The Lawsuits
The RealPage MDL consolidates more than 30 class action suits against RealPage Inc. and its clients, who are residential property owners, operators, and managers. The plaintiffs, who are renters of multifamily housing and student housing, allege that RealPage’s software feeds data provided by landlords into a “common algorithm” that calculates optimal prices for units in a specific geographic area. The plaintiffs claim that the landlords coordinated rental pricing by sharing non-public, commercially sensitive information with, and delegating price setting to, RealPage and then pricing units in line with RealPage’s suggested prices 80-90% of the time.
The Yardi case similarly alleges that Yardi Systems, Inc., which created a separate algorithmic pricing tool, conspired with the operators of multifamily rental properties who used its software “to coordinate on setting supracompetitive pricing on multifamily properties” in violation of Section 1 of the Sherman Act.
Both cases allege “hub-and-spoke” price-fixing conspiracies, in which the defendant software companies acted as the hubs through which the users of their products (the spokes) agreed to fix prices with the aid of the respective software’s algorithms.
DOJ’s Statements of Interest
DOJ’s RealPage Statement of Interest opposed the defendants’ motions to dismiss in the case. The Statement positioned algorithms as the latest stage in the evolution of price fixing, which has gone from “in-person handshakes” to “phone and fax, and later to email.” DOJ suggested that algorithms threatened even more competitive harm than these earlier iterations of price fixing, averring that “given the amount of information an algorithm can access and digest, this new frontier poses an even greater anticompetitive threat than the last.” While acknowledging that “not every use of an algorithm to set price qualifies as a per se violation of Section 1,” DOJ argued that when the “joint use of common algorithms [removes] independent decision making,” it must “be subject to the same condemnation as other price-fixing schemes,” with “no legal significance” attached to the fact that an algorithm facilitated the unlawful conduct.
An accompanying Memorandum of Law expanded on DOJ’s position. DOJ argued that, regardless of the methods employed, “Section 1 applies to collaborations that eliminate independent decisionmaking” and prohibits competitors from knowingly sharing competitively sensitive information “with, and then relying on pricing decisions from,” a common agent, whether human or algorithm.
DOJ’s argument proceeded in two parts, tracking Section 1, which prohibits (1) concerted action that (2) unreasonably restrains trade.
Regarding the first prong, DOJ criticized the defendants’ “overly narrow” view of concerted action under the Sherman Act, which it argued included “the joint delegation of competitive decisions” to a common entity, regardless of the existence of a formal agreement and without a requirement of “simultaneous action – or even action that is close in time.” Such concerted action, DOJ further argued, can be proved through either circumstantial or direct evidence, including an invitation to act in concert followed by a course of conduct consistent with accepting such an invitation.
The complaint, DOJ argued, sufficiently alleged concerted action of the type prohibited by the Sherman Act: both in its allegations that RealPage intended to have its users coordinate to increase revenues through raised rents and that its customers went along with the scheme by providing RealPage non-public and competitively sensitive data. These allegations, DOJ said, were bolstered by other allegations related to the monitoring and enforcement of the alleged agreement.
Regarding the second prong of the Section 1 analysis – which requires finding an unreasonable restraint of trade – DOJ argued that horizontal price fixing is per se unlawful and “[t]he analysis is no different simply because a software algorithm is involved.” Since 1940, DOJ said, the Supreme Court “has condemned as per se unlawful agreements to use the same ‘formula underlying price policies,’” and this includes the joint use of a pricing algorithm. The RealPage complaint, DOJ said, likewise satisfied this analytical prong.
First, it alleged a horizontal scheme, because use of the software eliminated independent pricing decisions by competing landlords. DOJ cited Supreme Court precedent that competitors cannot avoid antitrust liability simply by acting through an intermediary, including a vertically related entity. Second, the complaint alleged that the landlords delegated aspects of rental pricing to RealPage’s algorithm, which DOJ asserted is tantamount to agreeing to a common pricing formula. It further alleged that they acted in concert by sharing non-public and competitively sensitive pricing information with RealPage, with the “mutual understanding that other landlords … would do the same.” Evoking an oft-repeated line that dates back to 2017, DOJ’s memo said that “[i]t makes no difference that the confidential pricing information was shared through an algorithm rather than through ‘a guy named Bob” or “whether the competing landlords ever communicated with one another about prices.” Therefore, according to DOJ, whether or not the landlords ever communicated directly with each other about prices, because Section 1 includes wholly tacit agreements, the complaint sufficiently alleged per se unlawful price fixing.
DOJ’s more recent Statement of Interest in the Yardi case – filed jointly with the Federal Trade Commission – echoes and attaches its RealPage Statement of Interest, repeating its view of the “tremendous practical importance” of the “judicial treatment of the use of algorithms in price fixing.” DOJ also challenged the Yardi defendants’ position in their motion to dismiss that retention of pricing discretion is fatal to the price fixing claims, saying that “[a]lthough full adherence to a price-fixing scheme may render it more effective, the effectiveness of the scheme is not a requirement for per se illegality” and citing to court decisions holding that it is per se illegal to fix advertised list or sticker prices.
While these Statements of Interest are among the Antitrust Division’s most recent statements on the topic of algorithmic pricing tools, they are neither its first nor likely its last. DOJ has made similar public proclamations about algorithmic pricing on several other occasions. In February 2023, for instance, the Antitrust Division’s second-in-command said in a speech that “algorithms can lead to tacit or express collusion in the marketplace, potentially resulting in higher prices, or at a minimum, a softening of competition.” And in February 2024, a Division manager pledged that DOJ is “going to be focusing on the intent of the parties, including particularly their intent to use the same algorithm to share information or decide their prices for them and what was the purpose of that.”
