Colorado Technology Law Journal > Volume 23 > Issue 1 > Fixing Algorithmic Pricing? Competition Concerns and Solutions

Fixing Algorithmic Pricing? Competition Concerns and Solutions

By /

Henry Hauser[1]*

We are amidst a significant shift in how companies price their products and services. A rapidly increasing number of firms use pricing algorithms to recommend or determine prices. Like many technological transitions, the widespread adoption of pricing algorithms can raise important questions and implicate significant tradeoffs.

There are three core competitive scenarios around algorithmic pricing: (1) human agreements to fix prices with algorithmic implementation, (2) sharing competitively sensitive data through a shared algorithm to recommend or set prices or production, and (3) feeding public data into an algorithm to recommend or set prices. Each scenario reflects unique legal and factual considerations. Accordingly, each requires a tailored solution to address specific competitive concerns.

Part I of this article explains the legal framework for evaluating claims of collusion under Section 1 of the Sherman Antitrust Act. Part II analyzes three different scenarios involving algorithmic pricing. Part III discusses legislative proposals to address algorithmic pricing. Part IV offers my conclusion that each of the scenarios discussed in Part II requires a solution that is narrowly tailored.

 

Introduction

I. Legal Framework of the Sherman Act, Section 1

A. Concerted Action

B. Unreasonable Restraint of Trade

C. Interstate or Foreign Commerce

II. Competitive Concerns Around Algorithmic Pricing

A. Human Agreement to Fix Prices with Algorithm Implementation

B. Sharing Competitively Sensitive Data Through a Shared Algorithm

C. Feeding Public Data into an Algorithm

III. Proposals to Address Competitive Concerns

A. Preventing Algorithmic Collusion Act

B. Preventing Algorithmic Facilitation of Rental

Housing Cartels Act

C. Technical solutions

Conclusion

 

Introduction

We are amidst a significant shift in how companies price their products and services. A “rapidly increasing” number of firms use pricing algorithms to recommend or determine prices.[2] Pricing algorithms can leverage a litany of inputs, including costs, competitor pricing, market supply, inventory, production levels, capacity constraints, consumer demand, efficiencies, and business objectives. Like many technological transitions, the rapid advancement and widespread adoption of pricing algorithms can raise important questions and implicate significant tradeoffs.[3]

However, there is a striking lack of consensus over the nature and scope of these issues, including whether and how they can be addressed under the Sherman Antitrust Act (“Sherman Act”). A review of the legislative rhetoric, case law, and academic literature reveals three core competitive scenarios around algorithmic pricing: (1) human agreements to fix prices with algorithmic implementation, (2) sharing competitively sensitive data through a shared algorithm to recommend or set prices or production, and (3) feeding public data into an algorithm to recommend or set prices. As discussed below, each scenario reflects unique legal and factual considerations. Accordingly, each requires a tailored solution to address specific competitive concerns.

Some scholars and politicians believe legislative reform would be helpful to address competitive concerns arising from algorithmic pricing.[4] For example, in February 2024, Senator Amy Klobuchar, Chairwoman of the Senate Judiciary Subcommittee on Competition Policy, Antitrust, and Consumer Rights, introduced the “Preventing Algorithmic Collusion Act.”[5] Senator Klobuchar states that this bill would “[c]lose a loophole in current law by presuming a price-fixing ‘agreement’ when competitors share completely sensitive information through a pricing algorithm.”[6] In addition, Senator Wyden and Senator Welch have introduced the “Preventing the Algorithmic Facilitation of Rental Housing Cartels Act,” which would impose per se illegality on certain conduct in rental real estate markets. As discussed below, there are significant legal implications of per se illegality under the Sherman Act.

Others believe legislative reform is premature and unnecessary. In their view, “the expanding use of algorithms raises familiar issues that are well within the existing canon” of antitrust law and enforcement.[7] These commentators believe the Sherman Act is sufficient to address any harm to competition.

Still, others argue that algorithmic pricing is economically efficient, could have potential benefits,[8] and should not be assumed to cause competitive harm. They highlight that algorithms can enable firms to adjust the prices of products and services in real-time to reflect rapidly shifting market conditions. According to proponents, algorithms can also “save resources” and “lower barriers to entry or expansion in a given field” providing opportunities for new firms to emerge as disruptors by tracking and besting incumbents.[9]

Part I of this article explains the legal framework for evaluating claims of collusion under Section 1 of the Sherman Antitrust Act. Part II analyzes three different scenarios involving algorithmic pricing, including recent antitrust litigation reflecting these scenarios. Part III discusses the various legislative proposals to address algorithmic pricing. Part IV offers my conclusion that each of the three scenarios discussed in Part II requires a solution that is narrowly tailored. Stated differently, a one-size-fits-all approach is not optimal.

  1. Legal Framework of the Sherman Act, Section 1

Section 1 of the Sherman Act prohibits agreements, contracts, and combinations that unreasonably restrain trade. To have a successful claim under Section 1, a plaintiff must prove (1) concerted action that (2) unreasonably restrains trade and (3) affects interstate or foreign commerce.[10] Each element is discussed below.

  1. Concerted Action

Under Section 1, plaintiffs must show an agreement or concerted action. “Congress treated concerted behavior more strictly than unilateral behavior” based on its belief that it “inherently is fraught with anticompetitive risk” and “deprives the marketplace of the independent centers of decision making that competition assumes and demands.”[11]

“‘The crucial question’ is whether the challenged anticompetitive conduct ‘stem[s] from independent decision or from an agreement, tacit or express.’”[12] “[T]here must be evidence that tends to exclude the possibility that . . . [defendants] were acting independently.”[13] A combination of direct or circumstantial evidence may be used to show that defendants “‘deprive[d] the marketplace of independent centers of decisionmaking,’ and therefore of ‘diversity of entrepreneurial interests,’ and thus of actual or potential competition.”[14]

One way plaintiffs can satisfy their burden is by establishing “plus factors” to support an inference of a conspiracy.[15] Plus factors are “economic actions and outcomes that are largely inconsistent with unilateral conduct but largely consistent with explicitly coordinated action.”[16] These include uniformity of tactics, behavior contrary to economic self-interest, conduct that is too complicated to be explained by unilateral behavior, failure to price based on relative cost advantages, past history of collusion, opportunities to meet and communicate with competitors, and market concentration.[17]

