THE EVOLUTION OF ANTITRUST IN THE DIGITAL ERA: Essays on Competition Policy

All rights reserved.

No photocopying: copyright licences do not apply.

The information provided in this publication is general and may not apply in a specific situation. Legal advice should always be sought before taking any legal action based on the information provided. The publisher accepts no responsibility for any acts or omissions contained herein. Enquiries concerning reproduction should be sent to Competition Policy International at the address below.

Copyright © 2020 by Competition Policy International

111 Devonshire Street · Boston, MA 02108, USA

www.competitionpolicyinternational.com

[email protected]

Printed in the United States of America

First Printing, 2020

ISBN 978-1-950769-60-5 (Paperback)

ISBN 978-1-950769-61-2 (Hardcover)

ISBN 978-1-950769-62-9 (Ebook)

Publisher’s Cataloging-in-Publication Data

provided by Five Rainbows Cataloging Services

Names: Evans, David S. (David Sparks), 1954- author. | Tucker, Catherine, editor. | Fels, Allan, editor.

Title: The evolution of antitrust in the digital era : essays on competition policy, volume I / David S. Evans ; Catherine Tucker [and] Allan Fels AO, editors.

Description: 1st edition. | Boston : Competition Policy International, 2020.

Identifiers: LCCN 2020945593 (print) | ISBN 978-1-950769-60-5 (paperback) | ISBN 978-1-950769-61-2 (hardcover) | ISBN 978-1-950769-62-9 (ebook)

Subjects: LCSH: Antitrust law. | Competition, Unfair. | Electronic commerce--Law and legislation. | Big data. | Consolidation and merger of corporations. | Commercial law. | BISAC: LAW / Antitrust. | LAW / Commercial / General. | LAW / Mergers & Acquisitions.

Classification: LCC KF1649 .E93 2020 (print) | LCC KF1649 (ebook) | DDC 343.07/21--dc23.

Cover and book design by Inesfera. www.inesfera.com

Editors’ Note

The story of antitrust is the story of technology. The essays in this volume tell the latest chapter in this ongoing saga.

In the late 19th century, the disruptive technology of the day was the railroad. In the expanding U.S., local railroads were bought up and consolidated into broad systems by the trusts that gave the Sherman Antitrust Act of 1890, and the resulting worldwide body of law, its name. Moving on from transport, various technologies have formed the locus of economic growth, and therefore of antitrust scrutiny, throughout the past hundred years or so.

After the railroads came Standard Oil, and its control over the key input for 20th century economic growth. Again, this was a reflection of technology, both in other industries’ need for vast sources of energy, and the improved refining technology that led to scale in the oil industry itself. Antitrust enforcement, famously, split the company up. Then, mid-century, came the telecommunications revolution. In the U.S., concerns crystallized around the role of the Bell System as an incumbent technology provider. Once more, antitrust enforcement split it up. In the 1970s and 80s, IBM’s mainframe computing business became the target of enforcement. Following on from that, the banner cases of the 1990s in both the U.S. and Europe were against Microsoft’s practices in the desktop computing space. In the latter two instances, however, the consequences were less radical, due, perhaps, to the intervening Chicago School critique of earlier antitrust remedies.

Despite these different outcomes, at each step along the way, antitrust thinking has been defined by the technologies that gave rise to its greatest enforcement challenges. Since the dawn of this century, attention has turned to the current generation of innovators, in what today is termed the digital economy. The quandaries facing today’s legislators, enforcers, and public, are novel and multifaceted. Nonetheless, they bear comparison with the formative struggles that policymakers grappled with throughout the first century of antitrust.

The pieces in this volume draw on the lessons of the past to set out how competition rules might deal with this new set of concerns, in various jurisdictions around the world. Each one draws on general themes, yet nevertheless addresses specific aspects of the contemporary debate.

Much of today’s antitrust discussion concerns the businesses run by large companies such as Amazon, Apple, Facebook, Google, and Microsoft. Each has significant share in a given industry, and derives its revenues from what are described as platforms. But how are such platforms different from the incumbent businesses of the past? The answer to this is not clear. Yet queries surrounding the platforms’ alleged dominance, and whether their conduct amounts to an infringement of competition rules, have been a source of controversy for over a decade. The pieces in this volume address this dilemma head-on.

