The Federal Trade Commission has accused Amazon of illegally maintaining its monopoly, extracting supra-competitive fees on merchants that use Amazon’s platform. If and when the fact-finder determines that Amazon violated the antitrust laws, we propose structural remedies to address the competitive harms. Behavioral remedies have fallen out of favor among antitrust scholars. But the success of a structural remedy cannot be taken for granted.
To briefly review the bidding, the FTC’s Complaint alleges that Amazon prevents merchants from steering customers to a lower-cost platform—that is, a platform that charges a lower take rate—by offering discounts off the price it charges on Amazon. Amazon threatens merchants’ access to the Buy Box if merchants are caught charging a lower price outside of Amazon, a variant of a most-favored-nation (MFN) restriction. In other words, Amazon won’t allow merchants to share any portions of its savings with customers as an inducement to switch platforms; doing so would put downward pressure on Amazon’s take rate, which has climbed from 35 to 45 percent since 2020 per ILSR.
The Complaint also alleges that Amazon ties its fulfillment services to access to Amazon Prime. Given the importance of Amazon Prime to survival on Amazon’s Superstore, Amazon’s policy is effectively conditioning a merchant’s access to its Superstore on an agreement to purchase Amazon’s fulfillment, often at inflated rates. Finally, the Complaint alleges that Amazon gives its own private-label brands preference in search results.
These are classic exclusionary restraints that, in another era, would be instinctively addressed via behavioral remedies. Ban the MFN, ban the tie-in, and ban the self-preferencing. But that would be wrongheaded, as doing so would entail significant oversight by enforcement authorities. As the DOJ Merger Remedies Manual states, “conduct remedies typically are difficult to craft and enforce.” To the extent that a remedy is fully conduct-based, it should be disfavored. The Remedies Manual appears to approve of conduct relief to facilitate structural relief, “Tailored conduct relief may be useful in certain circumstances to facilitate effective structural relief.”
Instead, there should be complete separation of the fulfillment services from the Superstore. In a prior piece for The Sling, we discussed two potential remedies for antitrust bottlenecks—the Condo and the Coop. In what follows, we explain that the Condo approach is a potential remedy for the Amazon platform bottleneck and the Coop approach a good remedy for the fulfillment center system. Our proposed remedy has the merit of allowing for market mechanisms to function to bypass the need for continued oversight after structural remedies are deployed.
Breaking Up Is Hard To Do
Structural remedies to monopolization have, in the past, created worry about continued judicial oversight and regulation. “No one wants to be Judge Greene.” He spent the bulk of his remaining years on the bench having his docket monopolized by disputes arising from the breakup of AT&T. Breakup had also been sought in the case of Microsoft. But the D.C. Circuit, citing improper communications with the press prior to issuance of Judge Jackson’s opinion and his failure to hold a remedy hearing prior to ordering divestiture of Microsoft’s operating system from the rest of the company, remanded the case for determination of remedy to Judge Kollar-Kotelly.
By that juncture of the proceeding, a new Presidential administration brought a sea change by opposing structural remedies not only in this case but generally. Such an anti-structural policy conflicts with the pro-structural policy set forth in Standard Oil and American Tobacco—that the remedy for unlawful monopolization should be restructuring the enterprises to eliminate the monopoly itself. The manifest problem with the AT&T structural remedy and the potential problem with the proposed remedy in Microsoft is that neither removed the core monopoly power that existed, thus retaining incentives to engage in anticompetitive conduct and generating continued disputes.
The virtue of the structural approaches we propose is that once established, they should require minimal judicial oversight. The ownership structures would create incentives to develop and operate the bottlenecks in ways that do not create preferences or other anticompetitive conduct. With an additional bar to re-acquisition of critical assets, such remedies are sustainable and would maximize the value of the bottlenecks to all stakeholders.
Turn Amazon’s Superstore into a Condo
The condominium model is one in which the users would “own” their specific units as well as collectively “owning” the entire facility. But a distinct entity would provide the administration of the core facility. Examples of such structures include the current rights to capacity on natural gas pipelines, rights to space on container ships, and administration for standard essential patents and for pooled copyrights. These examples all involve situations in which participants have a right to use some capacity or right but the administration of the system rests with a distinct party whose incentive is to maximize the value of the facility to all users. In a full condominium analogy, the owners of the units would have the right to terminate the manager and replace it. Thus, as long as there are several potential managers, the market would set the price for the managerial service.
