Platforms should be allowed to sell on their own marketplaces
An increasing number of e-commerce companies are choosing to be a marketplace and a reseller at the same time. That is, they operate marketplaces for third-party products and are (re)sellers of products in their own name which (potentially) compete with these third-party products. In our academic work, we say that such companies operate in “dual mode”.
Examples include Amazon, Target, Walmart, Flipkart in India, JD.com in China, each of whom operates an online marketplace and also sells (or resells) products itself side-by-side with third party sellers on the same online marketplace. And many other platform companies also sell their own products on their marketplaces: Apple in its App Store, Google in its Play Store, Microsoft in Windows’ Apps, Intuit in its QuickBoks Apps, Salesforce in its App Exchange, etc.
Regulators around the world are increasingly proposing new antitrust laws that seek to ban large platforms from operating in dual mode. In this post we explain why such policies do not make sense.
The regulatory concerns
In recent years significant policy concerns have been raised about big-tech firms stifling competition. In late 2018, India passed a law that banned Amazon and Flipkart from selling products (on their respective marketplaces) of companies in which they have an equity stake, thus effectively banning the dual mode. And earlier this month new proposed antitrust laws have been unveiled in the United States, which among many other things, seem to be targeted at banning large platforms from continuing to operate in dual mode.
One proposed Act (the Ending Platform Monopolies Act) would make it unlawful for “covered platforms” to own or operate a business that presents a clear “conflict of interest”. Based on our reading of the Act, a covered platform is carefully designed so that it covers Alphabet/Google, Amazon, Apple, Facebook, and Microsoft, but no other firms. A conflict of interest arises when the “covered platform’s ownership or control of that line of business creates the incentive and ability for the covered platform to advantage the covered platform operator’s own products, services, or lines of business on the covered platform over those of a competing business…” Clearly this is aimed at stopping Amazon and Apple from operating in dual mode, given that they face a conflict of interest in selling their own competing products or apps in competition with third-parties on their respective marketplaces.
It is worth highlighting another proposed Act that is very much related. This is the American Choice and Innovation Online Act, which would prohibit many activities that involve self-preferencing or discrimination on covered platforms. The overall goal of this Act seems to be to stop any platform that faces a conflict of interest from acting on it.
Perhaps legislators proposing these bills are hedging their bets, but if the American Choice and Innovation Online Act is enacted, then there doesn’t seem to be much need for the more drastic Ending Platform Monopolies Act. Indeed, the concerns usually raised about platforms operating in dual mode are exactly the behaviors banned under the American Choice and Innovation Online Act.
Specifically, there are two main concerns that are commonly raised about platforms that operate in the dual mode:
That the platform obtains proprietary information on the third-party sellers’ products (e.g. detailed demand and pricing data, data on users' search behavior, what items they return and their reasons for doing so) via its marketplace, and then uses that information opportunistically to copy and compete on the more successful offerings, potentially leading to reduced incentives for third-party sellers to invest or innovate. For example, it has been claimed that some employees working as part of Amazon’s private labels group used data from its marketplace to inform their decisions about launching private label products, even though this is at odds with the company’s stated policies.
That the platform can steer consumers towards its own offerings (or affiliated products), rather than those offered by third-party sellers, by displaying its own offerings more prominently, a practice that has become known as “self-preferencing”. For example, Amazon can do this through its Buy Box, on which around 85% of all Amazon purchases are made. The Buy Box allocates a seller to the consumer when they click on “Buy now” or “Add to cart” according to a secret algorithm that Amazon controls, and oftentimes the allocated seller is Amazon itself.
Why banning the dual mode is bad policy
To begin with, it’s worth pointing out that if these concerns can be dealt with by behavioral regulations targeted at eliminating self-preferencing and the sharing of data across divisions that face a clear conflict of interest, then any remaining benefits of banning the dual mode are likely to be quite limited.
More importantly, the dual mode provides many clear efficiencies, which would be lost under such a heavy-handed, structural ban.
First, as noted in a previous post, some types of products may be more efficiently provided by the platform, while others may be more efficiently provided by third-party sellers. And there are clear economies of scale or scope from offering all products together on one site (i.e. in dual mode):
One-stop shopping benefits to consumers (search and checkout) relative to having to search and purchase on two different websites (e.g. Amazon’s marketplace and Amazon as an ecommerce retailer).
