Overview of the DC, California, and FTC Amazon Cases
Amazon operates the world’s largest online marketplace, accounting for significantly more sales share and number of sellers operating on its platform than any other online marketplace. There are two kinds of sellers on Amazon: (a) third-party sellers (TPSs) who go on to Amazon and sell their products directly to consumers; and (b) first-party sellers who sell their products to Amazon, who in turn sells those products on Amazon’s marketplace. Amazon, therefore, is both a seller on its marketplace competing with third-party sellers, and it operates the marketplace and collects fees from third-party sellers for sales that occur on its marketplace. Today, the majority of Amazon’s sales are through third-party sellers.
The District of Columbia, the State of California, and the FTC have separately filed suits challenging Amazon’s MFN-like policies on its marketplace. These lawsuits claim that third-party sellers are highly dependent on Amazon and that every time a third-party seller makes a sale on Amazon’s platform, Amazon receives about 30% to 40% of sales, as a commission. These payments to Amazon elevate sellers’ costs, which are reflected in the prices charged to consumers.
Sellers and Amazon monitor prices charged for the sellers’ products on other marketplaces to ensure compliance with Amazon’s policy. For example, according to the DC Complaint, rival marketplaces like Walmart “routinely field requests from merchants to raise prices on Walmart’s online retail platform because the merchants worry a lower price on Walmart will jeopardize their status on Amazon.” According to the FTC Complaint, whether done contractually or algorithmically, by monitoring prices and punishing sellers that lower prices elsewhere, Amazon requires sellers to keep prices off Amazon “as high or higher than prices on Amazon.” All three Complaints allege that Amazon’s MFN-like policy and related conduct raises prices on rival platforms even though those platforms charge sellers lower fees for selling through them.
The DC Complaint, brought under DC law, challenges Amazon’s MFNs relating to third-party sellers as both horizontal and vertical restraints of trade: (a) “Amazon and its TPSs unlawfully agree that TPSs will not offer their products through other competing online marketplaces at lower prices than the prices they offer them on Amazon’s marketplace;” (b) “Amazon has willfully maintained and enhanced its market power through its anticompetitive and exclusionary conduct;” and (c) Amazon is attempting “to monopolize the market for online marketplaces, control prices, exclude competitors, and suppress competition and innovation among online marketplaces.” The California Complaint, brought under the Cartwright Act, challenges Amazon’s MFNs related to third-party sellers as a horizontal restraint of trade: “Amazon has entered into contracts and/or combinations with its third-party sellers and wholesale suppliers for the purpose and effect to create and carry out restrictions in trade or commerce.” California also brings claims of unfair competition. The FTC Complaint alleges monopoly maintenance in the “online superstore market,” and in the “online marketplace services market,” under Section 2 of the Sherman Act and various state laws. The FTC also challenges Amazon’s “anti-discounting practices” and its manipulation of other online stores’ pricing algorithms into matching Amazon’s price increases under Section 5(a) of the FTC Act.
Economics of MFNs
MFNs are among a class of business practices known as vertical restraints. They typically appear in vertical agreements between firms operating at different levels of production or distribution. For example, a vertical agreement between a buyer (like Amazon) and a seller (like a retailer that sells through Amazon) might require the seller to assure the buyer that it is paying the lowest price given by the seller to any buyer.
During the panel, Dr. Das Varma explained that MFNs help ensure a trading partner is guaranteed specific beneficial terms received by other trading partners, without the need to overhaul the original trading agreement. Consider, for example, a price MFN that prohibits a seller from selling its product at a lower price to another buyer; if the seller sells its product to another buyer at a lower price, under its MFN with the first buyer, the seller would be obligated to give the first buyer the benefit of the same lower price. Similarly, a platform MFN is an agreement between sellers and platforms about the terms at which the sellers’ products are offered to buyers who purchase through rival platforms. For example, a platform MFN may prohibit a seller from offering products or services at a lower price on rival platforms. As Baker and Scott Morton explain, there are two types of platform MFNs: (a) wide MFNs constrain prices on all platforms, including the seller’s own website; (b) narrow MFNs constrain prices only on the seller’s own website. Based on the descriptions in the complaints, the Amazon MFN appears to be a wide MFN.
