With the growing market dominance of digital platforms like Amazon, Apple, Facebook and Google, calls for an effective policy response have become louder. New competition regulations are on the horizon, but international cooperation is needed for them to succeed.
As our use of information and communications technologies has increased, a number of large digital platforms – ‘Big Tech’ – have expanded and come to dominate their respective markets. In February 2021, Google accounted for nearly 87% of the global search market, compared with Bing, which took under 7%. Similarly, with over 1.4 billion users, Facebook controls much of our social media, while Apple and Google dominate the market for mobile phones and associated apps stores; and with 1.87 billion monthly users, Amazon has become the giant of e-commerce.
The dominance of these companies and their expansion into new markets – creating major digital ecosystems – have led to ever louder calls for regulation to promote more competition.
New regulatory regimes are being developed in a variety of jurisdictions, with detailed proposals now on the table in the UK, the European Union (EU) and the United States. These will set out a framework of basic rules by which the largest digital platforms will be required to abide. These moves follow the publication of a number of expert reports internationally, which advocate such regulation, notably the UK government-commissioned report Unlocking Digital Competition.
But what problem is such regulation trying to solve? And what are the key design challenges?
What is the problem with competition in digital platform markets?
In recent years, a small number of critical digital platform markets have become highly concentrated – in other words, they are dominated by just a few companies. Further, the market positions of the firms involved, such as Apple, Facebook and Google, have become increasingly set in stone. This leaves little space for new companies to join and establish themselves in the market.
Figures 1 and 2 show the situation in search and social media. The most popular platform companies are Facebook, which also includes Instagram and WhatsApp, and Google, with Google Search and YouTube. These are used to a far greater extent than even relatively well-known platforms, such as Microsoft’s Bing and LinkedIn.
We also see strong concentration in mobile phone operating systems, where Apple and Google Android share the market between them. Meanwhile, in business-to-consumer (B2C) e-commerce platforms, market leader Amazon had net UK sales in 2020 of $26 billion, dwarfing its nearest competitor, eBay ($1.7 billion). While these figures are for the UK, the situation is broadly similar in most Western economies, including in the EU and the United States.
Figure 1: Shares of supply by page referrals from January 2009 to April 2020
Source: Statcounter Global Stats
Note: Bing’s share represents Bing and MSN Search, MSN Search was rebranded as Bing in 1998; ‘Other’ consists of: AOL, AskJeeves, AVGSearch, Babylon, Baidu, Conduit, Webcrawler; Yandex; and ‘other’
Figure 2: Total user time spent on social media platforms from July 2015 to February 2020
Source: Comscore, Competition and Markets Authority (CMA)
Note: *Including Facebook, Messenger, Instagram and WhatsApp; ** Including Messenger
These high levels of concentration have partly come about as a result of the basic economics of these platform markets. First, they exhibit huge economies of scale and scope, in that they are expensive to set up but the costs of serving additional users or entering additional markets are relatively low.
Second, they benefit from significant ‘network effects’. These occur whenever users value a service more when there are higher numbers of other users. A social network is a classic example: each of us is more likely to use a particular service when more of our friends are on it. But there can also be ‘two-sided’ network effects: the more sellers there are to choose from, the more end-users are likely to use a marketplace and sellers are more likely to use a marketplace if there are more users. Third, data play a critical role in these digital platform markets. The main four platforms – Amazon, Apple, Facebook and Google – are estimated to hold around 1.2 billion gigabytes of data between them. Access to these immense amounts of user data are by-products of the service, but they are also a critical input into both the services provided – and their monetisation.
For example, Google’s search engine is being constantly improved, through using data on users’ query and click activity in its machine-learning algorithms. Facebook and Google both use their extensive data to achieve ever more effective targeting of advertising. Data, including those collected from across different services, can also be useful in the development of new and innovative products.
These three factors – data, network effects and economies of scale and scope – are important because they can lead to markets becoming highly concentrated, and the market positions of the successful firms becoming firmly embedded. People are unlikely to switch to a new rival to Facebook unless their friends switch too. New entrant firms with innovative ideas cannot access the data required to ‘train’ their own machine-learning algorithms, while access to the same data facilitates entry into new markets by the major existing platforms. This allows the dominant platforms to succeed even if their offering is less useful or innovative than a new entrant’s would be.
