United States – E-commerce, Big Data and Algorithms: Antitrust


While in Europe there has been an effort to adopt policies specifically relating to e-commerce and digital markets, antitrust enforcement in the US has remained consistent with how courts and the enforcement agencies have applied antitrust law to the traditional brick-and-mortar economy and has focused more on how to apply general antitrust principles in new contexts and to new tools. Much of the developments have stemmed from the significant distinction between how horizontal and vertical agreements are treated under Section 1 of the Sherman Act. Horizontal agreements among competitors on matters such as price are regarded as per se violations, which means that agreements are condemned as antitrust violations regardless of their actual impact.[2] Conversely, vertical agreements are analysed under the ‘rule of reason’, which means that the government or private plaintiffs bear the burden of showing that the challenged action had an actual adverse effect on competition as a whole in a relevant market, defendants are then allowed to offer procompetitive justifications, and conduct can only be found unlawful after a balancing of the procompetitive and anticompetitive effects to determine ‘whether the restraint imposed is such as merely regulates and perhaps thereby promotes competition or whether it is such as may suppress or even destroy competition’.[3]

The stark difference in how horizontal and vertical agreements are treated means that enforcement and litigation decisions often turn on the question of whether an agreement or practice can be characterised as horizontal in nature. This dichotomy has heavily influenced the cases and legal theories that have been pursued in both the e-commerce space and in connection with digital communications, and the emerging development and use of algorithms.


The e-books cases involved application of Section 1 of the Sherman Act to ‘agency’ agreements between Apple and individual publishers, which contained most favoured nation (MFN) clauses. Under the agency model, digital sellers, such as Apple or Amazon, would act as agents of the publisher and use prices set by the publisher while retaining a percentage commission. This was in contrast to the prior wholesale model pursuant to which publishers would sell e-books to digital sellers at a wholesale price that could then resell the e-books to consumers at a retail price chosen by the digital seller. In contrast to Europe, where there has been a focus on addressing vertical restraints in e-commerce markets, the Department of Justice (DOJ) declined to assert vertical claims against Apple and the publishers under the rule of reason. Consequently, there was no direct challenge to either the ‘agency’ model or to the MFN clause in each agreement that required each publisher to set retail prices for e-books on Apple’s platform at the lowest price the books were sold on any other platform. The DOJ instead pursued a horizontal theory, alleging that the MFN clauses were facilitating devices for a hub-and-spoke conspiracy with Apple at the centre and the publishers serving as the spokes (and providing the horizontality).[4] The court found that Apple had consciously orchestrated a horizontal conspiracy among the publishers and used the MFN clauses to effectively force each publisher to adopt an agency model with other retailers (or they otherwise would have been stuck with both the lower prices set by Amazon and less revenue from each book sold through Apple).[5] While the MFN clauses served the purpose of forcing the shift to an agency model, the government also relied heavily on more conventional evidence of collusion, including assurances by Apple to each publisher that a critical mass of major publishers would be making the shift to the agency model and communications among the publishers themselves.[6]

Similarly, in the online travel companies (OTC) litigation, which was brought by private plaintiffs, the case also focused on whether there were sufficient facts to infer a horizontal agreement.[7] In the case, hotels had allegedly entered into ‘rate parity’ agreements with OTCs pursuant to which the hotel would set the lowest allowed rate for a room and guarantee that the published rate given to an OTC would be as favourable as the hotels own rates or the rates given to any other OTC. The plaintiffs did not attempt to state a claim that the vertical agreements between each hotel and each OTC were anticompetitive and violated the rule of reason. Instead, the plaintiffs framed the case as an industry-wide conspiracy to fix prices and eliminate intra-brand competition. The court dismissed the case at the pleading stage because the plaintiffs had only alleged consciously parallel behaviour and there were insufficient allegations to make a horizontal conspiracy among the hotels plausible.[8] In contrast with the more detailed allegations of horizontal communications in the e-books litigation, the court found the plaintiffs had made only vague allegations of an opportunity to conspire at trade conferences. The court also noted that each hotel would have legitimate, independent reasons for wanting to enter the rate parity agreements and noted the lack of any allegation about a reduction in inter-brand competition.