At a fundamental level, there is the definitional threshold of what a platform even is, and what rules should apply to such a business. Then there is the question of whether platforms have a special responsibility towards downstream operators that rely on them to reach customers. In other words, can platform operators favor their own businesses in those related markets? Or do competition laws require them to treat all firms in the same way? What are the risks to competition if platforms are given free rein? In antitrust parlance, these questions are assessed under the rubric of self-preferencing, which has dominated recent headlines.

Pieces by Thomas Kramler and Robert D. Atkinson & Joe Kennedy report on this controversy from the trenches. The authors draw on their considerable experience in dealing with these issues to ask whether antitrust concerns in the digital economy can effectively be addressed within the confines of existing antitrust law and jurisprudence, or whether new rules are needed.

At the time of publication, this platform regulation debate is reaching its crescendo. In 2019, various jurisdictions, including the EU, Germany, Australia, and the Brexiting UK, commissioned detailed reports on whether competition rules need to be updated to deal with platforms, and self-preferencing specifically. The coming months and years will see legislatures take action on these reports. Much is at stake in how these reports’ conclusions are interpreted. The pieces in this volume form a vital part of that discourse.

Aside from these (almost existential) concerns, there is the question of how platforms interact with other actors in the economy. While it is productive for there to be broad discourse on the role of competition and digital regulatory policy, it is also vital for those rules to stay in their own lane. Otherwise, reforms grounded in the logic of antitrust could unduly expand its role, and counteract other policies. This debate has reached an advanced stage in Australia, where policy efforts have honed in on the media and news industry. Pieces by Simon Bishop & George Siolis, and Andrew Low & Luke Woodward, describe these developments, and discuss the risks of focusing on a narrow set of sector-specific concerns to derive broad antitrust solutions.

Then, there are even more specific concerns. Algorithms, anonymously executed in server farms, dominate modern commerce. Aside from mundane operational decisions, algorithms are increasingly used to set pricing and other commercial strategies. This can be pro-competitive and efficient. But algorithms, like people, can also restrict competition and harm consumers. If firms use algorithms that autonomously tacitly collude through deep machine learning, can the firms that run them be held liable? The pieces by Andreas Mundt and Gönenç Gürkaynak, Burcu Can & Sinem Uğur underline the need for further research on how such algorithms operate in real-life settings, before creating a new head of liability.

Technology allows consumers to access and interact with offers in the digital world with remarkable ease. But it has also created the potential for new forms of consumer exploitation, and facilitates highly individualised price discrimination. This creates opportunities for business models based on exploiting incumbents’ superior bargaining position, particularly in the business-to-business space. Platforms can make take-it-or-leave-it offers that allow the platform to enjoy all the surplus of trade. This notion of an abuse of a superior bargaining position is foreign to competition rules in certain jurisdictions, but is known in Japanese competition law, as discussed by Reiko Aoki & Tetsuya Kanda.

Moore's Law famously predicts that the number of transistors on a microchip will double every two years, though their cost will be halved. These remarkable advances, coupled with parallel developments in mass data gathering and storage, allow today’s computers to solve tasks of extraordinary complexity, including innovative, reliable, and lucrative predictive analytics. Yet this possibility raises profound privacy concerns, as reflected in laws such as the California Consumer Privacy Act and the EU’s General Data Protection Regulation. Such rules, in turn, raise novel competition issues.

This dynamic has profound implications for competition law, and how it interacts with privacy rules. Although competition and privacy law are separate disciplines, they are in tension with each other. As Maureen K. Ohlhausen & Peter Huston discuss, this problem came to the forefront in recent U.S. litigation between hiQ and LinkedIn. The latter, invoking the privacy rights of its members, employed technical measures to block hiQ’s automated bots from accessing data on LinkedIn’s servers. HiQ, in turn, alleged that LinkedIn’s actions were in reality an attempt to restrict competition.

