Big data does not undermine market competition
- Daniel Sokol1 & Roisin Comerford, Professor of Law, 2017, University of Florida and Senior Of Counsel Wilson, Sonisini Goodrich & Rosati, Does Antitrust Have A Role to Play in Regulating Big Data?, https://awa2017.concurrences.com/IMG/pdf/ssrn-id2723693.pdf
Andres Lerner (2014) argues that claims of Big Data presenting competitive concerns are unsupported by real world evidence. In particular, Lerner argues that in practice the oft-cited “feedback loops” do not have the strong effects with which they are commonly credited. Lerner discusses the procompetitive rationales for collection and use of consumer data online, including the potential for improved services, and the ability of firms to monetize effectively on the paid side so as to provide better services at lower prices or for free. He dismisses the idea that firms’ may have the incentive or ability to use data to entrench their dominant position (e.g., user data is non-rivalrous and no one firm controls a significant share of data) citing similar attributes of data as Ohlhausen and Okuliar. Lerner maintains that there is a complete lack of evidence that online markets have “tipped” to dominant firms, due in most part to the differentiated nature of online offerings. He concludes that without strong real-world evidence of anticompetitive effects, aggressive antitrust enforcement would hamper competition and chill innovation, injuring consumer welfare in the process. Although policy makers have dipped their toe into the antitrust in Big Data debate,5 antitrust agencies and the courts have not found a Big Data competition problem. In fact, that the FTC and DG Competition have thoroughly considered Big Data as an antitrust problem and completely dismissed it. The agencies in the United States and Europe have moved cautiously so far, which is not only proper, but also serves as a reminder that the distinct issues addressed by antitrust and consumer protection law, and the solutions that may be applied by each set of laws to prohibited behavior, are distinct for good reason, and are complements rather than substitutes (Muris and Zepeda 2012; Averitt and Lande 1997)
Making it more difficult for companies to monetize data means fewer services for consumers
- Daniel Sokol1 & Roisin Comerford, Professor of Law, 2017, University of Florida and Senior Of Counsel Wilson, Sonisini Goodrich & Rosati, Does Antitrust Have A Role to Play in Regulating Big Data?, https://awa2017.concurrences.com/IMG/pdf/ssrn-id2723693.pdf
Perhaps the most obvious and pervasive benefit to be realized in the Big Data era has been the ability of firms to offer heavily subsidized, often free, services to consumers as consumers give those firms permission to monetize consumer data on the other side of their business (Evans and Schmalensee 2014). In a competition law regime where lower prices for consumers are deemed highly desirable, this is undoubtedly a benefit to consumers. The monetization of the data in the form of targeted advertising sales for antitrust purposes is not suspect or harmful, but rather “economically-rational, profit-maximizing behavior,” that results in obvious consumer benefit (Lerner 2015). Were online platforms prevented or restricted from collecting and monetizing consumer data, competition for users would be inhibited, and harm to consumers would result, in the form of higher prices for services. Indeed switching costs are low regarding data and search (Edlin and Harris 2013).
More data improves products
- Daniel Sokol1 & Roisin Comerford, Professor of Law, 2017, University of Florida and Senior Of Counsel Wilson, Sonisini Goodrich & Rosati, Does Antitrust Have A Role to Play in Regulating Big Data?, https://awa2017.concurrences.com/IMG/pdf/ssrn-id2723693.pdf
As an input, online firms use data to improve and refine products and services in a number of ways, and to develop brand new innovative product offerings. For example, search engines, both general and niche, can use data to deliver more relevant, high quality search results. By learning from user search queries and clicks, search engines can identify what are the most relevant results for a particular query. “Click-and-query” data, as it is known, is a highly valuable input in delivering high quality search results (Salinger and Levinson 2015). Outside of just relevant results, search engines can use data to provide additional “value-added” services to users. Travel search engines, for instance, can use data to forecast price trends on flights for specific routes. Amazon and multiple other ecommerce sites use past purchase information and browsing history to make personalized shopping recommendations for users (Goldfarb 2012). Social networking platforms use data collected from users to suggest friends, celebrity or business pages, or articles that customers might be interested in. Online media outlets use browsing history and personal information to recommend other articles that a reader may be interested in.
