Regulating digital platforms via competition and data-based approaches
Parminder Singh calls on developing countries to take the lead in establishing the policy framework for a globally fair and just digital economy, one that is not dominated by a handful of corporate tech titans. Such a framework, he proposes, should combine competition regulation with policy that governs access to and use of the all-important digital resource, data.
Digital technology companies are facing a worldwide backlash as they are seen to be too powerful and not accountable.
The US is home to most of the world’s tech or digital companies, and also in general the global capital of libertarian and neoliberal values. But today even the US is witness to a strong anti-digital-corporation sentiment. Not just its government and politicians, but also the CEOs of top digital companies themselves, are seeking regulation of the market and social power of these companies.
For developing countries, this issue takes up a further geopolitical angle, as they fear finding themselves locked at the lower ends of the emerging digital value chains. They would like their domestic digital businesses to have the space and capability to develop and grow, and in time graduate to higher and higher levels of the digital value chains.
What options do developing countries have for achieving these twin objectives of regulating big tech companies, and enabling their domestic digital businesses to develop and grow?
Developing countries have for long been advised, largely by Northern actors, that a laissez faire approach is best in the digital sector. Now even the North seems quite concerned about unchecked and unaccountable digital power. The challenges faced by developing countries may, however, be somewhat unique.
Combine competition and data approaches
The European Union is largely taking a competition-enforcing approach to tech power, and some similar debates are afoot in the US. Initially focused on mergers and acquisitions, (competition policy) discussions have moved to regulating the relationships between platforms and smaller businesses dependent on them, and recently also to exploring breakups and structural separations.
At the same time, European countries like the UK and France, and some developing countries, are coming up with artificial intelligence (AI) strategies. These recognize that command over AI is the new industrial capacity, might and advantage. Such strategies centrally focus on facilitating access to data for domestic businesses, through new data institutions like data infrastructures, data trusts and data markets.
It is prima facie surprising that competition policy measures and AI/data policies in Europe seem to be pursued in silos, when they have considerable commonality of “operational area” and even of objectives. This can partly be explained by institutional separations, as traditional industrial-age agencies confront new digital realities.
While recommending use of tools of competition regulation for taming tech power, a recent paper from the EU competition regulator is also forthright about the limits of such approaches and the need for complementary data policy and regulation-based approaches.
The enormous digital power can be tamed in the public interest only by getting out of industrial-age thinking and institutional approaches. Traditional competition approaches need to seamlessly combine and work with data and AI policies. A single approach aimed at addressing the excesses of digital power is needed. It should proceed from a deep understanding of the data-based nature of such power, even if it outwardly wraps in industrial-age structures of traditional corporations and markets.
Such a holistic digital economy approach is best housed in a specialized digital economy unit. It should have adequate research capacity, developing the necessary new economic frameworks that are adequate to digital realities.
For developing countries, there are the special added issues. The first is of making sure that they do not fall prey to a new wave of digital colonization, and that their businesses are able to move to higher levels in global value chains rather than being further demoted.
Equally, they have to ensure that smaller, traditional economic actors, from MSMEs (micro, small and medium enterprises) and traders to farmers and workers, get fair treatment within the new digital economic ecosystems.
Even if competition policies provide some consumer protection and benefit, and possibly some de-concentration of digital power, by themselves they would not enable domestic digital businesses to develop and flourish.
As data flows into a few digitally powerful countries – basically the US and China – rendering them the hubs of AI, all economic controls will be exercised from there.
Any country’s digital industrial strength will basically be its AI or digital intelligence strength. And taking a cue from the earlier mentioned country AI strategies, these are crucially dependent on how a country manages and leverages its data.
(While the term “artificial intelligence” is more popular, it is better to speak of “digital intelligence”, which covers all kinds of data-based intelligence – starting right from basic data analytics – that are definitional to digital businesses.)
Need for data ownership policies
European countries still reckon that with their strong manufacturing strength and traditional consumer-facing businesses (like supermarkets), along with superior institutional capacities, they will be able to manage their data towards domestic digital strength just by developing appropriate data sharing/access institutions.
While even for the EU the effectiveness of such a strategy by itself is suspect, both the economic and institutional strength in developing countries are too low for them to develop the required data institutions without some degree of explicit, formal data rights and controls. The latter need to be provided by a specific data rights and ownership policy.
Such a policy would outline primary economic rights – or ownership – of the concerned national community, and various sub-communities, groups and individuals as contextually relevant, to “data about them”.
To repeat, without such data rights/ownership policies, it is practically impossible for developing countries to build the kind of AI industrial capacities that AI strategies of Northern countries propose for themselves, based on various new data institutions.
This is the principal justification for such policies and for opposing “global free flow of data” agreements, whose basic aim is to disable such possibilities.
Presented with a new, window-dressed framework for free global data flows at the recent G20 meeting in Osaka, BRICS (Brazil, Russia, India, China and South Africa) and some other developing countries resisted this renewed attempt at divesting developing countries from developing any economic policies on data even in the future.
Developing countries’ counter-term was “data for development”. This was a good pushback. However, it is time that developing countries worked deeper on concepts like “data for development”, implying the national right to preferentially use “their data” for their own development.
What conceptual, policy and legal frameworks does such a right derive from?
It is not so difficult to work in these directions, for instance, taking from the need for mandatory obligations to share data required for public purposes, like commuting data for smart traffic management; the need for developing data infrastructures, or even fair and well-functioning open data markets; and so on.
The kinds of questions to be explored are: to whom should a country’s agriculture, transportation, education, health, governance etc data belong, including the aggregate anonymized kind which does not even have any personal data protections?