As the authors discuss, this case represents the archetypal conflict between data privacy and competition, and will be repeated throughout the world in years to come. The policy dilemma between privacy rules and antitrust cannot be overstated. Protecting privacy by restricting data flows can hinder competition by denying new entrants access to the data they need to compete. On the other hand, ensuring that rivals have easy access to data can diminish privacy by distributing data in ways that consumers may not anticipate or want.

The foregoing should make clear that the story of antitrust in the digital economy is but one chapter in a saga that is still being written. Like all sagas, it draws from universal themes, and is self-referential within its canon. Yet it is all the more interesting as a result.

The editors would like to thank Elisa Ramundo, Sam Sadden, and Andrew Leyden for commissioning, compiling, and editing this volume.

The Antitrust Challenge of Digital Platforms: How a Fixation on Size Threatens Productivity and Innovation

By Robert D. Atkinson & Joe Kennedy

¹

Abstract

Over the last several years, much of the debate on antitrust policy has focused on the largest digital platforms, including, Amazon, Apple, Facebook, Google, and Microsoft. Each of these companies is large, has some significant market share in their narrowly defined industry, and most derive much of their revenues from running one or more multi-sided platforms. These factors have generated a backlash by anti-big firm activists and consternation among many European and U.S. competition policy officials. However, a careful review of both the individual markets and the general issues reveals that the challenges these companies pose, while slightly different from previous markets, are not entirely new. Moreover, in most cases legitimate competition policy issues, especially those related to structure, are limited and can be effectively addressed within the confines of existing antitrust law and jurisprudence.

I.INTRODUCTION

Over the last several years, much of the debate on antitrust policy has focused on the largest digital platforms, including, Amazon, Apple, Facebook, Google, and Microsoft. Each of these companies is large, has some significant market share in their narrowly defined industry, and most derive much of their revenues from running one or more multi-sided platforms. These factors have generated a backlash by anti-big firm activists and consternation among many European and U.S. competition policy officials. However, a careful review of both the individual markets and the general issues reveals that the challenges these companies pose, while slightly different from previous markets, are not entirely new. Moreover, in most cases legitimate competition policy issues, especially those related to structure, are limited and can be effectively addressed within the confines of existing antitrust law and jurisprudence.

These companies, and others like them, add significant value to the economy. Although their size and market share have led to several concerns, regulators should pause before taking action. As with all antitrust cases, any decisions should follow from and be guided by a careful study of all sides of the specific markets involved and be focused on maximizing overall economic welfare. When this is done, regulators will find that many of the problems being debated, such as vertical expansion into other markets, accumulation of market power through mergers, and threats to innovation, do not justify a change in either the substance or the enforcement of antitrust law. Where real problems related to conduct, as opposed to presumed ones, do appear, existing tools give agencies sufficient power to deal with them.

It is critical that policymakers and antitrust practitioners get this right. Given the potential transformation of many industries by digital technologies and the importance of both network effects and economies of scale, digitally powered platforms in a range of industries may well become the dominant form of organization going forward, similar to the rise of the large industrial corporation at the turn of the 20th century and large diversified companies in the decades after WWII. In response to the great expansion of economic growth and innovation enabled by those then-new organizational forms, the U.S. federal government interpreted the Sherman and Clayton Acts to let individual companies acquire the size needed to maximize efficiency, but established guardrails against mergers and conduct that conferred clear market power and business practices that limited economic growth or hurt consumers.²

The main antitrust laws were passed around the turn of the 19th Century and were originally focused on limiting trusts – formal agreements of cooperation and collusion between separate companies. One key response to banning trusts was to encourage mergers and the emergence of large industrial corporations as a way for companies to take advantage of new technologies which enabled and even required economies of scale. Teddy Roosevelt’s Progressive Party platform reflected the common view of the time, The corporation is an essential part of modern business. The concentration of modern business, in some degree, is both inevitable and necessary for national and international business efficiency.³ Thirty years later, his cousin Franklin Delano Roosevelt stated, Nor today should we abandon the principle of strong economic units called corporations, merely because their power is susceptible of easy abuse.⁴ This acceptance of large corporations was central to the ability of the United States to become the dominant economic power through the late 1970s. In contrast, Europe continued to allow cartels which reduced the incentives to create large corporations, which meant that Europe developed many fewer large, globally competitive corporations, imposing a structural burden on the EU economy that persists to this day.⁵