Big data is not anticompetitive
- Daniel Sokol1 & Roisin Comerford, Professor of Law, 2017, University of Florida and Senior Of Counsel Wilson, Sonisini Goodrich & Rosati, Does Antitrust Have A Role to Play in Regulating Big Data?, https://awa2017.concurrences.com/IMG/pdf/ssrn-id2723693.pdf
(c) Economic Characteristics of Big Data Protect Against Competitive Harm In addition to the affirmative pro-competitive benefits of Big Data expounded above, the economics of how Big Data works, as described below, damages claims that it should be feared, or reined in by antitrust. Additionally, the unique economic characteristics of data mean that its accumulation does not, by itself, create a barrier to entry, and does not automatically endow a firm with either the incentive or the ability to foreclose rivals, expand or sustain its own monopoly, or harm competition in other ways (Lambert and Tucker 2015). Lambrecht and Tucker explain that “For there to be a sustainable competitive advantage, the firm’s rivals must be unable to realistically duplicate the benefits of the strategy or input.” As we suggest below, both theory and actual cases support a finding that the characteristics of data are such that rivals cannot be foreclosed from replicating the benefits of Big Data enjoyed by larger online firms, and Big Data in the hands of large firms does not necessarily pose a significant antitrust risk. (i) Low Barriers to Entry Data driven markets are typically characterized by low entry barriers, as evidenced by innovative challengers emerging rapidly and displacing established firms with much greater data resources than themselves (Tucker and Welford 2014). While the existence or lack thereof of barriers to entry can, and will, differ from market to market, and a blanket determination cannot be made in the abstract, the history of the digital economy offers many examples, like Slack, Facebook, Snapchat, and Tinder, where a simple insight into customer needs enabled entry and rapid success despite established network effects. The data requirements of new competitors are far more modest and qualitatively different than that of more established firms. Little, if any, user data is required as a starting point for most online services. Instead, firms may enter with innovative new products that skillfully address customer needs, and quickly collect data from users, which can then be used towards further product improvement and success. As such, new entrants are unlikely to be at a significant competitive disadvantage relative to incumbents in terms of data collection or analysis (Tucker and Welford 2014). And, while a firm that has been operational for ten years may have a larger data store, lack of asset equivalence has never been a sufficient basis to define a barrier to entry in any cases as of yet. In brick and mortar retail, a new entrant may have a smaller showroom than an established competitor, but this does not render the need for a physical store location an insurmountable barrier to entry. Indeed, an established bricks and mortar store could have much more data on local customer preferences, but that has never been viewed as prohibitive to entry. -6- (ii) Data is Ubiquitous, Inexpensive, and Easy to Collect Data is ubiquitous, inexpensive, and easy to collect (Tucker 2013). Users are constantly creating data – increased internet and smartphone usage means customers are continuously leaving behind traces of their needs and preferences (Lambrecht and Tucker 2015). Data can be easily and quickly collected from consumers upon launch, and both data and the tools needed to store and analyze it are readily available from numerous third party sources. Big Data has near-zero marginal costs of production and distribution (Shapiro and Varian 1999). There are many alternative sources of data available to firms, reflecting the extent to which customers leave multiple digital footprints on the internet (Lambrecht and Tucker 2015). The fact that data can, therefore, be acquired from third party sources, means that even on the first day of product launch, before any user has interacted with the platform, a provider can already have benefitted from insights into consumer preferences, and designed a platform that can act quickly act as data is collected and processed. While some argue that the resources and effort expended by companies in pursuit of data is evidence enough that data collection and processing is both “costly” and “timeconsuming,” (Stucke and Grunes 2015a) it is important to distinguish between the collection of raw data, and the analysis any given firm puts the data through, which is what makes the data valuable. This is the firm’s “secret sauce.” It is also, incidentally, the part of a firm’s Big Data usage that requires the most resources. There is also plenty of off-the-shelf and