These reasons might already be considered enough to raise competition concerns. But in addition, a number of major digital platforms have been found to have breached competition law by engaging in ‘strategic anti-competitive conduct’. For example, the European Commission has found Google guilty of abusing its dominant position in three different cases – Google Shopping, Google Android and Google AdSense.
Further investigations are in progress, with the UK’s Competition and Markets Authority (CMA) investigating Google’s Privacy Sandbox, Apple’s App Store and Facebook Marketplace. At the same time, the European Commission is investigating Amazon, Apple and Facebook for similar breaches. Many other cases are being taken forward in other countries.
The platforms have also engaged in substantial merger and acquisitions activity, with Google, Facebook, Apple, Amazon and Microsoft accounting for over 400 mergers globally within one decade. For example, Facebook now owns Instagram, WhatsApp, Oculus Virtual Reality and a number of other companies. This has both consolidated the company’s position in its core markets, often through acquiring potential competitors, and facilitated its expansion into new markets.
Why does this matter?
The largest digital platforms have brought us huge innovation and valuable new services. These are often provided at a very low cost or even free to users. It is thus reasonable to ask whether their market dominance is really a problem.
An initial point to highlight is that while services may appear free, users should in fact be characterised as paying for them ‘in kind’, through giving up their own privacy and allowing their data to be collected. Unlike a price-based transaction, however, many users are unaware of the how much data they are giving up and are not especially engaged in the decision. The CMA recently concluded that Google and Facebook design their platforms to exacerbate this situation by encouraging consumers to sign up to their default privacy policies.
There are couple of key reasons why the strong market positions of the major platforms really do matter.
First, these platform markets are a vital gateway for businesses to reach potential customers. This gives them market power over those firms. Even though many of the services are provided free to end-users, they are monetised through charges to business users, for example, for advertising.
Substantial market power allows the platforms to charge high fees to business users. These increased costs will in turn squeeze the businesses’ profits, which will feed through to us all as higher prices or reduced quality. For example, if Google charges a high price for its online advertising, then this will increase the prices charged by those advertisers to their consumers.
Second, although we have seen much innovation in the past, there is a significant risk that it will soon be more limited. Potential rivals to the major platforms are at a significant competitive disadvantage in this space.
Successful innovation involves much more than a good idea. In technology markets, it requires access to data (to train machine-learning algorithms), to users and to a fair reward if successful. For many potential innovators, the biggest digital platform firms control some or all of these crucial elements. They control key relevant data, they are often the gateway to potential users and they can exploit their market power to extract an unfair share of any rewards from successful innovation.
The dominant platforms are also in a unique position that enables them to create rival services when they observe successful innovation, and thus extract a large share of any profits. Finally, their market power enables them to engage in strategic anti-competitive conduct to keep rivals out of the market or to limit their profitability.
Such market power thus reduces the incentives for innovation by third parties. But what about the Big Tech platforms themselves? After all, they have great access to all the relevant data, users and funding needed to facilitate plenty of innovation. There is some truth to this, and there is evidence of substantial research activity in certain areas, such as self-driving cars and digital personal assistants. But much innovation can be seen as an attempt to escape competition.
Where this is the case, then, just as in any running race, the incentives of the leader to expend effort will be lower if it is too far ahead of its rivals – in other words, if its competitive advantage is too great. In simple terms, we can expect monopolists to innovate less than firms in competitive markets.
Why isn’t existing competition law enough?
There is a growing consensus internationally that intervention is needed to create a more level playing field and to help open up platform markets to competition. But an obvious question is whether this simply requires a ramping up of existing competition law. Competition authorities globally may have been a little slow off the mark, but the biggest platform firms are now subject to a wide range of investigations.
Competition law certainly has an important role to play, in relation to the core issue of platform market power and also in relation to wider competition concerns, such as algorithmic collusion.
Merger law is also crucial, and the UK has recently published new Merger Assessment Guidelines, which are designed to signal an enhanced focus on the sorts of issues – relating to potential competition and risks to innovation – that tend to arise with mergers in digital markets. Indeed, there are proposals in both the UK and the United States to change the merger test for the largest digital platforms. This reflects concerns that they have been allowed to make anti-competitive acquisitions in the past, and that it can be hard for authorities to intervene in these markets, given their uncertain and dynamic nature.