If it is true that in the near future large intra-industry digital platforms in industries such as health care, banking, education, and logistics will emerge by combining data analytics, low transactions costs, and global reach with a platform-based business model, it is critical that competition policy evolve to enable, rather than stifle, these new, more productive forms of business organization. Just as antitrust doctrine in the early decades of the 20th century adapted to and indeed embraced the form of the large industrial corporation, antitrust policy needs to do the same today in embracing platform organizations. Applying more restrictive antitrust doctrines and practices to these new models threatens to squash European growth and do to the U.S. economy what Europe did to its economy through the mid-20th century. As such, the stakes of getting this right cannot be overestimated. The first place to start is to focus antitrust policy on what it is supposed to focus on: overall economic welfare.

II.THE BENEFITS DIGITAL PLATFORMS BRING

The dominant fact about digital platforms is that they deliver significant benefits to a wide range of users, including app developers, sellers of a wide variety of goods and services, advertisers, consumers, and tens of millions of people who use social media to stay in touch with family and friends.

The value of these benefits is hard to measure, in part because many services are offered for free. But even if they were not, the consumer surplus between their value to Internet users and the amount that users actually have to pay is very large. A recent study by MIT economists estimates the median Internet user would require compensation of $17,530 to give up search engines for one year. The equivalent estimates for email and digital maps are $8,414 and $3,648, respectively.⁶

A filing by scholars from the Mercatus Center lists five ways Internet platforms create value:⁷

•By allowing people to rent out other people’s cars, homes, and other property, they increase the value of underutilized capital.

•By connecting large numbers of buyers and sellers, they make both supply and demand more competitive and allow greater specialization among producers, leading to more choice for consumers.

•By lowering the transaction costs of finding willing partners, negotiating over price, ensuring quality, and monitoring performance, they increase the number of beneficial trades.

•By making it easy for both buyers and sellers to check on the past performance of potential counterparties, they increase the amount of information in the marketplace and reduce the risk to parties.

•By offering an alternative to traditional markets, whose regulators are often captured by existing producers, they create opportunities for new suppliers to satisfy the unmet needs of consumers and force incumbents to become more efficient.

These benefits tend to have progressive effects. The savings from lower prices and free services often benefit low-income consumers the most, because the savings represent a higher proportion of their total income. Moreover, higher-income users are more valuable to platforms because they are more likely to buy advertised goods and services, yet both higher income and lower-income consumers receive the same services.

These companies are also among the most innovative in the world. Amazon and Alphabet led all companies in investment in research and development in 2018. Microsoft and Apple came in sixth and seventh, while Facebook was 14th.⁸ Each company is constantly innovating its core business in order to respond to competitive threats, including from each other, and attract new users. In addition to their core businesses, they are among the leaders in investing in the next generation of general-purpose technologies, including artificial intelligence, autonomous vehicles, blockchain, quantum computing, and robotics. Development of these technologies will deliver significant economic and social benefits.

III.THE ALLEGED THREAT TO ANTITRUST

Antitrust concerns about the largest digital giants are driven largely by the difficulty for antitrust thinking to effectively adapt to the network age. At the turn of the 19th century, some saw large firms with a significant share of the market as at best suspect; at worst a serious problem. Today, some see platform-based businesses in a similar light.⁹ But, in the digital economy, platforms may very well become the dominant form of business organization, for precisely the same reasons large industrial organizations became dominant in the 20th century: they are the most efficient organizational form for the current technology.