By contrast, evidence that competitors have increased their prices in tandem or in lockstep, by itself, is generally insufficient to prove an agreement to fix prices. “Even conscious parallelism, a common reaction of firms in a concentrated market that recognize their shared economic interests and their interdependence with respect to price and output decisions is not in itself unlawful.”[18] For example, common supply and demand conditions may be driving similar pricing. As such, “parallel price increases, without more” are not enough to show an agreement to fix prices.[19]

The fact that competitors have exchanged pricing information, without more, is also insufficient to support an inference of an agreement to fix prices. “[I]nformation exchanges help to establish an antitrust violation when either (1) the exchange indicates the existence of an express or tacit agreement to fix or stabilize prices, or (2) the exchange is made pursuant to an express or tacit agreement that is itself a violation of §1 under a rule of reason analysis.”[20] Notably, the second category requires a showing of competitive effects, whereas the first category is a per se violation and does not.[21]

While proof of parallel pricing alone is insufficient to infer an agreement to fix prices, some courts have inferred anticompetitive agreements where information sharing was intended to, and did in fact, render price increases “more effective by ensuring that competitors could quickly learn of, and respond to” the price increase.[22] Nonetheless, courts may be reluctant to infer a conspiracy from the publication of information where such an inference could “significantly deter important legitimate conduct.”[23]

  1. Unreasonable Restraint of Trade

In determining whether a restraint of trade is “unreasonable” under the Sherman Act, most courts employ either the rule of reason analysis or the per se standard.[24] The rule of reason standard “requires courts to conduct a fact-specific assessment of ‘market power and market structure . . . to assess the [restraint]’s actual effect’ on competition.”[25] The goal is to construct “an enquiry meet for the case, looking to the circumstances, details, and logic of a restraint”[26] to differentiate “restraints with anticompetitive effect that are harmful to the consumer and restraints stimulating competition that are in the consumer’s best interest.”[27] The rule of reason seeks to strike a balance between condemning anticompetitive conduct while leaving in place a restraint that “merely regulates, and perhaps thereby promotes competition.”[28]

The rule of reason analysis consists of a four-step burden-shifting framework. First, the plaintiff must show that the challenged restraint has a substantial anticompetitive effect in a relevant market.[29] Second, if the plaintiff meets its initial burden, the defendant must offer a “procompetitive rationale” for their conduct that is not pretextual.[30] If the defendant’s asserted rationale is not cognizable, then the plaintiff prevails. Third, if the defendant meets its burden, the plaintiff can prevail by showing that “substantially less restrictive means exist to achieve any proven procompetitive benefit.”[31] Even when substantially less restrictive means do not exist, many courts balance anticompetitive harm against procompetitive benefits in a “fourth step” that leading scholars describe as “consistent with antitrust history and necessary to effectuate the principles underlying standard antitrust analysis.”[32]

Importantly, not all conduct is subject to the rule of reason analysis. Some practices are so “plainly anticompetitive”[33] and lacking in “any redeeming virtue”[34] that they are considered illegal per se. Examples of conduct arising under the per se category include price fixing,[35] bid rigging,[36] and market allocation conspiracies.[37] The choice of analytical standard has significant implications for the ultimate outcome in antitrust cases.[38]

  1. Interstate or Foreign Commerce

The third element, which requires interstate or foreign commerce, is generally easy for plaintiffs to satisfy. “The reach of the Sherman Act has expanded with the scope of congressional power under the Interstate Commerce Clause of the Constitution.”[39] This construction covers “virtually any economic act in our infinitely interconnected national economy.”[40]

Commentators note that the Supreme Court “has routinely upheld applications of the Sherman Act to restraints that harm consumers only locally . . . .”[41] In Summit Health, for example, the Court enabled “the Justice Department to inject a competitive element into numerous lines of business that were previously considered too localized for jurisdiction under the Act.”[42] Even “[w]holly local business restraints can produce the effects condemned by the Sherman Act.”[43] If interstate commerce “feels the pinch” of the defendant’s conduct, “it does not matter how local the operation which applies the squeeze.’”[44]

Therefore, we can presume that the interstate or foreign commerce requirement would be satisfied for cases involving algorithmic pricing, even where the affected market is local in nature. This element is straightforward to establish when pricing software is marketed, sold, or used in a different state from where it is developed.

  1. Competitive Concerns Around Algorithmic Pricing

The critics of algorithmic pricing raise competitive concerns along a spectrum with three main categories. First, humans may enter an unlawful agreement to fix prices, rig bids, or allocate markets and then use pricing software to carry out their conspiracy. Second, shared algorithms with access to competitively sensitive data from more than one company may leverage that data to recommend or set prices. Third, independent algorithms with access only to public data could recommend or set prices that are above the competitive level. Each scenario is discussed below.

  1. Human Agreement to Fix Prices with Algorithm Implementation

The first scenario, which occurs when humans agree to fix prices and leverage algorithms to do the dirty work of implementing, monitoring, and enforcing their agreement, is plainly covered under Section 1 of the Sherman Act.[45] The agreement to fix prices satisfies the concerted action requirement of Section 1, before algorithms are ever involved.[46] Proving this agreement requires no technical knowledge or analysis of the algorithm, although the algorithm can sometimes provide additional direct or circumstantial evidence of concerted action.[47]

An example of this scenario is the Department of Justice Antitrust Division’s prosecution of anticompetitive conduct involving posters.[48] There, two companies had been competing aggressively to sell posters on an e-commerce marketplace.[49] Other competitors’ posters were priced much higher.[50]

Prior to the conspiracy, each company had programmed its software to beat all competitors by a defined amount. The result was a race to the bottom, with some posters being listed for as little as a penny plus shipping. This caused headaches for the two firms because it resulted in significant price and margin erosion. But, instead of competing, they opted to collude. Through the exchange of emails and telephone calls, each company agreed to work together to increase the prices of posters sold online. At that point, the parties had entered a per se unlawful conspiracy to fix prices in violation of the Sherman Act.[51]

To execute their conspiracy, the defendants turned to their pricing algorithms. Firm 1 programmed its algorithm to ignore Firm 2’s pricing but beat all other sellers’ pricing. Firm 2 programmed its algorithm to match Firm 1’s pricing. The result was that the defendants went from pricing against each other to pricing only against the remaining (higher priced) fringe competitors. As a result, prices rapidly increased.

