California is about to run a live-fire experiment in antitrust—and the working hypothesis appears to be that decades of case law and economic learning were optional.
In January, I published a short post—“Rewriting Antitrust, California Style”—that touched on the inner workings (machinations?) of the California Law Review Commission (CLRC). I flagged concerns about the staff’s recommendations on single-firm conduct. I was hardly alone. The International Center for Law & Economics (ICLE) submitted comments, as did Bilal Sayyed and Tech Freedom; Joe Coniglio and the Information Technology & Innovation Foundation; Daniel Francis of New York University School of Law; and Herbert Hovenkamp, who offered a characteristically concise note on cross-market effects.
My own comments were unusually (for me) brief. I raised a high-level concern about efforts to distance California from federal antitrust law and promised more as the CLRC process unfolded. Time flies. In the interim, two—count ’em, two—antitrust bills have landed before the California Legislature.
The first is California Assembly Bill 1776, laboriously titled in search of an acronym: the Competition and Opportunity in Markets for a Prosperous, Equitable and Transparent Economy (COMPETE) Act. It is a direct descendant of the CLRC recommendations. The Assembly Judiciary Committee advanced the bill April 7 on a strict party-line vote and re-referred it to the Appropriations Committee. Jonathan Nuechterlein—a former Federal Trade Commission (FTC) general counsel—recently analyzed the bill, its prospects, and its defects.
The second is California Senate Bill 1074, the Blocking Anticompetitive Self-Preferencing by Entrenched Dominant Platforms—the “BASED” Act. The acronym tries a bit too hard—and lands a bit obscure. State Sen. Scott Wiener (D-San Francisco) introduced the bill, reportedly with enthusiastic backing (if not drafting) from Y Combinator and Economic Security California—a misnomer, but never mind.
Both bills appear to be based on faith in unconstrained judicial intervention and a general rejection of established antitrust principles–or, at least, a rejection of U.S. antitrust. The BASED Act in particular seems to embody a “neo-Brandeisian” animus towards large tech firms and a rejection of consumer welfare. Neither the facts of federal enforcement experience nor decades of economic learning seem to have made much of an impact.
Spoiler alert for the impatient reader who, on the one hand, reads my posts here at Truth on the Market but, on the other, cannot quite predict my reaction: I do not care for either bill. Taken together, they look like a recipe for disaster.
The COMPETE Act’s Anything-Goes Approach
AB 1776 is an odd piece of drafting, but its intent to break from established federal antitrust law is expressly stated and runs throughout the bill.
The statement of purpose opens with a sweeping declaration. The bill aims to promote and protect:
…free and fair competition, which is fundamental to a healthy marketplace that protects all trade participants, including workers and consumers, and to an environment that is conducive to the preservation of our democratic, political, and social institutions.
In other words, it protects everyone and everything—from democratic, political, and social institutions (or at least an “environment” conducive to them), to consumers, workers, and “all trade participants.” So much for the Supreme Court’s oft-repeated line, dating to Brown Shoe, that antitrust law protects “competition, not competitors.” Tradeoffs? What tradeoffs?
The rest of the bill—starting with the remainder of the purpose statement and continuing through the substantive provisions—both recites and rejects core federal antitrust principles. A proposed Section 16730 of California’s Business & Professions Code reminds us that “[t]he California Supreme Court has determined that the Cartwright Act is ‘broader in range and deeper in reach’ than the Sherman Anti-Trust Act.” It adds that courts “may consider federal case law as persuasive authority to the extent they find it consistent with California law”—but that federal precedent “is not binding on California state courts.”
That theme—citing and discarding established principles of federal antitrust law—runs throughout.