Today, antitrust concerns over platforms are driven by two common traits of multi-sided platforms. On the demand side, the push for bigness is caused by network externalities. The network’s value to each user is increased by each additional user. One platform that contains everyone is more valuable than two platforms, each of which contains half the users. This is because with one platform every user can reach every other user. For example, Facebook has announced plans to make Facebook Messenger, WhatsApp, and Instagram interoperable, since these services are all owned by Facebook, so that users on one app can message users on the other apps using whichever service they prefer.¹⁰ Internet users would be worse off if the Federal Trade Commission obtained an injunction preventing Facebook from merging these services, or worse, split these companies apart, because then users would have to create and maintain separate accounts on each of these services to communicate with all of their contacts.¹¹ Of course, not every network works this way, and mandating interoperability requirements for social networks could create security risks or create other problems for users, such as spam or harassment.¹² Even the classic example, the telephone, has lost its monopoly on intercommunication; people no longer need a phone to call each other. Internet-protocol standards allow voice packets to be generated and sent on a variety of different platforms. Users also have different interests, so often not everyone needs to communicate with everyone else, in which case the network advantage will fade out at a certain size. The net result is scale. As an Obama administration Council of Economic Advisers’ report noted, Some newer technology markets are also characterized by network effects, with large positive spillovers from having many consumers use the same product. Markets in which network effects are important, such as social media sites, may come to be dominated by one firm…¹³

On the supply side, firms often grow bigger to benefit from economies of scale. By growing larger, firms can reduce their average total cost of production by spreading their fixed costs over more units. But traditional economic theory also assumes that most firms will eventually face increasing marginal costs because of inefficiencies that come from being too large. These increasing marginal costs limit how large firms can grow, making it difficult for any one firm to capture the entire market. However, digital platforms usually enjoy fixed marginal costs that do not increase with size. This means that their average total cost continues to decline as they add more users, and they do not face the same constraints on their size or market share. These efficiencies benefit society.

Digitally powered business models, including platforms, also have the advantage of being able to have strong offerings along a number of dimensions. Traditional firms normally focus on and gain advantage in one, or possibly two of three aspects: price, quality or customization, in large part because there are significant tradeoffs between each. Customization comes at the expense of low cost, for example. Indeed, much of the business strategy literature is premised on firms identifying which of these market areas they should specialize in. But for many Internet platforms, digital technologies enable them to make strong offerings in all three aspects: low prices, higher quality, and customization.¹⁴

These advantages are not likely to be absolute, however. Economists Daniel Spulber & Christopher Yoo point out that market share due to network effects can be interrupted by periodic outbreaks of new competition for the market, raising the possibility that the dominant platform will be replaced.¹⁵ Two of the biggest drivers of this disruption are technology and demographics. Historically, technological innovation played a significant role in companies like IBM (mainframes), Digital Equipment Corporation (minicomputers), AT&T (telephony), Walmart (retail) and FedEx (delivery) losing dominant market shares. Indeed, important transitions such as the move from analog to digital, the rise of the Internet, and the advent of smart phones have been especially challenging for incumbents to spot and respond to.

As antitrust scholars Carl Shapiro & Hal Varian note, [T]he information economy is populated by temporary, or fragile, monopolies. Hardware and software firms vie for dominance, knowing that today’s leading technology or architecture will, more likely than not, be toppled in short order by an upstart with superior technology.¹⁶ And as IT industry expert David Moschella points out, today’s giants are more vulnerable than previous industry leaders in at least one way: the customer switching costs are mostly ones of changing habits, not conversion effort and cost, and this relative ease of transition could be an important factor sometime down the road.¹⁷ Today, rapid advances in technology continue to present platforms with new services and business models. Platforms that do not quickly adapt to these opportunities leave the door open for rivals.

In fact, Spulber & Yoo believe platforms are likely to face even more competition in the future, spurring more innovation.¹⁸ However, in order to enable this dynamic efficiency, regulators may have to allow static inefficiency for a limited period of time. Businesses with large upfront expenses and low marginal costs often need to earn higher rates of return to recoup their investments, and to fund the next big investments in innovation. But even then, their advantages may be temporary, particularly in a globally competitive economy. Similarly, the advantage of efficiencies of scale can be offset if competitors also enjoy zero marginal cost.

