What are you looking for?
ICLE Senior Scholar Lazar Radic took part in a recent panel hosted by the Dynamic Competition Initiative on interoperability and its impact on platform competition . . .
ICLE Senior Scholar Lazar Radic took part in a recent panel hosted by the Dynamic Competition Initiative on interoperability and its impact on platform competition and innovation. Video of the full event is embedded below.
Source: Dynamic Competition Initiative
When one thinks of America’s war in the 1960s during the administration of President Lyndon Johnson, the Vietnam War comes to mind. The American government . . .
When one thinks of America’s war in the 1960s during the administration of President Lyndon Johnson, the Vietnam War comes to mind. The American government spent about $175 billion and the lives of about 58,000 men in a futile attempt to stop the communist takeover of South Vietnam. But as costly as this war was, it pales in comparison with the other war President Johnson, and all subsequent presidents have fought in vain—the “war on poverty.” In an address to Congress in January 1964, President Johnson declared an “unconditional war on poverty.” The effort has been almost entirely futile and has cost the American taxpayers more than $30 trillion. This is more than three times the cost of all military wars from the American Revolution to the War on Terror.
Source: Gazeta do Povo
ICLE Chief Economist Brian Albrecht delivered remarks to Hillsdale College’s Hillsdale Praxis program on the topic “Market Power: How Competition Generates Prosperity.” Video of the . . .
ICLE Chief Economist Brian Albrecht delivered remarks to Hillsdale College’s Hillsdale Praxis program on the topic “Market Power: How Competition Generates Prosperity.” Video of the full event is embedded below.
Source: Hillsdale Praxis
I. Introduction The International Center for Law & Economics (ICLE) welcomes the opportunity to comment on the Ministry of Industry and Trade’s (MOIT) draft decree . . .
The International Center for Law & Economics (ICLE) welcomes the opportunity to comment on the Ministry of Industry and Trade’s (MOIT) draft decree replacing Decree No. 75/2019/NĐ-CP on sanctioning administrative violations in the field of competition (the Draft Decree).[1] These comments build on ICLE’s Feb. 2, 2026, and June 16, 2026, submissions on the draft law amending the Competition Law (the Draft Law) and should be read together with those submissions.[2]
The Draft Decree establishes the sanctions framework for the Draft Law. It sets fines, remedies, aggravating and mitigating factors, and the method for calculating illicit profit. It also restates the elements of the new platform-related abuses of dominance and unfair competition offenses.
In our June submission, we urged MOIT to keep the standards governing liability in primary legislation. The Draft Decree now partly defines those standards and fully defines their consequences. How MOIT will prove, remedy, and penalize the new abuses therefore matters as much as how the Draft Law describes them.
The remaining comments proceed in four parts. Section II identifies reforms MOIT should preserve. Section III explains why the conduct targeted by the new abuses does not warrant presumptive condemnation. Section IV proposes specific changes to the elements required to establish liability, remedies, fines, profit disgorgement (the recovery of profits from unlawful conduct), the new unfair competition offenses, and liability for facilitating violations. Section V sets out our recommendations.
Several changes in the Draft Decree reflect sound policy that reduces the risk and cost of enforcement errors. MOIT should retain them in the final text
Abolition of forced restructuring. The Draft Decree removes the remedy requiring a dominant or monopoly firm to restructure (Article 3(3)(c) of Decree No. 75/2019/NĐ-CP and the corresponding points in the abuse provisions). MOIT’s explanatory note offers sound reasons for this change. Restructuring intrudes deeply on property rights and corporate governance; affects employees, shareholders, and creditors; and has lacked a workable procedure. Behavioral remedies—orders governing a firm’s conduct—can address exclusionary effects with far less risk. This change reflects the principle stated in both of ICLE’s earlier submissions that targeted cease-and-desist orders are preferable to prescriptive or structural intervention.
Removal of ill-suited supplementary penalties. The Draft Decree removes license suspensions, confiscation of exhibits and instruments, and revocation of business registration. As MOIT observes, competition violations arise from commercial decisions, rather than physical instruments. Revoking a firm’s registration in addition to imposing high fines and disgorgement would be disproportionate and deter investment.
Proportionate treatment of procedural merger violations. Failing to notify a concentration or implementing it before clearance constitutes a procedural breach. The Draft Decree replaces turnover-based fines for these breaches with a fixed range of 1 billion to 2 billion Vietnamese dong (VND) (Article 22). It reserves turnover-based fines for prohibited concentrations and breaches of clearance conditions (Articles 20 and 21). This graduated approach matches the severity of the sanction to the severity of the harm. Section IV.A proposes applying the same logic to the new platform abuses.
Legal certainty. The five-year limitation period (Article 7), clearer allocation of sanctioning authority (Articles 38–40), and separation of competition procedure from general administrative procedure (Article 42) improve predictability. So do the three-year cap on the disgorgement period and the exclusion of disgorgement where the firm earned no profit (Article 8(2) and (8)).
Effects qualifiers and carve-outs. The foreclosure and harm requirements in Article 12(1)(g)–(k), and the exceptions for data protection, trade secrets, and cybersecurity in points (h) and (i), are welcome. Section IV.A explains how MOIT should complete these provisions.
Article 12(1)(g)–(k) identifies four practices as abuses of dominance: favoring the firm’s own or affiliated products, restricting users’ ability to switch, refusing or conditioning users’ access to their transaction data, and using nonpublic data generated by users or business partners to compete with them. Article 13(1)(a) extends all four to dominant firms, whose dominance the Competition Law can presume at a 30% market share.
These practices require careful, case-by-case analysis. The Draft Decree’s foreclosure and harm requirements are welcome, but a complete effects test must also account for efficiencies and consumer benefits. Evidence on vertical integration and self-preferencing supports that approach, especially because the provisions reach ordinary business practices beyond digital platforms. Several provisions would also compel firms to share assets or provide access to rivals, raising concerns about investment incentives. Liability standards and sanctions should reflect these competing effects.
The Draft Decree requires a showing of foreclosure—conduct that restricts rivals’ access to markets—and harm for each new category of abuse. We welcome that choice. A complete effects test, however, must also weigh harms against benefits. As drafted, Article 12(1) asks whether conduct has adverse effects but gives the firm no opportunity to show that efficiencies outweigh those effects. Combined with an undefined “likely” standard and near-automatic maximum fines, this approach risks treating much beneficial conduct as unlawful. The evidence must establish that these practices harm competition often enough to justify a test that disregards their benefits.
Competition law reserves per se treatment—automatic condemnation without examining competitive effects—for conduct that always or almost always harms competition. That principle led the U.S. Supreme Court to assess vertical restraints—or restrictions between firms at different levels of a supply chain—under the rule of reason in Continental T.V. v. GTE Sylvania and Leegin.[3] The rule of reason weighs a practice’s competitive harms and benefits. The Court of Justice of the European Union likewise required proof of exclusionary effects in Intel and, specifically for self-preferencing, in Google Shopping.[4]
To be clear, these practices can harm competition. But the evidence does not establish that they do so reliably enough to justify treating them as suspect by default.
Self-preferencing—favoring a firm’s own products or services—is a form of vertical integration. The empirical literature finds that vertical integration usually lowers costs and improves quality.[5] Studies of platforms point in the same direction. ICLE’s February submission reviewed this evidence in detail, including studies of Facebook’s integration of Instagram; Google Photos; console ecosystems; and an Amazon field experiment. In that experiment, hiding Amazon’s private-label products reduced consumer surplus—the value consumers receive above what they pay—by 5.5%, while demoting those products in rankings produced no gain.[6]
Some studies find harm in particular settings. Those findings underscore that effects depend on context and support case-by-case analysis.
Article 12’s breadth raises the stakes, as the provision extends beyond digital platforms. Under the Competition Law, a firm is presumed dominant when it holds a 30% share of a relevant market. Groups of firms can also be collectively dominant, even when each holds well below 30%.
Article 12(1)(g) would therefore apply to a supermarket chain favoring its own brands, a hospital group directing patients to its own laboratory, or a manufacturer supplying an affiliated distributor on better terms. Economies depend on these ordinary forms of vertical integration.
Three of the four practices would effectively impose a duty to deal—an obligation to provide access to rivals. Point (h) limits a firm’s control over switching and portability, or users’ ability to move their data between services. Point (i) requires access to data on terms that are not “unreasonable.” Point (g), read with the “other measures” remedy discussed in Section IV.B, could require access to technical specifications or interfaces.
Competition law has long treated forced sharing as exceptional. In Bronner, the Court of Justice of the European Union required that access be indispensable, that refusal eliminate all effective competition, and that no objective justification exist.[7] The General Court applied a version of that test to interoperability—the ability of systems to work together—in Microsoft.[8]
In Android Auto, the Court held that Bronner’s indispensability condition does not apply where a dominant firm developed a platform to host third parties rather than solely for its own use.[9] European Union competition scholars have criticized the judgment for abandoning the premise, shared by Magill and Bronner, that indispensability protects investment incentives. They also warn that the judgment leaves firms unable to determine in advance whether their platforms still qualify for Bronner’s protection.[10]
U.S. law takes an even more skeptical view of compelled access, as Trinko illustrates.[11] The economic rationale is straightforward. Mandated access lowers the return on the investment that created the asset and encourages rivals to rely on the incumbent’s infrastructure rather than build their own.
Points (h) and (i) contain none of these safeguards. Dominance, which the law can presume at a 30% market share, together with “unreasonable conditions,” would suffice to trigger the access obligation.
The Draft Decree should pair clear liability standards with proportionate remedies and sanctions. The following subsections recommend consistent proof of foreclosure and consumer harm, workable justifications, and safeguards for compelled access. They also propose limits on remedial discretion, fines calibrated to harm and legal certainty, and disgorgement confined to profits caused by the infringement. Finally, they recommend narrowing unfair competition and facilitation liability, preventing duplicate sanctions, and coordinating enforcement with Vietnam’s related regulatory frameworks.
The effects qualifiers in Article 12(1) provide a sound starting point, but the drafting leaves three gaps. MOIT should close them in the Draft Decree where the Draft Law permits, and amend the Draft Law where necessary. The catch-all provision also needs narrowing.
First, use a consistent standard of harm. Point (h) covers conduct that causes or is likely to cause harm to customers. Points (g), (i), and (k) cover only harm the conduct has caused. The text also separates foreclosure and harm with a semicolon, leaving unclear whether the authority must prove both.
Every item should use the same formula. The competition authority should bear the burden of proving, on the evidence, that the conduct forecloses or is likely to foreclose rivals that are at least as efficient, and that it significantly harms or is likely to harm consumers through higher quality-adjusted prices, lower output, or reduced innovation. Quality-adjusted prices account for differences in product quality. Harm to a less efficient competitor should be insufficient. “Likely” should mean more probable than not. The authority should have to establish both foreclosure and consumer harm.
Second, provide a general justification. The Draft Decree lacks broad, workable grounds for justifying the listed practices. Its only express exceptions concern data protection, trade secrets, and cybersecurity in points (h) and (i), and it offers no guidance on how the authority should assess them. Firms engage in these practices for many other legitimate reasons, including product quality, reliability, fraud prevention, cost recovery, and the integration of complementary features.
MOIT should expressly allow firms to justify conduct that is objectively necessary or whose efficiencies outweigh its restrictive effects. Once the firm raises that justification, the authority should bear the burden of rebutting it. If MOIT concludes that the Draft Law leaves no room for such a justification, the Draft Decree should, at minimum, treat demonstrated efficiencies as a mitigating factor under Article 6(1).
Third, add safeguards for access duties. For points (h) and (i), and any application of point (g) that would require access to interfaces or specifications, liability should require proof that access is indispensable for rivals to compete and that refusal would eliminate effective competition in the relevant market, in line with Bronner. Any access order should allow the firm to recover its costs and protect its users’ security.
Finally, narrow the catch-all in point (l). Point (l) sanctions any abuse of dominance “prohibited under other laws.” As ICLE noted in its June submission, this cross-reference makes liability depend on statutes outside competition law, including the Law on Digital Transformation, without an effects test. The Competition Law should expressly list any additional prohibited conduct and subject it to the same standard as the other items.
The Draft Decree’s move toward behavioral remedies is welcome. Two provisions nonetheless reintroduce the discretion that removing forced restructuring was meant to eliminate.
The “other measures” remedy. Article 12(2)(d) allows the authority to require “other measures where necessary” to remedy the restrictive effects of self-preferencing, switching restrictions, data access refusals, data use, and several other abuses. Article 13(2)(d) applies the same power to every abuse of a monopoly position, and Article 36(3)(b) extends it to facilitation. The text places no limits on these measures.
This clause could produce interoperability mandates, compelled data sharing, or product redesign orders. ICLE has documented the costs of such obligations in the European Union, including degraded search features and delayed product launches. We also described their security risks as a persistent “live wire” into user accounts. An open-ended remedy undermines the predictability the Draft Decree otherwise improves.
MOIT should replace the “other measures” clause with an exhaustive list of remedies. Each remedy should be necessary and proportionate to the harm found, and the authority should choose the least restrictive effective option. Access, interoperability, or data-sharing remedies should be available only where the conditions in Section IV.A are met. The decision should specify terms protecting security and allowing cost recovery. Firms should also have an opportunity to propose commitments addressing the authority’s concerns before it imposes a remedy.
Price control. Articles 12(2)(c) and 13(2)(c) allow the authority to place a dominant or monopoly firm’s buying prices, selling prices, or other trading conditions “under the control of the competent state authority.” Ongoing price supervision turns a competition authority into a price regulator—a concern ICLE raised in June about the “unreasonable terms” prohibition.
If MOIT retains this remedy, it should limit its use to proven exploitative pricing in markets with high, durable barriers to entry. Any price-control order should have a defined duration, undergo periodic review, and explain why a cease-and-desist order would be insufficient.
Articles 5 and 6 would produce severe fines for the new abuses almost automatically. That approach is ill-suited to conduct whose competitive effects the evidence shows to be ambiguous.
The midpoint rule and new aggravating factors. Article 5(9), which carries over Article 4(8) of Decree No. 75/2019/NĐ-CP unchanged, starts an abuse fine at the midpoint of the 1%–10% range, or 5.5% of turnover in the relevant market. One aggravating factor raises the fine above the midpoint, up to 7.75%. Two or more require the 10% maximum.
The Draft Decree replaces the existing, unquantified aggravating factor for large-scale violations (Article 5(2)(e) of Decree No. 75/2019/NĐ-CP) with three quantified factors in Article 6(2). They apply when the conduct involves at least 30% of the relevant market (point (h)), the value of the products concerned reaches VND 10 billion (point (i)), or the conduct lasts more than one year (point (k)).
These thresholds will be met in nearly every abuse case. Because the law presumes dominance at a 30% market share, point (h) effectively repeats an element of the offense. A firm large enough to be dominant will almost always exceed VND 10 billion in affected sales, and most exclusionary conduct lasts more than a year before an investigation concludes. A dominant firm found to have engaged in self-preferencing for more than a year would therefore face the 10% maximum by default, before the authority assesses the actual harm.
MOIT should make three changes. First, exclude abuse of dominance or monopoly from point (h), because treating an element of the offense as an aggravating factor counts the same fact twice. Second, set point (i)’s value threshold relative to the relevant market’s size or the firm’s turnover. Third, allow the authority to depart downward from the midpoint where the established harm is limited.
A graduated approach to new abuses. The Draft Decree already distinguishes procedural merger breaches from substantive violations. The same logic should apply to conduct that becomes unlawful for the first time. Article 12(1)(g)–(k) creates new legal standards whose boundaries firms and the authority cannot yet predict. Heavy fines under uncertain standards deter potentially lawful conduct—a particularly costly enforcement error in innovative markets.
For a defined transition period, a first finding under these provisions should lead to a cease-and-desist order or warning. Fines should apply only if the firm continues the conduct after the decision. At minimum, Article 6(1) should list reasonable reliance on the pre-amendment law as a mitigating factor.
Tenfold increases in fine levels. The Draft Decree substantially raises several fixed monetary fine ranges and maximums. It increases the maximum for “other” violations from VND 200 million to VND 2 billion and the fallback range for firms with zero turnover from VND 100 million–200 million to VND 1 billion–2 billion.
Monetary caps and ranges may improve predictability, but a tenfold increase requires justification based on the conduct’s seriousness, the need for deterrence, and the risk of overdeterrence. MOIT should ensure that actual fines remain proportionate to intent, material competitive harm, and the violation’s substantive or procedural nature. The authority should retain discretion to impose lower fines where the conduct caused no material harm, involved uncertain legal standards, or the firm promptly remedied it.
Removal of the criminal-law ceiling. Article 5(1) removes the requirement that administrative fines for cartels and abuse remain below the lowest criminal fine. ICLE does not oppose aligning the Draft Decree with the Draft Law. Removing this ceiling nevertheless raises the stakes of the concerns above and strengthens the case for calibrating fines to proven harm.
Articles 8 and 9 provide welcome structure for confiscating illicit profit, which previously lacked a stated calculation method. Their formula, however, would systematically overstate the gain attributable to an infringement.
Measure profit against lawful conduct. Article 9(1) defines illicit profit as the money received from the infringing conduct minus the direct costs of generating it. Under Article 9(2), that money includes all revenue from sales in the relevant market connected to the conduct during the infringement period. This formula treats the entire margin on affected sales as illicit.
But a firm found to have favored its own products would have sold most of those products anyway. Only the additional profit caused by the infringement is illicit. The correct measure is the difference between the profit the firm actually earned and what it would have earned had it complied with the law. Economists and courts use this comparison to quantify overcharges and lost profits in damages actions. It should apply here as well.
Allow deductions for common and indirect costs. Article 9(3) and (4) exclude costs that are not directly tied to producing and distributing the affected products. For many firms, particularly digital platforms, most costs are common or fixed. They include research and development, infrastructure, security, and operating the platform as a whole. Excluding these costs treats the return on those investments as unlawful gain. The Draft Decree should allow firms to allocate common and indirect costs to the affected products using recognized accounting methods.
Require reasoned estimates that firms can challenge. Article 9(5) and (6) allow the authority to estimate profit when a firm’s records are incomplete. Some estimation power is necessary. The authority should state its assumptions, use conservative values where data are uncertain, and allow the firm to challenge the estimate before the decision becomes final.
Credit disgorgement against the fine. A firm may face a fine of up to 10% of relevant turnover alongside confiscation of its profit. When the authority imposes both for the same conduct, it should credit the amount disgorged against the fine or account for it when setting the fine. Otherwise, the firm faces two penalties for the same gain.
The Draft Decree sets sanctions for three new unfair competition offenses created by the Draft Law. These offenses apply to firms of any size. All three raise concerns.
Abuse of distribution rights (Article 29). Article 29 imposes fines of VND 300 million–450 million for refusing to grant distribution or sales rights, granting them on unreasonable conditions, revoking or declining to renew them, or treating similarly situated distributors differently, in each case “without good reason.”
These are ordinary decisions about organizing distribution. Manufacturers select, terminate, and differentiate among distributors to protect quality, reward performance, prevent free riding on promotional investment, and maintain brand standards. When a supplier lacks market power, a distributor that loses its rights can turn to competing suppliers, and any dispute belongs in contract law. Condemning these decisions without regard to market power treats vertical restraints as suspect, contrary to the evidence discussed in Section III. It also unnecessarily interferes with freedom of contract.
MOIT should delete Article 29. If it retains the provision, liability should require that the supplier hold significant market power or that the distributor depend economically on it, and that the conduct harm competition in the downstream market—the market in which the distributor operates. The Draft Decree should also list legitimate reasons for these decisions, including performance, quality and safety standards, brand protection, compliance, and reorganization of the supplier’s distribution network.
Confusing signs (Article 30). Article 30 sanctions the use of confusing signs and indications. Trademark, unfair competition, advertising, and consumer protection rules already address this conduct. Parallel sanctions under the Draft Decree risk duplicative enforcement and overdeterrence.
Liability should require proof that the sign is likely to cause material confusion among an appreciable number of consumers and that the conduct materially affects competition. Similarity alone should be insufficient. The aggravating factor for conduct that reaches many customers online should depend on the scale and seriousness of the harm. Online activity or a large audience should not automatically trigger it.
Deceptive interface design (Article 31). Article 31 targets “dark patterns”—interface designs that mislead or manipulate users. It covers false scarcity or urgency signals, interfaces that highlight one option and obscure others, wording that places “psychological pressure” on customers, and unnecessary cancellation steps. Deceptive design can harm consumers, but the provision has three problems.
Liability under Article 31 should require proof that the design is likely to cause an average customer to make a decision they would not otherwise have made, a standard familiar from unfair commercial practices law. The authority should coordinate enforcement with consumer protection authorities, and an express rule should prevent sanctions under both regimes for the same conduct. Aggravation should depend on the scale of harm.
Where the alleged harm concerns consumer deception, the consumer protection regime should ordinarily provide the primary enforcement route. Competition-law sanctions should require a material effect on competitive conditions, such as foreclosure of rivals or harm to the competitive process.
Facilitation (Article 36). Article 36 sanctions organizing, coercing, encouraging, or assisting a firm to engage in a prohibited practice. It raises the fine for assisting a restrictive practice to VND 500 million–1 billion, up from VND 30 million–50 million under Decree No. 75/2019/NĐ-CP.
We welcome the requirement that assistance be intentional, which partly addresses ICLE’s June concerns about the “aiding” provision in Article 8 of the Draft Law. The provision still lacks a materiality requirement. Software developers, data providers, consultants, and trade associations routinely supply services that clients could misuse. Liability should require proof that the assistance substantially contributed to the infringement and exclude the ordinary provision of generally available goods and services. MOIT should also replace Article 36(3)(b)’s “other measures” remedy as proposed in Section IV.B.
Overlap with the Law on Digital Transformation. As ICLE explained in both earlier submissions, the Law on Digital Transformation (Law No. 148/2025/QH15) already regulates data access and portability for designated dominant platforms. Article 12(1)(h) and (i) would sanction closely related conduct under a separate regime that a different authority enforces.
The Draft Decree should expressly protect conduct that complies with the Law on Digital Transformation’s data-access and portability obligations from sanctions under points (h) and (i). It should also prevent sanctions under both regimes for the same conduct.
Coordination with the decree on market power. MOIT is also preparing a decree to replace Decree No. 35/2020/NĐ-CP, detailing market definition and market-share calculations under Articles 9 and 10 of the Competition Law. Its treatment of digital indicators of market power will determine how many firms fall within the provisions discussed above.
For the reasons given in ICLE’s June submission, user numbers, data, network effects, and ecosystem integration should inform an effects-based inquiry. These indicators should serve as evidence within that inquiry, without creating presumptions of market power.
The Draft Decree makes meaningful progress. It removes disproportionate, unworkable sanctions, calibrates merger penalties to the seriousness of the breach, and attaches effects requirements to the new platform-related abuses. The remaining weaknesses concern both liability and its consequences. Incomplete effects tests, open-ended remedies, near-automatic maximum fines, and overstated illicit profits risk deterring conduct that benefits consumers.
MOIT should preserve these reforms while ensuring that enforcement accounts for competitive benefits, protects investment incentives, and provides predictable standards across digital and traditional markets. We recommend that MOIT:
These changes would help Vietnam address demonstrable competitive harm while preserving the integration, investment, and product design that benefit Vietnamese consumers.
[1] Ministry of Indus. & Trade (Viet.), Comparison and Explanatory Table of the Draft Decree Replacing Decree No. 75/2019/NĐ-CP on Sanctioning Administrative Violations in the Field of Competition (Sept. 26, 2026), https://vibonline.com.vn/du_thao/du-thao-nghi-dinh-thay-quy-dinh-ve-xu-phat-vi-pham-hanh-chinh-trong-linh-vuc-canh-tranh. Article references refer to the Draft Decree unless otherwise stated.
[2] Dirk Auer, Dario Oliveira Neto, Lazar Radic, Selcukhan Ünekbas & Mario A. Zúñiga, Comments of the International Center for Law & Economics: Vietnam Ministry of Industry and Trade Proposed Amendments to Law No. 23/2018/QH14 (Competition Law) (Int’l Ctr. for L. & Econ., Feb. 2, 2026) [hereinafter Auer et al., February Comments]; Dirk Auer & Lazar Radic, Comments of the International Center for Law & Economics: Vietnam Ministry of Industry and Trade Draft Law Amending Law No. 23/2018/QH14 (Competition Law) (Int’l Ctr. for L. & Econ., June 16, 2026) [hereinafter Auer & Radic, June Comments].
[3] Cont’l T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36 (1977); Leegin Creative Leather Prods., Inc. v. PSKS, Inc., 551 U.S. 877 (2007).
[4] Case C-413/14 P, Intel Corp. v. Comm’n, ECLI:EU:C:2017:632 (Sept. 6, 2017); Case C-48/22 P, Google LLC & Alphabet Inc. v. Comm’n (Google Shopping), ECLI:EU:C:2024:726 (Sept. 10, 2024).
[5] Francine Lafontaine & Margaret Slade, Vertical Integration and Firm Boundaries: The Evidence, 45 J. Econ. Literature 629 (2007); James C. Cooper, Luke M. Froeb, Dan O’Brien & Michael G. Vita, Vertical Antitrust Policy as a Problem of Inference, 23 Int’l J. Indus. Org. 639 (2005).
[6] Auer et al., February Comments, supra note 2, at 5–6 (discussing, among other studies, Chiara Farronato, Andrey Fradkin & Alexander MacKay, Vertical Integration and Consumer Choice: Evidence from a Field Experiment (Nat’l Bureau of Econ. Rsch., Working Paper No. 34135, 2025)).
[7] Case C-7/97, Oscar Bronner GmbH & Co. KG v. Mediaprint, ECLI:EU:C:1998:569, ¶¶ 38–41 (Nov. 26, 1998).
[8] Case T-201/04, Microsoft Corp. v. Comm’n, ECLI:EU:T:2007:289 (Sept. 17, 2007).
[9] Case C-233/23, Alphabet Inc. v. Autorità Garante della Concorrenza e del Mercato (Android Auto), ECLI:EU:C:2025:110, ¶¶ 49, 52 (Feb. 25, 2025), https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex:62023CJ0233.
[10] Pablo Ibáñez Colomo, Case C-233/23, Android Auto (II): How the Judgment Departs from Magill/IMS Health (and Bronner), Chillin’Competition (Mar. 12, 2025), https://chillingcompetition.com/2025/03/12/case-c-233-23-android-auto-ii-how-the-judgment-departs-from-magill-ims-health-and-bronner; David Fåhraeus & Giulio Preti, A Landmark Judgment in the Refusal to Deal Doctrine: Android Auto (Case C-233/23) and Bronner’s Future in Digital Platform Markets, 11 Eur. Papers 195, § 9.2 (2026).
[11] Verizon Commc’ns Inc. v. Law Offs. of Curtis V. Trinko, LLP, 540 U.S. 398 (2004).
Source: Vietnam Ministry of Industry and Trade
Introduction and Summary The International Center for Law & Economics (ICLE) respectfully submits these comments on the Taiwan Fair Trade Commission’s (TFTC) Public Consultation Report . . .
The International Center for Law & Economics (ICLE) respectfully submits these comments on the Taiwan Fair Trade Commission’s (TFTC) Public Consultation Report and Policy Statement on Generative Artificial Intelligence and Competition.[1] We commend the TFTC for a thoughtful, economically grounded document whose method and conclusions reflect the best of the modern law & economics tradition.[2]
The Policy Statement adopts a sound analytical approach. It opens with the maxim ex facto jus oritur—law arises from the facts—and honors that commitment throughout.[3] It adopts four guiding principles: issue-driven and evidence-based analysis, attention to a nexus with Taiwan’s markets, preservation of market contestability (the ability to enter markets and challenge incumbents), and case-by-case assessment under the rule of reason.[4]
The Policy Statement cautions against equating scale, data, or computing power with market dominance. It rejects reliance on market structure alone, recognizes the potential benefits of partnerships, and distinguishes domestic firms’ competitive disadvantages from competition-law violations. The TFTC’s commitment to evidence-based, case-specific assessment offers a valuable alternative to the European Union’s prescriptive approach to digital competition under the Digital Markets Act.[5]
Part I explains why these principles deserve support. Part II suggests clarifications to help firms and enforcers apply them consistently, particularly to ecosystems, entry, talent, and integration. Our recommendations seek to preserve the flexibility and attention to evidence that make the Policy Statement valuable. Part III concludes.
The Policy Statement’s insistence on a concrete theory of competitive harm to guide enforcement priorities is among its most important features.[6] As the document recognizes, a theory of harm links specific conduct, a mechanism, and a competitive effect. The theory alone does not establish that the mechanism operates in the market concerned.
In rapidly evolving markets, a plausible account of how an incumbent might exclude rivals cannot substitute for evidence that the conduct is likely to harm competition. Prospective assessment nonetheless remains appropriate. Enforcers need not wait for harm to occur, provided evidence supports the predicted harm.
Enforcement in artificial intelligence (AI) markets should replace hype and preconceived views with careful, evidence-based analysis.[7] Uncertainty in nascent markets increases the risk that premature intervention will suppress beneficial conduct.[8] The TFTC’s approach appropriately requires sufficient evidence to justify intervention while leaving room for innovation and experimentation.
The Policy Statement rightly cautions against equating the scale and volume of data and computing power with market dominance or a lack of contestability.[9] The OpenAI–Cerebras computing agreement, reported discussions involving OpenAI and Anthropic over biotech data, the Adobe–Runway partnership, and the BMW–Innoviz collaboration illustrate the importance of specialized suppliers and complementary capabilities. Scale in one activity does not automatically establish market power in another.[10]
The economic reasons for this caution vary across AI activities. Inference—running a trained model to generate outputs—entails variable costs. Those costs do not, however, negate substantial fixed costs, sunk investments, or economies of scale in model development and cloud infrastructure. An entrant also need not match a frontier model’s capabilities across every task to compete effectively for a particular use. Smaller, specialized models may offer an attractive combination of cost and performance.[11] Treating AI as a single market or monolithic input obscures differences in demand, production costs, and opportunities for entry across the “AI stack”—the layers of infrastructure, models, and applications that support AI services.[12]
Oversimplification is particularly problematic when assessing the competitive significance of data access. That significance depends, among other things, on which data a task requires, whether public, licensed, or synthetic alternatives are adequate, and whether claimed advantages persist as methods improve. Data volume alone does not establish a bottleneck that excludes rivals.
Open-weight models, which make their trained parameters available for others to use or adapt, can also reduce development costs for entrants. The TFTC should assess their availability alongside proprietary alternatives without prescribing a preferred business model. Any intervention should identify the particular constraint that prevents effective competition.[13]
The Policy Statement rightly reflects the mainstream economic rejection of the structure-conduct-performance (SCP) paradigm, which links market structure to firms’ conduct and market outcomes.[14] It declines to presume illegality from market structure unless the evidence shows that structure determines the overall competitive effect. It also expressly directs attention to demand and user choice.[15] Concentration alone cannot establish durable market power or consumer harm, as it may also reflect economies of scale, superior performance, or successful innovation.[16]
A consumer-welfare approach should assess price, output, quality, and innovation, including privacy and security where users value them. The number of suppliers or standalone products is not itself a measure of consumer welfare. Moreover, integrated offerings can improve welfare even when they reduce some forms of choice.
The report records stakeholders’ concern that unnecessary burdens on mergers and acquisitions may weaken innovation, while the Policy Statement recognizes that strategic alliances can promote innovation and competition.[17] These observations support a receptive, effects-based approach to AI partnerships. Acquisition can provide an important exit strategy for startups. Partnerships can supply the funding, computing resources, and complementary capabilities needed to commercialize their work.
The TFTC should be especially wary of treating an incumbent’s acquisition as suspect merely because the incumbent might hypothetically have developed the target’s business itself. That “growth through acquisition” theory rests on a fringe reading of antitrust law that, if generalized, could diminish acquisition-driven growth across the economy.[18] A credible independent-entry counterfactual—a prediction of how the incumbent would have entered without the acquisition—requires evidence of likely entry, its timing, and its competitive significance. Technical or financial capacity alone is insufficient.[19]
Recent empirical work also counsels against categorical assumptions. Florian Ederer, Regina Seibel, and Timothy Simcoe find little support for the often-claimed position that digital incumbents’ acquisitions systematically “kill” innovation.[20] They conclude instead that such transactions often reflect complementary capabilities and wider technology diffusion.[21]
The same error is particularly salient in AI markets. Firms may face criticism for acquiring or partnering to obtain complementary capabilities, and also for building those capabilities internally. This regulatory bind risks chilling the investments and collaborations needed for innovation.[22] The broader literature finds that voluntary vertical integration—combining activities at different stages of production or distribution—frequently benefits consumers. That evidence supports requiring a demonstration of harm before treating integration as anticompetitive.[23]
The appropriate comparison is with the opportunities realistically available without the transaction. Alternative financing may arrive later, cost more, or lack the computing and commercial support a partner supplies. Contractual protections may enable investments tailored to a particular commercial relationship rather than exclude rivals. These considerations require evidence about the agreement’s terms and alternatives. They do not establish that every restriction is efficient. Preventing established technology firms from entering through partnerships can also protect today’s leading AI providers from an important source of rivalry.[24]
Among the Policy Statement’s clearest analytical strengths is its distinction between policies addressing domestic firms’ limited capital or lower turnover and conduct that harms competition.[25] Competition law protects the competitive process and users’ welfare. It should not engineer a particular industrial structure or favor firms because of their nationality.
The report’s discussion of incentives and subsidies need not imply endorsement. Any such measures require their own justification, accounting for their fiscal costs, effects on entry, and risk of protecting less efficient firms.
We likewise commend the TFTC’s competition advocacy under Article 46 of the Fair Trade Act—in particular its warning that proposed data-sharing requirements may weaken firms’ incentives to innovate and compete.[26] Scrutiny of regulatory barriers to entry complements enforcement. The same attention to incentives should inform any proposed access or interoperability remedy—a requirement to provide access to resources or enable systems to work together. Its expected benefits should justify the costs to investment, privacy, security, and implementation. Its scope and duration should also be no broader than necessary to address the demonstrated problem.
Competition advocacy could also examine whether data-localization requirements—rules governing where firms store or process data—and other compliance obligations unnecessarily restrict access to computing resources or raise adoption costs for smaller firms. That inquiry should distinguish customers’ voluntary preferences from regulatory constraints, assess the stated privacy and security objectives, and consider whether less restrictive measures could achieve them.[27]
The TFTC rightly recognizes that Taiwan’s competitive context differs from that of other jurisdictions and that comparative experience requires adaptation to local conditions. The European model, which some jurisdictions have rushed to adopt, has not demonstrated superior outcomes. Before adopting it, authorities should assess its economic rationale and suitability to local conditions.[28]
The Policy Statement also appropriately notes that earlier Microsoft tying cases—concerning the sale of products together—did not involve generative AI.[29] Those cases may inform analysis, but they cannot establish the effects of a different arrangement under different competitive conditions. Authorities should assess the benefits of integration alongside any restriction of choice. The availability of more standalone products does not by itself demonstrate greater consumer welfare.[30]
We offer the following observations in the same constructive spirit as the above. Clarifying how the TFTC will apply its principles would improve predictability for businesses and help preserve the Policy Statement’s evidence-based, case-specific approach.
The Policy Statement places particular importance on promoting competition as AI ecosystems form.[31] Early attention can help the TFTC identify conduct likely to exclude rivals before harm becomes difficult to remedy. The accompanying discussion of the Apple–Google arrangement and Google’s integration across AI activities identifies useful subjects for inquiry. The existence of a partnership or an integrated provider, however, should not itself create a presumption of harm.[32]
We respectfully encourage the TFTC to clarify that this principle calls for heightened vigilance—a readiness to investigate conduct in emerging ecosystems under the rule of reason. It should not authorize preventive rules that prescribe market structure.
The document’s own ecosystem analysis supports this distinction. Where firms compete “for the market”—to establish a leading ecosystem—the relevant focus is competition between ecosystems or brands. Preserving rivalry within an ecosystem is not an end in itself. Competition between ecosystems can discipline conduct within them, while restrictions on rivalry within an ecosystem may support investment and competition between integrated offerings.[33]
The TFTC should test the strength of this competitive constraint in each case, rather than assume it is either absent or sufficient. Rules that dictate how an ecosystem must be organized risk suppressing the integration and experimentation through which providers compete.
The Policy Statement’s attention to competition within and between ecosystems is welcome.[34] An ecosystem can itself constitute the relevant market where users choose among competing systems and the integration of complementary services determines the offering’s performance. Defining markets solely around individual components can exclude the very competition that constrains a provider.
In AI, customers may choose among combinations of models, computing resources, development tools, and applications whose usefulness depends on how they work together. Common governance, compatibility rules, and coordinated investment help produce that value. They form part of the competitive offering and should not carry a presumption that they restrict competition and require removal.[35]
Market definition should therefore examine users’ willingness to switch among offerings based on the dimensions on which they compete, including quality, functionality, and innovation. The question is whether an increase in price or a decline in performance would cause enough users to switch to make that change unprofitable. The analysis should account for changes in demand and participation across components that materially depend on one another.
Depending on the evidence, the relevant alternatives may include other ecosystems, less integrated combinations, or non-AI products and services. Providers may compete through different combinations of internal development, partnerships, and purchased inputs. Antitrust should not privilege one organizational form over another.[36]
This approach gives practical effect to the TFTC’s attention to demand. Neither a narrow product label nor the description of a collection of products as an ecosystem should predetermine the market.
The Policy Statement recognizes that current indicators—competitor counts, prices, sales, and market-share changes—may not adequately capture competition in a rapidly evolving sector. It therefore proposes examining the potential and probability of entry.[37] We agree with this forward-looking conception of contestability, associated with William J. Baumol. The threat of entry can discipline incumbents even where few firms operate. The relevant question is whether that threat constrains their conduct, rather than whether the market contains a preferred number of competitors.[38]
Left unspecified, “potentiality and probability of entry” can cut both ways. Without a concrete evidentiary test, it can become an open-ended license to intervene whenever an incumbent appears large. The TFTC should instead specify whether entry or expansion is likely, timely, and sufficient to constrain the conduct at issue. Neither current concentration nor the mere possibility of entry should be considered conclusive.
This economic conception of contestability differs from the European regulatory approach. Reducing an incumbent’s advantages or increasing the number of rivals does not establish that consumers are better off. Network effects (the value a service gains as more people use it), economies of scale, and accumulated capabilities may reflect successful competition. Compelling firms to share or surrender those advantages can weaken investment without strengthening competitive constraints. Contestability should discipline the exercise of market power, rather than protect competitors from the consequences of competition.[39]
Error-cost analysis—weighing the expected costs of mistaken intervention and nonintervention—should inform the enforcement decision. The TFTC should weigh the expected costs of mistaken intervention and nonintervention, including lost innovation and the ability of entry to correct market power.[40] Evidence of multi-homing—using multiple providers—and competition on data localization and security, which the TFTC identifies in Taiwan, is relevant to that assessment.[41] Its significance depends on whether users can shift meaningful workloads at reasonable cost, rather than simply maintain accounts with multiple providers. This distinction would make contestability a practical inquiry into competitive constraints while retaining appropriate caution about intervention.
The TFTC should likewise assess switching costs over the customer relationship and against realistic alternatives. Some costs reflect data transfer, retraining, or the distinctive functionality customers chose. Competition to attract customers may also produce discounts and other benefits that compensate for anticipated migration costs.
The relevant inquiry is whether a particular restriction impairs competition, accounting for total costs, customers’ ability to anticipate contractual terms, and available alternatives. Neither competition at the initial contracting stage nor the existence of switching costs is conclusive. Requiring uniform products or costless switching could sacrifice differentiation without improving consumer welfare.[42]
The separation of industrial policy and competition law should also guide the treatment of noncompete clauses. The Policy Statement refers to Taiwan’s interest in stemming the outflow of talent and key technologies.[43] We encourage the TFTC to distinguish that industrial-policy objective from competition-relevant justifications, such as protecting trade secrets or enabling investment in training. The TFTC should assess those justifications on their merits, irrespective of the nationality of the firms or workers concerned.
The competitive assessment should also distinguish among types of restraint. A restriction reasonably necessary to a productive collaboration raises different questions from an agreement among competing employers simply to refrain from hiring workers or competing on compensation. The economics support effects-based review of ancillary restraints—restrictions reasonably necessary to a broader productive arrangement. They do not support a blanket rule-of-reason safe harbor for labor-market cartels.[44]
Subject to Taiwan’s applicable legal rules, the analysis should consider the restraint’s scope, duration, effect on workers’ opportunities, and substantiated efficiencies. National talent retention should not substitute for competitive analysis.
The Policy Statement lists acquihires—arrangements to recruit a firm’s team—alongside no-poach and noncompete clauses among talent-related practices warranting attention.[45] We encourage the TFTC to resist the view that acquihires are inherently suspect and assess their likely effects under the applicable legal standards. Hiring a team does not itself establish that a competitor has disappeared. These arrangements can serve different economic functions, and their competitive effects depend on their terms and the realistic alternatives available to the firms involved.[46]
Three distinctions would assist that inquiry. First, the labor-market conditions identified in the TFTC’s report may cut either way. They do not necessarily establish a scarcity of AI workers. Rising wages may attract additional workers, while retraining may alleviate shortages. Rising wages may also reflect scarcity, and the TFTC’s record identifies different training requirements for different AI roles that may limit retraining’s effectiveness. The inquiry should examine the specific skills concerned, sources of additional labor (e.g., retraining, education, and immigration), available substitutes, and the time needed to recruit or train alternatives.[47]
Second, the structural-merger framework fits acquihires poorly. Labor is not a commodity that firms can “acquire,” and employment relationships cannot transfer without workers’ consent. A team’s movement also has different competitive effects from the permanent transfer of a productive asset. Retention, employee choice, contractual restrictions, and the capabilities remaining at the target all matter. The typical case nevertheless has few of the hallmarks of a traditional merger.[48]
Third, the transactions most often invoked as cautionary tales are poor examples. In AI inference, the cost of querying a model at GPT-3.5-level performance fell more than 280-fold between November 2022 and October 2024.[49] In such a market, portraying a talent-and-licensing deal as eliminating a uniquely formidable rival assumes an unlikely course of events without the deal. A credible case requires evidence that the target would otherwise have remained a meaningful competitive constraint and that the transaction is likely to remove that constraint.[50]
To be clear, an acquihire can still be anticompetitive. Where a transaction removes a meaningful competitive constraint, the TFTC should examine whether it falls within the applicable legal provisions and whether the evidence establishes harm.
But acquihires can also enable entrepreneurial exit, transfer practical knowledge embedded in cohesive teams, and sustain the labor mobility and reputational incentives on which startup ecosystems depend.[51] These benefits belong in an effects-based assessment and support caution about treating the transaction’s form as evidence of competitive injury.
As noted, the Policy Statement rightly cautions against equating scale with dominance.[52] It also expresses concerns that larger firms may continue to grow and that markets may tend toward winner-takes-all outcomes.[53]
The TFTC should clarify the Policy Statement’s exception for circumstances in which market structure could determine competitive effects. Any such inference should explain why the observed structure reflects or enables a restriction of competition rather than superior performance or efficient scale. An undefined exception risks restoring the very presumption the Policy Statement otherwise rejects. Specifying the necessary evidence and distinguishing merger review from conduct cases would avoid that result.
A large firm’s growth is not inherently a competitive problem, particularly when the sector also grows through more output, lower prices, better integration, and higher quality. The relevant distinction is between exclusionary conduct that harms competition and growth through competition on the merits. Enforcement should target the former. The latter remains lawful even when it enlarges a firm’s market power.
Access to capital, vertical integration, and scale can support lower costs, better products, and innovation. Market power acquired through superior performance is not itself an antitrust violation. Enforcement should protect competition rather than particular competitors.[54]
At the same time, sector-wide growth does not establish that every practice is benign. The relevant inquiry is how the conduct changes competition and consumer welfare compared with realistic conditions without it, accounting for likely efficiencies.[55]
Tying and bundling likewise require attention to likely exclusion and consumer benefits under the applicable standards. Integration alone cannot establish harm. Any proposed remedy should account for the agency’s limited information about product design and the risk of impairing beneficial integration.[56]
The consultation identifies uncertainty about AI-assisted pricing and recognizes these tools’ potential benefits. The TFTC’s evidence-based approach should distinguish independent use of pricing software from agreements or information exchanges that facilitate coordination. Common software, similar recommendations, or parallel prices should not alone establish concerted action.
Relevant questions include what information firms share, whether it is public or competitively sensitive, whether firms retain independent pricing discretion, and what evidence establishes coordination under the Fair Trade Act. This approach would preserve effective enforcement against unlawful coordination while allowing firms and consumers to benefit from lawful uses of pricing technology.[57]
The TFTC’s Policy Statement offers a valuable example of economically grounded, evidence-based competition policy. We commend its insistence on concrete theories of harm, its economically informed treatment of partnerships and acquisitions, its separation of competition law from industrial policy, and its commitment to local evidence and case-by-case assessment.
Our suggested clarifications would help preserve these commitments against exceptions that reinstate presumptions against scale, integration, or entrepreneurial exit. Recognizing competition between ecosystems, assessing entry and switching in their economic context, and requiring a credible assessment of what would happen without an acquisition would make the Policy Statement’s principles more effective in practice.
We also support addressing the identified competition concerns through the existing Fair Trade Act.[58] Before adopting AI-specific prohibitions, the TFTC should identify a concrete deficiency in existing law and assess whether the proposed rule’s benefits justify its compliance and innovation costs. Taiwan can protect consumers and competition while preserving the investment, integration, and experimentation through which AI services improve.
[1] Authored by Geoffrey A. Manne, president and founder of the International Center for Law & Economics (ICLE), and Lazar Radic, senior scholar at ICLE. ICLE is a nonprofit, nonpartisan research center based in Portland, Oregon, dedicated to building the intellectual foundations for sensible, economically grounded policy. ICLE promotes law & economics methods to inform public policy and has filed comments and analyses with competition and regulatory authorities across numerous jurisdictions.
[2] Taiwan Fair Trade Comm’n, Public Consultation Report and Policy Statement on Generative Artificial Intelligence and Competition (Mar. 2026), https://www.ftc.gov.tw/upload/833774aa-edbd-42a5-a41e-5d71045a3e6c.pdf [hereinafter Policy Statement]. Page references follow the document’s printed page numbers.
[3] Policy Statement, supra note 2, at 47 (quoting Louis D. Brandeis, The Living Law, 10 Ill. L. Rev. 461, 467 (1916)).
[4] Id. at 61.
[5] Regulation (EU) 2022/1925, 2022 O.J. (L 265) 1 (Digital Markets Act). See generally Dirk Auer, Geoffrey A. Manne & Sam Bowman, Should ASEAN Antitrust Laws Emulate European Competition Policy?, 67 Sing. Econ. Rev. 1637 (2022), https://doi.org/10.1142/S0217590821430025.
[6] Policy Statement, supra note 2, at 50–51 & n.27 (describing theories of competitive harm as narratives linking conduct, mechanism, and effect).
