From a wink and nod to a point and click – assessing agreement through algorithmic tools

Several major antitrust matters have been brought upon the basis of alleged conspiracies to fix prices through would-be competitors’ use of algorithmic pricing tools in recent months. These tools amass and process pricing information from competitors in an industry and issue price recommendations aimed at maximizing revenues.

Two matters – relating, respectively, to the use of algorithmic pricing by Las Vegas Strip hotels and by multifamily rental property companies – have now progressed to opinions on motions to dismiss with strikingly different results. To an outsider, it is unclear how much the decisions hang on the particular facts of the case.[1] Numerous other cases are now pending with respect to the use of algorithmic pricing tools in the health care market,[2] indicating that the question of how “agreements” can be inferred from competitors’ use of such pricing tools remains a significant one.

The Las Vegas Hotel action

The District of Nevada addressed algorithmic pricing among Las Vegas Strip Hotels in opinions in October 2023 and May 2024. The actions were brought by a class of directed purchasers of guest rooms in hotels on the Strip. The plaintiffs alleged that defendant hotels had, over a period of years, “agreed to use a shared set of pricing algorithms . . . that recommend supracompetitive prices to the hotel operators.”[3] The plaintiffs alleged that one algorithm’s marketing materials advertised that “pricing recommendations are accepted 90% of the time meaning that the prices for hotel rooms are set by [the algorithm] and not normal market forces.”[4]

Initially, the court dismissed the complaint, finding that plaintiffs failed to adequately plead the pricing algorithms used by specific defendants, and stating that “without an agreement to accept the elevated prices recommended by the pricing algorithm, there is no agreement that could . . . make out a Sherman Act violation . . .”.[5]

Following amendments to the complaint, the district court again dismissed the complaint, this time with prejudice. The court asserted “[t]his case remains a relatively novel antitrust theory premised on algorithmic pricing going in search of factual allegations that could support it.”[6]

The court acknowledged Supreme Court precedent that agreement need not occur at a single point in time, but ultimately determined that because “Defendants began licensing [pricing tools] at different times over an approximately 10-year period and never agreed to charge the prices [pricing tools] recommended to them, the only plausible inference that the Court can draw is that the timing ‘does not raise the specter of collusion.’”[7] The court’s cited basis for concluding that the defendants did not so agree to charge the recommended prices stemmed from a confidential witness’s statement that a pricing tool’s representative “engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendations.”[8]

The court highlighted that plaintiffs failed to allege the exchange of non-public information through the algorithm, but that the algorithm integrated public hotel prices starting in 2015.[9] The court highlighted this fact, stating “consulting your competitors’ public rates to determine how to price your hotel room—without more—does not violate the Sherman Act”[10] and emphasizing “[t]here is nothing unreasonable about consulting public sources to determine how to price your product.”[11]

The Court acknowledged allegations that the machine learning aspect of the pricing tools made use of confidential pricing information, but rejected an inference of conspiracy, stating that the machine learning aspect “merely suggests GuestRev or GroupRev might be compelling to a Hotel Defendant because it offers better pricing recommendations than it used to.”[12] To that end, the Court adopted an analogy offered by defense counsel, comparing the algorithm to an attorney whose experience and knowledge improve with each client, each of whom offer confidential information. Finally, the court rejected plaintiffs’ allegations of a vertical restraint, again pointing to the “never-ending battle” language, this time asserting that the complaint itself alleged that defendants often overrode recommended prices.[13]

Plaintiffs in the action have filed a notice of appeal to the Ninth Circuit.

The opinion appears likely to be challenged on at least a couple of bases. The most significant of these is the exceptional weight that the court placed on the plaintiffs’ allegation stating that the pricing vendor “engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendations.”[14]

The court relies upon this allegation extensively, quoting the “never-ending” language thrice throughout its opinion.[15] In each instance, the court appears to use this quotation to draw its own unpleaded inference – that this battle was “never-ending” because it was often unsuccessful – the court then characterizes its inference as plaintiffs’ allegation.[16]

In contrast, the court did not, in substantial part, discuss the barriers to price modification set forth in the same paragraph of the complaint[17] and dismisses a defendant’s marketing materials indicating that its prices were adopted 90% of the time.[18]

The court was assessing these allegations at the motion to dismiss stage, where it is required to interpret allegations in such a way that draws all reasonable inferences in favor of the non-moving party, here, the plaintiffs.[19] If it does not seem that the only plausible interpretation of the “never-ending battle” allegation is that defendants often rejected prices – in contradiction of one defendant’s own marketing materials – plaintiffs may find success on appellate review.