The Antitrust Division learned of this conduct[52] and swiftly prosecuted it.[53] Enforcers secured two individual felony pleas and one corporate plea.[54] The posters prosecution shows that antitrust laws clearly cover concerns around algorithmic pricing, where humans agree to fix prices and then use algorithms to implement, monitor, or enforce their agreement. As former Acting Associate Attorney General William Baer stated when the first guilty plea was announced, an agreement to fix prices runs afoul of the Sherman Act “whether it occurs in a smoke-filled room or over the Internet using complex pricing algorithms.”[55] Price fixing is per se illegal, and the presence of algorithms does not change that result.[56]

However, it is important to recognize that pricing algorithms offer a new wrinkle in cartel investigations. For example, algorithms can make anticompetitive agreements more durable by helping solve what are known as “cartel problems.”[57] Although the agreement to fix prices is what is forbidden, game theory holds that successful conspirators try to solve three challenges.[58] First, they must all agree on terms. “For example, they must reach a consensus to charge a fixed price, not pursue each other’s customers, or cut production.”[59] Second, they must monitor for compliance to ensure everyone is following their agreement.[60] A cartel won’t be effective if conspirators never implement or follow their agreement.[61] Third, cartelists may impose a “credible threat of retaliation” against companies that deviate from the agreement.[62] If one company tries to increase sales by discounting below the agreed-on price, the other conspirators must have some way to discipline it.

Unless these challenges are solved, the cartel may not long endure. Relevant here, pricing algorithms can quickly observe, synthesize, and respond to vast amounts of sales, purchases, and transaction data. This could make it easier to know whether a conspirator is cheating on its cartel partners by selling below a fixed price or by producing above an allocated cap.[63] It also makes it easier for firms “to monitor and match” each other on price or production.[64]

When algorithms are involved, another enforcement concern is that companies need not meet or even communicate directly to demonstrate to each other that they are complying with the agreement.[65] This means that proving concerted action through direct evidence may be more difficult when algorithms are involved.

On the other hand, the use of pricing algorithms could offer additional evidence of concerted action. Lines of code and commands that are programmed into a pricing algorithm “could contain data regarding parallel pricing or lockstep production cuts.”[66] Further, computer code could demonstrate the implementation of a cartel or reflect risk and reward features designed to punish companies that “cheat” on their anticompetitive agreement.

  1. Sharing Competitively Sensitive Data Through a Shared Algorithm

While antitrust law is already well suited to address situations where there is an agreement to fix prices and the conspirators use algorithms to effectuate the agreement,[67] some see open questions as to whether competition law reaches other concerns around algorithmic pricing. I now turn to the scenario where competitors, while not explicitly agreeing to fix prices, feed their competitively sensitive data into a shared pricing algorithm that recommends or sets prices. Reflecting the concerns arising under this scenario, Antitrust Division Principal Deputy Assistant Attorney General Doha Mekki has warned that “[w]here competitors adopt the same pricing algorithms . . . studies have shown that these algorithms can lead to tacit or express collusion in the marketplace, potentially resulting in higher prices, or at a minimum, a softening of competition.”[68]

It is unclear whether a change in law is technically necessary to address these concerns, as it could be addressed as a “hub-and-spoke” conspiracy.[69] Hub-and-spoke cases can arise when a company at one level of the supply chain facilitates unlawful agreements between competitors at a different level in the supply chain by serving as a conduit for the exchange of competitively sensitive information or impermissible communications. For example, a supplier that relays the pricing information of one reseller to other resellers could face antitrust exposure under this theory. In that scenario, the supplier (the “hub”) has facilitated a per se unlawful horizontal agreement between competing resellers (the “rim”) through bilateral communications (the “spokes”) that relay competitive information and signals between horizontal competitors. Another example of a hub-and-spoke conspiracy is when a vertically related entity invites horizontal competitors to participate in a common plan, and those competitors then act consistent with that invitation. The Federal Trade Commission and the Department of Justice have prevailed against Toys “R” Us and Apple Inc., respectively, on hub-and-spoke charges.[70] Notably, each case was analyzed under the per se standard despite the significant involvement of vertically related entities.

Courts have long held that plaintiffs can sufficiently allege a price-fixing agreement through circumstantial evidence.[71] For example, plaintiffs can allege antitrust plus factors to support an inference of conspiracy.[72] As noted above, these include “economic actions and outcomes that are largely inconsistent with unilateral conduct but largely consistent with explicitly coordinated action[,]”[73] such as uniformity of tactics, behavior contrary to economic self-interest, or a “departure from prior practice.”[74]

Therefore, application of existing case law suggests that legislative reform to address this scenario is not technically necessary. However, it could be useful to ensure consistent judicial outcomes under the per se standard.

  1. Feeding Public Data into an Algorithm

During a recent Senate Hearing, William Baer, former head of the Department of Justice’s Antitrust Division during the Obama administration, shared his views on another scenario. Baer expressed a concern that algorithms could gather public pricing information regarding a company’s competitors and react by terminating discounts or stabilizing or increasing prices.[75] The concern here is that profit-maximizing algorithms could reach collusive results, even without (1) an advance agreement to fix prices, (2) access to competitively sensitive information from multiple firms, or (3) competitors using a shared algorithm to recommend or set prices.[76]

It is unclear whether algorithms could achieve such a result based on existing technology.[77] Although studies have shown that pricing algorithms can result in higher prices,[78] others suggest that when firms independently adopt different pricing algorithms rather than shared algorithms, this does not always lead to higher average prices.[79] Some commentators argue that concerns around independent algorithms are currently overstated due to the “limited capabilities of algorithmic communication.”[80] However, given the rapid advancements in machine learning technology, concerns that this will become a competition problem in the near future should not be dismissed as premature.[81]