[7] Lazar Radic & Kristian Stout, What Is the Relevant Product Market in AI?, in Artificial Intelligence and Competition Policy (Alden Abbott & Thibault Schrepel eds., 2024) (author manuscript at 5, 20), https://laweconcenter.org/wp-content/uploads/2024/09/ssrn-4927505.pdf [hereinafter Relevant Product Market]. Manuscript pinpoints count PDF pages beginning with the first page.
[8] Relevant Product Market, supra note 7, manuscript at 14 (discussing Jonathan M. Barnett, The Case Against Preemptive Antitrust in the Generative Artificial Intelligence Ecosystem, in Artificial Intelligence and Competition Policy (Alden Abbott & Thibault Schrepel eds., 2024) (author manuscript at 8)).
[9] Policy Statement, supra note 2, at 58 (cautioning against equating the scale and volume of data and computing power with market dominance or a lack of contestability).
[10] Id. at 57–58 & nn.34–38.
[11] For example, Google’s January 2026 TranslateGemma release includes a translation-specific 12-billion-parameter model that outperformed the general Gemma 3 27-billion-parameter baseline on Google’s reported translation benchmark, using fewer than half as many parameters. See David Vilar & Kat Black, TranslateGemma: A New Suite of Open Translation Models, Google (Jan. 15, 2026), https://blog.google/innovation-and-ai/technology/developers-tools/translategemma.
[12] See Andrei Hagiu & Julian Wright, Artificial Intelligence and Competition Policy, 103 Int’l J. Indus. Org. 103134, at 9–10 (2025), https://doi.org/10.1016/j.ijindorg.2025.103134; Policy Statement, supra note 2, at 54; Relevant Product Market, supra note 7, manuscript at 1–2, 6–8.
[13] See Geoffrey A. Manne et al., ICLE Comments to DOJ on Promoting Competition in Artificial Intelligence §§ I.C–D (Int’l Ctr. for L. & Econ., July 15, 2024), https://laweconcenter.org/resources/icle-comments-to-doj-on-promoting-competition-in-artificial-intelligence [hereinafter DOJ AI Comments]; Hagiu & Wright, supra note 12, at 9–10; Kristian Stout & Thomas A. Hemphill, A Framework for Understanding and Evaluating AI Commercialization Strategies 5–8 (Int’l Ctr. for L. & Econ., Issue Brief, Aug. 27, 2025), https://laweconcenter.org/resources/a-framework-for-understanding-and-evaluating-ai-commercialization-strategies.
[14] See Steven Berry, Martin Gaynor & Fiona Scott Morton, Do Increasing Markups Matter? Lessons from Empirical Industrial Organization, 33 J. Econ. Persp. 44, 45–48 (2019), https://doi.org/10.1257/jep.33.3.44 (explaining why concentration–performance correlations cannot identify the competitive mechanism or its welfare consequences).
[15] Policy Statement, supra note 2, at 58–59.
[16] See Auer, Manne & Bowman, supra note 5, at 1679–80 (examining the limitations of concentration-based inferences); Berry, Gaynor & Scott Morton, supra note 14, at 45–48.
[17] Policy Statement, at 45–46 & 48.
[18] See Geoffrey A. Manne & Justin (Gus) Hurwitz, Build, Buy, or Both?: On the Antitrust Laws’ Supposed Preference for “Internal Growth” over Acquisitions, 26 Nev. L.J. 53, 80–87 (2025), https://scholars.law.unlv.edu/nlj/vol26/iss1/3.
[19] See, e.g., FTC v. Meta Platforms, Inc., 654 F. Supp. 3d 892, 934–35 (N.D. Cal. 2023) (requiring a feasible and reasonably probable path to independent entry rather than an inference from resources alone). See also Lazar Radic, Imaginary Consensus as a Legitimizing Philosophy of the New Antitrust Meta-Narrative 5 (Int’l Ctr. for L. & Econ., Issue Brief, Jan. 31, 2025) [hereinafter Imaginary Consensus].
[20] See, e.g., Cristina Caffarra, Gregory S. Crawford & Tommaso Valletti, “How Tech Rolls”: Potential Competition and “Reverse” Killer Acquisitions, CPI Antitrust Chron., May 2020, at 13, 14–16, https://www.competitionpolicyinternational.com/wp-content/uploads/2020/05/AC-May-II.pdf (arguing that acquisitions can eliminate both the target’s potential challenge and the acquirer’s own innovation effort, and favoring presumptions against such transactions).
[21] Florian Ederer, Regina Seibel & Timothy Simcoe, Digital (Killer?) Acquisitions 51 (Nat’l Bureau of Econ. Rsch., Working Paper No. 35762, 2026) (“[T]he evidence is more consistent with complementarity than with suppression: innovation does not decline after these deals, and the acquired technologies diffuse more widely.”).
[22] See Dirk Auer & Mario A. Zúñiga, AI Partnerships and Competition: Damned if You Buy, Damned if You Don’t 17–21 (Int’l Ctr. for L. & Econ., White Paper, Aug. 19, 2025) (forthcoming Eur. Competition L. Rev.), https://laweconcenter.org/wp-content/uploads/2025/08/AI-partnerships-and-competition.pdf.
[23] See Geoffrey A. Manne, Kristian Stout & Eric Fruits, The Fatal Economic Flaws of the Contemporary Campaign Against Vertical Integration, 68 U. Kan. L. Rev. 923, 957 (2020); Francine Lafontaine & Margaret Slade, Vertical Integration and Firm Boundaries: The Evidence, 45 J. Econ. Lit. 629, 680 (2007), https://doi.org/10.1257/jel.45.3.629. See also Imaginary Consensus, supra note 19, at 5.
[24] See Manne & Hurwitz, supra note 18, pt. III; DOJ AI Comments, supra note 13, at 22–24; Geoffrey A. Manne, Dirk Auer & Mario A. Zúñiga, ICLE Comments to UK Competition and Markets Authority on AI Partnerships (Int’l Ctr. for L. & Econ., May 9, 2024), https://laweconcenter.org/resources/icle-comments-to-uk-competition-and-markets-authority-on-ai-partnerships.
[25] Policy Statement, supra note 2, at 49–50.
[26] Id. at 50 & n.26. See also DOJ AI Comments, supra note 13, § I.D (discussing how mandatory data sharing can weaken investment incentives).
[27] See Policy Statement, supra note 2, at 8–10, 39, 50, 54 (discussing localization, computing access, and competition advocacy).
[28] See generally Auer, Manne & Bowman, supra note 5, at 1640 (arguing that the EU model’s appeal partly reflects its accommodation of politically useful populist narratives about bigness and robber barons that more evidence-based regimes resist).
[29] Policy Statement, supra note 2, at 49.
[30] See Auer, Manne & Bowman, supra note 5, at 1657–58 (criticizing the failure to weigh asserted consumer-choice benefits against integration efficiencies in the European Microsoft litigation).
[31] Policy Statement, supra note 2, at 56.
[32] Id. at 56 & nn.32–33 (noting the Apple–Google arrangement to adopt Gemini for Siri and Apple Intelligence, and concerns about Google’s full-stack vertical integration).
[33] See Auer, Manne & Bowman, supra note 5, at 1664–65; Policy Statement, supra note 2, at 59–60.
[34] Policy Statement, supra note 2, at 59–60.
[35] See Geoffrey A. Manne, Ecosystem Market Definition: Systems Competition, Dynamic Analysis, and the Limits of Ecosystem Theories of Harm, in Conglomerates and Ecosystems in the Digital Era ch. 12, §§ 12.3–12.4 (Justin (Gus) Hurwitz, Geoffrey A. Manne & Jeffrey Prince eds., Cambridge Univ. Press forthcoming) (Sept. 10, 2026 draft).
[36] See Manne, supra note 35, §§ 12.3–12.4; Manne & Hurwitz, supra note 18, pt. III; Relevant Product Market, supra note 7, manuscript at 14–20.
[37] Policy Statement, supra note 2, at 55–56.
[38] See William J. Baumol, Contestable Markets: An Uprising in the Theory of Industry Structure, 72 Am. Econ. Rev. 1, 3–4 (1982); Geoffrey A. Manne, Lazar Radic & Dirk Auer, Regulate for What? A Closer Look at the Rationale and Goals of Digital Competition Regulations, 22 Berkeley Bus. L.J. 201, 235–41 (2025), https://doi.org/10.15779/Z38M03Z022.
[39] See Manne, Radic & Auer, supra note 38, at 235–41.
[40] See Baumol, supra note 38, at 3–4; Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1, 2–3 (1984).
[41] Policy Statement, supra note 2, at 53–54, 59–60.
[42] See Brian Albrecht & Geoffrey A. Manne, ICLE Response to the FTC’s Cloud Computing RFI § II.B (Int’l Ctr. for L. & Econ., June 21, 2023), https://laweconcenter.org/resources/icle-response-to-the-ftcs-cloud-computing-rfi; Geoffrey A. Manne et al., ICLE Comments on the CMA’s Provisional Findings on the Cloud Services Market §§ IV–V (Int’l Ctr. for L. & Econ., Feb. 18, 2025), https://laweconcenter.org/resources/icle-comments-on-the-cmas-provisional-findings-on-the-cloud-services-market; Policy Statement, supra note 2, at 12–15, 51–52, 59–60.
[43] Policy Statement, supra note 2, at 54–55 & n.29.
[44] See Brief for International Center for Law & Economics and Scholars of Law and Economics as Amici Curiae Supporting Appellees at 17–19, Giordano v. Saks Inc., No. 23-600 (2d Cir. Nov. 3, 2023), https://laweconcenter.org/wp-content/uploads/2023/11/2023.11.03-Saks-ICLE-Amicus-Motion-and-Brief-As-Filed.pdf. This example applies U.S. law. Although the economics are the same, Taiwan’s law may differ.
[45] Policy Statement, supra note 2, at 52.
[46] See, e.g., Selcukhan Ünekbas, Acquihires and Other Antitrust Ghost Stories, Truth on the Mkt. (Mar. 25, 2026), https://truthonthemarket.com/2026/03/25/acquihires-and-other-antitrust-ghost-stories; Onyeka Aralu & Dirk Auer, Acquihires and Antitrust: When Buying the Team Isn’t Buying the Company, Truth on the Mkt. (Apr. 9, 2026), https://truthonthemarket.com/2026/04/09/acquihires-and-antitrust-when-buying-the-team-isnt-buying-the-company.
[47] Policy Statement, supra note 2, at 52; cf. Ünekbas, supra note 46.
[48] See John F. Coyle & Gregg D. Polsky, Acqui-Hiring, 63 Duke L.J. 281, 293–99, 303–05 (2013), https://scholarship.law.duke.edu/dlj/vol63/iss2/1; Beril Boyacioglu, Mahmut N. Özdemir & Samina Karim, Acqui-hires: Redeployment and Retention of Human Capital Post-Acquisition, 45 Strategic Mgmt. J. 205, 205–07 (2024), https://doi.org/10.1002/smj.3556. See also Aralu & Auer, supra note 46.
[49] See Stanford Inst. for Human-Centered Artificial Intelligence, Artificial Intelligence Index Report 2025 41 (2025), https://hai.stanford.edu/assets/files/hai_ai_index_report_2025.pdf (reporting a decline from $20 to $0.07 per million tokens for GPT-3.5-level MMLU performance, using a 3:1 weighted average of input and output token prices).
[50] See Eric Fruits et al., ICLE Comments to FTC and DOJ on Premerger Notifications § IV.B (Int’l Ctr. for L. & Econ., May 22, 2026), https://laweconcenter.org/resources/icle-comments-to-ftc-and-doj-on-premerger-notifications.
[51] See Coyle & Polsky, supra note 48, at 283–85, 294–95, 309–16; Boyacioglu, Özdemir & Karim, supra note 48, at 205–07; Aralu & Auer, supra note 46; Ünekbas, supra note 46.
[52] Policy Statement, supra note 2, at 58.
[53] Id. at 55, 59.
[54] See Imaginary Consensus, supra note 19, at 5; Lafontaine & Slade, supra note 23, at 680.
[55] Policy Statement, supra note 2, at 59.
[56] Id. at 32, 58–59.
[57] See Policy Statement, supra note 2, at 37, 46, 50 n.26; Eric Fruits et al., ICLE Comments on DOJ/FTC Guidance on Business Collaborations §§ IV.A–B (Int’l Ctr. for L. & Econ., 2026), https://laweconcenter.org/resources/icle-comments-on-doj-ftc-guidance-on-business-collaborations; Ian Adams et al., ICLE Comments to Competition Bureau Canada on Anti-Competitive Conduct and Agreements (Int’l Ctr. for L. & Econ., 2026), https://laweconcenter.org/resources/icle-comments-to-competition-bureau-canada-on-anti-competitive-conduct-and-agreements.
[58] See Policy Statement, supra note 2, at 39–41, 61.
Source: Taiwan Fair Trade Commission
Introduction The International Center for Law & Economics (ICLE) submits these comments on the Korea Fair Trade Commission’s (KFTC) proposed amendment to the Notification Guidelines . . .
The International Center for Law & Economics (ICLE) submits these comments on the Korea Fair Trade Commission’s (KFTC) proposed amendment to the Notification Guidelines for Business Combinations (Proposed Amendment), published for public comment on Sept. 9.[1] ICLE is a nonprofit, nonpartisan global research and policy center that applies law & economics methods to public-policy debates. Its scholars have written extensively on acquihires—transactions centered on hiring a firm’s team—and artificial intelligence (AI) partnerships.[2] ICLE has also commented on the KFTC’s merger review guidelines.[3]
The Proposed Amendment would treat transactions in which a firm hires a target’s key personnel, often alongside a license to its technology, as business transfers under Article 9(1)(iv) of the Monopoly Regulation and Fair Trade Act (MRFTA).[4] The KFTC describes the amendment as a clarification. In its view, acquihires already constitute business transfers under Article 9(1)(iv), and the amendment would make that clear so parties do not fail to notify. We welcome greater certainty for parties, but the proposed text goes beyond restating existing law.
The Proposed Amendment would expand the definition of a “business” to include organized personnel whose technology or knowledge enables them to perform a core business function. It would treat a “material part” of a business as transferred whenever the acquirer becomes able to carry on the same activities as the transferor. It would count any consideration paid to the transferor toward the transfer price, whatever its label, including payments to release rights over departing personnel and license fees. It would also treat the transferor’s cessation of business as implementation of the transfer.[5]
We share the principle behind the Proposed Amendment. Competition law should look to a transaction’s substance, not its form. A transaction that strips a firm of its competitive significance should not escape scrutiny merely because it takes the form of a hiring arrangement and no shares change hands. As drafted, however, the amendment would bring ordinary licensing and hiring within merger control. We recommend that the KFTC withdraw it or narrow it substantially. Our comments make three points.
First, the case for extending notification has not been made (Section I). No competition authority has found an acquihire anticompetitive. The United Kingdom’s Competition and Markets Authority (CMA) cleared the Microsoft/Inflection transaction in Phase 1. Brazil’s Administrative Council for Economic Defense (CADE) also cleared it through an expedited review, finding that “[t]he transaction is not likely to result in significant changes to the competitive landscape in Brazil and therefore does not require a more in-depth analysis.”[6] Structuring a transaction to stay below notification thresholds is not misconduct. It is consistent with shareholders’ interest in minimizing transaction costs. Nor does the concern that an acquihire might neutralize a nascent competitor warrant amending or clarifying competition law. Error-cost considerations counsel restraint: Wrongly deterring these transactions is costly and difficult to reverse, while errors from wrongly permitting them tend to correct themselves.
Second, most acquihires are not mergers in the competition-law sense (Section II). The Notification Guidelines define a business as a bundle of property rights. Employment contracts are not property rights in employees, who have a constitutional right to choose where they work.[7] An acquirer instead obtains a coordinated but revocable set of employment decisions, often alongside a nonexclusive license that leaves rivals free to use the same technology. Against this backdrop, the proposed “material part” test asks what the acquirer can do rather than what the transferor has lost. The proposed implementation rule also ties the parties’ obligations to decisions they do not control.
Third, the proposed transfer-price calculation would not track competitive significance and would create perverse incentives (Section III). It would count license fees paid for access without control. And because it would count only payments to the transferor—payments that return capital to a startup’s investors—it would burden compensated, cooperative transactions while leaving uncompensated raids on a startup’s workforce untouched.
The KFTC explains that the Proposed Amendment seeks to prevent parties from evading notification obligations, noting that large global technology firms have used acquihires to recruit key personnel from AI startups.[8] This concern echoes statements by U.S. enforcers and legislators.[9]
A transaction’s exclusion from a notification regime says little about its competitive effects. Every threshold invites parties to structure transactions around it. Notification regimes deliberately leave many transactions unreviewed to focus scarce resources on those most likely to alter competitive conditions and give parties a predictable basis for planning.[10]
Acquihires also arise partly because conventional acquisitions face heightened scrutiny. When firms respond with licensing and hiring arrangements that leave the target independent, extending merger control to those arrangements tells them they are “damned if they buy and damned if they don’t.”[11] The relevant question is whether the transactions the Proposed Amendment would capture are likely to restrain competition. The published materials offer no evidence that they are.
Before extending notification obligations to a new class of transactions, the KFTC should establish that the class poses a significant risk to competition, that existing law cannot adequately address that risk, and that the benefits of review outweigh the costs. The Proposed Amendment meets none of these conditions.
Published reviews and evidence from evolving AI markets favor restraint rather than a presumption of harm (Subsection A). Existing law reaches harmful hiring practices and allows authorities to weigh the evidence, reasonable justifications, and efficiencies (Subsection B). Acquihires also support team mobility, workers’ bargaining power, and startup investment. Wrongly deterring them would impose broad, often unobservable costs that could exceed those of wrongly permitting them, across sectors well beyond AI (Subsection C).
The KFTC states that it developed the Proposed Amendment through exchanges with authorities in the European Union, Germany, and the United Kingdom.[12] But if anything, those authorities’ experience points toward restraint. The CMA and CADE have both reviewed Microsoft’s arrangements with Inflection AI and cleared the transaction. These published reviews provide a more useful starting point than a presumption that acquihires harm competition.
Microsoft reportedly paid about $620 million for a nonexclusive license to Inflection’s models and $30 million for a waiver of claims arising from its hiring of Inflection staff, most of whom joined Microsoft.[13] The CMA asserted jurisdiction but cleared the transaction at Phase 1. It found that Inflection had a very small share of chatbot use, had struggled to grow or sustain its user base, and exerted no material competitive constraint on rival services.[14]
CADE likewise cleared the transaction through an expedited review, concluding that it was “not likely to result in significant changes to the competitive landscape in Brazil and therefore does not require a more in-depth analysis.”[15] CADE also observed:
Microsoft would not exceed a 30% market share in any of the segments related to the development and application of artificial intelligence tools, which would mitigate major vertical competition concerns.
It is worth noting that the artificial intelligence sector is rapidly evolving, with the boundaries of relevant markets not yet clearly defined; it is reasonable to assume that there remains scope for market contestability and significant gains before a state of stabilization is reached. Furthermore, there are no strong indications of a decline in rivalry that would warrant intervention.[16]
Germany’s Bundeskartellamt concluded that the transaction could, in principle, constitute a concentration but declined jurisdiction because Inflection had no substantial operations in Germany.[17] The European Commission took no decision after the Court of Justice curtailed referrals of transactions below notification thresholds in Illumina.[18] U.S. agencies have opened investigations but, to our knowledge, have brought no enforcement action against an acquihire.
The markets in which these transactions occur also show little sign of the entrenchment that critics predict. The cost of AI inference—running a trained model to produce an output—has fallen sharply. Epoch AI reports that “the price to achieve GPT-4’s performance on a set of PhD-level science questions fell by 40x per year. The rate of decline varies dramatically depending on the performance milestone, ranging from 9x to 900x per year.”[19]
New firms continue to enter at every layer of the AI supply chain, from computing resources to consumer services. Startups, rather than incumbents with the largest data holdings, produced the most successful early generative-AI services.[20] The empirical record on “killer acquisitions” also offers little basis for presuming harm. The study that popularized the term and prompted several investigations estimated that 5.3% to 7.4% of pharmaceutical acquisitions preempted future competition.[21] Evidence of comparable effects in digital markets remains limited.
One of that study’s authors, Florian Ederer, has since examined roughly 1,200 startup acquisitions by eight leading technology firms and found little evidence that they suppress innovation. Patenting in the relevant technology areas rose before acquisition and continued afterward where further acquisitions followed. Acquired patents received more citations, including from firms that hired the targets’ employees, and target inventors who stayed with the acquirer went on to patent more.[22]
We encourage the KFTC to take a wait-and-see approach while the evidence develops and generative AI remains in a formative stage. A notification obligation covering an entire class of transactions should rest on evidence that the class poses a significant risk to competition.
The harms critics attribute to acquihires—hollowing out a rival or hoarding talent to deny it to competitors—are forms of conduct the MRFTA already reaches. Article 45(1)(viii) prohibits unfair interference with another firm’s business activities. The Enforcement Decree identifies unjustly inducing or hiring another firm’s personnel as one form of interference when it significantly impedes that firm’s business.[23] The KFTC’s guidelines give examples that closely match the concerns raised about acquihires: recruiting a large share of a firm’s core personnel with excessive benefits, or recruiting key personnel the hiring firm does not need to obstruct a competitor. The guidelines also allow the KFTC to consider reasonable justifications and efficiency gains.[24]
The talent-hoarding theory also rests on thin evidence. The most prominent study of this theory infers that incumbents keep researchers idle using data that shows only that certain software firms retained more skilled employees after their patent protection weakened. Yet ordinary management decisions could explain the same pattern.[25] Where hoarding does occur and impedes a rival, the KFTC can assess it under the existing rules on the evidence.
Other provisions of Korean competition law address related conduct. Article 5 reaches abuse of a dominant position, including unfair interference with rivals’ business activities. Article 40 reaches agreements among competitors, including restraints on hiring. Korea’s trade-secret law addresses the misappropriation of confidential information by departing employees.[26] These laws allow authorities to assess conduct on an actual record without burdening the many transactions that raise no concern.
Acquihires serve important economic functions. They give founders and investors an opportunity for exit when a startup cannot raise further capital, preserve teams whose joint value exceeds the sum of individual hires, and reward entrepreneurial risk-taking. Sought-after engineers often prefer to move through an acquihire rather than leave individually, which could damage their standing with investors who might finance future ventures.[27]
Competition for AI talent has also boosted workers’ compensation.[28] Treating coordinated team moves as reportable business transfers would weaken those workers’ bargaining position. A team that can move together has a better alternative to its current employer—and therefore more bargaining power—than individuals who must move one at a time. Burdening employers who bid for teams would reduce competition for their services, contrary to competition law’s aims in labor markets.
For many target firms, failure is the realistic alternative to an acquihire. Practitioners and scholars describe acquihires as a soft landing for startups that cannot secure further funding, although the recent AI transactions that prompted regulatory scrutiny involved considerably better-capitalized targets.[29]
Making these exits costlier or less certain lowers the expected return on venture investment and, with it, the funding available to the next generation of startups.[30] Japan offers a useful contrast. Acquisitions now account for 88% of Japanese startup exits, and the Japanese government has made raising that share an explicit policy objective.[31]
The costs of enforcement errors are asymmetric. If the KFTC wrongly permits an acquihire, the error tends to correct itself because employees remain free to leave, join rivals, or found new firms. As Section II.B explains, acquired employees leave at higher rates than ordinary hires. If the KFTC wrongly deters acquihires, the costs fall on labor mobility and venture investment across the economy. Those costs are largely unobservable because the deterred transactions never occur.[32]
These costs would extend beyond AI and global technology firms. The Proposed Amendment applies to any acquirer that meets the MRFTA’s thresholds, and Chairman Joo Byung-ki has identified semiconductors and pharmaceuticals among the sectors of concern.[33] Their importance raises the costs of both mistaken intervention and inaction. Well-intentioned rules that chill procompetitive deals can do lasting harm to the competition they aim to protect.
The Proposed Amendment would bring acquihires within merger control by redefining a “business” under Article 9(1)(iv) of the MRFTA rather than amending the statute. This approach stretches provisions designed for transfers of assets, contracts, and an organized business to cover voluntary employee moves, often accompanied by a nonexclusive license.
The amendment departs from the existing definition of a business transfer and raises questions about the KFTC’s authority to expand statutory obligations through an administrative notice (Subsection A). Employees retain control of their human capital and remain free to leave, limiting the acquirer’s ability to retain a team (Subsection B). The proposed “material part” test overlooks what the transferor retains and whether rivals can still access its technology (Subsection C). Finally, the implementation rule ties the acquirer’s obligations to decisions it may not control, while conventional merger remedies cannot reliably restore a team to its former employer (Subsection D).
These problems call for a narrower approach tied to transferred property rights, the target’s loss of competitive significance, and acts within the parties’ control.
Article 9(1)(iv) of the MRFTA reaches the acquisition or lease of all or a material part of another company’s business, and the acquisition of all or a material part of its fixed operating assets.[34] The Notification Guidelines define a business as “a bundle of property rights organized for the company’s business purposes and functioning as an integrated whole,” including sales rights, intellectual-property rights, and permits of economic value.[35] A part of a business is “material” if it can operate as an independent business unit or its transfer significantly reduces the transferor’s sales.[36]
Both definitions describe something that changes hands. A business transfer moves assets, contracts, and the organization that uses them from one owner to another. The associated employment relationships generally pass to the transferee, subject to each employee’s right to object. In an acquihire, by contrast, employees resign and accept new offers, while the transferor retains its property.
The existing definition mentions personnel only as a component of sales rights, which include “sales-related organization, personnel, and agency relationships.”[37] Personnel therefore count when they form part of an organized business that changes hands together with its contracts and assets. The Proposed Amendment would make the movement of personnel and their know-how sufficient, even when the transferor transfers nothing else and continues to operate.
Adding organized personnel to a category defined by bundles of property rights makes the definition internally incoherent. Employees remain free to change employers after a deal. Their movement does not itself transfer property rights.
The amendment also raises a question of legal authority. The KFTC presents it as a clarification, but it would substantially expand statutory notification obligations. An administrative notice should not bring transactions within the statute’s scope when its existing provisions do not cover them. The National Assembly should decide whether to expand Article 9(1)(iv).
Article 15 of the Korean Constitution guarantees every citizen’s freedom to choose an occupation.[38] Other legal systems recognize similar principles. Section 6 of the U.S. Clayton Act provides that “the labor of a human being is not a commodity or article of commerce.” The House of Lords held in Nokes v. Doncaster Amalgamated Collieries that, absent a contrary statutory provision, an employment contract cannot transfer without the employee’s consent.[39] An employer offers terms that individuals remain free to accept or reject.
An acquirer therefore gains the prospect that a team will move together, rather than ownership of its human capital. Coordinated offers, bonuses tied to the team’s joining, and agreements not to sue align the incentives of individuals who could otherwise decide separately. Each employee remains free to decline, leave later, or join a rival.[40] Economic theory supports the same distinction. Under the property-rights theory of the firm, a firm owns the complementary assets around which employees organize their work, but cannot own their human capital.[41]
The evidence confirms that employee retention is uncertain. A study of roughly 230,000 acquired startup workers finds that “acquired workers show significantly higher turnover rates than regular hires,” with attrition concentrated among the highest earners.[42] Other research finds the same pattern among senior and technical staff, and finds that acquisitions increase acquired employees’ later entrepreneurship.[43] Acquihire theories of harm depend on the acquirer retaining a team. That assumption weakens fastest for the people whose departure from the target supposedly causes the harm.
Some commentators identify the relevant asset as the know-how the team carries.[44] This argument conflates two kinds of knowledge. Codified know-how—source code, model weights, and documentation—moves through licenses or assignments, to which existing legal analysis already applies. Tacit know-how—knowledge embedded in employees’ skills and experience—is inseparable from the people who hold it. Those employees remain free to leave. Where departing employees take information they have no right to use, trade-secret law supplies the remedy rather than merger control.
The Proposed Amendment adds a third way to satisfy the first component of the “material part” test. It would suffice that the acquirer becomes able, after the transaction, to carry on the same business activities as the transferor.[45] This test focuses on the acquirer’s capabilities rather than the transferor’s loss, treating hiring arrangements coupled with exclusive and nonexclusive licenses alike. An incumbent already active in the field will often satisfy it by definition, as will almost any hire that adds capability. As an alternative to the existing criteria, it also bypasses the requirement that the transferred part be capable of operating as an independent business unit—the requirement that gives “material part” its meaning.
The other jurisdictional requirements do little to narrow the test. The price threshold requires consideration equal to only 10% of the transferor’s total assets or 10 billion South Korean won (KRW).[46] The thresholds for the parties’ size reach any acquirer group with KRW 300 billion in assets or turnover and any transferor with KRW 30 billion.[47] Because total assets count toward these thresholds, a well-capitalized startup holding venture funding may qualify despite having little revenue.
Consider a hypothetical transaction. A Korean electronics group hires 15 engineers from a chip-design startup holding KRW 40 billion in venture funding and pays it KRW 12 billion for a nonexclusive license to its design tools. The startup retains its remaining staff, continues to serve other customers, and remains free to license the same tools to the group’s rivals. Under the Proposed Amendment, the transaction could qualify as a business transfer. The engineers form an organized team whose knowledge enables them to perform a core function, the group can now carry on the startup’s design activities, and the license fee exceeds KRW 10 billion. Yet nothing in this sequence suggests that the transaction has transferred or eliminated the startup’s competitive significance.
A more accurate test would focus on what the transaction removes from the market, including whether the acquirer’s rivals retain commercially meaningful access to the target’s technology.[48] When the target remains free to license its technology to others and continue operating, the acquirer gains a workforce it cannot own and a license its rivals can also obtain. By contrast, a formally nonexclusive license that operates as exclusive in practice, or prevents the target from using or licensing its technology, may transfer the target’s competitive position and properly qualify as a business transfer.
The transactions that prompted the Proposed Amendment—including those involving Inflection, Character.AI, Windsurf, and Groq—rested on nonexclusive licenses under which the target retained the right to use and license its technology.[49] The guidelines should distinguish these arrangements from transfers that remove a business’s competitive significance.
The KFTC has publicly indicated that ordinary recruitment and the hiring of a handful of employees will not count as business transfers.[50] The Proposed Amendment contains no such exclusion. The final guidelines should state it expressly.
The new implementation rule illustrates the mismatch. For organized personnel, the Proposed Amendment treats a business transfer as implemented when the transferor ceases its existing business.[51] Implementation determines when parties must submit a post-closing notification and what parties subject to pre-closing notification may do before clearance.[52]
The transferor’s board and investors decide when to cease operations, and the Proposed Amendment does not define cessation. A full wind-down would presumably qualify. A product pivot, a reduction in operations, or the sale of the remaining business to a third party is less clear. An acquirer could face liability for implementing a transaction before clearance because of a decision it neither made nor knew about.
The rule would also pressure startups to keep hollowed-out operations running during review, consuming investors’ capital to avoid triggering the acquirer’s obligations. The transitional provision compounds this uncertainty. It applies the amendment to business combinations whose notification obligation arises after its effective date, so a transaction signed years earlier could become notifiable if the transferor winds down afterward.[53]
Remedies present a further difficulty. Corrective measures for unlawful business combinations, such as orders to dispose of shares or transfer a business, presuppose assets the acquirer can relinquish.[54] The KFTC cannot order employees to return to a former employer without conflicting with their constitutional freedom to choose an occupation. Requiring the acquirer to dismiss the team would not restore competition, because nothing would ensure that the employees return to the target. Rescinding the license would deprive the target and its investors of the consideration that often keeps the target alive. A pre-closing hold would, in practice, prevent individuals from starting jobs they have already accepted.[55] A notification obligation that cannot lead to a workable remedy imposes costs without corresponding benefits.
Recommendation: Delete subparagraph (2) of the definition of a business, or limit it to personnel who move together with transferred property rights, such as exclusive intellectual-property rights, ownership of source code or model weights, or customer contracts. Replace the new “material part” criterion with a test focused on what the transferor loses and whether rivals retain access to the technology. Tie implementation to acts within the parties’ control.
Under the current Notification Guidelines, the “transfer price” includes consideration paid for the business or fixed operating assets, liabilities assumed, and annual fees for leases and management entrustment.[56] Each element reflects something the transferor gives up, such as ownership of assets or the right to operate a business for a specified period.
For organized personnel, the Proposed Amendment would count any economic consideration paid to the transferor for the transaction, “regardless of its designation.” Examples include payments to release rights relating to departing personnel and license fees for intellectual property needed for the business.[57] The KFTC presents this as a technical change to account for consideration spread across several contracts.[58] Its consequences are substantive. The transfer price determines whether a transaction meets the price requirement of the “material part” test. A personnel-based transfer that falls below that threshold is not notifiable, even if it satisfies the test’s other component.[59]
These rules would poorly distinguish transactions that warrant review. License fees and payments supporting lawful hiring do not establish that the acquirer gains control over the target’s competitive position, and their size does not establish competitive harm (Subsection A). Counting only payments to the transferor would burden cooperative deals, encourage uncompensated recruitment, and shift value toward selected employees at the expense of investors and those who remain. Unclear aggregation rules would compound the uncertainty and encourage parties to restructure payments to avoid notification (Subsection B).
A nonexclusive license fee buys access to technology without control over it. The licensor remains free to use the technology and license it to others, including the licensee’s rivals. For this reason, the U.S. Federal Trade Commission’s Premerger Notification Office has long treated exclusive licenses as potentially reportable asset acquisitions and nonexclusive licenses as outside the notification requirement.[60]
The transactions that prompted the Proposed Amendment illustrate the point. After Microsoft’s arrangements with Inflection, Inflection retained “the unfettered ability to commercialise” its models, including by licensing them to third parties. It also announced that it would offer its latest model through licensing partnerships and application programming interfaces (APIs) that let other software access its models.[61] Counting a nonexclusive license fee as the price of a business conflates access with control. The test proposed in Section II.C would capture a formally nonexclusive license that operates as exclusive in practice. The transfer-price calculation need not substitute for that inquiry.
The same problem affects the transaction-value threshold. Korea introduced it in 2021 on the premise that a high price for a target with little revenue may signal competitive significance that revenue-based thresholds miss. Authorities have rarely invoked it since.[62] Whatever the merit of that premise for acquisitions, it does not carry over to license-and-hire arrangements. Those payments buy a license the licensor can grant again and a set of employment decisions it cannot guarantee.
Payments to release rights over departing personnel serve a similar function. In the Microsoft/Inflection arrangements, the reported $30 million payment secured a waiver of claims arising from Microsoft’s hiring of Inflection staff. Such payments remove a litigation threat that would otherwise deter lawful mobility. They support the hiring arrangement without purchasing the target’s business.[63]
More generally, transaction prices reflect scarce talent, the opportunities new technology may create, the benefits of combining the target’s capabilities with the acquirer’s assets, and competition among bidders. None of these establishes that the transaction eliminates a competitive constraint imposed by the target.[64]
The price requirement compounds the problem. Measured against the transferor’s total assets, the 10% threshold captures the transaction’s size relative to the target rather than its effect on competition. For a small startup, almost any meaningful payment will meet it.
The Proposed Amendment counts only consideration paid to the transferor. These payments allow a startup’s investors, remaining employees, and option holders to recover value. An acquirer that recruits a startup’s team without paying the startup would not meet the price requirement. The rule would therefore burden cooperative transactions while leaving uncompensated raids on a startup’s workforce untouched. This result runs counter to the Proposed Amendment’s stated purpose of protecting venture companies.[65] It would encourage structures that leave investors with less and substitute litigation for negotiated settlements.
The rule would also encourage parties to shift value from payments to the transferor toward compensation paid directly to individual employees. This would arbitrarily favor founders and senior researchers over investors and employees who stay behind. The disputes following the Windsurf transaction illustrate these conflicts over how to distribute a deal’s proceeds.[66]
The press release explains that “regardless of its designation” aims to capture consideration spread across several contracts, but the Proposed Amendment provides no aggregation rules.[67] It does not identify which agreements form part of a transaction or the period over which payments count. Nor does it explain whether accompanying agreements for cloud computing, chips, or distribution count toward the price, or whether payments to the transferor’s shareholders count at all. These gaps prevent parties from determining whether they must notify and encourage them to stagger payments or route them through affiliates.
Recommendation: Exclude nonexclusive license fees and payments ancillary to hiring, such as releases of noncompete covenants and agreements not to sue, from the transfer price. Specify which agreements count, the aggregation period, and whether payments to the transferor’s shareholders count. If the KFTC retains a price rule for personnel-based transfers, set a separate, substantially higher threshold to reduce the risk of capturing harmless transactions, rather than treating price as a measure of competitive significance.
The concern that a firm might acquire a competitor’s business in substance without acquiring it in form supports a narrow rule. The Proposed Amendment instead would reach ordinary licensing and hiring, measure transactions by prices that do not establish competitive harm, and tie notification obligations to events outside the acquirer’s control. Employees remain free to leave, and conventional merger remedies cannot reliably restore a team to its former employer.
Published reviews and the evidence from AI markets instead favor restraint. Existing law already addresses harmful hiring practices, while acquihires can preserve teams, strengthen workers’ bargaining power, and provide exits that support startup investment. Wrongly deterring these transactions would impose broad, often unobservable costs extending beyond AI and global technology firms. Acquirers could shift toward uncompensated recruitment, staggered payments, and litigation, while costlier, less certain exits would reduce the returns that finance venture investment in Korea.
Firms would also face obligations they cannot reliably assess in advance under a combination of tests no peer jurisdiction has adopted. The United Kingdom, which the KFTC consulted, is moving toward narrower jurisdiction. Its government has proposed to narrow the open-ended share-of-supply test that gives the CMA broad discretion to assert jurisdiction, and the CMA has said it is less likely to investigate global transactions that other authorities are reviewing.
Guidelines serve their purpose when firms can predict how the KFTC will treat a transaction. Predictable merger control also helps make Korea an attractive place to found, finance, and acquire innovative companies. We recommend that the KFTC withdraw the Proposed Amendment and address harmful conduct through Articles 5, 40, and 45 of the MRFTA. Any expansion of statutory notification obligations should come from the National Assembly.
If the KFTC nonetheless proceeds, we recommend that it:
[1] Press Release, Korea Fair Trade Comm’n, Sinyuhyeong Gieop Gyeolhap-ui Sagakjidae Haeso-reul Wihan Gieop Gyeolhap-ui Singo Yoryeong Gaejeongan Haengjeong Yego [Advance Notice of Proposed Amendment to the Notification Guidelines for Business Combinations to Eliminate Blind Spots for New Types of Business Combinations] (Sept. 9, 2026) [hereinafter Press Release]; id. attach. (comparison of current and proposed provisions) [hereinafter Proposed Amendment]. This comment cites lettered subdivisions of the Notification Guidelines by their English equivalents (가 = a, 나 = b, and so on). Unless otherwise noted, translations of Korean-language materials are our own.
[2] See, e.g., Dirk Auer & Onyeka Aralu, ‘Reverse Acquihires’ and the Market for Talent: Antitrust Theory in Search of a Case, Concurrences No. 9-2026 (forthcoming) [hereinafter Auer & Aralu, Reverse Acquihires]; Dirk Auer & Mario A. Zúñiga, AI Partnerships and Competition: Damned if You Buy, Damned if You Don’t (Int’l Ctr. for L. & Econ., Working Paper, May 2026).
[3] Geoffrey A. Manne, Dirk Auer & Lazar Radic, Comment of the International Center for Law & Economics Concerning the Proposed Amendments to Korea’s Merger Review Guidelines (Int’l Ctr. for L. & Econ., Dec. 5, 2023).
[4] Dokjeom Gyuje mit Gongjeong Georae-e Gwanhan Beomnyul [Monopoly Regulation and Fair Trade Act], Act No. 3320, Dec. 31, 1980, amended by Act No. 17799, Dec. 29, 2020, art. 9(1)(iv) (S. Kor.) [hereinafter MRFTA].
[5] Proposed Amendment, supra note 1, §§ 4(b)(2), 4(c)(1), 4(d)(3), 4(f).
[6] Microsoft/Inflection, Merger Case No. 08700.005792/2026-71, Decision No. 475/2026/CGAA5/SGA1/SG, ¶ 19 (Braz. Admin. Council for Econ. Def. 2026), https://sei.cade.gov.br/sei/modulos/pesquisa/md_pesq_documento_consulta_externa.php?HJ7F4wnIPj2Y8B7Bj80h1lskjh7ohC8yMfhLoDBLddbNMaDk2AO6MZR6aBnSzp6KvSE3SnFkJaYX5WBvy-aRe0rVTcGut04mNHPDwDq9y_ebH8rs7cqfZFpufAf0t1Iw (“Neste contexto, entende-se que a Operação não possui o condão de acarretar alterações significativas ao ambiente concorrencial no Brasil, não demandando, deste modo, um maior aprofundamento da análise.”).
[7] Daehanminguk Heonbeop [Constitution] art. 15 (S. Kor.) (“All citizens shall enjoy freedom of occupation.”).
[8] Press Release, supra note 1.
[9] See Bloomberg Podcasts, FTC Chair Andrew Ferguson Talks Acquihires | Bloomberg Talks, YouTube (Jan. 16, 2026), https://www.youtube.com/watch?v=8u_hazxCplM; Letter from Sens. Elizabeth Warren, Ron Wyden & Richard Blumenthal to Gail Slater, Assistant Att’y Gen., Antitrust Div., U.S. Dep’t of Just., and Andrew N. Ferguson, Chairman, Fed. Trade Comm’n (Feb. 4, 2026), https://www.warren.senate.gov/wp-content/uploads/media/doc/final_-_warren_wyden_blumenthal_letter_to_the_department_of_justice_and_the_federal_trade_commission_on_big_tech_reverse_acqui-hires.pdf.
[10] See Auer & Zúñiga, supra note 2, at 6–7.
[11] Id. at 23–24; Brian J. Broughman, Matthew T. Wansley & Samuel N. Weinstein, No Exit, 100 N.Y.U. L. Rev. 1481, 1484–86 (2025) (situating license-and-hire transactions among responses to tougher merger enforcement).
[12] Press Release, supra note 1.
[13] Krystal Hu & Harshita Mary Varghese, Microsoft Pays Inflection $650 Mln in Licensing Deal While Poaching Top Talents, Source Says, Reuters (Mar. 21, 2024), https://www.reuters.com/technology/microsoft-agreed-pay-inflection-650-mln-while-hiring-its-staff-information-2024-03-21.
[14] Competition & Mkts. Auth., Microsoft Corporation’s Hiring of Certain Former Employees of Inflection AI and Its Entry into Associated Arrangements, Case ME/7103/24, Phase 1 Decision, ¶ 22 (Sept. 4, 2024), https://assets.publishing.service.gov.uk/media/6719ff5f549f63039436b3c8/__Full_text_decision__.pdf.
[15] Microsoft/Inflection, Decision No. 475/2026/CGAA5/SGA1/SG, ¶ 19.
[16] Id. ¶¶ 14–15.
[17] Press Release, Bundeskartellamt, Taking Over Employees May Be Subject to Merger Control in Germany—Bundeskartellamt Not Competent to Review Microsoft/Inflection Transaction as Inflection Has No Substantial Operations in Germany (Nov. 29, 2024), https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2024/29_11_2024_Microsoft.htm.
[18] See Joined Cases C-611/22 P & C-625/22 P, Illumina, Inc. v. Comm’n, ECLI:EU:C:2024:677 (Sept. 3, 2024), https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:62022CJ0611; Bundeskartellamt, supra note 17 (noting the Commission’s Sept. 18, 2024, announcement that it would not issue a decision).
[19] Ben Cottier, Ben Snodin, David Owen & Tom Adamczewski, LLM Inference Prices Have Fallen Rapidly but Unequally Across Tasks, Epoch AI (Mar. 12, 2025), https://epoch.ai/data-insights/llm-inference-price-trends.
[20] Auer & Zúñiga, supra note 2; Geoffrey A. Manne & Dirk Auer, From Data Myths to Data Reality: What Generative AI Can Tell Us About Competition Policy (and Vice Versa), CPI Antitrust Chron., Feb. 2024, at 2, 7–8.
[21] Colleen Cunningham, Florian Ederer & Song Ma, Killer Acquisitions, 129 J. Pol. Econ. 649 (2021).
[22] Florian Ederer, Regina Seibel & Timothy Simcoe, Digital (Killer?) Acquisitions (Nat’l Bureau of Econ. Rsch., Working Paper No. 35762, Sept. 2026), https://www.nber.org/papers/w35762.
[23] MRFTA, supra note 4, art. 45(1)(viii); Dokjeom Gyuje mit Gongjeong Georae-e Gwanhan Beomnyul Sihaengnyeong [Enforcement Decree of the Monopoly Regulation and Fair Trade Act], annex 2, item 8(b) (S. Kor.). This provision treats the conduct as an unfair trade practice rather than an abuse of dominance. The burden of proof is therefore lower: Plaintiffs need not show that the defendant holds a dominant position.
[24] Korea Fair Trade Comm’n, Bulgongjeong Georae Haengwi Simsa Jichim [Guidelines for Review of Unfair Trade Practices] pt. V, item 8(b) (S. Kor.).
[25] Geoffrey A. Manne, Gotta Catch ’Em All? Antitrust and the AI Talent Wars, Truth on the Mkt. (June 18, 2026), https://laweconcenter.org/resources/gotta-catch-em-all-antitrust-and-the-ai-talent-wars (discussing Shaolong Wu & Zefan Qian, Talent Hoarding and Upstream Innovation: Labor Market Distortions by Large Incumbents (working paper)).
[26] MRFTA, supra note 4, arts. 5, 40; Bujeonggyeongjaeng Bangji mit Yeongeopbimil Boho-e Gwanhan Beomnyul [Unfair Competition Prevention and Trade Secret Protection Act], Act No. 911, Dec. 30, 1961 (as amended) (S. Kor.).
[27] John F. Coyle & Gregg D. Polsky, Acqui-Hiring, 63 Duke L.J. 281 (2013).
[28] Auer & Zúñiga, supra note 2, at 14–15.
[29] See Elizabeth Pollman, Startup Failure, 73 Duke L.J. 327 (2023) (discussing soft-landing acquisitions and acquihires as alternatives to formal insolvency); Louis Lehot, What an Acqui-Hire Looks Like for an AI Startup Today, Medium (July 16, 2026) (noting that most acquihires involve companies that raised less than $5 million and could not raise another round).
[30] Geoffrey A. Manne, Samuel Bowman & Dirk Auer, Technology Mergers and the Market for Corporate Control, 86 Mo. L. Rev. 1047 (2022).
[31] Geoffrey A. Manne, Dirk Auer, Lazar Radic & Mario A. Zúñiga, Comments of the International Center for Law & Economics to the Japan Fair Trade Commission Re: Opinions on the Draft of the Business Combination Guidelines 3–4 (Int’l Ctr. for L. & Econ., Aug. 27, 2026).
[32] For an explanation of the error-cost framework, see Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1, 15 (1984).
