Online marketplaces are online retail platforms on which independent sellers price their products and share proceeds with the platform. The Amazon Marketplace, eBay, and the iOS and Android app stores are examples. This article addresses how online marketplaces differ from other types of retailers and the resulting implications for assessing the economics of retailer policies, such as the offering of private-label goods (vertical integration). In turn, the analysis suggests that concerns raised by EU and U.S. officials about the dual role of online retailers (who sell their own products alongside those of third-party sellers) cannot be immediately dismissed as harmless by analogy to the practices of brick and mortar retailers.
Here, I explore a theory of harm to competition based on reduced innovation incentives by sellers faced with vertically integrated platforms. For example, in this context, on-platform advertising systems may in some cases work to squeeze the profits of sellers rather than improve competition. Vertical integration by platforms may also require sellers to increase advertising expenditures, further squeezing seller profits.
Yet, these same essential differences between online marketplaces and other types of retailers might instead lead to substantial consumer benefits. Before assessing the prospects for both procompetitive and anticompetitive outcomes from platform vertical integration, I begin by describing the three essential differences between online marketplaces and other retailers. The first relates to ease of selling and market intelligence, the second to on-platform advertising, and the third to data practices.
It can be very easy to sell your product on an online marketplace. For example, the Amazon Marketplace reportedly has over 2 million active sellers, while Google Play and the Apple App Store offer about 2.5 million and 1.8 million apps, respectively. There are several reasons why it is so easy to sell on an online marketplace.
First, selling on an online marketplace is a seller-initiated decision—the marketplace itself need not have ever heard of the product, much less have any sense of the likely demand for the product or its optimal pricing. Rather, on sites such as eBay and the Amazon Marketplace a seller simply completes an online process and can then begin selling the product on the platform. The seller itself is in charge of making retail pricing decisions on the marketplace, lowering the informational requirement of the retail platform.
Second, unlike brick and mortar retailers, online marketplaces face much softer or non-existent constraints on physical shelf space. This feature makes it feasible for a platform to indeed become a marketplace where all sellers are welcome and where upfront fees to gain access to shelf space are often small or absent. For example, globally distributing an app on the Apple App Store requires enrollment in the Apple Developer Program, which carries a $99 annual fee (additional fees may be required, for example commission fees contingent on sales if an app is not free to consumers, but the main point is that the upfront fee is small). In contrast, some brick and mortar retailers charge upfront slotting fees in the thousands of dollars merely for a product to be carried regionally.
The ease of seller access provided by an online marketplace may affect the composition of products sold there, leading to a different product mix compared to traditional retailers. This mix may be relatively skewed towards newer, lesser known, or more niche products.
Consider the Amazon Marketplace’s advice to sellers (sell.amazon.com/sell.html): “Whether you’ve already got an established ecommerce business, a great idea for a new product, or you just have a passion for selling, here’s how to take that next step with Amazon.” Amazon explicitly encourages sellers to find existing products that are made by others and to resell them on the Marketplace, in addition to encouraging them to sell their own products: “Resellers find popular products that already exist and offer them in Amazon’s stores. Brand owners manufacture their own products—or source goods to sell under a private label—to offer shoppers unique selection. Lots of sellers do both. You can choose whichever method works for your goals.”
By allowing parties to resell the products of other firms, the number of products on the marketplace naturally increases. This further solidifies the role of a marketplace as a tool for market intelligence that helps the marketplace become quickly aware of new products and new product categories.
If an online marketplace succeeds in developing a large base of sellers then it will likely need a way to help consumers search for or otherwise discover those products. In addition to organic search, a marketplace has the option of letting sellers pay an advertising fee to boost the firm’s ranking in consumer search results. For example, various types of ads are sold by Amazon, including sponsored ads that are served to consumers based on keyword searches, ads placed in front of consumers based on their previous shopping history at Amazon or other metrics, and even ads placed on websites other than Amazon itself.
