Amazon collects a wealth of digital data about consumers and merchants on their marketplace. The data is used to optimize recommendation algorithms, customize search experiences, and even learn from third-party merchants what private-label products to offer. But what if we could use that same data to determine whether merchants and consumers would be better or worse off if Amazon faced more competition? The choice of which level of competition is best for platform ecosystems, such as Amazons, is not simply a dichotomous choice between complete monopoly and perfect competition. Our research on platforms and network effects suggests that the best policies may not lie at eitherextreme of laissez-faire or competition at all costs. Finding the best policy often requires a careful analysis of the trade-offs.
Much of the dominance of todays tech firms, from Amazon to Google to Facebook, arises from network effects. These platforms facilitate connections: they connect people to products, to friends, to apps, to drivers, and to homes that would not otherwise be available to them. As more people use these platforms, they can facilitate exponentially more connections, and thus provide more value to each person. It is Amazons connections to consumers (both for discovery and delivery) that make it so attractive to many businesses. Take Hunnibi, a Canadian start-up specializing in mess-free honey dispensers. Amazon allowed them to reach a large pool of honey enthusiasts, an unlikely outcome for a small company based in Dollard-Des Ormeaux, Canada.
But businesses depending on Amazon increasingly find themselves at the mercy of the giant platform. There are increasing concerns that dominant platforms may harm consumers and businesses, by taking an increasing share of the value created from enabling connections, or by crushing innovation. Take for example Fortem, a manufacturer of car accessories founded by two young entrepreneurs in 2016. Their trunk organizer soon became very popular on Amazon Marketplace, until Amazon launched a similar competing product through its own private label, Amazon Basics.Or take Peak Design, a stylish designer of camera gear. Amazon Basics camera bag is eerily similar to Peak Designs Everyday Sling, both in name and functionality.
The question of how much competition tech giants like Amazon should face is now at the center of worldwide policy debate, and reining in big tech is a common goal across the political divide. But traditional antitrust policy is ill-equipped to regulate network-based platforms for several reasons. First, markets with network effects may naturally tip toward one dominant network, so a platform may find itself in a monopolistic position without violating any antitrust laws. Second, to attract users and generate network effects, some platforms are offered free to users, charging in data or attention. For these platforms, standard antitrust rules that evaluate consumer welfare based on prices are not applicable to evaluate market power. Third, conditions in these markets evolve quickly and often unpredictably, requiring a nimbler response.
Historically, especially in the U.S., antitrust authorities have taken a laissez-faire approach under the assumption that it is better to err on the side of not intervening when there is uncertainty. This has allowed companies like Google and Facebook to go on a shopping spree to acquire early-stage competitors that could have become a threat if left independent. But recent signals, such as the appointment of Lina Khan as the Chair of the U.S. Federal Trade Commission, suggest that the tide may be turning, and big tech may find themselves in the position of having to either defend their dominance as beneficial to their ecosystems or risk losing it.
The most obvious potential downside of monopolies is that firms may charge high prices when consumers have few alternatives. In fact, existing antitrust policy in the U.S. and many other countries is based on the notion that if consolidation increases prices for consumers, it should be blocked. There are many such examples in platform mergers and acquisitions, although the definition of price is a bit more subtle. Take the merger of Singapore-based Grab by Ubers operations in that region in early 2018. The companies were fined by the Singapore Competition Authority because following the deal, not because Grab changed nominal prices, but rather reduced both the number of points earned by riders per dollar spent and the number and frequency of driver promotions and incentives.
In many digital platforms, however, the standard price-based approach may not provide much guidance. Eased by generous VC financing, platforms tend to charge low prices in the beginning to attract users and generate network effects, even if they may raise them later. This means that looking at price changes immediately following a merger provides only a partial picture of the long-term market power that such a merger may generate. When Rover, the largest pet-sitting platform in the U.S., acquired its fiercest competitor, DogVacay, in early 2017, prices and promotions did not budge.
Many platforms, such as Facebook or Google Search, are free to use, charging instead by collecting more user data, or by charging a different user group, like advertisers. These choices affect product quality even if customers dont have to pay for the product. For example, Facebooks acquisition of WhatsApp and Instagram may have affected the quality users experience, even if none of the services charge them directly. A key question is how competition affects quality and innovation. It is unclear whether competition will lead to higher quality or more innovation. On the one hand, pressure imposed by coexisting competitors competition in the market can lead platforms to innovate and increase quality. But on the other hand, the expectation of future dominance can lead to competition for the market: entrants may innovate on the prospect of becoming a dominant platform or being acquired by one. Competition for the market is nothing new. In fact, patent protection, which effectively grants temporary monopolies, forms part of the incentive that impacts innovation in pharmaceuticals.