[33] Kim Nam-myung, Korea to Screen “Acqui-Hire” Deals as Corporate Mergers, Seoul Econ. Daily (Sept. 10, 2026), https://en.sedaily.com/finance/2026/09/10/korea-to-screen-acqui-hire-deals-as-corporate-mergers.
[34] MRFTA, supra note 4, art. 9(1)(iv).
[35] Korea Fair Trade Comm’n, Gieop Gyeolhap-ui Singo Yoryeong [Notification Guidelines for Business Combinations] § 4(b) (S. Kor.) [hereinafter Notification Guidelines].
[36] Id. § 4(c).
[37] Notification Guidelines, supra note 35, § 4(b).
[38] S. Kor. Const. art. 15.
[39] 15 U.S.C. § 17; Nokes v. Doncaster Amalgamated Collieries Ltd. [1940] AC 1014 (HL).
[40] Auer & Aralu, Reverse Acquihires, supra note 2; Onyeka Aralu & Dirk Auer, Acquihires and Antitrust: When Buying the Team Isn’t Buying the Company, Truth on the Mkt. (Apr. 9, 2026), https://truthonthemarket.com/2026/04/09/acquihires-and-antitrust-when-buying-the-team-isnt-buying-the-company [hereinafter Aralu & Auer].
[41] Oliver Hart & John Moore, Property Rights and the Nature of the Firm, 98 J. Pol. Econ. 1119 (1990).
[42] J. Daniel Kim, Startup Acquisitions as a Hiring Strategy: Turnover Differences Between Acquired and Regular Hires, 9 Strategy Sci. 118 (2024).
[43] Weiyi Ng & Toby E. Stuart, Acquired Employees Versus Hired Employees: Retained or Turned Over?, 43 Strategic Mgmt. J. 1025 (2022); J. Daniel Kim, Startup Acquisitions, Relocation, and Employee Entrepreneurship, 43 Strategic Mgmt. J. 2189 (2022).
[44] See Nina Fridman, When a Mass Resignation Becomes a Merger: Rethinking Asset Acquisitions for the AI Era, 93 U. Chi. L. Rev. 999 (2026); David T. Wong, Comment, An Acquisition by Another Name: Reverse Acquihires Under the Clayton Act, 135 Yale L.J. 3175 (2026) (treating human capital as an asset while distinguishing it from employees themselves).
[45] Proposed Amendment, supra note 1, § 4(c)(1).
[46] Notification Guidelines, supra note 35, § 4(c); Proposed Amendment, supra note 1, § 4(c)(2).
[47] MRFTA, supra note 4, art. 11(1); Shin & Kim LLC, Merger Control in Korea Vol. 13—KFTC to Bring “Acquihires” Within the Scope of Merger Review (May 29, 2026), https://shinkim.com/eng/media/newsletter/3312 [hereinafter Shin & Kim].
[48] Aralu & Auer, supra note 40.
[49] Auer & Aralu, Reverse Acquihires, supra note 2; Press Release, Groq, Groq and Nvidia Enter Non-Exclusive Inference Technology Licensing Agreement to Accelerate AI Inference at Global Scale (Dec. 24, 2025).
[50] Kim Nam-myung, supra note 33.
[51] Proposed Amendment, supra note 1, § 4(f).
[52] MRFTA, supra note 4, art. 11(8); Notification Guidelines, supra note 35, § 4(e).
[53] Proposed Amendment, supra note 1, addendum ¶ 2.
[54] MRFTA, supra note 4, art. 14.
[55] Auer & Aralu, Reverse Acquihires, supra note 2.
[56] Notification Guidelines, supra note 35, § 4(d).
[57] Proposed Amendment, supra note 1, § 4(d)(3).
[58] Press Release, supra note 1 at 2.
[59] Proposed Amendment, supra note 1, § 4(c).
[60] Fed. Trade Comm’n, Premerger Notification Off., Informal Interpretation 9307006 (July 1993), https://www.ftc.gov/legal-library/browse/hsr-informal-interpretations/9307006; Eric Fruits, Daniel J. Gilman, Onyeka Aralu & Brian Albrecht, ICLE Comments to FTC and DOJ on Premerger Notifications pt. IV.B (Int’l Ctr. for L. & Econ., May 22, 2026).
[61] Competition & Mkts. Auth., supra note 14, at 21.
[62] Shin & Kim, supra note 47, § 3.
[63] Broughman, Wansley & Weinstein, supra note 11; Auer & Aralu, Reverse Acquihires, supra note 2, pt. III (describing such covenants as ancillary to lawful hiring).
[64] Manne, Bowman & Auer, supra note 30; Geoffrey A. Manne et al., Comments of the International Center for Law & Economics: EU Draft Merger Guidelines—Public Consultation pt. II.A.3 (Int’l Ctr. for L. & Econ., June 22, 2026) (explaining that acquisition prices reflect complementarity, option value, and talent).
[65] See Kim Nam-myung, supra note 33.
[66] Auer & Aralu, Reverse Acquihires, supra note 2, pt. IV.
[67] Press Release, supra note 1, at 2.
Source: Korea Fair Trade Commission
I. Introduction and Summary of Recommendations The International Center for Law & Economics (ICLE) is a nonprofit, nonpartisan research center based in Portland, Oregon. In . . .
The International Center for Law & Economics (ICLE) is a nonprofit, nonpartisan research center based in Portland, Oregon. In comments to the Federal Trade Commission (FTC) in 2014,[1] 2018–19,[2] and 2022,[3] and to the Canadian Competition Bureau in 2025[4] and 2026,[5] we have explained why price differences that expand output and lower prices for price-sensitive consumers should not be deemed presumptively harmful. We have made the same point in work on the Kroger/Albertsons merger,[6] state grocery-pricing laws,[7] food-delivery fees,[8] and the FTC’s proposed enforcement policy statement on personalized pricing (Proposed Statement).[9]
The Proposed Statement says that, where consumers reasonably expect prices not to vary with their personal data, a seller that personalizes prices should disclose that fact, the basis for the personalization, and the types of data used.[10] Omitting any of the three is “likely to constitute an unfair or deceptive act or practice in violation of Section 5.”[11] Yet the same document reports that “the effects of personalized pricing on consumers are unclear.”[12]
The Commission thus proposes to presume unlawful a practice whose prevalence it does not know and whose effects, by its own account, could benefit or harm consumers. Section 5(n) requires the Commission to weigh the costs of enforcement mistakes in unfairness matters,[13] and a presumption can substitute for evidence only where theory predicts a high probability of harm.[14] Where theory predicts mixed effects, a presumption turns the agency’s uncertainty into the regulated firm’s liability.
The two possible errors also impose different costs. A seller that drops targeted discounts to avoid a disclosure dispute takes those discounts from the consumers who would have received them, and market forces may not restore them.[15] A seller that personalizes prices without disclosure, in a market where consumers can compare offers, faces rivals that can discount to win its customers. Disclosure mandates produce the first kind of loss nearly as readily as prohibitions do: A mandate raises the cost of the conduct it covers relative to conduct it does not, and firms may respond by changing what they do rather than by becoming transparent about it.[16] The Proposed Statement moves from “unclear” effects to “likely” illegality without weighing these errors.
The FTC’s deception policy recognizes that omissions can mislead consumers. But a consumer’s belief alone cannot establish that a seller conveyed a misleading message. The Proposed Statement’s second deception theory finds deception when a consumer “reasonably believes” a price is static or widely offered and the seller “fails to disclose” that it is personalized.[17] This assumption is unsubstantiated. A displayed price tells a buyer what she may pay; standing alone, it does not say what everyone else pays. The Commission’s International Harvester decision declines to treat such “pure omissions” as deceptive.[18]
Unfairness requires substantial injury that consumers cannot reasonably avoid and that countervailing benefits do not outweigh. The 11th U.S. Circuit Court of Appeals called those conditions “mandatory” in FTC v. Corpay Inc.[19] The Proposed Statement applies none of them to its seven hypothetical examples. Having called the practice’s effects “unclear,” the Commission cannot simply conclude that it is nevertheless “injurious in its net effects.”[20]
The Proposed Statement cites six economic works for the claim that “the more sophisticated personalized pricing practices become, the less likely consumers are to benefit.”[21] None tests disclosure, and none supports the claim as stated. Where the papers find consumer losses, they trace them to market structure, a seller’s pricing power, or unequal access to data among rivals. In each of the three empirical studies, most consumers pay less; in two, the market grows.[22]
The Proposed Statement also assumes that the benefits of personalization may “be realized without concealing” it.[23] But a notice that lets buyers avoid a higher price changes the prices a seller can profitably offer. A seller forced to discount either for everyone or for no one will discount less. The Commission has not examined how its proposed notice would change prices.
The notice’s second and third elements compound the problem. They require a seller to explain how it set a price—more information than consumers generally care about or use, and more than International Harvester contemplates for mandatory disclosure.[24]
The Proposed Statement improves on the 2022 advance notice of proposed rulemaking that preceded it.[25] It acknowledges that Congress has not authorized the Commission to prohibit personalized pricing,[26] drops the term “surveillance pricing,”[27] and proceeds through guidance, as we urged.[28] But useful guidance would explain how the Commission will assess materiality, injury, countervailing benefits, and reasonable avoidability in the cases it describes. The Proposed Statement announces a conclusion instead.
A document that tells every seller what to disclose also operates as a disclosure rule, whatever its disclaimer of legal effect.[29] Firms will plan as though it binds them, without the protections of Section 18 rulemaking.[30]
The Commission should withdraw the Proposed Statement and proceed case by case, as Corpay shows it can. If the Commission nonetheless issues a statement, it should preserve the concessions the proposal already makes and adopt the six revisions laid out in Part VII:
Only after resolving these legal and evidentiary questions should the Commission test which of the avoidance steps it describes that a notice would enable consumers to take, just as it tested its mortgage disclosures.[31] It should separately assess how sellers would change their prices in response.
Finally, any prescription of a disclosure’s wording, placement, or format constitutes a rule, which requires Section 18 rulemaking procedures.
The Proposed Statement makes four concessions that any final statement should retain. First, Congress has not authorized the Commission to prohibit personalized pricing outright. Second, personalized pricing “is a long-established norm in some markets,” including car sales, residential real estate, and markets where buyers and sellers negotiate.[32] Third, prices vary for reasons unrelated to personalization, including local supply and demand, regional taxes, and the individualized risk assessments used in insurance and credit. Fourth, real-world personalization is imperfect. An analysis that takes these concessions seriously must account for competition and the limits of consumer data and pricing tools.
Ample evidence that personalized pricing generally harms consumers might make those circumstances less consequential. The studies the Commission cites provide no such evidence. They report mixed effects on consumer welfare and show that the circumstances often determine the outcome.
The Commission begins its economic account with a monopoly benchmark, then concedes that the benchmark does not describe most markets. Perfect personalized pricing “without business competition would allow a monopolist to capture the entire consumer surplus.”[33] But the Commission immediately acknowledges that “real-world personalized pricing would be less than perfect in its ability to identify consumers and determine their willingness to pay and, in the vast majority of circumstances, would have to contend with competitive pressures that limit pricing power.”[34] That is the Statement’s most significant concession: In actual markets, imperfect information and competition may limit or eliminate the harm predicted by the monopoly benchmark. Without empirical findings about particular markets, the Commission’s case rests heavily on what six economic studies say about consumer harm under realistic conditions. As discussed below, those studies do not support many of the Statement’s claims or proposals.
Personalized prices can intensify competition by letting a seller compete for a rival’s customers without cutting prices for all its existing buyers. Under uniform pricing, a seller must extend a price cut meant to attract new customers to buyers who would have paid the old price. That makes price cuts more costly and less likely. Targeted offers avoid that cost, and rivals can respond with targeted offers of their own, multiplying the potential benefits.
The literature on price discrimination—charging different buyers different prices—supports this dynamic, although its effects depend on the market. Thisse and Vives show that it can lower prices and profits in a duopoly market with two differentiated sellers.[35] Corts shows why the result depends on how firms rank customer groups and the groups’ relative importance.[36] In his model, price discrimination can lower prices in every segment or raise them in every segment. A calculation based on a monopolist cannot, however, predict what happens when competing sellers can respond. As the antitrust literature puts it, price discrimination “implies neither that a market is not performing ‘competitively’ nor that it would perform more ‘competitively’ without price discrimination.”[37]
Imperfect data and competing offers limit a seller’s ability to personalize prices and the effects it can plausibly achieve. Data are noisy and often dated, and sellers estimate pricing models from samples rather than individual buyers. A consumer who can buy elsewhere—or decline to buy—will not pay more than a purchase is worth compared with her best alternative. Better data can improve a seller’s estimate and change the prices it can profitably charge, but the result remains an estimate.
In some markets, prices must depend on personal information. As then-Commissioner Ferguson observed:
This practice is ancient and commonplace for many products and services. The prices of insurance policies or of credit, for example, necessarily turn on a consumer’s personal details because the price must reflect the risk of providing the good or service to the consumer.[38]
The Commission also concedes that it does not know how widely sellers use personalized pricing.[39] The available evidence suggests limited use in the markets studied. A 2018 European Commission mystery-shopping study of 160 e-commerce sites in eight member states found personalized price differences in only 6% of tests. The median difference was less than 1.6%.[40]
Alberto Cavallo found no evidence that large U.S. multichannel retailers varied online prices with a shopper’s location or browsing history. He attributed that restraint to retailers’ fear of “antagonizing their customers.”[41] Stefano DellaVigna and Matthew Gentzkow found that most large U.S. food, drug, and mass-merchandise chains charged nearly uniform prices across stores despite substantial differences in local demographics and competition. The median chain forwent about $16 million a year in profit by doing so.[42]
Customers’ reactions already constrain a practice that firms forgo despite its potential returns. That restraint does not make enforcement unnecessary in every case, but it raises questions about what a general presumption would accomplish and what it would cost to apply one to the discounts that make up much of the practice.
The Proposed Statement cites six works in support of its three economic claims: Personalized pricing is likely to increase business profits; it benefits some customers while harming others; and more “sophisticated” practices are less likely to benefit consumers.[43] The Commission does not claim that personalized pricing has no benefits, causes net harm, or is prevalent. Its first two claims are accurate as far as they go, but leave out consequential findings. The cited literature does not support the third.
The six studies differ in what they can establish. Three are empirical studies but compare the observed setting with modeled pricing alternatives. Jean-Pierre Dubé and Sanjog Misra conduct field experiments on a business-to-business hiring platform and model the seller as a monopolist. Benjamin Reed Shiller estimates what personalization might have done for Netflix’s DVD-by-mail service in 2006, when the company had some pricing power but did not personalize prices. Nicholas Buchholz and coauthors model a Prague-based taxi-hailing platform whose drivers bid in an auction that does not rely on personalized data.
Two other papers are theoretical models without empirical data: Andrew Rhodes and Jidong Zhou’s model of competing firms in an oligopoly setting, and Rosa-Branca Esteves and Francisco Carballo-Cruz’s unpublished model of entry over two periods. The sixth work, by Ginger Zhe Jin, Liad Wagman, and Mengyi Zhong, summarizes the literature without adding a model or data of its own.
None empirically studies personalized pricing in a consumer retail market like those in the Proposed Statement’s examples. None tests the proposed disclosures, and none of the three empirical studies measures what happens when competitors respond.[44]
The Appendix to these comments includes a table that summarizes each paper’s setting and findings, how the Commission uses it, and how well it supports the Commission’s claims. Briefly, with respect to the claims, the studies show the following:
Assessing welfare also requires separating new sales from the redistribution of gains on sales that would have occurred anyway. A higher price on a purchase that still occurs transfers surplus from buyer to seller. A discount that induces an additional purchase that would not otherwise have occurred can create gains from trade. At the same time, price discrimination can also reduce welfare if it allocates a fixed supply to buyers who value it less. The traditional monopoly analysis developed by Robinson, Schmalensee, and Varian and surveyed by Stole weighs the effect on output against this allocation effect. Under its assumptions, output must increase for total welfare to rise. But neither higher output nor higher total welfare establishes that consumers collectively benefit.[45]
The three empirical studies the Commission cites estimate profit gains from personalization, but none evaluates how rival sellers would respond. Rivals could offer deeper discounts, retain customers who would otherwise switch, or both, reducing the estimated gains.
The studies model some customer responses, but none measures how competing sellers would respond. Dubé and Misra expressly exclude a competitive response from other platforms.[53] In the abstract, that dynamic should uniformly reduce the estimated effects. The distinction matters for Liftago. Buchholz and coauthors note that Prague’s regulatory setting is “different from most U.S. cities in which there is nearly free driver entry into the ride-hail market with firms such as Uber and Lyft.”[54] Where drivers can switch easily between competing platforms, a rival’s better offer would limit how much profit a platform could shift from drivers to itself.
The size of that competitive effect is uncertain, but its direction is clear: It would reduce the platform’s estimated profit gain. The papers therefore show that personalization can increase profits in the settings they model, while leaving the effect of competitive responses unanswered. In the limit case, competition could reduce any expected profit increase to zero. Indeed, although the Commission does not cite it for this point, Andrew Rhodes and Jidong Zhou model a market in which, when the market is fully covered, personalized pricing by all firms reduces industry profit.[55]
As the Commission acknowledges, the cited papers find that some consumers gain from personalized pricing while others lose. None of the three empirical studies, however, examines an ordinary retail buyer looking at a posted price in a U.S. store—the kind of buyer the Proposed Statement would affect.
The Commission also omits a finding common to all three studies: Most consumers gain surplus—the value they receive beyond what they pay—or receive lower price offers. In Buchholz et al., 62.5% of consumers gain surplus, while losses fall chiefly on riders least sensitive to price and waiting time.[56] Dubé and Misra estimate that roughly three-fifths of potential customers would receive a lower offer under personalization than under the profit-maximizing uniform price.[57] Shiller finds that “most consumers receive lower prices when prices are personalized, and hence the majority are slightly better off.”[58]
Rhodes and Zhou’s theoretical model shows why the broader result depends on the market and on which competitors have access to data.[59] When the market is fully covered— all potential consumers buy at the prevailing uniform price—and all firms personalize, industry profits fall and consumers collectively gain relative to uniform pricing. That result cuts against the Proposed Statement’s implicit presumption of consumer harm.
The result can reverse when the uniform price excludes some consumers. Personalization then brings more buyers into the market, but firms may collect enough from existing buyers to raise industry profits and reduce aggregate consumer surplus. In the authors’ numerical examples, industry profits begin to rise when uniform pricing excludes roughly 10%–20% of consumers. Aggregate consumer surplus begins to fall when it excludes roughly 20%–30%. Those thresholds depend on costs and the number of firms, while the model holds firms’ information fixed at a perfect level.[60] The remedies the authors consider—data sharing or restrictions on firms with more data—address situations in which only some firms can personalize. Neither is a consumer-disclosure remedy.
The benchmark also matters when interpreting the empirical studies. In Buchholz et al., replacing Liftago’s actual driver auctions with a single price set by the platform would cut consumer surplus by 33% and increase the share of riders who walk away from 37% to 53%.[61] But loss comes from replacing driver competition with centralized pricing. Moving from centralized uniform pricing to centralized personalized pricing reduces consumer surplus by a further 2.25%. Thus, the vast majority of the consumer-surplus loss comes from the platform taking control of pricing, rather than using consumer data.[62]
Dubé and Misra find a find a similar distinction. Moving ZipRecruiter from its actual uniform price of $99 to a profit-maximizing uniform price of $327 would raise profits by 55% and halve the share of prospective customers who buy, from 25% to 12%. Again, this entails no personalization. The paper measures the consumer-surplus effects of personalization against that $327 price rather than against $99, because they expect a maturing firm to charge the optimal uniform price even without personalization. Against that benchmark, theoretically optimal personalized pricing reduces total consumer surplus by 25%. The $499-capped personalization ZipRecruiter implemented reduces it by 2.04% while increasing profits by more than 8%.[63]
Personalization matters in both studies, but their headline changes cannot all be attributed to it. In Buchholz et al., personalization accounts for a small fraction of the consumer-surplus loss. In Dubé and Misra, the implemented policy reduces consumer surplus by about 2% against a uniform-price benchmark that itself halves the share of prospective customers who buy.
The studies’ most important finding is that personalized pricing affects consumers differently. In Rhodes and Zhou’s model, when the prevailing uniform price excludes some buyers, personalization both lowers aggregate consumer surplus and brings previously priced-out consumers into the market.[64] The studies find that the net effect of personalized pricing can sometimes be a reduction in aggregate consumer surplus.[65] They also find that personalization can expand access, that most consumers in the settings studied gain or receive lower offers, and that the beneficiaries tend to be more sensitive to price.[66] The Commission acknowledges that personalized pricing produces winners and losers, but it does not explain what follows from that dichotomy.
The FTC’s Unfairness Policy Statement treats a practice as unfair only if it is “injurious in its net effects.”[67] But the Proposed Statement does not claim that personalized pricing lowers aggregate consumer welfare. Instead, it points to particular consumers who pay more. That raises a question about distribution—who gains and who loses—that the cited studies help answer.
In Buchholz et al., losses fall chiefly on riders least sensitive to both price and waiting time, a profile unlike that of a passenger in an emergency. In Dubé and Misra’s study, larger employers hiring full-time workers with benefits tend to pay more, while smaller firms hiring part-time staff are most likely to gain. In Shiller’s study, subscribers whose browsing histories predict the highest willingness to pay face higher prices.[68] These findings do not support the impression left by the Statement’s examples that those paying more are the consumers least able to choose otherwise.
Section 5(n) also requires the Commission to weigh any “substantial injury to consumers” against “countervailing benefits to consumers or to competition.”[69] In the studies cited, however, the measured losses are small. Against a uniform-price benchmark, the estimated consumer-surplus losses are 0.5% in Shiller and 2.25% in Buchholz et al. For the capped personalization ZipRecruiter implemented, Dubé and Misra estimate a 2.04% loss against the optimal uniform price, although their estimate for uncapped personalization is much larger: 25% against that benchmark.[70]
Whether those numbers suggest substantial injury or not, they are static, partial-equilibrium sums with equal dollar weights. They do not show who gains or loses, and they value new participation only by the estimated surplus of each additional buyer. A measure that also considers distribution may reach a different result and the sign can even reverse. Dubé and Misra find that under inequality-averse welfare measures, personalization is preferred to uniform pricing “because the allocative benefits outweigh the decline in total surplus,” and that those benefits accrue mainly to smaller firms.[71]
Similarly, the assessment of benefits “to competition” permitted by Section 5(n) contemplates a separate calculation from the static consumer-effect balancing, one that includes the dynamic consequences of a larger market (e.g., entry, scale, investment, spillovers), not just consumer surplus.[72] Market size and access may, in fact, be more important than the consumer-surplus sign. Output expansion is antitrust law’s own measure of competitive benefit,[73] and traditional monopoly analysis treats output growth as necessary, though not sufficient, for price discrimination to increase total welfare.[74] Two of the three empirical studies find that personalization expands participation: About 7% more requests result in a ride in Buchholz et al., and about 3% more households subscribe in Shiller.[75]
The Proposed Statement instead treats a measured aggregate loss as presumptive injury without accounting for the majority who gain, the consumers newly able to buy, or the benefits of market expansion. That applies a welfare criterion the Statement has not explained or defended and does not complete the balancing Section 5(n) requires.
The Proposed Statement’s cited authorities do not establish that “the more sophisticated personalized pricing practices become, the less likely consumers are to benefit.”[76]
The Commission cites Dubé and Misra to suggest that the effects on consumers depend on the data available to a seller—and, implicitly, that more data means less consumer welfare.[77] The study finds a more complicated relationship. The pages the Commission cites give an example in which more detailed data raise consumer surplus and conclude that “increasing the granularity of the consumer data available to a firm can increase consumer surplus and even reduce deadweight loss.”[78] The authors report “a nonmonotonic relationship between the granularity of data and consumer surplus under personalization.”[79] In other words, more detailed data do not consistently help or harm consumers.
Across 62 potential personalization schemes, the authors find that “allowing the firm to target on more granular data can be good for consumer surplus,” that “granularity per se does not harm consumers,” and that “granting the firm access to more information is not per se worse for the consumer, as it can lead to segmentation schemes that allocate more surplus to the consumer.”[80] They warn that “overregulation of the types of data that firms can use for personalized pricing purposes could exacerbate rather than offset some of the harm to consumers.”[81] All 62 schemes do lower total consumer surplus relative to the optimal uniform price, but the size of that loss does not move consistently with the granularity of the data.[82] That finding undercuts the Statement’s general claim about sophistication.
The Commission cites Rhodes and Zhou for the finding that “when only some companies in a market can personalize prices, consumers can be worse off than when either all or no companies do so.”[83] That summary is accurate, but it does not support the sophistication claim. The harm in the model comes from unequal access to personalized data across firms, not from increasingly detailed data.[84] Indeed, the finding cannot depend on the amount or sophistication of the data, because the model gives every personalizing firm perfect information about each consumer. The authors leave the effects of differing amounts of information for future work.[85]
Buchholz and coauthors provide the Commission’s best evidence for its sophistication claim, and it is thin. The paper finds that, relative to a uniform price set by the platform, personalizing prices based only on riders’ sensitivity to waiting time leaves aggregate consumer surplus unchanged. Adding their sensitivity to price lowers it by 2.25%.[86] Yet nearly all of the consumer loss comes from the platform taking control of pricing. Replacing driver auctions with a uniform platform price costs riders a third of their surplus. The platform’s other pricing policies, from a menu using no rider data to full personalization, differ by less than four percentage points.[87] As the authors put it, “the losses in consumer surplus from personalized pricing are mostly due to the platform’s ability to centralize pricing and not its knowledge of consumer preferences.”[88]
Adding information thus has either no effect or a small negative effect on aggregate consumer surplus in the settings Buchholz and coauthors model. It also brings important benefits: 62.5% of riders are better off, the average fare falls, and about 7% more requests result in a ride. Indeed, the authors reach the opposite conclusion from the Commission’s about total welfare: “Conditional on the platform exercising this pricing power, the more information the platform incorporates into its pricing policy, the larger the overall welfare gains.”[89] Meanwhile, higher fares fall on the riders least sensitive to price.[90] The paper primarily demonstrates that a dominant platform with unconstrained pricing power can take surplus from its least elastic customers. That is much stronger support for the Commission’s market-power admission than for its sophistication claim.
Jin, Wagman, and Zhong’s literature review likewise does not establish the Commission’s general claim.[91] First, it is a literature review with no original model or data; it summarizes and assesses other economists’ work and is not independent evidence. Second, the authors warn that “as a firm with market power collects extremely fine-grained personal data, it can approach this limit case and potentially hurt all consumers.”[92] The claim is confined to the market-power setting, and the paper’s next sentence adds that “the concern about first-degree price discrimination can be mitigated by market competition.”[93]
Even where the paper claims that “access to finer-grained consumer information” allows a firm to “selectively undercut its rivals’ customer base,”[94] its support for the Statement’s sophistication claim is limited. As the paper notes, the result is a function of “raising switching costs for targeted consumers and creating data asymmetry among firms and consumers.”[95] Asymmetry can arise at any level of sophistication (although presumably more sophisticated data in some firms’ hands makes asymmetry more likely).
The review’s conclusion emphasizes the conditions the Proposed Statement overlooks. It says that “in competitive markets where multiple firms have access to consumer data, personalized pricing can intensify competition and benefit consumers,” and that greater pricing sophistication holds “the potential for both significant consumer benefits and serious market failures.”[96] Even where a firm has market power, “the consumer benefits can exceed the costs in net, depending on the market and product specifics.”[97] The authors’ first policy recommendation addresses the Statement’s central defect: “Regulatory frameworks should focus on market structure and competitive dynamics rather than treating all forms of personalized pricing equally.”[98]
The remaining paper the Commission cites for its sophistication claim, Esteves and Carballo-Cruz’s theoretical duopoly model, attributes harm to the asymmetry of access to data, not the sophistication of the data used. Consumer harm arises when an incumbent’s exclusive data prevent a rival from entering, leaving the incumbent as a monopolist.[99] When entry does occur, aggregate consumer surplus is highest if both firms personalize, lower if only the incumbent does, and lowest if neither does.[100] In other words, consumers collectively benefit as more firms personalize.[101] The authors conclude that, when entry costs are low, “price discrimination intensifies competition and overall consumer surplus is above its non-discrimination counterpart.”[102] In those markets, they find “there are no reasons to block price discrimination.”[103]
Other research confirms that more information can raise or lower consumer surplus, under both competition and monopoly. Mark Armstrong and John Vickers show that data identifying captive buyers, who consider only one seller, let firms charge those buyers more while competing harder for buyers who compare offers. Under the paper’s demand conditions, discrimination harms consumers collectively when the firms are sufficiently similar and can benefit them when the firms are sufficiently asymmetric.[104] In Brian C. Albrecht’s model, complete information available to all firms can maximize aggregate consumer surplus even though sellers charge loyal buyers the full value of the product: Competition drives prices for contested buyers down to the cost of serving them.[105] Dirk Bergemann and coauthors show that perfect price discrimination leaves consumers with zero surplus, their worst feasible outcome. But they also identify partial market segmentations that leave consumers better off than uniform monopoly pricing would.[106]
Different information and different market conditions determine whether greater sophistication harms or helps consumers. Better estimates of willingness to pay, the ability to identify a rival’s customers, and wider access to the same data can affect prices and welfare differently. Buchholz and coauthors find one refinement that slightly lowers consumer surplus and another that leaves it unchanged. The competitive models identify conditions under which more informative or more widely available data benefit consumers.
Evidence that personalization can sometimes benefit consumers is not, by itself, an objection to disclosure; nor is disclosure obviously an answer to firms’ variable ability to harm consumers with personalized pricing. The studies compare pricing policies and access to information; the Proposed Statement concerns what sellers tell buyers. The Commission has yet to ask how its proposed notices would affect consumer choices, firms’ access to data, their incentives to change prices, and the resulting gains and losses.
There is, however, no basis for the Proposed Statement’s broad, unqualified claim about sophistication. A final statement should withdraw the claim or identify when more detailed information can harm consumers. The relevant conditions would include the extent of competition, the amount and distribution of data, differences among firms’ ability to personalize, and the ease of entry.[107] The Commission must also specify the pricing benchmark against which it measures harm and separately assess how disclosure would change consumer choices and seller prices.
The Proposed Statement advances two theories of deception. Under the first, retailers “may deceive consumers in violation of Section 5 when they represent, expressly or by implication, that a price is static or widely offered when in fact it is personalized.”[108] That theory follows settled deception law by asking what the seller’s express or implied representations conveyed.
Under the second, retailers “may similarly deceive consumers when a consumer reasonably believes that a price for a good or service is static or widely offered, and the merchant fails to disclose that the price is in fact personalized.”[109] This theory rests on a consumer’s expectation without identifying what, if anything, in the seller’s presentation conveyed that prices were uniform.
A price tag makes a narrow, useful offer: A buyer may purchase an item at the displayed price. Standing alone, it does not say that every other buyer sees the same price at the same moment. Whether a seller’s broader presentation conveys that message depends on how consumers understand it, and that understanding can differ across markets.
U.S. shoppers already encounter many forms of price variation: clipped, downloaded, and register-generated coupons; loyalty-card prices posted beside shelf prices; student, senior, and military discounts; airline fares that change between search and booking; retention offers available only when a subscriber tries to cancel; and negotiated prices for cars, homes, and professional services. Prices for insurance and credit products, as the Commission itself says, “necessarily vary from purchaser to purchaser.”[110]
The Commission acknowledges that personalized pricing is established in some markets. In others, it infers that consumers reasonably expect “that the price they see for a product or service is the same price that any other consumer at the same place and time would see.”[111] But the absence of established personalization does not show that consumers read a displayed price as a promise of uniform treatment. In a store with a loyalty program, that expectation is already false, and loyalty-card holders know it. The Commission might call a loyalty price “widely offered,” but that answer concedes the point: Consumers may expect to see the terms of a price without expecting everyone to pay the same amount.
The Proposed Statement also leaves the expectation undefined. How many consumers must hold it, and how consistently?[112] A new customer may expect something different from a customer placing his 100th order. Expectations can change as pricing practices change. In a 2015 report on differential pricing, the Council of Economic Advisers said disclosure rules “should be based on a comparison of the compliance costs and the expected benefits from increased transparency.” It predicted that buyers “may come to expect personalized pricing in specific online settings, just as they do offline.”[113] A trigger based on current expectations risks freezing a norm that, as the Commission notes, varies by market. It also gives sellers little guidance when they set prices.
The Commission can still bring a deception case when a seller’s presentation conveys a uniform-price claim and the evidence establishes that claim. The Deception Policy Statement says the Commission “will often be able to determine meaning through an examination of the representation itself.” In other situations, it “will require extrinsic evidence that reasonable consumers reach the implied claims,” such as copy tests, surveys, or consumer testimony.[114] An alleged implied promise of uniform pricing calls for that evidence. The policy’s presumptions of materiality—for express claims, intended implied claims, and information about central product characteristics such as cost—apply only after the Commission identifies a claim or omission likely to mislead. They do not establish that a price display conveys a promise of uniformity.[115]
The FTC’s deception test has three elements: A representation, omission, or practice must be likely to mislead; the Commission assesses it from the perspective of a consumer acting reasonably in the circumstances; and it must be material, meaning likely to affect the consumer’s conduct or decision.[116]
An omission can satisfy the first element only when the context gives the silence a misleading meaning. In International Harvester, the Commission distinguished a half-truth—“where a seller fails to disclose qualifying information necessary to prevent one of his affirmative statements from creating a misleading impression”—from a “pure omission,” or “a subject upon which the seller has simply said nothing, in circumstances that do not give any particular meaning to his silence.”[117]
The Commission explained that it “does not treat pure omissions as deceptive.” Otherwise, deception would expand “virtually beyond limits” because “[i]ndividual consumers may have erroneous preconceptions about issues as diverse as the entire range of human error.” Moreover, deception law provides no cost-benefit analysis of the disclosure that a finding would compel.[118] The Commission therefore held that a manufacturer’s failure to disclose a fuel-system hazard was unfair, but not deceptive, allowing “a cost-benefit analysis” before ordering disclosure.[119] The Proposed Statement’s second theory treats exactly such a preconception as a representation.
International Harvester recognized an exception when silence occurs “under circumstances that constitute an implied but false representation.” That can include “ordinary consumer expectations as to the irreducible minimum performance standards of a particular class of good,” such as the expectation that a car can reach highway speed.[120] The Proposed Statement’s second theory appears to invoke that category. But the category is defined by an expectation that virtually every buyer of the class of goods holds and that the seller knows its product does not fulfill. The Commission has acknowledged that personalized pricing is well established in some markets. Expectations of uniform pricing therefore vary by market and consumer. The Statement turns an expectation it has not identified into a representation it need not prove.
If a consumer’s mistaken preconception is enough to create a duty to speak, a duty to speak, sellers must anticipate every consequential misunderstanding a buyer might hold. Sellers cannot survey that set of assumptions, and enforcers cannot define it in advance.[121]
The Proposed Statement’s citation to P. Lorillard Co. v. FTC, introduced with a “cf.” signal, does not establish a general duty to disclose the fact, basis, or data sources behind personalized pricing.[122] Lorillard promoted Old Gold cigarettes as lowest in nicotine, tar, and resin while selectively invoking a Reader’s Digest test. Read as a whole, the test found the differences among brands practically insignificant and did not support a claim that any brand was less harmful. The court therefore found that Lorillard’s own partial affirmative statements created a misleading net impression. The problem “was not that it did not print all that the Reader’s Digest article said, but that it printed a small part thereof in such a way as to create an entirely false and misleading impression.”[123]
Lorillard did not hold that silence alone is deceptive or that firms must disclose everything consumers might consider relevant. The other case cited in the same footnote, FTC v. Standard Education Society, concerned salespeople who falsely told prospects that a set of books was free and that its regular price exceeded the offered price—affirmative statements the Commission found false.[124]
The Proposed Statement grounds materiality in steps an informed consumer might take to avoid a personalized price. Consumers unaware of personalization “cannot take steps to avoid the higher prices that may result from it, such as using a virtual private network or private browsing functionality, choosing a different retailer whose prices are static or widely offered, rather than personalized, or simply declining to complete the transaction.”[125]
A virtual private network masks the address a server uses to infer location, but does nothing about personalization keyed to a logged-in account, a loyalty number, or a stored payment card—the personalization the Commission’s model disclosure describes as a price derived “from data about that consumer’s previous purchases from the same retailer through the same login account.”[126] Private browsing can clear cookies after a session, but it leaves purchase history held by the seller intact. A consumer would need to know which other retailers use uniform prices before switching to one; a notice from the current seller does not supply that information. And a consumer can decline any purchase at a price she considers too high, personalized or otherwise.
Materiality then depends on something narrower than the Commission claims: Would learning that a price was personalized cause consumers to shop elsewhere or delay buying? The Commission can test that proposition but has not done so. Nor does a presumption that information about cost is material resolve the issue. Materiality is the third element of deception; a presumption about it cannot establish the first—an act or practice likely to mislead.
FTC v. Colgate-Palmolive Co. does not establish a general duty to disclose personalized pricing.[127] The passage the Commission cites says a seller deceives consumers by falsely claiming that a product usually sells at an inflated price and is now available at a special discount.[128] The Court offered that example by analogy. The case itself concerned a television commercial that presented an undisclosed mock-up as an actual product demonstration.
Even so, the discount example turns on an affirmative false claim. The seller represents that the higher reference price is genuine and the lower price is a bargain. The Court used it to show that a false claim can be material even when buyers receive a product worth what they paid. It did not hold that a seller must disclose every fact that influenced its price.
A retailer that advertises “$40, marked down from $80” makes a false claim if the item did not ordinarily sell for $80. A retailer that simply lists an item for $40 makes no comparable claim about a discount, price uniformity, or its pricing method. The Proposed Statement initially respects that distinction by tying possible liability to an express or implied representation that a personalized price is static or widely offered. Its second theory drops the requirement by treating a consumer’s belief as sufficient without a seller-created representation. Colgate-Palmolive does not support turning a rule against false price claims into a general duty to disclose personalized pricing.
Section 5(n) limits the Commission’s authority rather than listing factors for it to balance. It provides that the Commission “shall have no authority” to declare a practice unfair “unless the act or practice causes or is likely to cause substantial injury to consumers which is not reasonably avoidable by consumers themselves and not outweighed by countervailing benefits to consumers or to competition.”[129] As the 11th U.S. Circuit Court of Appeals explained in Corpay, “[a]s ‘unless’ conveys, the three factors that follow it are mandatory for a finding that a practice is ‘unfair.’”[130]
The Proposed Statement, however, casts each requirement as a possibility: Injury “may be” substantial, consumers “may not” be able to avoid it, and benefits “may also” be realized another way. It then presents seven hypothetical examples without applying the statutory requirements to any of them.
The Proposed Statement closes with seven hypothetical examples of personalized pricing that, “without adequate disclosure,” would raise Section 5 concerns. A food-delivery service quotes a higher price to a customer it believes cannot leave home. A grocery chain charges more for milk to a household with several children. A hotel charges more to a traveler it believes is attending a funeral. A rideshare service charges more to a rider with no competing apps installed, and more again to one headed to a medical facility during an apparent emergency. A retailer charges a recent crime victim more for a security camera and raises its website price for a shopper standing in its parking lot.[131]
The examples differ sharply on whether a consumer can avoid the higher price. The shopper in the parking lot can check the shelf price at little cost. A rider without a competing app may be able to download one, though that takes time and effort. Indeed, the example’s surcharge depends on the seller’s belief that the rider is less likely to switch. By contrast, a passenger headed to an emergency room may have no practical chance to compare fares. The Proposed Statement groups these situations together despite their differences on the very question Section 5(n)’s avoidability requirement asks.
A finding of substantial injury also needs a comparison price—the price the seller would have charged but for personalization. The Proposed Statement supplies neither that price nor a method for finding it. The comparison is not necessarily the lowest price offered to anyone, but the price the seller would have set had it not personalized. It might be a uniform price or a price based on different customer information. Data that could support a surcharge for a household could, for example, also support a family discount. The grocery example does not say which outcome would occur.
A defined comparison price would also distinguish a loss in total consumer surplus from a redistribution of gains among consumers. Higher prices injure the buyers who pay them. If personalization also lowers prices for others, those buyers gain. Section 5(n) then requires the Commission to weigh the injury against benefits to consumers or competition. It cannot do so without estimating who gains, who loses, and by how much.
The statute requires a finding that injury is “not outweighed by countervailing benefits to consumers or to competition.”[132] Yet the Commission says the effects of personalized pricing on consumers are unclear.[133] Instead of estimating those effects, the Proposed Statement assumes that “[a]ny benefits to consumers or competition from personalized pricing may also be realized without concealing the fact and nature of the personalized pricing.”[134] Whether sellers would keep personalizing under a disclosure requirement, and which benefits would survive, is a question about how firms respond to compliance costs. The Commission should test those responses using its Section 6(b) record.
The FTC’s Unfairness Policy Statement adds two constraints absent from the proposal. The Commission “is not concerned with trivial or merely speculative harms,” and “[e]motional impact and other more subjective types of harm . . . will not ordinarily make a practice unfair.”[135] It also “will not find that a practice unfairly injures consumers unless it is injurious in its net effects.”[136] The Proposed Statement cannot establish net injury while describing the practice’s effects as “unclear.” To the extent it relies on consumers’ reported sense that personalized pricing is unfair, it relies on the kind of subjective harm that the Unfairness Policy Statement and the legislative history of Section 5(n) exclude.[137]
The Commission also identifies three ways consumers might avoid a higher price. It says they may lack the tools to “modify their behavior to avoid triggering higher prices, dispute or correct inaccurate information collected about them that is leading to higher prices, or avoid the collection of that data in the first place.”[138] A notice might help consumers compare offers or limit data collection without explaining every feature of a seller’s pricing model. Disputing an error, though, requires knowing which information the seller used. Limiting collection requires knowing where the information came from. Categories such as “purchase history” or “location” answer neither question.
This does not mean the Commission should demand a more detailed notice. It means the avoidability analysis cannot depend on steps the prescribed notice would not enable. A final statement should identify which steps the notice would make possible and confine its analysis to them.
The proposal’s unfairness theory turns on concealment: Consumers may be unable to avoid a higher price “if the fact or nature of personalization of the price has been concealed by the retailer.”[139] Consider the rider headed to a medical facility during an apparent emergency. Telling him that the fare reflects data suggesting an emergency removes the concealment but leaves both his circumstances and the surcharge intact. The same problem applies to a hotel guest believed to be attending a funeral. Both examples depend on the possibility that some personalized prices are unfair even when fully disclosed—a question the Commission expressly sets aside.[140]
FTC v. Corpay Inc. shows what an unfairness case supported by evidence can look like. The 11th U.S. Circuit Court of Appeals upheld summary judgment for the Commission on a claim that Corpay charged fleet-card customers fees it had not disclosed.[141] The record showed what customers were told and what the company knew about their confusion. A telephone survey by the Commission’s expert found that Corpay told only 7.02% of affected customers about all the fees before charging them. The survey also measured advance notice for each fee.
Internal emails and complaints showed that sales staff omitted the fees. Corpay kept its terms and conditions offline for three years and instructed employees not to send electronic copies. Eight internal surveys from 2016 through 2020 showed that its fees confused and surprised customers. Corpay’s expert testified that customers bought no less fuel after a fee was disclosed. The court found that testimony did not answer the evidence that Corpay had failed to disclose the fees in the first place.
None of the Proposed Statement’s seven examples supplies comparable evidence about what consumers were told, what sellers knew, or how many consumers could have avoided the charge.
The Commission invokes the Fair Credit Reporting Act (FCRA) to support a duty under Section 5 to disclose the reasons for a personalized price. It says the FCRA requires notice “not only when an adverse action is based on a consumer report (for example, the charging of a higher price), but also of the specific basis for that adverse action.”[142] Section 1681m(a) does not require that explanation. The Proposed Statement’s own footnote describes the provision more accurately: It requires an adverse-action notice and tells consumers how to obtain a free copy of their report “so they can identify and dispute any inaccuracies.”[143]
The notice must identify the adverse action; provide any credit score used, along with the key factors that adversely affected the score; identify the consumer reporting agency; and explain the right to a free report and to dispute its contents. The key factors explain the score, not the creditor’s decision. FCRA also requires “a statement that the consumer reporting agency did not make the decision to take the adverse action and is unable to provide the consumer the specific reasons why the adverse action was taken.”[144]
The reason-giving duty the Commission describes comes from the Equal Credit Opportunity Act (ECOA), which it does not cite. ECOA entitles an applicant subject to adverse action to a statement of reasons, which suffices “only if it contains the specific reasons for the adverse action taken.”[145] Even there, a creditor may discharge the obligation by notifying the applicant of the right to request those reasons, rather than volunteering them.[146] The duty applies to specified credit decisions, including a denial or revocation, a change to the terms of an existing arrangement, or a refusal to grant credit on substantially the terms requested.[147]
FCRA’s other notice provisions are equally specific. If a lender uses a consumer report to offer credit “on material terms that are materially less favorable than the most favorable terms available to a substantial proportion of consumers,” Section 1681m(h) requires a risk-based pricing notice. That notice identifies the report’s role in setting the terms, the reporting agency, and the right to a free copy and contact information for obtaining it. If the lender used a credit score, the notice must also provide the score and specified accompanying information, including the factors that adversely affected it.[148]
FCRA does require disclosure of “the nature of the information” behind certain decisions based on information from someone other than a consumer reporting agency, including a denial of credit or an increase in its charge. But the duty arises only after a consumer makes a written request within 60 days.[149] Disclosure to a consumer who asks differs from a notice on every displayed price.
Congress enacted these targeted requirements. That suggests Section 5 did not already compel them across markets. The same point applies to the state insurance provisions the Commission cites: Each reflects a legislative or regulatory decision to require notice in a particular market.[150]
Section 5(n) allows the Commission to consider established public policies as evidence, but says “[s]uch public policy considerations may not serve as a primary basis for such determination.”[151] The Proposed Statement quotes that limit in footnote 21, then relies on FCRA and state insurance law as its only statutory support for a disclosure duty.[152] The FTC’s Unfairness Policy Statement says public policy most often supplies “additional evidence on the degree of consumer injury.” When the Commission relies on it heavily, the policy should be “widely shared, and not the isolated decision of a single state or a single court.”[153] Two state insurance statutes, one state insurance regulation, and a credit statute with narrower requirements do not establish the general duty the Proposed Statement proposes.
The Proposed Statement leaves unclear how much detail a personalized-pricing notice must provide. It calls for disclosure of personalization, its basis, and the types of data used.[154] Its example of an adequate notice describes a price “based on a consumer’s estimated willingness to pay derived from data about that consumer’s previous purchases from the same retailer through the same login account.”[155] That explanation identifies the purpose and source of the data without revealing the full pricing model.
If the Commission retains these three elements instead of adopting a safe harbor, it should clarify what satisfies them when prices depend on several data sources. That clarification should, however, stop at what a notice must convey. If the Commission goes on to mandate the format of the disclosure—its wording, placement, or presentation—the document ceases to be a statement of enforcement policy and becomes a disclosure rule, one that would require the procedures of Section 18 rulemaking.