In seeking leave to amend their complaint, the plaintiffs dropped the hotels as defendants and alleged a horizontal price fixing agreement among the OTCs.[9] Under this new theory, the rate parity agreements were a tool to effectuate a conspiracy among the OTCs and allegedly followed a prior cessation of price competition among the OTCs. But the court found that the new theory relied on the same allegations of consciously parallel behaviour and that ‘[p]laintiffs’ mere re-configuration of the culpable actors’ added ‘nothing new or suggestive to the mix.’[10] The outcome again turned on the lack of factual allegations plausibly suggesting a horizontal conspiracy.

These cases demonstrate that there is no distinct body of antitrust jurisprudence pertaining to e-commerce markets. As in any other industry, cases will turn on whether plaintiffs can muster sufficient facts and evidence to make a horizontal conspiracy plausible.

Algorithmic pricing

Pricing algorithms involve having a computer set prices based on a set step-by-step procedure utilising certain prescribed inputs, which could include publicly available data about a competitor’s prices. There is no prohibition under US law on using algorithms to set prices. In certain contexts, the use of algorithms can even be procompetitive by enabling a company to respond more quickly to its competition. Significantly, the mere decision of a single company to utilise algorithmic pricing constitutes unilateral conduct that is beyond the purview of Section 1 of the Sherman Act.

An example of the use of a pricing algorithm that does not raise antitrust concerns was the subject of a business review letter submitted by Amadeus Group LLC, relating to a subscription service that would involve subscribers inputting a variety of information relating to services for commercial mailings.[11] Once each subscriber input the details of their job request, Amadeus’s pricing algorithm would calculate different postage, packaging and transportation scenarios specific to each subscriber’s request to allow the subscriber to choose the most efficient logistics for its commercial mailings. In this scenario, the pricing algorithm did not raise any apparent horizontal collusion issues and the DOJ indicated that it would not be inclined to challenge it.[12] The competitive concerns stemmed not from the algorithm, but more from the fact that subscribers could be uploading pricing and other competitively sensitive information from competitors of Amadeus’s affiliate Mystic Logistics, which itself provides transportation services for commercial mailings. The DOJ was satisfied with firewalls and safeguards that would prevent Amadeus and Mystic from having access to data from its competitors and similarly preclude subscribers and third parties from accessing information from other subscribers.[13]

Algorithms do raise antitrust issues, however, where they can be used by competitors to coordinate their pricing or to effectuate a price-fixing conspiracy. That is true whether they result in identical pricing or not; the touchstone inquiry under Section 1 of the Sherman Act is whether there is an agreement among competitors that directly affects prices.[14]

The DOJ charged companies that sell wall posters on the Amazon Marketplace with using an algorithm as part of a conspiracy to match prices for specific posters.[15] The competitors allegedly used software to set matching prices just below that offered by the lowest non-conspiring competitor. In that regard, it was a pure horizontal price-fixing conspiracy that simply used an algorithm to set the agreed-upon price. While unremarkable in certain respects, the DOJ has cited the case as an example of how pricing algorithms can make price-fixing schemes easier to effectuate and more difficult to detect, since price changes can be effectuated automatically without need for ongoing monitoring, coordination and communication by its participants.[16] The DOJ has also expressed concern that algorithms may help facilitate more conscious parallelism and tacit collusion that is beyond the reach of Section 1 of the Sherman Act.[17]