Fees from such ads have helped Amazon become the third largest online advertising intermediary by revenue in the United States, behind only Google and Facebook. These advertising services are heavily promoted by Amazon. For example, the webpage for sellers interested in the Amazon Marketplace (sell.amazon.com/advertising.html) includes quotes from sellers such as, “Amazon Advertising is an essential part of our growth strategy.”
Compared to a brick and mortar retail setting, the advertising and promotional spending in online marketplaces is most analogous to the fees sellers pay to have their products carried in premium or eye-level shelf space. However, due to the greatly expanded ability to target ads in an online marketplace, there may be some important differences in terms of which sellers end up paying advertising fees.
To understand this distinction, consider a hypothetical area of a traditional brick and mortar grocery store that I will call the “salsa aisle.” The salsa aisle has premium eye-level shelf space available and also non-premium shelf space that displays products somewhat less prominently. Consider a hypothetical niche producer of salsa, selling organic salsa that is heavy on chipotle. Its main rivals are also salsas that are either organic, heavy on chipotle, or both. But being a niche category, neither this brand nor its main rivals pay for premium shelf space, instead ceding that to larger national brands selling more mainstream salsa varieties.
In an online environment, however, these niche products may need to compete against each other for advertising. The reason is that a consumer whose search terms or past purchasing behavior indicate an interest in organic chipotle salsa can now be auctioned off to brands selling such salsa. To avoid falling behind their organic chipotle rivals in sales, each brand may find it must compete for scarce ad slots. Importantly, despite the seemingly competitive nature of this advertising, the primary end effect may be to transfer profits from salsa manufacturers to the platform, as opposed to improving the efficiency of consumers’ product selection.
But how can increased competition for advertising placement not benefit consumers? Suppose that consumers know when they are in the mood for organic chipotle salsa. In a brick and mortar store they would scan the lower shelves (having not found organic chipotle salsa on the premium shelves), spying one or more of the relevant products and then making a choice, sometimes seeing one relevant brand first and being more likely to buy that brand, and sometimes seeing a different relevant brand first and being more likely to buy that brand.
In the online environment, effectively the same thing happens, only now which brand is seen first is not a random consequence of how a consumer’s eyes moved or which packaging stood out to that consumer. Instead, which brand is seen first is determined by whichever supplier paid an advertising fee (specifically, it may have bid higher in a pertinent keyword auction). In the online equilibrium, then, it may well be that most firms end up engaging in additional advertising, and sometimes one is shown first and sometimes another—more or less the same as in the brick and mortar world. This suggests that while online advertising need not substantially alter which products consumers ultimately purchase compared to the brick and mortar environment (or, absent paid search), sellers nonetheless fork over much more money to the retailer (not to the consumer!) in advertising fees than they otherwise would.
In sum, online marketplaces may allow for much better monetization of what is promoted to consumers, taking what brick and mortar retailers do to a new level. It is as if for each consumer and product niche, the online marketplace can create specialty shelf space featuring only the most relevant products. The platform can then induce advertising spending for premium placement on this specialty shelf. But the marketplace may well have been able to deduce the most relevant products and display them to consumers using organic search results instead. Thus, at least according to this one view, suppliers achieve lower profits, the platform achieves higher profits, but there is no definite improvement on overall consumer wellbeing.
Although any retailer may be aware of basic information, such as the prices and quantity sold of products featured in their stores, online marketplaces potentially have access to much more detailed information that is germane to the vertical integration decision. This is especially true for online marketplaces that succeed in vastly growing the number of sellers and available categories.
As already noted, one source of information comes from the ease with which firms can sell their own products on a marketplace. This allows the marketplace quickly to become aware of new products and new product categories, and more generally understand all the available options for consumers in a given category. This information can be quite valuable for the platform’s vertical integration decision. As explained above, even if the owner of a branded product does not wish to sell on the online marketplace, little prevents a different firm from acquiring the good and then reselling it on the marketplace. Thus, it is hard for a single owner to prevent the marketplace from gathering information about its product.