Disclosing what a consumer will pay differs from explaining how the seller arrived at the price. Clear final prices, honest reference prices, and understandable loyalty terms help consumers compare offers.[156] The Proposed Statement’s second and third elements instead ask sellers to describe their pricing methods and the data they use.
In International Harvester, the Commission advised that “[i]n most cases it is appropriate to limit mandatory disclosure to those core aspects of a transaction that virtually all consumers would consider essential to an informed decision.”[157] Price clearly meets that threshold. Whether the model behind the price does is far less clear. Even a 2014 White House report drew this line, calling it “implausible for consumers to be presented with the full parameters of the data and algorithms shaping their online and offline experience.”[158] The Council of Economic Advisers’ 2015 follow-up described a narrower option: a rule “requiring disclosure that prices or search results may vary across users.” It said such a rule “should be based on a comparison of the compliance costs and the expected benefits from increased transparency.”[159]
Petra Persson models how a firm can satisfy a disclosure mandate while defeating its purpose. The firm complies, then sends “an increasing swarm of irrelevant cues” that obscures the required information and leaves readers worse off.[160] Her model formalizes a prediction Frank H. Easterbrook and Daniel R. Fischel made in 1984: Firms “may cease disclosing some category of useful information and switch to some obfuscatory (but complying) information.”[161] Persson also shows how a firm told to disclose information in an easy-to-understand format can complicate its product until that requirement no longer helps.[162] She concludes that “simple disclosure rules can be completely impotent and have no impact on social welfare.”[163]
At the same time, the consumer who must read what the firm produces bears a cost of her own. Mandated disclosures “provide diminishing returns and increasing costs,” while “[o]ptimal disclosure is not maximal disclosure.” Yet commentators often “take for granted that increased disclosure is beneficial—and cheap.”[164] Additional information can “corrode[] the decision-making value of information” by forcing consumers to separate what matters from what does not.[165] The consumers least equipped to do that may be the very people a mandate aims to help.[166]
The Proposed Statement’s second and third disclosure elements ask a shopper at checkout to do exactly that. A notice saying a price reflects “estimated willingness to pay derived from data about that consumer’s previous purchases” describes a process.[167] It does not tell the shopper what she would pay elsewhere or what action to take. Nor would a better-drafted paragraph solve the broader problem. A single disclosure prescribed for every seller and market replaces judgments tailored to particular transactions with “a blunt ex ante standard of disclosure.” As Geoffrey A. Manne observes, “[t]here is no reason to believe that the costs imposed by the blanket standard, although different in nature, are any smaller in magnitude” than the costs it displaces.[168] The Commission should instead ask whether each added element improves consumers’ decisions sufficiently to justify its costs in attention and changes to sellers’ behavior. The Proposed Statement does not ask that question.[169]
The Organization for Economic Co-operation and Development (OECD) accordingly advises governments to “limit the prescriptiveness and extensiveness” of mandated disclosures to what is necessary. It also warns of an accumulation problem: Disclosures that seem defensible one by one can overwhelm consumers in combination. Quoting Omri Ben-Shahar and Carl Schneider—“[o]ne disclosure by itself may seem trivial, but en masse disclosures are overwhelming”—the OECD warns that “even a well-designed personalised pricing disclosure may be easily overlooked in the presence of a relatively more salient health warning.”[170]
Accumulation is a problem even at this Commission. The Rule on Unfair or Deceptive Fees governs how sellers display total prices for live-event ticketing and short-term lodging.[171] In March and April 2026, the Commission sought comment on rules for rental-housing fees[172] and for fees in online food and grocery delivery.[173] While each of those disclosures is confined to its market, a personalization notice would not be. It would apply to every seller that personalizes, joining the fee disclosure the Commission already requires, along with the privacy notices, cookie banners, and terms that already compete for the same checkout screen. The consumer’s attention to that screen does not grow with the number of notices on it.
Research by Lisa Robinson, Kip Viscusi, and Richard Zeckhauser suggests that generic warnings can weaken the force of warnings about serious risks. They found that 69% of surveyed consumers thought a generic Proposition 65 cancer warning on breakfast cereal conveyed a risk as great as a cigarette warning from 1966. “Taking cigarettes as the benchmark for wolves,” they write, “the Proposition 65 warnings are largely about puppies.”[174] A notice on every personalized price would likewise attach the same signal to a loyalty discount and to the emergency rideshare surcharge in the Commission’s example. The OECD’s review of online disclosures adds that consumers “might mistake, for example, neutral information for a warning, focus their attention unduly on disclosed information, rather than other product characteristics,” or react to it emotionally.[175]
Howard Beales and Todd J. Zywicki similarly warn that “disclosures may misdirect attention, if they imply that consumers should pay attention to information that has no real significance for them.”[176] A personalization notice could do just that for a shopper receiving a discount.
Anna Priester and coauthors, whom the Commission cites in support of disclosure, asked participants to imagine buying a digital movie and rate the fairness of the price they were told they received. Individually personalized prices drew lower fairness ratings than prices set for customer groups, though the difference was marginal in both studies.[177] Every participant was told the pricing rule, so the researchers did not test the effect of disclosing it. Their study therefore measures reactions to a described practice, not whether a notice helps consumers understand a price or act on it.
Finally, the academic case for algorithmic-pricing transparency is narrower than the Proposed Statement suggests. Oren Bar-Gill, Cass R. Sunstein, and Inbal Talgam-Cohen tie harm to algorithms that exploit consumers’ misperceptions. They describe pricing based on preferences and wealth as causing “limited harm,” and a disclosure that a seller charges wealthier consumers more as revealing a “(not particularly alarming) fact.”[178]
The OECD two personalized-pricing notices. Working for its Committee on Consumer Policy, Ireland’s Economic and Social Research Institute recruited 210 participants in Dublin and 210 in Santiago, Chile. Each completed eight purchase tasks on simulated retail sites. Some saw no notice. Others saw a statement that “[w]e used your personal data in calculating this/these price(s) for you.” A stronger version added that the price “may be higher or lower than for other customers.” The stronger notice increased stated awareness among Irish participants after they received a definition of personalized pricing, but awareness remained low. The study “failed to provide evidence of a significant impact of the disclosures on participants’ purchasing behaviour.”[179] Between 62% and 94% of Irish participants and between 85% and 96% of Chilean participants did not recall seeing a notice.[180] The authors concluded: “Neither disclosure was effective in protecting participants from paying more than they needed to.”[181]
While interesting, the experiment has two important limits. First, prices varied randomly rather than according to personal data, a limitation the authors identify themselves.[182] At the same time, 66% of Dublin participants and 51% of Santiago participants said personalized pricing should not be allowed.[183] The absence of a significant effect on purchases therefore does not show that participants approved of the practice. Second, results from simulated sites in two cities need testing before they can support a U.S. retail policy. The notices were prominent and repeated—likely more so than a notice at checkout[184]—yet most participants did not recall seeing them.[185]
Willem H. van Boom and coauthors tested how notice wording affected stated purchase intentions. They first studied 426 Dutch university students choosing among hotel rooms, then repeated the experiment with 186 U.S. participants recruited online.[186] Notices suggesting that personalization served the buyer’s interests—because the price might be lower than others paid or was “appropriate”—raised stated purchase intentions. In the U.S. study, the effect operated entirely through participants’ belief that the practice served their self-interest.[187] Notices also tended to raise stated purchase intentions compared with no notice.[188] It should be noted in this regard that, under the Proposed Statement, sellers would draft their own notices and choose the wording. The van Boom study suggests that wording, rather than disclosure alone, can shape consumers’ responses.
Allison Nicoletti and Christina Zhu studied roughly 22 million loan applications after the Truth in Lending Act–Real Estate Settlement Procedures Act (TILA-RESPA) Integrated Disclosure rule took effect. They compared covered mortgages with exempt open-end loans, which let borrowers draw credit repeatedly. Borrowers became more likely to withdraw covered applications or leave them incomplete, a result the authors interpret as consistent with lower information-processing costs. Approval rates also fell by about 5 percentage points. Compliance costs and frictions in selling loans to investors reduced credit availability.[189] The authors draw no conclusion about net welfare and call their findings “primarily descriptive.”[190]
Calorie labels can change purchasing behavior, although their average effect is small. A 2025 Cochrane review by Natasha Clarke and coauthors finds strong evidence of a small reduction in calories selected or purchased. The evidence on calories actually consumed is less certain.[191] Pasquale E. Rummo and coauthors found 24.7 fewer calories purchased per transaction across 2,329 Taco Bell restaurants. Most restaurants subject to labeling were in California. For restaurants outside California, the pooled estimate showed no statistically significant change over the three- to 24-month follow-up.[192]
These studies offer a narrower lesson than the Proposed Statement draws. Calorie labels are brief. Mortgage forms span several pages, but they use standardized fields and quantities that borrowers can compare when choosing a loan. A notice that prices may vary with personal data could be brief and standardized. An explanation of how a seller set a price may be longer and harder to standardize. It also gives consumers no alternative price to compare at the moment of purchase.
None of these studies establishes how consumers would use the Commission’s proposed notice at a U.S. retail checkout. The Commission should first resolve the legal and evidentiary questions discussed above and define the scope of any disclosure duty. If it proceeds, it should conduct or commission consumer testing comparable to its mortgage-disclosure research.[193] The tests should ask whether consumers understand the notice, can use it to compare offers, and make different purchasing decisions. Testing could inform a safe harbor or an eventual Section 18 rule. The Commission must separately assess how disclosure would change sellers’ prices and consumer welfare.
The Commission cannot assume that personalized pricing’s benefits would survive its proposed disclosure requirement.[194] If a notice helps consumers avoid higher prices, sellers may lower those prices, raise discounted prices, change how they make offers, or narrow eligibility for discounts. The Commission must assess the resulting gains and losses across consumers before concluding that disclosure would preserve the benefits.
The Proposed Statement assumes that “[a]ny benefits to consumers or competition from personalized pricing may also be realized without concealing the fact and nature of the personalized pricing.”[195] That assumption treats sellers’ conduct as fixed and the notice as a costless addition. But it is a mistake to “treat regulated behavior as static” when regulation makes some conduct more expensive.[196] As Geoffrey A. Manne explains, “[d]isclosure regulation makes some forms of behavior more expensive relative to others. Rational actors will respond by shifting some conduct into the comparatively cheaper outlets. And these alternative behaviors may actually be less beneficial than the regulated, deterred behavior.”[197]
The size of that shift depends on how closely other practices can substitute for personalized pricing.[198] Sellers have many alternatives to an individualized discount that requires a three-part notice and may still invite enforcement scrutiny.[199] They can charge everyone one higher price, vary prices by location or time of day, offer coupons through intermediaries, or use bundles and loyalty tiers. They can also let consumers choose among offers. Buchholz and his coauthors describe one example: pricing based on a rider’s expected wait time rather than purchase history.[200] These alternatives may preserve some gains from personalization, but they may miss consumers who would have received a targeted discount. The losses will be hard to observe. The Commission may see fewer personalized prices without seeing the discounts that sellers stopped offering.[201] And the benefits cannot be assumed: “It is insufficient either to ignore unanticipated consequences or to presume that the only relevant effects will be the salutary, intended ones.”[202]
Price discrimination also depends on limits to buyers’ ability to obtain another buyer’s lower price. If a notice lets a buyer qualify for a discount at a cost below the price difference, the seller must account for that possibility when setting both prices. It may lower the higher price or raise the discounted price until seeking the discount is no longer worthwhile. Consumers need not actually change their behavior for this constraint to affect prices. An assessment that holds prices fixed would miss these effects.[203]
Where disclosure helps consumers in these models, it does so because sellers change their behavior. Xi Li and Krista J. Li find that making it cheaper for consumers to manipulate the data used in pricing can reduce consumer surplus.[204] In their model, sellers raise prices for buyers with low willingness to pay, while other buyers incur costs to obtain discounts. Disclosing the scope of data collection, by contrast, weakly increases consumer surplus in the same model. Sellers collect less data, consumers spend less effort manipulating it, and prices fall for buyers with low willingness to pay.
Consumers’ information choices can also change the offers competing firms make. S. Nageeb Ali, Greg Lewis, and Shoshana Vasserman study buyers who choose which verifiable information to reveal to each seller.[205] A buyer can obtain a lower offer from a rival by revealing information to that rival while withholding it from a preferred seller. Under the model’s demand assumptions, the authors construct an equilibrium in which every consumer type benefits relative to a market without personalization. Competition for the buyer drives those gains. A notice alone may give consumers no comparable control over their information. To apply this result, the Commission must identify what its notice would enable consumers to do and how competing sellers would respond.
Consumers who currently receive lower personalized prices could lose out if disclosure prompts sellers to withdraw those offers. Dubé and Misra estimate that roughly three-fifths of potential business customers would receive lower offers under personalization than under a profit-maximizing uniform price.[206] Those customers’ potential losses belong in the analysis. Whether the discounts would disappear depends on whether they would remain profitable after consumers respond to the notice.
Evidence on geographic pricing shows that a move toward common prices can affect consumers unequally. DellaVigna and Gentzkow find that food-store prices near the 10th percentile of demand elasticity within a chain are only 0.4% higher than prices near the 90th percentile. A model with common marginal costs predicts a difference of about 17%.[207] Their calibration also suggests that uniform pricing raises prices at stores serving the lowest-income decile by roughly 10% relative to stores serving the highest-income decile, compared with flexible pricing. The authors caution that competitive responses, including entry and exit, could change those results.[208] That caution reinforces the need to account for consumers who benefit from existing price differences.
Rules that push sellers toward uniform prices offer another warning. When sellers must extend a discount to everyone or no one, they may discount less. The Antitrust Modernization Commission reached that conclusion about the Robinson-Patman Act, which has exposed sellers to litigation over price differences and pushed them toward rigid uniform pricing.[209] The FTC has likewise alleged that Amazon’s “anti-discounting” tactics, “[b]y taming price cutters into price followers,” suppress price competition and deprive shoppers of lower prices.[210] Competition can produce individualized discounts, and a disclosure regime that raises the cost or legal risk of offering them weakens sellers’ incentives to do so. Research on pay transparency points to a similar risk: when individualized terms become visible, firms may narrow the differences and offer fewer favorable terms.[211]
The effect of personalization on consumers depends on how firms compete for different buyers. If a notice changes which customers a seller can attract with a discount, rivals may change their offers in response. The welfare comparison must therefore account for prices across competing sellers, rather than treating the avoidance of one seller’s higher price as the entire consumer benefit.
The Commission needs evidence on these pricing responses before concluding that disclosure would preserve the benefits of personalization. Its January 2025 interim staff perspective reports initial findings from the Section 6(b) study and acknowledges limits on their representativeness and on conclusions about effects.[212] Further aggregated evidence should show which consumers receive higher and lower prices, what the proposed notice would enable them to do, and how sellers would adjust their offers. The Commission should compare gains and losses across consumers under the resulting prices.
The Proposed Statement concedes that personalized pricing’s effects on consumers are unclear. The literature and evidence do not support its broader inferences, and the Commission has not done the analysis needed to justify the proposed policy. The Commission should withdraw it. If it proceeds, we urge it to adopt the following recommendations:
Qualify the third economic claim and develop the evidence. Withdraw the claim that more sophisticated personalization is less likely to benefit consumers, or identify the conditions under which more detailed consumer information can cause harm. Build on the Section 6(b) release with aggregated evidence on pricing, consumer welfare, and disclosure’s likely effects. Once the Commission has addressed the evidentiary and legal concerns and clarified what it proposes to require, it should test whether consumers can use the notice to compare offers, correct information, or avoid higher prices. It should separately assess how sellers would change their prices and which consumers would gain or lose.
Clarify what a displayed price display conveys. State affirmatively that a displayed price, standing alone, does not represent that every consumer sees the same price at the same time. Whether a seller’s broader presentation conveys that claim depends on evidence from the market in which it appears.
Ground deception claims in established law. Strike the Proposed Statement’s second theory of deception, which finds that a seller deceives consumers whenever they “reasonably believe” a price is static or widely offered and the seller does not disclose personalization. Retain the first theory, which requires a seller’s express or implied representation. A false claim that a price is uniform or generally available can violate Section 5 if the Commission proves each element of the Deception Policy Statement’s three-part test. Where the claim is implied, the Commission should use extrinsic evidence to establish what consumers understood.
Provide a disclosure safe harbor. Replace the requirement to disclose the basis of personalization and the types of data used with a safe harbor: A clear and conspicuous statement that prices may vary based on customer data defeats a claim of implied uniformity. This would tell sellers how to avoid implying that prices are uniform without requiring an explanation of how they set each price.
Identify practices the Commission does not presume unlawful. Name individualized discounts, coupons, loyalty pricing, retention offers, and negotiated prices as practices the Commission does not regard as presumptively unlawful. That clarification would reduce the risk that uncertainty deters beneficial discounts. The protection should exclude cases where a seller inflates a nominal public price merely to present every individualized price as a discount. The public price must be offered in good faith to a meaningful number of consumers for a meaningful period. Within that limit, a personalized price no higher than the price generally offered at the same time should be safe.[213]
Analyze the examples or present them as questions. For each of the seven hypotheticals, the Commission should identify the relevant alternative price, explain the substantial injury and whether consumers can reasonably avoid it, and assess gains and losses across consumers and any countervailing benefits to competition. Otherwise, it should identify the examples as questions for further study. Corpay illustrates the evidence an unfairness claim requires: a customer survey, internal surveys and emails, complaint data, and evidence of what customers were told about each fee. The Commission also supported its deception claims with the advertisements and internal documents.[214]
The Proposed Statement relies on six works to summarize the economics of personalized pricing.[215] This table identifies each paper’s methodology, the claim the Commission cites it to support, its relevant findings—including those the Commission omits—and whether it supports the Commission’s claim.
Page references are to the cited works; note numbers (n.8–n.13) are the Proposed Statement’s. Quotations from the Statement are at 3–4.
[1] Berin Szoka, Tom Struble, Geoffrey A. Manne & Ben Sperry, Comments of TechFreedom & the International Center for Law & Economics in Response to Big Data: A Tool for Inclusion or Exclusion? 11–12 (FTC Project No. P145406, 2014), https://www.laweconcenter.org/resources/comments-response-ftc-workshop-big-data-tool-inclusion-exclusion (“tailoring prices more closely to an individual’s willingness to pay means that more transactions will take place than otherwise would have, with more consumers having access to the product and more consumers’ desires being satisfied. And price differentiation generally means that the poor pay less, not more, for the same goods.”).
[2] Dirk Auer, Justin (Gus) Hurwitz, Geoffrey A. Manne, Julian Morris & Kristian Stout, Privacy, Antitrust, and the Economic Approach to the Regulation of Consumer Data 5–7 (ICLE Comment 4, FTC Hearings on Competition & Consumer Protection in the 21st Century, Project No. P181201, 2018), https://laweconcenter.org/resources/icle-comments-the-regulation-of-consumer-data; Int’l Ctr. for Law & Econ., The Weakness of Interventionist Claims: Summary Comments on the FTC Hearings on Competition & Consumer Protection in the 21st Century 10, 139–40 (2019), https://laweconcenter.org/resources/icle-comments-ftc-hearings-on-competition-consumer-protection-in-the-21st-century-summary (“there are many cases where personalized pricing reduces deadweight loss and increases access to a product or service for those with a lower willingness-to-pay”).
[3] Geoffrey A. Manne, Daniel Gilman & Kristian Stout, Comments of the International Center for Law & Economics: FTC Advance Notice of Proposed Rulemaking, Trade Regulation Rule on Commercial Surveillance and Data Security (Docket No. FTC-2022-0053, R111004, Nov. 21, 2022), https://laweconcenter.org/wp-content/uploads/2022/11/ICLE-Commercial-Surveilance-ANPR-Comments-v4.pdf.
[4] Geoffrey A. Manne et al., Comments on Canadian Competition Bureau Discussion Paper on Algorithmic Pricing and Competition (Int’l Ctr. for Law & Econ., Aug. 4, 2025), https://laweconcenter.org/wp-content/uploads/2025/08/Canada-Algorithmic-pricing-and-competition-1.pdf.
[5] Ian Adams, Eric Fruits & Brian Albrecht, ICLE Comments to Competition Bureau Canada on Examining the Food Supply Chain 9–12, 27 (Int’l Ctr. for Law & Econ. 2026), https://laweconcenter.org/resources/icle-comments-to-competition-bureau-canada-on-examining-the-food-supply-chain.
[6] Brian C. Albrecht et al., Food-Retail Competition, Antitrust Law, and the Kroger/Albertsons Merger (Int’l Ctr. for Law & Econ., White Paper No. 2023-10-17, 2023), https://laweconcenter.org/wp-content/uploads/2023/10/Kroger-Albertsons-Merger-Full-Paper.pdf.
[7] Eric Fruits, The Waterbed Effect and the Price of ‘Pricing Fairness’ (Int’l Ctr. for Law & Econ., Issue Brief No. 2026-03-17, 2026), https://laweconcenter.org/wp-content/uploads/2026/03/Grocery-Pricing-Legislation-2026.pdf.
[8] Eric Fruits, Too Much on the Menu: Disclosure Overload and Crowding Out in Food-Delivery Regulation (Int’l Ctr. for Law & Econ., Issue Brief No. 2026-08-21, Aug. 21, 2026), https://laweconcenter.org/wp-content/uploads/2026/08/Delivery-Pricing-Literature-Review-v3-2026.pdf; Brian C. Albrecht et al., Comments of the International Center for Law & Economics: Rule on Unfair or Deceptive Fees in Online Food Delivery Services (FTC Project No. P267101, May 18, 2026) [hereinafter Online Food Delivery Comments], https://laweconcenter.org/wp-content/uploads/2026/05/FTC-Delivery-Fees-2026.pdf.
[9] See Daniel J. Gilman, Fine Print for Every Price: The FTC’s One-Size-Fits-All Guidance, Truth on the Mkt. (Aug. 28, 2026),https://truthonthemarket.com/2026/08/28/fine-print-for-every-price-the-ftcs-one-size-fits-all-guidance; Alden Abbott, Your Price May Vary: The FTC’s Personalized-Pricing Puzzle, Truth on the Mkt. (Sept. 2, 2026), https://truthonthemarket.com/2026/09/02/your-price-may-vary-the-ftcs-personalized-pricing-puzzle.
[10] Fed. Trade Comm’n, Proposed Enforcement Policy Statement Regarding Personalized Pricing, Docket No. FTC-2026-1057, Doc. No. FTC-2026-1057-0001 (Aug. 19, 2026), https://downloads.regulations.gov/FTC-2026-1057-0001/content.pdf [hereinafter Proposed Statement].
[11] Id. at 1.
[12] Id. at 3.
[13] See Geoffrey A. Manne, Bringing Antitrust’s Economic and Institutional Limits to the FTC’s Consumer Protection Authority 12–13 (Int’l Ctr. for Law & Econ. 2014), https://www.laweconcenter.org/resources/bringing-antitrusts-economic-institutional-limits-ftcs-consumer-protection-authority (explaining that Section 5(n)’s requirements “serve to impose an error cost approach on unfairness questions, limiting both the likelihood and harm of erroneous over-enforcement”).
[14] See Geoffrey A. Manne, Error Costs in Digital Markets, in The GAI Report on the Digital Economy 33, 48 (2020), https://laweconcenter.org/resources/error-costs-in-digital-markets (“Rebuttable presumptions are sometimes employed as a cost-saving substitute for direct evidence when economic theory predicts a relatively high probability of competitive harm.”).
[15] See Adams, Fruits & Albrecht, supra note 5, at 2 (“If enforcement condemns lawful discounting, loyalty pricing, or algorithm-driven responses to market conditions, firms may abandon those practices across their operations to avoid liability. Market forces cannot readily restore the lost benefits.”).
[16] See Geoffrey A. Manne, The Hydraulic Theory of Disclosure Regulation and Other Costs of Disclosure, 58 Ala. L. Rev. 473, 477, 485–86 (2007) [hereinafter Manne, Hydraulic Theory] (“Disclosure regulation makes some forms of behavior more expensive relative to others. Rational actors will respond by shifting some conduct into the comparatively cheaper outlets. And these alternative behaviors may actually be less beneficial than the regulated, deterred behavior.”).
[17] Proposed Statement, supra note 10, at 5.
[18] Int’l Harvester Co., 104 F.T.C. 949, 1059 (1984).
[19] 15 U.S.C. § 45(n) (2024); FTC v. Corpay, Inc., 164 F.4th 807, 840 (11th Cir. 2026) (“As ‘unless’ conveys, the three factors that follow it are mandatory for a finding that a practice is ‘unfair.’”).
[20] FTC Policy Statement on Unfairness, 104 F.T.C. 1070, 1073 (1984) (appended to Int’l Harvester Co., 104 F.T.C. 949 (1984)).
[21] Proposed Statement, supra note 10, at 4 & n.13.
[22] See infra Part II.C and the Appendix.
[23] Proposed Statement, supra note 10, at 6.
[24] Int’l Harvester, 104 F.T.C. at 1062.
[25] Trade Regulation Rule on Commercial Surveillance and Data Security, 87 Fed. Reg. 51,273, 51,293 (Aug. 22, 2022) (advance notice of proposed rulemaking) (dissenting statement of Commissioner Noah Joshua Phillips); id. at 51,298 (dissenting statement of Commissioner Christine S. Wilson).
[26] Proposed Statement, supra note 10, at 4.
[27] See Gilman, supra note 9.
[28] Manne, Gilman & Stout, supra note 3, at 15, 47 (urging the Commission to explain how its Section 5 authority applies to data practices and noting that “[s]uch guidance would, of necessity, account for the assessment of countervailing benefits too”); Berin Szóka & Geoffrey A. Manne, The Federal Trade Commission: Restoring Congressional Oversight of the Second National Legislature 67 (Int’l Ctr. for Law & Econ. 2016), https://laweconcenter.org/resources/wstate-manneg-20160524-sd004 (arguing that guidance should explain how the Commission will measure “concepts like materiality, weighing injury with benefits, and measuring reasonable avoidability”).
[29] Proposed Statement, supra note 10, at 8.
[30] See Brief of Amici Curiae International Center for Law & Economics and TechFreedom in Support of Petitioner at 26–27, LabMD, Inc. v. FTC, No. 16-16270 (11th Cir. 2017), https://laweconcenter.org/resources/amicus-brief-icle-techfreedom-labmd-inc-v-federal-trade (arguing that treating informal guidance as the source of “legally enforceable duties” creates a “de facto rulemaking[]” that bypasses Section 18’s safeguards).
[31] See James M. Lacko & Janis K. Pappalardo, Fed. Trade Comm’n Bureau of Econ., Improving Consumer Mortgage Disclosures: An Empirical Assessment of Current and Prototype Disclosure Forms ES-1 (2007), https://www.ftc.gov/sites/default/files/documents/reports/improving-consumer-mortgage-disclosures-empirical-assessment-current-and-prototype-disclosure-forms/p025505mortgagedisclosurereport.pdf.
[32] Proposed Statement, supra note 10, at 1; see also id. at 3–4.
[33] Id. at 3 (emphasis added).
[34] Id. at 4.
[35] See Lars A. Stole, Price Discrimination and Competition, in 3 Handbook of Industrial Organization 2221, 2239–43 (Mark Armstrong & Robert Porter eds., 2007) (reviewing the models of Thisse & Vives and of Corts).
[36] Id.
[37] James C. Cooper, Luke Froeb, Daniel P. O’Brien & Steven Tschantz, Does Price Discrimination Intensify Competition? Implications for Antitrust, 72 Antitrust L.J. 327, 342 (2005).
[38] Andrew N. Ferguson, Comm’r, Fed. Trade Comm’n, Concurring Statement, Surveillance Pricing Intermediaries 1 (Matter No. P246202, July 23, 2024), https://www.ftc.gov/system/files/ftc_gov/pdf/surveillance-pricing-6b-ferguson-concurrence.pdf.
[39] Proposed Statement, supra note 10, at 3. In January 2025, Commission staff described its summary of the Section 6(b) responses as preliminary and disclaimed any assessment of illegal conduct. Fed. Trade Comm’n, FTC Surveillance Pricing 6(b) Study: Research Summaries—A Staff Perspective 2 (Jan. 2025), https://www.ftc.gov/system/files/ftc_gov/pdf/p246202_surveillancepricing6bstudy_researchsummaries_redacted.pdf [hereinafter Research Summaries] (“The format of this work aims to highlight initial observations . . . . This document does not reflect or suggest any assessment as to whether anyone has engaged in illegal conduct.”).
[40] Competition & Mkts. Auth., Pricing Algorithms: Economic Working Paper on the Use of Algorithms to Facilitate Collusion and Personalised Pricing ¶ 7.14, at 38 (CMA94, 2018) (describing the European Commission’s 2018 mystery shopping study of 160 e-commerce websites in eight member states: “Price differences were only observed in 6% of tests, and the median price difference observed was less than 1.6%.”); see also Competition & Mkts. Auth., The Commercial Use of Consumer Data ¶¶ 3.69–3.70, at 92 (CMA38, 2015) (“we did not receive a clear indication of examples where price discrimination using consumer data was being used to the likely detriment of consumers”; “Firms may be wary of damaging their reputation or brand value by being seen to do this.”).
[41] Alberto Cavallo, Are Online and Offline Prices Similar? Evidence from Large Multi-Channel Retailers, 107 Am. Econ. Rev. 283, 295 (2017).
[42] Stefano DellaVigna & Matthew Gentzkow, Uniform Pricing in U.S. Retail Chains, 134 Q.J. Econ. 2011, 2011 (2019).
[43] Proposed Statement, supra note 10, at 3–4 nn.8, 11–13.
[44] The paper the Commission relies on most, citing it for each of its economic claims, models the seller as a monopolist and states that it “do[es] not consider the impact of personalized pricing in a competitive market.” Jean-Pierre Dubé & Sanjog Misra, Personalized Pricing and Consumer Welfare, 131 J. Pol. Econ. 131, 180 (2023).
[45] See Stole, supra note 35, at 2231–32 & n.19.
[46] Benjamin Reed Shiller, Approximating Purchase Propensities and Reservation Prices from Broad Consumer Tracking, 61 Int’l Econ. Rev. 847, 861 (2020).
[47] Id. at 862 tbl.4.
[48] Id. at 867.
[49] Dubé & Misra, supra note 44, at 157–58.
[50] Id. at 134–35.
[51] Nicholas Buchholz et al., Personalized Pricing and the Value of Time: Evidence from Auctioned Cab Rides, 93 Econometrica 929, 930, 934 (2025).
[52] Id. at 952–53 & tbl.VII.
[53] Dubé & Misra, supra note 44, at 159 (“we also do not expect any competitive response from other platforms”).
[54] Buchholz et al., supra note 51, at 934.
[55] Andrew Rhodes & Jidong Zhou, Personalized Pricing and Competition, 114 Am. Econ. Rev. 2141, 2152 (2024) (stating in Proposition 1 that, where the market is fully covered, “relative to uniform pricing, personalized pricing harms firms and benefits consumers in aggregate”).
[56] Buchholz et al., supra note 51, at 954.
[57] Dubé & Misra, supra note 44, at 167.
[58] Shiller, supra note 46, at 862.
[59] Rhodes & Zhou, supra note 55, at 2152–53, 2162–63; see also Proposed Statement, supra note 10, at 4 n.13.
[60] Rhodes & Zhou, supra note 55, at 2157 & n.23 (thresholds); id. at 2163, 2165 (remedies).
[61] Buchholz et al., supra note 51, at 952.
[62] Id. at 952–53 & tbl.VII.
[63] Dubé & Misra, supra note 44, at 157, 165–67 & tbl.8. Every personalized price exceeds the uninformative $99 status quo, ranging from $126 to $6,292. A comparison with $99 thus shows every buyer paying more. The authors use the optimal uniform price “as our base case, not $99, since the former reflects the textbook inverse-elasticity-rule pricing that would be predicted for a maturing company.” Id. at 158, 165.
[64] Rhodes & Zhou, supra note 55, at 2144 n.7, 2155 (“With partial market coverage, personalized pricing increases welfare by expanding demand, but it boosts profit so much that in aggregate consumers suffer from it.”).
[65] Against a uniform price, consumer surplus falls 0.50% in Shiller and 2.25% in Buchholz et al. Shiller, supra note 46, at 861 tbl.3; Buchholz et al., supra note 51, at 953 tbl.VII. Dubé and Misra estimate a 25% decline against the optimal uniform price, even though most potential customers would receive lower price offers. Dubé & Misra, supra note 44, at 166–67. In Buchholz et al., adding price sensitivity to personalization based on wait sensitivity alone also lowers consumer surplus, but only slightly. Buchholz et al., supra note 51, at 953 tbl.VII.
[66] Buchholz et al., supra note 51, at 953 tbl.VII, 954–55 & fig.5 (the inside-option share rises 7.28% relative to uniform pricing; 62.5% of consumers gain; the most price- and wait-sensitive riders gain most); Dubé & Misra, supra note 44, at 167, 171–72 (63% of personalized prices fall below the optimal uniform price; smaller firms with part-time staff are most likely to benefit); Shiller, supra note 46, at 861 tbl.3, 862 (subscribers rise 3.12%; “most consumers receive lower prices when prices are personalized, and hence the majority are slightly better off”); see also Rhodes & Zhou, supra note 55, at 2153, 2155 (under partial coverage, personalized pricing “expands total demand and strictly increases total welfare”).
[67] FTC Policy Statement on Unfairness, supra note 20, at 1073.
[68] See Buchholz et al., supra note 51, at 954–55; Dubé & Misra, supra note 44, at 172; Shiller, supra note 46, at 862, 867.
[69] 15 U.S.C. § 45(n) (2024).
[70] Shiller, supra note 46, at 861 tbl.3; Buchholz et al., supra note 51, at 953 tbl.VII; Dubé & Misra, supra note 44, at 166–67 & tbl.8. Dubé and Misra compare personalized prices capped at $499 with the optimal uniform price of $327. Uncapped personalization reduces total consumer surplus by 25% against that benchmark. Capped personalization reduces it by 11.45% against the $249 uniform price ZipRecruiter later adopted. Id. at 166 tbl.8, 167.
[71] Dubé & Misra, supra note 44, at 167, 179.
[72] The Unfairness Policy Statement lists “reduced incentives to innovation and capital formation” among the costs of a remedy that must be considered. FTC Policy Statement on Unfairness, supra note 20, at 1073–74.
[73] See, e.g., Ohio v. American Express Co., 585 U.S. 529 (2018).
[74] See Stole, supra note 35, at 2231–32 & n.19.
[75] Buchholz et al., supra note 51, at 953 tbl.VII; Shiller, supra note 46, at 861 tbl.3. Buchholz et al. note that increased platform use need not mean an equal increase in total travel because their model’s outside option includes taking “another form of transportation.” Buchholz et al., supra note 51, at 939. The relevant question is how pricing affects users’ decisions to book trips on the platform; the distance traveled matters less to that question.
[76] Proposed Statement, supra note 10, at 4 & n.13.
[77] Id.
[78] Dubé & Misra, supra note 44, at 142–43.
[79] Id. at 131 (emphasis added).
[80] Id. at 171–72.
[81] Id. at 180.
[82] Id. at 136, 172.
[83] Proposed Statement, supra note 10, at 4 n.13.
[84] Rhodes & Zhou, supra note 55, at 2163 (“[A]symmetry in information and timing allows the firms to better segment the market, increasing industry profit. Consequently, consumer surplus ends up being lowest in the mixed regime.”).
[85] Id. at 2142 (“We focus on the limit case of personalized pricing, i.e., first-degree price discrimination, where firms can perfectly infer individual consumers’ preferences”); id. at 2162 n.30 (“it would also be interesting to consider other information structures and investigate how the impact of price discrimination changes with the amount of information that firms have access to”).
[86] Buchholz et al., supra note 51, at 953 tbl.VII.
[87] Id. at 952–53 & tbl.VII.
[88] Id. at 953.
[89] Id. at 932.
[90] Id. at 953 tbl.VII, 954–55 & fig.5.
[91] Ginger Zhe Jin, Liad Wagman & Mengyi Zhong, Surveillance Pricing: A Cautionary Summary of Potential Harms and Solutions, CPI Antitrust Chron., July 2025, at 4.
[92] Id. (emphasis added).
[93] Id.
[94] Id. at 5.
[95] Id. (citing Rhodes & Zhou, supra note 55).
[96] Id. at 9 (emphasis added).
[97] Id.
[98] Id.
[99] Rosa-Branca Esteves & Francisco Carballo-Cruz, Access to Data for Personalized Pricing: Can It Raise Entry Barriers and Abuse of Dominance Concerns? 10, 15, 18 (NIPE Working Paper No. 05/2021, Mar. 2021). The Proposed Statement misdates the paper by four years. See Proposed Statement, supra note 10, at 4 n.13 (citing the paper as “NIPE Working Papers, NIPE, Universidade do Minho (May 2025)”).
[100] Esteves & Carballo-Cruz, supra note 99, at 14 tbl.1.
[101] Id. at 15 (“conditional on entry, consumer surplus is in fact maximized when both firms are able to use data for personalized prices”).
[102] Id. at 1.
[103] Id.
[104] Mark Armstrong & John Vickers, Discriminating Against Captive Customers, 1 Am. Econ. Rev.: Insights 257, 257–58, 263 (2019).
[105] Brian C. Albrecht, Price Competition and the Use of Consumer Data 1–2 (working paper, June 12, 2026), https://briancalbrecht.com/albrecht_price_competition_consumer_data.pdf.
[106] Dirk Bergemann, Benjamin Brooks & Stephen Morris, The Limits of Price Discrimination, 105 Am. Econ. Rev. 921, 921–23 (2015).
[107] See Rhodes & Zhou, supra note 55, at 2164; Esteves & Carballo-Cruz, supra note 99, at 1, 17–18.
[108] Proposed Statement, supra note 10, at 5.
[109] Id.
[110] Proposed Statement, supra note 10, at 3; see also id. at 1 (personalized pricing is “a long-established norm in some markets”).
[111] Id. at 1.
[112] Gilman, supra note 9 (“How many consumers must hold that expectation? How consistently? Must a clear majority share it?”); see also Albrecht et al., Online Food Delivery Comments, supra note 8, at 7 (“A first-time user’s expectations differ materially from those of a customer placing a hundredth order.”).
[113] Council of Econ. Advisers, Big Data and Differential Pricing 17 (2015).
[114] FTC Policy Statement on Deception, 103 F.T.C. 174, 176 & n.8 (1984) (appended to Cliffdale Assocs., Inc., 103 F.T.C. 110 (1984)).
[115] FTC Policy Statement on Deception, supra note 114, at 182; Int’l Harvester, 104 F.T.C. at 1057; see also Geoffrey A. Manne, R. Ben Sperry & Berin Szoka, In the Matter of Nomi Technologies, Inc.: The Dark Side of the FTC’s Latest Feel-Good Case 5, 10–11 (ICLE White Paper 2015-1, 2015), https://laweconcenter.org/resources/nomi-white-paper (objecting to the Commission’s treatment of the absence of an in-store notice as an implied representation and observing that “for an implied promise to be deemed material (and thus deceptive) under the DPS, the FTC must adduce some proof: either that it was, in fact, intended, or that it was, in fact, material”).
[116] FTC Policy Statement on Deception, supra note 114, at 175–76.
[117] Int’l Harvester, 104 F.T.C. at 1057, 1059.
[118] Id. at 1059–60.
[119] Id. at 1063–64, 1070.
[120] Id. at 1058 & n.29; see also id. at 1062 (limiting mandatory disclosure “to those core aspects of a transaction that virtually all consumers would consider essential to an informed decision”).
[121] See Int’l Harvester, 104 F.T.C. at 1060 (“Since the seller will have no way of knowing in advance which disclosure is important to any particular consumer, he will have to make complete disclosures to all.”).
[122] Proposed Statement, supra note 10, at 6 n.24 (quoting P. Lorillard Co. v. FTC, 186 F.2d 52, 58 (4th Cir. 1950)) (“to tell less than the whole truth is a well-known method of deception”).
[123] P. Lorillard Co. v. FTC, 186 F.2d 52, 58 (4th Cir. 1950) (“The fault with this advertising was not that it did not print all that the Reader’s Digest article said, but that it printed a small part thereof in such a way as to create an entirely false and misleading impression, not only as to what was said in the article, but also as to the quality of the company’s cigarettes.”). The Commission itself cites Lorillard as a half-truth case, not a silence case. Int’l Harvester Co., 104 F.T.C. at 1057 & n.25.
[124] FTC v. Standard Educ. Soc’y, 302 U.S. 112, 114–16 (1937) (salespeople represented that a set of books was free, that the prospect had been specially selected, and that the regular price exceeded the offered price; the Commission found each representation false).
[125] Proposed Statement, supra note 10, at 5.
[126] Proposed Statement, supra note 10, at 7.
[127] FTC v. Colgate-Palmolive Co., 380 U.S. 374, 387 (1965).
[128] Proposed Statement, supra note 10, at 5 n.19 (quoting Colgate-Palmolive, 380 U.S. at 387).
[129] 15 U.S.C. § 45(n) (2024) (emphasis added).
[130] Corpay, 164 F.4th at 840 (“As ‘unless’ conveys, the three factors that follow it are mandatory for a finding that a practice is ‘unfair.’”); see also Proposed Statement, supra note 10, at 5 n.17 (citing Corpay, 164 F.4th at 839).
[131] Proposed Statement, supra note 10, at 7–8.
[132] 15 U.S.C. § 45(n) (2024).
[133] Proposed Statement, supra note 10, at 3.
[134] Id. at 6.
[135] FTC Policy Statement on Unfairness, supra note 20, at 1073.
[136] Id. at 1073.
[137] See Proposed Statement, supra note 10, at 6 n.24 (citing Priester et al. and Xia et al. on consumers’ perceptions of fairness); Geoffrey A. Manne & Kristian Stout, Comments of the International Center for Law & Economics, Informational Injury Workshop 40–41 (FTC Project No. P175413, 2017), https://laweconcenter.org/resources/comments-international-center-law-economics-matter-informational-injury-workshop-ftc (quoting the legislative history of Section 5(n): “Emotional impact and more subjective types of harm alone are not intended to make an injury unfair.”).
[138] Proposed Statement, supra note 10, at 6.
[139] Id. at 6.
[140] Id. at 6 n.20.
[141] Corpay, 164 F.4th at 825–27, 838–42.
[142] Proposed Statement, supra note 10, at 6.
[143] Proposed Statement, supra note 10, at 6 n.22.
[144] 15 U.S.C. § 1681m(a)(1)–(4) (2024); see id. § 1681g(f)(1)(C) (key factors). The quoted statement appears in § 1681m(a)(3)(B).
[145] 15 U.S.C. § 1691(d)(2)–(3) (2024).
[146] 15 U.S.C. § 1691(d)(2)(B) (2024).
[147] 15 U.S.C. § 1691(d)(6) (2024).
[148] 15 U.S.C. §§ 1681m(h)(1), (h)(5), 1681g(f)(1)(B)–(E) (2024).
[149] 15 U.S.C. § 1681m(b)(1) (2024).
[150] Proposed Statement, supra note 10, at 6 n.23 (citing Cal. Ins. Code §§ 791.02, 791.10; Va. Code Ann. § 38.2-602; Wash. Admin. Code ch. 284-30A).
[151] 15 U.S.C. § 45(n) (2024).
[152] Proposed Statement, supra note 10, at 6 n.21.
[153] FTC Policy Statement on Unfairness, supra note 20, at 1075–76.
[154] Proposed Statement, supra note 10, at 6.
[155] Proposed Statement, supra note 10, at 7.
[156] Adams, Fruits & Albrecht, supra note 5, at 27 (“retailers should clearly disclose final prices and truthfully present discounts and reference prices”); Albrecht et al., Online Food Delivery Comments, supra note 8, at 7–8, 12.
[157] Int’l Harvester, 104 F.T.C. at 1062 (adding that “virtually any piece of information may be useful to some consumers,” and that a disclosure duty grounded in unfairness requires “a full cost-benefit analysis, in which we weigh the consumer benefits of disclosure against their likely costs”).
[158] Exec. Office of the President, Big Data: Seizing Opportunities, Preserving Values 65 (2014).
[159] Council of Econ. Advisers, supra note 113, at 17.
[160] Petra Persson, Attention Manipulation and Information Overload, 2 Behav. Pub. Pol’y 78, 96–97 (2018).
[161] Frank H. Easterbrook & Daniel R. Fischel, Mandatory Disclosure and the Protection of Investors, 70 Va. L. Rev. 669, 709 (1984).
[162] Persson, supra note 160, at 101.
[163] Id. at 97. Both results come from a formal model, illustrated with examples rather than tested against market data. They show that the seller responses Persson describes are available to any firm subject to a disclosure mandate, but they do not show how often firms use them.
[164] Manne, Hydraulic Theory, supra note 16, at 480–81 (“As this Article suggests, however, mandatory disclosure has important and ill-considered consequences; it is not merely regulatory ‘chicken soup.’”).
[165] Id. at 506.
[166] Id. at 506–07 (“It is precisely individual shareholders who are not in a position to evaluate complex information regarding, for example, executive pay packages. But it is also precisely these individuals at whom mandatory disclosures are directed.”).
[167] Proposed Statement, supra note 10, at 7.
[168] Manne, Hydraulic Theory, supra note 16, at 510–11.
[169] Like other agencies before it. See id. at 497 (“Unfortunately, the SEC does not appear to consider the relevant, marginal question. . . . The broad-brushed nature of federal regulation effectively precludes a firm-by-firm marginal analysis.”); see also id. at 491 (“Regulators are limited in their ability to predict the effects of their regulations . . . . Even well-intended regulation is bound to have unintended consequences . . . .”).
[170] OECD, Enhancing Online Disclosure Effectiveness 14, 31 (OECD Digital Econ. Papers No. 335, 2022) (quoting Omri Ben-Shahar & Carl E. Schneider, The Failure of Mandated Disclosure, 159 U. Pa. L. Rev. 647, 686 (2011)).
[171] See Trade Regulation Rule on Unfair or Deceptive Fees, 90 Fed. Reg. 2066, 2066 (Jan. 10, 2025) (codified at 16 C.F.R. pt. 464).
[172] See Rule on Unfair or Deceptive Rental Housing Fee Practices, 91 Fed. Reg. 12,325 (Mar. 13, 2026) (advance notice of proposed rulemaking).
[173] See Proposed Statement, supra note 10, at 1 n.1.
[174] Lisa Robinson, Kip Viscusi & Richard Zeckhauser, Efficient Warnings, Not “Wolf or Puppy” Warnings, in The Future of Risk Management 227, 241 (Howard Kunreuther et al. eds., 2019).
[175] OECD, Enhancing Online Disclosure Effectiveness, supra note 170, at 6.
[176] Howard Beales & Todd J. Zywicki, Junkyard Dogs: The Law and Economics of “Junk” Fees, CPI Antitrust Chron., Apr. 2023, at 6, https://laweconcenter.org/resources/junkyard-dogs-the-law-and-economics-of-junk-fees/.