Yet there are two reasons why competition law is not enough to address the concerns highlighted above.
First, a number of the factors driving concentration are a result of the structure of the digital market, and hard to describe as strategic anti-competitive behaviour. It is far from straightforward, if not impossible, to bring competition law cases where there is no deliberate anti-competitive conduct.
Second, even for conduct that could be addressed through competition law, these cases tend to take a long time and to be narrowly focused (which reduces the potential for wider deterrence benefits).
Remedies also tend to be highly retrospective and typically ineffective. The sanctions levied may seem huge: the European Commission has so far fined Google €9.4 billion across the three cases referenced above. Nevertheless, the threat of such sanctions is unlikely to change the conduct of the biggest platform firms in practice, given the profits they stand to make and their very deep pockets to litigate cases.
There is thus seen to be a need for ex ante regulation, which sets out the clear ‘rules of the game’ upfront and which is designed to be pro-active in opening up digital platform markets and creating a fairer environment for their users.
What will be the key elements of regulation?
In December 2020, the EU published detailed proposals for its Digital Markets Act, and the UK Digital Markets Taskforce published its advice for the UK government on implementing regulation. On 11 June 2021, the United States also announced five new bills, which would together create a similar regulatory framework, if enacted.
There is substantial similarity between these proposals. All would involve an upfront designation process, whereby the firms to be regulated would be identified, and there is significant alignment between the proposed criteria to be used for designation. Processes for enforcing against breaches will also be included. Significantly, these regulations allow (in principle) for high fines – up to 10% of global turnover for the UK and EU proposals, and up to 15% of US turnover in the US proposals.
A comparison of these proposals highlights a number of important differences. These reflect a number of key challenges in designing optimal regulation.
Challenge 1: Which firms to regulate?
The first challenge relates to determining which platforms should fall within the scope of the regulation – and how to set designation criteria to capture these and only these. There is broad consensus that any regulation should capture the five largest technology platforms (Amazon, Apple, Facebook, Google and Microsoft), but there is far more debate about the next tier of digital firms (for example, Booking.com, Spotify and Uber).
There are also concerns that ill-defined designation criteria could unintentionally include firms that offer rather different sorts of digital platforms, such as Mastercard, Visa, Oracle, SAP and the London Stock Exchange, among others.
The EU proposals partly solve this problem by specifying a number of core platform services that the regulation will cover, such as online search engines or online social networking services. The proposals also include a quantitative shortcut to designation, whereby firms will be presumed to fulfil the criteria where they are active in providing specified core platform services and have sufficient end-users and business users.
While this presumption is in principle rebuttable, this is expected to provide a simplified route to designation for the very largest digital platforms. There is flexibility within the EU proposals to include additional platforms and services, even if they do not meet the quantitative criteria, but it seems unlikely that this will be an early priority.
The US proposals go further still in terms of simplicity, by relying purely on quantitative criteria, specifically $600 billion in annual sales or market capitalisation, and at least 50 million active US users each month or 100,000 active US businesses. This approach is expected to narrow down designation to the five major platform firms.
By contrast, in the UK, the proposed regulator – the Digital Markets Unit – would be required to demonstrate that the specified qualitative criteria for designation are met. There is in principle no restriction on the digital platform services that may be specified. This provides for more flexibility than the EU or US approaches and is arguably more proportionate, but it could be more resource-intensive.
Challenge 2: How much flexibility to retain in rule-setting?
The second challenge relates to the extent to which obligations should be pre-defined in the legislation. The EU and US proposals both set out a number of specific obligations that will be placed on the designated platforms. These range from relatively precise rules, such as a requirement not to use data received from business users to compete against them to more overarching requirements relating to interoperability and data portability.
The UK proposal, by contrast, involves only very high-level objectives being set out in the legislation, with detailed rules to be developed for each platform during the designation process. This allows for the rules to be bespoke, reflecting each platform’s particular market position and business model.
The UK proposal has the benefit of allowing rules to be well targeted to particular concerns arising in relation to specific platforms and services. It also allows for greater flexibility over time. But some believe that it gives overly expansive powers to the regulator, while others are concerned that it will lead to extensive litigation around the evidence base for the rules, and thus that the regulator will struggle to introduce them within the intended timeframes.
