The Digital Transition of Services: Getting Regulation Right From the Start

The digital transition and the introduction of artificial intelligence (AI) in services are blurring the boundaries of both occupations and sectors. At the same time, new or rekindled trade-offs between different regulatory objectives arise, transgressing sectoral and occupational boundaries as we know them. Services regulations, however, are still typically sector or occupation specific. Reforming them is urgent to address the regulatory challenges of the digital era.

Regulation of the professions: fit for purpose?

The digital transition of professional services opens new opportunities for drawing on a multitude of experiences, skills, and capabilities to embark on an inclusive, sustainable, and digital future. In architecture, for instance, some of the activities that are traditionally reserved for licensed architects can now be performed by AI-enabled software in the hands of other professions including engineers, city planners, and the ultimate customer.

In the health sector, AI-enabled data analytics and software support diagnostics, choice of treatment, patient monitoring, prediction of outcomes, and automated medication.[1] Some of these technologies can be safely operated by semi-skilled health workers, and some can be fully automated. Patients can also more often go about their normal lives with automated medication, sensors monitoring e.g., heartbeat or blood sugar and alerting the family doctor, the local hospital, or emergency services, when needed.

Against this backdrop, traditional regulation of professions in the form of reserving a given set of tasks and functions for certain occupations may no longer serve its intended purpose. There is for instance some evidence that occupational licensing nowadays protects the professionals rather than the consumers.[2] Furthermore, such regulation may hold back the adoption of technology in services. That means lost opportunities for job creation in potentially competitive and rapidly growing mass markets for services that only high-income households can currently afford.

Telecommunications: trade-off between static and dynamic efficiency

At the heart of the digital transition lies data, the networks over which data are transmitted and the devices and services for storing and processing data. Therefore, telecommunications have warranted a separate chapter addressing regulatory issues in most trade agreements.

Uncompetitive behavior by incumbent telecommunications suppliers could easily undermine market access negotiated at the WTO and in free trade agreements. Recognizing the complementarity between market access and competition, the WTO introduced a reference paper on pro-competitive regulation to be added to countries’ schedules of commitments in the General Agreement on Trade in Services (GATS). It prescribes regulatory obligations to be imposed on the incumbent operator, forcing it to offer access to its network and to interconnect with rival entrants. These obligations are to be enforced by an independent and sufficiently resourced regulator.[3]

When the reference paper was introduced in the 1990s, telecommunications were mainly used for voice calls over fixed lines. Markets were well-defined and there were clear criteria and processes leading to regulatory interventions. These are necessary features for pro-competitive regulation to be effective and achieve its goals.

Fast forward to the 2020s, the digital transition has blurred the boundaries between telecommunications and e.g., internet services. Markets are harder to define, dominance more difficult to identify and the standards for what constitutes anti-competitive behavior have changed with rapidly evolving technology including the ubiquitous presence of digital platforms.

A trade-off between static and dynamic efficiency has emerged with the opening of telecommunications markets to trade and competition. In the short run, ex ante access regulation lowers entry barriers and keeps the incumbent from abusing its market power, ensuring static efficiency. In the long run such regulation may discourage investment in infrastructure such as fiber to the home or 5G mobile networks, both by the incumbent and entrants. Today, fewer and fewer markets are considered susceptible to non-transitory significant market power (SMP) and there is a global trend towards rolling back regulation and subject telecommunications to the general competition law and its enforcement.

Against this backdrop the reference paper on telecommunications clearly needs to be revisited, including assessing whether it has outlived its usefulness.

Competition and innovation: friends or foes?

The relationship between competition and innovation is quite complex. At an economy-wide level it is inverted u-shaped. When the economy starts off at a low level of competition, firms respond to rising competitive pressure by innovating to escape competition. They may engage in product innovation to stand out with new and different products, or they may focus on process innovation to reduce costs relative to competitors. As competitive pressure rises above a certain threshold, however, the margins become too small for this mechanism to work, and innovation declines with further increases in competition. When on the upward slope of the curve, competition and innovation reinforce each other, while when on the downward slope, the trade-off applies.

In the digital economy, innovation adds further complexities to defining the market for competition policy purposes as new products and services lead to convergence between communications and information services. The market definitions upon which regulation is based then change with the uptake of such innovations.[4] Social media platforms, for instance, offer services similar to telecommunications, and need to be taken into consideration when defining markets and identifying SMPs.

More generally, a distinction between market power as a capacity and anti-competitive behavior is gaining prominence in the policy debate. Market power may stem from economies of scale on the demand side, i.e., demand-driven network effects. In such cases consumers are better off the larger the platform. Market power as a capacity is thus not necessarily a problem unless combined with anti-competitive behavior. [5]

This point is particularly acute when data are a source of competitive advantage. Huge amounts of data are needed for innovation, notably for the development of AI applications such as image recognition, computer vision, speech recognition, language translation and reasoning. At the same time, huge amounts of data can be the source of market power – often generated by demand side network effects.