[177] Anna Priester et al., A Special Price Just for You: Effects of Personalized Dynamic Pricing on Consumer Fairness Perceptions, 19 J. Revenue & Pricing Mgmt. 99, 106–08 (2020).
[178] Oren Bar-Gill, Cass R. Sunstein & Inbal Talgam-Cohen, Algorithmic Harm in Consumer Markets, 15 J. Legal Analysis 1, 10, 25 (2023).
[179] OECD, The Effects of Online Disclosure About Personalised Pricing on Consumers 5, 9–13 (OECD Digital Econ. Papers No. 303, 2021).
[180] Id. at 5.
[181] Id. at 27.
[182] Id. at 27.
[183] Id. at 26.
[184] The OECD’s later report emphasizes the prominence and repetition of the tested notices: the disclosures were “succinct, substantial in size, seen repeatedly, and accompanied every single personalised price. OECD, Enhancing Online Disclosure Effectiveness, supra note 170, at 14.
[185] Id.
[186] Willem H. van Boom et al., Consumers Beware: Online Personalized Pricing in Action! How the Framing of a Mandated Discriminatory Pricing Disclosure Influences Intention to Purchase, 33 Soc. Just. Rsch. 331 (2020).
[187] Id. at 335–37.
[188] Id. at 343–45.
[189] Allison Nicoletti & Christina Zhu, Economic Consequences of Transparency Regulation: Evidence from Bank Mortgage Lending, 61 J. Acct. Rsch. 1827, 1846–51 (2023).
[190] Id. at 1832.
[191] Natasha Clarke et al., Calorie (Energy) Labelling for Changing Selection and Consumption of Food or Alcohol, 2025 Cochrane Database of Systematic Reviews, Art. CD014845, at 2, https://doi.org/10.1002/14651858.CD014845.pub2.
[192] Pasquale E. Rummo et al., Menu Labeling and Calories Purchased in Restaurants in a US National Fast Food Chain, 6 JAMA Network Open e2346851, tbl. 3 (2023).
[193] See Lacko & Pappalardo, supra note 31, at ES-1.
[194] Proposed Statement, supra note 10, at 6.
[195] Proposed Statement, supra note 10, at 6.
[196] Manne, Hydraulic Theory, supra note 16, at 476. The article discusses securities regulation, but its account of how regulated actors respond to disclosure mandates applies beyond that setting.
[197] Id. at 477.
[198] Id. at 485 (“[W]here behavior can be altered at a lower cost than the cost of disclosure, disclosure rules will induce behavioral changes rather than increased information flow.”); id. at 486 (“The magnitude of the hydraulic effect will turn primarily on the degree of cross-elasticity (substitutability) between the impeded (more costly) and the freer (less costly) sources of behavior.”).
[199] Proposed Statement, supra note 10, at 6 n.20 (declining “to take any position on whether some personalized pricing practices are unfair even when fully disclosed to consumers”).
[200] Buchholz et al., supra note 51, at 932, 952 (describing ETA-based pricing as “another non-personalized pricing policy” that platforms may adopt “due to consumer backlash,” under which “[w]ait-time pricing may then give consumers incentives to reveal their high wait-time sensitivity by picking a driver with shorter wait time and higher price”).
[201] Manne, Hydraulic Theory, supra note 16, at 477; see also id. at 487 (observing that, under seat-belt laws, a shift in behavior “does not impose the same risk on, say, pedestrians, who now suffer the ill effects of more reckless driving but earn none of the benefit of ‘safer’ accidents”).
[202] Id. at 477–78.
[203] Stole, supra note 35, at 2226 (conditions for discrimination). The price-gap example in the text is a conditional incentive argument, not an estimate of the proposed notice’s effect.
[204] Xi Li & Krista J. Li, Beating the Algorithm: Consumer Manipulation, Personalized Pricing, and Big Data Management, 25 Mfg. & Serv. Operations Mgmt. 36, 36 (2023).
[205] S. Nageeb Ali, Greg Lewis & Shoshana Vasserman, Voluntary Disclosure and Personalized Pricing, 90 Rev. Econ. Stud. 538, 558–60 (2023).
[206] Dubé & Misra, supra note 44, at 167, 171–72.
[207] DellaVigna & Gentzkow, supra note 42, at 2014, 2033.
[208] Id. at 2055–56 & fig.XI. See also Manne, Hydraulic Theory, supra note 16.
[209] See Fruits, Waterbed Effect, supra note 7, at 14–16 (the Robinson-Patman Act’s effect on discounting, the Antitrust Modernization Commission’s 2007 conclusion, and the pay-transparency evidence of Cullen and Pakzad-Hurson); Geoffrey A. Manne, Written Testimony, Reviving Competition, Part 5: Addressing the Effects of Economic Concentration on America’s Food Supply 19–20 (Int’l Ctr. for Law & Econ. 2022), https://laweconcenter.org/resources/testimony-of-geoffrey-a-manne-reviving-competition-part-5-addressing-the-effects-of-economic-concentration-on-americas-food-supply/ (“The RPA effectively imposes wholesale price uniformity on sellers which deters innovation (or, as noted, price reductions) and new entry.”).
[210] Complaint ¶¶ 7, 21–22, 262, FTC v. Amazon.com, Inc., No. 2:23-cv-01495 (W.D. Wash. Sept. 26, 2023).
[211] See, e.g., Zoë B. Cullen & Bobak Pakzad-Hurson, Equilibrium Effects of Pay Transparency, 91 Econometrica 765, 765 (2023) (finding that state laws protecting workers’ right to inquire about pay were followed by wage declines of about 2%, because employers “credibly refuse to pay high wages to any one worker to avoid costly renegotiations with others”).
[212] Fed. Trade Comm’n, Research Summaries, supra note 39, at 1–2, 10.
[213] See Abbott, supra note 9.
[214] Corpay, 164 F.4th at 838–42 (unfairness count proved through a customer survey, eight internal surveys, internal emails, complaint data, and evidence of what customers were told about the fees); see also id. at 825–27 (record evidence on what customers were told).
[215] Proposed Statement, supra note 10, at 3–4 nn.8, 11–13.
Source: Federal Trade Commission
I appreciate the opportunity to comment on the FTC’s Proposed Enforcement Policy Statement Regarding Personalized Pricing (“Proposed Policy”).[1] This comment offers my own views as . . .
I appreciate the opportunity to comment on the FTC’s Proposed Enforcement Policy Statement Regarding Personalized Pricing (“Proposed Policy”).[1] This comment offers my own views as a long-time student of and practitioner in the FTC’s consumer protection mission.[2]
The comment begins with a brief overview of personalized pricing and the issues it raises as a form of price discrimination in Section I. Section II then argues that when there are competing sellers, consumers can easily avoid any injury from personalized pricing by the simple expedient of searching for another price. Section III argues that the Commission’s premise that consumers generally expect one price for all customers is unsubstantiated. Consumers are well aware that different consumers pay different prices in ordinary purchase situations. Section IV considers discounts based on personal information. It argues that such practices benefit consumers, and that the Commission should create a safe harbor to avoid requiring disclosures for personalized discounts. Section V argues that disclosure is an inappropriate remedy for the problems the Commission seeks to address and should, in any event, be tested before it is required.
Some critics of the information economy have become concerned about personalized pricing, in which businesses use consumer-specific information to set different prices for different consumers. By definition, charging different prices for different consumers (assuming costs are the same) is price discrimination. The economics literature has long been clear that price discrimination can be either welfare increasing or welfare reducing, depending on the circumstances.[3] Whether a given instance of price discrimination is good or bad for consumers is thus an empirical question. Nevertheless, many scholars have argued that common forms of retail price discrimination lead to increased output and enhanced consumer welfare.[4]
As the Proposed Policy notes, “the extent to which businesses currently use personalized pricing is not well understood, and the effects of personalized pricing on consumers are unclear.”[5] Instead, the discussion is driven by concern about a wide variety of hypothetical business practices which are not known to have actually occurred, and which may never occur.
The hypothetical scenario that has attracted the most attention is the possibility that a monopolist could charge each individual consumer the maximum amount that consumer would be willing to pay, thus extracting all consumer surplus. Although a theoretically interesting polar case, such personalized pricing is not possible in practice: Statistical uncertainty remains. Even the best models cannot predict with certainty, and data, even in the physical sciences and especially in economics, are subject to measurement errors. Dubé and Misra show by example that, even in a monopoly model that incorporates statistical uncertainty, a database that allows different prices for different consumers may increase both profits and consumer surplus.[6] In their careful experimental study, personalized pricing never extracted more than 30% of consumer surplus that would result from uniform pricing, whereas “true perfect price discrimination … would extract all consumer surplus.”[7] Moreover, in their experiment, more than 60% of consumers received a lower price with personalized pricing. Why such a pricing outcome might warrant consumer protection enforcement is hardly obvious.
More important than the limitations of personalized pricing in the monopoly context is the fact that most industries are not monopolies. When there are competitors, the same tools and data that allow a monopolist to charge more to consumers who are more willing to pay also enable competitors to identify the consumers who are most overcharged and compete for their business through lower prices. Firms can also offer discounts to consumers who prefer their rival’s product, again pressuring the rival to lower price.[8] These uses of information-based pricing offer clear consumer benefits.
Empirical questions that underlie a public policy deserve empirical answers rather than adopting a policy that assumes an answer. The impact of price discrimination on consumer welfare is just such a question, as economists have recognized for decades. Empirical answers about the effects of personalized pricing as a method of price discrimination are, at present, lacking. Whether a particular implementation of personalized pricing benefits consumers or harms them should be central to the Commission’s enforcement decisions. Under the Proposed Policy, the Commission does not appear to ask that key factual question in deciding to bring a case.
The specific “personalized pricing” scenarios the Proposed Policy identifies as requiring disclosure highlight the important role of competition as a constraint. Marketing delivery services to those who are unable to leave home, milk to people with several children, hotels to people who need to travel, or home security to crime victims are examples that appear to be competitive markets. Information identifying these consumers makes competition, including offering a lower price, more likely, not less. Similarly, in the Commission’s example of a consumer standing in the store who is shown a higher price online, the consumer can buy the item off the shelf. Moreover, a competitor using the same information to offer online discounts to consumers who are in its rival’s stores would seem to produce clear consumer benefits. As discussed in the next section, consumers can easily avoid injury in such circumstances.
The most troubling of the Commission’s scenarios is the rideshare company charging a premium price for a consumer in dire need of transportation, particularly if the consumer has no reasonable alternatives. Here, however, the problem is more akin to duress, rather than a failure to disclose personalized pricing. The common law of contract has long policed closely related situations. For example, a seller may refuse to continue performance under a contract without a price higher than previously agreed. In another example, a seller who finds a prospective buyer in a sensitive position with few if any alternatives, raises the price to extraordinary levels. In either circumstance, common law courts scrutinize the contract with extra care and may refuse to enforce it.[9] Most importantly, although this exploitation may constitute a Section 5 violation,[10] the problem is not the failure to disclose. No amount of disclosure about how the price was set will either generate new alternatives or relieve the consumer of the dilemma.
Consumers cannot buy anything without first locating a seller. Although some may purchase from the first seller they find, consumers frequently gather information about the offerings of other sellers to choose the best option. That process of consumer search means that consumers do not come to a particular offering in ignorance. They have some information about prices at other sellers, or perhaps this seller at an earlier time, that provides context for the offered price. Even when a new seller or new product is first encountered, consumers can easily generate a context by searching for other offerings. In the online context that seems to be a primary concern of the proposed personalized pricing policy, the simple expedient of clicking the next link will generate additional information about whether the price is high or low.
Rather than focusing on the injury, the draft statement’s analysis of materiality and avoidability focuses on the process that allegedly generates the injury.[11] It is true that if consumers are unaware of personalized pricing they would not know to take steps like using a VPN or a private browser.[12] But the injury is not the process of setting the price, it is the high price that may result. Consumers can avoid the actual injury with the same simple process they routinely employ to avoid any price they think is high: check another seller. Consumers can therefore reasonably avoid any injury that might result from personalized pricing.
Of course, consumers may decide to purchase without additional search. That decision, however, is based on the perceived benefits and costs of further information gathering. The manner in which the seller determined the price does not affect the search decision. It is likely that consumers who choose not to search will pay a higher price on average, but not searching may nonetheless be a rational choice. In addition to avoiding the costs of search, some consumers prefer familiarity with one seller and will pay for that benefit. Choosing not to search also captures the Commission’s scenario of a rideshare company charging more for a consumer who has only one installed rideshare app. Moreover, installing another app would seem an easy way for a consumer to avoid injury. In any event, disclosure at the point of sale will not convert a non-searcher into a searcher and does nothing to reduce the costs of search.
The presence or absence of personalized pricing has no effect on search behavior in two of the most studied strategies in the economics literature. In a fixed sample size strategy, consumers choose a number of sellers to sample based on the costs of checking another price and the expected benefit of finding a lower price.[13] Personalized pricing does not affect either side of this equation. Another much-studied search strategy is an optimal stopping rule,[14] in which consumers continue to search until they find a price below some specified value. Again, nothing about the presence or absence of personalized pricing would affect this determination or the transaction price that results.
The information that consumers need to make optimal choices is information about the price itself, not how it was determined. As Nobel Laureate Friedrich Hayek observed in his seminal paper The Use of Knowledge in Society, the beauty of the price system is that the price aggregates and summarizes all information about value and cost into a single statistic, one that provides the basis for decisions without exploration of the underlying details.[15] Whatever consumers may expect about how it was determined, that summary remains the same for the particular consumer and the particular proposed transaction. In most circumstances consumers can remain blissfully ignorant of the process by which prices are set and the range of information used to set them.
Charging different prices to different consumers is as old as retailing. Consumers who are more willing to pay for a product or service, or less sensitive to price, are often charged higher prices. Consumer characteristics may serve as a proxy, as is the case with discounts for students or for the elderly. But many traditional price discrimination techniques rely on consumers who care more about price to self-identify, usually by their willingness to jump over some “hurdle” to qualify for a lower price. Rebate offers or discount coupons are common examples: Consumers who are more price sensitive are more willing to clip coupons or jump the hurdles necessary to claim the rebate. As one economics textbook noted, “[i]t is a rare product whose seller does not use the hurdle model of differential pricing.”[16]
Indeed, nothing is more common in the experience of every consumer than the simple fact that prices vary. Different sellers offer the same goods at different prices, in different locations, at different times, and to different consumers. Similar goods that are close substitutes are available at a similar array of prices. The only way for any consumer to know whether a particular price is a good one is to compare it to other prices—to search for a price they are willing to pay.
Even in circumstances most favorable to the Commission’s assertion that consumers have “reasonable expectations” of one price for all, it is highly likely that consumers understand that the transaction price will differ for different consumers. Consider a posted price for an item in a grocery store, arguably an express representation of one price for all. Consumers understand that at checkout some will pay the posted price, but others will pay less because they use a coupon on the purchase (and perhaps different coupons offering different discounts). Still others may receive a different loyalty program discount or take advantage of a volume discount. The range of prices expands even further when we consider competing brands, including store brands that may be functionally identical. Each of these arrangements means that different consumers pay different prices, based on their personal choices and behavior. Such pricing arrangements are ubiquitous in retailing, obvious to consumers, and make clear that the posted price is not “one price for all.”
The Proposed Policy cites evidence that consumers do not understand the nature and extent of data collection and use.[17] It does not follow that they therefore expect “one price for all,” despite their everyday experience to the contrary. Moreover, the Commission recognizes that consumers do expect individualized pricing in some circumstances, but it offers no principled distinction between those instances and others where consumers allegedly expect a single price, and no way for a company to assess what consumers “reasonably expect” in any particular instance.
In any event, consumer expectations about prices and how they are set are a foundation of quicksand for enforcement that actually benefits consumers. The Commission has a sad history of using claims of “deception” to attack discounters and other low-price sellers,[18] spurred on by competitors seeking to suppress competition that benefits consumers. Presumably most consumers who choose to buy “expect” that the price they are paying is a good one. By the logic of the Proposed Policy, however, the failure to disclose that the price is in fact higher than most other offerings in the market would seem to be a deceptive practice. That approach, however, would make the Commission a price regulator, something it clearly is not under its statute.
Although the possible downsides of personalized pricing are at this point largely hypothetical, concrete benefits of information-based pricing abound. Using the same data and techniques as personalized pricing, loyalty programs and discount email marketing are widespread. Industry estimates indicate that millions of consumers are members of loyalty programs, often multiple programs, that offer discounts and other promotional benefits based on the consumer’s history with the seller. Capital One Shopping reports that discounts are the strongest drivers of loyalty program participation.[19] Zoho, a provider of online platforms for loyalty programs, reports that 80% of Americans are members of at least one loyalty program.[20] EMARKETER reports that the average consumer is a member of a surprisingly high 17 loyalty programs,[21] and notes that the best programs tailor offers and content based on the individual member’s preferences and behavior. Email marketing also makes extensive use of consumer information. Capital One reports that 77% of consumers want to receive emails offering discounts, and that personalized promotional emails with discounts are opened 29% more often and clicked 41% more often than those without.[22]
These programs offer clear consumer benefits, reducing the costs of living for American families. They use the same kinds of data, and the same or similar analytical techniques, to determine how large a discount (if any) to offer any individual consumer. Unless the Commission creates a safe harbor for such discounts, the analysis of the proposed policy statement would seem to require exactly the same personalized pricing disclosure that the policy statement contemplates when personalization leads to higher prices. In the Commission’s example of a price for a security system based on knowledge that the consumer was a recent crime victim, the seller would have to disclose that “this price is personalized based on data showing you are a recent crime victim” whether the price was higher or lower than the usual price. When the price is a discount, however, the disclosure offers no benefits whatsoever. Defining “discounts” with enforceable precision is a difficult challenge, but the Commission can clearly state that it does not intend to require disclosure when information is used to offer discounts, and consider whether a given price is in fact a “discount” on a case-by-case basis.
Although the Proposed Policy is based on disclosure rather than a prohibition of personalized pricing, given the well-known fact that any communication can confuse or mislead a significant number of consumers,[23] required disclosure when prices are discounted risks harming consumers. The legal risk, the cost of the disclosure itself, and the risk of consumer misinterpretation of the disclosure will discourage sellers from offering these discounts. The Commission’s past attacks on discounting argue strongly for a safe harbor, rather than reliance on the Commission’s enforcement discretion.
The proposed policy statement would require that firms using personalized pricing “disclose not just that the price is personalized, but also the basis for that personalization and the types of data on which the personalization is based.”[24] In its evaluation of disclosures, the Taskforce on Federal Consumer Financial Law Report noted that “when disclosures are required, it should be with a clear, and articulated, goal in mind.”[25]
The contemplated disclosure fails this test. Is it simply intended to discourage personalized pricing? It will likely have that effect, but, as noted above, that will raise prices to the many beneficiaries of personalized discounts. Is it intended to motivate additional consumer search for a better price? If so, it would seem equally valuable to require any seller to disclose that its price is relatively high, whether personalized or not. Is the goal to encourage consumer use of behaviors and technologies that seek to conceal their information? If so, the disclosure should be at the point at which information is collected, when consumers can act on its message. Disclosure when a price is offered is in the wrong place and at the wrong time.
Rather than facilitating consumer choices, which should be the goal of all required disclosures, the proposed disclosure appears to be a “normative” disclosure – information provided not because consumers are interested, but because policymakers think consumers should take it into account.[26]
Like any disclosure, this one may be misunderstood. The Commission’s own research has shown that additional information can lead to worse consumer decisions, either because it is misunderstood or because consumers assign it more significance than it should have.[27] To ensure that consumers actually benefit from disclosure, the Commission should first test it, and only then consider requiring it.
In most circumstances, personalized pricing is unlikely to harm consumers. There may be harm, and a Section 5 violation, if there are instances in which personalized pricing borders on duress. Disclosure, however, does nothing to address the problem, and leaves a possibly better-informed consumer with the same unpalatable choice.
It is generally easy for consumers to avoid any injury from personalized pricing by the simple expedient of searching for a better price. The Commission’s premise that consumers generally expect one price for all is unsubstantiated. It seems highly likely that most consumers are aware that prices differ from one consumer to another because of coupon use, volume discounts, and loyalty programs. Because personalized discounts offer substantial consumer benefits, the Commission should create a safe harbor making clear that disclosures are not necessary when information is used to offer a discount. Disclosure, particularly untested disclosure, is an inappropriate remedy.
[1] Fed. Trade Comm’n, Proposed Enforcement Policy Statement Regarding Personalized Pricing, Matter No. P034101 (Aug. 19, 2026) [hereinafter “Proposed Policy”], https://www.ftc.gov/system/files/ftc_gov/pdf/p034101-ftc-enforcement-policy-statement-re-personalized-pricing-proposed-for-public-comment.pdf.
[2] I am currently Emeritus Professor of Strategic Management and Public Policy at the George Washington School of Business and a Senior Scholar at the GW Regulatory Studies Center. I was on the staff of the FTC in the Bureau of Economics and in the Bureau of Consumer Protection as an Assistant to the Director, Associate Director, and Acting Deputy Director from 1977 to 1986, and Director of the Bureau of Consumer Protection from 2001-2004. I am grateful to the International Center for Law and Economics for financial support.
[3] See Jean Tirole, The Theory of Industrial Organization 138 (1988) (“The welfare effects of third-degree price discrimination are ambiguous.”).
[4] See 14 Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application ¶ 2340c (2024) (“most price discrimination is socially beneficial in that it produces higher output and thus yields greater consumer benefits than forced nondiscriminatory pricing”); see also Benjamin Klein, Price Discrimination and Market Power, in 2 Issues in Competition Law and Policy 977, *20 (Wayne D. Collins ed., 2008), available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1657202 (“[P]rice discrimination can be used and in some circumstances accepted as a legitimate competitive business justification. . . .”).
[5] Proposed Policy, supra note 1, at 3.
[6] See Jean-Pierre Dubé & Sanjog Misra, Personalized Pricing and Consumer Welfare, 131 J. Pol. Econ. 131, 142 tbl. 1 (2023).
[7] Dubé & Misra, supra note 7, at 170. They note that “[e]ven with expanded data collection, it is unlikely that a firm could truly perfectly price discriminate using consumer data.” Id.
[8] See Jacques-François Thisse & Xavier Vives, On the Strategic Choice of Spatial Price Policy, 78 Am. Econ. Rev. 122 (1988).
[9] See, e.g., Restatement (Second) of Contracts §§ 175–76 (Am. L. Inst. 1981). See also Timothy J. Muris, Opportunistic Behavior and the Law of Contracts, 65 Minn. L. Rev. 521 (1981).
[10] The most promising theory is unfairness. There is substantial injury. By hypothesis, there is no alternative, and there are no offsetting benefits of the practice to consumers or competition. See, e.g., Holland Furnace Co. v. FTC, 295 F.2d 302 (7th Cir. 1961) (seller’s servicemen dismantled home furnaces and then refused to reassemble them until the consumers had agreed to buy services or replacement parts).
[11] Materiality is the injury inquiry in the context of deception: Would the consumer have done something differently to avoid the problem? Avoidability is the same question in the context of unfairness: Is there an injury-free alternative the consumer could reasonably have chosen?
[12] The proposed policy also points to correcting information and avoiding information collection in the first place as possible responses. But disclosure of personalization does not create an opportunity for correction, and existing disclosure requirements apply where correction rights exist. As discussed in Section V below, the point of sale is the wrong place for a disclosure aimed at information collection practices.
[13] Stigler examines a fixed sample size strategy in his seminal article on the economics of information. See George J. Stigler, The Economics of Information, 69 J. Pol. Econ. 213 (1961).
[14] See J. McCall, Economics of Information and Job Search, 84 Q.J. Econ. 113 (1970); Martin L. Weitzman, Optimal Search for the Best Alternative, 47 Econometrica 641 (1979).
[15] See Friedrich A. Hayek, The Use of Knowledge in Society, 35 Am. Econ. Rev. 519 (1945). Price conveys information even when the information is not observed directly, even when the information is not disclosed, and even when it may not be “consciously articulable” by any market participant. See Randal K. Quarles, Vice Chair for Supervision, Bd. of Governors of the Fed. Rsrv. Sys., Friedrich Hayek and the Price System Remarks at the Ninth Annual Conference of the William F. Buckley, Jr., Program at Yale 5 (Nov. 1, 2019), https://www.federalreserve.gov/newsevents/speech/files/quarles20191101a.pdf. For a formal model in a slightly different context, see Sanford Grossman, On the Efficiency of Competitive Stock Markets Where Trades Have Diverse Information, 31 J. Fin. 573 (1976).
[16] Robert H. Frank, Microeconomics and Behavior 434 (4th ed. 2006).
[17] See Proposed Policy, supra note 1, at 2–3 & nn. 4, 7.
[18] See Robert Pitofsky, Beyond Nader: Consumer Protection and the Regulation of Advertising, 90 Harv. L. Rev. 661 (1977); Robert Pitofsky, Randal Shaheen & Amy Mudge, Pricing Laws Are No Bargain for Consumers, 18 Antitrust 62 (2004); Timothy J. Muris & Paloma Zepeda, The Benefits, and Potential Costs, of FTC-Style Regulation in Protecting Consumers, 8 Competition L. Int’l 11 (2012).
[19] See Discount Marketing Statistics, Capital One Shopping (last updated June 15, 2026), https://capitaloneshopping.com/research/discount-statistics/.
[20] See 30+ Loyalty Program Statistics Businesses Should Know in 2026, Zoho Thrive (last visited Sept. 22, 2026), https://www.zoho.com/thrive/loyalty-program-software/statistics.html.
[21] See Arielle Feger, Why Loyalty Programs Are Hitting a Growth Ceiling, EMARKETER (July 31, 2026), https://www.emarketer.com/content/why-loyalty-programs-hitting-growth-ceiling.
[22] See Discount Marketing Statistics, supra note 20.
[23] As Karl Popper said, “Always remember that it is impossible to speak in such a way that you cannot be misunderstood: there will always be some who misunderstand you.” Karl Popper, Unended Quest: An Intellectual Autobiography 29 (1976) (emphasis in original). Academic studies of brief communications find that 20 to 30 percent of the audience misunderstand some aspect of the message. Regarding televised messages, see Jacob Jacoby, Wayne D. Hoyer & David A. Sheluga, Miscomprehension of Televised Communications 64 (1980). Regarding print communications, see Jacob Jacoby & Wayne D. Hoyer, The Comprehension and Miscomprehension of Print Communications: An Investigation of Mass Media Magazines (1987). Both studies compare advertisements with excerpts of editorial content designed to be roughly equal in length and find no significant differences in the extent of miscomprehension.
[24] Proposed Policy, supra note 1, at 1.
[25] Consumer Fin. Prot. Bureau, Taskforce on Federal Consumer Financial Law Report vol. I, at 326 (Jan. 2021, amended Jan. 2022) [hereinafter “Taskforce Report”], https://files.consumerfinance.gov/f/documents/cfpb_taskforce-federal-consumer-financial-law_report-volume-1_2022-01_amended.pdf. The CFPB’s Taskforce included four members with vast FTC experience, including me.
[26] See the discussion in the Taskforce Report, id. at 327–28 (noting that disclosures that reduce information costs help consumers achieve their goals but “normative disclosures seek to change their goals.”).
[27] See James M. Lacko & Janis K. Pappalardo, Fed. Trade Comm’n, Bureau of Econ., The Effect of Mortgage Broker Compensation Disclosures on Consumers and Competition: A Controlled Experiment 28–29 (Feb. 2004), https://www.ftc.gov/sites/default/files/documents/reports/effect-mortgage-broker-compensation-disclosures-consumers-and-competition-controlled-experiment/030123mortgagefullrpt.pdf. The experiment tested alternative disclosures on the closing forms required under the Real Estate Settlement Procedures Act, and found that when the broker’s compensation was disclosed, consumers were less likely to select the low-cost mortgage.
Source: Federal Trade Commission
I. Introduction and Summary The International Center for Law & Economics (‘ICLE’) welcomes the opportunity to respond to the Department for Business and Trade’s (‘DBT’) . . .
The International Center for Law & Economics (‘ICLE’) welcomes the opportunity to respond to the Department for Business and Trade’s (‘DBT’) consultation on Swifter and Simpler Competition Redress, Regulatory Appeals and Competition Enforcement.[1] ICLE is a nonprofit, nonpartisan research centre that promotes rigorous economic analysis in public-policy debates. Its scholars have written extensively on competition enforcement, private litigation, and regulatory design. ICLE has also responded to the government’s Refining Our Competition Regime consultation, the Competition and Markets Authority’s (‘CMA’) merger-efficiencies review, and its proposed first conduct requirements under the Digital Markets, Competition and Consumers Act 2024 (‘DMCC Act’).[2]
ICLE submits these comments by email as a single document rather than through the online survey. Headings identify any specific consultation questions addressed in a section. Paragraph references are to the Consultation unless otherwise indicated.
The Consultation’s three chapters share an ambition to make redress, appeals, and enforcement ‘swifter and simpler’. ICLE supports that aim. Speed and simplicity help when they lower the cost of reaching sound outcomes. They can, however, do harm when they remove safeguards needed to reach those outcomes. We therefore assess each proposal through the error-cost framework that has long guided law & economics analysis of enforcement institutions: Does the reform reduce the expected cost of error, including false positives, false negatives, and enforcement costs?[3]
Our principal conclusions follow.
The government’s diagnosis is broadly right, and most of its remedies point in the right direction. The regime has two linked problems. First, certification has become too easy, leaving funders’ appetite for returns to filter claims that the Competition Appeal Tribunal (‘CAT’) should assess for merit and proportionality. Seconds, even successful claims have returned very little to the consumers in whose name they are brought
ICLE supports codifying an absolute-suitability test with a weighted cost-benefit assessment (Q4–Q6), clarifying class representatives’ obligations through the CAT Rules (Q1–Q3), permitting damages-based agreements subject to disclosure (Q7–Q9), requiring budgets from both sides (Q13–Q14), and introducing settlement offers with automatic costs consequences (Q18–Q20). We oppose two proposals. Requiring the CAT to indicate at certification what funder return would be ‘reasonable’ (Q10) would import rate-of-return regulation into litigation finance. This risks blending into rate-of-return regulation of litigation finance with the associated cost inflation. It would be neither workable nor wise. Allowing the CMA to order redress based on a presumed overcharge (Q22–Q26) would turn an enforcement agency into an administrative damages tribunal without the experience, evidence, or standard of review that role requires.
ICLE supports transferring the CMA’s appellate functions to the CAT (Q44–Q45). Adjudication belongs in a judicial body, and the CMA’s resources are better spent on its primary functions.
That transfer should, however, preserve parties’ right to a meaningful appeal. Applying judicial-review principles ‘taking due account of the merits’ (Q47) would lower the standard that now governs energy appeals and replace water redeterminations with a materially more deferential process. Decisions that set the return on tens of billions of pounds of regulated investment warrant a genuine merits appeal.
Several proposals would sensibly reorganise procedures and institutions. Others would remove statutory safeguards and replace them with CMA guidance. Most notably, omitting Rule 3 (Q65) would allow the official who issued a statement of objections to sit in judgment on it.
The proposals would also withdraw the right to make confidentiality representations (Q67), narrow access to the file (Q75, Q77, Q78), and codify appeal waivers in settlement (Q68). Together, these would weaken checks on an agency whose infringement decisions bind courts in follow-on damages actions. ICLE supports faster competition-law procedures, but notes that procedural fairness protects the parties, lends credibility to enforcement decisions, and makes the UK a more attractive place to do business.
The opt-out regime began in 2015 with three objectives: better redress for victims of anticompetitive conduct, deterrence, and protection for defendants against unmeritorious claims.[4] A decade later, the evidence assembled in the Consultation suggests that the regime is falling short on redress and protection for defendants. It also pursues deterrence in a way that generates large private returns without commensurate public benefit.
The regime has grown rapidly since 2020. Claimants have filed cases with a cumulative estimated value exceeding £130 billion on behalf of some 655 million class members.[5] Claimant budgets routinely exceed £20 million, and defence budgets can be expected to match or exceed them.[6] Funders will not back a claim worth less than roughly £500 million.[7]
Yet in the only case to reach distribution, less than one per cent of the class made a claim. Approximately £216,000 reached consumers, while more than £10 million went to lawyers, funders, and other stakeholders.[8] In the only case to reach a judgment in the claimant’s favour, the Tribunal had to determine what a ‘fair’ commission for access to a digital platform would have been. As ICLE has explained elsewhere, this exercise in administered pricing is unlikely to produce stable or welfare-enhancing results. The judgment is now the subject of an application for permission to appeal.[9]
These outcomes reflect a well-understood divergence between the private and social incentives to litigate.[10] The lawyer or funder who selects and controls a claim captures much of its private return, while society bears the costs of adjudication and any over-deterrence. Research on ‘entrepreneurial litigation’ shows that when those who select claims are paid from the proceeds, they are likely to favour claims that promise returns to themselves over those that benefit the people they represent.[11] Opt-out proceedings sharpen that divergence because class members are dispersed, passive, and often unaware of the claim.
Two features of the UK regime make it especially susceptible to this dynamic. First, since Merricks, certification has been understood as a ‘relative suitability’ test that ‘does not involve a merits test’.[12] Second, because the regime covers only competition law, claimants have pleaded claims ‘not obviously rooted in competition’ as abuses of dominance. In some cases, the Tribunal has observed that the alleged conduct ‘seem[s] not to be competition law infringements at all’.[13]
Together, these features have made the regime increasingly attractive as an asset class for litigation finance rather than a route to redress: a portfolio of large standalone claims certified at a low threshold, whose value to investors depends on a handful settling or succeeding.
We agree with the government that the regime should be neither expanded nor further restricted at this stage.[14] The answer to the ‘dressing up’ problem is to enforce its existing boundaries. At certification, the Tribunal should require a pleaded competition claim with a real prospect of success, rather than accept a competition-law label alone. Cost-benefit analysis should also be undertaken; at the moment, only cost-effectiveness analysis is. This necessarily implies considering the market impact of the claim, including at the certification stage; this omission is the driver of claims which simply incur costs without improving markets (and sometimes harming them). The proposals discussed below would do much to achieve this.
ICLE supports all three proposals in paragraph 29. Together, they would restore the ‘strong process of judicial certification, including a preliminary merits test’ that the government promised when it created the regime and that subsequent case law has eroded.[15]
Intrinsic (absolute) suitability. A relative-suitability test asks whether collective proceedings are preferable to individual proceedings. Because individual claims for small losses are rarely viable, the answer is usually clear, and certification does little to filter claims. An intrinsic test asks whether the claims are fit to be grouped and determined collectively under the Act. It also asks whether collective proceedings offer a reasonable prospect of a just outcome at proportionate cost.[16] Where claimants seek aggregate damages, they must offer a methodology with a realistic prospect of establishing class-wide loss and evidence that the necessary data exist.[17]
Some mature class-action jurisdictions take a similar approach. In the United States, Rule 23 requires a ‘rigorous analysis’ that will ‘frequently … entail some overlap with the merits’.[18] The class plaintiff must also show that damages can be measured on a class-wide basis consistent with its theory of liability.[19]
Greater weight for costs and benefits, including expected take-up. This is, in our view, the Consultation’s most important reform. The error-cost logic is straightforward. Certification imposes immediate and certain costs: defence spending that can match a £20 million claimant budget, years of management distraction, and, critically, settlement pressure largely independent of a claim’s merit.[20] The benefit is uncertain and, on the evidence so far, small: only a fraction of a damages award reaches a fraction of the class.
A test that gives this comparison real weight and counts likely take-up among the benefits would screen out claims most likely to enrich stakeholders while compensating few class members. It would also encourage funders and law firms to design claims and distribution plans around the class rather than headline quantum.
To make the test operational, the CAT Rules should require the proposed class representative (‘PCR’) to file a costed distribution plan and a reasoned estimate of take-up at certification. A projected return to the class disproportionate to projected costs should be a presumptive ground for refusal. The Tribunal’s recent refusal to certify Waterside v Mowi because litigation costs far exceeded the likely benefit and distribution return to consumers shows that the existing Rules permit this reasoning. Codification would make it the norm rather than the exception.[21] Importantly, this should be cost-benefit analysis and not just cost-effectiveness analysis: there should be an assessment of whether the lawsuit will improve market performance, and not just its costs, or this analysis is unduly truncated.
Sufficient evidence for aggregate damages. Aggregate awards make opt-out proceedings viable, but they also make them risky. The Tribunal should distinguish injured from uninjured class members and tie awards to a sound theory of loss so that they reflect the harm class members actually suffered.[22] Rule 79(2)(f) already requires the Tribunal to consider whether claims are suitable for an aggregate award. Under the current approach, though, suitability is only one factor in a broader assessment of ‘all matters it thinks fit’.[23] Suitability for aggregate damages should instead be a condition of certification, supported by credible evidence grounded in the facts of the case.
The government should also address the ‘multiple bites of the cherry’ problem noted in paragraph 61. Repeatedly filing reformulated claims after a refusal imposes costs on defendants and the Tribunal. The Rules should require permission to refile and allow the Tribunal to make payment of the earlier application’s costs a condition of that permission.
The concern that a statutory merits element would generate satellite litigation (¶ 27) is overstated. Parties already contest certification at length. In Evans v Barclays, the Supreme Court held that the weakness of a claim was properly treated as a powerful factor against opt-out certification under Rule 79(3)(a), without requiring a separate merits trial.[24]
ICLE agrees that greater certainty would help in all three areas identified in paragraph 19: funding arrangements, costs specialists, and consultative panels.
The class representative is the only participant in the proceedings whose incentives are meant to align with the class. Yet a funder or law firm typically recruits the representative, who has no financial stake in the outcome and relies on the funder to meet an adverse costs order.[25]
Gutmann v First MTR South Western Trains illustrates the problem. The class representative, a private individual, could not have met costs beyond the £15 million indemnity his funder provided, so the defendants pursued the funder directly.[26] The funder had advanced about £6.58 million of the representative’s own costs—less than half the adverse costs cover it had provided. The solicitors absorbed later budget increases through deferred and contingent fees. When the defendants sought costs above the cap, the funder argued that the lead solicitors should bear the excess. The solicitors, in turn, wrote to the Tribunal to rebut the funder’s account. The representative sat between two parties financing the litigation, each with a contingent interest, without an independent adviser on the dispute. The agency problem identified in the class-action literature is acute.[27]
The Tribunal’s decisions in Riefa and Hammond ask the right questions: Does PCR understand and genuinely negotiate the funding terms? Has the PCR obtained independent costs advice? Is there a way to make informed decisions on settlement?[28]
The CAT Rules should answer those questions, with guidance on what is proportionate for a given claim. Rule 78 should require the PCR to (i) disclose the litigation funding agreement (‘LFA’) to the Tribunal, including provisions on funder control, termination, and payment priority; (ii) confirm that the PCR obtained independent advice on the funding terms; and (iii) explain how the representative will decide questions of settlement and distribution, including through any consultative panel.[29] This would address the concern in paragraph 17 that PCRs might overspend to cover every base. The requirement is informed independence, not a prescribed governance structure.
Damages-based agreements (Q7–Q9). ICLE cautiously supports lifting the prohibition on damages-based agreements (‘DBAs’) in opt-out proceedings. The economic case rests on competition. When third-party funders paid a multiple of their investment are the only source of finance, they gain substantial pricing power over the claims the regime exists to enable.[30] Allowing lawyers to take a share of recoveries would introduce a competing source of capital and—as the experience in Victoria, Australia suggests—may lower funding costs and improve returns to the class.[31] It would also align lawyers’ incentives more closely with the size of the recovery than hourly billing does.
The government identifies credible risks, but certification safeguards can address them. We recommend three: full disclosure of the DBA to the CAT through the Rule 78 process; approval of the percentage at settlement or judgment based on the sum actually distributed, rather than the headline award (see Section II.D); and CAT approval of ‘stacked’ arrangements in which a funder and law firm each take a return from the same recovery, so that the CAT can assess their combined cost to the class.
Funder returns: rate-of-return analysis of damages (Q10–Q12). The proposal in paragraph 52 responds to a genuine problem. Funders do not know what return the CAT will approve at the end of a claim, and that uncertainty raises the cost of capital.[32] But asking the CAT at certification to indicate a reasonable return and payment order would, in substance, regulate the rate of return on litigation finance, albeit reading on damages. It would bring the well-documented problems of that model into the CAT.
Rate-of-return regulation asks a decision-maker to determine in advance what return on invested capital is ‘reasonable’. Decades of utility regulation reveal three recurring problems.[33] First, the decision-maker cannot obtain all the information needed to set the rate correctly. The appropriate return on a funded claim depends on the funder’s cost of capital, how the claim’s risk relates to the rest of its portfolio, and the likelihood of success. The CAT cannot observe those factors at certification.[34]
Second, fixing a return as a multiple of cost, or damages, gives the regulated party an incentive to inflate its cost base or the damages estimate (e.g., by not settling). In litigation, an ‘approved’ multiple on invested capital or damages would reward spending more rather than litigating better. The Consultation’s evidence that claimant budgets have doubled within months of certification (¶ 60) suggests that this dynamic is already present. Third, an early indication would become an anchor for later decisions. The promised discretion to ‘revisit’ the assessment at the end of the claim (¶ 52) could prove illusory in practice.
A better response has three parts. First, funders should obtain certainty from their agreements and efficient operations so that they are competitive providers of finance, rather than advance CAT approval. If the PCR has disclosed the LFA, negotiated it with independent advice, and shown at certification that it preserves the PCR’s independence, the Rules should provide that the CAT will enforce the agreed return absent fraud, undue influence, or a material change in the claim. Second, DBAs and the transparency measures proposed above would discipline returns through competition. Third, the Rules should address returns that rise with the passage of time and thus reward delay past the next step-up date. They could bar recovery from the class of any increase triggered after a reasonable settlement offer of the kind discussed in Section II.D. That would target the incentive to delay without regulating the rate. Indeed, a regulated return on damages would likely discourage settlement and slow down lawsuits (e.g. by going to trial to extract more damages than settlement would give).
ICLE sees the force of the time-value argument for paying funders when the CAT makes an award or approves a settlement (¶ 53; Q11–Q12), but the proposed presumption raises a concern. If the funder receives its entire return from the gross sum before distribution, no participant retains a financial interest in whether the class receives anything. Gutmann shows how little may otherwise reach class members.[35]
We would support a presumption that the funder recovers its capital and costs on award or approval, while the CAT ties payment of the profit element to distribution. The funder could draw any remaining profit from undistributed funds only after the CAT finds that reasonable distribution efforts were made. This would preserve an incentive to design and carry out distribution effectively. The CAT’s residual discretion (Q12) is welcome but should not be the class’s only protection. Indeed, the major problem has been precisely that the claimant class is already too far back in the queue; further demoting their claims is not the answer here.
Budgets and costs conduct (Q13–Q14). ICLE supports requiring both parties to submit budgets after certification and to notify the CAT of material changes. Transparency on both sides reduces surprises when costs are assessed and helps the CAT exercise its powers under Rule 104.[36]
The Rules should not presume that vigorous defence of a claim worth hundreds of millions of pounds is a ‘scorched earth’ tactic (¶ 61). Costs sanctions should apply equally to both sides, turn on objective conduct—such as applications that fail comprehensively, repeated failure to follow directions, or refiling a refused certification application—and follow an assessment of the record. The cost-benefit test at certification offers the strongest discipline on claimant budgets. A realistic prospect that only meritorious claims will be certified offers the strongest discipline on defendant budgets.
Mediation (Q15–Q17). ICLE supports allowing the CAT to order mediation after the parties exchange expert reports and to account for an unreasonable refusal when awarding costs. This would align with the approach now taken under the Civil Procedure Rules.[37] At that point, both sides have enough information to value the claim. Mandatory ADR before filing would add costs without resolving disputes, as the government recognises (¶ 71).
Settlement offers with automatic costs consequences (Q18–Q20). ICLE supports extending a Rule 45-type mechanism to collective proceedings. The current asymmetry is striking. A funded PCR faces little downside from refusing a reasonable defendant offer. A defendant that refuses a Calderbank offer faces, at most, a discretionary costs sanction that, as Kent v Apple shows, may never be imposed.[38] Automatic consequences for both sides would create stronger incentives to settle.
Existing CAT approval under Rule 94 addresses the concern that such a mechanism might put undue pressure on PCRs (Q19). The CAT must find a settlement just and reasonable before approving it, so an offer cannot compel acceptance of a bad settlement.[39] The mechanism should otherwise follow Rule 45, with enough time for acceptance to accommodate CAT approval. The approval process itself (Q21) should develop through experience, with a review once the CAT has considered more settlements.
Distribution and undistributed sums (Q34–Q37). Listing certified claims on the CAT website is sensible and inexpensive. But when fees and returns depend on the gross award, no participant has a financial interest in getting money to the class. Evidence from the mature US regime shows that claims rates in consumer class actions are typically in single digits.[40]
The reforms proposed above would create that interest: counting expected take-up in the certification cost-benefit test, requiring a costed distribution plan, and tying payment of funder profit to distribution. Parties could also agree that undistributed sums revert to the defendant, an option the government rightly leaves to them (¶ 129).
Destination of undistributed damages (Q37). Large undistributed sums may show that the parties overstated the claim’s benefit to the class at certification. The CAT should be able to consider that record in later certification decisions involving the same representatives, funders, or firms.
ICLE also strongly advises against designating the Consumers’ Association (Which?), an active class representative and litigant before the CAT.[41] Directing undistributed damages to an organisation that brings collective proceedings creates a perception—and a real risk—of circularity: more certified claims and less distribution would mean more money for that organisation. US experience with cy pres awards to organisations connected to litigants has drawn scrutiny from the Supreme Court.[42] By contrast, the Access to Justice Foundation may have little connection to competition law, but it has no stake in the regime.
ICLE supports simplifying the Redress Scheme Regulations (Q23),[43] allowing the CMA to accept binding redress undertakings when a business volunteers them in a settlement or commitments decision, and making clear that payments under a redress scheme reduce any subsequent damages claim and count at certification (Q24, Q27). These measures would make voluntary redress easier without displacing the adjudication of loss.
We oppose the two proposals that go further: a statutory presumption of overcharge and a power for the CMA to direct a business to establish a redress scheme (Q22, Q25–Q26).
Presumed overcharge. Damages compensate loss actually suffered. The Organisation for Economic Co-operation and Development (OECD) impact-assessment methodology and meta-studies of historical cartels discussed in paragraph 92 can help estimate the aggregate benefits of enforcement across many cases. The issue is to identify the true overcharge, because correcting the overcharge is how the economic harm of principled cases is addressed. The underlying studies report overcharges ranging from zero to well above 50 per cent, so any single default figure will systematically overcompensate some purchasers and undercompensate others.[44] A presumed overcharge will therefore not achieve optimal deterrence.