Turning to the case law, an antitrust class action lawsuit against Las Vegas Strip hotel operators shows the difficulty of bringing cases involving algorithms that—in the court’s view—do not leverage competitors’ pooling of confidential or proprietary information. In Gibson v. MGM Resorts International, the plaintiffs alleged that hotel operators violated Section 1 by “artificially inflating the price of hotel rooms after agreeing to all use [a] pricing software.”[82] The plaintiffs further claimed that the defendants shared current pricing information with each other.[83]

However, the Gibson court dismissed the case with prejudice[84] largely because it interpreted the complaint as not alleging that competitors had pooled confidential or proprietary information or that competitors were required to accept the pricing recommendations that the software generated.[85]

  1. Proposals to Address Competitive Concerns
  2. Preventing Algorithmic Collusion Act

The Preventing Algorithmic Collusion Act (PACA) would presume concerted action under Section 1 of the Sherman Act when competitors share competitively sensitive information through a pricing algorithm. This presumption would apply to “any pricing algorithm that uses, incorporates, or was trained with nonpublic competitor data.”[86] The proposed bill would also require companies to disclose when they use algorithms to set prices.

PACA also directs the Federal Trade Commission to “study the use of pricing algorithms”[87] and their impact on competition. Such a study could be conducted through the agency’s investigative authority under Section 6(b) of the Federal Trade Commission Act.[88] Section 6(b) authorizes the Commission to require companies to submit “reports or answers in writing to specific questions” regarding an entity’s “organization, business, conduct, practices, management, and relation to other corporations, partnerships, and individuals.”[89] It further empowers the agency to conduct “wide-ranging studies that do not have a specific law enforcement purpose.”[90] In recent years, the FTC has issued 6(b) orders to better understand a myriad of markets, including social media and video streaming,[91] artificial intelligence,[92] pharmacy benefit management,[93] and broadband.[94] The issuance of 6(b) orders can be an efficient alternative to bringing an enforcement action before enforcers fully understand the market dynamics at issue.

Senator Klobuchar’s proposed legislation could help ensure consistent judicial outcomes, provide clear guidance to software developers and users, and prevent and deter anticompetitive conduct involving pricing algorithms that leverage “nonpublic competitor data.”[95]

Targeted solutions can deter anticompetitive conduct while leaving other behavior untouched. Here, PACA identifies a specific competitive concern (algorithms that use, incorporate, or are trained on “nonpublic competitor data”), directs regulators to conduct further study and investigation, and addresses a narrow range of conduct with a clear nexus to the underlying competitive concerns.

  1. Preventing Algorithmic Facilitation of Rental Housing Cartels Act

Senator Wyden and Senator Welch’s Preventing the Algorithmic Facilitation of Rental Housing Cartels Act (“Wyden Bill”), in some ways, goes further than PACA. The bill would make it per se unlawful for a rental property owner to use software that (1) collects “historical or contemporaneous prices, supply levels, or lease or rental contract termination and renewal dates” of rental housing from more than one property owner, (2) analyzes that information “to train an algorithm[,]” and (3) “recommend[s] rental prices, lease renewal terms, or ideal occupancy levels.”[96]

There are three main differences between the PACA and the Wyden Bill. First, PACA applies to all markets, whereas the Wyden Bill applies only to “residential dwelling units.”[97] Second, PACA presumes concerted action when nonpublic competition data is involved, whereas the Wyden Bill goes further and assigns per se liability to a broader range of data inputs. Third and potentially most importantly, the Wyden Bill is not limited to collecting and coordinating competitively sensitive or nonpublic information. Instead, it appears to apply to all prices, supply levels, and contract dates derived from more than one property owner, whether public or not. This could include conduct such as checking another property manager’s public website as part of a competitive benchmarking exercise.[98]

  1. Technical solutions

Several commentators have proposed technical solutions to these various concerns arising from algometric pricing. Some argue that “ex ante regulation has been unsuccessful in the past.”[99] Professor Massarotto suggests that we “connect the collusion problems we deal with in law to problems of agreement on a shared value in computer networks.”[100] In order to detect and prove the existence of cartels utilizing algorithmic pricing software, she proposes that we look in software code for red flags or mechanisms that help companies solve the cartel problems discussed above: forming an agreement, monitoring compliance, and punishing cheating.[101] Specifically, she identifies several elements that can facilitate adopting “cheat tolerant cartels,” enabling enforcers and courts to infer an agreement between competitors, including signed messages (cryptology), broadcasting, leader election, and private channels.[102]

Another potential solution is collusion failsafe code that would prevent algorithms from reaching anticompetitive outcomes. Yet another proposal seeks to ban companies from using competitor prices, whether public or not, to train or run their pricing algorithms.[103] Other proposed technical solutions range from compelling “a supplier to adopt a disruptive algorithm, effectively acting as a maverick” to requiring a “time-lag in pricing algorithms’ response to market conditions.”[104] However, these proposals all entail a heavy hand, and further research is necessary to evaluate their practicality and efficiency. Indeed, “any blanketed restrictions on either the design or the input should be carefully vetted so we do not unintentionally rule out competitive algorithms.”[105]

Research regarding the competitive effect of using public data to train independent pricing algorithms is still in its nascency. Further, it is unclear whether the use of such data has uniform effects across different products and services. For example, the competitive effects in lumber markets may be very different than for digital technology services. For these reasons, it may be premature to consider a legislative solution to this third scenario, which we are only beginning to comprehend.

Conclusion

Tailored solutions are often the best tool for targeting anticompetitive conduct while leaving anodyne practices intact. I offer three observations regarding competitive concerns around algorithmic pricing. First, the Antitrust Division’s successful track record of prosecuting conspiracies where defendants agree to fix prices and use algorithms to implement their agreement shows that Section 1 of the Sherman Act is up to the task of addressing this competitive concern.[106] The wall posters prosecution is the quintessential example that existing laws are sufficient. No additional legislation is necessary.

Second, sharing competitively sensitive information through a shared algorithm to set or recommend prices is already addressed under Section 1. Nonetheless, PACA could provide the additional benefits of consistent judicial treatment and clear guidance to businesses and software developers by clarifying the legal standard for concerted action.