The constant entry of new consumers can also present Internet platforms with a challenge. Young consumers have little invested in existing networks, tend to be very comfortable with the latest technology, are more concerned about communicating with a narrow group of friends rather than the whole world, and are less than awed by their parents’ technology. Younger users were the main drivers behind instant messaging, WhatsApp, Instagram, and most recently TikTok.

In both cases, antitrust regulators need to balance the benefits of competition for the market with the benefits of scale. However, this tradeoff only needs to be made if there is some harm to the consumer, such as a delay in innovation.

IV.LET ANTITRUST BE ANTITRUST

Nevertheless, the rapid growth of a few large companies that operate digital platforms has produced a lot of anxiety about their effects on society. Purported market dominance has been blamed for a host of ills including reduced privacy, poor data security, censorship, poor moderation of offensive and dangerous content, and excessive political power. These are all important issues, deserving their own policy response. But in most cases antitrust solutions would do nothing to solve them, and indeed may make them worse. Moreover, trying to incorporate them into antitrust policy would replace one widely supported policy goal, the consumer welfare standard, with a jumble of goals that often conflict with each other. Attempts to rank many policy goals on their own would not only give tremendous discretion to unelected regulators. It would threaten the discipline’s ability to accomplish the goals it is suited to. Even the best regulators would struggle to find the right balance between these goals. The worst would use the tremendous discretion given to them to reward political supporters.

Moreover, the development of separate policies for privacy and other concerns can have antitrust implications. Rules such as Europe’s General Data Protection Regulation impose large, somewhat fixed costs on firms that are subject to them, both in the cost of complying and the risk of punishment for inadvertent breaches. This gives larger firms an advantage because they can spread the fixed costs over a larger user base. Smaller firms therefore find it harder to compete. It can also drive firms from the market, further reducing the amount of competition. As economist Catherine Tucker notes, stringent privacy rules can strengthen advantages larger firms have when it comes to data.¹⁹

Nobel Laureate economist Jan Tinbergen developed the rule that achieving a desired number of policy targets requires regulators to have an equal number of policy instruments.²⁰ For this reason, social concerns should remain outside the boundaries of antitrust analysis and practice.

V.HOW ANTITRUST HANDLES CONCERNS ABOUT DATA SCARCITY

As mentioned above, in most cases, a separate policy goal requires a separate policy instrument. Thus, most of the issues raised in the previous section do not involve antitrust issues and deserve non-antitrust solutions. However, data collection does have an antitrust component. Two points are important to keep in mind about data. First, the antitrust aspect is only a small part of this issue and should not get confused with other aspects. Second, current antitrust policy built around the Consumer Welfare Standard works well in dealing with the antitrust issues surrounding data. The confusion comes when proponents of stronger regulation try to use antitrust remedies to fix issues that do not involve antitrust goals.

Proponents of stronger antitrust enforcement for platforms frequently point to data as a strategic asset. It is often said that data is the new oil, implying strategic importance for those who have it and vulnerability for those who do not.²¹ But this is simply the wrong metaphor. Unlike oil, data is non-rivalrous (the same data can be used more than once) and is often non-excludable (others may gather the same data that I already own). Even more relevant to its role in antitrust policy, data often has zero marginal costs (the cost of collecting, processing, and storing additional data often approaches zero) but rapidly diminishing returns to scale.

Most important, however, is that much of the most collected personal data often tends to be worth little by itself. Rather, it is the business model and algorithms used to analyze the data that give companies value. Reporting by the Financial Times shows that individual pieces of data are often sold cheaply.²² General information, such as age, gender and location is worth only $0.0005 per person. Information about someone shopping for a car is slightly more valuable at $0.0021 per person. Knowing a woman is in her second trimester of pregnancy bumps it up to $0.11 per person. The total for most individuals is less than a dollar.