The Consultation would also apply the presumption to abuse-of-dominance cases, where ‘overcharge’ is often a poor fit because the alleged harm concerns exclusion rather than price. And because the same agency would find the infringement, impose a penalty, and set the presumed sum, the presumption could operate in practice as a second penalty under another name. It would increase deterrence without improving compensation.[45]
Mandatory redress. The government rejected this power in 2013 for reasons that remain relevant: assessing loss lies outside the CMA’s expertise and would divert resources from investigation and enforcement.[46] The CMA’s role has since expanded considerably, while Chapter 3 of the Consultation proposes to reduce safeguards around its infringement decisions (see Section IV below).
Giving the CMA power to order compensation, subject to a standard of review the Consultation does not specify, would place investigation, prosecution, adjudication, penalties, and damages assessment in one administrative body. ICLE warned against that concentration of functions in its response to the January consultation.[47] It would also sit uneasily with the principle that redress should follow from the merits of a claim rather than regulatory direction. If the government wants more redress in follow-on cases, it should make the collective regime fit for purpose through the certification reforms discussed above
Leniency is the most effective cartel-detection tool available to competition authorities. The decline in Type A applications from an average of 21 a year in 2014–16 to eight a year in 2022–24 is a serious warning sign.[48] The economics are well understood: a firm will report its own conduct only if the expected benefit of immunity exceeds the expected cost of exposure. As private damages exposure has grown, the calculation has shifted against reporting.[49] Immunity from penalties offers little to a firm whose main exposure is civil liability.
ICLE therefore supports extending immunity from civil liability to Type A applicants in principle (Q28), subject to two safeguards (Q29–Q30). First, the government should retain the set-aside for unrecoverable damages proposed in paragraph 108 and define it using the Damages Directive’s model. Under that model, the immunity recipient remains liable when victims cannot obtain full compensation from the other infringers.[50]
Second, immunity should depend on continued cooperation with claimants as well as the CMA. The US model, for example, conditions relief from treble and joint-and-several liability on ‘satisfactory cooperation’ with civil plaintiffs.[51] That would make the immunity recipient a source of evidence for claims against the remaining cartel members and help address the concern about compensation. The objection that the recipient ‘escapes’ liability has less force when its report enables the cartel’s detection and makes redress possible (¶ 103).
ICLE supports transferring the CMA’s appellate functions to the CAT and bringing telecoms price-control disputes within its jurisdiction. Consolidation would give these disputes a judicial forum and reduce duplication. The CAT will need the capacity and sectoral expertise to take on the work. The transfer should also preserve meaningful scrutiny of decisions that shape billions of pounds in infrastructure investment. A genuine merits standard, with power to substitute a decision where the correct answer is clear, would allow the CAT to correct errors without unnecessary remittals.
ICLE supports transferring the CMA’s appellate functions to the CAT, converting the remaining reference functions into appeals, and extending the CAT’s jurisdiction to price-control matters in telecoms appeals.
The current allocation assumes that sectoral regulation is best reviewed by a body with deep experience assessing pricing decisions across sectors. The Competition Commission, and later the CMA, built that expertise through decades of price-control appeals and references. But deciding a dispute between a regulator and a regulated firm is a judicial function. The Consultation itself accepts that a judicial body with the right technical expertise ‘would be equally capable of dealing with these issues’ (¶ 166).[52]
The CAT can acquire that sectoral depth by selecting and, where necessary, supplementing its Ordinary Members (Q46). It would thereby gain independence without sacrificing technical competence. Transfer would also free CMA resources: a single large regulatory appeal can absorb the equivalent of two or three market studies (¶ 167). The CAT already hears appeals from the Office of Communications (Ofcom), the Office of Gas and Electricity Markets (Ofgem), the Payment Systems Regulator (PSR), and the Civil Aviation Authority (CAA). Consolidation would improve consistency and help precedent develop across sectors, as the Penrose report recommended five years ago.[53]
Extending the CAT’s jurisdiction to price-control matters in telecoms appeals (Q45) follows the same logic. The CAT currently hears a telecoms appeal but must refer any price-control element to the CMA, then incorporate the CMA’s conclusions into its judgment (¶ 174).[54] Splitting one appeal between a judicial body and an administrative agency adds procedure, lengthens timetables, and divides the reasoning between institutions with different processes. If the CAT can determine price-control appeals in water and energy, as the Consultation proposes, it can do so in telecoms.
The proposed transfer also bears on the CMA’s independent panel. In March, ICLE argued that the panel should not be abolished without introducing merits-based appellate review at the CAT. The panel provided an independent, quasi-judicial layer of decision-making that helped justify deferential judicial review.[55] The Consultation now cites the panel’s impending abolition as a reason to move regulatory appeals out of the CMA (¶ 169). While that is sound, abolishing the panel does not eliminate the need for the functions it performs. The CAT is the right place for regulatory appeals. For merger and market decisions, the same logic points to merits review at the CAT, which the government has yet to propose.
The government should assess the CAT’s staffing needs before the transfer takes effect. Comparing the expected caseload with the pool of people who have the necessary sectoral and financial expertise, and who are eligible and willing to serve as Ordinary Members, would let the CAT build capacity in advance. Without that assessment, the reform could improve the forum while lengthening the queue—undermining the timeliness gains it seeks.
The Consultation proposes that the CAT decide all regulatory appeals by applying ‘the same principles as would be applied by a court on an application for judicial review, while also taking due account of the merits of the case’. This follows the telecoms model in section 194A of the Communications Act 2003 (¶ 204(a)).[56] For most appeals within scope, it would reduce the intensity of review. Energy licence-modification appeals currently succeed when a decision is wrong on specified grounds, including errors of fact and failure to give appropriate weight to relevant matters. The CMA may substitute its own decision in price-control appeals, while water redeterminations involve a fresh decision on the merits.[57] The proposal would replace those approaches with a review in which the CAT ‘takes [the regulator’s] decision as a starting point’, accords it ‘an appropriate degree of respect’, and asks whether it is ‘materially wrong’ (¶ 202).
The government may argue that the telecoms formula also asks whether a decision is ‘wrong’ (¶ 201), but two differences matter. First, under the existing energy, water, and air regimes, an error of fact, a failure to give appropriate weight to relevant matters, or—in the air regimes—an error in the exercise of discretion can independently justify allowing an appeal. Under section 194A, the CAT considers the merits through the judicial-review grounds of illegality, irrationality, and procedural unfairness. A decision can therefore be wrong without meeting the threshold for intervention.
Second, treating the regulator’s decision as a starting point gives it a degree of deference. An appeal on specified grounds instead requires the reviewing body to form its own view of whether the decision was wrong, giving the regulator’s judgement only the weight the evidence warrants.
Research on regulatory disputes supports that distinction. Despoina Mantzari argues that deference to a regulator’s economic assessment should depend on the relative institutional competence of the regulator and the reviewing body, rather than on the regulator’s expertise alone.[58] The CAT’s own economic and technical expertise supports meaningful scrutiny, even when the regulator’s specialised judgement deserves weight.
ICLE’s concern is with the error costs of a deferential standard. Price controls set the allowed return on tens of billions of pounds of infrastructure investment, and regulatory error runs in both directions. If a regulator sets the cost of capital too low and an appeal cannot correct it, consumers may bear the cost of underinvestment for decades. If it sets the allowed return too high, consumers pay excess returns for a full control period. The Consultation itself acknowledges that the current arrangements ‘can … deter investments in UK markets’ (¶ 199).
From a law & economics perspective, the question is how the review standard allocates the risk of error among regulators, regulated firms, investors, and consumers. A merits standard, combined with the proposed standing for consumer bodies that ICLE supports, would allow errors in either direction to be corrected.
The Consultation’s principal argument for deference is that many elements of a price control, ‘e.g. the cost of capital’, are ‘complex matters requiring the exercise of regulatory judgement with no single right answer’ (¶ 199). That may be true, but a merits appeal can still give the regulator appropriate latitude within a range of defensible values. In the RIIO-2 appeals, the CMA upheld the networks’ challenges to the ‘outperformance wedge’ and, for five of the eight appellants, an innovation uplift unsupported by evidence. It nonetheless dismissed every challenge to the cost of equity.[59]
Fresh redetermination has not simply favoured regulated firms, either. In the PR24 redeterminations, the CMA raised the allowed return from 4.03 to 4.20 per cent but granted the five disputing companies £463 million in additional revenue—17 per cent of the £2.7 billion they sought.[60] A judicial-review standard would do more than respect regulatory judgement: it would leave wrong decisions standing when they fall short of irrationality. The lack of a single right answer strengthens the case for external review because a regulator may struggle to identify its own errors.
The standard also matters to the government’s growth agenda. Water and energy networks must raise substantial private capital over the coming decade, and investors account for the available appellate review when pricing risk. If an erroneous determination of the allowed return can be corrected only when it is ‘materially wrong’ under judicial-review principles, investors may demand a higher risk premium, with consequences for allowed returns and investment.
A merits appeal need not entail unrestricted de novo review. It asks whether the regulator’s decision has evidential support and accords with its statutory duties. It does not require the CAT to take over the regulator’s continuing administrative role. The weight due to a technical assessment will depend on the issue under review.
ICLE therefore recommends a genuine merits appeal set out in statute. A single list of grounds drawn from the existing energy, water, and air regimes should cover error of fact; error of law; failure to consider, or give appropriate weight to, relevant matters; failure to follow the applicable procedure; failure of a modification to achieve its stated effect; and error in the exercise of discretion. The CAT should decide whether the regulator was wrong on those grounds, giving its assessment the weight the evidence warrants.[61] This would largely extend the existing energy standard across sectors. The Consultation offers no evidence that the standard has produced excessive appeals or unmanageable proceedings.
Requiring a notice of appeal to identify whether each ground alleges an error of fact, law, or discretion (Q48) would help focus proceedings, provided a mistaken label does not bar a sound argument. ICLE also agrees with the approach to new evidence (Q49) and cross-examination (Q50) in paragraph 204(c): the CAT should retain discretion under its existing Rules to allow both. Restricting appeals to material available to the regulator could reward opaque reasoning, since an appellant may not know what evidence informed the decision until its publication. The principal barriers the proposals would create (Q52) are thus the review standard and the proposed rule on remedies.
On remedies (Q51), mandatory remittal should not be the default. When the CAT finds an error and the correct answer follows from that finding—for example, a discrete input to a price control—remittal would delay the outcome and invite another appeal. The CAT should be able to substitute its own decision on that point, as the CMA can in energy price-control appeals and as the appeal bodies can in the air and postal regimes (¶ 195).[62] It should remit cases when the regulator needs to exercise fresh judgement in light of the CAT’s findings. Confined in this way, substitution corrects an error without beginning a second round of regulation.
ICLE supports faster enforcement, but the CMA’s decisions must remain reliable when they bind courts in follow-on damages claims. The Rules should keep final decision-makers separate from the team that investigated a case, preserve a meaningful chance to protect confidential information and inspect the evidence, and maintain safeguards when the CMA imposes interim measures. Settlement procedures can be simpler, provided they give parties adequate access to the case against them and clearly define the scope of their admissions.
Rule 3 of the CMA Rules embodies a simple principle: the officials who decide to issue a statement of objections (‘SO’) should not decide whether those objections are well founded.[63] The Consultation proposes removing Rule 3 and letting the CMA set its decision-making model in guidance. One option is an ‘SRO + 2’ model, in which the senior responsible officer (‘SRO’) who issued the SO would make the final decision alongside two colleagues who had not worked on the case (¶ 229).
Basic procedural fairness weighs against this change. Separating investigative and decision-making functions within an enforcement agency is an important safeguard against confirmation bias: the tendency to interpret new evidence as supporting a theory one has already formed.[64] The concern is not misconduct by CMA officials. Any decision-maker can be influenced by a view formed earlier in a case, and institutional rules should account for that risk.
Critics have long challenged the European Commission’s combination of these functions. The European Court of Human Rights has accepted such an arrangement in an administrative competition authority where a court can fully review its decisions on the facts and the law.[65] The UK’s case-decision-group model was a deliberate improvement on the Commission’s approach.
The consumer-protection regime offers no reason to remove Rule 3. It also separates functions, but the CMA sets that safeguard itself and can revise it. The difference concerns where the rule sits—in CMA guidance or in a statutory instrument. It does not diminish the need for separation. Nor does the SRO’s familiarity with the case outweigh the risk of confirmation bias. Two uninvolved colleagues may give undue weight to the views of the officer who led it.
Three considerations make separation especially important. First, infringement decisions under the Competition Act 1998 bind courts in follow-on damages actions,[66] which now include opt-out collective claims worth hundreds of millions of pounds. The reliability of the underlying decision matters more than it did in 2014. Second, although the CAT hears appeals on the merits, [67] an appeal is costly and slow, and a party that settles waives it. For a meaningful share of decisions, internal separation is the only check. Third, moving the safeguard from the Rules to guidance would let the CMA change it after a consultation it controls, without parliamentary scrutiny.
If the government considers Rule 3 too prescriptive, it should retain the core requirement in the Rules: people who took part in the investigation or decided to issue the SO should not make the final infringement or penalty decision. The CMA could set the decision-makers’ composition and procedures in guidance. The same principle should govern settlements (Q66). A party’s admission does not relieve the CMA of its duty to establish an infringement, and someone outside the case team that negotiated the settlement should make that judgement.
Rule 7 (Q67). Rule 7 should be amended, not omitted. The Consultation may be right that the CMA must process too many blanket confidentiality claims that offer no reasons. It should therefore require parties to identify and explain each claim, and allow the CMA to disclose information without further representations when a claim is unreasoned or manifestly unfounded. Parties should retain the right to make representations on substantiated claims. Competition Act 1998 investigations can uncover trade secrets whose disclosure is irreversible. Even well-resourced authorities have failed to protect such information.[68] A chance to be heard before disclosure is a minimal safeguard.
Confidentiality rings (Q71–Q74). ICLE supports a standardised framework that gives economic as well as legal advisers access to confidentiality rings, so parties need not negotiate the arrangements afresh in every case. This is already the case following Practice Direction 2/2026 and the framework there, which amounts to a rebuttable presumption, seems apt. Predictability would benefit both the CMA and the parties it investigates. Civil sanctions for breaches (Q72) are acceptable in principle if they require fault, are proportionate to the breach, and provide a safe harbour for prompt self-reporting. Employer liability should apply to systemic failures, not automatically to every breach (Q73). Extending the framework to mergers and markets (Q74) is sensible if the same safeguards apply.
Access to the file (Q75–Q77). Aligning the Rule 6(2) exceptions with the consumer-enforcement rules would let the CMA withhold ‘routine correspondence’ and documents ‘to the extent that they identify an individual’ (¶ 253).[69] Neither category defines its own limits. In an abuse-of-dominance investigation, correspondence the CMA regards as routine may matter to the defence. Any new exception should be narrow and require a list of withheld documents, so a party can request specific items.
Disapplying Rule 6 altogether in settlement cases (Q77) is more troubling. A settling party admits an infringement that binds courts in follow-on claims and gives up its right of appeal. It should understand the full case file before doing so, rather than see only the documents the CMA chooses to cite. The CMA could offer streamlined access if the party agrees to it. The Rules should not impose it as the default.
Interim measures (Q78). ICLE opposes removing Rules 13(1) and 13(2). Interim measures are the CMA’s most intrusive tool short of a final decision. They impose binding obligations on the basis of a prima facie case, before any finding of infringement.[70] Removing the right to inspect the file could require a business to change its conduct without seeing the evidence said to justify the direction. A targeted rule could address the efficiency concern: give the business access to documents on which the CMA relies, through a confidentiality ring where needed, within a short, fixed period. The safeguard should remain in the Rules rather than move to guidance.
Settlement benefits both sides. It saves CMA resources and gives the settling party a penalty reduction and an earlier end to the investigation. But settlement decisions also affect third parties: they establish binding infringement findings on which follow-on collective claims may rest.
Together, the proposals would allow an infringement decision without an SO (Q69–Q70), full access to the file (Q77), or a right of appeal (Q68). Each proposal reflects existing practice in isolation.[71] In combination, they could produce a decision that no one outside the case team has tested against the full evidence, yet that carries the weight of a binding infringement finding in litigation worth many times the penalty.
ICLE does not object to codifying the agreement not to appeal. The Rules should, however, also make clear that the settling party admits only the facts and infringement set out in the settlement decision, and that follow-on claimants cannot extend that admission. The decision should state that it resulted from a settlement.
When all parties settle before an SO, a summary statement of facts may serve as the section 31 notice if it sets out the CMA’s case with the specificity that section 31 requires and the parties have seen the documents on which it relies. The consumer-enforcement rules should not set the benchmark for this process. The procedural rules should instead reflect the difference between conduct whose illegality is usually clear and conduct whose illegality turns on contested economic analysis.
The Consultation is right to seek swifter, simpler competition redress, regulatory appeals, and enforcement. Several proposals would also improve accuracy: certification that tests claims’ merit and proportionality, settlement offers with meaningful consequences, a judicial forum for regulatory appeals, and standardised confidentiality arrangements. ICLE supports these proposals.
Other proposals would sacrifice accuracy for speed. Regulating the rate of damages, allowing the CMA to order redress based on a presumed overcharge, weakening review of regulatory decisions worth tens of billions of pounds, and removing the separation between investigation and decision-making in Competition Act 1998 cases would concentrate discretion while weakening checks on it. These proposals draw on the DMCC Act’s consumer-enforcement provisions, which address a narrower set of issues and remain largely untested. Their effects should be understood before they become a model for more complex and contested cases.
The reforms most likely to meet the government’s aims will align private incentives with public purposes: claims brought on their merits, funders rewarded when consumers receive compensation, and binding enforcement decisions reached through fair procedures.
[1] Dep’t for Bus. & Trade, Swifter and Simpler Competition Redress, Regulatory Appeals and Competition Enforcement (17 July 2026) [hereinafter Consultation].
[2] Geoffrey A. Manne et al., Refining Our Competition Regime: ICLE Response to the Department for Business and Trade Consultation Published 20 January 2026 (Int’l Ctr. for L. & Econ., 31 March 2026) [hereinafter ICLE Refining Comments]; Dirk Auer, Selcukhan Ünekbas & Mario A. Zúñiga, Comments of the International Center for Law & Economics: UK Competition and Markets Authority Call for Evidence for Merger Efficiencies Review (Int’l Ctr. for L. & Econ., 25 February 2026); Geoffrey A. Manne et al., Comments of the International Center for Law & Economics: The CMA’s Proposed Steering Conduct Requirements for Apple and Google (Int’l Ctr. for L. & Econ., 28 July 2026) [hereinafter ICLE Steering Comments].
[3] See Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1 (1984); Geoffrey A. Manne & Joshua D. Wright, Innovation and the Limits of Antitrust, 6 J. Competition L. & Econ. 153 (2010).
[4] Consultation, supra note 1, ¶ 1.
[5] Id. ¶ 38 (citing CMS, European Class Action Report 2025, at 14, 36 (2025)).
[6] Consultation, supra note 1, ¶ 34.
[7] Id. ¶ 38.
[8] Id. ¶¶ 29, 116 (discussing Gutmann v First MTR South Western Trains Ltd [2025] CAT 72).
[9] Dr Rachael Kent v Apple Inc [2025] CAT 67 (application for permission to appeal pending); see ICLE Steering Comments, supra note 2, at 11 (discussing the Tribunal’s difficulty in determining a fee justified by the value provided).
[10] Steven Shavell, The Fundamental Divergence Between the Private and the Social Motive to Use the Legal System, 26 J. Legal Stud. 575 (1997); see also William M. Landes & Richard A. Posner, The Private Enforcement of Law, 4 J. Legal Stud. 1 (1975).
[11] John C. Coffee, Jr., Understanding the Plaintiff’s Attorney: The Implications of Economic Theory for Private Enforcement of Law Through Class and Derivative Actions, 86 Colum. L. Rev. 669 (1986); Jonathan R. Macey & Geoffrey P. Miller, The Plaintiffs’ Attorney’s Role in Class Action and Derivative Litigation: Economic Analysis and Recommendations for Reform, 58 U. Chi. L. Rev. 1 (1991); John C. Coffee, Jr., Entrepreneurial Litigation: Its Rise, Fall, and Future (2015).
[12] Merricks v Mastercard Inc [2020] UKSC 51, ¶¶ 59, 70, 74.
[13] Consultation, supra note 1, ¶ 7 (quoting Gormsen v Meta Platforms, Inc [2023] CAT 10, ¶ 30(1)); see also Gutmann v First MTR South Western Trains Ltd [2025] CAT 79 (holding that ‘[c]ompetition law is not a general law of consumer protection’).
[14] Consultation, supra note 1, ¶ 10.
[15] Dep’t for Bus., Innovation & Skills, Private Actions in Competition Law: A Consultation on Options for Reform—Government Response ¶ 5.55 (January 2013) [hereinafter 2013 Government Response]; Consultation, supra note 1, ¶ 21.
[16] Merricks v Mastercard Inc [2020] UKSC 51, ¶¶ 117–119.
[17] Id. ¶¶ 120–123.
[18] Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 350–51 (2011).
[19] Comcast Corp. v. Behrend, 569 U.S. 27, 34–35 (2013); see also In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 307 (3d Cir. 2008) (requiring courts to resolve factual disputes relevant to certification, even when they overlap with the merits).
[20] See In re Rhone-Poulenc Rorer Inc., 51 F.3d 1293, 1298–1300 (7th Cir. 1995) (Posner, J.) (describing the ‘intense pressure to settle’ that certification imposes even on defendants with strong defences); Merricks v Mastercard Inc [2020] UKSC 51, ¶¶ 118, 154, https://supremecourt.uk/uploads/uksc_2019_0118_judgment_6dd114686b.pdf (Lord Sales & Lord Leggatt, separate judgment) (noting that collective actions ‘confer substantial legal advantages on claimants and burdens on defendants which are capable of being exploited opportunistically’).
[21] Waterside Class Ltd v Mowi ASA [2026] CAT 32; Consultation, supra note 1, ¶ 29 & n.15; see also Competition Appeal Tribunal Rules 2015, SI 2015/1648, r. 79(1)–(2) (UK), https://www.legislation.gov.uk/uksi/2015/1648/article/79.
[22] See Comcast Corp. v. Behrend, 569 U.S. 27, 35–38 (2013).
[23] See Merricks v Mastercard Inc [2020] UKSC 51, ¶ 61 (‘The listed factors are not separate suitability hurdles, each of which the applicant for a CPO must surmount.’), https://supremecourt.uk/uploads/uksc_2019_0118_judgment_6dd114686b.pdf.
[24] Evans v Barclays [2025] UKSC 48.
[25] See Gutmann v First MTR South Western Trains Ltd [2025] CAT 21.
[26] Id.
[27] See Coffee, Understanding the Plaintiff’s Attorney, supra note 11; Macey & Miller, The Plaintiffs’ Attorney’s Role, supra note 11.
[28] Riefa v Apple Inc [2025] CAT 5, ¶¶ 89(5), 95, 114; Hammond v Amazon.com, Inc. [2025] CAT 42, ¶¶ 56, 67(1), 156(1).
[29] Competition Appeal Tribunal Rules 2015, SI 2015/1648, r. 78 (UK).
[30] See Jonathan T. Molot, Litigation Finance: A Market Solution to a Procedural Problem, 99 Geo. L.J. 65 (2010).
[31] Consultation, supra note 1, ¶ 42 (citing evidence from Victoria, Australia, after the introduction of group costs orders under s. 33ZDA of the Supreme Court Act 1986 (Vic.)).
[32] Consultation, supra note 1, ¶¶ 46–51 (discussing Merricks v Mastercard Inc [2025] CAT 28, ¶ 188; Gutmann v Apple Inc [2024] CAT 18, ¶ 38; and Gutmann v Apple Inc [2025] EWCA Civ 459, ¶ 78).
[33] See 1 Alfred E. Kahn, The Economics of Regulation: Principles and Institutions ch. 2 (1988); Harvey Averch & Leland L. Johnson, Behavior of the Firm Under Regulatory Constraint, 52 Am. Econ. Rev. 1052 (1962); Jean-Jacques Laffont & Jean Tirole, A Theory of Incentives in Procurement and Regulation (1993).
[34] See Friedrich A. Hayek, The Use of Knowledge in Society, 35 Am. Econ. Rev. 519 (1945).
[35] See Gutmann v First MTR South Western Trains Ltd [2025] CAT 21.
[36] Competition Appeal Tribunal Rules 2015, SI 2015/1648, r. 104 (UK).
[37] See Churchill v Merthyr Tydfil County Borough Council [2023] EWCA Civ 1416; Consultation, supra note 1, ¶¶ 74–76.
[38] Consultation, supra note 1, ¶¶ 78–80.
[39] Competition Appeal Tribunal Rules 2015, SI 2015/1648, r. 94 (UK).
[40] Fed. Trade Comm’n, Consumers and Class Actions: A Retrospective and Analysis of Settlement Campaigns 11–13 (September 2019) (finding a median claims rate of 9 per cent and a weighted mean of 4 per cent across the settlements studied); see also Consultation, supra note 1, ¶ 122 (noting that take-up in North America ‘will not often exceed ~10 per cent’).
[41] Consultation, supra note 1, ¶ 127.
[42] Marek v. Lane, 571 U.S. 1003 (2013) (Roberts, C.J., statement respecting denial of certiorari); Frank v. Gaos, 139 S. Ct. 1041 (2019) (per curiam); Martin H. Redish, Peter Julian & Samantha Zyontz, Cy Pres Relief and the Pathologies of the Modern Class Action: A Normative and Empirical Analysis, 62 Fla. L. Rev. 617 (2010).
[43] Competition Act 1998 (Redress Scheme) Regulations 2015, SI 2015/1587 (UK).
[44] See Oxera, Quantifying Antitrust Damages: Towards Non-Binding Guidance for Courts (December 2009) (study for the European Commission finding wide variation in estimated cartel overcharges); John M. Connor & Robert H. Lande, Cartel Overcharges and Optimal Cartel Fines, in 3 Issues in Competition Law and Policy 2203 (ABA Section of Antitrust Law 2008).
[45] See Wouter P.J. Wils, The Relationship Between Public Antitrust Enforcement and Private Actions for Damages, 32 World Competition 3 (2009).
[46] 2013 Government Response, supra note 15, ¶ 6.36; Consultation, supra note 1, ¶ 89.
[47] ICLE Refining Comments, supra note 2, at 5–8.
[48] Consultation, supra note 1, ¶¶ 103–104; see also Organisation for Economic Co-operation and Development (OECD), The Future of Effective Leniency Programmes: Advancing Detection and Deterrence of Cartels (2023).
[49] See Massimo Motta & Michele Polo, Leniency Programs and Cartel Prosecution, 21 Int’l J. Indus. Org. 347 (2003); Giancarlo Spagnolo, Leniency and Whistleblowers in Antitrust, in Handbook of Antitrust Economics (Paolo Buccirossi ed., 2008).
[50] Directive 2014/104/EU, art. 11(4), 2014 O.J. (L 349) 1.
[51] Antitrust Criminal Penalty Enhancement and Reform Act of 2004, Pub. L. No. 108-237, § 213, 118 Stat. 661, 666–67 (2004).
[52] See OECD, Recommendation of the Council on Transparency and Procedural Fairness in Competition Law Enforcement, OECD/LEGAL/0465 (2021) (calling for review by an impartial, independent adjudicative body).
[53] John Penrose MP, Power to the People: Independent Report on Competition Policy 18 (February 2021); Consultation, supra note 1, ¶¶ 152 n.62, 171.
[54] Communications Act 2003, c. 21, § 193 (UK).
[55] ICLE Refining Comments, supra note 2, at 5–8.
[56] Communications Act 2003, c. 21, § 194A(2) (UK); TalkTalk Telecom Group PLC v Office of Communications [2020] CAT 8, ¶¶ 137–140 (5 March 2020).
[57] Consultation, supra note 1, ¶¶ 192, 195 (table); see Gas Act 1986, c. 44, §§ 23D–23E (UK); Electricity Act 1989, c. 29, §§ 11E–11F (UK).
[58] Despoina Mantzari, Economic Evidence in Regulatory Disputes: Revisiting the Court-Regulatory Agency Relationship in the US and the UK, 36 Oxford J. Legal Stud. 565 (2016).
[59] Competition & Mkts. Auth., Energy Licence Modification Appeals 2021: Summary of Final Determination (28 October 2021) (upholding challenges concerning the outperformance wedge and the innovation-related ongoing-efficiency uplift, while rejecting challenges concerning the cost of equity).
[60] Competition & Mkts. Auth., Water PR24 References: Final Determinations—Summary (10 March 2026) (following an 18 March 2025 reference, raising the allowed return from 4.03 to 4.20 per cent and allowing £463 million in additional revenue, 17 per cent of the £2.7 billion sought).
[61] See Consultation, supra note 1, ¶ 195 (table); Gas Act 1986, c. 44, § 23D(4) (UK); Electricity Act 1989, c. 29, § 11E(4) (UK); Water Industry Act 1991, c. 56, § 12F(4) (UK); Transport Act 2000, c. 38, § 19B (UK); Civil Aviation Act 2012, c. 19, § 26 (UK).
[62] Consultation, supra note 1, ¶ 195 (table) (identifying substitution powers in energy price-control, air traffic and airport licence, and postal price-control appeals); Gas Act 1986, c. 44, § 23E (UK); Postal Services Act 2011, c. 5, § 59 (UK).
[63] Competition Act 1998 (Competition and Markets Authority’s Rules) Order 2014, SI 2014/458, sch., r. 3 (UK), https://www.legislation.gov.uk/uksi/2014/458/pdfs/uksi_20140458_en.pdf.
[64] See Ian S. Forrester, Due Process in EC Competition Cases: A Distinguished Institution with Flawed Procedures, 34 Eur. L. Rev. 817 (2009).
[65] A. Menarini Diagnostics S.r.l. v. Italy, App. No. 43509/08, ¶¶ 57–67 (Eur. Ct. H.R. 27 September 2011).
[66] Competition Act 1998, c. 41, § 58A (UK).
[67] Competition Act 1998, c. 41, sch. 8, ¶ 3 (UK).
[68] See ICLE Refining Comments, supra note 2, at 13–14 (discussing the inadvertent public disclosure of confidential Microsoft materials in the Microsoft/Activision proceedings).
[69] Competition & Mkts. Auth., Direct Consumer Enforcement Rules r. 5 (CMA201, April 2025).
[70] Competition Act 1998, c. 41, § 35 (UK).
[71] Competition & Mkts. Auth., Guidance on the CMA’s Investigation Procedures in Competition Act 1998 Cases ch. 14 (CMA8, December 2024).
Source: UK Department for Business and Trade
I. Introduction The International Center for Law & Economics (ICLE) supports Kuiper Systems LLC’s (Amazon Leo) application to provide direct-to-device (D2D) service in the 1.6/2.4 . . .
The International Center for Law & Economics (ICLE) supports Kuiper Systems LLC’s (Amazon Leo) application to provide direct-to-device (D2D) service in the 1.6/2.4 GHz mobile-satellite service (MSS) bands (ICFS File No. SAT-LOA-20260601-00224; Call Sign S00917).
Satellite broadband and D2D are emerging, capital-intensive markets. High fixed costs and long deployment cycles make it difficult to enter the consumer low-Earth-orbit (LEO) broadband market.[1] Amazon Leo’s deployment would help a competitor capable of operating at scale to overcome those hurdles, thereby strengthening competition and benefiting consumers.[2]
Amazon Leo’s proposed entry also shows how existing spectrum transactions can support new service. Amazon has agreed to acquire Globalstar, including its 1.6/2.4 GHz MSS authorizations, subject to the transaction’s closing.[3] The deal would pair Globalstar’s operating MSS network and spectrum authorizations with Amazon Leo’s planned D2D constellation. It would put scarce spectrum to more intensive use while giving companies a predictable basis for investment, consistent with the Federal Communications Commission’s (FCC) longstanding support for secondary-market spectrum transactions.
The FCC should grant the application and appropriate limited waivers so that another operator can expand D2D capacity using licensed spectrum that sees limited use. It should also address legitimate coexistence and interference concerns raised by terrestrial licensees and other satellite operators.
LEO satellite broadband remains a developing market with steep entry costs. Before serving customers, a provider must invest heavily in satellites, launches, ground infrastructure, and regulatory approvals. Building a constellation—and earning a return on it—takes years.[4]
Those costs, along with spectrum constraints, likely limit how many providers can operate at scale. There is no fixed number of competitors that every broadband market needs,[5] but entry by a credible additional provider can put pressure on incumbents and give consumers more choice.[6] Timely application review and workable spectrum-sharing rules can reduce the uncertainty and delay that compound the costs of entry.[7]
Starlink dominates commercial LEO broadband service in the United States and nothing in the record suggests that it gained that position through anticompetitive conduct. Still, another provider capable of operating at scale would give consumers and terrestrial partners an alternative.
Amazon Leo proposes a D2D constellation of 5,105 satellites in five orbital shells, at altitudes ranging from 510 to 580 kilometers.[8] It plans to use the 1.6/2.4 GHz band and standards developed by the 3rd Generation Partnership Project (3GPP) to connect with compatible commercial handsets, without specialized customer equipment.[9] That design could extend messaging, location sharing, roadside assistance, and emergency communications to people beyond the reach of terrestrial networks, including those in rural and remote areas.[10] D2D could also provide backup connectivity when disasters disrupt terrestrial service.[11]
The application and Amazon’s proposed acquisition of Globalstar could also make more intensive use of existing MSS authorizations. MSS bands support satellite links to consumer devices, but use varies considerably by band.[12] The 2 GHz MSS band sees relatively limited use,[13] while Globalstar and Iridium actively serve customers in the Big LEO band. Globalstar uses 1.6 GHz frequencies for uplinks and 2.4 GHz frequencies for downlinks; Iridium uses 1.6 GHz frequencies in both directions.[14] Satellite operators are also developing D2D services through agreements involving MSS rights and terrestrial mobile spectrum.[15]
If the transaction closes, Amazon Leo could pair Globalstar’s operating network and MSS authorizations with its planned constellation. That combination offers a practical way to expand the capacity and reach of spectrum that already supports D2D service.
Amazon Leo seeks limited waivers of several rules in Parts 2 and 25.[16] Some requests address filing procedures. Others concern frequency allocations, handset licensing, limits on signal strength, the treatment of links to users and ground stations, and rules for reviewing competing satellite applications.
The FCC may waive a rule for good cause when special circumstances warrant a departure and a waiver would serve the public interest. That includes cases in which strict application would frustrate the rule’s purpose.[17] The question for each request is whether the FCC can grant relief while retaining meaningful technical review and protection against interference.
Amazon Leo’s Schedule S request illustrates the distinction. Schedule S collects technical data for satellite applications. Amazon Leo entered a representative orbital plane in the form and supplied a separate spreadsheet with the full dataset the rules require.[18] If those materials let FCC staff assess individual satellites and potential interference, requiring Amazon Leo to enter the same data again in Schedule S would add work without strengthening review.
The FCC took a similarly tailored approach in its June 2026 Amazon Leo milestone order. It granted a limited waiver with conditions designed to preserve incentives for rapid deployment and efficient spectrum use.[19] The FCC should grant the requests here where Amazon Leo shows that substantive review and interference protections will remain intact, imposing conditions where needed to protect coexistence.[20]
Expedited review must still address credible technical concerns from terrestrial licensees and other satellite operators. The FCC should evaluate evidence of potential interference and, where needed, impose conditions tied to actual performance or require coordination. That approach can protect existing services while allowing more intensive non-geostationary orbit (NGSO) operations and more efficient spectrum use.[21]
Good-faith frequency coordination is essential to sharing spectrum. The FCC can address documented objections through conditions and coordination requirements, leaving operators room to negotiate the details of coexistence. Speculative disputes should not delay a decision on the application.
The FCC should grant Amazon Leo’s application and the limited waivers that preserve substantive technical review and safeguards against interference. High entry costs and long deployment timelines make a credible additional LEO provider especially valuable. Amazon Leo’s planned constellation could expand D2D service through compatible handsets, give consumers and network partners another option, and provide connectivity where terrestrial networks are unavailable or disrupted.
If Amazon’s acquisition of Globalstar closes, the combined system could put Globalstar’s operating network and MSS authorizations to more intensive use. The FCC should address documented engineering concerns through tailored conditions and coordination requirements. Timely approval on those terms would support competitive entry, protect existing services, and make more efficient use of scarce spectrum.
[1] Kristian Stout & Michael Calabrese, Competition in the Low-Earth-Orbit Satellite Industry, Truth on the Mkt. (Oct. 10, 2023), https://truthonthemarket.com/2023/10/10/competition-in-the-low-earth-orbit-satellite-industry [hereinafter Stout & Calabrese, LEO Competition]. See also Kuiper Systems LLC, Application for Extension or Waiver of the Milestone Deadline, Order, ICFS File No. SAT-MOD-20260129-00065, DA 26-553, ¶ 1 (Space Bur. June 5, 2026) (calling Amazon Leo a “new competitor-at-scale” in low-Earth-orbit broadband).
[2] Michael Calabrese et al., Low Earth Orbit Satellites: Policies to Promote Spectrum Sharing, Foster Competition, and Close Digital Divides (Int’l Ctr. for L. & Econ. & New America Oct. 30, 2025), https://laweconcenter.org/wp-content/uploads/2025/10/Low_Earth_Orbit_Satellites_FINAL.pdf [hereinafter LEO Policy Working Group].
[3] Approval of the Proposed Assignment and Transfer of Control of Licenses and Authorizations Held by Globalstar to Amazon, Public Notice, GN Docket No. 26-134, DA 26-550, at 1 (June 4, 2026) (accepting the transfer applications for filing).
[4] Stout & Calabrese, LEO Competition, supra note 1.
[5] Eric Fruits et al., Dynamic Competition in Broadband Markets: A 2024 Update (Int’l Ctr. for L. & Econ. June 4, 2024), https://laweconcenter.org/wp-content/uploads/2024/06/Broadband-Competition-2024-Update.pdf [hereinafter Fruits et al., Dynamic Competition]; Eric Fruits et al., A Review of the Empirical Evidence on the Effects of Market Concentration and Mergers in the Wireless Telecommunications Industry (Int’l Ctr. for L. & Econ. Sept. 17, 2019), https://laweconcenter.org/wp-content/uploads/2019/09/ICLE-Telco_Merger_Lit_Review_Jud_Rpt_FINAL.pdf; Thomas W. Hazlett & Robert W. Crandall, Competitive Effects of T-Mobile/Sprint: Analysis of a “4-to-3” Merger, in Proceedings of the TPRC2024: The Research Conference on Communications, Information and Internet Policy (2024), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4736059.
[6] Fruits et al., Dynamic Competition, supra note 5. See also Kuiper Systems LLC, DA 26-553, ¶ 1.
[7] LEO Policy Working Group, supra note 2.
[8] Application of Kuiper Systems LLC for Authority to Launch and Operate a Non-Geostationary Satellite Orbit System, Legal Narrative § II, ICFS File No. SAT-LOA-20260601-00224 (July 24, 2026) [hereinafter Kuiper D2D Application] (proposing 5,105 satellites in five orbital shells at altitudes of 510–580 km); see also Space Bureau Accepts for Filing Application of Amazon Leo for NGSO D2D System, Public Notice, ICFS File No. SAT-LOA-20260601-00224, DA 26-866, at 1 (Space Bur. Aug. 18, 2026) (accepting the application for filing).
[9] Kuiper D2D Application, supra note 8, at 7, 10.
[10] Id. at 10 (explaining that D2D would improve coverage for rural and remote users); see also id. at 2 (describing service for people beyond existing networks).
[11] Id. at 11 (explaining that D2D could keep rescue and emergency communications running when terrestrial networks fail); see also id. at ii (describing D2D as a fallback during terrestrial outages).
[12] LEO Policy Working Group, supra note 2.
[13] Timothy M. Farrar & J. Armand Musey, Spectrum for Emerging Direct-to-Device Satellite Operators (TMF Assocs. & Summit Ridge Grp. Jan. 27, 2025), https://summitridgegroup.com/wp-content/uploads/D2D-White-Paper-SRG-TMF-Final.pdf [hereinafter Farrar & Musey, SRG/TMF].
[14] Id.
[15] GSMA Intell., What Does the Early Data on D2D Usage Tell Us? (May 29, 2026), https://www.gsmaintelligence.com/blogs/what-does-the-early-data-on-d2d-usage-tell-us. See also Fed. Commc’ns Comm’n, Order, DA 26-398, ¶¶ 1–2 (Space Bur. Apr. 23, 2026), https://docs.fcc.gov/public/attachments/DA-26-398A1.pdf (reaffirming licensees’ exclusive mobile-satellite-service spectrum rights and encouraging secondary-market transactions).
[16] See Kuiper D2D Application, supra note 8, Ex. B (Waiver Requests), at 1, 6–9; 47 C.F.R. § 1.3 (2026); Ne. Cellular Tel. Co. v. FCC, 897 F.2d 1164, 1166 (D.C. Cir. 1990); WAIT Radio v. FCC, 418 F.2d 1153, 1159 (D.C. Cir. 1969).
[17] 47 C.F.R. § 1.3 (2026); Ne. Cellular Tel. Co. v. FCC, 897 F.2d 1164, 1166 (D.C. Cir. 1990); WAIT Radio v. FCC, 418 F.2d 1153, 1157, 1159 (D.C. Cir. 1969).
[18] Kuiper D2D Application, supra note 8, Ex. B (Waiver Requests), at 7.
[19] Kuiper Systems LLC, DA 26-553, ¶¶ 1–2.
[20] Kuiper Systems LLC, DA 26-553, ¶¶ 1–2; Gerald Adams, Out of Orbit: How Outdated Spectrum Rules Hold Back Satellite Broadband (Int’l Ctr. for L. & Econ. Aug. 12, 2026), https://laweconcenter.org/resources/out-of-orbit-how-outdated-spectrum-rules-hold-back-satellite-broadband [hereinafter Adams, Out of Orbit].
[21] See Adams, Out of Orbit, supra note 20; Eric Fruits & Kristian Stout, Comments of the International Center for Law & Economics Regarding Modernizing Spectrum Sharing for Satellite Broadband, SB Docket No. 25-157 (July 28, 2025), https://laweconcenter.org/wp-content/uploads/2025/07/2025-EPFD-Comments.pdf.
Source: Federal Communications Commission
I. Introduction and Summary The International Center for Law & Economics (ICLE) submits these comments in response to the Federal Communications Commission’s (FCC) Notice of Proposed Rulemaking (NPRM) on reducing...
The International Center for Law & Economics (ICLE) submits these comments in response to the Federal Communications Commission’s (FCC) Notice of Proposed Rulemaking (NPRM) on reducing local barriers to broadband deployment.[1] ICLE is a nonprofit, nonpartisan research center that applies law & economics to public policy. Its work promotes sound economic analysis and consumer welfare, particularly in dynamic, technology-driven markets such as telecommunications.
The NPRM proposes reforms to streamline the state and local approvals that providers must obtain before building wireline infrastructure. Deployment often requires numerous permits, rights-of-way authorizations, and access agreements. Slow, discretionary, or unpredictably priced approval processes increase the cost and risk of investment and can make otherwise viable projects uneconomical.
These barriers carry especially high costs in broadband markets. Broadband networks require substantial upfront investments that providers cannot recover if a project fails. They also generate broad economic and social benefits that providers cannot fully capture. Permitting delays and excessive fees therefore shape where and how quickly providers build next-generation networks—and whether they build them at all. The resulting costs fall hardest on rural and lower-income communities, the very communities that federal connectivity policy seeks to reach.
The FCC should adopt its proposals to require cost-based fees and establish firm deadlines, known as shot clocks, for local review of wireline-infrastructure applications. It should pair those deadlines with a strong deemed-granted remedy that treats an application as approved when a locality fails to act on time. The FCC should also prevent localities from evading fee caps by demanding noncash, or in-kind, contributions. Together, these reforms would reduce avoidable deployment costs and help broadband providers build the infrastructure needed to support next-generation applications, including artificial intelligence.
As the NPRM explains, providers must secure numerous permits, rights-of-way authorizations, and access agreements before deploying infrastructure.[2] Slow, inconsistent, or unpredictably priced approvals increase both the cost and risk of deployment. These barriers can deter investment even when a provider has capital available. They therefore shape the cost of individual projects, where and how quickly providers build next-generation infrastructure, and whether they build it at all.
Two features of broadband deployment make the FCC’s proposed reforms especially important. First, broadband networks require high fixed costs, and providers cannot recover much of that investment if a project fails.[3] A provider must incur most network costs before serving a single customer. If the network attracts too few customers, much of that spending is lost.[4]
Providers spread these fixed costs across their customer base. Projects serving large or affluent populations are therefore more likely to produce an adequate return, while projects in rural and lower-income communities often remain marginal.[5] Every avoidable regulatory cost or delay pushes more of those marginal projects below the threshold for investment.
Second, broadband creates broad social benefits that providers cannot capture through subscription revenue.[6] Reliable connectivity can raise property values, improve educational outcomes, expand access to telemedicine in ways that make rural living more viable, and allow remote workers to bring income into lower-cost communities.[7] Providers cannot charge for a farmhouse’s increased resale value or a student’s improved grades.
These spillover benefits create an underinvestment problem. A broadband project may produce substantial value for society while generating too little revenue to justify private investment. AI, augmented reality, and virtual reality will widen this gap because their benefits will extend far beyond what any single broadband provider can monetize. Policies that reduce deployment costs and risks therefore create value beyond the revenue that providers earn.
State and local requirements impose transaction costs and create holdout risks that can exceed the cost of the equipment itself. A single linear route may require agreements with several municipalities, county and state authorities, private landowners, and owners of thousands of utility poles.[8] Each party controls a segment of the project and has an incentive to seek the highest payment its approval can command. Yet no single party bears or fully accounts for the cumulative burden imposed on the network.
Many projects thus require consent from a long chain of independent parties, any one of which can block the entire build.[9] This structure creates a holdout problem. After a provider has invested in a route and obtained nearly every required approval, the final gatekeeper can demand far more than the underlying access is worth. The provider must then overpay, redesign the route, or abandon the investment already sunk into the rest of the project.
Delay and unpredictability often impose greater costs than the fees themselves. A project’s economics depend heavily on when it begins generating revenue relative to when its costs accrue. Protracted or discretionary review postpones revenue while financing and development costs continue to mount. Each month of delay also consumes part of an asset’s useful economic life and increases the risk of technological obsolescence.[10] A 10-year asset that spends three years awaiting approval must repay its debt from seven years of revenue instead of 10.
Open-ended standards compound these costs. A provider can budget for a fixed fee. It cannot price a discretionary decision with the same confidence, so it must demand a higher expected return to compensate for the uncertainty. Firm deadlines with meaningful consequences for inaction, paired with fees tied to a government’s actual and direct costs, reduce the risk premium built into deployment decisions. This benefit extends even to projects completed on schedule, because providers must evaluate them ex ante based on the possibility of delay.
These burdens fall most heavily on marginal projects. Excessive delay and denial risk raise the minimum return needed to justify deployment, eliminating projects in rural and lower-income communities first.
The FCC can directly address these problems by establishing firm deadlines backed by a presumption of approval, limiting fees to reasonable government costs, and counting in-kind demands against those limits. These safeguards target the delay, discretion, and rent extraction that impose the largest avoidable costs and risks on broadband deployment.