Depending on when and how they are used, MFNs can be either procompetitive or anticompetitive. Economists have theorized that MFNs can lower the transactions costs associated with negotiating prices and reduce buyer risk arising from uncertainty in valuing the seller’s product or from sellers acting opportunistically. For example, a buyer can invest in a downstream product or service without risking the possibility that the seller will take actions (like selling its product for less to a competing buyer) that would erode the value of the seller’s investment. In these situations, MFNs can be procompetitive.
Economists have also theorized that MFNs change a firm’s incentive, and under certain circumstances, they can: (a) facilitate coordination and dampen competition among non-coordinating rivals; and (b) discourage competition from entrants. In these situations, MFNs can be anticompetitive.
The economics literature recognizes potential efficiency rationales for MFNs. The first of these is that MFNs can address free-riding or the holdup-problem, which can erode investment incentives. Dr. Das Varma explained that some consumers might use Amazon to learn about a product and, absent its MFN policy, go to a rival platform like Walmart to buy at a lower price. Amazon needs to make the sale, or at least sufficient sales, to recoup its investment in information, otherwise it will not invest as much in the future. Other efficiency rationales involve reducing delays in transactions and reducing transaction costs.
How can a wide MFN facilitate coordination among rivals? Dr. Das Varma explained that an MFN makes price collusion among sellers more durable because it reduces the incentive of a participating sellers to lower price. Suppose there are several sellers involved in a price fixing agreement, and one of the sellers is contemplating deviating (or cheating) on the agreement. Under an MFN between the deviating seller and a platform the seller uses to reach customers, the deviating seller wishing to offer a discount to customers on one platform would also have to offer the discount to customer of the other platform. By preventing the deviating firm from selectively offering discounts on a rival platform, the wide MFN makes deviating from the agreement less attractive and thus makes the collusive scheme more durable. The bigger the platform imposing the wide MFN, the stronger this effect—because the greater will be the cost of offering a discount.
Discouraging Competition from Entrants
How can a wide MFN discourage platform entrants or smaller platform rivals from competing? The answer comes from understanding a platform’s incentives to reduce price (referring to the fees it charges merchants selling through its platform) or invest in quality. A platform lowers price or invests in quality to generate greater sales and higher profits. Therefore, any contract between market participants that prevents a platform from gaining share when it cuts its price can potentially discourage competition. In the context of the Amazon MFN, Dr. Das Varma explained that a rival platform like Walmart would want to reduce the fees it charges merchants that want to sell on Walmart.com only if the lower fee translates to lower a price at which the merchant sells there, so that Walmart.com can gain sales relative to Amazon. Amazon’s MFN policy may prevent merchants from lowering prices at Walmart, which in turn would dis-incentivize Walmart from reducing its own fees or investing in its website because it is less likely to recoup those investments through additional sales.
One can imagine a merchant that sells on a platform, also known as a purchaser of platform services, wishing to pay lower platform fees, would want to drive sales away from a dominant platform to a rival platform that charges it a lower commission for use of its platform. The obvious way to drive sales to the rival platform is to sell the product on the rival platform for less than on the dominant platform. So doing creates both short-term and long-term benefits for the merchant. In the short-term, by driving sales to the lower-cost rival platform, the seller can lower costs and potentially increase profits. In the long-term, the greater rivalry among the platforms will force the dominant platform to lower its fees. The wide MFN can discourage entry and expansion by rival platforms, and thereby allow the dominant platform to obtain or maintain market power.
Mr. Kass suggested the defense of Amazon’s MFN policy would be easy because it boils down to Amazon asking for the most favorable terms in exchange for giving sellers access to a large volume of loyal customers who love Amazon because Amazon offers them either the best price or close to the best price. The other panelists disagreed with Mr. Kass in several ways. Highlights of the discussion are summarized here.
Mechanism of Harm
According to Mr. Kass, the Amazon MFN policy amounts to Amazon asking for the lowest price, which cannot be anticompetitive, and the MFN helps Amazon to offer the best (or close to best) price to customers who shop on its marketplace. Mr. Kass analogizes that Amazon asking for the lowest price is akin to the familiar volume discount associated with buying (in first-party seller policy) and selling (in third-party seller policy) large volumes of merchandise. In his view, Amazon is simply demanding the lowest purchase price because it is a big buyer (from first-party sellers) and the lowest price seller on its platform (from third-party sellers) because it is delivering the most sales.