Third, technical solutions, rather than legislative reform, may present the optimal path for addressing concerns that independent algorithms using public data can harm competition. Given the complex dynamics spanning technology, law, and economics, further research into the price effects of using publicly available data to train independent algorithms should be conducted.[107] For example, price effects may be very different depending on the market and industry at issue. A deeper understanding of where the competitive concerns exist will help identify a solution that effectively prevents anticompetitive conduct while leaving other practices intact.

  1. * Henry Hauser is an adjunct professor at the University of Colorado Law School. The views expressed in this essay are those of the author. I am grateful to Professor Sam Weinstein, Ai Deng, and Dayna Zolle Hauser for their very helpful comments. I am also grateful to the Colorado Technology Law Journal for their careful edits and suggestions, especially Nicole Ela, Sydney Weigert, Annaliese Bennett, Jack Shapiro, and Gregory Ruff. ↑
  2. . See Antonio Capobianco, The Impact of Algorithms on Competition and Competition Law, ProMarket (May 23, 2023), https://www.promarket.org/2023/05/23/the-impact-of-algorithms-on-competition-and-competition-law [https://perma.cc/BW29-3DWQ]. (“[T]here is evidence to suggest that the use of pricing algorithms is rapidly increasing.”); Marco Bertini & Oded Koenigsberg, The Pitfalls of Pricing Algorithms, Harv. Bus. Rev. (Sept.-Oct. 2021), https://hbr.org/2021/09/the-pitfalls-of-pricing-algorithms [https://perma.cc/Z7JK-DT33]; Henry Hauser, Shylah Alfonso & Jon Jacobs, Artificial Intelligence in Antitrust Spring Meeting Spotlight, Westlaw Today (Apr. 15, 2024) https://today.westlaw.com/Document/I8a92e5f5fb5c11ee8921fbef1a541940/View/FullText.html?transitionType=Default&contextData=(sc.Default)&VR=3.0&RS=cblt1.0&firstPage=true [https://perma.cc/43UB-TRK3] (This source was written by the author and serves as a reference throughout the present article.) ↑
  3. . Zach Brown & Alexander MacKay, Are Online Prices Higher Because of Pricing Algorithms?, Brookings (July 7, 2022), https://www.brookings.edu/articles/are-online-prices-higher-because-of-pricing-algorithms [https://perma.cc/F4VY-YTYC].(“Economists largely agree that algorithms that adjust prices based on demand conditions and/or costs have the potential to increase efficiency. However, there is growing concern that other aspects of pricing algorithms could reduce competition and increase prices.”); see also Ariel Ezrachi & Maurice E. Stucke, Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy 89-100 (Harv. Univ. Press 2016). ↑
  4. . See Press Release, Wyden and Welch Introduce Legislation to Crack Down on Companies that Inflate Rents with Price-Fixing Algorithms, Ron Wyden United States Senator for Oregon (Jan. 30, 2024), https://www.wyden.senate.gov/news/press-releases/wyden-and-welch-introduce-legislation-to-crack-down-on-companies-that-inflate-rents-with-price-fixing-algorithms [https://perma.cc/M9YY-2U9Q]; see also Preventing the Algorithmic Facilitation of Rental Housing Cartels Act of 2024, S. 3692, 118th Cong. (2024). ↑
  5. . Press Release, Klobuchar, Colleagues Introduce Antitrust Legislation to Prevent Algorithmic Price Fixing, U.S. Senator Amy Klobuchar (Feb. 2, 2024), https://www.klobuchar.senate.gov/public/index.cfm/news-releases?ID=97FCD417-A580-4BF6-BFA9-2549D6B776E8 [https://perma.cc/2U8B-U7R5]; see also Preventing Algorithmic Collusion Act of 2024, S. 3686, 118th Cong. (2024). ↑
  6. . Id. ↑
  7. . See id. ↑
  8. . See Ai Deng, What Do We Know About Algorithmic Collusion Now? New Insights from the Latest Academic Research (July 29, 2023), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4521959 [https://perma.cc/6V9M-PEWE]. ↑
  9. . See Michal S. Gal and David Rubinfeld, Algorithms AI, and Mergers, 85.3 Antitrust L. J. 683, 687 (2024). ↑
  10. . 15 U.S.C. § 1 (1890); See Hobart-Mayfield, Inc. v. Nat’l Operating Comm. on Standards for Athletic Equip., 48 F.4th 656, 663 (6th Cir. 2022); Realcomp II, Ltd. v. FTC, 635 F.3d 815, 824 (6th Cir. 2011); In re Ins. Brokerage Antitrust Litig., 618 F.3d 300, 314 (3d Cir. 2010). ↑
  11. . Copperweld Corp. v. Indep. Tube Corp., 467 U.S. 752, 768-69 (1984). ↑
  12. . Bell Atl. Corp. v. Twombly, 550 U.S. 544, 553 (2007) (quoting Theatre Enter., Inc. v. Paramount Film Distrib. Corp., 346 U.S. 537, 540 (1954)). ↑
  13. . Monsanto Co. v. Spray-Rite Serv. Corp., 465 U.S. 752, 764 (1984). ↑
  14. . Am. Needle, Inc. v. Nat’l Football League, 560 U.S. 183, 195 (2010) (citing Fraser v. Major League Soccer, L.L.C., 284 F.3d 47, 56 (1st Cir. 2002); Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 214 (D.C. Cir. 1986). ↑
  15. . See In re Musical Instruments & Equip. Antitrust Litig., 798 F.3d 1186, 1194 (9th Cir. 2015) (“This court has distinguished permissible parallel conduct from impermissible conspiracy by looking for certain ‘plus factors.’” (citing In re Citric Acid Litig., 191 F.3d 1090, 1102 (9th Cir.1999))); Quality Auto Painting Ctr. Of Roselle, Inc. v. State Farm Indem. Co., 917 F.3d 1249, 1262 (11th Cir. 2019) (“[P]lus factors ‘remove [a plaintiff’s] evidence from the realm of equipoise and render that evidence more probative of conspiracy than of conscious parallelism.’” (quoting Williamson Oil Co. v. Philip Morris USA, 346 F.3d 1287, 1301 (11th Cir. 2003))). ↑