Cost savings to the platform from only having to build and manage one common website.
Cost savings for users if the shipping of the platform’s and the third-parties’ products can sometimes be combined into single orders.
The platform can make better recommendations between its own products and those of third-party sellers when they are all offered on the same site.
These economies of scale and scope should already make it clear that banning the dual mode across products does not make much sense.
Second, even when focusing on an individual product or product category, the dual mode allows the platform to create more competition within each product category. Indeed, by being able to step in and offer a product directly at a lower price, the platform ensures consumers will get more choice and lower prices on the marketplace in case established sellers have market power. In addition to creating price competition, the platform can also offer superior customer service or logistics, which puts pressure on third-party sellers to do likewise, thus benefiting consumers.
Third and relatedly, the platform can also help avoid stock-outs by increasing the supply of items that are in high-demand. It may even be willing to sacrifice margins to be the supplier of last resort given that it cares about customer experience on the marketplace much more than any individual third-party seller would (a consumer coming back to the marketplace may or may not buy from the same seller, but always benefits the platform).
Given the benefits of operating in the dual mode, it is not surprising that it is used by an increasing number of companies who open up marketplaces where they sell their own products alongside third-parties: Best Buy, Intuit, Salesforce, Shopify, Twilio, etc. In fact, there is now even a successful startup, Mirakl, which provides companies the necessary tools to create such marketplaces.
It is striking to us that the benefits of the dual mode are clear and obvious, while the harms that can arise from it are not. The harms are based on behaviors (self-preferencing and accessing third-party data, as outlined above) that have been alleged but have yet to be properly empirically documented at scale. The fact that a dual-mode platform can tilt the playing field on its own marketplace in its favor vs. third-party sellers does not necessarily mean it will always do so. For instance, if Amazon was systematically steering consumers to less than ideal products, or blatantly crushing its third-party merchants, over time it would lose both merchants and consumers, to its own long-term detriment. Moreover, if self-preferencing and accessing third-party data can be established as significant problems, they can be targeted via competition law or new regulations that address the concerns directly (such as those contained in the proposed American Choice and Innovation Online Act, or in Europe, by the proposed Digital Markets Act) rather than imposing a full-scale breakup.
Lastly, given the obvious implications this discussion has for Amazon and Apple, we should report that neither of us has done any consulting work for Amazon or Apple, nor (unfortunately) does either of us hold Amazon or Apple stock, least readers suspect we are in their pockets. On the other hand, together with our co-author Tat-How Teh, we have written an academic paper on this topic which provides a much more in-depth analysis of the issues discussed here (for the more mathematically-inclined readers).
If you enjoyed reading this, sign up now so you will receive our posts directly in your inbox as they are published.
In the meantime, tell your friends!
Having built my entire career as a 3rd party Amazon seller, I can tell you these “harms” of self-preferencing and third-party data are NOT major concerns of 3rd party sellers.
The very nature of a dual marketplace such as Amazon is to create a more honest and fair marketplace and shopping experience for customers by enabling the widest variety of competition as possible. Enacting these new regulations would undoubtedly force Amazon to remove 3rd party sellers from their marketplace and themselves attempt to become the sole seller of millions of diverse products which are sourced from all over the world. The inevitable ramifications of this would be:
- Complete monopolization of the biggest marketplace in the world, leading to unfair & subjective prices since there is no competition.
- An astronomical decrease in product options and product differentiation, leading to severely limited options for customers and a significant decrease in product quality since there is no influential reason to improve.
It’s hysterical how blatantly counterproductive these proposed “anti monopolies” and “anti trust” regulations would be—but I forget there are some really stupid people in this world who don’t have a lick of experience building and running a business but want to tell everyone else how to do it.
The not so hysterical part is that if these proposed laws go through, it will obliterate countless individual’s and family’s well-being who spent hundreds and thousands of hours building their businesses and careers as a 3rd party seller just to have it all ripped out from under our feet…and done so by ignorant, small minded people who are so ridiculously disconnected from the very issues they’re trying to solve.