Questions Remain
Despite the DOJ’s Statements of Interest staking out its position that algorithmic price fixing is per se unlawful, many questions remain about how DOJ will approach the application of antitrust laws to the use of algorithms in price setting.
Will the Wheel Have a Rim?
The DOJ’s Statements of Interest cite as authority a single Division plea agreement, U.S. v. Topkins – a case in which the conspirators “agree[d] to adopt specific pricing algorithms” – for the proposition that per se prohibitions of the use of a common pricing formula extend to “the use of the same pricing algorithm.”
Setting aside the precedential value of a Division plea agreement, Topkins differs in key ways from the allegations in the RealPage and Yardi complaints. In Topkins, the defendant, an online seller of posters, engaged in direct discussions with his competitors to fix prices, including agreeing to use specific pricing algorithms to coordinate changes in their prices and writing computer code that instructed a competitor’s pricing software to conform to the agreement. In other words, algorithms were used to implement an unlawful agreement in Topkins, but the agreement itself was between humans.
The RealPage and Yardi cases, however, do not allege direct communication between the defendants, much less direct agreements like in Topkins. These cases therefore pose the question of whether the unilateral decision to use a third-party algorithmic price-setting tool, by itself, can demonstrate the requisite intent to join a Sherman Act conspiracy. If not, at what point does the user’s interaction with the algorithm become unlawful? Hub-and-spoke conspiracies require a rim – in the form of an agreement – to complete the wheel. While the law recognizes that the Sherman Act prohibits wholly tacit agreements, there is an open question of what, if any, evidence will convince factfinders of such an agreement, and what countervailing evidence might suggest the absence of one.
Will DOJ bring its own enforcement actions?
It is one thing for DOJ to file Statements of Interest in private litigation to which it is not a party. It will be another if DOJ brings its own case alleging algorithmic price fixing and assumes the elevated burden of proof that attaches to criminal matters, which is how DOJ typically pursues per se violations of Section 1 of the Sherman Act.
While DOJ reportedly has opened an investigation into RealPage, it has not obtained a criminal indictment (or filed a civil complaint) to this point. If DOJ does pursue an enforcement action, the attorneys general of the District of Columbia and Arizona have filed lawsuits against RealPage and landlords that use its software that closely track the allegations in the private lawsuits (North Carolina's attorney general also launched an investigation), and these enforcement actions may provide a limited roadmap to federal prosecutors. But these cases will proceed on a lower burden of proof and therefore will have limited utility for DOJ should it choose to pursue a criminal case against algorithmic price fixing.
All eyes on AIs?
The RealPage and Yardi cases deal with algorithms, which the RealPage Statement of Interest labels “the new frontier.”
But the next frontier is already here with artificial intelligence (AI), and DOJ has been clear that it has its eyes on AI price fixing. As Deputy Attorney General Lisa Monaco said in a recent speech: “[p]rice fixing using AI is still price fixing.” Monaco further asserted that “[o]ur laws will always apply” and declared that DOJ’s “enforcement must be robust.” At the Antitrust Division, Assistant Attorney General Jonathan Kanter has trumpeted the “Project Gretzky” initiative to “skate where the puck is going” with respect to AI.
But AI pricing differs in meaningful ways from algorithm-driven pricing. An algorithm is an automated instruction that executes upon encountering a pre-programmed trigger. Algorithms can be simple, such as an “if-then” command that executes on the trigger, or they can consist of more complex mathematical equations. But all algorithms are deterministic and rely on human input to solve problems. AI, on the other hand, is built upon algorithms, but is able to alter its algorithms and create new ones in response to learned inputs and data, beyond just the data that it was originally programmed to recognize as triggers. Two AI-pricing tools may learn to coordinate with each other by trial and error, without being designed or instructed to collude by their human creators.
In that situation, could DOJ or a private plaintiff prove sufficient intent to support finding an unlawful agreement that violates Section 1 of the Sherman Act and therefore properly might be considered “AI price fixing?” DOJ has signaled that its investigations of AI pricing will use aggressive tools – including covert techniques such as wiretaps and undercover operations -- “to really get at that intent element and figure it out on the front end.” But will evidence of human intent even be sufficient in a situation where autonomous systems learn to adjust prices in relation to competitors when a human did not design or instruct the algorithm to coordinate competitor pricing and may not even realize that the algorithm has learned to optimize pricing in this way?
Even if the Division is successful in influencing judicial doctrine around algorithmic price setting, therefore, the durability of rulings that relate to algorithmic pricing, specifically with regard to their applicability to AI pricing, will remain an open question.
Conclusion
Algorithmic pricing software programs have become a common tool for businesses that engage in dynamic pricing of many consumer products and services. The DOJ has been vocal about its concern that algorithmic pricing tools risk becoming the most insidious instrumentality of price fixers in the history of the Sherman Act. Its Statements of Interest in the RealPage and Yardi litigations leave little doubt that DOJ may pursue some uses of algorithmic price-setting tools as per se unlawful violations of Section 1 of the Sherman Act. It remains to be seen, however, how the DOJ will grapple with several open questions and – if and when it does – whether judges and factfinders will agree with its positions.