  16. . See In re Musical Instruments, 798 F.3d at 1194. ↑
  17. . See, e.g., id.; see also Lifewatch Serv. Inc. v. Highmark Inc., 902 F.3d 323 (3d Cir. 2018); Prosterman v. Am. Airlines, Inc., 747 Fed. Appx. 458, 461 (9th Cir. 2018); In re Text Messaging Antitrust Litig., 782 F.3d 867, 871 (7th Cir. 2015). ↑
  18. . Twombly, 550 U.S. at 553-54 (quoting Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 227 (1993)). ↑
  19. . In re Musical Instruments & Equip. Antitrust Litig., 798 F.3d at 1197. ↑
  20. . In re Coordinated Pretrial Proc. in Petroleum Prod. Antitrust Litig., 906 F.2d 432, n.13 (9th Cir. 1990). ↑
  21. . See United States v. Socony Vacuum Oil Co., 310 U.S. 150, n.59 (1940) (stating that conduct may violate Section 1 “though it is not established that the conspirators had the means available for the accomplishment of their objective”). ↑
  22. . In re Coordinated Pretrial Proc. in Petroleum Prod. Antitrust Litig., 906 F.2d at 446. ↑
  23. . Id. at 448. ↑
  24. . There is also an intermediate scrutiny analysis known as the “quick look” test, which courts use when “a restraint is not conclusively presumed illegal . . . but the likelihood of anticompetitive effects is . . . obvious.” Realcomp II, Ltd., F.3d at 825. “Under a quick-look analysis, once a restraint is deemed facially anticompetitive, the burden shifts to its proponent for justification on procompetitive grounds.” Id. However, this standard is applied much less frequently than either the per se or rule of reason approach. ↑
  25. . Ohio v. Am. Express Co., 585 U.S. 529, 541 (2018) (citing Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752, 768). ↑
  26. . California Dental Ass’n v. F.T.C., 526 U.S. 756, 781 (1999). ↑
  27. . Leegin Creative Leather Prod., Inc. v. PSKS, Inc., 551 U.S. 877, 886 (2007). ↑
  28. . Bd. of Trade of the City of Chi. v. United States, 246 U.S. 231, 238 (1918). ↑
  29. . Ohio, 585 U.S. at 541. ↑
  30. . Id. ↑
  31. . Nat’l Collegiate Athletic Ass’n v. Alston, 594 U.S. 69, 100 (2021). ↑
  32. . See Michael A. Carrier, The Four-Step Rule of Reason, 33 Antitrust 50, 50 (2019). ↑
  33. . Nat’l Soc’y of Pro.l Eng’r v. United States, 435 U.S. 679, 692 (1979). ↑
  34. . N. Pac. R.R. Co. v. United States, 356 U.S. 1, 5 (1958). ↑
  35. . United States v. Trenton Potteries Co., 273 U.S. 392, 397 (1927). ↑
  36. . In re Ins. Brokerage Antitrust Litig., 618 F.3d at 336 (citing United States v. Heffernan, 43 F.3d 1144, 1147 (7th Cir.1994)) (Bid rigging and bid rotation “eliminate[s] all competition rather than just price competition”); United States v. Portsmouth Paving Corp., 694 F.2d 312, 317 (4th Cir.1982). ↑
  37. . United States v. Sealy, Inc., 388 U.S. 350, 357­–58 (1967). ↑
  38. . See Carrier, supra note 31, at 51 (between 1999 and 2009, courts dismissed 97% of rule of reason cases at the first stage of the rule of reason inquiry). ↑
  39. . Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application at 266a (Wolters Kluwer, 2024). ↑
  40. . Id. ↑
  41. . See Bruce Johnsen & Moin A. Yahya, The Evolution of Sherman Act Jurisdiction: A Roadmap for Competitive Federalism, 7:2 J. Const. L. 403, 407 (2004); see also Summit Health, Ltd. v. Pinhas, 500 U.S. 322 (1991). ↑
  42. . William F. Detwiler, Redefining Interstate Commerce Jurisdiction under the Sherman Act: Summit Health, Ltd. v. Pinhas, 37 Vill. L. Rev. 373, 408 (1992). ↑
  43. . United States v. Employing Plasterers Assn. of Chi., 347 U.S. 186, 189 (1954). ↑
  44. . Gulf Oil Corp. v. Copp Paving Co., 419 U.S. 186, 195 (1974). ↑
  45. . United States v. Topkins, 3:15-cr-00021-WHO, Information (Dkt. No. 1), https://www.justice.gov/sites/default/files/opa/press-releases/attachments/2015/04/06/topkins_information.pdf [https://perma.cc/U5FP-W8SE]; United States v. Aston, 3:15-cr-00419-WHO, Indictment (Dkt. No. 1), https://www.justice.gov/atr/file/840016/dl?inline [https://perma.cc/Z5UD-RG7N]; United States v. Trod Ltd., 3:15-cr-00419-WHO, Plea Agreement (Dkt. No. 41), https://www.justice.gov/d9/327965.pdf [https://perma.cc/6LB8-LYMX]. ↑
  46. . Topkins, supra note 44, at ¶ 6-7; Aston, supra note 44, at ¶ 2; Trod Ltd., supra note 44, at ¶ 4(b). ↑
  47. . See Topkins, supra note 44, at ¶ 8(c)-(e). ↑
  48. . The author led the wall posters investigation for the Antitrust Division. ↑
  49. . See Claudia Patricia O’Kane & Ioannis Kokkoris, A Few Reflections on the Recent Caselaw on Algorithmic Collusion, 1 Competition Pol’y Int’l Antitrust Chron. 47, 51 (2020). ↑
  50. . See generally John Egerton, DOJ Charges Poster Peddlers With Algorithmic Collusion on Amazon Marketplace, Next TV (Nov. 7, 2018), https://www.nexttv.com/news/doj-charges-poster-peddlers-algorithmic-collusion-amazon-marketplace [https://perma.cc/X92Z-FFB3]. ↑
  51. . Topkins, supra note 44, at ¶ 5; Aston, supra note 44, at ¶ 3; Trod Ltd., supra note 44 at ¶ 2. ↑