Additional information can come from several sources. First, advertising expenditures provide insight into the underlying variable profit margin of the seller. For example, suppose a seller is charging $20 for a product and submits a bid of $1 for an advertisement. This fee is only paid when a shopper clicks on the ad. Suppose 20 percent of clicks convert to a sale. Then a click on average generates 20 percent of $20, or $4 in revenue for this product. Because the seller was willing to pay $1 for the ad, this means that the variable costs of selling the product might reasonably be considered to be less than $3. This knowledge naturally may be useful to a retailer considering entering the category.
A second source of additional information is from consumer shopping behavior. This data can help an online store evaluate other products that are considered by consumers in a given category, how much time is spent considering them, and what consumers ultimately buy. Overall, on-site shopping behavior provides information about structural features of supply and demand. Most simply, the platform may be aware of how many alternatives are seriously considered by most buyers. A lack of alternatives might signal an opportunity for entry.
This type of information typically is not available to brick and mortar firms. The ability of such firms to track consumer behavior is more limited, and they may not have the same market intelligence as an expansive online marketplace. Indeed, the broader view of what is taking place in different categories may give online marketplaces privileged access to category-level information such as whether aggregate category demand is trending upwards.
On-platform product reviews provide a third source of additional information. These reviews provide valuable information for a firm considering entering a category. For example, reviews might suggest what other products are good substitutes, as access to shopping behavior data does. Reviews might also suggest why consumers buy various products, or what causes consumers to not buy certain products. Brick and mortar retailers do not typically operate review systems for products they sell. In principle, such stores could access publicly available reviews posted elsewhere (such as on an online marketplace). However, because such reviews would not be from customers who purchased at the brick and mortar outlet they might not be as useful as on-platform reviews are to the online marketplace. The reason is that the types of consumers at the brick and mortar outlet might not be the same as those who purchased online. And, a platform potentially stores consumer reviews in a format that is easier for it to directly and systematically search and analyze, again making such reviews more valuable for the platform than for other retailers.
There are numerous complex forces at work when considering vertical integration by an online marketplace, and limited evidence at this time regarding the likely effects. Here, I discuss some theoretical possibilities regarding the economic effects of such practices, beginning with possible anticompetitive effects and then turning to possible procompetitive effects.
Anticompetitive Effects. A worst-case scenario looks something like the following. The online marketplace invites sellers (especially smaller sellers offering new, lesser known, or niche products) in with the promise of easy access to consumers and minimal upfront carriage fees. This helps gather market intelligence. The marketplace then betrays this promise by quickly introducing its own competing product variant in an effort not only to gain sales but also to intensify the on-platform advertising battle in an attempt to extract a greater share of sellers’ profits for itself. Existing empirical research supports the intuition that more competition raises on-platform advertising revenues from keyword auctions and that firms sometimes bid on keywords most closely associated with rivals.
The incentive and opportunity to engage in such an “advertising price squeeze” to drive ad revenues may encourage the marketplace to self-preference its own products or otherwise alter the organic search process of the platform, especially if so doing further encourages ad spending. The related decisions about whether to vertically integrate and how to design a new product (and potentially other ongoing business decisions) are also guided by the marketplace’s resulting superior data, which allow it to identify categories that are vulnerable to new entry. Collectively, these platform decisions may work to raise the costs of independent sellers, as in the literature on raising rivals’ costs.
In this worst-case scenario, sellers certainly may be harmed by vertical integration. Whether it is good policy to protect firms from the threat of retailer vertical integration is a matter of debate. But, at a minimum, the discussion should accept that the composition of affected firms may differ in an online marketplace setting compared to traditional retailer settings. If it is true that smaller, innovative firms in lesser known or niche categories are harmed by the operational procedures and vertical integration decisions of online marketplaces, and if this harm is greater than in a traditional retail setting, then the prospect of harm to rivals may support policies aimed at protecting independent sellers. From a legal standpoint, one possibility is that vertical integration and the other marketplace practices described above might, together, be viewed as exploitative or unfair to sellers or certain classes of sellers—to the extent that falls into a cognizable claim.