As explained above, broadband providers need regulatory certainty when deciding whether to invest. ICLE supports the proposed 120-day shot clock, but the remedy for missed deadlines matters more than the precise timeframe.[11] If the record supports a shorter period to encourage deployment or a longer one to let localities manage their public rights-of-way, the FCC should adjust the deadline accordingly.
Whatever period the FCC chooses, it must provide swift and meaningful enforcement. A deadline offers little relief if a locality can ignore it, stall a deployment, and create a de facto moratorium on new builds. Although the NPRM declined to propose a deemed-granted remedy, it left the issue open, stating that the FCC “may revisit this decision if evidence submitted to the Commission suggests that a ‘deemed granted’ remedy is needed.”[12]
The economics of broadband deployment provide sufficient evidence of that need. Without an automatic consequence for inaction, providers remain exposed to the same uncertainty that the shot clock should eliminate. The FCC could wait for a more extensive record, but broadband deployment would suffer in the meantime. It should instead adopt a deemed-granted remedy under which an application receives approval when a locality fails to act before the deadline.
The FCC should also apply a single deadline to all authorizations required for access to a public right-of-way. Allowing localities to stack separate shot clocks for rights-of-way agreements, construction permits, road-closure permits, and other approvals would preserve the delay and uncertainty the rule seeks to eliminate.
The FCC should clarify that the shot clock begins to run when a provider initiates the first procedural step in the application process. A firm deadline does little to constrain delay if a jurisdiction can postpone its start by withholding acknowledgment of an application, imposing open-ended preapplication or completeness requirements, or excluding informal preliminary steps from the clock.[13]
These open-ended requirements introduce the same costly discretion as an unenforceable deadline. In the aggregate, they can recreate the de facto moratoria that the shot clock should prevent. Predictability requires an objective, provider-triggered starting point that localities cannot manipulate. The clock should therefore begin when the provider takes the first required procedural step, rather than when the locality later decides that the application is complete.
The FCC should also allow providers to submit batch applications. It has previously approved batching for small-cell deployments.[14] Although wireline projects differ in scale, the same logic applies when a provider proposes substantially uniform work, such as repeated attachments using the same design along a fiber route or materially identical installations using common construction methods.
Batch review would benefit providers and localities alike. Providers could avoid duplicative filings, fees, and staggered timelines. Localities could conduct one engineering and legal review of the shared elements instead of repeatedly examining identical specifications. That approach would accelerate deployment while easing the burden on resource-constrained local review staff.
The NPRM cites several examples of localities imposing fees that exceed the actual costs of processing permits and maintaining public rights-of-way.[15] These examples provide ample support for cost-based fee rules. Two additional disputes further illustrate what happens when gatekeepers can impose charges untethered from their costs.
Recent events in Maple Grove, Minnesota, show how directly some localities treat broadband deployment as a revenue opportunity. In March 2026, the city amended its code to require every “Broadband Service Provider” to obtain a cable-television franchise from the Northwest Suburbs Cable Communications Commission (NSCCC) before the city would act on any rights-of-way application.[16]
Gateway Fiber is a certificated telecommunications carrier that provides broadband internet access, rather than cable service. When it sought permits to continue construction, the city deferred its applications. A provider accustomed to receiving permits within four to six weeks suddenly could not obtain them at all.[17]
The local commission stated its financial motive plainly. It projected that cable-franchise-fee revenue would fall by about 10% annually and identified “new broadband fees” as the “most impactful” source of replacement revenue.[18] The city’s Request for Council Action likewise tied the ordinance to “declining cable franchise revenue.” It said nothing about coordinating construction, protecting pavement, or preventing conflicts among users of the rights-of-way—the interests that might justify such a requirement.[19] The ordinance sought to turn a broadband entrant into a replacement source of franchise revenue.
A recent pole-replacement dispute illustrates a related problem. Owners of essential infrastructure can use their gatekeeping position to impose charges untethered from cost. In February 2026, the FCC rejected Appalachian Power Company’s (APCo) “cost causer” policy. The policy required a company seeking to attach equipment to an APCo pole to pay upfront 100% of the cost of replacing a pole that already violated applicable standards, subject to a possible 50% refund.[20]
The FCC held that a new attacher may pay only the incremental cost of the taller or stronger pole its equipment requires. APCo could not charge the attacher for correcting a preexisting violation.[21] The dispute nonetheless continued after the ruling. Rather than apply the incremental-cost standard, APCo allegedly billed Comcast a flat share of roughly 20% of the full replacement cost, regardless of the costs Comcast’s attachment caused. APCo also reportedly treated 20% as a floor rather than a ceiling, prompting Comcast to file another complaint.[22]
Investor-owned utilities may have especially strong incentives to extract rents, but similar incentives operate when local governments and permitting offices control access needed for broadband deployment. A rule limiting fees to “reasonable” or “actual” costs will constrain those incentives only if the FCC defines the standard concretely and provides an effective remedy. Providers should be able to challenge systemic overcharges without relitigating the same issue one gatekeeper and one pole at a time.[23]
A fee cap that excludes in-kind contributions constrains the form of compensation while leaving the total burden unchecked. When a locality conditions access on free service to municipal buildings, dedicated fiber or conduit, capacity on an institutional network, or a comparable obligation, the provider bears the added cost of supplying that service or facility.[24]
An in-kind obligation requiring $1,000 in additional spending imposes the same direct burden as a $1,000 cash payment. Its total cost may be even greater if the obligation displaces commercial uses or forces the provider to expand capacity sooner than planned. The precise cost depends on the provider’s available capacity and the alternative uses it could reasonably expect.
A cap limited to cash payments therefore allows localities to collect compensation above the nominal ceiling. These demands can increase subscriber prices, reduce service quality, and discourage investment. Although the effects depend on market conditions and the nature of the obligation, they fall hardest on marginal projects that broadband policy should encourage.
Cable franchising offers a cautionary precedent. Congress defined “franchise fee” broadly to prevent local authorities from evading the 5% statutory ceiling by demanding services or facilities instead of cash.[25] Localities cannot circumvent Congress’ limit simply by changing the form of compensation.[26] The FCC recognized this principle in its 2019 Section 621 order, which held that nonincidental in-kind obligations count toward the 5% cap.[27]
The same principle applies to the fee limits in this proceeding. The FCC should adopt its proposal to count the value of state and local in-kind demands toward any safe-harbor fee level.[28] Covered demands should include dedicated capacity, free service to public facilities, donated equipment, and similar assessments that function as payment.
To make the limit durable, the FCC should require the locality to justify the value assigned to each in-kind demand. That valuation should reflect the provider’s marginal cost of meeting the demand and its opportunity cost from displaced commercial uses.
A threshold objection is that Section 253 protects the ability to provide “telecommunications service,” while the FCC currently classifies broadband internet access as a Title I information service. That objection misconstrues both the statute and the deployments at issue. The FCC can apply firm deadlines and cost-based fee limits to the facilities covered by this proceeding without changing the classification of retail broadband. Section 253 protects the deployment of infrastructure capable of carrying telecommunications, regardless of the regulatory classification assigned to a particular retail service.
Section 253(a) reaches any state or local requirement that prohibits or has the effect of prohibiting “the ability of any entity to provide any interstate or intrastate telecommunications service.”[29] This language extends beyond firms that sell telecommunications directly to consumers. It also covers entities that deploy infrastructure capable of providing telecommunications services.[30]
The same fiber that delivers mass-market broadband often carries—or can carry—interconnected voice over internet protocol (VoIP), business-data and enterprise-transport services, dark-fiber leases, and point-to-point connections. Providers building these networks also commonly hold state telecommunications certificates and file tariffs for regulated services.[31] The relevant considerations are the network’s telecommunications capabilities and the burden imposed on them. A provider’s use of the same fiber to offer an information service does not place the infrastructure beyond Section 253’s reach.
The FCC should therefore adopt its proposal to prohibit state and local governments from imposing additional requirements on wireline telecommunications infrastructure because the facilities “may be used to provide other services.”[32] This commingled-facilities rule would prevent localities from accomplishing indirectly what Section 253 forbids them to do directly.
The proposal also follows the approach the FCC has taken in the cable context. Its mixed-use rule prevents franchising authorities from regulating noncable services carried over a cable operator’s mixed-use system.[33] Applying the same principle here places the shot-clock and fee-cap rules on settled statutory ground, even when providers also use the infrastructure to offer an information service.
The proposed reforms reflect the basic economics of infrastructure investment. Broadband deployment requires large, largely sunk investments under substantial uncertainty. It also creates broad social benefits that no provider can fully capture through subscription revenue. State and local delay, discretion, and rent extraction raise the minimum return needed to justify deployment, eliminating marginal projects first. Rural and lower-income communities bear the greatest burden, even though federal broadband policy seeks to reach them most urgently.
Firm deadlines backed by a deemed-granted remedy, fees tied to a government’s actual, direct costs, and in-kind demands counted toward those limits would reduce the risk built into every deployment decision. The FCC can adopt these measures on settled statutory grounds. Section 253 reaches requirements that prohibit or effectively prohibit an entity’s ability to provide telecommunications service. That protection extends to infrastructure capable of carrying telecommunications, regardless of how the provider’s retail services are classified.
ICLE therefore urges the FCC to adopt a firm shot clock backed by a deemed-granted remedy, cost-based fee limits that include in-kind contributions, and a commingled-facilities rule that prevents localities from indirectly imposing requirements that Section 253 directly prohibits. These reforms would give providers the predictability needed to invest and help ensure that next-generation networks reach the communities that need them most.
[1] See Build America: Eliminating Barriers to Wireline Deployment, Notice of Proposed Rulemaking, WC Docket No. 25-253, FCC 26-40 (rel. June 25, 2026), https://docs.fcc.gov/public/attachments/FCC-26-40A1.pdf.
[2] See also Jeffrey Westling & Kristian Stout, The Permission Layer: Clearing the Way for AI-Era Broadband 13 (Int’l Ctr. for L. & Econ. Sept. 9, 2026), https://laweconcenter.org/resources/the-permission-layer-clearing-the-way-for-ai-era-broadband.
[3] Geoffrey A. Manne, Kristian Stout & Ben Sperry, A Dynamic Analysis of Broadband Competition: What Concentration Numbers Fail to Capture 20 (Int’l Ctr. for L. & Econ. 2021), https://laweconcenter.org/resources/a-dynamic-analysis-of-broadband-competition-what-concentration-numbers-fail-to-capture; see also Geoffrey A. Manne et al., Comments on Refreshing the Record in Restoring Internet Freedom and Lifeline Proceedings, WC Docket Nos. 17-108, 17-287 & 11-42, at 2 (Int’l Ctr. for L. & Econ. 2020), https://laweconcenter.org/resources/comments-on-refreshing-the-record-in-restoring-internet-freedom-and-lifeline-proceedings-in-light-of-the-d-c-circuits-mozilla-decision.
[4] See Comments of the International Center for Law & Economics, The State of Competition in the Communications Marketplace, GN Docket No. 26-78, at 8 (May 21, 2026), https://laweconcenter.org/resources/icle-comments-to-the-fcc-on-the-state-of-competition-in-the-communications-marketplace [hereinafter ICLE Competition Comments].
[5] Westling & Stout, supra note 2, at 4.
[6] Jeffrey Westling, The Last Mile Is a Paper Trail: Why Broadband Gets Stuck, Truth on the Mkt. (Apr. 22, 2026), https://truthonthemarket.com/2026/04/22/the-last-mile-is-a-paper-trail-why-broadband-gets-stuck.
[7] See Paraoma Sanyal et al., Fiber Deployment Has Significant Incremental Economic Benefits 32 (The Brattle Grp. 2024), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5370089.
[8] Kristian Stout, Infrastructure Is Destiny: The Geography of the Next Technology Race 18 (Int’l Ctr. for L. & Econ. Issue Brief 2026), https://laweconcenter.org/resources/infrastructure-is-destiny-the-geography-of-the-next-technology-race.
[9] Robert Cooter & Aaron Edlin, The Falcon’s Gyre: Legal Foundations of Economic Innovation and Growth 6.21 (2014), https://www.law.berkeley.edu/library/resources/cooter.pdf.
[10] ICLE Competition Comments, supra note 4, at 11.
[11] Ajit Pai, Comm’r, Fed. Commc’ns Comm’n, Remarks at PCIA’s 2014 Wireless Infrastructure Show 3 (May 20, 2014), https://www.fcc.gov/document/commissioner-pai-remarks-pcia-2014; see also Comments of the International Center for Law & Economics and TechFreedom, In re Applications of Cellco P’ship & SpectrumCo LLC, WT Docket No. 12-4, at 9 (Mar. 16, 2012), https://laweconcenter.org/resources/in-the-matter-of-application-of-cellco-partnership-spectrumco-llc-for-consent-to-assign-licenses-wt-docket-no-12-4 (noting that the FCC’s wireless shot clock “has had little effect in practice”).
[12] Build America, supra note 1, ¶ 32.
[13] See Comments of the Free State Foundation, Eliminating Barriers to Wireline Deployments, WC Docket No. 25-253, at 5 (Nov. 18, 2025), https://freestatefoundation.org/wp-content/uploads/2025/11/FSF-Comments-Eliminating-Barriers-to-Wireline-Deployments-Final-111825.pdf.
[14] Accelerating Wireless Broadband Deployment by Removing Barriers to Infrastructure Investment et al., Declaratory Ruling & Third Report and Order, 33 FCC Rcd. 9088, ¶ 113 (2018), https://docs.fcc.gov/public/attachments/FCC-18-133A1.pdf.
[15] Build America, supra note 1, ¶ 35.
[16] Gateway Fiber Petition for Preemption and Declaratory Ruling Pursuant to Section 253(d) of the Communications Act, WC Docket No. 26-166, at 3 (June 17, 2026), https://www.fcc.gov/ecfs/document/26109834869/1.
[17] Id. at 2.
[18] Id. at 6.
[19] Id.
[20] Comcast Cable Commc’ns, LLC v. Appalachian Power Co., Memorandum Opinion and Order, Proceeding No. 25-330, Bureau ID No. EB-25-MD-002, FCC 26-6, ¶¶ 26–27 (rel. Feb. 5, 2026), https://docs.fcc.gov/public/attachments/FCC-26-6A1.pdf.
[21] Id. ¶ 2.
[22] Complaint of Comcast Cable Communications, LLC at 18, Comcast Cable Commc’ns, LLC v. Appalachian Power Co., Proceeding No. 25-330 (July 7, 2026), https://www.fcc.gov/ecfs/document/26109998224/1.
[23] Jeffrey Westling, The Cost of Holding Up Broadband, Truth on the Mkt. (July 14, 2026), https://truthonthemarket.com/2026/07/14/the-cost-of-holding-up-broadband.
[24] Comments of the International Center for Law & Economics, Implementation of Section 621(a)(1) of the Cable Communications Policy Act of 1984, as Amended by the Cable Television Consumer Protection and Competition Act of 1992, MB Docket No. 05-311 (June 18, 2019), https://laweconcenter.org/resources/icle-comments-on-implementation-of-section-621a1-of-the-cable-communications-policy-act-of-1984.
[25] See 47 U.S.C. § 542(b) (capping franchise fees at 5% of gross revenue), (g)(1) (defining “franchise fee” to include “any tax, fee, or assessment of any kind”); Implementation of Section 621(a)(1) of the Cable Communications Policy Act of 1984, Third Report and Order, 34 FCC Rcd. 6844 (2019).
[26] See Comments of the International Center for Law & Economics, supra note 24.
[27] See id.
[28] Implementation of Section 621(a)(1) of the Cable Communications Policy Act of 1984, supra note 25.
[29] 47 U.S.C. § 253(a).
[30] See Crown Castle Fiber, L.L.C. v. City of Pasadena, 76 F.4th 425, 433 (5th Cir. 2023).
[31] See Comments of Crown Castle Fiber LLC, Build America: Eliminating Barriers to Wireline Deployments, WC Docket No. 25-253, at 15–16 (Nov. 18, 2025).
[32] Build America, supra note 1, ¶ 68.
[33] See 47 C.F.R. § 76.43 (“A franchising authority may not regulate the provision of any services other than cable services offered over the cable system of a cable operator, with the exception of channel capacity on institutional networks.”).
Source: Federal Communications Commission
I. Introduction and Overview The International Center for Law & Economics (ICLE) is a nonprofit, nonpartisan research centre that applies economic analysis to legal and . . .
The International Center for Law & Economics (ICLE) is a nonprofit, nonpartisan research centre that applies economic analysis to legal and regulatory questions. We appreciate the opportunity to respond to the Competition Bureau’s examination, Behind the Price Tag: Examining the Food Supply Chain.[1] These comments address Question of Interest No. 3, which covers retail pricing and six related sub-questions.
ICLE uses law & economics analysis to determine whether challenged conduct harms consumers by raising prices, reducing output, lowering quality, or diminishing innovation. Our analysis builds on submissions to the Bureau concerning algorithmic pricing,[2] anti-competitive conduct and agreements,[3] merger enforcement guidelines,[4] and the future of Canadian competition policy.[5] We have also studied grocery competition in work on the proposed Kroger/Albertsons merger,[6] state grocery-pricing laws,[7] and online food-delivery fees.[8]
Canadian grocery prices rose sharply after 2021,[9] and Canadians reasonably want to understand why. That concern has focused attention on loyalty programs, pricing algorithms, and less visible changes such as “shrinkflation,” which reduces package size without changing the nominal price, and “skimpflation,” which lowers quality or reformulates a product to reduce costs.
The Bureau should evaluate each practice by its demonstrated effects on competition and consumers, not by whether it is new, automated, or unpopular. That inquiry must account for error costs—the harms caused by mistaken intervention or nonintervention. Wrongly condemning beneficial pricing practices may suppress them across the market, while failing to address harmful conduct may allow it to persist where entry barriers are high or switching is costly. Ambiguous cases therefore require market-specific evidence rather than presumptions based on the form of the conduct.
That caution applies in particular to Section 78(1)(k) of the Competition Act, which identifies “excessive and unfair selling prices” as a potential anti-competitive act without defining either term. High prices alone do not establish exclusion or harm to competition. Treating them as sufficient could turn abuse-of-dominance enforcement into price regulation. The Bureau has recognized that the Competition Act does not authorize it to regulate grocery prices and that price controls are “blunt tools” because authorities cannot reliably identify or continuously adjust the right price.[10]
These comments instead examine the conditions under which consumers can find, compare, and act on prices, and those under which firms can coordinate or exclude rivals. Across Item 3’s six sub-questions, we focus on consumer information, switching costs, entry barriers, deception, coordination, and exclusion. Grocery-price levels reflect broader macroeconomic forces and supply-chain conditions that competition enforcement cannot directly manage. The relevant questions are what conduct occurred, how it changed the choices available to firms and consumers, and what the evidence shows about its competitive effects.
Competition enforcement operates under uncertainty. The error-cost framework, developed by Frank Easterbrook and extended in the modern literature, assesses legal rules by the consequences of mistaken decisions. A false positive, or Type I error, wrongly condemns conduct that benefits consumers or does not harm competition. A false negative, or Type II error, fails to stop conduct that does.[11]
These errors are not symmetric in a low-margin, high-volume industry such as grocery retail. False positives can prove both costly and durable. If enforcement condemns lawful discounting, loyalty pricing, or algorithm-driven responses to market conditions, firms may abandon those practices across their operations to avoid liability. Market forces cannot readily restore the lost benefits. False negatives are more likely to invite a market response because elevated margins attract entry, expansion, and substitution. As Easterbrook observed, a monopolist’s higher prices attract rivalry.
This asymmetry supports restraint in ambiguous cases, not inaction when the evidence clearly establishes harm. False negatives may persist where artificial barriers impede entry or consumers face high switching costs. Enforcers should require market-specific evidence of those conditions rather than presume that they exist.
Bill C-56 (2023) amended the Competition Act’s abuse-of-dominance provisions to identify the imposition of “excessive and unfair selling prices” as an anti-competitive act.[12] Yet the Act defines neither “excessive” nor “unfair,” creating significant interpretive uncertainty. High prices are not inherently exclusionary. They allocate scarce goods and signal opportunities for entry. Deciding that a price is excessive requires judgments about common-cost allocation, the value of intangible assets such as brands and intellectual property, and the appropriate return for risk. Competition authorities cannot make those judgments reliably, particularly for retailers that sell many products and platforms that serve several interdependent groups of users.
The European Union’s experience enforcing Article 102 of the Treaty on the Functioning of the European Union illustrates the problem.[13] In United Brands v. Commission, the European Court of Justice established a two-part test that asks whether the margin between price and cost is excessive and, if so, whether the price is unfair in itself or compared with competing products.[14] Despite that framework, the European Commission rarely brings excessive-pricing cases because measurement remains difficult and the risk of error remains high.[15]
Comparative studies likewise counsel intervention only in exceptional cases. Surveys of EU and national case law, along with Organisation for Economic Co-operation and Development (OECD) roundtables, show that these cases typically involve regulated or quasi-regulated sectors, legal privileges, or earlier exclusionary conduct. Authorities rarely challenge prices as exploitative without some related restraint on competition.[16]
David S. Evans and A. Jorge Padilla,[17] Massimo Motta and Alexandre de Streel,[18] and Lars-Hendrik Röller[19] similarly argue that excessive-pricing claims are most defensible when linked to exclusionary conduct or durable structural barriers that competition law can identify and address. Policing high prices in markets where entry remains feasible creates a serious risk of measurement error and turns competition authorities into price regulators—a role for which they lack both the tools and the institutional mandate.
The Bureau should therefore presume that pricing conduct benefits competition unless exclusionary conduct prevents rivals from entering or expanding. High prices alone should not create liability without evidence that they result from, or reinforce, conduct that substantially lessens competition. The 2025 expansion of private access to the Competition Tribunal heightens the risk by allowing more private parties to bring cases.[20] Canadian law also lacks a screen comparable to the U.S. antitrust-injury doctrine, which requires a private plaintiff to show that its injury flows from harm to competition rather than from competition itself.[21]
The practices covered by the Bureau’s six sub-questions do not have a common competitive effect and require individual assessment. Our response to Sub-Question 1 illustrates the point. Pricing algorithms may automate lawful market monitoring, or they may facilitate coordination by pooling confidential data. Loyalty discounts may expand output, or they may exclude rivals when combined with market power and substantial foreclosure. Sub-Questions 4 through 6 present similar trade-offs. Disclosure may reduce consumers’ search costs, but it may also help competitors detect departures from a coordinated price.[22]
The Bureau should ask what conduct occurred, how it changed the choices available to firms and consumers, and what the evidence shows about prices, output, quality, innovation, entry, and switching. The analysis should turn on the conduct’s mechanism and demonstrated effects.
The Bureau asks which pricing practices, including loyalty programs and pricing algorithms, help or harm competition. Most strengthen competition by improving price discovery, helping supply respond to demand, reducing waste and shortages, and intensifying rivalry for consumers. Algorithms often automate lawful pricing decisions, dynamic pricing helps markets adjust to changing conditions, and personalized prices and loyalty discounts can expand output and deliver genuine savings.
Competitive harm arises through narrower, fact-specific mechanisms. Firms may coordinate through shared confidential data or a common pricing agent, while dominant firms may use exclusive loyalty terms to foreclose rivals. Existing competition law already reaches those cases. The Bureau should target evidence of agreement or exclusion, not price variation, software use, or loyalty programs themselves. Questions about notice, transparency, and data use may warrant consumer-protection scrutiny, but they do not by themselves establish harm to competition.
Pricing algorithms usually automate tasks that human managers have long performed, including monitoring competitors’ public prices, tracking inventory, and estimating how demand responds to price. Airlines and hotels have used software to adjust fares and room rates for decades. Scale and speed have changed, but the underlying activity has not.
Maureen K. Ohlhausen, then acting chair of the U.S. Federal Trade Commission, captured this point with the “guy named Bob” test. If a pricing manager named Bob may lawfully review competitors’ public prices and adjust his firm’s prices, an algorithm may perform the same task.[23] Automating lawful conduct does not make it unlawful. Appendix A provides a fuller discussion of algorithmic pricing and competition.
The empirical evidence confirms that pricing algorithms can produce different effects in different markets. Zach Y. Brown and Alexander MacKay studied high-frequency price data for over-the-counter allergy medication sold online.[24] They found that retailers using algorithms to update prices within hours systematically charged lower prices than competitors that updated prices daily or weekly.
Sophie Calder-Wang and Gi Heung Kim studied rental housing and found that landlords using pricing algorithms reduced rents more quickly during demand downturns, including the Great Recession.[25] Although rents increased more quickly during expansions, the software improved occupancy and helped the market clear when demand weakened.
Stephanie Assad and her co-authors reached a different result in their prominent study of German retail-gasoline stations.[26] They found that algorithmic pricing increased average prices and margins by roughly 15%, but the effects varied sharply with market structure. Adoption had little effect in monopoly markets. In duopoly markets, where two stations competed and both adopted pricing software, margins rose by 38% relative to markets where neither station adopted.[27] The authors called the result “tacit collusion,” reasoning that the algorithms learned not to undercut because competitors would quickly match price cuts.[28]
The evidence, though, indicates conscious parallelism—firms independently making similar decisions while recognizing their mutual interdependence—rather than an agreement. The German stations did not communicate, signal conditional strategies, coordinate through software vendors, agree to use common software or pricing rules, or exchange competitively sensitive information. Each station independently adopted pricing software, and each algorithm independently learned that aggressive price cuts were unprofitable.[29] Similar tools responding to similar market conditions produced parallel outcomes without a meeting of the minds. As the Bureau’s proposed enforcement guidelines correctly recognize, conscious parallelism alone does not violate Canadian competition law.[30]
The 9th U.S. Circuit Court of Appeals applied this distinction in Gibson v. Cendyn. Competing hotels had independently entered vertical software-licensing agreements at different times. The plaintiffs alleged that the agreements collectively produced anti-competitive effects, even though “neither the terms nor the operation of the licensing agreements [were] alleged to have harmed competition by affecting the competitive incentives in the relevant market, nor did the licensing agreements restrain any party’s ability to compete.”[31] The court held that conclusory allegations could not survive even the pleading stage. Plaintiffs still had to allege “a restraint of trade that causes an actual adverse effect on competition.”[32]
The 3rd U.S. Circuit Court of Appeals’ 2026 decision in Cornish-Adebiyi v. Caesars marks the other side of the line.[33] The court reversed a dismissal because casino-hotel guests had plausibly alleged a horizontal price-fixing agreement under Section 1 of the Sherman Act. Competing hotels allegedly delegated room pricing to a shared algorithm, supplied it with real-time nonpublic pricing and occupancy data, and followed its recommendations roughly 90% of the time. The court held that staggered adoption over 14 years affected the weight of the evidence, not whether the allegations supported an inference of conspiracy.
The legal question is whether firms agreed to coordinate. Parallel price movements alone do not establish that consensus. The decisive allegation in Cornish-Adebiyi was that competitors pooled confidential, nonpublic data through a shared pricing agent—a mechanism capable of converting an ordinary competitive input into a coordination device.
A single fixed price can be too high during slack periods, leaving willing buyers unserved, and too low during peak periods, producing shortages and long waits. Dynamic pricing—prices that change as supply and demand change—can lower off-peak prices to encourage consumption and raise peak prices to allocate scarce supply to those who value it most.
The practice long predates modern pricing algorithms. Airlines have managed seat availability against forecast demand since Ken Littlewood developed his booking rule in 1972. By 1992, American Airlines estimated that its yield-management program, which adjusted fares and seat availability based on expected demand, had generated $1.4 billion in quantifiable benefits over three years.[34]
Evidence from airline markets shows that dynamic pricing can expand output and increase total welfare. Kevin R. Williams found that it increased output by 2.7% and total welfare by 1% relative to charging one price per flight. Dynamic fares could fall for early, price-sensitive travellers and adjust when demand surged. They also reduced stockouts. Under uniform pricing, 29.4% of flights sold out before departure, compared with 18.7% under dynamic pricing, and uniform-price flights sold out earlier.[35]
An efficiency gain does not mean every buyer pays less or receives the benefit. In Williams’ study, airline revenue increased by 7.6%, while total consumer surplus fell by 6.3% relative to uniform fares. Airlines captured much of the gain. Early-booking leisure travellers generally benefited, while late-booking business travellers paid more.
The evidence from ride-hailing is more favourable to consumers. Juan Camilo Castillo found that surge pricing in Houston’s Uber market increased total welfare by 2.15% of gross revenue, or roughly $0.25 per trip, relative to uniform pricing. Riders at every income level benefited through lower average prices, shorter pickup times, and more reliable trips, with low-income riders gaining the most.[36] The gains were not universal. Driver surplus fell by 0.98% of gross revenue, leading Castillo to conclude that drivers “might be right to complain.”
Andrew Sweeting documented the same basic logic in another market for a perishable product: Major League Baseball tickets, which lose all value once a game begins. Sellers reduced prices by 40% or more as game time approached, and the ability to reprice increased the average seller’s expected payoff by roughly 16%.[37]
Grocery shoppers encounter the same practice when stores mark down bakery items, meat, and produce approaching their sell-by dates. Grocery pricing remains much closer to uniform pricing than to surge pricing. Its principal dynamic elements—perishable-goods markdowns and discounts that respond to demand—help sell inventory that might otherwise spoil or remain unsold.
The relevant comparison is therefore not always a fixed price. It may be the nonprice rationing that replaces flexible pricing, including stockouts, queues, spoilage, and unused capacity. Time-varying prices are familiar even for regulated essential services. In the PowerCentsDC experiment, established through a settlement approved by the District of Columbia’s utility regulator, households facing dynamic electricity rates produced “stable, predictable, and sizable demand reductions” during critical peak periods.[38]
Price increases during periods of scarcity can still raise legitimate questions about transparency and notice. Those are consumer-protection questions about how firms disclose prices, and consumer-protection tools can address them. They do not establish that adjusting prices to market conditions harms competition.
Pricing practices that may harm competition operate through different mechanisms, including prices tied to a shopper’s identity and algorithmic coordination among rivals. We address the first in the next subsection and the second in Appendix A (ICLE comments to the Bureau on algorithmic pricing and competition).
Personalized pricing means offering different prices to different customers based on data-driven estimates of their willingness to pay. A targeted discount to a price-sensitive shopper who would otherwise buy nothing expands output and can benefit both the seller and the shopper. Brian Albrecht shows that, under competition, firms minimize consumer prices when they can identify customers who view their products as close substitutes because that information intensifies competition for those customers.[39] An empirical study of platform personalization similarly found that 62.5% of consumers would benefit. Losses were concentrated among the least price-sensitive customers, and most of the aggregate consumer harm arose from the platform’s market power rather than personalization itself.[40]
Price differentials are also common in competitive markets, so their presence in grocery retail does not show that competition has failed. James Dana demonstrates that these discount patterns can arise among firms that earn no economic profit and lack power to set the market price. As he observes, price discrimination “is common in many industries that appear to be extremely competitive.”[41]
Empirical studies reach the same conclusion. Joanna Stavins found that airline discounts tied to ticket restrictions were largest on the least concentrated routes: $253 at the 25th percentile of route concentration, compared with $165 at the 75th percentile. She cautioned that her sample covered only 12 routes.[42] Using a much larger sample, Severin Borenstein and Nancy Rose described competition’s significant positive effect on price dispersion—the variation in prices paid by different customers—as their strongest result.[43]
The Bureau also has evidence from Canada. Using Air Canada fare data, Ambarish Chandra and Mara Lederman of the University of Toronto found that competition reshapes price differentials rather than eliminating them. Each additional rival reduced fares in the middle of the distribution by 7% to 8% while barely changing the cheapest and most expensive fares. Some price gaps therefore narrowed while others widened, making aggregate measures of price dispersion potentially uninformative about how competition affects prices.[44]
The price differentials that persisted under rivalry reflected brand loyalty, which raises questions about consumer switching rather than market power. Canadian grocery shoppers do switch. The average Canadian regularly shops at 2.3 grocery stores, and multihoming—shopping across multiple stores and retail formats—is the industry norm.[45]
Businesses also appear to use personalized pricing rarely and on a limited scale. The European Commission conducted a mystery-shopping study across 160 e-commerce websites in eight member states. It found no price difference in 94% of 34,403 matched product pairs. Among the remaining 6%, the median difference was less than 1.6%.[46] Reputational concerns help explain firms’ caution. The OECD observes that businesses may avoid personalized pricing to protect their reputations and risk stigma when they adopt it.[47] We address the broader “surveillance pricing” narrative, and the gap between technical capability and observed conduct, in Sub-Question 3.
The Instacart episode illustrates how observers can mistake randomized price testing for personalized pricing—and how that mistake can produce legislation. A December 2025 study by the Groundwork Collaborative and Consumer Reports found that shoppers sometimes saw different prices for the same item at the same store and time. For example, a dozen Lucerne eggs at a Washington, D.C., Safeway appeared at five prices ranging from $3.99 to $4.79.[48]
The study’s own regressions undermine the claim that Instacart profiled individual shoppers. Across 193 cleaned submissions, price differentials associated with shoppers’ Instacart histories and demographic characteristics—including age, income, gender, and race—were “small and not statistically meaningful.” The study found no evidence that Instacart based prices on income, ZIP code, shopping history, or other individual characteristics.
The observed pattern instead resembles A/B testing, in which a company randomly assigns customers to groups to compare their responses to different prices. A random process, not a customer profile, determines the price each shopper sees. Instacart’s explanation that it tested randomized cohorts based on product category and location fits the study’s data.[49] The vendor’s marketing materials establish that its technology can target individual shoppers, but capability does not establish use. As Albrecht concludes, the episode “is a story about something that might happen, not something that did happen.”
Labelling the practice “surveillance pricing” assumes what the evidence failed to establish. Even scholars offering cautionary accounts of surveillance pricing recognize “complex trade-offs between efficiency gains and consumer welfare,” rather than uniform harm.[50]
The legislative response shows the cost of that mistaken inference. Maryland enacted its Protection from Predatory Pricing Act in April 2026 after the Instacart narrative dominated legislative debate. The law prohibits personalized prices based on a consumer’s personal data but exempts loyalty programs, memberships, subscriptions, and prices offered to consumers who consent to provide personal data. It therefore does not reach the A/B testing that prompted the debate.[51] When Colorado’s governor vetoed a similar bill, he identified the problem directly: It “would punish differentially lower prices, not just higher prices.”[52]
The United Kingdom’s Competition and Markets Authority (CMA) conducted a rigorous recent review of grocery loyalty pricing. It analysed a year of daily prices covering roughly 30,000 of the 86,000 loyalty promotions offered during that period, encompassing about 50,000 products at five supermarkets. These dual-price promotions offered one price to loyalty members and another to nonmembers.[53]
The CMA found that 92% of the products offered members genuine savings against the usual price. It applied a concrete test: The nonmember price during the promotion had to be the same as, or lower than, the price available to all shoppers during the preceding 28 days. The CMA found “very limited evidence of price changes which could indicate that supermarkets may have inflated prices to make their loyalty prices appear misleadingly attractive.” Across the five retailers, average member savings ranged from 17% to 25% of the nonmember price.
The pattern most likely to raise concern—a nonmember price higher than the price charged both before and after the promotion—appeared in only 0.6% of the roughly 50,000 products analysed. Other promotions in the surrounding days explained many of those cases. The CMA concluded that the potentially concerning examples “do not represent a systemic problem or suggest retailers are trying to mislead shoppers.”
The review also tested the suspicion that supermarkets fund loyalty discounts by inflating their regular prices. The data did not support it. Loyalty membership is nearly universal in the United Kingdom: 97% of shoppers belong to at least one scheme, and members hold three memberships on average. Yet 55% of surveyed shoppers believed that nonmember prices during loyalty promotions exceeded the usual price. That figure rose to 71% among shoppers aged 18 to 29. The CMA found that pattern in only 0.6% of the products analysed—roughly six in every 1,000.
Cross-retailer comparisons produced similar results. Of the 429 branded products the CMA compared, the loyalty price was the cheapest or tied for the cheapest available price for 368, or 86%. When compared with another retailer’s nonpromotional price, 52% of loyalty prices were cheaper, usually by 11% to 26%. Shoppers should still compare prices across stores. For 61 products, or 14%, the cheapest available price was not a loyalty price. That caveat supports better cross-retailer comparison tools, which we address in Sub-Question 6. It does not make the case against loyalty pricing.
Loyalty pricing is a longstanding feature of competitive markets, not a novel algorithmic harm. Frequent-flyer programs were already distinguishing loyal customers from occasional travellers in the deregulated U.S. airline industry during the 1980s.[54]
The CMA identified two practical caveats. First, some supermarkets could do more to ensure that shoppers without smartphones and those younger than 18 can access loyalty prices or understand how to obtain them. That calls for better program design, not abandoning loyalty pricing. Second, the CMA found no consumer-law concerns arising from supermarkets’ collection and use of personal data. Only 7% of nonmembers cited data concerns as their reason for not joining. The regulator that examined loyalty data most closely found no actionable evidence.
Genuine competitive harm from loyalty programs takes a different form: A dominant firm may impose exclusive terms that foreclose rivals or raise switching costs without providing offsetting value. We address those circumstances in “The Narrow Harm Cases” below, including our amicus brief in the EpiPen litigation. That brief explains why a legal test that cannot distinguish procompetitive discounting from exclusion will chill the discounting itself.
Three fact patterns can produce genuine competitive harm. Each depends on evidence of agreement or exclusion, not the presence of software or a loyalty card.
First, competitors may coordinate through a common vendor whose algorithm prices each firm’s products using a pool of rivals’ competitively sensitive data. That conduct differs from each firm independently using its own information and publicly available data. It raises a conspiracy question under section 45 of the Competition Act, whose criminal prohibition targets “naked restraints”—agreements such as price fixing that do not support a legitimate collaboration—and, on the civil side, under section 90.1.[55]
The Bureau has already drawn the relevant line. Its Competitor Collaboration Guidelines state that, “[i]n general, the Bureau does not consider publicly available information to be competitively sensitive.” Its proposed 2025 enforcement guidelines identify the common-vendor concern more specifically. If a third-party developer’s “algorithm uses commercially sensitive information provided to the developer by other firms,” the arrangement “could lead to coordinated outcomes without any direct information sharing between industry participants.”[56]
Second, a hub-and-spoke arrangement may include a “rim”—a horizontal agreement among competing firms to adopt and follow a common vendor’s recommendations. The rim distinguishes a conspiracy from a series of independent vertical contracts between the vendor and each firm. This, too, raises a section 45 question. The Competitor Collaboration Guidelines ask whether the parties reached a “meeting of the minds.” Conscious parallelism alone does not establish an agreement, although “parallel conduct coupled with facilitating practices, such as sharing competitively sensitive information,” may prove one.[57]
Third, a dominant grocer may impose exclusive loyalty terms that foreclose rivals or raise switching costs without providing offsetting value. That conduct may constitute abuse of dominance under sections 78 and 79. The Act’s enumerated anti-competitive acts include inducing a supplier to refuse to sell to a rival to impede its entry or expansion, and selling below acquisition cost to discipline or eliminate a competitor. Section 79 also requires that the effect on competition “is not a result of superior competitive performance.”[58]
The Bureau’s proposed guidelines take the same balanced approach. They identify volume and loyalty discounts as possible means of inducing exclusivity but recognize that “[e]xclusive dealing is not necessarily anti-competitive.” The concern arises when exclusivity forecloses rivals by denying them the scale, inputs, or sales channels needed to compete.[59] In each case, the agreement or exclusion creates the potential harm. The software or loyalty card does not.
The Bureau’s guidance already distinguishes these cases from lawful conduct, and U.S. algorithmic-pricing decisions show that courts can administer the distinction. Those decisions do not bind Canadian tribunals, but they help identify the allegations that can make a claim actionable.
In Gibson v. Cendyn, in which ICLE participated as amicus curiae, the 9th U.S. Circuit Court of Appeals explained that competing hotels could violate Section 1 of the Sherman Act by agreeing among themselves to follow a third party’s pricing recommendations. The pleaded claim failed because the software did not share one hotel’s confidential information with competing licensees. The court noted that allegations of such sharing could change the analysis.[60]
The district court reached a similar initial result in Cornish-Adebiyi v. Caesars, but the 3rd U.S. Circuit Court of Appeals reversed in July 2026. The appellate court read the complaint as alleging that each casino hotel supplied current, nonpublic room-pricing and occupancy data to a common platform, which processed those data alongside competitors’ information. Taken with the complaint’s other allegations, that pooling sufficed to plead a conspiracy.
In re RealPage involved the same critical allegation: The vendor pooled clients’ commercially sensitive pricing and supply data before returning pricing recommendations.[61] The relevant questions here concerned the flow of confidential data and any agreement to follow the common recommendation. A firm does not enter into a conspiracy merely by independently adopting software that relies on its own information and public data.
The Act also protects low prices because competition law seeks to promote them. An effects-based order under section 79 requires proof that the effect on competition “is not a result of superior competitive performance.” A lower price achieved through scale or efficiency reflects superior competitive performance, not abuse.[62] The Bureau’s proposed guidelines state the presumption directly: “Low prices are usually a sign of vigorous competition, which the Act is designed to encourage.”[63]
As ICLE and allied scholars argued as amici in the EpiPen litigation, loyalty discounts and the lower prices they produce routinely benefit consumers. A legal test that cannot distinguish procompetitive discounting from exclusion will chill the discounting itself.[64] A workable test should instead require objective evidence of foreclosure, including the share of the market covered and the duration of the challenged terms. Every sales contract “forecloses” the sales it wins, while short-term commitments that customers can easily terminate give rivals recurring opportunities to compete for their business.[65] A test based on the size of the savings rather than the extent of foreclosure condemns the discount itself—the costliest false positive in a low-margin industry.
The Bureau asks how these pricing practices affect small and medium-sized enterprises (SMEs). Modern pricing technology generally strengthens their ability to compete. Widely available third-party tools give smaller firms pricing capabilities once limited to large incumbents, reduce barriers to entry, and support new business models.
The greater risk comes from overbroad regulation. Same-terms mandates would erase legitimate cost efficiencies and raise consumer prices, while disclosure and documentation requirements impose fixed costs that fall most heavily on independent grocers. The Bureau should preserve smaller firms’ access to competitive tools and avoid rules that favour businesses large enough to absorb the cost of compliance.
Sophisticated pricing strategies once required extensive data collection, specialized analytics, and continuous monitoring. Large incumbents could afford those resources, giving them a pricing advantage over smaller rivals. Widely available third-party pricing services now give firms of any size access to tools once available only through an internal analytics department. New entrants and smaller firms can adopt advanced pricing techniques without developing them in-house.
These services narrow the pricing-capability gap. A startup can use the same revenue-management software as an industry leader, reducing the risk that it will misprice its products or leave revenue unrealized. Algorithms thus reduce large firms’ scale advantage in pricing by allowing smaller firms to purchase sophisticated analysis through a vendor. The OECD has recognized this benefit, noting that algorithms can “reduce barriers to entry by allowing smaller entrants to gain market insights or develop new disruptive products at lower cost.”[66]
Algorithms can also support entry through new business models. Uber and other ride-hailing platforms built dynamic pricing into their services. By raising prices when demand surged, the platforms attracted more drivers and helped match supply with demand in real time. This addressed the peak-period shortages common in traditional taxi markets and helped establish a new market for on-demand rides. Consumers can now find rides more reliably at almost any time.
Had regulators prohibited surge pricing from the outset as “price gouging,” the platform model might not have attracted enough drivers or proved viable. Algorithms also enable products such as real-time price-comparison services, personalized shopping assistants, and dynamic-discount apps. These tools increase transparency, help consumers find better offers, and create new ways for firms to compete.
Rules requiring suppliers to offer every buyer the same wholesale terms would eliminate cost advantages associated with large-format retail. Emek Basker’s review of Walmart’s role in grocery markets found that its grocery prices averaged about 10% less than competitors’ prices. A Walmart Supercenter’s entry typically prompted competing grocery stores to reduce prices by roughly 1% to 2% in the short run. Over four years, the expansion of superstores, mass merchandisers, and warehouse clubs led traditional supermarkets to reduce prices by an estimated 3%.[67] Walmart’s presence therefore lowers prices for its customers and for consumers who continue shopping at competing stores.[68]
Using household panel data, Jerry Hausman and Ephraim Leibtag similarly estimated that the expansion of large-format supercentres produced substantial consumer-welfare gains. Those gains disproportionately benefited low-income households, which shop at price-sensitive formats more often than higher-income households.[69] A same-terms rule that erodes large retailers’ cost advantages would therefore raise prices most for shoppers with the fewest alternatives.[70]
Grocery prices already display substantial uniformity across stores. A same-terms rule would entrench that condition rather than correct it. Stefano DellaVigna and Matthew Gentzkow found that most large U.S. food, drugstore, and mass-merchandise chains charged nearly uniform prices across locations despite wide differences in local demographics and competition. The median chain sacrificed an estimated $16 million in annual profit by doing so.[71]
Uniform pricing is not the consumer-friendly benchmark it may appear to be. DellaVigna and Gentzkow’s model indicated that uniform pricing raised prices in the poorest decile of ZIP codes by 10% relative to the richest decile. Alberto Cavallo found similar uniformity across sales channels. Among 56 large multichannel retailers in 10 countries, online and offline prices were identical about 72% of the time overall and 52% of the time in food retail. Across Canadian sectors, they matched more than 90% of the time.[72] Requiring still greater uniformity would eliminate the price variation that most benefits low-income shoppers.
Disclosure and documentation rules designed for national platforms would burden small grocers disproportionately because compliance costs do not fall with transaction volume. Requirements for transaction-specific cost records or custom disclosure formats impose fixed costs that a large chain can spread across billions of dollars in sales but a small independent grocer cannot. Our comments to the U.S. Federal Trade Commission on online food-delivery fees examine this asymmetry in detail. They explain that the burden falls most heavily on small first-party operators—businesses that run their own delivery websites.[73]
The Bureau asks whether access to data and technology affects firms’ ability to adopt modern pricing practices and compete. In grocery retail, both inputs are widely available at modest cost. Third-party services give smaller firms sophisticated pricing capabilities, while grocery data depreciate quickly, lack strong network effects, and remain available from multiple commercial sources. Common software also does not imply pooled data. The current record does not show that data or pricing technology constitute scarce assets that foreclose rivals or entrench incumbents.
Regulatory intervention may create the greater competitive risk. Mandated data sharing and disclosure of confidential terms can facilitate coordination, weaken incentives to collect and analyse data, and impose disproportionate compliance costs on smaller firms. Claims that retailers use personal data to charge each shopper the most they will pay likewise require evidence, not inferences drawn from randomized price testing. The Bureau should distinguish firms’ use of their own or public data from the pooling of competitors’ confidential information, and leave genuine privacy concerns to privacy and data-protection law.
In its parallel consultation on algorithmic pricing, the Bureau observed that “[u]sing common pricing algorithms or pooling data among competitors may raise issues under the Competition Act.”[74] That concern is valid in principle but narrow in practice. The risk arises when third-party software ingests one firm’s confidential data and transmits it to a competitor, either directly or through recommendations based on pooled inputs. The common algorithm then becomes a conduit for competitively sensitive information.