Larger developing countries like South Africa, India and Indonesia – that coordinated well at Osaka – should get together to shape the needed principles and frameworks. Data is a complex area, but it need not all be figured out and explained at one go. It is much more practical to first develop larger framework principles, both as a basis to keep working together and for representing their position to the outside world at various forums.
Rwanda has a data sovereignty policy. India has employed the terms “community data” and “national data” in its recent draft e-commerce policy. This constitutes recognition of data as a collective economic resource which the concerned community has a right to preferentially employ for its own best benefit. The draft Indian policy also says that enabling legal and technical means will be worked out in this regard. This is a very promising start. However, India is facing considerable pressure from the US to back off from these positions.
It is therefore important that developing countries, especially the larger ones to start with, work together in these respects. Domestic data rights and ownership policies do not at all mean cutting off from global value chains. It is about being able to negotiate better with and within them, and to find an appropriate, potentially improvable place for local businesses and other economic actors in them.
Why developing countries must take the lead
As it has been developed following the “Silicon Valley model”, the digital economy is congenitally global. Geo-economic considerations in this regard tend to trump national-level policy ones. This is why the US, otherwise an avid enforcer of competition policies, is reluctant to do so in the digital area.
This is also the reason that, even with nearly no global digital platforms and with further diminishing prospects in the face of China becoming the US’s main competitor in this area, the EU still sides with the US on global digital trade proposals.
Developing countries therefore cannot wait for appropriate digital policy frameworks and solutions to emerge in the North which they can then contextually adopt, as happens in many other areas. Developing countries, especially the larger ones, will have to take the lead to develop required policy principles and frameworks for a globally fair and just digital economy.
As mentioned, this will require appropriately combining industrial-age competition regulation with digital-age data and digital intelligence (or AI)-related policy perspectives. Such a combination must be organic, flowing from a deep understanding of the nature of the digital economy. It will be best done under the aegis of a specialized digital economy agency under an appropriate institutional framework.
As discussed, competition-enforcing tools by themselves will not be able to address digital market power. Conversely, data policies too will remain ineffective without appropriate complementary competition measures. For instance, data portability is a key issue in data policies with important economic implications. However, absent strong competitors in any given market segment, it will still beg the question – porting data to whom?
There are increasing demands that bridling the huge power of digital corporations or platforms requires structural separation approaches. India disallows foreign direct investment based e-commerce platforms from selling their own goods in competition with third-party goods. This is called a “marketplace only model”. A recent academic paper in the US has also advocated such separation of platforms from the commerce taking place over them.
Such structural separation will go a considerable length towards checking platform power. However, even without their own competing goods and services, platforms would become unacceptably powerful in any sector. This will increasingly be the case as AI dominates economic organization and relationships.
Uber would be too powerful an actor in the urban transportation sector even if it owns no cars and does not compete with independent drivers on its platform. Also, what happens when – as is its ultimate aim – Uber is basically running a network of autonomous cars? These cars will be so fundamentally dependent on Uber-provided digital intelligence that they would for all intents and purposes, if not formally, be Uber-owned.
Similarly, even though, unlike Amazon, Alibaba does not sell its own products, it is extraordinarily, and unacceptably, powerful in the consumer goods value chain wherever it operates.
A case of data-based structural separation
The tightly integrated digital value chains no doubt require structural separation. But this should primarily hit at the integration of data elements of the value chain, as the principal anchors of all digital economy activity and value processes.
It may be best to explore structural separation between data collection points/businesses and data-processing ones which convert the data into digital intelligence. Data collection takes place at both consumer-facing businesses, like retail, and sector-specific product/service developing businesses, like manufacturing. Meanwhile data processing is the forte of tech companies like Google that convert the data into sophisticated digital intelligence or AI. Data-derived digital intelligence is then applied back at consumer-facing as well as product/sector development levels or businesses.
Separation of the two therefore also amounts to structural separation of businesses providing AI or digital intelligence as a service, and those applying such services in specific sectoral contexts and situations.
To illustrate this with an example, Google can develop general transportation-related AI services and provide these services to, say, Ford Motor Company, but cannot compete with the latter to actually provide physical transportation products or services. In turn, Google will require some means to access transportation data that can only be collected at manufacturing and/or actual transportation ends.
Such separation will have interposed, on one side, various structured means of data sharing and access, like data infrastructures, data trusts and data markets. On the other side, there will exist an AI or digital intelligence services market, which will be relatively open and competitive. Various sectoral businesses, whether in production or retail, will buy AI or digital services but maintain their relative independence in an – expected – considerably competitive market for such services.
These are some preliminary points framing the larger principles of a model of data and AI services-based structural separation. These would of course need to be applied in a nuanced and contextual manner.
It would, for instance, not be amiss for retail and manufacturing businesses to process their data for developing AI internally or among their business networks. They may not, however, develop a generic AI-based application that provides services across the sector or generally. Similarly, the AI application providers may be able to themselves collect some kinds of data but not other important kinds.
Constraints of space do not allow further expansion here on this proposed model that combines competition and data approaches for taming mega tech or digital power. It may however be summarily mentioned that such a model aims in particular at enabling developing countries to:
(1) leverage the value of locally collected data, both as a collective resource in the form of data infrastructures – access to which can be conditional – and as private data of businesses in key positions to collect data that then gets monetized through data markets; and
(2) enable domestic businesses to develop increasing data-related capabilities and to make enough profits to be able to structurally change in the direction of graduating up the global digital value chains. (SUNS8957)
Parminder Singh is Executive Director of India-based IT for Change. The above is based on a presentation he made at a recent expert panel on “Regulation of Digital Platforms for Economic Development” arranged by the Centre for Competition, Regulation and Economic Development at the University of Johannesburg, and Department of Trade and Industry, South Africa.
Third World Economics, Issue No. 686, 1-15 April 2019, p14-16