As markets mature, policymakers are considering interventions that mix competition for the market and competition in the market, but which mix yields better outcomes is still an open question. In some markets it is feasible for each consumer to multi-home, or use many similar services simultaneously. On social media, for example, many users jump between Facebook, Twitter, and Instagram over the course of a single day. In these contexts, interventions like data portability can increase competition by making it easier to multi-home.
In markets where it is difficult to multi-home, policymakers might require competitors to make their networks interoperable. These policies have been debated in sub-Saharan Africa over the past two decades, as mobile phone networks have grown to be lifelines for communication. But what degree of interoperability is optimal is not so obvious. Take Bjrkegrens research, which examined the degree of interoperability across mobile phone operators in Rwanda. An extreme interoperability policy, making it completely free for competitors to connect their subscribers across networks, would have lowered incentives to invest in mobile phone coverage in rural areas by as much as 43%. Similarly, the other extreme of shutting down competition in the market does not maximize incentives to grow the network. A desirable policy lies in the middle: Competitors are allowed to connect users across networks, but have to pay the other network an interconnection fee that is 57% higher than what the government mandated at the time. Promoting competition under these terms would have increased the total utility of Rwandans of consumers, firms, and government by an amount equivalent to 1% of the countrys Gross Domestic Product.
Even though a larger network may be able to create more connections, different consumers may wish to have those connections take different forms than a single dominant platform could provide. For example, Snapchat introduced messages that disappear after they are sent, a popular innovation among a subset of social media users. In these cases, one downside of promoting competition through standards for interoperability is that these can freeze the forms that connections can take, limiting differentiation.
The few studies we have of platform competition point to three early conclusions. First, either extreme, granting dominant platforms free rein, or maximizing competition in the market, may not be optimal. Second, setting the best rules can have enormous impacts. Third, the functioning of these platforms depends on the structure of connections they facilitate. This structure is currently opaque to company outsiders. The main constraint to measuring the value of connections is, paradoxically, a lack of data. We know that many people use Facebook or Uber every day, but do not precisely know how much the presence of one user benefits another. However, researchers have demonstrated that it is possible to measure the effects of policies in network industries with data from the firms themselves. The paradox is that we often treat technology platforms as black boxes that can only be theorized about, while in fact, platforms collect and store more data about their own functioning than any other entities in history.
Even coarse statistics from platforms would allow policymakers to quantify some of the benefits and costs of competition versus dominance in platform industries. For example, this could include diagnostic data about usage, prices (over time and by user group), and quality (such as fraction of ads relative to content). Having data from multiple competing platforms could allow for the measurement of substitutability and switching costs, and to learn retrospectively from mergers that have already taken place. The same internal experiments that platforms use to fine-tune their services or understand their own demand can also reveal how policies are likely to impact the market.
Of course, tech giants would never volunteer their data to third parties to assess whether more competition would benefit their ecosystems. That might harm the firms interests. But a regulator could enforce this under clear and stringent privacy-preserving rules. Governments around the world have the authority to oversee businesses, and could build capacity to request and analyze this digital data, by either existing regulators or new authorities that have been proposed to specialize in digital markets (as is being commissioned in the UK, or suggested in the U.S.). That has historically happened in other industries, with the establishment, for example, of the Federal Communications Commission and the Federal Aviation Administration to coordinate radio communications and the use of airspace.Such entities could request and analyze data from platforms to learn the optimal course of action for each market.Given the increasing importance of these networks to daily life, such a regulator should use the most privacy-preserving level of data possible, and be transparent to constituents about what data is being used, and why.
There are substantial upsides to learning from platforms data. This does not mean that any intervention should be delayed while waiting for some perfect understanding, since many instances of dominant platforms can be evaluated using the best knowledge available. However, completely overhauling antitrust policy in the era of platform companies would require a careful balance of the benefits of scale and those of competition, a balance that we have only begun to study empirically in a handful of cases. A better understanding of these networks can help societies fully reap the benefits of digital innovation, while mitigating emerging harms.
Here is the original post:
Network-Based Platforms Must Be Regulated. But How? - Harvard Business Review