One counterargument to the idea that sellers are harmed by the actions of an online marketplace is that they would not engage with the platform if doing so would ultimately harm them. This argument is suspect for at least two reasons. First, as already discussed, even if a brand owner chooses not to sell on a marketplace, an intermediary may step in and resell the product on the marketplace after adding its own commission. Indeed, as pointed out above, Amazon itself encourages this behavior. Second, sellers, especially smaller ones, may not understand how their own decisions about selling on a marketplace may lead to the creation of more competition in the form of private-label products. This may be especially true if they do not understand that the data that a marketplace may gain access to (different from that obtainable by brick and mortar retailers) provides valuable information which may encourage retailer entry into select categories.
In this worst-case scenario, consumers may also be harmed, which of course is a critical link in most antitrust jurisdictions. This is because there may be a decrease in the economic returns to independent sellers that put effort and investment costs into developing new products. In the long run, consumers may have fewer innovations and variety. The platform, focused on its own profits, may be willing to accept fewer innovations and variety on its marketplace if vertical integration raises its own profits in some categories.
This tradeoff is similar to that facing a textbook monopolist that raises prices: yes, raising prices reduces quantity sold, but the overall move benefits the monopolist. Thus, because the marketplace may care more about the profits that it receives than the innovation benefits that consumers enjoy, it may be that the marketplace finds it optimal to risk harming innovation to some extent—e.g., because of the increased advertising fees that it (not consumers!) receives. Earlier theoretical work in economics, albeit focusing on a somewhat different setting, shows that a monopolist may optimally vertically integrate even if so doing harms innovation.
This theory of consumer harm is quite consistent with the 2020 U.S. Vertical Merger Guidelines. These Guidelines posit that an ability and incentive to raise rivals’ costs may raise competitive concerns, specifically noting as an example that rivals may be “deterred from innovation, entry, or expansion,” and further establish the potential for harm from access to competitively sensitive information:
In some circumstances, the merged firm can use access to a rival’s competitively sensitive information to moderate its competitive response to its rival’s competitive actions. For example, it may preempt or react quickly to a rival’s procompetitive business actions. Under such conditions, rivals may see less competitive value in taking procompetitive actions.
A difference is that the discussion regarding access to competitively sensitive information in the Guidelines seems to suggest that the act of vertical integration itself is what gives the firm access to information, whereas in the discussion here the marketplace has access to at least some of the information even absent vertical integration; this informs whether and how soon it vertically integrates, and other competitive decisions, potentially reducing incentives for other sellers to take procompetitive actions such as innovation and entry.
This theory of consumer harm is also related to the idea of innovation competition. The horizontal merger guidelines of the EU and the U.S. both recognize that horizontal mergers might in principle harm competition by curtailing incentives for innovation. The recent EU assessment of the proposed Dow/DuPont merger is an example. Despite being a horizontal case, the underlying premise is that the conduct in question tended to reduce industrywide innovation incentives, not merely those within precise and limited product categories. This is similar to the notion that the practices and features of online marketplace vertical integration, as well as inherent marketplace features such as those related to advertising, may likewise have a tendency to reduce innovation (potentially in certain classes of circumstances although not as a general rule).
Another source of potential harm stems from the ease with which a seller other than the manufacturer can resell a good on a marketplace without consent of the manufacturer. This may make it difficult for smaller retail platforms to assemble exclusive or limited selections that help them stand out from larger players such as Amazon. By not being able to differentiate their product lines, such small retail platforms may exit, eventually leading to fewer channel options for sellers. This feature can be seen as imposing, albeit indirectly, a form of a price-parity restriction, limiting the extent to which other retail platforms can offer better deals or unique selections.
Procompetitive Effects. I now discuss several reasons why the negative consequences just discussed may not come to pass. It is instead possible that the practices of online marketplaces, although different than those of other retailers, do not exhibit net negative consequences for either consumers or sellers.
