Crémer Report on Competition for the Digital Era Misses the Mark
This post is part of a series covering different studies on antitrust enforcement of digital services released by various academics, think tanks, and regulatory agencies. The second in the series is a review of the Crémer Report on Competition for the Digital Era.
One year after the Commissioner Margrethe Vestager appointed academics to look at competition challenges in the digital space, their long-awaited policy recommendations were released on April 4, 2019. The report co-authored by Jacques Crémer, Yves-Alexandre de Montjoye, and Heike Schweitzer titled Competition Policy for the Digital Era (“the Crémer Report”) comes at a critical time when regulators around the world are grappling with how competition law should apply in the digital space.
The Crémer Report characterizes three fundamental features of the digital economy in the following ways: first, the digital economy exhibits extreme returns to scale. Second, the convenience of using existing technologies or services due to an already significant number of users is called “network effects,” and can constitute an additional barrier to market entrants. Lastly, the ability to collect and use large amounts of data in the digital economy can affect competition in markets. All of these characteristics, the Report argues, make incumbent digital players very difficult to compete with. However, these features do not in fact suggest a lack of competition, and are based on either inaccurate or incomplete assumptions.
Increasing Returns to Scale Does Not Automatically Exclude Competitors
The first feature is the idea of extreme returns of scale, where the cost of production for digital services is proportionally much less than the number of customers served. In contrast to traditional industries, in the digital economy information can be transmitted to a large number of people at a very low cost. For example, once a search engine or mapping service has been developed and is running, it can serve hundreds of thousands of users relatively cheaply. According to the Crémer Report, extreme returns to scale would affect competition:
With increasing returns to scale, competition between two firms producing the same product will not allow them to cover their costs. Indeed, were they to cover their (total) costs, they would have to price above the cost of serving an additional consumer (the marginal cost) and each of them would find it profitable to lower their price to steal the other’s clients.
Therefore, the theory is that predominant digital services can leverage their higher margins to invest in more research and development or lowering their prices, which could drive out weaker rivals. The Report surmises that “[n]o firm, unless armed with a much superior and cheaper technology, would want to enter a market dominated by an incumbent, even when this incumbent is making large profits.”
Successful multi-sided business models do indeed enjoy increasing returns to scale, but the Report’s overly broad conclusion based on this fact is premature. Digital services compete aggressively with each other and in multiple markets. Trends and observations made on these businesses are based on a very short period of time—the Internet economy is, after all, only about 20 years old. Therefore, blanket conclusions in reliance of these observations can easily be inaccurate or fail to consider the bigger picture. Friendster, once the “king” of social media, was toppled by Myspace and then Facebook. Microsoft was viewed as the reigning services of the information technology industry, only to lose that status once mobile businesses were introduced. It is unequivocally wrong to surmise that firms do not want to compete due to increasing returns to scale unless they possess superior and cheaper technology.
There Is More Than Meets the Eye With Network Effects
The Crémer Report assumes that strong network effects pose competitive constraints. According to the Report, it is sometimes not enough for a new entrant to offer better quality or a lower price than an incumbent; it also has to find further ways attract users to its services. Therefore, network effects could “thus prevent a superior platform from displacing an established incumbent.” And while the Report mentions that “multi-homing” (people using multiple services with overlapping features) potentially reduces strong network effects, it fails to emphasize the importance of multi-homing and the pro-competitive role it plays. David Evans, an economist, explains that multi-homing enables new firms to attract new users, and allows for vigorous competition between incumbents and start-up firms:
Online platforms are more susceptible to attack by entrants than network industries of a century ago. Network effects and sunk costs made the monopolies around the turn of the 20th century difficult to challenge. Rivals had to sink massive amounts of capital into duplicating physical networks such as railroad tracks and telephone lines. Using multiple networks, or switching between them, was expensive for customers, even if a second network as available. However, online platforms can leverage the Internet to provide wired and wireless connections globally. People find it generally easy, and often costless, to use multiple online platforms, and many often do. The ease and prevalence of multi-homing have enabled new firms, as well as cross-platform entrants, to attract significant numbers of users and secure critical mass necessary for growth. Incumbent platforms then face serious competitive pressure from new entrants—because their networks are reversible.
As Evans mentioned, network effects are also reversible, a point that the Crémer Report neglects to consider. This means that while a large user base can attract more users, each lost user could induce others to leave as well. In fact, the reversibility of network effects may be even quicker in the digital economy because it is incredibly easy for users to switch online services. Users of a service can also typically try a new one for free, and can seamlessly shift over to the new one if they like that one better.
While the Crémer Report acknowledges the existence of multi-homing, its understanding of network effects is incomplete and ignores a more nuanced picture of how users on these multi-sided business models actually behave—from how users enjoy multiple services with the same function to how effortlessly they can switch to one another. Thus, the strength of network effects should not be the primary focus when discussing this particular phenomenon, and other observations should be factored into the conversation.
It’s Not How Much Data, But How That Data Is Used That Matters
Finally, the third misguided assumption that the Report presumes is that access to large amounts of data acts as a significant barrier to entry and gives companies an unfair advantage in the market. The Report argues:
However, a systematic strategy by dominant platforms to buy up or leverage their data assets to compete with innovative start-ups (even sometimes cutting off a competitor’s access) may result in early elimination of an emerging competitive threat or otherwise strengthen the firm’s dominance by increasing barriers to entry.
To be sure, data is an asset that could potentially act as a barrier to entry. But while the Report takes this assumption as truth, there is no empirical support at this point to conclude definitively that this is the case. Instead, the use and collection of large amounts of data does not by itself represent a threat to competition, and is instead critical for the economy’s most important innovations, especially in artificial intelligence. As the artificial intelligence race continues, it is essential that data be amassed and shared to encourage innovation––to enhance better matching products and services, improved image recognition, automatic video captioning, and more.
In addition, aggregation of data is also just one piece of the puzzle. Other inputs, such as skilled and creative labor, are what allows data to be analyzed, valued, and monetized. To give an example, the start-up dating service Tinder surpassed established rivals such as Match.com, eHarmony, and OkCupid not because Tinder had accumulated stockpiles of data, but because of its innovative features and unique algorithms. Therefore, the presumption that an incumbent with the ability to “buy up or leverage their data assets” can just eliminate start-ups fails to recognize that just because a company has more data does not necessarily mean it automatically has an unfair advantage in the market.
As a result of these misguided assumptions, the Crémer Report makes many controversial proposals, one of which is to change the existing structure of the consumer welfare standard so that successful firms must bear the burden of proving their conduct isn’t anticompetitive. However, with low entry costs and easy switching by consumers, dominant market positions are more transient than the Report acknowledges. A case-by-case analysis that takes into account evidence, economic analysis, and that is specific to the facts should continue to be the approach of the consumer welfare standard to ensure competition in the digital era. In contrast, requiring incumbent firms to show that their conduct is pro-competitive unduly burdens these firms and penalizes them for merely being successful.