  52. . S. Judicial Comm., The New Invisible Hand? The Impact of Algorithms on Competition and Consumer Rights (Dec. 13, 2023), https://www.judiciary.senate.gov/imo/media/doc/2023-12-13_pm_-_testimony_-_baer.pdf [https://perma.cc/55YD-ZD47] (spoken remarks of Hon. Bill Baer). ↑
  53. . Id. ↑
  54. . See Trod Ltd., supra note 44. ↑
  55. . See Former E-Commerce Executive Charged with Price Fixing in the Antitrust Division’s First Online Marketplace Prosecution, Dep’t of Just., Off. of Pub. Affs. (Apr. 6, 2015), https://www.justice.gov/opa/pr/former-e-commerce-executive-charged-price-fixing-antitrust-divisions-first-online-marketplace [https://perma.cc/F998-ASSY]. ↑
  56. . Id. ↑
  57. . See Salil K. Mehra, Price Discrimination-Driven Algorithmic Collusion: Platforms for Durable Cartels, 2 Stan. J. L. & Bus. Fin. (forthcoming 2021). ↑
  58. . See George S. Stigler, A Theory of Oligopoly, 72 J. Pol. Econ. 44, 44-46 (1964). ↑
  59. . Henry Hauser, AI Risks Hindering DOJ Antitrust Cases With Complex Pricing Data, Bloomberg Law (Dec. 18, 2023), https://news.bloomberglaw.com/us-law-week/ai-risks-hindering-doj-antitrust-cases-with-complex-pricing-data [https://perma.cc/2YP4-5UWX]. ↑
  60. . Stigler, supra note 57, at 46. ↑
  61. . Id. ↑
  62. . Gal, supra note 8, at 693; Stigler, supra note 57, at 47-48. ↑
  63. . See Terrell McSweeney & Brian O’Dea, The Implications of Algorithmic Pricing for Coordinated Effects Analysis and Price Discrimination Markets in Antitrust Enforcement, 32 AM. Bar Ass’n Antitrust Mag. 75, 75 (2017) (algorithms may facilitate the stability of certain price-fixing schemes by enabling firms to more quickly detect, and respond to, attempts to cheat on the collusive pricing agreement). ↑
  64. . Price-bots can collude against consumers, The Economist (May 6, 2017), https://www.economist.com/finance-and-economics/2017/05/06/price-bots-can-collude-against-consumers [https://perma.cc/GKX3-FX9T]. ↑
  65. . Id. ↑
  66. . Henry Hauser & Kim Ng et al., Balancing Blockchain’s Rewards With Its Antitrust Risks, Law360 (Jan. 31, 2023), https://www.law360.com/articles/1569473/balancing-blockchain-s-rewards-with-its-antitrust-risks [https://perma.cc/FR2Z-KP9V]. ↑
  67. . See, e.g., Former E-Commerce Executive Charged with Price Fixing in the Antitrust Division’s First Online Marketplace Prosecution, supra 54. ↑
  68. . Doha Mekki, Principal Deputy Assistant Att’y Gen., U.S. Dep’t of Just., Off. of Pub. Affs., Principal Deputy Assistant Attorney General Doha Mekki of the Antitrust Division Delivers Remarks at GCR Live: Law Leaders Global 2023 (Feb. 2, 2023), https://www.justice.gov/opa/speech/principal-deputy-assistant-attorney-general-doha-mekki-antitrust-division-delivers-0 [https://perma.cc/6RTK-FAF2] (citing Zach Brown & Alexander MacKay, Are online prices higher because of pricing algorithms, Brookings (July 7, 2022)); Matthew Leisten, Algorithmic Competition, with Humans (June 2022) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4733318 [https://perma.cc/9QMC-FFY9]. ↑
  69. . See Interstate Circuit, Inc. v. U.S., 306 U.S. 208 (1939). ↑
  70. . See Toys “R” Us, Inc. v. F.T.C., 221 F.3d 928 (7th Cir. 2000); U.S. v. Apple, Inc., 791 F.3d 290 (2d Cir. 2015). ↑
  71. . See In re Musical Instruments & Equip. Antitr. Litig., 798 F.3d 1186, 1194 (9th Cir. 2015) (citing In re Citric Acid Litig., 191 F.3d 1090, 1102 (9th Cir.1999)) (“This court has distinguished permissible parallel conduct from impermissible conspiracy by looking for certain ‘plus factors.’”). ↑
  72. . Id. ↑
  73. . Id. ↑
  74. . Interstate Circuit, Inc., 306 U.S. at 218. ↑
  75. . Hearing on “The New Invisible Hand? The Impact of Algorithms on Competition and Consumer Rights”, 118th Cong. 5 (2023) (statement of Bill Baer, Comm. on the Judiciary, Subcomm. on Competition Policy, Antitrust, and Consumer Rights). ↑
  76. . Similarly, the U.S. Department of Justice, Antitrust Division, in rescinding several longstanding policy statements regarding information exchange, highlighted that “the rise of data aggregation, machine learning, and pricing algorithms . . . can increase the competitive value of historical data.” Doha Mekki, Principal Deputy Assistant Att’y Gen., U.S. Dep’t of Just., Off. of Pub. Affs., Principal Deputy Assistant Attorney General Doha Mekki of the Antitrust Division Delivers Remarks at GCR Live: Law Leaders Global 2023 (Feb. 2, 2023), https://www.justice.gov/opa/speech/principal-deputy-assistant-attorney-general-doha-mekki-antitrust-division-delivers [https://perma.cc/9XU9-PUED]. ↑
  77. . See Salil K. Mehra, Robo-Seller Prosecutions and Antitrust’s Error-Cost Framework, 2 CPI Antitrust Chron., 36, 37 (May 2017) (noting that the “possibility of enhanced tacit collusion or big data turbocharged anticompetitive action remains theoretical.”); Axel Gautier, Ashwin Ittoo & Pieter Van Cleynenbreugel, AI Algorithms, Price Discrimination and Collusion: A Technical, Economic and Legal Perspective, Eur. J. L. Econ. 405, 430 (July 14, 2020); Ai Deng, Algorithmic Collusion and Algorithmic Compliance: Risks and Opportunities, The Glob. Antitr. Instit. Rep. on the Digit. Econ. 964, 966 (Nov. 2020). ↑
  78. . See Stephanie Assad, Robert Clark, Daniel Ershov & Lei Xu, Algorithmic Pricing and Competition: Empirical Evidence from the German Retail Gasoline Market, 132 J. Pol. Econ. 723, 725 (March 2024). ↑