A pricing algorithm also featured prominently in an antitrust class action against the founder of Uber alleging that he had conspired with the millions of Uber drivers to use Uber’s pricing algorithm to set the prices charged to Uber riders.[18] The plaintiff alleged that the algorithm was used to effectuate a horizontal conspiracy among all Uber drivers who do not compete on price and instead all charge the fares set by the Uber algorithm. In assessing the allegations at the pleading stage, which required the court to accept the plaintiff’s factual allegations as true, the court found that the plaintiff had adequately alleged a horizontal conspiracy and that Uber drivers agreed to participate in the conspiracy when they assented to the terms of Uber’s written agreement.[19] In rejecting the defendant’s argument that each driver’s agreement to follow Uber’s terms should be construed as a series of vertical agreements, the court cited the Apple e-books decision and other cases in which plaintiffs have successfully pled hub-and-spoke conspiracies.[20] Further, in rejecting the defendant’s argument that he was the mere purveyor of an ‘app’, the court noted that ‘[t]he advancement of technological means for the orchestration of large-scale price-fixing conspiracies need not leave antitrust law behind.’[21] The plaintiff also relied on additional facts to bolster his conspiracy allegations, including that Uber ‘organizes events for drivers to get together’.[22]

Taken together, these cases demonstrate that, as in the e-commerce context, enforcement and litigation decisions will often depend on what an algorithm is being used for and, in particular whether there is any basis to allege that it is part of an effort to coordinate prices among horizontal competitors.[23]

Digital communications

In their efforts to identify and prosecute horizontal conspiracies, an area that will be of increasing importance to the enforcement agencies is monitoring and investigating how advances in digital communications and social media provide new methods for conspirators to communicate. Much like how the development and widespread use of email changed the face of the typical conspiracy case, the same can be expected as the means of electronic communication continually advance and change. In criminal enforcement actions brought by the DOJ relating to online retail stores selling custom wristbands, the DOJ alleged that the defendants communicated ‘via text and online messaging platforms.’[24] In the trade secret litigation between Uber and Google, the court rebuked Uber for using a messaging app called Wickr, which allows for users to send encrypted, self-destroying messages.[25] While the issue in the case pertained to the destruction of evidence, Wickr and similar apps have the potential to impact antitrust litigation by making the discovery of communications between competitors harder to detect. Unlike email, this type of app is specifically designed to avoid any sort of paper trail.

Big data

Big data has not been the subject of much attention in the enforcement area outside of the merger context. Unlike algorithms and digital communications, which raise issues regarding potential horizontal collusion, any conduct concerns regarding big data relate more to potential monopolisation or attempted monopolisation under Section 2 of the Sherman Act. At this point, however, no cases have been pursued on such a theory and there is reason to be sceptical about how much concern the collection of data raises under Section 2.

Indeed, the US enforcement agencies have recognised that the collection and use of data provides opportunities for innovation and efficiencies in addressing consumer needs. Consequently, a consistent theme in public remarks by officials at the enforcement agencies has been that an overzealous reaction to unspecified concerns about big data could be counterproductive and actually stifle potential innovations that would be beneficial to consumers. Assistant Attorney General Makan Delrahim stated that regulators must be careful not to ‘kill the golden goose of innovation’ and said enforcers should bring only ‘evidence-based’ challenges.[26] Similarly, Deputy Assistant Attorney General Roger Alford has explained that ‘big data’ is a vague term, and that ‘evidence-based investigations are better than static, one-size-fits-all solutions.’[27] Mr Alford cautioned that an unprincipled push to address big data could cause serious problems, warning that regulators must be ‘careful in how we proceed in analyzing digital markets’.[28]

As to potential remedies, Deputy Assistant Attorney General Bernard A Nigro Jr warned that ‘[t]here are many reasons to be sceptical of using the antitrust laws to force the sharing of data.’[29] Mr Nigro cautioned that such a policy would raise free-riding concerns and reduce firms’ incentives to innovate.[30]

Economists have likewise cautioned that a firm’s possession of big data will rarely on its own confer an unfair competitive advantage.[31] This is true both because data relating to consumer preferences can go stale quickly and also because data is rarely truly unique. Data can be collected from a variety of different sources and provide similar insights into consumer behaviour and preferences. And in the rare case where there are anticompetitive concerns, nothing prevents regulators from pursuing ‘evidence-based’ challenges.


[1] Paul Eckles is a partner, and Jeremy Koegel is an associate, at Skadden, Arps, Slate, Meagher & Flom.

[2] See, e.g., United States v. Socony-Vacuum Oil Co, 310 US 150 (1940).