A condominium mode requires the easy separability of management of the bottleneck from the uses being made of it. The manager would coordinate the uses and maintain the overall facility while the owners of access rights can use the facility as needed.
Another feature of this model is that when the rights of use/access are constrained, they can be tradable; much as a condo owner may elect to rent the condo to someone who values it more. Scarcity in a bottleneck creates the potential for discriminatory exploitation whenever a single monopolist holds those rights. Distributing access rights to many owners removes the incentive for discriminatory or exclusionary conduct, and the owner has only the opportunity to earn rents (high prices) from the sale or lease of its capacity entitlement. Thus, dispersion of interests results in a clear change in the incentives of a rights holder. This in turn means that the kinds of disputes seen in AT&T’s breakup are largely or entirely eliminated.
The FTC suggests skullduggery in the operation of the Amazon Superstore. Namely, degrading suggestions via self-preferencing:
Amazon further degrades the quality of its search results by buying organic content under recommendation widgets, such as the “expert recommendation” widget, which display Amazon’s private label products over other products sold on Amazon.
Moreover, in a highly redacted area of the complaint, the FTC alleges that Amazon has the ability to “profitably worsen its services.”
The FTC also alleges that Amazon bars customers from “multihoming:”
[Multihoming is] simultaneously offering their goods across multiple online sales channels. Multihoming can be an especially critical mechanism of competition in online markets, enabling rivals to overcome the barriers to entry and expansion that scale economies and network effects can create. Multihoming is one way that sellers can reduce their dependence on a single sales channel.
If the Superstore were a condo, the vendors would be free to decide how much to focus on this platform in comparison to other platforms. Merchants would also be freed from the MFN, as the condo owner would not attempt to ban merchants from steering customers to a lower-cost platform.
Condominiumization of the Amazon Superstore would go a long way to reducing what Cory Doctorow might call the “enshittification” of the Amazon Superstore. Given its dominance over merchants, it would probably be necessary to divest and rebrand the “Amazon basics” business. Each participating vendor (retailer or direct selling manufacturer) would share in the ownership of the platform and would have its own place to promote its line of goods or services.
The most challenging issue is how to handle product placement on the overall platform. Given the administrator’s role as the agent of the owners, the administrator should seek to offer a range of options. Or leave it to owners themselves to create joint ventures to promote products. Alternatively, specific premium placement could go to those vendors that value the placement the most, rather than based on who owns the platform. The revenue would in turn be shared among the owners of the condo. Thus, the platform administrator would have as its goal maximizing the value of the platform to all stakeholders. This would also potentially resolve some of the advertising issues. According to the Complaint,
Amazon charges sellers for advertising services. While Amazon also charges sellers other fees, these four types constitute over [redacted] % of the revenue Amazon takes in from sellers. As a practical matter, most sellers must pay these four fees to make a significant volume of sales on Amazon.
Condo ownership would mean that the platform constituents would be able to choose which services they purchase from the platform, thereby escaping the harms of Amazon’s tie-in. Constituents could more efficiently deploy advertising resources because they would not be locked-into or compelled to buy from the platform.
Optimization would include information necessary for customer decision-making. One of the other charges in the Complaint was the deliberate concealment of meaningful product reviews:
Rather than competing to secure recommendations based on quality, Amazon intentionally warped its own algorithms to hide helpful, objective, expert reviews from its shoppers. One Amazon executive reportedly said that “[f]or a lot of people on the team, it was not an Amazonian thing to do,” explaining that “[j]ust putting our badges on those products when we didn’t necessarily earn them seemed a little bit against the customer, as well as anti-competitive.”
Making the platform go condo does not necessarily mean that all goods are treated equally by customers. That is the nature of competition. It would mean that in terms of customer information, however, a condominiumized platform would enable sellers to have equal and nondiscriminatory access to the platform and to be able to promote themselves based upon their non-compelled expenditures.
Turn Amazon’s Fulfillment Center in a Coop
The Coop model envisions shared user ownership, management, and operation of the bottleneck. Such transformation of ownership should change the incentives governing the operation and potential expansion of the bottleneck.