  79. . See Joseph E. Harrington, Jr., The Effect of Outsourcing Pricing Algorithms on Market Competition (July 19, 2021), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3798847 [https://perma.cc/DY7V-N5CE]. ↑
  80. . David Smith & Steven Tadelis, Algorithmic Pricing: What Every Antitrust Lawyer Needs to Know, 22 A.B.A Section of Antitr. L. Pricing Conduct Comm., 2021, at 1, 3 (citing Ulrich Schwalbe, Algorithms, Machine Learning, & Collusion, 14 J. COMPETITION L. & ECON. 568 (2018), https://doi.org/10.1093/joclec/nhz004). ↑
  81. . For a different viewpoint, see Gautier, Ittoo & Van Cleynenbreugel, supra note 76, at 420-21. ↑
  82. . See Gibson v. MGM Resorts Int’l, 2:23-cv-00140-MMD-DJA (D. Nev. May. 8, 2024), at *1. ↑
  83. . See id. at *3-4. ↑
  84. . Id. at *8-9. ↑
  85. . However, it is worth noting that the Gibson case is on appeal to the U.S. Court of Appeals for the Ninth Circuit. The U.S. Department of Justice, Antitrust Division, has filed an amicus brief in that case. ↑
  86. . Preventing Algorithmic Collusion Act, S. 3686, 118th Cong. § 4(a) (2024). ↑
  87. . Preventing Algorithmic Collusion Act, S. 3686, 118th Cong. § 7 (2024). ↑
  88. . 15 U.S.C. § 46(b). ↑
  89. . Id. ↑
  90. . See Fed. Trade Comm’n, A Brief Overview of the Federal Trade Commission’s Investigative, Law Enforcement, and Rulemaking Authority (May 2021) https://www.ftc.gov/about-ftc/mission/enforcement-authority [https://perma.cc/A68X-UBPA]. ↑
  91. . Lesley Fair, FTC Issues 6(b) Orders to Social Media and Video Steaming Services, Fed. Trade Comm’n (Dec. 14, 2020), https://www.ftc.gov/business-guidance/blog/2020/12/ftc-issues-6b-orders-social-media-and-video-streaming-services [https://perma.cc/G9T8-9SAC]. ↑
  92. . FTC Launches Inquiry into Generative AI Investments and Partnerships, Fed. Trade Comm’n (Jan. 24, 2024), https://www.ftc.gov/news-events/news/press-releases/2024/01/ftc-launches-inquiry-generative-ai-investments-partnerships [https://perma.cc/9RMM-LGQ8]. ↑
  93. . FTC Launches Inquiry Into Prescription Drug Middlemen Industry, Fed. Trade Comm’n (June 7, 2022), https://www.ftc.gov/news-events/news/press-releases/2022/06/ftc-launches-inquiry-prescription-drug-middlemen-industry [https://perma.cc/R8LJ-4RN6]; see also United States of America Before The Federal Trade Commission., F.T.C. Matter No. P221200, Order To File A Special Report (2022). ↑
  94. . FTC Seeks to Examine the Privacy Practices of Broadband Providers, F.T.C. (March 26, 2019), https://www.ftc.gov/news-events/news/press-releases/2019/03/ftc-seeks-examine-privacy-practices-broadband-providers [https://perma.cc/F4V5-LMHN]; see also United States of America Before The Federal Trade Commission, F.T.C. Matter No. P195402, Order To File A Special Report. ↑
  95. . Preventing Algorithmic Collusion Act, S. 3686, 118th Cong. § 4(a) (2024). ↑
  96. . Preventing the Algorithmic Facilitation of Rental Housing Cartels Act, S. 3692, 118th Cong. § 2(4)(A) (2024). ↑
  97. . Id. ↑
  98. . Additional economic research is necessary to determine whether foreclosing companies from offering evidence of legitimate business justifications would unduly chill competitive behavior in these markets. As the U.S. Supreme Court stated in the seminal case Broadcast Music v. CBS, “[i]t is only after considerable experience with certain business relationships that courts classify them as per se violations.” There, in declining to impose per se illegality on defendant BMI’s and ASCAP’s issuance of “blanket licenses to copyrighted musical compositions at fees negotiated by them, “the court highlighted that it had “never examined a practice like this one before.” See Broadcast Music, Inc. v. CBS, Inc., 441 U.S. 1, 4 (1979). ↑
  99. . Giovanna Massarotto, Antitrust Needs To Draw on Computer Science To Detect Algorithmic Collusion, Promarket (Jan. 23, 2024), https://www.promarket.org/2024/01/23/antitrust-needs-to-draw-on-computer-science-to-detect-algorithmic-collusion [https://perma.cc/777W-6M53]. ↑
  100. . Id. ↑
  101. . Id. ↑
  102. . Giovanna Massarotto, Using Computer Science to Detect Cheat Tolerant Cartels (Univ. of Pa. Carey L. Sch. Instit. for L. & Econ. 2023). ↑
  103. . See, e.g., Zach Brown & Alexander MacKay, Are Online Prices Higher Because of Pricing Algorithms?, Brookings (July 7, 2022), https://www.brookings.edu/articles/are-online-prices-higher-because-of-pricing-algorithms [https://perma.cc/MED6-DX3U]. ↑
  104. . Capobianco, supra note 1, at 4. ↑
  105. . Deng, supra note 7, at 10 (“In some cases, sophisticated algorithms can lead to much lower prices . . . and can disrupt seeming collusion”). ↑
  106. . See, e.g., Debra D. Bernstein, Can AI engage in price fixing?, Financier Worldwide (Aug. 2023), Can AI engage in price fixing? — Financier Worldwide (“The DOJ had no trouble prosecuting this agreement under traditional US antitrust frameworks because the anticompetitive conduct –humans agreeing to fix prices – was no different than traditional price-fixing agreements.”). ↑
  107. . See e.g., McSweeney and O’Dea, supra note 62, at 76 (“Future research may prove especially valuable in this area.”); Deng, supra note 7, at 10 (“there is still much that we do not know.”). ↑

 

Scroll to Top