Ordinary pricing software does not operate this way. Firms may independently use similar or even identical programs without pooling any data. A pricing algorithm, whether developed in-house or licensed from a vendor, generally relies on two types of information. The first consists of the firm’s commercially sensitive data, including sales volumes, inventory, and costs. Vendors can process that information while keeping it segregated from rivals. The second consists of lawfully collected public information, such as advertised prices and general market trends.
A hotel using revenue-management software, for example, may receive analysis of its own booking data, demand trends, and competitors’ advertised rates from public travel websites. It does not ordinarily receive a direct feed of a rival’s proprietary booking data. Reputable vendors design their systems to prevent the interfirm transfer of sensitive information because sharing it would expose their clients to liability.[75]
Monitoring competitors’ public prices is legitimate competitive conduct, whether performed manually or by software. Automation does not convert lawful information gathering into unlawful coordination. The Bureau’s Competitor Collaboration Guidelines generally treat public information as unproblematic and reserve concern for exchanges of competitively sensitive, nonpublic data.[76] They also distinguish direct exchanges between competitors, which present greater risk, from aggregated data supplied through an independent third party.
The controlling question is whether firms have moved beyond information gathering and agreed to coordinate their competitive decisions. Common analytical tools, similar market intelligence, and parallel outcomes do not establish such an agreement. An enforcer must still prove one.
Access to pricing data therefore does not ordinarily mean access to a rival’s secrets. It means access to a firm’s own records and publicly available market information. A policy that assumes algorithmic pricing necessarily pools competitors’ data would mischaracterize the technology and discourage lawful monitoring that helps discipline prices.
If access to technology affects firms’ ability to compete, pricing technology has broadened that access. This is the access dimension of the argument developed in Sub-Question 2, and it bears directly on the Bureau’s inquiry.
Sophisticated pricing once required scale. Firms had to collect data, acquire analytics expertise, and monitor markets continuously. Only large incumbents could justify those fixed investments, giving them a durable advantage over smaller rivals. Third-party pricing services have sharply reduced that advantage. They give a new entrant or small independent access to the revenue-management capabilities of an industry leader without requiring an in-house analytics department.
In economic terms, these services reduce the pricing advantages associated with firm size. A small firm can obtain sophisticated market analysis through a vendor’s platform. The OECD has recognized this benefit, observing that algorithms can lower barriers to entry by helping smaller firms gain market insights and develop new products at lower cost.[77]
Adoption data indicate broad diffusion rather than concentration. A 2020 survey by the Netherlands Authority for Consumers and Markets found that 80% of firms using pricing algorithms had developed them internally, while 20% relied on third-party providers.[78] In 2017, the European Commission found that 49% of European retailers monitored competitors’ prices. Of those retailers, 66.6% used price-monitoring software, meaning roughly one-third of all retailers used some form of algorithmic monitoring.[79] Among firms using monitoring software, 78% changed prices in response to competitors’ moves, and 35% used specialized algorithms to adjust prices automatically.[80] Surveys elsewhere in Europe show similar adoption patterns.[81]
Accessible pricing technology has also enabled new competitors and business models, including dynamic-pricing platforms, real-time price-comparison services, and discount apps. These products increase transparency and help consumers find lower prices. The competitive pressure they place on incumbents—often over incumbents’ objections—shows that the market remains open to entry.
The principal access risk comes from regulation itself. Compliance does not scale down. Disclosure, documentation, and audit requirements impose fixed costs that a national chain can spread across billions of dollars in sales but a small grocer cannot. Rules designed for the largest platforms could rebuild the entry barriers that accessible technology has lowered. We address this asymmetry in Sub-Question 2 and in our comments to the U.S. Federal Trade Commission on online food-delivery fees.[82] The Bureau should preserve pricing technology’s equalizing effect by avoiding compliance mandates that fall hardest on the smallest firms.
A recurring theory treats large grocers’ accumulated transaction and loyalty data as a barrier to entry. On this account, incumbents use a self-reinforcing data advantage to price and target customers in ways that rivals cannot match. That theory fits grocery retail poorly for three reasons. The Bureau should require evidence of the mechanism rather than presume it.
First, grocery transaction data depreciate quickly and produce diminishing returns. Prices, promotions, product assortments, and consumer preferences change continuously. Each additional historical observation therefore adds less value, while stale data offer little predictive insight. A firm does not need every household’s complete purchase history to price competitively. It needs a representative and reasonably current picture of the market, which costs far less to obtain than the incumbent-scale archive assumed by the barrier theory.
Second, grocery retail lacks the demand-side network effects that can entrench data-driven platforms. Such effects arise when each additional user makes a service more valuable to other users. One shopper’s choice of a grocery chain does not make that chain more valuable to the next shopper. Grocery data therefore do not create a self-reinforcing loop that locks consumers into one retailer and denies rivals the scale needed to compete. Data accumulate, but they do not produce the winner-take-most dynamic found in markets where an entry-barrier theory is more plausible.
Third, the relevant data are neither exclusive nor unavailable to entrants. Consumers shop across supermarkets, supercentres, warehouse clubs, discount grocers, dollar stores, and online retailers. No single retail banner can see a household’s entire grocery basket.[83] Competition therefore spans formats well beyond the local supermarket.
Entrants can also purchase comparable market intelligence. Commercial sources offer scanner data, consumer-panel data, third-party analytics, and audience information. An entrant can buy much of the information that an incumbent collects internally. Data that are nonexclusive, rapidly depreciating, and available from multiple suppliers do not constitute an essential facility—an indispensable input that rivals cannot reasonably obtain elsewhere. Treating them as one would risk the durable false positive that the error-cost framework counsels against.
This point bears only on the Bureau’s data-access question. It is not a merger analysis. Our Kroger/Albertsons work addresses merger-specific concerns on their own terms.[84] The narrower conclusion here is that the current record does not show that access to grocery data creates a barrier requiring intervention.
The sub-question focuses on whether some firms have too little access to data. A more serious risk could arise from rules that compel firms to share data or disclose confidential terms in the name of equal access or “transparency.” The Bureau should distinguish consumer-facing transparency from competitor-facing transparency.
Consumer-facing transparency—including clear final prices, unit prices, and legible loyalty terms—reduces search costs and helps shoppers switch. Competitor-facing transparency can have the opposite effect. Mandating disclosure of supplier contracts, rebates, promotional allowances, or forward-looking pricing information can help rivals detect and punish departures from a coordinated price.
Tacit coordination rarely breaks down because firms cannot identify a common price. It breaks down because secret discounting makes deviations difficult to detect and punish. Anonymising disclosed terms does not eliminate that risk in a market with only a few large buyers. Volumes, shipping patterns, and timing may reveal the parties’ identities, and modern analytical tools make reidentification easier. Recent research shows that large language models (LLMs) can reidentify individuals or entities in ostensibly anonymised text by extracting identifying features and matching them with external data.[85]
The risk increases when disclosure rules cover only a small number of retailers. Even if supplier contracts omit retailers’ names, market participants may infer which chain received which terms. Disclosure could then make prices more uniform and higher. We develop this mechanism, including its interaction with statutory most-favoured-nation and same-terms rules, in Sub-Question 6.[86]
The same caution applies to data-sharing and data-portability mandates intended to “level” competitors’ access. Beyond facilitating coordination, compelled sharing weakens incentives to collect and analyse data. It may therefore degrade the very capability the mandate seeks to spread. Any such rule would also rest on the unsupported theory that grocery data create a barrier to entry, discussed in the preceding subsection. It would impose immediate, durable costs to address a harm that the evidence has not established—a strong case for error-cost restraint.
Much of the concern about data access rests on an unsupported narrative: Retailers use consumers’ personal data to charge each shopper the most they will pay. The episode that drove this narrative shows how misreading a study can produce poorly targeted law.
In December 2025, the Groundwork Collaborative and Consumer Reports published a study showing that Instacart shoppers sometimes saw different prices for the same item at the same store and time.[87] The study expressly found no evidence that Instacart based prices on individual characteristics such as income, ZIP code, or purchase history. The observed pattern instead matched routine A/B testing, in which random assignment determines the price a shopper sees.[88]
Unlike personalized pricing, A/B testing does not estimate what a particular shopper will pay. It assigns prices across randomized groups to learn about demand without relying on the shopper’s identity. Yet observers recast the study’s unexplained price variation as evidence of personal-data profiling.
The legislative response extended far beyond that finding. Lawmakers have introduced more than 40 bills addressing algorithmic or “surveillance” pricing across at least two dozen U.S. states, and Maryland enacted the first statute of its kind.[89] That law illustrates the risks of mismatched drafting.[90]
The statute’s title invokes predatory pricing—a settled term for below-cost pricing intended to exclude rivals—but its operative provisions address an unrelated practice by prohibiting prices personalized with consumers’ data. The law then exempts loyalty programs, memberships, subscriptions, and prices offered with consumer consent. It does not reach the A/B testing that prompted the debate. The triggering study did not find the practice the law targets, the statute’s title describes a different practice, and its exemptions sharply limit its reach. That is legislation shaped by alarm rather than measured effects.
Two facts should constrain any similar response in Canada. First, personalized pricing remains uncommon and modest in scale. The European Commission’s mystery-shopping exercise found price differences in only about 6% of matched tests, with a median difference below 1.6%. Those differences concentrated in industries such as airlines and hotels, where price discrimination is well established.[91]
Second, concerns about how businesses handle personal information fall primarily within privacy and data-protection law. Treating privacy concerns as competition problems produces rules poorly suited to either task. The Bureau should require evidence that a data practice harms competition or consumers before intervening and leave genuine privacy questions to the legal framework designed to address them.
The Bureau asks how the pricing practices at issue affect consumers’ ability to compare prices and value across retailers and their willingness to switch. Canadian shoppers actively compare offers, shop across multiple formats, and respond to price. Digital tools have reduced the time and effort required to compare retailers, while the pronounced shift toward discount stores provides direct evidence of price-driven switching. Loyalty pricing, dynamic pricing, and personalized pricing can make some comparisons more complex, but the evidence does not show that these practices suppress switching. Loyalty programs also lower effective prices and can intensify competition for shoppers.
The proportionate response is consumer-facing disclosure, including harmonized unit pricing, clear final prices, and legible loyalty terms, as discussed in Sub-Question 6. Canada’s concentrated grocery market and limited independent hard-discount competition raise separate concerns because major grocers own many discount banners. Those are supply-side questions about entry and expansion, which the Bureau’s 2023 market study addressed through structural recommendations. They do not establish that pricing practices prevent consumers from comparing prices or switching retailers.
Most Canadians buy groceries from five companies—Loblaws, Sobeys parent Empire, Metro, Costco, and Walmart—which together account for roughly two-thirds or more of national grocery sales.[92] The Bureau has found that concentration has increased in recent years[93] and that Canada has little independent hard-discount competition.[94] he degree of concentration remains contested. The Retail Council of Canada argues that Canada is not concentrated relative to peer countries, citing the rising shares of Costco and Walmart and the large majority of Canadians who regularly shop at discount retailers.[95]
When Canadians trade down to a discount grocer, they often remain within the same corporate family. Industry observers describe discount banners as a way to keep deal-seeking customers “within [the] estate” rather than lose them to a rival.[96] Still, roughly 6,900 independent grocers operate across Canada, and larger regional chains compete directly with national firms in some markets. The Bureau has found, however, that independents face significant barriers to becoming national competitors.[97]
Digital tools have reduced the cost of comparing prices. Weekly flyers, retailer apps, online storefronts, and third-party comparison services allow shoppers to survey competing offers before leaving home. That visibility disciplines attempts to price above competitive levels because consumers who see better alternatives can act on them.
Canadians regularly use that capacity. The Bureau’s survey found that 79% visit multiple grocery stores to find the best price at least sometimes. That figure includes 41% who do so sometimes, 21% often, and 17% very often.[98]
The clearest evidence of price-driven switching is the shift toward discount stores. The discount channel’s market share rose from roughly 35% in 2008 to near parity with conventional stores. Discount sales continue to grow faster than conventional sales by both dollars and volume.[99]
Online shopping provides another low-friction channel for comparison. About 30% of Canadians used at least one online option to buy groceries during a recent three-month period. That included 17% who used third-party delivery, 12% who used curbside pickup, and 9% who used home delivery.[100] Switching costs in these closely related channels can approach zero, as shoppers can open a competing app and compare offers side by side within seconds.
Individualized or conditional prices and tiered loyalty offers can complicate the comparisons made easy by flyers and shelf tags. The proportionate response is to improve comparison through clear final-price disclosure and, above all, unit pricing, which allows shoppers to compare prices per standard unit. This approach follows Recommendation 3 of the Bureau’s 2023 study, prompted by its finding that consumers have difficulty comparing even a few items across grocery stores.[101]
Loyalty programs play a central role in Canadian grocery competition. About 75% of Canadians hold a loyalty or points card, and 61% say they are more likely to shop where they can earn rewards.[102] These programs affect switching in both directions. Points, targeted offers, and rewards earned across banners within the same corporate group make staying with one program more convenient, creating some switching friction. At the same time, they lower members’ effective prices. The Bureau has recognized that loyalty programs “can bring about lower prices for consumers” and “may also drive grocery stores to compete harder” for customers.[103]
The CMA’s 2024 loyalty-pricing review provides the most rigorous recent evidence. Of roughly 50,000 products offered through dual-price loyalty promotions, 92% delivered genuine savings. Average member savings ranged from 17% to 25%, and the CMA found very limited evidence that retailers inflated ordinary prices to make loyalty offers appear more attractive.[104]
Canadian consumers continue to compare prices and switch retailers. The pricing practices at issue have not prevented either behaviour. Switching friction alone does not establish competitive harm without evidence of deception, lock-in, exclusion, or market power. Restricting loyalty, dynamic, or personalized pricing based on an unsupported switching-harm theory would weaken the price competition that Canadians are driving through their shift toward discount retailers.
The Bureau should instead support consumer-facing tools that directly reduce comparison costs, including harmonized unit pricing, clear final prices, and legible loyalty terms. A wider range of competitors will strengthen comparison and switching more durably. Those competitors should remain free to use pricing tools that help them compete.
The Bureau asks how shrinkflation and skimpflation affect consumers’ ability to compare price and value, and whether they influence purchasing and switching. Shrinkflation reduces a product’s quantity while holding its shelf price steady. Skimpflation reduces quality or changes a product’s formulation to cut costs. Both are rational nonprice responses to rising input costs, and both occur in competitive and concentrated markets. Neither practice alone reveals the level of competition.
The principal competition concern is comparability. Shrinkflation can succeed because consumers notice price changes more readily than changes in package size. Standardized unit pricing addresses that problem more directly than controls on product design.
Aljoscha Janssen and Johannes Kasinger provide the most comprehensive evidence to date. They analysed a decade of U.S. store-level scanner data and found that downsizing was more than five times as common as upsizing on a sales-weighted basis. Size reductions “usually occur with little or no accompanying price changes,” while the price per unit of volume averaged roughly 12% more during the year after a reduction.[105] A shopper can accurately read every posted price and still miss that a product has become more expensive per gram.
That evidence does not prove market power or “gouging.” Resizing is a predictable response to rising marginal costs, while broad increases in the price level chiefly reflect macroeconomic forces rather than the conduct of one firm. The evidence also does not establish the motive for any particular reduction. Janssen and Kasinger describe downsizing as a way for firms “to increase margins or respond to cost pressures.”[106] The comparability problem, and its appropriate remedy, remain the same in either case.
Existing Canadian cartel law addresses agreements among competitors to reduce package size or quality.[107] The best-documented shrinkflation case involved the three largest U.S. canned-tuna producers, which agreed to reduce the standard can size from six ounces to five ounces beginning in 2008. The U.S. Department of Justice brought criminal charges. Two of the three producers had pleaded guilty by 2018, and a chief executive received a prison sentence.[108]
Skimpflation presents the same comparison problem but is even less visible. Consumers may struggle to detect a quality change at the shelf, and unit pricing cannot capture it. The systematic evidence remains thin. The term arose in popular reporting about declining service quality, and grocery products have no evidence base comparable to the shrinkflation literature.
Existing safeguards address the clearest harms. Deceptive-marketing law reaches false or misleading claims about quality and net quantity, while repeat purchasing disciplines quality in products consumers buy weekly. Treating every reformulation as suspect would instead subject routine cost-saving product changes and genuine quality improvements to regulatory review.
The remedy should match the conduct: standardized unit pricing for hard-to-see quantity changes, deceptive-marketing enforcement for false claims, and cartel law for agreements among competitors. Product-design controls would address none of those problems as precisely.
The Bureau asks which measures could improve transparency and competition in retail pricing. The answer depends on the audience. Consumer-facing transparency helps shoppers compare offers and switch retailers. Competitor-facing transparency may expose confidential terms, discourage selective discounts, and facilitate coordination.
We recommend harmonized unit pricing, clear final prices, and legible loyalty terms. The Bureau should reject same-terms mandates, compelled disclosure of confidential contracts, and price or margin controls. Enforcement should target demonstrated agreements, deception, and exclusion while preserving lawful discounting and independent pricing. The most durable reforms operate on the supply side by reducing barriers to entry and expansion, strengthening independent wholesale infrastructure, and widening the range of grocers available to Canadian consumers.
Our lead recommendation is accessible, harmonized unit pricing across every province and territory. Unit pricing displays the price per standard unit of measure, allowing shoppers to compare a 500-gram package with a 750-gram package at a glance. It addresses loyalty tiers, shrinkflation, package-size differences, and in-store and online comparisons without regulating prices.
This approach also follows the Bureau’s 2023 recommendation. The Bureau noted that only Quebec legally requires unit pricing and urged provincial and territorial governments to develop accessible, harmonized requirements.[109]
The rules should establish standard units for each product category and apply in stores, online, and in advertising. Retailers should display unit prices accurately, legibly, and close to the relevant item. Very small retailers could receive phased implementation or simplified compliance requirements. Unit pricing requires no disclosure of wholesale terms, margins, or supplier contracts. It provides transparency to shoppers without revealing competitively sensitive information to rivals.
Two additional consumer-facing measures follow the same principle. First, retailers should clearly disclose final prices and truthfully present discounts and reference prices—the prices used to frame an advertised saving. This would target misleading “was/now” offers and false scarcity claims without restricting dynamic pricing.
Second, retailers should present loyalty-program terms in plain language. Disclosures should compare the member price with the shelf price and clearly explain points’ redemption value, expiry dates, and material exclusions. We discuss those requirements further in Sub-Question 4.
The Bureau should reject same-terms and wholesale price-parity mandates. New York Senate Bill 8563, the proposed “Consumer Grocery Pricing Fairness Act,” offers a recent example. It would have required large suppliers to offer identical terms to retailers purchasing the same volume, capped at one truckload, and to disclose anonymized terms upon request.[110]
Such a requirement operates as a statutory most-favoured-nation clause. A most-favoured-nation clause promises one buyer terms at least as favourable as those offered to another. A privately negotiated clause limits a supplier’s ability to offer selective discounts after making that promise. A statutory mandate goes much further by imposing the obligation across the market. If a supplier must extend every discount to all similarly situated buyers, it has less incentive to discount at all. Suppliers will instead tend toward a uniform, higher wholesale price.[111]
The U.S. experience supports this concern. The Robinson-Patman Act discouraged discounting for decades by exposing sellers to litigation over price differences. In 2007, the bipartisan Antitrust Modernization Commission concluded that the law’s chilling effect on discounts likely raised consumer prices above the levels that would otherwise have prevailed.[112]
Supporters of same-terms rules often invoke the “waterbed effect,” under which a discount to a large buyer supposedly forces the supplier to charge smaller buyers more. Roman Inderst and Tommaso Valletti developed the leading model and derived a sufficient condition under which this effect would harm consumers.[113] Recent research questions whether their model supports that conclusion.[114]
Brian Albrecht shows that the condition for consumer harm is never satisfied in any equilibrium generated by Inderst and Valletti’s own model. Their analysis therefore identifies no equilibrium in which the waterbed effect harms consumers. Consumer surplus instead rises at the margin even when the smaller rival pays and charges more.
Empirically, the U.K. Competition Commission found little evidence of a systemic waterbed effect in grocery mergers, and courts have treated the theory skeptically, as in DeHoog v. Anheuser-Busch InBev, where the court noted that antitrust law protects the squeezed supplier, not the large buyer’s competitors.[115]
Any statutory cost-justification defence would offer little practical protection. A supplier would have to determine what portion of each discount reflected cost savings and maintain a continuous, transaction-level record to prove it. Those compliance costs would ultimately pass through to retail prices.[116]
Compelled disclosure of contract terms creates an additional coordination risk. George Stigler showed that the central obstacle to tacit coordination is detecting deviations from a common price. Publishing even anonymized terms lowers that obstacle in a market with only a dozen large buyers. Volumes, shipping patterns, and other details may reveal the parties’ identities, and modern LLMs make reidentification easier.[117] Once every buyer can demand a supplier’s lowest disclosed price, that price becomes a floor rather than a selective discount.
For the same reasons, the Bureau should avoid fee caps, margin caps, and cost-based pricing mandates. Its 2023 study recognizes that the Bureau cannot regulate grocery prices and that price controls are blunt tools because authorities struggle to identify the right price and adjust it as conditions change.[118] The food-delivery market illustrates the risk. A de facto cap can suppress the price signals needed to sustain service in high-cost areas and prompt providers to withdraw from them.[119]
The Bureau should enforce vigorously against conduct shown to cause harm. That includes price-fixing agreements, hub-and-spoke information sharing supported by a horizontal agreement among competitors, exchanges of confidential or forward-looking pricing data, exclusionary conduct that prevents entry or expansion, deceptive discounts or loyalty terms, hidden mandatory fees, and false claims about quantity or quality.
The Bureau should not presume liability from high prices, parallel price movements, ordinary monitoring of public prices, above-cost discounts, nonexclusionary loyalty programs, or algorithmic pricing without evidence of agreement or likely competitive harm. Its existing Competitor Collaboration Guidelines already provide a workable distinction between publicly available information and competitively sensitive information. That distinction should govern here.[120]
The durable way to discipline grocery prices is through greater entry and expansion. The most effective tools operate on the supply side: zoning and land-use reform, limits on restrictive property covenants, stronger wholesale and distribution infrastructure for independents, fewer interprovincial trade barriers, and support for independent, international, discount, and online grocers. The Bureau’s 2023 market study recommends several of these measures, including support for international and discount entrants and limits on property controls that impede competition.[121]
Wholesale-price regulation poorly addresses the food-access problem it often seeks to solve. U.S. evidence illustrates the mismatch. Hunt Allcott and his co-authors estimate that equalizing access and prices between low- and high-income households would close only about 10% of the gap in nutritional quality. Differences in demand account for the remaining 90%.[122] Zoning reform, targeted incentives, and transit investments address barriers to grocery access more directly than rules governing supplier prices.
The retail-pricing positions above follow the framework ICLE has applied across its submissions to the Bureau. Our algorithmic-pricing comments urged effects-based analysis rather than categorical rules. Our comments on the draft conduct guidelines emphasized asymmetric error costs, the uncertainty created by the excessive-pricing provision, and the need for workable market-power screens. Our merger comments treated structural presumptions as screening tools rather than verdicts. The common principles are demonstrated consumer harm, administrable standards, and recognition that false positives can prove especially costly in a low-margin industry.
The evidence reviewed here points in the same direction. Pricing algorithms usually automate lawful conduct. Dynamic pricing can expand output and reduce shortages, while personalized and loyalty pricing can lower effective prices and intensify competition. Competitive harm arises through narrower mechanisms, including agreements to pool confidential data, coordinated pricing, deception, and exclusion. High prices, parallel price movements, common software, or price differences alone do not establish that harm.
The Bureau can promote transparency without becoming a price regulator. Consumer-facing measures such as harmonized unit pricing, clear final prices, and legible loyalty terms help shoppers compare offers and switch retailers. They also address the comparability problems created by shrinkflation and complex pricing. Competitor-facing disclosure can instead expose sensitive terms, facilitate coordination, weaken investment incentives, and burden smaller firms disproportionately. Same-terms mandates, categorical presumptions against algorithmic pricing, and fee or margin caps carry similar risks.
Canadian consumers already compare prices, shop across formats, and switch in response to better offers. The more durable challenge lies on the supply side. Reforms that ease entry and expansion, limit restrictive property controls, improve wholesale access for independents, reduce interprovincial trade barriers, and support new retail formats will strengthen competition more effectively than controls on pricing practices.
The Bureau can best protect Canadian shoppers by making comparison easier, entry more feasible, and demonstrated coordination or exclusion riskier. We thank the Bureau for the opportunity to comment and welcome the chance to provide further analysis on any issue raised here.
[1] Competition Bureau Can., Notice: Behind the Price Tag: Examining the Food Supply Chain (June 16, 2026), https://competition-bureau.canada.ca/en/how-we-foster-competition/notice-behind-price-tag-examining-food-supply-chain [hereinafter Notice].
[2] Geoffrey A. Manne et al., Comments on Canadian Competition Bureau Discussion Paper on Algorithmic Pricing and Competition (Int’l Ctr. for Law & Econ., Aug. 4, 2025), https://laweconcenter.org/wp-content/uploads/2025/08/Canada-Algorithmic-pricing-and-competition-1.pdf (included as Appendix A).
[3] Ian Adams et al., Comments of the International Center for Law & Economics to the Competition Bureau Canada: Anti-Competitive Conduct and Agreements (Jan. 27, 2026), https://laweconcenter.org/wp-content/uploads/2026/01/Competition-Bureau-Canada-Comments-2026.pdf.
[4] Ian Adams et al., Comments of the International Center for Law & Economics to the Competition Bureau Canada: Proposed Merger Enforcement Guidelines (Feb. 10, 2026), https://laweconcenter.org/wp-content/uploads/2026/02/Competition-Bureau-Canada-Merger-Comments-2026.pdf.
[5] Geoffrey A. Manne et al., Comments from the International Center for Law and Economics on the Future of Competition Policy in Canada (Mar. 31, 2023), https://laweconcenter.org/wp-content/uploads/2023/03/ICLE-Canada-Comments.pdf.
[6] Brian C. Albrecht et al., Food-Retail Competition, Antitrust Law, and the Kroger/Albertsons Merger (Int’l Ctr. for Law & Econ., White Paper No. 2023-10-17, 2023), https://laweconcenter.org/wp-content/uploads/2023/10/Kroger-Albertsons-Merger-Full-Paper.pdf.
[7] Eric Fruits, The Waterbed Effect and the Price of “Pricing Fairness” (Int’l Ctr. for Law & Econ., Issue Brief No. 2026-03-17, 2026), https://laweconcenter.org/wp-content/uploads/2026/03/Grocery-Pricing-Legislation-2026.pdf.
[8] Brian Albrecht et al., Comments of the International Center for Law & Economics: Rule on Unfair or Deceptive Fees in Online Food Delivery Services, ANPRM, Federal Trade Commission Project No. P267101 (May 18, 2026), https://laweconcenter.org/wp-content/uploads/2026/05/FTC-Delivery-Fees-2026.pdf.
[9] Stat. Can., Consumer Price Index, February 2026 (Mar. 16, 2026), https://www150.statcan.gc.ca/n1/daily-quotidien/260316/dq260316a-eng.htm (reporting that grocery prices rose 30.1% from February 2021 to February 2026).
[10] Competition Bureau Can., Canada Needs More Grocery Competition: Retail Grocery Market Study Report 44 (2023), https://competition-bureau.canada.ca/sites/default/files/attachments/2023/CB-Retail-Grocery-Market-Study-Report-EN-2023-06-23.pdf.
[11] See Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1 (1984).
[12] Competition Act, R.S.C. 1985, c C-34, § 78(1)(k) (Can.). See also An Act to Amend the Excise Tax Act and the Competition Act, S.C. 2023, c 31, pt 2 (Can.) [hereinafter Bill C-56].
[13] Consolidated Version of the Treaty on the Functioning of the European Union art. 102, May 9, 2008, 2008 O.J. (C 115) 47.
[14] Case 27/76, United Brands Co. v. Comm’n, ECLI:EU:C:1978:22, ¶ 252 (Feb. 14, 1978) (“The questions therefore to be determined are whether the difference between the costs actually incurred and the price actually charged is excessive, and, [if] the answer to this question is in the affirmative, whether a price has been imposed which is either unfair in itself or when compared to competing products.”).
[15] See, e.g., Aline Blankertz et al., The European Commission Can and Must Act on Excessive Pricing, ProMarket (Oct. 8, 2025), https://www.promarket.org/2025/10/08/the-european-commission-can-and-must-act-on-excessive-pricing.
[16] See Org. for Econ. Coop. & Dev., Excessive Prices, OECD Doc. DAF/COMP(2011)18 (Feb. 7, 2012), https://www.oecd.org/content/dam/oecd/en/publications/reports/2012/02/excessive-prices_372e092c/8e1fd82e-en.pdf; Raphaël De Coninck, Excessive Prices: An Overview of EU and National Case Law, Concurrences e-Competitions Bull. (June 2018), https://ecp.crai.com/wp-content/uploads/2018/06/Excessive-Pricing-R.-de-Coninck-e-Competitions-2018.pdf.
[17] David S. Evans & A. Jorge Padilla, Excessive Prices: Using Economics to Define Administrable Legal Rules, 1 J. Competition L. & Econ. 97 (2005).
[18] Massimo Motta & Alexandre de Streel, Excessive Pricing and Price Squeeze Under EU Law, in The Pros and Cons of High Prices 14 (Konkurrensverket ed., 2007), https://www.konkurrensverket.se/globalassets/dokument/informationsmaterial/rapporter-och-broschyrer/pros-and-cons/rapport_pros-and-cons_2007_high_prices.pdf.
[19] Lars-Hendrik Röller, Exploitative Abuses (ESMT Bus. Brief No. BB-107-002, 2007), https://d-nb.info/1012903311/34.
[20] Competition Act, R.S.C. 1985, c C-34, § 103.1(7) (Can.); see also Competition Bureau Can., Information Bulletin on Private Access to the Competition Tribunal (June 2025), https://competition-bureau.canada.ca/en/information-bulletin-private-access-competition-tribunal.
[21] Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 489 (1977) (holding that antitrust injury must be “of the type the antitrust laws were intended to prevent” and “flow[] from that which makes defendants’ acts unlawful”).
[22] See George J. Stigler, A Theory of Oligopoly, 72 J. Pol. Econ. 44 (1964).
[23] Maureen K. Ohlhausen, Acting Chairman, Fed. Trade Comm’n, Should We Fear the Things That Go Beep in the Night? Some Initial Thoughts on the Intersection of Antitrust Laws and Algorithmic Pricing 10 (May 23, 2017), https://www.ftc.gov/system/files/documents/public_statements/1220893/ohlhausen_-_concurrences_5-23-17.pdf.
[24] Zach Y. Brown & Alexander MacKay, Competition in Pricing Algorithms, 15 Am. Econ. J.: Microeconomics 109 (2023), https://www.aeaweb.org/articles?id=10.1257/mic.20210158.
[25] Sophie Calder-Wang & Gi Heung Kim, Algorithmic Pricing in Multifamily Rentals: Efficiency Gains or Price Coordination? (Aug. 16, 2024) (unpublished manuscript), https://ssrn.com/abstract=4403058.
[26] Stephanie Assad et al., Algorithmic Pricing and Competition: Empirical Evidence from the German Retail Gasoline Market, 132 J. Pol. Econ. 723 (2024).
[27] Id. at 726 (finding that algorithmic adoption increased margins by roughly 15% overall and had no meaningful effect in monopoly markets, but increased margins by 38% in duopoly markets where both stations adopted).
[28] Id. at 727 (“Margins start to increase only about a year after market-wide adoption, suggesting that algorithms in this market learn tacitly collusive strategies. . . . [A]lgorithms learn that undercutting will not be profitable, since lower prices will be followed.”).
[29] Id. at 760 (“To the best of our knowledge, this occurs without explicit communication between competitors, making it legal according to current competition laws in many countries.”).
[30] Competition Bureau Can., Anti-Competitive Conduct and Agreements Enforcement Guidelines ¶ 121 (proposed Oct. 31, 2025), https://competition-bureau.canada.ca/en/how-we-foster-competition/consultations/anti-competitive-conduct-and-agreements.
[31] Gibson v. Cendyn Grp., LLC, 148 F.4th 1069 (9th Cir. 2025).
[32] Id.
[33] Cornish-Adebiyi v. Caesars Entm’t, Inc., No. 24-3006, slip op. (3d Cir. July 29, 2026), rev’g 2024 WL 4356188 (D.N.J. Sept. 30, 2024).
[34] See Ken Littlewood, Forecasting and Control of Passenger Bookings, 4 J. Revenue & Pricing Mgmt. 111 (2005) (reprinting a 1972 AGIFORS paper); Barry C. Smith et al., Yield Management at American Airlines, 22 Interfaces 8 (1992).
[35] See Kevin R. Williams, The Welfare Effects of Dynamic Pricing: Evidence from Airline Markets, 90 Econometrica 831 (2022).
[36] See Juan Camilo Castillo, Who Benefits from Surge Pricing?, 93 Econometrica 1811 (2025).
[37] See Andrew Sweeting, Dynamic Pricing Behavior in Perishable Goods Markets: Evidence from Secondary Markets for Major League Baseball Tickets, 120 J. Pol. Econ. 1133 (2012).
[38] See Frank A. Wolak, Do Residential Customers Respond to Hourly Prices? Evidence from a Dynamic Pricing Experiment, 101 Am. Econ. Rev. Papers & Proc. 83, 87 (2011).
[39] See Brian C. Albrecht, Price Competition and the Use of Consumer Data (Working Paper, June 12, 2026), https://briancalbrecht.com/papers/price-competition-data.
[40] See Manne et al., supra note 2, at 11–12 (discussing Nicholas Buchholz et al., Personalized Pricing and the Value of Time: Evidence from Auctioned Cab Rides, 93 Econometrica 929, 930 (2025)); see also John M. Yun, Should We Fear Personalized Pricing?, CPI Antitrust Chron. (July 2025).
[41] See James D. Dana Jr., Advance-Purchase Discounts and Price Discrimination in Competitive Markets, 106 J. Pol. Econ. 395, 395 (1998).
[42] See Joanna Stavins, Price Discrimination in the Airline Market: The Effect of Market Concentration, 83 Rev. Econ. & Stat. 200, 202 (2001).
[43] See Severin Borenstein & Nancy L. Rose, Competition and Price Dispersion in the U.S. Airline Industry, 102 J. Pol. Econ. 653, 672 (1994).
[44] See Ambarish Chandra & Mara Lederman, Revisiting the Relationship Between Competition and Price Discrimination, 10 Am. Econ. J.: Microeconomics 190, 193, 198, 210 (2018).
[45] See Retail Council of Can., The Great Canadian Shopping Trip (2019); see infra Sub-Question 4.
[46] See Eur. Comm’n, Consumer Market Study on Online Market Segmentation Through Personalised Pricing/Offers in the European Union 171 (2018).
[47] See Manne et al., supra note 2, at 9 (quoting OECD); Yun, supra note 40.
[48] See Katie Wells et al., Same Cart, Different Price: Instacart’s Price Experiments Cost Families at Checkout (Groundwork Collaborative & Consumer Reports, Dec. 9, 2025), https://groundworkcollaborative.org/wp-content/uploads/2025/12/Same-Cart-Different-Price.pdf.
[49] See Brian Albrecht, Instacart Didn’t Read Your Mind. It Ran an A/B Test., Truth on the Mkt. (Dec. 10, 2025), https://truthonthemarket.com/2025/12/10/instacart-didnt-read-your-mind-it-ran-an-a-b-test.
[50] See Ginger Zhe Jin et al., Surveillance Pricing: A Cautionary Summary of Potential Harms and Solutions, Competition Pol’y Int’l (July 14, 2025).
[51] See H.B. 895, 2026 Gen. Assemb., Reg. Sess. (Md. 2026) (enacted as 2026 Md. Laws ch. 154) (effective Oct. 1, 2026); Brian Albrecht, How a Panic over Egg Prices Hatched a Bad Law, Wash. Post (May 11, 2026); see infra Sub-Question 3.
[52] See Brian Albrecht, Eliminate ‘Personal Pricing’ and You Risk Harming Consumers, Fin. Times (July 2, 2026).
[53] See U.K. Competition & Mkts. Auth., Loyalty Pricing in the Groceries Sector—Summary (Nov. 27, 2024), https://assets.publishing.service.gov.uk/media/6749f28975bb645366b3a2a9/Executive_summary__2.pdf; U.K. Competition & Mkts. Auth., Loyalty Pricing in the Groceries Sector—Findings Report (Nov. 2024).
[54] See Borenstein & Rose, supra note 43.
[55] See Competition Bureau Can., Competitor Collaboration Guidelines § 1.1 (May 6, 2021), https://competition-bureau.canada.ca/sites/default/files/attachments/2022/CB-BC-CCGs-Eng.pdf.
[56] See id. § 3.7.2, at 42; Competition Bureau Can., supra note 30, ¶ 272.
[57] See Competition Bureau Can., supra note 55, § 2.2.
[58] See Competition Act, R.S.C. 1985, c C-34, §§ 78(1)(h)–(i), 79(1)(b)(ii) (Can.).
[59] See Competition Bureau Can., supra note 30, ¶¶ 171–176.
[60] Gibson, 148 F.4th at 1076, 1083 & n.8, 1088 n.12.
[61] See Cornish-Adebiyi v. Caesars Entm’t, Inc., No. 1:23-cv-02536-KMW-EAP, 2024 WL 4356188 (D.N.J. Sept. 30, 2024), rev’d, No. 24-3006, slip op. (3d Cir. July 29, 2026); Cornish-Adebiyi, No. 24-3006, slip op. at 9, 36–38, 42; In re RealPage, Inc., 709 F. Supp. 3d 478, 494, 512 (M.D. Tenn. 2023).
[62] See Competition Act, R.S.C. 1985, c C-34, § 79(1)(b)(ii) (Can.).
[63] See Competition Bureau Can., supra note 30, ¶ 236.
[64] See Br. of Amici Curiae Int’l Ctr. for Law & Econ. et al. at 2, In re EpiPen (Sanofi-Aventis U.S., LLC v. Mylan Inc.), No. 21-3005 (10th Cir. Sept. 22, 2021).
[65] See id. at 15.
[66] Org. for Econ. Coop. & Dev., Algorithmic Competition: Background Note, OECD Doc. DAF/COMP(2023)3 (June 14, 2023), https://one.oecd.org/document/DAF/COMP(2023)3/en/pdf.
[67] Emek Basker, The Causes and Consequences of Wal-Mart’s Growth, 21 J. Econ. Persp. 177, 188–89 (2007).
[68] Id. at 188.
[69] Jerry Hausman & Ephraim Leibtag, Consumer Benefits from Increased Competition in Shopping Outlets: Measuring the Effect of Wal-Mart, 22 J. Applied Econometrics 1157 (2007).
[70] Fruits, supra note 7.
[71] See Stefano DellaVigna & Matthew Gentzkow, Uniform Pricing in U.S. Retail Chains, 134 Q.J. Econ. 2011 (2019).
[72] See Alberto Cavallo, Are Online and Offline Prices Similar? Evidence from Large Multi-Channel Retailers, 107 Am. Econ. Rev. 283, 290, 292 (2017).
[73] Albrecht et al., supra note 8.
[74] Competition Bureau Can., Algorithmic Pricing and Competition: Discussion Paper (June 10, 2025), https://competition-bureau.canada.ca/en/how-we-foster-competition/education-and-outreach/publications/algorithmic-pricing-and-competition-discussion-paper. See infra Appendix A.
[75] See infra Appendix A.
[76] Competition Bureau Can., supra note 55, at 42–43.
[77] Org. for Econ. Coop. & Dev., supra note 66.
[78] Neth. Auth. for Consumers & Mkts., Position Paper: Oversight of Algorithms (2020), https://www.acm.nl/sites/default/files/documents/position-paper-oversight-of-algorithms.pdf.
[79] Eur. Comm’n, Commission Staff Working Document Accompanying the Final Report on the E-Commerce Sector Inquiry 17, 22, 24, 29–32, SWD(2017) 154 final (May 10, 2017), https://eur-lex.europa.eu/legal-content/EN/ALL/?uri=SWD:2017:154:FIN.
[80] Id.
[81] See Org. for Econ. Coop. & Dev., supra note 66, at 12. In 2020, the Norwegian Competition Authority found that 55% of surveyed firms used monitoring algorithms and 20% used pricing algorithms. In Denmark, 17% of e-commerce companies used pricing algorithms, with automation ranging from basic information provision to full algorithmic control. The Netherlands Authority for Consumers and Markets reported that 36% of firms used competitors’ pricing data. Of those firms, 16%—or 6% of all firms—used pricing algorithms.
[82] Albrecht et al. supra note 8.
[83] Albrecht et al., supra note 6.
[84] Id.
[85] See, e.g., Simon Lermen et al., Large-Scale Online Deanonymization with LLMs, arXiv:2602.16800 (Feb. 18, 2026), https://doi.org/10.48550/arXiv.2602.16800.
[86] Stigler, supra note 22; Fruits, supra note 7. See infra Sub-Question 6 for further discussion of compelled disclosure.
[87] Wells et al., supra note 48.
[88] Albrecht, supra note 49.
[89] H.B. 895, 2026 Gen. Assemb., Reg. Sess. (Md. 2026) (enacted as 2026 Md. Laws ch. 154) (effective Oct. 1, 2026).
[90] Albrecht, supra note 51.
[91] See Org. for Econ. Coop. & Dev., supra note 66 (reporting the European Commission’s mystery-shopping exercise); see also infra Appendix A.
[92] Estimates of the combined market share vary by scope and source. See U.S. Dep’t of Agric., Foreign Agric. Serv., Retail Foods Annual: Canada, GAIN Rep. No. CA2025-0040 (2025), https://www.fas.usda.gov/data/gain-report/2025/12/Retail%20Foods%20Annual_Ottawa_Canada_CA2025-0040.pdf (estimating that five leading retailers held more than 65% of the market in 2024: Loblaws 17%, Costco 15%, Sobeys 14%, Walmart 13%, and Metro 6%); Kevin Grier, Grocery Retail (UFCW Loc. 401, Nov. 2024), https://gounion.ca/wp-content/uploads/2024/11/UFCW-Local-401-Retail-Grier-November-2024-rev.pdf (estimating 2022 shares of 26% for Loblaws, 20% for Sobeys, 14% for Costco, and 12% each for Walmart and Metro).
[93] Competition Bureau Can., supra note 10.
[94] Id.
[95] Karl Littler, Retail Council of Can., Retail Council of Canada Submission to the Competition Bureau’s Market Study: Competition in Canada’s Grocery Sector (Dec. 16, 2022), https://competition-bureau.canada.ca/sites/default/files/attachments/2023/RCC-Market-Study-Submission.pdf (arguing that concentration had not recently increased, big-box retailers and new entrants had gained market share, regional and independent grocers remained significant, and 87% of Canadians regularly shopped at discount grocers).
[96] Ritika Dubey, Discount Is the New Name of the Game. What Does That Mean for Conventional Stores?, Can. Press (Apr. 3, 2026), https://www.thecanadianpressnews.ca/business/discount-is-the-new-name-of-the-game-what-does-that-mean-for-conventional-stores/article_50f71fe5-5410-5c00-ab63-b5d0655397f6.html (quoting Henry Chamber of Sentinel Management Consultants on grocers’ efforts to retain shoppers across their discount and conventional banners); Rebecca Harris, Retailers Are Betting on Discount. Will the Momentum Last?, Can. Grocer (Dec. 2, 2024), https://canadiangrocer.com/retailers-are-betting-discount-will-momentum-last (reporting that more than 62% of grocery shoppers had changed their primary store to secure better deals and nearly 30% chose stores based on in-store discounts and promotions).
[97] Competition Bureau Can., supra note 10.
[98] Id.
[99] Angela Altass, Budget-Conscious Shoppers Drive Increased Growth in Discount Grocery Channel, Grocery Bus. (July 9, 2025) (reporting that discount grocers’ market share rose from 35% in 2008 to roughly 50% and that their same-store sales outpaced conventional grocers 3-to-1 nationally by both dollars and tonnage).
[100] Competition Bureau Can., supra note 10.
[101] Id.
[102] Id.
[103] Id.
[104] U.K. Competition & Mkts. Auth., Loyalty Pricing in the Groceries Sector—Summary, supra note 53 (reporting that 92% of roughly 50,000 products offered through dual-price loyalty promotions yielded genuine savings, with average member savings of 17% to 25%).
[105] See Aljoscha Janssen & Johannes Kasinger, Shrinkflation and Consumer Demand, 45 Mktg. Sci. 142 (2026).
[106] Id. at 142.
[107] Competition Act, R.S.C. 1985, c C-34, § 45(1)(c) (Can.); Competition Bureau Can., supra note 55, § 2.4.3.
[108] See Xibo Wan et al., Shrinkflation, Unit Price Disclosure, and Consumer Welfare: Evidence from Canned Tuna, 142 Food Pol’y 103106 (2026); Press Release, U.S. Dep’t of Just., Former Bumble Bee CEO Sentenced to Prison for Fixing Prices of Canned Tuna (June 16, 2020), https://www.justice.gov/archives/opa/pr/former-bumble-bee-ceo-sentenced-prison-fixing-prices-canned-tuna.
[109] Competition Bureau Can., supra note 10, at 44.
[110] See S. 8563, 2025–2026 Reg. Sess. (N.Y. 2025).
[111] Fruits, supra note 7.
[112] See Antitrust Modernization Comm’n, Report and Recommendations (2007).
[113] Roman Inderst & Tommaso M. Valletti, Buyer Power and the “Waterbed Effect”, 59 J. Indus. Econ. 1 (2011).
[114] See Brian C. Albrecht, Consumer Harm from the Waterbed Effect: A Comment on Inderst and Valletti (2011) (Working Paper, Apr. 22, 2026), https://briancalbrecht.com/Albrecht-IV-Waterbed-Comment.pdf; Brian C. Albrecht, When Does the Waterbed Effect Harm Consumers? (Int’l Ctr. for Law & Econ., July 24, 2026), https://laweconcenter.org/resources/when-does-the-waterbed-effect-harm-consumers.
[115] DeHoog v. Anheuser-Busch InBev SA/NV, No. 1:15-cv-02250-CL, 2016 U.S. Dist. LEXIS 137759 (D. Or. July 22, 2016).
[116] Fruits, supra note 7.
[117] Stigler, supra note 22; Fruits, supra note 7.
[118] Competition Bureau Can., supra note 10.
[119] Albrecht et al., supra note 8.
[120] Competition Bureau Can., supra note 55.
[121] See Competition Bureau Can., supra note 10, recommendations 2 and 4.
[122] Hunt Allcott et al., Food Deserts and the Causes of Nutritional Inequality, 134 Q.J. Econ. 1793 (2019).
Source: Competition Bureau Canada