The EU and US proposals have the benefit of providing more clarity upfront and limiting the powers of the regulator. But they risk being overly prescriptive and insufficiently flexible to deal with issues that may arise over time as the markets, and our understanding of them, develop further. After all, the rules that might have been imposed five years ago would almost certainly have been very different to those being imposed today.
Questions have also been raised as to whether complex provisions relating to data portability, data access and interoperability are well suited to upfront obligations, as is currently intended in the EU. While specific provisions of this sort can be immensely valuable in opening up markets, this needs to be done in a careful and targeted way if it is to be effective. For rivals to be able to use such data to create new services, they need to be provided on a consistent basis and directly transferable to third parties (with consumer consent). There is a risk that the obligations currently proposed by the EU are too wide-ranging and imprecise to be effective.
By contrast, the US and UK proposals address data portability and interoperability requirements rather differently from the remainder of the regulatory rules. The UK proposes a separate power to impose these as bespoke pro-competitive interventions, and they are the subject of a separate bill in the United States.
Challenge 3: How to avoid unintended harm?
A third challenge relates to how to ensure that regulation does not have unintended harmful side effects. The proposed EU framework does not allow any significant role for firms to mount an ‘efficiency defence’ or ‘objective justification’ for breaching an obligation. The only bases on which an obligation may be suspended or exempted are public morality, public health, public security or endangering the economic viability of the relevant service in the EU.
By contrast, the UK framework proposes that an exemption can be given where it is necessary or objectively justified in terms of efficiencies, innovation or other competition benefits, although this would be at the discretion of the Digital Markets Unit. This proposal should help to protect against harmful consequences from the regulation, but risks making its enforcement harder.
The US proposal lies somewhere between the two. A platform can be exempted from a given rule if it can show that this would not harm the competitive process, or that it is required in order to enable compliance with (or prevent a violation of) the law or to protect user privacy.
Underpinning these challenges is the question of the right balance to strike between administrability and finding the perfect answer, noting that if ‘finding the perfect answer’ takes a long time, then it is in fact far from ideal.
The EU and US approaches could be seen as giving very substantial weight to administrability, in a context where the authorities are dealing with highly litigious firms with deep pockets. By contrast, the UK has given greater priority to getting the right answer, while nonetheless including a number of measures that are designed to enhance administrability.
For the moment though, all are just proposals. It remains to be seen whether any jurisdiction will move towards another’s position as the final regulations are implemented.
The need for international cooperation
A final challenge in this area is how to ensure effective international cooperation. Since these are global firms, with global systems, there are clearly significant risks associated with inconsistent or incoherent regulatory approaches across jurisdictions.
To limit this risk, it is vital that there is as much international coordination and cooperation as possible. Fortunately, this does appear to be happening, not least with the CMA convening two important meetings with relevant G7 policy-makers in 2021 to deepen collaboration. It is perhaps noteworthy that the publication of the US proposals coincided with the recent G7 summit.
Where can I find out more?
- The European Commission Digital Markets Act: A translation: Cristina Caffarra and Fiona Scott Morton examine the European Commission’s Digital Markets Act, including its shortcomings.
- The European proposal for a Digital Markets Act: A first assessment: Report from the Centre on Regulation in Europe (CERRE).
- Digital Markets Taskforce: A new pro-competition regime for digital markets: Advice to the UK government on the design and implementation of a pro-competition regime for digital Markets.
- House Dems unveil bills to rein in Silicon Valley giants — opening rift among Republicans.
Who are experts on this question?
- Amelia Fletcher, Centre for Competition Policy and Norwich Business School, University of East Anglia
- Tommaso Valletti, Imperial College London
- Diane Coyle, University of Cambridge
- Cristina Caffarra, Charles River Associates
- Sean Ennis, Centre for Competition Policy and Norwich Business School, University of East Anglia
- Gregory S. Crawford, University of Zurich and CEPR Competition Policy RPN
- Fiona Scott Morton, Yale School of Management
- Monika Schnitzer, Ludwig-Maximilians-University of Munich
- Jacques Crémer, Toulouse School of Economics