Economists Anja Lambrecht and Catherine Tucker argue that data seldom provides a company with a competitive advantage, especially in the face of a superior product offering.²³ Although Amazon Marketplace, for example, benefits from knowing what products users searched for on their website, its large selection, low prices, and superior logistics are far more valuable. Although data makes its business model better, it is a mistake to think data scarcity is the primary constraint to greater competition. Antitrust law should look at data the same as any other asset that occasionally has strategic value rather than as a unique resource conferring broad market power. In a more recent paper, Tucker cites previous studies to argue that neither network effects nor switching costs are likely to confer market power due to the mere possession of data.²⁴ One of these showed that the accuracy of Internet searches was not sensitive to the amount of historical data about past searches, implying that Google’s market share is due mainly to its present superiority rather than its past dominance.²⁵ Moreover, because most data is relatively worthless and ubiquitous, government requirements to share it will seldom be justified.

VI. SHOULD INTERNET COMPANIES BE BANNED FROM COMPETING WITH USERS ON THE PLATFORMS THEY RUN?

Another concern is that companies that run multi-sided platforms often compete with sellers on one or more sides of the market. For example, Amazon offers its own products on Amazon Marketplace, often in direct competition with those of its regular suppliers. Because Amazon can see what products are selling at what prices, critics claim it has an unfair advantage. Similarly, Apple was recently sued by Spotify, which alleged that Apple used unfair tactics in its App Store to suppress its competitors.²⁶ Senator Elizabeth Warren (D-MA) would prohibit this practice by designating the largest platforms utilities and prohibiting companies from both owning the platform and being a seller on the platform.²⁷

Of course, digital platforms are not the first, or even the biggest, marketplaces. Large retailers including CVS, Walmart, and Costco, offer shelf space to thousands of manufacturers. They also offer competing products under their own brand names, such as Costco’s Kirkland brand. These products usually sit on the same shelf, side-by-side with those of their suppliers, often undercutting them on price. And of course, the retail chains look at sales data when making decisions on what to sell.

The main point to remember when analyzing such cases is that this behavior is usually procompetitive. It may hurt the supplier of a competing good, but it benefits consumers by offering more choice at lower prices. As long as the supplier’s intellectual property is not violated and consumers are not confused, government should not dissuade this type of competition. Antitrust policy should protect consumers, not sellers hurt by legitimate competition. While there may be room for general fair-dealing requirements that require platforms to post their policies and enforce them evenly, platforms usually have a strong incentive to engage in fair-dealing: platforms that abuse their suppliers will lose them. This in turn will attract fewer buyers, leading to a downward spiral.

The widespread assumption that platforms’ dominant position in a particular market frees them from competition is based on a faulty definition of the relevant market. For platforms providing free services, including Facebook and Google, the relevant market is the total ad market. They derive a significant portion of their revenue from advertisements shown to their users. In many cases, the buyers of these ads are also large sophisticated companies that use analysis by Visual IQ and C3 Metrics to evaluate the return on investment across platforms, including television and radio. The competition for ads leads to a competition for eyeballs as platforms compete for users’ limited time and attention. Although YouTube does not offer the same functions as Facebook, users can only pay attention to one or the other at a time.

Platforms that make money from consumers must also compete with off-line firms which still represent the largest part of the economy. Amazon competes with other sellers on its Marketplace, the Internet sites of other retailers and, most significantly, brick and mortar stores. Its cloud-based services compete with those of established companies including Google, Microsoft and Oracle. Apple computers compete with Windows-based systems and its iOS devices, such as the iPhone and iPad, compete with Android-based tablets and smart phones.

Finally, possessing a dominant position in the market does not guarantee a profit. Amazon experienced net cash outlays for many years as it reinvested profits to expand its business. More tellingly, Uber continues to suffer losses with no end in sight. Heavy competition from rival services, including taxis, may prevent it from ever rising out of what is essentially a commodity business.

VII.FOUR ILLUSTRATIVE CASES OF COMPETITION INVOLVING PLATFORMS

A closer look at some of the more commonly cited instances of market abuse shows that the issues are far more complicated than many observers describe. Although the four instances described below may not be typical, the fact that they are commonly cited as showing the failures of market forces justifies going into more detail.

Amazon’s purchase of Quidsi: As mentioned above, the potential for companies to offer competing products on the markets they run is often cited as a threat to competition. The concern, however, seems directed at protecting