Focusing on Amazon’s MFN policy for third-party sellers, Ms. Bojedla and Dr. Das Varma pointed out that Mr. Kass’s arguments ignored the potential significance of price structure. They emphasized that there are actually two sets of prices at issue: (a) the prices at which third-party sellers sell their products on platforms, which is the price on which Mr. Kass focuses; and (b) the price the platform charges the third-party seller for access to the platform’s users. The latter is the cost of selling that the former has to cover. According to the DC and California complaints, Amazon charges merchants a significantly higher price (~30 to 40%) to use is platform relative to other platforms (some of whom charge nothing). Mr. Kass dismissed the significance of the second price as simply a distribution fee that merchants think is worth paying and that allows Amazon to cover its costs.
Dr. Das Varma explained the significance of the price structure. He pointed out that under a simple MFN, where the seller agrees to give the buyer the lowest price, the buyer plays no role in influencing the seller’s cost. Under a platform MFN, the seller agrees to sell at the lowest price on the platform and has to pay the platform a fee (the second price described above). These fees are selling costs for the seller, and the seller has to recoup these costs. So, by setting a high fee and requiring purchase price parity, the Amazon policy “externalizes” Amazon’s profits to other platforms. In other words, all else equal, a seller would tend to set higher prices on Amazon compared to rival platforms, because the seller has higher selling costs on Amazon. Conversely, because sellers have lower selling costs on rival platforms, they would tend to set lower prices there, but the Amazon policy props up the prices.
Mr. Sallet explained that the economics of MFNs make it clear that they have the capability of making prices higher. He quoted National Society of Professional Engineers, which in quoting older cases, said that “price is the central nervous system of the economy” and that “an agreement that interferes with the setting of price by free market forces is illegal on its face.” That is not to say that all conduct that affects price is necessarily per se illegal. Rather, he argued that the statutory policy of the antitrust laws precludes an inquiry into the question of whether competition is good or bad. Mr. Sallet (and others) agreed that, ultimately, there would need to be a factual investigation to see whether and how alleged conduct impacts prices.
Finally, on Mr. Kass’s comments regarding volume discounts, Mr. Sallet cautioned that one would not want to claim the fruits of monopoly maintenance as a justification for the conduct—for example if a company with 95% market share says that as a result of exclusionary conduct it can lower prices to consumers and maintain its monopoly. He would not at the outset regard such a defense as consistent with Section 2 of the Sherman Act.
Horizontal Theory of Harm
Mr. Kass questioned California’s horizontal theory of harm, which he says relies on Amazon being a horizontal competitor to the third-party merchants that sell on Amazon, because Amazon is both a reseller and a platform provider. He described this aspect of the complaint as a “nefarious spin on platform economics.” He said the difference between Amazon as the reseller and Amazon as the platform provider is simply about whether Amazon takes title, which is a decision that reflects ordinary business issues like taxes, insurance, revenue recognition, and the like. If Amazon takes title, it is a reseller; if it doesn’t, it is a third-party platform provider. Amazon’s decision to take title, Mr. Kass observed, has nothing to do with antitrust and antitrust should not care.
Mr. Sallet explained that all harm is horizontal. There are vertical mechanisms and horizontal mechanisms, but the harm is horizontal; for example, in an input-foreclosure claim to some group of downstream rivals. He said what the California Complaint does is expressly plead the case as horizontal and it asserts that it does so in its Complaint for the following reason: “Amazon’s third-party sellers and wholesale suppliers are not just vertically situated inputs to Amazon’s online retail stores, they are individually and collectively a powerfully horizontally-situated threat to Amazon because they can and do sell their products directly.” Mr. Sallet noted that he was not opining on the validity of the claim, particularly because he has no knowledge of the product markets being pled, but that California has put forward a theory that he thinks will be important to test in court.