The rental property action[20]

In In re RealPage Rental Software Antitrust Litigation, the middle district of Tennessee denied the bulk of defendants’ motions to dismiss complaints alleging algorithmic price-fixing filed on behalf of renters of multifamily housing units and students living in student housing. The court granted the motion to dismiss of a single defendant as to a complaint filed by students living in student housing,[21] denied a motion to dismiss filed by defendants who used a pricing tool called LRO, and denied a motion to dismiss filed by all defendants.

Plaintiffs alleged that RealPage’s revenue management solutions program was used to achieve price-fixing. Defendants were alleged to “separately contract with RealPage, paying RealPage periodic fees and, critically, providing RealPage their independent commercially sensitive pricing data.”[22]

LRO Defendants sought to dismiss the complaint, pointing to the Vegas action, but the court distinguished the opinion, stating “the Multifamily Complaint clearly alleges that RealPage's revenue management software inputs a melting pot of confidential competitor information through its algorithm and provides price recommendations based on that private competitor data.”[23] The court similarly rejected defendants’ contention that the absence of identifiable information about individual defendants precluded an inference of conspiracy. The Court found that because the private data was used to generate prices, “[i]t is irrelevant then in what form the Defendants monitor their competitors’ data—they still use that private data through their reliance on RealPage’s pricing algorithm.”[24] Moreover, the court emphasized that if, in fact, LRO did not use private data in generating its prices, the case would be properly resolved after discovery.[25]

Like in the Vegas actions, plaintiffs in RealPage pointed to a defendants’ continual efforts to ensure price adoption. Here, however, the court inferred evidence of agreement from these efforts. The court noted that RealPage’s own marketing materials indicated that defendants would beat the market by adopting their prices, which could only be achieved by ensuring clients actually adopt the recommended prices.[26] Unlike the court in the Vegas actions, the court in RealPage found the defendant’s continual efforts to ensure their prices were adopted – even at a lower rate than promoted by the Vegas defendant – was support for the plaintiffs’ claims.[27]

In its main opinion, the court found that plaintiffs had pleaded a vertical relationship between RealPage and its customers, specifically highlighting allegations that defendants knew their private pricing information would be used in generating prices. Although plaintiffs pointed to a statement by an LRO executive that “[W]e are all technically competitors, LRO helps us to work together ... to make us all more successful in our pricing ... LRO is designed to work with a community in pricing strategies, not work separately,”[28] the court declined to accept it as direct evidence of a conspiracy. Instead, the court continued to assess circumstantial evidence, finding the most critical factor to be defendants’ agreement that their data would be used by RealPage to generate prices and knowledge that their peers’ data would be similarly used.[29] Ultimately, the court concluded that circumstantial evidence of knowledge sufficed to state a claim.[30]

Synthesizing the opinions

The opinions are best reconciled on a public/non-public dichotomy. Hotels signal prices publicly, allowing algorithms to incorporate these public prices on a constant basis; apartments typically do not have publicly advertised prices. Thus, the RealPage action proceeds to discovery, the Vegas actions are dismissed.

This standard falls short of the view urged by the Department of Justice in its statement of interest in RealPage, underscoring that “the alleged scheme constitutes price fixing regardless of whether the competing landlords ever communicated with one another about prices.”  One could reason that apartment companies with market power would likely not hesitate to signal prices publicly, particularly in markets where the opportunity for new development is most limited. Companies marketing algorithmic pricing tools will simply know not to identify the competitors whose information is being used and be opaque as to the means used to generate a price and what happens to the data they are given.

One view is that the tools involved have the problematic capacity to achieve aligned prices through “conscious parallelism” that results in reduced – or eliminated – competition. Not even the paradigmatic smoke-filled room could achieve price alignment so immediately and so thoroughly. The effects could be both more immediate and more competition-stifling.

Both cases are premised upon the conclusion that an agreement to accept prices from the algorithm is a prerequisite to stating a claim; this falls short of the standard urged by the Department of Justice in its statement of interest in RealPage, which stated “it suffices to show that RealPage proposed the price-fixing scheme to competing landlords, who were each aware that its competitors were also being invited to participate in the scheme, and the competitors adhered to it—generating a common understanding among the competitors that they would increase prices collectively by using RealPage.” And, of course, even a subscriber intending to undercut the algorithmic pricing tool is still likely narrowing the range of competition by making use of its rivals’ data.

Although the courts have drifted far from U.S. v. Container Corp., its acknowledgment that even informal agreements that “chill… the vigor of price competition”[31] can be illegal is one that some will argue should apply equally here. The threat of using tools to achieve that which would be considered anticompetitive if done by people – or achieving anticompetitive aims no human could achieve – is a real one that a mere public/private distinction may not sufficiently address.

*Timothy Kearns is a Partner in Washington, D.C.

