Just before Christmas, furniture maker Williams-Sonoma filed a lawsuit against Amazon in the Northern District of California, alleging that the e-commerce giant copied furniture designs and abused its dominant tech platform to gobble up a new product line.
Analysts had deemed Williams-Sonoma less exposed to an Amazon invasion because the furniture maker’s West Elm line was more suited to in-store purchasing. Turns out no one is immune to the great gobbling.
In addition to trademark infringement claims, the lawsuit alleges what we already knew—that Amazon can (1) predict what its customers want based on its vast and proprietary database of prior purchasing and search behavior on the Amazon site; and (2) offer similar products to those of an independent merchant and steer customers to its private label via its discriminatory search algorithm. In this second sense, what Amazon does is similar to how Google discriminates in favor of its own content in local search. Exploiting one’s platform advantages to overtake a vertical is the opposite of competition on the merits.
The Williams-Sonoma lawsuit also alleges something that at least your fearless blogger didn’t know—that (3) Amazon advertises its private labels on Google using the same keywords shoppers use on Amazon, raising an independent rival’s costs to advertise on Google. As Sally Hubbard observes in The Capitol Forum, not only does Amazon control its own marketplace and uses that to favor its own goods, it also uses Google’s platform for the same purpose. Indeed, a former Amazon manager told Hubbard that “Google gives Amazon higher ranking than a brand itself can get” under Google’s search algorithm, which rewards Amazon’s high-traffic website. It’s as if the dominant platforms are coordinating in their assault on independent merchants! Gobble gobble.
The complaint also includes an unfair competition claim under the Lanham Act, as well as a common law unfair competition claim. Although claims for unfair competition imply that the alleged conduct harms competition, the complaint notably never asserts a violation of the federal antitrust laws, and it never uses the word “anticompetitive.”
A Gap in Antitrust Law
Nor could it. As Lina Khan explained in her pathbreaking law review article, Amazon’s predatory conduct raises anticompetitive concerns, but by fixating on short-term price effects, antitrust law fails to grasp them. Moreover, leveraging one monopoly into another or “monopoly leveraging theory” is arguably a dead letter in antitrust, except possibly when used to maintain a monopoly. But Amazon’s leveraging of its platform is designed to scoop up profits in ancillary markets, not to fend off a future e-commerce platform rival. Thus, the lack of an antitrust claim in the lawsuit is not a deficiency of the complaint, but instead reveals a gap in our competition laws.
Amazon has a history of vertically integrating to take profits from leading “complementors” to its platform. The dominant e-commerce website famously invaded verticals occupied by Diapers.com (diapers), BareBones WorkWear (clothing), and Beauty Bridge (cosmetics). Unlike the “pure appropriation” case asserted by Williams-Sonoma, however, these three trampled edge providers sold their products on Amazon before the appropriation; thus, Amazon followed its appropriation by discriminating against the complementor and in favor of its branded clone—for example, by putting affiliated cosmetic items in its “buy box” or at the top of the buying choices page. This difference proves critical in remedy design. More on that in a bit.
According to TJI Research, Amazon had more than 120 brands as of September 2018, most of which were introduced over the past two years. In October 2018, Amazon began selling AmazonBasics-branded memory foam mattresses, which compress into a box for easy shipping. The price of Amazon’s 12-inch California King was only $330, a considerable discount from prices for comparable mattresses sold by independents such as Leesa on Amazon’s site. Who can be opposed to lower prices?
Alas, the competitive injury here is not higher prices, but instead is decreased edge innovation. Even if you don’t think of mattresses in a box or designer chairs as the most innovative offerings, that Amazon can appropriate any innovator’s design and privilege its own brand—what Hubbard has coined “platform privilege”—is problematic given Amazon’s control of nearly 50 percent of U.S. retail e-commerce sales and given the likely spillover effects to other complementors who observe the appropriation. As noted by Professor J.P. Eggers of NYU’s Stern School of Business, the “effect of Amazon imitation should be to decrease innovation in categories sold on Amazon, as firms recognize imitation risk.” Gobble gobble gobble.
Edge innovation is notoriously hard to measure, but a decent proxy is firm exit by complementors. Economists Feng Zhu of Harvard University and Qihong Liu of the University of Oklahoma found that Amazon entered several new product lines (nearly 5,000 items) in a short span in 2013-14, and that those invasions had the effect of chasing independents off the platform. In particular, they found that “affected” sellers on Amazon’s platform, against which Amazon competes directly, reduce the number of products offered on Amazon by 24.1 percent relative to unaffected sellers. That an affected seller chased off Amazon might still exist as a shell of its former self does not repair the dampened incentives of future innovators.