Treatment of Vertical Restraints
Mr. Kass observed that the MFN is a vertical restraint, and GTE Sylvania decided that vertical restraints like these are generally “okay” because Amazon’s interests are aligned with consumers’ interests. Amazon wants to sell a lot and consumers want to buy a lot, and manufacturers are looking for the best distribution alternative. Mr. Kass argued that if manufacturers dislike it, they can avoid it by going to other platforms. To suggest otherwise, he said, is basically like saying Amazon is an essential facility, a concept that has been discredited by the Supreme Court.
Mr. Sallet noted that, in his view, there has been too much skepticism about vertical claims, and that there has been good analysis in cases like McWane v. FTC. Ultimately, he said he sees no reason why multiple theories cannot be pled, including horizontal theories, all of which can fit the facts.
Standard of Proof for the But-for World
Mr. Kass suggested that the benefits of Amazon’s MFN policy are tangible, and the harm is speculative. He expects the plaintiffs to argue that the actual price a seller charges consumers on Amazon is higher than an imaginary price that would have prevailed absent Amazon’s MFN policy, suggesting that this price would be hard to defend.
Mr. Sallet explained that the questions that need to be asked are: what are the systemic effects across a properly defined product market, and what would happen if there were greater competition? He said it was particularly important in a monopoly maintenance case to ask what a competitive world would look like. It was his view that it is the policy of the Sherman Act that price competition has been limited by exclusionary conduct, so the presumption should be that there is harm to competition and that it should not be up to antitrust enforcers or big companies to say what competition might or should be for a consumer. If the choice of consumers is artificially constricted by dominant monopoly power in a way that is not justified by procompetitive benefits, he said the law tells us that there has been harm to competition. It was also his view that exactly how competition had been harmed does not have to be proved to the fifth decimal point of specificity.
Post-Mortem from the Moderator
These cases involve more than a “simple MFN” where a buyer asks for a lowest price assurance. Here, the buyer or platform (Amazon) attaining the price assurance also controls the seller’s cost by charging a platform usage fee, and by so doing, may have the ability to elevate prices on rival platforms. Analysis of competitive effects will need to assess the implications of this price structure. For example, plaintiffs may argue that this price structure limits the ability of entrants with low-cost business models to establish themselves and grow by charging sellers lower fees, relative to the incumbent platform.
In addition, it may be possible to recreate the effects of an MFN without the explicit use of an MFN. Faced with the right combination of financial penalties and rewards, a seller may choose to give a buyer the lowest price, sometimes by raising the price to other buyers. Plaintiffs will argue that the financial incentives that give rise to such a choice are akin to an MFN, even though the seller is not “required” to give the buyer the lowest price.
In turn, Amazon may argue that its MFN (or MFN-like) policies benefit consumers in the largest marketplace by giving them assurance of the lowest prices. Amazon may argue that its pricing policy allows them to recoup investments in the Amazon platform that might otherwise would not have occured. Amazon may also argue that at least some of its higher platform usage fee stems from the higher value consumers place on the greater breadth and higher quality of one-stop searching convenience that Amazon’s marketplace provides.
Like with most cases, the success of these Amazon cases will depend on the ability (or inability) to tie the theory with the evidence of competitive effects. Identification of the but-for world, and sometimes the but-for price, is inherently elusive, but that does not mean one cannot gain a sense of directional effect or even begin to gain some sense of the magnitude of the effect. Here are some ideas of areas of empirical work that could potentially prove useful:
- What are the lessons from other industries where firms stopped imposing wide MFNs?
- How often and under what circumstances do Amazon’s third-party sellers ask rival platforms not to discount their prices?
- How often and under what circumstances do third-party sellers pull their products from other retail sites, out of concern about monitoring prices to comply with Amazon’s MFN?
- How often and in what circumstances does Amazon enforce its MFN?
- Are there any differences in the price distributions for sellers for whom Amazon is an important distribution platform versus others where Amazon is relatively less important?
- Are there investments that Amazon would not have made but-for the price assurances?
- What relationship, if any, is there between increases in platform usage fees and investments in the platform?
Of course, this list should be informed by the available discovery record. Ultimately, this is the type of work that would have to be undertaken to evaluate the competitive effects of the Amazon MFN, because as a matter of economics, asking for the lowest price through an MFN can discourage competition in some circumstances.