open source analytics software that could give small firms a head start. (iii) Data is Non-Exclusive and Non-Rivalrous Data is non-exclusive and non-rivalrous. No one firm can, or does, control all of the world’s data. Collection of a piece of data by one firm does not occur at the expense of another firm. “Multi-homing” is the norm among internet users – users can, and do, spread their data around the internet, using multiple different providers for multiple different services, or sometimes the same service. While multi-homing, a user shares data with multiple providers. Big Data has been likened to other inputs as it becomes an increasingly important asset. However, Big Data’s non-rivalrous and non-exclusive nature sets it apart from other key inputs. If one provider has a piece of data, another provider is not prevented from collecting that very same piece of data. Similarly, while conceivably one provider could at least theoretically hold all of the world’s oil resources, for example, no one provider can amass all available data. Furthermore, incumbent online providers do not have explicit or de facto exclusivity over user data. There are no exclusivity clauses in terms of service with users, and there are no structures (pricing or otherwise) that lock users into sharing their data with only one provider.
(iv) Data’s Value is Short-Lived -7- Data has a limited lifespan – old data is not nearly as valuable as new data – and the value of data lessens considerably over time. Additionally, the returns on scale diminish over time. Therefore, any competitive advantage that data provides is fleeting, and entrants are unlikely to be significantly disadvantaged relative to incumbents in terms of data collection and analysis (Chiou and Tucker 2014). The need for fresh, differentiated data means that a firm with a large volume of stale or generalized data does not, necessarily, benefit the holder and disadvantage a potential challenger. Potential competitors do not need to create a data store equivalent to the size of the incumbent; they rather need to devise a strategy to accumulate highly relevant and timely data (Shepp and Wambach 2015). (v) Data Alone is Not Enough Data does not typically provide value on a standalone basis. Mere possession of data alone therefore, even in large volume, does not secure competitive success – that can only be achieved through engineering talent, quality of service, speed of innovation, and attention to consumer needs. As such, the firm with the most data does not necessarily win. Take the online dating application, Tinder, initially launched in September 2012, as an example. Data is of particular value in industries where personalized experience is important, such as online dating. When Tinder launched, it had no access to user data, but nevertheless it became the market leader within a couple of years. Lambrecht and Tucker (2015) explain that even in this highly data driven industry, Tinder succeeded not through reliance on Big Data, but due to the strength of its underlying solution. A simple user interface and a precise attention to consumer needs resulted in massive gains for the new entrant. Similarly, despite facing competition from long established incumbents with access to huge volumes of data, amassed over years of customer service, WhatsApp was able to take on more established messaging and social networks because of its low cost and easy-to-use interface. Examination of these industries leads Lambrecht and Tucker to conclude that to build a sustainable competitive advantage from Big Data, a firm needs to focus on developing both the managerial toolkit and organizational competence that allows them to turn Big Data into value to consumers in previously impossible ways, rather than simply amassing tremendous amounts of data. (vi) Highly Differentiated Platforms Need Highly Differentiated Data Online platforms are highly differentiated, even in the provision of the same type of service, and as each entrant carves out a niche, the most useful data to them differs more and more from the data most useful to their rivals. Consumers are moving towards meeting more precise, niche consumer needs. A consumer looking to book a flight could use Kayak, Expedia, Orbitz, or a multitude of other travel-dedicated search engines. The same is true in internet shopping, online dating, social networking, product and service reviews, and a host of other online markets. In today’s online environment, successful firms must carve out their own niche, and increasingly, data that is useful (even crucial) to one firm may not be useful to its competitors (Schepp and Wamback 2015). An astute and innovative entrant will identify a niche where the incumbent does not have requisite data, and can very quickly “catch up” to the incumbent in terms of valuable data amassed.