Choosing Among Imperfect Remedies
There are basically three types of interventions to protect edge innovation from Amazon’s scheme: (1) expand intellectual property (IP) protections, (2) beef up antitrust enforcement, or (3) offer a new protection outside of IP and antitrust. This post will ignore IP-based fixes, except to note the obvious—one brute-force solution to the appropriation of IP is to extend the scope of IP protection. The effect of an IP adjustment could be to harm innovation in other ways, and such a move could have far-reaching unintended consequences.
An antitrust intervention could take the form of structural relief, which as the name suggests, entails changing the structure of the firm. Applied here, structural separation would entail drawing a pen around Amazon’s (or Google’s or Facebook’s) core mission, and barring Amazon from straying outside said pen. A structural remedy could be achieved as the culmination of a successful antitrust suit against Amazon, or via legislative fiat. The breakup of AT&T is the precedent for the antitrust path, and the Glass-Steagall Act, which prevented securities firms and investment banks from taking deposits a la commercial banks, is the template for the legislative path.
The antitrust path to an Amazon breakup is fraught with peril, however, as judges increasingly require antitrust plaintiffs to show a tangible consumer injury, which (with the possible exception of degraded quality of search) isn’t how the competitive harm manifests itself here. And even if the Federal Trade Commission (FTC) could prevail under the exacting consumer-welfare standard, the snail’s pace of antitrust would ensure that relief would not be obtained for several years (approaching a decade with appeals), at which point edge innovation might be irreparably harmed.
While the legislative path to a breakup overcomes the timeliness challenge, the structural remedy still raises the tricky question of how to draw the lines around Amazon’s core mission. Should Amazon be barred, for example, from competing against Google in selling video or display ads across the web? Many competition scholars, including Chicago’s Randy Picker, would welcome that form of competition, yet it clearly crosses the boundaries of a pure e-commerce platform provider. The structural approach is challenging precisely because it requires an express parsing of firm identity (or core mission), which is so easily gameable, as we’ve seen with tech platforms such as Uber.
Perhaps the line-of-business restriction could apply only to markets that are directly adjacent to a market in which a platform is dominant. That still leaves some issues. For example, should Amazon be permitted to offer products that are not yet sold by third parties on its platform? Should Google be allowed to offer answers to math problems or provide other commodity-like content in search results? Even after one solves the line-drawing exercise, a structural remedy would deny any economies of scope that exist between serving as the platform and producing content on the platform.
A less-invasive alternative to a structural solution is a nondiscrimination regime. In contrast to structural barriers, Amazon (or any dominant tech platform) would be free to appropriate a complementor’s innovation subject to any limitation in IP law. Under this complaint-driven process, independent merchants against whom Amazon discriminated could lodge a complaint at the FTC or some independent tribunal. (The ultimate framework could require that the factfinder determine whether the appropriated content or merchandise was a commodity and thus not an immediate threat to innovation; or it could provide relief even to commodity content under the rationale that any discrimination could harm innovation via spillover effects for other, innovative content. It bears noting that the nondiscrimination protections of the Cable Act did not create a safe harbor for copying and then discriminating in favor of affiliated, commodity-like programming such as home shopping.) This remedy would also require legislation, as the FTC lacks a nondiscrimination standard to enforce, and independent merchants lack a private right of action to bring a discrimination complaint.
When asked by Antitrust Subcommittee Chairman David Cicilline (D-RI) on December 11, 2018 whether he could abide by a nondiscrimination standard “designed to ensure that online firms with significant market power cannot harm the competitive process via discriminatory conduct,” Google’s CEO Sundar Pichai testified that “he is happy to engage constructively” with Congress on legislation. (See 2:08 at the Hearing Video). Presumably Amazon could abide as well.
As with structural approaches, the nondiscrimination regime also has its challenges. An independent merchant could not avail itself of the protections unless it sought access to Amazon’s platform, as disparate treatment of similarly situated content is a key element in the proof of discrimination. According to its complaint at paragraph 22, Williams-Sonoma has never sold its products through Amazon or any other online retail channel. Thus, while Diapers.com, BareBones WorkWear, Beauty Bridge and other complementors operating on Amazon’s site could seek relief at the tribunal from discrimination, Williams-Sonoma could not. This is an unfortunate gap in coverage of the nondiscrimination standard, but perhaps the risk of appropriation outside of the platform is better addressed via IP law.
Another demerit, relative to structural separation, is that the resources required to litigate a discrimination case against Amazon would not be trivial. Thus, the smallest Internet startups might not be able to avail themselves of protections under a complaint-driven process. It would be up to midsized and larger complementors to enforce the nondiscrimination regime.
No remedy is perfect. But it’s high time for Congress to do something about Amazon. A few more years without an intervention—even an imperfect one—and edge innovation could be gobble gobble gone.