ to their case (such as potential communication among competitors at industry conferences and the existence of government investigations into the matter in the EU) did not suggest a conspiracy or were irrelevant. Because the case was framed holistically as an industry-wide conspiracy, rather than focusing on the legality of the individual vertical agreements, the court did not complete a rule of reason analysis of the RPM agreements in its decision. As such, the case did not set further precedent on how to apply Leegin in e-commerce industries.
Vertical restraints in e-commerce raise important questions for antitrust regulators around the globe. Given the reach of many e-commerce sites, the regulatory framework applied to various vertical restraints must be carefully considered by businesses when crafting agreements. Meanwhile, regulators must consider the risks of weakened competition in the context of efficiencies and market expansion. Because e-commerce creates a natural complementarity between the aggregation of information and resources, and the distribution of goods and services, there is an increasing likelihood for firms to engage in both horizontal and vertical relationships simultaneously. These arrangements may change the antitrust analysis of vertical restraints, and particularly require the evaluation of both supply and demand side substitution patterns. The traditional tools of antitrust analysis are up to the challenge, so long as the time horizon over which effects are considered is correctly matched to the dynamism and uncertainty that accompanies e-commerce business models.
 Rebecca Kirk Fair is a managing principal, and Nikita Piankov and Emmanuel Frot are vice presidents, of Analysis Group, an economics consulting firm.
 Wallace, Tracey, ‘The 19 Ecommerce Trends + 147 Online Shopping Stats Fueling Sales Growth in 2018,’ BigCommerce, https://www.bigcommerce.com/blog/ecommerce-trends/ (last accessed 17 October 2018).
 ‘US Census Bureau News,’ US Department of Commerce, Economic Indicators Division, 17 August 2018; ‘E-commerce share of total retail sales in China from 2014 to 2019,’ Statista (eMarketer, InsideRetail Hong Kong), https://www.statista.com/statistics/379087/e-commerce-share-of-retail-sales-in-china/, (last accessed 17 October 2018).
 Orendorff, Aaron, ‘B2B in Ecommerce: How the Best Succeed in a $7.6 Trillion Online Industry,’ ShopifyPlus, 17 July 2017, https://www.shopify.com/enterprise/b2b-ecommerce (last accessed 17 October 2018).
 Opinion of the Court, Leegin Creative Leather Products, Inc. v. PSKS, Inc., US Supreme Court, No. 06-480, 28 June 2007. In Europe, RPM is largely seen as a restriction of competition ‘by object.’ The European Commission’s Guidelines on Vertical Restraints (2010) consider it a ‘hardcore restriction,’ although they also note that ‘RPM may not only restrict competition but may also, in particular where it is supplier driven, lead to efficiencies, which will be assessed under Article 101(3) [TFEU].’ In practice, RPM is an area of substantial antitrust intervention among EU national antitrust agencies. The European Commission itself added to this enforcement trend with its own recent decisions in July 2018, against four manufacturers that were found to have engaged in RPM with respect to the online sale of electronic goods (see http://europa.eu/rapid/press-release_IP-18-4601_en.htm).
 For example, Wal-Mart has faced only increasing competition from Amazon, as the types of items frequently purchased on Amazon continue to expand, including fresh food and drugstore items. See Stone, Brad and Matthew Boyle, ‘Can Wal-Mart’s Expensive New E-Commerce Operation Compete With Amazon?’ Bloomberg, https://www.bloomberg.com/news/features/2017-05-04/can-wal-mart-s-expensive-new-e-commerce-operation-compete-with-amazon.
 See Stucke, Maurice and Allen Grunes, Big Data and Competition Policy, 2016, Oxford University Press; Lambrecht, Anja, and Catherine E Tucker, ‘Can Big Data Protect a Firm from Competition?,’ Competition Policy International Antitrust Chronicle, January 2017.
 Economists find positive but modest costs, usually lower than the value of the time it would take consumers to travel to just one offline seller. See Bajari, Patrick and Ali Hortaçsu, ‘The Winner’s Curse, Reserve Prices, and Endogenous Entry: Empirical Insights from eBay Auctions,’ The RAND Journal of Economics, 34(2), 329-355, 2003; Brynjolfsson, Erik, Astrid A Dick and Michael D Smith, ‘A Nearly Perfect Market? Differentiation vs. Price in Consumer Choice,’ Quantitative Marketing and Economics, 8, 1-33, 2010; and Hong, Han and Matthew Shum, ‘Using Price Distributions to Estimate Search Costs,’ The RAND Journal of Economics, 37(2), 257-275, 2006.
 See Brynjolfsson, Erik, Yu (Jeffrey) Hu and Michael D Smith, ‘Consumer Surplus in the Digital Economy: Estimating the Value of Increased Product Variety at Online Booksellers,’ Management Science, 49(11), 1580-1596, 2003. They find that online book retailers offer 23 times as many titles as a typical brick-and-mortar firm does. Welfare gains generated by the greater variety are estimated to be 7 to 10 times larger than the gains from increased competition.
 A more recent example is Amazon’s interest in offering advertising. See Creswell, Julie, ‘Amazon Sets Its Sights on the $88 Billion Online Ad Market,’ The New York Times, 3 September 2018.
 Opinion of the Court, Leegin Creative Leather Products, Inc. v. PSKS, Inc., US Supreme Court, No. 06-480, 28 June 2007.
 Shim, Sunghee and Juwon Kwak, ‘Tying to Foreclose in Two-Sided Markets,’ Journal of Economic Analysis and Policy, 15(4), 1919-1937, 2015. See also Michael D Whinston, ‘Tying, Foreclosure, and Exclusion,’ American Economic Review, 80(4), 837-859, September 1990.
 The Herfindahl-Hirschman Index, which is used to measure market concentration.
 The Upward Pricing Pressure Index, which is used to assess the risk of price increases.
 Feinstein, Deborah L, ‘Revising the Merger Guidelines: Looking Back to Move Forward,’ Competition Policy International Antitrust Chronicle, December 2009 (1).
 Jacobs v. Tempur-Pedic Int’l., an antitrust action claiming that the defendant illegally enforced vertical RPM agreements, was dismissed by the US 11th Circuit court in 2010 because of the lack of both a defined, relevant market and any harm to competition. Further, in June 2018 the US Supreme Court ruled in Ohio et al. v. American Express Co. et al. that the rules in American Express’s merchant contracts, which barred merchants from encouraging American Express cardholders to use competing credit cards to pay for purchases, did not violate antitrust laws because of the services American Express provided to both sides of the platform.
 The publishers charged were Hachette Book Group Inc, HarperCollins Publishers LLC, Penguin Group USA Inc, Simon & Schuster Inc and Holtzbrinck Publishers LLC.
 Opinion of the Court, Leegin Creative Leather Products, Inc v. PSKS, Inc, US Supreme Court, No. 06-480, 28 June 2007, p. 2.
 Starr v. Sony BMG Music Entertainment, US Court of Appeals, 2nd Circuit, 592 F.3d 314, January 2010.
 United States of America v. Apple Inc, US Court of Appeals, 2nd Circuit, 13-3741, June 2015 (DLC 2013).
 EyeforTravel, Ltd, a travel industry news company based in the UK, was also a defendant because it allegedly facilitated the price fixing conspiracy through its annual industry conferences. See In re Online Travel Company (OTC) Hotel Booking Antitrust Litigation, 997 F. Supp. 2d 526 (N.D. Tex. 2014).
 ‘Memorandum Opinion and Order,’ In re Online Travel Company (OTC) Hotel Booking Antitrust Litigation, Case No. 3:12-cv-3513-B.