No anti-competitive feedback loops
- Daniel Sokol1 & Roisin Comerford, Professor of Law, 2017, University of Florida and Senior Of Counsel Wilson, Sonisini Goodrich & Rosati, Does Antitrust Have A Role to Play in Regulating Big Data?, https://awa2017.concurrences.com/IMG/pdf/ssrn-id2723693.pdf
The Perceived Strength of Scale, Network Effects, and Barriers to Entry Many, if not all, of the theories of harm attributed to Big Data rest on the perceived strength of the “feedback loop” and the consequential network effects enjoyed by large firms with access to tremendous amounts of data (Graef 2015). Big Data can give rise to network effects, and certainly, network effects can play a significant role in a sound antitrust analysis. However, agencies, policy makers, and scholars must resist any foregone conclusion that the presence of network effects in Big Data automatically results in anticompetitive harm. Big Data can lead to economies of scale via the alleged “feedback loop.” In search, some argue, “the availability of data on previous search queries is crucial” to competitive success (Graef 2015). There are two ways scale can be accomplished through the “feedback loop.” The “user feedback loop” presumes that as a platform gains more users, it can collect more user data, leading to better insights into consumers and their needs, which can be used to improve quality, attracting even more users. The “monetization feedback loop” claims that as a platform gains more users and collects more user data, it is better able to target ads and therefore sell ads, and so is better able to monetize its platform, gaining revenues which can be invested in improving quality of service, thereby attracting more users. Alongside these feedback loops, a number of distinct network effects come into play in online platforms that collect and use Big Data. Direct network effects occur when a product or service becomes more valuable to an individual user as more people use that particular product or service. In a modern context, social networking platforms, photo sharing platforms and chat applications may enjoy significant direct network effects. Indirect network effects occur when more users make the use of a product or service better or more attractive to consumers, though not because of direct interaction between users. Search engines benefit from indirect network effects as more users allow the search engine to essentially gain insight into what users want from user clicks, essentially learning by trial and error, and therefore improving the quality of search results. Some argue that network effects are particularly strong in two-sided platforms. A firm operating a two-sided platform can, it is argued, benefit from not only from traditional network effects, but also from cross platform network effects, where more users on one side of the platform makes the platform more attractive to users on the other side of the market (Graef 2015; Stucke and Grunes 2015a). While entry barriers naturally vary from industry to industry, and indeed change over time, these practitioners suggest that the economies of scale 14 Case COMP/M.6314—Telefónica UK/Vodafone UK/Everything Everywhere/JV, Comm’n Decision (Sept. 4, 2012). 15 Case No COMP/M.7023—Publicis/Omnicom (Jan. 9, 2014). -13- and network effects that characterize data-driven markets lead to a “winner takes all” result, and present insurmountable barriers to entry. In reality, the strength of the feedback loop may be grossly overstated. The feedback loop theory assumes smaller rivals and challengers will not be able to compete effectively as they lack comparable amounts of users, and therefore data, inhibiting their ability to improve quality and attract more users. As Lerner (2015) points out, however, these assumptions are unsupported by real-world evidence. The economics characteristics of Big Data weaken the claimed strength of the feedback loop. Chief among these characteristics is the fact that online providers can gain scale in users in ways that do not involve user data, and that access to data alone is not enough to improve quality and gain scale in users. Additionally, firms can gather data from other sources than users (e.g. data brokers), and can gain scale in data in alternative ways, such as entering into strategic distribution arrangements. As to network effects, even in classic cases of direct network effects such as social networking and communications applications, innovation can be strong enough to upend the market, and network effects have time and time again proven insufficient to prevent incumbents from disrupting established market leaders. In social networking for example, Friendster, the original “market leader” was replaced quickly by MySpace, which has now been rendered almost completely obsolete by Facebook. An innovative product is enough to cause users to switch, notwithstanding any network effect enjoyed by the incumbent. Among advertisers, network effects are diminished by the pricing