Footnotes

[1] Although Hausfeld is counsel in the RealPage matters, the author is not involved in any matter discussed herein; his analysis is based solely upon materials in the public record.
[2] See Adventist Health Sys. Sunbelt Healthcare Corp. v. MultiPlan, Inc., 1:23-cv-7031 (S.D.N.Y. Aug. 9, 2023); Allegiance Health Mgmt. v. MultiPlan, Inc., 1:24-cv-3223 (N.D. Ill. Apr. 22, 2024); Live Well Chiropractic PLLC v. Multiplan, Inc., 1:24-cv-03680 (N.D. Ill. May 6, 2024); Shoshany v. MultiPlan, Inc., 1-24-cv-5201 (N.D. Ill. June 21, 2024).
[3] Gibson v. MGM Resorts Int’l, 2:23-cv-140-MMD-DJA, 2023 WL 7025996, *1 (D. Nev. Oct. 24, 2023)
(“Gibson I”).
[4] Gibson v. Cendyn Group LLC, 2:23-cv-140-MMD-DJA, Dkt. 1 at ¶12 (D. Nev. Jan. 25, 2023).
[5] Id., at *3.
[6] Gibson v. Cendyn Group, LLC, 2:23-cv-140-MMD-DJA, 2024 WL 2060260, *3 (D. Nev. May 8, 2024) (“Gibson II”)
[7] Id. at *4.
[8] Id. at *3 (D. Nev. May 8, 2024) (“Plaintiffs … indeed allege … that Cendyn has difficulty getting its customers to accept the prices it recommends in GuestRev and GroupRev. (ECF No. 144 at 10 (‘CW 1 stated that Rainmaker engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendations[...]’).)”).
[9] See Gibson v. Cendyn Group LLC, 2:23-cv-140-MMD-DJA, Dkt. 144, at ¶137 (D. Nev. Nov. 25, 2023) (“On a daily basis, Revcaster collects advertised rate information from a hotel’s competitors.”).
[10] Id.
[11] Id. at *5.
[12] Id. at *6.
[13] Id., at *9 (“Plaintiffs do not allege that Hotel Defendants are required to accept the prices that GuestRev and GroupRev . . . recommend to them—and indeed allege that the recommendations are often rejected.”)
[14] Id, at *3 (“Plaintiffs … indeed allege … that Cendyn has difficulty getting its customers to accept the prices it recommends in GuestRev and GroupRev. (ECF No. 144 at 10 (‘CW 1 stated that Rainmaker engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendations[...]’).)”).
[15] Id.; Id. at *7 (“Plaintiffs allege in the FAC that Hotel Defendants are not required to accept the prices
that GuestRev proposes for their hotels. (ECF No. 144 at 10 (alleging that customers may override GuestRev's proposed prices, and indeed, that they often did because ‘Rainmaker engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendation’)”); Id. at *8 (“Plaintiffs go to some lengths in terms of allegations included in the FAC to allege that GuestRev's user interface is set up to encourage customers to accept GuestRev's pricing recommendations . . .but these allegations are ultimately contradicted by Plaintiffs’ allegation that “Rainmaker engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendation[.]” (ECF No. 144 at 10.))”)
[16] Id. at *7 (“(ECF No. 144 at 10 (alleging that customers may override GuestRev's proposed prices, and indeed, that they often did because ‘Rainmaker engaged in a ‘never-ending battle’ to convince clients not to override its pricing recommendation’)”) (emphasis added).
[17] See Gibson v. Cendyn Group, LLC, 2:23-cv-140, ECF. 144 at ¶8.
[18] Gibson II, 2024 Wl 2060260, at *8.
[19] Gibson II, 2024 WL 2060260, *2 n.4.
[20] Hausfeld LLP serves as co-lead counsel on behalf of direct purchasers in the multifamily rental property action. The author has no involvement in the matter and his analysis is based solely upon materials in the public record in the matter.
[21] In re: RealPage, Inc., Rental Software Antitrust Litig. (No. II), No. 3:23-md-03071, 2023 WL 9004816 (M.D. Tenn. Dec. 28, 2023) (“Student Opinion”).
[22] In re: RealPage, Inc., Rental Software Antitrust Litig. (No. II), No. 3:23-md-03071, 2023 WL 9002991, at *2 (M.D. Tenn. Dec. 28, 2023) (“LRO Opinion”)
[23] Id..at *3.
[24] Id. at *4.
[25] Id. at *5.
[26] In re: RealPage, Inc., Rental Software Antitrust Litig. (No. II), No. 3:23-md-03071, 2023 WL 9002991, 2023 WL 9004806 (M.D. Tenn. Dec. 28, 2023), at *3.
[27] See id. (quoting allegation that “Pricing Advisors often spen[d] considerable time ... educat[ing] clients on the pricing methodology and associated benefits of accepting all, or almost all RealPage pricing recommendations.”)
[28] Id. at *10.
[29] Id. at *17.
[30] Id.
[31] U.S. v. Container Corp., 393 U.S. 333, 337 (1969).