Take Ohio v. American Express Co. There, the Supreme Court held that courts must consider both sides of a two-sided platform when assessing competitive effects, at least with certain two-sided platforms. Drawing on work by Richard Schmalensee and David Evans, and by Benjamin Klein, Andres Lerner, Kevin Murphy and Lacey Plache, the Court explained that “[e]valuating both sides of a two-sided transaction platform is also necessary to accurately assess competition.” And accurately assessing competition, and the question of whether or not it had been impaired, was the task at hand, not formalistic category distinctions. That holding has sparked debate over how broadly to read Amex: Does it apply only to “transaction platforms”? Does simultaneity matter, and if so, why?
Hovenkamp captures the stakes succinctly:
Antitrust rarely does cross-market balancing [because] it is hard. Preventing it even when it isn’t is dumb: like doing cost-benefit analysis by looking only at costs. would you condemn the iPhone camera by looking only at the impact on camera makers, ignoring users of iPhone?
Wherever one lands on Amex, AB 1776 sidesteps the debate. The bill describes—and then repudiates—the Court’s holding (without citation). Harm on one side of a multisided platform “may constitute evidence of a violation,” but liability does not require harm on more than one side, nor any weighing of harms against benefits across sides.
The pattern repeats across doctrines.
Refusals to deal? Under federal law, Aspen Skiing Co. v. Aspen Highlands Skiing Corp. sets a demanding standard for the imposition of a duty to deal—a departure from the default rule that firms should be free to choose their trading partners. That was a sufficient departure from a sufficiently general presumption that the Supreme Court later described it as “at or near the outer boundary of §2 liability.”
Yet AB 1776 treats factors like terminating a prior course of dealing, differential treatment of rivals, or conduct that “makes no economic sense” as neither necessary nor sufficient for a court to impose a duty to deal. They may be evidence of anticompetitive conduct, but they are not required to make a case for court-ordered (and supervised) dealing.
Predatory pricing fares no better. Federal law requires below-cost pricing, likely exclusion, and a plausible path to recoupment. AB 1776 requires none of these. What does it require? The bill leaves that to the discretion of California courts.
Nuechterlein underscores the problem:
Although reasonable people can disagree about the doctrinal details, some such limiting principles are needed for sound competition policy. Then-Judge Stephen Breyer put it best when he observed in 1983 that “the consequence of a mistake here is not simply to force a firm to forego legitimate business activity it wishes to pursue; rather, it is to penalize a procompetitive price cut, perhaps the most desirable activity (from an antitrust perspective) that can take place in a concentrated industry where prices typically exceed costs.”
And the bill does not just relax federal standards—it abandons them:
[It] would impose state-level liability for price cuts that easily comply with the federal standard, and it would impose no alternative limiting principles instead. Without such principles, California would leave it to the subjective intuitions of California judges or juries about whether price cuts are excessive or unfair to rivals. Unlucky defendants in such cases would have to pay treble damages for the rivals’ lost business. And to avoid that fate, well-counseled companies would pull their punches and err on the side of charging consumers more—precisely the anticompetitive outcome that Breyer warned about.
All this follows a familiar script. The bill declares that the Cartwright Act is “broader in range and deeper in reach” than the Sherman Act. It distances California law from U.S. Supreme Court precedent. It cites—then disclaims—market-share and market-power thresholds recognized under Section 2 of the Sherman Act. They may serve as evidence, but they are not required, and the act specifies no substitutes (never mind for monopoly power).
In short, the COMPETE Act does not merely articulate a clear, plaintiff-friendly standard to replace federal law. Instead, it rejects a range of established limits on liability for unilateral (and, at times, joint) conduct, and invites California courts and juries to go further, all in service of “all trade participants.”
And so much for the consumer welfare standard.
The BASED Act’s Billion-Dollar Gatekeeping
Here, by the way, are some brief but useful comments on SB 1074—by my ICLE colleagues Geoffrey Manne, Ian Adams, Dirk Auer, and Eric Fruits. SB 1074 addresses what might seem a narrower domain: “self-preferencing” by large online platforms. It is arguably narrower than the COMPETE Act, in that it applies only to “self-preferencing” and to “covered providers”—firms that operate large digital platforms. Narrower, perhaps, but not at all narrow.