structures employed by most online platforms, by advertiser multi-homing due to the low cost in advertising on multiple platforms, and by advertiser “congestion.” The pay-per-click model means that while advertising on a “busier” platform may result in better conversion rates for an advertiser, it also involves proportionally higher costs, and more clicks means the advertiser has to pay more. As such, it may actually not be as economically advantageous for an advertiser to choose a larger online platform over a smaller one (contrary to real world platforms that are priced differently). Additionally, since fixed costs to advertise on any particular platform are low, advertisers may be incentivized to advertise on multiple different platforms as opposed to putting all their eggs in one basket. Finally, while more users on a platform might be good for advertisers, more advertisers on the platform can actually be detrimental. Limited available space for online ads and competition for users’ attention means that advertisers may be better off on smaller platforms with less congestion (Lerner 2015). Perhaps most importantly, cross platform network effects are also commonly overstated, and are actually one-sided. While advertisers certainly may flock to a search engine (or other online platform) with a strong user base with the hope of more impressions and hopefully more conversions, users, on the other hand, do not choose a search engine based on a greater number of advertisements. This weakening of the cross platform network effects argument in turn weakens the potential for a strong “feedback loop” that locks users and advertisers into a dominant platform. If a smaller entrant offers a better product or -14- service to users, users will switch, uninhibited by network effects, and advertisers will soon follow (Lerner 2015). The above discussion demonstrates how the feedback loop is not as effective as suggested in gaining scale, but the importance of scale is also misjudged by many. Big Data industries typically experience diminishing returns of scale. Statistically, as Lerner (2015) illustrates, the value of user data in returning relevant results to user search queries is subject to quickly diminishing returns, as the advantages of scale weaken or disappear at a low level. While returns are greater for less frequent queries (known as “tail” queries), both large and small search providers are faced with queries they have never seen before on daily basis, where both small and large platforms are at an equal disadvantage in delivering relevant results. Because of these rapidly diminishing returns, a larger provider may gain zero marginal value from incremental data after a certain point, and a smaller player may glean greater value from incremental data, incentivizing it to compete in attracting users at the margin by investing in quality and innovation. Even if scale is crucial to competitive success, smaller rivals do maintain both the ability and the incentive to compete. As to ability, many online players are well-funded, or at least have access to additional funding from investors, with which they can improve quality and performance of their platform. Furthermore, all online players have access to stores of data from third parties, which is readily
available and affordable, and can be deftly used to increase quality. As to incentive, economics tells us that investment incentive is based on marginal, not average effects. An investment in quality by a smaller firm will attract more incremental users than a similar investment by a larger firm. As such, the smaller firm’s incentives to invest in quality may actually be greater than that of its larger rival.
More competition means more platforms to the Russians to infiltrate
Clara Hendrickson and William A. GalstonWednesday, Brookings Institute, December 6, 2017, Big technology firms challenge traditional assumptions about antitrust enforcement, https://www.brookings.edu/blog/techtank/2017/12/06/big-technology-firms-challenge-traditional-assumptions-about-antitrust-enforcement/
So while fear that big tech can wield excessive influence in our democracy may reflect broader misgivings outside the realm of antitrust law and enforcement, some political concerns about big tech appropriately fall under the purview of antitrust regulation. As Sally Hubbard, a Senior Editor at the Capitol Forum who covers monopolization issues, recently stated in an interview with Vox’s Sean Illing, “Companies like Facebook and Google have had an outsize effect on political discourse because of the ways their algorithms help to promote and spread fake news and propaganda. Even if it’s not their intent, their business model invariably contributes to this problem.” More competition between rival platforms would have introduced a greater number of algorithms for Russian operatives to navigate, and probably would have mitigated the impact of the fake news that successfully targeted voters during the 2016 U.S. election