First, although online marketplaces do have the potential to host many new products, it may be that the great bulk of products are not new. More aptly, it may be that platform vertical integration does not diminish innovation incentives for many categories. According to the Wall Street Journal, Amazon asserts that the “vast majority of its private-label sales are staples such as batteries and baby wipes.” In such categories the main effect of vertical integration by a platform may be to reduce prices for consumers by bringing additional competition to under-served categories.
The theory of reduced innovation and consequent consumer harm discussed above assumes that an online marketplace might free ride on certain investments, such as the those related to exploring the viability of a category or product niche; these investments by third-party innovators generate an “informational externality” that benefits other firms, which might then discover the attractiveness of the category and decide to enter themselves.
But many other types of investments in a competitive market are not easy to free ride upon. For example, the returns on investments in cost reductions or incremental product improvements are most likely to be substantially enjoyed by the firms making the investments, not free-riding rivals. Indeed, in a category that exhibits little investment free riding, innovation and investment may increase following the vertical integration by the platform. Thus, it is possible innovation is spurred rather than chilled by vertical integration. Some empirical work suggests nuanced tradeoffs, for example finding that innovation falls in a category when Google offers its own version of an app in the Android app store, but that independent developers may increase their innovation efforts in other categories.
Second, although the prospect of higher advertising revenues may in principle heighten vertical integration incentives, it is also possible that the advertising system itself serves primarily a procompetitive role and does not appreciably drive entry decisions. Also, the advertising systems themselves may improve the matching of consumers to products in a way that is superior to the organic search capabilities of the marketplace, as opposed to mostly being tools for rent-extraction. Indeed, it is possible that the underlying organic search algorithms are trained or informed by the keyword advertising of sellers, suggesting that such advertising systems may in fact help improve the sorting of consumers to products even within organic search results.
Third, it may be that the superior market intelligence and accumulation of data by the online marketplace drives procompetitive entry decisions into under-served categories. The information generated by category-level analysis and reviews may also guide, procompetitively, how new products are designed to better serve consumers in that category or provide those consumers with new niche options.
Just as prospects for consumer harm may be low in some categories, concerns over harm to suppliers may also be low if the suppliers themselves are large, profitable firms. Unless the goal of policy is to transfer benefits from consumers to the shareholders and executives of such companies, limiting vertical integration by an online marketplace into such categories may not be wise. Therefore, the composition of sellers on an online marketplace is important. If new policies are to be put in place, the question arises as to whether such policies can or should target different types of sellers in different ways, so as to balance a broad set of concerns.
Online marketplaces differ from other types of retailers, and these differences matter for assessing the economics of retailer polices such as the offering of private-label goods (vertical integration). Therefore, concerns raised by EU and U.S. officials about the “dual role” of online retailers (who sell their own products alongside the products of third-party sellers) cannot be immediately dismissed as harmless by analogy to brick and mortar retailers that commonly offer their own private-label goods.
Most simply, it may be that vertical integration by a company such as Amazon is not simply a minor variant that is qualitatively equivalent to vertical integration by a brick and mortar retailer. Instead, online marketplaces may have heightened incentives to vertically integrate, and such integration may harm different types of sellers, for example, smaller sellers with newer or more niche products. On-platform advertising and promotional systems are potentially problematic. They may serve substantially as tools to extract rents from suppliers without necessarily helping consumers, and the desire to spur higher ad revenues from third-party sellers may further raise incentives for the marketplace to vertically integrate. Consumers may also be harmed.
However, negative outcomes are by no means certain. Vertical integration might encourage rather than chill innovation by sellers and spur entry into under-served categories. Similarly, on-platform advertising services might primarily work to improve the matching of consumers to products rather than to drive platform profits.
The author thanks Jeff Cohen, Richard Gilbert, Bruce Hoffman, and D. Daniel Sokol for helpful comments.