“Large digital platforms” are defined in two parts. First, a platform must provide “a digital interface that allows business users or sellers to connect with consumers or other business users.” Second, it must meet a scale requirement: “The platform, at any time in the last 24 months, had an average of 100,000,000 or more monthly active users in the United States.”
That is a large number. Even so, “at any time” means that a short-term spike in usage could sweep more firms into the scope of the bill than one might expect. And notably, the statute keys this threshold to nationwide users, not California users.
Second, the bill limits its reach to “covered providers,” defined as:
…a person or entity operating a platform that meets either of the following conditions: “(A) At any point prior to January 1, 2030, it is owned, controlled, or operated by an entity or person that in the last 24 months has had an average of one trillion dollars ($1,000,000,000,000) or greater in market capitalization. [or] (B) At any point on or after January 1, 2030, it is owned, controlled, or operated by an entity or person that in the last 24 months has had an average of one trillion dollars ($1,000,000,000,000) or greater in market capitalization or private valuation.”
Again, the bill sets a high threshold—this time using an average over a 24-month period, rather than a momentary spike. But it does not require that the platform itself (as opposed to the parent firm) meet that threshold. Nor does it require that the firm hold assets in California at or above that level, or that the line of business in question generates any particular income or has any particular valuation.
Consider Walmart. It is headquartered in Arkansas and operates more stores in Texas than in California. Still, it appears to qualify as a “covered provider,” with a market capitalization (as of April 8) exceeding $1 trillion and a platform—walmart.com—with more than 100 million monthly users. On the bill’s terms, its conduct would be subject to the stipulated prohibitions, whether or not Walmart holds market power in any relevant market.
One small but telling detail: the statute fixes a dollar threshold with no adjustment for inflation. That may sound like nitpicking, but it matters. Try an inflation calculator over a decade or so—especially using late-1970s rates of 9%, 13.3%, 12.5%, and 8.9%. Today’s $1 trillion threshold may not mean the same thing for long.
More fundamentally, as the ICLE comments note, “[t]hese criteria target a small number of firms without grounding in neutral legal principles. This approach subjects identical conduct to different legal standards based solely on firm size and invites strategic behavior.” To that, one might add: without any apparent grounding in the economic literature. The thresholds look either arbitrary or reverse-engineered from the market capitalizations of particular firms.
Even the “progressive” and ambitious CLRC (recall the COMPETE Act from two minutes ago) reached a different conclusion after a multi-year review:
It found that exclusionary conduct can arise in any industry and advised that any reform should apply across sectors. It also declined to recommend abuse-of-dominance provisions, citing concerns about vague and arbitrary thresholds. (internal citations omitted)
When ‘Self-Preferencing’ Means Whatever You Need It To
The focus on “self-preferencing” might sound narrow. It is not. The BASED Act prohibits—“but is not limited to”—a laundry list of conduct that is not generally harmful to competition or consumers. The seven categories of prohibited conduct in proposed Section 16851(a) sweep across a range of vertical integration practices—both within firms and in agreements with third parties—that may often be procompetitive or benign.
The bill offers an affirmative defense, but it comes at a price. Defendants must show that the conduct was both (a) “narrowly tailored, non-pretextual, and reasonably necessary to achieve a procompetitive purpose” and (b) that the “procompetitive benefits and actual effects clearly outweigh the competitive harms in the same market.” That is a demanding standard. For a broad and loosely defined set of conduct—much of it frequently procompetitive—the bill effectively presumes illegality. Defendants may rebut that presumption, but only under a burden more stringent than anything in federal antitrust law or current California law. Meanwhile, plaintiffs—including rivals—could survive motions to dismiss without satisfying established pleading standards for harm to competition or consumers.
Add private rights of action and treble damages, and the incentives become obvious. Expect suits across the spectrum—from meritorious, to questionable, to spurious.