[3] See, e.g., Chicago Bd of Trade v. United States, 246 US 231, 238 (1918).

[4] US v. Apple, Inc, 791 F.3d 290 (2d Cir. 2015).

[5] id. at 326-329.

[6] id. at 317-19.

[7] In re Online Travel Co (OTC) Hotel Booking Antitrust Litigation, 2014 WL 8276572 (ND Tex 2014)i

[8] id.

[9] In re: Online Travel Co. (OTC) Hotel Booking Antitrust Litig, 2014 WL 5460450 (N.D. Tex. 2014).

[10] id.

[11] See Kathleen M Porter, ‘Amadeus and Mystics Logistics Business Review Request Letter’ (26 Sep. 2016), available at https://www.justice.gov/atr/page/file/920796/download.

[12] See Renate B Hesse, ‘Amadeus and Mystics Logistics Department of Justice Business Review Letter’ (13 Dec. 2016), available at https://www.justice.gov/atr/page/file/920786/download.

[13] id.

[14] See Socony-Vacuum, 310 US at 843.

[16] Algorithms and Coordinated Effects, Remarks of Commissioner Terrell McSweeny (22 May 2017), available at . https://www.ftc.gov/system/files/documents/public_statements/1220673/mcsweeny_-_oxford_cclp_remarks_-_algorithms_and_coordinated_effects_5-22-17.pdf.

[17] See The Implications of Algorithmic Pricing for Coordinated Effects Analysis and Price Discrimination Markets in Antitrust Enforcement, Antitrust Vol. 32, at 75 (Fall 2017), by Terrell McSweeny and Brian O’Dea.

[18] Meyer v. Kalanick, 174 F Supp 3d 817 (SDNY 2016). The case was ultimately sent to arbitration so it is unlikely to result in further published opinions on the merits of the claim.

[19] id. at 822.

[20] id. at 824.

[21] id. at 825.

[22] id. at 825.

[23] Pricing algorithms can also help facilitate price discrimination, but the US agencies do not view that as a significant antitrust concern. In fact, they have noted that ‘[p]rice discrimination can increase market output, which we as competition enforcers generally view as a positive.’ ‘Algorithms and Coordinated Effects’, Remarks of Commissioner Terrell McSweeny (22 May 2017), available at

[24] United States v. Custom Wristbands, Inc 4:17-cr-00510 (SD Tex 22 Aug. 2017).

[25] Cara Bayles and Bill Donahue, ‘Waymo-Uber Trial Delayed Over Claims of Hidden Evidence’, Law360 (28 Nov. 2017), available at https://www.law360.com/articles/988725/waymo-uber-trial-delayed-over-claims-of-hidden-evidence.

[26] Makan Delrahim, ‘Don’t Stop Believin’: Antitrust Enforcement in the Digital Era’, Department of Justice, at 5-6 (19 Apr. 2018), available at https://www.justice.gov/opa/speech/assistant-attorney-general-makan-delrahim-delivers-keynote-address-university-chicagos; Makan Delrahim, ‘“Start Me Up”: Start-Up Nations, Innovation, and Antitrust Policy’, Remarks at the University of Haifa (17 Oct. 2018), available at https://www.justice.gov/opa/speech/assistant-attorney-general-makan-delrahim-delivers-remarks-university-haifa-israel.

[27] Deputy Assistant Attorney General Roger Alford Delivers Remarks at King’s College in London, Justice News, at 3 (23 Feb. 2018), available at https://www.justice.gov/opa/speech/deputy-assistant-attorney-general-roger-alford-delivers-remarks-kings-college-london.

[28] id.

[29] Bernard (Barry) A Nigro, Jr, ‘“Big Data” and Competition for the Market, Department of Justice’, at 3 (13 Dec. 2017), available at https://www.justice.gov/opa/speech/file/1017701/download.

[30] id. at 4.

[31] Anja Lambrecht and Catherine E Tucker, ‘Can Big Data Protect a Firm From Competition?’ Antitrust Chronicle 1, no. 12 (January 2017).

Get unlimited access to all Global Competition Review content