The individual owner-user stands to gain little by trying to impose a monopoly price on users including itself or by restricting access to the bottleneck by new entrants. So long as there are many owners, the primary objective should be to manage the entity so that it operates efficiently and with as much capacity as possible.
This approach is for enterprises that require substantial continued engagement of the participants in the governance of the enterprise. With such shared governance, the enterprise will be developed and operated with the objective of serving the interest of all participants.
The more the bottleneck interacts directly with other aspects of the users’ or suppliers’ activity, the more those parties will benefit from active involvement in the decisions about the nature and scope of the activity. Historically, cooperative grain elevators and creameries provided responses to bottlenecks in agriculture. Contemporary examples could include a computer operating system, an electric transmission system, or social media platform. In each, there are a myriad of choices to be made about design or location or both. Different stakeholders will have different needs and desires. Hence, the challenge is to find a workable balance of interests. That maximizes the overall value of the system for its participants rather than serving only the interests of a single owner.
This method requires that no party or group dominates the decision processes, and all parties recognize their mutual need to make the bottleneck as effective as possible for all users. Enhancing use is a shared goal, and the competing experiences and needs should be negotiated without unilateral action that could devalue the collective enterprise.
As explained above, Amazon tie-in effectively requires that all vendors using its platform must also use Amazon’s fulfillment services. Yet distribution is distinct from online selling. Hence, the distribution system should be structurally separated from the online superstore. Indeed, vendors using the platform condo may not wish to participate in the distribution system regardless of access. Conversely, vendors not using the condo platform might value the fulfillments services for orders received on their platforms. Still other vendors might find multi-homing to be the best option for sales. As the Complaint points out, multi-homing may give rise to other benefits if not locked into Amazon Distribution:
Sellers could multihome more cheaply and easily by using an independent fulfillment provider- a provider not tied to any one marketplace to fulfill orders across multiple marketplaces. Permitting independent fulfillment providers to compete for any order on or off Amazon would enable them to gain scale and lower their costs to sellers. That, in turn, would make independent providers even more attractive to sellers seeking a single, universal provider. All of this would make it easier for sellers to offer items across a variety of outlets, fostering competition and reducing sellers’ dependence on Amazon.
The FTC Complaint alleges that Amazon has monopoly power in its fulfillment services. This is a nationwide complex of specialized warehouses and delivery services. The FTC is apparently asserting that this system has such economies of scale and scope that it occupies a monopoly bottleneck for the distribution of many kinds of consumer goods. If a single firm controlled this monopoly, it would have incentives to engage in exploitative and exclusionary conduct. Our proposed remedy to this is a cooperative model. Then, the goal of the owners is to minimize the costs of providing the necessary service. These users would need to be more directly involved in the operation of the distribution system as a whole to ensure its development and operation as an efficient distribution network.
Indeed, its users might not be exclusively users of the condominiumized platform. Like other cooperatives, the proposal is that those who want to use the service would join and then participate in the management of the service. Separating distribution from the selling platform would also enhance competition between sellers who opt to use the cooperative distribution and those that do not. For those that join the distribution cooperative, the ability to engage in the tailoring of those distribution services without the anticompetitive constraints created by its former owner (Amazon) would likely result in reduced delivery costs.
Separation of Fulfillment from Superstore Is Essential for Both Models
We propose some remedies to the problems articulated in the FTC’s Amazon Complaint—at least the redacted version. Thus, we end with some caveats.
First, we do not have access to the unredacted Complaint. Thus, to the extent that additional information might make either of our remedies improbable, we certainly do not have access to that information as of now.
Second, these condo and cooperative proposals go hand in hand with other structural remedies. There should be separation of the Fulfillment services from the Superstore and Amazon Brands might have to be divested or restructured. Moreover, their recombination should be permanently prohibited. These are necessary conditions for both remedies to function properly.
Third, in both the condo and coop model, governance structures must be in place to assure that both fulfillment services and the Superstore are not recaptured by a dominant player. In most instances, a proper governance structure would bar that. The government should not hesitate to step in should capture be evident.
Peter C. Carstensen is a Professor of Law Emeritus at the Law School of University of Wisconsin-Madison. Darren Bush is Professor of Law at University of Houston Law Center.