The litigation risk does not stem only from the breadth of these prohibitions, but also from their ambiguity. Once we move beyond clear, limiting cases, the boundaries become hard to discern. For example, one might have a casual or intuitive sense that a firm might unduly favor its own products or services in search results or rankings. If so, an unanalyzed prohibition against the manipulation of “the order of search results or rankings to favor the products or services of the covered provider” might seem straightforward.
But straightforward compared to what? Why should a firm not promote its own products, so long as it is not materially deceptive and does not violate the Sherman Act in doing so? Even setting that aside, what does “manipulation” mean in practice for a firm that constantly designs, tests, and refines ranking algorithms or models? What is the neutral baseline against which “manipulation” becomes unlawful?
The same sorts of questions arise elsewhere. The bill would prohibit using “transaction data or nonpublic proprietary data collected from a third-party seller to market or develop the products of the covered provider.” Why should such use be categorically unlawful, regardless of the nature of the data or the conditions under which it was obtained?
In addition to the seven categories of prohibited “self-preferencing” under proposed Section 16851(a), the bill adds four additional categories of “independence or interoperability” obligations under proposed Section 16851(b). These are framed as prohibitions, but they operate as affirmative technical and managerial mandates, constraining how platforms collect data, and how they design and use data structures and computational systems.
Consider data portability. The bill would prohibit conduct that “[r]estrict[s] a business user or consumer from obtaining a copy of their data in a useful and portable format.” In limited contexts, that may be sensible. Stated at this level of generality, it is anything but. What counts as “useful”? To what extent would a covered person be liable to make such data useful? What counts as “their data”? Outside simple cases—such as clearly defined “personally identifiable information” (PII), “sensitive PII,” or standardized business inputs—the answers quickly become unclear. For complex data structures, making data “useful” to a particular user may be burdensome or even infeasible.
Similarly, the bill bars platforms from “[r]estrict[ing] a consumer from voluntarily providing data through a covered platform to a third party.” That may sound innocuous. It is not. What data? In what format? To which third parties? And with what implications for privacy and data security?
The bill also imposes interoperability obligations for artificial intelligence (AI), defined in strikingly broad terms, and under what may be the haziest definition of AI I have seen (and I’ve been reading such things since the 1970s):
“Artificial intelligence” means an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
That definition sweeps in a vast range of systems. The bill attempts a safe harbor: the AI provisions do not apply if the platform “consistently applies a neutral methodology” to both its own and third-party content, and any differences arise solely from that methodology. But the burden of proof rests with the platform. And the bill offers no clear definition of “neutral methodology,” nor any explanation of why a platform should ignore its own commercial interests when designing its systems.
The Economics the BASED Act Forgot
It has long been understood that vertical restraints—and vertical integration more generally—can be anticompetitive under particular facts and circumstances. That much is clear from the Vertical Merger Guidelines jointly adopted by the FTC and the U.S. Justice Department (DOJ) in June 2020 (if only to be summarily withdrawn by agency leadership under the Biden administration), and even from the deeply flawed and controversial Merger Guidelines adopted on partisan lines in December 2023.
Even so, federal courts—and many state courts—have moved away from per se condemnation of vertical integration. That shift reflects decades of enforcement experience and economic learning.
The BASED Act moves in the opposite direction. It does not merely relax standards of proof for certain types of conduct associated with competitive harm. It reverses them. It turns decades of judicial and enforcement experience on their head—and disregards a substantial body of economic literature in the process.
As the ICLE comments put it:
SB 1074 rests on the premise that self-preferencing and vertical integration are inherently anticompetitive. The empirical literature does not support that premise.
Indeed, the empirical literature points the other way. Francine Lafontaine—a former director of the FTC’s Bureau of Economics—and Margaret Slade conducted a comprehensive meta-analysis and found that profit-maximizing vertical integration typically benefits consumers:
In spite of the lack of unified theory, overall a fairly clear empirical picture emerges. The data appear to be telling us that efficiency considerations overwhelm anticompetitive motives in most contexts. Furthermore, even when we limit attention to natural monopolies or tight oligopolies, the evidence of anticompetitive harm is not strong.