Privacy and competition: another possible trade-off

A suggested way of preempting abuse of data-based market power is to mandate data portability such that consumers and corporate users of platforms can take their data to another platform with ease. The idea stems from experience with number portability in the mobile and fixed telephone market, and is embedded in for instance, the EU General Data Protection Regulation (GDRP). A related remedy is mandating data sharing. This is also inspired by telecommunications regulation where SMPs are obliged to offer entrants access to their network on non-discriminatory terms.

In either case, technical issues arise as platforms as well as entrants may use different formats and standards. Furthermore, smaller firms often outsource data management to third parties. The interface between the incumbent and entrants can thus constitute a weak link in the chain of protected data flows. Standardization is therefore a prerequisite for data portability and data sharing to work.[6]

The largest global platforms (Google, Facebook, Microsoft and Twitter) have already introduced a data transfer project that contributes to interoperability and facilitates data portability. Regulators and other stakeholders are not convinced that such self-regulation and market initiatives are sufficient, however.

Internal and external regulatory consistency

Services regulation is typically embedded in the economy’s institutional and social fabric which may vary substantially across economies. Some examples illustrate the point. Countries that have an extensive rights-based and government-funded social safety net, including health, education, pensions, and unemployment benefits, may see less need for regulating insurance services than countries that rely on private insurance for such safety nets.

The building industry is another example. Some countries focus regulation on the services providers, including the regulation of architects and engineers as well as requiring authorization for a host of crafts. Others regulate mainly through the building codes, the processes and requirements related to obtaining a building permit, and inspections at the building site during construction.

Services regulation is – at its best – part and parcel of a coherent and consistent domestic regulatory framework. Specific regulations – or lack thereof – may, however, be inconsistent with trading partners’ regulation. In such cases, mutual recognition of qualifications and standards may generate trade-offs between internal and external consistencies that cannot be fully resolved in the short term. In the long run, however, forward-looking regulatory reforms could alleviate such inconsistencies. During the digital transition, reforms could aim at common regulatory objectives, with built-in flexibility to accommodate new developments in technology and markets, while sharing best practice.

As we highlight in a recent APEC study, we live in transformative times where clear answers and well-proven policies are yet to be established. This poses unique opportunities for experimenting and collaboration on forward-looking regulation, which can best be realized within institutional frameworks of trust. Standing at the cusp of an AI-driven technological revolution, we have a unique opportunity to get this right from the start.



Aghion, P., Bloom, N., Blundell, R., Griffith, R., & Howitt, P. (2005). Competition and innovation: An inverted-U relationship. The quarterly journal of economics120(2), 701-728.

Blair, P. Q., & Chung, B. W. (2019). How much of barrier to entry is occupational licensing?. British Journal of Industrial Relations57(4), 919-943.

Borgogno, O., & Colangelo, G. (2019). Data sharing and interoperability: Fostering innovation and competition through APIs. Computer Law & Security Review35(5), 105314.

Corrado, C., et al. (2021), “New evidence on intangibles, diffusion and productivity”, OECD Science, Technology and Industry Working Papers, No. 2021/10, OECD Publishing, Paris,

Evenenett, S. and Fritz, J. (2022). Emergent digital fragmentation The perils of unilateralism.

ITU (2020). Digital Regulation Handbook.

Jiang, F., Jiang, Y., Zhi, H., Dong, Y., Li, H., Ma, S., … & Wang, Y. (2017). Artificial intelligence in healthcare: past, present and future. Stroke and vascular neurology2(4).

Jullien, B., & Sand-Zantman, W. (2021). The economics of platforms: A theory guide for competition policy. Information Economics and Policy54, 100880.

Kleiner, M. M. (2006). Licensing occupations: Ensuring quality or restricting competition?. WE Upjohn Institute.

OECD (2022a) OECD Services trade restrictiveness index: Policy trends up to 2022.

OECD (2022b). The evolving concept of market power in the digital economy. OECD Competition Policy Background Note.

Vogelsang, I. (2017). The role of competition and regulation in stimulating innovation–Telecommunications. Telecommunications Policy41(9), 802-812.

Yu, K. H., Beam, A. L., & Kohane, I. S. (2018). Artificial intelligence in healthcare. Nature biomedical engineering2(10), 719-731.

[1] See Jang et al. (2017) and Yu et al. (2018).

[2] See Kleiner (2006) and Blair and Chung (2019).

[3] The incumbent was typically a private or government-owned dominant supplier, also coined suppliers with significant market power (SMP) in trade agreements and legislation.

[4] See Vogelsang (2017)

[5] See Julien and Sand-Zantman (2021).

[6] See Borgogno and Colangelo (2019).