Similarly, James Cooper, Luke Froeb, Daniel O’Brien, and Michael Vita conclude:
Because the welfare effects of vertical practices are theoretically ambiguous, optimal decisions depend heavily on prior beliefs, which should be guided by empirical evidence. Empirically, vertical restraints appear to reduce price and/or increase output. Thus, absent a good natural experiment to evaluate a particular restraint’s effect, an optimal policy places a heavy burden on plaintiffs to show that a restraint is anticompetitive.
(FWIW, all four authors are former FTC economists, including three deputy directors and Froeb, who served as director of the FTC’s Bureau of Economics and as chief economist of the DOJ Antitrust Division.)
Work focused on digital platforms reinforces the point. See, e.g., Zhuoxin Li and Ashish Agarwal on Facebook’s acquisition and integration of Instagram and Jens Foerderer and coauthors on platform entry and innovation.
None of this suggests that vertical integration is always benign. It is not. Nor has anyone seriously argued that it should be per se lawful. The point is more modest—and more important: such conduct does not generally harm competition or consumers. The weight of the economic literature—and not just federal precedent—cuts strongly against the presumption of illegality at the core of the BASED Act.
SB 1074 would instead impose a de facto “guilty-until-proven-innocent” regime. That approach will generate false positives—many of them—condemning conduct that is procompetitive and pro-consumer. Mounting an affirmative defense will be costly and uncertain. At the same time, the presumption of illegality will encourage less efficient rivals to bring opportunistic suits.
Add treble damages and private rights of action, and the effect becomes predictable: heightened litigation risk around ordinary business conduct, including innovation and product improvement. The rational response for covered providers will be caution—excessive caution. That is a recipe for less competition, not more, particularly in dynamic markets where innovation depends on precisely the kinds of integration the bill would deter.
Two Bills, One Big Mistake
Either bill—the COMPETE Act or the BASED Act—would substantially expand antitrust liability for competitively ambiguous and, in many cases, procompetitive conduct. Each rejects long-established substantive and procedural standards of federal antitrust law. More than that, each turns its back on the Supreme Court’s oft-repeated observation that antitrust law protects “competition, not competitors.” (Brown Shoe initially, and, e.g., in many various subsequent cases.) One need not believe federal antitrust law is static—or optimal in all respects—to question the rationale for these alternatives.
Some competitors will benefit. That much seems clear. California courts will likely see heavier dockets. But reasons to expect gains in innovation, competition, or consumer welfare are thin—and often exceedingly so. Meanwhile, the costs will not stay within California. The bills’ interstate reach, through spillover effects, is not just likely—it is effectively baked in.
That prospect has prompted questions about federal preemption and constitutional limits. Alden Abbott notes precedent that affords states latitude to depart from federal antitrust standards, but suggests the COMPETE Act may push far enough to invite dormant Commerce Clause scrutiny. Eric Stock similarly explores limits on state antitrust autonomy in proposals from both California and New York. A Commerce Clause challenge may face long odds, and Jonathan Nuechterlein expresses skepticism about its viability, at least as to the COMPETE Act. Still, he identifies a core tension with federal policy:
…what can keep individual states from unilaterally knocking down the limiting principles that federal antitrust law has erected precisely to promote federal competition objectives?
At what point might Congress step in? It is not clear. The tension, however, is.
Whatever courts or Congress may do, the bills themselves are overbroad and poorly justified. Taken separately—or together—they would upend antitrust as we know it. If enacted, they are far more likely to do harm than good. Their reach also undercuts any claim that they function as contained “experiments” in state policy.
Whatever else they are, these bills do not reflect sober, incremental, evidence-based reform. Once more, and with feeling, California legislators should just say no.
